Fig. 2: Integrative analysis of chromatin states and gene expression in lean and obese mouse pancreatic islets revealed dynamic correlations between gene expression and H3K27ac/H3K4me1 levels.

UMAP embedding from single-nucleus joint profiles of transcriptome and histone modification (H3K27ac (A) or H3K4me1 (B)). Total number of nuclei: 37,847 in A and 18,919 in B. Genome tracks of aggregated normalized signals of H3K27ac (C) and H3K4me1 (D) of individual endocrine cell types (α, β, δ, Mki67-beta, and PP cells) at Gcg, Ins1, Sst, Mki67, and Ppy loci.Heatmaps of RNA expression (left) and the accumulated H3K27ac (E) or H3K4me1 (F) signals (right). The top 50 differentially enriched genes for each cell type were selected. Correlations between individual H3K4me1 (G) or H3K27ac (H) enhancer activity and the expression of Glrx3, a gene involved in glutathione metabolism and a marker for stressed β cells¹²⁵. Top 3 panels: genome tracks of pooled H3K4me1 (G) or H3K27ac (H) signals in α, β, and δ cells. Mid panel: H3K4me1 (G) or H3K27ac (H) peaks. Bottom panel: spider plot represents the strength of linkage between H3K4me1 (G) or H3K27ac (H) and Glrx3 expression. (I-J) UMAP shows the trajectory of β cell dysfunction from β-high to beta-low states in H3K27ac(I) and H3K4me1(J). Pseudo-time values were overlaid on the UMAP embedding; the smoothed line and arrow represent the visualization of the trajectory path from the spline fit. K Heatmaps showing the enrichment patterns of H3K27ac (left) and expression (right) of 1,657 genes of which the H3K27ac and RNA are correlated along the pseudo-time order. L Heatmaps showing the enrichment patterns of H3K4me1 (left) and expression (right) of 964 genes of which the H3K4me1 and RNA are correlated along the pseudo-time order.