Izasa.App-notes-Acea.Migracion-invasion

xCELLigence RTCA DP InstrumentFlexible Real-Time Cell Monitoring

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

Migración�e�Invasión��

RTCA Control Unit RTCA DP Analyzer

The RTCA DP Instrument expands the throughput and application options of the xCELLigence

Real-Time Cell Analyzer (RTCA) portfolio. Featuring a dual-plate (DP) format, the instrument

measures impedance-based signals in both cellular and cell invasion/migration (CIM)

assays – without the use of exogenous labels. With outstanding application flexibility, the RTCA

DP Instrument supports multiple users performing short-term and long-term experiments.

The xCELLigence RTCA DP Instrument Flexible Real-Time Cell Monitoring

Explore the wide range of applications

� Cell invasion and migration assays

� Compound- and cell-mediated cytotoxicity

� Cell adhesion and cell spreading

� Cell proliferation and cell differentiation

� Receptor-mediated signaling

� Virus-mediated cytopathogenicity

� Continuous quality control of cells

Figure 1: Reveal cytotoxic effects through continuous monitoring. HT1080 cells were seeded in an E-Plate at two different densities (5,000 and 10,000 cells) and treated 24 hours later with 12.5 nM Paclitaxel, or DMSO as a control. As shown by the Cell Index profile, which reflects cell adherence, the antimitotic effect of Paclitaxel was observed in HT1080 cells that were proliferating, whereas confluent cells showed no response.

Cytotoxicity Analysis in E-Plates

The xCELLigence System continuously and non-invasively detects cell responses throughout an experiment, without the use of exogenous labels that can disrupt the natural cell environment.

� Obtain complete, continuous data profiles from cell responses generated during in vitro experiments (Figure 1).

� Take advantage of real-time data to identify optimal time points for downstream assays.

� Combine real-time monitoring of cellular responses with complementary functional endpoint assays, and maximize data quality before, during, and after your experiment.

Compact. Convenient. Versatile.

The RTCA DP Instrument consists of two components: the RTCA Control Unit and the RTCA DP Analyzer with three integrated stations for measuring cell responses in parallel or independently.

� Choose from three types of impedance-based 16-well plates: — E-Plate 16 and E-Plate VIEW 16 for cellular assays — CIM-PLATE 16 for cell invasion/migration assays

� Use all three different plate types in any combination.

� Easily achieve optimal cell culture conditions by placing the RTCA DP Analyzer and plates into standard CO2 incubators.

2

E-Plate 16 and E-Plate VIEW 16: Cellular Assays in a 16-Well Format

� Quantitatively monitor changes in cell number, cell adhesion, cell viability, and cell morphology.

� Easily add compounds during an experiment.

� Assess short- and long-term cellular effects.

� With the E-Plate VIEW 16, observe measured changes using microscopes.

E-Plate 16

E-Plate 16 E-Plate VIEW 16

500 μm

Obtain detailed information about your cells with the versatile RTCA DP Instrument, which supports

up to three plates of any type – E-Plate 16, E-Plate VIEW 16, or CIM-Plate 16 – in any combination.

For example, cell invasion/migration assays and cytotoxicity assays or short- and long-term assays

may be run simultaneously.

E-Plates for the RTCA DP Instrument More Flexibility. More Data. More Insight.

Figure 3: Continuously monitor cells and determine optimal time points for assessing cytotoxicity. Cell proliferation and cell death were continuously monitored using the xCELLigence RTCA DP Instrument. The optimal time points for visual inspection of HeLa cells were determined and images taken 4 and 22 hours after compound treatment using a Z16 Apo Microscope with light base (Leica Microsystems).

Control

300 μM Antimycin A

Nor

mal

ized

Cel

l Ind

ex

1.9

0.9

-0.19 18 27 36 45 54

Time (Hours)

Antimycin A administration

Control5 μM100 μM300 μM

4 h 22 h

Figure 2: Easily visualize cells while measuring cell response with xCELLigence System E-Plate VIEW technology. A modified version of the standard E-Plate 16, the E-Plate VIEW 16 enables image acqui-sition using microscopes or automated cell-imaging systems. For the modification, four rows of microelectrode sensors were removed in each well to create a window for visualizing cells. Approximately 70% of each well bottom is covered by the microelectrodes, providing cell impedance measurements nearly identical to those obtained with the standard E-Plate 16. Both plate types can be used in parallel.

3

CIM-Plate 16

Lid

MicroelectrodesMicroporous Membrane

Adherent Cell

Upper ChamberCells

Gel Layer(user provided)

Chemoattractant

Lower Chamber

CIM-Plate 16: Quantitative Cell Invasion/Migration Analysis

� Monitor cell invasion and migration continuously in real time over the entire time course of an experiment.

� Eliminate time-consuming manual detection (Figure 4).

� Perform CIM analysis in a convenient one-well system (Figure 5).

Figure 4: Quantitatively measure the rate and onset of invasion while concurrently assessing migration. HT1080 cells (2 x 104) were seeded in the upper chamber of CIM-Plate wells coated with varying dilutions of Matrigel, or in wells with no coating. Serum was added to the lower chamber of selected wells as a chemoattractant. Invasion was observed and migration monitored continuously over a 70-hour period. All serum-starved samples resulted in base-line Cell Index levels, indicating the absence of invasion/migration, while those wells with chemoattractant induced migration.

Invasion/Migration Analysis in CIM-Plates

Figure 5: Analyze invasion/migration in real time with the CIM-Plate 16. The plate features two separable sections for ease of experimental setup. Cells seeded in the upper chamber move through the microporous membrane into the lower chamber that contains a chemoattractant. Cells adhering to the microelectrode sensors lead to an increase in impedance, which is measured in real time by the RTCA DP Instrument.

4

E-Plate Insert 16: Co-Culture in Real-Time

� Continuously monitor indirect cell-cell interactions.

� Assess short- and long-term cell response without labor-intensive labeling and microscopy.

� Co-culture different cell typtes under physiological conditions for a broad range of applications, including:

Cancer Research: Assess paracrine stimulation of cancer cell proliferation by fibroblasts.

Immunology: Investigate immune cell interactions.

Stem Cell Research: Monitor proliferation and differentiation in the presence of stimulation cells.

Toxicology: Determine cytotoxicity of agents and assess effects of cytokine release.

Figure 6. Real-time monitoring of co-culture-induced proliferation stimulation and its inhibition using the E-Plate Insert. Intercellular interactions play an important role in normal cell development and tumorigenesis. Results show that the proliferation of hormone-responsive tumor cells is likely mediated by hormones and growth factors exchanged between the two cell populations separated by the E-Plate Insert.

Elevated T47D cell proliferation on the E-Plate (green trace � ) was induced by hormone secretion of H295R cells in the insert, and inhibited

by the hormone synthesis inhibitor Prochloraz (blue trace � ). Incubation of T47D cells with only the E-Plate Insert did not affect proliferation (red

trace � ).

Lid

E-Plate 16

E-Plate Insert

5

1. Cell Invasion and Migration

MicroRNA-200c Represses Migration and Invasion of Breast Cancer Cells by Targeting Actin-Regulatory Proteins FHOD1 and PPM1Ferences.Jurmeister S, Baumann M, Balwierz A, Keklikoglou I, Ward A, Uhlmann S, Zhang JD, Wiemann S, Sahin O. Mol Cell Biol. 2012; 32(3):633–651.

c-Myb regulates matrix metalloproteinases 1/9, and cathepsin D: implications for matrix-dependent bre-ast cancer cell invasion and metastasis.Knopfová L, Beneš P, Pekar�íková L, Hermanová M, Masa�ík M, Pernicová Z, Sou�ek K, Smarda J. Mol Cancer. 2012; 11:15.

Comparative Analysis of Dynamic Cell Viability, Migration and Invasion Assessments by Novel Real-Time Technology and Classic Endpoint Assays.Limame R, Wouters A, Pauwels B, Fransen E, Peeters M, Lardon F, De Wever O, Pauwels P. PLoS One. 2012; 7(10): e46536.

2. Compound-mediated Cytotoxicity/Apoptosis

Screening and identification of small molecule compounds perturbing mitosis using time-dependent cellular response profiles.Ke N, Xi B, Ye P, Xu W, Zheng M, Mao L, Wu MJ, Zhu J, Wu J, Zhang W, Zhang J, Irelan J, Wang X, Xu X, Abassi YA. Anal Chem. 2010; 82(15):6495-503.

Kinetic cell-based morphological screening: prediction of mechanism of compound action and off-target effects.Abassi YA, Xi B, Zhang W, Ye P, Kirstein SL, Gaylord MR, Feinstein SC, Wang X, Xu X. Chem Biol. 2009; 16(7):712-23.

3. Cell-mediated Cytotoxicity

Real-time profiling of NK cell killing of human astrocytes using xCELLigence technology. Moodley K, Angel CE, Glass M, Graham ES. J Neurosci Methods. 2011; 200(2): 173-180.

Unique functional status of natural killer cells in metastatic stage IV melanoma patients and its modula-tion by chemotherapy.Fregni G, Perier A, Pittari G, Jacobelli S, Sastre X, Gervois N, Allard M, Bercovici N, Avril MF, Caignard A. Clin Cancer Res. 2011; 17(9): 2628–37.

4. Cell Adhesion and Cell Spreading

A role for adhesion and degranulation-promoting adapter protein in collagen-induced platelet activati-on mediated via integrin a2b1.Jarvis GE, Bihan D, Hamaia S, Pugh N, Ghevaert CJ, Pearce AC, Hughes CE, Watson SP, Ware J, Rudd CE, Farndale RW. Journal of Thromb Haemost. 2012; 10(2): 268–277.

Dynamic monitoring of cell adhesion and spreading on microelectronic sensor arrays.Atienza JM, Zhu J, Wang X, Xu X, Abassi Y. J Biomol Screen. 2005; 10(8): 795-805.

Selected Publications for the RTCA DP Instrument

6

5. Receptor-mediated Signaling

Impedance responses reveal b2-adrenergic receptor signaling pluridimensionality and allow classifica-tion of ligands with distinct signaling profiles.Stallaert W, Dorn JF, van der Westhuizen E, Audet M, Bouvier M. PLoS One. 2012; 7(1): e29420.

Label-free impedance responses of endogenous and synthetic chemokine receptor CXCR3 agonists cor-relate with Gi-protein pathway activation.Watts AO, Scholten DJ, Heitman LH, Vischer HF, Leurs R. Biochem Biophys Res Commun. 2012; 419(2):412-8.

Impedance measurement: A new method to detect ligand-biased receptor signaling.Kammermann M, Denelavas A, Imbach A, Grether U, Dehmlow H, Apfel CM, Hertel C. Biochem Biophys Res Commun. 2011; 412(3): 419-424.

6. Virus-mediated Cytopathogenicity

Novel, real-time cell analysis for measuring viral cytopathogenesis and the efficacy of neutralizing anti-bodies to the 2009 influenza A (H1N1) virus.Tian D, Zhang W, He J, Liu Y, Song Z, Zhou Z, Zheng M, Hu Y. PloS One. 2012; 7(2):e31965.

Real-time monitoring of flavivirus induced cytopathogenesis using cell electric impedance technology.Fang Y, Ye P, Wang X, Xu X, Reisen W. J Virol Methods. 2011; 173(2):251–8.

7. Quality of Control of Cells

Rapid and quantitative assessment of cell quality, identity, and functionality for cell-based assays using real-time cellular analysis.Irelan JT, Wu MJ, Morgan J, Ke N, Xi B, Wang X, Xu X, Abassi YA. J Biomol Screen. 2011; 16(3):313-22.

Live cell quality control and utility of real-time cell electronic sensing for assay development.Kirstein SL, Atienza JM, Xi B, Zhu J, Yu N, Wang X, Xu X, Abassi YA. Assay Drug Dev Technol. 2006; 4(5):545-53.

8. Endothelial Barrier Function

An inverted blood-brain barrier model that permits interactions between glia and inflammatory stimuli.Sansing HA, Renner NA, MacLean AG. J Neurosci Methods. 2012; 207(1):91–6.

A dynamic real-time method for monitoring epithelial barrier function in vitro.Sun M, Fu H, Cheng H, Cao Q, Zhao Y, Mou X, Zhang X, Liu X, Ke Y. Anal Biochem. 2012; 425(2):96–103.

Selected Publications continued

7

Ordering Information for xCELLigence RTCA DP System

Product Cat. No. Pack Size

xCELLigence RTCA DP Instrument RTCA DP Analyzer RTCA Control Unit

003806010500546975900105454417001

1 Bundled Package1 Instrument1 Notebook PC

E-Plate 16

E-Plate VIEW 16

E-Plate Insert 16

0546983000105469813001063247380010632474600106465382001

6 Plates 6 x 6 Plates 6 Plates6 x 6 Plates1 x 6 Devices (6 16-Well Inserts)

CIM-Plate 16 0566581700105665825001

6 Plates6 x 6 Plates

CIM-Plate 16, Assembly Tool 05665841001 1 Assembly Tool

Learn more about the enabling technology of the xCELLigence System and its broad range of applications at www.aceabio.com

XCELLIGENCE, E-PLATE, CIM-PLATE, and ACEA BIOSCIENCES are registered trademarks of ACEA Biosciences, Inc. in the US and other countries. All other product names and trademarks are the property of their respective owners.


Published by

ACEA Biosciences, Inc.

6779 Mesa Ridge Road Ste. 100

San Diego, CA 92121

U.S.A.

www.aceabio.com

© 2013 ACEA Biosciences, Inc.

All rights reserved.

�

�

�

�

�

�

�

�

�

�

��

ISSU

E 0

4

Focus ApplicationCell Migration


xCELLigence System Real-Time Cell Analyzer

IntroductionCell migration and invasion are mechanically integrated molecular processes and fundamental components of embryogenesis, vasculogenesis, immune responses, as well as pathophysiological events such as cancer cell metastasis (1, 2). Cell migration and invasion involve morphological changes due to actin cytoskeleton rearrangement and the emergence of protrusive membrane structures followed by contraction of the cell body, uropod detachment, and secretion of matrix degrading enzymes (1, 2). These multi-step processes are influenced by extracellular and intracellular factors and signaling events through specialized membrane receptors.

The integrated nature of cell migration is exemplified by angiogenesis. Angiogenesis or neo-angiogenesis refers to the formation of new blood vessels from pre-existing vessels and is critical for development, wound healing and tumor growth. Endothelial cell migration is an important compo-nent of angiogenesis, involving chemotactic, haptotactic and mechanotactic (shear stress) induced cell migration (3). Chemotactic cell migration is typically induced by soluble growth factors such as vascular endothelial growth factor (VEGF) and its isoforms, fibroblast growth factor (bFGF) and hepatocyte growth factor (HGF) amongst others. These growth factors interact with their cognate receptor tyrosine kinases on he surface of endothelial cells activating signaling pathways culminating in directed cell migration.

In the present study, we used the new CIM-Plate 16 with the xCELLigence RTCA DP Instrument to monitor growth factor-mediated migration of endothelial cells in realtime using label-free conditions. The CIM-Plate 16 is a 16-well modified Boyden chamber composed of an upper chamber (UC) and a lower chamber (LC). The UC and LC easily snap together to form a tight seal. The UC is sealed at its bottom by a microporous Polyethylene terephthalate (PET) mem-brane. These micropores permit the physical translocation of cells from the upper part of the UC to the bottom side of the membrane. The bottom side of the membrane (the side facing the LC) contains interdigitated gold microelectrode sensors which will come in contact with migrated cells and generate an impedance signal. The LC contains 16 wells, each of which serves as a reservoir for a chemoattractant solution on the bottom side of the wells, separated from each other by pressure-sensitive O-ring seals.

ResultsTo analyze endothelial cell migration using the CIM-Plate 16, human umbilical vein endothelial cells (HUVEC) from Lifeline Cell Technologies were cultured in Vasculife VEGF cell culture medium, according to the manufacturer’s recommendations. Cells were serum starved in Vasculife Basal medium, detached using a trypsin-EDTA solution, and the cell density was adjusted to 300,000 cells/mL. To assess general HUVEC cell migration in response to

Featured Study:Automated Continuous Monitoring of Growth Factor-Mediated Endothelial Cell Migration using the CIM-Plate 16 and xCELLigence RTCA DP Instrument

Jieying Wu and Jenny Zhu

ACEA Biosciences, Inc., San Diego, USA.

different growth factors encountered during angiogenesis, Vasculife VEGF medium containing VEGF, EGF, IGF, or bFGF with 2% fetal bovine serum, was serially diluted with Vasculife Basal medium and transferred to the lower chamber of the CIM-Plate 16 (see Figure 1). For optimal HUVEC migration, it was determined from previous experiments that extracellular matrix (ECM) proteins, such as fibronectin (FN) are necessary. The PET membrane was therefore coated on both sides with 20 μg/mL FN. After CIM-Plate 16 assembly, 100 μL of cell suspension (30,000 cells) were added to each well of the UC. The CIM-Plate 16 was placed in the RTCA DP Instrument equilibrated in a CO2 incubator. HUVEC migration was continuously monitored using the RTCA DP Instrument. Figure 1 shows the time- and dose-dependent directional migration of HUVEC cells from the upper chamber to the lower chamber. The combination of growth factors and serum provides a strong chemoattractant signal which together induce the directional migration of HUVEC cells through the micropores of the CIM-Plate 16. Migrating cells are detected by the electronic sensing microelectrodes, producing changes in the measured Cell Index values (see Figure 1). HUVEC migration has been shown to be influenced by a number of growth factors including VEGF, HGF, and bFGF. These growth factors are known to be secreted by tumors and cells within the tumor stroma, as well as endothelial cells inducing migration and angiogenesis.

To measure HUVEC migration in response to individual growth factors using the CIM-Plate 16, HUVEC cells from Lonza were starved for 6 hours and detached. At the same time, titration of HGF, and separately of VEGF, was performed in basal media (complete HUVEC media diluted at a ratio of 1:125 with EGM media from Lonza). Each of

these growth factors was then transferred to the wells of the lower chamber (see Figure 2). The PET membrane was coated with FN as described above.

After CIM-Plate 16 assembly, HUVEC cells were added at 30,000 cells/well for VEGF-induced migration and 15,000 cells/well for HGF-induced migration. Time-dependent HUVEC migration was monitored using the RTCA DP Instrument. Both VEGF and HGF induced the migration of HUVEC cells in a time- and dose-dependent manner (see Figure 2A and 2B). The RTCA Software 1.2 was used to calculate time-dependent EC50 values for both VEGF- and HGF-mediated HUVEC migration (see Figure 2C and 2D).

Angiogenesis is a compelling target for cancer therapy. Monoclonal antibodies targeting angiogenesis play an important role in colon and lung cancer therapy (4). For this reason, the migration of endothelial cells in response to angiogenic factors such as VEGF is a good in vitro model system for studying and screening potential inhibitors of this process (see Figure 3). For the quantitative and time-dependent assessment of inhibition of VEGF-induced endo-thelial cell migration, HUVEC cells were added to the CIM-Plate 16, as described above, in the presence of increasing amounts of a VEGF receptor inhibitor. As shown in Figure 3A, this inhibitor was found to potently block VEGF-induced cell migration in a time- and dose- dependent manner. The inhibition of VEGF-induced HUVEC cell migration by this compound was quantified using the RTCA Software 1.2. Time-dependent IC50 values, shown in Figure 3B, demonstrate that this particular VEGF receptor inhibitor blocks the kinase activity of all three VEGF receptor isoforms with IC50 values in the picomolar range.

Figure 1: Time- and dose-dependent directional migration of HUVEC cells from the upper to the lower CIM-Plate 16 chamber. To assess HUVEC cell migration, Vasculife VEGF medium, containing VEGF, EGF, IGF, bFGF and 2% fetal bovine serum, was serially diluted with Vasculife Basal medium and transferred to the lower chamber of the CIM-Plate 16.

0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0Time (in Hour)

6.5

5.5

4.5

3.5

2.5

1.5

0.5

-0.5

Cel

l Ind

ex

Complete Media

1:3

1:9

1:27

1:81

1:243

1:729Basal Media

Figure 2: HUVEC cell migration in response to the growth factors, VEGF (A, C) and HGF (B, D) using the CIM-Plate 16 and xCELLigence RTCA DP Instrument, showing Cell Index profiles (A, B) and EC50 plots (C, D); see text in Results for details.

0.0 3.0 6.0 9.0 12.0 15.0 18.0 21.0Time (in Hour)

VEGF-induced cell migration HGF-induced cell migration

3.5

2.5

1.5

0.5

-0.5

Cel

l Ind

ex

60 ng/mL20 ng/mL

6.7 ng/mL

2.2 ng/mL

0.7 ng/mL0.2 ng/mL0 ng/mL

EGM Media

2.0

1.0

0.00.0 3.0 6.0 9.0 12.0 15.0 18.0

Time (in Hour)

Cel

l Ind

ex

100 ng/mL33.3 ng/mL200 ng/mL

3.7 ng/mL

1.2 ng/mLEGM Media

11.1 ng/mL

3.6000

3.0000

2.4000

1.8000

1.2000

0.6000

0.0000

Cel

l Ind

ex

-10.50 -10.00 -9.50 -9.00 -8.50 -8.00 -7.50 -7.00 -6.50

Log of concentration (g/ml)

2.1000

1.8000

1.5000

1.2000

0.9000

0.6000

0.3000

0.0000

Cel

l Ind

ex-9.50 -9.00 -8.50 -8.00 -7.50 -7.00 -6.50 -6.0

Log of concentration (g/ml)

T1 EC50= 5.5 ng/mLT2 EC50= 5.3 ng/mL

T1 EC50= 3.2 ng/mLT2 EC50= 2.7 ng/mL

C

A B

D

A

0.04 nM

0.12 nM

0.4 nM

1.1 nM3.3 nM10 nM Inhibitor

0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 18.0 20.0Time (in Hour)

2.5

2.0

1.5

1.0

0.5

0.0

Cel

l Ind

ex

B2.4000

2.0000

1.6000

1.2000

0.8000

0.4000

0.0000

Cel

l Ind

ex

-11.00 -10.50 -10.00 -9.50 -9.00 -8.50 -8.00 -7.50

Log of concentration (M)

T1 EC50= 78 pMT2 EC50= 179 pM

Figure 3: Real-time continuous HUVEC cell monitoring showing the Cell Index profiles (A) and IC50 plot (B) for the inhibition of VEGF-induced cell migration by a VEGF receptor inhibitor; see text in Results for details.

ConclusionData presented in this application note demonstrate that growth-factor-mediated endothelial cell migration can be monitored quantitatively and in realtime using the CIM-Plate 16 with the RTCA DP Instrument. The xCELLigence System proved to be ideal for assessing and screening an inhibitor of endothelial cell migration and angiogenesis. The CIM-Plate 16 combines the benefits of continuous label-free impedance-based technology with the classic Boyden chamber permitting automated, real-time, and quantitative measurements of cell migration and invasion.

Classic cell migration techniques utilizing standard and transwell Boyden chambers are labor intensive, producing results that can be difficult to reproduce. The non-invasive CIM-Plate 16 does not require manual cell counting or cell labeling. Moreover, the continuous real-time data obtained using the CIM-Plate 16 identifies optimal time points for performing parallel gene expression and functional analyses of HUVEC migration. The above described features and benefits of the CIM-Plate 16 and the RTCA DP Instrument describe an ideal system for the in vitro analysis of the cellular and molecular events associated with cell migration and invasion.

