Potential conflict of interest: Nothing to report.
Supported by National Institutes of Health grants R01 DK53366 (to D. D. M.), RC4 DK90849 (to P. W.), and R01 DK54208 (to G. D. L.), and R01 DK062975 and a VA Merit Award (to G. A.).
Nuclear receptors (NRs) play crucial roles in the regulation of hepatic cholesterol synthesis, metabolism, and conversion to bile acids, but their actions in cholangiocytes have not been examined. In this study, we investigated the roles of NRs in cholangiocyte physiology and cholesterol metabolism and flux. We examined the expression of NRs and other genes involved in cholesterol homeostasis in freshly isolated and cultured murine cholangiocytes and found that these cells express a specific subset of NRs, including liver X receptor (LXR) β and peroxisome proliferator-activated receptor (PPAR) δ. Activation of LXRβ and/or PPARδ in cholangiocytes induces ATP-binding cassette cholesterol transporter A1 (ABCA1) and increases cholesterol export at the basolateral compartment in polarized cultured cholangiocytes. In addition, PPARδ induces Niemann-Pick C1-like L1 (NPC1L1), which imports cholesterol into cholangiocytes and is expressed on the apical cholangiocyte membrane via specific interaction with a peroxisome proliferator-activated response element (PPRE) within the NPC1L1 promoter. Conclusion: We propose that (1) LXRβ and PPARδ coordinate NPC1L1/ABCA1-dependent vectorial cholesterol flux from bile through cholangiocytes and (2) manipulation of these processes may influence bile composition with important applications in cholestatic liver disease and gallstone disease, two serious health concerns for humans. (HEPATOLOGY 2012;56:2288–2296)
Although the effects of nuclear receptors (NRs) upon hepatic cholesterol synthesis and degradation and conversion to bile acids have been studied extensively, their effects on cholangiocytes are unclear.1 Cholangiocytes are epithelial cells that actively regulate bile composition through a process of absorption and secretion; they modulate the content of water, organic anions and cations, and electrolytes in bile.2 Cholangiocytes can also reabsorb bile acids, a process that is important in cholestatic liver disease.3 Although cholesterol is present at high levels in bile, it is not yet known whether cholangiocytes influence bile cholesterol content or whether this process is regulated by NRs. This issue is important in devising a means of reducing cholesterol in bile to prevent/block cholestatic liver disease and gallstone formation.
NRs, including the liver X receptors (LXRs), the peroxisome proliferator-activated receptors (PPARs), farnesoid X receptor (FXR), and short heterodimer partner (SHP) regulate genes with roles in cholesterol and bile acid metabolism in a variety of cell types such as hepatocytes, cholangiocytes, and macrophages.4 For example, LXRα regulates CYP7A1, low-density lipoprotein receptor (LDLR), and sterol response element binding proteins (SREBPs) that mediates fatty acid regulation in hepatocytes.5 In macrophages, LXRβ and PPARs regulate ATP-binding cassette cholesterol transporter A1 (ABCA1), an ATP-binding cassette cholesterol transporter that mediates cholesterol efflux and reverse cholesterol transport from the periphery to the liver.4
Niemann-Pick C1-like L1 (NPC1L1) is another key component of cholesterol metabolism.6 The protein is the putative target for the cholesterol uptake inhibitor ezetimibe; it is expressed on the brush border of the small intestine, where it mediates dietary cholesterol absorption.7 NPC1L1 is also expressed in human liver, but its precise distribution is controversial with some studies showing high hepatocyte expression.8 while others show higher expression in gallbladder epithelial cells.9 Regulatory elements in the NPC1L11 promoter are partly characterized with suggestive evidence for direct roles of hepatocyte nuclear factor (HNF) 1α and SREBP2,10 as well as a possible indirect inhibitory role for PPARα.11
In this study, we investigated the roles of NRs in cholangiocyte physiology and cholesterol transport and metabolism. We examined the expression of NRs and other genes involved in cholesterol homeostasis in freshly isolated and cultured murine cholangiocytes and found that cholangiocytes express a subset of NRs that are different from hepatocytes, and that agonists for two of these receptors (LXRβ and PPARδ) induce vectorial cholesterol transport through the cholangiocyte basolateral membrane via ABCA1 induction. We also found that NPC1L1 is expressed on the apical cholangiocyte membrane and is regulated by PPARδ via specific interaction with a peroxisome proliferator-activated response element (PPRE) located within the NPC1L1 promoter, suggesting that PPARδ plays a direct role in NPC1L1 expression in cholangiocytes. We propose that LXRβ and PPARδ coordinate NPC1L1/ABCA1-dependent cholesterol flux from bile through cholangiocytes and that it will be possible to manipulate these processes to influence bile physiology in cholestatic liver disease and gallstones.
