Cell 100806

Noncoding RNAsOrchestrate p53

Volum

e 142 Num

ber 3 Pages 335–496 A

ugust 6, 2010

Volume 142

www.cell.com

Number 3

August 6, 2010

INSERT ADVERT

cell142_3.c1.indd 1cell142_3.c1.indd 1 7/30/2010 1:33:06 AM7/30/2010 1:33:06 AM

ANALYZE

AAEXPAND

EXP

IS

OLATE

PRODUCE

For Research Use Only. Not for use in diagnostic or therapeutic procedures.BD, BD Logo and all other trademarks are property of Becton, Dickinson and Company. © 2010 BD23-11458-00

BD Biosciences2350 Qume DriveSan Jose, CA 95131bdbiosciences.com

Powerful platforms and fl exible solutions that enable your work to fl ow.

From isolation through production, todayBD Biosciences delivers high quality products and services, meeting your needs at every step in the cell-based workfl ow.

As research moves toward clinical trials and ultimately new cell therapies, you can rely on BD’s proven experience in delivering quality GMP products and services as we deliver the nextgeneration of tools and systems to help you trans-form disease management in the next decade.

Systematic ExcellenceSM is our commitment to deliver both best-in-class and highly integrated products and services to help you reduce risk and to ensure superior quality in the production of clinical-grade applications.

BD Systematic ExcellenceSM: Powering the next generation of tools and systems to transform disease management in the next decade.

Visit us at bdbiosciences.com/stemcells.

Integrated solutions from BD

SYSTEMATICEXCELLENCESM

23-11458-00i_Cell.indd 123-11458-00i_Cell.indd 1 7/9/10 9:09:59 AM7/9/10 9:09:59 AM

ORDER 888.999.1371 • TECH 888.810.6168 • INQUIRIES [email protected]

Cell: trim 8.375x10.875, bleed .125, margin .25

FlowCytomix™ are bead-based immunoassays for multiple analyte detection on the � ow cytometer.

• Measure up to 20 analytes simultaneously from only 25 µl of sample

• Choose pre-con� gured panels or create your own custom combinations

• Easy-to-use analysis software included

• For use on most � ow cytometers, including single laser instruments

Th1/ Th2/ Th9/ Th17/ Th22Multiple Analyte Detection Kits! NEW! Human Th Di� erentiation 13plex Cat. No. BMS817FF (IFN-у, IL-1β, IL-2, IL-4, IL-5, IL-6, IL-9, IL-10, IL-12 (p70),

IL-13, IL-17A, IL-22, TNF-α)

NEW! Mouse Di� erentiation 13plex Cat. No. BMS822FF (IFN-у, IL-1α, IL-2, IL-4, IL-5, IL-6, IL-10, IL-13, IL-17A,

IL-21, IL-22, IL-27, TNF-α)

Visit www.eBioscience.com and search FlowCytomix.

5 T helper cell profiles in 1 kit!

FlowCytomix™

Naїve Th

Th1

APC

Th2

Th9

Th17

Th22

Q310014-FlowCytomix-Cell.indd 1 7/13/10 4:16 PM

Limited Time Free Offer. ($3,795 Value) For a third level of security during brief power outages SANYO offers a free, built-in Liquid N2 or Liquid CO2 back-up system with the purchase of any MDF-U500VXC now through September 30, 2010.

Limited Time Free Offer. ($3,795 Value)For a third level of security during brief power outages SANYO offers a free, built-in Liquid N or Liquid CO back-up system with the purchase of any

©2010 Sanyo Biomedical

Pictured: The 18.3 cu.ft. V.I.P.™ insulated MDF-U500VXC. Includes integrated LCD performance monitor and digital controller

for comprehensive system management, data logging, remote communications, alarms, predictive performance and

validation. Maintenance free, filterless design.

www.twinguardseries.com

FREEOFFER [ $3,795 VALUE ]

! GREENproduct

Nothing Beats 100% Protection.Except 200% Protection.Nothing Beats 100% Protection.Except 200% Protection.If you’re preserving the work of a lifetime, this -86°C freezer is for you.

SANYO introduces Twin Guard Series™, the most reliable dual, redundant autocascade freezer for the most critical -86°C ultra-low applications ever. The V.I.P.™ insulated MDF-U500VXC is powered by two completely separate -86 Dual˚Cool refrigeration systems, each built to run effi ciently – alone or even better together in energy-saving EcoMode™ – to safely preserve whatever you put inside. Learn more. Visit www.twinguardseries.com or call 800-858-8442.

Except 200% Protection.Except 200% Protection.Except 200% Protection.Except 200% Protection.Except 200% Protection.Except 200% Protection.Except 200% Protection.Except 200% Protection.Except 200% Protection.Except 200% Protection.Except 200% Protection.Except 200% Protection.Except 200% Protection.Except 200% Protection.Except 200% Protection.Except 200% Protection.Except 200% Protection.Except 200% Protection.Except 200% Protection.Except 200% Protection.Except 200% Protection.Except 200% Protection.If you’re preserving the work of a lifetime, this -86°C freezer is for you. If you’re preserving the work of a lifetime, this -86°C freezer is for you. If you’re preserving the work of a lifetime, this -86°C freezer is for you. If you’re preserving the work of a lifetime, this -86°C freezer is for you. If you’re preserving the work of a lifetime, this -86°C freezer is for you. If you’re preserving the work of a lifetime, this -86°C freezer is for you. If you’re preserving the work of a lifetime, this -86°C freezer is for you. If you’re preserving the work of a lifetime, this -86°C freezer is for you. If you’re preserving the work of a lifetime, this -86°C freezer is for you.

the most reliable dual, redundant autocascade

If you’re preserving the work of a lifetime, this -86°C freezer is for you. If you’re preserving the work of a lifetime, this -86°C freezer is for you. If you’re preserving the work of a lifetime, this -86°C freezer is for you. If you’re preserving the work of a lifetime, this -86°C freezer is for you. If you’re preserving the work of a lifetime, this -86°C freezer is for you. If you’re preserving the work of a lifetime, this -86°C freezer is for you. If you’re preserving the work of a lifetime, this -86°C freezer is for you. If you’re preserving the work of a lifetime, this -86°C freezer is for you. If you’re preserving the work of a lifetime, this -86°C freezer is for you. If you’re preserving the work of a lifetime, this -86°C freezer is for you. If you’re preserving the work of a lifetime, this -86°C freezer is for you. If you’re preserving the work of a lifetime, this -86°C freezer is for you. If you’re preserving the work of a lifetime, this -86°C freezer is for you. If you’re preserving the work of a lifetime, this -86°C freezer is for you. If you’re preserving the work of a lifetime, this -86°C freezer is for you. If you’re preserving the work of a lifetime, this -86°C freezer is for you. If you’re preserving the work of a lifetime, this -86°C freezer is for you. If you’re preserving the work of a lifetime, this -86°C freezer is for you. If you’re preserving the work of a lifetime, this -86°C freezer is for you.

the most reliable dual, redundant autocascade

B

-86˚C

A

Twin Guard Series™ with-86 Dual˚Cool Technology

Editor

Emilie Marcus

Senior Deputy Editor

Elena Porro

Scientific Editors

Karen Carniol

Connie M. Lee

Fabiola Rivas

Lara Szewczak

Leading Edge Editor

Orla M. Smith

Associate Editor

Robert Kruger

Assistant Editor

Michaeleen Doucleff

Senior Managing Editor

Meredith Adinolfi

Deputy Managing Editor

Jackie M. Smith

Illustration and

Design Supervisor

Nathan E. Bress

Lead Illustrator

Andrew A. Tang

Illustrators

Yvonne Blanco

Kate Mahan

Production Staff

Reyna Clancy

Editorial Assistant

Mary Beth O’Leary

Editorial Board

Abul Abbas

C. David Allis

Genevieve Almouzni

Uri Alon

Angelika Amon

Johan Auwerx

Richard Axel

Cori Bargmann

Konrad Basler

Bonnie Bassler

David Baulcombe

Jeffrey Benovic

Carolyn Bertozzi

Wendy Bickmore

Elizabeth Blackburn

Joan Brugge

Lewis Cantley

Joanne Chory

David Clapham

Andrew Clark

Hans Clevers

Stephen Cohen

Pascale Cossart

Jeff Dangl

Ted Dawson

Pier Paolo di Fiore

Marileen Dogterom

Julian Downward

Bruce Edgar

Steve Elledge

Anne Ephrussi

Ronald Evans

Witold Filipowicz

Marco Foiani

Elaine Fuchs

Yukiko Goda

Stephen Goff

Joe Goldstein

Douglas Green

Leonard Guarente

Taekjip Ha

Daniel Haber

Ulrike Heberlein

Nobutaka Hirokawa

Mark Hochstrasser

Arthur Horwich

Tony Hunter

James Hurley

Richard Hynes

Thomas Jessell

Narry Kim

Mary-Claire King

David Kingsley

Frank Kirchhoff

Richard Kolodner

John Kuriyan

Robert Lamb

Mark Lemmon

Beth Levine

Wendell Lim

Jennifer Lippincott-Schwartz

Dan Littman

Richard Losick

Scott Lowe

Tom Maniatis

Matthias Mann

Kelsey Martin

Joan Massague

Iain Mattaj

Satyajit Mayor

Ruslan Medzhitov

Craig Mello

Tom Misteli

Tim Mitchison

Alex Mogilner

Paul Nurse

Roy Parker

Dana Pe’er

Kathrin Plath

Carol Prives

Klaus Rajewsky

Venki Ramakrishnan

Rama Ranganathan

Anne Ridley

James Roberts

Alexander Rudensky

Helen Saibil

Joshua Sanes

Randy Schekman

Ueli Schibler

Joseph Schlessinger

Hans Scholer

Trina Schroer

Geraldine Seydoux

Kevan Shokat

Pamela Sklar

Nahum Sonenberg

James Spudich

Paul Sternberg

Bruce Stillman

Azim Surani

Keiji Tanaka

Craig Thompson

Robert Tjian

Jurg Tschopp

Ulrich von Andrian

Gerhard Wagner

Detlef Weigel

Alan Weiner

Jonathan Weissman

Matthew Welch

Tian Xu

Shinya Yamanaka

Marino Zerial

Xiaowei Zhuang

Huda Zoghbi

Cell Office

Cell, Cell Press, 600 Technology Square, 5th Floor, Cambridge, Massachusetts 02139

Phone: (+1) 617 661 7057, Fax: (+1) 617 661 7061, E-mail: [email protected]

Online Publication: http://www.cell.com

Cell (ISSN0092-8674) is published biweeklybyCell Press, 600TechnologySquare, 5th Floor,Cambridge, Massachusetts02139.The institutional subscription rate for

2010 is $1,360 (US and Canada) or $1,532 (elsewhere). The individual subscription rate is $202 (US and Canada) or $305 (elsewhere). The individual copy price is $50.

Periodicals postage paid at Boston, Massachusetts and additional mailing offices. Postmaster: send address changes to Elsevier Customer Service Americas,

Cell Press Journals, 11830 Westline Industrial Drive, St. Louis, MO 63146, USA.

The paper used in this publication meets the requirments of ANSI/NISO Z39.48-1992 (Permanence of Paper). Printed by Dartmouth Printing Company, Hanover, NH.

xCELLigence RTCA DP System

Leave endpoint assays behind and move to real time

Lid ——————————————————————

Upper Chamber (cells) ——————————————————

Membrane with pores ——————————————

Sensor on bottom

————————————————————————— surface

———————————————— Lower Chamber (media, chemoattractant)

CIM-Plate 16 for cell invasion and migration analysis on

the xCELLigence RTCA DP System.

Monitor cell invasion and migration continuously in real time. Go beyond the limits of endpoint assays by using

the xCELLigence RTCA DP Instrument and CIM-Plate 16 for label-free measurement of physiologically relevant cell

responses.

* Boyden Chamber, transwell

For more information, visit www.xcelligence.roche.com

For life science research only. Not for use in diagnostic procedures.XCELLIGENCE is a trademark of Roche. CIM-PLATE, E-PLATE and ACEA BIOSCIENCES

are registered trademarks of ACEA Biosciences, Inc. in the U.S.

© 2010 Roche Diagnostics. All rights reserved.

Roche Diagnostics Corporation Roche Applied Science Indianapolis, Indiana

CIM-Plate 16

CIM-Plate 16E-Plate 16

Current Assays *

RTCA-CIM_Ad_US.indd 1 7/14/2010 9:41:50 AM

Cell Press

President & CEO

Lynne Herndon

Editor in Chief, Vice President of Content Development

Emilie Marcus

Vice President of Marketing and Publishing

Els Bosma

Vice President of Web Development and Operations

Keith Wollman

Director of Marketing

Jonathan Atkinson

Production Supervisor

Meredith Adinolfi

Operations Supervisor

Daniel Fauxsmith

Press Officer

Cathleen Genova

Display Advertising

Midwest/Southeast/Eastern Canada:

Inez Herrero, ph: 585 678 4395; fax: 585 678 4722;

e-mail: [email protected]

Mid-Atlantic/New England: Vikki Macomber, ph: 508 928

1255; fax: 508 928 1256; e-mail: [email protected]

West Coast/Western Canada/Asia: Lynne Stickrod, ph: 415

931 9782; fax: 415 520 6940; e-mail: [email protected]

UK/Europe: James Kenney, ph: +44 20 7424 4216; fax: +44 18

6585 3136; e-mail: [email protected]

Classified Advertising

United States and Canada:

Gordon Sheffield, Key Account Manager, ph: 617 386 2189; fax:

617 397 2805; e-mail: [email protected]

UK, Europe, and Asia:

Sabrina Dodge, Key Account Manager, ph: +44 20 7424 4997;

fax: +44 18 6585 3136; e-mail: [email protected]

ª2010 Elsevier Inc. All rights reserved.

This journal and the individual contributions contained in it are protected

under copyright by Elsevier Inc., and the following terms and conditions

apply to their use:

Photocopying:

Single photocopies of single articles may be made for personal use as al-

lowed by national copyright laws. Permission of the Publisher and payment

of a fee arerequired forall otherphotocopying, including multipleorsystem-

atic copying, copying for advertising or promotional purposes, resale, and

all forms of document delivery. Special rates are available for educational

institutions that wish to make photocopies for nonprofit educational class-

room use. For information on how to seek permission, visit www.elsevier.

com/permissions or call (+44) 1865 843830 (UK) / (+1) 215 239 3804 (US).

Permissions:

For information on how to seek permission, visit www.elsevier.com/

permissions or call (+44) 1865 843830 (UK) / (+1) 215 239 3804 (US).

Derivative Works:

Subscribers may reproduce tables of contents or prepare lists of articles

including summaries for internal circulation within their institutions.

Permission of the Publisher is required for resale or distribution outside

the institution. Permission of the Publisher is required for all other deriv-

ative works, including compilations and translations (please consult

www.elsevier.com/permissions).

Electronic Storage or Usage:

Permission of the Publisher is required to store or use electronically any

material contained in this journal, including any article or part of an article

(please consult www.elsevier.com/permissions). Except as outlined

above, no part of this publication may be reproduced, stored in a retrieval

system, or transmitted in any form or by any means, electronic, mechan-

ical, photocopying, recording, or otherwise, without prior written permis-

sion of the Publisher.

Notice:

No responsibility is assumed by the Publisher for any injury and/or damage

to persons or property as a matter of products liability, negligence, or other-

wise, or from any use or operation of any methods, products, instructions,

or ideas contained in the material herein. Because of rapid advances in

the medical sciences, in particular, independent verification of diagnoses

and drug dosages should be made. Although all advertising material is

expected to conform to ethical (medical) standards, inclusion in this

publication does not constitute a guarantee or endorsement of the quality

or value of such product or of the claims made of it by its manufacturer.

Reprints:

Article reprints are available through Cell’s reprint service; for informa-

tion, contact Nicholas Pavlow (e-mail: [email protected]; ph: (+1)

212 633 3960).

Subscription Orders and Inquiries:

Mail, fax, or e-mail address changes to Elsevier Customer Service Amer-

icas, allowing 4–6 weeks for processing. Lost or damaged issues will be

replaced, subject to availability, if Cell Press is notified within the claim

period (US and airmail delivery: 3 months from issue date; surface deliv-

ery: 4 months from issue date). Periodical delivery in the US can take up

to 3 weeks. Airmail delivery can take 2–4 weeks.

The price of a single copy of Cell is $50 (excluding special issues). All

orders must be prepaid and in writing. Please include the volume and is-

sue number, payment (check or credit card, MasterCard, Visa, or Amer-

ican Express only), and a delivery address. Allow 4–6 weeks for delivery.

Mailing address: Elsevier Customer Service Americas, Cell Press

Journals, 11830 Westline Industrial Drive, St. Louis, MO 63146,

USA. Toll-free phone within USA/Canada: 866 314 2355; phone for

outside US/Canada: (+1) 314 453 7038; fax: (+1) 314 523 5170; e-mail:

[email protected]; internet: www.cellpress.com or <www.cell.com>.

Funding Body Agreements and Policies:

Elsevier has established agreements and developed policies to allow au-

thors whose articles appear in journals published by Elsevier to comply

with potential manuscript archiving requirements as specified as condi-

tions of their grant awards. To learn more about existing agreements and

policies, visit www.cellpress.com.

Guide for Authors:

For a full and complete guide for authors, please go to www.cell.com/

authors.

REMARKABLY FAST

www.neb.com

CLONING & MAPPING DNA AMPLIFICATION& PCR RNA ANALYSIS PROTEIN EXPRESSION

& ANALYSISGENE EXPRESSION

& CELLULAR ANALYSIS

Not all PCR polymerases are created equal

Amplifi cation of a 3.8 kb fragment from the human beta globin gene clearly illustrates the extreme speed and robustness offered by using Phusion DNA Polymerase. Reactions were performed according to the suppliers’ recommendations using varying extension times (shown above gel).

Phusion is a registered trademark of Finnzymes Oy.

Amplifi cation of a 3.8 kb fragment from the human beta globin gene clearly illustrates the

Phusion® High-Fidelity DNA Polymerase With Phusion® High-Fidelity DNA Polymerase, there is no need to compromise any aspect of your PCR performance.

A superior choice for cloning, this recombinant polymerase has an error rate 50-fold lower than Taq DNA Polymerase,

combining extreme precision with unparalleled speed and robustness.

Advantages:• Extreme fi delity

• High speed

• Robustness

Distributed byProduced by

Go to www.neb.com/phusion to fi nd out

how Phusion High-Fidelity DNA Polymerase can

improve your PCR performance.

• High yield

• Specifi city

Leading EdgeCell Volume 142 Number 3, August 6, 2010

IN THIS ISSUESTEM CELL SELECT

ANALYSIS

347 Stitching Together Cross-border Research C. Macilwain

CORRESPONDENCE

350 Successes of Genome-wide Association Studies R.J. Klein, X. Xu, S. Mukherjee,J. Willis, and J. Hayes

351 Strategies for Genetic Studies of Complex Diseases K. Wang, M. Bucan, S.F.A. Grant,G. Schellenberg, and H. Hakonarson

353 Response: Why It Is Time to Sequence J. McClellan and M.C. King

PREVIEWS

356 Fyn-Tau-Amyloid: A Toxic Triad C. Haass and E. Mandelkow

358 Noncoding RNAs: The Missing ‘‘Linc’’in p53-Mediated Repression

A.M. Barsotti and C. Prives

360 Stem Cells and DNA Damage: Persist or Perish? A.A. Lane and D.T. Scadden

362 Mitochondrial Matrix Reloaded with RNA T. Endo, K. Yamano, and T. Yoshihisa

MINIREVIEW

364 A MAP for Bundling Microtubules C.E. Walczak and S.L. Shaw

PERSPECTIVE

368 The Pioneer Round of Translation:Features and Functions

L.E. Maquat, W.-Y. Tarn, and O. Isken

SNAPSHOT

496 Nonhomologous DNA End Joining (NHEJ) M.R. Lieber and T.E. Wilson

© 2010 Cell Signaling Technology, Inc.

Cell Signaling Technology® and PhosphoSitePlus

® are trademarks of Cell Signaling Technology, Inc.

Orders (toll-free) 1-877-616-2355 | Technical support (toll-free) 1-877-678-8324 [email protected] | Inquiries [email protected] | Environmental Commitment eco.cellsignal.com

www.cellsignal.comUnparalleled product quality, validation and technical support.

PhosphoSitePlus® v3.0 (PSP) is an open web resource that integrates encyclopedic information on experimentally determined protein modifi cation sites, upstream and downstream regulation of these modifi cations, and powerful analytical tools for investigating the structural and biological signifi cance of protein modifi cations.

www.phosphosite.org

• Expansive and continuously curated content

• Molecular rendering to visualize the location of modifi cation sites

• On-the-fl y generation of kinase substrate sequence logos

• Browsing of high-throughput content by disease, cell line, and tissue

• New search interfaces that retrieve modifi cation sites and proteins by subcellular locations, sequence and motifs, domains, responsiveness to treatments, disease, tissue, and cell type

A Comprehensive Online Protein Modification Resource

provided by Cell Signaling Technology with grant support from the NIH

ArticlesCell Volume 142 Number 3, August 6, 2010

375 Direct Reprogramming of Fibroblastsinto Functional Cardiomyocytesby Defined Factors

M. Ieda, J.-D. Fu, P. Delgado-Olguin,V. Vedantham, Y. Hayashi, B.G. Bruneau,and D. Srivastava

387 Dendritic Function of Tau MediatesAmyloid-b Toxicity inAlzheimer’s Disease Mouse Models

L.M. Ittner, Y.D. Ke, F. Delerue, M. Bi, A. Gladbach,J. van Eersel, H. W€olfing, B.C. Chieng, M.D.J. Christie,I.A. Napier, A. Eckert, M. Staufenbiel, E. Hardeman,and J. G€otz

398 Single-Stranded DNA TranspositionIs Coupled to Host Replication

B. Ton-Hoang, C. Pasternak, P. Siguier,C. Guynet, A.B. Hickman, F. Dyda, S. Sommer,and M. Chandler

409 A Large Intergenic Noncoding RNAInduced by p53 Mediates Global GeneRepression in the p53 Response

M. Huarte, M. Guttman, D. Feldser, M. Garber,M.J. Koziol, D. Kenzelmann-Broz, A.M. Khalil,O. Zuk, I. Amit, M. Rabani, L.D. Attardi, A. Regev,E.S. Lander, T. Jacks, and J.L. Rinn

420 A Minimal Midzone Protein ModuleControls Formation and Length ofAntiparallel Microtubule Overlaps

P. Bieling, I.A. Telley, and T. Surrey

433 Insights into Antiparallel MicrotubuleCrosslinking by PRC1, a ConservedNonmotor Microtubule Binding Protein

R. Subramanian, E.M. Wilson-Kubalek, C.P. Arthur,M.J. Bick, E.A. Campbell, S.A. Darst, R.A. Milligan,and T.M. Kapoor

444 Aurora Kinases and Protein Phosphatase 1Mediate Chromosome Congressionthrough Regulation of CENP-E

Y. Kim, A.J. Holland, W. Lan, and D.W. Cleveland

456 PNPASE Regulates RNA Importinto Mitochondria

G. Wang, H.-W. Chen, Y. Oktay, J. Zhang, E.L. Allen,G.M. Smith, K.C. Fan, J.S. Hong, S.W. French,J.M. McCaffery, R.N. Lightowlers, H.C. Morse III,C.M. Koehler, and M.A. Teitell

468 Plzf Regulates Germline ProgenitorSelf-Renewal by Opposing mTORC1

R.M. Hobbs, M. Seandel, I. Falciatori, S. Rafii,and P.P. Pandolfi

(continued)

Working with stem cells ?

PromoCell GmbHNorth America: 1 - 866 - 251 - 2860 (toll free)Deutschland: 0800 - 776 66 23 (gebührenfrei) United Kingdom: 0800 - 96 03 33 (toll free)France: 0800 90 93 32 (ligne verte)Other countries: +49 6221 - 649 34 0

Go for PromoCell ! Your partner for stem cell cultivation

Mesenchymal Stem Cells & Pericytes

CD34 + & CD133+ Progenitor CellsCD14 + Monocytes

www.promocell.com

Stem Cell Promotion !www.promocell.com/promotion

new

480 Myc-Nick: A Cytoplasmic CleavageProduct of Myc that Promotes a-TubulinAcetylation and Cell Differentiation

M. Conacci-Sorrell, C. Ngouenet, and R.N. Eisenman

ERRATUM

494 SIRT1 Suppresses b-Amyloid Productionby Activating the a-Secretase Gene ADAM10

G. Donmez, D. Wang, D.E. Cohen, and L. Guarente

ANNOUNCEMENTS

POSITIONS AVAILABLE

On the cover: When p53 is active in cells, it, like an orchestra conductor, coordinates the

activation and repression of many cellular players. Among these are long intergenic noncod-

ing RNAs (lincRNAs) that contribute to the balanced modulation of transcription in the p53

response. Huarte et al. (pp. 409–419) show that lincRNA-p21, represented as a baton, is a di-

rect transcriptional target of p53 and is necessary for the repression of many target genes,

particularly those involved in apoptosis downstream of p53 activation. Artwork by Lauren

Solomon, John Rinn, Maite Huarte, and Sigrid Hart, Broad Institute; adapted from artwork

by iStockphoto.

Proteins

Antibodies

ELISAs

Assay Services

MultiAnalyte Pro� ling

Activity Assays

Stem Cells

ELISpot Kits

Flow Cytometry

Cell Selection

α1-Acid Glycoprotein Angiopoietin-like 3 Cathepsin V Clusterin Dkk-1 EGF R/ErbB1 EG-VEGF/PK1 Fetuin A FGF-21 Galectin-3 Gas 6 GDF-15 IL-17A/F Heterodimer IL-19 Lipocalin-2/NGAL MBL Proprotein Convertase 9/PCSK9 Periostin/OSF-2 Progranulin ST2/IL-1 R4 Thrombomodulin/CD141 Tie-1 TIM-1/KIM-1

NEW Quantikine ELISA Kits

R&D Systems, Inc. www.RnDSystems.com

R&D Systems Europe, Ltd. www.RnDSystems.co.uk

R&D Systems China Co., Ltd. www.RnDSystemsChina.com.cn

For research use only. Not for use in diagnostic procedures.

For more information visit our website at www.RnDSystems.com/go/ELISA

A direct measure of product quality is the frequency of citations in the scienti� c literature. R&D Systems has more than 20 years of experience designing, testing, and optimizing the most cited ELISA kits in the world. Find out why scientists trust R&D Systems ELISAs more than any other brand.

The Most Referenced Immunoassays

1.4%1.6%2.1%2.5%

3.7%

7.9%

17.3%

41.8%R&D Systems

R&D Systems is the Most Referenced ELISA Manufacturer

R&D Systems Quantikine® ELISAs

Approximately 42% of Referenced Immunoassays are Developed and Manufactured by R&D Systems. A survey of 860 manuscripts from 44 journals was conducted to compare the number of citations specifying the use of R&D Systems ELISAs to the number citing ELISAs from other commercial sources. A total of 433 ELISA citations referencing immunoassays from 66 di� erent vendors were identi� ed in the survey.

R&D Systems Tools for Cell Biology Research™

55870_Cell Press.indd 1 6/25/10 10:49 AM

Leading Edge

In This Issue

Reprogramming’s Got the BeatPAGE 375

Reprogramming of fibroblasts to induced pluripotent stem cells suggested that a somatic cell could be reprogrammed intoalternative fates. Ieda et al. now report that a combination of three developmental transcription factors, Gata4, Mef2c, andTbx5, rapidly and efficiently reprograms postnatal cardiac or dermal fibroblasts directly into cardiomyocyte-like cells withoutpassing through a cardiac progenitor state. Thus, reprogramming of endogenous or explanted fibroblasts might providea source of cardiomyocytes for regenerative approaches.

Fyn-ally, the Missing Link between Tau and AbPAGE 387

The peptide amyloid-b (Ab) and the tau protein are both found in toxic aggre-gates in the brains of Alzheimer’s disease (AD) patients. In this issue, Ittneret al. reveal a mechanism by which tau may collaborate with Ab in mediatingAD pathogenesis. The authors show that tau, generally thought of as an axonalprotein, targets Fyn kinase to dendritic spines, where it phosphorylates NMDAreceptors. This leads to increased excitotoxicity, which boosts the toxic effectsof Ab on neurons. Disruption of tau-mediated targeting of Fyn in a mouse modelof AD decreases excitotoxicity via NMDA receptors and ameliorates other path-ological features associated with Ab.

p53 Gets Linc-ed InPAGE 409

Mammalian genomes encode numerous noncoding RNAs, including a class of large intergenic noncoding RNAs (lincRNAs)associated with p53. In this issue, Huarte et al. show that lincRNA-p21 is a direct p53 transcriptional target that also mediatesp53-dependent transcriptional repression through its physical association with hnRNP-K. These results reveal a newaspect to the p53 transcriptional response and suggest that lincRNAs may serve as key regulatory hubs in transcriptionalpathways.

Lagging Strand Takes the Lead in TranspositionPAGE 398

For most transposons, excision and insertion require double-strandedtemplates. In one recently identified family of bacterial insertion sequences(IS200/IS605), however, the cutting and pasting is done with only a single-stranded DNA segment. Ton-Hoang et al. now show that these single strandsare preferentially excised and inserted from the lagging strand of replicatingDNA, coupling transposition to replication fork passage. In addition to identifi-cation of a novel transposition pathway, the unique excision and insertion prop-erties of the IS200/IS605 family may make them useful tools for probing thein vivo structures of ssDNA segments.

A-PRC-iating Midzone DynamicsPAGE 420 and PAGE 433

During cell division, microtubules are arranged in the mitotic spindle to segre-gate the duplicated chromosomes. In anaphase, tightly bundled antiparallel microtubules in the spindle center form the mid-zone structure. Using an in vitro reconstitution approach, Bieling et al. show how PRC1, an antiparallel microtubule bundler,and the processive motor kinesin-4 form the minimal module needed to dynamically organize the core structure of the verte-brate midzone. Related work from Subramanian et al. provides detailed insight into how PRC1 functions. Using structural andbiophysical approaches, they find that PRC1 dimers contain both structured and unstructured domains that crosslink anti-parallel microtubules, which allow the protein to track along the overlap at the midzone without substantially resisting relativefilament sliding.

Cell 142, August 6, 2010 ª2010 Elsevier Inc. 335

We set out to bring you superior data from a fl ow cytometer.

Attune™

ACOUSTIC FOCUSING CYTOMETER

The new Applied Biosystems® Attune™ Acoustic Focusing Cytometer is unlike traditional fl ow cytometers. It uses high-frequency sound waves to precisely focus cells before they pass through the point of interrogation. Now get:• Ease of use—without sacrifi cing complete control of your run• Better clarity—by providing amazing separation of signal peaks• Higher sample rates—faster rare-event detection

Find out how acoustics is revolutionizing cytometry. Visit us at www.appliedbiosystems.com/attune.

For research, forensic or paternity use only. Not intended for any animal or human therapeutic or diagnostic use. The trademarks mentioned herein are the property of Life Technologies Corporation or their respective owners. © 2010 Life Technologies Corporation.

We succeeded…and changed traditional fl ow cytometry in the process.

Mike Ward, PhDAssociate Director Instrumentation & EngineeringFlow Cytometry Systems

Greg Kaduchak, PhDDirector Instrumentation & EngineeringFlow Cytometry Systems

Chromosomes Turn on a PhosphatePAGE 444

Proper orientation of chromosomes in the mitotic spindle ensures accuratesegregation and guards against aneuploidy, a common feature of humancancers. In this issue, Kim at al. demonstrate that CENP-E, a kinetochore motorprotein, is phosphorylated at a single site by Aurora kinases at the spindle polesto promote chromosome congression. CENP-E is subsequently dephosphory-lated by PP1 at the outer kinetochore, enabling bistable orientation. These find-ings explain how spatially regulated phosphorylation can control chromosomemovements during mitosis.

Protecting the Family JewelsPAGE 468

The mTORC1-signaling pathway is a critical regulator of cell growth. AberrantmTORC1 activation is associated with stem cell depletion through poorly char-acterized targets. Now, Hobbs et al. demonstrate that Plzf, a transcription factor implicated in germline maintenance,opposes mTORC1 pathway activity in spermatogonial progenitor cells (SPCs). Further, they show that elevated mTORC1activity inhibits response of SPCs to GDNF, a niche-derived growth factor required for self-renewal. This study providesimportant insight into mechanisms of germline maintenance and defines a model by which mTORC1 activity is detrimentalto stem cell function.

RNA Takes the Autobahn to the MitochondriaPAGE 456

RNA import into mammalian mitochondria is poorly understood comparedwith protein import. Here, Wang and colleagues report the identification ofa mammalian mitochondrial RNA import factor, the enzyme PNPASE. Theyfind a 20 nucleotide stem-loop RNA structure that can mediate PNPASE-dependent mitochondrial RNA import. PNPASE-dependent imported RNAsregulate processing of long mitochondrial RNA transcripts and the translationof electron transport chain proteins for respiration. The study identifies acomponent of the mammalian mitochondrial RNA import pathway and a poten-tial new approach for selectively targeting RNAs to mitochondria.

Mini-Myc Goes NonnuclearPAGE 480

Myc family proteins are nuclear transcriptional regulators that control cellgrowth, proliferation, and apoptosis. Here, Conacci-Sorrell et al. report the characterization of Myc-nick, a truncated formof Myc generated by calpain cleavage of full-length Myc. Myc-nick lacks nuclear localization and DNA binding regions andis predominantly cytoplasmic. Myc-nick interacts with microtubules and the GCN5 acetyltransferase to promote a-tubulinacetylation, changes in cell morphology, and terminal differentiation. Proteolytic cleavage may provide a functional switchfrom the nuclear, proproliferation form of Myc to a cytoplasmic, prodifferentiation form.


Leading Edge

Stem Cell Select

Lineage commitment requires stem cells to loosen their grip on pluripotency, sequentially picking andchoosing among competing genetic programs to realize specific cell fates. New findings described in thisissue’s Stem Cell Select address the impact of history and cellular memory on differentiation events andreveal critical molecular mediators of stem cell reprogramming and pluripotency.

A Remembrance of Tissues PastAccording to two recent reports (Kim et al., 2010; Polo et al., 2010),how you achieve stem cell pluripotency has unexpected conse-quences. Kim et al. show that reprogrammed pluripotent cells frommice created by somatic cell nuclear transfer differ from thosecreated through the expression of transcription factor cocktails.Their findings suggest that cells reprogrammed by somatic cellnuclear transfer bear greater similarity to embryonic stem cellsthan induced pluripotent stem (iPS) cells reprogrammed with definedfactors, that is, at least prior to extensive passaging. Delving into themolecular basis for this difference reveals that iPS cells retainresidual DNA methylation signatures reflecting their cell type oforigin. Similar observations are made by Polo et al. in their examina-tion of the gene expression patterns and epigenetic marks ofa collection of iPS cells derived from different mouse tissues. Thetwo reports show that iPS cells retain a form of epigenetic memory

that makes it possible to identify which tissue iPS cells come from and, more importantly, biases their returnto that tissue type upon differentiation. Kim et al. show that further interventions, such as treatment of iPS cellswith chromatin-modifying agents or repeating the cycle of differentiation and reprogramming, alter the impact ofthis epigenetic memory, whereas Polo et al. report that the epigenetic memory is dissipated by continuouspassaging of the cells in culture. These findings highlight the notion that pluripotency is not a singular conditionbut a diverse spectrum of states. They also suggest that a cell’s history is not easily erased, bringing to mind thewell-known quote from William Faulkner: ‘‘The past is never dead. It’s not even past.’’K. Kim et al. (2010). Nature. Published online July 19, 2010. 10.1038/nature09342.J. M. Polo et al. (2010). Nat. Biotechnol. Published online July 19, 2010. 10.1038/nbt1667.

DNA Repair Gives Germ Cells a Fresh StartDNA methylation is one of the cell’s most stable epigenetic marks.Hajkova et al. (2010) now provide evidence that base excision repair(BER) is actively engaged in the removal of these marks in mouseprimordial germ cells (PGCs). On their journey to totipotency,PGCs go through a dramatic transformation in their chromatin land-scape, with a major reorganization in their nuclear architecture andchanges in histone and DNA modifications. In cells undergoing thesereprogramming events, the authors observe an activation of BERpathways, coincident with the removal of methylated cytosines.The activation of BER is also temporally linked to the occurrenceof single-stranded DNA (ssDNA) breaks, a step in BER. Consistentwith a direct role of BER in DNA demethylation, inhibitors of BERinterfere with the removal of methyl-cytosine from the paternalpronucleus of the zygote, another setting in which active DNA demethylation is reported to occur. In relationto the recent papers by Kim et al. (2010) and Polo et al. (2010) discussed above, future work may assess whetherpromotion of this BER pathway could be used to augment existing methods for the generation of induced plurip-otent stem cells.P. Hajkova et al. (2010). Science 329, 78–81.

Osteogenic colonies from fibroblast-derived

induced pluripotent stem cells stained with alizarin

red to detect calcium deposits. Image courtesy of

K. Kim.

Chromatin-bound XRCC1 (green) in the male

pronucleus of the zygote (shown) and in

primordial germ cells suggests the presence of

single-stranded DNA breaks at the time of DNA

demethylation. Image courtesy of P. Hajkova.


A Chronicle of Differentiation ForetoldHow do events in embryonic stem cells set the stage for tissue-specific expression later in development? Liberet al. (2010) follow the molecular events at an enhancer specific for pre-B cell differentiation to show how it isprimed in embryonic stem cells (ESCs) for activation at a later stage. The core of the pathway uncovered involvesa handover between the transcription factor Sox2, which binds the l5-VpreB1 enhancer in ESCs, and Sox4,which binds the same region in pro-B cells. In ESCs, Sox2 promotes histone 3 lysine 4 di- and trimethylation,which are activating histone marks, and modulates the recruitment of the Foxd3, a factor that maintains theenhancer and the surrounding regions in a repressed state. At the pro-B cell stage, the Sox and Fox binding sitescooperate (the former with Sox4 bound) to fully activate transcription, thereby enhancing expression of l5,a protein that acts as a critical surrogate for the immunoglobulin light chain during B cell differentiation. Themodel proposed by the authors is that factors in ESCs establish active epigenetic marks that then cooperatewith tissue-specific factors to drive transcription during differentiation. Future work is likely to address whetherESC factors have a more general role in gene priming in this and other lineage commitment events.D. Liber et al. (2010). Cell Stem Cell 7, 114–126.

Holding Back a Natural Killer InstinctBecoming a T cell from a hematopoietic progenitor means avoidingthe temptations of other possibilities along the way, including B cell,macrophage, dendritic cell, and natural killer (NK) cell fates. Threerecent papers reveal a transcription factor needed for T cell precur-sors to take the final step of commitment (Ikawa et al., 2010; Li et al.,2010a; Li et al., 2010b). They show that in the absence of this factor,Bcl11b, would-be T cells can instead be redirected to a natural killercell fate. NK cells are lymphocytes that directly kill cells recognizedas non-self. Li et al. (2010a) demonstrate that Bcl11b is neededboth for T lineage commitment and to limit self-renewal, as itpromotes the downregulation of both NK cell genes and regulatorygenes that are characteristic of stem and progenitor cells. Future

work may explore which of the regulated genes are direct targets of Bcl11b. Potential functional consequencesof reprogramming T cell progenitors are explored by Li et al. (2010b), who show that even mature T cells can beconverted to NK cells by the loss of Bcl11b, and these reprogrammed cells share with native NK cells thecapacity to hinder tumor establishment in a mouse model of lung metastasis. The findings of Ikawa et al. mayserve as a starting point for efforts to identify the in vivo signals that direct this lineage decision or that maintainprogenitors in a multipotent state. They show in culture that an arrest in T cell development can be promoted byinterleukin-7 (IL-7), thereby maintaining their myeloid and natural killer cell potential, due to failure to induceBcl11b. A topic for future examination is to determine how signaling by IL-7 or other cytokines intersects withBcl11b at the decision point for T cell commitment.T. Ikawa et al. (2010). Science 329, 93–96.L. Li et al. (2010a). Science 329, 89–93.P. Li et al. (2010b). Science 329, 85–89.

Robert P. Kruger

T cell lineage commitment depends on the tran-

scription factor Bcl11b. Image courtesy of L. Li

and E.V. Rothenberg.


How does your institution measure up?Scopus is the optimal data source for research performance measurement. No other database has so much breadth of content covering so many authors.

“As financial resources become more scarce, it is more critical to identify research and researchers who are the most productive and on the right track.”Peter BrimblecombeProfessor, Atmospheric ChemistrySchool of Environmental SciencesUniversity of East Anglia, UK

With Scopus you can identify authors’ papers, tracktheir citations and analyze their influence using the Scopus h-index. And, to evaluate the performance of journals, research projects and groups of researchersyou can measure the performance of a specifiedcollection of articles.

Now it’s easy to:• Evaluate and prioritize resource allocation by

departments or fields• Make informed decisions about tenure and promotion • Promote your institution for funding and recruitment

www.scopus.com

015.

A1.0

101.

F.U

S ©

200

9 E

pp

end

orf A

G.

Bac

kgro

und

: ele

ctro

por

ated

L-m

ouse

cel

ls.

Electroporation of mammalian cells

Electroporation of bacteria

Electroporation of yeasts

Cell fusion

Outstanding transfection

rates!

www.eppendorf.com • Email: [email protected] • Application hotline: 516-515-2258

In the U.S.: Eppendorf North America, Inc. 800-645-3050 • In Canada: Eppendorf Canada Ltd. 800-263-8715

Our Multiporator® is a compact, flexible system designed for the electroporation of eukaryotic cells, bacteria and other microorganisms, as well as the electrofusion of cells.

Unique features include Soft Pulse microsecond pulses, patented* electronically regulated pulses, a gentle hypoosmolar buffer system and a choice of electroporation and electrofusion options.

A safe and easy method for performing cell fusion: Microfusion chamber for viewing and optimization

of fusion parameters Helix chamber for larger cell volumes Applications include: monoclonal antibodies, plant

protoplast fusion, tumor research, cancer vaccines, cloning mammals

For more information visit www.eppendorf.com

One electroporator–many applications.

*U.S. Pat. 6,008,038

Next-gen GWAS. NOW.

It’s a content revolution.Up to 50% more coverage of common and rare

variants than all other arrays. Maximum power for

any population.

The Omni family of microarrays can propel

your studies into true next-gen GWAS. With

a clearly defined product path to the future.

Immediate utility. Future flexibility.

Get on the path to next-gen GWAS.

Now is the time. Get started at

www.illumina.com/GWAS

Expand your stem cell library and save today on the latest books on stem cells

and regenerative medicine

Cell Stem Cell subscribers save 25% on their book orderSecure ordering online at elsevierdirect.comEnter promo code 28024 at check outPrices and publication dates subject to change without notice.

Stem Cells Scientific Facts and FictionChristine Mummery, Ian Wilmut, Anja Van de Stolpe and Bernard RoelenNovember 2010 | 400 pages | Paperback | $79.95 | €57.95 | £48.99 | ISBN: 9780123815354

Principles of Regenerative Medicine, 2nd EditionAnthony Atala and Robert LanzaNovember 2010 | 1400 pages | Hardback | $199.95 | €143.00 | £125.00 | ISBN: 9780123814227

Heart Development and Regeneration, 2-Volume SetNadia Rosenthal and Richard P. HarveyJune 2010 | 1072 pp. | Hardback | $199.95 | €143.00 | £125.00 | AU$296.00 | ISBN: 9780123813329

Essentials of Stem Cell Biology, 2nd EditionRobert Lanza, Roger Pedersen, John Gearhart, E. Donnall Thomas, Brigid Hogan, James Thomson, Douglas Melton and Sir Ian WilmutJune 2009 | 600 pp. | Hardback | $199.95 | €134.00 | £125.00 | AU$302.00 | ISBN: 9780123747297

Foundations of Regenerative Medicine Clinical and Therapeutic ApplicationsAnthony Atala, Robert Lanza, James Thomson and Robert NeremSeptember 2009 | 750 pp. | Hardback | $99.95 | €66.95 | £60.99|AU$148.00 | ISBN: 9780123750853

Stem Cell Anthology From Stem Cell Biology, Tissue Engineering, Regenerative Medicine, Cloning and Stem Cell MethodsBruce M. CarlsonOctober 2009 | 450 pp. | Hardback | $150.00 | €100.00 | £95.00 |AU$222.00 | ISBN: 9780123756824

Essential Stem Cell Methods A Volume in the Reliable Lab Solutions SeriesRobert Lanza and Irina KlimanskayaApril 2009 | 628 pp. | Paperback | $75.00 | €50.95 | £45.99 |AU$111.00 | ISBN: 9780123750617

Tissue EngineeringClemens van Blitterswijk, Peter Thomsen, Jeffrey Hubbell, Ranieri Cancedda, Anders Lindahl Sahlgrenska,Jerome Sohier and David F. WilliamsMarch 2008 | 760 pp. | Hardback | $115.00 | €76.95 | £69.99 |AU$170.00 | ISBN: 9780123708694

Human Stem Cell Manual A Laboratory GuideJeanne F. Loring, Robin L. Wesselschmidt and Philip H. SchwartzJune 2007 | 488 pp. | Spiral bound | $88.95 | €59.95 | £53.99 |AU$132.00 | ISBN: 9780123704658

Handbook of Stem Cells 2-Volume Set with CD-ROM Vol. 1–2Vol. 1 – Embryonic Stem CellsVol. 2 – Adult & Fetal Stem CellsRobert Lanza, Roger Pedersen, Helen Blau, E. Donnall Thomas, John Gearhart, James Thomson, Brigid Hogan, Catherine Verfaillie, Douglas Melton, Irving Weissman, Malcolm Moore and Michael WestSeptember 2004 | 1,760 pp. | Hardback | $566.00 | €380.00 | £345.00 | AU$817.00 | ISBN: 9780124366435

Leading Edge

Analysis

Cell 142, August 6, 2010 ©2010 Elsevier Inc. 347

The nations of the European Union (EU) spent €80 billion ($100 billion) last year on public nonmilitary research and develop-ment, yet European science still seems to have a quality gap compared with the US. For example, the EU produces 33% of research papers published annually world-wide but garners only 34% of citations, compared with the US, which publishes 29% of papers but earns 41% of citations (http://www.nsf.gov/statistics/seind08/). Policymakers believe that one reason for this quality shortfall is the fragmentation of research spending in Europe. Accord-ing to the European Commission based in Brussels, 85% of public research funds in Europe are distributed through sepa-rate national programs run by the EU’s 27 member states. Many think that the way to get more bang for their euro is to tie these national activities more closely together.

Multiple attempts to unite different European research programs have failed, however. The €7 billion that the European Commission allocates for research annually through its Frame-work Programme is supposed to nur-ture cross-border collaborations but does so one project at a time. And other efforts—including the long-established European Cooperation in Science and Technology (COST) scheme and plans in the Framework 6 Programme (which ran from 2002 to 2006) for “integrated projects” and “networks of excellence”—tried and failed to link the national research pro-grams together. All of these efforts have foundered on a mixture of bureaucracy, nationalism, inertia, and the reluctance of top researchers, who are able to get funding in their own countries, to get involved. “It is very difficult for member states to come together on a common basis,” concedes Enda Connolly, chief executive of Ireland’s Health Research Board, which distributes €40 million annually for biomedical research in Ire-land. “They are all locked into their own programs.”

In the last 2 years, however, a new fix has been proposed for the problem: “joint programming” between national research agencies. The idea is to get interested nations to band together and agree on a detailed strategy for a given research field and then pick-and-choose which elements of that strategy to collaborate on. “Joint programming is critical to the future, but it is still in gestation,” says Frank Gannon, former director of the European Molecular Biol-ogy Organization and current member of the European Research Area Board, which advises the European Commis-sion. “From my point of view, it is cur-rently the most crucial, single thing that we have to put right.”

Last December, the Council of Ministers representing the EU member states con-firmed that the first joint programming pilot

project would focus on neurodegenerative disease research, which is particularly weak and fragmented in Europe (Figure 1). The European Commission’s research directorate estimates that US spending in this area ($856 million, or €527 million, in 2007) is almost ten times that of Europe ($93 million, or €57 million).

Alzheimer’s disease researcher Bart De Strooper of KU Leuven in Belgium agrees that neurodegenerative dis-ease research is lagging in Europe. “My impression is that in the United States, much more of a vision has been devel-oped with regard to problems of aging, and Alzheimer’s in particular,” he says, pointing to collaborations such as the Alzheimer’s Disease Neuroimaging Ini-tiative (http://www.loni.ucla.edu/ADNI/), which is supported by several institutes of the National Institutes of Health (NIH)

Stitching Together Cross-border Research

European research has been hampered by fragmented national research programs. Is joint programming the answer? Colin Macilwain investigates.

Figure 1. Europe’s Research LandscapeBrain disease research in Europe, including the study of neurodegenerative diseases such as Alzheimer’s and Parkinson’s, is weakly coordinated across Europe and receives less investment compared with the US. (x axis, degree of coordination; y axis, spending in Europe relative to the US). Source: European Commission, 2008 (ec.europa.eu/research/press/2008/pdf/com_2008_468_en.pdf).

macilwain.indd 347 7/28/2010 2:08:52 PM

348 Cell 142, August 6, 2010 ©2010 Elsevier Inc.

and the private sector. “We in Europe look to the US and are happy if some of us are incorporated in their initiatives,” he says. “We should be much more active on the international scene.”

The idea of a combined approach to boost neurodegenerative disease research was first proposed in 2008, when France held the rotating presi-dency of the EU and President Nicolas Sarkozy sought to push aging issues higher up Europe’s research agenda. In July of that year, the European Com-mission issued a paper advocating joint programming as a generic approach to improving coordination among national research bodies. Meanwhile, medical research agencies in France, Germany, and the UK were already seeking ways to strengthen their respective involve-ment in an area of biology where the US has a pronounced lead (http://www.mrc.ac.uk/Utilities/Documentrecord/index.htm?d=MRC004898).

Philippe Amouyel, an epidemiologist at the University of Lille in France, led a working group established in 2008 to look into the idea of joint programming in the discipline and is now chair of the board of the EU Joint Programme for Neurodegenerative Disease. (Connolly also sits on the initiative’s five-person management board.) Initially supported by INSERM, the French biomedical research agency, the working group now has a small €2 million European Com-mission grant to cover its administration costs.

The Joint Programme has appointed a fifteen-member scientific advisory board, which met in Stockholm in April and will now draw up a 5 year strate-gic plan for neurodegenerative disease research in Europe. The board is made up of five social scientists, five clini-cians, and five biologists: Jesus Avila of the University of Madrid, Bart De Strooper, John Hardy of University Col-lege, London, Leszek Kaczmarek of the Polish Academy of Sciences and the chairman, Thomas Gasser of the Uni-versity of Tubingen. There is no budget yet to implement the plan, but officials involved in the discussions say that it is likely to involve an investment of about €200 million over 5 years, mainly from national funding bodies. “For the very first time,” says Amouyel, “we’ll have

a common view in Europe of what we need to do in neurodegenerative dis-ease research.”

Most of the research supported will be in basic neurobiology, including sequencing the complete genomes of patients to find risk genes, the devel-opment and standardization of disease biomarkers, and developing better ani-mal models of these diseases. “The big problem is that we don’t have a pipeline for new therapies, so we need to better understand the fundamental pathophys-iology of these diseases,” says Rob Buckle, program manager for neurosci-ences and mental health at the UK Medi-cal Research Council (MRC). “This is an opportunity to do things in a different way in Europe.” Buckle says he hopes that the Joint Programme will do work that could be relevant to the treatment of several neurological diseases, including Parkinson’s and motor neuron disease, as well as Alzheimer’s. “Mechanistically, there’s a lot of overlap between these disorders,” he says.

Gasser, a neurologist who studies Parkinson’s disease, says the scien-tific advisory board will confer broadly before publishing its research plan in summer 2011. Three meetings early next year, involving interested biologists, cli-nicians, and social scientists, respec-tively, will help the process along. Gas-ser adds that there is a “huge political will” to make the Joint Programme work but notes that its directed approach “will never displace bottom-up, undirected basic research that can give us com-pletely new insights.”

There are already some examples of the kinds of collaboration that might proceed under the Joint Programme. On June 29th, for example, the UK MRC, the German Centre for Neurodegenera-tive Diseases (DZNE), and the Canadian Institutes of Health Research (CIHR) announced a £3 million ($4.6 million) col-laboration on methods, technologies, and data sharing in neurodegenerative disease research. (Canada’s involvement reflects the MRC’s desire to cooperate with partners outside Europe who have relevant expertise.) And Amouyel says that some test projects may go ahead under the Joint Programme before the strategic plan’s completion, in areas such as genomics, the standardization of bio-

markers, developing new therapeutics, and infrastructure for large clinical trials. “They could get going by the end of the year,” he says, adding that the initiative is moving quickly. “To get something from concept to practical action in three years is really new in European research policy.”

De Strooper adds that as a researcher in Belgium, he sees advantages in col-laborating with larger countries—France, Germany, and the UK—that are starting initiatives in neurodegenerative disease research. He says that he hopes the Joint Programme will help researchers to do animal modeling and drug screening, obtain microRNA profiles of patients, and do deep sequencing and annota-tion of expression profiles for Alzheim-er’s and other diseases. Alzheimer’s researcher and geneticist John Hardy says he hopes that the program can help researchers to find better biomark-ers for neurodegenerative diseases and to use rapid, full-genome sequencing of patients to find the risk genes for them. But he admits that it is too early to know how it will unfold. “You get involved in the process, but you never know if there’s going to be a good outcome,” he says. “There is a genuine need for it. But you do worry that the optimism behind the programme will end up clashing with the hard reality of budget cuts.”

Although the joint programming pilot has been broadly welcomed, many experienced researchers and research administrators have questioned whether the joint programming approach has enough backing or momentum to have much impact on fragmented European research. Critics charge that the overall approach is ill-defined, inadequately pro-moted, and—most of all—underfinanced. Physicist and European Research Advi-sory Board member Jerzy Langer of the Polish Academy of Sciences says he fears that it won’t get much further than a long line of prior Commission efforts to get national research agencies to work more closely together. “I’m not saying it is wrong,” he says. “But any initiative that starts from the bureaucracy tends to go nowhere.” Langer contends that joint programming still has little constituency among researchers. “The key question,” he says, “is when, and in what capacity, real researchers get involved.”



“There is serious added value to be achieved if we can use joint program-ming to get national funding used more effectively,” says Ian Halliday, president of the European Science Foundation in Strasbourg. “But politically and organi-zationally, this is not going to be easy.” And observers of the joint programming pilot say that the MRC and INSERM are already frustrated by the need to accom-modate participation by the 24 EU nations who have signed up for the pilot—many of whom cannot contribute much in the way of cutting-edge neurobiology. Some suggest that participants will have to learn to do what physicists have done at CERN, the successful European particle physics center in Switzerland, and find ways of accommodating partners who are less scientifically advanced.

Buckle denies that this is a problem for the MRC, arguing that the Joint Pro-gramme for Neurodegenerative Disease will be an “umbrella” for the sharing of information and will allow “different con-stellations” of nations to work together on different areas of interest. “Everyone involved will be signed up to the same top-level objectives,” he says, “but everyone will have different interpreta-tions of how to best achieve them.”

Backers of joint programming say it can work because it is the EU member states, through the Council of Ministers, who have endorsed the approach and have pledged to see it through. “For the first time, member states are coming

together at the ministerial level to iden-tify jointly areas where public research can contribute to tackling Europe’s major societal problems,” said Maire Geoghegan-Quinn, the newly appointed European Commissioner for research, in a statement. “It is precisely because it is underpinned by a high-level, strate-gic and structured process—and most importantly of all, by real political will—that joint programming is a very big step forward.”

The Commission’s approach to the idea is finely nuanced, however, because if joint programming is seen as a Com-mission project, member states will view it as a means of getting Commis-sion money. So Commission officials are orchestrating it from behind the scenes, hoping that national governments will take the actions needed to provide money and drive it forward. But there are hints that large-scale Commission fund-ing to support joint programming could become available under the next phase of the Framework Programme, FP8, which starts in 2014.

Governments are certainly watching the pilot with interest. In April, the Com-mission recommended three further pilots—in food security, healthy diets, and cultural heritage conservation. In tough economic times, at least in theory, joint programming could help to reduce duplication enabling research funds to be spent more efficiently. “In a way, a shrinking budget is an opportunity,” says

Amouyel, “because you have to allocate resources more efficiently.” And in fields such as neurodegenerative disease, where even the largest national research agencies cannot cover every aspect, the joint programming approach may help.

Yet the odds remain stacked against it making much difference: EU member states control five-sixths of Europe’s research budget, and they want to spend it on their own scientists. A 2009 study by Eurohorcs, the federation of national research councils, found that 14 out of 32 of the national agencies surveyed had legal prohibitions on supporting work outside their borders (http://www.era.gv.at/attach/EUROHORCs.pdf). At the larger agencies, around 5% of the budget is usually devoted to collaborations with European partners (some small agen-cies, such as the Foundation for Polish Science and the Greek National Hellenic Research Foundation, spend proportion-ally more). The perceived political draw-backs of spending dwindling research funds “abroad” remain daunting, however. And there is a strong imperative within national research agencies to hold on to control of what they have. “There’s always going to be people at national agencies who fear that they could lose out” from joint programming, agrees Connolly. “But there’s very strong commitment at the political level, and in the European Com-mission, to making this happen. Member states are coming to see that they can’t do everything on their own.”

Colin MacilwainEdinburgh, UKDOI 10.1016/j.cell.2010.07.031


Leading Edge

Correspondence


In a recent Essay in Cell, McClellan and King argue that genomic resequencing rather than genome-wide association studies (GWAS) will be necessary to understand the genetic basis of common disease (McClellan and King, 2010). Like the authors, we too are excited about the potential for emerging sequencing tech-nologies to facilitate discoveries that explain the missing heritability of com-mon diseases. However, we disagree with the implication that GWAS have not been successful to date. Instead, we propose that insofar as the goal of these studies is to understand the etiology of heritable diseases, GWAS have provided numerous tantalizing clues for us biolo-gists to decipher. Rather than disprove the common disease/common variant hypothesis, we find that results from GWAS support the contention that com-mon polymorphisms do directly contrib-ute to disease risk, validating the linkage disequilibrium-based GWAS approach for helping to identify variants underlying disease. Although we do not dismiss the likelihood that rare variants also contrib-ute to common diseases, we expect that whole-genome sequencing approaches will show that the full spectrum of alleles, from rare to common, play important roles in disease etiology. Here, we argue that the existence of common disease-causing polymorphisms is not inconsis-tent with population genetic theory and that actual results from GWAS suggest that the reported associations represent real biology rather than false positives.

The contention that deleterious alleles that cause human diseases are common in the population may seem paradoxical, but several mechanisms can explain how such pathogenic alleles can overcome negative selective pressure. First, accu-mulating evidence demonstrates that there is balancing selection in which a certain allele confers susceptibility to one disease while simultaneously conferring protection from another. The best known example is heterozygosity for sickle cell anemia, which affords protection against malaria. GWAS have identified other

such instances, for example, the TCF2 (or HNF1B) gene where alternate alleles are risk factors for type 2 diabetes and prostate cancer (Gudmundsson et al., 2007). More generally, several loci where alleles have opposite effects on the risk of developing type 1 diabetes and Crohn’s disease have been reported (Wang et al., 2010), and it is likely that more examples of balancing selection are yet to be dis-covered.

The argument that common pathogenic variants must have withstood selective pressure throughout human history is predicated on the assumption that mod-ern humans developed in the same envi-ronment that we exist in today. However, due to the rapid acceleration of human development in the recent evolution-ary timeframe, numerous environmental changes have occurred that may impact the risk of complex common diseases. For instance, variants that were ben-eficial in the past may well have turned against their carriers as human lifestyles changed. The “thrifty gene hypothesis” suggests that variants that predispose to type 2 diabetes and obesity may have conferred a selective advantage in times of famine (Neel, 1962). However, in devel-oped countries, where food tends to be in overabundance, type 2 diabetes and obesity have become common diseases. Furthermore, existing neutral variation may manifest positive or negative effects as new environmental modifiers come into play. A set of single-nucleotide polymor-phisms (SNPs) associated with lung can-cer and located at a locus encoding the nicotinic acetylcholine receptor appears to have a stronger effect on lung cancer risk in smokers born long ago relative to those born more recently (Landi et al., 2009); this effect has been attributed to changes in the composition of cigarettes over time. This demonstrates that recent environmental changes can alter the disease-influencing effect of a common variant. These examples are likely to be only the tip of the iceberg of phenotypic effects modulated by gene-environment interactions.

The fact that most SNPs identified by GWAS do not lie in coding regions or other known regulatory elements is expected from the study design and is not evidence of false positives. The assumption under-lying the design of GWAS and choice of genotyped SNPs is that the true functional allele will be nearby and correlated with the initial SNP through linkage disequi-librium. When one considers linkage dis-equilibrium, there is an observed excess of GWAS hits that influence promoter regions or change the protein-coding sequence of a gene and a relative paucity of hits in intergenic regions (Hindorff et al., 2009). Moreover, many disease-associated SNPs identified by GWAS are located in genes or pathways previously known or suspected to play a role in disease etiology. Recent GWAS for Alzheimer’s disease, Crohn’s disease, type 1 diabetes, and type 2 dia-betes have rediscovered SNP associations previously reported from candidate gene studies. The demonstration that GWAS can identify common disease suscepti-bility variants provides a positive control (Hindorff et al., 2009). More generally, the functional pathways of GWAS-identified genes often make sense; for instance, numerous inflammatory genes have been implicated by GWAS in inflammatory bowel disease (Hindorff et al., 2009). We have found a similar concordance between the literature and GWAS hits in our own work. In a GWAS for age-related macular degen-eration, an SNP in complement factor H strongly associates with disease risk; this is consistent with previous suggestions that the complement pathway plays a role in disease etiology (Klein et al., 2005). Sim-ilarly, using GWAS, we identified a germ-line variant in the intron of the JAK2 gene that is associated with myeloproliferative neoplasms; JAK2 is known to harbor acti-vating oncogenic somatic mutations in this disease (Kilpivaara et al., 2009). One would not expect such correlations between GWAS findings, genic regions, and known disease biology if these findings were ran-domly distributed false positives due to population stratification or other causes. As these associations are likely to be real, the most logical and parsimonious expla-nation is that, in general, GWAS success-fully identify disease-associated variants and that variants found through GWAS tag regions important for the biology of these diseases.

Successes of Genome-wide Association Studies

correspondence.indd 350 7/28/2010 2:09:03 PM


In light of this, it is likely that GWAS hits found in intergenic regions far from known genes are true associations whose biol-ogy is not yet understood, rather than false positives. The human genome is incompletely annotated. Regions where GWAS associations have been found, but no known genes are located, could easily harbor unidentified new genes or regula-tory elements. For instance, the authors point to the colon and prostate cancer risk SNP rs693267 located 335 kb upstream from the MYC gene on chromosome 8q24 (McClellan and King, 2010). This locus has been shown to physically interact with MYC and is associated with enhanced Wnt signaling. Therefore, although the biol-ogy of this locus is not fully understood, it suggests a paradigm where intergenic disease-associated SNPs alter enhancer elements, either directly or through linkage disequilibrium, and therefore cause differ-ential regulation of disease-related genes.

This observation leads to a broader point: A lack of biological understanding of how these disease-associated vari-ants are pathogenic does not mean that there is no biology to discover. Although our understanding of the mechanisms by which disease risk loci contribute to pathogenesis currently lags behind the pace at which new loci are discovered, promising stories continue to emerge. To continue the previous example, although no definitive correlation between the rs6983267 genotype and MYC expression has been demonstrated, MYC is known to be tightly regulated and the right develop-mental time point may need to be exam-

ined to see such a correlation. Although further work is necessary to uncover the elusive mechanism by which the SNP confers risk, we propose that the existing evidence supports rather than refutes this SNP as a true cancer risk allele. Another example is a non-protein-coding region of chromosome 9q21 in which SNPs have been robustly associated with arte-rial disease. A recent paper reported that targeted deletion of an orthologous region in mouse interferes with cis-regulation of nearby genes (Cdkn2a/Cdkn2b) and may influence vascular cell proliferation (Visel et al., 2010). As a third example, an intronic type 2 diabetes risk SNP (rs7903146) was recently found to overlap with a region of islet cell-selective chromatin, and the two alleles of rs7903146 correlate with the open/closed chromatin state of the region (Gaulton et al., 2010). Thus, understanding the mechanisms by which GWAS loci con-tribute to disease will require considerable effort and time. We take this not as a sign that the common disease-common vari-ant model has failed but rather that a chal-lenge exists for the scientific community—a challenge that must be addressed with both traditional experimental genetics and innovative new approaches.

Robert J. Klein,1,* Xing Xu,1 Semanti Mukherjee,1 Jason Willis,1 and James Hayes1

1Program in Cancer Biology and Genetics, Memorial Sloan-Kettering Cancer Center, New York, NY 10065, USA*Correspondence: [email protected] 10.1016/j.cell.2010.07.026

RefeRenceS

Gaulton, K.J., Nammo, T., Pasquali, L., Simon, J.M., Giresi, P.G., Fogarty, M.P., Panhuis, T.M., Mieczkowski, P., Secchi, A., Bosco, D., et al. (2010). Nat. Genet. 42, 255–259.

Gudmundsson, J., Sulem, P., Steinthorsdottir, V., Bergthorsson, J.T., Thorleifsson, G., Manolescu, A., Rafnar, T., Gudbjartsson, D., Agnarsson, B.A., Baker, A., et al. (2007). Nat. Genet. 39, 977–983.

Hindorff, L.A., Sethupathy, P., Junkins, H.A., Ra-mos, E.M., Mehta, J.P., Collins, F.S., and Mano-lio, T.A. (2009). Proc. Natl. Acad. Sci. USA 106, 9362–9367.

Kilpivaara, O., Mukherjee, S., Schram, A.M., Wadleigh, M., Mullally, A., Ebert, B.L., Bass, A., Marubayashi, S., Heguy, A., Garcia-Manero, G., et al. (2009). Nat. Genet. 41, 455–459.

Klein, R.J., Zeiss, C., Chew, E.Y., Tsai, J.Y., Sackler, R.S., Haynes, C., Henning, A.K., San-giovanni, J.P., Mane, S.M., Mayne, S.T., et al. (2005). Science 308, 385–389.

Landi, M.T., Chatterjee, N., Yu, K., Goldin, L.R., Goldstein, A.M., Rotunno, M., Mirabello, L., Ja-cobs, K., Wheeler, W., Yeager, M., et al. (2009). Am. J. Hum. Genet. 85, 679–691.

McClellan, J., and King, M.C. (2010). Cell 141, 210–217.

Neel, J.V. (1962). Am. J. Hum. Genet. 14, 353–362.

Visel, A., Zhu, Y., May, D., Afzal, V., Gong, E., Attanasio, C., Blow, M.J., Cohen, J.C., Rubin, E.M., and Pennacchio, L.A. (2010). Nature 464, 409–412.

Wang, K., Baldassano, R., Zhang, H., Qu, H.Q., Imielinski, M., Kugathasan, S., Annese, V., Du-binsky, M., Rotter, J.I., Russell, R.K., et al. (2010). Hum. Mol. Genet. 19, 2059–2067.

In a recent Essay published in Cell, McClellan and King discussed genetic heterogeneity and the potential role of rare genetic variants in complex human diseases (McClellan and King, 2010). These important issues, in particu-lar the application of high-throughput

sequencing techniques to discover disease genes, are highly relevant to genetics researchers. However, the authors allocated a substantial propor-tion of their efforts to being critical of the utility of genome-wide association studies (GWAS). These particular sec-

tions of the Essay may lead to misin-terpretation of published studies by us and others. For the broad readership of Cell and for the scientific community in general, we highlight our concerns in this Correspondence.

The authors refer to the fact that most single-nucleotide polymorphisms (SNPs) detected in GWAS reside in intergenic regions and consequently challenge the utility and reliability of GWAS with the question: “How did genome-wide asso-ciation studies come to be populated by

Strategies for Genetic Studies of complex Diseases



risk variants with no known function?” When addressed in the proper context, however, it is well established that GWAS do not attempt to identify functional SNPs but rather “tag” the approximate location of disease variants, typically down to 100 kb or less. This is made possible due to the linkage disequilib-rium (LD) patterns characterized by the International HapMap project, that is, the correlation of genotypes between the yet-to-be-determined underlying disease variant and neighboring SNPs. Indeed, the vast majority of SNPs used in GWAS are of unknown biological function, due to the fact that most SNPs reside outside of coding regions and that the manufac-turers of the SNP arrays selected SNPs from the HapMap to facilitate efficient tagging across the genome, that is, the priority was information capture rather than putative function. Furthermore, noncoding SNPs identified by GWAS may reveal intergenic regulatory ele-ments that are critical to understand, and it is now up to the genetics com-munity to develop approaches to interro-gate the function of regulatory variants. The authors did refer to LD as a potential explanation for noncoding variants yield-ing association in GWAS, but they failed to recognize that the design of GWAS is not to directly interrogate causal variants in the first instance. For example, the two GWAS that we have conducted for sickle cell anemia and hearing loss (Dickson et al., 2010) yielded top hits in intergenic regions, but these are in close proximity to the causal genes (HBB and GJB2) that were already well-established before the GWAS era. These studies demonstrate the reliability of GWAS for identifying the approximate locations of disease genes by noncoding SNPs. In addition, they represent two vivid examples of how GWAS can work by leveraging LD.

McClellan and King have also attrib-uted many published GWAS hits to population stratification. In the absence of scientific support or statistical deriva-tion, they claim that “an odds ratio of 3.0, or even of 2.0 depending on population allele frequencies” would be robustly interrogated by GWAS. The vast major-ity of published GWAS loci therefore fall below the threshold for “popula-tion stratification.” However, compared to candidate gene association studies,

the beauty of whole-genome SNP data is that inflation of test statistics due to population substructure can be read-ily identified and adjusted. Populations do not differ by just one or two SNPs; instead, they differ at many loci such that whole-genome data aid in identifying stratification, including extremely fine-scale subpopulations among Europeans (Novembre et al., 2008). The GWAS com-munity has established methods to deal with population stratification, and these methods effectively adjust for common variants without controversy. There are certainly challenges with the analyses of rare variants, hypervariable variants, or interrogation of recently admixed popu-lations, which are all active topics of cur-rent research. There are now standard practices to handle population stratifica-tion in GWAS, such as genomic control, EigenStrat, and multidimensional scal-ing. Furthermore, family-based study designs have the advantage of pro-tecting against stratification. Thus, the GWAS community has developed ways to address population stratification, and odds ratios of 1.5, 1.2, or even 1.1 have proven to be bona fide signals rather than artifacts of population stratification.

The Essay’s authors go on to claim that the autism locus reported by our group (Wang et al., 2009) is a “false posi-tive” due to population stratification, but the study was driven by family-based cohorts both at the discovery and repli-cation stages, with case-control cohorts further supporting the finding. Here, McClellan and King ignore the fact that family-based analysis is robust against stratification and have mistakenly con-sidered GWAS hits as “false positives” if allele frequencies varied across Euro-pean or HapMap populations. For exam-ple, they claim that our reported autism-associated SNP, rs4307059 (Wang et al., 2009), was a “particularly dramatic example of the perils of cryptic popula-tion stratification.” Their reasoning is that the frequency of the proposed risk variant varies from 0.21 to 0.77 across European populations and that it is monomorphic in African populations. However, they used data from the Human Genome Diver-sity Project that examined 938 samples from 51 worldwide populations (Coop et al., 2009), including 7 samples from Tuscany (allele frequency = 0.77) and 15

samples from Orkney (allele frequency = 0.21). When dealing with whole-genome data, small sample sizes often lead to biased estimates of allele frequencies, which we have now proven to be the case in this instance. We queried the 101 Tuscan samples included in the HapMap 3 project (http://www.sanger.ac.uk/humgen/hapmap3) and found that the allele frequency is 0.41, which is very similar to what we reported for our European American cohort (0.39 in Wang et al., 2009) and substantially different from 0.77 quoted by McClel-lan and King. Furthermore, even if the allele frequency measures are accurate, McClellan and King should not claim this SNP as “hypervariable” without appro-priate control experiments to compare it with other SNPs. They may not appre-ciate that the vast majority of SNPs do have variable allele frequencies between populations; in fact, 44% of SNPs on the Illumina array (http://hgdp.uchicago.edu/Browser_tracks/FST) have more extreme population divergence (measured by Fst values) than rs4307059. Similarly, another SNP, which has been replicated for its association with type 1 diabetes (Hakonarson et al., 2007) in over 20 inde-pendent studies, is ranked in the 29th percentile for Fst value. As such, it is our interpretation that the authors took a ran-dom SNP from the middle of a ranked list and inappropriately claimed it as a “par-ticularly dramatic” example of “popula-tion stratification.” Extending this further, we also wish to point out that replication of GWAS signals depends on many fac-tors, including the power of the study, the sample ascertainment scheme or diagnostic criteria, and the heterogene-ity of the study population (for example, sporadic versus familial cases). For neu-ropsychiatric diseases, it is well known that replication can be difficult with small sample sizes, as is the case for the study specifically referred to by the authors (Weiss et al., 2009), which interrogated 258 families with autism for replica-tion that were not part of AGRE (Autism Genetic Resource Exchange). Never-theless, the autism locus on 5p14.1 has been replicated in a family-based GWAS without case-control comparisons (Ma et al., 2009) and has been associated with communication-spectrum phenotypes through quantitative trait associations



in 7313 children from a UK birth cohort, eliminating concerns of population strati-fication (St Pourcain et al., 2010).

Like McClellan and King, we expect that with the development of high-throughput sequencing technologies, whole-genome sequencing will be an invaluable tool for the study of rare genetic variants contributing to com-plex diseases. However, the apparent importance of rare variants does not discount the contribution of common variants; the concerns about population stratification should not be overstated and certainly should not be presented as an argument to discredit many pub-lished GWAS signals with an odds ratio of less than 2. These sections within an otherwise excellent Essay must be countered so that Cell readers have a more balanced view of the current state of the field.

Kai Wang,1 Maja Bucan,2 Struan F.A. Grant,1,3 Gerard Schellenberg,4 and Hakon Hakonarson1,3,*1Center for Applied Genomics, Children’s Hospital of Philadelphia2Department of Genetics3Department of Pediatrics4Department of Pathology and Laboratory MedicineUniversity of Pennsylvania, Philadelphia, PA 19104, USA*Correspondence: [email protected] 10.1016/j.cell.2010.07.025

RefeRenceS

Coop, G., Pickrell, J.K., Novembre, J., Kudaravalli, S., Li, J., Absher, D., Myers, R.M., Cavalli-Sforza, L.L., Feldman, M.W., and Pritchard, J.K. (2009). PLoS Genet. 5, e1000500.

Dickson, S.P., Wang, K., Krantz, I., Hakonarson, H., and Goldstein, D.B. (2010). PLoS Biol. 8, e1000294.

Hakonarson, H., Grant, S.F., Bradfield, J.P., March-and, L., Kim, C.E., Glessner, J.T., Grabs, R., Casa-lunovo, T., Taback, S.P., Frackelton, E.C., et al.

(2007). Nature 448, 591–594.

Ma, D., Salyakina, D., Jaworski, J.M., Konidari, I., Whitehead, P.L., Andersen, A.N., Hoffman, J.D., Slifer, S.H., Hedges, D.J., Cukier, H.N., et al. (2009). Ann. Hum. Genet. 73, 263–273.


Novembre, J., Johnson, T., Bryc, K., Kutalik, Z., Boyko, A.R., Auton, A., Indap, A., King, K.S., Berg-mann, S., Nelson, M.R., et al. (2008). Nature 456, 98–101.

St Pourcain, B., Wang, K., Glessner, J.T., Golding, J., Steer, C., Ring, S.M., Skuse, D.H., Grant, S.F., Hako-narson, H., and Smith, G.D. (2010). Am. J. Psychia-try. Published online July 15, 2010. 10.1176/appi.ajp.2010.09121789.

Wang, K., Zhang, H., Ma, D., Bucan, M., Glessner, J.T., Abrahams, B.S., Salyakina, D., Imielinski, M., Bradfield, J.P., Sleiman, P.M., et al. (2009). Nature 459, 528–533.

Weiss, L.A., Arking, D.E., Gene Discovery Proj-ect of Johns Hopkins and the Autism Consortium, Daly, M.J., and Chakravarti, A. (2009). Nature 461, 802–808.

In response to our Essay “Genetic Het-erogeneity of Human Disease” (McClellan and King, 2010), Wang et al. and Klein et al. challenge our critique of GWAS findings. Both Correspondence suggest that our conclusions lack an understanding of the principles of the common variant-common disease model and its application in GWAS methodology. We respectfully disagree, and in fact our Essay directly addressed many of their points. We are happy to clar-ify further. The issue is not ignorance of GWAS methodology; the issue is reconcil-ing GWAS findings with population genet-ics, evolution, and biology.

There is no dispute that common genetic variants influence human traits. Alleles with the strongest influences on human traits are associated with benign phenotypes, such as hair color and eye color, that vary across individuals. Alleles for these traits lie in coding and known regulatory regions (Hindorff et al., 2009). Such variants have been influ-enced by selection in human evolution.

Thus, the best documented common variants with a substantial impact on disease risk typically are associated with illnesses presenting later in life, includ-ing Alzheimer’s disease (APOE), exfoliat-ing glaucoma (LOXL1), and age-related macular degeneration (CFH). These alleles persist in the population because the associated illnesses do not nega-tively influence reproductive fitness.

However, the existence of common alleles that truly influence disease does not imply that all GWAS findings are valid. To date, published GWAS have reported more than 500 single-nucle-otide polymorphisms (SNPs) associated with various diseases and traits, 88% of which lie in introns or intergenic regions, and for which the median odds ratio for effect size is 1.33 (Hindorff et al., 2009). Few of these associations have biologi-cal support.

In order to identify alleles that influence disease using GWAS, two fundamen-tal criteria must be met: (1) a common

variant influencing the trait in question must truly exist in the population being studied; and (2) the genotyped markers used in association analyses must either include the causal variant or be in link-age disequilibrium (LD) with the causal variant in the population being stud-ied. Common “risk SNPs” detected by GWAS could in theory be in LD with rare disease-causing alleles, if by chance rare causal alleles are disproportionately linked with a common variant (Dickson et al., 2010).

In addressing our critique, both Wang et al. and Klein et al. acknowl-edge that GWAS methodologies are designed to detect SNPs primarily found in intergenic or intronic regions, given the construction of standard SNP arrays. They argue that the risk SNPs implicated by GWAS are not expected to be causal but instead are in LD with true causal variants. Wang et al. main-tain that standard GWAS methodolo-gies adequately control for population stratification. Further, they note that a large proportion of HapMap SNPs vary widely across populations, and therefore it is not surprising that highly variable SNPs emerge as risk vari-

Response

Why It Is Time to Sequence



ants. These issues are at the crux of the debate and will be the focus of our response.

Genetic heterogeneity across human populations limits GWAS methodology and may result in the identification of “risk alleles” that are neither likely to be func-tional nor likely to be in LD with alleles that are. The genetic architecture of human disease is shaped by the same evolu-tionary forces that impact the human genome. Genetic architecture is dic-tated by four factors: mutation, selection, migration, and drift. Negative selection globally reduces variation at missense mutations, particularly in genes associ-ated with disease (Barreiro et al., 2008). Frequencies of missense alleles (i.e., nonsynonymous SNPs) are less diverse across populations than are frequencies of synonymous SNPs, SNPs in untrans-lated regions (UTRs), and intronic SNPs because selection reduces interpopula-tion variation for alleles influencing phe-notypic expression (Garte, 2010). Popula-tion differences in genic regions, primarily at nonsynonymous SNPs and 5`UTR vari-ants, may persist as the result of selec-tion for adaptive responses or by chance (Barreiro et al., 2008). For alleles influ-encing phenotypes, selection constrains overall variation in allele frequency while also influencing specific patterns of varia-tion defined by geographic ancestry.

Many SNPs, inversions/deletions (indels), and short tandem repeats vary widely in allele frequency among populations. This is especially true for variants that are not in coding or regu-latory regions. Such alleles vary with population clusters in patterns more consistent with neutral drift and migra-tion rather than with selection (Coop et al., 2009). The colonization of the world by modern humans was carried out by a series of founder populations with subsequent rapid expansion of popula-tion size. Neutral alleles emerging at the forefront of these expansions “surfed” waves of population growth. Variations in allele frequencies across populations stem from differences in the timing of the variant’s emergence in the expansion.

Intergenic or intronic SNPs that vary widely in frequency among populations are most likely neutral alleles reflect-ing the history of human migration. Wide variations in allele frequency also

characterize many “risk alleles” identi-fied by GWAS, including, for example, rs4307059 reported to be associated with autism (Wang et al., 2009), which is a major focus of the Correspondence by Wang et al. We suggest that associations based on such highly variable SNPs are often artifacts of cryptic population strat-ification. Wang et al. argue that standard GWAS strategies have been adopted to control for population stratification. How-ever, these methods control by person, not by SNP. Because populations from large geographic areas (e.g., Europe) are genetically heterogeneous, outlier SNPs that vary widely among subgroups of such populations are not excluded by these methods and often drive positive associations.

Wang et al. also assert that the asso-ciation of rs4307059 and autism was driven by family-based cohorts, which are robust to stratification. However, no SNPs reached genome-wide signifi-cance in their study using a family-based design (Wang et al., 2009). Significance was obtained by pooling subjects from both family-based and case-control samples, including thousands of unre-lated cases and controls. Analyses only comparing affected to unaffected persons within families will be robust to stratification; but comparing a mixed series of related and unrelated cases to controls can exacerbate stratification due to the clustering of shared ances-tries in the affected group.

In arguing for rs4037059 as a risk vari-ant for autism, Wang et al. dismiss the negative results of a replication study (Weiss et al., 2009) due to small sample size. Yet, they cite another study (Ma et al., 2009) as supporting an association with this genomic region (5p14.1), even though no SNPs in this study met genome-wide significance and rs4037059 was not even nominally significant in the discovery sample. Variable weak associations of different SNPs across the same genomic region in different cohorts do not consti-tute replication.

Both Klein et al. and Wang et al. suggest that the vast majority of GWAS risk alleles are in LD with causal mutations, and that intergenic and intronic risk variants rep-resent regulatory elements. In principle, either or both of these hypotheses could be true. However, thus far, virtually no

such mutations or elements have been found by following up on GWAS findings. Wide variations in allele frequency across populations argue against shared hap-lotype structure, which is necessary to detect causal variants in LD.

Common variants influencing traits or disease must withstand selection in every generation in order to be maintained at polymorphic frequencies worldwide. Regulatory elements are impacted by the same evolutionary forces as coding regions. Klein et al. discuss alternative methods whereby otherwise deleterious alleles are maintained in the population, citing both well-known and speculative examples of balancing selection and gene-environment interactions. How-ever, as is the case for much of this dis-cussion, a confirmed biological mecha-nism for one illness does not mean that all GWAS findings can be attributed to similar mechanisms.

Klein et al. suggest that a common risk allele may have been under neu-tral or positive selection in early human history, thus explaining its worldwide prevalence and maintenance in the population despite the association with disease. Yet evolution is an ongoing process. If a common variant decreases reproductive fitness during modern his-tory, coinciding with the vast expansion of the human population, the frequency of the variant will decline. Klein et al. also suggest that many common risk variants, either the identified SNP or a causal variant in LD, operate by vir-tue of some yet-to-be-determined bal-ancing selection or gene-environment interaction. This is obviously possible but needs to be demonstrated for any given variant. In the presence of a true balancing selection or gene-environ-ment interaction, variation in allele fre-quencies will correlate with geographic variation in disease prevalence and is likely to offer clues about specific envi-ronmental exposures or risk factors. The best characterized examples of balancing selection are found in spe-cific geographic populations sharing an environmental exposure, such as an infectious disease. In these examples, e.g., sickle cell disease, risk alleles are maintained at a much higher rate in the exposed population than in populations from other parts of the world.



It is a leap of theoretical faith to infer that SNPs with no known function and variable weak associations across different disease cohorts represent prima facie evidence of regulatory function, gene-environment interaction, or balanced selection. We are not suggesting that such phenomena do not occur. Deciphering such mechanisms will be a major scientific focus over the next decade. However, for the vast majority of SNPs identified by GWAS, these mecha-nisms are speculative in the absence of biological evidence.

Many of the examples cited by Klein et al. and Wang et al. in support of GWAS suffer from the same limitations that we highlighted in our Essay (McClellan and King, 2010). For example, as Klein et al. describe, there is robust epidemiologi-cal data in support of individuals with diabetes having a lower risk for prostate cancer. However, associations between putative risk/protective SNPs in HNF1B and JAZF1 and the two illnesses vary across cohorts, and these variants do not mediate the relationship between the two diseases (Stevens et al., 2010).

We understand that many believe that most GWAS findings are valid. Ultimately, the debate hinges on how definitive one views variable results with statistically

highly significant p values and very small effects that diminish with further study. Currently, GWAS results fail to explain the vast majority of genetic influence on any human illness. Further, most risk variants implicated by GWAS have no demonstrated biological, functional, or clinical relevance for disease.

We suggest that biological relevance needs to be established before assert-ing that positive correlations from GWAS are equivalent to disease risk. Such evi-dence must address the functional sig-nificance of the SNP, or a variant in LD with the SNP, rather than arguing for the appeal of a nearby gene. In their argu-ments, Wang et al. and Klein et al. simply restate GWAS principles, which serves to reify, not prove, the model. We make a plea for more rigorous analysis. Science ultimately advances by evidence.

Jon McClellan1 and Mary Claire King2,*1Department of Psychiatry2Department of Medicine and Department of Genome SciencesUniversity of Washington, Seattle, WA 98195, USA*Correspondence: [email protected] 10.1016/j.cell.2010.07.027

RefeRenceS

Barreiro, L.B., Laval, G., Quach, H., Patin, E., and Quintana-Murci, L. (2008). Nat. Genet. 40, 340–345.

Coop, G., Pickrell, J.K., Novembre, J., Kudaravalli, S., Li, J., Absher, D., Myers, R.M., Cavalli-Sforza, L.L., Feldman, M.W., and Pritchard, J.K. (2009). PLoS Genet. 5, e1000500.

Dickson, S.P., Wang, K., Krantz, I., Hakonar-son, H., and Goldstein, D.B. (2010). PLoS Biol. 8, e1000294.

Garte, S. (2010). Am. J. Hum. Biol. 22, 297–300.

Hindorff, L.A., Sethupathy, P., Junkins, H.A., Ra-mos, E.M., Mehta, J.P., Collins, F.S., and Mano-lio, T.A. (2009). Proc. Natl. Acad. Sci. USA 106, 9362–9367.

Ma, D., Salyakina, D., Jaworski, J.M., Konidari, I., Whitehead, P.L., Andersen, A.N., Hoffman, J.D., Slifer, S.H., Hedges, D.J., Cukier, H.N., et al. (2009). Ann. Hum. Genet. 73, 263–273.


Stevens, V.L., Ahn, J., Sun, J., Jacobs, E.J., Moore, S.C., Patel, A.V., Berndt, S.I., Albanes, D., and Hayes, R.B. (2010). Prostate 70, 601–607.

Wang, K., Zhang, H., Ma, D., Bucan, M., Glessner, J.T., Abrahams, B.S., Salyakina, D., Imielinski, M., Bradfield, J.P., Sleiman, P.M., et al. (2009). Nature 459, 528–533.

Weiss, L.A., Arking, D.E., Gene Discovery Project of Johns Hopkins and the Autism Consortium, Daly, M.J., Chakravarti, A. (2009). Nature 461, 802–808.


Leading Edge

Previews


Alzheimer’s disease (AD) is a devastat-ing public health problem for our aging societies. Although it is well established that amyloid β-peptide (Aβ) forms toxic oligomers in the brain (Haass and Sel-koe, 2007), it is not clear how Aβ initiates the amyloid cascade and causes the death of neurons. Tau, an axonal protein, seems to be an executor of Aβ toxicity even though it is localized to axons, and Aβ toxicity is primarily triggered through interactions of Aβ oligomers with den-dritic spines (Haass and Selkoe, 2007). Tau binds to microtubules, a process that is prevented by its abnormal phosphory-lation during AD pathogenesis. Loss of microtubule binding by tau is thought to cause the disassembly of microtubules followed by the aggregation and depo-sition of tau in pathogenic neurofibril-lary tangles. Although amyloid plaques and tau tangles are prominent markers of AD, one of the first and most obvious pathological abnormalities observed in brain tissue from AD patients is the relocalization of tau from axons to the somatodendritic compartment of neu-rons (Figure 1) (Ballatore et al., 2007). During brain development, tau and other microtubule-associated proteins are ini-tially distributed ubiquitously throughout neurons, but then, as differentiation pro-gresses, tau becomes sorted into axons (Figure 1). In AD and other neurodegen-erative diseases involving tau (termed “tauopathies”), this neat sorting pattern breaks down (Ballatore et al., 2007), perhaps because abnormal phospho-rylation of tau enables it to detach from microtubules and diffuse rapidly into other neuronal compartments (Konzack

et al., 2007). In this issue of Cell, Ittner et al. (2010) now shed light on an unex-pected dendritic function for normal tau that suggests how tau may mediate Aβ toxicity.

Most research on tau function has focused on whether it is important for microtubule stabilization, neurite out-growth, or the formation of tracks for cargo transport along axons. The cata-strophic redistribution of tau to the soma and dendrites of neurons in AD sug-gests that an efficient sorting mechanism normally keeps tau localized to axons. However, such “polarized” sorting like other cellular sorting pathways is never completely efficient, thus enabling small amounts of tau to become localized to the somatodendritic compartment even under physiological conditions. This led Ittner and colleagues to propose a microtubule-independent physiological function for tau that regulates signaling in dendritic spines. Their research was ini-tially triggered by accumulating evidence that subacute seizures occur not only in transgenic mouse models of AD, but also in AD patients (Palop and Mucke, 2009). Interestingly, tau deficiency decreases seizures induced by the Aβ-mediated overstimulation of excitatory N-methyl-D-aspartate (NMDA) receptors and improves survival in a transgenic AD mouse model (Roberson et al., 2007).

It is well established that tau binds not only to microtubules but also to several nonreceptor tyrosine kinases including Fyn through its N-terminal domain (Lee et al., 1998). This enables tau to seques-ter Fyn and to alter its localization in the neuron. Fyn phosphorylates subunit 2 of

the NMDA receptor, resulting in stabili-zation of this receptor’s interaction with PSD95, a scaffolding protein in the den-dritic spines of neurons. This stabiliza-tion, in turn, strengthens signaling by the excitotoxic neurotransmitter glutamate, which enhances Aβ toxicity (Figure 1). A tau-dependent increase in Fyn in den-dritic spines could boost excitotoxic sig-naling, and, conversely, sequestration of Fyn by tau or downregulation of tau expression could mitigate excitotoxicity.

To address whether this is the case, Itt-ner and coworkers first generated trans-genic mice that overexpressed a variant of tau (∆tau) that lacked the C terminus and thus could bind to Fyn but not to micro-tubules. Interestingly, ∆tau was completely excluded from dendrites and accumulated within the soma (Figure 1). Moreover, ∆tau efficiently competed with endogenous tau for binding to Fyn, resulting in sequestra-tion of Fyn in the soma. Similarly, loss of tau also prevented postsynaptic targeting of Fyn, and loss of Fyn in the dendrites was enhanced when ∆tau was expressed in mice deficient in normal tau. This sug-gests that the targeting of Fyn to dendrites depends on normal tau, a surprising find-ing for a protein believed to act only in axons. But how tau-based sorting of Fyn occurs and whether other proteins are required remain unclear.

Fyn phosphorylates NMDA receptor subunit 2 at tyrosine 1472 and stabilizes the interaction of this subunit with the postsynaptic protein PSD95 (Nakazawa et al., 2001). However, when Fyn is trapped within the soma by expression of ∆tau or reduced expression of endogenous tau, there is decreased phosphorylation of

Fyn-Tau-Amyloid: A Toxic TriadChristian Haass1,2,* and Eckhard Mandelkow3,4,*

1German Center for Neurodegenerative Diseases (DZNE), 80336 Munich, Germany2Adolf Butenandt Institute, Ludwig Maximilian University, 80336 Munich, Germany3Max Planck Unit for Structural Molecular Biology, 22607 Hamburg, Germany4German Center for Neurodegenerative Diseases (DZNE), 23175 Bonn, Germany*Correspondence: [email protected] (C.H.), [email protected] (E.M.)DOI 10.1016/j.cell.2010.07.032

The axonal protein tau and amyloid β-peptide (Aβ) are key players in the pathogenesis of Alzheim-er's disease, and tau mediates Aβ toxicity, but it is not clear how. Ittner et al. (2010) now report an unexpected physiological function for tau in neuronal dendrites that may explain how tau mediates Aβ toxicity.

previews.indd 356 7/28/2010 2:09:13 PM


the NMDA receptor, the interaction of the receptor with PSD95 is destabilized, and excitotoxic signaling decreases. Excito-toxicity has been implicated in Aß-medi-ated toxicity, and a reduction in tau ame-liorates the pathological action of Aβ in the AD transgenic mouse model (Rober-son et al., 2007). Similarly, overexpression of ∆tau in the AD mice prevented seizures and improved survival; indeed, these ben-efits were enhanced when expression of endogenous tau was reduced. Moreover, the AD mice overexpressing ∆tau showed improvements in memory, suggesting that Aβ toxicity was reduced when Fyn levels decreased in the dendritic spines. This was independent of Aβ production or the overall amyloid plaque load.

Notably, there is an interesting twist to the Fyn-tau-microtubule story. In oli-godendrocytes, tau is necessary for the outgrowth of cell processes and for transport of Fyn, which is important for myelinating neurons. Expression of the N-terminal domain of tau alone causes abnormal sorting of Fyn, resulting in poor myelination of neurons and seizures in rodents (Klein et al., 2002). Thus, there are two completely distinct settings for this potentially toxic triad, which implicates abnormal tau in seizure disorders as well as in neurodegenerative diseases.

But can the Fyn-tau connection in den-dritic spines be exploited to develop new therapeutic strategies for treating AD? A peptide that blocks phosphorylation of NMDA receptors by Fyn protects neu-rons from excitotoxic damage (Aarts et al., 2002). Strikingly, when Ittner and col-leagues treated their transgenic AD mice with this peptide, memory deficits were ameliorated and there was improved survival, similar to the results when Fyn was sequestered by ∆tau.

The Ittner et al. study may provide a missing link between extracellular deposits of Aβ and intracellular tau and pinpoints tau and Fyn as possible medi-ators of Aβ toxicity. Although the idea is tantalizing, the actual colocalization of tau and Fyn within dendritic spines and their modes of action remain to be shown under physiological conditions. A major surprise of this study is that a normal physiological function of tau, regarded as an axonal protein, mediates Aβ tox-icity at dendritic spines. Is this physio-logical function of tau affected during the

earliest stages of AD? This may be the case, as redistribution of full-length tau (and possibly tau fragments) to the som-atodendritic compartment occurs well before the formation of neurofibrillary tangles. The abnormal sorting of tau to the somatodendritic compartment may be due to changes in signaling cascades that alter kinase and phosphatase activi-ties in those dendritic spines affected by Aβ, resulting in local changes in tau sort-ing and cytoskeletal rearrangements. But do tau and Fyn still interact after tau redistribution, given that the Fyn bind-ing site on tau can be phosphorylated leading to disruption of this binding? If they do, then relocalized tau may boost Fyn action in dendrites and hence phos-phorylation of NMDA receptors, thus enhancing excitotoxic signaling and increasing the sensitivity of neurons to Aβ (Figure 1). However, before targeting the tau-Fyn interaction for therapeutic purposes, we need to know more about the mechanism by which tau medi-

ates the transport of Fyn to dendrites. If that transport mechanism is disturbed, what other neuronal functions would be affected? Clearly, not only Fyn, but also other signaling molecules containing SH3 domains bind to the PXXP motifs in the N-terminal domain of tau and of related microtubule-associated proteins; but we do not know whether inhibition of this binding would have deleterious effects. Also, is ∆tau just a scavenger that when overexpressed sequesters Fyn? Could proteins other than tau per-form this activity in vivo? The enhanced phenotypes upon reduction of endoge-nous tau seem to speak against this pos-sibility. The provocative work of Ittner et al. raises new challenging questions, but also brings us a significant step closer to understanding AD pathophysiology. Hopefully, these findings will help in the design of new therapeutic strategies for reducing the synaptic dysfunction and neuronal loss of AD and other neurode-generative diseases.

Figure 1. A Toxic Triad in Alzheimer’s Disease.Shown is the neuronal localization of tau protein (red) and Fyn kinase (blue) under physiological and pathological conditions. (Left) Normal tau is primarily located in axons but also interacts with Fyn kinase and targets it to den-drites. Fyn kinase phosphorylates subunit 2 of the NMDA receptor in dendritic spines, which results in stabilization of the receptor’s interaction with the postsynaptic density protein PSD95, leading to enhanced excitotoxicity. Excitotoxicity is known to increase the toxic effects of oligomers of Aβ (dark purple) on neurons. (Middle) Overexpression of a ∆tau variant lacking the C terminus, which binds to Fyn kinase but not to microtubules, results in the sequestration of Fyn in the soma, preventing Fyn from reaching the dendrites. Consequently, Fyn-mediated phosphorylation of NMDA receptors is decreased and Aβ-mediated toxicity is reduced (pale purple). (Right) Enhanced redistribution of abnormally hyperphosphorylated tau from axons to the somatoden-dritic compartment during AD pathogenesis may increase tau-dependent sorting of Fyn to the dendrites, boosting excitotoxic signaling and increasing the toxic effects of Aβ (black) on neurons.

previews.indd 357 7/28/2010 2:09:14 PM


ReFeRences

Aarts, M., Liu, Y., Liu, L., Besshoh, S., Arundine, M., Gurd, J.W., Wang, Y.T., Salter, M.W., and Tymi-anski, M. (2002). Science 298, 846–850.

Ballatore, C., Lee, V.M., and Trojanowski, J.Q. (2007). Nat. Rev. Neurosci. 8, 663–672.

Haass, C., and Selkoe, D.J. (2007). Nat. Rev. Mol. Cell Biol. 8, 101–112.

Ittner, L.M., Ke, Y.D., Delerue, F., Bi, M., Gladbach,

A., v Eersel, J., Wölfing, H., Chieng, B.C., Christie, M.J., Napier, I.A., et al. (2010). Cell, this issue.

Klein, C., Kramer, E.M., Cardine, A.M., Schraven, B., Brandt, R., and Trotter, J. (2002). J. Neurosci. 22, 698–707.

Konzack, S., Thies, E., Marx, A., Mandelkow, E.M., and Mandelkow, E. (2007). J. Neurosci. 27, 9916–9927.

Lee, G., Newman, S.T., Gard, D.L., Band, H., and Panchamoorthy, G. (1998). J. Cell Sci. 111,

3167–3177.

Nakazawa, T., Komai, S., Tezuka, T., Hisatsune, C., Umemori, H., Semba, K., Mishina, M., Manabe, T., and Yamamoto, T. (2001). Biol. Chem. 276, 693–699.

Palop, J.J., and Mucke, L. (2009). Arch. Neurol. 66, 435–440.

Roberson, E.D., Scearce-Levie, K., Palop, J.J., Yan, F., Cheng, I.H., Wu, T., Gerstein, H., Yu, G.Q., and Mucke, L. (2007). Science 316, 750–754.

The tumor suppressor protein p53 is one of the cell’s most important barri-ers against oncogenic transformation. By regulating the expression of thousands of genes, either directly or indirectly, p53 pro-foundly influences cell fate in response to stress. Several decades of research have established p53 as a transcriptional acti-vator with high sequence specificity. How-ever, p53 clearly also represses at least as many genes as it activates, if not more. Despite this, the mechanism of repression is less well characterized than the trans-activation mechanism by p53. Now, the informative study by Huarte et al. (2010) in this issue lays the framework for a new and elegant mechanism by which p53 globally downregulates a large subset of its repres-sion targets. These authors show that the long intergenic RNA-p21 (lincRNA-p21), a bona fide downstream target of p53, is a key inhibitor of gene expression through its interaction with heterogeneous ribonu-cleoprotein K (hnRNP-K).

Although the first reports of gene repression by p53 focused on suppres-sion of the basal transcriptional machin-

ery, subsequent studies identified more precise mechanisms occurring at specific genes (reviewed in Laptenko and Prives, 2006). These include p53 interacting with and inhibiting specific transcription factors; displacement of specific activa-tors from promoters due to the presence of overlapping binding sites; the recruit-ment by p53 of chromatin-modifying factors, such as histone deacetylases, which then block gene expression; and the binding of p53 to unique “repression” response elements. In addition, p53 may also inhibit genes indirectly by activat-ing transcription of factors that block expression of specific genes. Most nota-bly, many labs have demonstrated that the cell-cycle inhibitor p21 (especially within the context of Rb-family activa-tion) is a critical mediator of p53-depen-dent transcriptional repression (Figure 1, top) (Barsotti and Prives, 2009, and references therein). Recently, studies indicate that p53 regulates microRNAs, which either degrade mRNA targets or inhibit their translation into protein. The p53 protein facilitates not only the tran-

scriptional activation of microRNAs but also their processing into mature, active forms (Figure 1, middle) (Shi et al., 2010). Now, Huarte, Rinn, and their colleagues (Huarte et al., 2010) add an exciting new route through which p53 executes wide-spread gene repression, specifically by activating a long intergenic RNA (Figure 1, bottom).

LincRNAs are large RNA molecules (primary transcript ≥5 kb) that are evo-lutionarily conserved across mammalian genomes. Although these RNAs are tran-scribed by RNA polymerase II, 5′capped, and polyadenylated like normal mRNAs, they do not code for proteins (Guttman et al., 2009). Previous work by the Rinn group suggested that lincRNAs may repress transcription by targeting chro-matin-modifying complexes to specific genomic loci (Khalil et al., 2009).

In their new work, Huarte et al. (2010) sought to identify specific lincRNAs that operate within the p53 pathway. They con-structed a tiling microarray designed to detect the expression of ?400 lincRNAs. They then incubated this array with RNA

noncoding RnAs: The Missing “Linc” in p53-Mediated RepressionAnthony M. Barsotti1 and Carol Prives1,*1Department of Biological Sciences, Columbia University, New York, NY 10027, USA*Correspondence: [email protected] 10.1016/j.cell.2010.07.029

The tumor suppressor protein p53 coordinates the cellular response to stress through regulation of gene expression. Now, Huarte et al. (2010) identify a long intergenic noncoding RNA as a new player in p53-mediated repression of genes involved in apoptosis.

previews.indd 358 7/28/2010 2:09:14 PM


isolated from two mouse cell lines engineered with control-lable expression of p53. These experiments uncovered sev-eral lincRNAs that are poten-tial targets of p53, and the authors focused on lincRNA-p21, named for its proximity to the p53-target gene p21. Importantly, they confirmed that the human ortholog of the mouse lincRNA-p21 is con-served in both sequence and regulation.

To ascertain the role of lincRNA-p21 within the p53 pathway, the authors induced the p53 response pathway and then silenced either p53 or lincRNA-p21. By compar-ing microarrary data under these two conditions, the authors identified a large set of genes that are derepressed in response to disruption of both p53 and lincRNA-p21. Moreover, these specific genes are highly represented in the group of genes normally repressed by p53. This is a significant finding because it implicates lincRNA-p21 as a potent downstream mediator of p53-dependent transcriptional repression that acts on a very large scale.

As mentioned, p21 itself is also thought to inhibit numerous target genes of p53. Nevertheless, disruption of lincRNA-p21 does not alter levels of p21 mRNA or protein. Furthermore, genes regulated by lincRNA-p21 do not appear to overlap with those identified in previous studies as targets of p21. These results suggest that lincRNA-p21 acts independently of p21.

In response to stress, p53 initiates a cellular program that often results in cell-cycle arrest or cell death. Despite the large common set of genes regulated by p53 and lincRNA-p21, Huarte, Rinn, and coworkers found that silencing of lin-cRNA-p21 blocks programmed cell death (i.e., apoptosis) but not cell-cycle arrest. LincRNA-p21 represses the expression of several important prosurvival factors, which may help to explain this interesting phenotype. This hypothesis is supported by previous studies demonstrating that the repression of antiapoptotic genes

by p53 is crucial for p53-facilitated cell death (Ho and Benchimol, 2003). Further, Huarte et al. show that the activation of certain proapoptotic genes by p53, such as Noxa and Perp, depends on lincRNA-p21. Therefore, lincRNA-p21 may also play an important role in determining the promoter selectivity of p53 for gene acti-vation in addition to directing the repres-sion of target genes.

Next, Huarte and colleagues delved into the mechanism by which lincRNA-p21 represses transcription. With great satisfaction, they identify hnRNP-K as a key protein partner for lincRNA-p21. Best known for its role in chromatin remodeling and transcriptional regula-tion (initiation, elongation, and termina-tion), hnRNP-K binds to a wide range of molecules, including DNA, RNA, protein kinases, and factors that remodel chro-matin. Thus, hnRNP-K also participates in diverse processes, such as RNA splicing, mRNA stability, and translation (Bomsztyk et al., 2004).

After validating the specificity of the association between lincRNA-p21 and hnRNP-K and identifying the region of the RNA necessary for their binding, the authors again turned to microar-ray analyses to assess the importance

of this interaction. Indeed, a large percentage of genes commonly inhibited by p53 and lincRNA-p21 also require hnRNP-K for their repression. Further, hnRNP-K is recruited to promoters for genes that are downregulated by the p53/ l incRNA-p21/hnRNP-K axis, and this recruitment depends on lincRNA-p21. Taken together, these results imply that hnRNP-K plays a significant and direct role in executing gene repres-sion mediated by p53 and lincRNA-p21.

Supporting this model, previous studies found that hnRNP-K is present in a repressive complex with the linker histone H1.2. Moreover, this complex blocks chromatin acetylation and subsequent transcriptional activation by p53 and the transcription fac-tor complex p300 (Kim et al.,

2008). In contrast, however, Moumen et al. (2005) found that hnRNP-K cooperates with p53 as a coactivator of transcription in response to DNA damage. In the system used by Huarte and colleagues, hnRNP-K and lincRNA-p21 have independent roles as well, with hnRNP-K sharing an over-lapping set of target genes with p53 that are not regulated by lincRNA-p21. Thus, it is not clear that lincRNA-p21 and hnRNP-K are more closely related than p53 and hnRNP-K. Nevertheless, the large amount of data generated by the new microarray analyses supports a more widespread role for hnRNP-K in p53-dependent gene repression than in p53-dependent gene activation.

The Huarte et al. findings raise several key questions. Does the binding part-ner of hnRNP-K determine whether it acts as a positive or negative regulator of transcription? For example, does the interaction between p53 and hnRNP-K serve as a platform for other coactiva-tors, whereas the association between lincRNA-p21 and hnRNP-K favors for-mation of a repressive complex? This model would indeed explain the seem-ingly contradictory roles of hnRNP-K with regard to p53 activity. Further, how does lincRNA-p21 influence the binding

Figure 1. Activation for Repression’s sakeActivation of p53 in response to stress triggers expression of three distinct classes of targets that silence gene expression in different ways, resulting in different cellular outcomes.(Top) The p21 protein inhibits cyclin-dependent kinases, thereby activating the Rb family of proteins and inhibiting the E2F transcription factors. This ultimately halts progression of the cell cycle.(Middle) The microRNA miR-34a degrades or inhibits translation of the mRNAs of genes important for both cell-cycle progression and survival.(Bottom) A new class of p53 targets, exemplified by the long intergenic non-coding RNA lincRNA-p21, is activated by p53 and then inhibits gene expres-sion (Huarte et al., 2010). Although the exact mechanism of gene repression is unclear, lincRNA-p21 interacts with heterogeneous ribonucleoprotein K (hnRNP-K) to downregulate a large group of genes that are repressed by p53, including several that are important for cell survival.

previews.indd 359 7/28/2010 2:09:15 PM


of hnRNP-K to specific genomic loci? Is there functional significance to the remarkable proximity (?15 kb) of lin-cRNA-p21 to the p21 gene? If these loci are regulated interdependently, they may act as a key molecular switch between life and death. Finally, given the impor-tance of lincRNA-p21 to p53-dependent cell death, is lincRNA-p21 mutated in cancer? Answers to each of these ques-tions will certainly enrich our under-standing of the functional relationship between p53 and this powerful class of regulatory molecules, lincRNAs.

ReFeRences

Barsotti, A.M., and Prives, C. (2009). Oncogene 28, 4295–4305.

Bomsztyk, K., Denisenko, O., and Ostrowski, J. (2004). Bioessays 26, 629–638.

Guttman, M., Amit, I., Garber, M., French, C., Lin, M.F., Feldser, D., Huarte, M., Zuk, O., Carey, B.W., Cassady, J.P., et al. (2009). Nature 458, 223–227.

Ho, J., and Benchimol, S. (2003). Cell Death Differ. 10, 404–408.

Huarte, M., Guttman, M., Feldser, D., Garber, M., Koziol, M., Kenzelmann-Broz, D., Khalil, A., Zuk, O., Amit, I., Rabani, M., et al. (2010). Cell, this issue.

Khalil, A.M., Guttman, M., Huarte, M., Gar-ber, M., Raj, A., Rivea Morales, D., Thomas, K., Presser, A., Bernstein, B.E., van Oudenaarden, A., et al. (2009). Proc. Natl. Acad. Sci. USA 106, 11667–11672.

Kim, K., Choi, J., Heo, K., Kim, H., Levens, D., Kohno, K., Johnson, E.M., Brock, H.W., and An, W. (2008). J. Biol. Chem. 283, 9113–9126.

Laptenko, O., and Prives, C. (2006). Cell Death Dif-fer. 13, 951–961.

Moumen, A., Masterson, P., O’Connor, M.J., and Jackson, S.P. (2005). Cell 123, 1065–1078.

Shi, M., Liu, D., Shen, B., and Guo, N. (2010). Bio-chim. Biophys. Acta 1805, 218–225.

