University of Groningen Cholesterol, bile acid and triglyceride metabolism intertwined Schonewille, Marleen IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below. Document Version Publisher's PDF, also known as Version of record Publication date: 2016 Link to publication in University of Groningen/UMCG research database Citation for published version (APA): Schonewille, M. (2016). Cholesterol, bile acid and triglyceride metabolism intertwined [Groningen]: Rijksuniversiteit Groningen Copyright Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons). Take-down policy If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim. Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum. Download date: 14-07-2018
Transintestinal Cholesterol Excretion Can Drive Massive Cholesterol Elimination in Mice Marleen Schonewille, Jan Freark de Boer, Marije Boesjes, Henk Wolters, Vincent W. Bloks, Trijnie Bos, Theo H. van Dijk, Angelika Jurdzinski, Renze Boverhof, Jan M. van Deursen, Ronald P.J. Oude Elferink, Antonio Moschetta, Claus Kremoser, Henkjan J. Verkade, Folkert Kuipers, Albert K. Groen Submitted
ABSTRACT GRAPHICAL ABSTRACT Transintestinal Cholesterol Excretion (TICE) is a major pathway of cholesterol excretion that, under control conditions, accounts for about 30% of daily fecal cholesterol loss in mice. Using a panel of knock-out and transgenic mice as well as pharmacological manipulations we show that TICE is regulated by the concerted action of intestinal Farnesoid X Receptor (FXR) via its target Fibroblast Growth Factor 1/19 (FGF1/19), the cholesterol importer Niemann-Pick C1-Like 1 (NPC1L1), and the sterol exporting heterodimer ABCG/G8. Data show that a combination of intestinal FXR activation and blockade of NPC1L1 activity stimulates this pathway to such an extent that mice excrete up to 60% of their total cholesterol pool daily. Intriguingly, hydrophilicity of the bile salt pool, controlled by FXR and FGF1/19, is an important determinant of TICE. Translation of these results to humans may offer new modalities for prevention of cardiovascular disease. Transintestinal Cholesterol Excretion Can Drive Massive Cholesterol Elimination in Mice Introduction C ardiovascular disease (CVD) is still the leading cause of death worldwide. A main driving force in the development of CVD is atherosclerosis, which progresses in a complex interplay between plasma lipoprotein metabolism and inflammatory processes. 186 The concentration of LDL-cholesterol is by now an undisputed pathogenic factor in CVD and all successful therapies today aiming at preventing or reducing progress of atherosclerosis, target the concentration of LDL-cholesterol in blood mostly by increasing the activity of the LDL-receptor. 187 A recent study showed that heterozygous carriers of inactivating mutations in the gene encoding the intestinal cholesterol importer Niemann-Pick C1-Like 1 (NPC1L1) had a 3% decreased incidence of coronary heart disease. 188 Intriguingly, the reduction of plasma LDL cholesterol levels in those carriers was barely significant compared to controls. This suggests that, in addition to lowering plasma LDL cholesterol levels, increasing cholesterol turnover in the body, e.g. by modulating intestinal cholesterol absorption efficiency, might have beneficial effects on the incidence of CVD. We and others showed recently that, in addition to the classical biliary sterol secretion pathway, cholesterol turnover can also occur by direct transport from blood into the intestine followed by fecal excretion. This pathway is known as transintestinal cholesterol excretion or TICE. 24,189 Under control conditions this pathway accounts for approximately 30% of fecal cholesterol output in mice. Despite its importance for control of cholesterol homeostasis the molecular mechanism and control of TICE have not yet been elucidated. We have demonstrated that cholesterol excretion via the TICE pathway can be stimulated by LXR-agonism and plant sterols, indicating a regulatory role for the sterol transporting heterodimer Abcg/ Abcg8. 2,27 Le May et al. suggested a role for the multi drug transporters Abcb1a and b and proposed regulatory activities by PCSK9. 31 More recently flavin monooxygenase 3 (FMO3) has been implicated in TICE regulation. 98 Since TICE is a secretory pathway, activity of the cholesterol importer NPC1L1 could be expected to modulate the net amount of cholesterol excreted via this route. Ezetimibe is a specific inhibitor of NPC1L1 190 and treatment of experimental animals but also humans with ezetimibe strongly increases fecal neutral sterol (i.e. cholesterol and its derivatives) excretion. 28,191,192 Few groups, however, have actually measured whether the numbers add up, i.e. whether the increase in fecal neutral sterols (FNS) can be accounted for by inhibition of cholesterol absorption. Cholesterol balance studies to quantify the contribution of the various pathways during ezetimibe treatment have not yet been reported. 189,193,194 Cholesterol absorption is not only controlled by 6
NPC1L1, the presence of bile salts (BS) in the intestinal lumen is a conditio sine qua non of absorption to occur. 19 It is generally accepted that water insoluble products of dietary lipid digestion and cholesterol requires the formation of mixed micelles containing bile salts to be transported through the unstirred water layer towards the enterocytes. Of note, the potential contribution of a cholesterol export pathway has never been taken into account when delineating the regulatory impact of bile salts on cholesterol absorption. A considerable number of different bile salt species has been identified across the animal kingdom. The species differ in their physical chemical properties which directly influence their lipid solubilizing but also signaling properties (see Kuipers et al. 2 and Li et al. 64 for recent reviews). The Farnesoid X receptor (FXR/NR1H4) is the central actor controlling bile salt synthesis and it is expressed mainly in the small intestine and liver. 76,77,196 Intestinal FXR has been shown to play a pivotal role in regulating the bile salt pool via the secretion of Fibroblast Growth Factor 1 (FGF1;FGF19 in humans) that inhibits bile salt synthesis in the liver. 