Conditional ablation of Gsk-3β in islet beta cells results in expanded mass and resistance to fat feeding-induced diabetes in mice
- First Online:
- 1k Downloads
Glycogen synthase kinase 3β (GSK-3β) is an enzyme that is suppressed by insulin and when elevated results in insulin resistance in skeletal muscle and diabetes. Its role in beta cell development and function is little known. Because of the enzyme’s anti-proliferative and pro-apoptotic properties, the hypothesis to be tested here was that beta cell specific deficiency of GSK-3β in mice would result in enhanced beta cell mass and function.
Mice with beta cell deficiency of GSK-3β (β-Gsk-3β [also known as Gsk3b]−/−) were generated by breeding Gsk-3βflox/flox mice with mice overexpressing the Cre recombinase gene under the control of the rat insulin 2 gene promoter (RIP-Cre mice), and glucose tolerance, insulin secretion, islet mass, proliferation and apoptosis were measured. Changes in islet proteins were investigated by western blotting.
On a normal diet β-Gsk-3β–/– mice were found to have mild improvement of glucose tolerance and glucose-induced insulin secretion, and increased beta cell mass accompanied by increased proliferation and decreased apoptosis. On a high-fat diet β-Gsk-3β–/– mice exhibited improved glucose tolerance and expanded beta cell mass with increased proliferation relative to that in control mice, resisting fat-fed diabetes. Molecular mechanisms accounting for these phenotypic changes included increased levels of islet IRS1 and IRS2 proteins and phospho-Akt, suggesting enhanced signalling through the phosphatidylinositol 3-kinase (PI3K)/Akt pathway, and increased islet levels of pancreas/duodenum homeobox protein 1 (PDX1). Inhibition of GSK3 in MIN6 cells in vitro led to increased IRS1 and IRS2 protein levels through inhibition of proteosomal degradation.
These results are consistent with a mechanism whereby endogenous GSK-3β activity controls islet beta cell growth by feedback inhibition of the insulin receptor/PI3K/Akt signalling pathway.
KeywordsGlycogen synthase kinase 3β Islet beta cells Islet mass IRS1 IRS2 PDX1 Proliferation
Forkhead box O
Glycogen synthase kinase 3β
Intraperitoneal glucose tolerance test
Pancreas/duodenum homeobox protein 1
Terminal deoxynucleotidyl transferase-mediated dUTP-biotin nick-end labelling
The molecular mechanisms enabling beta cells to adapt to insulin resistance are being discovered primarily in animal models. A number of important genetic models have focused on the requirement for insulin signalling through beta cell insulin/IGF1 receptors, as reviewed in [1, 2]. In these models the primacy of phosphatidylinositol 3-kinase (PI3K)/Akt activity in expansion and postnatal maintenance of beta cell mass is apparent, as reviewed in . The remarkable ability of beta cell mass to expand via enhanced proliferation and reduced apoptosis in transgenic mice producing constitutively active Akt in beta cells illustrates the potential importance of this pathway for expanding beta cells [4, 5]. Knowing that increased production of Akt in beta cells leads to marked expansion, our studies have focused on the role of an important negatively regulated Akt substrate, glycogen synthase kinase 3 (GSK3).
GSK3 was originally identified as a serine/threonine kinase that inactivates glycogen synthase . This kinase is distinguished by the property of being constitutively active. Early studies showed that insulin inhibits GSK3 activity through PI3K/Akt-induced phosphorylation, promoting glycogen synthesis and glucose disposal. Later, many cellular processes, including transcription, translation, cell cycle regulation and apoptosis, were shown to be affected by GSK3 . The adverse consequences of GSK3 activity have been illustrated by its overproduction in peripheral tissues in diabetic animals and patients [7, 8, 9], and by reversal of diabetes with GSK3 inhibitors . While previous studies illustrated the undesirable consequences of overexpression of Gsk-3β (also known as Gsk3b) in peripheral tissues and islet beta cells, little is known about the consequences of reduced expression in islet beta cells. A recent in vitro study of GSK-3β activity in isolated human and rat islets showed that incubation with various chemical inhibitors resulted in enhanced DNA synthesis and cell cycle progression . These in vitro studies suggested that this enzyme might be rate-limiting for islet beta cell growth in the animal in vivo. The hypothesis to be tested here was that deficiency of endogenous GSK-3β would result in expansion of beta cell mass through enhanced proliferation. The present experiments used a conditional Gsk-3β mouse (Gsk-3βflox/flox) to create mice with beta cell-specific GSK-3β deficiency (β-Gsk-3β–/–). The results of these studies now confirm that GSK-3β is rate-limiting for islet beta cell mass in vivo, and suggest a mechanism involving feedback inhibition of the insulin receptor/Akt-signalling pathway.
