Hepatocyte nuclear factor-1α regulates transcription of the guanylin gene

J. A. Hochman, D. Sciaky, T. L. Whitaker, J. A. Hawkins, M. B. Cohen

Abstract

To study the molecular mechanisms controlling guanylin expression, we have cloned the mouse guanylin gene, including 2.7 kb of upstream sequence. We show that the first 133 base pairs (bp) of the upstream guanylin promoter are sufficient to drive near maximal (6-fold over basal) luciferase reporter gene expression in Caco-2 intestinal cells; at least 300 bp of upstream promoter are required for reporter gene expression in HT-29 intestinal cell lines. Using electromobility shift assays, we demonstrate that nuclear proteins bind to the hepatocyte nuclear factor-1 (HNF-1) consensus sequence in the guanylin promoter. The HNF-1 consensus sequence, located in the immediate 5′ flanking region, is required for transcriptional activation of the guanylin gene in both intestinal cell lines. Mutagenesis of the HNF-1 consensus sequence abolishes transcriptional activation of guanylin promoter-luciferase reporter gene constructs. Cotransfection of these constructs with HNF-1α augments transcriptional initiation of the reporter gene. In contrast, HNF-1β has no significant effect on transcription of the reporter gene. These experiments demonstrate that HNF-1α is an important regulatory element in the transcriptional activation of guanylin.

  • Caco-2 cells
  • HT-29 cells
  • guanylate cyclase

the identification of guanylin (9), which is 50% homologous to the heat-stable enterotoxin (STa) elaborated by enterotoxigenic Escherichia coli, provides a potential explanation for the endogenous function of the intestinal STa receptor, guanylate cyclase C (GCC). Guanylin is thought to be involved in fluid and electrolyte balance in the intestine through the same intracellular pathway as STa; that is, guanylin, like STa, binds to GCC (9). This binding results in increased guanosine 3′,5′-cyclic monophosphate production, which activates the cystic fibrosis transmembrane conductance regulator (CFTR) via a protein kinase intermediary (12, 26, 35, 46). Activation of the CFTR, in turn, results in increased chloride secretion, decreased sodium and chloride absorption, and possibly bicarbonate secretion (13).

Guanylin mRNA has been shown to be distributed in a complex, region-specific pattern within the human intestine (15), a pattern resembling that of its receptor GCC (18). Signal is present in the superficial epithelial cells of the human colon, villous cells of the human ileum, and crypt cells throughout the small intestine (15). Based on the cellular localization of guanylin mRNA, we tested the hypothesis that cis-active elements were present in the mouse guanylin promoter that could direct expression of the guanylin gene. Previous analysis of the guanylin gene (Guca2) revealed a number of similarities between the mouse and human gene; both genes are ∼1.7 kb and are composed of three exons (14, 42). To determine the molecular mechanisms governing guanylin expression in the intestine, we have cloned additional upstream sequence of the mouse guanylin gene. Using various human colonic cell lines that express guanylin, we now show that the mouse guanylin promoter is transcriptionally active in these cell lines.

A minimal (−133 bp) promoter is transcriptionally active in Caco-2 cells. This immediate 5′ flanking sequence is an evolutionarily conserved region of the guanylin gene with 76% homology to the human promoter region. The mouse promoter contains a putative binding site for hepatocyte nuclear factor-1 (HNF-1) at −53 to −41, an element that is identical between human and mouse guanylin genes.

HNF-1 proteins, including HNF-1α and HNF-1β, belong to a class of transcription factors that are important in the transcriptional activation of many genes, including albumin and α-1-antitrypsin (8,28). HNF-1α and HNF-1β are also known as HNF-1 and vHNF-1 or LF-B1 and LF-B3 (29, 38). Therefore, based on our initial data and our analysis of the promoter region, we hypothesized that the HNF-1 binding site is necessary for the transcriptional activation of the guanylin gene.

EXPERIMENTAL PROCEDURES

Tissue culture cell lines.

