| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Unité Mixte de Recherche 505, Institut National de la Santé et de la Recherche Médicale-Université Pierre & Marie Curie, 75006 Paris, France
Address all correspondence and requests for reprints to: Agnès Ribeiro, Unité Mixte de Recherche 505 Institut National de la Santé et de la Recherche Médicale-Université Pierre & Marie Curie, 15 rue de l’Ecole de Médecine, 75006 Paris, France. E-mail: agnes.ribeiro-pillet-u505{at}bhdc.jussieu.fr.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
and 
. Here, intraduodenal injections in mice and infections of human enterocytic Caco-2/TC7 cells with an adenovirus expressing a dominant-negative form of HNF-4 repress the expression of the apoA-IV gene, demonstrating that HNF-4 controls the apoA-IV gene expression in enterocytes. We show that HNF-4 and functionally interact with a second HRE present in the proximal region of the human apoA-IV promoter. New sets of transgenic mice expressing mutated forms of the promoter, combined with the human apo C-III enhancer, demonstrate that, whereas a single HRE is sufficient to reproduce the physiological cephalo-caudal gradient of apoA-IV gene expression, both HREs are required for expression that is restricted to villi. The combination of multiple HREs may specifically recruit regulatory complexes associating HNF-4 and either coactivators in villi or corepressors in crypts. |
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
The precise mechanisms involved in the regulation of the enterocyte differentiation program, based on a spatial restriction of the transcription of genes responsible for enterocyte functions, have not been determined. The intestinal expression of the human apoA-I and C-III genes requires a cooperation between their respective proximal hormone-responsive element (HRE) and a HRE located in the apoC-III enhancer. This has been demonstrated in vivo (7) and in vitro in hepatic HepG2 and enterocytic Caco-2 cell lines (8, 9, 10, 11, 12). Such a cooperation has also been reported in vitro between the HRE located in the apoC-III enhancer and the proximal (–128/–141) HRE of the human apoA-IV promoter (AIV-C) (13, 14), but its role in the enterocyte-specific expression of apoA-IV has not yet been tested in vivo. We have shown in transgenic mice that the human apoC-III enhancer is not sufficient to confer the spatial cephalo-caudal and crypt-villus gradients of expression to the genes of the A-I/C-III/A-IV cluster (15). Furthermore, we have recently identified a distal (–357/– 377) HRE (AIV-E) in the human apoA-IV promoter that is necessary to confer the physiological pattern of apoA-IV gene expression in the intestine (16). Altogether, these results raise questions about the respective roles of the multiple HREs present in the apoA-I, C-III, and A-IV promoters in the restriction of the expression of apolipoprotein genes to villus enterocytes. In enterocytes, the distal HRE (AIV-E) of the apoA-IV promoter binds two distinct isoforms of hepatocyte nuclear factor 4 (HNF-4), and (16), which are encoded by two different genes (17, 18). Their intestinal pattern of expression led us to hypothesize that they could be involved in intestinal differentiation and regulation of gut-specific gene expression, as previously has been reported for several other transcription factors, e.g. cdx1, cdx2 (19, 20, 21), and the GATA 4, 5, 6 family (22, 23).
HNF-4 is a highly conserved member (NR2A1) of the nuclear receptor superfamily (24, 25). The presence of fatty acids in the ligand binding domain has been reported (26, 27), but the functional role of this association is still a matter of debate (28, 29, 30). The highest amounts of HNF-4 are found in the liver, the kidney, the intestine, and the pancreas in mammals and in the homologous structures in invertebrates (25, 31). HNF-4 is expressed in the primitive endoderm as soon as it becomes morphologically distinct and is important for its differentiation (32, 33). HNF-4 is not necessary for specification and early development of the liver (34). However, Hnf-4-/– hepatoblasts fail to become mature hepatocytes, so this differentiation process requires HNF-4 (35, 36). The phenotype of a conditional knockout of Hnf-4 indicates that HNF-4 is also central to the maintenance of differentiated hepatocytes and acts in vivo as a major hepatic regulator of genes involved in the control of lipid metabolism, e.g. apoB, C-II, C-III, A-II, and A-IV genes (37). The function of the recently identified HNF-4 has not yet been clearly established (17, 18).
