Tumor Heterogeneity and Plasticity As Elusive Drivers for Resistance to MAPK Pathway Inhibition in Melanoma

REVIEW Tumor heterogeneity and plasticity as elusive drivers for resistance to MAPK pathway inhibition in melanoma

A Roesch

Despite the recent success of MAPK signaling-targeted drugs in melanoma, the majority of patients with metastatic melanoma still undergo disease progression after initial tumor shrinkage indicating gradually developing therapy resistance. This review will give an overview on currently suggested concepts of resistance to MAPK pathway inhibitors in melanoma with particular focus on inter- and intraindividual as well as intratumor heterogeneity. The high plasticity of melanoma cells promotes both the clonal evolution of genetic resistance, for example, because of mutations in the MAPK or PI3K/AKT/PTEN pathways, and the emergence of cell phenotypes that functionally and metabolically overcome MAPK inhibition. Like a ‘moving target’, melanoma cells are shifting between different metabolic, cell cycle and differentiation states reﬂecting a highly dynamic potential to adapt to exogenous stressors including drugs. The introduction of MAPK inhibitors into the clinics has tremendously pushed the ﬁeld of melanoma research not just because of the historic therapeutic success but also by providing a new tool to study human melanoma in its natural microenvironment.

Oncogene (2015) 34, 2951–2957; doi:10.1038/onc.2014.249; published online 11 August 2014

INTRODUCTION when applied to patients. Thus, the examination of drug Melanoma is an aggressive tumor causing the majority of skin resistance mechanisms increases our knowledge on melanoma cancer-related deaths in Caucasians with increasing incidence, at biology and provides a plethora of future therapeutic the moment 15–25 per 100 000 in Western countries. Unlike early innovations. In this respect, the first combinatorial therapies of stage melanomas that can be cured by surgical excision, advanced BRAF and MEK inhibitors are already being tested in clinical trials melanomas (AJCC 2009 stage IV) are associated with a median aiming to increase the number of cells being killed and to survival of 6–12 months.1,2 Until 2010, all major chemotherapeutic minimize the risk for development of resistance originating from and immunological approaches as well as radiotherapeutic surviving cells. A recent phase I/II trial suggests that the median interventions have failed to increase survival rates of stage IV progression-free survival in the combination group (9.4 months) could be improved by approximately 3.6 months as compared patients in randomized clinical trials. Recent phase III trials with 7 MAPK pathway-targeted drugs showed significant improvements with the monotherapy group (5.8 months). But also here, in progression-free survival and also overall survival when development of resistance has been observed with a delay of a compared with the former therapeutic standard dacarbazine.2 few months (see below). This review will give a brief overview on For example, mutant-specific BRAF kinase inhibitors like vemur- currently suggested concepts of resistance to MAPK pathway afenib or dabrafenib achieve ~ 50% objective response rate and inhibitors in melanoma with the particular focus on tumor an overall survival benefit of ~ 4 months in BRAFV600E-positive heterogeneity and plasticity. individuals (representing ~ 50% of all melanomas).3–5 The MEK1/2 inhibitor trametinib improved the median progression-free survival in BRAFV600E/K-positive individuals by ∼ 3 months when CURRENTLY DISCUSSED CATEGORIES OF RESISTANCE compared with dacarbazine. At 6 months of therapy, the rate of MECHANISMS IN MELANOMA overall survival was 81% in the trametinib group and 67% in the The number of suggested resistance mechanisms to MAPK dacarbazine group.6 pathway inhibition in melanoma is rapidly increasing. These Nevertheless, in the broad majority of patients treated with mechanisms can be subdivided into genomic vs phenotypic MAPK pathway inhibitors, metastases recur after a median mechanisms and according to their temporal occurrence into duration of ∼ 5–7 months indicating gradually developing intrinsic, adaptive and acquired resistance. Intrinsic resistance, for therapy resistance after initial (complete or partial) tumor example, to vemurafenib, according to RECIST criteria occurs in – shrinkage.3 6 Despite—or even because of—such developing ∼ 50% of all patients with 15% of patients showing no tumor resistance, those drugs represent valuable tools enabling shrinkage at all and 35% of patients achieving only limited tumor researchers to study melanoma and its evolutionary capability in reduction that does not fulfill the RECIST criteria for partial its natural environment. For example, melanoma cells that are response. The other 50% of patients initially show tumor shrinkage addicted to activated MAPK signaling are enforced to disclose 430% but subsequently undergo progressive disease with tumor their full genomic and phenotypic plasticity to survive the outgrowth because of acquired resistance.3,8–10 Adaptive resistance therapeutic selection pressure of BRAF or MEK inhibitors occurs already within hours after drug exposure and reflects the

University Hospital Essen, Department of Dermatology, Essen, Germany. Correspondence: Professor A Roesch, University Hospital Essen, Department of Dermatology, Hufelandstr. 55, D-45122 Essen, Germany. E-mail: [email protected] Received 29 April 2014; revised 22 May 2014; accepted 23 May 2014; published online 11 August 2014 Resistance to MAPK pathway inhibition in melanoma A Roesch 2952 dynamic re-adjustment of signaling pathways on the cellular level, other positive and negative regulators of this pathway (PI3K3CA, for example, ERK-controlled feedback loops regulating upstream PIK3CG, PIK3R2, PTEN and PHLPP1) were detected at an overall activators like RAS.11 However, all categorizations, which are based frequency of 22% of patients with BRAF inhibitor resistance. Of on clinical responses, have to be used with caution because they them, 18% also had overlapping alterations in the MAPK core do not fully reflect the resistance state of single melanoma cells. pathway.21 However, as is the case for MEK1 mutations, also PI3K For example, also, most responders still have remaining tumor pathway alterations do not necessarily preclude clinical response — — masses, which on the cellular level certainly display relative if detected in pre-treatment tumors. Although a recent tumor resistance compared to cells, which are being killed. Thus, as