197 In the present study, we reveal the complex interaction of key players in regulation of cholesterol absorption and its antagonist TICE. We show that NPC1L1 inhibition combined with FXR agonism greatly enhances the flux of cholesterol through the TICE pathway in a way that largely depends on ABCG/G8 function in the intestine. Furthermore, our data strongly indicate that the physical chemical properties of the bile salt pool control the rate of cholesterol removal from the body via TICE with hydrophilic bile salts acting as potent stimulators of this route. This implies that intestinal bile salts not only facilitate the uptake of dietary and biliary cholesterol by the enterocytes through solubilization in mixed micelles where hydrophobic bile salts play an important role (i.e., hydrophobic > hydrophilic), but also fulfil a so far unknown function in controlling the excretion of plasmaderived cholesterol from the enterocytes into the intestinal lumen where hydrophilic bile salts are more important. Hence, the balance between hydrophobic and hydrophilic bile salt species in the circulating bile salt pool represents a so far underestimated determinant of whole body cholesterol turnover that provides leads for the design of new strategies for prevention of cardiovascular diseases. Materials & Methods Animal experiments Male Age-matched wild-type C7Bl/6J mice, intestinal-specific FXR knock-out (ifxr KO) 213 and their wild-type (WT) littermates, intestinespecific FXR TG mice (ifxr TG; see supplemental materials and methods) and their non-transgenic littermates (FXR KO) 214 and AbcG8 knock-out mice and their WT littermates, 21 were housed in a light (12:12)- and temperaturecontrolled (21 C) facility and received laboratory chow (RMH-B, Hope Farms, Woerden, The Netherlands) ad libitum. When stated, mice received 10 mg/kg/day PX (Phenex Pharmaceuticals, AG, Heidelberg, Germany), 198,216 0.00% Ezetimibe ( Ezetrol; Pharmacy UMCG, Groningen, The Netherlands), or both of the compounds. Additionally, Abcg8 knock-out mice were injected with 1 x 10 11 particles/mouse of an Abcg8-encoding adenovirus or control virus containing and empty expression cassette (AdNull). Experiments were performed between day 3 and day after injection of the respective adenoviruses, a time frame when high and stable expression is achieved. Furthermore, ifxr KO mice were injected (i.p) with 1 mg/kg/day FGF19 (generous gift of dr. J. Nguyen, MGM Biopharmaceuticals San Francisco, USA) for 8 consecutive days when indicated (Genscript, Piscataway NJ, USA; Cat. No.: Z02738-1). Experimental procedures To assess fractional cholesterol absorption, mice received an intravenous dose of 0.3 mg (0.73 µmol) cholesterol-d (Medical Isotopes, Inc, Pelham, NH, USA) dissolved in Intralipid (20%, Fresenius Kabi, Den Bosch, The Netherlands) and an oral dose of 0.6 mg (1.3 µmol) cholesterol-d7 (Cambridge Isotope Laboratories, Inc, Andover, MA, USA) dissolved in medium-chain triglyceride oil (Pharmacy UMCG, Groningen, The Netherlands) 10 days prior to the end of the experiment. Mice received 13 C-acetate via the drinking water 3 days prior to the end of the experiment in order to determine cholesterol synthesis. Blood spots were collected from the tail at 0, 3, 6 and 12 hours and subsequently every 24 hours after administration of the stably labelled cholesterol. Bloodspots were taken at 9 AM and 6 PM during the 13 C-acetate. At the end of the experiment, mice were anesthetized by intraperitoneal injection with Hypnorm (1ml/kg) (Janssen Pharmaceuticals, Tilburg, The Netherlands) and Diazepam (10mg/kg) (Actavis, Baarn, The Netherlands). After ligation of the bile duct, the gallbladder was cannulated. Bile collected during the initial min after cannulation was disposed to ensure no gallbladder bile was collected. Subsequently, hepatic bile was collected for 30 min. 217 During the collection period, body temperature was stabilized in a humidified incubator. After bile had been collected, mice were sacrificed by cardiac puncture and tissues were rapidly excised. In a separate series of experiments Wistar rats (Sulzfeld, Germany) were surgically equipped with cannulas in bile duct and intestine. The outlets of the cannulas were fixed on the head of the animals. After surgery, the animals were allowed to recover for days. During this period, the cannulas from the bile duct and intestine were connected on the head of the animals, essentially restoring the enterohepatic circulation via an extended bile duct. After having obtained feces for determination of FNS excretion, the enterohepatic circulation was interrupted and all bile from the rats was quantitatively collected in a tube located outside of the cage. From the moment of interruption, model bile was infused into the intestinal cannula going to the intestine. All animals received identical model bile composed of 1 mm taurocholic acid and 2. mm phosphatidylcholine at a rate of 1.2 ml/hour. 66 67
Analytical procedures Hepatic lipids were extracted according to Bligh and Dyer. 112 Plasma and liver triglycerides, total cholesterol and free cholesterol concentrations were determined using commercially available kits (Roche Diagnostics, Mannheim, Germany and DiaSys Diagnostic Systems, Holzheim, Germany) or using gas chromatography after trimethylsilylation as described below when indicated. Cholesterol in bile was measured using lipid extraction following gas chromatography as described below for measurement of fecal neutral sterols. 218 Biliary and fecal bile salt composition were quantified using capillary gas chromatography (Hewlett-Packerd gas chromatograph; HP 6890) equipped with a FID and a CP Sil 19 capillary column; length 2 m, internal diameter 20 µm and a film thickness of 0.2 µm (Chrompack BV, Middelburg, The Netherlands). Bile salts were methylated with a mixture of methanol and acetyl chloride and trimethylsilylated with pyridine, N,O-Bis( trimethylsilyl) trifluoroacetamide and trimethylchlorosilane. The hydrophobicity index of the bile salt pool was calculated according to Heuman. 200 Fecal cholesterol and its derivatives (also known as neutral sterols) were trimethylsilylated with pyridine, N,O-Bis (trimethylsilyl) trifluoroacetamide and trimethylchlorosilane and measured on the GC equipped with the same column. Cholesterol enrichment was determined by capillary gas chromatography on a Agilent gas chromatograph (7890A; Amstelveen, The Netherlands) equipped with a 30 m 0.2 mm column, with a film thickness of 0.2 μm (ZB-; Bester, Amstelveen, The Netherlands) connected to a Agilent mass spectrometer (97C). Cholesterol was derivatized with N,O-Bis (trimethylsilyl) trifluoroacetamide containing -10 % trimethylchlorosilane. Isotope ratios were determined in the selected ion monitoring mode on m/z 48 (M0) to 46 (M7). TICE, Cholesterol absorption and synthesis were calculated as described elsewhere. 