Peroxisome proliferator-activated receptor α activates human multidrug resistance transporter 3/ATP-binding cassette protein subfamily B4 transcription and increases rat biliary phosphatidylcholine secretion


  • Nisanne S. Ghonem,

    1. Department of Internal Medicine, Liver Center, Yale University School of Medicine, New Haven, CT
    Current affiliation:
    1. Department of Pharmaceutical Sciences, School of Pharmacy, MCPHS University, Boston, MA
    Search for more papers by this author
  • Meenakshisundaram Ananthanarayanan,

    1. Department of Internal Medicine, Liver Center, Yale University School of Medicine, New Haven, CT
    Search for more papers by this author
  • Carol J. Soroka,

    1. Department of Internal Medicine, Liver Center, Yale University School of Medicine, New Haven, CT
    Search for more papers by this author
  • James L. Boyer

    Corresponding author
    1. Department of Internal Medicine, Liver Center, Yale University School of Medicine, New Haven, CT
    • Address reprint requests to: James L. Boyer, M.D., Ensign Professor of Medicine, Yale University School of Medicine, 300 Cedar St., TAC S240, New Haven, CT 06519. E-mail:; fax: 203-785-7273.

    Search for more papers by this author

  • Potential conflict of interest: Nothing to report.

  • Supported by National Institutes of Health T32 Grant (to N.S.G.), DK 25636 (to J.L.B.) and P30-34989 (Yale Liver Center).


Multidrug resistance transporter 3/ATP-binding cassette protein subfamily B4 (MDR3/ABCB4) is a critical determinant of biliary phosphatidylcholine (PC) secretion. Clinically, mutations and partial deficiencies in MDR3 result in cholestatic liver injury. Thus, MDR3 is a potential therapeutic target for cholestatic liver disease. Fenofibrate is a peroxisome proliferator-activated receptor (PPAR) α ligand that has antiinflammatory actions and regulates bile acid detoxification. Here we examined the mechanism by which fenofibrate regulates MDR3 gene expression. Fenofibrate significantly up-regulated MDR3 messenger RNA (mRNA) and protein expression in primary cultured human hepatocytes, and stimulated MDR3 promoter activity in HepG2 cells. In silico analysis of 5′-upstream region of human MDR3 gene revealed a number of PPARα response elements (PPRE). Electrophoretic mobility shift (EMSA) and chromatin immunoprecipitation (ChIP) assays demonstrated specific binding of PPARα to the human MDR3 promoter. Targeted mutagenesis of three novel PPREs reduced inducibility of the MDR3 promoter by fenofibrate. In collagen sandwich cultured rat hepatocytes, treatment with fenofibrate increased secretion of fluorescent PC into bile canaliculi. Conclusion: Fenofibrate transactivates MDR3 gene transcription by way of the binding of PPARα to three novel and functionally critical PPREs in the MDR3 promoter. Fenofibrate treatment further stimulates biliary phosphatidylcholine secretion in rat hepatocytes, thereby providing a functional correlate. We have established a molecular mechanism that may contribute to the beneficial use of fenofibrate therapy in human cholestatic liver disease. (Hepatology 2014;59:1030–1042)


ATP-binding cassette protein subfamily B4


adenosine triphosphate




chenodeoxycholic acid


chromatin immunoprecipitation assay




direct repeats


electrophoretic mobility shift assay




farnesoid X receptor




multidrug resistance transporter 2


multidrug resistance transporter 3




primary biliary cirrhosis




primary cultured human hepatocytes


peroxisome proliferator-activated receptor


PPAR response element


primary sclerosing cholangitis


retinoic acid receptor


transcription start site


ursodeoxycholic acid



The multidrug resistance transporter 3 (MDR3, Mdr2/Abcb4 in rodent), encoded by the adenosine triphosphate-binding cassette superfamily (ATP)-binding cassette proteins subtype B4 (ABCB4), is predominantly expressed in the liver[1] and localized to the canalicular membrane of hepatocytes, where it is the major determinant of biliary phosphatidylcholine (PC) secretion.[2] Specifically, MDR3/ABCB4 functions as a floppase and translocates PC from the inner to the outer leaflet of the canalicular membrane for bile salt extraction.[3] Together with cholesterol and bile acids, PC forms biliary micelles, which inactivate the toxic detergent action of bile salts and prevent damage to epithelial cells lining the bile duct as well as inhibiting the formation of cholesterol gallstones. The functional importance of MDR3 was first demonstrated in Mdr2 knockout mice which lack biliary phospholipids and develop bile duct injury and progressive liver disease, closely resembling that of primary sclerosing cholangitis (PSC) in humans.[4, 5] Clinically, mutations and polymorphisms of MDR3 contribute to cholestatic liver injury, including a subtype of progressive familial intrahepatic cholestasis type 3, intrahepatic cholestasis of pregnancy, low phospholipid cholelithiasis, drug-induced and idiopathic chronic cholestasis.[6] Thus, MDR3 represents an important pharmacological target.

