SEMA6A/RhoA/YAP axis mediates tumor-stroma interactions and prevents response to dual BRAF/MEK inhibition in BRAF-mutant melanoma

For SEMA6A protein expression two patient cohorts were used; the RECI1 cohort, including 59 BRAF^V600E and 53 BRAF-wt melanoma patients treated at the Regina Elena National Cancer Institute, was used to compare SEMA6A expression in BRAF-mut versus BRAF-wt tumors. Bioptic specimens were collected from the metastatic lesions of involved patients. The RECI2 cohort was used to assess the predictive role of SEMA6A expression in 14 BRAF-mut melanoma patients admitted to dabrafenib+trametinib at Regina Elena National Cancer Institute. Based on recurrence-free interval, RECI2 patients were categorized as short-term and long-term responders; a cut-off value of 12 months was established. Information on clinical and histopathological features, anti-tumoral therapies and related outcomes were retrieved from patients’ medical records by specifically trained research assistants; written informed consent from all patients was obtained in accordance with the Declaration of Helsinki. TCGA (DOI: https://​doi.​org/​10.​1016/​j.​cell.​2018.​03.​022) and DFCI (DOI: https://​doi.​org/​10.​1038/​s41591-019-0654-5) melanoma cohorts were used to investigate the prognostic role of SEMA6A in BRAF-mut melanoma. Melanoma cases were referred to as “early” when stage was 0, I, and II, and “advanced” when stage was III and IV. Genomic and clinical data from the two cohorts were downloaded from cBioportal. Only patients with BRAF mutation were considered for the analyses. RNA-seq data were obtained from Firebrowse (http://​firebrowse.​org/​). SEMA6A expression from RNA-seq data was quantified through TPM (Transcripts per Million) normalization. Patients were classified as SEMA6A High and Low expression by calculating the relative TPM value for each sample in the patient cohort (N = 20).

Formalin-fixed paraffin-embedded sections were analyzed as described [41]. Antigen retrieval was performed at 96 °C (10 mM/L citrate buffer, pH 6) for 20 min. Sections from specimens of RECI1 and RECI2 cohorts were incubated with the primary antibody anti-SEMA6A 1:50 (HPA031265; SIGMA-Aldrich, St. Luis, MO USA) for 30 min at room temperature. Immunoreactions were revealed by Bond Polymer Refine Detection Kit according to manufacturer’s procedure (Leica Biosystems) in an automated autostainer Bond III Leica Biosystems. Diaminobenzidine was used as chromogenic substrate. Microscope Nikon ECLIPSE 55i with digital camera HESP Technology was used. Scale bars 30 μm. The study was reviewed and approved by the ethical committee of Regina Elena National Cancer Institute, and informed consent was obtained from all patients. For IHC, the secondary antibody was used as internal control.

The Human Fibroblast BJ were obtained from the American Type Culture Collection (ATCC) and maintained in DMEM medium supplemented with 10% FBS and 1% penicillin/streptomycin (Gibco, Life Technologies, Milan, Italy). BRAF-wt/NRAS-wt ME1007 and BRAF-wt/NRAS-mut ME4405, derived from lymph node metastases, and BRAF-wt/NRAS-mut 2/17 and BRAF^V600E/NRAS-wt 2/59, derived from subcutaneous metastasis, were isolated from surgical specimens of melanoma patients, not previously subjected to chemotherapy and admitted to Fondazione IRCCS Istituto Nazionale dei Tumori, Milan (Sensi M et al., 2006; Daniotti M et al., 2004). M14 and C32 cell lines were obtained from the American Type Culture Collection (ATCC).

To obtain an inducible SEMA6A depletion, 2/59 cells were transduced using shERWOOD UltramiR lentiviral inducible shRNA (pZIP-TRE3G) according to the manufacturer’s guidelines (Transomic Technologies, Huntsville, AL USA). Stable cells were generated by antibiotic selection with 1 μg/mL puromycin for 14 days. We obtained a control cell line by using a non-specific shRNA (2/59 shCtrl) and two cell lines inducible for SEMA6A depletion by using two specific shRNAs (2/59 shSEMA6A A3 and H2). To induce shRNAs and GFP reporter gene expression, cells were treated with doxycycline at 0,5 μg/mL. All melanoma cell lines were maintained in RPMI medium supplemented with 10% FBS and 1% penicillin/streptomycin (Gibco, Life technologies, Milan, Italy). All cell lines were periodically tested for mycoplasma contamination.

