ERK Mutations and Amplification Confer Resistance to ERK-Inhibitor Therapy

Published OnlineFirst May 14, 2018; DOI: 10.1158/1078-0432.CCR-17-3674

Cancer Therapy: Preclinical Clinical Cancer Research ERK Mutations and Ampliﬁcation Confer Resistance to ERK-Inhibitor Therapy Bijay S. Jaiswal1, Steffen Durinck1, Eric W. Stawiski1, Jianping Yin2, Weiru Wang2, Eva Lin3, John Moffat4, Scott E. Martin3, Zora Modrusan1, and Somasekar Seshagiri1

Abstract

Purpose: MAPK pathway inhibitors targeting BRAF and MEK types to generate resistant lines. We have used in vitro modeling, have shown clinical efficacy in patients with RAF- and/or structural biology, and genomic analysis to understand the RAS-mutated tumors. However, acquired resistance to these development of resistance to ERK inhibitors and the mechanisms agents has been an impediment to improved long-term survival leading to it. in the clinic. In such cases, targeting ERK downstream of Results: We have identified mutations in ERK1/2, amplifica- BRAF/MEK has been proposed as a potential strategy for tion and overexpression of ERK2, and overexpression of overcoming acquired resistance. Preclinical studies suggest that EGFR/ERBB2 as mechanisms of acquired resistance. Structural ERK inhibitors are effective at inhibiting BRAF/RAS-mutated analysis of ERK showed that specific compounds that induced on- tumor growth and overcome BRAF or/and MEK inhibitor resis- target ERK mutations were impaired in their ability to bind tance. However, as observed with other MAPK pathway inhib- mutant ERK. We show that in addition to MEK inhibitors, ERBB itors, treatment with ERK inhibitors is likely to cause resistance in receptor and PI3K/mTOR pathway inhibitors are effective in the clinic. Here, we aimed to model the mechanism of resistance overcoming ERK-inhibitor resistance. to ERK inhibitors. Conclusions: These findings suggest that combination therapy Experimental Design: We tested five structurally different with MEK or ERBB receptor or PI3K/mTOR and ERK inhibitors ATP-competitive ERK inhibitors representing three different scaf- may be an effective strategy for managing the emergence of folds on BRAF/RAS-mutant cancer cell lines of different tissue resistance in the clinic. Clin Cancer Res; 1–12. 2018 AACR.

Introduction human cancers (7, 8). However, efforts to directly target RAS have not been successful so far (9–11). Several small-molecule inhi- The RAS/RAF/extracellular signal–regulated kinase (ERK) path- bitors that target key effector kinases of MAPK signaling cascade way is extensively studied owing to its involvement in the regu- downstream of RAS have been successfully developed (9, 12). Key lation of cell proliferation, differentiation, and survival (1). The FDA-approved MAPK pathway inhibitors include vemurafenib RAS–MAPK signaling cascade involves an upstream receptor and dabrafenib, which target BRAF, and trametinib, AZD6244 tyrosine kinase (RTK) that upon activation sequentially activates (selumetinib), and GDC-0973 (cobimetinib), which target MEK RAS GTPase, which in turn activates the RAF kinases (MAP3K; (13–16). In the clinic, these inhibitors have led to improved ref. 2). The RAF kinases phosphorylate and activate MEK progression-free survival and overall survival of melanoma and (MAP2K), which then phosphorylates ERK (MAPK) leading to colorectal cancer patients, either as single agents or as combina- its activation (1, 3, 4). Activated ERK then phosphorylates many tion therapy (13–16). However, despite their effectiveness and downstream targets, thereby controlling cellular proliferation, therapeutic successes, a majority of patients relapse within a year differentiation, and survival (1, 5, 6). due to acquired resistance to these agents (17). Analysis of drug- Gain-of-function mutations in RAS and BRAF leading to con- resistant tumors from patients showed reactivation of MEK/ERK stitutive activation of the MAPK pathway occur in about a third of signaling and sustained ERK activation involving multiple mechanisms (18–22). Acquired resistance to BRAF inhibitors has been shown to occur through acquisition of NRAS or KRAS 1Molecular Biology Department, Genentech Inc., South San Francisco, California. mutations (18, 23, 24), ampliﬁcation of BRAF V600E (24), 2Department of Structural Biology, Genentech Inc., South San Francisco, Cali- alternative splicing of BRAF (20), mutations that arise in MEK1 fornia. 3Discovery Oncology Department, Genentech Inc., South San Francisco, 4 or MEK2 (25), and loss of CDKN2A (23). Resistance to MEK California. Department of Biochemical and Cellular Pharmacology, Genentech inhibitors is known to occur due to MEK mutations (26, 27) or Inc., South San Francisco, California. BRAF ampliﬁcation (28). Note: Supplementary data for this article are available at Clinical Cancer Preclinical studies suggest that ERK inhibition may be effec- Research Online (http://clincancerres.aacrjournals.org/). tive in targeting RAS-mutated tumors (29–31). Also, ERK Corresponding Author: Bijay S. Jaiswal, Genentech Inc., South San Francisco, inhibition has been shown to be effective in overcoming CA 94080. Phone: 1-650-4671898; E-mail: [email protected]; and Somasekar acquired resistance to BRAF/MEK inhibitors (29, 30). Several Seshagiri, [email protected] ERK inhibitors including GDC-0994, MK-8353, LTT462, and doi: 10.1158/1078-0432.CCR-17-3674 BVD-523areinvariousstagesofclinicaldevelopment(32–35). 2018 American Association for Cancer Research. ERK inhibitors will expand the choice of targeted therapy for

www.aacrjournals.org OF1

Jaiswal et al.

concentration of indicated inhibitors. Cell growth was assessed Translational Relevance after 4 days using Cell Titer-Glo Luminescent cell viability assay kit Acquired resistance to targeted cancer therapy remains a (Promega). All cell viability data shown were mean SEM of at major challenge in the clinic. ERK inhibitors are under inves- least 3 to 6 replicates of a representative experiment that was tigation for treatment of RAF/RAS-mutated tumors or those repeated at least 2 times with similar results. IC50 values were resistant to BRAF/MEK inhibitors. Understanding the evolu- determined by fitting nonlinear regression curves using GraphPad tion of resistance to current ERK inhibitors will help guide the Prism 5.00 Software (GraphPad). development of better inhibitors and also aid in identifying strategies for combination therapy that can overcome clinical Western blot analysis resistance development. Western blotting was performed as described earlier (41). Briefly, 24 hours after treatment with the indicated drugs, cells were washed with cold PBS and lysed in the RIPA lysis buffer containing protease inhibitor (Roche) and PhosStop phosphatase MAPK pathway–deregulated cancers and also for treating inhibitor (Roche). Lysates were centrifuged at 10,000 g for tumors resistant to BRAF/MEK inhibitors. However, as 20 minutes at 4C. Proteins were resolved by SDS-PAGE and observed with other small-molecule inhibitors, tumors treated transferred to a nitrocellulose membrane using iBlot (Thermo with ERK inhibitors will likely develop resistance. Consistent Fisher Scientific), immunoblotted with indicated antibodies, with this, recent studies using mutagenesis and in vitro experi- HRP-conjugated secondary antibodies (Thermo Fisher Scientific), ments showed development of on-target resistance to ERK and detected with super signal chemiluminescence (Thermo inhibitors (36, 37). Fisher Scientific) as described earlier (41). Using ERK-inhibitor–sensitive cancer cell lines, we have followed the development of resistance upon treatment with multiple Extraction of DNA/RNA ERK inhibitors. In this study, we applied whole-exome sequencing Genomic DNA and total RNA were simultaneously extracted (WES), transcriptome sequencing (RNA-seq), and whole-genome from cell pellets using the All Prep DNA/RNA mini kit (Qiagen). sequencing (WGS) to understand mechanisms of acquired resistance to ERK inhibition. We found on-target and off-target mechan- WES and variant calling isms of resistance and identified strategies for overcoming or We performed WES of parental and resistant cells to identify managing ERK resistance using the resistant cell lines. acquired resistance mutations. Exome capture was performed using the SureSelect Human All Exome kit (50 Mb; Agilent Technologies), and resulting libraries were sequenced on HiSeq Materials and Methods 2500 (Illumina) to generate 2 75-bp-long paired-end data. Cell lines and antibodies A targeted mean coverage of 111 with 80% bases covered at A375, HCT116, MIA PaCa-2, and Panc1 cell lines were pur- 20 was achieved for the exome libraries. Sequencing reads chased from the ATCC. SKMEL30 and IPC298 were obtained from were mapped to the human genome (GRCh38) using BWA German Collection of Microorganisms and Cell Cultures software set to default parameters. Local realignment, duplicate (DSMZ). MelBR1 cell line was generated as described previously marking, and raw variant calling were performed as described (38, 39). Antibodies used in this study are as follows: p-ERK1/2 previously (42). Somatic variants were called by both Strelka (43) (Thr202/Tyr204), pS6-ribosomal protein (Ser235/6), ERK1/2, and MuTect (44), and mutants reported by both programs were RSK, and S6- ribosomal protein (Cell Signaling Technology); included for further evaluation. For on-target mutations, we pRSK (Ser359/363; Abcam); FLAG-M2 and b-ACTIN (Sigma Life included mutation called by either of the two programs. Potential Science); and horseradish peroxidase (HRP)–conjugated second- causal variants in the resistant lines were obtained by filtering out ary antibodies (Thermo Fisher Scientific). the variants observed in the parental lines.

