Farnesoid X Receptor Activation Increases Reverse Cholesterol Transport by Modulating Bile Acid Composition and Cholesterol Absorption in Mice

AMERICAN ASSOCIATION FOR THE STUDY OFLIVERD I S E ASES

HEPATOLOGY, VOL. 64, NO. 4, 2016 Farnesoid X Receptor Activation Increases Reverse Cholesterol Transport by Modulating Bile Acid Composition and Cholesterol Absorption in Mice

Yang Xu,1 Fei Li,2 Munaf Zalzala,1,3 Jiesi Xu,1 Frank J. Gonzalez,2 Luciano Adorini,4 Yoon-Kwang Lee,1 Liya Yin,1 and Yanqiao Zhang1

Activation of farnesoid X receptor (FXR) markedly attenuates development of atherosclerosis in animal models. However, the underlying mechanism is not well elucidated. Here, we show that the FXR agonist, obeticholic acid (OCA), increases fecal cholesterol excretion and macrophage reverse cholesterol transport (RCT) dependent on activation of hepatic FXR. OCA does not increase biliary cholesterol secretion, but inhibits intestinal cholesterol absorption. OCA markedly inhibits hepatic cholesterol 7a-hydroxylase (Cyp7a1) and sterol 12a-hydroxylase (Cyp8b1) partly through inducing small heterodimer partner, leading to reduced bile acid pool size and altered bile acid composition, with the a/b-muricholic acid pro- portion in bile increased by 2.6-fold and taurocholic acid (TCA) level reduced by 71%. Overexpression of Cyp8b1 or concurrent overexpression of Cyp7a1 and Cyp8b1 normalizes TCA level, bile acid composition, and intestinal cholesterol absorption. Conclusion: Activation of FXR inhibits intestinal cholesterol absorption by modulation of bile acid pool size and composition, thus leading to increased RCT. Targeting hepatic FXR and/or bile acids may be useful for boosting RCT and preventing the development of atherosclerosis. (HEPATOLOGY 2016;64:1072-1085)

he farnesoid X receptor (FXR) is a nuclear for secretion to the bile and feces. Macrophage RCT is hormone receptor that regulates multiple bio- known to be atheroprotective.(9,10) Consistent with the T logical processes. FXR plays an important role role of FXR in RCT and TG metabolism, activation in regulating bile acid, lipid and glucose metabolism, of FXR is shown to protect against development of inflammation, and energy homeostasis.(1-3) Activation atherosclerosis(11,12) and nonalcoholic fatty liver disease of FXR improves insulin sensitivity, lowers blood and (NAFLD).(13) Given all these beneficial effects, FXR hepatic triglycerides (TGs) and blood cholesterol, and has been considered an attractive therapeutic target for inhibits hepatic inflammatory response.(4-7) In addi- treatment of metabolic disorders.(14,15) tion, activation of FXR increases macrophage reverse FXR is highly expressed in the liver, intestine, kid- cholesterol transport (RCT),(8) a process that delivers ney, and adrenal gland, but is undetectable in macro- cholesterol from peripheral macrophages to the liver phages.(16) Thus, activation of FXR affects the

Abbreviations: Abcg5, ATP-binding cassette subgroup G member 5; Abcg8, ATP-binding cassette subgroup G member 8; acLDL, acetylated low-density lipoprotein; Bsep, bile salt export protein; CA, cholic acid; cDNA, complementary DNA; cpm, counts per minute; Cyp7a1, cholesterol 7a-hydroxylase; Cyp8b1, sterol 12a-hydroxylase; DCA, deoxycholic acid; DMEM, Dulbecco’s modiﬁed Eagle’s medium; EMEM, Eagle’s minimum essential medium; FBS, fetal bovine serum; FPLC, fast protein liquid chromatography; FXR, farnesoid X receptor; HDL, high-density lipoprotein; HDL-C, HDL-cholesterol; LDL, low-density lipoprotein; LSC, liquid scintillation counter; MCAs, muricholic acids; a/b-MCA, a/b-muricholic acid; Mdr2, multidrug resistance protein 2; mRNA, messenger RNA; NAFLD, nonalcoholic fatty liver disease; OCA, obeticholic acid; PBS, phosphate-buffered saline; PCR, polymerase chain reaction; RCT, reverse cholesterol transport; Shp, small heterodimer partner; SR-BI, scavenger receptor group B type 1; TCA, taurocholic acid; TCDCA, taurochenodeoxycholic acid; TDCA, taurodeoxycholic acid; TGs, triglycerides; TUDCA, tauroursodeoxycholic acid. This work was supported by NIH grants R01HL103227, R01DK095895, and R01DK102619 (to Y.Z.). Received December 1, 2015; accepted June 27, 2016. Additional Supporting Information may be found at onlinelibrary.wiley.com/doi/10.1002/hep.28712/suppinfo. Copyright VC 2016 by the American Association for the Study of Liver Diseases. View this article online at wileyonlinelibrary.com. DOI 10.1002/hep.28712 Potential conflict of interest: Dr. Adorini is employed by and owns stock in Intercept.

