Functional Analysis of Genes in Regions Commonly Amplified in High-Grade Serous and Endometrioid Ovarian Cancer

Published OnlineFirst January 29, 2013; DOI: 10.1158/1078-0432.CCR-12-3433

Clinical Cancer Human Cancer Biology Research

Functional Analysis of Genes in Regions Commonly Ampliﬁed in High-Grade Serous and Endometrioid Ovarian Cancer

Sally J. Davis1,4, Karen E. Sheppard2,4,5,6, Richard B. Pearson2,5,6, Ian G. Campbell1,4,6, Kylie L. Gorringe1,4,6, and Kaylene J. Simpson3,4,6

Abstract Purpose: Ovarian cancer has the highest mortality rate of all the gynecologic malignancies and is responsible for approximately 140,000 deaths annually worldwide. Copy number amplification is frequently associated with the activation of oncogenic drivers in this tumor type, but their cytogenetic complexity and heterogeneity has made it difficult to determine which gene(s) within an amplicon represent(s) the genuine oncogenic driver. We sought to identify amplicon targets by conducting a comprehensive functional analysis of genes located in the regions of amplification in high-grade serous and endometrioid ovarian tumors. Experimental Design: High-throughput siRNA screening technology was used to systematically assess all genes within regions commonly amplified in high-grade serous and endometrioid cancer. We describe the results from a boutique siRNA screen of 272 genes in a panel of 18 ovarian cell lines. Hits identified by the functional viability screen were further interrogated in primary tumor cohorts to determine the clinical outcomes associated with amplification and gene overexpression. Results: We identified a number of genes as critical for cellular viability when amplified, including URI1, PAK4, GAB2, and DYRK1B. Integration of primary tumor gene expression and outcome data provided further evidence for the therapeutic use of such genes, particularly URI1 and GAB2, which were significantly associated with survival in 2 independent tumor cohorts. Conclusion: By taking this integrative approach to target discovery, we have streamlined the translation of high-resolution genomic data into preclinical in vitro studies, resulting in the identification of a number of genes that may be specifically targeted for the treatment of advanced ovarian tumors. Clin Cancer Res; 19(6); 1411–21. 2013 AACR.

Introduction exacerbated by limited treatment options. Current treat- Ovarian cancer is associated with the highest mortality ment for ovarian cancer involves cytoreductive surgery rate of all gynecologic malignancies and is identified as the combined with platinum-based chemotherapy. In most seventh most common cause of cancer-related death for cases, this treatment regime is initially successful, with women, with approximately 140,000 deaths annually approximately 80% of patients responding to the first-line worldwide (1). This high mortality rate is commonly attrib- treatment (2). However, the majority of patients relapse uted to the diagnosis of advanced stage disease and is often within 12 months and currently there are no curative treatments for recurrent disease. "Molecular-targeted agents" are a new class of cancer therapeutics that directly interfere with the biologic events Authors' Affiliations: 1Cancer Genetics Laboratory, 2Oncogenic Signaling and Growth Control Program, and 3Victorian Centre for Functional Geno- that drive tumorigenesis. The increased selectivity of these mics, Peter MacCallum Cancer Centre, East Melbourne; and Departments agents towards malignant cells has the potential to not only 4 5 6 of Pathology and Biochemistry and Molecular Biology and Sir Peter increase the efficacy of cancer treatment, but also to facilitate MacCallum Department of Oncology, The University of Melbourne, Park- ville, Victoria, Australia a more combinatorial approach to cancer therapeutics without the risks associated with overlapping toxicities (3–5). Note: Supplementary data for this article are available at Clinical Cancer Research Online (http://clincancerres.aacrjournals.org/). Unlike traditional chemotherapeutics, which exert their effect upon rapidly proliferating cells, molecular-targeted I.G. Campbell, K.L. Gorringe, and K.J. Simpson contributed equally to this work. agents are designed against genetic events that distinguish malignant cell populations from normal cells. Inherent to Corresponding Author: Kylie Gorringe, Cancer Genetics Laboratory, Peter MacCallum Cancer Centre, Locked Bag 1, A'Beckett St, Melbourne, the development of molecular-targeted agents is the iden- Victoria 8006, Australia. Phone: 613-9656-1131; Fax: 613-9656-1411; tification of driver genomic events unique to tumor cells. In E-mail: [email protected] high-grade ovarian cancer, recent studies have indicated doi: 10.1158/1078-0432.CCR-12-3433 that somatic mutational activation of oncogenes is a rare 2013 American Association for Cancer Research. event (6). However, somatic copy number amplification is

www.aacrjournals.org 1411

Davis et al.

