Distinct Transcriptional Programming Drive Response to MAPK Inhibition in BRAFV600 Mutant Melanoma Patient Derived Xenografts

Author Manuscript Published OnlineFirst on September 16, 2019; DOI: 10.1158/1535-7163.MCT-19-0028 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

2 Distinct transcriptional programming drive response to MAPK inhibition in BRAFV600 mutant 3 melanoma patient derived xenografts

4 Tianshu Feng**, Javad Golji, Ailing Li†, Xiamei Zhang, David A. Ruddy, Daniel P. Rakiec, Felipe C. 5 Geyer, Jane Gu, Hui Gao‡, Juliet A. Williams, Darrin D. Stuart*, Matthew J. Meyer*,†

6 Oncology Drug Discovery, Novartis Institutes for BioMedical Research (NIBR), Cambridge MA, USA

8 Running title: Transcriptional programs predict response to MAPK inhibition

9 Keywords: Mitogen-activated protein kinase (MAPK); BRAF; metastatic melanoma; patient derived 10 xenograft (PDX); targeted therapy

12 *Corresponding authors: Darrin D. Stuart, Novartis Institutes for BioMedical Research, 250 13 Massachusetts Ave, Cambridge MA, USA; Phone: (617)871-5311; E-mail: [email protected] 14 and Matthew J. Meyer, Novartis Institutes for BioMedical Research, 250 Massachusetts Ave, Cambridge 15 MA, USA; Phone: (617)494-7466; E-mail: [email protected]

16 **Current address: Tanshu Feng, Tango Therapeutics, 100 Binney St, Cambridge MA 02142

17 †Current address: Ailing Li and Matthew J. Meyer, Discovery Oncology, Bristol-Myers Squibb, 100 18 Binney St, Cambridge MA 02142

19 ‡Current address: Hui Gao, AstraZeneca, 35 Gatehouse Dr., Waltham MA 02451

20 Disclosure of potential conflicts of interest: This research was funded by Novartis. All authors were 21 employees of Novartis at the time the study was performed.

23 Word count: 5,040

24 6 figures

25 6 supplementary figures

26 3 supplementary tables

Page 1 of 21

Downloaded from mct.aacrjournals.org on September 27, 2021. © 2019 American Association for Cancer Research. Author Manuscript Published OnlineFirst on September 16, 2019; DOI: 10.1158/1535-7163.MCT-19-0028 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

36 Abstract

37 Inhibitors targeting BRAF and its downstream kinase MEK produce robust response in patients with

38 advanced BRAFV600 mutant melanoma. However, the duration and depth of response vary significantly

39 between patients, therefore predicting response a priori remains a significant challenge. Here, we utilized

40 the Novartis collection of patient-derived xenografts to characterize transcriptional alterations elicited by

41 BRAF and MEK inhibitors in vivo, in an effort to identify mechanisms governing differential response to

42 MAPK inhibition. We show that the expression of a MITF-high, ‘epithelial-like’ transcriptional program is

43 associated with reduced sensitivity and adaptive response to BRAF and MEK inhibitor treatment. On the

44 other hand, xenograft models that express a MAPK-driven ‘mesenchymal-like’ transcriptional program are

45 preferentially sensitive to MAPK inhibition. These gene expression programs are somewhat similar to the

46 MITF high and low phenotypes described in cancer cell lines, but demonstrate an inverse relationship

47 with drug response. This suggests a discrepancy between in vitro and in vivo experimental systems that

48 warrants future investigations. Finally, BRAFV600 mutant melanoma rely on either MAPK or alternative

49 pathways for survival under BRAF and MEK inhibition in vivo, which in turn predict their response to

50 further pathway suppression using a combination of BRAF, MEK, and ERK inhibitors. Our findings

51 highlight the inter-tumor heterogeneity in BRAFV600 mutant melanoma, and the need for precision

52 medicine strategies to target this aggressive cancer.

Page 2 of 21

65 Introduction

66 Approximately 50% of cutaneous melanoma harbor oncogenic mutations of BRAF. The majority of BRAF

67 mutations occur in the V600 position, resulting in constitutive activation of the MAPK signaling pathway

68 (1,2). The inhibition of BRAF, either alone or in combination with MEK, provides remarkable clinical

69 benefits for patients with advanced BRAFV600 mutant melanoma (3-6). Nevertheless, despite the robust

70 initial response, most patients eventually develop resistance. Acquired resistance to MAPK inhibition

71 involves heterogeneous genetic and non-genetic mechanisms that activate MAPK or other compensatory

72 signaling pathways (7-9). As such, there is a need for patient stratification and treatment strategies in

73 order to improve response and delay the onset of disease progression.

74 A subset of BRAFV600 mutant melanoma patients progress rapidly upon treatment using targeted therapy

75 (3-6). In contrast, approximately 20% of patients benefit from durable response to BRAF and MEK

76 inhibition, as measured by progression free survival at three years (10,11). The disparity between these

77 two patient groups is not well understood. Clinical factors related to disease burden, including lactate

78 dehydrogenase levels and the number of metastases, are important prognostic markers (10,11). At the

79 molecular level, melanoma cell line sensitivity to BRAF-targeted therapy is linked to the melanoma cell

80 state (12,13). Gene expression profiling revealed that melanoma fall into two distinct transcriptional states

81 irrespective of their mutation status, characterized by proliferative or invasive features and MITF

82 expression levels (14-16). The invasive, or MITF-low and AXL-high phenotype, confers intrinsic

83 resistance to MAPK inhibition in BRAFV600 mutant melanoma in vitro (12,13). However, both high and low

84 MITF expression levels are implicated in therapeutic resistance against MAPK inhibition in melanoma

85 (17). It remains to be determined whether the melanoma cell state predicts response to BRAF inhibitor-

86 based therapy in the clinical setting.

87 In this study, we sought to elucidate mechanisms governing differential sensitivity to BRAF and MEK

88 inhibition, using our internally established patient-derived xenograft (PDX) collection (18). Work by our

89 group and others demonstrated the prognostic value of PDXs in modeling clinical response to targeted

90 therapies (18-20). In comparison to other preclinical experimental systems, PDX models conserve tumor

91 architecture and heterogeneity, and faithfully recapitulate patient response to chemo- and targeted

Page 3 of 21

92 therapies (18,19,21-23). Moreover, since obtaining high quality research biopsies from the clinic is

93 challenging, PDXs enable profiling of treatment and time matched tumors with statistically powered

94 replicates. Here, we characterized on-treatment PDX tumor samples to compare the pharmacodynamic

95 effects of BRAF and MEK inhibitors in models that achieved complete regression versus partial response

96 and stable disease. Transcriptome profiling revealed that in vivo, a MITF-high ‘epithelial-like’ phenotype is

97 associated with adaptive and intrinsic resistance, whereas a MAPK-high ‘mesenchymal-like’ phenotype is

98 linked to sensitivity to BRAF and MEK inhibitor treatment. This is different from findings using in vitro

99 cancer cell lines (12,13,17), and suggests an in vitro-in vivo discrepancy that warrants additional

100 investigation. In less sensitive models, the lack of tumor regression in response to treatment was model

101 dependent and mediated by either MAPK dependent or independent mechanisms. Our results suggested

102 that response of BRAFV600 mutant melanoma xenografts to BRAF and MEK inhibitors is heterogeneous,

103 and precision medicine-based combination strategies are required to target this aggressive cancer.

104 Importantly, this study provided insights into determinants of therapeutic response in BRAFV600 melanoma

105 in vivo.

106

107 Material and Methods

108 Mouse xenograft studies

109 Mice were maintained and handled in accordance to the Novartis Institutes for BioMedical Research

110 (NIBR) Animal Care and Use Committee protocols and regulations. PDX models were procured and

111 established as previously described (18). Surgical tumor tissues were obtained from treatment-naïve

112 cancer patients, all of whom provided informed written consent for the samples procured by Novartis Inc.

