Preexisting MEK1 Exon 3 Mutations in ^V600E/KBRAF Melanomas Do Not Confer Resistance to BRAF Inhibitors

Abstract

BRAF inhibitors (BRAFi) induce antitumor responses in nearly 60% of patients with advanced ^V600E/KBRAF melanomas. Somatic activating MEK1 mutations are thought to be rare in melanomas, but their potential concurrence with ^V600E/KBRAF may be selected for by BRAFi. We sequenced MEK1/2 exon 3 in melanomas at baseline and upon disease progression. Of 31 baseline ^V600E/KBRAF melanomas, 5 (16%) carried concurrent somatic BRAF/MEK1 activating mutations. Three of 5 patients with BRAF/MEK1 double-mutant baseline melanomas showed objective tumor responses, consistent with the overall 60% frequency. No MEK1 mutation was found in disease progression melanomas, except when it was already identified at baseline. MEK1-mutant expression in ^V600E/KBRAF melanoma cell lines resulted in no significant alterations in p-ERK1/2 levels or growth-inhibitory sensitivities to BRAFi, MEK1/2 inhibitor (MEKi), or their combination. Thus, activating MEK1 exon 3 mutations identified herein and concurrent with ^V600E/KBRAF do not cause BRAFi resistance in melanoma.

SIGNIFICANCE

As BRAF inhibitors gain widespread use for treatment of advanced melanoma, bio-markers for drug sensitivity or resistance are urgently needed. We identify here concurrent activating mutations in BRAF and MEK1 in melanomas and show that the presence of a downstream mutation in MEK1 does not necessarily make BRAF–mutant melanomas resistant to BRAF inhibitors.

INTRODUCTION

The majority of melanomas harbor an activating mutation in the RAS/RAF/mitogen-activated protein kinase (MAPK) signaling pathway. ^V600E/KBRAF mutations are found in >50% of melanomas (1), and their “druggability” in humans has been shown using the novel small-molecule BRAF inhibitors (BRAFi) PLX4032/vemurafenib and GSK2118436/dabrafenib (2-5). However, acquired resistance to BRAFi is common, and proposed mechanisms include upregulation of MAPK-redundant signaling (via receptor tyrosine kinase overexpression and AKT activation) and MAPK reactivation [via N-RAS mutations, COT expression, ^V600EBRAF alternative splicing, and ^V600EBRAF amplification (6-11)]. A missense MEK1 somatic activating mutation (C121S) in exon 3 has recently been proposed to account for acquired BRAFi resistance in one patient (12). Germline mis-sense MEK1/2 mutations have been found in patients with the developmental disorder known as cardio-facio-cutaneous syndrome (13). However, somatic activating MEK1/2 mutations are thought to be exceedingly rare among human malignancies (14, 15).

To assess the potential role of MEK1/2 exon 3 mutations in primary (innate) or secondary (acquired) drug resistance to BRAFi therapy, we analyzed samples from 31 patients treated with either vemurafenib or dabrafenib from whom there was available baseline (prior to BRAFi treatment) or patient-matched baseline and disease progression [DP (i.e., acquired BRAFi–resistant)] tissues. Contrary to the expectation, somatic MEK1 exon 3 mutations (^P124SMEK1 and ^I111SMEK1) were found in baseline melanoma tumors concurrent with ^V600E/KBRAF mutations. Importantly, the pattern of MEK1 exon 3 mutations could not account for either innate or acquired BRAFi resistance. Functional studies using BRAF/MEK1 double-mutant melanoma cell lines show that ^P124SMEK1 and ^I111SMEK1 do not determine BRAFi sensitivity. This clinical series thus offers important insight into the tumor response pattern of BRAF/MEK1 double-mutant melanomas to novel BRAF inhibitors.

RESULTS

Among 31 patients with BRAF–mutant melanomas treated with either the BRAFi, PLX4032/vemurafenib or GSK2118436/dabrafenib, 5 MEK1 exon 3 mutations (P124S in 4, I111S in 1) were detected in baseline melanoma tumors of 5 distinct patients (Table 1). In 4 of these 5 patients with available normal tissue-derived genomic DNA (gDNA), these MEK1 mutations were determined to be somatic (Supplementary Fig. S1). No MEK2 exon 3 mutation was detected in any baseline melanoma tumor (Table 1). Among 18 of these patients whose BRAFi acquired-resistant (DP) tumors were available, MEK1 exon 3 mutation was detected in only those patients with preexisting BRAF/MEK1, double-mutant baseline melanomas and none of the other 23 DP tumors [several patients have multiple DP tissue biopsies (Table 1)]. MEK1 exon 3 mutant alleles were detected in 9 tumor tissues (both baseline and DP) at a ratio of 1:1 to wild-type (WT) MEK1; no homozygous MEK1 mutation was detected (Supplementary Fig. S1), suggesting counterselection against MEK1-mutant homozygosity. MEK1 mutations concurred with ^V600EBRAF in 2 patients and ^V600KBRAF in 3 patients (Supplementary Fig. S1).