The Caco-2 human colon carcinoma cell line was cultured as previously described (6, 25). HT-29 cell lines used included HT-29-18-N2 and HT-29-18-C1 subclones (gifts of Cynthia Sears), which are committed to differentiate to specific cell types (16, 20). HT-29-18-N2 cells are mucin-secreting, goblet cell-like, whereas the HT-29-18-C1 cells are enterocyte-like. In addition, we used the undifferentiated cell line HT-29-CP [gift of John Barnard (3)]. All cell lines were routinely grown in Dulbecco’s modified Eagle’s medium containing 25 mM glucose, 50 IU/ml streptomycin, 10 μg/ml penicillin, and 10% fetal calf serum. In addition, HT-29 cell lines were supplemented with 10 μg/ml of human transferrin and 4 mM glutamine (16, 32). The 293 human embryonic kidney cells were cultured according to the recommendations of the American Type Culture Collection (Rockville, MD).

Isolation of promoter sequence.

Inverse polymerase chain reaction (PCR) (33) was initially used to obtain additional upstream sequence. Briefly, mouse genomic DNA (strain 129/Sv) was digested with BamH I. The DNA fragments were ligated to themselves via the BamH I compatible ends, and PCR was performed using oligonucleotides F3 MSP PCR INV (5′-CTGTTGAGCCCCATCAGATAAGC-3′) and B16 MSP PCR INV (5′-CCAGCCAGCCATATTTTCCC-3′). A 1,100-bp fragment was generated and ligated into pCRII (Invitrogen). Four separate clones were sequenced to discern any infidelity in the PCR product. Using this inverse PCR product (−890 to −78 and 232 to 557) of the mouse guanylin gene, we screened a λDASH IISau3A partial 129/Sv mouse genomic library (gift of Marcia Shull) by hybridization as previously described (42). Two clones, λ2mgg2 and λ2mgg5, were isolated, and restriction fragment digests of DNA from these clones yielded fragments of the predicted molecular weight based on genomic Southern blot analysis. The clone λ2mgg5 was identified as containing the most upstream sequence by Southern hybridization of an EcoR I, BamH I, or Hind III digest when hybridized with the inverse PCR product. DNA from λ2mgg5 was digested with either EcoR I or BamH I and cloned into the respectiveEcoR I or BamH I site of pTZ18U (gift of Charlotte Paquin). Subclones containing the appropriately sized fragments as determined by restriction digestion (not shown) were further verified by sequencing with the oligonucleotide P1 (5′-CAATGTGAATACCTCCCTG-3′). Additional sequence beyond −890 bp and confirmation of the inverse PCR sequence were performed using sequential oligonucleotides as primers. Sequence analysis was performed using MacVector (Kodak).

Cloning of promoter constructs into luciferase reporter gene vector.

The initial reporter gene construct consisted of 133 bp of upstream promoter sequence linked to the 5′ untranslated sequence cloned into the pGL3-basic (Promega). To accomplish this a 2.7-kbEcoR I fragment containing 133 bp of upstream sequence and the entire Guca2 gene were gel purified from the lambda clone λmgg10 (42). This EcoR I fragment was digested with BamH I, and a 689-bp fragment containing 133 bp of upstream sequence was cloned intoEcoR I/BamH I digested pTZ18U. A fragment containing 133 bp of upstream sequence and 5′ untranslated region was isolated by digestion of the 689-bp EcoR I/BamH I fragment withSau3A andNla III. The largest fragment (172 bp) of the digest containing the sequences from −133 to +39 was gel purified and cloned into pIC20R (27) using aSau3A compatibleBgl II site (reconstructing theBgl II site) and anNla III compatibleSph I site. TheNlaIII/Sph I fusion in pIC20R was converted to an Nco I site by PCR. The fragment containing this minimal promoter region was digested withBglII/Nco I and cloned into theBglII/Nco I site of pGL3-basic. The fidelity of this construct was confirmed by sequencing. Thus the intact relationship between upstream promoter sequence and the 5′ untranslated sequence of guanylin is retained in the luciferase reporter construct. Convenient restriction sites were used to subclone increasing lengths of genomic fragments into the reporter construct. To fuse the −839 bp guanylin promoter to luciferase, aBglII/Spe I (−133 to −2) digest of the pGL3/−133 plasmid was excised and replaced with aBamH I/Spe I (−839 to −2) fragment of the λ2mgg5 subclone. To fuse the −1,886 bp guanylin promoter to luciferase, anXhoI/Spe I (vector to −2) fragment of λ2mgg5 was subcloned and inserted intoXhoI/Spe I prepared pGL3/−133. To fuse the −300 bp guanylin promoter to luciferase, aBstXI/Sma I (−300 to −839) fragment of the pGL3/−839 construct was removed; the remaining plasmid was ligated to yield the pGL3/−300 construct. The fidelity of these constructs was confirmed by restriction digest analysis.

