48 Amplification and overexpression of vinculin are associated with increased tumour cell...

ORIGINAL PAPERJournal of PathologyJ Pathol 2011; 223: 543–552Published online 5 January 2011 in Wiley Online Library(wileyonlinelibrary.com) DOI: 10.1002/path.2828

Amplification and overexpression of vinculin are associatedwith increased tumour cell proliferation and progressionin advanced prostate cancerChristian Ruiz,1* David R Holz,2 Martin Oeggerli,1 Sandra Schneider,1 Irma M Gonzales,2 Jeffrey M Kiefer,2Tobias Zellweger,3 Alexander Bachmann,4 Pasi A Koivisto,5 Heikki J Helin,6 Spyro Mousses,2 Michael T Barrett,2David O Azorsa2 and Lukas Bubendorf1

1 Institute for Pathology, University Hospital Basel, Basel, Switzerland2 Pharmaceutical Genomics Division, Translational Genomics Research Institute, Scottsdale, Arizona, USA3 Division of Urology, St Claraspital, Basel, Switzerland4 Department of Urology, University Hospital Basel, Basel, Switzerland5 Departments of Neurology, Seinajoki and Vaasa Central Hospitals, Finland6 Division of Pathology, HUSLAB, Helsinki University Hospital, Finland

*Correspondence to: Christian Ruiz, Institute for Pathology, University Hospital Basel, Schoenbeinstrasse 40, 4031 Basel, Switzerland.e-mail: [email protected]

AbstractAndrogen withdrawal is the standard treatment for advanced prostate cancer. Although this therapy is initiallyeffective, nearly all prostate cancers become refractory to it. Approximately 15% of these castration-resistantprostate cancers harbour a genomic amplification at 10q22. The aim of this study was to explore the structure ofthe 10q22 amplicon and to determine the major driving genes. Application of high-resolution array-CGH usingthe 244k Agilent microarrays to cell lines with 10q22 amplification allowed us to narrow down the commonamplified region to a region of 5.8 megabases. We silenced each of the genes of this region by an RNAi screen inthe prostate cancer cell lines PC-3 and 22Rv1. We selected genes with a significant growth reduction in the 10q22amplified cell line PC-3, but not in the non-amplified 22Rv1 cells, as putative target genes of this amplicon.Immunohistochemical analysis of the protein expression of these candidate genes on a tissue microarray enrichedfor 10q22 amplified prostate cancers revealed vinculin as the most promising target of this amplicon. We founda strong association between vinculin gene amplification and overexpression (p < 0.001). Further analysis of443 specimens from across all stages of prostate cancer progression showed that vinculin expression was highestin castration-resistant prostate cancers, but negative or very low in benign prostatic hyperplasia (p < 0.0001).Additionally, high tumour cell proliferation measured by Ki67 expression was significantly associated with highvinculin expression in prostate cancer (p < 0.0001). Our data suggest that vinculin is a major driving gene ofthe 10q22 amplification in prostate cancer and that vinculin overexpression might contribute to prostate cancerprogression by enhancing tumour cell proliferation.Copyright 2011 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd.

Keywords: vinculin; 10q22; array-CGH; prostate cancer; tissue microarray

Received 14 September 2010; Revised 21 November 2010; Accepted 23 November 2010

Conflict of interest statement: The authors declare that the University Hospital Basel is in the process of filing a patent application for findingsdescribed in this article.

Introduction

Prostate cancer (PrCA) is the most frequent malig-nant tumour among males in Western countries andthe second leading cause of cancer-related death [1].More than an estimated 192 000 new prostate can-cers are diagnosed every year in the United States[1]. Although most tumours are now diagnosed at anearly stage, still many progress to metastatic disease.Androgen withdrawal is the standard therapy for thesepatients. Although this therapy is initially effective,nearly all prostate cancers become resistant to it after afew months or years, referred to as hormone-refractory

or castration-resistant (CR) PrCA [2]. The therapeuticoptions in CR PrCA are very limited [3]. Identificationof molecular targets responsible for this progressionis overdue, since their inhibition might help to atten-uate this process. The development of CR PrCA isaccompanied or driven by genomic aberrations, suchas mutations or genomic amplifications [4–6]. The lat-ter are of main interest, since they can pinpoint tooverexpressed genes located in the amplified genomicregion [7]. In PrCA, the most prevalent amplificationis that of the androgen receptor (AR), which prevails in20–30% of CR PrCAs and is associated with increasedAR expression [6,8].