For co-cultures, BJ were plated and 3 days after 2/59 shCtrl, A3 and H2 were added to BJ monolayer cultures without discard the culture medium.

For treatments in vitro, trametinib and dabrafenib (Selleckchem, Houston, TX, USA) were used at 5 nM and 0,1 μM concentrations respectively. The drugs were stored as stock solutions at − 20 °C and diluted just before use.

Anti-Sema6A (ab72369) was from Abcam (Cambridge, UK), anti-Lamin A (#86846), anti-Akt (#9272), anti-phospho-ser473-Akt (#9271), anti-Erk1/2 (#9102), anti-phospho-Thr202/Tyr204-Erk 1/2 (#9101), anti-YAP (#12395), anti-phospho-ser127-YAP (#13008), anti-p65 (#8242), anti-phospho-ser536-p65 (#3033), anti-Tubulin (#2125) were from Cell Signaling (Danvers, MA USA), anti-Hsp-70 (ab-83,392) and anti-GAPDH (ab-81,594) were from Immunological Sciences (Rome, Italy). HRP-conjugated secondary antibodies were from Bio-Rad (Hercules, CA USA).

All cell lines, before and after treatment, were lysed using RIPA buffer, analyzed by SDS-PAGE and probed (WB) with antibodies of interest and secondary HRP-conjugated antibodies. Signals were detected by Amersham ECL Prime Western Blotting Detection Reagent (GE Healthcare Life Sciences, Chicago, IL USA) and by LuminataTM Classico Western HRP substrate (Millipore, Burlington, MA USA). Same amount of total protein from three independent experiments were pooled and analyzed. Each experiment was repeated at least three times.

Nucleic and cytoplasmic fractions, derived from 2/59 shCtrl and shSEMA6A A3 and H2 cells 72 h post-induction, untreated or treated with RhoA activator I, were obtained using NE-PER Nuclear and Cytoplasmic Extraction Reagents (#78835; Thermo Scientific, Rockford IL USA) according to the manufacturer’s guidelines. Each experiment was repeated at least three times.

2/59 shCtrl and shSEMA6A A3 and H2 cells were plated on 100 mm dishes or poly-l lysine coated slides 24 h post-induction. After 48 h cells were treated with 1 U/mL RhoA activator I, according to the manufacturer’s guidelines (#CN01; Cytoskeleton, Denver, CO USA). The levels of RhoA activation in untreated or treated cells were analyzed by RhoA activation assay (#BK036; Cytoskeleton, Denver CO USA) performed according to the manufacturer’s guidelines.

For immunofluorescence experiments, cells plated at a subconfluent state, untreated or treated with RhoA activator I as described above, were stained with anti-YAP (sc-376,830; Santa Cruz, Dallas TX, USA) or Alexa Fluor 555 Phalloidin (#8953; Cell Signaling, Danvers, USA) and counterstained with Hoechst to highlight nuclei (SIGMA-Aldrich, St. Luis, MO USA). 2/59 and ME4405 cells plated on poly-l lysine coated slides were treated with trametinib, dabrafenib and their combination, and stained as above. The quantification of YAP localization and stress fiber positive cells was performed by inspecting at least five fields/slide using a 60X magnification in three independent experiments. The results were reported as percentage of whole number of counted cells.

Microscope OLYMPUS BX53 was used to evaluate fluorescence. Scale bars 10 μm. Each assay was carried out in triplicate and repeated at least three times.

Following doxycycline induction for 24 h, 2/59 shCtrl and shSEMA6A A3 and H2 cells plated on 24-well plates were treated with trametinib, dabrafenib and/or their combination at specified concentrations. Seventy-two hours later the cells were fixed with 4% formaldehyde for 15 min and stained with 0.1% crystal violet solution in 10% ethanol for 40 min. At the end of incubation the medium was removed, cells were allowed to dry at room temperature and dissolved in isopropanol, and optical density was measured at 570 nm using an ELISA plate reader. Each assay was carried out in quadruplicate and repeated at least three times.

BJ were plated on 12-well plates or 60 mm dishes and 3 days later shCtrl, A3 and A H2 2/59, following a 72 h-induction, were added to BJ monolayer cultures without discard the culture medium. The day after cells were treated with trametinib, dabrafenib or their combination.

