Transcriptional dynamics of bile salt export pump during pregnancy: Mechanisms and implications in intrahepatic cholestasis of pregnancy

Authors

  • Xiulong Song,

    1. Department of Biomedical and Pharmaceutical Sciences, Center for Pharmacogenomics and Molecular Therapy, College of Pharmacy, University of Rhode Island, Kingston, RI
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Alexander Vasilenko,

    1. Department of Biomedical and Pharmaceutical Sciences, Center for Pharmacogenomics and Molecular Therapy, College of Pharmacy, University of Rhode Island, Kingston, RI
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Yuan Chen,

    1. Department of Biomedical and Pharmaceutical Sciences, Center for Pharmacogenomics and Molecular Therapy, College of Pharmacy, University of Rhode Island, Kingston, RI
    Search for more papers by this author
  • Leila Valanejad,

    1. Department of Biomedical and Pharmaceutical Sciences, Center for Pharmacogenomics and Molecular Therapy, College of Pharmacy, University of Rhode Island, Kingston, RI
    Search for more papers by this author
  • Ruchi Verma,

    1. Department of Biomedical and Pharmaceutical Sciences, Center for Pharmacogenomics and Molecular Therapy, College of Pharmacy, University of Rhode Island, Kingston, RI
    Search for more papers by this author
  • Bingfang Yan,

    1. Department of Biomedical and Pharmaceutical Sciences, Center for Pharmacogenomics and Molecular Therapy, College of Pharmacy, University of Rhode Island, Kingston, RI
    Search for more papers by this author
  • Ruitang Deng

    Corresponding author
    1. Department of Biomedical and Pharmaceutical Sciences, Center for Pharmacogenomics and Molecular Therapy, College of Pharmacy, University of Rhode Island, Kingston, RI
    • Address reprint requests to: Ruitang Deng, Ph.D., Department of Biomedical and Pharmaceutical Sciences, Center for Pharmacogenomics and Molecular Therapy, College of Pharmacy, University of Rhode Island, 7 Greenhouse Road, Kingston, RI 02881. E-mail: DengR@mail.uri.edu; fax: 401-874-5787.

    Search for more papers by this author

  • Potential conflict of interest: Nothing to report.

  • Supported by the National Institutes of Health Grant R01DK087755. B.Y. is supported by the National Institutes of Health Grants R01GM61988, R01ES07965, and R15AT007705.

  • See Editorial on Page 1815

Abstract

Bile salt export pump (BSEP) is responsible for biliary secretion of bile acids, a rate-limiting step in the enterohepatic circulation of bile acids and transactivated by nuclear receptor farnesoid X receptor (FXR). Intrahepatic cholestasis of pregnancy (ICP) is the most prevalent disorder among diseases unique to pregnancy and primarily occurs in the third trimester of pregnancy, with a hallmark of elevated serum bile acids. Currently, the transcriptional regulation of BSEP during pregnancy and its underlying mechanisms and involvement in ICP are not fully understood. In this study the dynamics of BSEP transcription in vivo in the same group of pregnant mice before, during, and after gestation were established with an in vivo imaging system (IVIS). BSEP transcription was markedly repressed in the later stages of pregnancy and immediately recovered after parturition, resembling the clinical course of ICP in human. The transcriptional dynamics of BSEP was inversely correlated with serum 17β-estradiol (E2) levels before, during, and after gestation. Further studies showed that E2 repressed BSEP expression in human primary hepatocytes, Huh 7 cells, and in vivo in mice. Such transrepression of BSEP by E2 in vitro and in vivo required estrogen receptor α (ERα). Mechanistic studies with chromatin immunoprecipitation (ChIP), protein coimmunoprecipitation (Co-IP), and bimolecular fluorescence complementation (BiFC) assays demonstrated that ERα directly interacted with FXR in living cells and in vivo in mice. Conclusion: BSEP expression was repressed by E2 in the late stages of pregnancy through a nonclassical E2/ERα transrepressive pathway, directly interacting with FXR. E2-mediated repression of BSEP expression represents an etiological contributing factor to ICP and therapies targeting the ERα/FXR interaction may be developed for prevention and treatment of ICP. (Hepatology 2014;60:1992–2006)