HNF-1 site-directed mutagenesis.

Two vectors were constructed that contained a 3-bp substitution in the consensus HNF-1 sequence. To construct the pGL3/−133ΔHNF-1 vector, a 175-bp PCR fragment of the murine guanylin promoter from −133 to +42 was directionally subcloned into aXba I/Hind III site of pAlterI (Promega). A mutagenic oligonucleotide (5′-CAAGGCCCCAGGC̅A̅TAG̅TGAGTAACCCC-3′), which altered the HNF-1 consensus site, was used. The altered bases are underlined. Mutagenesis was performed according to the manufacturer’s instructions (Promega). Screening of plasmid colonies for incorporation of the desired mutation employed restriction analysis, taking advantage of the elimination of a Dde I site at the mutation. After the fidelity of the construct was confirmed by sequencing, the 175-bp fragment was subcloned into theNcoI/Bgl II site of pGL3-basic (Promega), yielding the pGL3/−133ΔHNF-1 construct. The pGL3/−300ΔHNF-1 construct was made by isolating anXcmI/Xcm I fragment of the pGL3/−133ΔHNF-1 construct and subcloning this into theXcmI/Xcm I site of the pGL3/−300 construct. Restriction digest analysis confirmed proper orientation and size of the construct. Mutagenesis of this site did not affect the AP1 site (GTAACC), which overlaps the HNF-1 site on the 3′ end. The bases modified from the wild-type HNF-1 site to yield the altered ΔHNF-1 site are underlined: 5′-G̅T̅TAC̅TGAGTAAC-3′.

Northern analysis.

Total RNA was extracted from tissue culture cells and from human intestinal tissues using acid guanidine isothiocyanate-phenol-chloroform extraction (5). Total RNA (20 μg) was fractionated by electrophoresis in a 1.5% agarose-1.9% formaldehyde gel, transferred to a nylon membrane (MagnaGraph, MSI, Westboro, MA) by capillary action, and crosslinked to the membrane using a Stratalinker (Stratagene, La Jolla, CA). The following cDNA probes, radiolabeled with [32P]CTP by random primer DNA synthesis, were used: a human guanylin fragment isolated from pMON 22305 (47), an HNF-1α fragment isolated from pBJ5 (gift of Gerald Crabtree), an HNF-1β fragment isolated from pBJ5 (gift of Gerald Crabtree), and a human GCC receptor fragment (23). The blots were hybridized under stringent conditions as previously described (19,21). Northern blots previously hybridized with these probes were rehybridized with a labeled oligonucleotide complementary to 18S ribosomal RNA (25) to quantitate relative amounts of total RNA loaded in each lane of the gels as previously described (19). Visualization and quantitation of positive signals were accomplished with the PhosphorImager system (Molecular Dynamics, Sunnyvale, CA).

Nuclear extract preparation.

Nuclear proteins were isolated by a modification of the method described by Schreiber et al. (41). Briefly, confluent cells in 75 cm2 flasks were trypsinized and suspended in 10 ml of freshly warmed media. After pelleting by centrifugation at 1,500 g for 5 min, the cells were washed with 10 ml of tris(hydroxymethyl)aminomethane-buffered saline (TBS) and then repelleted. The washed cells were resuspended in 1 ml of TBS, transferred to a 1.5-ml Eppendorf tube, and repelleted by spinning for 15 s in a microcentrifuge. The pellet was resuspended in ice-coldbuffer A [(in mM) 10N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid (HEPES), pH 7.9, 10 KCl, 0.1 EDTA, 0.1 ethylene glycol-bis(β-aminoethyl ether)-N,N,N′,N′-tetraacetic acid (EGTA), 1 dithiothreitol (DTT), and 0.5 phenylmethylsulfonyl fluoride (PMSF)]. After incubation on ice for 15 min, 25 μl of 10% Nonidet P-40 (Sigma) were added. The mixture was vortexed for 10 s and centrifuged for 30 s (14,000 g). The resulting nuclear pellet was resuspended in 50 μl of coldbuffer B [(in mM) 20 HEPES, pH 7.9, 0.4 NaCl, 1 EDTA, 1 EGTA, 1 DTT, and 1 PMSF]. After vigorous rocking on a shaking platform for 15 min at 4°C, the extract was centrifuged for 5 min in a Microfuge (14,000g). Ten microliters (∼5 μg) of this supernate were used for electromobility shift assays (EMSAs).

