Nuclear translocation of UDCA by the glucocorticoid receptor is required to reduce TGF-β1–induced apoptosis in rat hepatocytes

Authors


  • Potential conflict of interest: Nothing to report.

Abstract

Ursodeoxycholic acid (UDCA) inhibits classical mitochondrial pathways of apoptosis by either directly stabilizing mitochondrial membranes or modulating specific upstream targets. Furthermore, UDCA regulates apoptosis-related genes from transforming growth factor β1 (TGF-β1)–induced hepatocyte apoptosis by a nuclear steroid receptor (NSR)–dependent mechanism. In this study, we further investigated the potential role of the glucocorticoid receptor (GR) in the antiapoptotic function of UDCA. Our results with short interference RNA (siRNA) technology confirmed that UDCA significantly reduces TGF-β1–induced apoptosis of primary rat hepatocytes through a GR-dependent effect. Immunoprecipitation assays and confocal microscopy showed that UDCA enhanced free GR levels with subsequent GR nuclear translocation. Interestingly, when a carboxy-terminus deleted form of GR was used, UDCA no longer increased free GR and/or GR translocation, nor did it protect against TGF-β1–induced apoptosis. In co-transfection experiments with GR response element reporter and overexpression constructs, UDCA did not enhance the transactivation of GR with TGF-β1. Finally, using a flourescently labeled UDCA molecule, the bile acid appeared diffuse in the cytosol but was aggregated in the nucleus of hepatocytes. Both siRNA assays and transfection experiments with either wild-type or mutant forms of GR showed that nuclear trafficking occurs through a GR-dependent mechanism. In conclusion, these results further clarify the antiapoptotic mechanism(s) of UDCA and suggest that GR is crucial for the nuclear translocation of this bile acid for reducing apoptosis. (HEPATOLOGY 2005;42:925–934.)

Ursodeoxycholic acid (UDCA) is widely used in the treatment of cholestatic liver diseases. Its therapeutic effects have been attributed to several mechanisms,1 including modulation of classic mitochondrial pathways of apoptosis.2–4 Nevertheless, reduction of cell death by UDCA may also involve alternate and upstream molecular targets of the E2F-1/Mdm-2/p53 apoptotic pathway5 mediated through nuclear steroid receptors (NSRs).6 Transforming growth factor β1 (TGF-β1) is a multifunctional cytokine that induces growth arrest and apoptosis in hepatic cells, in part through the E2F-1 transcription factor.7, 8 Our previous studies indicated that UDCA modulates the expression of apoptotic target genes induced by TGF-β1 by increasing both expression and nuclear translocation of the glucocorticoid (GR) and the mineralocorticoid (MR) receptors.6

The activation of NSR is modulated in vivo by compounds that interact directly with either the receptor moiety or its associated proteins.9 UDCA is a cholesterol-derived molecule, suggesting a possible interaction between this bile acid and NSR. In fact, a number of bile acids bind and inactivate the farnesoid X-activated receptor (FXR).10, 11 Although UDCA itself does not bind FXR,10 it inhibits its activation by more hydrophobic bile acids.12 Interestingly, other studies have demonstrated that UDCA interacts with GR.13 The bile acid was shown to directly activate GR,14 as well as to interact with distinct regions of its ligand binding domain (LBD) without eliciting GR transcriptional function.15 Nevertheless, it is not known whether this potential interaction with the LBD of GR is required for the antiapoptotic effect of UDCA.

GR is a member of the nuclear receptor superfamily and is an important transcriptional regulator involved in diverse physiological functions.16 In the absence of ligand binding, GR is primarily inactive and located in the cytoplasm by association with a variety of chaperone proteins. After ligand binding, GR dissociates from these proteins and becomes actively involved in either transcriptional or nontranscriptional processes. In fact, GR can act as a ligand-dependent transcription factor that upregulates gene expression through interaction with DNA enhancer sequences of the glucocorticoid response elements (GREs).17 Further, GR can negatively regulate expression of inflammatory genes through direct protein–protein interaction with transcription factors in the absence of DNA binding.18

It has been reported that transactivation of NSR regulates cell death in several models of apoptosis.19, 20 Moreover, the therapeutic activities of glucocorticoids are thought to be mediated by GR.21 GR has heterogeneous functions in modulating apoptosis, influenced by alternative initiation sites and different effects of co-modulators, among others.22 Nevertheless, GR prevents liver cell death by upregulating antiapoptotic members of the Bcl-2 family, including Bcl-xL.23, 24

Hormone-bound GR rapidly and continuously shuttles between the nucleus and cytosol.25 UDCA may require GR to only reach the nucleus of cells where the bile acid itself can regulate several apoptosis-related genes without requiring GR transactivation. In fact, UDCA has been shown to accumulate in the nucleus of rat hepatocytes.26, 27 It is possible that a potential interaction between NSR and UDCA may result in regulation of several pathways during apoptosis.