Cholangiocytes and hepatocytes were isolated from normal and bile duct ligated (BDL) rat liver as described by Alpini et al.3 Normal rat cholangiocytes (NRCs) were isolated and established in culture as described.12 AAV-293 cells were obtained from Stratagene (La Jolla, CA). T0901317, GW501516, water-soluble cholesterol, 27-, 22-hydrocholesterol, fatty acid-free bovine serum albumin, apolipoprotein A1 (apoAI), Flag, and β-actin monoclonal antibodies and other chemicals were obtained from Sigma (St. Louis, MO). All cell culture reagents were obtained from Invitrogen (Carlsbad, CA). ABCA1 polyclonal antibody was obtained from Novus (Littleton, CO). PPARδ and LXRβ polyclonal antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Anti-NPC1L1 polyclonal antibody was obtained from Cayman Chemical (Ann Arbor, MI).
Quantitative Reverse-Transcription Polymerase Chain Reaction and Northern Blot Analysis.
For quantitative reverse-transcription polymerase chain reaction (qRT-PCR), total RNA was purified using an RNeasy mini kit following the manufacturer's instructions.
PCR fragments containing murine NPC1L1 5′ flanking region (−1045 to −1, relative to translation start ATG) were cloned into pGL3zeocin at MluI-XhoI to generate pGL3-1.1NPC1L1-Luc. The PPARδ coding region was amplified from the pCMX.mPPARδ template and inserted into p3XFLAG-CMV-10 (Sigma) at the Hind-III/BamH1 site to obtain pFLAG-PPARδ. Luciferase activity was measured using a dual-luciferase kit according to the manufacturer's instructions (Promega, Fitchburg, WI) and was reported as a ratio of firefly luciferase to Renilla luciferase.
The PPARδ target sequence (RNA interference [RNAi] #1: AATCCGCATGAAG CTCGAGTA, encoding amino acids 630-650) was cloned into pSilencer-2.1-U6-Hygro according to the manufacturer's instructions. The construct was verified by sequencing the junctions and inserts before transfection. To knock down PPARδ mRNA levels, PPARδ-RNAi was transfected separately using Lipofectamine 2000 (Invitrogen). pSilencer-2.1-U6-Hygro was used as a negative control.
Immunohistochemistry and Immunofluorescence.
For immunohistochemistry analysis, paraffin-embedded rat liver sections (4 μm) were dewaxed, and antigens were retrieved. After blocking, slides were incubated with primary antibodies. EnVision (Dako) kit was used for avidin-biotin complex method to visualize the signal following the manufacturer's instructions. For immunofluorescence analysis, cultured cholangiocytes were made permeable for 20 minutes at −20°C with acetone, washed, and incubated with the appropriate antibodies. After washing, slides were reprobed with goat anti-rabbit Alexa 594 (Invitrogen) and were mounted with Fluoromount G. Nuclei were counterstained with 4',6-diamidino-2-phenylindole (DAPI). Slides were analyzed using a Nikon A1 cofocal microscope system.
Western Blot Analysis.
Western blot analysis was performed using protein extraction from cultured cholangiocytes. Samples were separated by electrophoresis on 12% (wt/vol) polyacrylamide gels and electrophoretically transferred to a polyvinylidene difluouride membrane (Bio-Rad Laboratories). Blots were probed with primary antibodies, followed by horseradish peroxidase–conjugated secondary antibodies.
Cholesterol Efflux and Uptake.
