Elevated cholesterol metabolism and bile acid synthesis in mice lacking membrane tyrosine kinase receptor FGFR4.

Heparan sulfate-regulated transmembrane tyrosine kinase receptor FGFR4 is the major FGFR isotype in mature hepatocytes. Fibroblast growth factor has been implicated in the definition of liver from foregut endoderm where FGFR4 is expressed and stimulation of hepatocyte DNA synthesis in vitro. Here we show that livers of mice lacking FGFR4 exhibited normal morphology and regenerated normally in response to partial hepatectomy. However, the FGFR4 (-/-) mice exhibited depleted gallbladders, an elevated bile acid pool and elevated excretion of bile acids. Cholesterol- and bile acid-controlled liver cholesterol 7alpha-hydroxylase, the limiting enzyme for bile acid synthesis, was elevated, unresponsive to dietary cholesterol, but repressed normally by dietary cholate. Expression pattern and cholate-dependent, cholesterol-induced hepatomegaly in the FGFR4 (-/-) mice suggested that activation of receptor interacting protein 140, a co-repressor of feed-forward activator liver X receptor alpha, may mediate the negative regulation of cholesterol- and bile acid-controlled liver cholesterol 7alpha-hydroxylase transcription by FGFR4 and cholate. The results demonstrate that transmembrane sensors interface with metabolite-controlled transcription networks and suggest that pericellular matrix-controlled liver FGFR4 in particular may ensure adequate cholesterol for cell structures and signal transduction.

The fibroblast growth factor (FGF) 1 tyrosine kinase signaling complex is an intrinsic mediator of cell to cell communication in tissue remodeling in development and cellular homeostasis in adult organs (1,2). The FGF receptor kinase family consists of an extensive repertoire of alternately spliced products of four genes (fgfr1 to fgfr4), which are expressed in development in a temporal-and spatially specific mode (3) and in adult tissues in a cell type-specific mode (2). Disruption of the fgfr1 and fgfr2 genes in mice disrupt development, exhibit global proliferation defects, and are embryonic lethal (4,5). Mice in which fgfr3 was disrupted are viable, but exhibit severe skeletal dysplasias due to overgrowth of long bones, which is a consequence of loss of restraints on growth of chondrocytes during endochondral ossification (6,7).
FGF-1, FGF-2, and FGF-8 have been implicated in definition of the liver from foregut endoderm where FGFR4 is expressed (8). However, disruption of fgfr4 in the mouse germline resulted in no overt abnormalities (9). All four fgfr genes are expressed in adult liver (10), but only FGFR4 is expressed in mature hepatocytes (11). External administration of FGF-7 and FGF-18, or expression of FGF-18 under control of liver specific-promoters, elicits hyperplasia in the liver (12,13). FGF-1 and FGF-7 elicit DNA synthesis in primary liver cell cultures enriched in hepatocytes (14,15).
To determine whether FGFR4 plays a role in liver in vivo, we examined the morphology of the liver and associated organs in fgfr4 Ϫ /fgfr4 Ϫ mice, including the compensatory growth response after partial hepatectomy. Here we report no differences in liver architecture and kinetics or extent of restoration of liver mass between wild-type and fgfr4 Ϫ /fgfr4 Ϫ animals. However, examination of liver-associated organs revealed a small, depleted gallbladder in the FGFR4-deficient mice that prompted an analysis of cholesterol and bile acid metabolism. The results revealed that the FGFR4 deficiency caused a significant elevation of the excreted and total bile acid pools. Elevation of bile acid pools were coincident with constitutively elevated expression of Cyp7a, the limiting enzyme in the classical pathway for bile acid synthesis (16), and 3-hydroxy-3methylglutaryl-CoA (HMG-CoA) reductase, the rate-limiting step in cholesterol synthesis (17). The FGFR4 knockout mice exhibited a cholate-dependent, cholesterol-induced hepatomegaly. Analysis of gene expression in the hepatomegalic livers of the mutant mice suggested points where both FGFR4 and bile acids exert negative controls on liver bile acid synthesis. These results implicate the pericellular matrix-controlled FGFR4 kinase complex in hepatocytes in control of cholesterol metabolism and bile acid synthesis in a physiological setting.