We previously demonstrated that UDCA prevents TGF-β1–induced hepatocyte apoptosis by a NSR-dependent mechanism. Here we further explored the molecular events linking NSR and the protective role of UDCA during apoptosis of hepatocytes from TGF-β1. Our results show that a specific region of the LBD is essential in rat hepatocytes for nuclear translocation of UDCA and the potential modulation of gene expression for inhibiting apoptosis.

Abbreviations:

UDCA, ursodeoxycholic acid; NSR, nuclear steroid receptor; TGF-β1, transforming growth factor β1; GR, glucocorticoid receptor; MR, mineralocorticoid receptor; FXR, farnesoid X-activated receptor; LBD, ligand binding domain; GRE, glucocorticoid response element; TUDCA, tauroursodeoxycholic acid; GFP, green fluorescent protein; Wt, wild-type; NBD, nitrobenzoxadiazolyl; hsp90, heat shock protein 90.

Materials and Methods

Hepatocyte Isolation and Cell Culture.

Rat primary hepatocytes were isolated from male Sprague-Dawley rats (100-150 g) by collagenase perfusion,5, 28 resuspended in complete William's E medium (Invitrogen Corp., Grand Island, NY), and plated on Primaria tissue culture dishes (BD Biosciences, San Jose, CA) at either 2 × 104 cells/cm2 for transfection assays or 6 × 104 cells/cm2 for all other experiments.

Induction of Apoptosis.

Freshly isolated hepatocytes were cultured for 6 hours as described above, washed, and then incubated in 10% fetal bovine serum (Atlanta Biologicals Inc., Norcross, GA) William's E medium supplemented with either 100 μmol/L UDCA or tauroursodeoxycholic acid (TUDCA; Sigma Chemical Co., St. Louis, MO) or no addition (control) for 12 hours. Cells were then exposed to 1 nmol/L recombinant human TGF-β1 (R & D Systems Inc., Minneapolis, MN) for 24 hours. Attached and floating cells were combined to extract cytosolic and total proteins.

Morphologic Evaluation of Apoptosis and Caspase Activation.

Hoechst labeling of cells was used to detect apoptotic nuclei.5 General caspase-3–like activity was determined by enzymatic cleavage of chromophore p-nitroanilide (pNA) from the substrate N-acetyl-Asp-Glu-Val-Asp-pNA (DEVD-pNA; Sigma Chemical Co.) in cytosolic protein extracts (Complete; Roche Applied Science, Mannheim, Germany).5 In addition, caspase-3 cleavage was determined by immunoblotting.

Transfections with Short Interference RNA and GFP-GR Chimeric Proteins.

Transfection with siRNA was performed as previously described.6 In addition, hepatocytes were transiently transfected with the expression plasmids for the chimeric proteins of green fluorescent protein (GFP) and human GR, including the wild-type (Wt) form, GFP-GR Wt, the carboxy-terminus partially deleted form of LBD, GFP-GR (1-765), and the carboxy-terminus completely deleted form of LBD, GFP-GR (1-730). GFP-GR mutant forms were generated by cloning different polymerase chain reaction products into the GFP-GR Wt plasmid, all under CMV-IEP T7 enhancer/promoter control.29, 30 At 24 hours after plating, 40% confluent hepatocytes were transfected with 5 μg of each plasmid complexed with polyethylenimine.31

Immunoblotting.

Steady-state levels of GR, GFP, and heat shock protein 90 (hsp90) proteins, as well as caspase-3 cleavage were determined by Western blot, using primary rabbit polyclonal antibodies reactive to GR, hsp90, and caspase-3 (Santa Cruz Biotechnology, Santa Cruz, CA) or a primary mouse monoclonal antibody to GFP (BD Biosciences), as well as secondary antibodies conjugated with horseradish peroxidase (Bio-Rad Laboratories, Hercules, CA). Membranes were processed for protein detection using Super Signal substrate (Pierce, Rockford, IL). Protein concentrations were determined using the Bio-Rad protein assay kit.

Immunoprecipitation.