The aim of the present study was to demonstrate the role of HNF-4 in the intestinal expression of the apoA-IV gene and to determine the respective importance of the two HREs of the human apoA-IV promoter in its spatial pattern of expression. Using an adenovirus encoding a dominant-negative form of HNF-4, we showed that HNF-4 controls the intestinal expression of the apoA-IV gene in mouse models in vivo, and in human enterocytic Caco-2/TC7 cells. Finally, we characterized the properties of HNF-4 and binding to both HREs, and demonstrated in vivo, with transgenic mice, that both HREs are essential, in combination with the apoC-III enhancer, for the villus-specific expression of the apoA-IV gene.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Decreases apoA-IV Gene Expression in Vivo in Murine Small Intestine and ex Vivo in Caco-2/TC7 Cells and isoforms in intestinal apoA-IV gene expression, we used a dominant-negative form of HNF-4 lacking the F region and the AF-2 activation domain (38). The dominant-negative HNF-4 (DN-HNF4) binds to DNA as a dimer and inhibits the HNF-4 activity either by heterodimerization with the endogenous HNF-4 or by competitive binding to the HREs (38). We therefore investigated the functional role of the dominant-negative HNF-4 on the transactivation of the human apoA-IV promoter by HNF-4 (Fig. 1A
). We transiently transfected COS-7 cells with increasing amounts of the dominant-negative HNF-4, a constant amount of HNF-4 (or HNF-4 as a control), and the chloramphenicol acetyltransferase (CAT) reporter gene under the control of the human apo A-IV promoter. Stimulation of the promoter by either HNF-4 or HNF-4 was similarly inhibited by the dominant-negative HNF-4 in a dose-dependent manner (Fig. 1A). Thus, the dominant-negative HNF-4 impaired the transactivation activities of HNF-4 and HNF-4.
|
to transduce human Caco-2/TC7 cells. These cells differentiate spontaneously after confluence into enterocytic cells, representative of small intestine enterocytes. The apo A-IV gene is not expressed in dividing Caco-2/TC7 cells but is expressed after they have differentiated. Dose-response (Fig. 1B
) and time-course experiments (Fig. 1C) were thus performed on confluent cells using the dominant-negative HNF-4 or an adenovirus expressing GFP (the green fluorescent protein) (Ad-GFP) as a control. As a control, we used RT-PCR to check that the dominant-negative HNF-4 was transcribed in transduced Caco2/TC7 cells (Fig. 1B, inset). The sequence of the reverse primer used to detect this mRNA is absent in the endogenous HNF-4 mRNA; it is only present in the cloning vector and thus specific for the truncated HNF-4 (Table 1). At 5 d after infection, we observed a dose-dependent decrease in the apoA-IV mRNA level, with an 80% reduction at 50 pfu/cell (Fig. 1B). Adenovirus-mediated expression of the dominant-negative HNF-4 reduced the level of apoA-IV mRNA after 2 d, and this inhibition was still observed 5 d after infection (Fig. 1C). The specific effect of the dominant-negative HNF-4 was also assessed by measuring L19 mRNA levels. As expected, the mRNA levels of this large ribosomal polypeptide were not affected by the dominant-negative HNF-4 (Fig. 1D). Conversely, the dominant-negative HNF-4 induced a 60% decrease in the mRNA levels of apoB (Fig. 1D), a gene involved in lipid metabolism and known to be repressed in mice hepatocytes lacking HNF-4 (37).
|
). The complete renewal of the intestinal epithelium from stem cells located in the crypts takes 3–4 d in mice. Because both GFP fluorescence and luciferase activities were still detectable 9 d after in vivo gene transfer (not shown), the stem cells of crypts must have been infected. Validated in this way, we applied this adenoviral strategy to the expression of the dominant-negative HNF-4 in mouse intestine. After intestinal injection, we first used RT-PCR to verify the expression of the dominant-negative HNF-4 mRNA in epithelial cells. Specificity was confirmed by the lack of any RT-PCR product in the sample from control mice injected with the Ad-GFP construction (Fig. 2B, inset). As shown in Fig. 2B, the expression of the dominant-negative form of HNF-4 significantly reduced the apoA-IV mRNA level (by 30%, P 
0.05) in intestinal epithelial cells when compared with the control mice injected with the same amount of Ad-GFP. As an internal control, we analyzed the expression of the oxysterol receptor [liver X receptor (LXR) ] gene and sterol response element-binding protein (SREBP)-1c, two genes that are not affected by the hepatic knockout of the HNF-4 gene (37). LXR and SREBP-1c mRNA levels did not vary significantly (Fig. 2C). Altogether, these experiments demonstrate that HNF-4 controls the expression of the apoA-IV gene in vivo and ex vivo in enterocytes.
|
and Binding to the Proximal and Distal Regions of the apoA-IV Promoter) that binds the HNF-4 and isoforms (16). A second HRE (AIV-C, Fig. 3A) is present in the proximal region of the apoA-IV promoter (13). Comparison of the sequences of the human and mouse distal and proximal apo A-IV promoter regions (nucleotide –700 to +10, GenBank accession nos. gi 178756 and gi 192006), using the basic local alignment and search tool sequence alignment program, showed that the distal HRE (AIV-E) and the proximal HRE (AIV-C) were conserved in the two species (Fig. 3B).
