Differentiation of early foregut endoderm into pancreatic endocrine and exocrine cells depends on a cascade of gene activation events controlled by various transcription factors. Prior in vitro analysis has suggested that the forkhead/winged helix transcription factor Foxa2 (formerly HNF-3β) is a major upstream regulator of Pdx1, a homeobox gene essential for pancreatic development. Pdx1 is also essential for the maintenance of glucose homeostasis, as its human orthologue, IPF-1, is mutated in a subset of patients with early-onset type 2 diabetes (MODY4). To analyze the Foxa2/Pdx1 regulatory cascade during pancreatic β-cell differentiation, we used conditional gene ablation of Foxa2 in mice. We demonstrated that the deletion of Foxa2 in β-cell−specific knockout mice results in downregulation of Pdx1 mRNA and subsequent reduction of PDX-1 protein levels in islets. These data represent the first in vivo demonstration that Foxa2 acts upstream of Pdx1 in the differentiated β-cell.
The mammalian pancreas is comprised of exocrine, endocrine, and ductal cell types that are of endodermal origin. During mouse development, dorsal and ventral pancreatic primordia first appear as evaginations of the foregut endoderm on day 9 post coitum (p.c.). Prepancreatic endoderm is patterned in a stepwise manner, first by signals from mesoderm/ectoderm on day 7.5 p.c., and then by signals from the notochord during the next day (1,2). Shortly thereafter, endothelial cells provide additional signals to induce the prepatterned endoderm to differentiate into insulin-expressing β-cells (3). The exocrine pancreas is composed of a ductal system and a relatively homogenous population of acinar cells that secrete digestive enzymes. The endocrine compartment is organized into clusters of cells called islets of Langerhans that differentiate into four endocrine cell types—the α, β, δ, and pancreatic polypeptide (PP) cells—which express glucagon, insulin, somatostatin, and pancreatic polypeptide, respectively (4,5).
The development of the endocrine compartment is controlled by a hierarchy of transcription factors that include Isl1, Nkx2.2, Pax4, Pax6, and NeuroD/BΕΤA2 (6–8). These factors are expressed during the early stages of pancreas development, and function at several levels of endocrine cell differentiation. However, the initial steps of pancreatic development are not perturbed in mice homozygous for null mutations in the corresponding genes (6–8).
In contrast, the homeodomain protein PDX-1 is not only important for the differentiation of β-cells, but is also essential for the initial development of the pancreas. Pancreatic differentiation is arrested at a very early stage in mice homozygous for a null mutation in the Pdx1 gene, and pancreatic development is also severely affected in humans with homozygous mutation of the insulin promoter factor 1 (IPF-1) gene, the human orthologue of Pdx1 (9–11). Despite the dramatic phenotype of Pdx1 mutant mice, evagination of the foregut epithelium and formation of the dorsal pancreatic bud still occur. Likewise, the expression of early pancreatic markers is unaffected, suggesting the existence of additional genes upstream of Pdx1 involved in the regulation of the earliest stages of pancreatic development (9–12). Therefore, Pdx1 appears to act downstream of the initial specification of the gut endoderm to a pancreatic fate.
Recent evidence suggests that the forkhead/winged helix transcription factor Foxa2 (formerly HNF-3β) is a key regulator of foregut development that may play an essential role in the cell type−specific transcription of the Pdx1 gene in the pancreas (13,14). Foxa2 is expressed in the foregut endoderm before and at the onset of pancreatic development and persists to adulthood, where it is expressed throughout the islet and in acinar cells (14–16). Conserved FOXA (HNF-3) binding sites were identified in the promoter/enhancer of the Pdx1 gene in mice, rats, and humans, and the presence of FOXA-2 in the binding complexes was confirmed by EMSA (electrophoretic mobility shift assay) using nuclear extracts from multiple cell lines and primary pancreatic islet cell cultures (13–15,17,18). A transgene containing an enhancer fragment including the FOXA binding site fused to a β-galactosidase reporter directs islet-specific expression in transgenic mice (14). In addition, the FOXA binding site along with a NeuroD/BETA2 element form a composite enhancer in the rat Pdx1 gene that mediates the synergistic induction of Pdx1 promoter activity by Foxa2 and BETA2 in transfected HIT-T15 cells (13). When Pdx1 promoter activity is repressed by glucocorticoids in HIT-T15 cells, this repression is reversed by overexpression of FOXA-2 (13).
