Hepatocyte nuclear factor-1α (HNF-1α) is a modified homeodomain-containing transcription factor that has been implicated in the regulation of intestinal genes. To define the importance and underlying mechanism of HNF-1α for the regulation of intestinal gene expression in vivo, we analyzed the expression of the intestinal differentiation markers and putative HNF-1α targets lactase-phlorizin hydrolase (LPH) and sucrase-isomaltase (SI) in hnf1α null mice. We found that in adult jejunum, LPH mRNA in hnf1α−/− mice was reduced 95% compared with wild-type controls (P < 0.01, n = 4), whereas SI mRNA was virtually identical to that in wild-type mice. Furthermore, SI mRNA abundance was unchanged in the absence of HNF-1α along the length of the adult mouse small intestine as well as in newborn jejunum. We found that HNF-1α occupies the promoters of both the LPH and SI genes in vivo. However, in contrast to liver and pancreas, where HNF-1α regulates target genes by recruitment of histone acetyl transferase activity to the promoter, the histone acetylation state of the LPH and SI promoters was not affected by the presence or absence of HNF-1α. Finally, we showed that a subset of hypothesized intestinal target genes is regulated by HNF-1α in vivo and that this regulation occurs in a defined tissue-specific and developmental context. These data indicate that HNF-1α is an activator of a subset of intestinal genes and induces these genes through an alternative mechanism in which it is dispensable for chromatin remodeling.
- liver fatty acid binding protein
- intestinal gene expression
- chromatin modification
hepatocyte nuclear factor-1α (HNF-1α), a modified homeodomain-containing transcription factor originally identified in liver by its ability to interact with regulatory elements in the albumin and β-fibrinogen genes (7, 8), has since been implicated in the regulation of gastrointestinal genes. HNF-1α is expressed in liver, pancreas, and kidney (2, 3) as well as in the epithelium of the stomach, small intestine, and colon (19). In the small intestine of adult rodents, HNF-1α is expressed evenly from duodenum to ileum in the proliferative crypt compartment as well as in differentiated cells on villi (4, 39, 49). It is expressed in the small intestine throughout development (4, 39). HNF-1α binds in vitro to specific regulatory regions in multiple genes expressed in the small intestinal epithelium (4, 6, 9, 11, 12, 15, 18, 20, 23, 27, 37, 38, 43, 52, 54), and has been shown to activate the promoters of most of these genes in overexpression experiments in cell culture transient cotransfection assays (4, 6, 9, 14, 15, 18, 37, 52, 54). In germline hnf1α null mice, the mRNAs for CFTR, Claudin-2, and calbindin D9k were modestly reduced (∼50%) (27, 37, 52), whereas that of the apical sodium-dependent bile acid transporter (ASBT) was silenced (40). In transgenic mice expressing reporters under the control of the sucrase-isomaltase (SI) or liver fatty acid binding protein (fabpl) promoters, mutations in well-characterized HNF-1 binding sites resulted in significant reductions in transgene expression (6, 10). HNF-1α may interact with other intestinal transcription factors, including Hox C11 (24), Cdx-2 (4, 18, 25), and GATA factors (4, 10, 18, 49, 50). Taken together, these data suggest that HNF-1α controls the expression of multiple intestinal genes through direct binding to the promoter and likely regulates these genes through cooperation with other transcription factors.