Table 1.

Summary of BRAF, MEK1/2 mutation status, and patient characteristics

Study site	Patient number	BRAF exon 15 baseline	MEK1 exon 3 baseline	MEK2 exon 3 baseline	MEK1 exon 3 at progressive disease	Age and gender	Stage	Dose of BRAFi (mg)	Best overall response	Progression-free survival (days)	Sites of baseline tumors biopsieda	Responses of baselinea
UCLA	1	V600E	WT	WT	WT	65 F	M1c	1120 bid	−37%	279	—	—
					WT
					WT
	2	V600E	P124S	WT	P124S	46 M	M1c	960 bid	−32%	149	Axillary node	Day 15 PET uptake decrease (Fig. 1A)
	3	V600E	WT	WT	WT	66 F	M1b	960 bid	−53%	137	—	—
					WT
					WT
	4	V600E	WT	WT	WT	48 M	M1c	960 bid	−24%	126	—	—
					WT
	5	V600E	WT	WT	WT	54 M	M1c	960 bid	−75%	84	—	—
	6	V600E	WT	WT	WT	65 F	M1c	960 bid	−72%	373	—	—
					WT
	7	V600E	WT	WT	WT	51 M	M1c	960 bid	−60%	212	—	—
					WT
	8	V600E	WT	WT	WT	47 F	M1b	960 bid	−72%	97	—	—
	9b	V600E	WT	WT	WT	47 F	IIIc	150 bid	−31%	238	—	—
					WT
					WT
	10	V600E	WT	WT	DP N/A	52 M	M1c	960 bid	−70%	—	—	—
	11	V600K	P124S	WT	DP N/A	72 F	M1c	960 bid	−11%	80	Axillary node	Decrease by 25% in longest tumor dimension
	12b	V600E	P124S	WT	DP N/A	48 F	M1c	150 bid	−9%	104	Axillary node	Not palpable (Fig. 1B)
	13	V600K	WT	WT	DP N/A	66 M	M1c	960 bid	−30%	140	—	—
	14	V600E	WT	WT	DP N/A	57 M	M1c	960 bid	−47%	189	—	—
	15	V600E	WT	WT	DP N/A	55 M	M1c	960 bid	−50%	300	—	—
	16	V600E	WT	WT	Responding	53 F	M1a	960 bid	−91.7%	—	—	—
	17	V600E	WT	WT	Responding	45 F	M1c	960 bid	−73%	—	—	—
	18	V600E	WT	WT	DP N/A	29 F	M1c	960 bid	−34.5%	289	—	—
	19	V600E	WT	WT	Responding	74 M	IIIc	960 bid	−20%	—	—	—
	20	V600K	WT	WT	WT	65 M	M1c	960 bid	−22%	161	—	—
	21	V600K	WT	WT	Responding	61 M	M1c	960 bid	−34%	—	—	—
	22	V600E	WT	WT	Responding	60 F	M1a	960 bid	−42%	—	—	—
	23b	V600E	WT	WT	DP N/A	36 F	M1c	150 bid	−66%	85	—	—
	24b	V600E	WT	WT	DP N/A	65 F	M1c	150 bid	−39.5%	76	—	—
MIA	25b	V600K	WT	WT	WT	59 M	M1c	100 bid, dose escalated to 100 tid	−14%	224	—	—
	26	V600E	WT	WT	WT	71 M	M1c	960 bid	−53%	239	—	—
	27b	V600K	I111S	WT	I111S	38 F	M1a	35 bid, dose escalated to 100 tid 18 Dec 2009	−100%	119	Right inguinal node	Day 15 PET uptake decrease in single measurable lesion
					I111S
	28b	V600K	P124S	WT	P124S	39 M	M1c	150 bid	−56%	118	Right neck node	Day 15 PET uptake decrease (regional, Fig. 1C); SUV max reduced 48% overall metastases
	29b		WT	WT	WT	58 F	M1c	70 bid, dose escalated to 100 tid 29 Dec 2009	−63%	209	—	—
VI	30	V600E	WT	WT	WT	46 F	M1c	720	−59%	107	—	—
	31	V600E	WT	WT	WT	70 F	M1a	960	−70%	183	—	—

NOTE: Highlighting, information for patients whose tumors harbor MEK1 mutations. Dashes, data not applicable.