Binding of GR to hsp90 was detected by immunoprecipitation analysis. In brief, whole cell extracts were prepared by lysing cells in M-PER Mammalian Protein Extraction Reagent (Pierce). Immunoprecipitation experiments were carried out using antibodies to GR and the Ezview Red Protein G Affinity Gel (Sigma Chemical Co.). Typically, 200 μg of lysate were incubated with 1 μg primary rabbit polyclonal antibody to GR overnight at 4°C. Immunoblots were then probed with the mouse monoclonal anti-hsp90 antibody. GR expression was determined in the same membrane after stripping off the immune complex for the detection of hsp90. In parallel, 20 μg whole cell extract was independently used for immunodetection of GR and hsp90. Immunoprecipitation assays using high-detergent conditions as well as Western blot analysis showed absence of nonspecific binding of the GR antibody to hsp90. In addition, immunoprecipitation assays using the mouse monoclonal antibody reactive to GFP were performed to confirm the results obtained with the GR antibody. Finally, immunoprecipitation assays using the mouse monoclonal antibody reactive to β-actin demonstrated no association with either hsp90 or GR.

Subcellular Localization of GFP Fusion Proteins.

Analysis of intracellular trafficking of GFP-GR in primary rat hepatocytes was performed by using cells transiently transfected with chimeric proteins of GFP-GR Wt, GFP-GR (1-765), or GFP-GR (1-730). For detection of GFP-fluorescence, transfected hepatocytes were incubated at 30°C for 4 hours, fixed with 4% paraformaldehyde in phosphate-buffered saline (PBS), pH 7.4 at room temperature for 10 minutes, washed, and mounted using Fluoromount-GTM (Southern Biotech, Birmingham, AL). Fluorescence was visualized using a MRC1000 confocal microscope (Bio-Rad).

Transactivation of Nuclear Steroid Receptors.

Transactivation of NSRs was investigated by co-transfecting hepatocytes with a GR/MR-responsive reporter plasmid, pGRE/MRE-luciferase, and the Wt human GR, pRShGR, or the Wt human MR, pRShMR. The pGRE/MRE plasmid consisted of the entire human GR/MR-responsive promoter fused to the luciferase gene32; pRShGR and pRShMR overexpression plasmids were under SV40 enhancer/promoter control.29 Twelve hours after plating, hepatocytes at 40% confluence were transfected with 3 μg of pGRE/MRE-luciferase plasmid and 10 ng of each receptor expression plasmid, pRShGR or pRShMR. To assess transfection efficiency of the polyethylenimine complexes, hepatocytes were co-transfected with the chloramphenicol acetyltransferase reporter construct, PGL3-Control vector (Promega Corp., Madison, WI). The cells were harvested for luciferase assays (Promega Corp.) and CAT ELISA (Roche Applied Science, Mannheim, Germany).

Intracellular Distribution of UDCA.

Cellular distribution of UDCA was visualized by incubating hepatocytes with 3α-hydroxy-7-nitrobenzoxadiazolyl (NBD), 5β-cholan-24 oic acid (NBD-UDCA) fluorescent molecule, which appears to have transport properties similar to those of natural bile acids.33 Freshly isolated hepatocytes were cultured for 6 hours or transfected with either GR-siRNA or GFP-GR chimeric forms as described above, and then washed and incubated in fresh medium supplemented with 100 μmol/L of either NBD-UDCA or unlabeled UDCA for 5, 15, and 30 minutes, and 1 and 2 hours. Competition studies were performed by pretreating primary rat hepatocytes with 500 μmol/L unlabeled UDCA for 5 minutes, washing, and replacing the medium with 100 μmol/L NBD-UDCA molecule for 1 and 2 hours. The attached cells were washed three times with PBS and fixed with 4% paraformaldehyde in PBS, pH 7.4, for 10 minutes at room temperature. Following additional washes, samples were mounted using Fluoromount-G, and fluorescence was visualized using a MRC1000 confocal microscope.

Statistical Analysis.

Statistical analysis was performed using GraphPad InStat version 3.00 for Windows 95 (GraphPad Software, San Diego, CA) for the ANOVA and Bonferroni's multiple comparison tests. P values less than .05 were considered statistically significant.

Results

The Antiapoptotic Activity of UDCA Is Associated with GR/hsp90 Dissociation.

Approximately 20% of apoptosis was detected in primary rat hepatocytes after 24 hours of incubation with TGF-β1 (P < .01), and this was associated with a similar increase in both caspase-3–like activity and caspase-3 cleavage (P < .01) (Fig. 1). In addition, the protective effect of UDCA against TGF-β1–induced apoptosis was clearly demonstrated, and the taurine-conjugated derivative, TUDCA, showed similar protection (P < .01).

