Potential conflict of interest: Nothing to report.
This study was supported by the Biotechnology and Biological Sciences Research Council, the Institute of Obstetric and Gynaecology Trust, the Netherlands Organisation for Scientific Research, the University Utrecht High Potentials Program, the University Medical Center Utrecht (through an internationalization grant), the Biomedical Research Centre at the Imperial College Healthcare National Health Service Trust, and GlaxoSmithKline.
Pregnancy alters bile acid homeostasis and can unmask cholestatic disease in genetically predisposed but otherwise asymptomatic individuals. In this report, we show that normal pregnant mice have raised hepatic bile acid levels in the presence of procholestatic gene expression. The nuclear receptor farnesoid X receptor (FXR) regulates the transcription of the majority of these genes, and we show that both ablation and activation of Fxr prevent the accumulation of hepatic bile acids during pregnancy. These observations suggest that the function of Fxr may be perturbed during gestation. In subsequent in vitro experiments, serum from pregnant mice and humans was found to repress expression of the Fxr target gene, small heterodimer partner (Shp), in liver-derived Fao cells. Estradiol or estradiol metabolites may contribute to this effect because coincubation with the estrogen receptor (ER) antagonist fulvestrant (ICI 182780) abolished the repressive effects on Shp expression. Finally, we report that ERα interacts with FXR in an estradiol-dependent manner and represses its function in vitro. Conclusion: Ligand-activated ERα may inhibit FXR function during pregnancy and result in procholestatic gene expression and raised hepatic bile acid levels. We propose that this could cause intrahepatic cholestasis of pregnancy in genetically predisposed individuals. HEPATOLOGY 2010
The synthesis, metabolism, and enterohepatic circulation of bile acids is tightly regulated by nuclear hormone receptors.1 Farnesoid X receptor (FXR) is required for the basal maintenance of the enterohepatic circulation and its response to bile acid challenge.2 Bile acid–activated FXR directly induces genes that stimulate hepatic bile export [bile salt export pump (BSEP)3 and multidrug resistance protein 3 (MDR3)4] and metabolism (sulfotransferase family cytosolic 2A dehydroepiandrosterone-preferring member 1 and uridine diphosphoglucuronate glucuronosyltransferase 2 family polypeptide B45, 6). Additionally, FXR indirectly represses the expression of bile acid import [Na+-taurocholate cotransporting polypeptide (NTCP)7] and synthesis genes [cytochrome P450 7A1 (CYP7A1) and cytochrome P450 8B1 (CYP8B1)8] through the induction of a transcriptional repressor, small heterodimer partner (SHP), in the liver8 and a signaling hormone, fibroblast growth factor 19 (FGF19)/Fgf15, in the intestine.9 FXR therefore plays a central role in preventing the toxic accumulation of bile acids in the liver.
Intrahepatic cholestasis of pregnancy (ICP) is characterized by raised serum bile acid levels and abnormal liver function tests. The disease is associated with fetal distress, spontaneous preterm delivery, and unexplained intrauterine death.10 We have identified genetic variants of FXR, BSEP, and MDR3 that contribute to the etiology of ICP.11-13 However, it is currently not known how pregnancy unmasks cholestatic disease in these genetically predisposed but otherwise normal individuals. Importantly, gestation itself may be a state of impaired bile acid homeostasis because up to 40% of women develop asymptomatic hypercholemia of pregnancy,14 and an increase in the total bile acid pool has also been reported.15 As such, the mechanisms that affect bile acid homeostasis during normal pregnancy may also be relevant to the etiology of ICP.
For a number of reasons, estrogens are thought to contribute to the etiology of ICP.16 First, the disease usually develops in the third trimester of pregnancy when concentrations of estrogens are highest. Second, twin pregnancies have both a higher incidence of ICP and a more pronounced rise in estradiol concentrations.17 Third, ICP patients can present with cholestasis outside of pregnancy when they are taking oral contraceptives containing 17α-ethinylestradiol.18 High doses of estradiol and its metabolites also cause cholestasis in rodents,19 and mice lacking ER are resistant to these effects.20 Taken together, these findings imply that estrogens could dysregulate bile acid homeostasis in normal pregnant women and trigger cholestatic disease in genetically predisposed individuals. However, liver biopsy is not clinically indicated in the majority of ICP cases, so data on the response of the human liver to pregnancy and ICP are limited.