1072 HEPATOLOGY, Vol. 64, No. 4, 2016 XU ET AL. development of atherosclerosis likely through modulat- overexpression of CYP8B1 or CYP7A1 plus CYP8B1. ing plasma lipid levels. We have previously shown that These data demonstrate that activation of FXR activation of FXR lowers plasma high-density lipopro- increases RCT by regulating bile acid synthesis and tein (HDL) level by inducing scavenger receptor group composition and subsequent inhibition of intestinal B type 1 (SR-BI).(8) Overexpression of SR-BI is shown cholesterol absorption. to lower plasma HDL-cholesterol (HDL-C) level, pro- mote RCT, and protect against atherosclerosis.(17,18) Consequently, activation of FXR also increases macro- Materials and Methods (8) phage RCT. So far, the mechanism underlying FXR- MICE, FXR LIGANDS, AND induced of RCT is not well elucidated. Cholesterol absorption is an important component ADENOVIRUSES of RCT. Around 50% dietary or biliary cholesterol is C57BL/6 mice, Fxr2/2 mice, albumin-Cre mice, absorbed in the intestine. Therefore, inhibition of and vilin-Cre were purchased from The Jackson Labo- (19-21) intestinal cholesterol absorption promotes RCT. ratory (Bar Harbor, ME). Fxrfl/fl mice(24) and Shp2/2 As amphipathic molecules, bile acids are important mice(25) have been described previously. Liver- or regulators of cholesterol absorption. Among all the bile intestine-specific Fxr2/2 mice were generated by acids, cholic acid (CA) has the most capability to pro- crossing albumin-Cre mice or vilin-Cre mice with mote cholesterol absorption, whereas muricholic acids Fxrfl/fl mice. Where indicated, 10- to 12-week-old (MCAs) are the strongest inhibitors of cholesterol mice were gavaged with either vehicle (0.5% carboxy- (22) absorption. FXR plays an essential role in all aspects methycellulose; Sigma-Aldrich, St. Louis, MO) or of bile acid metabolism, including bile acid synthesis, vehicle containing GW4064 (30 mg/kg, twice a day) conjugation, secretion, absorption, and uptake by hep- or OCA (40 mg/kg) for 6-9 days. Adenoviruses (23) atocytes. Classic bile acid synthesis is controlled expressing rat Cyp7a1 (Ad-rCyp7a1),(26) mouse mainly by two rate-limiting enzymes, cholesterol 7a- Cyp8b1 (Ad-mCyp8b1),(27) FXRa1-VP16, and hydroxylase (CYP7A1) and sterol 12a-hydroxylase FXRa2-VP16(4) have been described. Unless otherwise (CYP8B1), which are inhibited by FXR. CYP8B1 stated, mice were euthanized 7 days postinjection. In determines the production rate of CA. general, all mice were fasted for 5-6 hours before In this study, we used global and tissue-specific euthanization. All animal experiments were approved 2/2 Fxr mice to investigate how FXR regulates RCT. by the institutional animal care and use committee at Our data show that the FXR agonist, obeticholic acid Northeast Ohio Medical University (Rootstown, OH). (OCA), increases fecal cholesterol excretion and macrophage RCT through activation of hepatic FXR. CHEMICAL REAGENTS AND Such an effect coincides with unchanged biliary choles- ANTIBODIES terol secretion. Activation of FXR inhibits intestinal cholesterol absorption by altering bile acid pool size TRIzol was purchased from Invitrogen (Carlsbad, and composition, and this effect is abolished by CA). A reverse-transcription kit was purchased from

ARTICLE INFORMATION:

From the 1Department of Integrative Medical Sciences, Northeast Ohio Medical University, Rootstown, OH; 2Laboratory of Metabolism, Center for Cancer Research, National Cancer Institute, NIH, Bethesda, MD; 3Department of Pharmacology and Toxicology, College of Pharmacy, University of Baghdad, Baghdad, Iraq; 4Intercept Pharmaceuticals, New York, NY. ADDRESS CORRESPONDENCE AND REPRINT REQUESTS TO:

Yanqiao Zhang, M.D. Liya Yin, M.D., Ph.D. Department of Integrative Medical Sciences Department of Integrative Medical Sciences Northeast Ohio Medical University Northeast Ohio Medical University 4209 Street, Route 44 4209 Street, Route 44 P.O. Box 95, Rootstown, OH 44272 P.O. Box 95, Rootstown, OH 44272 E-mail: [email protected] E-mail: [email protected] Tel: 11-330-325-6693; or Tel: 11-330-325-6521

1073 XU ET AL. HEPATOLOGY, October 2016

Applied Biosystems (Foster City, CA). CYP7A1 anti- bile acids were quantified on a Waters Acquity H- body was a general gift from Dr. David Russel at Uni- Class UPLC system using a Waters Acquity BEH versity of Texas Southwestern Medical Center. C18 column (2.1 3 100 mm) coupled to a Waters CYP8B1 antibody was purchased from Santa Cruz Xevo G2 QTOFMS, as described.(32) Biotechnology (Santa Cruz, CA). Calnexin and b- actin were purchased from Novus Biologicals (Little- ton, CO). GALLBLADDER CANNULATION Mice were fasted for 4 hours followed by anesthesia. REAL-TIME POLYMERASE CHAIN The lower end of the common bile duct was ligated REACTION and the gallbladder cannulated below the entrance of RNA was isolated using TRIzol Reagent. Messen- the cystic duct using a PE-10 catheter (Becton Dick- ger RNA (mRNA) levels were determined by quanti- sinson, Sparks, MD). After successful catheterization tative reverse-transcription polymerase chain reaction and flow of fistula bile, the cystic duct was doubly (PCR) on a 7500 real-time PCR machine from ligated. Hepatic bile was collected by gravity for the Applied Biosystems (Foster City, CA) by using SYBR first 1-2 hours. Green Supermix (Roche, Indianapolis, IN). Results were calculated using threshold cycle values and normalized to 36B4 mRNA level. IN VIVO RCT The in vivo RCT was performed essentially as WESTERN BLOTTING ASSAYS described.(8,33) Briefly, J774 macrophages (ATCC, Western blotting assays were performed using Manassas, VA), grown in suspension in Dulbecco’s whole-liver lysates(4) or nuclear extracts of the liver modified Eagle’s medium (DMEM) containing 10% samples,(28,29) as described. Gel band intensity was fetal bovine serum (FBS) and penicillin/streptomycin, were cultured in uncoated cell-culture dishes. To load quantified using ImageJ software (National Institutes 3 of Health, Bethesda, MD) and normalized to b-actin cells with H-cholesterol and acetylated low-density or histone. lipoprotein (acLDL), J774 cells were incubated in DMEM/10% FBS containing 5 lCi/mL of 3 l LIPID AND LIPOPROTEIN H-cholesterol and 25 g/mL of acLDL for 48 hours, followed by wash in phosphate-buffered saline (PBS) ANALYSIS and equilibration for 4 hours in DMEM supplemented Approximately 100 mg of liver tissue were homoge- with 0.2% bovine serum albumin and penicillin/strep- nized in methanol and lipids were extracted in chloro- tomycin. Before injection, cells were pelleted and form/methanol (2:1, v/v), as described.(30) Hepatic TG resuspended in Eagle’s minimum essential medium and cholesterol levels were then quantified using Infin- (EMEM; ATCC). All mice were caged individually ity reagents from Thermo Scientific (Waltham, MA). with free access to water and food. Mice were injected Plasma TG and cholesterol levels were also determined intraperitoneally with 0.5 mL of J774 cells (typically 6 6 using Infinity reagents. Fast protein liquid chromatog- 5 3 10 cells) in EMEM containing 6.5-10 3 10 raphy (FPLC) analysis of lipoprotein profile was per- counts per minute (cpm). Blood was collected at 24 3 formed as described.(29,31) and 48 hours, and plasma H-tracers were assayed using a liquid scintillation counter (LSC). Feces were BILE ACID LEVEL AND collected continuously from 0 to 48 hours. At the end COMPOSITION of the experiment, mice were perfused with cold PBS. A portion of the liver was removed for lipid and RNA Bile acids in liver, intestine, and gallbladder were extraction. extracted using ethanol. Bile acid levels in liver, intes- Approximately 100 mg of liver were homogenized tine, gallbladder or plasma were measured using a bile in methanol and lipids were extracted in chloroform/ acid kit (Diazyme, San Diego, CA). Total bile acids in methanol (2:1, v/v), as described.(30) Lipids were col- liver, intestine, and gallbladder were used for calcula- lected, dried, and resuspended in toluene and counted tion of total bile acids/bile acid pool size. Individual in an LSC.

