During pancreas development, endocrine progenitors differentiate into the islet cell subtypes, which undergo further functional maturation in postnatal islet development. In islet β-cells, genes involved in glucose-stimulated insulin secretion are activated, and glucose exposure increases the insulin response as β-cells mature. We investigated the role of H3K4 trimethylation in endocrine cell differentiation and functional maturation by disrupting TrxG complex histone methyltransferase activity in mouse endocrine progenitors. In the embryo, genetic inactivation of TrxG component Dpy30 in NEUROG3+ cells did not affect the number of endocrine progenitors or endocrine cell differentiation. H3K4 trimethylation was progressively lost in postnatal islets, and the mice displayed elevated nonfasting and fasting glycemia as well as impaired glucose tolerance by postnatal day 24. Although postnatal endocrine cell proportions were equivalent to controls, islet RNA sequencing revealed a downregulation of genes involved in glucose-stimulated insulin secretion and an upregulation of immature β-cell genes. Comparison of histone modification enrichment profiles in NEUROG3+ endocrine progenitors and mature islets suggested that genes downregulated by loss of H3K4 trimethylation more frequently acquire active histone modifications during maturation. Taken together, these findings suggest that H3K4 trimethylation is required for the activation of genes involved in the functional maturation of pancreatic islet endocrine cells.
Introduction
Endocrine cell specification from pancreatic progenitors is initiated by the induction of the proendocrine factor Neurog3 and subsequent exit from the cell cycle (1–3). NEUROG3 drives expression of downstream transcription factors, such as Neurod1, Nkx2-2, Nkx6-1, Pax6, and Pdx1, which determine further differentiation of endocrine progenitors into hormone-expressing endocrine cells that migrate and coalesce to form proto-islet structures (1,4). In the first few weeks after birth, endocrine cells undergo functional maturation, which involves acquisition of glucose sensing and hormone secretion machinery (1,4–6). For β-cells, this transition from an immature to a mature state involves activation of the maturity marker Ucn3 (6,7), metabolic gene expression changes (e.g., switch from high-affinity hexokinase [Hk] to low-affinity glucokinase [Gck]), and improved glucose-stimulated insulin secretion as a result of increased glucose exposure (5,8). In addition, mature β-cells are defined by the repression of immature or “disallowed” β-cell genes that are often detected in stem-cell–derived pancreatic β-like cells and models of diabetes (5,9,10).
The gene expression changes that occur during endocrine cell differentiation are associated with specific histone modification changes at surrounding cis-regulatory loci. In addition to histone H3 lysine 27 acetylation (H3K27ac), genes that become activated during pancreas and endocrine cell differentiation are associated with histone H3 lysine 4 monomethylation (H3K4me1) at enhancers and H3K4 trimethylation (H3K4me3) at promoters (11–13).
The majority of H3K4 methylation is catalyzed by trithorax group (TrxG) complexes, which each contain one of six mammalian histone methyltransferases (i.e., SET1A/B, MLL1–4) and a minimum of four core proteins (i.e., ASH2L, DPY30, RBPP5, WDR5) (14–16). Although H3K4 methylation is associated with active chromatin, several studies have suggested that H3K4 methylation is dispensable for gene expression (17–21). Previously, we reported that loss of Dpy30 and H3K4 methylation from PDX1+ progenitors increased the proportion of CPA1+ acinar progenitors while NEUROG3+ endocrine progenitors were reduced, suggesting a role for H3K4 methylation in endocrine cell specification (22). However, this model left unresolved whether H3K4me3 is required for the transcription of genes essential for islet endocrine cell differentiation or functional maturation. As such, in this study, our objective was to determine the role of H3K4me3 in the differentiation and functional maturation of mouse pancreatic endocrine cells.
Research Design and Methods
Mouse Strains
All mice were held at the British Columbia Children’s Hospital Research Institute animal care facility, and ethical procedures were followed according to protocols approved by the University of British Columbia animal care committee. Previously generated Dpy30flox/flox mice (22) were crossed to Neurog3-Cre driver mice (23) to obtain conditional deletion of Dpy30 in endocrine progenitors. In all studies, knockout mice (Dpy30ΔN, Neurog3-Cre; Dpy30flox/flox) were compared with Cre-negative littermate controls (Dpy30flox/flox or Dpy30flox/wt).
Pancreas and Islet Isolations
For pancreas isolations, dissected mouse pancreas was placed directly into 4% paraformaldehyde in PBS on ice followed by fixation overnight at 4°C and further processing for histology. For islet isolations, the common bile duct was perfused with 3 mL collagenase XI (1,000 units/mL; Sigma-Aldrich) in Hanks’ balanced salt solution, digested with collagenase XI and then filtered through a 40 μm filter. Handpicked islets were recovered overnight at 37°C in a 5% CO2 humidified incubator or used immediately for experiments.
Immunostaining
Paraffin slides were incubated with primary antibodies at 4°C overnight and secondary antibodies for 1 h at room temperature in a humidified dark chamber (see Supplementary Material for antibody information and details of histology, imaging, and analysis).
Blood Glucose Measurements
Mice were monitored after weaning at postnatal day 21 (P21) for body mass and nonfasting blood glucose. For intraperitoneal glucose tolerance tests (IPGTTs), mice were fasted for 6 h prior to injection of 2 g/kg i.p. 20% d-glucose (Sigma-Aldrich). Blood glucose measurements were obtained prior to injection (T0) and 15, 30, 45, 60, and 120 min after injection.
RNA Extraction and Quantitative PCR Analysis
RNA was extracted from lysed cells using TRIzol reagent (Thermo Fisher Scientific) with the PureLink RNA Mini Kit (Ambion). cDNA was generated using SuperScript III Reverse Transcriptase (Thermo Fisher Scientific), and quantitative PCR (qPCR) experiments were performed using 0.25–2 μL of cDNA per reaction with a ViiA 7 Real-Time PCR system (Thermo Fisher Scientific). Gene expression was normalized to β-actin and determined using the ΔΔCt method.