Figure 1.

UDCA inhibits apoptosis induced by TGF-β1 in primary rat hepatocytes. Cells were incubated with 1 nmol/L TGF-β1, or no addition (control), with or without 100 μmol/L UDCA, or TUDCA for 24 hours. In co-incubation experiments, hepatocytes were pretreated with either UDCA or TUDCA 12 hours prior to incubation with TGF-β1. Cells were fixed and stained for morphological evaluation of apoptosis, and cytosolic proteins were extracted for caspase activity as described in Materials and Methods. (A) Fluorescent microscopy of Hoechst staining (top); and percent of apoptosis (bottom) in control hepatocytes and cells exposed to TGF-β1 with or without bile acids. Original magnification ×400. (B) DEVD-specific caspase activity (left) and representative immunoblot of caspase-3 cleavage (right) in cytosolic fractions after incubation with TGF-β1 for 24 hours with or without bile acids. The results are expressed as mean ± SEM for at least 3 different experiments. *P < .01 from control; ‡P < .01 and †P < .05 from TGF-β1.

To confirm that GR is an important regulatory factor in the antiapoptotic function of UDCA, we performed posttranscriptional gene silencing experiments for the NSR. GR expression decreased by ∼70% in UDCA-treated hepatocytes after transfection with the siRNAs (P < .05) (Fig. 2A). More importantly, both changes in nuclear morphology (Fig. 2B) and caspase activation (Fig. 2C) indicated that GR-specific siRNA transfection abolished UDCA protection against TGF-β1–mediated apoptosis, whereas the nonspecific control duplex had no effect.

Figure 2.

The antiapoptotic activity of UDCA is mediated via GR in primary hepatocytes. Four hours after control or GR siRNA transfections, either vehicle or 100 μmol/L UDCA final concentration was added to cells. After an additional 12 hours, 1 nmol/L TGF-β1 was included in the cultures for 24 hours. Cells were fixed and stained for morphological evidence of apoptosis, and cytosolic proteins were extracted for caspase activity as described in Materials and Methods. (A) Representative immunoblot of GR in cells treated with UDCA and transfected with either control (C siRNA) or GR siRNA. β-Actin was used to control for lane loading. To assess GR silencing, protein levels of GR were quantified by immunoblotting; transfection efficiency of siRNA for GR was approximately 70% in UDCA-treated hepatocytes. (B) Fluorescent microscopy of Hoechst staining (top); and percent of apoptosis (bottom) in hepatocytes transfected with control (a) or GR (b) siRNA and incubated with TGF-β1 plus UDCA. Original magnification, ×400. (C) DEVD-specific caspase activity (left) and representative immunoblot of caspase-3 cleavage (right) in cells transfected with control (C siRNA) or GR siRNA and exposed to TGF-β1 with or without UDCA. The results are expressed as mean ± SEM for at least 3 different experiments. *P < .01 and §P < .05 from respective controls; ‡P < .01 and †P < .05 from cells exposed to TGF-β1.

The effect of TGF-β1 with or without UDCA on complex formation between GR and its chaperone hsp90 was analyzed by immunoprecipitation assays. TGF-β1 induced a slight increase in GR/hsp90 association (Fig. 3A). Notably, UDCA alone induced a ∼40% decrease of GR/hsp90 binding compared with controls (P < .01). More importantly, co-incubation with UDCA significantly decreased GR/hsp90 binding compared with TGF-β1–treated cells (P < .01), suggesting that UDCA-mediated increase of free GR may contribute to its antiapoptotic effect. Interestingly, parallel Western blots showed that TGF-β1 markedly reduced total levels of hsp90 (P < .01), which in turn may contribute to an increase of free GR levels in cells co-incubated with UDCA (Fig. 3B). Thus, these data suggested that UDCA either promotes the dissociation between GR and its cytosolic chaperone hsp90 and/or increases free GR levels by enhancing GR expression.6

Figure 3.