1074 HEPATOLOGY, Vol. 64, No. 4, 2016 XU ET AL.

FECAL CHOLESTEROL AND BILE attributed to decreased plasma HDL-C level (data not ACIDS shown). OCA also increased fecal free cholesterol level (Fig. 1B, left panel), but reduced fecal bile acid level Fecal total sterols, cholesterol, and bile acids were (Fig. 1B, right panel) in wild-type mice, but not in (8,33) extracted as described. Radioactivity was measured Fxr2/2 mice. These data demonstrate that OCA low- using an LSC. For mice that were not given any radio- ers plasma cholesterol and increases fecal free choles- isotopes, feces were collected consecutively for 48 terol level through activation of FXR. hours. After total sterols, cholesterol, and bile acids FXR is highly expressed in both the liver and were extracted, free cholesterol and total bile acids intestine. To determine whether OCA regulates cho- were measured using a free cholesterol kit (Wako lesterol metabolism through activation of hepatic USA, Richmond, VA) or total bile acid kit (Diazyme). and/or intestinal FXR, we treated liver-specific Fxr2/2 (L-Fxr2/2) mice, intestine-specific Fxr2/2 (I-Fxr2/2) INTESTINAL CHOLESTEROL mice, and their control littermates (Fxrfl/fl mice) with ABSORPTION either vehicle or OCA for 7 days. OCA lowered plasma cholesterol (Fig. 1C) and increased fecal free cho- Intestinal cholesterol absorption was performed as lesterol level (Fig. 1D, left panel) in control mice, but (8,34) described. Briefly, mice were injected with 2.5 not in L-Fxr2/2 mice. In addition, OCA reduced fecal 3 lCi of H-cholesterol in Intralipid (Sigma-Aldrich) by bile acid levels in both genotypes (Fig. 1D, right tail vein injection, followed by immediate gavage with panel). 14 1 lCi of C-cholesterol in median-chain TGs (MCT In contrast to what was observed in L-Fxr2/2 mice, oil; Mead Johnson, Evansville, IN). After 72 hours, OCA reduced plasma cholesterol and increased fecal blood and tissue were collected. The percent of the free cholesterol level in both control mice and I-Fxr2/ intestinal cholesterol absorption was calculated using the 2 mice (Fig. 1E,F, left panel), but reduced fecal bile 14 formula: [percent of intragastric dose ( C-cholesterol) acid level only in control mice (Fig. 1F, right panel). per ml plasma] 4 [percent of intravenous dose Collectively, the data of Fig. 1 indicate that OCA 3 ( H-cholesterol) per mL plasma] 3 100. reduces plasma cholesterol and increases fecal cholesterol level through activation of hepatic FXR, whereas STATISTICAL ANALYSIS reduction in fecal bile acids is dependent on activation of intestinal FXR. Statistical significance was analyzed using an unpaired Student t test or analysis of variance (Graph- OCA INHIBITS CYP7A1 AND Pad Prism; GraphPad Software Inc., La Jolla CA). All values are expressed as mean 6 standard error of the CYP8B1 EXPRESSION THROUGH mean. Differences were considered statistically signifi- ACTIVATION OF HEPATIC AND/ cant at P < 0.05. OR INTESTINAL FXR AND PARTLY THROUGH SMALL Results HETERODIMER PARTNER To determine whether OCA functioned properly in OCA LOWERS PLASMA animals, we analyzed hepatic gene expression. As CHOLESTEROL AND INCREASES expected, OCA induced Shp (small heterodimer part- FECAL CHOLESTEROL ner), a well-characterized FXR target gene, and repressed Cyp7a1 and Cyp8b1 mRNA levels in wild- EXCRETION THROUGH 2/2 ACTIVATION OF HEPATIC FXR type mice, but not in Fxr mice (Fig. 2A). However, repression of Cyp7a1 or Cyp8b1 was mitigated or abol- To investigate how FXR activation increases RCT, ished in L-Fxr2/2 mice (Fig. 2B) or I-Fxr2/2 mice we treated Fxr2/2 mice and their wild-type littermates (Fig. 2C), suggesting that OCA represses Cyp7a1 and with either vehicle or the semisynthetic FXR agonist, Cyp8b1 expression through activation of hepatic and/ OCA, for 7 days. OCA lowered plasma total choles- or intestinal FXR. Previous studies show that activa- terol in wild-type mice, but not in Fxr2/2 mice (Fig. tion of FXR induces intestinal expression of FGF15, 1A). Such a reduction in plasma cholesterol level was an enterohepatic hormone, to repress hepatic CYP7A1

1075 XU ET AL. HEPATOLOGY, October 2016

FIG. 1. OCA lowers plasma cholesterol and increases fecal cholesterol excretion by activation of hepatic FXR. (A,B) Fxr1/1 and Fxr2/2 mice were treated with either vehicle or OCA for 7 days (n 5 7-8). Plasma cholesterol (A), fecal free cholesterol (B, left panel), and fecal bile acid (B, right panel) levels were determined. (C,D) Fxrfl/fl and L-Fxr2/2 mice were treated with either vehicle or OCA for 7 days (n 5 8). Plasma cholesterol (C), fecal free cholesterol (D, left panel), and fecal bile acid (D, right panel) levels were determined. (E,F) Fxrfl/fl and I-Fxr2/2 mice were treated with either vehicle or OCA for 7 days (n 5 8). Plasma cholesterol (E), fecal free cholesterol (F, right panel), and fecal bile acid (F, right panel) levels were determined. *P < 0.05; **P < 0.01. Unless otherwise indicated, the difference is shown when compared to vehicle-treated control mice.

and CYP8B1 expression.(35,36) In I-Fxr2/2 mice, inhibit both Cyp7a1 and Cyp8b1, we gavaged wild- OCA treatment completely failed to inhibit Cyp7a1 type mice and Shp2/2 mice with vehicle or the speciﬁc and only partially inhibited Cyp8b1 (Fig. 2C), which FXR agonist, GW4064, for 7 days. The data of Fig. may explain why OCA treatment reduces fecal bile 2D show that GW4064 markedly reduced both acid levels dependent on activation of intestinal FXR Cyp7a1 and Cyp8b1 expression, and such a reduction (Fig. 1F). was greatly attenuated in Shp2/2 mice, indicating that SHP is known to inhibit bile acid synthesis.(25,37) activation of FXR represses Cyp7a1 and Cyp8b1 To determine whether Shp is required for FXR to expression partly through induction of SHP.

