SEMA6A is highly expressed and has prognostic relevance in BRAF-mut melanoma patients.
We have previously shown preferential expression of SEMA6A in BRAF-mut cell lines and melanoma lymph-node metastases compared with BRAF-wt ones (12). 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 Figure 1A and in representative IHC panels of Figure 1B, we found a statistically significant association between high SEMA6A and BRAF mutation (Pearson’s Chi-square test p<0,0001) (Figure 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) (Figures 1C and 1D). 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) (Figure 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.
SEMA6A regulates actin cytoskeleton remodeling by activating the RhoA/YAP axis in BRAF-mut melanoma cells.
We have previously identified SEMA6A as a BRAF-mut-associated protein that might contribute to actin cytoskeleton remodeling (12). 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 (Figure 2A), resulting in 20-30% reduction in the number of viable melanoma cells (Figure 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 (Figure 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 Figure 2D, SEMA6A depletion caused a reduction of RhoA activity compared with control cells. The densitometric analysis of pooled immunoblot data from both H2 and A3 revealed a statistically significant reduction of RhoA activity upon SEMA6A depletion (Figure 2E). Next, we found higher levels of YAP phosphorylation following SEMA6A-depletion (Figure 2A). Moreover, Wester Blot analysis of cytoplasmic and nuclear fractions of control and SEMA6A-depleted cells revealed increased phosphorylation levels of YAP and its cytoplasmic retention in A3 and H2 compared with shCtrl cells (Figure 2F, upper panels). The densitometric analysis of pooled data from both H2 and A3 shows increase of YAP phosphorylation in the cytoplasmic fraction of SEMA6A-depleted cells (Figure 2G). YAP cytoplasmic accumulation following SEMA6A depletion was confirmed by immunofluorescence (Fig. 2H, upper panels) and was associated with loss of stress fibers (Figure 2I, upper panels). 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 (Figure 2D and 2E) and the reduction of YAP cytoplasmic localization (Figure 2F and Figure 2H, upper panels) in shCtrl cells compared with their untreated counterpart. By contrast, treatment of SEMA6A-depleted cells with the RhoA activator did not rescue the level of active RhoA-GTP (Figure 2D and 2E), YAP cytoplasmic localization (Figure 2F, lower panels, and Figure 2H, lower panels), and actin cytoskeleton organization (Figure 2I, lower panels), indicating that SEMA6A is required for downstream activation of RhoA/YAP axis. These data show that SEMA6A regulates actin cytoskeleton remodeling by sustaining the activity of RhoA that in turn induces YAP nuclear localization.
Dual BRAF/MEK inhibition induces SEMA6A/RhoA/YAP axis.
YAP-induced actin cytoskeleton remodeling has been recently reported as a mechanism of resistance to BRAF inhibitors (44, 45). 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 (Figure 3A, left panels). The increase of SEMA6A expression was accompanied by a mild but reproducible induction of RhoA activity (Figure 3B and 3C) and reduced phosphorylation of YAP (Figure 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 (46) in NRAS-mut cell lines (Figure 3A, central panels) and were affected only to a marginal extent in the BRAF-wt/NRAS-wt ME1007 cells (Figure 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 (Figures 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.
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 hours dabrafenib+trametinib, BRAF-mut 2/59 cells showed complete disorganization of actin cytoskeleton and YAP nuclear translocation (Figure 3D, left upper panels, and Figure 4A, left upper panels). Afterwards, surviving cells gradually re-organized their actin cytoskeleton, as shown by images taken 96 hours post treatment (Figure 3D, left middle panels), and rescued stress fibers/cytoskeleton organization and YAP cytoplasmic localization after 7 days treatment (Figure 3D, left lower panels and Figure 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 (Figure 4A, left upper panels). Figure 4B shows the increase in percentage of nuclear YAP upon 48 hours 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 (Figure 4A, right panels), and trametinib only induced a mild YAP nuclear translocation (6,7% at 48 hours and 7 days) (Figure 4A, right panels, and Figure 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 Figure S1). As expected, SEMA6A depletion per se reduced the number of viable A3 and more efficiently that of H2 cells compared with shCtrl cells (Supplementary Figure S1, compare black bars of untreated cells); however, it did not further affect the number of viable cells following drug treatments (Supplementary Figure S1, 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.
SEMA6A is a mediator of surrounding fibrobalsts-induced melanoma cell growth.
Sorrounding fibroblasts have been previously reported to promote the growth of melanoma cells in vitro (48); 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 (Figures 5A); indeed, the fibroblasts-dependent fold growth induction was significantly reduced in SEMA6A-depleted cells compared to shCtrl cells (Figure 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 (Figure 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 (Figure 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, NFkB) pathways and indicate that SEMA6A plays a crucial role as a mediator in this process.
SEMA6A depletion rescues the efficacy of both BRAF and dual BRAF/MEK inhibition in fibroblasts-cocultured melanoma cells.
Tumor microenvironment is known to increase resistance to BRAF inhibitors (28, 29). 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 hours.
First, we observed reduced efficacy of dabrafenib in co-cultered cells. Indeed, as shown in Figure 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 (Figure 5D and 5E). Then, we found that in SEMA6A-depleted cells both dabrafenib and dabrafenib+trametinib treatments dramatically reduced cell proliferation only in the co-cultured condition (Figure 5F). As shown in Figure 5G, 156 hours 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 (Figure 5H); indeed, 180 hours post-treatment the viability of treated/untreated SEMA6A-depleted cells was significantly reduced as compared with shCtrl cells (Figure 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 (Figure 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.
SEMA6A predicts progression free survival on dual BRAF/MEK inhibition in BRAF-mut melanoma.
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) (Figures 6A and 6B), 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.