Cells were exposed to either 1 μM T0901317, GW501516, or vehicle (0.1% dimethyl sulfoxide) and incubated with 0.5 μCi/mL [3H]-cholesterol (Amersham) and 50 μg/mL cholesterol for 24 hours in serum-free medium containing 0.2% bovine serum albumin. Cells were washed twice with phosphate-buffered saline and efflux of cholesterol was initiated in serum-free medium in the presence or absence of 10 μg/mL apoA1. After 4 hours, medium was removed, cell debris was pelleted, and effluxed radioactivity was measured by scintillation counting. To obtain cell-associated [3H]-cholesterol, cells were lysed in 0.2 M NaOH and radioactivity was measured by scintillation counting. Percent efflux was determined by dividing radioactivity in the culture media by the sum of radioactivity in the cell lysate and the media. apoA1-dependent efflux was calculated by subtracting percent efflux without apoA1 from efflux with apoA1 present.
Electrophoretic Mobility-Shift Assay.
Nuclear extracts were prepared from PPARδ expressing AAV293 cells using a Nuclear/Cytosol Fractionation Kit according to manufacturer's instruction (BioVision, Mountain View, CA).
Chromatin Immunoprecipitation–Coupled qRT-PCR.
ChIP was performed using the Chromatin Immunoprecipitation (ChIP) Assay Kit (Millipore, Billerica, MA) according to the manufacturer's protocol.
Data represent the mean ± SEM. We used a t test and analysis of variance to calculate and determine statistical significance.
Cholangiocytes and Hepatocytes Express Different NRs and Genes Involved in Cholesterol Metabolism
To determine the expression of NRs in cholangiocytes and their possible roles in cholesterol transport and metabolism, we analyzed gene expression in freshly isolated hepatocytes, large and small cholangiocytes (Fig. 1A,C; Supporting Figs. 3-6), and cultured cholangiocytes (Fig. 1B; Supporting Figs. 1 and 2). Purity of cell preparations was verified by qRT-PCR with specific cholangiocyte and hepatocyte marker genes (Supporting Fig. 5).
We detected expression of multiple NRs in cholangiocytes, including LXRβ, PPARδ, thyroid hormone receptor (TR) α, estrogen receptors (ERs) α and β, retinoid X receptors (RXRs) β and γ, and vitamin D receptor (VDR) (Figs. 1A,B; Supporting Figs. 1, 3, and 4). Of these, several transcripts were preferentially expressed in cholangiocytes versus hepatocytes, including PPARδ (Fig. 1A), TRα, and VDR (Supporting Fig. 4). Conversely, many NRs that are expressed in hepatocytes such as PPARα, PPARγ, HNF4α, TRβ, LXRα, RXRα, FXR, SHP , pregnane X receptor (PXR), liver receptor homolog 1 (LRH1) and constitutive androstane receptor (CAR) were undetectable levels in cholangiocytes (Supporting Figs. 1, 3, and 4). Both PPARδ and LXRβ were highly expressed in rat cholangiocytes as shown by western blotting analysis and immunohistochemical analysis of rat liver sections (Fig. 1B,C). Expression of many of these NRs, including PPARδ, PPARγ, LXRβ, HNF4α, FXR, TRβ, and RXRα,β, was similar in both large and small cholangiocytes, although expression of TRα, ERα,β, RXRγ, and VDR are higher in small versus large cholangiocytes and the NR coregulators PGC1α and β were expressed at higher levels in large cholangiocytes, implying that NR actions may not be completely identical.
Hepatocytes and cholangiocytes displayed unique expression profiles of genes associated with bile acid and cholesterol metabolism. CYP27A1, which catalyzes conversion of cholesterol to oxysterols, was the only significantly expressed gene product with a role in cholesterol metabolism that could be detected in large cholangiocytes, albeit at a much lower level than hepatocytes (Fig. 1D; Supporting Fig. 1). Other enzymes typically involved in the conversion from cholesterol to bile acids (CYP7A1, CYP7B1, and CYP8B1) were undetectable (Fig. 1D; Supporting Fig. 1).