Copyright 2011 Pathological Society of Great Britain and Ireland. J Pathol 2011; 223: 543–552Published by John Wiley & Sons, Ltd. www.pathsoc.org.uk www.thejournalofpathology.com

544 C Ruiz et al

We previously identified amplification of the 10q22region in 15% of CR PrCAs [9]. Since this amplifi-cation is absent in benign prostatic controls, precursorlesions, and clinically localized PrCA, we suggestedthat this genomic region might harbour genes with arole in the progression to the CR state. We have alreadyshown the importance of the potassium large conduc-tance calcium-activated channel, subfamily M, alphamember 1 gene (KCNMA1 ) at 10q22 in CR PrCA [10].Despite the relatively high prevalence of this genomicamplification in PrCA, the detailed structure of thisaberration has never been elucidated. This is of par-ticular importance since the 10q22 region harbours alarge number of genes which could also be involvedand drive the progression of PrCA. In order to iden-tify the driver genes of this amplicon, we mapped thecommon amplified region (CAR) of the 10q22 amplifi-cation by high-resolution array-CGH and performed anRNA interference (RNAi) screen on the genes locatedin it. Intriguingly, two genes, whose involvement inPrCA progression has never been reported until now,stood out from this screen: the genes vinculin (VCL)and discs large homologue 5 (DLG5 ). Here, for thefirst time, we analyse the role of these two genes inthe context of prostate cancer and suggest a novel rolefor vinculin as a tumour-promoting protein in advancedPrCA.

Materials and methods

Cell culturePC-3, SK-BR-3, and ME-180 were purchased fromLGC Standards (Wesel, Germany). 22Rv1 and MFM-223 were obtained from DSMZ (Braunschweig, Ger-many). All cells were cultured in either RPMI orOptimem (Invitrogen, Carlsbad, CA, USA) containing10% FBS (Amimed, Basel, Switzerland) and 1% peni-cillin/streptamycin (Amimed) in a humidified incubatorat 37 ◦C and 5% CO2.

Array-CGH analysisCells were cultured as described above and DNA wasextracted using the DNeasy Blood and Tissue Kit(Qiagen, Valencia, CA, USA) according to the man-ufacturer’s protocol. For each hybridization, 1 µg ofextracted DNA and 1 µg of 46,XX reference (Promega,Madison, WI, USA) were digested with the restrictionenzymes Alu I and RSA I (Promega), and then labelledwith Cy-5-dUTP and Cy-3-dUTP, respectively, usingthe BioPrime Labeling Kit (Invitrogen). Labelling reac-tions were assessed by use of the Nanodrop (Nanodrop,Wilmington, DE, USA) prior to hybridization to the244k CGH arrays (Agilent Technologies, Santa Clara,CA, USA).

Microarray slides were scanned using an Agilent2565C microarray scanner. Images were analysedwith Agilent Feature Extraction software version 10.1

according to the recommendations of the supplier. Theresulting aCGH data were assessed with QC metricsand then analysed using an aberrant detection algorithm(ADM2).

The microarray files (aCGH, Agilent Technolo-gies) used in this study have been deposited inNCBI’s Gene Expression Omnibus and are accessi-ble through GEO Series accession number GSE24216(http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE24216).

RNAi screen

RNAi screening was performed using a custom-madefocused siRNA library (Qiagen) targeting 26 genesin the 10q22 amplicon using four siRNAs per gene.Stock siRNA was diluted in siRNA buffer (Qiagen)and 9.3 ng of each siRNA was printed in quadruplicatewells of white Corning 384-well plates (Fisher Scien-tific; Pittsburgh, PA, USA) along with negative controlnon-silencing siRNA and positive control lethal siRNA(Qiagen). RNAi screening was carried out by reversetransfection of cells as described previously [11]. After96 h, the total cell number was determined by the addi-tion of Cell Titer Glo (Promega, Madison, WI, USA)and relative luminescence units (RLU) were measuredusing an EnVision plate reader (Perkin-Elmer, Welles-ley, MA, USA). Cell viability was calculated as log2ratios of siRNA treatment over untreated wells. Thedifferential response of log2 ratios of 22Rv1 cells ver-sus PC3 cells was plotted from two independent RNAiscreens.

RNAi validation and proliferation assay

DLG5-specific, VCL-specific, and control non-silencing siRNA (nsRNA) were purchased from Qia-gen. Cells were transfected with Lipofectamine(Invitrogen) and RNAi (Qiagen) as described in theQiagen protocol. Briefly, siRNA transfection was per-formed in serum- and antibiotic-free medium for 24 h.After transfection, the medium was replaced with 10%FBS containing medium. Cells were harvested at regu-lar time points to determine expression knockdown andcell growth rate. Growth rate was determined by count-ing cells with the CASY cell counter (ScharfesystemGmbH, Reutlingen, Germany).

