A Comparative Genomic Approach for Identifying Synthetic Lethal Interactions in Human Cancer

Published OnlineFirst August 26, 2013; DOI: 10.1158/0008-5472.CAN-12-3956

Cancer Priority Report Research

Raamesh Deshpande1, Michael K. Asiedu3, Mitchell Klebig3,4, Shari Sutor3, Elena Kuzmin5, Justin Nelson2, Jeff Piotrowski7, Seung Ho Shin2, Minoru Yoshida6, Michael Costanzo5, Charles Boone5,6, Dennis A. Wigle2,3, and Chad L. Myers1,2

Abstract Synthetic lethal interactions enable a novel approach for discovering speciﬁc genetic vulnerabilities in cancer cells that can be exploited for the development of therapeutics. Despite successes in model organisms such as yeast, discovering synthetic lethal interactions on a large scale in human cells remains a signiﬁcant challenge. We describe a comparative genomic strategy for identifying cancer-relevant synthetic lethal interactions whereby candidate interactions are prioritized on the basis of genetic interaction data available in yeast, followed by targeted testing of candidate interactions in human cell lines. As a proof of principle, we describe two novel synthetic lethal interactions in human cells discovered by this approach, one between the tumor suppressor gene SMARCB1 and PSMA4, and another between alveolar soft-part sarcoma-associated ASPSCR1 and PSMC2. These results suggest therapeutic targets for cancers harboring mutations in SMARCB1 or ASPSCR1 and highlight the potential of a targeted, cross-species strategy for identifying synthetic lethal interactions relevant to human cancer. Cancer Res; 73(20); 6128–36. 2013 AACR.

Introduction essential gene in normal cells, BRCA mutant cells are depen- PARP fi Synthetic lethality is an exciting new avenue to disrupt dent on for their survival. The described ef cacy of an PARP cancer cells for targeted treatment. Two genes are said to be oral inhibitor, olaparib (AZD2281), in early-phase clinical BRCA synthetic lethal if mutations in both genes cause cell death, but trials for treating mutant tumors is a remarkable success a mutation in either of them alone is not lethal. In applying story for translational cancer therapeutics (2). Importantly, synthetic lethality to the discovery of cancer drugs, the goal strategies based on synthetic lethal interactions enable drug fi would be to identify a target gene that when mutated or targeting of cancer-speci c alterations in tumor suppressors chemically inhibited, kills cells that harbor a specific cancer- that might otherwise be untreatable by drugs. Several recent related alteration, but spares otherwise identical cells lacking studies have reported large-scale assays on the basis of on the cancer-related alteration (1). This concept has recently RNA-interference (RNAi) technology to discover synthetic been exploited in the development of PARP inhibitors as novel lethal interactions with common cancer mutations, including BRCA1/2 RAS – chemotherapeutics for breast cancer. Although PARP is not an and genes (3 9). These studies typically target cells with a well-defined genetic background using a library of short hairpin RNAs (shRNA) to identify combinations that result in cell death or growth inhibition. Although such app- Authors' Affiliations: 1Department of Computer Science and Engineering; 2Program in Biomedical Informatics and Computational Biology, University roaches have the potential to rapidly discover genetic inter- of Minnesota, Minneapolis; 3Department of Surgery, Mayo Clinic, Roche- actions at a full genome scale, a number of technologic 4 ster, Minnesota; Department of Laboratory Medicine & Pathology, Molec- challenges remain to be solved, and the number of indepen- ular Genetics Lab, Mayo Clinic, Rochester, Minnesota; 5Department of Molecular Genetics, Terrence Donnelly Centre for Cellular and Biomolec- dently validated interactions produced by these efforts has ular Research, University of Toronto, Ontario, Canada; 6Chemical Geno- been relatively limited to date (10). mics Research Group, RIKEN Advance Science Institute, Saitama, Japan; One complementary strategy to whole-genome screens in and 7Great Lakes Bioenergy Research Center, University of Wisconsin, Madison, Wisconsin cancer cell lines is motivated by the wealth of potentially relevant interaction data in model organisms. Publication of Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/). the first eukaryotic genome-scale genetic interaction map in yeast (Saccharomyces cerevisiae;ref.11),whereapproximate- R. Deshpande and M.K. Asiedu contributed equally to this work. ly 30% of all possible gene pairs were tested for interactions, Corresponding Authors: Chad. L. Myers, University of Minnesota - Twin provides a unique opportunity for discovering potentially Cities, 200 Union Street SE, Minneapolis, MN 55455. Phone: 612-624- 8306; Fax: 612-625-0572; E-mail: [email protected]; and Dennis A. therapeutic synthetic lethal interactions. For example, puta- Wigle, Department of Surgery, Mayo Clinic, 200 First Street SW, Rochester, tive synthetic lethal interactions in human could be inferred MN 55905. E-mail: [email protected] on the basis of yeast synthetic lethal interactions between doi: 10.1158/0008-5472.CAN-12-3956 conserved genes in yeast and human. These predicted 2013 American Association for Cancer Research. pairs of human genes provide a rich database of possible