The antiapoptotic activity of UDCA induces GR/hsp90 dissociation. Cells were incubated with either 1 nmol/L TGF-β1 or no addition (control), with or without UDCA for 24 hours. In co-incubation experiments, UDCA was added to hepatocytes 12 hours prior to incubation with TGF-β1. Total proteins were extracted for immunoprecipitation or Western blot assays as described in Materials and Methods. (A) Immunoprecipitation analysis of GR/hsp90 association. Representative immunoblots with hsp90- and GR-specific antibodies (top), and a histogram of GR/hsp90 association (bottom) in cells exposed to TGF-β1 for 24 hours with or without UDCA. All densitometry values for hsp90 were normalized to the respective GR expression, and the results are expressed as mean ± SEM arbitrary units for at least 6 different experiments. *P < .01 from respective control; ‡P < .01 from cells exposed to TGF-β1. (B) Representative immunoblot of total levels of hsp90 in cells exposed to TGF-β1 with or without UDCA for at least 4 different experiments. All densitometry was normalized to endogenous β-actin protein expression.

The COOH-terminal Region of GR LBD Is Required for UDCA Protection.

Since the dissociation of NSR requires ligand binding,9 we determined whether the LBD region is involved in UDCA-induced dissociation of GR/hsp90. Hepatocytes were transfected with plasmids expressing GFP fusion proteins with either GR Wt or LBD mutant forms of GR (Fig. 4A). Complex formation of the chimeric proteins with hsp90 was analyzed by immunoprecipitation after TGF-β1 with or without bile acid incubations (Fig. 4B). All chimeric proteins associated with hsp90. With UDCA incubation, both GR Wt and the partial deleted form of GR LBD (1-765), were readily dissociated from hsp90. In fact, the interaction between GR Wt and hsp90 decreased ∼30% with UDCA alone and ∼70% with UDCA plus TGF-β1 (P < .05). Interestingly, partial deletion of GR LBD (1-765) did slightly affect the interaction between UDCA and GR. Nevertheless, UDCA protection was still associated with a marked GR/hsp90 dissociation (P < .05). In contrast, deletion of the carboxy-terminus of GR LBD (1-730) completely abolished the effect of UDCA on GR/hsp90 association, indicating that UDCA requires this specific region to interact with the NSR. Curiously, in primary rat hepatocytes this carboxy-terminus of GR LBD appears to affect only UDCA interaction and not chaperone binding. Immunoprecipitation assays using a GFP antibody confirmed the results obtained with the GR antibody (data not shown). Thus, the data indicated that UDCA mediates GR/hsp90 dissociation, which in turn may induce nuclear translocation of GR.

Figure 4.

The carboxy-terminus of GR is required for the antiapoptotic effect of UDCA. Cells were transfected with Wt and LBD mutant forms of GR, GFP-GR Wt, GFP-GR (1-765), or GFP-GR (1-730) as described in Materials and Methods. Twelve hours later, vehicle or 100 μmol/L UDCA was added to cells. After an additional 12 hours, 1 nmol/L TGF-β1 was included in the cultures for 24 hours. Hepatocytes were harvested for immunoprecipitation assays or fixed for confocal microscopy analysis and Hoechst staining as described in Materials and Methods. (A) Schematic drawing of Wt and LBD mutant forms of GFP-GR plasmids. (B) Histogram of the association between GR Wt, GR (1-765), and GR (1-730) and hsp90 in cells exposed to TGF-β1 with or without UDCA. All densitometry for hsp90 was normalized to the respective GR expression. The results are expressed as mean ± SEM arbitrary units for at least 4 different experiments. (C) Fluorescent microscopy of GFP staining in control hepatocytes and cells exposed to TGF-β1 with or without UDCA. Representative photographs of at least 4 different experiments. Original magnification ×400. (D) Percentage of apoptosis in cells transfected with GR Wt, GR (1-765), and GR (1-730) and exposed to TGF-β1 with or without UDCA. The results are expressed as mean ± SEM for at least 3 different experiments. *P < .01 from respective controls; †P < .05 from cells exposed to TGF-β1.

To confirm this we analyzed the subcellular localization of Wt and LBD mutant forms of GFP-GR by confocal microscopy (Fig. 4C). GR Wt localized primarily to the cytosol in the absence of UDCA. In contrast, UDCA-treated hepatocytes resulted in significant nuclear accumulation of green fluorescence, confirming that UDCA-induced GR/hsp90 dissociation results in GR nuclear translocation. In addition, UDCA-induced nuclear translocation of the partial deletion mutant GR LBD (1-765) compared with controls and TGF-β1 alone. UDCA, however, did not translocate the completely carboxy-terminus deleted form of GR LBD (1-730), reinforcing the notion that the carboxy-terminus region of LBD is required for UDCA-induced nuclear translocation of GR. Thus, the interaction of UDCA with this specific region of the GR LBD results in nuclear translocation of the receptor, which in turn may be required for UDCA protection against TGF-β1–induced apoptosis.

