Restriction of apolipoprotein A-IV gene expression to the intestine villus depends on a hormone-responsive element and parallels differential expression of the hepatic nuclear factor 4alpha and gamma isoforms.

The apoA-I/C-III/A-IV gene cluster, like most intestine-specific genes, displays a specific pattern of expression along the intestinal cephalocaudal and crypt-to-villus axes. We have shown that this specific pattern of expression requires the distal apoA-IV promoter and the apoC-III enhancer. Using a new set of transgenic mice, we demonstrate here that the restriction of apoA-IV gene transcription to villus enterocytes requires a hormone-responsive element (HRE) located within the apoA-IV distal promoter. We showed, using nuclear extracts from villus or crypt epithelial cells, that this HRE bound the transcription factor hepatic nuclear factor 4 (HNF-4). We also found that the HNF-4gamma isoform was produced only in the villus, whereas the HNF-4alpha isoform was produced along the entire length of the crypt-to-villus axis. Our results demonstrate that the HRE of the distal apoA-IV promoter is responsible for the restriction of gene expression to villus epithelial cells and that this HRE binds HNF-4 isoforms. The in vivo observation of parallel gradients for apoA-IV and HNF-4gamma gene expression raises questions concerning whether this transcription factor plays a specific role in the control of enterocyte differentiation.

Maintenance of a functional intestinal epithelium is a dynamic process that requires the constant, rapid renewal of cells from the stem cells located in the crypts (1). Nutrients are absorbed by enterocytes, the most abundant intestinal epithelial cells, which differentiate and express a variety of specific genes as they exit the crypt and migrate toward the tip of the villus. The regulatory sequences responsible for the spatial pattern of gene expression in the intestine have been studied in vivo with transgenic mice expressing a human or a reporter gene (2)(3)(4)(5)(6)(7)(8)(9). Several transcription factors are thought to be involved in the regulation of gut-specific gene expression, based on their pattern of expression. However, the transcriptional mechanisms involved are still poorly understood. In vitro studies with the enterocyte-like cell line Caco-2 indicated that the transcription factors hepatic nuclear factor 1 (HNF-1), 1 HNF-4, Cdx-2, and GATA-4, -5, and -6 may be involved in the expression of several intestine-specific genes, including sucrase isomaltase (10,11), calbindin-D9K (12,13), intestinal fatty acidbinding protein (14), lactase-phlorizin hydrolase (15)(16)(17), and apolipoprotein B (apoB) (18).
The apolipoprotein (apo) A-I/C-III/A-IV gene cluster is a useful model for studies of the molecular mechanisms controlling the specific pattern of gene transcription along the crypt-tovillus and cephalo-caudal axes of the intestine. The human apoA-IV gene is specifically expressed in villus enterocytes (19,20). Within the cluster, located on chromosome 11, the apoA-I and A-IV genes are transcribed in the same direction, whereas the apoC-III gene is transcribed in the opposite direction. Therefore, the apoC-III/A-IV intergenic region constitutes a common 6.6-kilobase pair 5Ј-flanking region for both genes (21). Previous studies have shown that the Ϫ780/Ϫ580 enhancer region of the apoC-III gene directs the intestinal expression of the three genes of the cluster (20,22,23). This apoC-III enhancer contains a hormone-response element (HRE). In vitro, both nuclear hormone receptors and orphan receptors are able to bind the HRE (24). Specific mutations in this HRE abolish intestinal expression of the apoA-I and apoC-III genes (25). However, although the apoC-III enhancer is necessary to direct the intestinal expression of the cluster, it is not sufficient to restrict gene expression to villus cells, as observed for the endogenous gene. This suggests that other regulatory sequences are involved in conferring the pattern of expression of apolipoprotein genes observed in the intestine in vivo. We identified the Ϫ700/Ϫ310 distal region of the apoA-IV promoter as such a regulatory region. This region, in combination with the proximal promoter and the Ϫ890/Ϫ500 apoC-III enhancer, is necessary and sufficient to restrict the expression of both the apoC-III and apoA-IV genes to villus enterocytes (20).