Mean progression-free survival for patients with WT MEK1 melanomas 5 182.4 days (SD 84.7) versus mutant MEK1 melanomas 5 114 days [SD 25.1, (2-tailed P = 0.09)].Mean best overall response for patients with WT MEK1 melanomas = −50% (SD 20.0) versus mutant MEK1 melanomas = 41.6 [SD 37.8, 2-tailed P = 0.45)]. Partial response (PR) for patients with WT MEK1 melanomas (21/26 or 80%) versus mutant MEK1 melanomas (3/5 or 60%).

Abbreviations: bid, twice a day; DP, disease progression; MIA, Melanoma Institute, Australia; PET, positron emission tomography; SUV_max, maximum standard uptake value; tid, 3 times a day; UCLA, University of California, Los Angeles; VI, Vanderbilt-Ingram Cancer Center

^aInformation only for applicable patients with MEK1-mutant melanomas.

^bDabrafenib-treated patients. All others were treated with vemurafenib.

Four of 5 patients with BRAF/MEK1 double-mutant baseline melanomas displayed objective responses in the tumors biopsied, and 3 of 5 patients achieved overall partial response (by Response Evaluation Criteria in Solid Tumors 1.1) to BRAF inhibition (Table 1 and Fig. 1A–C). Both patients who did not achieve objective partial response experienced clinical response (Table 1 and Supplementary Fig. S2). The mean progression-free survival and best overall tumor response were not significantly different between BRAF single-mutant versus BRAF/MEK1 double-mutant melanomas (Table 1): mean progression-free survival for patients with WT MEK1 melanomas = 182.4 days (SD 84.7) versus mutant MEK1 melanomas = 114 days (SD 25.1), 2-tailed P = 0.09; mean best overall response for patients with WT MEK1 melanomas = −50% (SD 20.0) versus mutant MEK1 melanomas = 41.6 (SD 37.8), 2-tailed P = 0.45. Thus, we have uncovered a concurrence of MEK1 and BRAF mutations in metastatic melanomas prior to the onset of BRAFi selective pressure. The pattern of MEK1 exon 3 mutations and clinical responses could not account for either innate or acquired BRAFi resistance, warranting in vitro validation.

Figure 1.

Spectrum of clinical responses of ^V600E/KBRAF/^P124SMEK1 double-mutant melanomas to BRAF inhibitors. A, in patient 2, a right axillary nodal melanoma clearly responded to vemurafenib at day 15 [positron emission tomography (PET) scan], at which time a “baseline” biopsy of the same tumor revealed mutations in both BRAF and MEK1. Tumor tracking shows timing of baseline and DP tumor biopsies. Best overall response demonstrates a partial response, and biopsy locations are indicated. B, in patient 12, a 2-cm right axillary tumor (which was not a chosen target tumor tracked below), from which a 3-mm punch biopsy prior to dabrafenib initiation revealed both BRAF and MEK1 mutations, became nonpalpable within 14 days of starting on dabrafenib. Despite tumor tracking not reaching partial response before DP, initial tumor response in this patient was associated with a dramatic increase in right shoulder mobility as well as decreased right breast/axillary swelling (Supplementary Fig. S2). C, in patient 28, a baseline tumor biopsy in the right side of the neck taken 4 months and 11 days prior to dabrafenib initiation was shown to harbor both BRAF and MEK1 mutations. PET scans before and after dabrafenib initiation clearly showed a rapid metabolic response at day 15 on treatment. Best overall response indicates a PR, and the same neck lesion that responded on day 15 was biopsied when it progressed.