EXPERIMENTAL PROCEDURES
Animals and Diets-Disruption of the mouse fgfr4 locus was carried out in 129 Sv strain-derived ES cells as described (9). Wild-type 129 Sv mice were obtained from the Jackson Laboratory. FGFR4 (ϩ/Ϫ) mice were generated by crossing FGFR4 (Ϫ/Ϫ) mice with wild-type 129 Sv mice, or by further crossing FGFR4 (ϩ/Ϫ) with FGFR4 (Ϫ/Ϫ) mice. Only * This work was supported by Public Health Service Grants DK35310 and DK47039 from the NIDDK, National Institutes of Health and Grant CA59971 from the NCI, National Institutes of Health. 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.
ʈ To whom correspondence should be addressed: Inst. of Biosciences and Technology, 2121 W. Holcombe Blvd., Houston, TX 77030-3303. Tel.: 713-677-7522; Fax: 713-677-7512; E-mail: wmckeeha@ibt. tamu.edu. 1 The abbreviations used are: FGF, fibroblast growth factor; FGFR, FGF receptor; Cyp7a, cholesterol 7␣-hydroxylase; RIP140, receptor interacting protein 140; LXR, liver X receptor; HMG-CoA, 3-hydroxy-3methylglutaryl-CoA; PH, partial hepatectomy; RT-PCR, reverse transcriptase-polymerase chain reaction; ISBT, ileal sodium-dependent bile acid transporter; IBABP, intestinal bile acid-binding protein; RPA, ribonuclease protection; BrdUrd, bromodeoxyuridine; Cyp7b, oxysterol 7␣-hydroxylase; RXR␣, retinoid X receptor; Cyp27a, sterol 27-hydroxylase; FXR, farnesoid X receptor; nt, nucleotide(s); PBS, phosphatebuffered saline. mice 7-8 weeks old were used in the study. Male mice were used for partial hepatectomy (PH)-induced liver regeneration experiments, and female mice were used in all other protocols. Mice were maintained in 12-h light/12-h dark cycles and were given free access to food and water. Standard rodent chow containing 0.02% (w/w) cholesterol and the standard chow supplemented with 2% (w/w) cholesterol or both 2% (w/w) cholesterol and 2% (w/w) sodium cholate was obtained from Alief Purina Feed Store, Inc. (Alief, TX). Two to four mice were employed for each experimental group, as described in the specific figure legends. After the mice were weighed, anesthetized, and exsanguinated, the livers or other tissues were harvested at 10:00 a.m., except that of the mice used in the PH-induced liver regeneration, which were harvested at the times indicated. All procedures were performed in accordance with the Institutional Animal Care and Use Committee at the Institute of Biosciences and Technology, Texas A&M University System Health Science Center.
Analysis of mRNA-Total RNA was isolated from livers or ileal tissue with the Ultraspec RNA Isolation System (BL-10200, Biotecx Laboratories) and specific mRNAs were measured by ribonuclease protection (RPA) using the HybSpeed RPA kit (no. 1412, Ambion). About 50 g of liver or ileal RNA was hybridized with 1 ϫ 10 5 cpm of 32 P-labeled specific antisense and ␤-actin riboprobes in the same reaction mixture. After treatment with ribonuclease, protected products were analyzed on 5% polyacrylamide sequencing gels, followed by autoradiography. Size of protection products was determined from the product of a DNA sequencing reaction parallel to the protection assays. The amount of each radiographic product was quantitated using a PhosphorImager (Molecular Dynamics). The value of bands between samples was standardized by division of the value of the internal ␤-actin in each sample. Experimental values were expressed in units relative to the level of expression in wild-type mice on standard chow, which was assigned a value of one.