The aim of this study was to define the cis-regulatory sequence or sequences involved in this spatial pattern of intestine-specific expression. DNase I footprint analysis of the Ϫ700/Ϫ310 distal region of the apoA-IV promoter using mouse enterocyte nuclear factors led to the identification of a conserved HRE binding site. We demonstrated in vivo with a new set of transgenic mice that deletion of this HRE abolished the villus-specific gene expression. In parallel, we showed that this * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ Supported by a fellowship of the Fondation pour la Recherche Médicale.
§ To whom correspondence should be addressed. HRE bound the HNF-4␣ and -␥ isoforms, which display differential patterns of production along the crypt-to-villus axis.

EXPERIMENTAL PROCEDURES
Nuclear Extracts from Mouse Intestinal Epithelial Cells-The epithelium was isolated from villi as described by Perreault and Beaulieu (26). Briefly, the small intestine was everted, washed twice in phosphatebuffered saline, pH 7.2 (PBS), and cut into 5 mm fragments. Fragments were incubated overnight in 10 ml Matrisperse (BD PharMingen) at 4°C without shaking. The fragments were then shaken gently, and the isolated cells were filtered and washed twice with cold PBS. Crypt cells were isolated by shaking intestine fragments in a chelating buffer, as described by Flint et al. (27). Nuclear extracts from intestinal epithelial cells were then prepared as previously described (28).
Preparation of Whole-cell Extracts from COS-transfected Cells-COS-1 cells were transfected with vectors encoding HNF-4␣, ErbArelated protein 3/chicken ovalbumin upstream promoter transcription factor (COUP-TFI), and apoA-I regulatory protein-1/COUP-TFII by the calcium phosphate co-precipitation method. Whole-cell extracts were prepared as previously described for use in gel retardation assays (29).
DNase I Footprinting Assays-The DNA probes used in footprint analysis were obtained by PCR using the eCIII-AIV-CAT plasmid previously described (20) as a template and primer sets A and B. For each amplification by PCR, either the 5Ј or the 3Ј primer was labeled with T4 polynucleotide kinase and [␥-32 P]ATP. DNase I footprint analysis was carried out as previously described (30) with mouse enterocyte nuclear extract.
Electrophoretic Mobility Shift Assays (EMSA)-The human AIVE and mouse AIVE double-stranded oligonucleotides were used for EMSA as probes or competitors. The human AIVE and mouse AIVE correspond, respectively, to the human element E, between nucleotides Ϫ377 and Ϫ357 (5Ј-GGGAGATGTGGACTTTGCCCCCCATGAGCCC-3Ј) of the apoAIV promoter, and its homolog in mice, between nucleotides Ϫ146 and Ϫ114 (5Ј-GGGAGACTTGGACCTTGTTCTCTCAGACT-3Ј) (GenBank TM gi:192006). EMSAs were performed according to the protocol of Fried and Crothers (31), as described by Lacorte et al. (32), with 1 l of COS-1 whole-cell extract or 2 g of nuclear extract from villus or crypt epithelial cells. For competition assays, oligonucleotides were added in a 100-fold molar excess. Supershift experiments were performed with antibodies directed against HNF-4␣ (C-19X) and HNF-4␥ (C-18X), according to the manufacturer's instructions (Santa Cruz Biotechnology, Inc.).
