EUROPEANJOURNALOFCANCER46 (2010) 2889– 2895

available at www.sciencedirect.com

journal homepage: www.ejconline.com

Review A lethal combination for cancer cells: Synthetic lethality screenings for drug discovery

Elisa Ferrari a, Chiara Lucca a, Marco Foiani a,b,* a Fondazione IFOM (Istituto FIRC di Oncologia Molecolare), IFOM-IEO Campus, via Adamello 16, 20139 Milan, Italy b DSBB-Universita` degli Studi di Milano, Milan, Italy

ARTICLE INFO ABSTRACT

Article history: In recent years, cancer drug discovery has faced the challenging task of integrating the Received 2 July 2010 huge amount of information coming from the genomic studies with the need of developing Accepted 21 July 2010 highly selective target-based strategies within the context of tumour cells that experience Available online 17 August 2010 massive genome instability. The combination between genetic and genomic technologies has been extremely useful Keywords: and has contributed to efficiently transfer certain approaches typical of basic science to Synthetic lethality screens drug discover projects. An example comes from the synthetic lethal approaches, very pow- Cancer erful procedures that employ the rational used by geneticists working on model organisms. Saccharomyces cerevisiae Applying the synthetic lethality (SL) screenings to anticancer therapy allows exploiting the Drug discovery typical features of tumour cells, such as genome instability, without changing them, as opposed to the conventional anticancer strategies that aim at counteracting the oncogenic signalling pathways. Recent and very encouraging clinical studies clearly show that certain promising anti- cancer compounds work through a synthetic lethal mechanism by targeting pathways that are specifically essential for the viability of cancer cells but not of normal cells. Herein we describe the rationale of the synthetic lethality approaches and the potential applications for anticancer therapy. Ó 2010 Elsevier Ltd. All rights reserved.

1. Introduction Genetic instability is a hallmark of tumour cells. Cancer cells genetically differ from normal cells as they have accumu- The majority of chemotherapeutic drugs were identified for lated a large number of mutations including growth-promoting their ability to kill rapidly growing cells. Consequently, most mutations. In fact, the genetic and epigenetic alterations that of these drugs hit not only cancer cells but also normal divid- characterise cancer cells can be instrumental for developing ing cells like bone marrow haematopoietic precursors, stom- more selective pharmacological approaches. As Paul Workman ach, intestine and hair follicle cells.1 This lack of selectivity said, ‘What do cancer cells have that normal cells don’t?...They have for tumour cells is one of the major causes of chemothera- mutations, and you can take advantage of those’.2 Cancer genetic peutic failure in cancer treatment. instability may indeed provide the key to tumour vulnerability.

* Corresponding author at: Fondazione IFOM (Istituto FIRC di Oncologia Molecolare), IFOM-IEO Campus, via Adamello 16, 20139 Milan, Italy. E-mail address: [email protected] (M. Foiani). 0959-8049/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.ejca.2010.07.031 2890 EUROPEANJOURNALOFCANCER46 (2010) 2889– 2895