threshold of >1). Genomic Identification of Significant Translational Relevance Targets in Cancer (12) was also used to identify amplifica- Despite the availability of extensive genomic data for tion peaks in the subset of 240 cases with SNP6 data. These high-grade serous and endometrioid ovarian cancers, loci were included if they satisfied the same frequency identifying key driver genes remains challenging due to requirement. The union of all peaks was then assessed and the extensive and complex cytogenetic alterations char- any loci delineated by a copy number polymorphism were acteristic of these tumors. Using high-throughput siRNA removed or adjusted. screening technology, we integrated genomic and functional data to robustly identify key ovarian cancer driver Cell culture genes. These discoveries will expedite the development Ovarian cell lines were obtained from the National Can- of targeted therapies that may improve the outcome for cer Institute (OVCAR3, OVCAR4, OVCAR8, SKOV3, Caov3), patients with advanced ovarian tumors. American Type Culture Collection (ATCC: Caov4), European Collection of Cell Cultures (ECACC: OAW28, 59M), The Institute of Physical and Chemical Research (RIKEN: JHOC9, JHOS3), Health Science Research Resources Bank (HSRRB: highly prevalent indicating that this is a common mecha- Kuramochi, MCAS), Deutsche Sammlung von Mikroorganis- nism of oncogene activation in this tumor type. Copy men und Zellkuturen (DSMZ: EFO21, FUOV1), Stephen number amplification generally results in the overexpres- Howell at University of California, San Diego (2008), and sion of genes harbored within the amplified chromosomal Dr Nuzhat Ahmed at the Women’s Cancer Research Centre, region. A number of high-resolution genomic studies have Royal Women’s Hospital, Melbourne (OVCA432). Cell lines identified recurrent copy number amplifications that occur were maintained according to the conditions outlined in in ovarian tumors (6–8). However, the insights provided Supplementary Table S1. We used short tandem repeat by such studies regarding the identity of the driver gene(s) (STR) profiling to verify the authenticity of the panel of cell is limited due to the fact that a large proportion of genes lines. We amplified 8 STR loci using primer sequences listed within amplified regions will not impart a functional in Supplementary Table S2. All profiles were compared with effect. In this study, we describe the largest systematic the ATCC database or STR profiles published on Catalogue functional assessment of genes located in frequently ampli- of Somatic Mutations in Cancer (COSMIC; http://www. fied regions of the ovarian tumor genome in an extensive sanger.ac.uk/genetics/CGP/cosmic), where possible. STR panel of ovarian cancer cell lines that recapitulate the analysis occurred both before and after siRNA screening to amplifications observed in primary samples. ensure that cell line misidentification was avoided. We sought to capitalize on our high-resolution copy number analysis of high-grade serous and endometrioid High-throughput siRNA screening tumors (7) by integrating those data with RNA interference Cell lines were grown and transfected with SMARTpool (RNAi) to identify amplification-specific genetic dependen- siRNAs using Dharmafect (DF) 1, 2, or 3 lipid transfection cies in ovarian cancer cells. Unlike previous studies that reagents (Thermo Fisher Scientific) as indicated in Supple- focus on individual regions of amplification (9, 10), we mentary Table S1. The boutique siRNA library of 272 have designed a boutique siRNA library targeting every siGENOME complexes (gene targets and catalog numbers protein-coding gene located within regions that are fre- in Supplementary Table S3) was obtained in a 384-well quently amplified in high-grade serous and endometrioid plate format from Dharmacon RNAi Technologies (Thermo ovarian cancer. Using high-throughput screening technol- Fisher Scientific), hydrated, and diluted in the Victorian ogy, we have been able to simultaneously assess the loss-of- Centre for Functional Genomics (VCFG) to 1 nmol using function phenotypes associated with each gene and identify 1 siRNA buffer (Dharmacon). Each well in this library novel candidate genes including URI1, GAB2, and PAK4 contained a SMARTpool of 4 distinct siRNA species target- that may be targeted for the treatment of high-grade serous ing different sequences of the target transcript. Cells were and endometrioid ovarian tumors. reverse transfected (13) at 30% confluence to a final con- centration of 40 nmol/L with the SMARTpool library using the Sciclone ALH3000 (Caliper Life Sciences) and BioTek Materials and Methods 406 (BioTek) liquid handling robotics. Cell lines were Amplicon selection transfected in duplicate and 2 independent biologic repli- Amplicons were selected on the basis of the following cates were screened. Each cell line was optimized to reach procedure. First, a peak finding algorithm (11) on the confluence at 72 hours after transfection. At that point, cell combined copy number (CN) frequency of single-nucleo- viability for each well was quantified on the basis of the tide polymorphism (SNP) array data from 398 ovarian direct measurement of intracellular ATP using the Cell Titer cancer samples (7) was conducted to identify genomic Glo luminescent assay (Promega) at a 1:4 dilution. All regions with log2 ratio thresholds of more than 0.8 at 5% luminescent measurements were taken on the Synergy frequency and more than 1 at 2.5% frequency. Each peak H4 high-throughput multimode microplate reader (Bio- was extended to encompass the segmentation boundaries of Tek). To identify loss of viability, data was normalized and 3 samples (at a CN threshold of >0.8) or 1 sample (at a CN processed as described by Brough and colleagues (14).

1412 Clin Cancer Res; 19(6) March 15, 2013 Clinical Cancer Research

Functional Analysis of Candidate Ovarian Cancer Driver Genes

Briefly, raw luminescent signals from each well were log Survival analysis was conducted on copy number and transformed, plate centered by median, and normalized expression data from the TCGA and AOCS datasets using by Z score. The Z scores generated in each replicate screen the Cox proportional hazards model. Genes with a low were averaged to generate a single Z score for each data median expression value across samples within a platform point that was then ranked to identify screening positives were excluded: TCGA Agilent <4(BMP8B, CACNA1C, or "hits". SAMD4B, DYRK1B) and AOCS U133 <7(BMP8B, CAC- NA1C, DYRK1B). Gene expression was treated as a contin- Validation of siRNA gene silencing using real-time uous variable, with residual disease at none, up to 1 cm, or quantitative PCR more than 1 cm as a covariate. Copy number was treated as Quantitative real-time PCR (qRT-PCR) was used to categorical comparing loss/neutral (CN log2 ratio <0.1) validate siRNA-mediated gene silencing efficiency and with gained (CN log2 ratio >0.5) with residual disease as specificity. Transfection conditions were scaled to a a covariate. 96-well format, cells were transfected with 40 nmol/L SMARTpool siRNA targeting URI1, DYRK1B, GAB2, BMP8B, ZFP36, CACNA1C, SAMD4B,andPAK4,and Results RNA was collected 24 hours after transfection. RNA was Selection of ovarian cell lines for analysis extracted from cells using the RNeasy Mini Kit (Qiagen), We identified 25 regions of the genome with frequent in accordance with the manufacturer’s instructions. cDNA high-level copy number amplifications in high-grade ser- was generated from 100 ng of RNA using the Superscript ous and endometrioid ovarian cancer based on our pre- III VILO kit (Invitrogen), in accordance with the manu- vious study (ref. 7; Table 1). To specifically address the facturer’s instructions. Expression of URI1, DYRK1B, functional role of genes targeted by these amplifications, GAB2, BMP8B, ZFP36, CACNA1C, SAMD4B,andPAK4 we first sought to select ovarian tumor cell lines that was assessed using primers listed in Supplementary recapitulate the genetic changes that were observed in Table S2 on the Lightcycler 480 (Roche Diagnostics). clinical samples. We obtained data from SNP-based high- PCR was conducted using the Brilliant II SYBR Green resolution copy number studies for 36 ovarian cell lines QPCR Mix (Agilent Technologies) and relative expres- (K.E. Sheppard, unpublished data; COSMIC) to identify sion was determined according to the comparative lines with gene amplifications similar to those identified in CT (DDCT)method:DDCT Gene siRNA DNA DCT Non- clinical samples (data not shown). We identified 17 cancer targeting siRNA DNA. DCT (threshold cycles) is the CT of cell lines (Table 1) for our analysis, with most amplicons the reference gene minus the CT of the target gene. represented in at least 3 independent cell lines. However, C Remaining gene expression was calculated by 2 DD T. chromosome 19 amplicons at 19p13.13 and 19p13.12 were Hypoxanthine phosphoribosyltransferase 1 (HPRT1)was only represented in a single cell line and no appropriate cell used as an endogenously expressed reference gene for line was identified for the chromosome 20p13 amplicon. the purpose of quantifying relative gene expression. Each Epithelial ovarian cancer encompasses a class of ovarian qRT-PCR was conducted in triplicate using cDNA derived tumors and includes a number of different histologic sub- from 3 independent experiments. types including serous, endometrioid, mucinous, and clear cell, which are further differentiated by low- and high-grade Copy number and gene expression in primary tumors status. Although we defined our amplified regions based We obtained both SNP copy number (n ¼ 388) and upon data collected from primary high-grade serous and Agilent gene expression data (n ¼ 399) of high-grade endometrioid tumors, cell lines for this analysis were select- serous ovarian tumors from The Cancer Genome Atlas ed on the basis of their copy number characteristics inde- (TCGA) and high-grade serous and endometrioid ovarian pendent of histologic subtype, which has resulted in some tumor Affymetrix U133Plus2 expression (n ¼ 254) and heterogeneity in the subtypes represented in this panel. Cell SNP copy number (n ¼ 159) from the Australian Ovarian lines derived from clear cell (2/17, 12%), low-grade (Type I) Cancer Study [AOCS, Tothill and colleagues ref. (15) and serous (1/17, 6%), high-grade (Type II) serous (11/17, Etemadmoghadam and colleagues ref. (16)] and Gorringe 64%), endometrioid (2/17, 12%), and mucinous (1/17. and colleagues (7). Gene 1.0ST expression data with 6%) tumors are included in this analysis (Supplementary matching copy number (n ¼ 48) but with no survival Table S1). information was obtained from Ramakrishna and collea- Screening the same set of candidate genes through 17 gues (17). For Affymetrix U133 array data, the probe with different cell lines provides internal controls for the specific the highest median value for each gene across all samples detection of amplicon-associated phenotypes. An addition- was used. Copy number data (TCGA and Gorringe and al nontumorigenic ovarian epithelial cell line (IOSE532) colleagues; ref. 7) was normalized and segmented as was also included in this study to identify genes that are described (7). The genomic segment with the highest copy essential for normal ovarian survival. This cell line does not number ratio overlapping each gene was selected to rep- harbor any of the amplifications found in the tumor- resent that gene. Gene expression values were compared derived cell lines. The identity of all cell lines was confirmed with matching copy number values using a Spearman by microsatellite genotyping (data not shown, see Materials correlation. and Methods).

