Peroxisome Proliferator-activated Receptor γ (PPARγ) and Its Target Genes Are Downstream Effectors of FoxO1 Protein in Islet β-Cells

Background: The molecular mechanisms for islet β-cell compensation and failure are not fully known. Results: FoxO1/PPARγ signaling regulates key β-cell genes, with this network being up-regulated in nondiabetic insulin-resistant rats and impaired in rodents with diabetes. Conclusion: We examine the potential for the FoxO1/PPARγ network as a feature of β-cell compensation and failure. Significance: We identify targets for prevention of type 2 diabetes. The molecular mechanisms and signaling pathways that drive islet β-cell compensation and failure are not fully resolved. We have used in vitro and in vivo systems to show that FoxO1, an integrator of metabolic stimuli, inhibits PPARγ expression in β-cells, thus transcription of its target genes (Pdx1, glucose-dependent insulinotropic polypeptide (GIP) receptor, and pyruvate carboxylase) that are important regulators of β-cell function, survival, and compensation. FoxO1 inhibition of target gene transcription is normally relieved when upstream activation induces its translocation from the nucleus to the cytoplasm. Attesting to the central importance of this pathway, islet expression of PPARγ and its target genes was enhanced in nondiabetic insulin-resistant rats and markedly reduced with diabetes induction. Insight into the impaired PPARγ signaling with hyperglycemia was obtained with confocal microscopy of pancreas sections that showed an intense nuclear FoxO1 immunostaining pattern in the β-cells of diabetic rats in contrast to the nuclear and cytoplasmic FoxO1 in nondiabetic rats. These findings suggest a FoxO1/PPARγ-mediated network acting as a core component of β-cell adaptation to metabolic stress, with failure of this response from impaired FoxO1 activation causing or exacerbating diabetes.

The molecular mechanisms and signaling pathways that drive islet ␤-cell compensation and failure are not fully resolved. We have used in vitro and in vivo systems to show that FoxO1, an integrator of metabolic stimuli, inhibits PPAR␥ expression in ␤-cells, thus transcription of its target genes (Pdx1, glucose-dependent insulinotropic polypeptide (GIP) receptor, and pyruvate carboxylase) that are important regulators of ␤-cell function, survival, and compensation. FoxO1 inhibition of target gene transcription is normally relieved when upstream activation induces its translocation from the nucleus to the cytoplasm. Attesting to the central importance of this pathway, islet expression of PPAR␥ and its target genes was enhanced in nondiabetic insulin-resistant rats and markedly reduced with diabetes induction. Insight into the impaired PPAR␥ signaling with hyperglycemia was obtained with confocal microscopy of pancreas sections that showed an intense nuclear FoxO1 immunostaining pattern in the ␤-cells of diabetic rats in contrast to the nuclear and cytoplasmic FoxO1 in nondiabetic rats. These findings suggest a FoxO1/PPAR␥-mediated network acting as a core component of ␤-cell adaptation to metabolic stress, with failure of this response from impaired FoxO1 activation causing or exacerbating diabetes.
Peroxisome proliferator-activated receptor ␥ (PPAR␥) 2 is a member of the nuclear hormone receptor family of ligand-inducible transcription factors and plays a pivotal role in diverse biological processes such as glucose and lipid homeostasis, inflammation, and cellular proliferation and differentiation (1,2). In addition to adipocytes and several other cell types, it is expressed in islet ␤-cells (3). We have shown that PPAR␥ in ␤-cells regulates transcription of the prodifferentiation transcription factor Pdx1 (4,5) and the receptor for the incretin hormone glucose-dependent insulinotropic polypeptide (GIP) (6). Others have reported that PPAR␥ regulates expression of the anaplerotic enzyme pyruvate carboxylase in adipose tissue (7), but whether this occurs in ␤-cells is unknown. Collectively, these genes are key controllers of ␤-cell mass, mitochondrial fuel metabolism, and insulin secretion, suggesting the potential for a central role of PPAR␥ in ␤-cell compensation and/or failure. Consistent with this proposal, we have shown increased expression of PPAR␥ and its target genes in islets from rodents with successful ␤-cell adaption to an experimental reduction of ␤-cell mass (4,6) and to obesity and insulin resistance (6,8) and decreased expression in a diabetic rodent model (6,8). Also, clinical trials with thiazolidinedione PPAR␥ agonists (9) have shown a high success rate at stabilizing ␤-cell function and preventing type 2 diabetes (10).