long genetic analysis suggested a correlative trend (P = 0.059) between as the technical resolution of our diagnostic and molecular PTEN loss/mutation status and progression-free survival in screening procedures in patients is too low to routinely assess the dabrafenib-treated patients,22 single patients with PTEN altera- single cell level, a comprehensive categorization of the temporal fi and spatial development of resistance is difficult. More clinical tions may still bene t from BRAFV600E inhibition. Very lately, a studies are highly needed to address this question. Besides, future patient with a pre-existing PTENK128T mutation was reported to therapeutic concepts in melanoma will increasingly involve have passed 18 weeks of stable disease before developing 15 immunotherapeutic approaches such as anti-PD-1 or anti-CTLA4 progressive disease. checkpoint inhibition, which most likely will lead to additional With regard to the first clinical trials on combinatorial BRAF plus changes in the definition of resistance. MEK inhibition, two groups have identified a mutation in MAP2K2/ MEK2 (Q60P) in patients who progressed under dabrafenib/ trametinib.23,24 Because melanoma cells that had been chronically THE HETEROGENEITY AND EVOLUTION OF GENOMIC exposed to dabrafenib/trametinib remained sensitive to either RESISTANCE ERK inhibition24 or PI3K/mTOR inhibition,23 hope is raised that Interindividual genomic heterogeneity future triple therapies could also overcome double resistance in In 2010, Nazarian et al.12 initially demonstrated that melanoma the clinics. cells can maintain high levels of ERK1/2 phosphorylation despite Next to the MAPK and PI3K-AKT core pathways, a number of the presence of the BRAF inhibitor vemurafenib because of the other genomic modifications have been identified till date expression of mutant NRASQ61K. In the presence of BRAF including NF1, RAC1, HOXD8 and prominent melanoma drivers inhibitors, mutated RAS promotes dimerization of BRAFV600E such as the master lineage transcription factor MITF, which was with other (wild-type) RAF isoforms, which restores strong found focally amplified in 2 of 45 patients.21 Subsequent in vitro 11 activation of the MAPK pathway. Somatic mutations in NRAS experiments demonstrated that enforced overexpression of MITF (Q61K/R/L, G12D/R and G13R) were detected till date by whole- in BRAFV600E mutant cells promotes resistance towards inhibitors exome sequencing in ∼ 8–18% of BRAF inhibitor-resistant patients; 15 13–15 of BRAF, MEK and ERK. This observation is supported by a recent in most cases, as a late event beyond 12 weeks of therapy. systematic gain-of-function study, which individually expressed This is interesting because in primary melanomas co-existence of 16 more than 15 500 ORFs of a lentiviral expression library in a BRAF and NRAS mutations occurs rather rarely. Also, KRAS BRAFV600E mutant melanoma cell line. This approach revealed a mutations (G12C, G12R, Q61H) have been detected in resistant cyclic-AMP-dependent melanocytic signaling network to be melanoma cell lines and in up to 7% of BRAF inhibitor-treated associated with resistance to MAPK pathway inhibition including patients, although KRAS mutations are far less common in primary 14,17 cAMP response element binding protein (CREB) and downstream melanomas than NRAS mutations. 25 Two other, very frequent mechanisms that overcome targets like MITF. BRAFV600E inhibition in melanoma occur on the level of BRAF itself, but interestingly do not act through secondary mutations or interference with drug binding sites. Instead, resistant cells either Intraindividual and intratumoral genomic heterogeneity amplify the number of BRAF copies 18 or express alternatively There is emerging evidence that tumors within the same patient spliced variants of BRAFV600E (e.g, exons 4–8 encoding the RAS- can simultaneously develop different molecular mechanisms of binding domain) and promote dimerization with other RAF resistance, for example, NRAS mutations (Q61R or Q61K) in one 11,19 isoforms irrespective of the RAS status. Both scenarios foster metastasis and BRAF amplifications or MEK1 mutations in a solid re-activation of ERK1/2, also in the absence of other genetic another.15,26 As reported by Shi et al.,21 20% of all patients (9 of events, such as NRAS mutations. Mutant BRAF amplifications can 44) from their study developed at least two different mechanisms be found in ∼ 8–19% and alternative splicing variants in 13–32% 13–15 of resistance, either in different progressive lesions or even in the of resistant melanomas. same lesion. However, this rate is probably a vast underestimate of fi A mutation in MAP2K1/MEK1 (C121S) has been rst described the true level of intraindividual heterogeneity. A sub-analysis of by targeted sequencing of a cancer gene panel in a tumor patients who donated more than one biopsy showed that actually obtained from a patient with melanoma who developed 81% of patients displayed multiple mechanisms of resistance. To resistance after an initial dramatic response to vemurafenib.20 In investigate temporal and spatial heterogeneity, Shi et al.21 general, MAP2K1 mutations (e.g, K57N, C121S) can be detected in 13–15 performed whole-exome sequencing from nine different cuta- ∼ 3–16% of BRAF inhibitor-resistant patients; however, their significance for single BRAF inhibitor treatment is not clear yet, neous metastases of a single patient spanning 726 days of because their presence does not universally preclude clinical dabrafenib treatment. They found a broad spectrum of different response.15,21 On the basis of experiments with a doxycycline- genomic alterations across the nine biopsies in both core inducible expression system, Van Allen et al.15 concluded that resistance pathways including alternative splicing of BRAFV600E, overexpression of mutant MEK1 can confer resistance to RAF/MEK BRAF amplification, KRASG12C mutation and PTEN deletion. More inhibition in vitro and that a dynamic regulation of mutant MEK1 strikingly, they could compute a phylogenetic tree of clonal may modulate drug sensitivity. Altogether, genomic alterations of evolution based on the common occurrence or absence of single the MAPK pathway occur in up to 70% of all patients with BRAF nucleotide variants. All resistant clones originated from a common inhibitor resistance.14 ancestral node and followed a branched (non-linear) evolution As second core resistance pathway, PI3K-PTEN-AKT signaling pattern that was also not collinear with the timing of clinical was unraveled. AKT1/3 mutations (Q79K, E17K) and mutations in emergence.

