Ptf1a Binds to and Activates Area III, a Highly Conserved Region of the Pdx1 Promoter That Mediates Early Pancreas-Wide Pdx1 Expression^▿

Transgenic constructs and generation of transgenic mice.

Characterization of transgenic mice.

The morning of vaginal plug appearance was considered to be E0.5. Dissected internal organs were fixed, stained with 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-Gal) to detect β-Gal, embedded, and sectioned (12, 54). Immunohistochemical staining for glucagon and insulin was as described previously (12). The Pdx1 expression level was analyzed using a 1:1,000 dilution of rabbit anti-Pdx1 (32) and Vectastain ABC and DAB kits (Vector Labs). Whole-mount images of dissected digestive organs and embryos were taken using an Olympus XZS9 bifocal dissecting microscope with a Nikon Coolpix 4300 digital camera and a Nikon UR-E4 eyepiece adaptor. Images of sections were taken under brightfield illumination using an Olympus BX41 microscope and digital camera with Magnafire software (Optronics). The image brightness, contrast, and color variations were minimally adjusted using Adobe Photoshop 6.0 or PowerPoint (37). The adjustments were equivalent for all samples that were directly compared.

TRANSFAC analysis.

The highly conserved region of the mouse Pdx1 fragment ranging from nucleotide −3030 to −1694 was examined in silico by using a default setting for putative transcription factor binding sites using the TRANSFAC matrix table and Tfsearch software version 1.3 (http://www.cbrc.jp/htbin/nph-tfsearch).

Electrophoretic mobility shift assays.

Nuclear extracts were prepared from 8-week-old male B6D2F1/J (Jackson Labs) mouse pancreata as described in references 29 and 43, modified for use with whole tissue (35). Briefly, whole pancreata were homogenized with a Tissuemiser homogenizer in ice-cold lysis buffer (10 mM Tris-HCl [pH 7.4], 0.1 M NaCl, 3 mM MgCl₂, 0.5% NP-40, 0.5 mM dithiothreitol [DTT], 0.5 mM phenylmethylsulfonyl fluoride [PMSF], 0.5 mg/liter N-α-p-tosyl-l-phenyl chloromethyl ketone [TPCK], 0.5 mg/liter N-α-p-tosyl-l-lysine chloromethyl ketone [TLCK], 0.6 μM leupeptin, and 2 μM pepstatin). The homogenate was centrifuged (1,000 × g, 2 min, 4°C), and the pellets were resuspended in ice-cold lysis buffer and centrifuged (1,000 × g, 1 min, 4°C). The pellet was resuspended in ice-cold nuclear extraction buffer (20 mM HEPES [pH 7.9], 0.4 M NaCl, 0.90 mM spermidine, 0.2 mM EDTA, 2 mM EGTA, 25% glycerol, 2 mM DTT, 1 mM PMSF, 0.5 mg/liter TPCK, 0.5 mg/liter TLCK, 0.6 μM leupeptin, and 2 μM lepstatin). The nuclear extract was centrifuged (16,000 × g, 1 min, 4°C), and the supernatant was collected and stored at −80°C.

Electrophoretic mobility shift assays including antibody supershifts were performed as described previously (36, 40). The PTF1 control (PTF1 Ela1; bp −115 to −96 of the rat elastase 1 promoter) sequences were as follows: upper, 5′-gatcGTCACCTGTGCTTTTCCCTGC-3′, and lower, 5′-gatcGCAGGGAAAAGCACAGGTGAC-3′. The putative area III PTF1 binding site (AIII PTF1; bp −1730 to −1709 of the Pdx1 promoter) sequences were as follows: upper, 5′-gatcCACAGGTGGCTCTTTTCCCTG-3′, and lower, 5′-gatcCAGGGAAAAGAGCCACCTGTG-3′. Lowercase letters indicate 5′ overhangs for radioactive fill-in labeling. Other oligonucleotides are described in Fig. 6C. Mobility shift binding reaction mixtures, 20 μl in volume, were prepared on ice and contained 10 mM HEPES (pH 8.0), 90 mM NaCl, 1 mM EDTA, 0.1 M DTT, 6 μg bovine serum albumin, 2 μg poly(dI-dC), and 40 fmol of labeled probe. When specified, 5 μg of pancreatic nuclear extract, 5 μg of MIN-6 nuclear extract (gift from Jennifer Van Velkinburgh, Vanderbilt University), 1 μl of anti-Ptf1a (36), or a specified quantity ranging from 0.4 to 40 pmol of nonradioactive competitor oligonucleotides was added to the binding reaction mixture (see Fig. 6). The reaction mixtures were incubated for 15 min at 30°C. Totals of 20 μl of each reaction mixture were loaded and resolved at 4°C on a 4.0% nondenaturing polyacrylamide gel (40:1) in ice-cold Littman running buffer (50 mM Tris, 0.38 M glycine, 2 mM EDTA [pH 8.0]).