|
) with both probes. The HNF-4/DNA complex was completely supershifted with its corresponding antibody (Fig. 4A, lanes 2 and 8), but not with the anti-HNF-4 antibody (Fig. 4A, lanes 3 and 9). Likewise, the HNF-4/DNA complex (Fig. 4A, lanes 6 and 12), but not the HNF-4/DNA complex (Fig. 4A, lanes 5 and 11), was completely supershifted with the anti-HNF-4 antibody. We used whole cell extracts from these transfected COS-7 cells to determine the relative affinities of each HNF-4 isoform for the distal (probe E) and proximal (probe C) HREs of the apoA-IV promoter. For these studies, the radiolabeled HRE was competed out with increasing amounts of homologous nonradioactive competitor (Fig. 4, B and C) or with the other HRE in heterologous competition (Fig. 4D). The relative affinities of HNF-4 and for the HREs were estimated from the slopes of the linear relationships (Fig. 4, B–D). HNF-4 and HNF-4 displayed the same affinity for the distal HRE (AIV-E) (Fig. 4B) and for the proximal HRE (AIV-C) (Fig. 4C). Heterologous competitions revealed that HNF-4 (Fig. 4D) and HNF-4 (published as supplemental data Fig. S1 on The Endocrine Society’s Journals Online web site at http://mend.endojournals.org) display a better binding to the distal HRE (AIV-E) than to the proximal HRE (AIV-C) (1.5- and 4-fold, respectively). Taken together, these results indicated that HNF-4 and HNF-4 bind slightly better to the distal than to the proximal HRE and that the two isoforms display comparable affinities for each HRE.
|
). HNF-4 and both transactivated the eC3A4 construction in COS-7 cells in a similar dose-dependent manner (Fig. 4E). Moreover, cotransfection with both isoforms led to a similar dose-dependent transactivation (Fig. 4E). In Caco-2 cells, overexpression of the two isoforms led to similar transactivation levels (Fig. 4E), although they were lower than the levels reached in COS-7 cells (1.8-fold compared with 10- to 13-fold). This difference results from the high level of endogenous HNF-4 in Caco-2 cells. We concluded from these transient transfections that the two HNF-4 isoforms activated the apoA-IV promoter at comparable levels.
Ex Vivo Analysis of the Respective Roles of the Two HREs in the Expression of the apoA-IV Gene in Enterocytes
The involvement of the proximal HRE (AIV-C, Fig. 3A) in the apoA-IV gene expression in enterocytes was studied in cells transfected with the CAT reporter gene under the control of the human –700/+10 apoA-IV promoter linked to the –890/–520 apoC-III enhancer (eC3A4). We introduced mutations (described in Fig. 5A) in the distal HRE (AIV-E), in the proximal HRE (AIV-C) or in both elements (eC3A4Em, eC3A4Cm, and eC3A4EmCm; see Fig. 5C).
|
in Ref.13), and to the distal HRE (AIV-E), as shown in Fig. 5B. With the E probe, the single and specific complex observed with HNF-4 is not competed out with a 100-fold excess of mutated nonradiolabeled competitor (Fig. 5B, lane 3).
Because 1) Caco-2/TC7 cells can only be transfected when they are dividing and 2) the apoA-IV gene is only expressed in postconfluent and well-differentiated enterocytic Caco-2/TC7 cells, we generated permanent Caco-2/TC7 cell lines stably expressing the different constructs. Figure 5C shows that in the stable cell lines CAT activity decreased dramatically when a single site was altered. The activities of the mutated promoters remained significantly above background levels. From these data, we conclude that in enterocytic cells the full activity of the apoA-IV promoter requires the integrity of both the distal and the proximal HREs.
In Vivo Analysis of the Respective Roles of the Two HREs in the Expression of the apoA-IV Gene in Enterocytes
We investigated the respective binding pattern of the two HREs using nuclear extracts prepared from villus or crypt cells. Both nuclear extracts produced a single slowly migrating band with either the proximal HRE (AIV-C) (lanes 1 and 5, Fig. 6A) or the distal HRE (AIV-E) probe (lanes 9 and 13, Fig. 6A). In each case, HNF-4-directed antibodies produced a supershift (Fig. 6A, lanes 2, 6, 10, and 14), but a complex remained at the initial position when using nuclear extracts purified from villi (Fig. 6A, lanes 2 and 10). We have previously shown for the distal HRE (AIV-E) that this complex could not be displaced by increasing amounts of anti-HNF-4. Because different nuclear receptors can bind a HRE, we tested various antibodies and demonstrated that HNF-4 was present in the complex with the distal HRE (AIV-E), but not apolipoprotein A-I regulatory protein-1, v-erbA-related protein 3, retinoic acid receptor , or retinoid X receptor (16). However, their migration patterns in band-shift assays are different from those of HNF-4. Here, the presence of a supershifted complex with an antibody directed against HNF-4 clearly revealed that this isoform was present in the villi nuclear extracts and could bind to AIV-C and to AIV-E (Fig. 6A, lanes 3 and 11). Furthermore, when anti-HNF-4 and anti-HNF-4 antibodies were added together, the complexes formed with villus nuclear extracts were supershifted (Fig. 6A, lanes 4 and 12). Conversely, the complex formed with crypt nuclear extracts was almost completely supershifted by the anti-HNF-4 antibody (Fig. 6A, lanes 6 and 14), but not by the anti-HNF-4 antibody (Fig. 6A, lanes 7 and 15). This is consistent with the fact that HNF-4 is only expressed in villi (16). The slight supershift observed with the anti-HNF-4 antibody resulted from a contamination of the crypt cell fraction with the more abundant villus cells (supplemental data Fig. S2). Altogether, these results demonstrate that both proximal (AIV-C) and distal (AIV-E) HREs of the apoA-IV promoter display the same binding profile of villus and crypt nuclear extracts.