Chronic cholestasis, including primary biliary cirrhosis (PBC) and PSC, leads to liver fibrosis and cirrhosis, which eventually results in liver failure and the need for liver transplantation. The only therapeutic option available for these patients is ursodeoxycholic acid (UDCA), which slows the progression of PBC, particularly in stage I and II of the disease. However, some patients only partially respond to UDCA therapy, while more advanced cases usually do not respond. Furthermore, UDCA does not improve survival in patients with PSC, emphasizing the need for alternative therapies.

Peroxisome proliferator-activated receptors (PPAR) are ligand-activated transcription factors that belong to the superfamily of nuclear receptors. There are three PPAR isoforms: α, β/δ, and γ, and they regulate gene expression by forming a heterodimer with the retinoic acid receptor (RXR) and bind to peroxisome proliferator response elements (PPRE) containing direct repeats (DR-1) of the consensus sequence AGGTCA separated by a single nucleotide.[7] Fibrates are synthetic PPAR ligands used clinically for the treatment of dyslipidemia,[8] although each fibrate differs slightly in its specificity for the different PPAR subtypes.[9] PPARα is predominantly expressed in the liver and regulates the transcription of several genes involved in lipid metabolism. Interestingly, PPARα ligands, including fenofibrate, ciprofibrate, and clofibrate, reportedly up-regulate Mdr2 messenger RNA (mRNA) and protein expression[10, 11] in a PPARα-dependent manner.[10] Additionally, fenofibrate improves biomarkers of cholestasis following bile-duct ligation in mice.[12] PPARα has also been shown to inhibit the expression of genes involved in inflammation by negatively interfering with nuclear factor kappa B (NF-κB) signaling,[13] and fenofibrate up-regulates the human bile acid metabolizing enzymes of the UDP-glucuronosyltransferase family.[14-16] These actions have important and direct benefits for patients with chronic cholestatic liver disease. Clinical data shows that fenofibrate reduces symptoms and improves liver function abnormalities in some patients with PBC who do not fully respond to UDCA.[17] These data suggest that fenofibrate may have an important role as therapy for cholestatic liver disease, although the mechanism(s) responsible for its actions remain unknown.

This study examined the ability of fenofibrate to induce the expression of human MDR3 and the mechanisms that underlie this phenomenon. Our findings show that fenofibrate up-regulates the expression of MDR3 by activating PPARα. We further demonstrate that PPARα binds to specific PPREs in the human MDR3 promoter. We also show that fenofibrate increases canalicular excretion of fluorescently labeled PC. We present direct evidence for the first time that the human MDR3 gene is transactivated by PPARα and strongly suggest that PPARα up-regulates the expression of MDR3 to facilitate hepatic export of phospholipids. These findings provide a molecular mechanism by which fenofibrate may improve symptoms and liver function in patients with chronic cholestatic liver disease.

Materials and Methods

Materials, Animals, and Cell Cultures

See the Supporting Materials. Normal human hepatocytes (primary cultured human hepatocytes, PCHH), were obtained from the Liver Tissue Cell Distribution System (Pittsburgh, PA, funded by NIH Contract #N01-DK-7-0004/HHSN267200700004C) and maintained on collagen-coated plates in hepatocyte maintenance medium containing 1% penicillin-streptomycin-amphotericin B in a 37°C, 5% CO2 incubator. Treatment with the indicated agents was performed 24 hours after arrival. Rat hepatocytes were isolated from adult rats by collagenase perfusion as described.[18] Yale University's Institutional Animal Care and Use Committee (IACUC) approved all animal procedures and protocols.

Plasmid Constructs

The 5′-flanking regions of the MDR3 gene spanning 10 kb upstream of the transcription start site (TSS) were divided into five 2-kb fragments and amplified by polymerase chain reaction (PCR) using human genomic DNA or BAC clone 3147L21 (Invitrogen) with Pfu Ultra II Taq Polymerase. The upstream primers contained an internal Mlu I restriction site, the downstream primers an internal XhoI site. Following overnight digestion, PCR products were ligated into the pGL3-enhancer luciferase reporter gene vector, yielding the following MDR3 promoter constructs: p-1999/+1-luc, p-3999/-2000-luc, p-5999/-4000-luc, p-7999/-6000-luc, and p-9999/-8000-luc. Plasmid DNA was isolated using the Qiagen plasmid maxi kit. Mutagenesis of selected PPREs: -6775/-6797, -7197/-7219, and -8554/-8576 were performed using a QuikChange Site-Directed Mutagenesis kit (Stratagene). All primers are listed in Table 1.

Table 1. Oligonucleotide Primers for Human MDR3 Promoter Constructs, EMSA, ChIP, and Mutagenesis Assays
OligonucleotidePrimer Sequences
  1. a

    DR-1, Mut DR-1, and

  2. b

    APOA-IV, as previously reported,[19, 24] respectively. Positions of adaptor restriction enzyme sites for sub-cloning into the pGL3 enhancer vector appear in italics. Positions of mutated nucleotides in PPREs appear in bold.

p-3999/-2000 F5′-cg ACG CGT AGG GTT GTG GGT GTT TTC AGG-3′
p-3999/-2000 R5′-ccg CTC GAG GAA AGA AGG CTT GTG GTG ACT GG-3′
p-7999/-6000 F5′-cg ACG CGT TGT GAC AGG GAA GCA CTG AGT TC-3′
p-7999/-6000 R5′-ccg CTC GAG CAC CAG TGG GCT CAG TAA ATG TG-3′
p-9999/-8000 R5′-ccg CTC GAG AAT CAC CCA TCC CAG TTG TCG-3′





Quantitative Real-Time PCR Analysis (qRT-PCR)

Total RNA was extracted from PCHH and HepG2 cells using TRIzol (Invitrogen). TaqMan assays were used to detect message levels of human MDR3, BSEP, MDR1 (Applied Biosystems), and β-actin and MRP2 primers are listed in Table 2. qRT-PCR analysis was performed using a Roche LightCycler Detection System.