113 PDX fragments were implanted subcutaneously into the right flanks of female nude mice (Charles River)

114 using a trocar. Once the tumor volumes reached 300-500 mm3 in size, mice were randomized into

115 respective groups. For tissue collection, mice were dosed twice daily by oral gavage with encorafenib and

116 binimetinib at 20 mg per kg and 3.5 mg per kg, respectively. These doses were selected as they achieved

117 exposures in mice comparable to that in humans (4). For efficacy studies, encorafenib and binimetinib

Page 4 of 21

118 were administered via diet supplementation (200 mg encorafenib and 35 mg binimetinib per kg of chow;

119 Bio-Serv). The supplemented diet was designed to achieve similar drug exposure to twice daily oral

120 gavage. The ERK inhibitor VX-11e was administered orally at 50 mg per kg daily. Body weight and tumor

121 size were monitored twice per week. Tumor volume was measured using a caliper and calculated as

122 (length x width x width)/2. Best average response and response calls were determined as previously

123 described (18). All studies were performed using PDXs between passages 4-9.

124 RNA extraction and sequencing

125 Total RNA were extracted from snap frozen tumor fragments, using the RNeasy kit with the QIAcube

126 (QIAGEN) automated sample preparation platform. The RNA concentration and integrity were measured

127 using the RNA 6000 nano kit (Agilent Technologies) on the Agilent 2100 BioAnalyzer.

128 200 ng of high purity RNA was used as input to generate sample libraries using the TruSeq Stranded

129 mRNA library prep kit (Illumina), following manufacturer’s instructions on the Hamilton STAR robotics

130 platform. The PCR amplified RNA-seq library products were quantified using the Fragment Analyzer

131 Standard Sensitivity NGS Fragment Analysis Kit (Advanced Analytical Technologies). Samples were

132 diluted to 10 nM in Elution Buffer (QIAGEN), denatured, and loaded at a range of 2.5 to 4.0 pM on an

133 Illumina cBOT using the HiSeq® 4000 PE Cluster Kit (Illumina). RNASeq libraries were sequenced on a

134 HiSeq® 4000 at 75 base pair paired-end with 8 base pair dual indexes using the HiSeq® 4000 SBS Kit,

135 150 cycles (Illumina). The sequence intensity files were generated on instrument using the Illumina Real

136 Time Analysis software and resulting intensity files were demultiplexed with the bcl2fastq2.

137 Gene expression analysis

138 Gene level read counts were normalized and voom transformated (24) for differential expression using

139 the limma framework (25), and P-values were corrected for multiple testing using the Benjamini-Hochberg

140 method. Genes were considered differentially expressed if the log2-fold change > 1.5 in absolute value

141 and the adjusted P-value < 0.05. Log transformed and TMM normalized (26) expression values were

142 used for clustering analysis with NbClust (27). K=5 was selected as the most informative partition. A

143 hypergeometric distribution function was used to determine enrichment P-values for overlap of each

Page 5 of 21

144 cluster membership with gene sets from mSigDB (28), in addition to transcription factor and canonical

145 gene sets exported from MetaCore (Clarivate Analytics). Gene set enrichment P-values were adjusted

146 using the Benjamini-Hochberg method. Mutation and copy number calls were made as described in (18).

147 The RNA-seq data has been submitted to the Sequence Read Archive (SRA) database; submission

148 number SUB6174690.

149 Clinical dataset

150 Normalized and background corrected microarray data was downloaded from Gene Expression Omnibus

151 public database, accession number GSE99898 (29). The dataset was derived from 17 patients treated

152 with dabrafenib, vemurafenib, or dabrafenib and trametinib. 17 untreated, 8 on-treatment, and 13

153 resistant samples were collected in total. Patient response data was obtained from Table 1 in the

154 reference publication (29). Gene signature scores were calculated as sum of the normalized and z-

155 transformed gene expression values for all genes of interest.

156 Western blot analysis

157 Snap frozen tumor fragments were suspended in RIPA lysis buffer with protease (Roche) and

158 phosphatase (Sigma) inhibitor cocktails, and mechanically homogenized with tungsten beads using the

159 TissueLyser (QIAGEN). Protein concentrations were quantified using the Pierce BCA protein assay kit

160 (Thermo Fisher Scientific). Western blotting was performed using standard procedures. Antibodies

161 against MEK1/2, pMEK1/2 (S217/S222), ERK1/2, pERK1/2 (T202/Y204), FRA1, pFRA1 (S265),

162 RSK1/2/3, pRSK3 (T356/S360), FOSB, EGR1, AKT, pAKT (S473), S6, pS6 (S240/S244), Cyclin D1, pRB

163 (S780), N-cadherin, E-cadherin, vimentin, BIM, MITF, and GAPDH were purchased from Cell Signaling

164 Technology. Antibodies against ERF and pERF (T526) were purchased from Invitrogen. Signals were

165 developed using SuperSignal West Femto, Dura, or Pico chemiluminescent substrates (Thermo Fisher

166 Scientific), and visualized using the ChemiDoc Imaging System (Bio-Rad).

167 RT-PCR

168 RT-PCR reactions were performed using Taqman® Gene Expression master mix (Applied Biosystems),

169 FAM-labeled probe for DUSP6, and Beta-2-macroglobulin (B2M) as a normalization control. Samples

Page 6 of 21

170 were run on a 7900HT Real-Time PCR machine (Applied Biosystems) and data was analyzed and

171 normalized according to manufacturer’s instructions (2-ΔCt method).

172 Results

173 Characteristics of the BRAFV600 mutant melanoma PDXs

174 Of all the cutaneous melanoma models in the Novartis PDX encyclopedia, 13 models had BRAFV600

175 mutations (18). Fig. 1A shows the common genetic lesions identified in these models. CDKN2A loss-of-

176 function occurred at a higher frequency in these PDX models relative to human tumors (2), consistent

177 with results from a previous study (30). MITF copy number alterations and mutations were found in five

178 PDX models with potential oncogenic effects (1), but no trends associated with drug response were

179 observed.

180 The Novartis mouse clinical trial (18) demonstrated that response of the BRAFV600 mutant PDX models to

181 continuous treatment with encorafenib and binimetinib ranged from complete regression (CR), to partial

182 regression (PR) and stable disease (SD) (supplementary Fig 1). Consistent with the phase 3 COLUMBUS

183 trial where the disease control rate was >90% (4), none of the PDX models displayed outright resistance

184 (18) (supplementary Fig 1). In addition to the PDXs that were enrolled in the mouse clinical trial (18), one

185 additional model, X-20767, was also included in subsequent experiments (Fig. 1B).

186 BRAF and MEK inhibitor treatment elicits distinct transcriptional regulation in vivo

187 To identify factors associated with differential drug response in BRAFV600 mutant melanoma in vivo, we

188 characterized immediate and longer term transcriptional changes in response to encorafenib and

189 binimetinib treatment in our PDX models. RNA sequencing (RNA-seq) was performed on tumor fragments

190 harvested prior to treatment, eight hours post a single dose, or eight hours post the final dose on the ninth

191 day of continuous treatment (Fig. 1C). Three biological replicates were collected for each time point, with

192 the exception of X-1906 which only had two RNA-seq samples for day nine due to limitation in tumor size.

193 In total, two CR and four PR/SD models were profiled. The day nine samples for X-2613 were excluded

194 from the analysis, because they were made up of >80% mouse RNA.

Page 7 of 21

195 We applied limma-voom, a well-established method for RNA-seq differential expression analysis (24,25),

196 to identify differentially expressed genes (DEGs) between treated and untreated conditions for each

197 model (supplementary Table 1). For all six PDX models, more transcriptome reprogramming occurred

198 after nine days of treatment in comparison to a single dose (Fig. 1D). There was significant overlap in

199 differential gene expression between early and late time points, suggesting that the early transcriptome

200 changes were retained after prolonged treatment. Furthermore, only a small subset of DEGs were shared

201 across multiple models, whereas a significant number of DEGs were unique to each model

202 (supplementary Fig. 2), indicating heterogeneity among PDXs.

