Nuclear Factor-Y in Mouse Pancreatic β-Cells Plays a Crucial Role in Glucose Homeostasis by Regulating β-Cell Mass and Insulin Secretion Free

Because pancreatic β-cells are responsible for the biosynthesis and release of insulin that maintains the body’s glucose homeostasis, and β-cell mass plays an essential role in determining the amount of insulin, type 2 diabetes mellitus (T2DM) has been thought of as a disease of deficient β-cell mass and function unable to meet the demands of insulin resistance (1,2). Hence, the molecular mechanisms and pathways implicated in the β-cell mass expansion and insulin secretion have attracted considerable interest as potential therapeutic targets to prevent or delay the onset of T2DM.

Pancreatic β-cell mass and function are determined by complicated signaling cascades that require dynamic changes of transcription factor expression levels in response to altered metabolic demand (3,4). Nuclear factor-Y (NF-Y) is an evolutionarily conserved transcription factor binding to the CCAAT motif in the proximal promoter region, and it is composed of three subunits (i.e., NF-YA, NF-YB, and NF-YC) (5). Because of its general binding consensus and ubiquitous expression, NF-Y has been shown to regulate an increasing number of genes and play multiple roles in various contexts (6–9). In addition, NF-Y is induced by multiple extracellular signals, such as growth factors, apoptotic and inflammatory signals, and hormones, in a cell type–specific manner (10). Given the diversity of pathways that regulate NF-Y expression, it is likely that the biological functions of NF-Y also depend highly on cell type and context. However, the function of NF-Y in the control of β-cell mass and function has not previously been explored. We recently identified an association of Nf-ya genetic variants with the risk of diabetes (11) and reduced blood glucose levels in the Nf-ya liver-specific knockout mice (9). Moreover, recent evidence reveals that NF-Y acts as an activator in the promoter of G6pc2 (12), which encodes an islet-specific glucose-6-phosphatase–related protein (IGRP). IGRP has been proposed to modulate β-cell glycolytic flux and glucose-stimulated insulin secretion (GSIS) by opposing the action of glucokinase (13). Thus, NF-Y likely plays essential roles in the maintenance of β-cell homeostasis, and its in vivo function in β-cells warrants further investigation.

To investigate the physiological role of NF-Y in pancreatic β-cells, we set out to generate a conditional knockout of Nf-ya specifically in β-cells (Nf-ya βKO). In this study, we show that Nf-ya βKO mice exhibit diabetes-like symptoms characterized by impaired insulin secretion and glucose intolerance. These phenotypes are most likely due to both impaired islets mass and insulin secretion, suggesting critical roles of NF-Y in maintaining β-cell homeostasis. Mechanistically, we find that NF-Y is essential for β-cell proliferation by regulating the organization of cytoskeleton. In addition, our results demonstrate that loss of NF-Y in β-cells results in a reduced expression of GLUT2 and cellular mitochondrial dysfunction. Collectively, our studies demonstrate that NF-Y is an essential regulator of β-cell mass and insulin secretion in Nf-ya βKO mice and in isolated islets ex vivo, as well as in mouse insulinoma cell lines.

All animal experiments were approved by the Animal Care and Use Committee of Sichuan University. Mice were housed under a 12-h light/dark cycle with free access to food and water. Leptin-deficient (ob/ob) mice, C57BL/6J mice, and Ins2-Cre mice were purchased from The Jackson Laboratory (Bar Harbor, ME). Nf-ya exons 3–8 floxed mice were a gift from N. Nobuyuki’s laboratory at Juntendo University Graduate School of Medicine (Tokyo, Japan). Nf-ya flox (Nf-ya^fl/fl) mice bred with the Ins2-Cre mice generated Nf-ya^fl/+;Cre mice, which were then mated via brother and sister to generate Nf-ya^fl/fl;Ins2-Cre mice (referred as Nf-ya βKO) (Supplementary Fig. 1A). Littermate mice with the genotypes of Nf-ya^+/+Cre, Nf-ya^fl/+;Cre, and Nf-ya^fl/fl were used as controls. Mice were confirmed by genotyping with the primers shown in Supplementary Table 1 and Supplementary Fig. 1B. Mice were given either standard chow or a high-fat diet (HFD) containing 60% fat (TROPHIC Animal Feed High-Tech Co., Ltd, Shanghai, China) from 8 weeks of age. Only male mice were used for experiments.

