Development of Sirna Payloads to Target KRAS -Mutant Cancer

Published OnlineFirst August 6, 2014; DOI: 10.1158/2159-8290.CD-13-0900

RESEARCH ARTICLE

Development of siRNA Payloads to Target KRAS -Mutant Cancer

Tina L. Yuan 1, Christof Fellmann 2 , Chih-Shia Lee 3 , Cayde D. Ritchie 1 , Vishal Thapar 2,4, Liam C. Lee 3 , Dennis J. Hsu 3 , Danielle Grace 2,4 , Joseph O. Carver 3 , Johannes Zuber 2,5 , Ji Luo 3, Frank McCormick 1 , and Scott W. Lowe 2,4,6

ABSTRACT RNAi is a powerful tool for target identifi cation and can lead to novel therapies for pharmacologically intractable targets such as KRAS. RNAi therapy must combine potent siRNA payloads with reliable in vivo delivery for effi cient target inhibition. We used a functional “Sensor” assay to establish a library of potent siRNAs against RAS pathway genes and to show that they effi ciently suppress their targets at low dose. This reduces off-target effects and enables combination gene knockdown. We administered Sensor siRNAs in vitro and in vivo and validated the delivery of KRAS siRNA alone and siRNA targeting the complete RAF effector node (A/B/CRAF) as promising strategies to treat KRAS-mutant colorectal cancer. We further demonstrate that improved therapeutic effi cacy is achieved by formulating siRNA payloads that combine both single-gene siRNA and node-targeted siRNAs (KRAS + PIK3CA/B). The customizable nature of Sensor siRNA payloads offers a universal platform for the combination target identifi cation and development of RNAi therapeutics.

SIGNIFICANCE: To advance RNAi therapy for KRAS -mutant cancer, we developed a validated siRNA library against RAS pathway genes that enables combination gene silencing. Using an in vivo model for real-time siRNA delivery tracking, we show that siRNA-mediated inhibition of KRAS as well as RAF or PI3K combinations can impair KRAS -mutant colorectal cancer in xenograft models. Cancer Discov; 4(10); 1182–97. ©2014 AACR.

INTRODUCTION RNAi provides an alternative therapeutic approach to small- molecule and antibody-based therapeutics for inhibiting gene Mutations in the RAS family of small GTPases, particularly function. RNAi can, in principle, be applied to reversibly silence KRAS, occur in 30% of all human cancers and are often asso- any target gene (reviewed in refs. 8 and 9 ), thereby increasing ciated with resistance to chemotherapy and targeted therapy. the druggable landscape from 10% to virtually 100% of the Somatic mutations lock RAS in the GTP-bound state, leading genome ( 10, 11 ). Because all siRNAs bear structural similarity to constitutive activation of its downstream effector path- and are likely to have comparable pharmacokinetic profi les, ways. Despite intense efforts, pharmacologic inhibition of their use as therapeutics would facilitate drug formulation, KRAS itself ( 1 ) and inhibition of individual effector kinases preclinical testing, and development of combination therapies. downstream of KRAS such as RAF (2, 3) and MEK (4–6 ) have Although delivery is still a major challenge, more reliable lipid so far been unsuccessful in treating KRAS -mutant tumors. and polymeric nanoparticles are in development to deliver Combinations of MEK and PI3K inhibitors are currently siRNA payloads to target tissues (reviewed in ref. 12 ). Regard- being evaluated in clinical trials, although toxicity in normal less, treatment effi cacy would benefi t from the development tissue could limit their therapeutic window (7 ). Thus, effec- of optimized siRNA payloads that potently and specifi cally tive and targeted treatment against KRAS-driven cancer is an silence well-validated target genes at low dose. urgent and unmet clinical need. Beyond its potential as a therapeutic modality, RNAi is a useful tool for identifying and validating new drug targets. This has proved particularly powerful in cancer research, where 1Helen Diller Family Comprehensive Cancer Center, University of Cali- fornia, San Francisco, San Francisco, California. 2 Cold Spring Harbor shRNA or siRNA screens have been used to identify genes that Laboratory, Cold Spring Harbor, New York. 3Laboratory of Cancer Biol- are selectively required for the proliferation and survival of ogy and Genetics, Center for Cancer Research, National Cancer Institute, cancer cells. However, the identifi cation and current in silico 4 Bethesda, Maryland. Memorial Sloan Kettering Cancer Center, New York, prediction of effective shRNAs and siRNAs remains imprecise, New York. 5 Research Institute of Molecular Pathology, Vienna, Austria. 6Howard Hughes Medical Institute, New York, New York. resulting in low success rates. Consequently, screening high- order shRNA/siRNA combinations is not possible without Note: Supplementary data for this article are available at Cancer Discovery fi rst establishing a collection of functionally validated RNAi Online (http://cancerdiscovery.aacrjournals.org/). triggers. In addition, off-target effects, which can be due to T.L. Yuan, C. Fellmann, and C.-S. Lee contributed equally to this article. sequence-dependent and sequence-independent gene deregu- Current address for C. Fellmann: Mirimus, Inc., Cold Spring Harbor, New lation (reviewed in ref. 13 ), must be minimized for meaningful York. interpretation of phenotypic outcomes. Corresponding Authors: Scott W. Lowe, Memorial Sloan Kettering Cancer To overcome these limitations, we used a previously Center, 415 East 68th Street, Z-1114, New York, NY 10065. Phone: 646- described “Sensor” assay (14 ) to generate a functionally 888-3342; Fax: 646-888-3347; E-mail: [email protected] ; Frank McCor- validated library of RNAi sequences against RAS path- mick, [email protected] ; and Ji Luo, [email protected] . way genes. We show that Sensor siRNAs effi ciently ablate doi: 10.1158/2159-8290.CD-13-0900 their gene targets at low nanomolar concentrations in vitro, ©2014 American Association for Cancer Research. which decreased off-target effects and enabled the use of

OCTOBER 2014CANCER DISCOVERY | 1183

RESEARCH ARTICLE Yuan et al.

high-order siRNA combinations to codeplete multiple correlation between the initial algorithmic rank and the rank genes. By applying Sensor siRNAs in vitro and in vivo, we obtained through the biologic Sensor assay (Fig. 1E ), show- identifi ed single-gene and combination-gene payloads that ing that prediction tools alone are still insuffi cient to select inhibit the growth of KRAS -mutant colorectal cancer. Thus, the most potent shRNAs (see also Supplementary Fig. S1C the use of Sensor siRNAs for target discovery and develop- and S1D). Conversely, the Sensor assay identifi ed potent ment of customized siRNA payloads can lead to nanoparticle- shRNAs for nearly every gene, with top-ranked shRNAs show- based treatments for KRAS -mutant cancer and provide a ing similar scores across all genes ( Fig. 1F and Supplementary blueprint for similar strategies to target other key nodes in Table S2). More than 91% of the top fi ve human (344 of cancer maintenance. 375) and >88% of the top fi ve mouse (333 of 375) shRNAs scored >3 (Supplementary Fig. S1C and S1D), a threshold RESULTS score defi ned by positive-control shRNAs. This indicates that the preselection process of 65 shRNAs per gene provided A Functionally Validated RNAi Library Targeting suffi cient coverage to identify several very potent shRNAs KRAS Pathway Genes per gene. To identify potent shRNAs targeting the RAS network, we applied the Sensor assay to evaluate candidate shRNA Sensor siRNAs Are Potent and On-Target sequences targeting 75 human genes and their mouse Synthetic shRNAs and siRNAs enter the endogenous orthologs (Fig. 1A ). These genes encode many classes of microRNA pathway at different stages, but ultimately use proteins, including kinases, GTPases, and transcription fac- the same conserved machinery to downregulate their target tors. The Sensor assay interrogates large numbers of shRNAs genes. We thus hypothesized that potent shRNA sequences under conditions of single genomic integration (“single- could be directly converted into potent siRNA triggers. On copy”) for their ability to repress a cognate target sequence the basis of top-scoring Sensor shRNA sequences, we gener- placed downstream of a fl uorescent reporter expressed in cis ated corresponding 22mer Sensor siRNAs targeting human (14 ). Although we have previously established this assay and KRAS, KSR1, ARAF, BRAF, RAF1/CRAF, MAP2K1, MAP2K2, shown its potential to identify very potent shRNA sequences MAPK1, MAPK3, PIK3CA, PIK3CB, PIK3CD, AKT1, AKT2, and through full gene tiling, here we applied it to evaluate the effi - AKT3. We then measured the knockdown effi ciency of three ciency of preselected candidate sequences targeting a much top Sensor shRNAs and two corresponding siRNAs target- larger set of genes to establish a functionally validated RNAi ing endogenous KRAS , and found that Sensor siRNAs retain library. To assemble the initial candidates, 65 shRNAs per >80% mRNA knockdown when transfected at concentrations gene were selected using a combination of bioinformatics as low as 0.5 nmol/L (Fig. 2A and 2B and Supplementary predictions (15 ) and “Sensor rules” requiring shRNA-specifi c Fig. S2A). We extended this validation to ARAF , BRAF , and features (14 ). This resulted in a total of approximately 10,000 RAF1 siRNAs in a human osteosarcoma cell line, U2OS, shRNAs (split into sets targeting mouse and human genes), which does not have mutations in the RAS–MAPK pathway including 17 control shRNAs added at 1-fold representation and is not sensitive to knockdown of these genes. Transfec- to monitor the functional behavior and stoichiometric fl uc- tion of four of six of these Sensor siRNAs resulted in >70% tuations of the assay. Deep sequencing after two-step cloning mRNA knockdown and >80% protein knockdown at 0.1 of the shRNA–Sensor libraries (14 ) revealed that >99.8% of all nmol/L, and all six siRNAs conferred >80% protein knock- designed human (4,958 of 4,967) and mouse (4,828 of 4,837) down at 0.5 and 2 nmol/L ( Fig. 2C and Supplementary RAS set vectors were successfully constructed. Fig. S2A). Finally, we tested at least two of the top-scoring To evaluate shRNA effi ciency through the enrichment of siRNAs for select genes in U2OS cells. Among 46 Sensor potent shRNAs and depletion of nonfunctional constructs, siRNAs tested, approximately 90% (41) knocked down their fi ve iterative rounds of FACS were run (Sort 1–5), with gates target mRNA at >70% when transfected at 2 nmol/L (Fig. 2D ). set to progressively select for only the most potent, func- The Sensor assay thus serves as a powerful and reliable strat- tional shRNAs (14 ). Compared with our previous libraries egy to identify potent siRNA sequences to generate function- generated by complete tiling of transcripts, the preselected ally validated shRNA/siRNA libraries. RAS pathway libraries showed a less drastic reduction in To biologically validate KRAS Sensor siRNAs for specifi city, shRNA complexity during early selection cycles (Supplemen- we transfected two siRNAs, siKRAS_234 and siKRAS_355 , tary Fig. S1A and S1B), indicating that the bioinformatics into a panel of nine human colorectal cancer cell lines with shRNA preselection process enriched the pools for potent or without mutant KRAS alleles. We observed a strong cor- shRNAs. Importantly, shRNA representation in independent relation between siKRAS-induced cell death and KRAS muta- biologic replicates correlated throughout the sorting pro- tional status and a strong correlation between the behaviors cedure ( Fig. 1B and Supplementary Table S1), whereas the of the two siRNAs ( Fig. 2E ). The antiproliferative effect of the correlation to the initial population was progressively lost, siRNAs in KRAS-mutant cells was attributable to apoptosis showing that the assay can reproducibly identify potent shR- (Supplementary Fig. S2B), although, as previously reported, NAs regardless of their initial representation in the library heterogeneity among cell lines was observed ( 16 ). To verify (Fig. 1C and Supplementary Table S1). that the cell death observed in KRAS -mutant cells was an The recovery of all positive control shRNAs indicates that on-target effect, we generated SW1116 cells (KRAS -mutant the assay robustly enriched for potent single-copy shRNAs, and KRAS-dependent) stably expressing an siRNA-resistant, while depleting weak and intermediate ones (Fig. 1D and tetracycline-inducible KRAS G12V cDNA (Supplementary Table Supplementary Table S1). Strikingly, there was nearly no S2). As expected, both KRAS siRNAs effectively knocked