Table 2. Primer Sequences for Human ACTB and MRP2 for qRT-PCR Assays
OligonucleotideSequence (5′-3′)
  1. P = probe.

  2. a

    Dual-labeled probe, 5′-FAM/3′-BHQ.


Western Blot Analysis

Total protein expression of MDR3 from PCHH and HepG2 cells was measured by western immunoblotting. Proteins were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE; 7.5% TGX gel, BioRad), transferred to PVDF membranes, incubated overnight (4°C) with antihuman ABCB4 antibody detecting the P3II-26 epitope, then horseradish peroxidase (HRP)-conjugated secondary antibody. Immunodetection was performed using chemiluminescence (Thermo Scientific).

Cell Transfection and Luciferase Reporter Assays

HepG2 cells were cotransfected with 0.5 μg plasmid DNA, 0.1 μg each of PPARα and RXRα expression plasmid, and 5 ng pRL-CMV (Renilla luciferase plasmid) per well in complex with TransIT (Mirus) transfection reagent. At 16 hours posttransfection, fresh media containing charcoal-stripped FBS was added and cells were treated for an additional 24 hours with dimethyl sulfoxide (DMSO) or fenofibrate. Dual-Luciferase Reporter Assay (Promega) was used to measure luciferase activity.

Canalicular Membrane Localization of MDR3

HepG2 cells grown on glass coverslips were treated with the indicated agents for 48 hours. Cells were fixed in methanol for 10 minutes, blocked in 1% bovine serum albumin (BSA) for 20 minutes, then incubated overnight (4°C) with antibodies to MDR3 (Abcam) and Villin (Santa Cruz). Alexa488 antimouse immunoglobulin G (IgG) and Alexa555 antigoat IgG (Molecular Probes) were used as secondary antibody. Fluorescent images were taken in a blinded manner and visualized by confocal microscopy (Carl Zeiss).

Electrophoretic Mobility Shift Assay (EMSA)

Human PPARα and RXRα plasmids were in vitro transcribed and translated using the TNT T7 quick-coupled transcription/translation system (Promega). Oligonucleotides are listed in Table 1. EMSA assays were performed using the DIG Gel Shift kit (Roche). Per reaction, 2 μg of hPPARα and hRXRα protein was incubated in binding solution containing 15 pmol of digoxigenin (DIG)-labeled probes for 15 minutes (room temperature). Competition experiments were performed using a 100-fold molar excess of unlabeled oligonucleotide of the consensus DR-1 sequence or a mutated DR-1[19] added to the reaction 15 minutes before addition of DIG-probe. For supershift assays, 4 μg of anti-PPARα antibody (N-19) was added to the reaction on ice for 30 minutes before addition of DIG-probe. DNA-protein complexes were resolved on a 6% polyacrylamide gel (Invitrogen) and transferred to a positively charged nylon membrane. The blot was incubated with anti-digoxigenin conjugated with alkaline phosphatase, detected using a chemiluminescence substrate, and exposed to x-ray film.

Chromatin Immunoprecipitation (ChIP) Assay

ChIP assays were performed using HepG2 cells and PCHH (Millipore-Upstate, Temecula, CA). HepG2 cells were cotransfected with PPARα and RXRα 16 hours before treatment with DMSO or fenofibrate. Chromatin samples were lysed and sheared into fragments of 150 and 900 bp by sonication and precleared using Protein G Agarose/Salmon Sperm DNA for 1 hour (4°C). Precleared chromatin samples were incubated overnight (4°C) with 5 μg of anti-PPARα (H-98X), rabbit IgG, or RNA polymerase II (8WG16) antibody. Antibody-chromatin complexes were captured by incubation with Magna Protein G Agarose beads (Millipore). Following washings of the agarose beads, reversal of cross-linking, and purification, qRT-PCR of the precipitated DNA was performed for each reaction. Primers flanking the PPRE sites are listed in Table 1.

Canalicular Phosphatidylcholine (NBD-PC) Secretion

Primary rat hepatocytes plated on collagen-coated glass coverslips were used for live cell imaging. Hepatocytes were cultured in collagen sandwich configuration and after 48 hours of treatment with DMSO or fenofibrate cells were cooled on ice for 20 minutes then labeled with NBD-PC for 20 minutes on ice followed by a 30-minute incubation at 37°C for internalization of the fluorescent lipid. Afterwards, coverslips were washed three times with HEPES buffer on ice and observed by fluorescent microscopy using a Zeiss LSM 510 Microscope (Oberkochen, Germany).

Excretion of Fluorescent Bile Salt Analogs in BC

See the Supporting Methods.