2,27,166 RNA isolation and measurement of mrna levels by quantitative real-time PCR Total RNA was isolated using TRI-Reagent (Sigma, St. Louis, MO, USA). cdna was produced using Moloney-Murine Leukemia Virus (M-MLV) reverse transcriptase (RT) (Life Technologies, Bleiswijk, The Netherlands) with random primers. Gene expression was normalized to 36B4. Primer and probe sequences can be obtained elsewhere (http://www.rtprimerdb.org). Figure 1. Effect of PX20606 on plasma, liver and fecal lipid parameters in C6BL/6J mice (A-D) Cholesterol and triglyceride levels in plasma and liver. (E) Fecal bile salts. (F) Fractional cholesterol absorption. (G) Fecal neutral sterols. (H) Transintestinal cholesterol excretion (TICE). (I) Left: Time course of the fractional enrichments of cholesterol-d in plasma. Right: Plasma cholesterol turnover rate. (J) Excretion of intravenously injected cholesterol-d in feces. All parameters were analyzed in wildtype mice fed a control diet (CTRL) or a diet supplemented with the FXR agonist PX20606 (PX) for 2 weeks. (n=7 mice/group) Western blotting for ABCG protein expression Intestinal brush border membranes were isolated as described by Schmitz et al. 219 Protein concentration was determined according to Lowry et al. 220 Equal amounts of brush border membranes (10 µg protein) for detection of ABCG were electrophoresed through 7% polyacrylamide gels and blotted on Hybond ECL membranes (Amersham, Little Chalfont, UK). Staining with Ponceau S was performed to check for equal protein transfer. Membranes were blocked in TRIS-buffered saline (ph 7.4) containing 0.1% Tween 20 and 4 % skim milk powder. Membranes were incubated with anti-abcg 68 69
primary antibody. 221 After washing, immunocomplexes were detected using horseradish peroxidase-conjugated goat anti-rabbit (DAKO, Heverlee, Belgium ) and SuperSignal West Dura substrate (ThermoScientific, Rockford, IL, USA). Statistics Data are shown as either Tukey box plots, stacked vertical bar plots, XY plot and XY scatter. Statistical analysis was performedby Kruskal-Wallis H test followed by Conover post-hoc comparisons when more than 2 experimental groups were compared or Man-Whitney U test when only 2 experimental groups were present using Brightstat. 116 HI indexes were statistically tested with Spearman s correlations coefficient. Differences were considered statistically significant when p < 0.0. Significant differences between groups are indicated with * when significant to control treated animals of the same genotype, # when significant to ezetimibe-treated animals of the same genotype, $ when significant to PX-treated animals of the same genotype or PX-treated rats before bile diversion, when significant to littermates with same treatment or AdNull-injected animals, and + significant to FGF19 supplemented mice on control diet. Correlations were analyzed by Spearman s correlations coefficient using Graphpad Prism. Results Pharmacological activation of intestinal FXR markedly enhances fecal neutral sterol excretion in mice To delineate the role of bile salts in control of body cholesterol fluxes, we chose to modify bile salt synthesis and pool composition via modulation of the bile salt receptor FXR. C7BL/6J mice were treated with the novel FXR agonist PX20606 (PX) for two weeks. In line with data obtained in hyperlipidemic mice and monkeys, 198 the compound was well tolerated and had no effect on body weight or food intake (data not shown). Confirming results from that study, PX treatment decreased plasma cholesterol and triglycerides but increased hepatic levels of these lipids (Figures 1A-1D). Liver weight was about 20% higher in PX-treated animals. In keeping with the regulatory role of FXR in bile salt synthesis, PX-treatment decreased fecal bile salt output (Figure 1E) which, in steady state conditions, reflects hepatic bile salt synthesis. Stably labeled cholesterol tracers were used to obtain more detailed insight in the effects of PX on body sterol fluxes. PX treatment inhibited cholesterol absorption (Figure 1F) and strongly increased fecal neutral sterol (FNS) (Figure 1G) excretion and TICE as calculated via the cholesterol balance method described in Brufau and Groen (2011) 166 (Figure 1H). PX induced a rapid elimination of, intravenously injected, cholesterol-d from the plasma compartment (Figure 1I), suggesting that the cholesterol withdrawn from the plasma compartment might end up in feces. The size Figure 2. Effect of PX20606 on FNS and TICE requires expression of intestinal FXR (A)Fecal neutral sterols from wildtype (WT) and FXR KO mice. (B) Fecal neutral sterols from intestinal FXR KO (ifxr KO) mice and WT littermates. (C) Transintestinal cholesterol excretion (TICE) in ifxr KO mice and WT littermates. (D) Fractional cholesterol absorption in ifxr KO mice and WT littermates. (E) Plasma cholesterol of (ifxr TGand whole body FXR KO mice. (F) Plasma triglycerides of ifxr TG and whole body FXR KO mice. (G) Fecal neutral sterols of ifxr TG and whole body FXR KO mice. (H) Fractional cholesterol absorption of ifxr TG and whole body FXR KO mice. (I) Fecal bile acids of ifxr TG and whole body FXR KO mice. Parameters were analyzed in wild-type (WT), FXR KO, ifxr KO or ifxr TG mice treated with control diet (CTRL) or control diet supplemented with PX20606 (PX) for two weeks. During the treatment, cholesterol kinetics were studied using stably labeled tracers. (n=6-8 mice/group) 70 71
