Coregulator Sin3a Promotes Postnatal Murine β-Cell Fitness by Regulating Genes in Ca²⁺ Homeostasis, Cell Survival, Vesicle Biosynthesis, Glucose Metabolism, and Stress Response

Islet β-cell differentiation starts during early embryogenesis when a subset of pancreatic multipotent progenitor cells (MPCs) activates the expression of transcription factor (TF) neurogenin 3 (Ngn3 or Neurog3) and subsequently dozens of other islet cell TFs (1). During perinatal growth, the differentiated β-cells proliferate to increase their overall mass while also fine-tuning gene expression profiles to become mature, functional cells (2). Besides TFs, several transcriptional coregulators, including the mixed-lineage leukemia 3 and 4 methyltransferases (3) and Swi/Snf complexes (4), are essential for the production and/or maturation of β-cells. The standard model is that specific combinations of coregulators and TFs organize the chromatin landscape to shape the transcriptomic profiles of progenitors for cell development and function (4).

Swi-independent 3 (Sin3) is a well-established coregulator conserved from yeast to human cells (5). It does not bind DNA but uses several highly conserved amphipathic α-helices to interact with a variety of TFs and other coregulators, including P53, Mad1, Myc, REST, and ESET (5,6). In addition, Sin3 contains binding sites for histone deacetylases, histone lysine methylases, and demethylases (7–9). Thus, Sin3 primarily acts as a scaffold protein to assemble chromatin-modifying complexes that regulate gene transcription, and its target gene selectivity is determined by the DNA-binding TFs that recruit Sin3 (7–9). Notably, although Sin3 was commonly known as a “corepressor,” several studies suggest that it can coactivate gene expression in some cellular contexts (10).

Sin3 has two paralogs in mammalian cells, Sin3a and Sin3b, with evidence showing overlapping yet distinct functions (5,10). In mice, Sin3a, but not Sin3b, is essential in cell survival and/or differentiation during embryogenesis. Nullizygous Sin3a^−/− mouse embryos died shortly after implantation, whereas Sin3b-null mice were born with seemingly normal organs but died immediately after birth (5). Consequently, Sin3a was reported to be required for the development and/or survival of embryonic stem (ES) cells (11,12), muscle cells (13), male germ cells (14), lung progenitors (15), and some skin cells (16). In addition, gene expression and protein-DNA interaction studies showed that Sin3a could directly regulate molecules involved in cell proliferation, survival, metabolism, and stress responses (7,17,18), yet how Sin3a functions in postnatal organs has not been examined.

In pancreatic cells, Sin3a was detected in transcriptional complexes containing Myt, Mafa, and/or Foxo TFs, which are all required for β-cell function (3,19,20). In this study, we assessed the roles and mechanisms of Sin3, focusing on Sin3a, in embryonic development and postnatal function of mouse islet β-cells. We show that Sin3a is dispensable for islet cell differentiation but required for β-cell function and survival, that is, their postnatal “fitness.” In addition, although Sin3 activity, sufficiently provided by either Sin3a or Sin3b, is essential for endocrine specification from MPCs, it is not needed for the differentiation of endocrine progenitors into islet hormone-positive cells. Thus, our data reveal stage- and cell-type–specific roles of the Sin3 complex, with Sin3a being particularly important for postnatal β-cell fitness.

Mouse usage was supervised by the Vanderbilt University institutional animal care and use committee in compliance with Association for Assessment and Accreditation of Laboratory Animal Care regulations. Sin3a^F, Sin3a⁻, Sin3b^F, and Sin3b⁻ mice were previously described (17,21). Pdx1^Cre [Tg(Pdx1Cre)89.1Dam], Ai9 [Cg-Gt(ROSA)26Sor^{tm9(CAG-tdTomato)Hze}/J], and Neurog3-Cre [Tg(Neurog3-Cre)C1Able/J] mice were from The Jackson Laboratory. ICR (CD1) mice were from Charles River Laboratories. All analyzed mice had a mixed genetic background (∼25% CD1, ∼37.5% C57BL/6, and ∼37.5% 129/Sv, as estimated from crossing history). The day of vaginal plug appearance was counted as embryonic day 0.5 (E0.5), and the day of birth was counted as postnatal day 1 (P1). PCR-based genotyping used oligos listed in Supplementary Table 1.