RNA extraction and real-time PCR analysis

RNA extraction was performed by use of the RNeasyMini Kit (Qiagen) and reverse-transcribed using theTranscriptor First Strand cDNA Synthesis Kit (RocheDiagnostics, Rotkreuz, Switzerland). Real-time PCRanalysis was performed by use of the SYBR GreenKit (Applied Biosystems, Switzerland) according tothe manufacturer’s instructions. The primers used arelisted in the Supporting information, Supplementarymethods.


Vinculin expression in prostate cancer 545

Figure 1. Array-CGH analysis of 10q22 amplified cell lines. (A) aCGH analyses of chromosome 10 reveal a 5.8 Mb common amplified region.Characterization of the genes located in the CAR revealed a large functional heterogeneity, such as vesicle trafficking or budding (SEC24C,AP3M1), transferase activity (FUT11, NDST2, MYST4, COMTD1), kinase activity (CAMK2G, ADK ), elements of ion channels (KCNMA1, VDAC2)or components of the cytoskeleton like vinculin (VCL). (B) The common amplified region harbours 38 transcripts and consists of a large anda small core. Genes marked with an asterisk were considered for siRNA screening. All chromosomal positions are based on the hg assemblyfrom March 2006 (Build 36). CAR = common amplified region.

Protein extraction and western blot analysisWhole cell protein lysates were generated by use ofthe PARIS kit according to the manufacturer’s instruc-tions (Ambion, Rotkreuz, Switzerland). Protein lysateconcentration was determined by use of the RC DCProtein Assay (Bio-Rad Laboratories, Hercules, CA,USA), and equal amounts of proteins were resolvedby use of 4–20% precast Tris–glycine gels (Invit-rogen) and transferred to Immobilon-P (Millipore)membranes. The following primary antibodies wereused: anti-DLG5 (HPA000555; Sigma-Aldrich, Ger-many), anti-VCL (Mob226; Diagnostics BioSystems,Pleasanton, CA, USA), and anti-beta-tubulin (ab6046;Abcam, Germany). Visualization was carried out byuse of the Odyssey infrared detection system (Odyssey,Germany).

Tissue specimens and tissue microarray construction

Tissue microarrays (TMAs) were constructed as des-cribed before [12]. The 10q22 TMA was built byusing two to three tumour punches from CR PrCa of59 patients and from five benign prostatic hyperplasia(BPH) specimens. This TMA was enriched for PrCAwith 10q22 amplification identified in previous studiesby FISH and CGH [9,10]. The prostate progressionTMA has been described previously [13]. Use of theclinical samples for TMA construction was approvedby the Ethical Committee of the University of Basel.

Immunohistochemical analysis

Standard indirect immunoperoxidase procedures wereused for immunohistochemistry (IHC). Antibodies


546 C Ruiz et al

were tested on array sections from formalin-fixed,paraffin-embedded tissue and cell line samples. Anti-bodies, staining conditions, and the scoring system aredescribed in the Supporting information, Supplemen-tary methods. Images were obtained by use of a ZeissAXIO Imager.A1 microscope (Zeiss, Jena, Germany)equipped with an AxioCam (Zeiss) and AxioVision 4.6software (Zeiss).

Fluorescence in situ hybridizationFluorescence in situ hybridization (FISH) analyseswere performed with three digoxigenated BAC probes[RP11-354F19 (VCL gene), RP11-428P16 (KCNMA1 ),and RP11-21B16 (DLG5 ); imaGenes GmbH, Berlin,Germany] and the centromeric probe of chromosome10 (Vysis, Downers Grove, IL, USA). Hybridizationand post-hybridization washes were according to the‘LSI procedure’ (Vysis). Indirect labelling of the digox-igenated probes was carried out according to the ‘Flu-orescent Antibody Enhancer Set for DIG Detection’(Roche Applied Science, Rotkreuz, Switzerland). Forthe analysis of the three BAC FISH probes on the10q22 TMA, we counted the FISH signals in as manytumour cells as needed to draw an unequivocal conclu-sion about the amplification status. 10q22 amplificationwas defined as at least twice as many gene signals ascentromere 10 signals in the majority of the tumourcells. All FISH analyses were evaluated by two peo-ple (CR and SS). Images were obtained by use ofa Zeiss Axioplan 2 fluorescence microscope (Zeiss)equipped with an ISIS digital camera (MetaSystems,Altlussheim, Germany).