1076

FIG. 2. OCA inhibits CYP7A1 and CYP8B1 expression through activation of hepatic and/or intestinal FXR and partly through SHP. (A-C) Liver samples were collected from mice described in the legend of Fig. 1. Hepatic mRNA levels in Fxr1/1 or Fxr2/2 mice (A), Fxrfl/fl or L-Fxr2/2 mice (B) and Fxrfl/fl or I-Fxr2/2 mice (C) were determined (n 5 8). (D) Shp1/1 and Shp2/2 mice were gavaged with either vehicle (Veh) or GW4064 for 7 days (n 5 6). Hepatic mRNA levels were determined. (E) Fxrfl/fl mice, L-Fxr2/2 mice, and I-Fxr2/2 mice were gavaged with either vehicle or OCA for 7 days (n 5 6). Bile acid levels in liver, gallbladder, and intestine, as well as total bile acid levels, were determined. *P < 0.05; **P < 0.01. In (E), the difference was compared between the OCA-treated group and the vehicle-treated group of the same genotype.

XU ET AL. HEPATOLOGY, October 2016

Consistent with the involvement of both hepatic ACTIVATION OF FXR DOES NOT and intestinal FXR in regulating CYP7A1 and INCREASE BILIARY CYP8B1 expression, OCA treatment reduced bile CHOLESTEROL SECRETION acid levels in liver and gallbladder or intestine dependent on activation of hepatic or intestinal FXR, Biliary cholesterol secretion plays an important role respectively (Fig. 2E). In addition, both hepatic and in RCT. Given that hepatic FXR activation is required intestinal FXR were required for OCA to reduce bile for OCA-induced macrophage RCT, we investigated acid pool size (Fig. 2E) or plasma bile acid level whether activation of FXR increased biliary cholesterol (Supporting Fig. S1). OCA did not lower plasma secretion. FXR has four isoforms. Previous data sug- bile acid levels in global Fxr2/2 mice (data not gest that FXRa1 and FXRa2 may have similar func- (16) shown). Together, the data of Fig. 2 demonstrate tions as FXRa3 and FXRa4, respectively. that suppression of CYP7A1 and CYP8B1 expres- Therefore, we over-expressed FXRa1-VP16 or sion and bile acid pool size involves activation of FXRa2-VP16 in the liver to constitutively activate hepatic and/or intestinal FXR and its downstream FXRa1orFXRa2. As expected, activation of FXRa1 target, SHP. or FXRa2 markedly inhibited bile acid secretion (Fig. 4A, left panel). However, activation of FXRa2, but not FXRa1, inhibited bile flow (Fig. 4A, middle panel) HEPATIC FXR IS REQUIRED FOR and reduced biliary cholesterol secretion (Fig. 4A, right OCA TO INCREASE panel). Consistent with these observations, activation MACROPHAGE RCT of FXRa1orFXRa2 markedly inhibited Cyp7a1 and Cyp8b1 expression (Fig. 4B). Interestingly, activation The data of Fig. 1 suggest that OCA increases fecal of FXRa1 induced bile salt export protein (Bsep)by free cholesterol excretion and lowers plasma cholesterol 9-fold, whereas activation of FXRa2 induced Bsep level through activation of hepatic FXR. Analysis of only by 2.3-fold (Fig. 4B). In addition, activation of plasma lipoprotein profile by FPLC indicates that FXRa1, but not FXRa2, induced multidrug resistance OCA lowered plasma cholesterol levels in control lit- protein 2 (Mdr2) and ATP-binding cassette subgroup termates primarily by reducing HDL-C (Fig. 3A). In G member 5 (Abcg5) expression (Fig. 4B). Neither addition, there was an increase in low-density lipopro- 2/2 FXRa1 nor FXRa2 activation had an effect on ATP- tein (LDL)-cholesterol in L-Fxr mice (Fig. 3A). binding cassette subgroup G member 8 (Abcg8) expres- Given that HDL plays an important role in RCT, we sion (Fig. 4B). determined the role of hepatic FXR in macrophage We next investigated whether the specific FXR ago- RCT. nist, GW4064, had similar effects as constitutively Forty-eight hours after injection of J774 macro- 3 activated FXR (FXR-VP16) in the liver. Treatment of phages loaded with [ H]cholesterol and acLDL, blood mice with GW4064 had no effect on bile flow (Fig. radioactivity was significantly reduced (Fig. 3B), 4C) or biliary cholesterol secretion (Fig. 4D). In a 3 whereas fecal [ H]total sterols (Fig. 3C) as well as macrophage RCT study, we did not observe any 3 [ H]free cholesterol level (Fig. 3D) were significantly change in biliary [3H] radioactivity after GW4064 increased, in OCA-treated control littermates, but treatment (Fig. 4E). In addition, GW4064 treatment 2/2 these changes were not observed in L-Fxr mice. significantly induced Shp and Bsep, but inhibited 3 There was not much change in fecal [ H]bile acid level Cyp7a1 and Cyp8b1 (Fig. 4F), suggesting that (Fig. 3E) or hepatic radioactivity (data not shown) in GW4064 functioned properly in vivo. 3 both genotypes. The unchanged fecal [ H]bile acid Studies with another specific FXR agonist, OCA, level is likely a net result of increased transport of produced similar results. OCA treatment reduced bile 3 [ H]cholesterol from macrophages to the liver by SR- acid secretion in wild-type, but not in Fxr2/2, mice BI and an inhibition of bile acid biosynthesis. Last, (Supporting Fig. S3A), and had no effect on bile flow OCA treatment significantly reduced hepatic Cyp7a1 (Supporting Fig. S3B) or biliary cholesterol secretion mRNA level (Fig. 3F), but induced SR-BI mRNA (Supporting Fig. S3C) in wild-type or Fxr2/2 mice. and protein levels (Supporting Fig. S2), indicating that Further studies involving liver- or intestine-specific OCA functioned properly in vivo. Thus, the data of Fxr2/2 mice show that OCA inhibited biliary bile Fig. 3 demonstrate that hepatic FXR activation is acid secretion dependent on activation of both hepatic required for OCA to increase macrophage RCT. and intestinal FXR (Supporting Fig. S3D). OCA

