Fxr(-/-) mice adapt to biliary obstruction by enhanced phase I detoxification and renal elimination of bile acids.

Farnesoid X receptor knockout (Fxr−/−) mice cannot upregulate the bile salt export pump in bile acid loading or cholestatic conditions. To investigate whether Fxr−/− mice differ in bile acid detoxification compared with wild-type mice, we performed a comprehensive analysis of bile acids extracted from liver, bile, serum, and urine of naive and common bile duct-ligated wild-type and Fxr−/− mice using electrospray and gas chromatography mass spectrometry. In addition, hepatic and renal gene expression levels of Cyp2b10 and Cyp3a11, and protein expression levels of putative renal bile acid-transporting proteins, were investigated. We found significantly enhanced hepatic bile acid hydroxylation in Fxr−/− mice, in particular hydroxylations of cholic acid in the 1β, 2β, 4β, 6α, 6β, 22, or 23 position and a significantly enhanced excretion of these metabolites in urine. The gene expression level of Cyp3a11 was increased in the liver of Fxr−/− mice, whereas the protein expression levels of multidrug resistance-related protein 4 (Mrp4) were increased in kidneys of both genotypes during common bile duct ligation. In conclusion, Fxr−/− mice detoxify accumulating bile acids in the liver by enhanced hydroxylation reactions probably catalyzed by Cyp3a11. The metabolites formed were excreted into urine, most likely with the participation of Mrp4.

The nuclear farnesoid X receptor (FXR) mediates bile acid effects on the expression of various genes involved in bile acid metabolism and transport (1)(2)(3). The central role of FXR in bile acid homeostasis has been established in mice with targeted disruption of Fxr (Fxr 2/2 mice) under conditions of bile acid loading [i.e., cholic acid (CA) or ursodeoxycholic acid (UDCA) feeding (4-6) and common bile duct ligation (CBDL) (7)]. When studying the adaptive response of hepatic ABC transport proteins, we found that CBDL induced the expression of multidrug resistance-related protein 3 (Mrp3) and Mrp4 in wild-type mice and even more in Fxr 2/2 mice, whereas Mrp2 expression remained unchanged (7). FXR-independent induction of hepatic Mrp2-Mrp4 as well as of renal Mrp2 and Mrp4 expression was also seen during CA and UDCA feeding (5,8). In contrast, a striking FXR dependence was seen in the regulation of the bile salt export pump (Bsep), because Fxr 2/2 mice failed to upregulate Bsep in any bile acid-loading condition (7). Nevertheless, alanine aminotransferase levels and mortality rates did not differ between wild-type and Fxr 2/2 mice in obstructive cholestasis (7). Rather, Fxr 2/2 mice had significantly lower levels of bile acids in the liver tissue and serum than did wild-type animals. This was even more surprising in light of the inability of Fxr 2/2 mice to decrease the level of hepatotoxic CA (4,5) in the liver via downregulation of Cyp7a1 and Cyp8b1, as shown previously (7).
This study thus aims to determine i) whether differences in the bile acid profiles may account for a more efficient elimination in Fxr 2/2 mice than in their wild-type relatives, ii) whether these compounds are formed by enhanced phase I or phase II detoxification reactions, and iii) whether changes in renal bile acid transporter expression also contribute to increased bile acid elimination.

EXPERIMENTAL PROCEDURES
Animals C57/BL6 mice, 25-30 g, with targeted disruption of Fxr (4), obtained from Frank J. Gonzalez (National Cancer Institute, National Institutes of Health, Bethesda, MD), and wild-type littermates were housed with a 12/12 h light/dark cycle and permitted ad libitum consumption of water and a standard mouse diet. The experimental protocol was approved by the local Animal Care and Use Committee, according to criteria outlined in the "Guide for the Care and Use of Laboratory Animals" prepared by the National Academy of Sciences (National Institutes of Health publication 86- 23, revised 1985).

CBDL
To study the role of FXR in changes in bile acid metabolism and the expression of renal bile acid transporters in cholestasis, 2 month old male wild-type and Fxr 2/2 mice were subjected to bile duct ligation and cholecystectomy, as described previously (7). Sham-operated animals were subjected to the same surgical procedure, but without ligation of the common bile duct and removal of the gallbladder. Livers were excised under general anesthesia with avertin (400 mg/kg body weight, intraperitoneally) 3 and 7 days after surgery, respectively (three to five animals were studied in each group). Urine was collected in metabolic cages, and bile was sampled by puncturing of the gallbladder in naive mice or the dilated bile duct after CBDL.

