Identification of Functional Lung Adenocarcinoma Susceptibility Loci through Positional Integration with Human Alveolar Cell Epigenomes

Lung adenocarcinoma (LUAD) is the predominant lung cancer subtype. To date, 15 single nucleotide polymorphisms (SNPs) have been reproducibly associated with LUAD susceptibility. However, as most SNPs identified in genome-wide association studies (GWAS) are not located in the coding regions of genes and are co-inherited with hundreds of SNPs in linkage disequilibrium (LD), mechanistic links to disease outcomes remain largely unexplored. We hypothesized that some SNPs increase LUAD risk by affecting enhancers of gene transcription in alveolar epithelial cells (AECs), the purported cells of origin for LUAD. Forty-six LUAD risk-associated SNPs mapped to putative AEC enhancers, and 11 of these were located in FAIRE peaks. Of those, seven were predicted to form transcription factor binding motifs. For four of these, ChIPseq data confirmed the binding of the predicted transcription factors. We chose rs452384, rs6942067 and rs11364096 for functional validation. The reference (risk) allele for rs452384 forms an NKX2-1 binding site that is disrupted by the alternate allele, elicited a 42% increase in activity in H1648 LUAD cells relative to the alternative allele (P = 6.7 x 10-3), and was associated with elevated TERT levels in esophageal tissue (P = 6 x 10-8). The reference (risk) allele for rs6942067 forms an EP300 binding site disrupted by the alternate allele, elicits a 94% increase in enhancer activity relative to the alternative allele in H1648 LUAD cells (P = 2.7 x 10-2), and was associated with higher expression of potential oncogene DCBLD1 in blood (P = 5.2 x 10-16). Lastly, the alternate (risk) allele for rs11364096 disrupts binding sites for FOXA1, HDAC2 and SP1, elicits a 32% decrease in enhancer activity in A549 LUAD cells relative to the reference allele (P = 5.7 x 10-3), and LUAD data from The Cancer Genome Atlas indicates that this SNP is associated with altered expression in LUAD of four genes on other chromosomes. Our analyses provide possible mechanisms for the genetic susceptibility to LUAD, with the identification of three strong functional SNP candidates that appear to affect AEC enhancers. Further investigation of these enhancers and their target genes may ultimately yield more effective and personalized strategies for LUAD risk assessment, prevention and treatment. Overall design: To test this hypothesis, we performed ChIP-seq for enhancer histone marks using purified primary human AECs, and carried out formaldehyde-assisted identification of regulatory elements (FAIRE) to identify transcription factor accessible DNA. We integrated this data with LUAD index SNPs and SNPs in high LD (r^2=0.5). We performed transcription factor binding site predictions to identify SNPs likely to affect transcription factor binding. Next, we used luciferase reporter assays to test whether selected SNPs confer allele-specific enhancer activity in LUAD cell lines. Lastly, we performed expression quantitative trait locus (eQTL) analysis to identify potential target genes affected by selected SNPs.