www.aacrjournals.org Clin Cancer Res; 19(6) March 15, 2013 1413

Davis et al.

Table 1. Amplicons and cell lines for analysis

Tumors with Tumors with No. Region Location (Mb)a CN >1 (%)b CN >0.8 (%) genes Cell lines with CN gain 1p34.2 39.372–41.577 3.0 5.0 31 CAOV3, CAOV4, OAW28, KURAMOCHI, OVCA432 1p34.2 42.913–43.07 3.0 4.3 7 CAOV3, CAOV4, KURAMOCHI, OVCA432 1q21.1 145.415–145.831 2.5 4.5 3 CAOV4, KURAMOCHI, OVCAR4, JHOS3 3q26.2 170.244–170.406 4.8 12.3 2 EFO21, OVCAR3, JHOS3, CAOV3, CAOV4, FUOV1, OAW28, OVCAR4, 2008, OVCA432, JHOC9, SKOV3 3q26.31 173.564–174.253 1.8 8.0 6 EFO21, JHOS3, CAOV3, CAOV4, FUOV1, OAW28, 2008, JHOC9 3q26.32 178.213–178.589 2.3 7.0 1 EFO21, OVCAR3, JHOS3, CAOV3, CAOV4, FUOV1, OAW28, 2008, SKOV3 3q29 194.369–199.306 4.8 7.3 46 EFO21, OVCAR3, JHOS3, CAOV3, CAOV4, FUOV1, OAW28, OVCAR4, 2008, OVCA432, SKOV3, KURAMOCHI 6p21.1 40.899–42.031 2.5 3.0 22 JHOS3, CAOV3, OVCAR8 8q23.3 112.645–114.649 2.5 9.8 1 JHOS3, 59M, OAW28, KURAMOCHI, JHOC9 8q24.11 117.721–118.789 3.3 11.1 6 JHOS3, 59M, CAOV4, OAW28, KURAMOCHI, OVCAR8, JHOC9 8q24.12 121.039–121.517 3.8 12.6 3 EFO21, JHOS3, 59M, KURAMOCHI, JHOC9, OVCAR8 8q24.21 128.494–129.146 7.0 18.3 1 JHOS3, 59M, CAOV4, FUOV1, KURAMOCHI, JHOC9, OVCAR4, OVCAR3, OVCAR8 8q24.3 141.725–142.15 4.8 15.1 1 JHOS3, 59M, CAOV3, KURAMOCHI, JHOC9, OVCAR4, OVCAR8 8q24.3 146.202–146.275 2.3 6.3 1 JHOS3, 59M, CAOV3, KURAMOCHI, JHOC9, OVCAR4, OVCAR8 11q14.1 75.913–78.31 3.3 5.0 25 EFO21, ES2, OVCAR3 12p13.33-p13.32 1.205–4.214 3.5 6.3 19 ES2, CAOV3, CAOV4, FUOV1, OAW28 12p12.1 24.236–26.396 4.3 6.5 13 EFO21, JHOS3, MCAS, CAOV4, FUOV1, OAW28, KURAMOCHI, SKOV3 19p13.13 12.992–13.761 3.5 7.0 10 EFO21 19p13.12 15.021–15.314 2.8 6.5 7 EFO21 19q12 34.914–35.121 8.8 11.8 1 EFO21, CAOV4, FUOV1, OVCAR4, KURAMOCHI, OVCAR3, OVCAR8 19q12 35.135–35.198 9.0 11.6 1 EFO21, CAOV4, FUOV1, OVCAR4, KURAMOCHI, OVCAR3, OVCAR8 19q13.2 44.011–45.086 2.8 5.0 36 CAOV3, CAOV4, FUOV1, KURAMOCHI, OVCAR3 20p13 0.086–0.178 1.3 5.5 3 - 20q11.21 29.397–30.249 1.8 5.5 21 EFO21, JHOS3, CAOV4, FUOV1, OAW28, OVCAR4, 2008, KURAMOCHI, OVCAR3, SKOV3 20q13.2 50.408–52.416 2.3 5.5 5 EFO21, JHOS3, MCAS, CAOV4, OVCAR4, 2008, KURAMOCHI, OVCAR3, JHOC9

aBuild 36.1 2006 locations. bFrom Gorringe and colleagues (36).

Identification of amplicon target genes using siRNA ods). After 72 hours, cell viability was measured using the knockdown screening Cell Titer Glo luminescent assay (Promega), which allows A siRNA library targeting 272 annotated genes located for quantitation of cell number based on levels of cellular within regions of genomic amplification in high-grade ATP. The heterogeneity of cell line morphology and serous and endometrioid ovarian cancer was used to growth characteristics made it difficult to integrate all the transiently silence the expression of these genes in 18 data generated by the screen. To compare the data col- ovarian cell lines. The siRNA library was arrayed in a 384- lected across multiple cell lines, we applied the normal- well format to facilitate the simultaneous transfection of ization strategy of Brough and colleagues (14) to our data all genes in a single experiment (see Materials and Meth- set to generate Z scores as a normalized distribution

