ROR2 promotes EMT in melanoma cell lines
To investigate the role of ROR2 in melanoma progression we used both gain- and loss-of-function approaches based on A375 and UACC903 cell lines overexpressing ROR2 (Fig. 1A) and MeWo and M2 cell lines with ROR2 silencing (Fig. 1B). Upon generation of these cell lines, we observed that ROR2 expression induced marked morphological changes in both A375 and MeWo cells (Fig. 1C). This effect was also detected, albeit less noticeably, in UACC903 and M2 cells (data not shown). Compared to control cells, A375-ROR2 cells presented a more elongated spindle morphology, were more scattered, and made cell-cell contacts more randomly. Measurement of morphometric parameters revealed an increase in elongation and a decrease in roundness in A375-ROR2 cells (Fig. 1D). On the other hand, the typically thin and elongated MeWo cells turn to a more compact shape upon ROR2 silencing (Fig. 1C) that was evidenced by a decrease in elongation and an increase in roundness (Fig. 1D). The morphological features of ROR2 expressing cells induced by ROR2 are similar to those described in cells undergoing EMT . Although neural crest-derived melanocytes are not epithelial in origin, an EMT-like process is a well-documented phenomenon contributing to the metastatic potential of melanoma . To further evaluate the induction of EMT by ROR2 we determined the expression of both epithelial and mesenchymal markers upon ROR2 overexpression and silencing. We found that ROR2 decreased the expression of the epithelial marker E-cadherin and increased the expression of both the mesenchymal markers N-cadherin and Vimentin and the EMT-inducing transcription factors Twist, Slug, ZEB1, and Snail (Fig. 1E). In contrast, the expression levels of these proteins were altered in the opposite direction by ROR2 knocking-down in both M2 and MeWo cells (Fig. 1F). Furthermore, both overexpression and silencing experiments confirmed that ROR2 positively regulates the protein levels of HIF-1a, another major EMT driver (Fig. 1G). Finally, we measured ROR2-dependent changes in cell migration as a cellular read-out of EMT induction. We determined that ROR2 overexpression in both A375 and UACC903 cells increased cell migration (Fig. 1H and Suppl. Fig. S1). In contrast, silencing of ROR2 decreased cell migration in M2 cells (Fig. 1I). These results demonstrate that ROR2 promotes EMT in melanoma cells.
ROR2 promotes EMT in melanoma patients
In a previous report, we described six prognostic gene expression signatures (LMC classes) from the analysis of a large cohort of primary melanoma patients (LMC, Leeds Melanoma Cohort) . Among the six classes, ROR2 is highest in class 6, characterized by having a high EMT score, upregulation of both c-JUN and AXL, and a bad prognosis  (Fig. 2A). To confirm this role of ROR2 in EMT, we sought to identify Differentially Expressed Genes (DEGs) between melanoma samples from the independent dataset TCGA-SKCM with high and low ROR2 levels. In agreement with the observations above, the DEGs upregulated in tumors with high ROR2 expression (Supplementary Table S1) were significantly enriched in signatures associated with invasion, extracellular matrix organization, and EMT (Fig. 2B). Altogether, these results demonstrate that ROR2 promotes EMT in human melanoma.
Induction of EMT by ROR2 is mediated by ERK hyperactivation
Since it was recently shown that p38 signaling promoted EMT in breast cancer downstream of ROR2 , we evaluated the effect of ROR2 in p38 phosphorylation in melanoma cells. We found that ROR2 overexpression increased p38 phosphorylation and ROR2 silencing inhibited this pathway (Suppl. Fig. S2). However, inhibition of p38 using the pharmacological inhibitor SB202190 did not alter the expression of the EMT marker Vimentin (Suppl. Fig. S3). These results suggest that although ROR2 activates MAPK/p38, this pathway is not implicated in EMT in melanoma. Next, we looked at the effect of ROR2 on the MAPK/ERK pathway. Despite melanoma cells present constitutive activation of ERK, we found that ROR2 overexpression increased around 10-fold the phosphorylation of ERK (Fig. 3A). In contrast, silencing of ROR2 abolished ERK phosphorylation (Fig. 3B). To determine the role of ERK on EMT induction by ROR2 we used PLX-4032 (PLX), an inhibitor of the ERK pathway. Before doing so, we corroborated that 10 mM PLX similarly inhibited p-ERK levels in both A375-Empty and A375-ROR2 (Fig. 3C). The addition of PLX induced the expression of E-cadherin and potently inhibited Vimentin and N-cadherin expression in A375-ROR2 cells but not in A375-Empty cells (Fig. 3D). PLX also inhibited Snail levels in both cell lines although the inhibition observed in A375-ROR2 was much potent (Fig. 3D). Similarly, PLX abrogated both basal and ROR2-induced HIF-1a levels (Fig. 3E). These results demonstrate that ROR2 induces ERK hyperactivation and that the induction of EMT by ROR2 is highly dependent on this event.