Oncogene (2015) 2951 – 2957 © 2015 Macmillan Publishers Limited Resistance to MAPK pathway inhibition in melanoma A Roesch 2953 THE HETEROGENEITY AND PLASTICITY OF PHENOTYPIC activation of mitochondrial biogenesis and OXPHOS across RESISTANCE different tissue and cancer types, including brown and white 38,39 Epigenetic restoration of MAPK signaling and functional activation adipocytes as well as breast and colon cancer. Further of PI3K signaling despite presence of BRAF inhibitors analyses of gene expression signatures of diverse cancer types (lung, colon, breast, melanoma) revealed that PGC1α is exclusively Next to mutational pathway re-activation, MAPK restoration on fi 36 the level of RAF can also occur through increased expression and linked to melanocytic lineage-speci c regulation pathways. 27,28 Congruently, the master transcription factor of melanocytic subsequent dimerization of CRAF. Downstream and indepen- α dent of RAF, expression of the MAPK kinase kinase COT was differentiation, MITF, acts as a direct inducer of PGC1 as identified in a functional genomics approach as an additional demonstrated by promoter assays and chromatin immunopreci- mechanism of ERK reactivation.27 Further downstream and also pitation. MITF regulates key differentiation pathways in melano- independent of ERK1/2, repression of two important apoptotic cytes, for example, by controlling melanin producing enzymes such as DCT, TYRP1 or TYR, and is genomically amplified in up to BH3-only genes, Bim-EL and Bmf, has been observed in the 40 α context of resistance.29 As Bim-EL is silenced epigenetically, 30% of melanomas. Both factors, MITF and PGC1 , seem to be the authors tested the HDAC inhibitor vorinostat and found a part of a bioenergetic adaption program, which secures ATP re-sensitization of resistant melanoma cells to vemurafenib. production through OXPHOS in response to BRAFi treatment. Overexpression of MITF in BRAFV600E-positive melanoma cells The expression of receptor tyrosine kinases (RTKs) like EGFR, 36 β reduced the drop in ATP supply after BRAFi treatment. Forced PDGFR or IGF-1 R has been shown by several studies to confer α resistance of melanoma cells to vemurafenib in a MAPK- expression of PGC1 protected cells from BRAFi and the 12,30,31 combination of BRAFi plus inhibitors of the OXPHOS, like 2,4- independent way. This constellation is highly reminiscent fi of observations from BRAFV600E-positive colon cancer, in which DNP or oligomycin A, resulted in signi cant reduction of tumor enhanced EGFR signaling also can confer resistance.32,33 Regard- growth in murine melanoma xenografts. ing the mechanisms behind RTK regulation, SOX10 has been The inducible nature of MITF-mediated metabolic resistance is strongly reminiscent of the phenotypic switch model suggested identified as an epigenetic repressor of EGFR by using a screening 41–44 approach with 661 shRNAs targeting different chromatin previously for melanoma. According to this concept, regulators.30 Subsequent transcriptome profiling of SOX10- malignant progression does not necessarily require clonal knocked down melanoma cells unraveled the TGF-β pathway as evolution but can also be mediated by reversible reprogramming crucial signaling hub including the downstream targets SMAD2/3 of signaling networks. By global cDNA microarray analysis and hierarchic clustering of more than 80 cultured melanoma cell lines, and JUN. Accordingly, incubation of melanoma cells with fi recombinant TGF-β resulted in EGFR and PDGFRβ upregulation two gene signatures have been identi ed, which determine opposing differentiation phenotypes: a MITF-driven, melanocytic followed by vemurafenib resistance. Previously, ERBB3 (HER3) has ‘ ’ ‘ ’ β been identified as an adaptively regulated downstream target of ( differentiated ) phenotype and a de-differentiated , TGF- -driven phenotype with stem-like properties. This further suggests that the stemness factor FOXD3 in BRAF inhibitor-resistant melanoma β fl cells by combining chromatin immunoprecipitation sequencing the MITF-driven and the TGF- -driven cell population may re ect with microarray-based gene profiling.34 ERBB3 showed enhanced endpoints of a gradient of subpopulation identities. According to the ‘MITF rheostat model’ developed by Goding’s group, cells signaling through PI3K/AKT in response to vemurafenib treatment low βhigh and knockdown of FOXD3 re-sensitized intrinsically resistant switch from a quiescent stem-like phenotype (MITF /TGF- ) 34 to a proliferative phenotype (intermediate MITF expression) and melanoma cells to vemurafenib-induced apoptosis. Because then to a differentiated, again cell cycle-arrested phenotype co-treatment of melanoma cells with the EGFR inhibitor gefitinib (MITFhigh/TGF-βlow).45,46 Interestingly, the cross-comparison of did not block resistance to vemurafenib whereas PI3K inhibitors latest studies on drug resistance in melanoma indicates that both significantly did, multiple RTKs may collaboratively act as drivers 30 (non-proliferative) phenotypes, either MITF- or TGF-β-triggered, for resistance at the same time. Thus, particularly the PI3K can promote reduced sensitivity to MAPK inhibition.30,36 As a pathway, which pools signals from different RTKs, and its consequence, the proliferation state by itself and, tightly linked to downstream effector mTOR are being discussed as druggable this, the respective metabolic state could represent major targets for future combinatorial therapies.31,35 phenotypic parameters for the therapeutic susceptibility of melanoma cell subpopulations. Phenotypic drug resistance is dynamically linked to cell cycle Otto Warburg initially postulated that cancer cells rely on an heterogeneity, cell differentiation and metabolic reprogramming increase in glucose uptake, the enhancement of glycolysis and Oncogenic BRAF not only controls cell proliferation and survival, elevated lactate production along with the absence of oxidative but is also directly linked to metabolic reprogramming of respiration in the mitochondria despite the presence of sufficient melanoma cells. Haq et al.36 have recently demonstrated by gene oxygen concentrations.47 Against Warburg’s initial assumption, set enrichment analysis of previously published gene expression mitochondria in cancer cells are not commonly defective. The profiles that treatment of BRAFV600E-positive melanoma cells (but reason for the unexpected usage of nutrients in cancer cells is not wild-type cells) by BRAFi or MEKi induces mRNA expression of most likely the requirement of rapidly proliferating tissue for both mitochondrial genes that are involved in the citric acid cycle and production of bioenergy and