203 In an attempt to better understand patterns of transcriptional reprogramming by BRAF and MEK inhibitors

204 in vivo, we performed unsupervised clustering of treatment-matched samples from all six PDX models

205 using DEGs from the aforementioned analyses (supplementary table 1). Only the top 1000 most variable

206 DEGs across all samples were included in the clustering analysis, which may enrich for model- and time-

207 dependent gene expression differences in addition to post-treatment transcriptome alterations. Five major

208 clusters were identified (Fig. 2). To elucidate the biological functions of each cluster, gene set enrichment

209 was performed using the Broad Institutes’ Molecular Signatures Database (MSigDB) Hallmark (31) and

210 Metacore transcription factor gene sets (GeneGo; Fig. 2; supplementary Table 2). Genes in clusters

211 three and five exhibited potentially interesting model and treatment-dependent patterns, although we

212 were unable to find any functionally meaningful gene sets enriched in these two clusters.

213 The expression of cluster one genes was reduced with treatment across all PDX models (Fig. 2). Multiple

214 gene sets were enriched in cluster one, including NF-kB signaling, cell cycle and proliferation, and the

215 activation of several transcription factors downstream of MAPK signaling (32). In addition, cluster one

216 contained known ERK target genes including DUSP4, DUSP6, SPRY4, and CCND1 (32) (supplementary

217 Table 3). The expression pattern, in addition to gene set enrichment results, indicated that these genes

218 were downstream of MAPK signaling.

219 In contrast, the expression of cluster two genes increased with BRAF and MEK inhibition. Cluster two was

220 enriched for the MITF activation gene set, which contained known MITF target genes such as MLANA,

221 TYRP1, and BCL2A1 (12) (Fig. 2; supplementary Table 3). In addition, epithelial markers such as CDH1,

Page 8 of 21

222 SNAI2, and ZEB2 were also upregulated. This is consistent with the enrichment of a differentiated

223 melanocytic transcriptional state during the early phase of drug treatment. Although this cluster was

224 associated with MITF activity, many of the activated genes were not found in the in vitro derived gene

225 signature describing the high MITF, proliferative melanoma phenotype sensitive to MAPK inhibition

226 (15,33) (supplementary Fig. 3A). The upregulation of MITF transcriptional programming was described as

227 a mechanism of drug tolerance, which protects melanoma against MAPK inhibitor treatment (17,34).

228 Based on known biology of MITF and the expression pattern of cluster two, these genes appeared to be

229 linked to adaptive response to MAPK inhibition in melanoma.

230 Genes in cluster four were highly expressed in CR PDX models at baseline, and downregulated with

231 treatment. Gene set enrichment identified genes involved in epithelial-mesenchymal transition, NF-kB

232 signaling, and the activation of multiple transcription factors known to be downstream of ERK (32).

233 Although AXL was identified in this gene set, there was very little overlap between cluster four and the in

234 vitro derived gene signature for the invasive melanoma phenotype associated with drug resistance

235 (15,33) (supplementary Fig. 3B). Consistent with gene set enrichment results, cluster four contained

236 genes representing a more ‘mesenchymal-like’ phenotype, in addition to ERK target genes such as

237 FOSL1 (FRA1), EGR2, and DUSP5 (32). These results suggested that the genes in cluster four defined a

238 MAPK-dependent, ‘mesenchymal-like’ phenotype associated with sensitivity to BRAF and MEK inhibition.

239 Taken together, hierarchical clustering revealed distinct patterns of gene expression that were dependent

240 on MAPK inhibitor treatment, drug sensitivity, and baseline model variability. Unsupervised clustering

241 using the top 1000 most variable DEGs demonstrated that there are three major sets of genes related to

242 canonical MAPK signaling, drug tolerance, and drug sensitivity respectively. Whereas a MITF-dependent,

243 differentiated melanocyte transcription program was induced upon drug treatment, the expression of a

244 subset of MAPK-dependent genes featuring ‘mesenchymal-like’ markers was enriched in melanoma

245 PDXs that achieve complete response upon BRAF and MEK inhibition.

246 Identification of transcriptional programs associated with sensitivity and resistance to BRAF and

247 MEK inhibitors in PDXs

Page 9 of 21

248 Using RNA-seq data from untreated tumors, we asked whether baseline expression of the identified

249 clusters from Fig. 2 was correlated with response to encorafenib and binimetinib across all 13 BRAFV600

250 mutant melanoma PDXs (18). Genes associated with the MITF-high drug tolerance phenotype were more

251 highly expressed in PR/SD models relative to CR models at baseline (Fig. 3A). The calculated signature

252 scores were statistically significantly different between the two groups (Fig. 3B). On the other hand, the

253 CR models expressed elevated levels of drug sensitivity genes in comparison to the PR/SD models (Fig.

254 3C, D). There were no significant differences in the expression of canonical MAPK signaling genes

255 between CR and PR/SD PDXs (supplementary Fig. 4A; Fig. 3E).

256 To validate our findings, we assessed relevant markers and pathways at the protein level (Fig. 3F).

257 Several post-translational and transcriptional targets of ERK were highly expressed and/or

258 phosphorylated in CR models. These include FRA1 (FOSL1), fosB, and EGR1 (32), consistent with the

259 enrichment of relevant gene sets in the drug sensitivity cluster (Fig. 2). Furthermore, the complete

260 responders exhibited increased mesenchymal protein markers, whereas the epithelial markers were

261 highly expressed at the protein level in the less sensitive PDX models (Fig. 3F). The AKT-mTOR

262 pathway was also profiled given its known role in BRAFV600 mutant melanoma drug response (35,36),

263 though a clear trend was not observed.

264 We showed that in BRAFV600 mutant melanoma PDXs, the MITF-high, ‘epithelial-like’ phenotype is not

265 only associated with adaptive response, but also reduced sensitivity to MAPK inhibition. Inversely, the

266 expression of a MAPK-high, ‘mesenchymal-like’ transcriptional program is linked to sensitivity to BRAF

267 and MEK inhibitors. This finding is distinct from the MITF high (proliferative) and low (invasive)

268 phenotypes described in melanoma cell lines, which highlights important differences between in vivo and

269 in vitro experimental systems.

270 Gene signatures derived from PDXs do not predict clinical outcome

271 Next, we evaluated the ability of our gene sets to predict clinical outcome using a published cohort of

272 seventeen patients (29). Using a combined signature score made up of both drug tolerance and

273 sensitivity genes, there were no differences in best response or progression free survival between

Page 10 of 21

274 patients predicted to be sensitive and vice versa (Fig. 4A, B). However, analysis using on-treatment

275 patient samples revealed that the expression of drug tolerance genes was uniformly increased during the

276 course of treatment (Fig. 4C). The expression of drug sensitivity genes during the course of treatment

277 underwent heterogeneous changes among patients (Fig. 4D). The induction of the MITF-high, ‘epithelial-

278 like’ phenotype in all patients suggested that selection towards a more drug resistant phenotype occurs

279 during the course of treatment, consistent with our findings from PDXs described herein. Although the

280 baseline expression of these genes was not associated with patient outcome, it could be due to

281 numerous factors such as intra-tumor heterogeneity, small sample size, heterogeneous treatment

282 regimen, or tumor microenvironment.

283 Tumor regression in response to MAPK inhibition is concomitant with downregulation of multiple

284 pathways

285 To further characterize gene expression changes associated with tumor regression in vivo, we

286 constructed linear models to identify DEGs post treatment only in the CR, but not PR/SD PDX models

287 (25) (supplementary Fig. 4B; Fig. 5A). More DEGs specific to the CR models were identified after nine

288 days of treatment in comparison to after a single dose. Gene set enrichment was performed using

289 MSigDB Hallmark gene sets (31). Genes preferentially downregulated in the CR models relative to the

290 PR/SD models were enriched for epithelial mesenchymal transition (EMT), cell cycle and proliferation,

291 and MAPK signaling (Fig. 5B). Western blot revealed that these pathways were also inhibited in models

292 that showed PR/SD in response to BRAF and MEK inhibition, but the changes were most significant in

293 the complete responder X-1906 (Fig. 5C). Consistent with earlier gene expression analyses, BRAF and

294 MEK inhibitor treatment resulted in a shift towards a more ‘epithelial-like’ phenotype, whereas two less

295 sensitive models already exhibited high level of E-cadherin protein expression prior to treatment.

