Farnesoid X receptor (FXR), a bile acid-activated nuclear hormone receptor, plays an important role in the regulation of cholesterol and more specifically high-density lipoprotein (HDL) homeostasis. Activation of FXR is reported to lead to both pro- and anti-atherosclerotic effects. In the present study we analyzed the impact of different FXR agonists on cholesterol homeostasis, plasma lipoprotein profiles, and transhepatic cholesterol efflux in C57BL/6J mice and cynomolgus monkeys and atherosclerosis development in cholesteryl ester transfer protein transgenic (CETPtg) low-density lipoprotein receptor (LDLR) (−/−) mice. In C57BL/6J mice on a high-fat diet the synthetic FXR agonists isopropyl 3-(3,4-difluorobenzoyl)-1,1-dimethyl-1,2,3,6-tetrahydroazepino[4,5-b]indole-5-carboxylate (FXR-450) and 4-[2-[2-chloro-4-[[5-cyclopropyl-3-(2,6-dichlorophenyl)-4-isoxazolyl]methoxy]phenyl]cyclopropyl]benzoic acid (PX20606) demonstrated potent plasma cholesterol-lowering activity that affected all lipoprotein species, whereas 3-[2-[2-chloro-4-[[3-(2,6-dichlorophenyl)-5-(1-methylethyl)-4-isoxazolyl]methoxy]phenyl]ethenyl]benzoic acid (GW4064) and 6-ethyl chenodeoxycholic acid (6-ECDCA) showed only limited effects. In FXR wild-type mice, but not FXR(−/−) mice, the more efficacious FXR agonists increased fecal cholesterol excretion and reduced intestinal cholesterol (re)uptake. In CETPtg-LDLR(−/−) mice PX20606 potently lowered total cholesterol and, despite the observed HDL cholesterol (HDLc) reduction, caused a highly significant decrease in atherosclerotic plaque size. In normolipidemic cynomolgus monkeys PX20606 and 6-ECDCA both reduced total cholesterol, and PX20606 specifically lowered HDL2c but not HDL3c or apolipoprotein A1. That pharmacological FXR activation specifically affects this cholesterol-rich HDL2 subclass is a new and highly interesting finding and sheds new light on FXR-dependent HDLc lowering, which has been perceived as a major limitation for the clinical development of FXR agonists.
The bile acid-activated nuclear receptor FXR (farnesoid X receptor; NR1H4) has received some attention as a potential therapeutic target (Crawley, 2010; Mencarelli and Fiorucci, 2010; reviewed in Porez et al., 2012). First identified as a major transcriptional regulator of bile acid biosynthesis (Makishima et al., 1999; Parks et al., 1999), FXR has been shown to also play an important role in the regulation of cholesterol and triglyceride homeostasis (reviewed in Zhang and Edwards, 2008; Fuchs, 2012; Calkin and Tontonoz, 2012). FXR has a further indirect impact on intermediary metabolism through inducing fibroblast growth factor (FGF) 15/19 in the ileum. FGF15/19 binds to the FGF receptor 4/β-Klotho receptor complex on hepatocytes and induces a signal transduction cascade that overlaps with insulin receptor intracellular signaling (Inagaki et al., 2005; Kir et al., 2011). Both adenovirus-mediated ectopic expression of FXR and treatment with synthetic FXR agonists corrected hyperglycemia and lowered plasma cholesterol and triglyceride levels in different mouse models (Stayrook et al., 2005; Zhang et al., 2006a; Cariou et al., 2006; Ma et al., 2006). Furthermore, FXR agonist treatment of both male and female LDLR(−/−) and ApoE(−/−) mice protected against diet-induced aortic lesion formation (Hartman et al., 2009). Likewise, Mencarelli et al. (2009) reported antiatherosclerotic effects of the FXR agonist 6-ethyl chenodeoxycholic acid (6-ECDCA) in ApoE(−/−) mice. However, conflicting data were published based on findings in FXR(−/−) mice. Although FXR(−/−) mice display a proatherogenic lipid profile with elevated levels of plasma total cholesterol (TC) and total triglycerides (TG) (Sinal et al., 2000), two independent studies reported reduced atherosclerosis in FXR-deficient mice (Guo et al., 2006; Zhang et al., 2006b). Further concerns regarding the therapeutic value of FXR agonists were raised when it was shown that the FXR agonist isopropyl 3-(3,4-difluorobenzoyl)-1,1-dimethyl-1,2,3,6-tetrahydroazepino[4,5-b]indole-5-carboxylate (FXR-450) reduced the levels of high-density lipoprotein cholesterol (HDLc) in mice and hamsters, but not in Sprague-Dawley rats (Evans et al., 2009). This is consistent with the observation that plasma HDLc was increased in FXR(−/−) mice (Lambert et al., 2003). The finding that FXR activation might reduce apolipoprotein A1 (ApoAI) in human hepatocytes could, if confirmed, limit the clinical use of FXR agonists (Claudel et al., 2002). In more recent studies no impact of selective FXR agonists on ApoAI regulation was reported (Gardès et al., 2011). Next to the influence FXR exerts on cholesterol and HDL metabolism, FXR is expressed in the vasculature and in macrophages where it might also contribute to its overall impact on atherogenesis (Bishop-Bailey et al., 2004; Li et al., 2007). Thus, the question of whether FXR activation in sum translates into an atheroprotective or a proatherogenic effect has not yet been conclusively answered (Hageman et al., 2010; Porez et al., 2012). To date, only limited human or monkey data are available to unravel the role of FXR in controlling HDL metabolism in these species.