generation of biomass for cell oxidative phosphorylation (OXPHOS) of ATP. Also the density and duplication. It seems that glycolysis and interconnected pathways, number of mitochondria were found enriched in resistant for example, the pentose phosphate pathway, can satisfy both melanoma cells as indicated by Mito Tracker Green and electron needs in a metabolically balanced way, that is, ATP production microscopy analysis. In contrast, lactate levels were decreased. and generation of cellular building blocks, for example, riboses for Accordingly, Corazao-Rozas et al.37 found a stable increase in the nucleotide synthesis.48 However, it is likely that cancer hetero- oxygen consumption rate and mitochondrial production of geneity also applies to the level of energy metabolism following reactive oxygen species in melanoma cells treated with vemur- evolutionary conserved principles.49 Unicellular organisms and afenib. As a possible regulatory mechanism, an (up to 14-fold) cells from multicellular organisms pursue similar metabolic induction of the peroxisome proliferator-activated receptor γ strategies under similar environmental conditions. Microbes and coactivator 1 family member PGC1α has been suggested as a proliferating cells in a multicellular organism both metabolize result of BRAFi treatment in BRAV600E-positive melanoma cells glucose primarily through glycolysis. However, unicellular organ- but not in wild-type cells or BRAFV600E-positive colon carcinoma isms that lack nutrients rely on oxidative metabolism, as do cells in cells.36 PGC1 factors are commonly involved in the dynamic a multicellular organism that are not stimulated to proliferate, for

© 2015 Macmillan Publishers Limited Oncogene (2015) 2951 – 2957 Resistance to MAPK pathway inhibition in melanoma A Roesch 2954 example, when they are differentiated. As cells switch from slow to whereas glycolysis was downregulated as indicated by low rapid proliferation and vice versa, they exhibit a remarkable hexokinase expression. The addition of inhibitors of the mito- flexibility to switch also between oxidative and glycolytic ATP chondrial electron transport chain (phenformin, rotenone) or ATP production (Figure 1).49,50 synthase (oligomycin A, Bz-423) to melanoma drugs such as Recent studies by others and us confirmed that a hitherto vemurafenib significantly improved drug susceptibility and unknown phenotype of melanoma cells characterized by low cell eliminated the intrinsically resistant slow-cycling JARID1Bhigh doubling rate truly represents a major determinant for drug subpopulation in vitro and in mouse xenografts.53,54 Except for 51 resistance. Sharma et al. demonstrated that different types of phenformin, which has already been used in the clinics as an anticancer, including melanoma, harbor populations of quiescent cells diabetic drug, most of the so far available inhibitors of oxidative that are drug resistant and whose chromatin states can switch mitochondrial metabolism may not be suitable for combination dynamically and reversibly. Such quiescent cells were character- therapies because of their high toxicity. Clinical studies with ized by high expression of the chromatin-remodeling factor phenformin or novel, more tolerable mitochondrial inhibitors are fi JARID1A. Our group simultaneously identi ed the histone 3 lysine highly needed to show a benefit of this new combinatorial 4 demethylase JARID1B/KDM5b, a close homologue of JARID1A, as concept in humans. a marker for a small subpopulation of slow-cycling cells, which are In contrast to a preexisting OXPHOS-dependent slow cycling essential for continuous tumor growth and repopulation of fi 52 phenotype, which may mediate resistance already at the very rst melanoma. The slow-cycling subpopulation was found in 53 contact to drugs, absence of cell proliferation can also be melanomas irrespective of the cells’ genetic origin. In contrast to acquired in response to MAPK inhibition. It has been recently classic cancer stem cells, this slow-cycling phenotype did not shown that vemurafenib induces features of stress-induced follow a static tumor hierarchy, but could be acquired also by cells senescence in addition to apoptosis.55 This senescence phenotype of the rapidly proliferating tumor bulk. Depending on micro- was determined by heterochromatin formation, changes in cell environmental factors like the oxygen level, we observed a β dynamic interconversion between the rapidly proliferating JAR- shape and increased senescence-associated -galactosidase activ- ID1Blow and the slow-cycling JARID1Bhigh phenotype. Slow-cycling ity. Interestingly, the overexpression of RTKs like EGFR in MEKi- high resistant melanoma cells also was found accompanied by a slow- JARID1B melanoma cells displayed multi-drug resistance 30 irrespective of the agents used, for example, vemurafenib, growth phenotype. EGFR expression in A375 melanoma cells led cisplatin, temozolomide, bortezomib and so on, and were to hypophosphorylation of pRB, induction of the CDK inhibitors selectively enriched under therapy suggesting a high intrinsic CDKN1A/B and increased β-galactosidase staining. A possible capacity for survival of this subpopulation.53,54 As described by explanation how slow-cycling phenotypes can mediate both a Haq et al.36 for induced metabolic resistance against BRAF priori resistance resulting in selective phenotype selection and inhibition, also intrinsically resistant slow-cycling melanoma cells induced resistance in surviving cells could be the effective drug relied on OXPHOS and mitochondrial supply with ATP (Figure 2).53 concentration exposed to the cells. For example, Haferkamp 55 Liquid chromatography-mass spectrometry-based proteomic et al. challenged melanoma cells in vitro with 0.5 μM of profiling together with measurement of oxygen consumption vemurafenib up to 7 days, whereas we used 2.5–50 μM vemur- and mitochondrial ATP production unraveled a significant afenib for 72 hrs.53 Because the effective drug concentration may activation of the citrate cycle and OXPHOS in JARID1Bhigh cells also broadly vary in humans depending on factors like anatomical

Figure 1. Dynamic shifting between evolutionarily conserved metabolic phenotypes. Unicellular (a) and multicellular organisms (b) can dynamically switch between glycolytic and oxidative bioenergy production depending on microenvironmental parameters such as supply with nutrients or stimulation by growth signals. As an evolutionary principle, proliferating cells rely on cytoplasmic glycolysis, whereas quiescent cells rely on oxidation of pyruvate to CO2 during the mitochondrial citrate cycle and OXPHOS of ATP (according to Vander Heiden et al.49).