296 Furthermore, the CR model X-1906 exhibited the highest level of BIM upregulation upon treatment,

297 consistent with apoptotic induction and the observed tumor regression (37). Models that showed stable

298 disease also upregulated BIM upon BRAF and MEK inhibitor treatment, indicating that this is not sufficient

299 to yield tumor regression. Taken together, encorafenib and binimetinib treatment in BRAFV600 mutant

300 melanoma PDXs resulted in inhibition of MAPK signaling, cell cycle and proliferation, and transition

Page 11 of 21

301 towards a more ‘epithelial-like’ cell state. Complete regression in response to treatment is associated with

302 strong inhibition of these pathways and upregulation of the pro-apoptotic protein BIM.

303 BRAFV600 mutant melanoma PDXs exhibit varied dependence on MAPK signaling

304 The gene expression and signaling studies suggested that MAPK pathway suppression was more robust

305 in the CR versus PR/SD models. To further extend this observation, MAPK pathway suppression was

306 measured in untreated and on-treatment tumor fragments (Fig. 1C) using ERK phosphorylation and

307 DUSP6 mRNA (Fig. 6A; supplementary Fig. 5). As an ERK phosphatase tightly regulated by MAPK

308 output, DUSP6 expression is highly sensitive to MAPK pathway inhibition (32). Unsurprisingly, the

309 complete responder X-1906 displayed markedly durable pathway suppression. In comparison, the PR/SD

310 models exhibited varied levels of MAPK inhibition. Three of the five PDX models displayed either

311 sustained MAPK signaling or signaling recovery after nine days of treatment, whereas the remaining two

312 PDXs showed durable MAPK pathway suppression. This finding was consistent with on-treatment patient

313 samples (29), where MAPK inhibition as measured by DUSP6 mRNA was greater in responding tumors

314 but more variable in patients with less tumor shrinkage (Fig. 6B). In addition, AKT phosphorylation

315 exhibited variable changes upon treatment, highlighting inter-tumor heterogeneity in adaptive response to

316 MAPK inhibition (supplementary Fig. 5A). These findings indicated that BRAFV600 mutant melanoma

317 elicited both MAPK dependent and independent adaptive mechanisms to survive under MAPK inhibitor

318 therapy.

319 To evaluate the hypothesis that BRAFV600 mutant melanoma vary in their dependence on the MAPK

320 pathway for survival, we assessed the efficacy of triple combination with encorafenib, binimetinib, and the

321 ERK inhibitor Vx-11e (38). The three-drug combination was well-tolerated in vivo (supplementary Fig. 6).

322 We evaluated the triple combination in two PDX models that exhibited either MAPK signaling rebound (X-

323 20767) or continual pathway suppression (X-4849; Fig. 6A). In the MAPK activated model, combination

324 with the ERK inhibitor resulted in further tumor regression in comparison to BRAF and MEK inhibitors

325 alone (Fig. 6C, D). The ERK inhibitor Vx-11e stabilizes the phosphorylated form of ERK, therefore MAPK

326 pathway inhibition was assessed by the phosphorylation of ERK substrate RSK3 (32). Tumor samples

327 collected after nine days of treatment showed that the triple combination resulted in increased

Page 12 of 21

328 suppression of MAPK signaling, in addition to the upregulation of BIM (Fig. 6D). In contrast, triple

329 combination with ERK inhibitor did not improve antitumor activity in X-4849, which did not demonstrate

330 MAPK reactivation following BRAF and MEK inhibitor treatment (Fig. 6E). This was despite the triple

331 combination yielding deeper suppression of the MAPK pathway, as evidenced by reduced RSK3

332 phosphorylation relative to BRAF and MEK inhibitor double combination (Fig. 6F). Consistent with the

333 lack of tumor regression, we did not observe any BIM upregulation upon treatment. To identify additional

334 combinations that may be efficacious for MAPK inactivated PDX models, we assessed efficacy results

335 from the Novartis mouse clinical trial (18). For model X-4849, combination of encorafenib and the PI3K

336 inhibitor burparlisib resulted in markedly more tumor regression in contrast to treatment using encorafenib

337 and binimetinib (Fig. 6G). This suggests that targeting parallel pathways rather than vertical combination

338 could improve efficacy for a subset of BRAFV600 mutant melanoma that do not reactivate MAPK pathway

339 during treatment.

340 As shown, these data support the notion that some BRAFV600 mutant melanoma do not depend on

341 reactivation of the MAPK pathway to survive under BRAF and MEK inhibitor therapy. Inhibition of the

342 MAPK pathway could only achieve limited efficacy in these tumors, and additional work is required to

343 identify treatment strategies with curative benefit. On the other hand, triple combination with BRAF, MEK,

344 and ERK inhibitors presents as a viable treatment strategy to improve response in BRAFV600 mutant

345 melanoma that depend on the MAPK pathway for survival, where reactivation of the pathway occurs as

346 early as nine days following treatment in vivo.

347

348 Discussion

349 Here, we demonstrated that distinct transcriptional programs are associated with differential sensitivity to

350 MAPK inhibition in BRAFV600 mutant melanoma PDXs. Melanoma PDXs with MITF-high, ‘epithelial-like’

351 gene expression were less sensitive to BRAF and MEK inhibition. PDX models with increased levels of

352 MAPK signaling and express a ‘mesenchymal-like’ transcriptional program were preferentially sensitive to

353 MAPK inhibition. In addition, the lack of tumor regression in response to BRAF and MEK inhibition was

Page 13 of 21

354 associated with adaptive responses consisting of MAPK reactivation or alternative survival mechanisms,

355 which in turn dictated their response to further suppression of the MAPK pathway. This study highlighted

356 the inter-tumor heterogeneity in BRAFV600 mutant melanoma, and the relevant challenges in optimizing

357 therapeutic strategies for individual tumors.

358 In this study, we observed that elevated MITF activity and the ‘epithelial-like’ transcriptional program was

359 associated with therapeutic resistance rather than sensitivity in vivo. This finding contradicts in vitro

360 studies, where the ‘MITF-low’ invasive cell state is linked to intrinsic resistance to BRAF inhibitors and

361 vice versa (12,13). The reasons for such in vitro-in vivo discrepancies remain unclear. Although tumor

362 samples from melanoma patients could be classified into proliferative and invasive subtypes (39), there

363 were no obvious correlations between melanocyte lineage transcriptional programming with clinical

364 response (40). Intra-tumor heterogeneity could affect biological interpretations in vivo. Although tumor

365 samples from melanoma patients could be classified into proliferative and invasive subtypes (39), there

366 were no obvious correlations between melanocyte lineage transcriptional programming with clinical

367 response (40). Melanoma cells switch between proliferative and invasive phenotypes in vivo (41), and

368 these two cell types co-exist in patient tumors (42). Moreover, the transcriptional and post-translational

369 regulation of MITF is highly complex and can be modulated by the in vivo tumor microenvironment (17).

370 MITF plays pleiotropic roles under diverse cellular contexts (17), sometimes demonstrating opposing

371 effects on melanoma resistance to targeted therapy (13,43). In addition, downregulation of AXL

372 expression was identified in BRAF inhibitor resistant patient tumors (44). It is conceivable that both

373 extremely high and low levels of MITF could protect against the cytotoxic effects exerted by MAPK

374 pathway inhibition.

375 Low levels of MITF and increased ‘EMT’-like transcription signature in the CR PDX models could also be

376 a direct consequence of MAPK pathway output. ERK and its substrate p90RSK1 post-translationally

377 phosphorylate MITF (45). Increased MITF expression and activity occurs shortly after BRAF inhibition,

378 which induces a drug tolerant state preceding disease progression (34). Although EMT has long been

379 implicated in therapeutic resistance across a variety of cancer types and therapeutic modalities (46),

380 melanocytes originate from the neural crest and therefore do not undergo true ‘EMT’. Oncogenic BRAF

Page 14 of 21

381 was shown to modulate NF-ƙB activity (47) and elicit EMT transcription factor reprogramming in

382 melanoma (48). In an in vivo murine melanoma model, treatment with MEK inhibitor resulted in transition

383 towards a more epithelial phenotype (49). Furthermore, the induction of EMT by the MAPK pathway and

384 its effector FRA1 was described in several epithelial cancers (50-52). In support of this interpretation, we

385 observed that elevated baseline MAPK signaling, in addition to high levels of FRA1 expression and

386 phosphorylation, occurred in association with elevated transcriptional programs such as EMT and NF-ƙB.

