Fetal PGC-1α Overexpression Programs Adult Pancreatic β-Cell Dysfunction

Mouse islets were isolated with collagenase solution (1 mg/mL, Sigma-Aldrich, St. Louis, MO), separated on Histopaque (Sigma-Aldrich) gradient, and handpicked under a binocular microscope (Leica Microsystems GmbH, Wetzlar, Germany). Fetal rat islets were isolated from embryonic day 21 pancreata and cultured, as previously described (20). Islets were treated with 10⁻⁷ mol/L dexamethasone (Dex, Sigma-Aldrich). Min6 cells were similarly treated and cultured as described (21).

CD1 Swiss pregnant mice underwent food restriction (50% of the control daily food intake) from 11.5 days of pregnancy until 18.5 days, as previously described (8). At 18.5 days of pregnancy, fetal pancreata were dissected.

PGC-1α overexpression was achieved using an adenovirus expressing the human PGC-1α cDNA under the control of a cytomegalovirus promoter (22). Islets from mutant mice with a GR deletion in β-cells (GR^RipCre mice) (23) and control mice were isolated, as described above.

DNA were extracted from Min6 cells and cross-linked with formaldehyde, and the sonicated protein-DNA complexes were isolated, incubated overnight at 4°C with anti-GR (M20, Santa Cruz Biotechnology, Santa Cruz, CA) or anti–PGC-1α antibody, and immunoprecipitated. In another set of experiments, protein-DNA complexes were immunoprecipitated first with the anti-GR antibody then with the anti–PGC-1α antibody. PCR was performed with AmpliTaq Gold Polymerase (Applied Biosystems, Life Technologies). Primers sequences are available upon request.

Transgenic mice expressing the tetracycline transactivator from a bacterial artificial chromosome transgene containing the Insulin1 gene regulatory elements (Ins-tTA) were generated in our laboratory (24) as were transgenic mice carrying the tetracycline response element (TRE) controlling PGC-1α expression (TetO PGC-1α), which were described previously (25). The two mouse lines were crossed to generate Ins-PGC-1α double-transgenic mice. To stop PGC-1α overexpression from conception until adult age, pregnant and lactating mice were given 0.1 g/L doxycycline (Dox, Sigma-Aldrich) in their drinking water, and weaned mice received 1 g/L until adult age. Mice with PGC-1α overexpression never received Dox. All animal experiments were done according to the “Principles of Laboratory Animal Care” and the French law, authorization No. 75-435 delivered to B.V. by the French Ministry of Agriculture.

Glucose (2 g/kg body weight) was injected intraperitoneally to fasted mice, and blood glucose levels were measured before and 15, 30, 60, and 120 min after injection using a glucometer (Freestyle Papillon Mini; Abbott Diabetes Care, Abbott Park, IL). Serum insulin levels were measured by ELISA (Mercodia, Uppsala, Sweden).

After a 5-h fast, insulin (1 unit/kg body weight) was injected intraperitoneally. Blood glucose levels were measured before and 15, 30, 60, and 120 min after the insulin injection.

Pancreatic insulin contents were extracted at −20°C in acidic ethanol (1.5% [vol/vol] HCl in 75% [vol/vol] ethanol) and assayed by ELISA kit (Mercodia).

Pancreata from adult mice were dissected, fixed, and sectioned. β-Cell mass was evaluated on eight sections per animal, as previously described (8).

Total RNAs were extracted using RNeasy Plus extraction kit (Qiagen, Hilden, Germany). RNAs were reverse-transcribed into cDNA using Superscript II reverse transcriptase (Invitrogen). Gene expression was quantified by real-time PCR with SybrGreen supermix in a MyIQ thermocycler (Biorad, Hercules, CA). Gene expression was normalized for the 18S ribosomal RNA. Primers sequences are available upon request.

Total proteins were separated on SDS-PAGE gels and transferred to nitrocellulose membranes. Membranes were incubated with antibodies raised in rabbit against Pdx1 (gift of Dr. Raphael Scharfmann, INSERM, Paris, France), against PGC-1α (25) or against actin (Sigma-Aldrich). Band quantification was performed using a Kodak Imager (Kodak, Rochester, NY).

All results are expressed as means ± SD. The statistical significance of variations was tested by a Mann-Whitney nonparametric test. P values <0.05 were considered significant.