After 48 h melanoma cells seeded on 60 mm dishes were separated from the fibroblasts by BD FACSMelody Cell Sorter, gating on the GFP signal and collecting only clearly GFP positive melanoma cells. The purity of the FACS-separated melanoma fractions (i.e. no contamination with fibroblasts) is guaranteed since the gating was stringent. Total cell lysate extracted from sorted melanoma cells were analyzed by western blot.

Co-culture performed on 12-well plates were incubated in IncuCyte® and monitored non-invasively by a planned acquisition of images every 12 h for 156-180 h. Images were analyzed by software supplied by Sartorius. Each experiment was carried out in triplicate and repeated at least three times.

We have previously shown preferential expression of SEMA6A in BRAF-mut cell lines and melanoma lymph-node metastases compared with BRAF-wt ones [16]. We first verified the association of SEMA6A protein expression with BRAF mutation in the RECI1 cohort, which includes 112 metastatic advanced melanoma patients surgically treated at the Regina Elena National Cancer Institute. SEMA6A protein expression was measured by immunohistochemistry (IHC) and assigned an intensity score (IS) value ranging from 0 to 3. We then dichotomized SEMA6A into two categories: “low” if IS was 0 or 1 and “high” if score was 2 or 3. Table 1 shows the main patient- and tumor-related characteristics. Briefly, 59 patients (52,7%) were affected by BRAF-mut disease and 53 (47,3%) were BRAF-wt. SEMA6A was high in 36 (32,1%) and low in 76 (67,9%) patients. As shown in Fig. 1A and in representative IHC panels of Fig. 1B, we found a statistically significant association between high SEMA6A and BRAF mutation (Pearson’s Chi-square test p < 0,0001) (Fig. 1A). Notably, 100% of samples with SEMA6A IS 3 were BRAF-mut. In addition, we performed SEMA6A IHC in subsequent metastatic lesions at diagnosis and at progression in three patients. As shown in Table 2, SEMA6A levels progressively increased, suggesting that SEMA6A may have prognostic relevance in BRAF-mut melanoma. To verify this hypothesis, we analyzed mRNA data from TCGA and DFCI large datasets; advanced and early melanoma patients were analyzed separately. When focusing on patients with advanced melanoma (n = 149), patients with high SEMA6A expression had reduced PFS and OS than those with low SEMA6A (PFS: 6.76 months vs. 19.50 months; Log-Rank p = 0.013; OS: 31.6 months vs. 69.1 months; Log-Rank 0.0008) (Fig. 1C and D). In patients with early melanoma (n = 109), high SEMA6A was significantly associated with reduced relapse-free survival (RFS) (40.5 months vs. 55.5 months; Log-Rank p = 0.031) (Fig. 1E). Overall these data support the involvement of SEMA6A in the biology of BRAF-mut melanoma and indicate that SEMA6A is a prognostic indicator in this subset of patients.

We have previously identified SEMA6A as a BRAF-mut-associated protein that might contribute to actin cytoskeleton remodeling [16]. Our previous findings indicate that the depletion of SEMA6A by siRNA has dramatic effect in terms of cell death; thus, to unravel the molecular mechanism by which SEMA6A regulates actin stress fibers dynamics, we aimed to exploit a cell model that would allow us to modulate SEMA6A expression and avoid massive cell death. For this purpose, the BRAF-mut melanoma cell line 2/59 was engineered for inducible silencing of SEMA6A expression and induction of GFP reporter gene expression. We obtained three polyclonal cell populations: one shCtrl and two shSEMA6A (A3 and H2). Following induction, the expression of SEMA6A was efficiently downregulated in A3 and H2 compared with shCtrl cells (Fig. 2A), resulting in 20-30% reduction in the number of viable melanoma cells (Fig. 2B). In agreement with our previous findings, the depletion of SEMA6A in 2/59 cell model induced actin cytoskeleton remodeling and loss of actin stress fibers (Fig. 2C).