Abbreviations
ABCB11

ATP-binding cassette subfamily B member 11

BiFC

bimolecular fluorescence complementation

BSEP

bile salt export pump

CDCA

chenodeoxycholic acid

ChIP

chromatin immunoprecipitation

Co-IP

coimmunoprecipitation

DBD

DNA binding domain

DMSO

dimethyl sulfoxide

E2

17β-estradiol

eGFP

enhanced green fluorescence protein

ERα

estrogen receptor-α

ERβ

estrogen receptor-β

ERE

estrogen response element

EtOH

ethanol

FACS

fluorescence activated cell sorting

FXR

farnesoid x receptor

FXRE

FXR response element

GAPDH

glyceraldehyde 3-phosphate dehydrogenase

IACUC

institutional animal care and use committee

ICP

intrahepatic cholestasis of pregnancy

IR1

inverted repeat spaced by one nucleotide

IVIS

in vivo imaging system

LBD

ligand binding domain

MDR3

multidrug resistance protein 3

NR1H4

nuclear receptor subfamily 1 group H member 4

SHP

small heterodimer partner

WT

wild-type

Pregnancy causes profound maternal changes in physiology and metabolism.[1-4] One such change is the serum lipid profile with elevated bile acids, cholesterol, and triglycerides, especially in the third trimester of gestation.[1-3, 5-7] Most of those changes either directly or secondarily result from elevated pregnancy-related hormones, mainly estrogens, such as 17β-estradiol (E2) and progesterone. In a subgroup of pregnant women, when their bodies cannot adequately adapt to elevated reproductive hormones, pregnancy-related diseases occur.[8, 9]

Intrahepatic cholestasis of pregnancy (ICP) is the most prevalent disorder among diseases unique to pregnancy.[8-10] The incidence of ICP varies widely among ethnic groups, ranging from less than 1% to 14%.[11, 12] ICP occurs predominantly during the third trimester of pregnancy and spontaneously disappears after labor.[10] Although ICP has a mild impact on the mother, it poses significant risks to the fetus for complications, such as preterm delivery, respiratory stress, and prenatal mortality.[8-10, 13] ICP shares the characteristic manifestations of intrahepatic cholestasis with a hallmark of markedly elevated levels of serum bile acids, which has become the most accurate diagnosis for ICP.[6, 7, 14]

Bile acid homeostasis is achieved through a tightly regulated enterohepatic circulation and canalicular secretion through the bile salt export pump (BSEP, ABCB11) is the rate-limiting step.[15, 16] BSEP expression is coordinately regulated by distinct but related transactivation pathways, notably the bile acid / farnesoid x receptor (FXR, NR1H4) signaling pathway.[17-21] Activation of FXR by bile acids strongly induces BSEP expression in vitro and in vivo.[17, 18] Such feed-forward regulation of BSEP by bile acid/FXR is considered a major mechanism to prevent excessive accumulation of toxic bile acids in hepatocytes.

It has been recognized that the etiology of ICP is complex, with genetic and endocrine contributing factors.[10] Indeed, genetic variants of BSEP and FXR have been associated with an increased risk for ICP.[22-24] On the other hand, steroid hormones and their metabolites have been implicated in the development of ICP.[25-29] Currently, the transcriptional regulation of BSEP during pregnancy and its underlying mechanisms and involvement in ICP are not fully understood.

In this study the transcriptional dynamics of BSEP in vivo in the same group of pregnant mice before, during, and after gestation were established, resembling the clinical course of ICP in human. Further studies showed that BSEP transcription was inversely correlated with serum E2 levels during pregnancy and E2 repressed BSEP expression in vitro and in vivo through a nonclassical E2/ERα transrepressive pathway, directly interacting with FXR.

Materials and Methods

Plasmid Constructs

Human and mouse BSEP promoter reporters phBSEP(-2.6kb) and pmBSEP (-2.6kb) were described elsewhere.[20, 30] Human FXR, Flag-FXR, ERα, and ERβ were provided by Drs. David Mangelsdorf and Matthew Stoner. Construct eGFPn-FXR was made by fusing the N-terminal 172 residues of enhanced green fluorescence protein (eGFP) to human FXR, while ERα-eGFPc was generated by fusing the human ERα to the C-terminal fragment of eGFP with a linker (RSIATGS) in between. Promoter reporters phBSEP (-805b), phBSEP(-405b), phBSEP(-205b), phBSEP (-160b), and phBSEP(-120b) were described previously.[19] The estrogen response element (ERE) reporter pTK-2xERE was made by cloning two copies of the ERE consensus sequences into the pTK-Luc vector. FXR response element (FXRE) reporters pGL-2xFXREcon and pGL-2xhIR1 were constructed by cloning two copies of the FXRE consensus (5′-AGGTCA T TGACCT-3′) or IR1a (inverted repeat spaced by one nucleotide, 5′-GGGACA T TGATCC-3′) in human BSEP promoter into the pGL3/promoter vector.

Treatments of Human Primary Hepatocytes and Huh 7 Cells

Human primary hepatocytes obtained through the Liver Tissues Procurement and Distribution System and Huh 7 cells were treated with chenodeoxycholic acid (CDCA) (5 or 10 μM) or a combination of CDCA and various concentration of E2 (0, 1, 10, or 100 nM) for 30 hours in a phenol red-free Dulbecco's modified Eagle's medium (DMEM) containing 1% charcoal-stripped fetal bovine serum (FBS).

