SREBP1c-PAX4 Axis Mediates Pancreatic β-Cell Compensatory Responses Upon Metabolic Stress

Pancreatic islets regulate glucose homeostasis through the secretion of several hormones, particularly glucagon and insulin. Under hypoglycemic conditions, glucagon is released from pancreatic α-cells to promote hepatic glycogenolysis and gluconeogenesis to increase blood glucose levels (1). In contrast, insulin secretion from pancreatic β-cells is stimulated by hyperglycemia. Insulin effectively lowers the level of blood glucose via glucose uptake into peripheral tissues and hinders hepatic glucose production. On the other hand, metabolic stresses such as obesity and insulin resistance increase insulin demand and disrupt glucose homeostasis (2). In response to chronic fuel surfeit or hyperglycemia in obesity, pancreatic islets adaptively elevate the level of serum insulin through a β-cell compensatory mechanism. This process includes the expansion of β-cell mass and the augmentation of insulin production (3,4). Although β-cell neogenesis and protection from apoptosis can contribute to β-cell mass expansion, accumulating evidence suggests that an increase in β-cell proliferation would be the principal mechanism of postnatal β-cell growth (5,6). Thus, the potentiation of β-cell proliferation is important to maintain glucose homeostasis and to protect against the onset of diabetes.

SREBP1c is a basic helix-loop-helix leucine zipper transcription factor that governs de novo lipogenesis (7–10). During the postprandial state, insulin activates SREBP1c via AKT (protein kinase B) and mTORC1 (mammalian target of rapamycin C1) (11,12). SREBP1c stimulates de novo lipogenesis by elevating the expression of lipogenic target genes including fatty acid synthase, stearoyl-CoA desaturase 1, and acetyl-CoA carboxylase (12–14). Moreover, SREBP1c has been implicated in lipid metabolisms of pancreatic islets (15,16). For instance, it has been reported that the activation of SREBP1c by liver X receptor agonist increases intracellular lipid accumulation in pancreatic islets (16). SREBP1 is associated with cell growth through lipid metabolism and/or cell cycle progression in certain cell types (17,18). However, it remains unclear whether SREBP1c might be associated with pancreatic β-cell proliferation.

Paired box 4 (PAX4) belongs to the PAX gene family, a group of transcription factors that carry out essential roles in embryogenesis as well as in cellular plasticity in adults (19). PAX4 is mainly expressed in endocrine pancreas where it plays an essential role to induce differentiation toward β-cells (20). Also, PAX4 has been implicated in β-cell plasticity of adult pancreatic islets. For example, it has been revealed that ectopic expression of PAX4 in human or murine islets enhances β-cell proliferation by stimulating the transcription of cell cycle genes (21–23). Moreover, the overexpression of PAX4 protects β-cells from apoptosis induced by streptozotocin (STZ) (24). Recently, it has been reported that mutations of the PAX4 gene are associated with type 1 and 2 diabetes as well as ketosis-prone diabetes in various ethnic groups (25). Although the correlation between PAX4 dysfunction and diabetes has been demonstrated, the regulatory mechanism of PAX4 expression remains elusive.

In this study, we investigated SREBP1c^+/+ and SREBP1c^−/− mice under normal chow diet (NCD) or a high-fat diet (HFD) to understand the roles of SREBP1c in pancreatic islets. We demonstrate that pancreatic SREBP1c plays an important role in β-cell compensatory responses upon metabolic stress, accompanied with regulating β-cell proliferation and survival. In pancreatic islets, we identified that PAX4 could act as a downstream mediator of SREBP1c to control β-cell growth. In addition, transplantation of SREBP1c-overexpressing primary islets ameliorated glucose intolerance in diabetic animals, implying that elevated expression of SREBP1c would potentiate β-cell function. Collectively, our data suggest that the SREBP1c-PAX4 axis would play a pivotal role in compensatory responses of pancreatic islets under metabolic stress.