1184 | CANCER DISCOVERYOCTOBER 2014 www.aacrjournals.org

RNAi Therapy for KRAS-Mutant Cancer RESEARCH ARTICLE

A BC

1,000,000 1,000,000 r = 1.00 r = 0.61 100,000 100,000 Candidate shRNA library 10,000 10,000

1,000 1,000 hRAS V2 hRAS S3 100 100

rtTA + 10 10 reporter cell 1 1 1 10 100 1,000 10,000 100,000 1,000,000 1 10 100 1,000 10,000 100,000 1,000,000 TRE PGK Venus Target hRAS V1 hRAS V

1,000,000 1,000,000 r = 0.48 r = 0.87 Sensor assay 100,000 100,000

10,000 10,000

1,000 1,000 FACS 100 hRAS S5 100 hRAS S3R2

10 10

1 1 1 10 100 1,000 10,000 100,000 1,000,000 1 10 100 1,000 10,000 100,000 1,000,000 hRAS S3R1 hRAS V Potent shRNAs 1,000,000 16 r r = 0.71 14 = 0.04 100,000 12 10,000 10

1,000 8

hRAS S5R2 100

Sensor score 4 10 2

1 0 1 10 100 1,000 10,000 100,000 1,000,000 1 10 100 1,000 10,000 100,000 1,000,000 hRAS S5R1 hRAS V

DEρ 10 70 = 0.11 F 60 Strong control 50 Intermediate control 1 40

Weak control 30 Sensor score

RAS pathway Sensor rank 20 (65 shRNAs/gene) 10 16 0.1 Strong Intermediate Weak 0 14 010203040 50 50 70 12 Algorithm rank 10 8 6 Sensor score 4 2 0 C int JAK1 ICMT TBK1 AKT1 RAF1 JAK3 JAK2 PAK4 KSR1 PAK1 RGL1 AKT3 AKT2 RALB RALA BRAF RAC1 RGL3 ARAF RGL2 EGFR TlAM1 CHUK IGF1R RHOA MTOR IKBKB STAT3 PLCE1 NFKB1 PIK3R1 PSEN2 C weak ERBB2 VEGFA MAPK1 MAPK3 RPTOR PIK3R2 PIK3CA PIK3CB PIK3CD PIK3CG PFKFB3 C strong RALBP1 MAP3K1 MAP2K1 MAPK14 MAP2K2 RALGDS shRNAs (gene)

Figure 1. Functionally validated RAS pathway RNAi libraries for potent and specifi c gene silencing. A, the Sensor assay enables the generation of functionally validated shRNA libraries. The potency of candidate shRNAs was biologically probed in a pooled assay by quantifying knockdown of a Venus reporter cDNA fused to the cognate target site of shRNA. B–F, human RAS set (hRAS) Sensor assay results. Comparable results were obtained for the mouse RAS set (mRas, Supplementary Table S1 and S2). B, correlations in read numbers of technical vector pool replicates (V1 and V2), and biologic duplicates (R1 and R2) at different selection stages of the Sensor assay (S3, Sort 3; S5, Sort 5; r , Pearson correlation coeffi cient). C, correlations in read numbers between the initial vector library (mean of technical duplicates) and the endpoint populations after the indicated sorts (geometric mean of biologic replicates), and the fi nal Sensor scores. D, Sensor scores of control shRNAs of strong (orange), intermediate (blue), or weak (yellow) potency. E, rank correlation between input (Algorithm rank) and output library (Sensor rank), showing that the Sensor rank is not predicted by the informatics tool ( r, Spearman rank correlation coeffi cient). F, Sensor shRNA scores for a selection of direct “RAS effector” genes (for a full list, see Supplementary Table S2). The dotted line indicates a threshold for potent shRNAs, based on known controls. 65 shRNAs/gene; controls are represented multiple times for better visibility.

OCTOBER 2014CANCER DISCOVERY | 1185

RESEARCH ARTICLE Yuan et al.

A BC

ARAF_169 ARAF_381 BRAF_1296 BRAF_2015 RAF1_387 RAF1_2446 siNEG 2 2 0.5 0.1 2 0.5 0.1 2 0.5 0.1 2 0.5 0.1 2 0.5 0.1 2 0.5 0.1 siRNA (nmol/L)

siNEG KRAS_234 KRAS_355 ARAF None Lipid only

sh.Ren.713 sh. KRAS.234 sh. KRAS.355 sh. KRAS.449 2 2 0.5 0.1 2 0.5 0.1 siRNA (nmol/L) G12V HA-KRAS KRAS BRAF Endo KRAS

Actin GAPDH RAF1

GAPDH D 100 >80% KD 80 70%–80% KD <70% KD 60

20 mRNA knockdown (%) mRNA knockdown 0 hAKT3_203 hAKT3_257 hAKT2_267 hAKT1_515 hRAF1_387 hRAF1_562 hRAF1_119 hARAF_169 hBRAF_761 hARAF_381 hKRAS_234 hKRAS_355 hKRAS_449 hAKT2_1082 hAKT2_4452 hAKT2_4044 hAKT1_2122 hRAF1_2446 hKSR1_3427 hKSR1_1680 hKSR1_4447 hKSR1_2225 hBRAF_1961 hBRAF_2392 hBRAF_1296 hBRAF_2015 hMAPK3_633 hMAPK1_991 hPIK3CB_862 hPIK3CB_183 hMAPK3_1698 hMAPK1_1264 hMAP2K1_857 hPIK3CA_2328 hPIK3CA_1746 hPIK3CD_3918 hPIK3CD_3915 hMAP2K2_1279 hMAP2K2_1280 hMAP2K2_1718 hMAP2K1_1918 hMAP2K1_2527 hMAP2K2_1340 hMAP2K1_1916 hMAP2K1_1919 hMAP2K2_1715

EF GKRAS G12V – Dox no KRAS KRAS si si KRAS G12V 150 TFX Lipid siNEG _234 _355 120 + Dox KRAS mutant KRASWT – Dox BRAF –+––– + + + – + doxycycline mutant 100 KRAS WT + Dox KRAS and BRAF WT HA-KRASG12V

LS411N Endo KRAS 80 100 RKO

LS123 SW48 pMEK1/2 60 Caco-2 40 MEK1/2 50 Cell viability (%) LoVo 20 2 R = 0.94 pERK1/2 SW1116 P < 0.0001 SW403 0

si KRAS_355 viability (% siCTRL) SW620 0 ERK1/2 None

050100 150 siNeg

KRAS_234 siDeath si viability (% siCTRL) GAPDH Lipid only si KRAS_234 si KRAS_355

