Bile Acid Reduces the Secretion of Very Low Density Lipoprotein by Repressing Microsomal Triglyceride Transfer Protein Gene Expression Mediated by Hepatocyte Nuclear Factor-4 *

Microsomal triglyceride transfer protein (MTP) is involved in the transfer of triglycerides, cholesterol esters, and phospholipids to newly synthesized apolipoprotein (apo) B. It is therefore essential for lipoprotein synthesis and secretion in the liver and the small intestine. Although several recent experiments have revealed the transcriptional regulation of the MTP gene, little has been revealed to date about hepatocyte nuclear factor-4 (HNF-4)-dependent regulation. We here report that the human MTP gene promoter contains a pair of functional responsive elements for HNF-4 and HNF-1, the latter of which is another target gene of HNF-4. Chromatin immunoprecipitation assays provide evidence that endogenous HNF-4 and HNF-1 can bind these elements in chromatin. In Hep G2 cells overexpression of either a dominant negative form of HNF-4 or small interfering RNAs (siRNAs) against HNF-4 dramatically reduces the activities of both the wild type and the HNF-4 site mutant MTP promoter. This suggests that HNF-4 regulates MTP gene expression either directly or indirectly through elevated HNF-1 levels. When Hep G2 cells were cultured with chenodeoxycholic acid (CDCA), a ligand for the farnesoid X receptor (FXR), mRNA levels for MTP and apo B were reduced because of increased expression of the factor small heterodimer partner (SHP), which factor suppresses HNF-4 activities. Chenodeoxycholic acid, but not a synthetic FXR ligand, attenuated expression of HNF-4, bringing about a further suppression of MTP gene expression. Over time the intracellular MTP protein levels and apo B secretion in the culture medium significantly declined. These results indicate that two nuclear receptors, HNF-4 and FXR, are closely involved in MTP gene expression, and the results provide evidence for a novel interaction between bile acids and lipoprotein metabolism.

MTP, 1 expressed specifically in the liver and the small intestine, plays a critical role in the assembly and secretion of very low density lipoproteins (VLDLs) and chylomicrons. MTP exists in the lumen of the endoplasmic reticulum as a heterodimer with protein-disulfide isomerase and is involved in the transfer of triglycerides, cholesterol esters, and phospholipids to newly synthesized apo B (1,2). If the apo B protein is not properly folded or if the enrichment of lipids is insufficient, the apo B protein is degraded by a ubiquitin-dependent proteasome process instead of proceeding to the formation of lipoprotein particles (3)(4)(5). In human patients with abetalipoproteinemia the absence of functional MTP results in a defect in the assembly and secretion of plasma lipoproteins containing apo B (6). A vital role of MTP in the formation and secretion of apo B is further supported by the fact that conditional gene knock-out mice specifically lacking hepatic MTP are unable to form VLDLs in the liver (7). Moreover, specific inhibitors of MTP lipid transfer have been developed and have lowered plasma cholesterol levels successfully (8). These findings clearly indicate that changes in MTP activities under various physiological conditions can modulate lipoprotein production and secretion in the liver and intestine.
HNF-4 is a highly conserved member of the nuclear receptor superfamily. It is a liver-enriched transcription factor that, together with other factors, plays a key role in the tissuespecific expression of a large number of genes involved in lipid and glucose metabolism. The active form of HNF-4 is a homodimer, and it does not appear to heterodimerize with other members of the nuclear receptor family. Recent investigations have shown that coenzyme A derivatives of certain fatty acids activate the receptor, and these derivatives thus have been characterized as endogenous ligands for HNF-4 (9,10). A crucial role for HNF-4 in metabolic homeostasis was demonstrated by the finding that mutations in the HNF-4 gene cause the disorder known as maturity onset diabetes of the young (11). Conditional HNF-4 gene knock-out mice, which were produced using the Cre-loxP method with an albumin-Cre transgene, exhibit a great reduction in serum cholesterol and triglycerides because of the decreased levels of MTP and several apolipoproteins (12). Although this result indicates that MTP gene expression is under the control of HNF-4, little is known about the specific step.