We set out to more thoroughly compare the therapeutic value of different FXR agonists by investigating their effects on plasma cholesterol lipoprotein homeostasis in wild-type C57BL/6J mice and cynomolgous monkeys. Furthermore, we tested two FXR agonists for their effect on aortic plaque development in cholesteryl ester transfer protein (CETP) transgenic LDLR(−/−) mice, a model believed to more closely resemble the human lipoprotein profile. For in vivo studies, the selective and potent nonsteroidal FXR agonist 3-[2-[2-chloro-4-[[3-(2,6-dichlorophenyl)-5-(1-methylethyl)-4-isoxazolyl]methoxy]phenyl]ethenyl]benzoic acid (GW4064) has limited utility because of poor systemic exposure and rapid clearance by the liver. To address these major pharmacokinetic limitations, we synthesized 4-[2-[2-chloro-4-[[5-cyclopropyl-3-(2,6-dichlorophenyl)-4-isoxazolyl]methoxy]phenyl] cyclopropyl]benzoic acid (PX20606), a structurally related novel and selective FXR agonist with similar properties in in vitro assays compared with GW4064 but with considerably better systemic exposure, including the liver as a major target organ for FXR activation. 6-ECDCA and FXR-450 were included in the analysis because they represent independent chemotypes of FXR agonists; both were tested in human clinical studies (FXR-450 up to phase I; 6-ECDCA is now in phase III clinical studies, see www.interceptpharma.com). We show that the rank order efficacies of these different FXR agonists in inducing the hepatic steps of cholesterol excretion from the body correlate with their effects on plasma HDLc lowering. Furthermore, pharmacological FXR activation led to an overall reduction in athereosclerotic plaque formation in CETP transgenic LDLR(−/−) mice and a lowering of HDL2c but not HDL3c or ApoAI in cynomolgus monkeys. Thus, these data suggest that despite an apparent HDL-lowering activity, an FXR full agonist with potent effects on transhepatic cholesterol efflux may turn out to be a good candidate for a potential FXR-targeted pharmacotherapy of atherosclerosis or other cholesterol-related diseases.
Materials and Methods
Radiochemicals, FXR Agonists, and Other Chemicals.
[14C]cholesterol and [3H]cholesterol were supplied by PerkinElmer Life and Analytical Sciences (Rodgau, Germany), and [3H]sitostanol was supplied by American Radiochemicals (St. Louis, MO). GW4064, FXR-450, 6-ECDCA, and 4-[[[6-[[5-cyclopropyl-3-(2,6-dichlorophenyl)4-isoxazolyl]methoxy]-2-(trifluoromethyl)-3-pyridinyl]methylamino]methyl]benzoic acid (PX20350) were synthesized according to published synthesis protocols (Maloney et al., 2000; Pellicciari et al., 2002; Mehlmann et al., 2009; Abel et al., 2010) and checked for identity and purity by 1H-NMR and liquid chromatography/mass spectrometry before being tested in FXR activity assays. Synthesis of PX20606 has been described previously (Kremoser et al., 2011). All other chemicals used were of analytical grade or higher.
C57BL/6J mice were purchased from Elevage Janvier (Rennes, France) at the age of 8 weeks. After an acclimation period of 2 weeks, animals were maintained on a Surwit high-fat diet (HFD) (Ssniff, Soest, Germany) with 60 kcal% from fat plus 0.5% (w/w) extra cholesterol (Sigma-Aldrich, St. Louis, MO). Animals were prefed on this diet for 4 weeks, and two independent plasma samples were drawn within 7 days to monitor TC and TG. Only the upper 75% of animals in terms of TC, TG, and body weight were then randomized into treatment groups to yield a similar baseline for each group regarding these parameters. The test compounds were incorporated into the diet by grinding them with mortar and pestle into the prewarmed softened diet and thoroughly homogenizing them afterwards. The diet was pressed into 50-g rectangular blocks, allowed to dry overnight at room temperature, and stored at 4°C until use (typically within 5–7 days of preparation). The required daily doses were calculated as the percentage of test compound from the average food uptake and adjusted to the weekly food uptake actually observed.
Mice carrying the human CETP gene under the control of its natural promoter [CETP transgenic mice, B6.CBA-Tg(CETP)5203Tall/J] were obtained from The Jackson Laboratory (Bar Harbor, ME) (Jiang et al., 1992). CETPtg-LDLR(−/−) mice were generated by breeding CETPtg mice on a C57BL/6J background (The Jackson Laboratory) with LDLR(−/−) mice on a C57BL/6J background. Starting at 8 weeks of age, male CETPtg;LDLR(−/−) mice were fed a Western diet (containing 21% milk fat and 0.15% cholesterol; Harlan Teklad, Madison, WI) containing PX20606 (5 or 25 mg/kg) or 6-ECDCA (10 mg/kg) for 16 weeks. Animals were sacrificed after a 4-h fasting.
FXR(−/−) mice were kindly obtained from Dr. Frank Gonzalez (National Institutes of Health, Bethesda, MD) under an official licensing agreement. They were bred following the originator's instructions and treated in a similar fashion as their wild-type counterparts.
Male Sprague-Dawley rats were obtained from Elevage Janvier and maintained on standard rodent chow until they were changed to the same high-fat diet as the mice. After prefeeding for 2 weeks they were treated with either the vehicle diet or PX20606 in the high-fat diet in a similar fashion as the mice.
Male and female cynomolgus monkeys (age 2–3 years; body weight 2–3 kg), which were maintained on primate growth and maintenance feed (Beijing Keao Xieli Feed Co. Ltd., Beijing, Peoples Republic of China) under standard conditions for nonhuman primate animal husbandry as provided by the Association for Assessment and Accreditation of Laboratory Animal Care guidance, were allocated to either protocol A of dose escalation (n = 6 male monkeys per group, 1 and 10 mg/kg treatment) or protocol B for continuous 28 days testing (n = 3 male and 3 female animals per dosing group). In protocol A, blood was collected at baseline and at 4, 7, and 14 days after switching to the new compound dose followed by a washout of 7 days before new randomization of the animals and the start of the new dose treatment. In protocol B, animals were randomly assigned to treatment groups, and then blood was collected at baseline and after 28 days. In both protocols, the test compounds were administered by oral gavage at the indicated doses after suspending them in standard vehicle [0.5% (w/v) polyvinylpyrrolidone and 0.1% (v/v) Tween 80, in phosphate-buffered saline, pH 7.4].