of the exchangeable cholesterol pool was not affected by PX-treatment (data not shown). Notably, analysis of cholesterol tracer in feces revealed that PX-treated mice had excreted 68% of i.v. injected tracer into the feces at 72 h after injection whereas this was only 7.8% in control mice (Figure 1J). To control the selectivity and specificity of PX, we also performed experiments in whole-body FXR knock-out mice. PX did not impact FNS excretion (Figure 2A) in these mice nor on any of the other parameters studied, indicating that the observed effects were indeed mediated via FXR. PX has been shown to activate both hepatic and intestinal FXR. To disentangle the effects of hepatic and intestinal FXR we performed experiments in intestine-specific FXR (ifxr) knock-out mice. Interestingly and unexpectedly, the PX phenotype was almost completely abrogated in these mice; only marginal induction of FNS excretion (Figure 2B) and TICE (Figure 2C) was observed and in contrast to to WT mice cholesterol absorption was not affected by PX treatment in the ifxr KO mice. (Figure 2D). This indicates that intestinal FXR plays a major role in the control of whole-body cholesterol turnover. The role of intestinal FXR in control of cholesterol fluxes To more clearly delineate the role of intestinal FXR in control of cholesterol fluxes, we rescued intestinal FXR expression in a total body FXR KO background. Transgenic intestine-specific FXR expressing mice (ifxr TG) were generated using the murine FXR gene under the control of the villin promoter (Figure S1). No overt phenotypic differences between FXR KO and ifxr TG mice were noticed when kept under standard laboratory conditions. Treatment of ifxr TG mice with the agonist did not reduce plasma cholesterol and triglyceride levels (Figures 2E and 2F). This indicates that hepatic FXR rather than intestinal FXR is mediating the reduction of plasma lipids in response to PX. The compound also did not affect liver weight, bile flow or biliary cholesterol secretion (data not shown). Reinstitution of intestinal FXR expression in FXR KO mice rescued the effects of PX as it increased FNS excretion (Figure 2G), and decreased cholesterol absorption (Figure 2H) and fecal bile salt output (Figure 2I). However, induction of the fecal neutral sterols by PX is not as high in ifxr TG treated with PX as it is in PX-treated wildtypes, suggesting that hepatic FXR might contribute in part to the stimulation of FNS excretion by PX. Figure 3. Additive effects of PX20606 and ezetimibe on FNS and TICE in WT mice (A) Fecal neutral sterols. (B) Bile flow. (C) Biliary bile salt sectretion. (D) Biliary cholesterol secretion. (E) Transintestinal cholesterol excretion (TICE). (F) Fecal neutral sterols (FNS) over a period of 8 days. (G) Gene expression Abcg/g8 in liver, proximal and distal small intestine normalized to 36B4. (H) Protein levels of ABCG in brush border membranes isolated from proximal small intestine. All parameters were analyzed in wildtype C7BL/6J mice fed a control diet (CTRL) or a control diet supplemented with the cholesterol absorption inhibitor ezetimibe (EZE), the FXR agonist PX20606 (PX) or both (EZE/PX) for a period of 2 weeks. Cholesterol kinetics were assessed using stably labeled tracers. (n=7 mice/group) The role of cholesterol absorption blockade on cholesterol fluxes Part of the effect of PX on FNS excretion may have been caused by the observed inhibition of cholesterol absorption. We therefore investigated the influence of ezetimibe, an inhibitor of cholesterol absorption, on the stimulatory effect of PX on FNS excretion. For this purpose, ezetimibe was added to the diet of the mice. Confirming earlier reports, ezetimibe increased FNS excretion (Figure 3A) but had no effect on bile flow, biliary bile salt and only a limited effect on biliary cholesterol secretion (Figure 3B-3D). Ezetimibe did not affect bile salt secretion into feces nor did it influence bile salt species distribution noteworthy (Figure S2A). Since differences in dietary cholesterol intake and delivery of bile-derived cholesterol were 72 73
insufficient to account for the increase in FNS loss, most of the increase in FNS excretion is due to enhanced TICE. Interestingly, ezetimibe given in combination with PX induced an additive effect, leading to a massive FNS excretion and TICE in animals receiving the combined treatment (Figure 3A and 3E). Of note, 8 days of PX treatment was needed to reach the maximal effects on FNS excretion, while the maximal effect of ezetimibe was established much earlier (Figure 3F). This indicates that the compounds act via different mechanisms. Dual treatment had no effect on food intake or body weight of the mice but induced, compared to single treatments, an additional lowering of plasma cholesterol and triglycerides (Figure S2A and B). These effects were apparent in all lipoprotein classes (Figure S2A and B). Next, we addressed the mechanism via which cholesterol is secreted into the intestinal lumen. Being an established cholesterol transporter present at the luminal side of enterocytes, the Abcg/Abcg8 heterodimer is a potential mediator of the TICE pathway. As depicted in Figure 3G, ezetimibe slightly increased expression of Abcg and Abcg8 in the liver but decreased expression of the heterodimer in the proximal small intestine. PX increased Abcg/g8 expression in the liver but had no effect in the small intestine (Figure 3G). Figure 3H shows that the effect on intestinal gene expression was mirrored at protein level. Since expression levels of genes/proteins do not necessarily reflect activity, we investigated whether the heterodimer is involved in stimulation of FNS excretion and TICE by ezetimibe and PX using Abcg8 knock-out mice. The single Abcg8 knock-out phenocopies the double knock-out mice because the proteins need to dimerize to allow trafficking of the transporter to the apical membrane of hepatocytes and enterocytes. 199 We challenged Abcg8 knock-out mice and WT littermates with ezetimibe, PX or both and assessed the effects on body cholesterol fluxes. A severely blunted induction of FNS (Figure 4A), TICE (Figure 4B) and biliary cholesterol (Figure 4C) in response to treatment with ezetimibe, PX or both clearly demonstrates that ABCG/G8 is mediating the vast majority of cholesterol that is secreted via the TICE pathway under these stimulated conditions. Moreover, the data indicate that, although ezetimibe and PX act via different mechanisms, they share at least in part the same cholesterol efflux pathway. Interestingly, despite the blunted induction of FNS excretion, PX and dual treatment do considerably impact plasma cholesterol levels (Figure 4D) indicating that the decrease of plasma cholesterol induced by PX is not merely the consequence of increased intestinal excretion. Abcg/g8 is expressed in the liver and intestine. Therefore, reintroduction of Abcg8 in the liver via adenoviral transfection in Abcg8 -/- mice allows the investigation of the effects of Abcg/ g8 deficiency specifically in the intestine. Using this model, we investigated the impact of intestinal Abcg/g8 on plasma and biliary parameters, FNS output and TICE in response to PX and the combined treatment of ezetimibe and PX. Adenoviral rescue of Abcg8 in Abcg8 -/- mice had very little impact on FNS (Figure 4E) excretion and TICE (Figure 4F). It did, however, restore biliary cholesterol secretion (Figure 4G) and did seem to impair the PX and ezetimibe induced decrease in plasma