1414 Clin Cancer Res; 19(6) March 15, 2013 Clinical Cancer Research

Functional Analysis of Candidate Ovarian Cancer Driver Genes

Z across all the samples where negative scores represent Table 2. Summary of candidate genes loss of viability. identified by siRNA screen Cell line screening performance was assessed using a number of statistical methods as follows. Variability between replicates was calculated according to the percent- Number of affected a age coefficient of variation (%CV), which measured an cell lines (Z > 1.5) Genes average of 5.6% across replicate experiments for each gene 1 BCAS1, BCAT1, BCL9, BMP8B, 0 in all cell lines (Supplementary Fig. S1). We also used the Z RASSF8, URI1, CACNA1A, factor (18) as a means to establish the dynamic range CASC1, CASP14, CLDN19, between the positive and negative controls to provide a CYP24A1, DLG1, DYRK1B, measure of high-throughput screen performance. We used ECH1, ELKS, SLFNL1, XXYLT1, PLK1 as a positive control for cell death in all cell lines except RINL, FRS3, GAB2, GMFG, EFO21, MCAS, and OAW28, which were not affected by loss MED29, NCCRP1, KIAA0754, of PLK1. DLL3 was used as a positive control for these cell KRAS2, MDFI, C1ORF50, lines. Nontargeting control siRNA was used as a negative NDUFC2, SPATA16, PAK4, control for all cell lines assessed. The Z0 factor data for all cell PLAGL2, SARS2, SCMH1, lines was more than 0 with an average of 0.44, indicating a SOX5, SYCN, TFEB, THRSP, high-quality screen (ref. 19; Supplementary Fig. S1). MED20, ZDHHC19 For the purpose of this screen, "hits" were defined as those 2 ATP13A5, BCL2L1, CAP1, CPN2, siRNAs that resulted in an average Z score of < 1.5. This PIGX, FOXP4, CEP19, MYC, threshold was measured to be the equivalent of an average MYLK2, NCBP2, OPA1, PAK1, 40% decrease in cellular viability when compared with PPIH, SELV, TFRC, USP49, confluent cells (P < 0.0001) in all cell lines assessed ZFP36 (Supplementary Table S4). Applying this threshold, we 3 BYSL, CACNA1C, DEFB127, identified 74 genes that when silenced cause a statistically FYTTD1, C11ORF67, SUPT5H significant decrease in cellular viability (Table 2 and Sup- 4 FBL, MECOM, SAMD4B, ZNF684 plementary Table S5). Approximately half of the genes 5 RAD21, TPX2 identified (35/74, 47%) induced significant viability effects 6 DLL3, HNRNPL, RPS16 across multiple cell lines. The gene that significantly induced 7 RPL35A cell death in the largest number of cell lines was PTK2 8 ATP13A4 (also referred to as FAK), affecting viability in 9 independent 9 PTK2 cell lines. Notably, PTK2 is the subject of ongoing clinical aGenes highlighted in bold are significantly associated with trials in tumors that overexpress PTK2 including (but not gene amplification. limited to) breast, ovary, colon, liver, lung, and prostate tumors (GlaxoSmithKline 2012; clinicaltrials.gov identifier NCT01138033) and in hematologic malignancies such as be a more general response as the viability of IOSE532 cells lymphocytic leukemia (National Cancer Institute 2012; was also significantly impaired following loss of HNRNPL clinicaltrials.gov identifier NCT00563290). In addition, expression (Supplementary Table S5). Loss of TPX2 did not ribosomal genes such as FBL, RPL35A, RPS16, and HNRNPL induce a significant effect in IOSE532 cells (Supplementary were prominent with many hits in multiple cell lines. At least Fig. S2 and Supplementary Table S5) and this inverse 1 of these 4 ribosomal genes was detected as a hit in 67% of relationship between TPX2 amplification and TPX2 gene the cell lines analyzed (12/18; Supplementary Table S5). knockdown has also been reported by others (9). HNRNPL Since the primary purpose of this study was to identify and TPX2 were excluded from further analysis. genes that are amplified in ovarian cancers and also have a To ensure that the phenotypes observed in the screen were functional affect on tumorigenesis, all data were subjected attributable to on-target gene silencing, we quantified the to a 2-tailed t test to compare the Z scores in cell lines with efficiency of silencing by qRT-PCR in the 8 amplification- amplification at a specific gene locus to cell lines that were associated hits. Expression of each target gene was evaluated copy number neutral or showed copy number loss at that at24hoursaftertransfection,witheachtargetgeneassessedin locus. Applying this test, 8 of 74 (11%) of the genes were at least 2 cell lines with amplification alongside the IOSE532 significantly positively associated with gene amplification at cell line for comparison. In every case, highly efficient gene P < 0.05, namely CACNA1C, BMP8B, DYRK1B, GAB2, silencing was observed with an average of 79% silencing of PAK4, SAMD4B, URI1, and ZFP36 (Fig. 1 and Supplemen- gene expression across all cell lines analyzed (Fig. 1). tary Table S5). Interestingly, HNRNPL and TPX2 were found to have an inverse association with amplification. That is, Assessment of gene hits in primary tumors cell lines harboring amplification of the HNRNPL or TPX2 Having identified 8 genes with amplification-associated locus were protected from the decreased cellular viability viability effects (Fig. 1 and Supplementary Table S5) in cell effects that were observed in cell lines without amplification lines, we sought to validate this association in primary high- of this locus. The effect observed with siRNA HNRNPL may grade serous and endometrioid primary ovarian tumors.

www.aacrjournals.org Clin Cancer Res; 19(6) March 15, 2013 1415

Davis et al.

1416 Clin Cancer Res; 19(6) March 15, 2013 Clinical Cancer Research

Functional Analysis of Candidate Ovarian Cancer Driver Genes

Table 3. Correlation between copy number and gene expression of amplicon-associated siRNA hits in primary tumors

Spearman r2a Median expression value

AOCS TCGA Ramakrishna AOCS TCGA Ramakrishna BMP8B 0.41 0.062 0.30 5.43 3.11 6.92 CACNA1C 0.037 0.016 0.54 4.68 3.05 6.55 GAB2 0.63 0.48 0.70 8.61 6.52 7.79 URI1 0.71 0.71 0.39 9.69 7.24 8.76 ZFP36 0.16 0.18 0.10 8.78 7.31 6.75 SAMD4B 0.41 0.18 0.40 8.28 3.47 7.65 PAK4 0.47 0.71 0.50 8.19 6.15 7.99 DYRK1B 0.30 0.38 0.52 5.26 3.53 7.95

aBold values are signiﬁcant at P < 0.05.

