Effects of feeding bile acids and a bile acid sequestrant on hepatic bile acid composition in mice.

An improved ultra performance liquid chromatography-tandem mass spectrometry (UPLC/MS/MS) method was established for the simultaneous analysis of various bile acids (BA) and applied to investigate liver BA content in C57BL/6 mice fed 1% cholic acid (CA), 0.3% deoxycholic acid (DCA), 0.3% chenodeoxycholic acid (CDCA), 0.3% lithocholic acid (LCA), 3% ursodeoxycholic acid (UDCA), or 2% cholestyramine (resin). Results indicate that mice have a remarkable ability to maintain liver BA concentrations. The BA profiles in mouse livers were similar between CA and DCA feedings, as well as between CDCA and LCA feedings. The mRNA expression of Cytochrome P450 7a1 (Cyp7a1) was suppressed by all BA feedings, whereas Cyp7b1 was suppressed only by CA and UDCA feedings. Gender differences in liver BA composition were observed after feeding CA, DCA, CDCA, and LCA, but they were not prominent after feeding UDCA. Sulfation of CA and CDCA was found at the 7-OH position, and it was increased by feeding CA or CDCA more in male than female mice. In contrast, sulfation of LCA and taurolithocholic acid (TLCA) was female-predominant, and it was increased by feeding UDCA and LCA. In summary, the present systematic study on BA metabolism in mice will aid in interpreting BA-mediated gene regulation and hepatotoxicity.

genate was spiked with 10 µl of ISs, mixed, and equilibrated on ice for 10 min. An amount of 3 ml of ice-cold alkaline acetonitrile (5% ammonia in acetonitrile) was added to the homogenate, which was then vortexed vigorously and shaken continuously for 1 h at room temperature. The mixture was centrifuged at 12,000 g for 10 min, and the supernatant was collected. The pellet was resuspended in 1 ml of methanol, sonicated for 5 min, and centrifuged at 12,000 g for 10 min. The two supernatants obtained were combined, evaporated under vacuum, and reconstituted in 100 µl of 50% methanol. The suspension was transferred into a 0.2 µm Costar Spin-X HPLC microcentrifuge fi lter (purchased from Corning Inc., Corning, NY), and centrifuged at 20,000 g for 10 min. The supernatant was then ready for injection.

BA quantifi cation
Liquid chromatographic and mass spectrometric conditions are described in the supplementary information (supplementary Table I and Fig. I). For preparation of standard stock solutions, 10 mg/ml BAs and ISs were dissolved in methanol. Because BA sulfates were synthesized without further purifi cation, their stock solutions were prepared separately. IS stock solution was diluted with 50% MeOH to a fi nal concentration of 40 µg/ml for 2 H 4 -GCDCA and 20 µg/ml for 2 H 4 -CDCA. BA stock solutions were diluted with 50% MeOH and spiked with 10 µl of ISs to construct standard curves between 5 and 50,000 ng/ml. The fi nal concentration of 2 H 4 -G-CDCA and 2 H 4 -CDCA was 4 and 2 µg/ml, respectively. The assignment of target BAs in the UPLC profi le was conducted by comparing their retention behavior and molecular mass spectra with the available BA reference standards. Quantifi cation was performed via peak area ratios (analyte versus IS) by linear-weighted (1/x 2 ) least-squares calibration curves within dividual BA concentrations in livers of male and female C57BL/6 mice subjected to a diet containing either a primary BA (CA or CDCA), a secondary BA (DCA or LCA), a therapeutic BA (UDCA), or a BA binding resin (cholestyramine). The purpose of this study is to investigate BA metabolism and synthesis, along with potential gender differences in mice fed various BA-supplemented diets.

Validation of BA quantifi cation
To optimize the chromatographic conditions, we compared the method used by Alnouti et al. ( 13 ) with that by Hagio et al. ( 14 ). The Hagio method resulted in a better separation between -, ␣ -, and ␤ -MCA, whereas the signal intensity was only about 50% of the Alnouti method. Compared with the Alnouti method, more peaks were found in the chromatograph window for CDCA using the Hagio method (data not shown). Therefore, chromatographic conditions similar to Hagio's ( 14 ) were used in the present study to separate all BA standards in less than 28 min. The intraday and interday accuracy and precision were determined according to a previous method ( 13 ); their relative standard deviations were below 15% for all BA standards (data not shown). All standard curves were constructed using a 1/concentration 2 weighted quadratic regression, and the correlation coeffi cient (r 2 ) for all BAs was above 0.99. The limit of detection (signal/noise ratio = 3) for the various BAs was in the range of 5-10 ng/ml.