RNA and Chromatin Immunoprecipitation Sequencing Analysis
For RNA sequencing (RNA-seq), islets from three control and three Dpy30ΔN male mice were isolated at P24 and 100–200 islets per mouse were handpicked. The NEBNext Poly(A) mRNA Magnetic Isolation Module (New England Biolabs) was used to enrich mRNA from 1,000 ng of total RNA, and cDNA libraries were prepared with the NEBNext Ultra II Directional RNA Library Prep Kit for Illumina (New England Biolabs) with nine amplification cycles. Indexed libraries were analyzed for size distribution using the Agilent Bioanalyzer High Sensitivity DNA chip and quantified using the Qubit dsDNA HS Assay Kit (Thermo Fisher Scientific). Pooled libraries were sequenced on an Illumina NextSeq 500 platform for 2 × 43-nucleotide paired-end reads. Resulting FastQ sequencing files were trimmed with Trimmomatic (24). Salmon (25) was used to quasi-map, assemble, and quantify transcripts using the mouse GRCm38/VM24 transcriptome. Differential gene expression analysis was performed using DEseq2 (26). Gene set enrichment analysis was performed with g:Profiler (27).
Chromatin immunoprecipitation sequencing (ChIP-seq) data for H3K4me3, H3K4me1, H3K27ac, and H3K27me3 were obtained from E13.5 Neurog3HI cells (28) and adult islets (13,29). Transcription factor binding data from adult islets and β-cell lines were obtained for FOXA2, MAFA, PDX1, and NEUROD1 (13), as well as NKX2-2 (30), NKX6-1 (31), PAX6 (32), and RFX6 (33). All data were mapped with Bowtie 2 to the GRCm38/mm10 genome, and transcription factor peaks were called with MACS2. ChromHMM software (34) was used to identify chromatin states.
Statistics
Data are expressed as mean ± SD unless otherwise specified, and all experiments were carried out at a minimum in triplicate. Statistical analyses were performed using GraphPad Prism 8 software. Statistical significance was determined using unpaired, two-tailed Student t tests for comparisons between two groups, and two-way ANOVA with Šidák post hoc test for multiple comparisons was used for repeated-measures comparisons among more than two groups (unless otherwise specified). P < 0.05 was considered significant.
Data and Resource Availability
Data generated and/or analyzed during the current study are available from the Gene Expression Omnibus under accession no. GSE174751 or from the corresponding author upon request.
Results
H3K4me3 Is Lost Postnatally in Dpy30ΔN Islets
To investigate the role of H3K4me3 in endocrine cell differentiation and maturation, we disrupted the TrxG complex core component Dpy30 in NEUROG3+ endocrine progenitor cells using Neurog3-Cre driver mice (23). DPY30 is required for TrxG catalytic activity or H3K4 methylation without affecting complex formation or coactivator activity (35,36). Previously described Dpy30flox/flox mice (22) were bred to generate experimental Neurog3-Cre; Dpy30flox/flox mice (hereafter known as Dpy30ΔN) and littermate Dpy30flox/flox (Cre-negative) control mice. In Dpy30ΔN embryos, DPY30 staining was absent from endocrine cells detected with the pan-endocrine marker chromogranin A (CHGA) beginning at embryonic day 14.5 (E14.5) and at P0, P7, P14, and P24 (Fig. 1A). However, DPY30 staining was maintained in the surrounding Dpy30ΔN exocrine pancreas, validating deletion of Dpy30 in the endocrine lineage. At P24, Cre recombination was highly efficient in Dpy30ΔN islets, as DPY30 staining was not detected in 95% of CHGA+ endocrine cells (Fig. 1B). Costaining of endocrine cells with CHGA and H3K4me3 revealed a delay between loss of DPY30 protein and H3K4 methylation (Fig. 1C). Consistently, H3K4me3 staining was absent from 95% of CHGA+Dpy30ΔN endocrine cells at P24 (Fig. 1D). Measuring the median H3K4me3 intensity in CHGA+ islets relative to surrounding non-CHGA+ H3K4me3 intensity levels revealed that H3K4me3 intensity was maintained in E14.5 CHGA+Dpy30ΔN endocrine cells similar to controls (Fig. 1E). However, H3K4me3 intensity was progressively decreased from P0 onward and was consistently absent from P14 endocrine cells and thereafter (Fig. 1E). At P24, Western blot analysis confirmed that both H3K4me3 (∼4.5-fold, P < 0.01) and H3K4me1 (∼2.2-fold, P < 0.05) were significantly reduced in Dpy30ΔN islets (Fig. 1F). These data suggest that disruption of Dpy30 in NEUROG3+ endocrine progenitors results in loss of DPY30 protein from embryonic endocrine cells and in the delayed loss of H3K4me3 in postnatal islets.
Dpy30 deletion in pancreas endocrine progenitor cells leads to progressive loss of H3K4 methylation in Dpy30ΔN islets. A: Staining of DPY30, CHGA, and nuclei (blue) in control and Dpy30ΔN pancreas from E14.5 to P24. B: The percentage of CHGA+DPY30+ cells in P24 control and Dpy30ΔN pancreas. C: Staining of H3K4me3, CHGA, and nuclei (blue) in control and Dpy30ΔN pancreas from E14.5 to P24. D: The percentage of CHGA+H3K4me3+ cells in P24 control and Dpy30ΔN pancreas. E: The median H3K4me3 intensity in control and Dpy30ΔN CHGA+ cells relative to surrounding non-CHGA+ cell H3K4me3 intensity from E14.5 to P24. F: Western blot and densitometry for H3K4me3 and H3K4me1 relative to histone H3 from P24 control and Dpy30ΔN islets. Data are mean ± SD (n ≥ 3). Scale bars = 25 μm (E14.5, P0) and 75 μm (P7–P24). *P < 0.05, **P < 0.01, ****P < 0.0001, Dpy30ΔN vs. control by unpaired, two-tailed Student t test.