Figure 2. Metabolic and proliferative tumor heterogeneity. Proliferating tumor cells metabolize most of the glucose through glycolysis irrespective of the supply with oxygen; in particular, to fulﬁll their need to produce enough building blocks for subsequent cell doubling (Warburg effect). Slow cycling tumor cells do not require biomass supply for daughter cells and, thus, apply the more energy-efﬁcient process of OXPHOS to produce ATP (36 mol ATP vs 2 mol ATP from glycolysis) (according to Vander Heiden et al.49).

conditions, the level of vascularization or necrosis, both scenarios inﬂuence the dynamic shift and development of phenotypic could potentially occur in patients. resistance in melanoma.57

Influences of the microenvironment on resistant cell phenotypes Experimental modeling of resistance and future clinical relevance Alterations in RTK signaling are involved in the development of Das Thakur et al.61 have recently demonstrated in a time-course MAPKi resistance in many respects. In addition to the mechanisms xenograft model how melanoma cells dynamically adapt to discussed above, two groups have recently reported that stromal- vemurafenib treatment but—after a short period of ‘drug derived growth factors like HGF confer resistance to BRAFi and holidays’—can become vemurafenib-sensitive again. By using an MEKi in melanoma cells through c-MET activation of ERK1/2.56,57 intermittent on and off dosing schedule in their experiments, they The c-MET inhibitor crizotinib antagonized this resistance in vitro could double the total time of response from 100 to 200 days. and in xenograft models. Strikingly, the pre-treatment plasma HGF Resistance was conferred in their model through elevated BRAF level of melanoma patients from the BRIM2 trial was indicative for protein expression (in part, mediated through BRAFV600E the subsequent vemurafenib response as demonstrated by amplification), which restored MAPK signaling despite the significant differences in progression-free and overall survival presence of vemurafenib. Intriguingly, phospho ERK levels of between HGFhigh and HGFlow patients.57 Systematical co-culturing resistant cells were not only restored, but even exceeded those of of melanoma cells with 23 different stromal cell types further untreated parental cells. As a consequence of this ‘drug addiction’, emphasized the role of soluble factors, particularly HGF, as melanoma cells paradoxically died when vemurafenib was mediators of resistance in melanoma, colon cancer and glioblas- suddenly removed. This addiction-related death lasted for toma cells in vitro.56 Also in the stroma of melanoma patients approximately 10 days. Thereafter, the remaining melanoma cells under therapy, elevated HGF expression was detected by adapted to this new situation, normalized their phospho ERK immunostaining further supporting HGF’s role as a potential levels, and finally became sensitive again to BRAF inhibition. Cells biomarker for disease control. expressing mutant BRAF splice variants can undergo similar Hypoxia has been recently described as another environmental addiction processes.62 Also, the study by Sun et al.30 showed that factor influencing cell phenotypes and drug resistance in melanoma cells that acquired resistance to vemurafenib through melanoma. O’Connell et al.58 demonstrated that hypoxia- expression of high levels of EGFR could not survive when induced signal switching within the non-canonical Wnt5A path- vemurafenib was depleted from the culture medium. Although, way confers resistance to BRAFi. Moreover, they have found that re-sensitization of the cells was not in the focus of Sun’s study, it Wnt5A expression correlated with clinical response and therapy could be that any phenotypic resistance to MAPKi responds to resistance in melanomas treated with vemurafenib. These results intermittent or sequential combinatory schedules. Certainly, in the are highly compatible with other recently published observations real situation in humans, multiple resistance mechanisms may regarding the connection of dynamic phenotype switching and occur at the same time and, thus, could impede intermitting drug resistance. For example, Cheli et al.59 have reported that dosing regimens in the clinics. New clinical trials addressing this hypoxic conditions decrease MITF expression through an indirect, aspect are highly needed. HIF1α-dependent transcriptional mechanism and increase the In sum, all discussed studies suggest that not just melanomas of tumorigenic and metastatic properties of melanoma cells. In different patients, but also different metastases within one patient addition, our group observed a rapid, but reversible shift towards and even cells within one tumor can develop different kinds of the slow-cycling, multi-drug resistant cell phenotype after resistance mechanisms comprising genomic and phenotypic incubating cells at 1% of oxygen.52,53 It will be interesting to see mechanisms and that we cannot predict which of them will how targeting of microenvironmental influences like stromal actually occur. The recent work of Gowrishankar et al.63 illustrates growth factors (e.g., by crizotinib) or hypoxia (e.g., by HIF-1α- that both types of resistance, that is, genetic (e.g., activating inhibitors like YC-160) in a combined experimental setting will NRASQ61H mutation) and phenotypic (e.g., accumulation of