Among cholesterol and bile acid receptors, transporters, and apolipoprotein components, we detected high expression of ABCG1 and LDLR transcripts in cholangiocytes relative to hepatocytes (Supporting Fig. 6), whereas ABCA1, CD36, SR-BI, ABCG5, ABCG8, apolipoprotein E (apoE), fatty acid binding protein (FABP) and the hepatocyte bile acid transporters BSEP, NTCP, and OATP were low or essentially undetectable (Supporting Figs. 2, 5, and 6). The expression of ABCA1, CD36, SR-B1, and FABP was the same in both large and small cholangiocytes, whereas ABCG1 was higher and LDLR were lower in large cholangiocytes.
LXR-Dependent Basolateral Cholesterol Efflux in Cholangiocytes.
The pattern of LXR and PPAR subtype expression in cholangiocytes is more similar to macrophages, which function in reverse cholesterol transport, than hepatocytes. We therefore asked whether activation of LXRβ and PPARδ would mediate an analogous response in cultured cholangiocytes. Treatment of NRCs with the LXR agonist T0901317 increased the expression of ABCA1 and ABCG1, whereas FXR and PPARγ agonists chenodeoxycholic acid and rosiglitazone had no effect (Fig. 2A). The PPARδ agonist GW501516 also induced ABCA1 in cultured cholangiocytes, although it had no effect on ABCG1 expression levels (Fig. 2B). The combination of T0901317 and GW501516 elicited an additive effect on ABCA1 transcript and protein levels (Fig. 2C,D). Confocal fluorescence micrograph (x-z sections) of NRCs demonstrated that ABCA1 was localized within the basolateral membrane region (Fig. 2E). Expression of LXRβ and PPARδ was not affected by any treatment (data not shown).
Because ABCA1 mediates cholesterol efflux, we examined the effects of LXR and PPARδ ligands on cholesterol efflux from polarized NRCs cultured in a transwell plate assay (Fig. 2F). NRCs grown on polycarbonate membrane form monolayers with suitable polarization and retain the phenotypic and functional characteristics that define cholangiocytes in vivo.12 After labeling with [3H]-cholesterol for 24 hours, cells were treated with LXR ligand T0901317 and/or PPARδ ligand GW501516 for 24 hours. During the last 4 hours of ligand treatment, new media containing delipidated apoAI was added to apical and basolateral compartments as an acceptor for excreted cholesterol. Cholesterol release into the apical compartment was unaffected by either treatment (Fig. 2G). We observed an increased secretion of cholesterol into the basolateral compartment in response to increasing concentrations of T0901317 and the combination of T0901317 and GW501516 (Fig. 2G). Basolateral efflux was not increased by lower concentrations of T0901317 and GW501516 (1 μM) in the absence of the apoA1 acceptor protein, but was modestly elevated by higher concentrations. Thus, activation of LXRβ and PPARδ promote ABCA1-dependent basolateral cholesterol efflux in cholangiocytes.
PPARδ Induces NPC1L1 Expression in Cholangiocytes.
NPC1L1 mediates intestinal cholesterol uptake but is also expressed in the liver.7-9 We detected NPC1L1 transcripts in both the large and small cholangiocytes (Fig. 3A). Confocal fluorescence micrographs revealed that NPC1L1 protein was localized to the apical membrane (Fig. 3B,C) in NRCs. In addition, immunohistochemical analysis of rat liver sections revealed NPC1L1 protein throughout the bile duct cells (Fig. 4C).
PPARδ specifically induced NPC1L1. The PPARδ agonist (GW501516) induced the expression of NPC1L1 mRNA and protein within the cultured cholangiocytes (Fig. 3D,E). LXRβ agonist (T0901317) and PPARγ agonist (rosiglitazone) did not alter the expression level of NPC1L1. NPC1L1 expression was inhibited by specific depletion of PPARδ expression via RNAi, indicating that PPARδ is required for optimal expression in unstimulated conditions (Fig. 3F left); western blot analysis confirmed that RNAi eliminated PPARδ protein expression in cultured cholangiocytes (Fig. 3F, right).