Dpy30 deletion in pancreas endocrine progenitor cells leads to progressive loss of H3K4 methylation in Dpy30ΔN islets. A: Staining of DPY30, CHGA, and nuclei (blue) in control and Dpy30ΔN pancreas from E14.5 to P24. B: The percentage of CHGA+DPY30+ cells in P24 control and Dpy30ΔN pancreas. C: Staining of H3K4me3, CHGA, and nuclei (blue) in control and Dpy30ΔN pancreas from E14.5 to P24. D: The percentage of CHGA+H3K4me3+ cells in P24 control and Dpy30ΔN pancreas. E: The median H3K4me3 intensity in control and Dpy30ΔN CHGA+ cells relative to surrounding non-CHGA+ cell H3K4me3 intensity from E14.5 to P24. F: Western blot and densitometry for H3K4me3 and H3K4me1 relative to histone H3 from P24 control and Dpy30ΔN islets. Data are mean ± SD (n ≥ 3). Scale bars = 25 μm (E14.5, P0) and 75 μm (P7–P24). *P < 0.05, **P < 0.01, ****P < 0.0001, Dpy30ΔN vs. control by unpaired, two-tailed Student t test.
Loss of Dpy30 Does Not Affect NEUROG3+ Endocrine Progenitor Cell Specification
To determine whether deletion of Dpy30 in NEUROG3+ endocrine progenitors altered the proportion of islet endocrine cells, we first examined the ratio of NEUROG3+ endocrine progenitors to SOX9+ pancreas progenitors and determined that there was no significant difference in E14.5 control and Dpy30ΔN pancreas (Fig. 2A and B). Next, costaining for insulin+ β-cells, glucagon+ α-cells, and somatostatin+ δ-cells demonstrated that the sum of these endocrine cell types relative to total pancreas cells was not altered in the E18.5 Dpy30ΔN pancreas compared with controls (Fig. 2C and D). The relative proportion of Dpy30ΔN islet insulin+ β-cells, glucagon+ α-cells, and somatostatin+ δ-cells was also unaffected both in E18.5 pancreas (Fig. 2E) and postnatally in P24 Dpy30ΔN pancreas (Fig. 2F and G). In addition, there were no significant differences in Dpy30ΔN islet endocrine cell proportions when examined at 5 weeks of age (Fig. 2I and J). Although there were no changes in islet endocrine cell proportions, we noted that the insulin staining in Dpy30ΔN islets appeared reduced compared with controls (Fig. 2F). Furthermore, quantification of CHGA+ area at P24 revealed no changes in total endocrine cell area between control and Dpy30ΔN pancreas (Fig. 2H). These data suggest that disruption of Dpy30 in NEUROG3+ progenitors does not affect the proportion of differentiated endocrine cells in the embryo or postnatal pancreas.
The proportion of endocrine cells is not altered in Dpy30ΔN mice. A: Staining for SOX9, NEUROG3, and nuclei (blue) in E14.5 control and Dpy30ΔN pancreas. B: The NEUROG3+/SOX9+ cell ratio in E14.5 control and Dpy30ΔN pancreas. C: Costaining for insulin (INS), glucagon (GCG), somatostatin (SST), and nuclei (gray) in E18.5 control and Dpy30ΔN pancreas. D: The fraction of endocrine cells (sum of INS+, GCG+, and SST+ cells) relative to total pancreas cells in E18.5 control and Dpy30ΔN pancreas. E: The proportions of INS+ β-cells, GCG+ α-cells, and SST+ δ-cells relative to total endocrine cells in E18.5 control and Dpy30ΔN pancreas. F: Costaining for INS, GCG, SST, and nuclei (gray) in P24 control and Dpy30ΔN pancreas. G: The proportions of INS+, GCG+, and SST+ cells relative to total endocrine cells in P24 control and Dpy30ΔN pancreas. H: Quantification of CHGA+ area in P24 control and Dpy30ΔN pancreas. I: Costaining for INS, GCG, SST, and nuclei (gray) in control and Dpy30ΔN pancreas at 5 weeks. J: The proportions of INS+, GCG+, and SST+ cells relative to total endocrine cells in control and Dpy30ΔN pancreas at 5 weeks. Data are mean ± SD (n = 3). Scale bars = 75 μm. Dpy30ΔN vs. control by unpaired, two-tailed Student t test.
The proportion of endocrine cells is not altered in Dpy30ΔN mice. A: Staining for SOX9, NEUROG3, and nuclei (blue) in E14.5 control and Dpy30ΔN pancreas. B: The NEUROG3+/SOX9+ cell ratio in E14.5 control and Dpy30ΔN pancreas. C: Costaining for insulin (INS), glucagon (GCG), somatostatin (SST), and nuclei (gray) in E18.5 control and Dpy30ΔN pancreas. D: The fraction of endocrine cells (sum of INS+, GCG+, and SST+ cells) relative to total pancreas cells in E18.5 control and Dpy30ΔN pancreas. E: The proportions of INS+ β-cells, GCG+ α-cells, and SST+ δ-cells relative to total endocrine cells in E18.5 control and Dpy30ΔN pancreas. F: Costaining for INS, GCG, SST, and nuclei (gray) in P24 control and Dpy30ΔN pancreas. G: The proportions of INS+, GCG+, and SST+ cells relative to total endocrine cells in P24 control and Dpy30ΔN pancreas. H: Quantification of CHGA+ area in P24 control and Dpy30ΔN pancreas. I: Costaining for INS, GCG, SST, and nuclei (gray) in control and Dpy30ΔN pancreas at 5 weeks. J: The proportions of INS+, GCG+, and SST+ cells relative to total endocrine cells in control and Dpy30ΔN pancreas at 5 weeks. Data are mean ± SD (n = 3). Scale bars = 75 μm. Dpy30ΔN vs. control by unpaired, two-tailed Student t test.
