DOI: 10.1126/science.1218062, 1164 (2012);336 Science

et al.Dennis BreitsprecherDefined by Single-Molecule ImagingRocket Launcher Mechanism of Collaborative Actin Assembly

This copy is for your personal, non-commercial use only.

clicking here.colleagues, clients, or customers by , you can order high-quality copies for yourIf you wish to distribute this article to others

  here.following the guidelines

can be obtained byPermission to republish or repurpose articles or portions of articles

  ): June 5, 2012 www.sciencemag.org (this information is current as of

The following resources related to this article are available online at

http://www.sciencemag.org/content/336/6085/1164.full.htmlversion of this article at:

including high-resolution figures, can be found in the onlineUpdated information and services,

http://www.sciencemag.org/content/suppl/2012/05/30/336.6085.1164.DC1.html can be found at: Supporting Online Material

http://www.sciencemag.org/content/336/6085/1164.full.html#ref-list-1, 14 of which can be accessed free:cites 30 articlesThis article

http://www.sciencemag.org/cgi/collection/biochemBiochemistry

subject collections:This article appears in the following

registered trademark of AAAS. is aScience2012 by the American Association for the Advancement of Science; all rights reserved. The title

CopyrightAmerican Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005. (print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by theScience

on

June

5, 2

012

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

and rigidity to allow more efficient AtCERK1activation. The free hydroxyl groups at C4 andC1 (Fig. 2B) from the two ends of (NAG)4 arenot involved in AtCERK1-ECD binding, suggest-ing that chitin oligomers with more than eightunits can interact with more than two AtCERK1molecules simultaneously.Consistently,AtCERK1interacts strongly with polymeric chitin (17) be-cause of multiple AtCERK1-binding sites.

Like the mammalian receptor tyrosine ki-nases (RTKs) and toll-like receptors (TLRs), plantRLKs have emerged as critical regulators of keycellular processes. Structural analyses have dem-onstrated that ligand-induced dimerization is acommon theme for the activation of RTKs (27)and TLRs (28). The molecular mechanisms un-derlying RLK activation, however, remain muchless understood. Ligand-induced heterodimeriza-tion has been reported for several RLKs (29–32).Our results suggest that chitin-mediated cross-linking of AtCERK1s is required for immunesignaling, shedding light on a molecular mecha-nism of ligand-induced RLK activation.

References and Notes1. T. Boller, G. Felix, Annu. Rev. Plant Biol. 60, 379 (2009).2. C. Zipfel, Curr. Opin. Immunol. 20, 10 (2008).

3. J. E. Parker, Trends Plant Sci. 8, 245 (2003).4. S. H. Shiu et al., Plant Cell 16, 1220 (2004).5. A. Miya et al., Proc. Natl. Acad. Sci. U.S.A. 104, 19613

(2007).6. J. Wan et al., Plant Cell 20, 471 (2008).7. C. Zipfel et al., Cell 125, 749 (2006).8. C. Zipfel et al., Nature 428, 764 (2004).9. H. Kaku et al., Proc. Natl. Acad. Sci. U.S.A. 103, 11086

(2006).10. T. Boller, Annu. Rev. Plant. Physiol. Mol. Biol. 46, 189

(1995).11. G. Buist, A. Steen, J. Kok, O. P. Kuipers, Mol. Microbiol.

68, 838 (2008).12. S. Tanaka et al., BMC Plant Biol. 10, 288 (2010).13. J. Fliegmann et al., Plant Physiol. Biochem. 49, 709

(2011).14. R. Willmann et al., Proc. Natl. Acad. Sci. U.S.A. 108,

19824 (2011).15. S. Gimenez-Ibanez et al., Curr. Biol. 19, 423 (2009).16. S. Radutoiu et al., Nature 425, 585 (2003).17. E. K. Petutschnig, A. M. Jones, L. Serazetdinova, U. Lipka,

V. Lipka, J. Biol. Chem. 285, 28902 (2010).18. E. Iizasa, M. Mitsutomi, Y. Nagano, J. Biol. Chem. 285,

2996 (2010).19. K. M. Ramonell et al., Mol. Plant Pathol. 3, 301 (2002).20. N. Shibuya, E. Minami, Physiol. Mol. Plant Pathol. 59,

223 (2001).21. T. Ohnuma, S. Onaga, K. Murata, T. Taira, E. Katoh,

J. Biol. Chem. 283, 5178 (2008).22. L. P. Hamel, N. Beaudoin, Planta 232, 787 (2010).23. B. Zhang, K. Ramonell, S. Somerville, G. Stacey,

Mol. Plant Microbe Interact. 15, 963 (2002).24. B. Schulze et al., J. Biol. Chem. 285, 9444 (2010).

