Abstract 97: Genotype to Phenotype: Function of Common Noncoding and Rare Coding Variants In ANGPTL3

Human genetics studies have demonstrated a strong link between ANGPTL3 , which encodes lipoprotein lipase inhibitor Angiopoietin-like 3, and blood lipid phenotypes. Rare nonsense ANGPTL3 mutations were identified in patients with familial combined hypolipidemia, while common variants at the ANGPTL3 locus have been found by genome-wide association studies (GWASs) to associate with lower triglycerides (TGs) and low-density lipoprotein cholesterol. In light of the seemingly favorable clinical consequences of ANGPTL3 deficiency, we established an experimental framework to identify (1) causal common variants that regulate ANGPTL3 expression and (2) rare missense mutations that disrupt ANGPTL3 function. Using massively parallel reporter assays, we profiled the regulatory activity of all the common variants linked ( r 2 ≥ 0.5) to the lead GWAS SNP in the ANGPTL3 locus and found that rs10889356 demonstrated significant allele-specific enhancer activity. To validate this finding, we used CRISPR-Cas9 to alter the SNP in a human pluripotent stem cell line. When differentiated into hepatocytes, altered cells displayed a 67% increase in ANGPTL3 expression ( n = 4 wild-type and 4 mutant clones, P = 0.007). CRISPR interference using each of three guide RNAs targeting the SNP in HepG2 cells also substantially increased ANGPTL3 expression. These findings support rs10889356- ANGPTL3 as a causal SNP-gene set. Next, we examined the coding regions of ANGPTL3 in 20,000 sequenced individuals and sought to experimentally define rare missense variants using a mouse model. We used CRISPR-Cas9 to generate Angptl3 knockout mice, which exhibited decreased TG (61%, P < 0.001) and decreased cholesterol (31%, P < 0.002). We reconstituted the knockout mice to normal expression levels with adenoviruses expressing either wild-type ANGPTL3 or missense variant ANGPTL3 . So far we have assessed 28 rare missense variants computationally predicted to be deleterious, of which only 10—D42N, K58E, S117P, P264S, Q286H, L315S, L360Q, T383I, T383S, and Y417C—were validated as loss-of-function (conferring <25% of wild-type activity as assessed by changes in both TG and cholesterol levels), underscoring the need for functional characterization of variants of uncertain significance.