Generation of resistant cell lines RNA-seq and gene expression analysis Parental cells were grown in RPMI-1640 media with 10% FBS RNA-seq data were obtained from total RNA isolated from and were treated continuously for 4 to 6 months with increasing parental and ERKi-R cell lines. RNA-seq libraries were prepared concentrations of inhibitors, starting at 100 nmol/L, until cells using TruSeq RNA sample preparation kit v2 (Illumina). The capable of proliferating efﬁciently in 10 mmol/L drug were libraries were multiplexed and sequenced on HiSeq2500 to derived. obtain on average 50 million single-end (50 bp) reads per sample. RNA-seq reads were aligned to the human genome (GRCh38) Generation of ERK1/2 mutants overexpressing stable cell lines using GSNAP (45). Expression counts per gene were obtained ERK1/2 mutants used in the study were generated by muta- by counting the number of reads aligned uniquely to each gene genesis of wild-type (WT) pCMV6-ERK1/2 (Origene) using Quick locus as deﬁned by NCBI and Ensembl gene annotations and Change Site-Directed Mutagenesis Kit (Stratagene). FLAG-tagged RefSeq mRNA sequences. Differential gene expression analysis (n-terminal) WT and mutant ERK1/2 constructs were cloned into was performed using edgeR (46). pRetro-IRES-GFP retroviral vector, and stable cell lines were generated as described earlier (40). Copy-number analysis Low-pass WGS of parental and resistant cell lines was per- Cell viability assay formed to compute copy number. Alignment of paired-end 75-bp Cells were plated in a 96 well plate (10,000 cells/well in 100 mL reads to GRCh38 using BWA resulted in a median coverage of of media) for 24 hours. They were then treated with increasing 1.8. The genome was then divided in 10 kb bins, and the number

OF2 Clin Cancer Res; 2018 Clinical Cancer Research

Mechanisms of Acquired Resistance to ERK Inhibitors

of reads in each bin provided a count for the genomic bins. This previously (50) by molecular replacement with a known ERK2 count was used to estimate copy-number ratio by computing the structure (PDB code: 1ERK) as the search model using the ratio with the corresponding parental line and adjusting for total program Phaser (51). The structure was further refined with number of reads for each sample. The copy-number ratios were program REFMAC5 (52) and BUSTER (53) using the maximum then segmented using CBS (47), and the resulting segments were likelihood target functions, anisotropic individual B-factor re- used to assign a copy-number value for each gene. finement method, and translation–liberation–screw motions refinement method, to achieve convergence. Protein expression and purification The full-length human ERK2-G169D–mutant construct with Drug sensitivity screen an N-terminal noncleavable His-tag was cloned into a pET52b Drug sensitivity screens were performed on VI-3-R, G994-R and vector. The plasmid was transformed into BL21 (DE3) codon plus MK-ex6-R HCT116 and MIA PaCa2 cell lines as previously Escherichia coli cells (Stratagene), and single colony was inoculated described (54). Briefly, 1K (HCT116) or 2K (MIA PaCa2) cells into 50 mL lysogeny broth (LB) with 50 mg/mL Ampicillin and were dispensed into 384-well microplates in 25 mL of media in the cultured overnight at 37C in a shaking incubator to generate a presence or absence of each 2.5 mmol/L ERK inhibitor tested. Cells seed culture. One liter LB media containing 50 mg/mL Ampicillin were incubated overnight (37 C, 5% CO2) prior to the addition of were inoculated with 15 mL of the seed culture. The cells were compounds in 5 mL of media. All compounds were evaluated in a grown at 37 C in a shaking incubator until OD600 reached 0.4 to 9-point dose–response assay. After 96 hours, cell viability was 0.5. We then shifted the culture to 16 C for 30 minutes and added measured by Cell Titer-Glo assay (Promega). IC50 (concentration 0.5 mmol/L IPTG to induce the ERK protein. The cells were spun at yielding 50% reduction in viability) values were determined by 6,000 rpm for 15 minutes at 12 hours after induction and stored at fitting curves using Genedata Screener software (Genedata). IC50 80C for further processing. values were further used to identify compounds that synergize Cells were lysed in 50 mmol/L Tris, pH 8.0, 500 mmol/L NaCl, with ERK inhibitors in resistant cells. Compounds exhibiting 5 mmol/L BME, 10 mmol/L MgCl2, and 1 mmol/L PMSF using a 4-fold or greater increased sensitivity in at least one ERKi-R cell Microfluidizer. The supernatant was collected after centrifugation line were classified as "hits" and plotted as a heat map. at 10,000 rpm for 30 minutes and then loaded onto 5 mL HisTrap column (GE Healthcare). The column was washed with 50 mL of Statistical analysis 50 mmol/L Tris, pH 8.0, 500 mmol/L NaCl, 5 mmol/L BME, and The Student t test (two-tailed) was used for statistical analyses 10 mmol/L imidazole. The bound proteins were eluted from to compare treatment groups using GraphPad Prism 5.00 column using 50 mmol/L Tris pH 8.0, 5 mmol/L BME, Software (GraphPad). A P value <0.05 was considered statistically 500 mmol/L NaCl, and 10–200 mmol/L imidazole gradient over significant (, P < 0.05). 20 column volumes (protein elution peak fraction was at about 80 mmol/L imidazole). Fractions from HisTrap column were analyzed by SDS-PAGE gel, and peak fractions containing the Results HisERK2-G169D were pooled. The sample was diluted 20-fold Sustained inhibition of ERK in RAS/RAF-mutant cells leads to with Tris-TCEP buffer (25 mmol/L Tris, pH 8.5, and 1 mmol/L resistance TCEP) and loaded onto QHP 5 mL column (GE Health care) In this study, we tested five structurally different ATP-compet- and then washed with Tris-TCEP buffer until OD280 was flat. itive ERK inhibitors (ERKi-s): VTX-11e [V11e] (55), VTX-I-3 [VI-3] HisERK2-G169D protein was eluted with Tris-TCEP buffer con- (29, 55), GDC-0994 [G994] (56), SCH772984 [S984] (30), and taining 0 to 350 mmol/L NaCl gradient. There were two peaks MK example 6 [MK-ex6] (57), a compound structurally similar to resolved by this shallow gradient, one at about 150 mmol/L NaCl SCH772984 and in phase I clinical trials (refs. 33, 37, 57; and the other at 200 mmol/L NaCl, and both peaks were of Supplementary Fig. S1A). These five compounds belong to three HisERK2-G169D. The first one was unphosphorylated ERK, and different scaffold classes with V11e and VI-3 falling into one the second one contained His123-phosphorylated ERK. These class, S984 and MK-ex6 into a second class, and G994 into a two peaks were pooled separately and further purified on a S75 third class of its own. Although V11e, VI-3, and G994 are selective size exclusion column using 25 mmol/L Tris, pH 8.0, 150 mmol/L ATP-competitive ERK inhibitors that bind preferentially to the NaCl, and 1 mmol/L TCEP. Purified proteins were concentrated to active form of ERK (58), S984 and MK-ex6 bind to both active 10 mg/mL and stored at 80C. (phospho) and inactive ERK1/2 with a unique binding mode that prevents phosphorylation of ERK by MEK (30). Protein crystallization and structure determination We tested the five ERK inhibitors for activity in A375, IPC298, The protein was crystallized with hanging-drop vapor-diffusion SKMEL30, HCT116, MIA PaCa2, and Panc1 cells (Supplementary method. Ten mg/mL of protein was mixed with 20% PEG 3350, Fig. S1B). The cell lines used represent different cancer types and 10% isopropanol, and 0.1 mol/L Hepes (pH 7.5). Crystals grew carry BRAF or RAS mutations (Supplementary Table S2). We after 7 days and were cryoprotected in 25% glycerol, 20% PEG confirmed that the cell lines were sensitive to the ERKi-s tested. 3350, 10% isopropanol, and 0.1 mol/L Hepes (pH 7.5). Both peaks The IC50 for each inhibitor ranged from 45 to 1,000 nmol/L from QHP column were crystallized under the same condition. depending on the cell type (Fig. 1A; Supplementary Fig. S1B). The diffraction data for the unphosphorylated form of ERK2- To test if sustained treatment of sensitive cell lines with ERK G169D (QHP peak1) were collected at Stanford Synchrotron inhibitors leads to resistance, we cultured the cells with increasing Radiation Light-source beamline 11-1. The data reduction was concentration of ERKi-s ranging from 0.1 to 10 mmol/L over a done with programs XDS (48) and CCP4 suit (49). Data collec- period of 4 to 6 months (Fig. 1B). This resulted in cells that were tion and structure refinement statistics are summarized in able to proliferate in the presence of high concentrations of ERKi-s Supplementary Table S1. The structure was solved as described (10 mmol/L) compared with the parental lines, and they were

www.aacrjournals.org Clin Cancer Res; 2018 OF3

Jaiswal et al.