Lactase-phlorizin hydrolase (LPH) and SI are absorptive enterocyte-specific, microvillus membrane disaccharidases necessary for the terminal digestion of certain carbohydrates in mammals: LPH hydrolyzes the β-linkage in lactose, whereas SI hydrolyses the α-linkage in α-dissacharides (13). The genes encoding these disaccharidases are widely used as models to study intestine-specific gene expression and intestinal differentiation (16, 17, 35, 36, 44). In adult mammals, LPH and SI expression is confined to absorptive enterocytes on villi and is highest in the jejunum and proximal ileum, although SI is also expressed in the distal ileum (16, 36). LPH is highly expressed just before birth through suckling and declines at weaning, whereas SI is low before weaning and increases during weaning. Although this pattern coincides with a change from a milk-based diet to a diet of solid foods containing α-disaccharides, it is well known that LPH and SI are not regulated by their substrates (13). LPH and SI have well-characterized HNF-1 binding sites in their 5′-flanking regions, called cis-element-2c (CE-2c) in LPH (47) (−73 to −61 in the mouse LPH promoter) and SI footprint 3 (SIF3) in SI (55) (−174 to −155 bp in the mouse SI promoter). These sites mediate HNF-1α activation in transient cotransfection assays in cell culture models (4, 18), and SIF3 is necessary in vivo for the expression of a transgene under the control of the SI promoter (6). However, the association of HNF-1α with CE-2c or SIF3 of the endogenous gene in vivo has not yet been reported. We have shown that HNF-1α interacts with GATA factors, namely GATA-4, to cooperatively activate the LPH and SI promoters (4, 18, 49, 50) through an evolutionarily conserved mechanism that requires the DNA binding and activation domains of HNF-1α and the HNF-1 binding sites on target promoters (49, 50). Together, these data support a role for HNF-1α as a transcriptional activator of LPH and SI gene expression and validate the use of LPH and SI as models for HNF-1α regulation in the mammalian small intestine.
Targeted disruption of hnf1α has been reported by two laboratories from which the importance and underlying mechanism of HNF-1α regulation is beginning to be elucidated (21, 31). Hnf1α null mice display delayed growth, liver dysfunction, diabetes, and sterility, but there is no structural defect in the intestine (40). In these mice, hepatic expression of phenylalanine hydroxylase (pah) and fabpl was markedly reduced (1, 32), whereas the expression of pah in the pancreas was unaffected (29). In contrast, hepatic expression of the glucose transporter 2 (glut2) and L-type pyruvate kinase (L-PK) genes was expressed normally in hnf1α null mice but markedly attenuated in pancreas. These findings reveal a differential regulation of target genes in liver and pancreas by HNF-1α (29). Although HNF-1α was associated with these promoters in all tissues where these genes are expressed, HNF-1α was indispensable for transcriptional activation only in cellular- and promoter-specific contexts in which it was required to recruit histone acetyl transferase (HAT) activity (29). Whether HNF-1α is required for the recruitment of HAT activity in the activation of intestinal genes is unknown.
In the present study, we hypothesized that HNF-1α is required for the expression of the LPH and SI genes as well as other putative HNF-1α targets in vivo and activates these genes by mediating local histone acetylation at the promoters. In our experimental model, we used mice that do not express hnf1α rather than by relying on cell culture models dependent on overexpression of transcription factors and extrachromosomal promoter-reporter plasmids. We found that HNF-1α is essential for the expression of the LPH gene but is not necessary for that of SI. We further showed that HNF-1α is associated with the binding sites on the LPH and SI promoters in vivo, but this interaction does not necessarily correlate with transcriptional activation, as has been shown for genes expressed in liver and pancreas (29). Finally, we demonstrated that the histone acetylation state of the LPH and SI promoters is not affected by the presence or absence of HNF-1α, suggesting that, in contrast to liver and pancreatic genes (29), HNF-1α is not required for histone acetylation and subsequent activation of specific intestinal targets. These data are consistent with the hypothesis that HNF-1α regulates a subset of intestinal genes by a mechanism in which it is dispensable for chromatin modification.
MATERIALS AND METHODS
Mice segregating a null hnf1α allele on a C57BL/6J background were generated by deletion of the first exon using Cre-LoxP technology (21). Mice were housed under standard conditions in the Animal Research at Children's Hospital facility and provided food and water ad libitum. To identify hnf1α wild-type and null alleles, DNA was obtained from tail snips, and a 3-primer PCR strategy was employed (T. Akiyama, unpublished data). Absence of HNF-1α expression in the intestine of hnf1α−/− mice was confirmed by Western blot analysis and immunohistochemistry as described (49). To obtain tissue for study, mice were anesthetized and tissue was extracted through a midline incision. All adult study animals were 8–16 wk of age, and all tissue was collected between 1300 and 1600 to avoid any fluctuations in gene expression due to circadian cycles (34). Approval was obtained from the Institutional Animal Care and Use Committee for all experiments involving mice.
RNA was isolated from 30–50 mg of mouse small intestine, colon, and liver using the RNeasy kit (Qiagen, Valencia, CA). To ensure that all traces of DNA were removed, RNA samples were treated with DNase (DNA-free, Ambion, Austin, TX) for 1 h at 37°C following the manufacturer's instructions. RNA samples were quantified by optical density at A260 nm and checked for absence of degradation on an agarose gel.
