The complexity of microenvironment-mediated drug resistance

Since the FDA approval of vemurafenib for the treatment of disseminated melanoma in 2011, followed by approval of dabrafenib in 2013 and dabrafenib/trametinib combination in 2014, melanoma has become the poster child for targeted kinase therapy. Despite encouraging initial clinical responses, most patients ultimately develop drug resistance and relapse, prompting an enormous research effort focusing on the mechanisms of therapy escape. A great variety of drug resistance mechanisms have been identified since, including alternative BRAF splicing, mutations in MEK, mutations in RAS, amplified receptor tyrosine kinase (RTK) signaling, among many others [1]. However, more and more focus has shifted from studying cell-autonomous modes of resistance to determining the impact of the tumor microenvironment on drug sensitivity. The first evidence for stroma-mediated drug resistance in melanoma came from studies demonstrating the role of stroma-derived hepatocyte growth factor (HGF) in BRAF inhibitor resistance [2, 3]. Two recent publications from our group and recent work from Hirata et al. have added to these initial findings by demonstrating new mechanisms of bi-directional cross-talk between the tumor and stromal fibroblasts which allow the tumors to amplify cellautonomous adaptations and create a drug resistant niche [4-6]. Several groups, including our own, have confirmed that co-culturing melanoma cells with fibroblasts leads to a diminished therapeutic response in the melanoma cells [5-7]. Most interestingly, the protective effects observed are not one-dimensional but rather a complex culmination of signaling resulting from direct effects of the drug on melanoma cells, the ability of the drugs to activate normal fibroblasts and crosstalk between fibroblasts and melanoma cells. Our work has shown a subset of melanoma cells to secrete transforming growth factorbeta (TGF-β) in response to vemurafenib treatment, and this TGF-β, in turn, activates dermal fibroblasts that then express alpha-smooth muscle actin, produce fibronectin and secrete neuregulin (NRG-1) [5]. Intriguingly, we found that maximal fibroblast activation was dependent on both melanoma-derived TGFB-β and the direct effects of vemurafenib on the fibroblasts. We showed that vemurafenib had a direct effect on dermal fibroblasts through paradoxical ERK activation, a finding also reported in the recent publication by Hirata et al [5, 6]. Paradoxical ERK activation was shown to be responsible for fibroblast activation and HGF secretion. Accordingly, co-treatment with an inhibitor of MEK blocked the secretion of HGF from fibroblasts [5]. A number of established cell-autonomous resistance mechanisms highlight signaling through upregulated RTKs or through restoration of sensitivity to growth factors [1]. Our work demonstrates that fibronectin secreted in response to vemurafenib treatment can augment RTK