Apoptosis was evaluated in hepatocytes after transfection with the GFP-GR expression plasmids to determine whether the interaction of UDCA with the carboxy-terminus region of the GR LBD was critical for its antiapoptotic function. The results showed that, unlike GR Wt and the partial deleted form of GR LBD (1-765), GR LBD (1-730) completely abolished UDCA protection against TGF-β1–mediated apoptosis (Fig. 4D). Thus, UDCA requires a specific region of GR LBD to induce GR nuclear translocation and inhibit apoptosis.

The Protective Role of UDCA Does Not Require NSR Transactivation.

Although UDCA was shown to induce GR nuclear translocation in primary rat hepatocytes, the role of NSR in the antiapoptotic function of UDCA was unclear. Therefore, we measured GR transactivation in hepatocytes exposed to TGF-β1 with or without bile acid by co-transfecting cells with a GR/MR responsive element-reporter construct, pGRE/MRE-luciferase, and a GR overexpression plasmid, pRShGR. Curiously, the results revealed that the antiapoptotic effect of UDCA did not involve a GR transactivation–dependent mechanism (Fig. 5A). In fact, co-incubation with UDCA reduced GR response element activation compared with TGF-β1 alone. Moreover, when cells were co-transfected with the GR overexpression plasmid, pRShGR, GR activity was markedly increased (P < 0.01). Both TGF-β1 and UDCA further enhanced this activity, although TGF-β1–induced GRE activation was significantly decreased by co-incubation with UDCA (P < .05). Similar results were obtained in cells co-transfected with the MR overexpression plasmid, pRShMR (data not shown). These results indicate that during TGF-β1–induced apoptosis, UDCA pretreatment results in nuclear translocation of NSRs but no significant additional increase in their transactivation. Thus, UDCA-induced GR transactivation does not appear to be a prerequisite for UDCA protection against TGF-β1–induced apoptosis compared with GR transactivation triggered by TGF-β1 alone.

Figure 5.

GR mediates nuclear translocation of UDCA. (A) The antiapoptotic activity of UDCA is not a GR transactivation–dependent mechanism. Cells were co-transfected with a GR/MR responsive element-reporter construct, pGRE/MRE-luciferase, and a GR overexpression plasmid, pRShGR, as described in Materials and Methods. Twelve hours later, vehicle or 100 μmol/L of UDCA was added to cells. After an additional 12 hours, 1 nmol/L TGF-β1 was included in the cultures and cells harvested for luciferase assays. To assess transfection efficiency, hepatocytes were co-transfected with the CAT reporter construct, PGL3-Control vector. Based on this control, transfection efficiencies were ∼70% and did not differ between reporter and expression plasmids. Histogram of the GRE transactivation in cells exposed to TGF-β1 for 24 hours with or without UDCA, in the presence or absence of GR overexpression. Luciferase activity (relative light units [RLU]/mg protein) was normalized to control CAT expression and the results are expressed as mean ± SEM for 12 different experiments. *P < .01 and §P < .05 from respective control; ‡P < .01 and †P < .05 from cells exposed to TGF-β1. (B) Subcellular localization of UDCA in primary rat hepatocytes. Representative photographs of confocal microscopy of either unlabeled UDCA (a) or NBD-UDCA (b) staining in hepatocytes and of NBD-UDCA staining in cells after transfection with control (c) or GR (d) siRNA. Hepatocytes were either cultured for 6 hours or transfected with GR-siRNA as described in Materials and Methods. Four hours after platting or transfection with control or GR siRNA, 100 μmol/L of either unlabeled UDCA or NBD-UDCA was added to cells for 2 hours. Original magnification ×400. (C) Subcellular localization of UDCA in primary rat hepatocytes transfected with Wt and LBD mutant forms of GFP-GR plasmids. Representative photographs of either unlabeled UDCA (a) or NBD-UDCA staining in hepatocytes. Cells were transfected with either GR Wt (b), GR (1-765) (c), or GR (1-730) (d). Competition studies demonstrated the specificity of NBD-UDCA molecule in primary rat hepatocytes (data not shown). Original magnification ×400.

GR Modulates Nuclear Trafficking of UDCA.