6128 Cancer Res; 73(20) October 15, 2013

Comparative Genomics Approach to Cancer Synthetic Lethality

candidates for further study in the context of human disease. Luminescence Cell Viability Assay (Promega). This assay In fact, several interactions related to chromosome stability determines the number of viable cells in culture based on have already been mapped from yeast to worm to human the amount of ATP produced by the living cells and is (12), suggesting that such a strategy has the potential to yield designed for use with multiwell plate formats and high- promising new drug targets. In combination with the expo- throughput screening for cell proliferation and cytotoxicity nentially accumulating volumeofdataregardingtheland- assays. The addition of the assay reagent results in cell lysis scape of genomic alterations in human cancer, such an and generation of a luminescent signal proportional to the approach has the potential to become increasingly powerful amount of ATP present, which is directly proportional to the going forward. number of living cells present in each well. The intensity of We describe a combined computational and experimental the luminescent signal was measured in relative lumines- approach whereby yeast interactions between human ortho- cence units (RLU) using the Beckman Coulter DTX 880 logs are filtered by cancer association and interaction multimode plate reader. strength in yeast, and candidates from the prioritized list are then validated in human cell lines. Using this approach, we Immunoblotting discovered two previously unknown synthetic sick interac- Cell lysates from 293, A-204, G-401, and IMR90 control cells, tions: one between SMARCB1 (yeast SNF5)andPSMA4 (yeast and from shRNA-infected cells were extracted after five days of PRE9), and another between ASPSCR1 (yeast UBX4)and incubation and quantified using Bio-Rad Protein Assay PSMC2 (yeast RPT1). The predicted synthetic sick/lethal reagent. An equal amount of protein (50 mg) was subjected interactions between these genes were validated with shRNA to SDS–PAGE, transferred onto a polyvinylidene difluoride double knockdown in multiple cell lines and single knock- membrane and blocked with nonfat milk. The membranes down of PSMA4 in two cancer cell lines containing endoge- were then incubated in primary antibody overnight at 4C and nous SMARCB1 mutations. These interactions suggest poten- then with anti-mouse (1:6,000) or anti-rabbit (1:1,000) second- tially new therapeutic targets for SMARCB1 and ASPSCR1 ary antibody at room temperature for one hour. Primary mutated cancers and, more broadly, illustrate the potential antibodies rabbit anti-SNF5 (SMARCB1) and rabbit anti-TUG of this cross-species approach. (ASPSCR1) were purchased from Cell Signaling, whereas rabbit anti-PSMC2, 20S Proteosome a-4 (PSMA4), and anti-b-actin- Materials and Methods HRP were obtained from Santa Cruz Biotechnology. Protein Cell culture and shRNAs expression was detected using enhanced chemoluminiscence IMR90, 293TN, A-204, G-401, and 293 cell lines were obtained substrate (Pierce). from the American Type Culture Collection. IMR90, 293TN, and 293 cells were maintained in Dulbecco's Modified Eagle's Estimation of the significance of genetic interactions in Medium (DMEM) supplemented with 10% FBS, penicillin, and human shRNA experiments streptomycin. A-204 and G-401 cells were cultured in McCoy's Growth rate of a single or double shRNA knockdown relative 5A media containing 10% FBS, penicillin, and streptomycin. to the empty shRNA vector control were calculated using fi Bacterial stocks of control and validated gene-speci c shRNA- RLU Count expressing vectors, including PSMA4 and SMARCB1 shRNAs, f ¼ A A RLU Count were selected from the RNAi consortium database and pur- Control chased from Sigma-Aldrich. where fA is relative growth rate for a single or double knockdown experiment (A, B, or AB), and RLUCountA is the intensity Preparation of viral particles of the luminescent signal measured in relative luminescence Bacterial stocks of validated shRNAs clones were amplified units (RLU). Because fA is a ratio of two quantities that has and DNA extracted using the HiSpeed Plasmid purification kit error associated with it, error for fA is given by (Qiagen). 293TN cells were then transfected with shRNA vector sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi clones mixed with viral package vectors pMD2 and psPAX2 2 2 DRLU CountA DRLU CountControl using Lipofectamine 2000 transfection reagent (Invitrogen). sA ¼ þ fA After 48 hours, culture media containing viral particles were RLU CountA RLU CountControl mixed with polybrene and centrifuged at 10,000 rpm to pre- cipitate and concentrate the viral particles. where DRLU CountA is the SD in the n (6 for our experiment) observations of RLU CountA. RNAi-mediated gene knockdown Expected double mutant fitness and error associated with it IMR90 cells were seeded into 96-well plates and trans- is given by 0 duced with predetermined pairs of shRNAs to generate four fAB ¼ fA fB conditions with six replicates each: (i) control shRNAs; (ii) sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2 2 control shRNA/PSMA4; (iii) control shRNA/SMARCB1;and 0 sA sB 0 sAB ¼ þ fAB (iv) PSMA4/SMARCB1. A similar set-up of four conditions fA fB were used for other pairs of interactions tested. Cells from each treatment were cultured for eight and 10 days, and the To assess the significance of the interaction, we assumed a 0 0 number of viable cells determined by the CellTiter-Glo normal distribution for fAB and fAB, and compared fAB with fAB

www.aacrjournals.org Cancer Res; 73(20) October 15, 2013 6129

Deshpande et al.