|
, B and C, respectively). The CAT activity was determined along the small intestine for each transgenic mouse line (supplemental data Figs. S3 and S4). A decreasing cephalo-caudal gradient of CAT activity was observed when the reporter gene was controlled by eC3A4, eC3A4Cm or eC3A4Em promoters, i.e. constructions that retained one functional HRE in the human apoA-IV promoter (Fig. 6B). However, the gradient was suppressed in mice expressing the eC3A4CmEm-CAT transgene in which both HREs of the apoA-IV promoter are mutated (Fig. 6B). To further assess the role of the proximal HRE (AIV-C), we generated additional transgenic mice expressing the CAT transgene under the control of the (–520/–890) apoC-III enhancer fused to the (–310/+10) proximal apoA-IV promoter, named eC3p4. As already observed in mice expressing the eC3A4Em or eC3A4Cm construction, we measured a decreasing gradient of CAT activity along the cephalo-caudal axis in these mice (Fig. 6B). Thus, in vivo, the presence of a single HRE within the apoA-IV promoter is sufficient to maintain the cephalo-caudal gradient of expression.
CAT expression was then analyzed by immunohistochemistry along the crypt-villus axis of the transgenic mice (Fig. 6C). The specificity of the signal was verified using jejunum sections from non transgenic mice (Fig. 6C-a) and from eC3A4 transgenic mice as controls (Fig. 6C-b). In the latter, the transgene expression was restricted to villi, as previously shown by in situ hybridization (16). In transgenic mice in which the proximal HRE (AIV-C) (Fig. 6C-c) or the distal HRE (AIV-E) (Fig. 6C-d) was mutated, the CAT protein was detected in the villus epithelium (comparing the intensity of the signal in Fig. 6C, villus insets) and accumulated in crypts (comparing the fluorescence signals in Fig. 6C, crypt insets). Moreover, immunochemical analysis studies revealed that the CAT reporter gene was expressed in crypt and villus cells of eC3pA4 transgenic mice (Fig. 6C-e).
The CAT reporter gene expression patterns were confirmed by RT-PCR analysis of mRNA extracted from villus or crypt cells, isolated by laser microdissection of intestinal sections from these transgenic mice (Fig. 6D). The same mRNA samples were used to analyze endogenous apoA-IV expression by RT-PCR. The lack of apoA-IV signal from transgenic mice crypts and, to a lesser extent, from the control mice crypts when compared with the signals from villi, shows that these fractions were almost free of villus mRNA. As expected, CAT expression was only detected in villi of the eC3A4 transgenic mice. On the other hand, CAT expression was detected in crypts and villi from the eC3A4Cm, eC3A4Em, and eC3A4CmEm transgenic mice. 18S RNA was present at the same levels in crypt and villus fractions from all mice (Fig. 6D). Thus, in vivo, mutation of a single HRE lowers the apoA-IV promoter activity in enterocytes. More strikingly, this allows the expression of the reporter gene in crypts, abolishing the spatial restriction of the apoA-IV gene expression to the villi. Altogether, these results suggest that in transgenic mice, both HREs of the human apo A-IV promoter are necessary to restrict gene expression to villus epithelial cells.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
and HNF-4 isoforms bind similarly to the distal and proximal HREs of the human apoA-IV promoter. Both HREs are necessary to restrict transcription of a reporter gene to villus epithelial cells, whereas the expression pattern that decreases along the cephalo-caudal axis only requires one or the other HRE.