EMSAs.

For EMSAs, ∼100,000 counts/min of a32P-labeled guanylin promoter oligonucleotide (−64 to −30) containing the HNF-1 consensus sequence was added into four reaction mixtures. The first reaction, a negative control, contained a 1× gel shift binding buffer (Promega) without nuclear extract. In the remaining three reactions, nuclear extract was added with or without unlabeled competitor DNA oligonucleotides (∼100-fold excess). The reactions were allowed to incubate for 20 min before electrophoresis on a 4% nondenaturing acrylamide gel in 0.5× TBE buffer (45 mM Tris-borate, 1 mM EDTA).

Transient transfection assays.

Cells were transfected 24 h after subculture into six-well plates at ∼50–60% confluency. For Caco-2 and 293 cells, DNA (0.36 nM pGL3 vector with or without insert and 0.4 μg pSV40βgal to normalize for transfection efficiency) was suspended in 100 μl OPTI-MEM (GIBCO-BRL). For cotransfection experiments, 0.2–0.8 μg of either HNF-1α or HNF-1β in pBJ5 was added to the DNA suspension. Then 10 μl of lipofectin (GIBCO-BRL) were diluted in 100 μl OPTI-MEM, added to the DNA suspension, and incubated at room temperature for 10 min. This suspension was diluted to 1 ml with OPTI-MEM and added to the six-well plates. After 24 and 48 h, fresh media replaced the previous suspension. For HT-29 cell lines, a similar protocol was followed except cellfectin (GIBCO-BRL) was substituted for lipofectin and pCMVβgal or pRSVβgal was used to control for transfection efficiency. A 20-μl cell lysate was analyzed for both luciferase (Promega) and β-galactosidase (Tropix) according to the manufacturer’s instructions. Both enzyme activities were measured in a Berthold Lumat LB9501 luminometer. Reporter gene expression was calculated after a minimum of three transfection experiments with each construct.

RESULTS

Isolation of promoter sequence.

A 4.8-kb EcoR I fragment from the λmgg5 clone, whose size was predicted from Southern blots of mouse genomic DNA probed with guanylin, was subcloned into pTZ18U. The sequence of the promoter contained within this fragment is shown in Fig.1; also shown are the corresponding sequences of the human guanylin promoter up to −331 (14, 34). Matches to the consensus sequences of known regulatory elements (10) are also shown in Fig. 1. We had previously demonstrated by primer extension analysis the presence of a transcriptional start site spanning 3 bp at the beginning of exon 1 of the murine guanylin gene (42). There is a conserved element between −31 and −25 (TTTAAAA) between mouse and human, which was suggested as a TATA box (34). Also, sequence analysis reveals the presence of a hepatocyte nuclear factor (HNF-1) site at nucleotides −53 to −41. This sequence is also present in the human guanylin promoter (Fig. 1). In both human and mouse, this binding site has 11 of 13 (underlined) nucleotide matches with the HNF-1 consensus binding sequence (5′-G̅T̅T̅A̅AT̅G̅/̅T̅A̅ A/T T̅N̅A̅C̅-3′).

Fig. 1.

Nucleotide sequence of the promoter region of the mouse guanylin gene. Nucleotides are numbered with respect to the major transcription initiation site (+1). Sequence comparisons to the human guanylin promoter (6) and to previously identified consensus sequences of known regulatory elements are given. The nucleotide sequence data in this paper have been assigned the accession number U60528 in the NCBI GenBank database.

Northern analysis.

We evaluated several human intestinal epithelial cell lines for the presence of guanylin mRNA. As shown in Fig.2, the 650-bp guanylin mRNA is detectable by day 2 after Caco-2 cell subculture. Also, levels of guanylin mRNA increase after subculture, paralleling the increase previously reported in Caco-2 cellular differentiation and GCC expression (23). In addition, guanylin mRNA expression was seen in those HT-29 cell subclones committed to differentiate to either an enterocyte-like (HT-29-18-C1) or goblet cell-like (HT-29-18-N2) phenotype but not in the undifferentiated HT-29-CP cell line (Fig.2). In contrast, guanylin mRNA was not detectable in poly(A)+ RNA isolated from 293 cells (Fig. 2).

