Deletion of Ascl1 in pancreatic β-cells improves insulin secretion, promotes parasympathetic innervation, and attenuates dedifferentiation during metabolic stress

Abstract

Objective

ASCL1, a pioneer transcription factor, is essential for neural cell differentiation and function. Previous studies have shown that Ascl1 expression is increased in pancreatic β-cells lacking functional K_ATP channels or after feeding of a high fat diet (HFD) suggesting that it may contribute to the metabolic stress response of β-cells.

Methods

We generated β-cell-specific Ascl1 knockout mice (Ascl1^βKO) and assessed their glucose homeostasis, islet morphology and gene expression after feeding either a normal diet or HFD for 12 weeks, or in combination with a genetic disruption of Abcc8, an essential K_ATP channel component.

Results

Ascl1 expression is increased in response to both a HFD and membrane depolarization and requires CREB-dependent Ca²⁺ signaling. No differences in glucose homeostasis or islet morphology were observed in Ascl1^βKO mice fed a normal diet or in the absence of K_ATP channels. However, male Ascl1^βKO mice fed a HFD exhibited decreased blood glucose levels, improved glucose tolerance, and increased β-cell proliferation. Bulk RNA-seq analysis of islets from Ascl1^βKO mice from three studied conditions showed alterations in genes associated with the secretory function. HFD-fed Ascl1^βKO mice showed the most extensive changes with increased expression of genes necessary for glucose sensing, insulin secretion and β-cell proliferation, and a decrease in genes associated with β-cell dysfunction, inflammation and dedifferentiation. HFD-fed Ascl1^βKO mice also displayed increased expression of parasympathetic neural markers and cholinergic receptors that was accompanied by increased insulin secretion in response to acetylcholine and an increase in islet innervation.

Conclusions

Ascl1 expression is induced by stimuli that cause Ca²⁺-signaling to the nucleus and contributes in a multifactorial manner to the loss of β-cell function by promoting the expression of genes associated with cellular dedifferentiation, attenuating β-cells proliferation, suppressing acetylcholine sensitivity, and repressing parasympathetic innervation of islets. Thus, the removal of Ascl1 from β-cells improves their function in response to metabolic stress.

Highlights

• •Induction of Ascl1 in pancreatic β-cells requires CREB-dependent Ca²⁺ signaling.
• •Deletion of Ascl1 does not measurably alter islet function in the normal state.
• •β-cells lacking Ascl1 exhibit increased insulin secretion and proliferation on a HFD.
• •HFD-induced β-cell dedifferentiation is decreased in the absence of Ascl1.
• •Parasympathetic innervation is increased in mice lacking Ascl1 on a HFD.

1. Introduction

Type 2 diabetes (T2D) is a multi-system disorder that is characterized by insulin resistance, hyperglycemia, nutrient overload and metabolic stress-induced loss of pancreatic β-cell mass and/or function [[1], [2], [3], [4]]. Cellular dedifferentiation and the loss of β-cell identity are important contributors to β-cell dysfunction in the setting of insulin resistance and metabolic stress [5,6]. Dedifferentiation is characterized by the loss of key β-cell genes that control essential cellular processes, such as glucose sensing and insulin secretion, and the upregulation of genes associated with immature pancreatic endocrine cell states. The signaling pathways and transcriptional regulators that drive these changes in metabolically stressed β-cells are not understood [7,8].

ASCL1 (Achaete-scute homolog 1), also known as MASH1, is a basic helix-loop-helix (bHLH) transcription factor involved in neuronal fate determination [9]. Like other bHLH proteins, ASCL1 must either homodimerize or form heterodimers with other bHLH proteins to bind to the hexanucleotide binding motif CANNTG, also known as an E-box [10]. During development, the dynamic expression of Ascl1 promotes both the proliferation and differentiation of neural stem cells [11] with ASCL1 being both necessary and sufficient to induce the formation of functional neurons from non-neuronal cells [12]. Ascl1 is also critical for development of neuroendocrine cells in multiple tissues, including stomach, lung, adrenal medulla, thyroid and prostate [[13], [14], [15]]. Since ASCL1 targets both readily accessible and closed regions of chromatin in developing neural cells, it is considered a pioneer transcription factor [16,17].

We previously reported that Ascl1 expression is increased in two different models of metabolically stressed β-cells: HFD-fed mice and mice lacking the K_ATP channel subunit Abcc8 [18,19]. In Abcc8 knockout (KO) β-cells, a sustained increase in intracellular Ca²⁺ ([Ca²⁺]_i), or excitotoxicity, resulting from chronic membrane depolarization leads to impairments in islet morphology, glucose tolerance, and β-cell identity [19,20]. We have shown that expression of Ascl1 in β-cells is increased following treatments that cause Ca²⁺ influx, and have performed detailed transcriptomic analyses that have suggested that ASCL1 is an upstream regulator for many of the genes upregulated in Abcc8 KO islets, including for Aldh1a3, a marker of β-cell dedifferentiation [19]. Ascl1 has also been shown to be expressed at higher levels in less mature β-cells that exhibit increased [Ca²⁺]_i [21]. While Ascl1b regulates pancreatic endocrine cell fate in zebrafish [22], Ascl1 is not required for pancreatic endocrine cell specification or the expression of pro-endocrine transcription factor Neurog3 during mouse development [23]. ASCL1 is expressed in human islets, its gene exhibits an open chromatin structure in β-cells [24], and islet-specific super-enhancers are located nearby [25]. Since super-enhancers are often located near key cell identity genes [26], these observations further suggest that ASCL1 may be important for regulating the function of β-cells in a yet-to-be-defined manner. Altogether, these data led us to hypothesize that the induction of Ascl1 in response to metabolic stress contributes to β-cell failure in T2D by regulating the gene network that is induced in β-cells in response to metabolic stress and promoting β-cell dedifferentiation.

To determine the role of Ascl1 in metabolic-stress induced β-cell failure, we generated pancreatic β-cell-specific Ascl1 knockout mice and assessed β-cell function and gene expression in response to 1) a normal diet, 2) metabolic stress brought on by a HFD, and 3) excitotoxic stress caused by a genetic ablation of Abcc8. Our results indicate that ASCL1 contributes to the loss of β-cell identity and function in the setting of HFD-induced metabolic stress. Mice with a β-cell-specific ablation of Ascl1 exhibit improved glucose tolerance, increased β-cell proliferation, increased expression of β-cell identity genes and decreased expression of dedifferentiation markers. In addition, islets of HFD-fed that lack Ascl1 exhibit increased islet innervation and increased insulin secretion in response to acetylcholine.