SREBP1c^−/− mice were provided by Dr. Jay D. Horton at the University of Texas Southwestern Medical Center (Dallas, TX). db/+ and db/db male mice were obtained from Daehan Bio (Seoul, Korea). SREBP1c^−/− and SREBP1^+/+ littermates were maintained on a NCD (Zeigler) for 8 weeks. They were fed with NCD or 60% HFD (Research Diets Inc.) for 12 weeks. For fasted/refed experiments, blood samples were collected after fasting for 16 h and refeeding for 2 h. Mouse tissue specimens were immediately stored at −80°C. For the glucose tolerance test (GTT), mice were fasted for 16 h and intraperitoneally injected with glucose (1 g ⋅ kg⁻¹ body weight for mice). Blood glucose levels were measured in tail vein blood samples by using a Contour TS Blood Glucose Meter (Bayer). For the insulin tolerance test (ITT), ad libitum mice were intraperitoneally injected with insulin (1 unit/kg body weight for mice). All experiments with mice were approved by the Seoul National University Institutional Animal Care and Use Committee.

Cell proliferation rates were assayed using a Cell Counting Kit-8 reagent. Briefly, cell growth curves were generated using the sensitive colorimetric assay for viable cells according to the manufacturer protocol (#CK04-11; Dojindo Molecular Technologies).

Total RNAs were isolated from MIN6 and αTC1–6 cells using TRIzol reagent (Invitrogen). Islet RNAs were prepared using the RNeasy Mini Kit (Qiagen). Subsequently, equal amounts of RNA were synthesized to cDNA using RevertAid reverse transcriptase (#EP0441; Thermo Fisher Scientific). Relative amounts of mRNA were calculated by using a CFX Real-Time Quantitative PCR Detection System (Bio-Rad) after normalization to cyclophilin mRNA. The primer sequences used are listed in Supplementary Table 1.

Adenoviral plasmid were constructed as previously described (26). cDNAs for SREBP1c and PAX4 were incorporated into an AdTrack-CMV shuttle vector and an Ad-Easy vector. Adenoviruses were amplified in human embryonic kidney 293A (HEK293A) cells and isolated by CsCl density centrifugation. Green fluorescent protein was coexpressed from an independent promoter with inserted cDNA. Empty virus expressing only the gene for green fluorescent protein served as control (MOCK).

MIN6 cells were cultured in DMEM medium containing 15% FBS (HyClone). HEK293T, αTC1–6 cells were cultured in DMEM containing 10% FBS. Harvested primary islets were cultured in RPMI 1640 medium (HyClone) with 10% FBS. Two kinds of siRNAs were designed and produced by Bioneer (Daejeon, South Korea). The antisense siPAX4 sequences were as follows: antisense, UGGUACUCCUCACAGAAGG and ACUGUCAAAUAGAGGCCUC. siRNA was transfected using lipofectamine-iMAX (Thermo Fisher Scientific) into cell lines.

The pancreata were isolated from mice, fixed in 4% paraformaldehyde, and embedded in paraffin block. Paraffin blocks were cut into 5-μm-thick sections with 30-μm intervals. Pancreas sections from NCD- or HFD-fed mice and db/db mice were stained with anti-SREBP1 (SC-367; Santa Cruz Biotechnology), anti-insulin (ab7842; Abcam), anti-pax4 (ab42450; Abcam), anti-glucagon (ab10983; Abcam), and anti-Ki67-FITC (11-5698-80; eBioscience) antibodies. Species-specific secondary antibodies staining or diaminobenzidine staining was followed. For the islet morphology analysis, specimens were viewed on a confocal LSM 700 System (Carl Zeiss). The areas of the images were processed and measured using ImageJ software.