In this report, we investigate whether bile homeostasis is dysregulated in pregnant mice and whether this is due to impairment of Fxr function. We show that hepatic bile acids are raised in pregnant mice and that liver gene expression is procholestatic and resembles a state of Fxr inactivation. We provide in vivo and in vitro evidence showing that estrogen or its metabolites may be the underlying cause of Fxr dysfunction.
This study conformed to the guidelines outlined by the 1975 Declaration of Helsinki, and permission was obtained from the ethics committee of the Hammersmith Hospitals National Health Service Trust (London, United Kingdom; REC 97/5197). All ICP patients were diagnosed on the basis of clinical symptoms in combination with routine laboratory investigations, as described previously.13 The diagnosis was confirmed by raised serum liver aminotransferases and/or bile acids. Serum samples were pooled for in vitro assays. Serum from eight normal, nonpregnant women and nine normal, pregnant women (between 30 and 38 weeks of gestation) was used as well as serum from six individuals diagnosed with ICP (see Supporting Information Table 1 for clinical details).
Animals and Treatments.
Studies were conducted in accordance with the UK Animals (Scientific Procedures) Act of 1986. Female mice, 10 to 12 weeks old, were used throughout this study. Fxr−/−21 and all other mice were on a C57BL6 background. Nonpregnant animals were killed on the day of conception to synchronize them in their menstrual cycles. Pregnant animals were killed on the 18th day after conception. All pregnant mice had six to eight live fetuses inutero. All mice that received the 0.5% cholic acid (CA) diet (Special Diet Service, United Kingdom) were fed for 30 ± 2 days. Estrogen implant experiments were conducted as described previously.21 Briefly, female mice were ovariectomized and allowed to recover for 2 weeks. Subsequently, Silastic tubing containing 17β-estradiol (33 μg/mL) or vehicle (peanut oil) were implanted subcutaneously for 18 days. Tissues were harvested at 1 pm (zeitgeber time 6) after a 4-hour fasting period.
Quantification of Hepatic Bile Acids.
Hepatic bile acids were extracted in triplicate essentially as described.22 Tissues were removed, weighed, and snap-frozen. Care was taken not to contaminate the tissue with gallbladder bile. Tissues were subsequently homogenized in 75% ethanol (EtOH), incubated with shaking, and centrifuged, and the supernatant was assayed (Sentinel Diagnostics, Milan, Italy).
RNA Isolation, Microarray Hybridization, and Data Analysis.
Microarray analyses of livers from six mice per group were conducted. Total RNA was extracted with TRIzol (Invitrogen, Carlsbad, CA) and purified on a column (Qiagen, Valencia, CA). The RNA was reverse-transcribed, labeled, hybridized to Mouse Genome 430A 2.0 arrays (Affymetrix, Santa Clara, CA), and scanned. Rosetta Resolver (Rosetta Inpharmatics, Seattle, WA) was used to perform background correction and normalization and to construct Venn diagrams and two-dimensional clusters. For each two-way comparison, probe sets with a mean, normalized, scaled intensity of less than 30 U in both of the comparison groups were removed from the analysis (see also Schnoes et al.23). A false detection rate of 1% was used for construction of the Venn diagrams, and a rate of 5% was used for Ingenuity Pathway Analysis (Ingenuity Systems, Mountain View, CA).
Total RNA was isolated by TRIzol extraction and reverse-transcribed (Invitrogen). qRT-PCR was conducted with the SYBR Green reagent (Sigma-Genosys, Haverhill, United Kingdom). Each 25-μL reaction comprised 0.8 μM primers, 0.5 μL of a complementary DNA (cDNA) template, and 12.5 μL of SYBR Green. Data are presented as relative expressions normalized to cyclophilin.
Fao Cell Serum Treatment.