Mice were given BrdU intraperitoneally at a dose of 50 mg/kg body wt once daily for 3 days. Two hours after the final injection, mice were euthanized, and pancreata were processed for histology.

Serum concentrations of insulin, triglyceride, and cholesterol were determined as previously described (9). A glucose tolerance test (GTT), GSIS, and insulin tolerance test (ITT) were performed as previously described (14).

Mice islets were isolated using the intraductal collagenase digestion technique (15). Islets were sequentially incubated in Krebs-Ringer bicarbonate buffer containing 2.8 mmol/L and 16.7 mmol/L of glucose at 37°C for 1 h. The total insulin content of islets was harvested by overnight extraction in an ethanol/HCl buffer at 4°C. Secreted insulin in the buffer was measured with the insulin ELISA kit (catalog number 90080; Crystal Chem, Elk Grove Village, IL).

Isolated islets or NIT1 cells cultured in RPMI 1640 medium without glucose or carbon sources were stained with fluorescent derivative of glucose 2-(N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl)amino)-2-deoxyglucose (2-NBDG). The fluorescence was detected at excitation (λ) 475 nm and emission (λ) 550 nm on a microplate reader.

Hematoxylin and eosin staining, immunohistochemistry (IHC), immunofluorescence staining, and TUNEL were conducted according to the previous operations (14,16) with antibodies shown in Supplementary Table 2. Filamentous actin (F-actin) in islets was stained using Alexa Fluor 488–conjugated phalloidin.

The cytosolic Ca²⁺ concentration was measured using the Ca²⁺-sensitive dye Fluo-3 AM in accordance with the product instructions. Briefly, Fluo-3 AM dissolved in anhydrous DMSO was diluted to make a working solution of 10 to 20 μmol/L in Hanks’ and HEPES buffer. Islets were incubated in the working solution at 37°C for 60 min. Then, islets were perifused at 37°C at a flow of 2 mL/min with Hanks’ and HEPES buffer–based solutions that contained 16.7 mmol/L of glucose or 30 mmol/L of KCl. Fluorescence imaging was performed using a Nikon N-STORM and A1 confocal scanning microscope.

Islets were incubated with fluorescent probe JC-1 and imaged using 530-nm and 590-nm emission filters on an AxioVert fluorescent microscope.

ATP levels were measured using a cellular ATP assay kit (Toyo Ink, Tokyo, Japan) following the manufacturer’s instructions.

Mice pancreas sections were postfixed in 1% OsO₄, washed in 0.1 mol/L phosphate buffer, and dehydrated via a graded ethanol series. Samples were embedded in propylene oxide/Poly/Bed 812 epoxy resin overnight. Ultrathin sections (70-nm) stained with 2% uranyl acetate and 1% lead citrate were imaged under a Philips CM120 scanning transmission electron microscope (TEM).

Proteomics analysis of the mice islets was performed as described previously (17). Mice islets were lysed by boiling in 6 mol/L guanidine, and total proteins were precipitated with trichloroacetic acid and digested overnight at room temperature with trypsin. The protein digest was desalted using Strata-X columns (Phenomenex Inc., Torrance, CA) followed by an acidification with trifluoroacetic acid. Samples were then loaded onto the equilibrated Strata-X columns and eluted with 0.2% formic acid into clean tubes with 1 mL of 80% acetonitrile. After drying, peptides were reconstituted in 0.2% formic acid and then analyzed with liquid chromatography-tandem mass spectrometry using an Orbitrap Fusion Lumos Mass Spectrometer (Thermo Fisher Scientific, Waltham, MA). Data were processed with Proteome Discoverer 2.0 software (Thermo Fisher Scientific). Peptides were matched against the mouse reference proteome database from UniProt Knowledgebase (https://www.uniprot.org/). The processes of major biologic significance were determined on the basis of the gene ontology (GO) annotation function.

A series of the Glut2 upstream regions (−1,380 to +114 base pairs [bp], −624 to +114 bp, and −72 to +114 bp) were PCR-amplified with specific primers (Supplementary Table 3), and they were then cloned into the pGL3-basic luciferase reporter vector. The Nf-y binding site (CCAAT) in the constructs was mutated using a site-directed mutagenesis kit with the primers containing specific mutations (Supplementary Table 3). The cDNAs encoding mouse Nf-ya, Nf-yb, and Nf-yc amplified using gene-specific primers (Supplementary Table 3) were subcloned into the pLVX-puro vector, which was then transfected into HEK293 cells to allow amplification of the recombinant lentivirus.