PGC-1α is a coactivator that plays major roles for several nuclear receptors, including the GR, and GCs increase PGC-1α expression in the liver (18) where it cooperates with the GR to modulate adaptative metabolic responses. To define if PGC-1α could participate in the GCs effects in the β-cells, we first asked whether GCs regulate PGC-1α in these cells. In a previous work, we showed that maternal undernutrition leads to overexposure of the fetuses to GCs (8). Here we showed that this GCs elevation in vivo increases PGC-1α expression in vivo in the pancreata of these fetuses. Similarly, in vitro, a 24-h Dex treatment increases PGC-1α expression in isolated mouse islets, in clonal Min6 cells, and in rat fetal islets (Fig. 1A). Dex treatment increases PGC-1α protein in isolated mouse islets (Fig. 1B). This GCs-induced increase of PGC-1α expression was associated with decreased expression of key β-cell genes in Min6 cells, such as Pdx1, Nkx2.2, Pax4, Pax6, Isl1, MafA, Srebp1c, and the glucose transporter Glut2 induced by Dex treatment (Fig. 1C).

Because GCs repress most of the genes in β-cells but activate PGC-1α, we asked whether this increased PGC-1α expression could be implicated in the effects of GCs on β-cells. First, we made a time-course measurement of gene expression after Dex treatment. We observed that increased PGC-1α expression (1 h after treatment, Fig. 1D) preceded decreased Pdx1 expression (3 h after treatment, Fig. 1E), suggesting that GCs-stimulated PGC-1α may be involved in GC repression of Pdx1. To define this potential role, we overexpressed PGC-1α with an adenovirus in Min6 cells and found that it led to decreased mRNA levels of Pdx1 Nkx2.2, Pax4, Pax6, Isl1, MafA, Nkx6.1, Ins1, Srebp1c, and Hnf4a (Fig. 2A). The similarities between Dex treatment and PGC-1α overexpression on β-cell gene regulation suggested that PGC-1α participates in the GCs effects on β-cells.

Because PGC-1α associates with many nuclear receptors, we tried to specify if PGC-1α action on Pdx1 was mediated through the GR. Islets from mice deleted for the GR in β-cells, GR^RipCre (23), and from control mice were infected with an adenovirus expressing PGC-1α (Fig. 2B). In control islets, PGC-1α overexpression decreased Pdx1 expression, whereas in the absence of GR, PGC-1α overexpression had no effect on the Pdx1 mRNA level, demonstrating that Pdx1 repression by PGC-1α requires the presence of the GR.

We next studied whether GR and PGC-1α regulate Pdx1 through direct repression of the Pdx1 promoter. First, we aimed at defining if a complex containing GR and PGC-1α could bind to the Pdx1 promoter. Chromatin immunoprecipitation (ChIP) with an anti-GR antibody of sheared DNA from Min6 cells revealed specific amplification bands for areas II and III of the Pdx1 promoter (Fig. 2C). ChIP with an anti–PGC-1α antibody revealed PCR products for the same areas (Fig. 2C), showing that PGC-1α is recruited onto these DNA regions. Then, sequential ChIP, first with an anti-GR antibody and then with an anti-PGC-1α antibody, led to PCR products for areas II and III, demonstrating that a complex containing PGC-1α and GR binds to areas II and III. Additional experiments using mutations in areas II and III showed that GCs actually repress Pdx1 promoter activity through binding of the GR to area II (Supplementary Data).

Because we had demonstrated in vitro that PGC-1α was implicated in the regulation of Pdx1 and other β-cell genes, we generated mice with β-cell specific PGC-1α overexpression to define how PGC-1α regulates β-cells in vivo.

We used tetracycline-dependent inducible gene expression to generate mice overexpressing PGC-1α in β-cells. We crossed TetO-PGC-1α hemizygous mice carrying the cDNA coding PGC-1α under the control of a tetracycline-dependent promoter (25) with Ins-tTA mice hemizygous mice expressing the tetracycline transactivator (tTA) under regulatory elements of the Insulin1 gene recently generated in the laboratory (24). We obtained Ins-tTA + TetO-PGC-1α mice that will be referred as Ins-PGC-1α mice for the rest of the study and will be compared with Ins-tTA mice considered as controls.

