Expression of Heparin-binding Epidermal Growth Factor-like Growth Factor during Pancreas Development

The development of the pancreas appears to be regulated by various growth factors. We report here the expression of heparin-binding epidermal growth factor (EGF)-like growth factor (HB-EGF) in the developing pancreas. Immunostaining of fetal and neonatal rat pancreata, in which endocrine cells are visible as cell clusters often associated with primitive ducts or ductular cells, revealed that most of the cluster-forming cells and primitive ducts or ductular cells express HB-EGF protein. In contrast, the exocrine pancreas lacked HB-EGF expression. Based on findings that the expression pattern was similar to that of the homeodomain-containing transcription factor PDX-1 (IDX-1/STF-1/IPF1) and that the regulatory region of the HB-EGF gene contained sequences similar to the PDX-1-binding A element, we examined whether PDX-1 could be a potential activator of HB-EGF gene expression. The results of reporter gene analyses suggested that the HB-EGF gene promoter is PDX-1-responsive and that the activity of the promoter in pancreatic beta cell-derived βTC1 cells depends on the PDX-1 binding site-like sequences. Gel-mobility shift analyses using an anti-PDX-1 antibody indicated that PDX-1 is a specific and dominant binding factor for an A element-like sequence in the HB-EGF gene. These observations suggest the possible involvement of HB-EGF in pancreas development. While PDX-1 is essential for pancreas development, HB-EGF may function as a mediator of PDX-1 and thus be involved in the development of the endocrine pancreas.

The pancreas is an organ composed of two distinct cell populations: exocrine cells, which secrete digestive enzymes, and endocrine cells, which secrete hormones. The pancreas arises from the endoderm as a dorsal bud and a ventral bud which fuse together to form a single organ (1). Various growth factors expressed in the developing pancreas and its surrounding mesenchyme-derived cells are considered to be involved in the development of the endocrine and exocrine cells. Among those growth factors which have been shown to be expressed in pancreatic islets, the epidermal growth factor (EGF) 1 family of growth factors such as EGF, transforming growth factor-␣ (TGF-␣), and betacellulin can bind to the EGF receptor produced by pancreatic islet cells and have been shown to exert various effects on islet cell differentiation and proliferation (2,3).
Heparin-binding EGF-like growth factor (HB-EGF) is a member of the EGF family which was purified initially from the conditioned medium of macrophage-like U937 cells (4,5). It is synthesized as a membrane-anchored precursor (HB-EGF proform) that can be processed to release the soluble form. In terms of function, HB-EGF induces autophosphorylation of the EGF receptor and is known to be a potent mitogen for several cells (6 -10). Clinically, HB-EGF may be involved in several diseases such as atherosclerosis (11) or carcinogenesis (12,13). Like other EGF family growth factors (2,3), HB-EGF seems to play a role in tissue development; during myogenesis, for example, HB-EGF gene expression is activated by MyoD (14). To clarify the molecular mechanism for pancreas development, we examined the expression of HB-EGF in fetal, neonatal, and adult pancreata. We found that HB-EGF is abundantly expressed in endocrine pancreas cells and primitive duct cells from which the endocrine cells are derived in fetal rats. We also found that the expression pattern in the developing pancreas was similar to that of the homeodomain-containing transcription factor PDX-1 (IDX-1/STF-1/IPF1). While the regulatory sequences of the HB-EGF gene contained sequences similar to the PDX-1 binding motif, we have shown that PDX-1 can bind to the 5Ј-flanking region of the HB-EGF gene and activates its promoter. Thus our present observations suggest that HB-EGF, which is expressed in the developing pancreas and may be regulated in part by PDX-1, is involved in pancreas development.

MATERIALS AND METHODS
Animals and Tissue Processing-Pregnant female rats were anesthetized by injection of pentobarbital (0.5 mg/kg of body weight) and the uteri resected and placed in 10% formalin buffered to pH 7.4 with 0.1 M sodium phosphate. The embryos (days 14 and 20) were removed and fixed for 5 h at 4°C in the same solution. Next, pancreata were removed and fixed for 5 h at 4°C, and the tissues were routinely processed to prepare paraffin sections.