Figure 2. Sensor siRNA potency and KRAS siRNA sensitivity in colorectal cancer lines. A, SW1116 cells expressing an shRNA-resistant HA-KRAS G12V cDNA were infected with one of three KRAS shRNAs or a negative control (sh.Ren.713), and knockdown of endogenous KRAS protein was measured. B and C, U2OS cells were transfected with varying concentrations of Sensor siRNAs against KRAS (B) or RAF kinases (C) for 72 hours and knockdown of endogenous protein was measured (see Supplementary Fig. S2A for quantiﬁ cation). D, top-ranked siRNAs against selected RAS pathway genes were transfected at 2 nmol/L into U2OS cells, and mRNA levels were measured by RT-qPCR 72 hours after transfection. E, colorectal cancer cell lines with the indicated KRAS and BRAF mutational status were transfected with siKRAS_234 or siKRAS_355 at 5 nmol/L, and cell viability was correlated between the two siRNAs. Four of the ﬁ ve KRAS -mutant lines were sensitive to KRAS depletion, whereas KRAS WT lines were resistant. F, SW1116 cells carrying a doxycycline-inducible siRNA-resistant HA-KRAS G12V construct were treated ± doxycycline (100 ng/mL) 60 hours before siRNA transfection. Cell lysates were collected 48 hours after siRNA transfection for Western blot analysis. pERK, phospho-ERK. G, SW1116 cells carrying doxycycline-inducible HA- KRAS G12V or HA- KRAS WT constructs were treated ± doxycycline (100 ng/mL) 60 hours before siRNA transfection. Cell viability (normalized to uninduced cells) was assessed 5 days after siRNA transfection.

down endogenous KRAS, which led to a decrease in phospho- at low dose, and their cytotoxic effect on KRAS -mutant cells MEK and phospho-ERK (pERK; Fig. 2F ) and decreased cell is due to KRAS depletion. viability (Fig. 2G ). Doxycycline-induced expression of siRNA- resistant KRAS G12V , but not siRNA-resistant KRAS WT cDNA, Low-Dose shRNA and siRNA Minimize was sufﬁ cient to restore phospho-MEK and phospho-ERK Off-Target Effects ( Fig. 2F ) and rescue cell viability ( Fig. 2G ). Together, these Off-target effects remain a major concern with RNAi exper- results indicate that siKRAS_234 and siKRAS_355 are potent iments and can occur through several mechanisms, including

1186 | CANCER DISCOVERYOCTOBER 2014 www.aacrjournals.org

RNAi Therapy for KRAS-Mutant Cancer RESEARCH ARTICLE

(i) sequence-dependent cross-hybridization to unintended Sensor siRNAs Can Be Used in High-Order mRNA targets (17, 18); (ii) sequence-independent saturation Combinations to Identify Therapeutic of the RNAi machinery (19 ); and (iii) other as-yet-undefi ned siRNA Cocktails mechanisms. As competition of RNAi triggers for incorpo- ration and retention in the RNA-induced silencing com- We next tested whether the high potency of Sensor siRNAs plex (RISC) is concentration dependent (20 ) and as bulk allows for low-dose transfection of multiple siRNA species transfected siRNAs can outcompete endogenous miRNAs for without affecting the effi ciency of individual siRNAs. We RISC loading ( 19 ), we set out to specifi cally investigate dose- performed high-order combination siRNA delivery into dependent RNAi off-target effects by testing shRNAs and U2OS cells, such that individual siRNAs were transfected at siRNAs in cells in which their cognate target genes are absent. 2 nmol/L each in a pool of four different siRNAs. The major- In such a system, any changes in gene or miRNA expression ity of Sensor siRNAs exhibited no decrease in potency when should result from off-target effects. transfected in combination compared with when transfected We analyzed shRNA-based off-target effects by transduc- alone (Supplementary Fig. S4A). At the protein level, trans- ing Trp53 −/− mouse embryonic fi broblast s (MEF; ref. 21 ) at fection of a complex pool of seven potent Sensor siRNAs single copy and high copy with six well-characterized potent or at 2 nmol/L each—targeting ARAF, BRAF, RAF1/CRAF, weak Trp53 shRNAs ( 14 ). Small RNA sequencing showed that MEK1, MEK2, ERK1, and ERK2—resulted in strikingly effi - high-copy shRNA transduction resulted on average in 20- to cient depletion of all seven proteins simultaneously ( Fig. 4A ). 70-fold more shRNA-derived mature small RNAs than single- Notably, the potency of each siRNA was not compromised as copy transduction ( Fig. 3A ), suggesting that each high-copy siRNA complexity increased. In contrast, for siRNAs that are transduced cell contained at least 20 to 70 viral integrations. less potent, their knockdown effi ciency was preserved only Interestingly, endogenous microRNA levels were unchanged at when pooled with equally weak siRNAs, but further declined both conditions (Supplementary Fig. S3A and S3B), raising the when pooled with potent siRNAs (Supplementary Fig. S4B). possibility that common gene perturbations might be due to Apparently, only siRNAs of equally high potency can be used unintended competing endogenous RNA (ceRNA)–like sponge in high-order combinations to consistently achieve robust effects of vector-based mRNA transcripts ( 22 ) or increased multigene knockdown, further emphasizing the value of the engagement of the transcription and translation machinery. Sensor assay in identifying such equally potent sequences Gene-expression analysis revealed several hundred sequence- across various genes. based off-target effects at high-copy conditions (550 up and Next, we set out to use Sensor siRNAs to identify siRNA 301 down, P value <0.05), whereas only seven altered genes were combinations that could selectively inhibit the proliferation found under single-copy conditions (Supplementary Fig. S3C). of KRAS -mutant cell lines. As shown above ( Fig. 2E ), siRNAs Gene perturbations commonly found across all six shRNAs targeting KRAS itself are one viable option. To identify alter- numbered in the thousands at high-copy transduction (1,821 native targets, we assessed the functional dependency of up and 2,910 down, P value <0.05), but again were absent at different RAF and PI3K isoforms in colorectal cancer cells by single-copy ( Fig. 3B and Supplementary Fig. S3D). Interest- depleting single and multiple isoforms in cell lines carrying ingly, potent and weak shRNAs perturbed gene expression to wild-type KRAS and BRAF alleles (Caco-2 and SW48), mutant a similar degree. We also observed that potent Sensor shRNAs KRAS alleles (SW1116 and SW620), and mutant BRAF alleles lead to preferential retention of the designed guide strands (RKO and LS411N). Cell viability was measured to identify (Supplementary Fig. S3E; ref. 14 ), which might reduce passen- lethal siRNA combinations, and downstream signaling was ger strand–mediated off-target effects, and thus designed all monitored to validate on-target knockdown. Sensor siRNAs as 22mers—rather than conventional 21mers— We fi rst investigated how KRAS mutational status affects to guard the same sequence properties for both strands. sensitivity to knockdown of various RAF isoforms. Caco-2 To test dose-dependent off-target effects of such Sensor- and SW48 cells, two KRAS wild-type cell lines, were relatively derived siRNAs, we transfected siKRAS_234 and siKRAS_355, resistant to individual and combinations of RAF isoform which target both human and mouse KRAS , into Kras−/− MEFs knockdown, as measured by cell viability. As expected, the ( 23 ) at 0.2, 2, 20, and 50 nmol/L and measured global gene- two BRAF -mutant cell lines, RKO and LS411N, were uniquely expression changes. Higher concentrations (20 and 50 nmol/L) sensitive to BRAF depletion ( Fig. 4B ), indicative of their of siRNAs caused signifi cant gene-expression changes (>2-fold addiction to the BRAF oncogene. The KRAS -mutant cell downregulated genes, 12–16 for siKRAS_234 and 63–93 for lines, SW1116 and SW620, showed partial sensitivity to the siKRAS_355), whereas lower concentrations (0.2 and 2 nmol/L) depletion of either BRAF or CRAF ( Fig. 4B ). This fi nding is perturbed only a limited number of genes (>2-fold down- consistent with recent reports that CRAF plays a critical role regulated genes, 0–1 for siKRAS_234 and 3–6 for siKRAS_355; in RAS-driven lung carcinoma and melanoma (24–26 ). Com- Fig. 3C and Supplementary Fig. S3F–S3G). In addition, we bined depletion of BRAF and CRAF in these cells resulted found that at 20 and 50 nmol/L, siKRAS_355 downregulated in additive toxicity, although depletion of all three RAF iso- more genes than siKRAS_234 and that the differentially down- forms led to the strongest reduction in cell viability that was regulated genes were largely nonoverlapping (Supplementary comparable to that of KRAS depletion. Fig. S3H), indicating that these effects are sequence depend- To investigate the mechanism by which codepletion of ent. Together, our data show that although both shRNAs and multiple RAF isoforms in KRAS -mutant cells phenocopies siRNAs can induce aberrant gene perturbations at high concen- KRAS knockdown, we analyzed apoptosis, ERK phosphor- trations, lowering their dosage does not eliminate but critically ylation, and the accumulation of the proapoptotic protein reduces both sequence-specifi c and general off-target effects. BIM in these cells. Progressive depletion of RAF isoforms

OCTOBER 2014CANCER DISCOVERY | 1187

RESEARCH ARTICLE Yuan et al.