Transfection plasmids.

The following were the generous gift of Ray MacDonald (University of Texas, Southwestern Medical Center) and have been described previously (2): rat elastase 1 minimal promoter, bp −92 to +8 in the pGL3 basic vector (−92rEla1p/pGL3; Promega); a six-copy rat Elastase 1 PTF1 site (A element) and elastase 1 minimal promoter (A26/pGL3); HEB expressed under the control of cytomegalovirus (CMV) promoter (HEB/pcDNA1.1); mouse Ptf1a expressed under the control of CMV (pCMV-mp48); human RBP-Jκ expressed under the control of CMV (Myc-hRbpsuh/pcDNA3); and mouse RBP-L expressed under the control of CMV (CMV.RBbpsuh-L).

Oligonucleotides annealing in the pBluescript II (Stratagene) vector (ATTAACCCTCACTAAAG) and to the 3′ end of area III (GTCTCTGATTTCTTCAGttAAAAGAGCCACCTtTGCCCGTCAAGGGGCC) (11) were used to generate an E box and TC box mutant of the PTF1 binding site (bp −1730 to −1709); the lowercase nucleotides correspond to those targeted in other PTF1 binding mutants (2, 21, 36). Wild-type and area III mutant (Pdx1^mIII) sequences were subcloned into the HindIII site just upstream of the herpes simplex virus thymidine kinase (TK) minimal promoter driving the firefly luciferase expression cassette (TK-Luc) in pGL3 (Promega) (31). Wild-type and mutant Pdx1^I-II-III TK-Luc were generated by subcloning area I- to II-spanning sequences from the Pdx1^I-II-III transgene into either Pdx1^III TK-Luc or Pdx1^mIII TK-Luc. The correctness of the plasmids was verified by restriction digestion and partial DNA sequencing.

Transient transfections.

The Rattus norvegicus exocrine-derived cell line (AR42J) was obtained from ATCC and propagated according to the specifications provided. Cells were transiently transfected with Lipofectamine 2000 (Invitrogen) in 24-well plates at 2 × 10⁵ cells/well (0.8 μg DNA/well), according to the manufacturer's instructions. The data were analyzed for significance by one-way analysis of variance and Tukey's posttest.

The Homo sapiens embryonic-kidney-derived cell line (HEK 293) was propagated in Dulbecco modified Eagle medium with 10% fetal bovine serum according to specifications provided by ATCC. The cells were transiently transfected with Lipofectamine in 24-well plates, 1.5 × 10⁵ cells/well (0.5 μg DNA per well). Equimolar amounts of the wild-type and area III-mutant TK-Luc plasmids were used at a 1:1 molar ratio with the CMV-driven expression plasmids. The data were analyzed for significance by two-way analysis of variance and Bonferroni's posttest.

The TK-driven Renilla luciferase expression plasmid phRL-TK (10 ng; Promega) was cointroduced into AR42J and HEK 293 cells to control for transfection efficiency. Renilla and firefly luciferase activities were measured by using the dual luciferase assay (Promega) on cell extracts prepared 40 to 48 h after transfection. Each transfection condition was performed on at least four independent occasions. Statistical analysis and graph generation were performed using PRISM software (GraphPad).

Embryonic dorsal pancreas and gut tube dissections.

