Figure 5. Chromatin co-accessibility links diabetes risk variants to target genes.

(a) Distance-matched odds that co-accessible sites in beta cells overlap islet pcHi-C interactions at different threshold bins. (b) Single cell co-accessibility and islet pcHi-C interactions at the PDX1 promoter. (c) Enhancer harboring T2D variant rs231361 shows distal beta cell co-accessibility to the INS promoter and other non-promoter sites. The enhancer is accessible in INS^high but not INS^low beta and has decreasing accessibility across the beta cell trajectory. rs231361 disrupts an RFX motif and has predicted effects in INS^high beta. *FDR<0.1. CRISPR/Cas9 deletion of a 2.6 kb region (highlighted, ∆Enh) around the enhancer. (d) Expression (transcripts per million, TPM±SEM) of genes within 2 Mb of the enhancer in beta cell stage cultures for ∆Enh (n=6; 3 clones × 2 differentiations) and control (n=2; 1 clone × 2 differentiations), p-values from DESeq2, ns not significant. (e) Representative immunofluorescence staining for INS (green), NKX6–1 (red), and DAPI (blue) in beta cell stage cultures for ∆Enh (n=3 independent experiments) and control (n=3 experiments) (f) Quantification of INS median fluorescence intensity (MFI) and (g) INS content in beta cell stage cultures for ∆Enh (n=9; 3 clones × 3 differentiations) and control (n=6; 2 clones × 3 differentiations). (h) Relative expression of INS and CDKN1C, (i) Quantification of INS MFI, and (j) INS content in beta cell stage cultures for KCNQ1^A/A (n=9; 3 clones × 3 differentiations) and KCNQ1^G/G (n=6; 2 clones × 3 differentiations). Data shown are mean±SEM, p-values by two-sided Student’s T-test, ns, not significant.