It has been shown that the transcriptional activity of HNF-4 is regulated by interaction with small heterodimer partner (SHP), an atypical negative nuclear receptor lacking a DNAbinding domain (13,14). SHP, induced by FXR together with bile acids, controls the transcriptional activity of several other nuclear receptors including the constitutive androstane receptor, thyroid receptor, retinoid X receptor (RXR), retinoic acid receptor, estrogen receptors, peroxisome proliferator-activated receptors, the liver X receptor, and the liver receptor homolog-1 (14 -19). Moreover, recent findings have provided evidence that bile acids activate a MAPK pathway (20,21) and reduce the transactivation potential of HNF-4 (22). These findings prompted us to examine the effect of bile acids, which activate FXR and eventually induce SHP gene expression, on the HNF-4-dependent transcription of the MTP gene.
We here show that MTP gene expression is regulated by HNF-4 and HNF-1, which bind to the individual responsive element in the promoter of the human MTP gene. We also demonstrate that bile acids can down-regulate MTP transcription by impairing the transactivation potential of HNF-4 through the interaction with SHP and suppressing HNF-4 gene expression. In response to attenuated HNF-4 activity the transcription of other HNF-4-responsive genes including HNF-1 and apo B is also reduced, leading to decreased VLDL secretion. Taken together, this evidence suggests that bile acids are able to control lipoprotein synthesis and secretion via the FXRand HNF-4-mediated pathways.
Cell Culture-Hep G2 and HEK293 cells were cultured with a medium containing 10% fetal bovine serum (FBS) in collagen-coated dishes as described previously (23,24).
Construction of the Reporter Genes for Luciferase Assays-The reporter plasmid containing human MTP promoter (Ϫ204 to ϩ33), pMTP-204, was described previously (25). All of the mutant reporter constructs (pMTP-AKO, -BKO, and -CKO and pMTP⌬HNF1) were synthesized by a PCR-assisted method using a site-directed mutagenesis kit (Stratagene) following the manufacturer's instructions. All of the mutant HNF-4 and HNF-1 sites were replaced by the sequence shown by the italicized letters in Fig. 1. To generate pHNF4-1000, an 1-kb fragment containing the 5Ј-flanking region of the human gene (Ϫ1014 to ϩ55) was inserted into a pGL3 basic vector (Promega). The reporter plasmid containing human intestinal bile acid-binding protein (I-BABP) promoter (Ϫ862 to ϩ30), pI-BABP, was described previously (20).
Reporter Assays-Reporter assays were performed as described previously (20,26,27). HEK293 cells were transfected with 0.2 g of the indicated reporter construct, 0.01 g of pRL-CMV, an expression plasmid encoding Renilla luciferase (Promega), and 0.6 g of either pHNF4 or pHNF1. Hep G2 cells were transfected with 2 g of the indicated reporter construct and 0.1 g of pRL-CMV. In reporter assays in Fig. 8, Hep G2 cells were cultured with a medium containing 10% charcoalstripped FBS (20). After incubation for 48 h, the Dual-Luciferase TM reporter system (Promega) was used to determine luciferase activity. When Hep G2 cells were transfected with siRNA expression constructs, cells were harvested after 72 h of incubation.
Northern Blot Analysis-Hep G2 cells were set up on day 0 in medium A (Dulbecco's modified Eagle's medium, 100 units/ml penicillin, and 100 g/ml streptomycin) supplemented with 10% charcoal-stripped fetal calf serum. On day 1, the medium was removed, and the cells were then washed with phosphate-buffered saline and refed with medium A containing 5% LPDS supplemented with the indicated concentration of CDCA. After a 6-, 12-, 18-, or 24-h culture, total RNA was extracted, and Northern blotting was performed as described previously (20,27). A 670-bp fragment from human MTP, a 640-bp fragment from human apo B, a 980-bp fragment from human HNF-4␣, a 770-bp fragment from human SHP, a 1900-bp fragment of human HNF-1␣, and a 700-bp fragment from 36B4 were used as templates for 32 P-labeled probes.
Western Blot Analysis-Hep G2 cells were cultured as described above. On day 1, the cells were refed with medium A containing 5% LPDS supplemented with 100 or 200 M CDCA. On day 5, the cells were harvested, and Western blot analysis was carried out using a polyclonal antibody against human MTP (RS001, Ref. 25). To analyze apo B secretion, on day 4, the cells were refed with the same medium, and then the culture medium was collected on day 5. Western blot analysis was performed using a polyclonal antibody against human apo B (Chemicon International Inc.) as described previously (20,27).