Tissue preparation and immunofluorescence (IF) analysis followed routine protocols using 10-μm-thick frozen sections (19). Primary antibodies (1:500 to 1:2,000 dilution) were as follows: rabbit anti-Sin3a (LS-C331555; Lifespan Biosciences), rabbit anti-Sin3b (ab101841; Abcam), rabbit anti-Mafa (NBP1-00121; Novus Biologicals), rabbit anti-Mafb (IHC-00351; Bethyl Laboratories), rabbit anti-Nkx6.1 (gift from Palle Serup, University of Copenhagen, Copenhagen, Denmark), goat anti-Pdx1 (ab47303 [Abcam] or 06-13850 [Millipore]) (22), goat anti-Neurog3 (this laboratory, available upon request) (23), rabbit anti-Glut2 (GT21-A; Alpha Diagnostic), rabbit anti-Ucn3 (ab79121; Abcam), guinea pig anti-insulin (A0564; Dako), rabbit anti-glucagon (ab92517; Abcam), goat anti-somatostatin (sc-7819; Santa Cruz Biotechnology), rabbit anti–cleaved caspase-3 (#9661; Cell Signaling Technology), and rabbit anti-Ki67 (ab15580; Abcam). Secondary antibodies (1:500) used included the following: Alexa Fluor 488 donkey anti-guinea pig (706-545-148), Cy3 donkey anti-rabbit (711-165-152), Cy3 donkey anti-goat (705-165-147), Alexa Fluor 647 donkey anti-guinea pig (706-605-148), Alexa Fluor 647 donkey anti-rabbit (711-605-152), and Alexa Fluor 647 donkey anti-goat (705-605-147), all from Jackson ImmunoResearch.

Images were taken using a Nikon spinning disk confocal microscope and quantified with ImageJ 1.51j 14 software (National Institutes of Health [NIH]) under double-blind settings using 16-bit images. The presented results are the average intensity per pixel within the region of interest. The region of interest was manually selected to encircle the plasma membrane (for Glut2), cytoplasm (Ucn3), or nuclei (Mafa) of β-cells. We then normalized the relative levels of control cells to 1.0 for comparison. For each sample, 12–15 islet sections were examined. Transmission electron microscopy (TEM) and quantification followed published protocols (19).

One of every 12 sections (∼120 μm apart) of entire P1 and P7 pancreata and 1 of every 18 sections (∼180 μm apart) of P14 pancreata were scored. The sections were costained with DAPI and hormone antibodies to identify all cells and islet cells, respectively. Tissue sections were scanned/analyzed using Aperio ImageScope software (Leica Biosystems). The islet cell mass was calculated as follows: islet cell mass (mg) = (hormone + area / DAPI-positive area) × pancreas weight (mg).

Blood glucose was measured through tail snipping. Plasma insulin/glucagon assays used blood from the retro-orbital collection. Islet isolation and secretion assays followed published protocol (19). Handpicked islets, after overnight recovery in RPMI medium (10% v/v FBS and 11.0 mmol/L glucose), were incubated in Krebs-Ringer buffer (KRB), containing 2.8 mmol/L glucose, for 1 h, with 2.8 mmol/L glucose for 45 min, then in KRB with 20 mmol/L glucose for 45 min, and lastly in KRB with 25 mmol/L KCl and 2.8 mmol/L glucose for 45 min. Insulin was measured using an ELISA kit from ALPCO and glucagon with a kit from Mercodia (Uppsala, Sweden) in the Vanderbilt Hormone Assay and Analytical Services Core. Insulin tolerance test and intraperitoneal glucose tolerance test followed that in Huang et al. (19).

Ca²⁺ imaging followed published protocols (24). P5 islets were attached to poly-lysine–coated dishes overnight in RPMI medium (10% FBS and 11.0 mmol/L glucose). Islets were incubated in RPMI medium/10% FBS/2.8 mmol/L glucose basal media with 2.0 mmol/L Fura-2 AM (Invitrogen) for 20 min, followed by 20 min of washing in basal media. Ca²⁺ imaging was performed in recording solution (119 mmol/L NaCl, 2.5 mmol/L CaCl₂, 4.7 mmol/L KCl, 10 mmol/L HEPES, 1.2 mmol/L MgSO₄, and 1.2 mmol/L KH₂PO₄). The islets were perfused at 2.0 mL/min at 37°C with 2.8 mmol/L glucose (0–195 s), 20 mmol/L glucose (196–915 s), 2.8 mmol/L glucose (916–1,800 s), and 25 mmol/L KCl with 2.8 mmol/L glucose (1,801–2,090 s). Fluorescence images, excited at 340 and 380 nm (F340/F380), were taken/measured every 5 s with a Nikon Eclipse Ti2 microscope and Photometrics Prime 95B 25mm sCMOS camera.