Animal production and phenotypic analysis
The generation of mice expressing a conditional allele of Gsk-3β was described previously . Pancreatic beta cell-specific Gsk-3β knockout mice (β-Gsk3b–/–) were generated by breeding Gsk-3β floxed mice with mice that expressed the Cre recombinase gene under the control of the promoter of the rat insulin 2 gene (RIP-Cre mice) .The primers used for genotyping are listed in the Electronic supplementary material (ESM Table 1). For all studies, RIP-Cre mice were used as a control rather than wild-type mice because RIP-Cre mice have been shown to have glucose intolerance relative to wild-type mice . The mice were fed a normal rodent diet (Picolab Rodent Diet 20, catalogue no. 5053; Purina Mills, St Louis, MO, USA) or a synthetic high-fat diet (HFD) (Teklad Adjusted Calories Diet; 55/fat, catalogue no. TD.93075; Harlan Laboratories, Indianapolis, IN, USA) and were housed in colony cages and maintained on a 12 h light–dark cycle. The composition of the standard diet was 20% protein, 4.5% fat. The composition of the HFD diet was 21.2% protein, 54.8% fat from vegetable shortening.
Blood glucose and serum insulin concentrations were determined as previously described . This project was approved by the Animal Ethics Committee of Washington University School of Medicine.
Immunohistochemical and morphometric analysis of pancreatic islets
Pancreases were isolated from 3-month-old RIP-Cre control and β-Gsk-3β−/− mice and fixed. Immunohistological and morphometric analyses were as described previously . Antibodies used in this study are listed in the Electronic supplementary material (ESM) Table 1. The mass of beta cells was quantified using NIH Image 1.39j software . Islet beta cell proliferation and apoptosis were assessed as described previously . Quantitative data were obtained from at least three mice in each group unless indicated otherwise.
Assays of islet insulin secretion, insulin content and DNA content
Glucose-stimulated insulin secretion and insulin content were determined in islets isolated by collagenase distension/disgestion of the pancreas, as described previously .
Western blotting analysis and cell culture
Islets were isolated as described previously  and immunoblots were analysed as described previously . Extracts (20 μg total protein) were subjected to immunoblot analysis with antibodies (for specific antibodies see ESM Table 1). Immune complexes were revealed using the ECL Advance Western Blot Detection Kit (Amersham, Little Chalfont, UK) and images were acquired using a FluoroChemi 8800 digital camera acquisition system (Alpha Innotech, San Leandro, CA, USA). Band intensities in the blots were later quantified using ImageJ 1.39j  and α-tubulin or β-actin bands were used to adjust for loading differences.
MIN6 cells were maintained in DMEM . Cells with passage numbers <35 were used. For immunoblot analysis, we prepared lysates of MIN6 cells treated with either 20 mmol/l LiCl or 20 mmol/l NaCl for 30 min, then cotreated them with 50 μg/ml cycloheximide and 5 μmol/l MG132 (Calbiochem, San Diego, CA, USA) or cotreated them with DMSO and either 0.5% BSA or 400 μmol/l palmitate+0.5% BSA at the indicated time points. The lysates were probed with the antibodies listed above. Protein was extracted with a cell lysis buffer . Protein samples (20 μg) were subjected to immunoblot analysis with antibodies to IRS1 and IRS2 (Upstate, Lake Placid, NY, USA).
MIN6 cell lysates (1 mg) were incubated with antibody specific for IRS2 or IRS1 (Upstate, Lake Placid, NY, USA) overnight at 4°C with gentle agitation. Protein A/G beads were then added and incubation was continued for 4 h at 4°C. Immunoprecipitated proteins were subjected to SDS-PAGE and transferred to nitrocellulose membranes. The membranes were incubated with anti-phosphotyrosine antibody (Upstate, Lake Placid, NY, USA).
Quantitative data are presented as mean ± SEM from at least three independent experiments and at least 100 islets from more than three mice unless indicated otherwise. We assessed interactions among variables by two-way analysis of variance and used the Student’s t test to compare independent means, or the Wilcoxon–Mann–Whitney U test to compare means when sample sizes differed. A p value of <0.05 was considered statistically significant.