It has been demonstrated that bile acids are detectable within the nuclei of hepatocytes,26, 27, 34 where they may play an important role in controlling gene expression. Further, although not dependent on GR-induced transcriptional modulation, the antiapoptotic effect of UDCA is associated with GR nuclear translocation. Thus, we determined whether the association between UDCA and GR results in nuclear translocation of the bile acid. We investigated the intracellular trafficking of UDCA in primary rat hepatocytes by incubating cells with a fluorescent UDCA molecule, NBD-UDCA (Fig. 5B). The results revealed that unlabeled UDCA did not show any significant fluorescence. In contrast, the NBD-UDCA molecule was clearly detectable within the cells. The fluorescent signal was weak at 5 minutes; however, it became increasingly stronger with 15 minutes to 2 hours of incubation in hepatocytes (data not shown). Although the results indicated that UDCA was localized in the major subcellular compartments, NBD-UDCA staining appeared diffuse in the cytosol and aggregated in the nucleus.

To investigate the role of GR on nuclear translocation of UDCA, we performed posttranscriptional gene silencing experiments for GR and evaluated the subcellular localization of the bile acid (Fig. 5B). In fact, siRNA experiments showed that by silencing GR, the nuclear translocation of UDCA was significantly reduced, indicating that nuclear trafficking of UDCA is a GR-dependent mechanism. Similar results were obtained, following endogenous silencing of MR (data not shown).

To further evaluate the role of GR in nuclear trafficking of UDCA, we overexpressed Wt and LBD mutant forms of GFP-GR in hepatocytes and investigated the subcellular localization of the UDCA fluorescent molecule (Fig. 5C). GFP fluorescence was undetectable under the conditions used to visualize the NBD fluorescence from the bile acid. In addition, the overexpression of both GR Wt and GR LBD (1-765) resulted in significant nuclear accumulation of green fluorescence, confirming that GR induces UDCA nuclear translocation. GR LBD (1-730), however, did not translocate UDCA to the nucleus, reinforcing the notion that the carboxy-terminus region of LBD is required for GR-induced nuclear translocation of UDCA. Collectively, these data strongly suggest that UDCA is translocated into the nucleus of primary rat hepatocytes by interacting with a specific region of NSR, and this mechanism is essential for its antiapoptotic properties.

Discussion

The mechanisms by which UDCA triggers signaling pathways involved in the control of cell death are not fully understood. We have previously reported that UDCA reduces TGF-β1–induced hepatocyte apoptosis by both inhibiting classical mitochondrial pathways2–4 and modulating upstream targets, including the E2F-1/Mdm-2/p53 apoptotic pathway, in a caspase-independent manner.5 In addition, we have also shown that UDCA modulates apoptosis-related genes through an NSR-dependent mechanism.6 The results presented here provide an additional role for NSRs in the antiapoptotic function of UDCA. The bile acid appears to interact with GR through the carboxy-terminus region of LBD. Moreover, GR is required for nuclear translocation of UDCA as a prerequisite for its modulation of gene expression and apoptosis.

Several studies have described NSRs as intracellular bile acid–sensing transcription factors. In fact, many FXR target genes play critical roles in bile acid synthesis and transport. UDCA, a relatively hydrophilic bile acid compared with FXR ligands, does not activate FXR10 but rather acts as a biological response modulator of GR in hepatocytes.13, 35 We have recently demonstrated that in the absence of glucocorticoid ligands, nuclear levels of GR increase in primary rat hepatocytes during inhibition of TGF-β1–induced apoptosis by UDCA.6 However, it was unclear whether the bile acid interacts with GR and/or requires GR transactivation to protect hepatocytes against apoptosis.

In this study, we confirmed the protective effect of UDCA as well as the essential role of GR in TGF-β1–induced hepatocyte apoptosis using GR siRNAs. Moreover, we demonstrated that UDCA decreases the association between GR and its molecular chaperone, hsp90, and induces subsequent GR translocation into the cell nucleus. Interestingly, TGF-β1 was associated with a pronounced degradation of hsp90, which may further enhance UDCA-mediated increase of free GR, as seen in co-incubated hepatocytes. Two possibilities may explain the increase of free GR in the presence of UDCA. First, the ratio of GR/hsp90 binding might vary because UDCA interacts with GR, inducing its dissociation; or second, UDCA inhibits TGF-β1–induced degradation of GR,6 thus increasing levels of free GR after hsp90 degradation. Nevertheless, our results demonstrated that the LBD of GR is at least one target domain for UDCA. In fact, using Wt and mutant GFP-GR fusion constructs we showed that the specific carboxy-terminal region of the GR LBD is essential for UDCA-mediated decrease of GR/hsp90 binding, subsequent GR nuclear translocation, and protection from apoptosis. The COOH-terminal deleted form of GR (1-730) lacks Tyr735 of the steroid binding pocket, which is thought to be necessary for the hydrophobic contact between ligands and NSR.36 Finally, this specific region of GR is required for inhibition of apoptosis by the bile acid. UDCA may either directly bind to the specific COOH-terminal region of GR LBD or interact via a second messenger signal. Bile acids and oxysterols are natural ligands for several orphan nuclear receptors in hepatocytes, such as the liver X receptor (LXR) and FXR,10, 11 and UDCA mimics cholesterol in several molecular events.37 However, the nuclear translocation of UDCA in a GR-dependent manner strongly suggests that UDCA and GR may directly bind to each other as a UDCA-GR complex. Nevertheless, a direct interaction between the classical GR and the UDCA molecule has not been identified. Interestingly, UDCA appears to require the same region of the LBD to promote GR nuclear translocation and suppress inflammation in hepatic cells.15