Dpy30ΔN Mice Develop Hyperglycemia and Impaired Glucose Tolerance
To investigate whether the postnatal islet reduction in H3K4 methylation had a biological effect in Dpy30ΔN mice, we first monitored body mass and nonfasting blood glucose levels. Daily tracking between P24 and P38 indicated that while body mass was not altered in male Dpy30ΔN mice compared with controls (Fig. 3A), mean ad libitum glycemia was >18 mmol/L by P28 and worsened over time (Fig. 3B). A similar trend was observed in female Dpy30ΔN mice (Supplementary Fig. 1A and B). Following a 6-h fast at P23, blood glucose levels were equivalent in control and Dpy30ΔN male mice (Fig. 3C). IPGTTs at this stage (P23) revealed a marginal increase to ∼19 mmol/L at 30 min, although blood glucose measurements at other time points were otherwise comparable to controls (Fig. 3D). Quantification of the IPGTT area under the curve (AUC) at P23 demonstrated a significant impairment in Dpy30ΔN male mice (Fig. 3E). At P24, although fasting blood glucose levels remained similar between control and Dpy30ΔN male mice (Fig. 3F), blood glucose was significantly elevated in Dpy30ΔN males following an i.p. glucose challenge (Fig. 3G and H). We measured islet insulin content and detected a ∼40% reduction in insulin levels from P24 male Dpy30ΔN islets relative to controls (Fig. 3I). A ∼50% reduction in islet insulin content was also detected in female Dpy30ΔN mice (Supplementary Fig. 1C). At P25, fasting glycemia levels were elevated to ∼12 mmol/L in Dpy30ΔN male mice (Fig. 3J), and blood glucose levels were above the limit of detection (33.3 mmol/L) at 15- and 30-min postglucose injection (Fig. 3K). AUC measurements of P25 glucose tolerance tests confirmed a significant elevation in Dpy30ΔN blood glucose levels relative to controls (Fig. 3L). These data suggest that Dpy30ΔN mice develop nonfasting and fasting hyperglycemia and impaired glucose tolerance and have decreased islet insulin content.
Dpy30ΔN mice develop hyperglycemia and impaired glucose tolerance. A: Mouse body mass measurements between P24 and P38 in male control and Dpy30ΔN animals. B: Nonfasting blood glucose measurements from male Dpy30ΔN and control mice between P24 and P38. C: Blood glucose measurements after a 6-h fast in P23 male control and Dpy30ΔN mice. D: IPGTT of 2 g/kg body mass i.p. glucose in P23 male control and Dpy30ΔN mice following a 6-h fast. E: Quantification of AUC in D for P23 male control and Dpy30ΔN mice. F: Blood glucose measurements after a 6-h fast in P24 male control and Dpy30ΔN mice. G: IPGTT of 2 g/kg body mass i.p. glucose in P24 male control and Dpy30ΔN mice following a 6-h fast. H: Quantification of AUC in G for P24 male control and Dpy30ΔN mice. I: Islet insulin content relative to DNA from P24 male control and Dpy30ΔN islets. J: Blood glucose measurements after a 6-h fast in P25 male control and Dpy30ΔN mice. K: IPGTT of 2 g/kg body mass i.p. glucose in P25 male control and Dpy30ΔN mice following a 6-h fast. L: Quantification of AUC in K for P25 male control and Dpy30ΔN mice. Data are mean ± SD (n ≥ 3). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, Dpy30ΔN vs. control by multiple t tests with Holm-Šidák correction for multiple comparisons (A and B) or repeated-measures two-way ANOVA with Šidák post hoc test for multiple comparisons (D, G, and K). Note that statistical testing was not performed at 15 and 30 min (in K) because measurements were above the detection limit (33.3 mmol/L). An unpaired, two-tailed Student t test was performed in all other panels.
Dpy30ΔN mice develop hyperglycemia and impaired glucose tolerance. A: Mouse body mass measurements between P24 and P38 in male control and Dpy30ΔN animals. B: Nonfasting blood glucose measurements from male Dpy30ΔN and control mice between P24 and P38. C: Blood glucose measurements after a 6-h fast in P23 male control and Dpy30ΔN mice. D: IPGTT of 2 g/kg body mass i.p. glucose in P23 male control and Dpy30ΔN mice following a 6-h fast. E: Quantification of AUC in D for P23 male control and Dpy30ΔN mice. F: Blood glucose measurements after a 6-h fast in P24 male control and Dpy30ΔN mice. G: IPGTT of 2 g/kg body mass i.p. glucose in P24 male control and Dpy30ΔN mice following a 6-h fast. H: Quantification of AUC in G for P24 male control and Dpy30ΔN mice. I: Islet insulin content relative to DNA from P24 male control and Dpy30ΔN islets. J: Blood glucose measurements after a 6-h fast in P25 male control and Dpy30ΔN mice. K: IPGTT of 2 g/kg body mass i.p. glucose in P25 male control and Dpy30ΔN mice following a 6-h fast. L: Quantification of AUC in K for P25 male control and Dpy30ΔN mice. Data are mean ± SD (n ≥ 3). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, Dpy30ΔN vs. control by multiple t tests with Holm-Šidák correction for multiple comparisons (A and B) or repeated-measures two-way ANOVA with Šidák post hoc test for multiple comparisons (D, G, and K). Note that statistical testing was not performed at 15 and 30 min (in K) because measurements were above the detection limit (33.3 mmol/L). An unpaired, two-tailed Student t test was performed in all other panels.