Although the presence of a FOXA binding site in the Pdx1 promoter in mice, rats, and humans suggests that FOXA-2 is a transcriptional activator of Pdx1, the contribution of Foxa2 to the regulation of Pdx1 in pancreatic β-cells in vivo is unknown. There are now several examples where predictions about regulatory circuits involving Foxa genes derived from in vitro studies were not borne out when examined in vivo (19–22). Analysis of the regulation of Pdx1 by Foxa2 in Foxa2-null mice is precluded by the fact that these mice die shortly after gastrulation because of the essential function of this gene in notochord and yolk sac development (23,24). To circumvent this limitation, mice lacking Foxa2 specifically in pancreatic β-cells were generated using the Cre-loxP recombination system (25). We have previously shown that the FOXA-2 protein is deleted in 85% of β-cells on postnatal day 8 (P8) in these pancreatic β-cell−specific Foxa2 knockout mice (Foxa2loxP/loxP; Ins.Cre mice). In this study, we used these mice to investigate the regulation of Pdx1 by Foxa2 in β-cells in vivo.
RESEARCH DESIGN AND METHODS
The derivation of the Foxa2loxP/loxP mutant mouse and the Ins.Cre transgenic line has been previously reported (22,26,27). All mice were kept on a mixed outbred CD-1 background. Genotyping was performed by PCR analysis using genomic DNA isolated from the tail tip of newborn mice (22). We focused our studies on P8 mice, because mutant mice usually succumb to severe hypoglycemia 9–12 days after birth (25).
Tissues were fixed in 4% paraformaldehyde overnight at 4°C, embedded in paraffin, sectioned to 6 μm thickness, and applied to Probe-on Plus slides (Fisher Scientific). Deparaffinized and rehydrated slides were subjected to microwave antigen retrieval by being boiled for 6 min in a 10 mmol/l citric acid buffer (pH 6.0) and allowed to cool for 10 min at room temperature. Slides were washed in PBS and then blocked with protein blocking reagent (Immunotech) for 20 min at room temperature. The primary antibody was diluted in PBS containing 0.1% BSA and 0.2% Triton X-100 (PBT) and incubated with the sections overnight at 4°C. Slides were washed in PBS and then incubated with the appropriate secondary antibodies diluted in PBT for 2 h at room temperature. Slides were washed in PBS, mounted, and examined using confocal microscopy (Leica). The following antibodies were used at the indicated dilutions for immunofluorescence: rabbit anti-FOXA-2 (1:2,000; a gift from Dr. T.M. Jessell), guinea pig anti-insulin (1:1,000; Linco), rabbit anti-Pdx1 (IDX-1 253: 1:1,000), rabbit anti-GLUT2 (1:100; a gift from Dr. B. Thorens), Cy3-conjugated donkey anti-rabbit IgG (1:1,500; Jackson Immunoresearch), and Cy2-conjugated donkey anti-guinea pig IgG (1:400; Jackson).
Islet isolation and RNA analysis.