Reporter Luciferase Assays

Transient transfection and dual luciferase assays were carried out as described elsewhere.[31]

Quantitative Real-Time Polymerase Chain Reaction (PCR)

Total RNA isolation from human primary hepatocytes, Huh 7 cells, or liver tissues and subsequent TaqMan real-time PCR assays were performed as described previously.[20, 30]

Living Imaging With an In Vivo Imaging System (IVIS)

Before mating, 30 female CD-1 mice were hydrodynamically injected with mouse BSEP promoter reporter pmBSEP(-2.6kb) through the tail vein (0.5 μg/g). Hepatic luciferase expressions were monitored by IVIS[30] before, during, and after the gestation in both pregnant and nonpregnant mice. In the study with E2 treatment, 20 female CD-1 mice were randomly divided into E2 (5 mg/kg daily for 5 days subcutaneously) and vehicle ethanol (EtOH) group. All mice were hydrodynamically injected with pmBSEP(-2.6kb) plasmid prior to E2 treatment. Luciferase levels were detected by IVIS before and 7 days posttreatment. All animal studies were approved by the Institutional Animal Care and Use Committee (IACUC) at the University of Rhode Island.

ERα(−/−) Mice and Treatments

ERα(−/−) mice were obtained through in-house breeding with ERα(−/+) heterozygous male and female mice (Jackson Laboratories). Eleven ERα(−/−) mice were randomly divided into E2 treatment (5 mg/kg daily for 5 days subcutaneously) and control EtOH groups. ERα wild-type (WT) mice from the same colony were included as controls. Liver tissues were harvested and processed for BSEP messenger RNA (mRNA) and protein detection.

Bimolecular Fluorescence Complementation (BiFC) Assays

Huh 7 cells were cotransfected with 1 μg eGFPn-FXR and 1 μg ERα-eGFPc, ERα-LBD-Mut-eGFPc, or pcDNA5. Twenty-four hours posttransfection, the cells were treated with 100 nM E2 for 24 hours, followed by detection of fluorescence under a confocal microscope. Eighteen CD-1 female mice were randomly divided into three groups. Each group of mice were hydrodynamically coinjected with 5 μg eGFPn-FXR with 5 μg ERα-eGFPc, ERα-LBD-Mut-eGFPc, or pcDNA5 through the tail vein. Twenty-four hours postinjection, mouse primary hepatocytes were collected through liver perfusion. Hepatocytes were either cultured on slide chambers for florescence detection or subjected to fluorescence activated cell sorting (FACS) analysis.

Western Blotting

Cell lysates were prepared from Huh 7 cells or mouse liver tissues for detecting BSEP, ERα, and small heterodimer partner (SHP) by regular or capillary electrophoresis-based western blot as described previously.[30]

Serum E2 Detection

Serum samples were collected from mice through the tail vein before, during, and after gestation. Serum E2 levels were detected with an E2 enzyme-linked immunosorbent assay (ELISA) kit from Calbiotech.

Site-Directed Mutagenesis

Mutagenesis was performed using QuickChange mutagenesis kit (Stratagene). ERα DNA binding domain mutant (ERα-DBD-Mut) was made by substituting two amino acids at E207A and G208A in ERα WT. The ERα ligand binding domain mutants (ERα-LBD-Mut) and ERα-LBD-Mut-eGFPc were generated by replacing the glycine at 521 with arginine in ERα WT or ERα-eGFPc.

Coimmunoprecipitation (Co-IP) Assays

HEK 293T or Huh 7 cells were transfected with 2 μg Flag-FXR or FXR and ERα, followed by treatment with 100 nM E2 and 10 μM CDCA for 24 hours and subsequently lysed. Liver tissues from pregnant mice at gestation day 18 were homogenized and lysed. Cell lysates (500 μg total protein) underwent Co-IP with the Dynabeads Antibody Coupling kit (Life Technologies) using ERα (sc-542, Santa Cruz Biotechnology), custom-made FXR antibodies, or rabbit IgG, followed by western blotting with anti-Flag (Sigma-Aldrich), FXR, or ERα antibodies.

Chromatin Immunoprecipitation (ChIP) Assays

Chromatins were prepared from Huh 7 cells transfected with 2 μg ERα and FXR plasmids, followed by treatment of transfected cells with 10 μM CDCA and 100 nM E2 or 1% EtOH for 1 or 24 hours. ChIP assays were performed with the ChIP-IT Express Kit (Active Motif) following the procedure described elsewhere.[20, 32]

Electrophoretic Mobility Shift Assays (EMSA)

Nuclear extracts of Huh 7 cells transfected with FXR and ERα were prepared and EMSAs were performed using the LightShift Chemiluminescent EMSA kit (Pierce) as described previously.[32]

Statistical Analysis

A Student t test was applied to pairwise comparison for normally distributed data. One-way analysis of variance (ANOVA) was applied to analyze data with multiple groups, followed by a Tukey posthoc test for multiple comparisons. The nonparametric Mann-Whitney test was used for pairwise comparison for nonnormally distributed data. The Pearson correlation test was used to determine the correlation coefficient. P 0.05 or lower was considered statistically significant.