using the Welch t test (13). The significance of the difference were constructed in a SSL204 MATa strain and crossed with an 0 between fAB and fAB can be calculated using a one-tailed t test, isogenic rpt1-1 MATa strain. Diploid cells were sporulated at which requires the t test score t (13) and degree of freedom n 25C and dissected. Plates were incubated for 3 to 5 days at given by the Welch–Satterthwaite equation (14), as follows: either 25Cor30C. f 0 f t ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiAB AB 0 Yeast tetrad dissection s AB2 sAB2 0 þ n AB nAB Confirmations by tetrad analyses were conducted as described previously (17). 0 2 s AB2 sAB2 0 þ nAB nAB n ¼ 0 s AB4 sAB4 0 0 þ n AB2ðn AB1Þ nAB2ðnAB1Þ

0 Here, nAB ¼ nAB ¼ 6 because we have six replicate observations for control, single, and double knockdown experiments.

Orthology mapping InParanoid7 (15) was used to map yeast genes to human genes. Only 1:1 orthologs were used for our study (Supplemen- tary Table S1).

Collection and processing of yeast genetic interaction data Yeast genetic interaction data was taken from the report by Costanzo and colleagues in 2010 (11), which reported data for interactions between 1,711 query genes and 3,885 array genes. We applied a P value cutoff of less than 0.05 on all interactions. Furthermore, we applied an interaction cutoff in two ways: first, we considered stringent negative genetic interactions (e<0.2) and, second, we allowed intermediate interactions (e<0.08), which were reported in reciprocal screens. Spe- cifically, in the Costanzo and colleagues network, query genes were screened against the entire nonessential deletion array and, in some cases, genes present on the array were also screened as queries. For these cases, an interaction between genes A and B was tested in both screens: A (query) B (array) and B (query) A (array). In such cases, we applied an intermediate cutoff because an interaction appearing in both of these screens is of high confidence. Eleven new SGA screens were also used to generate candidate gene pairs, including screens for the following queries (human/yeast orthologs): XPC/RAD4, VTI1A/VTI1, NOP56/ NOP56, POLD2/POL31, MLH1/MLH1, XPO1/CRM1, UBA3/UBA3, ERCC4/RAD1, XPA/RAD14, PSMC2/RPT1, and PSMB1/PRE7. A screen involving a temperature-sensitive (TS) allele of yeast RPT1 (human PSMC2) was the basis for testing the human interaction PSMC2–ASPSCR1; therefore, the yeast interaction Figure 1. Comparative genomic approach for discovering cancer-related synthetic sick/lethal interactions in human. A, flowchart describing steps data supporting that inference are included here (Supplemen- to use the wealth of synthetic sick/lethal interactions available in yeast tary Table 2). A genome-wide screen for the RPT1 TS allele's and knowledge of genes commonly mutated in cancer (Sanger Institute genetic interactions was conducted as described by Baryshni- Cancer Gene Census) for discovery of novel cancer drug targets in kova and colleagues in 2010 (16). Briefly, a rpt1-1 mutant strain human. B, summary of yeast synthetic sick/lethal interaction network statistics and mapping of interactions between human orthologs. The marked with a nourseothricin (NatMX4) resistance cassette fi fi "Complete set" contains all signi cant synthetic sick or lethal interaction and harboring the SGA haploid-speci c markers and reporter pairs at an intermediate confidence cutoff as described previously (16) was mated to an array of approximately 4,000 viable S. (e<0.08; P < 0.05; ref. 11), and human totals include any genes with cerevisiae deletion mutants. Nourseothricin- and Geneticin- human orthologs. The "Filtered set" contains only high confidence e< P < resistant heterozygous diploid mutants were selected and interactions ( 0.2; 0.05) or interactions replicated in two independent experiments, and human totals include only gene pairs with sporulated and MATa rpt1-1 double mutants were subsequent- one-to-one orthologs (see Materials and Methods—Collection and ly selected (16). To confirm the SGA results, all gene deletions processing of yeast genetic interaction data).

