BAF60a-dependent chromatin remodeling preserves β cell function and contributes to the therapeutic benefits of GLP-1R agonists

Abstract

Impaired glucose-stimulated insulin secretion (GSIS) is a hallmark of β cell dysfunction in diabetes. Epigenetic mechanisms govern cellular glucose sensing and GSIS by β cells, but they remain incompletely defined. Here, we found that BAF60a functions as a chromatin regulator that sustains biphasic GSIS and preserves β cell function under metabolic stress conditions. BAF60a was downregulated in β cells from obese and diabetic mice, monkeys, and humans. β cell-specific inactivation of BAF60a in adult mice impaired GSIS, leading to hyperglycemia and glucose intolerance. Conversely, restoring BAF60a expression improved β cell function and systemic glucose homeostasis. Mechanistically, BAF60a physically interacted with Nkx6.1 to selectively modulate chromatin accessibility and transcriptional activity of target genes critical for GSIS coupling in islet β cells. A BAF60a V278M mutation associated with decreased β cell GSIS function was identified in human donors. Mice carrying this mutation, which disrupted the interaction between BAF60a and Nkx6.1, displayed β cell dysfunction and impaired glucose homeostasis. In addition, GLP-1R and GIPR expression was significantly reduced in BAF60a-deficient islets, attenuating the insulinotropic effect of GLP-1R agonists. Together, these findings support a role for BAF60a as a component of the epigenetic machinery that shapes the chromatin landscape in β cells critical for glucose sensing and insulin secretion.

Keywords: Beta cells; Cell biology; Diabetes; Endocrinology; Insulin.

Figure 1. Multi-omics profiling identifies BAF60a inactivation as a key mediator of β cell dysfunction in T2D.

(A) Heatmap of shared DAPs in islets from ob/ob and db/db mice. Scale bar: Z score. (B) GO analysis of genes annotated to DAPs in (A). (C) Schematic of siRNA screening pipeline. (D) Heatmap of chromatin remodeling genes downregulated in ob/ob (n = 3) and db/db (n = 5) islets. Scale bar: Z score. (E) Scatter plot of GSIS changes in Min6 cells transfected with siRNAs (n = 6 per gene). Significant hits are indicated in black, BAF60a in red. Control siRNA used for normalization. (F) Correlation between relative BAF60a expression and GSIS. Linear regression is shown with 95% CI (dashed). Mean ± SEM; n = 6 biological replicates. (G) Heatmap of altered BAF60a-binding peaks in db/db islets (|log2FC| > 0.5; FDR < 0.05). (H) GO analysis of genes linked to more-accessible (red) or less-accessible (blue) BAF60a-binding peaks. (I) Relative BAF60a, Slc2a2, and Pdx1 mRNA in HFD (n = 3), ob/ob (n = 4), and db/db (n = 3–5) islets. Mean ± SEM; 2-tailed unpaired Student’s t test. (J and K) Immunoblot and quantification of islet lysates (pooled from 3 mice/group). Mean ± SEM; 2-tailed unpaired Student’s t test. (L and M) Representative immunofluorescence images and quantification of mouse pancreas sections (n = 6). Scale bars: 100 μm and 10 μm. Quantification is shown per mouse (left) and per islet (right). Mean ± SEM; 2-tailed unpaired Student’s t test. (N–Q) Representative immunofluorescence images and quantification of monkey (n = 3/group in N and O) and human (n = 21/group in P and Q) pancreas sections. Scale bars: 100 μm and 10 μm. Gray dots indicate islets; colored dots indicate monkey or human individuals. Mean ± SEM; P values were determined by 1-way ANOVA (monkey individuals) or 2-tailed unpaired Student’s t test (human individuals). CTR, control; INS, insulin.

Figure 2. Ablation of BAF60a in β cells impairs insulin (Ins) secretion and glucose (Glu) homeostasis.

(A) Workflow for generating BaβKO mice. (B) Relative BAF60a mRNA in flox/flox and BaβKO islets. Mean ± SEM; n = 3–4/group; 2-tailed unpaired Student’s t test. (C) Immunoblot of islet lysates (pooled from 3 mice/group). (D) Pancreas sections stained with indicated antibodies. Scale bar: 100 μm and 10 μm. (E) Short-fast blood glucose levels. Mean ± SEM; n = 4/group; 2-tailed unpaired Student’s t test. (F) OGTT results. Mean ± SEM; n = 4–5/group; 2-way ANOVA. (G) In vivo GSIS results. Mean ± SEM; n = 4–6/group; 2-way ANOVA. (H) ITT results. Mean ± SEM; n = 4–5 per group. (I) Dynamic GSIS and biphasic insulin release from islets pooled from 3 mice/group. Mean ± SD; n = 3 technical replicates; 2-tailed unpaired Student’s t test. (J) Intracellular Ca²⁺ influx in response to 16.8 mM glucose. Islets pooled from 3 mice. Mean ± SEM; n = 42–49 islets/group; 2-way ANOVA. (K and L) GSIS (left) and Ca²⁺ influx (right) in Min6 cells with BAF60a knockdown (K) or KO (L). Mean ± SEM; GSIS: n = 3 biological replicates; Ca²⁺ influx: n = 94–96 (K) or 46–51 (L) cells/group; 2-way ANOVA. CTR, control.