25. R. de Jonge et al., Science 329, 953 (2010).26. J. L. Asensio et al., Chem. Biol. 7, 529 (2000).27. M. A. Lemmon, J. Schlessinger, Cell 141, 1117 (2010).28. J. Y. Kang, J. O. Lee, Annu. Rev. Biochem. 80, 917

(2011).29. D. Chinchilla et al., Nature 448, 497 (2007).30. A. Heese et al., Proc. Natl. Acad. Sci. U.S.A. 104, 12217

(2007).31. J. Li et al., Cell 110, 213 (2002).32. K. H. Nam, J. Li, Cell 110, 203 (2002).

Acknowledgments: We thank F. Yu and J. He at ShanghaiSynchrotron Radiation Facility (SSRF). J.C. was funded byState Key Program of National Natural Science of China(31130063) and Chinese Ministry of Science and Technology(2010CB835300). J.M.Z. was funded by Chinese Ministry ofScience and Technology (2011CB100700; 2010CB835300).J.B.C. was funded by the National Outstanding Young ScholarScience Foundation of National Natural Science Foundationof China (30825043). The coordinates and structural factorsfor free- and chitin-bound AtCERK1-ECD have been depositedin Protein Data Bank with accession codes 4EBY and 4EBZ,respectively.

Supplementary Materialswww.sciencemag.org/cgi/content/full/336/6085/1160/DC1Materials and MethodsTable S1Figs. S1 to S11References (33–40)

9 January 2012; accepted 23 April 201210.1126/science.1218867

Rocket Launcher Mechanism ofCollaborative Actin Assembly Definedby Single-Molecule ImagingDennis Breitsprecher,1 Richa Jaiswal,1 Jeffrey P. Bombardier,2 Christopher J. Gould,1

Jeff Gelles,2* Bruce L. Goode1*

Interacting sets of actin assembly factors work together in cells, but the underlying mechanismshave remained obscure. We used triple-color single-molecule fluorescence microscopy to image thetumor suppressor adenomatous polyposis coli (APC) and the formin mDia1 during filamentassembly. Complexes consisting of APC, mDia1, and actin monomers initiated actin filamentformation, overcoming inhibition by capping protein and profilin. Upon filament polymerization,the complexes separated, with mDia1 moving processively on growing barbed ends while APC remainedat the site of nucleation. Thus, the two assembly factors directly interact to initiate filament assemblyand then separate but retain independent associations with either end of the growing filament.

Regulation of actin assembly is a fundamen-tal requirement in all eukaryotic cells,and a growing number of factors have

been identified that either inhibit or promotethis process. For example, the combined pres-ence of the actin monomer–binding proteinprofilin and the filament end–binding cappingprotein (CapZ) strongly suppresses both sponta-neous filament nucleation and elongation. Thus,filament assembly in vivo requires nucleationand elongation factors to overcome these barriers

to assembly (1–3). The formation ofmost cellularactin structures depends on two or more suchfactors, which often interact directly. A formin isa component of many actin assembly–promotingfactor (APF) pairs that likely function togetherin vivo: the formins FMN/Cappucino (Capu)and Spire (4), the formin Bni1 and Bud6 (5),the formin mDia1 and adenomatous polypo-sis coli (APC) (6), and the formin dDia2 andDictyostelium vasodilator-stimulated phospho-protein (DdVASP) (7).