Results

BSEP Transcription Was Repressed in the Late Stages of Pregnancy

Serum bile acids are mildly elevated in normal pregnancy and markedly increased in ICP, especially in the third trimester of pregnancy.[6, 7, 14] BSEP is responsible for biliary secretion of bile acids, a rate-limiting step in the enterohepatic circulation of bile acids.[15, 16] To determine whether BSEP expression is transcriptionally altered during pregnancy, especially at the late stages of pregnancy, the dynamics of BSEP transcription in the same group of mice hydrodynamically injected with BSEP promoter reporter was monitored by IVIS before, during, and after gestation. As shown in Fig. 1A,B, significant decreases in BSEP transactivation were detected on days 2 and 1 prior to parturition and on the day of parturition with P 0.005, <0.001, and 0.003, respectively. The lowest levels were detected on the day of parturition and only 3.4% BSEP transactivation levels remained with one mouse below the detection level. One day after parturition, BSEP transactivation was immediately recovered to ∼50% of the premating levels. By days 6 and 11 postparturition, the BSEP transactivation levels were completely recovered. The data for individual mice are presented in Supporting Table 1. No significant changes were detected in nonpregnant mice during the same monitoring period (Fig. 1C,D). Taken together, for the first time the transcriptional dynamics of BSEP was established in the same group of mice during the entire pregnancy. Such dynamics resemble the clinical course of ICP in human, which occurs in the late stages of pregnancy and spontaneously disappears within a week after delivery.[10]

Figure 1.

The dynamics of BSEP transactivation before, during, and after gestation in the same group of pregnant mice. (A) Mouse BSEP promoter reporter pmBSEP(-2.6kb) was hydrodynamically injected (0.5 μg/g) before mating. Hepatic luciferase expression levels as luminescent signals were monitored by IVIS before, during, and after gestation in the same group of pregnant mice (n = 15). (B) Quantification of the luciferase activities as luminescent signals (photons) at various timepoints. (C) Hepatic luciferase expression levels were monitored in a group of nonpregnant mice (n = 10) in parallel with the pregnant mice shown in (A). (D) Quantification of the luciferase activities in nonpregnant mice. Median value of each group is indicated by a short line. Mann-Whitney's nonparametric test was applied for pairwise comparison. *P < 0.05 and **P < 0.01.

Serum E2 Levels Were Increased in the Late Stages of Pregnancy and Inversely Correlated With BSEP Transactivation

Among estrogens, E2 is the predominant hormone during pregnancy. As shown in Fig. 2A, serum E2 levels steadily increased throughout the pregnancy and peaked right before delivery. Significant increases in E2 levels were detected on days 3, 2, 1 prior to parturition and on the day of delivery with P 0.023, 0.005, 0.001, and 0.015, respectively. The E2 levels then quickly dropped to the premating levels on days 1, 2, and 7 postdelivery. The data for individual mice are presented in Supporting Table 2.

Figure 2.

Serum E2 levels were increased in the late stages of pregnancy and inversely correlated with BSEP transactivation. (A) Serum samples were collected before, during, and after gestation and E2 levels were determined in the same group of pregnant (n = 14) and (B) nonpregnant mice (n = 14). The median value of each group is indicated by a short line. Mann-Whitney's nonparametric test was used for pairwise comparison. *P < 0.05 and **P < 0.01. (C) Pearson correlation test was performed between BSEP transactivation and serum E2 levels in the same groups of pregnant mice. Group means for BSEP transactivation and serum E2 levels were used for the calculation on the following timepoints: premating, day −5/6, day −2, day −1, day 0, day +1, and day +6/7.

No significant alterations in E2 serum levels were detected in a group of nonpregnant mice at all timepoints except for one (Fig. 2B), in which the E2 level was statistically higher than the premating concentration (P = 0.021). Such a significant elevation in E2 might reflect the fluctuation of E2 levels during a normal estrus cycle.

To determine whether BSEP transactivation dynamics are correlated with the changes in serum E2 levels, a Pearson correlation test was performed. As shown in Fig. 2C, BSEP transactivation was negatively correlated with serum E2 concentrations (r = −0.902) with P 0.005. It was thus concluded that elevation of E2 concentrations in the late stages of pregnancy was a possible cause for the suppression of BSEP transactivation.

E2 Suppressed BSEP Expression In Vivo

To determine whether E2 directly suppresses BSEP transactivation, the effects of E2 on BSEP transactivation were investigated. As shown in Fig. 3A, BSEP transactivation levels were comparable among all the mice prior to E2 treatment. However, it was markedly suppressed by E2 (75% decrease) (P < 0.05) (Fig. 3A). To confirm the results, endogenous BSEP mRNA and protein levels were quantified with real-time PCR and western blot. Consistent with the IVIS data, E2 significantly decreased BSEP expression by ∼50% (P < 0.05) (Fig. 3B,C). It was thus concluded that E2 directly suppressed BSEP expression in mice.