Immunostaining of PDX-1, HB-EGF, Insulin, and Cytokeratin-The anti-PDX-1 antiserum was recently established by us (15) after immunizing a rabbit with the synthetic peptide SPQPSSIAPLRPQE repre-senting amino acid residues 269 -282 of the PDX-1 peptide (16). An antibody recognizing the cytoplasmic domain of the HB-EGF proform was also produced by immunizing a rabbit with another synthetic peptide (precursor COOH-terminal residues 185-208) as previously reported (11). All procedures were carried out at room temperature unless otherwise specified. The sections were rinsed with 0.05 M Tris-HCl buffer or 0.01 M phosphate-buffered saline three times before incubation at each step.
Immunohistochemical detection of HB-EGF protein was done using the streptavidin-biotin (LSAB) method (Dako LSAB Kit). The sections, which had been preincubated with 3% H 2 O 2 solution for 10 min to block endogenous peroxidase, were incubated for 20 min with blocking agent, for 20 min with rabbit polyclonal anti-HB-EGF antibody diluted 1:200 in phosphate-buffered saline containing 1% bovine serum albumin, and for 10 min with biotinylated swine anti-rabbit immunoglobulins in phosphate-buffered saline. They were then incubated with streptavidin conjugated to horseradish peroxidase in 0.05 M Tris-HCl buffer, and a positive reaction was visualized with 3-amino-9-ethylcarbazol. For the negative control, the primary antibody was preabsorbed with an excess amount of the peptide antigens. Before mounting, the sections were counterstained with Mayer's hematoxylin.
For detection of PDX-1, the avidin-biotin complex method (Vectastain ABC Kit, Vector Laboratories Inc., Burlingame, CA) was used. Sections were incubated with rabbit anti-rat PDX-1 antibody diluted 1:200 in phosphate-buffered saline containing 1% bovine serum albumin and then incubated with secondary antibodies using biotinylated goat anti-rabbit IgG diluted 1:2000. These sections were then incubated with avidin-biotin complex reagent for 30 min. A positive reaction was visualized by incubating with peroxidase substrate solution containing 3,3Ј-diaminobenzidine (Zymed Laboratories Inc., San Francisco, CA) for 3-5 min.
For double immunostaining for HB-EGF and insulin and for PDX-1 and duct cell-specific cytokeratin, the indirect immunofluorescence method was used. The sections, which had been immunostained for HB-EGF or PDX-1 as described above, were further incubated for 30 min with guinea pig anti-porcine insulin antibody or rabbit anti-bovine epidermal cytokeratin antibody (Nichirei Co., Ltd., Tokyo, Japan), respectively. The latter antibody was shown to react selectively to duct cell cytokeratin previously (17). They were then incubated with fluorescein-conjugated goat anti-guinea pig IgG or goat anti-rabbit IgG F(abЈ) 2 (Dako), respectively.
Molecular Cloning of the Human HB-EGF Gene Promoter Region-Recombinant plaques were propagated in Escherichia coli at 37°C, and approximately 8 ϫ 10 5 bacteriophage plaques from the human genome library were screened by standard methods with some modifications, which we described previously (18), using a 32 P-labeled 900-bp EcoRI-SmaI human HB-EGF cDNA fragment ([␣-32 P]dCTP; Amersham Corp.). Hybridization was performed at 37°C in a buffer containing 50% formamide and 10% dextran sulfate, and the filters were washed twice for 10 min at room temperature and three times for 10 min at 48°C in a solution containing 30 mM sodium chloride, 3 mM sodium citrate, pH 7.0, and 0.1% sodium dodecyl sulfate. Hybridizing plaques were detected by autoradiography. DNA from plaque-positive isolates was mapped after digestion with selected restriction enzymes and Southern hybridization with a 32 P-labeled HB-EGF cDNA fragment.
Preparation of Plasmids Used for Reporter Gene Analyses-HB-EGF gene promoter-luciferase reporter constructs were prepared by inserting various 5Ј-flanking sequences of the human HB-EGF gene into a promoterless vector, pA3Luc, which was a generous gift from Dr. W. M. Wood (University of Colorado Health Science Center, Denver, CO) and Dr. D. R. Helinski (University of California San Diego, San Diego, CA). To generate pHB-EGF/Luc880, pHB-EGF/Luc590, and pHB-EGF/ Luc470 plasmids, the 880-, 590-, and 470-bp fragments of the human HB-EGF gene promoter region, respectively, were isolated from the phage DNA using appropriate restriction enzymes (see Fig. 4). After both ends of those DNA fragments were made blunt-ended using T4 DNA polymerase, they were individually linked 5Ј to the luciferase reporter gene at the SmaI site in the pA3Luc plasmid. pA3/Luc0 was the insertionless pA3Luc plasmid used as a negative control. Sitedirected mutagenesis at A element-like sequences located in the 5Јflanking region of the HB-EGF gene was performed by using a Transformer site-directed mutagenesis kit (CLONTECH Laboratories, Inc., Palo Alto, CA). PDX-1 expression plasmid containing nucleotides 221-1151 of the mouse PDX-1 cDNA (16) was prepared as described previously (15).