Understanding why ␤-cell PPAR␥ expression is defective under hyperglycemic conditions requires characterizing its upstream regulation. In adipocytes, PPAR␥ gene expression is under inhibitory control by FoxO1 (11,12). FoxO1 is a member of the family of winged-helix/forkhead transcription factors that serve important roles in cellular differentiation, proliferation, apoptosis, and the response to cellular stress in many tissues. FoxO1 is highly expressed in ␤-cells and is a key regulator of ␤-cell development, mass, and function (13)(14)(15)(16). The best described FoxO1 target gene in ␤-cells is Pdx1 (17). The mechanism is reported to be competition for the FoxA2-binding element on the Pdx1 promoter, with phosphorylation of FoxO1 resulting in nuclear export and release of its blockade of FoxA2dependent Pdx1 expression. However, the evidence is mostly correlative: EMSA and competition assays for FoxO1 and FoxA2 on the Pdx1 promoter, and imaging studies showing cytoplasmic FoxO1 in Pdx1-positive ␤-cells and nuclear FoxO1 in Pdx1-negative ␤-cells (14,17). FoxO1 activity is modulated by post-translational modifications and protein-protein interactions that impact its intracellular trafficking and function (18). In the ␤-cell, insulin or growth factor PI3K/Akt-induced FoxO1 phosphorylation leads to its translocation from the nucleus to cytoplasm, relieving its inhibition of target gene transcription (14). The incretin hormones GIP and glucagon-like peptide 1 (GLP-1) also activate this sequence (19,20). Another is the balance between oxidative stress-induced FoxO1 acetylation and deacetylation (21). Thus, FoxO1 orchestrates ␤-cell transcriptional regulatory pathways for adaptive responses to many hormones, growth factors, and metabolic stresses. Whereas much is known about FoxO1 activity in ␤-cells, much less is known regarding its downstream target genes.
In this study we provide evidence for a central role for PPAR␥ in FoxO1-dependent regulation of Pdx1 expression and other key ␤-cell genes. Also, we show strong nuclear localization of FoxO1 in ␤-cells along with reduced expression of PPAR␥ and its target genes in diabetic rats, suggesting a novel mechanism for ␤-cell failure.

EXPERIMENTAL PROCEDURES
Cell Culture and in Vitro Assays-INS-1 (832/13) cells and ␤TC6 cells were maintained as described previously (5). PPAR␥ and FoxO1 siRNA were carried out in INS-1 cells using pooled SMART selection-designed siRNA duplexes with UU 3-overhang and 5Ј-phosphate on the antisense strand that targeted rat FoxO1 (NM_012560) and rat PPAR␥ (NM_013124) (Dharmacon RNA Technology). Transfections were carried out with Dharmafect 4 transfection reagent for 24 h at final siRNA concentrations of 100 pM. The protocol was two siRNA transfections at 0 h and 48 h; control cells were transfected with scrambled siRNA.
ChIP was performed with ␤TC6 cells using ChIP-IT kit (Active Motif) and a previously described method (5). Rabbit polyclonal FoxO1 antibody (Chemicon) was added to aliquots of precleared 300 -500-bp chromatin fragments and incubated overnight, with parallel samples incubated with the negativecontrol IgG provided with the kit. Protein G-agarose beads were added to the mixture for 1.5 h at 4°C. After reversing cross-links, DNA was isolated and PCR performed using primers for the mouse PPAR␥ promoter. Primer sequences were: forward, 5Ј-AACATCAACCATTGGAATACTACTGC-3Ј and reverse, 5Ј-TGGTTGAAACTCACACATCTGAAG-3Ј with an expected 281-bp PCR product. PCR conditions were 1 cycle at 94°C for 3 min, 40 cycles at 94°C for 30 s, 62°C for 1 min, 72°C for 1 min. A mouse-specific positive control was performed using a kit based on the binding of the transcription factor EFI-␣ by anti-RNA polymerase II (Active Motif).