T1 T2 T1 T2

T1 T2

Published by

ACEA Biosciences, Inc. 6779 Mesa Ridge Road Ste 100 San Diego, CA 92121U.S.A.

www.aceabio.com

© 2013 ACEA Biosciences, Inc. All rights reserved.

Ordering Information

References1. Lauffenburger DA, Horwitz AF. (1996).

“Cell migration: a physically integrated molecular process.” Cell 84(3):359-69.

2. Ridley AJ, Schwartz MA, Burridge K, et al. (2003). “Cell migration: integrating signals from front to back.” Science 302(5651):1704-9.

3. Lamalice L, Le Boeuf F, Huot J. (2007). “Endothelial cell migration during angiogenesis.” Circulation Research 24: 100: 782-794.

4. Folkman J. (2007). “Angiogenesis: an organizing principle for drug discovery?” Nature Reviews Drug Discovery 6(4): 273-286.


Trademarks:XCELLIGENCE, CIM-PLATE, E-PLATE, and ACEA BIOSCIENCES are registered trademarks of ACEA Biosciences, Inc. in the US and other countries.Other brands or product names are trademarks of their respective holders.

Key Words:Growth factor-mediated cell migration, xCELLigence System, RTCA DP Instrument, real-time migration monitoring, CIM-Plate 16, non-invasive and label-free migration detection

Product Cat. No. Pack Size

xCELLigence RTCA DP Instrument RTCA DP Analyzer RTCA Control Unit

003806010500546975900105454417001

1 Bundled Package1 Instrument1 Notebook PC

E-Plate 16

E-Plate VIEW 16

E-Plate Insert 16

0546983000105469813001063247380010632474600106465382001

6 Plates 6 x 6 Plates 6 Plates6 x 6 Plates1 x 6 Devices (6 16-Well Inserts)

CIM-Plate 16 0566581700105665825001

6 Plates6 x 6 Plates

�

�

�

�

�

�

�

�

�

��

Comparative Analysis of Dynamic Cell Viability,Migration and Invasion Assessments by Novel Real-TimeTechnology and Classic Endpoint AssaysRidha Limame1*, An Wouters1, Bea Pauwels1, Erik Fransen2, Marc Peeters1,3, Filip Lardon1, Olivier De

Wever4, Patrick Pauwels1,5

1Center for Oncological Research (CORE), University of Antwerp, Antwerp, Belgium, 2 StatUA Center for Statistics, University of Antwerp, Antwerp, Belgium, 3Department

of Oncology, Antwerp University Hospital, Edegem (Antwerp), Belgium, 4 Laboratory of Experimental Cancer Research, Department of Radiotherapy and Nuclear

Medicine, Ghent University Hospital, Ghent, Belgium, 5 Laboratory of Pathology, Antwerp University Hospital, Edegem (Antwerp), Belgium

Abstract

Background: Cell viability and motility comprise ubiquitous mechanisms involved in a variety of (patho)biological processesincluding cancer. We report a technical comparative analysis of the novel impedance-based xCELLigence Real-Time CellAnalysis detection platform, with conventional label-based endpoint methods, hereby indicating performancecharacteristics and correlating dynamic observations of cell proliferation, cytotoxicity, migration and invasion on cancercells in highly standardized experimental conditions.

Methodology/Principal Findings: Dynamic high-resolution assessments of proliferation, cytotoxicity and migration wereperformed using xCELLigence technology on the MDA-MB-231 (breast cancer) and A549 (lung cancer) cell lines. Proliferationkinetics were compared with the Sulforhodamine B (SRB) assay in a series of four cell concentrations, yielding fair to goodcorrelations (Spearman’s Rho 0.688 to 0.964). Cytotoxic action by paclitaxel (0–100 nM) correlated well with SRB (Rho.0.95)with similar IC50 values. Reference cell migration experiments were performed using Transwell plates and correlated by pixelarea calculation of crystal violet-stained membranes (Rho 0.90) and optical density (OD) measurement of extracted dye(Rho.0.95). Invasion was observed on MDA-MB-231 cells alone using Matrigel-coated Transwells as standard referencemethod and correlated by OD reading for two Matrigel densities (Rho.0.95). Variance component analysis revealedincreased variances associated with impedance-based detection of migration and invasion, potentially caused by thesensitive nature of this method.

Conclusions/Significance: The xCELLigence RTCA technology provides an accurate platform for non-invasive detection ofcell viability and motility. The strong correlations with conventional methods imply a similar observation of cell behaviorand interchangeability with other systems, illustrated by the highly correlating kinetic invasion profiles on differentplatforms applying only adapted matrix surface densities. The increased sensitivity however implies standardizedexperimental conditions to minimize technical-induced variance.

Citation: Limame R, Wouters A, Pauwels B, Fransen E, Peeters M, et al. (2012) Comparative Analysis of Dynamic Cell Viability, Migration and Invasion Assessmentsby Novel Real-Time Technology and Classic Endpoint Assays. PLoS ONE 7(10): e46536. doi:10.1371/journal.pone.0046536

Editor: Aamir Ahmad, Wayne State University School of Medicine, United States of America

Received April 24, 2012; Accepted August 31, 2012; Published October 19, 2012

Copyright: � 2012 Limame et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permitsunrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Funding: These authors have no support or funding to report.

Competing Interests: Olivier De Wever, listed as co-author, is currently serving as an academic editor for PLOS ONE. This does not alter the authors’ adherenceto all the PLOS ONE policies on sharing data and materials.

* E-mail: [email protected]

Introduction

Among the most fundamental hallmarks of cancer are loss of

pre-existing tissue architecture by sustained proliferation and

extracellular matrix infiltration of cancer cells. Cancer cells may

sustain proliferative signaling in an autocrine or paracrine fashion

by producing growth factors themselves, by overexpression of

growth factor receptors or by a constitutive activation of

downstream signaling components [1]. Monitoring of cell prolif-

eration and cell viability is critical in biomedical research, in order

to understand the pathways regulating proliferation and viability,

and to develop agents that modulate these processes. The

sulforhodamine B (SRB) test is a high throughput and reproduc-

ible colorimetric assay, based on the binding of SRB to protein

basic amino acid residues, providing a sensitive index of cellular

protein content that is linear over a cell density range [2].

Matrix penetration necessitates activation of the cellular motility

apparatus and can occur by either individual cells or cell strands,

sheets or clusters [3]. A phenomenon predominantly involved in

this process is chemotaxis, whereby cell movement is directed

along an extracellular chemical gradient of secreted factors in the

microenvironment [4]. Already in the early stages of embryogen-

esis, formation of complex tissues and organs is orchestrated by

fine-tuned chemotactic migration of cell chains. In malignant

processes however, cancer cells tend to adopt similar, if not

identical mechanisms to metastasize to distant organ sites [5].

Several well-established experimental approaches are available to

PLOS ONE | www.plosone.org 1 October 2012 | Volume 7 | Issue 10 | e46536

study cell migration and chemotaxis in vitro (reviewed in [6]). The

Transmembrane/Boyden chamber assay is based on a chemotactic-

driven cell transit through a filter [7]. An important feature of the

endpoint in this experimental set-up is that cells need to exhibit

active migratory behavior to end up at the other side of the

membrane.

The xCELLigence RTCA technology (Roche Applied Science) has

emerged as an alternative non-invasive and label-free approach to

assess cellular proliferation, migration and invasion in real time on

a cell culture level [8]. This system makes use of impedance

detection for continuous monitoring of cell viability, migration and

invasion (reviewed in [9]) (Fig. 1).

Here we report data of in vitro assessment of four cellular

processes (proliferation, cytotoxicity, migration and invasion) on

the MDA-MB-231 and A549 cancer cell lines using xCELLigence

RTCA DP (Roche Applied Science) in comparison with data

resulting from parallel experiments applying a previously existing

and well-established measuring method (to be considered as a

‘‘gold standard’’ method) for each process. Both these cell lines are

extensively characterized and used as models representing two

different highly incidental tumor types (breast cancer, lung cancer).

Furthermore, these cell lines show a strong degree of motility in

the wild-type state, thus providing useful examples for the

distinction between chemotactic and random motility.

Importantly, all of the comparative techniques are traditional

label-based endpoint assays that have been selected due to their

widespread application within the scientific community and

similarity in working principle with xCELLigence. Although they

have been slightly modified to match with the xCELLigence setup

and its ability to acquire time-dependent kinetics of cultured cell

behavior, the fundamental handling and detection principles of

each classic assay have been maintained. Cell proliferation and

cytotoxicity testing has been performed using the SRB assay [10],

with the microtubule stabilizer paclitaxel as cytotoxic agent in the

latter experiments. Being widely used for the treatment of a variety

of tumor types, inhibitory effects of this anti-mitotic compound on

cell proliferation have been described extensively. Furthermore,

previous reports on the use of the xCELLigence device included

paclitaxel as a reference compound in their studies [8,11], making

this a suitable agent for this study. Cell migration and invasion

experiments were performed using conventional Transwell plates

and quantified by both pixel area calculation of stained

membranes and optical density reading of solubilized dye. For

the first time, results from ‘‘tried-and-tested’’ assay setups are

confronted with parallel data recorded using a novel, commer-

cially available technology, providing an objective technical

comparison of dynamic observations on cultured cells in highly

standardized experimental conditions.

Results

ProliferationThe dynamic assessment of proliferation kinetics was modeled

by performing SRB testing on both MDA-MB-231 and A549 cells.

Growth curve studies were performed over a ten-day period and

proliferation curves were established for four different plating

densities. To correct for seeding area differences between SRB-

experiments and xCELLigence, cell seeding densities were synchro-

nized between both techniques (100, 500, 1000 and 2000 cells/

cm2). Corresponding experiments on the xCELLigence system were

performed in duplicates and the resulting high-resolution data

were extrapolated to the matching data points of the counterpart

method as described in the Materials and Methods section.

For MDA-MB-231 cells, cell doubling time was

27.7865.14 hours and 29.9262.85 hours measured with the

SRB assay and the xCELLigence system respectively. Similarly,

for A549 cells, doubling time was 27.9361.75 hours and

29.1861.87 hours measured with the SRB assay and the

xCELLigence system respectively.

Spearman’s Rho (r) correlations were calculated on global

average results that had been normalized to the highest value in

the data set per method (SRB or xCELLigence) to eliminate units of

measurement (Fig. 2A, B, shown as ‘‘scaled’’). Both MDA-MB-231

and A549 cells revealed fair to good correlation rates for all

applied seeding densities, noting however that proliferation did not

set off at the lowest cell seeding density (100 cells/cm2) of MDA-

MB-231 on RTCA (Fig. 2A). A549 cells showed only minimal

proliferative activity at this density as well (Fig. 2B). Correlations

observed between SRB-based and impedance-based quantitation

reached higher values at the medium cell seeding densities of 500

cells/cm2 and 1000 cells/cm2 for both cell lines tested (Spearman’s

r=0.835 resp. 0.790 for MDA-MB-231 and r=0.964 resp. 0.883

for A549). Altogether, correlation values ranged from 0.880 to

0.964 for A549 and 0.688 to 0.835 for MDA-MB-231. Variance

component analysis on proliferation data of A549 cells resulting

from both techniques indicated smaller intra- and inter-experi-

mental variances on xCELLigence when compared to SRB.

Conversely, detection of proliferation kinetics of MDA-MB-231

cells resulted in a higher degree of (intra- and interexperimental)

variance when performed with the xCELLigence system (quantified

as sw and sb in Table S1).

Figure 1. xCELLigence RTCA: impedance-based detection of cellviability and motility. Interdigitated gold microelectrodes on thewell bottom (viability – E-plate) or on the bottom side of a filtermembrane (motility – CIM-plate 16) detect impedance changes, causedby the presence of cells and expressed as a Cell Index. This detectionmethod is proportional to both cell number (left and above) andmorphology as increased cell spreading is reflected by a higher CellIndex value (right). When starting an experiment, a baseline Cell Indexvalue is recorded in medium only before cell addition.doi:10.1371/journal.pone.0046536.g001

Real-Time Technology Compared with Endpoint Assays




CytotoxicityCompound cytotoxicity was assessed after exposure of MDA-

MB-231 and A549 cells to a concentration range of paclitaxel (0–

100 nM for 72 hours). In both cell lines, a similar toxic response

was detected by both SRB and xCELLigence, with comparable IC50

values of 4.7860.90 nM and 6.4461.90 nM respectively (t-test,

p = 0.244) in MDA-MB-231 cells. The normalized toxic response

correlated highly between SRB-based and xCELLigence-based

detection for both cell lines (Spearman’s r=0.970 and 0.976 for

MDA-MB-231 and A549 resp.) (Fig. 2C).

Cell migrationConventional Transwell plates have been organized in a

sequential setup to perform dynamic observations of cancer cell

migration in a time-dependent manner, yielding a series of data

similar to the xCELLigence system (Fig. 3A, B). Both chemotactic

migration to medium containing 10% FBS and random migration

with SF medium on either side of the membrane have been

considered. High correlation values for both cell lines were

obtained when comparing area calculation and OD with

xCELLigence Cell Index (CI) (Fig. 4A). Importantly, it must be

noted that correlation coefficients were calculated for overall mean

values resulting from three independent experiments, each

performed in duplicates for all time points. As described in the

Materials and Methods section, all raw data obtained were

normalized to the single maximal value over three experiments per

quantitation method (CI, pixel area or OD) to eliminate units of

measurement. Subsequently, values from random (SF) migration

were subtracted from chemotactic (FBS) migration per time point

to generate signals of net chemoattraction (Fig. 4B). Pixel area

calculations, averaged over three fields per insert membrane

(Fig. 4C), correlated with xCELLigence CI measurements for both

MDA-MB-231 and A549 cells (Spearman’s r=0.90 for both cell

lines). However, OD measurements showed even stronger

correlations with xCELLigence data (Spearman’s r=0.96 and

1.00 for MDA-MB-231 and A549 resp.). Area calculation

correlated with OD measurements in a similar fashion as with

xCELLigence for both MDA-MB-231 and A549 (Spearman’s

r=0.89 and r=0.90 resp.) (Fig. 4A).

At later time points of the experiments, xCELLigence generated

larger variances between intra-experimental replicates when

compared to area calculations or OD values derived from classic

Transwells (Table S1). Briefly, variance between replicates within

one experiment increased over time, creating a funnel shaped

time-dependent pattern (Fig. 4D, E). Variance component analysis

indeed revealed an increase of intra-experimental variance on

xCELLigence data in comparison with Transwell data. However,

early (,10 h) pixel area values showed similar degrees of variance.

Additional experiments using an identical setup yielded similar

results (results not shown).

A significant difference was observed between background

(serum-free, SF) signals generated by the three methodologies of

cell migration quantitation. These signals derived from random

movement of cells without exposure to any chemoattractant and,

serving as a negative control, showed a different pattern when

generated by xCELLigence or by both quantitation techniques using

Transwells (Fig. 5). Paired comparisons between time-dependent

tracking of background migration were performed between all

techniques using a likelihood ratio test and revealed a significant

difference between xCELLigence and pixel area calculation

(p,0.001) for both MDA-MB-231 and A549. OD measurements

differed significantly from xCELLigence (p,0.001) for MDA-MB-

231 cells, but showed a similar pattern when compared with

xCELLigence data generated from A549 cells (p = 0.22).

Cell invasionImpedance-based detection of MDA-MB-231 cell invasion was

compared with results derived from a Transwell system as applied

for migration experiments, with Matrigel as extracellular matrix

component added on top of the microporous membranes (Fig. 6A).

It was found that Matrigel dilutions of 10% (v/v, SF medium) on

xCELLigence yielded high correlations when compared to a dilution

of 20% on Transwells in dynamic invasion profile recording

during a 48-hour incubation (Spearman’s r=0.939). Similarly,

invasion through a Matrigel dilution of 3.3% on xCELLigence

correlated highly with a dilution of 7.7% on Transwell plates

(Spearman’s r=0.927) (Fig. 6B, C). Variance component analysis

of Transwell and xCELLigence data revealed slightly increased

degrees of intra-experimental variance in the latter. The variance

between independent experiments was increased for xCELLigence

in comparison with Transwell assays (Table S1). All Matrigel

dilutions have been synchronized regarding seeding surface area

for both systems and theoretical calculations for correlating

Matrigel dilutions are shown in Table 1.

Discussion

xCELLigence technology measures impedance changes in a

meshwork of interdigitated gold microelectrodes located at the

well bottom (E-plate) or at the bottom side of a microporous

membrane (CIM16-plate). These changes are caused by the

gradual increase of electrode surface occupation by (proliferated/

migrated/invaded) cells during the course of time and thus can

provide an index of cell viability, migration and invasion. This

method of quantitation is directly proportional to cellular

morphology, spreading, ruffling and adhesion quality as well as

cell number [8,9] (Fig. 1). Cell proliferation and paclitaxel

cytotoxicity kinetics, as assessed by the xCELLigence platform, were

compared with an SRB-based approach, showing good correla-

tions for both cell lines tested, implying that both methods detect

similar process kinetics when performed in standardized condi-

tions. However, correlations were generally stronger for the A549

than for the MDA-MB-231 cell line, which may indicate a possible

cell type-dependent cause. MDA-MB-231 cells show a heteroge-

neous morphotype with round and spread cells, which may

differentially influence impedance based measurements or crystal

violet uptake. Paclitaxel was chosen as it is widely used as

treatment for a variety of tumor types and previous reports on the

use of the xCELLigence system as a tool for cytotoxicity screening

included paclitaxel as a reference compound [8,11]. Nevertheless,

it must be underlined that the highly correlative nature of

cytotoxic kinetics as detected by both techniques for paclitaxel may

not apply for certain other compounds. Indeed, the xCELLigence

Figure 2. Time-dependent proliferation and cytotoxicity profiles of MDA-MB-231 and A549. A. Proliferation curves of MDA-MB-231 cellsas generated by xCELLigence RTCA (red) and SRB (black) for different seeding densities of 100 (top left), 500 (bottom left), 1000 (top right) and 2000cells/cm2 (bottom right) during a ten-day incubation. B. Same as (A) for A549 cells. All graphs represent results from three independent experiments6SD. C. Cytotoxicity profiles relating to 72 hours of exposure to paclitaxel (0–100 nM). Cells were allowed to attach and propagate during 24 hoursprior to start of treatment. Toxicity data from xCELLigence RTCA were derived from normalized plots. All graphs represent results from threeindependent experiments 6 SD.doi:10.1371/journal.pone.0046536.g002



RTCA device has been reported to generate compound-specific

kinetic profiles on cultured cells, hereby demonstrating associa-

tions with the respective mechanisms of action [11].

To quantify cell migration through conventional setups,

detection by crystal violet was selected, followed by pixel area

and OD quantitation, as these methods are widespread within the

scientific community and also take morphologic features into

account. Both area calculation and OD measurement correlated

highly with cell migration, as detected by xCELLigence, confirming

that the observed kinetic cell behavior is strongly similar, provided

that equal cell seeding densities are applied. The closest

associations were found between OD measurements and RTCA

CI, generating nearly identical migration patterns. OD values

were determined on cell lysates with extracted crystal violet stain,

derived from entire Transwell membranes, and thus provide a

more reliable quantitation when compared with pixel area

calculation, which was based on averaging three microscopic

fields per Transwell membrane. This indicates that impedance-

based measurements have smaller limits of detection, resulting in

highly reliable migration estimates. Analysis of background signals,

resulting from random migration in a serum-free environment

(negative control), revealed a significant difference in signal

detection between xCELLigence and both classic techniques

(Fig. 5). Weak signals resulting from limited baseline cell migration

were indeed more accurately detected by the RTCA platform,

which implies a higher sensitivity, compared to classic detection

methods. OD measurements correlated more closely with

xCELLigence data in all experiments and did not show a significant

difference when background signals were compared with xCELLi-

gence for the A549 cell line, suggesting a smaller detection limit and

thus a more accurate method of quantifying cell migration when

using Transwells. Additionally, a smaller limit of detection can

explain the observed time-dependent increase of variability

between intra-experimental replicates during the course of an

experiment. Data derived from cell culture-based experiments are

subject to inter- as well as intra-individual variation regarding cell

counting, pipetting, preparation of chemotactic factors and

general cell culture handling. As a consequence, small handling

differences during the preparation of an experiment can result in

signal differences between technical replicates and this will be

reflected in variations increasing with time within one experiment

when compared to area calculation or OD measurement, that do

not detect this variability to this extent. This is also illustrated by

cell culture-based invasion experiments, as the uniform application

Figure 3. Conventional Transwell design for detection of time-dependent cell migration. A. A Transwell setup consists of an upperchamber (insert) that is placed onto a lower chamber (well). The insert contains a microporous membrane (8 mm pores) allowing passage of tumorcells. After a period of serum starvation a serum-free cell suspension is seeded in the insert and exposed to medium containing potentialchemoattractants (by default: medium+FBS). During incubation at 37uC and 5% CO2, cells migrate toward the bottom side of the membrane. B.Experimental design to assess time-dependent migratory behavior of cultured cells. Both migration toward FBS-containing medium and baselinemigration (toward SF medium, no chemoattraction) as a negative control were included. Two times two 24-well Transwell plates were used toexamine migration to FBS (positive control – top row) and baseline migration (negative control – bottom row). At ten time points during a 24-hourincubation period inserts were fixed and stained in duplicate. Two inserts containing cell-free media (grey fill) have been included throughout theexperiment and fixed and stained after 12 hours incubation to assess background absorption in optical density (OD) measurements. In addition, toexclude influence of inter-plate variability on observed migration rates, each plate contained duplicate two-hour control inserts.doi:10.1371/journal.pone.0046536.g003





of a Matrigel layer implies the introduction of an important added

variable. Consequent Matrigel thawing and handling on ice, using

only cooled consumables, and hands-on coating experience can

contribute to homogenous gelification and thus to enhanced assay

reproducibility. Comparative invasion results have shown corre-

lation between xCELLigence and Transwells when similar cell

seeding densities were applied and, more importantly, when the

amount of Matrigel per square area unit is synchronized. Diluting

the matrix barrier, and thus changing the degree of matrix

fenestration, gives rise to similar invasion rates on both setups with

different seeding areas, although equal volumes of Matrigel have

been applied.

In conclusion, the real-time label-free xCELLigence system

provides a suitable and accurate platform for high-throughput

kinetic screenings and for determination of cell motility dynamics.

In contrast with classic endpoint assays, the impedance-based

detection method is generally less labor-intensive, provides kinetic

information on the studied processes and does not affect cell

viability, potentially generating further experimentation possibil-

ities. Moreover, the correlating observations as performed with

conventional approaches make methods interchangeable to

perform functional studies when larger cell populations of interest

are needed. However although impedance measurement provides

a sensitive cell-based detection method, it should be applied as a

complementary tool to further functional confirmation.

Importantly, this is the first study illustrating the highly

correlative nature of invasion kinetics detected by two different

setups applying synchronized matrix densities. The increased

sensitivity, however, necessitates standardized experimental con-

ditions and user experience, to minimize variance increments on

the xCELLigence system.