Concentrations of BAs in livers of mice fed BAs
Using this method, 36 BAs were quantifi ed in the livers of mice fed various BAs or BA sequestrants ( Table 1 and supplementary fi gures). To simplify, we only list BA changes in livers of male mice after BA feedings. The predominant BAs in control mouse liver were TCA (86.1 ± 18.5 nmol/g) and CA (4.7 ± 0.9), T ␤ MCA (34.2 ± 8.8) and ␤ MCA (35.1 ± 5.0), and T MCA (18.1 ± 4.5) and MCA (11.8 ± 2.1). The major oxo-BAs in control mouse liver were 7-oxoDCA (5.1 ± 1.0) and .
In livers of mice fed CA, DCA, CDCA, LCA, or UDCA, the taurine conjugates of the fed BAs became the predominant BAs. This indicates that the fed BAs were absorbed from the intestine and delivered in portal blood to the liver, where a majority of the BAs were conjugated with taurine. Various BA biotransformations occurred subsequent to BA feeding. For example, during CA feeding, TCA (86.1 → 298.5 nmol/g) and TDCA (7.7 → 38.4) became the major constituents of hepatic BAs, followed by increased DCA (0.4 → 5.2), GCA (0.1 → 0.5), GDCA (0.0 → 0.1), isoDCA (0.0 → 0.3), and 12-oxoLCA (0.3 → 1.2). Mice fed a range of working standard concentrations. As insuffi cient T MCA and MCA were available, they were quantifi ed relatively by referring to T ␣ MCA and ␣ MCA, respectively.

Animals and treatments
Eight-week-old adult male and female C57BL/6 mice were purchased from Charles River Laboratories (Wilmington, MA). All mice were fed Teklad Rodent Diet #8604 (Harlan Laboratories, Madison, WI) ad libitum and housed according to American Animal Association laboratory animal care guidance. The control diet was prepared by grinding Harlan Teklad Rodent Diet #8064 (Harlan Laboratories). To prepare BA-supplemented diets, BAs were fi rst ground with a small amount of control diet using a mortar and pestle, and then mixed with a large amount of control diet in a Hobart food mixer (Hobart Corporation, Troy, OH). Cholestyramine-supplemented diet (2% resin by weight of diet) was prepared using the same method as BA-supplemented diets. During a preliminary study, mice were fed diets supplemented with different concentrations (0.01%, 0.1%, 0.3%, and 3% by weight of diet) of individual BAs. Concentrations that were nonlethal were selected for the present study. Individually housed C57BL/6 mice (n = 5/gender/group) were fed a control diet or a diet supplemented with 1% CA, 0.3% DCA, 0.3% CDCA, 0.3% LCA, 3% UDCA, or 2% resin for seven days. Mice were anesthetized between 8:00 AM and 12:00 AM on day 7, and gallbladders were carefully removed. Livers were then harvested, washed, frozen in liquid nitrogen, and stored at Ϫ 80°C until analysis.

Total RNA isolation
Total RNA was isolated using RNA-Bee reagent (Tel-Test Inc., Friendswood, TX) according to the manufacturer's protocol. Total RNA concentrations were quantifi ed spectrophotometrically at 260 nm. One microgram per microliter solutions were prepared from the stock RNA solutions by dilution in diethyl pyrocarbonate-treated deionized water. Integrity of RNA samples was determined by formaldehyde-agarose gel electrophoresis with visualization by ethidium bromide fl uorescence under UV light.

Multiplex suspension array
Liver mRNA was quantifi ed by multiplex suspension array (Panomics-Affymetrix, Inc., Fremont, CA). Individual gene accession numbers can be accessed at www.panomics.com. Glyceraldehyde-3-phosphate dehydrogenase (Gapdh) was used as the loading control. The detailed method was described previously ( 15 ). The mRNA data are presented as relative light units (RLU) normalized to Gapdh mRNA.