Immunoblot Analysis-Livers were homogenized in PBS containing 0.5% sodium deoxycholate and 0.1% SDS and centrifuged. The protein concentration was determined using the BCA protein assay reagent (no. 23225X, Pierce). A total of 25 g of protein was subjected to 12% SDS-PAGE, transferred to Hybond-P membrane (Amersham Pharmacia Biotech), incubated with 1/10,000 rabbit anti-mouse CYP7A antiserum (a gift from Dr. David W. Russell), washed, and then incubated with 1/20,000 goat anti-rabbit IgG conjugated to horseradish peroxidase (Bio-Rad). Development was carried out using the Amersham ECL-Plus detection regents (Amersham Pharmacia Biotech).
Bile Acid Analysis-Bile acids were measured enzymatically using the Bile Acids kit (no. 450-A, Sigma). To determine fecal bile acid excretion, the feces from individually housed mice were collected, weighed, and dried over a 72-h period. Then 0.5 g of dried feces was minced and extracted in 10 ml of 75% ethanol at about 50°C for 2 h. The extract was centrifuged, and 1-ml samples of supernatant were diluted for assay to 4 ml with a 25% PBS solution. The bile acid concentration was measured enzymatically. The daily feces output (g/day per 100 g of body weight) and fecal bile acid content (mol/g) were used to calculate the rate of bile acid excretion (mol/day/100 g of body weight). The total bile acid pool size was determined as bile acid content of the small intestine, the gallbladder, the liver, and their contents. After the mice were weighed, anesthetized, and exsanguinated, the fresh organs were collected, minced together, and extracted in 15 ml of 75% ethanol at about 50°C for 2 h. The extract was centrifuged, 1-ml samples of supernatant for assay were diluted to 4 ml with 75% ethanol, and then 1-ml samples were diluted to 4 ml with 25% PBS. Bile acids were determined enzymatically, and the pool size was expressed as micromoles of bile acid/100 g of body weight.
Measurement of Hepatic Cholesterol-To measure hepatic cholesterol level, 50 mg of fresh liver was minced, hydrolyzed, and extracted in 4 ml of 1 M KOH/methanol (diluted aqueous KOH (10 M) with 9 volumes of methanol) at 66°C for 2 h. The extract was centrifuged, and the cholesterol concentration in the supernatant was measured using a Cholesterol kit (no. 139050, Roche Molecular Biochemicals). The cholesterol level was determined and expressed as milligrams of cholesterol/g of liver weight.
Histological Procedures-Liver tissues were fixed overnight in Histochoice Tissue Fixative MB (no. H120 -4L, Amresco), dehydrated through a series of ethanol treatments, and embedded in paraffin according to standard procedure. Sections were prepared and stained with hematoxylin and eosin.
Partial Hepatectomy and DNA Synthesis-A 70% hepatectomy, consisting of removal of the anterior and left lateral hepatic lobes, was performed on male mice at 10:00 a.m. Two hours prior to sacrifice of the animals for analysis, 50 g/g of body weight of bromodeoxyuridine (BrdUrd) was administered intraperitoneally. Remnant livers were removed and weighed at different times. BrdUrd incorporation in fixed liver sections was visualized with an anti-BrdUrd monoclonal antibody (no. 2531, Sigma) and an alkaline phosphatase-conjugated second antibody. Positive hepatocytes were counted, and BrdUrd incorporation was expressed as the percentage of the number of labeled hepatocytes in four or five visual fields.
Statistical Analyses-Values are expressed as the mean Ϯ standard deviation (S.D.) with the number of replicates described in the legends to figures. The statistical significance of differences between mean values (p Ͻ 0.05) was evaluated using the two-tailed Student's t test.

Normal Liver Architecture and Regeneration after Partial
Hepatectomy in FGFR4-deficient Mice-A histological examination of the liver revealed no apparent abnormalities in the overall morphology, cellular content and arrangement of different compartments in wild-type (ϩ/ϩ) and FGFR4-deficient (Ϫ/Ϫ) mice. The knockout animals exhibited normal blood chemistry including glucose, protein, and aspartate aminotransferase and alanine aminotransferase levels (data not shown). Partial hepatectomy was performed on both FGFR4 (Ϫ/Ϫ) mice and FGFR4 (ϩ/ϩ) mice, and both DNA synthesis and restoration of liver mass was measured periodically for up to 168 h after the operation. DNA synthesis peaked at the expected 38-h time point, and both the rate and extent of recovery of liver mass were identical in both wild-type and mutant mice (Fig. 1, A and B). Thus, FGFR4 is either not directly involved in compensatory growth of the liver in response to loss of 67% of the liver mass, or it is fully compensated for by other proliferative regulators.