Recent years have witnessed a change in the drug discov- This effect can be derived from the loss-of-function of two ery strategies in the cancer field. Thanks to the genomic and genes that act in parallel redundant pathways, or belong to post genomic technologies, the integration between basic and the same essential pathway or act in two distant pathways translational research is becoming extremely productive. An that are needed to react to a specific cellular perturbation example comes from the synthetic lethal approach, a very (Fig. 1). promising drug discovery procedure that employs the rational used by geneticists working on model organisms. While the 2. The synthetic lethal approach for cancer conventional anticancer strategies aim at counteracting the therapy oncogenic signalling pathways, the synthetic lethal approach seeks to exploit the typical features of tumour cells without In 1997, Hartwell and colleagues pioneered the idea of apply- changing them. ing genetic approaches to drug discovery through the concept Synthetic lethality (SL) is a genetic phenomenon originally of synthetic lethality.5 They extended the use of the term be- observed in Drosophila melanogaster by Bridges in 1922, while yond classical genetics to all the cases in which the combina- the term was coined by Dobzhansky in 1946, to describe com- tion of a mutation and a drug causes cellular death, whereas plementary lethal systems in wild-type population of Dro- the presence of the mutation alone or the drug alone is viable. sophila pseudoobscura.3,4 It refers to cases in which the The rationale behind this approach is that the effect of a drug combination of two non-lethal mutations yields to lethality; that targets a specific gene product resembles the phenotype the less severe situation, in which the final phenotype corre- caused by a mutation in the gene encoding the same protein. sponds to reduced fitness, is defined as synthetic sickness The authors emphasised two fundamental advantages of (SS). relying on genetic screens for drug discovery: first, a gene The two synthetically lethal mutations have an addictive mutation represents an ideal model for designing a new drug negative impact on a function required for the cell viability. that can mimic the loss-of-function of a specific protein by

Fig. 1 – Different mechanisms leading to synthetic lethality. The possible SL gene pairs are reported on the right side of each panel. Synthetic lethality can arise form the absence of two genes acting in redundant parallel pathways (a1) or distant pathways (a2). Alternatively, it can originate from the lack of two subunits of an essential protein complex (b1) or two proteins of the same essential pathway (b2). EUROPEANJOURNALOFCANCER46 (2010) 2889– 2895 2891

inhibiting or poisoning it; second, genetic screens are unbi- trials studies. Methotrexate, for example, is currently under ased, without precluding any unexpected possibility. phase II evaluation for its efficacy in the treatment of The application of the synthetic lethality rationale offers advanced with defects in the DNA mis- new possibilities for cancer research as shown in Fig. 2. A can- match repair genes.9 Methotrexate is able to induce oxida- cer-related mutated gene can sensitise the tumour cells to a tive DNA lesions and it has been shown to be highly drug that specifically inhibits its synthetic lethal partner. In selective for MSH2-defective cells.10 Differently from MSH2 addition to this, the same drug should not affect normal cells, wild-type cells, in MSH2 cells these DNA