Stem cells have the immense responsi-bility of maintaining tissue and organism homeostasis over the lifetime of an indi-vidual. As such, stem cells are speculated to have evolutionary characteristics that offer protection against acute insults, allowing them to survive and to repopu-late their tissues in the short term. How-ever, they must also act as self-renew-ing guardians of the genome to ensure maximal integrity of the genomic code for future stem cells and their mature tis-sue progeny. The hematopoietic (blood) system is perhaps the best studied tis-sue in terms of its hierarchical develop-ment from a small number of long-term stem cells that are relatively quiescent, to progenitors that proliferate and dif-ferentiate, and then to mature blood cell lineages that are produced by the billion

each day. Hematopoietic stem cells are thought to be resistant to injury including DNA damage, which may be related to their specific gene expression programs, epigenetic factors, or exogenous protec-tion by the stem cell “niche.” Two new reports in Cell Stem Cell from the labora-tories of Emmanuelle Passegué (Mohrin et al., 2010) and John Dick (Milyavsky et al., 2010) further our understanding of how hematopoietic stem cells respond to radiation-induced DNA damage.

So how do quiescent stem cells han-dle genotoxic insults? Mohrin et al. (2010) found that murine hematopoietic stem and progenitor cells (HSPCs)—defined as bone marrow cells expressing the mark-ers: lineage−/c-Kit+/Sca-1+/Flk2−—were more resistant to apoptosis induced by a specific dose of ionizing radiation

than were more differentiated progenitor cells (Figure 1). The unique DNA damage response of mouse HSPCs involves the tumor suppressor protein p53 and is lost when stem cells are forced out of qui-escence and into the cell cycle by treat-ment with chemotherapy or cytokines. Not only are quiescent HSPCs poised to resist apoptosis as evidenced by their antiapoptotic gene expression program, but they are also able to repair their DNA by nonhomologous end joining (NHEJ). Repair of DNA damage through homolo-gous recombination (which has a lower error rate than NHEJ) requires that cells enter the cell cycle; thus, quiescent stem cells must rely on NHEJ as an alterna-tive. The reliance of quiescent adult tis-sue stem cells on NHEJ for the repair of DNA damage may in fact be a general

stem cells and DnA Damage: Persist or Perish?Andrew A. Lane1,2,* and David T. Scadden1,2,*1Center for Regenerative Medicine, Cancer Center, Massachusetts General Hospital, Boston, MA 02114, USA2Harvard Stem Cell Institute, Department of Stem Cell and Regenerative Biology, Harvard University, Boston, MA 02114, USA*Correspondence: [email protected] (A.A.L.), [email protected] (D.T.S.)DOI 10.1016/j.cell.2010.07.030

Stem cells repopulate tissues after injury while also renewing themselves, but this makes them vulnerable to genotoxic damage. Mohrin et al. (2010) and Milyavsky et al. (2010) now show that mouse and human hematopoietic stem cells make opposing decisions about whether to die or to persist in response to DNA damage.

previews.indd 360 7/28/2010 2:09:15 PM


phenomenon in mice, given the similar conclusions of a recent study using hair follicle stem cells as a model system (Sotiropoulou et al., 2010).

Unfortunately, Mohrin et al. (2010) also uncover a down-side to short-term radioresis-tance and rapid DNA repair through the error-prone NHEJ pathway. Spectral karyotyping revealed gross chromosomal aberrations in irradiated mouse HSPCs, and some of the same abnormal cytogenetic find-ings persisted in the progeny of irradiated HSPCs trans-planted into mouse recipients. Furthermore, despite their resistance to apoptosis imme-diately after injury, irradiated HSPCs were unable to con-tribute to long-term sustained hematopoiesis when seri-ally transplanted into mouse recipients. Such events would be of obvious risk to a long-lived organism as serial expo-sure of stem cells to genotoxic agents could readily result in leukemia or aplasia.

Does the human hemato-poietic system accept the same tradeoff between stem cell survival in the short term versus accumulation of del-eterious mutations in the long term? In a companion study, Milyavsky et al. (2010) addressed this question in human umbilical cord blood cells. They observed an enhanced sen-sitivity to apoptosis induced by low-dose ionizing radiation in these cells compared to more differentiated cells. In contrast to the findings of Mohrin et al. in the mouse, these authors noted that human hematopoietic stem and early multipotent progenitor cells were poised for apopto-sis in response to DNA damage. Survival and the clonogenic and reconstitutive capacity of the irradiated human HSPCs were rescued by blocking p53 expression or by overexpression of the antiapoptotic factor Bcl-2. However, irradiated human HSPCs lacking p53 were unable to sus-tain hematopoiesis and showed evidence of persistent DNA double-strand breaks when serially transplanted into recipi-

ent mice. Therefore, a short-term gain in survival could be achieved by human hematopoietic stem cells as found in the mouse, but the default setting for irradi-ated human HSPCs is an increase in p53 expression resulting in apoptosis.

The differences between these two studies may have a technical basis: mark-ers for stem and progenitor populations are more refined in the mouse than in the human so somewhat different stem and progenitor cell populations may have been analyzed. In addition, slightly differ-ent doses of radiation were used. In the in vivo experiments of Milyavsky et al., human HSPCs from umbilical cord blood were transplanted into the mouse bone marrow niche, which may have provided less efficient survival signals for human cells than for mouse cells. Also, human

umbilical cord blood HSPCs have a different biology than the bone marrow HSPCs of mouse. However, it is tempt-ing to see the different find-ings in the light of evolution, that is, as a reflection of the different challenges faced by mammals with different life spans and ages of reproduc-tive maturity.

Recent elegant studies from Bondar and Medzhitov (2010) and Marusyk et al. (2010) demonstrate that competitive selection takes place within tissue stem cell populations. These authors found that irra-diated p53-deficient HSPCs in the mouse have an initial survival advantage but that long-term fitness is balanced by complex interactions with neighboring HSPCs and the relative levels of p53 and DNA damage in stem cells and their neighbors. The studies from the Passegué and Dick labs indi-cate that hematopoietic cells within a tissue have adopted different means of handling DNA damage depending on their differentiation stage. That mouse and human stem cells may have acquired or under-gone selection for distinct responses to ionizing radiation is a reasonable notion. How-

ever, it remains to be seen which specific molecular mechanisms that differ between stem cells and progenitors, or between stem cells of different species, lead to these distinctive traits. We now have a set of reagents with which to discover and understand how such important yet differ-ent tissue stem cell traits have evolved.

There are other practical implications of the Passegué and Dick reports. Sec-ondary myelodysplasia and leukemia are believed to arise from DNA damage to HSPCs from the radiation or chemother-apy given to treat a primary malignancy. Mohrin et al. show intriguing evidence that NHEJ activity and chromosomal aberra-tions decrease when HSPCs are induced to enter the cell cycle prior to irradiation. Interestingly, a parallel evolving hypoth-esis in the study of cancer stem cells sug-

Figure 1. Hematopoietic cell Responses to DnA Damage(Top) Quiescent murine hematopoietic stem and progenitor cells (HSPCs) are poised to survive DNA damage induced by low-level ionizing radiation through a DNA repair process called nonhomologous end joining (NHEJ), which tends to be error prone (Mohrin et al., 2010). In contrast, mouse HSPCs and committed progenitors (CP) progressing through the cell cycle are more likely to undergo apoptosis or repair their DNA using higher-fidelity homolo-gous recombination. Although the short-term consequence of HSPC survival is maintenance of tissue integrity in the face of injury, long-term consequenc-es include genomic rearrangements that persist and HSPCs with a diminished functional capacity. (Bottom) In contrast, compared to more committed progenitors, the default program for damaged human HSPCs is to undergo apoptosis. However, a de-crease in p53 rescues human HSPCs from apoptosis immediately after low-level irradiation (Milyavsky et al., 2010). Despite interspecies differences in the short-term response to radiation, the long-term functional consequences of avoiding apoptosis for both mouse and human HSPCs include persistent DNA damage and decreased self-renewal capacity. The delicate balance be-tween tissue survival and the DNA damage response therefore could predis-pose surviving HSPCs to future malignant transformation.

previews.indd 361 7/28/2010 2:09:15 PM


gests that activating leukemia stem cells from quiescence prior to chemotherapy may result in more efficient elimination of these cancer-repopulating cells (Saito et al., 2010). An unexpected benefit of such a prestimulation strategy may be that nor-mal hematopoietic stem cells activated from quiescence would simultaneously be protected from accumulating long-term DNA damage. However, as shown by Milyavsky and colleagues, stem cell escape from acute damage, particularly if it involves a decrease in p53 activity, may lead to long-term deleterious effects on stem cell fitness and repopulating ability. The interplay between the response to

acute injury and long-term fitness needs to be more fully understood and will require both laboratory models and the thoughtful correlative study of stem cells from patients receiving genotoxic chemo-therapy. Understanding these events may point the way to methods for preserving short-term tissue reconstitution while maintaining long-term cell and genomic integrity.

ReFeRences

Bondar, T., and Medzhitov, R. (2010). Cell Stem Cell 6, 309–322.

Marusyk, A., Porter, C.C., Zaberezhnyy, V., and DeGregori, J. (2010). PLoS Biol. 8, e1000324.

Milyavsky, M., Gan, O.I., Trottier, M., Komosa, M., Tabach, O., Notta, F., Lechman, E., Hermans, K.G., Eppert, K., Konovalova, Z., et al. (2010). Cell Stem Cell 7. Published online July 7, 2010. 10.1016/j.stem.2010.05.016.

Mohrin, M., Bourke, E., Alexander, D., Warr, M.R., Barry-Holson, K., Le Beau, M.M., Mor-rison, C.G., and Passegué, E. (2010). Cell Stem Cell 7. Published online July 7, 2010. 10.1016/j.stem.2010.06.014.

Saito, Y., Uchida, N., Tanaka, S., Suzuki, N., Tomi-zawa-Murasawa, M., Sone, A., Najima, Y., Takagi, S., Aoki, Y., Wake, A., et al. (2010). Nat. Biotech-nol. 28, 275–280.

Sotiropoulou, P.A., Candi, A., Mascré, G., De Cler-cq, S., Youssef, K.K., Lapouge, G., Dahl, E., Se-meraro, C., Denecker, G., Marine, J.C., and Blan-pain, C. (2010). Nat. Cell Biol. 12, 572–582.

Mitochondria, the power plants of the eukaryotic cell, are bound by two mem-branes and contain 1000–1500 different proteins and tens of RNAs. Most of the genes that encode mitochondrial pro-teins are found in the nuclear genome and thus are translated in the cytosol and then imported into mitochondria. The pathways and machineries required for protein import into mitochondria have been extensively studied and are highly conserved among fungi, plants, and mammals (Endo and Yamano 2009; Chacinska et al., 2009). The mitochon-drial matrix also contains several kinds of noncoding RNAs that are also imported from the cytosol. However, in contrast to protein translocation, the mechanisms that mediate import of RNAs into mito-chondria remain enigmatic (Salinas et al.,

2008; Lithgow and Schneider, 2010). In this issue of Cell, Wang et al. (2010) shed light on this question, revealing that poly-nucleotide phosphorylase (PNPase) is a much sought after component of the RNA import apparatus in mammalian cells.

PNPases comprise an evolutionally conserved enzyme family (found in bac-teria, plants, flies, and mammals but not in yeast) that has 3′→5′ exoribonuclease and RNA-polymerase activities (Chen et al., 2007). Although bacterial PNPases are cytosolic, eukaryotic PNPases are mainly localized in mitochondria or chlo-roplasts. Prior work has established how PNPases get to the intermembrane space (IMS). After crossing the mito-chondrial outer membrane via the trans-locase of outer mitochondrial membrane 40 (TOM40) complex, the PNPase pre-

cursor engages with the translocase of the inner membrane 23 (TIM23) complex (Figure 1) (Chen et al., 2006; Rainey et al., 2006). After the PNPase presequence is removed by matrix processing pepti-dase (MPP), an AAA protease Yme1 in the inner membrane pulls PNPase into the IMS, where PNPase assembles into a trimeric complex (Figure 1).

Wang et al. now assess the function of mammalian PNPase by tissue-specific disruption of the PNPase gene in mouse hepatocytes. They find that mitochondria from hepatocytes deficient in PNPase display defects in oxidative phosphory-lation (OXPHOS), the major ATP-gener-ating metabolic pathway of respiration. This defect is shown to arise from the failure in the processing of polycistronic mitochondrial mRNAs encoding the

Mitochondrial Matrix Reloaded with RnAToshiya Endo,1,* Koji Yamano,1 and Tohru Yoshihisa2,3

1Department of Chemistry, Graduate School of Science, Nagoya University, Chikusa-ku, Nagoya 464-8602, Japan2Research Center for Materials Science, Nagoya University, Chikusa-ku, Nagoya 464-8602, Japan3Precursory Research for Embryonic Science and Technology, Japan Science and Technology Agency, Kawaguchi, Saitama 332-0012, Japan*Correspondence: [email protected] 10.1016/j.cell.2010.07.024

Although mitochondrial biogenesis requires the import of specific RNAs, the pathways and cellular machineries involved are only poorly understood. Wang et al. (2010) now find that polynucleotide phosphorylase in the intermembrane space of mammalian mitochondria facilitates import of several RNAs into the mitochondrial matrix.

previews.indd 362 7/28/2010 2:09:15 PM


subunits of the OXPHOS complexes. But how does PNPase in the IMS affect the processing of RNAs in the matrix?

A hint to the answer came from the observation that the RNA component of mammalian mitochondrial RNase P is markedly decreased by PNPase depletion. Given that RNase P medi-ates processing of mitochondrial trans-fer RNAs (tRNAs), which are encoded between open reading frames for OXPHOS components, failure in tRNA excision also impairs maturation of the OXPHOS subunit transcripts. In addi-tion, expression of human PNPase in yeast mitochondria, which do not pos-sess PNPase, enhances the import effi-ciency of heterologous human RNase P RNA, 5S ribosomal RNA, and the RNA component of MRP, a ribonucleopro-tein that mediates mitochondrial DNA replication. Similar PNPase-dependent import in vitro is confirmed with mito-chondria isolated from PNPase-defi-cient mouse liver or embryonic fibro-blasts. These findings suggest that PNPase in the IMS is a component of the RNA import system in human mito-chondria.

What is the mechanism of RNA import that is facilitated by PNPase? In yeast and plant mitochondria, substrate tRNAs are recognized by the mitochondrial surface components including the import recep-tor Tom20, and then move through the Tom40 channel or voltage-dependent anion channel (VDAC). This movement probably involves several mechanisms, including a piggyback mechanism in which the RNA substrate is associated with an appropriate escort protein (Sali-nas et al., 2008; Lithgow and Schneider, 2010). Although no receptor or translo-cation channel for RNA import is known for mammalian mitochondria, the pres-ent study reveals that PNPase binds to substrate RNAs with a specific stem-loop structure. When grafted onto other nonsubstrate RNAs, this stem-loop can direct the RNA into yeast mitochondria containing human PNPase. Therefore, after crossing the outer membrane by an unknown mechanism, substrate RNAs with a specific stem-loop structure are recognized by PNPase, which subse-quently allows the transfer of a subset of RNAs to downstream components of the import pathway. The binding of PNPase

to substrate RNAs may also contribute to the unidirectional translocation of RNAs across the outer membrane via a trap-ping mechanism or by preventing their backward movement to the cytosol.

The identity of the import chan-nel in the inner membrane is unknown, although Wang et al. find that translo-cation of RNase P RNA into yeast mito-chondria with human PNPase requires the membrane potential. This suggests that the membrane potential may help remove bound RNAs from PNPase or facilitate unidirectional translocation of RNAs across the inner membrane (Fig-ure 1). Translocation of RNAs through the inner membrane import channel may also require an additional protein, such as mitochondrial heat shock protein 70, as an import motor for presequence-con-taining mitochondrial proteins. Although many of the components of the yeast and human RNA import pathway remain to be identified, the efficient in vitro system developed by Wang et al. should acceler-ate the pace of discovery.

The study by Wang et al. casts a spot-light on mitochondrial RNA import, with mammalian PNPase taking its rightful

place center stage. Will the unfolding drama of RNA translocation in mitochon-dria reveal common principles or expose a diversity of organism specific mecha-nisms?

ReFeRences

Chacinska, A., Koehler, C.M., Milenkovic, D., Lithgow, T., and Pfanner, N. (2009). Cell 138, 628–644.

Chen, H.-W., Rainey, R.N., Balatoni, C.E., Dawson, D.W., Troke, J.J., Wasiak, S., Hong, J.S., McBride, H.M., Koehler, C.M., Teitell, M.A., and French, S.W. (2006). Mol. Cell. Biol. 26, 8475–8487.

Chen, H.-W., Koehler, C.M., and Teitell, M.A. (2007). Trends Cell Biol. 17, 600–608.

Endo, T., and Yamano, K. (2009). Biol. Chem. 390, 723–730.

Lithgow, T., and Schneider, A. (2010). Philos. Trans. R. Soc. Lond. B Biol. Sci. 365, 799–817.

Rainey, R.N., Glavin, J.D., Chen, H.-W., French, S.W., Teitell, M.A., and Koehler, C.M. (2006). Mol. Cell. Biol. 26, 8488–8497.

Salinas, T., Duchêne, A.-M., and Maréchal-Drouard, L. (2008). Trends Biochem. Sci. 33, 320–329.

Wang, G., Chen, H.-W., Oktay, Y., Zhang, J., Allen, E.L., Smith, G.M., Fan, K.C., Hong, J.S., French, S.W., McCaffery, J.M., et al. (2010). Cell, this issue.

Figure 1. PnPase Facilitates the Import of RnAs into the Mitochondrial MatrixThe polynucleotide phosphorylase (PNPase) precursor is imported into the intermembrane space with the aid of the translocase of outer mitochondrial membrane 40 (TOM40) complex, the translocase of the inner mitochondrial membrane (TIM23) complex, matrix processing peptidase (MPP), and AAA protease Yme1. The trimeric PNPase promotes the import of RNAs from the cytosol into the matrix. Mitochondrial mRNAs are processed in the matrix and their translation products are assembled into the oxidative phosphorylation (OXPHOS) complexes I–IV. The OXA complex integrates proteins from the matrix into the inner membrane. OM, outer membrane; IM, inner membrane. Heat shock protein 70 (Hsp70) is shown as an import motor for presequence-containing proteins. AAC, ADP-ATP carrier; ∆Ψ, membrane potential.

previews.indd 363 7/28/2010 2:09:17 PM

Leading Edge

Minireview


The microtubule cytoskeleton is a remarkable structure that can adopt diverse architectures uniquely suited to the indi-vidual needs of a particular cell type or process. For example, in vertebrate cells, the mitotic spindle, which separates chro-mosomes during anaphase, contains two antiparallel arrays of microtubules with their minus ends anchored at opposing centrosomes and their plus ends overlapping to form a bundle of crosslinked filaments in the middle of the spindle (Figure 1, bottom inset). In contrast, plant (angiosperm) cells do not pos-sess discrete microtubule organizing centers (i.e., centrioles) but instead rely primarily on specific interactions between microtubules to organize the filaments into crosslinked arrays (Ehrhardt, 2008).

The variety of microtubule structures observed across differ-ent cell types requires a diverse group of proteins to assemble, stabilize, and dynamically control these microtubule arrays. Microtubule-associated proteins (MAPs), which include both molecular motors and nonmotor proteins, regulate the global properties of microtubule structures by moving and crosslinking filaments. Although much is known about these individual pro-teins, key questions remain about how they interact to control the size, shape, and dynamics of microtubule arrays. Now, two studies in this issue of Cell (Bieling et al., 2010; Subramanian et al., 2010) demonstrate how the MAP65 protein PRC1 (protein regulator of cytokinesis 1) independently bundles microtubules into antiparallel arrays and works with two motors, kinesin-4 and kinesin-5, to control the global properties of these overlap-ping regions. Together, these papers suggest a model for how microtubule bundles can persist despite the action of numer-ous motor proteins acting along them.

Microtubule Structure and DynamicsMicrotubules are linear polymers inside the cell composed of α/β-tubulin heterodimers arranged head to tail into protofilaments. The protofilaments, typically 13, associate laterally to form a hollow tube with substantial flexural rigidity and inherent structural polar-ity, described as having plus and minus ends (Figure 1). Microtu-bules exhibit dynamic instability (Mitchison and Kirschner, 1984) wherein individual microtubules within a population interconvert between states of growth and shortening. In general, microtubule plus ends are more dynamic than minus ends.

The polarity of microtubules within a bundle of overlapping filaments is critical to the action of motor proteins that slide filaments past each other and drive the movement of cargoes, such as chromosomes, on these microtubules. For example, during late anaphase, the overlapping regions of microtubules at the center of the mitotic spindle elongate as the microtubules push the spindle poles to opposite sides of the cell. However, to separate daughter cells during cytokinesis, this overlapping region shortens and forms a dense, compact array of anti-parallel microtubules, called the midzone. How does the cell specify the size of the overlapping region in a bundle and man-age the timing and position of its remodeling? To answer these questions requires a better understanding of the key molecular players that govern the formation of microtubules.

MAP65 Family of Microtubule Crosslinking ProteinsOne major class of proteins that crosslink microtubules into arrays is the MAP65/Ase1/PRC1 family. Biochemical studies with plant extracts identified the first members of this family as 65 kD proteins capable of bundling microtubules (Chang-Jie and Sonobe, 1993). Subsequent studies demonstrated that MAP65 proteins crosslink microtubules in vitro with a spacing between microtubule filaments of 25 nm (Chan et al., 1999), consistent with in situ observations of microtubule bundles in plants. A genetic screen identified the yeast ortholog of MAP65 as Ase1, which is required for properly elongating the spindle during anaphase (Pellman et al., 1995). Ase1 was later shown to be a homodimer that also bundles microtubules in vitro (Schuyler et al., 2003), an activity that is critical for its role in sliding microtubule filaments past each other during anaphase. The vertebrate ortholog of MAP65/Ase1 is PRC1, and in mam-malian cells, PRC1 regulates the organization of the central spindle during cytokinesis (Jiang et al., 1998; Mollinari et al., 2002). Phosphorylation of PRC1 by cyclin dependent kinase 1 (Cdk1) negatively regulates the crosslinking activity of PRC1, which limits the bundling of microtubules by PRC1 until late stages of mitosis when they are needed for cytokinesis (Zhu et al., 2006).

Like other MAP65 proteins, PRC1 and Ase1 are not molec-ular motors themselves but instead work in concert with motor proteins to organize arrays of microtubules. In fis-

A MAP for Bundling MicrotubulesClaire E. Walczak1,* and Sidney L. Shaw2,*1Medical Sciences2Department of BiologyIndiana University, Bloomington, IN 47405, USA*Correspondence: [email protected] (C.E.W.), [email protected] (S.L.S.)DOI 10.1016/j.cell.2010.07.023

Microtubules assemble into arrays of bundled filaments that are critical for multiple steps in cell division, including anaphase and cytokinesis. Recent structural and functional studies, including two papers in this issue of Cell (Bieling et al., 2010; Subramanian et al., 2010), demonstrate how the MAP65 protein PRC1 crosslinks microtubules and cooperates with kinesin motors to control the dynamics and size of bundled regions.

walczak.indd 364 7/28/2010 2:09:27 PM


sion yeast, Ase1 dynamically controls the overlap of bundles by coordinating with the kinesin-14 motor klp2 (Janson et al., 2007). In mammalian cells, PRC1 is transported to the midzone of the spindle by Kif4 (kinesin family member 4), a kinesin-4 motor protein that is critical for positioning chro-mosomes and for cytokinesis in multiple organisms (Glotzer, 2009; Hornick et al., 2010). Disruption of either PRC1 or Kif4 perturbs the localization of the other protein, making it diffi-cult to elucidate whether the Kif4 motor recruits PRC1 to the microtubules or whether loss of PRC1 disrupts the localiza-tion of the central spindle and thus Kif4. Nevertheless, PRC1 is required to set up the central spindle before a number of other kinesins locate to the spindle. These kinesins include motors involved in finishing the assembly of the central spin-dle and in cytokinesis, including mitotic kinesin-like protein 1 (MKLP-1) and 2 and Kif14 (Glotzer, 2009; Hornick et al., 2010). Nevertheless, it is still unknown how PRC1 interacts with these motor proteins to direct the size, shape, and sta-bility of the central spindle.

Kinesin-4 Limits Microtubule Overlap LengthTo understand how PRC1 organizes microtubule arrays, Biel-ing and colleagues (2010) developed a total internal reflection fluorescence (TIRF) microscopy assay in which microtubule seeds are attached to a microscope slide and mixed with

tubulin dimers to polymerize dynamic microtubules. They then added PRC1 proteins labeled with fluorescent tags to crosslink the microtubules (Figure 1A). The PRC1 proteins bound prefer-entially to microtubules that overlap at antiparallel regions, showing decisively that PRC1 alone is sufficient to cross-

link antiparallel microtubules.Knowing that PRC1 interacts directly with Kif4 (Xklp1 in Xeno-

pus), Bieling et al. next determined how Xklp1 alters the forma-tion of microtubule bundles by PRC1. Previous studies showed that a truncated version of Xklp1 could inhibit both growth and shrinkage of microtubules at particular ends (Bringmann et al., 2004), suggesting that Xklp1 regulates the dynamics of micro-tubules within bundles. Indeed, bundles of antiparallel microtu-bules still formed when Xklp1 and PRC1 were added together to the polymerizing microtubules, but, remarkably, the bundles grew to a fixed size when Xklp1 was present. The authors show that this limit in bundle length is due to cessation of growth at the plus end of the microtubules and that the steady-state length of the overlapping region depends on the concentration of Xklp1 (Figure 1A).

These findings uncover an elegant and simple system capable of self-organizing into a bundle of microtubules with a defined length. Moreover, they show that only two additional proteins, PRC1 and Xklp1, are required for making stable microtubule bundles from highly dynamic polymers, with PRC1 generating the bundles and Xklp1 controlling their length.

To understand how Xklp1 “monitors” and regulates the size of the overlap region between microtubules, Bieling et al. next measured how Xklp1 changes the dynamics of individual microtubules that are not crosslinked. They found that Xklp1

Figure 1. PRC1 Controls Microtubule Assembly(A) Protein regulator of cytokinesis 1 (PRC1) can initiate crosslinking of dynamic microtubules (MT) that interact in an antiparallel fashion (top). Kine-sin-4 is a molecular motor directed at the plus ends of microtubules; it accumulates in the re-gion where microtubules crosslink as their plus ends grow (middle). The interaction of kinesin-4 with PRC1 increases the dwell time of kinesin-4 on microtubules, which in turn limits the length of the overlap region by blocking microtubule growth at the plus ends (bottom) (Bieling et al., 2010).(B) Kinesin-5 is also a molecular motor directed at the plus ends of microtubules, but kinesin-5 can slide microtubules past each other (top). When kinesin-5 is added to crosslinked microtubules (middle), PRC1 maintains the crosslinks despite the sliding action (bottom) (Subramanian et al., 2010).(C) PRC1 forms a homodimer that interacts through its central spectrin domains with two mi-crotubules to crosslink the antiparallel filaments. Although PRC1 is shown to associate predomi-nantly with α-tubulin, the current resolution of the structures presented by Subramanian et al. (2010) is not sufficient to distinguish between binding to α- or β-tubulin.

walczak.indd 365 7/28/2010 2:09:28 PM


alone blocks the growth of microtubules but only at much higher concentrations of Xklp1 than were required to limit the length of microtubule bundles in the presence of PRC1. Biel-ing et al. then demonstrate that the ability of Xklp1 to limit the bundle length of crosslinked microtubules requires the proces-sive motility of Xklp1. This suggests that Xklp1 motility plays an active role in bundle architecture rather than simply acting to regulate microtubule growth.

The key connection, however, came when the authors found that the presence of PRC1 actually increases the time that Xklp1 stays within the overlapping region of two microtubules (i.e., the dwell time). Together with the previous results, this sug-gests a model in which PRC1 forms antiparallel crosslinks of microtubules and recruits Xklp1 within the crosslinked regions, where it walks toward the plus ends of microtubules (Figure 1B). The effective concentration of Xklp1, which depends on the concentration of PRC1, determines the extent to which Xklp1 blocks microtubule growth and thus the steady-state length of the microtubule bundle. Thus, together Xklp1 and PRC1 deter-mine the size and stability of the overlapping zone between microtubules. Control of bundle length is vital because the central spindle must elongate during late anaphase but then shrink to a shorter size to form the midzone during cytokinesis. Changes in PRC1 binding or the activity of Xklp1 during these morphological transitions may provide a mechanism to control the morphology and function of the central spindle.

PRC1 Forms Compliant Crosslinks between MicrotubulesThe above study illustrates how a stable bundle of microtu-bules of a fixed size can form in the presence of two proteins known to function at the midzone in vertebrate cells. However, the work does not explain how PRC1 behaves under conditions where other forces may be acting on the bundles. For example, molecular motors actively slide crosslinked microtubules past each other during the late stages in mitosis.

To characterize how PRC1 and its crosslinking activity modulates or affects the sliding of bundled microtubules, Sub-ramanian et al. (2010) take advantage of elegant microscopy assays they developed in an earlier study to visualize microtu-bule filaments sliding past each other in vitro by the kinesin-5 motors (Kapitein et al., 2005) (Figure 1B). Kinesin-5 proteins are important for establishing the bipolarity of spindles during the early stages of mitosis by actively sliding apart microtubules of opposite polarity. Furthermore there is evidence that these molecular motors slide antiparallel microtubules apart during late anaphase (i.e., anaphase B).

Using this assay, Subramanian et al. now find that PRC1 displays two distinct behaviors in the presence of kinesin-5. In certain cases, the concentration of PRC1 in the overlap region and, thus the bundling length, stay constant while one microtubule slides relative to another filament in the cross-linked region. In the second case, the length of the crosslinked region reduces at a rate similar to that at which the filaments slide past each other. Although in this case PRC1 still tracked the microtubule overlap zone, the crosslinking protein did not reduce the velocity at which filaments slide even when excess PRC1 was present. Together, these results demonstrate that the crosslinks by PRC1 do not significantly resist the sliding

motion of the microtubules by the kinesin-5 motors. This is important because during cytokinesis these crosslinks must remain in place to preserve the organization of the microtu-bules in the midzone. At the same time, the crosslinks must allow the microtubules to slide past each other to achieve and maintain complete segregation of the chromosomes.

PRC1 Has a Spectrin-like Microtubule-Binding DomainHow does PRC1 maintain the structure of midzone bundles while simultaneously allowing the overlap region to adjust to the changing architecture of the central spindle? To answer this question requires a better understanding of the structural and biophysical aspects of how PRC1 bundles microtubules. PRC1 has three prominent domains: the N-terminal domain, which mediates homodimerization; the central domain, which contains the major site for binding to microtubules; and the C-terminal domain, which regulates the interaction with micro-tubules (Figure 1C). Using time-lapse TIRF microscopy, Subra-manian et al. found that PRC1 diffuses one-dimensionally along the microtubule lattice for an average of 7 s, and the C-terminal regulatory domain enhances this association. These findings support the idea that binding of PRC1 to microtubules is medi-ated by both the central microtubule-binding domain and the unstructured C-terminal region. This latter domain contains a large number of positively charged residues (i.e., lysines and arginines), which is a common feature of regions that interact with microtubules.

To gain further insight into how PRC1 interacts with micro-tubules, Subramanian and colleagues determined the X-ray crystal structure of the central microtubule-binding domain of PRC1 and cryo-electron microscopy (cryo-EM) reconstruc-tions of PRC1 fragments bound to microtubules. They found that the central domain of PRC1 consists of a three-helix bun-dle (?70 Å long) with connecting loops between the helices. A cluster of highly conserved and positively charged residues exists at the interface between α helix 1 and α helix 2 within this helical bundle. Mutation of these residues diminished but did not abolish microtubule binding by PRC1, further support-ing the hypothesis that PRC1 possesses two major surfaces that contact microtubules.

Interestingly, this domain shares structural homology with the spectrin domains found in actin-binding proteins (Djinovic-Carugo et al., 2002). Spectrin domains are not required for interaction with actin filaments; instead, they typically link together different functional domains of actin-binding proteins. The present findings by Subramanian and colleagues identify a new role for spectrin domains in regulating the microtubule cytoskeleton.

Cryo-EM reconstructions of microtubules interacting with a truncated fragment of PRC1, which includes the homodimeriza-tion domain and the spectrin domain but not the C-terminal domain, revealed that PRC1 crosslinks nearly all microtubules in an antiparallel manner with a spacing of ?35 nm between filaments (Figure 1C). Notably, the PRC1 molecules bound to crosslinked microtubules were more structured than those bound to a single microtubule. This suggests that crosslink-ing itself converts PRC1 from an inherently flexible molecule to a rigid one, which may enhance PRC1’s crosslinking activity.

walczak.indd 366 7/28/2010 2:09:28 PM


In both cases, the cryo-EM data indicate that PRC1 interacts with one α/β-tubulin heterodimer and extends as a single rod shape almost perpendicular to the microtubule lattice (Figure 1C). Remarkably, the crystal structure of the spectrin domain of PRC1 fits nicely into the cryo-EM density with the conserved basic residues between α helix 1 and α helix 2 residing at the microtubule surface. The binding site of PRC1 on the microtu-bule surface is slightly displaced relative to where many motor proteins interact, providing a possible clue for how motors and crosslinking proteins can bind simultaneously to the same sur-face of the microtubule.

The above studies clearly define the key microtubule-binding element of PRC1, but the single-molecule studies using trun-cated derivatives of PRC1 suggest that PRC1 has a second microtubule-binding domain at its C terminus or this C-termi-nal domain somehow regulates the interaction of the spectrin domain with the microtubule surface. To address this ques-tion, Subramanian and colleagues obtained a second cryo-EM reconstruction of a PRC1 construct containing the spectrin microtubule-binding domain and the C-terminal region. This fragment bound in a similar position on the microtubule lat-tice as the PRC1 construct containing only the dimerization and spectrin domains (Subramanian et al., 2010). Therefore, the most attractive hypothesis is that the C-terminal domain enhances the affinity of PRC1 with microtubules rather than forming a distinct second site for microtubule binding.

Future PerspectivesThe two current papers by Bieling et al. and Subramanian et al. provide critical insight into how inherently dynamic micro-tubules are organized into functional sub-assemblies, which are fundamental to multiple biological systems. It is remark-able that just two proteins are sufficient to reconstitute the morphological subassembly of the spindle midzone. However, it is essential to remember that these proteins do not work in isolation in vivo but rather function in the complex milieu of the central spindle. The finding that PRC1 induces bundles of microtubules that remain compliant to the action of kinesin-5 is key for understanding how the microtubules in the midzone slide apart while still maintaining an organized structure. It will be interesting to add kinesin-5 to the mixture of Xklp1 and PRC1 to see how the system responds to regulators of both microtubule growth and microtubule sliding, a situation that more closely reconstitutes the physiological one.

The work presented here also opens the doors to crucial structure-function and signaling studies on the PRC1 family of proteins. The identification of residues that clearly form the attachment site to microtubules will allow for the engineering of mutations in PRC1 that modulate the strength of its interaction with microtubules. Previous work has shown that phosphoryla-tion regulates PRC1’s interaction with the central spindle (Fu et al., 2007). It is interesting that those phosphorylation sites map to the unstructured C-terminal domain of PRC1 that is positively charged and shown to enhance microtubule binding. Interest-

ingly, another study recently found that phosphorylation of an unstructured region of the kinetochore attachment protein Hec1 also controls its affinity for microtubules (Guimaraes et al., 2008). Thus, phosphorylation of unstructured domains may play a general role in regulating the affinities of microtubule-binding proteins. Finally, understanding how distinct structural modifications of the PRC1 protein affect the morphology of microtubule cytoskeletal arrays in vivo will be an important avenue of future research endeavors.

ACknowledgMents

The authors thank Clive Lloyd, Yixian Zheng, and Stephanie Ems-McClung for helpful discussions. The authors are supported by National Institutes of Health grant GM059618 to C.E.W. and National Science Foundation grant 0920555 to S.L.S.

ReFeRenCes

Bieling, P., Telley, I.A., and Surrey, T. (2010). Cell, this issue.

Bringmann, H., Skiniotis, G., Spilker, A., Kandels-Lewis, S., Vernos, I., and Surrey, T. (2004). Science 303, 1519–1522.

Chan, J., Jensen, C.G., Jensen, L.C., Bush, M., and Lloyd, C.W. (1999). Proc. Natl. Acad. Sci. USA 96, 14931–14936.

Chang-Jie, J., and Sonobe, S. (1993). J. Cell Sci. 105, 891–901.

Djinovic-Carugo, K., Gautel, M., Ylänne, J., and Young, P. (2002). FEBS Lett. 513, 119–123.

Ehrhardt, D.W. (2008). Curr. Opin. Cell Biol. 20, 107–116.

Fu, C., Yan, F., Wu, F., Wu, Q., Whittaker, J., Hu, H., Hu, R., and Yao, X. (2007). Cell Res. 17, 449–457.

Glotzer, M. (2009). Nat. Rev. Mol. Cell Biol. 10, 9–20.

Guimaraes, G.J., Dong, Y., McEwen, B.F., and Deluca, J.G. (2008). Curr. Biol. 18, 1778–1784.

Hornick, J.E., Karanjeet, K., Collins, E.S., and Hinchcliffe, E.H. (2010). Semin. Cell Dev. Biol. 21, 290–299.

Janson, M.E., Loughlin, R., Loïodice, I., Fu, C., Brunner, D., Nédélec, F.J., and Tran, P.T. (2007). Cell 128, 357–368.

Jiang, W., Jimenez, G., Wells, N.J., Hope, T.J., Wahl, G.M., Hunter, T., and Fukunaga, R. (1998). Mol. Cell 2, 877–885.

Kapitein, L.C., Peterman, E.J., Kwok, B.H., Kim, J.H., Kapoor, T.M., and Schmidt, C.F. (2005). Nature 435, 114–118.

Mitchison, T., and Kirschner, M. (1984). Nature 312, 237–242.

Mollinari, C., Kleman, J.P., Jiang, W., Schoehn, G., Hunter, T., and Margolis, R.L. (2002). J. Cell Biol. 157, 1175–1186.

Pellman, D., Bagget, M., Tu, Y.H., Fink, G.R., and Tu, H. (1995). J. Cell Biol. 130, 1373–1385.

Schuyler, S.C., Liu, J.Y., and Pellman, D. (2003). J. Cell Biol. 160, 517–528.

Subramanian, R., Wilson-Kubalek, E.M., Arthur, C.P., Bick, M.J., Campbell, E.A., Darst, S.A., Milligan, R.A., and Kapoor, T.M. (2010). Cell, this issue.

Zhu, C., Lau, E., Schwarzenbacher, R., Bossy-Wetzel, E., and Jiang, W. (2006). Proc. Natl. Acad. Sci. USA 103, 6196–6201.

walczak.indd 367 7/28/2010 2:09:28 PM

Leading Edge

Perspective


IntroductionIn mammalian cells, newly synthesized transcripts are subject to a series of nuclear processing steps to become mature tem-plates for protein synthesis (Moore and Proudfoot, 2009). The 5′ ends of these transcripts acquire a 5′-m7GpppN cap struc-ture (where N is the first transcribed nucleotide) while being elongated by RNA polymerase II. The cap first binds to the cap-binding protein (CBP) heterodimer CBP80-CBP20 (CBC), which supports the pioneer round of mRNA translation (Isken and Maquat, 2008). This round involves the loading of one or more ribosomes, depending on the efficiency of translation ini-tiation and the length of the open translational reading frame (Isken and Maquat, 2008; Isken et al., 2008). The cap subse-quently binds to the eukaryotic translation initiation factor 4E (eIF4E), which directs steady-state rounds of mRNA translation (Isken and Maquat, 2008).

Although CBC-bound mRNAs are precursors to eIF4E-bound mRNAs, the two messenger ribonucleoprotein par-ticles (mRNPs) differ in significant ways (Figure 1). For exam-ple, spliced CBC-bound mRNAs differ from the eIF4E-bound mRNAs that derive from them because they are associated with one or more exon-junction complexes (EJCs) of proteins. By the time eIF4E replaces CBC at the mRNA cap, EJCs are no longer detectable, largely because most reside within the cod-ing region of mRNAs and therefore are displaced by translat-ing ribosomes during the pioneer round (Gehring et al., 2009; Sato and Maquat, 2009). As another example, the poly(A) tails of CBC-bound mRNAs are associated with the mostly nuclear but shuttling poly(A)-binding protein N1 (PABPN1) and the pri-marily cytoplasmic but likewise shuttling PABPC1; in contrast, eIF4E-bound mRNAs do not detectably bind to PABPN1, the replacement of which by PABPC1 is promoted by the pioneer round of translation (Sato and Maquat, 2009).

Despite these and other differences (see below), both CBC-bound and eIF4E-bound mRNAs most likely engage in similar mechanisms of translation initiation, elongation, and termina-tion. Therefore, it is not surprising that both CBC-bound and eIF4E-bound mRNAs use many of the same translation initiation

factors. These factors include not only PABPC1, which data indi-cate is important for activating the translation of both mRNPs, but also eIF4G, eIF3, eIF4B, eIF4A, eIF2 (Figures 1 and 2; Isken and Maquat, 2008), and undoubtedly many other factors that work in conjunction with ribosomes to synthesize proteins. Although both mRNPs support protein synthesis and can be tar-geted for translational activation or repression, the purpose for so doing is distinct: the translation of CBC-bound mRNAs pro-vides a means to control the quality of gene expression; in con-trast, the translation of eIF4E-bound mRNAs generates the bulk of cellular proteins (Isken and Maquat, 2008). Here, we discuss our growing understanding of how cells maintain the specialized functions of each mRNP via associations with particular transla-tion factors, RNA-binding proteins, and signaling targets.

The Pioneer Round of Translation Supports Nonsense-Mediated mRNA DecayIn higher eukaryotes, the vast majority of genes contain multi-ple introns that are removed from the primary transcript by the process of splicing. Splicing may be accompanied by routinely made mistakes so as to result in mRNAs that contain a prema-ture termination codon (McGlincy and Smith, 2008). Premature termination codons can also arise during splicing as a con-sequence of a conditionally regulated process, as exemplified by those pre-mRNAs whose splice site usage is influenced by the encoded RNA-binding protein (McGlincy and Smith, 2008). Nonsense-mediated mRNA decay is a translation-dependent mRNA surveillance pathway that detects and eliminates tran-scripts containing premature termination codons and thus have the potential to be deleterious by virtue of the truncated proteins they encode (see, e.g., Rebbapragada and Lykke-Andersen, 2009).

During the pioneer round of translation, nonsense-mediated mRNA decay is thought to be triggered by the first ribosome that translates newly processed CBC-bound mRNAs and arrives at a premature termination codon (or a normal termi-nation codon) that is situated more than ?50–55 nucleotides upstream of an EJC-bearing exon-exon junction, although

The Pioneer Round of Translation: Features and FunctionsLynne E. Maquat,1,* Woan-Yuh Tarn,1,3 and Olaf Isken2,*1Department of Biochemistry and Biophysics and Center for RNA Biology, School of Medicine and Dentistry, University of Rochester, Rochester, NY 14642, USA2Institute for Virology and Cell Biology, University of Lübeck, 23562 Lübeck, Germany3Institute of Biomedical Sciences, Academia Sinica, Taipei 115, Taiwan*Correspondence: [email protected] (L.E.M.), [email protected] (O.I.)DOI 10.1016/j.cell.2010.07.022

In mammalian cells, newly synthesized mRNAs undergo a pioneer round of translation that is important for mRNA quality control. Following maturation of messenger ribonucleoprotein particles during and after the pioneer round, steady-state cycles of mRNA translation generate most of the cell’s proteins. Translation factors, RNA-binding proteins, and targets of signaling pathways that are particular to newly synthesized mRNAs regulate critical functions of the pioneer round.

maquat.indd 368 7/28/2010 2:09:39 PM


there are exceptions to this rule. The CBC plays a critical role in nonsense-mediated mRNA decay not only because it com-prises the mRNP that harbors EJCs but also because CBP80 interacts directly with the nonsense-mediated mRNA decay factor, up-frameshift 1 (UPF1), enhancing the efficiency of this process (Isken and Maquat, 2008). In short, nonsense-medi-ated mRNA decay is thought to involve recognition of prema-ture termination codons by the SURF complex, which consists of the UPF1 phosphoinositide 3-kinase (PIK)-related protein kinase called SMG1, UPF1, and the two translation termina-tion factors referred to as eukaryotic release factor (eRF)1 and eRF3 (Figure 1; Kashima et al., 2006; Isken and Maquat, 2008). After recognition of the premature termination codon, SMG1

and UPF1 join the EJC that resides downstream of the pre-mature termination codon. Notably, the interaction between CBP80 and UPF1 promotes nonsense-mediated mRNA decay at two sequential steps (Hwang et al., 2010). The first is the association of SMG1-UPF1 with eRF1-eRF3 at a premature termination codon to form the SURF complex. The second is the association of SMG1-UPF1 with a downstream EJC, which results in SMG1-mediated UPF1 phosphorylation. Subse-quently, phosphorylated UPF1 elicits translational repression by binding to the eIF3 constituent of the 43S preinitiation com-plex that is poised at the translation initiation codon (Isken et al., 2008). Phosphorylated UPF1 also promotes the recruitment of mRNA decay activities (Isken et al., 2008). The importance

Figure 1. Pioneer and Steady-State Translation Initiation ComplexesShown is a CBP80-CBP20 (CBC)-bound mRNP from the pioneer round of translation and an eIF4E-bound mRNP from steady-state translation. CBC-bound mRNAs direct pioneer rounds of translation, whereas eIF4E-bound mRNAs, which derive from the remodeling of CBC-bound mRNAs, support the bulk of cel-lular protein synthesis. CBC-bound mRNAs are associated with at least one exon-junction complex (EJC) (provided they are the product of pre-mRNA splicing) and the poly(A)-binding proteins PABPN1 and PABPC1. PYM, which interacts with EJC components and the small 40S ribosomal subunit (not shown), and SKAR (S6 kinase 1 ALY-REF-like target), a component of EJCs, may help to activate the pioneer round of translation. eIF3e, a non-core subunit of the eukaryotic translation initiation factor eIF3, may regulate the translation of a specific set of CBC-bound mRNAs. CTIF (CBP80-CBP20-dependent translation initiation fac-tor) interacts directly with CBP80, as does eIF4G. It is currently unclear whether CTIF is the sole eIF4G-like molecule or if eIF4G also functions during pioneer rounds. eIF4G has been proposed to form a complex with poly(A)-bound PABPC1 to circularize and promote the translation of CBC-bound mRNAs, similar to how poly(A)-bound PABPC1 circularizes and promotes the translation of eIF4E-bound mRNAs. Importin (IMP)-β binds to IMP-α (which is a stable constituent of cap-bound CBC) and augments the translation-independent replacement of CBC by eIF4E. In contrast, the pioneer round of translation promotes the removal of EJCs and of associated RNA-binding proteins such as SF2/ASF (which also activates the pioneer round; not shown) and the replacement of PABPN1 by PABPC1. The insets depict how the SURF complex (comprising the PIK-related protein kinase SMG1, UPF1, eRF1, and eRF3) assembles together with an 80S stalled ribosome at a premature termination codon of CBC-bound mRNA. SMG1 subsequently phosphorylates UPF1 upon UPF1 and SMG1 binding to a downstream EJC during the process of nonsense-mediated mRNA decay. AUG, translation initiation codon; STOP, normal termination codon.

maquat.indd 369 7/28/2010 2:09:42 PM


of the CBC to nonsense-mediated mRNA decay is corrobo-rated by the finding that decay is still restricted to CBC-bound mRNAs in the case of mRNAs containing premature termina-tion codons that initiate translation in a cap-independent mode from an internal ribosome entry site (IRES), provided that IRES function depends on eIF3 (Isken and Maquat, 2008).

CBC-bound mRNAs form polysomes that are smaller than the polysomes associated with their eIF4E-bound counterparts, consistent with the replacement of CBC by eIF4E taking place on polysomes during the pioneer round of translation. This finding, and data demonstrating that nonsense-mediated mRNA decay involves a step of translational repression that targets eIF3, indi-cate that CBC-bound mRNAs, like eIF4E-bound mRNAs, are generally loaded with more than one ribosome, eIF3, and other well-characterized translation initiation factors.

Specialized Factors of the Pioneer Round of TranslationThe process of translation can be divided into four phases—initiation, elongation, termination, and ribosome recycling. The majority of regulation occurs during the initiation phase (Sonenberg and Hinnebusch, 2009). There is considerably more known about the translation of eIF4E-bound mRNAs than the translation of CBC-bound mRNAs, which is a relatively recent discovery. Nevertheless, there are likely to be many more similarities than differences between the translation of the two mRNPs, as available data have borne out. Factors that specifically regulate the pioneer round of translation will likely target, directly or via another protein, the CBC, the EJC, PABPN1, or other constituents that do not typify eIF4E-bound mRNA, just as factors that specifically regulate steady-state translation often target eIF4E.

Figure 2. Signaling Pathways and TranslationShown are the signaling pathways that regulate the pioneer round of translation (top) and subsequent steady-state translation (bottom). Under condi-tions that activate the mTOR signaling pathway, mTORC1 becomes activated and may bind to the translation initiation factor eIF3 associated with CBP80-CBP20 (CBC)-bound mRNA resulting in the phosphorylation and dissociation of activated S6 kinase 1 (S6K1). The EJC component SKAR (S6 kinase 1 ALY-REF-like target) then recruits phosphorylated (i.e., activated) S6K1. Activated S6K1 next mediates the phosphorylation of SKAR and other downstream effectors, including possibly eIF4B and PDCD4 (programmed cell death factor 4), that also associate with CBC-bound mRNA and thereby promote the pioneer round of translation for spliced mRNAs. The Cdc42-dependent phosphorylation of S6K1 also may promote the binding of CBC to cap structures, which would further stimulate the pioneer round. Another CBC-bound mRNA constituent, SF2/ASF, binds to the RNA export receptor TAP and also triggers S6K1 signaling via mTORC1 in ways that promote CBC-bound mRNA translation. Moreover, PYM interacts with the EJC core proteins Y14-MAGOH and with the 40S ribosomal subunit to somehow promote the translation of spliced mRNAs presumably during and certainly beyond the pioneer round. (Left) The boxes summarize the translational regulation of CBC-bound or eIF4E-bound mRNAs under various cell-growth conditions. Under conditions of heat shock, hypoxia, or serum starvation, the pioneer round of translation is favored over steady-state translation; steady-state translation is largely inhibited by phosphorylation of eIF2α (not shown) or by binding of 4E-BP1 to eIF4E.

maquat.indd 370 7/28/2010 2:09:43 PM


PYMCompared to nonspliced mRNA, it is well established that the process of pre-mRNA splicing endows spliced mRNA with a higher translational capacity during both pioneer and steady-state rounds of translation (Moore and Proudfoot, 2009). The abil-ity of splicing to promote translation appears to be mediated at least in part by EJCs, which data indicate are dynamic rather than either static or homogenous complexes. Different EJC constitu-ents may augment different steps of translation initiation (Lee et al., 2009). For example, PYM may move between mRNA-bound EJCs and translationally active ribosomes: PYM can interact with the EJC core proteins Y14-MAGOH and, via a separate domain, with the 40S ribosomal subunit and the 48S preinitiation complex to somehow promote the translation of spliced mRNAs during and beyond the pioneer round (Diem et al., 2007). The finding that PYM detectably coimmunoprecipitates with CBP80 but not eIF4E not only is consistent with the idea that PYM is present during, and gone after, the pioneer round of translation but also raises the possibility that PYM constitutes a fraction of ribosomes that may specifically function during pioneer rounds. For example, by coupling ribosome recruitment to EJC disassembly, PYM may ensure the preferential recruitment of newly synthesized CBC-bound mRNAs over eIF4E-bound mRNAs to the translational machinery. Alternatively, translation factors that are unique to CBC-bound mRNAs and/or EJCs could recruit ribosomes bound by PYM for the pioneer round. Along these lines, PYM is recruited to those newly exported mRNAs of Kaposi’s sarcoma-associated herpesvirus that are intronless and thus devoid of EJCs, by the viral open reading frame 57 protein (ORF57). ORF57 interacts with PYM and enhances viral mRNA translation via the PYM-mediated recruitment of preinitiation complexes (Boyne et al., 2010). Nota-bly, the recent report that PYM is an EJC disassembly factor that inhibits nonsense-mediated mRNA decay when overexpressed, i.e., when a fraction of PYM exists free of an association with 40S ribosomal subunits so as to promiscuously dissociate EJCs (Geh-ring et al., 2009), is compatible with the idea that EJC constituents can activate translation.SKARThe EJC constituent called SKAR, for S6 kinase 1 (S6K1) ALY-REF-like target, may also mark spliced CBC-bound mRNAs with a translational status that is distinct from the translational status of eIF4E-bound mRNAs and provides another example of how EJCs positively regulate the pioneer round of transla-tion. The mammalian target of the rapamycin (mTOR) signaling pathway promotes cellular translation depending on the pres-ence of growth factors, the absence of stress, and the availabil-ity of nutrients or other energy sources (Ma and Blenis, 2009). mTOR is a PIK-related protein kinase that, together with raptor and LST8, forms the mTOR complex 1 (mTORC1). Interestingly, signaling through mTORC1 modulates the translation of both CBC-bound mRNAs and eIF4E-bound mRNAs by distinct but undoubtedly overlapping mechanisms (Figure 2).

In the case of eIF4E-bound mRNAs, the recruitment of acti-vated mTORC1 by eIF3 into the proximity of the eIF4E-binding protein 1 (4E-BP1) activates translation by mediating the phos-phorylation of 4E-BP1 and S6K1. 4E-BP1 phosphorylation results in 4E-BP1 dissociation from mRNA-bound eIF4E, which activates translation by allowing eIF4G to interact with and thereby stabilize

the association of eIF4E, eIF4A, and PABPC1 with mRNA. S6K1 phosphorylation results in the dissociation of activated S6K1 from mRNA-bound eIF3. Released S6K1 promotes scanning of the 43S preinitiation complex to the AUG translation initiation codon and, therefore, translation by phosphorylating both the eIF4A activator eIF4B and the eIF4A inhibitor and tumor suppressor programmed cell death 4 (PDCD4).

In the case of CBC-bound mRNAs, activated mTORC1 may be recruited by eIF3 and, additionally, SF2/ASF (see below), promot-ing the phosphorylation-induced activation of S6K1 and then the dissociation of activated phosphorylated S6K1. EJC-bound SKAR then recruits activated S6K1 to CBC-bound spliced mRNAs, and SKAR-bound S6K1 promotes the pioneer round of translation by allowing for the phosphorylation of SKAR and other downstream effectors that could include eIF4B and PDCD4, which also asso-ciate with CBC-bound mRNAs (Ma et al., 2008). The finding that SKAR regulates cell growth (Richardson et al., 2004) raises the interesting question of how the mTORC1-induced translation of all or a subset of CBC-bound spliced mRNAs is transduced to a functionally significant increase in total-cell protein synthesis.SF2/ASFThe serine-arginine-rich protein SF2/ASF plays a central role in recruiting the splicing machinery to pre-mRNA splice sites and can either enhance or inhibit splicing. Like PYM, SF2/ASF coim-munoprecipitates with CBP80 but not detectably with eIF4E (Sato et al., 2008). Data indicate that SF2/ASF, when bound to exonic sequences, can travel as a constituent of CBC-bound mRNPs from nuclei into the cytoplasm, where it has the ability to promote the pioneer round of translation (Sato et al., 2008). Evidence indicates that SF2/ASF recruits a number of transla-tional activators to mRNAs, including the RNA export receptor TAP (Sato et al., 2008) and mTORC1 (Michlewski et al., 2008). Also, like PYM, SF2/ASF promotes mRNA translation beyond the pioneer round (Sato et al., 2008; Michlewski et al., 2008).CBP80Another effector of the pioneer round could be CBP80 as the binding of CBC to cap structures is stimulated by growth fac-tors during the G1/S phase of the cell cycle, activated forms of the Ras proto-oncogene protein, and the Rho-GTPase Cdc42. Cdc42 has been shown to transduce extracellular signals from G-coupled protein receptors, integrins, and growth factor receptors at least in part by binding to S6K1 in a GTP-depen-dent manner. Binding facilitates recognition of the Cdc42-S6K1 complex by upstream S6K-activating kinases (Chou and Blenis, 1996; Wilson and Cerione, 2000). Moreover, signaling from Cdc42 to CBP80 is transmitted in part by the S6K-cata-lyzed phosphorylation of CBP80, although the biological con-sequence remains unknown (Ma and Blenis, 2009).

As exemplified above, it appears that an efficient pioneer round of translation is followed by efficient steady-state cycles of translation. Growth factors may augment cellu-lar protein synthesis not only by promoting CBC binding to caps but also by expediting the replacement of CBC by eIF4E (see below). Despite not yet knowing mechanistic details, it is clear that activating or inhibiting pioneer rounds of translation depending on cellular growth conditions is likely to provide an important checkpoint mechanism for the bulk of cellular protein synthesis.

maquat.indd 371 7/28/2010 2:09:44 PM


Importin-αRegulated binding of CBC to mRNA caps is additionally influ-enced by importin-α, which stably associates with cap-bound CBC (Sato and Maquat, 2009). Cytoplasmic importin-β inter-acts directly with importin-α and CBP20, thereby promoting the replacement of CBC by eIF4E at mRNA caps (Figure 1; Dias et al., 2009; Sato and Maquat, 2009). In view of the recent structure of an importin-α-CBC-cap complex and a structural model of the importin-β-importin-α-CBC complex (Dias et al., 2009), it will be possible to study how signaling pathways, and the resulting posttranslational modifications of importin-α, importin-β, and/or CBC, might regulate importin-α or importin-β binding to cap-associated CBC to either stabilize or disrupt the interaction of CBC with mRNA caps. When considering the critical role of CBC in assuring the quality of gene expression through nonsense-mediated mRNA decay, and that CBC can be replaced by eIF4E even when the pioneer round of translation is inhibited (Sato and Maquat, 2009), it is important that timing the pioneer round rela-tive to the replacement of CBC by eIF4E be coordinated so that the pioneer round of translation largely takes place first.eIF3eMammalian eIF3, which is a large protein complex consisting of ?13 subunits, has multiple functions in stimulating translation initiation (Sonenberg and Hinnebusch, 2009). The eIF3 subunit eIF3e (also called p48/INT6) coimmunoprecipitates with CBP80 as well as the EJC constituent and nonsense-mediated mRNA decay factor UPF2, but it does not detectably coimmunoprecipi-tate with eIF4E and, like all eIF3 subunits that have been tested, is essential for nonsense-mediated mRNA decay (Morris et al., 2007; Isken et al., 2008). eIF3e, which is a core subunit of mam-malian eIF3 that is not present in all species, appears to contribute to the recruitment of ribosome-bound eIF3 to mRNA by directly associating with eIF4G (Sonenberg and Hinnebusch, 2009). How-ever, at least in fission yeast, eIF3e is not critical for global pro-tein synthesis, and there exist distinct eIF3 complexes that differ in their non-core subunits. Interestingly, those complexes that contain eIF3e associate with a restricted set of mRNAs, some of which are implicated in the response to cell stress (Zhou et al., 2005). Therefore, it is possible that a specialized eIF3e-containing eIF3 complex, if present in mammals, could specifically function during pioneer rounds of translation by recruiting a defined set of newly synthesized mRNAs to the translational machinery. By so doing, this specialized complex could route those mRNAs that are targets of nonsense-mediated mRNA decay to this pathway and activate the steady-state translation of those that are not non-sense-mediated mRNA decay targets.CTIFAnother initiation factor that is central to steady-state transla-tion is eIF4G, which binds directly to cap-associated eIF4E and serves critical roles as a scaffold for eIF3, PABPC1, and eIF4A. eIF4G, which also binds directly to CBC, is thought to function in a similar fashion during pioneer rounds of translation. Recently, an eIF4G-like protein called CBP80-CBP20-dependent transla-tion initiation factor (CTIF) has been implicated specifically during pioneer rounds. CTIF interacts directly with CBP80 as well as pre-mRNAs, coimmunoprecipitates with eIF3 in an RNase-resistant manner, and functions in the translation of CBC-bound mRNAs; however, it plays no detectable role in the translation of eIF4E-

bound mRNAs (Kim et al., 2009). CTIF could promote the recruit-ment of ribosomes to newly synthesized CBC-bound mRNAs. As one possibility, CTIF, which contains a middle domain of eIF4G (MIF4G) but lacks the eIF4E-binding domain of eIF4G, may func-tionally replace eIF4G during pioneer rounds if, for example, it (possibly together with CBP80, which contains three MIF4Gs) serves as an eIF4G-like scaffold for CBC-bound mRNPs. Alter-natively, CTIF may function in addition to eIF4G considering that CTIF lacks the PABPC1- and eIF4A-binding domains that typify eIF4G and that another MIF4G-containing protein, called SLIP1, interacts directly with eIF4G to promote the translation of histone mRNA (Cakmakci et al., 2008). Moreover, data indicate that eIF4G functions during pioneer rounds of translation. Regardless, as is possible for eIF3, the use of translation initiation factors that are unique to and activate pioneer rounds could be advantageous by shortening the time span between transcriptional activation and the start of protein production.

The Pioneer Round during Cell StressThe bulk of cellular cap-dependent translation is rapidly downreg-ulated by signaling pathways in response to most physiological and environmental stressors as part of a repertoire of events that mitigate cellular damage and promote cell survival (Figure 2). This global downregulation during, for example, viral infection, amino acid starvation, hypoxia, or heat shock is often accompanied by eIF2α phosphorylation. eIF2α phosphorylation decreases the translation of the majority of CBC-bound as well as eIF4E-bound mRNAs by limiting ternary complex abundance while mediating the selective translation of a subset of cellular mRNAs that ini-tiate translation in a mechanism that depends on an upstream open reading frame. These mRNAs encode specific transcription factors that help the cell adapt to stress and ultimately restore translation (Holcik and Sonenberg, 2005). In addition to eIF2α phosphorylation, stress-induced proteolysis of translation initia-tion factors can also compromise cellular translation.

There are variations to this theme that result in the differen-tial regulation of pioneer rounds of translation relative to sub-sequent rounds of translation. For example, the preferential translation of CBC-bound mRNAs is maintained during serum starvation and heat shock, when the translation of eIF4E-bound mRNAs is compromised (Oh et al., 2007b; Marín-Vinader et al., 2006). Serum starvation does not result in eIF2α phosphoryla-tion but instead activates 4E-BP1.

As another example, during early phases of hypoxia, activa-tion of the endoplasmic reticulum kinase PERK leads to eIF2α phosphorylation and the translational inhibition of both CBC-bound and eIF4E-bound mRNAs (Oh et al., 2007a). However, during late stages of hypoxia, eIF2α dephosphorylation allows for the resumption of CBC-bound mRNA translation (Oh et al., 2007a). Concomitantly, in an eIF2α-independent pathway, dis-ruption of eIF4E-eIF4G-eIF4A and sequestration of eIF4E by dephosphorylated 4E-BP1 and the eIF4E-transporter (which moves eIF4E into the nucleus and processing bodies) maintain the repression of eIF4E-bound mRNA translation (Koritzinsky et al., 2006; Lee et al., 2008). Conceivably, the restoration of CBC-bound mRNA translation as a first step toward transla-tional normalcy promotes cell survival after stress by allowing for the surveillance of newly synthesized mRNAs. Furthermore,

maquat.indd 372 7/28/2010 2:09:44 PM


the renewed engagement of newly made mRNAs with the translational machinery could speed up the cellular response after stress.

Spatial Regulation of the Pioneer Round of TranslationProper embryonic patterning and development as well as neu-ronal function often involve proteins that bind to the 3′ untrans-lated regions (3′ UTRs) of particular mRNAs in a sequence-spe-cific manner to couple mRNA translational activation and mRNA localization (Sonenberg and Hinnebusch, 2009). By so doing, an expression gradient of the encoded morphogens, in the case of oocytes, or of proteins that maintain synaptic function or plastic-ity, in the case of neurons, is spatially established.

There have been many reports of translational repression occurring prior to mRNA localization so as to inhibit ectopic protein production from unlocalized mRNAs. If the sole mecha-nism of translational repression targets eIF4E, then an mRNA should have undergone the pioneer round of translation prior to repression although, as noted above, it is possible for eIF4E to replace CBC without a pioneer round. In theory, other mech-anisms of repression might target both CBC-bound mRNA and eIF4E-bound mRNA.

Examples of translational repression include sequestrating mRNAs into translationally silenced particles, as Bruno does for oskar mRNA in Drosophila; recruiting the CCR4 deadeny-lase to shorten poly(A) tail length, as Bicaudal-C does for its own mRNA and certain other germline mRNAs in Drosophila; and blocked joining of the 60S ribosomal subunit to the 48S preinitiation complex, as Zipcode-binding protein 1 (ZBP1, also called IMP1) does for β-actin mRNA at the leading lamel-lipodia of fibroblasts or neurite growth cones (Besse and Ephrussi, 2008; Sonenberg and Hinnebusch, 2009). Notably, the findings that ZBP1 associates cotranscriptionally with β-actin mRNA (Besse and Ephrussi, 2008), and RNP granules undergoing ZBP1-mediated transport contain CBP80 and EJC constituents but lack detectable eIF4E (Jønson et al., 2007), suggest that ZBP1-mediated translational repression targets CBC-bound transcripts.

To date, studies of arc mRNA and other specific mRNAs in mammalian neurons provide the best examples of translational repression that targets CBC-bound mRNAs (Giorgi et al., 2007). arc mRNA, which harbors two 3′ UTR introns, is a natural tar-get of nonsense-mediated mRNA decay: When the pioneer round of translation terminates normally, arc mRNA undergoes nonsense-mediated mRNA decay. The arc gene is transcrip-tionally induced upon intense synaptic activation, and the resulting mRNA is translated and thus targeted for nonsense-mediated mRNA decay, once it localizes to activated synapses in a mechanism that depends on glutamatergic receptors. By essentially limiting ARC protein synthesis to the pioneer round of translation at activated synapses, improper protein synthe-sis at inopportune times and cellular locations is prevented. These findings are consistent with data indicating that CBP80-bound mRNAs migrate to spines when glutamatergic recep-tors of rat dendrites are stimulated (di Penta et al., 2009) and thus have yet to undergo the pioneer round of translation. At present, which signaling factors control these pioneer rounds of translation in neurons remain unknown.

Features of the pioneer translation initiation complex also appear to be important for proper localization of oskar mRNA to the posterior pole in Drosophila oocytes. For example, splic-ing of the first intron of oskar pre-mRNA (or another intron at the site of the first intron) is required for oskar mRNA localiza-tion, presumably by virtue of the resulting EJC in conjunction with nearby mRNA sequences (Hatchet and Ephrussi, 2004). In fact, demonstrated roles for the EJC constituents eIF4AIII, Tsunami-Magonashi, and Barentz (the latter two of which are orthologous to mammalian Y14-MAGOH and MLN51, respectively) in oskar mRNA localization and their colocalization with oskar mRNA at the posterior pole (Besse and Ephrussi, 2008) indicate that the pioneer round of translation and removal of the first EJC occur after the regulatory steps of translational repression and local-ization. As noted above, these regulatory steps may involve the RNA-binding protein Bruno. As oskar mRNA is also translation-ally repressed by Bruno-mediated recruitment of the Cup protein, which targets eIF4E (Besse and Ephrussi, 2008), there appear to be multiple pathways to translationally silence oskar mRNA. In fact, a number of mRNAs are repressed at different steps of translation when bound by CBC or eIF4E.

MicroRNAsIt was recently reported that the microRNA (miRNA) miR-2 medi-ates translational repression, which interferes with the interac-tion between eIF4E and eIF4G in a way that is predicted to have no consequence to the pioneer round of translation (Zdanow-icz et al., 2009). Therefore, mRNA caps bound by eIF4E could represent another signature target used by cells to differentially regulate the translation of CBC-bound mRNA relative to eIF4E-bound mRNA. Nevertheless, other mechanisms of miRNA-mediated translational repression can function simultaneously to repress the translation of CBC-bound mRNA. This became evident through the recent demonstration that (1) Ago2, which is known to inhibit the translation of its target mRNAs, is loaded onto both CBC-bound mRNAs and eIF4E-bound mRNAs, (2) the loading of endogenous microRNA-induced silencing com-plexes (miRISCs) onto the 3′ UTRs of CBC-bound mRNAs that contain a premature termination codon abrogates nonsense-mediated mRNA decay, and (3) several natural substrates of nonsense-mediated mRNA decay are stabilized by miRISC-mediated translational repression (Choe et al., 2010). It will be interesting to determine if factors that are specific for CBC-bound mRNA and the pioneer round of translation are recog-nized during miRNA-mediated translational repression.

Future DirectionsMammalian cells have evolved two mRNP templates for protein synthesis that manifest a precursor-product relationship and serve distinct functions. Although these two templates share many constituent proteins that homogenize their regulation, at least some unique features impart a means for their differential control. Examples also exist of distinct regulatory mechanisms that coordinate CBC-bound mRNA translation and eIF4E-bound mRNA translation by targeting factors that are particular to each mRNP. As eIF4E-bound mRNA supports the bulk of cellular protein synthesis, it is the logical target for a rapid response to changes in physiological conditions. However, CBC-bound mRNA holds

maquat.indd 373 7/28/2010 2:09:44 PM


the unique capability of enhancing subsequent rounds of transla-tion, downregulating further translation, or completely stopping not only CBC-supported mRNA surveillance but also eIF4E-supported protein production. Future studies are expected to illuminate molecular aspects of known and unforeseen regulatory mechanisms.

ACknowledgmenTS

We thank Chenguang Gong and Hanae Sato for helpful comments and ac-knowledge NIH R01 grants GM074593 and GM59614 (L.E.M.), the National Science Council of Taiwan (W.-Y.T.), and start-up funds from the University of Lübeck (O.I.) for support.

ReFeRenCeS

Besse, F., and Ephrussi, A. (2008). Translational control of localized mRNAs: re-stricting protein sythesis in space and time. Nat. Rev. Mol. Cell Biol. 9, 971–980.

Boyne, J.R., Jackson, B.R., Taylor, A., Macnab, S.A., and Whitehouse, A. (2010). Kaposi’s sarcoma-associated herpesvirus ORF57 protein interacts with PYM to enhance translation of viral intronless mRNAs. EMBO J. 29, 1851–1864.

Cakmakci, N.G., Lerner, R.S., Wagner, E.J., Zheng, L., and Marzluff, W.F. (2008). SLIP1, a factor required for activation of histone mRNA translation by the stem-loop binding protein. Mol. Cell. Biol. 28, 1182–1194.

Choe, J., Cho, H., Lee, H.C., and Kim, Y.K. (2010). microRNA/Argonaute 2 regu-lates nonsense-mediated messenger RNA decay. EMBO Rep. 11, 380–386.

Chou, M.M., and Blenis, J. (1996). The 70S kDa S6 kinase complexes with and is activated by the Rho family of G proteins Cdc42 and Rac1. Cell 85, 573–583.

Dias, S.M., Wilson, K.F., Rojas, K.S., Ambrosio, A.L., and Cerione, R.A. (2009). The molecular basis for the regulation of the cap-binding complex by the import-ins. Nat. Struct. Mol. Biol. 16, 930–937.

di Penta, A., Mercaldo, V., Florenzano, F., Munck, S., Ciotti, M.T., Zalfa, F., Mer-canti, D., Molinari, M., Bagni, C., and Achsel, T. (2009). Dendritic LSm1/CBP80-mRNPs mark the early steps of transport commitment and translational control. J. Cell Biol. 184, 423–435.

Diem, M.D., Chan, C.C., Younis, I., and Dreyfuss, G. (2007). PYM binds the cyto-plasmic exon-junction complex and ribosomes to enhance translation of spliced mRNAs. Nat. Struct. Mol. Biol. 14, 1173–1179.

Gehring, N.H., Lamprinaki, S., Kulozik, A.E., and Hentze, M.W. (2009). Disassem-bly of exon junction complexes by PYM. Cell 137, 536–548.

Giorgi, C., Yeo, G.W., Stone, M.E., Katz, D.B., Burge, C., Turrigiano, G., and Moore, M.J. (2007). The EJC factor eIF4AIII modulates synaptic strength and neuronal protein expression. Cell 130, 179–191.

Hatchet, O., and Ephrussi, A. (2004). Splicing of oskar RNA in the nucleus is coupled to its cytoplasmic localization. Nature 428, 959–963.

Holcik, M., and Sonenberg, N. (2005). Translational control in stress and apopto-sis. Nat. Rev. Mol. Cell Biol. 6, 318–327.

Hwang, J., Sato, H., Tang, Y., Matsuda, D., and Maquat, L.E. (2010). UPF1 as-sociation with the cap-binding protein, CBP80, promotes nonsense-mediated mRNA decay at two distinct steps. Mol. Cell. Published online August 5, 2010. 10.1016/j.molcel.2010.07.004.

Isken, O., and Maquat, L.E. (2008). The multiple lives of NMD factors: balancing roles in gene and genome regulation. Nat. Rev. Genet. 9, 699–712.

Isken, O., Kim, Y.K., Hosoda, N., Mayeur, G.L., Hershey, J.W., and Maquat, L.E. (2008). Upf1 phosphorylation triggers translational repression during nonsense-mediated mRNA decay. Cell 133, 314–327.

Jønson, L., Vikesaa, J., Krogh, A., Nielsen, L.K., Hansen, T.O., Borup, R., Johns-en, A.H., Christiansen, J., and Nielsen, F.C. (2007). Molecular composition of IMP1 ribonucleoprotein granules. Mol. Cell. Proteomics 6, 798–811.

Kashima, I., Yamashita, A., Izumi, N., Kataoka, N., Morishita, R., Hoshino, S., Ohno, M., Dreyfuss, G., and Ohno, S. (2006). Binding of a novel SMG-1-Upf1-

eRF1-eRF3 complex (SURF) to the exon junction complex triggers Upf1 phos-phorylation and nonsense-mediated mRNA decay. Genes Dev. 20, 355–367.

Kim, K.M., Cho, H., Choi, K., Kim, J., Kim, B.W., Ko, Y.G., Jang, S.K., and Kim, Y.K. (2009). A new MIF4G domain-containing protein, CTIF, directs nuclear cap-binding protein CBP80/20-dependent translation. Genes Dev. 23, 2033–2045.

Koritzinsky, M., Magagnin, M.G., van den Beucken, T., Seigneuric, R., Savelkouls, K., Dostie, J., Pyronnet, S., Kaufman, R.J., Weppler, S.A., Voncken, J.W., et al. (2006). Gene expression during acute and prolonged hypoxia is regulated by dis-tinct mechanisms of translational control. EMBO J. 25, 1114–1125.

Lee, H.C., Cho, H., and Kim, Y.K. (2008). Ectopic expression of eIF4E-transporter triggers the movement of eIF4E into P-bodies, inhibiting steady-state translation but not the pioneer round of translation. Biochem. Biophys. Res. Commun. 369, 1160–1165.

Lee, H.C., Choe, J., Chi, S.G., and Kim, Y.K. (2009). Exon junction complex en-hances translation of spliced mRNAs at multiple steps. Biochem. Biophys. Res. Commun. 384, 334–340.

Ma, X.M., and Blenis, J. (2009). Molecular mechanisms of mTOR-mediated trans-lational control. Nat. Rev. Mol. Cell Biol. 10, 307–318.

Ma, X.M., Yoon, S.O., Richardson, C.J., Jülich, K., and Blenis, J. (2008). SKAR links pre-mRNA splicing to mTOR/S6K1-mediated enhanced translation efficien-cy of spliced mRNAs. Cell 133, 303–313.

Marín-Vinader, L., van Genesen, S.T., and Lubsen, N.H. (2006). mRNA made dur-ing heat shock enters the first round of translation. Biochim. Biophys. Acta 1759, 535–542.

McGlincy, N.J., and Smith, C.W. (2008). Alternative splicing resulting in nonsense-mediated mRNA decay: what is the meaning of nonsense? Trends Biochem. Sci. 3, 385–393.

Michlewski, G., Sanford, J.R., and Cáceres, J.F. (2008). The splicing factor SF2/ASF regulates translation initiation by enhancing phosphorylation of 4E-BP1. Mol. Cell 30, 179–189.

Moore, M.J., and Proudfoot, N.J. (2009). Pre-mRNA processing reaches back to transcription and ahead to translation. Cell 136, 688–700.

Morris, C., Wittmann, J., Jäck, M., and Jalinot, P. (2007). Human INT6/eIF3e is required for nonsense-mediated mRNA decay. EMBO Rep. 8, 596–602.

Oh, N., Kim, K.M., Choe, J., and Kim, Y.K. (2007a). Pioneer round of translation mediated by nuclear cap-binding proteins CBP80/20 occurs during prolonged hypoxia. FEBS Lett. 581, 5158–5164.

Oh, N., Kim, K.M., Choe, J., and Kim, Y.K. (2007b). Pioneer round of transla-tion occurs during serum starvation. Biochem. Biophys. Res. Commun. 362, 145–151.

Rebbapragada, I., and Lykke-Andersen, J. (2009). Execution of nonsense-medi-ated mRNA decay: what defines a substrate? Curr. Opin. Cell Biol. 21, 394–402.

Richardson, C.J., Bröenstrup, M., Fingar, D.C., Jülich, K., Ballif, B.A., Gygi, S., and Blenis, J. (2004). SKAR is a specific target of S6 kinase 1 in cell growth con-trol. Curr. Biol. 14, 1540–1549.

Sato, H., and Maquat, L.E. (2009). Remodeling of the pioneer translation initia-tion complex involves translation and the karyopherin importin β. Genes Dev. 23, 2537–2550.

Sato, H., Hosoda, N., and Maquat, L.E. (2008). Efficiency of the pioneer round of translation affects the cellular site of nonsense-mediated mRNA decay. Mol. Cell 29, 255–262.

Sonenberg, N., and Hinnebusch, A.G. (2009). Regulation of translation initiation in eukaryotes: Mechanisms and biological targets. Cell 136, 731–745.

Wilson, K.F., and Cerione, R.A. (2000). Signal transduction and post-transcrip-tional gene expression. Biol. Chem. 381, 357–365.

Zdanowicz, A., Thermann, R., Kowalska, J., Jemielity, J., Duncan, K., Preiss, T., Darzynkiewicz, E., and Hentze, M.W. (2009). Drosophila miR2 primarily targets the m7GpppN cap structure for translational repression. Mol. Cell 35, 881–888.

Zhou, C., Arslan, F., Wee, S., Krishnan, S., Ivanov, A.R., Oliva, A., Leatherwood, J., and Wolf, D.A. (2005). PCI proteins eIF3e and eIF3m define distinct translation initiation factor 3 complexes. BMC Biol. 3, 14.

maquat.indd 374 7/28/2010 2:09:44 PM

Direct Reprogramming of Fibroblastsinto Functional Cardiomyocytesby Defined FactorsMasaki Ieda,1,2,3,6,* Ji-Dong Fu,1,2,3 Paul Delgado-Olguin,1,2,4 Vasanth Vedantham,1,5 Yohei Hayashi,1

Benoit G. Bruneau,1,2,4 and Deepak Srivastava1,2,3,*1Gladstone Institute of Cardiovascular Disease2Department of Pediatrics3Department of Biochemistry and Biophysics4Cardiovascular Research Institute5Department of Medicine

University of California, San Francisco, San Francisco, CA 94158, USA6Present address: Departments of Cardiology and of Clinical and Molecular Cardiovascular Research, Keio University School of Medicine,

Shinanomachi 35, Shinjuku-ku, Tokyo 160-8582, Japan

*Correspondence: [email protected] (M.I.), [email protected] (D.S.)

DOI 10.1016/j.cell.2010.07.002

SUMMARY

The reprogramming of fibroblasts to induced plurip-otent stem cells (iPSCs) raises the possibility thata somatic cell could be reprogrammed to an alterna-tive differentiated fate without first becoming a stem/progenitor cell. A large pool of fibroblasts exists inthe postnatal heart, yet no single ‘‘master regulator’’of direct cardiac reprogramming has been identified.Here, we report that a combination of three develop-mental transcription factors (i.e., Gata4, Mef2c, andTbx5) rapidly and efficiently reprogrammed post-natal cardiac or dermal fibroblasts directly into differ-entiated cardiomyocyte-like cells. Induced cardio-myocytes expressed cardiac-specific markers, hada global gene expression profile similar to cardio-myocytes, and contracted spontaneously. Fibro-blasts transplanted into mouse hearts one day aftertransduction of the three factors also differentiatedinto cardiomyocyte-like cells. We believe these find-ings demonstrate that functional cardiomyocytescan be directly reprogrammed from differentiatedsomatic cells by defined factors. Reprogrammingof endogenous or explanted fibroblasts might pro-vide a source of cardiomyocytes for regenerativeapproaches.

INTRODUCTION

Heart disease is a leading cause of adult and childhood mortality.

The underlying pathology is typically loss of cardiomyocytes that

leads to heart failure or improper development of cardiomyo-

cytes during embryogenesis that leads to congenital heart

malformations. Because postnatal cardiomyocytes have little

or no regenerative capacity, current therapeutic approaches

are limited. Embryonic stem cells possess clear cardiogenic

potential, but efficiency of cardiac differentiation, risk of tumor

formation, and issues of cellular rejection must be overcome

(Ivey and Srivastava, 2006; Laflamme et al., 2007; Nussbaum

et al., 2007; van Laake et al., 2008). The ability to reprogram

fibroblasts into induced pluripotent stem cells (iPSCs) with four

defined factors might address some of these issues by providing

an alternative source of embryonic-like stem cells (Takahashi

and Yamanaka, 2006). However, generating sufficient iPSC-

derived cardiomyocytes that are pure and mature and that can

be delivered safely remains challenging (Zhang et al., 2009).

The human heart is composed of cardiomyocytes, vascular

cells, and cardiac fibroblasts. In fact, cardiac fibroblasts com-

prise over 50% of all the cells in the heart (Baudino et al.,

2006; Camelliti et al., 2005; Snider et al., 2009). Cardiac fibro-

blasts are fully differentiated somatic cells that provide support

structure, secrete signals, and contribute to scar formation

upon cardiac damage (Ieda et al., 2009). Fibroblasts arise from

an extracardiac source of cells known as the proepicardium,

and do not normally have cardiogenic potential (Snider et al.,

2009). The large population of endogenous cardiac fibroblasts

is a potential source of cardiomyocytes for regenerative therapy

if it were possible to directly reprogram the resident fibroblasts

into beating cardiomyocytes. Unfortunately, although embryonic

mesoderm can be induced to differentiate into cardiomyocytes

(Takeuchi and Bruneau, 2009), efforts to accomplish this in

somatic cells have thus far been unsuccessful, and to our knowl-

edge, no ‘‘master regulator’’ of cardiac differentiation, like MyoD

for skeletal muscle (Davis et al., 1987), has been identified

to date.

The generation of iPSCs suggests that a specific combination

of defined factors, rather than a single factor, could epigeneti-

cally alter the global gene expression of a cell and allow greater

plasticity of cell type than previously appreciated. Consistent

Cell 142, 375–386, August 6, 2010 ª2010 Elsevier Inc. 375

with this, the bHLH transcription factor, Neurogenin 3, in combi-

nation with Pdx1 and Mafa, can efficiently reprogram pancreatic

exocrine cells into functional b cells in vivo, although the exocrine

cells were known to have some potential to become islet cells

in vitro and share a common parent cell with islet cells (Baeyens

et al., 2005; Zhou et al., 2008). A combination of three factors,

Ascl1, Brn2, and Myt1l, converts dermal fibroblasts to functional

neurons (Vierbuchen et al., 2010), although the degree of global

reprogramming of the neurons is unknown.

In this study, we examined whether key developmental

cardiac regulators could reprogram cardiac fibroblasts into car-

diomyocytes. We found that out of a total of 14 factors, a specific

combination of three transcription factors, Gata4, Mef2c, and

Tbx5, was sufficient to generate functional beating cardiomyo-

cytes directly from mouse postnatal cardiac or dermal fibro-

blasts and that the induced cardiomyocytes (iCMs) were globally

reprogrammed to adopt a cardiomyocyte-like gene expression

profile.

A

D

E

F G H

C

B

Phase αMHC-GFP

αMHC promoter GFPOff

TFs

FACSOn

Cardiac fibroblasts Cardiomyocytes

αMHC promoter GFP

Thy1

100 101 102 103 104100

101

102

103

104

100%

0%

0 1 102 103 104100

101

102

103

104

76.5%

20.6%

10 10

αMHC-GFP

PI

Control Gata4/Mef2c/Mesp1/Tbx5

Control αMHC-GFP explant

αMHC-GFP

Thy1

0 102 103 104 105

0

102

103

104

105

0 102 103 104 105

0

102

103

104

1050% 0%

0%100%

68% 0%

0.6%31.4%

100 101 102 103 104100

101

102

103

104

93.4%

5.1%

100 101 102 103 104100

101

102

103

104

98.9%

0.5%

αMHC-GFP

PI

14 factors - Gata4 14 factors - Pitx2c

100 101 102 103 104100

101

102

103

104

100 101 102 103 104100

101

102

103

104

Control 14 factors

100%

0%

97.3%

1.7%

0123456

Con

trol

Gat

a4B

af60

c

Han

d2H

opx

Hrt2 Isl1

Mef

2cM

esp1

Myo

cdN

kx2.

5P

itx2c

Sm

yd1

Srf

Tbx5

14 factors - 1

*

Con

trol

Gat

a4 Isl1

Mef

2cM

esp1

Myo

cdN

kx2.

5S

myd

1S

rfTb

x5

9 factors - 1

9 fa

ctor

s0

5

10

15

20

αM

HC

-GFP

+ ce

lls (%

)

αM

HC

-GFP

+ ce

lls (%

)

αM

HC

-GFP

+ ce

lls (%

)

** * * * *

05

101520

Con

trol

Gat

a4M

ef2c

Mes

p1M

yocd

Nkx

2.5

Tbx5

6 factors - 1

9 fa

ctor

s6

fact

ors

**

* * *

Exp

lant

Con

trol

Gat

a4M

ef2c

Mes

p1M

yocd

Tbx5

5 factors - 1

6 fa

ctor

s5

fact

ors0

5

10

15

20

I

14 fa

ctor

s

14 fa

ctor

s

***

*

**

*** ****

αM

HC

-GFP

+ ce

lls (%

)

Figure 1. Screening for Cardiomyocyte-

Inducing Factors

(A) Schematic representation of the strategy to test

candidate cardiomyocyte-inducing factors.

(B) Morphology and characterization of fibroblast-

like cells migrating from aMHC-GFP heart

explants. Phase contrast (left), GFP (middle), and

Thy-1 immunostaining (right). Insets are high-

magnification views. See also Figure S1.

(C) Thy-1+/GFP� cells were FACS sorted from

explant cultures for reprogramming.

(D) Summary of FACS analyses for a-MHC-GFP+

cells. Effect on GFP+ cell induction with 14 factors

or the removal of individual factors from the pool of

14 factors (n = 3). Removal of Baf60c, Hand2,

Hopx, Hrt2, or Pitx2c did not decrease the percent

of GFP+ cells and were excluded for further anal-

yses. See also Figure S2.

(E) FACS plots for analyses of GFP+ cells. GFP+

cells were analyzed 1 week after 14 factor trans-

duction. The number of GFP+ cells were reduced

by removal of Gata4, but increased by removal of

Pitx2c from 14 factors.

(F–H) Effect on GFP+ cell induction of the removal

of individual factors from the pool of 9 (F), 6 (G), or 5

(H) factors (n = 3). Factors that did not decrease

efficiency upon removal were excluded from

further study.

(I) GFP+ (20%) cells were induced from fibroblasts

by the combination of four factors, Gata4, Mef2c,

Mesp1, and Tbx5. Representative data are shown

in each panel. PI, propidium iodine. All data are pre-

sented as means ± SD. *p < 0.01; **p < 0.05 versus

relevant control. Scale bars represent 100 mm.

See also Figures S1 and S2.

RESULTS

Screening for Cardiomyocyte-Inducing FactorsWe developed an assay system in

which the induction of mature cardiomyo-

cytes from fibroblasts could be analyzed quantitatively by

reporter-based fluorescence-activated cell sorting (FACS) (Fig-

ure 1A). To accomplish this, we generated aMHC promoter-

driven EGFP-IRES-puromycin transgenic mice (aMHC-GFP), in

which only mature cardiomyocytes expressed the green fluores-

cent protein (GFP) (Gulick et al., 1991). We confirmed that only

cardiomyocytes, but not other cell types such as cardiac fibro-

blasts, expressed GFP in the transgenic mouse hearts and in

primary cultured neonatal mouse cardiac cells (Figure S1 avail-

able online).

To have enough cardiac fibroblasts for FACS screening, we

obtained GFP� cardiac fibroblasts from neonatal aMHC-GFP

hearts by explant culture. Fibroblast-like cells migrated from

the explants after 2 days and were confluent after 1 week. The

migrating cells did not express GFP, but expressed Thy1 and

vimentin, markers of cardiac fibroblasts (Figure 1B and data

not shown) (Hudon-David et al., 2007; Ieda et al., 2009). To avoid

contamination of cardiomyocytes, we filtered the cells by cell

376 Cell 142, 375–386, August 6, 2010 ª2010 Elsevier Inc.

strainers to remove heart tissue fragments and isolated Thy1+/

GFP� cells by FACS (Figure 1C). Using FACS, we confirmed that

Thy1+/GFP� cells did not express cardiac troponin T (cTnT),

a specific sarcomeric marker of differentiated mature cardio-

myocytes (Figure S1) (Kattman et al., 2006). With these proce-

dures, we had no cardiomyocyte contamination in the fibroblast

culture and could generate greater than twice the number of

cardiac fibroblasts than by conventional fibroblast isolation tech-

niques (Ieda et al., 2009).

To select potential cardiac reprogramming factors, we used

microarray analyses to identify transcription factors and epige-

netic remodeling factors with greater expression in mouse

cardiomyocytes than in cardiac fibroblasts at embryonic day

12.5 (Ieda et al., 2009). Among them, we selected 13 factors

that exhibited severe developmental cardiac defects and

embryonic lethality when mutated (Figure S2). We also included

the cardiovascular mesoderm-specific transcription factor

Mesp1 because of its cardiac transdifferentiation effect in

Xenopus (David et al., 2008). We generated individual retrovi-

ruses to efficiently express each gene in cardiac fibroblasts

(Figure S2).

We transduced Thy1+/GFP� neonatal mouse cardiac fibro-

blasts with a mixture of retroviruses expressing all 14 factors

or with DsRed retrovirus (negative control) (Hong et al., 2009).

We did not observe any GFP+ cells in cardiac fibroblasts 1 week

after Ds-Red retrovirus infection or 1 week of culture without any

viral infection. In contrast, transduction of all 14 factors into fibro-

blasts resulted in the generation of a small number of GFP+ cells

(1.7%), indicating the successful activation of the cardiac-

enriched aMHC gene in some cells (Figures 1D and 1E).

To determine which of the 14 factors were critical for activating

cardiac gene expression, we serially removed individual factors

from the pool of 14. Pools lacking five factors (Baf60c, Hand2,

Hopx, Hrt2, and Pitx2c) produced an increased number of

GFP+ cells, suggesting they are dispensable in this setting

(Figures 1D and 1E). Of note, removing Gata4 decreased the

percentage of GFP+ cells to 0.5%, and removing Pitx2c

increased it to 5%. Removal of the five factors listed above

resulted in an increase in the percentage of GFP+ cells to 13%

(Figure 1F). We conducted three further rounds of withdrawing

single factors from nine-, six-, and five-factor pools, removing

those that did not decrease efficiency upon withdrawal, and

found that four factors (Gata4, Mef2c, Mesp1, and Tbx5) were

sufficient for efficient GFP+ cell induction from cardiac fibro-

blasts (Figures 1F–1H). The combination of these four factors

dramatically increased the number of fibroblasts activating the

aMHC-GFP reporter to over 20% (Figure 1I).

Gata4, Mef2c, and Tbx5 Are Sufficient forCardiomyocyte InductionNext, we examined the expression of cTnT by FACS. We found

that 20% of GFP+ cells expressed cTnT at high enough levels

to detect by FACS 1 week after the four-factor transduction.

Again removing individual factors from the four-factor pool in

transduction, we found that Mesp1 was dispensable for cTnT

expression (Figures 2A and 2B). In contrast, we did not observe

cTnT+ or GFP+ cells, when either Mef2c or Tbx5 was removed.

Removal of Gata4 did not significantly affect the number of

GFP+ cells, but cTnT expression was abolished, suggesting

Gata4 was also required. Whereas the combination of two

factors, Mef2c and Tbx5, induced GFP expression but not

cTnT, no combination of two factors or single factor induced

both GFP and cTnT expression in cardiac fibroblasts (Figure 2C).

These data suggested that the combination of three factors,

Gata4, Mef2c, and Tbx5, is sufficient to induce cardiac gene

expression in fibroblasts.

We found that 30% of GFP+ cells expressed cTnT 1 week after

the three-factor transduction. Next, to confirm our screening

results, we transduced cardiac fibroblasts with three factors

(Gata4, Mef2c, and Tbx5, hereafter referred to as GMT) plus

Nkx2-5, a critical factor for cardiogenesis but excluded by our

initial screening. Surprisingly, adding Nkx2-5 to GMT dramati-

cally inhibited the expression of GFP and cTnT in cardiac

fibroblasts. We also transduced cardiac fibroblasts with the

combination of Baf60c, Gata4, and Tbx5, which can transdiffer-

entiate noncardiogenic mesoderm to cardiomyocytes in mouse

embryos (Takeuchi and Bruneau, 2009). We found that this

combination did not efficiently induce cTnT or GFP expression

above that of Tbx5 alone, confirming our screening results

(Figure 2D).

To determine if other cardiac genes were enriched in GFP+

cells, we sorted GFP+ cells and GFP� cells 7 days after transduc-

tion with GMT and compared gene expression of cardiomyo-

cyte-specific genes, Myh6 (a-myosin heavy chain), Actc1

(cardiac a-actin), Actn2 (actinin a2), and Nppa (natriuretic

peptide precursor type A) by quantitative RT-PCR (qPCR). We

found that these cardiac genes were upregulated significantly

more in GFP+ than in GFP� cells (Figure 2E). Next, we used

immunocytochemistry to determine if cardiac proteins were

expressed in GFP+ cells. Despite the detection of cTnT in only

30% of GFP+ cells, most GFP+ cells induced with the three fac-

tors expressed sarcomeric a-actinin (a-actinin) and had well-

defined sarcomeric structures, similar to neonatal cardiomyo-

cytes (Figure 2F; Figure S1). In addition to a-actinin, some

GFP+ cells also expressed cTnT and ANF (atrial natriuretic

factor), indicating GFP+ cells expressed several cardiomyo-

cyte-specific markers (Figures 2G and 2H). We also confirmed

that neither GFP+ nor GFP� cells expressed smooth muscle or

endothelial cell markers (Figure S2), suggesting specificity of

GMT effects.

Induced Cardiomyocytes Originate from DifferentiatedFibroblasts and Are Directly ReprogrammedWe next isolated neonatal cardiac fibroblasts by the conven-

tional fibroblast isolation method in which hearts were digested

with trypsin and plated on plastic dishes (Ieda et al., 2009).

More than 85% of the cells expressed Thy1, and we isolated

Thy1+/GFP� cells by FACS to exclude cardiomyocyte contami-

nation (Figure 3A). Fibroblasts transduced with GMT expressed

GFP, cTnT, and actinin after 1 week at the same level as fibro-

blasts isolated from explant cultures (Figures 3B and 3C). Similar

results were obtained on introduction of GMT into adult cardiac

fibroblasts, with full formation of sarcomeric structures (Fig-

ure 3D; Figure S2).

To determine if the induced cardiomyocyte-like cells (iCMs)

were arising from a subpopulation of stem-like cells, we


analyzed c-kit expression (Beltrami et al., 2003) in the Thy1+/

GFP� cells. Most c-kit+ cells coexpressed Thy1, whereas 15%

of Thy1+ cells expressed c-kit, which is consistent with a previous

report of cardiac explant-derived cells (Davis et al., 2009). We

isolated GFP�/Thy1+/c-kit+ cells and GFP�/Thy1+/c-kit� cells

by FACS and transduced each population of cells with GMT.

We found 2–3-fold more cardiomyocyte induction in GFP�/

Thy1+/c-kit� cells than in GFP�/Thy1+/c-kit+ cells (Figure S3).

These results suggest that most of the iCMs originated from

a c-kit-negative population.

We then sought to more definitively exclude the possibility of

rare cardiac progenitors giving rise to iCMs. We tested the

potential of mouse tail-tip dermal fibroblasts to generate iCMs.

We found that sorted Thy1+/GFP� tail-tip dermal fibroblasts

transduced with GMT expressed GFP at the same level as

GMT-transduced cardiac fibroblasts, although the percentage

of cTnT+ cells was less than cardiac fibroblast-derived iCMs

A

αMHC-GFP100 101 102 103 104

100

101

102

103

104

0% 0%

0%100%100 101 102 103 104

100

101

102

103

104

0.7% 4.3%

24.7%70.3%100 101 102 103 104

100

101

102

103

104

1.0% 6.5%

19.7%72.7%100 101 102 103 104

100

101

102

103

104

0.1% 0.5%

28.3%70.9%100 101 102 103 104

100

101

102

103

104

0.2% 0.1%

1.2 %98.6%100 101 102 103 104

100

101

102

103

104

0.7% 0%

0.3%98.9%

cTnT

4 f - Gata4 4 f - Mef2c 4 f - Mesp1 4 f - Tbx5Control 4 factors

D F

αMHC-GFP cTnT Merged αMHC-GFP ANF Merged

100 101 102 103 104100

101

102

103

104

0.8% 4.0%

13.0%82.2%100 101 102 103 104

100

101

102

103

104

0.3% 0.9%

2.4%96.4%

αMHC-GFP

cTnT

3 factors 3 f + Nkx2.5

100 101 102 103 104100

101

102

103

104

0.3% 0.5%

2.7%96.5%

Baf60c/Gata4/Tbx5

E

αMHC-GFP α-actinin Merged

control

3 factors

3 factors

αMHC-GFP+ cells cTnT+ cells

0

5

10

15

20

Con

trol

Gat

a4

Mef

2c

Mes

p1

Tbx5

4 factors - 1 factor

4 fa

ctor

s

25

30C

ontro

l

Gat

a4

Mef

2c

Tbx5

3 factors - 1

3 fa

ctor

s

Gat

a4

Mef

2c

Tbx5

1 factor

0

5

10

15

20

25

30B C

* ** **

*

* ** ** ** ****

(%) (%)

**

*

*0

50

100

150

200

250

Myh6

Actn2Actc1

Nppa

Rel

ativ

e m

RN

A e

xpre

ssio

n

(αMHC-GFP+ vs. αMHC-GFP- cells)

G H

Figure 2. Combination of Three Transcrip-

tion Factors Induces Cardiac Gene Expres-

sion in Fibroblasts

(A) FACS analyses for a-MHC-GFP and cardiac

Troponin T (cTnT) expression. Effects of the

removal of individual factors from the pool of four

factors on GFP+ and cTnT+ cell induction.

(B) Quantitative data of GFP+ cells and cTnT+ cells

in (A) (n = 3).

(C) Effect of the transduction of pools of three, two,

and one factors on GFP+ and cTnT+ cell induction

(n = 3).

(D) FACS analyses for a-MHC-GFP and cTnT

expression. Effects of GMT plus Nkx2.5 and

Baf60c/Gata4/Tbx5 transduction are shown.

(E) The mRNA expression in GFP+ and GFP� cells

7 days after GMT transduction was determined by

qPCR (n = 3).

(F) Immunofluorescent staining for GFP, a-actinin,

and DAPI. The combination of the three factors,

GMT, induced abundant GFP, and a-actinin

expression in cardiac fibroblasts 2 weeks after

transduction. High-magnification views in insets

show sarcomeric organization. See also Figures

S1 and S2.

(G) Induced cardiomyocytes expressed cTnT by

immunocytochemistry with clear sarcomeric orga-

nization 4 weeks after transduction. Insets are

high-magnification views.

(H) Induced cardiomyocytes expressed ANF at

perinuclear sites 2 weeks after transduction.

All data are presented as means ± SD. *p < 0.01

versus relevant control. Scale bars represent

100 mm.


(Figures 3E–3G). Like cardiac fibroblasts,

tail-tip fibroblast-derived GFP+ cells ex-

pressed a-actinin and had well-defined

sarcomeric structures (Figure 3H; Fig-

ure S3), suggesting noncardiac fibro-

blasts can also be reprogrammed into

cardiomyocytes by GMT induction.

These results excluded the possibility that the iCMs arose from

contamination of cardiomyocytes or cardiac progenitors before

cardiac induction in the fibroblast population.

We also determined whether the reprogramming of fibroblasts

to differentiated cardiomyocytes was a direct event or if the

fibroblasts first passed through a cardiac progenitor cell fate

before further differentiation. To distinguish between these two

possibilities, we used mice expressing Isl1–yellow fluorescent

protein (YFP) obtained by crossing Isl1-Cre mice and R26R-

EYFP mice (Srinivas et al., 2001) (Figure S3). Isl1 is an early

cardiac progenitor marker that is transiently expressed before

cardiac differentiation. If iCMs generated from fibroblasts

passed through a cardiac progenitor state, they and their

descendants would permanently express YFP (Laugwitz et al.,

2005). We isolated Isl1-YFP�/Thy1+cells from Isl1-YFP heart

explants by FACS and transduced the cells with GMT. The

resulting cTnT+ cells did not express YFP in significant numbers,


suggesting that the iCMs were not first reprogrammed into Isl1+

cardiac progenitor cells (Figures 3I and 3J). Moreover, these

results provided supportive evidence that the iCMs did not orig-

inate from a rare population of cardiac progenitor cells that might

exist in neonatal hearts.

Whereas Isl1 marks most early cardiac progenitors, a subpop-

ulation of cardiac progenitors remains Isl1 negative. Mesp1 is

the earliest pan-cardiovascular progenitor cell marker that is

transiently expressed in nascent mesoderm before further

cardiovascular differentiation (Figure S3) (Saga et al., 1999).

We therefore generated Mesp1-YFP mice by crossing Mesp1-

Cre and R26R-EYFP mice to determine if iCMs were reprog-

rammed into early cardiac mesoderm before further differentia-

tion. We isolated Mesp1-YFP�/Thy1+ tail-tip dermal fibroblasts

by FACS and transduced the cells with GMT (Figures 3K and

2 3 4 5

0

102

103

104

105 0% 0%

0%100%

A B

Thy1

cTnT

Control αMHC-GFP CF Control Gata4/Mef2c/Tbx5

J

100 101 102 103 104100

101

102

103

104

8.6% 0.4%

1.3%89.7%

cTnT

Isl1-YFP

Gata4/Mef2c/Tbx5

I

0 102 103 104 105

0

102

103

104

105 35.5% 27.5%

17.5%19.4%0 102 103 104 105

0

102

103

104

105 0% 0%

0%100%

Isl1-YFP

Thy1

Isl1-YFP Explant Control

0 10 10 10 10 0 102 103 104 105

0

102

103

104

105 86.6% 0%

0.1%13.2%100 101 102 103 104

100

101

102

103

104

0% 0%

0%100%100 101 102 103 104

100

101

102

103

104

1.9% 6.5%

18.2%73.3%

αMHC-GFP

C

αMHC-GFP

αMHC-GFP α-Actinin Merged

D

cTnT

αMHC-GFP100 101 102 103 104

100

101

102

103

104

3.5% 5.2%

17.4%73.9%

Adult αMHC-GFP ExplantGata4/Mef2c/Tbx5

0 102 103 104 105

0

102

103

104

105

0 102 103 104 105

0

102

103

104

105

100 101 102 103 104100

101

102

103

104

100 101 102 103 104100

101

102

103

104

Thy1

cTnT

GFE αMHC-GFP TTF Gata4/Mef2c/Tbx5

LK

Thy1

Mesp1-YFP

cTnT

Mesp1-YFP

Mesp1-YFP TTF Gata4/Mef2c/Tbx5

05

101520

CF

2530 αMHC-GFP+ cells

cTnT+ cells

*

(%)

TTF

88.4% 0%

0%11.6%

87.7% 0.2%

0.2%11.9%

1.5% 2.5%

21.5%74.5%

5.7% 0%

0%94.3%

αMHC-GFP α-Actinin Merged

H

αMHC-GFP CF-derived iCMs

αMHC-GFP TTF-derived iCMs

αMHC-GFP αMHC-GFP One week

Figure 3. Induced Cardiomyocytes Origi-

nate from Differentiated Fibroblasts and

Are Directly Reprogrammed

(A) Cardiac fibroblasts (CF) isolated by the conven-

tional isolation method. Most cells were positive

for Thy1, and Thy-1+/GFP� cells were sorted by

FACS for transduction.

(B) FACS analyses for aMHC-GFP and cTnT

expression in cardiac fibroblasts isolated in (A)

1 week after transduction by GMT.

(C) Immunofluorescent staining for GFP, a-actinin,

and DAPI in the GMT induced cardiomyocytes

originated from (A).

(D) Cardiac fibroblasts isolated from adult aMHC-

GFP hearts were transduced with three factors.

See also Figure S2.

(E) Thy-1+/GFP� tail-tip dermal fibroblasts (TTFs)

were sorted by FACS for transduction.

(F) FACS analyses for GFP and cTnT expres-

sion in TTFs isolated in (E) 1 week after GMT

transduction.

(G) Quantitative data of GFP+ cells and cTnT+ cells

indicated in (F) (n = 3 in each group).

(H) Immunofluorescent staining for GFP, a-actinin,

and DAPI in TTF-derived iCMs. See also Figure S3.

(I) Isl1-YFP�/Thy1+cells were sorted from Isl1-Cre/

Rosa-YFP heart explants and transduced with

GMT. See also Figure S3.

(J) The vast majority of cTnT+ cells induced from

Isl1-YFP�/Thy1+cells was negative for YFP.

(K) Mesp1-YFP�/Thy1+cells were sorted from

Mesp1-Cre/Rosa-YFP TTFs and transduced with

GMT. See also Figure S3.

(L) All cTnT+ cells induced from Mesp1-YFP�/

Thy1+cells were negative for YFP.

All data are presented as means ± SD. *p < 0.01

versus relevant control. Scale bars represent

100 mm. See also Figure S3 for analyses of c-kit+

cells.


3L). The resulting cTnT+ cells did not

express YFP, suggesting that the iCMs

were not converted into the cardiac

mesoderm cell state for reprogramming,

but rather they were directly reprog-

rammed into differentiated cardiomyocytes by the three factors

(Figure 3L).

Induced Cardiomyocytes Resemble PostnatalCardiomyocytes in Global Gene ExpressionWe next analyzed the time course of cardiomyocyte induction

from cardiac fibroblasts. GFP+ cells were detected 3 days after

induction and gradually increased in number up to 20% at day

10 and were still present after 4 weeks (Figure 4A). GFP+ cells

were less proliferative than GFP� cells and, over time, decreased

in percentage relative to the total number of cells. Importantly, the

percentage of cTnT+ cells among the a-MHC-GFP+ iCMs and the

intensity of cTnT expression increased significantly over time

(Figures 4B and 4C). We sorted GFP+ cells at 1, 2, and 4 weeks

after transduction with GMT and compared cardiac gene


expression with cardiac fibroblasts and neonatal cardiomyo-

cytes. The cardiomyocyte-specific genes, Actc1, Myh6, Ryr2

(ryanodine receptor 2), and Gja1 (connexin43), were significantly

upregulated in a time-dependent manner in GFP+ cells, but

were not detected in cardiac fibroblasts by qPCR (Figure 4D).

Col1a2 (collagen 1a2), a marker of fibroblasts, was dramatically

downregulated in GFP+ cells from 7-day culture to the same level

as in cardiomyocytes. These data indicated that the three factors

induced direct conversion of cardiac fibroblasts to cardiomyo-

cytes rapidly and efficiently, but full maturation was a slow

process that occurred over several weeks. Total gene expression

of the three reprogramming factors was upregulated 6- to 8-fold in

iCMs over neonatal cardiomyocytes. However, only endogenous

expression of Gata4 was upregulated in iCMs to the same level as

in neonatal cardiomyocytes, whereas endogenous Mef2c and

Tbx5 expression was lower in iCMs than in cardiomyocytes,

potentially reflecting negative autoregulatory loops (Figure S4).

A B

D

100 101 102 103 104100

101

102

103

104

100%

0%

100 101 102 103 104100

101

102

103

104

59.2%

31.8%

100 101 102 103 104100

101

102

103

104

55.9%

35.5%

100 101 102 103 104100

101

102

103

104

51%

45.1%

C

E

cTnT

Control 1 week 2 weeks 4 weeks

0

0.4

0.6

0.8

1.0

ND0.2

*

**

Actc1

0

8

12

16

4

*

Col1a2

CF

iCMs (1W)

iCMs (2W)

iCMs (4W)CM

Relative mRNA expression

ND**

Myh6

0

0.4

0.6

0.8

1.0

0.2ND

**

Ryr2

0

0.4

0.6

0.8

1.0

0.2

*

Gja1

0

0.4

0.6

0.8

1.0

0.2

(vs. 1W iCMs)

0

400

800

1200

1600

2000Relative

cTnT intensity

0 1 2 4Time (weeks)

*

*

(vs. 1W)

05

1015202530 αMHC-GFP+ cells

0 1 3 7 10 14 28 Time (days)

**

***

(vs. 3d)

Percent of cTnT+ cells in αMHC-GFP+ iCMs

2W GFP- 2W iCMsCF CM

FDR-adj p<0.0001 in at least one comparison

4W GFP- 4W iCMs

1

2

Per

cent

(%) o

f tot

al c

ells

Figure 4. Gene Expression of Induced Car-

diomyocytes Is Globally Reprogrammed

(A) The percent of aMHC-GFP+ cells after GMT

transduction (n = 3). The number of GFP+ cells

was counted by FACS at each time point and

divided by the number of plated cells.

(B) FACS analyses of percent of cells with cTnT

expression among aMHC-GFP+ iCMs. Note that

cTnT + cell number and cTnT intensity were both

increased over time (n = 4).

(C) Quantitative data of cTnT intensity in (B) (n = 4).