Decrease in Weight/Volume of the Gallbladder Suggested Abnormal Bile Acid Metabolism in the FGFR4-deficient Mice-During surgical manipulation of livers for partial hepatectomy, we noted that the gallbladders of FGFR4 (Ϫ/Ϫ) mice, inclusive of contents, were smaller and weighed on average about 30% of those from FGFR4 (ϩ/ϩ) mice (Fig. 1C). The differential was maintained in mice fed a high cholesterol/cholate diet, although FGFR4 and Bile Acid Synthesis gallbladders exhibited an expected 4-fold increase in total weight (Fig. 1D). Histological analysis revealed no apparent difference in gallbladder morphology and structure between mutant and wild-type mice (data not shown).
Analysis of combined liver, gallbladder, and small intestine, as well as feces from both male and female FGFR4 (Ϫ/Ϫ), FGFR4 (ϩ/Ϫ), and FGFR4 (ϩ/ϩ) mice, revealed that bile acids were elevated 2-3-fold in FGFR4-deficient mice (Fig. 2, A and  B). Surgical ablation of the gallbladders from wild-type and FGFR4-deficient mice had no effect on the fecal bile acid excretion rate ( Fig. 2A). This suggested that an abnormality in bile acid metabolism and flow was the cause of the smaller gallbladders in mutant mice, rather than a defect in architecture and function of the gallbladders. It has been shown previously that acceleration of bile acid synthesis, by blocking bile acid feedback inhibition by blocking intestinal uptake, can accelerate gallbladder emptying and a decrease in gallbladder volume (18,19). Although the bile acid pool increased by an expected 60% in FGFR4 (ϩ/ϩ) mice on a high cholesterol diet (20), the diet had little effect on the already elevated pool observed in the FGFR4-deficient animals (Fig. 2C). Both newborn FGFR4 (ϩ/ϩ) and FGFR4 (Ϫ/Ϫ) animals exhibited a pool size of 20 mol/100 g of body weight, which rose to 30 mol/100 g on day 3 (Fig. 2D). By day 6, the pool in FGFR4 (Ϫ/Ϫ) mice was 2 times that of normal at 60 mol/100 g, and continued to increase through day 12, while pools were static in normal mice. During weaning, which causes an increase in the bile acid pools of both normal (21) and mutant mice, the pool in FGFR4 (Ϫ/Ϫ) mice rose to 3 times (250 mol/100 g of body weight) that of wildtype adult levels (about 80 mol/100 g of body weight) at 21 days. Levels in mutant mice dropped to about 160 mol/100 g after 1 month and remained static thereafter. This indicated that an elevation of bile acid pools in the FGFR-deficient mice occurs prior to maturation of mechanisms for reabsorption of bile acids in the ileum and liver (22,23) and the secondary acid pathway for synthesis of bile acids (24). Expression of mRNA for IBABP and ISBT, whose transcription is activated and repressed by bile acids, respectively (25,26), was increased and depressed, respectively, in mature FGFR4-deficient mice (Fig.  2, E and F). These observations suggested a role of FGFR4 in bile acid metabolism at the site of synthesis in the liver.
Elevation of Liver HMG-CoA Reductase and Cyp7a Expression in FGFR4-deficient Mice-Elevation of bile acids may result from accelerated conversion from cholesterol, or indirectly through increased availability of cholesterol substrate through its synthesis or deposition in the liver. Enzymes whose levels are regulated at transcription by substrates and products regulate both liver pathways (27). We first measured mRNA levels of HMG-CoA reductase, the rate-limiting enzyme in cholesterol biosynthesis in animals on standard rodent chow (Յ0.02% cholesterol w/w). HMG-CoA reductase mRNA, which is repressed by sterols and activated when they are deficient (28), was elevated by 7-fold in the FGFR4 (Ϫ/Ϫ) animals, but downregulated to near equal levels in both FGFR4 (ϩ/ϩ) and FGFR4 (Ϫ/Ϫ) animals fed a cholesterol-rich diet (Fig. 3A).