HNF-4-dependent Regulation of the Human MTP Promoter-
As it has been reported that all of the putative positive and negative response elements for the liver-specific MTP gene expression are localized within the human MTP promoter Ϫ142 bp region (30), we focused on the promoter activity of the first 200 bp 5Ј to the transcription start site to identify the cis acting element for HNF-4. We found three putative HNF-4responsive elements in this region ( Fig. 1) and performed luciferase assays using wild type or mutant versions of reporter genes to confirm certain functional element(s) among them. When Hep G2 cells, which endogenously express HNF-4, were transfected with one of the reporter genes, the luciferase activities were significantly decreased only by disruption of the B site ( Fig. 2A). When HEK293 cells were transfected, the pro- moter activity was undetectable unless an HNF-4 expression plasmid was introduced, suggesting that HNF-4 is responsible for cell type-specific expression of the MTP gene (Fig. 2B). Only mutation at the B site elicited a tremendous suppression of luciferase activities in the presence of HNF-4. These results indicate that the B site is important for the HNF-4-dependent induction of the MTP promoter activity.
HNF-4 Can Bind to the B Site in the Human MTP Promoter-Next, gel mobility shift assays were performed to demonstrate direct binding of HNF-4 to the B site. Fig. 3 shows that a nucleotide probe that carried the binding sequence for HNF-4 (28) was shifted after the addition of nuclear extracts of HEK293 cells transfected with an expression plasmid for FLAG-tagged HNF-4 and was supershifted by the addition of anti-FLAG antibodies (lanes 1-3). Excess amounts of unlabeled A site or C site probes (wild type and mutant forms) did not affect the formation of DNA-HNF-4 complex (Fig. 3, lanes 4 -6 and lanes 10 -12). However, the addition of excess amounts of an unlabeled B site probe (wild type), but not the mutant variant, inhibited the formation of the complex (Fig. 3, lanes 7-9), suggesting that the Ϫ49 to Ϫ61 region is a binding site for HNF-4. These results are in good accord with the results in Fig. 2.
HNF-1-dependent Regulation of the Human MTP Promoter-Because there exists a putative HNF-1-responsive element in the 200-bp promoter region (Fig. 1), we examined whether this site is functional. Reporter assays showed that the luciferase activities in Hep G2 cells transfected with a mutant version of reporter gene lacking the HNF-1-responsive element were much lower than those with a wild type reporter gene (Fig. 4A).
To determine whether HNF-1 can bind to the putative HNF-1-responsive element, gel shift assays were performed with a nucleotide probe covering this element. A specific protein-DNA complex, which was not detected with the nuclear extracts from mock-transfected cells (Fig. 4B, lane 1), was detected in the presence of the nuclear extracts from FLAG-HNF-1-expressing cells (lane 2). This band was migrated to the same position as a control complex with a probe containing a consensus HNF-1 binding sequence (Fig. 4B, lane 6). The addition of excess amounts of unlabeled wild type probe (Fig. 4B, lanes 3 and 4) inhibited the formation of the complex, whereas the addition of unlabeled mutant probe did not affect it (lane 5). These results clearly indicate that the Ϫ98 to Ϫ111 region is a binding site for HNF-1 and crucial for the MTP promoter activity.
To investigate whether both endogenous HNF-1 and -4 bind to the MTP promoter, we performed chromatin immunoprecipitation. As shown in Fig. 4C, endogenous HNF proteins bound to the promoter of the MTP gene (second and third lanes). An unrelated anti-Gal4 DNA-binding domain antibody did not generate any PCR products (Fig. 4C, fourth lane). These data indicate that MTP gene expression is driven by the binding of endogenous HNF-1 and -4 to the individual responsive element in chromatin. Endogenous HNF-4 Plays a Crucial Role for MTP Gene Expression-Although Figs. 2 and 3 clearly show that the MTP promoter activity requires both the HNF-4 and HNF-1 functions, we do not know how these factors coordinately regulate MTP gene expression. HNF-1 has been shown to be a direct transcriptional target of HNF-4 in liver (31). Therefore, one reasonable hypothesis is that HNF-4 stimulates MTP gene expression indirectly by increasing HNF-1 levels, which in turn activates the MTP promoter. To confirm the role of HNF-4, we constructed an expression plasmid for DN-HNF-4 that possesses a functional domain but lacks a DNA-binding domain and therefore suppresses the activity of endogenous HNF-4. As shown in gel mobility shift assays (Fig. 5A), a complex of HNF-4 and a probe containing the B site in the MTP promoter (lane 1) was replaced by the addition of excess amounts of DN-HNF-4 (lane 4), suggesting that this dominant negative form suppresses the activity of HNF-4. To distinguish a direct action of HNF-4 from an indirect action mediated through HNF-1, we compared the activity of wild type and mutant promoters in the presence or absence of DN-HNF-4 in Hep G2 cells (Fig. 5B). The activities of wild type and ⌬HNF1 promoters significantly declined by expression of DN-HNF-4 through the intact B site. Furthermore, expression of DN-HNF-4 led to reduction in the activity of the promoter with the mutation in the HNF-4 site, confirming that HNF-4 also stimulates MTP gene expression through an indirect effect via HNF-1. Alternatively, endogenous HNF-4 functions were repressed by specific siRNAs, which had already been shown to be effective in our previous report (24). It turns out that quite similar results were obtained by both methods, i.e. weakening of endogenous HNF-4 activities (Fig. 5, B and C). Although in these assay systems it is difficult to compare quantitatively the direct with the indirect action of HNF-4, the significant effects of DN-HNF-4 and siRNAs clearly indicate that both actions are substantial.
