p53 Is a Key Molecular Target of Ursodeoxycholic Acid in Regulating Apoptosis*

p53 plays an important role in regulating expression of genes that mediate cell cycle progression and/or apoptosis. In addition, we have previously shown that the hydrophilic bile acid ursodeoxycholic acid (UDCA) prevents transforming growth factor β1-induced p53 stabilization and apoptosis in primary rat hepatocytes. Therefore, we hypothesized that p53 may represent an important target in bile acid-induced modulation of apoptosis and cell survival. In this study we demonstrated that UDCA reduces p53 transcriptional activity, thereby preventing its ability to induce Bax expression, mitochondrial translocation, cytochrome c release, and apoptosis in primary rat hepatocytes. More importantly, bile acid inhibition of p53-induced apoptosis was associated with decreased p53 DNA binding activity. Subcellular localization of p53 was also altered by UDCA. Both events appear to be related with increased association between p53 and its direct repressor, Mdm-2. In conclusion, these results further clarify the antiapoptotic mechanism of UDCA and suggest that modulation of Mdm-2/p53 interaction is a prime target for this bile acid.

The tumor suppressor protein p53 is a transcription factor that plays an important role in regulating expression of genes that mediate cell cycle arrest and/or apoptosis in response to a wide variety of cellular stress factors (8,9). Although its role in suppressing cell cycle progression has been extensively described (10,11), less is known about the mechanisms by which p53 induces apoptosis. Curiously, recent studies have revealed that p53 can mediate apoptosis by a transcription-independent process (12). Nevertheless, cells in which wild-type p53 was replaced by a transcriptionally inactive mutant showed loss of both cell cycle arrest and apoptotic functions, supporting the idea that transcriptional activity is of paramount importance in these cellular responses (13,14). In fact, it is thought that p53 signals apoptosis through its activity as a sequence-specific transcriptional activator of proapoptotic target genes, such as bax, Noxa, or PUMA (15)(16)(17). These proteins are translocated to mitochondria, where they promote loss of the mitochondrial membrane potential and cytochrome c release, thus activating the Apaf-1/caspase-9 apoptotic cascade (18). Indeed, Apaf-1 itself has also been described as a transcriptional target for p53 (19).
Interestingly, in unstressed conditions p53 is a short-lived protein. Pivotal to its regulation is the function of the murine double minute-2 (Mdm-2) protein (20). p53 is tightly regulated by a negative feedback loop where the tumor suppressor induces Mdm-2 transcription, which in turn binds to p53 and inhibits its function (21) (22). The Mdm-2 protein has been shown to inhibit p53 activity by binding to its transactivation domain, targeting it to ubiquitination, transporting it to the cytoplasm, and promoting its degradation by the proteasome. The precise mechanism(s) involved in p53 stabilization with response to stress remains unclear. It has been shown that it involves a series of post-translational modifications to both p53 and Mdm-2, which in turn may facilitate the dissociation of the Mdm-2/p53 complex (23).
We have previously demonstrated that UDCA decreases E2F-1 transcriptional activation, thus preventing the downstream events of transforming growth factor-␤1-induced cell death associated with Mdm-2 degradation and p53 stabilization (6). Moreover, microarray analysis revealed that UDCA incubation increased an expressed sequence tag (EST) highly similar to Mdm-2 protein while decreasing another EST for p53-apoptosis-associated target (24) in primary rat hepatocytes. Finally, UDCA was shown to alter Mdm-2 protein levels in several cell types (6,25). Here we further explore the molecular events underlying the cytopro-tective role of UDCA in p53-induced apoptosis of hepatocytes.
Our results indicate that UDCA inhibits p53 transactivation and its DNA binding activity in hepatocytes by preventing nuclear accumulation of this tumor suppressor protein, in part, through a p53/Mdm-2 binding-dependent mechanism.

EXPERIMENTAL PROCEDURES
Hepatocyte Isolation and Cell Culture-Rat primary hepatocytes were isolated from male Sprague-Dawley rats (100 -150 g) by collagenase perfusion as described previously (26). In brief, rats were anesthetized with phenobarbital, and the livers were perfused with 0.05% collagenase. Hepatocyte suspensions were obtained by passing collagenase-digested livers though 125-m gauze and washing cells in William's E medium supplemented with 26 mM sodium bicarbonate, 23 mM HEPES, 0.01 units/ml insulin, 2 mM L-glutamine, 10 nM dexamethasone, 100 units/ml penicillin, 100 units/ml streptomycin (Sigma-Aldrich), and 20% heat-inactivated fetal bovine serum (Invitrogen). Cell viability was determined by trypan blue exclusion and was typically 80 -85%. After isolation, hepatocytes were resuspended in William's E medium and plated on Primaria TM tissue culture dishes (BD Biosciences) at either 5 ϫ 10 4 cells/cm 2 for cell morphology assays or 2.5 ϫ 10 5 cells/cm 2 for all other experiments. The cells were maintained at 37°C in a humidified atmosphere of 5% CO 2 for 4 h to allow attachment. Plates were then washed with medium to remove dead cells and incubated in William's E medium containing 10% heat-inactivated fetal bovine serum.
