Massively parallel reporter assays combined with cell-type specific eQTL identified a functional melanoma risk variant in HIV-1 restriction gene, MX2 [MPRA NGS]

Genome-wide association studies (GWAS) have identified twenty melanoma susceptibility loci. To identify susceptibility genes and variants simultaneously from multiple GWAS loci, we integrated massively-parallel reporter assays (MPRA) with melanocyte-specific expression quantitative trait locus (eQTL) profiling. We identified thirty-nine candidate functional variants displaying allelic transcriptional activity, of which nine from four loci were also correlated with local gene expression in melanocytes (CTSS, CASP8, MX2, and MAFF). Among these, we further characterized the locus in MX2 gene on chromosome band Chr21q22.3 and validated a functional variant, rs398206. The functional variant mediates allelic transcriptional activity via binding of the transcription factor, YY1. This allelic transcriptional regulation largely accounts for a significant cis-eQTL of the HIV-1 restriction gene, MX2, in primary melanocytes, where the melanoma risk-associated A allele is correlated with higher MX2 levels. Melanocyte-specific transgenic expression of human MX2 in a zebrafish model demonstrated an accelerated melanoma formation in a BRAFV600E background. Thus, using an efficient scalable approach to streamline GWAS follow-up functional studies, we uncovered a pleiotropic function of MX2 in melanoma susceptibility. We tested allelic transcriptional activities of 840 melanoma-associated common variants using MPRA. 832 variants were selected based on their colocalization to open chromatin and active histone marks in melanoma-relevant cell types. The other 8 were selected as controls that do not overlap with these functional genomic signatures. Luciferase reporter libraries were constructed by taking 145bp genomic sequence encompassing the risk or protective allele of each variant. Each 145bp sequence was tested in both forward and reverse directions and was assigned 10 different unique 10bp barcode sequences. We also included a scrambled sequence for each variant, where 21 bases encompassing the variant were scrambled to serve as a pseudo-baseline. Resulting 50,400 unique oligo sequences were cloned into two separate MPRA libraries with or without the minimal promoter (TATA). Libraries were then transfected into a melanoma cell line (UACC903) and the HEK293FT cell line. Resulting transcribed output as well as DNA input were then quantified by sequencing. A total of 22 samples were sequenced including 4 input DNA samples representing 2 libraries with (Enh) or without (Pro) TATA element, and 8 RNA output samples from the transfections into UACC903 representing 2 libraries with (Enh) or without (Pro) TATA in two transfection replicates each, and 10 RNA output samples from the transfections into HEK293FT representing 2 libraries with (Enh) or without (Pro) TATA in two transfection replicates each, except for four replicates for library 1 Enh constructs.