Generation and Characterization of Transgenic Mice-The Ϫ700/ Ϫ410 apoA-IV promoter region was amplified by PCR using oligonucleotides spanning the Ϫ700 to Ϫ680 sequence with an XbaI restriction site and the Ϫ410 to Ϫ387 sequence with XbaI and BamHI restriction sites as sense and antisense primers, respectively. The resulting apoA-IV fragment was digested with XbaI, gel-purified, and inserted into the XbaI site of the C3-CAT plasmid (20) downstream from and in the opposite direction to the C3-CAT region, yielding the dA4⌬⌭-C3-CAT plasmid. The dA4⌬⌭-C3-CAT plasmid was sequenced to check that the cloning had been successful, and the transgene was excised from the plasmid by digestion with BamHI and gel-purified. The transgene (6 ng/l) was microinjected into fertilized eggs from C57BL/6J ϫ CBA/J females mated with males of the same strain according to standard procedures (33). Founder mice were identified by extracting DNA from the tails of 10 -15-day-old pups and analyzing it by PCR with oligonucleotide primers corresponding to sequences ϩ163 to ϩ189 (coding strand) and ϩ696 to ϩ670 (non coding strand) of the CAT gene. Founder mice were further analyzed by Southern-blotting, and copy number was estimated by densitometric scanning of autoradiographs as previously described (33). F 0 mice expressing the dA4⌬⌭-C3-CAT transgene were out-bred to generate lines of heterozygous mice.
CAT Assay-Individual tissues samples were homogenized and assayed for CAT activity as previously described (20). The soluble protein concentration was determined with the Bio-Rad protein assay. The proportion of chloramphenicol converted to acetylated forms was determined by scraping individual spots from the thin-layer chromatogram and counting them in a scintillation counter. CAT activity is expressed as % acetylation/min/mg of protein after subtracting the background for each tissue from control mice, which do not express the CAT gene In Situ Hybridization-Adult mice were killed by cervical disloca-tion. The entire small intestine was rapidly removed and divided into three parts corresponding to the proximal, middle, and distal regions of the small intestine. Samples were immediately embedded in tissuefreezing medium (Tissue-TEK) and frozen in isopentane cooled by liquid N 2 . In situ hybridization was performed with a modified version of the method of Braissant et al. (34). Tissue sections (12 m) were cut at Ϫ20°C and mounted on superfrost slides (Menzel-Glazer). Sections were fixed by incubation for 10 min in 4% paraformaldehyde in PBS at 4°C. They were then incubated twice for 15 min each in 0.1% active diethyl pyrocarbonate in PBS and equilibrated by incubation for 15 min in 5ϫ SSC (1ϫ SSC ϭ 0.15 M NaCl and 0.015 M sodium citrate, pH 7.0) supplemented with 0.1% active diethyl pyrocarbonate. Sections then incubated for 2 h at 58°C in 50% formamide, 5 ϫ SSC, 40 g/ml salmon sperm DNA and then hybridized by incubation for 40 h at 58°C in the same mixture to which 10 6 cpm of antisense or sense CAT riboprobe was added per slide. Sense and antisense CAT riboprobes were synthesized using T7 and T3 RNA polymerases, respectively, incorporating [␣-33 P]UTP. The template used to produce the two riboprobes was a linearized pSK-CAT plasmid containing a 265-bp fragment of the CAT gene (20). Slides were washed for 30 min in 2ϫ SSC at room temperature, 1 h in 2ϫ SSC at 65°C and twice, for 30 min each, in 0.1ϫ SSC at 65°C. Sections were then dehydrated in a graded series of ethanol solutions to achieve total desiccation and dipped in Hypercoat LM1 emulsion. Slides were stored in the dark at 4°C for 1-4 days and then developed. Images were recorded with a Zeiss Axiophot microscope. Laser Microdissection and mRNA Detection-Jejunal sections (7 m) were cut at Ϫ20°C and mounted on foil-covered microscope slides (Leica). Sections were rapidly stained with methyl green and dried at 37°C. Laser microdissection was performed, with an AS LMD Leica Laser microdissection system used to collect crypt and villus cells. Total RNA was extracted from crypt and villus samples with the Absolutely RNA nanoprep kit (Stratagene), and first-strand cDNA was generated with SensiScript reverse transcriptase (Qiagen) using random hexamer primers. The cDNAs were used for PCR amplification with CAT sense 5Ј-TATCCCAATGGCATCGTAAAGA-3Ј and antisense 5Ј-GGTATTCA-CTCCAGAGCGATG-3Ј primers. We amplified the 18 S RNA gene as a control for RNA extraction and reverse transcription with sense 5Ј-A-CGRACCAGAGCGAAAGCAT-3Ј and antisense 5Ј-GGACATCTAAGG-GCATCACAGAC-3Ј primers. PCR products were detected by Southern blot analysis using CAT or 18 S DNA as a probe. The 18 S DNA probe was used as an internal standard (data not shown).