Cyp7a, which converts cholesterol into 7␣-hydroxycholesterol, is the rate-limiting enzyme in the classical route of bile acid synthesis (16). Cyp7a is positively regulated at transcription in a feed-forward mode by oxysterol metabolites of cholesterol (29), and negatively regulated in feedback mode by bile acids. The expression of liver Cyp7a mRNA was elevated by 2.5-fold in FGFR4 (Ϫ/Ϫ) mice compared with wild-type animals (Fig. 3B). Immunoblot analysis revealed that the elevation of the level of CYP7A protein was similar to the rise in mRNA (Fig. 3C). Although the high cholesterol diet increased Cyp7a expression by 1.5-fold in wild-type mice, the increase failed to reach the elevated level of Cyp7a expression observed in FGFR4 (Ϫ/Ϫ) mice (Fig. 3B). No increase over the elevated Cyp7a mRNA levels in the FGFR4-deficient animals was observed as a consequence of the high cholesterol chow. Cholate is is constitutively elevated in its absence.
Hepatomegaly in FGFR4-deficient Mice on High Cholesterol and Cholate-FGFR4 (ϩ/ϩ) and FGFR4 (Ϫ/Ϫ) mice on a high cholesterol diet (Fig. 4, A and B) exhibited no significant difference in liver size or liver cholesterol concentration. When challenged with a diet containing both 2% (w/w) cholesterol and cholate that strongly repressed Cyp7a (Fig. 4C), liver cholesterol concentration exhibited an expected increase of nearly 30 times in both wild-type and mutant mice. Unexpectedly, the liver weight in the FGFR4-deficient animals doubled within 2 weeks on the combined high cholesterol/cholate diet, and was 1.8 times larger than wild-type livers after 1 month (Figs. 4D and 5A). Administration of cholate alone had no effect (data not shown). The cholesterol/cholate-induced hepatomegaly was confirmed by the 32% fewer hepatocytes per visual field in sections of livers from FGFR4 (Ϫ/Ϫ) mice (Fig. 5B). A separate analysis of DNA synthesis by incorporation of BrdUrd (data not shown) confirmed that the hepatomegaly in FGFR4 (Ϫ/Ϫ) mice was due to hepatocyte hypertrophy rather than an increase in hepatocyte number.
Why is cholesterol-induced hepatomegaly dependent on dietary cholate, and why does it occur specifically in the FGFR4deficient mice, which appear to be more capable of disposal of cholesterol? When Cyp7a is deficient (31), the secondary acid pathway of bile acid synthesis compensates by generation and disposal of potentially hepatotoxic oxysterols from cholesterol (24,32). Moreover, the pathway is less stringently feedbackinhibited by bile acids than the classical pathway (33). We examined expression of cholesterol 27␣-hydroxylase (Cyp27a) and oxysterol 7␣-hydroxylase (Cyp7b), rate-limiting enzymes of the alternative acid pathway. Nuclease protection analysis revealed no difference between expression levels of Cyp27a or Cyp7b mRNA between FGFR4 (ϩ/ϩ) and FGFR4 (Ϫ/Ϫ) mice on standard chow. However, Cyp27a was repressed twice as effectively in mutants as in wild-type mice on the high cholesterol/ cholate combination (Fig. 6A). We also observed a decrease in Cyp27a in the mutant mice fed high cholesterol without cholate. No difference in Cyp7b expression was detected under any of the conditions (data not shown). The exaggerated depression of Cyp27a in absence of FGFR4 under conditions where Cyp7a is severely repressed may contribute to the selective hepatomegaly in the FGFR4-deficient mice.
