Abstract
Differential HLA-C levels influence several human diseases, but the mechanisms responsible are incompletely characterized. Using a validated prediction algorithm, we imputed HLA-C cell surface levels in 228 individuals from the 1000 Genomes dataset. We tested 68,726 SNPs within the MHC for association with HLA-C level. The HLA-C promoter region variant, rs2395471, 800 bp upstream of the transcription start site, gave the most significant association with HLA-C levels (p = 4.2 × 10(-66)). This imputed expression quantitative trait locus, termed impeQTL, was also shown to associate with HLA-C expression in a genome-wide association study of 273 donors in which HLA-C mRNA expression levels were determined by quantitative PCR (qPCR) (p = 1.8 × 10(-20)) and in two cohorts where HLA-C cell surface levels were determined directly by flow cytometry (n = 369 combined, p < 10(-15)). rs2395471 is located in an Oct1 transcription factor consensus binding site motif where the A allele is predicted to have higher affinity for Oct1 than the G allele. Mobility shift electrophoresis demonstrated that Oct1 binds to both alleles in vitro, but decreased HLA-C promoter activity was observed in a luciferase reporter assay for rs2395471_G relative to rs2395471_A on a fixed promoter background. The rs2395471 variant accounts for up to 36% of the explained variation of HLA-C level. These data strengthen our understanding of HLA-C transcriptional regulation and provide a basis for understanding the potential consequences of manipulating HLA-C levels therapeutically.