Materials and Methods

Cell cultureTwo malignant cell lines (A549, lung adenocarcinoma; MDA-

MB-231, breast adenocarcinoma), obtained from the American

Type Culture Collection (ATCC, Manassas, VA, USA) (http://

Figure 4. Time-dependent migratory pattern of MDA-MB-231 and A549. A. MDA-MB-231 (left) and A549 (right) cell migration profiles,detected by Transwell experiments (black) and xCELLigence (red). Graphs represent scaled signals (0–1) of net chemoattraction after subtraction of therandom migration signal (empty squares in panel A, B), with associated Spearman’s Rho values. All results originate from three independent duplicateexperiments 6 SD. B. Normalization procedure of migration patterns. Raw data (left panel) were normalized to a (0–1) scale (middle panel) throughdivision of all data by the maximum value obtained in three independent experiments. Subsequently, randommigration (SF) signals (triangle markers)were subtracted from the positive (FBS) control counterparts (circle markers) per experiment to obtain a pure chemotactic signal (right panel).Example shown is the migratory pattern of MDA-MB-231 cells estimated by pixel area calculation in three experiments (exp 1 - red, exp 2 - green, exp3 - black). Triangle and circle markers represent negative (SF) and positive (FBS) control data respectively. C. ImageJ-based picture processing. Originalpictures were color thresholded to obtain a binary image displaying cellular content as saturated black areas on a white background. Thresholdedimages were masked to exclude non-cellular particles from the final area calculation. Pictures shown are migrated MDA-MB-231 cells after four hours(top row) and 16 hours (bottom row) of incubation. D. Migratory behavior of MDA-MB-231 cells toward medium+FBS (positive control – filled squares)and background migration (empty squares) as detected by conventional Transwell experiments at ten time points spread over 24 hours of incubation.Area calculation (left) of stained cells and optical density (OD – middle) were compared to the xCELLigence migration pattern, reconstructed from theoriginal high-resolution plot by extrapolating data from the corresponding time points (right). All results represent original data from threeindependent duplicate experiments 6 SD. Picture string (obj. 2.56) shows migratory status of MDA-MB-231 cells, stained as described, at fivedifferent stages during 24 hours of incubation. E. Same as (D) for A549 cells.doi:10.1371/journal.pone.0046536.g004

Figure 5. Time-dependent random migration profile of MDA-MB-231 and A549. Comparison of random migration signals (negativecontrol – SF) between three quantitation methods: pixel area calculation – black, OD - red, xCELLigence - green. A likelihood ratio test revealed asignificant difference in slope between area calculation and OD (p,0.001) and area calculation and xCELLigence (p,0.001) for both cell lines and ODand xCELLigence (p,0.001) for MDA-MB-231 only. OD and xCELLigence slopes did not differ significantly (p = 0.22) for A549 cells.doi:10.1371/journal.pone.0046536.g005



www.lgcstandards-atcc.org), were cultured in DMEM and

RPMI1640 respectively, each supplemented with 10% fetal bovine

serum (FBS), 1% penicillin/streptomycin, 1% L-glutamine and

additionally, 1% sodium pyruvate was added to RPMI1640 only.

All cell culture reagents were purchased from Invitrogen NV/SA

(Merelbeke, Belgium). For proliferation and cytotoxicity experi-

ments, normal growth medium containing FBS was used. Cell

lines were maintained at 37uC and 5% CO2/95% air in a

humidified incubator and confirmed free of mycoplasma infection

through regular testing (MycoAlertH Mycoplasma Detection Kit, Lonza,

Belgium). All cell lines have been validated in-house by short

tandem repeat (STR) profiling using the Cell IDTM System

(Promega, Madison, WI, USA) according to the manufacturer’s

Figure 6. Time-dependent invasion profile of MDA-MB-231. A. Experimental design to quantify MDA-MB-231 Matrigel invasion. Two timestwo 24-well Transwell plates were used to examine invasion to FBS through a 20% (v/v) (top row) and 7.7% (v/v) Matrigel layer (bottom row) after24 hours of serum starvation. At ten time points during a 48-hour incubation period inserts were fixed and stained in duplicate. Two insertscontaining cell-free media (grey fill) have been included throughout the experiment and fixed and stained after 24 hours incubation to assessbackground absorption in optical density (OD) measurements. In addition, to exclude influence of inter-plate variability on observed migration rates,each plate contained duplicate 24-hour control inserts. B. MDA-MB-231 dynamic cell invasion profiles, generated by Transwell experiments (black)and xCELLigence (red). Graphs represent normalized signals (scaled values 0–1) of invasion through 20% (open circles), 10% (open squares), 7.7% (filledcircles) and 3.3% (filled squares) to medium+10% FBS with associated Spearman’s rank correlation coefficients (Rho). All results are from threeindependent duplicate experiments with SD. C. Sequential pictures showing invasive MDA-MB-231 cells at the indicated time points during a 48-hourincubation on Transwells coated with 20% (top row) and 7.7% Matrigel (bottom row). Pictures (obj. 2.56) show cells fixed and stained in 20%methanol/0.1% crystal violet.doi:10.1371/journal.pone.0046536.g006

Table 1. Matrigel surface densities corresponding withdegree of dilution for a fixed volume of 20 mL.

Matrigel density

xCELLigence RTCA Transwell

% mg/cm2* %

10.0 189.50 23.1

3.3 63.17 7.7

*Matrigel (Basement Membrane Matrix, growth factor reduced, BD Biosciences)delivered as a 613.55 mg/mL stock.doi:10.1371/journal.pone.0046536.t001



instructions. The obtained STR profiles were matched with

reference ATCC DNA fingerprints (www.lgcstandards-atcc.org)

and with the Cell Line Integrated Molecular Authentication

(CLIMA) database (http://bioinformatics.istge.it/clima) [12] to

authenticate cell line identity.

xCELLigence Real-Time Cell Analysis (RTCA): proliferationand cytotoxicityExperiments were carried out using the xCELLigence RTCA DP

instrument (Roche Diagnostics GmbH, Mannheim, Germany)

which was placed in a humidified incubator at 37uC and 5% CO2.

Cell proliferation and cytotoxicity experiments were performed

using modified 16-well plates (E-plate, Roche Diagnostics GmbH,

Mannheim, Germany). Microelectrodes were attached at the

bottom of the wells for impedance-based detection of attachment,

spreading and proliferation of the cells. Initially, 100 mL of cell-

free growth medium (10% FBS) was added to the wells. After

leaving the devices at room temperature for 30 min, the

background impedance for each well was measured. Cells were

harvested from exponential phase cultures by a standardized

detachment procedure using 0.05% Trypsin-EDTA (Invitrogen

NV/SA, Merelbeke, Belgium) and counted automatically with a

Scepter 2.0 device (Merck Millipore SA/NV, Overijse, Belgium),

Fifty mL of the cell suspension was seeded into the wells (20, 40, 80,

100, 200, 400 and 800 cells/well for proliferation, 1000 cells/well

for cytotoxicity experiments). The cell concentrations of 20, 100,

200 and 400 cells/well were considered for correlation with the

SRB method described below. After leaving the plates at room

temperature for 30 min to allow cell attachment, in accordance

with the manufacturer’s guidelines, they were locked in the RTCA

DP device in the incubator and the impedance value of each well

was automatically monitored by the xCELLigence system and

expressed as a Cell Index value (CI). Water was added to the space

surrounding the wells of the E-plate to avoid interference from

evaporation. For proliferation assays, the cells were incubated

during ten days in growth medium (10% FBS) and CI was

monitored every 15 min during the first six hours, and every hour

for the rest of the period. Two replicates of each cell concentration

were used in each test. For cytotoxicity experiments, CI of each

well was automatically monitored with the xCELLigence system

every 15 min during the overnight recovery period. Twenty-four

hours after cell seeding, cells were treated during a period of

72 hours with paclitaxel (0, 1, 2, 5, 10, 20, 50 and 100 nM)

dissolved in phosphate buffered saline (PBS). PBS alone was added

to control wells. Each concentration was tested in duplicate within

the same experiment. CI was monitored every 15 min during the

experiment. Three days after the start of treatment with paclitaxel,

CI measurement was ended.

xCELLigence Real-Time Cell Analysis (RTCA): migrationand invasionCell migration and invasion experiments were performed using

modified 16-well plates (CIM-16, Roche Diagnostics GmbH,

Mannheim, Germany) with each well consisting of an upper and a

lower chamber separated by a microporous membrane containing

randomly distributed 8 mm-pores. This setup corresponds to

conventional Transwell plates with microelectrodes attached to

the underside of the membrane for impedance-based detection of

migrated cells. Prior to each experiment, cells were deprived of

FBS during 24 hours. Initially, 160 mL and 30 mL of media was

added to the lower and upper chambers respectively and the CIM-

16 plate was locked in the RTCA DP device at 37uC and 5% CO2

during 60 minutes to obtain equilibrium according to the

manufacturer’s guidelines. After this incubation period, a

measurement step was performed as a background signal,

generated by cell-free media. To initiate an experiment, cells

were detached using TrypLE ExpressTM (Invitrogen, Merelbeke,

Belgium), resuspended in serum-free (SF) medium, counted and

seeded in the upper chamber applying 36104 cells in 100 mL.After cell addition, CIM-16 plates were incubated during

30 minutes at room temperature in the laminar flow hood to

allow the cells to settle onto the membrane according to the

manufacturer’s guidelines. To prevent interference from evapora-

tion during the experiments, SF medium was added to the entire

empty space surrounding the wells on the CIM-16 plates. Lower

chambers contained media with or without FBS in order to assess

chemotactic migration when exposed to FBS and background

migration to SF medium as a negative control accordingly. Signals

representing net chemoattraction were obtained by subtracting

background (SF) values from the positive control (medium

containing FBS) signals. Each condition was performed in

quadruplicate with a programmed signal detection schedule of

each three minutes during the first 11 hours of incubation followed

by each five minutes for three hours and finally each 15 minutes to

24 hours of incubation.

A protocol identical to the above migration experiments was

followed for invasion experiments added with the application of a

layer of Matrigel on the upper side of the membranes and dynamic

process follow-up during 50 hours. Aliquoted Matrigel (Basement

Membrane Matrix, growth factor reduced, BD Biosciences,

Erembodegem, Belgium) was thawed overnight on ice and mixed

with ice cold SF medium to obtain two dilutions corresponding

with 6190 mg/mL (10%, v/v) and 663 mg/mL (3.3%, v/v)

(Table 1). All Matrigel handling materials as well as the sealed

packs containing CIM-16 upper chambers were stored ice cold

overnight. Establishment of a Matrigel layer on the CIM-16 upper

chamber membranes was achieved by adding 50 mL of the

dilution sequentially on top of four membranes followed by

removal of 30 mL, leaving a total of 20 mL Matrigel dilution.

Subsequently, the coated upper chambers were incubated at 37uCto homogenously gelify during a minimum of four hours, followed

by addition of 160 mL media to the lower and 30 mL SF media to

the upper chambers. Equilibration and cell addition was carried

out as described above.

All data have been recorded by the supplied RTCA software (vs.

1.2.1). As described below in the ‘‘Prestatistical data processing

and statistical analysis’’ section, original high-resolution data sets

generated by xCELLigence were exported to MS Excel and

reconstructed at a lower resolution by selecting only the data

points corresponding with the respective time points of signal

detection by the endpoint methods (Fig. 7). CI-data from

cytotoxicity experiments have been normalized using the RTCA

software to the last data point prior to treatment start. All other

results (proliferation, migration and invasion) are based on raw

data without CI-normalization and were processed as described

above for comparison with conventional methodology.

SRB assay: proliferationCells were harvested from exponential phase cultures by

trypsinization, counted and plated in 48-well plates. To determine

a proliferation curve and calculation of the doubling time, seeding

densities ranged from 100 to 2000 A549 or MDA-MB-231 cells/

well. Each concentration was tested six times within the same

experiment. General cell culture conditions and culture medium

used for this method were similar to those applied for the

xCELLigence counterpart experiments, as well as applied cell

densities (100, 500, 1000 and 2000 cells/cm2). Every day, one



plate was fixed by the first step of the SRB assay: culture medium

was aspirated prior to fixation of the cells by addition of 200 mlcold 10% trichloroacetic acid. After one hour incubation at 4uC,cells were washed five times with deionized water and left to dry.

After collection of all plates during ten days, the following steps of

the SRB test were performed as described previously [13,14].

Shortly, the cells were stained with 200 ml 0.1% SRB dissolved in

1% acetic acid for at least 15 minutes and subsequently washed

four times with 1% acetic acid to remove unbound stain. The

plates were left to dry at room temperature and bound protein

stain was solubilized with 200 ml 10 mM unbuffered TRIS base

(tris(hydroxymethyl)aminomethane) and transferred to 96 wells

plates for optical density reading at 540 nm (Biorad 550

microplate reader, Nazareth, Belgium). Cell doubling time was

calculated from the exponential phase of the growth curve.

SRB assay: cytotoxicityCells were harvested as described above. In order to assure

exponential growth during the experiments, seeding density was

103 A549 cells per well and 103 MDA-MB-231 cells/well. After an

overnight recovery period, treatment with paclitaxel (0–100 nM)

dissolved in PBS was started. Control wells were added with PBS.

Figure 7. Prestatistical data processing. A. Schematic depiction of processing kinetic data generated by SRB, xCELLigence and Transwells. Rawdata with high time resolution (filled and empty circles), resulting from independent xCELLigence experiments (1, 2, 3 and grey arrows) are reduced to alower time resolution by selecting only the data points corresponding with the time points of endpoint detection (filled circles only). Subsequently,data have been normalized by dividing all values by the highest value recorded over all experiments per method, resulting in a modified Y-axis scalethat ranges from 0 to 1. Finally, the normalized data have been averaged with calculation of SD for the three independent experiments per method.B. Reduction of high-resolution data, generated by xCELLigence, to a low resolution comparable with data from conventional assays. The exampleshows migration (left) and invasion (right) of MDA-MB-231 cells through two densities of Matrigel. The ten time points in the Transwell method (blackarrows) were selected from the xCELLigence plots (grey and blue) to reconstruct a low-resolution graph (black), directly comparable to the Transwelldata. An identical approach was applied for all other processes studied.doi:10.1371/journal.pone.0046536.g007



Each concentration was tested six times within the same

experiment. After 72 hours incubation with paclitaxel, survival

was determined by the SRB assay as described above. IC50 values,

representing the drug concentration causing 50% growth inhibi-

tion, were calculated using WinNonlin software (Pharsight,

Mountain View, USA).

Transwell migration assayComparative migration experiments were conducted using a

conventional 24-well Transwell system (6.5 mm TranswellH(#3422), CorningH, NY, USA) with each well separated by a

microporous polycarbonate membrane (10 mm thickness, 8 mmpores) into an upper (‘‘insert’’) and a lower chamber (‘‘well’’). After

24 hours of serum deprivation, cells were detached using

TrypLETM Express (Invitrogen, Merelbeke, Belgium), counted

and resuspended in media without FBS to obtain equal cell

densities (2.16105 cells/cm2) as applied in the xCELLigence RTCA

DP system with respect to the membrane seeding surface of both

techniques (classic TranswellH membrane surface 0.33 cm2,

RTCA DP 0.143 cm2). A volume of 250 mL containing 76104

cells was plated to each insert and 600 mL medium was added to

the wells. For each experiment, both chemotactic migration to

medium containing 10% FBS and random migration with SF

medium on both sides of the membrane have been assessed in

parallel Transwell plates. At ten predetermined time points after

incubation start (Fig. 3B), inserts were fixed and stained in

duplicates and migration was quantitated using two commonly

used methods. At each time point, cells were fixed and stained in a

20% methanol/0.1% crystal violet solution during three minutes

at room temperature, followed by washing in deionized water to

remove redundant staining [15]. Non-migrated cells remaining at

the upper side of the membranes were carefully removed with

cotton swabs and inserts were dried in darkness overnight. As

fixing and staining was performed per set of two inserts with the

remaining inserts to be further incubated until the following time

point, inserts to be fixed and stained at a time point were

transferred to a companion 24-well plate and the remaining inserts

immediately replaced in the incubator. Using this approach

unfavorable influences caused by continuous switching incubating

cells between 37uC and ambient temperatures could be avoided.

The following day stained membranes were pictured in three

random non-overlapping fields at 106objective and 106eyepiece

on a transmitted-light microscope (Leica DMBR, Leica Micro-

systems GmbH, Wetzlar, Germany) equipped with an AxioCam

HRc camera (Carl Zeiss MicroImaging GmbH, Jena, Germany).

A first method of quantitation was performed by processing all

obtained images using ImageJ software (http://rsbweb.nih.gov/ij/

). Each image (Fig. 4C, left) was color thresholded to obtain a

binary (black & white – 8 bit) image with cellular material

portrayed as saturated black areas (Fig. 4C, middle). As a next step

all non-cellular artifacts, predominantly visible shadows of empty

pores and debris, were removed from each image by performing

the particle analysis function with a mask excluding all particles

smaller than 100 to 250 pixels dependent on the experiment

(Fig. 4C, right). Masking thresholds were set by comparing binary

images with their original phase contrast counterparts [16].

Degree of migration for each time point per experiment was

determined by calculating the average pixel area of the three fields

in duplicate. Inserts were subsequently submerged in 300 mL 1%

SDS/16 PBS in order to lyse migrated cells and extract crystal

violet stain [17]. Submerged inserts were incubated in darkness

overnight on a plate shaker at medium speed to ensure complete

lysis. The following day 200 mL of each lysate was transferred to a

96-well plate for optical density (OD) measurement at 590 nm

using a Powerwave X microplate scanning spectrophotometer

(Bio-Tek, Bad Friedrichshall, Germany), representing a second cell

migration quantitation method. Cell-free inserts containing only

medium had been included in duplicate throughout each

experiment as OD background controls. Reported OD data

represent average background-corrected values 6 SD obtained

from three independent experiments in duplicate.

Transwell Matrigel Invasion assayReference cell invasion experiments were carried out using a

Transwell plate system as described for migration experiments,

added with the application of Matrigel as extracellular matrix

component. Matrigel dilutions were prepared as described above

for the xCELLigence RTCA invasion assay. In order to obtain

Matrigel surface area densities synchronized with the CIM-16

plates used for xCELLigence, two dilutions of 20% and 7.7% (v/v)

have been prepared in ice cold SF medium corresponding with

6190 mg/mL and 663 mg/mL respectively, as applied in a volume

of 20 mL per insert membrane, identical to the CIM-16 upper

chamber coating volume (Table 1). All other conditions regarding

culturing, cell seeding density and serum deprivation were

identical to the Transwell migration assays described above. At a

panel of ten predetermined time points, inserts were fixed and

stained in duplicates and invasion was quantified by OD reading

at 590 nm after overnight extraction of the crystal violet stain.

Reported OD data represent average background-corrected

values 6 SD obtained from three independent experiments in

duplicate.

Prestatistical data processing and statistical analysisAll data recorded using the xCELLigence RTCA system have

been processed using MS Excel in order to obtain data series with

the same resolution as the data recorded by the conventional

reference methods (SRB, Transwell). This has been performed by

selecting only the xCELLigence-generated values corresponding to

the time points that have been used in the reference methods, thus

leading to a reconstruction of the studied process dynamics at a

lower time resolution (Fig. 7A, B).

Furthermore, to eliminate differences in units of measurement

between the compared methods (xCELLigence CI, OD, pixel area),

all data have been reduced to a (0–1) scale. This was done by

considering all data gathered over three performed experiments

per method and subsequently dividing all data by the single

maximal value obtained, thus reducing this value to one (Fig. 4B

and 7A). As these interventions do not influence proportionality

nor variance levels of the data, comparable series of dynamically

generated results were obtained for further statistical processing.

For cell migration experiments, random migration signals (SF)

were subtracted from the chemotactic migration signals (FBS) per

time point to generate dynamic profiles of net chemoattraction

(Fig. 4B).

All statistical analyses were performed using the statistical

package R, version 2.13.1 (www.r-project.org). Correlations were

calculated according to the Spearman’s rank correlation method.

Intra- and inter-experimental variances were assessed for each

quantitation method separately using a mixed model approach

with time as fixed and biological replicate as random effect. The

standard deviation of the random intercept (inter-experimental

variance) as well as the residual standard deviation (intra-

experimental variance) was obtained through a variance compo-

nent analysis. Due to differences in these values regarding the cell

migration data, calculations were split up into ‘‘early’’ (before ten

hours incubation) and ‘‘late’’ (after ten hours incubation)

measurements. All cell migration data analyses were performed



on background (SF)-reduced signals representing pure chemoat-

traction. Comparison of background migratory (negative control –

SF) signal detection between the three quantitation techniques

(pixel area calculation, OD and RTCA CI) was performed by

fitting a mixed linear model where the signal was regressed on

time, technique and their interaction. Biological replicates were

entered as a random effect. A likelihood ratio test was performed

to test the significance, expressed as a p-value, of the interaction

term time-technique.

Supporting Information

Table S1 Variance component analysis of proliferation,cytotoxicity, migration and invasion. All values expressed as

the square root of the variance (s2). sb Variance between

independent experiments (‘‘between’’). sw Variance within one

experiment (‘‘within’’). Low 6102 and 656102 cells/cm2. High

6103 and 626103 cells/cm2. Early Before 10 hours incubation.

Late After 10 hours incubation. ND Not done. * Matrigel dilution

in SF medium (v/v).

(DOCX)

Acknowledgments

We thank Ken Op de Beeck (CORE and Center for Medical Genetics,

University of Antwerp) for assistance with cell line authentication, the

Laboratory for Pathophysiology (University of Antwerp) for granting access

to the microscope for picturing Transwell membranes and the Laboratory

for Experimental Medicine and Pediatrics (University of Antwerp) for

granting access to the microplate scanning spectrophotometer for OD

measurements. We gratefully acknowledge Marc Baay and Johan Ides

(CORE, Antwerp) for constructive discussions.

Author Contributions

Conceived and designed the experiments: RL AW BP EF MP FL ODW

PP. Performed the experiments: RL AW BP. Analyzed the data: RL AW

BP EF. Contributed reagents/materials/analysis tools: MP FL PP. Wrote

the paper: RL. Evaluated and interpreted results: RL AW BP EF MP FL

ODW PP. Evaluated manuscript text: AW EF MP FL ODW PP.

References

1. Hanahan D, Weinberg RA (2011) Hallmarks of cancer: the next generation. Cell

144: 646–674. doi:10.1016/j.cell.2011.02.013.

2. Skehan P, Storeng R, Scudiero D, Monks A, McMahon J, et al. (1990) New

colorimetric cytotoxicity assay for anticancer-drug screening. J Natl Cancer Inst

82: 1107–1112.

3. Friedl P, Gilmour D (2009) Collective cell migration in morphogenesis,

regeneration and cancer. Nat Rev Mol Cell Biol 10: 445–457. doi:10.1038/

nrm2720.

4. Roussos ET, Condeelis JS, Patsialou A (2011) Chemotaxis in cancer. Nat Rev

Cancer 11: 573–587. doi:10.1038/nrc3078.

5. Friedl P, Wolf K (2003) Tumour-cell invasion and migration: diversity and

escape mechanisms. Nat Rev Cancer 3: 362–374. doi:10.1038/nrc1075.

6. Hulkower KI, Herber RL (2011) Cell Migration and Invasion Assays as Tools

for Drug Discovery. Pharmaceutics 3: 107–124. doi:10.3390/pharmaceu-

tics3010107.

7. BOYDEN S (1962) The chemotactic effect of mixtures of antibody and antigen

on polymorphonuclear leucocytes. J Exp Med 115: 453–466.

8. Ke N, Wang X, Xu X, Abassi YA (2011) The xCELLigence system for real-time

and label-free monitoring of cell viability. Methods Mol Biol 740: 33–43.

doi:10.1007/978-1-61779-108-6_6.

9. Atienza JM, Yu N, Kirstein SL, Xi B, Wang X, et al. (2006) Dynamic and label-

free cell-based assays using the real-time cell electronic sensing system. Assay

Drug Dev Technol 4: 597–607. doi:10.1089/adt.2006.4.597.