A B 7 No Single-copy High-copy shRNA shRNA shRNA

Endo miRNAs (r = 0.998) 3 Trp53.393 Trp53.703 Trp53.814 2 Trp53.1090

Mean high-copy (log reads) Mean high-copy Trp53.1404 Trp53.1525 1 1 10 100 1,000 High/single copy 0 01234567 Mean single-copy (log reads)

C siRNA Concentration (nmol/L) Lipid only 0.0 siKRAS_234 0.2 siKRAS_355 2.0 20.0 50.0 –2 –1 0 12 345678 –4 40 Row Z-score P-value (–In) Gene expression (log2)

Figure 3. Low-dose shRNA and siRNA minimizes off-target effects. A, Trp53 −/− MEFs were infected at single or high copy with one of six Trp53 shRNAs. RNA deep sequencing showed that high-copy shRNA transduction resulted in increased levels of mature Trp53 shRNAs in all cases. Endogenous microRNA expression was unperturbed (see also Supplementary Fig. S3A, S3B, and S3E and Supplementary Table S3 for details; r , Pearson correlation coeffi cient). B, microarray analysis was performed on Trp53 −/− MEFs not expressing any shRNA (WT and empty vector) or transduced at single or high copy with one of 6 Trp53 shRNAs. Shown are the top 1,500 genes that are sequence-independently upregulated and downregulated across all six shRNAs. Signifi cant perturbations were observed only at high-copy transduction (P < 0.05). For more details, see Supplementary Fig. S3C and S3D. C, gene-expression changes were analyzed in Kras −/− MEFs transfected with Sensor siKRAS_234 or siKRAS_355 at various concentrations after 72 hours. Unsupervised clustering of signifi cantly downregulated genes (>2-fold) showed a similar gene-expression pattern between control cells and cells transfected with low concentrations (0.2 and 2 nmol/L) of siRNAs. More profound gene perturbation occurred in cells transfected with high concentrations (20 and 50 nmol/L) of siRNAs (see also Supplementary Fig. S3F–S3H).

in the KRAS -mutant SW620 cells led to progressive reduction tiple RAFs led to a stronger apoptotic response in the KRAS- in pERK levels, such that BRAF/CRAF and ARAF/BRAF/ mutant cells ( Fig. 4C ), indicating that depletion of multiple CRAF codepletion reduced pERK to levels comparable with RAF isoforms is necessary to effectively inhibit ERK activa- those seen upon KRAS knockdown (Supplementary Fig. S4C). tion and induce apoptosis in the KRAS -mutant context. In This correlated with higher accumulation of BIM protein the KRAS wild-type Caco-2 cells, individual and combined (Supplementary Fig. S4C). Furthermore, codepletion of mul- RAF depletion also downregulated pERK, but this did not

1188 | CANCER DISCOVERYOCTOBER 2014 www.aacrjournals.org

RNAi Therapy for KRAS-Mutant Cancer RESEARCH ARTICLE

A siARAF + ++ + + siBRAF + +++ + Figure 4. Sensor siRNAs can be used in high-order siCRAF + +++ + combinations. A, U2OS cells were transfected with siMEK1 + ++ siMEK2 ++ + Sensor siRNAs targeting MAPK pathway genes. Each siERK1 +++ siRNA was transfected at 2 nmol/L ± a nontargeting siERK2 +++ siNEG for a total of 14 nmol/L siRNA. Target protein siNEG ++++++++++++++ knockdown was assessed by Western blot 72 hours ARAF after transfection. B, viability of colorectal cancer cells was measured 5 days after siRNA transfection BRAF or treatment with PLX4032 (1 μmol/L). C, apoptosis of colorectal cancer cells was assessed by measur- CRAF ing caspase-3/7 activity 72 hours after siRNA transfection. Apoptosis was adjusted by viability MEK1 and normalized to a siNEG-transfected control. D, colorectal cancer cells were treated for 72 hours MEK2 with 5 nmol/L of the indicated siRNA(s) ± siNEG for p-ERK1 a total of 30 nmol/L siRNA. Where indicated, cells p-ERK2 were treated with AZD6244 (1 μmol/L), BEZ235 ERK1 (1 μmol/L), or PLX4032 (10 μmol/L for SW48, Caco2, SW620, and SW1116; 1 μmol/L for RKO and GAPDH LS411N) for 24 hours. Lysates were analyzed for pMEK, pERK, pAKT, and pS6 levels with the ViBE B Cell viability bioanalyzer. ND, not determined.

0.0 0.2 0.4 0.6 0.8 1.0 1.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 siNEG siDEATH siKRAS siARAF siBRAF siCRAF siA/BRAF siA/CRAF siB/CRAF siA/B/CRAF PLX4032 Caco-2 SW1116 RKO SW48 SW620 LS411N KRAS and BRAF WT KRAS mutant BRAF mutant C Apoptosis (fold change)

01234560123456 0123456 siNEG siKRAS siARAF siBRAF siCRAF siA/BRAF siA/CRAF siB/CRAF siA/B/CRAF Caco-2 SW1116 RKO SW48 SW620 LS411N D pMEK pERK pAKT pS6 pMEK pERK pAKT pS6 pMEK pERK pAKT pS6 pMEK pERK pAKT pS6 pMEK pERK pAKT pS6 pMEK pERK pAKT pS6 siNEG

siKRAS_234 10 siARAF siBRAF siCRAF siA/BRAF

siA/CRAF change) -fold siB/CRAF 2

A/B/CRAF Phospho-protein si (log –1 siPIK3CA siPIK3CB

siPIK3CD 1 siPIK3CA/B siPIK3CA/D siPIK3CB/D siPIK3CA/B/D siRAFs+PIK3Cs

BEZ235 Cell viability

PLX4032 0 AZD6244 SW48 Caco-2 SW620 SW1116 RKO LS411N KRAS and BRAF WT KRAS mutant BRAF mutant

OCTOBER 2014CANCER DISCOVERY | 1189

RESEARCH ARTICLE Yuan et al.

strongly induce BIM (Supplementary Fig. S4C) or lead to (Supplementary Fig. S5C). To monitor nanoparticle deliv- a signifi cant apoptotic response ( Fig. 4C ). Finally, in the ery, nanoparticles containing siANC_22 and siDsRed, here- BRAF-mutant LS411N cells, pERK was primarily sensitive to after denoted NP(ANC_22+DsRed), were intravenously BRAF depletion (Supplementary Fig. S4C and Fig. 4C ). These administered to mice bearing DsRed;EGFP–positive subcu- results emphasize the importance of inhibiting complete taneous xenografts at varying frequencies. Untreated mice effector nodes, including all related gene isoforms, presum- displayed minor fl uctuations in DsRed fl uorescence, whereas ably due to functional redundancy that can lead to compensa- mice treated two or three times a week displayed two or tion or feedback reactivation of the pathway. In comparison, three distinct reductions in DsRed fl uorescence per week, the small-molecule BRAF inhibitor PLX4720 did not inhibit respectively (Supplementary Fig. S5D). Moreover, untreated the proliferation of KRAS -mutant cells (Fig. 4B ), as previously tumor sections displayed homogeneous DsRed and EGFP reported ( 2, 3 , 26 ). Thus, RAF inhibition by siRNA and small staining, whereas treated tumors displayed heterogeneous molecules may exert distinct biologic effects, probably owing DsRed staining, with particularly low DsRed levels proximal to the unique mechanism of RAF activation (27, 28). to blood vessels (Supplementary Fig. S5E). Together, these We next investigated how depletion of various PI3K iso- results indicate that the nanoparticles were capable of suc- forms in combination with RAF depletion would modulate cessfully delivering an siRNA payload. cell viability. Although the KRAS wild-type Caco-2 and SW48 To test the processing effi ciency of a combination siRNA cells were resistant to individual and combinations of RAF payload targeting two genes, we treated mice with nano- isoform knockdown (Fig. 4B ), the PI3K node, on the other particles packaging siKRAS and siDsRed, NP(KRAS+DsRed), hand, in particular the PIK3CA gene, is critical for their and harvested the tumors 2 days later. Tumors were enzy- viability ( Fig. 4D ). Across all cell lines tested, ablation of all matically digested into single-cell suspensions and FACS three PI3K isoforms decreased cell viability to some extent, sorted for transduced (EGFP high;DsRed low) and untransduced and the combined knockdown of all RAF and PI3K isoforms (EGFPhigh ;DsRed high ) populations (Supplementary Fig. S5F). most severely decreased viability ( Fig. 4D ). However, the fact Western blot analysis showed that both KRAS and DsRed that this siRNA cocktail impaired viability in all cell lines were suppressed in transduced cells compared with untrans- tested indicates a lack of selectivity for KRAS -mutant cells. duced cells, confi rming successful combination siRNA pay- Hence, our experiments support the notion that targeting load delivery. complete effector nodes, particularly the RAF kinases, can inhibit proliferation in KRAS -mutant colorectal cancer cells Nanoparticle-Mediated Delivery of siKRAS to and confer a selective therapeutic benefi t in KRAS-mutant KRAS -Mutant Tumors Impairs Tumor Growth colorectal cancer. To test the therapeutic effi cacy of siKRAS_234 delivery to tumors, we treated mice bearing DsRed;EGFP-positive Sensor siRNAs Can Be Effectively SW620 xenografts (KRAS G12V ) with nanoparticles packaging Delivered In Vivo differing doses of siKRAS_234 (Fig. 5A ). Mice were treated To test the therapeutic effi cacy of Sensor siRNAs in vivo, twice a week beginning when tumors reached 100 mm3 , and we used previously characterized cyclodextrin-polymer nano- tumor growth and nanoparticle uptake were monitored daily. particles (29, 30) to deliver Sensor siRNAs to tumors in mice. Over 3 weeks, siKRAS 4 mg/kg treatment signifi cantly slowed These particles are transferrin-tagged for endocytic uptake in tumor growth compared with control siANC_22 treatment tumor cells and size-selected to extravasate from leaky tumor ( Fig. 5B ). Lower doses of 2 mg/kg and 1 mg/kg siKRAS also vessels. Preclinical studies indicate that they are safe in ani- elicited antitumor effects, although with greater variability mals (31, 32) and, indeed, these particles have been used in (Supplementary Fig. S5G). We suspect that the amount of human clinical trials for the treatment of solid tumors ( 30 ). siRNA released into each cell is limiting, which emphasizes To carefully monitor nanoparticle-payload delivery, we devel- the importance of using very potent siRNAs that achieve tar- oped a fl uorescent reporter system to noninvasively and lon- get knockdown at low concentration. gitudinally monitor nanoparticle transduction using stable In most siKRAS-treated tumors, resected tumor tissue expression of DsRed and enhanced GFP (EGFP) in tumors. (Fig. 5C ) and FACS-isolated transduced tumor cells (Supple- Successful transduction of cells with nanoparticles carrying mentary Fig. S5H) showed decreased levels of KRAS protein. siDsRed resulted in loss of DsRed expression as monitored in Accordingly, concomitant decreases in pERK and pCRAF and real time (Supplementary Fig. S5A). This system thus reports increased levels of p21 were also observed (Fig. 5C ). To moni- on the terminal step of nanoparticle transduction, which is tor the effects of KRAS knockdown during tumorigenesis, target knockdown within tumor cells. EGFP fl uorescence can EGFP fl uorescence was measured daily to track tumor cell be concomitantly tracked to monitor tumor size and treat- viability. siKRAS-treated tumors displayed a dose-dependent ment effi cacy. decrease in EGFP signal during the fi rst week of treatment, These cyclodextrin nanoparticles assemble at a fi xed poly- indicating tumor regression (Fig. 5D , left). Synchronous mer-to-payload ratio of 3:1. Given that high-dose RNAi leads dips in DsRed and EGFP fl uorescence further indicated that to signifi cant changes in gene expression (Fig. 3 and Sup- siRNA delivery and cell death occur concordantly, thus vali- plementary Fig. S3), we designed a nontargeting, non-RISC dating our real-time reporter approach ( Fig. 5D , right). binding “fi ller” siRNA, siANC_22 (Supplementary Fig. S5B), Tumor regression was not maintained after the fi rst that facilitates particle assembly through charge neutraliza- week of treatment. However, the growth-suppressive effects tion, but does not compete with processing of Sensor siRNAs of siKRAS treatment persisted throughout the treatment when transfected at up to 40-fold higher concentrations course. Immunohistochemistry on tumor sections harvested