2. Materials and methods

2.1. Mouse lines and husbandry

The Ascl1^Flox (Ascl1^tm2Fgu) [27], Ins^Cre (Ins1^{tm1.1(cre)Thor}) [28], Gt(ROSA)26^{Sortm9(CAG-tdTomato)Hze} [29] and Abcc8 KO (Abcc8^tm1.1Mgn) [20] alleles were all maintained as C57Bl/6J congenic lines (stock 000664 at The Jackson Laboratory). At weaning (3–4 weeks of age), mice were fed normal chow (NC) (4.5% fat content, PicoLab, 5L0D) or a HFD (60% fat content, D12492, Research Diets) for 12 weeks. All animal experimentation was performed under the oversight of the Vanderbilt Institutional Animal Care and Use Committee. Mice were genotyped by PCR using primers listed in Table S1.

2.2. Glucose homeostasis

Blood glucose concentrations were determined in tail blood samples using a BD Logic glucometer. Intraperitoneal glucose tolerance tests (IPGTT) were performed following a 16-hour overnight fast. Blood glucose concentrations were measured at 0, 15, 30, 60, and 120 min after administering d-glucose (2 mg/g body mass). Plasma insulin concentrations were measured by radioimmunoassay (Millipore, PI-13K) and performed in triplicate by the Vanderbilt Hormone Assay and Analytical Services Core.

2.3. Islet isolation and culture

Islets were isolated by the Vanderbilt Islet and Pancreas Analysis (IPA) core by injecting 0.6 mg/mL collagenase P (Roche) into the pancreatic bile duct followed by a Histopaque-1077 (Sigma) fractionation and hand-picking. Isolated islets were cultured were cultured overnight in low glucose DMEM (Thermo, 11966-025) containing 1 g/L glucose and supplemented with 10% FBS (Atlanta Biologicals), 4.6 mM HEPES, 1% penicillin/streptomycin (Thermo), 1% non-essential amino acids (Sigma) in a sterile cell incubator at 37 °C with 5% CO₂ infusion and 95% humidity. Islet responses were assessed by treatments with culture media containing high glucose (20 mM), 100 μM tolbutamide (Sigma, T0891), 50 μM verapamil (Sigma, V4629) or 1 μM for CREB-inhibitor 666-15 (MedChemExpress, HY-101120). Control samples were treated only with the DMSO vehicle (0.1%). To measure static glucose-stimulated insulin secretion, islets from HFD-fed male mice were cultured overnight then equilibrated in KRHB (Krebs-Ringer HEPES Buffer, pH7.4) for 30 min followed by 30 min incubation with KRHB containing either low glucose (3.3 mM), high glucose (16.7 mM), high glucose with 100 nM Extendin-4 (MedChemExpress, HY-13443) or 1 μM acetylcholine (MedChemExpress, HY-B0282). Insulin concentrations in post-incubation supernatants were measured by radioimmunoassay.

2.4. RNA isolation and RT-qPCR

RNA was isolated from whole islets using the Maxwell 16 LEV simplyRNA Tissue Kit (Promega, TM351). For qPCR, reverse transcription was performed using a High Capacity cDNA Reverse Transcription Kit (Thermo). 2 ng of cDNA was used to perform real-time qPCR using Power SYBR Green PCR master mix (Thermo) and a CFX96 Real-Time PCR system (Bio-Rad). Relative expression was determined by the ΔCt method by normalizing to the expression of Actb. Primers are listed in Table S1.

2.5. RNA sequencing and data analysis

RNA quality was assessed using an Agilent 2100 Bioanalyzer and only those samples with an RNA integrity number (RIN) 7 or above were used to produce cDNA libraries using a Low Input Library Prep Kit. 150 nucleotide paired-end reads were obtained by Novogene Corp. using an Illumina NovaSeq6000 instrument that resulted in at least 40 M raw sequencing reads per sample. The Spliced Transcripts Alignment to a Reference (STAR v2.6.0c) application [30] was used to perform sequence alignments to the mm10 (GRCm38) mouse genome reference and GENCODE comprehensive gene annotations (Release M17). Overall, 85–87% of the raw sequencing reads were uniquely mapped to genomic sites. DESeq2 was employed for additional sample-level quality control analysis and downstream pairwise comparisons [31]. Gene ontology analysis of differentially expressed genes was performed using Metascape [32]. LISA analysis was used to predict transcriptional regulators of differentially expressed genes [33].

For analysis of ASCL1 expression in human β-cells, scRNA-seq data for human normal and T2D islets were obtained from the human pancreas atlas program (HPAP) [34] and downloaded in the anndata.h5ad format. Raw counts were normalized and Log + 1 transformed using Scanpy v1.9.3 [35]. To explore ASCL1 and INS expression, cells were filtered for β-cells based on annotations provided by HPAP. ASCL1⁺ β-cells were defined as cells with greater than 0 normalized counts for ASCL1. β-Cells were binned into three discretized bins using the pandas qcut function with q = 3 and using expression of INS as the input. Cells with fewer than 8.343 log normalized counts for INS were defined as low, those with greater than 8.631 log normalized counts as high, and those in between as moderate.

2.6. Immunohistochemistry and morphometric analysis

Whole pancreata were fixed for 2 h in 4% paraformaldehyde, incubated overnight at 4 °C in 30% sucrose, embedded in OCT compound (Tissue Tek), frozen on dry ice, and sectioned at a depth of 10 mm. Immunostaining was performed as previously described [36]. Primary antibodies used were guinea pig anti-insulin (Linco, 1:1000), rabbit anti-glucagon (Linco, 1:1000), rabbit anti-somatostatin (Linco 1:1000), guinea pig anti-pancreatic polypeptide (Linco, 1:1000), goat anti-dsRed (Santa Cruz Biotechnology, 1:1000), rabbit anti-ALDH1A3 (Novus Biologicals,1:500), and mouse anti-TUJ (BioLegend, 1:2000). Secondary antibodies were donkey anti-guinea pig and anti-mouse IgG Alexa-488 conjugated, donkey anti-goat IgG Alexa-555 conjugated and donkey anti-rabbit IgG Alexa-647 conjugated (Thermo, 1:1000). After antibody staining, slides were mounted with Invitrogen Prolong Gold Antifade Mountant with DAPI (Thermo). Images were acquired using an Aperio ScanScope CS imaging system or Zeiss LSM Meta 510 confocal microscope. Quantifications of relative areas were done on evenly spaced pancreatic sections at 150 μm apart by using ImageJ software [37]. Insulin⁺ β-cell areas were calculated as a percentage of the whole pancreas areas (DAPI⁺), relative hormone+ areas, and relative TUJ+ areas as a percentage of total islet areas and β-cell proliferation as a percentage of MKI+ cell per insulin+ cells. Images in figures are representative of the phenotype observed in at least three different animals per genotype.

2.7. Statistical analysis

p-Values were determined by ANOVA or by a two-tailed unpaired Student's t-test where applicable.