Identification of a Conserved HRE in the Distal Region of the
ApoA-IV Promoter Region-We previously reported that the Ϫ700/Ϫ310 distal promoter region of the human apoA-IV gene was required to confer on a reporter gene the specific pattern of expression of the endogenous apoA-IV gene along the cephalocaudal and crypt-to-villus axes of the intestine (20). In vitro DNase I footprinting analysis of this distal region using nuclear extracts prepared from mouse intestinal epithelial cells identified four protected regions, designated E to G, at positions Ϫ377 to Ϫ356, Ϫ458 to Ϫ430, Ϫ585 to Ϫ523, and Ϫ640 to Ϫ610, respectively (Fig. 1, A-C). Footprints E and H were flanked by 4 and 2 DNase I hypersensitive sites, respectively. Another hypersensitive site was detected on the non-coding strand, within footprint G (Fig. 1B). Comparison of the sequences of the human and mouse distal apoA-IV promoter regions (nucleotide Ϫ700 to Ϫ310, GenBank TM accession num- bers gi:178756 and gi:192006), using the BLAST sequence alignment program, showed the two sequences to be 64% identical within footprint E (Fig. 1D), whereas no significant identity was found between the human and mouse sequences for the other footprints (data not shown). Footprint E contains a putative HRE, conserved in the human and mouse promoters, with a direct repeat 1 (DR1) (Fig. 1D). Electrophoretic mobility shift assays using extracts from COS-1 cells producing HNF-4␣, apoA-I regulatory protein-1, and ErbA-related protein 3 revealed that these factors bound to this motif in footprint E (Fig. 2). Thus, the motif identified in footprint E actually functions as an HRE.
Deletion of the Distal ApoA-IV HRE Abolishes the Cephalocaudal Gradient of Gene Expression-We investigated whether this distal HRE was involved in the specific pattern of expression of the apoA-IV gene by generating dA4⌬E-C3-CAT transgenic mice. These mice expressed the CAT reporter gene under the control of the Ϫ890/ϩ24 apoC-III promoter (C3) fused to the distal Ϫ700/Ϫ410 fragment of the apoAIV promoter (dA4), the Ϫ377/Ϫ357 HRE (⌬E) thus being deleted (Fig. 3B). Three foun-der mice, each of which expressed 10 -20 copies of the dA4⌬E-C3-CAT transgene, were identified.
The pattern of expression of the reporter gene in the intestine was determined for each founder (Fig. 4) and compared with that of the three previously published transgenic mice expressing the CAT reporter (i) under the control of the human Ϫ890/ϩ24 apoC-III promoter (C3-CAT), (ii) under the control of the Ϫ890/ϩ24 apoC-III promoter fused to the Ϫ310/Ϫ700 apoA-IV distal promoter region (dA4-C3-CAT), and (iii) under the control of the Ϫ500/Ϫ890 apoC-III promoter fused to the Ϫ700/ϩ10 apoA-IV promoter (eC3-A4-CAT) (schematic diagram in Fig. 3A). The last two of this transgenic lines displayed similar patterns of transgene expression. In dA4⌬E-C3-CAT mice, CAT activity was constant and similar in the proximal, middle, and distal segments of the intestine (Fig. 4A), as observed in C3-CAT mice (20). Conversely, we observed a gradient of CAT activity, decreasing along the cephalocaudal axis, in mice expressing the dA4-C3-CAT transgene that retained the distal apoA-IV HRE (20). Thus, deletion of the distal apoA-IV HRE results in abolition of the cephalocaudal gradient of expression.