Using an adenovirus coding for a dominant-negative form of HNF-4, we reduced the expression of apoA-IV in enterocytic Caco-2 cells in a time- and dose-dependent manner. This dominant-negative form has already been used to impair the effect of glucose on the hepatic expression of apoA-II gene, another target of HNF-4 (41). The dominant-negative HNF-4 lacks the C-terminal transactivating domain but is still able to dimerize and bind to DNA (38). Considering the overall similarities and identities in the amino acid sequences of HNF-4 and , we might expect that the dominant-negative HNF-4 could also dimerize with HNF-4. We found no experimental evidence for such heterodimerization despite many attempts and different approaches. Two helical domains, H9 and H10, play an essential role for the dimerization of HNF-4 polypeptides (39). The HNF-4 sequence is well conserved among human, rat, and mouse; this is also the case for HNF-4 between human and mouse (the rat sequence has not been reported). It is noteworthy that the sequence of helix 10 is fully conserved between the and isoforms, in human and mouse, whereas that of helix 9 diverges (six differences out of 21 residues). We thus believe that inhibition results from a competitive binding to the different HREs. The binding of another nuclear receptor is unlikely because, in intestinal nuclear extracts, only HNF-4 and bind to the HREs of the apoA-IV promoter (see also Ref.16).
We infected the small intestine epithelium of mice with this adenovirus and observed a significant decrease in the apoA-IV gene expression. Differences in adenovirus transduction efficiency ex vivo and in vivo, as well as the rapid self-renewal of the intestinal epithelium, most likely account for the smaller decrease in apoA-IV expression observed in vivo as compared with that observed in cell cultures. Furthermore, the dominant-negative HNF-4 also affects the expression of genes involved in the lipid metabolism of the intestine, as has already been demonstrated for the liver with a conditional knockout of Hnf-4 (37). These results show that recombinant adenoviruses can be used to investigate transcriptional regulation in the intestinal epithelium. In the liver, HNF-4 plays a central role in hepatocyte differentiation and in the maintenance of differentiated hepatocytes by regulating the expression of genes involved in the control of lipid metabolism (35, 37). Our data demonstrate that the HNF-4 transcription factor family is also a major regulator of cell-specific gene expression in the intestine and could play a central role in digestive epithelium development, enterocyte differentiation, and the maintenance of enterocyte functions, and particularly, that of lipid metabolism.
The second question addressed by this study was to determine the respective roles of the different HREs in the control of apoA-IV gene expression along the intestinal crypt-villus and cephalo-caudal axes. HREs, the regulatory sequences bound by HNF-4 transcription factors, have been shown to play a major role in the expression of the apoA-I/C-III/A-IV gene cluster. The intestinal expression of these genes requires the HRE located in the apoC-III enhancer (7, 8, 42, 43). In this work, we demonstrate that the physiological gradient of apoA-IV gene expression along the cephalo-caudal axis requires either one of the HREs located within the promoter (Fig. 7A). Conversely, the two HREs of the apoA-IV promoter are necessary to restrict the expression to the villi (Fig. 7A). Thus, distinct elements appear to be necessary to restrict apoA-IV expression to enterocytes and to direct transcription of the gene along the duodenum-ileal axis. This has also been reported for the rat intestinal fatty acid binding protein (FABP-I) gene. The FABP-I promoter (–103/+28) is sufficient to direct transcription of this gene along the duodenum-colon axis. However, upstream sequences are needed to confine FABP-I expression to the differentiated enterocytes of the villus. In particular, a 20-bp element located between nucleotides –263 and –244 of the promoter prevents FABP-I expression in crypt cells (44, 45). This is also the case for the rat liver FABP and the sucrase-isomaltase genes (46, 47).
|
is expressed in villi and crypts, whereas HNF-4 is only expressed in villi (16), one could hypothesize that HNF-4 is the factor responsible for apoA-IV gene restriction to villi. In vitro, both isoforms display comparable affinities for the apoA-IV HREs and similar transactivating properties in COS-7 and CaCo-2 cells. Therefore, it is unlikely that the intestinal expression of the apoA-IV gene results solely from a competition between HNF-4 and HNF-4.
The expression of apoA-IV along the crypt-villus axis relies on a two-step process: gene silencing in epithelial cells while they remain in crypts, followed by gene activation when these cells reach the villi as differentiated enterocytes. The presence of CAT activity in the crypts of transgenic mice with mutated apoA-IV promoters demonstrates that the integrity of the two HREs is required for gene silencing in crypts. Because only HNF-4 binds to these HREs in nuclear extracts from crypts, this suggests its involvement in the apoA-IV gene silencing. HNF-4, like other nuclear receptors, interacts with coactivator and corepressor complexes, containing histone acetyltransferase or histone deacetylase (HDAC) activities (for review see Ref.48). Histone acetyltransferase and HDAC ultimately control the balance between chromatin decondensation, a transcriptionally permissive state, and chromatin condensation, a transcriptionally silent state (for review see Ref.49). In accordance with this, it has been shown recently that in the developing mouse intestine the Fabpi, Fabpl, and apoA-I genes, which are expressed in differentiated enterocytes, are specific targets for developmental modulation by class I HDAC (50). Thus, recruitment of endogenous class I HDAC by HNF-4 may participate in villus-restricted gene expression.