3. Results

3.1. Ascl1 expression is elevated in dedifferentiated and metabolically stressed β-cells and requires Ca²⁺ signaling through CREB

To better understand the developmental expression pattern of Ascl1 during pancreas and pancreatic β-cell development, we began by extracting a temporal expression profile for Ascl1 from our prior transcriptional network analysis of FACS-purified cell populations [38]. As shown in Figure 1A, Ascl1 is expressed during the pre-pancreatic and multipotent endocrine progenitor cell stages (embryonic days (E) 9.5–10.5) and is then down-regulated beginning at E15.5 in normal pre-endocrine and β-cell lineages. The highest levels of Ascl1 expression were observed in two less differentiated progenitor cell states caused either by the genetic deletion of Neurog3 or Insm1 [38]. While Ascl1 is expressed at very low levels in normal adult β-cells, we have previously shown that it is induced 2- and 30-fold, respectively, by metabolic stress due to overnutrition (5 weeks on a HFD) or by excitotoxicity due to the lack of functional K_ATP channels (Abcc8 KO) (Figure 1B) [18]. In this study, islets isolated from HFD-fed (12 weeks) and Abcc8 KO mice had similarly increased levels of Ascl1 (Figure 1C). 24-hour in vitro incubations of wild type (WT) islets with the K_ATP channel inhibitor tolbutamide, which causes membrane depolarization, leads to increased Ascl1 expression that is diminished with the addition either of the voltage-dependent Ca²⁺ channel (VDCC) inhibitor verapamil or CREB inhibitor 666-15 (Figure 1D). CREB (cAMP Response Element-Binding protein) is an important Ca²⁺-signaling activated regulator of transcription [39], and analysis of available CREB ChIP data shows that promoters for human and mouse Ascl1 contain nearby CREB-bound regions (Figure S1). Analysis of scRNA-seq data from human islets of control and T2D individuals [34] showed that higher ASCL1 expression occurs in β-cells that have lower expression of INS in T2D (Figure 1E, F). Combined, these data indicate that Ascl1 expression is increased in metabolically stressed β-cells, that less differentiated endocrine cells exhibit higher expression than mature β-cells, and that the induction of Ascl1 in adult β-cells occurs in response to treatments that cause an increase [Ca²⁺]_i and requires CREB-dependent Ca²⁺-signaling to the nucleus.

Figure 1.

Increases in [Ca²⁺]_i and CREB-dependent Ca²⁺ signaling increase Ascl1 expression in β-cells. A) Expression levels of Ascl1 mRNA profiled by RNA-seq in sorted pancreatic developmental lineages: gut tube endoderm (Sox17⁺ at E8.0/8.5), posterior foregut endoderm (Pdx1⁺ at E9.5), pancreatic multipotent progenitor cells (Ptf1a⁺ at E10.5), endocrine progenitor cells (Neurog3⁺ and Insm1⁺ at E15.5), nascent β-cells (Ins⁺ at E16.5), and adult β-cells (Ins⁺ at P60). Two profiled mutant conditions for endocrine progenitor cells (Neurog3^−/− and Insm1^−/− at E15.5) indicate that Ascl1 is increased in the absence of Neurog3 and Insm1. N = 3. B) RNA-seq of purified β-cells shows that Ascl1 expression is increased in wild type (WT) mice on HFD (WT + HFD, 5 weeks) and in Abcc8 KO mice. N = 8 (males and females, 8 weeks old). C) RT-qPCR analysis of isolated islets shows the increase in Ascl1 expression on HFD and in Abcc8 KO mice in comparison to ND mice. N = 4 (males, 14–17 weeks old). D) RT-qPCR analysis of islets from in vitro incubations of WT islets shows that tolbutamide or high glucose (Glu, 20 mM) lead to an increase in Ascl1 expression that is decreased with the addition of verapamil and CREB inhibitor (N = 4, males, 8 weeks old). Error bars: ±SEM. ∗∗p ≤ 0.01; ∗p ≤ 0.05. p-value determined by ANOVA. (E, F) Analysis of scRNA-seq data of human β-cells from the human pancreas atlas program (HPAP). E) Scatter plot of insulin (INS) versus ASCL1 expression in ASCL1-expressing human β-cells from normal (935 cells) and T2D (280 cells) individuals shows that higher ASCL1 expression is associated with lower insulin expression in T2D. F)ASCL1-expressing human β-cells were divided into three groups based on levels of INS expression. The bar graph shows that the percentage of ASCL1-expressing cells per group is higher in low-insulin expressing cells in T2D.

3.2. Glucose homeostasis and islet morphology are unaffected in adult Ascl1^βKO mice fed a normal diet (ND)

To determine the role of Ascl1 in β-cells, we crossed Ascl1^Flox and Ins1^Cre mice to obtain animals that lack Ascl1 specifically in β-cells (Ascl1^βKO). In addition, to assess the efficiency of Cre-recombination and to lineage trace β-cells we also introduced a Cre-activated red fluorescent protein (RFP) reporter Rosa26^tdTomato allele in the breeding scheme (Figure 2A). Analysis of the number of RFP⁺ cells per Ins⁺ cells showed that we achieved more than 95% efficiency of Cre recombination in β-cells (Figure 2B) and observed an expected decrease in Ascl1 in islets by RT-qPCR (Figure 2C). Next, to assess the impact of Ascl1 on normal β-cell development and function we analyzed glucose homeostasis and islet morphology of adult 14 weeks old mice fed a ND. Male Ascl1^βKO animals, wild type (WT) controls, and mice heterozygous for the Ascl1^Flox allele (Het) all had similar weight, blood glucose and insulin levels, glucose tolerance and insulin responses (Figure 3A–G) with female mice also showing no differences (Figure S2). Immunofluorescent staining of pancreatic sections and quantification showed no difference in the percentage of β-cell area per total pancreatic area, INS⁺ to RFP⁺ area ratio (a measure of maintenance of β-cell identity), or the percentage of hormone⁺ area (insulin, glucagon, somatostatin and pancreatic polypeptide) per islet area (Figure 3H–K). Islet architecture also appeared normal (Figure 3J). Since Ascl1^βKO mice survive into adulthood and demonstrate no significant differences in glucose homeostasis and islet composition we conclude that Ascl1 does not measurably affect β-cell development or function in a metabolically unstressed state.

Figure 2.

Generation and analysis of β-cell specific Ascl1 knockout mice (Ascl1^βKO). A) Summary of alleles that were bred together to create Ascl1^βKO and control heterozygous (Het) and wild type (WT) mice, animal cohorts that were used in the study including normal diet (ND), high fat diet (HFD) and Abcc8 homozygous KO groups, and analyses that were performed. B) Immunofluorescent co-staining of Ascl1 Het islets with anti-insulin and anti-RFP antibodies demonstrates high efficiency Cre-recombination in β-cells. Scale bar: 50 μM. C) RT-qPCR analysis of islets of Ascl1^βKO mice in ND, HFD and Abcc8 KO conditions shows a significant decrease in Ascl1 expression in comparison to Het controls. N = 4 (males 14–17 weeks old). Error bars: ±SEM. ∗∗∗p ≤ 0.001. p-value determined by t-test.