Figure 3.

E2 suppressed BSEP expression in vivo in mice and in vitro in human primary hepatocytes and hepatoma Huh 7 cells. (A) Twenty CD-1 mice hydrodynamically injected with pmBSEP(-2.6kb) (0.5 μg/g) by way of the tail vein prior to treatment were randomly divided into treatment (E2, 5 mg/kg daily for 5 days) and control groups (vehicle EtOH). Luciferase activities were quantified by IVIS before and after treatment. (B) Endogenous BSEP mRNA and (C) protein levels in E2 or EtOH-treated mice were detected by real-time PCR or western blot with glyceraldehyde 3-phosphate dehydrogenase (GAPDH) as internal standard. (D) Human primary hepatocytes or (E) Huh 7 cells were treated with CDCA (5 μM) or a combination of CDCA and various concentrations of E2, followed by detection of BSEP mRNA by real-time PCR and (F) BSEP protein levels by western blotting. The Student t test was applied to pairwise comparison in (A-C). One-way ANOVA was applied to analyze data in (D,E), followed by the Tukey posthoc test for multiple comparisons. *P < 0.05 and **P < 0.01.

E2 Suppressed BSEP Expression in Human Primary Hepatocytes and Huh 7 Cells

As expected, treatment with CDCA significantly induced BSEP mRNA expression in both human primary hepatocytes and Huh 7 cells. However, such inductions were significantly suppressed by E2 (Fig. 3D,E). Consistently, BSEP protein levels were dose-dependently suppressed by E2 (Fig. 3F). Taken together, it was concluded that E2 suppressed BSEP expression in both human primary hepatocytes and Huh 7 cells primarily through transcription.

E2 Repressed BSEP Promoter Transactivation Through ERα In Vitro

To determine whether E2-mediated repression of BSEP expression is mediated through ERα or ERβ, the effects of E2 on BSEP promoter transactivation were evaluated in the presence of ERα or ERβ. As shown in Fig. 4A, cotransfection of ERα significantly decreased basal as well as CDCA-mediated BSEP transactivation. More important, E2 further significantly decreased BSEP transactivation. In contrast, ERβ had minimal effects on BSEP transactivation in the presence or absence of E2. In the dose response studies, both ERα and E2 dose-dependently repressed BSEP transactivation (Fig. 4B,C). The results thus established that E2 repressed BSEP transactivation through ERα but not ERβ, and such repression was dose-dependent of ERα and E2.

Figure 4.

E2 repressed BSEP promoter transactivation in vitro through ERα. (A) Huh 7 cells were transfected with phBSEP(-2.6kb), FXR and ERα or ERβ, followed by treatment with CDCA (10 μM) or CDCA plus E2 (10 nM). Luciferase activities were detected by the dual luciferase assays. (B) Dose-response studies were performed with ERα and (C) E2. Data are presented as means ± SD of at least three individual experiments. One-way ANOVA was applied to analyze data in (A,C), followed by a Tukey posthoc test for multiple comparisons. The Student t test was applied to pairwise comparison in (B). *P < 0.05 and **P < 0.01.

ERα Was Required for E2 to Repress BSEP Expression In Vivo

After demonstrating that E2 repressed BSEP transactivation through ERα in vitro, we next carried out experiments to confirm such findings in ERα−/− mice in vivo. As shown in Fig. 5A,B, BSEP mRNA levels were significantly decreased in WT mice treated with E2 (P < 0.01) but the decrease was not observed in ERα(−/−) mice. Consistently, BSEP protein levels were markedly reduced by E2 in WT mice, whereas such repression was not detected in ERα(−/−) mice (Fig. 5C,D). The data thus firmly established that ERα was required for E2 to repress BSEP expression.

Figure 5.

ERα was required for E2 to repress BSEP expression in vivo. (A) ERα(−/−) and (B) ERα WT mice from the same colony were treated with E2 (5 mg/kg) or vehicle daily for 5 days, followed by detection of BSEP mRNA levels with real-time PCR. The mean and SD are indicated in each group. (C) BSEP protein expression was detected by capillary electrophoresis-based western blotting with GAPDH as internal standard. (D) Quantification of BSEP protein levels. Mann-Whitney's nonparametric test was used for pairwise comparison. *P < 0.05 and **P < 0.01.

ERα Expression in Huh 7 Cells and Mice Liver Before, During, and After Pregnancy

Both ERα mRNA and protein were readily detected in Huh 7 cells and markedly increased in ERα-transfected cells (Supporting Fig. 1A,B). E2 had no significant effects on ERα mRNA levels but decreased ERα protein levels, consistent with the notion that ligand-binding to ERα promotes ERα protein turnover. ERα expression was increased during pregnancy at both mRNA and protein levels (Supporting Fig. 2A,B) but the increases were not statistically significant.