Statistical analysis
Bars represent Mean ± SEM (n = 5). Differences between mean values were tested for statistical signifi cance ( P < 0.05) by the twotailed Student's t -test.

Extraction of BAs from mouse livers
During preliminary experiments, different BA extraction methods were compared. In previous BA analyses, livers were either homogenized in 50% methanol ( 13 ) or ground as frozen tissue ( 14 ). In the present study, livers were homogenized in fi ve vol of water to obtain a homogenate that was easily transferred with a pipette. To optimize BA extraction, liver homogenates were divided into small

Concentrations of total BAs in livers of mice fed BAs
Total BAs in livers of mice are expressed as the sum of the values of each BA analyzed ( Table 2 and supplementary Fig. XI). Feeding CA, DCA, CDCA, and LCA surprisingly had little effect on total BA concentration in mouse livers. In contrast, feeding UDCA markedly increased total BAs, which may be due to the higher dose of UDCA (3%) a Signifi cant difference between the same gender of control and BA-fed groups ( P < 0.05). b Signifi cant difference between male and female mouse livers in the same group ( P < 0.05).   3. The mRNA levels of BA synthetic genes in livers of mice fed BAs and resin. Total RNA from livers of control and BA-fed mice (n = 5/gender/group) were analyzed by multiplex suspension array. The mRNA level of each gene was normalized to GAPDH. All data are expressed as mean ± SEM. for fi ve mice in each group. *Statistically signifi cant difference between the same gender of control and BA-fed groups ( P < 0.05). # Statistically signifi cant difference between male and female mouse livers in the same group ( P < 0.05). BA, bile acid; CA, cholic acid; Cont, control; CDCA, chenodeoxycholic acid; Cyp, Cytochrome P450; DCA, deoxycholic acid; LCA, lithocholic acid; UDCA, ursodeoxycholic acid; WT, wild-type. creased by feeding CA, but markedly suppressed by feeding CDCA ( Fig. 2A, B ). TCDCA7S was markedly increased by feeding CDCA or LCA, but suppressed by feeding CA or DCA ( Fig. 2C ). In addition, feeding UDCA markedly suppressed TCA7S, CA7S, and TCDCA7S. Interestingly, feeding the resin increased both TCA7S and TCDCA7S in mice.

mRNA expression of BA-synthetic genes in livers of mice fed BAs
To investigate the effects of BA feedings on BA biosynthesis, mRNA levels of BA-synthetic enzymes, such as Cyp7a1, 8b1, 27a1, and 7b1, were quantifi ed. As shown in Fig. 3 , Cyp7a1 was decreased markedly by feeding CA (M by 92%; F by 85%), CDCA (M by 95%; F by 83%), DCA (M by 92%; F by 64%), and UDCA (M by 99%; F by 99%). Cyp8b1 was higher in livers of male than female control mice. Cyp8b1 was signifi cantly decreased by feeding CA (M by 99%; F by 99%), CDCA (M by 50%; F by 77%), DCA (M by 98%; F by 98%), and UDCA (M by 99%; F by 94%). In addition, feeding LCA decreased Cyp8b1 in male mice (44%) but not female mice. Cyp27a1 was decreased by feeding CA (M by 27%; F by 23%) and UDCA (M by 54%; F by 63%). In addition, Cyp27a1 was decreased by feeding CDCA (22%) and DCA (27%) in male mice but not female mice. Control male mice have much higher Cyp7b1 mRNA expression than female mice. Cyp7b1 was markedly decreased by feeding UDCA (M by 85%; F by 64%). Moreover, feeding CA decreased Cyp7b1 in male mice (38%) but not female mice. Feeding the resin had little BAs (162.9 → 271.6), unconjugated BAs (62.7 → 1188.5) were increased much more in livers of mice fed UDCA.