Transfections and Chloramphenicol Acetyltransferase (CAT) Assays-Transfections were performed using two expression plasmids, pCMV-p53 wt and pCMV-p53 179 , and a bax promoter-driven CAT reporter construct. The bax-CAT construct consisted of a 371-base pair SmaI/SacI fragment of human bax gene subcloned into the HindIII site of the promoterless CAT plasmid, pUCSV0CAT (15). Overexpression plasmids were generated by cloning either wild-type p53 (pCMV-p53 wt ) or a mutant inactive form of human p53 (pCMV-p53 179 (H179E)), all under cytomegalovirus enhancer/promoter control (27). Twelve hours after plating, hepatocytes were treated with vehicle or 100 M of UDCA and co-transfected with 2 g of both expression and reporter plasmids using Lipofectamine 2000 (Invitrogen). For normalization, cells were cotransfected with 0.5 g of the luciferase reporter construct, PGL3-Control vector (Promega Corp., Madison, WI). Transfection efficiencies of ϳ70% were determined in primary rat hepatocytes using a reporter plasmid expressing ␤-galactosidase and did not differ between reporter and expression plasmids. At 48 h post-transfection, attached cells were harvested for CAT enzyme-linked immunosorbent assay (Roche Applied Science) and luciferase assays (Promega), according to the manufacturers' instructions. In parallel experiments, hepatocytes were pretreated with vehicle or 100 M UDCA 12 h before transfection with 8 g of each expression plasmid. Cells overexpressing either wild-type or mutant p53 were harvested for protein extraction and immunoblot analysis. Nuclear, mitochondrial, and cytosolic protein fractions or total proteins were prepared at 36 or 60 h after transfection. Attached cells were fixed for morphologic evaluation of apoptosis and culture medium used for lactate dehydrogenase viability assays.
Short Interference-mediated Silencing of the mdm-2 Gene-A pool of 4 short interference RNA (siRNA) nucleotides designed to knock down mdm-2 gene expression in rats was purchased from Dharmacon (Waltham, MA). A control siRNA containing a scrambled sequence that does not lead to the specific degradation of any known cellular mRNA was used as control. Four hours after p53 overexpression, the culture medium was changed, and hepatocytes were transfected using INTERFERin TM Transfection reagent for siRNA (Polyplus-transfection, Illkirch, France) according to the manufacturer's instructions for additional 48 h. The final concentration of siRNAs was 10 nM. UDCA was re-added to the cultures after the first 4 h of silencing. Floating and attached cells were harvested for preparation of total, nuclear, and cytosolic protein extracts, which were then subjected to immunoblot analysis. Attached cells were fixed for Hoechst staining.
Measurement of Cell Death and Caspase Activity-Cell viability was measured by the lactate dehydrogenase (Sigma-Aldrich) viability assay according to the manufacturer's instructions. In addition, Hoechst labeling of cells was used to detect apoptotic nuclei. Briefly, the medium was gently removed to prevent detachment of cells. Attached hepatocytes were fixed with 4% paraformaldehyde in phosphate-buffered saline (PBS), pH 7.4, for 10 min at room temperature, incubated with Hoechst dye 33258 (Sigma) at 5 g/ml in PBS for 5 min, washed with PBS, and mounted using PBS:glycerol (3:1, v/v). Fluorescent nuclei were scored blindly and categorized according to the condensation and staining characteristics of chromatin. Normal nuclei showed non-condensed chromatin dispersed over the entire nucleus. Apoptotic nuclei were identified by condensed chromatin, contiguous to the nuclear membrane, as well as nuclear fragmentation of condensed chromatin. Five random microscopic fields per sample of ϳ100 nuclei were counted, and mean values were expressed as the percentage of apoptotic nuclei. In addition, caspase activity was determined in cytosolic protein extracts after harvesting and homogenization of cells in isolation buffer containing 10 mM Tris-HCl buffer, pH 7.6, 5 mM MgCl 2 , 1.5 mM potassium acetate, 2 mM dithiothreitol, and protease inhibitor mixture tablets (Complete; Roche Applied Science). General caspase-3-like activity was determined by enzymatic cleavage of chromophore p-nitroanilide from the substrate N-acetyl-Asp-Glu-Val-Asp-p-nitroanilide (Sigma-Aldrich). The proteolytic reaction was carried out in isolation buffer containing 50 g of cytosolic protein and 50 M DEVDp-nitroanilide. The reaction mixtures were incubated at 37°C for 1 h, and the formation of p-nitroanilide was measured at 405 nm using a 96-well plate reader. In addition, caspase-3 cleavage was determined by immunoblotting.
