A key approach we take is to develop melanoma models driven by expression of oncogenes in melanocytes. In humans, these specialised pigment cells largely reside in the skin, where one of their major functions is to provide protection from the harmful effects of the ultraviolet radiation (UVR) that is present in sunlight. However, melanocytes also reside in organs such as the brain, eyes, ears and heart where they are presumed or known to perform other specialist functions. We previously demonstrated that oncogenic BRAF drives melanomagenesis, but although it can initiate this process, it does so by cooperating with additional genetic abnormalities. In 2013, we reported developmental melanoma models driven by oncogenic NRAS (Pedersen et al, 2013). We found that somatic mutations in NRAS in embryonic melanocytes appear to be a risk factor for leptomeningeal melanosis, a rare, but inevitably fatal form of childhood melanoma of the CNS. These studies have improved our understanding of the genetics underlying this particular disease and provided a relevant and tractable model system for its further study.
Drugs that inhibit BRAF can drive a curious paradox in cells. When BRAF is mutated they inhibit MEK/ERK signalling, but when NRAS is mutated, they hyper-activate MEK/ERK signalling. This is because in the presence of active RAS, BRAF inhibitors drive BRAF into a complex with a closely related protein called CRAF, activating CRAF and thereby MEK/ERK signalling. In 2013, we demonstrated that a third closely related protein called ARAF is not functionally redundant with CRAF in this process (Rebocho and Marais, 2013). It appears also to be activated by BRAF, but rather than binding to BRAF, it appears to stabilise the BRAF:CRAF complex, at least in some cells. These data add another layer of complexity to the paradox, but the consequences of this are currently unclear.
In 2012, we demonstrated that a consequence of the BRAF inhibitor paradox is induction of secondary non-melanoma skin cancers in ~30% of patients (keratoacanthomas and squamous cell carcinomas). This is because the paradox accelerates growth of pre-existing tumours, bearing mutations in one of the RAS genes, whose growth is then accelerated by the drugs. Single lesions can be easily removed by surgical approaches, but removal of multiple lesions can prove problematic. We demonstrated that topical application of 5-fluorouracil elicits regression in paradox-induced lesions, providing a relatively simple and inexpensive treatment for patients with large fields of tumours for whom surgery is not desirable (Viros et al, 2013).
Figure 1. The EGF receptor confers BRAF inhibitor resistance in BRAF-mutant melanoma cells. LHS upper panels: phospho-protein arrays for A375 and A375/R cell lines; lower panels: the SFKs confer BRAF inhibitor resistance in BRAF-mutant melanoma cells - phospho-protein arrays. RHS drugs that target both the BRAF (BRAFi) and EGFR (EGFRi) pathways may provide a treatment strategy for patients who develop resistance to BRAF inhibitors.
Although BRAF drugs are effective in the majority of melanoma patients with a BRAF mutation, approximately 20% of melanoma patients do not respond to these drugs despite the presence of a BRAF mutation (primary resistance), and the majority of patients who do respond, relapse on treatment due to the development of secondary resistance. In 2013, we reported that hyper-activation of epidermal growth factor receptors (EGFR) can drive resistance in some patients (Girotti et al. 2013). We developed cell lines resistant to BRAF inhibitors and discovered that they had elevated EGFR, and SRC family kinase (SFK) signalling. Importantly, the combination of EGFR and BRAF drugs was able to inhibit the growth of the resistant lines both in vitro and in vivo. Importantly, we observed increased EGFR and SFK activity in tumours from patients who had developed resistance to BRAF drugs and showed that tumours from resistant patients were susceptible to the combination of BRAF and EGFR drugs. These data established that increased EGFR signalling can mediate resistance to BRAF drugs, adding to the complexity of mechanisms of resistance to these drugs (Figure 1). Our studies suggest that drugs that target both the BRAF and EGFR pathways may delay the onset of resistance in some patients and may even provide second-line treatments for patients who have failed treatment.
Finally, we have continued to use next generation sequencing to characterise the genomes of human melanomas. We performed whole genome, or whole exome sequencing on ten mucosal melanomas. The mutation signature did not implicate UVR exposure in the genesis of this disease and we observed a significantly lower number of single nucleotide variants (SNVs), but a significantly higher number of copy number variations and structural changes than are seen in mucosal melanoma (Furney et al, 2013a). These data show that mucosal and common cutaneous melanomas are different diseases with a distinct genesis. The data suggest that structural variations in the chromosomes play a more significant role in mucosal than in common cutaneous melanomagenesis, but the consequences of these differences to disease progression and treatment are currently unclear.
We have also performed whole genome sequencing on 12 uveal melanoma samples (Furney et al, 2013b). Uveal melanoma is the most common eye malignancy and it has a very poor prognosis, being fatal in about half of patients. Surprisingly, uveal melanoma had a very low tumour burden, with very small number of SNVs (only ~2,000 per genome) and a small number of structural chromosome variations. Again we did not observe a UVR mutation signature, suggesting that UVR does not play a role in uveal melanomagenesis. In addition to recurrent mutations in GNAQ or GNA11 (11/12 samples) and BAP1 (7/12 samples), previously described as the most common oncogenes and tumour suppressors in uveal melanoma, we also identified mutations in SF3B1 in ~15% of the tumours. The tumours with SF3B1 mutations had a better prognosis than those with loss of chromosome 3 and these events generally appear to be mutually exclusive. SF3B1 encodes for a component of the spliceosome machinery and accordingly, the mutations in SF3B1 were associated with differential splicing of specific coding and non-coding RNA. Again, these data suggest that uveal melanoma is a distinct disease that is driven by distinct genetics, highlighting the need to improve our understanding of melanoma biology to allow us to develop new treatment approaches for melanoma patients.