Dorsal pancreas bud dissections were performed as illustrated in Fig. 8, panels A to F. Gut tube dissections were performed as previously described (3). Tissue purity was assayed by detection of Ptf1a or alb1 mRNA isolated from dissected tissues. mRNA from dissected tissues was isolated by using an RNeasy micro kit (QIAGEN) and cDNA was synthesized by using an iScript cDNA synthesis kit (Bio-Rad). PCR analysis was performed using iQ SYBR green supermix (Bio-Rad) according to the manufacturer's instructions. The following primers were used at a concentration of 0.3 μM in 25-μl reaction mixtures: Ptf1a, GGTTATCATCTGCCATCGAGG and GCTGTTTTTCATCAGTCCAGGA; alb1, AGCACACAAGAGTGAGATCGCC and TGGCATGCTCATCGTATGAGC; and Hprt, GCTTGCTGGTGAAAAGGACCT and TGCGCTCATCTTAGGCTTTGTA. The iCycler iQ and optical system software version 3.1 (Bio-Rad) were used to define the threshold cycle (C_T) value for each PCR. The C_T values of the Hprt amplification were subtracted from the C_T values of Ptf1a or alb1 for normalization. The relative expression was calculated as 2^(−ΔC_T).

Ptf1a ChIP.

Chromatin isolation and chromatin immunoprecipitation (ChIP) for embryo tissues were modified from established protocols (6, 50). Crude chromatin preparations were pooled from 163 dissected dorsal pancreatic buds and 77 gut tubes of E11.5 embryos. Dissected material was fixed for 10 min at 25°C in 1% formaldehyde in phosphate-buffered saline, pelleted, transferred to 125 mM glycine in phosphate-buffered saline, and incubated on ice for 5 to 10 min. Fixed cells were pelleted, resuspended in radioimmunoprecipitation assay (RIPA) buffer (50 mM Tris-HCl [pH 8.0], 0.15 M NaCl, 0.5% deoxycholic acid, 0.1% sodium dodecyl sulfate [SDS], 1% IgePal CA-630, 5 mM EDTA, 0.5 mM PMSF, 1 mM benzamidine, 5 μg/ml antipain, 5 μg/ml leupeptin, 5 μg/ml trypsin inhibitor), and disrupted in a tissue homogenizer to liberate the nuclei. The nuclei were pelleted; resuspended in 50 mM Tris-HCl (pH 8.0), 10 mM EDTA, 0.5 mM PMSF, 1 mM benzamidine, 5 μg/ml antipain, 5 μg/ml leupeptin, 5 μg/ml trypsin inhibitor; and incubated on ice for 10 min prior to being flash-frozen and stored at −80°C. Nuclear isolates were pooled and sonicated to an average size of 150 to 400 bp. Equal amounts of chromatin (1 μg as estimated by DNA concentration) were diluted in RIPA (modified to 1 mM EDTA) and precipitated by using normal rabbit immunoglobulin G (IgG) (Santa Cruz) or anti-Ptf1a rabbit IgG (gift from Ray MacDonald, UT Southwestern). Chromatin-antibody complexes were precipitated with protein A-Sepharose (protein A-Sepharose CL-4B; catalogue no. 17-0780-01, Amersham). Protein A beads were washed with RIPA (modified to 0.5 M NaCl) and eluted with 10 mM Tris-HCl (pH 8.0), 1% SDS, 1 mM EDTA for 20 min at 42°C. The cross-links were reversed by incubation at 68°C overnight, and the DNA was digested in 10 mM Tris-HCl (pH 8.0), 0.1 M NaCl, 1 mM EDTA, 1% SDS, 0.1 mg/ml proteinase K and phenol chloroform extracted prior to PCR. Material assayed by ChIP and input material were analyzed by PCR with the following primers: Pdx1 area I, CTGGGACAGAGTCTCAGCAGAAG and CGTCCTGATAGTCCTCCCTGAT; Pdx1 area II, AGCAGCGAGCTTGTTTTTCTG and CACTTCTGGTCTAATTGCATGCA; Pdx1 area III, TGCCCCGGCCTTTCA and GGGAGAGTGTCTCTGATTTCTTCAG; Pdx1 area IV, GCTCCAATGCCATTTGTCAA and TGGATCCTGACTGCGTCTTG; and elastase 1 promoter, TTGACTTAAAATTTGTTCATTTGT and ACCCTCTTTATACGGCTCTT. PCRs were analyzed by using a 2100 bioanalyzer and dsDNA 1000 lab chips (Agilent Technologies). The PCR product concentration and molarity were calculated using 2100 Bioanalyzer Expert software (Agilent Technologies).