Rat hepatoma Fao cells (ECACC 85061112) were cultured in 24-well cell culture plates. Twenty-four hours after seeding, the normal medium (Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum) was removed and replaced with phenol red–free Dulbecco's modified Eagle's medium containing 20% mouse serum in the presence of dimethyl sulfoxide or the anti-estrogen fulvestrant (ICI 182780; 10 μM).24 In addition, Fao cells were incubated with pooled serum from nonpregnant women, normal, pregnant women, or women with ICP. After 24 hours of incubation, total RNA was isolated.
pcDNA-RXR, pcDNA-FXRα2, pcDNAGal4-DBD-FXR-LBD, and pCMV-Renilla have been described elsewhere.11 pcDNA-ERα was a kind gift from Eric Kalkhoven. Glutathione S-transferase (GST)–FXR was generated by the cloning of FXRα2 into pGEX-4T-2 and was verified by sequencing.
Human embryonic kidney cells (HEK293T; ECACC 05030204) were plated onto 96-well plates in a phenol red–free medium supplemented with 5% dextran charcoal–stripped fetal serum. Cells were transfected with pcDNA-RXR and pcDNA-FXRα2, pcDNA-ERα/β together with pGL3-SHP promoter, and pCMV-Renilla. Alternatively, HEK293T cells were transfected with pcDNAGal4-DBD-FXR-LBD fusion constructs together with ERα and pGL3-Gal4 promoter. On the next day, fresh medium with or without 1 μM 3-(2,6-dichlorophenyl)-4-(3′-carboxy-2-chlorostilben-4-yl)oxymethyl-5-isopropylisoxazole (GW4064) and/or 10 nM estradiol was applied to the cells. After 24 hours, the luciferase activity was determined with the Promega dual-luciferase reporter assay system, and the Renilla luciferase activity was measured with a Centro LB 960 luminometer (Berthold Technologies, Bad Wildbad, Germany) to correct for the transfection efficiency. Transfection experiments were performed at least three times, and the results are shown as mean values of quadruplicates and standard deviations.
GST Pull-Down Assays.
Rosetta pLysS-competent bacteria (Novagen, EMD Chemicals, Inc., Darmstadt, Germany) were transformed with GST-FXRα2 expression plasmids. GST fusion protein expression was induced with 1 mM isopropyl β-d-1-thiogalactopyranoside and was purified as described elsewhere.25 GST fusion proteins were eluted from glutathione-Sepharose beads (Amersham Biosciences) with an elution buffer [20 mM glutathione, 100 mM trishydroxymethylaminomethane (pH 8.0), and 120 mM sodium chloride]. The full-length coding sequence of ERα/β in the pcDNA3.1 expression vector was transcribed and translated in vitro in a reticulocyte lysate in the presence of [35S]methionine (Amersham Biosciences) according to the manufacturer's protocol (TNT T7 coupled transcription/translation kit, Promega, Madison, WI). 35S-labeled proteins were incubated with GST fusion proteins in an NETN buffer [20 mM trishydroxymethylaminomethane (pH 8.0), 100 mM sodium chloride, 1 mM ethylene diamine tetraacetic acid, and 0.5% Nonidet P40] containing protease inhibitors (Complete, Roche Applied Bioscience). Samples were subsequently washed and subjected to sodium dodecyl sulfate–polyacrylamide gel electrophoresis. Gels were incubated with Amplify (GE Healthcare) to enhance the detection efficiency of 35S-labeled proteins. Coomassie brilliant blue staining was used to stain for the GST proteins.
Data are presented as means and standard errors of the mean of six animals or at least three in vitro experiments. Significance was calculated by analysis of variance. Significance is defined as P < 0.05.
Raised Hepatic Bile Acid Levels in Pregnant Mice.
Raised serum bile acid levels during pregnancy have been reported in mice21 and humans.14 Because elevated bile acid levels can occur in the serum for a number of reasons and do not necessarily indicate a defect in the hepatocytes, we aimed to determine whether hepatic bile acid concentrations are affected by pregnancy. We found that in normal, pregnant mice, hepatic bile acid concentrations were significantly higher than those in nonpregnant controls (Fig. 1). Indeed, the hepatic bile acids in pregnant mice were measured at levels comparable to those in cholate-fed mice (a model of bile acid overload) or Fxr−/− mice (a genetic model of cholestasis).