p53-DNA Binding Enzyme-linked Immunosorbent Assay-The TransAM TM p53 transcription factor assay kit (Active Motif, Carlsbad, CA) was used according to the manufacturer's protocol. Nuclear extracts were diluted to 2 g/ml total protein with lysis buffer. Extracts were applied to plates containing immobilized 20-mer oligonucleotide with a p53 consensus binding site (5Ј-GGACATGCCCGGGCATGTCC-3Ј). After 1 h of incubation at room temperature, plates were washed and incubated with diluted p53 antibody (1:1000) for an additional 1 h. Diluted anti-rabbit horseradish peroxidase-conjugated antibody (1:1000) was then added to previously washed plates, and developing solution was added and incubated for 5-8 min to allow color development. The reaction was stopped, and absorbance was read at 450 nm with a reference wavelength of 650 nm. In addition, both nuclear and cytosolic protein fractions were analyzed for p53 by immunoblotting.
Bax Translocation and Cytochrome c Release-Cellular distribution of Bax and cytochrome c was determined using mitochondrial and cytosolic protein extracts. Cells were harvested and centrifuged at 600 ϫ g for 5 min at 4°C. The pellets were washed once in ice-cold PBS and resuspended with 3 volumes of isolation buffer containing 20 mM HEPES/KOH, pH 7.5, 10 mM KCl, 1.5 mM MgCl 2 , 1 mM EDTA, 1 mM EGTA, and 1 mM dithiothreitol supplemented with protease inhibitor mixture tablets in 250 mM sucrose. After chilling on ice for 15 min, cells were disrupted by 40 strokes of a glass homogenizer, and homogenates were centrifuged twice at 2500 ϫ g for 10 min at 4°C to remove unbroken cells and nuclei. The mitochondrial fraction was then centrifuged at 12,000 ϫ g for 30 min at 4°C, and the pellet was resuspended in isolation buffer and frozen at Ϫ80°C. For cytosolic proteins, the 12,000 ϫ g supernatants were removed, filtered sequentially through 0.2-and 0.1-m Ultrafree MC filters (Millipore, Bedford, MA) to remove other cellular organelles, and frozen at Ϫ80°C.
Immunoblotting-For Bax and cytochrome c detection, 40 g of mitochondrial and cytosolic proteins were separated by 14% SDS-PAGE. In addition, steady-state levels of Bax and p53 as well as p53 cellular distribution and caspase-3 cleavage were also determined. After electrophoretic transfer onto nitrocellulose membranes, the immunoblots were incubated with 15% H 2 O 2 for 15 min at room temperature. After blocking with a 5% milk solution, the membranes were incubated overnight at 4°C with primary mouse monoclonal antibodies reactive to Bax (B-9), p53 (Pab 240), p-p53 (mSer 20), caspase-3 (H-227), cytochrome c oxidase subunit II (K-20), lamin A/C (346) (Santa Cruz Biotechnology, Santa Cruz, CA), p-p53 (phospho-Ser-15, Ab-3; Calbiochem), and cytochrome c (7H8.2C12; Pharmingen) and finally with secondary goat anti-mouse or anti-rabbit IgG antibody conjugated with horseradish peroxidase (Bio-Rad) for 3 h at room temperature. The membranes were processed for protein detection using the SuperSignal substrate (Pierce). ␤-Actin (AC-15, Sigma-Aldrich), lamin A/C, and cytochrome c oxidase were used as loading controls for total, nuclear, and mitochondrial proteins, respectively. Protein concentrations were determined using the Bio-Rad protein assay kit according to the manufacturer's specifications.
Immunoprecipitation-Binding of p53 to Mdm-2 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 a primary mouse monoclonal antibody to Mdm-2 (SMP 14; Santa Cruz Biotechnology) and the Ezview Red Protein G Affinity Gel (Sigma-Aldrich). Typically, 200 g of lysate were incubated with 1 g of Mdm-2 antibody overnight at 4°C. Immunoblots were then probed with the mouse monoclonal anti-p53 antibody. Mdm-2 expression was determined in the same membrane after stripping off the immune complex for the detection of p53. Immunoprecipitation assays using high detergent conditions as well as immunoblot analysis showed an absence of nonspecific binding of the Mdm-2 antibody to p53. In addition, immunoprecipitation assays using the mouse monoclonal antibody reactive to ␤-actin demonstrated no association with either p53 or Mdm-2.