We introduced MEK1 WT and MEK1 P124S into 2 ^V600EBRAF (Fig. 2) and 1 ^V600KBRAF (Supplementary Fig. S3) human melanoma cell lines using a lentiviral expression vector with a doxycycline-repressible promoter. MEK1 WT or P124S-regulated expression was achieved with 10 ng/mL (no expression), 0.1 ng/mL (low expression, mimicking a 1:1 endogenous vs. exogenous MEK1 expression), and 0 ng/mL (high expression, artificially maximizing any observed effect) of doxycycline in the culture media (Fig. 2A). Expression of exogenous MEK1 WT or P124S was confirmed by a FLAG epitope tag Western blot 2 days after doxycycline removal. Additionally, the relative expression of endogenous MEK1 versus total levels after exogenous MEK1 WT or P124S mutant induction was shown by MEK1-specific Western blotting. Interestingly, regulated expression of MEK1 WT or P124S at “physiologic” or high levels did not alter the downstream p-ERK levels (Fig. 2A), suggesting that ^V600EBRAF is dominant over MEK1 WT or P124S mutant in regulating cellular p-ERK levels. Because 3 of 5 patients’ melanomas harbored MEK1 exon 3 mutations concurrent with the ^V600KBRAF allele, we also showed that MEK1 P124S coexpression in the ^V600KBRAF melanoma cell line, YULAC, had no appreciable impact on cellular p-ERK levels and sensitivities to BRAFi or MEK1/2 inhibitor [MEKi (Supplementary Fig. S3A and S3B)]. Moreover, we chose M238, a cell line heterozygous for ^V600EBRAF, with regulated expression of MEK1 WT or P124S, to determine a potential impact of MEK1 P124S on p-ERK levels under reducing levels of ^V600EBRAF activity (or increasing levels of acute vemurafenib treatment for 1 hour). Regulated expression of MEK1 WT or P124S in M238 clearly had no significant impact on cellular p-ERK levels modulated by vemurafenib treatment (Fig. 2B).

Figure 2.

Regulated ^P124SMEK1 expression cannot alter p-ERK levels or sensitivity to BRAF or MEK inhibition in ^V600EBRAF melanoma cell lines. A, doxycycline-repressible expression of vector (control), FLAG-MEK1 WT, versus FLAG-MEK1 P124S in ^V600EBRAF melanoma cell lines (M229, M238). Protein lysates (48 hours post seeding at 0, 0.1, and 10 ng/mL doxycycline) were probed by Western blotting for the indicated phospho- and total protein levels. Tubulin, loading control. B, dose-dependent suppression of p-ERK levels by vemurafenib (PLX4032) in a ^V600EBRAF background with or without concurrent ^P124SMEK1. Indicated M238 stable cell lines were washed free of doxycycline for 48 hours, inducing exogenous FLAG-MEK1 WT or FLAG-MEK1 P124S expression, and treated for 1 hour with increasing doses (μM) of vemurafenib: 0 (DMSO), 0.01, 0.1, 1.0, and 10. Cell lysates were then probed for the indicated protein levels. C, stable MEK1 knockdown in the naturally occurring ^V600EBRAF/^P124SMEK1 double-mutant melanoma short-term culture (YUKSI). Protein lysates were probed for the indicated protein levels. D, impact of ^P124SMEK1 on cellular p-ERK levels in BRAF WT versus V600E backgrounds. Indicated FLAG-tagged expression constructs were transiently transfected into HEK293T cells (BRAF WT) with either pBABE-PURO (empty vector) or pBABE-PURO-^V600EBRAF. After 72 hours, cell lysates were probed for the indicated protein levels by Western blotting. E, stable cell lines (M229, top; M238, bottom) were either maintained with doxycycline (10 ng/mL) or washed and released incrementally (0.1 or 0 ng/mL) from doxycycline-mediated suppression of FLAG-MEK1 WT or FLAG-MEK1 P124S gene expression for 24 hours prior to treatment with increasing concentrations of vemurafenib/PLX4032 (black) or selumetinib/AZD6244 (red). Survival curves are shown after 72 hours of drug treatments, and data represent percent surviving cells relative to DMSO-treated controls (mean ± SEM, n = 5). The dashed line corresponds to 50% cell killing. F, the BRAF/MEK1 double-mutant melanoma short-term culture, YUKSI, was infected with either a control or shMEK1 virus and subjected to PLX4032 or AZD6244 treatments for 72 hours. G, indicated M238 stable cell lines maintained with doxycycline (100 ng/mL) or washed free of doxycycline, and seeded at single-cell density. At 24 hours after seeding, cells were treated with indicated concentrations of PLX4032. Cellular colonies were visualized by staining with crystal violet at 12 days after drug treatments. Photographs are representative of 2 independent experiments and time points.