RNase protection assays.
To determine the effect of the absence of hnf1α on LPH and SI mRNA abundance, RNase protection assays were conducted as described (16, 17). To construct a plasmid template for the synthesis of an antisense mouse LPH RNA probe, mouse LPH cDNA sequence (+285 to +535bp) was amplified and subcloned into pBluescript II KS(+). The plasmid was linearized with XbaI and transcribed using SP6 RNA polymerase. The template for a mouse SI probe was kindly provided by Dr. P. Traber (University of Pennsylvania) (22). A mouse β-actin probe (17) was used as a control for tissue RNA. Gel-isolated 32P-labeled probes were hybridized to RNA at 68°C in 50% formamide overnight, digested with RNase A and T1, and the protected fragments were separated on 6% denaturing polyacrylamide gels and revealed by autoradiography.
Real-time quantitative RT-PCR.
To quantify LPH and SI mRNA levels in wild-type and hnf1α null mice, real-time quantitative RT-PCR (qRT-PCR) was performed using an iCycler and iQ SYBR Green Supermix (Bio-Rad). Specific primer pairs (Fig. S1; supplemental data for this article may be found at ajpgi.physiology.org/cgi/content/full/00359.2005/DC1) were designed using Beacon Designer software (Biosoft International), which enables selection of primers that are highly specific for intended targets and do not form primer-dimers. A temperature gradient and melt curve were obtained for all primer pairs to define optimal cycling conditions and to confirm the absence of primer-dimer formation, respectively. All reactions were conducted in triplicate using 10 pmol of both forward and reverse primers at an annealing temperature of 59°C and an extension temperature of 72°C. To calculate the efficiency of the PCR reaction, a standard curve was produced using 10-fold dilutions of calibrator RNA. Calibrator RNA was a pool of intestinal RNA obtained from wild-type adult mouse jejunum. RNA abundance was corrected for gapdh and expressed relative to the calibrator using the Pfaffl method (30), which takes into account the efficiency of amplication of each primer pair. No-RT controls were used for all samples to confirm the absence of DNA contamination. Significant differences in mRNA abundance between hnf1α wild-type and null mice were determined by Student's t-test.
To define the effects on other transcriptional regulators and potential intestinal HNF-1α targets, semiquantitative RT-PCR was performed on total RNA from jejunum and duodenum of wild-type and hnf1α null mice. Primers as shown in Fig. S1 were designed with Beacon Designer software as described in Real-time quantitative RT-PCR. Optimal temperature and cycling conditions in the linear range were determined for each primer set with wild-type cDNA using an iCycler (Bio-Rad). cDNA was prepared with RNA (0.8 μg) from two wild-type and two hnf1α null mice using the iScript cDNA synthesis kit (Biorad) following the manufacturer's instructions. For each reaction, 2 μl cDNA were used in a 25-μl PCR reaction. The PCR products were separated on a 2% agarose gel.
To define protein-DNA interactions, EMSAs were carried out as previously described (18) using probes and/or competitors shown in Fig. S1 and nuclear extracts from isolated epithelial cells (49). For supershift analyses, antibodies (0.1 μg/μl) were preincubated with the nuclear extracts for 20 min before the addition of the probe. The antibodies used in EMSAs were purchased from Santa Cruz Biotechnology (Santa Cruz, CA) as the concentrated gel shift stock (HNF-1α, sc-6547X; HNF-1β, sc-7411X). Both antibodies were verified by the specific supershift complex produced in control experiments using in vitro transcribed and translated proteins. All experiments were conducted on at least three different animals.
Chromatin immunoprecipitations assays.