Figure 3. BAF60a regulates β cell function by modulating glucose metabolism–related genes.

(A) Left: Differential expression analysis of RNA-Seq from flox/flox and BaβKO islets (RNA pooled from 3 mice; n = 3 biological replicates; |log2FC| > 0.5; FDR < 0.05). Right: GO analysis of significantly differentially expressed genes (DEGs) showing top nonredundant biological processes with gene counts and P values. (B) Heatmap of selected pathway-associated genes from (A). (C) Left: Uniform Manifold Approximation and Projection (UMAP) visualization of transcriptome-based clustering in flox/flox and BaβKO islets. Islets pooled from 3 mice; 4,077 and 4,078 cells analyzed, respectively. Right: GO analysis of DEGs in β cell cluster with top biological processes. (D) Workflow of ¹³C-glucose metabolic flux analysis in Min6 cells. (E) ¹³C-glucose tracing in control (CTR) and BAF60a-KO Min6 cells. M, unlabeled metabolite; M+n, metabolite mass with n labeled carbons. Mean ± SEM; n = 5 biological replicates; 2-tailed unpaired Student’s t test. (F) Molar percent enrichment (MPE) of ¹³C in glycolysis and TCA metabolites in CTR and BAF60a-KO Min6 cells. Mean ± SEM; n = 5 biological replicates; 2-tailed unpaired Student’s t test. (G) Intracellular ATP content in CTR and BAF60a-KO Min6 cells stimulated with 16.8 mM glucose. Mean ± SEM; n = 5–6 biological replicates; 2-way ANOVA.

Figure 4. AAV-mediated overexpression of BAF60a preserves β cell function and glucose homeostasis in diabetic mice.

(A) Pancreatic ductal AAV infusion and experimental design for (B–K). (B) Relative BAF60a mRNA in islets pooled from 3 mice/group. Mean ± SD; n = 3 technical replicates; 2-tailed unpaired Student’s t test. (C–F) Random glucose (C), ipGTT (D), in vivo GSIS (E), and ITT (F) in BKS-db/db mice. Mean ± SEM; n = 5–6 (C, E, and F) or n = 6 (D); 2-way ANOVA. (G) Biphasic insulin secretion in islets pooled from 3 mice/group. Mean ± SD; n = 3 technical replicates; 2-tailed unpaired Student’s t test. (H) Intracellular Ca²⁺ influx in islets pooled from 3 mice/group. Mean ± SEM; n = 32–33 islets; 2-way ANOVA. (I–K) Transcriptomics of AAV-GFP– versus AAV-BAF60a–treated islets, heatmap (I), relative mRNA expression (J), and GO analysis (K) of differentially expressed genes. (L–N) Design (L), immunoblot (M), and biphasic insulin secretion (N) in islets pooled from 3 mice/group. Mean ± SD; n = 3 replicates; 2-tailed unpaired Student’s t test. (O–R) Design (O), fasting blood glucose (P), OGTT (Q), and in vivo GSIS (R) in STZ-treated and islet-transplanted mice. Mean ± SEM; n = 6–9 (P and Q) or 6–8 (R); 2-tailed unpaired Student’s t test (P) or 2-way ANOVA (Q and R).

Figure 5. BAF60a modulates β cell function via Nkx6.1-dependent chromatin remodeling.

(A) Metagene heatmap of ATAC-Seq signals across –3 kb to +3 kb of TSS (n = 3 biological replicates). (B) Differential accessibility analysis of ATAC peaks. MAPs (red) and LAPs (blue) defined by |log2FC| > 0.5, FDR < 0.05 (n = 3). (C) Metagene heatmap of BAF60a CUT&Tag-Seq signals and the Encyclopedia of DNA Elements (ENCODE) cis-regulatory elements around MAPs (red) and LAPs (blue). (D) GO enrichment of genes linked to LAPs. (E) Motif enrichment within LAPs with consensus sequence, transcription factor (TF) name, P value, and percentage of peaks containing motif. (F) ChIP-Seq enrichment of TFs at LAPs with fold enrichment and hypergeometric P values. (G) CUT&Tag-Seq heatmap and GO analysis of BAF60a-binding LAPs. (H) Browser tracks of Slc2a2 locus from RNA-Seq, ATAC-Seq, and CUT&Tag-Seq. (I) ATAC-Seq signal distribution at Nkx6.1-binding peaks (left) and TF footprinting of Nkx6.1 (right). (J) BAF60a interactome in Min6 cells. Red lines from BioID–mass spectrometry analysis, and gray lines from STRING datasets. (K and L) Immunoblots of BAF60a interactomes detected by BioID/Flag IP (K), and GST pull-down with GST-tagged BAF60a truncated proteins and His-Nkx6.1 (L). (M) BioID-immunoblots of BAF60a-Nkx6.1 interaction with or without TNF-α (50 ng/mL) and PA (0.5 mM) treatment. (N) GSIS in Min6 cells overexpressing BAF60a with or without Nkx6.1 KO under PA/TNF-α treatment. Mean ± SEM; n = 3 biological replicates; 2-way ANOVA. Strep, streptavidin.

