In Vivo and in Vitro Studies of a Functional Peroxisome Proliferator-activated Receptor γ Response Element in the Mouse pdx-1 Promoter*

We reported that peroxisome proliferator-activated receptor γ (PPARγ) transcriptionally regulates the β-cell differentiation factor pancreatic duodenal homeobox (PDX)-1 based on in vitro RNA interference studies. We have now studied mice depleted of PPARγ within the pancreas (PANC PPARγ-/-) created by a Cre/loxP recombinase system, with Cre driven by the pdx-1 promoter. Male PANC PPARγ-/- mice were hyperglycemic at 8 weeks of age (8.1 ± 0.2 mm versus 6.4 ± 0.3 mm, p = 0.009) with islet cytoarchitecture and pancreatic mass of islet β-cells that were indistinguishable from the controls. Islet PDX-1 mRNA (p = 0.001) and protein levels (p = 0.003) were lowered 60 and 40%, respectively, in tandem with impaired glucose-induced insulin secretion and loss of thiazolidinedione-induced increase in PDX-1 expression. We next identified a putative PPAR-response element (PPRE) in the mouse pdx-1 promoter with substantial homology to the corresponding region of the human PDX-1 promoter. Electrophoretic mobility supershift assays with nuclear extracts from β-cell lines and mouse islets, also in vitro translated PPARγ and retinoid X receptor, and chromatin immunoprecipitation analysis demonstrated specific binding of PPARγ and retinoid X receptor to the human and mouse pdx-1 × PPREs. Transient transfection assays of β-cells with reporter constructs of mutated PPREs showed dramatically reduced pdx-1 promoter activity. In summary, we have presented in vivo and in vitro evidence showing PPARγ regulation of pdx-1 transcription in β-cells, plus our results support an important regulatory role for PPARγ in β-cell physiology and thiazolidinedione pharmacology of type 2 diabetes.

Peroxisome proliferator-activated receptor ␥ (PPAR␥) 4 is a member of the nuclear hormone receptor superfamily of ligand-inducible transcription factors (1) and contributes significantly to diverse biological processes such as glucose homeostasis, lipid metabolism, cellular proliferation, and differentiation (2)(3)(4). PPAR␥ regulates transcriptional activity of target genes by forming a heterodimer with retinoid X receptor (RXR), and binding to a specific PPAR␥ response element sequence (PPRE) within the promoter region (5,6). The PPRE consists of two hexamer repeats (DR1 and DR2) separated by a single nucleotide, with PPAR␥ and RXR occupying the 5Ј-and 3Ј-half-sites, respectively (7,8). Thiazolidinediones are PPAR␥ agonists used for the treatment of type 2 diabetes based on insulin-sensitizing effects in adipose tissue and muscle (9). However, a recent finding is PPAR␥-mediated anti-proliferation and pro-survival of islet ␤-cells (10 -12). This is of interest as clinical trials showing glycemic benefits of thiazolidinediones in prediabetes and early type 2 diabetes (13)(14)(15)(16)(17) have usually been interpreted that the insulin sensitization effect unloads overstimulated insulin secretion, so-called "␤-cell rest." On the other hand, the possibility of direct PPAR␥ regulatory effects in islet ␤-cells is controversial, as a ␤-cell-specific PPAR␥ knock-out mouse was normoglycemic basally and after fat feeding (11). Also little is known about PPAR␥ target genes in ␤-cells.
PPAR␥ regulates the function and survival of tissues by acting through a triad of effects as follows: anti-proliferation, prosurvival, and pro-differentiation (18). Because PPAR␥-mediated pro-survival and anti-proliferation had been reported for ␤-cells (10 -12), one might also expect pro-differentiation properties. Our group has studied a model of ␤-cell adaptation to a loss of ␤-cell mass, 60% pancreatectomy (Px) Sprague-Dawley rats. They are normoglycemic because of partial ␤-cell regeneration during the 1st week post-Px (19,20) that is followed by ␤-cell hyperfunction secondary to increased glucokinase activity (21). We investigated the transition phase, 14 days post-Px, to determine how these events are coordinated, and we found increased nuclear PPAR␥ expression in isolated islets at mRNA and protein levels (22). We also noted increased nuclear expression of pancreatic duodenal homeobox (PDX)-1 that is an essential regulator of the function and survival of mature ␤-cells (23)(24)(25)(26)(27)(28). We then studied INS-1 ␤-cells following RNAiinduced 75% knockdown of PPAR␥, finding PDX-1 mRNA and protein levels were lowered 80 and 60%, respectively. Based on these collective results, we proposed that pdx-1 is a physiologically regulated target gene for PPAR␥ in ␤-cells (22).