A 1,500 V11e C G994 VI-3 MK-ex6 IPC298 SKMEL30 HCT116 MIA PaCa2 Panc1 S984 1,000 P + V11e P + V11e P + V11e P + V11e P + V11e 160 V11e-R + V11e 160 V11e-R + V11e 160 V11e-R + V11e 160 V11e-R + V11e 160 V11e-R + V11e P + G994 P + G994 140 140 P + G994 P + G994 P + G994 G994-R + G994 G994-R + G994 140 140 140 G994-R + G994 G994-R + G994 G994-R + G994 120 120 120 120 120 (nmol/L) 100 100 100 100 100 50 500 80 80 80 80 80 IC 60 60 60 60 60 Cell viability Cell viability Cell viability Cell viability 40 40 40 Cell viability 40 40 20 20 20 20 20 0 0 0 0 0 0 -9 -8 -7 -6 -5 -4 -9 -8 -7 -6 -5 -4 -9 -8 -7 -6 -5 -4 -9 -8 -7 -6 -5 -4 -9 -8 -7 -6 -5 -4 Drugs, log [mol/L] Drugs, log [mol/L] Drugs, log [mol/L] Drugs, log [mol/L] A375 Drugs, log [mol/L] Panc1 IPC298 HCT116 SKMEL30 MIA PaCa2 BRAF-mut NRAS-mut KRAS-mut

10,000 BDNA NA NA A375 0.1 μmol/L 10 μmol/L NA NA IPC298 8,000 NA NA SKMel30 6,000 HCT116 ERKi MIA PaCa2 4,000 Parental (P) Resistant (R) NA NA Panc1 2,000 P R P R P R P R P R V11e VI-3 G994 MK-ex6 S984 IC50

Figure 1.

Sustained ERK-inhibitor treatment leads to resistance. A, IC50 of ERK inhibitors in BRAF/RAS-mutant cell lines. B, Cartoon depicting the generation of ERK-inhibitor–resistant cell lines. C, Dose–response curve of V11e and G994 ERK inhibitors on parental (P) and resistant (R) lines. D, Heatmap showing

IC50 for parental (P) and resistant (R) lines. NA, not available (resistant line generation using the particular drug was not attempted).

termed as V11e-R, VI-3-R, G994-R, MK-ex6-R, and S984-R to acquired ERK mutations were scaffold-specific and arose in denote their ERKi-resistant (ERKi-R) status (Fig. 1C and D). response to treatment with V11e or VI-3. The ERK2 mutations Overall, ERKi-R cells were between 10 and 100 times less sensitive include Y36H in HCT116-V11e-R, C65F in A375-V11e-R, G37A in to ERK inhibitors compared with the parental cell lines (Fig. 1C SKMEL30-VI-3-R, and G37C in HCT116-VI-3-R (Fig. 2B and C). and D; Supplementary Table S3). An A191V mutation in MIA PaCa2-VI-3-R and G186D mutation in HCT116-S984-R were found in ERK1 (Fig. 2B and C). Consis- ERK inhibitor–resistant lines show activated MAPK signaling tent with our findings, using a mutagenesis approach, ERK1 Acquired resistance to BRAF and MEK targeting has been (G186D) and ERK2 (Y36N, G37S, C65Y) mutants have been attributed to sustained MAPK signaling even in the presence of reported to promote in vitro resistance to cell growth inhibition by RAF/MEK inhibitors (19, 28, 38, 59). To understand the mech- V11e in A375 melanoma cells (36). Recently, a G186D ERK1 anism of ERKi resistance, we tested the status of MAPK pathway mutation was reported in S984-resistant HCT116 cells (37). The activation in ERKi-R cells. Treatment of parental KRAS-mutant on-target ERK1/2 mutations observed were inhibitor scaffold- HCT116 colon cancer and MIA PaCa2 pancreatic cancer cell lines specific as we did not find ERK mutations in the G994-R cells, with ERK inhibitors VI-3 or G994 resulted in dose-dependent indicating acquired resistance in these cells evolved through a inhibition of phosphorylation of ERK substrate p90 ribosomal S6 different mechanism. Analysis of the exome data did not identify Kinase (RSK) and S6 ribosomal protein (S6-RP), whereas phos- additional mutations in other MAPK pathway genes such as RAS, phorylation levels of these proteins remained elevated in both RAF, or MEK in the G994-R or other ERKi-R lines (Supplementary VI-3-R and G994-R cells even at 3 times the effective concentration Table S4). of the inhibitors observed in the parental lines (Supplementary Fig. S2). Similarly, sustained pRSK and pS6-RP levels were ERK1/2-mutant ERKi-R cells are sensitive to an alternate ERK observed in SKMEL30-VI-3-R cells compared with parental lines and MEK inhibitor with VI-3 treatment (Fig. 2A). These data show that ERK inhibitors Given that most of the ERK mutations in ERKi-R cells arose in are not effective in inhibiting MAPK signaling in ERKi-R cell lines response to treatment with compounds from a specific scaffold, when compared with the parental lines. we tested whether the acquired resistance can be overcome by treatment with a compound from a different scaffold class. We On-target ERK mutations confer resistance to ERK inhibitors assessed the sensitivity of SKMEL30-VI-3-R, HCT116-V11e-R, and Acquired resistance to BRAF or MEK inhibitors has been shown HCT116-S984-R carrying ERK2-G37A, ERK2-Y36H, and to occur due to BRAF amplification and splice-site alterations in ERK1-G186D, respectively, against indicated ERK inhibitors from BRAF (20, 24, 28, 59) and KRAS G13D, NRAS Q61K/L, MEK1 an alternate scaffold class (Fig. 2D). As expected, we found that P124L, or MEK Q60P mutations (17, 19). To identify acquired SKMEL30-VI-3-R, HCT116-V11e-R, and HCT116-S984-R were resistance mechanisms to ERK inhibition, we performed exome resistant to VI-3, V11e or S984, respectively. However, sequencing of our ERKi-R lines. We found acquired on-target SKMEL30-VI-3-R, HCT116-V11e-R, and HCT116-S984-R were mutations in both ERK1 and ERK2 in some of the resistant lines sensitive to MK-ex6, S984, and V11e (Fig. 2D), respectively, indi- (Fig. 2B and C; Supplementary Table S4). A majority of the cating that the on-target ERKi resistance acquired in response to a

OF4 Clin Cancer Res; 2018 Clinical Cancer Research

Mechanisms of Acquired Resistance to ERK Inhibitors

A D SKMEL30 P + GDC-0973 P 140 P 140 140 VI-3-R + GDC-0973 SKMEL30 VI-3-R VI-3-R P + AZD6244 Parental VI-3-R 120 120 120 VI-3-R + AZD6244 100 100 100

0 80 80 80 0 10 0.1 1.0 0.1

VI-3 (μmol/L) 10 1.0 60 60 60 pRSK) (S359/63) Cell viability Cell viability Cell viability 40 40 40

RSK 20 20 20

pERK (T202/Y204) 0 0 0 -9 -8 -7 -6 -5 -4 -9 -8 -7 -6 -5 -4 -9 -8 -7 -6 -5 -4 ERK VI-3, log [mol/L] MK-ex6, log [mol/L] MEKi, log [mol/L] pS6-RP (S235/6) HCT116 S6-RP P + GDC-0973 140 P 140 P 140 V11e-R + GDC-0973 V11e-R V11e-R P + AZD6244 β-ACTIN 120 120 120 V11e-R + AZD6244

100 100 100

80 80 80 B 60 60 60 Cell viability Cell viability 40 Cell viability 40 40

20 20 20

0 0 0 -9 -8 -7 -6 -5 -4 -9 -8 -7 -6 -5 -4 -9 -8 -7 -6 -5 -4 V11e, log [mol/L] S984, log [mol/L] MEKi, log [mol/L] HCT116

P + GDC-0973 P 140 140 P 140 S984-R + GDC-0973 S984-R S984-R P + AZD6244 120 120 120 S984-R + AZD6244

100 100 100

80 80 80 C 60 60 60 Cell viability Cell viability 40 Cell viability 40 40 Genes mutated 20 20 20 0 0 0 Resistant cell lines ERK1 ERK2 -9 -8 -7 -6 -5 -4 -9 -8 -7 -6 -5 -4 -9 -8 -7 -6 -5 -4 HCT116-V11e-R WT Y36H S984, log [mol/L] V11e, log [mol/L] MEKi, log [mol/L] HCT116-VI-3-R WT G37C E SKMEL30 F SKMEL30 SKMEL30-VI-3-R WT G37A Parental VI-3-R Parental VI-3-R 0 A375-V11e-R WT C65F 0

μ 0.1 1.0 10 0 0.1 1.0 10 μ MK-ex6 ( mol/L) GDC-0973 ( mol/L) 0.1 1.0 10 0 0.1 1.0 10 HCT116-S984-R G186D WT pRSK (S359/63) pRSK (S359/63) MIA PaCa2-VI-3-R A191V WT RSK RSK pERK (T202/Y204) pERK (T202/Y204) ERK ERK pS6-RP (S235/6) pS6-RP (S235/6) S6-RP S6-RP β-ACTIN β-ACTIN