cholesterol levels (Figure 4H). Figure 4. Effect of PX20606 and ezetimibe on cholesterol fluxes ABCG8-/- mice (A) Fecal neutral sterols. (B) Transintestinal cholesterol excretion (TICE). (C) Biliary cholesterol secretion. (D) Plasma cholesterol. (E) Fecal neutral sterols in mice injected with Abcg8-encoding or control adenovirus. (F) TICE in mice injected with Abcg8-encoding or control adenovirus. (G) Biliary cholesterol secretion in mice injected with Abcg8-encoding or control adenovirus. (H) Plasma cholesterol in mice injected with Abcg8-encoding or control adenovirus. Parameters were analyzed in Abcg8-/- mice and their WT littermates (A-D) or Abcg8-/- mice infected with control adenovirus (AdNull) or adenovirus encoding Abcg8 (AdAbcg8) (E-H). Mice received either control diet or control diet supplemented with ezetimibe (EZE), PX20606 (PX) or both compounds (EZE/PX) for two weeks. Cholesterol kinetics were studied using stably labeled tracers. (n=3-6 mice/group for panels A-D, n=4 per group for panels E-H) 74 7
Taken together, these data pinpoint intestinal ABCG/G8 as the transporter mediating most of the cholesterol excretion in mice treated with either PX alone or in combination with ezetimibe. The mechanism by which intestinal FXR activation and ezetimibe induce TICE FXR is expressed along the small intestine but is considered to act under physiological conditions mainly via bile salt signaling in the distal part of the ileum. Yet, PX may activate FXR all along the small intestine and thereby stimulate TICE at the site where it is most active, i.e. in the proximal small intestine. 24 To analyze which genes/pathways were targeted by PX, we performed microarray analysis in the proximal intestine at an FDR of %. Figure S3A shows the results for KEGG pathway analysis; genes involved in Parkinson, Huntington disease and oxidative phosphorylation popped up in the proximal part of the small intestine. KEGG pathway analysis of the array data derived from the distal small intestine gave a very different pattern (Figure S3B). In this case, metabolic pathways involved in nitrogen and carbohydrate metabolism were the most regulated ones. In contrast, cholesterol synthesis was the most prominent regulated pathway in the liver, indicating that the mobilized cholesterol was replenished at least in part by the de novo synthesis in the liver (Figure S3C). Since the small intestine is the site where TICE is most active, we compared small intestinal gene expression patterns after treatment with either PX or ezetimibe. At an FDR of 10%, FXR activation induced differential expression of 4022 genes in the proximal small intestine and from which 1 genes overlapped with ezetimibe treatment (Figure S3D). At this relatively high FDR genes involved in cholesterol synthesis surfaced. Both treatments apparently increased intestinal cholesterol synthesis somewhat. No known genes involved in cholesterol transport were commonly upregulated. Figure. FGF1/19 mediates fecal neutral sterol excretion (A) Gene expression of Cyp7a1 and Cyp8b1 in C7BL/6J mice. (B) Cholesterol synthesis analyzed after during a 3 day administration of 13C-acete via the drinking water in C7BL/6J mice. (C) Fecal neutral sterols in ifxr KO mice and WT littermates. (D) Gene expression of Cyp7a1 and Cyp8b1 in ifxr KO mice and WT littermates. (E) Bile salt species distribution in ifxr KO mice and WT littermates. (F) Correlation between the hydrophobicity index (HI) of biliary bile salts versus fecal neutral sterols in ifxr KO mice and WT littermates. Parameters from panels A and B were analyzed in C7BL/6J mice which either received control diet (CTRL) or diet supplemented with ezetimibe (EZE), PX20606 (PX) or both compounds. Parameters from panels C-F were analyzed in ifxr KO mice and WT littermates fed control diet or diet supplemented with PX for two weeks receiving i.p injections with either vehicle or FGF19 during the last 8 days of PX treatment (n=3-7 mice/group). Validation by PCR confirmed the results from the KEGG pathway analysis in liver and showed the expected changes in hepatic gene expression of sterol 12-alpha-hydroxylase (Cyp8b1) and cholesterol 7-alpha-hydroxylase (Cyp7a1) (Figure A). Application of mass isotopomer distribution analysis after administration of 13C-acetate confirmed the anticipated upregulation of cholesterol synthesis (Figure B). In the small intestine, FXR activation induced the expected upregulation of FGF1 (data not shown) which is thought to mediate most of the physiological relevant effects of intestinal FXR on the liver. When FGF1 mediates the effects of PX on FNS excretion and TICE, treatment of ifxr mice with FGF1/19 should at least partially rescue the lack of effect of PX in ifxr -/- mice. Therefore, PX-treated ifxr mice were treated for eight days with the human orthologue FGF19 and FNS output was compared to mice only receiving PX. As evident from the data shown in Figure C, FGF19 rescued FNS excretion thereby identifying FGF1/19 as a crucial factor mediating increased FNS loss in response to FXR agonism by PX. Next, we explored the mechanistic basis underlying the restoration of 76 77
PX-induced stimulation of FNS excretion by FGF1/19. Both, PX and Fgf1, control bile salt synthesis but via a differential effect on expression of Cyp7a1 and Cyp8b1 (Figure D), leading to an altered balance between the classic and acidic bile salt synthesis pathways. We therefore determined bile salt species composition in the ifxr -/- mice treated with PX and/or FGF19 (Figure E). Interestingly, FGF19 seemed to be required for the shift of the balance towards the relatively hydrophilic muricholic acids instead of the more hydrophobic cholic acid and deoxycholic acid. This interesting observation let us to study the relation between the hydrophobicity index (HI), calculated according to Heuman, 200 with FNS excretion. A clear negative correlation emerged (Figure F) suggesting that hydrophilic bile salts stimulate FNS excretion via the TICE pathway. Similar to hydrophilicity changes in biliary bile salts, the hydrophilicity of the fecal bile salt pool changed gradually over a period of 8 days during PX treatment (Figure S4A), in line with the gradual increase in FNS. Of note, in Abcg8 -/- mice the hydrophobicity of the bile acid pool changed in identical way as was observed in WT mice (data not shown) but this translated into only limited effects on FNS excretion and TICE, suggesting an interaction of hydrophilic bile acids with the ABCG/G8 transporter