(D) Actc1, Myh6, Ryr2, Gja1, and Col1a2 mRNA

expression in cardiac fibroblasts (CF), induced

cardiomyocytes (iCMs) (1 week (W), 2 W, 4 W after

transduction), and neonatal cardiomyocytes (CM),

determined by qPCR (n = 3).

(E) Heatmap image of microarray data illustrating

differentially expressed genes among CF, a-

MHC-GFP�, iCMs (FACS sorted 2 and 4 weeks

after transduction), and CM (n = 3 in each group).

The scale extends from 0.25- to 4-fold over

mean (�2 to +2 in log2 scale). Red indicates

increased expression, whereas green indicates

decreased expression. Group 1 includes the

genes upregulated only in CM, and group 2

includes the genes upregulated in CM and 4W-

iCMs compared to CF. Lists of genes are shown

in Table S1 and Table S2. All data are presented

as means ± SD. *p < 0.01; **p < 0.05 versus rele-

vant control. See also Figure S4 for endogenous

and exogenous expression of reprogramming

factors and Table S1 and Table S2 for differentially

expressed genes.

See also Tables S1 and S2 and Figure S4.

We next compared the progressive

global gene expression pattern of iCMs,

neonatal cardiomyocytes, and cardiac

fibroblasts by mRNA microarray anal-

yses. We sorted GFP+ cells and GFP�

cells 2 and 4 weeks after GMT transduc-

tion. The iCMs at both stages were similar

to neonatal cardiomyocytes, but were

distinct from GFP� cells and cardiac fibroblasts in global gene

expression pattern (Figure 4E). We found that functionally impor-

tant cardiac genes were upregulated significantly more in 4 week

iCMs than in 2 week iCMs, including Pln (phospholamban),

Slc8a1 (sodium/calcium exchanger), Myh6, Sema3a (sema-

phorin 3a), Id2 (inhibitor of DNA binding 2), and Myl2 (myosin, light

polypeptide 2, regulatory, cardiac, slow, also known as MLC2v)

(Table S1). Conversely, some genes were downregulated more

in 4 week iCMs than in 2 week iCMs (Table S1). The array analyses

also identified genes that were upregulated more in neonatal car-

diomyocytes than in 4 week iCMs or cardiac fibroblasts (group 1

in Figure 4E), including Bmp10 (bone morphogenetic protein 10),

Erbb4 (v-erb-a erythroblastic leukemia viral oncogene homolog

4), Irx4 (Iroquois related homeobox 4), and Atp1a2 (ATPase,

Na+/K+ transporting, a 2 polypeptide) (Table S2). We also identi-

fied genes that were expressed to a greater extent in both cardi-

omyocytes and 4 week iCMs than in fibroblasts (group 2 in


Figure 4E), including Actc1, Myl7 (myosin, light polypeptide 7,

regulatory, also known as MLC2a), Tnnt2 (troponin T2, cardiac),

Tbx3 (T-box 3), and Srf (serum response factor) (Table S2).

Thus, iCMs were similar, but not identical, to neonatal cardiomyo-

cytes, and the reprogramming event was broadly reflected in

global gene expression changes.

Fibroblasts Are Epigenetically Reprogrammedto a Cardiomyocyte-like State by Gata4/Mef2c/Tbx5To determine if iCMs have gained a cardiomyocyte-like chro-

matin state, we analyzed the enrichment of histone modifications

in the promoter regions of the cardiac-specific genes Actn2,

Ryr2I, and Tnnt2. We analyzed the enrichment of trimethylated

histone H3 of lysine 27 (H3K27me3) and lysine 4 (H3K4me3),

which mark transcriptionally inactive or active chromatin,

respectively (Li et al., 2007), in cardiac fibroblasts, 4 week

iCMs, and neonatal cardiac cells by chromatin immunoprecipita-

tion, followed by qPCR (Figure 5A). After reprogramming,

H3K27me3 was significantly depleted at the promoters of all

the genes analyzed in iCMs, reaching levels comparable to those

in cardiac cells, whereas H3K4me3 increased on the promoter

regions of Actn2 and Tnnt2 in iCMs, as compared with cardiac

fibroblasts. Ryr2 had similar levels of H3K4me3 in iCMs as in

fibroblasts, suggesting that its activation reflects the resolution

of a ‘‘bivalent’’ chromatin mark (Bernstein et al., 2006). These

results suggested that cardiac fibroblast-derived iCMs gained

a chromatin status similar to cardiomyocytes at least in some

cardiac specific genes. Intriguingly, H3K27me3 levels were

higher in tail-tip fibroblasts than cardiac fibroblasts on all three

genes analyzed and, despite a significant reduction upon

reprogramming to iCMs, remained somewhat higher than in

cardiac cells and cardiac fibroblast-derived iCMs.

A

αMHC promoter GFP

Off

WT TTF

iCMs

OnαMHC promoter GFP

αMHC-GFP TTF

1 day 6 days

5´-LTR tetO GFP 3´-LTR

5´-LTR CMV rtTA 3´-LTR

iCMs

5´-LTR CMV rtTA 3´-LTR

5´-LTR tetO Gata4/Mef2c/Tbx5 3´-LTR

1 week2 weeks

B

Off Dox 1d, +Dox 1d, -Dox 3d, -Dox 6d, -Dox

C

E

CF

CM

iCMs

αMHC-GFP-

Nppa Myh6

infection

+Dox -Dox

infection

+Dox -Dox

D

F

Fold

enr

ichm

ent/I

gG

H3K4me3H3K27me3

Act

n2R

yr2

Tnnt

2

0

345

21

0

345

21

0

10

15

5

0

345

21

0

10

15

20

5

0

10

15

20

5

CF iCMs (CF) TTF iCMs (TTF) Cardiac cells

** *

* **

** *

** * *

**

G

αMHC-GFP α-actinin Merged αMHC-GFP α-actinin Merged

-Dox +Dox +Dox

GFP

Figure 5. Fibroblasts Are Stably Reprog-

rammed into iCMs by Gata4, Mef2c, and

Tbx5

(A) The promoters of Actn2, Ryr2, and Tnnt2 were

analyzed by ChIP for trimethylation status of

histone H3 of lysine 27 or 4 in cardiac fibroblasts

(CF), CF-derived iCMs, tail-tip fibroblasts (TTF),

TTF-derived iCMs, and neonatal cardiac cells.

Data were quantified by qPCR.

(B) The promoters of Nppa and Myh6 were

analyzed with bisulfite genomic sequencing for

DNA methylation status in CF, a-MHC-GFP� cells,

a-MHC-GFP+ iCMs (FACS sorted 4 weeks after

transduction), and neonatal CM. Open circles indi-

cate unmethylated CpG dinucleotides; closed

circles indicate methylated CpGs.

(C) Schematic representation of the strategy to

test expression kinetics of the doxycycline (Dox)-

inducible lentiviral system.

(D) Wild-type TTFs were infected with pLVX-tetO-

GFP and pLVX-rtTA and imaged before (off Dox),

1 day after Dox addition, and at time points after

Dox withdrawal (–Dox). All images were taken

using constant exposure times and identical

camera settings.

(E) Schematic representation of the strategy to

determine temporal requirement of Gata4/Mef2c/

Tbx5 in reprogramming. Thy1+/GFP� TTF were

infected with the pLVX-tetO-GMT and pLVX-rtTA

lentiviruses, and Dox was added for 2 weeks and

thereafter withdrawn for 1 week.

(F) Immunofluorescent staining for GFP, a-actinin,

and DAPI in iCMs 2 weeks after lentiviral infection

and Dox induction.

(G) Immunofluorescent staining for GFP, a-actinin,

and DAPI 1 week after Dox withdrawal. iCMs

maintained a-MHC GFP expression and had

a-actinin positive sarcomeric structures. High-

magnification views in insets show sarcomeric

organization. Representative data are shown in

each panel. All data are presented as means ±

SD. *p < 0.01; **p < 0.05 versus relevant control.

Scale bars represent 100 mm.


The DNA methylation status of specific loci also reflects the

stability of the reprogramming event and we therefore investi-

gated such changes during reprogramming from cardiac fibro-

blasts to iCMs. We performed bisulfite genomic sequencing in

the promoter regions of Nppa and Myh6 in cardiac fibroblasts,

4 week GFP� cells, iCMs, and neonatal cardiomyocytes. Both

promoter regions were hypermethylated in cardiac fibroblasts

and GFP� cells, as expected from the cardiomyocyte-specific

expression of these genes, but were comparatively demethy-

lated in iCMs, similar to neonatal cardiomyocytes (Figure 5B).

These results indicated that reprogramming by Gata4, Mef2c,

and Tbx5 induced epigenetic resetting of the fibroblast genome

to a cardiomyocyte-like state.

To further assess the stability of the reprogramming event, we

generated a doxycycline-inducible lentiviral system in which

transgene expression of the reprogramming factors was con-

trolled by doxycycline administration. We first transduced wild-

type tail-tip fibroblasts with a mixture of lentiviruses containing

pLVX-tetO-GFP and pLVX-rtTA to determine the expression

kinetics of this system (Figure 5C). We confirmed that the

majority of fibroblasts infected with both viruses expressed

GFP within 1 day after doxycycline induction, and the GFP

expression was instantly diminished by withdrawal of doxycy-

cline and disappeared within 6 days (Figure 5D). Thy1+/GFP�

tail-tip fibroblasts were harvested from aMHC-GFP neonatal

mice, transduced with a pool of lentiviruses containing inducible

Gata4, Mef2c, and Tbx5, along with pLVX-rtTA, and subse-

quently treated with doxycycline (Figure 5E). We found that

aMHC-GFP expression was induced from tail-tip fibroblasts

after doxycycline administration and that the iCMs had well-

defined sarcomeric structures marked with an anti-a-actinin

antibody after 2 weeks of culture (Figure 5F). Doxycycline was

withdrawn after 2 weeks of culture, and cells were subsequently

cultured without doxycycline for 1 week to fully remove exoge-

nous expression of the reprogramming factors (Figure 5E). The

iCMs maintained aMHC-GFP expression and had sarcomeric

structures after doxycycline withdrawal, suggesting that the

fibroblasts were stably reprogrammed into iCMs after 2 weeks

exposure to Gata4, Mef2c, and Tbx5 (Figure 5G).

Induced Cardiomyocytes ExhibitSpontaneous ContractionTo determine if iCMs possessed the functional properties char-

acteristic of cardiomyocytes, we analyzed intracellular Ca2+

flux in iCMs after 2–4 weeks of culture. Around 30% of cardiac

fibroblast-derived iCMs showed spontaneous Ca2+ oscillations

and their frequency was variable, resembling what was observed

in neonatal cardiomyocytes (Figures 6A, 6B, and 6D; Movie S1).

We observed that tail-tip dermal fibroblast-derived iCMs also

exhibited spontaneous Ca2+ oscillations, but the oscillation

frequency was lower than that of cardiomyocytes and cardiac

fibroblast-derived iCMs (Figures 6C and 6E; Movie S2).

In addition to the characteristic Ca2+ flux, cardiac fibroblast-

derived iCMs showed spontaneous contractile activity after

4–5 weeks in culture (Movies S3 and S4; Figure S5). Single-cell

extracellular recording of electrical activity in beating cells

revealed tracings similar to the potential observed in neonatal

cardiomyocytes (Figure 6F). Intracellular electrical recording of

iCMs displayed action potentials that resembled those of adult

mouse ventricular cardiomyocytes (Figure 6G). Thus, the reprog-

ramming of fibroblasts to iCMs was associated with global

changes in gene expression, epigenetic reprogramming, and

the functional properties characteristic of cardiomyocytes.

Transplanted Cardiac Fibroblasts Transducedwith Gata4/Mef2c/Tbx5 Reprogram In VivoTo investigate whether GMT-transduced cardiac fibroblasts can

be reprogrammed to express cardiomyocyte-specific genes in

their native environment in vivo, we harvested GFP�/Thy1+

cardiac fibroblasts 1 day after viral transduction and injected

them into immunosuppressed NOD-SCID mouse hearts. GMT-

infected cells did not express GFP at the time of transplantation

(Figure 4A). Cardiac fibroblasts were infected with either the

mixture of GMT and DsRed retroviruses or DsRed retrovirus

(negative control) to be readily identified by fluorescence.

Cardiac fibroblasts infected with DsRed did not express a-acti-

nin or GFP, confirming cardiomyocyte conversion did not

happen in the negative control (Figures 7A and 7B). Despite

being injected into the heart only 1 day after viral infection,

a subset of cardiac fibroblasts transduced with GMT and DsRed

expressed GFP in the mouse heart within 2 weeks (Figure 7B).

Importantly, the GFP+ cells expressed a-actinin and had sarco-

meric structures (Figure 7C). These results suggested that

cardiac fibroblasts transduced with Gata4, Mef2c, and Tbx5

can reprogram to cardiomyocytes within 2 weeks upon trans-

plantation in vivo.

DISCUSSION

Here we demonstrated that the combination of three transcrip-

tion factors, Gata4, Mef2c, and Tbx5, can rapidly and efficiently

induce cardiomyocyte-like cells from postnatal cardiac and

dermal fibroblasts. iCMs were similar to neonatal cardiomyo-

cytes in global gene expression profile, electrophysiologically,

and could contract spontaneously, demonstrating that func-

tional cardiomyocytes can be generated from differentiated

somatic cells by defined factors. Although much refinement

and characterization of the reprogramming process will be

necessary, the findings reported here raise the possibility of

reprogramming the vast pool of endogenous fibroblasts that

normally exists in the heart into functional cardiomyocytes for

regenerative purposes.

The three reprogramming factors, Gata4, Mef2c, and Tbx5,

are core transcription factors during early heart development

(Olson, 2006; Srivastava, 2006; Zhao et al., 2008). They interact

with one another, coactivate cardiac gene expression (e.g.,

Nppa, Gja5 [Cx40], and Myh6), and promote cardiomyocyte

differentiation (Bruneau et al., 2001; Garg et al., 2003; Ghosh

et al., 2009; Lin et al., 1997). Gata4 is considered a ‘‘pioneer’’

factor and might open chromatin structure in cardiac loci (Cirillo

et al., 2002), thus allowing binding of Mef2c and Tbx5 to their

specific target sites and leading to full activation of the cardiac

program. Although the reprogramming event appears stable at

the epigenetic level, as marked by histone methylation and

DNA methylation, the global gene expression of iCMs is similar

but not identical to neonatal cardiomyocytes. Whether they are


more similar to adult ventricular cardiomyocytes or other

subpopulations remains to be determined. Additional epigenetic

regulators, microRNAs, or signaling proteins may be leveraged

to increase the efficiency and robustness of the reprogramming

event. Furthermore, other combinations of factors likely also

induce cardiac reprogramming, much like the experience in the

iPSC field.

Several lines of evidence suggest that the iCMs we describe

here originated from differentiated fibroblasts. We found that

any potential rare cardiac ‘‘progenitor-like’’ cells, marked by

c-kit or Isl1, were dispensable for cardiomyocyte induction

(Beltrami et al., 2003). Furthermore, the high efficiency of

cardiac induction (up to 20%) does not favor the interpretation

that rare stem or progenitor cells were the origin of induced

cardiomyocytes. Most importantly, the ability to reprogram

dermal fibroblasts into iCMs supports the conclusion that

cardiac progenitors are not the target cells for the reprogram-

A

B

D

αMHC-GFP Ca2+ max Ca2+ min

C Rhod-3 intensity(Tail tip fibroblast-derived iCMs)

E

Ca2+ max Ca2+ min

Tail tip fibroblast-derived iCMs

Cardiac fibroblast-derived iCMs

F

Rhod-3 intensity (Neonatal cardiomyocytes)

1 s 1 s

Rhod-3 intensity (Cardiac fibroblast-derived iCMs)

1 s 1 s

GCardiac fibroblast-derived iCMsextracellular electrical recording

50 ms

20 m

V

0 mV

0 mV

Cardiac fibroblast-derived iCMs

Adult mouse ventricular cardiomyocytes

50 ms

20 m

V

1 s

Ca2+ max Ca2+ min

Neonatal cardiomyocytes

1 s

2 m

V

Neonatal cardiomyocytesextracellular electrical recording

1 s

2 m

V

αMHC-GFP

Figure 6. Induced Cardiomyocytes Exhibit

Spontaneous Ca2+ Flux, Electrical Activity,

and Beating

(A and B) Cardiac fibroblast (CF)-derived iCMs

showed spontaneous Ca2+ oscillation with varying

frequency (A), similar to neonatal cardiomyocytes

(B). Rhod-3 intensity traces are shown.

(C) Tail-tip dermal fibroblast (TTF)-derived iCMs

showed spontaneous Ca2+ oscillation with lower

frequency. The Rhod-3 intensity trace is shown.

(D) Spontaneous Ca2+ waves observed in CF-

derived a-MHC-GFP+ iCMs (white dots) or

neonatal cardiomyocytes (arrows) with Rhod-3 at

Ca2+ max and min is shown. Fluorescent images

correspond to the Movie S1.

(E) Spontaneous Ca2+ oscillation observed in the

TTF-derived a-MHC-GFP+ iCMs with Rhod-3 at

Ca2+ max and min is shown. Fluorescent images

correspond to the Movie S2.

(F) Spontaneously contracting iCMs had electrical

activity measured by single cell extracellular elec-

trodes. Neonatal cardiomyocytes showed similar

electrical activity.

(G) Intracellular electrical recording of CF-derived

iCMs cultured for 10 weeks displayed action

potentials that resembled those of adult mouse

ventricular cardiomyocytes. Representative data

are shown in each panel (n = 10 in A–F, n = 4 in

G). See also Figure S5 and Movies S1, S2, S3

and S4.

See also Movies S1, S2, S3, and S4 and Figure S5.

ming factors. Remarkably, reprogram-

ming of cardiac fibroblasts to myocytes

occurred in a relatively short period,

with the first GFP+ cells appearing at

day 3, in contrast to iPSC reprogram-

ming, which typically takes 10–20 days

and occurs with much lower efficiency

(<0.1%) (Takahashi and Yamanaka,

2006). Despite the early initiation of

reprogramming, the process appears to

continue for several weeks, with progres-

sive changes in gene expression, contractile ability, and electro-

physiologic maturation.

Although many questions remain regarding the mechanisms

of reprogramming, we were able to genetically test the ‘‘route’’

of cell fate alteration. Our findings suggest that cardiomyocytes

were directly induced from cardiac fibroblasts without reverting

to a cardiac progenitor cell state, which may explain the rapid

early reprogramming process. This conclusion was supported

by the absence of Isl1-Cre-YFP or Mesp1-Cre-YFP activation

during the process of reprogramming, which would have marked

any cells that transiently expressed Isl1 or Mesp1 (Laugwitz

et al., 2005; Saga et al., 1999).

The ability to reprogram endogenous cardiac fibroblasts into

cardiomyocytes has many therapeutic implications. First, the

avoidance of reprogramming to pluripotent cells before cardiac

differentiation would greatly lower the risk of tumor formation

in the setting of future cell-based therapies. Second, large


amounts of an individual’s own fibroblasts can be grown from

a cardiac biopsy or skin biopsy in vitro for transduction with

the defined factors, followed by delivery of cells to damaged

hearts. Third, and most promising, is the potential to introduce

the defined factors, or factors that mimic their effects, directly

into the heart to reprogram the endogenous fibroblast popula-

tion, which represents more than 50% of the cells, into new car-

diomyocytes that can contribute to the overall contractility of the

heart. Our observation that injection of fibroblasts into the heart

only 1 day after induction of Gata4/Mef2c/Tbx5 resulted in

reprogramming of the transplanted cells suggests that this

may be possible. Future studies in human cells and advances

in safe delivery of defined factors will be necessary to advance

this technology for potential regenerative therapies.

EXPERIMENTAL PROCEDURES

Generation of aMHC-GFP, Isl1-YFP, and Mesp1-YFP Mice

To generate aMHC-GFP mice, EGFP-IRES-Puromycin cDNA was subcloned

into the expression vector containing a-myosin heavy chain promoter (Gulick

et al., 1991). Pronuclear microinjection and other procedures were performed

according to the standard protocols (Ieda et al., 2007). PCR primers are listed

in the Extended Experimental Procedures. Isl1-YFP mice were obtained by

crossing Isl1-Cre mice and R26R-EYFP mice (Srinivas et al., 2001). Mesp1-

YFP mice were obtained by crossing Mesp1-Cre mice and R26R-EYFP mice

(Saga et al., 1999).

A

DsRed αMHC-GFP Merged

DsRed αMHC-GFP Merged

DsRed-cell injected DsRed-cell injected DsRed-cell injected

GMT/DsRed-cell injected GMT/DsRed-cell injected GMT/DsRed-cell injected

B

αMHC-GFP Merged

GMT-cell injected GMT-cell injected GMT-cell injectedC

α-actinin Merged

DsRed-cell injected DsRed-cell injected DsRed-cell injected

DsRed

α-actinin

Figure 7. Transplanted Cardiac Fibroblasts Trans-

duced with Gata4/Mef2c/Tbx5 Can Be Reprog-

rammed to Cardiomyocytes In Vivo

(A) DsRed infected cardiac fibroblasts (DsRed-cell) were

transplanted into NOD-SCID mouse hearts 1 day after

infection and cardiac sections were analyzed by immuno-

cytochemistry after 2 weeks. Transplanted fibroblasts

marked with DsRed did not express a-actinin (green).

(B) Cardiac fibroblasts infected with DsRed or Gata4/

Mef2c/Tbx5 with DsRed (GMT/DsRed-cell) were trans-

planted into NOD-SCID mouse hearts 1 day after infection

and visualized by histologic section. Note that a subset of

GMT/DsRed cells expressed a-MHC-GFP. Data were

analyzed 2 weeks after transplantation.

(C) Gata4/Mef2c/Tbx5-transduced cardiac fibroblasts

(GMT-cell) were transplanted into mouse hearts and histo-

logic sections analyzed. A subset of induced GFP+ cells

expressed a-actinin (red) and had sarcomeric structures.

Insets are high-magnification views of cells indicated by

arrows. Data were analyzed 2 weeks after transplantation.

Representative data are shown in each panel (n = 4 in each

group). Scale bars represent 100 mm. Note that GMT/

DsRed or GMT-infected cells did not express GFP at the

time of transplantation (Figure 4A).

Cell Culture

For explant culture, isolated neonatal or adult mouse

hearts were minced into small pieces less than 1 mm3 in

size. The explants were plated on gelatin-coated dishes

and cultured for 7 days in explant medium (IMDM/20%

FBS) (Andersen et al., 2009). Migrated cells were har-

vested and filtered with 40 mm cell strainers (BD) to avoid

contamination of heart tissue fragments. aMHC-GFP�/

Thy1+, Isl1-YFP�/Thy1+, aMHC-GFP�/Thy1+/c-kit�, or

aMHC-GFP�/Thy1+/c-kit+ live cells (as defined by the

lack of propidium iodine staining) were isolated using FACS Aria 2 (BD Biosci-

ences). For conventional isolation of neonatal cardiac fibroblasts, hearts were

digested with 0.1% trypsin and plated on plastic dishes (Ieda et al., 2009).

For isolation of tail-tip fibroblasts, tails were digested with 0.1% trypsin and

plated on plastic dishes. Attached fibroblasts were cultured for 7 days and

aMHC-GFP�/Thy1+ or Mesp1-YFP�/Thy1+ cells were sorted and cultured in

DMEM/M199 medium containing 10% FBS at a density of 104/cm2. Cells

were transduced by retroviruses or lentiviruses after 24 hr.

Isolation of Cardiomyocytes

To isolate cardiomyocytes, neonatal aMHC-GFP+ ventricles were cut into

small pieces and digested with collagenase type II solution (Ieda et al.,

2009). A single-cell suspension was obtained by gentle triturating and passing

through a 40 mm cell strainer. aMHC-GFP+ live cells were isolated by FACS

Aria 2. To obtain cardiac cells, cells were plated on gelatin-coated plastic

dishes and treated with Ara C (Sigma) to inhibit nonmyocyte proliferation.

Molecular Cloning and Retroviral/Lentiviral Infection

Retroviruses or inducible lentiviruses containing the cardiac developmental

factors were generated as described and as detailed in the Extended Experi-

mental Procedures (Kitamura et al., 2003; Takahashi and Yamanaka, 2006).

The pMXs-DsRed Express retrovirus infection in cardiac fibroblasts resulted

in >95% transduction efficiency (Hong et al., 2009).

FACS Analyses and Sorting

For GFP expression analyses, cells were harvested from cultured dishes and

analyzed on a FACS Calibur (BD Biosciences) with FlowJo software. For

aMHC-GFP/cTnT expression, cells were fixed with 4% PFA for 15 min, per-

meabilized with Saponin, and stained with anti-cTnT and anti-GFP antibodies,


followed by secondary antibodies conjugated with Alexa 488 and 647 (Katt-

man et al., 2006).

For aMHC-GFP�/Thy1+, Isl1-YFP�/Thy1+, and Mesp1-YFP�/Thy1+ cell

sorting, cells were incubated with PECy7 or APC-conjugated anti-Thy1

antibody (eBioscience) and sorted by FACS Aria 2 (Ieda et al., 2009).

For aMHC-GFP�/Thy1+/c-kit� and aMHC-GFP�/Thy1+/c-kit+ cell sorting,

PECy7-conjugated anti-Thy1 and APC-conjugated anti-c-kit antibodies

(BD) were used. We used bone marrow cells as a positive control for c-kit

staining.

Cell Transplantation

Fibroblasts were harvested 1 day after retroviral infection. A left thoracotomy

was carried out in NOD-SCID mice, and 106 cultured cells were injected

into the left ventricle. After 1–2 weeks, the hearts were excised for

immunohistochemistry.

Histology and Immunocytochemistry

Cells or tissues were fixed, processed and stained with antibodies against

numerous proteins in standard fashion as detailed in the Extended Experi-

mental Procedures.

Quantitative RT-PCR

Total RNA was isolated from cells, and qRT-PCR was performed on an ABI

7900HT (Applied Biosystems) with TaqMan probes (Applied Biosystems),

which are listed in the Extended Experimental Procedures. To quantify endog-

enous-specific transcripts and both endogenous and transgene common tran-

scripts, primers were designed using Vector NTI, and SYBR green technology

was used. Primer information is available on request. mRNA levels were

normalized by comparison to Gapdh mRNA.

Microarray Analyses

Mouse genome-wide gene expression analyses were performed using

Affymetrix Mouse Gene 1.0 ST Array. aMHC-GFP+ cardiomyocytes were

collected by FACS. Three-factor transduced GFP+ cells and GFP� cells

were collected by FACS after 2 and 4 weeks of culture. Cardiac fibroblasts

were also collected after 4 weeks of culture. RNA was extracted using Pico-

Pure RNA Isolation (Arcturus). Microarray analyses were performed in triplicate

from independent biologic samples, according to the standard Affymetrix

Genechip protocol. Data were analyzed using the Affymetrix Power Tool

(APT, version 1.8.5). See the Extended Experimental Procedures for additional

statistical methods.

Chromatin Immunoprecipitation Assay

Chromatin immunoprecipitations were performed on cardiac fibroblasts, tail-

tip dermal fibroblasts, iCMs, and neonatal cardiac cells. Immunoprecipitations

were done using the Imprint Chromatin Immunoprecipitation Kit (Sigma)

following the manufacturer instructions. Antibodies against H3K27me3 and

H3K4me3 were from Active motif, and normal rabbit IgG was from Cell

Signaling Technology. Primer sequences for qPCR custom TaqMan gene

expression assays (Applied Biosystems) are listed in the Extended Experi-

mental Procedures.

Bisulfite Genomic Sequencing

Bisulfite treatment was performed using the Epitect Bisulfite Kit (QIAGEN)

according to the manufacturer’s recommendations. PCR primers are listed

in the Extended Experimental Procedures. Amplified products were cloned

into pCR2.1-TOPO (Invitrogen). Ten randomly selected clones were

sequenced with the M13 forward and M13 reverse primers for each gene.

Ca2+ Imaging

Ca2+ imaging was performed according to the standard protocol. Briefly, cells

were labeled with Rhod-3 (Invitrogen) for 1 hr at room temperature, washed,

and incubated for an additional 1 hr to allow de-esterification of the dye.

Rhod-3-labeled cells were analyzed by Axio Observer (Zeiss) with MiCAM02

(SciMedia).

Electrophysiology

After 4 week transduction with GMT, the electrophysiological activities of iCMs

were analyzed using extracellular electrode recording with an Axopatch 700B

amplifier and the pClamp9.2 software (Axon Instruments). iCMs were visually

identified by GFP expression and spontaneous contraction. Glass patch

pipettes, with typical resistances of 2–4 MU, were directly attached on single

GFP+ cells for extracellular recording in Tyrode’s bath solution. For recording

intracellular action potentials, single GFP+ cells were held at �70 mV

membrane potential with a stimulation of 0.1–0.5 nA for 5 ms to elicit

a response after 10-week transduction with GMT.

Statistical Analyses

Differences between groups were examined for statistical significance using

Student’s t test or ANOVA. p values of <0.05 were regarded as significant.

ACCESSION NUMBERS

Microarray data have been submitted and can be accessed by the Gene

Expression Omnibus (GEO) accession number GSE22292.

SUPPLEMENTAL INFORMATION

Supplemental Information includes Extended Experimental Procedures, five

figures, two tables, and four movies and can be found with this article online

at doi:10.1016/j.cell.2010.07.002.

ACKNOWLEDGMENTS

We are grateful to members of the Srivastava lab, to K. Tomoda for critical

discussions and comments on the manuscript, to J. Olgin and C. Ding for elec-

trophysiology assistance, to Z. Yang and K. Worringer for help with lentivirus

experiments, to Y. Huang for cell transplantation experiments, to B. Taylor,

G. Howard, and S. Ordway for editorial assistance and manuscript prepara-

tion, to C. Barker and L. Ta in the Gladstone Genomics core, to C. Miller and

J. Fish in the Gladstone Histology core, to A. Holloway in the Gladstone Bioin-

formatics core, and to S. Elmes in the Laboratory for Cell Analysis in UCSF. We

also thank S. Yamanaka for helpful discussions and providing pMXs-DsRed

Express plasmid, T. Kitamura for Plat-E cells, J. Robbins for a-MHC promoter

plasmid, T.M. Jessell for Isl1-Cre mice, and F. Costantini for R26R-EYFP mice.

M.I. and Y.H. are supported by a grant from the Uehara Memorial Foundation.

V.V. is supported by grants from the GlaxoSmithKline Cardiovascular

Research and Education Foundation and the NIH/NHLBI. P.D.-O. was a post-

doctoral scholar of the California Institute for Regenerative Medicine. D.S. and

B.G.B are supported by grants from NHLBI/NIH and the California Institute for

Regenerative Medicine. The J. David Gladstone Institutes received support

from a National Center for Research Resources Grant RR18928-01. D.S. is

a member of the Scientific Advisory Board of iPierian, Inc., and RegeneRx.

B.G.B. is a member of the Scientific Advisory Board of iPierian Inc.

Received: February 18, 2010

Revised: May 18, 2010

Accepted: June 25, 2010

Published: August 5, 2010

REFERENCES

Andersen, D.C., Andersen, P., Schneider, M., Jensen, H.B., and Sheikh, S.P.

(2009). Murine ‘‘cardiospheres’’ are not a source of stem cells with cardiomyo-

genic potential. Stem Cells 27, 1571–1581.

Baeyens, L., De Breuck, S., Lardon, J., Mfopou, J.K., Rooman, I., and Bou-

wens, L. (2005). In vitro generation of insulin-producing beta cells from adult

exocrine pancreatic cells. Diabetologia 48, 49–57.

Baudino, T.A., Carver, W., Giles, W., and Borg, T.K. (2006). Cardiac fibroblasts:

friend or foe? Am. J. Physiol. Heart Circ. Physiol. 291, H1015–H1026.

Beltrami, A.P., Barlucchi, L., Torella, D., Baker, M., Limana, F., Chimenti, S.,

Kasahara, H., Rota, M., Musso, E., Urbanek, K., et al. (2003). Adult cardiac


stem cells are multipotent and support myocardial regeneration. Cell 114,

763–776.

Bernstein, B.E., Mikkelsen, T.S., Xie, X., Kamal, M., Huebert, D.J., Cuff, J., Fry,

B., Meissner, A., Wernig, M., Plath, K., et al. (2006). A bivalent chromatin struc-

ture marks key developmental genes in embryonic stem cells. Cell 125, 315–

326.

Bruneau, B.G., Nemer, G., Schmitt, J.P., Charron, F., Robitaille, L., Caron, S.,

Conner, D.A., Gessler, M., Nemer, M., Seidman, C.E., et al. (2001). A murine

model of Holt-Oram syndrome defines roles of the T-box transcription factor

Tbx5 in cardiogenesis and disease. Cell 106, 709–721.

Camelliti, P., Borg, T.K., and Kohl, P. (2005). Structural and functional charac-

terisation of cardiac fibroblasts. Cardiovasc. Res. 65, 40–51.

Cirillo, L.A., Lin, F.R., Cuesta, I., Friedman, D., Jarnik, M., and Zaret, K.S.

(2002). Opening of compacted chromatin by early developmental transcription

factors HNF3 (FoxA) and GATA-4. Mol. Cell 9, 279–289.

David, R., Brenner, C., Stieber, J., Schwarz, F., Brunner, S., Vollmer, M., Men-

tele, E., Muller-Hocker, J., Kitajima, S., Lickert, H., et al. (2008). MesP1 drives

vertebrate cardiovascular differentiation through Dkk-1-mediated blockade of

Wnt-signalling. Nat. Cell Biol. 10, 338–345.

Davis, D.R., Zhang, Y., Smith, R.R., Cheng, K., Terrovitis, J., Malliaras, K., Li,

T.S., White, A., Makkar, R., and Marban, E. (2009). Validation of the cardio-

sphere method to culture cardiac progenitor cells from myocardial tissue.

PLoS ONE 4, e7195.

Davis, R.L., Weintraub, H., and Lassar, A.B. (1987). Expression of a single

transfected cDNA converts fibroblasts to myoblasts. Cell 51, 987–1000.

Garg, V., Kathiriya, I.S., Barnes, R., Schluterman, M.K., King, I.N., Butler, C.A.,

Rothrock, C.R., Eapen, R.S., Hirayama-Yamada, K., Joo, K., et al. (2003).

GATA4 mutations cause human congenital heart defects and reveal an inter-

action with TBX5. Nature 424, 443–447.

Ghosh, T.K., Song, F.F., Packham, E.A., Buxton, S., Robinson, T.E., Ronksley,

J., Self, T., Bonser, A.J., and Brook, J.D. (2009). Physical interaction between

TBX5 and MEF2C is required for early heart development. Mol. Cell. Biol. 29,

2205–2218.

Gulick, J., Subramaniam, A., Neumann, J., and Robbins, J. (1991). Isolation

and characterization of the mouse cardiac myosin heavy chain genes.

J. Biol. Chem. 266, 9180–9185.

Hong, H., Takahashi, K., Ichisaka, T., Aoi, T., Kanagawa, O., Nakagawa, M.,

Okita, K., and Yamanaka, S. (2009). Suppression of induced pluripotent

stem cell generation by the p53-p21 pathway. Nature 460, 1132–1135.

Hudon-David, F., Bouzeghrane, F., Couture, P., and Thibault, G. (2007). Thy-1

expression by cardiac fibroblasts: lack of association with myofibroblast

contractile markers. J. Mol. Cell. Cardiol. 42, 991–1000.

Ieda, M., Kanazawa, H., Kimura, K., Hattori, F., Ieda, Y., Taniguchi, M., Lee,

J.K., Matsumura, K., Tomita, Y., Miyoshi, S., et al. (2007). Sema3a maintains

normal heart rhythm through sympathetic innervation patterning. Nat. Med.

13, 604–612.

Ieda, M., Tsuchihashi, T., Ivey, K.N., Ross, R.S., Hong, T.T., Shaw, R.M., and

Srivastava, D. (2009). Cardiac fibroblasts regulate myocardial proliferation

through beta1 integrin signaling. Dev. Cell 16, 233–244.

Ivey, K.N., and Srivastava, D. (2006). The paradoxical patent ductus arteriosus.

J. Clin. Invest. 116, 2863–2865.

Kattman, S.J., Huber, T.L., and Keller, G.M. (2006). Multipotent flk-1+ cardio-

vascular progenitor cells give rise to the cardiomyocyte, endothelial, and

vascular smooth muscle lineages. Dev. Cell 11, 723–732.

Kitamura, T., Koshino, Y., Shibata, F., Oki, T., Nakajima, H., Nosaka, T., and

Kumagai, H. (2003). Retrovirus-mediated gene transfer and expression

cloning: powerful tools in functional genomics. Exp. Hematol. 31, 1007–1014.

Laflamme, M.A., Chen, K.Y., Naumova, A.V., Muskheli, V., Fugate, J.A., Dup-

ras, S.K., Reinecke, H., Xu, C., Hassanipour, M., Police, S., et al. (2007).

Cardiomyocytes derived from human embryonic stem cells in pro-survival

factors enhance function of infarcted rat hearts. Nat. Biotechnol. 25, 1015–

1024.

Laugwitz, K.L., Moretti, A., Lam, J., Gruber, P., Chen, Y., Woodard, S., Lin,

L.Z., Cai, C.L., Lu, M.M., Reth, M., et al. (2005). Postnatal isl1+ cardioblasts

enter fully differentiated cardiomyocyte lineages. Nature 433, 647–653.

Li, B., Carey, M., and Workman, J.L. (2007). The role of chromatin during tran-

scription. Cell 128, 707–719.

Lin, Q., Schwarz, J., Bucana, C., and Olson, E.N. (1997). Control of mouse

cardiac morphogenesis and myogenesis by transcription factor MEF2C.

Science 276, 1404–1407.

Nussbaum, J., Minami, E., Laflamme, M.A., Virag, J.A., Ware, C.B., Masino, A.,

Muskheli, V., Pabon, L., Reinecke, H., and Murry, C.E. (2007). Transplantation

of undifferentiated murine embryonic stem cells in the heart: teratoma forma-

tion and immune response. FASEB J. 21, 1345–1357.

Olson, E.N. (2006). Gene regulatory networks in the evolution and develop-

ment of the heart. Science 313, 1922–1927.

Saga, Y., Miyagawa-Tomita, S., Takagi, A., Kitajima, S., Miyazaki, J., and In-

oue, T. (1999). MesP1 is expressed in the heart precursor cells and required

for the formation of a single heart tube. Development 126, 3437–3447.

Snider, P., Standley, K.N., Wang, J., Azhar, M., Doetschman, T., and Conway,

S.J. (2009). Origin of cardiac fibroblasts and the role of periostin. Circ. Res.

105, 934–947.

Srinivas, S., Watanabe, T., Lin, C.S., William, C.M., Tanabe, Y., Jessell, T.M.,

and Costantini, F. (2001). Cre reporter strains produced by targeted insertion

of EYFP and ECFP into the ROSA26 locus. BMC Dev. Biol. 1, 4.

Srivastava, D. (2006). Making or breaking the heart: from lineage determination

to morphogenesis. Cell 126, 1037–1048.

Takahashi, K., and Yamanaka, S. (2006). Induction of pluripotent stem cells

from mouse embryonic and adult fibroblast cultures by defined factors. Cell

126, 663–676.

Takeuchi, J.K., and Bruneau, B.G. (2009). Directed transdifferentiation of

mouse mesoderm to heart tissue by defined factors. Nature 459, 708–711.

van Laake, L.W., Passier, R., Doevendans, P.A., and Mummery, C.L. (2008).

Human embryonic stem cell-derived cardiomyocytes and cardiac repair in

rodents. Circ. Res. 102, 1008–1010.

Vierbuchen, T., Ostermeier, A., Pang, Z.P., Kokubu, Y., Sudhof, T.C., and

Wernig, M. (2010). Direct conversion of fibroblasts to functional neurons by

defined factors. Nature 463, 1035–1041.

Zhang, J., Wilson, G.F., Soerens, A.G., Koonce, C.H., Yu, J., Palecek, S.P.,

Thomson, J.A., and Kamp, T.J. (2009). Functional cardiomyocytes derived

from human induced pluripotent stem cells. Circ. Res. 104, e30–e41.

Zhao, R., Watt, A.J., Battle, M.A., Li, J., Bondow, B.J., and Duncan, S.A.

(2008). Loss of both GATA4 and GATA6 blocks cardiac myocyte differentiation

and results in acardia in mice. Dev. Biol. 317, 614–619.

Zhou, Q., Brown, J., Kanarek, A., Rajagopal, J., and Melton, D.A. (2008). In vivo

reprogramming of adult pancreatic exocrine cells to beta-cells. Nature 455,

627–632.


Dendritic Function of Tau MediatesAmyloid-b Toxicity inAlzheimer’s Disease Mouse ModelsLars M. Ittner,1,6,* Yazi D. Ke,1,6 Fabien Delerue,1 Mian Bi,1 Amadeus Gladbach,1 Janet van Eersel,1 Heidrun Wolfing,1

Billy C. Chieng,2 MacDonald J. Christie,2 Ian A. Napier,2 Anne Eckert,3 Matthias Staufenbiel,4 Edna Hardeman,5

and Jurgen Gotz1,*1Alzheimer’s and Parkinson’s Disease Laboratory2Neuropharmacology LaboratoryBrain and Mind Research Institute, University of Sydney, Sydney NSW 2050, Australia3Neurobiology Research Laboratory, Psychiatric University Clinic, University of Basel, Basel CH-4025, Switzerland4Novartis Institutes for BioMedical Research, Basel CH-4002, Switzerland5Neuromuscular and Regenerative Medicine Unit, University of New South Wales, Sydney NSW 2052, Australia6These authors contributed equally to this work

*Correspondence: [email protected] (L.M.I.), [email protected] (J.G.)

DOI 10.1016/j.cell.2010.06.036

SUMMARY

Alzheimer’s disease (AD) is characterized byamyloid-b (Ab) and tau deposition in brain. It hasemerged that Ab toxicity is tau dependent, althoughmechanistically this link remains unclear. Here, weshow that tau, known as axonal protein, hasa dendritic function in postsynaptic targeting of theSrc kinase Fyn, a substrate of which is the NMDAreceptor (NR). Missorting of tau in transgenic miceexpressing truncated tau (Dtau) and absence of tauin tau�/� mice both disrupt postsynaptic targetingof Fyn. This uncouples NR-mediated excitotoxicityand hence mitigates Ab toxicity. Dtau expressionand tau deficiency prevent memory deficits andimprove survival in Ab-forming APP23 mice, a modelof AD. These deficits are also fully rescued witha peptide that uncouples the Fyn-mediated interac-tion of NR and PSD-95 in vivo. Our findings suggestthat this dendritic role of tau confers Ab toxicity atthe postsynapse with direct implications for patho-genesis and treatment of AD.

INTRODUCTION

Alzheimer’s disease (AD) is characterized by two hallmark

lesions, amyloid-b (Ab) plaques and neurofibrillary tangles

(NFTs) (Ballatore et al., 2007). Ab is derived from the amyloid-b

precursor protein (APP) by proteolytic cleavage (Haass et al.,

1992; Selkoe, 1997). The major constituent of NFTs is tau,

a microtubule (MT)-associated protein (Goedert et al., 1988). In

the course of AD, tau becomes phosphorylated, forming aggre-

gates that deposit as NFTs and neuropil threads (Geschwind,

2003). Tau can also form aggregates in the absence of an overt

Ab pathology, for example in frontotemporal dementia (FTD),

where familial mutations have been identified in the tau-encod-

ing MAPT gene (Ballatore et al., 2007). Evidence that tau

pathology in AD is induced by Ab comes from our previous

observation that intracerebral Ab injections exacerbate hyper-

phosphorylation of tau and NFT formation in transgenic mice

that express FTD mutant P301L tau (Gotz et al., 2001b). A similar

finding was obtained by crossing transgenic mice with NFT and

plaque pathologies (Lewis et al., 2001).

Ab-plaque formation along with memory impairment and tau

pathology with increased phosphorylation, in the absence of

deposition and NFT formation, has been reproduced in several

transgenic mouse lines that express human APP together with

pathogenic mutations identified in familial AD (Gotz and Ittner,

2008; Hsiao et al., 1996; Mucke et al., 2000; Sturchler-Pierrat

et al., 1997). In one of these, PDAPP, tau deficiency (tau�/�)

was shown to rescue lethality and memory deficits by an uniden-

tified mechanism (Roberson et al., 2007).

Tau is known as axonal protein that regulates MT stability

and MT-dependent processes (Dixit et al., 2008; Drechsel

et al., 1992; Lee et al., 1988), while Ab likely exerts toxicity

at the postsynapse (Selkoe, 2002; Shankar et al., 2008;

Zhao et al., 2006). Although in AD, hyperphosphorylated tau

accumulates in the somatodendritic compartment of neurons

(Ballatore et al., 2007), given the spatial separation it remains

unknown how tau is involved in mediating Ab toxicity when AD

is initiated.

Seizures characterize several APP transgenic strains (Minkevi-

ciene et al., 2009; Palop et al., 2007; Palop and Mucke, 2009) and

have been associated with AD; the extent of their contribution to

pathology, however, remains to be established (Minkeviciene

et al., 2009; Palop et al., 2007; Palop and Mucke, 2009). Excito-

toxicity results from overactivation of N-methyl-D-aspartate

(NMDA) receptors (NRs). Interestingly, tau reduction decreases

susceptibility to excitotoxic seizures in vivo, which may explain

the concomitant improvement of the PDAPP phenotype (Rober-

son et al., 2007). How tau prevents excitotoxic damage at

a molecular level is not understood.


Tau interacts via its amino-terminal projection domain (PD)

with the kinase Fyn (Figure 1A) (Lee et al., 1998). Fyn phosphor-

ylates the NR subunit 2 (NR2) to facilitate interaction of the NR

complex with the postsynaptic density protein 95 (PSD-95)

(Nakazawa et al., 2001; Rong et al., 2001; Tezuka et al., 1999),

linking NRs to synaptic excitotoxic downstream signaling (Salter

and Kalia, 2004). Disruption of the NR/PSD-95 interaction

prevents excitotoxic damage in cultured neurons and a rat

model of stroke, without affecting synaptic NMDA currents

(Aarts et al., 2002). Reduction of Fyn in APP transgenic mice

prevents Ab toxicity, while overexpression enhances it (Chin

et al., 2005; Chin et al., 2004).

To address how tau confers Ab toxicity, we generated trans-

genic mice (Dtau74) that express only the amino-terminal projec-

tion domain (PD) of tau and crossed them with Ab-forming

APP23 and tau�/� mice. We found that tau has an important

dendritic function, as in Dtau74 and tau�/� mice, postsynaptic

Fyn localization is reduced, resulting in reduced NR phosphory-

lation, destabilized NR/PSD-95 interaction, and protection from

excitotoxicity.

RESULTS

Truncated Tau Is Excluded from DendritesTau comprises an amino-terminal projection domain, an MT

binding (MTB) domain that mediates interaction with MTs

(Butner and Kirschner, 1991; Lee et al., 1988) and is essential

for tau aggregation (Crowther et al., 1989; Ksiezak-Reding and

Yen, 1991) and a carboxy-terminal tail region (Figure 1A). We

generated truncated (Dtau) transgenic mice that express the

projection domain of tau in neurons, intended to compete with

functions of endogenous tau. Four phenotypically normal lines

expressed Dtau throughout the brain (Figure 1B) at comparable

levels, with line Dtau74 expressing the transgene at 1.4-fold

higher levels than endogenous tau (Figure 1C). Expression of

Dtau neither affected levels nor distribution of endogenous tau

(Figure 1C and Figures S1A–S1C available online). Consistent

with previous in vitro findings (Maas et al., 2000), Dtau localized

to the cell membrane, as indicated by coimmunostaining with

cadherin and subcellular fractionation of membranes (Figures

S1D and S1E). In AD and also full-length P301L mutant tau trans-

genic pR5 mice, tau is hyperphosphorylated and redistributed

into the somatodendritic compartment (Figure 1D) (Gotz et al.,

2001a). In contrast to full-length tau, Dtau, while in the soma,

was virtually excluded from dendrites (Figure 1D). In pR5

mice, tau becomes progressively hyperphosphorylated and

insoluble, and eventually the mice develop NFTs. Surprisingly,

Dtau in Dtau74 mice is hardly phosphorylated at all (Figure S1F).

Postsynaptic Targeting of Fyn Is Tau DependentDifferent from full-length human tau in pR5 mice, in the absence

of an MTB domain, Dtau fails to interact with MTs, as determined

by MT precipitation from hippocampi (Figure 2A). However, Dtau

contains motifs that mediate interaction with the Src kinase Fyn,

as shown in vitro (Lee et al., 1998). Accordingly, Fyn can be coim-

munoprecipitated with Dtau from Dtau74 hippocampi in vivo,

using a human tau-specific antibody (HT7) (Figure 2B). Immuno-

precipitation (IP) with tau-specific antibodies to epitopes not

present on the Dtau construct reveals a significantly reduced

interaction of Fyn with endogenous tau (Figure 2B). Likewise,

IP with Fyn antibodies shows a reduced interaction with endog-

enous tau in Dtau74 mice (Figure 2B). Together, this suggests

a dominant negative effect of Dtau on the normal interaction of

Fyn and endogenous tau. A similar effect on the Fyn/Tau

cx

am

hp

∆tau74wt

5037

5037

kD

4

tau

leve

ls(fo

ld o

f wt)

3210

totaltau

endogenoustau

∆tau74wt

C

B

Tau-5

HT7Gapdh

A

PD MTB C’

Fyn binding site

1 255mThy1.2 ∆tau

1 441htau40

wt ∆tau74

S

D

pR5∆tau74H

T7 D

AP

ID

Figure 1. Truncated Tau Is Excluded from

Dendrites in Dtau74 Mice

(A) The longest human tau isoform (htau40; 441 aa)

is composed of an amino-terminal projection

domain (PD), the microtubule-binding (MTB)

domain with four repeats (gray boxes), and the car-

boxy-terminal tail (C0). Dtau transgenic mice

express only the PD of tau under control of the

neuronal mThy1.2 promoter. Dtau lacks the MTB

domain and therefore the MT-binding and aggre-

gation properties of full-length tau, but contains

a Fyn binding site.

(B) Expression pattern of Dtau in Dtau74 brains.

Immunohistochemistry (IHC) with a human tau-

specific antibody (HT7; brown) reveals Dtau

expression within several brain regions, including

hippocampus (hp), cortex (cx), and amygdala (am).

(C) Western blotting of wild-type and Dtau74

hippocampal extracts reveals endogenous murine

tau (50 kD) in all and Dtau (37 kD) only in transgenic

samples. Quantification shows comparable levels

of endogenous tau, while endogenous tau and

Dtau levels add up to 2.4-fold increased total levels

in Dtau74 compared to WT mice.

(D) IHC of the hippocampal CA1 region reveals that in Dtau74 mice, Dtau localizes to the soma (S) but is excluded from dendrites (D), whereas expression of P301L

mutant full-length tau in pR5 mice results in a somato-dendritic localization of transgenic tau (HT7; reactive with Dtau and pR5 tau, but not endogenous tau,

in red). The scale bar represents 50 mm.

Error bars represent the standard error. See also Figure S1.


interaction was obtained by overexpression of full-length tau in

pR5 mice (Figure 2B). Given the dendritic exclusion of Dtau in

Dtau74 in contrast to full-length tau in pR5 mice (Figure 1D),

we speculated that the aberrant Dtau/Fyn interaction might

affect the normal intracellular distribution of Fyn. Immunohisto-

chemistry showed that Fyn colocalized with drebrin in wild-

type (WT) brain, consistent with postsynaptic targeting, while in

Dtau74 brains it accumulated in the soma, an effect enhanced

by crossing of Dtau and tau�/� (Figures 2C and 2D). Together

with reduced dendritic Fyn staining, this suggests impaired

postsynaptic targeting of Fyn. To determine the role of tau in

dendritic localization of Fyn, we also analyzed tau�/� mice. Fyn

also accumulated in the soma (Figures 2C and 2D), suggesting

that postsynaptic targeting of Fyn is, at least in part, tau depen-

dent. This is consistent with reduced localization of Fyn-DsRED

in primary hippocampal neurons either from tau�/�mice or mice

coexpressing Dtau (Figure S2). Interestingly, further truncation of

Dtau shows that the Fyn-interactive motif, PXXP (Lee et al.,

1998), is critical for Fyn localization.

Despite changes in the localization of Fyn, its total levels

and activity were comparable in Dtau74, tau�/�, and WT mice,

as determined by total and phosphorylation site-specific anti-

Fyn antibodies (Figures 2E and 2F). To quantify changes in the

subcellular localization of Fyn, we prepared synaptosomes

from WT, Dtau74, and tau�/� hippocampi. Consistent with the

immunohistochemical findings of reduced postsynaptic target-

ing, levels of synaptic Fyn were reduced by 73% and 62% in

Dtau74 and tau�/�mice, respectively, compared to WT controls

(Figure 2G). Taken together, both the presence of Dtau and

absence of endogenous tau impair synaptic localization of Fyn.

Uncoupled NMDA Receptors and PSD-95 in Dtauand tau�/� SynapsesThe postsynaptic NR subunit NR2b is a known substrate of

Fyn (Nakazawa et al., 2001). NR2b phosphorylation at Y1472

strengthens the NR/PSD-95 interaction (Rong et al., 2001). In

both Dtau74 and tau�/� mice, Y1472 phosphorylation is signifi-

cantly reduced compared to the WT, while total levels of NR1,

NR2a, and NR2b are unaffected (Figure 3A). To determine

whether this affects the stability of NR/PSD-95 complexes,

we performed coimmunoprecipitations (coIPs). Markedly less

NR1, NR2a, and NR2b coimmunoprecipitated with PSD-95

E F G

0

0.5

1.0

Fyn

(fold

of w

t)

wt

∆tau74 tau

-/-

Fyn

PSD-95

wt ∆tau74 tau-/-

Fyn

Gapdh

wt ∆tau74 tau-/-

wt ∆tau74

tau-/-

pR5B

mTau

inpu

tIP

: mTa

uIP

: HT7

IP: F

yn

HT7Fyn

Gapdh

mTau

HT7Fyn

mTau

Fyn

Fyn

mTau

HT7

human taumurine tau∆tau

ApR

5

Tau-5human taumurine tau∆tau

tubulinactin

extract no MT MT prec

∆tau74

pR5

∆tau74

pR5

∆tau74

Fyn

Dre

brin

∆tau74

∆tau74.tau-/-

wt

tau-/-

C

S

D

S

D

D

* * *

* * *

wt

∆tau74

∆tau74.tau-/-

tau-/-

soma dendrites0

10

20

30

40

Fluo

resc

ence

inte

nsity

(AFU

)

Fyn

(fold

of w

t)

0

0.5

1.0

wt

∆tau74 tau

-/-

p420-Fyn

IP:Fyn

wt ∆tau74 tau-/-

p531-Fyn

* *

Figure 2. Dtau Impairs Tau-Dependent

Dendritic Targeting of the Src Kinase Fyn

(A) Dtau from Dtau74 mice does not interact with

microtubules. Endogenous murine tau, but not

Dtau, precipitates with microtubules in extracts

from Dtau74 mice. In contrast, both full-length

human and endogenous murine tau precipitate

with microtubules in extracts from pR5 mice.

(B) Expression of Dtau results in a 74% ± 6%

(n = 8; *p < 0.01) reduced interaction of Fyn with

endogenous murine tau (mtau) compared to the

wild-type (wt), as revealed by coimmunoprecipita-

tion (coIP) with antibodies to endogenous murine

tau (mTau). In Dtau74 mice, Fyn instead coimmu-

noprecipitates with Dtau, as revealed by antibody

HT7. Similarly, in pR5 mice, Fyn precipitates with

full-length human tau (htau). CoIP with Fyn anti-

bodies predominantly pulled down endogenous

tau in WT, Dtau in Dtau74, and full-length human

tau in pR5 mice. No precipitation was observed

from tau�/� tissue.

(C) Fyn accumulates in cell bodies in Dtau74,

tau�/�, and Dtau74.tau�/�mice. While Fyn staining

(red) colocalizes with dendritic drebrin (green) in

WT CA1 neurons, Fyn staining is evident in the

soma (S) and is reduced in the dendrites (D) of

Dtau74 and tau�/� neurons. The insets show

higher magnification of dendritic staining. The

scale bar represents 50 mm.

(D) Quantification of fluorescence intensity of Fyn

staining in cell bodies and dendrites shows accu-

mulation of Fyn in cell bodies of Dtau74, tau�/�,

and Dtau74.tau�/� mice (n = 15, *p < 0.0001).

(E) Total Fyn levels are not reduced in Dtau74 and

tau�/� mice. Western blots of hippocampal

extracts from WT, Dtau74, and tau�/� brains

show comparable levels of Fyn, normalized to

Gapdh (n = 6).

(F) Phosphorylation of activating (Y420) and inactivating (Y531) sites of immunopurified Fyn from WT, Dtau74, and tau�/� brains is similar.

(G) Hippocampal synaptosomal preparations reveal reduced levels of Fyn in Dtau74 and tau�/� postsynapses compared to the WT (n = 6, *p < 0.005).



from Dtau74 and tau�/� compared to WT extracts, consistent

with a decreased interaction of NR and PSD-95 in both strains

(Figure 3B). The PSD-95-interacting proteins Homer and

nNOS, however, were coimmunoprecipitated to a similar extent

from WT, Dtau74, and tau�/� brains, suggesting intact interac-

tions. The NR/PSD-95 interaction facilitates stable anchoring

of NRs in the postsynaptic density (PSD) (Roche et al., 2001).

Therefore, we next extracted purified synaptosomes from WT,

Dtau74, and tau�/�mice that show similar levels of NR subunits,

but reduced NR2b phosphorylation at Y1472 (Figure S3A), using

buffers of increasing stringency (Phillips et al., 2001). In line with

a strong PSD association in WT synaptosomes, NR subunits

were mostly found in the SDS fraction (Figure 3C and 3D). In

contrast, they were markedly reduced in Dtau74 and tau�/�,

appearing instead in earlier fractions, suggestive of a weakened

anchoring in the PSD (Figure 3D). Interestingly, endogenous

tau that was enriched by synaptosome preparation, recovered

in WT SDS fractions and coimmunoprecipitated with PSD-95

from WT, and to a lesser degree from Dtau74, brains (Figures

3B and 3C and Figure S1C).

0

0.5

1.0

Y14

72 (f

old

of w

t)

D

C

Y1472

Gapdh

wt ∆tau74 tau-/-

NR1

NR2a

NR2b

Fyn

SNAP25

B

WB

wt ∆tau74 tau-/-

A

NR2b

wt

pH8

NR1

NR2a

NR2b

Fyn

PSD-95

SDS

∆tau74 tau-/-

NR1

NR2a

wt∆tau74tau-/-

PSD-95

E wt ∆tau74

wt ∆tau74 ∆tau74.tau-/-tau-/-

50 ms

50 pA

F

G

H

50 ms

50 pA

IP:PSD-95

mTau

HT7

Fyn

NR1

NR2a

NR2b

nNOS

Homer

*

*

*

*

*

**

*

*

*

*

NR

1

1.0

0.5

0

NR

2a

1.0

0.5

0

NR

2b

1.0

0.5

0

mTa

u 1.0

0.5

0

Fyn

1.0

0.5

0

Gap43

SNAP25

PSD-95

mTau

Fyn

NR1

NR2a

NR2b

total pH6 pH8 SDS

0

50

25

75

100

EPSC

(% N

R2b

)

wt

∆tau74

∆tau74

.tau-/-

tau-/-

0

5

4

3

2

1Freq

uenc

y (s

-1)

wt

∆tau74

∆tau74

.tau-/-

tau-/-

0

80

60

40

20

Min

i EPS

C (p

A)

wt

∆tau74

∆tau74

.tau-/-

tau-/-

0

6

43

5

21

EPSC

(NM

DA

/AM

PA)

wt

∆tau74

∆tau74

.tau-/-

tau-/-

∆tau74.tau-/-tau-/-

Figure 3. Destabilized NMDA Receptors in

the Postsynaptic Density of Dtau74 and

tau�/� Mice

(A) Levels of NR subunits NR1, NR2a and NR2b,

and PSD-95 are comparable in extracts from WT,

Dtau74, and tau�/� brains, whereas phosphoryla-

tion of NR2b at the Fyn site, Y1472, that is known

to stabilize NR/PSD-95 complexes (Roche et al.,

2001), is significantly reduced in Dtau74 and

tau�/� than in the WT (n = 6, *p < 0.005).

(B) PSD-95 antibodies coimmunoprecipitate much

less NR subunits NR1, NR2a, and NR2b from

Dtau74 and tau�/� than from WT hippocampi.

Similarly, coimmunoprecipitation (coIP) of Fyn

with PSD-95 is reduced in Dtau74 and tau�/�

compared to WT hippocampi, while that of nNOS

and Homer was unaffected. Endogenous murine

Tau (mTau) coprecipitates with PSD-95 from WT

hippocampi, while much less mTau, but no Dtau

(HT7), is recovered from Dtau74 hippocampi.

mTau is absent in tau�/� coIPs, consistent with

tau deficiency. (n = 3, *p < 0.005.)

(C) Sequential extraction of synaptosomes. Puri-

fied WT synaptosomes were further fractionated

with buffers of increasing stringency (pH6 <

pH8 < SDS), to purify proteins that are stably asso-

ciated with the PSD (Phillips et al., 2001). Brain

extracts (total) are loaded for comparison. NR

subunits NR1, NR2a and NR2b, PSD-95, tau,

and Fyn are purified in the SDS fraction, consistent

with strong anchoring in the PSD. Soluble proteins,

such as GAP43 and proteins that are not (such as

SNAP25) or less stably associated with the PSD,

are extracted with less stringent pH 6 and pH 8

buffers, respectively.

(D) SDS fractions from synaptosomes show that

stable anchoring of NRs in the PSD is reduced in

Dtau74 and tau�/�mice. While NRs are recovered

in the SDS fraction of WT synaptosomes, they are

primarily found in the pH 8 and hardly at all in SDS

fractions from Dtau74 and tau�/� mice.

(E) Representative traces of AMPAR- (gray) and

NR- (black) mediated components of electrically

evoked (e) EPSCs in CA1 hippocampal neurons from WT, Dtau74, tau�/�, and Dtau74.tau�/�mice (average of 12 sweeps per neuron) normalized with AMPAR-

mediated component. Neurons were voltage clamped and held at +40 mV. AMPAR-mediated eEPSCs are inverted for clarity. There is no significant difference of

NMDA/AMPA ratios between genotypes (n = 19–20).

(F) Representative traces of eEPSCs (average of 12 sweeps per neuron) separating total NMDAR-mediated and NR2b subunit-mediated components

(black traces, total NR eEPSC minus component in CP-101,606 [5 mM]) normalized to the amplitude of the total NR eEPSC. Neurons were voltage clamped

at +20 mV. NR2b EPSCs were obtained by subtraction of EPSCs generated in CP101 606 (5 mM) from total NR-mediated EPSCs, i.e., before CP applications.

There is no significant difference in the percentage of NR2b component between genotypes (n = 18–20).

(G and H) Mean amplitude (G) and rate (s�1) (H) of AMPAR-mediated mEPSCs (recorded in 1 mM TTX) were unaffected in Dtau74, tau�/�, and Dtau74.tau�/�mice

(n = 10–12).



The organization of NRs within the PSD is important for coor-

dinated signal transduction (Kim and Sheng, 2004). Hence, alter-

ations of NRs in Dtau74 and tau�/� mice may affect synaptic

currents. Therefore, we determined excitatory postsynaptic

currents (ESPCs) in acute hippocampal slices from WT,

Dtau74, tau�/�, and Dtau74.tau�/� mice. In Dtau74, tau�/�,

and Dtau74.tau�/� mice, we found np significant changes in

synaptic currents (Figure 3E). Similarly, no significant reduction

emerged in the contribution of NR2b-containing NRs to ESPCs

in Dtau74, tau�/�, and Dtau74.tau�/� mice (Figure 3F). Baseline

miniature amplitudes and frequency were also comparable

(Figures 3G and 3H and Figures S3B–S3D). Taken together,

these data indicate that both expression of Dtau or tau deficiency

reduces the interaction of NRs with PSD-95 without affecting

synaptic NR levels and currents.

Dtau Expression Prevents Premature Lethalityand Memory Deficits in APP23 MiceIt has been shown previously that perturbing the interaction of

NRs with PSD-95 had no effect on NR-mediated currents but

reduced the resilience of neurons to NMDA-mediated excitotox-

icity (Aarts et al., 2002). Interestingly, excitotoxicity has been

proposed to contribute to Ab toxicity in PDAPP mice, which

was reduced when the mice were crossed onto a tau�/� back-

ground (Roberson et al., 2007). APP expression per se may

contribute to toxicity in mice; however, primary disease-related

effects are attributed to Ab, as suggested by reverted deficits

in Ab-immunized APP models (Roskam et al., 2010) and absence

of seizure-induced hippocampal remodeling in APP transgenic

mice with low Ab levels (Palop et al., 2007). Excitotoxicity has

been linked to premature lethality in APP transgenic mice (Chis-

hti et al., 2001; El Khoury et al., 2007; Leissring et al., 2003;

Roberson et al., 2007). Hence, we speculated that in Dtau74

alterations in NR/PSD-95 interaction might similarly rescue the

early lethality that characterizes APPswe mutant APP23 mice

(Figures S4A and S4B) (Sturchler-Pierrat et al., 1997). APP23

mice have high Ab levels already at a very young age (Kuo

et al., 2001; Van Dam et al., 2003), eventually forming plaques

and presenting with neuronal loss and memory deficits (Calhoun

et al., 1998; Kelly et al., 2003; Sturchler-Pierrat et al., 1997).

When we crossed APP23 either with Dtau74 or tau�/� mice

(Figure S4C), this caused both a significantly delayed onset of

mortality and an improved overall survival (Figure 4A). Whereas

any rescue (either on a tau�/� background or by expressing

Dtau) was partial, expression of Dtau on a heterozygous or

homozygous tau-deficient background rescued lethality

completely, suggesting complementary beneficial effects of

tau deficiency and Dtau expression on survival (Figure 4A).

In contrast, crossing of APP23 mice with pR5 mice with an

increased dendritic accumulation of tau (Figure 1D) (Gotz et al.,

BA

C

0 2 4 6 8 1050

75

100

APP23.tau+/-

APP23.∆tau74

APP23.tau-/-

APP23

APP23.∆tau74.tau-/-

APP23.∆tau74.tau+/-

Age (months)

Surv

ival

(% a

live

mic

e)

0

1

2

3

4

Acquisition2h24h

Erro

rs (T

-maz

e)

APP23wt

APP23.∆tau

74

∆tau74

APP23.ta

u-/-

tau-/-

APP23.∆tau

74.ta

u-/-

** ****

** **

****

*

****

****

wt APP23

APP23.∆tau

74

APP23.ta

u-/-

APP23.∆tau

74.ta

u-/-

Fyn

(fold

of w

t)

Y147

2 (fo

ld o

f wt)

Y1472

NR2b

Fyn

PSD-95

2.0

* ***

* ***

1.01.0

2.0

3.0

1.5

0.5

0 0

*

D

E

G

F

0

25

75

100

Plaq

ues

per

sect

ion

APP23APP23.∆tau74

APP23.tau-/-

APP23.∆tau74.tau+/-

0.0

0.5

1.0

1.5

hAPP

mR

NA

(fold

of A

PP23

)

APP23.∆tau74APP23

APP23.∆tau74.tau+/-APP23.tau-/-

0

30

60

90

Aβ1-

42(n

g/m

g)

0

100

200

300

A1-

40(n

g/m

g)β

Figure 4. Dtau Expression Improves

Memory and Ameliorates Premature

Mortality of APP23 Mice

(A) APPswe transgenic APP23 mice (n = 76) present

with a pronounced premature mortality that is

ameliorated by reducing tau levels in APP23.tau+/�

(n = 41, p < 0.001) and even more in APP23.tau�/�

mice (n = 108, p < 0.0001). Expression of Dtau

improves the survival of APP23.Dtau74 mice

(orange, n = 43, p < 0.01) similar to APP23.tau+/�.

Interestingly, combination of Dtau expression with

tau reduction completely rescues APP23.Dtau74.-

tau+/� (purple, n = 38, p < 0.0001) and

APP23.Dtau74.tau�/� (red, n = 52, p < 0.0001)

mice from lethality.

(B) Improved memory acquisition of APP23.D

tau74, APP23.tau�/�, and APP23.Dtau74.tau�/�

compared to APP23 mice in the T maze, 2 and 24

hr after a five-trial acquisition, at 8 months of age.

While WT, Dtau74, and tau�/�mice only make few

errors during the trials, memory deficits of APP23

mice are obvious from the continuously high

numbers of errors made during the entire test. In

contrast, both APP23.Dtau74, APP23.tau�/�, and

APP23.Dtau74.tau�/� mice presented with WT-

like numbers of errors (n = 8, *p < 0.05, **p < 0.01).

(C) In synaptosomal preparations obtained from 4-month-old APP23, both Fyn levels and NR2b phosphorylation at Y1472 are increased as compared to wild-

type (wt) mice (n = 6, *p < 0.05). However, in synaptosomes from APP23.Dtau74 and APP23.tau�/�, and even more in APP23.Dtau74.tau�/�, both levels of Fyn

and NR2b phosphorylation are significantly lower than in APP23 mice (n = 6, *p < 0.05, **p < 0.01). Representative western blots from three independent exper-

iments are shown.

(D–G) Dtau expression and tau deficiency do not affect APP mRNA expression, Ab levels, or plaque burden.

(D) Levels of APP mRNA are not altered in APP23 mice in the presence of Dtau or when tau is absent (tau�/�).

(E) Ab1–40 and Ab1–42 levels are comparable in APP23, APP23.Dtau74, and APP23.tau�/� mice.

(F and G) Thioflavine S staining (green) reveals Ab plaques (arrows; insets) at similar numbers (F) and with similar morphology (G) in APP23 mice, independent of

coexpression of Dtau or tau reduction.



2001a), resulted in increased premature lethality, with no survival

beyond 4 months of age (Figure S4D). Interestingly, both Fyn

levels and Y1472 phosphorylation of NR2b are increased in

pR5 synaptosomes (Figure S4E). Because of possible confound-

ing effects of APP overexpression and Ab formation in APP

mutant mouse strains, we used also primary neurons treated

with Ab, in the absence of APP overexpression, as a model. In

tau�/� and Dtau74-expressing neurons, acute Ab toxicity was

markedly reduced (Figure S2E). Interestingly, deletion of the

Fyn-interacting motif, PXXP (Lee et al., 1998), from Dtau abro-

gated the protective effect.

We next determined whether Dtau expression or tau reduction

also improves memory functions in APP23 mice. Memory defi-

cits were both improved to WT levels in APP23.Dtau74

and APP23.tau�/� mice using the water T maze (Figure 4B).

Consistent with the findings in Dtau and tau�/� mice, both

synaptic Fyn levels and NR2b phosphorylation were reduced

in APP23.Dtau74 and APP23.tau�/� and even more so in

APP23.Dtau74.tau�/� synaptosomes, while they were increased

in APP23 compared to WT brains (Figure 4C). Interestingly, in

APP23 mice, neither Dtau expression nor tau reduction affected

human APP messenger RNA (mRNA) levels (Figure 4D), Ab levels

(Figure 4E), or plaque burden (Figures 4F and 4G). Similarly,

phosphorylation of endogenous tau was comparable in APP23

and APP23.Dtau74 mice (data not shown). Taken together,

expression of Dtau in APP23 or crossing of APP23 with tau�/�

mice reduces Fyn-mediated NR2b phosphorylation, attenuates

premature mortality, and improves memory deficits without

changing Ab levels or plaque load.

Dtau Reduces Susceptibility to Excitotoxic SeizuresAb-induced aberrant excitatory neuronal activity may contribute

to the deficits that characterize AD mouse models (Busche et al.,

2008; Palop and Mucke, 2009). APP23 mice show spontaneous

seizures (Lalonde et al., 2005), similar to other APP transgenic

strains (Minkeviciene et al., 2009; Palop et al., 2007; Palop and

Mucke, 2009). Hence, reduced mortality of APP23.Dtau74 and

APP.tau�/� mice may be related to a reduced susceptibility to

excitotoxic seizures. We therefore first induced convulsions in

Dtau74, tau�/�, Dtau74.tau�/�, and WT mice using the g-amino-

butyrate (GABA) antagonist pentylenetetrazole (PTZ). Seizure

severity was significantly reduced in Dtau74, tau�/�, and

Dtau74.tau�/� compared to the WT (Figure 5A), while the latency

to develop severe convulsion increased (Figure 5B). Next, we

induced seizures in APP23, APP23.Dtau74, APP23.tau�/�, and

APP23.Dtau74.tau�/� mice. APP23 mice presented with a

reduced convulsion latency and showed the most severe seizure

response, with the lowest survival rate (1/11) and all mice reach-

ing status epilepticus (n = 11) (Figure 5C). However, when APP

expression was combined with Dtau expression or tau defi-

ciency, this significantly decreased seizure severity, reduced

fatality, and increased convulsion latency (Figures 5C and 5D).

The double mutant Dtau74.tau�/� prevented severe seizures

better than Dtau74 or tau�/� alone, on both WT and APP23 back-

grounds, in agreement with the survival data (Figure 4A). Inter-

estingly, we found a similar degree of protection from PTZ-

induced seizures as in Dtau74, tau�/�, APP23.Dtau74, or

APP23.tau�/� mice when we pretreated WT and APP23 mice,

**

B

D

A

wt

∆tau74

∆tau74

.tau-/-

tau-/-

0

1

2

3

4

5

6

7

Seiz

ure

Seve

rity

** *****

**

1 2 3 4 5 6 70

2

4

6

8

Seizure Severity

Late

ncy

(min

)

∆tau74tau-/-

∆tau74.tau-/-

wt

***

C

APP23

APP23.∆tau

74

APP23.ta

u-/-

APP23.∆tau

74.ta

u-/-

0

1

2

3

4

5

6

7

Seiz

ure

Seve

rity **

*

FE

vehicle MK801

0

1

2

3

4

5

6

7

Seiz

ure

Seve

rity

1 2 3 4 5 6 70

2

4

6

8

Seizure Severity

Late

ncy

(min

)wt + vehiclewt + MK801

* ***

1 2 3 4 5 6 70

2

4

6

8

Seizure Severity

Late

ncy

(min

)

APP23.∆tau74

APP23.tau-/-

APP23

APP23.∆tau74.tau-/-

APP23 + vehicleAPP23 + MK801

wt APP23

Figure 5. Dtau Expression Reduces Susceptibility to Excitotoxic

Seizures

(A) When excitotoxic seizures were induced by i.p. injection of PTZ (50 mg/kg),

mean seizure severity was significantly reduced in both Dtau74, tau�/�, and

Dtau74.tau�/� compared to WT mice (n = 10, **p < 0.01, ***p < 0.001).

(B) Similarly, the latency to more severe seizure stages is increased in Dtau74,

tau�/�, and Dtau74.tau�/� mice.

(C) In APP23 mice, PTZ-induced seizures are mostly lethal (10 of 11), whereas

in APP23.Dtau74, APP23.tau�/�, and APP23.Dtau74.tau�/� seizure severity is

markedly reduced (n = 10, *p < 0.05, **p < 0.01, ***p < 0.001).

(D) APP23.Dtau74, APP23.tau�/�, and APP23.Dtau74.tau�/� mice show an

increased latency to more severe seizures compared to APP23 mice.

(E and F) Pretreatment of WT or APP23 mice with MK801 (0.1 mg/kg) reduced

seizure severity (n = 8, p < 0.05) (E) and increased latency to more severe

seizures (F).

Error bars represent the standard error.


respectively, with the NR-antagonist MK801 (Figures 5E and 5F).

Hence, reduced susceptibility to excitotoxicity is consistent with

a reduced NR contribution and may contribute to reduced

mortality in APP23 mice in the presence of Dtau or absence of

endogenous tau.

Targeted Uncoupling of NR and PSD-95 PreventsPremature Death and Memory Seficits in APP23 MiceProvided that disturbed NR/PSD-95 complexes with a reduced

dendritic Fyn localization in Dtau74 and tau�/� mice contribute

to improved memory functions and survival of APP23.Dtau74

and APP23.tau�/� mice, targeted perturbation of the NR/PSD-

95 interaction, independent of tau or Fyn, should also decrease

Ab toxicity. Therefore, we treated primary cortical cultures with

the Tat-NR2B9c peptide composed of carboxy-terminal amino

acids of NR2b (including Y1472) fused to a HIV1-Tat peptide to

achieve cell membrane permeability (Figure 6A). Tat-NR2B9c

has been shown previously to protect from NMDA-induced exci-

totoxicity (Aarts et al., 2002; Kornau et al., 1995). As a negative

control, we included Tat-NR2BAA in which critical amino acids

were replaced by alanine (Aarts et al., 2002; Kornau et al.,

1995). NMDA and Ab both induced pronounced cell death, while

a combined NMDA/Ab treatment did not further increase cell

death, consistent with shared signaling pathways mediating their

toxicity (Figures 6B and 6C). Cell death induced by NMDA and

Ab, both separate and in combination, was significantly reduced

by preincubation with Tat-NR2B9c, but not when induced by

hydrogen peroxide or staurosporine (Figures 6B and 6C). Tat-

NR2BAA had no protective effects. Hence, perturbing the NR/

PSD-95 interaction with Tat-NR2B9c ameliorates Ab-mediated

toxicity in vitro.

Next, we tested in vivo whether APP23 mice would also benefit

from treatment with Tat-NR2B9c. A single dose of this peptide

has previously been shown to confer virtually complete protec-

tion from excitotoxic damage in a rat model of stroke (Aarts

et al., 2002). First, we determined whether sufficient NR/

PSD-95 uncoupling was achieved by intracerebroventricular

(i.c.v.) Tat-NR2B9c treatment, using osmotic minipumps. We

delivered either Tat-NR2B9c or Tat-NR2BAA for 1 week and

then performed coIP with a PSD-95 antibody (Figure 6D). This

revealed a reduced NR/PSD-95 interaction upon Tat-NR2B9c,

but not Tat-NR2BAA, treatment. The level of reduction was

similar to that found in Dtau74 and tau�/� brains (Figure 6D).

Sufficient uptake of peptides by the brain was further confirmed

by protection from PTZ-induced seizures by Tat-NR2b9c, but

not Tat-NR2BAA (Figure 6E). Next, we implanted minipumps

pre-tr

eatm

ent

treatm

ent

wash

PI analy

sis

1h 1h 24h 5’

A

DOBi.c

.v. pu

mp

pumpch

ange

d

pumprem

oved

6wks 28d 28d

F

1st p

ump

2nd p

ump

0 30 60 2400

25

50

75

100

Tat-NR2B9c (n=17)

Tat-NR2BAA (n=9)

aCSF (n=11)

days after implantation

Perc

enta

ge s

urvi

val

GcontrolH2O2StaurosporineNMDAAβ (1μM)Aβ (0.1μM)Aβ (1μM) + NMDAAβ (0.1μM) + NMDA

C

0

20

40

60

80

100

vehicle Tat-NR2BAA Tat-NR2B9c

** *

* *

## ######

####

##

##

########

#

#

% o

f dyi

ng c

ells

DwtTa

t-NR2B

AA

Tat-N

R2B9c

∆tau74

tau-/-

PSD-95

inpu

tIP

NR1

Gapdh

Tau-5

PSD-95NR1

E

Tat-N

R2BAA

Tat-N

R2B9c

0

1

2

3

4

5

6

7

Seiz

ure

Seve

rity

*B AβNMDAcontrol

vehi

cle

Tat-N

R2B

9cTa

t-NR

2BA

A

PI/Hoechst

Erro

rs (T

-maz

e)

H

0

4

3

2

1

Tat-N

R2B9c

untre

ated

aCSF

***

*

4mo 8mo

Figure 6. Peptide-Driven Uncoupling of the

NR/PSD-95 Interaction Reduces Ab Toxicity

and Improves Survival and Memory of

APP23 Mice

(A) Twenty-day-old primary cortical neurons were

pretreated with 100 nM Tat-NR2B9c peptide,

which disrupts the NR/PSD-95 interaction (Aarts

et al., 2002), prior to treatment with the toxins

NMDA, Ab, H2O2, and staurosporine. Twenty-

four hours after treatment, cell death was deter-

mined by propidium iodide (PI) uptake. Control

cells were pretreated with vehicle or 100 nM Tat-

NR2BAA (inactive peptide).

(B and C) Tat-NR2B9c (bottom row) significantly

reduces toxicity of NMDA and Ab to cortical

neurons, as indicated by lower numbers of PI-

positive cells (red; arrows) compared to pretreated

vehicle (top row) or Tat-NR2BAA (middle row)

controls. Nuclei were stained with Hoechst

(blue). Treatment with H2O2, staurosporine,

NMDA, Ab, or NMDA/Ab causes significant cell

death (#p < 0.005, ##p < 0.0001), which for

NMDA- and Ab-treated neurons is reduced by

pretreatment with Tat-NR2B9c, but not for H2O2-

and staurosporine-treated neurons (*p < 0.05,

**p < 0.01). One hundred cells each were counted

in three independent experiments. The scale bar

represents 25 mm.

(D) Immunoprecipitation (IP) with a PSD-95 antibody from hippocampus of WT mice that were i.c.v. infused with Tat-NR2B9c and Tat-NR2BAA for 1 week, and

from untreated WT, Dtau74, and tau�/� mice. Less NRs were coimmunoprecipitated upon Tat-NR2B9c, but not Tat-NR2BAA treatment. The reduction was

comparable to Dtau74 and tau�/� mice.

(E) One week of i.c.v. infusion of Tat-NR2B9c reduced PTZ-induced seizure severity significantly, compared to inactive Tat-NR2BAA (n = 10, *p < 0.01).

(F) APP23 mice treated with vehicle (artificial cerebrospinal fluid [aCSF]) alone or together with Tat-NR2B9c or Tat-NR2BAA, using osmotic mini pumps. DOB,

date of birth.

(G) Survival of APP23 mice upon i.c.v. delivery of Tat-NR2B9c (n = 17) is markedly improved compared to vehicle (aCSF)-treated (n = 11, p < 0.01) and Tat-

NR2BAA-treated (n = 9, p < 0.05) controls. Gray boxes indicate time of drug delivery from two consecutively implanted pumps.

(H) Tat-NR2B9c-treated APP23 mice show markedly improved memory functions at 4 and 8 months after initiating treatment, compared to aCSF-treated or age-

matched untreated APP23 mice (n = 8, n = 4 for aCSF, *p < 0.05, **p < 0.01).

Error bars represent the standard error.


into 6-week-old APP23 mice for i.c.v. delivery of artificial cere-

brospinal fluid (aCSF), with and without Tat-NR2B9c or Tat-

NR2BAA (Figure 6F). Mice in the aCSF and Tat-NR2BAA control

groups died frequently (7 of 11 and 4 of 9, respectively), whereas

only 1 of 17 mice died in the Tat-NR2B9c group (Figure 6E).

Finally, we tested whether Tat-NR2B9c-treatment has long-

term effects on memory in APP23 mice. The T maze revealed

comparable memory deficits in age-matched aCSF-treated

and untreated APP23 mice (Figure 6F). However, treatment

with Tat-NR2B9c resulted in a significantly improved perfor-

mance. Thus, perturbing NR/PSD-95 interaction is sufficient

to prevent premature lethality and memory deficits in APP23

mice.

DISCUSSION

Dendritic Localization of Fyn Is Tau-DependentOur data reveal a dendritic function of the ‘‘axonal’’ protein tau, in

targeting the kinase Fyn to the dendrite (Figure 7). We also found

an association of tau with the PSD complex by using coIP, PSD

purification, and immunohistochemistry with enhanced antigen

retrieval. It is important to note that levels of tau in the dendritic

compartment are much lower than in axons, suggesting that

under physiological conditions a major function of tau is in axonal

MT stabilization and regulation of MT-dependent processes

(Dixit et al., 2008; Weingarten et al., 1975). Here, we show that

the additional role of tau in dendrites becomes pivotal in disease,

in particular in mediating early Ab toxicity.

In both Dtau74 and tau�/� mice, dendritic targeting of Fyn is

significantly reduced, as revealed by immunohistochemistry

and synaptosomal purification and confirmed in primary

neurons. In Dtau74 mice, this is due to a competition of Dtau

with endogenous tau in the interaction with Fyn. Both the abun-

dance of Dtau in the cell body and its exclusion from dendrites

result in ‘‘trapping’’ of Fyn in the soma. Tau�/�mice, in compar-

ison, show a similar accumulation of Fyn, suggesting that post-

synaptic Fyn targeting requires tau. This difference in mediating

aberrant sorting of Fyn between Dtau74 and tau�/� mice

(Figure 7) may explain the additive effects on seizure suscepti-

bility and survival in Dtau.tau�/� crosses. Reduced levels of

postsynaptic Fyn in Dtau74 and tau�/� mice are associated

with reduced phosphorylation of the Fyn-substrate NR2b at

Y1472. Consistent with a critical role of Y1472 phosphorylation

in facilitating the interaction of NRs with PSD-95 (Rong et al.,

2001), this complex is reduced and destabilized in Dtau74 and

tau�/� brains. Whether the Fyn-mediated stabilization of NR/

PSD-95 complexes in the PSD under physiological conditions

involves a direct interaction with tau and what the exact mecha-

nism(s) of tau-mediated dendritic Fyn localization are remains to

be established.

In a rat model of stroke, targeted disruption of the NR/PSD-95

interaction prevented excitotoxic damage and reduced the

lesion size (Aarts et al., 2002). Consistent with this, reduced

NR/PSD-95 complexes in Dtau74 and tau�/� mice were associ-

ated with a reduced susceptibility to excitotoxicity. Interestingly,

NR-mediated currents were not affected in Dtau74 and tau�/�

mice, which is in line with normal synaptic activity upon treat-

ment with Tat-NR2B9c (Aarts et al., 2002). Normal NR-mediated

currents in Dtau74 and tau�/� mice may be explained by

reduced, but not totally depleted, synaptic Fyn in Dtau74 and

tau�/� mice, comparable to the situation in heterozygous fyn-

deficient mice that have no overt deficits (Yagi et al., 1993), while

in homozygous fyn-deficient mice these are pronounced (Grant

et al., 1992).

Dtau and tau�/� Prevent Deficits of APP23 MiceExcitotoxicity is increasingly recognized as a mechanism of how

Ab exerts toxicity in AD. Accordingly, we found that crossing

∆tau

P

Ca2+

FynTauPSD-95

PSD

NMDAR

NR2b

P

Ca2+

FynPSD-95

PSD

NMDAR

NR2b

∆tau74 tau-/-

P

Ca2+

FynTauPSD-95

PSD

NMDAR

NR2b

wtA B C

dend

ritic

spi

ne

Figure 7. Simplified Scheme of the

Proposed Mechanism Underlying Reduced

Excitotoxicity in Dtau74 and tau�/� Mice

Compared to the Wild-Type

(A) Postsynaptic NMDA receptors (NRs) are het-

eromeric complexes predominantly formed by

subunits NR1, NR2A, and NR2B. The Src kinase

Fyn localizes to the postsynapse in a tau-depen-

dent manner and associates with the postsynaptic

density (PSD; gray box), where it phosphorylates

(P) the NR subunit NR2b at Y1472 in the extreme

carboxy terminus. This phosphorylation facilitates

the interaction of NRs with the scaffolding protein

PSD-95. This interaction increases the stability of

NRs within the PSD and couples NRs to excito-

toxic downstream signaling (skull). NR-mediated

currents (ESPC trace), however, do not depend

on this NR/PSD-95 interaction. Whether tau is

associated with the PSD via Fyn or another inter-

action partner remains to be elucidated.

(B) In Dtau74 mice, Dtau is excluded from entering dendrites. Since Fyn interacts with Dtau (red bar) in the cell body of neurons, it is therefore trapped and less

localized to dendrites. Also, phosphorylation of NR2b and the interaction of NRs and PSD-95 are markedly reduced. Hence, excitotoxic downstream signaling is

uncoupled from NRs and their stability within the PSD is reduced. As NR-mediated currents are not dependent on this interaction, they are not affected.

(C) As for Dtau74 mice, in tau�/�mice, tau-dependent localization of Fyn to the postsynapse is also markedly reduced. NR2b phosphorylation and the interaction

of NRs and PSD-95 are decreased. Thus, excitotoxic downstream signaling is uncoupled from NRs, and their stability within the PSD is reduced. Again, NR-medi-

ated currents are not affected.


of Dtau74 and tau�/� mice, both characterized by reduced

susceptibility to excitotoxicity, with Ab-forming APP23 mice

ameliorated premature mortality and memory deficits of APP23

mice. In contrast, early lethality was more pronounced in

APP23 mice crossed with pR5. Similarly, tau deficiency or Dtau

expression conferred protection from Ab-induced toxicity in

primary neuronal cultures. However, Ab levels and plaque forma-

tion, as well as endogenous tau phosphorylation (in APP23.D

tau74), were comparable in APP23 mice, suggesting an alterna-

tive mechanism for protection. Interestingly, in APP23 mice we

found both increased postsynaptic Fyn and Y1472 phosphoryla-

tion of NR2b that was completely reverted in APP23.Dtau74 and

APP23.tau�/� mice. Further reduction of post-synaptic Fyn in

APP23.Dtau74.tau�/� mice suggests additional tau-indepen-

dent mechanisms in dendritic Fyn localization, which are partially

competed with by Dtau. Consistent with a role for Fyn in Ab

pathology, Fyn transgenic mice present with seizures and

premature mortality (Kojima et al., 1998). This is exacerbated in

Fyn/APPmut double-transgenic mice (Chin et al., 2004). More-

over, APP-associated mortality is reduced on a fyn�/� back-

ground (Chin et al., 2004). Hence, our findings in APP23.Dtau74

and APP23.tau�/� mice are consistent with previous data (Chin

et al., 2005; Chin et al., 2004). Furthermore, they are in line

with the recent observation that crossing of PDAPP mice onto

a tau�/� background reverses Ab-associated defects (Roberson

et al., 2007).

Mechanistically, our data suggest that stable NR/PSD-95

complex formation is required for Ab toxicity in APP23 mice.

This is likely to contribute to disease together with other tau-

dependent and -independent mechanisms of Ab toxicity. In

support of our findings, we used a tau/Fyn-independent

approach to disrupt this interaction, by delivering the Tat-

NR2B9c peptide to young APP23 mice. This peptide has been

shown to protect from excitotoxicity in vitro and in vivo.

We show specifically that perturbing the NR/PSD-95 interac-

tion with the Tat-NR2B9c peptide improves survival and

memory functions of APP23 mice. The data suggest that disrup-

tion of the NR/PSD-95 interaction is sufficient to prevent

Ab toxicity involving NR signaling. Remarkably, Tat-NR2B9c-

treated APP23 mice survived long term, suggesting that treat-

ment within a short therapeutic window is sufficient to prevent

lethality.

In summary, we reveal a dendritic role for the ‘‘axonal’’

protein tau in postsynaptic targeting of Fyn. This involves inter-

action of Fyn with the tau projection domain (Lee et al., 1998).

Accordingly, dominant negative effects of Dtau expression or

tau deficiency result in reduced postsynaptic Fyn, decreased

phosphorylation of its substrate NR2b and instability of NR/

PSD-95 complexes (in Dtau74 and tau�/� mice). Importantly,

this additional function of tau appears to be pivotal for medi-

ating Ab toxicity, in that premature lethality, memory deficits,

and seizure susceptibility of APP23 mice were mitigated in

APP23.Dtau74 and APP23.tau�/� mice. Hence, reduction of

tau levels or targeting of tau-dependent mechanisms, such as

the Fyn-mediated interaction of NRs and PSD-95, are suitable

strategies in the treatment of AD and related disorders, high-

lighting tau as an attractive drug target, in addition to Ab

(Ashe, 2007).


Animals

APP23 and pR5 transgenic and tau�/� mice have been generated previously

(Gotz et al., 2001a; Sturchler-Pierrat et al., 1997; Tucker et al., 2001). The

generation of Dtau74 mice is described in the Extended Experimental Proce-

dures. Two- to three-month-old mice were analyzed in age- and sex-matched

groups, unless stated otherwise. All animal experiments were approved by the

Animal Ethics Committee of the University of Sydney.

Histology, Western Blotting, IP, and Synaptosome Preparation

Detailed protocols are provided in the Extended Experimental Procedures.

Electrophysiology

Electrophysiological recording were done in acute hippocampal slices

obtained from 4- to 8-week-old wild-type, Dtau74, tau�/�, and Dtau74.tau�/�

mice as described in detail in the Extended Experimental Procedures.

Experimental Seizures

Seizures were induced by intraperitoneal (i.p.) injection of (50 mg/kg body

weight) pentylenetetrazole (PTZ; Sigma) as described (Roberson et al.,

2007). Where indicated, mice were injected i.p. with (0.1 mg/kg body weight)

MK801 (Sigma) 30 min prior to PTZ administration. Mice were video moni-

tored, and seizure severity was rated by an independent, blinded person, as

follows: 0, no seizures; 1, immobility; 2, tail extension; 3, forelimb clonus; 4,

generalized clonus; 5, bouncing seizures; 6, full extension; and 7, death.

i.c.v. Treatment with Osmotic Pumps

Six-week-old APP23 and WT mice were anesthetised with ketamine/xylazine,

and i.c.v. delivery cannulas (Alzet; brain infusion kit #3 with one spacer) were

implanted with a stereotaxic frame (KOPF Instruments) at the following

coordinates according to the bregma: AP, �0.25 mm; ML, 1 mm; and DV,

�2.5 mm. Osmotic mini pumps (Alzet; model #1004) were filled with aCSF

(Alzet) with and without Tat-NR2B9c or Tat-NR2BAA peptide (750 mM) and

equilibrated in 0.9% NaCl at 37�C for 48 hr. They were attached to the i.c.v.

cannula tubing and subcutaneously implanted at the back. After 28 days,

the pumps were replaced with a second batch of pumps via a small skin inci-

sion for another 28 days. Then, they were removed and the tubing was ligated.

Statistics

Statistics was done with the Prizm 4 software (GraphPad) with Student’s t or

two-way ANOVA test. Values are given as mean ± standard error.


Supplemental Information includes Extended Experimental Procedures and

four figures and can be found with this article online at doi:10.1016/j.cell.

2010.06.036.

ACKNOWLEDGMENTS

The authors thank Yves-Alain Barde for providing tau�/�mice and him as well

as Hanns Mohler and Nikolas Haass for helpful comments. This research was

supported by grants from the University of Sydney, National Health and

Medical Research Council, Australian Research Council, and Deutsche

Forschungsgemeinschaft. J.G. is a Medical Foundation Fellow.

Received: December 14, 2009

Revised: April 6, 2010

Accepted: May 28, 2010

Published online: July 22, 2010

REFERENCES

Aarts, M., Liu, Y., Liu, L., Besshoh, S., Arundine, M., Gurd, J.W., Wang, Y.T.,

Salter, M.W., and Tymianski, M. (2002). Treatment of ischemic brain damage


by perturbing NMDA receptor- PSD-95 protein interactions. Science 298,

846–850.

Ashe, K.H. (2007). A tale about tau. N. Engl. J. Med. 357, 933–935.

Ballatore, C., Lee, V.M., and Trojanowski, J.Q. (2007). Tau-mediated neurode-

generation in Alzheimer’s disease and related disorders. Nat. Rev. Neurosci. 8,

663–672.

Busche, M.A., Eichhoff, G., Adelsberger, H., Abramowski, D., Wiederhold,

K.H., Haass, C., Staufenbiel, M., Konnerth, A., and Garaschuk, O. (2008).

Clusters of hyperactive neurons near amyloid plaques in a mouse model of

Alzheimer’s disease. Science 321, 1686–1689.

Butner, K.A., and Kirschner, M.W. (1991). Tau protein binds to microtubules

through a flexible array of distributed weak sites. J. Cell Biol. 115, 717–730.

Calhoun, M.E., Wiederhold, K.H., Abramowski, D., Phinney, A.L., Probst, A.,

Sturchler-Pierrat, C., Staufenbiel, M., Sommer, B., and Jucker, M. (1998).

Neuron loss in APP transgenic mice. Nature 395, 755–756.

Chin, J., Palop, J.J., Yu, G.Q., Kojima, N., Masliah, E., and Mucke, L. (2004).

Fyn kinase modulates synaptotoxicity, but not aberrant sprouting, in human

amyloid precursor protein transgenic mice. J. Neurosci. 24, 4692–4697.

Chin, J., Palop, J.J., Puolivali, J., Massaro, C., Bien-Ly, N., Gerstein, H.,

Scearce-Levie, K., Masliah, E., and Mucke, L. (2005). Fyn kinase induces

synaptic and cognitive impairments in a transgenic mouse model of

Alzheimer’s disease. J. Neurosci. 25, 9694–9703.

Chishti, M.A., Yang, D.S., Janus, C., Phinney, A.L., Horne, P., Pearson, J.,

Strome, R., Zuker, N., Loukides, J., French, J., et al. (2001). Early-onset

amyloid deposition and cognitive deficits in transgenic mice expressing

a double mutant form of amyloid precursor protein 695. J. Biol. Chem. 276,

21562–21570.

Crowther, T., Goedert, M., and Wischik, C.M. (1989). The repeat region of

microtubule-associated protein tau forms part of the core of the paired helical

filament of Alzheimer’s disease. Ann. Med. 21, 127–132.

Dixit, R., Ross, J.L., Goldman, Y.E., and Holzbaur, E.L. (2008). Differential

regulation of dynein and kinesin motor proteins by tau. Science 319,

1086–1089.

Drechsel, D.N., Hyman, A.A., Cobb, M.H., and Kirschner, M.W. (1992). Modu-

lation of the dynamic instability of tubulin assembly by the microtubule-

associated protein tau. Mol. Biol. Cell 3, 1141–1154.

El Khoury, J., Toft, M., Hickman, S.E., Means, T.K., Terada, K., Geula, C., and

Luster, A.D. (2007). Ccr2 deficiency impairs microglial accumulation and

accelerates progression of Alzheimer-like disease. Nat. Med. 13, 432–438.

Geschwind, D.H. (2003). Tau phosphorylation, tangles, and neurodegenera-

tion: the chicken or the egg? Neuron 40, 457–460.

Goedert, M., Wischik, C.M., Crowther, R.A., Walker, J.E., and Klug, A. (1988).

Cloning and sequencing of the cDNA encoding a core protein of the paired

helical filament of Alzheimer disease: identification as the microtubule-associ-

ated protein tau. Proc. Natl. Acad. Sci. USA 85, 4051–4055.

Gotz, J., and Ittner, L.M. (2008). Animal models of Alzheimer’s disease and

frontotemporal dementia. Nat. Rev. Neurosci. 9, 532–544.

Gotz, J., Chen, F., Barmettler, R., and Nitsch, R.M. (2001a). Tau filament

formation in transgenic mice expressing P301L tau. J. Biol. Chem. 276,

529–534.

Gotz, J., Chen, F., van Dorpe, J., and Nitsch, R.M. (2001b). Formation of neuro-

fibrillary tangles in P301l tau transgenic mice induced by Abeta 42 fibrils.

Science 293, 1491–1495.

Grant, S.G., O’Dell, T.J., Karl, K.A., Stein, P.L., Soriano, P., and Kandel, E.R.

(1992). Impaired long-term potentiation, spatial learning, and hippocampal

development in fyn mutant mice. Science 258, 1903–1910.

Haass, C., Schlossmacher, M.G., Hung, A.Y., Vigo-Pelfrey, C., Mellon, A.,

Ostaszewski, B.L., Lieberburg, I., Koo, E.H., Schenk, D., Teplow, D.B., et al.

(1992). Amyloid beta-peptide is produced by cultured cells during normal

metabolism. Nature 359, 322–325.

Hsiao, K., Chapman, P., Nilsen, S., Eckman, C., Harigaya, Y., Younkin, S.,

Yang, F., and Cole, G. (1996). Correlative memory deficits, Abeta elevation,

and amyloid plaques in transgenic mice. Science 274, 99–102.

Kelly, P.H., Bondolfi, L., Hunziker, D., Schlecht, H.P., Carver, K., Maguire, E.,

Abramowski, D., Wiederhold, K.H., Sturchler-Pierrat, C., Jucker, M., et al.

(2003). Progressive age-related impairment of cognitive behavior in APP23

transgenic mice. Neurobiol. Aging 24, 365–378.

Kim, E., and Sheng, M. (2004). PDZ domain proteins of synapses. Nat. Rev.

Neurosci. 5, 771–781.

Kojima, N., Ishibashi, H., Obata, K., and Kandel, E.R. (1998). Higher seizure

susceptibility and enhanced tyrosine phosphorylation of N-methyl-D-aspar-

tate receptor subunit 2B in fyn transgenic mice. Learn. Mem. 5, 429–445.

Kornau, H.C., Schenker, L.T., Kennedy, M.B., and Seeburg, P.H. (1995).

Domain interaction between NMDA receptor subunits and the postsynaptic

density protein PSD-95. Science 269, 1737–1740.

Ksiezak-Reding, H., and Yen, S.H. (1991). Structural stability of paired helical

filaments requires microtubule-binding domains of tau: a model for self-asso-

ciation. Neuron 6, 717–728.

Kuo, Y.M., Beach, T.G., Sue, L.I., Scott, S., Layne, K.J., Kokjohn, T.A.,

Kalback, W.M., Luehrs, D.C., Vishnivetskaya, T.A., Abramowski, D., et al.

(2001). The evolution of A beta peptide burden in the APP23 transgenic

mice: implications for A beta deposition in Alzheimer disease. Mol. Med. 7,

609–618.

Lalonde, R., Dumont, M., Staufenbiel, M., and Strazielle, C. (2005). Neurobe-

havioral characterization of APP23 transgenic mice with the SHIRPA primary

screen. Behav. Brain Res. 157, 91–98.

Lee, G., Cowan, N., and Kirschner, M. (1988). The primary structure and

heterogeneity of tau protein from mouse brain. Science 239, 285–288.

Lee, G., Newman, S.T., Gard, D.L., Band, H., and Panchamoorthy, G. (1998).

Tau interacts with src-family non-receptor tyrosine kinases. J. Cell Sci. 111,

3167–3177.

Leissring, M.A., Farris, W., Chang, A.Y., Walsh, D.M., Wu, X., Sun, X., Frosch,

M.P., and Selkoe, D.J. (2003). Enhanced proteolysis of beta-amyloid in APP

transgenic mice prevents plaque formation, secondary pathology, and

premature death. Neuron 40, 1087–1093.

Lewis, J., Dickson, D.W., Lin, W.L., Chisholm, L., Corral, A., Jones, G., Yen,

S.H., Sahara, N., Skipper, L., Yager, D., et al. (2001). Enhanced neurofibrillary

degeneration in transgenic mice expressing mutant tau and APP. Science 293,

1487–1491.

Maas, T., Eidenmuller, J., and Brandt, R. (2000). Interaction of tau with the

neural membrane cortex is regulated by phosphorylation at sites that are

modified in paired helical filaments. J. Biol. Chem. 275, 15733–15740.

Minkeviciene, R., Rheims, S., Dobszay, M.B., Zilberter, M., Hartikainen, J.,

Fulop, L., Penke, B., Zilberter, Y., Harkany, T., Pitkanen, A., and Tanila, H.

(2009). Amyloid beta-induced neuronal hyperexcitability triggers progressive

epilepsy. J. Neurosci. 29, 3453–3462.

Mucke, L., Masliah, E., Yu, G.Q., Mallory, M., Rockenstein, E.M., Tatsuno, G.,

Hu, K., Kholodenko, D., Johnson-Wood, K., and McConlogue, L. (2000). High-

level neuronal expression of abeta 1-42 in wild-type human amyloid protein

precursor transgenic mice: synaptotoxicity without plaque formation.

J. Neurosci. 20, 4050–4058.

Nakazawa, T., Komai, S., Tezuka, T., Hisatsune, C., Umemori, H., Semba, K.,

Mishina, M., Manabe, T., and Yamamoto, T. (2001). Characterization of Fyn-

mediated tyrosine phosphorylation sites on GluR epsilon 2 (NR2B) subunit of

the N-methyl-D-aspartate receptor. J. Biol. Chem. 276, 693–699.

Palop, J.J., and Mucke, L. (2009). Epilepsy and cognitive impairments in

Alzheimer disease. Arch. Neurol. 66, 435–440.

Palop, J.J., Chin, J., Roberson, E.D., Wang, J., Thwin, M.T., Bien-Ly, N., Yoo,

J., Ho, K.O., Yu, G.Q., Kreitzer, A., et al. (2007). Aberrant excitatory neuronal

activity and compensatory remodeling of inhibitory hippocampal circuits in

mouse models of Alzheimer’s disease. Neuron 55, 697–711.

Phillips, G.R., Huang, J.K., Wang, Y., Tanaka, H., Shapiro, L., Zhang, W., Shan,

W.S., Arndt, K., Frank, M., Gordon, R.E., et al. (2001). The presynaptic particle


web: ultrastructure, composition, dissolution, and reconstitution. Neuron 32,

63–77.

Roberson, E.D., Scearce-Levie, K., Palop, J.J., Yan, F., Cheng, I.H., Wu, T.,

Gerstein, H., Yu, G.Q., and Mucke, L. (2007). Reducing endogenous tau

ameliorates amyloid beta-induced deficits in an Alzheimer’s disease mouse

model. Science 316, 750–754.

Roche, K.W., Standley, S., McCallum, J., Dune Ly, C., Ehlers, M.D., and Went-

hold, R.J. (2001). Molecular determinants of NMDA receptor internalization.

Nat. Neurosci. 4, 794–802.

Rong, Y., Lu, X., Bernard, A., Khrestchatisky, M., and Baudry, M. (2001). Tyro-

sine phosphorylation of ionotropic glutamate receptors by Fyn or Src differen-

tially modulates their susceptibility to calpain and enhances their binding to

spectrin and PSD-95. J. Neurochem. 79, 382–390.

Roskam, S., Neff, F., Schwarting, R., Bacher, M., and Dodel, R. (2010). APP

transgenic mice: the effect of active and passive immunotherapy in cognitive

tasks. Neurosci. Biobehav. Rev. 34, 487–499.

Salter, M.W., and Kalia, L.V. (2004). Src kinases: a hub for NMDA receptor

regulation. Nat. Rev. Neurosci. 5, 317–328.

Selkoe, D.J. (1997). Alzheimer’s disease: genotypes, phenotypes, and treat-

ments. Science 275, 630–631.

Selkoe, D.J. (2002). Alzheimer’s disease is a synaptic failure. Science 298,

789–791.

Shankar, G.M., Li, S., Mehta, T.H., Garcia-Munoz, A., Shepardson, N.E.,

Smith, I., Brett, F.M., Farrell, M.A., Rowan, M.J., Lemere, C.A., et al. (2008).

Amyloid-beta protein dimers isolated directly from Alzheimer’s brains impair

synaptic plasticity and memory. Nat. Med. 14, 837–842.

Sturchler-Pierrat, C., Abramowski, D., Duke, M., Wiederhold, K.H., Mistl, C.,

Rothacher, S., Ledermann, B., Burki, K., Frey, P., Paganetti, P.A., et al.

(1997). Two amyloid precursor protein transgenic mouse models with Alz-

heimer disease-like pathology. Proc. Natl. Acad. Sci. USA 94, 13287–13292.

Tezuka, T., Umemori, H., Akiyama, T., Nakanishi, S., and Yamamoto, T. (1999).

PSD-95 promotes Fyn-mediated tyrosine phosphorylation of the N-methyl-D-

aspartate receptor subunit NR2A. Proc. Natl. Acad. Sci. USA 96, 435–440.

Tucker, K.L., Meyer, M., and Barde, Y.A. (2001). Neurotrophins are required for

nerve growth during development. Nat. Neurosci. 4, 29–37.

Van Dam, D., D’Hooge, R., Staufenbiel, M., Van Ginneken, C., Van Meir, F., and

De Deyn, P.P. (2003). Age-dependent cognitive decline in the APP23 model

precedes amyloid deposition. Eur. J. Neurosci. 17, 388–396.

Weingarten, M.D., Lockwood, A.H., Hwo, S.Y., and Kirschner, M.W. (1975).

A protein factor essential for microtubule assembly. Proc. Natl. Acad. Sci.

USA 72, 1858–1862.

Yagi, T., Aizawa, S., Tokunaga, T., Shigetani, Y., Takeda, N., and Ikawa, Y.

(1993). A role for Fyn tyrosine kinase in the suckling behaviour of neonatal

mice. Nature 366, 742–745.

Zhao, L., Ma, Q.L., Calon, F., Harris-White, M.E., Yang, F., Lim, G.P., Morihara,

T., Ubeda, O.J., Ambegaokar, S., Hansen, J.E., et al. (2006). Role of p21-acti-

vated kinase pathway defects in the cognitive deficits of Alzheimer disease.

Nat. Neurosci. 9, 234–242.


Single-Stranded DNA TranspositionIs Coupled to Host ReplicationBao Ton-Hoang,1,* Cecile Pasternak,2 Patricia Siguier,1 Catherine Guynet,1 Alison Burgess Hickman,3 Fred Dyda,3

Suzanne Sommer,2 and Michael Chandler1,*1Laboratoire de Microbiologie et Genetique Moleculaires, Centre National de Recherche Scientifique, Unite Mixte de Recherche 5100,

118 Route de Narbonne, F31062 Toulouse Cedex, France2Universite Paris-Sud, Centre National de Recherche Scientifique, Unite Mixte de Recherche 8621, Laboratoire de Recherche Correspondantdu Commissariat a l’Energie Atomique 42V, Institut de Genetique et Microbiologie, Batiment 409, Orsay, France3Laboratory of Molecular Biology, National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, Bethesda,

MD 20892, USA

*Correspondence: [email protected] (B.T-H.), [email protected] (M.C.)DOI 10.1016/j.cell.2010.06.034

SUMMARY

DNA transposition has contributed significantly toevolution of eukaryotes and prokaryotes. Insertionsequences (ISs) are the simplest prokaryotic trans-posons and are divided into families on the basisof their organization and transposition mechanism.Here, we describe a link between transposition ofIS608 and ISDra2, both members of the IS200/IS605 family, which uses obligatory single-strandedDNA intermediates, and the host replication fork.Replication direction through the IS plays a crucialrole in excision: activity is maximal when the ‘‘top’’IS strand is located on the lagging-strand template.Excision is stimulated upon transient inactivation ofreplicative helicase function or inhibition of Okazakifragment synthesis. IS608 insertions also exhibit anorientation preference for the lagging-strand tem-plate and insertion can be specifically directed tostalled replication forks. An in silico genomicapproach provides evidence that dissemination ofother IS200/IS605 family members is also linked tohost replication.