H3K4me3 Is Necessary for the Expression of Mature Pancreatic Islet Genes
To understand the underlying mechanism for hyperglycemia and impaired glucose tolerance in Dpy30ΔN mice, we next examined whether islet dysfunction was the result of reduced endocrine cell maturity. To assess islet maturation, islets from male Dpy30ΔN and control mice were collected at P24, and RNA-seq was performed. Differential gene expression analysis demonstrated that endocrine cell hormones (e.g., Gcg, Ins1, Ins2, Ppy, Sst) and genes involved in insulin secretion or β-cell maturation (e.g., Abcc8, Glp1r, Kcnj11, Slc2a2, Slc30a8, Ucn3) were downregulated in Dpy30ΔN islets compared with controls (Fig. 4A and B). Interestingly, the glucagon receptor (Gcgr) was one of the most downregulated genes (∼16-fold reduction) in Dpy30ΔN islets. Genes in the glucose-stimulated insulin secretion pathway were particularly affected by decreased H3K4 methylation in maturing pancreatic islets (Fig. 4C). Elevated genes in Dpy30ΔN islets included genes associated with immature β-cells (e.g., Aldh1a3, Gast, Hk1/2, Ldha, Npy, Pdgfra, Rest, Slc16a1) and several genes involved in cell adhesion (e.g., Apoa1, Ccn3, Cldn2, Col16a1, Itga11, Lamc3) (Fig. 4A and B). RNA-seq analysis also revealed that neither mature islet transcription factors (e.g., Mafa, Neurod1, Nkx2-2, Nkx6-1, Pax6, Pdx1) nor genes involved in mTOR signaling (e.g., Akt1, Mtor, Pdk1, Rptor) were differentially expressed in Dpy30ΔN islets (Supplementary Fig. 3). Gene Ontology (GO) term analysis indicated that genes involved in ion and transmembrane transport and hormone and insulin secretion were reduced in Dpy30ΔN cells, while genes involved in extracellular matrix formation and cell adhesion were elevated (Fig. 4D).
Islet maturity is compromised in Dpy30ΔN mice. A: Volcano plot of differentially expressed genes in P24 Dpy30ΔN vs. control islets with log2 fold change plotted against the adjusted P value [−log10(padj)]. Downregulated genes are in blue, upregulated genes are in red, and stable genes are in black. B: Heatmaps of select gene expression (log2 fold change) comparing three biological replicates across P24 control and Dpy30ΔN islets. C: Overview of the glucose-stimulated insulin secretion pathway with affected genes in P24 Dpy30ΔN islets represented by log2 fold change values. D: g:Profiler GO term analysis of downregulated and upregulated genes from P24 control and Dpy30ΔN islets represented as −log10(padj). E: Heatmap of transcription factor occupancy from published adult islet ChIP-seq data at downregulated, upregulated, and stable genes in P24 Dpy30ΔN vs. control islets. F: Histone modification enrichment plot of published adult islet ChIP-seq data indicating repressed (H3K27me3), unmarked, poised (H3K4me1/3), active (H3K4me1/3 and H3K27ac), and bivalent (H3K4me1 and H3K27me3) chromatin states. G: Heatmap showing the fraction of downregulated, upregulated, and stable genes in P24 Dpy30ΔN vs. control islets in each chromatin state identified in F. BP, biological process; CC, cellular component; FC, fold change; KO, knockout; TCA, tricarboxylic acid; WT, wild type.
Islet maturity is compromised in Dpy30ΔN mice. A: Volcano plot of differentially expressed genes in P24 Dpy30ΔN vs. control islets with log2 fold change plotted against the adjusted P value [−log10(padj)]. Downregulated genes are in blue, upregulated genes are in red, and stable genes are in black. B: Heatmaps of select gene expression (log2 fold change) comparing three biological replicates across P24 control and Dpy30ΔN islets. C: Overview of the glucose-stimulated insulin secretion pathway with affected genes in P24 Dpy30ΔN islets represented by log2 fold change values. D: g:Profiler GO term analysis of downregulated and upregulated genes from P24 control and Dpy30ΔN islets represented as −log10(padj). E: Heatmap of transcription factor occupancy from published adult islet ChIP-seq data at downregulated, upregulated, and stable genes in P24 Dpy30ΔN vs. control islets. F: Histone modification enrichment plot of published adult islet ChIP-seq data indicating repressed (H3K27me3), unmarked, poised (H3K4me1/3), active (H3K4me1/3 and H3K27ac), and bivalent (H3K4me1 and H3K27me3) chromatin states. G: Heatmap showing the fraction of downregulated, upregulated, and stable genes in P24 Dpy30ΔN vs. control islets in each chromatin state identified in F. BP, biological process; CC, cellular component; FC, fold change; KO, knockout; TCA, tricarboxylic acid; WT, wild type.
Transcription factor occupancy and histone modification status at downregulated, upregulated, and stably expressed genes in Dpy30ΔN islets were examined using previously published adult mouse islet ChIP-seq data (13,29–33). The islet transcription factors NKX2-2, NKX6-1, and RFX6 were marginally depleted at upregulated genes (Fig. 4E). We used ChromHMM to stratify the histone modification ChIP-seq data into chromatin states, including repressed (H3K27me3), unmarked, poised (H3K4me1/3), active (H3K4me1/3 and H3K27ac), and bivalent (H3K4me1 and H3K27me3) chromatin (Fig. 4F). Histone modifications associated with active chromatin were relatively increased at the promoters of downregulated and stable genes (Fig. 4G). In contrast, the cis-regulatory elements governing the upregulated genes were frequently in an inactive chromatin state, with either unmarked histones or enrichment of a repressive, bivalent, or poised chromatic state (Fig. 4G). Together, these results suggest that hyperglycemia and impaired glucose tolerance in Dpy30ΔN mice manifests from the reduced expression of genes required for islet β-cell maturity and glucose-stimulated insulin secretion.