Statistical Analysis

Data are expressed as mean ± SD. Statistical analyses of differences were carried out using the Student t-test or one-way analysis of variance (ANOVA), followed by Dunnett's post-hoc multiple comparisons using GraphPad Prism. Differences with P < 0.05 were considered statistically significant.


PPARα Activators Induce MDR3 mRNA Expression

To determine if fibrates regulate human MDR3 mRNA expression, PCHH and HepG2 cells were treated for 24 hours with fenofibrate (FF), gemfibrozil (GF), bezafibrate (BF), and chenodeoxycholic acid (CDCA), a farnesoid X receptor (FXR) agonist reported to regulate MDR3,[20] and qRT-PCR analysis was performed. Treatment with Wy-14,643 (Wy), a PPARα agonist, served as positive control. In PCHH, FF had the strongest effect, compared to the other fibrates tested, and significantly increased MDR3 mRNA expression by 6.5-fold (Fig. 1A). In HepG2 cells, treatment with FF also significantly increased MDR3 mRNA levels, albeit to a lesser extent than in PCHH (Fig. 1B). This effect is likely due to lower levels of endogenously expressed PPARα in HepG2 cells.[21] Similar findings of fibrates displaying varying potency have been reported.[11] In PCHH, induction of MDR3 mRNA expression by FF was dose- (Fig. 1C) and time-dependent (Fig. 1D).

Figure 1.

Quantitative RT-PCR analysis reveals that fenofibrate significantly up-regulates MDR3 mRNA expression in (A) PCHH (50 μM) and (B) HepG2 cells (125 μM) treated for 24 hours. In PCHH, fenofibrate induced MDR3 mRNA expression in a (C) dose- and (D) time-dependent manner. Wy-14,643 (Wy, 10 μM), bezafibrate (BF), fenofibrate (FF), gemfibrozil (GF), and chenodeoxycholic acid (CDCA), each 50 μM. Results were normalized to β-actin using the 2-ΔΔCt method and expressed as fold-increase over DMSO (vehicle control). Data are expressed as the mean ± SD (n = 3-4 individual cases of PCHH), *P < 0.05, **P < 0.01 versus DMSO.

To determine if the induction by FF is specific for the MDR3 gene, the mRNA levels of other canalicular membrane transport proteins, including MRP2, BSEP, and MDR1 in PCHH were also measured by qRT-PCR. Treatment with FF or Wy-14,643 did not significantly up-regulate the mRNA levels of MRP2, BSEP, or MDR1, confirming the specificity of FF on MDR3 (Supporting Fig. 1). Also, CDCA and UDCA increased MDR3 mRNA expression by only 2- and 1-fold, respectively. These results demonstrate that fenofibrate specifically up-regulates human MDR3 mRNA expression and, further, indicates that PPARα is likely to be involved in human MDR3 gene transcription.

Fenofibrate Up-Regulates Human MDR3 Protein Expression

The effect of fibrates on human MDR3 protein expression was assessed after 48 hours of treatment by western immunoblotting. FF and Wy significantly up-regulated human MDR3 protein expression by 3.2- and 2.9-fold, respectively, in PCHH versus DMSO-treated cells (Fig. 2A). Similarly, effects were observed in HepG2 cells, although not reaching statistical significance (Fig. 2B). The discrepancy between MDR3 message and protein levels in HepG2 cells suggests posttranscriptional regulation of human MDR3 expression.

Figure 2.

Western blot analysis shows that fenofibrate up-regulates MDR3 protein expression in (A) PCHH (50 μM) and (B) HepG2 cells (125 μM) treated for 48 hours. (C,D) Representative immunoblotting, respectively. Data were normalized to SH-PTP1 protein levels. Wy-14,643 (Wy, 10 μM), bezafibrate (BF), fenofibrate (FF), gemfibrozil (GF) and chenodeoxycholic acid (CDCA, 50 μM). Protein band intensity was analyzed using FOTODYNE Imaging and TL-100 image analysis software. Data are expressed as mean ± SD (n = 3-4), **P < 0.01 versus DMSO.

PPARα Stimulates Human MDR3 Expression at the Canalicular Membrane

The pseudo-canalicular expression and localization of human MDR3 in HepG2 cells was investigated by immunofluorescent microscopy. To ensure identification of the apical membrane, colocalized expression of MDR3 (green) plus villin (red), a marker of BC,[22] was visualized in orange (Fig. 3A,B). Following 48 hours of treatment with DMSO, Wy, BF, FF, GF, or CDCA, fenofibrate significantly up-regulated the colocalized expression of human MDR3 and villin by 3-fold (Fig. 3C). FF treatment did not affect villin expression (Fig. 3D). These findings suggest that fenofibrate up-regulated canalicular localization of human MDR3.

Figure 3.