Figure 2. On-target ERK mutation leads to resistance. A, Western blot assessment of the activation status of ERK and downstream signaling molecules. B, Quilt plot showing ERK mutations in ERKi-R cells. C, ERK1 and ERK2 mutations identified in ERKi-R cells. D, Dose–response curve of resistant ERK-mutant lines against cross-class ERK inhibitors and indicated MEK inhibitors. Parental lines (P). E and F, Western blot analysis showing effect of ERK inhibitor MK-ex6 (E) and MEK inhibitor GDC-0973 (F) on MAPK signaling of SKMEL30 cells. compound of a particular class can be overcome by another ERK in a dose-dependent manner in SKMEL30-VI-3-R cells (Fig. 2F). inhibitor belonging to a different scaffold. Consistent with this, Together, these results indicate that the mutant ERK in the sustained MAPK and PI3K/mTOR signaling of SKMEL30-VI-3-R resistant lines is not constitutively active and is still dependent cells was inhibited by alternate ERK inhibitor MK-ex6, as indi- on MEK for it activation. Thus, these results indicate that MEK cated by a decrease in pRSK and pS6-RP levels (Fig. 2A and E). inhibitor treatment is a viable strategy for overcoming ERKi Interestingly, though both MK-ex6 and VI-3 inhibit ERK and resistance. block downstream signaling, they lead to sustained phosphory- To confirm that the on-target ERK1/2 mutations directly con- lation of ERK (Fig 2A and E) perhaps by uncoupling feedback tributed to the resistance observed, we generated stable cell lines inhibition (60). expressing the ERK mutants and tested them for sensitivity to ERK Analysis of the site of on-target ERKi-R mutations within ERK inhibitors (Fig. 3A–E). As expected, expression of ERK2 Y36H or indicated that these mutations likely do not lead to constitutive C65F in HCT116 or SKMEL30 cells promoted resistance to ERKi ERK activation. This along with the observation that inhibitors V11e, when compared with paternal cells or WT ERK2-expressing from alternate scaffold classes were able to block ERKi-R lines cells (Fig. 3A and D; Supplementary Fig. S3). Similarly, expression with ERK mutations led us to predict that MEK inhibitors of ERK2 G37C or G37A in HCT116 or SKMEL30 cells and ERK1 would be effective in overcoming on-target ERK mutation– A191V in HCT116 conferred resistance to VI-3 inhibitor (Fig. 3B mediated resistance. Consistent with this, we found MEK inhi- and E; Supplementary Fig. S3). Further expression of ERK1 bitors GDC-0973 and AZD6244 to be effective in blocking the G186D in HCT116 cells led to resistance to S984 (Fig. 3C; growth of ERKi-resistant SKMEL30-VI-3-R, HCT116-V11e-R, Supplementary Fig. S3A). Consistent with our findings, expres- and HCT116-S984-R cells (Fig. 2D). Further, treatment with sions of ERK1 and ERK2 mutants were previously reported to MEK inhibitor GDC-0973 reduced the pRSK and pS6-RP level confer ERKi resistance in A375 and HCT116 cells (36, 37, 61).

www.aacrjournals.org Clin Cancer Res; 2018 OF5

Jaiswal et al.

AB C F BRAF-V600E-GEMM Parental ERK2-WT 160 Parental 160 160 Parental EV EV ERK2-G37A ERK1-WT EV 140 ERK2-WT 140 ERK2-G37C 140 ERK1-WT ERK2-Y36H ERK1-A191V ERK1-G186D 120 ERK2-C65F 120 120 Tumor Vemurafinib Tumor Vemurafinib Tumor 100 100 100 formation regressed 80 80 80 relapse 60 60 60 Cell viability Cell viability Cell viability 40 40 40 ERKi 20 20 20 0 0 0 ERKi-Resistant Generated Vemurafinib -10 -4-5-6-7-8-9 -10 -4-5-6-7-8-9 -10 -4-5-6-7-8-9 MelBR1 cell lines MelBR1 cell lines resistant tumors V11e, log [mol/L] VI-3, log [mol/L] S984, log [mol/L] MelBR1 G 140 P 140 P V11e-R G994-R 120 120

100 100 D 160 Parental ERK2-Y36H E 160 Parental ERK2-G37A 80 EV ERK2-C65F EV ERK2-G37C 80 140 ERK2-WT 140 ERK2-WT 60 60 120 120 Cell viability 40 viability Cell 40 100 100 20 80 80 20 60 60 0 0 -4-5-6-7-8-9 -4-5-6-7-8-9 Cell viability Cell viability Cell 40 40 V11e, log [mol/L] G994, log [mol/L] 20 20 0 0 G55A (Mouse)/G37A (Human) -4-5-6-7-8-9-10 -4-5-6-7-8-9-10 H V11e, log [mol/L] VI-3, log [mol/L]

Mouse_ERK1 41 PRYTQLQYIGEGAYGMVSSAYD 62 Human_ERK2 23 PRYTNLSYIGEGAYGMVCSAYD 44

Figure 3. Ectopic expression of ERK mutants confers resistance to ERK inhibitors. A–E, Effect of V11e (A and D), VI-3 (B and E), or S984 (C) on proliferation of HCT116 (A–C) and SKMEL30 (D and E) cells stably expressing ERK1/2 WT, empty vector (EV), or the indicated mutants. F, Schematics of generation of ERK-inhibitor–resistant cells using vemurafenib-resistant BRAF V600E–mutant MelBR1 mouse tumor cell line. G, Dose–response curve of ERK inhibitors for parental and ERKi-R MelBR1 cells. Parental line (P). H, Sequence alignment of mouse ERK1 and human ERK2 proteins. Amino acid coordinates are shown on either side of the sequence.

These findings confirm that the on-target mutations are sufficient with ERK2 (PDB: 4FV6). Mapping of the resistance mutations on to confer resistance to ERK inhibitors against which they arose. to the ERK2 structure (Fig. 4A and B) revealed that the ERK2 It has been shown that, in a genetically engineered mouse mutations shown in Fig 2C are located in the vicinity of residue model, skin-specific expression of BRAF V600E leads to the Tyr36, suggesting protein–ligand interactions in this region are development of melanoma (62). The BRAF V600E–mutant mel- important for the inhibitory function of VI-3. We noted that the anoma is sensitive to BRAF inhibitor vemurafenib (39). However, chlorine atom within VI-3 makes a "face-on" Cl–p interaction sustained treatment with vemurafenib leads to resistance (38, 39). with Tyr36 at a distance of 3.5Å (Fig. 4B). Tyr36, in turn, engages A melanoma cell line, MelBR1, has been established from vemur- p–p stacking with Tyr64 in aC-helix. The sandwich-like structure afenib-resistant tumors (Fig. 3F; ref. 39). We exposed the MelBR1 of VI-3–Tyr36–Tyr64 stabilizes the complex. Y36H mutation cell line to increasing concentrations of V11e and G994 over the modified the centerpiece of the sandwich. As Imai and colleagues course of 3 to 5 months (Fig. 3F) and generated a MelBR1-ERKi– reported (63), Cl–p type interaction with histidine favors "edge- resistant cell lines (Fig. 3F and G). As expected, MelBR1-V11e-R on" over "face-on" conformation and prefers a longer interaction and G994-R cells were less sensitive to ERKi V11e and G994, distance of 4.0Å. We reason that when VI-3 binds, this part of respectively, when compared with parental MelBR1 cell lines ERK2 structure becomes too crowded that prevents the optimal (Fig. 3G). Exome sequencing of MelBR1-resistant lines identified conformation for a histidine residue. V11e shares identical chem- a G55A ERK1 mutation in V11e-R cells (Fig. 3H; Supplementary ical structure with VI-3 in this region and is expected to bind ERK2 Table S4). However, we did not observe on-target mutations in in the same manner. Cys65 is a buried residue adjacent to Tyr64 ERK1 or ERK2 in MelBR1-G994-R cells. Alignment of human and (Fig. 4B). Switching to a bulky phenylalanine C65F is likely to mouse ERK protein sequences showed that the mouse ERK1 perturb Tyr64 orientation and consequently weaken the inhibitor Gly55 is equivalent to Gly37 of human ERK2 (Fig. 3H). This is binding. Two other resistant mutations G37A and G37C appear to consistent with the Gly37 to Ala resistance mutation observed in block VI-3 and V11e binding in the pocket under the glycine-rich the ERKi-R human cell lines treated with VI-3, a compound in the loop (G-loop). Interestingly, S984 was sensitive to above ERK2 same class as V11e (Fig. 2C). These observations collectively mutations (Fig 2D, center plot). This phenomenon could be indicate that on-target ERK resistance will likely occur in BRAF- explained by the crystal structure of S984/ERK2 complex (PDB: mutant patients resistant to BRAF inhibitors when treated with 4QTB). Unlike VI-3, which binds underneath the G-loop, the long ERK inhibitors from the VI-3/V11e scaffold class. piperazine–phenyl–pyrimidine moiety of S984 wraps around the outside of G-loop (Fig. 4C and D). Residues Tyr36 and Gly37 no Structural analysis of ERKi-resistant mutants longer make specific interactions with the inhibitor; therefore, To investigate the structural basis of resistant mutations found mutation in these positions does not affect the inhibitory ability in ERK1/2, we analyzed the crystal structure of VI-3 in complex of S984.

OF6 Clin Cancer Res; 2018 Clinical Cancer Research

Mechanisms of Acquired Resistance to ERK Inhibitors

A C P-loop

P-loop S984

VI-3 A-loop A-loop

B D

G37 Y36 C65 S984 Y64

D167ERK2 ERK2 L170ERK2 D184ERK1 D169 VI-3 D186ERK1 L187ERK1

Figure 4. Crystal structure of ERK2 in complex with ERK inhibitor. A, ERK2 crystal structure in complex with VI-3 (PDB: 4FV6, magenta). Resistant mutations found in ERK2 (colored in orange) are located in G-loop and C-helix. B, A close-up view of VI-3 interaction with ERK2 P-loop and C-helix. The color scheme is the same as in A. Dotted lines indicate the stacking interactions between VI-3 chloro-group Y36 and Y64. C, Crystal structure of ERK2 in complex with S984 (PDB: 4QTB, yellow) superimposed onto ERK2-G169D structure (blue). D, A close-up view at residue Gly169 of ERK2, which is equivalent to the resistant ERK1-G186D mutation. Addition of Asp side chain at D169ERK2 forced D167ERK2 to adopt a rotamer conformation that blocks inhibitor binding. Also note that Y34 is tucked under the P-loop upon S984 binding.