We also identified a naturally occurring BRAF/MEK1 double-mutant melanoma short-term culture, YUKSI, and confirmed the RNA expression of the MEK1 P124S mutant allele (Supplementary Fig. S4). The ratio of WT and P124S alleles in this short-term culture is approximately 1:1, consistent with that observed in tumor tissues (Supplementary Fig. S1), and these 2 MEK1 alleles appear to be expressed at similar RNA and protein levels, with the latter extrapolation based on WT and P124S MEK1 having similar protein half-lives (Supplementary Fig. S1; data not shown). We then examined the effect of short hairpin RNA (shRNA)-mediated MEK1 knockdown on the p-ERK level (Fig. 2C). Consistent with the inducible expression data, stable knockdown of MEK1 did not have an appreciable effect on the p-ERK level, suggesting that in BRAF–mutant melanoma cell lines the MEK1 level or mutational status at P124 is not a limiting modulator of p-ERK output. To examine potential long-term effects, M238 with regulated expression of MEK1 WT or P124S was followed by cell counting for growth assessment every 4 days over a span of 20 days. Neither ^WTMEK1 nor ^P124SMEK1 expression altered the rate of population doubling regardless of expression levels (or doxycycline concentrations), suggesting once again that ^P124SMEK1 does not act in a dominant fashion over ^V600EBRAF in ERK phosphorylation or cell growth regulation in melanoma cell lines (Supplementary Fig. S5A and S5B).

When transiently overexpressed in a genetic background devoid of mutant BRAF (HEK293T), ^P124SMEK1, but not MEK1 WT, induced an increased level of cellular p-ERK (Fig. 2D), consistent with ^P124SMEK1 harboring intrinsically enhanced kinase activity toward recombinant ERK (Supplementary Fig. S6). On coexpression of ^V600EBRAF, which, as expected. led to baseline increases in p-MEK and p-ERK levels in HEK293T, ^P124SMEK1 no longer enhanced the cellular p-ERK level (Fig. 2D and Supplementary Fig. S6). Notably, although ^P124SMEK1 compared with WT MEK1 displayed increased kinase activity toward ERK1/2 in a ^WTBRAF genetic background, this difference was lost in a ^V600EBRAF background. Taken together, these data support a dominant role of ^V600E/KBRAF over ^P124SMEK1 in determining the cellular p-ERK output when these mutant alleles coexist in the same melanoma tumor or cell line.

BRAF/MEK1 double-mutant melanoma cell lines were then subjected to vemurafenib (a BRAFi) or AZD6244/selumetinib (a MEKi) titration in survival assays over 3 days (Fig. 2E). Regulated expression of ^P124SMEK1 in the background of ^V600EBRAF (Fig. 2A) or ^V600KBRAF (Supplementary Fig. S3B) and stable MEK1 knockdown in the context of a naturally occurring BRAF/MEK1 double-mutant short-term melanoma culture, YUKSI (Fig. 2C), consistently produced no significant differences in vemurafenib (Fig. 2E and F, black) or AZD6244 (Fig. 2E and F, red) sensitivities. Similarly, regulated exogenous ^P124SMEK1 expression in the context of ^V600EBRAF (M238) did not result in relative resistance to BRAF inhibition, compared with vector control or MEK1 WT in a clonogenic assay where cells were replenished with fresh vemurafenib every 2 days over 12 days (Fig. 2G). Additionally, regulated ^P124SMEK1 expression at day 8 did not alter cellular p-ERK levels in a clonogenic assay (Supplementary Fig. S5B). We also performed isobologram analysis of quantitative synergy, additivity, or antagonism using combination treatments with constant ratios of vemurafenib and AZD6244/selumetenib. Cotargeting of BRAF and MEK1 yielded strong synergy (log₁₀ confidence interval from −0.52 to −1.0) in ^V600EBRAF– mutant melanoma cell lines (Supplementary Fig. S7). However, regulated expression of ^P124SMEK1 in BRAF–mutant melanoma cell lines failed to significantly alter the degree of vemurafenib and AZD6244 synergy and did not confer cross-resistance to BRAFi and MEKi (Supplementary Fig. S7).