Chromatin immunoprecipitation (ChIP) assays were conducted as previously described (26). Briefly, extracted proteins from total cell lysates were immunoprecipitated with anti-SREBP1 antibody (557036; BD Biosciences) or IgG (sc-2025; Santa Cruz Biotechnology) for 2 h. Precipitated DNA fragments were analyzed by quantitative real-time PCR (qRT-PCR) using primer sets that encompassed the sterol regulatory element–containing region of the mouse PAX4 gene promoter and non–sterol regulatory element negative control region. The primer sequences for the ChIP assay were as follows: sense, 5′-GTATAATTGTGAGCAGATGGCG-3′ and antisense, 5′-GGCCTAGCAAGCCCAAA-3′; negative control primer: sense, 5′-TTCCAGGCAAGAACTCACCT-3′ and antisense, 5′-TATCGTTTCCCAGCCATCA-3′.

The pancreatic islets were isolated from male mice using the intraductal collagenase technique. The pancreas was dissected from the surrounding tissues and digested in a shaking bath for 12 min at 37°C with collagenase P (Roche). The islets were purified using Ficoll gradient solutions (29%, 24%, and 15% [w/v] in Hanks’ balanced salt solution). After centrifugation for 20 min, the islets were collected and washed with cold Hanks’ balanced salt solution. The isolated islets were cultured in RPMI 1460 medium supplemented with 10% FBS (Gibco). For ex vivo experiments or transplantation, primary islets were incubated with adenovirus (40 multiplicity of infection) for 12 h in RPMI 1640 serum-free medium and replaced to RPMI 1640 medium with 10% FBS media for 2 days. For static glucose-stimulated insulin secretion assays with mouse islets, ∼20 islets were hand picked, incubated for 2 h in Krebs-Ringer bicarbonate buffer at 37°C, 5% CO₂, and then incubated for 60 min in 2.8 mmol/L glucose in the presence or absence of 30 mmol/L KCl or 22.2 mmol/L glucose. Secreted insulin was measured using an insulin ELISA kit (Morinaga Ultra Sensitive Mouse Insulin ELISA kit; Morinaga Institute of Biological Science), and normalized to the total amounts of protein. Total insulin contents of the pancreas were extracted with acidic ethanol (0.2 mol/L HCl in 75% ethanol) and measured with an insulin ELISA kit.

Transplantation of primary islets was performed as previously described (27). Mice were intraperitoneally injected 150 mg/kg STZ. After 3 days of injections, primary islets were transplanted into recipient mice showing hyperglycemia >350 mg/dL. For islet transplantation, ∼200 freshly isolated islets were aspirated into a 200-μL pipette tip, and the pipette tip was connected to a silicon tube (internal diameter × outer diameter = 0.58 × 0.965 mm; Becton Dickinson). Under anesthesia, the right kidney of the recipient mouse was exposed through an incision and primary islets were injected under the outer surface of the kidney. The tube was removed, and the capsulotomy was cauterized. After transplantation, the insulin level was analyzed (fasting [16 h]/refeeding [2 h]), and a GTT was performed (fasting 16 h).

Apoptosis rates were determined using an in situ cell death detection kit (12156792910; Roche) according to the manufacturer protocol. Fluorescein-labeling images were taken by a fluorescence microscope (Olympus) using an excitation wavelength of 450–500 nm and a detection wavelength of 515–565 nm. Peroxidase-labeling images were monitored under a microscope, and statistical analyses of the obtained images were performed using LSM510 software (Carl Zeiss).

Trypsinized pancreatic islets were washed with PBS and fixed and permeabilized using Fixation/Permeabilization Concentrate (eBioscience) for 30 min. Fixed cells were incubated with propidium iodide solution containing 0.1% Nonidet P-40, 100 μg/mL RNase, and 2.5 μg/mL propidium iodide for 30 min. Stained cells were analyzed and quantified for each stage by flow cytometry using a FACSCanto II System (BD Biosciences).