To evaluate the role of NPC1L1 in cholesterol uptake in cholangiocytes, we examined the effects of GW501516 and the NPC1L1 inhibitor ezetimibe on cholesterol uptake in polarized NRCs. The treatment of cholangiocytes with GW501516 leads to an increase in apical cholesterol uptake, whereas the addition of ezetimibe inhibited this effect (Fig. 3G). This finding suggests that PPARδ may regulate NPC1L1 expression and NPC1L1-dependent cholangiocyte apical cholesterol uptake.
NPC1L1 regulates the absorption of cholesterol in the small intestine,7 so we speculated that dramatic increases in cholesterol as observed during cholestasis would have a significant impact on the expression of NPC1L1 in cholangiocytes. Accordingly, we observed an increase in both PPARδ and NPC1L1 expression in both large and small cholangiocytes after bile duct ligation (Fig. 4A-C).
Because cholangiocytes express CYP27A1, which converts cholesterol to oxysterols that act as LXR ligands, we also investigated whether cholesterol influx into cholangiocytes would elicit LXR-dependent changes in gene expression in NRCs. Multiple LXR target genes, including ABCA1, ABCG1, Sult2b1, ABCC1, and CYP46A1, were increased in NRCs in response to cholesterol loading (Fig. 4D). Similar changes in gene expression were observed after the treatment of NRCs with either oxysterols 27- and 22-hydroxysterols (data not shown). These data are consistent with the idea that oxysterol production is increased in cholangiocytes in response to elevations of cholesterol.
PPARδ Regulated NPC1L1 Expression in Cultured Cholangiocytes.
Three lines of evidence suggest that PPARδ directly regulates the NPC1L1 promoter. First, GW501516 increased the activity of an NPC1L1 promoter (−1045/−1)-driven luciferase reporter in cultured cholangiocytes (Fig. 5A, top). Second, PPARδ overexpression activated the NPC1L1 promoter, whereas PPARδ with a dominant negative in which the mutation impairs DNA binding had no effect (Fig. 5A, middle). Finally, PPARδ RNAi reduced basal NPC1L1 promoter activity (Fig. 5A, bottom).
We searched the NPC1L1 5′ flanking region for putative PPREs and identified one possible PPRE at −142 bp from the transcription start site (Fig. 5B). Mutation of this site eliminated GW501516-dependent increases in NPC1L1 promoter activity (Fig. 5C); this specific RXR-PPAR association with the PPRE was confirmed by gel shifts using nuclear extracts of transfected cells and in vitro–translated normal mouse cholangiocytes (NMCs) (Fig. 5D,E). Finally, ChIP assays revealed PPARδ association with the fragment that contains the PPRE in cholangiocytes (Fig. 5F). Thus, PPARδ regulates NPC1L1 expression by associating with a PPRE that lies at the 5′ location of the translational start site.
We profiled cholangiocyte expression of NRs and key enzymes and lipid transporters implicated in cholesterol and bile acid metabolism. Whereas hepatocytes express LXRα, LRH1, PPARα, and other NRs with established roles in liver cholesterol metabolism, cholangiocytes express LXRβ and PPARδ. Further, cholangiocytes expressed high levels of CYP27, which converts cholesterol to oxysterols that can serve either as precursors for bile acids or as LXR ligands. Thus, cholangiocytes and hepatocytes exhibit marked differences in the expression of NRs that modulate bile acid and cholesterol metabolism and gene products that regulate key steps in these processes. Although many NRs and lipid transporters are expressed at similar levels in large and small cholangiocytes, there are some differences in expression pattern, and further studies of differences in gene expression and protein function between small and large cholangiocytes will likely be of great interest.
Agonists for both LXRβ and PPARδ induced expression of the cholesterol pump ABCA1 in cholangiocytes. This is accompanied by increased cholesterol export at the cholangiocyte basolateral membrane. This action of LXRβ and PPARδ agonists resembles regulation of ABCA1-dependent cholesterol transport mechanisms in peripheral macrophages, where LXRβ and PPARs cooperate to promote reverse cholesterol transport from the periphery back to the liver.13 The LXRβ agonist also induces the cholesterol transporter in ABCG1 in cultured cholangiocytes. ABCG1 cooperates with ABCA1 by addition of additional lipids to apoAI, resulting in maturation of HDL particles.14 Thus, our data suggest that ABCA1 and ABCG1 could cooperate in basolateral extrusion of cholesterol from cholangiocytes.