Colocalized human MDR3 and villin canalicular membrane expression in HepG2 cells. To ensure pseudo-canalicular membrane localization, only staining of colocalized MDR3 + villin were quantified. (A) DMSO- or (B) fenofibrate-treated indicate MDR3 (left, green), villin (middle, red), and arrows (white) indicate colocalized MDR3+villin (right, orange), respectively. (C) Quantification of colocalized MDR3+villin canalicular membrane expression from the average of six images per coverslip per experiment. Data are expressed as mean ± SD (n = 4-5),*P < 0.05, **P < 0.01 versus DMSO. (D) Representative immunoblotting of MDR3, SH-PTP1, and Villin protein. Wy-14,643 (Wy, 10 μM), bezafibrate (BF), fenofibrate (FF), gemfibrozil (GF), each 125 μM, and chenodeoxycholic acid (CDCA, 50 μM). Images were acquired in a blinded manner, visualized with a Zeiss LSM510 confocal scanning microscope and processed with Adobe PhotoshopCS (Mountain View, CA).

Fenofibrate Activates the Human MDR3 Promoter

In silico analysis of the human MDR3 promoter ( identified 21 putative PPREs. To determine which, if any, of these PPREs activate human MDR3 transcription, we performed reporter assays using HepG2 cells transiently transfected separately with five 2-kb MDR3 promoter-luciferase constructs (spanning a 10 kb upstream region) and cotransfected with human PPARα and RXRα expression plasmids. We and others[16, 21] have found it necessary to transfect PPARα together with RXRα expression plasmids in the reporter gene activity assays due to low endogenous levels of expression of these proteins in HepG2 cells. Twenty-four hours after treatment with fenofibrate, MDR3 promoter activity was significantly increased in the p-7999/-6000 and p-9,999/-8000 regions (Fig. 4A). Cotransfection with PPARα and RXRα and treatment with DMSO (vehicle control) also increased MDR3 promoter activity, which was further up-regulated by fenofibrate, suggesting the presence of endogenous ligands capable of activating PPARα when exogenously expressed. Fenofibrate enhanced luciferase activity when the rat acetyl-CoA oxidase (AOX) promoter, which contains an active PPRE,[23] and when a reporter gene driven by three copies of the consensus DR-1 cloned in front of the TK promoter (tk-PPREx3-luc), were cotransfected with PPARα and RXRα expression plasmids in HepG2 cells (Fig. 4B), serving as positive controls and thereby confirming activation of human MDR3 promoter by PPARα.

Figure 4.

PPARα activates human MDR3 promoter in a reporter luciferase assay. (A) Schematic showing in silico analysis of 5′-upstream region of human MDR3 gene and the localized 21 PPREs. (B) pGL3e-ABCB4-luc constructs were transfected together with hPPARα and hRXRα expression plasmids in HepG2 cells and treated with DMSO or FF (125 μM) for 24 hours. Data are expressed as the mean ± SD (n = 3), *P < 0.05 versus DMSO. (C) Reporter assay of tk-PPREx3-luc and rat AOX promoter show significant activation by PPARα following fenofibrate treatment, as positive controls. MDR3 promoter activity was calculated by normalizing the firefly luminescence to the Renilla luminescence signal. Data are expressed as the average fold-increase over DMSO (vehicle control) ± SD (n = 3), *P < 0.05 versus DMSO.

To determine whether activation of MDR3 by PPARα is due to binding of the predicted PPREs to the human MDR3 promoter, we performed an EMSA assay using PPARα and RXRα proteins and oligonucleotide probes corresponding to the PPRE at -7197/-7219 bp upstream of the TSS. This specific PPRE was selected due to the increased luciferase activity within the -6/-8 kb region, as well as its very high sequence similarity to the consensus sequence. Absence of the DIG-probe confirmed no interfering proteins in the TNT mix (Fig. 5, lane 1). A PPARα/RXRα heterodimer bound to the DIG-PPRE, which resulted in a bound DNA-protein complex (lane 2). Band shift was completely abolished by adding unlabeled competitor corresponding to the DR-1 sequence for the PPRE motif (lane 3). In contrast, band shift was not affected by the addition of unlabeled mutated DR-1 competitor (lane 4). Incubation with PPARα antibody (sc-1985) resulted in super shift (lane 5), demonstrating specificity of the bound complex. As a positive control, oligonucleotides corresponding to the consensus DR-1 sequence were also labeled with digoxigenin, incubated with protein, and tested in competition assays (Supporting Fig. 2). These results strongly indicate that PPARα is bound to the human MDR3 promoter at -7197/-7219 bp upstream of the TSS.

Figure 5.

Electrophoretic mobility shift assays were performed using in vitro translated protein and a digoxigenin-labeled probe corresponding to the human MDR3 promoter containing the PPRE located at -7219/-7197 bp upstream of the TSS. Incubation with PPARα/RXRα protein alone did not result in band shift (lane 1). Incubation of protein with DIG-PPRE resulted in a band shift (lane 2). Competitive analysis was performed in the presence of a 100-fold molar excess of unlabeled wild-type consensus DR-1 (wt, lane 3) and mutated DR-1 (m, lane 4) as indicated. Addition of the PPARα antibody super shifted the bound PPRE complex (lane 5), and confirmed specificity of PPARα-mediated complex formation. The positions of the DNA-protein complexes are indicated by the arrow.