The hepatomegalic phenotype in FGFR4 (Ϫ/Ϫ) mice induced by the high cholesterol/cholate combination was similar to that induced by cholesterol alone in mice devoid of the gene for the nuclear oxysterol receptor and transcription factor LXR␣ (20). We observed no change in the expression of LXR␣ or LXR␤ Arrow denotes return to standard chow. The data shown in C and D are from one representative experiment of three independent trials. mRNAs in all described conditions in wild-type and FGFRdeficient mice (data not shown). However, a screen for diet-dependent differences in expression of candidate co-activator/corepressors of LXR␣ in the wild-type and FGFR4 knockout mice revealed that expression of the multi-functional co-activator and co-repressor RIP140 was responsive to the dietary manipulation (Fig. 6B). Expression of RIP140 was depressed by 40% in wild-type mice fed high cholesterol, but elevated 1.8-fold in mice fed both cholesterol and cholate. RIP140 mRNA in FGFR4-deficient animals on standard chow was 30% of that in wild-type animals; the high cholesterol diet caused no further decrease, but surprisingly expression levels increased by over 10-fold in the mutant animals when cholate was added to the high cholesterol chow. RIP140 has been demonstrated to be a repressor of transcriptional activation by LXR␣/retinoid X receptor (RXR␣), as well as peroxisome proliferator-activated receptor ␣-RXR␣ heterodimers through interaction with LXR␣ or peroxisome proliferator-activated receptor ␣ (34). Our results suggest that, in addition to direct repression of Cyp7a expression through negative bile acid response elements at transcription, bile acid products also repress Cyp7a through activation of a co-repressor (RIP140) of the oxysterol receptor and feed-forward activator LXR␣. The exaggerated level of RIP140 in FGFR4-deficient animals induced by the presence of cholate may also contribute to the cholesterol-induced hepatomegaly observed in the mutant animals. Finally, the constitu-tively reduced levels of normally an FGFR4-activated co-repressor of LXR␣ activity provide a basis for the elevated Cyp7a expression and consequent elevated bile acid synthesis observed in the FGFR4 (Ϫ/Ϫ) mice on standard chow.

DISCUSSION
Transcriptional Networks Controlling Cholesterol Metabolism to Bile Acids in Mice-Mouse genetics has recently yielded major insight into the intricate regulation of disposal of cholesterol to bile acids. A knockout of Cyp7a, the rate-limiting enzyme of the classical bile acid pathway (Fig. 7A), caused reduction in bile acid synthesis, fat-soluble vitamin deficiency, and liver failure in 3 weeks in 90% of newborn mice. The other 10% of mice recovered to the normal phenotype in the period. The 90% could be rescued by temporary dietary supplementation with vitamins and cholate, which could be discontinued 3 weeks after birth (35). This confirmed the compensatory capability of the late onset secondary or alternate pathway of bile acid synthesis (Fig. 7A). The secondary pathway is rate-limited by sterol hydroxylases that convert cholesterol to oxysterols. Disruption of the gene for sterol 27-hydroxylase gene (Cyp27a) in mice resulted in decreased bile acid synthesis and excretion of fecal bile acids to 15-20% of normal coincident with elevation of transcription of Cyp7a by 9-fold, presumably due to release of bile acid-mediated feedback repression. HMG-CoA reductase was concurrently elevated by 2-3-fold, presumably due to reduced uptake of cholesterol and/or depletion by elevated Cyp7a (36).
The nuclear transcription factor, LXR␣, has been identified as an oxysterol receptor that activates the cyp7a gene in a feed-forward mode. Targeted disruption of the LXR␣ locus in mice had little effect on phenotype beyond a small decrease in bile acid pool size. However, when the knockout mice were fed high cholesterol, they failed to induce Cyp7a and cholesterol sufficiently accumulated in the liver to cause hepatomegaly (20). Still another orphan nuclear receptor, farnesoid X receptor (FXR), has been identified as a bile acid receptor. FXR regulates bile acid synthesis by repression of liver Cyp7a at transcription and also accelerates uptake in the ileum by activation of IBABP (37,38). Disruption of the FXR gene is predicted to result in elevated levels of excreted bile acids, depletion of bile acid pools, elevation of Cyp7a due to loss of feedback regulation and potentially concurrent induction of HMG-CoA reductase due to cholesterol depletion (39). These developments have revealed the exquisite metabolite-controlled transcriptional networks that balance the concentration of harmful, but essential, cholesterol and its metabolites in a physiological context (Fig. 7).