Lineage tracing reveals that areas I and II drive expression in pancreatic endocrine but not exocrine tissue.

We previously reported that the 1-kb PstI-BstEII fragment (Pdx1^PB; bp −3007 to −2011), which contains areas I and II from the Pdx1 locus (and none of area III), drives the expression of a β-Gal reporter transgene exclusively in endocrine cells as early as E11.5 and continuing into adulthood (11, 12, 49). A specific role for areas I and II in regulating endocrine expression is supported here by lineage-tracing analysis using Pdx1^PBCre mice. Pdx1^PBCre transgenic mice were crossed with the lineage-independent Gt(ROSA)26Sor^tm1sor (R26R) reporter strain (45) to determine whether Pdx1^PB was activated at an earlier stage of pancreatic development within a common endocrine/exocrine pancreatic progenitor. In this case, we should observe β-Gal activity in all pancreatic-bud derivatives (acini, ducts, and islets). As shown in Fig. 2, Pdx1^PBCre-mediated recombination within the pancreas was detected only in islet endocrine cells at postnatal day 1 (P1). Recombination was also observed in a few scattered cells within the stomach and duodenum (sites of endogenous Pdx1 expression) (Fig. 2) and ectopically in dorsal root ganglia (data not shown). When X-Gal staining was observed in the duodenal lumen, it was present in both transgenic and nontransgenic neonatal littermates and is most likely due to endogenous intestinal β-Gal activity (11). Within the islets, there was recombination throughout the insulin-expressing β-cell core and in some glucagon-expressing cells (Fig. 2B). Despite the generation of only one Pdx1^PBCre transgenic line, our findings are consistent with our previously published results regarding the Pdx1^PBhsplacZ transgenic mice, the Pdx1^PBCre-ER; R26R bigenic mice (11, 54), and the expression of endogenous Pdx1 in islets after birth (18). Thus, our combined analyses indicate that areas I and II are not sufficient to direct expression to early pancreatic progenitors or differentiated pancreatic exocrine tissue.

Addition of area III to areas I and II is sufficient to drive early pancreas-wide expression.

Four independent Pdx1^I-II-IIIlacZ transgenic mice with detectable expression were generated in which lacZ was under the control of Pdx1^PB (containing areas I and II) plus area III (Fig. 1). Pdx1^I-II-IIIlacZ transgenic mice showed high levels of β-cell-enriched expression in the mature pancreas but at earlier stages showed pancreas-wide expression at apparently equivalent levels throughout the endodermally derived tissue (Fig. 3 and 4).

Two Pdx1^I-II-IIIlacZ lines were propagated and characterized. Adult pancreata from both Pdx1^I-II-IIIlacZ lines were X-Gal stained to assess the mature-organ expression specificity of the transgene. Both in whole mount and in sections, a pattern of β-Gal activity consistent with islet-specific expression in the pancreas was observed, preferentially in β cells (Fig. 3). The two Pdx1^I-II-IIIlacZ lines showed differing levels of expression in the adult (Fig. 3A and B), and the expression pattern within islets was sometimes variegated (Fig. 3C), but both lines showed expression within the same cell types. Exocrine expression was not detectable in adult pancreata (Fig. 3C and D). The expression pattern at this stage was similar to that seen in Pdx1^PBCre; R26R mice (Fig. 2) and to that previously reported for the Pdx1^PBhsplacZ (11, 51) and Pdx1^PBCre-ER; R26R transgenic mice (54).