6130 Cancer Res; 73(20) October 15, 2013 Cancer Research

Comparative Genomics Approach to Cancer Synthetic Lethality

Results tially buffering effects of paralog functional redundancy (Sup- To discover cancer-associated genetic interactions in plementary Table S1; ref. 19). Applying these relatively strin- human cells, we first selected a set of highly significant int- gent criteria, we obtained 1,522 putative synthetic sick/lethal eractions between yeast genes from the large network of interactions between human orthologs of yeast genes, of which synthetic genetic interactions that has recently been mapped 70 interactions involved a gene that has been previously in yeast. A recent study reported testing genetic interactions implicated in some form of human cancer (Fig. 1B; Supple- for 5.4 million yeast gene pairs, consisting of instances where mentary Table S4). In addition to these published interactions, two nonessential genes were deleted in combination, or a we applied the same criteria to 11 previously unpublished yeast temperature-sensitive mutation of an essential gene was used screens involving human orthologs (see Materials and Meth- together with a deletion of a nonessential gene (11). In total, ods; Supplementary Table S2). Candidate interaction pairs approximately 116,000 pairs were reported as having a detect- involving cancer-associated mutations (Sanger Institute Can- able synthetic sick or lethal interaction, of which approxi- cer Gene Census) were ranked on the basis of the strength of mately 24,000 interactions connect two genes that both have the yeast interactions, and were selected in order up to a human orthologs (Fig. 1). More than 500 of these latter maximum of three interactions per gene. In total, 21 pairs of interactions involve at least one gene that has been previously genes representing mutations associated with a diverse set of associated with mutations in cancer (Sanger Institute Cancer cancers were selected for further experiments in human cell Gene Census; Fig. 1B; ref. 18), suggesting a large number of lines (Fig. 2). candidate pairs can be generated by this approach (Supple- The candidate synthetic sick or lethal pairs derived from the mentary Table S3). yeast genetic interaction network were screened in normal To narrow the candidate list for testing in human cells, we human IMR90 fibroblast cells using an RNAi approach. IMR90 first applied a very stringent cutoff on interactions in yeast, cells were chosen because the cell line was established from either requiring a high-magnitude effect, high-confidence the lungs of a 16-week female fetus and have the advantage of interaction to be reported (e<0.2; P < 0.05) or selecting early passage and a low likelihood of accumulated genetic gene pairs for which interactions were reproduced in two alterations. This stable genetic background allowed us to reciprocal screens (see Materials and Methods for details). assess the validity of candidate interactions with the lowest Furthermore, we restricted our search to genes with one-to- possibility of unknown, confounding genetic alterations. We one orthologs in human to increase the likelihood of functional screened the selected 21 pairs of potential interactions using a conservation between yeast and human, and to avoid poten- CellTiter-Glo luminescence viability assay. We found evidence

Figure 2. Interaction testing of 21 selected candidate synthetic sick/lethal interactions in human fibroblast cell lines. A, the interaction scores for all human interactions tested. The interaction score is the difference between observed and expected growth rate based on a multiplicative model. The significant negative genetic interactions are shown as dark bars and the strength of the significance is denoted by the number of asterisks according to the legend shown. B, results for each significant interaction tested. The number of days the fibroblast cells were grown in the presence of shRNAs is indicated in each plot. The error bars for both A and B represent twice the width of the standard error in the interaction scores and growth rates.

www.aacrjournals.org Cancer Res; 73(20) October 15, 2013 6131

Deshpande et al.

Figure 3. Yeast data for the candidate synthetic sick/lethal interaction between SNF5-PRE9 (human SMARCB1- PSMA4) and UBX4-RPT1 (human ASPSCR1-PSMC2). A, fitness of the single and double mutants relative to wild-type for the SNF5-PRE9 interaction. The interaction score (e) was estimated by comparing the observed double mutant fitness with the fitness expected based on the single mutant fitness. B, confirmation of the synthetic sick/lethal interaction using tetrad dissection analysis for the SNF5-PRE9 double mutant. Each tetrad is oriented horizontally and represents four meiotic progeny of a heterozygous double mutant between pre9D::natMX4/PRE9 and snf5D::kanMX4/SNF5. Four representative tetrads are shown. The genes knocked out are identified by the presence of the natMX and kanMX markers, respectively. The identified double knock-out spore colonies are enclosed in circles whereas single gene knockout strains are enclosed in squares or diamonds, and wild-type strains are not enclosed. C and D present similar data for query mutant rpt1-1, a temperature-sensitive conditional mutant of RPT1, and yeast ubx4D.