Failed Activation of Proendocrine cis-Regulatory Elements in Dpy30ΔN Mice
To explore whether downregulated, upregulated, and stably expressed genes in Dpy30ΔN islets require distinct chromatin state changes during pancreatic endocrine cell development to become active, we compared the histone modification profiles of these genes in Neurog3HI cells and in adult islets (13,28,30–33). Overall, downregulated, upregulated, and stable genes had similar H3K4me3, H3K4me1, H3K27ac, and H3K27me3 profiles at their promoters in Neurog3HI endocrine progenitors as in adult islets (Fig. 5A–D), although an overall increase in H3K4me3 at these genes was evident in adult islets compared with in Neurog3HI cells. Furthermore, H3K4me1 profiles revealed, though, that promoters of downregulated and upregulated genes were blocked in Neurog3HI cells (37), whereas H3K4me1-marked histones shifted away from the transcription start site (TSS) in adult islets (Fig. 5B). We next used ChromHMM (34) to annotate chromatin states identified in Neurog3HI endocrine progenitors and those identified in mature endocrine islet cells at downregulated, upregulated, and stable genes (Fig. 5E). This analysis showed that a significantly increased fraction of genes downregulated in Dpy30ΔN islets transitioned to an active chromatin state (H3K4me1/3 and/or H3K27ac) during development (Fig. 5F). Conversely, we found a significant reduction in repressed chromatin (H3K27me3 or bivalent) in adult islets at genes that were downregulated in Dpy30ΔN islets (Fig. 5F). University of California, Santa Cruz, Genome Browser views at the TSS of Glp1r, Slc2a2, and Slc30a8 demonstrate acquisition of active histone marks (increased H3K4me3 and H3K27ac) during endocrine cell maturation at these loci (Fig. 5G). Overall, these data suggest that downregulated genes in Dpy30ΔN islets are more likely to lose H3K27me3 and instead acquire active histone modifications during development.
Failure to activate proendocrine cis-regulatory elements in Dpy30ΔN mice. A–D: Comparison of histone modification profiles in E13.5 Neurog3HI cells vs. adult islets around the TSS of downregulated, upregulated, and stable genes in P24 Dpy30ΔN vs. control islets. Distributions for H3K4me3 (A), H3K4me1 (B), H3K27ac (C), and H3K27me3 (D) are shown ± 3.0 kb around the TSS. E: Percentage of total genes in each chromatin state (unmarked, repressed [H3K27me3], bivalent [H3K27me3 and H3K4me1], K4me1, K4me3 and K4me1, K4me1 and K27ac, K4me3 and K27ac, K4me1/3 and K27ac) for P24 islet Dpy30ΔN downregulated, upregulated, and stable genes in Neurog3HI cells and islets. F: Sum of active and repressed chromatin as defined in E for P24 islet Dpy30ΔN downregulated, upregulated, and stable genes in Neurog3HI cells and islets. Active chromatin includes all chromatin states containing H3K4me3, and repressed chromatin includes both repressed and bivalent states. Statistical testing was performed using Fisher exact test. G: Comparison of University of California, Santa Cruz, Genome Browser tracks at Glp1r, Slc2a2, and Slc30a8 for H3K4me3, H3K4me1, H3K27ac, and H3K27me3 in Neurog3HI cells and islets. *P < 0.05.
Failure to activate proendocrine cis-regulatory elements in Dpy30ΔN mice. A–D: Comparison of histone modification profiles in E13.5 Neurog3HI cells vs. adult islets around the TSS of downregulated, upregulated, and stable genes in P24 Dpy30ΔN vs. control islets. Distributions for H3K4me3 (A), H3K4me1 (B), H3K27ac (C), and H3K27me3 (D) are shown ± 3.0 kb around the TSS. E: Percentage of total genes in each chromatin state (unmarked, repressed [H3K27me3], bivalent [H3K27me3 and H3K4me1], K4me1, K4me3 and K4me1, K4me1 and K27ac, K4me3 and K27ac, K4me1/3 and K27ac) for P24 islet Dpy30ΔN downregulated, upregulated, and stable genes in Neurog3HI cells and islets. F: Sum of active and repressed chromatin as defined in E for P24 islet Dpy30ΔN downregulated, upregulated, and stable genes in Neurog3HI cells and islets. Active chromatin includes all chromatin states containing H3K4me3, and repressed chromatin includes both repressed and bivalent states. Statistical testing was performed using Fisher exact test. G: Comparison of University of California, Santa Cruz, Genome Browser tracks at Glp1r, Slc2a2, and Slc30a8 for H3K4me3, H3K4me1, H3K27ac, and H3K27me3 in Neurog3HI cells and islets. *P < 0.05.
H3K4me3 Is Required for Pancreatic Endocrine Cell Maturation
To determine whether the endocrine cell terminal markers Ins1, Ins2, Gcg, Sst, and Ppy that require activation during endocrine development fail to fully activate or instead lose expression in Dpy30ΔN islets, we used qPCR to quantify their expression at P7, P14, and P24. We found that expression of these markers at P7 was very similar between Dpy30ΔN and control islets (Fig. 6A). The expression of all of these markers significantly increased in control islets by P24, but no increase was detected in Dpy30ΔN islets at P14 or P24 compared with at P7 (Fig. 6A). In contrast, Npy and Gast, which are known to have elevated expression in immature islets, were significantly increased at P24 in Dpy30ΔN islets compared with control islets (Fig. 6A). Staining pancreas sections from P24 control and Dpy30ΔN mice confirmed that gastrin (GAST) protein levels were also elevated in Dpy30ΔN islets (Fig. 6B). In addition, we detected dramatically reduced levels of the mature β-cell proteins GLUT2 and glucagon-like peptide 1 receptor (GLP-1R) in P24 Dpy30ΔN islets (Fig. 6B). To determine whether reductions in islet hormones and proteins in the insulin secretion pathway altered Dpy30ΔN islet function, we performed a glucose-stimulated secretion assay. While P24 Dpy30ΔN islets responded to low and high glucose similar to control islets, we detected a significant reduction in insulin secretion from Dpy30ΔN islets when stimulated with KCl (Fig. 6C). Combined, these data suggest that H3K4me3 is necessary for the appropriate activation of mature endocrine cell genes during postnatal islet functional maturation.