1190 | CANCER DISCOVERYOCTOBER 2014 www.aacrjournals.org

RNAi Therapy for KRAS-Mutant Cancer RESEARCH ARTICLE

AC Nanoparticle formulations ANC_22 4 mg/kg KRAS 4 mg/kg KRAS 2 mg/kg Empty NP KRAS KRAS ANC_22 4 mg/kg 2 mg/kg 4 mg/kg siDsRed_200 KRAS 8 mg/kg 2 siKRAS_234 capacity 44 4 44 ANC_22 2 si pERK1/2

SW620 Tumor pCRAF Dose 1 Dose 2 Dose 3 Dose 4 Dose 5 Harvest

pS6 Day: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 B p21 KRAS 4 mg/kg 10 GFP KRAS 2 mg/kg ANC_22 GAPDH 8 D

Day13 2 4567 6 1.2 *** 0.8

4 0.4 DsRed

Relative tumor size Relative EGFP 0 4 mg/kg si KRAS 02468 2 2 1.5 ***, P < 0.001 0 1 0510 15 20 0.5 Day 0 2 mg/kg si KRAS 02468 F *** 5 80 *** fluorescence Fold-change 4 60 3 2 40 1 20 KRAS 4 mg/kg Ki67 0

4 mg/kg si ANC_22 02468 Ki67 positivity (%) 0 Day

ANC_22 4 mg/kg KRAS 4 mg/kg ANC_22 KRAS 4 mg/kgKRAS 2 mg/kg E ANC_22 4 mg/kg Ki67 NP treatment

2.0 ****

1.5

1.0 pERK1/2

KRAS 4 mg/kg CC3 0.5

CC3 positivity (%) 0.0 pS6 ANC_22 CC3 KRAS 4 mg/kgKRAS 2 mg/kg ANC_22 4 mg/kg NP treatment

Figure 5. SW620 xenografts treated with varying doses of siKRAS. A, three doses of NP(siKRAS+siDsRed) were prepared by varying the ratios of siKRAS_234 and siANC_22. Nanoparticles were administered via tail-vein injection on days 1, 3, 8, 10, and 15, beginning when tumors reached 100 mm3 . B, NP treatment of siKRAS at 4 mg/kg was able to signifi cantly slow tumor growth compared with control siANC_22–treated tumors. C, at day 19, whole- tumor lysate was analyzed by Western blot for KRAS expression and markers of MAPK signaling and cell-cycle progression. D, tumor volume and relative tumor viability were measured daily with optical imaging of EGFP fl uorescence. Left, raw EGFP images of two representative tumors from each treatment group are shown, where NP siRNAs were injected on days 1 and 3. Right, viability of transduced cells was monitored by concomitant tracking of DsRed and EGFP. Concurrent drops in DsRed and EGFP signals (synchronous arrowheads) illustrate siRNA-induced lethality, whereas DsRed drops that do not induce EGFP drops (opposing arrowheads) indicate no siRNA effect on viability. E, tumors were analyzed by immunohistochemistry for pERK and pS6 levels. F, immunohistochemistry staining and quantifi cation of Ki67- and CC3-positive cells in tumors (***, P < 0.001; ****, P < 0.0001).

OCTOBER 2014CANCER DISCOVERY | 1191

RESEARCH ARTICLE Yuan et al.

at the terminal time point revealed decreased pERK staining a week for 2 weeks at 2 mg/kg per individual siRNA (Fig. 7A ). in 4 mg/kg siKRAS-treated tumors, particularly in regions As observed previously ( Fig. 5B and D ), siKRAS alone induced proximal to blood vessels (Fig. 5E ). Ki67 positivity was also an initial tumor regression that was not sustained over time signifi cantly decreased in 4 mg/kg and 2 mg/kg siKRAS- ( Fig. 7B ). Cotargeting of PIK3C-A/B together with KRAS treated tumors, and cleaved caspase-3 (CC3) positivity was induced a similar level of tumor inhibition during the early increased in 4 mg/kg siKRAS-treated tumors compared with stages of treatment but, importantly, provided signifi cantly siANC-treated tumors ( Fig. 5F ). These data collectively indi- better inhibition of tumor growth during later stages of treat- cate that loss of KRAS protein via siKRAS_234 delivery atten- ment, even surpassing the effi cacy of the high-dose (4 mg/kg) uates, but does not ablate, SW620 tumor growth through siKRAS treatment (Fig. 5B ). inhibition of the MAPK pathway, decreased proliferation, Resected tumor tissue showed at least partial knockdown and increased apoptosis. of all intended target genes (Fig. 7C ), and for all experimental treatment arms, concomitant dips in both the red and green Nanoparticle-Mediated Delivery of Combination channels indicate loss of cell viability concurrent with target Payloads Allows for Multigene Knockdown In Vivo knockdown ( Fig. 7D ). As shown previously, the loss of cell As we were able to concomitantly knock down up to viability in siKRAS-treated tumors resulted from decreased seven genes in vitro using high-order combinations of Sensor proliferation and increased apoptosis (Fig. 7E ), whereas siRNAs, we evaluated the possibility of using Sensor siRNAs siPIK3CA/B–treated tumors primarily underwent increased to concurrently knock down multiple genes in vivo. Our apoptosis. Accordingly, the combination therapy only slightly in vitro data indicated that combined knockdown of the further decreased proliferation and increased apoptosis but RAF effector node (A/B/C RAF) mimics KRAS knockdown not in an additive manner. This indicates that the sustained in SW1116 and SW620 colorectal cancer cell lines (Fig. 4B ). growth inhibition was likely a consequence of perturbing We thus packaged 2 mg/kg of each individual siRNA in vari- other KRAS- and PI3K-regulated cellular functions. Taken ous combinations into nanoparticles and administered these together, these data show that the therapeutic effi cacy of formulations 3 times a week to animals carrying SW1116 anti- KRAS RNAi therapies can be increased through the ( KRASG12A ) xenografts (Fig. 6A ). delivery of combination payloads. In the absence of small Over the 2.5-week treatment course, siKRAS and siA/B/ molecules against most putative cancer targets, the ability CRAF treatment equally inhibited tumor growth (Fig. 6B ), to validate such combination targets in vivo with siRNA also whereas siB/CRAF treatment had no therapeutic effi cacy, has the potential to accelerate both RNAi therapy and small- as predicted by in vitro experiments (Fig. 4B). Resected molecule development. tumor tissue (Fig. 6C ) and FACS-sorted, transduced tumor cells (Supplementary Fig. S5I) showed decreased levels of DISCUSSION all intended target genes, albeit with some variability at this low dose. It is possible that the low vascularity of In this study, we used the Sensor assay (14 ) to build two this xenograft model prevented penetration of the nan- focused, functionally validated RNAi libraries targeting 75 oparticles to the entire tumor. Accordingly, siKRAS and human RAS pathway genes and their mouse orthologs (Sup- si A/B/CRAF treatment only inhibited tumor growth in this plementary Table S2). Inhibiting the RAS signaling pathway xenograft model and did not induce regression, as indicated has been a challenging but important area in translational by relatively stable levels of EGFP in tumor tissue ( Fig. 6D ). research. Our library of KRAS pathway genes was curated Nevertheless, a signifi cant decrease in Ki67 positivity and to contain both direct KRAS effectors (RAF, PI3K, and corresponding increase in CC3 positivity was observed in RAL) and indirect and noncanonical effectors (apoptosis, si KRAS - and siA/B/CRAF –treated tumors compared with metabolism, and stress pathway genes), and we believe that siANC- or si B/CRAF –treated tumors (Fig. 6E ). These data this library will be a useful tool for the broader research demonstrate that four genes can be concurrently knocked community. Sensor siRNAs against RAS pathway genes down in tumor tissue, allowing for inhibition of complete effi ciently depleted their targets at nanomolar concentra- effector nodes and the potential to rationally design com- tions in vitro ( Figs. 2 and 3 ) and accordingly reduced both plex combination therapies. sequence-dependent and sequence-independent off-target effects. Furthermore, low-dose activity facilitates higher- Concomitant Knockdown of KRAS and PI3K order combinatorial gene knockdown by eliminating inter- Improves KRAS -Mutant Tumor Growth Inhibition ference between siRNA species. To improve the therapeutic effi cacy of single-agent siKRAS Using Sensor siRNAs targeting KRAS , we demonstrate treatment in SW620 tumors (Fig. 5 ), we formulated a combi- KRAS addiction in a panel of KRAS -mutant colorectal can- nation payload targeting both KRAS and the PI3K node. In cer cell lines and provide in vivo proof-of-principle evidence the mouse lung, PI3K activation is dependent on oncogenic that direct targeting of KRAS by RNAi is a viable strategy to KRAS and is required for tumor maintenance ( 33, 34 ). How- inhibit KRAS-mutant cancer ( Figs. 2 and 5 ). These results ever, in other tissues, PI3K can be activated through RAS- are corroborated by a parallel study from Xue and col- independent mechanisms, such as direct binding to receptor leagues ( 35 ), in which nanoparticle-mediated delivery of tyrosine kinases. To more thoroughly eliminate PI3K activity, KRAS siRNA and other small RNAs was shown to inhibit we sought to target the PI3K node directly in combination the growth of KRAS -mutant lung cancers. We further show with KRAS. We administered nanoparticles targeting KRAS that targeting the complete RAF effector node by codeplet- alone, PIK3C-A/B, KRAS+PIK3C-A/B, or ANC_22 three times ing all three RAF kinases, but not any single RAF kinase or