Sample sizes were chosen based on pilot experiments that ensured adequate statistical power with similar variances. Multiple comparisons were performed by one-way ANOVA, or by two-way ANOVA when two conditions were involved. Statistical significance was assessed by the Student t test and are presented as the mean ± SD determined from at least three independent experiments. Values of P < 0.05 were considered to be statistically significant differences. All n values defined in the legends refer to biological replicates, unless otherwise indicated. If technical failures such as the failure of intraperitoneal injection occurred, these samples were excluded from the final analysis.

It has been well established that SREBP1c is closely associated with metabolic disorders, including obesity and insulin resistance (28,29). However, it is not thoroughly understood whether SREBP1c might be involved in insulin metabolism in obese pancreatic islets. To address this, we investigated the levels of SREBP1c protein and mRNA in pancreatic islets. In lean mice, SREBP1c protein was abundantly expressed in the nuclei of insulin-positive β-cells (Fig. 1A). In pancreatic islets, the levels of SREBP1c protein and mRNA were higher in HFD-fed obese mice than in NCD-fed lean mice (Fig. 1A–C). Similarly, pancreatic expression of SREBP1c was upregulated in db/db mice compared with nondiabetic db/+ mice (Fig. 1D–F). In contrast, SREBP1a, another SREBP1 isoform, was not significantly elevated in pancreatic islets of HFD-fed obese mice and db/db mice (Supplementary Fig. 1A and B). These results suggest that SREBP1c expression would be upregulated in obese and diabetic pancreatic islets.

Given that pancreatic β-cell dysregulation leads to impaired glucose homeostasis, we raised the question of whether SREBP1c deficiency may alter glucose and/or insulin metabolism. To answer this, we investigated metabolic phenotypes of SREBP1c-deficient mice upon being fed a NCD or HFD. Both SREBP1c^+/+ and SREBP1c^−/− mice showed similar body weight gains with NCD or HFD feeding (Fig. 2A). Nonetheless, NCD- or HFD-fed SREBP1c^−/− mice showed decreased serum insulin and increased blood glucose under the refed condition (Fig. 2B and C). Because refed SREBP1c^−/− mice exhibited reduced levels of serum insulin, we hypothesized that SREBP1c deficiency might affect systemic glucose homeostasis. To test this, SREBP1c^+/+ and SREBP1c^−/− mice were subjected to a GTT and an ITT. Upon NCD or HFD feeding, SREBP1c^−/− mice were shown to be glucose intolerant compared with SREBP1c^+/+ mice (Fig. 2D and E). However, NCD-fed SREBP1c^−/− mice did not exhibit significant differences in ITT (Fig. 2F), implying that decreased levels of postprandial insulin in SREBP1c-deficient mice might be uncoupled to insulin sensitivity in lean animals. Together, these findings suggest that SREBP1c deficiency might alter the function of pancreatic islets, leading to glucose intolerance.

Reduced β-cell mass has been considered to be one of the major characteristics of diabetic pancreatic islets (30). To determine whether impaired glucose tolerance in SREBP1c^−/− mice might result from decreased β-cell mass, we histologically analyzed pancreatic islets. In the embryo stage, there was no significant difference in the development or distribution of insulin-positive cells from SREBP1c^+/+ and SREBP1c^−/− mice (Supplementary Fig. 2A and B), implying that SREBP1c deficiency might have little, if any, effect on β-cell differentiation during embryonic development. In adult stage, the insulin-positive β-cell area in SREBP1c^−/− mice was smaller than that in SREBP1c^+/+ mice (Fig. 3A and B). Although the insulin-positive β-cell area was expanded by HFD feeding, the extent of increase was much less in HFD-fed SREBP1c^−/− mice. In accordance with these data, total insulin contents in the pancreata of SREBP1c^−/− mice were lower than those in the pancreata of SREBP1c^+/+ mice (Fig. 3C). In addition, the numbers of pancreatic islet clusters were not different between SREBP1c^+/+ and SREBP1c^−/− mice (Fig. 3D). These data imply that a decrease in β-cell numbers in islet clusters might confer low insulin contents in the pancreata of SREBP1c^−/− mice.