Islets from two to five pancreata from P8 mice were isolated using the standard collagenase procedure and hand picked under a light microscope (28). Total RNA from islets was isolated in TRIzol (GIBCO) according to the manufacturer’s instructions. The amount of total RNA recovered per islet was ∼25 ng and did not differ between control and Foxa2loxP/loxP;Ins.Cre mice. RT-PCR analysis was performed as described (29,30). Briefly, reverse transcription was performed using random hexamers and SuperScript II Reverse Transcriptase (GIBCO). PCR conditions were one cycle of 94°C for 3 min, followed by 28 cycles of 94°C for 45 s, 60°C for 45 s, 72°C for 90 s, and one cycle of 72°C for 5 min in a buffer containing 1.5 mmol/l MgCl2, 10 μmol/l primers, 0.05 μCi 32P-dATP, and 200 μmol/l dNTPs. PCR products were separated on 5% acrylamide gels that were dried and exposed to phosphorimager screens. The forward and reverse primers used were as follows:
Hprt (130bp): 5′-GGCCATCTGCCTAGTAAAGCT and 5′-GCTGGCCTATAGGCTCATAGT,
Pdx1 (553bp): 5′-TGGAGCTGGCAGTGATGTTGA and 5′-TCAGAGGCAGATCTGGCCAT,
Glut2 (193bp): 5′-CTGCTACTGCTCTTCTGTCCA and 5′-CATCCGTGAAGAGCTGGATCA.
Under the conditions used, determination of mRNA levels by RT-PCR was quantitative, as phosphorimager analysis (ImageQuant, Molecular Dynamics) of amplified products from serially diluted cDNAs confirmed the linearity of the assay for all primers used.
Protein isolation and Western blot analysis.
Pancreatic islets from P8 mice were isolated on ice as described above. Protein was extracted in 50 mmol/l Tris (pH 7.5), 150 mmol/l NaCl, 1% NP40, 0.5% sodium deoxycholic acid and Complete Protease Inhibitor Cocktail Tablets (Roche) on ice. We obtained ∼150–200 ng protein per islet, with yields comparable between control and mutant mice. Proteins were separated on 10% acrylamide Tris-HCl gels (Biorad), transferred to nitrocellulose membranes (Millipore), and blocked with 10% nonfat milk at room temperature for 1 h. Anti-IDX-1 253 was used at 1:10,000 in 1% milk for 1 h at room temperature (31). Goat α-rabbit conjugated to horseradish peroxidase was used at 1:10,000 (Biorad) for 1 h at room temperature. Then three 15-min washes were performed after each antibody incubation. The ECL-Plus detection system was used to detect the signal (Amersham Pharmacia). For quantification of band intensities, films were scanned and images imported into the ImageQuant (Molecular Dynamics) program. Intensities obtained for PDX-1 were normalized to those calculated for α-tubulin.
RESULTS AND DISCUSSION
Foxa2 is a transcriptional activator of Pdx1 in β-cells.
Because prior in vitro studies have suggested Foxa2 as an important regulator of Pdx1 (13,14,17,18), we examined Pdx1 mRNA expression in 8-day-old Foxa2loxP/loxP;Ins.Cre mice in which the Foxa2 gene is deleted from 85% of all β-cells (25). When we measured Pdx1 mRNA levels in whole pancreas using a RNAse Protection Assay, no differences were observed between Foxa2loxP/loxP;Ins.Cre and control mice (data not shown). This result can be explained by the fact that Pdx1 transcripts are present not only in islets but also in ducts and acinar cells. Consequently, because acinar and duct cells comprise >90% of pancreatic mass, normal Pdx1 expression levels in these cells would overshadow any alteration in β-cell−specific Pdx1 expression. Therefore, we isolated pancreatic islets from Foxa2loxP/loxP;Ins.Cre mice and analyzed Pdx1 mRNA levels by quantitative RT-PCR. Hypoxanthine-phosphoribosyltransferase (Hprt) was used as an internal control (Fig. 1A). We observed a 68% downregulation of Pdx1 mRNA in Foxa2loxP/loxP;Ins.Cre islets when compared to control islets (Fig. 1B). The observed effect is likely an underestimation of the true contribution of Foxa2 to the regulation of Pdx1 in β-cells, because 1) islets also contain Pdx1-expressing δ-cells in which Foxa2 was not deleted and 2) Foxa2 is still present in 15% of β cells in P8 islets (25). Because Foxa2loxP/loxP;Ins.Cre mice are hypoglycemic (25), it is possible that the reduction of Pdx1 expression observed in these mice might be the result of hypoglycemia rather than a direct effect of Foxa2 deficiency. However, recent studies on the impact of glucose levels on Pdx1 expression in pancreatic β-cells suggest that although persistent hyperglycemia (16 mmol/l glucose) extinguishes Pdx1 mRNA and protein expression, hypoglycemia (0.8 mmol/l glucose) actually restores Pdx1 levels (32). Therefore, it appears likely that the downregulation of Pdx1 expression in Foxa2-deficient β-cells is not a consequence of the hypoglycemia observed in Foxa2loxP/loxP;Ins.Cre mice, but rather a direct effect of Foxa2 deficiency. Our data showed that Foxa2 is an essential upstream factor that regulates Pdx1 mRNA levels, directly or indirectly, in β-cells in vivo.