Cell Culture and DNA-mediated Gene Transfection-Mouse pancreatic beta cell-derived ␤TC1 cells and MIN6 cells and human hepatoblastoma cell line HepG2 cells were grown in RPMI 1640, Dulbecco's modified Eagle's medium, and EMEM medium (Nacalai Tesque, Japan), respectively, supplemented with 10% fetal calf serum (ICN Biomedical, Inc.), 100 units/ml penicillin, and 0.1 mg/ml streptomycin sulfate in a humidified atmosphere of 5% CO 2 at 37°C. Transfection was performed by the calcium phosphate precipitation method followed 4 h later by a glycerol shock (19). HepG2 cells and ␤TC1 cells were replated at a density of 5 ϫ 10 6 /10 cm diameter tissue culture dish 24 h before transfection. Then the cells were co-transfected with 4 g of an HB-EGF promoter-reporter plasmid and 4 g of an internal control, pMSV␤Gal plasmid, which contained the E. coli ␤-galactosidase gene driven by the murine sarcoma virus promoter (20). When required for HepG2 cells, 4 g of the PDX-1 expression plasmid or the empty vector was also co-transfected.
Luciferase and ␤-Galactosidase Assays-Forty-eight hours after the transfection, the cells were harvested for luciferase and ␤-galactosidase assays. Preparation of cellular extracts and luciferase assays were performed using a Pica Gene Kit (Toyo Ink Inc., Tokyo) as we reported previously (19). Light emission was measured by integration over 20 s of reaction using Lumat LB9501 (Belthold, Postfach, Germany). ␤-Galactosidase assays were also performed as described previously (19). The luciferase results were normalized with respect to transfection efficiency assessed from the results of the ␤-galactosidase assays.
Gel-mobility Shift Assay-Nuclear extracts of ␤TC1 cells and MIN6 cells were prepared following a reported procedure (21). Two micrograms of nuclear extract was incubated with 2 g of poly(dI-dC) at 4°C in 20 l of reaction buffer (10 mM HEPES pH 7.8, 0.1 mM EDTA, 75 mM KCl, 2.5 mM MgCl 2 , 1 mM dithiothreitol, and 3% Ficoll). The binding reaction was initiated by adding 5Ј end 32 P-labeled double-stranded oligonucleotide probes and, when required, nonradioactive competitor oligonucleotides, followed by incubation at room temperature for 30 min. In some of the binding assays, anti-PDX-1 antiserum or preimmune serum was added to the binding reactions 1 h before addition of the DNA probes. After the binding reactions, the samples were analyzed by separation on 5% polyacrylamide gel (150 V, 1 h) in 1 ϫ TBE buffer (45 mM Tris, 45 mM boric acid, 1 mM EDTA) followed by autoradiography.

Immunohistochemical Detection of HB-EGF and PDX-1 in
Developing Pancreas-To examine the possible implication of HB-EGF in pancreas development, immunohistochemical analyses of fetal rat pancreata were performed. As shown in Fig. 1, developing endocrine cells had not formed islet-like cell clus- ters at embryonic day 14 (Fig. 1A). At day 20, islet-like endocrine cell clusters could be observed, although they were not clearly separated from the exocrine cells (Fig. 1C).
In terms of HB-EGF expression, positive immunostaining was observed in the primitive ducts or ductular cells in the pancreas at day 14 ( Fig. 1A). At day 20, the islet-like endocrine cell clusters, but not the surrounding exocrine cells, showed positive immunostaining for HB-EGF (Fig. 1C).
The pattern of HB-EGF expression in the developing pancreas appeared to be similar to that of PDX-1 reported previously (22). Also, we found several regions which were similar to the CT box (C/TTAATG, recently renamed the A element), a cis motif known as the PDX-1 binding site, in 5Ј-flanking sequences of the human HB-EGF gene (23). Because these findings suggested a possible involvement of PDX-1 in the HB-EGF gene expression, we performed immunostaining for PDX-1 to find whether HB-EGF protein was co-localized with PDX-1.