in the intestine. If hydrophilic bile salts would indeed be responsible for the increased FNS excretion as a result of increased TICE, PX would fail to stimulate FNS excretion in bile diverted animals. However, it is known that cholesterol excretion via TICE requires the presence of acceptor particles (i.e. bile acids and phospholipid micelles) in the intestine. Therefore, we used rats that were treated with PX and non-treated controls and performed bile duct diversion experiments in non-anaesthesized rats. 201 Animals were allowed to recover for days after surgery, in which permanent catheters were placed in the bile duct and duodenum. During this period, cannulas from the bile duct and small intestine were connected on top of the head of the animals, essentially restoring the enterohepatic circulation via an extended bile duct. After having obtained feces for determination of FNS excretion, the enterohepatic circulation was interrupted and all bile from the rats was quantitatively collected in a tube located outside of the cage. From the moment of interruption, model bile was infused into the intestinal cannula. All animals received identical model bile composed of taurocholate and phosphatidylcholine, ensuring identical bile salt actions in the intestine. If PX would mediate its effects by affecting the bile salt pool composition, then the effect of PX would be lost after interruption of the enterohepatic circulation and start of the model bile infusions. Indeed, whereas PX-induced FNS excretion in the rats when the enterohepatic circulation was still intact, this effect was completely abrogated when the enterohepatic circulation was interrupted and the rats were receiving the identical model bile (Figure 6A). Of note, gene expression analysis confirmed upregulation of the FXR target gene Small heterodimer partner (NR0b2/Shp) and downregulation of Cyp7a1 and Cyp8b1 in the PX-treated rats, indicating the FXR agonist was functionally active in these bile- diverted animals (Figure 6B). Discussion In this study we demonstrate that the intestine itself has the capability to facilitate elimination of humongous amounts of cholesterol from the body via the TICE pathway. The large capacity of TICE that became evident from the current studies highlights the potential of the intestine as a target for future cholesterol-lowering therapies. Importantly and intriguingly, this study demonstrates for the first time that the presence of bile salts in the intestinal lumen is not only a prerequisite for uptake of dietary and biliary cholesterol by the enterocytes but, additionally, an important determinant of excretion of plasma-derived cholesterol by the enterocytes back into the intestinal lumen and hence of whole-body cholesterol turnover. The large capacity of the intestine to remove cholesterol can be appreciated from a comparison between the measured values of TICE with estimates of the total amount of cholesterol that is present within the mouse body. The total cholesterol pool in mice has been estimated to be about 0 mg, an amount shown to be remarkably constant across a variety of genotypes. 202 About 2% of this pool is present in the brain and does not exchange with the remainder of the body pool. 203 Dual treatment with Ezetimibe and PX increased FNS excretion to about 30 mg per 24 h, implying that an amount of about 80% of the entire exchangeable pool is excreted each day. Since the sum of biliary secretion and dietary intake of cholesterol in our experimental set up amounted to maximally mg/day, TICE has to account for at least 80% of this cholesterol disposal. Earlier studies from our laboratories indicated that the sterol transporting heterodimer Abcg/ g8 plays a prominent Figure 6. PX20606-induced fecal neutral sterol excretion is mediated via bile (A) Fecal Neutral Sterols. (B) gene expression of Cyp7a1, Cyp8b1 and Shp. Experiments were performed in rats treated with a control diet (CTRL) with or without PX20606 (PX). Rats were surgically equipped witht cannulas in the bile duct and duodenum. After surgery, the animals were allowed to recover for days. During this period, the bile duct and intestinal cannulas were connected on the head of the animals, essentially restoring the enterohepatic circulation via an extended bile duct. After having obtained feces for determination of FNS excretion, the enterohepatic circulation was interrupted, and bile diverted and quantitatively was collected. From the moment of interruption, model bile was infused into the intestinal cannula. All animals received identical model bile composed of 1 mm taurocholic acid and 2. mm phosphatidylcholine infused at 1.2 ml/hour, ensuring identical bile salt actions in the intestine. After sacrife, the liver was collected for measurement of gene expression. (n=3-6 rats/group). 78 79
role in the TICE pathway. 2,27,28 Our present results demonstrate that in addition to Abcg/g8, the cholesterol importer NPC1L1 and luminal bile salts, and in particular their physical chemical properties, take part in the control of TICE. Using a number of transgenic mouse models, pharmacological manipulations and in vivo flux measurements, we were able to visualize and to quantify the interaction between the different steps in the TICE pathway. On the apical membrane domain of the enterocyte, the sterol transporters NPC1L1 and ABCG/G8 were demonstrated to be the major mediators of TICE flux, at least under the TICE-stimulating conditions employed in this study. Both transporters are most highly expressed in duodenum and jejunum. Although lack of suitable specific antibodies has precluded visualization of co-localization, it seems reasonable to assume that they are present in the same cells. The activity of an importer and exporter in the same cell introduces a potential futile cycle. However, such systems often operate in biology to ensure versatile flux regulation. It is worth mentioning that in absence of Abcg/g8 activity dual treatment with PX and ezetimibe still activated TICE up to 33 µmol/day/100g (1% of WT values), suggesting that there is another transporter present that can partly compensate for Abcg/ g8. In support of this, Wang et al. showed recently that feeding Abcg/g8 DKO mice with ursodeoxycholate stimulated FNS excretion considerably. 