1078 HEPATOLOGY, Vol. 64, No. 4, 2016 XU ET AL.

FIG. 3. OCA increases macrophage reverse cholesterol transport by activation of hepatic FXR. (A) Fast protein liquid chromatography FPLC analysis of plasma lipoprotein profile in Fxrfl/fl and L-Fxr2/2 mice (n 5 8). (B-F) Fxrfl/fl and L-Fxr2/2 mice were gavaged with either vehicle (Veh) or OCA (n 5 8). After 7 days, mice were injected intravenously with J774 macrophages loaded with [3H]cholesterol and RCT was performed as described in Materials and Methods. Plasma radioactivity (B), fecal [3H]total sterols (C), fecal [3H]free cholesterol (D), fecal [3H]bile acids (E), and hepatic mRNA levels (F) were determined. *P < 0.05; **P < 0.01. Unless otherwise indicated, the difference is shown when compared to vehicle-treated control mice. Abbreviation: VLDL, very-low-density lipoprotein.

treatment had no effect on bile flow in Fxrfl/fl mice, L- component of RCT. Bile acids play a critical role in Fxr2/2 mice, or I-Fxr2/2 mice (Supporting Fig. regulating cholesterol absorption, with CA being the S3E). Together, the data of Fig. 4 and Supporting Fig. most efficient bile acid in promoting cholesterol S3 demonstrate that activation of FXR does not absorption. In the classic pathway of bile acid biosyn- increase biliary cholesterol secretion and inhibits bile thesis, CYP7A1 catalyzes the rate-limiting step, acid secretion through activation of both hepatic and whereas CYP8B1 determines rate of CA biosynthesis intestinal FXR. (Fig. 5A). OCA treatment significantly inhibited cholesterol absorption (Fig. 5B). This inhibition was ACTIVATION OF FXR REDUCES attributed to activation of hepatic FXR because OCA INTESTINAL CHOLESTEROL was able to inhibit intestinal cholesterol absorption in 2/2 2/2 ABSORPTION PRIMARILY I-Fxr mice, but not in L-Fxr mice (Supporting THROUGH INHIBITION OF Fig. S4). Given that FXR activation markedly inhib- HEPATIC CYP8B1 ited both Cyp7a1 and Cyp8b1 expression, we asked whether recapitulation of hepatic Cyp7a1 and/or Intestinal cholesterol absorption plays an important Cyp8b1 levels would prevent FXR-induced inhibition role in cholesterol homeostasis and is an important of cholesterol absorption. Overexpression of Cyp7a1

1079 XU ET AL. HEPATOLOGY, October 2016

FIG. 4. Activation of FXR does not increase biliary cholesterol secretion. (A,B) C57BL/6 mice were injected intravenously with Ad- VP16 (control), Ad-FXRa1-VP16 (FXRa1), or Ad-FXRa2-VP16 (FXRa2). After 7 days, gallbladder cannulation was performed. The rates of bile acid secretion (A, left panel), bile flow (A, middle panel), and biliary cholesterol secretion (A, right panel), as well as hepatic mRNA levels (B), were determined. (C-F) C57BL/6 mice were gavaged with either vehicle or GW4064. Bile flow rate (C) and biliary cholesterol secretion rate (D) were determined. In (E), C57BL/6 mice were also injected intraperitoneally with J774 macrophages loaded with [3H]cholesterol and bile [3H]-radioactivity was determined after 48 hours. In (F), hepatic mRNA levels were quantified. *P < 0.05; **,#P < 0.01. Unless otherwise indicated, the difference is shown when compared to the control mice or vehicle treatment group.

(from rat complementary DNA [cDNA]) and/or (Fig. 5B). OCA markedly inhibited endogenous Cyp8b1 (from mouse cDNA) had no effect on intesti- Cyp7a1 mRNA level (Fig. 5C), and reexpression of nal cholesterol absorption in vehicle-treated mice Cyp7a1 was able to recapitulate hepatic CYP7A1 pro- (Supporting Fig. S5A-D). Concurrent expression of tein level to the normal level (Fig. 5D). The data of Cyp7a1 and Cyp8b1 or overexpression of Cyp8b1 alone, Fig. 5D,E show that OCA markedly inhibited hepatic but not Cyp7a1 alone, recovered intestinal cholesterol levels of Cyp8b1 mRNA and protein, which were absorption to the normal level in OCA-treated mice reversed by Cyp8b1 overexpression. Taken together,

1080 HEPATOLOGY, Vol. 64, No. 4, 2016 XU ET AL.

FIG. 5. Reexpression of CYP8B1 alone or CYP7A1 plus CYP8B1 prevents OCA from inhibiting intestinal cholesterol absorption. (A) Diagram showing that CYP7A1 and CYP8B1 are key enzymes in the classic pathway of bile acid biosynthesis. (B-E) C57BL/6 mice were injected intravenously with a control virus, Ad-rCyp7a1 and/or Ad-mCyp8b1, on day 1. On day 2, mice were gavaged with either vehicle or OCA (n 5 6). On day 8, mice were given [3H]cholesterol and [14C]cholesterol and intestinal cholesterol absorption was performed using the dual-isotope plasma ratio method (B). Hepatic Cyp7a1 mRNA level (C), protein levels (D), and Cyp8b1 mRNA level (E) were determined. *P < 0.05; **P < 0.01. Unless otherwise indicated, the difference is shown when compared to the vehicle treatment group. Abbreviation: ND, not detectable.