Hepatic Gene Expression Profiling of Pregnant, Cholate-Fed, and Fxr−/− Mice.
In order to investigate the rise in hepatic bile acids in pregnant mice, we initially aimed to compare the transcriptional effects of pregnancy, cholate feeding, and Fxr deficiency. To this end, we conducted gene expression microarrays.
Venn diagrams and hierarchical clustering were used to explore similarities between groups at the transcriptional level. Of the 27 genes regulated by both pregnancy and Fxr deficiency (Fig. 2A), 80% were affected in the same direction (increased/decreased expression) under both conditions (Fig. 2B). In contrast, the number of genes affected in the same direction by both pregnancy and cholate feeding (Fig. 2C) was the proportion that would have been affected by chance alone (52% of the 53 commonly affected genes; Fig. 2D).
To define the functional networks of the differentially expressed genes, data were analyzed with Ingenuity Pathway Analysis. These unsupervised investigations revealed that a gene network associated with cholestatic liver disease was the most statistically overrepresented network in pregnant, cholate-fed, and Fxr−/− mice (Fig. 2E). Therefore, global expression analysis demonstrates that pathways and networks regulated under conditions of bile acid overload or genetic cholestasis are also significantly affected by pregnancy.
Procholestatic Hepatic Gene Expression in Pregnant Mice.
We aimed to determine whether hepatic genes respond to the accumulation of bile acids during pregnancy or are more likely to be causative for the raised bile acid concentrations. To this end, we performed qRT-PCR assays on genes known to maintain bile acid homeostasis. These assays also served to confirm changes detected by microarrays.
As expected, in cholate-fed mice, the data were consistent with the adaptation of gene expression of bile acid homeostasis genes by Fxr activation. As such, cholate feeding induced hepatic Shp and Bsep expression, whereas Cyp7a1 and Ntcp expression was repressed (Fig. 3). In contrast, Fxr−/− mice showed elevated Cyp7a1 levels and reduced Bsep, Shp, and Mdr1a levels (Fig. 3).
Despite the presence of elevated hepatic bile acid concentrations, there was no evidence of Fxr activation in pregnant mice. Although Fxr protein levels were unaltered, the messenger RNA (mRNA) expression of the Fxr target gene Shp was significantly repressed (−2.8-fold; P < 0.01) during pregnancy, whereas its targets for repression, the bile acid biosynthesis enzymes Cyp7a1 and Cyp8b1,8 were up-regulated 1.6-fold (P < 0.05; Fig. 3). Similarly, pregnancy significantly reduced the expression of hepatic import genes [Ntcp, organic anion-transporting polypeptide 2 (Oatp2), and Oatp4] and export genes [Bsep, Mdr1a, and multidrug resistance–associated protein (Mrp3); Fig. 3]. Defective Fgf15 signaling from the intestine could contribute to the observed bile acid phenotype,9 but this does not seem to be the case because the expression of Fgf15 in the terminal ileum was unaffected by pregnancy (data not shown). Therefore, raised hepatic bile acid concentrations during pregnancy do not result in hepatic or intestinal Fxr activation. Instead, hepatic gene expression during pregnancy is procholestatic and resembles a state of Fxr inactivation because the majority of these genes are directly or indirectly regulated by Fxr.
Hepatic Bile Acids Do Not Accumulate in Pregnant, Cholate-Fed, or Fxr−/− Mice.
Our data indicate that hepatic bile acids accumulate in pregnant mice as a result of procholestatic gene expression resembling reduced Fxr function. We therefore assessed whether enhanced Fxr target gene transcription is sufficient to prevent further accumulation of hepatic bile acids during pregnancy. Indeed, hepatic bile acids did not further accumulate in pregnant cholate-fed mice (Fig. 4) in which anticholestatic mechanisms (such as the induction of Shp, Bsep, Oatp2, Mrp3, and Mdr1a and the repression of Cyp7a1 and Cyp8b1) were already induced (see Fig. 3). In addition, the hepatic accumulation of bile acids during pregnancy did not occur in Fxr−/− mice (Fig. 4), and this suggests that Fxr may be involved in causing this phenotype. These findings support our proposal that a gestational signal perturbs the feedback regulation of the enterohepatic circulation via Fxr.