Luciferase reporter gene assays were carried out in 60 -70% confluent INS-1 cells in 6-well plates incubated overnight in antibiotic-free medium. Co-transfections were performed with the FoxO1 overexpression cassette (pcDNA3-FLAG-FKHR from Dr. Kun-Liang Guan, University of Michigan) and the Pdx1 promoter pTAL-PPRE-Pdx-1 luciferase plasmid (Dr. Roland Stein, Vanderbilt University) or 3ϫ PPRE-luciferase reporter vector (Dr. Speigelman laboratory, Addgene) with Lipofectamine 2000 transfection reagent (Invitrogen). Renilla luciferase reporter plasmid (pRL-TK, Promega) was included (0.05 g) in all transfections as internal control. Cells were lysed 48 h after transfection, and luciferase assay performed in a TD 20/20 luminometer (Turners Design) using a dual luciferase assay kit (Promega). Firefly luciferase activity was normalized to Renilla luciferase and expressed as relative luciferase activity of the reporter constructs with and without transfection of the FoxO1 overexpression cassette.
Animal Studies-All protocols were approved by the University of Vermont Institutional Animal Care and Use Committee. FoxO1-haplodeficient mice (17) and mice that overexpress a constitutively nuclear FoxO1 transgene (FoxO1 S253A ) in ␤cells and liver (13,22) were obtained from D. Accili (Columbia University) on a mixed background and were backcrossed for Ͼ10 generations to the C57BL/6 background. Mice with PPAR␥ deficiency restricted to pancreatic epithelium (PANC PPAR␥ Ϫ/Ϫ ) were generated by crossing Pdx1-Cre mice (original source, D. Melton, Harvard University) and mice with two floxed PPAR␥ alleles as detailed previously (5). Some animals underwent glucose tolerance testing after an overnight fast that consisted of 2 g/kg intraperitoneal glucose with serum glucose measured at 0, 30, 60, 90, and 120 min (Freestyle glucose meter). Zucker lean (ZL, fa/ϩ or ϩ/ϩ) and Zucker fatty (ZF, fa/fa) male rats (Harlan) underwent 60% pancreatectomy (Px) or sham Px surgery at 6 weeks of age as described previously (8). Islets were isolated by pancreas duct perfusion with collagenase, Histopaque gradient separation, and hand picking.
Results are expressed relative to control gene expression (cyclophilin B).
GLP-1 Induction of Pdx1 Gene Expression-Isolated islets from PANC PPAR␥ Ϫ/Ϫ and WT mice were cultured overnight in RPMI 1640 medium, 10% FBS, 11.2 mM glucose, 100 units/ml penicillin, 100 g/ml streptomycin, followed by overnight culture in serum-free medium with 3% BSA. Islets were incubated in serum-free medium with 3% BSA and 11.2 mM glucose plus freshly dissolved 100 nM mouse GLP-1 (Phoenix Pharmaceutical) or vehicle for 8 h followed by quantitative PCR analysis.
Pyruvate Carboxylase (PC) Activity-Freshly isolated islets from 3 week postsurgery sham and 60% Px ZF and ZL rats were frozen in liquid nitrogen and shipped to Dr. Ye Q. Liu at the University of Louisville. PC activity was measured as described previously (23).
Pancreas Immunofluorescence and Islet Morphometry-Excised pancreata were immersion-fixed overnight in 4.0% paraformaldehyde in 0.1 M phosphate buffer at 4°C followed by embedding in paraffin. ␤-Cell mass was quantified using a previously detailed computer planimetric method (24). For multiple-labeling immunofluorescence studies of FoxO1 and Pdx1, sections were stained with guinea pig anti-insulin (Linco/Millipore) to mark ␤-cells, mouse anti-Pdx1 (Developmental Studies Hybridoma Bank, University of Iowa), and rabbit anti-FoxO1 (Cell Signaling, clone C29H4). Secondary antibodies were multiple-labeling grade anti-species-specific IgG conjugated to either CY2, CY3, or Alexa Fluor 647, respectively (Jackson ImmunoResearch or Molecular Probes/Invitrogen). Sections were imaged confocally using an LSM 510 META (University of Vermont Microscopy Imaging Center).