Dpy30ΔN endocrine islet cells do not undergo functional maturation. A: Relative expression of Ins1, Ins2, Gcg, Sst, Ppy, Npy, and Gast transcripts in P7, P14, and P24 control and Dpy30ΔN islets as determined by qPCR. B: Staining for GAST, GLUT2, GLP-1R, CHGA, and nuclei (blue) in P24 control and Dpy30ΔN pancreas. C: Insulin secretion from P24 control and Dpy30ΔN islets stimulated with low glucose (2.8 mmol/L), high glucose (16.7 mmol/L), and KCl (30 mmol/L). Data are mean ± SD (n ≥ 3). Scale bar = 50 μm. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, Dpy30ΔN vs. control, unpaired, two-tailed Student t test.
Dpy30ΔN endocrine islet cells do not undergo functional maturation. A: Relative expression of Ins1, Ins2, Gcg, Sst, Ppy, Npy, and Gast transcripts in P7, P14, and P24 control and Dpy30ΔN islets as determined by qPCR. B: Staining for GAST, GLUT2, GLP-1R, CHGA, and nuclei (blue) in P24 control and Dpy30ΔN pancreas. C: Insulin secretion from P24 control and Dpy30ΔN islets stimulated with low glucose (2.8 mmol/L), high glucose (16.7 mmol/L), and KCl (30 mmol/L). Data are mean ± SD (n ≥ 3). Scale bar = 50 μm. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, Dpy30ΔN vs. control, unpaired, two-tailed Student t test.
Discussion
TrxG complex proteins are enriched in mature islets (38) and interact with the maturation factor MAFA in mature β-cells (39), suggesting a role for TrxG complexes and/or H3K4 methylation in maintenance of islet identity. Deletion of the TrxG gene Ncoa6 in embryonic β-cells does not affect Mafa expression but results in reduced downstream activation of MAFA target genes and impaired glucose-stimulated insulin secretion (39). Importantly, decreased gene expression was correlated with reductions in H3K4me3 and RNA polymerase II at the TSS of affected genes. In addition, H3K4me3 deposited by the SET7/9 histone methyltransferase was linked to transcriptional maintenance of genes involved in glucose-stimulated insulin secretion in primary islets (40). Together, these studies suggested that H3K4me3 may play a role in maintaining β-cell gene expression, but whether H3K4me3 is required during the process of endocrine cell maturation has not been examined.
In this study, we demonstrate that genetic inactivation of Dpy30 in NEUROG3+ cells does not alter the proportion of pancreatic endocrine progenitors or endocrine cells. Global loss of H3K4me3 occurred postnatally during islet cell maturation. H3K4me1 levels were reduced but not to the same extent, suggesting that the bulk of the effects seen in our mice are due to the loss of H3K4me3 or H3K4 methylation reductions in general.
By P28, Dpy30ΔN mice displayed elevated nonfasting and fasting glycemia in addition to impaired glucose tolerance by P24. These phenotypes could not be explained by changes in Dpy30ΔN endocrine cell area or the relative proportions of islet cells. While Neurog3 is also highly expressed in the brain, we found several intrinsic islet defects that explain the Dpy30ΔN phenotype. At P24, islet insulin content and expression of insulin, and genes associated with insulin secretion and β-cell maturity, were reduced in Dpy30ΔN islets. We also detected impaired activation of Ins1 and Ins2 during the postnatal islet maturation period. In P24 Dpy30ΔN islets, the critical glucose transporter GLUT2 was dramatically reduced, and KCl-stimulated insulin secretion was impaired. Together, these data suggest that H3K4me3 is required to activate genes involved in β-cell functional maturation.
Hyperglycemia and Impaired Glucose Tolerance Correlates With Loss of Islet H3K4me3
As discussed above, loss of endocrine cell H3K4me1 and H3K4me3 was delayed in Dpy30ΔN mice. Although DPY30 was absent from CHGA+ endocrine cells at E14.5, loss of H3K4 methylation did not occur until P14. This difference is likely, at least in part, the result of failed H3K4 methylation reestablishment during cell division. At the time when Dpy30 is genetically inactivated, NEUROG3+ cells exit the cell cycle and differentiate into endocrine cells (3). These cells remain nonproliferative until the early postnatal period when endocrine cell replication increases (41), suggesting that there is progressive loss of H3K4 methylation after each round of endocrine cell division. Of note, we observed no phenotype in Dpy30ΔN mice prior to the loss of H3K4 methylation, suggesting that loss of DPY30 itself has a minimal effect and that the impaired endocrine cell maturation in Dpy30ΔN mice is due to the absence of H3K4 methylation.
The loss of H3K4 methylation from Dpy30ΔN islets at P14 overlaps with the period of endocrine cell maturation. A key determinant of β-cell maturity is the ability to effectively respond to glucose, which includes efficient glucose transport and metabolism as well as regulated insulin secretion. During the first few weeks after birth, endocrine cells undergo functional maturation and develop glucose-stimulated hormone secretion by P14 (5,6,42). Further β-cell maturation is triggered by exposure to a carbohydrate-rich diet after weaning at P21 (8). The transition to glucose metabolism as the primary energy source exposes maturing islets to increased glucose levels, and this stimulates β-cell proliferation, enhances insulin secretion, and promotes oxidative phosphorylation (8). Following weaning, Dpy30ΔN mice developed hyperglycemia and impaired glucose tolerance beginning at P24. Measurement of islet insulin content and our gene expression data confirmed that insulin levels were significantly reduced in P24 Dpy30ΔN islets. We detected no significant difference in P24 islet area relative to controls, suggesting that hyperglycemia did not result from impaired β-cell proliferation. Glucose-stimulated insulin secretion assays revealed that Dpy30ΔN islets respond to low- and high-glucose stimulation similarly to controls; however, the response to depolarizing KCl stimulation was significantly reduced. This result suggested that in addition to reduced insulin levels, Dpy30ΔN β-cells have an intrinsic functional defect in first-phase insulin secretion and are not fully mature (43). Taken together, our data suggest that H3K4 methylation is required for the activation of genes essential for complete β-cell function and maturation.