All animal work was conducted according to the national guidelines for animal care in Germany, Spain, the People's Republic of China, and the United States, all of which follow the Declaration of Helsinki for humane animal handling.
Extraction of Hepatic Lipids.
Hepatic lipids were extracted by using the method of Hara and Radin (1978) with modifications to adjust the extraction of small tissue samples (Rodríguez-Sureda and Peinado-Onsurbe, 2005). In brief, 30 mg of ground frozen liver tissue were extracted twice by using 0.5 ml of hexane/isopropanol (3:2). The organic extracts were pooled, and 0.35 ml of Na2SO4 (0.47 M) was added. Tubes were vortexed for 2 min and centrifuged at 1000g for 10 min at 4°C. The upper phase was transferred to a new preweighted tube and dried by Speed Vac centrifugation (Concentrator 5301; Eppendorf AG, Hamburg, Germany). The resulting pellet was dissolved by sonication in 0.5 ml of LPL buffer [28.75 mM 2-(N-morpholino)ethanesulfonic acid, 57.41 mM MgCl2, and 0.1% (w/v) SDS], and total triglyceride and cholesterol were measured immediately.
Plasma and Hepatic Lipids.
For the determination of triglycerides and total cholesterol from heparin plasma or hepatic extracts, standard enzymatic kits (Wako Pure Chemicals, Neuss, Germany) were used according to the manufacturer's instruction.
Lipoprotein Separation by Fast Performance Liquid Chromatography.
Serum samples were pooled within treatment groups and subjected to fast performance liquid chromatography (FPLC) analysis using a Superose 6 column on an ÄKTA purifier system (GE Healthcare, Chalfont St. Giles, Buckinghamshire, UK). In brief, the column was equilibrated with buffer containing 0.154 M NaCl and 1 mM EDTA, pH 8.0, at a flow rate of 0.5 ml/min. A 250-μl aliquot of pooled plasma sample was injected and separated at a flow rate of 0.4 ml/min, and 0.15-ml fractions were collected and analyzed for cholesterol and triglyceride content by using enzymatic assays (Wako Pure Chemicals).
En Face Analysis of Aortae.
For en face analysis, mice were euthanized, and the aortae were dissected out, opened longitudinally from heart to the iliac arteries, and stained with Sudan IV to determine lesion area as described previously (Miyazaki-Anzai et al., 2010). Images were captured by a Zeiss Axiocam-CCD (Carl Zeiss GmbH, Jena, Germany) video camera and analyzed by a single technician who was blinded to the study protocol and used AxioVision image analysis software (Carl Zeiss GmbH). The extent of lesion formation was expressed as the percentage of the total aortic surface area covered by lesions.
Separation of Lipoproteins by Nondenaturing Gradient Gel Electrophoresis and Western Blotting.
Serum samples were prestained with a lipophilic dialkylaminostyryl fluorophore and separated by 4 to 16% polyacrylamide gradient gel electrophoresis according to the instructions of the manufacturer (Invitrogen, Carlsbad, CA). Fluorescent-stained lipoproteins were detected with a Typhoon scanner (GE Healthcare) and analyzed with ImageQuant software (GE Healthcare). Separated lipoproteins were electrotransferred to polyvinylidene difluoride membranes. ApoAI-containing lipoproteins were detected by Western blot analysis with a polyclonal antibody (Millipore Corporation, Billerica, MA) against human ApoAI. Band intensity was analyzed by densitometric measurement using ImageJ software (National Institutes of Health).
Plasma Cholesterol Efflux Experiments.
Animals that had been in treatment for at least 2 weeks were subjected to a protocol that followed Kruit et al. (2005) where [14C]cholesterol was directly formulated into Intralipid (Fresenius Kabi, Bad Homburg, Germany), which allows for the monitoring of cholesterol efflux from plasma to feces. For this purpose, 1 μCi of [14C]cholesterol with 1 μCi of [3H]sitostanol was formulated in 20% (v/v) Intralipid and injected intravenously into the tail vein. The addition of [3H]sitostanol to the intravenous Intralipid bolus was necessary to normalize to the rate of sitostanol elimination, which was found to be increased upon high-fat diet feeding. Plasma samples were drawn at 4, 24, and 48 h and subjected to liquid scintillation counting as described above. Feces were collected for 0 to 24, 24 to 48, 48 to 72, and 72 to 96 h postgavage and subjected to the extraction or resuspension procedure described below. The dataset shown is a representative one from a set of three independent experiments.
Following the procedures described previously (Grundy et al., 1965; Miettinen et al., 1965), fecal samples from the cholesterol efflux experiments were pooled from one treatment group and one sampling period (0 to 24 h). Feces were lyophilized, and neutral sterols were separated after incubation at 95°C for 1 h with 1 N NaOH followed by three extractions with n-hexane. The aqueous phase was acidified with concentrated HCl to pH 1.0, and bile acids were extracted with one volume chloroform/methanol (2:1) followed by two extractions with chloroform. The organic phases containing neutral sterols or bile acids were dried down, resuspended in 100 μl of chloroform/methanol (4:1), and checked for separation purity by using authentic samples of cholesterol, coprostanol, cholic acid, chenodeoxycholic acid, and muricholic acid by thin-layer chromatography. Then a 10-μl aliquot was quantified for [3H] and [14C] radioactivity by adding 3 ml of Rotiszint (Roth, Karlsruhe, Germany) and subsequent liquid scintillation counting.
Fractional Cholesterol Uptake Experiments.
Fractional cholesterol absorption from an oral bolus was determined by using the plasma dual-isotope method according to Wang and Carey (2003) with slight modifications. In brief, mice were pretreated with FXR agonists in a high-fat diet or vehicle for 2 weeks before subjecting them to the plasma dual-isotope procedure. For the plasma dual-isotope method, the drug-pretreated animals were first gavaged with 2 μCi of [3H]cholesterol dissolved in a 150-μl lipid bolus as described above, followed by an injection with 0.5 μCi of [14C]cholesterol dissolved in 20% Intralipid into the tail vein. Plasma was collected at 4, 24, 48, and 72 h, and feces were collected for 0 to 24, 24 to 48, 48 to 72, and 72 to 96 h postgavage. Plasma samples from each animal were diluted as 10 μl in 3 ml of Rotiszint (Roth) and subjected to standard liquid scintillation counting. Fecal samples were analyzed following the procedure described below.