for significant synthetic sick or lethal interactions for 6 of which has been widely used in the genetic interaction com- the 21 tested pairs (see Supplementary Fig. S1 for data and munity (see Materials and Methods for details; ref. 20). Similar Supplementary Fig. S2 for significant interactions). We focused results were observed when different shRNA clones for PSMA4/ further validation efforts on the strongest 2 of the 6 significant SMARCB1 and ASPSCR1/PSMC2 knockdown were used (Fig. 4C interactions: SMARCB1/PSMA4 and ASPSCR1/PSMC2 (panels 1 and F). In addition, we confirmed the effectiveness of shRNA and 2 for Fig. 2B). silencing of the targeted genes by conducting protein expres- To further validate these two interactions, we first retested sion analyses using Western blots (Fig. 4A and D), which them in yeast cells by dissecting tetrads (Fig. 3), which indeed showed greatly reduced protein levels in the shRNA-infected confirmed a strong synthetic sick effect between the pairs cells. of yeast orthologs, SNF5/PRE9 (human SMARCB1/PSMA4) and The discovery of cancer-related synthetic lethal interactions UBX4/RPT1 (human ASPSCR1/PSMC2; Fig. 3). In human cells, can directly impact therapeutic potential, as the synthetic we repeated the same viability assay and additionally con- lethal interactor of a cancer related gene can be targeted ducted knockdowns with independent targeting shRNAs selectively to kill cancer cells. To test the clinical relevance of for both pairs of genes. After simultaneous depletion of the the PSMA4 and SMARCB1 interactions, we identified an epi- targeted gene pairs, the number of cells that survived was thelial muscle rhabdosarcoma cell line (A-204) and a renal significantly reduced in all cases (Fig. 4B, C, E, and F). Impor- rhabdoid sarcoma cell line (G-401), each harboring SMARCB1 tantly, the extent of survival was significantly lower than the mutations, and used embryonic kidney HEK-293 cells expres- expected survival of double knockdowns estimated from the sing wild-type SMARCB1 as a control (Fig. 5). We observed single shRNA effects (Fig. 4B of PSMA4/SMARCB1; Welch t test; that PSMA4 knockdown almost completely kills the cell lines Score ¼ 8.11 at day 8, and 8.90 at day 10; Fig. 4E for ASPSCR1- harboring SMARCB1 mutations and that this effect was PSMC2; Welch t test score ¼ 14.86 at day 7 and 20.95 at day 10; exaggerated compared with controls when following the P < 0.0001 in all cases; ref. 13). Expected double knockdown cells to later time points (day 7, Fig. 5A–D). In addition, we effects were calculated assuming a multiplicative null model, showed the complete absence of SMARCB1 protein in cell

6132 Cancer Res; 73(20) October 15, 2013 Cancer Research

Comparative Genomics Approach to Cancer Synthetic Lethality

Figure 4. Validation of two candidate synthetic sick/lethal interactions in human ﬁbroblast cell lines. A, a Western blot of IMR90 cells transduced with PSMA4 and SMARCB1 shRNA virus showing downregulation of respective protein expression. B and C, cell viability analyses of PSMA4 and SMARCB1 interaction using two different clones that showed decreased cell survival in cells depleted of both PSMA4 and SMARCB1 compared with cells expressing shRNAs of individual genes and compared with the expected effect from depletion of both genes. D, immunoblotting showing knockdown of PSMC2 and ASPSCR1 expression in IMR90 cells treated with viral particles encoding PSMC2 or ASPSCR1 shRNA. E and F, synthetic lethal interaction effect of PSMC2 and ASPSCR1 in IMR90 cells as shown by signiﬁcantly decreased survival in cells expressing shRNAs of both genes compared with the expected effect from depletion of both genes. lines A-204 and G-401 by Western blot analysis (Fig. 5E). A- strategy to large-scale RNAi screens that are in progress by 204 carries a TC deletion of codons 181 and 182 in exon 5 several other groups. whereas G-401 harbors a homozygous deletion of exons 1 to On the basis of the synthetic sick/lethal interaction, we 9 (21). In both cell lines harboring SMARCB1 mutation, the discovered between SMARCB1 and PSMA4, we hypothesize decrease in growth is greater than expected by the multi- that targeting PSMA4 in therapeutic approaches could selec- plicative combination of the individual SMARCB1 mutation tively inhibit the growth of cancer cells harboring SMARCB1 and PSMA4 knockdown effects, as estimated from the con- mutations. Human PSMA4 is a proteasome subunit component trol cell line (Fig. 5; P < 2.5 10 6 for all days, all replicates, expressed across numerous tissues. PSMA4 mRNA levels are andbothcelllines). increased in lung tumors compared with normal lung tissues, and downregulation of PSMA4 expression in lung cancer cell Discussion lines decreases proteasome activity and induces apoptosis (22). We describe an experimental pipeline where we prioritized Human SMARCB1 is a core component of the BAF ATP-dep- synthetic genetic interactions from the global map of yeast endent chromatin-remodeling complex, known to play impor- interactions to test candidate synthetic sick/lethal pairs involv- tant roles in cell proliferation and differentiation, and inhibi- ing cancer-associated mutations in human cells. We propose tion of tumor formation. Deletions in SMARCB1 are associated this general approach, involving computational prioritization with epitheliod sarcomas (23), and are a known cause of followed by experimental validation, as a complementary rhabdoid tumor predisposition syndrome (RTPS), a highly

www.aacrjournals.org Cancer Res; 73(20) October 15, 2013 6133

Deshpande et al.