Pdx1^I-II-IIIlacZ transgenic embryos were examined at E9.5, E10.5, and E11.5, the time during which pancreatic buds first become visible and outgrowth begins. Pdx1^I-II-IIIlacZ expression was undetectable at E9.5 (data not shown) but was detected broadly and robustly at both E10.5 and E11.5 within the dorsal and ventral pancreatic buds, but not in the adjoining gut tube (Fig. 4D and E). Consistent with the lack of transgene expression in the antral stomach and rostral duodenum, X-Gal staining colocalized with Pdx1 protein in the pancreatic region at E10.5, but was absent from the Pdx1-labeled stomach (Fig. 4C). At E14.5 and E16.5, whole-mount β-Gal expression was observed throughout the pancreas at relatively equal levels (Fig. 4F and data not shown). In contrast, whole-mount analysis at E18.5 and P1 revealed an expression pattern more consistent with forming islet clusters (Fig. 4G and H). Examination of sections at these time points showed β-Gal expression in both exocrine and endocrine cells; however, the level of β-Gal expression observed in endocrine cells was more intense than that observed in exocrine cells, and expression in exocrine cells was variegated (data not shown). We collected, at late embryonic stages, two independent F₀ transgenic Pdx1^I-II-IIIlacZ mice which had patterns of expression consistent with our propagated Pdx1^I-II-IIIlacZ transgenic lines. Thus, the expression of Pdx1^I-II-IIIlacZ gradually increases within endocrine cells and diminishes from exocrine cells between E14.5 and P1, eventually becoming restricted to β cells in adult tissue, similar to endogenous Pdx1 expression within the pancreas.

Pdx1^I-II-IIIlacZ mice exhibited β-Gal activity beginning at E10.5 in the dorsal neural tube caudal to the midbrain/hindbrain junction (Fig. 4A and B), which was observed in all four transgenic animals generated, suggesting that this ectopic expression was inherent to the transgene. In the two lines characterized at adult stages, Pdx1^I-II-IIIlacZ mice also had low levels of ectopic expression in the kidneys and hair follicles (data not shown), similar to the levels in Pdx1^PBCre-ER; R26R bigenic mice (54). Expression was not observed in other sites of endogenous Pdx1 expression, such as the antral stomach or rostral duodenum (30), at any time point analyzed.

The area III-containing XhoI-BglII fragment drives expression early throughout the Pdx1 domain.

The Pdx1^XB fragment drives reporter gene expression transiently in β cells; this expression was observed at E14.5 and P1 but not in the adult (Fig. 1). Earlier embryonic stages were not examined in these transgenic lines (11). To trace the cumulative effect of the activity of this fragment throughout development, we generated four independent Pdx1^XBCre transgenic mouse lines. All four lines shared the same overall pattern of recombination (Fig. 5), although in one line, recombination was not as robust (data not shown). We identified cells and their progeny that expressed Cre at high enough levels to cause recombination of the R26R reporter (45), even after Pdx1^XB activity ceased. In all Pdx1^XBCre; R26R transgenic mice, we observed recombination throughout the endogenous Pdx1 domain at both E14.5 and P1 (Fig. 5F and G). The pattern of recombination observed in whole mounts included pancreas, duodenum, antral stomach, and bile duct, which are all tissues that arise from cells expressing endogenous Pdx1 (30). Analysis of sections from P1 Pdx1^XBCre; R26R animals revealed complete recombination throughout both the endocrine and exocrine pancreas (Fig. 5C), suggesting an early recombination event. Indeed, examination of earlier embryonic stages revealed that the Pdx1^XBCre transgene is active in the posterior foregut region as early as E9.5 (data not shown). At E10.5, R26R recombination was observed in the entire dorsal and ventral pancreatic buds of the dissected digestive organs and was also detectable in the adjoining gut tube (Fig. 5A and B). By E11.5, R26R recombination was also apparent in the bile duct (Fig. 5D), consistent with later stages. Occasionally, whole mice underwent ubiquitous recombination (data not shown), but otherwise, no ectopic expression was observed; recombination was consistently restricted to the endogenous Pdx1 expression domain.

Ptf1a binds to area III in vitro.

Collectively, the data demonstrate that area III is involved in directing early, pancreas-wide Pdx1 expression. Area III sequences were thus analyzed in silico using Tfsearch and the TRANSFAC database to identify a potential transcription factor(s) responsible for controlling this expression. A PTF1-like binding site was found at the 3′ end (nucleotides −1730 to −1709) of area III, which was highly relevant since Ptf1a is coexpressed with Pdx1 throughout the pancreatic buds at E10.5 (23). This site had high identity within both the E and TC boxes that define Ptf1a-HEB-RBP-Jκ/L (PTF1) activator binding (Fig. 6C) (2, 7, 36), whereas a PTF1-like site was not found in area I or II.