Generation of beta cell-specific Gsk-3β knockout mice
Tissue-specific deletion of GSK-3β was demonstrated by observing that the abundance of GSK-3β was similar in the brain, hypothalamus, adipose tissue, liver, skeletal muscle and heart in β-Gsk-3β–/– and control mice (ESM Fig. 2), consistent with previous observations .
Metabolic characteristics of β-Gsk-3β–/– mice
Fasting and fed glucose and insulin values at 12 weeks of age for RIP-Cre control and β-Gsk-3β–/– mice
Blood glucose concentration (mmol/l)
Serum insulin concentration (pmol/l)
No. of mice
No. of mice
No. of mice
No. of mice
4.3 ± 0.1
8.3 ± 0.2
2.1 ± 0.14
6.3 ± 0.14
3.9 ± 0.1*
7.5 ± 0.2**
2.8 ± 0.14
6.7 ± 0.14
Morphology of islets of β-Gsk-3β–/– mice
To assess another possible mechanism for the improved glucose tolerance and glucose-stimulated insulin secretion, the effects of islet GSK-3β deficiency on insulin secretion from isolated islets was examined. As shown in Fig. 3d, insulin release from islets of β-Gsk-3β–/– mice at 16.8 mmol/l glucose appeared to be increased compared with that of islets of control mice, although the increase was not significant (p = 0.06). Islet insulin contents of β-Gsk-3β–/– mice were indistinguishable from those of control mice (Fig. 3e). Thus, the increased insulin levels in β-Gsk-3β–/– mice may be due to increased insulin secretion per islet as well as increased islet mass.
Signalling mechanisms underlying expanded beta cell mass and function in β-Gsk-3β–/– mice
Further support for the hypothesis that reduced GSK-3β activity in β-Gsk-3β–/– mice can increase IRS1 and IRS2 levels and insulin signalling was obtained by in vitro experiments with a potent and selective inhibitor of GSK-3 activity, SB216763 . Islets were isolated from 12-week-old male wild-type mice, serum-deprived for 6 h, then treated with either SB216763 or DMSO as a vehicle for 16 h in RPMI medium. Western blot analysis of cell lysates showed that inhibition of GSK3 was associated with enhanced levels of IRS1 and IRS2 and PDX1 protein, along with increased phospho-Akt (Fig. 4b), findings consistent with those obtained in islets isolated from β-Gsk-3β–/– mice.
GSK-3β has been shown to phosphorylate several intracellular components of the Wnt signal transduction pathway . It is well known that wingless-type MMTV integration site family member 2 (WNT) signalling regulates cellular proliferation and development. GSK-3β activity promotes phophorylation of β-catenin and protein degradation. As shown in ESM Fig. 5a, active β-catenin measured with antibody specific for dephosphorylation at serine 37 and threonine 41  was increased in islets from β-Gsk-3β–/– mice relative to control mice. These results are consistent with activation of β-catenin as a contributor to the observed expanded beta cell mass in β-Gsk-3β–/– mice.
The effects of GSK-3β activity on cell cycle regulatory proteins in beta cells have been little studied. Irs2 knockout mice have been shown to have fourfold elevation of GSK-3 activity in islets that is associated with an elevated level of the cell cycle inhibitor p27Kip1 and GSK-3 activity in insulinoma cells has been shown to stabilise p27 protein . Similarly, in other cell types GSK-3 activity has been shown to regulate degradation, for example the stability of p27 [15, 27]. The protein levels of the cyclin-dependent kinase inhibitors p27kip1 and p21cip1 in islets from β-Gsk-3β–/– mice did not appear to be altered (ESM Fig. 5b).
Evidence that GSK3B inhibition increases insulin signalling and reduces proteasomal degradation of IRS proteins in vitro
After phosphorylation by GSK-3, a number of proteins have been demonstrated to be regulated at the post-translational level by either enhancing stability  or increasing degradation [29, 30]. One of the mechanisms whereby GSK-3 activity increases insulin resistance in skeletal muscle is by phosphorylating and promoting proteasomal degradation of IRS1 , and in liver by a similar post-translational reduction in IRS2 . An increased IRS protein level in islets of β-Gsk-3β–/– mice could be accounted for by effects of GSK-3β activity on the stability of proteins. An experiment was designed to test this hypothesis.