ERK1 and ERK2 are highly homologous (84% identical), and ERK2 amplification confers resistance to ERK inhibitors most of the ERK inhibitors including G994 and S984 displayed Amplification of KRAS has been implicated as a mechanism similar inhibitory potency toward ERK1 and ERK2 (30, 56). of resistance to anti-EGFR antibody treatment as well as BRAF- Therefore, it is expected that these inhibitors bind to ERK1 and and MEK-inhibitor therapies (29, 59, 64). Similarly, HGF/MET ERK2 in a similar manner. Not surprisingly, some of the resistant amplification has been implicated in the resistance to EGFR mutations arose in ERK1, while others in ERK2, confirming the therapy (65, 66). Amplification of BRAF V600E has been functional and structural redundancy. The two acquired ERK1 reported as a cause of resistance to BRAF inhibitor or BRAF resistance mutations we identified, A191V and G186D, reside in inhibitor/anti-EGFR therapy (24, 59). To understand the the activation-loop (A-loop). We investigated the S984-resistant mechanism of ERKi resistance in G994-R, MK-ex6-R, and other mutation G186DERK1 in the context of S984 and ERK2 complex ERKi-R lines, we assessed copy-number alterations using crystal structure (PDB: 4QTB). As shown in Fig 4D, Asp167 side WGS data. Our analysis identified ERK2 focal copy gains on chain needs to flip out in order to allow S984 to fit into the WT chromosome 22 in IPC298-G994-R and IPC298-V11e-R ERK2 pocket. To understand the impact of G186DERK1 mutation, (Fig. 5A; Supplementary Table S5) and to a modest level in we determined the crystal structure of ERK2 with the equivalent MIA PaCa2-S984-R cells (Supplementary Table S5). Consistent residue Gly169 mutated to aspartate (G169DERK2). Compared with the amplification, expression of ERK2 was elevated in with Gly169, Asp169 occupied additional space and pushed these cells as assessed by RNA-seq (Fig. 5B). In addition to Asp167 toward the ligand-binding pocket (Fig. 4D). This struc- ERK2 amplification, IPC298-GDC-0994-R cells also showed tural change occluded part of the S984-binding pocket and amplification of KRAS on chromosome 12 (Fig. 5A, top thereby prevented compound binding. In contrast to S984, VI-3 panel). However, we did not observe an increase in KRAS contains a relatively small hydroxyl-methyl group interacting with expression, indicating the ERK2 amplicon to be the most Asp167. The flexibility associated with this could tolerate different likely relevant driver in these ERKi-R cells. Besides ERK2 conformations of Asp167, which could explain the sustained amplification, we also found focal amplification of MITF on activity of VI-3 against G186DERK1-mutant cells (Fig. 2D, low- chromosome 3 in SKMEL30-V11e-R, SKMEL30-VI-3-R, and er-middle plot). Mutation A191V is further away from the active SKMEL30-G994-R lines (Supplementary Fig. S4A). Consistent site, and as such, we were not able to identify a basis for its with the amplification, we found elevated expression of resistance based on structure and thus will require further studies. MITF in these lines (Supplementary Fig. S4B). Interestingly,

www.aacrjournals.org Clin Cancer Res; 2018 OF7

Jaiswal et al.

A IPC298-G994-R B C IPC298-G994-R

IPC298-V11e-R Parent EV ERK2-WT

pRSK (S359/63) RSK pERK (T202/Y204) ERK2 Flag-ERK2 Average RPKM Average ERK IPC298-V11e-R MIA PaCa2-S984-R pS6-RP (S235/6) S6-RP 50 100 150 200 250 300 β-ACTIN 0.0 0.5 1.0 1.5 2.0

Log2 copy number ratio

ERK2

D IPC298-G994-R MIA PaCa2-S984-R E IPC298-ERK2-WT F IPC298-G994-R 140 P 140 140 Parental 140 Parental 140 P 140 P P EV EV G994-R G994-R S984-R G994-R 120 120 120 ERK2-WT 120 ERK2-WT 120 120

100 100 100 100 100 100

80 80 80 80 80 80

60 60 60 60 60 60 Cell viability Cell viability Cell viability 40 Cell viability 40 Cell viability 40 Cell viability 40 40 40

20 20 20 20 20 20

0 0 0 0 0 0 -9 -8 -7 -6 -5 -4 -9 -8 -7 -6 -5 -4 -10-9-8-7-6-5-4 -10-9-8-7-6-5-4 -10-9-8-7-6-5-4 -10-9-8-7-6-5-4 G994, log [mol/L] S984, log [mol/L] G994, log [mol/L] V11e, log [mol/L] GDC-0973, log [mol/L] AZD6244, log [mol/L] MEK inhibitor H μ μ μ G MIA PaCa2-S984-R G994 ( mol/L) GDC-0973 ( mol/L) AZD6244 ( mol/L) Parental Parental G994-R Parental G994-R 140 P 140 P G994-R S984-R S984-R 0 0 120 0 0

120 0 0 10 10 10 0.1 1.0 0.1 0.1 1.0 0.1 0.1 1.0 0.1 10 1.0 10 1.0 10 1.0 100 100 pRSK (S359/63) 80 80 RSK 60 60 pERK (T202/Y204) Cell viability 40 Cell viability 40 ERK 20 20 pS6-RP (S235/6) 0 0 -9 -8 -7 -6 -5 -4 -9 -8 -7 -6 -5 -4 S6-RP GDC-0973, log [mol/L] AZD6244, log [mol/L] β-ACTIN

Figure 5. ERK2 amplification confers resistance to ERK inhibitors. A, Copy-number analysis of ERKi-R cells shows focal amplification of ERK2. B, Copy number versus expression of ERK2 in ERKi-R lines. Resistant lines that showed a positive correlation are labeled. C, Western blot showing ERK2 overexpression and downstream signaling in IPC298 cells. D and E, Effect of indicated ERKi on the proliferation of ERK2-amplified ERKi-resistant cells (D) or IPC298 cells overexpressing ERK2 (E). F and G, Effect of MEK inhibitors GDC-0973 and AZD6244 on proliferation of ERK2-amplified IPC298-G994-R (F) or MIA PaCa2-S984-R (G) cells. H, Western blot showing the MAPK and PIK3/mTOR signaling status in ERK2-amplified IPC298-G994-R cells treated with G994 ERKi or the indicated MEK inhibitors.

amplification/overexpression of MITF in melanoma cell lines amplified IPC298-G994-R (Fig. 5F) and MIA PaCa2-S984-R has been shown to confer resistance to BRAF/MEK-inhibitor cells (Fig. 5G) and IPC298 cells overexpressing ERK2-WT treatment (22, 67). (Supplementary Fig. S5A). The sensitivity of the ERK2- To further confirm that ERK2 amplification can confer resis- amplified ERKi-R lines and ERK2-overexpressing IPC298 cells tance to ERKi, we stably overexpressed ERK2-WT in IPC298 to MEK inhibitors was comparable with the sensitivity of the (Fig 5C) and tested the effect of ERKi on cell proliferation. parental cells (Fig. 5F and G; Supplementary Fig. S5A). In Consistent with the resistance observed in ERK2-amplified contrast to the observation with ERKi-R lines containing on- IPC298-G994-R and MIA PaCa2-S984-R with G994 and target ERK1/2 mutation, ERK2-amplified resistant cells were S984, respectively (Fig. 5D), ERK2-overexpressing IPC298 cells not sensitive to inhibitors from alternate scaffold classes showed resistance to both G994 and V11e (Fig 5E), confirming (Supplementary Fig. S5B and S5C). Furthermore, treatment that the elevated level of ERK2 expression is sufficient to with GDC-0973 or AZD6244 blocked MAPK and AKT/mTOR promote resistance to these compounds. signaling in ERK2-amplified IPC298-G994-R cells as indicated We hypothesized that the ERK2-amplified cells would be by decreased pRSK (S359/363), pERK1/2, and pS6-RP sensitive to MEK inhibition as they would be dependent on (S235/236; Fig. 5H). Taken together, our data indicate that upstream MAPK components for ERK activation. Consistent ERK2 amplification–mediated ERKi resistance can be over- with this, we found that MEK inhibitors GDC-0973 and come by treatment with upstream MEK inhibitors (Fig. 5F–H; AZD6244 were effective in blocking the growth of ERK2- Supplementary Fig. S5A).