© 2015 Macmillan Publishers Limited Oncogene (2015) 2951 – 2957 Resistance to MAPK pathway inhibition in melanoma A Roesch 2956 COT1), can even originate from a single melanoma cell within a 18 Shi H, Moriceau G, Kong X, Lee MK, Lee H, Koya RC et al. Melanoma whole-exome relatively short period of time. This has crucial implications sequencing identifies (V600E)B-RAF amplification-mediated acquired B-RAF inhi- for future personalized medicine in melanoma patients: (i) The bitor resistance. Nat Commun 2012; 3: 724. predictive value of pre-therapeutic tumor screening is uncertain. 19 Poulikakos PI, Persaud Y, Janakiraman M, Kong X, Ng C, Moriceau G et al. RAF inhibitor resistance is mediated by dimerization of aberrantly spliced BRAF Tumors may still respond, for example, despite MEK mutations or (V600E). Nature 2011; 480:387–390. PI3K-AKT-PTEN pathway alterations. (ii) Because most likely every 20 Wagle N, Emery C, Berger MF, Davis MJ, Sawyer A, Pochanard P et al. Dissecting patient develops a broad spectrum of resistance, single follow-up therapeutic resistance to RAF inhibition in melanoma by tumor genomic profiling. biopsies cannot assess the entire time-and-spatial heterogeneity J Clin Oncol 2011; 29: 3085–3096. of all metastases. Consequently, there is a high need for novel 21 Shi H, Moriceau G, Kong X, Koya RC, Nazarian R, Pupo GM et al. Preexisting MEK1 diagnostic imaging and molecular screening technologies as well exon 3 mutations in V600E/KBRAF melanomas do not confer resistance to BRAF 2 – as new combination therapies that also consider phenotypic inhibitors. Cancer Discov 2012; : 414 424. 22 Nathanson KL, Martin AM, Wubbenhorst B, Greshock J, Letrero R, D'Andrea K heterogeneity. The prevention of development of resistance could et al. Tumor genetic analyses of patients with metastatic melanoma treated be a helpful strategy, for example by eliminating all melanoma with the BRAF inhibitor dabrafenib (GSK2118436). Clin Cancer Res 2013; 19: cells right from the beginning of therapy, particularly targeting 4868–4878. mechanisms of intrinsic resistance. 23 Villanueva J, Infante JR, Krepler C, Reyes-Uribe P, Samanta M, Chen HY et al. Concurrent MEK2 mutation and BRAF amplification confer resistance to BRAF and MEK inhibitors in melanoma. Cell Rep 2013; 4: 1090–1099. CONFLICT OF INTEREST 24 Wagle N, Van Allen EM, Treacy DJ, Frederick DT, Cooper ZA, Taylor-Weiner A et al. The author declares no conflict of interest MAP kinase pathway alterations in BRAF-mutant melanoma patients with acquired resistance to combined RAF/MEK inhibition. Cancer Discov 2014; 4: 61–68. REFERENCES 25 Johannessen CM, Johnson LA, Piccioni F, Townes A, Frederick DT, Donahue MK et al. A melanocyte lineage program confers resistance to MAP kinase pathway 1 Balch CM, Gershenwald JE, Soong SJ, Thompson JF, Atkins MB, Byrd DR et al. Final inhibition. Nature 2013; 504: 138–142. fi 27 version of 2009 AJCC melanoma staging and classi cation. J Clin Oncol 2009; : 26 Romano E, Pradervand S, Paillusson A, Weber J, Harshman K, Muehlethaler K et al. – 6199 6206. Identification of multiple mechanisms of resistance to vemurafenib in a patient 2 Schadendorf D, Hauschild A. Melanoma in 2013: Melanoma--the run of success with BRAFV600E-mutated cutaneous melanoma successfully rechallenged after 11 – continues. Nat Rev Clin Oncol 2014; :75 76. progression. Clin Cancer Res 2013; 19: 5749–5757. 3 Chapman PB, Hauschild A, Robert C, Haanen JB, Ascierto P, Larkin J et al. 27 Johannessen CM, Boehm JS, Kim SY, Thomas SR, Wardwell L, Johnson LA et al. Improved survival with vemurafenib in melanoma with BRAF V600E mutation. COT drives resistance to RAF inhibition through MAP kinase pathway reactivation. 364 – New Engl J Med 2011; : 2507 2516. Nature 2010; 468:968–972. 4 Hauschild A, Grob JJ, Demidov LV, Jouary T, Gutzmer R, Millward M et al. 28 Montagut C, Sharma SV, Shioda T, McDermott U, Ulman M, Ulkus LE et al. Elevated Dabrafenib in BRAF-mutated metastatic melanoma: a multicentre, open-label, CRAF as a potential mechanism of acquired resistance to BRAF inhibition in 380 – phase 3 randomised controlled trial. Lancet 2012; : 358 365. melanoma. Cancer Res 2008; 68: 4853–4861. 