The ratio of orally versus intravenously administered differently labeled cholesterol was determined in mouse plasma 72 h postadministration and converted into the percentage of cholesterol absorbed by this formula: percentage of cholesterol absorbed = [percentage of oral [3H]cholesterol dose × 100/ml plasma]/[percentage of intravenous [14C]cholesterol dose/ml plasma].
Tissue Gene Expression Analysis by Quantitative Real Time PCR (qRT-PCR).
Frozen tissues were ground to fine powder under liquid nitrogen with a mortar. Total RNA from 25 mg of powdered tissue was isolated by using RNAzol RT Reagent (Molecular Research Center, Cincinnati, OH) and an RNA Isolation Kit (RNeasy 96; QIAGEN GmbH, Hilden, Germany) following the manufacturer's instructions. The concentration and integrity of the total RNA was assessed by gel electrophoresis. cDNAs were synthesized from 0.5 μg of total RNA by using Superscript II reverse transcriptase (Invitrogen) and primed with 50 pmol of random nonamers. Quantitative real time PCR was performed and analyzed by using Absolute QPCR Rox Mix (Invitrogen) on a 384-format ABI 7900HT Sequence Detection System (Applied Biosystems, Foster City, CA). Primer and probe sequences are listed in Supplemental Table 1. All samples were run in duplicate from the same RT reaction. Gene expression was expressed in arbitrary units and normalized relative to the mRNA of the housekeeping gene TATA box binding protein.
Homogenous Time-Resolved Fluorescent Resonance Energy Transfer Assays and GAL4-Hybrid Cellular Reporter Assays for Mouse, Rat, and Human FXR.
To assess the FXR agonist activity by time-resolved fluorescence resonance energy transfer (FRET) a biotinylated SRC-1 peptide b-CPSSHSSLTERHKILHRLLQEGSPS-COOH (0.4 μM), 200 ng of streptavidine-allophyco-cyanine, and 6 ng of europium labeled-anti-GST were mixed with either purified human FXRaa187–472 fused to GST, mouse FXRaa202–484 fused to GST, or rat FXRaa203–469 fused to GST (2.5 ng) in 25-μl assay buffer (20 mM Tris/HCl pH 7.5, 5 mM MgCl2, 60 mM KCl, 1 mM dithiothreitol, and 0.9 g/l bovine serum albumin). After 60 min of incubation, the FRET efficiency was determined by using a spectrofluorometer (Envision; PerkinElmer Life and Analytical Sciences, Waltham, MA). Measurements were done in duplicate. FRET values are given in nanometers, and the maximum efficacy of the compound relative to the control FXR agonist GW4064 was determined.
As cellular assay for FXR agonist activity, a GAL4-FXR fusion construct reporter system was used. Human embryonic kidney 293 cells were transiently transfected with a reporter plasmid pFRluc (Agilent Technologies, Santa Clara, CA) encoding a GAL4 promoter-driven firefly luciferase and a given FXR-LBD (human FXRaa187–472, mouse FXRaa202–484, or rat FXRaa203–469) expression plasmid fused C-terminally to a GAL4 DNA binding domain under transcriptional control of the cytomegalovirus promoter in pCMV-BD (Agilent Technologies). In addition, a second reporter plasmid pRL-TK constitutively expressing the Renilla luciferase under control of a thymidine kinase promoter was included as an internal control. Four hours after transfection cells were treated with serial dilutions of the test FXR agonists. Sixteen hours after treatment cells were lysed with Passive Lysis Buffer (Promega, Madison, WI), and firefly and Renilla luciferase activities were measured sequentially in the same cell extract by using a Dual-Light-Luciferase-Assay system and microplate luminometer (LUMIStar OPTIMA; BMG Labtech GmbH, Offenburg, Germany). Relative luciferase activity was calculated and reported as the ratio between firefly luciferase and Renilla luciferase activity. EC50 values were calculated from at least three independent experiments.
Statistical significance was analyzed by using the unpaired Student's t test for unequal variance. Values are expressed as mean ± S.E.M.
Different FXR Agonists Have Distinct Effects on Plasma Cholesterol Lowering and the Resulting Lipoprotein Profile.
Five synthetic FXR agonists representing three separate chemical classes (Fig. 1) were evaluated for their plasma lipid-lowering capabilities in C57BL/6J mice on a high-fat diet regimen. PX20350 (Abel et al., 2010) and PX20606 are structurally related to GW4064, whereas FXR-450 represents a structurally diverse chemotype of a nonsteroidal, fully synthetic FXR agonist. 6-ECDCA is closely related to the natural FXR ligand chenodeoxycholic acid albeit with higher potency at the human FXR compared with its natural counterpart (Tables 1 and 2). FXR-450, PX20350, and PX20606 treatment for 2 weeks resulted in significant plasma total cholesterol and triglyceride lowering at both doses used (10 and 30 mg/kg) (Fig. 2, A and B). The FXR agonists GW4064 and 6-ECDCA only displayed a trend toward TC and TG lowering, even at the highest dose of 30 mg/kg (Fig. 2, A and B). FPLC plasma lipoprotein fractionation further confirmed these results. PX20350, PX20606, and FXR-450 lowered cholesterol in all three major fractions (HDLc, LDLc, and VLDLc) (Fig. 2C). It is noteworthy that VLDLc and LDLc were reduced to levels close to the detection limit, whereas HDLc was reduced by approximately 50%. In contrast, GW4064 and 6-ECDCA treatment caused no change in LDLc levels, a trend toward increasing VLDLc, and in the case of 6-ECDCA a small increase in HDLc.
Cholesterol Lowering by PX20606 and FXR-450 Versus GW4064 and 6-ECDCA Correlated with Their Potency in Inducing Transhepatic Plasma Cholesterol Efflux.