Immunofluorescence-Primary rat hepatocytes overexpressing wild-type p53 in the presence or absence of UDCA were fixed with 4% paraformaldehyde in PBS, pH 7.4, for 30 min. After fixation, cells were blocked for 1 h in PBS containing 0.1% Triton-X-100, 1% fetal bovine serum, and 10% normal donkey serum. Cells were then sequentially incubated with monoclonal antibody to p53 (Santa Cruz Biotechnology) at a dilution of 1:200 in blocking solution, overnight at 4°C. Subsequently, after three washes with PBS, cells were incubated with aminomethyl coumarin (AMCA)-conjugated donkey anti-mouse secondary antibody (Jackson ImmunoResearch Laboratories, Inc.) for 2 h along with fluorescein isothiocyanate-tagged UDCA. Subcellular localization of p53 and UDCA was visualized using an Axioskop fluorescence microscope (Carl Zeiss, Jena, Germany).
Densitometry and Statistical Analysis-The relative intensities of protein bands were analyzed using the Quantity One Version 4.6 densitometric analysis program (Bio-Rad) and normalized to the respective loading controls. Statistical analysis was performed using GraphPad InStat Version 3.00 (GraphPad Software, San Diego, CA) for the analysis of variance and Bonferroni's multiple comparison tests. Values of p Ͻ 0.05 were considered significant.

UDCA Reduces p53
Transactivation-In addition to its role in cell cycle progression and senescence, p53 also induces apoptosis. Furthermore, we have previously shown that UDCA prevented transforming growth factor-␤1-induced p53 stabilization and apoptosis in primary rat hepatocytes (6). We now hypothesize that p53 may represent an important target in bile acid-induced modulation of apoptosis. Immunoblot analysis confirmed that total p53 expression was increased in cells transfected with pCMV-p53 wt and pCMV-p53 179 overexpressing plasmids as compared with endogenous levels of pCMV-transfected cells (Fig. 1A). Our results showed significant levels of apoptosis in cultured primary rat hepatocytes after transfection with wild-type p53 overexpression plasmid, as assessed by changes in nuclear morphology (Fig. 1B), caspase activity, and processing ( Fig.  1C) (p Ͻ 0.01). Mutant p53 only slightly increased cell death (data not shown), implying a transcription-independent role of p53 in apoptosis. Notably, pretreatment with UDCA reduced the levels of apoptosis to those of controls transfected with mutant p53 (p Ͻ 0.01) while inhibiting caspase-3-like activity and caspase-3 processing by 90% (p Ͻ 0.05). In addition, camptothecin, a topoisomerase I inhibitor known to induce p53 accumulation, resulted in ϳ25% hepatocyte apoptosis, and this was inhibited with UDCA by ϳ80% (p Ͻ 0.01). Similar results were obtained with the lactate dehydrogenase viability assay (data not shown).
It is well established that p53 binds to specific sequences within the promoter of various genes to modulate their expres-sion. Proapoptotic Bax was the first member of the Bcl-2 gene family shown to be induced by p53 (15,28). Therefore, primary rat hepatocytes were cotransfected with a reporter gene construct containing the bax gene promoter to drive transcription of CAT in combination with wild-type or mutant p53 overexpression plasmids. At 48 h after transfection cell death was associated with a strong transactivation of the bax gene promoter by Ͼ5-fold compared with cells overexpressing mutant p53 (p Ͻ 0.01) (Fig. 2A). Notably, when hepatocytes were incubated with UDCA, p53-induced bax transcription was reduced by ϳ50% (p Ͻ 0.05) (Fig. 2B). Taurine-and glycine-conjugated derivatives of UDCA were similarly protective. In contrast, treatment with hydrophobic bile acids, such as deoxycholic, lithocholic, and chenodeoxycholic acids, resulted in increased bax transcriptional activation up to 25-fold (p Ͻ 0.05) (supplemental data). This may explain the toxicity and increased levels of apoptosis seen in hepatocytes incubated with these bile acids (data not shown). Thus, UDCA is a specific, strong repressor of p53-driven bax transcription.