lesions thus allowing higher therapeutic selectivity. Moreover, in are not repaired and so the methotrexate treatment should principle, this approach is applicable to any type of cancer be lethal. Therefore, the subtype of colorectal cancer charac- mutation, not only loss-of-function mutations in tumour sup- terised by a deficiency in mismatch repair may represent a pressors but also gain-of-function mutations leading to onco- selective hit of methotrexate. gene expression. These examples clearly pinpoint the potential offered by In retrospect, the mechanism of action of many clinical the synthetic lethal approach when applied to drug discovery. compounds is based on synthetic lethality as documented The synthetic lethality strategy can be relevant in cancer re- by an increasing number of cases in the literature. For exam- search in different ways: ple, the rapamycin derivative CCI-779 exhibits enhanced activity against tumours with mutations in PTEN compared (I) It can contribute to the identification of novel pharma- to tumours with normal PTEN.6 PTEN is a tumour suppressor cological targets. Most of cancer therapies aim at inhib- gene encoding a phosphatase that regulates the PI3-K/AKT iting hyperactive oncoproteins, but often these targets pathway, which plays a central role in growth and anti-apop- are not ‘druggable’. Moreover, only 10% of the most totic mechanisms. The target of CCI-779 is the protein kinase common cancer genes are oncogenes, while the vast mTOR that acts downstream of the PI3-K/AKT pathway, majority is tumour suppressor genes. Synthetic lethal- which is, in turn, up-regulated in PTEN null cancers. These ity interactions may provide a source of cancer-selec- observations explain the higher responsiveness to mTOR tive drug targets: a cancer gene that is frequently inhibition of PTEN-deficient cells compared to PTEN-profi- inactivated in tumours embodies the first SL partner cient cells, even if CCI-779 inhibits the mTOR pathway in both while any other gene that once mutated exhibits SL cells types. interactions with the original cancer gene represents Another example, which is becoming a paradigm for SL the second SL partner and, therefore, a potential drug applications, is the genetic interaction between BRCA1 or target. BRCA2 and PARP1.7,8 (II) Synthetic lethal screens may contribute to the identifi- PARP is involved in the repair of DNA single strand breaks: cation of novel biomarkers by unmasking those genes it binds to the break region and, through autopoly(ADP-rybo- that are absolutely required for cell viability following sil)ation, attracts proteins involved in the repair process. Con- treatment with a specific drug. This approach allows versely, the products of BRCA1 and BRCA2 genes are the identification of not only those genetic profiles that implicated in homology-directed DNA double-strand break sensitise the cells to the drug of interest but also those repair, so both pathways participate in the repair of DNA le- mutations that cause resistance to the pharmacological sions. Therefore, the inhibition of PARP1 in BRCA1-or compound. BRCA2-defective cells results in the accumulation of DNA le- (III) Another application comes from the integration of the sions that cannot be repaired and causes cell lethality. synthetic lethal genetic profiles with the chemical- The validation of synthetic lethal interactions in human genetic profiles.11 A loss-of-function mutation in a gene cells may also provide a mechanistic rationale for clinical that encodes the target of an inhibitory drug mimics