Negative Regulation of Bile Acid Synthesis by Liver Transmembrane Kinase FGFR4 -Here we implicate the transmembrane tyrosine kinase FGFR4 in control of cholesterol metabolism to bile acids by targeted gene disruption in mice. In contrast to the Cyp7a and Cyp27a knockout mice, both the excreted and total bile acid pools are elevated. This indicates that FGFR4 normally exerts a negative control on cholesterol metabolism to bile acids, which is abrogated by disruption of the fgfr4 gene. A defect in uptake and recycling underlying the elevated fecal bile acid levels in the FGFR4-deficient mice was unlikely, since 1) the elevation of bile acid pools was significant prior to developmental maturation of intestinal uptake and recycling mechanisms (Fig. 2D); and 2) induced malabsorption of ileal bile acids accompanied by elevation of fecal bile acid content, Cyp7a, and HMG-CoA reductase, causes a decrease in the total bile acid pool size (40,41). The normal sensitivity of Cyp7a expression in the FGFR4-deficient mice to dietary bile acids argued against an alteration in the bile acid-mediated feedback regulation in which the bile acid receptor FXR has FIG. 7. Model for regulation of bile acid biosynthesis by FGFR4. A, the biosynthesis of primary bile acids from cholesterol occurs by two pathways. The classical neutral pathway for bile acid synthesis is rate-limited by Cyp7a, and the acidic pathway is initiated and rate-limited by Cyp27a. Oxysterols, which are proportional to animal cholesterol levels, activate LXR␣, which activates transcription of Cyp7a. Bile acids activate FXR, which represses Cyp7a transcription. FGFR4 and FXR repress Cyp7a expression by different mechanisms (see "Discussion"). A magnified depression of Cyp27a expression in knockout mice due to dietary cholate may indicate a positive role of FGFR4 (Fig. 6A). B, multilevel repression of Cyp7a expression by FGFR4 and FXR through activation of co-repressor RIP140. Depression of RIP140 expression in mice on high cholesterol may indicate that LXR␣ also down-regulates RIP140 expression (Fig. 6B). recently been implicated. Although the anticipated phenotype of FXR knockout mice is similar to that which we describe here for FGFR4-deficient mice, the mechanism of negative control exerted by FGFR4 and FXR appears to be different, e.g. one does not mediate or compensate for the other.
On low cholesterol chow (Ͻ0.02%), de novo synthesis determined by the level of rate-limiting HMG-CoA reductase provides sufficient cholesterol to compensate for metabolism and the 5% of the bile acid pool that normally escapes uptake and recycling (42). In the FGFR4-deficient mice, we also observed an elevation of HMG-CoA reductase concurrent with the elevation of bile acid pools and Cyp7a. On the one hand, elevation of HMG-CoA reductase and cholesterol biosynthesis can elevate Cyp7a by feed-forward activation. On the other hand, elevation of Cyp7a and flux to bile acids can cause elevated HMG-CoA reductase through depletion of cholesterol substrate. If FGFR4 is normally a repressor of HMG-CoA reductase, then repression occurs independent of cholesterol-mediated regulation, since there was no defect in the dietary cholesterol-induced repression of HMG-CoA reductase in the FGFR4-deficient animals. Since Cyp7a remains elevated under the same dietary conditions that repressed HMG-CoA reductase, the elevation of Cyp7a expression as solely a consequence of increased cholesterol synthesis was ruled out. The elevation of HMG-CoA reductase observed in the FGFR4-deficient mice is more likely a consequence of cholesterol depletion due to accelerated bile acid synthesis.
Finally, a negative role of FGFR4 on the secondary acid pathway of bile acid synthesis was ruled out since the elevated bile acid pools in FGFR4 (Ϫ/Ϫ) mice was evident in neonatal mice at a time when Cyp7b and activity of the pathway is absent (24). Taken together, these findings demonstrate that FGFR4 exerts a negative control on cholesterol metabolism and bile acid synthesis in liver at the level of expression of Cyp7a that cannot be compensated for in its absence.