UDCA Prevents p53-induced Bax Expression and Translocation-It has been demonstrated that UDCA plays an important role in preventing apoptosis through the mitochondrial pathway (3)(4)(5). To further characterize the mechanism by which UDCA modulates p53-mediated apoptosis, we analyzed the expression and subcellular distribution of its downstream target, Bax. Total protein extracts were prepared from primary rat hepatocytes overexpressing either wild-type or mutant p53 in the presence or absence of the bile acid. In addition, at 60 h post-transfection, hepatocytes overexpressing wild-type p53 showed an almost 50% increase in Bax protein compared with control cells expressing mutant p53 (p Ͻ 0.05) (Fig. 3A). Notably, the levels of total p53 remained unaltered. In contrast, pretreatment with UDCA completely abolished p53-driven Bax expression (p Ͻ 0.01). As expected, cells transfected with wildtype p53 showed significantly increased Bax translocation from the cytosol to mitochondria, which in turn was significantly prevented by UDCA (p Ͻ 0.05) (Fig. 3B). The release of cytochrome c into the cytosol was consistent with the observed changes in Bax. Thus, these results suggest that UDCA prevents hepatocyte apoptosis associated with wild-type p53 overexpression by abrogating p53 transcriptional activity. This then leads to reduced total levels of Bax in the cell and decreased Bax translocation to the mitochondria.
UDCA Inhibits DNA Binding Activity of p53-Activation of p53 involves stabilization of the protein and enhancement of its DNA binding activity (29). When stabilized, p53 accumulates in the nucleus and regulates the expression of numerous proapoptotic genes by binding to their promoter sequences (30). To further investigate the mechanism by which UDCA inhibits p53 transactivation, we evaluated DNA binding activity of p53 in hepatocytes with or without UDCA. Nuclear extracts were prepared 48 h after transfection with p53 overexpression plas-mids, and DNA-bound p53 was assayed in the nuclear lysates. As shown in Fig. 4, levels of DNA-bound p53 were 6-fold increased in cells overexpressing wild-type p53 relative to the mutant form (p Ͻ 0.01). Notably, when cells were pretreated with UDCA, there was a reduction of ϳ50% in DNA binding activity of p53 (p Ͻ 0.05). The presence of the bile acid in cells with mutant p53 did not alter the transcription factor binding activity. This suggests that UDCA-mediated decrease in p53-DNA binding may result from either a direct interference or indirect inhibition of p53 stabilization and/or accumulation in the nucleus. To address this question we prepared nuclear protein extracts from both wild-type and mutant-transfected cells, incubated the extracts with UDCA, and analyzed p53-DNA binding activity. Curiously, the results showed that the bile acid was unable to prevent p53-DNA binding by directly interfering with nuclear proteins and/or DNA (Fig. 4). In addition, preliminary circular dichroism results suggest that UDCA does not interact directly with the p53 core DNA-binding domain (data not shown). However, it is possible that UDCA requires cooperation with other cellular factors to reduce nuclear p53 stability, thus compromising its transcriptional activity. . UDCA prevents p53 transcriptional activation in primary rat hepatocytes. Cells were cotransfected with a bax promoter-driven CAT construct and a luciferase control plasmid in combination with either mutant p53 (pCMV-p53 179 ) or wild-type p53 (pCMV-p53 wt ) overexpression plasmids. Vehicle or 100 M UDCA was included in the culture medium at the time of transfections. At the indicated time points after transfection, cells were harvested for the CAT enzyme-linked immunosorbent assay and luciferase assays as described under "Experimental Procedures." A, bax promoter activity in cells overexpressing either mutant or wild-type p53 for the indicated times. B, bax promoter activity in cells overexpressing either mutant or wildtype p53 Ϯ UDCA for 48 h. CAT activity (absorbance/mg) was normalized to control luciferase expression, and the results are expressed as the mean Ϯ S.E. arbitrary units for at least five different experiments. *, p Ͻ 0.01 from control; †, p Ͻ 0.05 from wild-type p53 alone. Mut, mutant; Wt, wild-type. UDCA Represses p53 Transcriptional Activity via Increased Binding to Mdm-2-The p53 inhibitor Mdm-2 is a key regulator of p53 abundance and activity. It inhibits the transcriptional activity of p53 and, more importantly, promotes its shuttling to the cytoplasm and subsequent degradation by the proteasome (21,22). Moreover, we have previously reported that UDCA prevented the decrease of Mdm-2 protein levels associated with transforming growth factor-␤1-induced cell death (6).