10. Vichai V, Kirtikara K (2006) Sulforhodamine B colorimetric assay for

cytotoxicity screening. Nat Protoc 1: 1112–1116. doi:10.1038/nprot.2006.179.

11. Abassi YA, Xi B, Zhang W, Ye P, Kirstein SL, et al. (2009) Kinetic Cell-BasedMorphological Screening: Prediction of Mechanism of Compound Action and

Off-Target Effects. Chem Biol 16: 712–723. doi:10.1016/j.chem-

biol.2009.05.011.12. Romano P, Manniello A, Aresu O, Armento M, Cesaro M, et al. (2009) Cell

Line Data Base: structure and recent improvements towards molecularauthentication of human cell lines. Nucleic Acids Res 37: D925–D932.

doi:10.1093/nar/gkn730.

13. Pauwels B, Korst AEC, de Pooter CMJ, Lambrechts HAJ, Pattyn GGO, et al.(2003) The radiosensitising effect of gemcitabine and the influence of the rescue

agent amifostine in vitro. Eur J Cancer 39: 838–846.14. Pauwels B, Korst AEC, De Pooter CMJ, Pattyn GGO, Lambrechts HAJ, et al.

(2003) Comparison of the sulforhodamine B assay and the clonogenic assay for

in vitro chemoradiation studies. Cancer Chemother Pharmacol 51: 221–226.doi:10.1007/s00280-002-0557-9.

15. Zhang D, LaFortune TA, Krishnamurthy S, Esteva FJ, Cristofanilli M, et al.(2009) Epidermal Growth Factor Receptor Tyrosine Kinase Inhibitor Reverses

Mesenchymal to Epithelial Phenotype and Inhibits Metastasis in InflammatoryBreast Cancer. Clinical Cancer Research 15: 6639–6648. doi:10.1158/1078-

0432.CCR-09-0951.

16. de Wever O, Hendrix A, de Boeck A, Westbroek W, Braems G, et al. (2010)Modeling and quantification of cancer cell invasion through collagen type I

matrices. Int J Dev Biol 54: 887–896. doi:10.1387/ijdb.092948ow.17. Huang J, Bridges LC, White JM (2005) Selective modulation of integrin-

mediated cell migration by distinct ADAM family members. Mol Biol Cell 16:

4982–4991. doi:10.1091/mbc.E05-03-0258.



Mechanistic modeling of the effectsof myoferlin on tumor cell invasionMarisa C. Eisenberga,1, Yangjin Kimb, Ruth Lic, William E. Ackermanc, Douglas A. Knissc,d, and Avner Friedmana,e,1

aMathematical Biosciences Institute, Ohio State University, Columbus, OH 43210; bDepartment of Mathematics, University of Michigan, Dearborn, MI48128; cDepartment of Obstetrics and Gynecology, Ohio State University, Columbus, OH 43210; dDepartment of Biomedical Engineering, Ohio StateUniversity, Columbus, OH 43210; and eDepartment of Mathematics, Ohio State University, Columbus, OH 43210

Contributed by Avner Friedman, October 5, 2011 (sent for review August 18, 2011)

Myoferlin (MYOF) is a member of the evolutionarily conservedferlin family of proteins, noted for their role in a variety of mem-brane processes, including endocytosis, repair, and vesicular trans-port. Notably, ferlins are implicated in Caenorhabditis eleganssperm motility (Fer-1), mammalian skeletal muscle developmentand repair (MYOF and dysferlin), and presynaptic transmission inthe auditory system (otoferlin). In this paper, we demonstrate thatMYOF plays a previously unrecognized role in cancer cell invasion,using a combination of mathematical modeling and in vitro experi-ments. Using a real-time impedance-based invasion assay (xCELLi-gence), we have shown that lentiviral-based knockdown of MYOFsignificantly reduced invasion of MDA-MB-231 breast cancer cellsin Matrigel bioassays. Based on these experimental data, we de-veloped a partial differential equation model of MYOF effectson cancer cell invasion, which we used to generate mechanistichypotheses. Themathematical model predictions revealed thatma-trix metalloproteinases (MMPs) may play a key role in modulatingthis invasive property, which was supported by experimental datausing quantitative RT-PCR screens. These results suggest thatMYOFmay be a promising target for biomarkers or drug target for meta-static cancer diagnosis and therapy, perhaps mediated throughMMPs.

cancer invasion ∣ RNAi ∣ partial differential equation models ∣ metastasis

Amajority of cancer deaths are related not to the primarytumor itself, but rather the formation of disseminated me-

tastases (1). Cancer spread requires that cells achieve atypicalmotility, which enables them to invade surrounding tissues andvessels of the blood and lymphatic systems (2–4). Thus, under-standing the mechanisms and signaling processes that lead toinvasive cell behavior may lead to new therapeutic approaches forcontrolling and treating cancer.

The fundamental mechanisms of invasive cancer cell move-ment are largely conserved across a wide range of cell types, withsome of the protease dependent and protease independent move-ment types demonstrated by cancer cells also seen in organisms asdiverse as unicellular organisms, slime molds, and white bloodcells. The ferlin family is an evolutionarily ancient family of pro-teins (5), which are known to affect processes crucial to migrationand invasion, including membrane fusion and repair, vesicletransport, endocytosis, protein recycling and stability, and cellmotility (6–13). Thus, one might expect the ferlin family to begood candidates for cancer proteins, although they have notpreviously been investigated in this capacity. In Caenorhabditiselegans, spermatozoa exhibit amoeboid movement, and muta-tions in the fer-1 gene [an orthologue of myoferlin (MYOF)]result in immobility and infertility (13). In humans, MYOF hasbeen implicated in a variety of cellular processes, including myo-blast fusion, growth factor receptor stability, endocytosis, andendothelial cell membrane repair (6, 8, 10–12); however untilnow its role in cancer cell movement has not been explored.Although information on MYOF is currently limited, it has beenshown to be upregulated in breast cancer biopsies (14) andexpressed in breast cancer cell lines (15). Immunohistochemical

evidence available from the Human Protein Atlas (16) suggeststhat MYOF is strongly expressed in several cancer types includ-ing colorectal, breast, ovarian, cervical, endometrial, thyroid,stomach, pancreatic, and liver cancer (14, 15, 17–26).

To explore the function of MYOF in cancer, a stable lineof MYOF-deficient malignant breast carcinoma cells (MDA-MB-231) was generated using lentivirus-based delivery of shRNAconstructs targeting human MYOF mRNA (Sigma). A stable,lentiviral control cell line was generated in tandem using lentivir-al particles carrying a nonhuman gene targeting shRNA (Sigma).MYOF depletion was validated by immunoblotting (SI Appendix,Fig. 2). We used an electrode-impedance-based invasion assay[xCELLigence (27)] to probe the effect of MYOF deficiency oncell invasion. Compared to the control MBA-MB-231 cells,MYOF-knockdown (MYOF-KD) cells exhibited reduced inva-sive capacity (28).

Motivated by these experimental results, we developed a math-ematical model that examines the role of MYOF in cancer cellinvasion. Because relatively little is known about the function ofMYOF in cancer, there is a useful opportunity for mathematicalmodeling to suggest hypotheses, which can then be tested experi-mentally. The model is described by a system of partial differen-tial equations (PDEs). It builds on previous work on cancer cellmigration/invasion (29), now incorporating a submodel forMYOF-mediated growth factor receptor recycling.

Using multiple MYOF-related datasets (8–12), we determinedseveral parameters which differed between wild-type/controland MYOF-deficient cells. Our simulations suggest that one keyparameter—the matrix metalloproteinase (MMP) productionrate—is enough to reproduce the experimental data showingreduced MYOF-KD cell invasion. Based on the mathematicalmodel, we hypothesized that MYOF affects MMP productionand/or secretion in MDA-MB-231 cells. Preliminary experimen-tal results thus far confirm our hypothesis. Indeed, pilot PCRresults presented in this work show that MMPs may be signifi-cantly downregulated by MYOF depletion.

We propose that MYOF may serve as a fundamental playerin cancer cell movement, by regulating the local behavior ofthe plasma membrane and affecting trafficking of receptors andproteins to and from the membrane. In particular, MYOF effectson MMP production and release may be key to its role in regulat-ing tumor cell invasivity. Metastasis requires cancer cells to de-velop increased invasive capability, suggesting that MYOF mayplay an important role in the ability of tumor cells to metastasize.

Author contributions: M.C.E., Y.K., R.L., W.E.A., D.A.K., and A.F. designed research,performed research, and wrote the paper.

The authors declare no conflict of interest.1To whom correspondence may be addressed. E-mail: [email protected] [email protected].

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1116327108/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1116327108 PNAS Early Edition ∣ 1 of 6

APP

LIED

MAT

HEM

ATICS

MED

ICALSC

IENCE

S

Mathematical ModelSpatial Setup. The conventional modified Boyden chamber setupincludes two chambers with a semipermeable membrane betweenthem. There is typically laminin-rich matrix (e.g., Matrigel) ontop of the semipermeable membrane, which replicates the ECM,and invasion is measured by the number of cells which invadethrough the matrix and cross the membrane from the upper tothe lower chamber. For the xCELLigence data, the membraneis coupled with a microelectrode array at the bottom of the upperchamber. Cells begin in the upper chamber (Ω in Fig. 1), fromwhere they can migrate and invade through the ECM (region Sin Fig. 1) to reach the microelectrode array. They then cross themembrane/microelectrode array and attach to the bottom side ofthe array upon crossing. Growth factors (GF) may be introducedin the bottom well below the microelectrode array, to act as achemoattractant for the cells.

We measure the approximate number of cells which haveadhered to the microelectrode array by measuring the change inimpedance, the cell index [although we note that cell index is alsodependent on other factors, such as cell adhesion and spreading(27)]. To simulate these conditions, we can use a similar setupas for the conventional modified Boyden chamber simulationsgiven in refs. 29 and 30, with several modifications to incorporatethe microelectrode array and measurement of cell index. Fullmathematical description and details on the spatial setup andboundary/initial conditions are given in SI Appendix.

State Variable Definitions. We introduce the following variables:R1, free growth factor receptor (GFR) (number∕cell)R2, surface-bound GF-GFR complex (number∕cell)R3, internalized GF-GFR complex (number∕cell)ρ, concentration of ECM (g∕cm3)P, MMP concentration (g∕cm3)G, GF concentration (g∕cm3)n, density of breast cancer cells (cells∕cm3)Although MMPs are a diverse family of proteins with overlap-

ping yet distinct functions, for simplicity we model their combinedeffects with the variable P, which represents generic/nonspecificMMP concentration. Similarly, because our experimental datause fetal calf serum as a chemoattractant, the variable G repre-sents a combination of multiple growth factors (further detailsgiven in SI Appendix).

Model Equations. The model equations are based on the modelvariable interactions shown in Fig. 2. MYOF has been shownin other contexts to affect membrane processes (6, 8, 11) andreceptor/protein recycling/transport (10, 12). Parameters relatingto these functions, e.g., receptor recycling parameters, are un-derlined in Eqs. 1–7, indicating they are considered MYOF-dependent. We developed two versions of the cell-relatedparameters—one for wild-type/control cells and another forMYOF-KD cells, as described below.

∂R1

∂t¼ ðλ1 − k21R1Gþ k13R3 − k01R1Þ

nn0

[1]

∂R2

∂t¼ ðk21R1G − k32R2 þ k23R3Þ

nn0

[2]

∂R3

∂t¼ ½k32R2 − ðk13 þ k23 þ k03ÞR3�

nn0

[3]

∂ρ∂t

¼ −λ21Pρ|fflfflffl{zfflfflffl}degradation

þ λ22ρ

�1 −

ρ

ρ0

�|fflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflffl}reconstructionðsmallÞ

in S [4]

Fig. 1. xCELLigence well setup. The upper (Ω) and lower chambers are se-parated by a semipermeable membrane and microelectrode array at x1 ¼ 0.Matrigel/ECM sits on top of the microelectrode array (indicated by region S).

Fig. 2. Model for receptor recycling and cell migration and invasion, with processes marked in red affected byMYOF. Cells (n) proliferate, migrate, and invadefollowing these processes, with cell movement dependent on the growth factor gradient (chemotaxis) and the extracellular matrix gradient (haptotaxis).

2 of 6 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1116327108 Eisenberg et al.

∂P∂t

¼ Dp∇2P þ λ31nρ|fflffl{zfflffl}production by cells

− λ32P|ffl{zffl}degradation

[5]

∂G∂t

¼ DG∇2G − k21CmwR1nG|fflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflffl}binding

− λ10G|ffl{zffl}degradation

[6]

∂n∂t

¼ Dn∇2nþ λ11nð1 −nn0

− μρÞ R2

R20|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}proliferation

− ∇ ·�χn

n R2

R20∇Gffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

1þ λGj∇Gj2p

|fflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflffl}chemotaxis

þ χ0nIS

n∇ρffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1þ λρj∇ρj2

q|fflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}

haptotaxis

�[7]

Eqs. 1–3 constitute the receptor recycling ordinary differentialequation (ODE) submodel, collectively representing growthfactor receptor binding, internalization, degradation, and returnto the cell surface. We assume that all the internalized ligand isdegraded so that ligand return to the surface may be neglected.The receptor recycling submodel equations are multiplied byn∕n0 to scale the concentrations to the local cell density.

The remaining PDEs are largely based on a previous modelof tumor growth and movement in a Boyden chamber (29), up-dated and with modifications to incorporate MYOF effects andour particular experimental setup. The ECM, in Eq. 4, undergoesdegradation by MMP (31) and includes a small remodeling term(30, 32–34). In Eq. 5, MMP is produced by the cells to degradethe matrix, and then degrades and diffuses. Because MMP isproduced in response to the presence of extracellular matrix,we have modified this term from ref. 29 to be dependent on bothn and ρ. Growth factor in Eq. 6 diffuses from below the xCELLi-gence well bottom, where it binds to free receptors on the cellsurface, and is degraded at a constant rate.

Lastly, tumor cells Eq. 7 begin above the ECM in the upperchamber, from which they then undergo dispersion, chemotaxisfollowing the GF concentration gradient, haptotaxis through theECM, following the ECM concentration gradient, and growthfactor dependent proliferation based on the level of surface-bound growth factor (Fig. 2). Cell proliferation is modeled aslogistic growth, to which we add an additional ECM-dependentterm to account for additional crowding effects in the presenceof ECM (35). The diffusion constants and other parameters arepositive constants.

Parameter Estimation. As discussed above, MYOF is known toaffect a variety of membrane processes (6, 8, 11) and receptor/protein recycling and transport (10, 12). Thus, we developedtwo versions of the membrane-related model parameters, under-lined in Eqs. 1–7, each characterizing the wild-type/lentiviralcontrol and MYOF-KD cell types represented in our study.

The two versions of the receptor recycling submodel para-meters in Eqs. 1–3 were determined by fitting to experimentalreceptor internalization data from wild-type and MYOF-nullmyoblasts [MYOF knockout (MYOF-KO)] cells (12), with thefull details of the model parameterization given in SI Appendix.The resulting parameter estimates suggest that MYOF-deficientcells yield decreased GFR production, and increased receptorrecycling pathways leading to receptor degradation.

There are three MYOF-dependent parameters in the full PDEmodel which remain unaccounted for—the MMP secretion rateand the chemotactic and haptotactic sensitivity parameters. AsMYOF-KD cells show decreased invasivity compared to wild-type/control cancer cells, we expect these parameters to decreasein the MYOF-KD case. The ODE submodel parameters show

between a 13 and 40% change between wild-type and MYOF-KD parameter values, so we suppose χnm ¼ 0.75χn andχ0nm ¼ 0.75χ0n. As MMP has been shown to associate with MYOF(36) and MMP secretion is directly membrane-related, we wouldexpect that λ31 will be more strongly affected by the MYOF-KD.Indeed, we found the best fits to the xCELLigence invasion datafor a significantly lower value of λ31m, so that we take λ31m ¼λ31∕100. The remaining non-MYOF-dependent PDE modelparameters for both the wild-type/control and MYOF-KD cellswere determined based on literature values (29, 30, 36, 37) (seeSI Appendix, Tables 1 and 2 for individual references).

Simulation Results. The overall model behavior for the wild-type/control and MYOF-KD cells in the xCELLigence wells with 20%Matrigel coating are shown in SI Appendix, Figs. 3 and 4. Tumorcell invasion is more significant in wild-type/control than MYOF-KD cells, matching experimental data. Bound and internalizedreceptors tend to follow the invading front of tumor cells, withcells toward the top of the upper well tending to have moreunbound receptors (as the GF has not diffused completely up thechamber). MMPalso tends to roughly follow the invading front oftumor cells, as does degradation of ECM, with MMP productiondropping off outside the ECM region.

Applications to Cancer Cell InvasionDecreased Invasion in MYOF-KD Cells. Figs. 3 and 4 show modelsimulations compared to cell index experimental data in xCEL-Ligence wells for wild-type/control and MYOF-KD cells at 20%and 100% Matrigel concentrations. Model simulations recoverthe qualitative behavior of the experimental data, with a moresignificant decrease in invasivity for MYOF-KD cells in 100%Matrigel compared to 20% Matrigel.

0 5 10 15 20 250

0.05

0.1

0.15

0.2

0.25

0.3

0.35

Time (h)

Cel

l ind

ex

ControlMYOF KDMMP KD

A

B

Fig. 3. Comparison between experimental data and simulation results with20% Matrigel. (A) Experimental results for lentiviral control (LTV-ctrl) andMYOF-KD cells. (B) Simulation results for wild-type/control, MYOF-KD, andhypothetical MMP KD cells.

Eisenberg et al. PNAS Early Edition ∣ 3 of 6

APP

LIED

MAT

HEM

ATICS

MED

ICALSC

IENCE

S

MMP is a Key Target for Myoferlin Effects. We also simulated thecell index for a hypothetical MMP-KD case, where the modelparameters were fixed at wild-type/control values except for theMMP production rate, λ31, which was set equal to the decreasedMYOF-KD value λ31m. We found that loss of MMP alone wasenough to account for most of the changes in cell invasivity seenin MYOF-KD cells (Figs. 3 and 4).

Preliminary Experimental Validation. Based on these results, weused a PCR array to examine whether MYOF depletion affectsMMPs. We found that MMP1 was over 100-fold downregulatedin MYOF-KD MDA-MB-231 cells compared to lentiviral con-trols, matching the model predictions and suggesting that MMPsmay indeed be a target of MYOF. Additionally, MYOF has beenfound to be a substrate for membrane type 1 MMP (MT1-MMP)in MDA-MB-231 cells (38). These results serve to partially vali-date our simulation predictions that MMP may be important toMYOF effects on cell invasion.

Myoferlin and Receptor Tyrosine Kinases (RTKs).Because MYOF hasbeen shown to stabilize RTKs such as Tie-2, insulin-like growthfactor (IGF) receptors and VEGF receptors (10, 12, 39), we hy-pothesized that MYOF may provide a generalized mechanismof RTK (and other receptor) stabilization in cancer cells, perhapsby modulating receptor recycling so that vesicles are targeted forrecycling rather than the lysosomal degradation pathway [as sug-gested for myoblast fusion and muscle growth in (12)]. Thus, wemodeled MYOF effects using a generalized receptor, whereMYOF-KD affects the recycling/degradation rates for interna-lized receptors. Preliminary experimental results confirm thatphosphorylated forms of several RTKs (EphB4, FGFR2, Hck,IGF-1R, JAK2, tyrosine protein kinase, and VEGFR2) are sig-

nificantly downregulated in MYOF-KD MDA-MB-231 cells(SI Appendix, Fig. 5).

Predictive Simulations of Myoferlin Effects on Chemotaxis and Hapto-taxis. Next, we simulated varying concentrations of Matrigel (P)and growth factor (G) to explore how wild-type/control andMYOF-KD cells may behave under a range of experimentalconditions. We simulated varying ECM concentrations from10–100%, and growth factor concentrations for 0.5, 1, and 2 timesnormal. Simulated MYOF-KD cells showed a more dramaticdecrease in invasion than wild-type/control cells, consistent withthe significant effects on invasion related parameters in themodel (SI Appendix, Fig. 6). By contrast, the effects of varyingGF were relatively small in both wild-type and MYOF-KD simu-lations. Simulated MYOF-KD cell invasion was less affectedthan wild-type/control simulations, perhaps as a result of thedecreased sensitivity of MYOF-KD cells to GF due to changesin chemotactic sensitivity and growth factor receptor recycling(SI Appendix, Fig. 7).

In Vivo Tumor Growth Simulations.All the models and experimentaldata persented here thus far have considered cells in vitro. How-ever, we can use the model to explore hypothetical cancer cellbehavior in a simulated tumor in vivo, and in particular, to exam-ine how a heterogeneous microenvironment affects cell invasion.We simulated a two-dimensional scenario of a clump of wild-typeor MYOF-KD tumor cells surrounded by ECM, using differingdiffusion coefficients (normal and 1∕10 normal based on SIAppendix, Table 1) on the left and right halves of the domain.In wild-type cells, cell invasion was significantly reduced for lowerdiffusion coefficients (SI Appendix, Fig. 8). MMP followed theinvading front of tumor cells, as in the xCELLigence wells simu-lations. Although GF concentrations were fairly similar for bothhalves of the domain, bound, unbound, and internalized GFreceptor concentrations were quite different for the two diffusioncoefficient values. For MYOF-KD cells, tumor invasion wasmarkedly less than in the wild-type case, so that after 24 h, thetumor cells had not significantly invaded the surrounding tissue(SI Appendix, Fig. 9). MMP, ECM, and GF levels change less thanin the wild type case, and the differences between the two diffu-sion coefficient microenvironments were less evident, likely dueto decreased overall invasion.

DiscussionMYOF belongs to the evolutionarily ancient ferlin family, whichhas been associated with a variety of processes important to cellmigration and invasion, such as cell motility, growth factor recep-tor stability, endocytosis, and membrane repair (7–10, 11, 13).Although MYOF and the ferlin family have not previously beenstudied for their role in cancer, several reports suggest an asso-ciation between MYOF overexpression and metastatic cancer(14, 15, 17–26). In this work we used a combination of mathema-tical modeling and experimental validation to uncover MYOF asa cancer protein, which we propose may be involved in regulatingcancer cell invasion.

We developed a PDE model of MYOF-mediated cancer cellinvasion, as a tool to generate hypotheses regarding the mechan-isms behind MYOF effects on breast cancer cell invasion. Themodel extends existing models of cancer cell movement (29, 30)to incorporate receptor recycling and membrane regulatoryeffects of MYOF, based on experimental results using xCELLi-gence assays (27) showing that MYOF-KD MDA-MB-231 cellsexhibited decreased invasivity (Figs. 3 and 4, with immunoblotvalidation of MYOF specific depletion shown in SI Appendix,Fig. 2).

Model simulations confirm the experimental observation ofdecreased invasion in MYOF-KD cells (Figs. 3 and 4), andpredictive simulations suggest that MYOF-KD cells may be less

0 5 10 15 20 250

0.05

0.1

0.15

0.2

0.25

0.3

0.35

Time (h)

Cel

l ind

ex

ControlMYOF KDMMP KD

A

B

Fig. 4. Comparison between experimental data and simulation results with100% Matrigel. (A) Experimental results for lentiviral control (LTV-ctrl) andMYOF-KD cells. (B) Simulation results for wild-type/control, MYOF-KD, andhypothetical MMP KD.


able to invade (and thus metastasize) in vivo as well. The modelsimulations also suggest that the effects of MYOFon cell invasionmay be in large part mediated by MMPs, as similar invasion pro-files could be obtained in simulation by MYOF-mediated lossof MMP production/secretion alone. Based on these predictions,we proposed that MMPs may be key targets of MYOF, and thatthe effects of MYOF-KD on tumor cell invasiveness may be duein large part to the loss of MYOF effects on MMP productionand/or secretion.