387 BRAF and MEK inhibitor treatment resulted in a shift towards more ‘epithelial’ melanocytic program in

388 PDXs, which could either be the consequence of MAPK pathway inhibition or selection towards more

389 resistant populations. In patient samples, the increased expression of melanocytic antigens in response

390 to MAPK inhibition is concomitant with elevated CD8+ T cell infiltration (53). Immune evasion is frequently

391 associated with acquired resistance to BRAF inhibitor-based therapy, as immune modulation plays a

392 significant part in BRAF inhibitor-mediated efficacy in vivo (44). The potential synergy and robust single

393 agent activity make the combination of targeted therapy with immune checkpoint inhibitors an incredibly

394 promising therapeutic strategy for BRAFV600 mutant melanoma (54). To this end, a major limitation of

395 PDXs is the inability to model the tumor immune microenvironment. Current efforts in developing

396 humanized and syngeneic mouse models will help address this gap in recapitulating human tumor

397 biology.

398 Interestingly, analysis using a curated ten-gene MAPK signature demonstrated that high MAPK pathway

399 activity was correlated with improved response to BRAF inhibition in BRAFV600 mutant melanoma patients

400 (55). This is consistent with our finding that PDXs with elevated MAPK pathway activity responded better

401 to BRAF and MEK inhibitors. However, we were unable to demonstrate the predictive value of our gene

402 sets using a publicly available clinical dataset. This could be due to intrinsic variability in small clinical

403 datasets, or additional factors such as the immune system or disease severity that were not captured by

404 the PDX system. Larger preclinical and clinical data sets will be required to further elucidate the links

405 between basal MAPK pathway output, MITF transcriptional activity, and clinical response to BRAF and

406 MEK inhibitors.

Page 15 of 21

407 Analysis of on-treatment PDX tumors revealed that response to BRAF and MEK inhibition involved

408 significant downregulation of cell cycle and proliferation genes. This is expected as the gene expression

409 changes corresponded to sustained inhibition of MAPK signaling. However, robust MAPK inhibition was

410 also observed in a subset of PR/SD PDX models, indicating co-dependence on alternative signaling

411 pathways and necessity to target these parallel pathways to achieve tumor regression. PTEN deficiency

412 has long been associated with resistance against MAPK inhibitors in BRAFV600 mutant melanoma

413 (35,36,56), although PTEN loss-of-function alone was insufficient to predict MAPK dependency in the

414 PDX models. On the other hand, three drug combination of BRAF, MEK, and ERK inhibitors achieved

415 tumor regression in PDX models that reactivated the MAPK pathway in response to BRAF and MEK

416 inhibitors alone. It was shown that the triple vertical combination could overcome the emergence of

417 MAPK-reactivated therapeutic resistance (57). Our data suggests that early adaptive response to MAPK

418 inhibition may be an important predictor of response to triple combination of BRAF, MEK, and ERK

419 inhibitors. To this end, profiling of early on-treatment biopsies may serve as a patient stratification

420 strategy for different BRAF and MEK inhibitor-based therapies. Ongoing clinical trials of BRAF and MEK

421 inhibitor-based triple combinations will help identify additional therapeutic options for this heterogeneous

422 and aggressive cancer.

423 In summary, we presented a detailed pharmacodynamic characterization of BRAFV600 mutant melanoma

424 PDXs treated with BRAF and MEK inhibitors. The in vivo model system revealed patterns of therapeutic

425 sensitivity distinct from in vitro studies, which underscores the necessity for diverse preclinical

426 experimental systems. Results from this study demonstrated the inter-tumor heterogeneity in metastatic

427 melanoma, and the need for improved precision medicine approach to ultimately cure this aggressive

428 disease.

429

430 Acknowledgements

431 We thank the PDX team, especially J. Green and A. Loo for establishing and maintaining the Novartis 432 PDX collection; L. Fan and O. Iartchouk for performing RNA sequencing; J. Korn and A. Johnson for 433 bioinformatic support; G. Caponigro, M. Niederst, V. Cooke, and J. Engelman for helpful discussions.

Page 16 of 21

434

435 References

436 1. Tsao H, Chin L, Garraway LA, Fisher DE. Melanoma: from mutations to medicine. Genes Dev 437 2012;26(11):1131-55. 438 2. Cancer Genome Atlas N. Genomic classification of cutaneous melanoma. Cell 2015;161(7):1681- 439 96. 440 3. Chapman PB, Hauschild A, Robert C, Haanen JB, Garbe C, Testori A, et al. Improved survival 441 with vemurafenib in melanoma with BRAF V600E mutation. N Engl J Med 2011;364(26):2507-16. 442 4. Dummer R, Ascierto PA, Gogas HJ, Arance A, Mandala M, Liszkay G, et al. Encorafenib plus 443 binimetinib versus vemurafenib or encorafenib in patients with BRAF -mutant melanoma 444 (COLUMBUS): a multicentre, open-label, randomised phase 3 trial. Lancet Oncol 2018;19(5):603- 445 15. 446 5. Larkin J, Ascierto PA, Dreno B, Atkinson V, Liszkay G, Maio M, et al. Combined vemurafenib and 447 cobimetinib in BRAF-mutated melanoma. N Engl J Med 2014;371(20):1867-76. 448 6. Robert C, Karaszewska B, Schachter J, Rutkowski P, Mackiewicz A, Stroiakovski D, et al. 449 Improved overall survival in melanoma with combined dabrafenib and trametinib. N Engl J Med 450 2015;372(1):30-9. 451 7. Shi H, Hugo W, Kong X, Hong A, Koya RC, Moriceau G, et al. Acquired resistance and clonal 452 evolution in melanoma during BRAF inhibitor therapy. Cancer Discov 2014;4(1):80-93. 453 8. Wagle N, Van Allen EM, Treacy DJ, Frederick DT, Cooper ZA, Taylor-Weiner A, et al. MAP 454 kinase pathway alterations in BRAF-mutant melanoma patients with acquired resistance to 455 combined RAF/MEK inhibition. Cancer Discov 2014;4(1):61-8. 456 9. Long GV, Fung C, Menzies AM, Pupo GM, Carlino MS, Hyman J, et al. Increased MAPK 457 reactivation in early resistance to dabrafenib/trametinib combination therapy of BRAF-mutant 458 metastatic melanoma. Nat Commun 2014;5:5694. 459 10. Long GV, Weber JS, Infante JR, Kim KB, Daud A, Gonzalez R, et al. Overall survival and durable 460 responses in patients with BRAF V600-mutant metastatic melanoma receiving dabrafenib 461 combined with trametinib. J Clin Oncol 2016;34(8):871-8 doi 10.1200/JCO.2015.62.9345. 462 11. Long GV, Flaherty KT, Stroyakovskiy D, Gogas H, Levchenko E, de Braud F, et al. Dabrafenib 463 plus trametinib versus dabrafenib monotherapy in patients with metastatic BRAF V600E/K-mutant 464 melanoma: long-term survival and safety analysis of a phase 3 study. Ann Oncol 465 2017;28(7):1631-9. 466 12. Konieczkowski DJ, Johannessen CM, Abudayyeh O, Kim JW, Cooper ZA, Piris A, et al. A 467 melanoma cell state distinction influences sensitivity to MAPK pathway inhibitors. Cancer Discov 468 2014;4(7):816-27. 469 13. Müller J, Krijgsman O, Tsoi J, Robert L, Hugo W, Song C, et al. Low MITF/AXL ratio predicts 470 early resistance to multiple targeted drugs in melanoma. Nat Commun 2014;5:5712. 471 14. Bittner M, Meltzer P, Chen Y, Jiang Y, Seftor E, Hendrix M, et al. Molecular classification of 472 cutaneous malignant melanoma by gene expression profiling. Nature 2000;406(6795):536-40. 473 15. Hoek KS, Schlegel NC, Brafford P, Sucker A, Ugurel S, Kumar R, et al. Metastatic potential of 474 melanomas defined by specific gene expression profiles with no BRAF signature. Pigment Cell 475 Res 2006;19(4):290-302. 476 16. Jeffs AR, Glover AC, Slobbe LJ, Wang L, He S, Hazlett JA, et al. A gene expression signature of 477 invasive potential in metastatic melanoma cells. PLoS One 2009;4(12):e8461. 478 17. Wellbrock C, Arozarena I. Microphthalmia-associated transcription factor in melanoma 479 development and MAP-kinase pathway targeted therapy. Pigment Cell Melanoma Res 480 2015;28(4):390-406. 481 18. Gao H, Korn JM, Ferretti S, Monahan JE, Wang Y, Singh M, et al. High-throughput screening 482 using patient-derived tumor xenografts to predict clinical trial drug response. Nat Med 483 2015;21(11):1318-25. 484 19. Bertotti A, Migliardi G, Galimi F, Sassi F, Torti D, Isella C, et al. A molecularly annotated platform 485 of patient-derived xenografts ("xenopatients") identifies HER2 as an effective therapeutic target in 486 cetuximab-resistant colorectal cancer. Cancer Discov 2011;1(6):508-23.