Another ^V600BRAF–mutant concurrent MEK1 exon 3 mutation, found in the baseline tumor and both DP tumors of patient number 27, results in an I111S substitution. In the same cell line and expression system, we compared the signaling and drug sensitivity impact resulting from the regulated expression of ^I111SMEK1 versus ^C121SMEK1, an exon 3 mutant reported to mediate acquired vemurafenib resistance (12). In contrast with ^C121SMEK1, regulated expression of ^I111SMEK1 in the ^V600EBRAF melanoma cell line M229 did not alter p-ERK levels (Fig. 3A) or sensitivity to BRAFi (Fig. 3B, left) or to MEKi (Fig. 3B, right). Consistently, exogenous expression of ^I111SMEK1 did not result in relative p-ERK resistance in response to vemurafenib treatment at increasing concentrations, whereas regulated expression of ^C121SMEK1 restored the p-ERK levels in the presence of vemurafenib (Fig. 3C). In a manner similar to ^P124SMEK1, ^I111SMEK1 also showed increased activity toward ERK in a ^WTBRAF but not in a ^V600EBRAF background (Fig. 3D). In contrast, ^C121SMEK1 conferred increased phosphorylation of ERK in both ^WTBRAF and ^V600EBRAF backgrounds (Fig. 3D), although this latter effect was more robust in a melanoma cell line (Fig. 3A and C).

Figure 3.

^I111SMEK1 expression fails to modulate p-ERK levels or melanoma sensitivity to BRAF or MEK inhibitors in the presence of ^V600EBRAF. A, doxycycline-repressible expression vector, FLAG-MEK1 I111S, or FLAG-MEK1 C121S in the ^V600EBRAF melanoma cell lines M229. Protein lysates (48 hours post seeding at 0, 0.1, and 10 ng/mL doxycycline) were probed by Western blotting for the indicated phospho- and total protein levels. Tubulin, loading control. B, M229 stable cell lines were either maintained with doxycycline (10 ng/mL) or washed and released incrementally (0.1 or 0 ng/mL) from doxycycline-mediated suppression of gene expression for 24 hours prior to treatment with increasing concentrations of vemurafenib/PLX4032 (black) or selumetinib/AZD6244 (red). Survival curves are shown after 72 hours of drug treatments, and data represent percent surviving cells relative to DMSO-treated controls (mean ± SEM, n = 5). The dashed line corresponds to 50% cell killing. C, dose-dependent suppression of p-ERK levels by vemurafenib/PLX4032 in a ^V600EBRAF background with or without concurrent ^WTMEK1, ^C121SMEK1, or ^I111SMEK1 expression. Indicated M229 stable cell lines were treated for 1 hour with increasing doses (μM) of PLX4032: 0 (DMSO), 0.01, 0.1, 1.0, and 10. Cell lysates were then probed for the indicated protein levels. D, impact of indicated MEK1 mutants on cellular p-ERK levels in BRAF WT versus V600E backgrounds. Indicated FLAG-tagged expression constructs were transiently transfected into HEK293T cells (BRAF WT) with either pBABE-PURO (empty vector) or pBABE-PURO-^V600EBRAF. After 72 hours, cell lysates were probed for the indicated protein levels by Western blotting.

DISCUSSION

Landmark studies have revealed largely mutually exclusive N-RAS, BRAF, and c-KIT activating mutations driving the MAPK/ERK pathway among human melanomas and defining therapeutically relevant melanoma subsets (16, 17). We report here the concurrence of activating mutations in BRAF and MEK1 in human melanomas and the dominant role of ^V600E/KBRAF mutants over these somatically acquired MEK1 exon 3 mutants (^I111SMEK1 and ^P124SMEK1) in determining and maintaining a critical level of ERK activation needed for the growth and survival of melanoma cells. The dominance of ^V600E/KBRAF over these MEK1 exon 3 mutants, which preexist prior to BRAFi or MEKi therapy, helps to explain the clear evidence of clinical responses of such double BRAF/MEK1–mutant melanomas to the BRAFi vemurafenib and dabrafenib (approved by the U.S. Food and Drug Administration and in advanced stages of clinical development, respectively). Thus, contrary to logical prediction, the reported MEK1 exon 3 mutations (I111S and P124S) are unlikely to be an important determinant of primary or innate sensitivity to this class of targeted agents.

The relative impact of MEK1 exon 3 mutants (I111S and P124S) versus ^V600EBRAF on p-ERK levels (Fig. 2D and 3D) may reflect the relative strength of kinase activation/loss of negative feedback and suggest a potential dependence of MEK1 exon 3 mutant hyperactivity on adequate upstream activation (i.e., MEK1 exon 3 mutations are not constitutively activating but render MEK1 more readily activated). These intriguing findings on concurrent activating mutations in BRAF and MEK1 suggest that exon 3 MEK1 mutants may subserve ERK-independent effects during melanoma progression and occur also in BRAF WT melanomas. Additionally, the trend toward a high ratio of double MEK1/^V600KBRAF mutations relative to MEK1/^V600EBRAF mutations found in this study warrants further validation. In additional tumors we have analyzed from patients with BRAF–mutant melanoma not treated with BRAFi, we have detected 3 additional MEK1 exon 3 mutants (data not shown). All 3 are P124S substitutions, concurrent with 1 ^V600RBRAF and 2 ^V600KBRAF mutations.