Figure 3.

Glucose homeostasis and islet composition are unaffected in adult Ascl1^βKO mice fed a normal diet (ND). A) Body weight measurements at the time of the experiment and fasted (B) and fed (C) blood glucose concentration measurements showed no significant differences between Ascl1^βKO (KO) and control Het and WT animals. N = 12. D) Random-fed plasma insulin levels. N = 4. E) Intraperitoneal glucose tolerance test (GTT) and area under the curve (AUC) measurements for (F). N = 12. G) Insulin secretion in response to glucose injection in GTT for Ascl1^βKO and control Het samples. Both genotypes show an equal increase in plasma insulin concentration 15 min after glucose injection. N = 4. H–K) Immunofluorescent staining and quantification of pancreatic islet tissues from Ascl1^βKO (KO) and control (cont.) Het mice. H) Quantification of β-cell (INS⁺) area as the percentage of total pancreas area. N = 4. I) Quantification of INS+ to RFP+ area ratio in islets as a measurement of β-cell dedifferentiation. N = 4. J) Representative images of immunofluorescent staining for red fluorescent protein (RFP, red) and pancreatic hormones insulin (INS, green), glucagon (GCG, blue), somatostatin (SST, blue) and pancreatic polypeptide (PPY, green) of Ascl1^βKO and control Het samples. Scale bar: 50 μM. K) Quantification of hormone-positive areas per total islet areas. N = 4. All mice are males, 14 weeks old. Error bars: ±SEM. ∗p ≤ 0.05. p-value was determined by ANOVA or by t-test were applicable. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

3.3. Ascl1^βKO mice have decreased blood glucose, improved glucose tolerance and increased β-cell proliferation after 12 weeks on a HFD

To investigate the role of Ascl1 in β-cell response to metabolic stress due to nutrient overload we fed Ascl1^βKO and control mice a HFD for 12 weeks prior to accessing glucose homeostasis and the β-cell area and proliferation. While male animals gained similar amounts of weight, Ascl1^βKO males exhibited a decrease in their fed blood glucose levels, an increase in glucose tolerance, and a trend towards higher plasma insulin levels (Figure 4A–G). At the same time, immunostaining and quantification revealed a trend in the increased total β-cell area (p = 0.09), no changes in the INS⁺ to RFP⁺ area ratio and a significant increase in the number of proliferating MKI67⁺ β-cells in Ascl1^βKO male mice (Figure 4H–K). We did not observe any significant changes in glucose homeostasis in HFD-fed female mice (Figure S3). The combination of decreased blood glucose levels, improved glucose tolerance, and increased β-cell proliferation in the absence of Ascl1 in male animals fed a HFD indicates that metabolic stress-induced expression of Ascl1 impairs both β-cell function and proliferation.

Figure 4.

Ascl1^βKO mice have decreased blood glucose, improved glucose tolerance and increased β-cell proliferation after 12 weeks on a high fat diet (HFD). A) Body weight measurements at the time of the experiment and fasted (B) and fed (C) blood glucose concentration measurements between Ascl1^βKO (KO) and control Het and WT animals on HFD. Ascl1^βKO mice have decreased fed blood glucose concentration. N = 10. D) Random-fed plasma insulin levels for Ascl1^βKO and control Het samples. N = 6. E) Intraperitoneal glucose tolerance test (GTT) and its area under the curve (AUC) measurements (F) show that Ascl1^βKO mice have improved glucose tolerance on HFD. N = 10. G) Insulin secretion in response to glucose injection in GTT for Ascl1^βKO and control Het samples on HFD. N = 5. H–K) Immunofluorescent staining and quantification of pancreatic islet tissues from Ascl1^βKO (KO) and control (cont.) Het mice on HFD. H) Quantification of β-cell (INS⁺) area as the percentage of total pancreas area. N = 4. I) Quantification of INS+ to RFP+ area ratio in islets as a measurement of β-cell dedifferentiation. N = 4. J) Representative images of immunofluorescent staining for red fluorescent protein (RFP, red) that marks β-cells and cell proliferation marker MKI67 (green) of Ascl1^βKO and control (cont.) Het samples. Nuclei are stained with DAPI (blue). Arrows show MKI67-positive β-cells. Scale bar: 50 μM. K) Quantification of β-cell proliferation. N = 4. All mice are males, 17 weeks old. Error bars: ±SEM. ∗∗∗∗∗∗p ≤ 0.0001, ∗∗p ≤ 0.01, ∗p ≤ 0.05. p-value was determined by ANOVA or by t-test were applicable. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

3.4. Ascl1^βKO mice do not show an improvement in glucose homeostasis or islet architecture on a global Abcc8 KO background

Since Ascl1 expression is strongly up-regulated in a genetic model of excitotoxicity (Abcc8 knockout (KO) mice) we next sought to determine whether the loss of Ascl1 affects β-cell function on a global Abcc8 KO background. Adult Abcc8 KO mice develop glucose intolerance after 12 weeks of age and have a disrupted islet architecture and compromised β-cell identity [18,19]. After interbreeding, we obtained Ascl1^βKO; Abcc8 KO mice as well as Abcc8 KO controls which were assessed at 14 weeks of age on a ND for alterations in glucose homeostasis and islet composition. Surprisingly, we did not observe any improvements in blood glucose levels, glucose tolerance or insulin secretion in this cohort of mice (Figure 5A–H, Figure S4). We also did not observe any changes in the β-cell area, the INS⁺ to RFP⁺ area ratio, or islet composition. Islets retained the disrupted architecture characteristic of Abcc8 KO mice with α- and δ-cells being found intermingling with β-cells within the core of the islets (Figure 5H–K). These results indicate that disruption of Ascl1 is not sufficient by itself to rescue the β-cell dysfunction of Abcc8 KO mice.

Figure 5.