The dimeric formin–homology 2 (FH2) do-main of formins processively tracks the growingbarbed end of the actin filament, protecting itfrom capping proteins (8–10). Adjacent FH1domains recruit profilin-actin complexes and canincrease the rate of elongation at barbed ends

(11). Whereas profilin enhances formin-mediatedfilament elongation, its presence also stronglysuppresses filament nucleation by formins (12).Collaboration of formins with other APFs thatbind multiple actin monomers (4–6, 13) maybe required to overcome the inhibitory effectsof profilin and capping protein. However, directevidence for this hypothesis has been lacking.To address this, we reconstituted mDia1-APC–mediated actin assembly with purified, fluores-cently tagged proteins and used multiwavelengthsingle-molecule TIRFM (total internal reflectionfluorescencemicroscopy) to directly visualize anddefine the mechanisms promoting collaborativefilament assembly.

For single-molecule imaging, we purified asoluble, modified O6-alkylguanine-DNA alkyl-transferase (AGT)-tagged (also known asSNAP–tag) C-terminal fragment of APC (APC-C):residues 2130–2843, encompassing its “Basic”domain, which is sufficient to mediate actinnucleation, and the domain that binds EB1 (amicrotubule end–binding protein) (6, 14). SNAP–APC-C labeled with SNAP-surface-647 (hereinnamed SNAP-647-APC-C) displayed activitiesidentical to those of APC-C in pyrene-actin as-sembly assays (fig. S1A). Photobleaching datasuggested that most SNAP-647-APC-C mole-cules are dimeric (fig. S1, B and C; and movieS1), consistent with hydrodynamic studies onmaltose-binding protein–tagged APC-C (6).

APC, like Spire and Bud6, has been proposedto catalyze actin nucleation by binding actinmonomers to form an F-actin seed (4–6). Weused dual-color TIRFM to directly visualizesurface-adsorbed SNAP-647-APC-C molecules,appearing as discrete spots, during the assembly

1Department of Biology, Brandeis University, Waltham, MA02454, USA. 2Department of Biochemistry, Brandeis Uni-versity, Waltham, MA 02454, USA.

*To whom correspondence should be addressed. E-mail:goode@brandeis.edu (B.L.G.); gelles@brandeis.edu (J.G.)

1 JUNE 2012 VOL 336 SCIENCE www.sciencemag.org1164

REPORTS

on

June

5, 2

012

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

of Oregon Green (OG)–actin filaments (Fig. 1Aand movie S2). From some of these spots, weobserved actin filaments emerge and grow pri-marily from their barbed ends, and APC did notalter the growth rate in the presence or absence ofprofilin (fig. S2A). Further, APC-C did not blockpolymerization at pointed ends (fig. S2B), whichsuggests that it stays bound to the filament at thesite of nucleation.

For the majority of surface-tethered filaments(88 out of 106 observed), APC fluorescence re-mained visible at the nucleation site 10 min afternucleation, which suggested a thermodynamical-ly stable association. We also observed filamentswith APC-C associated at their ends freely dif-fusing near the surface, which suggested that thestable association is not due to surface tethering.At concentrations ≥10 nM SNAP-647-APC-C,we occasionally observedAPC-C accumulating onfilament sides, which led to bundling (fig. S3 andmovie S3), consistent with previous reports (15).

Quantifying the oligomeric state of APC-C atthe onset of nucleation suggested that an APC-Cdimer was sufficient to nucleate filament as-sembly, although a minority of the filamentsoriginated from spots consisting of more than

two APC-C subunits (Fig. 1B). When we in-cluded latrunculin B (latB), which associateswith actin monomers and blocks polymeriza-tion, OG-actin still accumulated in some of theSNAP-647-APC-C spots; however, no filamentspolymerized (Fig. 1C and movie S4). Theseobservations suggest that nucleation by APC in-volves monomer recruitment, as opposed to cap-ture of spontaneously formed F-actin intermediates.