Deletion of the Distal ApoA-IV HRE Abolishes the Restriction of Gene Expression to Villus Enterocytes-We analyzed CAT gene expression along the crypt-to-villus axis of dA4⌬E-C3-CAT mice by in situ hybridization. In dA4⌬E-C3-CAT mice (Fig. 4B, a), CAT gene transcripts were detected in epithelial cells along the entire length of the crypt-to-villus axis but were more abundant in crypts than in villi. In eCIII-AIV-CAT mice (Fig. 4B, b), the expression of the transgene was restricted to villus cells (20). The specificity of the signal was assessed by hybridization of the sense CAT riboprobe to jejunum sections from dA4⌬E-C3-CAT transgenic mice and of the antisense CAT riboprobe to jejunum sections from non-transgenic mice. Fig.  4B, c and d, show a few scattered grains corresponding to the background signal for dark-field microscopy. The expression pattern of the CAT reporter gene was confirmed by RT-PCR analysis of mRNA isolated from villus or crypt cells obtained by laser microdissection of intestinal sections from transgenic mice (Fig. 4C). We also probed each section for 18 S RNA, to check the efficiency of RNA extraction and reverse transcription (data not shown). CAT mRNA was present in both crypt and villus cells from dAIV⌬E-CIII-CAT mice but was detected only in villus cells from eCIII-AIV-CAT mice. Overall, these results demonstrate that deletion of the distal apoA-IV HRE abolishes intestinal villus-specific gene expression.
HNF-4 ␣ and ␥ Are Differentially Expressed in Crypt and Villus Cells-We carried out electrophoretic mobility shift and antibody-mediated supershift assays with nuclear extracts prepared from mouse intestinal villi to identify the intestinal transcription factors that bound to the distal apoA-IV HRE. As expected from the results shown in Fig. 2, the human distal apoA-IV HRE formed a single major complex with nuclear extracts prepared from villi (Fig. 5A, lane 1). This complex was partly due to HNF-4␣ binding, as demonstrated by anti-HNF-4␣ antibody supershift assay (lane 4). However, the complex was not totally supershifted with increasing amounts of anti-HNF-4␣ antibody (lanes 4 -6). This suggests that a number of different complexes may form with the HRE. Supershift assays were performed with several antibodies, directed against various nuclear receptors known to bind an HRE, in addition to the anti-HNF-4␣ antibody. ApoA-I regulatory protein-1, ErbA-related protein 3, RAR␣, and RXR␣ were not involved in these subcomplexes (data not shown). Only an antibody specifically directed against HNF-4␥ resulted in a total supershift of the complex in association with the anti-HNF-4␣ antibody (lane 7) or a partial supershift of the complex  3. Generation of dA4⌬E-C3-CAT transgenic mice. A, schematic representation of the ϳ15-kilobase pair human AI-CIII-AIV gene cluster and of that of transgenic mice, as previously described (20). The direction of transcription is indicated by an arrow for each gene. The proximal promoters are sufficient for in vivo hepatic expression of apoA-I and apoC-III. The Ϫ500/Ϫ890 apoC-III enhancer is required for in vivo expression of apoA-I in the intestine. Arrows indicate the orientation of the region, from 5Ј to 3Ј. B, map of the CAT construct dA4⌬E-C3-CAT, driven by the entire human Ϫ890/ϩ24 apoC-III promoter (C3-CAT), fused to the human Ϫ700/Ϫ410 apoA-IV promoter region. C, Southern-blot analysis of the dA4⌬E-C3-CAT transgene in transgenic mice. Three transgenic mouse lines, d⌬E43, d⌬E95, and d⌬E138, were analyzed for the presence of the dA4⌬E-C3-CAT transgene, as described under "Experimental Procedures." The number of transgene copies in each line was determined by comparison with lanes containing 1-20 copies of transgene DNA. alone (lane 8). Similar results were obtained with the mouse distal apoA-IV HRE used as a probe (Fig. 5B). The anti-HNF-4␥ antibody clearly did not interact with the HNF-4␣ isoform because no supershift of the complex formed with extracts prepared from COS-1 cells producing HNF-4␣ but not HNF-4␥ was observed with this antibody (Fig. 5C).