Ascl1^βKO mice do not show an improvement in glucose homeostasis or an islet architecture on a global Abcc8 KO background. A) Body weight measurements at the experiment and fasted (B) and fed (C) blood glucose concentration measurements showed no significant differences between Ascl1^βKO (KO) and control Het and WT animals on Abcc8 KO background. N = 12. D) Random-fed plasma insulin levels. N = 6. E) Intraperitoneal glucose tolerance test (GTT) and its area under the curve (AUC) measurements for (F). N = 12. G) Insulin secretion in response to glucose injection in GTT. Both genotypes show defects in the increase in plasma insulin concentration 15 min after glucose injection. N = 4. H–K) Immunofluorescent staining and quantification of pancreatic islet tissues from Ascl1^βKO (KO) and control (cont.) Het mice on Abcc8 KO background. H) Quantification of β-cell (INS⁺) area as the percentage of total pancreas area. N = 4. I) Quantification of the INS+ to RFP+ area ratio in islets as a measurement of β-cell dedifferentiation. N = 4. J) Representative images of immunofluorescent staining for red fluorescent protein (RFP) and pancreatic hormones insulin (INS), glucagon (GCG), somatostatin (SST) and pancreatic polypeptide (PPY) of Ascl1^βKO and control Het samples. Nuclei are stained with DAPI (blue). Scale bar: 50 μM. K) Quantification of hormone-positive areas per total islet areas. N = 4. All mice are males, 14 weeks old. Error bars: ±SEM. p-value was determined by ANOVA or by t-test were applicable. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

3.5. Transcriptional profiling of Ascl1^βKO islets in three different conditions reveals variable changes in secretory and neural genes

To determine how the absence of Ascl1 in β-cells affects different cellular responses, we next performed bulk RNA-seq of islets from male Ascl1^βKO and Het littermate control male mice from the ND, HFD and Abcc8 KO cohorts (n = 4 for each condition, 14–17 weeks old). Principal component analysis of each transcriptome revealed a clear separation between three different conditions with the HFD condition demonstrating the largest difference between Ascl1^βKO and control islets (Figure 6A). Differential expression (DE) analysis revealed 605, 1685 and 589 genes whose expression was altered (FDR p_adj < 0.05) in ND, HFD and Abcc8 KO conditions, respectively (Table S2), with the most profound ASCL1-dependent transcriptional changes occurring in the HFD-fed islets, consistent with our phenotyping results. Overlap analysis of the genes affected by the absence of Ascl1 in the three different conditions showed only small similarities, indicating that each of the different treatment conditions caused a distinct response (Figure 6B). Notably, among the genes that were commonly dysregulated in all conditions three are required for neural exocytosis: Syt1, Pak6, and Dclk1. Despite the small number of shared genes, comparative gene ontology (GO)-based functional enrichment analysis of the datasets from the different conditions indicated that ASCL1-dependent genes shared several functional categories (Figure 6C, Table S3). The shared GO categories include protein localization, cell secretion and export from cell, synaptic signaling, neural system development, cell junction and adhesion, and ER-to-Golgi vesicle traffic (Figure 6D), indicating that ASCL1 regulates genes involved with the secretory function of islets.

Figure 6.

RNA-seq of Ascl1^βKO islets in three different conditions reveals variable changes in secretory and neural genes. A) PCA plot of RNA-seq samples. ND, normal diet (light green, Ascl1^βKO ND vs control (cont.) ND comparison), HFD, high fat diet (light blue, Ascl1^βKO HFD vs cont. HFD comparison), Abcc8 KO (pink, Ascl1^βKOAbcc8 KO vs control Abcc8 KO comparison). Male mice, 14–17 weeks old. B) Venn diagram indicating overlap between differentially expressed (DE) genes (p_adj < 0.05) identified from pairwise comparisons for each condition, with select dysregulated genes indicated for each overlap. ND, normal diet (light green, Ascl1 KO ND vs cont. ND comparison), HFD, high fat diet (light blue, Ascl1 KO HFD vs cont. HFD comparison), Abcc8 KO (pink, Ascl1 KO Abcc8 KO vs cont. Abcc8 KO comparison). C) Cord diagrams show genes (purple curves) and GO terms/pathways (blue curves) shared among lists of DE genes from three comparisons. ND, green; HFD, blue; Abcc8 KO, red. D) Enrichment network visualization of enriched functional GO terms/pathways shared among DE genes from the three comparisons. Node size is proportional to the number of genes in the GO category, with pie charts indicating a proportion of genes from each comparison: ND, green; HFD, blue; Abcc8 KO, red. Intensity of a node border color indicates the GO category enrichment p-value (from 10⁻⁴⁸ to 10⁻²). E) Venn diagrams indicating overlap between the top 100 transcriptional regulators predicted to drive the expression of upregulated DE genes (positive regulators) and downregulated genes (negative regulators) for each of the three comparisons. bHLH transcription factors are shown in callouts for each overlap. ND, light green; HFD, light blue; Abcc8 KO, pink. ASCL1 is among the predicted regulators. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

To predict the transcriptional co-regulators that might act in concert with ASCL1 we used the epigenetic Landscape In Silico deletion Analysis (LISA) tool. This analysis method has been used to identify upstream transcription factors and chromatin regulators that mediate the perturbation of specific DE gene sets [33]. Our use of LISA revealed a diverse and only partially overlapping set of positive and negative drivers of upregulated and downregulated DE genes, respectively (Figure 6E, Table S3). Notably, ASCL1 was identified as the upstream driver of genes that are upregulated in the absence of Ascl1 in all three conditions, implying that ASCL1 functions as a repressor of gene expression in β-cells. Conversely, in the HFD condition, ASCL1 was identified as one of the top drivers of down-regulated genes, implying that it activates gene expression in this setting. These findings suggest that transcriptional regulation by ASCL1 in β-cells is complex and dependent on both the treatment condition and genotype. The observed complexity is consistent with ASCL1 action being dependent on other co-regulators, including other bHLH neuronal transcription factors, thereby differentially regulating many neural and secretory genes in response to the ND, HFD and Abcc8 KO conditions.

3.6. β-Cell identity and metabolic stress genes affected by HFD are oppositely regulated in Ascl1^βKO islets

While ASCL1 regulates a wide spectrum of neural and secretory genes, nearly three times more genes were affected in the HFD compared to the ND and Abcc8 KO conditions. Although we did not detect any significant alterations in β-cell function in the ND fed Ascl1^βKO islets, transcriptional profiling revealed 309 up-regulated and 296 down-regulated genes. Functional enrichment analysis indicates that the upregulated genes are involved in cell morphogenesis, neuronal projection, and ECM organization, whereas the down-regulated genes are involved in ER protein transport and cell cycle regulation (Figure S5, Tables S2 and S3). Similarly, even though Ascl1 is highly expressed in Abcc8 KO β-cells, its deletion affected only 268 up-regulated and 321 down-regulated genes in the Ascl1^βKO; Abcc8 KO islets. This relatively low number is in marked contrast to the 7393 dysregulated genes in FACS-purified Abcc8 KO β-cells [18], suggesting that ASCL1 is only partially responsible for the many gene expression changes brought on by the lack of functional K_ATP channels. Interestingly, the top genes upregulated in Ascl1^βKO islets on Abcc8 KO background are involved in ER stress response (Hspa1a/b, Ddit3, Ddit4), autophagy (Depp1, Gabarapl1) and chromatin maintenance (H3c6, H4c12) and down-regulated genes involved in hormone secretion and vesicle transport (Chgb, G6pc2, Kdelr) (Figure S6, Tables S2 and S3). The increases in stress genes and decreases in secretory genes in the absence of Ascl1 indicate that it may play a limited protective role in β-cells in the absence of Abcc8, however its deletion does not result in any significant functional changes in Abcc8 KO β-cells.