Because β-cell proliferation and apoptosis are crucial factors to decide β-cell mass (6,31), we next investigated what factors may be involved in the reduction of β-cell mass in SREBP1c^−/− mice. As shown in Fig. 3E and F, the number of Ki67-positive β-cells, which represent proliferating β-cells, was lower in pancreatic islets from NCD- and HFD-fed SREBP1c^−/− mice compared with SREBP1c^+/+ mice. Moreover, mRNA levels of cell cycle–related genes were declined in pancreatic islets from SREBP1c^−/− mice (Supplementary Fig. 3), indicating that SREBP1c deficiency might lead to deterioration of β-cell proliferation. On the other hand, the number of TUNEL-positive apoptotic β-cells in SREBP1c^−/− mice was higher than that of SREBP1c^+/+ mice (Fig. 3G and H). Together, these results propose that SREBP1c deficiency would decrease β-cell mass, accompanied with reduced β-cell proliferation and increased β-cell apoptosis.

Since SREBP1c is a transcriptional activator, we hypothesized that SREBP1c could modulate β-cell proliferation by regulating β-cell–specific target genes. To identify SREBP1c target genes in β-cells, we scrutinized gene expression profiles in SREBP1c-overexpressing β-cell lines and primary islets (Supplementary Fig. 4A). Among several candidates, we found that PAX4 might be a potential target gene of SREBP1c. Previously, it has been reported that PAX4 is selectively expressed in pancreatic endocrine cells and plays critical roles in the proliferation and survival of β-cells (21). As shown in Fig. 4A and B, the level of PAX4 protein was reduced in pancreatic islets of NCD- or HFD-fed SREBP1c^−/− mice. Consistently, the level of PAX4 mRNA was decreased in pancreatic islets from SREBP1c^−/− mice compared with SREBP1c^+/+ mice (Fig. 4C). As the proximal promoter regions of the PAX4 gene contain putative binding sites for SREBP1c (Supplementary Fig. 4B), we decided to test the idea whether SREBP1c might upregulate the promoter activity of the PAX4 gene. As shown in Fig. 4D, ectopic expression of SREBP1c transactivated mouse PAX4 promoter activity in a luciferase reporter assay. Accordingly, ChIP assays with SREBP1 antibody revealed physical bindings of SREBP1 to the promoter of mouse PAX4 gene (Fig. 4E). Moreover, the level of PAX4 mRNA was increased by SREBP1c overexpression in β-cells (Fig. 4F), indicating that PAX4 would be a novel target gene of SREBP1c in pancreatic islets.

It has been reported that pancreatic PAX4 plays profound roles in β-cell proliferation and insulin secretion (21,22). To examine whether PAX4 could modulate β-cell proliferation, we analyzed gene expression with PAX4 modulation using siRNA or adenovirus. Consistent with previous reports (21,22), suppression of PAX4 via siRNA decreased the growth rate of β-cells and reduced the expression of cell cycle genes (Supplementary Fig. 5A–C). In contrast, PAX4 overexpression increased β-cell proliferation and elevated the expression of cell cycle genes (Supplementary Fig. 5D and E), indicating that PAX4 expression has a positive correlation with β-cell proliferation. Next, to investigate whether PAX4 would indeed act as a downstream mediator of SREBP1c in β-cell proliferation and insulin secretion, we examined the effects of PAX4 suppression in SREBP1c-overexpressing MIN6 β-cells. Although ectopic expression of SREBP1c in pancreatic β-cells upregulated expression of cell cycle genes and secretion-related genes, PAX4 suppression attenuated the effects of SREBP1c on gene expression (Fig. 5A). Consistently, PAX4 knockdown downregulated the rate of SREBP1c-induced β-cell proliferation (Fig. 5B). Moreover, SREBP1c overexpression elevated insulin secretion upon high glucose stimuli, whereas PAX4 suppression diminished this increase (Fig. 5C), indicating that PAX4 would be a downstream player of SREBP1c-mediated β-cell proliferation and secretory function. In accordance with these data, PAX4 overexpression rescued mRNA expression of cell cycle genes and secretion-related genes whose expression was downregulated in pancreatic islets of SREBP1c^−/− mice (Fig. 5D). Taken together, these data evidently suggest that the SREBP1c-PAX4 axis could promote β-cell proliferation and insulin secretion by regulating gene expression.