The level of PDX-1 protein is decreased in Foxa2loxP/loxP;Ins.Cre mice.
Next we examined PDX-1 protein levels in Foxa2loxP/loxP;Ins.Cre mice by indirect immunofluorescence (Fig. 2A–D). PDX-1 is normally expressed at high levels in β- and δ-cells and at a low level in acinar cells (Fig. 2A and C). In Foxa2loxP/loxP;Ins.Cre animals, β-cells that lack Foxa2 still express PDX-1 immunoreactivity similar to the extent seen in control islets (compare Fig. 2B and D with 2A and C). Because the determination of protein expression by immunofluorescence is at best semiquantitative, we next examined PDX-1 protein expression in islets by Western blot analysis (Fig. 2E). The level of PDX-1 protein was decreased by ∼69% in Foxa2loxP/loxP;Ins.Cre islets when compared to controls (Fig. 2F), which is similar to the reduction in Pdx1 mRNA levels shown above. This reduction was not the result of a complete loss of PDX-1 protein in a subset of β-cells, as was demonstrated by the uniform distribution of PDX-1 protein in β-cells from Foxa2loxP/loxP;Ins.Cre mice (Fig. 2A–D). Because the residual amount of PDX-1 protein found in Foxa2loxP/loxP;Ins.Cre mice was comparable to the PDX-1 protein level (50%) in Pdx1/IPF-1 heterozygotes, it can be hypothesized that Foxa2loxP/loxP;Ins.Cre mice would develop impaired glucose tolerance, similar to Pdx1 heterozygous mice and MODY4 (IPF-1) patients (31,33–35). However, in part because of the requirement in Foxa2 for the expression of both subunits of the ATP-sensitive K+ channel, Foxa2loxP/loxP;Ins.Cre mice succumb to hypoglycemia between P9 and P12 (25). Because the impaired glucose tolerance caused by Pdx1/IPF-1 heterozygocity is apparent only late in life (33,34), these long-term consequences of reduced Pdx1 levels secondary to Foxa2 deletion could not be evaluated in Foxa2loxP/loxP;Ins.Cre animals.
Glut2 expression is not altered in the Foxa2loxP/loxP;Ins.Cre pancreas.