CDCA Affects MTP Gene Expression-It has been shown that CDCA induces SHP gene expression through activation of FXR and that SHP in turn inactivates HNF-4 functions (14). Taking into account the fact that apo B is also one of the target genes for HNF-4, it might be that bile acids reduce MTP-mediated secretion of apo B-containing lipoproteins from hepatocytes. To investigate this possibility, Hep G2 cells were cultured with a medium containing CDCA for 24 h, and Northern blot analyses were carried out. Because of a long cascade from FXR to MTP through SHP and HNF-4, we analyzed changes in mRNA levels for several genes up to 24 h so as not to overlook any effects of CDCA (Fig. 6). In addition, because we had demonstrated previously that CDCA activates the MAPK pathway (20), the cells were cultured with the indicated CDCA concentration to stimulate this pathway. As shown in Fig. 6, after a 6-h incubation, only mRNA levels for SHP, a target gene of FXR, were upregulated, whereas the others were unaffected. MTP mRNA levels were reduced after 12 h and longer incubation with 100 M CDCA, and the levels were reduced more robustly with 200 M CDCA. Similar patterns were observed in terms of apo B, HNF-1, and HNF-4, which are all direct transcriptional targets of HNF-4. These results imply that the FXR-mediated activation of SHP might lead to suppression of HNF-4 target gene expression. It is likely that SHP mRNA levels were reduced after 24 h because of a self-regulatory mechanism (32).
SHP Suppresses Transcription of the MTP Gene-To investigate the direct effect of SHP on MTP promoter activity, luciferase assays were performed. When Hep G2 cells were transfected with an SHP expression plasmid, the MTP promoter activity was reduced in a dose-dependent manner (Fig. 7A). It appears that overexpressed SHP directly inhibits HNF-4 activities. To confirm an SHP-mediated reduction of HNF-4 activities, the transcriptional activity of a Gal4-HNF-4 fusion protein was examined in the presence of SHP (Fig. 7B). The overexpression of SHP resulted in inhibition of the activity of the fusion protein. This inhibitory pattern is in good accord with the pattern in Fig. 7A, suggesting that the inhibitory effect observed in Fig. 7A is mainly caused by the repression of HNF-4 activities mediated by SHP. Next, we tested whether SHP might directly suppress HNF-4 gene expression. The HNF-4 promoter containing ϳ1.0 kb of the 5Ј-flanking region of the human HNF-4 gene did not respond to SHP (Fig. 7C). A previous investigation demonstrated that this 1.0-kb upstream region exhibits full promoter activity and contains functional biding sites for several transcription factors including HNF-1, HNF-4, HNF-6, GATA-6, and Sp1 (33). These results indicate that CDCA-mediated suppression of MTP gene expression is partly attributable to the functions of SHP. Moreover, it appears that HNF-4 gene expression might be regulated by an alternative mechanism rather than the FXR-SHP pathway.
To address the underlying mechanisms by which CDCA affects HNF-4 gene expression, the effects of CDCA were compared with those of a synthetic FXR ligand, GW4064. When Hep G2 cells were treated with either CDCA or GW4064 for 18 h, MTP mRNA levels were reduced more powerfully by the addition of CDCA than by the addition of GW4064 (Fig. 8A). SHP mRNA levels, which appeared to be declining already from the peak levels observed at ϳ6 h (Fig. 6), were slightly increased by both compounds almost equally. HNF-4 gene expression was not affected by the addition of a synthetic ligand but was suppressed by CDCA.