To investigate whether Mdm-2 was an important regulatory factor in the antiapoptotic function of UDCA, we performed posttranscriptional mdm-2 gene silencing experiments. As a control for the specificity of siRNA, cells were transfected with a nonspecific pool of siRNAs. Western blot analysis confirmed that Mdm-2 expression decreased by ϳ70% after transfection with specific Mdm-2 siRNAs (Fig. 5A). Notably, Mdm-2 silencing reduced UDCA protection against p53-induced nuclear fragmentation by ϳ60% (p Ͻ 0.01) (Fig. 5B). Caspase activity was similarly reduced (data not shown). Thus, it appears that Mdm-2 plays a key role during UDCA-mediated modulation of p53-induced apoptosis.
Next, we investigated whether UDCA-mediated p53 inhibition was associated with an increase of p53 repression by Mdm-2. Complex formation of p53 and its inhibitor Mdm-2 was analyzed by immunoprecipitation assays after transfection with both wild-type and mutant p53 plasmids in the presence or absence of the bile acid (Fig. 6A). Overexpression of wild-type p53 induced a 2-fold increase of p53/Mdm-2 dissociation compared with controls (p Ͻ 0.01). Notably, preincubation of hepatocytes with UDCA completely abolished this increase (p Ͻ 0.05), indicating that UDCA protection is associated with a marked Mdm-2-dependent reduction of free p53 nuclear levels. In fact, both immunofluorescence and immunoblot analysis revealed marked differences in the subcellular distribution of p53 between cells overexpressing wild-type p53 in the presence or absence of UDCA (Fig. 6, B and C). Untreated hepatocytes exhibited a predominant nuclear staining of p53. In contrast, when cells were preincubated with the bile acid, p53 appeared . UDCA inhibits p53-DNA binding activity. Cells were transfected with either mutant (Mut) p53 (pCMV-p53 179 ) or wild-type (Wt) p53 (pCMV-p53 wt ) overexpression plasmids. Vehicle or 100 M UDCA was included in the culture medium at 12 h before transfection. At 36 h after transfection nuclear protein extracts were prepared as described under "Experimental Procedures." The level of p53 present in nuclear lysates that can bind to its DNA consensus recognition sequence was determined by the TransAM TM p53 enzyme-linked immunosorbent assay and expressed as -fold change relative to the control. In parallel experiments nuclear protein extracts were prepared from both wild-type and mutant-transfected cells, incubated with UDCA, and analyzed as p53-DNA binding activity. The results are expressed as the mean Ϯ S.E. for at least three different experiments. *, p Ͻ 0.01 from control; †, p Ͻ 0.05 from wild-type p53 alone. UDCAbb, UDCA added in the binding buffer. localized primarily in the cytoplasm, confirming that UDCAinduced p53/Mdm-2 complex formation results in increased p53 nuclear export. In both cases, a punctuate pattern of p53 was detected in the cytosol, which may indicate that p53 is also localized to mitochondria. The bile acid appeared diffusely in the cytosol and nucleus of hepatocytes. Moreover, we analyzed the phosphorylation status of two serine residues of the N-terminal domain of p53, Ser-15 and Ser-20. Both are thought to be involved in p53 apoptotic activity and its dissociation from Mdm-2 (31,32). Expression of p53-phosphorylated Ser-20 was low or undetectable (data not shown). However, a significant increase in phosphorylation at Ser-15 was detected in hepatocytes overexpressing wild-type p53 (Fig. 6C). This increase was greatly inhibited in hepatocytes pretreated with UDCA, reinforcing the notion that this bile acid affects binding of p53 and Mdm-2. Finally, silencing of Mdm-2 resulted in increased accumulation of p53 in the nucleus even in the presence of UDCA (Fig. 6D), thus confirming the crucial role of Mdm-2 in the bile acid-protective effect. Taken together, these results strongly suggest that UDCA inhibits p53 transcriptional activation in part by inducing Mdm-2/p53 association, which in turn results in increased p53 degradation.

DISCUSSION
The precise molecular mechanisms by which UDCA modulates cell survival and apoptosis are still a matter of debate. We have previously reported that UDCA is a pleiotropic agent that prevents apoptosis in primary rat hepatocytes by inhibiting the mitochondrial pathway (4,5) and interfering with the E2F1/ Mdm-2/p53 apoptotic cascade (6). The results presented here suggest that UDCA modulates p53-induced cell death by altering p53 transactivation and DNA binding activity and preventing its accumulation in the nucleus.