To analyze the association of HNF-1α and the acetylated forms of histone 3 and histone 4 with the mouse LPH and SI promoters, chromatin immunoprecipitations (ChIP) assays were performed by a procedure adapted from protocols used for mammalian cells (29, 42). The small intestine of anesthetized mice was removed and placed on a glass plate on a bed of ice, rinsed with ice cold 1× PBS, and 5 cm of midjejunum was isolated and opened longitudinally. Epithelial cells were collected by scraping the mucosa with glass microscope slides, washed in 1× PBS, and cross-linked for 10 min at room temperature in a final concentration of 1% formaldehyde (Sigma) in 1× PBS. Fixation was terminated by replacing the formaldehyde solution with 0.125 M glycine in 1× PBS. Cells were collected and washed with ice-cold 1× PBS and allowed to swell on ice for 10 min in lysis buffer [25 mM HEPES, pH 7.8, 1.5 mM MgCl2, 10 mM KCl, 0.1% Nonidet P-40 (NP-40), 1 mM DTT, and protease inhibitor cocktail (Sigma)]. After homogenization with a Dounce homogenizer (10 strokes), the nuclei were collected by centrifugation, resuspended in sonication buffer (50 mM HEPES, pH 7.8, 140 mM NaCl, 1 mM EDTA, 1% Triton X-100, 0.1% sodium deoxycholate, 0.1% SDS, and protease inhibitors) and sonicated on ice (10 times 20-s pulses, Ultrasonics, model W-220F ) until the chromosomal DNA ranged from 0.1 to 1.0 kb in length. Cellular debris was removed by centrifugation at 14,000 rpm, and the supernatant containing the fragmented chromatin was precleared with protein G-Sepharose (Active Motif). Immunoprecipitation was conducted on the precleared chromatin (20 μg) using 7.5 μl anti-HNF-1α antibody (sc-6547, Santa Cruz) or 4 μl anti-acetylated histone H3 or anti-acetylated histone H4 antibodies (06–599, 06–866, Upstate). The immune complexes were purified by adsorption to protein G-Sepharose beads, which were washed once with sonication buffer, once with sonication buffer containing 500 mM NaCl, once with 20 mM Tris, pH 7.8, 1 mM EDTA, 250 mM LiCl, 0.5% NP-40, 0.5% Na-deoxycholate, and twice with Tris-EDTA (TE) buffer. SDS was omitted in the wash buffers for the HNF-1α immunoprecipitations. The beads were incubated in elution buffer (50 mM Tris, pH 8.0, 1 mM EDTA, and 1% SDS) at 65°C for 15 min, then elution buffer containing 200 mM NaCl at 65°C for 4 h to reverse the formaldehyde cross-links. Samples were then treated with 1 μg proteinase K (25 mg/ml) for 1 h at 45°C followed by purification with PCR purification columns (Qiagen). PCR reactions were performed in a 25-μl reaction volume containing 2 μl immunoprecipitate or input DNA (fragmented, reverse cross-linked chromatin) and primer mixtures designed to amplify segments containing the transcription initiation site and mCE-2c for LPH or mSIF3 for SI (see Fig. S1). Amplification of LPH or SI coding regions was used as negative controls.
LPH mRNA is reduced in the jejunum of hnf1α null mice.
On the basis of in vitro, cell culture, and transgenic data, HNF-1α is hypothesized to be an activator of the LPH and SI genes in vivo. To test this hypothesis, the abundance of LPH and SI mRNAs in the adult jejunum of hnf1α null mice was compared with that in heterozygous or wild-type littermates. As shown in Fig. 1A, LPH mRNA was greatly reduced in an hnf1α−/− mouse compared with an hnf1α−/+ littermate using RNase protection assays, whereas SI mRNA was not affected by the absence of HNF-1α. To quantitatively determine the magnitude of mRNA reduction in the absence of HNF-1α, real-time qRT-PCR was conducted on RNA isolated from adult midjejunum using optimized primers for both LPH and SI mRNAs. LPH mRNA was reduced 95% in hnf1α null mice compared with wild-type controls (P < 0.01, n = 4), whereas SI mRNA in hnf1α null mice was similar to that in wild-type mice (Fig. 1B). In all analyses, LPH and SI expression was similar between wild-type and hnf1α+/− mice.
HNF-1α intestinal target genes are reduced throughout the length of the small intestine.
To define the importance of HNF-1α for LPH and SI gene expression throughout the length of the intestine, both of which display regulated proximal-to-distal patterns (16, 36), LPH and SI mRNA abundance was determined throughout the intestine of wild-type and hnf1α null mice using RNase protection assays. As shown in Fig. 2, LPH mRNA was greatly reduced throughout the length of the mouse small intestine of hnf1α−/− mice compared with hnf1α+/+ controls, whereas SI mRNA of hnf1α null mice was similar to that of wild-type mice in all segments, with the possible exception of the proximal intestine, where SI mRNA was higher in the hnf1α null mice. This pattern was confirmed by real-time qRT-PCR on postweaning mice (n = 4 in each group), which also revealed a greatly reduced LPH mRNA abundance and a variable increase in SI mRNA levels throughout the intestine of hnf1α null mice (not shown).