204 As suggested by Le May et al., the multidrug resistance transporters Abcb1a and b may be responsible for at least part of the ABCG/G8-independent TICE flux. 31 We did not detect increased expression of the Abcb1a and b genes under any of the stimulated conditions. However, we also did not observe increased expression of Abcg/g8, even under very high flux conditions. This suggests that these proteins have a very high intrinsic transport capacity. Rescue of hepatic Abcg/g8 activity by adenoviral transfection fully restored biliary cholesterol secretion. However, this barely impacted on FNS excretion, illustrating that most of the induced FNS flux under ezetimibe and or PX-stimulated conditions indeed is generated by intestinal Abcg/g8 activity. Whereas ABCG/G8 controls the outward flux of cholesterol from the enterocyte, NPC1L1 mediates the inward flux and thereby impacts net TICE. The contribution of Npc1l1 activity to TICE under the different conditions can be deduced from the effect of ezetimibe. Under control conditions, the inhibitor increased FNS excretion to 78 µmol/day/100g. Of note, ezetimibe has no effect on FNS in the absence of NPC1L1, 20 hence does not stimulate activity of Abcg/g8 or other cholesterol transporters directly. Assuming that ezetimibe fully inhibits activity of Npc1l1, an uptake rate of cholesterol under basal conditions of 9 µmol/day/100g can be calculated. Interestingly, the calculated flux through Npc1l1 increased up to 10 µmol/day/100g when PX was also present. This came as a surprise in view of the apparent inhibiting effect of PX on cholesterol absorption measured with the dual isotope method. 206 Apparently this method underestimates true fractional cholesterol absorption when a high activity of cholesterol export is present at the same time and a substantial part of the absorbed cholesterol is re-secreted at the luminal side rather than secreted at the basolateral membrane of the enterocyte. The effect of PX on Npc1l1 activity is not due to increased gene expression, hence other factors must be responsible. PX was shown previously to potently decrease plasma cholesterol levels and progression of atherosclerosis in CETPtransgenic/LDL-receptor knock-out mice. 198 At first sight, our results indicate that massive mobilization of cholesterol, driven by increased TICE, may explain the cholesterol lowering effect in plasma. Indeed, plasma cholesterol was also reduced in PX treated wild-type mice in our study. However, when increased TICE would underlie the reduction of plasma cholesterol levels, the same phenomenon should have been observed in the intestine-specific FXR transgenic mice on a FXR knock-out background. However, despite a considerable rescue of TICE activity in these mice there was very little effect on plasma cholesterol. It is therefore most conceivable that reduction of plasma cholesterol upon PX treatment in mice that do express hepatic FXR is due to enhanced expression of scavenger receptor class B-I (SR-BI) upon FXR activation as reported previously by others. 207 In keeping with this hypothesis, we found upregulation of hepatic SR-BI expression on mrna as well as protein level (data not shown). Thus, de novo cholesterol synthesis by the liver apparently compensates for the increased loss of cholesterol by TICE mediated sterol excretion. Inspection of the effect of PX on global gene expression in the proximal 10 cm of the intestine revealed that more than 4000 genes were significantly regulated by the agonist (Figure S3). Interestingly, cholesterol synthesis or transport were not in the top list in KEGG pathway analyses, suggesting that the stimulatory effect of PX on cholesterol trafficking was not determined by intestinal cholesterol production. We then focused the attention to the well-known ileal FXR target; FGF1/19. Supplementation of FGF19 to ifxr -/- mice fully rescued the effect of PX on FNS and TICE. Indeed, FGF19 alone was sufficient to induce cholesterol mobilization. Studies from the group of Kliewer and Mangelsdorf demonstrated direct effects of this intestinal hormone on glucose, lipid and bile acid metabolism, however intestinal cholesterol trafficking was not addressed. 208,209 The role of FGF19 in regulation of bile salt synthesis has been described in detail and these results are confirmed in the present study. 209 Treatment with PX had a strong impact on the ratio in which the classic and acidic pathways contribute to overall bile salt synthesis. Consequently, the ratio of muricholates to cholate increased considerably. We have shown in an earlier study that ABCG/G8-mediated cholesterol efflux in cultured cells is stimulated more strongly by hydrophilic bile salts compared to hydrophobic bile salts. 210 This difference between bile salt species now also appears to be of great physiological relevance in vivo. As a matter of fact, our data indicate that bile salt species distribution actually underlies the stimulatory effect of PX on FNS excretion and TICE. As shown in Figure F, a strong linear relationship exists between the magnitude of FNS excretion and the bile salt hydrophobcity index as first defined by Heuman based on HPLC retention times. 211 To validate this surprising effect of bile salt hydrophobicity in an 80 81
independent manner, we performed a series of experiments in PX-treated rats in which the enterohepatic circulation was diverted by surgical means to allow replacement of native bile by a model bile. Although the induction of FNS excretion by PX in rats was more modest compared to what was observed in mice, we clearly demonstrated that FNS excretion and TICE indeed strongly decrease when the PX-induced hydrophilic native bile salt pool is replaced by taurocholate, supporting a pivotal role for bile salt hydrophobicity. In another study we have shown that TICE is also active in humans and, similar to rodents, ezetimibe accelerates FNS by increasing TICE. In contrast to mice, ezetimibe treatment influences bile salt secretion possibly via modulation of the intestinal microbiota in humans. Also in humans we found a linear trend between bile salt hydrophobicity and TICE suggesting that in humans as well, bile salt hydrophilicity may regulate the rate of cholesterol excretion from the body. In conclusion, this is the first study to demonstrate that the intestine is capable of eliminating large quantities of cholesterol from the body via a direct blood-borne pathway. Net flux through the TICE pathway is regulated by a complex interplay between the cholesterol importer Npc1l1, cholesterol exporter Abcg/g8 and the physical chemical properties of bile salts within the intestinal lumen. It is noteworthy that loss of function mutations and pharmacological inhibition of NPC1L1 in humans have been shown to ameliorate cardiovascular disease and ABCG8 has been linked to cardiovascular disease in GWAS studies. 