Statistical analysisStatistical summaries (Tables 1 and 2) and analyses(ANOVA, Student’s t-test) were performed by use ofJMP 7.0 software (SAS Corporation, NC, USA).

Results

High-resolution array-CGH confines the commonamplified region (CAR) of the 10q22 ampliconWe screened the in silico SNP microarray data pro-vided by the Wellcome Trust Sanger Centre in orderto identify cell lines harbouring a genomic ampli-fication of the 10q22–23 region. Two breast carci-noma cell lines (SKBR-3 and MFM-223) and oneprostate (PC-3) and one cervix carcinoma cell line(ME-180) were identified and their DNA was sub-jected to whole genome high-resolution aCGH anal-ysis (Figure 1). These data allowed us to narrow downthe CAR of the 10q22 amplicon to a region spanning5.8 megabases (Mb). In-depth analysis of the CARrevealed a two-core structure of the amplicon: a largecore with a size of 5.72 Mb harbouring 35 transcriptsand a small core (111 kb) comprising three transcripts.Twelve of the 38 transcripts were excluded from further

Figure 2. RNAi screening identifies VCL and DLG5 as genespreferential to PC3 cell growth and survival. 22Rv1 and PC3cells were transfected with siRNA targeting 26 genes (4 siRNAs pergene) in the 10q22 amplicon and assessed for their effects on cellgrowth. The difference in the growth log2 ratios between 22Rv1and PC3 cells from two independent RNAi screens is plotted foreach siRNA. Two siRNAs targeting DLG5 (yellow circles) and threesiRNAs targeting VCL (red circles) were found to be outside the1.65 SD limit (blue lines).

analyses since they represented either pseudo-genes orpredicted transcripts with no mRNA, and expressedsequence tag evidence in public databases.

RNAi screen of the genes in the CAR reveals twocandidate genesIn order to define the potential contribution of the 26genes identified above to cell growth in 10q22 ampli-fied PrCA, we applied a targeted RNAi screen usinga focused siRNA set across the two defined ampli-fied cores in the 10q22 region. We chose two prostatecarcinoma cell lines for this purpose: the 10q22 ampli-fied PC-3 cell line, which was originally isolated froma bone metastasis and is androgen-independent [14],and the androgen-sensitive cell line 22Rv1, which doesnot harbour any genomic aberrations in chromosome10 [15]. The 10q22 siRNA focused set contained fourdifferent siRNAs targeting each of the 26 genes, andwe examined their effect on cell growth 96 h aftertransfection. Only down-regulation of the genes DLG5and VCL resulted in a statistically highly significantreduction of cell growth in the amplified PC-3 cellsbut not in the non-amplified 22Rv1 cells. This effectwas observed with at least two independent siRNAs inboth replicate experiments (Figure 2). Based on thesefindings, we selected the genes DLG5 and VCL aspotentially the most important target genes of the 10q22amplicon and subjected them to further validation.

Validation of the two candidate genes VCLand DLG5In order to validate the results obtained from the RNAiscreen, 22Rv1 and PC3 cells were grown in six-wellplates and subjected to a growth assay using siRNAspecific for the human gene DLG5 and for the geneVCL, respectively (Figure 3). As expected, knockdownof these two genes showed a specific growth effectin 10q22 amplified PC-3 cells (Figure 3A), but not



Figure 3. VCL and DLG5 siRNA validation on PC3 and 22Rv1 cells. (A) SiRNA treatment with either VCL or DLG5 leads to attenuated cellgrowth only in PC3 cells. Scrambled siRNA (ns siRNA) was used as a negative control. (B) Real-time PCR analysis showing RNA knockdownof DLG5 and VCL after siRNA treatment 24 h after transfection. (C) Western blot analysis showing DLG5 and VCL protein knockdown aftersiRNA treatment 3 days after transfection. u.c. = arbitrary units for living cells (average values ± SEM); n.s. = not significant, u.e. =arbitrary units for RNA expression ± SD. ∗p < 0.05; ∗∗p < 0.0001.

in non-amplified 22Rv1 cells. Knockdown of vinculinexpression led to a growth reduction by approximately30% in PC-3 cells (p < 0.05). The effect of DLG5silencing was more dramatic: the growth of PC-3 cellswas reduced by more than 65% (p < 0.05).