Gel shift analyses were performed using the PTF1-like binding site from area III and the well-characterized PTF1-binding site from the elastase 1 gene (Ela1) (Fig. 6) (36). Nuclear extracts enriched in PTF1 from adult mouse pancreata illustrated that the area III site formed a complex with mobility similar to that of the Ela1 PTF1 element (Fig. 6A, lanes 9 and 12), suggesting that the putative area III PTF1 site bound to the same protein component(s) as the Ela1 PTF1 control. Furthermore, the Ptf1a antibody completely retarded the mobility of the comigrating complex of both the area III and Ela1 PTF1 probes (Fig. 6A, lanes 10 and 13). In contrast, a complex of quite different mobility, which was insensitive to the Ptf1a antibody, was formed with extracts from the MIN-6 mouse β-cell line, which does not produce Ptf1a (24) (Fig. 6A, lanes 14 and 15). These results suggested that a complex containing PTF1a was capable of binding to the bp −1730 to −1709 element of area III.

A PTF1 complex binding site contains both a bHLH E box (CANNTG) for Ptf1a-HEB binding and a contiguous TC box (TTTCCC) (the invariant nucleotides are underlined) for the binding of RBP-J or RBP-L (Fig. 6C) (2, 7, 36). At one or two helical turns apart center to center, the spacing between these elements is critical for PTF1 binding (36). Gel shift competition experiments were next performed to determine if each of these binding site components was necessary for the formation of the Ptf1a-containing complex on the bp −1730 to −1709 probe. The mutation(s) introduced within each competitor oligonucleotide(s) is known to impact Ptf1a-HEB and/or RBP-Jκ/L binding in other contexts (2, 21, 36). The competition pattern demonstrated that the Ptf1a-HEB-RBP-Jκ/L complex is present in the area III Ptf1a-containing binding complex.

PTF1 is important for area III-mediated activation.

To determine if the PTF1 site in area III was involved in Pdx1 promoter activation, the activity of a binding-defective mutant constructed in the context of either a Pdx1^III or Pdx1^I-II-III reporter was compared to its wild-type version in transfected AR42J cells. This pancreatic acinar cell line produces both PTF1 (7, 53) and Pdx1 (52). The PTF1 site mutation significantly reduced Pdx1^I-II-III TK-Luc and Pdx1^III TK-Luc activity (Fig. 7A). Moreover, Pdx1^III activity was stimulated in a PTF1 site-dependent manner upon cotransfection of Ptf1a, HEB, and RBP-Jκ/L in HEK 293 cells, a nonpancreatic cell line that does not produce PTF1 (Fig. 7B) (2). Experiments performed with a reporter driven by six copies of the elastase 1 PTF1 binding site verified the PTF1-dependent activity in both AR42J and HEK 293 cell lines (data not shown). Collectively, these results demonstrate that PTF1 is capable of specifically binding to and stimulating activity through area III.

Endogenous Ptf1a binds the Pdx1 promoter in vivo.

To investigate whether Ptf1a bound Pdx1 upstream regulatory sequences during pancreatic organogenesis, we developed conditions for performing a ChIP assay from tissues that were microdissected from E11.5 mouse embryos. Similarly staged embryos from Pdx1^lacZ/+ knock-in mice (30) stained for β-Gal served as controls for developing a protocol to dissect wild-type dorsal pancreatic buds from the gut tube (Fig. 8, panels A to E). Real-time reverse transcriptase-PCR analysis of the RNA revealed that the microdissected buds expressed the Ptf1a gene, whereas RNA, from the gut tube, heart, and liver bud were negative for this gene, as expected (Fig. 8, panel G). Conversely, the Ptf1a-positive buds were negative for the liver gene, albumin 1 (alb1) (Fig. 8, panel G), also as expected. We pooled 163 such pancreatic buds, cross-linked the chromatin in the tissues, and performed ChIP with a Ptf1a-specific antibody. Pooled gut tubes served as negative controls. The scaled-down ChIP reactions required analysis of the PCR products on a microfluidics workstation; the data depict the results of two analytical runs per PCR. PCR analysis of the ChIP products with primers to the elastase 1 gene promoter yielded markedly higher signals from the dorsal pancreatic buds than from the gut tube control (Fig. 8, panel H). These data confirm the validity and specificity of the embryonic ChIP assay for Ptf1a.