Although glucocorticoids are well-known inhibitors of apoptosis in hepatic cells,24, 38, 39 we showed that the antiapoptotic function of UDCA does not require induction of GRE-dependent reporter gene expression transactivation. Other studies have shown that UDCA attenuates the interaction between GR and its co-activators.14, 15, 40 Here, we demonstrated that UDCA alone increases the activation of GRE in hepatocytes overexpressing GR, but does not protect hepatocytes against TGF-β1–induced apoptosis by further increasing GR transactivation.

To explore other possible roles for UDCA-induced nuclear translocation of GR during hepatocyte apoptosis, we considered that GR may function as nuclear transporter of UDCA. Using a fluorescently labeled UDCA molecule we clearly showed nuclear localization of UDCA in primary rat hepatocytes. Bile acids have previously been detected in the nucleus of rat hepatocytes.26, 27, 34 Further, it has also been demonstrated that only a small proportion of the total bile acid pool resides within the liver, suggesting that bile acids are taken up selectively by the hepatocyte nuclei.27 This observation, however, does not exclude an effect of UDCA in other cellular compartments; UDCA has been shown to co-localize with mitochondria in hepatic cells.41 The precise role of UDCA in the nucleus during hepatocyte apoptosis is still unclear.

UDCA was visualized as aggregates in the nucleus, indicating that the bile acid may either form complexes itself or interact with other protein complexes. Interestingly, bile acids may directly bind to DNA; 32P-labeling analysis of DNA-reactive bile acids provided evidence of DNA-adduct formation.42 Conversely, others have shown that bile acids do not bind covalently to DNA.43 It is also possible that UDCA interacts with transcriptional factors, thereby modulating several genes of the E2F-1/Mdm-2/p53 apoptotic pathway, as previously demonstrated.3

Finally, to assure that nuclear translocation of UDCA is a NSR-dependent mechanism, we determined nuclear translocation of the bile acid after both endogenous silencing of GR and overexpression of chimeric forms of GFP-GR. GR-induced translocation of UDCA to the nucleus6, 13 suggests that UDCA binds to cytosolic GR and is translocated as part of the ligand-receptor complex. In fact, GR has been shown to shuttle between the nucleus and the cytoplasm in addition to recycling between a naive, chaperone-associated cytoplasmic complex and a ligand chaperone-free nuclear form.25, 44 Curiously, MR and GR may have similar roles in promoting nuclear trafficking of UDCA, suggesting that the receptors may somehow interact with each other, possibly as a heterodimer, and induce UDCA nuclear translocation. Thus, UDCA appears to reach the nucleus through mechanisms similar to those used by steroid hormones. However, it remains to be determined whether GR is only required for the nuclear transport of the bile acid or if it is also transrepressed during the antiapoptotic function of UDCA. It has been postulated that negative regulation of NSRs may result from a ligand interaction that also binds to other transcription factors without requiring DNA binding.45

In conclusion, our study involves GR in the nuclear trafficking of UDCA during apoptosis of primary rat hepatocytes. The results demonstrated that UDCA targets a specific region of the COOH-terminal region of the GR LBD and dissociates GR from its cytosolic chaperone, hsp90. GR is then translocated into the nucleus together with UDCA, ultimately protecting hepatocytes from undergoing apoptosis. Finally, since GR transactivation is not a prerequisite for the antiapoptotic effect of UDCA, identification of additional co-modulators may result in the development of novel therapeutic strategies for apoptosis in liver diseases.

Acknowledgements

We are deeply grateful to Dr. Alan F. Hoffman, University of California, San Diego, for the generous gift of NBD-UDCA. The authors also thank Ms. Xiaoming Ma for skillful technical assistance.

Ancillary