Absence of hnf1α does not affect the expression of other intestinal transcription factors.
To ensure specificity of the hnf1α null effect on its target genes, the expression of other intestinal transcription factors implicated in the regulation of LPH or SI was determined in wild-type and hnf1α−/− mice (Fig. 3A). As expected, hnf1α mRNA was expressed in the adult jejunum of wild-type mice but was absent in that of the hnf1α null mice (Fig. 3A). Absence of HNF-1α was also confirmed by Western analysis and immunohistochemistry (Fig. 3, B and C, respectively), as previously documented in the liver of these mice (1). None of the other intestinal transcription factors expressed in jejunum, including hnf1β, gata4, gata5, gata6, cdx1, and cdx2, was affected by the absence of hnf1α. However, duodenal pdx1, a putative target of HNF-1α (11) and repressor of LPH gene expression (53), was reduced in the hnf1α null mice. Because duodenal expression of LPH is attenuated rather than increased in the hnf1α null mice, it is unlikely that decreased pdx1 affects duodenal LPH expression in this model. Noteworthy, is the lack of a compensatory response in hnf1β gene expression as shown in the liver of hnf1α null mice (41). Together, these data demonstrate that the observed effects on the expression of LPH are unlikely due to secondary effects by these transcriptional regulators. However, these findings do not rule out the possibility that as yet unidentified transcriptional regulators of LPH are affected by the absence of HNF-1α.
HNF-1α binds the mouse LPH and SI promoters both in vitro and in vivo.
Although previous reports using EMSAs demonstrated that HNF-1α interacts with specific sites on the LPH and SI promoters in vitro (4, 18), these studies have generally been conducted using rat, pig, or human probes and in vitro transcribed and translated proteins or proteins derived from cells grown in culture. Thus, to define the association of HNF-1α with the previously identified HNF-1 binding sites on the mouse LPH and SI promoters under stoichiometric conditions similar to those present in the intestinal epithelium, EMSAs were conducted using mouse probes (mCE-2c and mSIF3) and nuclear extracts obtained from isolated intestinal epithelial cells from adult mouse jejunum (Fig. 4A). For both the LPH and SI probes, a specific protein-DNA complex was formed, as indicated by specific competition, that completely supershifted when an HNF-1α antibody was added; no detectable supershift complex was formed when an HNF-1β antibody was added. These data indicate that both promoters are capable of binding HNF-1α at physiological levels in vitro and that HNF-1α is the predominant protein in nuclear extracts from intestinal epithelium that binds to these promoters.
To test the hypothesis that HNF-1α is associated with the LPH and SI promoters in vivo, ChIP assays were conducted on chromatin isolated from the epithelium of adult mouse jejunum using an HNF-1α antibody. A critical step for valid ChIP assays on chromatin isolated from animal tissue is to establish consistent conditions for chromatin fragmentation before immunoprecipitation. Optimal results were achieved when chromatin fragments ranged from 0.1 to 1.0 kb (Fig. 4B). In formaldehyde cross-linked chromatin from hnf1α+/+ mice, both of the LPH and SI promoters were immunoprecipitated with the HNF-1α antibody (Fig. 4C). These data, which were replicated on three different animals, demonstrate that HNF-1α is associated with the promoters of both the LPH and SI genes in vivo.
The histone acetylation state at the LPH and SI promoters is not regulated by HNF-1α.
In liver and pancreas, HNF-1α has been shown to be indispensable for the recruitment of HAT activity to the promoters of target genes in these tissues. To test the hypothesis that HNF-1α is also necessary for histone acetylation and subsequent activation of intestinal target genes, ChIP assays were conducted on the LPH and SI promoters using antibodies specific for acetylated tails of histone 3 and histone 4 (Fig. 5). In wild-type mice, the histones at both the LPH and SI promoters were highly acetylated, which is consistent with an activated or poised state. However, in the absence of HNF-1α, histone 3 and histone 4 remained highly acetylated at both promoters. These data were confirmed in two other pairs of wild-type and hnf1α null mice. Although these data are consistent with the hypothesis that HNF-1α is dispensable for the acetylation of histones at the LPH promoter, they do not rule out the possibility that subtle changes in histone acetylation occur that are below the limits of detection of this assay or that redundant or compensatory mechanisms take over in the absence of HNF-1α.