These modeling results led us to survey MMP expressionfollowing MYOF-KD. We used quantitative RT-PCR arrays toexamine gene regulatory changes in MYOF-KD cells, and foundan unexpected link between MYOF and MMPs, namely that lossof MYOF results in 100-fold downregulation of MMP-1, match-ing model predictions. These results confirm that MYOF doesindeed have a significant effect on MMP, and highlight how in-teractions between experiments and mathematical modeling canprovide a fruitful method for generating and testing hypotheses.

In this model we also hypothesized that MYOF may providea generalized mechanism of RTK stabilization in cancer cells,perhaps via changes in the receptor recycling/degradation path-ways (motivated by ref. 12). We have partially confirmed thishypothesis using RTK phosphorylation arrays (SI Appendix,Fig. 5). Although the present paper is focused on cancer cellinvasion, this process is only one facet of the larger picture ofmetastasis, which involves a wide variety other processes (includ-ing migration, angiogenesis, and proliferation at secondary sites).Subsequent work in progress on migration and other aspects ofMYOF in cancer provide an opportunity for this generalizedmodel to be specialized for individual receptors and the interac-tions between MYOF and RTKs to be further examined.

To conclude, MYOFappears to be a promising target in cancertherapy. Our mathematical and experimental results highlightthe connection between membrane trafficking processes andcell invasion, and suggest that MYOF plays a significant role inpromoting invasive behavior. Moreover, based on our model pre-dictions together with preliminary experimental validation, wepropose that MYOF may play a significant role in regulatingMMPs in cancer. Indeed, based on our model predictions, wehypothesize that these effects on MMPs are an important com-ponent of the regulatory changes associated with MYOF. Pre-vious studies (29, 40, 41) have highlighted the effects of thetumor microenvironment on cell movement, suggesting that mod-ulating cell adhesion or blocking MMPs may be a way to controland treat cancer. MYOF may be a promising target for suchtherapeutic approaches, or perhaps a biomarker for metastasis.As more information about the regulatory network of MYOF incancer is uncovered, our model may be extended to include adynamic equation for MYOF within a larger protein network,and used to further test potential therapeutic targets.

Materials and MethodsMathematical Modeling Methods. All model simulations were per-formed using a finite volume method and clawpack (http://www.amath.washington.edu/∼claw/) with fractional step method (42) as well as thenonlinear solver nksol for algebraic systems. The model equations weresolved on a regular uniform spatial grid and an adaptive time step. Simulatedcell index was assumed to be proportional to the fraction of cells that reachthe well bottom, where we found a proportionality constant of 0.5 matchedthe experimental data well.

Cell Culture and Generation of Stably Transduced Cell Lines. MDA-MB-231(American Type Culture Collection, HTB-26) cells were maintained in DMEM

with 4.5 g∕L D-glucose supplemented with 10% FBS. Recombinant lentiviralparticles containing nontarget control shRNA (SHC002V) and human myofer-lin (TRCN0000010628) targeted shRNA in the pLKO.1 vector were purchasedfrom Sigma-Aldrich (MISSION). For lentiviral infection, cells were seededin 24-well culture plates and incubated overnight at 37 °C in 5% CO2 in ahumidified atmosphere. Media was replaced with media containing 8 μg∕mLhexadimethrine bromide (Sigma) and lentiviral particles were added tocultures that were approximately 70% confluent at a multiplicity of infectionof 1. After overnight incubation (37 °C, 5% CO2), virus-containing super-natant was removed and replaced with complete media and incubatedovernight once more. Stably transduced cultures were selected in mediacontaining an appropriate amount of puromycin as predetermined by apuromycin kill curve. Wild-type and MYOF-deficient cells were validatedthrough short tandem repeat profiling at Johns Hopkins University’s Frag-ment Analysis Facility.

Real-Time Invasion Monitoring. The invasion assays were completed on CIM-16plates with 8 μm pore membranes (Roche). Wells were coated with 20 μL of20% or 100%Matrigel and allowed to gel at 37 °C, 5% CO2 for 4 h. After 4 h,the wells of the bottom chamber were filled with 160 mL of 10% serum con-taining media and the top and bottom portions of the CIM-16 plates wereassembled together. The assembled CIM-16 plate was allowed to equilibratefor 2 h at 37 °C, 5% CO2 after the addition of 50 μL of serum-freemedia to thetop chamber wells. For seeding, cells were rinsed with PBS, trypsinized for2 min, centrifuged at 150 g for 3 min, and washed with serum-free DMEMbefore resuspension in serum-free DMEM. Cells (8 × 104 cells∕well) wereseeded onto the top chambers of CIM-16 plates and placed into the xCELLi-gence system for data collection after a 30 min incubation at room tempera-ture. The xCELLigence software was set to collect impendence data (reportedas cell index) at least once every hour. Percentage invasion was calculatedby the ratio of the cell index of invaded cells (with Matrigel coating) to thecell index of migrated cells (no Matrigel coating).

Quantitative RT-PCR Array. RT2 Profiler PCR array from SABiosciences wasperformed for human extracellular matrix and adhesion molecules followingmanufacturer’s protocol. Briefly, total RNA was extracted from lentiviralcontrol and MYOF-deficient MDA-MB-231 cells that had been in culturefor 10 consecutive passages using TRIzol reagent (Invitrogen), according tomanufacturers’ protocol up to the chloroform extraction and centrifugationsteps. The resulting aqueous phase was mixed with an equal volume of70% ethanol and applied to an RNeasy mini column (QIAGEN) and processedaccording to the manufacturer’s protocol with DNase on column digestion.Total RNAwas quantified with a NanoDrop 2000, and 1 μg of total RNA fromeach cell type was reverse-transcribed to cDNA using the RT2 First Strand Kit(SABiosciences). Equal amounts of diluted cDNA was mixed with LightCycler480 SYBR Green I Master mix (Roche) and aliquoted to each well of thePCR array plate containing the prefilled gene-specific primer sets, and PCRwas performed according to manufacturer’s instructions for the Roche Light-Cycler 480. The LightCycler 480 software (Roche) was used to calculate thethreshold cycle (crossing point, Cp) values for all the transcripts in the array.The Cp values were then exported into an Excel-based PCR array data analysistemplate (SABiosciences) to calculate fold changes in gene expression forpairwise comparisons using the ΔΔCt method (where Ct is threshold cycle).

RTK Array. Initial screen of relative levels of phosphorylation of a panel ofRTKs in MDA-MB-231 lentiviral control and MYOF-depleted cells were deter-mined by an antibody array (RayBio Human RTK Phosphorylation AntibodyArray 1, RayBiotech, Inc.), using 600 μg∕mL of protein for each cell type andfollowing the manufacturer’s protocol. The membranes were quantifiedby densitometry (Quantity One software, BioRad), with global backgroundsubstraction of the average value of negative control spots on the array.

ACKNOWLEDGMENTS. This work was supported in part by National ScienceFoundation Award 0635561 (to M.E.), Ohio State University PerinatalResearch and Development Fund (D.A.K., W.E.A., R.L.), K08 HD49628(to W.E.A.), National Science Foundation Award EEC-0425626 (to R.L.), andRoche Applied Science (D.A.K., W.E.A.).

1. Eccles SA, Welch DR (2007) Metastasis: Recent discoveries and novel treatment

strategies. Lancet 369:1742–1757.

2. Howlett NG, et al. (2002) Biallelic inactivation of BRCA2 in fanconi anemia. Science

297:606–609.

3. Thiery JP, Acloque H, Huang RY, Nieto MA (2009) Epithelial-mesenchymal transitions

in development and disease. Cell 139:871–890.

4. Lopez-Novo JM, Nieto A (2009) Inflammation and EMT: An alliance towards organ

fibrosis and cancer progression. EMBO Mol Med 1:303–314.

5. Lek A, Lek M, North K, Cooper S (2010) Phylogenetic analysis of ferlin genes reveals

ancient eukaryotic origins. BMC Evol Biol 10:231.

6. Doherty KR, et al. (2008) The endocytic recycling protein EHD2 interacts with myofer-

lin to regulate myoblast fusion. J Biol Chem 283:20252–20260.

Eisenberg et al. PNAS Early Edition ∣ 5 of 6

APP

LIED

MAT

HEM

ATICS

MED

ICALSC

IENCE

S

7. Davis DB, Doherty KR, Delmonte AJ, McNally EM (2002) Calcium-sensitive phospholipidbinding properties of normal and mutant ferlin C2 domains. J Biol Chem277:22883–22888.

8. Cipta S, Patel HH (2009) Molecular bandages: Inside-out, outside-in repair of cellularmembranes. Focus on “myoferlin is critical for endocytosis in endothelial cells”. Am JPhysiol Cell Physiol 297:C481–C483.

9. Davis DB, Delmonte AJ, Ly CT, McNally EM (2000) Myoferlin, a candidate gene andpotential modifier of muscular dystrophy. Hum Mol Genet 9:217–226.

10. Bernatchez PN, et al. (2007) Myoferlin regulates vascular endothelial growth factorreceptor-2 stability and function. J Biol Chem 282:30745–30753.

11. Bernatchez PN, Sharma A, Kodaman P, Sessa WC (2009) Myoferlin is critical for endo-cytosis in endothelial cells. Am J Physiol Cell Physiol 297:C484–C492.

12. Demonbreun AR, et al. (2010) Myoferlin is required for insulin-like growth factorresponse and muscle growth. FASEB J 24:1284–1295.

13. Nelson GA, Roberts TM, Ward S (1982) Caenorhabditis elegans spermatozoan locomo-tion: Amoeboid movement with almost no actin. J Cell Biol 92:121–131.

14. Amatschek S, et al. (2004) Tissue-wide expression profiling using cDNA subtraction andmicroarrays to identify tumor-specific genes. Cancer Res 64:844–856.

15. Adam PJ, et al. (2003) Comprehensive proteomic analysis of breast cancer cell mem-branes reveals unique proteins with potential roles in clinical cancer. J Biol Chem278:6482–6489.

16. Ponten F, Jirstrom K, Uhlen M (2008) The human protein atlas—a tool for pathology.J Pathol 216:387–393.

17. van’t Veer LJ, et al. (2002) Gene expression profiling predicts clinical outcome of breastcancer. Nature 415:530–536.

18. Stec J, et al. (2005) Comparison of the predictive accuracy of DNA array-based multi-gene classifiers across cDNA arrays and Affymetrix GeneChips. J Mol Diagn 7:357–367.

19. DeNardo DG, et al. (2005) Global gene expression analysis of estrogen receptor tran-scription factor cross talk in breast cancer: Identification of estrogen-induced/activatorprotein-1-dependent genes. Mol Endocrinol 19:362–378.

20. Labhart P, et al. (2005) Identification of target genes in breast cancer cells directlyregulated by the SRC-3/AIB1 coactivator. Proc Natl Acad Sci USA 102:1339–1344.

21. Iacobuzio-Donahue CA, et al. (2003) Exploration of global gene expression patterns inpancreatic adenocarcinoma using cDNA microarrays. Am J Pathol 162:1151–1162.

22. Zhao C, Putnik M, Gustafsson JA, Dahlman-Wright K (2009) Microarray analysis of al-tered gene expression in ERbeta-overexpressing HEK293 cells. Endocrine 36:224–232.

23. Kristiansen G, et al. (2005) Expression profiling of microdissected matched prostatecancer samples reveals CD166/MEMD and CD24 as new prognostic markers for patientsurvival. J Pathol 205:359–376.

24. Skubitz AP, Pambuccian S, Argenta P, Skubitz K (2006) Differential gene expressionidentifies subgroups of ovarian carcinoma. Transl Res 148:223–248.

25. Hasegawa S, et al. (2002) Genome-wide analysis of gene expression in intestinal-typegastric cancers using a complementary DNA microarray representing 23,040 genes.Cancer Res 62:7012–7017.

26. Ma C, et al. (2008) Mycobacterium tuberculosis culture supernatant induces cancer cellapoptosis and cell cycle arrest. Open Cancer J 2:31–41.

27. Bird C, Kirstein S (2009) Real-time, label-free monitoring of cellular invasion andmigration with the xCELLigence system. Nat Methods 6:v–vi Application Note.

28. Li R (2011) The expression and effect of myoferlin depletion in breast cancer cells. PhDthesis (Ohio State University, Columbus, OH).

29. Kim Y, Friedman A (2009) Interaction of tumor with its micro-environment: A math-ematical model. Bull Math Biol 72:1029–1068.

30. Kim Y, Friedman A, Wallace J, Li F, Ostrowski M (2009) Transformed epithelial cellsand fibroblasts/myofibroblasts interaction in breast tumor: A mathematical modeland experiments. J Math Biol 61:401–421.

31. Hansen R, Bissell M (2000) Tissue architecture and breast cancer: The role of extracel-lular matrix and steroid hormones. Endocr Relat Cancer 7:95–113.

32. Mueller M, Fusenig N (2004) Friends or foes—bipolar effects of the tumour stroma incancer. Nat Rev Cancer 4:839–849.

33. Zhao Y, et al. (2004) Endometase/matrilysin-2 in human breast ductal carcinoma in situand its inhibition by tissue inhibitors of metalloproteinases-2 and -4: A putative role inthe initiation of breast cancer. Cancer Res 64:590–598.

34. Shiomi T, Okada Y (2003) MT1-MMP and MMP-7 in invasion and metastasis of humancancers. Cancer Metastasis Rev 22:145–152.

35. Perumpanani A, Byrne H (1999) Extracellular matrix concentration exerts selectionpressure on invasive cells. Eur J Cancer 35:1274–1280.

36. Niiya D, et al. (2009) Identification and characterization of lutheran blood groupglycoprotein as a new substrate of membrane-type 1 matrix metalloproteinase 1(MT1-MMP). J Biol Chem 284:27360–27369.

37. Kim Y, Lawler S, Nowicki M, Chiocca E, Friedman A (2009) A mathematical model forpattern formation of glioma cells outside the tumor spheroid core. J Theor Biol260:359–371.

38. Butler GS, Dean RA, Tam EM, Overall CM (2008) Pharmacoproteomics of a metallopro-teinase hydroxamate inhibitor in breast cancer cells: Dynamics of membrane type 1matrix metalloproteinase-mediated membrane protein shedding. Mol Cell Biol28:4896–4914.

39. Yu C, et al. (2011) Myoferlin gene silencing decreases tie-2 expression in vitro andangiogenesis in vivo. Vasc Pharmacol 55:26–33.

40. Cheng J, Weiner L (2003) Tumors and their microenvironments: Tilling the soil. ClinCancer Res 9:1590–1595.

41. Jones P (2001) Extracellular matrix and tenascin-c in pathogenesis of breast cancer.Lancet 357:1992–1994.

42. Tyson R, Stern L, LeVeque R (2000) Fractional step methods applied to a chemotaxismodel. J Math Biol 41:455–475.


Published Ahead of Print 5 December 2011. 2012, 32(3):633. DOI: 10.1128/MCB.06212-11. Mol. Cell. Biol.

David Zhang, Stefan Wiemann and Özgür SahinIoanna Keklikoglou, Aoife Ward, Stefan Uhlmann, Jitao Sarah Jurmeister, Marek Baumann, Aleksandra Balwierz, and PPM1FTargeting Actin-Regulatory Proteins FHOD1Invasion of Breast Cancer Cells by MicroRNA-200c Represses Migration and

http://mcb.asm.org/content/32/3/633Updated information and services can be found at:

These include:SUPPLEMENTAL MATERIAL Supplemental material

REFERENCEShttp://mcb.asm.org/content/32/3/633#ref-list-1at:

This article cites 49 articles, 16 of which can be accessed free

CONTENT ALERTS more»articles cite this article),

Receive: RSS Feeds, eTOCs, free email alerts (when new

http://journals.asm.org/site/misc/reprints.xhtmlInformation about commercial reprint orders: http://journals.asm.org/site/subscriptions/To subscribe to to another ASM Journal go to:

on May 31, 2013 by guest

http://mcb.asm

.org/D

ownloaded from

MicroRNA-200c Represses Migration and Invasion of Breast CancerCells by Targeting Actin-Regulatory Proteins FHOD1 and PPM1F

Sarah Jurmeister,* Marek Baumann,* Aleksandra Balwierz, Ioanna Keklikoglou, Aoife Ward, Stefan Uhlmann, Jitao David Zhang,*Stefan Wiemann, and Ozgür Sahin

Division of Molecular Genome Analysis, German Cancer Research Center (DKFZ), Heidelberg, Germany

MicroRNA-200c (miR-200c) has been shown to suppress epithelial-mesenchymal transition (EMT), which is attributed mainlyto targeting of ZEB1/ZEB2, repressors of the cell-cell contact protein E-cadherin. Here we demonstrated that modulation ofmiR-200c in breast cancer cells regulates cell migration, cell elongation, and transforming growth factor � (TGF-�)-inducedstress fiber formation by impacting the reorganization of cytoskeleton that is independent of the ZEB/E-cadherin axis. We iden-tified FHOD1 and PPM1F, direct regulators of the actin cytoskeleton, as novel targets of miR-200c. Remarkably, expression lev-els of FHOD1 and PPM1F were inversely correlated with the level of miR-200c in breast cancer cell lines, breast cancer patientsamples, and 58 cancer cell lines of various origins. Furthermore, individual knockdown/overexpression of these target genesphenocopied the effects of miR-200c overexpression/inhibition on cell elongation, stress fiber formation, migration, and inva-sion. Mechanistically, targeting of FHOD1 by miR-200c resulted in decreased expression and transcriptional activity of serumresponse factor (SRF), mediated by interference with the translocation of the SRF coactivator mycocardin-related tran-scription factor A (MRTF-A). This finally led to downregulation of the expression and phosphorylation of the SRF targetmyosin light chain 2 (MLC2) gene, required for stress fiber formation and contractility. Thus, miR-200c impacts on metas-tasis by regulating several EMT-related processes, including a novel mechanism involving the direct targeting of actin-regulatory proteins.

Expression of miR-200 family members is frequently down-regulated in metastases compared to that in primary tumors

(11, 18, 30), and reduced miR-200 levels are associated with a pooroutcome in several human epithelial malignancies (16, 47, 49).Furthermore, overexpression of miR-200 was demonstrated tosuppress metastasis in mouse models of lung adenocarcinoma andbreast cancer (1, 11). Metastasis-suppressing effects of miR-200family members have thus far been attributed mostly to their abil-ity to inhibit epithelial-mesenchymal transition (EMT), a processthat is thought to be central in the metastatic progression of manycancer types (42). This has been shown to be mediated via miR-200-induced downregulation of the transcriptional repressorsZEB1 and SIP1/ZEB2 (13, 22, 31). While targeting of ZEB1 andZEB2 by miR-200 and the resulting upregulation of E-cadherinwere shown to contribute to inhibition of motility (20), reexpres-sion of E-cadherin by targeting both ZEB1 and ZEB2 was insuffi-cient to fully reverse EMT, as characterized by failed remodeling ofthe actin cytoskeleton (5). Two recently identified miR-200 tar-gets, the cytoskeleton-associated protein moesin and the extracel-lular matrix protein fibronectin 1, have already been implicated inmiR-200-induced suppression of migration in one endometrialand one breast cancer cell line (15); however, the physiologicalrelevance of this mechanism still remains to be demonstrated, andadditional target genes are likely to be involved.

In this study, we demonstrated that miR-200c, the predomi-nant member of the miR-200 family (13, 17, 47), can inhibit mi-gration and invasion of breast cancer cells in a ZEB1/ZEB2-independent manner by interfering with actin cytoskeletalorganization. Using a combination of genome-wide expressionprofiling and computational and molecular biology approaches,we identified the actin-regulatory proteins formin homology 2domain containing 1 (FHOD1) and protein phosphatase, Mg2�/Mn2� dependent, 1F (PPM1F) as novel direct targets of miR-200c

and demonstrated that they contribute to miR-200c-induced in-hibition of migration and invasion through regulation of stressfiber formation and function by modulating several downstreammediators.

MATERIALS AND METHODSCell culture and growth factor stimulation. Two human breast cancercell lines (MDA-MB-231 and MCF-7) were obtained from the Amer-ican Type Culture Collection (Manassas, VA). Culturing media andsupplements for the two cancer cell lines were described previously(33). For stimulation with transforming growth factor � (TGF-�),cells were starved in serum-free medium for 24 h and subsequentlytreated with 10 ng/ml TGF-� (Peprotech, Rocky Hill, NJ) for 5 h.HEK293FT cells were grown in D-MEM high-glucose medium (Invit-rogen, Carlsbad, CA) containing 10% fetal bovine serum (FBS), 100U/ml penicillin-streptomycin, and 500 �g/ml Geneticin. Transfectionand starvation media were deprived of penicillin-streptomycin andFBS, respectively.

Received 31 August 2011 Returned for modification 13 September 2011

Accepted 23 November 2011

Published ahead of print 5 December 2011

Address correspondence to Ozgür Sahin, [email protected], or Stefan

Wiemann, [email protected].

* Present address: S. Jurmeister, Cancer Research UK Cambridge Research

Institute, Cambridge, United Kingdom; M. Baumann, Medical Theoretical Center,

Technische Universität Dresden, Dresden, Germany; J. D. Zhang, Computational

Biology, Translational Research Sciences, F. Hoffmann-La Roche AG,

Pharmaceutical Research and Early Development, Basel, Switzerland.

Supplemental material for this article may be found at http://mcb.asm.org/.

Copyright © 2012, American Society for Microbiology. All Rights Reserved.

doi:10.1128/MCB.06212-11

The authors have paid a fee to allow immediate free access to this article.

0270-7306/12/$12.00 Molecular and Cellular Biology p. 633–651 mcb.asm.org 633


http://mcb.asm

.org/D

ownloaded from

Transfection with siRNAs, miRNA mimics, miRNA hairpin inhibi-tors, and expression constructs. All transfections were carried out usingthe Lipofectamine 2000 transfection reagent as described previously (33).For silencing of genes of interest, either pools of four small interferingRNAs (siRNAs) per gene or individual siRNAs were used (for sequences,see Table S1 in the supplemental material). siRNAs, microRNA (miRNA)mimics (see Table S2), and miRNA hairpin inhibitors (see Table S3) (allfrom Dharmacon, Lafayette, CO) were used at final concentrations of 40,25, and 100 nM, respectively. For efficient inhibition of the miR-200bc/429 cluster, equal amounts of inhibitors directed against miR-200c andmiR-429 were combined. Expression vectors for FHOD1 (pCMV5-FHOD1-HA) and PPM1F (pCDNA-Dest47-PPM1F) open readingframes (ORFs), as well as respective empty-vector controls (pCMV6 andpCDNA-Dest47), were transfected at 200 ng per well of a six-well plate.

Cell lysis and Western blotting. Preparation of protein lysates andWestern blotting were done as previously described (33). Fifteen micro-grams of total protein was separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and exposed to primary antibodies(see Table S4 in the supplemental material). Horseradish peroxidase-conjugated secondary antibodies were from Santa Cruz Biotechnology(Santa Cruz, CA).

