Identification of PAM as a novel monogenic diabetes gene

Various biological mechanisms involved in Type 2 Diabetes (T2D) have been unveiled through converging evidence in common frequency variants from Genome wide association studies (GWAS) and rare or low frequency variants (RV/LFVs) from Exome wide association studies (EXWAS). We hypothesized that a two-pronged approach interrogating both common variants and RV/LFVs could yield identification of new diabetes genes. Per gene protein-altering RV/LFV burden was tested for association with T2D risk in an EXWAS of 11,731 cases and 149,613 controls in the UKBiobank. We used a logistic regression model adjusted for age, sex, and the top 20 genetic principal components of the Caucasian European cohort. Proteins encoded by significant genes were then investigated using two-sample Mendelian randomization (MR) for causal association with T2D and related variables, including fasting glucose, fasting insulin, post-prandial insulin secretion, and HbA1c. Protein quantitative trait loci were estimated in the Prospective Urban and Rural Epidemiological study (N=11,020) and compared to GWAS-summary statistics of T2D and related outcomes from the UKBiobank (N=407,701), FinnGen (N=207,318), MAGIC (N=58,074), DIAGRAM (N=898,130), and NEO (N=3,526). At the exome-wide significance threshold (P=0.05/19,589=2.55×10-6), we identified loss-of-function RV/LFVs in GCK and PAM, which encode the proteins glucokinase and peptidylgycine α-amidating monooxygenase respectively, to be associated with T2D risk. While GCK mutations are established causes of monogenic diabetes, deleterious mutations in the PAM gene were associated with a 29% increase in T2D risk (OR per RV/LFV 1.29; 95% CI, 1.17 to 1.43; P=7.17×10-7). This association was driven by a low frequency variant, rs78408340-G (p.Ser539Trp) present in 2.6% of T2D cases and in 1.8% of controls. Two-sample MR analyses confirmed that 1 SD decrease in genetically predicted PAM circulating levels was associated with increased T2D risk (OR=1.16; 95% CI, 1.14 to 1.19; P=9.54×10-52) and HbA1c (%) (b=0.0047; 95% CI, 0.00051 to 0.0090; P=0.028), decreased post-prandial insulin (mU/L) secretion (b=-0.060; 95% CI, -0.083 to -0.036; P=8.32×10-7), but decreased fasting glucose (pmol/L) level (b=-0.027; 95% CI, -0.034 to -0.20; P=1.76×10-13). No association was found with fasting insulin levels. Our findings identified PAM variants as a novel cause of monogenic diabetes and demonstrated a strong causal effect of circulating PAM levels on T2D risk and post-prandial insulin secretion. This supports further investigation of PAM as a therapeutic T2D target.