Normalized gene expression data, copy number data, and cytogenetic complexity and heterogeneity of high-grade survival information from patients with ovarian cancer was serous and endometrioid ovarian tumors has made it dif- obtained from recent publications [TCGA; ref. 6), AOCS ficult to determine which gene(s) within an amplicon rep- (15, 16), Ramakrishna and colleagues (17), Gorringe and resent the genuine oncogenic driver. Previous studies that colleagues (7)]. A significant positive correlation between have used copy number amplification as an aid to functional expression and copy number was observed with all 8 genes identification of novel oncogenes have been limited to in at least 1 dataset (Table 3) and for 5 genes across all 3 single regions of recurrent amplification (9, 10, 23). We datasets (GAB2, URI1, SAMD4B, DYRK1B, and PAK4). have used high-throughput siRNA screening technologies Discordant results between datasets were frequently to conduct the first systematic analysis of all recurrent and observed when the measured expression of the gene was highly amplified genes in high-grade serous and endome- low, which could indicate lack of expression or poor probe trioid ovarian cancer in a large panel of ovarian tumor cell performance (CACNA1C, BMP8B). lines. We further enhanced our dataset by integrating data We also looked for an association of altered copy number from large clinical cohorts of primary tumors that have been and expression with overall and progression-free survival assessed for copy number, gene expression, and patient using the TCGA (6) and AOCS (15) datasets. It is well outcome. By taking this integrative approach, we have recognized that an inverse relationship exists between post- streamlined the translation of high-resolution genomic data operative residual disease and patient survival (20–22); into preclinical in vitro studies, resulting in the identification consequently, we included residual disease in the model of a number of genes that may be specifically targeted for the as a known prognostic factor (Table 4). Elevated URI1 was treatment of advanced ovarian tumors. consistently associated with worse outcome. GAB2 ampli- As proof-of-concept for our experimental and analysis fication and increased expression was associated with approach, one of the most robust hits was URI1 prefoldin- improved overall survival, but not progression-free survival like chaperone (URI1), which has recently been identified (Fig. 2). Expression of BMP8B or CACNA1C was not sig- as an ovarian oncogene (10). URI1 encodes a molecular nificantly associated with overall or progression-free sur- chaperone that is specifically associated with the mitochon- vival; however, their expression was low across the cohorts. drial inhibition of PPIg and PPIg-mediated inhibition of The 4 genes on the 19q13.2 amplicon (DYRK1B, SAMD4B, S6K1-BAD survival signaling. Using in vitro and in vivo PAK4, and ZFP36) were not consistently associated with analyses, Theurillat and colleagues showed that ovarian survival across the 2 cohorts, but this may be a reflection of cancer cells that harbor amplification of chromosome the limited power of the AOCS CN cohort. 19q12 were "addicted" to the aberrant expression of URI1 (10), such that short hairpin RNA (shRNA) knockdown of Discussion URI1 in ovarian tumor cells overexpressing URI1 resulted in Copy number amplification has long been recognized to proliferative arrest and cell death. Interestingly, URI1 lies be a mechanism of oncogene activation. However, the approximately 100 kb downstream of CCNE1, which has

Figure 1. Amplicon-associated loss of viability detected by high-throughput siRNA screening. Waterfall plots of Z scores collected across all cell lines and corresponding gene silencing for (A) BMP8B,(B)CACNA1C,(C)DYRK1B,(D)GAB2,(E)PAK4,(F)SAMD4B,(G)URI1, and (H) ZFP36. For viability plots, the average of 2 biologic replicates is shown. Error bars are representative of SEM, n ¼ 2. Z scores of cell lines with and without amplification of gene locus were compared by 2-tailed t test. All hit genes showed decreased cellular viability that was significantly associated with the presence of gene amplification (P < 0.05; data not shown). On-target gene silencing of more than 50% was observed for amplified and nonamplified cell lines for all target genes as measured by qRT-PCR. The average of three biologic replicates is shown. Error bars are representative of SEM, n ¼ 3. In all plots, red bars represent amplified cell lines and blue bars represent nonamplified cell lines.

www.aacrjournals.org Clin Cancer Res; 19(6) March 15, 2013 1417

Davis et al.

Table 4. Survival analyses by Cox proportional hazards

Overall survival Progression-free survival

Gene Locus TCGA Agilent AOCS U133 TCGA CN AOCS CN TCGA Agilent AOCS U133 TCGA CN AOCS CN URI1 19q12 ns GAB2 11q14.1 P ¼ 0.09 ns ns PAK4 19q13.2 ns P ¼ 0.05 ns ns P ¼ 0.08 ZFP36 19q13.2 ns ns ns ns P ¼ 0.08 SAMD4B 19q13.2 NA ns NA ns P ¼ 0.08 DYRK1B 19q13.2 NA NA ns NA NA P ¼ 0.09 Ns BMP8B 1p34.2 NA NA ns ns NA NA ns ns CACNA1C 12p13 NA NA ns NA NA ns P ¼ 0.07

NOTE: , P < 0.05; , P < 0.01; , P < 0.001. Abbreviations: A, Agilent 244K Custom Gene Expression; CN, copy number; NA, median gene expression low; ns, not signiﬁcant; U133, Affymetrix Gene Chip Human Genome U133 Plus 2.0 expression array.

previously been identified as the target gene for the 19q12 ing of high-grade serous and endometrioid ovarian tumors amplicon in high-grade serous and endometrioid ovar- including BMP8B, CACNA1C, DYRK1B, GAB2, PAK4, ian tumors (9). CCNE1 was included in our boutique SAMD4B, and ZFP36. Copy number gain was correlated siRNA library, but did not induce a significant decrease in with aberrant expression in primary tumors in at least one cell viability using our stringent criteria in cell lines har- independent cohort of primary tumors for all genes [TCGA, boring amplification at 19q12. This outcome is consistent AOCS, and Ramakrishna and colleagues (17)]. Important- with previous studies that report a limited viability effect ly, 4 of these genes were significantly correlated across all 4 of CCNE1 using transient siRNA knockdown, as com- datasets, namely, DYRK1B, GAB2, PAK4, and SAMD4B. pared with assays of anchorage-independent growth (9). Furthermore, when clinical outcome was taken into con- Therefore, we provide evidence to support the conclusion sideration, GAB2 and PAK4 emerged as strong candidates that other genes within the 19q12 amplicon, includ- for therapeutic intervention due to altered overall and/or ing URI1, may have important functionality in ovarian progression-free survival when copy number and gene tumorigenesis. expression were assessed using the TGCA and AOCS data- In addition to URI1, we identified 7 genes located in 4 sets. It should be noted that the TCGA dataset was a amplicons that are strong candidates for therapeutic target- substantially larger cohort of primary tumors; therefore,

ABGAB2 TCGA CN vs. Expr GAB2 AOCS CN vs. Expr 3 1.5 Figure 2. GAB2 ampliﬁcation 1.0 2 correlates with increased gene 0.5 expression and is associated with 1 patient outcome. Correlation 0.0 between GAB2 copy number and 0 gene expression measured by –0.5 microarray in (A) TCGA (R2 ¼ 0.48; –1 10864 6 8 10 12 P < 0.05) and (B) AOCS (R2 ¼ 0.63; Copy number (log2 ratio) number Copy (log2 ratio) number Copy –1.0 Expression Expression P < 0.05) datasets. Kaplan–Meier analysis of overall survival of (C) CD TCGA and (D) AOCS datasets GAB2 overall survival TCGA GAB2 overall survival AOCS using GAB2 copy number as a 100 Gain >0.5 100 Gain >0.5 categorical variable comparing Neutral/Loss <0.1 Neutral/Loss <0.1 80 80 loss/neutral copy number (CN log2 ratio <0.1) with gained (CN log 60 P 60 * P < 0.05 2 * < 0.05 ratio >0.5). Cox proportion hazards 40 40 test with residual disease as a 20 P Percent survival Percent survival Percent 20 copredictor values reported. 0 0 0 1,000 2,000 3,000 4,000 0 20406080100 OS (d) OS (mo)