Mouse insulinoma cells (MIN6) were treated with cycloheximide to block new protein synthesis and with either LiCl to inhibit GSK-3β activity or MG132 to inhibit proteasome-mediated protein degradation . Fatty acid treatment of MIN6 insulinoma cells was shown to induce endoplasmic reticulum stress and reduce insulin receptor/Akt signalling , so cotreatment with fatty acid (palmitate 400 μmol/l) for 4 h was used in an attempt to induce endoplasmic reticulum stress and further activate GSK-3β activity. Cell lysates were analysed by western blotting, as shown in Fig. 5b. Addition of cycloheximide for 4 h resulted in a reduction of approximately 30–35% in IRS1 and IRS2 protein levels. The addition of LiCl had a significant effect on the preservation of these proteins. Addition of fatty acid resulted in further reduction of IRS1 and IRS2 levels, and again LiCl addition significantly increased protein levels. MG132 cotreatment preserved both the IRS1 and the IRS2 protein level. These results suggest that endogenous GSK-3β regulates IRS1 and IRS2 protein levels in beta cells through proteasome-mediated degradation.
β-Gsk-3β–/– mice on a high-fat diet exhibited resistance to diabetes
Previous studies of islets with chemical inhibitors of GSK-3β suggested that this enzyme might be rate-limiting for beta cell proliferation . In the present study we sought to examine the hypothesis that GSK-3β is rate-limiting for islet beta cell mass and function, and that targeted deletion in β-Gsk-3β–/– mice would result in expanded beta cell mass and function. These mice were found to have improved glucose tolerance, enhanced glucose-induced insulin secretion, and increased beta cell mass accompanied by enhanced proliferation and decreased apoptosis. Additionally, β-Gsk-3β–/– mice on a high-fat diet exhibited resistance to fat feeding-induced glucose intolerance relative to control mice. Molecular mechanisms accounting for these phenotypic changes in chow-fed mice included increased levels of islet IRS1 and IRS2 proteins and phospho-Akt, suggesting enhanced signalling through the insulin receptor/PI3K/Akt pathway and increased islet PDX1 levels. Inhibition of GSK3 in MIN6 cells in vitro led to increased IRS1 and IRS2 protein through inhibition of proteasomal degradation. These results are consistent with a mechanism whereby endogenous GSK-3β activity controls islet beta cell growth by feedback inhibition of the insulin receptor/PI3K/Akt signalling pathway.
An increase of more than 70% in islet IRS2 protein levels in β-Gsk-3β–/– mice relative to control mice on a normal diet was noted (Fig. 4a). The importance of IRS2 protein to islet beta cell function was established with IRS2-deficient mice that developed profound diabetes with severe loss of beta cells at a young age . Our finding of increased IRS2 and phospho-Akt in β-Gsk-3β–/– mice (Fig. 4a), along with in vitro evidence in MIN6 cells that GSK-3β phosphorylates IRS2 and targets it for proteasome-mediated degradation (Fig. 5b), suggests a possible inhibitory mechanism for GSK-3β in insulin receptor/PI3K/Akt signalling in islet beta cells. An increase of more than 70% in islet IRS1 protein levels in β-Gsk-3β–/– mice relative to control mice was also noted, although the role of IRS1 in islet beta cell function relative to that of IRS2 is less clear.
The expanded beta cell mass in β-Gsk-3β–/– mice was accompanied by enhanced proliferation and decreased apoptosis, and one of the effectors could be PDX1. Previous in vitro experiments in insulinoma cells showed that GSK-3 activity phosphorylates PDX1 and promotes its proteasomal degradation . In the present in vivo study we observed in βGsk-3β–/– mice a significant increase in islet PDX1 protein level relative to that in control mice, consistent with the reduction in PDX1 degradation and with increased insulin receptor/PI3K/Akt signalling. We did not measure PDX1 phosphorylation, however, as an antibody to the serine site phosphorylated by GSK3 was not available. Another in vivo example of GSK-3β regulation of PDX1 levels was observed in Irs2−/− mice rescued by crossing with Gsk-3β-haploinsufficient mice . Other proteins known to regulate islet beta cell proliferation include β-catenin and D-cyclins, and GSK-3 activity is well known to affect the levels of these proteins . In the present study preliminary experiments showed that islets of β-Gsk-3β–/– mice had an increase in active β-catenin (ESM Fig. 5a), which may have contributed to the expansion of beta cell mass.
The phenotypes observed in β-Gsk-3β–/– mice leave unanswered the question of the role of islet GSK3A in beta cell development and function. Conditional deletion of the Gsk-3α (also known as Gsk3a) gene in islet beta cells may result in further expansion of mass and function, but this has yet to be determined.