HNF-1α regulates a subset of intestinal genes in vivo.
The differential effect on LPH and SI gene expression in the hnf1α null mice has led us to investigate other genes expressed in the intestine, including those hypothesized to be regulated by HNF-1α in cell cultures studies and/or those expressed in the intestine but shown to be regulated by HNF-1α in nonintestinal tissues in vivo (Fig. 6). In addition to LPH and SI, these include fabpl (1, 9), α1-antitrypsin (15), aldolase B (48), guanylin (14), sodium-glucose cotransporter 1 (SGLT1) (23), hnf4α (12), L-PK (41), and neuroD1 (41). Surprisingly, of eight putative HNF-1α target genes indicated by cell culture assays, only LPH, fabpl, α1-antitrypsin, and guanylin mRNAs were reduced (Fig. 6) in the absence of HNF-1α in adult mouse jejunum; expression of the other putative target genes was indistinguishable from that in wild-type mice. Of the five target genes whose expression is reduced in the absence of HNF-1α in adult pancreas or liver, the mRNAs of two (fabpl and α1-antitrypsin) were also reduced in the intestine, revealing a tissue specificity associated with regulation by HNF-1α. Noteworthy, SGLT1 expression was modestly increased in adult jejunum of hnf1α null mice, which we believe is a response to the diabetic phenotype (see discussion).
Because many of the putative HNF-1α targets are developmentally regulated, the importance of HNF-1α in newborn mice was also determined (Fig. 6). LPH, fabpl, α1-antitrypsin, and guanylin were all reduced in the absence of HNF-1α in both the newborn and adult small intestine. However, although α1-antitrypsin in newborn hnf1α−/− mice was reduced similarly to that in adults, LPH and fabpl were reduced modestly in newborn hnf1α−/− mice but almost completely in adult hnf1α−/− mice. Guanylin, on the other hand, was greatly reduced in newborn hnf1α−/− mice but only modestly in adults. These data suggest that HNF-1α is required for the expression of a particular subset of intestinal target genes, but in a specific developmental context.
Establishment and maintenance of intestinal function requires a complex interplay of multiple transcription factors that together coordinate intestinal gene expression. HNF-1α is a transcriptional regulator that has been implicated as an activator of intestinal genes but whose function in the intestine in vivo is only beginning to be elucidated (27, 37, 40, 52). In the present study, we defined the importance of HNF-1α for the expression of intestinal genes using mice that do not express hnf1α rather than by relying on in vitro and cell culture models. We found that LPH gene expression was markedly reduced as anticipated, whereas SI gene expression was surprisingly unaffected by the absence of HNF-1α. Despite the differential requirement of HNF-1α for LPH and SI gene expression, both the LPH and SI promoters bind HNF-1α in the in vivo context of enterocyte chromatin. Analysis of the histone acetylation state at the LPH and SI promoters revealed that, in contrast to HNF-1α target genes in liver and pancreas where HNF-1α is required for recruitment of HAT activity and subsequent transcriptional activation, histone acetylation was not affected by the presence or absence of HNF-1α. Finally, we show that a subset of hypothesized intestinal target genes is regulated by HNF-1α, and this regulation occurs in a defined tissue-specific and developmental context.
Due to compelling in vitro and cell culture data demonstrating that HNF-1α binds and activates the LPH and SI promoters, it was hypothesized that HNF-1α regulates LPH and SI gene expression in vivo. Whereas a significantly reduced expression of LPH in the absence of HNF-1α was anticipated, it was unexpected that SI mRNA was unaffected by the absence of HNF-1α in the small intestines of either adult or newborn hnf1α null mice. Analyses of other genes previously shown to be targets of HNF-1α using in vitro and cell culture assays revealed that some are in vivo targets of HNF-1α, including fabpl, α1-antitrypsin, and guanylin, whereas others are not regulated by HNF-1α in vivo, including SGLT1, aldolase B, hnf4α, and L-PK. Thus these data indicate that cell culture studies, although valuable to define putative importance and underlying mechanisms of transcriptional activation, should be interpreted in the context of the defined model and may not always predict in vivo function.