Using primers specific for amplification of each of the previously identified highly conserved regions within the Pdx1 promoter (areas I to IV), we next analyzed whether Ptf1a interacted specifically with area III or any of the other known regulatory regions. Strikingly, both area IV and III exhibited strong binding by Ptf1a in E11.5 embryonic pancreatic buds compared to its binding in the gut tube controls. By contrast, the islet-specific areas I and II did not exhibit binding to Ptf1a. These data further support a role for PTF1 in the expression of Pdx1 in early exocrine and endocrine progenitors.

Sequences within Pdx1^I-II-III direct dynamic Pdx1 expression in the pancreas.

The Pdx1^I-II-IIIlacZ transgene differs from the previously published endocrine-specific Pdx1^PBhsplacZ only in the addition of area III. Pdx1^I-II-IIIlacZ was expressed throughout the pancreatic buds in precursors to all cell types at E10.5; as development proceeded, expression was gradually diminished in acinar cells and intensified in endocrine cells. Thus, sequences within area III likely confer early pancreas-wide expression to the islet endocrine-specific Pdx1^PB fragment. Pdx1^I-II-III was unable to drive expression within the Pdx1 domain outside the pancreas (the antral stomach, rostral duodenum, or common bile duct). These expression data are consistent with the phenotype of mice containing a global deletion of Pdx1^I-II-III (Pdx1^ΔI-II-III) in which Pdx1 expression is dramatically reduced in pancreatic buds but is essentially normal in stomach and duodenum (9). Consequently, Pdx1^I-II-III deletion resulted in limited pancreatic bud outgrowth and differentiation with little effect on stomach and duodenum, in contrast to the results of the global Pdx1 deletion (26, 30).

Sequences in Pdx1 area III mediate embryonic pancreas-wide expression.

Similar to Pdx1^I-II-III reporter expression, the Pdx1^XB fragment drove Cre recombinase activity throughout the pancreatic buds early in pancreatogenesis. The extent of recombination observed in Pdx1^XBCre; R26R mice is reminiscent of that mediated by Pdx1Cre (17), in which Cre is driven by a 5.5-kb Pdx1 promoter fragment which contains sufficient sequence to recapitulate the endogenous Pdx1 expression pattern (11, 47). In contrast to the expression driven by Pdx1^I-II-III, Pdx1^XBCre-mediated recombination was also detected at high levels in other areas of endogenous Pdx1 expression (antral stomach, common bile duct, and rostral duodenum). The conserved region in common between the Pdx1^I-II-III and Pdx1^XB fragments is area III (Fig. 1), suggesting that sequences within this region are responsible for the early, broad expression of Pdx1 within the pancreas. The region 3′ to area III within the Pdx1^XB fragment shows no significant sequence conservation between species and is incapable of driving robust lacZ expression to Pdx1-producing cell types in vivo (11, 14). It is possible that antral stomach and duodenal expression are normally driven by area III in vivo but are repressed by sequences within the Pdx1^PB fragment in the Pdx1^I-II-III transgene. This seems unlikely, however, given the phenotype of the Pdx1^ΔI-II-III mice, where in the stomach and duodenum, only the enteroendocrine cells are affected, with, specifically, a 50% reduction in stomach gastrin cells and duodenal GIP cells. Alternatively, sequences 3′ to area III within the Pdx1^XB fragment may mediate extrapancreatic expression of Pdx1 but require interaction with factors binding area III to stabilize, or enhance, their activity. To date, a duodenal or gastric regulatory element has not been identified in the Pdx1 gene. Our in vivo transgenic analysis supports a role for area III in the regulation of early, broad expression of Pdx1 throughout the pancreatic bud epithelium.

Area III mediates acinar expression of Pdx1.