The results obtained from two sets of mirror-image sections (Fig. 1, A and B and C and D) revealed that positive staining for both HB-EGF and PDX-1 was detected in primitive ducts or ductular cells and in the putative immature islet cells. Although the difference in cellular localization, cytosolic for HB-EGF and nuclear for PDX-1, made it difficult to do cell-to-cell comparison, the cells with positive immunostaining for HB-EGF (A and C) and for PDX-1 (B and D) in the developing pancreas correspond fairly well to each other.
To investigate the expression of HB-EGF in a later stage of pancreatic development, the same set of experiments was done with neonatal pancreata (3 weeks old; Fig. 2, A and B). To identify islet cells and duct cells in the sections, insulin (C) and duct cell-specific cytokeratin (D) were also immunostained using the same sections as A and B, respectively. The results shown in Fig. 2 revealed that HB-EGF (A) is also expressed in the neonatal pancreas, and the localization pattern seems to be similar to that of PDX-1 (B). The PDX-1 and HB-EGF double positive cells were cytokeratin-positive ductal cells and the endocrine cells, most of which were insulin-producing cells (Fig. 2, A-D). In adult rat pancreata (8 weeks old), the endocrine cells could be recognized clearly as isolated islets. Al-though the intensity of immunostaining was weaker than that in the developing pancreas, they were also positive for HB-EGF (Fig. 3A). Although the physiological significance is not known, clear subcellular localization of PDX-1 in the nucleus was observed in neonatal and adult rat pancreas but not in fetal rat pancreas (Fig. 1, B and D; Fig. 2B; Fig. 3B).
The co-localization of HB-EGF and PDX-1 in the pancreas, along with the existence of regions similar to the PDX-1 binding site (A element) in the 5Ј-flanking sequences of the human HB-EGF gene, suggested that PDX-1 may directly affect HB-EGF gene transcription in the pancreatic endocrine cells of the developing pancreas.
Evaluation of Promoter Activities of HB-EGF Gene in ␤TC1 Cells-To clarify the molecular basis of HB-EGF gene regulation in pancreatic islet cells, we searched for regions within the 5Ј-flanking region of the HB-EGF gene which are important for its promoter activity in pancreatic beta cells. By screening a phage genome library, we isolated and mapped an approximately 12-kilobase genome DNA fragment, containing a 6kilobase fragment of the 5Ј-flanking DNA of the human HB-EGF gene (data not shown). Using this phage clone, reporter gene plasmids containing various lengths of the HB-EGF promoter linked to a luciferase reporter gene were constructed (Fig. 4).
Within the 880-bp 5Ј-flanking DNA contained in the pHB-EGF/Luc880 plasmid, there were four A element-like regions to which PDX-1 may bind (arbitrarily named TAAT1, TAAT2, TAAT3, and TAAT4; Fig. 4). Among them, TAAT2 displayed a perfect match to the previously described CT box (A element) consensus (C/TTAATG), while TAAT4 was a stretch of AT-rich sequences containing repeated TAAT sequences, which are known as an important core motif of the cis-acting element. The pHB-EGF/Luc590 plasmid, containing 590-bp 5Ј-flanking DNA, had only two A element-like regions, including TAAT4 with repeated TAAT sequences (Fig. 4). The pHB-EGF/Luc470 plasmid contained only 470-bp 5Ј-flanking DNA and lacked the A element-like sequences. The basal promoter activities of those 5Ј-flanking DNA of the human HB-EGF gene were evaluated in the beta cell-derived ␤TC1 cells and hepatoblastomaderived HepG2 cells. Although both cells express the HB-EGF gene (data not shown), the former cells express PDX-1 but the latter cells do not. The results (Fig. 5B) indicated that, in HepG2 cells, all the fusion plasmids containing 470 bp or longer segments of 5Ј-flanking DNA induced comparable luciferase activity, which was well above the background (pA3/Luc0). However, in ␤TC1 cells (Fig. 5A), a stepwise reduction of the promoter activity was observed; the pHB-EGF/Luc470 plasmid containing no A element-like motifs displayed a promoter activity barely detectable over the background (pA3/Luc0). These observations indicate that certain cis motifs located between Ϫ880 and Ϫ590 and also between Ϫ590 and Ϫ470 function in a cell type-dependent manner; i.e. they are important for gen- erating promoter activity in ␤TC1 cells but not in HepG2 cells.