We now set out to verify these findings in vivo by studying mice with a pancreas-specific ablation of PPAR␥ based on a Cre/loxP recombinase system with Cre driven by the pdx-1 promoter (29). Our findings confirmed the expected reduction of PDX-1 expression in isolated islets. Moreover, these mice were characterized by glucose intolerance, impaired glucoseinduced insulin secretion, and loss of the pharmacological effect of thiazolidinediones to up-regulate PDX-1 expression. In addition, we have identified and characterized a novel functional PPRE within the mouse and human pdx-1 promoter regions.

EXPERIMENTAL PROCEDURES
Animal Studies-Mice with PPAR␥ deficiency restricted to pancreatic islets, ducts, and acini (PANC PPAR␥ Ϫ/Ϫ ) were generated from crossing pdx-1-Cre mice (original source D. A. Melton, described in Ref. 30) and mice with two floxed PPAR␥ alleles as detailed previously (29). Controls were littermate Cre negative PPAR␥ floxed mice. All protocols were in accordance with the principles of laboratory animal care and were approved by the Institutional Animal Use and Care Committee of the University of Vermont. At 8 weeks of age, tail vein blood sampling for blood glucose (Freestyle glucose monitor) was performed from nonanesthetized normally fed mice. Some animals underwent an intraperitoneal glucose tolerance test that consisted of an intraperitoneal injection of 2 g/kg glucose with blood glucose measured at 0, 30, 60, 90, and 120 min. Islets were isolated by pancreas duct infiltration with collagenase, Histopaque gradient separation, and hand picking.
Pancreas Histology and Morphometrics-Pancreata were rapidly excised, cleared of fat and lymph nodes, and blotted before immersion-fixing overnight in 4.0% paraformaldehyde in 0.1 mM phosphate buffer at 4°C. After washing in several changes of phosphate-buffered saline, tissues were dehydrated and embedded in paraffin. Pancreas sections were incubated overnight at 4°C in antibody mixtures of sheep anti-amylase (Biogenesis), guinea pig anti-insulin (Linco), and rabbit antiglucagon (Linco). ␤-Cell mass was quantified using a computerized planimetric method and ␤-cell proliferation frequency using Ki-67 as the cell cycle marker, using methods detailed previously (31).
Islet Insulin Secretion-Freshly isolated islets were incubated 1 h in RPMI 1640 media supplemented with 10% fetal calf serum, 10 mM HEPES, 2 mM L-glutamine, 1 mM sodium pyruvate, 100 units/ml penicillin, 100 g/ml streptomycin, 11.1 mM glucose, at 37°C in a 5% CO 2 incubator. Duplicate batches of 15-25 islets were preincubated in 12-well plates for 2 h in RPMI 1640 medium at 2.8 mM glucose and then 1 h in KRBH plus 0.5% bovine serum albumin and 2.8 mM glucose. Insulin secretion was assessed during a subsequent 1-h incubation in KRBH, 0.5% bovine serum albumin at 2.8 or 16.7 mM glucose at 37°C, followed by ultrasensitive mouse insulin enzyme-linked immunosorbent assay (Mercodia) of the medium. Islet insulin content was measured after acid-ethanol (0.2 mM HCl in 75% ethanol) extraction, and the insulin secretion results were normalized to the total insulin content.
Islet Culture with Troglitazone-Isolated islets were incubated in culture media supplemented with 10 M troglitazone or vehicle (DMSO) for 72 h with the media changed each 24 h. After the incubation period, islets were lysed and immunoblots were performed (20 g) using the technique described above, with rabbit anti-PDX-1 (Chris Wright, Vanderbilt University) followed by reprobing with anti-␤-actin (Sigma).