Fig. 2 – Synthetic lethality in chemotherapy. Differently from healthy cells, cancer cells are characterised by mutations; in the figure, yfg2 represents the cancer mutation (YFG: your favourite gene). If YFG1 and YFG2 represent a SL-pair, a drug that inhibits YFG1 can selectively damage the tumour cells, without affecting the normal cells. 2892 EUROPEANJOURNALOFCANCER46 (2010) 2889– 2895

the effect of the compound. Therefore, the comparison by a genetic marker and uniquely tagged with two 20-nucleo- between the chemical-genetic profiles and genetic syn- tide TAGs (molecular barcodes).15 Four different YKOs were thetic lethal interactions panels may help to identify generated: Mata, Mat haploids and homozygous diploids for those pathways that are altered by the drug treatment. non-essential genes, heterozygous diploids for both essential and non-essential genes. The libraries have been widely used In the following paragraph, we will review the available for high-throughput synthetic lethality analysis: in 2001, Tong technologies for large-scale identification of gene–gene and and colleagues set up a systematic approach called synthetic gene–compound SL interactions (GGSL and GCSL screens, genetic array (SGA).16 Using robotic stations, haploid double respectively12; Fig. 3). are generated by mating and meiotic recombination: the query strain (in question) carrying the mutation of inter- 3. High-throughput synthetic lethal screens est is crossed to the array of deletion mutants to produce het- erozygous diploids that can be easily induced to sporulate, Synthetic lethal studies have been carried out in many differ- thus generating the haploid combinations. The final selection ent model organisms including D. melanogaster and Caernor- steps are aimed at identifying those double mutants exhibit- habditis elegans by employing RNA-interference (RNAi) ing synthetic lethal or synthetic sick phenotypes. approaches13, but the concept of synthetic lethality has been In 2003, Ooi and colleagues developed an alternative tech- mainly applied in the budding yeast Saccharomyces cerevisiae. nique termed synthetic lethality analysis by microarray S. cerevisiae represents a powerful tool for studying basic (SLAM) for the investigation of global synthetic lethality inter- 17 cellular functions of eukaryotic cells, including those pro- actions. This method exploits the transformation of the cesses controlling genome integrity. Thanks to its genetic YKO library with the query mutation to create double mu- amenability and versatility, it has received enormous atten- tants. The strains are grown in pools and the genomic DNA tion and is a widely used model organism for studying a vari- isolated from the transformants is amplified by PCR using ety of eukaryotic cellular processes including those that have the molecular barcodes. The TAGs are flanked by universal relevance for human health. The budding yeast represents a priming sites, which allow the amplification of all the strains key experimental option for determining the function of a in the same PCR. By hybridising the amplified molecular bar- conserved gene of interest through the phenotypic analysis codes to DNA microarray and by evaluating the hybridisation of the corresponding mutants. Gene ablation can be easily intensities, it is possible to estimate the growth rate of the dif- performed, in haploid or diploid cells, using a polymerase ferent strains. A third technology called genetic interaction chain reaction (PCR)-based strategy. The Saccharomyces Gen- mapping (GIM) has been described. It is very sensitive as it al- ome Deletion Consortium has created a complete knockout lows the identification of subtle synthetic and epistatic inter- 18 collection of all the annotated yeast genes, the yeast knockout actions. This method combines SGA and SLAM approaches: library (YKO).14 Each open reading frame (ORF) was replaced it relies on mating yeast cells to obtain double mutants, which