Mapping of the cis-Acting Element Mediating E2/ERα Transrepression to FXRE

To localize the cis-element, a series of human BSEP promoter constructs with various lengths (Fig. 6A) were used to evaluate their ability to respond to E2. As shown in Fig. 6B, all the promoter constructs exhibited similar patterns in responding to E2. The minimal promoter which retained its ability to respond to E2 is phBSEP(-120b). It should be mentioned that further deletion of the minimal promoter resulted in a total loss of the promoter activity.

Figure 6.

Mapping of the cis-acting element mediating E2/ERα transrepression to FXRE. (A) Human BSEP promoter reporters with various lengths. (B) Transrepression of various BSEP promoter reporters and (C) FXRE consensus pGL-2xFXREcon and human BSEP IR1a element pGL-2xhIR1 reporters by E2 (10 nM) in the presence of CDCA (10 μM). (D) Chromatins were prepared from Huh 7 cells transfected with ERα and FXR, and subsequently treated with CDCA and E2. Chromatins were precipitated with anti-ERα antibodies or IgG, followed by PCR detection with a primer set flanking or upstream (negative control) the FXRE in the human BSEP promoter. (E) Huh 7 cells were transfected with phBSEP(-2.6kb), FXR and pcDNA5, ERα, ERα-LBD-Mut or ERα-DBD-Mut, followed by treatment with CDCA or CDCA and E2. Luciferase activities were determined. Data are presented as means ± SD of at least three individual experiments. The Student t test was applied to pairwise comparison in (B,C). One-way ANOVA was applied to analyze data in (E), followed by a Tukey posthoc test for multiple comparisons. *P < 0.05 and **P < 0.01.

Bioinformatic analysis of the minimal promoter revealed a number of potential cis-elements including FXRE. Given the fact that E2/ERα repressed basal as well as CDCA-induced BSEP transactivation, the FXRE was the first target to be investigated. Two FXRE reporters, pGL-2xFXREcon and pGL-2xhIR1a, were tested. E2 exhibited a similar transrepressive pattern on the two FXRE reporters (Fig. 6C) as to BSEP promoter reporters (Fig. 6B). Thus, the data strongly suggested that E2/ERα repressed BSEP transactivation through the FXRE in the BSEP promoter. Indeed, ERα was specifically recruited to the FXRE in the BSEP promoter in the ChIP assays (Fig. 6D).

Ligand but Not DNA Binding Was Required for ERα to Mediate Transrepression on BSEP

To investigate whether ERα directly binds to the FXRE, an ERα DNA-binding mutant (ERα-DBD-Mut) was tested for its ability to transrepress BSEP. As shown in Fig. 6E, ERα-DBD-Mut exhibited similar transrepressive activity as ERα WT, while ERα ligand binding domain mutant ERα-LBD-Mut totally lost its ability to repress BSEP transactivation. As expected, both ERα-DBD-Mut and ERα-LBD-Mut diminished their ability to respond to E2 with an ERE reporter pTK-2xERE (Supporting Fig. 3). The data thus indicated that DNA-binding is not required for ERα to mediate the transrepression. Therefore, it is unlikely that ERα directly binds to the FXRE to interfere in the FXR signaling pathway. More logically, ERα may interrupt FXR signaling through a protein-protein interaction with FXR.

ERα and FXR Were Coimmunoprecipitated in Cells and Livers From Pregnant Mice

To determine whether ERα forms a protein complex with FXR in the cells, Flag-tagged FXR or FXR and ERα were overexpressed in 293T or Huh 7 cells, followed by detection of ERα-FXR protein complex. As shown in Fig. 7A,B, FXR and ERα formed a protein complex, which was immunoprecipitated with anti-Flag or anti-ERα antibodies, followed by detection with anti-ERα or anti-FXR antibodies.

Figure 7.

Association of ERα with FXR in cell lines and livers from pregnant mice. (A) HEK 293T cells were transfected with Flag-FXR and ERα, followed by treatment with CDCA (10 μM) and E2 (100 nM). Whole cell lysates were precipitated with rabbit anti-ERα polyclonal antibodies or rabbit IgG as negative control. The precipitated proteins were analyzed by western blotting with anti-Flag antibodies. (B) Huh 7 cells were transfected with FXR and ERα. Cell lysates were immunoprecipitated with anti-FXR, followed by detection by western blotting with anti-ERα antibodies. (C) Liver tissues were collected from pregnant mice at gestation day 18, homogenized, and lysed. Cell lysates were immunoprecipitated with anti-ERα antibodies, followed by western blotting with anti-FXR antibodies. (D) Cell lysates from the mouse liver tissues were precipitated with anti-FXR antibodies, followed by detection with anti-ERα antibodies on western blot. (E) Nuclear extracts were prepared from Huh 7 cells transfected with FXR and ERα. EMSA assays were performed using the FXRE of human BSEP as probe in the presence of dimethyl sulfoxide (DMSO) (0.1%), EtOH (0.1%), E2 (100 nM), CDCA (10 μM), or a combination of E2 and CDCA. (F) Chromatins were prepared from Huh 7 cells transfected with FXR and ERα and treated with E2 (100 nM) or EtOH (0.1%), followed by immunoprecipitation with rabbit anti-FXR, ERα, or control rabbit IgG. ChIP assays were performed with the ChIP-Express Kit from Active Motif.