In transgenic mice expressing a reporter under the control of the SI or fabpl promoters, mutations in well-characterized HNF-1 binding sites reveal a strong reduction in promoter activation compared with mice carrying a wild-type transgene (6, 10), demonstrating that these sites are necessary for promoter activation in vivo. Because HNF-1α was the predominant protein from intestinal nuclear extracts that binds to these sites (Fig. 4) (4), it was hypothesized that interaction of HNF-1α with these sites is required for SI and fabpl gene expression in vivo (4, 10). Our data showing that fabpl is a transcriptional target of HNF-1α in vivo (Fig. 6) supports this hypothesis for fabpl gene expression, but the lack of a requirement of HNF-1α for SI gene expression (Figs. 1 and 2) is inconsistent with this hypothesis. It is possible that the SI promoter used in these transgenes does not contain all of the HNF-1 sites necessary for activation. It is also possible that the mutations in SIF3 interrupt the binding of another critical activator, such as HNF-1β. Although the ability of HNF-1β to activate SI in cell culture models is minimal (6, 18), it binds to SIF3 in vitro (6) and is thus a candidate activator of SI gene expression in vivo. It is also possible that an as yet unknown protein binds to SIF3. Precedence for alternative binding to well-characterized cis-acting elements is supported by similar data on the SI promoter where CDP-Cux, a transcriptional repressor of SI in vivo (5), binds to sequence that overlaps a conserved GATA binding site in the 5′-flanking region. In conclusion, SI is not a transcriptional target of HNF-1α, but its well-characterized HNF-1 binding site, SIF3, may be required to mediate SI activation in vivo.
It can be argued that the effects we observed in the intestine of the hnf1α null mice are secondary to the diabetic phenotype. It has been well established in numerous studies in diabetic humans and rodent models that enzymes and transporters involved in intestinal glucose metabolism are increased (26, 28, 45, 56), which probably is a secondary response to hyperphagia, hypoinsulinemia, or sensing of glucosuria. Consistent with these studies, we detected a modest increase in the mRNA abundance of SI (Fig. 3 and data not shown) and SGLT1 (Fig. 6) of adult hnf1α null mice. However, LPH mRNA levels were significantly reduced, and we therefore conclude that the reduced expression of LPH in adult hnf1α null mice is a direct effect of the lack of HNF-1α, rather than secondary to diabetes.
In ChIP assays, we found in adult wild-type mice that HNF-1α was associated with both the LPH and SI promoters in vivo, yet only LPH is regulated by HNF-1α. This discordant relationship between in vivo HNF-1α binding to target promoters and HNF-1α gene activation was also found for genes expressed in liver and pancreas. For example, HNF-1α is associated with the promoters of pah, glut2, and L-PK in liver and pancreas where these genes are all expressed. However, HNF-1α is only required for pah expression in liver and glut2 and L-PK expression in pancreas. These and our data therefore reveal that association of HNF-1α with a putative target promoter does not necessarily predict regulation by HNF-1α.
Our data reveal that in the absence of HNF-1α, the LPH promoter remains hyperacetylated (Fig. 5), and thus the 95% reduction in LPH gene expression in adult intestine is not a result of local hypoacetylation of the promoter. This finding contrasts with that in liver and pancreas, where HNF-1α dependency was always associated with histone hyperacetylation of the promoters. For example, in the presence of HNF-1α, such as in wild-type mice, the pah gene was expressed in liver where the histones at the promoter were highly acetylated, and the DNA in this region was hypomethylated indicating an active or poised state (29). However, in hnf1α null mice, pah gene expression was silenced, and the histones and DNA at the promoter were hypoacetylated and hypermethylated, respectively (29, 32). The glut2 and L-PK genes are endogenously expressed in multiple tissues, but in the absence of HNF-1α, glut2 and L-PK gene expression is silenced only in the pancreas. The histones at the glut2 and L-PK promoters, which are normally hyperacetylated in pancreas and liver, were hypoacetylated only in the pancreas of hnf1α null mice (29). From these studies, a model was proposed in which HNF-1α promotes and maintains the formation of a transcriptionally competent state through local recruitment of HAT activity (51). Our data show that the acetylation state of the LPH promoter is not different in the epithelium of adult hnf1α null mice compared with controls (Fig. 6). Although it is possible that modest alterations in histone acetylation are below the detection limits of the assay or that HNF-1α plays a role in histone acetylation, but in its absence other factors compensate, our data for LPH are fundamentally consistent with the hypothesis that HNF-1α is dispensable for the recruitment of HAT activity. These data therefore indicate that other factors are responsible for chromatin modification of the LPH promoter.