1192 | CANCER DISCOVERYOCTOBER 2014 www.aacrjournals.org

RNAi Therapy for KRAS-Mutant Cancer RESEARCH ARTICLE

A Nanoparticle formulations C ANC_22 KRAS A/B/CRAF B/CRAF Empty NP KRAS A/B/CRAF B/CRAF ANC_22 2 mg/kg 2 mg/kg 2 mg/kg 6 mg/kg KRAS siDsRed_200 ARAF siKRAS_234 8 mg/kg 222 22 2 siARAF_381 BRAF_1296 BRAF capacity 4 si 2 22 22 6 siRAF1_2446 siANC_22 CRAF

SW1116 GFP Tumor GAPDH Dose 1 Dose 2 Dose 3 Dose 4 Dose 5 Dose 6 Dose 7 Harvest D Day: 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Day 7 810129111314 DsRed 4 EGFP 3 2 1 B si KRAS 4.5 0 6 8101214 ANC_22 6 4 KRAS 5 A/B/CRAF 4 3 B/CRAF 2 3.5 1

si A/B/CRAF 0 6 8 10 12 14 3 8 6 2.5 4 2 si B/CRAF Fold-change fluorescence Fold-change Relative tumor size Relative 2 0 6 8 10 12 14

10 1.5 8 6 **, P < 0.01 4 1

si ANC_22 2 050 10 15 0 Day 681012 14 Day

E 50 2.5 6 ANC_22 6 KRAS 6 A/B/CRAF 6 B/CRAF 40 2.0 5 5 5 5 4 4 4 4 30 1.5 3 3 3 3 20 1.0 2 2 2 2 10 0.5 CC3 positivity (%) 1 1 1 1 Ki67 positivity (%) 0 0.0

Relative tumor size Relative 0 0 0 0 0 4 8 12 16 0 4 8 12 16 0481216 0481216 KRAS KRAS Day B/CRAF ANC_22 B/CRAF ANC_22 A/B/CRAF A/B/CRAF NP treatment NP treatment

Figure 6. SW1116 xenografts treated with siKRAS or combination siA/B/CRAF. A, four NP treatments delivering 2 mg/kg of each siRNA species were prepared as depicted. Nanoparticles were administered via tail-vein injection on days 1, 3, 5, 8, 10, 12, and 15, beginning when tumors reached 100 mm3 . B, siKRAS and siA/B/CRAF treatment slowed tumor growth to a similar extent, compared with siANC_22 and siB/CRAF–treated tumors. The inset (bottom) shows growth of all individual tumors analyzed. C, at day 17, whole-tumor lysate was analyzed by Western blot for KRAS and RAF isoform expression. D, tumor volume and relative viability were measured daily with optical imaging in the EGFP channel, and DsRed ﬂ uorescence tracked nanoparticle delivery, as described in Fig. 5C . NP siRNAs were injected on days 8, 10, and 12. E, tumors were analyzed by immunohistochemistry for Ki67- and CC3-positive cells (**, P < 0.01; ***, P < 0.001; ****, P < 0.0001).

dual combination, can phenocopy the therapeutic efﬁ cacy of be important to consider entire effector nodes as targets for siKRAS in KRAS -mutant colorectal cancer cells in vitro and drug development. Although such combination targeting in vivo ( Figs. 4 and 6 ). Functional redundancy between RAF can most easily be achieved using RNAi, pan-speciﬁ c small isoforms probably allows for compensatory kinase activ- molecules may achieve similar ends. ity when fewer than all three genes are targeted, thereby The ability to multiplex Sensor siRNAs for combina- incompletely shutting off RAF signaling. As most major tion target knockdown is particularly useful in the context RAS effectors are indeed composed of multi-isoform fami- of KRAS-driven cancers, where numerous effector path- lies (MEK1/2, ERK1/2, PIK3CA/B/D, RALA/B, etc.), it may ways are activated. We show that KRAS knockdown alone

OCTOBER 2014CANCER DISCOVERY | 1193

RESEARCH ARTICLE Yuan et al.

ACNanoparticle formulations KRAS + KRAS ANC_22 KRAS PIK3CA/B PIK3CA/B Empty NP KRAS PIK3CA/B +PIK3CA/B ANC_22 2 mg/kg 2 mg/kg 2 mg/kg 6 mg/kg KRAS siEGFP_413 KRAS_234 α 8 mg/kg 22222 2 si p110 siPIK3CA_1746 capacity 4 2 22 2 62 siPIK3CB_183 siANC_22 p110β

SW620 Actin Tumor D Dose 1 Dose 2 Dose 3 Dose 4 Dose 5 Dose 6 Harvest Day 1 2435 67 Day: 0 1 2 3 4 5 6 7 8 9 101112131415 6 EGFP 5 DsRed B 4 3 2 1 3.5 0

KRAS si KRAS+ PIK3CA/B 02468 PIK3CA/B 6 KRAS+PIK3CA/B 5 3 ANC_22 4 3 2 si KRAS 1 2.5 0 02468

6 5 2 4

Fold-change fluorescence Fold-change 3 2

si PIK3CA/B 1 0

Relative tumor size Relative 1.5 02468

6 5 1 4 3 2

*, P = 0.0164 si ANC_22 1 0 0.5 02468 0 2 4 6 8 10 12 14 Day

Day E n.s.

ANC_22 KRAS PIK3CA/B KRAS+ 40 2.5 55 5 5 PIK3CA/B 2.0 4 4 4 4 30 1.5 3 3 3 3 20 2 2 2 2 1.0 1 1 1 1 10 0.5 CC3 positivity (%) Ki67 positivity (%)

Relative tumor size Relative 0 0 0 0 0 0.0 0 3 6 9 12 15 0 3 6 9 12 15 0 3 6 9 12 15 0 3 6 9 12 15

Day PI3K PI3K KRAS KRAS ANC_22 ANC_22 KRAS+PI3K KRAS+PI3K NP treatment NP treatment

Figure 7. SW620 xenografts treated with combination siKRAS+siPIK3CA/B. A, four NP treatments delivering 2 mg/kg of each siRNA species were prepared as depicted. Nanoparticles were administered via tail-vein injection on days 1, 3, 5, 8, 10, and 12, beginning when tumors reached 70 to 100 mm3 . B, siKRAS and siPIK3CA/B treatment induced mild tumor regression and slowed the rate of tumor growth during the ﬁ rst week of treatment compared with siANC_22–treated tumors. siKRAS+PIK3CA/B treatment potentiated the growth inhibition throughout the course of treatment. The inset (bottom) shows growth of all individual tumors analyzed. C, whole-tumor lysate was collected on day 15 and analyzed by Western blot for KRAS, p110α, and p110β expression. D, tumor volume and relative viability were measured daily with optical imaging in the DsRed channel, and EGFP ﬂ uorescence tracked nanoparticle delivery, as described in Fig. 5C . NP siRNAs were injected on days 1, 3, and 5. E, tumors were analyzed by immunohistochemistry for Ki67- and CC3-positive cells (*, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001).

conferred considerable growth inhibition in vivo. However, (36 ), which can be circumvented through direct targeting of we further show that targeting KRAS in combination with effector nodes in addition to KRAS. Furthermore, concur- the PI3K effector node potentiated this effect ( Fig. 7 ). Thus, rent targeting of effector nodes may also be effective in dis- the delivery of a combination payload targeting both the abling feedback-mediated pathway reactivation or in delaying driver oncogene and an effector node offers the potential acquired resistance. Our system provides a robust and rapid to more thoroughly disable mitogenic and survival signal- means for exploration of such combinatorial space, allowing ing pathways. Many KRAS effectors, including RAF, PI3K, for the rational discovery and validation of optimal target RAL, and RAC, have RAS-independent means of activation combinations for RAS-driven cancers.