Next, we investigated the involvement of Rho small GTPases and downstream YAP in SEMA6A-induced actin cytoskeletal remodeling. First, we measured the protein levels of Rho GTPs in 2/59 cells, and found the expression of RhoA but not of RhoC small GTPase (data not shown); thus we analyzed the level of active RhoA-GTP upon induction of SEMA6A silencing. As shown in Fig. 2D, SEMA6A depletion caused a reduction of RhoA activity compared with control cells. The densitometric analysis of RhoA-GTP normalized to total RhoA revealed a statistically significant reduction of RhoA activity upon SEMA6A depletion in both H2 and A3 cell populations (Fig. 2E). Next, we found higher levels of YAP phosphorylation following SEMA6A-depletion (Fig. 2A). Moreover, western blot analysis of cytoplasmic and nuclear fractions of control and SEMA6A-depleted cells revealed increased phosphorylation levels of YAP and its cytoplasmic accumulation in A3 and H2 compared with shCtrl cells (Fig. 2F, upper panels). The densitometric analysis of phosphorylated YAP normalized to total YAP confirms increase of YAP phosphorylation in the cytoplasmic fraction of both H2 and A3 SEMA6A-depleted cells (Fig. 2G). YAP cytoplasmic accumulation following SEMA6A depletion was confirmed by immunofluorescence (Fig. 2H and Supplementary S1A, upper panels), and was associated with loss of stress fibers (Fig. 2I, Supplementary S1B, upper panels, and L). We then investigated whether induction of RhoA activity by a specific activator can rescue RhoA-YAP signaling in SEMA6A-depleted cells. To this aim, shCtrl, A3 and H2 cells were treated with RhoA activator and analyzed as above. We observed a significant induction of RhoA activity (Fig. 2D and E) and the reduction of YAP cytoplasmic localization in shCtrl cells compared with their untreated counterpart (Fig. 2F, H, and Supplementary S1A). By contrast, treatment of both A3 and H2 SEMA6A-depleted cell populations with RhoA activator prevented activation of RhoA (Fig. 2D and E) and reduction of YAP cytoplasmic localization (Fig. 2F, H, and Supplementary S1A, lower panels), indicating that SEMA6A is required for downstream activation of RhoA/YAP axis. In agreement, we observed approximately 80% reduction of stress fibers in both A3 and H2 SEMA6A-depleted cells compared with shCtrl cells regardless of stimulation with RhoA activator (Fig. 2I, L, and Supplementary S1B).

These data show that SEMA6A sustains the activity of RhoA that in turn induces YAP nuclear localization. In accordance, SEMA6A depletion is associated with loss of actin stress fibers.

YAP-induced actin cytoskeleton remodeling has been recently reported as a mechanism of resistance to BRAF inhibitors [48, 49]. Thus, we investigated the activity of SEMA6A/RhoA/YAP axis in response to targeted therapy by treating BRAF-mut cell lines 2/59, M14, and C32 with the BRAF inhibitor dabrafenib, the MEK inhibitor trametinib, and the combination of the two drugs. All treatments inhibited the phosphorylation of ERK 1/2 (p-ERK) as expected, and induced a marked increase of SEMA6A expression in all the three BRAF-mut cell lines (Fig. 3A, left panels). The increase of SEMA6A expression was accompanied by a mild but reproducible induction of RhoA activity (Fig. 3B and C) and reduced phosphorylation of YAP (Fig. 3A, left panels). In order to assess whether SEMA6A/RhoA/YAP induction by BRAF and MEK inhibition specifically occurs in a BRAF-mut context, we administered same treatments to 2/17 and ME4405 (BRAF-wt/NRAS-mut), and ME1007 (BRAF-wt/NRAS-wt) cell lines. As expected, p-ERK were inhibited by trametinib and paradoxically induced by dabrafenib, as previously described [51], in NRAS-mut cell lines (Fig. 3A, central panels) and were affected only to a marginal extent in the BRAF-wt/NRAS-wt ME1007 cells (Fig. 3A, right most panel). The treatments did not induce SEMA6A expression, nor the levels of phosphorylated YAP were affected in the BRAF-wt cell lines analyzed (Fig. 3A right panels). These results indicate that BRAF inhibition by dabrafenib, MEK inhibition by trametinib, and dual BRAF/MEK inhibition by dabrafenib+trametinib induce the SEMA6A/RhoA/YAP axis in BRAF-mut but not in BRAF-wt melanoma cells.