To examine the binding of PPARα to the human MDR3 promoter in vivo, we performed ChIP assays and probed with oligonucleotides flanking PPRE motifs at p-6775/-6797, p-7197/-7219, and p-8554/-8576 bp upstream of the TSS. HepG2 cells and PCHH were fixed and total chromatin was extracted 48 hours after treatment with DMSO or FF. The binding of PPARα to the MDR3 promoter in HepG2 cells and in PCHH was analyzed by qRT-PCR (Fig. 6A,B, respectively). Cells treated with fenofibrate and immunoprecipitated with a PPARα antibody resulted in amplicons spanning the -6775/-6797, -7197/-7219, and -8554/-8576 regions of the MDR3 promoter (Fig. 6C-E, respectively). In contrast, in DMSO-treated cells little to no PPARα binding was detected. We also analyzed the binding of PPARα to the human APOA-IV promoter, which contains an active PPRE[24] and observed a 6-fold enrichment in fenofibrate-treated HepG2 cells. RNA Polymerase II antibody was used as a control for active gene expression and a band was observed in all samples. As negative controls, no enrichment was detected when we used primers to amplify a region of the MDR3 gene that does not contain any reported PPREs (Fig. 6F). These findings demonstrate that PPARα was specifically recruited by FF and bound to -6775/-6797, -7197/-7219, and -8554/-8576 sites in the MDR3 promoter. These results strongly indicate that PPARα activates human MDR3 transcription by way of specific binding to upstream PPREs.

Figure 6.

ChIP assay demonstrates that PPARα directly binds to the PPREs located at -6775/-6797 bp, -7197/-7219 bp, and -8554/-8576 upstream regions of human MDR3 promoter in vivo. Chromatins prepared from (A) HepG2 cells (n = 3) and (B) PCHH (n = 1) treated with DMSO or FF (125 μM) for 48 hours were immunoprecipitated with antibodies against rabbit IgG, PPARα, and RNA polymerase II (control); 1% of the chromatin was saved as input. qRT-PCR detection was performed using primers for the -6775/-6797, -7197/-7219, and -8554/-8576 regions of the MDR3 promoter. The human APOA-IV gene was used as positive control. Representative qRT-PCR reaction was electrophoresed through a 2.5% agarose gel and stained with ethidium bromide to confirm the amplicon size. (C) -6797/-6775 bp, (D) -7219/-7197 bp, (E) -8576/-8554 bp, and (F) region containing no predicted PPRE, as negative control, upstream of the TSS. The positions of a 100-bp DNA ladder are indicated on the right. Data were normalized as ratio of PPARα to IgG and expressed as fold-enrichment (mean ± SD), *P < 0.05 versus DMSO. In HepG2 cells, each qPCR experiment was performed in duplicate and three independent experiments were performed.

Mutation of PPREs Abolished Fenofibrate-Mediated Activation of the MDR3 Promoter

To further examine the functional importance of the three novel PPREs identified within the human MDR3 promoter, site-directed mutagenesis of PPREs located at -6775/-6797 bp, -7197/-7219 bp, and -8554/-8576 bp upstream of the TSS was preformed in the context of the p-7999/-6000-luc and p-9999/-8000 constructs. These three particular PPREs were selected due to increased luciferase activity within the -7999/-6000kb and -9999/-8000 region and their high sequence similarity to the consensus sequence. Within each of the 2-kb sequences, however, several PPREs were predicted. Thus, it was possible that another active PPRE might compensate and maintain promoter activity when a single site was mutated. To address this possibility, a double-site mutation (DSM) was constructed for the elements at -6775/-6797 bp plus -7197/-7219 bp upstream of the TSS. HepG2 cells were transiently cotransfected with PPARα and RXRα expression plasmids and either wild-type or mutant construct. After 16 hours of transfection, fresh media containing DMSO or FF was applied for 24 hours and reporter assays were performed. Mutation of p-6775/-6797, p-7197/-7219, and DSM completely abolished PPARα-induced MDR3 promoter activity (Fig. 7A), while mutation of p-8554/-8576 reduced promoter activity (Fig. 7B). These results indicate that MDR3 promoter activation is mediated by the PPREs located at -6775/-6797, -7197/-7219, and -8554/-8576 bp regions and that mutation of these PPREs greatly reduced MDR3 transcription, demonstrating the functional importance of these three novel PPREs.

Figure 7.

Reporter assay demonstrates that mutation of PPREs abolishes MDR3 promoter activity. Mutated (A) -6775/-6797 bp, -7197/-7219 bp, DSM (-6775/-6797 plus -7197/-7219 bp), and (B) -8554/-8576 bp and their corresponding wild-type (WT) constructs were cotransfected with hPPARα/RXRα expression plasmids into HepG2 cells for 16 hours, then fresh media containing DMSO or FF (125 μM) was applied for 24 hours. Renilla luminescence was used to normalize the firefly luminescence. The results are presented as fold-change, by setting the firefly luciferase activity of DMSO-treated cells to 1.0. Data are expressed as the mean ± SD (n = 3). *P < 0.05 and **P < 0.01 versus DMSO.