NIT-1 β-cells were coinfected with recombinant Nf-ya, Nf-yb, and Nf-yc lentiviruses and selected using puromycin. Nf-y stable cells were sequentially stained with 5-ethynyl-2′-deoxyuridine (EdU) and Hoechst 33342 dye. Digital images were captured by a fluorescence microscope.

Luciferase activity assay was carried out as described previously (9). After transfection of Glut2 promoter construct or control vector, HEK293 cells were infected with vehicle or Nf-ya, Nf-yb, and Nf-yc constructs. The luciferase activities were measured using the Dual-Glo luciferase system.

Total RNA from mouse islets was extracted using TRIzol reagent (Invitrogen, Carlsbad, CA). First-strand cDNA was prepared from 500 ng of total RNA using a reverse-transcription kit (Invitrogen). Quantitative PCR was conducted as previously described (18) using primers shown in Supplementary Table 4.

Protein extraction and Western blotting with the indicated primary antibodies (Supplementary Table 2) were performed as described elsewhere (19).

All of the data represent at least three independent experiments and are presented as mean ± SD. Statistical significance with P ≤ 0.05 was determined by an unpaired, two-tailed Student t test.

The data sets generated and/or analyzed during the current study are available from the corresponding author upon reasonable request.

Immunostaining of mice pancreas sections showed that all three subunits (NF-YA, -YB, and -YC) of NF-Y were highly expressed in pancreatic β-cells (Fig. 1A). We examined potential changes of NF-Y levels in pancreatic islets under various pathologic conditions linked to diabetes. The leptin-deficient ob/ob mouse is generally accepted as a T2DM model displaying a dramatic increase in β-cell mass to compensate for increased insulin demand (20), and pancreatic IHC for NF-Y showed an increase in abundance in ob/ob mice compared with the lean mice (Fig. 1B). Similarly, HFD-induced obese mice showed higher pancreatic expression levels of NF-Y compared with their control groups (Fig. 1C). Multiple low doses of streptozotocin result in a mild impairment in insulin secretion that more closely resembles the later stages of T2DM (21), and streptozotocin-treated mice exhibited markedly decreased pancreatic NF-Y protein levels compared with the control mice (Fig. 1D). T2DM is also an aging-associated disease (22); pancreatic NF-YA was detected at higher levels in aged pancreas as compared with that in the young islets (Fig. 1E).

Subsequently, we examined the physiological role of NF-Y in pancreatic β-cell homeostasis using β-cell–specific Nf-y knockout (Nf-ya βKO) mice. For the metabolic studies, we included cohorts of mice carrying each single allele as controls to ensure these individual mutations were not contributing to the observed metabolic defects (Fig. 2). The NF-Y protein in islets isolated from Nf-ya βKO mice was reduced to 34.5% of that from Nf-ya^fl/fl littermates (Fig. 2A). Immunofluorescence staining confirmed the largely reduced expression of NF-Y in β-cells (Supplementary Fig. 1C). Nf-ya βKO mice showed no obvious phenotype with respect to body weight (Fig. 2B), feeding (Fig. 2C), and plasma lipid levels (Supplementary Fig. 2A). Compared with the control (Nf-ya^+/+Cre, Nf-ya^fl/+;Cre, and Nf-ya^fl/fl) mice, the Nf-ya βKO mice showed significantly elevated blood glucose levels under either feeding states or fasting conditions (Fig. 2D) by 12 weeks of age. This hyperglycemic phenotype of Nf-y KO mice was even observed at 2 weeks of age (Supplementary Fig. 2B). To determine whether the higher glucose levels in βKO mice are related to insulin secretion defect, we determined their blood insulin concentrations, which were significantly lower in Nf-y βKO mice under both feeding and fasting conditions than those in controls (Fig. 2E), indicating that Nf-y–deficient β-cells secrete insufficient amounts of insulin to maintain normal glucose levels.

To determine whether the disruption of NF-Y in β-cells corresponds to abnormalities in the temporal control of glucose metabolism, we analyzed glucose homeostasis by GTT in 12-week-old mice. Glucose concentrations during GTT were higher in the Nf-ya βKO mice (Fig. 2F), and the area under the blood glucose curve (AUC) was increased by 27.78% (P < 0.01) in the Nf-ya βKO mice compared with control littermates (Fig. 2G). However, the effectiveness of insulin in lowering blood glucose as shown in ITT was quite the same between Nf-y βKO and control mice (Fig. 2H). Finally, GSIS assay indicated that the significantly increased insulin secretion in response to glucose stimulation observed in control mice was blunted in βKO mice (Fig. 2I and J).