We next purified SNAP-tagged mDia1-C—which contains FH1, FH2, and diaphanous au-toregulatory (DAD) domains—and fluorescentlylabeled it (SNAP-549-mDia1-C). Single-moleculephotobleaching experiments suggested that theSNAP-549-mDia1-C preparation is predominant-ly dimeric but also contains higher-order oligo-mers (fig. S4, A and B). SNAP-549-mDia1-C andunlabeled mDia1-C stimulated indistinguish-able rates of pyrene-actin assembly (fig. S4C).Using dual-color TIRFM, we observed singleSNAP-549-mDia1-C molecules translocatingwith the growing barbed ends of filaments bothin the presence and absence of profilin (Fig. 1, Dand E; fig. S5; and movies S5, S6, and S7).Processive association of the formin was verifiedby additional lines of evidence (figs. S5 and S6

andmovie S7).mDia1-C andSNAP-549-mDia1-Ceach elongated filaments at equivalent rates inpresence and absence of profilin (Fig. 1F). Sim-ilar results were obtained with SNAP-549-mDia1-C in the presence of 1 nM CapZ. Ourobservations directly confirm that individualformin molecules processively track filamentbarbed ends,with orwithout profilin, and accelerateelongation in a profilin-dependent manner (11).

APC binds mDia formins in solution (16).Therefore, we used dual-color TIRFM to inves-tigate interactions between SNAP-647-APC-CandSNAP-549-mDia1-Cmolecules.Aftermixingthe two labeled proteins in solution, we examinedthe composition of complexes that adsorbed tothe surface. We found that 51 T 2.3% (SEM) ofSNAP-549-mDia-C spots colocalized withSNAP-647-APC-C spots (Fig. 2A). Coincidentalcolocalization was minimal (1.7%) (14). Takinginto account the labeling stoichiometries of bothproteins, our analysis is consistent with a modelin which most of the fluorescent spots containtwo SNAP-647-APC-C and two SNAP-549-mDia1-C molecules (Fig. 2, A and B). Similarresults were obtained in the presence of latB-sequestered actin monomers [60 T 3.9% (SEM)

Fig. 1. Single-molecule analysis of the APC-C– and mDia1-C–mediated actinassembly processes. (A) Dual-color TIRFM of OG-actin filaments (1 mM, 10%labeled, green) originating from SNAP-647-APC-C spots (10 nM, red). Greenarrowheads mark the growing barbed ends. (B) Fluorescence intensity analysisof number of SNAP-647-APC-C subunits at filament ends. (C) Micrographs ofOG-actin accumulation at SNAP-647-APC-C spots in the absence and presence

of latB. (D and E) Dual-color TIRFM of OG-actin filaments (1 mM, 10%labeled, green) being elongated by single SNAP-549-mDia1-C molecules (red)without (D) and with (E) profilin. Time is given in seconds. Red arrowheads in(E) mark the formin molecule associated with the barbed end. (F) Averageelongation rates of mDia1-C– and SNAP-549-mDia1-C–assembled filamentswithout and with profilin. Error bars represent SE, n ≥ 12, N = 3.

www.sciencemag.org SCIENCE VOL 336 1 JUNE 2012 1165

REPORTS

on

June

5, 2

012

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

colocalization] (Fig. 2, C and D). These datasuggest that actin monomers neither interferewith APC-formin interactions nor change theoligomeric state of the complex and may evenmodestly promote APC-formin interactions. Usingtriple-color TIRFM, we observed 54.8 T 6.6%(SEM) of APC/mDia1 complexes stably occu-pied by latB–OG-actin (Fig. 2, E and F). Theseassemblies may represent nucleation intermedi-ates that are normally short-lived during filamentassembly but can be trapped by latB. In theabsence of mDia1, OG-actin accumulation wasobserved in 40.5 T 4.6% (SEM) of SNAP-647-APC-C spots, but in the absence of APC-C,OG-actin accumulation was observed in only4.0 T 0.7% (SEM) of SNAP-549-mDia1-Cspots (Fig. 2F). These observations suggest thatAPC-C is the primary actinmonomer–recruitingfactor in the APC-mDia1 complex. Profilin didnot change OG-actin accumulation in SNAP-647-APC-C spots but resulted in an increasein OG-actin accumulation to 20.2 T 1.8% (SEM)in SNAP-549-mDia1 spots, which likely repre-sents profilin-actin recruitment by FH1 domains.Consistent with this view, competition with a50-fold excess of profilin over OG-actin reducedthe number of formin spots, with OG-actin accu-mulation to 8.5 T 0.9% (SEM).