INTRODUCTION

DNA transposition involves movement of discrete DNA seg-

ments (transposons) from one genomic location to another.

It occurs in all kingdoms of life and has contributed significantly

to evolution of eukaryotes and prokaryotes. Transposable

elements can represent a significant proportion of their host

genomes (Biemont and Vieira, 2006). They have been particularly

well studied in bacteria where they are major motors of broad

genome remodeling, play an important role in horizontal gene

transfer, and can sequester and transmit a variety of genes

involved in accessory cell functions, such as resistance to anti-

microbial agents, catabolism of unusual compounds, and path-

ogenicity, virulence, or symbiosis. They are also important as

genetic tools in identifying specific gene regulatory regions by

insertion and are being developed as delivery systems for gene

therapy applications.

A variety of structurally and mechanistically distinct enzymes

(transposases) have evolved to carry out transposition by several

different pathways (Turlan and Chandler, 2000; Curcio and

Derbyshire, 2003). They all possess an endonuclease activity

allowing them to cleave, excise, and insert transposon DNA

into a new location. Depending on the system (Curcio and Derby-

shire, 2003), different types of nucleophile can be used by trans-

posases to attack a phosphorus atom of a backbone phospho-

diester bond and cleave DNA. These include water (generally

activated by enzyme-bound metal ions), a hydroxyl group at

the 50 or 30 end of a DNA strand, or a hydroxyl group of an amino

acid of the transposase itself, such as serine or tyrosine.

Many mobile DNA elements move using a ‘‘cut-and-paste’’

mechanism by excision of a double-stranded copy from one

genomic location and insertion at another. Recently, a family of

bacterial insertion sequences (ISs), the IS200/IS605 family, has

been found that uses a completely different pathway and an

unusual transposase with a catalytic tyrosine (a Y1 transposase).

Studies of one member, IS608 (Figure 1A), provided a detailed

picture of their transposition (Ton-Hoang et al., 2005; Ronning

et al., 2005; Guynet et al., 2008; Barabas et al., 2008). In vitro,

this requires single-stranded DNA (ssDNA) substrates and is

strand specific: only the ‘‘top’’ strand is recognized by the

element-encoded transposase, TnpA, and is cleaved and trans-

ferred, whereas the ‘‘bottom’’ strand does not transpose. Exci-

sion of the top strand as a transposon circle with joined left

and right ends is accompanied by rejoining of the DNA flanks.

The circle junction then undergoes TnpA-catalyzed integration

into an ssDNA target in a sequence-specific reaction. Insertion

involves transfer of both the 50 and 30 ends of the single-strand

circle junction into the ssDNA target. The left (50) IS608

end always inserts specifically just 30 of the tetranucleotide,

50-TTAC-30 (Kersulyte et al., 2002), which is also essential for

subsequent transposition (Ton-Hoang et al., 2005).

The obligatorily single-stranded nature of IS200/IS605 trans-

position in vitro raises the possibility that it is limited in vivo by

the availability of its ssDNA substrates. A number of cellular


processes generate or occur using ssDNA, including DNA repair,

natural transformation, conjugative plasmid transfer, single-

stranded phage infection, and replication (where the DNA

serving as the template for Okazaki fragment synthesis on the

lagging strand of the replication fork is single stranded).

Here, we investigate the link between IS608 transposition and

the availability of ssDNA during replication in vivo. Our results

demonstrate that transposition of the IS200/IS605 family is

closely integrated into the host cell cycle and takes advantage

of the presence of ssDNA on the lagging-strand template at

the replication fork for dissemination. We also show that IS608

transposition is affected by perturbing the fork: transitory inacti-

vation of crucial replication proteins increased excision from the

lagging-strand template, and stalling of the fork resulted in inser-

tions directed to the lagging strand of the blocked fork.

Our results also suggest that insertion and excision of the

related element, ISDra2, also depends on the lagging-strand

template in its host, the radiation-resistant bacterium Deinococ-

cus radiodurans, and that this dependency can be abolished

with irradiation. We have extended our analysis to a number of

related IS200/IS605 elements in a variety of sequenced bacterial

genomes. The results of this in silico analysis are also consistent

with a strong bias of insertion into the lagging-strand template in

these organisms. Together, the results demonstrate the impor-

tance of the lagging-strand template for IS608 and ISDra2

activity and suggest that all IS200/IS605 family members have

evolved a mode of transposition that exploits ssDNA at the repli-

cation fork.

RESULTS

IS608 Excision Depends on the Direction of ReplicationTo investigate whether replication direction affects IS608 trans-

position, we used a plasmid assay in E. coli to monitor the exci-

sion step of transposition (Ton-Hoang et al., 2005). In this assay,

the IS-carrying plasmids included an IS608 derivative in which

the tnpA and tnpB genes (Figure 1A) were replaced by a chloram-

phenicol resistance (CmR) cassette (Figure 1C). In one case, the

active (top) IS608 strand was located in the lagging-strand

template (pBS102), and in the second, the replication origin

was inverted (Figures 1B and 1C), placing the transposionally

active top strand on the leading-strand template (pBS144).

A second compatible plasmid supplied TnpA in trans under

Figure 1. IS608 Excision Depends on the Direction of Replication

(A) IS608 organization and a simplified transposition model. Gray arrows, tnpA

and tnpB orfs; red and blue boxes, left (LE) and right (RE) ends (color code

retained throughout).

(Ai) Schematized single-stranded IS608 showing secondary structures of LE

and RE, the flanking TTAC, and cleavage positions at the ends (vertical black

arrows).

(Aii) Excision and ssDNA circle formation with an RE-LE junction and a sealed

donor joint (black line) retaining TTAC.

(Aiii) TnpA brings together the transposon junction with a new TTAC-carrying

target (dotted black line). Vertical black arrows, points of cleavage and strand

transfer.

(Aiv) IS608 insertion into the target.

(B) Orientation of the IS608 derivative with respect to replication direction. The

disposition of the IS608 active (top) strand with respect to replication direction

is shown when the fork approaches from one direction (left) when it is part of

the lagging-strand template or the other (right) when it is part of the leading

strand. This is described in more detail in Figure S1 and its legend.

(C) Excision measured directly in vivo by the appearance of ‘‘donor joint’’ plas-

mids, deleted for the IS608 derivative. Left: pBS102 with the active IS608

strand as part of the lagging-strand template. Right: pBS144 with the active

IS608 strand as part of the leading-strand template. Ori, pBR322 origin of repli-

cation; Cm, bla, SpSm, chloramphenicol, b-lactamase and, streptomycin/

spectinomycin resistance genes; Plac, lac promoter; tnpA-his, C-terminal

his6-tagged tnpA gene. Directions of DNA replication and transcription are

indicated. Agarose gel (0.8%) showing separation of plasmid DNA from over-

night cultures of strains carrying pBS102 + pBS21 (lane 2) and pBS144 +

pBS121 (lane 3). Lane 1, 1 kb standard.

(D) Mating-out assays. Left-hand column, transposon donor plasmid; middle,

presence or absence of TnpA (relevant plasmids in parentheses); right,

measured transposition frequencies (standard error in parentheses; n > 3).

Plasmids pBS102ter1 and pBS144ter1 carry a single set of origin proximal

terminators; pBS102ter2 and pBS144ter2 carry two sets of terminators flank-

ing both transposon ends (Extended Experimental Procedures).

See also Figure S1A.


control of plac (Figure 1C). After overnight growth, IS excision

was monitored by detection of reclosed donor backbone mole-

cules from which the IS had been deleted.

As shown in Figure 1C, when the IS608 active strand was

located on the lagging-strand template, the donor backbone

species could be clearly identified along with the parental

plasmid and the plasmid used to supply transposase (lane 2).

However, when the replication origin was inverted and the active

IS strand was located on the leading strand, formation of the

excised donor backbone species was only barely detectable

(lane 3).

This effect of replication direction on IS608 transposition was

also observed in mating-out assays (Galas and Chandler, 1982)

that measure overall transposition frequency. We monitored

transposition was monitored by following movement of IS608

from a nonmobilizable donor plasmid into a conjugative plas-

mid. When the IS608 active strand was on the lagging-strand

template (Figure 1D, line 2), the transposition frequency was

5.6 3 10�5, but when it was on the leading-strand template,

the frequency dropped 27-fold to 2.1 3 10�6 (line 3). To ensure

that this was not due to possible changes in transcription

resulting from the inversion introduced during cloning to

switch the orientation of the replication origin (where bla was

inverted together with ori), we inserted transcriptional termina-

tors (Simons et al., 1987) on either one (lines 4–5) or both (lines

6–7) sides of the IS608 derivatives to insulate them from

impinging transcription. In these cases, the observed effect of

replication direction on transposition frequency remained

unchanged.

Effect of Mutant Primase and Replicative Helicaseon IS ExcisionSince IS608 excision is sensitive to replication direction, we

asked whether it was affected by perturbation of the replication

fork. We used two temperature-sensitive replication mutants:

dnaG308ts, encoding a mutant DNA primase, DnaG (Wechsler

and Gross, 1971), and dnaB8ts, encoding a mutant of the essen-

tial replication fork DNA helicase, DnaB (Carl, 1970). Replication

in these mutants occurs at 30�C but is interrupted after a shift to

the restrictive temperature of 42�C. Inhibition of either DnaG or

DnaB activity is expected to increase the amount of ssDNA at

the fork, principally on the lagging-strand template (Louarn,

1974; Fouser and Bird, 1983; Belle et al., 2007) (see also the

Discussion).

To test the effect of inhibition of DnaG and DnaB activities, we

used a genetic screen in which a b-lactamase gene is interrupted

by an IS608 derivative (pAM1, Figure S2; Experimental Proce-

dures). TnpA-catalyzed precise excision (using TnpA provided

in trans) results in reconstitution of the b-lactamase gene

(Figure 2A) and the appearance of ampicillin resistant (ApR)

colonies.

As shown in Figures 2Bi, 2Bii, and 2Biii for the wild-type and

dnaB and dnaG mutants, respectively, after overnight growth

at 30�C (a permissive temperature) without TnpA induction, the

frequency of ApR colonies was low for the wild-type and mutant

hosts (column a). Induction of TnpA expression resulted in

a nearly 3-fold increase in excision with little difference between

wild-type, dnaBts, and dnaGts strains (column b). However,

when the growth protocol was modified to include a 30 min

temperature shift to 42�C (and further incubation of 3 hr at

30�C to allow replication to recover), excision increased about

10- and 7-fold in the dnaB and dnaG mutants, respectively,

compared to the wild-type (column d). Omission of the 42�Cstep resulted in indistinguishable basal levels for wild-type and

mutant hosts (column c). Thus, inactivation of DnaB helicase

function or inhibition of initiation of Okazaki fragment synthesis

with a dnaGts mutation stimulated the excision step of transpo-

sition, consistent with the notion that ssDNA at the replication

fork is a substrate for IS excision.

Effect of Transposon Size on ExcisionIf excision occurs at the replication fork and requires ssDNA, it

seemed possible that IS length might influence excision

frequency since the probability that both IS ends are within the

single-stranded region of the lagging-strand template should

decrease with increasing IS length. We therefore examined the

effect of IS length in the excision assay by using a set of IS608

derivatives with varying spacing between the left end (LE) and

right end (RE) and found that excision decreased strongly as

a function of increasing IS length and that these frequencies

were strongly modified in a strain carrying a dnaGts mutation.

Increasing the IS length from 0.3 to 2 kb, resulted in a 7- to

10-fold decrease in excision frequency with a log-linear relation-

ship, consistent with the notion that excision occurs more effi-

ciently when both ends are located within the short single

lagging-strand region of about 1.5–2 kb upstream of the first

complete Okazaki fragment at the replication fork (see Johnson

and O’Donnell, 2005, for review). As the IS length was further

increased, excision decreased only slightly, at least up to a trans-

poson length of 4 kb (Figure 2C) with a possible inflection

between 1.5 and 2 kb.

The length of ssDNA on the lagging-strand template depends

on the initiation frequency of Okazaki fragment synthesis, in turn

determined by DnaG. Progressive inactivation of DnaG activity

by growth of the dnaGts mutant at increasing but sublethal

temperatures should reduce this frequency and increase the

mean length of ssDNA at the replication fork upstream of the first

complete Okazaki fragment. We therefore analyzed excision of

the IS608-derived transposons in wild-type and dnaGts mutants

at different temperatures (Figure 2D). While profiles were indis-

tinguishable for the wild-type strain at 30�C and 33�C, the

dnaGts mutant exhibited a generally higher excision frequency

at 30�C and showed a lower length dependent slope, revealing

that the replication fork is affected by the mutation even at the

normal permissive temperature. However, an inflection still

appeared to occur. At 33�C, excision increased significantly,

particularly for the longer transposons.

To further examine this, we cloned the wild-type dnaG allele

downstream of the tnpAIS608 gene in the TnpA-providing plasmid

so that both were under control of the same promoter (Experi-

mental Procedures). When this plasmid, pBS179, was intro-

duced into the dnaGts strain, it clearly suppressed the dnaGts

defect at 33�C (Figure 2E). Moreover, when introduced into the

wild-type dnaG strain, the excision frequencies were even

further reduced and the length-dependant slope was increased

(Figure 2F).


IS608 Insertion into the E. coli ChromosomeThe circular E. coli chromosome replicates bidirectionally from

the replication origin, oriC. If IS608 insertions target the

lagging-strand template, they should occur in one orientation

on one side of ori and in the opposite orientation on the other.

To test this, we isolated IS608 insertions in the E. coli

Figure 2. Excision Frequency: Effects of dnaBts and dnaGts Mutants and IS Length

(A) Schematic of the excision assay. The relevant features of the generic IS-carrying donor plasmid, pAM1, and the product after IS excision, DpAM1, are shown

together with the plasmid used to supply transposase, pBS135. These are described in detail in Figure S1 and its legend.

(B) Transitory inactivation of DnaB or DnaG. Stimulation factors for excision from wild-type, dnaBts, and dnaGts strains: i, ii, and iii, respectively. Values are

normalized to those of an overnight culture without IPTG at 30�C for each strain (0.8 3 10�2, 0.6 3 10�2, and 2.2 3 10�2 corresponding to column a), overnight

cultures diluted at 30�C with IPTG for 4 hr (column b), TnpA induction at 30�C (45 min) and incubation at 30�C (3 hr 30 min) without IPTG (column c), and TnpA

induction at 30�C (45 min) followed by a shift to 42�C without IPTG (30 min) and then at 30�C (3 hr) (column d).

(C) Effect of IS length on excision frequency. IS608 derivatives are 0.3, 0.5, 0.8, 1.1, 1.4, 1.9, 3, and 4 kb long. The curve shows results expressed as ApRTpRKmR /

TcRKmR with DH5a at 37�C. Error bars are the standard deviation (SD) of more than four independent experiments.

(D) Effect of dnaG on IS length-dependant excision. MC4100, 30�C (dark blue) and 33�C (yellow); MC4100dnaGts, 30�C (light blue); and MC4100dnaGts+dnaGwt,

33�C (purple).

(E) Effect of DnaGwt overexpression on excision frequency in dnaGts strains. MC4100dnaGts, 33�C (purple), and MC4100dnaGts+dnaGwt, 33�C (light blue).

(F) Effect of DnaGwt overexpression on excision frequency in wild-type strains. MC4100, 33�C (yellow), and MC4100+dnaGwt, 33�C (purple).

See also Figure S1B.


chromosome using a temperature-sensitive plasmid as the IS

donor and supplying TnpA in trans (Figure 3A). Insertions were

localized by an arbitrary PCR procedure (Experimental Proce-

dures) followed by DNA sequencing. We observed a dramatic

skew in strand specificity of IS608 insertion relative to the origin

of replication, oriC.

Insertions either in the left or right replicores were in the

orientation expected for transposition into the lagging-strand

template (Figure 3B). It is interesting to note that while insertions

were distributed around the chromosome, many appeared in the

vicinity of the highly transcribed rRNA genes (Table S1).

We also obtained insertions into the TnpA donor plasmid in the

same experiment. In contrast to the chromosome, this plasmid

replicates unidirectionally and the IS608 insertion pattern was

completely (Figure 3C) different. All occurred into only one

strand, the lagging-strand template, a result that has strong

statistical support (Figure 3 legend).

In all cases, both plasmid and chromosome insertions

occurred 30 to a TTAC target, as expected. As these are distrib-

uted equally on both strands of the chromosome and the TnpA

donor plasmid (data not shown), the observed strand biases

for insertion cannot be explained by a bias in target sequence

distribution.

Targeting Stalled Plasmid Replication Forks:Tus-Ter SystemSince our accumulating data suggested that IS608 targets

ssDNA at the replication fork, we asked whether insertion could

also be observed into forks stalled at a predefined location. For

this, we used the E. coli Tus/Ter system in which a Ter site, when

bound by the protein Tus, strongly reduces replication fork

progression through Ter in the nonpermissive orientation (Ternp)

(see Bierne et al., 1994) but not in the opposite, permissive,

orientation (Terp) (Neylon et al., 2005).

The assay (Figure 4A and Figure S1) used a suicide conjuga-

tion system (Demarre et al., 2005) in which the unidirectionally

replicating target plasmid was carried by a recipient strain with

an inactivated chromosomal tus gene. The target plasmid

carried a functional tus gene and a Ter site in either the permis-

sive or nonpermissive orientation (Figures 4B and Extended

Experimental Procedures). The plasmid-based tus gene was

transitorily induced only for the duration of the experiment to

avoid plasmid loss. We also inserted a stretch of DNA with

several spaced copies of the TTAC target tetranucleotide on

both strands upstream of the Ter sites.

Mapping of the IS insertion sites revealed that all occurred at

TTAC sequences and were distributed over the entire length of

both target plasmids, largely in an orientation expected for inser-

tion into the lagging-strand template (Figure 4B, black arrow-

heads). The overall distribution was the same regardless of Ter

site orientation. As for the other target plasmids used here, we

confirmed that TTAC sequences were distributed roughly

equally on both DNA strands (data not shown).

An additional set of insertions was observed in the target

plasmid carrying Ternp when compared to that with Terp (Figures

4B and 4C). To precisely map these, we used PCR analysis with

primers complementary to a sequence upstream of the Ter sites

and either LE (for lagging-strand insertions) or RE (for leading-

strand insertions). When the target plasmid carried Ternp, ampli-

fication products from four independent experiments (Figure 4C,

lanes 1–4) revealed insertions adjacent to the Ter site on the

lagging-strand template (Figure 4B, filled red arrowheads; Fig-

ure 4C), although occasionally an insertion was observed on

the opposite strand (lane 5). Although there was a degree of vari-

ability in the insertion distribution between experiments (com-

pare lanes 1–4), a consistently strong signal was obtained from

the TTAC site located at 63 bp from Ternp, suggesting that this

is the preferred insertion site. However, insertions were also

observed at the closest site, located only 26 bp from Ter

(Figure 4C, lanes 1, 3, and 4).

With Terp, no strong Tus-dependent insertions were observed

(Figure 4D). In four independent experiments we only once iso-

lated an insertion in this region (in the plasmid population from

Figure 3. Insertions into the E. coli Chromosome and into a Resident

Plasmid

(A) Experimental system. The temperature-sensitive transposon donor

plasmid, pBS156 (middle), and the plasmid used for supplying TnpA,

pBS135 (left), are shown together with a schematic of the host chromosome

(right). Growth at the nonpermissive temperature results in loss of pBS156

and IS insertion into the chromosome or pBS135. This is described in detail

in Figure S1 and its legend.

(B) Chromosomal insertions. The chromosome is shown linearized at its

terminus. Red spot, origin of replication, ori; rrn, ribosomal RNA genes; black

arrows, positions of insertions (Table S1). Multiple insertions are indicated by

a number above the larger arrows. Replication occurs in both directions into

the left and right replicores, blue and orange, respectively: the lagging-strand

template of the left (bottom) and right (top) replicores. Fisher’s exact test gives

a p value of 7.122e�6.

(C) Plasmid insertions. Replication of the p15A-based plasmid occurs from left

to right. Orange triangle, origin of replication; KmR, tnpA, and lacIQ, kanamycin

resistance, transposase and lac repressor genes, respectively. Arrows indi-

cate their direction of transcription. Vertical arrowheads show the position of

the insertions obtained from the same experiments as in (A) above.

See also Figure S1C and Table S1.


experiment 1, lane 1), and this had occurred into a TTAC site

within the Ter sequence itself rather than immediately upstream.

Thus, the lagging strand of replication forks blocked by Tus is

a preferential target for IS608 insertions and these can occur in

close proximity to the Tus binding site, Ter.

Another IS200/IS605 Family Member: ISDra2

ISDra2 is an IS200/IS605 family member from the highly radiation

resistant D. radiodurans. It has a similar organization to IS608

and inserts specifically 30 to a TTGAT pentanucleotide (Islam

et al., 2003). Like IS608, it transposes using ssDNA intermedi-

ates (Pasternak et al., 2010). We had observed specific stimula-

tion of ISDra2 transposition upon g- or UV-irradiation (Mennecier

et al., 2006; Pasternak et al., 2010) related to the availability of

ssDNA generated during radiation-induced fragmentation and

reassembly of the host genome, called extended synthesis

dependent strand annealing (ESDSA) (Zahradka et al., 2006).

To determine whether spontaneous ISDra2 transposition is

influenced by host chromosome replication, we measured

ISDra2 excision in D. radiodurans as a function of replication

direction through the IS. D. radiodurans replication is thought

to be bidirectional (Hendrickson and Lawrence, 2006). We con-

structed D. radiodurans strains with an ISDra2 derivative,

ISDra2-113 (Pasternak et al., 2010), placed in one or the other

orientation at the same chromosomal location (Extended Exper-

imental Procedures). Excision of these ISs reconstitutes a func-

tional TcR gene. Consistent with our results with IS608, no exci-

sion (<4.8 3 10�9; Figure 5A) was detected without transposase,

whereas in the presence of TnpAISDra2, relatively high excision

levels (2.1 3 10�3) were observed when the IS derivative was

in one orientation but was reduced about 30-fold to 6.7 3 10�5

in the other (Figure 5A).

However, the orientation bias of excision was lost in g-irradi-

ated D. radiodurans strains: after g-irradiation, excision of

ISDra2-113 reached the same and maximal level of about 2 3

10�2 for both orientations relative to the direction of replication

(Figure 5A). Since there should be no strand bias in generating

ssDNA after irradiation during ESDSA, the lack of excision and

insertion bias in irradiated D. radiodurans suggests that insertion

targets are no longer limited to those rendered accessible during

the replication.

Finally, we isolated 23 independent spontaneous insertions of

ISDra2-113 in D. radiodurans chromosome I by scoring for TcR

CmR colonies. All occurred into a TTGAT target and exhibited

an orientation bias (18/23) on the lagging-strand template (Fig-

ure 5B and Table S2) without irradiation. In contrast, 21 inser-

tions isolated from g-irradiated D. radiodurans occurred essen-

tially equally on both strands (Figure 5C and Table S3).

Genomic Analysis Reveals Orientation Biasin Other IS200/IS605 Family MembersIn view of the strong orientation bias observed for IS608 and

ISDra2 insertions, we wondered whether similar patterns might

be found in other bacterial genomes as vestiges of members

of the IS200/IS605 family occur widely (ISfinder: http://www-is.

biotoul.fr). We identified several genomes with a significant

number of copies of various members of this family and deter-

mined the orientation of insertion relative to the origin of

Figure 4. IS608 Insertions into Tus/Ter Stalled Replication Forks

(A) Experimental design. The suicide conjugative IS-carrying donor plasmid,

pBS167b, is shown (top). It cannot propagate in the recipient, and IS insertions

were monitored into the Ter-carrying target plasmid, pBS172-3, as described

in detail in Figure S1 and its legend. Plasmids are described in the Extended

Experimental Procedures.

(B) Insertions with Ter in permissive and nonpermissive orientations. Replica-

tion from ori is from left to right; gray boxes, araC repressor, tus, and bla

genes; *, Ter-proximal TTAC sequences; horizontal red arrowhead, Ternp

(nonpermissive, pBS172); horizontal black arrowhead, Terp (permissive,

pBS173); black vertical arrowheads, IS608 insertions; red vertical arrowheads,

multiple insertions upstream of Ternp or an insertion inside of Terp. The position

of the oligonucleotides used to localize insertions is also shown.

(C) Detailed analysis of insertions in the Ternp region. PCR amplification prod-

ucts were separated on a 5% acrylamide gel. The distribution and orientation

of insertions close to Ter were determined in four independent experiments

(LE, lanes 1–4; RE, lanes 5–8). Distribution of TTAC on the lagging-strand

and leading-strand template are shown in the cartoon (left and right, respec-

tively).

(D) Detailed analysis of insertions in the Terp region. Identical experiment as

shown in (C) with Terp.

See also Figure S1D.


replication (as defined by GC skew). The genomic ISs exhibited

a strong orientation bias, supporting our notion that the lagging-

strand template provides the most attractive ssDNA for IS

targeting.

The results for representative bacterial genomes are shown in

Figure 6. For example, S. enterica CT18 carries 27 full copies of

IS200 (Deng et al., 2003) (Figure 6A). Thirteen of these are on

one replicore and 14 on the other, while all but five occur in

the orientation expected for insertion into the lagging-strand

template. Another example is Yersinia pseudotuberculosis

(IP31758) (Chain et al., 2004) (Figure 6C) with 16 full IS1541

copies (nine in the left and seven in the right replicore), 13 of

these in an orientation consistent with insertion into the

lagging-strand template. Photobacterium profundum SS9 (Vezzi

et al., 2005; Figure 6D) carries 28 ISPpr13 copies distributed

between its two chromosomes, with 26 in the expected orienta-

tion, and in Shewanella woodyi with 13 copies of ISShwo2, all

except one are in the expected orientation (Karpinets et al.,

2010; Figure 6E). Fisher’s exact test provided strong statistical

support (Figure 6 legend).

Interestingly, the distribution of the 42 IS1541 copies in

Y. pestis (Microtus) (Figure 6F) and the 50 copies of it in Y. pestis

(Pestoides F) (Figure 6G) are clearly different and show a complex

pattern with no apparent relationship to the replication direction.

However, these Y. pestis strains revealed aberrant GC skews

resulting from a series of inversions and rearrangements

throughout the chromosome (Song et al., 2004; Garcia et al.,

2007). When these are taken into account, an astoundingly

close correlation emerges between GC skew and IS1541

orientation.

Similar reasoning may explain those insertions that initially

appear in contradiction with a lagging-strand template targeting

preference. For example, of the five IS200 insertions in S. enter-

ica strain CT18 oriented in the opposite direction (Figure 6A), one

is located in a short chromosome region with an inverted GC-

skew compared to the neighboring sequences. This implies

that this region has undergone inversion and that the associated

IS insertion likely had occurred prior to this event. The insertion

pattern observed in S. enterica (Ty2) (Deng et al., 2003;

Figure 6B) is quite similar to that of S. enterica CT18; both carry

an identical number of IS200 copies. The differences in IS200

distribution between these two strains can be entirely explained

by a large interreplicore inversion between S. enterica (Ty2) and

S. enterica CT18. These genomic results strongly support the

idea that insertion into the lagging-strand template of the chro-

mosomes of their host is a general characteristic of IS200/

IS608 family members.

DISCUSSION

The transposition pathway of IS200/IS605 IS family members

involves ssDNA substrates and intermediates. In vitro IS excision

requires both transposon ends to be single stranded, and inser-

tion of the excised single-stranded circular transposon interme-

diate requires access to an ssDNA target (Guynet et al., 2009).

We now demonstrate that excision and insertion of two family

members, IS608 and ISDra2, occur preferentially in the lagging-

strand DNA template in vivo.

Excision is dependent on replication direction through the IS: it

is high when the active IS strand is on the lagging-strand

template but substantially lower when part of the leading-strand

template. We also examined insertions of a plasmid-localized

IS608 into normally replicating E. coli chromosomes and sponta-

neous insertions of ISDra2 into D. radiodurans chromosome I.

They were largely directed to the lagging-strand template, result-

ing in a skew of strand-specific insertion on either side of the

replication origin in these bidirectionally replicating chromo-

somes. For a unidirectionally replicating E. coli plasmid, inser-

tions occurred into only one strand. Importantly, the orientation

effect for ISDra2 insertion was abolished when transposition

was triggered by g-irradiation, accompanied by an increase in

transposition frequency, consistent with the observation that

g-irradiation induces a repair pathway resulting in massive

amounts of ssDNA with no obvious strand bias.

IS608 insertions into the E. coli chromosome were fairly evenly

distributed but a significant number occurred in the highly tran-

scribed rrn genes (which are oriented in the sense of replication),

suggesting that high transcription levels might also provide

accessible ssDNA for IS608 insertion, e.g., by generating R loops

or by affecting replication fork passage. Replication in E. coli

Figure 5. Excision and Insertion of ISDra2-113 in Deinococcus radio-

durans

(A) ISDra2-113 excision frequencies depend on whether the active (top) IS-

Dra2-113 strand is located on the lagging- or leading-strand template. The

middle column indicates whether TnpA was provided and whether the cells

were subjected to g-irradiation. Strain constructions are presented in the

Extended Experimental Procedures.

(B) Spontaneous ISDra2-113 insertions into the D. radiodurans chromosome I.

The origin (ori, red ellipse) and direction of replication of the D. radiodurans

chromosome was assumed to be that proposed by Hendrickson and Law-

rence (2006). All insertions occurred 30 to TTGAT pentanucleotide sequences.

Detailed mapping is presented in Table S2. Fisher’s exact test gave a p value of

0.01108. Multiple insertions are indicated by a number above the larger

arrows.

(C) ISDra2-113 insertions into the D. radiodurans chromosome I after g-irradi-

ation. Detailed mapping is presented in Table S3. Fisher’s exact test gave

a p value of 0.1344, suggesting that the pattern is random.

See also Tables S2 and S3.


has been estimated to be approximately 20-fold faster than is

transcription (800 nt/s versus 20 to 50 nt/s) (Kornberg and Baker,

1992). Replication forks may stall at transfer RNA and other

highly expressed genes, possibly because transcription

complexes collide ‘‘head on’’ with the forks. Replication forks

also stall upon codirectional encounters with RNA polymerase

(Elıas-Arnanz and Salas, 1997; Mirkin et al., 2006), possibly as

a result of a trapped RNA polymerase not readily displaced

from DNA by fork progression.

That ssDNA at the replication fork facilitates IS608 transposi-

tion is reinforced by studies which perturb the fork using temper-

ature sensitive DnaG (primase) or DnaB (helicase) mutants.

DnaG inactivation prevents initiation of Okazaki fragment syn-

thesis, increasing the average length of ssDNA upstream of the

first complete Okazaki fragment on the lagging-strand template.

(Louarn, 1974; Fouser and Bird, 1983). DnaB inactivation results

in accumulation of large amounts of ssDNA mostly likely arising

from degradation of both the nascent DNA and leading-strand

template or from uncoupled leading-strand synthesis (Belle

et al., 2007). We observed a significant stimulation of IS608 exci-

sion after transitory inactivation of both dnaGts and dnaBts

mutants.

If excision occurs at the replication fork and requires ssDNA,

the probability that both ends are within the single-stranded

region of the lagging-strand template should decrease with

increasing IS length, and thus the size of the IS should influence

excision frequency exactly as we observe. Moreover, the effi-

ciency of excision was generally higher in the dnaGts mutant

even at the permissive growth temperature of 30�C and the slope

of the curve was less steep. At the sublethal temperature of

33�C, excision was even higher and the length dependence

even less marked. This is consistent with an increase in ssDNA

length of on the lagging-strand template resulting from a lower

Okazaki fragment initiation frequency in the dnaGts mutant

even at temperatures permissive for growth and suggests that

the slope is a function of the ssDNA length available on the

lagging-strand template.

We also investigated the effect of DnaG overproduction.

Expression of a cloned wild-type dnaG gene not only sup-

pressed the dnaGts phenotype but also resulted in an even

Figure 6. Orientation of IS200/IS605 Family Members in Bacterial Genomes

Overall GC skew (G – C / G + C) is indicated in blue and orange. Black line, calculated GC skew with Artemis (http://www.sanger.ac.uk/Software/Artemis/) with

a 10 kb window; red ellipse, origin of replication; vertical arrow heads (above or below the genome depending on their orientation), individual ISs.

(A) S. enterica (typhi) CT18 (NC_003198); IS200, 709 bp (Fisher’s exact test p value, 0.002054).

(B) S. enterica (typhi) Ty2 (NC_004631); IS200, 709 bp (p value, 0.004799).

(C) Y. pseudotuberculosis IP31758 (NC_009708); IS1541, 708–709 bp (p value, 0.02144).

(D) P. profundum SS9 (NC_006371 and NC_006370); ISPr13, 596 bp (chromosome 1 p value, 0.0005139; chromosome 2 p value, 0.002806).

(E) S. woodyi (NC_010506); ISShwo2, 613 bp (p value, 0.02297).

(F) Y. pestis Microtus 91001 (NC_005810); IS1541, 711 bp (p value, 1.139e�06).

(G) Y. pestis Pestoides F (NC_009381); IS1541D, 711 bp (p value, 2.479e�06).

In (F) and (G), p values were calculated using the potential number of target sequences taking into account the regions of GC skew inversion.


more pronounced length dependence in the wild-type back-

ground, suggesting that, as observed in vitro (Zechner et al.,

1992; Sanders et al., 2010), DnaG concentration controls the

frequency of initiation and Okazaki fragment size. More impor-

tantly, this result would suggest that DnaG is not saturating at

the normal replication fork in vivo.

While the slopes of the curves presumably reflect the length of

ssDNA on the lagging-strand template, the explanation for the

apparent inflection of the curves for the longer transposons is

less clear. It is possible that we are observing the combined

effects of two phenomena: for example the initial probability

that both ends are in a single-stranded form and the probability

that both ends find each other. Further analysis is required to

determine the exact reason behind this behavior.

Although, for simplicity, we describe the lagging-strand

template as single stranded at the fork, it is important to note

that it is not naked but protected by proteins such as single-

strand binding protein (Ssb) (Shereda et al., 2008). This implies

that the transposition machinery can access the DNA through

the protecting protein and raises the question of whether TnpA

can recognize Ssb or other components of the replisome. Exper-

iments to investigate this are in progress.

We also used the natural Tus/Ter replication fork termination

system (Neylon et al., 2005; Kaplan and Bastia, 2009) to examine

whether blocked forks might also attract IS608 insertions. The

Tus-Ter complex forms a transient barrier to the replicative heli-

case, DnaB, when a fork arrives in the nonpermissive direction

(Neylon et al., 2005). When provided with appropriate target tet-

ranucleotides, Tus-dependent IS608 insertions readily occurred

close (26–77 nt) to the Ter site on the lagging-strand template,

consistent with nucleotide resolution mapping of the terminated

nascent DNA in vitro and in vivo showing that the final lagging-

strand primer sites are arrested 50–70 nucleotides upstream of

Ter (Hill and Marians, 1990; Mohanty et al., 1998). While our

data are consistent with the idea that stalled forks favor IS608

insertion, we do not yet know whether the ssDNA substrates

are directly available at blocked forks or are generated during

their processing, e.g., during replication restart or repair of ds

breaks caused by replication arrest (Michel et al., 1997; Bierne

and Michel, 1994).

To determine whether other IS200/IS605 family members

might locate suitable ssDNA substrates for transposition, we

annotated several complete bacterial genomes for ISs and iden-

tified several that harbor multiple copies of these family

members. In the majority, these ISs showed a similar orientation

bias to IS608 and ISDra2 relative to replication direction. Thus,

targeting of the ssDNA available on the lagging-strand template

appears to be a general theme among IS200/IS605 family

members.

Replication direction and therefore identification of the lagging

strand is generally implied from GC skew, the preference for G

over C on the leading strand thought to be the result of differen-

tial repair (Lobry, 1996; Grigoriev, 1998). More strictly speaking,

we observed that IS orientation was correlated with the GC skew

of the region into which they were inserted rather than with repli-

cation direction per se, suggesting that the insertions predated

the genome rearrangements whose scars are revealed by

changes in GC skew. This has two important implications: either

that transposition is infrequent or that transposition events

become genetically fixed in the population.

IS200/IS605 family members are not alone in showing asym-

metric strand preference in insertion. Other transposable

elements such as Tn7 and IS903 also appear to do so (Peters

and Craig, 2001; Hu and Derbyshire, 1998). Tn7 is targeted to

replication forks during conjugative plasmid transfer and inserts

in a specific orientation. The transposon-encoded TnsE protein

is instrumental in targeting the transpososome to the junction

between single- and double-stranded DNA by interaction with

the b clamp component of the replication apparatus (Parks

et al., 2009; Chandler, 2009). Insertion of Tn7 into a replication

fork likely occurs within the dsDNA covered by an Okazaki frag-

ment. IS903 insertion bias in the conjugative F plasmid might

also reflect targeting to the conjugative replication fork. It is

worthwhile noting that IS10 and IS50 have also exploited host

replication: both are activated by transient formation of hemime-

thylated DNA after fork passage (Roberts et al., 1985; Yin et al.,

1988; Dodson and Berg, 1989).

However, there are major mechanistic differences between

Tn7, IS903, and members of the IS200/IS605 family, suggesting

that different pathways are at play. Perhaps most importantly,

Tn7 transposes using a dsDNA intermediate in contrast to the

ssDNA species used by IS608 and ISDra2. For Tn7 and many

other transposons with dsDNA intermediates, strand transfer

into a dsDNA target occurs using the 30 OH groups on each com-

plementary strand at each end. Insertion of the first strand into

a single-strand target would leave a fatal break. Thus, whereas

an overarching theme in DNA transposition may be the exploita-

tion of replication forks, different elements do so in different ways.

The unique excision and insertion properties of the IS200/

IS605 family may make them useful tools for probing ssDNA

structures in vivo. The relationship between excision and IS

length might be used to determine the effect of various factors

on the state of the replication fork in vivo. For example, treat-

ments leading to reduced fork velocity, Okazaki fragment

synthesis, uncoupling of lagging from leading-strand synthesis,

or simply forks blocked by different factors could all be probed

using an excision assay as an experimental readout. We are

aware that the topology of the replication fork of small, multicopy

plasmids may differ in some respects from that of the chromo-

some, and this system may also be used to explore these

possible differences. We are at present testing these possibilities

experimentally.


Bacterial Strains and Media

Bacterial strains are described in the Extended Experimental Procedures; see

also Figure S2 and Table S4. E. coli cultures were grown in Luria broth (LB) sup-

plemented where necessary with chloramphenicol (Cm, 10 or 30 mg/ml), kana-

mycin (Km, 20 mg/ml), ampicillin (Ap, 100 mg/ml), gentamycin (Gm, 5 mg/ml),

spectinomycin (Sp, 30 mg/ml) and streptomycin (Sm, 20 mg/ml), tetracycline

(Tc, 15 mg/ml), or 2,6-Diaminoheplanediole acid (DAP, 0.006%). D. radiodurans

media and growth conditions were as described (Pasternak et al., 2010; Bona-

cossa de Almeida et al., 2002).

Plasmids

Details of plasmid constructions are available upon request; see Figure S3,

Table S4, and the Extended Experimental Procedures for schematics of


plasmids used and additional information. The dnaG-carrying plasmid pBS179

was constructed by insertion of a DNA fragment carrying the wild-type dnaG

allele with its natural ribosome binding site isolated by PCR with flanking

oligonucleotides, dnaGN, and dnaGC into the SphI site of transposase

supplying plasmid downstream of the tnpA gene, placing both genes under

control of plac. The clone was verified by sequencing and complements the

dnaGts allele.

Excision Assay with pAM1 Derivatives

Effect of dnaG308 and dnaB8 on Excision

E. coli MG4100 wild-type and dnaG308 and dnaB8 mutant strains were grown

overnight at 30�C, the permissive temperature, in LB+KmTc, and cultures were

diluted at 30�C and grown with 0.5 mM IPTG for 4 hr or 45 min. IPTG was

removed by centrifugation and cells were resuspended in prewarmed medium

at 42�C for 30 min and incubated at 30�C for 3 hr. As a control, the 42�C step

was omitted. Cells were harvested and dilutions plated on LA+KmTc and

LA+KmTcAp.

Effect of Transposon Length

E. coli DH5a harboring pBS135 and pAM1-derivative plasmids was grown

overnight at 37�C in LB+KmTc. Cultures were diluted in fresh medium at

37�C with KmTc and 0.5 mM IPTG to induce TnpA expression. Cells were

harvested after 5 hr and dilutions plated on LA+KmTc and LA+KmTcAp.

Effect of Transposon Length in the dnaG308 Mutant Strain

Overnight cultures of MC4100 and MC4100 dnaGts were grown at 30�C and

diluted in fresh medium at 30�C or 33�C containing KmTc and 0.5 mM IPTG.

Cells were harvested after 5 hr and dilutions plated on LA+KmTc and

LA+KmTcAp at 30�C.

Tus/Ter Experiments

Overnight donor and recipient strains were grown in LB+DAP+SpSmCm

and LB+GmKmAp, respectively. Glucose (0.5%) was added to fully repress

the para used to drive Tus expression. Cultures were diluted in fresh LB medium

without antibiotics (with DAP for donor cells). At an OD600 of 0.5, donor

cells were incubated without agitation, the recipient was diluted to an OD600

of 0.15, and TnpA was induced with 0.5 mM IPTG. After 60 min, Tus

was induced for 30 min by addition of 0.08% arabinose. Donor and recipient

strains were mixed, incubated for 2 hr and plated on Cm-containing LA plates

supplemented with 0.5% glucose. Bulk plasmid DNA was isolated and the

distribution of insertions in the population was analyzed by PCR with

six sets of primers: B248, B249, tus as forward primers, and LE and RE

as reverse primers (Extended Experimental Procedures) to determine the

insertion orientation. PCR amplification used Phusion DNA Polymerase

(Finnzyme) in GC buffer as follows: 30 s 98�C, 35 cycles of 10 s 98�C, 30 s

64�C, and �1 min 72�C. Fifty nanograms of bulk plasmid DNA were used for

each reaction.

Sequencing IS608 Insertions in the E.coli Chromosome

IS608 insertions into the E. coli chromosome and coresident plasmid pBS135

were sequenced by arbitrary PCR as described (Guynet et al., 2009).

Measurement of In Vivo Spontaneous and g-Induced Excision

Frequencies of ISDra2-113

Excision frequencies were determined with individual CmR TcS colonies puri-

fied from strains GY14302 or GY14303 (Extended Experimental Procedures)

grown with or without TnpA expression in trans as described (Pasternak

et al., 2010). The mean and standard deviations were calculated from six inde-

pendent experiments. ISDra2 insertion sites were mapped by arbitrary-primed

(AP)-PCR with DyNazyme EXT DNA polymerase (Finnzymes) as described

(Extended Experimental Procedures).

Statistical Analysis of Insertion Orientation

This was performed with Fisher’s exact test with the ‘‘R’’ statistical package

(http://www.r-project.org/) taking into account the number of potential target

sites on the leading and lagging strands. For the Yersinia species, the number

of potential target sites and of insertion sites was calculated taking into

account GC skew inversion.


Supplemental Information includes Extended Experimental Procedures, three

figures, and four tables and can be found with this article online at doi:10.1016/

j.cell.2010.06.034.

ACKNOWLEDGMENTS

We would like to thank members of the ‘‘Mobile Genetic Elements’’ group,

A. Bailone, P. Polard, and D. Lane for discussions, G. Coste and B. Marty

for expert technical assistance, L. Lavatine for guiding us through the

mysteries of statistics, A. Varani for identifying a representative oligonucleotide

sequence used in the PCR mapping for the D. radiodurans genome, and the

Institut Curie for the use of the 137Cs irradiation system. This work was sup-

ported by: intramural funding from the Centre National de Recherche Scienti-

fique (France), in its later stages by ANR grant Mobigen (M.C. and S.S.), Euro-

pean contract LSHM-CT-2005-019023 (M.C.), and the Commissariat a

l’Energie Atomique and Electricite de France (France; S.S.). At the National

Institutes of Health, this work was supported by the Intramural Program of

the National Institute of Diabetes and Digestive and Kidney Diseases.





REFERENCES

Barabas, O., Ronning, D.R., Guynet, C., Hickman, A.B., Ton-Hoang, B.,

Chandler, M., and Dyda, F. (2008). Mechanism of IS200/IS605 family DNA

transposases: activation and transposon-directed target site selection. Cell

132, 208–220.

Belle, J.J., Casey, A., Courcelle, C.T., and Courcelle, J. (2007). Inactivation of

the DnaB helicase leads to the collapse and degradation of the replication fork:

a comparison to UV-induced arrest. J. Bacteriol. 189, 5452–5462.

Biemont, C., and Vieira, C. (2006). Genetics: junk DNA as an evolutionary force.

Nature 443, 521–524.

Bierne, H., and Michel, B. (1994). When replication forks stop. Mol. Microbiol.

13, 17–23.

Bierne, H., Ehrlich, S.D., and Michel, B. (1994). Flanking sequences affect

replication arrest at the Escherichia coli terminator TerB in vivo. J. Bacteriol.

176, 4165–4167.

Bonacossa de Almeida, C., Coste, G., Sommer, S., and Bailone, A. (2002).

Quantification of RecA protein in Deinococcus radiodurans reveals involve-

ment of RecA, but not LexA, in its regulation. Mol. Genet. Genomics 268,

28–41.

Carl, P.L. (1970). Escherichia coli mutants with temperature-sensitive

synthesis of DNA. Mol. Gen. Genet. 109, 107–122.

Chain, P.S., Carniel, E., Larimer, F.W., Lamerdin, J., Stoutland, P.O., Regala,

W.M., Georgescu, A.M., Vergez, L.M., Land, M.L., Motin, V.L., et al. (2004).

Insights into the evolution of Yersinia pestis through whole-genome compar-

ison with Yersinia pseudotuberculosis. Proc. Natl. Acad. Sci. USA 101,

13826–13831.

Chandler, M. (2009). Clamping down on transposon targeting. Cell 138,

621–623.

Curcio, M.J., and Derbyshire, K.M. (2003). The outs and ins of transposition:

from Mu to Kangaroo. Nat. Rev. Mol. Cell Biol. 4, 865–877.

Demarre, G., Guerout, A.M., Matsumoto-Mashimo, C., Rowe-Magnus, D.A.,

Marliere, P., and Mazel, D. (2005). A new family of mobilizable suicide plasmids

based on broad host range R388 plasmid (IncW) and RP4 plasmid (IncPalpha)

conjugative machineries and their cognate Escherichia coli host strains. Res.

Microbiol. 156, 245–255.

Deng, W., Liou, S.R., Plunkett, G., 3rd, Mayhew, G.F., Rose, D.J., Burland, V.,

Kodoyianni, V., Schwartz, D.C., and Blattner, F.R. (2003). Comparative


genomics of Salmonella enterica serovar Typhi strains Ty2 and CT18. J. Bac-

teriol. 185, 2330–2337.

Dodson, K.W., and Berg, D.E. (1989). Factors affecting transposition activity of

IS50 and Tn5 ends. Gene 76, 207–213.

Elıas-Arnanz, M., and Salas, M. (1997). Bacteriophage phi29 DNA replication

arrest caused by codirectional collisions with the transcription machinery.

EMBO J. 16, 5775–5783.

Fouser, L., and Bird, R.E. (1983). Accumulation of ColE1 early replicative inter-

mediates catalyzed by extracts of Escherichia coli dnaG mutant strains.

J. Bacteriol. 154, 1174–1183.

Galas, D.J., and Chandler, M. (1982). Structure and stability of Tn9-mediated

cointegrates. Evidence for two pathways of transposition. J. Mol. Biol. 154,

245–272.

Garcia, E., Worsham, P., Bearden, S., Malfatti, S., Lang, D., Larimer, F.,

Lindler, L., and Chain, P. (2007). Pestoides F, an atypical Yersinia pestis strain

from the former Soviet Union. Adv. Exp. Med. Biol. 603, 17–22.

Grigoriev, A. (1998). Analyzing genomes with cumulative skew diagrams.

Nucleic Acids Res. 26, 2286–2290.

Guynet, C., Hickman, A.B., Barabas, O., Dyda, F., Chandler, M., and Ton-

Hoang, B. (2008). In vitro reconstitution of a single-stranded transposition

mechanism of IS608. Mol. Cell 29, 302–312.

Guynet, C., Achard, A., Hoang, B.T., Barabas, O., Hickman, A.B., Dyda, F., and

Chandler, M. (2009). Resetting the site: redirecting integration of an insertion

sequence in a predictable way. Mol. Cell 34, 612–619.

Hendrickson, H., and Lawrence, J.G. (2006). Selection for chromosome archi-

tecture in bacteria. J. Mol. Evol. 62, 615–629.

Hill, T.M., and Marians, K.J. (1990). Escherichia coli Tus protein acts to arrest

the progression of DNA replication forks in vitro. Proc. Natl. Acad. Sci. USA 87,

2481–2485.

Hu, W.Y., and Derbyshire, K.M. (1998). Target choice and orientation prefer-

ence of the insertion sequence IS903. J. Bacteriol. 180, 3039–3048.

Islam, S.M., Hua, Y., Ohba, H., Satoh, K., Kikuchi, M., Yanagisawa, T., and

Narumi, I. (2003). Characterization and distribution of IS8301 in the radioresist-

ant bacterium Deinococcus radiodurans. Genes Genet. Syst. 78, 319–327.

Johnson, A., and O’Donnell, M. (2005). Cellular DNA replicases: components

and dynamics at the replication fork. Annu. Rev. Biochem. 74, 283–315.

Kaplan, D.L., and Bastia, D. (2009). Mechanisms of polar arrest of a replication

fork. Mol. Microbiol. 72, 279–285.

Karpinets, T.V., Obraztsova, A.Y., Wang, Y., Schmoyer, D.D., Kora, G.H., Park,

B.H., Serres, M.H., Romine, M.F., Land, M.L., Kothe, T.B., et al. (2010).

Conserved synteny at the protein family level reveals genes underlying Shewa-

nella species’ cold tolerance and predicts their novel phenotypes. Funct. In-

tegr. Genomics 10, 97–110.

Kersulyte, D., Velapatino, B., Dailide, G., Mukhopadhyay, A.K., Ito, Y.,

Cahuayme, L., Parkinson, A.J., Gilman, R.H., and Berg, D.E. (2002). Transpos-

able element ISHp608 of Helicobacter pylori: nonrandom geographic distribu-

tion, functional organization, and insertion specificity. J. Bacteriol. 184,

992–1002.

Kornberg, A., and Baker, T.A. (1992). DNA Replication (New York: W.H.

Freeman).

Lobry, J.R. (1996). Asymmetric substitution patterns in the two DNA strands of

bacteria. Mol. Biol. Evol. 13, 660–665.

Louarn, J.M. (1974). Size distribution and molecular polarity of nascent DNA in

a temperature-sensitive dna G mutant of Escherichia coli. Mol. Gen. Genet.

133, 193–200.

Mennecier, S., Servant, P., Coste, G., Bailone, A., and Sommer, S. (2006).

Mutagenesis via IS transposition in Deinococcus radiodurans. Mol. Microbiol.

59, 317–325.

Michel, B., Ehrlich, S.D., and Uzest, M. (1997). DNA double-strand breaks

caused by replication arrest. EMBO J. 16, 430–438.

Mirkin, E.V., Castro Roa, D., Nudler, E., and Mirkin, S.M. (2006). Transcription

regulatory elements are punctuation marks for DNA replication. Proc. Natl.

Acad. Sci. USA 103, 7276–7281.

Mohanty, B.K., Sahoo, T., and Bastia, D. (1998). Mechanistic studies on the

impact of transcription on sequence-specific termination of DNA replication

and vice versa. J. Biol. Chem. 273, 3051–3059.

Neylon, C., Kralicek, A.V., Hill, T.M., and Dixon, N.E. (2005). Replication termi-

nation in Escherichia coli: structure and antihelicase activity of the Tus-Ter

complex. Microbiol. Mol. Biol. Rev. 69, 501–526.

Parks, A.R., Li, Z., Shi, Q., Owens, R.M., Jin, M.M., and Peters, J.E. (2009).

Transposition into replicating DNA occurs through interaction with the proces-

sivity factor. Cell 138, 685–695.

Pasternak, C., Ton-Hoang, B., Coste, G., Bailone, A., Chandler, M., and Som-

mer, S. (2010). Irradiation-induced Deinococcus radiodurans genome frag-

mentation triggers transposition of a single resident insertion sequence.

PLoS Genet. 6, e1000799.

Peters, J.E., and Craig, N.L. (2001). Tn7 recognizes transposition target struc-

tures associated with DNA replication using the DNA-binding protein TnsE.

Genes Dev. 15, 737–747.

Roberts, D., Hoopes, B.C., McClure, W.R., and Kleckner, N. (1985). IS10 trans-

position is regulated by DNA adenine methylation. Cell 43, 117–130.

Ronning, D.R., Guynet, C., Ton-Hoang, B., Perez, Z.N., Ghirlando, R., Chan-

dler, M., and Dyda, F. (2005). Active site sharing and subterminal hairpin recog-

nition in a new class of DNA transposases. Mol. Cell 20, 143–154.

Sanders, G.M., Dallmann, H.G., and McHenry, C.S. (2010). Reconstitution of

the B. subtilis replisome with 13 proteins including two distinct replicases.

Mol. Cell 37, 273–281.

Shereda, R.D., Kozlov, A.G., Lohman, T.M., Cox, M.M., and Keck, J.L. (2008).

SSB as an organizer/mobilizer of genome maintenance complexes. Crit. Rev.

Biochem. Mol. Biol. 43, 289–318.

Simons, R.W., Houman, F., and Kleckner, N. (1987). Improved single and

multicopy lac-based cloning vectors for protein and operon fusions. Gene

53, 85–96.

Song, Y., Tong, Z., Wang, J., Wang, L., Guo, Z., Han, Y., Zhang, J., Pei, D.,

Zhou, D., Qin, H., et al. (2004). Complete genome sequence of Yersinia pestis

strain 91001, an isolate avirulent to humans. DNA Res. 11, 179–197.

Ton-Hoang, B., Guynet, C., Ronning, D.R., Cointin-Marty, B., Dyda, F., and

Chandler, M. (2005). Transposition of ISHp608, member of an unusual family

of bacterial insertion sequences. EMBO J. 24, 3325–3338.

Turlan, C., and Chandler, M. (2000). Playing second fiddle: second-strand pro-

cessing and liberation of transposable elements from donor DNA. Trends

Microbiol. 8, 268–274.

Vezzi, A., Campanaro, S., D’Angelo, M., Simonato, F., Vitulo, N., Lauro, F.M.,

Cestaro, A., Malacrida, G., Simionati, B., Cannata, N., et al. (2005). Life at

depth: Photobacterium profundum genome sequence and expression anal-

ysis. Science 307, 1459–1461.

Wechsler, J.A., and Gross, J.D. (1971). Escherichia coli mutants temperature-

sensitive for DNA synthesis. Mol. Gen. Genet. 113, 273–284.

Yin, J.C., Krebs, M.P., and Reznikoff, W.S. (1988). Effect of dam methylation

on Tn5 transposition. J. Mol. Biol. 199, 35–45.

Zahradka, K., Slade, D., Bailone, A., Sommer, S., Averbeck, D., Petranovic, M.,

Lindner, A.B., and Radman, M. (2006). Reassembly of shattered chromo-

somes in Deinococcus radiodurans. Nature 443, 569–573.

Zechner, E.L., Wu, C.A., and Marians, K.J. (1992). Coordinated leading- and

lagging-strand synthesis at the Escherichia coli DNA replication fork. II.

Frequency of primer synthesis and efficiency of primer utilization control

Okazaki fragment size. J. Biol. Chem. 267, 4045–4053.


A Large Intergenic Noncoding RNAInduced by p53 Mediates Global GeneRepression in the p53 ResponseMaite Huarte,1,2,* Mitchell Guttman,1,3 David Feldser,3,4 Manuel Garber,1 Magdalena J. Koziol,1,2

Daniela Kenzelmann-Broz,5,6 Ahmad M. Khalil,1,2 Or Zuk,1 Ido Amit,1 Michal Rabani,1 Laura D. Attardi,5,6 Aviv Regev,1,3

Eric S. Lander,1,3,7 Tyler Jacks,3,4 and John L. Rinn1,2,*1The Broad Institute of MIT and Harvard, Cambridge, MA 02142, USA2Department of Pathology, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, MA 02215, USA3Department of Biology, Massachusetts Institute of Technology, Cambridge, MA 02139, USA4The Koch Institute for Integrative Cancer Research, Cambridge, MA 02139, USA5Department of Radiation Oncology6Department of and Genetics

Stanford University School of Medicine, Stanford, CA 94305, USA7Department of Systems Biology, Harvard Medical School, Boston, MA 02114, USA

*Correspondence: [email protected] (M.H.), [email protected] (J.L.R.)DOI 10.1016/j.cell.2010.06.040

SUMMARY

Recently, more than 1000 large intergenic noncodingRNAs (lincRNAs) have been reported. These RNAsare evolutionarily conserved in mammalian genomesand thus presumably function in diverse biologi-cal processes. Here, we report the identification oflincRNAs that are regulated by p53. One of theselincRNAs (lincRNA-p21) serves as a repressor inp53-dependent transcriptional responses. Inhibitionof lincRNA-p21 affects the expression of hundredsof gene targets enriched for genes normally re-pressed by p53. The observed transcriptional repres-sion by lincRNA-p21 is mediated through the phys-ical association with hnRNP-K. This interaction isrequired for proper genomic localization of hnRNP-K at repressed genes and regulation of p53 mediatesapoptosis. We propose a model whereby transcrip-tion factors activate lincRNAs that serve as keyrepressors by physically associating with repressivecomplexes and modulate their localization to sets ofpreviously active genes.

INTRODUCTION

It has become increasingly clear that mammalian genomes

encode numerous large noncoding RNAs (Mercer et al., 2009;

Ponting et al., 2009; Mattick, 2009; Ponjavic et al., 2007). It

has been recently reported the identification of more than 1000

large intergenic noncoding RNAs (lincRNAs) in the mouse

genome (Carninci, 2008; Guttman et al., 2009). The approach

to identify lincRNAs was by searching for a chromatin signature

of actively transcribed genes, consisting of a histone 3-lysine

4 trimethylated (H3K4me3) promoter region and histone 3-lysine

36 trimethylation (H3K36me3) corresponding to the elongated

transcript (Guttman et al., 2009). These lincRNAs show clear

evolutionary conservation, implying that they are functional

(Guttman et al., 2009; Ponjavic et al., 2007).

In an attempt to understand the potential biological roles of

lincRNAs, a method to infer putative function based on correla-

tion in expression between lincRNAs and protein-coding genes

was developed. These studies led to preliminary hypotheses

about the involvement of lincRNAs in diverse biological pro-

cesses, from stem cell pluripotency to cell-cycle regulation

(Guttman et al., 2009). In particular, we observed a group of

lincRNAs that are strongly associated with the p53 transcrip-

tional pathway. p53 is an important tumor suppressor gene

involved in maintaining genomic integrity (Vazquez et al., 2008).

In response to DNA damage, p53 becomes stabilized and trig-

gers a transcriptional response that causes either cell arrest or

apoptosis (Riley et al., 2008).

The p53 transcriptional response involves both activation and

repression of numerous genes. While p53 is known to transcrip-

tionally activate numerous genes, the mechanisms by which p53

leads to gene repression have remained elusive. We recently

reported evidence that many lincRNAs are physically associated

with repressive chromatin modifying complexes and suggested

that they may serve as repressors in transcriptional regulatory

networks (Khalil et al., 2009). We therefore hypothesized that

p53 may repress genes in part by directly activating lincRNAs,

which in turn regulate downstream transcriptional repression.

Here, we show that lincRNAs play a key regulatory role in the

p53 transcriptional response. By exploiting multiple independent

cell-based systems, we identify lincRNAs that are transcriptional

targets of p53. Moreover, we find that one of these p53-activated

lincRNAs—termed lincRNA-p21—serves as a transcriptional

repressor in the p53 pathway and plays a role in triggering apo-

ptosis. We further demonstrate that lincRNA-p21 binds to


hnRNP-K. This interaction is required for proper localization of

hnRNP-K and transcriptional repression of p53-regulated genes.

Together, these results reveal insights into the p53 transcrip-

tional response and lead us to propose that lincRNAs may serve

as key regulatory hubs in transcriptional pathways.

RESULTS

Numerous LincRNAs Are Activatedin a p53-Dependent MannerAs a first attempt to dissect the functional mechanisms of

lincRNAs, we focused on a strong association in the expression

patterns of certain lincRNAs and genes in the p53 pathway

(Guttman et al., 2009). In order to determine whether these

lincRNAs are regulated by p53, we employed two independent

experimental systems that allow us to monitor gene expression

changes atdifferent times after p53 induction (Venturaetal., 2007).

The first system uses mouse embryonic fibroblasts (MEFs)

derived from mice where the endogenous p53 locus is inacti-

vated by insertion of a transcriptional termination site flanked

by loxP sites (LSL) in the first intron. This endogenous p53 locus

(p53 LSL/LSL) is restorable by removal of the stop element by Cre

recombination (Ventura et al., 2007). The p53 LSL/LSL MEFs were

treated with AdenoCre virus expressing the Cre recombinase to

reconstitute the normal p53 allele or AdenoGFP control virus to

maintain the inactive p53 LSL/LSL allele. Then we compared the

transcriptional response between the p53-reconstituted and

p53 LSL/LSL MEFs after 0, 3, 6, and 9 hr of DNA damage treatment

with doxorubicin (we will refer to this system as ‘‘MEFs’’)

(Figure 1A). The second system uses a lung tumor cell line

DNA damage

p53 restoration

Doxorubicin

p53

MEFs

mRNAs

LincRNAs

microarrayprofiling

MEF38

MEF&

KRAS11

KRAS32

fold inductionsequence strand

13.5 +/-1.6 5.9E-06GAACATGCCTGGCACTTGCAAA -

24.6 +/-2.3 1.3E-05GGCCTTGCCCGGGCTTGCCT -+

H3K4me3

lincRNA-Mkln1

lincRNA-p21 chr17:29214831-29220756

Name induction P value

A

B

D

C

E

F

G

KRAS p53LSL/LSL

Lung Tumor

H

sA

NRcnil 83

sA

NRcnil 23

MEF645

MEF &

KRAS289

KRAS995

0 3 6 9

0 8 16 24 40 48time (hours)

time (hours)99

5 m

RN

As

645

mR

NA

s

0h 3h 6h 9h 0h 3h 6h 9h0h 8h 16h 40h48h24h 0h 8h 16h 40h48h24h

-8 +8 -8 +8

0

5

10

15

20

25

30

lincR

NA-p2

1

ΔlincR

NA-p2

1

lincR

NA-Mkl

n1

ΔlincR

NA-Mkl

n1

(Nor

mal

ized

Fire

fly L

ucife

rase

Act

ivity

)

Cdkn1

a

cont

rol

p53p53

+/+

-/-

0

5

10

15

20

25

30

35

117

)GgI/

PIhC 35p(

147

lincR

NA-p2

1

lincR

NA-Mkl

n1

Cdkn1

a

cont

rol-1

cont

rol-2

p53p53

+/+

-/-

cdkn1a CTGGGCATGTCTGGGCAGCGATCchr17:28818329-28822000 16.8 +/-1.2 3.7E-05

Rel

ativ

e R

epor

ter

Indu

ctio

n

tnemhcirn

E dloF evitale

R

LSL/LSL

Cre+/-

Cre+

chr6:31061148-31061169

Figure 1. Several LincRNAs Are p53 Tran-

scriptional Targets

(A) Experiment layout to monitor p53-dependent

transcription. p53-restored (+Cre) or not restored

(�Cre) p53LSL/LSL MEFs were treated with

500 nM dox for 0, 3, 6, and 9 hr (upper left).

KRAS (p53LSL/LSL) tumor cells were treated with

hydroxytamoxifen for p53 restoration for 0, 8, 16,

24, 40, or 48 hr (lower left). RNA was subjected

to microarray analysis of mRNAs and lincRNAs.

(B and C) Venn diagrams showing the number of

shared and distinct mRNAs (B) or lincRNAs (C)

induced in a p53-dependent manner in the MEF

or KRAS systems.

(D and E) mRNAs (D) and lincRNAs (E) activated by

p53 induction (FDR < 0.05) in MEF or KRAS

system. Colors represent transcripts above (red)

or below (blue) the global median scaled to

8-fold activation or repression, respectively.

(F) Promoter region, conserved p53 binding

motif, promoter orientation, and p53-dependent

fold induction in reporter assays of lincRNA

promoters induced in a p53-dependent manner

(values are average of at least three biological

replicates [±STD]; p values are determined by

t test).

(G) p53-dependent induction of lincRNA pro-

moters requires the consensus p53 binding ele-

ments. Relative firefly luciferase expression

driven by promoters with p53 consensus motif

(lincRNA-p21, lincRNA-Mkln1) or with deleted

motif (DlincRNA-p21 and DlincRNA-Mkln1) in

p53-restored p53LSL/LSL (p53+/+) or p53LSL/LSL

(p53�/�) cells. Values are relative to p53�/� and

normalized by renilla levels (average of three repli-

cates ±STD).

(H) p53 specifically binds to p53 motifs in lincRNA

promoters. p53 ChIP enrichment in p53+/+ and

p53�/�MEFs on regions with p53 motifs (lincRNA-

p21, lincRNA-Mkln1, Cdkn1a) or two irrelevant

regions (controls). Enrichment values are relative

to IgG and average of 3 replicates (±STD).

See also Figure S1 and Table S1.


derived from mice expressing an oncogenic K-Ras mutation

(K-RasG12D) and a restorable p53 knockout allele (p53 LSL/LSL),

similar to that described above (D.F. and T.J., unpublished data).

We compared the transcriptional response at different times

(0, 8, 16, 24, 40, and 48 hr) after restoration of p53 expression

by Cre recombination (Experimental Procedures) (we will refer

to this system as ‘‘KRAS’’) (Figure 1A).

To assess the transcriptional responses in each of these

systems, we isolated total RNA at each time point before

and after p53 restoration and performed DNA microarray anal-

ysis to monitor protein-coding gene expression levels. In the

MEF system and KRAS systems, we identified a total of 1067

(645 activated, 422 repressed) and 1955 (995 activated, 960

repressed) genes, respectively, that were regulated in a p53-

dependent manner (false discovery rate [FDR] < 0.05) (Figures

1B and 1D and Table S1, parts A and B, available online). The

sets of p53-induced genes identified in the two systems showed

significant (p < 10�8) overlap, including such canonical p53

targets as Cdnk1a, Mdm2, Perp, and Fas (Table S1, parts A

and B). There are also a number of p53-induced genes unique

to each system, likely reflecting specific properties of each

cell-type (Levine et al., 2006) (Figure S1C and Table S1, parts

A and B).

Functional analysis of the classes of genes that are enriched

among the genes regulated by p53 in both the MEF and KRAS

systems showed strong enrichment for known p53-regulated

processes, such as the cell cycle and apoptosis (Figures S1A

and S1B). Moreover, gene set enrichment analysis (GSEA) (Sub-

ramanian et al., 2005) of previously published microarray anal-

yses revealed a significant overlap with the p53 regulated genes

identified here (Table S1, parts H and I). Together these results

demonstrate that these two systems are largely reflective of

canonical p53 transcriptional responses.

We next examined lincRNAs regulated by p53 in these two

systems across the same time course, by analogously using

a custom tiling microarray representing 400 lincRNAs and

analyzing the data with previously described statistical methods

(Guttman et al., 2009) (Experimental Procedures). We found 38

and 32 lincRNAs induced by p53 in the MEF and KRAS systems,

respectively (Figures 1C and 1E and Table S1, parts C–G). Inter-

estingly, 11 lincRNAs were induced by p53 in both model

systems (Figure 1C and Table S1, part C), many more than

expected by chance (p < 10�6). These results confirm that, in

a manner similar to canonical p53 protein coding gene targets,

numerous lincRNAs are temporally regulated by p53.

LincRNAs Are Direct Transcriptional Targets of p53We sought to identify lincRNAs that might be canonical p53

target genes. As a first approach, we examined the promoters

of p53-induced lincRNA for enrichment of evolutionarily con-

served p53-binding motifs (Garber et al., 2009) (Extended

Experimental Procedures). The promoters of the p53-induced

lincRNAs were highly enriched for conserved p53 motifs relative

to the promoters of all lincRNAs (p < 0.01). We selected two

lincRNAs whose promoter regions contain highly conserved

canonical p53-binding motifs (el-Deiry et al., 1992; Funk et al.,

1992); we termed these lincRNA-p21 and lincRNA-Mkln1 (with

the names referring to the neighboring gene). We next performed

transcriptional reporter assays for these lincRNAs. Specifically,

we cloned their promoters (as defined by the H3K4me3 peaks

[Guttman et al., 2009]) into a luciferase reporter vector (Experi-

mental Procedures) and transfected the constructs along with

a vector to normalize transfection efficiency. Both promoters

showed significant induction of firefly luciferase in p53-wild-

type but not in p53 null cells (p < 0.01) (Figures 1F and 1G).

To determine whether the canonical p53-binding motif is

required for the observed transactivation, we repeated these

experiments in the absence of the p53-binding motif. Mutant

promoters resulted in the abolition of the observed transactiva-

tion for both lincRNA-p21 and lincRNA-Mkln1 in p53+/+ cells

(Figure 1F). Finally, we performed chromatin immunoprecipita-

tion (ChIP) experiments to determine whether p53 directly binds

to the sites containing the consensus motifs in vivo. Indeed, p53

is bound to the site containing the consensus motif in the

promoters of both lincRNA-p21 and lincRNA-Mkln1 in p53+/+

but not p53�/� MEFs treated with doxorubicin, and it is not

enriched at negative control sites of irrelevant regions (Figure 1H

and Extended Experimental Procedures). Together, these

results demonstrate that lincRNA-p21 and lincRNA-Mkln1 are

bona fide p53 transcriptional targets.

LincRNA-p21 Is Induced by p53 in Different Cell SystemsWe were intrigued by the p53 transcriptional target lincRNA-p21,

which resides�15 Kb upstream of the gene encoding the critical

cell-cycle regulator Cdkn1a (also known as p21), a canonical

transcriptional target of p53 (Riley et al., 2008) (Figures 2A–2C,

Figure S2A, and Table S1, parts C–E). Given the proximity of

lincRNA-p21 to the Cdkn1a gene, we sought to ensure that the

lincRNA transcript is distinct from that of the Cdkn1a gene. To

this end, we cloned the full-length transcriptional unit of lincRNA-

p21 using the 50 and 30 RACE method (Schaefer, 1995); the

transcript contains two exons comprising 3.1 Kb (Figure 2B).

In support of lincRNA-p21 being an independent transcript,

lincRNA-p21 is transcribed in the opposite orientation from the

Cdkn1a gene. Furthermore, the analysis of chromatin structure

in mouse embryonic stem cells (mESCs) (Mikkelsen et al., 2007)

indicates that these are distinct genes with distinct promoters

(Figure S2A).

We next examined the transcriptional regulation of lincRNA-

p21 in two additional cancer-derived cell lines. Specifically, we

irradiated p53LSL/LSL mice to induce lymphomas and sarcomas

and then restored p53 expression in tumor-derived cells (Exper-

imental Procedures) (Ventura et al., 2007). In cells derived in both

tumor types, lincRNA-p21 was strongly induced after p53 resto-

ration. Moreover, the induction of lincRNA-p21 followed similar

kinetics as those of p53 and Cdnk1a, consistent with lincRNA-

p21 being a primary transcriptional target of p53 (Figure 2D

and Figure S2B).

We further investigated the orthologous lincRNA-p21 locus in

the human genome. We first mapped the promoter (H3K4me3

domain) of lincRNA-p21 to human genome. Interestingly, this

region corresponds to one of four intergenic p53-binding sites

identified from a study by Wei et al. (2006) (Figure 2B) performing

p53 ChIP followed by sequencing. Next, we mapped the

lincRNA-p21 exonic structures to the human genome to deter-

mine whether this region is expressed and induced by DNA


damage in human fibroblasts. Indeed, qRT-PCR showed that the

orthologous 50 exon region (adjacent to observed p53 ChIP

binding site by Wei et al.) of human lincRNA-p21 is expressed

and strongly induced in human fibroblasts upon DNA damage

(Figure 2C and Extended Experimental Procedures).

Collectively, these results provide evidence that both the

human and mouse lincRNA-p21 promoters are bound by p53

resulting in transcriptional activation in response to DNA

damage. Moreover, lincRNA-p21 is induced by p53 in diverse

biological contexts, including multiple different tumor types

(Figure 2D and Figure S2B), suggesting that lincRNA-p21 plays

a role in the p53 pathway.

LincRNA-p21 as a Repressor in the p53 PathwayWe next investigated the consequence of the loss of lincRNA-

p21 function in the context of the p53 response. We reasoned

that, if lincRNA-p21 plays a role in carrying out the p53 transcrip-

tional response, then inhibition of lincRNA-p21 would show

effects that overlap with inhibition of p53 itself. To test this

hypothesis, we performed RNA interference (RNAi)-mediated

knockdown of lincRNA-p21 and p53 separately and monitored

the resulting changes in mRNA levels by DNA microarray

analysis.

Toward this end, we first designed a pool of small interfering

RNA (siRNA) duplexes targeting lincRNA-p21, a pool targeting

p53 or nontargeting control sequences. We validated that each

pool was effective at knocking down its intended target genes

in p53LSL/LSL restored MEFs (Figures 3A and 3B). We then

used microarray analysis to examine the broader transcriptional

consequences of knockdown of p53 and lincRNA-p21. We iden-

tified 1520 and 1370 genes that change upon knockdown of p53

and lincRNA-p21, respectively (relative to nontargeting control

siRNA, FDR < 0.05). We observed a remarkable overlap of

930 genes in both the lincRNA-p21 and p53 knockdowns, vastly

more than would be expected by chance (p < 10�200) (Figure 3C,

Figure S3A, and Table S2). Strikingly, 80% (745/930) of the

common genes are derepressed in response to both p53 and

lincRNA-p21 knockdown, much higher proportion than ex-

pected by chance (p < 10�10) (Figure 3C and Table S2) when

compared to all genes affected by the p53 knockdown (Fig-

ure S3A). This observation suggests that lincRNA-p21 partici-

pates in downstream p53 dependent transcriptional repression.

To demonstrate that the observed derepression upon

lincRNA-p21 knockdown is indeed p53 dependent and is not

due to off target effects of the RNAi-mediated knockdown, we

performed several additional experiments and analyses. First,

we repeated the knockdown experiments with four individual

siRNAs targeting lincRNA-p21, transfected separately rather

than in a pool and confirmed the derepression effect on select

target genes (Figure S3F and Table S2). Second, we confirmed

that the same genes that were derepressed in the lincRNA-p21

and p53 knockdown experiments correspond to genes that are

A

B

D

[[

H3K4Me3

H3K36Me3

Sfrs3 Cdkn1a lincRNA-p21Chr17

75 kb

lincRNA-p21[RNA

LUNG TUMOR

*

AT T T

G G

C C C T

C

C A A AT C C G T C A A T T T T G T A G GCA A G A C GT T T G A A A A G G T C T T

G G A C A T G C C C G G G C A T G T CC

PETS

PET overlapdensity

153 bp500 bp

p53 motif

3073 bp

*

LYMPHOMA

0

0.20.40.60.811.2

0 12 24 36 48 72time after p53 restoration

(hours)

level A

NR evit al e

R

0 8 16 24 40 480

0.30.71

1.31.72

time after p53 restoration (hours)

level A

NR evit al e

R 0

0.20.40.60.811.2

0 12 24 36 48

SARCOMA

time after p53 restoration (hours)

level A

NR evit al e

R

lincRNA-p21Cdkn1ap53

0

1

2

3

4

5

6

Rel

ativ

e R

NA

leve

l

UntreatedDox

RT-PCR -RT

C

human lincRNA-p21

Figure 2. LincRNA-p21: A p53 Target Gene

Induced in Different Tumor Models

(A) Schematic representation of the chromosomal

location of the lincRNA-p21 gene locus. Arrow-

heads indicate the orientation of transcription.

(B) Promoter and transcript structure of lincRNA-

p21 gene locus. Chromatin structure at the

lincRNA-p21 locus is shown as mESC ChIP-Seq

data (Mikkelsen et al., 2007); for each histone

modification (green, H3K4me3; blue, H3K36me3),

ChIP-seq results are plotted as number of DNA

fragments obtained at each position relative to

the genomic average. Red stars indicate the posi-

tion of the p53-binding motif. The promoter region

where p53 ChIP-PET fragments (black segments)

map is enlarged (Wei et al., 2006). PET overlap

density (gray) and p53 motif sequence are shown.

The structure of the full-length lincRNA-p21 is rep-

resented with red boxes as exons and arrowed

lines as the intronic sequence.

(C) Human lincRNA-p21 is induced by DNA

damage. Relative RNA levels of human lincRNA-

p21 determined by qRT-PCR (RT-PCR) or qPCR

(�RT) from untreated human fibroblasts or 500 nM

DOX-treated for 14 hr. PCR primers map on the

human region orthologus to the first exon of the

mouse gene.

(D) LincRNA-p21 is induced by p53 in different

tumor cell lines. LincRNA-p21, p53, and Cdkn1a

relative RNA levels at different times after p53

restoration.

Values in (C) and (D) are the median of four tech-

nical replicates (±STD).

See also Figure S2.


normally repressed upon p53 induction in both the KRAS and

MEF systems, in the absence of RNAi treatment (GSEA FDR <

0.002) (Figure 3D). Third, we demonstrated that enforced

expression of lincRNA-p21 (Experimental Procedures) also per-

turbed the expression of genes that are normally regulated by

p53 in both the KRAS and MEF systems (GSEA FDR < 0.01)

(Figure S3H). Finally, we repeated the siRNA experiments in

the absence of p53 (dox/-AdCre) and demonstrated that dere-

pression of these genes did not occur upon siRNA-mediated

knockdown in the absence of p53 (Figure S3I). Collectively,

these results indicate that lincRNA-p21 acts to repress many

genes in p53-dependent transcriptional response.

lincRNA-p21 Regulates ApoptosisThe activation of the p53 pathway has two major phenotypic

outcomes: growth arrest and apoptosis (Levine et al., 2006).

Consistent with this, our microarray analysis demonstrates that

p53 and lincRNA-p21 both regulate a number of apoptosis

and cell-cycle regulator genes (Figure 3E, Figure S3G, and

Table S2, parts A and B). Thus, we aimed to determine the phys-

iological role of lincRNA-p21 in these processes.

Toward this end, we used RNAi-mediated knockdown of

lincRNA-p21 in dox-treated or untreated primary MEFs. We simi-

larly performed RNAi-mediated knockdown of p53 (as a positive

control) or used the nontargeting siRNA pool (as a negative

control) under the same conditions. We observed a significant

increase in viability after DNA damage of cells treated with

siRNAs targeting either lincRNA-p21 or p53 compared to those

treated with the control siRNA pool (Figures 4A and 4B). The

increase in viability was greater for knockdown of p53, but was

still highly significant for knockdown of lincRNA-p21 (p < 0.01).

We observed similar results when using three individual siRNA

duplexes targeting lincRNA-p21, as well as two different control

siRNA pools (Figure 4B and Figures S4A–S4C). These results

suggest that lincRNA-p21 plays a physiological role in regulating

cell viability upon DNA damage in this system, although they do

not discriminate whether the effect is due to misregulation of the

cell cycle or apoptosis.

To distinguish between these two possibilities, we first exam-

ined whether cell-cycle regulation in response to DNA damage is

affected by knockdown of p53 and lincRNAp-21. Specifically,

we assayed 5-bromo-2-deoxyuridine (BrdU) incorporation and

propidium iodide staining of the cells by fluorescence-activated

cell sorting (FACS) analysis. Consistent with the ability of p53

to inhibit cell-cycle progression, knockdown of p53 caused a

significant increase in BrdU incorporation in response to DNA

damage (p < 0.01). In contrast, knockdown of lincRNA-p21

showed no significant changes in either BrdU levels or in the

percentages of cells in any of the cell-cycle phases (S, G1, or

G2) with or without dox treatment (Figure 4C). These results

suggest that lincRNA-p21 does not substantially contribute to

cell-cycle arrest upon DNA damage.

siRNAcontrol

siRNAp53

siRNAlincRNA-p21

p531520

p53 &

lincRNA-p21930

lincRNA-p211370

80%

20%

P< 10-10

0

0.2

0.4

0.6

0.8

1

1.2

siRNAcontrol

siRNAlincRNA-p21

siRNAp53

p53lincRNA-p21

A B C

p53

βActin

contr

ollin

cRNA

-p21

p53

siRNA

D

Rel

ativ

e R

NA

leve

l

-8 +8

p53

APOPTOSIS

LinRNA-p21

Mdm2 Cdkn1a Cdkn1bBcl2l3 Mapk3Perp G2e3Stat3 Atf2

CELL CYCLE ARREST

ReprimoFas BaxCabc1 Noxa Brca1Wt1Cyclin G Cdk4CyclinD2

E

Phka2 Mtap4Vcan

0

0.2

0.4

-0.2

KRAS

0.2

0.0

-0.2

MEF

FDR<0.001 FDR<0.002

Repressed by p53

Inducedby p53

Repressed by p53

Inducedby p53

* *

Enr

ichm

ent S

core

Enr

ichm

ent S

core

* *

0.4

-0.4 -0.4

Cdkn2a

Figure 3. LincRNA-p21 Is a Global

Repressor of Genes in the p53 Pathway

(A) RNAi-mediated knockdown of lincRNA-p21

and p53. Relative RNA levels determined by

qRT-PCR in p53-reconstitued p53LSL/LSL MEFs

transfected with the indicated siRNAs and treated

with DOX (median of four technical replicates

±STD).

(B) p53 protein levels after lincRNA-p21 and p53

knockdown from cells treated as in (A). bActin

levels are shown as loading control.

(C) Many genes are corepressed by lincRNA-p21

and p53. Top: Venn diagram of differentially ex-

pressed genes (FDR < 0.05) upon p53 knockdown

(left) or lincRNA-p21 knockdown (right); cells were

treated as in (A) and subjected to microarray anal-

ysis. Bottom: expression level of genes in lincRNA-

p21 and p53 siRNA-treated cells relative to control

siRNA experiments. Expression values are dis-

played in shades of red or blue relative to the

global median expression value across all experi-

ments (linear scale).

(D) Genes derepressed by lincRNA-p21 and p53

knockdown overlap with the genes repressed by

p53 restoration in the MEF and KRAS systems.

The black line represents the observed enrichment

score profile of genes in the lincRNA-p21/p53

derepressed gene set to the MEF or KRAS gene

sets, respectively.

(E) Genes corregulated by lincRNA-p21 and p53

are part of the p53 biological response. Examples

of genes affected by lincRNA-p21 and/or p53

siRNA-knockdown (FDR < 0.05). Downregulated

and upregulated genes are indicated with blue

arrows and red lines respectively.



We then examined the impact of lincRNA-p21 and p53 knock-

downs on apoptosis. To this end, we assayed the proportion of the

cell population undergoing apoptosis by measuring Annexin-V by

FACS analysis. We observed a significant decrease in the number

of apoptotic cells after DNA damage in both the lincRNA-p21 and

p53 depleted cells relative to the siRNA control (p < 0.01) (Figures

4D and 4E). We also observed a decrease in Caspase 3 cleavage

after knockdown of both p53 or lincRNA-p21, relative to controls

(Figure 4F). We next sought to determine whether, conversely to

lincRNA-p21 knockdown, the enforced expression of lincRNA-

p21 would result in an increased apoptosis. Indeed, lincRNA-

p21 overexpression in a lung cancer cell line harboring a KRAS

mutation (referred to as LKR) and in NIH/3T3 MEFs caused a sig-

nificant decrease in cell viability (Experimental Procedures, Fig-

ure 4G, and Figure S4E). This decrease in viability was due to

increased apoptosis in response to DNA damage (p < 0.01) and

not to an effect in cell-cycle regulation (Figures 4H and 4I and

Figure S4G). Together, these results demonstrate a reproducible

and similar reduction of apoptotic cells in response to DNA

damage in both lincRNA-p21 and p53 knockdown experiments.

A B

D

C

E

F

G

siRNAcontrol

siRNAp53

siRNAcontrol

siRNAp53

0

5

10

15

20

25

0 24 48 72

rebmunll ec

evit al eR

time after selection (hours)

untreated DOX

012345678

control lincRNA-p21

slleccit ot popaf o

%

*DOX

control vector

010203040506070

sllecfo%

G1 G2 S

lincRNA-p21

DOX

siRNAlincRNA-p21

siRNAlincRNA-p21

DOX

untreated

0

0.5

1

1.5

2

2.5rebmunll ec

evit al eR

0 24 48 72time after selection (hours)

control vector

lincRNA-p21

control vector

lincRNA-p21

* *

0

5

10

15

20

25

30

slleccit ot popaf o

%

0

1

2

3

4

G1G2S

62 20 18 59 2417 55 22 23

0

1

2

3

4

G1G2S

57 34 9 56 33 11 29 44 27

*

βActinAnnexin-V Annexin-V

1.9 14.3

1.2683.1Annexin-V

7-A

AD

Annexin-V

1.9 5.96

1.2690.9Annexin-V

7-A

AD

1.81 14.6

1.64Annexin-V

7-A

AD

81.9

siRNA control-1 siRNA control-2 siRNA p53

1.87 9.39

1.6187.17-A

AD

siRNA lincRNA-p21-1

1.8 6.56

0.9291.37-A

AD

siRNA lincRNA-p21-2

1.93 5.5

1.3592.497-A

AD

siRNA lincRNA-p21-3

H

siRNA control-1 siRNA control-2

siRNA p53-1 siRNA p53-2

siRNA lincRNA-p21-1

siRNA lincRNA-p21-2

siRNAcontrol

siRNAp53

siRNA

lincRNA-p21

siRNA

I

0

1

2

3

4

5

6

0 24 48 72

0

0.4

0.8

1.2

1.6

2.0

0 24 48 72

rebmunllec

evitaleR

rebmunllec

evitaleR

DOX

siRNA lincRNA-p21siRNA controlsiRNA p53

time post transfection (hours)

time post transfection (hours)

untreated

siRNA lincRNA-p21siRNA controlsiRNA p53

cont

rol

p53

lincp

RNA

-p21

Cleavedcaspase 3

Fol

d ch

ange

rel

ativ

eto

siR

NA

con

trol

Fol

d ch

ange

rel

ativ

eto

siR

NA

con

trol

Figure 4. LincRNA-p21 Is Required for

Proper Apoptotic Induction

(A) Increased cell viability of lincRNA-p21-

depleted cells. Relative number of siRNA-trans-

fected MEFs treated with 400 nM DOX from

24 hr after transfection (right) or untreated (left)

determined by MTT assay.

(B) Knockdown of lincRNA-p21 with individual

siRNAs increases cell viability. Images of MEFs

treated with different individual siRNAs after

48 hr of DOX treatment (72 hr after transfection).

(C) LincRNA-p21 knockdown doesn’t affect cell-

cycle regulation. Relative cell numbers in each

cell-cycle phase determined by FACS of BrdU

incorporation and PI staining of MEFs treated as

in (A). Numbers inside bars represent percentages

of cells in each phase.

(D) LincRNA-p21 knockdown causes a decrease

in cellular apoptosis. p53-reconstituted p53LSL/LSL

MEFs transfected with three individual siRNAs

targeting lincRNA-p21 (bottom), two independent

control siRNAs (upper left and middle) or a siRNA

pool targeting p53 (upper right). Twenty-four hours

after transfection, cells were treated with 400 nM

doxorubicin and 14 hr later were harvested and

subjected to FACS analysis. The x axis represents

Annexin-V and the y axis 7-AAD staining. The

percentage of cells in each quadrant is indicated.

(E) Decreased apoptosis caused by lincRNA-p21

knockdown. Percentage of Annexin-V-positive

cells (FACS) at 38 hr after transfection (14 hr of

400 nM DOX treatment) in MEFs treated as in (A).

(F) LincRNA-p21 knockdown in p53-reconstituted

p53LSL/LSL MEFs causes decrease in Caspase 3

cleavage. Levels of cleaved Caspase 3 or control

bActin in p53 reconstituted-p53LSL/LSL MEFs

treated with the indicated siRNA pools and

500 nM DOX for 14 hr.

(G) Decreased cell viability caused by lincRNA-p21

overexpression. Relative numbers of LKR cells

overexpressing lincRNA-p21 or control plasmid

determined by MTT assay.

(H) Overexpression of lincRNA-p21 causes cellular

apoptosis under DNA damage induction. Per-

centage of Annexin-V-positive LKR cells overex-

pressing lincRNA-p21 or control vector treated

with 500 nM DOX.

(I) LincRNA overexpression doesn’t affect cell-

cycle regulation. Cell-cycle analysis of DOX-

treated LKR cells overexpressing lincRNA-p21 or

control plasmid.

All values are the average of 3 biological replicates

(±STD). * p < 0.01 relative to controls.

Also see Figure S4.


Although MEFs typically respond to DNA damage by under-

going cell-cycle arrest rather than apoptosis (Kuerbitz et al.,

1992), several additional lines of evidence are consistent with

the observed apoptosis phenotype in response to knockdown

on p53 and lincRNA-p21. First, certain critical cell-cycle regula-

tors, such as Cdkn1a/p21, Cdkn2a, and Reprimo, are regulated

by p53 but not lincRNA-p21. For example, knockdown of

lincRNA-p21 perturbs neither the transcript levels of Cdkn1a/

p21 nor the protein stability (Figure S3E); this may explain why

lincRNA-p21 knockdown is insufficient to cause a cell-cycle

phenotype, yet the p53 knockdown is. Second, we observed

that both lincRNA-p21 and p53 knockdowns resulted in the

repression of apoptosis genes (Noxa and Perp) and derepres-

sion of cell survival genes (Bcl2l3, Stat3, and Atf2, among others)

(Figure 3E and Table S2). Moreover, the decrease of apoptotic

cells in response to knockdown of lincRNA-p21 was comparable

to that caused by knockdown of p53 (Figures 4D and 4E and

Figures S4D and S4E). Third, the apoptosis phenotype is depen-

dent on the dosage of dox-induced DNA damage (Figure S4D).

Thus, the apoptosis response is both p53 dependent and

lincRNA-p21 dependent, with this dependence confirmed in

multiple cell types and conditions (Figures 4B, 4D, 4F, and 4H

and Figures S4A–S4C). Collectively, these observations demon-

strate that lincRNA-p21 plays an important role in the p53-

dependent induction of cell death.

LincRNA-p21 Functions through Interactionwith hnRNP-KWe next wanted to investigate the mechanism by which

lincRNA-p21 mediates transcriptional repression. We have

recently reported that many lincRNAs regulate gene expression

through their interaction with several chromatin regulatory com-

plexes (Khalil et al., 2009). Thus, we hypothesized that lincRNA-

p21 could affect gene expression in a similar manner.

To test this, we first performed nuclear fractionation experi-

ments and confirmed that lincRNA-p21 is enriched in the nucleus

(Figure S5A). We next sought to identify proteins that are associ-

ated with lincRNA-p21 by an RNA-pulldown experiment. Specif-

ically, we incubated in vitro-synthesized biotinylated lincRNA-

p21 and antisense lincRNA-p21 transcripts (negative control)

with nuclear cell extracts and isolated coprecipitated proteins

with streptavidin beads (Experimental Procedures). We resolved

the RNA-associated proteins on a SDS-PAGE gel, cut out the

bands specific to lincRNA-p21, and subjected them to mass

spectrometry (Figures 5A and 5B). In all six biological replicates,

mass-spectrometry analysis identified heterogeneous nuclear

A B

C

Streptavidinbeads

Mass Spectrometry

Dan

itsen

se

RNA

lincR

NA-p21

hnRNP-K60 kDa

BiotinylatedRNA

Nuclearextract

0

1

2

3

4

5

6

7

1 2 3 4 5 6 7 8 9 10

123456

78

910

5’ 3’ length778 nt938 nt1459 nt2125 nt3073 nt3073 nt (antisense)2844 nt2633 nt1889 nt735 nt

hnRNP-K

RNA

E

F

0

10

20

30

40

hnRNP-Kantibody 1

hnRNP-Kantibody 2

lincRNA-p21p53βActin

Fol

d E

nric

hmen

t (R

IP/Ig

G)

lincRNA-p21βActinGapdh

G

Fol

d E

nric

hmen

t (R

IP/Ig

G)

Native RIP

X-linked RIP

hnRNP-K

Nono

5’ 778 nt

Full length

3’ 2633 nt

3’ 1889 nt

hnRNP-KInteraction

++--

% a

popt

opto

tic c

ells

0

4

8

12

16

vector

* ** *

ApoptosisInduction

+--

-

5’ 778 nt

Full length

3’ 2633 nt

3’ 1889 nt

Figure 5. LincRNA-p21 Physically Interacts

with hnRNP-K

(A) Schematic representation of RNA pulldown

experiments to identify associated proteins. Bioti-

nylated lincRNA-p21 or antisense RNA were incu-

bated with nuclear extracts, targeted with strepta-

vidin beads, washed, and associated proteins

resolved in a gel. Specific bands were cutout and

identified by mass spectrometry.

(B) LincRNA-p21 and hnRNP-K specifically

interact in vitro. SDS-PAGE gel of proteins bound

to lincRNA-p21 (right lane) or antisense RNA (left

lane). The highlighted region was submitted for

mass spectrometry identifying hnRNP-K as the

band unique to lincRNA-p21.

(C) Western blot analysis of the specific associa-

tion of hnRNP-K with lincRNA-p21. A nonspecific

protein (NONO) is shown as a control.

(D) Association between endogenous lincRNA-

p21 and hnRNP-K in the nucleus of DNA damaged

MEFs in native conditions. RNA Immunoprecipita-

tion (RIP) enrichment is determined as RNA asso-

ciated to hnRNP-K IP relative to IgG control.

(E) Physical association between lincRNA-p21 and

hnRNP-K after chemical crosslinking of life cells.

hnRNP-K was immunoprecipitated from nuclear

extracts of formaldehyde-crosslinked DNA-dam-

aged MEFs, and associated RNAs were detected

by RT-qPCR. The relative enrichment is calcu-

lated as in (D) and is the median of three techni-

cal replicates of a representative experiment

(±STD).

(F) LincRNA-p21 binds hnRNP-K through its 50

terminal region. RNAs corresponding to dif-

ferent fragments of lincRNA-p21 or its antisense

sequence (middle and bottom) were treated as in (A) and associated hnRNP-K was detected by western blot (top).

(G) Percentage of Annexin-V-positive LKR cells overexpressing the indicated lincRNA-p21 fragments or empty vector as control (average of three replicates

[±STD]). * p < 0.001.

See also Figure S5.


ribonucleoprotein K (hnRNP-K) as specifically associated with

the sense (but not antisense) strand of lincRNA-p21. We inde-

pendently verified this interaction by western blot analysis

(Figure 5C). hnRNP-K has been shown to play various roles in

the p53 pathway (Kim et al., 2008; Moumen et al., 2005). Interest-

ingly, among these roles, Kim et al. (2008) demonstrated that

hnRNP-K is a component of a repressor complex that acts in

the p53 pathway, consistent with our evidence that lincRNA-

p21 plays a role in global repression in this pathway.

To further validate the interaction between lincRNA-p21 and

hnRNP-K in our cell-based systems, we performed RNA immu-

noprecipitation (RIP) with an antibody against hnRNP-K from

nuclear extracts of MEFs subjected to DNA damage. We

observed an enrichment of lincRNA-p21 (but not other unrelated

RNAs) with hnRNP-K antibody as compared to the nonspecific

antibody (IgG control) (Figure 5D). We further performed analo-

gous RIP experiments with formaldehyde crosslinked cells

followed by stringent washing conditions (Ule et al., 2005) to

rule out potential nonspecific interactions. Consistent with

a bona fide interaction, we observed a greater and very signifi-

cant enrichment of lincRNA-p21 in the hnRNP-K RIP relative to

the IgG control RIP with two hnRNP-K different antibodies

(Figure 5E).

We further performed deletion-mapping experiments to deter-

mine whether hnRNP-K interacts within a specific region of

lincRNA-p21. To this end, we carried out RNA pulldown experi-

ments with truncated versions of lincRNA-p21 followed by

western blot detection of bound hnRNP-K. These analyses

identified a 780 nt region at the 50 end of lincRNA-p21 required

for the interaction with hnRNP-K (Figure 5F). Interestingly, RNA

folding analyses of this region based on sequence conservation

and compensatory changes across 14 mammalian species

(Hofacker, 2003) predict a highly stable 280 nt structure of

lincRNA-p21 with deep evolutionary conservation (Figures S5B

and S5C). Together, the RNA pulldown, native RIP, crosslinked

RIP, and deletion mapping results demonstrate a specific asso-

ciation between hnRNP-K and lincRNA-p21.

We next sought to determine the functional relevance of the

interaction between lincRNA-p21 and hnRNP-K. To do so, we

monitored the ability of different truncated versions of lincRNA-

p21 to induce cellular apoptosis when overexpressed in LKR

cells (Experimental Procedures). This revealed that the deletion

of the 50 end of lincRNA-p21, which mediates the hnRNP-K inter-

action, abolishes the ability of lincRNA-p21 to induce apoptosis

(Figure 5G). Interestingly, the 780 nt fragment at the 50 end of

lincRNA-p21 alone does not induce apoptosis, indicating that

this fragment is required but not sufficient for lincRNA-p21-medi-

ated cellular apoptosis.

We hypothesized that hnRNP-K is required for proper tran-

scriptional repression of target genes shared between p53 and

lincRNA-p21. If so, knockdown of hnRNP-K should result in

derepression of these shared targets. We tested this hypothesis

by performing siRNA-mediated knockdown of hnRNP-K,

lincRNA-p21, and p53 in p53-restored p53LSL/LSL MEFs, treating

the cells with dox and profiling the changes in gene expression

by microarray analysis.

Consistent with our previous data, we observed a strong over-

lap of 582 genes affected in the hnRNP-K, lincRNA-p21, and p53

knockdowns (FDR < 0.05) (Figure 6 and Figure S6). Remarkably,

83% of these common genes were derepressed in all three

knockdown experiments (Figure 6A and Figure S6D). The genes

previously identified as coregulated by lincRNA-p21 and

p53 also were strongly enriched (GSEA FDR < 10�4) among

those regulated by hnRNP-K (Figure 6B and Table S3). Thus,

lincRNA-p21 and hnRNP-K play roles in repressing a significant

common set of genes in the p53-dependent response to DNA

damage.

We further reasoned that if hnRNP-K is involved in the repres-

sion of genes corepressed by p53 and lincRNA-p21, then

hnRNP-K might also be physically bound to the promoters of

these genes. To test this, we performed ChIP experiments with

antibodies against hnRNP-K, followed by hybridization to DNA

tiling microarrays covering 30,000 gene promoters. We identified

1621 promoter regions with significant occupancy by hnRNP-K

(FDR < 0.05) (Figure 6 and Table S3). Notably, these promoter

regions exhibit a significant overlap with genes that were differ-

entially expressed upon hnRNP-K knockdown (GSEA FDR <

0.001) (Figure S6E). Moreover, hnRNP-K localizes to a significant

fraction (FDR < 0.001) of the genes corepressed by lincRNA-p21

and p53 (Figure 6C), suggesting that these are primary sites of

hnRNP-K regulation.

We next wanted to determine whether lincRNA-p21 plays

a role in hnRNP-K localization at promoters of p53-repressed

genes. To this end, we determined whether siRNA-mediated

knockdown of lincRNA-p21 affected the localization of hnRNP-K

after induction of p53. Specifically, we performed hnRNP-K

ChIP in dox-treated p53-restored p53LSL/LSL MEFs transfected

with either siRNAs targeting lincRNA-p21 or nontargeting con-

trol siRNAs. These experiments revealed that the depletion of

lincRNA-p21 causes a significant reduction in the association

of hnRNP-K at the promoter regions of genes that are normally

repressed in a lincRNA-p21- and p53-dependent fashion, as

determined by ChIP-qPCR (Figure 6D). Specifically, 12 of the

15 tested promoter regions exhibited loss of hnRNP-K enrich-

ment, in two biological replicate experiments, upon depletion

of lincRNA-p21. Thus, hnRNP-K is bound to the promoters of

genes that are normally repressed in a p53- and lincRNA-p21-

dependent manner, and this localization requires lincRNA-p21.

Collectively, our results indicate that lincRNA-p21 is a direct

p53 transcriptional target in response to DNA damage, acts to

repress genes that are downregulated as part of the canonical

p53 transcriptional response, is necessary for p53 dependent

apoptotic responses to DNA damage in our cell-based systems,

and functions at least in part through interaction with hnRNP-K

by modulating hnRNP-K localization.

DISCUSSION

It is clear that mammalian genomes encode numerous large

noncoding RNAs (Carninci, 2008; Guttman et al., 2009; Mattick,

2009; Ponjavic et al., 2007). Here, we demonstrate that

numerous lincRNAs are key constituents in the p53-dependent

transcriptional pathway. Moreover, we observed that some of

these lincRNAs are bound by p53 in their promoter regions and

sufficient to drive p53-dependent reporter activity that requires


the consensus p53-binding motif, suggesting that these

lincRNAs are bona fide p53 transcriptional targets.

Having discovered multiple lincRNAs in the p53 pathway, we

decided to focus on one such lincRNA in particular: lincRNA-

p21. Intrigued by its properties (genomic location upstream of

p21, p53-dependent activation requiring the consensus p53

motif, which is bound by p53 and conserved p53-dependent

activation of this gene in both human and mouse cell-based

systems), we explored the functional roles of lincRNA-p21. Our

studies revealed a role for lincRNA-p21 in a p53-dependent

apoptotic response after DNA damage.

We further observed that siRNA-mediated inhibition of lincRNA-

p21 affects the expression of hundreds of gene targets that are

enriched for genes normally repressed by p53 in both the MEF

and RAS cell-based systems. Strikingly, the vast majority of these

common target genes are derepressed upon inhibition of either

p53 or lincRNA-p21—suggesting that lincRNA-p21 functions as

a downstream repressor in the p53 transcriptional response.

We gained mechanistic clues into how lincRNA-p21 functions

to repress such a large subset of the p53 transcriptional

response by biochemical experiments that identified a specific

interaction between lincRNA-p21 and hnRNP-K. This interaction

is supported by RNA-pulldown, native RIP, crosslinked RIP, and

deletion-mapping experiments. Moreover, we identified a 780 nt

50 region of lincRNA-p21 that is required for hnRNP-K binding

and subsequent induction of apoptosis. Interestingly, this region

is much more highly conserved than the remainder of the tran-

script. This suggests that patches of conservations, previously

determined to be unique to lincRNAs, (Guttman et al., 2009),

may point to functional elements for binding interactions within

A

B

C

hnRNP-K

D

p53

Enr

ichm

ent S

core

0.0

0.5

Enr

ichm

ent S

core

0.0

0.5

siRNA control

siRNA hnRNP-K

siRNA lincRNA-p21

siRNA p53

P< 10-80

17%

83%

-8 +8

582

gene

s

0

-2

+3

Wt1

0

-2

+3

Cdkn2a

0

-2

+3

Rb1

0

-2

+3

G2e3

0

-2

+3

Mtap4

0

-2

+3

Suv39h1

0

-2

+3

Vcan

P<0.001

lincRNA-p21/p53

repressed

hnRNP-Kbound

FDR<.001 FDR<.001

Rela

tive

Bind

ing

Enric

hmen

t (hn

RNP-

K C

hIP/

IgG

)

**

**

0

2

4

6

8

10

12

14

siRNA control siRNA lincRNA-p21

115

35

45 39

2620 31

Rela

tive

Fold

Enr

ichm

ent

(hnR

NPK

ChI

P/Ig

G)

E

VcanCxcr6

Hus1Jmjd3

Zbtb20

Atf2Rb1

Mtpa4

Pdlim2

Usp25

Zfp386

Cdkn2a

Gapdh

Lpp

hnRNP-K

p53

?

lincRNA-p21

hnRNP-K

POL II

lincRNA-p21

hnRNP-K

p53

582

Figure 6. LincRNA-p21 and hnRNP-K Corepress Genes in the p53

Transcriptional Response

(A) Many genes are coregulated by p53, lincRNA-p21, and hnRNP-K. Genes

affected by knockdown of lincRNA-p21, p53, or hnRNP-K in p53-restored-

DNA-damaged p53LSL/LSL MEFs determined by microarray analysis (FDR <

0.05). Shades of red or blue represent expression values relative to global

median across experiments. Percentages of up- and downregulated genes

are indicated.

(B) Genes repressed by lincRNA-p21 are significantly enriched in genes

repressed by p53 and hnRNPK. GSEA comparing the genes upregulated on

knockdown of LincRNA-p21 and those upregulated upon knockdown of p53

(left) or hnRNP-K (right). The black line represents the observed enrichment

score profile of genes in the lincRNA-p21 gene set to the p53 or hnRNP-K

gene sets, respectively.

(C) hnRNP-K associates to promoters of genes corepressed by lincRNA-p21

and p53. Examples of promoters of genes repressed by p53 and lincRNA-

p21 (G2e3, Mtap4, Suv39h1, and Vcan) or repressed by lincRNA-p21

but not p53 (Rb1) bound by hnRNP-K (blue) determined by ChIP-chip of

hnRNP-K in dox-trated p53-reconstituted p53LSL/LSL MEFs (FDR < 0.05).

Cdkn2a and Wt1 are negative controls (gray).

(D) hnRNP-K binding to lincRNA-p21 and p53 corepressed genes is depen-

dent on lincRNA-p21. Relative enrichment of hnRNP-K (ChIP-qPCR) in the

indicated promoter regions in p53-reconstituted p53LSL/LSL MEFs transfected

with siRNA lincRNA-p21 or siRNA control and dox-treated determined by

ChIP-qPCR (representative of two biological replicates shown ±STD).

(E) Proposed models for the function of licRNA-p21 in the p53 transcriptional

response. Induction of p53 activates the transcription of lincRNA-p21 by

binding to its promoter (upper left). LincRNA-p21 binds to hnRNP-K, and this

interaction imparts specificity to genes repressed by p53 induction (upper right).



lincRNAs (as was also recently determined for Xist binding to

PRC2) (Zhao et al., 2008).

hnRNP-K is known to interact with other repressive complexes

such as the Histone H1.2 or members of the Polycomb-group

(PcG) (Kim et al., 2008; Denisenko and Bomsztyk, 1997). The

physical interaction between lincRNA-p21 and hnRNP-K is likely

required for lincRNA-p21-mediated gene repression, as loss of

hnRNP-K function results in the derepression of the same genes

that are repressed by both p53 and lincRNA-p21. Importantly,

genome-wide ChIP-chip analysis revealed hnRNP-K binding at

the promotersof thesecorepressed gene loci, suggestiveof direct

regulation by hnRNP-K and lincRNA-p21. Moreover, we observed

a lincRNA-p21 dependent binding of hnRNP-K at several of these

corepressed promoter regions. While hnRNP-K has been previ-

ously shown to activate one gene in the p53 pathway (Moumen

et al., 2005), our analyses suggest that it plays a much more

widespread role in repression. Together, these results implicate

lincRNA-p21 as an important repressor in the p53 pathway, by

interacting with and modulating the localization of hnRNP-K.

Our results raise the possibility that many transcriptional

programs (beyond the p53-pathway) may involve inducing

protein factors that activate specific sets of downstream genes

and lincRNAs that repress previously active sets of genes. The

notion of a noncoding RNA being involved in silencing-specific

gene loci is consistent with our recent observation that many

lincRNAs (including lincRNA-p21) bind to chromatin complexes

(such as PRC2) and are required to mediate repression at key

gene loci (Khalil et al., 2009). Moreover, there are several exam-

ples of lincRNAs involved in repression of known target genes—

including HOTAIR-dependent repression of HOXD genes (Rinn

et al., 2007) and XIST, AIR, and KCNQ1OT1, involved in genomic

imprinting and silencing of several genes in cis (Nagano et al.,

2008; Pandey et al., 2008; Zhao et al., 2008).

The precise mechanism by which lincRNA-p21 contributes to

repression at specific loci remains to be defined. Various possi-

bilities include that (1) lincRNA-p21 might direct a protein

complex to specific loci by Crick-Watson base pairing; (2)

lincRNAs might act by forming DNA-DNA-RNA triple-helical

structures, which do not require Crick-Watson base-pairing,

such as reported for a large noncoding RNA that forms a

triple-helix upstream of the Dihydrofolate Reductase (DHFR)

promoter resulting in repression of DHFR (Martianov et al.,

2007); or (3) lincRNAs might alter the binding specificity of

DNA-binding proteins (such as hnRNP-K) to influence their target

preference (Figure 6D). Further experiments are needed to

distinguish between these and other possibilities.

Aside from the general interest in gene regulation, we note that

lincRNA-p21 and several other lincRNAs function in an important

pathway for cancer. It is tempting to speculate that other

lincRNAs may also play key roles in numerous other tumor-

suppressor and oncogenic pathways, representing a hitherto

unknown paradigm in cellular transformation and metastasis.

It will be important for future studies to determine whether

lincRNAs genes can serve as tumor suppressor genes or

oncogenes.

In summary, lincRNAs may point to new mechanisms of gene

regulation, components in disease pathways and potential

targets for the development of therapies.


Cell Lines and In Vivo Models

KRAS lung tumor-derived cell lines were isolated from individual tumors (D.F.

and T.J., unpublished data). Isolation of matched p53+/+ and p53�/� MEFs,

p53LSL/LSL MEFs, lymphomas, and sarcomas, and p53 restoration were

done as described (Ventura et al., 2007). Primary WT MEFs and NIH/3T3

MEF cells were purchased from ATCC. Transfection, infection, and treatment

conditions are described in the Extended Experimental Procedures.