To further characterize the effect of NF-Y deficiency, we therefore established HFD-induced obesity. Nf-ya βKO and control mice demonstrated similar weight gain upon feeding with the HFD (Supplementary Fig. 3A). Although high-fat feeding induced hyperglycemia in mice of both genotypes, the increased blood glucose level in Nf-ya βKO mice was significantly higher than in Nf-ya^fl/fl mice (Supplementary Fig. 3B). GTT assay revealed a greater impairment in glucose homeostasis in HFD-fed Nf-ya βKO mice relative to diet-matched wild-type mice (Supplementary Fig. 3C), indicating a worsening of diabetes. In contrast, Nf-y βKO and Nf-ya^fl/fl mice fed HFD demonstrated an equivalent degree of insulin resistance, as shown in the ITT assay (Supplementary Fig. 3D). Furthermore, similar to mice fed a chow diet, HFD-fed Nf-ya^fl/fl mice had a prolonged increase in insulin secretion following glucose injection, while GSIS was unchanged in the HFD-fed Nf-ya βKO mice (Supplementary Fig. 3E). Collectively, these results indicate that the absence of Nf-y in β-cells reduces both glucose tolerance and GSIS and worsens HFD-induced glucose intolerance and diabetes, but has no effect on systemic insulin sensitivity.

The unaltered insulin sensitivity in peripheral tissues suggested the glucose intolerance in Nf-y βKO mice may arise from reduced insulin production or secretion from β-cells. First, we examined the morphology of the pancreatic islets. Compared with age-matched Nf-ya^fl/fl mice, the islet area of Nf-ya βKO mice was diminished by ∼39% (Fig. 3A and B). Additionally, the islet size (Fig. 3C), the density (Fig. 3D), and the total number of islets (Fig. 3E) were markedly decreased in Nf-ya βKO mice. Subsequently, we determined pancreatic β-cell mass and insulin production. Although the whole pancreas weight was comparable between Nf-ya βKO and Nf-ya^fl/fl mice (Supplementary Fig. 4A), Nf-ya βKO mice exhibited significantly lower pancreatic insulin staining (Fig. 3F and Supplementary Fig. 2C) and decreased total insulin content per pancreas weight (Fig. 2G and Supplementary Fig. 2D), indicating that NF-Y deficiency results in a defect in insulin production. In line with the diminished pancreatic insulin content, the abundance of transcripts or proteins of Ins1 and Ins2, along with β-cell markers of MAFA, PDX1, NEUROD, and GLUT2, were also significantly reduced in islets from Nf-ya βKO mice (Fig. 3H and Supplementary Fig. 4B). Moreover, Nf-ya βKO islets were more disorganized than those in Nf-ya^fl/fl littermates (Fig. 3I), and the ratio of β-cells/α-cells was significantly decreased in Nf-ya βKO islets compared with that in Nf-ya^fl/fl islets (Fig. 3J and K). Together, these findings suggest that NF-Y deficiency in islets leads to a reduction in β-cell mass.

Additionally, we also determined whether NF-Y was required for normal β-cell mass expansion in response to HFD-induced insulin resistance. The islet compensation was very obvious in control mice after HFD versus chow diet, especially regarding the size of the islets (Supplementary Fig. 3F). However, the islet showed almost no expansion in Nf-ya βKO mice after HFD (Supplementary Fig. 3F). The diminished islet expansion was at least in part caused by a significant reduction in average β-cell mass (Supplementary Fig. 3G and H). These results indicate that normal NF-Y expression is indispensable to the compensatory β-cell mass expansion in the context of diet-induced insulin resistance.

To better understand when β-cell mass is first affected in Nf-ya βKO mice, we monitored the newly born β-cells in control and Nf-ya βKO mice from postnatal day 1 (P1) until 14 days after birth (P14). Nf-ya βKO mice showed comparable body weight in comparison with control animals at all times (Fig. 4A), but they displayed higher fasting blood glucose levels by P14 (Fig. 4B). Next, we measured the relative insulin⁺ area of islets and found that the β-cell mass was comparable between Nf-ya^fl/fl and Nf-ya βKO mice at P1, but significantly decreased from P7 onwards in Nf-ya βKO animals (Fig. 4C), showing that NF-Y is required for postnatal expansion but not for establishing prenatal β-cell mass. Furthermore, we investigated the molecular defects in Nf-ya βKO β-cells via immunofluorescence for various proteins important for β-cell development and function. While no change was observed in the levels of transcriptional factors MAFA and PDX1 at P1, the levels of these β-cell markers were reduced at P7, which became more pronounced at P14 in Nf-ya βKO mice (Fig. 4D and E). Another β-cell signature gene is GLUT2, which is essential for glucose sensing of mature β-cells (23). Immunostaining for GLUT2 revealed Nf-ya βKO animals exhibited a striking reduction of GLUT2 at P7 and a striking reduction at P14 (Fig. 4D and E), supporting the loss of β-cell maturation and function in these mice. Taken together, these data suggest that the NF-Y defect in β-cells results in impaired postnatal β-cell maturation.