Fig. 3 – Gene–gene and gene–compound SL interactions (GGSL and GCSL). The figure represents one of the different technologies available for large-scale analysis of GGSL and GCSL: generation of S. cerevisiae double mutants through synthetic genetic array (SGA) (a), analysis of drug-sensitivity of the Yeast Knock Out library (b). The concept of synthetic lethality is extended to cancer research: a drug can mimic the absence of a protein generating a gene–compound SL interaction. See the text for details. EUROPEANJOURNALOFCANCER46 (2010) 2889– 2895 2893

are grown in pools; the fitness of the query population relative oped living cell microarrays that allow the screening of large to a reference population is assessed by quantifying micro- collections of RNAi and, by reducing the expression of two array hybridisation signals. genes simultaneously, the identification of synthetic lethal A systematic mapping of synthetic lethal interactions has interactions.22 recently been performed in S. cerevisiae by SGA analysis.19 In 2001 it was demonstrated that RNA-interference is also Over 1700 query genes have been screened and a total of feasible in mammalian cells and systematic SL studies be- nearly 5.4 million gene pairs analysed allowing the identifica- came possible.23 Several approaches are now available to gain tion of 170,000 interactions. This represents the first exam- RNA-interference: short duplex RNAs (called small interfer- ple of a global genome scale interaction map: it comprises ence RNA (siRNA)), short hairpin vector encoded RNAs 75% of (all) yeast genes and embodies a huge source of (shRNA) and endoribonuclease prepared-siRNAs (esiRNA).24 information. siRNAs and esiRNAs can be used for high-throughput sin- All those genes, whose mutation or deletion enhances or gle-well assays, in which each well contains a single siRNA re- reduces the activity of a chemical compound, represent the agent. shRNAs can be used in both single-well or polyclonal chemical-genomic profile and have an enormous importance assays, in which a single dish of cells is infected with a pool for drug discovery. The chemical-genetic profile can be assim- of shRNA vectors.25 ilated to a synthetic lethal profile. Numerous studies em- A recent study identified the genetic interactors of the hu- ployed yeast YKO libraries to analyse the differential man oncogene KRAS in cancer cells by targeting 1011 human response of the mutant collections to drug treatments. genes in eight cancer cell lines through shRNA constructs; These studies have relevant implications: (i) the chemical- STK33 turned out to be a component of a signalling pathway genetic profile identifies all those genes whose function is essential in the context of mutant KRAS, thus establishing a crucial for the efficacy of the drug treatment and some of potential drug target in cancers bearing KRAS mutations.26 these genes may represent useful biomarkers. (ii) The chem- High-throughput chemical SL screens in mammalian cells ical-genetic profiles may change according to the concentra- have been performed in various ways over the past decade, tion of the drug, thus providing key information for but the most recent approaches employed RNA-interference. designing the treatment protocol. (iii) Comparing the chemi- The two categories of GCSL screens described above for the cal-genetic profile of a compound of interest with the ones yeast model can be now performed in mammalian cells. of other known drugs may help to unmask potential synergis- In 2008, following the successful identification of synthetic tic interactions between drugs. (iv) Integrating the chemical- lethality between PARP1 and BRCA1 or BRCA2, an RNAi library genetic interaction profiles with the genetic interaction data made up by genes with known roles in DNA repair was tested provided by the SGA studies may help in elucidating the for KU0058948 (PARP inhibitor) sensitivity. This screen identi- mechanism of action of the drug. fied novel determinants of the PARP inhibition response such The GCSL screens also include screens aimed at testing as the transcription coupled DNA repair proteins DDB1 and the effect of a collection of chemicals on a mutant of inter- XAB2.27 These novel SL partners may contribute to the devel- est. This type of GCSL screens represents nowadays a very opment of PARP inhibitors-based pharmacological strategies. powerful drug discovery tool. A large variety of small-mole- High-throughput chemical screens in mammalian cells cule libraries are now available, ranging from collections of have been realised by means of different approaches: one Drug Administration-approved drugs or compounds with method is based on the complementation of the mutated known activities to collections of novel and uncharacterised gene of interest through a low-copy number unstable epi- chemicals. somal plasmid expressing the wild-type copy of the gene. In 2002, Dunstan and colleagues screened a library of more The retention of the plasmid is forced in all the cases in which than 85,000 compounds on yeast strains deficient in DNA there is a synthetic lethal interaction between the gene and double-strand break repair (rad50 and rad52 mutants) and the drug.28,29 were able to identify 126 compounds that showed higher tox- As illustrated for yeast, another approach consists in the icity towards these mutants.20 treatment of the cell line with the genotype of interest with The results obtained from the small-molecule library- compound libraries; cells are grown in multi-well plates and based screens can then be integrated with the genetic interac- tested for viability. Recently, Ji et al. performed a chemical-ge- tion information derived from the SL screens by clustering the netic screen aimed at the identification of drugs able to selec- profiles. tively target pancreatic cancer cells with gain-of-function the Recently, an increasing number of studies are facing the prob- KRAS mutation. KRAS is an oncogene mutated in more than lem of investigating GGSL and GCSL interactions in metazoans. 90% of pancreatic cancers, playing an essential role in the ini- The major setback to these strategies in higher organisms tiation and progression of these tumours. The authors is the mode of delivery of the small interference RNA (siRNA) screened almost 3200 chemical compounds and they found molecule. one KRAS synthetic lethal inhibitor that may be further In C. elegans, RNA-interference (RNAi) can be easily in- characterised.30 duced by feeding, injection, soaking and in vivo-delivery of double-stranded RNA (dsRNA21); RNA-interference can be combined with a query strain characterised by a loss-of-func- 4. Conclusions tion mutation to obtain a panel of genetic interactions. Although the introduction of dsRNA in D. melanogaster is more Cancer is a heterogeneous disease and this diversity results in complicated than in C. elegans, Wheller and colleagues devel- different and often unpredictable responses to the therapies. 2894 EUROPEANJOURNALOFCANCER46 (2010) 2889– 2895