Next, we used triple-color TIRFM to definethe sequence of events and spatial locations ofSNAP-549-mDia1-C and SNAP-647-APC-Cduring collaborative filament assembly. We in-cluded CapZ and profilin to reconstitute cellularbarriers to actin assembly. Tripartite complexesconsisting of SNAP-549-mDia1-C, SNAP-647-APC-C, and OG-actin were observed (asterisk inFig. 3A). Filaments grew from some of thesespots. Concomitant with filament growth, APCand mDia1 invariably separated (n > 100), whichleft SNAP-647-APC-C at the pointed end andSNAP-549-mDia1-C translocating on the grow-ing barbed end (Fig. 3A, fig. S8, and movie S8).The same was observed in the absence of profilin(fig. S9 and movie S9). Most nucleation com-plexes had already separated by the start ofimaging, as indicated by the presence of manyshort filaments carrying SNAP-549-mDia1-C attheir barbed ends and SNAP-647-APC-C at theirpointed ends (fig. S10). Note that SNAP-647-APC-C molecules were never seen bound toprocessively moving formins (n > 100), whichsuggested that mDia1 molecules in the process ofcatalyzing barbed-end polymerization may notbe capable of interacting with APC-C. Filamentsbearing APC-C on one end and mDia1 on theother end showed rates of elongation indistin-

guishable from rates observed for mDia1-Cwithout APC-C (fig. S11). Thus, elongation offilaments is likely to be purely mDia1-catalyzedwith no contribution from APC.

Increasing concentrations of APC-C producedincreasing numbers of SNAP-549-mDia1-C–elongated filaments in the presence of profilinand CapZ (Fig. 3B and fig. S12A). Quantitative-ly similar results were obtained using unlabeledAPC-C and mDia1-C (Fig. 3C). Moreover, lon-ger preincubation of mDia1-C and APC-C in-creased the number of mDia1-C–elongatedfilaments by about another twofold (fig. S12B).These effects of APC-Cwere less pronounced inthe absence of profilin (Fig. 3B), which suggestedthat APC-C plays a key role in overcoming thenucleation barrier imposed by profilin.

Two other APFs (Spire and Bud6) wererecently shown to bind to formin C-terminal tailsequences (4, 5, 13, 17). Similarly, we found thatAPC-C–coated beads depleted soluble mDia1-Cbut notmDia1-FH1FH2 (which lacks tail sequences)from supernatants (Fig. 3D), which suggests thatAPCbindsmDia1 tail sequences.Moreover,APC-Cfailed to increase actin assembly activity in com-bination with mDia1-FH1FH2 (red bars, Fig. 3Cand fig. S13). These observations strongly suggestthat direct interactions between APC and mDia1

Fig. 2. Association of APC-C and mDia1-C and formation of tripartite com-plexes with G-actin. (A) Colocalization of SNAP-647-APC-C and SNAP-549-mDia1-C in surface-adsorbed complexes. The proteins (200 nM each) werepreincubated for 10 min, diluted 200-fold, and visualized by TIRFM. Spotsemitting fluorescence from SNAP-549, SNAP-647, or both were counted. (B)Oligomeric state of APC-C–mDia1-C complexes (n = 91) determined byfluorescence intensity analysis for mDia1-C and photobleaching step analysisfor APC-C (see methods). (C and D) Same as (A) and (B), except that 20 mM

latB-sequestered actin monomers were included during preincubation (100 nMin final reaction). In (D), n = 43. Error bars represent SE. (E) Tripartite com-plexes formed by preincubating 200 nM SNAP-647-APC-C, 200 nM SNAP-549-mDia1-C, or 8 mM latB–OG-actin and then diluting 200-fold before surfaceadsorption. Field-of-view (top) and time-lapse record of a single spotcontaining all three proteins (bottom). (F) Fraction of APC-mDia1 spots thatcontain detectable OG-actin in the presence or absence of profilin (see thesupplementary materials).