Targets for Negative Regulation by FGFR4 -At least eight transcriptional activators and their corresponding sequence response elements have been identified for Cyp7a (43). Positive transcriptional activators of Cyp7a expression are potential targets for negative modulation by the FGFR4 membrane kinase-signaling complex. Among these are the oxysterol receptor LXR␣ (29), which works in partnership with RXR, which has been demonstrated to be indispensable in the induction of Cyp7a and tolerance of a high cholesterol challenge in the diet (20). Although both FGFR4 knockout and normal mice tolerated high dietary cholesterol, the addition of 2% (w/w) sodium cholate to the high cholesterol chow reproduced the hepatomegalic phenotype described in LXR␣ (Ϫ/Ϫ) mice administered only the high cholesterol (20). This paradoxical phenotype in mice with elevated bile acid synthesis in absence of dietary cholate yielded clues to the mechanism of both FGFR4-and the bile acid-mediated repression of bile acid synthesis. The cholatedependent, cholesterol-induced hepatomegaly in the FGFR4 knockout mice was coincident with over a 10-fold increase in expression of RIP140, a repressor of LXR␣/RXR-mediated transcription (34). The cholate-induced increase was over a depressed level of RIP140 expression in the knockout mice relative to levels in normal mice on both the low and high cholesterol diets. This suggests that FGFR4 may repress bile acid synthesis through induced expression of the co-repressor RIP140. The marked elevation of RIP140 expression induced by dietary cholate also suggested a bile acid/FXR-mediated feedback regulation at the feed-forward stage of cholesterol disposal, in addition to direct repression of Cyp7a through cis-acting elements in the cyp7a gene (37).
Why does the expression of RIP140 overshoot wild-type lev-els in the presence of dietary cholate? This paradoxical effect can be explained by the phenomena of co-factor sharing (44) between bile acid-activated FXR and a putative FGFR4-activated transcription factor, both of which activate RIP140. In wild-type mice, bile acid-dependent FXR must compete with an FGFR4-activated transcription factor for a shared co-factor, which limits its maximum activation potential. The absence of FGFR4 allows maximum activation by FXR, which now has a monopoly on available co-factor. Overall, our results illustrate a novel and elegant multilevel network of regulation of Cyp7a transcription and bile acid synthesis that will be the subject of future study.
Integration of Transmembrane Signaling with Metabolitecontrolled Transcriptional Networks-Our results show that a transmembrane signaling complex, which mediates cell to cell communication and monitors changes in the tissue environment, is coupled to the metabolite-controlled transcriptional network that maintains cholesterol and bile acid homeostasis. It is conceivable that the FGFR complex directly senses cholesterol, bile acids, and intermediates through yet undefined cofactors. It is more likely that the pericellular matrix-controlled FGFR kinase complex in hepatocytes transmits changes in the tissue microenvironment that call for a rise in liver cholesterol and lipid metabolism in its anabolic roles for liver cells locally or for the organism (45). The liver response to acute infection triggered by endotoxins and cytokines causes transient cholesterol accumulation and hyperlipidemia coincident with depression of Cyp7a (45). Liver Cyp7a and serum 7␣-hydroxy-cholesterol levels are also depressed after partial hepatectomy and prior to the regenerative response (46). In recent years, cholesterol has been implicated in an increasing number of membrane signaling functions, including caveolar function (47), covalent modification of hedgehog signaling proteins (48), and assembly of integrin-G protein complexes (49). Inherited autosomal dominant mutations in fgfr genes other than fgfr4 result in constitutively active signaling complexes, which cause a variety of developmental abnormalities (50). Chronic deficiency or constitutive activity of FGFR4 may underlie defects in cholesterol and bile acid homeostasis. Our findings suggest the FGF-heparan sulfate-FGFR4 signaling complex as a target for prevention or therapy in maladies of cholesterol and bile acid metabolism. Both FGFR4 knockout mice and mice overexpressing FGFR4 in the liver provide new models for study of cholesterol and bile acid abnormalities.