Prioritization of osteoporosis-associated GWAS SNPs using epigenomics and transcriptomics

Genetic risk factors for osteoporosis, a prevalent disease associated with aging, have been examined in many genome-wide association studies (GWAS). A major challenge is to prioritize transcription-regulatory GWAS-derived variants that are likely to be functional. Given the critical role of epigenetics in gene regulation, we have used an unusual epigenetics- and transcription-based approach to identify credible regulatory SNPs relevant to osteoporosis from 38 reported BMD GWAS. Using Roadmap databases, we prioritized SNPs based upon their overlap with strong enhancer or promoter chromatin preferentially in osteoblasts relative to 11 heterologous cell culture types. The selected SNPs also had to overlap open chromatin (DNaseI-hypersensitive sites) and DNA sequences predicted to bind to osteoblast-relevant transcription factors in an allele-specific manner. From >50,000 GWAS-derived SNPs, we identified 16 novel and credible regulatory SNPs (Tier-1 SNPs) for osteoporosis risk. Their associated genes, BICC1, LGR4, DAAM2, NPR3, or HMGA2, are involved in osteoblastogenesis or bone homeostasis and regulate cell signaling or enhancer function. Four of them are preferentially expressed in osteoblasts. BICC1, LGR4, and DAAM2 play important roles in canonical Wnt signaling, a pathway critical to bone formation and repair. The transcription factors that are predicted to bind to the Tier-1 SNP-containing DNA sequences also have bone-related functions. For the seven Tier-1 SNPs near the 5′ end of BICC1, examination of eQTL overlap and the distribution of BMD-increasing alleles suggests that at least one SNP in each of two clusters contributes to inherited osteoporosis risk. Our study not only illustrates a method that can be used to identify novel BMD-related causal regulatory SNPs for future study, but also reveals evidence that some of the Tier-1 SNPs exert their effects on BMD risk indirectly through little-studied noncoding RNA genes, which in turn may control the nearby bone-related protein-encoding gene.