To determine whether such an FXR-ERα protein complex forms in vivo in pregnant mice, liver tissues were collected from mice at gestation day 18 and similar Co-IP assays were performed. Consistent with the data from transfected cells, protein complexes of FXR-ERα were readily detected in the liver of pregnant mice (Fig. 7C,D).

Effects of E2 on FXR and ERα Recruitment to FXRE

In the EMSA assays, E2 decreased FXR binding to the FXRE. However, such a decrease largely disappeared when CDCA was present (Fig. 7E). In the ChIP assays, E2 increased ERα recruitment to the FXRE of BSEP promoter without affecting FXR recruitment (Fig. 7F).

Direct Interaction Between ERα and FXR in Living Cells and In Vivo

To determine whether ERα directly interacts with FXR, BiFC assays were performed.[33] After construction and confirming the activities of the fusion proteins eGFPn-FXR and ERα-eGFPc (Fig. 8A; Supporting Fig. 4A,B), direct physical interaction between the two fusion proteins was evaluated in vitro in Huh 7 cells and in vivo in mice. Minimal fluorescent signals were detected in cells transfected with eGFPn-FXR and vector pcDNA5. However, significant fluorescence was observed in the cells cotransfected with eGFPn-FXR and ERα-eGFPc (Fig. 8B,C), indicating a direct interaction between ERα and FXR. Mutation of the LBD in ERα-LBD-Mut-eGFPc markedly decreased the fluorescent signals, indicating that ligand binding is required for ERα to achieve maximal interaction with FXR. Similar results were obtained with the in vivo mouse study (Fig. 8D,E). Minimal fluorescent signals were detected in hepatocytes derived from mice injected with eGFPn-FXR and pcDNA5. However, strong fluorescent signals were detected in hepatocytes from the mice injected with eGFPn-FXR and ERα-eGFPc. Much weaker signals were detected in hepatocytes from mice injected with eGFPn-FXR and ERα-LBD-Mut-eGFPc. It was thus concluded that ERα directly interacted with FXR in vitro in living cells and in vivo in mice, and maximal interaction required ligand binding to ERα.

Figure 8.

Direct interaction between ERα and FXR in vitro in Huh 7 cells and in vivo in mice. (A) Construction of fusion proteins eGFPn-FXR, ERα-eGFPc, and ERα-LBD-Mut-eGFPc. (B) Huh 7 cells were transfected with eGFPn-FXR and either pcDNA5 vector, ERα-eGFPc, or ERα-LBD-Mut-eGFPc, followed by treatment with E2 (100 nM). Fluorescent signals were detected under a confocal microscope. Images were captured at a magnification of 40× with fluorescent and differential interference contrast (DIC) settings. (C) Quantification of fluorescent signals in (B). (D) Three groups of mice were hydrodynamically coinjected eGFPn-FXR with either ERα-eGFPc, ERα-LBD-Mut-eGFPc, or pcDNA5 vector by way of the tail vein. Twenty-four hours postinjection, primary hepatocytes collected through liver perfusion were cultured on slide chambers for fluorescence detection under confocal microscope or directly subjected to FACS analysis. (E) Quantification of the cells gated in M2 window showed in (D). (F) Cellular localization of ERα and FXR interaction as fluorescent signals in Huh 7 cells and primary hepatocytes. One-way ANOVA was applied to analyze data in (C,E), followed by a Tukey posthoc test for multiple comparisons. *P < 0.05 and **P < 0.01.

It is our expectation that the interaction between ERα and FXR should occur in the nuclei. Indeed, fluorescent signals were almost exclusively observed in the nuclei of transfected Huh 7 cells (upper panel, Fig. 8F). However, it was noticed that fluorescent signals were observed in various cellular compartments in mouse hepatocytes in vivo with a predominant presence around the nuclear membrane likely associated with endoplasmic reticulum and to a less extent in the cytoplasm possibly associated with vesicle-like structures and nucleus (lower panel, Fig. 8F).