To investigate further the difference between CDCA and GW4064, we carried out reporter assays with these FXR ligands. The MTP promoter activity was reduced by CDCA in a dose-dependent manner (Fig. 8B). Although GW4064 also repressed the MTP promoter activity, its effect was weaker than the CDCA effect (a 20% decrease at 10 M GW4064 versus a 50% decrease at 200 M CDCA). This result is consistent with the result of MTP mRNA levels in Fig. 8A. On the other hand, the human I-BABP promoter, which contains a functional FXRresponsive element (20), was almost equally regulated by both ligands (Fig. 8C). These results further support the notion that CDCA exerts its effect on transcription of the HNF-4 gene by an FXR-independent mechanism.
CDCA Treatment Reduces the MTP Protein Levels and Apolipoprotein B Secretion-To further confirm that CDCA-mediated suppression of MTP gene expression leads to a decrease in the intracellular MTP protein levels and in turn a reduction of apo B secretion, Western blot analyses were performed using Hep G2 cells treated with CDCA and their culture medium. The intracellular MTP protein levels indeed did decline with the addition of CDCA in a dose-dependent manner (Fig. 9), consistent with the results of a decline in their mRNA levels (Fig. 6). Concomitantly, the apo B protein levels in the culture medium were also reduced by CDCA treatment. Taken together, these results make it likely that the suppression of apo B secretion is attributable to the combined effects of the decreased MTP activity and the reduced apo B synthesis. DISCUSSION In the current study we demonstrate that the human MTP gene promoter contains a pair of functional HNF-4 and HNF-1 binding sites. The finding that disruption of one of these sites led to a dramatic suppression of the activity of the promoter in Hep G2 cells ( Figs. 2A and 4A) implies the importance of these transcription factors in the tissue-specific MTP gene expression. This notion is further supported by the observation that overexpression of HNF-4 in HEK293 cells remarkably stimulated promoter activity (Fig. 2B). The actions of these two nuclear factors are not independent but are modulated by each other as schematized in Fig. 10. The data on the use of DN-HNF-4 or siRNAs against HNF-4 demonstrate that HNF-4 is able to control the MTP promoter activity directly as well as indirectly, the latter through HNF-1, its direct target gene. Such a relationship between HNF-4 and HNF-1 has also been observed in activation of the insulin promoter (34). Although it is not possible as of yet to estimate quantitatively the relative contribution of the HNF-4 effects on activation of the MTP promoter as compared with the HNF-1 effects, HNF-4, which is considered to be at the top of the hierarchy of the transcription factor cascade driving hepatocyte differentiation, is likely to be a major and essential regulator of the MTP gene. Indeed, the MTP mRNA levels were undetectably low in the HNF-4-null mouse liver (12), whereas HNF-1␣-deficient mice exhibited normal hepatic levels for HNF-4 and MTP (35).
SHP is an unusual orphan nuclear receptor that contains a putative ligand-binding domain but lacks a conserved DNAbinding domain (13). It has been reported that SHP can negatively regulate the transcriptional activity of a number of nuclear receptors. Because SHP is one of the FXR target genes, bile acids can exert a modulation of lipid metabolism through the actions of SHP. Therefore, we hypothesized that bile acids might be able to regulate expression of HNF-4 target genes, MTP and apo B, by modulating HNF-4 functions and in turn lead to reduction of VLDL secretion from hepatocytes. The current data clearly demonstrate that CDCA treatment stimulates SHP gene expression, which suppresses the MTP promoter activity, decreases MTP and apo B mRNA levels, and eventually elicits reduced apo B secretion from Hep G2 cells (Fig. 10). While we prepared this paper, several investigators reported that bile acid can regulate the transcription of various genes involved in blood pressure, gluconeogenesis, and triglyceride metabolism through the SHP-dependent pathway (36 -38). In the current study we provided evidence that an alternative pathway as well as the SHP-dependent process might participate in the regulation of MTP gene expression by bile acid. A synthetic FXR ligand as well as CDCA was able to increase SHP mRNA levels, but its inhibitory effect on MTP gene expression was weaker than that of CDCA (Fig. 8, A and B). This suggests that the FXR-independent activity must be responsible for the CDCA effects. Two previous reports on SHP-null mice demonstrated that the