We investigated the specificity of villus nuclear extracts by comparing their binding pattern with that of nuclear extracts prepared from crypt cells. The crypt nuclear extracts formed an identical major complex with the distal apoA-IV HRE (Fig. 5A,  lanes 9 -14). Interestingly, this complex was almost completely supershifted by the anti-HNF-4␣ antibody (lane 12) and was supershifted to a much lesser extent by the anti-HNF-4␥ antibody (lane 14). The slight supershift observed with the anti-HNF-4␥ antibody may have resulted from contamination of the crypt cell fraction by the more abundant villus cells, as confirmed by a sucrase-isomaltase assay, sucrase-isomaltase activity being a specific marker of differentiated enterocytes (data not shown). These results suggest that HNF-4␥ is mostly produced in the villus, whereas HNF-4␣ is produced in both the villus and crypts. This differential pattern of expression of HNF-4 isoforms in the intestine was confirmed by immunolocalization of the two isoforms along the crypt-to-villus axis. Immunochemical analysis with specific anti-HNF-4␣ and anti-HNF-4␥ antibodies confirmed that endogenous HNF-4␣ was produced in the crypt and villus epithelial cells (Fig. 6A). Conversely, HNF-4␥ labeling was restricted to the villus epithelial cells (Fig. 6B).

DISCUSSION
Regulation of the enterocyte differentiation program involves mechanisms based on spatially restricted transcription, controlling the expression of terminal differentiation genes responsible for enterocyte function. In vivo analysis is required to improve our understanding of these mechanisms. We showed in a previous study that the distal region of the apoA-IV promoter is required to restrict apoC-III and apoA-IV gene expression to the villus enterocytes of the small intestine (20). In this study, we identified within this region the sequence responsible for this spatial restriction of gene expression.
This sequence contains an HRE that has never before been described. HREs have also been identified in the proximal promoters of each of the genes of the apoA-I/C-III/A-IV cluster and in the apoC-III enhancer, raising questions about the respective functions of these elements. In vitro transfection assays in the hepatic HepG2 and enterocytic Caco-2 cell lines showed that transcription of the apoA-IV, apoA-I, and apoC-III genes requires the corresponding proximal HRE in each case (23,29,(35)(36)(37)(38)(39). This has been confirmed in vivo for the apoA-I gene (25). The HRE of the apoC-III enhancer is required in vivo for intestinal expression of the apoA-I and apoC-III genes (25). However, we have shown that the apoC-III enhancer is not sufficient to confer the spatial gradient of expression of the genes of this cluster (20).
HREs are known to bind numerous transcription factors, particularly HNF-4␣ (for review, see Ref. 40). HNF-4␣ has been shown to be involved in the onset of intestinal development in vivo in Drosophila, in which intestinal development stops in the absence of the HNF-4␣ homolog (41). Specific inactivation of HNF-4␣ in the liver leads to abolition of the expression of apolipoprotein genes (42). We report here that the distal HRE of the apoA-IV promoter also binds HNF-4␣. These results were obtained with crude nuclear extract obtained from intestinal villi or crypt epithelial cells. Previous studies with Caco-2 nuclear extracts show that the proximal HRE of the three genes of the cluster and the HRE of the FIG. 4. Distribution of CAT activity along the cephalocaudal and the crypt-to-villus axes of the small intestine in dA4⌬E-C3-CAT transgenic mice. A, distribution of CAT activity along the length of the small intestine in dA4⌬E-C3-CAT transgenic mouse lines. Samples from homogenates of proximal, middle, and distal segments of the small intestine were assayed for CAT activity, as described under "Experimental Procedures." The CAT activity of each intestinal segment (proximal, middle, and distal) is expressed as a percentage of total CAT activity in the small intestine of each transgenic mouse line. B, dark-field images of CAT mRNA in jejunal sections from a dA4⌬E-C3-CAT transgenic mouse hybridized in situ with an antisense CAT riboprobe (a) and similar images for mRNA from jejunal sections from a eCIII-AIV-CAT transgenic mouse (b). The control was performed by hybridizing a jejunal section from a transgenic mouse with a sense CAT riboprobe (c) or a jejunal section from a control mouse with an antisense CAT riboprobe (d); d, only scattered grains are present in the sections. Note that CAT mRNA levels were higher in the epithelial cells of crypts than in those of villi from dA4⌬E-C3-CAT transgenic mice. Bar, 0.1 mm. C, RT-PCR of mRNA samples from crypts and villi obtained by the laser microdissection of jejunal sections from transgenic mice. RNA extraction, RT-PCR, and Southern-blot analysis were performed as described under "Experimental Procedures." Note that CAT mRNA was detected in both crypts and villi from dA4⌬E-C3-CAT mice but only in villi from eC3-A4-CAT mice.