Effect of PDX-1 Overexpression on HB-EGF Gene Promoter Activity-To discuss whether PDX-1 would be a possible factor that contributes to the cell type-dependent promoter activity of the HB-EGF gene, we investigated whether the HB-EGF gene is a potential target of PDX-1 transactivation. Exogenous expression of PDX-1 in HepG2 cells was induced by transient transfection of a PDX-1-expressing plasmid, and the effect on the HB-EGF promoter activity was evaluated. As shown in Fig.  6, PDX-1 expression in HepG2 cells caused a 20-fold increase in the promoter activity of the 880-bp promoter (pHB-EGF/ Luc880). A slightly weaker induction (12-fold) was observed with the 590-bp promoter (pHB-EGF/Luc590). However, for the 470-bp 5Ј-flanking DNA which contained no A element-like sequences (pHB-EGF/Luc470), the PDX-1-responsive induction of the promoter activity was trivial, if any. These results indicate that the human HB-EGF promoter is PDX-1-responsive and suggest that the A element-like sequences located in the promoter may be important for the responsiveness.

Evaluation of Roles of A Element-like Sequences in the HB-EGF Gene
Promoter-To examine whether the A element-like sequences function as cis-active elements, we disrupted those sequences and evaluated the effects on the basal promoter activity and PDX-1-induced promoter activation. Using the pHB-EGF/Luc 880 plasmid as a template, five mutant reporter plasmids, in which one or two of the four A element-like sequences were disrupted, were prepared (Mut1-5; Fig. 7).
As shown in Fig. 8, Mut2 and Mut4 mutants, in which TAAT2 or TAAT4 is disrupted, respectively, revealed reduced basal promoter activity as evaluated in the beta cell-derived ␤TC1 cells, whereas Mut1 or Mut3 mutations showed no effects. The Mut5 mutant, in which both TAAT2 and TAAT4 are disrupted, revealed further decrease in the promoter activity.
Using the same mutant reporter plasmids, the PDX-1 responsiveness was also examined. As described above, PDX-1 expression in HepG2 cells caused a 20-fold increase in the promoter activity of the 880-bp promoter (pHB-EGF/Luc880, Fig. 6). The magnitude of the induction was lower in Mut2 and Mut4 mutants but not in the Mut1 or Mut3 mutant (Fig. 9). When the reporter plasmid had both TAAT2 and TAAT4 mutations (Mut5), the promoter activity was not substantially induced by PDX-1 overexpression (Fig. 9). These results together indicate that there are two A element-like sites which play a major role in generating promoter activity in ␤TC1 cells and in the PDX-1 responsiveness of the promoter: TAAT2, which matches the CT box (A element) consensus (C/TTAATG), and TAAT4, the repeated TAAT sequences.
Identification of A Element-like Region-binding Factor in Beta Cell-derived Cells-Because two of the A element-like regions (TAAT2 and TAAT4) in the HB-EGF gene promoter were important for the PDX-1 responsiveness (Fig. 9), as well as for the basal promoter activity in ␤TC1 cells (Fig. 8), it seemed likely that PDX-1, which is expressed throughout beta cell development (22) (Figs. 1-3), binds to those regions and thus plays a primary role in generating promoter activity in the cells. This led us to perform gel-mobility shift analyses for those two A element-like regions using an anti-PDX-1 antibody. Nuclear extracts were isolated from two pancreatic beta cell-derived cell lines, ␤TC1 cells and MIN6 cells, and allowed to bind to two double-stranded oligodeoxynucleotide probes representing each of the two functional A element-like regions (Figs. 7-9) in the promoter. The results of the gel-mobility shift analyses (Fig. 10, A and C) revealed that there is one major nuclear factor in ␤TC1 cells and MIN6 cells which binds specifically to either of the A element-like regions (TAAT2 or TAAT4). Because the gel-shift complexes were supershifted by addition of an anti-PDX-1 antibody (Fig. 10, B and D), PDX-1 seems to be the dominant DNA-binding factor for the A element-like regions (TAAT2 and TAAT4) in the HB-EGF gene promoter in ␤TC1 cells and MIN6 cells. Since both TAAT2 and TAAT4 were shown to be important for the promoter activity, PDX-1 may play an essential role in the transcriptional activation of the HB-EGF gene in the endocrine pancreas. DISCUSSION In the present study, we identified the expression of HB-EGF in the developing pancreas. To date, various growth factors have been shown to be expressed in the pancreas and the surrounding