Aside from the setting of primary or innate drug sensitivity, a ^P124LMEK1 mutation has been reported to confer acquired resistance in one patient treated with the MEKi AZD6244/selumetinib (18). This ^P124LMEK1 mutant was proposed to mediate cross-resistance in vitro to PLX4720, a preclinical version of PLX4032/vemurafenib. However, the short-term culture derived from this MEK1 mutant, AZD6244-resistant melanoma, termed M307, was later found to harbor high levels of COT expression, which has been proposed more recently as a mediator of acquired BRAFi resistance in melanoma (6). Thus, it is possible that COT overexpression, rather than MEK1 mutation, accounts for BRAFi resistance in this melanoma culture. Alternatively, ^P124LMEK1 may be more catalytically active than ^P124SMEK1, with the latter likely being the most common MEK1-mutant allele detected concurrently with ^V600BRAF mutants without any prior BRAFi or MEKi exposure or selective pressure. Moreover, another MEK1 exon 3 allele, C121S, has been proposed as a determinant of acquired resistance to vemurafenib (12). In vitro data from our study corroborate this conclusion, although we have not yet detected the ^C121SMEK1 allele in DP tumors from our patients.

In our clinical series, MEK1 mutations were detected in acquired resistant or DP tumors only when the same mutations were also observed in the baseline melanoma tumors. Thus, these preexisting MEK1 mutations cannot account for acquired drug resistance. Notably, none of these patients had been exposed to MEKi prior to starting on the BRAFi therapy. The objective response rate of these 5 patients with preexisting double BRAF/MEK1 mutations is in the same range of the 53% objective response rate in a large series of patients treated with vemurafenib (19).

In our work, regulated expression of ^P124SMEK1 and ^I111SMEK1 to mimic endogenous expression levels (1:1 ratio of mutant to WT MEK1) consistently produced no appreciable effect on cellular p-ERK levels and vemurafenib sensitivity in several genetic backgrounds harboring either ^V600EBRAF or ^V600KBRAF alleles. Outside the context of ^V600E/KBRAF (e.g., in HEK293T), however, MEK1 P124S and I111S mutants have demonstratively higher levels of kinase activity toward ERK compared with MEK1 WT. Additionally, when double BRAF/MEK1-mutant melanoma cell lines were exposed acutely to vemurafenib, resulting in reduced ^V600EBRAF activity, the activating MEK1 P124S and I111S mutants still could not upregulate (or reactivate) cellular p-ERK levels, suggesting continued ^V600E/KBRAF oncogene addiction. Together, these data preclude a critical role of somatic MEK1 exon 3 mutations preexisting prior to BRAFi therapy in conferring primary resistance to BRAF inhibitors. These data also caution against relying on detection of every MEK1 exon 3 mutant allele as a biomarker of acquired BRAFi resistance.

Emerging clinical data with combined BRAF and MEK inhibitors (20) showing preliminary safety and a high response rate of BRAF–mutant melanomas suggest improved durability of clinical response (compared with single-agent therapy). In this context, our data showed a strong degree of BRAFi and MEKi synergy in both BRAF single- and BRAF/MEK1 double-mutant genotypes, supporting the utility of such a combinatorial approach as an upfront therapy or a regimen to overcome defined mechanisms of acquired BRAFi resistance (7, 8, 10). The coexistence of BRAF and MEK1-mutant alleles in the same tumor cell and its validation in an additional cohort of metastatic melanoma patients (data not shown) suggest potential ERK-independent roles of activating MEK1 mutants. Potential ERK-independent roles of somatic MEK1 mutants in melanoma pathogenesis would further support the combined BRAF and MEK inhibition therapeutic approach.

In conclusion, we identified a subset of BRAF–mutant melanomas that harbors concomitant MEK1 exon 3 mutations. Although the relevance of these concurrent mutations to multistep melanoma progression is at present unclear, contrary to expectation, their coexistence cannot preclude a clinical response to BRAF inhibitors. Data from cell line modeling support the BRAF mutant as dominant over the MEK1 mutants described herein with regard to determining cellular p-ERK output at a level critical for melanoma growth and survival. Together, the present study helps explain why preexisting MEK1 exon 3 mutations do not determine sensitivities of BRAF–mutant melanomas to the BRAF inhibitors vemurafenib/PLX4032 or dabrafenib/GSK2118436.