To test the idea that SREBP1c might modulate the physiological functions of pancreatic islets, we infected primary islets with SREBP1c adenovirus (Ad-SREBP1c). In primary islets, SREBP1c overexpression stimulated mRNA levels of cell cycle genes and secretion-related genes (Fig. 6A). Consistently, ectopic expression of SREBP1c in pancreatic islets led to an increased S and G2/M phase population (Supplementary Fig. 6). Moreover, SREBP1c-overexpressing islets exhibited augmented insulin secretion upon glucose stimulation (Fig. 6B). To further investigate whether SREBP1c would mediate functional compensation of pancreatic islets and restore systemic glucose metabolism in diabetic animals, we transplanted SREBP1c-overexpressing primary islets into the STZ-induced diabetic mice and analyzed glucose metabolism. As expected, STZ potently destroyed β-cells and led to hyperglycemia, whereas transplantation of wild-type primary islets infected with MOCK adenovirus relieved hyperglycemia in diabetic mice (Fig. 6C). More importantly, there was a stronger lowering effect on blood glucose in the SREBP1c-overexpressing islet-transplanted group (Fig. 6C). In addition, islet transplantation restored the serum insulin level, which was further potentiated by SREBP1c-overexpressing islet transplantation (Fig. 6D). Next, we investigated the effects of islet transplantation from SREBP1c^+/+ and SREBP1c^−/− mice on glucose metabolism. In GTT analysis, the SREBP1c knockout islet-transplanted group exhibited glucose intolerance compared with the SREBP1c wild-type islet-transplanted group (Fig. 6E). Moreover, the SREBP1c^−/− islet transplantation group showed decreased serum insulin levels after 30 min of glucose challenge (Fig. 6F). Collectively, these data suggest that SREBP1c would be crucial for the quantity and quality of pancreatic islets.

Dysregulated interaction between pancreatic endocrine cells and hyperglucagonemia has been recognized as another contributor to imbalanced glucose homeostasis in patients with diabetes (32,33). In addition, it has been reported that PAX4 is involved in the suppression of glucagon expression and the transdifferentiation of α-cells into β-cells (34). Thus, we decided to test whether the SREBP1c-PAX4 axis might also affect glucagon metabolism in pancreatic islets. Relative to SREBP1c^+/+ mice, SREBP1c^−/− mice exhibited higher levels of serum glucagon (Fig. 7A) and glucagon mRNA (Fig. 7B). Accordingly, the degree of glucagon-positive α-cells was higher in SREBP1c^−/− mice than in SREBP1c^+/+ mice (Fig. 7C and D). Furthermore, the overexpression of either SREBP1c or PAX4 downregulated the level of glucagon mRNA in α-cell lines (Fig. 7E), implying that the SREBP1c-PAX4 axis might be a negative regulator of glucagon expression in pancreatic α-cells.