Islet PCR and Immunoblots-Islet total RNA was extracted using the RNeasy Micro kit plus single step on-column DNase digestion (Qiagen). cDNAs were synthesized using 500 ng of extracted RNA with ImProm-II reverse transcriptase (Promega), dNTPs, and random hexamer primers. PCR analyses were carried out in a PTC-200 Peltier Thermal Cycler (MJ Research) using the cDNAs together with TaqDNA polymerase (Promega) and primer combinations (sequences available on request). The thermal cycle program was denaturing step at 95°C for 2 min followed by 35 cycles for PPAR␥, PPAR␣, and glucagon or 25 cycles for PDX-1, GLUT2, INS-1, and INS-2, at 94°C for 15 s, 56°C for 30 s, and 72°C for 60 s, with extension step of 5 min at 72°C. ␣-Tubulin and cyclophilin A were used as internal controls. Gel images were captured using the Gel Doc EQ documentation system (Bio-Rad) and analyzed using ImageJ software (National Institutes of Health), with the results expressed relative to the control genes.
PDX-1 and ␤-actin immunoblots were performed as detailed previously. Immunoblot and PCR results are expressed as mean Ϯ S.E. from three pairs of littermate floxed control and PANC PPAR␥ Ϫ/Ϫ mice. Statistical significance was determined by Student's t test.
Chromatin Immunoprecipitation Assay-Chromatin immunoprecipitation assay was performed with mouse-derived ␤TC6 cells using ChIP-IT kit (Active Motif). Cells were fixed using formaldehyde solution, and the chromatin was sheared by enzymatic digestion according to the instruction manual to DNA fragments that averaged 300 -500 bp in length. Mouse monoclonal antibody against PPAR␥ (E8, Santa Cruz Biotechnology) was added to aliquots of precleared chromatin and incubated overnight, with parallel samples incubated with the negative-control IgG provided with the kit. Protein G-agarose beads were added, and the mixture was incubated for 1.5 h at 4°C. After reversing the cross-links, DNA was isolated, and PCRs were performed with primers for the mouse pdx-1 promoter region PPRE (GenBank TM accession number AF192495). Primer sequences were as follows: FW 5Ј-ACAC-ACTCACTCACTCACTCATTGGG-3Ј; REV 5Ј-CTGAGAT-ACCCAGCCATTAGGCAAGA-3Ј (expected PCR product 312 bp). A second primer pair was as follows: FW 5Ј-CAATC-TAGTCCAAACCAGCCTTTGGC-3Ј; REV 5Ј-TGAGATAC-CCAGCCATTAGGCAAGAG-3Ј (expected PCR product 265 bp). As negative control, a primer pair (FW 5Ј-GCGCTGAGT-TCTGCAAGCATTTCT-3Ј; REV 5Ј-CGCGAACACCTGCA-CTTGTTTCAA-3Ј) was selected that amplified a stretch of 450 bp of DNA 2.56 kb downstream of the mouse pdx-1 ϫ PPRE. PCR conditions were 1 cycle of 94°C for 3 min, 39 cycles of 94°C for 30 s, 62°C for 1 min, 72°C for 1 min. A mouse-specific positive control for appropriate shearing of DNA and co-immunoprecipitation was performed using a kit (Active Motif) based on binding of transcription factor EFI-␣ by anti-RNA pol II.
Luciferase Reporter Gene Assay-80 -90% confluent INS-1 cells in a 6-well format were incubated overnight in antibioticfree media. Cells were transfected using Lipofectamine 2000 transfection reagent (Invitrogen) with 2 g of pTAL empty vector (firefly luciferase vector) or equivalent wild type or mutated pTAL ϫ PPRE pdx-1 promoter vector. Renilla luciferase reporter plasmid (pRL-TK, Promega) was included (0.05 g) in all transfections as internal control. Twenty four h post-transfection, cells were incubated another 24 h with 10 M troglitazone or DMSO, and the luciferase assay performed in a TD 20/20 luminometer (Turners Design) using dual luciferase assay kit (Promega). Firefly luciferase activity was normalized with Renilla luciferase and expressed as relative luciferase activity. Each experimental condition was performed in triplicate with the results calculated as relative (%) luciferase value of the wild type PPRE controls.