Recruitment of such coregulators by target promoters occurs mainly through direct physical interactions with the transactivation domains of nuclear receptors, with these interactions being modulated by the binding of the ligands. Wisely et al. (27) showed that HNF-4 constitutively binds fatty acids as structural ligands, raising the question of the mechanism of the switch from corepressor to coactivator complexes (30). Interestingly, it has been reported that the recruitment of different coregulators, and the subsequent promoter responses, differed between two HNF-4 isoforms (51, 52), and that this recruitment is mediated by the F domain of HNF-4 (53). This F domain exhibits only 32% identity between and isoforms (18).
As observed in the cellular model (Fig. 5C) and in vivo (Fig. 6C) mutation of a single HRE lowers the apo A-IV promoter activity in enterocytes. Thus, apoA-IV gene expression is not constitutive in enterocytes, but requires activating signals provided by interactions between HNF-4 or dimers and enterocyte-specific coactivators. Conversely in crypts, where HNF-4 is not expressed, the repression of apoA-IV expression could be due to the recruitment of corepressors by the HNF-4 homodimers (Fig. 7B). Our observations in cell lines and in transgenic mice strongly suggest that the activating or the inhibiting complexes should comprise at least three HNF-4 homodimers.
In conclusion, our results demonstrate that in the intestinal epithelium the specific pattern of apoA-IV gene expression along the crypt-villus axis is controlled by the binding of HNF-4 transcription factors to the two HREs located in the promoter and in the apoC-III enhancer. The switch observed in apoA-IV gene activity in epithelial cells from a silenced status in crypts to an active status in villi most likely requires the formation of specific transcriptional complexes recruiting corepressors in crypts and coactivators in villi to remodel chromatin. Such a mechanism, which has to be demonstrated, would confer a major role for HNF-4 in the onset and maintenance of differentiated functions in enterocytes, and particularly in lipid metabolism.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
, expressing a truncated version of rat HNF-42 lacking the C-terminal residues 361–465, was constructed from pCDNA1-HNF-4-Cd1b (38). pCDNA1-HNF-4-Cd1b was digested with HindIII and NotI; the Cd1b-containing fragment was then subcloned by blunt ligation into the EcoRI restriction site of the pTg-6600 vector. The recombinant adenovirus was then generated by in bacteria homologous recombination with the pKp1.3 vector. The production of infectious particles was performed by the GVPN (Gene Vector Production Network of Genethon, Evry, France). A recombinant adenovirus, Ad-GFP, containing a cytomegalovirus (CMV) promoter/GFP gene cassette was used as a control. The in vivo adenovirus infection experiments were performed as described in Cheng et al. (40) with modifications. The day before adenovirus injection, female B6CBA mice were fasted for 5 h and were orally loaded at 14 and 18 h with 0.3 ml of a mixture containing 100 mg of maternalized milk (Bledilait HA, Danone, France), 20% of acetylcysteine (wt/vol) (Mucomyst, Bristol-Myers Squibb, Paris, France) and ampicillin (60 mg/kg) in coloscopic liquid (Biopeg, Fresenius, Paris, France). The next day, they were anesthetized by ip injection of 2,2,2-tribromoethanol (0.25 mg/g, avertine, Sigma, St. Louis, MO). After a small abdominal incision on the right side, 100 µl of a mix containing glucagon (0.1 mg/kg, Sigma), dexamethasone (2 mg/kg, Dexadreson, Intervet, the Netherlands) and ranitidine (2 mg/kg, Raniplex, Fournier, France) in PBS was administered. Mice were divided into two groups (n = 3 in each group). The first group received an Ad-GFP injection and the second, an Ad-DN-HNF4 injection. Adenovirus (100 µl of 2.5 x 109 pfu/mouse in PBS) was injected into the intraduodenal lumen, the injection point was marked with a surgical thread, and the incision was closed. Mice were allowed to recover and given a regular chow diet. Eight days later, mice were euthanized and the small intestine (10 cm from the surgical mark) was removed, washed with PBS, and frozen in liquid nitrogen for further RNA analysis. The experiment was repeated five times. Animal care and the experimental procedures used in this study conform to the French guidelines for animal studies. For in vitro adenovirus infection experiments, Caco-2/TC7 cells were seeded in six-well plates with 8 x 104 cells per well and cultured as previously described (54). Ten days later, the cells were infected with Ad-DN-HNF4 or Ad-GFP (5 or 50 pfu/cell). Two, 3, or 5 d after infection, the medium was removed and cells were recovered for RNA analysis. Experiments were performed in triplicate and repeated three to five times.