Immunofluorescence staining and microscopy. Cells were seeded onsquare coverslips in six-well plates and transfected as described above.Subsequently, cells were fixed with 2% paraformaldehyde for 15 min. Forpermeabilization, 0.2% Triton X-100 in phosphate-buffered saline (PBS)was applied for 5 min. Specimens were blocked for 30 min with 3% bovineserum albumin (BSA)-PBS. For detection of pMLC2 (Thr18/Ser19), spec-imens were incubated with anti-pMLC2 antibody (1:250; Cell SignalingTechnology) overnight at 4°C. Alexa Fluor 647-labeled secondary anti-body (Invitrogen) was diluted 1:1,000 and applied for 1 h at room tem-perature. For staining of filamentous actin, cells were incubated with Al-exa Fluor 488-phalloidin (1:40; Invitrogen) for 30 min at roomtemperature. For nuclear staining, cells were treated with 1 �g/�l of 4=,6-diamidino-2-phenylindole (DAPI) for 10 min. All antibodies and stainingreagents were diluted in 3% BSA-PBS, and cells were washed three timeswith PBS after each step of the staining procedure. Coverslips weremounted with ProLong Gold antifade reagent. Images were taken using aZeiss LSM 510 Meta confocal microscope.

Plasmid construction and site-directed mutagenesis. The 3= un-translated regions (3=-UTRs) of the FHOD1 and PPM1F genes were am-plified by PCR using genomic DNA of the MDA-MB-231 cell line andcloned downstream of the Renilla luciferase open reading frame in thepsiCHEK-2 vector (Promega) using XhoI and NotI restriction sites.FHOD1 primers (GCGCGCCTCGAGGAGAAGGACGTTGAAGAGTGand GCGCGCGCGGCCGCGGATTAGCTAAGAAAATTTTATTTTG)were designed to bind to positions 17806 to 17825 and 18107 to 18131 ofthe FHOD1 genomic sequence (according to NC_000016.9); PPM1Fprimers (GCGCGCCTCGAGAGGCTCCACCAAGAAGCTA and GCGCGCGCGGCCGCGAGTTCAGAACTTGTGGTTTATTG) were designedto bind to positions 29766 to 29785 and 33433 to 33445 of the PPM1Fgenomic sequence (according to NC_000022.10). Four point mutationswere introduced into each target site by mutagenesis PCR.

Luciferase reporter assay. For luciferase reporter assays, cells wereseeded in 96-well plates and cotransfected with either 25 nM miRNAmimics or 100 nM hairpin inhibitors together with 15 ng/wellpsiCHECK-2 reporter vectors. Forty-eight hours after transfection, lucif-erase activity was measured using the dual-luciferase reporter assay sys-tem kit (Promega) according to the manufacturer’s instructions using aTecan M200 luminescence reader. Values were double normalized to fire-fly luciferase activity and to cells transfected with empty psiCHECK-2control vectors.

Measurement of SRF activity. For the measurement of serum re-sponse factor (SRF) activity, luciferase reporter assays were performedusing the pGL4.34 [luc2P/SRF-RE/Hygro] reporter vector (Promega),which drives transcription of a luciferase reporter gene under the control

of the SRF response element (SRE). HEK293FT cells were seeded in 96-well plates and cotransfected with one of the following (per well): 20 nMsiRNAs, 25 nM miRNAs, 100 nM miRNA inhibitors, or 20 ng/well over-expression vectors, together with 100 ng/well pGL4.34 and 10 ng/wellpRL-TK vector (Promega). Twenty-four hours after transfection, cellswere starved for 24 h and stimulated with 10 ng/ml TGF-� for 6 h beforecell lysis. Luciferase activity was measured as described above. Values werenormalized to Renilla luciferase activity.

Wound-healing assay. Wound-healing assays were carried out usingmigration culture dish inserts from Ibidi (Martinsried, Germany). Cellswere seeded in the chambers of the culture dish insert and transfected.Forty-eight hours after transfection, the insert was removed and freshculture medium was added to start the migration process. Pictures wereacquired after 0 h and 8 h (in the case of MDA-MB-231) or 36 h (in thecase of MCF-7) using a Zeiss Axiovert 24 light microscope and anAxiocam MRc camera. Image analysis was performed using CellProfiler.

RTCA invasion and migration assays. RTCA (real-time cell analyzer)invasion or migration assay measures the effect of any perturbations in alabel-free real-time setting. As cells migrate or invade (when coated) fromthe upper chamber through the membrane into the bottom chamber inresponse to a chemoattractant, they contact and adhere to the electronicsensors on the underside of the membrane, resulting in an increase in theelectrical impedance. The increase in the impedance correlates with in-creasing numbers of migrated or invaded cells on the underside of themembrane (36).

For RTCA invasion or migration experiments, transfections were per-formed as described above. Cells were then starved in serum-free mediumfor 24 h and seeded in RTCA Cim-16 plates (RTCA; xCELLigence Roche,Penzberg, Germany) in serum-free medium. Full growth medium wasused as a chemoattractant in the lower chamber. Measurements wereperformed in a time-resolved manner using the RTCA device (RTCA,xCELLigence Roche, Penzberg, Germany). For invasion assays, theCim-16 plates were initially coated with Matrigel (BD Biosciences, Bed-ford, MA) diluted in serum-free medium at a ratio of 1:20. MDA-MB-231cells were stimulated to invade in the presence of TGF-�.

Matrigel invasion assay. Cells were transfected with siRNAs ormiRNAs as described above and seeded in BioCoat Matrigel invasionplates (Becton Dickinson Bioscience, Franklin Lakes, NJ). TGF-� (10 ng/ml) was added to cells in the upper chamber, and after 24 h, the number ofinvaded cells was determined by flow cytometry.

Quantitative RT-PCR of mRNAs. Total RNA was isolated using theRNeasy minikit (Qiagen), and cDNA was synthesized from RNA using theRevert Aid H Minus first-strand cDNA synthesis kit (Fermentas, St. Leon-Rot, Germany) according to the manufacturer’s instructions. For reversetranscriptase PCRs (RT-PCRs), the TaqMan Abgene universal mix(Thermo Scientific, Rockford, IL) and probes from the Universal ProbeLibrary (Roche, Penzberg, Germany) were used. Oligonucleotide primerswere synthesized at MWG (Ebersberg, Germany). HPRT (hypoxanthinephosphoribosyltransferase) and ACTB or TFRC was used as ahousekeeping-gene control. Sequences of primers and the respective Uni-versal Probe Library (UPL) probe numbers are given elsewhere (see TableS5 in the supplemental material). Data were analyzed according to the��CT method (23).

Quantitative RT-PCR of miRNAs. For miRNAs, the TaqManmicroRNA reverse transcription kit and TaqMaq gene-specific mi-croRNA assays (Applied Biosystems, Weiterstadt, Germany) were used.For the quantitative RT-PCRs (qRT-PCRs), RNU44 and RNU48 wereused as housekeeping controls. Data were acquired using a HT-7900 Taq-Man instrument (Applied Biosystems) and analyzed with the delta-deltaCT algorithm (ddCt; Bioconductor).

MRTF translocation assay. MDA-MB-231 cells were transfected withmiRNA mimics or siRNAs as described above. After washing with PBS,cells were fixed in 4% paraformaldehyde (PFA) for 20 min and permeab-ilized with 0.2% Triton X-100 in PBS for 5 min. Cells were blocked with1% BSA in PBS for 1 h at room temperature and exposed to mycocardin-

Jurmeister et al.

634 mcb.asm.org Molecular and Cellular Biology


http://mcb.asm

.org/D

ownloaded from

related transcription factor A (MRTF-A) antibody (Santa Cruz Biotech-nology). Fluorescein-conjugated Alexa Fluor 488 secondary antibody (In-vitrogen) was used. All antibodies were used at a 1:1,000 dilution inblocking buffer with 4°C overnight incubation. Nuclei were stained withDAPI dihydrochloride (Sigma-Aldrich, St. Louis, MO). Images for quan-titative analysis at magnification �20 were acquired using an OlympusScanR high-content screening microscope (Olympus, Hamburg, Ger-many) from approximately 1,000 cells for each sample and analyzed byOlympus ScanR analysis software.

Computational target prediction. For the identification of miR-200ctargets, two different software algorithms were used to find conservedtarget sites throughout mammalian transcriptomes: TargetScan release5.1 (http://www.targetscan.org) and the PITA target site catalogue fromthe Segal Lab of Computational Biology (http://genie.weizmann.ac.il).The two lists obtained from mRNA expression profiling upon miR-200bc/429 overexpression in MDA-MB-231 cell lines (43) and in silicocomputational target predictions were compared. The intersection ofboth resulted in a list of 34 common genes as potential targets of miR-200c, depicted in Table S6 in the supplemental material.

Correlation analysis. Expression data for 101 human primary breasttumors was obtained from a public data set deposited in the NCBI GEOdatabase (GEO accession no. GSE19783 [8]). miR-200c, FHOD1, andPPM1F expression data for 11 breast cancer cell lines (SK-BR-3, BT474,ZR7530, HCC1419, MCF-7, T47D, BT483, HCC202, UACC812, MDA-MB-231, and MDA-MB-468) and two mammary epithelial cell lines(MCF-10A and MCF-12A) were determined by qRT-PCR as describedabove. Expression data sets for FHOD1, PPM1F, and miR-200c in the celllines of the NCI60 panel were obtained from http://dtp.nci.nih.gov/index.html (experiment IDs 372534, 124184, and 27785, respectively). Pearsoncorrelation coefficients (r) and two-tailed P values of normalized, log2-transformed data sets were computed in the R statistical software pro-gram.

To study the association of microRNA and target gene expressionlevels with epithelial or mesenchymal characteristics, mammary epithelialand breast cancer cell lines were divided into epithelial, mesenchymal, andundefined groups based on data on their phenotypes, invasive behaviorsand EMT marker gene expression (2, 24, 29, 37) and on E-cadherinmRNA expression. BT474, ZR7530, MCF-7, T47D, BT483, HCC202, andUACC812 were scored as epithelial in accordance with all available data.SK-BR-3 cells and MDA-MB-468 cells are negative for the mesenchymalmarkers vimentin and fibronection (2) but express low levels ofE-cadherin (2; also data not shown) and have been described to haveeither an epithelial cell-like or intermediate phenotype (24, 37); they werethus placed in the undefined category. For HCC1419 cells, insufficientinformation was available to make a confident decision; they were thusalso included in the undefined category. For the NCI60 panel, sorting ofepithelial, mesenchymal, and undefined subclusters was based on data byPark and colleagues (31), where the ratio of E-cadherin expression tovimentin expression was determined in these cell lines.

Statistical analysis. A two-tailed Student t test was used to estimateintergroup differences if not otherwise stated (indicated in figures as fol-lows: �, P � 0.05; ��, P � 0.01; ��, P � 0.001).

RESULTSOverexpression of miR-200c in metastatic breast cancer cellsinhibits invasion, migration, cell polarization, and stress fiberformation. To study the role of the miR-200 family in cell inva-sion and motility, we first assessed the invasion capacity of meta-static MDA-MB-231 breast cancer cells after transfection with amiR-200c mimic in real time and in endpoint invasion assaysusing the RTCA (real-time cell analyzer) and Matrigel assays, re-spectively. Consistent with earlier observations (3), overexpres-sion of miR-200c had a strong inhibitory effect on the invasion ofthe MDA-MB-231 cell line in both assays (Fig. 1A and B). Simi-larly, overexpression of miR-200c also inhibited migration of

MDA-MB-231 cells as assessed by a wound-healing assay (Fig.1C). Fluorescence microscopy results revealed that upon transfec-tion of miR-200c mimic, the loss of migratory capability was ac-companied by a loss of cell polarization (Fig. 1D, left). To quantifythis effect, the ratios of long to short axes of cells transfected witheither control mimic or miR-200c mimic were calculated. Elon-gation of cells transfected with miR-200c mimic was significantlyless than that of control cells (two-sided t test, P � 0.001) (Fig. 1D,right). Furthermore, actin cytoskeletal structures that are charac-teristic of mesenchymal cell migration, namely, the actin networkat the leading edge and long fibers spanning the cell body, werereplaced by a diffuse distribution of actin staining. It has beenshown that overexpression of miR-200c induces apoptosis (34,43); however, the observed effect on migration is not due to apop-tosis, since decreased cell viability became detectable only 48 h ormore after transfection (43). Furthermore, the nuclei of miR-200c-transfected cells were intact as visualized by DAPI stainings(Fig. 1D, left).

To further investigate the effect of miR-200c on cytoskeletalreorganization, we next tested the ability of MDA-MB-231 cells toform stress fibers, since these provide the contractile force re-quired for mesenchymal cell-like motility. Upon stimulation withTGF-�, which has been demonstrated to induce stress fibersthrough activation of the RhoA signaling pathway (6, 35), themajority of control-transfected MDA-MB-231 cells formedprominent stress fibers (Fig. 1E). However, in cells transfectedwith miR-200c mimic, actin filaments remained diffusely distrib-uted in the cytoplasm. Analysis of activated RhoA levels in MDA-MB-231 cells by GTPase activation assays revealed that overex-pression of miR-200c had no effect on RhoA activation (data notshown). This indicated that the regulatory effect of miR-200c onstress fiber formation was likely to act downstream of RhoA, po-tentially on the level of actin-regulatory effector proteins, ratherthan on upstream signaling events.

To further validate our observations with another cell linemodel, we inhibited miR-200c in noninvasive MCF-7 breast can-cer cells, which express high endogenous levels of this microRNA,by transfection with hairpin inhibitors. Upon stimulation withTGF-� for 5 h, cells transfected with the miR-200c hairpin inhib-itors indeed adopted a more elongated shape than control cells(Fig. 1F). Moreover, inhibition of miR-200c increased the abilityof MCF-7 cells to form stress fibers (Fig. 1G). Taken together,these results indicate that the modulation of miR-200c levels im-pinges on cell invasion, migration, elongation, and stress fiberformation in breast cancer cells.

Actin-regulatory proteins FHOD1 and PPM1F are direct tar-gets of miR-200c. ZEB1 and ZEB2 are established direct targets ofmiR-200c, and these transcription factors impact on cell-cell con-tact formation as well as cell migration (13, 22, 31). However,stress fiber organization was entirely unperturbed by depletion ofthese miR-200c target genes (Fig. 2). We thus reasoned that miR-200c would target additional genes mediating its effect on actincytoskeletal organization. To identify novel candidate targets ofmiR-200c, we used genome-wide mRNA expression profiling datathat we had obtained from MDA-MB-231 cells transfected withmimics of the miR-200bc/429 cluster sharing the same seed se-quence (43). We found 160 genes that were strongly downregu-lated by miR-200c overexpression (Fig. 3A). To identify directmiR-200c target genes, the data were merged with a list of poten-tial miR-200bc/429 targets predicted by two microRNA target

Regulation of Actin Reorganization by miR-200c

February 2012 Volume 32 Number 3 mcb.asm.org 635


http://mcb.asm

.org/D

ownloaded from



http://mcb.asm

.org/D

ownloaded from

prediction tools, yielding 34 candidate target genes (see Table S6in the supplemental material). Among those, the formin homol-ogy domain-containing protein 1 (FHOD1) and Mg2�/Mn2�-dependent protein phosphatase 1F (PPM1F) were of particularinterest due to their described functions in the regulation of actincytoskeletal organization (9, 10, 19, 46, 48). The 3=-UTRs ofFHOD1 and PPM1F are predicted to contain one and two miR-200c target sites, respectively, which all have a high degree of con-servation among mammalian species (Fig. 3B).

To validate regulation of FHOD1 and PPM1F by miR-200c, wefirst tested whether overexpression of miR-200c in MDA-MB-231cells would have an effect on FHOD1 and PPM1F protein levels.To demonstrate the specificity of the antibodies and to evaluatethe magnitude of the changes in FHOD1 and PPM1F protein lev-els after transfection of miR-200c mimic, nontargeting siRNA andsiRNAs directed against FHOD1 or PPM1F were used as controls.The siRNAs significantly reduced levels of their target proteinsafter 24 h, and transfection of miR-200c mimic also led to a sig-nificant reduction in both FHOD1 and PPM1F protein levelswithin the same time frame (Fig. 4A). Conversely, inhibition ofmiR-200bc/429 cluster miRNAs in MCF-7 cells by transfection ofmicroRNA hairpin inhibitors increased FHOD1 and PPM1F pro-tein levels. These findings were also validated at the mRNA levelupon miR-200c overexpression (in MDA-MB-231) or inhibition

(in MCF-7) (Fig. 4B), confirming mRNA degradation mecha-nisms for miR-200c targeting of these two genes.

To validate FHOD1 and PPM1F as direct targets of miR-200c,we next performed luciferase reporter assays combined with site-directed mutagenesis. For this purpose, the 3=-UTRs of the candi-date genes were cloned downstream of the Renilla luciferase ORFin a dual-luciferase reporter vector and cotransfected with controlor miR-200c mimics into MDA-MB-231 cells. Relative luciferaseactivity was significantly reduced for both candidate genes’ 3=-UTRs (Fig. 4C), indicating that FHOD1 and PPM1F are potentialdirect targets of miR-200c. Mutations in the predicted miR-200ctarget sites abrogated inhibition by miR-200c mimic, confirmingthe functionality of these target sites. We could further validatethese results in MCF-7 cells (Fig. 4D) and in an unrelated nontu-mor cell line, HEK-293FT (Fig. 4E), showing that the observedeffects are not restricted to our model systems. Finally, inhibitionof miR-200c in MCF-7 cells resulted in an increase in relativeluciferase activity (Fig. 4F). In conclusion, the results of the lucif-erase assay confirmed that FHOD1 and PPM1F are indeed noveldirect targets of miR-200c.

Expression of miRNA-200c targets, FHOD1 and PPM1F,correlates with a mesenchymal cell-like phenotype in breast andother cancer cell lines. To determine if targeting of FHOD1 andPPM1F is also relevant in vivo and particularly in the context of

FIG 1 miR-200c regulates invasion, migration, elongation, and stress fiber formation in breast cancer cells. (A) An RTCA (real-time cell analyzer) invasion assayof MDA-MB-231 cells transfected with microRNA mimics. MDA-MB-231 cells were transfected, starved in serum-free medium for 24 h, and seeded in RTCACIM-16 plates covered with Matrigel. Cells were stimulated to invade in the presence of TGF-� (10 ng/ml), and impedance measurements were performed in atime-resolved manner. Means for four replicates � standard deviations are shown; a t test was performed for the last time point. (B) Matrigel invasion assay ofMDA-MB-231 cells transfected with microRNA mimics. Cells were transfected, seeded in Matrigel-coated invasion plates, and stimulated to invade in thepresence of TGF-� (10 ng/ml). Invaded cells were quantified by flow cytometry. (C) Wound-healing assay of MDA-MB-231 cells transfected with microRNAmimics. MDA-MB-231 cells were seeded in migration chambers and transfected with 25 nM control or miR-200c mimic. Images of the migration area werecaptured 0 h and 8 h after removal of the migration chamber. For migration area quantification, the cell-free area of pictures was defined manually and filled withblack for a subsequent gray-black selection process performed by the CellProfiler software program (n � 2). (D) Analysis of cell elongation by fluorescencemicroscopy in MDA-MB-231 cells. Cells were transfected with control mimic or miR-200c mimic and fixed 24 h after transfection. Actin filaments werevisualized by staining with Alexa Fluor 488-phalloidin (green), and cell nuclei were stained with DAPI (blue). Boxes in the upper left corners of the imagesillustrate cell morphology with higher resolution. Cell elongation was quantified by measuring the long and short axes of cells using the Zeiss LSM ImageExaminer software. Box plots represent data from two independent experiments, with 40 cells measured for each condition. (E) TGF-�-induced stress fiber assayin MDA-MB-231 cells. Cells were transfected with control mimic or miR-200c mimic and starved for 24 h. Stress fibers were induced by treatment with TGF-�for 5 h, and cells were stained with Alexa Fluor 488-phalloidin and DAPI. Boxes in the upper left corners of the images demonstrate stress fibers with higherresolution. The percentage of stress fiber-containing cells was determined by counting 200 to 300 cells per experiment; data shown represent means for threeindependent experiments. (F) Analysis of cell elongation by fluorescence microscopy in MCF-7 cells. Cells were transfected with control microRNA hairpininhibitor or miR-200c hairpin inhibitor, starved for 24 h starting 2 days after transfection, and stimulated with TGF-� for 5 h. Staining, microscopy, and imageanalysis were carried out as for panel D. (G) Stress fiber assay in MCF-7 cells. Cells were transfected with control microRNA hairpin inhibitor or miR-200c hairpininhibitor and starved for 24 h starting 2 days after transfection. Stress fibers were induced by treatment with TGF-� for 5 h. Staining, microscopy, and imageanalysis were carried out as described for panel E.

FIG 2 Silencing of ZEB1 or ZEB2 does not interfere with stress fiber formation. (A) MDA-MB-231 cells were transfected with control siRNA (siAllStar) or siRNAdirected against ZEB1 or ZEB2 and starved for 24 h. Stress fibers were induced by treatment with TGF-� for 5 h, and cells were stained for actin with Alexa Fluor488-phalloidin (green), with the nucleus stained with DAPI (blue). (B) The percentage of stress fiber-containing cells was determined by counting 200 to 300 cellsper experiment; data shown represents means of data from two independent experiments.




http://mcb.asm

.org/D

ownloaded from

breast cancer, we performed a correlation analysis between theexpression of miR-200c and its targets using expression data frombreast cancer patients. For this purpose, publicly available mRNAand microRNA microarray data for 101 primary breast tumorswas obtained from the NCBI GEO database (8) (GEO accessionno. GSE19783). Pearson correlation analysis showed a significantnegative correlation between FHOD1 and miR-200c expressionlevels (r � �0.32; P � 9.81e�4) and between PPM1F and miR-200c expression levels (r � �0.45; P � 2.66e�6) (Fig. 5A). Thissuggests that miR-200c might also contribute to the regulation ofFHOD1 and PPM1F in breast cancer patients.

We then compared the expression levels of miR-200c and itstarget genes in 13 different breast cancer and mammary epithelialcell lines. Consistent with the observations made with clinicalsamples, FHOD1 and PPM1F expression levels were negativelycorrelated with miR-200c levels in these cell lines as well (r ��0.61 and P � 0.03, and r � �0.64 and P � 0.02, respectively)(Fig. 5B). Several studies have compared the expression levels ofmesenchymal and epithelial marker genes in different breast can-cer cells lines and linked a mesenchymal phenotype to increasedmotility and invasiveness (2, 24, 29, 37). Making use of the dataprovided by these studies, we divided the 13 cell lines into epithe-lial cell-like (n � 8), mesenchymal cell-like (n � 3), and undefined(n � 2) groups. Consistent with its role in the suppression ofEMT, miR-200c was expressed at higher levels in cell lines with anepithelial phenotype than in cell lines having a mesenchymal phe-notype. Conversely, both FHOD1 and PPM1F expression levelswere significantly higher in mesenchymal cell lines than in epithe-lial ones (Fig. 5C).

Intrigued by the finding that expression of miR-200c and itstarget genes FHOD1 and PPM1F inversely correlate both in breast

cancer specimens and in breast cancer cell lines, we were inter-ested in testing whether the negative correlation of miR-200c withFHOD1 and PPM1F expression levels would also hold true be-yond breast cancer. The NCI60 panel (http://dtp.nci.nih.gov/index.html) of the U.S. National Cancer Institute encompasses60 human cancer cell lines representing 9 different tumor types.We retrieved NCI60 expression data sets for miR-200c, FHOD1,and PPM1F. Again, FHOD1 and PPM1F showed a significant neg-ative correlation with miR-200c levels (r � �0.30 and P � 0.02;r � �0.35 and P � 0.01) (Fig. 5D). Moreover, Park and colleagues(31) had previously demonstrated that the NCI60 panel can bedivided into an epithelial group, a mesenchymal group, and anundefined subcluster according to their ratio of E-cadherin ex-pression to vimentin expression and that the expression of miR-200 family members is strongly associated with the epithelial sub-cluster. Using the subclusters defined by Park et al., we found thatboth miR-200c targets FHOD1 and PPM1F were significantlymore highly expressed in cell lines with a mesenchymal phenotype(Fig. 5E), confirming and extending our findings to a wide panelof cancer cell lines.