Our data show that HNF-1α regulates target genes in defined tissue-specific and developmental contexts. For example, hnf4α, L-PK, and neuroD1 are all targets of HNF-1α in nonintestinal tissue in vivo but are not regulated by HNF-1α in the intestine. Furthermore, LPH, fabpl, and guanylin reveal a developmental-specific reduction in gene expression in the absence of HNF-1α in adult intestine. Differential tissue regulation by HNF-1α has been previously shown for target genes expressed both in liver and pancreas (29, 41). Together, these data argue that HNF-1α controls the tissue- and developmental-specific expression of target genes through a complex mechanism, likely involving other transcriptional regulators.
The data presented here and in the literature (33) support a role for HNF-1α as a regulator of terminal metabolic genes in different tissues. The intestinal genes shown to be regulated by HNF-1α in the present study have diverse functions, including terminal digestion of nutrients (LPH), intracellular lipid transport (FABP-L), inflammatory response (α1-antitrypsin), and fluid and electrolyte balance (guanylin). Thus disruption in intestinal HNF-1α regulatory pathways could alter important metabolic processes within the intestine, although this was not specifically studied in the present investigation. Specific mutations in HNF-1α in humans are associated with MODY3 (33), and some of these mutations result in a disrupted interaction with other transcription factors causing a decreased activation of intestinal target genes, such as fabpl (9). Thus it is possible that individuals with HNF-1α mutations that cause MODY3 could have metabolic consequences in the small intestine.
HNF-1α selectively regulates specific genes in diverse tissues and cell types by a mechanism that is dependent on information in the promoters of these target genes and the panel of transcriptional regulators present in the tissues or cell types that express these genes. Many intestinal gene promoters have conserved binding sites not only for HNF-1 but also for the Cdx and GATA families of transcription factors (46). We have previously shown that HNF-1α physically interacts with GATA factors, namely GATA-4, to cooperatively activate specific promoters, including that of LPH and fabpl (10, 18, 49, 50) through an evolutionarily conserved mechanism (49, 50). We have hypothesized that the overlapping expression and subsequent interaction of HNF-1α and GATA-4 in the intestinal epithelium are a means to achieve high levels of intestine-specific gene expression of specific targets (49, 50). HNF-1α expression is constant throughout the length of the adult mouse small intestine (49), whereas GATA-4 expression declines in distal intestine coincident with a distal decline in LPH gene expression. Thus we propose that although HNF-1α is necessary for LPH gene expression in the adult mouse small intestine, its specific pattern is defined by coregulation with GATA-4. Similar to LPH, the fabpl promoter contains conserved HNF-1 and GATA binding sites (10) and is regulated in vivo by both HNF-1α (Fig. 6) and GATA-4 (10), further supporting a mechanism of cooperativity between these two transcription factors. These studies indicate that HNF-1α is necessary for the expression of a subset of intestinal genes and likely interacts with other transcriptional regulators such as GATA-4 to mediate the specific pattern of target gene expression. The specific pattern of expression is therefore dependent on a dynamic relationship between the target gene promoter and the transcriptional regulators present within a specific spatial and developmental context.
This work was supported by National Institute of Diabetes and Digestive and Kidney Diseases Grants R01-DK-061382 (to S. D. Krasinski) and R37-DK-32658 (to R. J. Grand), the Harvard Digestive Disease Center (5P30-DK-34854), and by the Nutricia Research Foundation (to T. Bosse).
We thank Dr. S. K. Ray (New England Medical Center, Boston) for technical support with ChIP assays), C. E. Hogen Esch (Leiden University) for help with immunohistochemistry and Dr. P. G. Traber (University of Pennsylvania) for providing us with the mouse SI probe.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2006 the American Physiological Society