Diminished β-cell mass in Nf-ya βKO mice could arise from a change in β-cell proliferation or viability. Compared with littermate controls, the β-cell proliferation monitored by either the number of Ki67-positive cells (Fig. 5A) or BrdU incorporation (Supplementary Fig. 5A) was markedly decreased in Nf-ya βKO pancreatic sections. Simultaneously, both the mRNA expression and protein levels of PCNA and cyclin D1 implicated in β-cell proliferation were significantly decreased in Nf-ya βKO islets (Fig. 5B and C). To further confirm the effect of NF-Y expression on the β-cell proliferation, all three subunits of NF-Y were expressed simultaneously in mouse NIT-1 insulinoma cells. The forced expression of NF-Y significantly induced PCNA, cyclin B1, and cyclin D1 protein levels (Fig. 5D). Correspondingly, NF-Y overexpression led to a significant increase of β-cell proliferation as indicated by a high EdU incorporation into β-cells (Fig. 5E). However, β-cell apoptosis monitored by TUNEL staining was not significantly altered in Nf-ya βKO mice compared with Nf-ya^fl/fl mice (Supplementary Fig. 5B). Collectively, these results indicated that impaired proliferation contributed to the reduced islet mass in Nf-ya βKO mice.

To further understand how the absent NF-Y influences β-cell mass and proliferation, we measured the whole-islet proteomes of Nf-ya βKO mice and the age-matched Nf-ya^fl/fl mice. A total of 21,263 unique peptides, corresponding to 2,674 quantified proteins for each sample, were detected. The hierarchical clustering yielded 100 proteins with an absolute fold change >1.5 and a P value <0.01 between groups (Supplementary Fig. 6). Of these, 54 were downregulated and 46 were upregulated in Nf-ya βKO mice compared with the control mice (Supplementary Tables 5 and 6). The most highly downregulated proteins in islets from Nf-ya βKO mice are enriched for GO term “cytoskeleton organization” (Fig. 5F). Immunofluorescence staining showed that F-actin cytoskeleton had a similar appearance, with long filaments in β-cells of Nf-ya βKO and control mice, while the cortical actin density was decreased in Nf-ya βKO β-cells (Fig. 5G and Supplementary Fig. 7), indicating a cytoskeleton perturbation by Nf-ya deletion. The cytoskeleton organization regulates cell cycle progression (24), and an abnormal cytoskeleton arrangement delays progression of mitosis (25). Therefore, the impaired β-cell proliferation might be partly attributed to the cytoskeleton perturbation.

Impaired glucose tolerance may also arise from the insufficient β-cell function. Thus, we examined whether NF-Y is essential for β-cell function in isolated islets under ex vivo culture. Nf-ya βKO islets showed a trend toward a reduction in the basal insulin secretion but without statistical significance relative to the Nf-ya^fl/fl islets (Fig. 6A). After being challenged with high-concentration glucose (16.7 mmol/L), the wild-type islets exhibited a significantly elevated insulin secretion, which was blunted in Nf-ya βKO islets (Fig. 6A), consistent with our in vivo findings. These observations suggest that impaired glucose tolerance in Nf-ya βKO mice may be also attributed to an insufficient insulin secretion in addition to the islet integrity.

Since glucose stimulates insulin secretion by inducting depolarization and Ca²⁺ influx (19), we measured intracellular Ca²⁺ levels in perifused islets exposed to 16.7 mmol/L glucose. Although Nf-ya βKO islets displayed some glucose-evoked Ca²⁺ influx, which was significantly compromised in comparison with control islets (Fig. 6B–D). β-Cell Ca²⁺ entry is induced by the closure of K⁺ channels with consequent depolarization; we therefore examined the potential effect of Nf-ya deletion on the K⁺ channel activity in β-cells. Ca²⁺ entry induced by 30 mmol/L KCl in the presence of basal 2.8 mmol/L glucose was equivalent in Nf-ya^fl/fl and Nf-ya βKO islets (Fig. 6E–G). In line with these observations, Nf-ya βKO islets showed a normal insulin secretory response to KCl challenge (Fig. 6H). These data indicated that the defect of GSIS in Nf-ya βKO islets was independent of K_ATP channels, and NF-Y–modulating insulin secretion involves events upstream of membrane depolarization.