OF8 Clin Cancer Res; 2018 Clinical Cancer Research

Mechanisms of Acquired Resistance to ERK Inhibitors

A B HCT116-parental H R-e11V-611TC HCT116-G994-R R-6xe-KM-611TCH

140 Canertinib 140 V11e 140 G994 140 MK-ex6 GDC-0980 Canertinib Canertinib + V11e Canertinib Canertinib + G994 Canertinib Canertinib + MK-ex6 120 120 GDC-0980 GDC-0980 + V11e 120 GDC-0980 GDC-0980 + G994 120 GDC-0980 GDC-0980 + MK-ex6

100 100 100 100

80 80 80 80

60 60 60 60 Cell viability Cell viability Cell viability 40 40 40 Cell viability 40

20 20 20 20

0 0 0 0 -10-9-8-7-6-5-4 -10-9-8-7-6-5-4 -10-9-8-7-6-5-4 -10-9-8-7-6-5-4 Compound Conc. log [mol/L] Compound Conc. log [mol/L] Compound Conc. log [mol/L] Compound Conc. log [mol/L]

C MIA PaCa2-parental MIA PaCa2-V11e-R MIA PaCa2-G994-R MIA PaCa2-MK-ex6-R

100 100 100 100

80 80 80 80

60 60 60 60 Cell viability Cell viability Cell viability 40 40 Cell viability 40 40

20 20 20 20

0 0 0 0 -10-9-8-7-6-5-4 -10-9-8-7-6-5-4 -10-9-8-7-6-5-4 -10-9-8-7-6-5-4 Compound Conc. log [mol/L] Compound Conc. log [mol/L] Compound Conc. log [mol/L] Compound Conc. log [mol/L]

F HCT116-G994-R Mia PaCa2-G994-R

Log2 FC Canertinib GDC-0980 Canertinib GDC-0980 0 0 0 0

μ 0.3 1.0 0.3 1.0 D HCT116 E MIA PaCa2 Drug ( mol/L) 0.3 1.0 0.3 1.0 80 80 pRSK (S359/63) EGFR EGFR ERBB2 ERBB2 * RSK 60 60 * * * pERK (T202/Y204)

40 40 ERK

RPKM * RPKM * pS6-RP (S235/6) 20 20 S6-RP

0 0 β-ACTIN l al -R -R -R a R R R e t 4- 1 n 1e- x6- 1 e 1 e arent V G994 ar P P V G99 MK-ex6 MK-

Figure 6.

RTK and PI3K/AKT/mTOR inhibitors overcome ERK-inhibitor resistance. A, Heatmap of IC50 fold change for compounds exhibiting a 4-fold or greater effect in at least one ERKi-R cell line tested (see Materials and Methods). B and C, Dose–response curve of parental or indicated ERKi-R HCT116 (B) and MIA PaCa2 (C) cells treated with ERKi or canertinib or GDC-0980 in the absence or presence of 2.5 mmol/L of ERKi. D and E, EGFR and ERBB2 expression assessed by RNA-seq (n ¼ 2) in parental and ERKi-R HCT116 (D) and MIA PaCa2 (E) cells. Data shown are mean SD of RPKM (read per kilobase per millions mapped reads). , P < 0.05 compared with respective parental cells when present. F, Western blot analysis of MAPK and downstream signaling in cells treated with canertinib or GDC-0980.

RTK and PI3K/mTOR inhibitors overcome acquired ERKi 0084, GDC-0068, and GDC-0980. We also found several resistance Aurora kinase inhibitors to be effective against HCT116 Although we found that on-target–related ERKi resistance ERKi-R lines but not in MIA PaCa2 ERKi-R cells. We further can be overcome with MEK inhibitors, we sought to identify validated the effect of ERBB inhibitor canertinib and PI3K/ additional inhibitors that might overcome resistance particu- mTOR inhibitor GDC-0980 for their effectiveness against larly in cells where the mechanism of resistance to ERKi does ERKi-R cells. We found canertinib to be effective in blocking not involve on-target mutations. Using cell viability as a read the growth of HCT116-V1e-R, HCT116-G994-R, HCT116-MK- out, we screened HCT116 and MIA PaCa2 ERKi-R–resistant ex6-R, MIA PaCa2-V11e-R, MIA PaCa2-G994R, and MIA cell lines against 474 compounds, including several approved PaCa2-MK-ex6-R in a dose-dependent manner, either by itself drugs, in the absence and presence of ERKi. Effective combi- or in the presence of ERKi (Fig. 6B and C; Supplementary nations were identified by determining the shift in IC50 values Fig. S6). Similarly, PI3K/mTOR inhibitor GDC-0980 sup- in the presence versus absence of ERKi (Fig. 6A). We found pressed cell viability in all the ERKi-resistant lines tested in that many compounds tested showed a differential effect on a dose-dependent manner (Fig. 6B and C; Supplementary cell viability in both HCT116 and MIA PaCa2 ERKi-R cells Fig. S6). Consistent with the efficacy of canertinib, we found (Fig. 6A). A majority of such compounds were either RTK evidence for increased expression of ERBB2 in HCT116-ERKi-R inhibitors or PI3K/AKT/mTOR inhibitors. RTK-inhibitor hits (Fig. 6D; Supplementary Table S6) and EGFR in MIA PaCa2- included canertinib (EGFR inhibitor), PD173074 (FGFR1 ERKi-R cells (Fig. 6E; Supplementary Table S6). Western blot inhibitor), and ponatinib (ABL kinase inhibitor). The PI3K/ analysis found that both canertinib and GDC-0980 blocked AKT/mTOR class of inhibitors included GDC-0032, GDC- MAPK and PI3K/mTOR signaling as confirmed by decreased

www.aacrjournals.org Clin Cancer Res; 2018 OF9

Jaiswal et al.

levels of pRSK and pS6-RP in both HCT116-G994-R and MIA that may arise from combination therapy involving MEK and ERK PaCa2-G994-R cells (Fig. 6F). inhibitors. However, we did not find any specific mutation or amplification of target ERK in several ERKi-R lines including all Discussion the MK-ex6-R lines indicating that diverse mechanisms of resistance can arise with different ERK inhibitor scaffold. Nevertheless, The MAPK pathway is a major therapeutic target for many these results provide a strong rationale for testing ERK mutation human cancers as it is deregulated or mutated in a third of cancers and amplification status, in addition to other described resistance (7, 8, 68). However, despite an initial dramatic response, patients mechanisms, upon patient relapse treated in the clinic following treated with RAF/MEK inhibitors eventually relapse because of ERK inhibitor treatment. This may help in designing alternate acquired resistance resulting from reactivation of the MAPK strategies for treatment that will likely provide benefit. pathway. Furthermore, preclinical studies suggest that ERK inhi- Our drug sensitivity screen identified RTK and PI3K/AKT/ bitors can be effective in overcoming resistance to RAF/MEK- mTOR pathway inhibitors as a key mediator of ERKi sensitivity inhibitor–based therapies (19, 30, 59) that show reactivation of in all ERKi-R HCT116 and MIA PaCa2 cell lines. Thus, our data MAPK signaling. Although it is early to conclude that ERK will be a suggest that in addition to MEK inhibitors, combination therapy better target for MAPK-driven tumors, it is plausible that even if involving ERK inhibitor with either RTK inhibitors such as pan- ERK inhibitors are successful in the clinic, resistance to these ERBB inhibitors or PI3K/AKT/mTOR inhibitors might also agents will likely emerge, as observed with most other kinase prevent resistance development. Although combination therapy small-molecule inhibitor therapies (14, 17, 59, 69). Preclinical has the potential to increase toxicity, it provides an attractive models can provide valuable tools to understand mechanisms of alternative where lower concentrations of each drug may prove to resistance to targeted therapies even before they emerge in clinic. be efficacious and safe. Alternatively, multiple drugs targeting Understanding resistance emergence can guide in the develop- the MAPK pathway administered sequentially can be effective, ment of effective strategies for clinical management of acquired as recently proposed (70). Thus, a rationally designed therapeutic resistance. strategy as described above can provide survival benefits to Using RAS/RAF-mutant cell lines, we have modeled the devel- patients by preventing onset of resistance or overcoming resis- opment of resistance to ERK inhibitors. We have identified several tance altogether. on-target ERK mutations that resulted in resistance to ERK inhibitors V11e and VI-3, both from a common scaffold class. We also fl found an on-target ERK mutation that arose in response to Disclosure of Potential Con icts of Interest All authors are employees of Genentech Inc. and shareholders of Roche treatment with S984. A previous study using random mutagenesis fl fi Holdings Inc. No potential con icts of interest were disclosed by the other identi ed some of the on-target ERK mutations that confer authors. resistance to V11e (36). Consistent with our findings, a recent study involving S984 ERKi identified the G186D ERK1 mutation Authors' Contributions in HCT116 cells as a cause of resistance to S984 (37). Structural Conception and design: B.S. Jaiswal, S. Seshagiri analysis of the ERK indicated that the mutations affect the binding Development of methodology: B.S. Jaiswal, J. Yin, S. Seshagiri of ERK inhibitors and thus prevent them from blocking ERK Acquisition of data (provided animals, acquired and managed patients, activity. Further, ERK1/2-mutant–resistant cells in this study, provided facilities, etc.): B.S. Jaiswal, J. Yin, W. Wang, E. Lin, S.E. Martin, although cross-resistant to the ERK inhibitors from same scaffold Z. Modrusan Analysis and interpretation of data (e.g., statistical analysis, biostatistics, class, were sensitive to ERK inhibitors belonging to alternate computational analysis): B.S. Jaiswal, S. Durinck, E.W. Stawiski, W. Wang, fi scaffold classes. Taken together, these ndings suggest that the E. Lin, J. Moffat, S.E. Martin, S. Seshagiri ERK mutations are likely to be a major mechanism of resistance to Writing, review, and/or revision of the manuscript: B.S. Jaiswal, E.W. Stawiski, ERK inhibitors in patients treated with V11e or other compounds J. Yin, W. Wang, J. Moffat, S.E. Martin, S. Seshagiri from this scaffold class. Administrative, technical, or material support (i.e., reporting or organizing In addition to on-target ERK mutations, for the first time, we data, constructing databases): B.S. Jaiswal Study supervision: B.S. Jaiswal, S. Seshagiri identified ERK2 amplification as a mechanism of resistance to fi ERK inhibitors. This mode of resistance was not scaffold-speci c. Acknowledgments – Consistent with this, unlike the ERK-mutant resistant lines, The authors thank members of the NGS Sequencing lab for their help ERK2-amplified lines were not sensitive to cross-scaffold inhibi- with sequencing and Kate Sanger for editing and critical comments on tors. However, both ERK-mutated and ERK-amplified resistant the article. They also thank Aju Antony, SciGenom Labs, and Derek lines were sensitive to MEK inhibitors, indicating ERKi resistance Vargas, MedGenome Inc., for their help with generation of the ERK-mutant can be managed with MEK inhibitors. Perhaps a combination constructs. therapy involving ERK and MEK inhibitors might limit or elim- inate the emergence of acquired resistance to either of the drugs. It The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked is plausible that simultaneous treatment of MEK and ERK inhi- advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate bitors will give synergistic toxicities. In this scenario, sequential this fact. treatment with ERK inhibitors first followed by MEK inhibitors may produce survival benefits to patient by delaying and/or Received December 7, 2017; revised March 30, 2018; accepted May 8, 2018; limiting the emergence of acquired resistance and reduce toxicities published first May 14, 2018.