Estradiol in the Serum of Pregnant Mice and Humans Represses Shp mRNA Expression in Fao Cells.
We aimed to determine whether a factor in the serum of pregnant mice is responsible for causing procholestatic hepatic gene expression. To this end, we incubated bile acid–responsive Fao cells with serum taken from nonpregnant and pregnant mice. We chose to determine the expression of the Fxr target gene Shp because this gene was robustly and consistently repressed in pregnant mice. In Fao cells, serum from pregnant mice caused a 2.5-fold reduction in Shp expression (Fig. 5A). The effect on Shp expression occurred despite the presence of significantly (P = 0.047) higher bile acid levels in the pregnant serum (15 μM) versus the nonpregnant control serum (2 μM). In this system, repression of Shp was also found in association with increased Cyp7a1 (P < 0.01; data not shown), and this suggests that the serum from pregnant mice reduced both the expression and function Shpin vitro. Because estrogens are implicated in the etiology of ICP in humans,16 we coincubated Fao cells with mouse serum and the ER antagonist fulvestrant (ICI 182780).24 Fulvestrant (10 μM) largely prevented the effect of the pregnant serum on Shp expression (Fig. 5A). Using Silastic implants, we found that pregnancy levels of 17β-estradiol alone were sufficient to reduce Shp expression in vivo (Fig. 5B). This regimen, however was not sufficient to cause raised hepatic bile acid levels (data not shown). Together, these data suggest that the increase in circulating estrogens in pregnant mice may contribute to the procholestatic gene expression observed during gestation.
Serum from pregnant women also reduced Shp expression in vitro (−1.3-fold) despite a small but significant increase (P = 0.04) in the serum bile acid concentration in the samples from pregnant women (8 μM) versus the samples from nonpregnant controls (2 μM) (Fig. 5C). Finally, serum from patients diagnosed with ICP (mean serum bile acid concentration = 61 μM) was used in these experiments. Despite the presence of raised bile acid levels that were expected to activate Fxr,26 the ICP serum caused a degree of repression of Shp similar to that caused by serum from pregnant individuals without ICP (Fig. 5C).
ERα Interacts with and Inhibits FXR Function in a β-Estradiol–Dependent Manner.
Because fulvestrant prevented the serum of pregnant mice from repressing Shpin vitro (Fig. 5A), we investigated whether estrogens (17β-estradiol) via hERα (the ER isoform expressed in hepatocytes) could inhibit the activity of hFXR. As expected, the activity of the hSHP promoter was increased upon stimulation with the FXR synthetic ligand GW4064 (Fig. 5D). However, in the presence of estradiol and ERα, FXR activity was repressed in reporter assays (Fig. 5D). Similar experiments showed that progesterone in the presence of progesterone receptor was unable to inhibit FXR function (data not shown). Next, by using an FXR-Gal4DBD fusion expression plasmid, we investigated whether ERα represses FXR activity regardless of direct DNA binding. Activity of the Gal4 promoter was induced by GW4064 incubation, but in the presence of ERα and β-estradiol, Gal4 promoter activity was reduced (Fig. 5E). GST pull-down assays subsequently revealed a physical interaction between FXR and ERα (Fig. 5F; data not shown), which was most abundant in the presence of both FXR and ER ligands. Together, these data demonstrate that ER can interact with FXR and perturb its function in an estradiol-dependent manner in vitro.
For previously unknown reasons, pregnancy alters bile acid homeostasis in humans14, 15 and can unmask cholestatic disease in predisposed but otherwise asymptomatic individuals.10 In this report, we show raised hepatic bile acid levels in normal pregnant mice, and we provide evidence of procholestatic gene expression caused by a functional, estradiol-dependent interaction between ER and FXR.
