Identification of structural motifs critical for human G6PC2 function informed by sequence analysis and an AlphaFold2-predicted model

Biochemical and metabolic characterization of a G6PC2 inhibitor

Three glucose-6-phosphatase catalytic subunits, that hydrolyze glucose-6-phosphate (G6P) to glucose and inorganic phosphate, have been identified, designated G6PC1-3, but only G6PC1 and G6PC2 have been implicated in the regulation of fasting blood glucose (FBG). Elevated FBG has been associated with multiple adverse clinical outcomes, including increased risk for type 2 diabetes and various cancers. Therefore, G6PC1 and G6PC2 inhibitors that lower FBG may be of prophylactic value for the prevention of multiple conditions. The studies described here characterize a G6PC2 inhibitor, designated VU0945627, previously identified as Compound 3. We show that VU0945627 preferentially inhibits human G6PC2 versus human G6PC1 but activates human G6PC3. VU0945627 is a mixed G6PC2 inhibitor, increasing the Km but reducing the Vmax for G6P hydrolysis. PyRx virtual docking to an AlphaFold2-derived G6PC2 structural model suggests VU0945627 binds two sites in human G6PC2. Mutation of residues in these sites reduces the inhibitory effect of VU0945627. VU0945627 does not inhibit mouse G6PC2 despite its 84% sequence identity with human G6PC2. Mutagenesis studies suggest this lack of inhibition of mouse G6PC2 is due, in part, to a change in residue 318 from histidine in human G6PC2 to proline in mouse G6PC2. Surprisingly, VU0945627 still inhibited glucose cycling in the mouse islet-derived βTC-3 cell line. Studies using intact mouse liver microsomes and PyRx docking suggest that this observation can be explained by an ability of VU0945627 to also inhibit the G6P transporter SLC37A4. These data will inform future computational modeling studies designed to identify G6PC isoform-specific inhibitors.

MultiSuSiE improves multi-ancestry fine-mapping in All of Us whole-genome sequencing data

Leveraging data from multiple ancestries can greatly improve fine-mapping power due to differences in linkage disequilibrium and allele frequencies. We propose MultiSuSiE, an extension of the sum of single effects model (SuSiE) to multiple ancestries that allows causal effect sizes to vary across ancestries based on a multivariate normal prior informed by empirical data. We evaluated MultiSuSiE via simulations and analyses of 14 quantitative traits leveraging whole-genome sequencing data in 47k African-ancestry and 94k European-ancestry individuals from All of Us. In simulations, MultiSuSiE applied to Afr47k+Eur47k was well-calibrated and attained higher power than SuSiE applied to Eur94k; interestingly, higher causal variant PIPs in Afr47k compared to Eur47k were entirely explained by differences in the extent of LD quantified by LD 4th moments. Compared to very recently proposed multi-ancestry fine-mapping methods, MultiSuSiE attained higher power and/or much lower computational costs, making the analysis of large-scale All of Us data feasible. In real trait analyses, MultiSuSiE applied to Afr47k+Eur94k identified 579 fine-mapped variants with PIP > 0.5, and MultiSuSiE applied to Afr47k+Eur47k identified 44% more fine-mapped variants with PIP > 0.5 than SuSiE applied to Eur94k. We validated MultiSuSiE results for real traits via functional enrichment of fine-mapped variants. We highlight several examples where MultiSuSiE implicates well-studied or biologically plausible fine-mapped variants that were not implicated by other methods.