Promoter Reporter Assays

LincRNA promoters were cloned into pGL3-basic vector (Promega), and motif

deletions were performed by mutagenesis. p53-reconstituted or control

p53LSL/LSL MEFs were transfected with 800 ng pGL3 and 30 ng TK-Renilla

plasmid per 24-well. Twenty-four hours later, cells were treated with 500 nM

dox for 13 hr, and cell extracts were assayed for firefly and renilla luciferase

activities (Promega E1910).

LincRNA and Gene-Expression Profiling and Informatic Analyses

RNA isolation, lincRNA expression profiling, and ChIP-chip analyses (Nimble-

gen arrays), as well as Affymetrix gene-expression profiling and analyses, were

performed as described (Guttman et al., 2009) (Extended Experimental Proce-

dures). Structure predictions were performed using the Vienna RNA package

(Hofacker, 2003).

Antibodies

The following antibodies were used: anti-p53, Novocastra (NCL-p53-CM5p)

(western blot) and Vector Labs (CM-5) (ChIP); anti-hnRNP-K, Santa Cruz

Biotechnology (sc-25373) (western blot) and Abcam (Ab70492 and Ab39975)

(ChIP and RIP); and control rabbit IgG Abcam (Ab37415-5) (RIP and ChIP-

chip).

Viability and Apoptosis Assays and Cell-Cycle Analysis

MTT assays were performed with Cell Proliferation Kit I from Roche

(11465007001). For apoptosis quantification, the Apoptosis Detection Kit I

from BD Biosciences (cat#559763) and FACS (van Engeland et al., 1996)

were used. Cell-cycle analysis was performed as described (Brugarolas

et al., 1995).

Cloning, RNA Pulldown, Deletion Mapping, RIP, and ChIP

50 and 30 RACE cloning of lincRNA-p21 were performed from total RNA of dox-

treated MEFs with RLM-RACE Kit (Ambion). RNA pulldown and deletion

mapping were performed as described (Rinn et al., 2007) with 1 mg mESC

nuclear extract and 50 pmol of biotinylated RNA. Mass spectrometry was per-

formed as described (Shevchenko et al., 1996). Native RIP was carried out as

described (Rinn et al., 2007). For crosslinked RIP, cells were crosslinked with

1% formaldehyde, antibody incubated overnight, recovered with protein G

Dynabeads, and washed with RIPA buffer. After reverse crosslink, RNA was

analyzed by qRT-PCR. p53 ChIP and hnRNP-K ChIP experiments were per-

formed as previously described (Rinn et al., 2007) (Extended Experimental

Procedures).

RNA Interference and LincRNA-p21 Overexpression

siRNA transfections were done with 75 nM siRNA and Lipofectamine 2000

(Invitrogen). For overexpression, lincRNA-p21 or truncated forms were cloned

into the pBABE vector. After transfection, cells were selected with 2 mg/ml

puromycin.

ACCESSION NUMBERS

The accession number for the full-length mouse lincRNA-p21 sequence

reported in this paper is HM210889 (bankit1350506). All primary data are avail-

able at the Gene Expression Omnibus (GSE21761).



Supplemental Information includes Extended Experimental Procedures, six

figures, and four tables and can be found with this article online at doi:10.

1016/j.cell.2010.06.040.

ACKNOWLEDGMENTS

We would like to thank Loyal A. Goff (Massachusetts Institute of Technology

[MIT]) for bioinformatic support, Nadya Dimitrova (MIT) for input on the manu-

script, David Garcia (MIT) for experimental assistance, and Sigrid Hart (Broad

Institute) for illustration support. J.L.R. is a Damon Runyon-Rachleff, Searle,

and Smith Family Foundation Scholar. J.L.R. and A.R. are Richard Merkin

Foundation Scholars. This work was supported by the National Institutes of

Health (NIH) Director’s New Innovator Award, Smith Family Foundation,

Damon Runyon Cancer Foundation, Searle Scholar Program, and NIH

1R01CA119176-01.

Received: October 6, 2009



Published online: July 29, 2010

REFERENCES

Brugarolas, J., Chandrasekaran, C., Gordon, J.I., Beach, D., Jacks, T., and

Hannon, G.J. (1995). Radiation-induced cell cycle arrest compromised by

p21 deficiency. Nature 377, 552–557.

Carninci, P. (2008). Non-coding RNA transcription: turning on neighbours. Nat.

Cell Biol. 10, 1023–1024.

Denisenko, O.N., and Bomsztyk, K. (1997). The product of the murine homolog

of the Drosophila extra sex combs gene displays transcriptional repressor

activity. Mol. Cell. Biol. 17, 4707–4717.

el-Deiry, W.S., Kern, S.E., Pietenpol, J.A., Kinzler, K.W., and Vogelstein, B.

(1992). Definition of a consensus binding site for p53. Nat. Genet. 1, 45–49.

Funk, W.D., Pak, D.T., Karas, R.H., Wright, W.E., and Shay, J.W. (1992). A tran-

scriptionally active DNA-binding site for human p53 protein complexes. Mol.

Cell. Biol. 12, 2866–2871.

Garber, M., Guttman, M., Clamp, M., Zody, M.C., Friedman, N., and Xie, X.

(2009). Identifying novel constrained elements by exploiting biased substitu-

tion patterns. Bioinformatics 25, i54–i62.

Guttman, M., Amit, I., Garber, M., French, C., Lin, M.F., Feldser, D., Huarte, M.,

Zuk, O., Carey, B.W., Cassady, J.P., et al. (2009). Chromatin signature reveals

over a thousand highly conserved large non-coding RNAs in mammals. Nature

458, 223–227.

Hofacker, I.L. (2003). Vienna RNA secondary structure server. Nucleic Acids

Res. 31, 3429–3431.

Khalil, A.M., Guttman, M., Huarte, M., Garber, M., Raj, A., Rivea Morales, D.,

Thomas, K., Presser, A., Bernstein, B.E., van Oudenaarden, A., et al. (2009).

Many human large intergenic noncoding RNAs associate with chromatin-

modifying complexes and affect gene expression. Proc. Natl. Acad. Sci.

USA 106, 11667–11672.

Kim, K., Choi, J., Heo, K., Kim, H., Levens, D., Kohno, K., Johnson, E.M.,

Brock, H.W., and An, W. (2008). Isolation and characterization of a novel

H1.2 complex that acts as a repressor of p53-mediated transcription.

J. Biol. Chem. 283, 9113–9126.

Kuerbitz, S.J., Plunkett, B.S., Walsh, W.V., and Kastan, M.B. (1992). Wild-type

p53 is a cell cycle checkpoint determinant following irradiation. Proc. Natl.

Acad. Sci. USA 89, 7491–7495.

Levine, A.J., Hu, W., and Feng, Z. (2006). The P53 pathway: what questions

remain to be explored? Cell Death Differ. 13, 1027–1036.

Martianov, I., Ramadass, A., Serra Barros, A., Chow, N., and Akoulitchev, A.

(2007). Repression of the human dihydrofolate reductase gene by a non-

coding interfering transcript. Nature 445, 666–670.

Mattick, J.S. (2009). The genetic signatures of noncoding RNAs. PLoS Genet.

5, e1000459.

Mercer, T.R., Dinger, M.E., and Mattick, J.S. (2009). Long non-coding RNAs:

insights into functions. Nat. Rev. Genet. 10, 155–159.

Mikkelsen, T.S., Ku, M., Jaffe, D.B., Issac, B., Lieberman, E., Giannoukos, G.,

Alvarez, P., Brockman, W., Kim, T.K., Koche, R.P., et al. (2007). Genome-wide

maps of chromatin state in pluripotent and lineage-committed cells. Nature

448, 553–560.

Moumen, A., Masterson, P., O’Connor, M.J., and Jackson, S.P. (2005). hnRNP

K: an HDM2 target and transcriptional coactivator of p53 in response to DNA

damage. Cell 123, 1065–1078.

Nagano, T., Mitchell, J.A., Sanz, L.A., Pauler, F.M., Ferguson-Smith, A.C., Feil,

R., and Fraser, P. (2008). The Air noncoding RNA epigenetically silences tran-

scription by targeting G9a to chromatin. Science 322, 1717–1720.

Pandey, R.R., Mondal, T., Mohammad, F., Enroth, S., Redrup, L., Komorowski,

J., Nagano, T., Mancini-Dinardo, D., and Kanduri, C. (2008). Kcnq1ot1 anti-

sense noncoding RNA mediates lineage-specific transcriptional silencing

through chromatin-level regulation. Mol. Cell 32, 232–246.

Ponjavic, J., Ponting, C.P., and Lunter, G. (2007). Functionality or transcrip-

tional noise? Evidence for selection within long noncoding RNAs. Genome

Res. 17, 556–565.

Ponting, C.P., Oliver, P.L., and Reik, W. (2009). Evolution and functions of long

noncoding RNAs. Cell 136, 629–641.

Riley, T., Sontag, E., Chen, P., and Levine, A. (2008). Transcriptional control of

human p53-regulated genes. Nat. Rev. Mol. Cell Biol. 9, 402–412.

Rinn, J.L., Kertesz, M., Wang, J.K., Squazzo, S.L., Xu, X., Brugmann, S.A.,

Goodnough, L.H., Helms, J.A., Farnham, P.J., Segal, E., and Chang, H.Y.

(2007). Functional demarcation of active and silent chromatin domains in

human HOX loci by noncoding RNAs. Cell 129, 1311–1323.

Schaefer, B.C. (1995). Revolutions in rapid amplification of cDNA ends: new

strategies for polymerase chain reaction cloning of full-length cDNA ends.

Anal. Biochem. 227, 255–273.

Shevchenko, A., Wilm, M., Vorm, O., Jensen, O.N., Podtelejnikov, A.V., Neu-

bauer, G., Shevchenko, A., Mortensen, P., and Mann, M. (1996). A strategy

for identifying gel-separated proteins in sequence databases by MS alone.

Biochem. Soc. Trans. 24, 893–896.

Subramanian, A., Tamayo, P., Mootha, V.K., Mukherjee, S., Ebert, B.L.,

Gillette, M.A., Paulovich, A., Pomeroy, S.L., Golub, T.R., Lander, E.S., and

Mesirov, J.P. (2005). Gene set enrichment analysis: a knowledge-based

approach for interpreting genome-wide expression profiles. Proc. Natl.

Acad. Sci. USA 102, 15545–15550.

Ule, J., Jensen, K., Mele, A., and Darnell, R.B. (2005). CLIP: a method for iden-

tifying protein-RNA interaction sites in living cells. Methods 37, 376–386.

van Engeland, M., Ramaekers, F.C., Schutte, B., and Reutelingsperger, C.P.

(1996). A novel assay to measure loss of plasma membrane asymmetry during

apoptosis of adherent cells in culture. Cytometry 24, 131–139.

Vazquez, A., Bond, E.E., Levine, A.J., and Bond, G.L. (2008). The genetics of

the p53 pathway, apoptosis and cancer therapy. Nat. Rev. Drug Discov. 7,

979–987.

Ventura, A., Kirsch, D.G., McLaughlin, M.E., Tuveson, D.A., Grimm, J., Lintault,

L., Newman, J., Reczek, E.E., Weissleder, R., and Jacks, T. (2007). Restoration

of p53 function leads to tumour regression in vivo. Nature 445, 661–665.

Wei, C.L., Wu, Q., Vega, V.B., Chiu, K.P., Ng, P., Zhang, T., Shahab, A., Yong,

H.C., Fu, Y., Weng, Z., et al. (2006). A global map of p53 transcription-factor

binding sites in the human genome. Cell 124, 207–219.

Zhao, J., Sun, B.K., Erwin, J.A., Song, J.J., and Lee, J.T. (2008). Polycomb

proteins targeted by a short repeat RNA to the mouse X chromosome. Science

322, 750–756.


A Minimal Midzone Protein ModuleControls Formation and Length ofAntiparallel Microtubule OverlapsPeter Bieling,1,2 Ivo A. Telley,1 and Thomas Surrey1,*1Cell Biology and Biophysics Unit, European Molecular Biology Laboratory, Meyerhofstrasse 1, 69117 Heidelberg, Germany2Present address: Department of Cellular and Molecular Pharmacology, University of California, San Francisco, San Francisco,

CA 94158, USA

*Correspondence: [email protected] 10.1016/j.cell.2010.06.033

SUMMARY

During cell division, microtubules are arranged in alarge bipolar structure, the mitotic spindle, to segre-gate the duplicated chromosomes. Antiparallel mic-rotubule overlaps in the spindle center are essentialfor establishing bipolarity and maintaining spindlestability throughout mitosis. In anaphase, this anti-parallel microtubule array is tightly bundled formingthe midzone, which serves as a hub for the recruit-ment of proteins essential for late mitotic events.The molecular mechanism of midzone formationand the control of its size are not understood. Usingan in vitro reconstitution approach, we show herethat PRC1 autonomously bundles antiparallel micro-tubules and recruits Xklp1, a kinesin-4, selectively tooverlapping antiparallel microtubules. The proces-sive motor Xklp1 controls overlap size by overlaplength-dependent microtubule growth inhibition.Our results mechanistically explain how the twoconserved, essential midzone proteins PRC1 andXklp1 cooperate to constitute a minimal proteinmodule capable of dynamically organizing the corestructure of the central anaphase spindle.

INTRODUCTION

The various architectures adopted by the microtubule cytoskel-

eton are determined by distinct combinations of mechanochem-

ical protein activities. How global properties of a complex

cytoskeletal system, for instance characteristic shape and size,

emerge from the local biochemical properties of the constituting

molecules is an open question. During cell division, the microtu-

bule cytoskeleton forms the mitotic spindle to segregate the

duplicated set of chromosomes (Walczak and Heald, 2008).

Spindles consist of two antiparallel arrays of microtubules that

overlap in the spindle center, an organization that ensures

spindle function. At anaphase onset, the central spindle reorga-

nizes to form the midzone, a dense specialized array of antipar-

allel microtubules (Glotzer, 2009; Khmelinskii and Schiebel,

2008). The midzone size in metazoans is typically 2–3 mm in

length (Mastronarde et al., 1993). Several proteins essential for

spindle elongation, faithful chromosome segregation, and pro-

per cytokinesis relocalize selectively to the central spindle as a

consequence of changes in their phosphorylation patterns in

anaphase: PRC1 (protein required for cytokinesis 1), kinesin-4,

the centralspindlin complex (containing a kinesin-6), the chro-

mosome passenger complex, and several other molecular

motors, nonmotor microtubule-bundling proteins, and regula-

tors (Glotzer, 2009). How the organization of the midzone is de-

termined by the midzone proteins is, however, not understood.

The conserved vertebrate-bundling protein PRC1 (Jiang et al.,

1998; Kurasawa et al., 2004; Mollinari et al., 2002; Zhu and Jiang,

2005) and its orthologs (Ase1 in yeasts [Loiodice et al., 2005;

Schuyler et al., 2003], SPD-1 in C. elegans [Verbrugghe and

White, 2004], Feo in D. melanogaster [Verni et al., 2004], and

MAP65 in plants [Muller et al., 2004]) have emerged as key

players for midzone formation. Ase1 in budding yeast has been

proposed to be the core component of the spindle midzone.

Genetic experiments have demonstrated that localization of all

midzone proteins depends on Ase1, whose localization is inde-

pendent of most other midzone proteins (Khmelinskii et al.,

2007). Interestingly, fission yeast Ase1 has been shown to bind

with some preference to antiparallel versus parallel microtubule

bundles in vitro (Janson et al., 2007; Kapitein et al., 2008), indi-

cating that recognition of antiparallel microtubules is an autono-

mous property of Ase1 in yeast.

In contrast to yeast, PRC1 in higher eukaryotes interacts with

motor proteins of the kinesin-4 type. These chromokinesin

motors relocalize to the central spindle in anaphase (Kurasawa

et al., 2004; Kwon et al., 2004; Lee and Kim, 2004; Powers

et al., 2004; Vernos et al., 1995; Zhu and Jiang, 2005). Correct

localization of both PRC1 and kinesin-4 to the midzone depend

on each other (Kurasawa et al., 2004; Zhu and Jiang, 2005; Zhu

et al., 2005). However, whether kinesin-4 transports PRC1 to the

midzone (Zhu and Jiang, 2005) or PRC1 selectively binds the

central region of the anaphase spindle thereby recruiting kine-

sin-4 (Kurasawa et al., 2004) is presently unclear. Importantly,

it is unknown how the size of the midzone is determined.

Whereas absence of PRC1 results in complete loss of midzone

microtubule bundles in mammalian cells (Kurasawa et al.,

2004; Mollinari et al., 2002; Zhu and Jiang, 2005), absence of


kinesin-4 leads to increased anaphase spindle length (Kurasawa

et al., 2004; Zhu and Jiang, 2005). Interestingly, a fragment of

Xenopus kinesin-4 has been shown to decrease microtubule

growth and shrinkage rates in vitro (Bringmann et al., 2004), an

activity different from other regulators of microtubule dynamics

(Howard and Hyman, 2007). Hence, kinesin-4 might contribute

to midzone formation by regulating midzone microtubule

dynamics. The minimal set of protein activities required for

specific organizational features of the spindle midzone and the

logic behind the combination of the required activities are,

however, unknown.

We have developed an in vitro reconstitution approach to

investigate the combinatorial action of purified full-length

Xenopus kinesin-4 (Xklp1) and Xenopus PRC1 on microtubule

organization and dynamics. We show that Xklp1 is recruited to

antiparallel microtubule overlaps by PRC1, which autonomously

discriminates between parallel and antiparallel microtubules with

high selectivity. Xklp1 ensures the maintenance of microtubule

overlap size by selectively inhibiting the growth of overlapping

microtubules in a manner that is dependent on overlap length.

Our results demonstrate that only two proteins are necessary

and sufficient for the selective formation and for adaptable length

control of antiparallel microtubule overlaps in vitro, as required

for the formation of the midzone in vivo.

RESULTS

PRC1 Selectively Binds to Antiparallel MicrotubuleOverlaps with Nanomolar AffinityWe developed a fluorescence microscopy assay to visualize

dynamic microtubule encounters in vitro. We attached short,

Alexa 568- and biotin-labeled, stabilized microtubule seeds

under flow to glass slides, which were functionalized with a

NeutrAvidin-biotin-polyethylene glycol layer. The majority of

the seeds aligned, and microtubules growing from these seeds

in the presence of Alexa 568–tubulin occasionally encountered

each other in a plus end-to-plus end configuration (Figure 1A).

Multicolor time-lapse TIRF microscopy revealed that antiparallel

microtubules connected in the presence of 5 nM purified PRC1–

Alexa 647 as soon as their plus ends encountered, forming an

antiparallel microtubule overlap (Figure 1B, Movie S1 available

online). Interestingly, PRC1–Alexa 647 strongly accumulated in

the growing antiparallel overlap region under conditions at which

it did not localize significantly to individual microtubules or to

parallel microtubule pairs (Figure 1C, Movie S1). This demon-

strates that vertebrate PRC1 does not require an additional

factor to selectively target antiparallel microtubule configura-

tions. Autonomous binding to antiparallel microtubules, there-

fore, appears to be conserved between PRC1 orthologs from

vertebrates and yeast (Janson et al., 2007).

The length of antiparallel microtubule overlaps increased

linearly with time after the microtubule encounter (Figure 1D,

top) indicating constant average overlap growth velocity. Over-

lap length as measured from plus end to plus end of the two

antiparallel microtubules increased approximately twice as fast

(4.9 ± 0.9 mm/min, mean ± standard deviation [SD]) as the length

of single microtubules that were not part of an antiparallel micro-

tubule pair (2.3 ± 0.3 mm/min) (Figure 1D, bottom), demonstrating

that the growth velocity of two microtubules of an antiparallel

overlap is similar to that of individual microtubules and therefore

not significantly affected by the strong accumulation of PRC1.

To investigate the mechanism by which vertebrate PRC1

specifically recognizes antiparallel overlaps, we quantified its

affinity and selectivity of binding to distinct microtubule configu-

rations. We measured the average fluorescence intensity of

PRC1–Alexa 647 in antiparallel overlaps as a function of the

PRC1–Alexa 647 concentration (Figure 1E, Figure S1A). PRC1

binding to antiparallel overlaps was strong and positively coop-

erative with a microscopic dissociation constant of 7.6 ± 0.4 nM

(mean ± standard error of the mean [SEM]) and a Hill coefficient

of 1.9 ± 0.1 (Figure 1E). Under standard imaging conditions, we

did not detect any accumulation of PRC1 on individual microtu-

bules or between parallel microtubules even at the highest PRC1

concentration used (31.5 nM) (Figure S1B). We estimate that

PRC1 binds at least 28 times more strongly to antiparallel

microtubule pairs than to individual or parallel microtubules

(Extended Experimental Procedures), indicating considerably

higher selectivity for antiparallel microtubules than previously

reported for Ase1 from fission yeast (Janson et al., 2007).

To investigate the mechanism of this high binding selectivity,

we imaged single PRC1 molecules at higher temporal resolution.

Many binding events of individual PRC1–Alexa 647 molecules

were detected in antiparallel overlaps (Figure 1F, right), whereas

hardly any events were observed in parallel microtubule pairs

(Figure 1F, left). Individual PRC1 molecules were not statically

bound but diffused in antiparallel overlaps (Figures S1C and

S1D). Despite the high affinity for antiparallel microtubules,

turnover of PRC1 in antiparallel overlaps was fast with a single-

molecule half-life of around 1 s (Figure 1G, Figure S1E).

Therefore, selective binding to antiparallel microtubules is an

intrinsic property of the individual PRC1 dimer and it is very

dynamic. This appears to distinguish PRC1 from Ase1 whose

moderate selectivity for antiparallel microtubule binding was

suggested to be a consequence of the formation of slowly

dissociating oligomers (Kapitein et al., 2008).

Reconstitution of a Minimal Midzone with StableAntiparallel Microtubule Overlaps In VitroWe next tested the additional effect of Xenopus kinesin-4 Xklp1

on microtubule overlap formation. In the presence of both 5 nM

purified PRC1–Alexa 647 and 15 nM purified Xklp1–GFP (Fig-

ure S2), antiparallel microtubules connected to each other as

soon as their plus ends encountered, forming an antiparallel

microtubule overlap that remarkably reached a constant size

after awhile (Figure 2A, Movie S2, and Movie S3). PRC1–Alexa

647 and also Xklp1–GFP were observed to strongly accumulate

selectively in antiparallel but not parallel overlap regions (Fig-

ure 2B, Movie S2). Xklp1 had only a minor positive effect on

the amount of overlap-associated PRC1 (Figure 2C). In contrast,

no Xklp1 was observed in the absence of PRC1 in antiparallel

microtubule overlaps (Figures 2D–2F). Microtubules often failed

to form a tight bundle in the presence of Xklp1 only (Movie S4),

a behavior never observed when PRC1 was present.

These TIRF microscopy data indicate that Xklp1 interacts

directly with PRC1, which we confirmed by pull-down experi-

ments (Figure 2G, lane 2). Consequently, an N-terminal motor


fragment, Xklp506–GFP, that does not interact with PRC1 (Fig-

ure 2G, lane 3) failed to accumulate in antiparallel microtubule

overlaps despite the presence of PRC1–Alexa 647, as observed

by TIRF microscopy (Figure 2H, right). This shows that a direct

interaction between Xklp1 and PRC1 is required for PRC1-depen-

dent recruitment of Xklp1 to antiparallel microtubule overlaps.

The most striking observation in the presence of both midzone

proteins was, however, that the growth of microtubule plus ends

PRC1microtubules

0:23

0:00

0:47

1:10

1:33

1:57

evanescent field

Alexa568-Tubulinmicrotubule seed

bundling protein/complex

anti-parallel overlap

PRC1 microtubulesschemeC

A

0 1 2 3time [min]

2

0

4

6g

row

th v

elo

city

[µm

/min

]

B

D

F

microtubule imagebefore time-lapse

kymograph ofPRC1 time-lapse

microtubule imageafter time-lapse

scheme

parallel overlap antiparallel overlap

0 2 4 6 8 10 1210-2

10-1

100

dwell time [s]

1 −

CD

F

G

E

0 1 2 3time [min]

5

0

10

15

len

gth

[µm

]

free microtubulesantiparallel overlap

PRC1microtubules

PRC1 (single molecules)microtubules

0 10 20 30PRC1 concentration [nM]

fluo

r. si

gn

al in

AP

over

lap

[x10

4 au

]

0

1

2

Figure 1. PRC1 Preferentially Binds to and

Diffuses in Antiparallel Microtubule Over-

laps

(A) Scheme of the experimental setup. Dynamic

microtubules were grown in the presence of free

Alexa 568-labeled tubulin (light blue) and fluores-

cently labeled midzone proteins (red) from stabi-

lized microtubule seeds (dark blue) attached to a

PEG-passivated glass surface by means of

biotin–NeutrAvidin (yellow and green) links. Micro-

tubules occasionally encountering each other in a

plus end-to-plus end configuration forming an

antiparallel microtubule overlap and fluorescently

labeled midzone proteins were observed by multi-

color time-lapse TIRF microscopy.

(B) Time sequence of overlaid TIRF microscopy

images of PRC1–Alexa 647 (red, 5 nM) and a

dynamic Alexa 568–microtubule pair (blue) form-

ing an antiparallel overlap taken at the indicated

times in min:s. The time-lapse was recorded at 1

frame per 4.67 s. Scale bar, 10 mm.

(C) TIRF images showing several microtubules

growing from immobilized microtubule seeds, re-

corded 4–9 min after start of the experiment

(top), and a kymograph (time-space plot) of a

selected microtubule pair (bottom) in the presence

of PRC1–Alexa 647. The kymograph begins

�1 min after start of the experiment. Left: sche-

matic illustration of the microtubule configurations

visible in the image and kymograph: microtubule

segments (gray, plus ends labeled with ‘‘+’’)

growing from seeds (white) form antiparallel (red)

or parallel (green) overlaps. Parallel overlaps

form from parallel seed bundles. Right: Overlaid

and single-channel TIRF microscopy images and

kymographs for PRC1–Alexa 647 and Alexa 568

microtubules as indicated. Concentrations and

frame rate as in (B). Horizontal scale bars, 10 mm.

Vertical scale bar, 1 min.

(D) Average length (top) and average instantaneous

growth velocity (bottom) of antiparallel microtubule

overlaps (red, n = 7) and of individual microtubules

not part of a pair from the same experiment (blue,

n = 14) forming in the presence of PRC1–Alexa

647 as a function of time. For antiparallel microtu-

bule overlaps t = 0 is the moment of the plus end-

to-plus end encounter. Error bars are standard

deviation (SD). Concentrations are as in (B).

(E) Average fluorescence signal of PRC1–Alexa

647 bound to antiparallel microtubule overlaps

plotted against the total PRC1–Alexa 647 concentration. Data (black points) were fitted either using the Hill equation (black, straight line) or a hyperbolic function

(red, dotted line). At least a total of 100 mm of antiparallel overlap were evaluated per condition. Error bars are standard error of the mean (SEM).

(F) Single-molecule imaging of PRC1–Alexa 647 in a parallel (left) and an antiparallel (right) microtubule pair. Top: schematic representation of the pairs. Below:

kymographs, positioned between TIRF microscopy images of microtubule overlaps before (top) and after (bottom) fast recording of individual PRC1–Alexa 647

molecules, showing the behavior of individual PRC1–Alexa 647 molecules (green). The PRC1–Alexa 647 concentration was 50 pM. Frame rate was 1 frame per

100 ms. Horizontal scale bar, 10 mm. Vertical scale bar, 5 s.

(G) Dwell time distribution plotted as 1 � cumulative distribution function (CDF) of individual PRC1–Alexa 647 molecules in antiparallel overlaps (black, n = 687).

The red dashed curve is a biexponential fit with characteristic times t1 = 0.91 ± 0.02 s and t2 = 4.47 ± 0.24 s (mean ± 95% confidence interval [CI]), and relative

amplitudes 87% t1 and 13% t2. The black dashed curve is a monoexponential fit. The mean bleaching time for Alexa 647 was 32.0 ± 4.5 s. Consequently t1 is

practically unaffected by bleaching and the corrected t2 is 5.2 s.

See also Figure S1 and Movie S1.


being part of an antiparallel overlap slowed down until it ceased

completely in all observed cases (n > 100) (Figure 2A, kymograph

in 2B, 2H, left, Movie S2, Movie S3). The growth of overlap

microtubules gradually stopped within about 1 min after the

end-to-end encounter (Figure 2I, green). In contrast, individual

microtubules not engaged in overlaps in the same experiment

continued growing with constant velocity of 2.1 ± 0.3 (mean ±

SD) mm/min (Figure 2I, blue). Inhibition of antiparallel microtubule

overlap growth was a consequence of the simultaneous pres-

ence of PRC1 and full-length Xklp1. The presence of neither

PRC1 alone (Figures 1B–1D) nor Xklp1 alone (Figure 2F) had a

significant effect on microtubule growth in antiparallel overlaps.

Hence, stable antiparallel overlaps with a constant size of several

micrometers formed only in the presence of both midzone pro-

teins (Figures 2A, 2B, and 2I; Movie S3). Such antiparallel micro-

tubule overlaps maintained a constant size over long periods of

time despite the complete inhibition of microtubule growth. This

is remarkable, given that slowdown of microtubule growth

usually results in frequent transitions to microtubule depolymer-

ization (catastrophes) (Arnal et al., 2004; Janson et al., 2003;

Walker et al., 1988).

Our results show that a ‘‘minimal midzone’’ can be reconsti-

tuted in vitro with dynamic microtubules and two proteins only:

PRC1 is necessary and sufficient to recognize and bundle

antiparallel microtubules. Xklp1 alone does not bind to antipar-

allel overlaps but is instead recruited by overlap-associated

PRC1. This recruitment results in local inhibition of microtubule

growth without causing microtubule depolymerization, leading

to antiparallel microtubule overlaps of constant size.

Xklp1 Inhibits Turnover of Tubulin at the GrowingMicrotubule EndTo better understand how Xklp1 affects microtubule dynamics,

we measured the effect of varying concentrations of Xklp1

alone on the dynamics of individual microtubules. Similar to an

N-terminal fragment of Xklp1 (Bringmann et al., 2004), full-length

Xklp1 alone reduced the growth velocity of microtubule plus

ends in a concentration-dependent manner (Figures 3A and 3B).

This effect could only be observed at Xklp1 concentrations of

several hundred nanomolar, well above the Xklp1 concentra-

tions in the lower nanomolar range that lead to inhibition of

microtubule growth in antiparallel overlaps in the presence of

PRC1. Interestingly, the catastrophe frequency remained very

low at increasing Xklp1 concentrations (Figure 3C), despite the

strong reduction in microtubule growth velocity, which is ex-

pected to lead to more frequent catastrophes (Arnal et al.,

2004; Janson et al., 2003; Walker et al., 1988). The effect of

Xklp1 on the catastrophe frequency is more evident when

comparing microtubules growing with similar velocities (Fig-

ure 3D). The catastrophe frequencies of microtubules in the

presence of Xklp1, plotted as a function of their growth velocities

(Figure 3D, red), are significantly lower than those of microtu-

bules growing at similar velocities in the absence of Xklp1 at

reduced tubulin concentrations (Figure 3D, blue). This shows

that Xklp1 inhibits catastrophes, explaining why local inhibition

of microtubule growth by PRC1-recruited Xklp1 in antiparallel

microtubule pairs (Figures 2A, 2B, and 2I) does not result in

destabilization of the overlap.

To gain a better understanding of the effect of Xklp1 on the

microtubule catastrophe frequency, we measured the microtu-

bule growth velocity at different tubulin concentrations with

and without 300 nM Xklp1 (Figure 3E). This allowed us to deter-

mine the association and dissociation rate constants of guano-

sine-50-triphosphate (GTP)-tubulin at the growing plus end

(Drechsel et al., 1992; Walker et al., 1988). We found that

300 nM Xklp1 reduced the on-rate of GTP-tubulin by a factor

of about two and the off-rate by a factor of around three (Fig-

ure 3F). This shows that Xklp1 has the unique property of making

microtubules less dynamic by reducing the overall turnover of

GTP-tubulin at growing microtubule ends, distinguishing it

from other known mitotic regulators of microtubule dynamics

(Brouhard et al., 2008; Helenius et al., 2006; Varga et al., 2006).

Dynamic Control of Antiparallel Microtubule OverlapLengthTo investigate how the inhibitory effect of Xklp1 on tubulin turn-

over affects the length of antiparallel microtubule overlaps in the

presence of PRC1, we varied the Xklp1:PRC1 ratio in antiparallel

microtubule encounter experiments, leaving the PRC1 concen-

tration constant (Figure 4A, Movie S5). At Xklp1:PRC1 ratios lower

than 1, microtubule plus-end growth persisted over long times

(>10 min) after formation of an antiparallel overlap, albeit at greatly

reduced speeds as compared to microtubules that were not part

of an overlap (Movie S5). At Xklp1:PRC1 ratios R1, microtubule

plus ends within the antiparallel overlap stopped growing soon

after their initial encounter leading to stable antiparallel overlaps

with a distinct characteristic steady-state length for each

Xklp1:PRC1 ratio (Figure 4B). The average steady-state overlap

length decreased with increasing Xklp1 concentration, ranging

from 4.2 mm to 1.2 mm for ratios from 1 to 20, respectively (Fig-

ure 4B). Thus the concentration of Xklp1 determines the steady-

state length of the antiparallel microtubule overlap.

Next, we investigated whether the steady-state overlap

lengths could reversibly respond to changes in the Xklp1 con-

centration (Figure 4C). We preformed short overlaps at high

Xklp1:PRC1 ratios, followed by reduction of the Xklp1 concen-

tration, keeping the PRC1 and tubulin concentrations constant.

Strikingly, we observed that microtubules of short preformed

overlaps started to grow again until they had reached the length

characteristic for the new Xklp1:PRC1 ratio (Figures 4C–4E,

Movie S6). These results demonstrate that the Xklp1:PRC1 ratio

can dynamically control the steady-state overlap length.

Why do higher Xklp1:PRC1 ratios result in shorter antiparallel

overlaps? The major contribution of Xklp1 in determining overlap

length could be either its inhibitory effect on microtubule growth

or, alternatively, a consequence of Xklp1 sliding apart antiparallel

microtubules, similar to kinesin-5 (Kapitein et al., 2005). Indeed,

we observed that microtubules engaged in antiparallel overlaps

frequently buckled over time (Figure S3A), indicative of plus-

end-directed microtubule sliding. The average sliding length

(Extended Experimental Procedures) was, however, in all cases

shortascompared to theoverlap lengthatsteadystate (Figure4B).

A plot of the individual sliding lengths of all microtubule pairs

against their individual overlap lengthalso did not revealanycorre-

lation (Figure S3B). Therefore, antiparallel microtubule sliding

does not significantly contribute here to overlap length control.


A

Xklp1PRC1 microtubulesscheme

Xklp1 + PRC1microtubules

0:23

0:00

0:47

1:10

1:33

1:57

Xklp1microtubules

0:23

0:00

0:47

1:10

1:33

1:57

Xklp1 microtubulesscheme

0 1 2 3time [min]

2.5

0

5

7.5

leng

th [µ

m]

5

0

10

15

20

PRC1

sig

nal i

n A

P ov

erla

p[x

103 au

]

+ Xklp1- Xklp1

250 kDa

150 kDa

100 kDa

75 kDa

50 kDa

37 kDa

Xklp1-GFP

PRC1-SNAPXklp1506-GFP

tubulin

M M M

InputBeads

1 2 3 1 2 3

2

0

4

6

8

Xklp

1 si

gnal

in A

P ov

erla

p[x

103 au

]

- PRC1+PRC1

B

IC

D

F

E

H

G

0 1 2 3time [min]

2

0

4

6

grow

th v

eloc

ity [µ

m/m

in]

antiparallel overlapfree microtubules


Xklp1microtubules


Xklp1506+ PRC1microtubules

Figure 2. A Minimal Protein System Forms Stable Antiparallel Overlaps In Vitro

(A) Time sequence of overlaid TIRF microscopy images of PRC1–Alexa 647 (red, 5 nM) and Xklp1–GFP (green, 15 nM) in a dynamic Alexa 568-labeled microtubule

pair (blue) forming an antiparallel overlap at the indicated times in min:s.

(B) TIRF images, recorded 4–9 min after the start of the experiment, showing several microtubules growing from immobilized microtubule seeds (top) and a kymo-

graph of a selected microtubule pair (bottom). Left: schematic illustration of the microtubule configurations. Color coding as in Figure 1C. Right: overlaid and

single-channel TIRF microscopy images and kymographs for PRC1–Alexa 647, Xklp1–GFP, and Alexa 568 microtubules as indicated. Concentrations as in (A).

(C) Average fluorescence intensity signal of PRC1–Alexa 647 bound to antiparallel microtubule overlaps in the presence (n = 45 overlaps) or absence (n = 18

overlaps) of Xklp1–GFP, obtained from intensity line scans. Error bars are SEM.

(D) Time sequence of overlaid TIRF images of 15 nM Xklp1–GFP (green) and a dynamic Alexa 568-microtubule pair (blue) forming an antiparallel overlap taken at

the indicated times in min:s.


Xklp1 Is a Processive MotorTo elucidate the mechanism by which Xklp1 sets the size of

antiparallel microtubule overlaps, we inspected the behavior

of the motor more closely. Occasionally, antiparallel overlaps

formed by lateral encounters between one microtubule end

and one microtubule segment distant from the growing end

(Figure 5A). In such cases, Xklp1–GFP loaded onto the overlap

region by PRC1 also accumulated at the plus end of the

microtubule extending beyond the overlap (Figure 5A). Inter-

estingly, accumulation of Xklp1 at these free microtubule ends

also resulted in inhibition of microtubule growth (Movie S7).

This suggests that (1) it is the plus-end-associated fraction

of Xklp1 that inhibits microtubule growth and (2) Xklp1 can

target the plus end by means of processive motility over con-

siderable distances. This is surprising, given that an N-terminal

fragment of Xklp1 has been proposed to be at most weakly

processive based on enzymatic activity assays (Bringmann

et al., 2004).

0 nM Xklp1 300 nM Xklp1 600 nM Xklp1

A

B D

- Xklp1

+ 300 nM Xklp1

kon

[µM-1 s-1]

4.03 ± 0.04[3.951 ; 4.101]

1.86 ± 0.02[1.820 ; 1.906]

koff

[ s-1]

9.60 ± 0.85[8.626 ; 10.571]

3.08 ± 0.49[2.125 ; 4.027]

0 200 400 600Xklp1 concentration [nM]

gro

wth

vel

oci

ty [µ

m/m

in]

0

1

2

3

0 200 400 600Xklp1 concentration [nM]

cata

stro

ph

e fr

equ

ency

[10-2

eve

nts

/min

]

0

1

2

3

C

E F

cata

stro

ph

e fr

equ

ency

[10-2

eve

nts

/min

]

0

5

10

growth velocity [µm/min]0 1 2 3

-Xklp1+Xklp1

0 5 10 15 20 25 30 35

050

100

150

tubulin concentration [µM]

gro

wth

rate

[dim

ers/

s]

-Xklp1+Xklp1

0 8

-10

010

4

20

Figure 3. Xklp1 Inhibits Turnover of Tubulin at the

Growing Microtubule End

(A) Representative kymographs of Alexa 568-labeled

microtubules growing in the absence or presence of

Xklp1–GFP (not shown) at the indicated concentrations.

Concentration of soluble tubulin is 17.5 mM. The frame

rate was 1 frame per 3 s. Horizontal scale bar, 10 mm.

Vertical scale bar, 1 min.

(B) Average microtubule growth velocity and (C) catas-

trophe frequency plotted against the total Xklp1–GFP

concentration. Sixty microtubules from two independent

experiments were analyzed per condition.

(D) Catastrophe frequency as a function of microtubule

growth velocity. Red data points are from experiments

with 17.5 mM tubulin and varying Xklp1–GFP concentra-

tions, as shown in (B) and (C). Blue data points are from

experiments with varying tubulin concentrations (7, 12,

and 17.5 mM) in the absence of Xklp1–GFP.

(E) Average growth rate in the absence (blue circles) or in

the presence of 300 nM Xklp1–GFP (red circles) as a func-

tion of total tubulin concentration. A multiple linear regres-

sion model with a categorical factor g (+Xklp1: g = 0,

�Xklp1: g = 1) revealed that both the abscissa and the

slope were significantly different (p < 0.001, see Experi-

mental Procedures). The resulting regression lines repre-

sent vg = kon c – koff with the net assembly rate vg in tubulin

dimers per second and the time constants for association

kon in s�1mM�1 and dissociation koff in s�1 of GTP-tubulin

at the growing microtubule end. Inset: regression lines

and their 95% confidence intervals (dashed) at the origin,

showing that not only kon (slope) but also koff (intersection

with abscissa) is different for the two conditions. 185 to

210 microtubules from two independent experiments were

analyzed for each tubulin concentration per condition.

(F) Values for kon and koff (mean ± SEM, with 95% confi-

dence in brackets) as obtained from fits in (E).

Error bars in (B)–(E) are SD.

(E) Average fluorescence intensity signal of Xklp1–GFP bound to the antiparallel microtubule overlap in the presence (n = 42 overlaps) or absence of PRC1–Alexa

647, as obtained from intensity line scans. Error bars are SEM.

(F) Kymograph of the microtubule pair shown in (D). Color code as in (B).

(G) Pull-down of Xklp1 by PRC1–SNAP immobilized on beads. Coomassie-stained SDS-gel shows fractions bound to the beads after incubation with different

input solutions. Lanes 1: negative control with beads lacking PRC1–SNAP incubated with a mixture of 2 mM full-length Xklp1–GFP, 2 mM truncated Xklp1506–GFP,

and 5 mM soluble tubulin. Only weak nonspecific binding of the two Xklp1 proteins is seen. Lanes 2 and 3: beads with immobilized PRC1–SNAP incubated with

either 2 mM full-length Xklp1–GFP (lanes 2) or truncated 2 mM Xklp1506–GFP (lanes 3) in the presence of 5 mM soluble tubulin.

(H) Representative kymographs of antiparallel microtubule overlaps (blue) in the presence of PRC1–Alexa 647 (red, 5 nM) and either full-length Xklp1–GFP

(green, 15 nM) (left) or truncated Xklp1506–GFP (green, 15 nM) (right).

(I) Average length (left) and average instantaneous growth velocity (right) of antiparallel overlaps (green, n = 9) and individual microtubules that are not part of a pair

from the same experiment (blue, n = 18) formed in the presence of PRC1 and full-length Xklp1 as a function of time (conditions as in A and B). For comparison, the

length of microtubule overlaps formed in the presence of only PRC1 (from Figure 1D) is shown (red dashed line). t = 0 is the moment of plus end-to-plus end

encounter. Error bars are SD. Concentrations are as in (A). Horizontal scale bars are 10 mm. Vertical scale bars are 1 min. The frame rate for all time-lapse movies

was 1 frame per 4.67 s. Kymographs start �1 min after the start of the experiment.

See also Figure S2 and Movie S2, Movie S3, and Movie S4.


We therefore imaged single molecules of full-length Xklp1–GFP

on stabilized, individual microtubules in the absence of PRC1. We

observed that at low ionic strength Xklp1–GFP dimers indeed

moved processively (Figure 5B) with an average run length of 1.2

mm (Figure 5C) and an average velocity of 0.8 mm/s (Figure 5F,

red). Reduction in ionic strength was necessary because binding

of Xklp1 to individual microtubules was very weak. Processivity

was an intrinsic property of the dimerized motor domain, as

demonstrated by imaging a GFP-labeled motor fragment (Figures

5D–5G). Run length (0.9 mm, Figure 5E) and velocity (0.9 mm/s,

Figure 5F, black) of Xklp1506–GFP were in a similar range as the

values for the full-length motor. The longer dwell times of the

full-length motor (Figure S4A) are probably a consequence of

a secondary microtubule-binding site (Figure S4B). These results

demonstrate that Xklp1 is a processive motor, suggesting that it

might measure the overlap length by processive movement.

Xklp1 Determines Overlap Length by a ProcessiveOverlap Exploration MechanismTo test the hypothetical mechanism of processive overlap

length measurement, we imaged single full-length Xklp1–GFP

molecules (500 pM) in the presence of excess PRC1–Alexa 647

(5 nM) in dynamic antiparallel microtubule pairs at high time reso-

lution. Strikingly, Xklp1–GFP was very mobile in PRC1-containing

antiparallel overlaps. Xklp1–GFP showed processive movement

often abruptly changing direction, indicating a transition from one

to the other microtubule of the pair (Figure 6A, left, Figure S5A,

Movie S8). This highly mobile behavior of the motor was very

different from that of single PRC1–Alexa 647 molecules (5 pM)

in the presence of excess PRC1–Alexa 488 (5 nM) in the overlap,

which only diffused for very short distances (Figure 6A, right). The

dwell time distribution of Xklp1–GFP in PRC1-containing antipar-

allel overlaps was roughly exponential (Figure 6B, black dots)

with a calculated average dwell time of �5.9 s (dashed red line).

This demonstrates that Xklp1 dwells longer in a PRC1-decorated

antiparallel overlap as compared to on individual microtubules in

the absence of PRC1 at low ionic strength (Figure 6B, gray

circles), and also longer than the individual PRC1 molecules in

the antiparallel overlap (Figure 6B, blue dots). This shows that

PRC1 increases the residence time of Xklp1 in the overlap prob-

ably by offering additional binding sites. Mean squared displace-

ment (MSD) analysis confirmed that Xklp1 was very mobile in

PRC1-containing overlaps (Figure 6C, black dots), in contrast

to slowly diffusing PRC1 molecules (Figure 6C, blue dots;

Figure S1C). However, Xklp1 explored shorter distances in

PRC1-containing overlaps than on individual microtubules in

the absence of PRC1 under low ionic strength conditions

(Figure 6C, gray dots). The MSD curveof Xklp1 in antiparallel over-

laps could be well described using a ‘‘persistent random walk’’

model (Figure 6C, dashed red line, Extended Experimental Proce-

dures) (Othmer et al., 1988;Tranquillo and Lauffenburger, 1987). A

fit to the data yielded a characteristic persistence time between

A

Xklp1:PRC1 ratio

leng

th [µ

m]

1 2 3 5 10 200

2

4

overlap lengthsliding length

Xklp1 PRC1 microtubles

20 : 1 Xklp1 : PRC1

5 : 1Xklp1 : PRC1

1 : 1Xklp1 : PRC1

time [min]0 3 6 9

over

lap

leng

th [µ

m]

0

2

4

dilu

tion

imag

ing

SS 1 SS 2overlap growth DXklp1 PRC1 microtubles

00:00

01:22

02:44

03:06

04:28

06:00

Xklp1PRC1merge

B

C

E

Figure 4. Dynamic Control of Antiparallel Overlap

Length

(A) Representative triple-channel TIRF microscopy images

of PRC1–Alexa 647 (red, 5 nM) and Xklp1–GFP (green,

100, 25, and 5 nM) at PRC1:Xklp1 ratios as indicated in

a dynamic Alexa 568–microtubule pair (blue) forming an

antiparallel overlap. Images were recorded 7–12 min after

start of the experiment. Scale bar, 10 mm.

(B) Average overlap length (red) or average sliding length

(blue) as a function of the Xklp1:PRC1 ratio. Error bars

are SD. Overlap lengths and sliding lengths were

measured for at least 15 overlaps from at least three

experiments per condition after steady state had been

reached. Yellow shading indicates the regime where no

steady-state length was established.

(C) Reversibility experiment: Antiparallel microtubule over-

laps were allowed to form at 100 nM Xklp1–GFP and 5 nM

PRC1–Alexa 647 (Xklp1:PRC1 ratio 20:1) and to reach

their steady-state length (black dot with SD bars in red

area). Subsequently, Xklp1–GFP was diluted to 5 nM

keeping the PRC1–Alexa 647 and tubulin concentrations

constant (new Xklp1:PRC1 ratio 1:1, details in Experi-

mental Procedures). Quantification of overlap lengths

versus time for three examples (dark green, blue, and

red dots in green area) shows that after �6 min the over-

laps reach a new steady-state length characteristic for

the new, lower Xklp1 concentration (purple area).

(D) Example time series of overlaid triple-channel TIRF

microscopy images at the indicated times (min:s) after

Xklp1–GFP dilution and (E) corresponding kymographs

showing a representative growing overlap. The frame rate

was 1 frame per 4.67 s. Colors and horizontal scale bar as

in (A). Vertical scale bar, 1 min. See also Figure S3 and

Movie S5 and Movie S6.


directional switches of �2.3 s and a velocity of �0.5 mm/s for

Xklp1 in the overlap. For time intervals larger than the persistence

time, the movement of Xklp1 was effectively diffusive.

To estimate which distances Xklp1 can explore in PRC1-

decorated antiparallel microtubule overlaps, we calculated the

distribution of average exploratory distances (Figure 6D, black

dots, Extended Experimental Procedures). This analysis reveals

that Xklp1 can explore average lengths of up to �6 mm. This is

in strong contrast to the low exploratory distances of PRC1

(Figure 6D, blue dots), which can only move for distances of up

to �0.2 mm. Strikingly, the average exploratory distances of

individual Xklp1 molecules coincide well with the range of

steady-state lengths of antiparallel microtubule overlaps (Fig-

ure 4B). This suggests that, by processive movement, Xklp1 rea-

ches the plus ends of antiparallel overlapping microtubules,

where it can modulate microtubule dynamics. To test whether

the processive nature of Xklp1’s motility is indeed central to the

mechanism of steady-state length establishment, we replaced

adenosine-50-triphosphate (ATP) by adenosine-50-diphosphate

(ADP) in the TIRF microscopy experiment, thereby preventing

processive motility of Xklp1 but not its recruitment by PRC1. We

observed that under these conditions the two midzone proteins

failed to produce stable steady-state overlap lengths (Figure S6).

Instead, overlap microtubules continued growing despite the

presence of PRC1 and Xklp1 in the overlap. We conclude that

processive overlap exploration allows Xklp1 to reach and modu-

late the behavior of microtubule plus ends and that its finite

average exploratory range limits the maximum size of the

steady-state antiparallel microtubule overlap length to 5–6 mm

(Figure 4B, Figure 6D). This defines a processive exploration

mechanism for antiparallel microtubule overlap length control.

DISCUSSION

The Mechanism of Antiparallel Overlap Generationand Length Control by PRC1/Kinesin-4 ExplainsAnaphase Spindle Characteristics In VivoWe have elucidated a molecular mechanism capable of selective

formation and adaptable length control of antiparallel microtubule

overlaps by two conserved midzone proteins. Central to this

mechanism is the combination of four distinct molecular activities

(Figure 7): (1) bundling of microtubules with high selectivity for

antiparallel orientation by PRC1, (2) recruitment of kinesin-4 by

PRC1 to antiparallel overlaps, and (3) processive microtubule

plus-end-directed motility together with (4) an inhibitory effect

on microtubule growth by the motor protein kinesin-4. The combi-

nation of these activities constitutes a minimal two-component

system capable of formation of a minimal midzone in vitro.

Our in vitro reconstitution explains why loss of either PRC1 or

kinesin-4 leads to nonequivalent defects in midzone formation

in vivo (Kurasawa et al., 2004; Mollinari et al., 2002; Zhu and

Jiang, 2005). We show that selective antiparallel microtubule

crosslinking is an intrinsic property of vertebrate PRC1, an

activity which appears to be evolutionarily conserved (Janson

A

B D F

Xklp1 PRC1 microtubles

+

+

scheme

00:00 00:47 01:33 02:20 03:07

Xklp1 microtubles

GC

Xklp1506 microtubles

E

0 5 10 15 20run length [µm]

0

100

200

300

400

cou

nts

L = 1.20 µm[1.122 ; 1.273]

Xklp1

200 5 10 15run length [µm]

0

100

200

300

cou

nts

L = 0.91 µm[0.860 ; 0.967]

Xklp1506

intensity [102 au]

0

100

200

300

cou

nts

Xklp1506Xklp1

dKin401

2 10 2064 8 12 14 16 18

0 1 2 3 4 5 60

5

10

15

20

25

30

35

time [s]

MSD

[µm

2 ]

Xklp1fit, v = 0.82 µm/sXklp1506fit, v = 0.93 µm/s

Figure 5. Xklp1 Is a Processive Motor

(A) Time sequence (min:s) of multichannel TIRF micros-

copy images of PRC1–Alexa 647 (red, 5 nM) and Xklp1–

GFP (green, 15 nM) in a dynamic Alexa 568-labeled

microtubule pair (blue) forming an antiparallel overlap by

a lateral encounter between one microtubule plus end

and one microtubule segment distant from its growing

end, as illustrated in the scheme (right). Scale bar, 5 mm.

(B) Representative kymograph of individual full-length

Xklp1–GFP dimers (green, 2.5 nM) on an individual stabi-

lized microtubule (red) at low ionic strength (see Extended

Experimental Procedures) imaged at high time resolution

(10 frames/s). Horizontal scale bars, 10 mm. Vertical

scale bar, 5 s.

(C) Histogram of the run length distribution of full-length

Xklp1–GFP with a monoexponential fit (red dashed);

mean run length was 1.20 mm (95% CI in brackets).

(D) Representative kymograph of individual Xklp1506–GFP

dimers (green, 20 nM) on an individual taxol-stabilized

microtubule (red) at very low ionic strength conditions

(see Extended Experimental Procedures). Imaging and

bars as in (B).

(E) Histogram of the run length distribution of Xklp1506–

GFP with a monoexponential fit (red dashed); mean run

length was 0.91 mm (95% CI in brackets).

(F) Mean squared displacement (MSD) curve of full-length

Xklp1–GFP (red) and truncated Xklp1506–GFP (black) and

parabolic fits (dashed lines, see Experimental Proce-

dures). Error bars are SEM.

(G) Histograms of the initial brightness of Xklp1–GFP,

Xklp1506–GFP, and dimeric Kin401–GFP as a control (Telley

et al., 2009), indicating that both fragment and full-length

Xklp1 are dimeric.

See also Figure S4 and Movie S7.


et al., 2007). Autonomous bundling of antiparallel microtubules

places PRC1 at the core of the activities essential for midzone

formation and explains why PRC1 is absolutely required for the

bundling and stabilization of antiparallel microtubule overlaps,

as indicated by the complete loss of central anaphasic microtu-

bule bundles in the absence of PRC1 in vivo (Kurasawa et al.,

2004; Mollinari et al., 2002; Zhu and Jiang, 2005).

Vertebrate PRC1 has previously been shown to interact with

kinesin-4 during anaphase, an interaction necessary for proper

midzone organization (Kurasawa et al., 2004; Zhu and Jiang,

2005). Here, we establish a hierarchy of interactions: full-length

microtubule/PRC1 imagebefore time-lapse

kymograph of single-molecule time-lapse

scheme

microtubule/PRC1 image after time-lapse

Xklp1 (single molecules)microtubules

PRC1 (bulk protein in overlap)

Xklp1 in overlapA

PRC1 (single molecules)microtubules

PRC1 (bulk protein in overlap)

PRC1 in overlap

0 5 10 15 20 25 3010-2

10-1

100

dwell time [s]

1−C

DF

Xklp1 in overlapXklp1 on single MTPRC1 in overlap

0 2 4 6 8 10 12 14 160

5

10

15

20

25

30

Xklp1 on single MTXklp1 in overlap

PRC1 in overlap

time [s]

MSD

[µm

2 ]

0 2 4 6 8 100

0.2

0.4

0.6

0.8

1.0

avg. exploratory distance [µm]

calc

ula

ted

pro

bab

ility

Xklp1 in overlapXklp1 on single MTPRC1 in overlap

0 2 4 6 8 1010-2

10-1

100

B C

D

Figure 6. Xklp1 Determines Antiparallel Microtu-

bule Overlap Length by Processive Overlap

Exploration

(A) Single-molecule imaging of either individual Xklp1–

GFP molecules (500 pM, green) in a PRC1–Alexa 647

(5 nM, blue) containing antiparallel microtubule overlap

(red) (right panel) or individual PRC1–Alexa 647 molecules

(5 pM, green) in a PRC1–Alexa 488 (5 nM, blue) containing

antiparallel microtubule overlap (red) (left panel). Scheme

(top) and triple-channel TIRF microscopy images of the

microtubule pair before (upper image) and after (lower

image) fast time-lapse imaging of individual Xklp1–GFP

or PRC1–Alexa 647 molecules, as illustrated in a kymo-

graph (middle). The frame rate was 10 s�1. Horizontal

scale bar, 10 mm. Vertical scale bar, 5 s.

(B) Dwell time distribution (shown as 1 – CDF) of single

Xklp1–GFP molecules moving in antiparallel microtubule

overlap regions (conditions as in A, right; black dots, n =

226) and a monoexponential fit (dashed red); mean dwell

time is 5.85 ± 0.12 s (mean ± 95% CI). For comparison,

the distributions for Xklp1–GFP on single taxol-microtu-

bules (gray circles, under conditions as in Figure 5B) and

for PRC1–Alexa 647 in a PRC1–Alexa 488-containing anti-

parallel overlap (blue dots, under conditions as in A, left)

are shown.

(C) Mean squared displacement (MSD) curves for the same

conditions as shown in (B), and a fit to the data of Xklp1–

GFP in antiparallel overlaps using the ‘‘persistent random

walk’’ model, yielding persistence time Tp = 2.31 ± 0.68 s

and mean velocity v = 0.52 ± 0.06 mm/s (mean ± 95% CI,

see Experimental Procedures). Error bars are SEM.

(D) Calculated probability distribution of the average

exploratory distance (black dots, see Extended Experi-

mental Procedures) of Xklp1–GFP molecules in antiparallel

overlaps and the predicted probability curve (dashed red)

calculated from the persistent random walk model using

the parameter values as obtained from the fits in (B) and

(C). For comparison, calculated average exploratory

distance distributions are plotted for Xklp1–GFP on single

stabilized microtubules and PRC1–Alexa 647 in overlaps

(as in B and C). Inset: semi-logarithmic plot of average

exploratory distances.

See also Figure S5, Figure S6, and Movie S8.

Xenopus kinesin-4 Xklp1 binds only weakly to

microtubules. However, by interacting with

PRC1, Xklp1 is recruited to antiparallel microtu-

bule overlaps, reminiscent of selective recruit-

ment of sliding motors of the kinesin-5 and kine-

sin-6 families by Ase1 in budding yeast and

fission yeast, respectively (Fu et al., 2009; Khme-

linskii et al., 2009). This recruitment results in the local inhibition of

microtubule plus-end growth in antiparallel overlaps by Xklp1,

leading to a defined steady-state overlap length. This explains

why the midzone unnaturally extends but half-spindles remain

connected upon lossofkinesin-4 invivo, resulting in long, distorted

anaphase spindles (Kurasawa et al., 2004; Zhu and Jiang, 2005).

Processive Overlap Exploration by Kinesin-4 EnsuresAdaptive Overlap Length ControlXklp1 reduces the dynamicity of microtubules by lowering the

turnover of tubulin at the growing microtubule end. On individual


microtubules this activity can be observed in vitro only at highly

elevated Xklp1 concentrations (several hundred nanomolar) well

above physiological levels (Bieling et al., 2010a). In contrast,

recruitment of Xklp1 to antiparallel microtubule overlaps by

PRC1 results in complete inhibition of microtubule growth

already at lower, close-to physiological concentrations at which

individual microtubules remain unaffected in their growth. As

a consequence, specific recruitment of Xklp1 ensures the local

inhibition of growth of only those microtubules engaged in anti-

parallel, PRC1-decorated microtubule overlaps. Activation of

the PRC1/kinesin-4 system, through dephosphorylation at

anaphase onset, therefore results in tight bundling and growth

inhibition of antiparallel microtubule overlaps in the central

spindle (Kurasawa et al., 2004; Mollinari et al., 2002; Zhu and

Jiang, 2005), whereas microtubules residing outside of the mid-

zone are likely not affected by the PRC1/kinesin-4 system.

Xklp1 motors move processively with velocities of around

0.5 mm/s within the PRC1-decorated antiparallel microtubule

overlap (Figure 6C, Figure S5B). The drag force exerted by

PRC1 on Xklp1 can be estimated to be low (Extended Experi-

mental Procedures) compared to typical stall forces of kinesins

(Carter and Cross, 2005; Svoboda and Block, 1994), explaining

why Xklp1 is slowed down only mildly by interacting with

PRC1. This explains why Xklp1 efficiently explores the antipar-

allel overlap instead of generating substantial microtubule sliding

forces. This contrasts Xklp1 to antiparallel microtubule sliding

motors like kinesin-5, which reside relatively static between

the sliding microtubules (Kapitein et al., 2005; Kapoor and

Mitchison, 2001; Miyamoto et al., 2004; Uteng et al., 2008).

How does processive movement of Xklp1 within the antipar-

allel overlap region result in length control? Accumulation of

Xklp1 at growing microtubule ends (Figure 5A) is likely to be

responsible for the dose-dependent inhibition of microtubule

growth (Figure 4). The targeting of Xklp1 to growing microtubule

ends is reminiscent of the manner by which the microtubule de-

polymerizing kinesins Kip3 (kinesin-8) (Varga et al., 2006, 2009)

and MCAK (kinesin-13) (Helenius et al., 2006) reach the ends of

artificially stabilized microtubules. Both proteins are ‘‘collected’’

from an extended region near the end of stabilized microtubules

in vitro either by directional (Kip3) or diffusive (MCAK) motility.

Xklp1 (kinesin-4) is also collected from an extended region,

however, in this case from the microtubule overlap by processive

bidirectional motility (Figure 6A, Figure S5A). For such an

‘‘antenna mechanism’’ to be effective under conditions where

microtubules grow, it is important that the velocity of Xklp1 in

the overlap (�0.5 mm/s [Figure 6C, Figure S5B]) is considerably

faster than the velocity with which the microtubules elongate

(�35 nm/s in our in vitro experiments and �200 nm/s in vivo

[Rusan et al., 2001; Tournebize et al., 1997]). This might explain

why Xklp1 is a fast motor in comparison to other mitotic kinesins.

The lowest Xklp1 concentration (5 nM) required to form antipar-

allel microtubule overlaps with stable steady-state lengths

resulted in overlaps of around 5 mm (Figure 4B). This length is in

remarkably good agreement with the maximal exploratory

distance of individual Xklp1 molecules in PRC1-enriched antipar-

allel overlaps (Figure 6D). This suggests that the threshold amount

of Xklp1 required to induce a complete stopof microtubule growth

is reached at this concentration by exploiting the entire explor-

atory range of Xklp1 molecules in the overlap. At higher concen-

trations, the threshold amount is already reached by ‘‘collecting’’

from shorter overlaps, leading to characteristic overlap lengths as

a function of the Xklp1 concentration. In Xenopus egg extract, the

concentration of Xklp1 is approximately 100 nM (Bieling et al.,

2010a). Our experiments predict that such concentrations lead

to overlap lengths of �2 mm (Figure 4B), which is in agreement

with a midzone length of 2–3 mm as measured in vivo (Kurasawa

et al., 2004; Mastronarde et al., 1993; Zhu and Jiang, 2005).

The PRC1/kinesin-4 system offers adaptive, self-regulatory

control of overlap size by introducing a negative feedback

between overlap length and microtubule growth. Shortening

the microtubule overlaps as a consequence of motor-driven

microtubule sliding would result in lowered amounts of kinesin-

4 at the microtubule ends because the motor could be collected

now only from shorter distances. Similar to our reversibility

experiment where we reduced the global concentration of

Xklp1 (Figures 4C–4E), the system would be predicted to dynam-

ically respond to antiparallel sliding resulting in shortening

overlaps by initiating microtubule elongation. Local, selective

inhibition of microtubule growth by kinesin-4 in antiparallel

overlaps at conditions that globally favor microtubule elongation

therefore allows for a dynamic control of overlap length.

Antiparallel Microtubule Overlaps in Yeastsand MetazoansMaintenance of stable antiparallel microtubule overlaps in the

spindle center requires coordination between microtubule

(de)polymerization and microtubule sliding. In particular, spindle

elongation during anaphase B must not compromise spindle

stability by potentially reducing the length of the antiparallel

-

+ -

-++-

- ++

PRC1

Xklp1

-

-+

+

+

-

MT seed

STOP

STOP

-

+ --

growth

GrowthInhibition

ProcessiveExploration

Recruitment

Recognition

-+

MT pair

single MT

Figure 7. A Two-Component System Controls Antiparallel Microtu-

bule Overlaps

Model illustrating schematically the activities that lead in their combination to

antiparallel microtubule overlap formation and length control (see Discussion).


overlap region. Several proposals have been made for mecha-

nisms to locally regulate microtubule dynamics in the anaphase

spindle (Cheerambathur et al., 2007; Gardner et al., 2008; Pear-

son et al., 2006). Interestingly, recent work in fission yeast has

linked the conserved microtubule-binding protein CLASP to

the stabilization of antiparallel microtubule bundles (Bratman

and Chang, 2007). Ase1-dependent recruitment of CLASP to

antiparallel microtubule overlaps has been suggested to prevent

microtubule depolymerization past the overlap by inducing

microtubule rescue (Bratman and Chang, 2007). A similar

PRC1-dependent stabilizing function of CLASP for antiparallel

microtubule overlaps during anaphase appears also to exist in

cultured human cells (Liu et al., 2009).

Although local induction of microtubule rescue might explain

how overlap stability is ensured, it does not readily explain how

microtubule overgrowth is prevented and thus how a defined

overlap size is established. This is especially the case for meta-

zoan systems in which microtubule growth velocities are high

(Kinoshita et al., 2001). Our in vitro experiments have demon-

strated how the PRC1/kinesin-4 system can control antiparallel

overlap length at conditions that globally favor microtubule elon-

gation. It might be the combination of inhibition of microtubule

growth by kinesin-4 and the promotion of rescues by CLASP

that ensures the stabilization and length control of antiparallel

microtubule overlaps in metazoans. The adaptive nature of this

mechanism would allow for independent regulation of antiparallel

microtubule sliding activities at different stages of anaphase.

In contrast to metazoan anaphase spindles, the length of

antiparallel microtubule overlaps does not appear to be tightly

controlled in yeast. Microtubule plus ends are dynamic in fission

yeast anaphase spindles as well as in bundled interphase arrays

(Sagolla et al., 2003). The slower microtubule growth velocities

in yeast might be one reason for less stringent control of the

overlap size in this organism. In addition, yeast uses the proces-

sive motor kinesin-8 (Klp5/6 in fission yeast and Kip3 in budding

yeast) as the major microtubule plus-end-destabilizing factor

(Cottingham et al., 1999; Gupta et al., 2006; West et al., 2001).

Length-dependent destabilization of microtubules by kinesin-8

(Tischer et al., 2009; Varga et al., 2006) might be sufficient for

a less stringent control of overlap stability. This might explain

the absence of kinesin-4 from both fission and budding yeast.

During interphase, fission yeast also uses antiparallel microtu-

bule overlaps as means to organize the microtubule cytoskeleton.

Recently, a model for the formation of such overlaps was

proposed based on the combined action of the sliding motor

Klp2 (a kinesin-14) and Ase1 (Janson et al., 2007). In this model,

antiparallel microtubules slide with respect to each other driven

by minus-end-directed motors targeted selectively to microtu-

bule plus ends. Sliding stops either as a consequence of plus

ends growing beyond the overlap region or possibly by friction ex-

erted by overlap-enriched Ase1 (Janson et al., 2007). This leads to

the formation of antiparallel microtubule overlaps with inverted

polarity and dynamically varying sizes, a mechanism that differs

from that of tight overlap size control during metazoan anaphase.

ConclusionWe have reconstituted a two-component system necessary and

sufficient to form antiparallel microtubule overlaps with precisely

defined length in vitro, an essential process for proper midzone

formation in metazoan anaphase spindles. The in vitro reconsti-

tution allowed us to dissect the mechanism of autonomous

antiparallel microtubule recognition by PRC1 and of local micro-

tubule growth inhibition and overlap length control by kinesin-4.

In the future, the extension of our reconstitution approach by

including other midzone proteins promises to yield further mech-

anistic understanding of aspects of the organization and function

of the late mitotic spindle.


Dynamic Microtubule Overlap Assay

A flow chamber assembled from a biotin-PEG functionalized glass coverslip

and a PLL-PEG passivated glass slide separated by two stripes of double-

sided tape (Bieling et al., 2010b) was filled with a series of solutions: (1)

Pluronic F-127 and k-casein, (2) NeutrAvidin and k-casein, (3) buffer, (4) short

GMP–CPP stabilized, biotinylated brightly labeled microtubule seeds (contain-

ing 26% Alexa Fluor 568-labeled tubulin), (5) buffer. Microtubule growth was

initiated by flowing in dimly labeled tubulin (containing 10% Alexa Fluor

568-labeled tubulin) together with midzone proteins at the indicated concen-

trations. The final assay buffer was 80 mM K-PIPES, pH 6.8, 85 mM KCl,

85 mM K-acetate, 4 mM MgCl2, 1 mM GTP, 2 mM MgATP, 1 mM EGTA,

10 mM b-mercaptoethanol, 0.25% [wt/wt] Brij-35, 0.1% methyl cellulose,

100 mg/ml b-casein, and oxygen scavengers. If not stated otherwise, the

concentrations of proteins were 5 nM PRC1–Alexa 647, 15 nM Xklp1–GFP

or Xklp1506–GFP, and 17.5 mM dimly labeled tubulin.

To test the reversibility of antiparallel microtubule overlap formation, over-

laps were allowed to form in the presence of 5 nM PRC1–Alexa 647, 100 nM

Xklp1–GFP (PRC1:Xklp1 ratio of 1:20), and 17.5 mM dimly labeled tubulin for

7–10 min at 28�C ± 1�C. Then the PRC1:Xklp1 ratio was changed to 1:1 by

flowing in a solution with only 5 nM Xklp1–GFP, leaving all other concentrations

constant.

Multicolor time-lapse imaging using TIRF microscopy was always

performed at 28�C ± 1�C.

Single-Molecule Imaging

Single full-length Xklp1–GFP molecules were imaged at a concentration of

2.5 nM on individual paclitaxel-stabilized, biotinylated, and fluorescently

labeled microtubules (5% Alexa 568-labeled tubulin) in low ionic strength

buffer (80 mM K-PIPES, pH 6.8, 85 mM K-acetate, 4 mM MgCl2, 2 mM MgATP,

1 mM EGTA, 10 mM b-mercaptoethanol, 10 mM paclitaxel, 0.25% [wt/wt]

Brij-35, 100 mg/ml b-casein, and oxygen scavengers). Xklp1506–GFP at 20

nM, or Kin401–GFP at 0.1 nM as a control, was imaged in very low ionic strength

buffer (12 mM K-PIPES, pH 6.8, 2 mM MgCl2, 2 mM MgATP, 1 mM EGTA,

10 mM b-mercaptoethanol, 10 mM paclitaxel, 100 mg/ml b-casein, and oxygen

scavengers).

For imaging individual PRC1–Alexa 647 molecules in dynamic antiparallel

microtubule overlaps, the PRC1–Alexa 647 concentration was lowered to 50

pM or to 5 pM in the presence of 5 nM PRC1–Alexa 488. For imaging individual

Xklp1–GFP molecules in dynamic, PRC1-containing antiparallel microtubule

overlaps, the Xklp1–GFP concentration was lowered to 500 pM.

Data Analysis

The movements of single fluorescent molecules were analyzed using auto-

mated particle tracking (Kalaimoscope, TransInsight, Dresden, Germany)

(Telley et al., 2009). For dwell time distributions, the cumulative distribution

function (CDF) of data was calculated using the ‘‘ecdf’’ function in Matlab

(MathWorks), and either a monoexponential or a biexponential function was

fitted to (1–CDF) data using a ‘‘least-squares’’ approach. Histograms of run

lengths were fitted with a monoexponential function as described (Telley et al.,

2009). MSD curves were calculated and fitted using the function v2 Dt2 + 2D

Dt + offset, with the mean velocity v and the diffusion coefficient D. For the

case of Xklp1 in the overlap the MSD curve was fitted to a ‘‘persistent

random walk’’ model (Othmer et al., 1988; Tranquillo and Lauffenburger,


1987): MSD = 2v2Tp [Dt – Tp (1 – exp(–Dt/Tp)] with the persistence time Tp and

mean velocity v, and hence v2Tp the effective diffusion coefficient. The

‘‘average exploratory distance’’ distribution in microtubule overlaps was

calculated by combining information from the dwell time distribution and the

MSD curve in antiparallel microtubule overlaps, thereby eliminating time.

Microtubule dynamics was determined by kymograph analysis using ImageJ.

The growth versus tubulin concentration relationship in the presence or

absence of Xklp1 was analyzed using a multiple linear regression model with

a categorical factor g (+Xklp1: g = 0, –Xklp1: g = 1) and y = b1 + m1 x + g

(b2 + m2 x). Generally, this analysis provides a significance test for differences

in the slope m and abscissa b of two linear relationships. Here, the slope and

abscissa represent the kinetic parameters kon and �koff, respectively, of the

microtubule dynamics.


Supplemental Information includes Extended Experimental Procedures, six

figures, and eight movies and can be found with this article online at doi:10.

1016/j.cell.2010.06.033.

ACKNOWLEDGMENTS

We thank Vladimir Rybin for analytical ultracentrifugation, Jan Ellenberg,

Elmar Schiebel, Johanna Roostalu, and Scott Hansen for critically reading

the manuscript, and the DFG, HFSPO, the European Commission (STREP

‘‘Active Biomics’’), and the Swiss National Science Foundation for financial

support.

Received: January 4, 2010




REFERENCES

Arnal, I., Heichette, C., Diamantopoulos, G.S., and Chretien, D. (2004).

CLIP-170/tubulin-curved oligomers coassemble at microtubule ends and

promote rescues. Curr. Biol. 14, 2086–2095.

Bieling, P., Kronja, I., and Surrey, T. (2010a). Microtubule motility on reconsti-

tuted meiotic chromatin. Curr. Biol. 20, 763–769.

Bieling, P., Telley, I.A., Hentrich, C., Piehler, J., and Surrey, T. (2010b).

Fluorescence microscopy assays on chemically functionalized surfaces for

quantitative imaging of microtubule, motor, and +TIP dynamics. Methods

Cell Biol. 95, 555–580.

Bratman, S.V., and Chang, F. (2007). Stabilization of overlapping microtubules

by fission yeast CLASP. Dev. Cell 13, 812–827.

Bringmann, H., Skiniotis, G., Spilker, A., Kandels-Lewis, S., Vernos, I., and

Surrey, T. (2004). A kinesin-like motor inhibits microtubule dynamic instability.

Science 303, 1519–1522.

Brouhard, G.J., Stear, J.H., Noetzel, T.L., Al-Bassam, J., Kinoshita, K.,

Harrison, S.C., Howard, J., and Hyman, A.A. (2008). XMAP215 is a processive

microtubule polymerase. Cell 132, 79–88.

Carter, N.J., and Cross, R.A. (2005). Mechanics of the kinesin step. Nature 435,

308–312.

Cheerambathur, D.K., Civelekoglu-Scholey, G., Brust-Mascher, I., Sommi, P.,

Mogilner, A., and Scholey, J.M. (2007). Quantitative analysis of an anaphase B

switch: predicted role for a microtubule catastrophe gradient. J. Cell Biol. 177,

995–1004.

Cottingham, F.R., Gheber, L., Miller, D.L., and Hoyt, M.A. (1999). Novel roles

for saccharomyces cerevisiae mitotic spindle motors. J. Cell Biol. 147,

335–350.

Drechsel, D.N., Hyman, A.A., Cobb, M.H., and Kirschner, M.W. (1992).

Modulation of the dynamic instability of tubulin assembly by the

microtubule-associated protein tau. Mol. Biol. Cell 3, 1141–1154.

Fu, C., Ward, J.J., Loiodice, I., Velve-Casquillas, G., Nedelec, F.J., and Tran,

P.T. (2009). Phospho-regulated interaction between kinesin-6 Klp9p and

microtubule bundler Ase1p promotes spindle elongation. Dev. Cell 17,

257–267.

Gardner, M.K., Bouck, D.C., Paliulis, L.V., Meehl, J.B., O’Toole, E.T., Haase,

J., Soubry, A., Joglekar, A.P., Winey, M., Salmon, E.D., et al. (2008).

Chromosome congression by Kinesin-5 motor-mediated disassembly of

longer kinetochore microtubules. Cell 135, 894–906.

Glotzer, M. (2009). The 3Ms of central spindle assembly: microtubules, motors

and MAPs. Nat. Rev. Mol. Cell Biol. 10, 9–20.

Gupta, M.L., Jr., Carvalho, P., Roof, D.M., and Pellman, D. (2006). Plus

end-specific depolymerase activity of Kip3, a kinesin-8 protein, explains its

role in positioning the yeast mitotic spindle. Nat. Cell Biol. 8, 913–923.

Helenius, J., Brouhard, G., Kalaidzidis, Y., Diez, S., and Howard, J. (2006).

The depolymerizing kinesin MCAK uses lattice diffusion to rapidly target

microtubule ends. Nature 441, 115–119.

Howard, J., and Hyman, A.A. (2007). Microtubule polymerases and depoly-

merases. Curr. Opin. Cell Biol. 19, 31–35.

Janson, M.E., de Dood, M.E., and Dogterom, M. (2003). Dynamic instability of

microtubules is regulated by force. J. Cell Biol. 161, 1029–1034.

Janson, M.E., Loughlin, R., Loiodice, I., Fu, C., Brunner, D., Nedelec, F.J., and

Tran, P.T. (2007). Crosslinkers and motors organize dynamic microtubules to

form stable bipolar arrays in fission yeast. Cell 128, 357–368.

Jiang, W., Jimenez, G., Wells, N.J., Hope, T.J., Wahl, G.M., Hunter, T., and

Fukunaga, R. (1998). PRC1: A human mitotic spindle-associated CDK

substrate protein required for cytokinesis. Mol. Cell 2, 877–885.

Kapitein, L.C., Peterman, E.J., Kwok, B.H., Kim, J.H., Kapoor, T.M., and

Schmidt, C.F. (2005). The bipolar mitotic kinesin Eg5 moves on both microtu-

bules that it crosslinks. Nature 435, 114–118.

Kapitein, L.C., Janson, M.E., van den Wildenberg, S.M., Hoogenraad, C.C.,

Schmidt, C.F., and Peterman, E.J. (2008). Microtubule-driven multimerization

recruits ase1p onto overlapping microtubules. Curr. Biol. 18, 1713–1717.

Kapoor, T.M., and Mitchison, T.J. (2001). Eg5 is static in bipolar spindles

relative to tubulin: evidence for a static spindle matrix. J. Cell Biol. 154,

1125–1133.

Khmelinskii, A., and Schiebel, E. (2008). Assembling the spindle midzone in the

right place at the right time. Cell Cycle 7, 283–286.

Khmelinskii, A., Lawrence, C., Roostalu, J., and Schiebel, E. (2007).

Cdc14-regulated midzone assembly controls anaphase B. J. Cell Biol. 177,

981–993.

Khmelinskii, A., Roostalu, J., Roque, H., Antony, C., and Schiebel, E. (2009).

Phosphorylation-dependent protein interactions at the spindle midzone

mediate cell cycle regulation of spindle elongation. Dev. Cell 17, 244–256.

Kinoshita, K., Arnal, I., Desai, A., Drechsel, D.N., and Hyman, A.A. (2001).

Reconstitution of physiological microtubule dynamics using purified compo-

nents. Science 294, 1340–1343.

Kurasawa, Y., Earnshaw, W.C., Mochizuki, Y., Dohmae, N., and Todokoro, K.

(2004). Essential roles of KIF4 and its binding partner PRC1 in organized

central spindle midzone formation. EMBO J. 23, 3237–3248.

Kwon, M., Morales-Mulia, S., Brust-Mascher, I., Rogers, G.C., Sharp, D.J., and

Scholey, J.M. (2004). The chromokinesin, KLP3A, dives mitotic spindle pole

separation during prometaphase and anaphase and facilitates chromatid

motility. Mol. Biol. Cell 15, 219–233.

Lee, Y.M., and Kim, W. (2004). Kinesin superfamily protein member 4 (KIF4) is

localized to midzone and midbody in dividing cells. Exp. Mol. Med. 36, 93–97.

Liu, J., Wang, Z., Jiang, K., Zhang, L., Zhao, L., Hua, S., Yan, F., Yang, Y.,

Wang, D., Fu, C., et al. (2009). PRC1 cooperates with CLASP1 to organize

central spindle plasticity in mitosis. J. Biol. Chem. 284, 23059–23071.

Loiodice, I., Staub, J., Setty, T.G., Nguyen, N.P., Paoletti, A., and Tran, P.T.

(2005). Ase1p organizes antiparallel microtubule arrays during interphase

and mitosis in fission yeast. Mol. Biol. Cell 16, 1756–1768.


Mastronarde, D.N., McDonald, K.L., Ding, R., and McIntosh, J.R. (1993).

Interpolar spindle microtubules in PTK cells. J. Cell Biol. 123, 1475–1489.

Miyamoto, D.T., Perlman, Z.E., Burbank, K.S., Groen, A.C., and Mitchison, T.J.

(2004). The kinesin Eg5 drives poleward microtubule flux in Xenopus laevis egg

extract spindles. J. Cell Biol. 167, 813–818.

Mollinari, C., Kleman, J.P., Jiang, W., Schoehn, G., Hunter, T., and Margolis,

R.L. (2002). PRC1 is a microtubule binding and bundling protein essential to

maintain the mitotic spindle midzone. J. Cell Biol. 157, 1175–1186.

Muller, S., Smertenko, A., Wagner, V., Heinrich, M., Hussey, P.J., and Hauser,

M.T. (2004). The plant microtubule-associated protein AtMAP65-3/PLE is

essential for cytokinetic phragmoplast function. Curr. Biol. 14, 412–417.

Othmer, H.G., Dunbar, S.R., and Alt, W. (1988). Models of dispersal in

biological systems. J. Math. Biol. 26, 263–298.

Pearson, C.G., Gardner, M.K., Paliulis, L.V., Salmon, E.D., Odde, D.J., and

Bloom, K. (2006). Measuring nanometer scale gradients in spindle microtubule

dynamics using model convolution microscopy. Mol. Biol. Cell 17, 4069–4079.

Powers, J., Rose, D.J., Saunders, A., Dunkelbarger, S., Strome, S., and

Saxton, W.M. (2004). Loss of KLP-19 polar ejection force causes misorienta-

tion and missegregation of holocentric chromosomes. J. Cell Biol. 166,

991–1001.

Rusan, N.M., Fagerstrom, C.J., Yvon, A.M., and Wadsworth, P. (2001). Cell

cycle-dependent changes in microtubule dynamics in living cells expressing

green fluorescent protein-alpha tubulin. Mol. Biol. Cell 12, 971–980.

Sagolla, M.J., Uzawa, S., and Cande, W.Z. (2003). Individual microtubule

dynamics contribute to the function of mitotic and cytoplasmic arrays in fission

yeast. J. Cell Sci. 116, 4891–4903.

Schuyler, S.C., Liu, J.Y., and Pellman, D. (2003). The molecular function of

Ase1p: evidence for a MAP-dependent midzone-specific spindle matrix.

Microtubule-associated proteins. J. Cell Biol. 160, 517–528.

Svoboda, K., and Block, S.M. (1994). Force and velocity measured for single

kinesin molecules. Cell 77, 773–784.

Telley, I.A., Bieling, P., and Surrey, T. (2009). Obstacles on the microtubule

reduce the processivity of Kinesin-1 in a minimal in vitro system and in cell

extract. Biophys. J. 96, 3341–3353.

Tischer, C., Brunner, D., and Dogterom, M. (2009). Force- and kinesin-8-

dependent effects in the spatial regulation of fission yeast microtubule

dynamics. Mol. Syst. Biol. 5, 250.

Tournebize, R., Andersen, S.S., Verde, F., Doree, M., Karsenti, E., and Hyman,

A.A. (1997). Distinct roles of PP1 and PP2A-like phosphatases in control of

microtubule dynamics during mitosis. EMBO J. 16, 5537–5549.

Tranquillo, R.T., and Lauffenburger, D.A. (1987). Stochastic model of leukocyte

chemosensory movement. J. Math. Biol. 25, 229–262.

Uteng, M., Hentrich, C., Miura, K., Bieling, P., and Surrey, T. (2008). Poleward

transport of Eg5 by dynein-dynactin in Xenopus laevis egg extract spindles.

J. Cell Biol. 182, 715–726.

Varga, V., Helenius, J., Tanaka, K., Hyman, A.A., Tanaka, T.U., and Howard, J.

(2006). Yeast kinesin-8 depolymerizes microtubules in a length-dependent

manner. Nat. Cell Biol. 8, 957–962.

Varga, V., Leduc, C., Bormuth, V., Diez, S., and Howard, J. (2009). Kinesin-8

motors act cooperatively to mediate length-dependent microtubule depoly-

merization. Cell 138, 1174–1183.

Verbrugghe, K.J., and White, J.G. (2004). SPD-1 is required for the formation of

the spindle midzone but is not essential for the completion of cytokinesis in C.

elegans embryos. Curr. Biol. 14, 1755–1760.

Verni, F., Somma, M.P., Gunsalus, K.C., Bonaccorsi, S., Belloni, G., Goldberg,

M.L., and Gatti, M. (2004). Feo, the Drosophila homolog of PRC1, is required

for central-spindle formation and cytokinesis. Curr. Biol. 14, 1569–1575.

Vernos, I., Raats, J., Hirano, T., Heasman, J., Karsenti, E., and Wylie, C. (1995).

Xklp1, a chromosomal Xenopus kinesin-like protein essential for spindle

organization and chromosome positioning. Cell 81, 117–127.

Walczak, C.E., and Heald, R. (2008). Mechanisms of mitotic spindle assembly

and function. Int. Rev. Cytol. 265, 111–158.

Walker, R.A., O’Brien, E.T., Pryer, N.K., Soboeiro, M.F., Voter, W.A., Erickson,

H.P., and Salmon, E.D. (1988). Dynamic instability of individual microtubules

analyzed by video light microscopy: rate constants and transition frequencies.

J. Cell Biol. 107, 1437–1448.

West, R.R., Malmstrom, T., Troxell, C.L., and McIntosh, J.R. (2001). Two

related kinesins, klp5+ and klp6+, foster microtubule disassembly and are

required for meiosis in fission yeast. Mol. Biol. Cell 12, 3919–3932.

Zhu, C., and Jiang, W. (2005). Cell cycle-dependent translocation of PRC1 on

the spindle by Kif4 is essential for midzone formation and cytokinesis. Proc.

Natl. Acad. Sci. USA 102, 343–348.

Zhu, C., Zhao, J., Bibikova, M., Leverson, J.D., Bossy-Wetzel, E., Fan, J.B.,

Abraham, R.T., and Jiang, W. (2005). Functional analysis of human microtu-

bule-based motor proteins, the kinesins and dyneins, in mitosis/cytokinesis

using RNA interference. Mol. Biol. Cell 16, 3187–3199.

Note Added in Proof

In a manuscript by Subramanian et al., which appears in this issue of Cell, the

authors arrive at similar conclusions regarding the nature of microtubule cross-

links formed by PRC1.


Insights into Antiparallel MicrotubuleCrosslinking by PRC1, a ConservedNonmotor Microtubule Binding ProteinRadhika Subramanian,1 Elizabeth M. Wilson-Kubalek,2 Christopher P. Arthur,2 Matthew J. Bick,3 Elizabeth A. Campbell,3

Seth A. Darst,3 Ronald A. Milligan,2 and Tarun M. Kapoor1,*1Laboratory of Chemistry and Cell Biology, The Rockefeller University, New York, NY 10065, USA2Department of Cell Biology, The Scripps Research Institute, La Jolla, CA 92037, USA3Laboratory of Molecular Biophysics, The Rockefeller University, New York, NY 10065, USA*Correspondence: [email protected]

DOI 10.1016/j.cell.2010.07.012

SUMMARY

Formation of microtubule architectures, required forcell shape maintenance in yeast, directional cellexpansion in plants and cytokinesis in eukaryotes,depends on antiparallel microtubule crosslinkingby the conserved MAP65 protein family. Here, wecombine structural and single molecule fluorescencemethods to examine how PRC1, the human MAP65,crosslinks antiparallel microtubules. We find thatPRC1’s microtubule binding is mediated by a struc-tured domain with a spectrin-fold and an unstruc-tured Lys/Arg-rich domain. These two domains, ateach end of a homodimer, are connected by a linkagethat is flexible on single microtubules, but forms well-defined crossbridges between antiparallel filaments.Further, we show that PRC1 crosslinks are compliantand do not substantially resist filament sliding bymotor proteins in vitro. Together, our data showhow MAP65s, by combining structural flexibility andrigidity, tune microtubule associations to establishcrosslinks that selectively ‘‘mark’’ antiparallel over-lap in dynamic cytoskeletal networks.

INTRODUCTION

The dynamic reorganization of microtubule networks plays a crit-

ical role in diverse biological processes, including cell migration,

neuronal transport and cell division. It is now clear that different

cytoskeletal architectures arise from the interplay between

motor proteins, which can crosslink and move microtubules rela-

tive to one another, and nonmotor microtubule associated

proteins (MAPs), which can crosslink microtubules to stabilize

specific orientations (Glotzer, 2009; Manning and Compton,

2008). While we have good biophysical and structural models

for motor proteins that crosslink microtubules, much less is

known about nonmotor microtubule crosslinking proteins.

Several nonmotor MAPs that crosslink microtubules (e.g.,

MAP65, NuMA, NuSAP and Mia1p) are now known to play

important roles in dividing and nondividing cells (Ribbeck et al.,

2006; Sasabe and Machida, 2006; Schuyler et al., 2003; Thadani

et al., 2009; Zeng, 2000). Current models for the functions of

these proteins are based on cellular localizations and loss-of-

function studies. However, we lack any structural data to explain

how microtubule crosslinking is achieved by these MAPs.

Recently, there have been several advances in our under-

standing of the structure of nonmotor MAPs. Among the best

characterized class of MAPs are the +TIP proteins (e.g.,

XMAP215, EB1 and CLIP170), which can dynamically track the

growing end of a microtubule. Microtubule binding in these

proteins is mediated by calponin-homology, CAP/Gly or TOG

domains (Slep and Vale, 2007). Similarly, structural work on

Ndc80, a conserved mitotic MAP, has revealed how a calpolin-

homology domain may be used to establish kinetochore-micro-

tubule associations during cell division (Ciferri et al., 2008; Wei

et al., 2007; Wilson-Kubalek et al., 2008). However, due to lack

of similarity in primary sequence, it is not very likely that these

structural models will shed light on nonmotor MAPs that can

crosslink two microtubules.

As a step toward developing structural models for how non-

motor MAPs may crosslink microtubules, we focused on the

conserved MAP65 family, which plays key roles in microtubule

organization in eukaryotes. Since their initial discovery in bud-

ding yeast, microtubule crosslinking functions of the MAP65

proteins have been shown to be required for cell shape mainte-

nance in yeast cells, directional cell expansion in plants and

formation of the central spindle in eukaryotes (Chan et al.,

1999; Jiang et al., 1998; Loiodice et al., 2005; Yamashita et al.,

2005). Currently, at least three activities have been ascribed to

these proteins. First, MAP65s can selectively crosslink microtu-

bules in an antiparallel orientation (Gaillard et al., 2008; Loiodice

et al., 2005). Second, these nonmotor crosslinkers can oppose

filament movements driven by motor proteins. For example,

Ase1, the fungal MAP65, is proposed to antagonize kinesin-14

driven filament sliding required for organizing microtubules

during interphase (Janson et al., 2007). Third, these crosslinking

proteins can recruit signaling proteins or kinesins to the microtu-

bule structures they stabilize. For example, the recruitment

of Polo-like kinase to the central spindle during cytokinesis is

mediated via interactions with PRC1, the human MAP65 (Neef


et al., 2007) and kinesin-5 driven microtubule sliding during

anaphase depends on Ase1 (Khmelinskii et al., 2009). Currently,

we do not have a structural framework to explain how these

MAPs specifically crosslink antiparallel microtubules. Moreover,

the activities of MAP65s have not been reconstituted in the pres-

ence of motor proteins to test if MAP65s resist filament sliding by

motor proteins or if their main function is to act as ‘‘marks’’ that

recruit other proteins to regions of antiparallel microtubule

overlap in dynamic networks.

Here we show, using single molecule fluorescence microscopy

assays, X-ray crystallography and electron microscopy that

PRC1 uses structured and unstructured domains to bind micro-

tubules. These domains at each end of a PRC1 homodimer are

connected by a linker that adopts a rigid conformation only when

crosslinking microtubules. We also show, in assays combining

TIRF and fluorescent speckle microscopy (FSM), that PRC1

does not substantially resist filament sliding by kinesin-5. Based

on these results, we propose a model for how a crosslinking MAP

can achieve specific and compliant crosslinking of microtubules

by balancing structural rigidity and flexibility.

RESULTS

Structured and Unstructured Domains MediateMicrotubule Binding in PRC1PRC1, like other Map65 family proteins, has a modular architec-

ture with an N-terminal coiled-coil domain, a central region that

can mediate microtubule binding, and a C-terminal regulatory

domain (Figure 1A). While the central domain is thought to

be required for microtubule binding, the contributions of the

C-terminal domain remain poorly characterized. To address

this, we analyzed the microtubule binding activity of PRC1 using

two approaches, a TIRF microscopy assay to examine the prop-

erties of single molecules and a microtubule cosedimentation

assay to analyze equilibrium binding.

For visualizing PRC1 molecules by fluorescence microscopy

we expressed and purified recombinant GFP-tagged full-length

PRC1 in bacteria. We found that C-terminal tags on PRC1

Figure 1. Single-Molecule Analysis of Microtubule Binding by PRC1

(A) Schematic of PRC10s domain organization and a guide for constructs used

in the fluorescence microscopy assays (purple: coiled-coil domain; green:

microtubule binding domain; black: C-terminal domain).

(B) Fluorescence intensity analysis of two PRC1 constructs, GFP-PRC1-FL

(aa: 1–620; intensity = 2.5 3 104 ± 0.9 3 104, N = 469) and GFP-PRC1-NS

(aa: 1–466; intensity = 2.0 3 104 ± 0.8 3 104, N = 156). Dimeric-Eg5-GFP

(Intensity = 2.5x104 ± 1.0x104, N = 377) and tetrameric-Eg5-GFP (Intensity =

4.2x104 ± 2.2 3 104, N = 290) were used as references. Intensities are reported

as mean ± SD.

(C–H) Single molecule TIRF assay was used to examine the association of

PRC1 constructs (green) with microtubules (orange) immobilized on a glass

surface. (C) Schematic for assay showing the two constructs, GFP-PRC1-FL

and GFP-PRC1-NSDC (aa 1–486). Single frames showing two-color overlays

(top) and associated kymographs (below) of GFP-PRC1-FL (D and E) or

GFP-PRC1-NSDC (F and G). (H) Distribution of microtubule association life-

times for GFP-PRC1-FL (blue) and GFP-PRC1-NSDC (red).

(I–K) Microtubule association of GFP-PRC1-NSDC under different ionic

strength conditions. Representative kymographs from assays at 0.75x motility

buffer (I), motility buffer (J), motility buffer+20 mM KCl. (K) The scale bar repre-

sents 1.5 mm, 10 s. See also Figure S1 and Figure S3.


resulted in constructs that were highly unstable and therefore

used a construct with GFP fused to the N-terminus of PRC1

(GFP-PRC1-FL). We first examined the oligomerization state of

single GFP-PRC1-FL molecules. Analysis of the intensities of

single fluorescent spots of GFP-PRC1-FL immobilized on a glass

coverslip, when compared to intensities of known dimeric and

tetrameric reference constructs, indicated that GFP-PRC1-FL

is a dimer (Figure 1B), similar to the yeast homolog, Ase1 (Kapi-

tein et al., 2008a; Schuyler et al., 2003). We also confirmed that

untagged PRC1 is a homo-dimer and has an extended confor-

mation (see below, and data not shown). Previous studies have

suggested that PRC1 exists as a homo-tetramer, but gel filtration

chromatography alone can sometimes be unreliable for such

analyses of proteins with elongated structures (Zhu et al.,

2006). To examine the interaction of single PRC1 molecules

with a microtubule, we used streptavidin to attach X-rhoda-

mine-labeled, biotinylated microtubules to a glass surface. After

blocking the glass surface, we added low concentrations of

GFP-PRC1-FL (20 pM) (Figure 1C). Similar to what is observed

for Ase1, we found that single GFP-PRC1-FL molecules diffuse

in 1-D along the microtubule surface (Figure S1D and 1E; (Kapi-

tein et al., 2008a)). Analysis of time-lapse sequences using

kymographs showed that individual GFP-PRC1-FL molecules

can maintain associations with the microtubule lattice that can

last several seconds (t1/2 = 7 ± 0.4 s; Figure 1H).

To examine the contribution of the C terminus of PRC1 to

microtubule binding, we first generated a construct comprised

of just the N-terminal dimerization and the microtubule-binding

domain (aa 1–466, hereafter GFP-PRC1-NS). Intensity distribu-

tion of GFP-PRC1-NS indicated that this construct, like the

full-length protein, is a dimer (Figure 1B). We were unable to

detect microtubule associations of single molecules of GFP-

PRC1-NS at the concentrations needed to resolve single

molecules using TIRF microscopy (<10 nM; data not shown),

suggesting that residues in the C terminus domain contribute

to microtubule binding. Secondary structure algorithms predict

that the C-terminal domain in PRC1 is likely to be unstructured.

Consistent with this prediction, we find that PRC1’s C terminus

is prone to proteolysis. A highly susceptible cleavage site resides

between two Lys/Arg-rich basic clusters in this domain (aa. Asn

500; Figure S1 available online). As basic amino acids are often

implicated in mediating interactions with the negatively charged

microtubules surface, we generated a GFP-fused PRC1 con-

struct with one of the two basic-residue clusters (aa 1–486,

hereafter GFP-PRC1-NSDC). Under TIRF microscopy condi-

tions we could readily observe microtubule binding by single

GFP-PRC1-NSDC molecules. Diffusion on the microtubule was

apparent for the subset of binding events that lasted several

seconds (Figures 1F and 1G). Analysis of the life-times of micro-

tubule association indicated that this construct has a higher

microtubule unbinding rate than the full-length protein (Fig-

ure 1H, lifetime is less than the lower limit of reliable event

detection, < 3 s). Increasing salt concentrations reduced the

GFP-PRC1-NSDC binding lifetimes, consistent with a charge-

dependent microtubule-binding interaction mediated by

PRC1’s C terminus (Figures 1I–1K).

We next carried out microtubule cosedimentation assays for

an independent analysis of the contribution of the C-terminal

domain to PRC1’s microtubule interaction. For these experi-

ments we generated PRC1 constructs lacking the dimerization

domain so that we could exclude the potentially complex effects

of filament crosslinking/bundling on this analysis (Figure 2A).

These PRC1 constructs were confirmed to be monomeric by

size exclusion chromatography (data not shown). The microtu-

bule binding affinity (Kd) of a construct comprising of the central

domain and the C terminus (aa 341–620, named PRC1-SC) was

0.6 ± 0.3 mM (Figures 2B and 2D). A construct comprising of the

central domain alone (aa 341–466, hereafter named PRC1-S)

had a �5-fold weaker binding (Kd: 3.3 ± 1.8 mM; Figures 2C

and 2D). Interestingly, we observe cooperative microtubule

binding by PRC1-SC (Hill coefficient of 3 ± 1). This suggests

Figure 2. Microtubule Cosedimentation Assays to Determine Equi-

librium Dissociation Constants for PRC10s Microtubule-Binding

Domains

(A) Schematic for constructs used in this assay: PRC1-S (aa 341–466) and

PRC1-SC (aa 341–620). SDS-PAGE analysis of cosedimentation assays for

PRC1-SC (B) and PRC1-S (C). Arrows in (B) indicate C terminus proteolysis

products in PRC1-SC. Bands marked with red boxes in (B) indicate the relative

tubulin concentration at which 50% of the different PRC1-SC truncation prod-

ucts cosediment with microtubules. (D) Band intensities from the gels in (C)

were used to determine fraction protein bound, and plotted against microtu-

bule concentration (n = 3, mean ± SD). The data were fit to a modified Hill equa-

tion to determine Kd’s (PRC1-S: 3.3 ± 1.8 mM; PRC1-SC: 0.6 ± 0.3 mM).


that the C terminus may be responsible for interactions between

PRC1 molecules on the microtubule lattice analogous to what

has been previously reported for Ase1 (Kapitein et al., 2008a).

Together, these data indicate that PRC1’s microtubule binding

is mediated by two regions, one that is predicted to be structured

and another, unstructured region, rich in Lys/Arg residues.

PRC1’s Microtubule-Binding Domain Has a Spectrin FoldTo determine the structural basis of microtubule binding by

PRC1, we focused on its central microtubule-binding domain.

As this domain has no obvious homology to any known microtu-

bule binding motif, we used X-ray crystallography to determine

its structure. Nonisomorphous crystals of the PRC1-S construct

were obtained from both the native protein, which diffracted to

1.75 A resolution, and the selenomethionyl-substituted protein,

which diffracted to 2.0 A resolution. The structure of the seleno-

methionyl-substituted protein was solved by single-wavelength

anomalous dispersion, refined, and, subsequently used as

a model to solve the structure of the native protein using molec-

ular replacement (Table S1). We found that the microtubule-

binding domain of PRC1 is an�70 A long 3-helix bundle (labeled

helix-1, -2, and -3), with connecting loops (labeled loop1-2 and

loop2-3) and N- and C- termini at opposite ends (Figure 3A

and Figures S2A and S2B). Residues from all three helices

contribute to a hydrophobic core which dominates the interface

between the three helices (Figure 3B). This core is flanked on

both ends of the helix bundle by two salt bridges mediated by

conserved residues (Figure 3C).

Using DALI, we found that PRC1-S has high structural similarity

to spectrin domains (Figure 3D), motifs commonly found in

proteins associated with the actin cytoskeleton (e.g., alpha-

actinin, spectrin and dystrophin; (Djinovic-Carugo et al., 2002)).

However, a spectrin fold has not thus far been found in any other

known MAP. Interestingly, in actin-binding proteins, the elon-

gated and rigid triple-helix of a spectrin domain does not directly

mediate actin binding, but acts as a spacer between canonical

actin-binding motifs, such as calponin homology domains.

Therefore, it appears that the spectrin fold has been evolutionarily

‘recycled’ and re-engineered to act as a microtubule interacting

domain in the Map65 protein family. These findings establish

a new role for the spectrin motif as a microtubule-binding domain.

A Conserved Basic Region Forms the Microtubule-Binding Surface in the Spectrin Domain of PRC1As it was unclear how the spectrin fold would bind microtubules,

we analyzed the electrostatic surface potential and the conser-

vation of surface residues for this domain. A map of the electro-

static surface potential shows a positively charged region

comprising of loop1-2 and residues from all three helices that

are proximal to this loop in the three dimensional structure

(Figure 3E). Interestingly, a cluster of highly conserved residues

is also located within this region (Figure 3F), suggesting that

the junction of helix-1 and helix-2 could form the microtubule

binding region in the spectrin fold. To test this hypothesis we

generated constructs of PRC1’s spectrin domain in which basic

residues in this region were mutated to alanine (Figure S2C).

Each of these constructs was characterized to be monomeric

by size-exclusion chromatography (Figure S2D). We used micro-

tubule cosedimentation assays to compare microtubule binding

of mutant and wild-type PRC1-S constructs (Figure 3G). We

found that mutation of each of the basic residues in the most

conserved region reduced microtubule binding 2- to 4-fold,

though none of the individual mutations abolished microtubule

binding (Figure 3H). To test that the observed microtubule

binding is not simply a nonspecific electrostatic interaction, we

tested two other constructs. First, we made a C terminus trunca-

tion in PRC1-SC (labeled del453–466; aa 341–452) to remove

three surface exposed lysines on helix-3. Second, we generated

a construct in which a lysine residue on helix-2 (K407), distal to

the potential microtubule binding site, was mutated to alanine.

Both of these constructs show less than 50% reduction in

microtubule binding. From this analysis, we propose that the

spectrin domain in PRC1 uses a basic surface comprising of

conserved residues at one end of the triple helix bundle to

mediate microtubule binding (Figure 3F, circle).

Spectrin Domain Fits with an Optimal Orientationinto the Cryo-EM Density Map of the PRC1-MicrotubuleComplexTo investigate the interaction of PRC1 with the microtubule

lattice we examined the structure of dimeric PRC1 bound to

single microtubules by cryo-electron microscopy (Cryo-EM)

and helical image analysis. As PRC1-FL extensively bundles

microtubules at concentrations needed for this analysis, we

used our findings from single molecule experiments to engineer

a construct PRC1-NSDC (aa 1–486; Figure 4A), which was

more suitable for this analysis (Figure 1A). This construct was

confirmed to be dimeric by HPLC Size Exclusion Chromatog-

raphy/Laser Light Scattering Analysis (Figure S3), and found to

retain sufficient binding to fully decorate single microtubules

without extensive bundling.

Diffraction of individual cryo-EM images of PRC1-NSDC

bound to microtubules showed an 80 A layer line, indicating

that one PRC1-NSDC molecule binds a a/b-tubulin heterodimer.

The 3D reconstruction of microtubule bound PRC1-NSDC

showed a single rod-shaped density corresponding to PRC1-

NSDC, protruding approximately perpendicular to the microtu-

bule lattice (microtubule side view, Figure 4B; top view,

Figure 4C). Interestingly, the PRC1 density in the reconstruction

is of the same size as PRC1’s spectrin domain. The optimal fit of

the spectrin domain crystal structure to the reconstructions

oriented the conserved, basic residues at the junction of helix-1

and helix-2 toward the microtubule lattice (Figure 4D and 4E).

This binding model is also consistent with our site-directed

mutagenesis analysis (Figure 3H). Comparison of the 3D recon-

struction of this complex with cryo-EM structures of motor

proteins (e.g kinesin and dynein’s microtubule-binding domain)

bound to microtubules suggests that the binding site for PRC1

partially overlaps with, but is not identical to, the microtubule-

binding surface used by motor proteins (Carter et al., 2008;

Figure S4B). Surprisingly, we see no extra density distal to the

microtubule surface that would correspond to the N-terminus

coiled-coil dimerization domain in PRC1-NSDC. This indicates

that the coiled-coil domain is likely to be flexible and can

have more than one conformation when bound to a single

microtubule.


To further understand the PRC1-microtubule interaction, we

obtained a second EM reconstruction with a monomeric

construct that lacked most of the oligomerization domain but

contained the spectrin domain and the entire C terminus

(aa 303–620). The observed density for this construct overlaps

with the PRC1-NSDC density close to the microtubule lattice

(Figures S4C–S4E). We also observe some additional density

distal to the microtubule surface which is not observed in

PRC1-NSDC. It is possible that this density corresponds to

PRC1’s C terminus folding back to interact with the spectrin

domain or to the amino acids at PRC1’s N-terminus that may

be ordered when this construct binds microtubules. As we do

not observe any additional density past the C terminus of the

spectrin domain that is close to the microtubule lattice, we favor

the possibility that the Lys/Arg-rich domain is disordered even

when PRC1 is bound to a microtubule. Together, these data

Figure 3. PRC10s Conserved Microtubule-Binding Domain Adopts a Spectrin Fold

(A) Ribbon diagram shows the overall structure. Five disordered residues in the loop between helix-1 and 2 are indicated by dots.

(B) Hydrophobic residues from helix-1 (cyan), helix-2 (yellow) and helix-3 (green), which form the core of the triple helix bundle, are highlighted (sticks).

(C) Salt bridges between charged residues from helix-1 (cyan), helix-2 (yellow) and helix-3 (green) are indicated (spheres).

(D) Overlay of PRC10s spectrin domain (orange) with its closest structural homolog, which is a spectrin repeat in Plectin (blue, PDB = 2odu-A, Z-score = 7.7,

rmsd = 3.4 A calculated using DALI).

(E) Surface representation showing electrostatic potential of the spectrin domain in PRC1 (Red to blue is �10 kbT to +10 kbT, as calculated using APBS).

(F) Residues with low (cyan), intermediate (white), and high (magenta) conservation on the surface of the PRC10s spectrin domain. The labeled residues were

selected for mutagenesis studies.

(G) SDS-PAGE analysis of microtubule cosedimentation assays of PRC1-S mutants at 27 mM tubulin.

(H) Band intensities from (G) were quantified to determine the fraction of PRC1 bound to microtubules (n = 3, error bars indicate SE). See also Figure S2.


show that PRC1’s spectrin domain is highly ordered and less

flexible than the other domains when associated with a single

microtubule.

Cryo-Electron Tomography Reveals Ordered PRC1Crosslinks between Two Antiparallel MicrotubulesHelical reconstructions suggest that PRC1 may not be an inher-

ently rigid molecule when bound to one microtubule. It is difficult

to explain how a flexible protein with two microtubule interaction

surfaces at opposite ends of dimer can achieve the crosslinking

specificity reported for MAP65 proteins (Gaillard et al., 2008;

Janson et al., 2007). The only other structural analysis of cross-

linking by the MAP65 family is a cryo-EM study of the plant

homolog of PRC1, which show that the protein can form dense

crossbridges between two microtubules (Gaillard et al., 2008).

However, the conformational differences between crosslinking

and noncrosslinking molecules could not be visualized in this

study. It also remains unknown whether establishing ordered

crossbridges is a property of single MAP65 molecules or results

from protein-protein interactions that may be involved in PRC1’s

co-operative microtubule binding (Figure 2D). To address these

questions, we employed cryo-electron tomography to obtain

a higher resolution structure of two microtubules crosslinked

by PRC1 (Figure 5A). For this analysis we used the PRC1

construct we had used for helical reconstruction (PRC1-NSDC;

Figure 4A). 3D reconstructions from cryo-electron tomography

of microtubules densely crosslinked by PRC1-NSDC shows

rod-like striations connecting two microtubules (Figure 5B

and inset). From analyzing the overall direction of crosslinking

Figure 5. The Dimerization Domain in PRC1 Is Ordered When Cross-

linking Two Microtubules(A) Schematic highlights the section viewed in the cryotomographic recon-

structions shown in (B) and (C), which encompasses the two microtubules

and the PRC1 crosslinks.