References 1. Raman M, Chen W, Cobb MH. Differential regulation and properties of 2. Avruch J, Khokhlatchev A, Kyriakis JM, Luo Z, Tzivion G, Vavvas MAPKs. Oncogene 2007;26:3100–12. D, et al. Ras activation of the Raf kinase: tyrosine kinase

OF10 Clin Cancer Res; 2018 Clinical Cancer Research

Mechanisms of Acquired Resistance to ERK Inhibitors

recruitment of the MAP kinase cascade. Recent Prog Horm Res 28. Corcoran RB, Dias-Santagata D, Bergethon K, Iafrate AJ, Settleman J, 2001;56:127–55. Engelman JA. BRAF gene amplification can promote acquired resistance 3. Dhillon AS, Hagan S, Rath O, Kolch W. MAP kinase signalling pathways in to MEK inhibitors in cancer cells harboring the BRAF V600E mutation. Sci cancer. Oncogene 2007;26:3279–90. Signal 2010;3:ra84. 4. Wellbrock C, Karasarides M, Marais R. The RAF proteins take centre stage. 29. Hatzivassiliou G, Liu B, O'Brien C, Spoerke JM, Hoeflich KP, Haverty PM, Nat Rev Mol Cell Biol 2004;5:875–85. et al. ERK inhibition overcomes acquired resistance to MEK inhibitors. Mol 5. Yoon S, Seger R. The extracellular signal-regulated kinase: multiple Cancer Ther 2012;11:1143–54. substrates regulate diverse cellular functions. Growth Factors 2006;24: 30. Morris EJ, Jha S, Restaino CR, Dayananth P, Zhu H, Cooper A, et al. 21–44. Discovery of a novel ERK inhibitor with activity in models of acquired 6. Chang L, Karin M. Mammalian MAP kinase signalling cascades. Nature resistance to BRAF and MEK inhibitors. Cancer Discov 2013;3:742–50. 2001;410:37–40. 31. Wong DJ, Robert L, Atefi MS, Lassen A, Avarappatt G, Cerniglia M, et al. 7. Davies H, Bignell GR, Cox C, Stephens P, Edkins S, Clegg S, et al. Mutations Antitumor activity of the ERK inhibitor SCH772984 [corrected] against of the BRAF gene in human cancer. Nature 2002;417:949–54. BRAF mutant, NRAS mutant and wild-type melanoma. Mol Cancer 2014; 8. Roberts PJ, Der CJ. Targeting the Raf-MEK-ERK mitogen-activated protein 13:194. kinase cascade for the treatment of cancer. Oncogene 2007;26:3291–310. 32. Genentech. A study of the safety, tolerability, and effects of cobimetinib 9. Fernandez-Medarde A, Santos E. Ras in cancer and developmental diseases. and GDC-0994 in patients with locally advanced or metastatic solid Genes Cancer 2011;2:344–58. tumors. In ClinicalTrials.gov [Internet]. Bethesda, MD: National Library 10. Baines AT, Xu D, Der CJ. Inhibition of Ras for cancer treatment: the search of Medicine. 2017. Available from: https://clinicaltrials.gov/ct2/show/ continues. Future Med Chem 2011;3:1787–808. NCT02457793. 11. Cox AD, Fesik SW, Kimmelman AC, Luo J, Der CJ. Drugging the undrug- 33. Moschos SJ, Sullivan RJ, Hwu WJ, Ramanathan RK, Adjei AA, Fong PC, et al. gable RAS: mission possible? Nat Rev Drug Discov 2014;13:828–51. Development of MK-8353, an orally administered ERK1/2 inhibitor, in 12. Downward J. RAS's cloak of invincibility slips at last? Cancer Cell patients with advanced solid tumors. JCI Insight 2018;3:pii: 92352. 2014;25:5–6. 34. Novartis Pharmaceuticals. A phase I clinical study with investigational 13. Chapman PB, Hauschild A, Robert C, Haanen JB, Ascierto P, Larkin J, et al. compound LTT462 in adult patients with specific advanced cancers. In Improved survival with vemurafenib in melanoma with BRAF V600E ClinicalTrials.gov [Internet]. Bethesda, MD: National Library of Medicine. mutation. N Engl J Med 2011;364:2507–16. 2016. Available from: https://clinicaltrials.gov/ct2/show/NCT02711345. 14. Flaherty KT, Infante JR, Daud A, Gonzalez R, Kefford RF, Sosman J, et al. 35. Sullivan RJ, Infante JR, Janku F, Wong DJL, Sosman JA, Keedy V, et al. Combined BRAF and MEK inhibition in melanoma with BRAF V600 First-in-class ERK1/2 inhibitor ulixertinib (BVD-523) in patients with mutations. N Engl J Med 2012;367:1694–703. MAPK mutant advanced solid tumors: results of a phase I dose-escalation 15. Flaherty KT, Puzanov I, Kim KB, Ribas A, McArthur GA, Sosman JA, et al. and expansion study. Cancer Discov 2018;8:184–95. Inhibition of mutated, activated BRAF in metastatic melanoma. N Engl J 36. Goetz EM, Ghandi M, Treacy DJ, Wagle N, Garraway LA. ERK mutations Med 2010;363:809–19. confer resistance to mitogen-activated protein kinase pathway inhibitors. 16. Flaherty KT, Robert C, Hersey P, Nathan P, Garbe C, Milhem M, et al. Cancer Res 2014;74:7079–89. Improved survival with MEK inhibition in BRAF-mutated melanoma. 37. Jha S, Morris EJ, Hruza A, Mansueto MS, Schroeder GK, Arbanas J, et al. N Engl J Med 2012;367:107–14. Dissecting therapeutic resistance to ERK inhibition. Mol Cancer Ther 17. Trunzer K, Pavlick AC, Schuchter L, Gonzalez R, McArthur GA, Hutson TE, 2016;15:548–59. et al. Pharmacodynamic effects and mechanisms of resistance to vemur- 38. Long GV, Fung C, Menzies AM, Pupo GM, Carlino MS, Hyman J, et al. afenib in patients with metastatic melanoma. J Clin Oncol 2013;31: Increased MAPK reactivation in early resistance to dabrafenib/trametinib 1767–74. combination therapy of BRAF-mutant metastatic melanoma. Nat Com- 18. Nazarian R, Shi H, Wang Q, Kong X, Koya RC, Lee H, et al. Melanomas mun 2014;5:5694. acquire resistance to B-RAF(V600E) inhibition by RTK or N-RAS upregula- 39. Long JEW, Chung MJ, Wang WJ, Riegler B, Li J, Nannini J, et al. Therapy- tion. Nature 2010;468:973–7. induced stromal remodeling enables evolution of non-genomic resistance 19. Wagle N, Van Allen EM, Treacy DJ, Frederick DT, Cooper ZA, Taylor-Weiner in BRAF mutant melanoma. In press. A, et al. MAP kinase pathway alterations in BRAF-mutant melanoma 40. Jaiswal BS, Kljavin NM, Stawiski EW, Chan E, Parikh C, Durinck S, et al. patients with acquired resistance to combined RAF/MEK inhibition. Oncogenic ERBB3 mutations in human cancers. Cancer Cell 2013;23: Cancer Discov 2014;4:61–8. 603–17. 20. Poulikakos PI, Persaud Y, Janakiraman M, Kong X, Ng C, Moriceau G, et al. 41. Jaiswal BS, Janakiraman V, Kljavin NM, Eastham-Anderson J, Cupp JE, RAF inhibitor resistance is mediated by dimerization of aberrantly spliced Liang Y, et al. Combined targeting of BRAF and CRAF or BRAF and PI3K BRAF(V600E). Nature 2011;480:387–90. effector pathways is required for efficacy in NRAS mutant tumors. PLoS 21. Johannessen CM, Boehm JS, Kim SY, Thomas SR, Wardwell L, Johnson LA, One 2009;4:e5717. et al. COT drives resistance to RAF inhibition through MAP kinase pathway 42. DePristo MA, Banks E, Poplin R, Garimella KV, Maguire JR, Hartl C, et al. reactivation. Nature 2010;468:968–72. A framework for variation discovery and genotyping using next-generation 22. Van Allen EM, Wagle N, Sucker A, Treacy DJ, Johannessen CM, Goetz EM, DNA sequencing data. Nat Genet 2011;43:491–8. et al. The genetic landscape of clinical resistance to RAF inhibition in 43. Saunders CT, Wong WS, Swamy S, Becq J, Murray LJ, Cheetham RK. Strelka: metastatic melanoma. Cancer Discov 2014;4:94–109. accurate somatic small-variant calling from sequenced tumor-normal 23. Shi H, Hugo W, Kong X, Hong A, Koya RC, Moriceau G, et al. Acquired sample pairs. Bioinformatics 2012;28:1811–7. resistance and clonal evolution in melanoma during BRAF inhibitor 44. Cibulskis K, Lawrence MS, Carter SL, Sivachenko A, Jaffe D, Sougnez C, et al. therapy. Cancer Discov 2014;4:80–93. Sensitive detection of somatic point mutations in impure and heteroge- 24. Shi H, Moriceau G, Kong X, Lee MK, Lee H, Koya RC, et al. Melanoma neous cancer samples. Nat Biotechnol 2013;31:213–9. whole-exome sequencing identifies (V600E)B-RAF amplification-mediat- 45. Wu TD, Nacu S. Fast and SNP-tolerant detection of complex variants and ed acquired B-RAF inhibitor resistance. Nat Commun 2012;3:724. splicing in short reads. Bioinformatics 2010;26:873–81. 25. Wagle N, Emery C, Berger MF, Davis MJ, Sawyer A, Pochanard P, et al. 46. Robinson MD, McCarthy DJ, Smyth GK. edgeR: a Bioconductor package for Dissecting therapeutic resistance to RAF inhibition in melanoma by tumor differential expression analysis of digital gene expression data. Bioinfor- genomic profiling. J Clin Oncol 2011;29:3085–96. matics 2010;26:139–40. 26. Emery CM, Vijayendran KG, Zipser MC, Sawyer AM, Niu L, Kim JJ, et al. 47. Olshen AB, Venkatraman ES, Lucito R, Wigler M. Circular binary segmen- MEK1 mutations confer resistance to MEK and B-RAF inhibition. Proc Natl tation for the analysis of array-based DNA copy number data. Biostatistics Acad Sci U S A 2009;106:20411–6. 2004;5:557–72. 27. Wang H, Daouti S, Li WH, Wen Y, Rizzo C, Higgins B, et al. Identification of 48. Kabsch W. XDS. Acta Crystallogr D Biol Crystallogr 2010;66:125–32. the MEK1 (F129L) activating mutation as a potential mechanism of 49. Winn MD, Ballard CC, Cowtan KD, Dodson EJ, Emsley P, Evans PR, et al. acquired resistance to MEK inhibition in human cancers carrying the Overview of the CCP4 suite and current developments. Acta Crystallogr B-RafV600E mutation. Cancer Res 2011;71:5535–45. D Biol Crystallogr 2011;67:235–42.

