Genetically Engineered Mouse Models to Study Melanoma Genesis and Progression

Division Of Basic Sciences - Nci

Investigators

Linked publications & trials

Abstract

The Cancer Modeling Section seeks to elucidate the complex molecular/genetic program governing melanoma genesis and progression through the development and analysis of GEM models of human cancer. Exposure to UV is considered a causal agent in approximately 80% of melanoma. Previously, we tested this hypothesis in a GEM model in which the receptor tyrosine kinase MET was deregulated by virtue of ectopic expression of its ligand, hepatocyte growth factor/scatter factor (HGF); this pathway is highly relevant in human melanoma genesis and drug resistance. We discovered that a single neonatal dose of burning UV in these GEM mice was necessary and sufficient to induce tumors reminiscent of human melanoma with shortened latency (Noonan et al., Nature 413: 271-2, 2001). A critical role for the INK4A/ARF locus, which helps regulate the pRb and p53 pathways and is widely regarded as a key melanoma suppressor in human patients, was also confirmed in our animal model (Recio et al., Cancer Res. 62: 6724-30, 2002). We also used albino HGF transgenic mouse to show that UVB, but not UVA, alone is able to induce the full melanoma phenotype in the absence of pigment (DeFabo et al., Cancer Res. 64: 6372-6, 2004). However, we also showed that UVA is highly melanomagenic in pigmented HGF/SF-transgenic mice (Noonan et al., Nat. Commun. 3: 884, 2012), demonstrating that melanin is associated with oxidative DNA damage and mutagenesis, and thus represents a double-edged sword with respect to melanoma risk. To determine the roles of UV in melanoma in vivo, we developed a mouse model (iDCT-GFP) that allows melanocytes, specifically and inducibly labeled with green fluorescent protein (GFP), to be isolated from disaggregated mouse skin by FACS following UV irradiation in vivo. We identified a pattern of UVB induced gene expression changes in melanocytes isolated from mice that are consistent with inflammatory alterations and may spare melanocytes post-UV remodeling-associated destruction. We identified an interferon (IFN)-gamma signaling signature arising in melanocytes after neonatal UV irradiation. The source was a type of macrophage recruited to the skin after UV exposure; IFN-gamma in turn activated melanocytes and the expression of genes that could facilitate immunoevasion. Transplanted neonatal macrophages were found to significantly enhance melanoma growth in vivo in an IFN-gamma-dependent fashion. This was surprising considering that IFN-alpha has been used to treat melanoma patients, albeit with limited success. We hypothesized that melanomas escape immune destruction by co-opting these pathways already hard-wired in melanocytes, and suggested that the IFN-gamma signaling pathway may represent a promising therapeutic target for melanoma (Zaidi et al., Nature 469:548-553, 2011). To address the UV mutation question we are also subjecting in vivo-exposed melanocytes from our HGF GEM model to whole exome and RNA sequencing. We have subjecting GFP-labeled melanocytes from all stages of melanoma development relevant to human disease, including nevi, to whole exome and RNA sequencing to catalog their precise genomic alterations. This in vivo HGF mouse, which represents a relevant model for so-called triple wild type melanoma (no mutations in BRAF, NRAS, NF1) is now providing novel insights into the nature of UV-induced damage, and the mechanisms by which UV provokes melanoma. Recurring mutations have been identified that may be associated with the initiation of nevus formation by UV-induced mutagenesis. In particular, mutations in Gnaq, Gna11, and other PLCB4 pathway members, become important in melanocyte transformation in the absence of dominant oncogenic mutations. This model has the added benefit of serving to test drugs effective in uveal melanoma, often driven my GNAQ/11 mutations. We have also hypothesized that late stage melanoma cells can co-opt pathways hard-wired into normal developing melanoblasts to achieve a more aggressive and metastatic phenotype. Both the embryonic melanoblast and the metastatic melanoma cell must undergo a similar EMT and become invasive, highly migratory, and survive to colonize at distant sites. We again employed our iDCT-GFP GEM model to isolate embryonic melanoblasts from key stages of melanocyte development. RNA sequencing and microarray-based gene expression profiling have been performed from representative developmental stages. Genes have been identified whose expression is characteristically up-regulated in both melanoblasts and metastatic melanoma but poorly expressed in adult melanocytes (Met/Dev genes), which may represent new therapeutic targets against metastasis in melanoma. Novel candidates, which include genes that regulate ER stress, autophagy, and neural development have been evaluated for a role in metastasis using RNAi-based knockdown and CRISPR-based knockout in metastatic human and mouse melanoma cells, and then tail vein injections. Several candidates have now been shown to regulate metastatic behavior, and to correlate with melanoma patient survival. We have focused on our top hit, the ER stress gene KDELR3, which we have linked to metastatic behavior in melanoma cell lines through analysis of the TCGA database and assessment of melanoma patient survival. We have linked KDELR3 pro-metastatic activity to ERAD-based degradation of the metastasis suppressor KAI1, and to mitochondrial metabolism (Marie et al., Nature Comm 11:333, 2020). We believe this set of Met/Dev genes will prove to be a valuable resource for new insights and mechanisms associated with metastatic spread. Other hits that were linked to both melanoma metastasis and melanocyte development were knocked out in zebrafish and several interesting phenotypes were observed, including one regulating melanocyte proliferation. These are now being evaluated and characterized. We have also integrated gene and microRNA expression data from our mouse models of highly and poorly malignant melanocytic tumors, and human melanoma databases, and discovered that miR-32/MCL1 pathway members are novel candidate anti-melanoma drug targets, and that their inhibition may enhance the efficacy of BRAFV600E inhibitors in the clinic (Mishra et al., PLoS One 11:e0165102, 2016). We discovered that PTEN phosphatase inhibits metastasis, at least in part, by negatively regulating the N-glycosylation and folding promoter Entpd5/IGF1R pathway, offering new therapeutic targets for PTEN-mutant metastatic melanomas (Yu et al., npj Precision Oncol, in press). Moreover, we found that PHLPP1 expression was significantly down-regulated or lost in metastatic melanoma tissue, which correlated with metastatic potential of melanoma cell lines (Yu et al., Oncogene 37: 2225-36, 2018). PHLPP1 also functions as a metastasis suppressor through its phosphatase activity and represents a new diagnostic marker for metastatic melanoma. We also found that AXL/AKT axis-mediated resistance to BRAF inhibition depended on PTEN status (Zuo et al., Oncogene 37: 3275-89, 2018). AXL is a key upstream regulator of AKT pathway-associated resistance to BRAF inhibition in WT PTEN, but not impaired PTEN melanomas. AXL has also come up as a key regulator of plasticity in our in vivo melanoma models. Thus, AXL could represent a promising therapeutic target for BRAF mutant melanoma patients with WT PTEN. Finally, we discovered and reported that PERK plays a critical role in BRAF inhibitor-acquired resistance in melanoma with impaired PTEN, providing a new strategy for combating BRAF inhibitor *TRUNCATED*