Next, the correlation of FHOD1 and PPM1F expression a withmesenchymal cell-like phenotype prompted us to ask if there is across talk between these two novel miR-200c target genes and theregulation of EMT by known miR-200c targets ZEB1/ZEB2. Thus,we tested whether silencing of ZEB1 in MDA-MB-231 cells wouldhave any effect on FHOD1 or PPM1F expression. Although theknockdown of ZEB1 was very efficient, neither FHOD1 norPPM1F protein levels were affected (Fig. 6A). Furthermore, wetested potential effects of silencing FHOD1 or PPM1F on ZEB1,ZEB2, and E-cadherin expression. As expected, overexpression ofmiR-200c in MDA-MB-231 cells led to a strong induction of

FIG 3 Identification of potential miR-200c target genes. (A) mRNA expression profiling of MDA-MB-231 cells transfected with miR-200c mimic was performedusing the Illumina HumanWG-6 v3.0 expression array chip (43). miR-200c targets were predicted using the TargetScan release 5.1 and PITA target predictionprograms, and the lists of downregulated genes and predicted targets were merged, resulting in 34 genes as candidates for direct targeting by miR-200c. (B)miR-200c target sites in the 3=-UTRs of FHOD1 and PPM1F and interspecies conservation of seed matching sequences (gray box).

Jurmeister et al.



http://mcb.asm

.org/D

ownloaded from

FIG 4 Validation of FHOD1 and PPM1F as direct targets of miR-200c. (A) MDA-MB-231 cells were transfected with microRNA mimics or siRNAs, and totalprotein was isolated after 24 h. MCF-7 cells were transfected for 48 h with microRNA hairpin inhibitors prior to protein isolation. FHOD1 and PPM1F weredetected by Western blotting; �-actin was used as a loading control. (B) MDA-MB-231 cells were transfected with microRNA mimics or siRNAs for 24 h, andMCF-7 cells were transfected with microRNA hairpin inhibitors for 48 h. RNA was isolated, and FHOD1 and PPM1F transcript levels were quantified byqRT-PCR. ACTB and HPRT were used as housekeeping-gene controls. (C) MDA-MB-231 cells were cotransfected with microRNA mimics and psiCHECK-2expression constructs containing the 3=-UTR of FHOD1 (n � 3) or PPM1F (n � 4) downstream of the Renilla luciferase gene. Forty-eight hours after transfection,luciferase activity was measured. Renilla luciferase activity was normalized first to firefly luciferase activity and then to the values measured for the parental vectorpsiCHECK-2. In the mutated constructs, the miR-200c target sites were disrupted by site-directed mutagenesis of four nucleotides within the seed sequence. (D)Luciferase assay with MCF-7 cells was carried out in the same manner as with MDA-MB-231 cells (n � 3 for FHOD1 3=-UTR; n � 2 for PPM1F 3=-UTR). (E)Luciferase assay with HEK-293FT cells was carried out in the same manner as with MDA-MB-231 cells (n � 2 for FHOD1 3=-UTR and PPM1F 3=-UTR). (F)MCF-7 cells were cotransfected with luciferase reporter constructs as before (n � 3 for FHOD1 3=UTR and PPM1F 3=-UTR) and microRNA hairpin inhibitors.Forty-eight hours after transfection, luciferase activity was measured.



http://mcb.asm

.org/D

ownloaded from

FIG 5 ExpressioncorrelationofFHOD1andPPM1FwithmiR-200candwithepithelial-mesenchymalstatesofthecells. (A)mRNAandmicroRNAprofilingdatafor101breastcancerspecimenswasobtainedfromtheNCBIGEOdatabase(GEOaccessionno.GSE19783);expressionlevelswereprovidedaslog2-transformed,normalizeddatasets.Pearsoncorrelation coefficients (r) and P values (two-tailed) were computed. Dashed lines indicate a first-order linear model fitting the data. (B) Transcript levels of miR-200c, FHOD1,and PPM1F in 13 breast cancer and mammary epithelial cell lines were quantified by qRT-PCR. Expression levels were normalized to data for MCF-7 cells and log2 transformed.PearsoncorrelationcoefficientsandPvalueswerecomputed.(C)Breastcancercell linesweredividedintoepithelial,undefined,andmesenchymalphenotypesbasedonpublisheddata(2,24,29,37),andexpressionlevels frompanelBwerecomparedbetweenthegroups.(D)Datasets fortheexpressionofmiR-200c,FHOD1,andPPM1FintheNCI60panelof cancer cell lines were obtained from http://dtp.nci.nih.gov/index.html (experiment IDs 372534, 124184, and 27785). Data were log2 transformed, and Pearson correlationcoefficientsandPvalueswerecomputed.(E)NCI60cell lineswereclassifiedasepithelial,undefined,andmesenchymalaccordingtoreference31.ExpressionlevelsofFHOD1andPPM1F were compared between the groups using the data from panel D.

Jurmeister et al.



http://mcb.asm

.org/D

ownloaded from

E-cadherin expression (Fig. 6B), and ZEB1 and ZEB2 mRNA lev-els were reduced by 80% (Fig. 6C and D). However, knockdown ofFHOD1 or PPM1F did not affect the expression of any of thesethree genes. This shows that although high levels of FHOD1 andPPM1F are characteristics of a mesenchymal cell-like phenotype,their expression is not regulated by ZEB1/ZEB2 transcription fac-tors and vice versa. Hence, we concluded that miR-200c couldregulate migration and invasion through two separate pathwaysthat do not influence each other, at least at the transcriptionallevel.

Silencing of FHOD1 or PPM1F mimics the effects of miR-200c overexpression on invasion, migration, cell polarization,and stress fiber formation. To assess if targeting of FHOD1 andPPM1F could contribute to miR-200c-induced phenotypes, wenext investigated whether individual knockdown of FHOD1 orPPM1F would phenocopy the effects of miR-200c expression oninvasion, cell migration, polarization, and stress fiber formation.Indeed, individual silencing of FHOD1 or PPM1F resulted in adecrease in the invasion of MDA-MB-231 cells as measured bothby real-time (Fig. 7A) and Matrigel (Fig. 7B) invasion assays, with-out affecting the viability of the cells (Fig. 7C). Furthermore,knockdown of FHOD1 or PPM1F also reduced migration of cellsin wound-healing assays (Fig. 7D). To exclude that these effectsare potentially caused by off-target effects of the siRNA pools usedfor gene silencing, we then tested the knockdown efficiency of thefour individual siRNAs directed against FHOD1 or PPM1F byqRT-PCR. All individual siRNAs except siFHOD1 3 and siPPM1F3 strongly reduced their respective target gene’s expression (Fig.7E). We then chose two individual siRNAs for each gene and

tested their effects on cell motility compared to that of siRNApools in a real-time migration assay (Fig. 7F). Importantly, theindividual siRNAs against FHOD1 or PPM1F reduced cell migra-tion to a degree similar to that of the respective siRNA pools,supporting the specificity of the observed effects.

We next investigated whether simultaneous silencing of bothFHOD1 and PPM1F would be sufficient to rescue the activatoreffect on migration induced by a miR-200c inhibitor. For thispurpose, we transfected MCF-7 cells with a miR-200c hairpin in-hibitor and assessed their migratory capacity in real-time migra-tion and wound-healing assays (Fig. 8A and B). As expected, in-hibition of miR-200c stimulated migration of MCF-7 cells.However, when FHOD1 and PPM1F were silenced in the presenceof the miR-200c inhibitor, this stimulatory effect on migrationwas completely abrogated (Fig. 8A and B), highlighting the essen-tial role of these two target genes in miR-200c-mediated regula-tion of migration.

We then asked whether targeting of FHOD1 or PPM1F couldalso explain the observed effects of miR-200c on the actin cyto-skeleton. Indeed, upon silencing of FHOD1 or PPM1F, MDA-MB-231 cells lost their elongated shape and instead adopted arounded shape that strongly resembled that of cells overexpress-ing miR-200c (Fig. 9B). Moreover, after stimulation with TGF-�,the fraction of MDA-MB-231 cells that formed visible stress fiberswas reduced by 2-fold when they were treated with siRNA againstFHOD1 or PPM1F compared to results for control-transfectedcells (Fig. 9C). In order to validate our results in a complementaryapproach with MCF-7 cells, we employed expression constructsfor FHOD1 and PPM1F; protein expression was validated by

FIG 6 Effects of FHOD1 and PPM1F on the actin cytoskeleton are independent of the ZEB/E-cadherin axis. (A) MDA-MB-231 cells were transfected withcontrol siRNA (siAllStar) or siRNA directed against ZEB1. Forty-eight hours after transfection, protein was isolated and expression levels of ZEB1, FHOD1, andPPM1F were assessed by Western blotting. �-Actin was used as a loading control. (B to D) MDA-MB-231 cells were transfected for 24 h with microRNA mimicsor siRNAs. RNA was isolated, and E-cadherin, ZEB1, and ZEB2 transcript levels were quantified by qRT-PCR. ACTB and HPRT were used as housekeeping-genecontrols.




http://mcb.asm

.org/D

ownloaded from

Jurmeister et al.



http://mcb.asm

.org/D

ownloaded from

Western blotting (Fig. 9A). Overexpression of either FHOD1 orPPM1F in MCF-7 cells was sufficient to induce increased cell elon-gation and polarization upon TGF-� treatment, mimicking theeffects of miR-200c inhibition in this cell line (Fig. 9D). Moreover,overexpression of FHOD1 or PPM1F also increased the ability ofMCF-7 cells to form stress fibers (Fig. 9E). Altogether, these datasuggest that targeting of either FHOD1 or PPM1F by miR-200ccontributes to the effects of miR-200c on migration-related pro-cesses.

Regulation of MLC2 by miR-200c and its target genes. Al-though silencing of both FHOD1 and PPM1F was sufficient toinduce both a loss of polarity and elongation (Fig. 9), these genesare unlikely to be the sole contributors to this aspect of the miR-200c overexpression-induced phenotypes. Other target genes,namely, WAVE3 and MARCKS (7, 38), have been proposed to also

have an impact on the loss of polarity observed upon miR-200overexpression. In contrast, the regulation of stress fibers by miR-200c via targeting of FHOD1 and PPM1F among all miR-200ctarget genes identified so far is unique. We thus focused our effortson elucidating the molecular mechanisms underlying the effect ofmiR-200c and its target genes on stress fiber formation. FHOD1 isproposed to act as an actin nucleating protein, directly inducingthe formation of actin filaments (41). Besides actin nucleation, thephosphorylation of myosin light chain 2 (MLC2) at Thr18 andSer19 is a second key process in stress fiber formation, which fa-cilitates the assembly of myosin into bipolar filaments. Myosinfilaments are shown to contribute to stress fiber formation bycross-linking actin filaments. Furthermore, they are required forstress fiber function during migration, since they provide contrac-tile activity (44).

FIG 7 Individual silencing of FHOD1 or PPM1F phenocopies the effect of miR-200c on invasion and migration. (A) RTCA invasion assay of MDA-MB-231 cellstransfected with siRNAs. MDA-MB-231 cells were transfected, starved in serum-free medium for 24 h, and seeded in RTCA CIM-16 plates covered with Matrigel.Cells were allowed to invade in the presence of TGF-� (10 ng/ml), and impedance measurements were performed in a time-resolved manner. Mean for fourreplicates � standard deviations are shown; a t test was performed for the last time point. (B) Matrigel invasion assay of MDA-MB-231 cells transfected withmicroRNA mimics. Cells were transfected, seeded in Matrigel-coated invasion plates, and stimulated to invade in the presence of TGF-� (10 ng/ml). Invaded cellswere quantified by flow cytometry. (C) Viability assay of MDA-MB-231 cells. Cells were transfected with microRNA mimics or with siRNAs. Cell viability wasmeasured using the Cell Titer Glo assay 72 h posttransfection. Viability values are normalized to control mimic for miR-200c and to siAllStar for siRNAs. (D)Wound-healing assay of MDA-MB-231 cells transfected with siRNAs. MDA-MB-231 cells were seeded in migration chambers and transfected with 40 nMsiRNA. Images of the migration area were captured 0 h and 8 h after removal of the migration chamber. Migration area quantification was done as for Fig. 1C.(E) Effect of individual siRNAs on FHOD1 and PPM1F transcript levels. MDA-MB-231 cells were transfected with 40 nM either individual siRNAs or a pool of4 siRNAs directed against FHOD1 or PPM1F. After 48 h, RNA was isolated, and FHOD1 and PPM1F transcript levels were quantified by qRT-PCR. ACTB andHPRT were used as housekeeping-gene controls. (F) RTCA migration assay of MDA-MB-231 cells transfected with siRNAs. MDA-MB-231 cells were transfected,starved in serum-free medium for 24 h, and seeded in RTCA CIM-16 plates. Full medium was used as a chemoattractant, and impedance measurements wereperformed in a time-resolved manner.

FIG 8 Silencing of FHOD1 and PPM1F is sufficient to prevent the stimulatory effect of miR-200c inhibition on migration. (A) RTCA migration assay of MCF-7cells transfected with miRNA inhibitors together with siRNAs. MCF-7 cells were transfected, starved in serum-free medium for 24 h, and seeded in RTCA CIM-16plates. Full medium was used as a chemoattractant, and impedance measurements were performed in a time-resolved manner. Means for four replicates �standard deviations are shown; a t test was performed for the last time point. (B) Wound-healing assay of MCF-7 cells transfected with microRNA inhibitors andsiRNAs. MCF-7 cells were seeded in migration chambers and transfected with microRNA inhibitors and siRNAs. Images of the migration area were captured 0h and 36 h after removal of the migration chamber. For migration area quantification, the cell-free areas of the pictures were defined manually and filled withblack for the subsequent gray-black selection process performed by CellProfiler software (n � 5 to 8).




http://mcb.asm

.org/D

ownloaded from

Thus, we investigated whether miR-200c and its target genesFHOD1 and PPM1F could affect MLC2 phosphorylation. For thispurpose, we induced stress fibers in MDA-MB-231 cells by treat-ment with TGF-� after transfection with microRNA mimics orsiRNAs and detected MLC2 by immunofluorescence using an an-tibody specific for the diphosphorylated (pThr18/pSer19) protein(Fig. 10A). As expected, pMLC2 colocalized with actin fibers incontrol cells. However, the pMLC2 signal was strongly reduced incells transfected with miR-200c mimic. Similarly, silencing ofFHOD1 or PPM1F also resulted in a decrease in pMLC2 staining.Quantification of fluorescence intensities confirmed the signifi-

cance of the observed effects (Fig. 10B). To assess total MLC2 andphospho-MLC2 levels, we detected pMLC2 and total MLC2 byWestern blotting (Fig. 10C). Consistent with the results from im-munofluorescence analysis, transfection with miR-200c mimic re-sulted in a considerable reduction in pMLC2. Silencing of FHOD1or PPM1F was sufficient to replicate this effect. We also assessedthe monophosphorylated (pSer19) form of MLC2 and observedthat it was also reduced by overexpression of miR-200c and bysilencing of FHOD1 or PPM1F (data not shown). Interestingly,overexpression of miR-200c or silencing of FHOD1 also stronglydecreased total MLC2 levels, suggesting a regulation of MLC2 ex-

FIG 9 FHOD1 and PPM1F regulate elongation and stress fiber formation of breast cancer cells. (A) Validation of FHOD1 and PPM1F overexpression constructs.MCF-7 cells were transfected with expression constructs for the FHOD1 ORF (pCMV5-HA-FHOD1), PPM1F ORF (pcDNA-Dest47-PPM1F), and respectiveempty vector controls for 48 h. Protein was isolated, and FHOD1 as well as PPM1F expression levels were analyzed by Western blotting. �-Actin was used as aloading control. (B) Analysis of cell elongation by fluorescence microscopy in MDA-MB-231 cells. Cells were transfected with control siRNA (siAllStar) orsiRNAs directed against FHOD1 and PPM1F and fixed 24 h after transfection. Actin filaments were visualized by staining with Alexa Fluor 488-phalloidin (green)and the nucleus by staining with DAPI (blue). Cell elongation was quantified by measuring the long and short axes of cells using the Zeiss LSM Image Examiner.Box plots represent data from two independent experiments, each time measuring 40 cells per condition. (C) Stress fiber assay of MDA-MB-231 cells. Cells weretransfected with siRNAs and starved for 24 h. Stress fibers were induced by treatment with TGF-� for 5 h, and cells were stained with Alexa Fluor 488-phalloidinand DAPI. The percentage of stress fiber-containing cells was determined by counting 200 to 300 cells per experiment; data shown represents means of data fromthree independent experiments. (D) Analysis of cell elongation by fluorescence microscopy in MCF-7 cells. Cells were transfected with expression constructs forFHOD1 or PPM1F or respective empty vectors, starved for 24 h starting 2 days after transfection, and stimulated with TGF-� for 5 h. Staining, microscopy, andimage analysis were carried out as described for panel B. (E) Stress fiber assay of MCF-7 cells. Cells were transfected with expression constructs for FHOD1 orPPM1F or respective empty vectors, starved for 24 h starting 2 days after transfection, and stimulated with TGF-� for 5 h for induction of stress fibers. Staining,microscopy, and image analysis were carried out as described for panel C.

Jurmeister et al.



http://mcb.asm

.org/D

ownloaded from

pression by miR-200c and its target FHOD1. In contrast, silencingof PPM1F did not result in a reduction of total MLC2 levels, indi-cating that its effect is potentially due to changes in signaling path-ways regulating phosphorylation of MLC2, rather than throughregulation of MLC2 expression.

In MCF-7 cells, inhibition of miR-200c resulted in an increasein pMLC2/MLC2 levels as assessed by both immunofluorescence(Fig. 11A and B) and Western blotting (Fig. 11C). Similarly, over-expression of FHOD1 also resulted in higher levels of both totaland diphosphorylated MLC2 (Fig. 11A to C). In contrast, overex-pression of PPM1F increased only the phosphorylation of MLC2(Fig. 11A to C) but not its expression (Fig. 11C), as shown with theknockdown experiments (Fig. 10). This suggests that the effect ofmiR-200c on MLC2 expression may be mediated primarilythrough targeting of FHOD1 and not PPM1F, which affects onlythe phosphorylation of MLC2. Altogether, these results suggestthat miR-200c negatively regulates expression of MLC2 throughtargeting of FHOD1, leading to a concomitant decrease in the levelof the active form, pMLC2.

SRF connects miR-200c and its target FHOD1 to MLC2 ex-pression. We next aimed to investigate the mechanism underlyingthe regulation of MLC2 expression by miR-200c and FHOD1.FHOD1 has been previously implicated in the regulation of serum

response factor (SRF) (40, 45, 46, 48). SRF is a key regulator of theactin cytoskeleton and contractile processes (27). MLC2 is an es-tablished direct target gene of SRF, offering an attractive potentialexplanation for the effect of FHOD1 on MLC2 expression (12, 26).To test the hypothesis that miR-200c and FHOD1 might regulateMLC2 expression through SRF, we first tested their effects on SRFprotein levels (Fig. 12A). Interestingly, both overexpression ofmiR-200c and silencing of FHOD1 decreased SRF expression inMDA-MB-231 cells. Analysis of SRF transcript levels supportedthese findings, since transfection of MDA-MB-231 cells with miR-200c mimic or siRNA against FHOD1 resulted in a modest butsignificant decrease in SRF transcript levels (Fig. 12B). We thenassessed whether miR-200c and its target gene FHOD1 also af-fected the transcriptional activity of SRF by making use of serumresponse element (SRE) luciferase reporter gene constructs. Over-expression of miR-200c resulted in a strong reduction of SRF ac-tivity, and this effect was mimicked by silencing of FHOD1 (Fig.12C). Conversely, inhibition of miR-200c or overexpression ofFHOD1 stimulated SRF transcription (Fig. 12D).

It has been well established that myocardin-related transcriptionfactors (MRTFs) function as coactivators of SRF. MRTFs are seques-tered in the cytosol by monomeric G-actin, coupling SRF transcrip-tional activity to actin cytoskeletal dynamics (28, 39). Furthermore, it

FIG 10 Overexpression of miR-200c or silencing of its target genes reduces MLC2 phosphorylation. (A) MDA-MB-231 cells were transfected with microRNAmimics and siRNAs for 24 h and starved for an additional 24 h. Stress fibers were induced by treatment with TGF-� for 5 h, and cells were stained for actin withAlexa Fluor 488-phalloidin (green), for pThr18/pSer19-MLC2 (pMLC2; red), and for the nucleus with DAPI (blue). (B) Quantification of pMLC2 staining.Immunofluorescence images of MDA-MB-231 cells transfected with microRNA mimics and siRNAs were acquired as described for panel A. Mean fluorescenceintensities of approximately 40 cells per condition were quantified using the ImageJ software program. For representation in box plots, data were normalized tothe median of control mimic or siAllStar, respectively. (C) MDA-MB-231 cells were transfected, starved and stimulated as for panel A, and protein was isolated.pThr18/pSer19-MLC2 (pMLC2), total MLC2 (MLC2), FHOD1, and PPM1F were detected by Western blotting. Tubulin was used as a loading control.




http://mcb.asm

.org/D

ownloaded from

has been proposed that FHOD1 may also regulate SRF activitythrough modulating intracellular G-actin levels due to its actin-nucleating activity (9). To investigate this, we analyzed the intracel-lular localization of MRTF-A in MDA-MB-231 cells upon transfec-tion of miR-200c mimic or siFHOD1 (Fig. 12E). In both cases, weobserved a decrease in the nuclear localization of MRTF-A, althoughthe results did not score as significant in the case of miR-200c over-expression (P � 0.09, two-tailed Student’s t test). Next, we testedwhether inhibition of SRF could explain the effect of miR-200c over-expression or FHOD1 knockdown on MLC2 expression. Silencing ofSRF induced a clear reduction of pMLC2 levels in immunofluores-cence experiments (Fig. 13A). This decrease in pMLC2 was con-firmed by Western blotting (Fig. 13B). Moreover, knockdown of SRFalso reduced total MLC2 levels, confirming that the effect on pMLC2levels is mediated through regulation of gene expression (Fig. 13B).Thus, silencing of SRF phenocopies the effects of both miR-200coverexpression and FHOD1 knockdown on MLC2. Furthermore,MDA-MB-231 cells transfected with siRNA directed against SRF dis-played decreased invasion in a Matrigel invasion assay (Fig. 13C),suggesting that regulation of SRF by miR-200c could at least partlyaccount for miR-200c-mediated inhibition of invasive capability. Si-

lencing of SRF showed no effect on cell viability in an MDA-MB-231cell line model, excluding that the observed effect could be caused bycytotoxic effects of the knockdown (Fig. 13D). In order to examinethe effect of SRF on cell migration, we first tested the knockdownefficiency of a siRNA pool and individual siRNAs. All four siRNAsdirected against SRF, as well the siRNA pool, efficiently reduced SRFtranscript levels (Fig. 13E). In real-time migration assays using MDA-MB-231 cells, both individual siRNAs directed against SRF and thepool resulted in a significant reduction of migration (Fig. 13F), sug-gesting that regulation of SRF is involved not only in invasion but alsoin migration as a downstream mediator of miR-200c and its targetFHOD1.