1194 | CANCER DISCOVERYOCTOBER 2014 www.aacrjournals.org

RNAi Therapy for KRAS-Mutant Cancer RESEARCH ARTICLE

METHODS Fast Real-Time PCR System platform (Applied Biosystems). GAPDH mRNA was measured as an endogenous control. Each quantitative Sensor Library Design and Construction RT-PCR assay was performed in triplicate. To measure the protein The shRNA–Sensor libraries were constructed as previously level, total cell lysates were subjected to immunoblotting as previ- described (14 ), with a few improvements to enhance pooled cloning ously described (37 ). performance. A recipient vector for the fi rst pooled cloning step, pTNL (TREtight -Neo R-5 ′miR30–BamHI–MfeI–EcoRI), was cloned Cell Viability and Apoptosis Assay by replacing the Xho I/ Cla I fragment in pSENSOR, with a linker Cell viability was measured 5 days after transfection by perform- containing a Bam HI– Mfe I– Eco RI multiple cloning site (CTCGAG ing the CellTiter-Glo Luminescent Cell Viability Assay (Promega), GGATCCCAATTGGAATTCATCGAT). To generate the RAS pathway whereas apoptosis was measured 3 days posttransfection by per- shRNA–Sensor libraries, we designed approximately 5,000 185mer forming the Caspase-Glo 3/7 Assay (Promega) following the man- oligonucleotides per library, each containing a 101mer miR30- ufacturer’s instructions. Cell viability and apoptosis rate were shRNA fragment, an Eco RI/ Bam HI/ Mlu I cloning site, the cognate normalized against AllStars Negative Control siRNA-transfected 50 nt Sensor cassette, and an 18 nt primer-binding site (CCTAG cells. GATCGACGCGGAC). For the 75 human RAS pathway genes and their 75 mouse orthologs (Supplementary Table S2), we generated Off-Target Effects and Gene-Expression Analysis 65 candidate shRNAs each by taking the common transcript of all gene variants and designing the Top300 DSIR predictions (15 ). We Trp53 −/− MEFs were transduced with retroviral LMP vectors ( 38 ) then further selected and ranked these shRNAs using Sensor rules expressing Trp53 shRNAs at single copy (11%–21% infection effi - (14 ) that demand features recurrently found in potent shRNAs and ciency) and high copy (>98% infection effi ciency), selected on puro- excluded sequences containing restriction sites of endonucleases mycin, and grown in the absence of the selection agent before used for cloning. The 65 top-ranked candidates for each gene (Sup- harvest. Uninfected Trp53 −/− MEFs and Trp53 −/− MEFs infected with plementary Table S2) were synthesized on custom 27k oligonucle- an empty vector control served as “no shRNA” reference. Total RNA otide arrays (Agilent Technologies) that also contained 17 control was extracted from at least 5 × 10 6 cells per sample using TRizol shRNAs–Sensor oligonucleotides included at 1-fold representation. (Invitrogen), followed by acidic phenol:chloroform:IAA (125:24:1, The fi nal libraries were constructed according to the previously pH 4.5; Ambion) purifi cation and isopropanol precipitation. All described two-step cloning procedure, with an additional Mfe I diges- samples were DNase I (Roche) treated and again purifi ed by acid tion of the ligation product from cloning step 1 and Bam HI diges- phenol:chloroform:IAA extraction and isopropanol precipitation. tion of the ligation product from cloning step 2. Briefl y, in step Gene-expression analysis was performed on MoGene-1_0-st-v1 1, oligonucleotides were amplifi ed via PCR with the Sens5′Xho microarrays (Affymetrix). To test siRNA off-target effect, RAS-less (TACAATACTCGAGAAGGTATATTGCTGTTGACAGTGAGCG) MEFs stably expressing exogenous NRAS were transfected with and Sens3′Mfe (ATTCATCACAATTGTCCGCGTCGATCCTAGG) siRNA KRAS_234 and KRAS_355 at 0.2, 2, 20, and 50 nmol/L in trip- primers, Xho I/ Mfe I digested, and ligated into Xho I/ Eco RI cut pTNL licate for 72 hours. Total RNA was extracted and subjected to micro- backbone vector. The ligation product was then digested with Mfe I array hybridization on Mouse Gene ST 2.0 microarrays (Affymetrix). to reduce background noise from concatemers. In step 2, the miss- Total RNA samples extracted from cells transfected with lipid only ing 3′miR30-PGK-Venus fragment was cloned into the Eco RI/ Mlu I were used as a gene-expression baseline control. Gene-expression site, separating each shRNA and its cognate target Sensor, followed analysis was performed using Partek Genomics Suite 6.0. The data by Bam HI digestion of the ligation product to further reduce back- discussed in this publication have been deposited in the NCBI’s ground noise. During each cloning step, a representation of at least Gene Expression Omnibus and are accessible through GEO Series 1,000-fold the complexity of the library was maintained. accession numbers GSE59952 and GSE59823.

Cell Lines and Reagents Xenografts and Tumor Imaging The human osteosarcoma cell line U2OS and human colorectal SW620 and SW1116 cells stably expressing DsRed and EGFP were cancer cell lines SW620, SW403, SW1116, LoVo, LS123, Caco-2, made by lentiviral infection. Cells were FACS sorted for double-pos- SW48, RKO, and LS411N were obtained from Dr. Thomas Reid itive cells and pooled populations were propagated for engraftment. (NCI, Bethesda, MD) and grown in McCoy’s 5A media (Lonza) sup- All mouse experiments were done in compliance with UCSF and plemented with 10% FBS. Trp53 −/− MEFs (21 ) were grown in DMEM Institutional Animal Care and Use Committee (IACUC) policies. Two supplemented with 10% FBS. Kras −/− MEFs were obtained from to 3 × 10 6 cells in PBS/Matrigel (1:1) were subcutaneously injected Dr. Mariano Barbacid (Centro Nacional de Investigaciones Oncologi- bilaterally into each rear fl ank of nude mice. Upon engraftment, ° = cas, Spain). All cell lines were cultured at 37 C in a humidifi ed 5% CO2 tumor volume was measured daily with calipers (volume 0.5233 incubator. No cell line authentication was performed by the authors. × W × L × H ). In addition, tumors were optically imaged for DsRed and EGFP fl uorescence daily using a G1 Xenogen IVIS 100 System. Sensor siRNA Transfection and Validation Optical data were analyzed using the Living Image software v. 4.2 (PerkinElmer). Cells were transfected with Sensor siRNAs at the indicated concentrations by using Lipofectamine RNAiMAX (Invitrogen) following Disclosure of Potential Confl icts of Interest the manufacturer’s instructions. AllStars Negative Control siRNA and AllStars Human Cell Death Control siRNA (Qiagen) were trans- C. Fellmann is the CSO of Mirimus, Inc. J. Zuber is a consultant/ fected as controls. To validate the Sensor siRNAs and to test knock- advisory board member for Mirimus, Inc. No potential confl icts of down effi ciency, U2OS cells were transfected with Sensor siRNAs interest were disclosed by the other authors. at the indicated concentrations. Cells were collected 72 hours posttransfection for total cell lysate preparation or total RNA isolation Authors’ Contributions using the RNeasy Mini Kit (Qiagen) followed by cDNA conversion Conception and design: T.L. Yuan, C. Fellmann, C.-S. Lee, J. Zuber, using the High Capacity cDNA Reverse Transcription Kit (Applied J. Luo, F. McCormick Biosystems). To measure the mRNA level, quantitative real-time Development of methodology: T.L. Yuan, C. Fellmann, C.-S. Lee, PCR was performed using the SYBR Green method on a 7900HT C.D. Ritchie, L.C. Lee, J. Zuber, J. Luo

OCTOBER 2014CANCER DISCOVERY | 1195

RESEARCH ARTICLE Yuan et al.