Numerous studies have reported insulin resistance in animals fed an HFD. We compared RIP-Cre control with β-Gsk-3β–/– mice on both a normal diet and an HFD. Fasting plasma glucose levels in β-Gsk-3β–/– and RIP-Cre mice on the normal diet and the HFD are shown for comparison in ESM Fig. 4. Results of an intraperitoneal glucose tolerance test (IPGTT) showed that on the HFD the glucose tolerance of β-Gsk-3β–/– mice at 24 weeks of age was significantly improved relative to RIP-Cre control mice at the same age (Fig. 6a) on the HFD. We also used the area under the curve to assess IPGTT results (ESM Fig. 6a, b). The results showed that, on the normal diet, blood glucose was about 31% lower in β-Gsk-3β–/– mice compared with RIP-Cre control mice, and on the HFD it was reduced by about 27% in β-Gsk-3β–/– mice compared with RIP-Cre controls (ESM Fig. 6a, b). We also compared the beta cell mass of β-Gsk-3β–/– and RIP-Cre mice on the two diets, although the data for the normal diet were on 12-week-old mice while those for the HFD were on 24-week-old mice. The beta cell mass of βGsk3b–/– mice on the normal diet was 22% higher than that of RIP-Cre control mice on the normal diet, while on the HFD the beta cell mass of β-Gsk-3β–/– mice was 45% higher than that of RIP-Cre control mice. These results indicate that β-Gsk-3β–/– mice resist fat feeding-induced diabetes at least in part as a result of increased beta cell mass and proliferation.
A number of genetic models that alter the insulin signalling pathway in mice have provided substantial evidence that this pathway is a major contributor to beta cell mass and islet function. Signalling through the PI3K/Akt pathway has been shown to be critical for the eventual outcome of beta cell mass and function. There are at least three major targets of this signalling, including mechanistic target of rapamycin (serine/threonine kinase) (mTOR)/S6 kinase, FOXO and GSK-3. The present study provides evidence that GSK-3β is rate-limiting for beta cell growth by feedback inhibition of the insulin signalling pathway through its inhibitory effects on IRS1 and IRS2. There are few direct data on FOXO activity in pancreatic islet beta cells. In vivo studies with Irs2−/− mice showed reduced beta cell mass and diabetes, and crossing these mice with mice haploinsufficient for Foxo1 restored beta cell mass and proliferation in these mice . An acute in vitro study using an adenovirus expressing a gain of function form of FOXO showed decreased beta cell glucose utilisation and insulin secretion .
While both GSK-3β and FOXO appear to have negative effects on beta cell proliferation and mass, mTOR complex 1 (mTORC1)/S6 kinase 1 signalling appears to have the opposite effect. In vitro studies with rapamycin showed inhibition of proliferation , and in vivo knockout of S6k1 (also known as S6k) resulted in reduced beta cell mass and diabetes . Thus, the present study shows that GSK-3β is only one part of the insulin-signalling pathway controlling islet beta cell mass and function.
In skeletal muscle and liver GSK-3β is constitutively active, and in the fasted state with low insulin levels this enzyme inhibits glycogen storage. With feeding, insulin secretion inhibits the enzyme, facilitating glycogen synthesis. In the present study, tissue-specific ablation of GSK-3β in pancreatic islet beta cells resulted in a phenotype that suggested the normal function of this enzyme. In contrast to peripheral tissues, the present findings suggest that GSK-3β controls the expansion of beta cell proliferation in the islet. The finding that absence of the enzyme in beta cells during the stress of fat feeding results in normal glucose tolerance and enhanced beta cell mass indicates that this enzyme is a potentially important therapeutic target for treating diet-induced diabetes in man.
The authors would like to thank other members of Dr Permutt’s lab for helpful discussions. The Washington University Diabetes Research Training Center (P60DK020579-30) is acknowledged for assistance from the Immunoassay and Morphology Cores and the Adipocyte Biology Core, Clinical Nutrition Research Unit (NIH DK56351).
This work was supported by National Institutes of Health grants (R37 DK16746, to MAP) and the Canadian Institutes of Health Research (J. Woodgett) and Grants-in-Aid for Scientific Research (grant no. 22590984, to K. Tanabe) from the Ministry of Education, Culture, Sports, Science and Technology of Japan. K. Tanabe was funded by the Takeda Science Foundation. Y. Liu was supported by an American Diabetes Association Mentor Based Fellowship Award.
Duality of interest
The authors declare that there is no duality of interest associated with this manuscript.