The transcriptional effects of the Ascl1 deletion were more pronounced in HFD-fed mice where 923 genes were up-regulated and 757 down-regulated (Figure 7A). To compare Ascl1-driven changes on HFD to transcriptional alterations in normal islets caused by HFD feeding, we also compared the control ND and control HFD samples. The results show that metabolic stress caused by HFD feeding leads to major changes in islet transcriptome with 4003 dysregulated genes (2173 up- and 1830 down-regulated) (Figure S7A, Table S2). The GO categories of up-regulated genes included lipid metabolism, autophagy, ER stress and immune response whereas those for down-regulated genes included cell morphogenesis, neuron projection, cell secretion, synapse organization and cell adhesion (Figure S7B,C). All of these changes are consistent with the now well-established view that metabolically stressed β-cells undergo ER stress, inflammation, and a progressive loss of cell identity. However, and in marked contrast, in HFD-fed islets that lack Ascl1 many of these adverse changes are reversed. Functional enrichment analysis indicated that upregulated genes are involved in cell morphogenesis, neuron projection, cell secretion, synaptic signaling, cell junction and adhesion, and downregulated genes are involved in immune response to virus, lipid metabolism, lysosome, autophagy and apoptotic signaling (Figure 7B, Table S3), a response profile that is nearly opposite to that of the HFD islets. Indeed, the directional overlap analysis of islet genes dysregulated by a HFD showed that most of the 1225 genes that are either up- or down-regulated in response to the HFD are oppositely regulated in the absence of Ascl1 (Figure 7C). Notably, the genes down-regulated by the HFD but up-regulated in the Ascl1^βKO on HFD include many involved in neural signaling (Grp, Crhr1, Vip), neurotransmitter receptors (Grik2, Gabrg3, Chrna3), cell junction and adhesion molecules (Nptx2, Pcdh7, Slit2, Chl1) and many genes that are involved in positive regulation of insulin secretion (Trpm5, Nnat, Ffar4, Syt1, Slc2a2, Glp1r, Adcy5, Baiap3). Additionally, key β-cell transcription factors (Mafa, Mafb, Myt1, Nkx6-1, Neurod1), early stress response genes (Atf3, Egr3, Egr4), and essential β-cell proliferation genes (Foxm1, Mki67) are also up-regulated. Conversely, genes normally up-regulated on HFD but down-regulated in the absence of Ascl1 include well established markers for β-cell dedifferentiation (Aldh1a3, Aass, Serpin7a) [40], immune response (Irf7, Ifit3), β-cell dysfunction and hyperglycemia (S100z) [41], developmental genes (Wnt5b, Bambi), genes involved in protein palmitoylation (Zdhhc22) and ubiquitination (Asb9, Asb11), and negative regulators of the cell cycle (Nupr1, Cdkn1a) (Figure 7C,D). Combined, these data indicate that transcriptional regulation by ASCL1 and, potentially, other co-regulators that are simultaneously induced by metabolic stress contribute substantially to the loss of β-cell function through the activation of dedifferentiation program and repression of a network of β-cell identity genes involved in insulin secretion and proliferation.

Figure 7.

β-cell identity and metabolic stress genes affected by HFD are oppositely regulated in Ascl1^βKO islets. A) Volcano plot (Log₂ FoldChange (FC) over FDR p_adj-value) showing differentially expressed (DE) genes with red dots (p_adj < 0.05) in Ascl1^βKO HFD vs control (cont.) comparison. HFD islets from 17-week-old male mice (N = 4). Select top differentially expressed genes (based on p_adj-value) are indicated by names and total numbers of downregulated (green) and upregulated (red) genes are provided. B) Functional enrichment analysis of upregulated and downregulated DE genes in Ascl1^βKO HFD vs control HFD islet comparison. Select top enriched pathways are shown. C) Venn diagram indicating directional overlaps between DE genes identified from pairwise comparisons for effects of HFD (Control HFD vs control ND) and for effects of Ascl1 on HFD (Ascl1^βKO HFD vs control HFD). A large proportion of genes (shown in red) that are affected by HFD are changed in a opposite direction when Ascl1 is deleted. Select oppositely regulated genes are shown in callouts. D) Differential expression of select top upregulated and downregulated DE genes in Ascl1^βKO HFD vs control HFD islets comparison that are also oppositely regulated in control HFD vs control ND comparison. Text indicates gene functional associations. Log₂FC: Log₂ Fold Change. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

3.7. Decreased dedifferentiation and increased insulin secretion and islet innervation in Ascl1^βKO islets on HFD

To validate the beneficial transcriptional changes that the deletion of Ascl1 has on islets from HFD-fed mice, we performed qPCR for selected genes and confirmed that genes critical for insulin secretion (Atf3, Trpm5, Chrna3, Foxm1, Slc2a2, Glp1r, Nptx2) are up-regulated in the Ascl1^βKO islets and genes indicating β-cell dysfunction (Aldh1a3, Serpina7, Asb11, Nupr1, Me3, S100z, Cdkn1a) are down-regulated (Figure 8A, B). Immunofluorescent staining of islet sections of HFD-fed mice also confirmed the decrease in levels of the β-cell dedifferentiation marker ALDH1A3 in Ascl1^βKO islets compared to controls (Figure 8C). Together, these results further confirm that deletion of Ascl1 from β-cells diminish the HFD-induced expression of dedifferentiation markers, including ALDH1A3.

Figure 8.