Increased vinculin protein expression in 10q22amplified prostate carcinomas

Subsequently, we investigated the association between10q22 amplification and protein expression of thetwo potential amplification target genes DLG5 andVCL. We constructed a TMA enriched for PrCAwhose 10q22 amplification status had been previouslyassessed by fluorescence in situ hybridization (FISH)(Figure 4). Immunohistochemical protein expressionanalysis of vinculin and DLG5 revealed membranousas well as cytoplasmic localization in most of the

positive samples (Figures 4D–4F). In the BPH controlspecimens, secretory epithelial cells were negative forvinculin, whereas the surrounding basal cells as wellas the cells of the fibromuscular stroma were stronglypositive (Figure 4G). Vinculin was highly expressedin 41 of 103 PrCA samples (40%) (Table 1). Therewas a strong association between 10q22 amplificationand vinculin overexpression (p < 0.001, Figure 4Aand Table 1), suggesting that 10q22 gene amplifica-tion is a driving mechanism for the overexpression ofthe VCL gene. In contrast, the DLG5 protein expres-sion was restricted to a small cohort of PrCAs: only7 of 99 PrCa samples revealed significant (moder-ate or strong) expression, but none of them showed10q22 amplification. Obviously, there was no correla-tion between 10q22 amplification and DLG5 proteinexpression (p = 0.22). Since only vinculin expressionwas strongly associated with 10q22 amplification,


548 C Ruiz et al

Figure 4. Tissue microarray analysis of vinculin genomic amplification and protein expression. (A) 10q22 amplified prostate carcinomas(PrCas) show higher vinculin protein expression than PrCas without 10q22 amplification. (B) FISH analysis showing 10q22 amplificationin a castration-resistant (CR) PrCa. Green channel: FITC (RP11-354F19); red channel: Spectrum Orange (centromere 10); blue channel:DAPI. (C) FISH analysis showing normal 10q22 genomic status in a CR PrCa. (D) A 10q22 amplified CR PrCa with high vinculin proteinexpression. (E) A CR PrCa with normal 10q22 gene status showing weak vinculin protein expression. (F) Strong vinculin protein expression(left) correlated with a higher Ki67 labelling index (right). (G) Epithelial cells of the BPH lack vinculin protein expression (black arrow).Smooth muscle cells (red arrow) and the basal cells (green arrow) of benign glands express high levels of vinculin.

we focused our subsequent work on the analysis ofvinculin.

Vinculin protein expression analysis in differentprostate diseases

In order to investigate the protein expression pro-file of vinculin across the whole spectrum of prostatecancer progression, we applied immunohistochemistry

to a prostate progression TMA [13]. This microar-ray consists of 443 prostate samples including BPH,high-grade prostatic intraepithelial neoplasia (HGPIN),clinically localized prostate cancers, and CR PrCArecurrences. Representative pictures of the vinculinprotein expression are shown in Figures 4D–4G.Table 2 and Figure 5A summarize the protein expres-sion results of different stages of PrCA progressionand BPH. Vinculin was significantly expressed in 50%



Table 1. Immunohistochemical analysis of VCL and DLG5 on the 10q22 TMASpots on Analysable (n) Analysable (n) Average Analysable (n) Averagearray (n) (10q22 FISH) (IHC VCL) (VCL score) (IHC DLG5) (DLG5 score)

Prostate carcinomas (all) 122 108 103 (62∗/41†) 114.0 99 (92∗/7†) 28.810q22 amplified 15 14 (6/8) 188.6 11 (11/0) 10.910q22 normal 93 82 (51/31) 101.2 80 (73/7) 33.6Benign prostatic hyperplasia 10 7‡ 7 (7/0) 31.4 6 (6/0) 33.3

This TMA consists of 122 PrCa specimens from 59 patients and ten benign prostatic hyperplasia (BPH) specimens from five patients. 10q22 amplified PrCas showsignificantly higher vinculin protein expression than non-amplified samples (p < 0.001). This association is not true for the protein expression of DLG5. The IHC scoringsystem ranges from 0 (no expression) to 300 (strong expression in all of the tumour cells), as described in detail in the Materials and methods section. FISH status wasassessed by the evaluation of three different FISH probes, all of them covering genomic regions of the common amplified region (see Materials and methods section).All 10q22 amplified PrCa samples showed genomic amplification with each of the three probes. Amplification was defined as the presence of more than twice as many10q22 signals as centromere 10 signals. ∗Number of TMA specimens with negative or weak protein expression. †Number of TMA specimens with moderate or strongprotein expression (IHC score > 100). ‡None of the BPH specimens showed 10q22 amplification.