Quantitative RT-PCR (qRT-PCR) followed routine procedures using Universal SYBR Green Supermix (Bio-Rad) on a Bio-Rad CFX96 thermal cycler. Transcript abundance was normalized to Gapdh. Oligos are listed in Supplementary Table 1.

For single-cell RNA sequencing (scRNA-seq), two batches of P4 Sin3a^F/+ and Sin3a^Δendo islets (two to four mice of each genotype used for each batch) were isolated and dissociated into single cells using trypsin-EDTA. After live cell selection, the inDrop platform (1CellBio) was used to encapsulate and barcode single cells used for CEL-Seq–based library preparation and sequencing (NextSeq 500; Illumina). Reads were assigned to individual cells. After adaptor sequences were trimmed and cell doublets corrected, batch alignment assigned the reads to specific gene loci or unique molecular identifiers. We next identified β-cells based on insulin 1 and insulin 2 gene expression, which were then compared with the Mouse Cell Atlas to verify their cell identity (25). Afterward, the number of reads of each unique molecular identifier within each sample were combined and analyzed as one bulk sample using DESeq2. Log(fold change [FC]) was calculated as log(2){[(level in mutant) − (level in control)] / (level in control)}. Statistical analyses used the Wilcoxon rank sum test, and the adjusted P values were derived using the Bonferroni posttest correction on the basis of the total number of genes in the data set. Gene set enrichment assays (GSEAs) followed that in Subramanian et al. (26). Gene ontology–based Database for Annotation, Visualization, and Integrated Discovery analysis of potential Sin3a-binding genes followed that in Huang et al. (27).

A Magna ChIP-HiSens Chromatin Immunoprecipitation Kit (MilliporeSigma) was used. MIN6 cells (passage #38–43) were fixed for 20 min using 2.0 mmol/L disuccinimidyl glutarate followed by an additional 12 min with 1% formaldehyde. Sheared chromatin (200–500 bp) was prepared using a Bioruptor Pico (Diagenode) (∼30 cycles of 30 s sonication + 30 s rest). For each immunoprecipitation, 2.0 μg of chromatin and 2.0 μg of antibodies were used. The normal rabbit IgG was from Cell Signaling Technology (Cat# 2729). The rabbit anti-Sin3a (LS-C331555) was from Lifespan Biosciences.

SPSS version 25 statistical software (IBM Corporation) was used. Data are shown as mean ± SD. Student t test or one-way repeated-measures ANOVA was used to compare two groups or more than two groups of data, respectively. P < 0.05 was considered significant.

The data sets generated and/or analyzed during the current study are deposited in Gene Expression Omnibus (GEO) under GSE146474 and are freely available for all. The data are also available from the corresponding author upon reasonable request.

At E10.5, E15.5, and P1, both Sin3a and Sin3b proteins were detected in most pancreatic cells (Fig. 1A–C), including Neurog3⁺ endocrine progenitors (Fig. 1B) and hormone-positive islet cells (Fig. 1C). After P1, their expression further increased, and they became highly enriched in islet cells (Fig. 1D and E). However, both Sin3a and Sin3b remained detectable in some nonislet cells in the 3-month-old pancreas (Fig. 1E).

Sin3a^F/−; Neurog3-Cre mice (termed Sin3a^Δendo) were derived to inactivate Sin3a in the endocrine progenitors and some neuronal cells (28). IF and qRT-PCR assays supported the efficient deletion of Sin3a in Sin3a^Δendo islets (Fig. 2A and B). We examined male and female Sin3a^F/−, Sin3a^F/+, and Sin3a^F/+; Neurog3-Cre mice. No significant differences were observed in body weight, ad lib–fed blood glucose, islet structure, or expression of several diagnostic endocrine markers (e.g., Mafa, Pdx1, Nkx6.1, hormones), except that Sin3a^F/− female mice had slightly lower body weights compared with Sin3a^F/+ females at 2 weeks of age (Supplementary Fig. 1A–F). Thus, we used Sin3a^F/− mice as controls for most studies unless noted.