Page 17 of 21

487 20. Migliardi G, Sassi F, Torti D, Galimi F, Zanella ER, Buscarino M, et al. Inhibition of MEK and 488 PI3K/mTOR suppresses tumor growth but does not cause tumor regression in patient-derived 489 xenografts of RAS-mutant colorectal carcinomas. Clin Cancer Res 2012;18(9):2515-25. 490 21. Tentler JJ, Tan AC, Weekes CD, Jimeno A, Leong S, Pitts TM, et al. Patient-derived tumour 491 xenografts as models for oncology drug development. Nat Rev Clin Oncol 2012;9(6):338-50. 492 22. Hidalgo M, Amant F, Biankin AV, Budinska E, Byrne AT, Caldas C, et al. Patient-derived 493 xenograft models: an emerging platform for translational cancer research. Cancer Discov 494 2014;4(9):998-1013. 495 23. Cassidy JW, Caldas C, Bruna A. Maintaining tumor heterogeneity in patient-derived tumor 496 xenografts. Cancer Res 2015;75(15):2963-8. 497 24. Law CW, Chen Y, Shi W, Smyth GK. voom: precision weights unlock linear model analysis tools 498 for RNA-seq read counts. Genome Biol 2014;15(2):R29. 499 25. Ritchie ME, Phipson B, Wu D, Hu Y, Law CW, Shi W, et al. Limma powers differential expression 500 analyses for RNA-sequencing and microarray studies. Nucleic Acids Res 2015;43(7):e47 doi 501 10.1093/nar/gkv007. 502 26. Robinson M, Oshlack A. A scaling normalization method for differential expression analysis of 503 RNA-seq data. Genome Biol 2010;11(3):R25. 504 27. Charrad M, Ghazzali N, Boiteau V, Niknafs A. NbClust: an R package for determining the relevant 505 number of clusters in a data set. Journal of Statistical Software 2014;61(6):1-36. 506 28. Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL, Gillette MA, et al. Gene set 507 enrichment analysis: A knowledge-based approach for interpreting genome-wide expression 508 profiles. Proc Natl Acad Sci USA 2005;102(43):15545-50. 509 29. Kakavand H, Rawson RV, Pupo GM, Yang JYH, Menzies AM, Carlino MS, et al. PD-L1 510 Expression and Immune Escape in Melanoma Resistance to MAPK Inhibitors. Clin Cancer Res 511 2017;23(20):6054-61 doi 10.1158/1078-0432.CCR-16-1688. 512 30. Garman B, Anastopoulos IN, Krepler C, Brafford P, Sproesser K, Jiang Y, et al. Genetic and 513 genomic characterization of 462 melanoma patient-derived xenografts, tumor biopsies, and cell 514 lines. Cell Rep 2017;21(7):1936-52. 515 31. Liberzon A, Birger C, Thorvaldsdottir H, Ghandi M, Mesirov JP, Tamayo P. The Molecular 516 Signatures Database (MSigDB) hallmark gene set collection. Cell Syst 2015;1(6):417-25. 517 32. Yoon S, Seger R. The extracellular signal-regulated kinase: multiple substrates regulate diverse 518 cellular functions. Growth Factors 2006;24(1):21-44. 519 33. Widmer DS, Cheng PF, Eichhoff OM, Belloni BC, Zipser MC, Schlegel NC, et al. Systematic 520 classification of melanoma cells by phenotype-specific gene expression mapping. Pigment Cell 521 Melanoma Res 2012;25(3):343-53. 522 34. Smith MP, Brunton H, Rowling EJ, Ferguson J, Arozarena I, Miskolczi Z, et al. Inhibiting drivers of 523 non-mutational drug tolerance is a salvage strategy for targeted melanoma therapy. Cancer Cell 524 2016;29(3):270-84. 525 35. Paraiso KH, Xiang Y, Rebecca VW, Abel EV, Chen YA, Munko AC, et al. PTEN loss confers 526 BRAF inhibitor resistance to melanoma cells through the suppression of BIM expression. Cancer 527 Res 2011;71(7):2750-60. 528 36. Xing F, Persaud Y, Pratilas CA, Taylor BS, Janakiraman M, She QB, et al. Concurrent loss of the 529 PTEN and RB1 tumor suppressors attenuates RAF dependence in melanomas harboring 530 (V600E)BRAF. Oncogene 2012;31(4):446-57. 531 37. O'Connor L, Strasser A, O'Reilly LA, Hausmann G, Adams JM, Cory S, et al. Bim: a novel 532 member of the Bcl-2 family that promotes apoptosis. EMBO J 1998;17(2):384-95. 533 38. Aronov AM, Tang Q, Martinez-Botella G, Bemis GW, Cao J, Chen G, et al. Structure-guided 534 design of potent and selective pyrimidylpyrrole inhibitors of extracellular signal-regulated kinase 535 (ERK) using conformational control. J Med Chem 2009;52(20):6362-8. 536 39. Verfaillie A, Imrichova H, Atak ZK, Dewaele M, Rambow F, Hulselmans G, et al. Decoding the 537 regulatory landscape of melanoma reveals TEADS as regulators of the invasive cell state. Nat 538 Commun 2015;6:6683. 539 40. Wongchenko MJ, McArthur GA, Dreno B, Larkin J, Ascierto PA, Sosman J, et al. Gene 540 expression profiling in BRAF-mutated melanoma reveals Patient subgroups with poor outcomes 541 to vemurafenib that may be overcome by cobimetinib plus vemurafenib. Clin Cancer Res 542 2017;23(17):5238-45.