Pancreatic Endocrine Cell Maturation Is Impaired in Dpy30ΔN Mice
Specification of the pancreatic endocrine lineage begins with activation of Neurog3 in a subset of SOX9+ pancreas progenitors in the developing embryo. NEUROG3+ cells delaminate and migrate away from the trunk epithelium into the surrounding mesenchyme, where they eventually differentiate into endocrine cells and coalesce into mature spherical islet structures (4). Extracellular matrix proteins, such as cadherins, collagens, integrins, and laminins, have an essential role in cell migration and cell-cell adhesion during endocrine cell aggregation into islet clusters (4,43,44). Interestingly, we detected a disproportionate upregulation of many genes associated with the extracellular matrix and cell-cell adhesion, including Cldn2, Apoa1, Col16a1, Ccn3, Lamc3, and Itga11, in P24 Dpy30ΔN islets, that play important roles in endocrine cell migration, isletogenesis, and maintenance of intercellular communication (45).
The transition from immature to mature β-cells involves important gene expression changes, where genes exclusive to immature β-cells (e.g., Hk1, Ldha, Rest, Pdgfra) become repressed, and mature genes associated with insulin secretion machinery (e.g., Gck, Slc2a2, Slc30a8) are induced (5). In Dpy30ΔN islets, the expression of several mature genes required for glucose transport (Slc2a2), glucose sensing (Gck), and insulin secretion (Abcc8, Kcnj11) was significantly reduced in addition to the mature β-cell marker Ucn3 (6). Meanwhile, several genes associated with endocrine cell immaturity, such as Aldh1a3, Gast, Hk2, Pdgfra, and Rest, were elevated in Dpy30ΔN islets. In addition, we detected GAST+ endocrine cells exclusively in P24 Dpy30ΔN islets as well as dramatic reductions in GLUT2 and GLP-1R. Transcript levels of islet hormones (e.g., Ins1, Ins2, Gcg, Ppy, Sst) were also significantly reduced in Dpy30ΔN islets. Furthermore, we detected a significant decrease in Dpy30ΔN islet insulin secretion in response to KCl stimulation, suggesting that β-cell depolarization was incomplete. Together, these data suggest that at P24, Dpy30ΔN islet cells fail to fully mature. It is important to note that this reduced maturity is without significant changes to the expression of key islet transcription factors (e.g., Mafa, Neurod1, Nkx2-2, Nkx6-1, Pax6, Pdx1) (Supplementary Fig. 2). We also did not observe significant gene expression changes in the mTOR pathway in Dpy30ΔN islets (Supplementary Fig. 2), which plays an important role in islet functional maturation (46–48).
H3K4 Methylation Is Acquired at Mature β-Cell Genes During Functional Maturation
We found that genes upregulated in P24 Dpy30ΔN islets were slightly more likely to become H3K27me3 marked or “unmarked” in islets compared with in Neurog3HI cells. This agrees with our gene expression data and suggests that a subset of these genes normally is more abundant in progenitors or immature endocrine cells and become “turned off” in mature pancreas endocrine cells. It is less clear why some genes, specifically endocrine cell terminal markers such as Ins1, Ins2, Gcg, Sst, and Ppy, and genes with functional relevance in β-cells fail to be activated in Dpy30ΔN endocrine cells, while other genes are unaffected. Our analysis of the chromatin state changes at endocrine cell-specific genes from Neurog3HI to adult islets suggests that they are only slightly more likely to be activated and to gain active chromatin marks and are otherwise very similar in chromatin state to genes unaffected by the loss of H3K4me3. Other studies have reported similar results and have found that H3K4me3 loss primarily affects the expression of lineage-specific genes (49,50). Why this is the case and whether H3K4me3 is required purely to activate such genes, or whether it is also essential to maintain lineage-specific expression, are unknown and warrant further investigation.
Regardless, the results in this study suggest that H3K4me3 is required for the expression of pancreatic islet genes involved in endocrine cell functional maturation. This conclusion is supported by evidence that H3K4 methylation is established at these genes during the endocrine cell maturation period. Overall, our data suggest that islet endocrine cells do not completely mature in the absence of H3K4 methylation.
This article contains supplementary material online at https://doi.org/10.2337/figshare.15124854.
Article Information
Acknowledgments. The authors would like to thank the staff of the animal care facility at British Columbia Children’s Hospital Research Institute for daily maintenance of the mouse colonies.
Funding. This work was funded by the British Columbia Children’s Hospital Research Institute, the Natural Sciences and Engineering Research Council of Canada (RGPIN-2016-04292), and the Canadian Institute of Nutrition, Metabolism and Diabetes (RN310864-375894).
Duality of Interest. No potential conflicts of interest relevant to this article were reported.
Author Contributions. S.A.C., J.B., C.L.M., B.V., and T.L.S. conducted the experiments. S.A.C. and B.G.H. conceptualized the study and wrote the manuscript. B.G.H. provided funding and supervision. B.G.H. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Prior Presentation. Parts of this study were presented in abstract form at the 53rd Annual Meeting of the European Association for the Study of Diabetes, Lisbon, Portugal, 11–15 September 2017. This study has been posted on the bioRxiv preprint server.