1 JUNE 2012 VOL 336 SCIENCE www.sciencemag.org1166

REPORTS

on

June

5, 2

012

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

tail sequences are required for their collaborativeeffects in actin assembly.

Taken together, our results suggest a mecha-nism (Fig. 3E) in which APC-C dimers recruitactin monomers and bind the tail region ofmDia1-C to form a tripartite nucleation complex.At the onset of actin polymerization, the complexseparates, leaving APC-C stably associated nearthe pointed end, while mDia1-C is propelled awayon the rapidly growing barbed end (Fig. 3E). Bythis mechanism, APC plays the central role inassembling the nucleation seed, and the formin

then detaches and serves as the elongation cat-alyst. It remains to be seen whether the samemechanism applies to other pairs of APFs that areknown to interact, e.g., Spire with Capu and Bud6with Bni1 (4, 5). This work may have importantimplications for APC biological function. Full-lengthAPC is a large protein that binds to numerousother cellular factors and functions in a wide varie-ty of regulatory pathways (18). Thus, the mech-anism forAPC-mDia1 collaboration described heremay result in site-specific actin assembly and,possibly, tethering of polymerizing actin filaments

at their pointed ends to APC-binding partners(e.g., IQGAP1, microtubule plus ends) (19, 20).These phenomena may serve to promote directedcell migration or other functions that require co-ordinated reorganization of the actin cytoskeletonwith respect to other cellular structures.

References and Notes1. T. D. Pollard, J. A. Cooper, Science 326, 1208 (2009).2. M. A. Chesarone, B. L. Goode, Curr. Opin. Cell Biol. 21,

28 (2009).3. K. G. Campellone, M. D. Welch, Nat. Rev. Mol. Cell Biol.

11, 237 (2010).

Fig. 3. Single-molecule visualization of APC-C and mDia1-C collaborating to assemble actinfilaments. (A) Triple-color TIRFM of the assembly of 1 mMOG-actin (green) in the presence of 50 pMSNAP-549-mDia1-C (red), 5 nM SNAP-647-APC-C (blue), 1 nM CapZ, and 5 mM profilin. Asteriskindicates APC-mDia1-actin nucleation complex. Images (left; time in seconds) and correspondingprofiles of fluorescence intensity along the length of the filament (right). Arrowheads mark thebarbed end (green), mDia1 (red), and APC (blue). (B) (Left) Density of formin-elongatedfilaments per 100- × 100-mm area for reactions containing 80 pM SNAP-549-mDia1-C, 1 nMCapZ, or variable concentrations of SNAP-APC-C, T5 mM profilin. Error bars represent SD, N = 3.(Right) APC-C–dependent fold increase in density of formin-elongated filaments in the presence(black) and absence (gray) of profilin. (C) Same as (B), except using non–SNAP-tagged APC-C,mDia1-C and tail-less mDia1 FH1FH2. (D) Western blot with antibody against the six-histidine tag (6His) showing levels of mDia1-C and mDia1 FH1FH2 (bothtagged with 6His) in supernatants after depletion by glutathione S-transferase (GST)–tagged (control) or GST–APC-C beads. Error bars represent SD, N = 2. (E)Proposed “rocket launcher” model for mDia1-APC collaboration during actin filament assembly (details in text).

www.sciencemag.org SCIENCE VOL 336 1 JUNE 2012 1167

REPORTS

on

June

5, 2

012

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om

4. M. E. Quinlan, S. Hilgert, A. Bedrossian, R. D. Mullins,E. Kerkhoff, J. Cell Biol. 179, 117 (2007).

5. B. R. Graziano et al., Mol. Biol. Cell 22, 4016 (2011).6. K. Okada et al., J. Cell Biol. 189, 1087 (2010).7. A. Schirenbeck et al., Proc. Natl. Acad. Sci. U.S.A. 103,

7694 (2006).8. D. Pruyne et al., Science 297, 612 (2002).9. H. Mizuno et al., Science 331, 80 (2011).