The concept of synthetic lethality applied to drug discovery is Hence, the SL interactions found in yeast can help elucidat- nowadays receiving increasing interest and a growing num- ing the mammalian SL interactions; in 2009, McManus and col- ber of clinically relevant SL interactions are proving its leagues exemplify the validation of a yeast prediction in efficacy. mammalian cells. The 50–30 exonuclease Rad27 was shown to The integration of the data obtained from GGSL and GCSL be SL with mutants, members of screens owns powerful applications. Firstly, knowing the the Rad52 epistasis group.33 Differently from RAD54-proficient mutations that are responsible for particular types of tu- cells, RAD54B-deficient human colorectal cancer cells resulted mours, the concept of SL permits the identification of drugs sensitive to shRNA-mediated silencing of FEN1, confirming the that spare normal tissues while selectively killing cancer cells prediction made in a model organism such as S. cerevisiae. characterised by a specific background. In summary, the concept of synthetic lethality has huge As was pointed out above, gene–gene synthetic lethality potentialities for anticancer drug discovery; its most impor- data represent also a source of potential targets to address tant features consist in the possibility to selectively affect the discovery of novel clinically useful compounds. cancer cells and to hit ‘undruggable targets’. Once verified, the SL interactions may be exploited before planning clinical trials. Actually this information allows the Conflict of interest statement stratification of the patients in subpopulations on the basis of responsiveness prediction: the knowledge of which genetic None declared. alterations lead to drug-sensitivity/resistance may foretell which patients will benefit from the treatment and which will not or even be harmed by it. In the end, this approach may al- Acknowledgements low to save lives, time and money. As shown above, new methods to identify genetic and We wish to thank Linda Cairns and all members of our labo- chemical-genetic SL interactions are emerging in the litera- ratory for helpful comments. Work in M.F. laboratory is sup- ture. RNAi collections are now available, covering the whole ported by grants from Italian Association for Cancer mammalian genome and being compatible with high- Research, from Telethon-Italy, European Community, Regione throughput studies also in mammals. Lombardia and Italian Ministry of Health. Despite the enormous progress made in recent years, obtaining SL data for metazoans is still technically complex and has some limitations as gene inactivation efficacy can REFERENCES vary. However, an increasing number of systematic studies based on RNA-interference are coming out, some with prom- ising results. Therefore, SL studies are possible and are show- 1. Kaelin Jr WG. The concept of synthetic lethality in the context ing nowadays their enormous potentialities also in mammals. of anticancer therapy. Nat Rev Cancer 2005;5(9):689. Moreover, Zender and colleagues recently performed an RNAi 2. Garber K. Synthetic lethality: killing cancer with cancer. J Natl screen for shRNAs that promote tumourigenesis in a mouse Cancer Inst 2002;94(22):1666. 31 cancer model showing the feasibility of in vivo RNAi ; this 3. Bridges CB, Brehme KS. The mutants of Drosophila new technique paves the way to in vivo mammalian SL melanogaster. Carnegie Inst Wash Publ 1944;552:1. screens. 4. Dobzhansky Th. Genetics of natural populations. XIII. Is yeast still meaningful in drug discovery despite the pro- Recombination and variability in populations of Drosophila pseudoobscura. Genetic 1946;31:269. gress that took place in mammalian system? S. cerevisiae 5. Hartwell LH et al. Integrating genetic approaches into the shows a high degree of conservation with mammalian cells: discovery of anticancer drugs. Science 1997;278(5340):1064. 40% of yeast proteins share aminoacid sequence similarity 6. Neshat MS et al. Enhanced sensitivity of PTEN-deficient 32 with a human protein. Thanks to conservation, the data ob- tumors to inhibition of FRAP/mTOR. Proc Natl Acad Sci USA tained in this model system may be used to predict new met- 2001;98(18):10314. azoans genetic interactions. 7. Bryant HE et al. Specific killing of BRCA2-deficient tumours RNAi is a powerful technique but it possesses some draw- with inhibitors of poly(ADP-ribose) polymerase. Nature 2005;434(7035):913. backs: apart from the variation in gene expression abolish- 8. Farmer H et al. Targeting the DNA repair defect in BRCA ment, it is time-consuming and expensive. Moreover, mutant cells as a therapeutic strategy. Nature differently from yeast cells, the high genomic redundancy 2005;434(7035):917. that characterised human cells can impede the discovery of 9. Hewish M et al. Mismatch repair deficient colorectal cancer in novel genetic interactions. the era of personalized treatment. Nat Rev Clin Oncol As described previously, Costanzo et al. revealed the genet- 2010;7(4):197. 10. Martin SA et al. Methotrexate induces oxidative DNA damage ic interaction profile for 75% of yeast genes; the integration of and is selectively lethal to tumour cells with defects in the this huge amount of data with the other SL studies available DNA mismatch repair gene MSH2. EMBO Mol Med 2009;1(6–7): in the literature can open to new research directions, focusing 323. on interesting interactions and validating them in higher 11. Parsons AB et al. Integration of chemical-genetic and genetic eukaryotic systems. This approach allows reducing the num- interaction data links bioactive compounds to cellular target ber of combinations that have to be analysed in mammalian pathways. Nat Biotechnol 2004;22(1):62. systems and, as a consequence, diminishing the costs. 12. Ooi SL et al. Global synthetic-lethality analysis and yeast functional profiling. Trends Genet 2006;22(1):56. EUROPEANJOURNALOFCANCER46 (2010) 2889– 2895 2895