Fig. 2.

Northern blot analysis of guanylin expression.A: total RNA (20 μg) from Caco-2 cells at various times after subculture (2–28 days) was hybridized with a fragment from pMON 22305 encoding human guanylin cDNA. Migration of 28S and 18S ribosomal markers is shown for comparison. Blots were reprobed with a labeled oligonucleotide complementary to 18S ribosomal RNA to demonstrate equal loading of gel lanes.B: poly(A)+ RNA (10 μg) from 293 human embryonal kidney cells was hybridized with the same pMON 22305 probe. Human colon RNA (5 μg) is shown as a control.C: total RNA (20 μg) from several HT-29 subclones was probed with the same pMON 22305 probe. Human jejunal RNA (5 μg) is shown as a control.

To examine the possible role of the HNF-1 consensus binding sequence in guanylin gene expression, we scanned several tissue culture lines to determine which lines express HNF-1α and HNF-1β. As shown in Table1, both HNF-1α and HNF-1β mRNA were detected in human intestinal cell lines that express guanylin mRNA, including Caco-2 cells and differentiated HT-29 cells. Neither HNF-1α nor HNF-1β mRNA expression was seen in the nonguanylin mRNA-expressing human embryonal 293 kidney cells or the 3T3 fibroblast cell line. A notable difference between the two HNF-1 proteins occurs in the undifferentiated human intestinal HT-29-CP and the COS kidney cell lines. These cell lines expressed HNF-1β but not HNF-1α or guanylin mRNA.

View this table:
Table 1.

HNF-1α and HNF-1β mRNA expression as determined by Northern analysis

EMSAs.

The ability of the nuclear proteins to bind to the guanylin promoter oligonucleotide containing the HNF-1 consensus sequence was examined by EMSAs. As shown in Fig. 3, extracts from two guanylin-expressing intestinal cell lines, Caco-2 and HT-29-18-C1, contained factors that bound the HNF-1-guanylin promoter oligonucleotide probe. The nuclear protein-probe complex binding was specifically inhibited by addition of 100-fold excess of unlabeled HNF-1 competitor DNA but not by the addition of 100-fold excess of nonspecific oligonucleotide DNA (T cell receptor-β). In contrast, extracts from the non-guanylin-expressing 293 cell line did not bind to the HNF-1 probe. Minor nonspecific binding of the oligonucleotide probe occurred in all three EMSAs.

Fig. 3.

Hepatocyte nuclear factor-1 (HNF-1) proteins bind to the guanylin promoter oligonucleotide. Nuclear extracts from Caco-2, HT-29-18-C1, or 293 cells were used in electromobility shift assays (EMSAs). In each EMSA, lane 1 is a negative control with no nuclear extract. In lane 2, in which nuclear extract is added, a protein-probe complex is present in the Caco-2 and HT-29-18-C1 EMSAs but not in the 293 EMSA. Inlane 3, in which a 100-fold excess of cold competitor double-stranded HNF-1 oligonucleotide DNA (5′-CCCAGGGTTACTGAGTAACCCCAA-3′) is added, the protein-probe complex disappears. In lane 4, in which a nonspecific oligonucleotide cDNA, T cell receptor-β (5′-TGTCAAACCACATCCTGTTGTGAC-3′) is added, the complex is not competed.

Transient transfection assays.

The transcriptional activity of the guanylin promoter in three guanylin mRNA-expressing intestinal cell lines was examined by transient transfection as shown in Fig. 4. Transfection results are expressed relative to the luciferase activity seen with the pGL3-SV40 promoter, which encodes the luciferase gene under the control of the SV40 promoter. Results of basal activity for each cell line, obtained by transfection with pGL3-basic, are also shown.

Fig. 4.