Preparation of total RNA and determination of mRNA levels by real-time PCR
The preparation of total RNA, reverse transcription into cDNA, TaqManT real-time PCR, and primer/probe sequences for Cyp2b10 and Cyp3a11 were described previously (13,14).
Preparation of kidney membranes and analysis of renal bile acid transporter protein levels by Western blotting Kidney membranes were prepared and Mrp2-Mrp4 and organic anion-transporting polypeptide 1 (Oatp1) protein levels were determined as described previously (15). Apical sodium-dependent bile acid transporter (Asbt) protein expression levels were determined using a polyclonal rabbit antibody against Asbt (dilution, 1:2,500; kindly provided by Dr. Paul A. Dawson, Wake Forest University School of Medicine, Winston-Salem, NC). Blots were reprobed with an anti-b-actin antibody (1:5,000; Sigma) to confirm the specificity of changes in transporter protein levels. Apical membrane targeting of Asbt was confirmed by immunohistochemistry (13) using the Asbt antibody provided by Dr. Dawson. No differences were found between wild-type and Fxr 2/2 mice, either in naive or 7 d bile duct-ligated animals (data not shown).

Statistical analysis
In each group, three to five animals were studied. Data are reported as arithmetic means 6 SD. For statistical analysis, ANOVA with Bonferroni posttest testing (for multiple comparisons) or Student's t-test (for single time points of two groups) was used with the SigmaStat statistics program (Jandel Scientific, San Rafael, CA). P , 0.05 was considered significant.

Quantitative changes in bile acid levels during obstructive cholestasis
Liver tissue bile acid levels increased during CBDL; in wild-type mice, from 0.07 6 0.04 mmol/g at baseline to 1.00 6 0.06 mmol/g at day 3 and to 0.92 6 0.43 mmol/g at day 7 after CBDL. Table 1 shows that liver tissue bile acid levels in Fxr 2/2 mice did not differ from those in wild-type mice at baseline (0.10 6 0.07 mmol/g) but were significantly lower at day 7 after CBDL (0.36 6 0.016 mmol/g; P , 0.05), as described previously (7).
The higher urinary bile acid elimination in Fxr 2/2 mice could principally be explained by i) excretion of compounds that might differ from those in wild-type mice,ii) differences in expression levels of bile acid-transporting membrane proteins, iii) an increased bile acid production rate, or a combination of these factors. To answer these questions, we performed i) a comprehensive bile analysis and ii) a comparison of putative renal bile acid transport proteins. The total urinary excretion of bile acids in CBDL was assumed to reflect total bile acid synthesis, which would be expected to be increased in the absence of FXR.
ES-MS is a highly sensitive method for the detection of different types of conjugated bile acids. Screening by this method did not reveal the presence of bile acid sulfates or glucuronides in liver, serum, bile, or urine. The analyses were performed both before and after cholylglycine hydrolysis to exclude the suppression of low-abundance anions by the predominant ions of taurine conjugates.
ES-MS revealed differences in the proportion of polyhydroxylated bile acids between FXR genotypes. In naive wild-type liver, only anions at m/z 498 and 514, indicative of taurine-conjugated dihydroxylated and trihydroxylated bile acids, at a ratio of 1:4, were found. Spectra of liver from naive Fxr 2/2 mice showed an anion at m/z 530 of minor intensity (,10%), indicative of taurine-conjugated tetrahydroxylated bile acids. This particular ion gained in relative intensity in the spectra of liver and serum after CBDL and became most abundant in urine. The relative intensity of m/z 530, compared with m/z 514, was always higher in Fxr 2/2 mice than in wild-type mice. There was an additional ion at m/z 546, indicative of taurine-conjugated pentahydroxylated bile acids, in spectra of liver (Fig. 1) and serum from Fxr 2/2 mice at day 7 of CBDL. This ion was also found to various extents in spectra of urine from both genotypes. Together, the ES-MS analyses excluded significant phase II (sulfation and glucuronidation) detoxification of bile acids in mice with obstructive cholestasis but indicated that phase I detoxification (hydroxylation) was enhanced in Fxr 2/2 mice. Ions indicative of hydroxylated C-27 bile acid precursors (8) were not observed in any material studied.