Figure 5. Validation of the PSMA4-SMARCB1 synthetic lethal interaction in cancer cell lines harboring SMARCB1 loss-of-function mutations. A, cell viability analyses of cell lines with (A-204) or without (293) endogenous SMARCB1 mutation, grown with or without PSMA4 shRNA knock-down (shPSMA4-1), show the therapeutic potential for this cancer-associated synthetic lethal interaction. B, the experiment in A is repeated with a different PSMA4 shRNA construct (shPSMA4-2). C and D, the experiment in A and B is repeated with a different SMARCB1-deﬁcient cell line, G-401. E, a Western blot showing the complete absence of SMARCB1 protein in cell lines A-204 and G-401, which have endogenous null SMARCB1 mutations, and normal expression of SMARCB1 in the control 293 cell line. The endogenous PSMA4 and b-actin protein levels detected serve as loading controls.

6134 Cancer Res; 73(20) October 15, 2013 Cancer Research

Comparative Genomics Approach to Cancer Synthetic Lethality

malignant group of neoplasms that usually occur in early The appeal of a targeted approach for identifying synthetic childhood (24, 25). No described direct protein interaction sick/lethal interaction candidates is strengthened by the fact exists between PSMA4 and SMARCB1. Although the clinical that there are currently large numbers of tumor genome implications of this synthetic lethal interaction await further sequencing efforts in progress that will produce new, poten- study, one potential application could be in the use of existing tially lengthy lists of mutations associated with various types of proteasome inhibitors such as bortezomib for the treatment of cancers. As we gain richer knowledge of the spectrum of tumors harboring SMARCB1 mutations. Interestingly, interac- mutations present in cancer, we can continue to directly screen tions between other SWI/SNF subunits and the proteasome the most promising candidate synthetic sick/lethal interac- were also observed in yeast (11), suggesting the possibility that tions involving these genes. The identification of specific perturbations in multiple combinations of subunits across cancer subtypes harboring specific mutations may provide these complexes could have the same effect. Whether inter- therapeutic opportunities for synthetic lethal approaches that actions exist in human between other genes encoding the SWI– are not currently appreciated. Furthermore, in future studies, SNF complex and the proteasome remains to be determined, we intend to leverage data beyond sequence-similarity and but this merits further study because mutations in other literature-derived functional information to prioritize interac- subunits of SWI/SNF have been observed in many other types tions for testing across species. For example, the large collec- of cancer (26, 27). tions of functional genomic data in both yeast and human The direct clinical implications of the ASPSCR1–PSMC2 could allow for a more robust and unbiased assessment of the synthetic sick interaction are less clear, but this case also likelihood of functional conservation of genes and conserved merits further study. ASPSCR1 is a relatively uncharacterized synthetic lethal interactions between them. Our initial results gene that has been associated with alveolar soft-part sarcoma highlight the feasibility of this comparative genomic approach, (ASPS), a rare class of tumors that typically occur in younger and suggest its potential use for rapid translation of novel patients (28). Most cases of this cancer are associated with an sequence variants into new therapeutic targets. We believe this unbalanced translocation der(17)t(X;17) (p11;q25) that results approach has the potential to provide a dramatic increase in in an ASPSCR1–TFE3 fusion protein. The fusion protein the number of therapeutic targets beyond those currently appears to act as an aberrant transcription factor, inducing available for drug development. unregulated transcription of TFE3-regulated genes (28, 29). This fusion truncates one ASPSCR1 allele, leaving the other Disclosure of Potential Conflicts of Interest fl allele intact in most cases (28, 29). How this genetic interaction No potential con icts of interest were disclosed. could be leveraged for therapeutic purposes awaits further Disclaimer investigation, but one possibility is the potential combined The funding entities had no role in study design, data collection and analysis, effect of reduced expression of ASPSCR1 in conjunction with decision to publish, or preparation of the article. proteasome inhibition. Authors' Contributions Interestingly, the two strongest synthetic sick/lethal interac- Conception and design: R. Deshpande, M.K. Asiedu, M. Klebig, M. Yoshida, D.A. tions we observed involved components of the proteasome, Wigle, C.L. Myers although we tested a variety of genes from multiple pathways Development of methodology: R. Deshpande, M.K. Asiedu, M. Klebig, S. Sutor, D.A. Wigle, C.L. Myers that were produced by our approach. These data suggest the Acquisition of data (provided animals, acquired and managed patients, proteasome may be a rich target for synthetic lethal approaches provided facilities, etc.): M.K. Asiedu, E. Kuzmin, C. Boone, D.A. Wigle Analysis and interpretation of data (e.g., statistical analysis, biostatistics, in human cancer therapy, and indeed, successful cancer treat- computational analysis): R. Deshpande, M.K. Asiedu, E. Kuzmin, J. Nelson, ment involving proteasome inhibitors has been reported J. Piotrowski, C. Boone, D.A. Wigle, C.L. Myers recently in a number of different contexts (30). The availability Writing, review, and/or revision of the manuscript: R. Deshpande, M. Asiedu, M. Klebig, J. Piotrowski, S.H. Shin, M. Costanzo, C. Boone, D.A. Wigle, of several approved proteasome inhibitors may make such C.L. Myers interactions between cancer-associated mutations and the Administrative, technical, or material support (i.e., reporting or orga- proteasome immediately translatable to several clinical set- nizing data, constructing databases): M.K. Asiedu, M. Klebig, S. Sutor, D.A. Wigle, C.L. Myers tings. These results also highlight the potential for discovering Study supervision: M. Yoshida, D.A. Wigle, C.L. Myers interactions within core biologic pathways with strong yeast/ human homology. Importantly however, one of the limitations Grant Support of our approach is its dependence on genes and proteins with This work was financially supported by a grant from the Minnesota Part- fl nership for Biotechnology and Medical Genomics program to C.L. Myers and D.A. such homology, and an inability to re ect many known onco- Wigle. R. Deshpande was funded by a University of Minnesota Doctoral Disser- genic pathways where yeast/human homology does not exist. In tation Fellowship and Biomedical Informatics and Computational Biology addition, we note that a recent study identified both PSMA4 and (BICB) traineeship. C.L. Myers and R. Deshpande are also partially supported PSMC2 by a grant from the NIH (1R01HG005084-01A1) and a grant from the National as two of a set of 56 genes (and the only proteasomal Science Foundation (DBI 0953881). C. Boone is supported by the Canadian components) for which gene knockdown inhibited the growth Institutes of Health Research (grant nos. MOP-102629, MOP-97939, and MOP- 57830), the Ontario Research Fund (grant no. GL2-01-22), and the NIH (grant no. of cells with partial copy number loss in the same gene (31). Our 1R01HG005853-01). S.H. Shin was supported by a BICB fellowship. C. Boone, M. independent finding of synthetic sick/lethal interactions for Yoshida, and J. Piotrowski are supported by the RIKEN President's Discretionary these same proteasomal subunits is intriguing and suggests that Fund. C.L. Myers and C. Boone are partially supported by the Canadian Institute perturbations of these subunits have a relatively unique effect for Advanced Research (CIFAR) Genetic Networks Program. on proteasome function that may not be replicated by manip- Received October 19, 2012; revised July 14, 2013; accepted August 5, 2013; ulation of its other components. published OnlineFirst August 26, 2013.