Functional analysis of 5′ deletions of the mouse guanylin promoter. Relative luciferase activity of the mouse guanylin promoter-luciferase fusion constructs in Caco-2, HT-29-18-C1, and HT-29-18-N2 cells. The constructs tested are denoted along they-axis with the pGL3-basic vector denoted as “LUC” without attached promoter. Data are expressed as a percentage of the activity (arbitrarily assigned as 100%) of the pGL3-SV40 positive control and are normalized for transfection efficiency (using β-galactosidase). Values are means ± SE;n ≥ 3.

Reporter gene expression in these human intestinal cell lines was transcriptionally activated by various lengths of the guanylin promoter. In Caco-2 cells, the minimal promoter containing −133 to +39 bp was sufficient to obtain near maximal reporter gene expression (6.0 ± 0.5-fold over basal). However, in the HT-29 cell lines, an additional length of upstream sequence beyond the first 133 bp was required for reporter gene expression above basal levels. In all three intestinal cell lines, the −300 construct was sufficient to initiate transcription. In the non-guanylin mRNA-expressing cell lines, 293 kidney cells and HT-29-CP intestinal cells, no increase over basal reporter gene expression was seen with any length of the guanylin promoter used (data not shown).

Cotransfection with HNF-1α or HNF-1β.

Having demonstrated that the 5′ flanking sequence of the guanylin gene was transcriptionally active in a number of cell lines, we wished to determine the functional significance of the HNF-1 binding site. The effects of cotransfection with HNF-1α or HNF-1β are shown in Figs.5 and 6. In the top panels of Fig. 5, A-D, the effect on reporter gene expression of a 3-bp substitution in the HNF-1 site is displayed. In the Caco-2 cell line, mutagenesis of the HNF-1 site reduced reporter gene expression to basal levels with the −133ΔHNF-1 construct. In the 293 kidney cell line, no increase over basal reporter gene expression was seen with the wild-type or the −133ΔHNF-1 constructs. In the two HT-29 cell lines, which require a longer promoter construct for transcriptional activation, alteration of the HNF-1 sequence in the −300 bp construct abolished reporter gene expression.

Fig. 5.

Role of the HNF-1 site on the transcriptional activation of the mouse guanylin promoter. Relative luciferase activity of three mouse guanylin-luciferase fusion constructs (pGL3-basic, pGL3/−133ΔHNF-1, and pGL3/−133) is shown for 4 cell lines. A: Caco-2 cells.B: 293 kidney cells.C: HT-29-18-C1 cells.D: HT-29-18-N2 cells. Data are expressed as described in Fig. 4. The bottom panel for each cell line represents the same constructs cotransfected with pBJ5/HNF-1α (0.4 μg in Caco-2, 0.3 μg in the other cell lines).

Fig. 6.

Dose-response relationship of cotransfection with HNF-1α or HNF-1β. Relative luciferase activity of the mouse guanylin −133 bp luciferase fusion construct with varying amounts of pBJ5/HNF-1α or pBJ5/HNF-1β (0 to 0.8 μg, with 0 serving as the control) in Caco-2 cells. Data are expressed as described in Fig. 4. Data in this experiment were also normalized to the appropriate controls (not shown) including pGL3-basic, pGL3-basic + pBJ5/HNF-1α, and pGL3-basic + pBJ5/HNF-1β.

In the bottom panels of Fig. 5, A-D, the effect on reporter gene expression of cotransfection with HNF-1α is shown. In each panel, cotransfection with HNF-1α resulted in an augmentation of reporter gene expression (the 3 intestinal cell lines) or acquisition of reporter gene expression (293 kidney cell line) when an intact or wild-type guanylin promoter-luciferase construct was used. In the Caco-2 cell line, HNF-1α cotransfection increased reporter gene expression to 195% of the constitutive SV40 promoter (21.7 ± 1.7-fold over basal). Similarly, in the HT-29-18-C1 cell line, HNF-1α cotransfection increased reporter gene expression to 62% of the constitutive promoter (12.4 ± 1.8-fold over basal); an increase to 75% of the constitutive promoter activity (25.0 ± 3.3-fold over basal) was seen in the HT-29-18-N2 cell line. In contrast, no increase over basal reporter gene expression was seen when HNF-1α was cotransfected with the mutagenized ΔHNF-1 promoter constructs.