Targeted inactivation of the Pdx1 gene in β-cells leads to disturbed glucose homeostasis (34). This is caused by the progressive loss of Glut2 expression and gradual decrease of insulin expression (34). In addition, in Pdx1 heterozygous animals, Glut2 expression is markedly reduced at age 18 weeks (34). Given the reduction in Pdx1 expression in Foxa2loxP/loxP;Ins.Cre mice, we investigated whether the remaining PDX-1 protein in pancreatic islets of Foxa2loxP/loxP;Ins.Cre mice is sufficient to activate downstream targets of Pdx1 such as Glut2 and insulin. In addition, overexpression of Foxa2 in rat INS-1 insulinoma cells resulted in a dramatic (90%) decrease in Glut2 expression, suggesting that Foxa2 might be a negative regulator of Glut2 gene expression in β-cells (36). To address this question, we determined Glut2 protein and mRNA levels by immunofluorescence (Fig. 3A and B) and quantitative RT-PCR (Fig. 3C), respectively. Glut2 mRNA levels were analyzed in total RNA samples from whole pancreas. Because Glut2 expression is limited to β-cells, the dilution by exocrine cell mRNA did not affect the analysis. No significant differences were observed in Glut2 mRNA level or protein expression between control and Foxa2loxP/loxP;Ins.Cre islets (Fig. 3A–C). The discrepancy with the results obtained by overexpression of Foxa2 in INS-1 cells (36) might be explained by the nonphysiological levels of Foxa2 expression achieved in the INS-1 system. We have previously shown that another important target of Pdx1 regulation, namely insulin, is not affected by the deletion of Foxa2 from pancreatic β-cells (25). No differences were observed in insulin immunofluorescence, total pancreatic insulin content, or insulin mRNA levels between control and Foxa2loxP/loxP;Ins.Cre mice (25). Thus, the residual amount of PDX-1 protein present in the β-cells of Foxa2loxP/loxP;Ins.Cre mice is sufficient to maintain activation of the two Pdx1 target genes, Glut2 and insulin, at least until P8, after which time the Foxa2loxP/loxP;Ins.Cre mice succumb to the severe hypoglycemia caused by the absence of Foxa2 from β-cells (25).
In summary, the molecular analysis of Foxa2loxP/loxP;Ins.Cre mice presented here has provided the first in vivo evidence of transcriptional regulation of Pdx1 by Foxa2 in β-cells, indicating that FOXA-2 acts as a transcriptional activator that is required for the maintenance of Pdx1 transcription. Consistent with the finding of FOXA binding sites in the Pdx1 promoter (13,14,17,18), we have shown downregulation of Pdx1 mRNA and protein levels in β-cells from which Foxa2 has been deleted. These in vitro and in vivo data have concurrently placed Foxa2 upstream of Pdx1 in the transcription factor hierarchy in the differentiated β-cell.
Future studies will be directed at developing a Cre-transgene that will allow for the deletion of Foxa2 before the formation of the pancreatic primordia. This will allow the assessment of the role for Foxa2 during the initial transcriptional activation of the Pdx1 gene. We have shown that Foxa2 is indeed required for the full activation of the Pdx1 gene in pancreatic β-cells in vivo; however, Foxa2 clearly is not the only activator of the Pdx1 gene, as some Pdx1 expression is maintained in cells devoid of Foxa2. Our findings support the notion expressed by others (37) that mutations in the human FOXA-2 gene might also contribute to abnormal glucose homeostasis and diabetes.
This study was facilitated by the University of Pennsylvania Diabetes Center (P30-DK-19525) and the Penn Center for Molecular Studies in Digestive and Liver Disease (P30-DK-50306). This work was supported by the National Institute of Diabetes and Digestive and Kidney Diseases (RO1-DK-55342 to K.H.K.).
The authors are grateful to Drs. J. A. Blendy, N. Perreault, and N. R. May for their critical reading of the manuscript. The authors are thankful to Dr. B. Calvi for help in confocal microscopy and to K. A. O’Shea for maintaining the mouse colony and genotyping the animals.
Address correspondence and reprint requests to Dr. Klaus H. Kaestner, Department of Genetics, University of Pennsylvania School of Medicine, 415 Curie Blvd., Philadelphia, PA 19104-6145. E-mail: firstname.lastname@example.org.
Received for publication 16 January 2002 and accepted in revised form 6 May 2002.
C.S.L. and N.J.S. contributed equally to this article.
Hprt, hypoxanthine-phosphoribosyltransferase; IPF-1, insulin promoter factor 1; Pn, postnatal day n; PBT, PBS containing 0.1% BSA and 0.2% Triton X-100; p.c., post coitum; PP, pancreatic polypeptide.