To reveal the underlying cause for the observed difference in plasma cholesterol lowering, we used different radioisotope protocols to assess the impact of the FXR agonists on cholesterol efflux from plasma into feces and on the intestinal absorption of cholesterol from the diet. C57BL/6J mice fed either normal chow or a high-fat diet were treated with FXR agonists PX20606, FXR-450, GW4064, and 6-ECDCA (30 mg/kg) or vehicle for 2 weeks and subsequently intravenously coinjected with [3H]sitostanol and [14C]cholesterol. Treatment with PX20606 and FXR-450 resulted in significantly decreased plasma [14C]cholesterol levels, whereas 6-ECDCA and GW4064 showed only minimal effects on plasma [14C]cholesterol (Fig. 3A). The separation of plasma samples from this experiment by FPLC and subsequent analysis of [14C]cholesterol radio counts showed that PX20606 treatment resulted in a decrease in HDL [14C]cholesterol compared with vehicle (Fig. 3C). The difference in HDL [14C]cholesterol radio counts between PX20606 and vehicle was maintained at 24 h postinjection, albeit at a lower overall level, indicating that the effects persist for at least 1 full day.
It is noteworthy that the cholesterol-lowering effect of PX20606 treatment was more pronounced in mice fed normal chow as opposed to a high-fat diet. To verify that the effects observed were FXR-specific, we dosed FXR(−/−) mice fed a high-fat diet with PX20606 or vehicle. As expected, vehicle-treated FXR(−/−) mice displayed higher plasma [14C]cholesterol levels compared with wild-type animals (Sinal et al., 2000). In contrast, PX20606 treatment had no effect on plasma cholesterol levels in FXR(−/−) mice, indicating that the cholesterol-lowering activity of PX20606 depends on the presence of a functional FXR (Fig. 3A).
In line with the lowering of [14C]cholesterol in plasma, PX20606 and FXR-450 treatment also caused a significant increase in fecal output of [14C]sterols indicated by a higher ratio of [14C] radioactivity to [3H] radioactivity in feces (Fig. 3B). The increase in fecal [14C] counts was found to be only in the neutral sterol fraction and not in the acidic sterol fraction, which was unchanged with regard to radio counts between PX20606 and vehicle treatment (data not shown). The effect of 6-ECDCA but not GW4064 on fecal [14C]-neutral sterol output was significant and higher than expected given its marginal effect on plasma counts but was substantially lower than with FXR-450 or PX20606. Consistent with the supposed FXR dependence, fecal [14C]sterols were reduced in FXR-deficient animals compared with wild-type mice, and no change upon PX20606 treatment could be detected (Fig. 3B).
Because an increase in transhepatic cholesterol efflux can result only in significant plasma cholesterol lowering when intestinal reuptake is efficiently prevented, we tested for a change in intestinal cholesterol absorption by a different dual-isotope method. In this independent dual-isotope experiment we combined an intravenous injection of [14C]cholesterol with an oral gavage of [3H]cholesterol to determine the fraction of cholesterol absorbed, a technique known as the plasma dual-isotope method (see Materials and Methods and references therein). Before administering the radiolabeled cholesterol by oral gavage, C57BL/6J mice fed either normal chow or a high-fat diet were treated with FXR agonists PX20606, FXR-450, GW4064, and 6-ECDCA (30 mg/kg) or vehicle for 2 weeks.
All FXR agonists tested caused a reduction of fractional cholesterol absorption in wild-type animals fed normal chow or a high-fat diet (Fig. 3D). It is noteworthy that plasma [3H]cholesterol counts reflecting foremost the effect of intestinal cholesterol uptake show marked differences between the more potent FXR agonist PX20606 and FXR-450 and the weaker GW4064 and 6-ECDCA; however, all of these values were significantly below the vehicle control (Fig. 3E). The plasma [14C]cholesterol data, however, reflect the transhepatic efflux rate overlaid to some extent by intestinal reuptake (Fig. 3E). The fact that GW4064 and 6-ECDCA display some effect, although lower than with PX20606 or FXR-450, on [3H]cholesterol lowering but not on [14C] count lowering indicates that in the former the intestinal uptake is dominant, whereas in the latter the transhepatic cholesterol efflux is prevailing.
Overall, the combined data from both sets of different dual isotope experiments indicate that all four FXR agonists reduce intestinal cholesterol absorption, but only PX20606 and FXR-450 significantly induce transhepatic cholesterol efflux from plasma into feces.
The Different Effects of FXR Agonists on Transhepatic Cholesterol Efflux Can Be Linked to Differential Activation of FXR Target Genes in the Liver.
To investigate the influence of FXR activation on the expression of FXR-controlled genes that are involved in transhepatic cholesterol efflux we quantified the mRNAs for scavenger receptor BI (SR-BI), endothelial lipase (EL), ATP-binding cassette (ABC) G5, and ABCB4 (synonym: Mdr2) by qRT-PCR. C57BL/6J wild-type or FXR(−/−) mice were fed a high-fat diet for 2 weeks before administration of a single dose of 30 mg/kg FXR-450, PX20606, GW4064, and 6-ECDCA. Although all FXR agonists from the panel were able to regulate the well established FXR direct target genes in liver and ileum (Fig. 4), we observed major differences in the regulation of genes involved in transhepatic cholesterol transport. The expression of small heterodimeric partner (SHP) and bile salt export pump (BSEP), two direct FXR target genes, were significantly increased after treatment with FXR-450, PX20606, and GW4064 (Fig. 4A). Although not significant, 6-ECDCA treatment was associated with a low induction of these genes compared with vehicle. However, only PX20606 and FXR-450 significantly up-regulated mRNAs for SR-BI, EL, and ABCB4 mRNA, whereas ABCG5 expression was induced by PX20606, FXR-450, and GW4064. FXR target genes FGF15 and ileal fatty acid binding protein (IBAP) were significantly increased by all four FXR agonists in the ileum (Fig. 4B). All compounds tested induced no change in the expression of SHP, SR-BI, EL, BSEP, ABCG5, or ABCB4 in the livers or intestines of FXR(−/−) mice, which proves that the transcriptional effects on these genes, which are crucial for cholesterol homeostasis, are strictly FXR-mediated.