Decreased dedifferentiation and increased insulin secretion and islet innervation in Ascl1^βKO islets on HFD. A) RT-qPCR analysis of islets from HFD-fed Ascl1^βKO (KO) and control mice. Het HFD samples confirms upregulation of genes important for insulin secretion and downregulation of genes associated with β-cell dysfunction. N = 4. B) Representative images of immunofluorescent staining of pancreatic islet tissues from Ascl1^βKO (KO) HFD and control (cont.) Het HFD mice for red fluorescent protein (RFP, red) that marks β-cells and β-cell dedifferentiation marker ALDH1A3 (green). Nuclei are stained with DAPI (blue). Arrows show ALDH1A3-positive β-cells. Control samples have higher levels of ALDH1A3 detected in β-cells. Scale bar: 50 μM. C) Glucose stimulated insulin secretion of Ascl1^βKO (KO) HFD and control islets. Het HFD islets from in vitro static islet incubations. LoG (low glucose, 3.3 mM), HiG (high glucose 16.7 mM), Ext.4 (extendin-4, 1 μM), Ach (acetylcholine, 10 μM). Ascl1^βKO samples show a stronger insulin secretion response to high glucose and acetylcholine than control samples. #p ≤ 0.01 Cont. samples compared to LoG control #p ≤ 0.01 KO samples compared to LoG KOs. D) Representative images of immunofluorescent staining for red fluorescent protein (RFP, red) that marks β-cells and neuronal marker TUJ (green) of Ascl1^βKO and control Het samples. Nuclei are stained with DAPI (blue). E) Quantification of islet innervation (TUJ-positive areas per total islet areas). N = 4. All mice are males, 17 weeks old. Error bars: ±SEM. ∗∗∗p ≤ 0.001, ∗∗p ≤ 0.01, ∗p ≤ 0.05. p-values were determined by t-test. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Finally, given the increase in pro-secretory genes, including Chrna3, a receptor for the parasympathetic neurotransmitter acetylcholine, we measured glucose-stimulated insulin secretion in static islet incubations of Ascl1^βKO and control islets from HFD-fed mice. Both genotypes showed significant increases in insulin secretion in response to high glucose, high glucose and extendin-4, a GLP1 receptor agonist, and acetylcholine, compared to low glucose alone (Figure 8D). While the Ascl1^βKO samples exhibited a trend towards more insulin secretion in response to of all tested secretagogues, the combination of high glucose and acetylcholine resulted in the most robust and significant increase (Figure 8D). Given the enhanced response to acetylcholine, increased expression of neuropeptides Vip and Grp, and the elevated amounts of Slc18a3 and n-NOS (Nos1), two well established markers of parasympathetic nerves, we next sought to determine whether Ascl1^βKO islets might have increased density of cholinergic nerve fibers. Immunostaining for TUJ (TUBB3), a neuron-specific form of tubulin, and measurement of TUJ⁺ area per islet area revealed an increase in the number of neuronal processes that both penetrate and surround pancreatic islets, indicating that neural innervation in the Ascl1^βKO islets from HFD-fed mice is increased (Figure 8E,F). These findings indicate that Ascl1^βKO islets secrete more insulin in response to acetylcholine and exhibit increased parasympathetic innervation.

4. Discussion

In this study we assessed the role of Ascl1 in mediating the metabolic stress response of pancreatic β-cells. Our findings indicate that ASCL1 contributes to a loss of β-cell function both by activating a dedifferentiation program and by suppressing the expression of secretory and innervation genes in response to HFD feeding. Due to these effects, deletion of Ascl1 improves β-cell function during metabolic stress brought on by overnutrition.

4.1. Ca²⁺ signaling induces Ascl1 expression

In β-cells, Ca²⁺ signaling is critical for glucose-stimulated insulin secretion [42] but prolonged increases in [Ca²⁺]_i due to chronic metabolic stress may cause β-cell dysfunction [43,44]. It was previously shown [4,45] that there is a rise in [Ca²⁺]_i in β-cells in HFD-fed mice and db/db mice. Abcc8 KO mice have a sustained increase in [Ca²⁺]_i, increased Ca²-signaling to the nucleus, and marked alterations in the β-cell transcriptome that impair cell function [18,19]. In neurons and cardiomyocytes, two other excitable cell types, excess Ca²⁺-signaling, or excitotoxicity [46], is an established cause of cellular dysfunction [[47], [48], [49]]. While β-cells have functional and transcriptional similarities with neurons [50], and transcriptional regulation by ASCL1 has a well-established role in neural cell differentiation and function, the involvement of Ascl1 in physiological responses of β-cells has not been previously explored.

Here we show that during pancreatic β- and endocrine cell development Ascl1 is expressed in multipotent pancreatic progenitor cells and is strongly increased in the absence of Neurog3 or Insm1, two TFs necessary for the functional endocrine cell differentiation [51,52]. While Ascl1 is not essential for formation of islets in mice [23], our findings indicate that Ascl1 is most highly expressed in mis-differentiated or dedifferentiated pancreatic endocrine progenitor cells. Conversely, in adult mouse β-cells in normal condition Ascl1 is expressed at very low levels. Thus, while Ascl1 does not appear to play any significant role in determining endocrine cell fates, it acts as a response factor that is induced in response to elevations of [Ca²⁺]_i, such as those brought on in animals either by HFD feeding or the deletion of Abcc8, or in isolated islets by short-term stimulation with high glucose or tolbutamide treatment. Our observations that treatment of isolated islets with verapamil and a CREB inhibitor abrogate tolbutamide-induced expression of Ascl1, and the prior identification of multiple CREB-binding sites near the Ascl1 promoter (Figure S1), are all consistent with Ascl1 gene transcription being regulated by depolarization induced entry of Ca²⁺ through VDCCs on the plasma membrane and signaling to the nucleus via CREB. Similarly, in neurons, increases in Ca²⁺ oscillations [53] and CREB activation [54] were shown to increase Ascl1 expression. While Ca²⁺-dependent transcriptional regulation by CREB is critical for the response of excitable cells to environmental cues, prolonged CREB activation is unhealthy and has adverse effects [55,56]. We show that Ascl1 expression is dependent on Ca²⁺-signaling and that it is activated in metabolically stressed β-cells. However, other pathways are likely to also contribute to Ascl1 regulation in response to a HFD. While this study was done in mice, our findings likely extend to humans since β-cells from people with T2D with low INS gene expression have higher ASCL1 expression. Furthermore, in human islets the chromatin surrounding ASCL1 is open [57], suggesting that the gene is poised to respond to an increase in [Ca²⁺]_i.

4.2. Role of Ascl1 in pancreatic β-cells and metabolic stress response

Analysis of Ascl1^βKO mice fed a normal diet revealed no differences in glucose handling, insulin secretion, β-cell area, or hormone cell composition in adult islets. While transcriptional analysis revealed changes in the islet transcriptome of Ascl1^βKO compared to wild type mice, these data further confirm that Ascl1 is not required for normal β-cell development and function. Although changes occur in the transcriptome of islets from Ascl1^βKO mice on ND that include many secretory and ER genes, they do not cause any readily detectable functional impairments. We also did not observe any changes in glucose homeostasis or islet structure when Ascl1 was eliminated from β-cells of Abcc8 KO mice. This is surprising since Ascl1 is highly expressed in Abcc8 KO mice, and was predicted to regulate dedifferentiation genes up-regulated in β-cell by excitotoxicity [19]. The lack of any improvement in islet function in the double KO mice (Ascl1^βKO; Abcc8 KO) indicates that ASCL1 does not contribute to the impairments in β-cell function brought on by the lack of functional K_ATP channels. Instead, genes involved in ER stress and the unfolded protein response, including the pro-apoptotic gene Ddit3 (Chop) [58], are upregulated while hormone secretion genes are downregulated in the absence of Ascl1. Since ER stress due to increased insulin secretory demand contributes to β-cell dysfunction [47], ASCL1-driven dedifferentiation may serve to protect β-cells from the adverse effects of temporary insulin hypersecretion brought on by chronic membrane depolarization [59]. In any case, the disruption of Ascl1 in mice lacking Abcc8 does not result in further impairment of β-cell function. Thus, other yet to be identified transcription factors may be principally responsible for the many gene expression changes observed in Abcc8 KO β-cells.