The tumor suppressor p53 is a transcription factor that coordinates a complex network of cellular proteins. In response to diverse stress factors, p53 is activated and induces the expression of different subsets of genes leading to cell cycle arrest, DNA repair, senescence, or apoptosis. However, the mechanism(s) by which p53 mediates the apoptotic process is still a matter of intense study. Numerous studies have recently described the importance of p53 transcriptional regulation in both the intrinsic and extrinsic pathways of apoptosis (33). In fact, several target gene products have pivotal roles in p53-dependent apoptosis. Proapoptotic Bax is one example of a direct and evolutionary conserved transcriptional target of p53 (15,28). During apoptosis, cytosolic Bax is translocated to the outer mitochondrial membrane where it induces cytochrome c release FIGURE 6. UDCA induces Mdm-2/p53 association and p53 nuclear export. Cells were transfected with either mutant (Mut) p53 (pCMV-p53 179 ) or wildtype (Wt) p53 (pCMV-p53 wt ) overexpression plasmids. Vehicle or 100 M UDCA were included in the culture medium at 12 h before transfection. At 36 h after transfection total proteins were extracted for immunoprecipitation (IP), and nuclear and cytosolic protein fractions were extracted for Western blot (WB) analysis. In addition, cells were labeled with a fluorescent UDCA molecule (fluorescein isothiocyanate (FITC)-UDCA) and fixed for immunofluorescence analysis as described under "Experimental Procedures." A, immunoprecipitation analysis of Mdm-2/p53 dissociation. Shown are representative immunoblots with p53-and Mdm-2-specific antibodies (top) and a histogram of Mdm-2/p53 dissociation (bottom). All densitometry values for p53 were normalized to respective Mdm-2 expression, and the results are expressed as the mean Ϯ S.E. arbitrary units for at least three different experiments. *, p Ͻ 0.01 from control; †, p Ͻ 0.05 from cells overexpressing wild-type p53 alone.
B, subcellular localization of p53 in fluorescein isothiocyanate-UDCA-labeled primary rat hepatocytes. Fluorescent microscopy of p53 staining in cells transfected with wild-type p53 and treated with either vehicle (a-c) or UDCA (d-f). Scale bar, 10 m. C, p53 translocation from the nucleus to the cytosol (Cyto) and phosphorylation status of phospho (p)-Ser-15-p53. Representative immunoblots of nuclear and cytosolic p53 are shown. Blots were normalized to endogenous lamin or ␤-actin. D, effect of Mdm-2 silencing on nuclear p53. Cells were transfected for 48 h with either control or mdm-2 siRNA 4 h after the initial transfection with p53 overexpression plasmids. Representative immunoblot of nuclear p53 is shown. The blot was normalized to endogenous lamin. (34,35). Once in the cytoplasm, cytochrome c functions as a cofactor with apoptosis protease-activating factor 1 (Apaf-1) to promote the cleavage of procaspase-9, initiating apoptosis (36). Recently, it has been demonstrated that after hypoxia or DNA damage, a small fraction of p53 translocates to mitochondria, where it interacts with Bak, Bcl-xL , or Bcl-2 and antagonizes their antiapoptotic function (37,38). Cytosolic p53 was also shown to induce apoptosis through direct activation of Bax (39). Nevertheless, most data suggest that the activities of p53 are due to its sequence-specific transcriptional activation.
Our data show that UDCA inhibits p53-induced apoptosis of primary rat hepatocytes. In fact, wild-type p53 overexpression resulted in high levels of apoptosis, which were associated with strong transactivation of the bax gene promoter, Bax mitochondrial translocation, cytochrome c release, and caspase-3 activation. Notably, preincubation with UDCA abrogated all apoptotic changes. We have previously shown that UDCA prevented the increase of Bax protein levels induced by p53 (6).
Here we demonstrate that UDCA preincubation leads to a significant decrease of p53-driven bax promoter activation, thus, via modulation of p53 transcriptional activity. Moreover, the presence of UDCA in hepatocytes overexpressing wild-type p53 resulted in down-regulation of Bax but not in a reduction of p53 total protein levels, reinforcing the notion that the transcription activity of p53 is impaired in cells pretreated with UDCA.
It has been reported that bile acids are detectable within the nuclei of rat hepatocytes (40 -42) where they may play an important role in controlling gene expression. More recently, nuclear translocation of UDCA mediated by nuclear steroid receptors was shown to be essential for its antiapoptotic properties (43). Once in the nucleus, UDCA might regulate gene expression by interacting directly with chromatin. Interestingly, 32 P-labeling analysis of DNA-reactive bile acids provided evidence of DNA-adduct formation (44). Conversely, other authors have reported that bile acids do not bind covalently to DNA (45). It is also possible that UDCA interacts with nuclear proteins, such as transcription factors, thereby indirectly modulating the expression of several genes. In this regard it has recently been shown that UDCA suppresses DCA-induced DNA binding activity of activator protein-1 in a human colon cancer cell line (46). In this report we show that UDCA reduces the ability of p53 to bind DNA through at least two possible mechanisms. First, UDCA may interact directly with p53, preventing an effective p53-DNA binding, or second, UDCA induces p53 destabilization in the nucleus, which may involve second messenger signals. Although direct interaction between p53 and the UDCA molecule remains to be proven, UDCA appears to inhibit the ability of p53 to bind DNA through indirect mechanism(s). Moreover, UDCA was previously shown to prevent transforming growth factor-␤1-induced p53 stabilization and Mdm-2 degradation. Therefore, it seems more plausible that UDCA affects p53 nuclear stabilization possibly via its repressor Mdm-2.