The rat has been described as a species that is insensitive to FXR-induced plasma cholesterol lowering (Willson et al., 2001; Evans et al., 2009). Evans et al. attributed this species difference to the inability of the rat FXR to alter expression of SR-BI and EL in the liver. We aimed to reproduce this effect by administering PX20606 at 30 mg/kg for 2 weeks in Sprague-Dawley rats. The plasma total cholesterol levels were not significantly altered, which was paralleled by a missing induction of SR-BI, ABCB4, or EL mRNAs in the livers of these rats compared with the vehicle control (Supplemental Fig. 1), thus confirming the published results and providing further evidence that the plasma cholesterol lowering by FXR agonists is linked to the induction of these genes.
FXR Agonists PX20606 and 6-ECDCA Reduce Atherosclerotic Plaque Sizes in CETPtg-LDLR(−/−) Mice Concomitant with an Apparent HDL and Total Cholesterol Lowering.
Next, we wanted to determine whether pharmacological FXR activation and the associated cholesterol- and HDLc-lowering effects would translate into proatherosclerotic or antiatherosclerotic effects. We chose the CETP transgenic LDLR(−/−) mouse as a model system, because these mice become highly hypercholesterolemic when maintained on a Western high-fat diet. The loss of function of the LDL receptor should not affect the liver clearance of HDLc if it was controlled mainly by the expression of SR-BI, EL, or ABCB4. Furthermore, the expression of human CETP in these animals enables the cholesteryl ester transfer between LDL and HDL, which generates a cholesterol lipoprotein distribution closer to the human profile (Agellon et al., 1991). Mice were administered PX20606 (5 or 25 mg/kg) or 6-ECDCA (10 mg/kg) for 12 weeks. After sacrifice the aortae were explanted, mounted, stained with Sudan IV, and subjected to quantitative image analysis to determine the extent of atherosclerotic plaque deposition. Plasma clinical chemistry including HDL cholesterol levels, were analyzed as well as liver total cholesterol and triglycerides. PX20606 showed a marked dose-dependent reduction of TC, whereas 10 mg/kg 6-ECDCA treatment showed significant TC lowering, albeit to a lesser extent than the 5 mg/kg dose of PX20606 (Table 3). Plasma triglycerides were significantly lowered in all three FXR agonist treatment groups. Plasma glucose and calcium were not significantly affected (data not shown). Liver cholesterol and triglycerides both were significantly reduced by PX20606 and 6-ECDCA treatment (Fig. 5A). Whereas both doses of PX20606 treatment yielded a significant reduction of the total percentage of Sudan IV-positive plaque area, 6-ECDCA treatment did not translate into a significant reduction despite the moderate reduction of total plasma cholesterol (Fig. 5B). This difference between 6-ECDCA- and PX20606-treated animals was also seen comparing HDLc levels. PX20606 at 25 mg/kg reduced HDLc by more than 60%, whereas in 6-ECDCA-treated animals HDLc levels were not significantly altered (Fig. 5C). It is noteworthy that the observed differences between PX20606 and 6-ECDCA treatment also extend to the level of FXR target genes. The mRNA for SHP was induced by 10 mg/kg 6-ECDCA and 5 mg/kg PX20606 to a comparable extent, but the genes that are presumably involved in the transhepatic cholesterol efflux, SR-BI, ABCB4, and EL, are up-regulated only by both doses of PX20606 but not by 10 mg/kg 6-ECDCA (Fig. 5D).
Both FXR Agonists, 6-ECDCA and PX20606, Lower Total Cholesterol in Normolipidemic Cynomolgus Monkeys in a Dose-Dependent Fashion; PX20606 Reduces HDL2 but not HDL3 at High Doses without a Total Reduction of ApoAI in Plasma.
Given that CETPtg mice still have some differences in cholesterol metabolism to humans, we sought to test for the cholesterol-lowering effect of the FXR agonist PX20606 and the clinical stage comparator 6-ECDCA in cynomolgus monkeys, a nonhuman primate species that has a lipoprotein profile close to humans. Animals were kept on standard diet and treated for 7 days with a twice-daily oral gavage of 1 or 10 mg/kg PX20606 or 6-ECDCA. Doses as low as 1 and 10 mg/kg b.i.d. of PX20606 and 6-ECDCA, respectively, led to a reduction of total plasma cholesterol (Fig. 6A). To determine the effect of more pronounced FXR activation, we treated cynomolgus monkeys with 20, 50, or 100 mg/kg PX20606 for 28 days, which resulted in a dose-dependent lowering of plasma total cholesterol (Fig. 6D). In the FPLC lipoprotein fractionation of plasma samples derived from this study HDLc levels were substantially reduced, whereas LDLc levels were not altered significantly (Fig. 6B). To elucidate the apparent changes in the main HDL subclasses further monkey plasma samples were subjected to native gradient gel electrophoresis with a subsequent Western blot for ApoAI. Fluorescence staining of neutral lipids after one-dimensional gel electrophoresis showed a dose-dependent reduction of the cholesteryl ester-rich HDL2 subclass and a redistribution of lipid mass from HDL2 to HDL3 in monkeys from both genders for PX20606 treatment (Fig. 6C). Quantification of the ApoAI protein expression on this Western blot provided a strong indication that PX20606 treatment did not lead to a lowering of the total amount of ApoAI protein but to a redistribution among the various HDL subfractions (Fig. 6, E and F). To yield data from an alternative separation method, HDL2 and HDL3 particles were separated by ultracentrifugation (Chapman et al., 1981). Subsequent quantification of cholesterol and ApoAI by an enzymatic assay and an enzyme-linked immunosorbent assay, respectively, indicated no lowering of ApoAI levels in either subfraction of HDL upon treatment with PX20606 (Supplemental Fig. 2B), whereas the cholesterol content was significantly lowered only in the HDL2-containing fractions. Thus, treatment with PX20606 decreased the cholesterol-to-ApoAI ratio (Supplemental Fig. 2C) specifically in HDL2 particles.