Fenofibrate Stimulates Canalicular PC Secretion in Primary Rat Hepatocytes

Phosphatidylcholine is localized to the inner leaflet of the membrane bilayer and is excreted into the biliary canaliculi by the action of MDR3.[3] Primary rat hepatocytes cultured in a collagen sandwich form closed bile canalicular spaces which permits assessment of canalicular excretion of fluorescent substrates.[25] To determine if transactivation of the Mdr2 gene resulted in an increase of Mdr2 function, we examined the biliary secretion of NBD-PC in primary rat hepatocytes by fluorescent microscopy after exposure of the treated cultures to fluorescent NBD-PC. After 48 hours of treatment, DMSO-treated cells showed positive canalicular membrane labeling (arrows) as well as distinct diffuse cytoplasmic staining, with varying degrees of fluorescence within the BC lumen (Fig. 8A). In contrast, fenofibrate-treated cells showed a stronger and more localized fluorescence, specifically within the BC lumen (Fig. 8B). Quantitative analysis of fluorescent intensity, performed in a blinded manner, revealed a significantly greater percentage of positive BC in the fenofibrate-treated cells, versus DMSO-treated cells (37 ± 9% versus 18 ± 5.5%, respectively, Fig. 8C). Active secretion of the fluorescent bile acid CGamF further demonstrated that bile excretory function was intact in these isolated hepatocytes (Supporting Fig. 3). These results demonstrate that fenofibrate treatment enhances canalicular PC secretion and suggests that PPARα increases Mdr2 function. The rat Mdr2 gene is ∼90% similar to the nucleotide sequence of the human MDR3 gene[26] and thus it is plausible to extend these findings to that of MDR3 function in human hepatocytes.

Figure 8.

Fluorescently labeled phosphatidylcholine (NBD-PC) is observed within bile canaliculi (BC) of primary rat hepatocytes cultured for 3-4 days in collagen sandwich gel configuration. At 48 hours of treatment with DMSO (0.1%) or FF (125 μM), hepatocytes were labeled with NBD-PC (4 μM) and incubated at 37°C for 30 minutes. Seven images per coverslip were obtained in a blinded fashion using a Zeiss LSM 510 Microscope equipped with a 63× objective and a green fluorescence filter (FT 510 nm dichroic mirror and LP 505 nm emission filter, Carl Zeiss) and the aperture set to 3 μm optical space. Areas of the coverslip with BC were first selected using brightfield phase optics, then scanned for NBD-PC using 30% laser power. Using Photoshop, the BC were delineated in the phase-contrast image and this layer was superimposed over the identical fluorescent image. (A) DMSO-treated cells showed positive canalicular membrane labeling of NBD-PC (left, arrows) as well as distinct diffuse cytoplasmic staining. (B) Fenofibrate-treated cells showed a stronger and more localized fluorescent PC within the lumen of bile canaliculi (left). BC were then delineated in the phase-contrast image (middle) and then the outlines were superimposed over the identical fluorescent image (right), respectively. (C) Quantitative analysis of the fluorescent intensity within the bile canaliculi revealed a significantly greater percentage of positive BC in the fenofibrate-treated cells, compared with DMSO-treated cells. The mean fluorescent intensity of the luminal space was determined using ImageJ software. Data reflect the fluorescent intensity calculated from 180-200 bile canaliculi collected from three independent experiments. Data represent mean ± SD (n = 3), *P < 0.05 versus DMSO. Scale bar = 25 μm.


Numerous case reports and pilot studies have demonstrated that bezafibrate and fenofibrate (reviewed[27]) can reduce serum biomarkers of cholestasis and liver function abnormalities in patients with PBC or PSC who have experienced an incomplete response to UDCA (13-15 mg/kg/day) monotherapy. The majority of these studies showed efficacy, yet the mechanism(s) by which fibrates reduce biochemical markers of cholestasis remains unclear. We hypothesized that fenofibrate's effect on liver function might be mediated by up-regulating human MDR3/ABCB4 gene expression by way of PPARα. Collectively, the findings in the present study demonstrate that the human MDR3 gene is directly transactivated by PPARα by way of the PPREs located in the MDR3 promoter and, further, suggest that PPARα regulates MDR3/Mdr2 function leading to enhanced hepatic export of phosphatidylcholine.

Shoda et al.[28] previously reported that bezafibrate treatment increased MDR3 localization and NBC-PC within pseudo-canaliculi of HepG2 cells despite a lack of change in the level of MDR3 protein. This finding was based on an increase in fluorescence over time, not necessarily specific for canalicular secretion of PC. Also, bezafibrate increased the expression of MRP2 and BSEP mRNA, two major canalicular transporters, to the same degree as MDR3.[28] PC is a phospholipid compound and a major component of biological membranes, whose canalicular secretion of PC is a function exclusively of MDR3/Mdr2 expression. While our studies also do not provide in vivo evidence that fenofibrate increases phospholipid excretion into bile, we were able to demonstrate increases in the fluorescent intensity of NBD-PC secreted into the canalicular lumen, using primary rat hepatocytes in a collagen sandwich culture, which closely mimic the in vivo situation. These findings show that fenofibrate increased biliary PC secretion 2-fold, versus untreated controls (Fig. 8B). Increased biliary excretion of PC would be expected to enhance micelle solubilization of bile acids and thus suggest that fenofibrate improves biochemical markers of cholestasis by enhancing the excretion of phosphatidylcholine into bile.