Finally, since MDA-MB-231 cells and MCF-7 cells representcellular models of the miR-200c low/highly invasive/migratoryand miR-200c high/noninvasive/less-migratory phenotypes, re-spectively, we then compared the expression levels of the FHOD1/PPM1F/SRF/MLC2 cellular motility axis described above in thesetwo cell lines (Fig. 13G). Downstream effectors of miR-200c thatpromote migration, i.e., FHOD1, PPM1F, SRF, and MLC2, wereconsistently expressed at higher levels in MDA-MB-231 cells thanin MCF-7 cells. Furthermore, levels of the phosphorylated, active

FIG 11 Inhibition of miR-200c or overexpression of its target genes increases MLC2 phosphorylation. (A) MCF-7 cells were transfected with microRNAinhibitors or expression constructs for 24 h and starved for an additional 24 h. Stress fibers were induced by treatment with TGF-� for 5 h, and cells were stainedfor actin with Alexa Fluor 488-phalloidin (green), for pThr18/pSer19-MLC2 (pMLC2; red), and for the nucleus with DAPI (blue). (B) Quantification of pMLC2staining. Immunofluorescence images of MCF-7 cells transfected with microRNA inhibitors or expression constructs were acquired as described for panel A.Mean fluorescence intensities of approximately 40 cells per condition were quantified using ImageJ. For representation in box plots, data were normalized to themedian of control mimic or siAllStar, respectively. (C) MCF-7 cells were transfected, starved, and stimulated as for panel A, and protein was isolated. pThr18/pSer19-MLC2 (pMLC2) and total MLC2 (MLC2) were detected by Western blotting. Tubulin was used as a loading control.

Jurmeister et al.



http://mcb.asm

.org/D

ownloaded from

form of MLC2 were also higher in MDA-MB-231 cells than inMCF-7 cells.

DISCUSSION

Changes in cell shape that are as profound as the transition fromcobblestone-like epithelial cells to elongated, spindle-shaped mes-enchymal cells almost invariably require remodeling of the actincytoskeleton. Such an epithelial-mesenchymal transition pre-

cedes cell migration and metastasis in cancer but is also re-quired during normal differentiation processes, such as meso-derm and neural tube development (32). While the activity ofthe miR-200 family of miRNAs in relation to EMT-related phe-notypes has been extensively studied (13, 14, 22, 31), its poten-tial contribution to the organization and dynamics of the actincytoskeleton has mostly not been investigated thus far. Here wepresent the first detailed characterization of the molecular

FIG 12 miR-200c and its target gene FHOD1 regulate SRF. (A) MDA-MB-231 cells were transfected with microRNA mimics or siRNAs. Protein wasisolated 48 h after transfection, and SRF protein levels were determined by Western blotting. Tubulin was used as a loading control. (B) MDA-MB-231cells were transfected with microRNA mimics or siRNAs for 48 h. RNA was isolated, and SRF transcript levels were determined by qRT-PCR, using HPRTand TFRC as housekeeping genes; n � 5 except for siSRF (n � 3). (C) SRF response element reporter assay with microRNA mimics and siRNAs.HEK293FT cells were cotransfected with the pGL4.34 [luc2P/SRF-RE/Hygro] reporter vector, pRL-TK vector, and microRNA mimics or siRNAs.Twenty-four hours after transfection, cells were starved for 24 h and stimulated with 10 ng/ml TGF-�1 for 6 h before cell lysis. Luciferase activity wasmeasured and normalized to Renilla luciferase activity (n � 8). (D) SRF response element reporter assay with microRNA hairpin inhibitors and expressionconstructs. HEK293FT cells were cotransfected with the pGL4.34 [luc2P/SRF-RE/Hygro] reporter vector, pRL-TK vector, and microRNA hairpininhibitors or expression constructs. SRF transcriptional activity was analyzed as described for panel C (n � 8). (E) MRTF nuclear translocation assay.MDA-MB-231 cells were transfected with miRNA mimics or siRNAs and stained for MRTF-A and nuclei (DAPI). Images for quantitative analysis atmagnification �20 were acquired and analyzed by using Olympus ScanR analysis software. Cells with a ratio of nuclear/cytoplasmic localization ofMRTF-A higher than 1 were considered positive for nuclear localization, while cells with a ratio of �1 were considered negative for nuclear localization.Data shown represent fold changes in the ratio of MRTF-A nuclear localization with standard deviations.




http://mcb.asm

.org/D

ownloaded from

FIG 13 Regulation of MLC2 by miR-200c and FHOD1 is mediated through SRF. (A) MDA-MB-231 cells were transfected with control siRNA (siAllStar) or siRNA directedagainst SRF for 24 h. Cells were starved for an additional 24 h, and stress fibers were induced by stimulation with 10 ng/ml TGF-� for 5 h. Cells were stained for actin with AlexaFluor488-phalloidin(green),forpThr18/pSer19-MLC2(pMLC2;red),andforthenucleuswithDAPI(blue).(B)MDA-MB-231cellsweretreatedasforpanelA,andproteinwasisolated. SRF, pMLC (Thr18/Ser19), and MLC were detected by Western blotting, using tubulin as a loading control. (C) Matrigel invasion assay of MDA-MB-231 cellstransfected with siRNAs. Cells were transfected, seeded in Matrigel-coated invasion plates, and stimulated to invade in the presence of TGF-� (10 ng/ml). Invaded cells werequantified by flow cytometry. (D) Viability assay of MDA-MB-231 cells transfected with siRNAs. Cell viability was measured using the Cell Titer Glo assay 72 h posttransfection.Viability values are normalized to control mimic for miR-200c and to siAllStar for siRNAs. (E) Effect of individual siRNAs on siSRF transcript levels. MDA-MB-231 cells weretransfected with 40 nM either individual siRNAs or pools of 4 siRNAs directed against SRF. After 48 h, RNA was isolated, and SRF transcript levels were quantified by qRT-PCR.HPRTandTFRCwereusedforhousekeeping-genecontrols. (F)RTCAmigrationassayofMDA-MB-231cells transfectedwithsiRNAsdirectedagainstSRF.MDA-MB-231cellswere transfected, starved in serum-free medium for 24 h, and seeded in RTCA CIM-16 plates. Full medium was used as a chemoattractant, and impedance measurements wereperformed in a time-resolved manner. Means for four replicates � standard deviations are shown; a t test was performed for the last time point. (G) Comparison of miR-200cdownstream effector levels in MDA-MB-231 cells and MCF-7 cells. Total protein was isolated from MDA-MB-231 cells and MCF-7 cells. FHOD1, PPM1F, SRF, pMLC(Thr18/Ser19), and total MLC were detected by Western blotting, using tubulin as a loading control. (H) Proposed model for miR-200c-mediated regulation of stress fibers andmigration/invasion.miR-200ctargetsFHOD1,resultingindecreasedactinpolymerizationandthuspreventingstressfiberformation.Additionally,MRTFsaresequesteredinthecytosol by increasing levels of monomeric actin, resulting in inhibition of SRF transcriptional activity and decreased expression of MLC2. Phosphorylated MLC2 is a componentofactivemyosin,whichcontributestostressfiberformationandfunctionthroughcross-linkingofactinfilaments,andbyprovidingcontractileactivity.AsecondmiR-200ctargetgene,PPM1F, increasesphosphorylation levelsofMLC2independentlyofMLC2expression,andtargetingofPPM1Fmight thus furtherenhancetheeffectofmiR-200constressfiber formation.



http://mcb.asm

.org/D

ownloaded from

mechanisms by which miR-200c exerts effects on the actin cy-toskeleton in relation to EMT-like phenotypes. Upon transfec-tion with miR-200c mimic, MDA-MB-231 cells lost the elon-gated shape associated with motile, mesenchymal cells andadopted a rounded, unpolarized shape. Importantly, the acqui-sition of a rounded cell shape was accompanied by a loss ofactin structures associated with cell polarization and perturbedstress fiber formation. These phenotypes are mediated byFHOD1 and PPM1F, which we found to be direct targets ofmiR-200c. Silencing of these target genes not only reducedmigration and invasion of breast cancer cells in both two-dimensional (2D) and three-dimensional (3D) model systemsof cancer cell motility but was also sufficient to completelyabrogate the stimulatory effect of miR-200c inhibition on mi-gration of MCF-7 cells. While some previous evidence had al-ready linked PPM1F to invasion in breast cancer (40), ourstudy is the first to implicate FHOD1 in the promotion of can-cer cell invasion. Furthermore, we have demonstrated thatFHOD1 and PPM1F promote elongation, polarization, andstress fiber formation in breast cancer cells, all of which are keyprocesses for migration and invasion. Our in vitro findingswere supported by an inverse correlation of the expression lev-els of FHOD1 and PPM1F mRNAs with miR-200c in breastcancer specimens, breast cancer and mammary cell lines, andthe cell lines of the NCI-60 panel, indicating that this effectappears to be a general mechanism rather than being restrictedto specific cell types.

Investigation of the molecular mechanisms mediating theeffects of miR-200c on the actin cytoskeleton revealed that bothmiR-200c and FHOD1 regulate the expression of MLC2. It hasbeen reported that FHOD1 stimulates SRF-mediated tran-scription (9, 25, 45), and our own results elaborate this in thecontext of miR-200c-mediated inhibition of migration and in-vasion. Moreover, while it had been previously suggested thatthis effect of FHOD1 may be due to its ability to decrease theintracellular G actin concentration and concomitant nucleartranslocation of the SRF coactivator MRTF, we have demon-strated for the first time that silencing of FHOD1 indeed resultsin decreased nuclear localization of the SRF coactivatorMRTF-A. G-actin sequesters myocardin-related transcriptionfactors (MRTFs) in the cytoplasm, which function as SRF co-activators (28, 39). Formins, such as mDia1, have been shownto regulate SRF through this mechanism (4), and our resultssuggest that FHOD1, which is a formin homolog, may act in asimilar way. Since SRF also stimulates its own gene transcrip-tion (27), inhibition of its transcriptional activity could alsoexplain the decreased SRF transcript and protein levels that weobserved upon silencing of FHOD1. It should be noted thatwhile we did observe a tendency for decreased nuclear localiza-tion of MRTF-A upon miR-200c overexpression, the resultfailed to reach significance (P � 0.09). However, the pro-nounced rounding of cells upon miR-200c overexpression in-terfered with the quantification of MRTF-A localization, andwe believe that this may have impacted the result.

Taken together, our data suggest a model in which miR-200c regulates migration and invasion by inhibiting stress fiberformation that is primarily mediated via targeting of FHOD1.FHOD1 promotes stress fiber formation by activating SRF, re-sulting in increased expression of MLC2. PhosphorylatedMLC2 then contributes to stress fiber formation and function

through cross-linking of actin filaments by myosin and by pro-viding contractile activity. Additionally, stimulation of actinnucleation by FHOD1 is expected to further contribute tostress fiber formation. Furthermore, PPM1F increases phos-phorylation levels of MLC2 independent of MLC2 expression,and targeting of PPM1F by miR-200c might thus further en-hance the effect of miR-200c on stress fiber formation (Fig.13H). It has been previously reported that PPM1F inhibitsPAK, resulting in increased phosphorylation of myosin lightchain phosphatase (MYPT) at inhibitory sites (19). However,our results demonstrated that while overexpression of miR-200c increased the phosphorylation of MYPT at Thr696 or atThr853, silencing of PPM1F did not affect any phosphorylationsites on MYPT (data not shown). This suggests that yet anothermechanism has to be involved. Therefore, further studies arewarranted to elucidate the exact mechanism through whichPPM1F regulates MLC2 phosphorylation.

It has been shown that reexpression of miR-200c in mesen-chymal cell lines reverses EMT (13, 22, 31). This raises thequestion of whether the targeting of EMT transcription factorsZEB1 and ZEB2 by miR-200c would be sufficient to explain thiseffect as well as the concomitant decrease in cell migration. Arecent study demonstrated that while silencing of ZEB1 andZEB2 can indeed restore E-cadherin expression in mousemammary epithelial cells that had undergone TGF-�-inducedEMT, this is not sufficient to completely reverse the mesenchy-mal phenotype (5). Strikingly, cells failed to fully remodel theiractin cytoskeleton to an epithelial organization characterizedby a lack of stress fibers and the appearance of cortical actinfilaments. Only the combination of ZEB knockdown with aROCK inhibitor was able to fully reverse EMT, suggesting thatinterference with both ZEB transcription factors and the Rhopathway is required. Another recent study has shown that sta-ble cell lines expressing miR-200 family members adopted anepithelial cell-like cell phenotype; however, cells stably express-ing E-cadherin maintained a mesenchymal phenotype (21).These findings are in line with the model of miR-200c action wepropose here. While miR-200c does not influence the activa-tion level of RhoA directly, its target genes FHOD1 and PPM1Fact downstream of the RhoA signaling pathway. The best-characterized outputs of the RhoA pathway are increased actin-nucleating activity, promotion of actin-myosin assembly, sta-bilization of actin filaments, and activation of SRF-mediatedtranscription. Together, FHOD1 and PPM1F are involved in allof these processes. Our findings led us to hypothesize that miR-200c can reverse EMT by targeting both the two ZEB transcrip-tion factors and the actin cytoskeleton via FHOD1/PPM1F.Thus, we propose that miR-200c could regulate the inductionand maintenance of the epithelial phenotype through twolargely separate pathways that complement each other: miR-200c downregulates ZEB1 and ZEB2, resulting in an inductionof epithelial molecules such as E-cadherin. This results in in-creased formation of stable cell-cell contacts, which impairsmigration. However, this is not sufficient to fully reverse themesenchymal phenotype, as demonstrated by the inability ofZEB1/ZEB2 knockdown to prevent stress fiber formation.Thus, miR-200c additionally targets genes involved in cyto-skeletal organization, namely, FHOD1 and PPM1F, and poten-tially other cytoskeletal regulators, such as the already-identified target genes WAVE3 and MARKCS (7, 38). This




http://mcb.asm

.org/D

ownloaded from

prevents the formation of stress fibers and cell polarization andshifts actin cytoskeletal organization toward the nonmotile,epithelial phenotype. Targeting of other regulators of cellularmotility, such as moesin and fibronectin (15), may further con-tribute to miR-200c-mediated inhibition of motility.

In conclusion, we have provided evidence that in addition toits well-established role in regulating cell-cell contacts, miR-200c also has a role in repressing breast cancer cell invasion andmigration through modulation of the accompanying actin cy-toskeleton reorganization during EMT. Additional studies willbe required to elucidate the roles of the miR-200c targets,FHOD1 and PPM1F, during metastasis development in vivo.However, our findings already suggest that therapeutic reex-pression of miR-200c, which has been proposed as a futureantimetastasis treatment, may have the benefit of regulatingmultiple pathways involved in the regulation of metastasis, in-cluding actin cytoskeletal dynamics.

ACKNOWLEDGMENTS

This work was supported by the National Genome Research Network(grants 01GS0816 and 01GS0864) of the German Federal Ministry of Ed-ucation and Research (BMBF) and Wilhelm Sanderstiftung (grant2009.051.1). J.D.Z. was supported by the DKFZ International Ph.D. Pro-gram.

We thank Moritz Küblbeck for excellent technical assistance. We alsothank Oliver Fackler (Department of Virology, University HospitalHeidelberg) for kindly providing us with the pCMV5-HA-FHOD1 ex-pression construct.

REFERENCES1. Ahmad A, et al. 2011. Phosphoglucose Isomerase/Autocrine Motility

Factor mediates epithelial-mesenchymal transition regulated by miR-200in breast cancer cells. Cancer Res. 71:3400 –3409.

2. Blick T, et al. 2008. Epithelial mesenchymal transition traits in humanbreast cancer cell lines. Clin. Exp. Metastasis 25:629 – 642.

3. Burk U, et al. 2008. A reciprocal repression between ZEB1 and membersof the miR-200 family promotes EMT and invasion in cancer cells. EMBORep. 9:582–589.

4. Copeland JW, Treisman R. 2002. The diaphanous-related formin mDia1controls serum response factor activity through its effects on actin poly-merization. Mol. Biol. Cell 13:4088 – 4099.

5. Das S, Becker BN, Hoffmann FM, Mertz JE. 2009. Complete reversal ofepithelial to mesenchymal transition requires inhibition of both ZEB ex-pression and the Rho pathway. BMC Cell Biol. 10:94.

6. Edlund S, Landstrom M, Heldin CH, Aspenstrom P. 2002. Transform-ing growth factor-beta-induced mobilization of actin cytoskeleton re-quires signaling by small GTPases Cdc42 and RhoA. Mol. Biol. Cell 13:902–914.

7. Elson-Schwab I, Lorentzen A, Marshall CJ. 2010. MicroRNA-200 familymembers differentially regulate morphological plasticity and mode ofmelanoma cell invasion. PLoS One 5:e13176.

8. Enerly E, et al. 2011. miRNA-mRNA integrated analysis reveals roles formiRNAs in primary breast tumors. PLoS One 6:e16915.

9. Gasteier JE, et al. 2003. Activation of the Rac-binding partner FHOD1induces actin stress fibers via a ROCK-dependent mechanism. J. Biol.Chem. 278:38902–38912.

10. Gasteier JE, et al. 2005. FHOD1 coordinates actin filament and microtu-bule alignment to mediate cell elongation. Exp. Cell Res. 306:192–202.

11. Gibbons DL, et al. 2009. Contextual extracellular cues promote tumorcell EMT and metastasis by regulating miR-200 family expression. GenesDev. 23:2140 –2151.

12. Gilles L, et al. 2009. MAL/SRF complex is involved in platelet formationand megakaryocyte migration by regulating MYL9 (MLC2) and MMP9.Blood 114:4221– 4232.

13. Gregory PA, et al. 2008. The miR-200 family and miR-205 regulateepithelial to mesenchymal transition by targeting ZEB1 and SIP1. Nat.Cell Biol. 10:593– 601.

14. Gregory PA, et al. 2011. An autocrine TGF-�/ZEB/miR-200 signalingnetwork regulates establishment and maintenance of epithelial-mesenchymal transition. Mol. Biol. Cell. 22:1686 –1698.

15. Howe EN, Cochrane DR, Richer JK. 2011. Targets of miR-200c mediatesuppression of cell motility and anoikis resistance. Breast Cancer Res.13:R45.

16. Hu X, et al. 2009. A miR-200 microRNA cluster as prognostic marker inadvanced ovarian cancer. Gynecol. Oncol. 114:457– 464.

17. Hurteau GJ, Carlson JA, Roos E, Brock GJ. 2009. Stable expression ofmiR-200c alone is sufficient to regulate TCF8 (ZEB1) and restoreE-cadherin expression. Cell Cycle 8:2064 –2069.

18. Iliopoulos D, et al. 2010. Loss of miR-200 inhibition of Suz12 leads topolycomb-mediated repression required for the formation and mainte-nance of cancer stem cells. Mol. Cell 39:761–772.

19. Koh CG, Tan EJ, Manser E, Lim L. 2002. The p21-activated kinase PAKis negatively regulated by POPX1 and POPX2, a pair of serine/threoninephosphatases of the PP2C family. Curr. Biol. 12:317–321.

20. Kong D, et al. 2009. miR-200 regulates PDGF-D-mediated epithelial-mesenchymal transition, adhesion, and invasion of prostate cancer cells.Stem Cells 27:1712–1721.

21. Korpal M, et al. 2011. Direct targeting of Sec23a by miR-200s influencescancer cell secretome and promotes metastatic colonization. Nat. Med.17:1101–1108.

22. Korpal M, Lee ES, Hu G, Kang Y. 2008. The miR-200 family inhibitsepithelial-mesenchymal transition and cancer cell migration by direct tar-geting of E-cadherin transcriptional repressors ZEB1 and ZEB2. J. Biol.Chem. 283:14910 –14914.

23. Livak KJ, Schmittgen TD. 2001. Analysis of relative gene expression datausing real-time quantitative PCR and the 2(-Delta Delta C(T)) method.Methods 25:402– 408.

24. Lombaerts M, et al. 2006. E-cadherin transcriptional downregulation by pro-motermethylationbutnotmutationisrelatedtoepithelial-to-mesenchymaltran-sition in breast cancer cell lines. Br. J. Cancer 94:661–671.

25. Madrid R, et al. 2005. Oligomerization of the diaphanous-related forminFHOD1 requires a coiled-coil motif critical for its cytoskeletal and tran-scriptional activities. FEBS Lett. 579:441– 448.

26. Medjkane S, Perez-Sanchez C, Gaggioli C, Sahai E, Treisman R. 2009.Myocardin-related transcription factors and SRF are required for cyto-skeletal dynamics and experimental metastasis. Nat. Cell Biol. 11:257–268.

27. Miano JM, Long X, Fujiwara K. 2007. Serum response factor: masterregulator of the actin cytoskeleton and contractile apparatus. Am. J.Physiol. Cell Physiol. 292:C70 –C81.

28. Miralles F, Posern G, Zaromytidou AI, Treisman R. 2003. Actin dynamicscontrol SRF activity by regulation of its coactivator MAL. Cell 113:329–342.

29. Neve RM, et al. 2006. A collection of breast cancer cell lines for the studyof functionally distinct cancer subtypes. Cancer Cell 10:515–527.

30. Olson P, et al. 2009. MicroRNA dynamics in the stages of tumorigenesiscorrelate with hallmark capabilities of cancer. Genes Dev. 23:2152–2165.

31. Park SM, Gaur AB, Lengyel E, Peter ME. 2008. The miR-200 familydetermines the epithelial phenotype of cancer cells by targeting theE-cadherin repressors ZEB1 and ZEB2. Genes Dev. 22:894 –907.

32. Row RH, et al. 2011. Completion of the epithelial to mesenchymal tran-sition in zebrafish mesoderm requires Spadetail. Dev. Biol. 354:102–110.

33. Sahin O, et al. 2009. Modeling ERBB receptor-regulated G1/S transition to findnovel targets for de novo trastuzumab resistance. BMC Syst. Biol. 3:1.

34. Schickel R, Park SM, Murmann AE, Peter ME. 2010. miR-200c regulatesinduction of apoptosis through CD95 by targeting FAP-1. Mol. Cell 38:908 –915.

35. Shen X, et al. 2001. The activity of guanine exchange factor NET1 isessential for transforming growth factor-beta-mediated stress fiber forma-tion. J. Biol. Chem. 276:15362–15368.

36. Solly K, Wang X, Xu X, Strulovici B, Zheng W. 2004. Application ofreal-time cell electronic sensing (RT-CES) technology to cell-based assays.Assay Drug Dev. Technol. 2:363–372.

37. Sommers CL, Byers SW, Thompson EW, Torri JA, Gelmann EP. 1994.Differentiation state and invasiveness of human breast cancer cell lines.Breast Cancer Res. Treat. 31:325–335.

38. Sossey-Alaoui K, Bialkowska K, Plow EF. 2009. The miR200 family ofmicroRNAs regulates WAVE3-dependent cancer cell invasion. J. Biol.Chem. 284:33019 –33029.

39. Sotiropoulos A, Gineitis D, Copeland J, Treisman R. 1999. Signal-regulated activation of serum response factor is mediated by changes inactin dynamics. Cell 98:159 –169.

Jurmeister et al.



http://mcb.asm

.org/D

ownloaded from

Atención al cliente 902 20 30 80 [email protected] izasa.es

AppNotes-Migración 1/2013

Health & Medicine

Izasa.App-notes-Acea.Migracion-invasion