ROR2 induces in vivo EMT, invasion, and necrosis
To evaluate the role of ROR2 in vivo we injected both A375-Empty and A375-ROR2 cells into nude mice. After 6 weeks, tumors from A375-ROR2 were significantly smaller than those from control mice (Fig. 4A). Histological analysis of tumor sections revealed that the tumor cells from mice injected with A375-Empty had a cobblestone-like shape whereas those from the A375-ROR2 derived tumors had an elongated shape (Fig. 4B), suggestive of epithelial-like and mesenchymal-like phenotypes, respectively. In line with this observation, western blot analysis revealed that A375-ROR2 tumors express low levels of the epithelial marker E-cadherin and high levels of the mesenchymal markers Vimentin, N-cadherin, ZEB1, Twist, Slug, Snail, and Fibronectin together with higher expression of p-ERK (Fig. 4C). The upregulation of p-ERK, Vimentin, and Snail in A375-ROR2 tumors was confirmed by immunohistochemistry of tumor sections (Fig. 4D). In agreement with this mesenchymal profile and the increased in vitro migration of A375-ROR2 cells, tumors from mice injected with these cells were found to invade adjacent muscles (Fig. 4E).
Interestingly, unlike tumors from control mice, those from mice injected with A375-ROR2 cells presented extensive necrosis (Fig. 5A). The induction of necrosis was confirmed by the observation of an increased expression of the necrotic markers 14-3-3 and HSP90  (Fig. 5B) and necrotic PARP cleavage  (Fig. 5C). The necrotic areas from A375-ROR2 tumors were delimited by cells presenting high levels of both p-ERK and Snail nuclear staining (Fig. 5D). In contrast, the cells delimiting the small regions of necrosis present in control tumors displayed both negative Snail and low p-ERK staining (Fig. 5D). These results demonstrate that ROR2 promotes in vivo necrosis, EMT, and tumor cells with invasive features.
ROR2 expression at latter stages reduces melanoma survival
Following our observation that ROR2 promotes EMT and invasion and previous reports showing that ROR2 increased lung colonization in mouse models of tail vein injection, we sought to evaluate whether ROR2 contributes to metastasis in melanoma patients. To this end, we determined the impact of ROR2 levels on melanoma-specific overall survival, a direct indicator of distant metastasis. To avoid the confounding effect of comparing primary tumors with ample differences in their clinico-pathological features (i.e. AJCC stage) we restricted our analysis to patients with lymph node metastasis (LNM), an early step on the metastatic cascade that is also a good indicator of distant metastasis and survival. Thus, the relationship between ROR2 levels in LNM samples and survival was evaluated in a cohort of 130 samples from dataset GSE65904. High levels of ROR2 significantly correlated with shorter overall survival (Fig. 6A). Although ROR2-high and ROR2-low patients presented similar distant metastasis-free survival (dMFS), the subgroup of patients with the highest ROR2 levels (Z-score > 1) exhibited a significantly shorter dMFS (Fig. 6B). These results indicate that tumors expressing high levels of ROR2 are prone to form metastasis and have shorter survival. Furthermore, ROR2 not only favors metastasis but its expression is also important in established metastasis as revealed by transcriptomic data showing that ROR2 levels are significantly higher in metastasis than in primary tumors in both the TCGA-SKCM and GSE46517 datasets (Fig. 6C and 6D). Moreover, analysis of matched primary and metastatic samples from the same patients revealed a significant ROR2 increase in the latter with seven out of nine patients displaying higher ROR2 levels in the metastatic sample (Fig. 6E). Altogether, our results demonstrate that ROR2 promotes EMT and invasion and contributes to melanoma metastasis.