We measured increased PGC-1α mRNA (Fig. 3A) and protein (Fig. 3B) levels in islets from Ins-PGC-1α mice at age 6 months compared with control islets. Ins-PGC-1α mice were hyperglycemic in the fed state (9.9 ± 2.7 vs. 7.1 ± 1.6 mmol/L in controls, P < 0.05), with lower blood insulin levels (4.2 ± 1.6 vs. 7.8 ± 3.72 pmol/L, P < 0.05). The intraperitoneal glucose tolerance test (ipGTT) revealed glucose intolerance in mice overexpressing PGC-1α in β-cells (Fig. 3C) with altered glucose-stimulated insulin secretion (GSIS) during the ipGTT (Fig. 3D). Whole-body insulin sensitivity assessed by the insulin tolerance test showed no apparent differences between the two groups (Fig. 3E). Thus, β-cell–specific PGC-1α overexpression impairs glucose tolerance and decreases GSIS.

In islets from 6-month-old Ins-PGC-1α mice, β-cell PGC-1α overexpression was associated with dramatic reductions in the mRNA levels of several key β-cell genes, including Pdx1, MafA, Nkx2.2, Ins1, Ins2, ZnT-8, Glut2, glucokinase (GK), Kir6.2, Sur1, and Hnf4α (Fig. 4A). No change in glucagon expression was observed, suggesting that α-cells were not altered.

Analysis of the endocrine pancreas from Ins-PGC-1α mice showed a strong reduction of β-cell fraction (Fig. 4B), absolute β-cell mass (Fig. 4C), β-cell mass per unit of body weight (Fig. 4D), and pancreatic insulin content (Fig. 4E). Precise analysis revealed a decreased number of islets per tissue surface unit (Fig. 4F), decreased mean islet size (Fig. 4G), and β-cell hypotrophy (Fig. 4H) in Ins-PGC-1α mice.

To clearly prove that the phenotype of Ins-PGC-1α mice was caused by PGC-1α overexpression, control and Ins-PGC-1α mice were treated with Dox in their drinking water from conception to adulthood to inhibit PGC-1α overexpression. PGC-1α mRNA levels in islets from Dox-treated Ins-PGC-1α mice were comparable with Dox-treated controls (Fig. 5A), and these animals presented normal glucose tolerance (Fig. 5B) and GSIS (Fig. 5C).

To determine whether β-cell–specific PGC-1α overexpression starting from adult stage could impair β-cell function and glucose tolerance, Ins-PGC-1α mice were given Dox from conception to adulthood, and Dox treatment was stopped at age 4 months to induce PGC-1α overexpression. Four months after the arrest of Dox treatment, we observed that although PGC-1α mRNA levels in the islets were increased as expected (Fig. 5D), Ins-PGC-1α mice presented normal glucose tolerance (Fig. 5E) and GSIS (Fig. 5F). Yet, the increased PGC-1α expression in Ins-PGC-1α islets was still associated with decreased Pdx1, MafA, Glut2, GK, or Ins2 mRNA levels, whereas Ins1, Kir6.1 and Sur-1 mRNA levels remained unchanged (Fig. 5G).

To investigate the consequences of PGC-1α overexpression only during fetal life, we stopped PGC-1α overexpression by giving Dox to Ins-PGC-1α mice from birth onwards. At birth (P1), just before Dox treatment, we observed decreased β-cell fraction in Ins-PGC-1α neonates (Fig. 6A) together with PGC-1α overexpression (Fig. 6B, P1). When Dox treatment was given from birth, PGC-1α overexpression was stopped, as illustrated by normal mRNA levels in Ins-PGC-1α at postnatal day 21 (P21; Fig. 6B). Dox treatment was continued until adulthood, ∼6 months old. In mice with PGC-1α expression only during fetal life, we found no difference at adult age in islet PGC-1α or other β-cell gene mRNA levels compared with control mice (Fig. 6C). Yet, those mice showed impaired glucose tolerance (Fig. 6D) and decreased GSIS (Fig. 6E). Interestingly, their β-cell mass was similar to that of controls (Fig. 6F). Thus, β-cell–specific PGC-1α overexpression during fetal life only was sufficient to impair adult β-cell function without altering β-cell mass.

Exciting hypotheses suggest that type 2 diabetes may originate from alterations in the development of organs involved in glucose homeostasis as a consequence of an adverse fetal environment (26). We proposed that β-cell development is impaired when fetal homeostasis is perturbed, especially when fetuses are exposed to an excess of GCs (6,8). Here, we have characterized the molecular mechanisms through which GCs regulate β-cell development and function. We first showed that GCs upregulate the expression of the GR coregulator PGC-1α in β-cells and that PGC-1α participates in the GCs effects on these cells. Then, we showed that GCs potently repress the expression of the key transcription factor, Pdx1, in β-cells through a direct binding of a GR/PGC-1α complex to the Pdx1 promoter on a crucial region for β-cell–specific promoter activity. We generated mice overexpressing PGC-1α in β-cells that display impaired glucose tolerance associated with decreased insulin secretion, Pdx1 expression, and β-cell mass with β-cell hypotrophy. Finally, we demonstrated that PGC-1α overexpression during fetal life only is sufficient to induce β-cell dysfunction in adult mice.