In agreement with two articles,27, 28 we measured a slight reduction in hepatic Fxr mRNA expression during gestation; however, this did not result in reduced Fxr protein expression. Importantly, pregnancy was associated with raised hepatic bile acid concentrations. This did not result in hepatic Fxr activation but rather seems to have been caused by pregnancy-associated inhibition of Fxr target gene transcription. Specifically, we observed reduced expression of transporter genes (Ntcp and Oatp2 for import and Bsep, Mrp3, and Mdr1a29 for export) important in bile homeostasis in combination with increased expression of bile acid synthesis genes (Cyp7a1 and Cyp8b1). Because most of these genes are under the direct or indirect regulation of Fxr, pregnancy is most likely to cause impaired Fxr activity, and this in turn is likely to be the cause of the raised hepatic bile acid concentrations in the pregnant mice. Notably, increased Cyp8b1 expression may result in more CA production versus chenodeoxycholic acid (CDCA) production. Indeed, the rate of CA (but not CDCA) synthesis has been reported to be higher in pregnant women versus nonpregnant controls,15 and the CA/CDCA ratio is increased in the serum of ICP cases versus women with uncomplicated pregnancies.30, 31
We propose that circulating estradiol likely contributes to the rise in hepatic bile acids during pregnancy. This is suggested by several lines of evidence. First, serum from pregnant mice represses Shp expression in vitro, and the effect was blocked by the ER antagonist fulvestrant. Second, slow-release implants mimicking pregnancy levels of estradiol also repressed Shp expression in ovariectomized mice. Third, ER interacted with FXR in the presence of estradiol and repressed its function in vitro. However, pregnancy concentrations of 17β-estradiol alone were not sufficient to significantly repress Shp expression in bile acid–treated Fao cells (data not shown). Together, these findings imply that metabolites of 17β-estradiol, highly present in the serum during pregnancy, may in fact be more potent inhibitors of Fxr function than 17β-estradiol itself.
The data presented here are broadly in agreement with studies reporting that estrogens are cholestatic in wild-type mice but not in ERα−/− mice.20 FXR has been reported to directly interact with ER in Michigan Cancer Foundation 7 (MCF-7) cells, and this results in repressed ER expression.32 Here we describe a different physiological setting in which ER and FXR interact with the result of reduced FXR function. Repression of Shp expression during pregnancy has been reported before.28 However, injections containing high doses of ethinyl estradiol have also been shown to activate Shp gene expression via Erα and possibly via Erα response elements in the Shp promoter.33 Although the reason for this inconsistency is at present unclear, it might be that this high dose of ethinyl estradiol, which is known to generate hepatotoxicity by itself,20 induces Shp expression. We propose that during pregnancy, Fxr loses its ability to respond to raising bile acid concentrations, and estradiol and its metabolites contribute to this perturbed function.
Impaired FXR function during pregnancy may explain the rise in serum bile acids observed in many pregnant women14 and is in agreement with the genetic etiology of ICP.11-13 However, in addition to raised levels of circulating estradiol and its metabolites, other procholestatic challenges are also likely to occur during pregnancy. For example, data from our laboratory suggest that metabolites of progesterone may disrupt bile acid homeostasis through nongenomic actions.34 In addition, changes in eating patterns and light/dark cycles occur during pregnancy,35 and similar alterations disrupt bile acid homeostasis in rodents.22 Dietary changes during gestation could also alter gut flora, and this situation is known to affect liver metabolism.36 ICP may manifest as a result of a combination of these environmental influences and is more likely to arise in individuals harboring functional variants of genes that maintain bile salt homeostasis.
In summary, we have found that hepatic bile acids accumulate during pregnancy in mice as a result of procholestatic gene expression, which is indicative of reduced Fxr function. Our data also imply that estrogens or estrogen metabolites, highly present in serum during pregnancy, may inhibit FXR function in vivo via their receptor ER. Our findings provide novel insights into the mechanisms by which pregnancy can unmask cholestatic disease in individuals harboring a genetic predisposition, such as functional variants in FXR.
The authors thank Shadi Abu-Hayyeh, Jenny Chambers, Victoria Geenes, Eric Kalkhoven, and Georgia Papacleovoulou for their assistance with this work.