(B and C) Slices through cryotomographic 3D reconstructions of microtubules

crosslinked by PRC1-NSDC. Slices represent the central area of the microtu-

bules and the bound PRC1-NSDC. The top and bottom of the microtubules are

excluded in these views. Examples of microtubule pairs with dense (B) and

sparse (C) PRC1 occupancy. Insets show 4-fold enlargements of the cross-

links between microtubules. Small red arrows in the inset highlight individual

cross-bridges. Blue arrow in (B) indicates a PRC1 molecule that does not

form crosslinks with another microtubule. Long red arrows in (B) indicate the

polarity of microtubules, portions of which are highlighted in blue for clarity.

The top two microtubules in (B) share PRC1 crossbridges as do the bottom

two microtubules, however, the middle two microtubules are separated by

another microtubule (dotted blue line) that is positioned below the plane and

only the top of its tubulin lattice can be observed.

Figure 4. The Spectrin Domain Fits into the Cryo-EM Density Map of

the PRC1-Microtubule Complex with an Optimal Orientation

(A) Schematic comparing full-length PRC1 to the construct PRC1-NSDC

(aa 1–486) used for Cryo-EM analysis.

(B) Surface rendered side-view of 3D EM density map of the microtubule-

PRC1-NSDC complex.

(C) Top view of cryo-EM density map of undecorated tubulin (gold) superim-

posed with the microtubule-PRC1-NSDC complex (purple).

(D and E) (D) Side- and (E) top-view of the crystal structure (cyan ribbon

diagram) docked into the PRC1 density (purple mesh) protruding from an

a/b tubulin dimer (gold). Residues R377 and K387, which are involved in micro-

tubule binding, are indicated in red and black respectively. Microtubule

polarity indicated in (B) was determined by comparison with a 3D EM structure

of a dynein-microtubule complex, which has a well defined polarity (Carter

et al., 2008). See also Figure S4.


protein density, we estimated that 90% (n = 20) of the cross-

linked microtubules were antiparallel. This shows that the C

terminus truncation construct has antiparallel crosslinking spec-

ificity similar to full-length MAP65 and Ase1 (Gaillard et al., 2008;

Schuyler et al., 2003). We find that the average distance between

two crosslinked filaments was 35 nm (±2 nm; n = 20) and the

crossbridge angle of PRC1 linkages was �70� (±5; n = 15) rela-

tive to the microtubule surface. As helical reconstruction shows

that the spectrin domain in PRC1 protrudes almost perpendic-

ular to the microtubule surface, the crossbridge orientation likely

results from �20� hinges between the oligomerization and the

spectrin domains.

Consistent with what is seen from the helical reconstruction

analysis, the tomograms show that the PRC1 densities observed

on the microtubule surface distal to crosslinked microtubule (i.e.,

the noncrosslinking PRC1 molecules, Figure 5B, blue arrow)

were less ordered than those between two filaments. This indi-

cates that noncrosslinking and crosslinking PRC1 molecules

have distinct conformations.

We next examined if the regular crossbridge geometry and the

intermicrotubule distances observed are retained in regions of

sparse PRC1 decoration between crosslinked microtubules.

We find that single PRC1 molecules still form crossbridges

with orientations similar to those observed in regions of dense

crosslinks (Figure 5C and inset). Interestingly, we found that

PRC1 crosslinks were not observed in regions where filament

spacing was significantly greater than 35 nm, indicating that

PRC1 molecules exhibit specificity for a narrow range of inter-

microtubule distances. These findings agree with our observa-

tions in fluorescence microscopy assays in which GFP-PRC1-

FL accumulation is seen only at a narrow range of microtubule

crosslinking angles (unpublished data). Together, our data indi-

cate that although the dimerization domain is not entirely rigid

when PRC1 is bound to a single microtubule, this domain in indi-

vidual PRC1 molecules can adopt a specific conformation upon

crosslinking two microtubules. The rigidity in the linker, together

with oriented binding by the spectrin domain, can allow PRC1 to

be a selective crosslinker of antiparallel microtubules aligned

with a narrow range of interfilament spacing.

PRC1 Is a Compliant Crosslinker of AntiparallelMicrotubulesOur analysis of microtubule binding by the structured and

unstructured domains in PRC1 shows that these interactions

have moderate affinity. Together with the observed 1-D diffusion

of single PRC1 molecules on microtubules, these data predict

that PRC1 crosslinks may not substantially resist filament sliding

by motor proteins. To test this hypothesis, we devised an in vitro

assay to visualize near-simultaneously PRC1 localization and

motor-protein driven sliding of pairs of crosslinked microtubules.

We attached biotinylated microtubules to a glass surface via

streptavidin. These microtubules also incorporated low levels

of fluorescent tubulin for analysis of relative microtubule motion

using fluorescent speckle microscopy (FSM). GFP-PRC1-FL

was added to the immobilized microtubules, followed by nonbio-

tinylated microtubules to generate a ‘sandwich’ comprised of

two microtubules crosslinked by PRC1 (Figure 6A). As would

be expected, based on studies of Ase1, we find that GFP-

PRC1-FL shows a 10 ± 4 -fold preference for regions where

two microtubule overlap relative to regions of single microtu-

bules (N = 15; Figures S5A–S5C; (Kapitein et al., 2008a)). To drive

the relative sliding of microtubules in this ‘sandwich’, we added

kinesin-5, a well characterized motor protein needed for cell

division.

Under these conditions we observed two types of events.

First, kinesin-5 could move a shorter microtubule relative to

a longer filament in the ‘sandwich’ such that the amount of

overlap tracked by GFP-PRC1, remained unchanged (Figures

6B–6D). Second, we observed events in which the microtubule

overlap in the ‘sandwich’ reduced at the rate of filament motion

(Figure 6E and Figures S5D and S5E) and GFP-PRC1 dynami-

cally tracked the microtubule overlap zone (Figure 6F). Remark-

ably, in this set of events, the velocity of relative filament sliding

by kinesin-5 did not increase as the extent of microtubule over-

lap reduced (Figures 6G and 6H). Fluorescence intensity analysis

indicated that the number of PRC1-crosslinks in the overlap

region decreased proportionately (data not shown). Analysis of

all data show that kinesin-5’s sliding velocity reduced only

�2.4 fold over an 18-fold change in PRC1 concentration

(Figure 6I). At higher PRC1 concentrations in this assay extensive

microtubule bundling was observed and it was very difficult to

reliably detect pairs of crosslinked microtubules. The few events

that could be analyzed under these conditions indicated that

a further reduction in kinesin-5 driven filament sliding was not

observed as PRC1 concentration was increased (Figure S5F-I).

We estimated the ratio of PRC1 and kinesin-5 at microtubule

overlap zones in our assays. Using GFP-kinesin-5 at concentra-

tions similar to the un-tagged kinesin-5 used in the above

assays, we found that microtubule sliding was driven by a few

motor protein molecules, and unlike PRC1, kinesin-5 did not

show a preference for microtubule overlap zones (data not

shown and (Hentrich and Surrey, 2010)). The average fluores-

cence intensity of single GFP-PRC1-FL molecules was then

used to determine the number of GFP-PRC1-FL molecules at

microtubule overlap regions. These data indicate that the ratio

of GFP-PRC1-FL to kinesin-5 molecules at microtubule overlap

zones was �25 (total solution concentration: GFP-PRC1-FL =

0.54 nM, kinesin-5 = 1.8 nM). This shows that crosslinks formed

by multiple PRC1 molecules do not substantially resist the rela-

tive microtubule movement driven by fewer kinesin-5 molecules.

DiscussionMembers of the MAP65 family organize microtubules by prefer-

entially crosslinking filaments with an antiparallel orientation. In

this study, we provide a structural framework for how three

structurally distinct domains in human MAP65 (PRC1) are

combined to achieve selective and efficient crosslinking of anti-

parallel microtubules, as would be needed during the self-orga-

nization of dynamic cytoskeletal networks.

The conserved microtubule-binding domain in PRC1 has

a spectrin fold. First identified in spectrin, a constituent of the

membrane skeleton, this fold has since been seen in several

actin crosslinking proteins such as alpha-actinin and dystrophin

(Djinovic-Carugo et al., 2002). In these proteins, spectrin-

domains appear as repeated units that connect other actin-

binding domains and regulate properties of the cytoskeletal


structures. In PRC1, this domain appears to have evolved to

mediate microtubule binding, an entirely different role for this

protein fold. Further, the spectrin fold is unrelated to the other

protein folds (e.g Cap/Gly motifs and calponin-homology

domains) that are known to mediate interactions with the micro-

tubule lattice. It has been proposed that some protein surfaces

are evolutionarily selected to act as a versatile protein interaction

platform. This is best exemplified in the Fc domain of IgG, which

is structurally adapted to interact with several different protein

scaffolds (DeLano et al., 2000). Similarly, it appears that the

microtubule surface has evolved to bind a large variety of unre-

lated structural motifs and thereby accommodate diverse

MAPs that can carry out a wide range of functions (Amos and

Schlieper, 2005).

Microtubule binding by the spectrin domain is augmented by

a Lys/Arg rich unstructured domain in PRC1. Synergistic microtu-

bule binding by structured and unstructured domains have also

been seen in other MAPs such as Ndc80 (Guimaraes et al.,

2008; Miller et al., 2008), suggesting that this feature may be

a frequent adaptation in MAPs. There are also at least two other

functions ascribed to these unstructured microtubule-binding

domains. First, these unstructured domains are often sites of

phospho-regulation (Holt et al., 2009). Interestingly, Cdk1 phos-

phorylation sites in PRC1 map to the unstructured Lys/Arg

domain (Zhu et al., 2006). Our data suggest that these phosphor-

ylations would directly attenuate microtubule affinity by reducing

the net positive charge of the domain. Dephosphorylation of these

residues would activate PRC1’s microtubule binding at

anaphase, as would be needed for PRC1’s functions during the

final stages of cell division. Second, as has been suggested for

Ndc80, these domains are proposed to allow a mode of attach-

ment which does not significantly resist microtubule movement.

We find that PRC1 does not strongly oppose microtubule sliding

by kinesin-5, indicating that moderate binding affinities and diffu-

sive microtubule interactions of PRC1 can permit microtubule

movement while maintaining attachment. This result is also

consistent with the reported diffusion constant of Ase1 which indi-

cates that > 100 crosslinker molecules would be needed to

generate a resistive forcegreater than 1.5 pN, which is in the range

of the force required to inhibit kinesin-5 movement (Bormuth et al.,

2009; Kapitein et al., 2008a; Korneev et al., 2007; Valentine and

Block, 2009). This would be relevant in vivo, when microtubule

bundling by Ase1 and sliding by kinesin-5 are both required for

spindle elongation in anaphase B (Khmelinskii et al., 2009).

Crossbridges between two PRC1-crosslinked microtubules

are formed by PRC1’s dimerization domain. These are seen to

project at an angle of 70� relative to the microtubule lattice.

Binding at a fixed angle relative to the microtubule lattice has

Figure 6. PRC1 Crosslinks Do Not Substantially Resist the Relative

Sliding of Two Microtubules by Kinesin-5

(A) Schematic illustrating the assay used to examine the effect of PRC1 (green)

on kinesin-5 (cyan)-mediated relative sliding of two microtubules (orange),

when extent of overlap between microtubules is unchanged. Near-simulta-

neous dual-mode microscopy was used to image microtubules (via wide-field

fluorescent speckle microscopy) and GFP-PRC1-FL (via Total Internal Reflec-

tion Fluorescence (TIRF) microscopy).

(B) Frames from a time-lapse sequence (1 min interval) show GFP-PRC1-FL

(green) enriched at regions where two crosslinked microtubules (red) overlap.

(C) Corresponding kymograph shows the surface-attached static microtubule

(vertical streaks) and a moving microtubule (diagonal streaks, 16.5 nm/s).

(D) Kymograph shows that the GFP-PRC1-FL decorated region moves at the

velocity of the moving microtubule.

(E) Schematic illustrating the assay when overlap between microtubules

decreases during relative microtubule sliding.

(F–H) Frames from a time-lapse sequence (2 min interval) (F) and corresponding

kymographs showing microtubule movement (7 nm/s) (G) and GFP-PRC1-FL

localization to overlap region that reduces due to relative sliding of filaments (H).

(I) Velocity distributions for kinesin-5 driven microtubule sliding at 1.8 nM

kinesin-5 and 0.034 (V = 23.1 ± 6.3 nm/s, N = 43), 0.14 nM (V = 17.7 ±

3.5 nm/s, N = 39), and 0.54 nM GFP-PRC1-FL (V = 9.7 ± 3.1 nm/s, N = 41).

Velocities are reported as mean ± SD. The scale bars represent 1.5 mm,

100 s. See also Figure S5.


also been reported for Ndc80 (Wilson-Kubalek et al., 2008) but

the implications of defined projection angles for any microtu-

bule-binding protein is thus far unknown. The crossbridge also

determines the intermicrotubule spacing between two microtu-

bules. This intermicrotubule distance could also affect the ability

of motor proteins to bind and slide PRC1 crosslinked microtu-

bules, providing an additional mechanism for activating or

deactivating specific motors at these structures. For example,

the reported length of kinesin-5 motor is approximately 95 nm,

which is�2-fold greater than the 37 nm intermicrotubule spacing

of PRC1 crosslinked microtubules (Kashina et al., 1996). Hence,

the reduction in the velocity of microtubule sliding by kinesin-5

seen in our experiments at high PRC1 concentrations could

result from the inability of kinesin-5 to bind properly and

efficiently walk along both microtubules with dense PRC1 cross-

links. Such modulation of motor activity through control of

interfilament spacing has been proposed to affect the magnitude

of active forces generated in striated muscles during muscle

contraction (Millman, 1998).

Based on our results, we propose a structural model for how

polarity-specific crosslinking is mediated by PRC1 (Figures 7A

and 7B). The ability of a microtubule associated protein to distin-

guish parallel and antiparallel filaments relies on two factors.

First, PRC1 molecules must decode filament polarity when

bound to one microtubule. Our results show that the spectrin

domain is the most ordered region in a PRC1-microtubule com-

plex and uses a well-defined surface for microtubule binding.

This suggests that this domain makes contacts with the microtu-

bule lattice that decode filament orientation. Second, the micro-

tubule polarity needs to be transmitted across the linker to

the second spectrin domain in the PRC1 homodimer. Our data

show that the dimerization domain in PRC1 has a single confor-

mation when crosslinking two microtubules. The structural

rigidity of this domain is likely to be responsible for PRC1’s selec-

tivity for antiparallel microtubules. Though specificity can be

achieved by oriented binding of the spectrin domain and rigidity

in the linker domain, the weak microtubule binding affinity of the

spectrin domain alone does not explain the extensive filament

bundling induced by PRC1. To increase binding and conse-

quently crosslinking efficiency, PRC1 uses an unstructured posi-

tively charged domain for maintaining long-lived associations

with microtubules. Additionally, the flexibility in the linker domain

of PRC1, which appears to have more than one conformation

when bound to one microtubule, may allow for an initial contact

with a second microtubule that could have a wide-range of

orientations. Relative to a highly rigid structure, such flexibility

could increase the crosslinking efficiency of PRC1 molecules.

These features would enable PRC1 to stay associated with the

first microtubule encountered, explore its length by 1-D diffu-

sion, and thereby increase the probability of capturing a second

filament for establishing antiparallel linkages between two

microtubules.

Many cellular processes require recognition of a ‘‘mark’’ at

precise locations. Post-translational modifications such as ubiq-

uitination and methylation are some of the common marking

mechanisms for proteins and DNA in cells. In the microtubule

cytoskeleton, the +TIP-proteins track and identify microtubule

growing microtubule plus-ends (Akhmanova and Steinmetz,

2008). Similarly, by recognizing specific microtubule geometries

and forming compliant crosslinks, the MAP65 proteins can mark

antiparallel microtubule overlap during the self-organization of

dynamic cytoskeletal networks.


Crystallization of PRC1-S

PRC1-S (341-466) was concentrated to 50 mg/ml in motility buffer (80 mM

PIPES (pH 6.8), 1 mM MgCl2, 1 mM EGTA) and 150 mM KCl and crystallized

by hanging drop vapor diffusion at 4�C using an equal volume of protein

sample and crystallization solution consisting of 0.1 M CHES (pH 9.5) and

30%–35% PEG 3350. Crystals (needles of dimension 25 X 25 X 300 mm)

appeared between 1 - 3 weeks. Microseeding was used to obtain crystals of

the selenomethionyl derivative of the construct. Crystals were frozen in liquid

nitrogen after a 5 min soak in a solution comprising of the crystallization solu-

tion with 5%–10% v/v glycerol. Both native and SAD datasets were collected

along single needles at NE-CAT beamline 24ID-E at APS at the peak selenium

wavelength 0.97918 A and processed as described in the supplement.

Cryo-EM and Image Analysis

Microtubules were polymerized as previously described (Wilson-Kubalek

et al., 2008). Microtubules (0.166 mg/ml in motility buffer) were applied to

plasma cleaned C-flat grids and incubated with PRC1-NSDC (0.53 mg/ml) or

PRC1-SC’ (0.9 mg/ml). The sample was then vitrified in liquid ethane, using

a manual plunger. The data sets were collected using a GATAN cryoholder

Figure 7. PRC1 Is a Compliant, Microtubule-Overlap Tracking

Protein that Tunes Structural Rigidity to Specifically Crosslink Two

Antiparallel Microtubules

(A) A model for how PRC1 can align microtubules into antiparallel arrays. The

spectrin domain in PRC1 can make oriented contacts with the microtubule to

decode filament polarity. The unstructured domain acts to enhance the

binding affinity, while allowing diffusion along microtubules. The dimerization

domain is not entirely rigid on a single microtubule but adopts a specific

conformation when crosslinking two microtubules.

(B) The rigidity and flexibility of the different domains in PRC1 can facilitate

sorting of randomly oriented microtubules into antiparallel arrays and allow

PRC1 to function as a selective ‘mark’ for microtubule overlap regions.


and a FEI Tecnai F20 transmission electron microscope equipped with a Gatan

Ultrascan 4000 3 4000 pixel CCD camera. Images of the decorated microtu-

bules were recorded in low-dose conditions (<10 e/A) at a magnification of

�29,000 and a nominal defocus range from 1.5 to 2.0 mm. For image analysis

of the PRC1-microtubule complexes, PRC1-decorated 15-protofilament

microtubules were selected from CCD images. 3D reconstructions were calcu-

lated using Phoelix, essentially as described elsewhere (Whittaker et al., 1995).

3D EM Reconstructions

Surface representations of side- and top- views of the cryo-EM reconstruc-

tions (Figure 4 and S4) were produced using the Chimera software package

(Pettersen et al., 2004). To interpret the PRC1 densities protruding from the

microtubule density, we manually docked the crystal structure of the spectrin

domain into the EM density reconstructions using Chimera.

Cryotomography

Tilt-series images were acquired using the Serial EM software package (Mas-

tronarde, 2005) on a Tecnai F20 microscope through a range of +/� 50� at 1.5�

increments. The specimen was subjected to a total dose of �100 electrons.

The tomographic image reconstruction was performed using the IMOD soft-

ware package (Kremer et al., 1996).

Fluorescence Microscopy

All experiments were performed on an instrument described previously (Kapi-

tein et al., 2008b). GFP was imaged with TIRF illumination from a 491-nm laser

source (Cobalt Calypso 50; Solamere Technology) and X-rhodamine was

imaged with wide-field illumination. For all experiments, oxygen scavenging

mix (OS) comprising of 25 mM glucose, 40 mg/ml glucose oxidase, 35 mg/ml

catalase and 0.5% beta-mercaptoethanol was included in the final buffer.

Kinesin-5/PRC1 sliding experiments were performed as described previ-

ously with some modifications (Kapitein et al., 2008b). Briefly, X-rhodamine-

labeled biotinylated taxol-stabilized microtubules were immobilized on the

coverslip by first coating the surface with biotinylated BSA followed by strepta-

vidin. After a brief incubation with casein to block nonspecific binding to

the surface, GFP-PRC1-FL was added and allowed to bind the immobilized

microtubules. This was followed by addition of X-rhodamine-labeled microtu-

bules which generated a ‘sandwich’ with two microtubules crosslinked by

PRC1. After washing the solution microtubules, the final assay mix comprising

of GFP-PRC1-FL and/or kinesin-5 in motility buffer supplemented with

80 mM KCl, 2 mM MgCl2, 1 mg/mL k-casein, 1 mM MgATP, 20 mM taxol and

OS mix was added to the flow chamber. TIRF and wide-field images were

acquired near simultaneously with 0.2 s exposure, 0.1 s-1 frame rate and EM

gain set to 100 and 200 for TIRF and wide-field respectively. Each pair of moving

microtubule was analyzed by kymographs using MetaMorph (MDS Analytical

Technologies). Sliding velocities were calculated from the slopes of the diag-

onal streaks generated by the movement of one speckled microtubule over

an immobilized microtubule. In a few rare events (<5%), we observed an abrupt

�2-fold change in microtubule movement velocity in the kymographs. Based

on our analyses of kinesin-5 sliding alone (without PRC1), we believe this is

due to kinesin-5 switching between walking on both microtubules it crosslinks

to walking on only one of the two microtubules. This change in motility mode

leads to a 2-fold reduction in relative filament sliding. In these cases, both

velocities were measured and are included in the histograms shown.

ACCESSION NUMBERS

Coordinates have been deposited with PDB IDs 3NRX (native crystal structure)

and 3NRY (SelenoMet crystal structure) and EMDB accession codes

EMD-5205 (EM map of PRC1 [aa 1-486]) and EMD-5212 (EM map of PRC1

[aa 303-620]).


Supplemental Information includes Extended Experimental Procedures, five

figures, and one table and can be found with this article online at doi:

10.1016/j.cell.2010.07.012.

ACKNOWLEDGMENTS

We thank K. Rajashankar (Argonne Advanced Photon Source) and Deena Oren

(Rockefeller University Structural Biology Resource Center (SBRC)) for tech-

nical assistance. T.M.K. is grateful to the NIH (GM65933 and GM65933-S1)

for support. R.S. is a recipient of the Rockefeller Women and Science postdoc-

toral fellowship. R.A.M. acknowledges support from the NIH (GM52468). We

also acknowledge the National Resource for Automated Molecular Micros-

copy (NIH RR17573), the SBRC (NIH 1S10RR022321-01), and the Keck

Facility at Yale University (NIH 1S10RR023748-01) for instrument use.

Received: May 10, 2010

Revised: June 21, 2010

Accepted: July 6, 2010


REFERENCES

Akhmanova, A., and Steinmetz, M.O. (2008). Tracking the ends: a dynamic

protein network controls the fate of microtubule tips. Nat. Rev. Mol. Cell

Biol. 9, 309–322.

Amos, L.A., and Schlieper, D. (2005). Microtubules and maps. Adv. Protein

Chem. 71, 257–298.

Bormuth, V., Varga, V., Howard, J., and Schaffer, E. (2009). Protein friction

limits diffusive and directed movements of kinesin motors on microtubules.

Science 325, 870–873.

Carter, A.P., Garbarino, J.E., Wilson-Kubalek, E.M., Shipley, W.E., Cho, C.,

Milligan, R.A., Vale, R.D., and Gibbons, I.R. (2008). Structure and functional

role of dynein’s microtubule-binding domain. Science 322, 1691–1695.

Chan, J., Jensen, C.G., Jensen, L.C., Bush, M., and Lloyd, C.W. (1999). The

65-kDa carrot microtubule-associated protein forms regularly arranged fila-

mentous cross-bridges between microtubules. Proc. Natl. Acad. Sci. USA

96, 14931–14936.

Ciferri, C., Pasqualato, S., Screpanti, E., Varetti, G., Santaguida, S., Dos Reis,

G., Maiolica, A., Polka, J., De Luca, J.G., De Wulf, P., et al. (2008). Implications

for kinetochore-microtubule attachment from the structure of an engineered

Ndc80 complex. Cell 133, 427–439.

DeLano, W.L., Ultsch, M.H., de Vos, A.M., and Wells, J.A. (2000). Convergent

solutions to binding at a protein-protein interface. Science 287, 1279–1283.

Djinovic-Carugo, K., Gautel, M., Ylanne, J., and Young, P. (2002). The spectrin

repeat: a structural platform for cytoskeletal protein assemblies. FEBS Lett.

513, 119–123.

Gaillard, J., Neumann, E., Van Damme, D., Stoppin-Mellet, V., Ebel, C.,

Barbier, E., Geelen, D., and Vantard, M. (2008). Two microtubule-associated

proteins of Arabidopsis MAP65s promote antiparallel microtubule bundling.

Mol. Biol. Cell 19, 4534–4544.

Glotzer, M. (2009). The 3Ms of central spindle assembly: microtubules, motors

and MAPs. Nat. Rev. Mol. Cell Biol. 10, 9–20.

Guimaraes, G.J., Dong, Y., McEwen, B.F., and Deluca, J.G. (2008). Kineto-

chore-microtubule attachment relies on the disordered N-terminal tail domain

of Hec1. Curr. Biol. 18, 1778–1784.

Hentrich, C., and Surrey, T. (2010). Microtubule organization by the antago-

nistic mitotic motors kinesin-5 and kinesin-14. J. Cell Biol. 189, 465–480.

Holt, L.J., Tuch, B.B., Villen, J., Johnson, A.D., Gygi, S.P., and Morgan, D.O.

(2009). Global analysis of Cdk1 substrate phosphorylation sites provides

insights into evolution. Science 325, 1682–1686.

Janson, M.E., Loughlin, R., Loiodice, I., Fu, C., Brunner, D., Nedelec, F.J., and

Tran, P.T. (2007). Crosslinkers and motors organize dynamic microtubules to

form stable bipolar arrays in fission yeast. Cell 128, 357–368.

Jiang, W., Jimenez, G., Wells, N.J., Hope, T.J., Wahl, G.M., Hunter, T., and

Fukunaga, R. (1998). PRC1: a human mitotic spindle-associated CDK

substrate protein required for cytokinesis. Mol. Cell 2, 877–885.


Kapitein, L.C., Janson, M.E., van den Wildenberg, S.M., Hoogenraad, C.C.,

Schmidt, C.F., and Peterman, E.J. (2008a). Microtubule-driven multimerization

recruits ase1p onto overlapping microtubules. Curr. Biol. 18, 1713–1717.

Kapitein, L.C., Kwok, B.H., Weinger, J.S., Schmidt, C.F., Kapoor, T.M., and

Peterman, E.J. (2008b). Microtubule cross-linking triggers the directional

motility of kinesin-5. J. Cell Biol. 182, 421–428.

Kashina, A.S., Baskin, R.J., Cole, D.G., Wedaman, K.P., Saxton, W.M., and

Scholey, J.M. (1996). A bipolar kinesin. Nature 379, 270–272.

Khmelinskii, A., Roostalu, J., Roque, H., Antony, C., and Schiebel, E. (2009).

Phosphorylation-dependent protein interactions at the spindle midzone

mediate cell cycle regulation of spindle elongation. Dev. Cell 17, 244–256.

Korneev, M.J., Lakamper, S., and Schmidt, C.F. (2007). Load-dependent

release limits the processive stepping of the tetrameric Eg5 motor. Eur.

Biophys. J. 36, 675–681.

Kremer, J.R., Mastronarde, D.N., and McIntosh, J.R. (1996). Computer visual-

ization of three-dimensional image data using IMOD. J. Struct. Biol. 116,

71–76.

Loiodice, I., Staub, J., Setty, T.G., Nguyen, N.P., Paoletti, A., and Tran, P.T.

(2005). Ase1p organizes antiparallel microtubule arrays during interphase

and mitosis in fission yeast. Mol. Biol. Cell 16, 1756–1768.

Manning, A.L., and Compton, D.A. (2008). Structural and regulatory roles of

nonmotor spindle proteins. Curr. Opin. Cell Biol. 20, 101–106.

Mastronarde, D.N. (2005). Automated electron microscope tomography using

robust prediction of specimen movements. J. Struct. Biol. 152, 36–51.

Miller, S.A., Johnson, M.L., and Stukenberg, P.T. (2008). Kinetochore attach-

ments require an interaction between unstructured tails on microtubules and

Ndc80(Hec1). Curr. Biol. 18, 1785–1791.

Millman, B.M. (1998). The filament lattice of striated muscle. Physiol. Rev. 78,

359–391.

Neef, R., Gruneberg, U., Kopajtich, R., Li, X., Nigg, E.A., Sillje, H., and Barr, F.A.

(2007). Choice of Plk1 docking partners during mitosis and cytokinesis is

controlled by the activation state of Cdk1. Nat. Cell Biol. 9, 436–444.

Pettersen, E.F., Goddard, T.D., Huang, C.C., Couch, G.S., Greenblatt, D.M.,

Meng, E.C., and Ferrin, T.E. (2004). UCSF Chimera–a visualization system

for exploratory research and analysis. J. Comput. Chem. 25, 1605–1612.

Ribbeck, K., Groen, A.C., Santarella, R., Bohnsack, M.T., Raemaekers, T.,

Kocher, T., Gentzel, M., Gorlich, D., Wilm, M., Carmeliet, G., et al. (2006).

NuSAP, a mitotic RanGTP target that stabilizes and cross-links microtubules.

Mol. Biol. Cell 17, 2646–2660.

Sasabe, M., and Machida, Y. (2006). MAP65: a bridge linking a MAP kinase to

microtubule turnover. Curr. Opin. Plant Biol. 9, 563–570.

Schuyler, S.C., Liu, J.Y., and Pellman, D. (2003). The molecular function of

Ase1p: evidence for a MAP-dependent midzone-specific spindle matrix.

Microtubule-associated proteins. J. Cell Biol. 160, 517–528.

Slep, K.C., and Vale, R.D. (2007). Structural basis of microtubule plus end

tracking by XMAP215, CLIP-170, and EB1. Mol. Cell 27, 976–991.

Thadani, R., Ling, Y.C., and Oliferenko, S. (2009). The fission yeast TACC

protein Mia1p stabilizes microtubule arrays by length-independent crosslink-

ing. Curr. Biol. 19, 1861–1868.

Valentine, M.T., and Block, S.M. (2009). Force and premature binding of ADP

can regulate the processivity of individual Eg5 dimers. Biophys. J. 97,

1671–1677.

Wei, R.R., Al-Bassam, J., and Harrison, S.C. (2007). The Ndc80/HEC1

complex is a contact point for kinetochore-microtubule attachment. Nat.

Struct. Mol. Biol. 14, 54–59.

Whittaker, M., Carragher, B.O., and Milligan, R.A. (1995). PHOELIX: a package

for semi-automated helical reconstruction. Ultramicroscopy 58, 245–259.

Wilson-Kubalek, E.M., Cheeseman, I.M., Yoshioka, C., Desai, A., and Milligan,

R.A. (2008). Orientation and structure of the Ndc80 complex on the microtu-

bule lattice. J. Cell Biol. 182, 1055–1061.

Yamashita, A., Sato, M., Fujita, A., Yamamoto, M., and Toda, T. (2005). The

roles of fission yeast ase1 in mitotic cell division, meiotic nuclear oscillation,

and cytokinesis checkpoint signaling. Mol. Biol. Cell 16, 1378–1395.

Zeng, C. (2000). NuMA: a nuclear protein involved in mitotic centrosome

function. Microsc. Res. Tech. 49, 467–477.

Zhu, C., Lau, E., Schwarzenbacher, R., Bossy-Wetzel, E., and Jiang, W. (2006).

Spatiotemporal control of spindle midzone formation by PRC1 in human cells.

Proc. Natl. Acad. Sci. USA 103, 6196–6201.

Note Added in Proof

In a manuscript by Bieling et al., which appears in this issue of Cell, the authors

arrive at similar conclusions regarding the nature of microtubule crosslinks

formed by PRC1.


Aurora Kinases and Protein Phosphatase 1Mediate Chromosome Congressionthrough Regulation of CENP-EYumi Kim,1,2,4 Andrew J. Holland,1,4 Weijie Lan,1 and Don W. Cleveland1,3,*1Ludwig Institute for Cancer Research2Department of Molecular and Cellular Biology, University of California, Berkeley, CA 94720, USA3Department of Cellular and Molecular Medicine

University of California, San Diego, La Jolla, CA 92093, USA4These authors contributed equally to this work

*Correspondence: [email protected]

DOI 10.1016/j.cell.2010.06.039

SUMMARY

Opposing roles of Aurora kinases and protein phos-phatase 1 (PP1) during mitosis have long been sug-gested. Here, we demonstrate that Aurora kinasesA and B phosphorylate a conserved residue on thekinetochore motor CENP-E. PP1 binds CENP-E viaa motif overlapping this phosphorylation site andbinding is disrupted by Aurora phosphorylation.Phosphorylation of CENP-E by the Auroras is en-riched at spindle poles, disrupting binding of PP1and reducing CENP-E’s affinity for individual micro-tubules. This phosphorylation is required for CENP-E-mediated towing of initially polar chromosomestoward the cell center. Kinetochores on such chro-mosomes cannot make subsequent stable attach-ment to spindle microtubules when dephosphoryla-tion of CENP-E or rebinding of PP1 to CENP-E isblocked. Thus, an Aurora/PP1 phosphorylationswitch modulates CENP-E motor activity as anessential feature of chromosome congression frompoles and localized PP1 delivery by CENP-E to theouter kinetochore is necessary for stable microtu-bule capture by those chromosomes.

INTRODUCTION

Accurate chromosome segregation during mitosis requires the

bipolar attachment of duplicated chromosomes to spindle

microtubules emanating from opposite poles. Each time a cell

divides, a specialized proteinaceous structure called the kineto-

chore assembles on the surface of each centromere, and it is

the kinetochore that binds to spindle microtubules and directs

chromosome motion during mitosis (Cleveland et al., 2003).

Microtubule capture by the kinetochore is a stochastic process.

Initial kinetochore attachment is often mediated via an interac-

tion with the lateral surface of a microtubule, and kinetochores

attached in this manner undergo rapid, dynein-mediated pole-

ward motion (Rieder and Alexander, 1990). Although some

chromosomes achieve biorientation without being transported

to the spindle pole, dynein-mediated transport is an important

mechanism to collect chromosomes to a common microtu-

bule-dense region, where kinetochores have a greater chance

of promoting efficient chromosome alignment.

Congression of polar-localized chromosomes to a metaphase

position is powered by a processive, plus end-directed kineto-

chore motor CENP-E (Kinesin-7) (Kapoor et al., 2006; Kim

et al., 2008). In various cell types and organisms, removal or

inhibition of CENP-E leads to a failure in complete metaphase

chromosome alignment, with a few unattached chromosomes

found close to the spindle poles (Putkey et al., 2002; Wood

et al., 1997; Yao et al., 2000; Yucel et al., 2000). Even the kinet-

ochores that do become bioriented and fully aligned in the

absence of CENP-E stably bind only half as many microtubules

(Putkey et al., 2002). Our finding that CENP-E possesses a highly

flexible and very long coiled-coil (Kim et al., 2008) raises the

possibility that, while it can work advantageously for initial

capture, CENP-E may also contribute, in part, to the inappro-

priate attachments of kinetochores. Indeed, the process of

capturing spindle microtubules by kinetochores is prone to

errors. Undesirable attachment frequently occurs in early prom-

etaphase, with a single kinetochore capturing microtubules from

both spindle poles (merotelic attachment), or both sister kineto-

chores attached to the same pole (syntelic attachment) (Cimini

and Degrassi, 2005). These improper kinetochore attachments,

if not resolved, can lead to chromosome missegregation and

aneuploidy (Holland and Cleveland, 2009).

Correction of aberrant kinetochore attachment requires

a conserved Ser/Thr kinase Aurora/Ipl1 (Lampson et al., 2004;

Tanaka et al., 2002). While budding yeast has a single Aurora

kinase Ipl1, metazoans express at least two Aurora kinases,

Aurora A and B. Like Ipl1, Aurora B is a component of the chromo-

some passenger complex (together with INCENP, Survivin, and

Borealin/Dasra) and is targeted to the inner centromere from

prophase to metaphase (Ruchaud et al., 2007). Aurora B is

thought to aid chromosome biorientation by destabilizing the

kinetochore-microtubule interaction of improperly attached

chromosomes (Cimini et al., 2006). Several proteins directly


involved in microtubule capture at the kinetochore, including

Dam1 in budding yeast and the core kinetochore microtubule

binding components in metazoans (Ndc80 and KNL1), are known

Aurora B substrates (Cheeseman et al., 2002, 2006; Gestaut

et al., 2008; Liu et al., 2010), and phosphorylation by Aurora B

has been shown to decrease the affinity of these proteins for

microtubules (Cheeseman et al., 2006; Gestaut et al., 2008).

Despite the high sequence similarity with Aurora B, Aurora A

plays distinct roles during mitosis. Localized to the centrosomes

during interphase and at the spindle poles during mitosis, Aurora

A has been implicated in promoting mitotic entry and is required

for centrosome maturation and separation (Marumoto et al.,

2005). Inhibition of Aurora A has also been reported to cause

chromosome congression defects (Hoar et al., 2007; Kunitoku

et al., 2003; Marumoto et al., 2003); however, how Aurora A

acts to promote chromosome alignment is unknown.

Genetic evidence in yeast (Francisco et al., 1994) and in verte-

brates (Emanuele et al., 2008; Liu et al., 2010) suggest that the

Aurora kinase activity is opposed by the ubiquitous Ser/Thr phos-

phatase, protein phosphatase 1 (PP1/Glc7). In vertebrates, PP1

isoforms a and g can be detected at outer kinetochores (Trinkle-

Mulcahy et al., 2006; Trinkle-Mulcahy et al., 2003), and PP1 has

been shown to stabilize kinetochore-microtubule attachment by

counteracting Aurora B kinase activity (Liu et al., 2010; Sassoon

et al., 1999). Recently, the non-essential yeast protein Fin1 and

conserved kinetochore protein KNL1 have been identified to target

some PP1 to yeast and vertebrate kinetochores, respectively

(Akiyoshi et al., 2009; Liu et al., 2010). However, whether the kinet-

ochore possesses multiple docking modules for PP1 is not known.

Phosphorylation of the C-terminal tail of CENP-E by Cdk1,

MAPK, or Mps1 has been previously proposed either to regulate

CENP-E motor activity prior to its binding to kinetochores (Espeut

et al., 2008) or inhibit a microtubule binding site in the tail that may

link anti-parallel, midzone microtubules in anaphase (Liao et al.,

1994; Zecevic et al., 1998). Additionally, among a proteomic

screen of 260 mitotic phosphoproteins, CENP-E was identified

to be multiply phosphorylated during mitosis (Nousiainen et al.,

2006). However, the significance of the phosphorylations of

CENP-E has not been established. Using purified components,

selective inhibitors and a phosphospecific antibody, here we

demonstrate that Aurora kinases, both A and B, phosphorylate

a single conserved residue close to the CENP-E motor domain.

We also identify a docking motif for PP1 that overlaps the site

of phosphorylation and demonstrate that PP1 binding to CENP-E

is disrupted by Aurora-mediated phosphorylation. Our findings

establish an Aurora/PP1 phosphorylation switch that is required

not only for congression of polar chromosomes through modula-

tion of the intrinsic motor properties of CENP-E, but also for

subsequent stable biorientation of those chromosomes by

CENP-E’s delivery of PP1 to the outer kinetochore.

RESULTS

A Conserved Phosphorylation Site Near the NeckDomain of CENP-E Is Phosphorylated by AuroraA and B In VitroIn searching for the origin of the one-dimensional diffusion found

in CENP-E motility (Kim et al., 2008), we identified a highly

conserved stretch of basic residues downstream of the CENP-E

coiled-coil neck (Figure 1A). Consisting of four or more consec-

utive arginines or lysines, this basic stretch and the following

threonine (422 in human and 424 in Xenopus laevis) are

conserved in almost all the eukaryotes that possess a clear

CENP-E homolog. Interestingly, the conserved threonine resides

in a consensus motif for phosphorylation by Aurora kinase

(Cheeseman et al., 2002) and has been previously mapped as

a phosphorylation site in a mass spectrometry-based proteomic

screen of mitotic spindles (Nousiainen et al., 2006).

To test whether CENP-E T422 (424 in Xenopus laevis) is phos-

phorylated by Aurora kinases, we performed in vitro kinase

assays using purified Aurora kinases and portions of Xenopus

CENP-E as a substrate. Xenopus Aurora B, together with its

activator INCENP, phosphorylated both full-length and a motor

fragment of CENP-E (CENP-E1-473). However, Aurora B failed

to phosphorylate CENP-E1-473 in which threonine 424 had

been converted to alanine (Figure 1B). Xenopus CENP-E

T424 was also readily phosphorylated by Aurora A (Figures

S1A and S1B available online), confirming that the conserved

threonine located close to the CENP-E motor domain is phos-

phorylated by both Aurora A and B in vitro. The stoichiometry

of CENP-E1-473 phosphorylation by Aurora A saturated at two

moles of PO4 per mole of CENP-E1-473, most likely with an

additional phosphorylation site located C-terminal to T424, as

a shorter CENP-E1-415 fragment was not phosphorylated by

either Aurora kinase (Figure S1B).

Aurora A and B Contribute to Phosphorylationof CENP-E T422 in CellsTo examine the phosphorylation of CENP-E T422 in vivo, a rabbit

polyclonal antibody was generated against a phosphopeptide of

human CENP-E surrounding T422 (Figure S1C). The affinity

purified anti-pT422 antibody recognized recombinant human

CENP-E1-429 only in the presence of active kinase (Figure 1C)

and recognition of phosphorylated Xenopus CENP-E1-428 by

the anti-pT422 antibody was abolished by the mutation T424A

(Figure S1D). The anti-pT422 antibody also recognized wild-

type (WT) CENP-E immunoprecipitated from nocodazole-

arrested human (DLD-1) cells, but not CENP-E containing a

T422A mutation or WT CENP-E that had been incubated with

l-phosphatase (Figure 1D). Together, these results demonstrate

that the anti-pT422 antibody specifically recognizes CENP-E

phosphorylated at T422.

To establish whether Aurora A or B phosphorylates CENP-E

T422 in cells, we took advantage of the anti-pT422 antibody

and a series of small molecule inhibitors that specifically inhibit

either one or both of the Aurora kinases. As expected, treatment

with the dual Aurora kinase inhibitor VX-680 (Harrington et al.,

2004) abolished phosphorylation of the Aurora A substrate

Transforming acidic coiled-coil 3 (p-TACC3) and the Aurora B

substrate histone H3 (p-Histone H3) (Figure 1E). VX-680 treat-

ment abolished phosphorylation of CENP-E at T422, whereas

treatments with an Aurora A specific inhibitor (MLN8054) or an

Aurora B specific inhibitor (ZM447439) (Ditchfield et al., 2003;

Manfredi et al., 2007) resulted in only a partial reduction in

T422 phosphorylation, indicating that inhibition of either Aurora

kinase alone is not sufficient to eliminate the phosphorylation


of CENP-E T422. However, when cells were treated with

MLN8054 and ZM447439 together to inhibit both Aurora A

and B, phosphorylation of T422 was completely inhibited (Fig-

ure 1E). Thus, we conclude that both Aurora A and B contribute

to the phosphorylation of CENP-E at T422 in vivo.

Phosphorylation of CENP-E T422 Is Enrichedon Kinetochores Close to the Spindle PolesIn unperturbed PtK2 cells, pT422 staining was uniformly detect-

able at individual kinetochores in early prometaphase, which

colocalized with the centromere components recognized by

autoantisera containing centromere antibodies (ACA) (Figure 1F).

The kinetochore-localized pT422 signal was reduced on chro-

mosomes congressed to the equator of the cells, but remained

enriched at the kinetochores of unaligned chromosomes that

are close to the spindle poles (Figure 1F). In nocodazole-treated

HeLa cells, the pT422 antibody recognized a large crescent

around kinetochore pairs, which colocalized with CENP-E and

the outer kinetochore protein Bub1 (Figure S2A). Kinetochore-

localized pT422 disappeared following depletion of CENP-E by

siRNA (Figure S2B), confirming the specificity of the pT422 stain-

ing at kinetochores. Inhibition of Aurora kinases with VX-680

sharply reduced kinetochore-localized pT422 signal (Figure 1G).

When normalized to the total level of CENP-E at the kinetochore

(which is also reduced in VX-680 treated cells (Ditchfield et al.,

2003)), a > 90% reduction in T422 phosphorylation was seen

following VX-680 treatment (Figure 1H), demonstrating that

kinetochore-localized CENP-E is a substrate for Aurora kinases

in vivo.

Aurora-Mediated Phosphorylation of CENP-E T422Reduces Its Affinity for MicrotubulesTo determine if phosphorylation of T422 affects the motor

properties of CENP-E, we phosphorylated T424 of Xenopus

CENP-E motor (CENP-E1-473) and measured CENP-E’s microtu-

bule-stimulated ATPase activity in the presence of an increasing

HCENP-EG

***

Nor

mal

ized

to p

T422

to to

tal C

EN

P-E

ratio 1.2

1.0

0.8

0.6

0.4

0.2

0

n>15

DMSO VX680

A 1

CENP-E motor core324

neck

a d a d ada dd aad

Neck linker

VNE VS TDE AL LKRYRKE I MDLKKQLEE VS - - LE TRAQAMEKDQL AQL LEEKDL LQK VQNEK I ENL TRML VTSSS L T LQQE LK AKRKRRVTWCLGVNE VSNDE AL LKRYRRE I ADLRKQLEE VN - - TK TRAQEMEKDQL AQL LDEKDL LQK VQDEK I NNLKRML VTSSS I ALQQE LE I KRKRRVTWCYGVNE VLDDDAL LKRYRKE I LDLKKQLEE VS - - MK TQ I HAMEKHQL AHL LEEKNS LQKMQEDR I RNL TEML VTS AS FSSKQNAK ARRRRRVTWAPGVNE VLDDE AL LKRYRKE I LDLKKQLENLESSSE TK AQAMAKEEHTQL L AE I KQLHKEREDR I WHL TN I VVASSQES - QQDQRVKRKRRVTWAPGVNEMVSDATMMKRLERE I K VLKDK L AEEE - - - - - - - - - - - - - - - - - RKNENQQK VEHLERQ I KHDMHK I I CGHS LSDKGQQ- - - - KRRRTWCP T

335 345 355 365 375 385 395 405 415 425

HumanMouse

ChickenXenopus

Drosophila

0

0.5

Paircoil scoreα-helical coiled-coil

Phosphothreonine

B

Coomassie

175

83

62

47.5

XCEN

P-E

XCE

1-47

3

-

175

83

62

47.5

32P

XCE

1-47

3 T424

AXC

ENP-

EXC

E1-

473

- XCE

1-47

3 T424

A

kDa kDa

Aurora B kinase assay

λ-PPase: +- -

WT

WT

T422

A

Myc

pT422

MycLAP-CENP-E:

IP

Myc

LAP

-CE

NP

-E(3

40 k

Da)

C Aurora B kinase assay

HC

EN

P-E

1-42

9

(49

kDa)

pT422

Coomassie

- WT

KDAurora B:

HCENP-E

D

E

F MergeDNA pT422ACATubulin

Ear

lypr

omet

aLa

tepr

omet

aLa

tepr

omet

a

MergepT422

Noc

+D

MS

ON

oc +

VX

-680

Inhibit

IP

Input

Figure 1. A Conserved Threonine Close to

the Motor Domain of CENP-E Is Phosphory-

lated by Aurora A and B on the Kinetochores

of Unaligned Chromosomes

(A) Alignment of CENP-E protein sequences using

ClustalW algorithm. a-helical coiled-coil in the

CENP-E neck were predicted for human CENP-E

using Paircoil (Berger et al., 1995). Heptad repeat

positions (a and d) in the coiled-coil are indicated.

(B) In vitro kinase assays using Aurora B/INCENP

to phosphorylate Xenopus full-length CENP-E

or CENP-E1-473 with or without T424A mutation.

Coomassie staining of purified proteins and auto-

radiogram showing incorporation of g-32P ATP.

(C) Coomassie staining and immunoblot for pT422

using human CENP-E1-429 incubated with wild-

type (WT) or kinase dead (KD) Aurora B.

(D) CENP-E immunoprecipitates from nocoda-

zole-treated DLD-1 cells expressing either WT or

T422A MycGFP-CENP-E were blotted with the

pT422 antibody. Half the WT immunoprecipitate

was treated with l-phosphatase.

(E) Nocodazole-arrested DLD-1 cells expressing

MycGFP-CENP-E were treated with Aurora kinase

inhibitors and MG132 and blotted for P-TACC3

(Aurora A substrate), P-Histone H3 (Aurora B

substrate), and tubulin (loading control). CENP-E

immunoprecipitates using a Myc antibody were

blotted with the pT422 antibody.

(F) PtK2 cells were stained for DNA (Blue), Tubulin

(Green), ACA, and pT422 (Red).

(G) Nocodazole-arrested HeLa cells were treated

with VX-680 and MG132 and stained for CENP-E

(Green), pT422 (Red) and DNA (Blue).

(H) pT422 fluorescence intensity was normalized

to the total CENP-E fluorescence. Plots show the

mean of > 15 cells per condition from two indepen-

dent experiments. ***p < 0.0001 by t test. Error

bars represent SEM. See also Figure S1 and

Figure S2.


concentration of microtubules (Figure 2A). The maximal ATP

turnover rate (kcat) was not affected by Aurora A phosphorylation

(13 ± 0.6 s-1 for CENP-E1-473, n = 3; 14 ± 1.3 s-1 for CENP-E1-473

plus Aurora A, n = 3) (Figure 2A). However, the concentration of

microtubules required to reach the half maximal ATPase rate

(KmMT) was increased by > 3 fold following phosphorylation

(0.17 ± 0.03 mM for CENP-E1-473, n = 3; 0.64 ± 0.15 mM

CENP-E1-473 plus Aurora A, n = 3) (Figure 2A).

KmMT reflects CENP-E’s affinity for microtubules. In the

absence of microtubules, kinesins are tightly bound to ADP in

solution and the rate of ADP release is extremely low (Hackney,

1988). However, binding of ADP-bound kinesin to microtubules

greatly accelerates the rate of ADP release, and the kinesin

proceeds to complete its enzymatic cycle. Since phosphorylation

of CENP-E increased KmMT without significantly affecting kcat

and the gliding speed (data not shown), it is likely that the phos-

phorylation of T424 reduces CENP-E’s microtubule affinity

primarily in its ADP bound state without affecting the rate-limiting

step in CENP-E enzymatic cycle (Woehlke et al., 1997). To test

this hypothesis, the extent of Xenopus CENP-E1-473 binding to

microtubules was determined with or without prior phosphoryla-

tion by Aurora kinase (Figure 2B). Phosphorylation of WT

CENP-E1-473 by Aurora A reduced the amount of CENP-E that

cosedimented with microtubules by�50% with a corresponding

50% increase in apparent Kd(MT) (2 mM for CENP-E1-473; 3 mM for

CENP-E1-473 plus Aurora A). By contrast, Aurora A did not affect

microtubule binding of T424A CENP-E1-473 (apparent Kd(MT) of

3.5 mM T424A CENP-E1-473; 3.4 mM for T424A CENP-E1-473

plus Aurora A), confirming that phosphorylation at T424 reduces

the affinity of CENP-E for microtubules in the ADP state.

Duration of binding (sec)

1-cu

mul

ativ

e pr

obab

ility

1

0 25 50 75 100

0.1

0.01

0.001

BA ATPase assays

15

10

5

00 1 2 3 4 5

Microtubules (μM)

ATP

ase

rate

(s-1

)

C D

Microtubule pelleting assays

Duration of binding (sec)

Freq

uenc

y of

eve

nts

(%)

0 20 40 60 80 1000

10203040506070

3 mM MgADP, 33

112 sec

time

˚C

3 mM MgATP, 33

time

112 sec

2 μm

XCE1-473 XCE1-473 + Aurora A

XCE1-473 XCE1-473 + Aurora A

(2 mM MgADP)

E

17 sec12 sec

1-cu

mul

ativ

e pr

obab

ility

1

0.1

0.01

0.0010 2 4 6 8 10

Run length (μm)

Run length (μm)

Freq

uenc

y of

eve

nts

(%)

0 1 2 3 4 5 6 7 8

60

50

40

30

20

10

0

F

1.6 μm1.2 μm

Microtubule (μM):S P S P S P S P S P S P S P

XCE1-473MT

MTXCE1-473

Aurora A

T424A XCE1-473

MT

MTT424A XCE1-473

5210 0.125 0.25 0.5

Aurora A

2 μm

010203040506070

0 1 2 3 4

CE

NP

-E b

ound

(%)

T424A (Xenopus)

T424A + Aurora A

5

WT + Aurora A

WT

Microtubule (μM)

Mic

rotu

bule

s

XCE1-473

5 μm

G˚C

n = 3

XCE1-473

XCE1-473 + Aurora AXCE1-473

XCE1-473 + Aurora A

Figure 2. Phosphorylation of T424 of Xeno-

pus CENP-E by Aurora Kinase Reduces

CENP-E’s Affinity for Microtubules and

Reduces Its Run Length In Vitro

(A) ATPase rates of Xenopus CENP-E1-473 and

phosphorylated CENP-E1-473 measured with

increasing concentrations of microtubules. Plots

show the mean of three independent experiments

and error bars represent SEM. kcat and KmMT

values are represented ± SE.

(B) Equilibrium binding of WT or T424A CENP-E1-473

(incubated with or without Aurora A) to microtu-

bules in the ADP state. S, supernatant; P, pellet.

Percent of CENP-E1-473 bound is shown below

(n = 3; Error bars represent SD).

(C) Kymographs showing diffusive motion of

CENP-E1-473-RFP preincubated with or without

Aurora A in 3 mM MgADP at 33�C.

(D) Duration of binding (t) was distributed exponen-

tially, and the mean binding time (tmean) was

determined by fitting the data into a cumulative

distribution function (exp [-t/tmean]). Inset shows

1-cumulative probability of CENP-E1-473-RFP

binding time plotted on a log scale. The tmean is

17 ± 0.13 s for CENP-E1-473-RFP, n = 231 and

12 ± 0.07 s for CENP-E1-473-RFP plus Aurora A,

n = 240. Values represented ± SE of the curve fit.

Probability < 0.0001 that the duration of binding

for CENP-E1-473-RFP plus Aurora A was distrib-

uted the same as the duration of binding for

CENP-E1-473-RFP (by Kolmogorov-Smirnov Test).

(E) Kymographs showing processive motion of

CENP-E1-473-RFP in the presence of 3 mM MgATP

at 33�C. See Movie S1.

(F) Run length of CENP-E1-473-RFP was determined

by fitting the data into a cumulative distribution func-

tion. Inset shows 1-cumulative probability of CENP-

E1-473-RFP run length plotted on a log scale. Mean

run length is 1.6 ± 0.02 mm for CENP-E1-473-RFP,

n = 337 and 1.2 ± 0.02 mm for CENP-E1-473-RFP

plus Aurora A, n = 294. See Movie S1. Probability <

0.05 that the run lengths for CENP-E1-473-RFP

plus Aurora A were distributed the same as the

CENP-E1-473-RFP run lengths (by Kolmogorov-

SmirnovTest). Values represent± SE of the curve fit.

(G) Accumulated CENP-E (green) at the end of the

microtubules (red). See also Figure S3.


Phosphorylation Reduces CENP-E ProcessivityTotal Internal Reflection Fluorescence (TIRF) microscopy was

used to determine how Aurora phosphorylation affects properties

of individual CENP-E molecules. Xenopus CENP-E1-473 was

tagged with the monomeric, photostable red fluorescent protein

TagRFP-T (CENP-E1-473-RFP) (Shaner et al., 2008). Oregon

Green 488-labeled GMPCPP microtubules were tethered to

a coverslip in a flow chamber (Figure S3A) and CENP-E1-473-

RFP was added in the presence of apyrase to induce rigor

binding. As expected, CENP-E1-473-RFP was stably bound in

the absence of nucleotides, and fluorescence signals were

photobleached in one or two steps 89% of the time (75 double

steps and 68 single step; n = 160), consistent with a dimeric state

for the CENP-E1-473 motor (Figure S3B). When CENP-E1-473-RFP

was introduced into the flow chamber in a buffer containing ADP,

both phosphorylated and unphosphorylated CENP-E1-473-RFP

remained loosely bound to microtubules without displaying

directional motility (Figure 2C), supporting our previous observa-

tion that CENP-E motility contains a diffusive mode that does not

Tubulin

CENP-E

55

275

kDa100 50 30 20 10 100

Control siRNA C-EsiRNA

Dilution (%):

275CENP-E

Tubulin 55

275Myc

CENP-E siRNA: kDa+- +- +-

WT

Rigor

+-

10A

T422A

+-+-

9AParenta

l

+ -+- +- +- +-+-Control siRNA:

MycGFP-CENP-EC D

T422 S454 S611 S1211 T1268 S2509S2512

S2555S2558

S2543

Myc GFP

MycGFPtag

Motordomain

MTbinding

KTtargeting

A

Plate cells TransfectsiRNA

Replatecells

0 24 48 60Time (h): 12 36 72 80

Add TetBegin filming/

harvest proteinfor immunoblot

CENP-E siRNA targetedagainst the 3'UTR ofendogenous CENP-E

siRNA resistantCENP-E transgene

induced with Tet

B

E

MycGFP-CENP-E

n>90

F

DN

AG

FP/T

ubul

in

T422AWT

Discontinuous coild-coil

Tim

e in

mito

sis

(min

)

Parental WT Rigor 10A 9A T422A

Figure 3. Phosphorylation of CENP-E at

T422 Is Required for Chromosome Align-

ment

(A) Diagram of MycLAP-CENP-E transgene with

previously identified phosphorylation sites. Splice

variations exist within the CENP-E coiled-coil

domain and phosphorylation sites are numbered

with respect to their position in the CENP-E clone

used in this study.

(B) Schematic of replacing endogenous CENP-E

with a siRNA resistant transgene.

(C) Immunoblot showing depletion of CENP-E by

siRNA.

(D) Immunoblot showing knockdown with control

(GAPDH) or CENP-E siRNA in cells expressing

various MycLAP-CENP-E transgenes.

(E) Box and whisker plots showing the time spent in

mitosis for cells expressing MycLAP-CENP-E

following transfection of control (GAPDH, green) or

CENP-E (red) siRNA. >90 cells per condition are

plotted fromat least three independent experiments.

(F) Immunofluorescence images of cells in which

endogenous CENP-E has been replaced with WT

or T422A MycGFP-CENP-E. GFP (green); Tubulin

(red). See alsoFigure S4, Figure S5, FigureS6, Movie

S2, and Movie S3.

require ATP hydrolysis (Kim et al., 2008).

Following phosphorylation, the duration

of CENP-E1-473-RFP binding to microtu-

bules was shortened by �30% in the

presence of ADP (t = 17 ± 0.13 s for

CENP-E1-473-RFP, n = 231; t = 12 ± 0.07 s

for CENP-E1-473-RFP plus Aurora A,

n = 240) (Figure 2D), consistent with the

observation that phosphorylation of

T424 reduces CENP-E’s affinity to micro-

tubules in the ADP bound state.

Processivity of CENP-E in the presence

of ATP was reduced after phosphoryla-

tion on T424, with run lengths of phosphorylated CENP-E1-473-

RFP on individual microtubules 25% shorter than those of the

unphosphorylated motor (1.6 ± 0.02 mm for CENP-E1-473-RFP,

n = 337; 1.2 ± 0.02 mm for CENP-E1-473-RFP plus Aurora A,

n = 294) (Figures 2E and 2F). Importantly, once reaching a micro-

tubule end using its plus end-directed motility, individual

CENP-E dimers did not immediately dissociate (Figure 2G), but

remained bound there for 5.8 s (±0.8 SEM, n = 26), a feature

previously observed for several other processive kinesins

(Case et al., 1997; Okada and Hirokawa, 1999; Varga et al.,

2006).

Phosphorylation of CENP-E T422 Is Requiredfor Chromosome CongressionCENP-E is phosphorylated during mitosis on at least ten sites

(Figure 3A) (Nousiainen et al., 2006; Zecevic et al., 1998), albeit

the significance of these phosphorylations has not been tested.

To determine the consequence of preventing CENP-E phos-

phorylation in human cells, we developed a strategy to replace


endogenous CENP-E with phosphorylation defective transgenes

(Figure 3B). Full-length CENP-E fused at the N-terminus to

a MycGFP epitope tag (Figure 3A) was integrated at a predefined

genomic locus in DLD-1 cells using FRT/Flp-mediated recombi-

nation and expression was induced by addition of tetracycline

(Figure S4A). Time-lapse microcopy revealed that the subcellular

distribution of WT MycGFP-CENP-E closely mirrored that of

endogenous CENP-E, localizing to kinetochores after nuclear

envelope breakdown and relocating to the spindle midzone in

anaphase and to the midbody during cytokinesis (Figure S5A).

Transfection of siRNA targeting the 30 untranslated region of

CENP-E mRNA depleted endogenous CENP-E by�90% across

the population, yielding it undetectable at the kinetochores of

most mitotic cells (Figure 3C and data not shown). As expected,

depletion of CENP-E extended the average duration of mitosis

(to 221 min) compared to control transfected cells (71 min)

(Figures 3D and 3E). Importantly, this delay was largely rescued

by the expression of MycGFP-CENP-E (producing an average of

100 min in mitosis). Replacing endogenous CENP-E with a rigor

mutant (T93N) (Figure 3D) strongly exacerbated the mitotic delay

(to 506 min on average) with a few chromosomes chronically

misaligned near the spindle poles (Figure 3E), confirming our

previous finding that the motor activity of CENP-E is essential

for metaphase chromosome alignment (Kim et al., 2008).

Replacement of endogenous CENP-E with a variant (10A) with

all 10 phosphorylation sites abolished (by substitution to

alanines) produced a robust mitotic delay (505 min on average).

On the other hand, abolishing phosphorylation of the nine sites

other than T422 (9A) (Figure 3D) had little effect on mitotic

progression (114 min on average in mitosis) (Figure 3E). Surpris-

ingly, preventing phosphorylation of T422 alone was sufficient to

produce a substantial mitotic delay (425 min on average),

demonstrating that of these 10 CENP-E phosphorylation sites,

phosphorylation at T422 makes the largest contribution to timely

mitotic progression.

Replacing endogenous CENP-E with the T422A mutant

prevented complete metaphase chromosome alignment, with

a few chromosomes remaining close to the spindle poles in

> 85% of cells (Figure 3F, Figures S4B–S4D, and Figure S5B),

a phenotype highly reminiscent of that observed with diminished

levels of CENP-E (Weaver et al., 2003). Phosphorylation of T422

was not required for the kinetochore recruitment of CENP-E

(Figures S6A and S6B). To eliminate the possibility that mutation

of T422 caused defects other than simply preventing phosphor-

ylation, we created an additional CENP-E phosphodeficient

mutant, in which two arginines in the Aurora consensus motif

were converted to lysines (RR:KK) (Figure S6C). Mutation of

RR:KK did not abolish the epitope of the pT422 antibody (data

not shown). However, recombinant Xenopus CENP-E1-428

carrying the RR:KK mutation could not be efficiently phosphory-

lated by Aurora A and B in vitro (Figure S6D and data not shown)

and the RR:KK mutant was not phosphorylated on T422 in

human cells (Figure S6E). Indeed, replacing endogenous

CENP-E with the RR:KK mutant caused a mitotic delay (Fig-

ure S6F) similar to that observed with the T422A mutant with

a few chromosomes remaining close to the spindle poles, con-

firming that phosphorylation of CENP-E at T422 is required for

chromosome congression.

CENP-E Phosphorylation by Aurora Is Requiredfor Congression of Polar ChromosomesCENP-E has been implicated in powering chromosome congres-

sion by transporting mono-oriented chromosomes to the spindle

equator along mature kinetochore fibers of already bioriented

chromosomes (Kapoor et al., 2006). To test whether phosphory-

lation of T422 is required for this process, we adopted a method

to enrich mono-oriented, polar chromosomes (Kapoor et al.,

2006; Lampson et al., 2004) in cells in which endogenous

CENP-E was replaced with the WT or T422A MycLAP-CENP-E

(Figures 4A and 4B). Cells were first treated with monastrol

to generate monopolar spindles with a high frequency of synteli-

cally-attached chromosomes and released from monastrol in

the presence of an Aurora kinase inhibitor (ZM) to allow bipolar

spindles to form while preserving improper kinetochore attach-

ments. Following the removal of ZM, congression of mal-

oriented chromosomes was assessed (Figures 4A and 4B). As

a control, cells were treated in parallel with DMSO to determine

the extent of chromosome misalignment in an unperturbed

mitosis. The enrichment of improper kinetochore attachments

significantly increased the number of polar chromosomes in cells

defective in T422 phosphorylation, but not in cells expressing

WT CENP-E (Figures 4C–4F). Live cell imaging demonstrated

that, following reactivation of the Aurora kinases, improperly

attached chromosomes were frequently moved to either spindle

pole in cells expressing WT or T422A CENP-E (Figure S7A–S7C).

However, these chromosomes remained closely associated with

those poles in cells expressing T422A CENP-E (Figure S7C),

establishing that phosphorylation of CENP-E on T422 by Aurora

kinases is required for the congression of polar chromosomes.

Aurora-Mediated Phosphorylation of CENP-E T422Opposes Direct Binding of CENP-E to the CatalyticSubunit of PP1Following CENP-E T422 is a highly conserved tryptophan, thereby

producing a RRVTW sequence that conforms to the docking motif

for protein phosphatase 1 (PP1) (Figure 5A) (Hendrickxetal., 2009).

Indeed, our mass spectrometry analysis of tandem affinity purified

CENP-E from mitotic human cells identified the catalytic subunit of

PP1 to be associated with CENP-E (data not shown) and PP1 was

also present in CENP-E immunoprecipitates from nocodazole-

arrested DLD-1 cells (Figure 5D). The interaction between CENP-E

and PP1 is direct, as recombinant CENP-E motor (CENP-E1-473)

was recovered together with PP1g in a pulldown experiment using

Microcystin-beads (Figure 5B). Recovery of a stoichiometric (1:1)

complex of CENP-E and PP1 required addition of > 5 molar excess

of CENP-E over PP1, indicating a weak affinity between CENP-E

and PP1. Further, CENP-E with a W425A substitution had mark-

edly reduced binding to PP1 (Figure 5B), demonstrating that the

interaction between CENP-E and PP1 is mediated through the

PP1 docking motif. To test whether phosphorylated T422 is

a substrate for PP1, phosphorylated CENP-E1-473 was incubated

with either PP1g or PP1g preinactivated with the inhibitor Micro-

cystin (Figure 5C). Monitoring of CENP-E’s phosphorylation status

with the pT422 antibody revealed that PP1g rapidly dephosphory-

lated CENP-E T422 (Figure 5C).

Previous reports have shown that phosphorylation of serine or

threonine overlapping the PP1 docking motif impairs the binding


to PP1 (Hubbard and Cohen, 1989). Given that CENP-E T422 is

overlapped by a consensus motif for Aurora kinases and

a conserved motif for PP1 binding, we tested whether Aurora

phosphorylation at T422 disrupts PP1’s binding to CENP-E.

Following in vivo inhibition of T422 phosphorylation with the

pan Aurora inhibitor VX-680, the amount of PP1 associated

with CENP-E was dramatically increased (Figure 5D). Moreover,

phosphorylation of CENP-E1-473 by Aurora A resulted in a

�10-fold reduction in the binding of CENP-E to the catalytically

inactive (D64N) PP1g in vitro (Figure 5E), demonstrating that

Aurora-mediated phosphorylation of CENP-E T422 opposes

direct binding of CENP-E to PP1.

Dephosphorylation of CENP-E T422 by PP1 Is Requiredfor Stable Biorientation of the ChromosomesCongressed by CENP-EThe pT422 antibody inhibited PP1-mediated dephosphorylation

of Xenopus CENP-E1-473 at T424 (the position homologous to

T422 in human CENP-E) in vitro (Figure 6A). Thus, to test the

DMS O MGMon ZM+MG MG

WT T422A

CENP-E siRNA

MycGFP-CENP-E

DMS O MGMon ZM+MG MG

WT T422A

CENP-E siRNA

MycGFP-CENP-E

***

Mon ZM+MG MG

WT T422A

DNAKNL1

DNAKNL1

DMSO MG

WT T422A

DNAKNL1

DNAKNL1

C

Monast rol2 hr

ZM + MG1321 hr

MG1321.5 hr

Fix for IF

DMSO3 hr

MG1321.5 hr

DMSO MG

Mon ZM+MG MG

ZM + MG132Monast rol MG132

?

?

A

B

D

E F

Num

ber o

f pol

ar c

hrom

osom

es

Cel

ls w

ith p

olar

chr

omos

omes

(%)

100

80

60

40

20

0

Figure 4. CENP-E Phosphorylation by

Aurora Is Required for Congression of Polar

Chromosomes

(A) Drug treatment scheme to enrich mono-

oriented chromosomes near spindle poles.

(B) Diagram illustrating attachment status of

chromosomes following treatment with drug

regime in (A).

(C and D) Immunofluorescence images of drug-

treated cells in which endogenous CENP-E was

replaced with either WT or T422A MycLAP-

CENP-E. KNL-1 (green); DNA (red).

(E) Percentage of cells from (C) and (D) with polar

chromosomes.

(F) Number of polar chromosomes in cells from (C)

and (D) displaying chromosome misalignment.

Bars represent the mean of four independent

experiments. ***p < 0.0001 by t test. Error bars

represent SEM. See also Figure S7 and Movie S4.

in vivo significance of the dephosphoryla-

tion of CENP-E T422 by PP1, we microin-

jected rhodamine-labeled pT422 anti-

bodies into HeLa cells stably expressing

histone H2B-YFP. Consistent with our

immunofluorescence analysis (Figure 1F),

the microinjected rhodamine-labeled

pT422 antibody was virtually absent

from aligned kinetochores, but accumu-

lated to high levels at the kinetochores of

chromosomes positioned close to the

spindle poles (Figure 6B). Microinjection

of the pT422 antibody substantially de-

layed the duration of mitosis compared

to control injected cells (average 125 min

for rhodamine-labeled rabbit IgG injected

cells; 619 min for rhodamine-labeled

pT422 injected cells) (Figures 6C–6E).

Interestingly, antibody-mediated preser-

vation of phosphorylation on CENP-E T422 promoted dynamic

chromosome movements distinct from the chromosome

behaviors observed when T422 phosphorylation is abolished

(Figure S5B). Polar chromosomes congressed to the equator of

the cell, but most failed to make stable microtubule attachments

and fell back out of the spindle equator or continued to move

forward to the other pole (Figure 6C). Consistently, the microin-

jected pT422 antibody remained enriched on the kinetochores

of chromosomes juxtaposed to the metaphase plate that did

not form stable microtubule attachments (Figure 6B). Thus,

despite CENP-E-mediated congression of chromosomes to the

proximity of the spindle equator, stable kinetochore attachment

(and subsequent biorientation) does not occur when dephos-

phorylation of CENP-E by PP1 is blocked.

DISCUSSION

Here, we demonstrate that phosphorylation by Aurora kinases

of a conserved residue (T422) close to the CENP-E motor


domain is essential to promote the congression of polar chro-

mosomes and dephosphorylation of this site is required for

the stable biorientation of these kinetochores. Aurora-medi-

ated phosphorylation of this site regulates the intrinsic motor

properties of CENP-E and disrupts the binding of the

opposing phosphatase PP1 to CENP-E, thereby establishing

a bistable phosphoswitch for regulation of CENP-E (see sche-

matic in Figure 7).

The Aurora phosphorylation site on CENP-E is adjacent to its

coiled-coil neck, next to several conserved positively charged

amino acids. Phosphorylation at T422 diminishes the basic

charge of what we propose to be an electrostatic tether directly

involved in microtubule binding (analogous to similar domains

identified in other kinesin family members [Okada and Hirokawa,

1999; Ovechkina et al., 2002; Thorn et al., 2000]). Consistently,

phosphorylation at T422 reduces CENP-E’s affinity for microtu-

bules and allows the motor to dissociate more readily during

processive runs. Phosphorylation of CENP-E 422 is highest on

the kinetochores close to the spindle poles. Since Aurora A is

concentrated at the poles, it is likely to be responsible for phos-

phorylation of T422 on such polar-oriented chromosomes.

Aurora phosphorylation reduces the proportion of time that

each motor molecule is bound unproductively to the many

dynamic astral microtubules nucleated near the pole (Figure 7B).

Phosphorylation-dependent reduction in CENP-E residence

time on an individual microtubule of a kinetochore fiber, on the

other hand, will be of little consequence, as rapid rebinding to

an adjacent microtubule is likely, given the high local concentra-

tion of parallel microtubules that comprise the fiber (Figure 7C).

E

A B

KRKRRVTWCLKRKRRVTWCYRRRRRVTWAP

KRKRRVTWAP

- - KRRRTWCP

425

HumanMouse

Chicken

Xenopus

Drosophila

420

PP1 docking motif

Auroraphosphorylation

C D

Myc

pT422

PP1

275

275

IP: IgG Myc

DMSO

VX-6

80

MycGFP-CENP-E

kDa

36

PP1γ

CENP-E1-473

Input (1/10)

1x 5x2x- --

1x 5x2x+-

- --

PP1-pulldown

1x 5x2x 5x+ + + -PP1γ

1x 5x2x 5xCENP-E1-473WT W425A WT W425A

+ + + -

40

57

kDa

+ KD AurA

Input (1/10)

1x 5x2x- --

1x 5x2x+-

- --

PP1-pulldown

1x 5x2x 5x+ + + - D64N PP1γ

1x 5x2x 5x CENP-E1-473+ AurA

+ + + -

+ KD AurA + AurA

D64N PP1γ

CENP-E1-473

Aurora AKD Aurora A

CENP-E1-473Aurora A

30 minp-CENP-E1-473

PP1 PP1 + Microcystin

35 40 45 60

Coomassie

t = 0 min t = 30 min

35 40 45 600 30

Aurora A

CENP-E1-473

Time (min):

57 kDapT422

Figure 5. Aurora Phosphorylationof CENP-E

T422 Opposes Direct Binding of CENP-E to

PP1

(A) CENP-E T422 overlaps the conserved PP1

docking motif (R/K-R/K-V-X-F/W).

(B) Coomassie staining of the PP1-bound

complexes purified with Microcystin-agarose. 1,

2, or 5 molar excess of WT or W425A Xenopus

CENP-E1-473 was incubated with Xenopus PP1g.

(C) In vitro phosphatase assay using PP1g (with or

without Microcystin) and phosphorylated Xenopus

CENP-E1-473 as substrate. Immunoblot shows

phosphorylation of T422 and coomassie stain

shows purified proteins.

(D) Nocodazole-arrested DLD-1 cells expressing

MycGFP-CENP-E were treated with VX-680 and

MG132 before harvesting. CENP-E immunoprecip-

itates were blotted for pT422 and PP1.

(E) Coomassie staining of the PP1-bound

complexes purified with Microcystin-agarose. 1, 2

or 5 molar excess of WT CENP-E1-473 was preincu-

bated with either WT or kinase-dead Aurora A

before incubating with catalytically inactive

(D64N) PP1g.

Thus, Aurora-mediated destabilization of

CENP-E tethering to individual spindle

microtubules yields a variant of kinetic

proofreading (Hopfield, 1974), with local,

destabilized attachment as a means to

eliminate incorrect initial attachments, while allowing productive

CENP-E-powered movement along a kinetochore microtubule

bundle.

A requirement for Aurora A in modulating CENP-E offers

a mechanistic explanation for prior reports that Aurora A inhibi-

tion causes chromosome misalignment with a few chromo-

somes found close to the spindle poles (Hoar et al., 2007;

Kunitoku et al., 2003; Marumoto et al., 2003). Although Aurora

A-mediated phosphorylation of the centromere-specific histone

H3 variant CENP-A has previously been proposed to promote

chromosome congression (Kunitoku et al., 2003), we conclude

that CENP-E is the kinetochore substrate whose Aurora

A-dependent phosphorylation is directly required for chromo-

some congression. For Aurora B, the absence of tension exerted

on mono-oriented polar kinetochores and the juxtaposed posi-

tion of sister kinetochores on syntelically attached chromo-

somes would bring it in close proximity to the highly elongated

and flexible CENP-E, allowing Aurora B phosphorylation to

modulate processivity of CENP-E attached to kinetochores

with reduced tension (Figures 7A and 7B). Further, Aurora

B-dependent phosphorylation in and around the inner centro-

meres of sister kinetochores would also be expected to prefer-

entially destabilize any incorrect attachments made by the

230 nm long CENP-E to microtubules that reach across the

inter-kinetochore space.

Recent evidence has demonstrated that KNL1, one of the core

microtubule binding components thought to be responsible for

end-on attachment at metazoan kinetochores (Cheeseman

et al., 2006), binds PP1 on chromosomes aligned at metaphase.


Binding is through a motif for PP1 docking with an overlapping

Aurora phosphorylation site (Liu et al., 2010), a situation similar

to what we now report for CENP-E. Thus, the vertebrate kineto-

chore has evolved multiple modules for recruiting PP1, with

recruitment by KNL1 and CENP-E each providing different func-

tions. Blocking KNL1 recruitment of PP1 increased the number

of kinetochores without cold stable microtubules and decreased

the level of PP1 recruited to kinetochores. Nevertheless, it did

not affect congression or chromosome alignment, but did lead

to an unexplained inhibition of cell growth (Liu et al., 2010). In

contrast, we have now shown that once CENP-E tows initially

misoriented chromosomes to the cell center, its subsequent

dephosphorylation and rebinding of PP1 is essential for stable

microtubule attachment to the kinetochores on these chromo-

somes (Figure 7D). Thus, we propose a model in which CENP-E

powers chromosome movement away from the high Aurora

activity at poles and then exploits its flexible coiled-coil and

plus end-directed motility to deliver PP1 phosphatase activity

within its �230 nm reach at the outer kinetochore (Figures 7E

and 7F). For the kinetochores on these chromosomes, our

0:00 1:00 1:15 1:20 2:25

Inje

ct:

Inje

ct:

Rab

bitI

gG-R

hod 0:100:05

Histone H2B-YFP

0:00

4

C

44

33

3

11 1

4:50 5:053:20 3:50 4:00

5:20 7:35 8:508:457:25 8:15

2 2 2

Pro

ject

ion

45°

rota

tion

Histone H2B-YFP

B

CENP-E1-473Aurora A

t = 30 minp-CENP-E1-473

0Time (min): 65 70 7530

+pT422antibody

+Rabbit IgG

p-CENP-E1-473

PP1

60 90 65 70 7560 90

t = 0 min

t = 60 min

Boil and probe with

p-CENP-E1-473

PP1

49kD

a

A

ED

Non-Inj Inj Non-inj Injo Non-Inj Inj Non-Inj Inj

pT422-Rhod Ab injected cellspT422-Rhod Ab

pT422-Rhod Ab

pT42

2-R

hod

Ab

pT422-Rhod

pT422-Rhod

Myc

Ab injected: Rabbit IgG-Rhod pT422-Rhod Ab injected: Rabbit IgG-Rhod pT422-Rhod

Tim

e in

mito

sis

(min

)

Abn

orm

al m

itosi

s (%

) 100

80

60

40

20

0

1000

750

500

250

0

Figure 6. Dephosphorylation of CENP-E

T422 by PP1 Is Required for the Stable Bio-

rientation of Chromosomes Congressed

from a Spindle Pole by CENP-E

(A) Phosphorylated Xenopus CENP-E1-473 (1 mM)

was preincubated with rabbit IgG or the pT422

antibody (3 mM) and subsequently mixed with

PP1g (0.2 mM). Phosphorylation of T422 was

determined using rhodamine-labeled pT422 anti-

body and visualized using the rhodamine fluoro-

phore. Anti-Myc immunoblot shows the CENP-

E1-473 loading.

(B) Reconstructed Z-sections of rhodamine-

labeled pT422 antibody-injected live HeLa cells

expressing Histone H2B-YFP. Histone H2B-YFP

(purple); pT422-Rhod antibody (green).

(C) Time-lapse images of antibody-microinjected

(green) HeLa cells stably expressing Histone

H2B-YFP (purple). Numbered arrows track the

movement of four individual chromosomes which

congress to the equator, but fail to stably biorient.

Stills are taken from Movie S5 and Movie S6.

(D) Box and whisker plots showing the time spent

in mitosis for microinjected cells. Uninjected cells

in the same field of view were also examined. For

each antibody, >120 microinjected cells were

observed from two independent experiments. (E)

Bar graphs showing the percentage of abnormal

mitosis for the cells observed in (D).

evidence implicates dephosphorylation

of the core microtubule-binding proteins

(e.g., KNL1 and the four member Ndc80

complex) by CENP-E-bound PP1 as an

essential step in reversing their prior inac-

tivation by Aurora-dependent phosphor-

ylation.

Finally, the spatial regulation of CENP-

E by Aurora kinases and PP1 may provide

an insight into the classic observation that

phosphorylation controls the directionality of two opposing

kinetochore motors on isolated chromosomes (Hyman and

Mitchison, 1991). To coordinate prometaphase chromosome

movement, this phosphorylation-dependent switch must turn

off the minus end-directed motor and turn on the plus end-

directed motor at the spindle poles. Here, we have shown that

the plus-end directed motor properties of CENP-E (and binding

of PP1) are altered by a gradient of Aurora kinase activity

emanating from the spindle poles. This provides spatial informa-

tion within the mitotic spindle to regulate CENP-E activity

according to the position of chromosome.


Constructs

The full-length human CENP-E open reading frame was cloned into a pcDNA5/

FRT/TO based vector (Invitrogen) modified to contain an amino-terminus

Myc-LAP epitope tag. The LAP tag consists of GFP-TEV-S-peptide as previ-

ously described (Cheeseman et al., 2004). TagRFP-T (a kind gift from Roger

Tsien, UCSD) was cloned into pET23d vector (Novagen) containing Xenopus

CENP-E (aa1-473). This cloning strategy generates a 16 amino acid-long linker


(MASMTGGGGMGRLR) between CENP-E and TagRFP-T. All human and

Xenopus CENP-E mutants were generated by site-directed mutagenesis

(QuickChange, Stratagene).

Cell Culture and siRNA Treatment

Cells were maintained at 37�C in a 5% CO2 atmosphere in Dulbecco’s

modified eagle medium (DMEM) containing 10% tetracycline free fetal bovine

serum (Clontech), 100 U/ml penicillin, 100 U/ml streptomycin and 2 mM L-

glutamine. For siRNA treatment, 1.5 3 105 cells were plated in a 6-well plate

and duplexed siRNAs were introduced using Oligofectamine (Invitrogen).

siRNAs directed against CENP-E (50-CCACUAGAGUUGAAAGAUA-30) and

GAPDH (50-UGGUUUACAUGAUCCAAUA-30) were purchased from Dharma-

con. Cells were processed for immunofluorescence microscopy or live cell

imaging 48 hr after transfection.

Generation of Stable Cell Lines

Stable DLD-1, H2B-RFP cell lines expressing CENP-E were generated using

the FRT/Flp-mediated recombination as described previously (Holland et al.,

2010). Small molecules were used at the following final concentrations: noco-

dazole, 0.2 mg/ml; taxol, 10 mM; monastrol, 20 mM; S-Trityl-L-cysteine, 5 mM;

MG132, 20 mM; ZM447439 (Tocris Bioscience), 3 mM; VX-680 (Selleck)

0.5 mM and MLN8054 (a kind gift of Patrick Eyers), 0.25 mM. All small molecules

were from Sigma-Aldrich unless otherwise specified.

Immunofluorescence Microscopy

Cells were pre-extracted for 90 s in MTSB (100 mM PIPES [pH 6.8], 0.1%

Triton X-100, 0.1 mM CaCl2, 1 mM MgCl2) and fixed in 2% formaldehyde in

MTSB. Cells were blocked in 2.5% FBS, 0.2 M glycine, 0.1% Triton X-100 in

PBS for 1 hr. For the pT422 staining, cells were extracted and fixed in the pres-

ence of 500 nM Microcystin-LR (EMD). Antibody incubations were conducted

in blocking solution for 1 hr. DNA was detected using DAPI and cells were

mounted in ProLong (Invitrogen). Images were collected using a DeltaVision

Core system (Applied Precision) controlling an interline charge-coupled device

camera (Cooldsnap, Raper). Kinetochore signal intensity was determined

using MetaMorph (Molecular Devices), by measuring integrated fluorescence

intensity with a 10 3 10 pixel square. Background signal was subtracted

from an area adjacent to the kinetochore. The mean integrated fluorescence

intensity of at least 10 kinetochore pairs per cell was calculated. Antibodies

used are specified in the Extended Experimental Procedures.

Single Molecule Assays

CENP-E single molecule assays were performed as previously described (Kim

et al., 2008) with the following modifications. Slides and 22 3 22-mm square

+

+

-

+

- +

+

+

-

+

+

-

Aurora Bactivity

+-

+

A

B

C

D

E

F

Aurora Aactivity

Kinetochore

+

+

Kinetochore

+

+

+

--

K-fiber

PP1P

PP

P

P

PP

Dynein-dependent poleward movementalong a laterally attached microtubule;

microtubule binding by KMN networkis inhibited by phosphorylation

Aurora A phosphorylation of CENP-Eat pole releases PP1

Processive tracking of p-CENP-E onlyalong a K-fiber

PP1 bound to CENP-E dephosphorylates Ndc80 and KNL1, enabling their end-on

microtubule attachment

PP1

P

P

P

P

P

P

PP

K-fiber

-

+

+

-

P

K-fiber+

-

-

-

+

ATP, Aurora A

PP1

PP1

P

P

P

P

P

P

P

P

+ -

PP1

PP1

PP1PP1

PP1

P

P

P

P

P

+ -

PP1

PP1

Cytoplasmic dynein

Centromere protein E

Centrioles

PP1

Protein phosphatase 1

Mis12complex KNL1

Ndc80 complex

KMN network

Phosphate

P

PP1-mediated dephosphorylation ofCENP-E away from polar Aurora A

Microtubules

Stable K-fiber attachment by PP1-boundKNL1/ Mis12 and Ndc80 complex

-

+

Phosphorylation by Aurora B

PP1

PP1

P

P P

P

P

P

P

Figure 7. A Model of CENP-E Regulation by Aurora Kinases and PP1

(A) Unattached chromosomes are translocated poleward along a single astral microtubule in a dynein-dependent manner (Rieder and Alexander, 1990).

(B) Aurora A phosphorylates CENP-E at T422 near the spindle poles, releasing PP1 from CENP-E. Phosphorylation of CENP-E provides selectivity toward

bundles of kinetochore microtubules (K-fiber) for CENP-E to glide along.

(C) Our evidence here demonstrates that CENP-E powers the congression of polar chromosomes to the spindle equator along the K-fiber of an already bioriented

chromosome, as earlier proposed (Kapoor et al., 2006).

(D) As chromosomes congress, kinetochores move away from the Aurora A gradient concentrated at the spindle poles and CENP-E is dephosphorylated.

(E) Dephosphorylated CENP-E recruits a high local concentration of PP1 to the outer kinetochores of chromosomes it has translocated away from a pole. CENP-E

delivered PP1 is essential for stable, microtubule capture by the kinetochores of these towed chromosomes through its dephosphorylation of key components at

the kinetochore-microtubule interface (e.g., Ndc80 and KNL1), thereby increasing their affinity for microtubules.

(F) Upon biorientation, kinetochore substrates become separated from the inner centromeric Aurora B. Kinetochore-recruited PP1 through its direct binding to

KNL1 and CENP-E stabilizes kinetochore-microtubule interactions.


coverslips were silanized as described (Helenius et al., 2006). A flow chamber

was incubated with 50 mg/ml of a rat monoclonal anti-tubulin antibody (YL1/2,

Serotec) for 5 min, followed by 1% Pluronic F-127 (Invitrogen) in BRB80 for

15 min and Oregon Green 488-labeled GMPCPP microtubules for 10 min.

�0.2 mg/ml of Xenopus CENP-E1-473-RFP was incubated with 50 mg/ml of

Aurora A in 20 mM Tris (pH 7.5), 25 mM KCl, 1 mM MgCl2, 1mM DTT,

0.1 mM MgATP for 15 min at room temperature and diluted to�0.5 nM before

imaging in motility buffer (25 mM K-PIPES, [pH 6.9], 5 mM MgCl2, 1 mM EGTA,

1 mM DTT, 0.25% Brij35, 0.5 mg/ml casein, 4.5 mg/ml glucose, 0.2 mg/ml

glucose oxidase, 0.35 mg/ml catalase, 0.5% bME) containing either 3 mM

MgATP or 3 mM MgADP. Frames were captured every 500 ms with 200 ms

exposure, and the typical duration of imaging was 2–3 min. Note, that since

imaging was performed at an elevated temperature (33�C) and in higher

MgCl2, the speed of CENP-E movement was faster than that measured at

room temperature in our previous study (Kim et al., 2008).


Supplemental Information includes Extended Experimental Procedures,

Supplemental References, seven figures, and six movies and can be found

with this article online at doi:10.1016/j.cell.2010.06.039.

ACKNOWLEDGMENTS

The authors would like to thank Stephen Taylor, Patrick Eyers, Todd Stuken-

berg, and Arshad Desai for providing reagents. This work was supported by

a grant (GM29513) from the National Institutes of Health to D.W.C., who

receives salary support from the Ludwig Institute for Cancer Research.

A.J.H. is supported by a European Molecular Biology Organization (EMBO)

Long-Term Fellowship.





REFERENCES

Akiyoshi, B., Nelson, C.R., Ranish, J.A., and Biggins, S. (2009). Quantitative

proteomic analysis of purified yeast kinetochores identifies a PP1 regulatory

subunit. Genes Dev. 23, 2887–2899.

Berger, B., Wilson, D.B., Wolf, E., Tonchev, T., Milla, M., and Kim, P.S. (1995).

Predicting coiled coils by use of pairwise residue correlations. Proc. Natl.

Acad. Sci. USA 92, 8259–8263.

Case, R.B., Pierce, D.W., Hom-Booher, N., Hart, C.L., and Vale, R.D. (1997).

The Directional Preference of Kinesin Motors Is Specified by an Element

outside of the Motor Catalytic Domain. Cell 90, 959–966.

Cheeseman, I.M., Anderson, S., Jwa, M., Green, E.M., Kang, J.-s., Yates, J.R.,

Chan, C.S.M., Drubin, D.G., and Barnes, G. (2002). Phosphoregulation of

Kinetochore-Microtubule Attachments by the Aurora Kinase Ipl1p. Cell 111,

163–172.

Cheeseman, I.M., Niessen, S., Anderson, S., Hyndman, F., Yates, J.R.,

Oegema, K., and Desai, A. (2004). A conserved protein network controls

assembly of the outer kinetochore and its ability to sustain tension. Genes

Dev. 18, 2255–2268.

Cheeseman, I.M., Chappie, J.S., Wilson-Kubalek, E.M., and Desai, A. (2006).

The Conserved KMN Network Constitutes the Core Microtubule-Binding

Site of the Kinetochore. Cell 127, 983–997.

Cimini, D., and Degrassi, F. (2005). Aneuploidy: a matter of bad connections.

Trends Cell Biol. 15, 442–451.

Cimini, D., Wan, X., Hirel, C.B., and Salmon, E.D. (2006). Aurora Kinase

Promotes Turnover of Kinetochore Microtubules to Reduce Chromosome

Segregation Errors. Curr. Biol. 16, 1711–1718.

Cleveland, D.W., Mao, Y., and Sullivan, K.F. (2003). Centromeres and

Kinetochores: From Epigenetics to Mitotic Checkpoint Signaling. Cell 112,

407–421.

Ditchfield, C., Johnson, V.L., Tighe, A., Ellston, R., Haworth, C., Johnson, T.,

Mortlock, A., Keen, N., and Taylor, S.S. (2003). Aurora B couples chromosome

alignment with anaphase by targeting BubR1, Mad2, and Cenp-E to kineto-

chores. J. Cell Biol. 161, 267–280.

Emanuele, M.J., Lan, W., Jwa, M., Miller, S.A., Chan, C.S.M., and Stukenberg,

P.T. (2008). Aurora B kinase and protein phosphatase 1 have opposing roles in

modulating kinetochore assembly. J. Cell Biol. 181, 241–254.

Espeut, J., Gaussen, A., Bieling, P., Morin, V., Prieto, S., Fesquet, D., Surrey,

T., and Abrieu, A. (2008). Phosphorylation Relieves Autoinhibition of the

Kinetochore Motor Cenp-E. Mol. Cell 29, 637–643.

Francisco, L., Wang, W., and Chan, C.S. (1994). Type 1 protein phosphatase

acts in opposition to IpL1 protein kinase in regulating yeast chromosome

segregation. Mol. Cell. Biol. 14, 4731–4740.

Gestaut, D.R., Graczyk, B., Cooper, J., Widlund, P.O., Zelter, A., Wordeman,

L., Asbury, C.L., and Davis, T.N. (2008). Phosphoregulation and depolymeriza-

tion-driven movement of the Dam1 complex do not require ring formation. Nat.

Cell Biol. 10, 407–414.

Hackney, D.D. (1988). Kinesin ATPase: rate-limiting ADP release. Proc. Natl.

Acad. Sci. USA 85, 6314–6318.

Harrington, E.A., Bebbington, D., Moore, J., Rasmussen, R.K., Ajose-

Adeogun, A.O., Nakayama, T., Graham, J.A., Demur, C., Hercend, T., Diu-

Hercend, A., et al. (2004). VX-680, a potent and selective small-molecule

inhibitor of the Aurora kinases, suppresses tumor growth in vivo. Nat. Med.

10, 262–267.

Helenius, J., Brouhard, G., Kalaidzidis, Y., Diez, S., and Howard, J. (2006).

The depolymerizing kinesin MCAK uses lattice diffusion to rapidly target

microtubule ends. Nature 441, 115–119.

Hendrickx, A., Beullens, M., Ceulemans, H., Den Abt, T., Van Eynde, A.,

Nicolaescu, E., Lesage, B., and Bollen, M. (2009). Docking Motif-Guided

Mapping of the Interactome of Protein Phosphatase-1. Chem. Biol. 16,

365–371.

Hoar, K., Chakravarty, A., Rabino, C., Wysong, D., Bowman, D., Roy, N., and

Ecsedy, J.A. (2007). MLN8054, a Small-Molecule Inhibitor of Aurora A, Causes

Spindle Pole and Chromosome Congression Defects Leading to Aneuploidy.

Mol. Cell. Biol. 27, 4513–4525.

Holland, A.J., and Cleveland, D.W. (2009). Boveri revisited: chromosomal

instability, aneuploidy and tumorigenesis. Nat. Rev. Mol. Cell Biol. 10,

478–487.

Holland, A.J., Lan, W., Niessen, S., Hoover, H., and Cleveland, D.W. (2010).

Polo-like kinase 4 kinase activity limits centrosome overduplication by

autoregulating its own stability. J. Cell Biol. 188, 191–198.

Hopfield, J.J. (1974). Kinetic proofreading: A new mechanism for reducing

errors in biosynthetic processes requiring high specificity. Proc. Natl. Acad.

Sci. USA 71, 4135–4139.

Hubbard, M.J., and Cohen, P. (1989). Regulation of protein phosphatase-1G

from rabbit skeletal muscle. 1. Phosphorylation by cAMP-dependent protein

kinase at site 2 releases catalytic subunit from the glycogen-bound

holoenzyme. FEBS 186, 701.

Hyman, A.A., and Mitchison, T.J. (1991). Two different microtubule-based

motor activities with opposite polarities in kinetochores. Nature 351, 206–211.

Kapoor, T.M., Lampson, M.A., Hergert, P., Cameron, L., Cimini, D., Salmon,

E.D., McEwen, B.F., and Khodjakov, A. (2006). Chromosomes Can Congress

to the Metaphase Plate Before Biorientation. Science 311, 388–391.

Kim, Y., Heuser, J.E., Waterman, C.M., and Cleveland, D.W. (2008). CENP-E

combines a slow, processive motor and a flexible coiled coil to produce an

essential motile kinetochore tether. J. Cell Biol. 181, 411–419.

Kunitoku, N., Sasayama, T., Marumoto, T., Zhang, D., Honda, S., Kobayashi,

O., Hatakeyama, K., Ushio, Y., Saya, H., and Hirota, T. (2003). CENP-A

Phosphorylation by Aurora-A in Prophase Is Required for Enrichment of


Aurora-B at Inner Centromeres and for Kinetochore Function. Dev. Cell 5,

853–864.

Lampson, M.A., Renduchitala, K., Khodjakov, A., and Kapoor, T.M. (2004).

Correcting improper chromosome-spindle attachments during cell division.

Nat. Cell Biol. 6, 232–237.

Liao, H., Li, G., and Yen, T.J. (1994). Mitotic regulation of microtubule

cross-linking activity of CENP-E kinetochore protein. Science 265, 394–398.

Liu, D., Vleugel, M., Backer, C.B., Hori, T., Fukagawa, T., Cheeseman, I.M.,

and Lampson, M.A. (2010). Regulated targeting of protein phosphatase 1 to

the outer kinetochore by KNL1 opposes Aurora B kinase. J. Cell Biol. 188,

809–820.

Manfredi, M.G., Ecsedy, J.A., Meetze, K.A., Balani, S.K., Burenkova, O., Chen,

W., Galvin, K.M., Hoar, K.M., Huck, J.J., LeRoy, P.J., et al. (2007). Antitumor

activity of MLN8054, an orally active small-molecule inhibitor of Aurora A

kinase. Proc. Natl. Acad. Sci. USA 104, 4106–4111.

Marumoto, T., Honda, S., Hara, T., Nitta, M., Hirota, T., Kohmura, E., and Saya,

H. (2003). Aurora-A kinase maintains the fidelity of early and late mitotic events

in HeLa cells. J. Biol. Chem. 278, 51786–51795.

Marumoto, T., Zhang, D., and Saya, H. (2005). Aurora-A - A guardian of poles.

Nat. Rev. Cancer 5, 42–50.

Nousiainen, M., Sillje, H.H., Sauer, G., Nigg, E.A., and Korner, R. (2006).

Phosphoproteome analysis of the human mitotic spindle. Proc. Natl. Acad.

Sci. USA 103, 5391–5396.

Okada, Y., and Hirokawa, N. (1999). A Processive Single-Headed Motor:

Kinesin Superfamily Protein KIF1A. Science 283, 1152–1157.

Ovechkina, Y., Wagenbach, M., and Wordeman, L. (2002). K-loop insertion

restores microtubule depolymerizing activity of a ‘‘neckless’’ MCAK mutant.

J. Cell Biol. 159, 557–562.

Putkey, F.R., Cramer, T., Morphew, M.K., Silk, A.D., Johnson, R.S., McIntosh,

J.R., and Cleveland, D.W. (2002). Unstable Kinetochore-Microtubule Capture

and Chromosomal Instability Following Deletion of CENP-E. Dev. Cell 3,

351–365.

Rieder, C.L., and Alexander, S.P. (1990). Kinetochores are transported

poleward along a single astral microtubule during chromosome attachment

to the spindle in newt lung cells. J. Cell Biol. 110, 81–95.

Ruchaud, S., Carmena, M., and Earnshaw, W.C. (2007). Chromosomal

passengers: conducting cell division. Nat. Rev. Mol. Cell Biol. 8, 798–812.

Sassoon, I., Severin, F.F., Andrews, P.D., Taba, M.-R., Kaplan, K.B., Ashford,

A.J., Stark, M.J.R., Sorger, P.K., and Hyman, A.A. (1999). Regulation of

Saccharomyces cerevisiae kinetochores by the type 1 phosphatase Glc7p.

Genes Dev. 13, 545–555.

Shaner, N.C., Lin, M.Z., McKeown, M.R., Steinbach, P.A., Hazelwood, K.L.,

Davidson, M.W., and Tsien, R.Y. (2008). Improving the photostability of bright

monomeric orange and red fluorescent proteins. Nat. Methods 5, 545–551.

Tanaka, T.U., Rachidi, N., Janke, C., Pereira, G., Galova, M., Schiebel, E.,

Stark, M.J.R., and Nasmyth, K. (2002). Evidence that the Ipl1-Sli15 (Aurora

Kinase-INCENP) complex promotes chromosome biorientation by altering

kinetochore-spindle pole connections. Cell 108, 317–329.

Thorn, K.S., Ubersax, J.A., and Vale, R.D. (2000). Engineering the Processive

Run Length of the Kinesin Motor. J. Cell Biol. 151, 1093–1100.

Trinkle-Mulcahy, L., Andersen, J., Lam, Y.W., Moorhead, G., Mann, M., and

Lamond, A.I. (2006). Repo-Man recruits PP1g to chromatin and is essential

for cell viability. J. Cell Biol. 172, 679–692.

Trinkle-Mulcahy, L., Andrews, P.D., Wickramasinghe, S., Sleeman, J.,

Prescott, A., Lam, Y.W., Lyon, C., Swedlow, J.R., and Lamond, A.I. (2003).

Time-lapse imaging reveals dynamic relocalization of PP1gamma throughout

the mammalian cell cycle. Mol. Biol. Cell 14, 107.

Varga, V., Helenius, J., Tanaka, K., Hyman, A.A., Tanaka, T.U., and Howard, J.

(2006). Yeast kinesin-8 depolymerizes microtubules in a length-dependent

manner. Nat. Cell Biol. 8, 957–962.

Weaver, B.A.A., Bonday, Z.Q., Putkey, F.R., Kops, G.J.P.L., Silk, A.D., and

Cleveland, D.W. (2003). Centromere-associated protein-E is essential for the

mammalian mitotic checkpoint to prevent aneuploidy due to single chromo-

some loss. J. Cell Biol. 162, 551–563.

Woehlke, G., Ruby, A.K., Hart, C.L., Ly, B., Hom-Booher, N., and Vale, R.D.

(1997). Microtubule Interaction Site of the Kinesin Motor. Cell 90, 207–216.

Wood, K.W., Sakowicz, R., Goldstein, L.S.B., and Cleveland, D.W. (1997).

CENP-E Is a Plus End-Directed Kinetochore Motor Required for Metaphase

Chromosome Alignment. Cell 91, 357–366.

Yao, X., Abrieu, A., Zheng, Y., Sullivan, K.F., and Cleveland, D.W. (2000).

CENP-E forms a link between attachment of spindle microtubules to kineto-

chores and the mitotic checkpoint. Nat. Cell Biol. 2, 484–491.

Yucel, J.K., Marszalek, J.D., McIntosh, J.R., Goldstein, L.S.B., Cleveland,

D.W., and Philp, A.V. (2000). CENP-meta, an Essential Kinetochore Kinesin

Required for the Maintenance of Metaphase Chromosome Alignment in

Drosophila. J. Cell Biol. 150, 1–12.

Zecevic, M., Catling, A.D., Eblen, S.T., Renzi, L., Hittle, J.C., Yen, T.J.,

Gorbsky, G.J., and Weber, M.J. (1998). Active MAP Kinase in Mitosis:

Localization at Kinetochores and Association with the Motor Protein

CENP-E. J. Cell Biol. 142, 1547–1558.


PNPASE Regulates RNA Importinto MitochondriaGeng Wang,1 Hsiao-Wen Chen,4 Yavuz Oktay,1 Jin Zhang,5 Eric L. Allen,5 Geoffrey M. Smith,5 Kelly C. Fan,5

Jason S. Hong,5 Samuel W. French,5 J. Michael McCaffery,6 Robert N. Lightowlers,7 Herbert C. Morse III,8

Carla M. Koehler,1,2,* and Michael A. Teitell2,3,5,*1Department of Chemistry and Biochemistry2Molecular Biology Institute3Jonsson Comprehensive Cancer Center, Broad Stem Cell Research Center, California NanoSystems Institute, and Center for Cell Control

University of California at Los Angeles, Los Angeles, CA 90095, USA4Center for Molecular and Mitochondrial Medicine and Genetics, University of California at Irvine, Irvine, CA 92697, USA5Department of Pathology and Laboratory Medicine, David Geffen School of Medicine at UCLA, Los Angeles, CA 90095, USA6Integrated Imaging Center, Department of Biology, Johns Hopkins University, Baltimore, MD 21218, USA7Mitochondrial Research Group, Institute for Ageing and Health, Newcastle University, Newcastle upon the Tyne, UK8Laboratory of Immunopathology, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Rockville,MD 20852, USA

*Correspondence: [email protected] (C.M.K.), [email protected] (M.A.T.)

DOI 10.1016/j.cell.2010.06.035

SUMMARY

RNA import into mammalian mitochondria is consid-ered essential for replication, transcription, andtranslation of the mitochondrial genome but thepathway(s) and factors that control this import arepoorly understood. Previously, we localized polynu-cleotide phosphorylase (PNPASE), a 30/ 50 exoribo-nuclease and poly-A polymerase, in the mitochon-drial intermembrane space, a location lackingresident RNAs. Here, we show a new role for PNPASEin regulating the import of nuclear-encoded RNAsinto the mitochondrial matrix. PNPASE reductionimpaired mitochondrial RNA processing and polycis-tronic transcripts accumulated. Augmented import ofRNase P, 5S rRNA, and MRP RNAs depended onPNPASE expression and PNPASE–imported RNAinteractions were identified. PNPASE RNA process-ing and import activities were separable and a mito-chondrial RNA targeting signal was isolated thatenabled RNA import in a PNPASE-dependentmanner. Combined, these data strongly support anunanticipated role for PNPASE in mediating the trans-location of RNAs into mitochondria.

INTRODUCTION

Much is understood about the mechanisms that regulate

nuclear-encoded protein import into mitochondria (Chacinska

et al., 2009). By contrast, much less is known about the factors

that regulate mitochondrial RNA import. Almost every organism

with mitochondria imports tRNAs and aminoacyl-tRNA synthe-

tases (Alfonzo and Soll, 2009; Duchene et al., 2009). The number

of imported tRNAs ranges from one in yeast to all in trypano-

somes, with mammalian mitochondria importing several

different tRNAs both in vitro and in vivo (Kolesnikova et al.,

2004; Rubio et al., 2008). Recently, microRNAs were isolated

from mitochondria (Kren et al., 2009) and the nuclear-encoded

5S rRNA was identified as one of the most abundant RNAs in

human mitochondria (Entelis et al., 2001; Smirnov et al., 2008).

RNase MRP and RNase P enzyme complexes localize and

function in mammalian mitochondria and may contain RNAs

that are encoded within the nucleus. RNase MRP functions

as a site-specific endoribonuclease involved in primer RNA

cleavage during the replication of mitochondrial DNA (Chang

and Clayton, 1987). Mammalian RNase P functions in the

processing of tRNAs during the maturation of mitochondrial

transcripts that encode oxidative phosphorylation (OXPHOS)

components. Large polycistronic RNA transcripts are generated

from the heavy and light strand promoters in mammalian mito-

chondria (Bonawitz et al., 2006). tRNAs often separate the

coding regions for OXPHOS subunits and are processed and

removed by RNase P (Doersen et al., 1985). Earlier studies

showed that the RNA component of RNase P was imported

into mammalian mitochondria (Doersen et al., 1985) and the

RNase P RNA and processing activity has been copurified

from mitochondria (Puranam and Attardi, 2001). By contrast,

RNase P RNA is encoded in the mitochondrial genome in

Saccharomyces cerevisiae (Hollingsworth and Martin, 1986).

Recently, an additional RNase P enzyme, consisting of three

protein subunits, has been purified from human mitochondria.

This alternative RNase P enzyme processes single tRNA 50

precursor sequences in vitro without an RNA component

(Holzmann et al., 2008; Walker and Engelke, 2008). The identifi-

cation of two enzymes with RNase P activity in mammalian

mitochondria warrants further investigation.

Critical open questions about RNA import into mammalian

mitochondria include the selectivity of RNAs, the factor(s) target-

ing RNA from the cytosol, and the translocation pathway(s)

across the mitochondrial membranes (Duchene et al., 2009).


Different import pathways have been proposed for a subset of

precursors in different species and the details of the components

involved remain ill-defined. For example, the TOM and TIM

protein import complexes have been implicated in the import

of tRNALys in yeast mitochondria (Tarassov et al., 1995). By

contrast, a protein import pathway-independent mechanism

may exist and involve a 600 kDa multi-subunit RNA import

complex (RIC) in Leishmania (Mukherjee et al., 2007). Thus,

inconsistencies among RNA import systems suggest a critical

need to learn more about the nature of the RNAs that are

imported and the factor(s) mediating this import.

Previously we showed an unexpected location in the mito-

chondrial intermembrane space (IMS) for mammalian polynucle-

otide phosphorylase (PNPASE) (Chen et al., 2007; Chen et al.,

2006; Rainey et al., 2006), a 30 / 50 exoribonuclease and

poly-A polymerase that uses phosphorolysis to degrade RNA

(Yehudai-Resheff et al., 2001). This was a surprise because

We expected that PNPASE would instead localize in the RNA-

abundant mitochondrial matrix. Therefore, initial studies on

mammalian PNPASE focused on a general role in maintaining

mitochondrial homeostasis, potentially by regulating adenine

nucleotide levels (Chen et al., 2006; Portnoy et al., 2008). Here,

we show that PNPASE has a central role in augmenting the

import of small RNA components required for DNA replication

and RNA processing into the mitochondrial matrix. We suggest

that PNPASE regulates adenine nucleotide levels and mitochon-

drial homeostasis at least partly by regulating RNA import to

control the abundance of the electron transport chain (ETC)

components.