We had previously shown with rodent models of fetal undernutrition that GCs are potent inhibitors of β-cell development (6) by targeting pancreatic precursor cells engaged in the endocrine lineage (23). We also suggested that a similar regulation was likely to take place in humans because the GR was expressed early in the human fetal pancreas in precursor cells at the time when they cluster and start to express Pdx1 (27). Moreover, we recently provided genetic evidence in mice that the GR was required for the negative effects of GCs on β-cell development (8). However, the precise molecular mechanisms and partners implicated remained unclear. During pancreas development and β-cell differentiation, Pdx1 has been shown to play crucial roles (28,29). Pdx1 levels are tightly regulated, and any impairment of this regulation results in the inability to regulate glucose levels (30) and β-cell mass (31). We, and others, have reported that GCs treatment leads to a decreased Pdx1 expression in rodent β-cells and islets (32–34). Here, we show that GR binds to area II of the mouse Pdx1 promoter and decreases its activity. Binding of GR is thought to require GCs response elements (GRE), but we could not identify any GRE in area II, based on consensus sequence (35). Yet, physical interaction of the GR to area II was proven by ChIP experiments. Because the area II region is conserved in the Pdx1 promoter in humans (36), it is likely that GR would bind to and also inhibit the human Pdx1 promoter, a hypothesis that requires further investigation. In line with these results, GC has been shown to inhibit the human insulin gene through the binding of the GR on a negative GRE located upstream of the gene (37).

GR is a nuclear receptor that can activate or repress gene expression, a role that depends on coregulators. PGC-1α is known to coregulate several nuclear receptors, including the GR (38). We thus investigated if PGC-1α could participate in Pdx1 regulation together with the GR. PGC-1α plays a major role in the control of gluconeogenic genes in the liver and in mitochondrial biogenesis (39), and its expression is upregulated by GC treatment in the liver (18). Here, we present data showing that GCs increase PGC-1α expression in β-cells and that PGC-1α contributes to the repression of Pdx1 expression mediated by the GCs. This places PGC-1α as an actor of the GCs regulation of β-cells. PGC-1α is usually described as a coactivator of nuclear receptors (17). We propose that PGC-1α also represses gene expression, thereby acting as a corepressor. In fact, a study showed that PGC-1α is involved in the GR-mediated repression of the human insulin gene (40). We suggest that PGC-1α plays a similar role in the GR-mediated repression of Pdx1 in β-cells. In line with this, a recent article showed that GCs treatment decreased Pdx1, increased PGC-1α expression, and inhibited porcine neonatal pancreas cell cluster differentiation into β-cells (41). Thus, similarly to what we show in murine fetal pancreas, PGC-1α is stimulated by GCs in porcine cells.

We found that GCs and PGC-1α had similar effects on most of the β-cell genes in vitro. Yet, Neurod1 and Glut2 seem to be differentially regulated. We found that Neurod1 expression is unchanged by GCs treatment and decreased by PGC-1α overexpression. Similar observations are found in the literature: PGC-1α downregulates its expression in islets (42), and in neurons GCs have no effect on Neurod1 expression (43). We found that Glut2 expression is decreased by GCs and increased by PGC-1α overexpression. GCs-induced Glut2 inhibition has been reported (44). Glut2 upregulation by PGC-1α may be due to an interaction between PGC-1α and members of the peroxisome proliferator–activated receptor (PPAR) family because a PPAR response element upregulating Glut2 expression has been identified in the Glut2 promoter (45).