Table 2. Immunohistochemical analysis of vinculin on the prostate progression TMANegative or weak Moderate Strong

Samples on Average VCL expression (n) expression (n) expression (n)array (n) Analysable (n) score Score: 0–100 Score: 101–200 Score: 201–300

Benign prostatic hyperplasia (BPH) 65 60 11.5 58 (97%) 2 (3%) 0High-grade prostatic 78 57 58.2 47 (82%) 10 (18%) 0intraepithelial neoplasia (HGPIN)Clinically localized prostate 95 65 32.2 60 (92%) 5 (8%) 0cancer T1a/bpT2a–pT3b (RP) 85 74 50.1 66 (89%) 8 (11%) 0Castration-resistant prostate 120 90 131.8 45 (50%) 31 (34%) 14 (16%)cancer

Vinculin protein expression is differentially expressed across stages of prostate cancer progression. Strong vinculin expression was observed only in castration-resistant(CR) prostate cancers. RP = radical prostatectomy specimen.

of the CR PrCAs (45 out of 90) and was expressedat the lowest degree in BPH (p < 0.0001). Clinicallylocalized PrCA showed intermediate but distinct vin-culin expression, which was significantly lower than inCR PrCA (p < 0.0001) but higher than in BPH (p <

0.005). Classification of the expression scores intothe three-tiered IHC expression categories (Table 2)emphasizes these findings: strong vinculin expressionis restricted to the group of CR PrCAs, and even therate of moderately expressing prostate samples (34%)is much higher than in the other groups analysed(p < 0.0001). Since the data from the RNAi screen(Figure 2) as well as the data from the validation exper-iments suggested that vinculin plays a role in enhancedproliferation, we investigated a potential correlationbetween vinculin protein expression and proliferation(Ki67): high tumour cell proliferation in PrCA was sig-nificantly associated with high vinculin protein expres-sion (p < 0.0001, Figure 5B). Notably, this associationremained true when the analysis was restricted to theCR PrCAs (p < 0.05, Figure 5C).

Discussion

The molecular mechanisms that drive the progressionto CR PrCA are incompletely understood. We recentlyfound a considerable fraction of CR PrCAs character-ized by a 10q22 genomic amplification that might con-tribute to this progression [10]. Here, after exploringthe structure of the 10q22 amplicon, an RNAi screen

strongly suggested the genes DLG5 and VCL as the twocandidate genes of this amplicon. We have shown thatvinculin protein expression is strongly associated with10q22 amplification and that its down-regulation bysiRNA represses the proliferation of 10q22 amplifiedprostate cancer cells. This strongly advocates for vin-culin as a major driving gene of this amplification. Thisis further emphasized in vivo by the strong associationof vinculin protein expression with cell proliferation inPrCA samples.

Our results are provocative in the light of previ-ous reports where tumour-suppressing properties wereattributed to vinculin (reviewed in ref 16). Vinculin ismostly known as an important component of the focaladhesion complex, and its down-regulation has beenassociated with increased cell migration and motil-ity [17–19]. However, most of these studies wereperformed on planar substrata. There is growing evi-dence that this experimental situation is reversed indense three-dimensional (3D) matrices [20]. Recently,Mierke et al reported the necessity for vinculin expres-sion for the invasion of 3D matrices [21]. It is evidentthat a 3D environment mimics the situation of clinicalPrCA samples better than cell culture experiments onplanar substrata. Furthermore, it has to be consideredthat vinculin is amplified in a considerable fraction ofCR PrCAs and these samples are characterized by sig-nificantly higher vinculin protein expression comparedwith non-amplified PrCA.

Amplification and the associated overexpression of agene are typically attributed to proteins with oncogenic


550 C Ruiz et al

Figure 5. TMA analysis of vinculin protein expression across different stages of prostate cancer progression and correlation with tumourcell proliferation. (A) Statistical analysis reveals high vinculin expression in CR PrCa and very low expression in BPHs. (B) Strong associationbetween the expression of the proliferation marker Ki67 and vinculin expression in prostate carcinomas. (C) This association is also true ifthis analysis is restricted to CR PrCa.

properties. Moreover, we found enhanced prolifera-tive activity in vinculin-expressing samples, suggest-ing that vinculin has a role in promoting proliferation,which might be unrelated to the well-known role ofvinculin in cell adhesion. It should also be consid-ered that although vinculin is mentioned in more than2400 pubmed entries, its protein expression has beenanalysed in only a few PrCA specimens and cell lines[22,23].