METHODS

Patients and Samples

Patients participated in the vemurafenib phase I dose-escalation study (NCT00405587), vemurafenib phase II study (NCT00949702), or the dabrafenib phase I/II study (NCT00880321). All patients had ^V600E/KBRAF mutation-positive, previously treated metastatic melanoma (none with MAPK-targeted drugs) and received either 960 or 1120 mg of vemurafenib or 35, 70, or 150 mg of dabrafenib orally twice daily. All consented to the genetic analysis of their tissue biopsies or samples. The clinical trials included optional biopsies at baseline or upon DP, and we selected 31 patients from whom samples from a biopsy at baseline, or both at baseline and DP, were available for analysis.

Genomic DNA and cDNA Sequencing

gDNAs were isolated using the Flexi Gene DNA Kit (QIAGEN) or the QIAamp DNA FFPE Tissue Kit. MEK1 exon 3 was amplified using the forward (CCTGTTTCTCCTCCCTCTACC) and the reverse (ACACCCACCAGGAATACTGC) primers. MEK2 exon 3 was amplified using the forward (TTGACCACTGTTGGGAACGCC) and the reverse (TCTGTTCCGTGGAGGCCCTG) primers. Total RNA was extracted using the RiboPure Kit (Ambion), and reverse-transcription reactions were performed using the SuperScript First-Strand Synthesis System (Invitrogen). PCR products were purified using QIAquick PCR Purification Kit (QIAGEN) followed by bidirectional sequencing using BigDye v1.1 (Applied Biosystems) in combination with a 3730 DNA Analyzer (Applied Biosystems).

Cell Culture, Constructs, Infections, and Transfections

All cell lines were maintained in DMEM with 10% or 20% heat-inactivated FBS (Omega Scientific), 2 mmol/L glutamine in a humidified, 5% CO₂ incubator, and 10 ng/mL doxycycline and/or puromycin, when applicable. Wild-type and mutant MEK1 were cloned, epitope-tagged, sequence-verified, and cloned into the doxycycline-repressible lentiviral vector pLVX-Tight-Puro (Clontech, Inc.). Knockdown of MEK1 was achieved using MISSION shRNA lentiviral transduction particles [clone IDs 455, 1163, 612, 1015, 2-753 (Sigma)]. ^V600EBRAF construct was purchased from Addgene. HEK293T cells were transfected using Lipofectamine2000 (Invitrogen).

Drug Sensitivity, Protein Detection, and Kinase Assay

Cell proliferation experiments were performed in a 96-well format (5 replicates) and drug treatments initiated at 24 hours post-seeding for 72 hours. Stocks and dilutions of PLX4032/vemurafenib (Plexxikon) and AZD6244/selumetenib (Selleck Chemicals) were made in dimethyl sulfoxide (DMSO). Cells were quantified using CellTiter-GLO Luminescence (Promega) following the manufacturer’s recommendations or by counting of trypan blue–positive cells following trypsinization. Clonogenic assays were performed by plating cells at single-cell density in 6-well plates and provided fresh media, doxycycline (if applicable) and PLX4032/vemurafenib (vs. DMSO) every 2 days. Colonies were then fixed by 4% paraformaldehyde and stained with crystal violet at 0.05%. Cell lysates for Western blotting were made in RIPA (Sigma) with protease (Roche) and phosphatase (Santa Cruz Biotechnology) inhibitor cocktails. Western blots were probed with antibodies against p-ERK1/2 (T202/Y204), total ERK1/2, MEK1, FLAG (Cell Signaling Technology), and tubulin (Sigma). In vitro kinase assays were performed as described (7) except recombinant ERK is now used as a substrate.

Study Oversight

Data were generated and collected by the study investigators and analyzed in collaboration between the senior authors, who vouch for the completeness and accuracy of the data and analyses. The corresponding author prepared the initial draft of the manuscript. All the authors made the decision to submit the manuscript for publication.

Data Processing

Statistical analyses were performed using InStat 3 Version 3.0b (GraphPad Software); graphical representations using DeltaGraph or Prism (Red Rock Software); and combination index calculation using CalcuSyn V2.1 (Biosoft).