1418 Clin Cancer Res; 19(6) March 15, 2013 Clinical Cancer Research

Functional Analysis of Candidate Ovarian Cancer Driver Genes

some of the associations observed in the TCGA dataset and could be potential therapeutic targets. DYRK1B knock- could not be replicated in the AOCS dataset possibly due down showed the most significant effect in association with to a lack of statistical power. amplified cell lines, and represents an attractive therapeutic GAB2 is a scaffolding adapter protein that binds the p85 target as a kinase; however, its expression was not strong by regulatory subunit of phosphoinositide-3-kinase (PI3K), microarray analysis. In contrast, PAK4 expression was sig- stimulating PI3K activity. We previously reported that GAB2 nificantly correlated with copy number, and showed the was gained in 31% (log2 ratio >0.3); amplified in 10% (log2 most consistent association with clinical outcome of the 4 ratio >0.6), and highly amplified in 5% (log2 ratio >0.8) of genes in the locus. PAK4 encodes a serine/threonine p21- high-grade serous and endometrioid tumors (7), which is activating kinase that is a major effecter of Rho GTPases consistent with other studies that have reported a frequency including Rac1 and Cdc42, which play an essential role in of approximately 15% amplification in ovarian tumors (24, cytoskeleton organization, cell morphology, apoptosis, sur- 25). This gene is located on 11q14, a region also amplified vival, and angiogenesis (28). Rho GTPases are generally in other tumor types such as breast (26). The amplicon is considered to be difficult to directly target therapeutically complex and distinct from the proximal CCND1 locus, and activating mutations in this family of proteins are rare which is seldom highly amplified in ovarian cancer, in (29). We observed PAK4 copy number gain (log2 ratio >0.3) contrast to breast cancer. A number of genes have been in 21%, amplification in 8% (log2 ratio >0.6), and high implicated as drivers for this amplicon, including EMSY/ amplitude amplification in 5% (log2 ratio >0.8) of high- C11ORF30 and PAK1 (24) suggesting an amplified gene grade serous and endometrioid ovarian tumors (7). Aber- cassette. However, we found no effect of EMSY knockdown, rant expression of PAK4 has been observed in a broad range and while reduced PAK1 showed an effect in 2 cell lines, this of tumor types including lung (30), ovary (30), colon (30, was not amplification associated. Thus, the functional data 31), breast (30), and pancreas (32, 33). This has led to the presented here combined with the genomic data provide suggestion that PAK4 could be an excellent molecule for convincing evidence to support GAB2 as an important targeted therapy and a number of small-molecule inhibitors driver of ovarian cancer for this locus that may be targetable targeting PAK4 have been developed (34, 35). In the context in the treatment of advanced tumors. Our clinical data of ovarian cancer, overexpression of PAK4 has been suggest that increased GAB2 copy number results in a reported at both the RNA and protein levels (36). In ovarian significant survival advantage in both the TCGA and AOCS tumor cells, PAK4 has been shown to regulate cell migration datasets (P < 0.05, respectively) indicating that GAB2 ampli- and invasion in a kinase-dependent fashion through the fication could drive a less aggressive cancer phenotype. activation of c-Src, MEK1, ERK1/2, and MMP2. Further- However, this survival analyses is conducted relative to more, PAK4 was also shown to regulate cell proliferation in other cancers of the same type; therefore, the outcome a c-Src- and EGFR-dependent manner (36). Despite the remains poor even when GAB2 amplification is present. genetic and functional evidence supporting the role of PAK4 Mechanistically, recent data suggest that GAB2 overex- in cancer, little is known about the mechanism under which pression enhances epithelial-to-mesenchymal (EMT)-like PAK4 operates in a biologic context. Specifically, there is behavior of ovarian tumor cells by enhancing properties little in the literature regarding PAK4 that gives a clear such as migration and invasion via activation of the PI3K- indication of molecules that operate downstream. This is Akt pathway and E-cadherin (27). This direct association a particular problem for the therapeutic targeting of PAK4 between GAB2 overexpression phenotypes and activation of due to the fact that there is no bona fide target or biomarker to PI3K has led to the use of small-molecule inhibitors for PI3K accurately detect the on-target inhibition of PAK4 function. and its downstream signaling molecule mTOR for the This issue must be addressed for PAK4 as a target to progress treatment of ovarian tumors that overexpress GAB2.In through to clinical development. preclinical studies, the use of such inhibitors significantly The genes included in our screen covered 25 distinct reduces the in vitro EMT-like behavior of GAB2 overexpres- amplicons that occur frequently in ovarian cancer. We have sing ovarian tumors cells (27). Furthermore, increased identified the potential drivers (strictly defined in this study sensitivity to a dual PI3K and mTOR inhibitor (PF- by an amplification-associated effect of gene knockdown) 04691502) has been shown to correlate with increased in 5 of these regions: 19q12 (URI1), 11q14 (GAB2), 12p13 expression of GAB2 mRNA in a panel of ovarian cancer (CACNA1C), 1p34 (BMP8B) and 19q13 (DYRK1B, PAK4, cell lines (Sheppard and colleagues, unpublished data). ZFP36, and SAMD4B). However, it is recognized that the These data suggest that PI3K/mTOR inhibitors may be used cell lines used for this analysis are limited in the extent to to treat ovarian tumor cells with amplification of 11q14 and which they recapitulate the types of amplicons that are consequent overexpression of GAB2 despite the inverse observed in patients with high-grade serous and endome- correlation we observed between GAB2 copy number and trioid ovarian cancer. For example, amplicons on 19p and overall survival. 20p are less frequently observed in cancer cell lines than The 19q13.2 amplicon was notable for identifying 4 would be expected from the primary tumor data. In addi- amplification-associated hits: PAK4, DYRK1B, ZFP36, and tion, it is plausible that many genes may only impart a SAMD4B. This amplicon measures approximately 1 Mb but phenotypic effect when silenced in conjunction with adja- is gene-rich with 36 genes included in the screen. Our data cent genes in the same region of amplification or with genes suggests that multiple genes may be drivers in this amplicon that lie on distal chromosomal locations. Functional

www.aacrjournals.org Clin Cancer Res; 19(6) March 15, 2013 1419

Davis et al.