10. S. H. Zigmond et al., Curr. Biol. 13, 1820 (2003).11. D. R. Kovar, E. S. Harris, R. Mahaffy, H. N. Higgs,

T. D. Pollard, Cell 124, 423 (2006).12. M. A. Chesarone, A. G. DuPage, B. L. Goode, Nat. Rev.

Mol. Cell Biol. 11, 62 (2010).13. C. L. Vizcarra et al., Proc. Natl. Acad. Sci. U.S.A. 108,

11884 (2011).

14. Materials and methods are available as supplementarymaterials on Science Online.

15. J. B. Moseley et al., J. Biol. Chem. 282, 12661 (2007).16. Y. Wen et al., Nat. Cell Biol. 6, 820 (2004).17. K. Zeth et al., J. Biol. Chem. 286, 30732 (2011).18. B. M. McCartney, I. S. Näthke, Curr. Opin. Cell Biol. 20,

186 (2008).19. T. Watanabe et al., Dev. Cell 7, 871 (2004).20. Y. Mimori-Kiyosue, N. Shiina, S. Tsukita, J. Cell Biol. 148,

505 (2000).

Acknowledgments: We thank B. McCartney, H. Higgs,J. Moseley, S. Reck-Petersen, and B. Smith for helpfulcomments on the manuscript; K. Okada and A. Deaconescufor purifying CapZ and generating the SNAP-APC-C plasmid;

and B. Smith for advice on single-molecule imaging.The project was supported by grants from DeutscheForschungsgemeinschaft (BR 4116/1-1 to D.B.), NSF(DMR-MRSEC-0820492 to J.G and B.G), and NIH (GM43369and GM81648 to J.G., GM083137 to B.G.).

Supplementary Materialswww.sciencemag.org/cgi/content/full/336/6085/1164/DC1Materials and MethodsFigs. S1 to S13References (21–31)Movies S1 to S9

19 December 2011; accepted 11 April 201210.1126/science.1218062

The Amyloid Precursor Protein Has aFlexible Transmembrane Domainand Binds CholesterolPaul J. Barrett,1* Yuanli Song,1* Wade D. Van Horn,1 Eric J. Hustedt,2 Johanna M. Schafer,1

Arina Hadziselimovic,1 Andrew J. Beel,1 Charles R. Sanders1†

C99 is the transmembrane carboxyl-terminal domain of the amyloid precursor protein that iscleaved by g-secretase to release the amyloid-b polypeptides, which are associated with Alzheimer’sdisease. Nuclear magnetic resonance and electron paramagnetic resonance spectroscopy show thatthe extracellular amino terminus of C99 includes a surface-embedded “N-helix” followed by a short“N-loop” connecting to the transmembrane domain (TMD). The TMD is a flexibly curved a helix,making it well suited for processive cleavage by g-secretase. Titration of C99 reveals a binding sitefor cholesterol, providing mechanistic insight into how cholesterol promotes amyloidogenesis.Membrane-buried GXXXG motifs (G, Gly; X, any amino acid), which have an established role inoligomerization, were also shown to play a key role in cholesterol binding. The structure andcholesterol binding properties of C99 may aid in the design of Alzheimer’s therapeutics.

Alzheimer’s disease (AD) currently afflictsmore than 20 million people worldwide(1). The production and subsequent ag-

gregation of amyloid-b (Ab) peptides are widelythought to play a central role in most forms ofAD (1, 2); accordingly, factors that increase Abproduction and oligomerization or that reduce itselimination increase the risk of AD. Elevatedneuronal cholesterol levels increase the genera-tion of Ab (3–5), but the underlying mechanismshave yet to be elucidated.

Amyloidogenic cleavage of full-length amy-loid precursor protein (APP) by b-secretase gen-erates the transmembrane protein C99 (APP672-770,also known as b-CTF, fig. S1). The transmembranedomain (TMD) of C99 is then processively butimprecisely cleaved by g-secretase to release theAb polypeptides (6, 7). Structural information onC99 may provide insight into amyloidogenesis;however, previous structural studies have em-ployed only low-resolution methods or have fo-cused on TMD-containing fragments (8–12).