Qualitative changes in bile acid levels during obstructive cholestasis
GC-MS analysis data of the bile acid composition of wildtype and Fxr 2/2 mice are shown for liver in Table 1, for serum and bile in Table 2, and for urine in Table 3. The relative amount of the major primary bile acid CA decreased significantly in the liver of wild-type mice, from 39% to 12% at day 7 of CBDL, whereas the contribution of CA remained unchanged in Fxr 2/2 mice, as we described previously (7). The relative amount of b-MCA increased continuously in livers of wild-type mice but decreased in livers of Fxr 2/2 mice ( Table 1). The decline of b-MCA was even more pronounced in serum of cholestatic Fxr 2/2 mice ( Table 2).
As expected, the secondary bile acids lithocholic acid (LCA) and deoxycholic acid (DCA) virtually disappeared during obstructive cholestasis (Tables 1-3). This was paralleled by the disappearance of 1b-hydroxy-DCA, in particular from livers of Fxr 2/2 mice. There was also some D 22 -b-MCA in livers and serum of naive and cholestatic wild-type mice, indicating partial b-oxidation as a pathway for bile acid metabolism not only in rats (9) but also in mice. Notably, D 22 -b-MCA was not found in livers and serum of Fxr 2/2 mice and was not excreted in bile and urine of any of the genotypes (Tables 1-3). The levels of major murine bile acids in bile of naive animals were, within statistical margins, the same as those found by Kok et al. (6) ( Table 2). The larger relative amount of biliary CA in Fxr 2/2 mice increased even more after CBDL. This might be attributable to the lack of FXR-mediated inhibition of cholesterol 7a-hydroxylase in the classical pathway leading to CA. Another new finding was the presence of small amounts of tetrahydroxylated bile acids in bile of naive Fxr 2/2 mice and in bile of both genotypes after 7 days of CBDL (Table 2).

Polyhydroxylated bile acids
The most important differences in bile acid profiles between wild-type and Fxr 2/2 mice were observed particularly in liver, serum, and urine as higher abundances of polyhydroxylated compounds in Fxr 2/2 mice. The formation of these compounds was obviously activated already in naive Fxr 2/2 mice, as seen by the occurrence of 4.3, 5.7, and 12.2% of tetrols in liver, bile, and serum, respectively, of these animals (Tables 1, 2). Tetrols were also found in the urine of naive Fxr 2/2 mice, but the total urinary excretion at baseline was too low for a quantitative analysis. During cholestasis, the relative amounts of polyhydroxylated bile acids were always higher in Fxr 2/2 mice compared with their wild-type littermates, and after 7 days of CBDL, these acids constituted z30, 45, and 100% of the total bile acids in liver, serum, and urine, respectively, from Fxr 2/2 mice (Tables 1-3).
The major part of polyhydroxylated bile acids consisted of hydroxylation products of CA, carrying an additional hydroxyl group in the 1b, 2b, 6a, 6b, 22, or 23 position. Mass spectra of the latter two compounds are shown in Fig. 2. There was also a pentahydroxylated compound found in liver, serum, and urine that was tentatively identified as CA hydroxylated in both the 6 and 22 positions. Thus, after 7 days of CBDL, 60-100% of the total polyhydroxylated bile acids found in liver, serum, and urine of Fxr 2/2 mice were hydroxylation products of CA, possibly reflecting the lack of FXR-mediated inhibition of the biosynthetic pathway to CA.

mRNA levels of Cyp2b10 and Cyp3a11
To determine whether the increase in bile acid hydroxylation is also accompanied by an increase in cytochromes P450, mRNA levels of Cyp2b10 and Cyp3a11 in the liver and kidney were determined. Two-fold higher basal expression levels of Cyp2b10 (P , 0.05) and Cyp3a11 (not significant) were found in livers of naive Fxr 2/2 mice (Fig. 3). After 3 d of CBDL, Cyp2b10 gene expression levels were increased 10-fold in both genotypes (P , 0.05 compared with naive animals; not significant between genotypes). At 7 d of CBDL, Cyp2b10 gene expression levels had returned to the baseline in both wild-type and Fxr 2/2 mice (Fig. 3). In contrast, Cyp3a11 expression levels in Fxr 2/2 mice were significantly higher at both 3 and 7 d of CBDL compared with those in wild-type mice (3.5-and 4.0-fold, respectively) (Fig. 3). No significant Cyp2b10 or Cyp3a11 gene expression was found in the kidney of naive or cholestatic animals. These data indicate the involvement of hepatic Cyp3a11 in the hydroxylation of bile acids.

Renal bile acid transporter expression profiles
To investigate whether changes in renal bile acid export (Mrp2 and Mrp4) and reuptake transporters (Oatp1 and Asbt) could account for the differences in urinary bile acid clearance, protein levels of the respective transporters were determined. In contrast to the situation in rats (15), no changes were observed in the expression levels of renal Mrp2 and Asbt (Fig. 4). Also, Oatp1 remained unchanged. However, renal Mrp4 was significantly induced in both genotypes after 3 days of CBDL, 2.9-fold in wild-type mice and 2.3-fold in Fxr 2/2 mice. Mrp3 was not expressed or induced in the kidney of any of the genotypes (data not shown). Notably, significant differences in baseline levels were not observed in any of the renal transporters studied (data not shown). Together, these data suggest a coordinated alternative excretory pathway for bile acids via hepatic (7) and renal Mrp4.