assessment of combinations of genes within single or mul- strong basis for further analysis of amplified genes in tiple regions of amplification is outside the scope of this alternative assays and model systems. study. Therefore, it is likely that we may have failed to In summary, we have described the first systematic identify a number of driver genes within these tumor types. functional analysis of all amplified genes in high-grade Recent studies have indicated (37, 38) that while primary serous and endometrioid ovarian tumors. The integration tumors show clear evidence of a monoclonal ancestry, there of high-throughput siRNA loss-of-function screening are often multiple subclones with varying genetic land- with comprehensive copy number and expression geno- scapes. Thus, the genetic and expression profiles obtained mic analyses and clinical outcome has resulted in the from a primary tumor depend on the part of the tumor identification of a number of convincing targets for ther- taken for nucleic acid extraction. The amplicons studied apeutic intervention. here were chosen on the basis of frequency (indicating positive selection in multiple individual instances) and Disclosure of Potential Conflicts of Interest high level of amplification (suggesting that they represent No potential conflicts of interest were disclosed. the dominant clone in the tissue taken) and are therefore highly likely to represent genuine drivers and valid targets Authors' Contributions for therapy. Nonetheless, future applications of targeted Conception and design: S.J. Davis, I.G. Campbell, K.L. Gorringe, K. Simpson therapies derived from such data will need to recognize that Development of methodology: S.J. Davis, I.G. Campbell, K.J. Simpson a single target is unlikely to be effective in all tumor cells, Acquisition of data (provided animals, acquired and managed patients, and multiple simultaneous strands of therapy will be provided facilities, etc.): S.J. Davis, K.E. Sheppard, K.J. Simpson Analysis and interpretation of data (e.g., statistical analysis, biosta- required. This multiplicity of targets will require complex tistics, computational analysis): S.J. Davis, R.B. Pearson, K.L. Gorringe, K. screening strategies and careful design of clinical trials as J. Simpson eligible patients will likely be limited. Writing, review, and/or revision of the manuscript: S.J. Davis, K.E. Sheppard, R.B. Pearson, I.G. Campbell, K.L. Gorringe, K.J. Simpson These studies provide important targets for rational Administrative, technical, or material support (i.e., reporting or orga- design of next generation therapeutics. Although the loss nizing data, constructing databases): S.J. Davis, I.G. Campbell, K.J. Simpson of viability data presented here is not sufficient to conclu- Study supervision: I.G. Campbell, K.L. Gorringe, K.J. Simpson sively characterize driver genes in high-grade serous and endometrioid ovarian cancer, these data provide additional Acknowledgments evidence to justify a more detailed biologic and molecular The authors thank Daniel Thomas and Yanny Handoko (Victorian Centre investigation into the target genes identified. Further inves- for Functional Genomics, Peter MacCallum Cancer Centre) for their tech- tigation will be required to define drivers for other loci, with nical assistance and screening support, Joshy George (Peter MacCallum Cancer Centre) for bioinformatics advice, and Jason Ellul (Peter MacCallum either insufficient cell lines with gain (20p13, 19p13.12, Cancer Centre) for bioinformatics support. and 19p13.12) or conversely with too many cell lines with gain ( 8 cell lines: 4 regions on 3q, 12p12, 8q24.21, and Grant Support two regions on 20q). Both situations lead to a lack of power This study was funded by an Australian National Health and Medical and in the latter case perhaps a lack of specificity in the Research Council project grant (#566603) and the Emer Casey Foundation. S.J. Davis is supported by a Cancer Council of Victoria Postgraduate Schol- identification of amplification-associated hits. Other arship. The Victorian Centre for Functional Genomics (VCFG) is funded by regions may contain drivers activated by alternative the Australian Cancer Research Foundation (ACRF), the Victorian Depart- mechanisms in addition to amplification, again reducing ment of Industry, Innovation and Regional Development (DIIRD), the Australian Phenomics Network supported by funding from the Australian the capacity to detect them using an amplification-associ- Government’s Education Investment Fund through the Super Science Ini- ated criterion. In addition, there are likely biologic reasons tiative, the Australasian Genomics Technologies Association (AMATA), and why some amplicons had no hits, including amplicons the Brockhoff Foundation. The costs of publication of this article were defrayed in part by the where the driver was selected for a cancer phenotype not payment of page charges. This article must therefore be hereby marked measured in this assay such as anchorage-independent advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate growth (e.g. CCNE1), cell motility, invasion, or other inter- this fact. actions with surrounding tissues, angiogenesis, and Received November 5, 2012; revised January 13, 2013; accepted January immune evasion. Nonetheless, this study represents a 16, 2013; published OnlineFirst January 29, 2013.

References 1. Ferlay J SH, Bray F, Forman D, Mathers C, Parkin DM. Cancer 4. Soria JC, Blay JY, Spano JP, Pivot X, Coscas Y, Khayat D. Added value Incidence and Mortality Worldwide: IARC CancerBase No. 10. of molecular targeted agents in oncology. Ann Oncol 2011;22: Lyon, France: International Agency for Research on Cancer; 1703–16. 2010 [cited 2012 04/04/2012]. Available from: http://globocan. 5. Stegmeier F, Warmuth M, Sellers WR, Dorsch M. Targeted cancer iarc.fr. therapies in the twenty-ﬁrst century: lessons from imatinib. Clin Phar- 2. Cooke SL, Brenton JD. Evolution of platinum resistance in high-grade macol Ther 2010;87:543–52. serous ovarian cancer. Lancet Oncol 2011;12:1169–74. 6. Cancer Genome Atlas Research Network. Integrated genomic analy- 3. DiCosimo S, Baselga J. Management of breast cancer with targeted ses of ovarian carcinoma. Nature 2011;474:609–15. agents: importance of heterogeneity. [corrected]. Nat Rev Clin Oncol 7. Gorringe KL, George J, Anglesio MS, Ramakrishna M, Etemadmo- 2010;7:139–47. ghadam D, Cowin P, et al. Copy number analysis identiﬁes novel