Immunohistochemical analysis of vinculin in PrCArevealed that its expression pattern was surprisinglydifferent from what would have been expected for acomponent of the focal adhesions: vinculin expressionwas not restricted to the (sub)membranous compart-ment. Samples with high vinculin expression showedprominent accumulation of vinculin in the cytoplasm.Although cytoplasmic localization of vinculin has beendescribed, its functional role remains unclear [24]. Thecurrent model proposes the existence of a cytosolicpool of auto-inhibited vinculin, which is recruited tothe cytoskeleton as required [16]. The fact that wefound cytoplasmic vinculin to be associated with pro-liferation and advanced stage of PrCA strongly arguesfor a functional role of this protein in the cytoplasm.However, it remains to be studied if cytoplasmic vin-culin in PrCA is in an activated form and interacts

with partners involved in signalling pathways. Inter-estingly, we also observed that vinculin was virtuallyabsent in the secretory epithelial cells of the benignprostate glands. Only the fibromuscular stroma, whichis abundant in the prostate, showed very strong vin-culin expression. Altogether, these findings furthercontribute to our suggested tumour-promoting role ofvinculin.

High-resolution array-CGH for the exact definitionof the CAR, followed by siRNA profiling of thegenes is a powerful technique for the identificationof the driver genes of an amplicon. Nevertheless,this approach has certain limitations. While the RNAiprofiling was designed to identify genes that impactviability and proliferation, genes with other oncogenicproperties such as enhanced angiogenesis or migrationcannot be highlighted by this technique. In addition,the two cell lines in our RNAi model system might notcover all biological aspects of the 10q22 amplificationin prostate cancer. For example, KCNMA1 knockdowndid result in a mild decrease of the cell numberin PC3 cells, but also in 22Rv1 cells; thus, therewas no significant difference in the overall effect ofKCNMA1 RNAi between the two cell lines, althoughdata from our previous study implicated KCNMA1 as acandidate amplification target gene [10]. Furthermore,



candidate genes emerging from RNAi profiling studiesin vitro need to be validated in vivo. In our study, theresults from the array-CGH and the RNAi experimentssuggested DLG5 as a compelling candidate gene. Butdue to the missing correlation between amplificationand expression in vivo, we decided not to consider thisgene further as a putative driving gene of the 10q22amplicon. Nevertheless, DLG5 immunohistochemistryflagged a small subset of advanced CR PrCAs withdistinct DLG5 expression. DLG5 is a member ofthe MAGUK family (membrane-associated guanylatekinase) and is mainly unexplored in cancer. Our resultsfrom the silencing experiments, as well as from theprotein expression analysis, suggest a role for DLG5in the context of PrCA. However, further studies areneeded to elucidate this role.

In summary, we have systematically analysed theprotein expression of vinculin in a large number ofhuman PrCA specimens for the first time. Our resultsstrongly advocate for vinculin as a main driving geneof the 10q22 amplicon and suggest a novel tumour-promoting role for vinculin in PrCa. Our findings ofthis well-known gene are surprisingly different fromthe functions that have hitherto been attributed to it.A limitation of our study is that we cannot providethe molecular mechanism by which vinculin performsthese functions. Further studies including analyses ofsignalling pathways are needed to clarify the exact roleand mechanisms of the vinculin action in PrCA.

AcknowledgmentWe thank Rita Epper, Thuy Nguyen, Rosemarie Chaf-fard, Alex Rufle, and Alex Robeson for their excellenttechnical support. This study was supported by theSwiss National Science Foundation (3100AO-105413)and the Oncosuisse Foundation (02285-08-2008).

Author contribution statement

CR, DRH, MO, SS, and IMG conceived and carriedout experiments. JMK, TZ, AB, PAK, HJH, DOA, andMTB collected and analysed data. CR and LB wrote thepaper. All authors discussed the results and commentedon the manuscript.

References

Note: Reference 25 is cited in the Supporting information tothis article.

1. Jemal A, Siegel R, Ward E, et al . Cancer statistics, 2009. CACancer J Clin 2009; 59: 225–249.

2. Petrylak DP, Tangen CM, Hussain MH, et al . Docetaxel and estra-mustine compared with mitoxantrone and prednisone for advancedrefractory prostate cancer. N Engl J Med 2004; 351: 1513–1520.

3. Lassi K, Dawson NA. Emerging therapies in castrate-resistantprostate cancer. Curr Opin Oncol 2009; 21: 260–265.

4. Dutt SS, Gao AC. Molecular mechanisms of castration-resistantprostate cancer progression. Future Oncol 2009; 5: 1403–1413.