5 McArthur GA, Chapman PB, Robert C, Larkin J, Haanen JB, Dummer R et al. Safety 29 Shao Y, Aplin AE. BH3-only protein silencing contributes to acquired resistance to fi and ef cacy of vemurafenib in BRAF(V600E) and BRAF(V600K) mutation-positive PLX4720 in human melanoma. Cell Death Differ 2012; 19: 2029–2039. melanoma (BRIM-3): extended follow-up of a phase 3, randomised, open- 30 Sun C, Wang L, Huang S, Heynen GJ, Prahallad A, Robert C et al. Reversible and 15 – label study. Lancet Oncol 2014; :323 332. adaptive resistance to BRAF(V600E) inhibition in melanoma. Nature 2014; 508: 6 Flaherty KT, Robert C, Hersey P, Nathan P, Garbe C, Milhem M et al. Improved 118–122. survival with MEK inhibition in BRAF-mutated melanoma. New Engl J Med 2012; 31 Villanueva J, Vultur A, Lee JT, Somasundaram R, Fukunaga-Kalabis M, Cipolla AK 367 – :107 114. et al. Acquired resistance to BRAF inhibitors mediated by a RAF kinase switch in 7 Flaherty KT, Infante JR, Daud A, Gonzalez R, Kefford RF, Sosman J et al. Combined melanoma can be overcome by cotargeting MEK and IGF-1 R/PI3K. Cancer Cell BRAF and MEK inhibition in melanoma with BRAF V600 mutations. New Engl J Med 2010; 18: 683–695. 2012; 367: 1694–1703. 32 Corcoran RB, Ebi H, Turke AB, Coffee EM, Nishino M, Cogdill AP et al. EGFR- 8 Flaherty KT, Puzanov I, Kim KB, Ribas A, McArthur GA, Sosman JA et al. Inhibition mediated re-activation of MAPK signaling contributes to insensitivity of BRAF of mutated, activated BRAF in metastatic melanoma. New Engl J Med 2010; 363: mutant colorectal cancers to RAF inhibition with vemurafenib. Cancer Discov 809–819. 2012; 2: 227–235. 9 Hartsough E, Shao Y, Aplin AE. Resistance to RAF Inhibitors Revisited. J Invest 33 Prahallad A, Sun C, Huang S, Di Nicolantonio F, Salazar R, Zecchin D et al. Unre- Dermatol 2014; 134:319–325. sponsiveness of colon cancer to BRAF(V600E) inhibition through feedback acti- 10 Sosman JA, Kim KB, Schuchter L, Gonzalez R, Pavlick AC, Weber JS et al. Survival in vation of EGFR. Nature 2012; 483: 100–103. BRAF V600-mutant advanced melanoma treated with vemurafenib. New Engl J 34 Abel EV, Basile KJ, Kugel CH 3rd, Witkiewicz AK, Le K, Amaravadi RK et al. Mela- Med 2012; 366: 707–714. noma adapts to RAF/MEK inhibitors through FOXD3-mediated upregulation 11 Lito P, Rosen N, Solit DB. Tumor adaptation and resistance to RAF inhibitors. Nat of ERBB3. J Clin Invest 2013; 123: 2155–2168. Med 2013; 19: 1401–1409. 35 Shi H, Kong X, Ribas A, Lo RS. Combinatorial treatments that overcome 12 Nazarian R, Shi H, Wang Q, Kong X, Koya RC, Lee H et al. Melanomas acquire PDGFRbeta-driven resistance of melanoma cells to V600EB-RAF inhibition. Cancer resistance to B-RAF(V600E) inhibition by RTK or N-RAS upregulation. Nature 2010; Res 2011; 71: 5067–5074. 468: 973–977. 36 Haq R, Shoag J, Andreu-Perez P, Yokoyama S, Edelman H, Rowe GC et al. Onco- 13 Rizos H, Menzies AM, Pupo GM, Carlino MS, Fung C, Hyman J et al. BRAF Inhibitor genic BRAF regulates oxidative metabolism via PGC1alpha and MITF. Cancer Cell Resistance Mechanisms in Metastatic Melanoma: Spectrum and Clinical Impact. 2013; 23: 302–315. Clin Cancer Res 2014; 20:1965–1977. 37 Corazao-Rozas P, Guerreschi P, Jendoubi M, Andre F, Jonneaux A, Scalbert C et al. 14 Shi H, Hugo W, Kong X, Hong A, Koya RC, Moriceau G et al. Acquired resistance Mitochondrial oxidative stress is the Achille's heel of melanoma cells resistant to and clonal evolution in mMelanoma during BRAF inhibitor therapy. Cancer Discov Braf-mutant inhibitor. Oncotarget 2013; 4: 1986–1998. 2013. 38 Girnun GD. The diverse role of the PPARgamma coactivator 1 family of tran- 15 Van Allen EM, Wagle N, Sucker A, Treacy DJ, Johannessen CM, Goetz EM et al. The scriptional coactivators in cancer. Semin Cell Dev Biol 2012; 23: 381–388. genetic landscape of clinical resistance to RAF inhibition in metastatic melanoma. 39 Lin J, Handschin C, Spiegelman BM. Metabolic control through the PGC-1 family Cancer Discov 2014; 4:94–109. of transcription coactivators. Cell Metab 2005; 1: 361–370. 16 Trunzer K, Pavlick AC, Schuchter L, Gonzalez R, McArthur GA, Hutson TE et al. 40 Garraway LA, Widlund HR, Rubin MA, Getz G, Berger AJ, Ramaswamy S et al. Pharmacodynamic effects and mechanisms of resistance to vemurafenib in Integrative genomic analyses identify MITF as a lineage survival oncogene patients with metastatic melanoma. J Clin Oncol 2013; 31: 1767–1774. amplified in malignant melanoma. Nature 2005; 436:117–122. 17 Su F, Bradley WD, Wang Q, Yang H, Xu L, Higgins B et al. Resistance to selective 41 Cheli Y, Giuliano S, Botton T, Rocchi S, Hofman V, Hofman P et al. Mitf is the key BRAF inhibition can be mediated by modest upstream pathway activation. Cancer molecular switch between mouse or human melanoma initiating cells and their Res 2012; 72: 969–978. differentiated progeny. Oncogene 2011; 30:2307–2318.