Mdm-2 regulates p53 through negative feedback loops that include inhibition of the transcriptional activity of p53 and its degradation via the ubiquitin-proteasome pathway. Mdm-2 also plays a role in regulating the subcellular localization of the tumor suppressor protein. The ubiquitin ligase activity of Mdm-2 contributes to the efficient nuclear export of p53 (47,48), which depends on the nuclear export sequence identified in the C terminus of p53 (49). Here we report that UDCA increases the association between p53 and its repressor Mdm-2. In addition, UDCA induces p53 nuclear export to cytoplasm, which is strongly correlated with Mdm-2 levels. All together, our results clearly show that inhibition of p53 by its direct repressor Mdm-2 is an essential step of UDCA regulation. By inducing Mdm-2/p53 complex formation, UDCA reduces p53 activity by simultaneously blocking its transactivation domain and enhancing its export to cytosol. However, the precise mechanism by which UDCA induces Mdm-2/p53 binding remains to be determined.
The binding of p53-derived peptides to Mdm-2 results in extensive conformational changes along the N-terminal domain of the repressor (50). Furthermore, various peptidic and nonpeptidic agents have been reported to bind to the Mdm-2/p53 transactivation domain binding cleft (51)(52)(53). Although the majority of these molecular agents are inhibitors of Mdm-2/p53 interaction, it is plausible that other ligands favor this interaction. In fact, molecular dynamics simulation studies revealed that the p53 binding cleft of Mdm-2 is highly flexible and adaptable to differential binding of ligands that could potentially induce global conformational changes and biological function (54). UDCA was already shown to interact with several proteins, namely the glucocorticoid receptor (GR) (43,55) or the p65 subunit of NF-B (56). Interestingly, we have recently shown that GR is required for UDCA anti-apoptotic function by facilitating its translocation into the nucleus (7). Furthermore, GR and p53 were found to repress each other's function either at the transcriptional level or through GR-mediated cytoplasmic anchoring of p53 (57). In fact, GR itself could be cofactor in the modulation of p53/Mdm-2 binding by UDCA. Other nuclear steroid receptors are also regulated by UDCA. This bile acid is a relatively strong agonist of the pregnane X receptor and a weaker agonist of the farnesoid X receptor. Although UDCA does not itself bind farnesoid X receptor, it does inhibit receptor activation by more hydrophobic bile acid species, such as chenodeoxycholic, deoxycholic, and lithocholic acids, which may also contribute to its protective effects (58). Finally, the taurine-conjugated derivative of UDCA has been described as a chemical chaperone that reduces endoplasmic reticulum stress and restores glucose homeostasis in a mouse model of type 2 diabetes (59). Therefore, it is possible that UDCA interacts with the Mdm-2/p53 transactivation domain-binding cleft, thus stabilizing the complex.
Alternatively, UDCA may stimulate Mdm-2/p53 binding via the phosphoinositide 3Ј-OH kinase survival pathway. Phosphoinositide 3Ј-OH kinase in turn activates the serine/threonine kinase Akt, which then phosphorylates a range of targets that function to promote cell survival, including Mdm-2 (60). Phosphorylation of Mdm-2 by Akt results in activation and nuclear accumulation of the Mdm-2 protein and consequent destabilization of p53, which may serve to protect cells from p53-induced apoptosis (61). Interestingly, UDCA has previously been suggested to protect from apoptosis in several cell types by stimulating phosphoinositide 3Ј-OH kinase/Akt-dependent survival signaling (62)(63)(64).
In conclusion, our studies show that Mdm-2/p53/Bax apoptotic pathway is a prime target of UDCA modulation. Furthermore, bile acid inhibition of p53-induced apoptosis involves Mdm-2-dependent shuttling of p53 to the cytoplasm and decreased p53 transcriptional and DNA binding activities. It is of paramount importance to investigate the precise mechanism by which UDCA interacts with p53 and/or Mdm-2 in an effort to develop novel effective therapeutics for apoptosis-related liver diseases.