Nevertheless, PPARα is a ligand-activated transcription factor with a wide range of additional actions including cholesterol homeostasis, primarily by way of down-regulation of bile acid synthesis through cytochrome P450 (CYP) enzyme 7A1[23] and detoxification by way of up-regulation of CYP8B1,[29] CYP3A4 and other members of the CYP450 and UDP-glucuronosyltransferases family of enzymes,[14, 16, 30] as well as SULT2A1[31] and ASBT.[19]

In rodents, there is conflicting data on the effects of fibrates on expression of Mdr2 mRNA. Ciprofibrate and Wy-14,643 treatment up-regulated Mdr2 mRNA expression in primary hepatocytes and liver tissue of wild-type mice but not those of Pparα-null mice, yet wild-type mice displayed an increased bile flow in spite of reduced biliary phospholipid and bile salt concentrations.[10] However, immunofluorescent analysis of Mdr2 protein was not performed in that study due to a reported lack of specific antibodies. In contrast, Chianale et al.[11] reported that a 0.5% fenofibrate diet increased Mdr2 mRNA levels but did not increase biliary phospholipid output in wild-type mice, while clofibrate and ciprofibrate increased biliary phospholipid levels in wild-type animals, but not in Pparα-null mice. Interestingly, Miranda et al.[32] also reported that phthalates and fish oil, both Ppar activators, up-regulated Mdr2 mRNA and protein expression in mouse liver. The latter two studies,[11, 32] however, used the C219 antibody to detect an epitope common to all isoforms of the Mdr family, and thus are not specific for Mdr2. In fact, the detected band (160-180 kDa) is closer to that of Mdr1 (170 kDa) than to Mdr2 (140 kDa). These effects of clofibrate and ciprofibrate on phospholipid output were independent of changes in bile acid output and treatment and did not alter the bile acid pool composition.[11] In contrast, Li et al.[33] found that Pparα plays a crucial role not only in bile acid synthesis, but also in transport and secretion, as demonstrated by a reduced phospholipid concentration and mRNA expression of Abcb4, Abca1, Abcg5, and Abcg8 in Pparα-null mice. Significant changes in the bile acid pool composition were also reported, resulting in disrupted bile acid homeostasis in cholic acid fed Pparα-null mice versus cholic acid-treated wild-type and control animals.[33]

In human liver biopsy samples, MDR3 mRNA is increased by fibrate treatment.[34] Importantly, PPARα also reduces inflammation, a hallmark of liver injury,[13] and Tnfα mRNA expression is significantly increased in Pparα-null mice.[33] In addition, fenofibrate has significant effects on serum bile acids and the bile acid pool composition in noncholestatic adult volunteers, resulting in a reduction in circulating toxic primary and secondary bile acids, including CDCA, deoxycholic acid, and lithocholic acid.[35] To clarify these discrepancies on the effects of fibrates on rodent versus human MDR3/Mdr2 mRNAs, we performed in silico analysis of the mouse Abcb4 promoter and found 25 Ppar response elements located within the 10 kb upstream region. In particular, in the 6-10 kb upstream region, there are three Ppres located very close to the three PPREs reported in the current study (Supporting Table 1). These findings support the conclusion that Pparα directly regulates Abcb4 expression in both rodents and humans. The opposite effects of other fibrates are likely due to the result of species-specificity or affinity of the different fibrates for PPARα/Pparα, as previously reported.[36] Lastly, mRNA data in the present study were obtained using primary human hepatocytes cultured in a sandwich configuration between two layers of gelled extracellular matrix proteins, which is considered the preferred model for the study of xenobiotic metabolism and transport.[37]

Nevertheless, it is likely that fenofibrate has additional beneficial effects that complement its stimulation of biliary phosphatidycholine excretion and together account for the benefit of fenofibrate therapy observed in patients with chronic cholestatic liver disease. For example, Huang et al.[20] identified an FXR response element in the proximal region of the human MDR3 promoter and showed a corresponding increase in MDR3 mRNA levels following treatment with CDCA. Despite these findings, FXR may not be a direct activator of Abcb4 regulation since mice with a homozygous deletion of the FXR gene fed a cholic acid-enriched diet showed an increase in Mdr2 mRNA, similar to control mice.[38] These opposing results introduce the possibility that FXR-mediated effects may occur by way of PPARα activation. Indeed, Pineda Torra et al.[39] identified an FXR response element within the proximal region of the human PPARα promoter and showed that combination treatment with CDCA and GW7647, a PPARα agonist, resulted in enhanced PPARα activation compared to either agent alone. However, in the present study fenofibrate treatment increased MDR3 mRNA and protein expression to a greater extent compared to CDCA, and synergist effects of CDCA and fenofibrate were not observed (Supporting Fig. 4). Interestingly, many of the actions reported by PPARα, such as up-regulation of UGT2B4[40] and down-regulation of CYP7A1, are also described for FXR.

In summary, the current study provides the first evidence that PPARα transactivates human MDR3 and adds further to our understanding of the transcriptional control of MDR3 gene expression. It is likely that fenofibrate improves cholestatic liver function by increasing the biliary excretion of phosphatidylcholine by this mechanism in addition to other beneficial actions of PPARα activation in liver. Altogether, these findings strongly support the use of PPARα agonists as therapeutic alternatives for patients with cholestatic liver diseases, specifically those with an incomplete response to UDCA.


We thank Dr. Shi-Ying Cai and Ms. Kathy Harry for technical support.