apoC-III enhancer also bind HNF-4␣ (23,29,35,36). HNF-4␣ mRNA has also been detected in the epithelial cells lining both the villi and the crypts of the intestine (36). Our immunohistochemical analyses provide further evidence that the HNF-4␣ protein is present along the entire length of the crypt-to-villus axis. Therefore, HNF-4␣ alone is unlikely to be responsible for the spatial restriction of gene expression to villus enterocytes.
Various mechanisms may account for the silencing of apolipoprotein genes in the crypt cells. First, the access of HNF-4␣ to the distal HRE of the apoA-IV promoter may be impeded by the structure of chromatin in this region, thereby affecting transcription in the crypt cells. Second, the activity of HNF-4␣ in crypt epithelial cells may be affected by the impairment of posttranscriptional modifications such as phosphorylation (43)(44)(45) and acetylation (46) or by specific protein-protein interactions (47). Third, the transcription of apolipoprotein genes in villus enterocytes may require another HRE binding transcription factor produced in the villus but not in the crypt. HRE motifs act as binding sites for homo-or heterodimers of ligand-dependent nuclear receptors and for homodimers of orphan receptors. Using specific antibodies directed against nuclear receptors and nuclear extracts from villus epithelial cells, we showed by supershift assay that the HRE of the distal apoA-IV promoter binds only the HNF-4␥ isoform in addition to the ␣ isoform. Immunohistochemical analysis also showed that HNF-4␥ was restricted to the villi.
The functional role of HNF-4␥ is unclear (48,49). Our results suggest that the ratio of HNF-4␥ to HNF-4␣ may play a specific role in controlling the spatial pattern of gene expression along the crypt-to-villus axis in vivo. Variations of the ratio between different isoforms of transcription factors, such as HNF-1 and GATAs, have been implicated in the control of gene expression in Caco-2 cells. The differential binding of HNF-1␣ and -␤ to their target on the sucrase-isomaltase promoter plays a role in the transcriptional control of this gene (10,11). GATA Ϫ4, Ϫ5, and Ϫ6 display differential patterns of expression in crypt and villus epithelial cells, and these factors activate intestinal genes such as intestinal fatty acid-binding protein (50), lactasephlorizin hydrolase (17,51), and sucrase-isomaltase (52) differently.
In conclusion, our results demonstrate that the presence of an HRE in the distal part of the apoA-IV promoter is necessary in vivo to restrict the expression of a reporter gene to villus epithelial cells. Our results suggest that the spatial pattern of expression of apolipoprotein genes may be controlled by the HNF-4␥/HNF-4␣ ratio, indicating for the first time a potential role for HNF-4␥ in the specific pattern of gene expression. Furthermore. in vivo cell type-specific conditional gene invalidation studies will provide further insight into the respective roles of HNF-4 isoforms in the spatial pattern of gene expression in the intestine.