RESULTS
PANC PPAR␥ Ϫ/Ϫ Mice-We studied mice with PPAR␥ deficiency within the pancreas created by a Cre/loxP recombinase system, with Cre driven by the pdx-1 promoter (29). PDX-1 is highly expressed in all pancreatic epithelial cells during embryogenesis, with continued high expression in ␤-cells postnatally as opposed to a marked reduction in other pancreatic tissues (32). Thus, the pdx-1-driven Cre creates a PPAR␥ knock-out in ␤-cells although there remains the possibility of additional effects related to PPAR␥ deletion in ducts, acini, and other islet cells. The original description of these mice reported increased pancreas weight, with larger islets and normal pancreas histology by gross inspection, com-pared with control mice (29). We studied 8 -10-week-old male animals and confirmed the pancreas specificity of the PPAR␥ deletion (Fig. 1A). We also found no abnormal pancreas histological features, as islet cytoarchitecture and insulin and glucagon immunostaining intensities were indistinguishable from the controls (Fig. 1B). In contrast, there was no difference in pancreas weight, ␤-cell mass, or ␤-cell proliferation rate between the PANC PPAR␥ Ϫ/Ϫ and the floxed control mice (Fig. 1C). PANC PPAR␥ Ϫ/Ϫ mice had the same body weight (29 Ϯ 1 g versus 28 Ϯ 1 g, n ϭ 7) as the controls, but unexpectedly, nonfasting glucose values were higher (8.1 Ϯ 0.2 mM versus 6.4 Ϯ 0.3 mM, n ϭ 4, p ϭ 0.009) and there was glucose intolerance during an intraperitoneal glucose tolerance test (Fig. 1D).
Isolated Islet Studies-Analysis of mRNA levels from isolated islets confirmed the ablation of PPAR␥ expression (p Ͻ 0.001) as opposed to PPAR␣ that was unaffected, demonstrating the specificity of the gene inactivation ( Fig. 2A). As predicted, islet PDX-1 mRNA (43 Ϯ 6% of control islets, n ϭ 3, p ϭ 0.001) and protein levels (58 Ϯ 2% of control islets, n ϭ 3, p ϭ 0.003) were markedly lowered in the PANC PPAR␥ Ϫ/Ϫ mice compared with the controls (Fig. 2, A and B, respectively). GLUT2 mRNA also was lowered (p ϭ 0.006) as opposed to no change in glucagon, insulin 1, or insulin 2 mRNA levels.
Functional assessment showed a near-total absence of glucoseinduced insulin secretion in the PPAR␥ Ϫ/Ϫ islets, based on no statistical difference in the in vitro insulin response to 16.7 mM glucose versus 2.8 mM glucose (1.3 Ϯ 0.1-fold), compared with a 2.5 Ϯ 0.3-fold increase in the control islets (Fig. 2C). Also, control islets cultured for 3 days with troglitazone had a near-doubling of Pdx-1 levels. In contrast, troglitazone induced no change of the lowered PDX-1 level in PPAR␥ Ϫ/Ϫ islets (Fig. 3).
PPRE Sequence on Mouse and Human pdx-1 Promoters-We next investigated the molecular basis for PPAR␥ regulation of   pdx-1 transcription. A PPRE consists of hexamer repeats (DR1 and DR2) separated by one nucleotide, with the consensus sequence being RGGTCA-A-AGGTCA. However, a characteristic feature is a high degree of homology for the 3Ј-half-site but substantial variance for the 5Ј-half-site (33), reflecting highly efficient RXR binding to the 3Ј DR2 half-site as part of the PPAR␥/RXR heterodimer as compared with PPAR␥ binding to the 5Ј DR1 half-site (34). As such, we searched for a PPRE on the mouse pdx-1 promoter using the DR2 consensus sequence to probe for a homologous area within the 4.531-kb mouse pdx-1 promoter region (Gen-Bank TM accession number AF192495). A putative PPRE consisting of GGGTAC-C-TCCAGT was identified at positions Ϫ2716/Ϫ2698 relative to the transcription initiation site in area 1 of the mouse pdx-1 promoter (Table 1A) as characterized by Gerrish et al. (35). The identified DR2 element was fully homologous to the consensus sequence. In contrast, we compiled a data base of known PPREs (data not shown), and we noted the DR1 sequence was unique from reported PPREs. Comparison of the putative mouse pdx-1 ϫ PPRE with the analogous human PDX-1 promoter sequence revealed partial homology (a three-nucleotide mismatch) for the DR-1 element and full homology for the DR-2 (Table 1, part B).