Up to P14, both male and female Sin3a^Δendo mice had body weights indistinguishable from control littermates but showed significant growth retardation afterward (Fig. 2C). The Sin3a^Δendo mice showed normal glycemia at P7 but significantly higher blood glucose levels afterward (Fig. 2D). The temporal phenotypic development is similar in both male and female mice (Fig. 2C and D). We therefore used both sexes interchangeably and presented our findings together for the rest of the studies unless noted.

Sin3a^Δendo mice had significantly higher fasting glucose levels on P28 but not on P14 (Fig. 2E). Compromised glucose tolerance was obvious in P14 Sin3a^Δendo mice (Fig. 2F), which had normal insulin sensitivity on P14 but increased sensitivity on P21 (Fig. 2G). P7 and P28 Sin3a^Δendo mice showed reduced plasma insulin (Fig. 2H), and plasma glucagon levels in P7 Sin3a^Δendo mice were reduced (Fig. 2I). These combined findings suggest that reduced circulating insulin, but not increased glucagon or compromised insulin response, causes the hyperglycemic phenotype of Sin3a^Δendo mice. Thus, our following studies focused on β-cells.

We assayed insulin secretion of P7 Sin3a^Δendo islets, the oldest stage when intact mutant islets could be readily isolated. These islets secreted more insulin under 2.8 mmol/L glucose (Fig. 3A) but less under 20 mmol/L glucose (Fig. 3B) or 25 mmol/L KCl (Fig. 3C). Sin3a^Δendo β-cells had compromised Ca²⁺ influx induced by glucose or KCl, with higher levels of basal but a lower stimulated Ca²⁺ influx (Fig. 3D–F). In addition, Sin3a^Δendo β-cells had more vesicles per unit of cytoplasmic area and a decreased size of the insulin dense core within each vesicle and more vesicle-like membrane structures that lacked detectable insulin crystals (Fig. 3G–I).

P7 and P14, but not P1, Sin3a^Δendo mice had significantly reduced β-cell mass (Fig. 4A and B) accompanied by increased apoptosis starting from P1 (Fig. 4C and D) but no change of proliferation (Fig. 4E and F). In addition, Sin3a^Δendo mice had reduced α-cell mass but no change in δ-cell mass in P14 (Supplementary Fig. 2). These combined findings suggest that Sin3a is not required for β-cell differentiation (i.e., producing insulin-positive cells) during embryogenesis but is required for their fitness after birth. This conclusion led us to examine the detailed molecular mechanisms on how Sin3a regulates this fitness.

We used a candidate approach and examined the expression of several TFs (Mafa, Mafb, Pdx1, and Nkx6.1) that are important for β-cell differentiation and maturation markers (Glut2 and Ucn3) that are important for function. Pdx1, Nkx6.1, or Mafb levels showed no difference in Sin3a^Δendo β-cells compared with controls (Supplementary Fig. 3), but Glut2, Ucn3, and Mafa expression was progressively reduced. Specifically, Ucn3 expression was significantly downregulated by P4, while Glut2 and Mafa expression was downregulated by P7 at both protein and mRNA levels (Fig. 5 and Supplementary Fig. 4). However, because the glycemic defects of Ucn3- or Mafa-mutant mice developed much later than the Sin3a^Δendo animals (29,30) and the increased β-cell death was observed before the downregulation of Ucn3, Mafa, and Slc2a2, these three genes are unlikely the major mediators of Sin3a function. Thus, we comprehensively defined the Sin3a-regulated genes in β-cells by RNA-seq approaches.

scRNA-seq was used to examine gene expression of P4 control and Sin3a^Δendo islets, allowing us to distinguish β-cells from other islet cell types. P4 Sin3a^Δendo mice had no recognizable physiological defects, avoiding complications imposed by hyperglycemia, but had increased β-cell death, underscoring the existing molecular defects in the P4 Sin3a^Δendo β-cells.

Two highly reiterative scRNA-seq data sets (Fig. 6A) were obtained, which showed clear islet cell type separations (Supplementary Fig. 5). Differences in expression were seen between control and Sin3a^Δendo β-cells, with both downregulated and upregulated genes (e.g., Ucn3 that recapitulates qRT-PCR results, Hspe1 that represents newly identified Sin3a-regulated genes, respectively) (Fig. 6B).