Then, we investigated whether SEMA6A expression level is associated with the sensitivity of BRAF-mut cell lines to MAPK inhibition. First, we treated 2/59 and M14 cell lines expressing respectively high (SEMA6A^hi) and low (SEMA6A^low) SEMA6A and comparable levels of p-ERK, with different doses of dabrafenib. As shown in Fig. 3D, SEMA6A^hi 2/59 cells showed significantly reduced sensitivity to BRAF inhibition compared with SEMA6A^low M14. Second, we analyzed by western blot analysis A375 melanoma cell line sensitive to dabrafenib and its dabrafenib-resistant counterpart. We found increased expression of SEMA6A in resistant A375, in association with p-ERK reduction (Fig. 3E). Moreover, YAP expression level was significantly higher in resistant compared with sensitive A375 cells, whereas the two cell lines showed comparable levels of p-YAP, indicating a lower p-YAP/YAP ratio in resistant cells. These data support the involvement of SEMA6A in the mechanisms of resistance of melanoma cells to BRAF inhibition.

To investigate the dynamics of SEMA6A/RhoA/YAP axis activation and actin cytoskeleton remodeling in response to BRAF and MEK inhibition, we treated BRAF-mut 2/59 and BRAF-wt ME4405 cells up to 7 days and analyzed them by immunofluorescence. Following 48 h dabrafenib+trametinib, BRAF-mut 2/59 cells showed YAP nuclear translocation and loss of actin stress fibers (Fig. 3F, upper panels, and Fig. 4A, left upper panels). Afterwards, surviving cells gradually re-organized their actin cytoskeleton, as shown by images taken 96 h post treatment (Fig. 3F, middle panels), and rescued stress fibers/cytoskeleton organization and YAP cytoplasmic localization after 7 days treatment (Fig. 3F, lower panels and Fig. 4A, left lower panels). Loss of stress fibers and YAP nuclear translocation were mainly induced in 2/59 cells by dabrafenib and dabrafenib+trametinib combination, and only to a lesser extent by trametinib (Fig. 4A, left upper panels). Figure 4B shows the increase in percentage of nuclear YAP upon 48 h and 7 days trametinib (34,4% and 22,2%), dabrafenib (81,6% and 71,3%), and their combination (95,2% and 76,5%) compared with untreated cells (5% and 3,6%). By contrast, in BRAF-wt ME4405 cells, the actin cytoskeleton was not affected by dual BRAF/MEK inhibition (Fig. 4A, right panels), and trametinib only induced a mild YAP nuclear translocation (6,7% at 48 h and 7 days) (Fig. 4A, right panels, and Fig. 4B, right graphs). These data suggest that, in response to dabrafenib and dabrafenib+trametinib, BRAF-mut melanoma cells activate the axis SEMA6A/RhoA/YAP as a compensatory mechanism aimed to sustain actin cytoskeleton remodeling.

Next, we assessed the viability of shCtrl and shSEMA6A cells following drugs treatments. Dabrafenib and dabrafenib+trametinib treatments induced a marked reduction in the number of viable shCtrl cells (Supplementary Fig. S2). As expected, SEMA6A depletion per se reduced the number of viable A3 and more efficiently that of H2 cells compared with shCtrl cells (Supplementary Fig. S2, compare black bars of untreated cells); however, it did not further affect the number of viable cells following drug treatments (Supplementary Fig. S2, compare grey bars of treated cells), suggesting that some other factors may be required for SEMA6A/RhoA/YAP axis to regulate sensitivity to targeted therapies in patients.

Sorrounding fibroblasts have been previously reported to promote the growth of melanoma cells in vitro [52]; thus we asked whether microenvironment might play a role in SEMA6A signaling, and exploited a co-culture condition in which melanoma cells are cultured on a monolayer of BJ human fibroblasts. Fibroblasts-cocultured (from hereon, co-cultured) shCtrl had a markedly higher proliferation rate compared with the same cells cultured in the absence of fibroblasts (from hereon, mono-cultured). SEMA6A depletion nearly abolished the proliferative stimulus conferred by fibroblasts (Fig. 5A); indeed, the fibroblasts-dependent fold growth induction was significantly reduced in SEMA6A-depleted cells compared to shCtrl cells (Fig. 5B).