www.aacrjournals.org Cancer Res; 73(20) October 15, 2013 6135

Deshpande et al.

References 1. Hartwell LH, Szankasi P, Roberts CJ, Murray AW, Friend SH. Inte- romyces cerevisiae and Schizosaccharomyces pombe. Methods grating genetic approaches into the discovery of anticancer drugs. Enzymol 2010;470:145–79. Science 1997;278:1064–8. 17. Amberg DC, Burke DJ, Strathern JN. Tetrad dissection. Cold Spring 2. Tutt A, Robson M, Garber JE, Domchek SM, Audeh MW, Weitzel JN, Habor Protoc 2006; doi:10.1101/pdb.prot4181. et al. Oral poly(ADP-ribose) polymerase inhibitor olaparib in patients 18. Futreal PA, Coin L, Marshall M, Down T, Hubbard T, Wooster R, et al. with BRCA1 or BRCA2 mutations and advanced breast cancer: a A census of human cancer genes. Nat Rev Cancer 2004;4:177–83. proof-of-concept trial. Lancet 2010;376:235–44. 19. VanderSluis B, Bellay J, Musso G, Costanzo M, Papp B, Vizeacoumar 3. Scholl C, Frohling S, Dunn IF, Schinzel AC, Barbie DA, Kim SY, et al. FJ, et al. Genetic interactions reveal the evolutionary trajectories of Synthetic lethal interaction between oncogenic KRAS dependency duplicate genes. Mol Syst Biol 2010;6:429. and STK33 suppression in human cancer cells. Cell 2009;137:821–34. 20. Mani R, St Onge RP, Hartman JL, Giaever G, Roth FP. Defining genetic 4. Luo J, Emanuele MJ, Li D, Creighton CJ, Schlabach MR, Westbrook interaction. Proc Natl Acad Sci U S A 2008;105:3461–6. TF, et al. A genome-wide RNAi screen identifies multiple synthetic 21. DeCristofaro MF, Betz BL, Wang W, Weissman BE. Alteration of lethal interactions with the Ras oncogene. Cell 2009;137:835–48. hSNF5/INI1/BAF47 detected in rhabdoid cell lines and primary rhab- 5. Barbie DA, Tamayo P, Boehm JS, Kim SY, Moody SE, Dunn IF, et al. domyosarcomas but not Wilms' tumors. Oncogene 1999;18:7559–65. Systematic RNA interference reveals that oncogenic KRAS-driven 22. Liu Y, Liu P, Wen W, James MA, Wang Y, Bailey-Wilson JE, et al. cancers require TBK1. Nature 2009;462:108–12. Haplotype and cell proliferation analyses of candidate lung cancer 6. Bryant HE, Schultz N, Thomas HD, Parker KM, Flower D, Lopez E, et al. susceptibility genes on chromosome 15q24-25.1. Cancer Res 2009; Specific killing of BRCA2-deficient tumours with inhibitors of poly 69:7844–50. (ADP-ribose) polymerase. Nature 2005;434:913–7. 23. Modena P, Lualdi E, Facchinetti F, Galli L, Teixeira MR, Pilotti S, et al. 7. Fong PC, Boss DS, Yap TA, Tutt A, Wu P, Mergui-Roelvink M, et al. SMARCB1/INI1 tumor suppressor gene is frequently inactivated in Inhibition of poly(ADP-ribose) polymerase in tumors from BRCA muta- epithelioid sarcomas. Cancer Res 2005;65:4012–9. tion carriers. N Engl J Med 2009;361:123–34. 24. Schneppenheim R, Fruhwald MC, Gesk S, Hasselblatt M, Jeibmann A, 8. Marcotte R, Brown KR, Suarez F, Sayad A, Karamboulas K, Krzyza- Kordes U, et al. Germline nonsense mutation and somatic inactivation nowski PM, et al. Essential gene profiles in breast, pancreatic, and of SMARCA4/BRG1 in a family with rhabdoid tumor predisposition ovarian cancer cells. Cancer Discov 2012;2:172–89. syndrome. Am J Human Genet 2010;86:279–84. 