To resolve hyperglycemia upon metabolic stress, pancreatic islets adaptively increase the level of serum insulin through β-cell compensation. In this regard, the capacity of β-cell compensation is crucial for glucose homeostasis and protection against type 2 diabetes (35). Nonetheless, the mechanisms underlying β-cell compensation are not thoroughly understood. Our data provide compelling evidence that pancreatic SREBP1c could play a key role in regulating compensatory responses in β-cells. We found that SREBP1c expression upregulated pancreatic islets in obese and diabetic animals. In addition, pancreatic SREBP1c potentiated the capacity of β-cells to proliferate and to secrete insulin upon increased insulin demands. Particularly, SREBP1c modulated various gene expressions in β-cells through a novel target gene, PAX4. On the contrary, SREBP1c deficiency hindered this compensatory process, resulting in glucose intolerance (Fig. 8). Collectively, our findings propose that the SREBP1c-PAX4 axis would play an essential role in β-cell compensation.

SREBP1c^−/− mice exhibited normal growth and displayed body weight gain similar to that of age-matched SREBP1c^+/+ mice. However, SREBP1c^−/− mice showed postprandial hyperglycemia and glucose intolerance, whereas they were not defective to insulin sensitivity. In pancreatic islets of SREBP1c^−/− mice, the levels of insulin secretion with glucose stimuli were decreased. These data raised the possibility that SREBP1c deficiency might mediate pancreatic islet dysfunction, leading to glucose intolerance. In this study, several lines of evidence suggest that SREBP1c plays key roles in regulating β-cell mass and function. First, compared with SREBP1c^+/+ mice, the β-cell area was smaller and total insulin contents was lower in the pancreata of SREBP1c^−/− mice. Second, SREBP1c expression was upregulated in pancreatic islets of obese and diabetic animals. Third, SREBP1c overexpression in β-cells stimulated mRNA levels of cell cycle genes and increased β-cell growth rate. Last, transplantation of SREBP1c-overexpressing islets ameliorated glucose intolerance in diabetic mice. These findings unravel a novel mechanism in which pancreatic SREBP1c would play an important role in β-cell compensation.

It has been reported (18) that SREBP1c is associated with lipid metabolism and cell cycle progression. For instance, SREBP1 influences cell growth through providing lipid metabolites as building blocks for cell division (17). Furthermore, SREBP1 is able to regulate expression of certain set of genes associated with cell cycle progression (18,36). To date, it remains unclear whether SREBP1c activation is involved in β-cell proliferation. Here, we demonstrate that SREBP1c augments β-cell proliferation through a novel target gene, PAX4. We observed that SREBP1c binds to the promoter of the PAX4 gene and transactivates its promoter in β-cells. Also, SREBP1c increased the expression of PAX4 mRNA and protein in pancreatic islets. Moreover, the expression levels of PAX4 and cell cycle genes were downregulated in pancreatic islets of SREBP1c^−/− mice. Conversely, PAX4 upregulation rescued the expression of cell cycle genes that were decreased in SREBP1c-deficient pancreatic islets. Given that PAX4 suppression significantly attenuated SREBP1c-induced β-cell growth, it is likely that SREBP1c could promote β-cell proliferation, at least in part, by upregulating PAX4. Together, these data imply that the SREBP1c-PAX4 axis would play a key role in β-cell proliferation.

Dysregulation of two pancreatic hormones, characterized by insulin insufficiency and hyperglucagonemia, leads to hyperglycemia and contributes to increased diabetes incidence (37). Recently, it has been reported (25,38) that single nucleotide polymorphisms and mutations of PAX4 gene are closely associated with diabetes in numerous ethnic groups. Since PAX4 has opposite roles in the development and function of pancreatic endocrine cells (39), we hypothesized that the dysregulation of the SREBP1c-PAX4 axis might be associated with an imbalance of pancreatic endocrine cells. In pancreatic islets of SREBP1c^−/− mice, we observed that the degree of TUNEL-positive apoptotic β-cells was elevated compared with SREBP1c^+/+mice. Accordingly, the number of β-cells was decreased in SREBP1c^−/− mice. To our surprise, we could also detect increased numbers of pancreatic α-cells and hyperglucagonemia in SREBP1c^−/− mice, implying that SREBP1c deficiency may influence the ratio of two primary pancreatic endocrine cells, such as α-cells and β-cells. Intriguingly, PAX4 and SREBP1c overexpression suppressed glucagon expression in pancreatic α-cells. Given that ectopic expression of PAX4 has dual roles in enhancing β-cell survival and inhibiting glucagon expression (40), it is plausible to speculate that the SREBP1c-PAX4 axis could participate in the determination of ratio and functions of pancreatic endocrine cells. In future, it needs to be elucidated whether developmental defects or indirect pathways might be attributable to SREBP1c-deficient pancreatic islets. Nonetheless, our data propose that dysregulated SREBP1c-PAX4 axis might be involved in pancreatic hormonal imbalance in obesity-induced diabetes.