Concentrations of BAs in livers of mice fed 2% resin
The effect of BA sequestrant feeding on hepatic BA metabolism was investigated by feeding mice a 2% resin-supplemented diet for seven days. As shown in supplementary

Sulfation of CA and CDCA in livers of mice fed BAs
By comparing the retention time and mass spectra with reference BA sulfates, CA and CDCA were found to be sulfated at . Due to the lack of pure standards, the peak areas of BA sulfates were normalized with those of internal standards to quantify their concentrations. As shown in Fig. 2 , TCA7S, CA7S, and TCDCA7S were higher in control male than female mice. This indicates that sulfation of CA and CDCA is male-predominant in the liver. TCA7S and CA7S were in-urine of patients with varying degrees of cholestasis, BAs were sulfated at either the 3-OH position (about two thirds) or the 7-OH position (about one third). However, Almé et al. ( 20 ) reported that monosulfates of BAs in human urine were sulfated at the 3-OH position. It was shown that monosulfates at the 7-OH position were the only BA sulfates detected in mouse feces ( 21 ). In the present study, both CA and CDCA are found to be sulfated at the 7-OH position in mouse livers. This sulfation is likely specifi c to the 7 ␣ -OH position, because DCA or UDCA feeding did not increase DCA or UDCA sulfates (data not shown). Sulfation is an important detoxifi cation pathway of LCA in man, chimpanzee, and rodents ( 22 ). However, in the present study, LCAS could only be detected in female mice fed UDCA, whereas TLCAS could be detected in female mice fed LCA or UDCA. One likely reason for this fi nding is that feeding the high dose of UDCA markedly increased both LCA and TLCA, whereas feeding LCA only increased TLCA. Sulfation activity has been shown to be genderdependent in rodents. For example, in germ-free rats, the percentage of BA sulfates was about 10-fold higher in female than male fecal contents ( 23 ). The female-predominant sulfation of LCA is likely due to Sult2a, which is predominantly expressed in female mouse livers but is essentially absent from male mouse livers ( 24 ). Unlike sulfation of LCA, sulfations of CA and CDCA are malepredominant in mouse liver. CA and CDCA feedings can suppress each other's sulfation, indicating that CA and CDCA are sulfated by the same enzyme. This enzyme activity could be inhibited by high concentrations of UDCA, effect on Cyp27a1 or Cyp7b1, but markedly increased Cyp7a1 (M by 197%; F by 1650%) and Cyp8b1 (M by 92%; F by 177%).