The present study aimed at answering the following questions: 1) Do all FXR agonists show the same pharmacological effects on cholesterol lowering? 2) Which general mechanism is mainly responsible for the potent HDL-cholesterol-lowering effects, and is it conserved throughout different species? 3) Will the HDL lowering result in pro- or anti-atherosclerotic effects in sum?
To address the first question we studied the effects of multiple dose administrations of four different FXR agonists of different chemotypes, GW4064, 6-ECDCA, FXR-450, and the newly synthesized PX20606, on cholesterol plasma levels and its intestinal uptake and efflux from plasma into feces. It turned out that the four FXR agonists under investigation fell into two main classes. Whereas FXR-450 and PX20606 yielded potent reductions in total plasma cholesterol, GW4064 and 6-ECDCA showed only minor effects. Two different sets of dual-radioisotope experiments showed that the two different classes of FXR agonists vary in their capabilities to induce transhepatic cholesterol efflux from the plasma into feces, whereas all reduced intestinal cholesterol uptake. The missing effects in the FXR(−/−) mice upon PX20606 treatment indicate the strict dependence of this effect on the presence of FXR.
Because published data pointed to an up-regulation of those genes that mediate the clearance of cholesteryl esters from HDL and their excretion into feces via the bile, we investigated the expression of the mRNAs of relevant genes known to be controlled by FXR. SR-BI, the receptor to facilitate HDL cholesterol ester efflux (Zhang et al., 2005; Chao et al., 2010), endothelial lipase, an enzyme known to hydrolyze phospholipids in HDL which facilitates HDL remodeling and cholesterol clearance (Evans et al., 2009), and ABCG5 and ABCB4, the former known as a direct cholesterol efflux transporter from liver into the bile and the latter described as a canalicular membrane efflux pump for phospholipids that drags cholesterol molecules along in its efflux mechanism (Huang et al., 2003; Yu et al., 2005). Indeed, all of these genes were up-regulated upon treatment with PX20606 and FXR-450 but not with GW4064 and 6-ECDCA in the liver. The induction of these genes was absent in FXR(−/−) mice. Thus, the differential capabilities of PX20606 and FXR-450 in inducing these cholesterol clearance-relevant genes parallel their capabilities in inducing transhepatic cholesterol efflux. Why PX20606 and GW4064 show these different pharmacological effects, despite very similar chemical structures and similar potencies at the human FXR in cellular and biochemical assays, might be explained by GW4064 reaching only 13% of the liver levels of PX20606. However, we cannot exclude the possibility that GW4064 is a gene-selective FXR agonist (selective bile acid receptor modulator), able to up-regulate the target genes SHP and BSEP in the liver but unable to induce the cholesterol-lowering relevant genes SR-B1, EL, and ABCB4. The reason for the missing in vivo potency of 6-ECDCA is probably a different one, because this compound and its equally active tauro conjugate metabolite achieve similar levels in the liver compared with PX20606. The weaker in vitro potency at the murine compared with human FXR might explain why 6-ECDCA does not perform in this species (see Supplemental Table 2). FXR-450, despite rather low levels, is at least one magnitude more potent in in vitro assays compared with the other FXR agonists, which explains the reduced need for liver exposure to exert similar effects as PX20606.
The fact that PX20606 induces SHP but not SR-BI, EL, and ABCB4 in rat livers, which goes along with no significant effect on plasma cholesterol, further highlights the role of these selectively induced cholesterol-relevant genes as a mechanistic basis for the transhepatic cholesterol efflux effect and points to the species differences.
Whereas SHP, BSEP, and ABCB4 are described in the literature as direct transcriptional target genes of FXR (Goodwin et al., 2000; Lu et al., 2000; Ananthanarayanan et al., 2001; Huang et al., 2003), EL and ABCG5 seem to be indirectly regulated by FXR in the liver (Evans et al., 2009; Zhang et al., 2010). Zhang et al. claim that the hepatic induction of SR-BI depends on an increased level of hepatic nuclear factor 4; but a recent publication showed that the murine and the human SR-BI genes have a functional FXR response element in an intron region (Li et al., 2012). In the latter study the missing up-regulation of SR-BI in FXR(−/−) mice led to higher HDLc levels compared with wild-type mice, which was consistent with our findings. Independent from the FXR-dependent regulation it has been demonstrated in several studies that SR-BI overexpression has a beneficial antiatherosclerotic effect in different mouse models despite an apparent lowering of plasma HDLc (Kozarsky et al., 1997, 2000; Arai et al., 1999; Zhang et al., 2005). Functional up-regulation of EL is clearly associated with decreased HDL levels in humans (Badellino et al., 2006), and EL is a key enzyme involved in the uptake of cholesteryl esters via SR-BI by the liver and in the remodeling of cholesteryl ester-rich HDL into smaller HDL particles (Nijstad et al., 2009). In this sense the co-up-regulation of SR-BI and EL upon FXR activation by potent agonists synergize in their effects on increasing HDL cholesterol clearance by the liver and excretion into feces.
Therefore, it is important to study whether the FXR-mediated changes in HDL metabolism are in sum pro- or anti-atherogenic, and we decided to use the CETP-transgenic LDLR(−/−) mouse as a model system. These mice are highly hypercholesterolemic because of the absence of the LDL receptor, and they develop atherosclerotic lesions when placed on a high-fat, high-cholesterol (Western) diet. Yet, they display a lipoprotein pattern that is closer to the human profile because of the presence of a functional CETP. Because the presumed mode of action of FXR agonists, induction of transhepatic cholesterol efflux, would not require a functional LDL receptor, this model seems to be well suited for studying the effects of FXR agonists on cholesterol lowering and the resulting net effect on atherosclerotic plaque deposition. Indeed, PX20606 dose-dependently and potently reduced total plasma cholesterol in this model. Here, 6-ECDCA also showed some reduction in total cholesterol as opposed to the wild-type mice on HFD, but it failed to demonstrate significant HDL lowering and did not have a positive effect on atherosclerotic plaque formation. PX20606, however, potently reduced aortic plaques in a dose-dependent manner, which correlated with its total cholesterol-lowering capabilities.