It has been reported that intracellular lipid metabolites have various effects on β-cell functions and their proliferation (41–43). Here, we found that SREBP1c expression was enhanced in pancreatic islets of obese and diabetic mice. Accordingly, the level of triglycerides was upregulated in obese pancreatic islets (Supplementary Fig. 7). Previously, it has been suggested that an uncontrolled increase of SREBP1c expression could induce detrimental effects on pancreatic β-cells, eventually leading to lipotoxicity and cell death (44,45). However, our in vivo data showed that an increase of SREBP1c expression in pancreatic islets of HFD-fed animals was not sufficient to provoke lipotoxicity. Instead, we could observe that β-cell apoptosis was increased in SREBP1c-deficient islets, implying that the moderate induction of SREBP1c within physiological range might have crucial roles in pancreatic β-cells. Although we could not exclude the possibility that aberrant upregulation of SREBP1c might result in β-cell lipotoxicity under pathophysiological conditions, our in vivo data propose that an appropriate increase of SREBP1c expression in obese pancreatic islets would be required for β-cell compensation.

It has been previously proposed that deficiency of SREBP1 does not cause significant effects on islet volume or insulin contents in pancreas (44). It appears that a previous report is somewhat contradictory to our finding from SREBP1c^−/− mice. This discrepancy might result from using different animal models. Here, we analyzed SREBP1c isoform-specific knockout mice that have the SREBP1a isoform (46). Instead, Takahashi et al. (44) have examined SREBP1^−/− mice that do not have both isoforms such as SREBP1a and SREBP1c. Unlike SREBP1c^−/− mice, it has been reported that SREBP1^−/− mice exhibited developmental defects and high lethality (50–85% died in utero at embryonic day 11) (47). Moreover, surviving SREBP1^−/− mice produced a truncated form of SREBP1 protein, which might distort pancreatic phenotypes.

In conclusion, we have newly identified the regulatory mechanism of β-cell compensation through the SREBP1c-PAX4 axis. Our data suggest that pancreatic SREBP1c plays an important role in fine-tuning β-cell function upon insulin demands to confer systemic glucose homeostasis. Given that the SREBP1c-PAX4 axis is crucial to maintain functional β-cell mass, it is likely that the regulation of SREBP1c activity might be a potential target to treat metabolic disease as well as pancreatic dysfunction.

Acknowledgments. The authors thank the members of the laboratory of adipocyte and metabolism research for helpful discussion. The authors also thank Dr. Jay Horton at the University of Texas Southwestern Medical Center for providing SREBP1c^−/− mice.

Funding. This work was supported by the National Creative Research Initiative Program of the National Research Foundation funded by the Korea government (Ministry of Science, ICT and Future Planning, 2011-0018312). G.L., Y.Y.K., and J.P. were supported by the BK21 program.

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

Author Contributions. G.L. executed most experiments, designed the project, prepared the manuscript, contributed to performing animal experiments, and discussed the data. H.J. designed the project. Y.Y.K. prepared the manuscript, contributed to performing animal experiments, and discussed the data. S.S.C., J.K., and S.-S.I. prepared the manuscript. I.H. and J.P. contributed to performing animal experiments and discussed the data. J.B.K. designed the project and prepared the manuscript. J.B.K. 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.