DISCUSSION
To understand the effects of BA feedings on hepatic gene regulation and hepatotoxicity, it is important to know the changes of liver BA concentrations and composition after BA feeding. The data from the present study showed that mouse liver has a remarkable ability to maintain total BA concentrations during BA feedings. The liver BA concentration can be determined by relative rates of BA hepatic uptake and canalicular secretion, as well as BA biosynthesis and biotransformation. The purpose of the present study was to investigate the BA biotransformation in mice during various BA feedings.
BA sulfation and glucuronidation have been thought to be important pathways to detoxify and eliminate BAs ( 16,17 ). The present study showed that BA glucuronidation is a minor BA metabolic pathway in mice. BA glucuronidation could be detected in mouse livers only after UDCA feeding (supplementary Fig. V), which is not surprising because the massive dose of UDCA (3%) may have overloaded the capacity of enzymes conjugating BA with amino acids (taurine or glycine). Dr. Alan F. Hofmann ( 18 ) has summarized the reasons why glucuronidation is not a major metabolic pathway for BAs. Confl icting data have been reported regarding the position of sulfate on the steroid nucleus of BAs. Raedsch et al. ( 19 ) reported that in the biosynthesis. Feeding CA markedly inhibited the mRNA expression of BA-synthetic enzymes (Cyp7a1, 27a1, 8b1, and 7b1), which may partly contribute to the suppression of liver BAs in mice.
The concentrations of various BAs in mouse liver after DCA feeding changed similarly to that after CA feeding ( Fig. 5 ). The majority of the fed DCA were conjugated with taurine in mouse liver. DCA feeding tended to increase hepatic TCA and CA, which was statistically significant when mice were fed a higher concentration of DCA (data not shown). Therefore, DCA and TDCA can be rehydroxylated to CA and TCA in mouse liver, a process known as BA "repair" ( 3 ). IsoDCA and 12-oxoDCA were minor metabolites during DCA feeding. DCA can undergo hepatic 3 ␣ -dehydrogenation to form 3-dehydroDCA in the guinea pig ( 27 ). In mice, 3-dehydroDCA is also a possible intermediate between DCA and isoDCA. However, due to lack of a standard, we did not quantify 3-dehydroDCA in the present study. Shefer et al. ( 28 ) showed that isoBAs can undergo hepatic transformation to their corresponding 3 ␣ -hydroxy epimers in the rat. Therefore, the DCA metabolite isoDCA could be converted back to DCA in mouse liver. Similar to feeding CA, feeding DCA also decreased most BAs in mouse livers, which may be partly due to the suppression of hepatic Cyp7a1, 8b1, and 27a1.
A scheme for CDCA biotransformation in mice is proposed in Fig. 6 . The fed CDCA is metabolized to LCA and UDCA by intestinal bacteria. The majority of them are because UDCA feeding markedly suppressed both CA and CDCA sulfations.
Based on the increased BAs in male mouse livers after BA feeding, metabolic schemes are proposed for each BA biotransformation in mice ( Figs. 4-8 ). During CA feeding, the fed CA can be metabolized to its secondary BA (DCA) by intestinal bacteria ( Fig. 4 ). Both CA and DCA are absorbed from the intestine and travel to the liver, where the majority of them are taken up and conjugated with taurine. Therefore, feeding CA markedly increased hepatic TCA and TDCA. Only a small fraction of CA and DCA were conjugated with glycine in mouse liver, because mouse BA CoA:amino acid N -acyltransferase is specifi c for taurine ( 25 ). IsoDCA is a possible product by DCA epimerization by intestinal bacteria, and 12-oxoLCA is a possible product of DCA oxidation in mouse livers. Both of them are minor metabolites during CA feeding. A previous study showed that incubation of CA with human liver microsomes produced 3-dehydroCA as the only metabolite ( 5 ). In the present study, 3-dehydroCA was not detected in livers of mice fed CA. Interestingly, feeding CA decreased most BAs in livers, especially the muricholic acids (>70%). This fi nding is consistent with the previous report that the percentage of T ␤ MCA in hepatic bile was markedly decreased in mice fed CA and DCA for seven days ( 26 ). The decrease in liver BAs, other than the metabolites of the fed BA, may be due to their dilution, enhanced excretion, biotransformation, or suppressed ify LCA ( 8,22,29 ). LCA can be hydroxylated to form HDCA, MDCA, and CDCA in human liver microsomes ( 30 ). Cyp3A has been suggested to mediate LCA hydroxylation in human, rat, and mouse (31)(32)(33). In the present study, LCA could be hydroxylated at its 6-or 7-position to produce MDCA, HDCA, or CDCA. The majority of them were conjugated with taurine in mouse liver. Among them, MDCA was increased more than other metabolites, suggesting that the 6 ␤ -position of LCA is more readily hydroxylated than other positions. Consistently, 6-oxoLCA was increased more than 7-oxoLCA after feeding LCA. In rat, MDCA, isoLCA, and dehydroLCA were the major LCA metabolites produced by rat liver microsomes, whereas 6-oxoLCA and UDCA were minor metabolites ( 34 ). In the present study, isoLCA and dehydroLCA were minor LCA metabolites in mice. Similar to CDCA feeding, LCA feeding increased T ␣ MCA and THCA, but decreased CA in male mouse livers. This result may be partly due to the suppression of the classic pathway of BA-synthetic enzymes Cyp7a1 and Cyp8b1 after LCA feeding.