Figure 6. Human BAF60a^V278M mutation impairs β cell insulin secretion.

(A) Domain structure of human BAF60a showing SWIB and predicted coiled-coil domains (C1–C3, orange) and variant position. (B) Evolutionary conservation of V278 across species. (C) Predicted 3D models of BAF60a^WT and BAF60a^V278M. (D) Serum insulin levels of BAF60a^WT and BAF60a^V278M groups after OGTT assays. Mean ± SEM (n ≥ 13,777 people in BAF60a^WT group and ≥57 people in BAF60a^V278M group); P value was determined by a nonparametric test (Mann-Whitney test) for a skewed distribution. (E) AUC of serum insulin levels in (D). Values are presented as median with interquartile range (n = 13,766 people in BAF60a^WT group and 58 people in BAF60a^V278M group); 2-tailed unpaired Student’s t test (F) The difference of δ30 [(30-minute serum insulin – fasting serum insulin)/(30-minute plasma glucose – fasting plasma glucose)] in (D). Values are presented as median with interquartile range (n = 13,376 people in the BAF60a^WT group and 54 people in the BAF60a^V278M group); 2-tailed unpaired Student’s t test.

Figure 7. Mice carrying the BAF60a^V278M mutation exhibit impaired insulin secretion and glucose homeostasis.

(A) Experimental design for (B–K). (B) Plasma insulin in chow-fed BAF60a^WT and BAF60a^V278M mice. Mean ± SEM; n = 5–6; 2-tailed unpaired Student’s t test. (C) ipGTT in chow-fed mice. Mean ± SEM; n = 6; 2-way ANOVA. (D) Biphasic insulin secretion in islets pooled from 3 mice. Mean ± SD; n = 3 technical replicates; 2-tailed unpaired Student’s t test. (E) Ca²⁺ influx in chow-fed islets pooled from 3 mice. Mean ± SEM; n = 60–64; 2-way ANOVA. (F) Overnight fasting blood glucose level in HFD-fed mice. Mean ± SEM; n = 5–6; 2-tailed unpaired Student’s t test. (G) ipGTT in HFD-fed mice. Mean ± SEM; n = 8–9; 2-way ANOVA. (H) ipITT in HFD-fed mice. Mean ± SEM; n = 5–6; 2-way ANOVA. (I) In vivo GSIS in HFD-fed mice. Mean ± SEM; n = 5–6; 2-way ANOVA. (J) Biphasic insulin secretion in HFD-fed islets pooled from 3 mice. Mean ± SD; n = 3 technical replicates; 2-tailed unpaired Student’s t test. (K) Ca²⁺ influx in islets pooled from 3 mice. Mean ± SEM; n = 45–53; 2-way ANOVA. (L) Relative mRNA expression in islets isolated from HFD-fed mice. Mean ± SEM; n = 4 biological replicates; 2-tailed unpaired Student’s t test. (M) Flag IP and immunoblot analysis of BAF60a^WT/BAF60a^V278M and Nkx6.1 interactions in HEK293T cells. (N and O) Experimental design (N), and fasting/refeeding glucose levels (O) in STZ-treated mice transplanted with islets transduced with indicated AAVs. Mean ± SEM; n = 6–8; 1-way ANOVA. (P) Immunoblots (left), perifusion (middle), and quantification of phase I/II insulin secretion (right) in human islets transduced with indicated AAVs. Mean ± SD; n = 3 technical replicates; 1-way ANOVA.

Figure 8. BAF60a contributes to the therapeutic benefits of GLP-1R agonists by modulating GLP-1R expression.

(A and B) Representative RNA-Seq, ATAC-Seq, and CUT&Tag-Seq tracks of Gipr (A) and Glp1r (B) loci. RNA-Seq and ATAC-Seq tracks for flox/flox and BaβKO islets. (C) Relative mRNA in isolated islets from indicated groups. Mean ± SEM; n = 3 replicates; 2-tailed unpaired Student’s t test. (D) Relative mRNA in human islets transduced with control or BAF60a siRNA. Mean ± SEM; n = 3 replicates; 2-tailed unpaired Student’s t test. (E) Immunoblots (left) and GSIS (right) of WT islets transduced with control or BAF60a siRNA (–3). Mean ± SEM; n = 4 replicates; 2-way ANOVA. (F) Relative mRNA in isolated islets from indicated groups. Mean ± SEM; n = 3 replicates; 2-tailed unpaired Student’s t test. (G) Dynamic (left) and quantification of phase I/II insulin secretion (right) in islets pooled from 3 mice. Mean ± SD; n = 3 technical replicates; 2-tailed unpaired Student’s t test. Interaction effect: P < 0.0001 (phase I) and P = 0.0016 (phase II), 2-way ANOVA.