The most pronounced, and beneficial, effect of eliminating Ascl1 in β-cells occurred in the setting of HFD-driven metabolic stress where we observed a decrease in blood glucose, increased glucose tolerance and increased β-cell proliferation in comparison to controls. Furthermore, we observed broader changes in transcriptome with changes in many genes that are affected by HFD being reversed in Ascl1^βKO islets and included decreases in the dedifferentiation genes Aldh1a3 and S100z [41,60], and the immature and aging β-cell marker Bambi [61]. We also observed an increase in expression of genes associated with insulin secretion (Slc2a2, Glp1r, Mafa) and β-cell proliferation (Foxm1) further indicating that the transcriptional regulation of these and other genes by ASCL1 contributes to β-cell failure during metabolic stress, and that the deletion of Ascl1 protects β-cells from negative effects of HFD.

Notably, some unexpected genes were increased in Ascl1^βKO islets from animals on a HFD. These included several neuronal peptides including gastrin-releasing peptide (Grp) and vasoactive intestinal peptide (Vip). While these peptides can be produced by islet cells, they are mostly secreted by parasympathetic neurons that innervate islets and stimulate insulin secretion [62,63]. We also detected an increase in Slc18a3, the acetylcholine transporter (or VAchT) and Nos1 (n-NOS). Because both are markers of parasympathetic nerves, we studied the Ascl1^βKO mice further and found an increase in neural fibers in Ascl1^βKO islets and improved secretion of isolated islets in response to acetylcholine. In islets, acetylcholine released by parasympathetic neurons and α-cells potentiate insulin secretion via muscarinic 3 cholinergic receptors [64]. While our RNA-seq results did not show any changes in muscarinic receptor expression, there was an increase in expression of the nicotinic cholinergic receptors (nAChRs) Chrna3, Chrnb4 and Chrna7. These changes could either be the result of increased innervation or could be caused by changes in Ascl1 KO β-cells since nicotinic receptors have been shown to be expressed in β-cells [[65], [66], [67]]. In either case, signaling through nAChRs has previously been shown to increase insulin secretion in β-cells [65,68], to improve β-cell function, and to protect β-cells from apoptosis [69,70] and ER stress [71]. ASCL1 has also been shown to regulate CHRNA3/B4 in lung cancer [72]. Thus, it is possible that an increase in nAChR expression in the absence of Ascl1 could also cause the increase in acetylcholine response and improvement in β-cell function that we observed. Parasympathetic signaling was also shown to contribute to FOXM1-driven β-cell proliferation in obesity [73]. Together, these findings suggest that the improvements in β-cell function and increased proliferation that occur in Ascl1^βKO mice fed a HFD may be at least partially due to an increase in islet innervation and increased acetylcholine signaling.

It is not clear how a deletion of Ascl1 in β-cells leads to an increase in islet innervation, however, β-cells are known to produce factors that promote the ingrowth of nerve fibers [74]. In Ascl1^βKO islets the observed increases in expression of neuropeptides and genes encoding neural guidance adhesion molecules (Nptx2, Pcdh7, Slit2) could all contribute to increased innervation. Paracrine signaling between neural and endocrine cells helps to adapt islet function to body insulin demand and the term “neuro-islet plasticity” has been proposed to describe an increase in islet innervation in order to maintain insulin secretion under metabolic stress [75]. The increases we observe in both islet innervation and function in metabolically stressed islets in the absence of Ascl1 suggest that transcriptional regulation by ASCL1 represses this process.

4.3. Transcriptional regulation by ASCL1 in β-cells is context-dependent

The marked variations in transcriptional responses in Ascl1^βKO in different conditions suggest a complex and context-dependent nature of transcriptional regulation by ASCL1 in β-cells. ASCL1 is known to form heterodimers with other bHLH neuronal transcription factors [10] and to interact with various transcriptional co-regulators [76] to exert its effects. Thus, both the binding and actions of ASCL1 may be highly dependent on the combination of transcription factors expressed in β-cells at a given condition. Indeed, it has been suggested that combinatorial binding by various bHLH in heterodimers forms a “bHLH code” where different combinations of bHLH factors bind slightly different E-box sequences to differentially drive the expression of target genes [77]. For example, in neurons, ASCL1 and NEUROG2 alone or in combination together bind to different sequences [78]. In β-cells, metabolic-stress induced ASCL1 and co-regulators may stimulate or block the binding of other bHLH proteins to their cognate elements, thereby affecting gene expression in a highly complex manner. For instance, prediction analysis of TFs that drive genes upregulated on HFD in the absence of Ascl1 predicts NEUROD1, a TF important for mature β-cell function [63], as one of the top regulators. Therefore, it is possible that the induction of ASCL1 by metabolic stress may displace NEUROD1 from target sites, down-regulating the expression of key target genes and impeding β-cell function. In addition, since ASCL1 is a pioneer transcription factor [16] that can bind to a closed chromatin by itself or in combination with other transcriptional regulators, the induction of ASCL1 may affect the epigenetic state of β-cells, impairing their ability to maintain their identity. Further characterization of ASCL1 binding sites and analysis of interacting partners in β-cells during metabolic stress is needed to precisely identify target genes and to better understand how a Ca²⁺-dependent gene network contributes to the metabolic-stress responses of β-cells.

5. Conclusions

Ascl1 expression is induced by metabolic stress and increases in Ca²⁺ signaling to the nucleus and contributes to β-cell failure by activating dedifferentiation program and suppressing genes important for mature β-cell function and parasympathetic innervation. These data provide new insights into the mechanisms by which the metabolic stress associated with overnutrition may lead to a loss in β-cell function and the development of T2D.

Funding

This study was supported by institutional and philanthropic funds from Vanderbilt University. The Vanderbilt Islet and Pancreas Analysis Core is supported by NIH grant DK020593, the Hormone Assay and Analytical Services Core is supported by NIH grants DK059637 and DK020593, Vanderbilt Cell Imaging Shared Resource (CISR) Core is supported by NIH grants CA68485, DK20593, DK58404, DK59637 and EY08126.

Author contributions

ABO: Conceptualization, Investigation, Data Curation, Formal Analysis, Methodology, Validation, Visualization, Supervision, Writing – Original Draft Preparation, Writing – Review & Editing.

FYZ: Investigation, Data Curation, Methodology, Validation.

JJC: Investigation, Data Curation, Methodology, Validation.

LTT: Investigation, Data Curation.

MAC: Data Curation, Methodology, Software, Visualization.

SS: Data Curation, Methodology, Software, Visualization.

JPC: Data Curation, Methodology, Software, Visualization.

MAM: Conceptualization, Project Administration, Funding Acquisition, Supervision, Resources, Writing – Review & Editing.

Data statement

RNA-Seq data are available in ArrayExpress (https://www.ebi.ac.uk/arrayexpress) under accession number E-MTAB-13355.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is the Supplementary data to this article:

Data availability

No data was used for the research described in the article.