UDCA is a primary BA in some mammals (e.g., bear, beaver, and nutria) and has been used to treat cholesterol gallstones, primary biliary cirrhosis, and cholestasis of pregnancy ( 3,35 ). The metabolic scheme for UDCA is proposed in Fig. 8 . UDCA can be dehydroxylated to LCA by intestinal bacteria. The high dose of UDCA (3%) could conjugated with taurine in mouse liver. During CDCA feeding, mouse liver can hydroxylate CDCA at the 6 ␤ -position to form ␣ MCA, which can be further dehydroxlated to form MDCA in the intestine. Wang et al. ( 26 ) reported that CDCA feeding increased T ␤ MCA in mouse hepatic bile. In the present study, CDCA feeding increased T ␣ MCA but not T ␤ MCA. This fi nding may be due to the difference of BA analytical methods, because it is diffi cult to separate ␣ MCA from ␤ MCA on the HPLC column. Incubation of CDCA with human hepatic microsomes suggests that 3-dehydroCDCA and HCA were major metabolites of CDCA, whereas 7-oxoLCA and CA were minor metabolites of CDCA ( 5 ). Our study suggests HCA is a minor metabolite of CDCA in mice, whereas liver CA was decreased by CDCA feeding. Two major peaks with the same mass spectra as 3-dehydroCDCA were found in livers of mice fed CDCA, whereas they were almost undetectable in livers of control mice (supplementary Fig. V). The mRNA expression of Cyp7a1 and Cyp8b1, two enzymes involved in the classic pathway of BA synthesis, were markedly suppressed by feeding CDCA. This fi nding may partly contribute to the decreased TCA, CA, TDCA, and DCA in mouse livers after CDCA feeding.
The BA profi le in mouse livers after LCA feeding undergoes changes similar to those after CDCA feeding ( Fig. 7 ). Hydroxylation and sulfation are major pathways to detox-  pression of liver Cyp7a1 and Cyp8b1 were markedly increased after feeding resin.
Male mice tend to have higher concentrations of BA metabolites than females during BA feedings ( Table 3 ). Feeding CA increased hepatic CA and DCA more in male than female mice. Moreover, 12-oxoLCA was increased only in male mice after CA feeding. Feeding DCA increased TDCA, GDCA, DCA, and 12-oxoLCA more in male than female mice. During CDCA feeding, male mice had higher hepatic CDCA, LCA, 6-oxoLCA, and 7-oxoLCA than female mice. After LCA feeding, male mice had more TMDCA, TLCA, 7-oxoLCA, 12-oxoLCA, and dehydroLCA than female mice. Gender differences in BA metabolism may explain the gender-different response of the BA-synthetic enzymes following BA feedings. For example, feeding CA suppressed Cyp7b1 in male mice but not in female mice, and feeding DCA decreased Cyp7a1 more in male mice than female mice. The gender difference in hepatic BAs was not prominent during UDCA feeding, except that male mice tended to have higher oxo-BAs than female mice. During resin feeding, female mice expressed more hepatic Cyp7a1 and Cyp27a1 than male mice. This fi nding may be due to lower liver CA concentrations in female mice than male mice after resin feeding. Because Cyp27a1 initiates the alternative pathway of BA synthesis, the induction of Cyp27a1 may explain why CDCA and LCA were increased in livers of female mice fed resin. saturate BA-conjugation enzymes, and the majority of UDCA and LCA were unconjugated in mouse livers. In mouse liver, LCA can be further hydroxylated at the 6 ␤and 7 ␣ -positions. UDCA feeding increased liver MDCA more than CDCA, which is consistent with the previous hypothesis that the 6 ␤ -position of LCA is more readily hydroxylated than the 7 ␣ -position. However, UDCA feeding had little effect on 6-oxoLCA, but it increased 7-oxoLCA. Therefore, 7-oxoLCA may be produced by UDCA instead of LCA oxidation. UDCA feeding markedly suppressed T ␣ MCA, T ␤ MCA, T MCA, TCA, TDCA, and CA, which is consistent with the previous report by Wang et al. ( 26 ) that UDCA feeding decreased T ␤ MCA and TCA in mouse hepatic bile. This suppression may be partly due to the marked inhibition of both classic (Cyp7a1 and Cyp8b1) and alternative (Cyp27a1 and Cyp7b1) pathways of BA synthesis in mice after UDCA feeding.
BA-binding resins (e.g., cholestyramine, colestipol, and colesevelam) are used to treat hypercholesterolemia and type 2 diabetes ( 36 ). Feeding the resin decreased most BAs in mouse livers. However, it tended to increase CDCA and LCA in male mice, and the increase was statistically signifi cant in female mice. This result is consistent with a previous fi nding in rats that feeding resin markedly increased CDCA but decreased ␤ MCA in bile ( 37 ). Feeding resin markedly decreased both conjugated and unconjugated BAs in mouse livers. As a feedback, the mRNA ex-BA metabolism in mice fed CA, CDCA, DCA, LCA, UDCA, or resin. Accordingly, the metabolic pathways of each BA in vivo are proposed and can be used to interpret BA-mediated gene regulation and hepatotoxicity.