RESULTS

PNPASE Forms a Trimer in Yeast and MammalianMitochondriaTo examine PNPASE in the IMS, a coimmunoprecipitation (IP)

assay was performed to identify potential binding partners. A

6XHis-Protein-C (HisPC) tag was added to the C terminus of

PNPASE and stable PNPASE-HisPC expressing HEK293 cells

were generated. The PNPASE-HisPC protein was detected in

isolated mitochondria by immunoblot (Figure S1A available

online). PNPASE-HisPC was isolated from mitochondria using

sequential purification over Ni2+ and Protein-C affinity columns,

followed by elution, and copurifying proteins were identified by

Sypro Ruby staining (Figure S1B) and liquid chromatography-

tandem mass spectrometry (LC-MS/MS) (data not shown). All

of the identified bands originated from PNPASE, suggesting

that PNPASE lacks partner proteins in vivo. Bands of lower

molecular weight than the �85 kDa of PNPASE monomers

were likely degradation products. The assembly state of

PNPASE was also investigated. Mitochondria from yeast ex-

pressing human PNPASE (Rainey et al., 2006) were detergent

solubilized and separated on blue-native (BN) gels. Immunoblot

showed PNPASE in a complex of�240 kDa (Figure S1C), similar

to the trimeric complex of endogenous mouse hepatocyte

PNPASE (Chen et al., 2006) and bacterially-expressed human

PNPASE (French et al., 2007). PNPASE-HisPC isolated from

HEK293 mitochondria also migrated in a similarly-sized complex

(Figure S1D). Overall, PNPASE assembled identically in yeast

and mammalian mitochondria into a homo-oligomeric

complex, consisting of a trimer or a ‘dimer of trimers’ (Symmons

et al., 2002). These results strongly suggest that PNPASE

assembles and may function similarly in yeast and mammalian

mitochondria.

Pnpt1 Knockout Cells Show Altered MitochondrialMorphology and Impaired RespirationWe used several approaches to determine the function of

PNPASE in mitochondria. First, the gene encoding PNPASE

(Pnpt1) was knocked out (KO) in C57BL/6 mice (Figure S2 and

Figure S3). Homozygous Pnpt1neo-flox mice, in which exon 2

was flanked by loxP recombination sites, were viable and fertile.

A complete KO of Pnpt1 exon 2 was generated by crossing

CMVCRE expressing mice with Pnpt1WT/neo-flox heterozygotes

followed by inter-crossing the Pnpt1WT/KO progeny. Pnpt1KO/KO

mice were embryonic lethal (Figure 1A).

Our prior cell line studies showed that RNAi targeting PNPASE

reduced ATP production by OXPHOS and slowed cell growth

(Chen et al., 2006). Recently, a viable liver KO of the COX10

gene, which is required for cytochrome c oxidase and ETC func-

tion, was produced, suggesting that targeted disruption of Pnpt1

in hepatocytes might be tolerated (Diaz et al., 2008). Therefore,

a liver-specific KO (HepKO) of Pnpt1 was generated by the

cross AlbCRE/WT/Pnpt1neo-flox/neo-floxx AlbWT/WT/Pnpt1neo-flox/neo-flox,

which produced fertile progeny at the expected frequency

(Figure S2C). Quantitative real-time PCR (QPCR) from HepKO

liver showed reduced Pnpt1 transcripts containing targeted

exon 2 compared with those containing untargeted exon 28

(Figure 1B). PNPASE protein expression was also markedly

reduced in HepKO liver compared with sex-matched littermate

WT liver (Figure 1B). The expression of Pnpt1 transcripts and

PNPASE protein in HepKO liver likely arises from heterogeneous

liver elements. Also, the AlbCRE deleting strain is hepatocyte-

specific but incomplete (Postic and Magnuson, 2000). The

physiology of these mice will be reported elsewhere.

The ultrastructure of HepKO liver mitochondria was investi-

gated by transmission electron microscopy (TEM). Rather than

displaying ordered, linear cristae with convolutions as in WT

mitochondria, the HepKO mitochondria showed disordered

circular and smooth IM cristae (Figure 1C), similar to mitochon-

dria that are impaired for OXPHOS (Mandel et al., 2001) and to

Pnpt1 RNAi mammalian cell lines (Chen et al., 2006). These

and prior results suggesting that reduced PNPASE caused a

decrease in ATP production prompted the evaluation of O2

consumption from HepKO liver mitochondria. Oxygen electrode

studies showed a �1.5- to 2-fold decrease in the activity of

Complex IV and Complexes II+III+IV when normalized to citrate

synthase activity in HepKO compared to WT mitochondria

(Figure 1D). Combined, these data establish an in vivo role for

PNPASE in mitochondrial morphogenesis and respiration

through an unknown mechanism.

PNPASE Is Required for the Processingof Mitochondrial RNA TranscriptsThe data showing decreased respiration in HepKO mitochon-

dria suggested a reduction in functional ETC complexes. There-

fore, RNA processing and translation were examined in cells


with decreased PNPASE. HEK293 cells with > 75% reduced

PNPASE expression were generated by RNAi, followed by

mitochondrial RNA (mtRNA) transcript quantification using

QPCR normalized to cytosolic GAPDH RNA. All mtRNAs tested

were reduced in Pnpt1 RNAi cells compared to WT cells

(Figure 1E). Proteins translated from mtRNAs were decreased

in HEK293 Pnpt1 RNAi cells (data not shown) and HepKO liver

cells (Figure 2A), strongly suggesting that a decrease in

functional ETC complexes was responsible for decreased

respiration.

The processing of polycistronic mtRNAs was investigated

because reduced PNPASE could cause an accumulation of large

precursor transcripts, resulting in reduced ETC proteins. Tran-

script processing requires RNase P excision of the tRNAs

between ETC gene coding regions. Primers were designed to

test processing between adjacent Cox1 and Cox2 transcripts

that are separated by tRNAser and tRNAasp (Figure 2B). The

primer set Cox1f and Cox1r generates a 450-bp fragment

whereas the primer pair Cox1f and Cox2r generates a 900-bp

fragment when tRNAser and tRNAasp are not excised from large

Figure 1. Deletion of Pnpt1 in Hepatocytes

Impairs Mitochondrial Function

(A) Breeding strategy (top) and results (bottom) for

attempting to generate a PNPASE KO mouse.

(B) Hepatocyte-specific Pnpt1 KO (HepKO) in

4-week old mice. Top: QPCR for liver Pnpt1

expression using an exon 2 – exon 3 primer pair

versus a primer pair within exon 28. Bottom:

PNPASE immunoblot from 4-week old WT and

HepKO mouse livers.

(C) HepKO mitochondria have altered cristae. Left:

TEM of 6-week old littermate livers shows circular,

smooth HepKO IM cristae in contrast to linear,

stacked cristae of WT mitochondria. Right: Anal-

ysis of cristae morphology in which a single normal

cristae within a mitochondrion was scored as

normal. Indet = indeterminate.

(D) Decreased respiration in isolated HepKO mito-

chondria. Oxygen consumption (nmol/min/mg

protein) for ETC complexes IV and II+III+IV was

measured using an O2 electrode. Mitochondrial

mass was determined by citrate synthase (CS)

activity using a spectrophotometer. Respiratory

activities are shown normalized to CS activity.

(E) Decreased mature mtRNAs in HEK293 cells

with RNAi to PNPT1. Transcripts were quantified

relative to cytosolic GAPDH expression by QPCR

from HEK293 cells 7d post-infection (nadir) with

scramble (Scr) or PNPT1 RNAi retroviral

constructs. See also Figure S1, Figure S2, and

Figure S3.

precursor transcripts. Isolated mitochon-

drial RNAs were examined by RT-PCR. A

900-bp fragment was detected from

HepKO but not from WT liver mitochon-

dria (Figure 2B). Similar results were

obtained using the same primers in

PNPASE KO mouse embryonic fibro-

blasts (MEFs) (Figure S3 and Figure S4A).

To query RNA processing at a second site, primers were gener-

ated for adjacent Cox2 and Atp8/6 loci, separated by tRNAlys.

Again, polycistronic transcripts accumulated in the PNPASE

KO MEFs (Figure S4B). The sizes of Cox1 and Cox3 transcripts

were investigated using specific probes and Northern blot

(Figure 2C). In addition to the mature Cox1 and Cox3 transcripts,

a range of larger precursor transcripts was seen in HepKO liver

cells. Also, the mature 0.9-kb Cox3 transcript was more

abundant in WT than HepKO liver. The abundance of mitochon-

dria-encoded COX3 and ND6 proteins was determined

(Figure 2D). In HepKO liver mitochondria, the steady-state

abundance of PNPASE was decreased by �2-fold compared

to the WT, similar to a �2-fold decrease for COX3 and ND6

proteins. Controls TOM40, MORTALIN, TIM23, and BAP37

showed that the amount of nuclear-encoded mitochondrial

proteins, and therefore the mitochondrial mass, was similar

between HepKO and sex-matched WT littermate liver cells.

Thus, the processing of polycistronic mtRNAs was impaired in

mitochondria with reduced PNPASE, resulting in fewer mature

mtRNAs and reduced ETC complexes.


RNase P RNA Binds to PNPASE and May Functionin PNPASE-Dependent mtRNA ProcessingAlthough an in vitro form of the RNase P enzyme can function

without nucleic acid (Holzmann et al., 2008; Walker and Engelke,

2008), the RNA component of RNase P localizes to mitochondria

(Puranam and Attardi, 2001) and has not been excluded from

mtRNA processing in vivo (Holzmann et al., 2008). Therefore,

the abundance of RNase P RNA in HepKO liver mitochondria

was determined by RT-PCR and QPCR (Figure 3A). Reproduc-

ibly, RNase P RNA was decreased by �75% in HepKO versus

WT liver mitochondria, suggesting that PNPASE may help import

and/or stabilize RNase P RNA. Therefore, we determined

whether RNase P RNA directly bound to PNPASE in HEK293

cells stably expressing dual-tagged PNPASE-HisPC. Isolated

mitochondria were treated with nuclease and tagged PNPASE

was purified. RNase P RNA was amplified by RT-PCR and

copurified with PNPASE (Figure 3B, lane 4). Importantly, control

IMS-localized TIM23-HisPC in stably-expressing HEK293 cells

did not bind RNase P RNA (Figure 3B, lane 2). PNPASE also

Figure 2. HepKO Liver Mitochondria Do Not

Efficiently Process mtRNA Precursors

(A) In organello protein synthesis. WT and HepKO

mitochondria (100 mg) were treated with micro-

coccal nuclease S7, and in organello translation

was performed using [35S]-MET. TOM40 immuno-

blot shows equivalent mitochondria in each assay.

(B) RNA was isolated from WT and HepKO liver

mitochondria followed by DNase I treatment to

remove contaminating DNA. RT-PCR was per-

formed for Cox1 and Cox2 with primers shown in

the schematic (upper) and separated on a 1.5%

agarose gel.

(C) Northern blot of mtRNA from WT and HepKO

mouse liver mitochondria using a Cox1 or Cox3

DNA probe. Asterisks mark larger precursor

mtRNAs and the arrow shows the mature mtRNA.

(D) Steady-state expression of nuclear- and mito-

chondrial-encoded proteins in WT and HepKO

liver mitochondria. Equivalent nuclear-encoded

protein expression shows that HepKO reduced

mitochondria-encoded protein expression was

not due to differing mitochondrial content between

WT and HepKO liver cells. See also Figure S4.

did not adventitiously bind RNA because

the RNA transcripts for Cox1, GAPDH,

mitochondrial 12S rRNA and mitochon-

drial tRNAtrp were not bound to PNPASE

(Figure 3B). Furthermore, PNPASE-

HisPC, but not TIM23-HisPC, bound

in vitro transcribed and imported RNase

P and MRP RNAs, but not control mito-

chondrial RNA, in cross linking IP assays

(Figure S5A) Thus, RNase P and MRP

RNA bound specifically to PNPASE.

A protein-only RNase P complex was

shown to process the 50-terminus of a

single mitochondrial tRNAlys (Holzmann

et al., 2008; Walker and Engelke, 2008).

However, most mammalian mitochondrial tRNA genes are

grouped together in the mitochondrial genome and lack an

intervening pre-sequence, causing them to be expressed with

the ETC genes in polycistronic transcripts. Therefore, we

examined whether protein-only RNase P can efficiently process

paired mitochondrial tRNAs, as must occur in vivo. As done

previously (Holzmann et al., 2008), mitoplast lysates were

untreated or treated with micrococcal nuclease, followed by

inactivation with EGTA and EDTA. tRNAlys or abutting tRNAs,

tRNAhistRNAser, were then added to the lysates. Nuclease-

treated mitoplast lysates cleaved the 50-precursor sequence of

the single and abutting tRNA substrates as efficiently as the

lysates without treatment, as shown before (Figure 3C). By

contrast, nuclease-treated lysates were impaired in cleaving

the two abutting tRNAs into individual tRNAs, strongly sug-

gesting an additional nucleic acid component is required for

efficient processing. Interestingly, mitoplast lysates from

HepKO liver showed the same defect on abutting tRNA matura-

tion as the nuclease-treated WT mitoplast lysates (Figure 3D).


Furthermore, the in vivo processing and separation of an

endogenous paired tRNAhistRNAser substrate was inhibited in

HepKO compared to WT liver mitochondria, whereas a linked

12 s rRNA-tRNAval substrate was processed equivalently (Fig-

ure S5B). These results, combined with prior data (Puranam and

Attardi, 2001), suggest protein-only and RNase P RNA-containing

RNase P complexes coexist in mitochondria, with efficient tRNA

processing requiring PNPASE-dependent RNase P RNA.

PNPASE Augments the Import of RNase P, 5S rRNA,and MRP RNAs into Yeast MitochondriaSimilar PNPASE complexes in yeast and mammalian mitochon-

dria (Chen et al., 2006) (Figure S1) suggest yeast as a model for

studying the importof nuclear-encodedRNAs. Importantly, added

human PNPASE did not alter yeast mitochondrial morphology,

rate of proliferation, or extent of cell death, supporting this choice

Figure 3. RNase P RNA Binds to PNPASE

and May Function in PNPASE-Dependent

tRNA Processing

(A) Left: RNA was isolated from WT and HepKO

liver mitochondria following nuclease treatment.

RT-PCR was performed with primers that

amplify nuclear-encoded RNase P RNA (212 bp).

Right: QPCR analysis of RNase P RNA expression

relative to TOM40 protein in isolated mitochondria.

(B) PNPASE-HisPC (PNP) or TIM23-HisPC (TIM23)

was purified from stably transfected HEK293 cells.

Candidate interacting RNAs that copurified in the

final eluate with PNPASE-HisPC and TIM23-

HisPC were identified by primer-specific RT-

PCR. T is the total lysate (0.3% of the reaction)

before mitochondrial purification and B is the

bound fraction. Note that only RNase P RNA

bound to PNPASE-HisPC (lane 4).

(C) Mitoplast extract (10 mg) was prepared from

nuclease-treated, intact WT liver mitochondria.

The extract was then treated with nuclease (+),

as indicated, and then inactivated with EDTA

and EGTA. The nuclease-treated or untreated

extract was incubated with abutted tRNAs

(tRNAHistRNASer) or a single tRNA (tRNALys) at

25�C for 10 or 30 min. RNA was separated on an

urea-acrylamide gel and detected by autoradiog-

raphy. A MORTALIN immunoblot shows equiva-

lent mitoplast extract in each assay.

(D) Mitoplast extract was prepared from nuclease-

treated, intact WT or HepKO liver mitochondria.

The enzymatic assay was performed as described

for panel (C). See also Figure S5.

(Figure S6). Mitochondria isolated from

WT yeast or yeast expressing human

PNPASE were incubated with in vitro tran-

scribed human RNase P RNA in import

buffer. The reaction was treated with

nuclease to remove nonimported RNA

followed by RNA isolation and RT-PCR.

RNase P RNA abundance was increased

in mitochondria containing PNPASE com-

pared to WT mitochondria (Figure 4A). This

RNA increase was specific for certain RNAs because cytosolic

GAPDH RNA was not increased in the same mitochondria

(Figure 4B). Osmotic shock was used to identify the location of

the imported RNase P RNA (Koehler et al., 1998). Mitochondria

were incubated in hypotonic buffer to rupture the outer membrane

and the mitoplasts (P, pellet fraction that contains the matrix and

IM) and the supernatant (S, contains the soluble IMS contents)

were separated by centrifugation (Figure 4C). RNase P RNA was

detected by RT-PCR and was localized in the mitochondrial

matrix. Detergent exposed the matrix to verify that the nuclease

degraded the RNase P RNA. To confirm that osmotic shock did

not disrupt the IM, antibodies against cytochrome b2 (cyt b2;

IMS) and a-ketoglutarate dehydrogenase (KDH; matrix) showed

that cyt b2 was sensitive to protease in the IMS, but KDH was

resistant to protease until the IM was lysed with Triton X-100.

Thus, RNase P RNA import was augmented, or RNase P RNA


was stabilized, in the yeast mitochondrial matrix when exogenous

PNPASE was present in the IMS.

To confirm the RT-PCR results and assay other imported

RNAs, we performed the in vitro RNA import assay with yeast

mitochondria and radiolabeled human RNAs (Figure 4D). Two

different RNA volumes were used and the imported RNA was

isolated and separated on a urea-acrylamide gel followed by

autoradiography. RNase P, 5S rRNA, and MRP RNAs showed

augmented import or stability in mitochondria expressing

PNPASE relative to control mitochondria (Figure 4D). Again,

this increase was RNA-type specific as PNPASE did not

augment GAPDH RNA levels. When the mitochondrial

membrane potential was dissipated, the RNase P RNA level

was not increased in this assay system (Figure 4E).

PNPASE Mutations that Inactivate RNA ProcessingDo Not Affect RNA Import or StabilitySpecific point mutations in conserved regions of PNPASE impair

its RNA exonuclease or 30 poly-A polymerase activity in

biochemical assays (Portnoy et al., 2008), although the effects

of these mutations on PNPASE in vivo are unknown. To deter-

mine whether the RNA import or stabilization activity of PNPASE

was separable from its RNA processing activities, RNase P RNA

import was studied when different PNPASE mutants were

expressed in yeast mitochondria (Figure 5A). The point mutants

generated and tested were based on results from Schuster and

colleagues (Portnoy et al., 2008). Mutants D135G and S484A

lacked poly-A polymerase and RNA degradation activities

in vitro. Mutant D544G and double mutant R445E/R446E

showed enhanced in vitro poly-A polymerase activity but com-

promised degradation activity. Of the four mutants, PNPASE

S484A and R445E/R446E supported the import or stabilization

of RNase P RNA, whereas mutants D135G and D544G were

defective in this function (Figure 5A). The abundance of WT

and the four mutant PNPASE proteins were similar between

yeast strains and all of the PNPASE proteins assembled into

�240 kDa complexes without impairment (Figure 5A, lower

panel). This contrasts with the expectation that mutant D135G

Figure 4. PNPASE Augments RNase P, 5S

rRNA, and MRP RNA Import into Yeast Mito-

chondria

(A) Upper: In vitro transcribed human RNase P

RNA was incubated with yeast mitochondria

expressing human PNPT1 (PNP) or an empty

vector (Vec) control. Nonimported RNA was

digested with nuclease and the imported RNA

was detected by RT-PCR. PNPT1-expressing

mitochondria without added RNase P RNA was

included as a specificity control for import and

RT-PCR (lane 2: Std, 1% of the reaction). Lower:

Control showing equivalent total mitochondrial

nucleic acid in each reaction.

(B) Upper: As in panel (A), with cytosolic human

GAPDH RNA used as a substrate. Middle- Control

showing equivalent total mitochondrial nucleic

acid in each reaction. Lower: Western blot

showing PNPASE expression and PORIN immu-

noblot showing equivalent mitochondria in each

import assay.

(C) After import as in panel (A), mitochondria

were subjected to osmotic shock, fractionated

by centrifugation into soluble (S) and pellet (P)

fractions, followed by proteinase K and nuclease

additions where indicated. The pellet fraction

was solubilized with Triton X-100 to expose the

matrix. Localization was determined by RT-PCR

for RNase P RNA and immunoblot for KDH (matrix)

and cyt b2 (IMS) proteins.

(D) Upper: Radiolabeled RNase P, MRP, 5S rRNA,

and GAPDH human RNAs were in vitro transcribed

and then incubated with yeast mitochondria

expressing PNPASE or an empty vector control.

Nonimported RNA was digested with nuclease,

followed by RNA isolation, separation on a urea-

acrylamide gel, and autoradiography. Import reac-

tions were repeated with 13 and 23 amounts of

RNA. Lower: Control showing equivalent total

mitochondrial nucleic acid in each reaction.

(E) Upper: As in panel A except that the mitochon-

drial membrane potential (Dc) was dissipated prior

to import. Lower: Control showing equivalent total

mitochondrial nucleic acid in each reaction.


would fail to form a trimeric complex from prior studies (Portnoy

et al., 2008). Therefore, the mitochondrial RNA import or stabili-

zation function of PNPASE is separable from its poly-A poly-

merase or exoribonuclease activities.

To determine whether PNPASE augmented either RNA import

or stabilization in mitochondria, the enzymatic properties of

the WT and S484A mutant protein were examined with respect

to RNA turnover in vitro and in isolated yeast mitochondria.

For in vitro studies, WT and S484A PNPASE were immunoprecip-

itated from yeast mitochondria and tested in an in vitro degrada-

tion assay with radiolabeled RNase P RNA (Figure 5B). WT

PNPASE degraded the RNase P RNA, but the S484A mutant

was impaired, consistent with prior results (Portnoy et al., 2008).

In addition, radiolabeled RNase P RNA was imported into mito-

chondria, followed by incubation at 25�C for up to 90 min

(Figure 5C). The internalized RNase P RNA was separated on

a urea-acyrlamide gel and quantified during this time course using

a phosphorimager. The rate of degradation of RNase P RNA was

Figure 5. PNPASE Mutations that Inactivate

RNA Processing Do Not Affect RNA Import

or Stability

(A) Upper: Schematic for the positions of point

mutations made in the PNPASE protein. Listed

are the in vitro effects of mutations on 30 poly-

merase and RNA degrading activities, from (Port-

noy et al., 2008). Middle: Import reactions were

performed as in Figure 4A. Radiolabeled RNase

P RNA was incubated with isolated yeast mito-

chondria expressing an empty vector or the listed

PNPASE constructs. Lower panels: Immunoblot of

WT and point mutant PNPASE yeast transfectants

used in panel A import assay. A PORIN immuno-

blot confirms the colocalization of PNPASE WT

and mutants in yeast mitochondria. The assembly

state of WT and point mutant PNPASE was deter-

mined by solubilization with 1% digitonin and

separation on a 6%–16% BN gel, followed by

PNPASE immunoblot.

(B) Upper: WT and S484A PNPASE IPs from yeast

mitochondria were used to analyze RNA degrada-

tion properties. Lower: WT or S484A mutant

PNPASE was incubated with radiolabeled RNase

P RNA for 10 min at 25�C to assess degradation

activity. The asterisk marks degradation products.

(C) Left: Following in vitro import of radiolabeled

RNase P RNA and nuclease treatment to remove

nonimported RNA, mitochondria were incubated

for up to 90 min at 25�C and aliquots removed at

the indicated time points. The RNA was then

resolved by urea-acrylamide gel electrophoresis.

Right: RNase P RNA that was not degraded was

quantified using FX imager; n = 3.

similar for degradation competent WT

and incompetent mutant PNPASE pro-

teins, supporting a role for PNPASE in

augmenting the import of specific RNAs

into the mitochondrial matrix. This result

further supports PNPASE localizing to

the IMS because a greater amount of

WT PNPASE imported into the matrix could cause a relative

increase in the rate of turnover of matrix localized RNase P RNA.

A Predicted Stem-Loop RNA Structure MediatesPNPASE-Dependent RNA ImportResults showing PNPASE-dependent RNA import into mito-

chondria (Figure 3, Figure 4, and Figure 5) do not exclude an

indirect role. To establish a direct PNPASE role, a systematic

search was used to identify PNPASE-dependent RNA import

sequences. Primers were designed to generate distinct

segments of the 340 nucleotide (nt) RNase P RNA full length

sequence. RPf1 lacked the 50 70 nt, RPf2 lacked the 50 140 nt,

and RPr1 lacked the 30 148 nt of WT PNPT1 (Figure 6A). Import

assays were performed using full length or truncated in vitro

transcribed RNase P RNAs (Figures 6B and 6C). Augmented

RPf1 and RPr1 import into yeast mitochondria depended upon

PNPASE, as did the full length RNase P RNA. In striking contrast,

RPf2 was not efficiently imported into yeast mitochondria,


implicating the sequence between nt 71 and 140 in PNPASE-

augmented RNA import. To further refine this import signal,

RNA sequences lacking the 50 86 (RPf3) or 102 (RPf4) nts

were generated (Figure 6A). Augmented RPf3 and RPf4 import

into yeast mitochondria was PNPASE-dependent (Figure 6D),

further implicating an import signal between nt 103 and 140.

The most obvious, predicted secondary structure of RNase P

RNA in this region was a 20 nt stem-loop (Figure 6F). Interest-

ingly, a similarly-predicted stem-loop structure was also identi-

fied in MRP RNA. To determine whether one or both stem-loop

structures could mediate mitochondrial targeting of non-

imported GAPDH RNA, each 20 nt stem-loop sequence was

fused to the 50-terminus of the GAPDH RNA, which is not

imported (Figure 4D and Figure 7D). Strikingly, the RNase P

and MRP stem-loop structures licensed the PNPASE-depen-

dent import of GAPDH RNA into yeast mitochondria (Figure 6E).

By contrast, a control random 20nt sequence could not mediate

this import. Human mitochondrial tRNAtrp with the RNase P RNA

step-loop structure, but not tRNAtrp itself, was imported into

Figure 6. A Stem-Loop Structure Mediates

PNPASE-Dependent RNA Import

(A) Schematic depiction of human RNase P RNA

and deletion fragments.

(B) Import of full length RNase P RNA into yeast

mitochondria expressing PNPASE (PNP) or

control (Vec) vectors, as in Figure 4A.

(C) Import of the indicated RNase P RNA frag-

ments.

(D) Import of RNase P RNA fragments RPf3 and

RPf4.

(E) Import of human GAPDH mRNA or GAPDH

mRNA with control (CR), MRP RNA, or RNase P

RNA 20 nt sequences fused to the 50 end, as

shown in panel (F).

(F) The secondary structures and sequences of

mitochondrial RNA targeting signals in RNase P

(RP) and MRP (MRP) RNAs. A random sequence

(CR) was used as a control.

(G) Isolated mitochondria from HEK293 cells

stably expressing IMS-localized PNPASE-HisPC

or TIM23-HisPC (control) dual-tagged proteins

were incubated with [32P]-CTP labeled CR-tRNAtrp

or RP-tRNAtrp, followed by UV-cross linking,

tag-IP, separation by SDS-PAGE, and autoradiog-

raphy. See also Figure S6.

isolated mouse liver mitochondria, with

the tRNAtrp-PNPASE interaction cap-

tured using UV-cross linking (Figure 6G).

These results strongly implicate the

structural specificity of mitochondrial

RNA import (Figure 6F) and the direct

involvement of PNPASE in this process.

PNPASE Augments RNA Importinto Yeast and MammalianMitochondria In VivoTo explore in vivo RNA import into mito-

chondria, a construct was generated in

which the human RNase P RNA was expressed from the yeast

NME1 promoter (Figure 7A). When expressed in control yeast,

RNase P RNA localized to mitochondria, consistent with

nuclear-encoded RNA import by a PNPASE-independent mech-

anism, since yeast normally lack PNPASE. By contrast, RNase P

RNA import increased by �2-fold in mitochondria from yeast

expressing PNPASE compared with control cells (Figures 7A

and 7B). Importantly, an RNA similar in size to RNase P RNA

(340 nt), HOT13, that is translated in the cytosol and imported

as a protein into mitochondria, was not localized to mitochon-

dria. Also, mitochondrial-encoded RPM1, which codes for the

yeast homolog of RNase P RNA, was sequestered in the mito-

chondrion at a level equivalent to control yeast mitochondria,

as expected. These data indicate that PNPASE augments the

import of RNase P RNA into yeast mitochondria in vivo. Finally,

replacement of the human RNase P RNA stem-loop sequence

with the 20nt random sequence blocked augmented RNase P

RNA import into yeast mitochondria in vivo (Figure S7), confirm-

ing the role of the stem-loop in PNPASE-regulated import.


PNPASE augmented import of RNase P RNA into yeast

mitochondria is nonphysiologic. Therefore, we developed WT,

PNPASE KO, WT expressing human PNPASE, and PNPASE

KO expressing human PNPASE MEFs for import assays.

PNPASE abundance in each MEF line was confirmed by

immunoblot (Figure 7C). Radiolabeled RNase P RNA was not

imported into mitochondria from the PNPASE KO MEFs, but

was imported into mitochondria that contained mouse and/or

human PNPASE. The in vitro import of RNase P, MRP, 5S

rRNA, and GAPDH RNAs was also tested in liver mitochondria

isolated from the HepKO mouse and WT littermates. Again,

RNase P, 5S rRNA, and MRP RNAs were imported into mito-

chondria expressing PNPASE, whereas cytosolic GAPDH RNA

was not imported (Figure 7D). As expected, more than half of

the imported MRP RNA was processed into the mature

�130 nt form (Figure 7D, S7) (Chang and Clayton, 1987). By

contrast, mitochondrial RNA import was severely compromised

in HepKO liver mitochondria. Combined, these results strongly

support PNPASE as the first RNA import factor that mediates

the translocation of specific RNAs into the mammalian

mitochondrial matrix.

DISCUSSION

In contrast to the protein translocation, very little is understood

about the import of specific RNAs into the mitochondrion. A

confounder is that the spectrum of RNAs imported and the

import factors and mechanisms seem to vary greatly among

different organisms. At an extreme is mammalian mitochondria,

in which despite strong evidence for RNA import (Alfonzo and

Soll, 2009; Tarassov et al., 1995), no factors have thus far been

identified. Here we implicate PNPASE as the first RNA import

factor for mammalian mitochondria. Our results show that

PNPASE KO disrupts mitochondrial morphology and respiration

in mouse liver cells, at least partially by inhibiting the import of

RNAs that control the transcription and translation of the ETC

proteins. Our data also suggest that a nucleic acid component

of the RNase P RNA processing complex, possibly RNase P

RNA (Puranam and Attardi, 2001), is imported in vivo to process

linked tRNAs in long mitochondrial transcripts. PNPASE medi-

ated RNA delivery into the mitochondrial matrix and this import

was augmented over background. Strikingly, PNPASE RNA

import and RNA processing functions were separable and

Figure 7. PNPASE Augments RNA Import

into Yeast and Mammalian Mitochondria In

Vivo

(A) Upper: Human RNase P RNA yeast expression

construct is driven by the RPM1 RNA promoter,

NME1. Lower: Mitochondria from yeast express-

ing human RNase P RNA and either PNPASE

(PNP) or an empty vector (Vec) were isolated and

treated with nuclease. RNA was then isolated

from the total cell lysate or from nuclease-treated

mitochondria (Mito) and analyzed by primer-

specific RT-PCR.

(B) QPCR for Cox1 and RNase P RNAs isolated

from mitochondria in panel (A), normalized to the

total mitochondrial RNA obtained.

(C) Radiolabeled, in vitro transcribed RNase P

RNA was imported into mitochondria from MEF

cell lines WT (expressing mouse PNPASE,

mPNP), Pnpt1 knockout (KO), PNPT1 overexpres-

sion (expressing mPNP and hPNP), or Pnpt1

knockout plus PNPT1 overexpression (expressing

hPNP). Upper panel is an immunoblot for mouse

and human PNPASE expression. Middle panel is

an immunoblot of b-ACTIN, a loading control.

Lower panel is an autoradiogram of RNase P

RNA import into isolated MEF mitochondria.

(D) Radiolabeled, in vitro transcribed RNAs were

incubated with WT or HepKO liver mitochondria

for 10 min at 25�C. Nonimported RNA was

removed with nuclease, followed by RNA isolation

and separation on a urea-acrylamide gel. Import

reactions were repeated with 13 and 23 amounts

of synthesized RNAs. TOM40 immunoblot pro-

vides a mitochondrial loading control. See also

Figure S7.


predicted stem-loop structures were identified in two imported

RNAs that could transfer PNPASE-dependent import potential

to nonimported RNAs. Combined, these results open a new

chapter for studies into the pathway(s) and mechanism(s) of

RNA import into mammalian mitochondria.

A key question that this study presages is how PNPASE

regulates the import of specific cytosolic RNAs. For this, the

location of PNPASE first needs to be considered. Previously,

we localized PNPASE to the IMS. Carbonate extraction studies

also indicated that PNPASE was bound to the IM facing the

IMS (Chen et al., 2006; Rainey et al., 2006). However, more

recently, others have shown a weak interaction with the

matrix-localized RNA helicase hSUV3, suggesting a matrix

localization for PNPASE (Szczesny et al., 2009; Wang et al.,

2009). Our sub-fractionation studies were highly reproducible

and, in contrast, we favor the idea that hSUV3 bound to PNPASE

after all of the mitochondrial sub-compartments were exposed

during purification. The methods used to identify this interaction

do not exclude this possibility and we have already shown an

unanticipated interaction between PNPASE and the cytosol-

localized oncoprotein TCL1 using similar limited resolution

methods (French et al., 2007). Our failure to again identify a

hSUV3 and PNPASE interaction using an ultra-sensitive dual-

tag expression and purification system (Claypool et al., 2008)

further supports this interpretation (Figure S1). An alternative

explanation that we cannot exclude or confirm with current

methodologies is that a small amount of PNPASE could get

into the matrix and interact with hSUV3.

A second interesting area opened by the current results is to

determine how PNPASE controls RNA import into mitochondria.

In concept, PNPASE could import RNAs from the cytosol into

the IMS and then pass this RNA to another protein or complex

that would assist it through the IM into the matrix (Figure 4C).

Interestingly, PNPASE augments import in yeast, which does

not have a PNPASE homolog, indicating a distinct RNA import

mechanism that PNPASE can augment directly or indepen-

dently. In mouse liver and MEF mitochondria, PNPASE KO is

incomplete because cells without PNPASE are nonviable, so it

is unclear whether it is absolutely essential for all RNA import

or whether the amount detected in import assays is still mediated

by the minimal residual amount of PNPASE present required

for cell survival. Furthermore, it is not clear if this imported

RNA in PNPASE KO mitochondria is stuck in the IMS and not

in the matrix, which could provide further insight for the detailed

function of PNPASE in RNA import.

PNPASE has two external domains (KH and S1) that bind RNA

near the opening of a central processing pore in a trimeric

complex (Carpousis, 2002; Symmons et al., 2000). It is not clear

whether the same domains are used indiscriminately or in some

distinct manner to trigger PNPASE RNA processing versus

import functions. It is possible that the stem-loop structures

identified in RNase P and MRP RNAs interact with PNPASE

in a manner that triggers only import rather than processing.

Interestingly, GAPDH RNA can be a target of PNPASE degrada-

tion activity in vitro (French et al., 2007), although when either

of the identified stem-loop structures is appended to the

50-terminus, GAPDH RNA is efficiently imported into mitochon-

dria (Figure 6E). Indeed, RNA structural elements regulate

PNPASE function in chloroplasts and prokaryotes (Lisitsky

et al., 1996; Yehudai-Resheff and Schuster, 2000) and a stem-

loop structure protects the RNA from degradation by PNPASE

in chloroplasts (Yehudai-Resheff et al., 2001). A detailed dissec-

tion of what constitutes a trigger sequence for processing versus

import activities is clearly warranted. Finally, since the overall

sequence homology between the identified stem-loop struc-

tures on RNase P and MRP RNAs is not high, binding interactions

between specific RNAs and PNPASE may be stronger or

weaker, allowing or inhibiting detection by standard in vitro

techniques such as IP, which could make the verification of

additional candidate imported RNAs challenging without the

functional import assay system for validation.

Studies and important applications that use RNA import in

mammalian cells have been hampered by the inconsistent

requirements between systems and their study in vitro versus

in vivo. Although the exact mechanism for how PNPASE

augments and licenses RNA import is not yet known, the identifi-

cation of PNPASE represents the first receptor-like component

that binds RNA in mammalian cells to mediate RNA import into

the mitochondrial matrix. This finding, along with the identification

of import signal sequences, should open up additional studies to

determine what other pathway components are involved and

what the RNA sequence or structure rules tell us about how

PNPASE may decipher between processing and import.


Protein and RNA Purification

Purification of the PNPASE-HisPC protein complex was performed as before

(Claypool et al., 2008). For protein-RNA interactions, mitochondria (1 mg/ml)

were solubilized in lysis buffer (300 mM NaCl, 10 mM imidazole, 10% glycerol,

0.25% Triton X-100, 2 mM DTT, 20 mM HEPES [pH 6.6]) containing protease

inhibitor (Roche) and RNase inhibitor (Invitrogen). Insoluble material was

removed by spinning and extracts transferred to Eppendorf tubes. 50 ml of

Ni2+NTA resin (QIAGEN) was incubated in 1 ml lysis buffer with 100 mg/ml

ssDNA for 1h at 4�C. The resin was then mixed with the mitochondrial lysates

in the presence of 100 mg/ml ssDNA for 1 hr at 4�C. After incubation, the resin

was washed 103with lysis buffer containing RNase inhibitor. The protein-RNA

complex was eluted with elution buffer (300 mM NaCl, 10 mM imidazole, 10%

glycerol, 0.25% Triton X-100, 20 mM citrate [pH 5.5]) containing RNase

inhibitor. RNA was isolated from the eluate using TRIzol reagent (Invitrogen).

Isolation of Mitochondrial RNA and DNA

Mitochondria (1 mg/ml) were treated with 25 mg/ml of micrococcal nuclease S7

in nuclease buffer (0.6 M Sorbitol, 20 mM MgCl2, 5 mM CaCl2, 20 mM Tris

[pH 8.0]) for 30 min at 27�C. The reaction was stopped by addition of 20 mM

EGTA. Mitochondria were collected and solubilized in SDS buffer (100 mM

NaCl, 1% SDS, 20 mM Tris pH 7.4) at 65�C for 5 min. RNA was purified using

TRIzol reagent, and treated with RNase-free DNase I (Roche) for 1h at 37�C.

DNase I was inactivated by heating at 65�C for 10 min. Phenol-chloroform

(EM Science) extractions were used for DNA purification from the mitochon-

drial lysates.

In Vitro Transcription

RNAs were synthesized as before (Portnoy et al., 2008). For radiolabeled RNA

synthesis, [32P]-CTP (MP Biochemicals) was incorporated. The RNAs were

purified using TRIzol reagent.

RNA Import Assay

Yeast mitochondria were isolated from cells grown in selection medium until

stationary phase and mammalian mitochondria were isolated as previously

described (Chen et al., 2006; Rainey et al., 2006). In vitro RNA import assays


were performed in a 200-ml volume containing 0.5 mg of RNA, 100 mg of mito-

chondria, 0.6 M sorbitol, 2 mM KH2PO4, 50 mM KCl, 10 mM MgCl2, 2.5 mM

EDTA, 5 mM L-methionine, 1 mg/ml BSA, 5 mM ATP, 2 mM DTT, 5 mM

NADH, 50 mM HEPES, [pH 7.1], at RT for 10 min. Mitochondria were spun

at 11,000 x g for 5 min and washed once with wash buffer (0.6 M sorbitol,

20 mM Tris, [pH 8.0]). Mitochondria were spun again and resuspended in

200 ml nuclease buffer containing 25 mg/ml of micrococcal nuclease S7 and

incubated for 30 min at 27�C. Mitochondria were collected and solubilized in

SDS buffer at 65�C for 5 min. RNA was purified using TRIzol reagent. For

import into mammalian mitochondria, 0.25 M sucrose instead of 0.6 M sorbitol,

and 20 mM succinate instead of 5 mM NADH, were used. For import with

radiolabeled RNA, the purified RNAs were analyzed by urea-acrylamide gel

and autoradiography.

RNA Degradation Assay

The RNA processing activity of WT and mutant PNPASE was done as before

(Portnoy et al., 2008). [32P]-RNA was incubated with the corresponding

proteins in buffer E (20 mM HEPES, [pH 7.9], 60 mM KCl, 12.5 mM MgCl2,

0.1 mM EDTA, 2 mM DTT, and 17% glycerol, 0.1 mM Pi) at 25�C for 5 min.

Following incubation, the RNA was isolated and analyzed by urea-acrylamide

gel and autoradiography.

Additional Procedures

Osmotic shock was performed by incubating mitochondria for 30 min on ice in

0.03 M sorbitol and 20 mM HEPES-KOH, (pH 7.4) (Claypool et al., 2008). Blue

native gel electrophoresis was performed on a 6%–16% linear polyacrylamide

gradient using 50 mg of digitonin solubilized material (Chen et al., 2006).

Northern blotting was performed as previously described (Tollervey, 1987).

Total mtRNA was separated on a 12% agarose-formaldehyde gel and

transferred to a nylon membrane. Hybridization was carried out with

[32P]-dCTP (MP Biochemicals) labeled DNA probes. In organello protein

synthesis assays were performed as before with minor changes (Stuart and

Koehler, 2007). 100 mg mouse liver mitochondria were incubated in 100 ml

translation buffer (250 mM sucrose, 100 mM KCl, 1 mM MgCl2, 10 mM Tris

(pH 7.4), 10 mM K2HPO4 (pH 7.4), 10 mM glutamate, 10 mM malate, 5mM

NADH, 1 mM ADP, 1 mg/ml BSA, 100 mg/ml emetine, 100 mg/ml cyclohexi-

mide, and 30 mM of amino acid mix without methionine) with 5 ml of L-[35S]

methionine (MP Biochemicals) at 37�C for 30 min. The mitochondria were

then precipitated and proteins resolved by 12% SDS PAGE.



seven figures and can be found with this article online at doi:10.1016/j.cell.

2010.06.035.

ACKNOWLEDGMENTS

We thank Michelle Husain for expertise in TEM. Supported by NIH grants

R01GM061721 and R01GM073981 (C.M.K.), R01CA90571 (M.A.T.) and

PN2EY018228 (Roadmap for Medical Research Nanomedicine Initiative)

(M.A.T.), and K22CA120147 (S.W.F.), the Intramural Research Program of

the NIH, National Institute of Allergy and Infectious Diseases (H.C.M. III), the

Muscular Dystrophy Association 022398 (C.M.K.), the American Heart Associ-

ation 0640076N (C.M.K.), and the California Institute of Regenerative Medicine

(RS1-00313 and RB1-01397, M.A.T.). C.M.K. is an Established Investigator of

the American Heart Association and M.A.T. is a Scholar of the Leukemia and

Lymphoma Society.


Revised: March 20, 2010



REFERENCES

Alfonzo, J.D., and Soll, D. (2009). Mitochondrial tRNA import–the challenge to

understand has just begun. Biol. Chem. 390, 717–722.

Bonawitz, N.D., Clayton, D.A., and Shadel, G.S. (2006). Initiation and beyond:

multiple functions of the human mitochondrial transcription machinery.

Mol. Cell 24, 813–825.

Carpousis, A.J. (2002). The Escherichia coli RNA degradosome: structure,

function and relationship in other ribonucleolytic multienzyme complexes.

Biochem. Soc. Trans. 30, 150–155.

Chacinska, A., Koehler, C.M., Milenkovic, D., Lithgow, T., and Pfanner, N.

(2009). Importing mitochondrial proteins: machineries and mechanisms. Cell

138, 628–644.

Chang, D.D., and Clayton, D.A. (1987). A mammalian mitochondrial RNA

processing activity contains nucleus-encoded RNA. Science 235, 1178–1184.

Chen, H.W., Koehler, C.M., and Teitell, M.A. (2007). Human polynucleotide

phosphorylase: location matters. Trends Cell Biol. 17, 600–608.

Chen, H.W., Rainey, R.N., Balatoni, C.E., Dawson, D.W., Troke, J.J., Wasiak,

S., Hong, J.S., McBride, H.M., Koehler, C.M., Teitell, M.A., and French, S.W.

(2006). Mammalian polynucleotide phosphorylase is an intermembrane

space RNase that maintains mitochondrial homeostasis. Mol. Cell. Biol. 26,

8475–8487.

Claypool, S.M., Oktay, Y., Boontheung, P., Loo, J.A., and Koehler, C.M. (2008).

Cardiolipin defines the interactome of the major ADP/ATP carrier protein of the

mitochondrial inner membrane. J. Cell Biol. 182, 937–950.

Diaz, F., Garcia, S., Hernandez, D., Regev, A., Rebelo, A., Oca-Cossio, J., and

Moraes, C.T. (2008). Pathophysiology and fate of hepatocytes in a mouse

model of mitochondrial hepatopathies. Gut 57, 232–242.

Doersen, C.J., Guerrier-Takada, C., Altman, S., and Attardi, G. (1985).

Characterization of an RNase P activity from HeLa cell mitochondria. Compar-

ison with the cytosol RNase P activity. J. Biol. Chem. 260, 5942–5949.

Duchene, A.M., Pujol, C., and Marechal-Drouard, L. (2009). Import of tRNAs

and aminoacyl-tRNA synthetases into mitochondria. Curr. Genet. 55, 1–18.

Entelis, N.S., Kolesnikova, O.A., Martin, R.P., and Tarassov, I.A. (2001). RNA

delivery into mitochondria. Adv. Drug Deliv. Rev. 49, 199–215.

French, S.W., Dawson, D.W., Chen, H.W., Rainey, R.N., Sievers, S.A.,

Balatoni, C.E., Wong, L., Troke, J.J., Nguyen, M.T., Koehler, C.M., and Teitell,

M.A. (2007). The TCL1 oncoprotein binds the RNase PH domains of the

PNPase exoribonuclease without affecting its RNA degrading activity. Cancer

Lett. 248, 198–210.

Hollingsworth, M.J., and Martin, N.C. (1986). RNase P activity in the mitochon-

dria of Saccharomyces cerevisiae depends on both mitochondrion and

nucleus-encoded components. Mol. Cell. Biol. 6, 1058–1064.

Holzmann, J., Frank, P., Loffler, E., Bennett, K.L., Gerner, C., and Rossmanith,

W. (2008). RNase P without RNA: identification and functional reconstitution of

the human mitochondrial tRNA processing enzyme. Cell 135, 462–474.

Koehler, C.M., Jarosch, E., Tokatlidis, K., Schmid, K., Schweyen, R.J., and

Schatz, G. (1998). Import of mitochondrial carriers mediated by essential

proteins of the intermembrane space. Science 279, 369–373.

Kolesnikova, O.A., Entelis, N.S., Jacquin-Becker, C., Goltzene, F., Chrzanow-

ska-Lightowlers, Z.M., Lightowlers, R.N., Martin, R.P., and Tarassov, I. (2004).

Nuclear DNA-encoded tRNAs targeted into mitochondria can rescue a

mitochondrial DNA mutation associated with the MERRF syndrome in cultured

human cells. Hum. Mol. Genet. 13, 2519–2534.

Kren, B.T., Wong, P.Y., Sarver, A., Zhang, X., Zeng, Y., and Steer, C.J. (2009).

MicroRNAs identified in highly purified liver-derived mitochondria may play

a role in apoptosis. RNA Biol. 6, 65–72.

Lisitsky, I., Klaff, P., and Schuster, G. (1996). Addition of destabilizing poly

(A)-rich sequences to endonuclease cleavage sites during the degradation

of chloroplast mRNA. Proc. Natl. Acad. Sci. USA 93, 13398–13403.

Mandel, H., Hartman, C., Berkowitz, D., Elpeleg, O.N., Manov, I., and Iancu,

T.C. (2001). The hepatic mitochondrial DNA depletion syndrome: ultrastruc-

tural changes in liver biopsies. Hepatology 34, 776–784.


Mukherjee, S., Basu, S., Home, P., Dhar, G., and Adhya, S. (2007). Necessary

and sufficient factors for the import of transfer RNA into the kinetoplast

mitochondrion. EMBO Rep. 8, 589–595.

Portnoy, V., Palnizky, G., Yehudai-Resheff, S., Glaser, F., and Schuster, G.

(2008). Analysis of the human polynucleotide phosphorylase (PNPase) reveals

differences in RNA binding and response to phosphate compared to its

bacterial and chloroplast counterparts. RNA 14, 297–309.

Postic, C., and Magnuson, M.A. (2000). DNA excision in liver by an albumin-

Cre transgene occurs progressively with age. Genesis 26, 149–150.

Puranam, R.S., and Attardi, G. (2001). The RNase P associated with HeLa cell

mitochondria contains an essential RNA component identical in sequence to

that of the nuclear RNase P. Mol. Cell. Biol. 21, 548–561.

Rainey, R.N., Glavin, J.D., Chen, H.W., French, S.W., Teitell, M.A., and

Koehler, C.M. (2006). A new function in translocation for the mitochondrial

i-AAA protease Yme1: import of polynucleotide phosphorylase into the inter-

membrane space. Mol. Cell. Biol. 26, 8488–8497.

Rubio, M.A., Rinehart, J.J., Krett, B., Duvezin-Caubet, S., Reichert, A.S., Soll,

D., and Alfonzo, J.D. (2008). Mammalian mitochondria have the innate ability to

import tRNAs by a mechanism distinct from protein import. Proc. Natl. Acad.

Sci. USA 105, 9186–9191.

Smirnov, A.V., Entelis, N.S., Krasheninnikov, I.A., Martin, R., and Tarassov, I.A.

(2008). Specific features of 5S rRNA structure - its interactions with macromol-

ecules and possible functions. Biochemistry (Mosc.) 73, 1418–1437.

Stuart, R.A., and Koehler, C.M. (2007). In vitro analysis of yeast mitochondrial

protein import. Curr. Protoc. Cell Biol. Chapter 11, Unit 11.19.

Symmons, M.F., Jones, G.H., and Luisi, B.F. (2000). A duplicated fold is the

structural basis for polynucleotide phosphorylase catalytic activity, processiv-

ity, and regulation. Structure 8, 1215–1226.

Symmons, M.F., Williams, M.G., Luisi, B.F., Jones, G.H., and Carpousis, A.J.

(2002). Running rings around RNA: a superfamily of phosphate-dependent

RNases. Trends Biochem. Sci. 27, 11–18.

Szczesny, R.J., Borowski, L.S., Brzezniak, L.K., Dmochowska, A., Gewartow-

ski, K., Bartnik, E., and Stepien, P.P. (2009). Human mitochondrial RNA

turnover caught in flagranti: involvement of hSuv3p helicase in RNA surveil-

lance. Nucleic Acids Res. 38, 279–298.

Tarassov, I., Entelis, N., and Martin, R.P. (1995). An intact protein translocating

machinery is required for mitochondrial import of a yeast cytoplasmic tRNA. J.

Mol. Biol. 245, 315–323.

Tollervey, D. (1987). A yeast small nuclear RNA is required for normal process-

ing of pre-ribosomal RNA. EMBO J 6, 4169–4175.

Walker, S.C., and Engelke, D.R. (2008). A protein-only RNase P in human mito-

chondria. Cell 135, 412–414.

Wang, D.D., Shu, Z., Lieser, S.A., Chen, P.L., and Lee, W.H. (2009).

Human Mitochondrial SUV3 and Polynucleotide Phosphorylase Form a

330-kDa Heteropentamer to Cooperatively Degrade Double-stranded RNA

with a 30-to-50 Directionality. J. Biol. Chem. 284, 20812–20821.

Yehudai-Resheff, S., Hirsh, M., and Schuster, G. (2001). Polynucleotide

phosphorylase functions as both an exonuclease and a poly(A) polymerase

in spinach chloroplasts. Mol. Cell. Biol. 21, 5408–5416.

Yehudai-Resheff, S., and Schuster, G. (2000). Characterization of the E.coli

poly(A) polymerase: nucleotide specificity, RNA-binding affinities and RNA

structure dependence. Nucleic Acids Res. 28, 1139–1144.


Plzf Regulates Germline ProgenitorSelf-Renewal by Opposing mTORC1Robin M. Hobbs,1 Marco Seandel,2 Ilaria Falciatori,3 Shahin Rafii,3 and Pier Paolo Pandolfi1,*1Cancer Genetics Program, Beth Israel Deaconess Cancer Center, Departments of Medicine and Pathology, Beth Israel Deaconess

Medical Center, Harvard Medical School, Boston, MA 02115, USA2Department of Surgery3Howard Hughes Medical Institute, Department of Genetic MedicineWeill Cornell Medical College, New York, NY 10065, USA


DOI 10.1016/j.cell.2010.06.041

SUMMARY

Hyperactivity of mTORC1, a key mediator of cellgrowth, leads to stem cell depletion, although theunderlying mechanisms are poorly defined. Usingspermatogonial progenitor cells (SPCs) as a modelsystem, we show that mTORC1 impairs stem cellmaintenance by a negative feedback from mTORC1to receptors required to transduce niche-derivedsignals. We find that SPCs lacking Plzf, a transcrip-tion factor essential for SPC maintenance, haveenhanced mTORC1 activity. Aberrant mTORC1 acti-vation in Plzf �/� SPCs inhibits their response toGDNF, a growth factor critical for SPC self-renewal,via negative feedback at the level of the GDNFreceptor. Plzf opposes mTORC1 activity by inducingexpression of the mTORC1 inhibitor Redd1. Thus, weidentify the mTORC1-Plzf functional interaction asa critical rheostat for maintenance of the spermato-gonial pool and propose a model whereby negativefeedback from mTORC1 to the GDNF receptorbalances SPC growth with self-renewal.

INTRODUCTION

Maintenance of a wide array of adult tissues is dependent on the

presence of a resident stem cell pool with self-renewal potential

that generates differentiating progeny. Factors regulating the

balance between stem cell self-renewal and differentiation

ensure tissue homeostasis whereas disruption of these regula-

tory mechanisms can lead to tissue degeneration or cancer (Ito

et al., 2009). One factor central to stem cell homeostasis is

mammalian TOR complex 1 (mTORC1), a signaling complex

that promotes protein translation and cell growth by phosphory-

lating components of the translation machinery (Ma and Blenis,

2009). mTORC1 is regulated in response to diverse stimuli

including nutrient availability, energy status, growth factors,

and cellular stress. Persistent mTORC1 activation in certain

tissues leads to increased proliferation but subsequent exhaus-

tion of the stem cell compartment, demonstrating that aberrantly

activated mTORC1 is detrimental to stem cell maintenance

(Castilho et al., 2009; Gan and DePinho, 2009; Yilmaz et al.,

2006). It is proposed that inappropriate mTORC1 activation

drives stem cell depletion through aberrant translation of down-

stream targets and subsequent activation of tumor suppressive/

fail-safe mechanisms resulting in cellular senescence or apo-

ptosis (Ito et al., 2009). However, the molecular mechanisms

and targets of mTORC1 in this context are currently unknown.

Interestingly, inhibition of mTORC1 also extends organism life-

span (Harrison et al., 2009; Schieke and Finkel, 2006), consistent

with the notion that declining stem cell potential underlies aging

(Rossi et al., 2008).

Undifferentiated germline cells of the testis (spermatogonial

progenitor cells; SPCs) are formed from gonocytes during post-

natal development of the mouse testis and possess self-renewal

potential (de Rooij and Russell, 2000). A major advance in the

study of male germline biology was the development of culture

systems allowing long-term SPC expansion in vitro while main-

taining stem cell potential. Key to this was the observation that

mice heterozygous for the glial cell-derived neurotrophic factor

(GDNF) cytokine gene had a depletion of SPC activity (Meng

et al., 2000). GDNF is produced by Sertoli cells within the testis,

and signals via the GFRa1/c-Ret receptor to promote SPC self-

renewal and growth through activation of Src family kinases and

Akt (Lee et al., 2007b; Oatley et al., 2007). Culture of SPCs with

GDNF plus a variety of additional factors (including basic fibro-

blast growth factor; bFGF) preserves self-renewal capabilities

and allows essentially unlimited cell expansion while maintaining

in vivo differentiation potential; assessed by the ability to repopu-

late depleted recipient testis (Kanatsu-Shinohara et al., 2003;

Kubota et al., 2004; Seandel et al., 2007). Although some cellular

signaling pathways involved in SPC self-renewal have been

described, it remains unclear how SPCs integrate signals from

general mitogenic stimuli with those required for self-renewal

to balance stem cell maintenance and differentiation.

A limited number of cell intrinsic factors have also been impli-

cated in SPC function, foremost among which is promyelocytic

leukemia zinc finger (PLZF). PLZF was identified from the trans-

location breakpoint in t(11;17) acute promyelocytic leukemia

(Chen et al., 1993) and encodes a transcription factor belonging

to the POZ-Kruppel (POK) family. PLZF binds DNA through

carboxy-terminal Kruppel-type zinc fingers and recruits histone


deacetylases (HDACs) via an amino-terminal POZ domain (David

et al., 1998). Recruitment of HDACs to target promoters can

result in gene repression although PLZF is also able to activate

gene expression (Doulatov et al., 2009; Labbaye et al., 2002).

Male mice lacking Plzf expression undergo progressive germ

cell loss and testis atrophy with age causing infertility (Buaas

et al., 2004; Costoya et al., 2004). Plzf is expressed by SPCs

and is needed in a cell autonomous fashion for maintenance

of the germ lineage. A male patient with biallelic PLZF loss-of-

function and gonad hypoplasia has been recently reported

(Fischer et al., 2008), emphasizing the role played by PLZF in

germ cell biology.

SPC maintenance is dependent on Plzf plus key growth

factors such as GDNF. As mTORC1 is activated in response to

growth factor signaling we hypothesized that both Plzf and

mTORC1 are involved in the GDNF response of SPCs and that

Plzf and mTORC1 may crosstalk. To assess the roles of Plzf

and mTORC1 in GDNF-dependent SPC maintenance, we devel-

oped systems for isolation and culture of SPCs from wild-type

(WT) and Plzf �/� prepubertal mice (subsequent to formation

of the SPC pool and prior to overt germ cell depletion in the

Plzf �/� mouse). We find that Plzf opposes mTORC1 activity

and define a cellular signaling network controlling SPC homeo-

stasis whereby mTORC1 integrates mitogenic signals received

by SPCs and determines their sensitivity to self-renewal stimuli.

RESULTS

Isolation, Culture, and Comparative Analysisof Plzf +/+ and Plzf �/� SPCsThe testis is composed of a heterogeneous mix of mitotic and

meiotic germ cells plus multiple types of somatic cells. SPCs

are rare within the testis and systems for their purification are

poorly developed. We therefore sought to take advantage of

the expression of Plzf in SPCs to enrich for this cell type. Plzf-

expressing cells were identified from juvenile testis by intracel-

lular staining and flow cytometry using a newly developed

monoclonal antibody (Figure 1A). Our aim was to correlate the

Plzf-expressing population to sets of cell surface markers that

would allow subsequent purification of those live cells. By adapt-

ing markers able to enrich for stem cell activity from cryptorchid

testis (Kubota et al., 2003; Shinohara et al., 2000), we found

that the Plzf-positive population was av-integrin negative and

expressed low levels of Thy-1 (Figure 1B). Negative selection

for av-integrin combined with Thy-1 positive selection allowed

isolation of Plzf-positive cells and minimized somatic cell

contamination (Figures S1A and S1B available online) (Virtanen

et al., 1986). c-Kit expression is associated with SPC differentia-

tion (Schrans-Stassen et al., 1999) and the avneg Thy-1low popu-

lation was composed primarily of c-Kit negative cells with

a smaller c-Kit positive fraction (Figure 1C, left). The c-Kitneg cells

were largely in G1/G0 phase of cell cycle whereas the c-Kitpos

population contained more cells in S plus G2 phases (Figure 1C,

right). As SPCs are more quiescent than differentiating sper-

matogonia (Takubo et al., 2008), the avneg Thy-1low population

contains both SPCs (c-Kitneg, quiescent) and cells undergoing

differentiation (c-Kitpos, proliferating). To confirm that the avneg

Thy-1low c-Kitneg fraction was enriched for stem cells, we

compared the ability of avneg Thy-1low c-Kitneg and unfractio-

nated cells to repopulate recipient depleted testis (Ogawa

et al., 1997). Donor cells were isolated from transgenic mice

expressing GFP from the b-actin promoter, allowing visualization

of donor-derived colonies (Figure 1D, left). Numbers of GFP-

positive colonies were scored 2 months posttransplant (Fig-

ure 1D, right), representing numbers of engrafted stem cells.

Stem cell activity was enriched �25-fold in the avneg Thy-1low

c-Kitneg population compared to unfractionated cells. As the

avneg Thy-1low c-Kitneg fraction represents 3%–5% of total

cells (�1/25th of testis cellularity) from 10 to 14 days postnatal

testis (Figures 1B, 1C, and 1G), a 25-fold enrichment of stem

cell activity indicates that almost all stem cells are contained

within this fraction. Given the number of colonies obtained

from avneg Thy-1low c-Kitneg cells (Figure 1D) and an estimated

10% engraftment efficiency (Nagano et al., 1999), there are

1265 stem cells per 105 cells in this fraction (�1 in 80 cells

have stem cell potential).

We next analyzed prepubertal Plzf +/+ and Plzf �/� testis by

flow-cytometry to determine if Plzf �/� mice display aberrant

SPC activity before overt germ cell depletion. Although the char-

acteristic av/Thy-1 flow profile was present in juvenile Plzf �/�

testis (2 and 3 weeks postnatal), the avneg Thy-1low population

was depleted relative to the WT (Figure 1E and Figure S1C)

and the percentage of c-Kitpos cells within the Plzf �/� avneg

Thy-1low fraction was increased (Figure 1F and Figure S1D).

The decreased frequency and numbers of avneg Thy-1low c-Kitneg

cells in Plzf �/� testis (Figure 1G) confirms that Plzf loss is detri-

mental to SPCs whereas the increase in c-Kitpos cells within the

avneg Thy-1low fraction is suggestive of increased differentiation

commitment. As Plzf directly represses c-Kit expression

(Filipponi et al., 2007), this can be reflective of a de-repression

of the c-Kit locus. However, the Plzf �/� avneg Thy-1low fraction

still contains c-Kitneg cells (Figure 1F and Figure S1D), indicating

that aberrant c-Kit expression is unlikely to fully explain the

Plzf �/� phenotype. Additionally, both Plzf +/+ and Plzf �/� avneg

Thy-1low c-Kitneg cells show substantial enrichment in expres-

sion of the SPC-associated factors Pou5f1/Oct4, Ngn3, Bcl6b,

and Pou3f1 (Figure 1H and Figure S1E) (Oatley et al., 2006;

Ohbo et al., 2003; Wu et al., 2010; Yoshida et al., 2006), confirm-

ing the identity of this fraction. We also noted that Plzf �/� avneg

Thy-1low cells had decreased levels of b1 and a6 integrins but

equivalent levels of CD9 compared to controls (Figures S1F

and S1G) (Kanatsu-Shinohara et al., 2004b; Shinohara et al.,

1999). Our data indicate that SPCs from juvenile Plzf �/� mice

have an impaired ability to maintain an undifferentiated state,

which precedes overt germ cell loss.

We next attempted in vitro culture of isolated Plzf +/+ and

Plzf �/� SPCs, adapting previously developed culture systems

(Kubota et al., 2004; Seandel et al., 2007). As anticipated, germ

cell colony-forming activity under SPC culture conditions was

essentially entirely contained within the avneg Thy-1low fraction

(Figure S2A). We successfully derived SPC lines from both

Plzf +/+ and Plzf �/� avneg Thy-1low cells, which grew as discrete

colonies on mouse embryonic fibroblast (MEF) feeder cells

(Figure 2A) and maintained their cell surface marker identity

during exponential growth over 1 year of culture (Figure S2B).

WT SPC lines expressed Plzf (Figure 2B) whereas Plzf �/� SPC


lines were more difficult to establish (not shown) consistent with

a role for Plzf in SPC function. Early passages of Plzf �/� SPCs

had a slower growth rate than WT cells (Figures S2C and S2D)

but could be maintained long-term and growth became more

comparable to WT cells at later passages (Figure 5C). The ability

to culture Plzf �/� SPCs seemed counterintuitive, but the gener-

ation of these cultures became critical for further dissection of

Plzf function and to the solution of this apparent contradiction.

An additional indicator of SPC potential is the ability to derive

embryonic stem-like cells from the cultured lines (multipotent

adult spermatogonial-derived stem cells; MASCs) (Kanatsu-

Shinohara et al., 2004a; Seandel et al., 2007). We observed

spontaneous formation of MASC colonies from two of seven

WT lines (Figure S2E). MASC formation was not observed in

four Plzf �/� SPC lines, possibly indicating a defective capability

to generate MASCs. The multipotent capabilities of MASC lines

were verified by teratoma formation assay (Figure S2F). On SPC

to MASC conversion, expression of the pluripotency-associated

factors Pou5f1/Oct4, Sox2 and Nanog were substantially

increased whereas Plzf expression was lost (Figure S2G and

data not shown) (Seandel et al., 2007). Formation of MASCs indi-

cates that SPC potential is maintained in our culture system,

Figure 1. Isolation and Analysis of Plzf +/+

and Plzf �/� SPCs

(A) Detection of Plzf-expressing spermatogonia

from juvenile testis (10–14 days postnatal) by intra-

cellular staining and flow cytometry using anti-Plzf

antibody.

(B) Plzf expressing cells correlate to a discrete av-

integrin negative, Thy-1 low population by flow cy-

tometry. CD45 marker excludes contaminating

leukocytes.

(C) Left panel: flow cytometric analysis of c-Kit

within the Plzf-expressing av-integrin negative,

Thy-1 low population. Right panels: cell cycle

status of c-Kit negative and positive populations

plus percentage of cells in G1/S/G2 cell cycle

phases from a representative sample.

(D) Left panels: seminiferous tubules repopulated

with donor GFP-positive av integrin negative,

Thy-1 low, c-Kit negative cells in testis transplant

assay. Fluorescent image (top) and brightfield

image of same recipient testis (bottom) are shown.

Right panel: numbers of GFP positive colonies ob-

tained from av integrin negative, Thy-1 low, c-Kit

negative (sorted) and unsorted testis cells in recip-

ient testes 2 months posttransplant. Data is pre-

sented as mean number of colonies per 1 3 105

donor cells together with standard error of the

mean (SEM). Fold enrichment of stem cell activity

in sorted populations is indicated (***p < 0.0001).

Values are averaged from two independent exper-

iments. Thirteen recipient testes were analyzed for

sorted populations and 12 for unsorted.

(E) Representative av-integrin/Thy-1 flow profiles

of Plzf +/+ and Plzf �/� littermate mouse testes at

2 weeks postnatal age. The av-integrin negative,

Thy-1 low gate, containing Plzf-positive cells in

WT testis is indicated.

(F) Representative flow cytometry analysis of c-Kit

expression within the av-integrin negative, Thy-1

low testis cell fractions from (E).

(G) Quantification of frequency and absolute

numbers of av integrin negative, Thy-1 low, c-Kit

negative testis cells from mice of the indicated

postnatal ages and Plzf genotypes (n = 3 per geno-

type and age group, **p < 0.002, *p < 0.02).Graphs

represent mean values ± SEM.

(H) Quantitative RT-PCR analysis of Pou5f1 mRNA

expression in testis cell populations from juvenile

Plzf +/+ and Plzf �/� littermate mice: total (unfrac-

tionated testis cells), SPCs (av-integrin negative, Thy-1 low, c-Kit negative), early differentiating spermatogonia (av-integrin negative, Thy-1 low, c-Kit positive),

late differentiating spermatogonia (av-integrin negative, Thy-1 negative, c-Kit positive), and somatic cells (av-integrin positive). mRNA levels are normalized to

those of b-actin and standard deviations from duplicate reactions are shown.

See also Figure S1.


whereas the apparent inability of Plzf �/� SPCs to form MASCs

can be consistent with their defective function.

Plzf �/� SPCs Show Enhanced mTORC1 ActivityFrom analysis of cultured SPC lines, we noticed that Plzf �/� cells

were physically larger than WT SPCs, measured by the FSC

parameter of flow cytometry (Figures 2C and 2D). Alterations in

cell size are associated with changes in activity of mTORC1

acting through its downstream targets S6Kinase1 (S6K1), which

phosphorylates ribosomal S6 protein (RPS6), and 4EBP1 (Fingar

et al., 2002; Ruvinsky et al., 2005). Given the role of mTORC1 in

hematopoietic stem cells (Gan and DePinho, 2009), we consid-

ered that aberrant mTORC1 activity in Plzf �/� SPCs could

contribute to their defective maintenance. Importantly, Plzf �/�

SPCs had elevated levels of phosphorylated RPS6 compared

to WT cells, confirming increased mTORC1 activity (Figure 2E

and Figure 3C). Inhibition of mTORC1 with rapamycin decreased

the size of Plzf �/� SPCs to that of WT cells suggesting that

elevated mTORC1 activity was responsible for the size increase

(Figure 2F). Rapamycin was confirmed to inhibit RPS6 phos-

phorylation in Plzf �/� SPCs (Figure 2E). Increased mTORC1

activity in Plzf �/� SPCs was associated with elevated levels of

cellular protein (Figure S2H), consistent with the role of mTORC1

in regulating protein translation. Importantly, freshly isolated

Plzf �/� SPCs were also physically larger (Figures 2G and 2H)

and rapamycin treatment of Plzf �/� mice normalized SPC size

(Figures 2G and 2H). Together, our data indicate that Plzf inhibits

mTORC1 activation in SPCs.

Growth Factor-Mediated Regulation of mTORC1 in SPCsGiven the possibility that mTORC1 activity can regulate SPC

function, we next defined key regulatory inputs of mTORC1 in

WT SPCs. Growth factor signaling represents a major activating

input of mTORC1 (Ma and Blenis, 2009; Shaw and Cantley,

2006) and mTORC1 inhibition interferes with mitogenic stimuli

causing cell cycle arrest (Brown et al., 1994). Indeed, rapamycin

Figure 2. Increased mTORC1 Pathway

Activity in Plzf �/� SPCs

(A) SPC lines established from culture of av-integ-

rin negative, Thy-1 low cells from juvenile Plzf +/+

and Plzf �/� littermate mice immunostained for

the germ cell marker Mvh and counterstained

with DAPI to detect DNA. Scale bar represents

100 mm.

(B) Cultured SPCs of the indicated genotype

analyzed for Plzf expression by intracellular stain-

ing and flow cytometry (left panel) and by western

blot (right panel). The germ cell marker Mvh plus

actin are used as loading controls.

(C) Analysis of SPC size by flow cytometry.

Forward-scatter (FSC) profiles of two indepen-

dently derived sets of littermate Plzf +/+ and

Plzf �/� SPC lines (1 and 2) are shown.

(D) Quantification of flow cytometry profiles from

(C) showing averaged mean FSC with standard

deviation and p value.

(E) Control (DMSO) and rapamycin-treated Plzf +/+

and Plzf �/� SPCs were fixed and permeabilized

48 hr post-treatment and levels of Phospho-

RPS6 analyzed by flow cytometry.

(F) Normalization of increased Plzf �/� SPC size by

mTORC1 inhibition. FSC profiles of Plzf �/� SPCs

treated with DMSO (control) or rapamycin for

48 hr compared to untreated Plzf +/+ SPCs.

(G) Overlay of representative av-integrin negative,

Thy-1 low, c-Kit negative testis cell fraction FSC

profiles from Plzf +/+ and Plzf �/� littermate mice

treated from 10 days postnatal age for a period

of 1 week with vehicle or rapamycin.

(H) Quantification of relative mean FSC values of

av-integrin negative, Thy-1 low, c-Kit negative

testis cells of mice treated as in (G) (n = 4 per geno-

type and treatment group, *p < 0.05, **p < 0.01).

Horizontal bars represent mean values.

See also Figure S2.


suppressed SPC colony growth and caused an accumulation of

cells in G1 phase of the cell cycle (Figures S3A and S3B).

Growth factor receptors activate mTORC1 by signaling

through phosphoinositide 3-kinase (PI3K)/Akt and Ras/Erk

MAPK, which inhibit the TSC1/TSC2 complex (Huang and

Manning, 2008); TSC1/TSC2 negatively regulates mTORC1 via

its GTPase-activating protein activity toward the small G protein

Rheb. Treatment of WT SPCs with PI3K/Akt or Erk MAPK

pathway inhibitors significantly reduced levels of phosphory-

lated RPS6 (Figure 3A), indicating that efficient mTORC1 activa-

tion in SPCs requires both PI3K/Akt and Erk MAPK. Inhibition of

Src family kinases, implicated in the GDNF response of SPCs

(Oatley et al., 2007), had modest effects on phospho-RPS6

(Figure 3A). Therefore, despite the ability of Src family kinases

to activate PI3K/Akt downstream GDNF (Encinas et al., 2001;

Oatley et al., 2007), this SPC self-renewal pathway couples inef-

ficiently to mTORC1.

SPCs are maintained with a cocktail of growth factors

(Kanatsu-Shinohara et al., 2003; Seandel et al., 2007) and are

responsive to others (Hamra et al., 2007); thus we next assessed

contributions of these factors to mTORC1 activation in WT

SPCs. We found that GDNF triggered a modest phosphorylation

of RPS6 whereas other mitogens (including bFGF) induced much

Figure 3. Upstream Pathways and Growth Factors

Regulating mTORC1 in SPCs

(A) WT SPCs fed with complete SPC medium containing

inhibitors to the indicated signaling pathways were har-

vested 5 hr after treatment and analyzed by western blot

for the indicated proteins and phospho (P)-proteins.

(B) WT SPCs were starved overnight in basal SPC medium

lacking cytokine supplements. The indicated cytokines

were then added and SPCs harvested 20 min later then

analyzed by western blot. Cytokines were used at the

concentrations present in complete SPC medium. Neure-

gulin1 (NRG1) is not a standard supplement for SPC

medium although SPCs are responsive to it (see text).

(C) Western blot analysis of two independently derived

sets of littermate Plzf +/+ and Plzf �/� SPC lines (1 and 2)

under steady-state conditions is shown to the left. Quanti-

fication of the relative levels of phospho-Akt, Erk, and

RPS6 (corrected to the total levels of respective proteins)

are shown in panels on the right.

See also Figure S3.

higher levels of phospho-RPS6 (Figure 3B). The

relative abilities of GDNF and bFGF to activate

mTORC1 correlated to their efficiency of Akt

and Erk activation (Figure 3B). However, this

correlation did not hold true for epidermal

growth factor (EGF) stimulation, possibly re-

flecting additional pathways by which EGF acti-

vates mTORC1 (Fan et al., 2009). In summary,

the self-renewal signal GDNF poorly activates

mTORC1 when compared to more general

mitogens for SPCs (bFGF, EGF), in agreement

with the Src kinase inhibition data (see above).

However, we note that although GDNF is

considered the key SPC self-renewal signal,

SPC expansion and self-renewal in vitro requires a combination

of GDNF plus other factors (e.g., bFGF) that stimulate mTORC1

more efficiently (Lee et al., 2007b).

Plzf Uncouples mTORC1 Activity from Growth FactorSignaling in SPCs through Redd1 ModulationGiven that mTORC1 activity in SPCs is regulated by growth

factors present in the culture medium and resultant PI3K/Akt

and Erk-MAPK activation, we next assessed whether the

elevated mTORC1 activity of Plzf �/� SPCs correlated with an

enhanced growth factor response. However, in comparison to

WT cells, the Akt and Erk activities of steady-state Plzf �/�

SPCs were decreased whereas phospho-RPS6 was increased

(Figure 3C), indicating that the elevated mTORC1 activity is not

due to increased activity of growth factor pathways.

We next considered alternative mechanisms by which

mTORC1 activity would be increased in Plzf �/� SPCs. Cellular

stress-response pathways provide additional regulatory inputs

resulting in mTORC1 inhibition when conditions such as low

energy availability or hypoxia require temporary arrest of cell

growth (Huang and Manning, 2008). We hypothesized that

mTORC1 hyperactivity in Plzf �/� SPCs could be due to reduced

activity of stress response pathways. Accordingly, we first


assessed levels of upstream regulatory proteins as stress

pathways converge on mTORC1 at the level of TSC1/TSC2.

However, we did not find significant differences in levels of

Tsc1, Tsc2, Rheb, or the mTOR kinase itself in Plzf �/� SPCs

compared to WT cells (Figure 4A).

By contrast, we noticed that levels of Redd1 (also known as

Ddit4, Rtp801, and Dig2) were substantially lower in Plzf �/�

SPCs compared to controls, at both protein and mRNA levels

(Figures 4A and 4B). Redd1 is induced by multiple types of cell

stress (e.g., hypoxia, DNA damage) and by developmental

signals and inhibits mTORC1 through regulation of TSC1/TSC2

(Brugarolas et al., 2004; Corradetti et al., 2005; Ellisen et al.,

2002). As a role for Redd1 in SPCs is not described, we analyzed

the distribution of Redd1 expression in isolated spermatogonial

fractions and found a relative enrichment of Redd1 mRNA

in SPCs (Figure 4C). We also confirmed that freshly isolated

Plzf �/�SPCs had lower Redd1 expression compared to controls

(Figure 4C). Furthermore, shRNA knockdown of Redd1 in WT

SPCs increased mTORC1 activity (Figure 4D and Figure S4).

Together, our data suggest that Plzf opposes mTORC1 by main-

taining Redd1 expression and that Redd1 is a key mTORC1

regulator in SPCs.

Plzf Is a Transcriptional Activator of the mTORC1Inhibitor Redd1 in SPCsRedd1 expression was decreased at the mRNA level in

Plzf �/� SPCs. We therefore tested whether Plzf could directly

induce Redd1 expression. We first performed a chromatin

Figure 4. Plzf Regulates Expression of the

mTORC1 Pathway Inhibitor Redd1

(A) Western blot analysis for components of the

mTORC1 pathway in two independently derived

sets of littermate Plzf +/+ and Plzf �/� SPC lines

(1 and 2) under steady-state conditions.

(B) Quantitative RT-PCR analysis of Redd1 mRNA

expression in independently derived Plzf +/+ and

Plzf �/� SPC lines. mRNA levels are normalized

to those of b-actin and standard deviations from

duplicate reactions are shown.

(C) Quantitative RT-PCR analysis of Redd1 mRNA

expression in spermatogonial fractions from

juvenile Plzf +/+ and Plzf �/� littermate mice:

SPCs (av-integrin negative, Thy-1 low, c-Kit nega-

tive), early differentiating spermatogonia (av-

integrin negative, Thy-1 low, c-Kit positive), late

differentiating spermatogonia (av-integrin nega-

tive, Thy-1 negative, c-Kit positive). mRNA levels

are normalized to those of b-actin and standard

deviations from duplicate reactions are shown.

Dashed line indicates Redd1 expression levels in

unfractionated WT testis cells.

(D) WT SPCs infected with control shRNA (against

GFP) or two independent shRNA constructs

against Redd1 were analyzed by western blot for

the indicated proteins. Quantification of relative

levels of phospho-RPS6 (corrected to total RPS6

levels) is shown under respective lanes. Cells

were under steady-state culturing conditions

(see also Figure S4).

(E) Top panel: Redd1 promoter (from translation

start site at +1 to 2 kb upstream) with location of

previously described transcription factor binding

sites. The promoter is divided into proximal and

distal regions (PP and DP, respectively). Positions

of ChIP amplicons are indicated. Bottom panel:

ChIP assay. Quantitative PCR for Redd1 promoter

regions from WT SPC chromatin pulled down

with Plzf antibody. Fold enrichment is shown rela-

tive to background of chromatin pulled down from

Plzf �/� SPCs with the same antibody. Standard

deviations from duplicate PCR reactions are

indicated.

(F) Luciferase reporter assay with constructs con-

taining the PP+DP Redd1 promoter regions or PP

alone. 293HEK cells were transfected with the

appropriate luciferase reporter together with PLZF constructs or vector control as indicated. Reporter activities were normalized to that of TK-Renilla and empty

vector luciferase reporter controls (*p < 0.001). Mean values from triplicate wells and associated standard deviations are indicated.


immunoprecipitation (ChIP) assay to assess whether Plzf was

bound to the Redd1 promoter in SPCs. We scanned a 2-kb

region upstream Redd1 that contains binding sites for multiple

transcription factors known to regulate Redd1 expression

(Figure 4E) (Ellisen et al., 2002; Lin et al., 2005; Shoshani et al.,

2002). Importantly, we could detect Plzf association with distal

promoter (DP) regions but not proximal promoter (PP) regions

(Figure 4E), indicating that Plzf regulates Redd1 expression

through recruitment to the DP. To confirm this we performed

luciferase assays with REDD1 promoter constructs containing

both DP+PP elements and the PP element alone (Figure 4F).

In agreement with our ChIP results, PLZF activated the DP+PP

construct but not the PP construct. Further, deletion of the

PLZF POZ domain prevented PP+DP reporter activation, indi-

cating a requirement for this protein-protein interaction domain.

We conclude that Plzf directly activates Redd1 through the DP to

inhibit mTORC1 in SPCs.

Active mTORC1 Inhibits SPC Self-Renewal Pathwaysvia a Negative Feedback LoopAs Plzf �/� SPCs show increased mTORC1 but decreased Akt

activities (Figure 3C), we considered that negative feedback

from mTORC1 could prevent Plzf �/� cells from activating Akt

in response to growth factors. In cells lacking TSC1/TSC2,

increased mTORC1-signaling induces a negative feedback

loop from the mTORC1 downstream target S6K to upstream

signaling components causing inhibition of PI3K/Akt (Harrington

et al., 2004; Shah et al., 2004). Therefore, we tested whether

aberrant mTORC1 activity inhibited the response of Plzf �/�

SPCs to GDNF, a growth factor required for SPC self-renewal,

which could explain the maintenance defect of Plzf �/� SPCs.

We compared the response of starved Plzf +/+ and Plzf �/�

SPCs to GDNF in the absence or presence of rapamycin to

vary mTORC1 activity (Figures 5A and 5B). Consistent with

previous data, Plzf �/� SPCs showed lower levels of basal

and GDNF-stimulated Akt activity when compared to WT

SPCs, indicating a substantially reduced ability of Plzf �/� cells

to activate PI3K/Akt. However, on rapamycin treatment, the

ability of Plzf �/� SPCs to activate Akt in response to GDNF

was significantly increased and became comparable to that of

WT cells. This rescue of GDNF responsiveness by rapamycin

indicates that a negative feedback response from aberrantly

activated mTORC1 suppresses the response of Plzf �/� SPCs

to GDNF. Indeed, if GDNF levels in the media were reduced,

Figure 5. A Negative Feedback Loop from

mTORC1 to the GDNF Receptor Is Activated

in Plzf �/� SPCs

(A) Plzf +/+ and Plzf �/� SPCs starved overnight in

basal SPC medium supplemented with DMSO

(vehicle control) or rapamycin were stimulated

with 10ng/ml GDNF for 20 min before harvesting

for western blot analysis.

(B) Quantification of relative levels of phospho-Akt

and phospho-RPS6 (corrected to total levels of

respective proteins) from (A).

(C) Plzf +/+ and Plzf �/� SPCs were plated in

medium containing decreasing concentrations of

GDNF, harvested one week later and counted to

determine fold cell recovery as an indicator of

growth. The concentration of GDNF was varied

from 4 ng/ml (left bars) to 1 ng/ml (right bars). Stan-

dard deviations from duplicate wells are indicated

(**p < 0.01, *p < 0.02).

(D) Quantitative RT-PCR analysis of GDNF

receptor components in Plzf +/+ and Plzf �/�

SPCs treated with DMSO (vehicle control) or rapa-

mycin for 48 hr. mRNA levels are normalized to

those of b-actin and standard deviations from

duplicate reactions are shown.

(E) Flow cytometry analysis of Gfra1 levels from

SPC lines treated as in (D).

(F) Quantitative RT-PCR analysis of GDNF

receptor components in av-integrin negative,

Thy-1 low, c-Kit negative testis cell fractions

pooled from 2-week-old Plzf +/+ and Plzf �/�

mice. mRNA levels are normalized to those of

b-actin and standard deviations from duplicate

reactions are shown.

See also Figure S5.


Plzf �/� SPC growth was inhibited more substantially than that of

the WT cells (Figure 5C). A decreased sensitivity to GDNF due

to negative feedback from mTORC1 can explain the defect in

Plzf �/� SPC maintenance in vivo when levels of GDNF are

limiting (Meng et al., 2000) and how Plzf �/� SPCs can be

cultured in vitro when GDNF is supplied to excess. Importantly,

GDNF expression in Plzf �/� testis was equivalent to that of

the WT (Figure S5), ruling out noncell autonomous defects in

production of niche factors.

Activated mTORC1 Suppresses Expression of GDNFReceptor ComponentsWe next sought to determine the mechanism by which activated

mTORC1 inhibits response of Plzf �/� SPCs to GDNF. Negative

feedback from mTORC1 to PI3K/Akt was originally character-

ized in the context of insulin responsiveness and signaling

through IRS1/2 proteins (Harrington et al., 2005). Although IRS

proteins have been implicated in activation of PI3K by the

GDNF receptor component c-Ret (Hennige et al., 2000), we

noticed that expression of the GDNF receptor components

GFRa1 and c-Ret in SPCs were responsive to rapamycin

(Figures 5D and 5E). Cultured Plzf �/� SPCs expressed lower

levels of GFRa1 and c-Ret compared to WT cells (correlating

with reduced GDNF-responsiveness) but on mTORC1 inhibition,

expression of these GDNF receptor components was increased

back to WT levels. Expression of GFRa1 and c-Ret were also

lower in freshly isolated Plzf �/� SPCs compared to WT SPCs

(Figure 5F), confirming that Plzf loss disrupts GDNF receptor

expression in vivo. Our data suggest that aberrantly activated

mTORC1 inhibits the response of Plzf �/� SPCs to GDNF by

opposing expression of the receptor. Interestingly, mTORC1

inhibits upstream signaling events in MEFs through transcrip-

tional inhibition of the PDGF receptor (Zhang et al., 2007). In

SPCs, negative feedback between mTORC1 and the GDNF

receptor will lead to loss of self-renewal when mTORC1 is hyper-

activated, such as occurs on loss of Plzf expression (Figure S6A).

This feedback loop would inversely couple cell growth to self-

renewal, so that on expansion of the stem cell pool by mitogenic

stimulation and resultant mTORC1 activation, response of the

cells to self-renewal signals is inhibited, restricting stem cell

numbers. Interestingly, GDNF activates mTORC1 weakly in

SPCs when compared to other mitogens (e.g., bFGF) (Figure 3B),

suggesting that GDNF would trigger a limited activation of the

negative feedback response whereas bFGF could cause more

extensive inhibition of GDNF signaling via mTORC1.

Rapamycin Treatment Attenuates the SPC MaintenanceDefect of Plzf �/� Mice and Increases SPCs in WTControl MiceAs inhibition of aberrant mTORC1 activity in Plzf �/� SPCs in vitro

rescues their response to GDNF, we next asked whether

mTORC1 inhibition in vivo could rescue the Plzf �/� SPC mainte-

nance defect. We initiated rapamycin treatment of Plzf �/� and

WT control mice during early stages of postnatal development

(10 days postnatal) and continued daily treatment for 1 week,

a regimen able to normalize the increased size of Plzf �/� SPCs

(Figures 2G and 2H). Subsequent analysis of SPC compartment

status demonstrated that rapamycin increased both frequency

and numbers of avneg Thy-1low c-Kitneg cells in Plzf �/� testis,

approaching those values found in vehicle-treated WT controls

(Figures 6A and 6B). Rapamycin treatment of Plzf �/� mice

increased both the frequency of avneg Thy-1low cells and normal-

ized the increased percentage of c-Kitpos cells within this fraction

compared to controls (Figure 6A). We conclude that aberrant

mTORC1 activation can explain, at least in part, the SPC mainte-

nance defect of Plzf �/� mice.

Intriguingly, rapamycin also increased the frequency of SPCs

in WT mice (Figure 6B), suggesting that mTORC1 inhibition

stimulates SPC function in a WT setting. Indeed, in WT SPCs,

rapamycin enhanced Akt activation in response to GDNF

(Figures 5A and 5B) and increased GDNF receptor expression

(Figures 5D and 5E). Importantly, prolonged rapamycin treat-

ment of juvenile WT mice increased the number of cells with

high levels of Plzf expression in the testis (Figures 6C and 6D

and Figure S6B) and enhanced GDNF receptor expression

(Figure S6C), suggesting that physiological mTORC1 activity

controls size and function of the SPC pool in vivo.

DISCUSSION

Disruption of genes encoding for negative regulators of the PI3K/

Akt/mTORC1 pathway (e.g., Pten, Tsc1, and Pml) leads to loss of

hematopoietic stem cell (HSC) quiescence and subsequent

exhaustion. Rapamycin prevents HSC depletion, demonstrating

that aberrant mTORC1 activation is responsible for stem cell loss

(Chen et al., 2008; Gan et al., 2008; Ito et al., 2008; Yilmaz et al.,

2006). The negative effects of mTORC1 on HSC maintenance

have been attributed to its downstream targets in cell growth

pathways (Chen et al., 2008) and aberrant mTORC1 activity

elicits activation of tumor suppressive mechanisms that can

contribute to stem cell failure (Figure S6D) (Alimonti et al.,

2010; Lee et al., 2007a). By studying Plzf function in SPCs, we

now implicate negative feedback responses from the mTORC1

pathway to receptors required to transduce niche-derived self-

renewal signals in the loss of stem cell potential (Figure S6E).

HSC maintenance is also linked to niche signals acting

through growth factor receptors (Arai et al., 2004; Yoshihara

et al., 2007) thus it will be of interest to translate our findings

into the hematopoietic system. In addition, the decreased sensi-

tivity of Plzf �/� SPCs to GDNF mimics a reduced production of

niche-signals; the decline in stem cell numbers in aging mouse

testis is due partly to declining niche function and GDNF produc-

tion (Ryu et al., 2006; Zhang et al., 2006), thus loss of Plzf expres-

sion can represent a premature aging phenotype. Our studies

therefore provide further insight into the association between

stem cell maintenance, mTORC1, and aging (Harrison et al.,

2009; Rossi et al., 2008).

A correlation between aging and cancer is also apparent, likely

due to the accumulation of multiple genetic hits over time for

tumor development (Rossi et al., 2008). As stem cells are resi-

dent long-term in tissues and possess self-renewal capability

(a feature of cancer stem cells), they are suggested to be the

source of many cancers. However, loss of tumor suppressors

(e.g., Pten, Pml) trigger stem cell depletion (Ito et al., 2009) and

subsequent cancer development is associated with additional

genetic hits (Guo et al., 2008). It is proposed that oncogenic


hits in stem cells elicit activation of fail-safe mechanisms that

have to be evaded to allow cancer development (Figure S6D).

These fail-safe mechanisms prevent propagation of clones of

cells derived from stem cells with oncogenic mutations and the

consequent risk of acquiring additional mutations that lead to

cancer. The mTORC1 hyperactivation that occurs in response

to loss of tumor suppressors such as Pten and Pml is likely

a trigger of this fail-safe mechanism as rapamycin prevents

stem cell exhaustion. Based on our new model, mTORC1

hyperactivation in response to oncogenic stimulation may also

desensitize the cancer stem cell to niche-signals that are

required for self-renewal; thus additional mutations supporting

niche-independent self-renewal could allow cancer to develop

(Figure S6E).

Figure 6. mTORC1 Inhibition Attenuates

SPC Maintenance Defect of Plzf �/� Mice

(A) Top panels: representative av-integrin/Thy-1

flow profiles of testis cells from Plzf +/+ and

Plzf �/� mice treated with vehicle or rapamycin

from 10 days postnatal age for a period of

1 week and analyzed the day after completing

treatment. Bottom panels: Flow cytometry anal-

ysis of c-Kit expression within the corresponding

av-integrin negative, Thy-1 low fractions.

(B) Quantification of frequency and absolute

numbers of av integrin negative, Thy-1 low, c-Kit

negative testis cells from mice of the indicated

genotypes treated with Vehicle or rapamycin as

in (A) (n R 5 per genotype and treatment group,

**p < 0.02, *p < 0.05). Graphs indicate mean

values ± SEM.

(C) WT juvenile mice (2–3 weeks postnatal) were

treated daily for 2 weeks with rapamycin or vehicle.

Testes were harvested and subject to immunohis-

tochemistry for Plzf. Representative images are

203.

(D) Quantification of cells showing strong positivity

for Plzf in seminiferous tubules of testis from (C),

representing SPCs. Results from duplicate exper-

iments (1 and 2) are shown together with SEM

(**p < 0.01).

(E) Illustration of the effects of aberrant mTORC1

activity or inhibition of physiological mTORC1

activation with rapamycin on SPC numbers and

function.

See also Figure S6.

PLZF was first identified from involve-

ment in acute promyelocytic leukemia

(APL) (Chen et al., 1993). Thus, our model

of Plzf-dependent SPC maintenance

can have implications for leukemia patho-

genesis. The ability of Plzf to inhibit

mTORC1 through Redd1 induction may

be relevant in this context given that

perturbed PLZF function and elevated

mTORC1 activity are associated with

leukemia development (He et al., 2000;

Martelli et al., 2007; McConnell and Licht,

2007). In addition, REDD1 is involved in

leukemic and myeloid cell differentiation (Gery et al., 2007;

Nishioka et al., 2009), suggesting that this gene can be a

relevant PLZF target in hematopoietic cells. PML, also identified

from its involvement in chromosomal translocations with the

RARA gene in APL, is required for HSC maintenance and inhibits

mTORC1 (Bernardi et al., 2006; Ito et al., 2008); thus illustrating

a striking commonality of function between the RARA partners

of APL.

We have identified a role for Plzf-mediated Redd1 expression

in inhibiting mTORC1 activation in SPCs. As mTORC1 activity

has a significant impact on the response of SPCs to GDNF, the

tight regulation of this pathway is critical. Redd1 is also induced

in response to cellular stress and in particular, by hypoxia

(Brugarolas et al., 2004; Shoshani et al., 2002). Interestingly,


the HSC niche has been suggested to exist in a hypoxic state

(Kubota et al., 2008; Parmar et al., 2007) that, if the SPC niche

had similar characteristics, could be responsible for Redd1

induction. However, SPCs reside close to the vasculature

in vivo (Yoshida et al., 2007) and Redd1 expression is maintained

when SPCs are cultured at atmospheric oxygen levels. Thus

Redd1 expression in this case depends on a Plzf tonic transcrip-

tional activity and its induction is seemingly independent of

hypoxia.

As mTORC1 inhibition enhances SPC function even in a WT

setting, our data also have important therapeutic implications.

It is tempting to speculate that rapamycin could be used to

enhance SPC activity in humans; potentially allowing improved

SPC culture from testicular biopsies and generation of therapeu-

tically relevant MASCs (Figure 6E). Furthermore, our findings

provide a novel and straightforward explanation for the require-

ment of Plzf in germline maintenance.


Mouse Maintenance and Manipulation

Plzf �/�mice are previously described (Costoya et al., 2004). Mice were treated

with rapamycin (LC Laboratories) at a daily dose of 4 mg/kg as described

(Yilmaz et al., 2006). Transgenic mice expressing EGFP from the b-actin

promoter were from Jackson Laboratories. Testis transplant assays were

performed as described (Seandel et al., 2007) with 50 3 103 unfractionated

or 5 3 103 avneg Thy-1low c-Kitneg testis cells injected per recipient testis.

Teratoma formation assays in NOD/SCID mice (Jackson Laboratories) were

performed as described (Seandel et al., 2007).

Antibody Generation

Monoclonal anti-PLZF antibody (clone 9E12) was raised in Armenian hamster

against a pool of keyhole limpet hemocyanin (KLH)-conjugated peptides at the

Memorial Sloan-Kettering Cancer Center (MSKCC) antibody facility and reacts

to a peptide within the hinge domain of mouse Plzf (RSKEGPGTPTRRSVIT

SARE). Antibody was directly conjugated to Alexa 488 or Alexa 647 for use

in flow cytometry (MSKCC antibody facility).

Flow Cytometry

Single cell suspensions were prepared from testis as described (Ogawa et al.,

1997) and resuspended in phosphate-buffered saline (PBS) containing 2%

fetal bovine serum (FBS) for subsequent staining and analysis (Filipponi

et al., 2007). For a detailed description of flow cytometry methods and anti-

bodies used, refer to the Extended Experimental Procedures.

Cell Culture and Treatment

SPC fractions from testis were plated directly onto mitotically inactivated MEF

feeder cells. Medium for SPC culture was as previously described (Seandel

et al., 2007) supplemented with 10 ng/ml IGF-I (Peprotech). MASCs and

mouse ESCs (v6.5) were cultured in embryonic stem cell medium on MEF

feeders (Seandel et al., 2007). Knockdown of Redd1 expression in SPCs by

shRNA was performed using pLKO.1 lentiviral vectors (Open Biosystems).

A detailed description of methods is included in the Extended Experimental

Procedures.

Immunofluorescence, Immunohistochemistry, and Western Blot

Tissue and cultured cells were fixed in 4% paraformaldehyde before

processing for immunofluorescence (IF) and immunohistochemistry (IHC) as

described (Costoya et al., 2004; Filipponi et al., 2007). For western blot anal-

ysis, cells were lysed in RIPA buffer and processed as described (Costoya

et al., 2008). Detailed methods are included in the Extended Experimental

Procedures.

Chromatin Immunoprecipitation (ChIP)

SPCs were trypsinized and resuspended in SPC medium then incubated in

tissue culture dishes for 30 min to remove contaminating (adherent) MEFs.

Preparation and immunoprecipitation of chromatin was performed using

a SimpleChIP Enzymatic Chromatin IP kit (Cell Signaling Technology)

according to manufacturers instructions and monoclonal anti-PLZF antibody

(Calbiochem). DNA samples were analyzed by quantitative PCR using a

QuantiTect SYBR Green PCR kit (QIAGEN) and Light Cycler (Roche). Primer

sequences for the Redd1 promoter are included in the Extended Experimental

Procedures.

Luciferase Assay

REDD1 promoter constructs are described (Lin et al., 2005) and were cloned

from U2OS genomic DNA into the pGL3-enhancer vector (Promega). Lucif-

erase assays were performed with 293HEK cells as described (Barna et al.,

2002). Please refer to Extended Experimental Procedures for details.

Quantitative Reverse Transcription Polymerase Chain Reaction

(qRT-PCR)

RNA was harvested using Trizol reagent (Invitrogen) then used for first strand

cDNA synthesis and subsequent quantitative PCR analysis as described

(Filipponi et al., 2007). For primer sequences and detailed methodology please

see Extended Experimental Procedures.

Statistical Analysis

Results from cell growth assays, testis transplants, and flow cytometry anal-

ysis were assessed for statistical significance by a standard two-tailed t test.



six figures and can be found with this article online at doi:10.1016/j.cell.

2010.06.041.

ACKNOWLEDGMENTS

We thank current and past members of the Pandolfi lab and in particular Taka-

hiro Maeda and Rosa Bernardi for helpful discussion and advice. We would

also like to thank Antonella Papa for generating luciferase reporters, Sharmila

Fagoonee for experimental help, Leif Ellisen for Redd1 promoter constructs,

and the flow cytometry facilities of BIDMC and MSKCC for expert support.

M.S. is a Stanley and Fiona Druckenmiller Fellow of the New York Stem Cell

Foundation. This work was supported by NIH grants to P.P.P.





REFERENCES

Alimonti, A., Nardella, C., Chen, Z., Clohessy, J.G., Carracedo, A., Trotman,

L.C., Cheng, K., Varmeh, S., Kozma, S.C., Thomas, G., et al. (2010). A novel

type of cellular senescence that can be enhanced in mouse models and human

tumor xenografts to suppress prostate tumorigenesis. J. Clin. Invest. 120,

681–693.

Arai, F., Hirao, A., Ohmura, M., Sato, H., Matsuoka, S., Takubo, K., Ito, K., Koh,

G.Y., and Suda, T. (2004). Tie2/angiopoietin-1 signaling regulates hematopoi-

etic stem cell quiescence in the bone marrow niche. Cell 118, 149–161.

Barna, M., Merghoub, T., Costoya, J.A., Ruggero, D., Branford, M., Bergia, A.,

Samori, B., and Pandolfi, P.P. (2002). Plzf mediates transcriptional repression

of HoxD gene expression through chromatin remodeling. Dev. Cell 3, 499–510.

Bernardi, R., Guernah, I., Jin, D., Grisendi, S., Alimonti, A., Teruya-Feldstein, J.,

Cordon-Cardo, C., Simon, M.C., Rafii, S., and Pandolfi, P.P. (2006). PML

inhibits HIF-1alpha translation and neoangiogenesis through repression of

mTOR. Nature 442, 779–785.


Brown, E.J., Albers, M.W., Shin, T.B., Ichikawa, K., Keith, C.T., Lane, W.S., and

Schreiber, S.L. (1994). A mammalian protein targeted by G1-arresting rapamy-

cin-receptor complex. Nature 369, 756–758.

Brugarolas, J., Lei, K., Hurley, R.L., Manning, B.D., Reiling, J.H., Hafen, E.,

Witters, L.A., Ellisen, L.W., and Kaelin, W.G., Jr. (2004). Regulation of mTOR

function in response to hypoxia by REDD1 and the TSC1/TSC2 tumor

suppressor complex. Genes Dev. 18, 2893–2904.

Buaas, F.W., Kirsh, A.L., Sharma, M., McLean, D.J., Morris, J.L., Griswold,

M.D., de Rooij, D.G., and Braun, R.E. (2004). Plzf is required in adult male

germ cells for stem cell self-renewal. Nat. Genet. 36, 647–652.

Castilho, R.M., Squarize, C.H., Chodosh, L.A., Williams, B.O., and Gutkind,

J.S. (2009). mTOR mediates Wnt-induced epidermal stem cell exhaustion

and aging. Cell Stem Cell 5, 279–289.

Chen, C., Liu, Y., Liu, R., Ikenoue, T., Guan, K.L., and Zheng, P. (2008). TSC-

mTOR maintains quiescence and function of hematopoietic stem cells by

repressing mitochondrial biogenesis and reactive oxygen species. J. Exp.

Med. 205, 2397–2408.

Chen, Z., Brand, N.J., Chen, A., Chen, S.J., Tong, J.H., Wang, Z.Y., Waxman,

S., and Zelent, A. (1993). Fusion between a novel Kruppel-like zinc finger gene

and the retinoic acid receptor-alpha locus due to a variant t(11;17) transloca-

tion associated with acute promyelocytic leukaemia. EMBO J. 12, 1161–1167.

Corradetti, M.N., Inoki, K., and Guan, K.L. (2005). The stress-inducted proteins

RTP801 and RTP801L are negative regulators of the mammalian target of

rapamycin pathway. J. Biol. Chem. 280, 9769–9772.

Costoya, J.A., Hobbs, R.M., Barna, M., Cattoretti, G., Manova, K., Sukhwani,

M., Orwig, K.E., Wolgemuth, D.J., and Pandolfi, P.P. (2004). Essential role of

Plzf in maintenance of spermatogonial stem cells. Nat. Genet. 36, 653–659.

Costoya, J.A., Hobbs, R.M., and Pandolfi, P.P. (2008). Cyclin-dependent

kinase antagonizes promyelocytic leukemia zinc-finger through phosphoryla-

tion. Oncogene 27, 3789–3796.

David, G., Alland, L., Hong, S.H., Wong, C.W., DePinho, R.A., and Dejean, A.

(1998). Histone deacetylase associated with mSin3A mediates repression by

the acute promyelocytic leukemia-associated PLZF protein. Oncogene 16,

2549–2556.

de Rooij, D.G., and Russell, L.D. (2000). All you wanted to know about

spermatogonia but were afraid to ask. J. Androl. 21, 776–798.

Doulatov, S., Notta, F., Rice, K.L., Howell, L., Zelent, A., Licht, J.D., and Dick,

J.E. (2009). PLZF is a regulator of homeostatic and cytokine-induced myeloid

development. Genes Dev. 23, 2076–2087.

Ellisen, L.W., Ramsayer, K.D., Johannessen, C.M., Yang, A., Beppu, H.,

Minda, K., Oliner, J.D., McKeon, F., and Haber, D.A. (2002). REDD1, a develop-

mentally regulated transcriptional target of p63 and p53, links p63 to regulation

of reactive oxygen species. Mol. Cell 10, 995–1005.

Encinas, M., Tansey, M.G., Tsui-Pierchala, B.A., Comella, J.X., Milbrandt, J.,

and Johnson, E.M., Jr. (2001). c-Src is required for glial cell line-derived neuro-

trophic factor (GDNF) family ligand-mediated neuronal survival via a phospha-

tidylinositol-3 kinase (PI-3K)-dependent pathway. J. Neurosci. 21, 1464–1472.

Fan, Q.W., Cheng, C., Knight, Z.A., Haas-Kogan, D., Stokoe, D., James, C.D.,

McCormick, F., Shokat, K.M., and Weiss, W.A. (2009). EGFR signals to mTOR

through PKC and independently of Akt in glioma. Sci. Signal. 2, ra4.

Filipponi, D., Hobbs, R.M., Ottolenghi, S., Rossi, P., Jannini, E.A., Pandolfi,

P.P., and Dolci, S. (2007). Repression of kit expression by Plzf in germ cells.

Mol. Cell. Biol. 27, 6770–6781.

Fingar, D.C., Salama, S., Tsou, C., Harlow, E., and Blenis, J. (2002). Mamma-

lian cell size is controlled by mTOR and its downstream targets S6K1 and

4EBP1/eIF4E. Genes Dev. 16, 1472–1487.

Fischer, S., Kohlhase, J., Bohm, D., Schweiger, B., Hoffmann, D., Heitmann,

M., Horsthemke, B., and Wieczorek, D. (2008). Biallelic loss of function of

the promyelocytic leukemia zinc finger (PLZF) gene causes severe skeletal

defects and genital hypoplasia. J. Med. Genet. 45, 731–737.

Gan, B., and DePinho, R.A. (2009). mTORC1 signaling governs hematopoietic

stem cell quiescence. Cell Cycle 8, 1003–1006.

Gan, B., Sahin, E., Jiang, S., Sanchez-Aguilera, A., Scott, K.L., Chin, L.,

Williams, D.A., Kwiatkowski, D.J., and DePinho, R.A. (2008). mTORC1-depen-

dent and -independent regulation of stem cell renewal, differentiation, and

mobilization. Proc. Natl. Acad. Sci. USA 105, 19384–19389.

Gery, S., Park, D.J., Vuong, P.T., Virk, R.K., Muller, C.I., Hofmann, W.K., and

Koeffler, H.P. (2007). RTP801 is a novel retinoic acid-responsive gene associ-

ated with myeloid differentiation. Exp. Hematol. 35, 572–578.

Guo, W., Lasky, J.L., Chang, C.J., Mosessian, S., Lewis, X., Xiao, Y., Yeh, J.E.,

Chen, J.Y., Iruela-Arispe, M.L., Varella-Garcia, M., et al. (2008). Multi-genetic

events collaboratively contribute to Pten-null leukemia stem-cell formation.

Nature 453, 529–533.

Hamra, F.K., Chapman, K.M., Nguyen, D., and Garbers, D.L. (2007). Identifica-

tion of neuregulin as a factor required for formation of aligned spermatogonia.

J. Biol. Chem. 282, 721–730.

Harrington, L.S., Findlay, G.M., Gray, A., Tolkacheva, T., Wigfield, S., Rebholz,

H., Barnett, J., Leslie, N.R., Cheng, S., Shepherd, P.R., et al. (2004). The

TSC1-2 tumor suppressor controls insulin-PI3K signaling via regulation of

IRS proteins. J. Cell Biol. 166, 213–223.

Harrington, L.S., Findlay, G.M., and Lamb, R.F. (2005). Restraining PI3K:

mTOR signalling goes back to the membrane. Trends Biochem. Sci. 30, 35–42.

Harrison, D.E., Strong, R., Sharp, Z.D., Nelson, J.F., Astle, C.M., Flurkey, K.,

Nadon, N.L., Wilkinson, J.E., Frenkel, K., Carter, C.S., et al. (2009). Rapamycin

fed late in life extends lifespan in genetically heterogeneous mice. Nature 460,

392–395.

He, L.Z., Bhaumik, M., Tribioli, C., Rego, E.M., Ivins, S., Zelent, A., and

Pandolfi, P.P. (2000). Two critical hits for promyelocytic leukemia. Mol. Cell

6, 1131–1141.

Hennige, A.M., Lammers, R., Arlt, D., Hoppner, W., Strack, V., Niederfellner,

G., Seif, F.J., Haring, H.U., and Kellerer, M. (2000). Ret oncogene signal trans-

duction via a IRS-2/PI 3-kinase/PKB and a SHC/Grb-2 dependent pathway:

possible implication for transforming activity in NIH3T3 cells. Mol. Cell. Endo-

crinol. 167, 69–76.

Huang, J., and Manning, B.D. (2008). The TSC1-TSC2 complex: a molecular

switchboard controlling cell growth. Biochem. J. 412, 179–190.

Ito, K., Bernardi, R., Morotti, A., Matsuoka, S., Saglio, G., Ikeda, Y., Rosenblatt,

J., Avigan, D.E., Teruya-Feldstein, J., and Pandolfi, P.P. (2008). PML targeting

eradicates quiescent leukaemia-initiating cells. Nature 453, 1072–1078.

Ito, K., Bernardi, R., and Pandolfi, P.P. (2009). A novel signaling network as

a critical rheostat for the biology and maintenance of the normal stem cell

and the cancer-initiating cell. Curr. Opin. Genet. Dev. 19, 51–59.

Kanatsu-Shinohara, M., Inoue, K., Lee, J., Yoshimoto, M., Ogonuki, N., Miki,

H., Baba, S., Kato, T., Kazuki, Y., Toyokuni, S., et al. (2004a). Generation of

pluripotent stem cells from neonatal mouse testis. Cell 119, 1001–1012.

Kanatsu-Shinohara, M., Ogonuki, N., Inoue, K., Miki, H., Ogura, A., Toyokuni,

S., and Shinohara, T. (2003). Long-term proliferation in culture and germline

transmission of mouse male germline stem cells. Biol. Reprod. 69, 612–616.

Kanatsu-Shinohara, M., Toyokuni, S., and Shinohara, T. (2004b). CD9 is

a surface marker on mouse and rat male germline stem cells. Biol. Reprod.

70, 70–75.

Kubota, H., Avarbock, M.R., and Brinster, R.L. (2003). Spermatogonial stem

cells share some, but not all, phenotypic and functional characteristics with

other stem cells. Proc. Natl. Acad. Sci. USA 100, 6487–6492.

Kubota, H., Avarbock, M.R., and Brinster, R.L. (2004). Growth factors essential

for self-renewal and expansion of mouse spermatogonial stem cells. Proc.

Natl. Acad. Sci. USA 101, 16489–16494.

Kubota, Y., Takubo, K., and Suda, T. (2008). Bone marrow long label-retaining

cells reside in the sinusoidal hypoxic niche. Biochem. Biophys. Res. Commun.

366, 335–339.