Electrophoretic Mobility Shift and Mutational Studies-We determined whether the putative mouse and human pdx-1 ϫ PPREs bind PPAR␥ using nuclear extracts of isolated mouse islets, mouse-derived ␤TC6 cells, and rat-derived INS-1 cells, incubated with 32 P-labeled doublestranded oligos of the mouse (Fig. 4A) and human (Fig. 4B) pdx-1 ϫ PPRE sequences. An oligo for the PPRE of acyl-CoA oxidase (36) was included as internal control. In all tissues, complexes were apparent with the pdx-1 ϫ PPREs that migrated at an identical position as the acyl-CoA oxidase PPRE complex. The specificity of the PPAR␥⅐PPRE complex was confirmed using PPAR␥ antibody that supershifted the mouse pdx-1 ϫ PPRE band (Fig. 4C, lane D) in the same fashion as the PPRE in acyl-CoA oxidase (Fig. 4C, lane B). We also determined the ability of the mouse pdx-1, human PDX-1, and acyl-CoA oxidase PPRE probes to bind in vitro translated PPAR␥ and RXR-␣, alone and in combination (Fig. 5, A and B). The translated PPAR␥ and RXR-␣ proteins individually and in combination resulted in a complex with identical mobility using all three probes (Fig. 5A, lanes 3-5 and 10 -12; Fig. 5B, lanes 2-4 and 8 -10). Analysis by immunoblot demonstrated the reticulocyte lysate used for programming the in vitro translated proteins contained residual amounts and then islet lysates underwent PDX-1 immunoblotting. Membranes were stripped and reprobed with ␤-actin antibody. A representative gel is shown, and the graph contains the mean Ϯ S.E. band intensities from the three separate experiments expressed as % intensity of each experimental condition compared with Cre Ϫ islets incubated with DMSO. NS ϭ not significant.

TABLE 1 PPRE within the mouse and human pdx-1 promoters
A, partial sequence of the mouse pdx-1 promoter region (GenBank TM accession number AF192495). The underlined region represents the putative PPRE and complementary sequence. B, comparison of PPRE sequence of mouse pdx-1 promoter to the analogous sequence stretch of area 1 within the human PDX-1 promoter (35).
of PPAR␥ and RXR-␣ accounting for the complexes observed in Fig. 5. Addition of PPAR␥ antibody or N terminus RXR-␣ antibody supershifted the complex. In contrast, pan-RXR antibody to the DNA and ligand binding domain of RXR protein, but lacking the N terminus, prevented formation of the complex, indicating that it interfered with PPAR␥ and RXR-␣ heterodimerization and DNA binding.
Specificity of binding was shown with mutational studies, in which up to a 20-fold excess of unlabeled oligos that contained mutations in the DR1 half-site of the mouse pdx-1 ϫ PPRE, DR2 half-site, and combined DR1-DR2 mutations failed to elicit competition as reflected in no change in band intensity (Fig. 6, B-D) in contrast to the wild type probe (Fig. 6A). In a reverse competition experiment, increasing concentrations of unlabeled acyl-CoA oxidase PPRE successfully competed with the radiolabeled mouse pdx-1 ϫ PPRE probe (Fig. 6E).
