High-throughput characterization of functional variants highlights heterogeneity and polygenicity underlying lung cancer susceptibility

Genome-wide association studies (GWAS) have identified multiple lung cancer risk loci including those that are distinct in the major histological types. However, most of these loci have not been functionally characterized. Here we employed massively parallel reporter assays (MPRA) to assess allelic transcriptional activity of risk-associated variants en masse. We tested 2,245 variants in 42 loci from 3 recent GWASs of East Asian and European populations in the context of lung adenocarcinoma and squamous cell carcinoma cells while incorporating exposure to a tobacco-smoke carcinogen, benzo[a]pyrene. At FDR < 0.01, we identified 844 MPRA-significant variants with allelic transcriptional activity across 88% of lung cancer loci. Further variant scoring using lung-specific epigenomic annotation demonstrated that 63% of the loci harbored multiple equally functional variants in linkage disequilibrium. Cell-type-specific variants were observed in 72% of the loci, a subset of which aligned with histology-specific association in the GWAS. Distinct subsets of transcription factors were predicted to bind to cell-type-specific variants and those affecting multiple GWAS loci in trans. Linking MPRA-significant variants to target genes based on four different approaches identified candidate susceptibility genes including essential genes for lung cancer cell growth. CRISPR-interference of a high-scoring MPRA-significant variant validated multiple variant-gene connections from different datasets, including RTEL1, SOX18, and ARFRP1. Our data provide a comprehensive catalog of functional characterization of lung cancer GWAS loci and the molecular basis of heterogeneity and polygenicity of lung cancer susceptibility. Targeted amplicon RNA-sequencing of ectopic luciferase constructs with sequence tags transfected into A549 human lung adenocarcinoma cell line and H520 human lung squamous cell carcinoma cell line with or without exposure to benzo(a)pyrene (n = 5 transfections for each condition) together with amplicon DNA sequncing of input luciferase constructs for normalization control