the data of Fig. 5 and Supporting Fig. S5 demonstrate overexpression of Cyp7a1 and/or Cyp8b1 increased that activation of FXR inhibits intestinal cholesterol bile acid levels in the intestine, liver, and/or gallblad- absorption primarily through inhibition of hepatic der, as well as total bile acid levels, in vehicle-treated CYP8B1 expression. mice (Supporting Fig. S6A-D). Activation of FXR by OCA signiﬁcantly reduced (Fig. 6A,B) or tended to (Fig. 6C) reduce bile acid level in the intestine (Fig. OVEREXPRESSION OF CYP8B1 OR 6A), liver (Fig. 6B), and gallbladder (Fig. 6C). As a CYP7A1 PLUS CYP8B1 PREVENTS result, OCA signiﬁcantly reduced bile acid pool size FXR-INDUCED INCREASE IN by 76% (Fig. 6D). Overexpression of Cyp7a1 alone in MURICHOLIC ACID LEVEL AND OCA-treated mice partly or completely recovered bile DECREASE IN TAUROCHOLIC acid levels in the intestine (Fig. 6A), liver (Fig. 6B), ACID LEVEL and gallbladder (Fig. 6C) and partly recovered total bile acid levels (Fig. 6D). In contrast, concurrent To determine how overexpression of CYP7A1 and expression of both Cyp7a1 and Cyp8b1 or Cyp8b1 CYP8B1 prevents FXR-induced inhibition of intesti- overexpression alone completely recovered or nal cholesterol absorption, we analyzed bile acid pool increased bile acid levels in the intestine, liver, and size and bile acid composition. As expected, gall bladder as well as total bile acid levels in OCA-

1081 XU ET AL. HEPATOLOGY, October 2016

FIG. 6. Reexpression of CYP8B1 alone or CYP7A1 plus CYP8B1 prevents OCA from altering TCA and MCA levels. (A-F) C57BL/6 mice were injected intravenously with a control virus, Ad-rCyp7a1 and/or Ad-mCyp8b1. One day later, mice were gavaged with either vehicle or OCA (n 5 6). Bile acid levels in the intestine (A), liver (B), and gallbladder (C), as well as total bile acid levels (including intestine, liver, and gallbladder) (D), were analyzed. Individual bile acid levels in bile were quantiﬁed (E), and the composition (%) of individual bile acids under different treatments is shown (F). *P < 0.05; **P < 0.01. Unless otherwise indicated, the difference is shown when compared to the vehicle treatment group.

treated mice (Fig. 6A-D). Interestingly, although mice (Fig. 6A), Cyp8b1, but not Cyp7a1,overexpres- overexpression of Cyp7a1 or Cyp8b1 increased intesti- sion was able to normalize cholesterol absorption nal bile acid levels to a similar extent in OCA-treated (Fig. 5B).

1082 HEPATOLOGY, Vol. 64, No. 4, 2016 XU ET AL.

Analysis of individual bile acid levels in bile indi- Our previous data show that activation of FXR cated that OCA treatment reduced, by 71%, levels of increases macrophage RCT through a pathway that taurocholic acid (TCA), the most abundant bile requires SR-BI.(8) Given that FXR activation induces acid, whereas concurrent expression of both Cyp7a1 SR-BI expression in the liver,(8) one may speculate and Cyp8b1 or Cyp8b1 overexpression alone recapit- that FXR activation likely increases biliary cholesterol 2 2 ulated TCA level to the normal level (Fig. 6E). secretion. By using tissue-specific Fxr / mice, we OCA treatment also significantly reduced deoxy- show that the synthetic FXR agonist, OCA, increases cholic acid (DCA) and taurodeoxycholic acid macrophage RCT through activation of hepatic FXR. (TDCA) levels, which could not be recovered to However, neither constitutive activation of FXR in the normal levels by Cyp8b1 or Cyp7a1 plus Cyp8b1 liver nor GW4064 or OCA treatment increases biliary overexpression (Fig. 6E). In contrast, OCA treat- cholesterol secretion. This latter observation supports ment had not much of an effect on the levels of CA, the recent findings that the intestine is capable of mod- (38) a/b-muricholic acid (a/b-MCA), taurochenodeoxy- ulating RCT directly. In addition, our data show cholic acid (TCDCA), or tauroursodeoxycholic acid that FXR activation significantly inhibits biliary bile (TUDCA; Fig. 6E). acid secretion. Given that FXR activation is known to Further analysis of bile acid composition in bile increase phospholipid secretion, our observations raise indicated that OCA treatment increased relative levels that possibility that FXR activation may increase the of a/b-MCA from 11.3% to 29.4%, and this increase formation of cholesterol gallstone. was blunted by overexpression of Cyp8b1 alone or The role of FXR in controlling bile acid homeostasis Cyp7a1 plus Cyp8b1, but not by overexpression of has been well established. Our data indicate that OCA Cyp7a1 alone (Fig. 6F). Given that MCAs are the represses both CYP7A1 and CYP8B1 expression most hydrophilic bile acids and the most potent inhibi- through activation of hepatic and intestinal FXR and tors of cholesterol absorption whereas cholic acids (CA partly through inducing SHP. Overexpression of and TCA) are the most potent promoters of cholester- CYP8B1 or CYP7A1 plus CYP8B1 is sufficient to ol absorption among all the bile acids,(22) the data of prevent OCA-induced inhibition of intestinal choles- Fig. 6 suggest that the changes in MCA and CA levels terol absorption, demonstrating that activation of FXR are responsible for the reduced cholesterol absorption inhibits intestinal cholesterol absorption by modulating after OCA treatment. bile acid metabolism. Gardes et al. have reported that synthetic FXR agonists, including OCA (also called 6E-CDCA), GW4064, R05186026, and X-Ceptor Discussion compound, all inhibit intestinal cholesterol absorption, reduce bile acid pool size, and decrease fecal bile acid Atherosclerosis is the leading cause of coronary heart excretion in Ldlr2/2 mice that are fed a Western disease. Activation of FXR has been shown to increase diet,(39) a known atherogentic mouse model. Thus, the (8) RCT and protects against development of athero- studies by us and Gardes et al. indicate that activation 2/2 2/2 (11,12) sclerosis in Ldlr or Apoe mice. In addition, of FXR inhibits intestinal cholesterol absorption in activation of FXR also lowers TG levels and inhibits both wild-type (normal) mice and atherogenic mice by inflammation in the liver, thus preventing development regulating bile acid metabolism. Activation of intesti- (13) of NAFLD. Finally, activation of FXR improves nal FXR has been shown to inhibit hepatic CYP7A1 (4-6) insulin sensitivity in diabetic mice or humans. expression through fibroblast growth factor 15/19.(35) These observations have suggested that FXR is a ther- However, recapitulation of hepatic CYP7A1 expres- apeutic target for treatment of common metabolic dis- sion has no effect on FXR-induced inhibition of intes- orders, including NAFLD, atherosclerosis, and insulin tinal cholesterol absorption, supporting the finding resistance. However, the mechanism underlying the that hepatic FXR activation is responsible for induc- regulation of RCT by FXR is not well elucidated. By tion of RCT. 2 2 using tissue-specific Fxr / mice, herein we show that Hydrophobic bile acids, such as CA and DCA, pro- activation of FXR increases RCT by suppression of mote cholesterol absorption whereas hydrophilic bile hepatic bile acid synthesis, alternation of bile acid com- acids, such as MCAs, are potent inhibitors of choles- position, and subsequent inhibition of intestinal cho- terol absorption.(22) In hamsters, TCA is identified as lesterol absorption, a novel pathway that is a potential plasma biomarker of early atheromatous independent of the canonical biliary secretion route. plaque formation.(40) Our data show that OCA