ChIP Assay-Binding of PPAR␥ to the pdx-1 ϫ PPRE in intact ␤-cells was confirmed using the ChIP assay. Preparations of 300 -500-bp chromatin fragments from mouse-derived ␤TC6 cells underwent immunoprecipitation with PPAR␥ monoclonal antibody or negative control IgG, and PCR of the immunoprecipitated DNA and input DNA using mouse pdx-1 ϫ PPRE region flanking primers. The expected 312-bp PCR product was generated with both input DNA and PPAR␥ antibody-immunoprecipitated DNA from two separate chromatin preparations, whereas only a very faint PCR product was observed from the negative control IgG-immunoprecipitated DNA (Fig. 7A). In contrast, a nonspecific primer pair for a 450-bp PCR product 2.56 kb downstream of the mouse pdx-1 ϫ PPRE showed a PCR product only with input DNA (Fig. 7B). The positive control showed the expected 233-bp PCR product with the EFI primer pair in the RNA pol II antibody-immunoprecipitated DNA but not the negative control IgG-immunoprecipitated DNA (Fig. 7C). Also, the negative control primers generated the expected 245-bp PCR product only with input DNA (Fig. 7D). Luciferase Reporter Gene Assay-The 1-kb Pst-Bst (Ϫ2916to Ϫ1920-bp) fragment of the mouse pdx-1 promoter region carrying the PPRE was subcloned into the pTAL luciferase   NOVEMBER 21, 2008 • VOLUME 283 • NUMBER 47 reporter vector and used as template to perform site-directed mutagenesis to incorporate the same mutations on the DR1 and DR2 regions as used for the in vitro gel shift competition assay. As shown in Fig. 8, both mutations lowered basal luciferase reporter activity (Mut 1 73 Ϯ 4% of wild type pdx-1 ϫ PPRE, p Ͻ 0.001; Mut 2 56 Ϯ 5% of wild type pdx-1 ϫ PPRE, n ϭ 3, p Յ 0.001). Also, 24 h of stimulation with the PPAR␥ agonist, troglitazone, of the cells transfected with the wild type pdx-1 ϫ PPRE reporter construct induced a 4-fold increase in luciferase reporter activity (438 Ϯ 15% of nontroglitazone-treated wild type, p Ͻ 0.001) that was lowered to less than 2-fold with both mutations (p Ͻ 0.001 for troglitazone-treated wild type versus troglitazone-treated Mut 1 and troglitazone-treated Mut 2, respectively). These findings confirmed physiological activity of the identified pdx-1 ϫ PPRE in INS-1 cells basally, and with induction by a PPAR␥ agonist.

DISCUSSION
This study stems from our work in the 60% pancreatectomy rat model of ␤-cell adaptive compensation, in which an initial period of partial ␤-cell regeneration is followed by enhanced ␤-cell function, so normoglycemia is preserved (19 -21). We investigated the transition between these phases of adaptation, and we identified heightened ␤-cell nuclear expression of PPAR␥ and the ␤-cell differentiation factors PDX-1 and NKX6.1 (22). The known effects of these factors in ␤-cells (anti-proliferation and pro-survival for PPAR␥ (10 -12), pro-survival and enhanced insulin secretion for PDX-1 (23)(24)(25)(26)(27)(28), and suppressed glucagon expression plus enhanced glucose-induced insulin secretion with NKX6.1 (37)) could explain the transition. On the other hand, PPAR␥ acts in tissues through a triad of effects, antiproliferation, pro-survival, and pro-differentiation (18), and PPAR␥-mediated proliferation arrest and pro-survival effects were described for ␤-cells (10 -12). As such, another possibility was PPAR␥-mediated regulation of these ␤-cell differentiation factors. We focused on PDX-1, as it is the most studied transcription factor in mature ␤-cells and is a crucial regulator of ␤-cell function (23)(24)(25), viability (26), and compensatory capacity (38,39). Support for PPAR␥-mediated regulation of PDX-1 expression was obtained in INS-1 cells by showing a tripling of PDX-1 expression with the PPAR␥ agonist, troglitazone, and a lowering of pdx-1 transcription following RNAiinduced reduction of PPAR␥ expression (22). The current results have confirmed in vivo PPAR␥ regulation of pdx-1 transcription in mouse ␤-cells, with 40% of PDX-1 expression being PPAR␥-dependent. They also have shown an absolute requirement for pancreas PPAR␥ expression in terms of normal ␤-cell function (glucose-induced insulin secretion) and whole animal glucose tolerance. Of potential clinical importance, we have confirmed in primary mouse ␤-cells the action of thiazolidinediones to augment ␤-cell PDX-1 expression that we had previously seen in INS-1 cells (22). Collectively, this panoply of findings supports a key role of PPAR␥ to regulate PDX-1 expression in ␤-cells physiologically and pharmacologically,  500 -600-bp chromatin preparations of BTC6 cells were prepared as described under "Experimental Procedures." A, they