DISCUSSION
This study presents detailed analyses of bile acids in liver, serum, bile, and urine of cholestatic Fxr 2/2 mice and parallel measurements of the expression of major renal bile acid transporters. Our data provide an explanation for the surprising finding that Fxr 2/2 mice eliminate the systemic and hepatic bile acid load more rapidly in biliary obstruction than their wild-type littermates, despite their inability to upregulate the canalicular bile acid export pump, Bsep. Instead, this inability may result in the efflux of bile acids from the hepatocytes into the blood for subsequent excretion into urine.
Fxr 2/2 mice most efficiently synthesize polyhydroxylated bile acids, and the majority of these compounds are hydroxylation products of CA, the most abundant bile acid in mice. This prototype phase I detoxification reaction is of particular importance because Fxr 2/2 mice, in contrast to their littermates, are unable to downregulate Cyp8b1 (6, 7) and actually were shown to have a 2-fold increased synthesis rate of CA (6). Thus, the metabolism we describe here preferably detoxifies CA, which is the major bile acid retained in cholestasis and has a notorious toxicity in feeding experiments on mice (4,5).
This mechanism is of obvious importance in Fxr 2/2 mice. We estimate that Fxr 2/2 mice, with a mean urinary bile acid excretion rate of z2 mmol/day during the first 7 days after CBDL, eliminate z50% of the bile acid load during obstructive cholestasis into urine. This estimation is based on recent data on the total bile acid pool size and CA production rates in Fxr 2/2 mice with normal intestinal bile acid absorption (6).
Although sulfation and glucuronidation are of relevance in the urinary elimination of bile acids in humans, our ES-MS data show that significant amounts of sulfates or glucuronides are not present in CBDL mice. This is in agreement with data obtained in rats using a similar experimental approach (16). Thus, these phase II detoxification mechanisms are most likely of no importance in rodents with obstructive cholestasis. The induction of hydroxysteroid sulfotransferase in LCA-fed Fxr 2/2 mice described recently (17) may be specific for LCA but apparently does not lead to an increased sulfation of other bile acids. LCA is a very minor compound in naive Fxr 2/2 mice and is virtually absent in obstructive cholestasis, as shown in this study and a previous study (6).
Our gene expression studies support the involvement of Cyp3a11 in the hydroxylation of CA. The expression of pregnane X receptor (PXR)-dependent Cyp3a11 (18,19) was significantly more increased after 3 and 7 d of CBDL in Fxr 2/2 mice than in wild-type mice. In agreement with previous reports (7,20), we found enhanced expression of Cyp3a11 already in naive animals, but this difference was not statistically different. Cyp3a11 is homologous to human CYP3A4, which has been shown in both in vitro (18,21) and in vivo (21) experiments to hydroxylate bile acids in the 1b, 6a, and 6b positions. Cyp2b10 gene expression was also studied, because its human homolog, CYP2B6, can hydroxylate 5b-cholestane-3a,7a,12a-triol, a minor CA precursor, in the 25 position, although to a much lesser extent than CYP3A4 (22). We did not find significant differences in Cyp2b10 gene expression levels in cholestatic mice of either genotype, supporting the assumption that this enzyme is not involved in the bile acid hydroxylations seen in this study. Increased formation of polyhydroxylated bile acids was also found in naive wildtype Swiss albino mice given constitutive androstane receptor (CAR)-and PXR-stimulating agents (14). However, the structures of these bile acids were not established.
The formation of hydroxylation products is a prototype phase I detoxification reaction. A larger number of hydroxyl groups are associated with higher hydrophilicity and usually with lower toxicity. In the case of bile acid metabolism in humans, this form of detoxification involves CYP3A4 in a feed-forward manner. Although LCA has been shown to bind to and activate the nuclear Cyp3a11 (lower graph) in biliary obstruction. Increases of Cyp2b10 mRNA expression levels are observed in both genotypes at day 3. In contrast, mRNA expression levels of Cyp3a11 differ significantly between wild-type and Fxr 2/2 mice in obstructive cholestasis on both days 3 and 7, which is consistent with the larger proportion of polyhydroxylated bile acids formed in Fxr 2/2 mice. Controls were naive Fxr 2/2 wild-type mice.
Enhanced phase I detoxification of bile acids in Fxr 2/2 mice receptor PXR that enhances CYP3A4 expression, resulting in the formation particularly of 6a-hydroxylated bile acids (18), this mechanism has also been shown for CBDL mice with undetectable LCA levels. This suggests that accumulation of bile acids other than LCA also could activate this distinct detoxification pathway (23).