To minimize the saturation and scarcity issues inherent to scRNA-seq (31), we analyzed the scRNA-seq results by combining identical cell types into bulk expression data within each sample (32). This revealed 772 downregulated and 3,668 upregulated genes in Sin3a^Δendo β-cells, with adjusted P < 0.05 and at least a onefold increase or difference in expression (i.e., the expression level in one sample is at least twice as much as the other; see the table “Sin3 candidate gene list.xlsm” in GEO [GSE146474]). We observed the changed transcript level of Ucn3 but not Mafa, Nkx6.1, or Pdx1, consistent with IF (Supplementary Fig. 3) and qRT-PCR results (Fig. 5P). GSEA of the scRNA-seq data set revealed several altered molecular pathways (Fig. 6C). The terms include protein/membrane transport/endoplasmic reticulum function, tricarboxylic acid cycle/mitochondrial activities, and oxidative stress responses.

Sin3a was reported to associate with several widely expressed DNA-binding proteins, including P53, Myc, Mad1, and Foxo1 (20,33). Sin3a was also detected in a Mafa-containing transcriptional complex, highly expressed in islet β-cells (3). We postulated that Sin3a shares common target genes among different cell types. Bioinformatics analyses were used to identify putative Sin3 target genes in β-cells by comparing the list of Sin3a-dependent genes with published Sin3a-based chromatin immunoprecipitation sequencing (ChIP-seq) results in mouse ES cells (34), epiblast stem cells (35), and muscle cells (13). We also included the Mafa-bound enhancers in mouse islets, postulating that the Sin3a associates with these sites through Mafa (35,36). This analysis revealed 2,847 gene loci having at least one reported Sin3a/Mafa-binding site, with 335 being downregulated and 2,512 being upregulated in Sin3a^Δendo β-cells (see the table “Sin3 candidate gene list.xlsm” in GEO [GSE146474]). Gene ontology–based analyses (Database for Annotation, Visualization, and Integrated Discovery) revealed that these Sin3a-associated genes likely regulate processes in mitochondrion, endoplasmic reticulum/protein transport, ribosome, cell redox homeostasis, and synapses (Fig. 6C and D), similar to pathways revealed by GSEA analysis of all differentially expressed genes.

ChIP-PCR was used to verify whether some of these predicted genes are direct targets of Sin3a in MIN6 β-cells. We prioritized several genes that have established molecular functions that correlated with the cellular defects of Sin3a^Δendo β-cells (Table 1). These include downregulated Ins1 and Kcnh2 required for insulin production and cell membrane polarization (37). They also include upregulated genes Esyt1 and Calr (related to defective Ca²⁺ homeostasis) (38,39); Bnip3, Casp3, and Ing1 (cell death) (40,41); Arl6, Cltb, Ergic3, and Rab11a (lipid transport and vesicular biosynthesis) (42–45); Aldoa and Idh3a (glucose metabolism); and Hsp90b1 and Hspe1 (stress responses) (46). Mafa was also included because Sin3a enrichment was detected on its putative regulatory elements (Table1). Ucn3, Slc2a2, Gcg, Gapdh, and albumin were included as controls because Sin3a was not reported to directly regulate these genes (Fig. 6E and Supplementary Fig. 6).

For ChIP-PCR, we selected several putative regulatory regions (RRs) in the 5′ region of each gene with putative Myc/P53-binding sites (47) (Supplementary Table 1). We found significantly enriched Sin3a occupancy in the 5′ RRs of Kcnh2, Esyt1, Bnip3, Casp3, Ing1, Cltb, Rab11a, Aldoa, Hsp90b1, and Hspe1 (Fig. 7A). Corresponding to these results, qRT-PCR assays detected significantly decreased transcription of Kcnh2 but increased expression of Bnip3, Casp3, Ing1, Rab11a, Aldoa, and Hspe1 in P4 Sin3a^Δendo islets, matching the scRNA-seq results (Fig. 7B and Table 1). We did not detect Sin3a enrichment in the tested regions of Ins1, Gcg, Calr, Arl6, Ergic3, and Ucn3 genes (Supplementary Fig. 6). Intriguingly, we also found enriched Sin3a binding to the 5′ regions of Slc2a2 and Mafa, although their expression levels remain unchanged in P4 Sin3a^Δendo β-cells (Table 1). It is not clear whether Sin3a directly regulates these two genes in older β-cells.

