Abstract 5678: Developing a preclinical immunotherapy platform using syngeneic mouse models of human melanoma

Melanoma is an aggressive and lethal disease with no efficacious therapies for a broad subset of late stage patients. Current immunotherapies, including immune checkpoint blockade (ICB), may prolong survival in certain patients, but generate responses in less than 40% of the treated cohort. This demands a better understanding of molecular mechanisms underlying the lack of response and acquired resistance to ICB. However, functional studies are limited in patients, and current preclinical studies are handicapped by the absence of appropriate mouse models that recapitulate the pathological and immunological diversity of human melanomas. Here we develop four syngeneic melanoma mouse models with human-relevant genetic modifications and carcinogenic agents, which we hypothesize will mirror the spectrum of responses to ICB and offer a platform for future mechanistic studies in melanoma. The models are: 1) neonatal ultraviolet radiation (UV)-induced melanoma in a HGF-transgenic mouse, in which melanocyte localization at the epidermal-dermal junction mimics human distribution (HU); 2) 7,12-Dimethylbenz(a)anthracene (DMBA)-induced melanoma in a HGF-tg and Cdk4R24C mouse (HC4D); 3) UV-induced melanoma in a BrafCA/+; HGF-tg; Cdkn2aflox/+; Tyr-CreERT2-t mouse (BHCU); and 4) UV-induced melanoma in a BrafCA/+; Ptenflox/+; Cdkn2aflox/+; Tyr-CreERT2-tg mouse (BPCU). Exome sequencing of the four models reveals a high correlation with mutational subtypes previously described in human melanoma. BPCU and BHCU represent different Braf mutant patient populations and HU and HC4D represent triple wildtype melanoma (non-BRAF, -NRAS, -NF1). The four mouse models demonstrate distinct responses to ICB with anti-CTLA-4 treatment. While HU and HC4D melanomas show high or partial sensitivity to anti-CTLA-4, respectively, BPCU and BHCU do not respond to treatment. In vivo vaccination assays demonstrate that the anti-CTLA-4 response in our models is linked to increased tumor immunogenicity. However, the number of non-synonymous mutations and antigen presentation functionality do not correlate with ICB efficacy. Tumor infiltration by T cells was assessed by CD3 and FoxP3 immunostaining and gene expression analysis. Although clear differential gene expression profiles are noted among the four models and in those tumors responding to the treatment, we unexpectedly found that “hot” melanomas (e.g., showing upregulation of inflammatory pathways and high T-cell infiltration) do not necessarily