Discussion

Pregnancy is associated with dramatic physiological changes in the mother. One of the changes during pregnancy is elevated levels of serum bile acid.[3, 5-7] When such elevation reaches in excess of 10 μM, ICP develops.[6, 7, 14] Canalicular effluxers including BSEP and multidrug resistance protein 3 (MDR3) have long been suspected as the targets for causing ICP.[22, 23, 27] Indeed, genetic variants of BSEP are associated with ICP.[22, 23] Also, estrogen and its metabolites have been reported to inhibit BSEP function or cell surface expression.[27, 28] In this study for the first time, the dynamics of BSEP transactivation during pregnancy was established in the same group of mice and it was demonstrated that BSEP expression was significantly transrepressed during the late stages of pregnancy. Furthermore, such transrepression was inversely correlated with elevated E2 levels during pregnancy and E2 repressed BSEP expression in vitro and in vivo. These findings support the notion that down-regulation of BSEP expression by elevated E2 during the late stages of pregnancy is a contributing factor to the pathogenesis of ICP. Consistent with a previous report,[26] SHP mRNA levels were significantly decreased during and after pregnancy (Supporting Fig. 5A). However, the decreases were not statistically significant at the protein level (Supporting Fig. 5B).

Since most women do not experience ICP during their normal pregnancy, it becomes apparent that repression of BSEP expression by E2 alone is not sufficient to cause ICP. Other factors, notably genetics, are also risk factors for ICP. The prevalence of ICP varies markedly among ethnic groups, ranging from less than 1% to 14%,[11, 12] with the highest in South America. Such variations in ICP incidence among ethnic groups strongly indicates that certain genetic traits predispose women to ICP. It is therefore reasonable to speculate that women carrying those genetic changes increase their sensitivity to elevated E2 during the late stages of pregnancy, leading to ICP.

The question remains regarding the physiological significance of BSEP transrepression by E2/ERα through attenuating the FXR signaling during pregnancy, especially at the later stages of pregnancy. Transrepression of BSEP leads to decreased canalicular disposition of bile acids, thus causing retention of bile acids in the liver. Such an increase in bile acid levels in turn slows down the conversion of cholesterol into bile acids through feedback mechanisms,[34] subsequently leading to conservation of cholesterol. Cholesterol is an essential building block for cell membranes and a precursor for steroid hormone production. A growing fetus requires cholesterol for rapid growth and steroid hormone production, especially in the late stages of pregnancy. Therefore, down-regulation of BSEP, as a result of attenuation of the FXR signaling by E2/ERα, may represent an adaptive mechanism by the mother to meet the increasing demand of the growing fetus for cholesterol. Consistent with such speculation, serum cholesterol levels are elevated in pregnant women.[1-3]

In the classical signaling pathway estrogens, including E2, regulate their target genes through activating ERα or ERβ,[35] which directly binds to ERE in the promoter of the target genes and subsequently results in the transactivation of the target genes.[35] In this study, we demonstrated that E2 transrepressed BSEP expression through ERα, which directly interacts with FXR. Such transrepression through a protein-protein interaction represents another nonclassical estrogen signaling pathway.[36] In support of our findings, it has been shown that ERα coimmunoprecipitated with FXR in a breast cancer cell line MCF-7[37] and ERα interacted FXR in a GST-pulldown assay.[26]

As FXR and ERα are both nuclear receptors, we expected that their interaction occurs in the nucleus. Indeed, the interaction in Huh 7 cells was almost exclusively in the nucleus. However, the interaction was more widely distributed in mice in vivo, with a predominant presence around the nuclear membrane, especially outside the membrane. The precise reasons for such a discrepancy are not clear. The predominant presence of FXR and ERα complexes outside the nucleus in mouse liver raised an interesting question as to whether retention of FXR in the cytoplasm as an ERα/FXR complex represents a possible mechanism for E2/ERα to attenuate the FXR signaling pathway.

We consistently demonstrated that ERα markedly decreased basal and CDCA-induced transactivation of BSEP promoter in the absence of exogenously added E2 (Figs. 4A,B, 6E). One possible explanation is that there are endogenous estrogenic compounds in cells or cell medium that bind to ERα to mediate the transrepression. Another possible explanation is that ERα has both ligand-dependent and -independent transrepressive activities on BSEP.

Given the fact that in addition to bile acid homeostasis the FXR signaling pathway is involved in many cellular functions including cholesterol, lipids, glucose, energy, and vascular regulations,[38-40] it is reasonable to assume that many other physiological and metabolic changes during pregnancy are direct results of estrogen's effects through the ERα/FXR interaction.

Acknowledgment

Instrumental support from the RI-INBRE Core Facility in the College are greatly appreciated, which is supported by an Institutional Development Award (IDeA) from the National Institutes of General Medical Sciences of the National Institutes of Health under grant P20 GM103430-13. We thank Dr. David Mangelsdorf for providing human FXRα1 expression plasmid, Dr. Matthew Stoner for providing expression plasmids for Flag-FXR, ERα, and ERβ, and Dr. Chang-Deng Hu for providing the expression constructs of the N- and C-terminal fragments of eGFP, pBiFC-VN173 (pFLAG-Venus 1-172), and pBiFC-VC155 (pHA-Venus 155-238).

Ancillary