Labbaye, C., Quaranta, M.T., Pagliuca, A., Militi, S., Licht, J.D., Testa, U., and

Peschle, C. (2002). PLZF induces megakaryocytic development, activates

Tpo receptor expression and interacts with GATA1 protein. Oncogene 21,

6669–6679.


Lee, C.H., Inoki, K., Karbowniczek, M., Petroulakis, E., Sonenberg, N., Henske,

E.P., and Guan, K.L. (2007a). Constitutive mTOR activation in TSC mutants

sensitizes cells to energy starvation and genomic damage via p53. EMBO J.

26, 4812–4823.

Lee, J., Kanatsu-Shinohara, M., Inoue, K., Ogonuki, N., Miki, H., Toyokuni, S.,

Kimura, T., Nakano, T., Ogura, A., and Shinohara, T. (2007b). Akt mediates

self-renewal division of mouse spermatogonial stem cells. Development

134, 1853–1859.

Lin, L., Stringfield, T.M., Shi, X., and Chen, Y. (2005). Arsenite induces a cell

stress-response gene, RTP801, through reactive oxygen species and tran-

scription factors Elk-1 and CCAAT/enhancer-binding protein. Biochem. J.

392, 93–102.

Ma, X.M., and Blenis, J. (2009). Molecular mechanisms of mTOR-mediated

translational control. Nat. Rev. Mol. Cell Biol. 10, 307–318.

Martelli, A.M., Tazzari, P.L., Evangelisti, C., Chiarini, F., Blalock, W.L., Billi,

A.M., Manzoli, L., McCubrey, J.A., and Cocco, L. (2007). Targeting the phos-

phatidylinositol 3-kinase/Akt/mammalian target of rapamycin module for

acute myelogenous leukemia therapy: from bench to bedside. Curr. Med.

Chem. 14, 2009–2023.

McConnell, M.J., and Licht, J.D. (2007). The PLZF gene of t (11;17)-associated

APL. Curr. Top. Microbiol. Immunol. 313, 31–48.

Meng, X., Lindahl, M., Hyvonen, M.E., Parvinen, M., de Rooij, D.G., Hess,

M.W., Raatikainen-Ahokas, A., Sainio, K., Rauvala, H., Lakso, M., et al.

(2000). Regulation of cell fate decision of undifferentiated spermatogonia by

GDNF. Science 287, 1489–1493.

Nagano, M., Avarbock, M.R., and Brinster, R.L. (1999). Pattern and kinetics of

mouse donor spermatogonial stem cell colonization in recipient testes. Biol.

Reprod. 60, 1429–1436.

Nishioka, C., Ikezoe, T., Yang, J., Gery, S., Koeffler, H.P., and Yokoyama, A.

(2009). Inhibition of mammalian target of rapamycin signaling potentiates the

effects of all-trans retinoic acid to induce growth arrest and differentiation of

human acute myelogenous leukemia cells. Int. J. Cancer 125, 1710–1720.

Oatley, J.M., Avarbock, M.R., and Brinster, R.L. (2007). Glial cell line-derived

neurotrophic factor regulation of genes essential for self-renewal of mouse

spermatogonial stem cells is dependent on Src family kinase signaling. J.

Biol. Chem. 282, 25842–25851.

Oatley, J.M., Avarbock, M.R., Telaranta, A.I., Fearon, D.T., and Brinster, R.L.

(2006). Identifying genes important for spermatogonial stem cell self-renewal

and survival. Proc. Natl. Acad. Sci. USA 103, 9524–9529.

Ogawa, T., Arechaga, J.M., Avarbock, M.R., and Brinster, R.L. (1997). Trans-

plantation of testis germinal cells into mouse seminiferous tubules. Int. J. Dev.

Biol. 41, 111–122.

Ohbo, K., Yoshida, S., Ohmura, M., Ohneda, O., Ogawa, T., Tsuchiya, H.,

Kuwana, T., Kehler, J., Abe, K., Scholer, H.R., et al. (2003). Identification and

characterization of stem cells in prepubertal spermatogenesis in mice small

star, filled. Dev. Biol. 258, 209–225.

Parmar, K., Mauch, P., Vergilio, J.A., Sackstein, R., and Down, J.D. (2007).

Distribution of hematopoietic stem cells in the bone marrow according to

regional hypoxia. Proc. Natl. Acad. Sci. USA 104, 5431–5436.

Rossi, D.J., Jamieson, C.H., and Weissman, I.L. (2008). Stems cells and the

pathways to aging and cancer. Cell 132, 681–696.

Ruvinsky, I., Sharon, N., Lerer, T., Cohen, H., Stolovich-Rain, M., Nir, T., Dor,

Y., Zisman, P., and Meyuhas, O. (2005). Ribosomal protein S6 phosphorylation

is a determinant of cell size and glucose homeostasis. Genes Dev. 19,

2199–2211.

Ryu, B.Y., Orwig, K.E., Oatley, J.M., Avarbock, M.R., and Brinster, R.L. (2006).

Effects of aging and niche microenvironment on spermatogonial stem cell

self-renewal. Stem Cells 24, 1505–1511.

Schieke, S.M., and Finkel, T. (2006). Mitochondrial signaling, TOR, and life

span. Biol. Chem. 387, 1357–1361.

Schrans-Stassen, B.H., van de Kant, H.J., de Rooij, D.G., and van Pelt, A.M.

(1999). Differential expression of c-kit in mouse undifferentiated and differen-

tiating type A spermatogonia. Endocrinology 140, 5894–5900.

Seandel, M., James, D., Shmelkov, S.V., Falciatori, I., Kim, J., Chavala, S.,

Scherr, D.S., Zhang, F., Torres, R., Gale, N.W., et al. (2007). Generation of

functional multipotent adult stem cells from GPR125+ germline progenitors.

Nature 449, 346–350.

Shah, O.J., Wang, Z., and Hunter, T. (2004). Inappropriate activation of the

TSC/Rheb/mTOR/S6K cassette induces IRS1/2 depletion, insulin resistance,

and cell survival deficiencies. Curr. Biol. 14, 1650–1656.

Shaw, R.J., and Cantley, L.C. (2006). Ras, PI(3)K and mTOR signalling controls

tumour cell growth. Nature 441, 424–430.

Shinohara, T., Avarbock, M.R., and Brinster, R.L. (1999). beta1- and alpha6-

integrin are surface markers on mouse spermatogonial stem cells. Proc.

Natl. Acad. Sci. USA 96, 5504–5509.

Shinohara, T., Orwig, K.E., Avarbock, M.R., and Brinster, R.L. (2000).

Spermatogonial stem cell enrichment by multiparameter selection of mouse

testis cells. Proc. Natl. Acad. Sci. USA 97, 8346–8351.

Shoshani, T., Faerman, A., Mett, I., Zelin, E., Tenne, T., Gorodin, S., Moshel, Y.,

Elbaz, S., Budanov, A., Chajut, A., et al. (2002). Identification of a novel

hypoxia-inducible factor 1-responsive gene, RTP801, involved in apoptosis.

Mol. Cell. Biol. 22, 2283–2293.

Takubo, K., Ohmura, M., Azuma, M., Nagamatsu, G., Yamada, W., Arai, F.,

Hirao, A., and Suda, T. (2008). Stem cell defects in ATM-deficient undifferen-

tiated spermatogonia through DNA damage-induced cell-cycle arrest. Cell

Stem Cell 2, 170–182.

Virtanen, I., Kallajoki, M., Narvanen, O., Paranko, J., Thornell, L.E., Miettinen,

M., and Lehto, V.P. (1986). Peritubular myoid cells of human and rat testis

are smooth muscle cells that contain desmin-type intermediate filaments.

Anat. Rec. 215, 10–20.

Wu, X., Oatley, J.M., Oatley, M.J., Kaucher, A.V., Avarbock, M.R., and Brinster,

R.L. (2010). The POU domain transcription factor POU3F1 is an important

intrinsic regulator of GDNF-induced survival and self-renewal of mouse

spermatogonial stem cells. Biol. Reprod. 82, 1103–1111.

Yilmaz, O.H., Valdez, R., Theisen, B.K., Guo, W., Ferguson, D.O., Wu, H., and

Morrison, S.J. (2006). Pten dependence distinguishes hematopoietic stem

cells from leukemia-initiating cells. Nature 441, 475–482.

Yoshida, S., Sukeno, M., and Nabeshima, Y. (2007). A vasculature-associated

niche for undifferentiated spermatogonia in the mouse testis. Science 317,

1722–1726.

Yoshida, S., Sukeno, M., Nakagawa, T., Ohbo, K., Nagamatsu, G., Suda, T.,

and Nabeshima, Y. (2006). The first round of mouse spermatogenesis is

a distinctive program that lacks the self-renewing spermatogonia stage.

Development 133, 1495–1505.

Yoshihara, H., Arai, F., Hosokawa, K., Hagiwara, T., Takubo, K., Nakamura, Y.,

Gomei, Y., Iwasaki, H., Matsuoka, S., Miyamoto, K., et al. (2007). Thrombo-

poietin/MPL signaling regulates hematopoietic stem cell quiescence and

interaction with the osteoblastic niche. Cell Stem Cell 1, 685–697.

Zhang, H., Bajraszewski, N., Wu, E., Wang, H., Moseman, A.P., Dabora, S.L.,

Griffin, J.D., and Kwiatkowski, D.J. (2007). PDGFRs are critical for PI3K/Akt

activation and negatively regulated by mTOR. J. Clin. Invest. 117, 730–738.

Zhang, X., Ebata, K.T., Robaire, B., and Nagano, M.C. (2006). Aging of male

germ line stem cells in mice. Biol. Reprod. 74, 119–124.


Myc-Nick: A Cytoplasmic CleavageProduct of Myc that Promotes a-TubulinAcetylation and Cell DifferentiationMaralice Conacci-Sorrell,1 Celine Ngouenet,1 and Robert N. Eisenman1,*1Division of Basic Sciences, Fred Hutchinson Cancer Research Center, Seattle, WA 98109, USA


DOI 10.1016/j.cell.2010.06.037

SUMMARY

The Myc oncoprotein family comprises transcriptionfactors that control multiple cellular functions andare widely involved in oncogenesis. Here we reportthe identification of Myc-nick, a cytoplasmic form ofMyc generated by calpain-dependent proteolysis atlysine 298 of full-length Myc. Myc-nick retainsconserved Myc box regions but lacks nuclear locali-zation signals and the bHLHZ domain essential forheterodimerization with Max and DNA binding. Myc-nick induces a-tubulin acetylation and altered cellmorphology by recruiting histone acetyltransferaseGCN5 to microtubules. During muscle differentiation,while the levels of full-length Myc diminish, Myc-nick and acetylated a-tubulin levels are increased.Ectopic expression of Myc-nick accelerates myo-blast fusion, triggers the expression of myogenicmarkers, and permits Myc-deficient fibroblasts totransdifferentiate in response to MyoD. We proposethat the cleavage of Myc by calpain abrogates thetranscriptional inhibition of differentiation by full-length Myc and generates Myc-nick, a driver of cyto-plasmic reorganization and differentiation.

INTRODUCTION

The Myc family (c-Myc, N-Myc, and L-Myc) of basic-helix-loop-

helix-zipper (bHLHZ) transcription factors controls the expres-

sion of a large number of target genes and noncoding RNA

loci. These Myc targets mediate the physiological effects of

Myc on cell proliferation, metabolism, apoptosis, growth, and

differentiation (Eilers and Eisenman, 2008). To promote tran-

scriptional activation at target genes, Myc forms heterodimers

with its partner Max and recruits chromatin-modifying com-

plexes to E-box-containing promoters. Myc is also involved in

transcriptional repression through the inhibition of the transcrip-

tional activator Miz1 (Kleine-Kohlbrecher et al., 2006). Aberrant

elevation of Myc levels has been shown to contribute to the

genesis of many types of human tumors (Hanahan and Wein-

berg, 2000).

Myc family proteins contain highly conserved regions termed

Myc boxes (MB) that are essential for Myc’s biological activities

(see Figure 1E). A major determinant of Myc transcriptional

function is MBII, which is the site of recruitment of coactivator

complexes containing histone acetyltransferases (HATs) such

as GCN5 (McMahon et al., 2000) and TIP60 (Frank et al.,

2003). MBI functions as a phosphorylation-dependent binding

site for the ubiquitin ligase Fbw7 (Welcker et al., 2004), whereas

MBII is one of the binding sites for the ligase SKP2 (Kim et al.,

2003; von der Lehr et al., 2003). Fbw7 and Skp2 both contribute

to the rapid degradation of Myc protein (t1/2 �20 min). The C

terminus of Myc harbors nuclear localization signals and the

bHLHZ motif that mediates dimerization with Max and DNA

binding.

Several variant forms of Myc protein have been previously

identified. All of them are nuclear-localized, low-abundance

proteins generated by alternative translation initiation. A weak

CUG translational initiation site, upstream and in-frame of the

predominant AUG codon, produces an N-terminally extended

form of c-Myc called c-Myc1 (Hann et al., 1988). Another Myc

protein variant is MycS, generated by internal translational initia-

tions at two AUG codons located �100 amino acids from the

normal N terminus (Spotts et al., 1997). MycS lacks MBI but

contains MBII and retains much of full-length Myc’s biological

activity (Xiao et al., 1998).

As expected, given their broad role as transcriptional regula-

tors, Myc family proteins are predominantly localized to the

cell nucleus during proliferation. Surprisingly, however, there

have been multiple reports of cytoplasmically localized Myc,

mostly in differentiated cells. For example, N-Myc localiza-

tion was shown to change from nuclear to cytoplasmic in

differentiating neurons of the neural crest, retinal ganglion

cells, neurons of spinal ganglia (Wakamatsu et al., 1993, 1997),

and Purkinje cells (Okano et al., 1999; Wakamatsu et al.,

1993). Cytoplasmic Myc was also reported in tumors with

diverse origins (Bai et al., 1994; Calcagno et al., 2009; Pietilai-

nen et al., 1995). These studies relied on immunostaining pro-

tocols and the form of the Myc protein involved was not char-

acterized.

Interestingly, association of Myc with several cytoplasmic

proteins has been reported. The best characterized is the

interaction of c-Myc with tubulins (Alexandrova et al., 1995)

(Koch et al., 2007; Niklinski et al., 2000). Myc has also been

reported to interact with other proteins that are predominantly


cytoplasmic, such as cdr2 (Okano et al., 1999) and AMY-1 (Taira

et al., 1998). However the nature of the cytoplasmic Myc protein

and its potential function remains an enigma. Here we report the

identification of Myc-nick, a cytoplasmically localized cleavage

product of Myc, and provide evidence for its role in cytoskeletal

organization and cell differentiation.

RESULTS

Myc-Nick Is a Truncated Form of Myc LocalizedPredominantly in the CytoplasmWhile studying regulation of c-Myc degradation, we noticed an

inverse correlation between the levels of full-length c-Myc and

Tubulin

+ - + c-MycDense CulturesSparse

Total lysates

CIB: anti Myc 143

62

49

38

Sin3A

Tubulin

Myc

c-Myc

Cell density

X 105 Cells50 5 .5 50 5 .5

6249

38

281714

IB: anti Myc 9E10

Max

20µM

Myc

Myc-nick

B

Sin3A

Tubulin

X 105 Cells50 5 .5 50 50 5 .5

Nuclear Cytoplasmic

Myc

Myc-nick

c-Myc c-Myc

Cell density

62

49

38

Vector

IB: anti Myc N262

IgG HA HAIgGN262

N262

IB: anti Myc N262

Nuclear Cytoplasmic

Nuclear Cytoplasmic

Myc

Myc-nick

143......................................... +274................................ +

N262 ....................................... +

9E10 -C19 -M

ycantibodies

Myc-nickE

MycS

IP:

Sparse Dense

Ab: N262 9E10

A

D

F

c-M

yc

DA

PI

Figure 1. Identification of Myc-Nick in the Cytoplasm of Cells Grown at High Density

(A) Total cell lysates of Rat1 myc null fibroblasts infected with c-Myc or empty retroviral vectors were prepared for western blot by adding boiling sample buffer.

(B and C) Nuclear and cytoplasmic fractions of HFF cells expressing c-Myc were prepared 48 hr after plating at the indicated increasing densities.

(D) Immunoprecipitation of HA-c-Myc (N-terminal tag) with anti-N262, anti-HA, and normal IgG from nuclear and cytoplasmic fractions of HFFs. Note that Max is

only coimmunoprecipitated along with nuclear c-Myc.

(E) Schematic representation of antibody mapping.

(F) HFF cells infected with c-myc-expressing retrovirus were cultured for 4 days after reaching confluency (middle and right panels) and compared with a

subconfluent culture (left panel) by immunofluorescence using N262 and 9E10 antibodies.

See also Figure S1.


a cytoplasmic 42 kDa protein in anti-Myc immunoblots derived

from confluent fibroblast cultures (Figures 1A and 1B). As

described below, this protein, which we have named Myc-nick,

is a cytoplasmic cleavage product of full-length c-Myc gener-

ated at high cell density (Figure 1B). Myc-nick is recognized by

three antibodies against the N-terminal two-thirds of c-Myc

(anti-Myc N262, 274, 143; Figures 1A and 1B and Figure S1A

available online) but not by anti- C-terminal antibodies (anti-

Myc 9E10, C19; Figure 1C). Furthermore, an anti-HA antibody

immunoprecipitates Myc-nick from cytoplasmic extracts of cells

expressing N-terminally HA-tagged c-Myc (Figure 1D). In addi-

tion, cytoplasmic Myc bearing N-terminal but not C-terminal

epitopes is detected by imunofluorescence in confluent cultures

(see below, Figure 1F). Together, these results indicate that

Myc-nick is a truncated protein lacking the C-terminal portion

of c-Myc while preserving an intact N terminus comprising

Myc boxes I–III (Figure 1E). This makes Myc-nick distinct from

any other previously identified form of Myc (see Introduction).

We have detected endogenously expressed Myc-nick in the

cytoplasm of a large number of cell lines including human

foreskin fibroblasts (HFFs), Wi38, L cells, HCT116, SW480, HeLa

(Figure S1B), C2C12 (Figure 7D), 293T, A431, Rat1, U2OS, ES

cells, and mouse neurospheres (not shown), in addition to mouse

tissues such as muscle, brain, and cerebellum (Figure 7A and

Figure S7A). We also observed Myc cytoplasmic localization

by immunofluorescence of confluent cultures of HFFs (Figure 1F

and Figure S5A) and Rat1 myc null cells (not shown) expressing

c-myc. Therefore a protein with the size and properties of Myc-

nick is very widely expressed. In some settings we observe

relatively low and variable amounts of Myc-nick in the nucleus

(e.g., Figure 2A).

Myc-Nick Is Generated by Proteolytic Cleavageof c-Myc in the CytoplasmWe considered the possibility that Myc-nick is generated in the

cytoplasm because the nuclear export inhibitor Leptomycin B

had no effect on the production or cytoplasmic localization of

Myc-nick (Figure 2A). To determine whether a cytoplasmic

activity could convert full-length c-Myc into Myc-nick, we incu-

bated in vitro-translated [35S]-methionine-labeled c-Myc or puri-

fied full-length recombinant c-Myc with nuclear or cytoplasmic

extracts from Rat1 myc null cells. Only cytoplasmic extracts

(CE) were capable of producing a protein (Figures 2B–2D) having

the same apparent molecular weight as Myc-nick that was

recognized by antibodies against the N terminus but not C

terminus of c-Myc (Figure S2A). Increasing the incubation time

with CE augmented Myc-nick production and decreased the

amounts of full-length c-Myc input (Figure 2C). In agreement

with our observations made in vivo, CE of dense cultures were

more efficient in cleaving c-Myc than CE of sparse cultures

(Figure S2B). In experiments designed to characterize the cyto-

plasmic activity responsible for formation of Myc-nick, we found

that inhibitors of transcription or translation had no effect on

Myc-nick formation (Figure 2E) whereas heating or adding

protease inhibitors to the CE blocked Myc-nick, consistent

with proteolysis (Figure 2F). Whereas MG132, an inhibitor of the

proteasome (and other cysteine proteases), blocked Myc nick

formation, specific proteasome inhibitors such as Epoxomycin

and Lactacystein failed to block the formation of Myc-nick

in vitro and in vivo (Figures 2G and 2H). In addition, inhibition

of either the Fbw7 or Skp2 degradation pathways, known to be

responsible for proteasomal Myc turnover, had no effect on

Myc-nick formation (data not shown). Moreover, incubation of

Myc with 20S and 26S proteasomes failed to produce Myc-

nick (not shown). These results indicate that full-length c-Myc

is converted into Myc-nick by a cytoplasmic protease that is

independent of the proteasome.

Myc Is Cleaved by a Calcium-Activated Calpainto Produce Myc-NickIn a systematic search for the proteases mediating cytoplasmic

cleavage of Myc, we ruled out both caspases and lysosomal

proteases on the basis of inhibitors and cleavage conditions

(Figure S2C). However, we found that all calpain inhibitors tested

including calpeptin, calpain inhibitor XII (Figure 3A), and calpain

inhibitor VI (Figure 3B) blocked the formation of Myc-nick in vitro

and in vivo. Moreover, MG132, although well known as a protea-

some inhibitor, has also been shown to inhibit calpains (Mailhes

et al., 2002). Calpains had been linked to Myc stability earlier, but

the antibodies used in those studies would not have detected

Myc-nick (Gonen et al., 1997; Small et al., 2002). Calpains

comprise a large family of calcium-dependent cytoplasmic

cysteine proteases that function at neutral pH and are primarily

associated with partial protein cleavage rather than complete

protein degradation. The most well studied members of this

family are mcalpain and mcalpain. These ubiquitously expressed

catalytic subunits form functional heterodimers with a calpain

regulatory subunit (calpain r) to bind calcium. To determine

whether Myc-cleavage is regulated by calcium-activated cal-

pains, we employed siRNA to knock down calpain r in cells

expressing c-Myc either endogenously or under control of

a retroviral vector. In both settings a partial silencing of calpain

r correlated with decreased Myc-nick and increased full-length

c-Myc (Figure 3C). Knockdown of m or mcalpains alone did not

block the formation of Myc-nick, most likely due to the presence

of other redundant calpains (not shown). Because calpain

activity is calcium dependent we next examined the effects on

Myc-nick of modulating calcium levels. Treatment with the

calcium chelators Bapta or EGTA reduced the ability of CE to

cleave c-Myc (Figures 3D and 3E). Incubation of either IVT

c-Myc (Figure 3F) or recombinant c-Myc (Figure 3G) with purified

m or mcalpain (together with calpain r) produced Myc-nick.

These results demonstrate that Myc-nick is directly generated

by calcium-dependant calpain cleavage of full-length c-Myc.

Lysine 298 Is the Primary Calpain Cleavage Site in c-MycTo map the calpain cleavage region on c-Myc, we used a series

of internal deletion mutants lacking 60 residue segments (c-Myc

DA-DG) (Tworkowski et al., 2002) (Figure 3H) and determined

whether any failed to generate a shorter c-Myc protein. Only

the deletion of residues 252–315 (DE) resulted in loss of a Myc-

nick-like product (Figure 3I). This region contains a high scoring

PEST domain, often present in unstable proteins and frequently

associated with calpain cleavage sites. We further narrowed our

search to residues 270–315, based on the ability of antibody 274

to detect Myc-nick (Figure S1A and Figure 3H). Although there is


no universal consensus for a calpain cleavage site, a comparison

of 106 sites present in 49 known calpain substrates indicated

a preference for calpain cleavage after K or R, and to lesser

extent Y, especially when these amino acids are flanked by P,

V, and L (Figure 3J) (Tompa et al., 2004). We noted a region local-

ized C-terminal to the PEST domain containing the sequence

PLVLKRC (Figure 3H, marked in red). This region is evolutionarily

conserved in c-Myc and N-Myc but not L-Myc (Figure S3A),

consistent with the fact that c-Myc and N-Myc, but not L-Myc,

are cleaved (Figures S3B and S3C). To determine whether this

region functions as a calpain cleavage site, we synthesized

a 10 amino acid peptide corresponding to the putative c-Myc

calpain cleavage region (291–300) and to another nearby region

of c-Myc (236–245) (Figure 3H, in red and blue, respectively) and

asked whether they could act as competitive inhibitors of Myc-

nick formation in vitro. Addition of increasing amounts of the

peptide containing the putative calpain cleavage site blocked

the formation of Myc-nick in vitro whereas the control peptide

B

- - - + CE- + - - NE

MycMyc-nick

A

10' 1 2 4 16 16 16 hours

C

+ + + + + - + CE

- - - - - - + MG- - - - - + - NE

MycMyc-nick

0 10 30 0 10 30 ng LBNuclear Cytoplasmic

MycMyc-nick

IB: N262

µM MG132 mM Lacta mM Epoxo

0 0.1 1 10 0 0.1 1 100 1 10 100

MycMyc-nick

D

MycMyc-nick

- + + - + + CE- - - - - + PI- - + - - - Boiled CE

FE

- + + + + + + + CE

G

MycMyc-nick

- + + CEMW Myc

Silver IB: 143 1 2 3 4 5 6 7 8

in vitroN C

- + - + - +MG132 (80nM) Lacta (200nM) Epoxo (400nM)

Myc

Myc-nick

N C N C

- + - + - +

IB: N262

in vivo

H

.01 .1 .5mg/µl CHX

.01 .1 .5mg/µl ActD

Figure 2. Myc-Nick Is a Product of c-Myc Cytoplasmic Cleavage Independent of the Proteasome

(A) CRM1-dependant nuclear export is not involved in the formation or localization of Myc-nick. Rat1 myc null cells infected with c-Myc were treated with

leptomycin B (LB) for 4 hr before harvesting.

(B) c-Myc gives rise to Myc-nick in vitro when incubated with cytoplasmic extracts. Radiolabeled IVT c-Myc was incubated for 2 hr with cytoplasmic (CE) or

nuclear extracts (NE) from Rat1 c-myc null cells.

(C) Timecourse of in vitro cleavage of c-Myc.

(D) Recombinant c-Myc is cleaved in the presence of CE. One microgram of recombinant c-Myc was incubated with 30 mg of CE and processed for western blot.

(E) IVT c-Myc was incubated with CE for 4 hr in the presence of of Actinomycin D (ActD) or cyclohexamide (CHX).

(F) The cleavage of c-Myc is inhibited by protease inhibitors and by heat inactivation. CE was boiled prior to incubation with IVT c-Myc and the protease inhibitor

(PI) was added to the incubation mixture.

(G and H) The cleavage of c-Myc is inhibited by MG132, but not by Lactacystein or Epoxomycin. (G) IVT c-Myc was incubated with CE, in the presence of

increasing amounts of MG132, Lactacystein, and Epoxomycin for 1 hr. (H) Rat1 myc null cells expressing c-Myc were incubated with MG132, Lactacystein,

or Epoxomycin for 2 hr prior to harvesting. Nuclear (N) and cytoplasmic (C) fractions are shown.

See also Figure S2.


had no effect (Figure 3K). Next, we generated a deletion mutant

of Myc lacking residues 291–300 and found it to be resistant to

cleavage by CE (Figure 3L). When this mutant was ectopically

expressed in 293T cells the cleavage was also reduced in

comparison to the wild-type (WT) c-Myc (Figure 3N), indicating

that this is the major calpain cleavage site within c-Myc. Next,

we made point mutations in the calpain cleavage region (labeled

with asterisks in Figure 3H) and assessed their cleavage. We

found that the K298A mutation reduced the cleavage of c-Myc

in vitro (Figure 3M) and in vivo when ectopically expressed in

293T cells (Figure 3N). Although K298 appears to be the major

calpain cleavage site, a K298A mutant was still cleaved in vivo

most likely because when this residue is mutated the cleavage

is shifted to R299 (and perhaps L297, V296). We also identified

weaker calpain cleavage sites localized in the C terminus of

Myc that become more pronounced in the cleavage-deficient

mutants.

Expression of Myc-Nick Alters Cell Morphologyand Increases Acetylation of a-TubulinTo study Myc-nick function, we generated a truncated form of

Myc containing amino acids 1–298 (referred to as Myc-nick*).

We found that ectopically expressed Myc-nick* is localized

predominantly to the cytoplasm and migrates on SDS-PAGE

with the same apparent molecular weight as Myc-nick generated

by cleavage of full-length c-Myc (Figure 4A). Myc-nick is

degraded at a comparable rate to full-length c-Myc with a half-

life of about 30 min (data not shown). Because Myc-nick

contains the MBI phosphodegron, with its GSK3b phosphoryla-

tion and Fbw7-binding sites, we would expect Myc-nick to be

N

- + + +0 .1 1 10

PHSPLVLKRC(aa 291-300)

SPEPLVLHEE(aa 236-245)

I

HA- c-Myc

c-Myc

c-Myc

ΔA

c-Myc

ΔB

c-Myc

ΔC

c-Myc

ΔD

c-Myc

ΔE

c-Myc

ΔF

c-Myc

ΔG

IB: 143+274

MycMyc-nick

L

MycMyc-nick

- - + + CE+ - + - Wt c-Myc- + - + Δ291-300

M

1 43963 126 189 252 315 378

A B C D E F G

SPEPLVLHEETPPTTSSDSEEEQEDEEEIDVVSVEKRQAPGKRSESGSPSAGGHSKPPHSPLVLK RCHVSTHQHNYAAPPST274 antibody

PEST domain** ****** **

- + + + CE0 .1 1 10 μg

K

A B

Sin 3A

Calpain r

Tubulin

MycMyc-nick

HCT116 HFF-Myc

C

- + - + Reg siRNA- + - + inhibitor VIN C

F

- - + - - Calpeptin

μ m Calpain

MycMyc-nick

+ + Purified c-Myc

Myc

Myc-nick

- + + + + + CE0 0 .1 1 10 100 μM EGTA

D

Myc

Myc-nick

- + + CE- + - Bapta

MycMyc-nick

E

MycMyc-nick

- + + + + +0 0 0.1 1 10 100

μM Calpeptin

0 1 10 100

μM inhibitor XII

+ + + + CE

H

P3 P2 P1V,L,F L,V RKY

V L K

J

G0 1 1 1 ng Calpain

- + - - - CE

MBIIIMBIIMBI NLS BHL LZH

MycMyc-nick

WT L295A K298A R299A

0 2 4 0 2 4 0 2 4 0 2 4 h incubation

Myc

Myc-nick

WT

Δ291-3

00

Vector

K298AK29

7A

IB: 143

MycMyc-nick

Tubulin

IB: 143+274

IB: N262 IB: N262

- + μCalpain

Figure 3. Myc-Nick Is Generated by Calpain

Cleavage of Full-Length Myc

(A) IVT c-Myc was incubated with CE for 1 hr in the

presence of the calpain inhibitors calpeptin and

calpain inhibitor XII.

(B) Dense cultures of HFF cells infected with a

c-Myc retroviral vector were incubated with cal-

pain inhibitor VI for 2 hr prior to harvesting.

(C) siRNA for calpain regulatory subunit (Reg)

reduces the formation of Myc-nick in HCT116

and HFF-Myc cells.

(D) Rat1 myc null cells were incubated for 2 hr in the

presence of the calcium chelant Bapta, and then

cytoplasmic extracts were prepared and incubated

with IVT c-Myc (as in 2B).

(E) IVT c-Myc was incubated with CE for 1 hr in the

presence of increasing amounts of EGTA.

(F) IVT c-Myc was incubated with purified recombi-

nant mcalpain or mcalpain and r subunit for 1 hr

on ice.

(G) One hundred nanograms of purified mcalpain

and r subunit was incubated with 2 mg of recombi-

nant c-Myc for 30 min on ice.

(H) Schematic representation of c-Myc protein

indicating A-G deletion regions (deletion end-

points indicated above), putative calpain cleavage

region in red.

(I) c-Myc and the deletion mutants DA-DG were

expressed in 293T cells, and 48 hr later the pres-

ence of a Myc-nick-like protein in cytoplasmic

extracts was determined.

(J) Amino acid preference for calpain cleavage

region according to Tompa et al. (2004). P1–P3

indicate the position of preferred residues in rela-

tion to the cleavage site, bold letters indicate the

c-Myc calpain cleavage site.

(K) IVT c-Myc was incubated with CE in the pres-

ence of a peptide that contains the potential

calpain cleavage site (amino acids 291–300 in

red) or a nearby sequence (amino acids 236–245

in blue).

(L) WT and D291–300 IVT c-Myc were incubated

with CE for 1 hr (as in 2B).

(M) IVT WT, L295A, K298A, and K299A c-Myc were

incubated with CE for the indicated time points.

(N) Cleavage products produced from the indi-

cated c-Myc mutants in 293T cells.

See also Figure S3.


degraded through similar proteasomal pathways as full-length

Myc. Indeed, blocking proteasome activity by pharmacological

inhibitors or by silencing E3 ligases and components of the pro-

teasome induced the stabilization of both full-length Myc and

Myc-nick (data not shown). Calpain inhibitors are incapable of

inducing accumulation of Myc-nick, indicating that calpains are

unlikely to play a major role in Myc-nick turnover in vivo.

Whereas overexpression of full-length c-Myc in Rat1a myc

null fibroblasts is associated with increased proliferation

and apoptosis, the ectopic expression of Myc-nick* had no

detectable effect on either cell doubling time or survival (data

not shown). However, Myc-nick*-expressing cells displayed

dramatic morphological changes—they appeared spindle-like

with long protrusions that occasionally formed intercellular con-

tacts (Figure 4C, Figure S4, Figure S5C). This morphology was

specific for Myc-nick* expression and could not be produced

by expressing the C-terminal 100 residues of c-Myc (not shown).

Introducing a scratch wound across a confluent cell monolayer

showed that whereas control cells migrated into the wound by

extending lamellipodia, Myc-nick*-expressing cells aligned

parallel to each other and extended long protrusions into the

wound (Figure 4B).

The morphological changes induced by Myc-nick* are

suggestive of altered cell-cell contacts and/or major cytoskeletal

reorganization. The elongated cellular protrusions present in

Myc-nick-expressing cells resemble specialized structures

formed by stable microtubules. Because stable microtubules

display increased acetylation of a-tubulin on lysine 40 (Hubbert

et al., 2002), we stained Myc-nick*-expressing cells with anti-

bodies against a-tubulin and acetylated a-tubulin. Figure 5A

shows that although acetylated a-tubulin immunostaining

in vector and c-Myc-expressing cells is low, Myc-nick* cells

display intense staining of elongated structures (Figures 5A

and 5F). This was confirmed by immunoblotting for acetylated

a-tubulin in Rat1 myc null or 293T cells expressing Myc-nick*

(Figure 5C, Figure S6B). Importantly neither the levels of total

cellular acetylated lysine (Figure S6A) nor the levels of tyrosi-

nated tubulin (not shown) were affected. To ask whether Myc-

nick* expression increased microtubule stability, we treated cells

expressing Myc-nick*, c-Myc, or vector with nocodozole for

15 min. This treatment disrupted unstable microtubules with

a half-life of about 10 min but not stable microtubules with

a half-life >2 hr. Only Myc-nick*-expressing cells possessed

microtubules that survived nocodozole treatment and these

stained strongly with antiacetylated a-tubulin (Figure 5B).

Myc-Nick Interacts with a- and b-TubulinsWe observed partial cytoplasmic colocalization between Myc

and a-tubulin in several cell types expressing c-Myc or Myc-

nick (Figure 5E; Figures S5A–S5C). Using 293T cells expressing

GFP-tubulin and Myc-nick we performed immunoprecipitation

with anti-GFP and immunobloted with anti-Myc. Myc-nick was

coimmunoprecipitated with GFP-tubulin but not GFP-EB1,

a microtubule-binding protein (Figure S5D). In addition, immuno-

precipitation of either c-Myc or endogenous a-tubulin from cyto-

plasmic extracts indicated that Myc-nick interacts with a-tubulin

in vivo (Figure 5D). To examine in vitro interactions we incubated35[S]-methionine-labeled IVT c-Myc with purified brain tubulins

vector Myc Myc-nick*- - + - MG132

N

C

A

C

MMMyyycccVVVeeeccctttooorrr

MMMyyyccc---nnniiiccckkk *** MMMyyyccc---nnniiiccckkk ***VVVeeeccctttooorrr

MMMyyyccc---nnniiiccckkk***MMMyyyccc---nnniiiccckkk***

Myc

Myc-nick

Myc

Myc-nick1 2 3 4

IB: N262

666000μμμMMM

B

MMMyyyccc---nnniiiccckkk ***VVVeeeccctttooorrr

DA

PI/αα

-tu

bu

lin

Figure 4. The Expression of Myc-Nick* Promotes Changes in Cell Morphology

(A) Myc-nick* (1–298) is cytoplasmic and has the same apparent molecular weight as Myc-nick derived from full-length c-Myc (compare lanes 2 and 4, upper

panel). Rat1 myc null cells expressing empty vector, full-length c-Myc, and Myc-nick were fractionated into nuclear (N) and cytoplasmic (C) fractionations.

(B) Myc-nick*-expressing cells extend protrusions at the wound edge. A confluent monolayer of Rat1 myc null cells expressing either vector or Myc-nick was

scratched using a 100 ml tip and phase contrast images were taken at 12 hr.

(C) Rat1 myc null cells expressing empty vector, c-Myc, and Myc-nick at 14 days after selection.

See also Figure S4.


for 1 hr, then immunoprecipitated a, b, g, bIII and acetylated

a-tubulin and exposed the gel to detect radioactive Myc. Full-

length c-Myc was coimmunoprecipitated mostly with b-tubulin

(Figure S5E, middle panel). We also incubated IVT c-Myc with

CE for 1 hr to produce Myc-nick and then performed immuno-

precipitation for tubulins as above. The results showed that

Myc-nick interacts with a- and b-tubulin (Figure S5E, upper

panel). The binding of Myc-nick to tubulin is consistent with the

previous report of an N-terminal region of Myc capable of asso-

ciating with tubulin (Alexandrova et al., 1995).

Myc-Nick Directly Regulates a-Tubulin Acetylationthrough GCN5Because Myc-nick lacks the Myc C-terminal dimerization and

DNA-binding domains, we surmised that the increase in acety-

lated a-tubulin induced by Myc-nick* expression is independent

of Myc transcriptional activity. One possibility is that Myc-nick

αα−t

ub

ulin

/D

AP

I

VVVeeeccctttooorrr ccc---MMMyyyccc MMMyyyccc---nnniiiccckkk***

A

B

111000μμμMMM

acetyl. α−tub/DAPI

ac

ety

l.

α−t

ub

/D

AP

I VVVeeeccctttooorrr ccc---MMMyyyccc MMMyyyccc---nnniiiccckkk***

α−t

ub

ulin

ac

ety

l.

α−t

ub

10μM

Noc

odaz

ole

15'



α-tubulin

γ-tubulin

acetyl α-tub

actin

Puro c-Myc

Myc-ni

ck*C

HDAC 6

HDAC 3

MycMyc-nick

IP

CN CN CN

Input IP:MycIB:α

-tub CN CN

α-tubulin

DMyc

-nick

*

IB:M

yc

Input α-tub IgG

666000μμμMMM

666000μμμMMM

Myc-nick*

acetyl α-tub

Merge

***

E F

Figure 5. Myc-Nick* Cells Display In-

creased Levels of Acetylated a-Tubulin

and Microtubule Stabilization

(A) Immunofluorescence for a-tubulin (upper

panels) and acetylated a-tubulin (lower panels) of

Rat1 myc null cells infected with empty vector,

c-Myc, or Myc-nick*.

(B) As for (A) but incubated in the presence of

nocodazole for 15 min prior to fixation.

(C) Immunoblotting of Rat1 cell extracts using anti-

bodies against the indicated proteins.

(D) NE or CE of Rat1 myc null cells express-

ing c-Myc were immunoprecipitated with anti

a-tubulin or normal mouse IgG and immunoblotted

for Myc (top panel), or immunoprecipitated with

anti-c-Myc N262 antibody, and immunoblotted

with anti-a-tubulin.

(E) Immunofluorescence for Myc and acetylated

a-tubulin in A431 lung epithelial cells cell trans-

fected with Myc-nick.

(F) Detail of Myc-nick-expressing cell stained for

acetylated a-tubulin.

See also Figure S5.

recruits an acetyltransferase to microtu-

bules. We observed marked acetylation

of a-tubulin in cytoplasmic extracts incu-

bated in the presence of either recombi-

nant c-Myc (Figure 6A) or IVT c-Myc

(Figure 6B and Figure S6C). Importantly,

Myc proteins are known to associate

with the acetyltransferases GCN5 and

TIP60 (Frank et al., 2003; Sterner and

Berger, 2000) via TRRAP that interacts

with MBII (residues 106–143) (McMahon

et al., 1998). We therefore tested a Myc-

nick deletion mutant lacking Myc box II

(DMBII) and found that its ability to pro-

mote a-tubulin acetylation was reduced

(Figures 6C and 6D). In addition, the

DMBII Myc-nick* mutant failed to induce

the cell morphological changes that we

had detected with WT Myc-nick* whereas a comparably sized

deletion in a region adjacent to MBII was similar to WT (Figure 6E).

The dependence on MBII for tubulin acetylation and altered

cell morphology suggests involvement of the acetyltranferases

known to bind this region. We detected substantial amounts of

GCN5 and TRRAP in the cytoplasm (Figure S6D), whereas

Tip60 was predominantly nuclear (not shown). To test GCN50sinvolvement in a-tubulin acetylation we performed siRNA-medi-

ated knockdown of TRRAP and GCN5 in Myc-nick*-expressing

cells and found that decreasing either one of these proteins

reduced the levels of acetylated a-tubulin (Figure 6F) and

reverted the changes in cell morphology induced by Myc-nick

to that of vector-infected cells (Figure 6G). Moreover, GCN5

coimmunoprecipated with a-tubulin and Myc in cytoplasmic

extracts of 293T cells transfected with GCN5 (Figure 6H).

Both ectopic expression of GCN5 in 293T cells (Figures 6D

and 6I) and addition of full-length recombinant GCN5 to


cytoplasmic extracts (Figure 6J, upper panel) resulted in

increased levels of acetylated a-tubulin. The addition of Myc

further increased the levels of a-tubulin acetylation induced by

GCN5 in cytoplasmic extracts (Figure 6J). Full-length recombi-

nant GCN5 also induced the acetylation of a-tubulin present

in assembled microtubules (Figure 6K) and synergized with

c-Myc to promote further acetylation (Figure 6K, compare

lanes 1 to 4 and 2 to 5). The GCN5 catalytic domain alone was

L+ - + + - + - c-Myc 0.5μg

- + + + + + - microtubules 5μg

Ponceau

TubulinMyc

GCN5 catalycticdomain

- - - - + + + GCN5 0.1μg

acetyl α-tub

α-tubulin

β-tubulin

acetyl α-tub

- - + + c-Myc IVT+ + - - vector IVT

B

α-tubulin

- + + + GTP- - 0.1 1 μg c-Myc+ + + + CE

acetyl α-tub.

A

TRRAP

GCN5

acetyl α-tub

Myc-nick

α-tubulin

si RNATRRAP

GCN5co

ntrol

HDAC6

F

+

acetyl α-tub

acetyl α-tub

J

- - - + + c-MycGCN5

- - + + c-Myc

GCN510% input

IP: α-tubulinIB: GCN5

H

IP: MycIB: GCN5

- - - + + + c-Myc 0.2μgGCN5

p300- - - + + + c-Myc 0.2μg

G

acetyl α-tub

Myc-ni

ck*

vecto

r

Myc IVT

c-MycC

α-tubulin

acetyl α-tub

GCN5

α-tubulin

Vector

GCN5I

D

α-tubulin

*

Vector Myc-nick*

Myc-nick* Δ106-143(ΔMBII) Myc-nick* Δ145-160

Myc-nick

GCN5

E

GCN5nic

k*ΔMBII

vecto

r

Myc-ni

ck*

K

acetyl α-tub

acetyl α-tub

acetyl α-tub

invi

troC

Ein

vivo

293T

cells

p300 1 2 3 4 5 6 7

1.0 1.6 1.9 1.8 2.4 2.7 fold change

TRRAP si RNAGCN5 si RNAControl si RNA

nick*

ΔMBII

0μM0μM

Vector

GCN5

30μg

CE

5μg

mic

rotu

bule

s

1.0 1.1 1.1 2.4 2.6 2.5 fold change

1.3 2.7 1.0 3.1 4.2 fold change

0.9 1.0 1.2 1.1 fold change

66

Figure 6. c-Myc and the HAT GCN5 Promote a-Tubulin Acetylation

(A) CE was incubated with recombinant c-Myc and c-Myc dilution buffer or (B) with c-Myc IVT or vector IVT for 1 hr at 37�C. The samples were immunoblotted as

indicated.

(C) CE was incubated with IVT vector, c-Myc, Myc-nick, and Myc-nick DMBII (D106–143) for 30 min at 37�C, then processed for immunoblotting.

(D) 293T cells were transfected with empty vector, GCN5, Myc-nick, and Myc-nick DMBII and processed for immunoblotting after 48 hr.

(E) Rat1 myc null cells were infected as indicated and photographed 10 days after selection.

(F and G) Rat1 myc null cells expressing Myc-nick were transfected with 100 nM of control, TRRAP, or GCN5 siRNA and 76 hr later processed for immunoblotting

(F) or photographed (G).

(H) CE of 293T cells transfected with empty or GCN5 vectors were immunoprecipitated with anti-a-tubulin or anti-Myc (143+274) and immunoblotted for GCN5.

(I) 293T cells were transfected with control or GCN5-expressing vectors and 48 hr later processed for immunoblotting.

(J–L) GCN5 induces acetylation of a-tubulin. (J) CE were incubated with 200 ng of Myc, 100 ng or 300 ng of GCN5 (upper panel), or 100 ng or 500 ng of p300 (lower

panel). (K) Assembled microtubules were incubated with 100 ng or 500 ng of recombinant full-length GCN5 (upper panel) or p300 (lower panel) in the presence or

absence of 200 ng of purified c-Myc. (L) Purified assembled microtubules were incubated with recombinant c-Myc and GCN5-catalytic domain. The asterisk

indicates a nonspecific bacterial band copurified with GCN5-catalytic domain.

See also Figure S6.


Tubulin

S M D D DensityGM DM

Tropomyosin

MycMyc-nick

MycMyc-nick

αT5

8Pα

N26

2

C2C12

D

GM DM

aacceettyyll αα--ttuubb //DDAAPPII

6600μμMM2200μμMM

cc--MMyycc//DDAAPPIIcc--MMyycc

6600μμMM

2200μμMM

VVeeccttoorr MMyycc--nniicckk

MMyycc--nniicckk**

VVeeccttoorr

6600μμMM

aacceettyyll αα--ttuubb //DDAAPPIIaacceettyyll αα--ttuubb //DDAAPPII

6600μμMM2200μμMM

C2C12 DM

cc--MMyycc//DDAAPPII

Ecc--MMyycc

MycMyc-nick

prim

ary

mou

sem

yobl

asts

S D D DensityGM DM

α-tubulin

TropomyosinTroponin C

acetyl α-tub

Mycnick*nick* ΔMBII

vecto

rΔMBII

RD

Rha

bdom

yosa

rcom

a

nick*

S D S D S D S DHFF RHJT RD RH1

alveolar embryonal

1 2 3 8 16 weeks

mouse hindlimb muscles

Myc

Myc-nick

Tropomyosin

A

myc -/-MyoD

TroponinC

Myogenin

α-tubulin

GM GM DM - S GM DM - S GM

myc +/+MyoDC C

O

Troponin C

Desmin

acetyl α-tubα-tubulin

Myc-nick*

nick*

vecto

r

hum

anm

yobl

asts

(pec

tora

l)

H

Myogenin

TroponinCGM GM DM GM DM

MyoD MyoD+nicknick

Rat

1m

yc-/-

α-tubulin

TroponinC

Tubulin

B

G

Tro

po

nin

C/D

AP

IP

hase

6600μμMM

2200μμMM

Rhabdomyosarcoma

MyoD MyoD+Myc-nick*Rat1 myc+/+

N

1.0

0.20.40.60.8

1.21.41.6

C

Cal

pain

activ

ity

1.8

GM DM

primary myoblasts

Calpain3acetyl α-tub

α-tubulin

Myc-nick*

hum

anm

yobl

ast

(rect

usab

dom

inus

)

VVeeccttoorr MMyycc--nniicckk

hum

anm

yobl

ast

(rec

tus

abdo

min

us)

nick*

vecto

r

ML

MMyycc--nniicckk**

VVeeccttoorr

6600μμMM

aacceettyyll αα--ttuubb //DDAAPPII

F

Subtype

I

Troponin C

Myogenin

J

K

P

Myogenin

GM DM

TroponinC

Caveolin 3

Desmin

MyoD

α-tubulin

- + - + Myc-nick*

acetyl α-tub

Tropomyosin

C2C

12

C2C

12

Figure 7. Myc-Nick Accelerates Muscle Cell Differentiation

(A) Hindlimb muscles dissected from 1-, 2-, 3-, 8-, and 16-week-old mice were processed for immunoblotting using N-terminal anti-Myc sera (anti 143+274).

(B) Mouse primary myoblasts isolated from hindlimb muscles of 8-week-old mice were cultured as sparse cultures for 24 hr or as dense cultures in the presence of

either growth medium (GM) or differentiation medium (DM) for 3 days. Total cell extracts were immunoblotted for c-Myc (anti 143+274) and indicated proteins.

(C) Mouse primary myoblasts were cultured in GM or DM for 3 days, lysed in buffer A, and total calpain activity was measured using Suc-LLVY-AMC synthetic

substrate (n = 2; calpain activity in DM was compared with calpain activity in GM; set to 1 ± standard error of the mean [SEM]).

(D) C2C12 mouse myoblasts cultured at sparse (S), medium (M), or high densities (D). Dense cultures were harvested or switched to DM for 7 days. Total cell

extracts were immunoblotted for c-Myc using antibodies against total c-Myc (N262) or against phosphorylated T58 Myc, a signal for Myc degradation.

(E) C2C12 cells cultured in DM for 7 days and stained for endogenous c-Myc (anti-N262).

(F) C2C12 cells grown in the presence of GM or DM for 5 days and stained with anti-acetylated a-tubulin.

(G) Western blotting for Myc in rhabdomyosarcoma cell lines grown as dense or sparse cultures for 3 days.

(H and I) Human myoblasts expressing vector, or Myc-nick were cultured in DM and processed for immunoblotting after 4 days.

(J) Human myoblasts expressing vector or Myc-nick were cultured in DM and photographed after 2 days.


sufficient to induce a-tubulin acetylation, but no synergy with

Myc was detected, most likely because association between

these two proteins occurs outside of GCN50s active site (Fig-

ure 6L). As a further control, we tested p300, a HAT that binds

the C terminus of Myc (Faiola et al., 2005). p300 neither induced

tubulin acetylation nor did it synergize with Myc (Figures 6J and

6K, bottom panels). These data indicate that GCN5 can specifi-

cally acetylate tubulin and that Myc augments tubulin acetylation

by binding to tubulin and recruiting GCN5.

Myc-Nick Is Produced in Differentiating Cellsand TissuesWhen examining the expression of Myc-nick in adult mouse

tissues, we found that brain, cerebellum, and skeletal muscle

express significantly higher levels of Myc-nick than any other

tissue (Figure 7A, Figure S7A). Interestingly, both neuronal and

muscle differentiation require major cytoskeletal rearrangements

that have been associated with increased microtubule stability

and elevated levels of acetylated a-tubulin. For example, it has

been demonstrated that the acetylation of a-tubulin by the

Elp3 acetyltransferase is required for proper cortical neuronal

migration and differentiation (Creppe et al., 2009). Increased

levels of acetylated a-tubulin are found during myogenic differ-

entiation (Gundersen et al., 1989). Importantly, inhibiting the

activity of the deacetylases HDAC6 and Sirt2 (known to deace-

tylate a-tubulin) generates augmented levels of acetylated

a-tubulin and promotes differentiation of myoblasts (Iezzi et al.,

2004). Furthermore, the presence of the primary cilium, a micro-

tubule-based antenna-like structure composed of acetylated

a-tubulin, is essential for cardiomyocyte differentiation (Clement

et al., 2009).

The extensive cytoskeletal changes that occur during muscle

differentiation are regulated by calcium and calpains (Dedieu

et al., 2004). The inhibition of calpain activity either by pharmaco-

logical inhibitors or by overexpression of Calpastatin (an endog-

enous inhibitor of calpains) blocks muscle cell differentiation

(Dedieu et al., 2004). In addition, both types of ubiquitous

calpains (m and m) were shown to regulate muscle cell differen-

tiation in vitro (Moyen et al., 2004). Moreover, mutations in the

calpain 3 gene (the muscle-specific calpain) cause limb girdle

muscle dystrophy 2A (LGMD2A) (Richard et al., 1995). Finally,

knocking out calpain r (the regulatory subunit shared by all

calcium-dependent calpains) is lethal due to impaired cardiovas-

cular development (Arthur et al., 2000).

To study the cleavage of Myc to Myc-nick during the process

of muscle differentiation, we employed human primary myo-

blasts purified from pectoral girdle, mouse myoblasts purified

from hindlimb muscles, and C2C12 cultured mouse myoblasts.

When stimulated to differentiate, these cells displayed low

levels of full-length c-Myc and high levels of Myc-nick when

compared to undifferentiated cycling cells (Figures 7B and 7D;

Figure S7B). Interestingly, Myc-nick levels were higher in a rhab-

domyosarcoma cell line from the alveolar subtype (Figure 7G),

a very aggressive tumor derived from partially differentiated

muscle cells.

The increased levels of Myc-nick in differentiated primary

mouse myoblasts correlate with an increase in calpain 3 levels

(Figure 7B; Figure S7B) and in total calpain activity (Figure 7C).

Similarly, when C2C12 cells were stimulated to differentiate,

they also displayed an increase in total calpain activity (Fig-

ure S7C) and in the ability to cleave Myc in vitro (Figure S7D).

Importantly, the levels of acetylated a-tubulin were also elevated

during the process of differentiation (Figure 7B; Figures S7B and

S7F). Acetylation of a-tubulin is accompanied by myoblast fusion

to form multinucleated myotubes (Figure 7F; Figure S7F). When

myoblasts are stimulated to differentiate they fuse into multinu-

cleated myotubes. In addition to displaying increased levels of

acetylated a-tubulin, these myotubes show decreased immu-

nostaining for nuclear Myc and increased cytoplasmic Myc

staining (Figure 7E). Conversely, undifferentiated satellite cells

in the same culture display nuclear staining for Myc (Figure S7E).

This is consistent with our results showing conversion of full-

length Myc into predominantly cytoplasmic Myc-nick.

In C2C12 cells, concomitant with increased Myc-nick abun-

dance, we observed a decrease in Myc-nick phosphorylation

at threonine 58 (T58) after the switch to differentiation conditions

(Figure 7D). Phospho-T58 mediates Fbw7-dependent degrada-

tion of Myc by the proteasome. Decreased phosphorylation at

this site is consistent with the notion that stabilization of Myc-

nick contributes to its elevated levels. Interestingly we found

that Myc-nick phosphorylation at T58 is also reduced in adult

muscle, brain, and cerebellum (Figure S7A and data not shown).

In summary, we found that during muscle differentiation there

is an increase in Myc-nick, concomitant with an elevation in cal-

pain activity and tubulin acetylation (Figures 7B–7D; Figure 7F;

Figures S7B–7D; Figure S7F).

Myc-Nick Accelerates Muscle DifferentiationWe examined the effects of ectopic expression of Myc-nick* in

human primary myoblasts isolated from the pectoral girdle

(Figure 7H) and rectus abdominus (Figures 7I–7J), from the

human rhabdomyosarcoma cell line RD (Figure 7K), and in

C2C12 cells (Figures 7L and 7M). In all cells we found that

Myc-nick* expression accelerated muscle cell differentiation,

augmented a-tubulin acetylation, and elevated expression of

muscle-specific proteins (Figures 7H–7M).

(K) RD rhabdomyosarcoma cells expressing vector, Myc-nick, or Myc-nick DMBII (D106–143) were grown in DM for 4 days and processed for immunobloting.

(L) C2C12 cells expressing vector or Myc-nick were grown as dense cultures and stimulated to differentiate for 4 days and then photographed.

(M) C2C12 cells expressing vector or Myc-nick were grown to confluency and harvested or stimulated to differentiate for 3 days and then harvested. Total cell

lysates were immunoblotted with antibodies against the indicated proteins.

(N) Rat1 (myc+/+) cells expressing MyoD or MyoD + Myc-nick were cultured in DM for 3 days and photographed (lower panels) or stained for Troponin C and DAPI

(upper panels).

(O) Rat1 (myc+/+) or Rat1 myc null (myc�/�) cells expressing vector (lane C) or MyoD were cultured in GM, DM, or serum-free medium (lane �S) for 4 days and

processed for immunobloting.

(P) Rat myc null (myc�/�) cells expressing Myc-nick, MyoD, or MyoD + Myc-nick were grown in GM or DM for 3 days and processed for immunoblotting.

See also Figure S7.


The expression of Myc-nick* in C2C12 myoblasts promoted

an increase in cell fusion not only in confluent cultures (Fig-

ure 7L; Figure S7H) but also in sparse cultures (Figure S7G).

The role of Myc-nick in promoting muscle cell differentiation is

partially dependent on MBII, as the differentiation of RD and

C2C12 cells by Myc-nick can be delayed but not inhibited by

the deletion of MBII (Figure 7K; Figure S7I).

To examine the requirement for Myc-nick production during

differentiation further, we first tested calpain inhibitor XII and

found that it reduces differentiation and tubulin acetylation

(Figure S7J). Next we employed c-Myc mutant (D291–300),

which is deficient in cleavage to Myc-nick. Similar to full-length

WT c-Myc, this mutant can induce proliferation, apoptosis,

fibrillarin expression, and phosphorylated H2AX in Rat1 myc

null cells (Figures S7L–S7M). However the D291–300 c-Myc

mutant significantly reduces C2C12 cell differentiation when

compared to c-Myc, even though the proteins are expressed

at equal levels. Whereas full-length c-Myc only blocks the

fusion of C2C12 myoblasts, the c-Myc mutant (D291–300) with

reduced ability to generate Myc-nick dramatically reduces the

expression of muscle markers in addition to blocking fusion

(Figures S7N–S7O).

Myc-Nick Renders Rat1 myc Null Cells Competentto DifferentiateMyoD has been long known to induce transdifferentiation in

diverse cell types. We found that MyoD induces expression of

muscle-specific markers in Rat1 fibroblasts but not in Rat1

myc null fibroblasts (Figure 7O), indicating that myc is necessary

for the transdifferentiation process. Expressing Myc-nick

together with MyoD in myc+/+ Rat1 fibroblasts further elevated

the levels of muscle-specific markers and promoted cell fusion

compared to MyoD alone (Figure 7N). Importantly, we find that

in Rat1 myc null cells, the coexpression of Myc-nick (Figure S7K)

permitted transdifferentiation to muscle in response to MyoD

(Figure 7P). This experiment indicates that Myc-nick is sufficient

for MyoD-induced transdifferentiation, supporting the idea that

conversion of full-length Myc to Myc-nick is important in

MyoD-induced differentiation.

DISCUSSION

Recent studies have demonstrated that Myc directly regulates

transcriptional activation through the three RNA polymerases,

as well as transcriptional repression, and DNA replication (Eilers

and Eisenman, 2008). Here we identified and characterized Myc-

nick, a novel form of Myc performing transcription-independent

functions in the cytoplasm. One of these functions is to regulate

a-tubulin acetylation in cooperation with the HAT GCN5. More-

over, Myc-nick levels are elevated in differentiated muscle

tissues and Myc-nick overexpression accelerates muscle cell

differentiation.

Despite its widespread expression, Myc-nick has not been

previously characterized. There are several likely reasons for this.

First, the most commonly used antibodies against Myc (such as

9E10) recognize its C terminus and would not detect Myc-nick.

Second, the proportion of Myc-nick to Myc increases when

cells are grown as confluent cultures, conditions that are not

commonly employed when studying Myc proteins. Third, most

studies use total protein or nuclear extracts and have not analyzed

the cytoplasmic pool of Myc. Fourth, Myc-nick may have been

mistaken for the similarly sized MycS protein (Hann et al., 1988),

a nuclear localized product of internal translation initiation.

Calpain Cleavage as a Posttranslational FunctionalSwitchWe found that Myc-nick is generated in the cytoplasm by

calpain-mediated cleavage of full-length Myc. Cytoplasmic

cleavage by calpains has been reported for over 100 proteins

including many cytoskeletal proteins, membrane receptors,

and transcription factors (Tompa et al., 2004). This number is

probably an underestimate because the lack of a clearly defined

consensus cleavage site for calpains makes the identification of

novel calpain substrates difficult. For most substrates the role of

calpain cleavage is not known. However, there are several note-

worthy examples of calpain cleavage operating as a functional

switch (Abe and Takeichi, 2007; Yousefi et al., 2006). Here we

have shown that the cleavage of Myc by calpain converts this

predominantly nuclear transcription factor into Myc-nick, a cyto-

solic factor that regulates a-tubulin acetylation. Based on our

findings and the examples in the literature, we surmise that the

partial proteolytic cleavage of proteins by calpains functions as

an irreversible posttranslational modification.

A Role for Myc-Nick in a-Tubulin AcetylationWe have shown that Myc-nick mediates the acetylation of

a-tubulin by forming a complex with microtubules and the HAT

GCN5. GCN5 has been long known to associate with nuclear

Myc through the highly conserved Myc box II, a region retained

in Myc-nick. Although tubulin acetylation was first described

20 years ago, little is known about the enzymes that catalyze

this reaction. Recently Elp3, a histone acetyltransferase, was

demonstrated to be critical in acetylation of a-tubulin in cortical

projection neurons, an event linked to neural differentiation and

migration (Creppe et al., 2009). We surmise that GCN5 repre-

sents another acetyltransferase targeting tubulin. Deacetylation

of a-tubulin is mediated by both HDAC6 and Sirt2 (Hubbert

et al., 2002; Matsuyama et al., 2002; North et al., 2003). However

the increase in acetylated a-tubulin by Myc-nick does not occur

through modulation of HDAC activity (Figures S6E and S6F).

A connection between a-tubulin acetylation and calpain acti-

vation was previously suggested by two independent studies.

First, calcium depletion, which inactivates calpains, was shown

to decrease the levels of acetylated a-tubulin in epithelial cells

(Ivanov et al., 2006). Second, the ectopic expression of calpain

6 causes microtubule stabilization, elevates the levels of acety-

lated a-tubulin, and impairs cytokinesis in HeLa cells (Tonami

et al., 2007). Our findings suggest that the induction of a-tubulin

acetylation by calpains could be mediated at least in part by

Myc-nick and GCN5.

The Role of Myc-Nick in Terminal DifferentiationThe role of Myc family members in differentiation is complex.

Endogenous Myc is strongly downregulated during terminal

differentiation of many cell types. Moreover, ectopic expression

of Myc has been shown to block terminal differentiation. The


ability of Myc to negatively regulate differentiation is consistent

with its role in maintaining pluripotency in ES cells and in the

generation of induced pluripotent stem (iPS) cells (Cartwright

et al., 2005; Takahashi et al., 2007). However, Myc has also

been implicated in promoting both proliferation and differentia-

tion in specific cellular contexts such as in progenitor cells of

the skin (Gandarillas and Watt, 1997; Gebhardt et al., 2006), B

lymphocytes (Habib et al., 2007), and hematopoietic stem cells

(Wilson et al., 2004). Our data suggest that whereas full-length

Myc blocks differentiation at the transcriptional level, Myc-nick

may be involved in promoting differentiation through transcrip-

tion-independent mechanisms. In agreement with this hypoth-

esis there have been numerous reports of Myc antigenicity

localized predominantly in the cytoplasm of differentiated cells

(see Introduction). In addition, we show that ectopic expression

of Myc-nick accelerates muscle cell differentiation and renders

Rat1 myc null cells competent to differentiate into muscle

following introduction of MyoD.

A number of studies have shown a strong correlation between

muscle cell differentiation, calcium influx, and calpain activation

(Dedieu et al., 2004; Kumar et al., 1992). In addition, limb girdle

muscular dystrophy 2A (LGMD2A) is caused by mutations in

calpain 3 that affect its activity (Richard et al., 1995). During

the process of normal differentiation, myoblasts elongate and

fuse into syncytial myotubes. An early event during this process

is the remodeling of the microtubule cytoskeleton, involving

disassembly of the centrosome and the alignment of stable

microtubules into a parallel array along the long axis of the cell.

We observed an increase in the levels of acetylated a-tubulin

(an indicator of microtubule stabilization and cytoskeletal reorga-

nization) and an elevation in Myc-nick abundance during muscle

cell differentiation. Induction of a-tubulin acetylation and micro-

tubule stabilization are likely to be important events in terminal

differentiation especially when cells must establish and maintain

a new shape. Acetylation of a-tubulin was also demonstrated to

regulate cortical neuron differentiation and migration (Creppe

et al., 2009).

We propose a model for the regulation of Myc by calpains and

for the function of Myc-nick. During proliferation, full-length Myc

is rapidly synthesized and transported to the nucleus where it

transcriptionally activates growth- and proliferation-related

genes and represses genes involved in differentiation. External

cues that stimulate differentiation, such as cell-cell contact,

and calcium influx can lead to the activation of calpains. Acti-

vated calpains interact with and may retain Myc in the cytoplasm

where it is cleaved to produce Myc-nick. As the cleavage

removes the nuclear localization sequence (NLS) and the DNA-

and Max-binding domains, Myc-nick is predominantly cyto-

plasmic and transcriptionally inactive. Indeed we have shown

that Myc-nick, when expressed in cells devoid of full-length

Myc, does not stimulate proliferation, growth, or apoptosis—

the transcription-dependent functions of full-length Myc. In

such Myc-nick-expressing cells we instead observe morpholog-

ical changes and increased tubulin acetylation, which are unique

to Myc-nick. During muscle differentiation Myc-nick appears to

be stabilized by dephosphorylation of threonine 58 within the

Myc box I phosphodegron. We propose that one of the functions

carried out by Myc-nick in the cytoplasm involves binding tubulin

and recruiting GCN5 to mediate a-tubulin acetylation. Myc-nick

is also likely to have additional functions and partners in the cyto-

plasm as the deletion of MBII in Myc-nick, which should block

the interaction with GCN5, only partially reduced the ability of

Myc-nick to promote muscle differentiation. We predict that

the functions carried out by Myc-nick in the cytoplasm coop-

erate to promote cytoskeletal changes that can further drive

terminal differentiation. In our model, the cleavage of Myc by

calpains has two roles. First, it helps diminish nuclear Myc abun-

dance preventing newly synthesized Myc from entering the

nucleus, therefore eliminating the transcriptional blockade to

differentiation caused by Myc. Second, the production of Myc-

nick influences cytoskeletal organization and facilitates terminal

differentiation. We predict that Myc-nick will play a general role

linking myc to both nuclear functions and cytoplasmic organiza-

tion during differentiation of a wide variety of cell types.


Retroviral Infection and Transfection

For retroviral production, 293T cells were cotransfected with pBabe-puro

vector expressing Myc clones and amphotropic helper. Infected cells were

selected with 2 mg/ml puromycin for 4 days. Rat1 myc null cells were harvested

10–14 days after infection and C2C12 cells were stimulated to differentiate

after selection. For overexpression experiments, 293T (transfected with

Fugene-Roche) cells were harvested 3 days after transfection with a change

in culture medium 24 hr before harvesting. Sparse cultures (S) are about

20% confluent, medium cultures (M) are about 40%–60% confluent, and

dense cultures (D) are grown as confluent cultures for 2–4 days.

Total Cell Lysates and Nuclear and Cytoplasmic Fractionation

For total extracts, cells were lysed in either boiling sample buffer or in RIPA

buffer. For cellular fractionation, cells were lysed in buffer A on ice for 20 min,

centrifuged for 3 min, and the supernatant employed as the cytoplasmic frac-

tion. The pellet was resuspended in buffer B, rotated for 20 min, sonicated, and

centrifuged for 10 min. The supernatant was employed as the nuclear fraction.

Ten to twenty micrograms of total extracts or 10 mg of nuclear and 30 mg of cyto-

plasmic extracts were probed overnight with indicated antibodies. See

Extended Experimental Procedures for composition of buffers.

Immunofluorescence

Cells were grown on glass coverslips and fixed with 4% paraformaldehyde

for 20 min, permeabilized with 0.5% Triton X-100 for 5 min, and blocked

with Image IT FX signal enhancer (Invitrogen) for 30 min. Primary and

secondary antibodies were diluted in PBS at 1:200 and incubated for 1 hr.

In Vitro Cleavage of Myc

All cDNAs were cloned into pCS2+ and were transcribed from the SP6

promoter using the Promega wheat germ system, in the presence of cold

(Promega) or 35[S]-labeled methionine (Perkin Elmer) according to the manu-

facturer’s instructions. For the in vitro cleavage experiments, Myc (1 ml IVT

Myc or 1 mg recombinant Myc) was incubated with 30 mg of nuclear or cyto-

plasmic extracts in 20 ml of buffer G at 37�C and the samples were processed

for autoradiography or western blot as specified. Nuclear and cytoplasmic

fractions were dissolved in the same buffer and adjusted to identical salt

concentration for these experiments. For Myc cleavage using recombinant

calpains, 1 ml IVT c-Myc or 0.25 mg of purified c-Myc was incubated with the

indicated amounts of calpains in the presence of 20 ml buffer G. Total calpain

activity was measured using a Calbiochem kit.

In Vitro Tubulin Acetylation Assays

CE (30 mg) or purified assembled microtubules (1 mg) were incubated with 1 ml

of IVT c-Myc or IVT vector or with recombinant Myc for 1 hr in the presence of


acetylation buffer. Assembled microtubules were incubated at 30�C and CE

at 37�C.

See Extended Experimental Procedures for buffer composition, constructs,

siRNA sequence, pharmacological inhibitors, antibodies, cell lines, and culture

media.



seven figures and can be found with this article online at doi:10.1016/j.cell.

2010.06.037.

ACKNOWLEDGMENTS

We are grateful to Nao Ikegaki, William Tansey, John Sedivy, and Peter Hurlin

for reagents essential for this work. We also thank Stephen Tapscott, Maura

Parker, Hector Rincon, Bill Carter, Linda Wordeman, and members of the

Eisenman lab for advice and discussion. M.C.-S. was a recipient of an

EMBO Long-Term Fellowship. This work was supported by NIH/NCI grant

CA20525 (R.N.E.).

Received: November 9, 2009

Revised: March 2, 2010



REFERENCES

Abe, K., and Takeichi, M. (2007). NMDA-receptor activation induces calpain-

mediated beta-catenin cleavages for triggering gene expression. Neuron 53,

387–397.

Alexandrova, N., Niklinski, J., Bliskovsky, V., Otterson, G.A., Blake, M., Kaye,

F.J., and Zajac-Kaye, M. (1995). The N-terminal domain of c-Myc associates

with alpha-tubulin and microtubules in vivo and in vitro. Mol. Cell. Biol. 15,

5188–5195.

Arthur, J.S., Elce, J.S., Hegadorn, C., Williams, K., and Greer, P.A. (2000).

Disruption of the murine calpain small subunit gene, Capn4: calpain is essen-

tial for embryonic development but not for cell growth and division. Mol. Cell.

Biol. 20, 4474–4481.

Bai, M.K., Costopoulos, J.S., Christoforidou, B.P., and Papadimitriou, C.S.

(1994). Immunohistochemical detection of the c-myc oncogene product

in normal, hyperplastic and carcinomatous endometrium. Oncology 51,

314–319.

Calcagno, D.Q., Guimaraes, A.C., Leal, M.F., Seabra, A.D., Khayat, A.S.,

Pontes, T.B., Assumpcao, P.P., De Arruda Cardoso Smith, M., and Burbano,

R.R. (2009). MYC insertions in diffuse-type gastric adenocarcinoma.

Anticancer Res. 29, 2479–2483.

Cartwright, P., McLean, C., Sheppard, A., Rivett, D., Jones, K., and Dalton, S.

(2005). LIF/STAT3 controls ES cell self-renewal and pluripotency by a Myc-

dependent mechanism. Development 132, 885–896.

Clement, C.A., Kristensen, S.G., Mollgard, K., Pazour, G.J., Yoder, B.K.,

Larsen, L.A., and Christensen, S.T. (2009). The primary cilium coordinates

early cardiogenesis and hedgehog signaling in cardiomyocyte differentiation.

J. Cell Sci. 122, 3070–3082.

Creppe, C., Malinouskaya, L., Volvert, M.L., Gillard, M., Close, P., Malaise, O.,

Laguesse, S., Cornez, I., Rahmouni, S., Ormenese, S., et al. (2009). Elongator

controls the migration and differentiation of cortical neurons through acetyla-

tion of alpha-tubulin. Cell 136, 551–564.

Dedieu, S., Poussard, S., Mazeres, G., Grise, F., Dargelos, E., Cottin, P., and

Brustis, J.J. (2004). Myoblast migration is regulated by calpain through its

involvement in cell attachment and cytoskeletal organization. Exp. Cell Res.

292, 187–200.

Eilers, M., and Eisenman, R.N. (2008). Myc’s broad reach. Genes Dev. 22,

2755–2766.

Faiola, F., Liu, X., Lo, S., Pan, S., Zhang, K., Lymar, E., Farina, A., and Martinez,

E. (2005). Dual regulation of c-Myc by p300 via acetylation-dependent control

of Myc protein turnover and coactivation of Myc-induced transcription. Mol.

Cell. Biol. 25, 10220–10234.

Frank, S.R., Parisi, T., Taubert, S., Fernandez, P., Fuchs, M., Chan, H.M.,

Livingston, D.M., and Amati, B. (2003). MYC recruits the TIP60 histone

acetyltransferase complex to chromatin. EMBO Rep. 4, 575–580.

Gandarillas, A., and Watt, F.M. (1997). c-Myc promotes differentiation of

human epidermal stem cells. Genes Dev. 11, 2869–2882.

Gebhardt, A., Frye, M., Herold, S., Benitah, S.A., Braun, K., Samans, B., Watt,

F.M., Elsasser, H.P., and Eilers, M. (2006). Myc regulates keratinocyte

adhesion and differentiation via complex formation with Miz1. J. Cell Biol.

172, 139–149.

Gonen, H., Shkedy, D., Barnoy, S., Kosower, N.S., and Ciechanover, A. (1997).

On the involvement of calpains in the degradation of the tumor suppressor

protein p53. FEBS Lett. 406, 17–22.

Gundersen, G.G., Khawaja, S., and Bulinski, J.C. (1989). Generation of

a stable, posttranslationally modified microtubule array is an early event in

myogenic differentiation. J. Cell Biol. 109, 2275–2288.

Habib, T., Park, H., Tsang, M., de Alboran, I.M., Nicks, A., Wilson, L.,

Knoepfler, P.S., Andrews, S., Rawlings, D.J., Eisenman, R.N., et al. (2007).

Myc stimulates B lymphocyte differentiation and amplifies calcium signaling.

J. Cell Biol. 179, 717–731.

Hanahan, D., and Weinberg, R.A. (2000). The hallmarks of cancer. Cell 100,

57–70.

Hann, S.R., King, M.W., Bentley, D.L., Anderson, C.W., and Eisenman, R.N.

(1988). A non-AUG translational initiation in c-myc exon 1 generates an

N-terminally distinct protein whose synthesis is disrupted in Burkitt’s

lymphomas. Cell 52, 185–195.

Hubbert, C., Guardiola, A., Shao, R., Kawaguchi, Y., Ito, A., Nixon, A., Yoshida,

M., Wang, X.F., and Yao, T.P. (2002). HDAC6 is a microtubule-associated

deacetylase. Nature 417, 455–458.

Iezzi, S., Di Padova, M., Serra, C., Caretti, G., Simone, C., Maklan, E., Minetti,

G., Zhao, P., Hoffman, E.P., Puri, P.L., et al. (2004). Deacetylase inhibitors

increase muscle cell size by promoting myoblast recruitment and fusion

through induction of follistatin. Dev. Cell 6, 673–684.

Ivanov, A.I., McCall, I.C., Babbin, B., Samarin, S.N., Nusrat, A., and Parkos,

C.A. (2006). Microtubules regulate disassembly of epithelial apical junctions.

BMC Cell Biol. 7, 12.

Kim, S.Y., Herbst, A., Tworkowski, K.A., Salghetti, S.E., and Tansey, W.P.

(2003). Skp2 regulates Myc protein stability and activity. Mol. Cell 11,

1177–1188.

Kleine-Kohlbrecher, D., Adhikary, S., and Eilers, M. (2006). Mechanisms of

transcriptional repression by Myc. Curr. Top. Microbiol. Immunol. 302, 51–62.

Koch, H.B., Zhang, R., Verdoodt, B., Bailey, A., Zhang, C.D., Yates, J.R., 3rd,

Menssen, A., and Hermeking, H. (2007). Large-scale identification of c-MYC-

associated proteins using a combined TAP/MudPIT approach. Cell Cycle 6,

205–217.

Kumar, A., Shafiq, S., Wadgaonkar, R., and Stracher, A. (1992). The effect of

protease inhibitors, leupeptin and E64d, on differentiation of C2C12 myoblasts

in tissue culture. Cell. Mol. Biol. 38, 477–483.

Mailhes, J.B., Hilliard, C., Lowery, M., and London, S.N. (2002). MG-132, an

inhibitor of proteasomes and calpains, induced inhibition of oocyte maturation

and aneuploidy in mouse oocytes. Cell Chromosome 1, 2.

Matsuyama, A., Shimazu, T., Sumida, Y., Saito, A., Yoshimatsu, Y.,

Seigneurin-Berny, D., Osada, H., Komatsu, Y., Nishino, N., Khochbin, S.,

et al. (2002). In vivo destabilization of dynamic microtubules by HDAC6-medi-

ated deacetylation. EMBO J. 21, 6820–6831.

McMahon, S.B., Van Buskirk, H.A., Dugan, K.A., Copeland, T.D., and Cole,

M.D. (1998). The novel ATM-related protein TRRAP is an essential cofactor

for the c-Myc and E2F oncoproteins. Cell 94, 363–374.


McMahon, S.B., Wood, M.A., and Cole, M.D. (2000). The essential cofactor

TRRAP recruits the histone acetyltransferase hGCN5 to c-Myc. Mol. Cell.

Biol. 20, 556–562.

Moyen, C., Goudenege, S., Poussard, S., Sassi, A.H., Brustis, J.J., and Cottin,

P. (2004). Involvement of micro-calpain (CAPN 1) in muscle cell differentiation.

Int. J. Biochem. Cell Biol. 36, 728–743.

Niklinski, J., Claassen, G., Meyers, C., Gregory, M.A., Allegra, C.J., Kaye, F.J.,

Hann, S.R., and Zajac-Kaye, M. (2000). Disruption of Myc-tubulin interaction

by hyperphosphorylation of c-Myc during mitosis or by constitutive hyper-

phosphorylation of mutant c-Myc in Burkitt’s lymphoma. Mol. Cell. Biol. 20,

5276–5284.

North, B.J., Marshall, B.L., Borra, M.T., Denu, J.M., and Verdin, E. (2003). The

human Sir2 ortholog, SIRT2, is an NAD+-dependent tubulin deacetylase. Mol.

Cell 11, 437–444.

Okano, H.J., Park, W.Y., Corradi, J.P., and Darnell, R.B. (1999). The cyto-

plasmic Purkinje onconeural antigen cdr2 down-regulates c-Myc function:

implications for neuronal and tumor cell survival. Genes Dev. 13, 2087–2097.

Pietilainen, T., Lipponen, P., Aaltomaa, S., Eskelinen, M., Kosma, V.M., and

Syrjanen, K. (1995). Expression of c-myc proteins in breast cancer as related

to established prognostic factors and survival. Anticancer Res. 15, 959–964.

Richard, I., Broux, O., Allamand, V., Fougerousse, F., Chiannilkulchai, N.,

Bourg, N., Brenguier, L., Devaud, C., Pasturaud, P., Roudaut, C., et al.

(1995). Mutations in the proteolytic enzyme calpain 3 cause limb-girdle

muscular dystrophy type 2A. Cell 81, 27–40.

Small, G.W., Chou, T.Y., Dang, C.V., and Orlowski, R.Z. (2002). Evidence for

involvement of calpain in c-Myc proteolysis in vivo. Arch. Biochem. Biophys.

400, 151–161.

Spotts, G.D., Patel, S.V., Xiao, Q., and Hann, S.R. (1997). Identification of

downstream-initiated c-Myc proteins which are dominant-negative inhibitors

of transactivation by full-length c-Myc proteins. Mol. Cell. Biol. 17, 1459–1468.

Sterner, D.E., and Berger, S.L. (2000). Acetylation of histones and transcrip-

tion-related factors. Microbiol. Mol. Biol. Rev. 64, 435–459.

Taira, T., Maeda, J., Onishi, T., Kitaura, H., Yoshida, S., Kato, H., Ikeda, M.,

Tamai, K., Iguchi-Ariga, S.M., and Ariga, H. (1998). AMY-1, a novel C-MYC

binding protein that stimulates transcription activity of C-MYC. Genes Cells

3, 549–565.

Takahashi, K., Tanabe, K., Ohnuki, M., Narita, M., Ichisaka, T., Tomoda, K.,

and Yamanaka, S. (2007). Induction of pluripotent stem cells from adult human

fibroblasts by defined factors. Cell 131, 861–872.

Tompa, P., Buzder-Lantos, P., Tantos, A., Farkas, A., Szilagyi, A., Banoczi, Z.,

Hudecz, F., and Friedrich, P. (2004). On the sequential determinants of calpain

cleavage. J. Biol. Chem. 279, 20775–20785.

Tonami, K., Kurihara, Y., Aburatani, H., Uchijima, Y., Asano, T., and Kurihara,

H. (2007). Calpain 6 is involved in microtubule stabilization and cytoskeletal

organization. Mol. Cell. Biol. 27, 2548–2561.

Tworkowski, K.A., Salghetti, S.E., and Tansey, W.P. (2002). Stable and

unstable pools of Myc protein exist in human cells. Oncogene 21, 8515–8520.

von der Lehr, N., Johansson, S., Wu, S., Bahram, F., Castell, A., Cetinkaya, C.,

Hydbring, P., Weidung, I., Nakayama, K., Nakayama, K.I., et al. (2003). The

F-box protein Skp2 participates in c-Myc proteosomal degradation and acts

as a cofactor for c-Myc-regulated transcription. Mol. Cell 11, 1189–1200.

Wakamatsu, Y., Watanabe, Y., Shimono, A., and Kondoh, H. (1993). Transition

of localization of the N-Myc protein from nucleus to cytoplasm in differentiating

neurons. Neuron 10, 1–9.

Wakamatsu, Y., Watanabe, Y., Nakamura, H., and Kondoh, H. (1997). Regula-

tion of the neural crest cell fate by N-myc: promotion of ventral migration and

neuronal differentiation. Development 124, 1953–1962.

Welcker, M., Orian, A., Jin, J., Grim, J.E., Harper, J.W., Eisenman, R.N., and

Clurman, B.E. (2004). The Fbw7 tumor suppressor regulates glycogen syn-

thase kinase 3 phosphorylation-dependent c-Myc protein degradation. Proc.

Natl. Acad. Sci. USA 101, 9085–9090.

Wilson, A., Murphy, M.J., Oskarsson, T., Kaloulis, K., Bettess, M.D., Oser,

G.M., Pasche, A.C., Knabenhans, C., Macdonald, H.R., and Trumpp, A.

(2004). c-Myc controls the balance between hematopoietic stem cell self-

renewal and differentiation. Genes Dev. 18, 2747–2763.

Xiao, Q., Claassen, G., Shi, J., Adachi, S., Sedivy, J., and Hann, S.R. (1998).

Transactivation-defective c-MycS retains the ability to regulate proliferation

and apoptosis. Genes Dev. 12, 3803–3808.

Yousefi, S., Perozzo, R., Schmid, I., Ziemiecki, A., Schaffner, T., Scapozza, L.,

Brunner, T., and Simon, H.U. (2006). Calpain-mediated cleavage of Atg5

switches autophagy to apoptosis. Nat. Cell Biol. 8, 1124–1132.


Erratum

SIRT1 Suppresses b-Amyloid Productionby Activating the a-Secretase Gene ADAM10Gizem Donmez, Diana Wang, Dena E. Cohen, and Leonard Guarente**Correspondence: [email protected]

DOI 10.1016/j.cell.2010.07.034

(Cell 142, 320–332; July 23, 2010)

As a result of an error during figure preparation, the model in Figure 6D was omitted. Additionally, in Figure S1B, Tau-pSer399 was

incorrectly typed. This has been corrected to Tau-pSer396 both in the figure and in the text. Corrected versions of both figures are

shown below.

These errors in no way affect the conclusions of the paper, and the authors apologize for any confusion caused by these errors.

The online version of the article has been corrected.

Figure 6. SIRT1 Deacetylates RARb


Figure S1. Related to Figures 1 and 2


Cell is seeking a scientist to join its editorial team. The minimum qualification for this position is a PhD in a relevant area of biomedical research, although additional post-doctoral or editorial experience is preferred. This is a superb opportunity for a talented individual to play a critical role in promoting science by helping researchers disseminate their findings to the wider community.

As an editor, you would be responsible for assessing submitted research papers and overseeing the refereeing process, and you would commission and edit material for Cell’s Leading Edge. You would also travel frequently to scientific conferences to follow developments in research and to establish and maintain close ties with the scientific community. You would have opportunities to pioneer and contribute to new trends in scientific publishing. The key qualities we look for are breadth of scientific interest and the ability to think critically about a wide range of scientific issues. The successful candidate will be highly motivated and creative and able to work inde-pendently as well as in a team.

This is a full-time, in-house editorial position, based at the Cell Press office in Kendall Square near MIT in Cambridge, Massachusetts. Cell Press offers an attractive salary and benefits pack-age and a stimulating working environment that encourages innovation.

To apply, please send your CV and a letter describing your qualifications, general research inter-ests, and motivation for pursuing a career in scientific publishing as soon as possible to our online jobs site: http://reedelsevier.taleo.net/careersection/51/jobdetail.ftl?lang=en&job=CEL0000R.

No phone inquiries, please. Cell Press in an equal opportunity/affirmative action employer, m/f/d/v.

Scientific Editor Position Available

editorad.indd 1 3/24/2010 2:17:01 PM

The American Society of Human Genetics is seeking an Editor for The American Journal of Human Genetics. The Editor leads one of the world’s oldest and most prestigious journals publishing pri-mary human genetics research.

Among the Editor’s responsibilities are determining the scope and direction of the scientific con-tent of The Journal, overseeing manuscripts submitted for review and their publication, selecting and supervising a staff consisting of an Editorial Assistant and doctoral-level Deputy Editor, direct-ing interactions with the publisher (currently Cell Press), reviewing quarterly reports provided by the publisher, evaluating the performance of the publisher, and if required, supervising the process of the selection a new publisher. The Editor serves as a member of the Board of Directors of the Ameri-can Society of Human Genetics (ASHG), as well as the ASHG Finance Committee, and presents semiannual reports to the Board. All Associate Editors of The Journal are appointed by the Editor, who also determines their duties. At the ASHG annual meeting, the Editor presides over a meeting of the Associate Editors and presents an annual report to the ASHG membership.

The term of the appointment is five years and includes a yearly stipend. The new Editor will be selected by the end of 2010 and will begin receiving manuscripts approximately in September 2011; there will be partial overlap with the Boston office. Applicants should be accomplished scientists in the field of human genetics and should have a broad knowledge and appreciation of the field. Nominations, as well as applications consisting of a letter of interest and curriculum vitae, should be sent to:

AJHG Editorial Search CommitteeAmerican Society of Human Genetics9650 Rockville PikeBethesda, MD 20814

The American Journal of Human Genetics Editor Position Available

editorad.indd 1 5/7/2010 12:25:11 PM

cell1423cla.indd 1cell1423cla.indd 1 7/30/2010 12:02:30 AM7/30/2010 12:02:30 AM

Positions Available

The University of Vermont College of Medicine is seeking applications from, or nominations of, outstanding candidates for the Chair of Biochemistry. This position is part of a strategic initiative in the College to build on areas of strength and increase interactions among the basic and clinical sciences. The College is particularly interested in candidates with substan-tive credentials in molecular cancer biology especially in areas related to environmental carcinogenesis. It is expected that candidates would have credentials appropriate for a tenured position. The Chair will be responsible for the development and oversight of the educational and research programs of the Department and have the opportunity and resources to implement a visionary approach to the discipline while promoting translational research. Current strengths of the Department are in Solution and Structural Studies of Macromolecules, and the Cell and Molecular Biology of cancer and of thrombosis.

The successful candidate will be an internationally recognized scientist with a Ph.D. or equivalent degree who has demonstrable skills as a leader, and substantial experience in the administrative, teaching and research activities of an academic department. The College seeks an innovative Chair who will position the Department as a national academic and research leader.

The University of Vermont is located in Burlington, a vibrant community with year-round recreational opportunities, safe communities and excellent schools. This progressive community has been frequently cited as one of the most livable cities in the U.S.

Women and those from diverse racial, ethnic and cultural backgrounds are encouraged to apply. Applications will be accepted until the position is filled. A curriculum vitae and letter of interest should be sent electronically to Linda Thatcher at the address below.

Susan S. Wallace, Ph.D.Chair, Biochemistry Search Committee

c/o Linda Thatcher ([email protected])E-126 Given Building

University of Vermont College of MedicineBurlington, VT 05405-0068

Or apply online at http://www.uvmjobs.com

The University of Vermont is an Equal Opportunity/Affirmative Action Employer.


Positions Available

Faculty PositionDepartment of Molecular Biology

Princeton University

The Molecular Biology Department at Princeton University invites appli-cations for a tenure-track faculty position at the Assistant Professor level in Developmental Biology. We are seeking an outstanding investigator using multicellular model systems to address fundamental questions in cell signaling, gene regulation, pattern formation, morphogenesis and/or pathogenesis. The University has a strong commitment to interdis-ciplinary studies, especially in the areas of systems biology, imaging, genomics, biophysics and neuroscience. The department has high level computing and microscope facilities, DNA array and high through-put sequencing technologies, mass spectrometry, modern transgenic mouse facilities and state of the art vivarium.

Applicants must have an excellent record of research productivity and demonstrate the ability to develop a rigorous research program. All applicants must have a Ph.D. or equivalent degree and a commitment to teaching at the undergraduate and graduate levels.

Applications must be submitted online at http://jobs.princeton.edu, requisition # 1000527, and should include a cover letter, curriculum vitae, a two-page research description, and contact information for three references. All materials must be submitted as PDF fi les. For full consideration, applications should be received by November 1, 2010.

Princeton University is an equal opportunity employer and complies with applicable EEO and affi rmative action regulations.

The Department of Microbiology at the University of Washington in Seattle is conducting a search for an Assistant or Associate Professor in the fi elds of bacterial-host interactions and/or bacterial physiology. We are looking for an innovative investigator who currently has or will develop an independent research program studying bacterial pathogenesis, bacterial interactions with the immune system, mutually benefi cial relationships between microbes and hosts, or bacterial physiology relevant to a variety of microbial systems. The position is a 12-month, full-time, tenure track position in the School of Medicine. In addition to research, the new faculty member will support the department’s teaching mission, including teaching at the undergraduate or graduate level. All University of Washington Faculty engage in teaching, research and service.

Salary and benefi ts are competitive and will be commensurate with the qualifi cations and experience of the applicant. The Uni-versity of Washington is an affi rmative action, equal opportunity employer and has built a culturally diverse faculty. Applications from female and minority candidates are especially encouraged. Expanding opportunities for women and other underrepresented groups in the faculty is an important goal of the department.

Applicants with a Ph.D., 1-2 years postdoctoral experience and a strong publication record should send their CV, a one or two page statement of research interests and the names and contact information for three references to:

Chair, Bacteriology Search CommitteeDepartment of Microbiology

Box 357242, Room K357B, University of Washington1705 N.E. Pacifi c Street, Seattle WA 98195

Application materials may also be sent by email c/o Bonnie Hightower at [email protected]. Applications should be submitted by September 15, 2010, but search will remain open until the position is fi lled.

Bacteriology Search:Tenure Track or Tenured Faculty

Department of MicrobiologyUniversity of Washington


Positions Available

FACULTY POSITION IN AGING AND AGE-RELATED DISEASE

Assistant Professor Level The Buck Institute invites applications for the Faculty position at the Assistant Professor level. The Buck Institute is the first independent research facility in the country focused solely on aging and age-related disease. The candidate cannot have held a previous faculty position. The successful candidate is expected to develop and maintain an extra-murally funded research program in basic mechanisms of aging or age-related disease. Applicants with expertise in molecular, cellular, animal models, imaging, and/or computational experimental approaches related to understanding the mechanisms of aging or disease in human health are encouraged to apply. Areas of interest include but are not limited to inflammation or metabolism. Candidates must have a Ph.D., M.D., or equivalent degree and at least two years of relevant postdoc-toral research experience. The position will be supported with five years’ salary support, startup funds and excellent laboratory space in a first class facility.

Please electronically submit curriculum vitae, a brief statement of current and future research interests, and contact information for three references to [email protected]. Applicants are encouraged to apply immediately for full consideration. The position could start immediately. This recruitment is through the NIH Recovery and Reinvestment Act.

years of leadership in human genetics research,

education and service.

1948–2008www.ashg.org

60

Look Again. Discover More.• Access to the 14 Cell Press

primary research journals and 14 Trends reviews title, all on the same platform

• Improved, more robust article and author search

• Easy to navigate home page, articles pages and archive

www.cell.com


careers.cell.com

Find Your Ideal Job!

See online version for legend and references.496 Cell 142, August 6, 2010 ©2010 Elsevier Inc. DOI 10.1016/j.cell.2010.07.035

SnapShot: Nonhomologous DNA End Joining (NHEJ)Michael R. Lieber1 and Thomas E. Wilson2

1University of Southern California, Los Angeles, CA 90089, USA and 2University of Michigan, Ann Arbor, MI 48109, USAMichael R. Lieber1 and Thomas E. Wilson2

1University of Southern California, Los Angeles, CA 90089, USA and 2University of Michigan, Ann Arbor, MI 48109, USA

lieber.indd 1 7/29/2010 1:24:59 PM

Webinar Part I: Assay development for broad kinome coverage, including di� cult targets

This introductory webinar will help you:

→ Understand the basic principles of assay design and where the LanthaScreen® Eu Kinase Binding Assay may fi t into your workfl ow

→ Discover how this assay platform can alleviate the assay development challenges of diffi cult kinase targets

→ Learn how easy this assay is to use for kinases with low purity, non-activated kinases, or kinases without a known substrate

Webinar Part II: Advanced applications of the LanthaScreen® Eu Kinase Binding Assay Platform

Learn how the LanthaScreen® Eu Kinase Binding Assay:

→ Can augment the characterization of compounds by simplifying on- and off -rate experiments

→ Can detect multiple classes of kinase inhibitors

→ Can be utilized for cell lysate–based applications

Each webinar series is being off ered in North American and European time zones.

View the dates and times of each webinar and register to attend at www.invitrogen.com/kbawebinar.

Webinar series: Applications of a kinase binding assay Solutions for advancing your workflow from challenging targets to downstream applications

Webinar Announcement

www.invitrogen.com

For research use only. Not intended for any animal or human therapeutic or diagnostic use, unless otherwise stated. © 2010 Life Technologies Corporation. All rights reserved. The trademarks mentioned herein are the property of Life Technologies Corporation or their respective owners. These products may be covered by one or more Limited Use Label Licenses (see Invitrogen catalog or www.invitrogen.com). By use of these products you accept the terms and conditions of all applicable Limited Use Label Licenses. CO21534 0710

Join us for this free two-part webinar series designed to highlight the versatility of the LanthaScreen® Eu Kinase

Binding Assay to help solve your assay development challenges and provide a simplifi ed solution for multiple

downstream applications.

MGC16169 (TBCK)

GUCY2D (CYGD)GUCY2F (CYGF)

NPR1 (ANPA)NPR2 (ANPB)

GUCY2C (HSER)

RNASELSTK31 (SGK396)

PIK3R4SCYL2LOC400687 (SGK424)

TEX14 (SGK307)PINK1

DKFZP761P0423 (SGK223)

KIAA2002 (SGK269)

EIF2AK4 (GCN2)EIF2AK2 (PKR)

EIF2AK3 (PEK)

EIF2AK1 (HRI)

BUB1B (BUBR1)

LOC91461 (SGK493)

PXK (SLOB)

TP53RK (PRPK)NRBP2

PACE-1 (SCYL3)

SCYL1

FLJ23356 (SGK196)

BUB1CDC7GSG2 (HASPIN)WNK3WNK1

WNK2

WNK4

NRBP (NRBP1)

SBK1 (SBK)LOC649023 (SGK110)

LOC390975 (LOC646643)

CHUK (IKK alpha)

IKBKB (IKK beta)

TBK1IKBKE (IKK epsilon)

ERN1 (IRE1)

ERN2 (IRE2)

PKMYT1 (MYT1)

WEE1WEE2 (WEE1B)

MOS

RIPK5 (SGK496)

PBK (TOPK)

AURKC (AURC)

AURKB (AURB)

AURKA (AURA)

PLK3 (FNK)

PLK2 (SNK)

PLK1

PLK4ULK1ULK2

ULK3

CAMKK2

CAMKK1CSNK2A1 (CK2 alpha 1)

CSNK2A2 (CK2 alpha 2)

TLK2TLK1

TTK

NEK2

NEK3NEK1

NEK5

NEK4

NEK11

NEK9NEK8

NEK6

NEK7

FLJ32685 (NEK10)

STK36 (FUSED)

ULK4

STK16 (MPSK1)GAK

AAK1BMP2K (BIKE)

STK35 (CLIK1)

PDIK1L (CLIK1L)

C9ORF96 (SGK071)

UHMK1 (KIS)

CSNK1G1 (CK1 gamm

a 1)

CSNK1G2 (CK1 gamm

a 2)

CSNK1G3 (CK1 gamm

a 3)

VRK3VRK1VRK2

TTBK1TTBK2

CSNK1A1 (CK1 alpha)CSNK1A1L (CK1 alpha 2)CSNK1D (CK1 delta)CSNK1E (CK1 epsi lon)

MAST

4MA

ST2

MAST

1MA

ST3

CDC4

2BPA

(MRC

K al

pha)

CDC4

2BPB

(MRC

K be

ta)CD

C42B

PG (D

MPK2

)DM

PK (D

MPK

1)ROCK

2RO

CK1ST

K38 (

NDR1

)

STK3

8L (N

DR2)

CIT (

CRIK

)LATS

1LA

TS2

STK3

2B (Y

ANK2

)

STK3

2A (Y

ANK1

)STK3

2C (Y

ANK3

)MAST

L

FLJ2

5006

(SGK

494)

GRK4

(GPR

K4)

GRK5

(GPR

K5)

GRK6

(GPR

K6)

GRK1

(RHO

K)

GRK7

(GPR

K7)

ADRB

K1 (G

RK2,

BARK

1)

ADRB

K2 (G

RK3,

BARK

2)

RPS6

KA4 (

MSK2

)

RPS6

KA5 (

MSK1

)

RPS6

KB2 (

P70S

6K b

eta)

RPS6

KB1 (

p70S

6K)

RPS6

KA1 (

RSk1

)

RPS6

KA3 (

RSK2

)

RPS6

KA2 (

RSK3

)RP

S6KA

6 (RS

K4)

AKT1

(PKB

alph

a)

AKT2

(PKB

bet

a)AKT3

(PKB

gamma)

SGK

SGK3

SGK2

PRKC

A (P

KC al

pha)

PRKC

B (PK

C bet

a)PRKC

G (PKC

gamma)

PRKC

E (PK

C eps

i lon)

PRKC

H (PKC

eta)PR

KCD (P

KC delt

a)

PRKC

Q (PKC

thet

a)PR

KCI (P

KC io

ta)

PRKC

Z (PKC

zeta)

PKN1

PKN2PK

N3

PDPK1 (PDK1)

PRKACA (PKA alpha)

PRKACB (PKA beta

)PRKACG (PKA gam

ma)PRKYPRKXPRKG1 (P

KG1)

PRKG2 (PKG2)

RPS6KC1 (RSKL1)

RPS6KL1 (RSKL2)

PDK2 (PDHK2)

ADCK1

ADCK2

ADCK4

CABCI (ADCK3)

ADCK5

ALPK2 (ALPHAK2)

ALPK3 (ALPHAK1)

FASTK

HSPB8 (H11)

TRPM6 (CHAK2)

TRPM7 (CHAK1)STK19 (G11)

BCR

BRD2BRD3

BRD4BRDT

TAF1L (TAF1_D2)TAF1

TWF2 (PTK9L)TWF1 (PTK9)

ALPK1 (ALPHAK3)TRIM28 (TIF1B)TRIM33 (TIF1G)

TRIM24 (TIF1A)

RIOK3RIOK1EEF2K

RIOK2

ATMATR

FRAP1 (mTOR)

PRKDC (DNA-PK)TRRAP

SGM1

BCKDK

PDK4 (PDHK4)

PDK1 (PDHK1)

PDK3 (PDHK3)

CCRK

CRK7CDC2L5 (CHED)

CDK9CDK10CDC2L2

CDC2L1 (PITSLRE)

CDK7

CDC2L6 (CDK11)CDK8ALS2CR7 (PFTAIRE2)PFTK1 (PFTAIRE1)PCTK3 (PCTAIRE3)PCTK1 (PCTAIRE1)PCTK2 (PCTAIRE2)

CDK5CDC2 (CDK1)CDK 2CDK3

CDK6CDK4

MAPK6 (ERK3)MAPK4 (ERK4)

ERK8 (ERK7)NLK

MAPK7 (ERK5)M A PK3 (E RK1)M A PK1 (E RK 2)

M A PK9 (JNK 2)

M A PK10 (JNK3)M A PK8 (JNK1)

M A PK12 (p38 gamma)

M A PK13 (p38 delta)

M A PK11 (p38 beta)

M A PK14 (p38 alpha) CDKL5

CDKL3CDKL2

CDKL1CDKL4

RAGE (MOK)

ICKMAK

GSK3A (GSK3 alpha)

GSK3B (GSK3 beta) PRPF4B (PRP4)

STK 23 (MSSK1)

SRPK 2

SRPK1CLK 2

CLK3

CLK4CLK1

DY RK1B

DY RK1A

DY RK4

DYRK2

DY RK3

HIPK4

HIPK3 (DY RK6)

HIPK1

HIPK2

PRKD2 (PKD2)

PRKD3 (EPK2)

PRKD1 (PKCM)CHEK 2 (CHK 2)

MAPK APK 2

MAPK APK3

MAPK APK5 (PR AK)

MKNK2 (MNK2)

MKNK1 (MNK1)

RPS6KA6_D2 (RSK4_D2)




RPS6KA5_D2 (MSK1_D2)

RPS6KA4_D2 (MSK2_D2)

DCAMKL3

DCAMKL2

DCAMKL1

PSKH2

PSKH1

CAMK V (VACAMKL)

CAMK4

CAMK1G (CAMKI gamma)

CAMK1D (CAMKI delta)

CAMK1 (CAMKI alpha)

PNCK (CAMKI beta)

CASK

CAMK2A (CAMKII a lpha)

CAMK2D (CAMKII delta)

CAMK2G (CAMKII gamma)

CAMK2B (CAMKII beta)

PHKG1 (PHK gamma 1)

PHKG2 (PHK gamma 2)

STK40 (SGK495)

TRIB1 (TRB1)

TRIB2 (TRB2) TRIB3 (TRB3)STK33HUNK

MGC42105 (NIM1) SNRKMELK

NUAK2 (SNARK)NUAK1 (ARK5)

PRKAA2 (AMPKA2)

PRKAA1 (AMPKA1)BRSK1

BRSK2

KIAA0999 (QSK)

SNF1LK (SIK)SNF1LK2 (QIK)

MARK4

MARK1MARK2

MARK3

PASK

PIM2

PIM1PIM3

CHEK1 (CHK1)

TSSK4

STK22C (TSSK3) SSTK

TSSK2 (STK22B)TSSK1 (STK22D)

STK11 (LKB1)

STK17A (DRAK1)STK17B (DRAK2) DAPK2

DAPK1DAPK3

TTN

MYLK (SMMLCK)

MYLK2 (SKMLCK)

MLCK (CAMLCK)LOC340156 (SGK085)

TRIOTRAD

OBSCN_D2APEG1_D2 (SPEG_D2) OBSCN

APEG1 (SPEG)

MAP3K7

M A P3K5 (ASK1)

MAP3K6

MAP3K4

YSK4 (FLJ23074)M A P3K3 (ME K K3)

M A P3K 2 (ME K K 2)

MAP3K1MAP3K14 (NIK)

M A P3K8 (COT)

E IF2 A K4 _ D2 (GCN2 _ D2)M A P2K1 (ME K1)M A P2K 2 (ME K 2)

M A P2K5M A P2K3 (ME K3)M A P2K6 (MK K6)

MAP2K4MAP2K7

LYK5 (STLK5)ALS2CR2OXSR1 (OSR1)STK39 (STLK3)PA K4PA K7PA K6

PA K1PA K3PA K 2TAOK 2 (TAO1)TAOK1 (MARKK, PSK2)

TAOK3 (J IK)

STK 24 (MST3)M ASK (MST4)STK 25 ( YSK1)STK3 (MST2)STK4 (MST1)

MINK1 (ZC3/MINK)M A P4K4 (ZC1/HGK)

TNIKNRK (ZC4)

MYO3AMYO3B

MAP4K3 (KHS2)MAP4K5 (KHS1)

MAP4K 2 (GCK)MAP4K1 (HPK1)

STK10 (LOK)SLK

ACVR2 (ACTR2)

ACVR2B (ACTR2B)

TGFBR2 (TGF beta R2)

AMHR2 (MISR2)

BMPR2

ACVRL1 (ALK1)BMPR1A

BMPR1BACVR1B (ALK4)

TGFBR1 (TGF beta R1)

ACVR1C (ALK7)

ILK

IR AK1

IR AK3

IRAK2

IR AK4

MLKL

RIPK1RIPK3

RIPK 2 (R ICK) RIPK4 (ANKRD3)

ANKK1 (SGK288)

LRRK2

LRRK1

AR AF BR AFR AF1 (cR AF)

KSR (KSR1)

KSR2

MAP3K12 (DLK)

MAP3K13 (LZK)MAP3K9 (MLK1)

M A P3K11 (MLK3)

KIAA1804 (MLK4)

M A P3K10 (MLK 2)

Z A K

M A P3K7 (TA K1)

TNNI3K (HH498)

LIMK 2LIMK1TESK1

TESK2

ACVR1 (ALK2)

ABL1

(ABL

)

ABL2

(ARG

)BM

XBT

KIT

KTE

CTX

KBL

KHC

KLY

N

LCK

FGR

FYN

SRC

PTK6

(BRK

)

YES1

(YES

)

CSK

MAT

K (H

YLTK

, CTK

)SRM

S (SR

M)

FRK

(RAK

, GTK

)

FER

(TYK

3)FE

S (FP

S)

FGFR

1FG

FR2

FGFR

3FG

FR4

RET

FLT1

(VEG

FR1)

KDR

(VEG

FR2)

FLT4

FLT3

(VEG

FR3)

K IT (

cKIT

)

CSF1

R (F

MS)

PDGF

RA (

PDGF

R al

pha)

PDGF

RB (

PDGF

R be

ta)

TIE1

TEK

(TIE

2)

ALK

LTK

IGF1

R

ROS1

(ROS

)

INSR

INSR

R (IR

R)

AXL

MERT

K (ME

R)

TYRO

3 (Dt

k)ME

T (cM

ET)

MST1

R (RO

N)RY

KTN

K2 (A

CK)

TNK1

JAK1

_D2

TYK2

_D2

JAK2

_D2

JAK3

_D2

STYK

1 (SU

RTK1

06) AA

TK (L

MR1)

LMTK

2 (LM

R2)

LMTK

3 (LM

R3)

PTK7

(CCK

4)

DDR1DDR2

MUSK

NTRK1

(TRK

A) NTRK2 (

TRKB)

NTRK3

(TRKC

)

ROR1

ROR2

EGFR (Erb

B1)ERBB2 (H

ER2)

ERBB4 (HER4)

ERBB3 (H

ER3) JA

K1TYK2

JAK2

JAK3

SYKZAP70

PTK2 (FAK)

PTK2B (PYK2)

EPHA1

EPHA2

EPHA3

EPHA4EPHA5

EPHA6

EPHA7

EPHB1

EPHB2

EPHB3

EPHB4

EPHA8

EPHA10

EPHB6

CAMK

CM

GC

ATYPICAL

AGCCK1

OTHER

STE

TKLTK

The Human Kinome

CO21534_KBA_webinar_Cell.indd 1 7/7/10 4:53 PM

B: 218 mm

CLIENT: Santa Cruz Biotechnology, Inc. PROJECT: 2010 ChemCruz Biochemicals Ad VERSION: Size “B” DATE: 04.08.10

PUBLICATION: CELL (MASTER/DIGITAL DOWNLOAD w/MATCH PROOF)

BLEED: 218mm W x 282mm H TRIM: 212mm W x 276mm H LIVE: 200mm W x 264mm H

SAN TA CRUZ B I OT ECH N O LO GY, I N C . SANTA CRUZ

B I O T E C H N O L O G Y

The Power to Question

®

Leading (#1 worldwide) supplier of specialty biochemicals for proteomics research.

ChemCruz™

HEADQUARTERS

TOLL FREE: 800.457.3801 TEL: 831.457.3800FAX: 831.457.3801 E-MAIL: [email protected]

EUROPEAN SUPPORT

TOLL FREE: +00800.4573.8000 PHONE: +49.6221.4503.0 FAX: +49.6221.4503.45 E-MAIL: [email protected]

ASIAN SUPPORT

JAPAN TOLL FREE: (010) 800.40402026 S. KOREA TOLL FREE: 00798.1.1.002.0297 N. CHINA TOLL FREE: 10.800.711.0752 S. CHINA TOLL FREE: 10.800.110.0694 FAX: 831.457.3801 E-MAIL: [email protected]

Products include over 71,300 unique biochemicals sorted alphabetically, by CAS number, by formula and on the basis of functional activity.

See our web site for product details and ordering information.

www.scbt.com

150 linescreenmp

Documents

Cell 100806