Statistical Analysis-Data are presented as mean Ϯ S.E. or S.D., as indicated. Each data point from the animal studies represents an individual animal. Statistical significance was determined by the unpaired Student's t test or two-way ANOVA.

Identification and Characterization of FoxO1 Binding Site on
Mouse PPAR␥ Promoter-The initial studies sought to identify a biological basis for FoxO1 regulation of PPAR␥ expression. The consensus binding sequence for FoxO1 is GTAAACA (25). We searched GenBank and identified a putative FoxO1 binding site within the mouse PPAR␥ promoter at positions Ϫ825 to Ϫ831 from the transcription start site (GenBank AY236530) with one nucleotide mismatch from the consensus sequence (Fig. 1A). MatInspector software identified homologous sequences in the rat and human PPAR␥ promoters. We confirmed that this sequence binds FoxO1 by performing EMSA and FoxO1 antibody inhibition studies using ␣-32 P-labeled complementary strand oligomers of the identified sequence and nuclear extracts from mouse-derived ␤TC6 and rat-derived INS-1 cells (Fig. 1B). Also, the specificity of binding was established with a competition assay using excess unlabeled probe.
ChIP assay was performed in ␤TC6 cells to confirm FoxO1 binds to this sequence in an intact insulin-containing cell line, using FoxO1 antibody-precipitated DNA fragments as the PCR template and a primer pair that spanned the region containing the FoxO1 binding site ( Fig. 2A). Input DNA and the FoxO1 antibody-precipitated chromatin fragments both contained the correct 281-bp product in contrast to the nonimmune IgG negative control (Fig. 2B). Also shown is the positive control based on the binding of RNA polymerase II to the transcription factor EFI-␣ (Fig. 2C).
Expression studies confirmed FoxO1 inhibits PPAR␥ expression in an insulin-containing cell line. A FoxO1 siRNA protocol was established in INS-1 cells that resulted in a 60% knockdown of FoxO1 mRNA level and 47% reduction in FoxO1 protein by 96 h (Fig. 3A). PCR analysis showed a 2.0 Ϯ 0.1-fold increase in PPAR␥ mRNA level compared with the scrambled siRNA control, and a similar increase in Pdx1 mRNA level: 2.2 Ϯ 0.2-fold (Fig. 3B). Also, FoxO1 was overexpressed by transient transfection in INS-1 cells and assayed for PPAR␥ promoter activity using two PPRE luciferase reporters, Pdx1-PPRE and 3ϫPPRE. A 48-h overexpression of FoxO1 lowered Pdx1-PPRE driven luciferase activity by 41% compared with no FoxO1 overexpression (Fig. 3C). In addition, 3ϫ PPAR␥-PPRE reporter activity was decreased 44% with the FoxO1 overexpression (Fig. 3D). Together, these in vitro findings confirmed that FoxO1 binds to the identified binding site sequence and inhibits PPAR␥ promoter activity in insulin-containing cell lines.
In Vivo Studies-FoxO1 inhibition of PPAR␥ expression in native ␤-cells was investigated using isolated islets from genetically modified mice. We first tested global FoxO1-haploinsufficient mice. Null mice are embryonically lethal, but the haploinsufficient mice are normoglycemic with relatively normal
We next studied mice that hyperexpress a constitutively nuclear FoxO1 transgene in ␤-cells and liver; the transgene encodes an amino acid substitution at the Ser 253 phosphorylation site that prevents insulin-stimulated translocation of the mutant FoxO1 to the cytoplasm and is under control of the transthyretin promoter that directs expression to liver and ␤-cells (22). Others have shown that these mice on a mixed background have reductions in islet Pdx1 levels and early onset mild diabetes (13,22). We backcrossed the transgene to the C57BL/6 background and confirmed glucose intolerance in 8 -10-week-old mice, also noting a 25% reduction in ␤-cell mass compared with the littermate WT mice (Fig. 5A). As predicted, islet lysate immunoblots revealed a 45% reduction of PPAR␥ and a similar loss of Pdx1 versus WT islets (Fig. 5B).