1420 Clin Cancer Res; 19(6) March 15, 2013 Clinical Cancer Research

Functional Analysis of Candidate Ovarian Cancer Driver Genes

interactions between genomic loci in ovarian cancer. PLoS ONE 23. Rui L, Emre NC, Kruhlak MJ, Chung HJ, Steidl C, Slack G, et al. 2010;5:pii:e11408. Cooperative epigenetic modulation by cancer amplicon genes. Cancer 8. Gorringe KL, Jacobs S, Thompson ER, Sridhar A, Qiu W, Choong DY, Cell 2010;18:590–605. et al. High-resolution single nucleotide polymorphism array analysis of 24. Brown LA, Kalloger SE, Miller MA, Shih Ie M, McKinney SE, Santos JL, epithelial ovarian cancer reveals numerous microdeletions and ampli- et al. Amplification of 11q13 in ovarian carcinoma. Genes Chromo- fications. Clin Cancer Res 2007;13:4731–9. somes Cancer 2008;47:481–9. 9. Etemadmoghadam D, George J, Cowin PA, Cullinane C, Kansara M, 25. ShihIe M, Kurman RJ. Ovarian tumorigenesis: a proposed model Gorringe KL, et al. Amplicon-dependent CCNE1 expression is based on morphological and molecular genetic analysis. Am J Pathol critical for clonogenic survival after cisplatin treatment and is 2004;164:1511–8. correlated with 20q11 gain in ovarian cancer. PLoS ONE 2010;5: 26. Curtis C, Shah SP, Chin SF, Turashvili G, Rueda OM, Dunning MJ, et al. e15498. The genomic and transcriptomic architecture of 2,000 breast tumours 10. Theurillat JP, Metzler SC, Henzi N, Djouder N, Helbling M, Zimmer- reveals novel subgroups. Nature 2012;486:346–52. mann AK, et al. URI is an oncogene amplified in ovarian cancer cells 27. Wang Y, Sheng Q, Spillman MA, Behbakht K, Gu H. Gab2 regulates the and is required for their survival. Cancer Cell 2011;19:317–32. migratory behaviors and E-cadherin expression via activation of the 11. Gorringe KL, Campbell IG. Large-scale genomic analysis of ovarian PI3K pathway in ovarian cancer cells. Oncogene 2012;31:2512–20. carcinomas. Mol Oncol 2009;3:157–64. 28. Kumar R, Gururaj AE, Barnes CJ. p21-activated kinases in cancer. Nat 12. Beroukhim R, Getz G, Nghiemphu L, Barretina J, Hsueh T, Linhart D, Rev Cancer 2006;6:459–71. et al. Assessing the significance of chromosomal aberrations in can- 29. Preudhomme C, Roumier C, Hildebrand MP, Dallery-Prudhomme E, cer: methodology and application to glioma. Proc Natl Acad Sci U S A Lantoine D, Lai JL, et al. Nonrandom 4p13 rearrangements of the 2007;104:20007–12. RhoH/TTF gene, encoding a GTP-binding protein, in non-Hodgkin's 13. Ziauddin J, Sabatini DM. Microarrays of cells expressing defined lymphoma and multiple myeloma. Oncogene 2000;19:2023–32. cDNAs. Nature 2001;411:107–10. 30. Callow MG, Clairvoyant F, Zhu S, Schryver B, Whyte DB, Bischoff JR, 14. Brough R, Frankum JR, Sims D, Mackay A, Mendes-Pereira AM, et al. Requirement for PAK4 in the anchorage-independent growth of Bajrami I, et al. Functional viability profiles of breast cancer. Cancer human cancer cell lines. J Biol Chem 2002;277:550–8. Discov 2011;1:260–73. 31. Parsons DW, Wang TL, Samuels Y, Bardelli A, Cummins JM, DeLong L, 15. Tothill RW, Tinker AV, George J, Brown R, Fox SB, Lade S, et al. Novel et al. Colorectal cancer: mutations in a signalling pathway. Nature molecular subtypes of serous and endometrioid ovarian cancer linked 2005;436:792. to clinical outcome. Clin Cancer Res 2008;14:5198–208. 32. Kimmelman AC, Hezel AF, Aguirre AJ, Zheng H, Paik JH, Ying H, et al. 16. Etemadmoghadam D, deFazio A, Beroukhim R, Mermel C, George J, Genomic alterations link Rho family of GTPases to the highly invasive Getz G, et al. Integrated genome-wide DNA copy number and expres- phenotype of pancreas cancer. Proc Natl Acad Sci U S A 2008; sion analysis identifies distinct mechanisms of primary chemoresis- 105:19372–7. tance in ovarian carcinomas. Clin Cancer Res 2009;15:1417–27. 33. Mahlamaki EH, Kauraniemi P, Monni O, Wolf M, Hautaniemi S, Kallio- 17. Ramakrishna M, Williams LH, Boyle SE, Bearfoot JL, Sridhar A, Speed niemi A. High-resolution genomic and expression profiling reveals 105 TP, et al. Identification of candidate growth promoting genes in ovarian putative amplification target genes in pancreatic cancer. Neoplasia cancer through integrated copy number and expression analysis. 2004;6:432–9. PLoS ONE 2010;5:e9983. 34. Zhang J, Wang J, Guo Q, Wang Y, Zhou Y, Peng H, et al. LCH-7749944, 18. Zhang JH, Chung TD, Oldenburg KR. A simple statistical parameter for a novel and potent p21-activated kinase 4 inhibitor, suppresses use in evaluation and validation of high throughput screening assays. J proliferation and invasion in human gastric cancer cells. Cancer Lett Biomol Screen 1999;4:67–73. 2012;317:24–32. 19. Birmingham A, Selfors LM, Forster T, Wrobel D, Kennedy CJ, Shanks 35. Murray BW, Guo C, Piraino J, Westwick JK, Zhang C, Lamerdin J, et al. E, et al. Statistical methods for analysis of high-throughput RNA Small-molecule p21-activated kinase inhibitor PF-3758309 is a potent interference screens. Nat Methods 2009;6:569–75. inhibitor of oncogenic signaling and tumor growth. Proc Natl Acad Sci 20. Griffiths CT. Surgical resection of tumor bulk in the primary treatment of U S A 2010;107:9446–51. ovarian carcinoma. Natl Cancer Inst Monogr 1975;42:101–4. 36. Siu MK, Chan HY, Kong DS, Wong ES, Wong OG, Ngan HY, et al. p21- 21. Hoskins WJ, Bundy BN, Thigpen JT, Omura GA. The influence of activated kinase 4 regulates ovarian cancer cell proliferation, migra- cytoreductive surgery on recurrence-free interval and survival in small- tion, and invasion and contributes to poor prognosis in patients. Proc volume stage III epithelial ovarian cancer: a Gynecologic Oncology Natl Acad Sci U S A 2010;107:18622–7. Group study. Gynecol Oncol 1992;47:159–66. 37. Gerlinger M, Rowan AJ, Horswell S, Larkin J, Endesfelder D, Gronroos 22. Hoskins WJ, McGuire WP, Brady MF, Homesley HD, Creasman WT, E, et al. Intratumor heterogeneity and branched evolution revealed by Berman M, et al. The effect of diameter of largest residual disease on multiregion sequencing. N Engl J Med 2012;366:883–92. survival after primary cytoreductive surgery in patients with suboptimal 38. Shah SP, Roth A, Goya R, Oloumi A, Ha G, Zhao Y, et al. The clonal and residual epithelial ovarian carcinoma. Am J Obstet Gynecol 1994; mutational evolution spectrum of primary triple-negative breast 170:974–9. cancers. Nature 2012;486:395–9.

www.aacrjournals.org Clin Cancer Res; 19(6) March 15, 2013 1421

Functional Analysis of Genes in Regions Commonly Amplified in High-Grade Serous and Endometrioid Ovarian Cancer

Sally J. Davis, Karen E. Sheppard, Richard B. Pearson, et al.

Clin Cancer Res 2013;19:1411-1421. Published OnlineFirst January 29, 2013.

Updated version Access the most recent version of this article at: doi:10.1158/1078-0432.CCR-12-3433

Supplementary Access the most recent supplemental material at: Material http://clincancerres.aacrjournals.org/content/suppl/2013/01/29/1078-0432.CCR-12-3433.DC1

Cited articles This article cites 37 articles, 9 of which you can access for free at: http://clincancerres.aacrjournals.org/content/19/6/1411.full#ref-list-1

Citing articles This article has been cited by 8 HighWire-hosted articles. Access the articles at: http://clincancerres.aacrjournals.org/content/19/6/1411.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at Subscriptions [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/19/6/1411. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.