Additional experiments studying a dose-response relationship between cotransfection of the HNF-1 proteins and reporter gene expression in Caco-2 cells are shown in Fig. 6. A dose-related increase in reporter gene expression is seen with increasing quantity of HNF-1α cotransfection (0.2–0.8 μg). The maximal reporter gene expression, seen with the 0.8 μg pBJ5/HNF-1α cotransfection, was 290% relative to the constitutive promoter (41.4 ± 3.9-fold over the basal cotransfection). This also represents a 6.9-fold increase over the pGL3/−133 control. With HNF-1β cotransfections, no significant increase over the pGL3/−133 control construct is seen with any quantity of cotransfection. These data also indicate the lack of effect of the pBJ5 vector that was common to the HNF-1α and HNF-1β cotransfections. No additive effect of HNF-1β was seen when both HNF-1α and HNF-1β (0.2 μg of each vector) were cotransfected with the −133 construct (Fig. 6).

DISCUSSION

Analysis of an HNF-1 site common to both mouse and human guanylin promoters demonstrates the importance of this consensus sequence in the transcription of the guanylin gene. Nuclear proteins from several intestinal cell lines bind the HNF-1 consensus sequence located in the immediate 5′ flanking region of the mouse guanylin gene. By demonstrating that mutagenesis of this consensus sequence abolishes reporter gene expression, we show that the HNF-1 binding site is required for reporter gene expression in Caco-2 cells, which have near maximal expression with the minimal promoter, and that the HNF-1 binding site is also necessary in HT-29 cell lines, which require an additional length of promoter for reporter gene expression. Furthermore, we have identified HNF-1α as a potent activator of transcriptional activation through cotransfection experiments. In addition to augmenting reporter gene expression in guanylin-expressing intestinal cell lines, HNF-1α enables acquisition of reporter gene expression in 293 kidney cells that express neither endogenous guanylin nor HNF-1α mRNA.

Expression of guanylin mRNA in intestinal cell lines.

We have identified several intestinal cell lines that may be useful for the study of transcriptional control of guanylin expression. Caco-2 cells are a human colon carcinoma cell line that spontaneously differentiates after achieving confluence into cells that closely resemble small intestinal enterocytes (37). HT-29 cells are a family of cell lines originally derived from a human colon adenocarcinoma. These cells differentiate into multiple epithelial cell types (2, 11, 16), and a number of subclones have been characterized. HT-29-18-C1 and HT-29-18-N2 cell types are committed to develop into either an enterocyte-like or mucin-secreting phenotype, respectively (16, 36). The enterocyte-like cell line is phenotypically distinguishable from Caco-2 cells. Unlike Caco-2 cells (6), this cell line does not express GCC (data not shown). More recently, Caco-2 cells have been shown to be competent for transcriptional activity of the GCC promoter (24). Previous studies, from our laboratory and others, have demonstrated that GCC mRNA or GCC protein (ligand binding) is expressed in villous epithelial cells of the small intestine (7) and in some species in crypt epithelial cells as well (1, 18, 45). GCC mRNA is also robustly expressed throughout the superficial and deep epithelial cells of the colonic glands (22). Thus, as is seen in the Caco-2 cell line, it is likely that GCC and guanylin mRNA are expressed in vivo in the same cell types, consistent with a paracrine or an autocrine function for guanylin.

Role of HNF-1 in transcriptional control of the guanylin gene.

A notable difference between the transcriptional activation of guanylin in the Caco-2 and the HT-29 cell lines is the ability of the short −133 bp promoter to be fully active in the Caco-2 cell line but not in the HT-29 cell lines. This promoter fragment contains a consensus binding site for HNF-1. In the human guanylin promoter, a minimal promoter (−157 to −5) construct containing the HNF-1 site also demonstrated near maximal reporter gene expression in T84 intestinal cells (34). Although the HNF-1 site alone is not sufficient to direct transcriptional initiation in HT-29 cells as indicated by the inability of the short promoter to be transcriptionally active, the alteration of the HNF-1 site in the transcriptionally active −300 construct abolishes reporter gene expression. Therefore, the HNF-1 site appears to be required for guanylin expression in HT-29 cell lines as well as Caco-2 cells.