9. Cheung HW, Cowley GS, Weir BA, Boehm JS, Rusin S, Scott JA, et al. 25. Lee RS, Stewart C, Carter SL, Ambrogio L, Cibulskis K, Sougnez C, Systematic investigation of genetic vulnerabilities across cancer cell et al. A remarkably simple genome underlies highly malignant pediatric lines reveals lineage-specific dependencies in ovarian cancer. Proc rhabdoid cancers. J Clin Invest 2012;122:2983–8. Natl Acad Sci U S A 2011;108:12372–7. 26. Weissman B, Knudsen KE. Hijacking the chromatin remodeling 10. Iorns E, Lord CJ, Turner N, Ashworth A. Utilizing RNA interference to machinery: impact of SWI/SNF perturbations in cancer. Cancer Res enhance cancer drug discovery. Nat Rev Drug Discov 2007;6:556–68. 2009;69:8223–30. 11. Costanzo M, Baryshnikova A, Bellay J, Kim Y, Spear ED, Sevier CS, 27. Reisman D, Glaros S, Thompson EA. The SWI/SNF complex and et al. The genetic landscape of a cell. Science 2010;327:425–31. cancer. Oncogene 2009;28:1653–68. 12. McManus KJ, Barrett IJ, Nouhi Y, Hieter P. Specific synthetic lethal 28. Folpe AL, Deyrup AT. Alveolar soft-part sarcoma: a review and update. killing of RAD54B-deficient human colorectal cancer cells by FEN1 J Clin Pathol 2006;59:1127–32. silencing. Proc Natl Acad Sci U S A 2009;106:3276–81. 29. Ladanyi M, Lui MY, Antonescu CR, Krause-Boehm A, Meindl A, Argani 13. Welch BL. The generalisation of student's problems when several P, et al. The der(17)t(X;17)(p11;q25) of human alveolar soft part sar- different population variances are involved. Biometrika 1947;34: coma fuses the TFE3 transcription factor gene to ASPL, a novel gene at 28–35. 17q25. Oncogene 2001;20:48–57. 14. Satterthwaite FE. An approximate distribution of estimates of variance 30. Hideshima T, Richardson P, Chauhan D, Palombella VJ, Elliott PJ, components. Biometrics 1946;2:110–4. Adams J, et al. The proteasome inhibitor PS-341 inhibits growth, 15. Ostlund G, Schmitt T, Forslund K, Kostler T, Messina DN, Roopra S, induces apoptosis, and overcomes drug resistance in human multiple et al. InParanoid 7: new algorithms and tools for eukaryotic orthology myeloma cells. Cancer Res 2001;61:3071–6. analysis. Nucleic Acids Res 2010;38:D196–203. 31. Nijhawan D, Zack TI, Ren Y, Strickland MR, Lamothe R, Schumacher 16. Baryshnikova A, Costanzo M, Dixon S, Vizeacoumar FJ, Myers CL, SE, et al. Cancer vulnerabilities unveiled by genomic loss. Cell Andrews B, et al. Synthetic genetic array (SGA) analysis in Saccha- 2012;150:842–54.

6136 Cancer Res; 73(20) October 15, 2013 Cancer Research

A Comparative Genomic Approach for Identifying Synthetic Lethal Interactions in Human Cancer

Raamesh Deshpande, Michael K. Asiedu, Mitchell Klebig, et al.

Cancer Res 2013;73:6128-6136. Published OnlineFirst August 26, 2013.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-12-3956

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2013/08/29/0008-5472.CAN-12-3956.DC1

Cited articles This article cites 30 articles, 11 of which you can access for free at: http://cancerres.aacrjournals.org/content/73/20/6128.full#ref-list-1

Citing articles This article has been cited by 6 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/73/20/6128.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://cancerres.aacrjournals.org/content/73/20/6128. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.