The impaired Ca²⁺ response to glucose challenge in Nf-ya βKO islets may result from the abnormal GLUT2, which functions as a major glucose transporter in rodent islet β-cells (23). Both mRNA expression (Fig. 3H) and protein levels (Supplementary Fig. 4B) of GLUT2 were significantly downregulated in Nf-ya βKO islets. Consistent with these observations, the degree of GLUT2 membrane localization and the signal intensity per islet was significantly reduced in Nf-ya βKO mice (Fig. 7A). Furthermore, the glucose uptake monitored by the uptake of 2-NBDG was markedly decreased in βKO islets compared with Nf-ya^fl/fl islets (Fig. 7B), suggesting that NF-Y was required for normal GLUT2 expression and GLUT2-mediated glucose uptake into β-cells. This was further confirmed in mouse NIT-1 β-cells overexpressing all three subunits of NF-Y simultaneously. The forced expression of NF-Y significantly induced GLUT2 protein level (Fig. 7C) and, as a result, enhanced uptake of 2-NBDG in NIT-1 cells (Fig. 7D). These findings suggest that NF-Y might be involved in the transcriptional processed of Glut2.

NF-Y often binds to CCAAT motif in regulated gene promoters. Analysis of the proximal region of mouse Glut2 gene showed the presence of three potential CCAAT motifs (Fig. 7E, top). To characterize a functional CCAAT motif involved in regulating the Glut2 promoter activity, we generated serially deleted 5′ flanking regions (from −1,380, −624, and −72 to +114) in the pGL3 luciferase vector and examined their responsiveness to the forced overexpression of Nf-y in the 293T cell line. The promoter activity was comparable between the region from −1,380 to +114 and the region from −624 to +114, although it was significantly higher than the baseline levels for pGL3-Basic vector transfection. Deletion of the construct to position −72, however, resulted in a dramatic elevation in the luciferase activity (Fig. 7E, middle). These results suggest that the CCAAT motif in region −72 to +114 is essential to Glut2 promoter activity, while the motif in region −1,380 to −72 has an inhibitory effect on Glut2 promoter activity. This was further confirmed by PCR site-directed mutagenesis assay; the luciferase expression was significantly increased in M1 and M2 mutant constructs compared to their original constructs (set as of 1), respectively, while the luciferase expression was significantly decreased in the M3 mutant construct compared to the original −72 to +114 construct (Fig. 7E, bottom).

Mitochondrial activation and upregulated ATP synthesis are key steps in insulin secretion (26,27). The ATP content was comparable between Nf-ya βKO and control islets in low glucose (2.8 mmol/L) conditions (Fig. 8A). However, Nf-ya βKO islets exhibited significantly lower ATP content relative to control islets in high glucose (16.7 mmol/L) conditions (Fig. 8A), indicating an absent ATP synthesis in response to glucose stimulation in the mitochondria of Nf-ya βKO pancreatic β-cells. Since ATP synthesis is driven by the glucose-stimulated changes in mitochondrial membrane potential (MMP), we then examined MMP and found that Nf-ya βKO islets showed significantly lower glucose-induced potential than Nf-ya^fl/fl islets (Fig. 8B and C). Given that an increase of uncoupling protein 2 (UCP2) in β-cells results in impairment of glucose-induced MMP and GSIS (28), we examined the UCP2 contents and mRNA expression in isolated islets. Unexpectedly, UCP2 expression levels were significantly lower in Nf-ya βKO islets compared with Nf-ya^fl/fl islets (Fig. 8D and E). These results indicate that the impaired GSIS in Nf-ya βKO islets was not attributed to the mitochondrial uncoupling. The expression of mRNAs encoding mitochondrial respiratory complex components was significantly decreased in Nf-ya βKO islets relative to wild-type islets (Fig. 8F), suggesting a mitochondrial dysfunction in Nf-ya βKO pancreatic β-cells. Indeed, ultrastructural analysis by TEM revealed that mitochondrial swelling was evident in most Nf-ya–deficient β-cells (Fig. 8G). In addition, the mitochondrial copy number (Fig. 8H) and the mRNA expression of Pgc1a (Fig. 8E) relevant to mitochondrial biogenesis were upregulated in Nf-ya βKO islets, possibly as a compensatory response to mitochondrial dysfunction. Overall, these data suggest that NF-Y is essential for the maintenance of mitochondrial integrity in β-cells.