In vivo, HNF-1α and HNF-1β have a wide tissue distribution including the liver (where guanylin is not expressed), intestine, and kidney (30); therefore these proteins alone cannot be sufficient to direct intestinal-specific expression of the guanylin gene. However, in vitro guanylin mRNA is expressed only in the presence of HNF-1α. Both HNF-1α and HNF-1β mRNA are expressed in epithelial cells along the entire length of the mouse small intestine and colon (43). This is consistent with a role for HNF-1 in expression of guanylin mRNA. HNF-1α mRNA is expressed in nearly equal levels in all cells along the crypt-villus axes (31); in contrast, guanylin expression is not uniform (15). In the human proximal intestine, expression is restricted to the crypts; in the ileum, there is expression in both the crypt and villus cells. Therefore, HNF-1α mRNA expression alone does not determine the pattern of guanylin expression in the intestine. However, because translational and posttranslational regulation of expression of HNF-1α and HNF-1β have been shown to be important (4, 29), the distribution of HNF-1 protein may be much different than the pattern of mRNA expression.

Role of HNF-1 in transcriptional activity of hepatic and nonhepatic genes.

HNF-1 was first found in the liver as a transcription factor binding to the β-fibrinogen promoter and since that time has been shown to regulate many genes, including albumin and α-1-antitrypsin (8, 28). Based on the patterns of expression of genes regulated by HNF-1, several authors initially concluded that the expression of HNF-1 in tissues other than the liver functioned in the regulation of genes that are expressed in both the liver and in other tissues (28). There is new evidence to dispute this conclusion, as several nonhepatic genes, including sucrase-isomaltase and villin, have been shown to be regulated by HNF-1 (40, 48). Guanylin is another example of a nonhepatic gene that is highly expressed in the intestine and is regulated by HNF-1.

Our results of guanylin gene transcriptional activity requiring an intact HNF-1 binding site parallel the observations of several authors studying other HNF-1-regulated genes. With sucrase-isomaltase, which is expressed in a strict tissue-, position-, and cell-dependent pattern (48), both HNF-1α and HNF-1β were capable of binding two regulatory elements, SIF2 and SIF3; however, HNF-1α had a potent effect on transcriptional activation, whereas HNF-1β had a minimal effect. Our results also are consistent with other studies of nonintestinal genes that have shown that HNF-1α is a potent inducer of gene transcription, whereas HNF-1β is a weak inducer (39) of transcription or has no effect (29). The observation that HNF-1 participates in the regulation of guanylin transcriptional activation broadens the range of regulatory functions for HNF-1. Furthermore, unlike the sucrase-isomaltase promoter, the HNF-1 site in the guanylin promoter acts as a transcription binding sequence in the absence of a Cdx homeodomain binding site (44). This suggests an alternate mechanism for the HNF-1 sequence to regulate intestinal gene expression.

HNF-1 in vivo.

An HNF-1 gene-targeted mouse has recently been described (38). Mice lacking HNF-1α develop failure to thrive, die around weaning, and have marked liver enlargement (38). The transcription rates of genes such as albumin and α-1-antitrypsin are reduced; the absence of any transcriptional activity for phenylalanine hydroxylase gives rise to a phenotype of phenylketonuria. No intestinal disorder has been identified macroscopically or histologically, but these authors have not specifically evaluated the effect of targeted disruption of HNF-1α on intestinal function or gene expression. Further investigation of these mice may allow characterization of the in vivo effects of HNF-1α on guanylin gene regulation.

In summary, the finding that HNF-1α regulates transcription of the guanylin gene expands the range of function for a transcription factor that was initially described as liver specific. Clearly, HNF-1α is important in the transcriptional activation of guanylin in vitro. In some cell lines (e.g., HT-29-18-C1 and HT-29-18-N2), the HNF-1 site is functionally active but not sufficient for guanylin gene expression. We have identified several cell lines (Caco-2, HT-29-18-C1, and HT-29-18-N2) that may be useful for understanding the interactions of HNF-1 with other modulators of guanylin gene regulation. Additional studies will be needed to better understand the molecular mechanisms responsible for the complex spatial pattern of guanylin gene expression along the cephalocaudal and crypt-villus intestinal axes.

Acknowledgments

This work was supported in part by National Institute of Diabetes and Digestive and Kidney Diseases Grants DK-47318 and DK-07727–02.

Footnotes

  • Address for reprint requests: M. B. Cohen, Div. of Pediatric Gastroenterology and Nutrition, Children’s Hospital Medical Center, 3333 Burnet Ave., Cincinnati, OH 45229.

REFERENCES

View Abstract