Acquisition of data (provided animals, acquired and managed 12. Zhou J , Shum KT , Burnett JC , Rossi JJ . Nanoparticle-based deliv- patients, provided facilities, etc.): T.L. Yuan, C. Fellmann, C.-S. ery of RNAi therapeutics: progress and challenges. Pharmaceuticals Lee, C.D. Ritchie, D.J. Hsu, D. Grace, J.O. Carver, F. McCormick, 2013 ; 6 : 85 – 107 . S.W. Lowe 13. Fellmann C , Lowe SW . Stable RNA interference rules for silencing . Analysis and interpretation of data (e.g., statistical analysis, biosta- Nat Cell Biol 2014 ; 16 : 10 – 8 . tistics, computational analysis): T.L. Yuan, C. Fellmann, C.-S. Lee, 14. Fellmann C , Zuber J , McJunkin K , Chang K , Malone CD , Dickins R A, C.D. Ritchie, V. Thapar, L.C. Lee, D.J. Hsu, J. Luo, F. McCormick et al. Functional identifi cation of optimized RNAi triggers using a Writing, review, and/or revision of the manuscript: T.L. Yuan, massively parallel sensor assay. Mol Cell 2011 ; 41 : 733 – 46 . 15. Vert JP , Foveau N , Lajaunie C , Vandenbrouck Y . An accurate and C. Fellmann, C.-S. Lee, C.D. Ritchie, V. Thapar, J. Luo, F. McCormick, interpretable model for siRNA effi cacy prediction. BMC Bioinfor- S.W. Lowe matics 2006 ; 7 : 520 . Administrative, technical, or material support (i.e., reporting 16. Singh A , Greninger P , Rhodes D , Koopman L , Violette S , Bardeesy or organizing data, constructing databases): C.-S. Lee, D.J. Hsu, N , et al. A gene expression signature associated with “K-Ras addic- J. Zuber tion” reveals regulators of EMT and tumor cell survival. Cancer Cell Study supervision: T.L. Yuan, J. Luo, F. McCormick, S.W. Lowe 2009 ; 15 : 489 – 500 . 17. Jackson AL , Bartz SR , Schelter J , Kobayashi SV , Burchard J , Mao M , Acknowledgments et al. Expression profi ling reveals off-target gene regulation by RNAi. The authors thank Beicong Ma and Colin Merrifi eld for excel- Nat Biotechnol 2003 ; 21 : 635 – 7 . lent technical assistance, Mark Davis, Jeremy Heidel, and Thomas 18. Birmingham A , Anderson EM , Reynolds A , Ilsley-Tyree D , Leake D , Schluep for providing nanoparticles and helpful discussion, and Fedorov Y , et al. 3 ′ UTR seed matches, but not overall identity, are Mariano Barbacid for cell line reagents. associated with RNAi off-targets. Nat Methods 2006 ; 3 : 199 – 204 . 19. Khan AA , Betel D , Miller ML , Sander C , Leslie CS , Marks DS. Trans- Grant Support fection of small RNAs globally perturbs gene regulation by endogenous microRNAs. Nat Biotechnol 2009 ; 27 : 549 – 55 . This work was supported by funding from the Susan G. Komen 20. Haley B , Zamore PD . Kinetic analysis of the RNAi enzyme complex. Foundation (KG111338 to T.L. Yuan), Daiichi Sankyo (to T.L. Nat Struct Mol Biol 2004 ; 11 : 599 – 606 . Yuan and F. McCormick), the NCI Intramural Research Program 21. Livingstone LR , White A , Sprouse J , Livanos E , Jacks T , Tlsty T D. (1ZIABC011437 to J. Luo), the Lustgarten Foundation, and the Altered cell cycle arrest and gene amplifi cation potential accompany Howard Hughes Medical Institute (to S.W. Lowe). loss of wild-type p53. Cell 1992 ; 70 : 923 – 35 . 22. Tay Y , Kats L , Salmena L , Weiss D , Tan SM , Ala U , et al. Coding- Received November 20, 2013; revised August 1, 2014; accepted independent regulation of the tumor suppressor PTEN by competing August 5, 2014; published OnlineFirst August 6, 2014. endogenous mRNAs. Cell 2011 ; 147 : 344 – 57 . 23. Drosten M , Dhawahir A , Sum EY , Urosevic J , Lechuga CG , Esteban LM , et al. Genetic analysis of Ras signalling pathways in cell proliferation, migration and survival . Embo J 2010 ; 29 : 1091 – 104 . REFERENCES 24. Blasco RB , Francoz S , Santamaria D , Canamero M , Dubus P , Char- 1. Gysin S , Salt M , Young A , McCormick F . Therapeutic strategies for ron J , et al. c-Raf, but not B-Raf, is essential for development of targeting ras proteins. Genes Cancer 2011 ; 2 : 359 – 72 . K-Ras oncogene-driven non-small cell lung carcinoma. Cancer Cell 2. Hatzivassiliou G , Song K , Yen I , Brandhuber BJ , Anderson DJ , A lvar- 2011 ; 19 : 652 – 63 . ado R , et al. RAF inhibitors prime wild-type RAF to activate the 25. Karreth FA , Frese KK , DeNicola GM , Baccarini M , Tuveson DA . C-R af MAPK pathway and enhance growth. Nature 2010 ; 464 : 431 – 5 . is required for the initiation of lung cancer by K-Ras(G12D). Cancer 3. Poulikakos PI , Zhang C , Bollag G , Shokat KM , Rosen N . RAF inhibi- Discov 2011 ; 1 : 128 – 36 . tors transactivate RAF dimers and ERK signalling in cells with wild- 26. Heidorn SJ , Milagre C , Whittaker S , Nourry A , Niculescu-Duvas I , type BRAF. Nature 2010 ; 464 : 427 – 30 . Dhomen N , et al. Kinase-dead BRAF and oncogenic RAS cooperate 4. Rinehart J , Adjei AA , Lorusso PM , Waterhouse D , Hecht JR , Natale to drive tumor progression through CRAF. Cell 2010 ; 140 : 209 – 21 . RB , et al. Multicenter phase II study of the oral MEK inhibitor, 27. Freeman AK , Ritt DA , Morrison DK . Effects of Raf dimerization CI-1040, in patients with advanced non-small-cell lung, breast, colon, and its inhibition on normal and disease-associated Raf signaling . and pancreatic cancer. J Clin Oncol 2004 ; 22 : 4456 – 62 . Molecular Cell 2013 ; 49 : 751 – 8 . 5. Haura EB , Ricart AD , Larson TG , Stella PJ , Bazhenova L , Miller V A, 28. Holderfi eld M , Merritt H , Chan J , Wallroth M , Tandeske L , Zhai H , et al. A phase II study of PD-0325901, an oral MEK inhibitor, in previ- et al. RAF inhibitors activate the MAPK pathway by relieving inhibi- ously treated patients with advanced non-small cell lung cancer. Clin tory autophosphorylation. Cancer Cell 2013 ; 23 : 594 – 602 . Cancer Res 2010 ; 16 : 2450 – 7 . 29. Davis ME . The fi rst targeted delivery of siRNA in humans via a self- 6. Hainsworth JD , Cebotaru CL , Kanarev V , Ciuleanu TE , Damyanov D , assembling, cyclodextrin polymer-based nanoparticle: from concept Stella P , et al. A phase II, open-label, randomized study to assess the to clinic. Mol Pharm 2009 ; 6 : 659 – 68 . effi cacy and safety of AZD6244 (ARRY-142886) versus pemetrexed in 30. Davis ME , Zuckerman JE , Choi CH , Seligson D , Tolcher A , Alabi C A, patients with non-small cell lung cancer who have failed one or two et al. Evidence of RNAi in humans from systemically administered prior chemotherapeutic regimens . J Thorac Oncol 2010 ; 5 : 1630 – 6 . siRNA via targeted nanoparticles. Nature 2010 ; 464 : 1067 – 70 . 7. Shimizu T , Tolcher AW , Papadopoulos KP , Beeram M , Rasco DW , 31. Hu-Lieskovan S , Heidel JD , Bartlett DW , Davis ME , Triche TJ . Smith LS , et al. The clinical effect of the dual-targeting strategy Sequence-specifi c knockdown of EWS-FLI1 by targeted, nonviral involving PI3K/AKT/mTOR and RAS/MEK/ERK pathways in delivery of small interfering RNA inhibits tumor growth in a patients with advanced cancer. Clin Cancer Res 2012 ; 18 : 2316 – 25 . murine model of metastatic Ewing’s sarcoma. Cancer Res 2005 ; 65 : 8. Hannon GJ . RNA interference. Nature 2002 ; 418 : 244 – 51 . 8984 – 92 . 9. Bartel DP . MicroRNAs: genomics, biogenesis, mechanism, and func- 32. Heidel JD , Yu Z , Liu JY , Rele SM , Liang Y , Zeidan RK , et al. Admin- tion. Cell 2004 ; 116 : 281 – 97 . istration in non-human primates of escalating intravenous doses of 10. Hopkins AL , Groom CR . The druggable genome. Nat Rev Drug Dis- targeted nanoparticles containing ribonucleotide reductase subunit cov 2002 ; 1 : 727 – 30 . M2 siRNA. Proc Natl Acad Sci U S A 2007 ; 104 : 5715 – 21 . 11. Patel MN , Halling-Brown MD , Tym JE , Workman P , Al-Lazikani B . 33. Castellano E , Sheridan C , Thin MZ , Nye E , Spencer-Dene B , Die fen- Objective assessment of cancer genes for drug discovery. Nat Rev bacher ME , et al. Requirement for interaction of PI3-kinase p110alpha Drug Discov 2013 ; 12 : 35 – 50 . with RAS in lung tumor maintenance. Cancer Cell 2013 ; 24 : 617 – 30 .

1196 | CANCER DISCOVERYOCTOBER 2014 www.aacrjournals.org

RNAi Therapy for KRAS-Mutant Cancer RESEARCH ARTICLE

34. Gupta S , Ramjaun AR , Haiko P , Wang Y , Warne PH , Nicke B , et al. 37. Weng MT , Lee JH , Wei SC , Li Q , Shahamatdar S , Hsu D , et al. Evo- Binding of ras to phosphoinositide 3-kinase p110alpha is required lutionarily conserved protein ERH controls CENP-E mRNA splicing for ras-driven tumorigenesis in mice. Cell 2007 ; 129 : 957 – 68 . and is required for the survival of KRAS mutant cancer cells . Proc 35. Xue W , Dahlman JE , Tammela T , Khan OF , Sood S , Dave A , et al. Natl Acad Sci U S A 2012 ; 109 : E3659 – 67 . Small RNA combination therapy for lung cancer. Proc Natl Acad Sci 38. Dickins RA , Hemann MT , Zilfou JT , Simpson DR , Ibarra I , Hannon U S A 2014 ;111:E3553–61 . GJ , et al. Probing tumor phenotypes using stable and regulated syn- 36. Karnoub AE , Weinberg RA . Ras oncogenes: split personalities. Nat thetic microRNA precursors. Nat Genet 2005 ; 37 : 1289 – 95 . Rev Mol Cell Biol 2008 ; 9 : 517 – 31 .

OCTOBER 2014CANCER DISCOVERY | 1197

Development of siRNA Payloads to Target KRAS-Mutant Cancer

Tina L. Yuan, Christof Fellmann, Chih-Shia Lee, et al.

Cancer Discovery 2014;4:1182-1197. Published OnlineFirst August 6, 2014.

Updated version Access the most recent version of this article at: doi:10.1158/2159-8290.CD-13-0900

Supplementary Access the most recent supplemental material at: Material http://cancerdiscovery.aacrjournals.org/content/suppl/2014/08/08/2159-8290.CD-13-0900.DC1

Cited articles This article cites 38 articles, 9 of which you can access for free at: http://cancerdiscovery.aacrjournals.org/content/4/10/1182.full#ref-list-1

Citing articles This article has been cited by 20 HighWire-hosted articles. Access the articles at: http://cancerdiscovery.aacrjournals.org/content/4/10/1182.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://cancerdiscovery.aacrjournals.org/content/4/10/1182. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.