Nonhuman primates such as the cynomolgus monkey serve as model organisms for testing cholesterol-lowering drugs such as statins in a species as close as possible to humans. Echoing the potent cholesterol-lowering effects in wild-type mice on a high-fat diet, doses of PX20606 and 6-ECDCA as low as 2 × 1 mg/kg/day significantly lowered total cholesterol in this species, and 100 mg/kg/day even yielded a reduction of −50% compared with baseline, an unprecedented pharmacological effect of a single agent in this species (compare with Liang et al., 2012). In addition, in this species the bulk of the lowering came from a reduction in HDL cholesterol, which was evidenced by FPLC separation. The fine-grained analysis of different HDL subfractions (Fig. 6; Supplemental Fig. 2) found that the cholesterol lowering by PX20606 can be observed only within the HDL2 subclass and not within the HDL3 subclass. ApoAI levels were nearly unchanged among all HDL fractions, which is in line with similar published findings (Gardès et al., 2011).
Thus, these data suggest that FXR activation by PX20606 in monkeys up-regulates similar pathways as in the C57BL/6J mice, resulting in cholesteryl ester clearance from HDL2 particles into the liver and subsequent excretion into feces. A similar pattern of HDL2-restricted cholesterol depletion was observed upon transgenic or viral overexpression of SR-BI in mice when infused with purified human HDL2 as a substrate (Ueda et al., 1999; Webb et al., 2002, 2004). In those studies it was also reported that the resulting remnants seem to keep their ApoAI and re-enter the pool of HDL3 particles, potentially becoming subject to “refilling” with cholesteryl esters by the action of CETP, lecithin-cholesterol acyltransferase, and phospholipid transfer protein. In this context it needs to be kept in mind that antiatherogenic properties of circulating HDL are not only determined by the sheer content of cholesterol or ApoAI but also by its functional cholesterol acceptor and antioxidant capabilities (Camont et al., 2011; Khera at al., 2011; Rye and Barter, 2012). Thus, whether the HDL3 particles, including the remnants from HDL2 depletion in PX20606-treated animals, really maintain their functional cholesterol scavenger and antioxidant capabilities needs to be elucidated mechanistically in more detail in follow-up studies.
In summary, we show in the present study that different FXR agonists induce different pharmacological effects on liver gene expression and HDL-derived cholesterol clearance into feces. Moreover, the differences in hepatic gene induction between PX20606 and 6-ECDCA seem to account for the differences in aortic plaque formation in the CETPtg LDLR(−/−) mouse model. A similar HDL cholesterol reduction was also observed in normolipidemic cynomolgus monkeys, suggesting that the basic mechanisms of FXR-mediated HDL cholesterol clearance are conserved in mice and nonhuman primates with the exception of the rat. The depletion of HDL2 cholesterol with unchanged absolute levels of ApoAI induced by a synthetic FXR agonist is a novel and exciting mechanism that might give rise to further mechanistic studies in other model organisms as well as in humans to unravel the functional impact of these HDL modifications on atherosclerosis and other cardiovascular diseases. FXR activation thus might be a means of increasing the overall flux through the reverse cholesterol pathway by opening the drainage from HDL particles into feces, a mechanism so far not being addressed pharmacologically (Degoma and Rader, 2011).
Participated in research design: Burnet, Levi, Schmitz, Miyazaki, and Kremoser.
Conducted experiments: Hambruch, Miyazaki-Anzai, Hahn, Matysik, Boettcher, Schlüter, Kinzel, Krol, Deuschle, Burnet, and Miyazaki.
Performed data analysis: Hambruch, Miyazaki-Anzai, Perović-Ottstadt, Burnet, Miyazaki, and Kremoser.
Wrote or contributed to the writing of the manuscript: Hambruch, Matysik, Perović-Ottstadt, Deuschle, Burnet, Miyazaki, and Kremoser.
We thank Dr. Jürgen Lehmann for careful and critical reading of our manuscript.
This work was supported in part by the German Federal Ministry of Education and Research [Grant 0315470].
Article, publication date, and citation information can be found at http://jpet.aspetjournals.org.
- farnesoid X receptor
- ATP-binding cassette
- bile salt export pump
- cholesteryl ester transfer protein transgenic
- 6-ethyl chenodeoxycholic acid
- endothelial lipase
- fibroblast growth factor
- fast performance liquid chromatography
- fluorescence resonance energy transfer
- isopropyl 3-(3,4-difluorobenzoyl)-1,1-dimethyl-1,2,3,6-tetrahydroazepino[4,5-b]indole-5-carboxylate
- glutathione transferase
- 3-[2-[2-chloro-4-[[3-(2,6-dichlorophenyl)-5-(1-methylethyl)-4-isoxazolyl]methoxy]phenyl]ethenyl]benzoic acid
- high-density lipoprotein
- HDL cholesterol
- high-fat diet
- ligand binding domain
- low-density lipoprotein
- LDL cholesterol
- LDL receptor
- 4-[[[6-[[5-cyclopropyl-3-(2,6-dichlorophenyl)4-isoxazolyl]methoxy]-2-(trifluoromethyl)-3-pyridinyl]methylamino]methyl]benzoic acid
- 4-[2-[2-chloro-4-[[5-cyclopropyl-3-(2,6-dichlorophenyl)-4-isoxazolyl]methoxy]phenyl]cyclopropyl]benzoic acid
- small heterodimeric partner
- scavenger-receptor B1
- plasma total cholesterol
- total triglycerides
- very-low-density lipoprotein cholesterol
- wild type.
- Received May 23, 2012.
- Accepted August 22, 2012.
- Copyright © 2012 by The American Society for Pharmacology and Experimental Therapeutics