PPAR␥ Is Required for GLP-1 Induction of Pdx1 Gene Expression in Vitro-We used GLP-1, a known mediator of FoxO1 nuclear translocation (20), to probe the FoxO1/PPAR␥/Pdx1 system. Normal mouse islets were incubated for 8 h with or without 100 nM GLP-1(7-36 amide) at 11.1 mM glucose followed by real time PCR analysis. PPAR␥ and Pdx1 signals were increased 2-2.5-fold whereas MafA and FoxA2 were unchanged, showing the selectivity of the effect (Fig. 6A). We used this protocol with islets from PANC PPAR␥ Ϫ/Ϫ mice. In agreement with the prior results, GLP-1 stimulation nearly doubled PPAR␥ and Pdx1 mRNA expression in the floxed control mice. In contrast, in the PANC PPAR␥ Ϫ/Ϫ islets a GLP-1induced increase in Pdx1 expression was absent (Fig. 6B), demonstrating that PPAR␥ is essential for in vitro GLP-1 activation of Pdx1 expression.
PPAR␥ Regulates PC Expression in Islet ␤-Cells-We have shown that PPAR␥ transcriptionally regulates Pdx1 (4, 5) and GIP receptor genes (6) in ␤-cells. Others have shown PPAR␥ regulation of PC expression in white and brown adipose tissue (7). We previously studied ZF rats that are obese, hyperlipidemic, and insulin-resistant secondary to a mutated leptin receptor, but are virtually normoglycemic because of compensatory increases in ␤-cell mass and function (26,27). Those studies showed that PC plays a critical role in their ␤-cell adaptive response by preventing dysmetabolism-induced defects in ␤-cell mitochondrial fuel metabolism (27). Also, reduced islet PC expression is found in rodents (28,29) and humans with type 2 diabetes (30).
We investigated whether ␤-cell PC expression is PPAR␥-regulated using an established siRNA for PPAR␥ in INS-1 cells that lowers PPAR␥ mRNA and protein 75% (4). Cells were treated 72 h with or without the PPAR␥ agonist troglitazone. PPAR␥ siRNA reduced PC mRNA band intensity Ͼ60% (Fig. 7A, first and fourth lanes) whereas troglitazone enhanced it 3-fold (second lane); that effect was lost when the PPAR␥ siRNA and troglitazone were used together (third lane).
We previously developed and characterized a diabetic rat model by performing a 60% Px in ZF rats (8). Px ZF rats became hyperglycemic (ϳ15 mM) by 3 weeks after surgery with the same free fatty acid levels and obesity as normoglycemic ZF shams. Additional controls were sham and 60% Px ZL rats that were euglycemic and metabolically indistinguishable. Islet immunoblots from 3 week postsurgery sham ZF rats had a doubling of PPAR␥ protein and its transcriptional target GIP receptor compared with both groups of ZL rats, whereas the diabetic Px ZF rats displayed 20 -30% reductions versus the ZL rats (6), supporting our hypothesis of importance for PPAR␥ signaling in ␤-cell compensation and failure. We predicted a similar pattern for PC and have now assessed PC expression using streptavidin A affinity labeling of SDS-PAGE-resolved islet extracts from 3-week postsurgery Px and sham ZL and ZF rats. As expected, the band intensity in the sham ZF islets was nearly twice that of the ZL rats (171 Ϯ 10%) whereas it was subnormal in the diabetic Px ZF rats (67 Ϯ 12%, Fig. 7B). Comparable results were obtained for PC activity of islet extracts (Fig. 7C), with a 60% increase in sham ZF islets versus ZL rats (p Ͻ 0.001), but no increase in the Px ZF islets (p Ͻ 0.002 sham ZF versus Px ZF).