NF-Y plays a critical role of in multiple cellular processes (7) and has been related to many human diseases (9,29,30). In this study, we provided evidence that β-cell–derived NF-Y contributes to glucose homeostasis by coordinating islet architecture and insulin secretion from islets. First, the NF-Y protein levels in the islets were changed in in vitro and in vivo diabetic conditions, suggesting that NF-Y is a regulator of β-cell homeostasis under pathophysiological conditions. Second, mice with β-cell–specific Nf-y deletion have reduced β-cell mass and exhibit a defect in GSIS, which together result in impaired glucose tolerance. Third, the reduction in β-cell mass appears to be due to the loss of β-cell maturity and improperly organized cytoskeleton-induced downregulation of proliferation, whereas insufficient insulin secretion results from a concomitant deceased Glut2 expression and mitochondrial dysfunction. However, the actin cytoskeleton has been linked to roles in GSIS (31,32); therefore, cytoskeleton perturbation by Nf-ya deletion may also contribute to the compromised insulin secretion in response to glucose.

The maintenance of adult β-cell mass is critical for the appropriate regulation of glucose homeostasis. The ability of β-cell replication in response to cell cycle regulators is essential for adult β-cell expansion (33,34). We observed that the β-cell area was markedly decreased and the total insulin content of the whole pancreas was also diminished in the adult Nf-ya βKO mice. Moreover, the reduction of the β-cell area and pancreatic insulin content was even observed in the 2-week-old Nf-ya βKO mice. Consistent with these findings, islets from the Nf-ya βKO mice showed decreased gene expression of Pdx1, Beta2, NeuroD, and Mafa, which are well known to encode for key transcription factors of β-cell maturation and insulin transcription (35). Intriguingly, our current work showed that the onset of reduced β-cell proliferation occurred in postnatal Nf-ya βKO mice beginning with food intake, and the nutrient sensor GLUT2 was expressed at significantly lower levels in Nf-ya–deleted β-cells, suggesting that NF-Y could enable β-cells to respond to nutrient-dependent inducers of β-cell proliferation. Supporting this notion, glucose is an important stimulus of β-cell proliferation (36), and similar to Nf-ya βKO mice, GLUT2-deficient mice exhibit reduced β-cell proliferation during the early postnatal period (37). Collectively, our results suggest that the NF-Y–mediated pathways in the pancreatic β-cells contribute to glucose homeostasis through regulation of the β-cell mass during development.

The cytoskeleton not only defines the cell shape and structural integrity, but also acts as an essential component of various aspects of cell physiology. The actin cytoskeleton has been shown to be involved in the regulation of cell proliferation, and perturbation of proper cytoskeleton reorganization leads to cell cycle arrest (38). Our proteomics study identified that proteins that decreased in abundance in Nf-ya βKO islets are associated with the cytoskeleton perturbation, which may be responsible for the impaired β-cells proliferation in the Nf-ya βKO mice. However, it is important to note the limitations of protein quantification in islets. The ratio of β-cells/α-cells was significantly decreased in Nf-ya βKO islets, raising the possibility that a greater amount on α-cell proteins and less β-cell proteins are represented in the Nf-ya βKO islets. Analysis on a per-cell basis would be the most informative measurement. Additionally, the function of NF-Y in cell proliferation is often related to tumor inhibitor p53, which suppresses the expressions of cell cycle genes (39). It is likely that p53 participates in β-cell cycle control and consequently contributes to glucose homeostasis. As such, NF-Y deficiency in β-cells disrupts its interaction with p53, which in turn results in the inactivation of target genes involved in cell proliferation, contributing to the observed reduction of cyclin B1, cyclin D1, and cyclin D2 in Nf-ya βKO islets. Further study is needed to clarify the possibility of the interaction between NF-Y and p53 in β-cells.