1083 XU ET AL. HEPATOLOGY, October 2016 reduces hydrophobic bile acid (CA and DCA) level by a b 5) Mudaliar S, Henry RR, Sanyal AJ, Morrow L, Marschall HU, 71% and increases the relative level of / -MCAs by Kipnes M, et al. Efficacy and safety of the farnesoid X receptor 2.6-fold. Overexpression of CYP8B1 alone or agonist obeticholic acid in patients with type 2 diabetes and non- CYP7A1 plus CYP8B1, but not CYP7A1 alone, alcoholic fatty liver disease. Gastroenterology 2013;145:574- reverses OCA-induced changes in bile acid level 582.e1. a b 6) Ma K, Saha PK, Chan L, Moore DD. Farnesoid X receptor is (TCA), bile acid composition (TCA and / -MCAs), essential for normal glucose homeostasis. J Clin Invest 2006;116: and cholesterol absorption. Interestingly, overexpres- 1102-1109. sion of CYP8B1 alone or CYP7A1 plus CYP8B1 does 7) Li Y, Jadhav K, Zhang Y. Bile acid receptors in non-alcoholic not restore DCA/TDCA levels, suggesting that OCA fatty liver disease. Biochem Pharmacol 2013;86:1517-1524. 8) Zhang Y, Yin L, Anderson J, Ma H, Gonzalez FJ, Willson TM, inhibits cholesterol absorption through inhibition of Edwards PA. Identification of novel pathways that control farne- CA/TCA synthesis and induction of a/b-MCA levels. soid X receptor-mediated hypocholesterolemia. J Biol Chem Overexpression of CYP7A1 normalizes bile acid pool 2010;285:3035-3043. size by 65%, but does not affect cholesterol absorption, 9) Cuchel M, Rader DJ. Macrophage reverse cholesterol transport: suggesting that change in bile acid pool size may key to the regression of atherosclerosis? Circulation 2006;113: per se 2548-2555. not contribute to reduced cholesterol absorption. Giv- 10) Rosenson RS, Brewer HB, Jr., Davidson WS, Fayad ZA, Fuster en that intestinal cholesterol absorption is an impor- V, Goldstein J, et al. Cholesterol efflux and atheroprotection: tant component of RCT, our data suggest that advancing the concept of reverse cholesterol transport. Circula- activation of FXR increases RCT by altering bile acid tion 2012;125:1905-1919. 11) Hartman HB, Gardell SJ, Petucci CJ, Wang S, Krueger JA, synthesis and composition and subsequently inhibiting Evans MJ. Activation of farnesoid X receptor prevents atheroscle- intestinal cholesterol absorption. rotic lesion formation in LDLR-/- and apoE2/2 mice. J Lipid Our current findings are consistent with previous Res 2009;50:1090-1100. observations that inhibition of bile acid synthesis 12) Flatt B, Martin R, Wang TL, Mahaney P, Murphy B, Gu XH, 2/2 (41,42) 2/2 et al. Discovery of XL335 (WAY-362450), a highly potent, increases RCT. Cyp7a1 mice or Cyp8b1 selective, and orally active agonist of the farnesoid X receptor (43) mice have reduced intestinal cholesterol absorption (FXR). J Med Chem 2009;52:904-907. and increased fecal sterol secretion. Loss of Cyp8b1 in 13) Li Y, Jadhav K, Zhang Y. Bile acid receptors in non-alcoholic Apoe2/2 mice is shown to inhibit development of ath- fatty liver disease. Biochem Pharmacol 2013;86:1517-1524. (44) 14) Zhang Y, Edwards PA. FXR signaling in metabolic disease. erosclerosis. Therefore, targeting bile acid synthesis FEBS Lett 2008;582:10-18. may be a useful strategy for prevention of 15) Porez G, Prawitt J, Gross B, Staels B. Bile acid receptors as tar- atherogenesis. gets for the treatment of dyslipidemia and cardiovascular disease. Accumulating data have indicated that FXR is an J Lipid Res 2012;53:1723-1737. 16) Zhang Y, Kast-Woelbern HR, Edwards PA. Natural structural attractive target for treatment of diverse metabolic dis- variants of the nuclear receptor farnesoid X receptor affect tran- orders, including atherosclerosis. The data presented scriptional activation. J Biol Chem 2003;278:104-110. here provide important insights into the mechanism 17) Kinoshita M, Fujita M, Usui S, Maeda Y, Kudo M, Hirota D, underlying FXR-medicated inhibition of atherosclero- et al. Scavenger receptor type BI potentiates reverse cholesterol transport system by removing cholesterol ester from HDL. Ath- sis. In addition, our studies suggest that FXR and bile erosclerosis 2004;173:197-202. acids may be targeted for enhancing RCT and thus for 18) van der Velde AE, Groen AK. Shifting gears: liver SR-BI drives preventing the development of atherosclerosis. reverse cholesterol transport in macrophages. J Clin Invest 2005; 115:2699-2701. 19) Wang J, Mitsche MA, Lutjohann D, Cohen JC, Xie XS, Hobbs REFERENCES HH. Relative roles of ABCG5/ABCG8 in liver and intestine. J Lipid Res 2015;56:319-330. 1) Lefebvre P, Cariou B, Lien F, Kuipers F, Staels B. Role of bile 20)HuangY,WangJ,QuanG,WangX,YangL,ZhongL.Lac- acids and bile acid receptors in metabolic regulation. Physiol Rev tobacillus acidophilus ATCC 4356 prevents atherosclerosis via 2009;89:147-191. inhibition of intestinal cholesterol absorption in apolipoprotein 2) Zhang Y. Farnesoid X receptor—acting through bile acids to E-knockout mice. Appl Environ Microbiol 2014;80: treat metabolic disorders. Drugs Future 2010;35:635-641. 7496-7504. 3) Fang S, Suh JM, Reilly SM, Yu E, Osborn O, Lackey D, et al. 21) Davidson MH, Voogt J, Luchoomun J, Decaris J, Killion S, Intestinal FXR agonism promotes adipose tissue browning and Boban D, et al. Inhibition of intestinal cholesterol absorption reduces obesity and insulin resistance. Nat Med 2015;21: with ezetimibe increases components of reverse cholesterol trans- 159-165. port in humans. Atherosclerosis 2013;230:322-329. 4) Zhang Y, Lee FY, Barrera G, Lee H, Vales C, Gonzalez FJ, 22) Wang DQ, Tazuma S, Cohen DE, Carey MC. Feeding natural et al. Activation of the nuclear receptor FXR improves hypergly- hydrophilic bile acids inhibits intestinal cholesterol absorption: cemia and hyperlipidemia in diabetic mice. Proc Natl Acad Sci studies in the gallstone-susceptible mouse. Am J Physiol Gastro- U S A 2006;103:1006-1011. intest Liver Physiol 2003;285:G494-G502.