Nuclear Localization of FoxO1 in ␤-Cells of Hyperglycemic
Rats-We sought an explanation for the markedly lowered expression of PPAR␥ and its target genes in diabetic ZF rats by examining the subcellular distribution of ␤-cell FoxO1 in 3-week postsurgery Px and sham ZL and ZF rats using confocal microscopy (Fig. 8). In both ZL groups, FoxO1 immunoreactivity was chiefly cytoplasmic (the hyperintense staining areas are erythrocytes in capillaries). In sham ZF rats, the FoxO1 staining was both nuclear and cytoplasmic. In contrast, in the Px ZF rats a striking increase in nuclear FoxO1 immunoreactivity was observed. In addition, in both ZL groups and sham ZF rats there was strong nuclear Pdx1 staining, whereas in the Px ZF rats the nuclear Pdx1 signal was markedly reduced. We conclude from these findings that in the diabetic ZF rats a defect in FoxO1 activation and/or translocation causes loss of the adaptive increase in the FoxO1/PPAR␥ pathway that is present in the nondiabetic insulin-resistant ZF rats.

DISCUSSION
The current results have established PPAR␥ expression in ␤-cells is under transcriptional restraint by FoxO1. That conclusion is supported by our having identified a FoxO1 binding site in the mouse PPAR␥ promoter with homologous sequences in rats and humans, and in vitro studies demonstrating FoxO1 binding and inhibition of PPAR␥ expression. Also, transgenic mice with altered FoxO1 expression or defective islet FoxO1 nuclear translocation exhibited changes in islet PPAR␥ expression and parallel findings for its target gene Pdx1. Of particular importance was our demonstrating that PPAR␥ is necessary for GLP-1 mediated augmentation of Pdx-1 expression in vitro, as GLP-1 is an established FoxO1 activator (20). We have also shown that PPAR␥ regulates PC gene expression in ␤-cells, as has previously been characterized using molecular and in vivo techniques in white and brown adipose (7), making it is the third PPAR␥ target gene we have identified in ␤-cells along with the Pdx1 (4, 5) and GIP receptor (6) genes.

FoxO1 and PPAR␥ Signaling in ␤-Cells
These results are especially notable because of the importance of the identified PPAR␥ target genes for regulating ␤-cell development and postnatal ␤-cell mass and function. Pdx1 is an essential transcription factor for pancreas morphogenesis and postnatally is a key regulator of ␤-cell differentiation, maturation, and survival (31). It is required for the normal adaptive increase in ␤-cell mass and function to insulin resistance (32,33). Distal from Pdx1 is a network of genes that impact diverse aspects of ␤-cell signaling, mass, and function (34) including glucokinase with its well known regulatory effects on glycolysis, insulin secretion, and ␤-cell proliferation (35).
PC catalyzes mitochondrial pyruvate conversion to malate. In most cells, the normal pathway for pyruvate metabolism is through pyruvate dehydrogenase with entry into the citric acid cycle for ATP production. However in ␤-cells there is unusually high PC expression so that 50% of pyruvate undergoes carboxylation (36) with entry into the anaplerotic pathways of insulin secretion (37). Obese, insulin-resistant, hyperlipidemic ZF rats normally do not develop diabetes because of compensatory increases in ␤-cell mass and function (27). We previously looked for the Randle effect (inhibition of pyruvate dehydrogenase activation by elevated levels of free fatty acid) in islets from ZF rats and found 30% lowered pyruvate dehydrogenase activity, but, paradoxically, glucose oxidation and glucose-induced insulin secretion were increased. The explanation was increased PC expression with diversion of pyruvate metabolism to the anaplerotic pathways (this study and Ref. 27). We observed the same protective effect in high glucose-cultured rat islets (38). Also, Liu and co-workers showed a role for this enzyme in the ␤-cell mass and functional adaptation to a 60% Px in rats (23,39). As such, PC is believed to play a key role in the ␤-cell adaptation to insulin resistance and dysmetabolism by preserving mitochondrial fuel metabolism (37,40).