repression of cholesterol 7␣-hydroxylase gene expression, which is partly caused by SHP, is retained in SHP-null mice fed bile acids, demonstrating the existence of compensatory pathways of bile acid signaling FIG. 7. Effects of SHP on the luciferase activities of the MTP promotercontaining reporter gene and HNF-4␣ promoter-containing reporter gene. A, Hep G2 cells were cotransfected with pMTP-204 together with pRL-CMV and pSHP, an expression plasmid for human SHP. B, HEK293 cells were cotransfected with pG5Luc containing five copies of Gal4 binding sites (27) together with pRL-CMV, pGal4-HNF4, and pSHP. C, Hep G2 cells were cotransfected with pHNF4-1000 together with pRL-CMV and pSHP. The cells were incubated with a medium containing 10% FBS for 48 h, and then luciferase assays were performed as described under "Experimental Procedures." The luciferase activities in the absence of SHP and Gal4-HNF-4 are considered as "1." The values given are the averages of data from three experiments. Data are expressed as means Ϯ S.D.  (21,39). We had shown previously that CDCA treatment activated the extracellular signal-regulated kinase (ERK) 1/2 in Hep G2 cells and stimulated the low density lipoprotein receptor gene expression in an FXR-independent manner (20). It also was reported that the c-Jun N-terminal kinase pathway was activated by bile acid (21,22). Although we treated Hep G2 cells with several MAPK inhibitors to prove the importance of the MAPK pathway, SHP gene expression was unexpectedly affected by these inhibitors (data not shown). Thus, it was hard to distinguish the effects of the MAPK pathway from the SHP action in the presence of these inhibitors. At present we do not know which pathway participates in an SHP-independent effect of bile acid. It is quite important to elucidate the complex network between bile acid and the SHP-independent pathways.
It also has been reported that transcription of apo CIII, another HNF-4 target gene, is regulated by MAPK (40). In that report it was demonstrated that an ERK 1/2 inhibitor stimulated HNF-4 promoter activity, resulting in an increase in apo CIII gene expression. This finding supports the notion that the ERK 1/2 pathway activated by bile acid might be involved in the regulation of HNF-4 gene expression observed in the current study (Figs. 6 and 8). Alternatively, it is possible that MAPK phosphorylates HNF-4, a phosphoprotein (41), and reduces its transcriptional activity. Nuclear receptors exhibit a modular structure with six distinct regions (A-F), which correspond to functional domains. Interestingly, the HNF-4 F region is adjacent to ligand-binding domain E, a designated negative regulatory domain (42) in which several potential phosphorylation sites are located. It will be of considerable interest to clarify the physiological significance of the inhibitory function of this domain on the HNF-4 activity in response to activation of the MAPK pathway.
Several reports have indicated that MTP promoter activity is controlled by insulin, sterol-regulatory element-binding proteins, and chicken ovalbumin upstream promoter transcription factor II (COUP-TFII) (25,30,43). It has been shown that COUP-TFII can bind to the direct repeat 1 (DR1) sequence (Ϫ39 to Ϫ50 in Fig. 1) and suppress transcription by competing with RXR␣ (43). Although this DR1 overlaps with the HNF-4 binding site B by two nucleotides, the current observation that the mutant reporter gene pMTP-BKO, which contains the intact DR1 but lacks the HNF-4 binding motif, did not respond to HNF-4 ( Fig. 2) suggests no involvement of this DR1 site in the HNF-4-dependent regulation of the MTP gene expression. Further studies are needed to elucidate the precise mechanisms for the network among these several transcription factors responsible for the regulation of MTP transcription.
In conclusion, our results reveal a novel pathway of regulation of VLDL secretion by bile acid that involves three nuclear receptors, FXR, HNF-4, and SHP. Bile acid impairs transcription of both the MTP and apo B genes, which are regulated mainly by HNF-4, by the inhibitory effect of SHP on the transactivation potential of HNF-4 and an attenuated expression of HNF-4. This study provides evidence for molecular mechanisms regulating the complicated links between bile acid and cholesterol metabolism. Open arrows depict the increase or decrease in the mRNA or protein levels. The activation or inactivation of proteins is indicated by plus or minus signs, respectively. Because the effect of the MAPK pathway on HNF-4 gene expression remains unclear in the current report, the filled arrow from MAPK to HNF-4␣ is marked by a question mark.