were immunoprecipitated using mouse monoclonal PPAR␥ and negative control IgG, followed by PCR of the immunoprecipitated and nonimmunoprecipitated DNA (input DNA) using flanking primer pairs to mouse pdx-1 ϫ PPRE. The shown bands are 312-bp (expected length) PCR product from two separate experiments, along with absence of PCR product in the control IgG lanes on the right. B, negative control was performed using primer pairs for a 450-bp area 2.7 kilobases downstream of the mouse pdx-1 ϫ PPRE. The expected length band was obtained with input DNA but not the PPAR␥-immunoprecipitated DNA. C, as a positive control, in parallel chromatin preparations were precipitated with RNA pol II antibody and underwent PCR of the immunoprecipitated and input DNA using EFI primer pairs. The shown bands are 250-bp (expected length) PCR product from two separate experiments in the RNA pol II antibody and input DNA lanes compared with the expected absence of a PCR product in the control IgG lanes. D, negative control for the RNA pol II immunoprecipitated and input DNA using the negative control primer pairs for RNA pol II. and thus to have pro-differentiation effects on ␤-cell function and survival.
An unexpected observation was the different phenotype of the PANC PPAR␥ Ϫ/Ϫ mice versus that reported for mice with a ␤-cell-specific knock-out of PPAR␥ using a Cre/loxP recombinase system based on the rat insulin promoter (RIP) (11). The ␤-cell phenotype of that model consisted of twice normal pancreatic ␤-cell mass because of loss of the PPAR␥ anti-proliferation effect, and blunting of the thiazolidinedione-induced augmentation of glucose-induced insulin secretion in isolated islets, confirming PPAR␥ regulation of ␤-cell mass and function (11). Regardless, the normoglycemia of these mice basally and after fat feeding led the investigators to conclude those observations were inconsequential. This study investigated mice with Cre expression driven by the pdx-1 promoter that creates PPAR␥ deletion in ␤-cells, but also with the possibility of impacting other pancreatic epithelial cells such as non-␤ islet cells, ducts, and acini (29). The potential for this additional effect is most pronounced during fetal development when PDX-1 is expressed in all pancreatic epithelial cells, whereas post-birth, PDX-1 is expressed at high levels only in ␤-cells (32). In contrast to the RIP-Cre PPAR␥ knock-out mice, male pdx-1-Cre PPAR␥ knock-out mice were hyperglycemic on a normal chow diet at 8 weeks of age. The basis for this disparity is not fully understood, although concern over the rat insulin II promoter for Cre expression has been raised because of modest hypothalamic expression of the transgene (32) and glucose intolerance of RIP-Cre mice with no loxP gene deletion (40). In contrast, this line of pdx-1-Cre mice is known to have no defect in glucose tolerance (41). Indeed there is an ongoing debate over Cre/loxP technology for ␤-cell gene deletions as reflected in a recent discussion of the different phenotypes for mice with leptin deletions using RIP and pdx-1 Cre promoters (42). We considered the possibility that the whole pancreas epithelial deletion of PPAR␥ had caused developmental abnormalities that were responsible for the hyperglycemia. However, in this current study and a previous analysis of this model (29), no abnormal pancreatic histology was noted, plus we measured ␤-cell mass and found no reduction compared with the control mice. Alternatively, we had concern over the potential for enhanced glucagon expression in the PANC PPAR␥ Ϫ/Ϫ mice, as PPAR␥ is highly expressed in islet ␣-cells (43,44) and acts to repress glucagon transcription (37). However, there was no change in islet glucagon expression by mRNA analysis or immunohistochemistry. Furthermore, PDX-1 expression is lacking in adult ␣-cells. Rather, the results in the PANC PPAR␥ ϩ/Ϫ mice agree with our prior in vitro results in terms of PPAR␥ regulation of ␤-cell PDX-1 expression (22). Unanswered is why the PANC PPAR␥ ϩ/Ϫ mice failed to show the ␤-cell hyperproliferation and larger ␤-cell mass of the RIP-Cre PPAR␥ null mice (11), although those measurements were performed in PANC PPAR␥ ϩ/Ϫ mice that already were hyperglycemic. Another issue for consideration is the phenotype of the PANC PPAR␥ ϩ/Ϫ mice is more severe than that reported for PDX-1 haploinsufficient mice, and they also have glucose intolerance, but glucose-induced insulin secretion is unimpaired in isolated islets (25,26), leading us to speculate there are additional PPAR␥ regulatory effects in ␤-cells that are PDX-1-independent.