Several studies in animal models and in humans suggest a link between PGC-1α and diabetes (39). Here, we show that β-cell–specific PGC-1α overexpression in transgenic mice leads at adult age to impaired GSIS and strongly decreased β-cell mass, associated with impaired glucose tolerance. So far, the cellular mechanisms explaining these defects are unknown. PGC-1α overexpression in cultured islets or β-cell lines suppressed glucose-stimulated membrane depolarization, induced glucose-6-phosphatase, and thereby reduced insulin secretion together with changes in expression of genes involved in energy metabolism and control of β-cell function (19). In addition, reduced insulin secretion has been observed after PGC-1α–induced uncoupling protein 2 (UCP2) expression in β-cells (46). An increased respiration uncoupling in islet mitochondria through UCP2 might lead to a lower ATP/ADP ratio, which could explain the impaired insulin secretion described here. Such modifications will be investigated in Ins-PGC-1α mice. In addition, we have shown that Ins-PGC-1α mice display β-cell hypotrophy, a defect that may participate in the phenotype because another study negatively correlated β-cell size with GSIS (47). The deletion of the S6K gene reduces β-cell size (48), while deletion of LKB1 increases β-cell size (49). The implication of these molecules in the PGC-1α–mediated β-cell hypotrophy deserves further investigation.

A growing body of evidence suggests that prenatal stress, most likely resulting in fetal GCs overexposure, is a major adverse intrauterine environment that alters organ development, with important implications for the developmental programming of adult diseases, including hypertension, dysregulations of the hypothalamo-pituitary-adrenal axis, and glucose homeostasis disorders (50). Our results place PGC-1α as a GCs-induced gene implicated in the β-cell programming by the fetal environment. In fact, in our model of fetal food restriction, the increased GC concentrations stimulated PGC-1α expression and reduced Pdx1 in the pancreas (8). This reinforces the idea that PGC-1α may act as an intermediate of GCs’ negative action on β-cells. Moreover, PGC-1α overexpression during fetal life only is sufficient to permanently impair adult glucose homeostasis, reinforcing the implication of this transcriptional regulator in the fetal programming of β-cells by GCs.

The precise mechanisms underlying the effects of PGC-1α on β-cell programming remain to be defined. Such alterations could be mediated through epigenetic modulations occurring during the fetal period of PGC-1α overexpression. Actually, PGC-1α has been shown to control gene expression through epigenetic mechanisms: it regulates the expression of Bmal1, a clock gene involved in β-cell function, through alterations of histone methylation and acetylation (51). How epigenetic modifications could be involved in the fetal programming by PGC-1α remains to be investigated.

In conclusion, our data point out the implication of the GR and its coregulator PGC-1α in the negative control of Pdx1 expression and β-cell function. We depict the molecular mechanisms underlying the control of β-cell mass and function by GCs and also propose that PGC-1α may participate in the GCs inhibition of β-cells. Finally, we point out β-cell PGC-1α overexpression during fetal life as a potential mechanism participating in the programmed dysfunction of adult β-cells.

B.V. received a fellowship from the Ministère de l’Enseignement Supérieur et de la Recherche and financial support from the Fondation pour la Recherche Médicale. These studies were supported by INSERM, CNRS, the European Union (FP5 Nutrix QLK1-CT-2000-00083 and FP6 Earnest Food-CT-2005-007036 programs), the Agence Nationale de la Recherche (Grant Gramsy No. A05245DS), Société Francophone du Diabète to J.-P.R., B.Bl., and B.Br.; Lilly to J.-P.R.; and Institut Benjamin Delessert to B.Bl. No other potential conflicts of interest relevant to this article were reported.

B.V. and J.-P.R. designed the studies, performed experiments, analyzed the data, and wrote, reviewed, and edited the manuscript. P.Z., A.S.-E., M.A., N.V., and A.B. performed experiments and analyzed the data. D.P.K. developed and provided the TRE-PGC-1α mice and reviewed the manuscript. F.T. developed and provided the GR floxed mice and reviewed the manuscript. P.F. and J.-F.G. reviewed the manuscript. B.Br. and B.Bl. designed the stud ies, and wrote, reviewed and edited the manuscript. B.Bl. 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.

The authors thank Delphine Dorchene (INSERM, Paris, France) for mouse husbandry, Caroline Peschard (INSERM, Paris, France), Isabelle Hainault (INSERM, Paris, France), and Adriana Rodriguez-Trejo (Universidad Nacional Autónoma de México, Mexico City, Mexico) for technical assistance, and the Centre d’Imagerie Cellulaire et Cytometrie of UMRS 872 (INSERM, Paris, France). The authors thank Domonique Langin (INSERM, Toulouse, France) and Raphael Scharfmann (INSERM, Paris, France) for sharing reagents and Eric Ravussin (Pennington Biomedical Research Center, Baton Rouge, LA) and Catherine Postic (INSERM, Paris, France) for helpful advice.