An intriguing finding is the upregulation of Sin3b in the absence of Sin3a (Table 1). We therefore tested whether inactivating Sin3b could further exacerbate the defects in Sin3a^Δendo β-cells. Inactivating Sin3b alone in Sin3b^F/−; Neurog3-Cre mice produced no detectable phenotypes (data not shown), yet Sin3a^F/−; Sin3b^F/−; Neurog3-Cre (Sin3a/3b^Δendo) mice became diabetic before P5, at least 1 week earlier than Sin3a^Δendo mice (Supplementary Fig. 7A), accompanied by largely reduced β-cell production (Supplementary Fig. 7B). These findings support the redundant function of Sin3a and Sin3b for the production/survival of islet β-cells.

Finally, we tested the roles of Sin3a and Sin3b in the pancreatic MPCs. Sin3a^F/F; Sin3b^F/F; Pdx1^Cre (yielding Sin3a/3b^ΔPanc mice), and Sin3a^F/F; Sin3b^F/F; Pdx1^Cre; Ai9 mice were derived. The Ai9 Cre-reporter allele in the latter marked Sin3a/Sin3b-deficient pancreatic cells with tdTomato (23). At E15.5 and P1, there was substantially reduced production of β- and α-cells in Sin3a/3b^ΔPanc pancreata (Supplementary Fig. 7C–E), accompanied by reduced Neurog3⁺ cell production (E15.5) (Supplementary Fig. 7F) and increased β-cell death (P1) (Supplementary Fig. 7G). Notably, the majority of pancreatic cells in Sin3a^F/F; Sin3b^F/F; Pdx1^Cre; Ai9 mice expressed tdTomato, suggesting that Sin3a/3b-deficient pancreatic cells do not die immediately after Sin3 inactivation (Supplementary Fig. 7H).

The mammalian Sin3a and Sin3b paralogs are scaffold proteins for the overall Sin3 coregulator complex that can associate with common TFs, such as Foxo, Mad1, Myc, and P53, and β-cell TFs Mafa and Myt (3). These associations recruit histone deacetylases, histone lysine methylases, and demethylase to regulate gene transcription in a highly cell context–dependent manner (5). We report here that inactivating Sin3a singly in the early mouse pancreatic endocrine lineage has little effect on islet cell differentiation but substantially reduces postnatal β-cell fitness without affecting the survival of δ-cells. The Sin3a-deficient β-cells showed defective insulin secretion, cell survival, Ca²⁺ influx, and insulin vesicle biogenesis. We further showed that Sin3a associates with putative enhancers of several genes involved in ion transport/Ca²⁺ homeostasis, cell death, membrane trafficking, glucose metabolism, and stress response in β-cells. Intriguingly, inactivating both Sin3a and Sin3b in the same endocrine lineage resulted in further reduction of endocrine cell numbers by birth, while coinactivating both Sin3a and Sin3b in the early pancreatic MPCs substantially reduced production of endocrine progenitors without preventing their differentiation into hormone-positive islet cells.

Our discoveries highlight a few features of Sin3 function in pancreatic cells. First, although Sin3a but not Sin3b is required for producing postnatal β-cells with similar “fitness qualities,” both paralogs can prevent cell death. Thus, inactivating Sin3b expedited islet cell loss in Sin3a-null islets. It is possible that Sin3a and Sin3b regulate a similar set of molecular targets, yet Sin3a contributes a higher proportion of such activity because of a higher affinity for transcriptional effectors. Alternatively, Sin3a and Sin3b may regulate different sets of target genes that have similar functions. Future comparisons of transcriptomic alterations between Sin3a-null and Sin3b-null β-cells together with comprehensive ChIP-seq analyses should address these possibilities. In either case, it appears that the total Sin3 activity has to achieve a certain threshold to maintain normal β-cell fitness.

The second feature is that different islet cell types have different levels of Sin3a dependency, with α- and β-cells, but not δ-cells, requiring Sin3a for postnatal survival. It is possible that Sin3a regulates similar gene sets in all islet cell types but that the cellular context dictates resistance to cell death pathways. Alternatively, different islet cell types may use different Sin3a-TF complexes to regulate transcription. Follow-up identification of these complexes using proteomic studies could address this question.

Third, although Sin3 (either Sin3a or Sin3b) is required in MPCs to promote endocrine specification, their overall activities are dispensable for the differentiation of committed endocrine progenitors into hormone-expressing islet cells. Therefore, our collective findings, combined with Sin3a’s reported roles in the differentiation of several nonpancreatic cells (13–15,17,48), highlight the idea that pancreatic cells use the Sin3-TF complexes differentially in a stage- and/or cell type–specific manner for differentiation, function, and survival.