13. Hannon GJ. RNA interference. Nature 2002;418(6894):244. Science 2002;296(5567):550; 14. Winzeler EA et al. Functional characterization of the S. Buchholz F, Kittler R, Slabicki M, Theis M. Enzymatically cerevisiae genome by gene deletion and parallel analysis. prepared RNAi libraries. Nat Methods 2006;3(9):696. Science 1999;285(5429):901. 25. Mullenders J, Bernards R. Loss-of-function genetic screens as 15. Shoemaker DD et al. Quantitative phenotypic analysis of a tool to improve the diagnosis and . yeast deletion mutants using a highly parallel molecular bar- Oncogene 2009;28(50):4409. coding strategy. Nat Genet 1996;14(4):450. 26. Scholl C et al. Synthetic lethal interaction between oncogenic 16. Tong AH et al. Systematic genetic analysis with ordered KRAS dependency and STK33 suppression in human cancer arrays of yeast deletion mutants. Science 2001;294(5550):2364. cells. Cell 2009;137(5):821. 17. Ooi SL, Shoemaker DD, Boeke JD. DNA helicase gene 27. Lord CJ et al. A high-throughput RNA interference screen for interaction network defined using synthetic lethality DNA repair determinants of PARP inhibitor sensitivity. DNA analyzed by microarray. Nat Genet 2003;35(3):277. Repair (Amst) 2008;7(12):2010. 18. Decourty L et al. Linking functionally related genes by 28. Simons A et al. Establishment of a chemical synthetic sensitive and quantitative characterization of genetic lethality screen in cultured human cells. Genome Res interaction profiles. Proc Natl Acad Sci USA 2008;105(15):5821. 2001;11(2):266. 19. Costanzo M et al. The genetic landscape of a cell. Science 29. Canaani D. Methodological approaches in application of 2010;327(5964):425. synthetic lethality screening towards anticancer therapy. Br J 20. Dunstan HM et al. Cell-based assays for identification of Cancer 2009;100(8):1213. novel double-strand break-inducing agents. J Natl Cancer Inst 30. Ji Z et al. Chemical genetic screening of KRAS-based 2002;94(2):88. synthetic lethal inhibitors for pancreatic cancer. Front Biosci 21. Timmons L. Delivery methods for RNA interference in C. 2009;14:2904. elegans. Methods Mol Biol 2006;351:119. 31. Zender L et al. An -based in vivo RNAi screen 22. Wheeler DB et al. RNAi living-cell microarrays for loss-of- identifies tumor suppressors in liver cancer. Cell function screens in Drosophila melanogaster cells. Nat Methods 2008;135(5):852. 2004;1(2):127. 32. Parsons AB, Geyer R, Hughes TR, Boone C. Yeast genomics and 23. Elbashir SM et al. Duplexes of 21-nucleotide RNAs mediate proteomics in drug discovery and target validation. Prog Cell RNA interference in cultured mammalian cells. Nature Cycle Res 2003;5:159. 2001;411(6836):494. 33. Symington LS. Homologous recombination is required for 24. Brummelkamp TR, Bernards R, Agami R. A system for stable the viability of rad27 mutants. Nucleic Acids Res 1998;26(24): expression of short interfering RNAs in mammalian cells. 5589.