Guest post by John Morris (ASHG talk: Wed Oct 17, 9:00am – 9:15am in Room 6D)
Osteoporosis is a common, aging-related disease characterized by decreased bone strength and, consequently, increased fracture risk. Bone mineral density (BMD), a non-invasive measurement, is the most clinically relevant risk factor for diagnosing osteoporosis and is highly heritable (i.e. determined by genetics). To understand the genetic determinants of osteoporosis, we performed a genome-wide association study (GWAS) in 426,824 UK Biobank participants to identify regions of the genome associated with BMD estimated from quantitative heel ultrasound (eBMD). This approach is unbiased in that it systematically tests millions of single nucleotide polymorphisms (SNPs) in the human genome—sites of common, uncommon, or rare genetic variation in the general population—for association with eBMD measurements. BMD-associated SNPs can then be used to identify novel bone genes, but such genes would require further study in human cells or animal models to understand their function. Therefore, we collaborated with the Origins of Bone and Cartilage Disease (OBCD, www.boneandcartilage.com) to examine genes in knockout mice. Such genes, when validated by knockout mouse skeletal phenotyping, represent strong candidates for developing new therapies to prevent and treat osteoporosis.
Our eBMD GWAS identified 518 significant regions of the genome, 301 of which were novel findings. Next, to identify target genes, we performed statistical fine-mapping and integrative bone cell functional genomics data analyses. First, by leveraging SNP association summary statistics and SNP-by-SNP correlations, we can identify a subset of plausibly causal SNPs. Then, by intersecting this list of plausibly causal SNPs with genomic characteristics that indicate function (e.g. coding SNPs, osteoblast open chromatin, osteoblast 3D contacts with gene promoters), we can identify a list of target genes likely to be functional in bone cells. These orthogonal approaches resulted in a list of 515 target genes, identified by plausibly causal and putatively functional SNPs, that we found were strongly enriched for known bone genes and osteoporosis drug targets. We sought to examine the effects of as many of these genes as possible in knockout mice and found the OBCD had skeletal phenotyping data on 126. Importantly, the OBCD receives all knockout mouse lines for skeletal phenotyping at random from the International Mouse Phenotyping Consortium (IMPC), therefore it is not known beforehand if a given knockout mouse has a skeletal phenotype. These 126 genes were found to be enriched for outlier skeletal phenotypes, providing strong evidence that our target genes are disease-relevant, and we focused on one such gene in further detail: disheveled-associated activator of morphogenesis 2 (DAAM2).
Mice with hypomorphic Daam2 alleles were found to have increased cortical porosity and markedly reduced bone strength, even though all other cortical bone parameters, including BMD, were normal. We performed further analyses on DAAM2, such as CRISPR-Cas9 mediated knockouts in human osteoblast cell lines, revealing a decreased ability of this crucial bone-forming cell to mineralize. We concluded that DAAM2 is a novel risk gene for osteoporosis meriting further study and highlighted five other strong candidates for follow-up: CBX1, WAC, DSCC1, RGCC, and YWHAE. In summary, we have generated an atlas of genetic influences on osteoporosis in humans and mice, more fully describing its genetic architecture. Human and mouse genetics identified DAAM2 and other genes previously unknown to function in bone biology. We expect the genes identified here to include new drug targets for the treatment of osteoporosis, where novel therapeutic options are a health priority.