The ␤-cell GIP receptor mediates that incretin hormone regulation of insulin secretion and glucose tolerance. GIP receptor gene knock-out mice are glucose-intolerant (41). Also, GIP receptor allelic variations correlate with the heterogeneity in serum glucose and insulin levels after oral glucose challenge in nondiabetic humans (42).
The second aspect of our results relates to evidence that this signaling pathway is enhanced with successful ␤-cell compensation to insulin resistance, whereas defective nuclear translocation of FoxO1 resulting in persistent inhibition of PPAR␥ expression is a feature of failed ␤-cell compensation and hyperglycemia. We previously studied ZF rats made hyperglycemic by a 60% pancreatectomy and observed subnormal islet expression of PPAR␥ and its target gene GIP receptor versus the hyperexpression in nondiabetic ZF rats (6). The current study has shown the same pattern for PC along with reduced nuclear Pdx1 immunostaining in Px ZF rats, which are in accord with our previous report showing subnormal PC and Pdx1 mRNA expression in Px ZF islets (8).
Insight into the impaired PPAR␥ expression with hyperglycemia came with our finding a distinct change in the subcellular localization of FoxO1 in ␤-cells of ZF diabetic rats, specifically, instead of the nuclear and cytoplasmic pattern seen in the con-

FoxO1 and PPAR␥ Signaling in ␤-Cells
AUGUST 30, 2013 • VOLUME 288 • NUMBER 35 trol rats, there was an intense nuclear distribution. This finding is not unique to the Px ZF model, as nuclear FoxO1 and reduced overall islet FoxO1 expression were found in insulin-resistant diabetic mice created by deletion of insulin receptors in GLUT4-expressing tissues (43). Also, below normal islet expression of Pdx1, GIP receptor, and PC is well known to occur in animals (6,28,29,44) and humans (30,45,46) with type 2 diabetes. Based on this collective evidence, we propose that nuclear retention of FoxO1 impairing PPAR␥-mediated defenses against metabolic stresses is an unrecognized feature of failed ␤-cell adaptation.
The mechanism of the defective nuclear to cytoplasmic translocation of FoxO1 with diabetes is not known, as the regulation of FoxO1 activity related to compartmentalization in ␤-cells is multifactorial related to several post-translational modifications and protein interactions (14,18,21). Best known for the ␤-cell is insulin or growth factor PI3K/Akt pathwayinduced phosphorylation causing nuclear export that removes its inhibition of target gene transcription (14). Supporting a defect at this level of the signaling pathway is a study performed in New Zealand obese mice that became diabetic when fed a carbohydrate-rich diet; reductions in islet FoxO1 phosphorylation were noted (47). Furthermore, islets from these mice, and MIN 6 cells, incubated with high glucose and palmitate for 48 h underwent dephosphorylation of FoxO1 along with its upstream kinase Akt (47). Another study reported reduced insulin receptor and insulin receptor substrate 2 expression in islets from humans with type 2 diabetes, and it noted nuclear retention of FoxO1 in islets of mice with a ␤-cell-specific deletion of insulin receptors (48). Alternatively, FoxO1 activity and its subcellular compartmentation in ␤-cells is regulated by other serine/threonine kinases as well the balance between acetylation and deacetylation (21). Recently ␤-cell oxidative stress was linked to hyperexpression of a B55a-containing PP2A holoenzyme that dephosphorylates FoxO1 and promotes its nuclear location (49).
The importance of PPAR␥ in ␤-cells has been controversial. A ␤-cell-specific knock-out mouse for PPAR␥ based on a rat insulin promoter Cre/loxP recombinase system was reported to  ). B, PC expression assessed with streptavidin A affinity labeling of PAGE-resolved islet extracts from 3 weeks after surgery 60% Px and sham ZL and ZF rats. Sham (Sh) ZF rats are obese and insulin-resistant, but normoglycemic because of compensatory increases in ␤-cell mass and function. In contrast, Px ZF rats become hyperglycemic (ϳ15 mM) by 3 weeks after surgery with the same obesity and free fatty acid levels as the normoglycemic ZF shams. As we have seen with other PPAR␥-regulated genes in ␤-cells, the PC band intensity in the sham ZF islets was nearly twice that of the ZL rats and below normal in the diabetic Px ZF rats. Top, representative gel. Bottom, mean Ϯ S.D. (error bars) results of four rats in each group. C, PC enzyme activity measured in islet extracts from 3 weeks after surgery 60% Px and sham ZL and ZF rats. Results are mean Ϯ S.D. of three rats in each group.