We next investigated the molecular basis for PPAR␥ regulation of pdx-1 transcription. A typical PPRE consists of two hexamer repeats, DR1 and DR2, separated by a nucleotide, with a consensus sequence of RGGTCA-A-AGGTCA. However, there is considerable divergence in the DR1 sequence, with a comparison of 73 reported functional PPREs showing only 2 with the consensus sequence, and 8% having five or more mismatches (33). In contrast, 20 of 73 matched the ideal DR2, and another 30 had only 1 nucleotide mismatch. These observations are consistent with the greater importance of RXR binding to the DR2 half-site as part of the PPAR␥/RXR heterodimer in terms of PPRE functionality, compared with PPAR␥ binding to the DR1 half-site, leading to the suggestion that the PPRE consensus sequence may need modification (34). We searched the mouse pdx-1 promoter region for homology to the DR2 consensus sequence, and found a site (CCCATG-G-AGGTCA, Ϫ2720 to Ϫ2708) that matched the DR2 consensus sequence. In contrast, comparison of the associated DR1 hexamer to reported PPREs showed its sequence was unique. Notably, the so-called area I encompassing the DNase I-hypersensitive site I within the mouse pdx-1 promoter is a regulatory site for ␤-cellspecific expression of PDX-1 (35), and our identified sequence was located in this area. Also, the human sequence of this area is 89% homologous with the mouse sequence and plays a similar regulatory role for PDX-1 expression (35). Investigation of the analogous human PDX-1 sequence showed partial homology for the DR1 element and 100% homology for DR2 compared with the putative mouse pdx-1 ϫ PPRE.
We went on to show binding of radiolabeled oligos for the mouse and human pdx-1 ϫ PPREs with nuclear extracts from mouse islets, mouse-derived ␤TC6 cells, and rat-derived INS-1

Functional PPRE in pdx-1 Promoter
cells. Supershift analysis with PPAR␥ antibody confirmed formation of a PPAR␥⅐PPRE complex. Binding studies with in vitro translated PPAR␥ and RXR-␣ showed a heterodimerized complex binds to the mouse and human pdx-1 ϫ PPREs. Binding specificity of PPAR␥ to the mouse PPRE was shown in vitro and in intact cells by competition assay and chromatin immunoprecipitation, followed by showing its capacity to regulate pdx-1 transcription. The latter entailed making DR1 and DR2 mutations for a luciferase reporter gene assay, in the absence and presence of 24 h of troglitazone treatment. As expected, troglitazone enhanced reporter activity of the wild type pdx-1 ϫ PPRE 4-fold, and this increase was lowered more than half with the mutated PPREs. Thus, we have identified and characterized a functional PPRE within the mouse pdx-1 promoter that has a unique sequence from previously reported PPREs and is highly conserved in humans.
In summary, we have provided in vivo and in vitro evidence showing PPAR␥ regulation of pdx-1 transcription, and thus indirectly ␤-cell function and mass. Also, we have shown in primary ␤-cells that thiazoledionediones increase PDX-1 expression through a PPAR␥-mediated mechanism. Given the crucial role of PDX-1 in the function (23)(24)(25), survival (26), and compensatory ability (38,39) of mature ␤-cells, along with the ␤-cell benefits attributed to thiazoledinediones in prediabetes and type 2 diabetes (13-17), our results strongly support an important regulatory role for PPAR␥ in ␤-cell physiology and disease pharmacology.