The compromised β-cell fitness in Sin3a-null mice is consistent with Sin3a being a coregulator recruited by Mafa and Foxo1, both of which interact with Sin3a and are essential for β-cell fitness (20,49). However, Sin3a-null mice developed overt diabetes before weaning, while Mafa- and Foxo1-deficient mice only did so a significant period afterward. An implication is that Sin3a inactivation may be equivalent to coinactivation of Mafa and Foxo1. Alternatively, Sin3a may also mediate the function of additional TFs required for β-cell function and/or survival (e.g., Myc and P53), which are ubiquitous in most cell types. Our ChIP-PCR assays detected Sin3a enrichment in putative Myc/P53-binding enhancers of several genes in β-cells whose products regulate cell death, ion transport, lipid trafficking/vesicular biosynthesis, metabolism, and stress responses (Table 1). Thus, our collective findings are consistent with the idea that Myc/P53 could recruit Sin3a in β-cells to promote β-cell fitness by ensuring insulin vesicle biosynthesis, stimulus secretion coupling, and survival. Note that the complete list of key Sin3a/Sin3b targets in β-cells remains unknown. Future genome-wide ChIP-seq studies of β-cells using specific Sin3a and Sin3b antibodies, combined with transcriptomic analysis of the few Sin3a/3b^ΔEndo and/or Sin3a/3b^ΔPanc β-cells that are formed, may shed more light on how Sin3a and Sin3b regulate β-cells.

Sin3 has been functionally diagnosed as both a coactivator and a corepressor. Consistent with this notion, we found both downregulated and upregulated genes in Sin3a-deficient β-cells, with many being reported as directly regulated by Sin3a in other cell types. Our candidate ChIP-PCR assays corroborated enrichment of Sin3a in putative regulatory regions of Sin3-repressed (e.g., Hspe1, Casp3) and -activated genes (Kcnh2) in β-cells, underscoring the bidirectional regulatory roles of Sin3a.

In summary, our findings show that the Sin3 coregulator plays essential roles in islet-cell production and postnatal β-cell fitness, with Sin3a being the major contributor. Thus, modulating Sin3a levels or activities could be explored to protect β-cell fitness and to control diabetes initiation and progression.

Acknowledgments. The authors thank Dr. Roland Stein (Vanderbilt University) for useful discussions about the Sin3 activities, Drs. Chin Chiang and Lei Chen (Vanderbilt University) for help with ChIP-PCR, and William Gu (Montgomery Bell Academy, Nashville, TN) for text editing.

Funding. This study is supported by grants from the National Institute of Diabetes and Digestive and Kidney Diseases (DK-106228 to I.K. and G.G., DK-103831 and CA-095103 to K.S.L., and DK-065949 to G.G.) and JDRF (1-2009-371 to G.G.). Confocal and TEM imaging was performed with Vanderbilt University Cell Imaging Shared Resource (supported by National Institutes of Health [NIH] grants CA-68485, DK-20593, 500 DK-58404, DK-59637, and EY-08126). Vanderbilt University Medical Center Hormone Assay and Analytical Services Core was supported by NIH grants DK-059637 (Mouse Metabolic Phenotyping Center) and DK-020593 (Diabetes Research and Training Center) and the Islet Procurement & Analysis Core by DK-020593. DNA sequencing was done by Vanderbilt Technologies for Advanced Genomics, supported by the Vanderbilt Ingram Cancer Center (P30-CA-68485), the Vanderbilt Vision Center (P30-EY-08126), and NIH/National Center for Research Resources (G20-RR-030956).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. X.Y., S.M.G., K.-H.H., D.A.J., K.S.L., and G.G. analyzed data. X.Y., S.M.G., A.J.S., and G.G. performed the experiments. X.Y., D.A.J., I.K., C.V.E.W., K.S.L., and G.G. designed the experiments. X.Y., K.S.L., and G.G. prepared figures. C.N.H., B.C., A.N.S-S., and K.S.L. performed the scRNA-seq analysis. G.D. provided the mice harboring Sin3a/Sin3b floxed alleles. All authors helped with data interpretation, participated in manuscript writing/proofreading, and the approved final version of the manuscript. G.G. 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 at the 11th Annual Midwest Islet Club, St. Louis, MO, 14-15 May 2018, and 12th Annual Midwest Islet Club, Ann Arbor, MI, 19–20 May 2019.