Reverse Transcription (RT) Real-Time PCR
Total RNA from infected mouse intestine and Caco-2/TC7 cells was isolated with RNAplus reagent (qBiogene Illkirch, France) according to the manufacturer’s instructions. The RT reactions were performed with 1 µg of RNA in a 20-µl reaction. mRNA was quantified using the Light Cycler System according to the manufacturer’s procedures (Roche Molecular Biochemicals, Meylan, France). PCRs were performed with a 1:1000 final dilution of the RT product as previously described (41). The amount of human apoA-IV mRNA was quantified using hybridization probes designed by Genset (Proligo, Paris, France). The PCR conditions were one step of denaturation (8 min at 95 C) followed by 40 amplification cycles (each cycle consisted of 15 sec at 95 C, 10 sec at 58 C and 15 sec at 72 C). Mice apoA-IV, Cd1b, LXR, and SREBP-1c and human apo B and L19 mRNA levels were quantified using the SYBR Green I dye. The PCR conditions were one step of denaturation (8 min at 95 C) followed by 40 amplification cycles (each cycle consisted of 10 sec at 95 C, 10 sec at 60 C and 10 sec at 72 C). The sequences of primers are shown in Table 1. The quantification of 18S RNA (using Taqman probes from Applied Biosystems, Courtaboeuf, France) was used as an internal control. Results were expressed as the ratio between the mRNA of interest and 18S RNA.
Nuclear Extracts from Mouse Intestinal Epithelial Cells
The epithelium was isolated (55), and villi and crypt cells were isolated by shaking intestine fragments in a chelating buffer (56), as previously described (16). Nuclear extracts from intestinal epithelial cells were then prepared according to Shapiro et al. (57).
Preparation of Whole-Cell Extracts from COS-Transfected Cells
The vector encoding rat HNF-42 (pMT2-HNF-4) is as previously described (11). Mouse HNF-4 cDNA was subcloned into the EcoRI site of pMT2 vector (11) after amplification from pCMV-HNF-4 (18). COS-7 cells were transfected with these two expression vectors by the calcium phosphate coprecipitation method. Whole-cell extracts were prepared as previously described for use in gel retardation assays (58).
EMSAs
The AIV-E and AIV-C double-stranded oligonucleotides used as probes or competitors for EMSAs are described in Fig. 1B. EMSAs were performed as previously described (58) with 2 µl of COS-7 whole cell extract or with 2 µg of nuclear extract from villus or crypt epithelial cells. For competition assays and affinity estimations, the unlabeled double-stranded oligonucleotide and the radioactive probe were added simultaneously. 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., Santa Cruz, CA).
CAT Gene Reporter Plasmids
The details concerning construction of the plasmids eC3pA4, eC3A4Cm, eC3A4Em, and eC3A4CmEm are presented in the supplemental data Fig. S5; eC3A4 has already been described (15). We verified that the mutations were present in the different constructions by sequencing. The mutations introduced in the HREs in elements E and C are indicated in Fig. 3B.
Transient Cell Tranfections
Caco-2/TC7 and COS-7 cells were seeded into six-well plates with 2 x 105 and 105 cells per well, respectively, and were allowed to adhere overnight. Cells were transfected by incubation for 24 h with CAT expressing vector (2.5 µg), HNF-4 expressing vectors (0.5 µg or variable amounts as indicated in the figures for COS-7 cells), and ß-galactosidase expressing vector (0.5 µg) in opti-MEM supplemented with 8 µg of lipofectin reagent (Invitrogen Life Technologies, Carlsbad, CA), according to the manufacturer’s instructions. After transfection, the medium was replaced with supplemented DMEM-glutamax (25 mM glucose, 20% fetal bovine serum), and the cells were incubated for a further 24 h. Experiments were performed in triplicate and repeated three to five times.
Stably Transfected Cells
Caco-2/TC7 cells were cotransfected using the calcium phosphate coprecipitation method with different plasmids expressing the CAT reporter gene as indicated in Fig. 4 and a selection plasmid coding for the neomycin resistance gene. The whole populations of stably transfected cell lines were selected with G418 (400 µg/ml) for 3 wk as previously described (54). To assay the CAT activity in the different lines, the stably transfected cells were grown on a semipermeable insert with an asymmetrical distribution of medium between the apical side (with 20% fetal calf serum) and the basal side (without fetal calf serum) for 21 d to allow the complete epithelium organization of Caco-2/TC7 cells, as previously described (59). Experiments were performed in triplicate and repeated three times.
Generation and Characterization of Transgenic Mice
Transgenic mice were obtained as previously described (15, 16). Briefly, the transgenes were excised from plasmids and microinjected into fertilized eggs from C57BL/6J x CBA/J females. DNA from pups’ tails was analyzed by PCR for the presence of the CAT gene. The DNA from positive mice was then further analyzed by Southern blot to estimate the copy number of the transgene (supplemental data Fig. S6). Three founders of the eC3A4 Cm-CAT transgene were obtained with 2, 1, and 20–30 copies (per genome), respectively. One founder each of the eC3A4 Em-CAT transgene and of the eC3A4 CmEm-CAT transgene were obtained with 1 and 10–20 copies, respectively. Two founders of the eC3pA4-CAT transgene with one and two to five copies, respectively, were identified. The transgenes are integrated at a single locus in each line.