212 Therefore, the current study provides a strong rationale for the consideration of combination therapies including hydrophilic bile salts for the management of lipid disorders and protection of atherosclerosis and cardiovascular disease. Supplemental Data Supplemental Experimental Procedures Plasmid construction and generation of intestine-specific FXRα2 mice The coding sequences of murine FXRα2 was amplified using specific primer pairs (forward; TACTACGGATCCACCATGAATCTGATTGGGGACTCC, reverse; TACTACGCGGCCGCTCACTGCACAT CCCAGATCT) harboring a Kozak consensus ATG initiation codon and a BamH1 restriction side at the end and a Not1 restriction side at the 3 end. As a template source, cdna was synthesized from hepatic RNA from C7BL/6J mice. The products were cloned into pcdna/frt/ TO (Life Technologies, Bleiswijk, The Netherlands). The presence of the correct gene was sequence-verified. Subsequently, the MYTG insertion was deleted from FXRα1 using inverse PCR primers (forward; TTGTTAACTGAAATCCAGTG, reverse; ACATTCAGCCAACATCCCCA), to generate FXRα2. FXRα2 was cloned behind the villin promoter in the pbsii-12.4kbvillin plasmid 213 (kind gift from Deborah Gumucio, University of Michigan Medical School, Ann Arbor MI, USA). FXRα2 was restricted from pcdna/frt/to using PmeI and XhoI and was ligated into the 12.4 kbvillin plasmid restricted using SmaI and XhoI. The presence of the correct insert and orientation was sequence-verified. Subsequently, the vector backbone was deleted by restriction using PmeI in the 12.4 kbvillin plasmid. The linear Villin-FXRα2 and Villin-FXRα4 sequences were heat-inactivated and purified using gel electrophoresis. Intestine-specific murine-fxrα2 transgenic mice were generated by microinjection of the construct into fertilized FVB/NHsd eggs. Clones with the highest murine FXR mrna expression levels were recovered. Mice were backcrossed to whole body FXR-deficient mice, 214 on a C7BL/6J background, generating a model with intestine-specific expression of FXRα2 on a whole-body FXR knock-out background. Cell culture, transfections and reporter assays CV1 cells (a kind gift from Ronald M. Evans, Salk Institute, San Diego, USA) were maintained in DMEM (Gibco, Breda, The Netherlands) supplemented with 10% FCS (Sigma Aldrich Chemie BV, Zwijndrecht, The Netherlands), 100 U/ml penicillin and 100 µg/ml streptomycin (Gibco, Breda, The Netherlands). Cultures were maintained at 37 C and % CO2 in a humidified incubator. Cells were transiently transfected using FuGENE 6 transfection reagent (Promega, Leiden, The Netherlands). Cell lysis and luciferase assay were performed using a dual luciferase reporter assay system (Promega, Leiden, The Netherlands). Briefly, the human PGL4-SHP (CHR16_M0432_R1 and CHR1_M0312_R1, Switchgear Genomics, Menlo Park, CA, USA) promoter reporter was cotransfected together with the pcdna/frt/ TO murine FXR isoforms and the heterodimer RXR alpha/nr2b1 for 48 hrs. Cells remain untreated or were treated with 0µM CDCA (Sigma-Aldrich), 1µM GW4064 (Sigma-Aldrich) or 1µM PX 198,216 (Phenex Pharmaceuticals, Heidelberg, Germany) 24 prior to cell lysis. 82 83
Figure S1: Generation of intestine-specific FXR transgenic mice (A) Transgene construct of murine FXRα2. The mouse villin promoter was ligated to the murine Fxrα2 gene. The construct was restricted from the backbone using PmeI and the resulting villin-fxrα2 fragment was quantified for microinjection. (B) Transcriptional activity of murine FXRα2 was tested in response to several FXR ligands. A luciferase assay was performed using an luciferase reporter containing tandem copies of FXRE from SHP and activity is shown in fold induction. (C) FXR protein expression levels were measured in mucosa of the distal ileum of chow-fed FXR KO, ifxr TG and WT mice. FXR protein levels were corrected for GAPDH protein levels. Data are presented as means ± SD (n= mice/group). (D) Immunohistochemistry staining using FXR antibody in the distal ileum of wildtype (E) FXR knockout and (F) ifxr transgenic mice. Figure S2: Plasma lipid distribution upon inhibition of intestinal cholesterol absorption and FXR activation (A) Biliary bile acid distribution in mice fed control diet with or without ezetimibe (EZE) for two weeks. (B) Left: Plasma cholesterol levels in mice fed control diets with or without EZE and/or PX20606 (PX). Right: Cholesterol concentrations in lipoprotein fractions after separation of pooled plasma samples by fast protein liquid chromatography (FPLC). (C) Left: Plasma triglyceride levels in mice fed control diets with or without EZE and/or PX. Right: Triglyceride concentrations in lipoprotein fractions after separation of pooled plasma samples by FPLC. (n=7 per group) 84 8
Intestinal immunohistochemical protein expression of FXR Briefly, after fixation in formaldehyde, intestinal pieces were embedded in paraffin and cut into 4μm sections. For immunohistochemistry, sections were deparaffinized and rehydrated. Antigen-retrieval was done in 0.01 M citrate buffer ph 6 at 100 C for 20 min. Endogenous peroxidase activity was blocked by incubation for 30 min in 3% H 2 O 2 in phosphate-buffered saline (PBS). Sections were blocked with % goat serum in PBS for 30 min. Immunostaining was performed overnight at 4 C using anti-fxr (Perseus Proteomics, Tokyo, Japan; PP-A9033A-00) 0.01 mg/ml in 2% normal goat serum/1% bovine serum albumin in PBS. Sections were washed in PBS followed by a 60 min incubation with polyclonal goat anti-mouse immunoglobulins/biotinylated (Dako, Denmark; 1/200 dilution), and subsequent by a 60 min incubation with streptavidin/hrp (Dako, Denmark; 1/200 dilution), both in PBS containing 1% normal goat serum. Visualization of the immune complexes was performed with diaminobenzidine, followed by hematoxylin counterstaining for 60 sec. Sections were dehydrated and mounted with Eukitt (Sigma-Aldrich, USA). Figure S4: Fecal bile acid distribution during 8 days of PX treatment (A) Fecal bile acid distribution in feces of mice treated with PX20606 (PX). (n=7 mice/group). Figure S3: KEGG pathways analysis in liver, proximal and distal small intestine KEGG pathway analyses of mrna expression in (A) proximal small intestine, (B) distal small intestine and (C) Liver after 2 weeks of PX20606 (PX) treatment. (D) Overlay of gene expression after 2 weeks of treatment with PX or ezetimibe (EZE). 86 87