mesenchyme-derived cells during development and also in adults. They include hepatocyte growth factor, expressed in mesenchyme-derived tissue, and EGF, TGF-␣, betacellulin, nerve growth factor (NGF), and transforming growth factor-␤1 expressed in the pancreas itself (2,3,24,25). The growth factors expressed in the developing pancreas may function independently or in cooperation with other growth factors. Among them, most of the EGF family growth factors share a common receptor of the ErbB receptor family, EGF receptor (ErbB1), which has been known to be expressed also in the endocrine and exocrine pancreas and is thought to exert FIG. 7. Schematic representation of site-directed mutagenesis of A element-like sequences. Mutations were introduced to the A element-like sequences in the HB-EGF gene promoter within the pHB-EGF/Luc880 plasmid (Wild-type) and five mutant reporter plasmids (Mut 1-Mut 5) were obtained. In Mut1, Mut2, Mut3, and Mut4 plasmids, TAAT1, TAAT2, TAAT3, and TAAT4 was disrupted, respectively. In Mut5, both TAAT2 and TAAT4 were disrupted. Substituted nucleotides are underlined.

FIG. 8. Effects of mutations in A element-like regions on promoter activity in ␤TC1 cells. Effects of disruption of A element-like
regions on the basal promoter activity in ␤TC1 cells were evaluated. A wild-type or mutant reporter plasmid was co-transfected with the pMSV␤gal plasmid, and 48 h after the transfection, luciferase and ␤-galactosidase assays were performed. The data are expressed as relative light units with that of the wild-type plasmid arbitrarily set at 1. The luciferase results were normalized with respect to transfection efficiency using the results of ␤-galactosidase assays. All data are presented as means Ϯ S.D. of three independent experiments.

FIG. 9. Effects of mutations in A element-like regions on PDX-1-responsive promoter activation.
Effects of disruption of A element-like regions on the PDX-1 responsiveness of the HB-EGF gene promoter were evaluated. A wild-type or mutant reporter plasmid was co-transfected into HepG2 cells with PDX-1 expression plasmid or the control plasmid (empty vector). To allow normalization of the luciferase results, the pMSV␤gal plasmid was also co-transfected. Forty-eight hours after the transfection, cellular extracts were obtained, and luciferase and ␤-galactosidase assays were performed. The results are expressed as n-fold induction over the basal promoter activity of each reporter plasmid obtained in HepG2 cells transfected with the control plasmid (empty plasmid). The luciferase results were normalized with respect to transfection efficiency using the results of ␤-galactosidase assays, and the data are presented as means Ϯ S.D. of three independent experiments. autocrine or paracrine effects. EGF seems to promote ductal development in the fetus (3) and increases the cell proliferation rate of fetal pancreatic cells (24). Also, another member of the EGF family, betacellulin, which was originally isolated from the conditioned medium of beta cell-derived cells (26), seems to function as an inducer of beta cell differentiation (27,28). Our present study has identified another EGF family growth factor expressed in the pancreas. During pancreas development, HB-EGF is expressed abundantly in ductal cells and immature endocrine cells, which were considered to be differentiated from ductal cells to form islets, but is not expressed in exocrine tissues.
The expression pattern of HB-EGF during pancreas development is clearly in contrast to that of TGF-␣, which can be detected in both endocrine and exocrine tissues (2), and also with NGF localized in nonendocrine cells surrounding the islets (25). Although the EGF receptor, a common receptor for HB-EGF and TGF-␣, is expressed in both endocrine and exo-crine tissues, the expression pattern of HB-EGF in the developing pancreas suggested that HB-EGF is likely to be exclusively involved in the development of the endocrine pancreas. It is difficult to clarify the in vivo physiological role of HB-EGF; however, evaluation of its effects using different tissue components of the developing pancreas, as previously done for EGF or other growth factors (3), may provide useful information. Also, mice homozygous for a targeted mutation in the HB-EGF gene, when they become available, would be useful for clarifying the physiological significance of HB-EGF expression during pancreas development.