Oncogene (2015) 2951 – 2957 © 2015 Macmillan Publishers Limited Resistance to MAPK pathway inhibition in melanoma A Roesch 2957 42 Hoek KS, Eichhoff OM, Schlegel NC, Dobbeling U, Kobert N, Schaerer L et al. In vivo 54 Yuan P, Ito K, Perez-Lorenzo R, Del Guzzo C, Lee JH, Shen CH et al. Phenformin switching of human melanoma cells between proliferative and invasive states. enhances the therapeutic benefit of BRAFV600E inhibition in melanoma. Proc Natl Cancer Res 2008; 68:650–656. Acad Sci USA 2013; 110: 18226–18231. 43 Hoek KS, Goding CR. Cancer stem cells versus phenotype-switching in melanoma. 55 Haferkamp S, Borst A, Adam C, Becker TM, Motschenbacher S, Windhovel S et al. Pigment Cell Melanoma Res 2010; 23:746–759. Vemurafenib induces senescence features in melanoma cells. J Invest Dermatol 44 Widmer DS, Cheng PF, Eichhoff OM, Belloni BC, Zipser MC, Schlegel NC et al. 2013; 133: 1601–1609. Systematic classification of melanoma cells by phenotype-specific gene expres- 56 Straussman R, Morikawa T, Shee K, Barzily-Rokni M, Qian ZR, Du J et al. Tumour sion mapping. Pigment Cell Melanoma Res 2012; 25: 343–353. micro-environment elicits innate resistance to RAF inhibitors through HGF 45 Carreira S, Goodall J, Denat L, Rodriguez M, Nuciforo P, Hoek KS et al. Mitf reg- secretion. Nature 2012; 487:500–504. ulation of Dia1 controls melanoma proliferation and invasiveness. Genes Dev 57 Wilson TR, Fridlyand J, Yan Y, Penuel E, Burton L, Chan E et al. Widespread 2006; 20: 3426–3439. potential for growth-factor-driven resistance to anticancer kinase inhibitors. 46 Saez-Ayala M, Montenegro MF, Sanchez-Del-Campo L, Fernandez-Perez MP, Nature 2012; 487:505–509. Chazarra S, Freter R et al. Directed phenotype switching as an effective anti- 58 O'Connell MP, Marchbank K, Webster MR, Valiga AA, Kaur A, Vultur A et al. melanoma strategy. Cancer Cell 2013; 24:105–119. Hypoxia induces phenotypic plasticity and therapy resistance in melanoma 47 Jose C, Bellance N, Rossignol R. Choosing between glycolysis and oxidative via the tyrosine kinase receptors ROR1 and ROR2. Cancer Discov 2013; 3: phosphorylation: a tumor's dilemma? Biochim Biophys Acta 2011; 1807:552–561. 1378–1393. 48 Levine AJ, Puzio-Kuter AM. The control of the metabolic switch in cancers by 59 Cheli Y, Giuliano S, Fenouille N, Allegra M, Hofman V, Hofman P et al. Hypoxia and oncogenes and tumor suppressor genes. Science 2010; 330: 1340–1344. MITF control metastatic behaviour in mouse and human melanoma cells. Onco- 49 Vander Heiden MG, Cantley LC, Thompson CB. Understanding the Warburg effect: gene 2012; 31: 2461–2470. the metabolic requirements of cell proliferation. Science 2009; 324: 1029–1033. 60 Kluza J, Corazao-Rozas P, Touil Y, Jendoubi M, Maire C, Guerreschi P et al. Inac- 50 Berridge MV, Herst PM, Tan AS. Metabolic flexibility and cell hierarchy in tivation of the HIF-1alpha/PDK3 signaling axis drives melanoma toward mito- metastatic cancer. Mitochondrion 2010; 10:584–588. chondrial oxidative metabolism and potentiates the therapeutic activity of 51 Sharma SV, Lee DY, Li B, Quinlan MP, Takahashi F, Maheswaran S et al. pro-oxidants. Cancer Res 2012; 72: 5035–5047. A chromatin-mediated reversible drug-tolerant state in cancer cell subpopula- 61 Das Thakur M, Salangsang F, Landman AS, Sellers WR, Pryer NK, Levesque MP tions. Cell 2010; 141:69–80. et al. Modelling vemurafenib resistance in melanoma reveals a strategy to fore- 52 Roesch A, Fukunaga M, Schmidt E, Zabierowski S, Brafford P, Vultur A et al. stall drug resistance. Nature 2013; 494: 251–255. A temporarily distinct subpopulation of slow-cycling melanoma cells is required 62 Hartsough EJ, Basile KJ, Aplin AE. Beneficial Effects of RAF Inhibitor in Mutant for continuous tumor growth. Cell 2010; 141:583–594. BRAF Splice Variant-Expressing Melanoma. Molecular cancer research: MCR 2014; 53 Roesch A, Vultur A, Wang H, Speicher D, Bogeski I, Hoth M et al. Overcoming 12:795–802. Intrinsic Multidrug Resistance in Melanoma by Blocking the Mitochondrial 63 Gowrishankar K, Snoyman S, Pupo GM, Becker TM, Kefford RF, Rizos H. Acquired Respiratory Chain of Slow-Cycling JARID1B high Cells. Cancer Cell 2013; 23: resistance to BRAF inhibition can confer cross-resistance to combined BRAF/MEK 811–825. inhibition. J Invest Dermatol 2012; 132: 1850–1859.