CAT Assays
Individual tissues were homogenized as previously described (15) and transiently or stably transfected cells were recovered. The soluble protein concentration was determined with the Bio-Rad (Hercules, CA) protein assay. CAT activities were measured using a liquid phase assay, as previously described (58). CAT activity is expressed as percentage of acetylation per minute per milligram of protein, after subtracting the background for each tissue from control mice, which do not express the CAT gene. In transient transfections, ß-galactosidase activity was determined to normalize the variability in transfection efficiency (58). In stably transfected cells, the CAT activity was normalized with the protein concentration.
Immunohistochemistry
Sections of small intestine from mice were fixed by incubation for 10 min in 4% paraformaldehyde in PBS and washed six times, for 5 min each, in PBS. The sections were incubated for 30 min in blocking solution (2% BSA, 0.2% Tween, 100 mM glycine in PBS) and then for 1 h at room temperature with anti-CAT-digoxigenin antibody [dilution: 1/50; Roche Molecular Biochemicals (Indianapolis, IN); no. 1 465 066]. The sections were washed six times with PBS, then incubated with blocking solution, followed by the secondary antibody, anti-digoxigenin-rhodamin (dilution:1/20; Roche; no. 1 207 750). After several washes, the samples were mounted with antifading mixture Fluoprep (Biomerieux, Marcy l’Etoile, France). Images were acquired with a Zeiss LSM-510 laser-scanning confocal microscope (Carl Zeiss, Oberkochen, Germany) equipped with Zeiss Axiovert 100M (plan neofluar 16x NA 0.5 oil immersion objective).
Laser Microdissection and mRNA Detection
Laser microdissections of jejunal sections (7 µm) were performed, with an AS LMD Leica system used to collect crypt and villus cells, as described in Sauvaget et al. (16). RNA extractions and RT reactions were performed as previously described (16). The cDNAs were analyzed by real-time PCR amplification after 1/10 dilution. Detection of mouse apoA-IV mRNA and 18S RNA is as described in the RT real-time PCR section. The CAT mRNA was detected using the SYBR Green I dye. The PCR conditions were one step of denaturation (8 min at 95 C) followed by 40 amplification cycles (10 sec at 95 C, 10 sec at 60 C, and 10 sec at 72 C), with CAT primers described in Table 1. PCR products were visualized by agarose gel electrophoresis with ethidium bromide staining.
Statistical Analysis
Results were expressed as means ± SEM. The statistical significance of differences was determined by the unpaired Student’s t test using Excel software.
|
ACKNOWLEDGMENTS |
|---|
expression vector, V. Becette (Department of Anatomo-pathology, Centre René Huguenin, Saint-Cloud, France) for access to the laser microdissection system, and L. Garcia (Généthon, Evry, France) for providing the GFP-expressing adenovirus. We thank the Production and Control Department of Genethon, which is supported by the Association Française contre les Myopathies (AFM), for production of the adenoviral stocks in the context of the GVPN network (http://www.gvpn.org). We also thank M. Séau and C. Lasne (Laboratoire de Transgenèse, IFR 58, Institut des Cordeliers, Paris, France) for technical assistance in the generation of transgenic mice, A. Bobard [by Institut National de la Santé et de la Recherche Médicale (INSERM) Unité 465, Institut des Cordeliers, Paris, France] for PCRs of SREBP-1c and LXR and J-M. Lacorte and R. Vidal (INSERM Unité 505, Institut des Cordeliers, Paris, France) for stably transfected Caco-2/TC7 lines. The confocal microscopy analysis was performed thanks to the cellular imaging platform IFR58 (Institut des Cordeliers, Paris, France). We thank Rachel Carol (Ecole Pratique des Hautes Etudes, Paris, France) for editing and rewriting the manuscript. |
FOOTNOTES |
|---|
First Published Online May 31, 2005
Abbreviations: Ad-GFP, Adenovirus expressing GFP; apo, Apolipoprotein; CAT, chloramphenicol acetyltransferase; CMV, cytomegalovirus; FABP, fatty acid binding protein; GFP, green fluorescent protein; GVPN, Gene Vector Production Network; HDAC, histone deacetylase; HNF, hepatic nuclear factor; HRE, hormone-responsive element; LXR, liver X receptor; RT, reverse transcription; SREBP, sterol response element-binding protein.
Received for publication November 17, 2004. Accepted for publication May 18, 2005.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
and isoforms. J Biol Chem 277:34540–34548. Biochim Biophys Acta 1490:21–32[Medline]
ligand binding domain in complex with endogenous fatty acid ligand. J Biol Chem 277:37973–37976 activation by ligand and coactivator binding. J Biol Chem 279:23311–23316 (HNF-4). J Biol Chem 277:23988–23999