1084 HEPATOLOGY, Vol. 64, No. 4, 2016 XU ET AL.

23) Lee FY, Lee H, Hubbert ML, Edwards PA, Zhang Y. FXR, a enterohepatic signal to regulate bile acid homeostasis. Cell Metab multipurpose nuclear receptor. Trends Biochem Sci 2006;31:572- 2005;2:217-225. 580. 36) Kong B, Wang L, Chiang JY, Zhang Y, Klaassen CD, Guo 24) Sinal CJ, Tohkin M, Miyata M, Ward JM, Lambert G, GL. Mechanism of tissue-specific farnesoid X receptor in sup- Gonzalez FJ. Targeted disruption of the nuclear receptor FXR/ pressing the expression of genes in bile-acid synthesis in mice. BAR impairs bile acid and lipid homeostasis. Cell 2000;102:731- HEPATOLOGY 2012;56:1034-1043. 744. 37) Kerr TA, Saeki S, Schneider M, Schaefer K, Berdy S, Redder T, 25) Wang L, Lee YK, Bundman D, Han Y, Thevananther S, Kim et al. Loss of nuclear receptor SHP impairs but does not elimi- CS, et al. Redundant pathways for negative feedback regulation nate negative feedback regulation of bile acid synthesis. Dev Cell of bile acid production. Dev Cell 2002;2:721-731. 2002;2:713-720. 26) Pandak WM, Schwarz C, Hylemon PB, Mallonee D, Valerie K, 38) Brufau G, Groen AK, Kuipers F. Reverse cholesterol transport Heuman DM, et al. Effects of CYP7A1 overexpression on cho- revisited: contribution of biliary versus intestinal cholesterol lesterol and bile acid homeostasis. Am J Physiol Gastrointest excretion. Arterioscler Thromb Vasc Biol 2011;31:1726-1733. Liver Physiol 2001;281:G878-G889. 39) Gardes C, Chaput E, Staempfli A, Blum D, Richter H, Benson 27) Pandak WM, Bohdan P, Franklund C, Mallonee DH, GM. Differential regulation of bile acid and cholesterol metabo- Eggertsen G, Bjorkhem I, et al. Expression of sterol 12alpha- lism by the farnesoid X receptor in Ldlr2/2 mice versus hamsters. hydroxylase alters bile acid pool composition in primary rat hepa- J Lipid Res 2013;54:1283-1299. tocytes and in vivo. Gastroenterology 2001;120:1801-1809. 40) Jove M, Ayala V, Ramirez-Nunez O, Serrano JC, Cassanye A, 28) Zhang Y, Yin L, Hillgartner FB. SREBP-1 integrates the Arola L, et al. Lipidomic and metabolomic analyses reveal poten- actions of thyroid hormone, insulin, cAMP, and medium-chain tial plasma biomarkers of early atheromatous plaque formation in fatty acids on ACCalpha transcription in hepatocytes. J Lipid hamsters. Cardiovasc Res 2013;97:642-652. Res 2003;44:356-368. 41) Schwarz M, Russell DW, Dietschy JM, Turley SD. Alternate path- 29) Xu J, Li Y, Chen W, Xu Y, Yin L, Ge X, et al. Hepatic carbox- ways of bile acid synthesis in the cholesterol 7alpha-hydroxylase ylesterase 1 is essential for both normal and farnesoid X- knockout mouse are not upregulated by either cholesterol or chole- receptor-controlled lipid homeostasis. HEPATOLOGY 2014;59: styramine feeding. J Lipid Res 2001;42:1594-1603. 1761-1771. 42) Erickson SK, Lear SR, Deane S, Dubrac S, Huling SL, Nguyen 30) Bligh EG, Dyer WJ. A rapid method of total lipid extraction L, et al. Hypercholesterolemia and changes in lipid and bile acid and purification. Can J Biochem Physiol 1959;37:911-917. metabolism in male and female cyp7A1-deficient mice. J Lipid 31) Xu Y, Zalzala M, Xu J, Li Y, Yin L, Zhang Y. A metabolic Res 2003;44:1001-1009. stress-inducible miR-34a-HNF4alpha pathway regulates lipid 43) Li-Hawkins J, Gafvels M, Olin M, Lund EG, Andersson U, and lipoprotein metabolism. Nat Commun 2015;6:7466. Schuster G, et al. Cholic acid mediates negative feedback regulation 32)JiangC,XieC,LiF,ZhangL,NicholsRG,KrauszKW, of bile acid synthesis in mice. J Clin Invest 2002;110:1191-1200. Cai J, et al. Intestinal farnesoid X receptor signaling promotes 44) Slatis K, Gafvels M, Kannisto K, Ovchinnikova O, Paulsson- nonalcoholic fatty liver disease. J Clin Invest 2015;125:386- Berne G, Parini P, et al. Abolished synthesis of cholic acid 402. reduces atherosclerotic development in apolipoprotein E knock- 33) Naik SU, Wang X, Da Silva JS, Jaye M, Macphee CH, Reilly out mice. J Lipid Res 2010;51:3289-3298. MP, et al. Pharmacological activation of liver X receptors promotes reverse cholesterol transport in vivo. Circulation 2006;113: Author names in bold designate shared co-first 90-97. 34) Turley SD, Herndon MW, Dietschy JM. Reevaluation and authorship. application of the dual-isotope plasma ratio method for the mea- surement of intestinal cholesterol absorption in the hamster. J Lipid Res 1994;35:328-339. Supporting Information 35) Inagaki T, Choi M, Moschetta A, Peng L, Cummins CL, McDonald JG, et al. Fibroblast growth factor 15 functions as an Additional Supporting Information may be found at onlinelibrary.wiley.com/doi/10.1002/hep.28712/suppinfo.

1085