Page 18 of 21

543 41. Hoek KS, Eichhoff OM, Schlegel NC, Dobbeling U, Kobert N, Schaerer L, et al. In vivo switching 544 of human melanoma cells between proliferative and invasive states. Cancer Res 2008;68(3):650- 545 6. 546 42. Tirosh I, Izar B, Prakadan SM, Wadsworth MH, II, Treacy D, Trombetta JJ, et al. Dissecting the 547 multicellular ecosystem of metastatic melanoma by single-cell RNA-seq. Science 548 2016;352(6282):189-96. 549 43. Van Allen EM, Wagle N, Sucker A, Treacy DJ, Johannessen CM, Goetz EM, et al. The genetic 550 landscape of clinical resistance to RAF inhibition in metastatic melanoma. Cancer Discov 551 2014;4(1):94-109. 552 44. Hugo W, Shi H, Sun L, Piva M, Song C, Kong X, et al. Non-genomic and immune evolution of 553 melanoma acquiring MAPKi resistance. Cell 2015;162(6):1271-85. 554 45. Wu M, Hemesath TJ, Takemoto CM, Horstmann MA, Wells AG, Price ER, et al. c-Kit triggers dual 555 phosphorylations, which couple activation and degradation of the essential melanocyte factor 556 Mitf. Genes Dev 2000;14:301-12. 557 46. Tsukasa S, Weinberg RA. EMT, CSCs, and drug resistance: the mechanistic link and clinical 558 implications. Nat Rev Clin Oncol 2017;14(10):611. 559 47. Liu J, Suresh Kumar KG, Yu D, Molton SA, McMahon M, Herlyn M, et al. Oncogenic BRAF 560 regulates beta-Trcp expression and NF-kappaB activity in human melanoma cells. Oncogene 561 2007;26(13):1954-8. 562 48. Caramel J, Papadogeorgakis E, Hill L, Browne GJ, Richard G, Wierinckx A, et al. A switch in the 563 expression of embryonic EMT-inducers drives the development of malignant melanoma. Cancer 564 Cell 2013;24(4):466-80. 565 49. Brighton HE, Angus SP, Bo T, Roques J, Tagliatela AC, Darr DB, et al. New Mechanisms of 566 Resistance to MEK Inhibitors in Melanoma Revealed by Intravital Imaging. Cancer Res 567 2018;78(2):542-57. 568 50. Lemieux E, Bergeron S, Durand V, Asselin C, Saucier C, Rivard N. Constitutively active MEK1 is 569 sufficient to induce epithelial-to-mesenchymal transition in intestinal epithelial cells and to 570 promote tumor invasion and metastasis. Int J Cancer 2009;125(7):1575-86. 571 51. Diesch J, Sanij E, Gilan O, Love C, Tran H, Fleming NI, et al. Widespread FRA1-dependent 572 control of mesenchymal transdifferentiation programs in colorectal cancer cells. PLoS One 573 2014;9(3):e88950. 574 52. Bakiri L, Macho-Maschler S, Custic I, Niemiec J, Guio-Carrion A, Hasenfuss SC, et al. Fra-1/AP-1 575 induces EMT in mammary epithelial cells by modulating Zeb1/2 and TGFbeta expression. Cell 576 Death Differ 2015;22(2):336-50. 577 53. Frederick DT, Piris A, Cogan AP, Cooper ZA, Lezcano C, Ferrone CR, et al. BRAF inhibition is 578 associated with enhanced melanoma antigen expression and a more favorable tumor 579 microenvironment in patients with metastatic melanoma. Clin Cancer Res 2013;19(5):1225-31. 580 54. Griffin M, Scotto D, Josephs DH, Mele S, Crescioli S, Bax HJ, et al. BRAF inhibitors: resistance 581 and the promise of combination treatments for melanoma. Oncotarget 2017;8(44):78174-92. 582 55. Wagle M-C, Kirouac D, Klijn C, Liu B, Mahajan S, Junttila M, et al. A transcriptional MAPK 583 Pathway Activity Score (MPAS) is a clinically relevant biomarker in multiple cancer types. NPJ 584 Precis Oncol 2018;2(1):7. 585 56. Nathanson KL, Martin AM, Wubbenhorst B, Greshock J, Letrero R, D'Andrea K, et al. Tumor 586 genetic analyses of patients with metastatic melanoma treated with the BRAF inhibitor dabrafenib 587 (GSK2118436). Clin Cancer Res 2013;19(17):4868-78. 588 57. Xue Y, Martelotto L, Baslan T, Vides A, Solomon M, Mai TT, et al. An approach to suppress the 589 evolution of resistance in BRAF(V600E)-mutant cancer. Nat Med 2017;23(8):929-37.

590

591

592 Figure legends

593 Figure 1. Description and transcriptome analysis of BRAFV600 mutant melanoma PDXs.

Page 19 of 21

594 (A) Genetic features of BRAFV600 mutant melanoma models in the Novartis PDX collection, arranged in

595 order of their best average response to encorafenib and binimetinib as published in Gao et al. (18). (B)

596 Representative efficacy studies showing examples of complete response (X-2613) and stable disease (X-

597 20767), upon continuous treatment with encorafenib and binimetinib for 28 days. (C) Schematic of

598 sample collection from mice treated with encorafenib and binimetinib. Samples were harvested from

599 tumor bearing mice (n=3) at baseline, eight hours post the first dose, and eight hours post the last dose

600 after nine days of continuous dosing. (D) Venn diagrams showing genes that were differentially

601 expressed after a single dose or nine days of treatment, relative to the untreated controls for each PDX

602 model. The cut off for differential gene expression analysis was log2 fold change > |1.5| and p-value <

603 1.05.

604 Figure 2. Unsupervised clustering of pre- and on-treatment melanoma PDX models.

605 Heat map displays unsupervised clustering results using 1000 most variable DEGs across all treatment-

606 matched samples. Left panel shows gene set enrichment results. Bottom panel shows the summary

607 scores for the clusters, calculated as the sum of z-transformed, normalized expression values.

608 Figure 3. Distinct transcriptional programs are associated with sensitivity to MAPK inhibition in

609 melanoma in vivo.

610 Heat maps showing baseline expression of genes belonging to the drug tolerance (A) and drug sensitivity

611 (C) clusters in BRAFV600 mutant melanoma models from the Novartis PDX collection, arranged in order of

612 previously published best average response to encorafenib and binimetinib (18). Expression of drug

613 tolerance (B), drug sensitivity (D), and canonical MAPK (E) gene clusters were quantified as signature

614 scores, and compared between models that achieved either CR or PR/SD in response to drug treatment.

615 Signature score for each PDX model was calculated as the sum of normalized, z-transformed gene

616 expression values. Statistical analyses were performed using Student’s t-test (***p<0.001; n.s. = not

617 significant). (F) Western blot analysis of relevant protein markers in tumor fragments collected from

618 untreated mice.

619 Figure 4. Gene signatures derived from PDXs do not predict drug response in the clinic.

Page 20 of 21

620 Signature scores were calculated using a dataset of 17 patients (26) by combining the drug tolerance and

621 sensitivity genes, where drug tolerance genes were assigned negative values and drug sensitivity genes

622 were given positive values. There were no significant differences in best response measured by RECIST

623 (A; Student’s t-test; n.s. = not significant), or progression free survival (B; Cox Proportional-Hazards)

624 between patients that were predicted to be sensitive or insensitive based on overall median of the

625 signature scores. Drug tolerance (C) and drug sensitivity (D) signature scores are shown in pre- and on-

626 treatment matched tumor samples.

627 Figure 5. Tumor regression in response to BRAF and MEK inhibition is concomitant with the

628 downregulation of cell cycle and proliferation, MAPK signaling, and EMT pathways in vivo.

629 (A) Volcano plot shows differentially expressed genes in complete response, but not stable disease PDX

630 models after nine days of treatment relative to untreated tumors. (B) Gene set enrichment of the complete

631 response models only, differentially expressed genes using MSigDB Hallmark gene sets. (C) Western blot

632 analysis of relevant protein markers from gene set enrichment analyses, using untreated and day nine

633 treatment tumor samples in triplicates.

634 Figure 6. A subset of PR/SD models do not depend on the MAPK pathway for survival.

635 (A) Quantified phosphorylated ERK protein level in tumor samples harvested pre-treatment, 8 hours after

636 a single dose, or 8 hours after the final dose on the ninth day of continuous treatment. Based on ERK

637 phosphorylation levels, the PR/SD models were classified into MAPK activated or inactivated groups. (B)

638 The degree of MAPK pathway suppression in patient samples as measured by DUSP6 log2 fold change,

639 relative to patient response to BRAF inhibitors with or without MEK inhibitor. Efficacy results of the MAPK

640 activated model X-20767 (C) and the MAPK inactivated model X-4849 (E), treated with BRAF and MEK

641 inhibitors with or without ERK inhibitor, or ERK inhibitor alone. Statistical analysis was performed on the

642 end point measurements using Student’s t-test (****p<0.0001; n.s. = not significant). Western blot

643 analysis was performed on tumors harvested after nine days of treatment with drug combinations as

644 shown, in X-20767 (D) and X-4849 (F). (G) Best average response in X-4849 to various drug

645 combinations (18), extracted from supplementary tables in the reference publication.