The backbone structure of C99 in lyso-myristoylphosphatidylglycerol (LMPG) detergentmicelles was determined by nuclear magneticresonance (NMR) restraints that included re-sidual dipolar couplings and distances derivedfrom paramagnetic relaxation enhancement ex-periments (Fig. 1, figs. S2 and S3, and tableS1). To prevent dimerization (11–14), we usedan 800:1 LMPG-to-C99 molar ratio (see supple-mental materials and methods). C99 is composedof three helical domains. A short extracellular“N-helix”(residues 688 to 694) is connected byan interfacial “N-loop” (695 to 699) to the helicalTMD (700 to 723). The N-helix is embedded inthe membrane surface and dynamically samplesa range of orientations around the TMD helixaxis (Fig. 1B). A third helix at the C terminus(residues 762 to 770) is surface-associated butis structurally uncoupled from the TMD by theintervening 38-residue “C-loop” (734 to 761).Power saturation electron paramagnetic resonance(EPR) measurements for spin-labeled C99 (Fig. 2and fig. S4) confirm that the span of the TMD isthe same in both micelles and lipid vesicles andthat the N- and C-helices are surface-associated.

The NMR structure reveals that the TM helixof C99 is highly curved, with the apex of cur-vature being located near glycine residues 708and 709, close to the center of the micelle (Fig.1). Pulsed EPR double electron-electron reso-

nance (DEER) experiments confirmed that theTMD curvature also occurs in lipid bilayers (Fig.2 and fig. S5). The measured end-to-end averagedistance in micelles (34.2 T 1.1 Å) is consistentwith the NMR structure and nearly identical tothe distance measured in lipid vesicles (33.5 T1.0 Å). With the use of EPR, we found that mu-tation of G708G709 to L708L709 (G, Gly; L, Leu)only modestly straightens the helix (average dis-tance now 35.3 T 0.5 Å), suggesting that thecurvature of the TMD derives only partially fromthe glycine pair (Fig. 2C). However, the rangesof distances sampled around the mean by theGly708→Leu708 (G708L) and G708L/G709Lmu-tants are dramatically reduced compared withthose of the wild type (Fig. 2C). This suggeststhat Gly708 confers flexibility to the TMD, aspredicted by molecular dynamics simulations (15).The flexibly curved nature of the TMD may bewell suited for its interactions with g-secretase.Medium-resolution electron microscopy structuresof the fully assembled and activated g-secretase(16,17) suggest a sluice-like active site that mightbest accommodate a substrate with a curved TMD(see docking model in fig. S6). Analogous activesites have been observed for other intramem-brane proteases (18, 19). TMD flexibility mayalso promote the processive cleavage of C99 byg-secretase, with flexibility colluding with ran-dom thermal motion to allow the TMD to slidethrough the active-site channel. Curvature mayalso play a role in exposing scissile bonds for pro-teolytic access.

The surface-associated N-helix and N-loopimmediately following the cleavage site (K687; K,Lys) for nonamyloidogenic a-secretase processingcontain a number of familial AD mutation sites(20) and, in conjunction with the extracellularend of the TMD, appear to play crucial roles indetermining the ratio of short versus the moretoxic long forms of the Ab peptides released byg-secretase (13, 14). A space-filling surface rep-resentation (Fig. 1C) of C99 suggests that thesesegments may also partially occlude approach byanotherTMD to the glycine zipperGXXXGXXXG(X, any amino acid) sequence located on the extra-cellular end of the TMD, which helps to explain theweakness of C99 self-association.

The organization of the N-loop and N-helixwith respect to the membrane surface and theTMD is consistent with the possibility of a lipid

1Department of Biochemistry, Center for Structural Biology andInstitute of Chemical Biology, Vanderbilt University School ofMedicine, Nashville, TN 37232 USA. 2Department of MolecularPhysiology and Biophysics, Vanderbilt University School ofMedicine, Nashville, TN 37232 USA.

*These authors contributed equally to this work.†To whom correspondence should be addressed. E-mail:chuck.sanders@vanderbilt.edu

1 JUNE 2012 VOL 336 SCIENCE www.sciencemag.org1168

REPORTS

on

June

5, 2

012

ww

w.s

cien

cem

ag.o

rgD

ownl

oade

d fr

om