www.aacrjournals.org Clin Cancer Res; 2018 OF11

Jaiswal et al.

50. Burdick DJ, Wang S, Heise C, Pan B, Drummond J, Yin J, et al. Fragment- 60. Lito P, Pratilas CA, Joseph EW, Tadi M, Halilovic E, Zubrowski M, et al. based discovery of potent ERK2 pyrrolopyrazine inhibitors. Bioorg Med Relief of profound feedback inhibition of mitogenic signaling by RAF Chem Lett 2015;25:4728–32. inhibitors attenuates their activity in BRAFV600E melanomas. Cancer Cell 51. McCoy AJ, Grosse-Kunstleve RW, Adams PD, Winn MD, Storoni LC, 2012;22:668–82. Read RJ. Phaser crystallographic software. J Appl Crystallogr 2007;40: 61. Brenan L, Andreev A, Cohen O, Pantel S, Kamburov A, Cacchiarelli D, et al. 658–74. Phenotypic characterization of a comprehensive set of MAPK1/ERK2 52. Murshudov GN, Vagin AA, Dodson EJ. Refinement of macromolecular missense mutants. Cell Rep 2016;17:1171–83. structures by the maximum-likelihood method. Acta Crystallogr D Biol 62. Dankort D, Curley DP, Cartlidge RA, Nelson B, Karnezis AN, Damsky WE Jr, Crystallogr 1997;53:240–55. et al. Braf(V600E) cooperates with Pten loss to induce metastatic mela- 53. Smart OS, Womack TO, Flensburg C, Keller P, Paciorek W, Sharff A, et al. noma. Nat Genet 2009;41:544–52. Exploiting structure similarity in refinement: automated NCS and target- 63. Imai YN, Inoue Y, Nakanishi I, Kitaura K. Cl-pi interactions in protein- structure restraints in BUSTER. Acta Crystallogr D Biol Crystallogr ligand complexes. Protein Sci 2008;17:1129–37. 2012;68:368–80. 64. Misale S, Yaeger R, Hobor S, Scala E, Janakiraman M, Liska D, et al. 54. Haverty PM, Lin E, Tan J, Yu Y, Lam B, Lianoglou S, et al. Reproducible Emergence of KRAS mutations and acquired resistance to anti-EGFR pharmacogenomic profiling of cancer cell line panels. Nature 2016;533: therapy in colorectal cancer. Nature 2012;486:532–6. 333–7. 65. Engelman JA, Zejnullahu K, Mitsudomi T, Song Y, Hyland C, Park JO, et al. 55. Aronov AM, Tang Q, Martinez-Botella G, Bemis GW, Cao J, Chen G, et al. MET amplification leads to gefitinib resistance in lung cancer by activating Structure-guided design of potent and selective pyrimidylpyrrole inhibi- ERBB3 signaling. Science 2007;316:1039–43. tors of extracellular signal-regulated kinase (ERK) using conformational 66. Bean J, Brennan C, Shih JY, Riely G, Viale A, Wang L, et al. MET ampli- control. J Med Chem 2009;52:6362–8. fication occurs with or without T790M mutations in EGFR mutant lung 56. Blake JF, Burkard M, Chan J, Chen H, Chou KJ, Diaz D, et al. Discovery of tumors with acquired resistance to gefitinib or erlotinib. Proc Natl Acad Sci (S)-1-(1-(4-Chloro-3-fluorophenyl)-2-hydroxyethyl)-4-(2-((1-methyl- U S A 2007;104:20932–7. 1H-pyrazol-5-y l)amino)pyrimidin-4-yl)pyridin-2(1H)-one (GDC-0994), 67. Garraway LA, Widlund HR, Rubin MA, Getz G, Berger AJ, Ramaswamy S, an extracellular signal-regulated kinase 1/2 (ERK1/2) inhibitor in early et al. Integrative genomic analyses identify MITF as a lineage survival clinical development. J Med Chem 2016;59:5650–60. oncogene amplified in malignant melanoma. Nature 2005;436:117–22. 57. Lim J, Kelley EH, Methot JL, Zhou H, Petrocchi A, Mansoor UF, et al. ; Merck 68. McCubrey JA, Steelman LS, Chappell WH, Abrams SL, Montalto G, Sharp & Dohme Corp., assignee. Novel compounds that are erk inhibitors. Cervello M, et al. Mutations and deregulation of Ras/Raf/MEK/ERK and Patent WO2013063214 A1. 2013 May 2. PI3K/PTEN/Akt/mTOR cascades which alter therapy response. Oncotarget 58. Rudolph J, Xiao Y, Pardi A, Ahn NG. Slow inhibition and conformation 2012;3:954–87. selective properties of extracellular signal-regulated kinase 1 and 2 inhi- 69. Gainor JF, Dardaei L, Yoda S, Friboulet L, Leshchiner I, Katayama R, et al. bitors. Biochemistry 2015;54:22–31. Molecular mechanisms of resistance to first- and second-generation ALK 59. Ahronian LG, Sennott EM, Van Allen EM, Wagle N, Kwak EL, Faris JE, et al. inhibitors in ALK-rearranged lung cancer. Cancer Discov 2016;6:1118–33. Clinical acquired resistance to RAF inhibitor combinations in braf-mutant 70. Xue Y, Martelotto L, Baslan T, Vides A, Solomon M, Mai TT, et al. An colorectal cancer through MAPK pathway alterations. Cancer Discov approach to suppress the evolution of resistance in BRAF(V600E)-mutant 2015;5:358–67. cancer. Nat Med 2017;23:929–37.

OF12 Clin Cancer Res; 2018 Clinical Cancer Research

ERK Mutations and Amplification Confer Resistance to ERK-Inhibitor Therapy

Bijay S. Jaiswal, Steffen Durinck, Eric W. Stawiski, et al.

Clin Cancer Res Published OnlineFirst May 14, 2018.

Updated version Access the most recent version of this article at: doi:10.1158/1078-0432.CCR-17-3674

Supplementary Access the most recent supplemental material at: Material http://clincancerres.aacrjournals.org/content/suppl/2018/05/12/1078-0432.CCR-17-3674.DC1

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 Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/early/2018/06/26/1078-0432.CCR-17-3674. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.