Glucose transports into β-cells through GLUT2, followed by glucose metabolism and generation of ATP, resulting in the activation of voltage-dependent Ca²⁺ channels and Ca²⁺ influx that in turn triggers the release of insulin granules. Nf-ya βKO islets showed some, but significantly compromised, glucose-stimulated Ca²⁺ influx in comparison with control islets. Unexpectedly, Nf-ya βKO mice did not show GSIS. This dissociation of insulin release and Ca²⁺ signaling may reflect a compensatory mechanism to overcome insufficient insulin secretion, which may explain the eventual suppression of blood glucose in Nf-ya βKO mice 2 h post–glucose challenge. In contrast to the impaired Ca²⁺-dependent insulin exocytosis response to glucose challenge, Nf-ya βKO islets displayed a normal insulin secretory response to KCl, which activates voltage-dependent Ca²⁺ channels by bypassing mitochondrial ATP synthesis. Thus, the impaired Ca²⁺ response to glucose in Nf-ya βKO islets involves events upstream of K_ATP channels.

NF-Y is an evolutionally conserved CCAAT box-binding transcriptional factor. In this study, we show that NF-Y transcriptionally regulates the expression of Glut2, which is decreased in islets from the Nf-ya βKO mice. Three CCAAT motifs are localized in the mouse Glut2 promoter. Surprisingly, NF-Y regulated the Glut2 promoter activities in opposite directions. The CCAAT box positioned at −72 to +114 bp activates the Glut2 promoter activity by positively responding to NF-Y, while the motif contained in the region ranging from −1,380 to −72 bp suppresses Glut2 promoter activity. Indeed, the bidirectional effects of NF-Y on gene expression are also revealed in von Willebrand factor (40). NF-Y can induce and repress the transcription of von Willebrand factor by binding to the CCAAT motifs positioned at −18 to −14 bp and the first exon, respectively. The exact mechanism underlying the bidirectional regulation of gene promoter activity by NF-Y is worthy of further investigation.

Physiologically, GLUT2 has a primary function in the control of GSIS, and its inefficiency causes reduction in insulin secretion. However, evidence also suggests that GSIS can proceed normally even in the presence of low levels of GLUT2 (23,41). Therefore, the moderate reduction of GLUT2 level observed in Nf-ya βKO islets may not be enough to explain the impaired glucose tolerance. UCP2 has been implicated in β-cell dysfunction, and its overexpression is associated with impaired GSIS (28), while its deletion results in improvement in GSIS (42). In the current study, we show that the expression of UCP2 is significantly decreased in Nf-ya βKO islets; thus, it is unlikely that the impaired GSIS of these mice is caused by the mitochondrial uncoupling. However, mitochondrial morphology and MMP were affected in Nf-ya βKO islets. MMP, as an essential component of the proton motive force, is essential for ATP production and essential for metabolic coupling factors including reactive oxygen species, which itself is essential for GSIS (43). Furthermore, the expression of genes encoding mitochondrial respiratory chain complex components was also decreased in Nf-ya βKO islets, indicating that mitochondrial respiration complexes in β-cells are functionally damaged. Additionally, Nf-y deletion in β-cells leads to decreased cortical actin density, indicating a cytoskeleton perturbation, which may directly attenuate insulin exocytosis in response to glucose (31,32). Therefore, the combined dysfunction of GLUT2, reduced mitochondrial function, and impaired cytoskeleton might have contributed to the significant impairment of insulin secretion of Nf-ya βKO mice.

As a whole, our study, for the first time, demonstrates that NF-Y is required for the maintenance of both pancreatic β-cell mass and function in vivo, such that in its absence, glucose homeostasis is perturbed. We show that deletion of NF-Y in β-cells results in cell immaturity and impaired cell proliferation and function. The absence of NF-Y leads to not only a diminished β-cell mass due to a loss of maturity and proliferation, but also a reduction in ATP production resulting from the disruption of mitochondrial integrity, which then results in impaired insulin secretion and glucose tolerance. This study reveals a novel dual role of NF-Y in integrating β-cell maturation, proliferation, and cell function, and this could be applicable to the processes of T2DM.

Funding. This study was supported by the National Natural Science Foundation of China (81770814), the Sichuan Province Science and Technology Support Program (2020YF0192), the National Clinical Research Center for Geriatrics, West China Hospital, Sichuan University (Z20201010), and the Graduate Student’s Research and Innovation Fund of Sichuan University (2018YJSY118).

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

Author Contributions. Y.L., S.H., R.Z., X.Z., S.Y., and D.D. researched data. C.Z., X.Y., and Y.C. contributed to providing reagents. Z.S. wrote the manuscript. Z.S. is the guarantor of this work and, as such, had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.