Because polyhydroxylated bile acids did not appear in great amounts in bile but were found almost exclusively in serum and urine, an efficient export of these compounds via hepatic basolateral and renal tubular export seems most likely. The rapid declines in liver tissue and systemic bile acid levels were paralleled with an enhanced expression of Mrp4 in both the liver (5) and the kidney. Thus, it is attractive to speculate that these two adaptive mechanisms work together (i.e., that Mrp4 actively excretes tetrahydroxylated bile acids both from the liver at the basolateral side and from the kidney at the apical side). However, we did not find the same relationships between enhanced bile acid hydroxylation and enhanced Mrp4 expression in wild-type and Fxr 2/2 genotypes, which indicates that the changes of Mrp4 expression and bile acid hydroxylation are differently regulated in the two genotypes. Whether passive glomerular filtration or another, as yet undefined transport system contributes to the efficient renal clearance of polyhydroxylated bile acid metabolites remains to be investigated. The fact that excretion of these compounds into bile did not increase during CBDL argues against Mrp2 as a potential transporter. Mrp3, which is upregulated to a greater extent in the liver of CBDL Fxr 2/2 mice (5), is not even expressed/ induced in the kidney. Preliminary results of renal Mrp6 expression after bile acid treatment indicate that this tubular apical transport system also did not contribute to the increased urinary bile acid excretion (8). Furthermore, decreases in renal Oatp1 and Asbt, which would be predicted to enhance renal bile acid output by reduced bile acid reuptake, were not detected in CBDL mice, in contrast to the downregulation of renal Asbt in CBDL rats (16).
The mechanisms by which Fxr 2/2 mice protect themselves against bile acid toxicity more efficiently than their wild-type littermates may have implications for our understanding of protective mechanisms in human cholestasis. The elimination of bile acids in urine may be considerable in patients with extrahepatic or intrahepatic cholestasis (24). Most of the hydroxylation products of CA that we have found to increase substantially in Fxr 2/2 mice have been described in urine of humans with various cholestatic liver diseases (12,(25)(26)(27). In fact, it was concluded that the occurrence of these compounds most typically distin- Fig. 4. Effects of biliary obstruction on putative renal bile salt transporters in wild-type and Fxr 2/2 mice. Kidney membranes were isolated from wild-type and Fxr 2/2 mice before (day 0) and after 3 and 7 days of CDBL and analyzed by Western blotting as described in Experimental Procedures. On day 3 of CBDL, a significant induction of multidrug resistance-related protein 4 (Mrp4) was observed in both genotypes (P , 0.05). Asbt, apical sodium-dependent bile acid transporter; Oatp1, organic anion-transporting polypeptide 1.
guished normal from cholestatic conditions (23), and it was suggested that cholestatic infants with a good capacity for hydroxylation reactions have a better prognosis (24). Notably, in these studies of humans, the tetrahydroxylated bile acids were amidated but otherwise unconjugated, as found here in the mouse. Large amounts of sulfated bile acids are excreted in human urine during cholestasis, but sulfation is most active with less polar monohydroxy and dihydroxy bile acids. Besides chenodeoxycholic acid, these include the secondary bile acids LCA and DCA (24), which also decrease in humans with time of cholestasis as a result of decreased reabsorption from the intestine and enhanced 6a-hydroxylation (12,(25)(26)(27)(28). Glucuronidation then emerges as a specific detoxification of 6a-hydroxylated bile acids (29).
Our data show that Fxr 2/2 mice have higher biliary bile acid concentrations than their wild-type littermates not only at baseline, as also shown by Kok et al. (6), but also after CBDL. However, this does not necessarily indicate larger amounts of bile acids in the bile duct. The latter is the case only in wild-type mice, in which we found a significant increase in bile duct pressure in CBDL as a result of continuous bile acid pumping via upregulated Bsep (7). In Fxr 2/2 mice, biliary bile acids may be excreted via other as yet undefined bile acid transporters. This is probably also the case in Bsep (spgp) knockout mice (30,31).
In summary, we describe hydroxylation reactions to be the first line of defense against bile acid toxicity in biliary obstruction. We speculate that the more rapid and efficient hydroxylation in Fxr 2/2 mice, together with enhanced Mrp4 expression, favor these animals for studies of treatments aimed at an enhancement of cytochrome P450dependent phase I detoxification reactions [e.g., by the administration of PXR ligands such as pregnenolone-16acarbonitrile (PCN) or statins].