Our present study has also shown that the expression of HB-EGF can be potentially regulated by the transcription factor PDX-1. It was originally isolated as IPF1 in mouse (16) and as STF-1/IDX-1 in rat (29,30) and has been shown to bind to the A elements of insulin gene and to activate its transcription (1,16,22,31,32). PDX-1 is selectively expressed in pancreatic islets and in the duodenum. At an early stage of embryonic development, PDX-1 is initially expressed in the gut region when the foregut endoderm becomes committed to common pancreatic precursor cells. During the development of the pancreas, PDX-1 expression is maintained in multipotential precursors that coexpress several hormones, and later it becomes restricted to beta cells. Mice homozygous for a targeted mutation in the PDX-1 gene have been shown to lack a pancreas (1,33,34). These observations support the crucial role of PDX-1 in pancreas development.
Although data are accumulating for the expression pattern of PDX-1 during development, it is largely unknown how PDX-1 drives pancreas development. Because PDX-1 is a DNA-binding transcription factor, its action should be exerted by transactivating genes through its binding to their regulatory sequences. To date, only four beta cell-specific genes, insulin, glucokinase, IAPP, and Glut2 genes, all of which have the A element-like motifs in their regulatory region, are known as putative targets of the PDX-1 regulation (15,28,(35)(36)(37). However, the simple induction of these four genes is unlikely to be enough to explain all of the in vivo effects of PDX-1. Recent data obtained with PDX-1-deficient mice showed that PDX-1 is essential for induction of the morphogenesis of pancreatic epithelium as well as the progression of differentiation of the endocrine cells (34). These observations cannot be rationally explained by the already known function of PDX-1 as a transcription factor of those four genes. Therefore, we are tempted to consider that PDX-1 may induce gene expression of one or more growth factors and that the growth factor(s) may, in turn, exert physiological effects on pancreas development. Thus, HB-EGF, which was shown to co-localize with PDX-1 during development and also revealed potential PDX-1 responsiveness in its promoter activity, may be a good candidate for such growth factors which would mediate PDX-1 effects.
Since AT-rich regions are known to be binding sites for a wide range of homeoproteins, we assume it unlikely that PDX-1 is the only transcription factor involved in HB-EGF gene transcription in the pancreas. To date, various homeodomain-containing transcription factors as well as basic helix-loop-helix proteins, which are either beta cell-specific or ubiquitous, are known to be expressed in pancreatic islets. To fully understand the regulation of HB-EGF gene expression during pancreas development, the possible involvement of those factors in HB-EGF gene activation needs to be studied.
In conclusion, HB-EGF is expressed in the developing pancreas and may be involved in the differentiation or growth of endocrine cells. Also, in our present study, PDX-1 was suggested as being a possible regulator of the HB-EGF gene expression in the pancreas. PDX-1-responsive induction of the FIG. 10. Analyses of protein bindings to A element-like sequences in HB-EGF gene promoter. Gel-mobility shift analyses were performed using nuclear extracts obtained from mouse beta cellderived ␤TC1 cells (A and B) and MIN6 cells (C and D). For A and C, two kinds of double-stranded oligonucleotides were used as binding probes: probe a (sequence of sense strand: 5Ј-GTTCAAAAATCCTAATGTC-CCCTG-3Ј) representing TAAT2 (Fig. 7), and probe b (sequence of sense strand: 5Ј-GATAGATGAATTAATTATTATTTGATACTTGC-3Ј), representing TAAT4 (Fig. 7). The 32 P-labeled binding probe was incubated with the nuclear extracts and, when required, 50-fold excess of unlabeled wild-type or mutated-type competitor was added to the binding reactions. The sequences of the sense strand of the mutated-type competitors were 5Ј-GTTCAAAAATCCTCCTGTCCCCTG-3Ј for probe a and 5Ј-GATAGATGAATTCCTTCTTCTTTGATACTTGC-3Ј for probe b. For B and D, gel-mobility shift analyses were performed as described for A and C except that an anti-PDX-1 antiserum or preimmune serum was added to the binding reactions before addition of the binding probe. Products were electrophoresed at 150 V for 1 h on 5% polyacrylamide gels in 1 ϫ TBE buffer, and dried gels were analyzed by autoradiography.
growth factor gene expression, if it occurs in vivo, may at least partially explain the essential and extensive roles of PDX-1 in the morphogenesis of the pancreas, as suggested by phenotypes of PDX-1-deficient mice.