Acinar expression of β-Gal has not been observed in any mouse line carrying transgenes driven by the Pdx1^PB fragment containing areas I and II (Pdx1^PBhsplacZ [11, 49, 51], Pdx1^PBCre [this study], or Pdx1^PBCre-ER [54]), suggesting that elements within area III are responsible for endogenous Pdx1 acinar expression. Since Ptf1a expression is initially expressed broadly throughout the pancreatic buds but becomes restricted to acinar tissue and is capable of binding to sequences upon which area III activity is dependent, we hypothesize that the dynamic changes in Pdx1 expression within acinar tissue are mediated, at least in part, by PTF1. Recent studies indicate that the expression of another component of the PTF1 complex, the mammalian Suppressor of Hairless orthologue RBP-Jκ/L, is also developmentally regulated. RBP-Jκ is more highly expressed during embryonic pancreas development, while RBP-L is exclusively found in adult pancreata (2). As RBP-Jκ and RBP-L differ in their transcriptional activity (2), this developmental switch in the TC box binding factor may explain the change in acinar Pdx1 expression.

PTF1 activates area III and binds to this control region in embryonic pancreas.

PTF1 was a likely candidate for early, broad regulation of Pdx1, since its pancreas-specific component, Ptf1a, is expressed pancreas-wide as early as E9.5 (23, 28), correlating well with the early pancreatic pattern of β-Gal expression that we observed in the Pdx1^I-II-IIlacZ (Fig. 4D and E) and the Pdx1^XBCre; R26R mice (Fig. 5A and B). Also, the ectopic β-Gal activity exhibited by the Pdx1^I-II-IIIlacZ mice in the dorsal neural tube at E10.5 is identical to the β-Gal expression in the Ptf1a^Cre/+; R26R mouse in those regions (16, 23). Analysis of area III in silico revealed a putative PTF1 site: a conserved CANNTG (E box) and TTTCCC (TC box) separated center to center by one turn of DNA (36). Ptf1a present in nuclear extracts from adult mouse pancreata bound the putative PTF1 site in area III with relatively high affinity (Fig. 6), and activity mediated by area III was dependent on this site (Fig. 7). The in vivo relevance of the PTF1 site within area III is demonstrated by the binding of endogenous Ptf1a to area III within the endogenous Pdx1 promoter in developing pancreatic buds at E11.5 (Fig. 8). Our data do not exclude the possibility that other transcription factors affect transcription of Pdx1 during early pancreatic bud formation and outgrowth (4, 33), but they suggest that PTF1 activates, at least in part, early pancreas-wide expression of Pdx1 through area III.

The significance of Ptf1a binding to area IV remains to be determined. The role of area IV in the regulation of Pdx1 gene expression is only now beginning to be addressed. Pdx1 upstream regulatory sequences lacking area IV are capable of driving reporter transgene expression in a pattern temporally and spatially indistinguishable from endogenous Pdx1 (11, 47). Transgene-based complementation experiments on Pdx1 null mice suggest, however, that area IV is required for appropriate levels of Pdx1 expression in the postnatal stomach and duodenum, although it is not required to rescue the pancreatic defects caused by Pdx1 deficiency (4, 10).

In vivo relevance of Ptf1a binding area III.

In vivo deletion of areas I, II, and III from the Pdx1 promoter results in a dramatic arrest in pancreas development, similar to that observed in globally Pdx1 null mice (9, 30). Since we show that areas I and II are only active in endocrine tissue within the pancreas (Fig. 2), the failure of the pancreas to develop in the Pdx1^{ΔI-II-III/−} mice is most likely due to the loss of area III, suggesting that sequences within area III, in particular the PTF1 site, are important for expression in the developing pancreas. Pdx1 expression precedes and is in a broader domain within the posterior foregut than Ptf1a expression. In Ptf1a null mice, Pdx1 expression is maintained in the dorsal pancreas (23). Taken together, these data suggest Ptf1a is not necessary to initiate Pdx1 expression in the pancreas but is involved in maintaining or augmenting Pdx1 expression within the developing pancreatic buds. We would predict that removal of area III containing the PTF1 binding site in vivo would result in an apancreatic phenotype without affecting the development of other tissues and cell types in which Pdx1 is expressed, similar to the phenotype observed with deletion of areas I, II, and III.