FoxO1 and PPAR␥ Signaling in ␤-Cells
be normoglycemic with a standard chow diet and also after fat feeding (50). Another publication from these authors (51) using a tamoxifen-inducible Pdx1 Cre/loxP system that lowered islet PPAR␥ mRNA by 90% found no ␤-cell phenotype after 7 months of normal chow; glucose tolerance, insulin secretion, and ␤-cell mass were unchanged from WT mice as was expression of multiple islet genes including Pdx1, pyruvate carboxylase, and GIP receptor. In addition, 11 weeks of a high fat diet induced the same hyperglycemia, obesity, and insulin resistance in the PPAR␥ knock-out and WT mice. Also, there was no difference in ␤-cell mass or islet gene expression profiles between the fat-fed WT and PPAR␥ knock-out mice although the atypical finding of no effect of fat feeding on these parameters in WT mice makes these latter findings difficult to interpret. The authors concluded that PPAR␥ in ␤-cells is inconsequential. In contrast, we studied noninducible Pdx-1 Cre/loxP PPAR␥-null mice (52) and found hyperglycemia at 8 weeks of age with standard chow along with near total absence of glucose-induced and GIP-induced insulin secretion in isolated islets, and a marked reduction in islet expression of Pdx-1 and GIP receptor (5,6). There was no difference in ␤-cell mass or pancreas histology from control mice (5). As such, this mouse showed a phenotype of functional ␤-cell failure, with an expression profile that agreed exactly with the proposed signaling pathway in this study. Also, others have reported PPAR␥ regulation of gene expression for GLUT2 (53), glucokinase (54), and GPR40 (55) in ␤-cells, plus key modulatory effects of PPAR␥ in ␤-cell endoplasmic reticulum stress related to a cytokine-induced loss of SERCA2 expression have been shown (56,57). In addition, numerous ␤-cell effects have been attributed to PPAR␥ from thiazolidinedione studies in islets, insulin-containing cell lines, and various animal models, although direct proof that the findings are PPAR␥-mediated is mostly lacking. Thus, although the basis for the different ␤-cell phenotypes in the mouse models is unknown, we conclude that there is extensive support for an important regulatory role for PPAR␥ in ␤-cells, especially in the adaptation response to metabolic stresses.
In summary, our results are consistent with a novel mechanism for ␤-cell compensation and failure based on an integral relationship between FoxO1 and PPAR␥, whereby FoxO1 exerts hierarchal control over several key ␤-cell genes, and PPAR␥ is a vital intermediate. FoxO1 is well poised to be a master transcriptional regulator over ␤-cell adaptive responses, as its expression and activity are influenced by numerous stimuli such as nutrient overload, growth factors, incretin hormones, and oxidative stress (13,14). Downstream from PPAR␥ is a network of genes that exert important regulatory control over prodifferentiation processes, incretin effects, glucose and mitochondrial fuel metabolism, and ␤-cell compensation to obesity and insulin resistance (13,27,32,33). We have shown that islet PPAR␥ and its target genes are hyperexpressed in nondiabetic insulin-resistant rats (present study and Refs. 6,8) and rats during the ␤-cell adaptation after a partial pancreatectomy (4, 6), plus we and others have shown impaired expression of these same genes in diabetic animals (6,29,44) and humans with type 2 diabetes (30,45,46). Collectively, these findings support, but do not prove, a FoxO1/PPAR␥-mediated transcriptional network functioning as a core element of how ␤-cells adapt to metabolic stress, with failure of this response causing or exacerbating diabetes.