Because ABCA1 mediates basolateral extrusion of cholesterol from cholangiocytes, we also studied mechanisms that could promote transport of cholesterol across the apical membrane from bile. Several studies indicated that the NPC1L1 cholesterol transporter may be expressed in the biliary system,9, 15 and we found that NPC1L1 transcripts and protein are expressed in cultured cholangiocytes and detected NPC1L1 protein in bile ducts in rat. We find that NPC1L1 is induced by PPARδ agonist (GW501516), and this effect is accompanied by GW501516-dependent increases in cholesterol uptake across the apical membrane of polarized cultured cholangiocytes. Moreover, these actions are blocked by ezetimibe, which targets NPC1L1. Evidence from NPC1L1 knockout mice is consistent with the idea that NPC1L1 mediates reabsorption of cholesterol from bile; these mice have a 41% increase in biliary cholesterol compared with wild-type mice.16
Together, our findings suggest a model for coordinated NR-dependent cholesterol flux through cholangiocytes (Fig. 6). First, PPARδ-dependent increases in NPC1L1 expression promote apical cholesterol import from bile into the cholangiocyte. Second, LXRβ/PPARδ-dependent increases in ABCA1 expression and, possibly, LXRβ-dependent increases in ABCG1 expression promote cholesterol efflux from the basolateral membrane.17 Because cholangiocytes express CYP27, it is likely that elevated levels of cholesterol cholangiocytes result in increased production of oxysterol LXRβ ligands, which would induce ABCA1/ABCG1-dependent cholesterol export from cells,18 and the fact that our experiments with cholesterol-loading into cultured cholangiocytes mimic effects of LXR agonist treatment supports this idea.
To our knowledge, our data represent the demonstration that PPARδ plays a direct role in NPC1L1 expression in cholangiocytes. There is evidence that PPARα activation results in decreased NPC1L1 expression in the intestine,11 but this effect appears to be indirect, because it is delayed and there is no identifiable PPARα effect on the NPC1L1 promoter. In this regard, it will be interesting to ask whether similar effects of PPARδ regulation of NPC1L1 with effects on cholesterol excretion occur in other cell types, including enterocytes.
We do not know why cholangiocytes exhibit the capacity for NR-dependent cholesterol flux, one possibility is that cholangiocytes prevent excess cholesterol accumulation in the bile, and that coordinated cholesterol flux protects cholangiocytes from high levels of intracellular cholesterol and its oxidized metabolites,19 which are toxic to cells20 due to induction of unfolded protein response21 and FAS-dependent apoptosis.22 We further recognize that strategies to manipulate vectorial cholesterol transport may be useful in cholestatic liver disease, in which local abnormalities in lipid metabolism promote cholangiocyte injury with severe toxic effects on local hepatocytes and increased levels of bile acids in the circulation, and possibly in treatment and prevention of gallstone disease, where increased concentrations of biliary cholesterol promote stone formation. Because NRs are major targets for pharmaceutical development, it is important to consider possible applications of NR ligands in these conditions. LXR ligand treatment alone is not likely to be useful, because LXR agonists increase cholesterol saturation index in the bile, leading to formation of cholesterol crystals in the gallbladder23; this lithogenic effect is caused by increased biliary efflux of cholesterol and phospholipids, a result of increased expression of hepatic cholesterol transporters Abcg5, Abcg8, and Abca1. However, it may be possible to use PPARδ ligands or combinations of PPARδ and LXRβ ligands to manipulate bile cholesterol. Because other NRs are expressed in cholangiocytes, it will also be interesting to understand effects of these receptors on cholesterol transport and reabsorption from bile.
The authors would like to thank Dr. Glenn E. Winner for critical reading of the manuscript. JAG is indebted to the Robert A. Welch Foundation for an endowment.