5. Stubbs AP, Abel PD, Golding M, et al . Differentially expressedgenes in hormone refractory prostate cancer: association with chro-mosomal regions involved with genetic aberrations. Am J Pathol

1999; 154: 1335–1343.6. Visakorpi T, Hyytinen E, Koivisto P, et al . In vivo amplification

of the androgen receptor gene and progression of human prostatecancer. Nature Genet 1995; 9: 401–406.

7. Pollack JR, Sorlie T, Perou CM, et al . Microarray analysis revealsa major direct role of DNA copy number alteration in thetranscriptional program of human breast tumors. Proc Natl Acad

Sci U S A 2002; 99: 12963–12968.8. Bubendorf L, Kononen J, Koivisto P, et al . Survey of gene ampli-

fications during prostate cancer progression by high-throughoutfluorescence in situ hybridization on tissue microarrays. CancerRes 1999; 59: 803–806.

9. El Gedaily A, Bubendorf L, Willi N, et al . Discovery of newDNA amplification loci in prostate cancer by comparative genomichybridization. Prostate 2001; 46: 184–190.

10. Bloch M, Ousingsawat J, Simon R, et al . KCNMA1 gene amplifi-cation promotes tumor cell proliferation in human prostate cancer.Oncogene 2007; 26: 2525–2534.

11. Azorsa DO, Gonzales IM, Basu GD, et al . Synthetic lethal RNAiscreening identifies sensitizing targets for gemcitabine therapy inpancreatic cancer. J Transl Med 2009; 7: 43.

12. Simon R, Mirlacher MSG. Tissue microarrays. In Molecular Diag-nosis of Cancer. Roulsten J, Barlett J (eds). Humana Press: Tatowa,NJ, 2004; 377–389.

13. Zellweger T, Ninck C, Bloch M, et al . Expression patterns ofpotential therapeutic targets in prostate cancer. Int J Cancer 2005;113: 619–628.

14. Kaighn ME, Narayan KS, Ohnuki Y, et al . Establishment andcharacterization of a human prostatic carcinoma cell line (PC-3).Invest Urol 1979; 17: 16–23.

15. Sramkoski RM, Pretlow TG 2nd, Giaconia JM, et al . A newhuman prostate carcinoma cell line, 22Rv1. In Vitro Cell Dev Biol

Anim 1999; 35: 403–409.16. Ziegler WH, Liddington RC, Critchley DR. The structure and

regulation of vinculin. Trends Cell Biol 2006; 16: 453–460.17. Ezzell RM, Goldmann WH, Wang N, et al . Vinculin promotes cell

spreading by mechanically coupling integrins to the cytoskeleton.Exp Cell Res 1997; 231: 14–26.

18. Rodriguez Fernandez JL, Geiger B, Salomon D, et al . Suppressionof vinculin expression by antisense transfection confers changes incell morphology, motility, and anchorage-dependent growth of 3T3cells. J Cell Biol 1993; 122: 1285–1294.

19. Subauste MC, Pertz O, Adamson ED, et al . Vinculin modulationof paxillin–FAK interactions regulates ERK to control survival andmotility. J Cell Biol 2004; 165: 371–381.

20. Mierke CT. The role of vinculin in the regulation of the mechanicalproperties of cells. Cell Biochem Biophys 2009; 53: 115–126.

21. Mierke CT, Kollmannsberger P, Zitterbart DP, et al . Vinculinfacilitates cell invasion into three-dimensional collagen matrices.J Biol Chem 2010; 285: 13121–13130.

22. Schmelz M, Cress AE, Scott KM, et al . Different phenotypesin human prostate cancer: alpha6 or alpha3 integrin in cell-extracellular adhesion sites. Neoplasia 2002; 4: 243–254.

23. Lang SH, Hyde C, Reid IN, et al . Enhanced expression ofvimentin in motile prostate cell lines and in poorly differentiatedand metastatic prostate carcinoma. Prostate 2002; 52: 253–263.

24. Lee S, Otto JJ. Vinculin and talin: kinetics of entry and exit fromthe cytoskeletal pool. Cell Motil Cytoskeleton 1997; 36: 101–111.

25. Ruiz C, Seibt S, Al Kuraya K, et al . Tissue microarrays forcomparing molecular features with proliferation activity in breastcancer. Int J Cancer 2006; 118: 2190–2194.


552 C Ruiz et al

SUPPORTING INFORMATION ON THE INTERNET

The following supporting information may be found in the online version of this article.

Supplementary methods

Primer sequences used for real-time PCR, antibodies, and scoring system for immunohistochemistry experiments.


48 Amplification and overexpression of vinculin are associated with increased tumour cell...

Documents

Transcript of 48 Amplification and overexpression of vinculin are associated with increased tumour cell...