In addition, we sorted GFP-positive shCtrl and SEMA6A-depleted cell populations from co-cultures and analyzed the phosphorylation levels of YAP and of survival and proliferation effectors AKT, P65, and ERK in co-cultured and mono-cultured cells. In agreement with their higher proliferation rate, co-cultured shCtrl cells showed higher phosphorylation levels of YAP, AKT, P65, and ERK compared with their mono-cultured counterpart (Fig. 5C). SEMA6A depletion increased the levels of phosphorylated and total YAP in mono-cultured condition, as expected, and also in co-cocultured cells. Of relevance, SEMA6A depletion reduced the phosphorylation levels of AKT, P65, and ERK only in the presence of sorrounding fibroblasts (Fig. 5C).

These findings recapitulate previous evidence indicating that surrounding fibroblasts sustain the growth of melanoma cells by the activation of pro-survival (PI3K/AKT) and pro-proliferative (MAPK, ΝFκΒ) pathways, and indicate that SEMA6A plays a crucial role as a mediator in this process.

Tumor microenvironment is known to increase resistance to BRAF inhibitors [32, 33]. Since SEMA6A supports surrounding fibroblasts-induced growth of melanoma cells, we asked whether SEMA6A might affect the efficacy of targeted therapy in co-cultured condition. Thus, we treated mono-cultured and co-cultured shCtrl and SEMA6A-depleted A3 and H2 cell populations with dabrafenib or dabrafenib+trametinib combination, and periodically monitored cell viability over time up to 156 h.

First, we observed reduced efficacy of dabrafenib in co-cultered cells. Indeed, as shown in Fig. 5D, the proliferation rate of both mono-cultured and co-cultured shCtrl cells increased irrespective of treatments; dabrafenib rather conferred a paradoxical growth advantage in co-cultured condition (Fig. 5D and E). Then, we found that in SEMA6A-depleted cells both dabrafenib and dabrafenib+trametinib treatments dramatically reduced cell proliferation only in the co-cultured condition (Fig. 5F). As shown in Fig. 5G, 156 h post-dabrafenib and dabrafenib+trametinib, the number of co-cultured viable SEMA6A-depleted cells was significantly reduced compared with untreated cells. Data from co-cultured shCtrl and SEMA6A-depleted A3 and H2 cells, reported as fold change number of treated/untreated viable cells, show that SEMA6A depletion rescued the efficacy of both dabrafenib and dabrafenib+trametinib (Fig. 5H); indeed, 180 h post-treatment the viability of treated/untreated SEMA6A-depleted cells was significantly reduced as compared with shCtrl cells (Fig. 5I).

Next, we analyzed the levels of phosphorylated and total AKT, ERK, and YAP of both shCtr and shSEMA6A cells mono-cultured and sorted from coculture, following dabrafenib+trametinib treatment. The combination treatment induced SEMA6A expression in mono-cultured shCtrl cells, as expected, and also in co-cultured shCtrl cells. Of relevance, upon dabrafenib+trametinib, SEMA6A depletion reduced the levels of phosphorylated AKT and ERK compared with treated shCtrl cells only in co-cultured condition (Fig. 5L).

Overall these data indicate that surrounding fibroblasts protect shCtrl cells from the inhibitory effect of both BRAF and dual BRAF/MEK inhibition. SEMA6A depletion abrogates this protective effect and rescues the efficacy of both dabrafenib and dabrafenib+trametinib, indicating that SEMA6A plays a key role in mediating fibroblasts-induced insensitivity to targeted therapy.

Based on the results we obtained in vitro, in the presence of surrounding fibroblasts as a microenvironment-mimicking condition, we investigated the predictive potential of SEMA6A on dual BRAF/MEK inhibition in vivo. To this aim, we analyzed the RECI2 cohort, which includes 14 BRAF-mut advanced melanoma patients admitted to dabrafenib+trametinib therapy at the Regina Elena Cancer Institute. Based on a cut-off value of 12 months of progression-free interval, eight of these patients were considered long-term responders and six short-term responders. Table 3 shows clinicopathological charachteristics of RECI2 cohort. SEMA6A expression was analyzed by IHC and reported both as percentage of positive cells and IS. Patients with high SEMA6A expression had a significantly reduced median PFS and OS compared with those with low SEMA6A (PFS: 10 months vs. 60 months; Log-Rank p = 0.001; OS: 27.5 months vs. not reached; Log-Rank p = 0.021) (Fig. 6A and B), indicating that SEMA6A might be a good candidate predictor of low efficacy of dual BRAF/MEK inhibition by dabrafenib+trametinib in BRAF-mut melanoma patients.