Page 21 of 21

sensitive resistant A CR PR SD D X-1906 (CR)

BRAF Known COSMIC gain-of-function mutation late vs. untreated NRAS Amplification (>5 copies)

NF1 Heterozygous deletion PTEN 81 366 1176 Homozygous deletion PIK3CA Novel mutation early vs. untreated MITF Heterozygous truncation, frameshift, or known COSMIC loss-of-function mutation CDKN2A Homozygous truncation, frameshift, TP53 or known COSMIC loss-of-function mutation

X-1906X-3483X-4644X-2613X-3746X-3211X-4538X-2723X-4530X-2602X-3676X-4668X-4849 X-20767 (SD) X-4530 (PR)

late vs. untreated late vs. untreated B Complete response Stable disease X-2613 X-20767 1500 2000 11343 1216 559 341

) Untreated 3 BRAFi + MEKi (chow)

(m m 1500 early vs. untreated early vs. untreated 1000

olum e 1000

500 500 umo r v T X-4668 (SD) X-4849 (SD) 0 0 0 5 10 15 20 25 30 0 5 10 15 20 25 30 late vs. untreated late vs. untreated Days post first dose Days post first dose C 77 695 40 778 11 5 early vs. untreated BRAFi + MEKi BRAFi + MEKi early vs. untreated

untreated early treatment (single dose) late treatment (day 9)

Downloaded from mct.aacrjournals.org on September 27, 2021. © 2019 American Association for Cancer Research. Author Manuscript Published OnlineFirst on September 16, 2019; DOI: 10.1158/1535-7163.MCT-19-0028 Feng et al. FigAuthor 2 manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

Complete response Stable disease

X-1906 X-2613 X-20767 X-4530 X-4668 X-4849 Untreated Early treatment Late treatment Untreated Early treatment Untreated Early treatment Late treatment Untreated Early treatment Late treatment Untreated Early treatment Late treatment Untreated Early treatment Late treatment Cluster

TNFα signaling via NF-ƙB G2M checkpoint CREB1 activation SP1 activation E2F targets E2F activation 1 FOXM1 activation p53 inhibition Oct-3/4 activation TFIIB activation

02468101214

TFE3 activation MITF activation 2

02468101214

Epithelial mesenchymal transition TNFα signaling via NF-ƙB SP1 activation Hypoxia AP-2A activation c-Jun activation 4 SMAD3 activation IL2 STAT5 signaling p53 activation EGR1 activation KRAS signaling up

02468101214 -log10 adjusted p-value 5

Avg log2 TPM (scaled)

3 0 −3

Cluster summary score 5

-200 -100 0 100 200

A Drug tolerance genes (cluster 2) B C Drug sensitivity genes (cluster 4) sensitive resistant sensitive resistant Drug tolerance CR PR SD genes (cluster 2) CR PR SD

150 ***

100

-50 Signature score -100

-150

CR models PR/SDmodels X-1906 X-3483 X-4644 X-2613 X-3746 X-3211 X-4538 X-2723 X-4530 X-2602 X-3676 X-4668 X-4849 X-2613 X-3746 X-3211 X-4538 X-2723 X-4530 X-2602 X-3676 X-4668 X-4849 X-1906 X-3483 X-4644

Avg log2 TPM (scaled) Avg log2 TPM (scaled)

3 0 −3 3 0 −3 F D E sensitive resistant Drug sensitivity Canonical MAPK genes (cluster 4) genes (cluster 1) CR PR SD

n.s. 300 *** 100 X-1906 X-3483 X-4644 X-2613 X-3746 X-3211 X-4538 X-2723 X-4530 X-2602 X-3676 X-4668 X-4849 200 50 pERK core 100 tERK 0 0 pRSK3 tRSK -50 Signature s

-100 Signature score pFRA RAF-MEK-ERK -200 -100 tFRA

CR CR models PR/SDmodels models PR/SDmodels FosB

EGR1 pAKT tAKT pS6 PI3K-AKT-mTOR tS6

MITF EMT/MITF E-cadherin N-cadherin GAPDH

Downloaded from mct.aacrjournals.org on September 27, 2021. © 2019 American Association for Cancer Research. Author Manuscript Published OnlineFirst on September 16, 2019; DOI: 10.1158/1535-7163.MCT-19-0028 Feng et al. FigAuthor 4 manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

A GSE99898 B GSE99898 C Drug tolerance genes D Drug sensitivity genes (cluster 2) (cluster 4) 20 n.s. predicted sensitive 60 100 100 predicted insensitive 0 40 50 burden

(% ) 20 -20 IST umor 50 0 0 -40 in t -20 Signature score by REC -50 -60 -40 Progression free survival Change -80 0 -60 -100 0 5 10 15 20 25 Time (months) Pre- On- Pre- On- treatment treatment Predicted Predicted treatment treatment sensitive insensitive

Downloaded from mct.aacrjournals.org on September 27, 2021. © 2019 American Association for Cancer Research. Author Manuscript Published OnlineFirst on September 16, 2019; DOI: 10.1158/1535-7163.MCT-19-0028 Feng et al. FigAuthor 5 manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

A C CR SD

X-1906 X-20767 X-4668 X-4849 BRAFi + MEKi - - - + + + - - - + + + - - - + + + - - - + + + (day 9) pERK tERK

pRSK3 RAF-MEK-ERK tRSK pFRA

tFRA pAKT tAKT PI3K-AKT-mTOR pS6

tS6

B MITF Downregulated EMT/MITF Hallmark gene set SIZE ES NES NOM p-val FDR q-val FWER p-val E-cadherin G2M checkpoint 198 -0.748 -1.996 0 0 0 N-cadherin Epithelial mesenchymal transition 189 -0.726 -1.920 0 0 0 E2F targets 199 -0.726 -1.899 0 0 0 Cyclin D1 KRAS signaling (up) 164 -0.702 -1.829 0 0 0 Cell cycle/ TNFα signaling via NFƙB 189 -0.683 -1.807 0 0 0 pRb (S780) apoptosis Upregulated BIMEL Hallmark gene set SIZE ES NES NOM p-val FDR q-val FWER p-val GAPDH Interferon α response 93 0.636 1.662 0 0.009 0.018

Feng et al. Fig 6

MAPK activated MAPK inactivated A B DUSP6 logFC CR PR/SD RECIST 20 0

X-1906 X-20767 X-3676 X-4530 X-4668 X-4849 baseline) vs. (treatment 150 150 150 150 150 0 -1 logFC DUSP6

100 100 100 100 100 -20 -2

50 50 50 50 50 50

by RECIST (%) -40 -3 relativeuntreated to Change in tumor burden %ERKphosphorylation 0 0 0 0 0 0 -60 -4 0 0 0 0 0 0 24 12 24 12 24 12 24 12 24 12 24 12 7 2 0 1 7 6 3 204 216 192 192 216 192 204 216 192 204 216 192 216 192 204 216 204 204 1 1 1 1 1 Time post first drug adminstration (hour) Patient

C X-20767 - MAPK activated D BRAFi BRAFi + MEKi Untreated Untreated + MEKi + ERKi ERKi 2000 BRAFi + MEKi

) pERK 3 BRAFi + MEKi + ERKi tERK 1500 ERKi pRSK3

1000 tRSK pAKT 500 tAKT

Tumor volume (mm **** BIMEL 0 0 5 10 15 20 25 30 GAPDH Days post treatment

E X-4849 - MAPK inactivated F BRAFi G BRAFi + MEKi Gao et al., 2015 Untreated Untreated + MEKi + ERKi ERKi 20 BRAFi + MEKi 1500 BRAFi + MEKi pERK BRAFi + PI3Ki )

3 BRAFi + CDK4/6i BRAFi + MEKi + ERKi tERK 0 1000 ERKi pRSK3 tRSK -20 pAKT 500 -40 n.s. tAKT BIM Tumor volume (mm EL -60 0 GAPDH 0 5 10 15 20 25 30 -4849 Days post treatment X Downloaded from mct.aacrjournals.org on September 27, 2021. © 2019 American Association for Cancer Research. Author Manuscript Published OnlineFirst on September 16, 2019; DOI: 10.1158/1535-7163.MCT-19-0028 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

Distinct transcriptional programming drive response to MAPK inhibition in BRAFV600 mutant melanoma patient derived xenografts

Tianshu Feng, Javad Golji, Ailing Li, et al.

Mol Cancer Ther Published OnlineFirst September 16, 2019.

Updated version Access the most recent version of this article at: doi:10.1158/1535-7163.MCT-19-0028

Supplementary Access the most recent supplemental material at: Material http://mct.aacrjournals.org/content/suppl/2019/09/14/1535-7163.MCT-19-0028.DC1

Author Author manuscripts have been peer reviewed and accepted for publication but have not yet been Manuscript edited.

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://mct.aacrjournals.org/content/early/2019/09/14/1535-7163.MCT-19-0028. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.