Ursodeoxycholic acid modulates E2F-1 and p53 expression through a caspase-independent mechanism in transforming growth factor beta1-induced apoptosis of rat hepatocytes.

Transforming growth factor beta1 (TGF-beta1)-induced hepatocyte apoptosis is associated with activation of E2F transcription factors and p53 stabilization through Mdm-2, thus potentially modulating a number of target genes. In previous studies, we have shown that ursodeoxycholic acid (UDCA) prevents TGF-beta1-induced hepatocyte apoptosis by inhibiting the mitochondrial pathway of cell death. In this study we examined the role of p53 in the induction of apoptosis by TGF-beta1, and identified additional antiapoptosis targets for UDCA. Our data show a significant transcriptional activation of E2F-1 in primary rat hepatocytes incubated with TGF-beta1, as well as a 5-fold increase in p53 and a 2-fold decrease in its inhibitor, Mdm-2 (p < 0.05). In addition, bax mRNA expression was significantly induced at 36 h (p < 0.01), resulting in increased levels of Bax protein. In contrast, Bcl-2 transcript and protein levels were decreased at all time points (p < 0.01). Notably, UDCA inhibited E2F-1 transcriptional activation, p53 stabilization and Bcl-2 family expression (p < 0.05), in part, through a caspase-independent mechanism. Moreover, in the absence of TGF-beta1, UDCA prevented induction of p53 and Bax by overexpression of E2F-1 and p53, respectively (p < 0.05). In addition, UDCA inhibited TGF-beta1-induced degradation of nuclear factor kappaB (NF-kappaB) and its inhibitor IkappaB (p < 0.05). In conclusion, these results demonstrate that UDCA inhibits E2F-1 transcriptional activation of hepatocyte apoptosis, thus modulating p53 stabilization, NF-kappaB degradation, and expression of Bcl-2 family members.

Transforming growth factor ␤1 (TGF-␤1)-induced hepatocyte apoptosis is associated with activation of E2F transcription factors and p53 stabilization through Mdm-2, thus potentially modulating a number of target genes. In previous studies, we have shown that ursodeoxycholic acid (UDCA) prevents TGF-␤1-induced hepatocyte apoptosis by inhibiting the mitochondrial pathway of cell death. In this study we examined the role of p53 in the induction of apoptosis by TGF-␤1, and identified additional antiapoptosis targets for UDCA. Our data show a significant transcriptional activation of E2F-1 in primary rat hepatocytes incubated with TGF-␤1, as well as a 5-fold increase in p53 and a 2-fold decrease in its inhibitor, Mdm-2 (p < 0.05). In addition, bax mRNA expression was significantly induced at 36 h (p < 0.01), resulting in increased levels of Bax protein. In contrast, Bcl-2 transcript and protein levels were decreased at all time points (p < 0.01). Notably, UDCA inhibited E2F-1 transcriptional activation, p53 stabilization and Bcl-2 family expression (p < 0.05), in part, through a caspase-independent mechanism. Moreover, in the absence of TGF-␤1, UDCA prevented induction of p53 and Bax by overexpression of E2F-1 and p53, respectively (p < 0.05). In addition, UDCA inhibited TGF-␤1induced degradation of nuclear factor B (NF-B) and its inhibitor IB (p < 0.05). In conclusion, these results demonstrate that UDCA inhibits E2F-1 transcriptional activation of hepatocyte apoptosis, thus modulating p53 stabilization, NF-B degradation, and expression of Bcl-2 family members.
Ursodeoxycholic acid (UDCA) 1 is widely used in the treatment of certain cholestatic disorders of the liver. In particular, it has significantly improved both the clinical and biochemical indices of patients with primary biliary cirrhosis (1,2). The therapeutic effects of UDCA have been attributed to three major mechanisms of action, including protection of cholangiocytes against the cytotoxicity of hydrophobic bile acids, stimulation of hepatobiliary secretion, and inhibition of liver cell apoptosis. It has been recently demonstrated that this hydrophilic bile acid plays a unique role in modulating the apoptotic threshold in both hepatic and nonhepatic cells by preventing apoptosis through the mitochondrial pathways (3Ϫ5). Nevertheless, the precise mechanism(s) by which UDCA prevents cell death remains speculative, and may involve molecular targets other than mitochondria.
Transforming growth factor ␤1 (TGF-␤1) is a multifunctional cytokine, whose activities include cell cycle control, regulation of early development, differentiation, and induction of apoptosis, among others (6,7). It has been shown to induce rapid growth arrest and apoptosis in hepatic cells by suppressing phosphorylation of the retinoblastoma protein (pRb) as well as inhibiting its expression (8). In fact, TGF-␤1 may act through pRb to negatively regulate c-fos and c-myc gene expression (9). pRb, in turn, is extensively regulated by its phosphorylation state, which is modified by the activities of various cyclin and cyclin-dependent kinase complexes and by protein phosphatases (10). The hypophosphorylated pRb is a growth inhibitory form of the protein, while hyperphosphorylated pRb mediates progression through G 1 into S phase (11,12). Inactivation of pRb either by phosphorylation, mutation, degradation, or binding to an oncoprotein reduces its ability to sequester E2F-1, thereby increasing levels of free E2F-1 (13,14). E2F-1 is the best-characterized member of the E2F family of transcription factors that regulates the expression of genes involved in cell cycle, proliferation (15), and apoptosis (16). It is typically bound to unphosphorylated pRb (17). Under certain stress conditions and during the cell cycle, pRb is inactivated and/or degraded (18), thus releasing E2F-1 to transactivate its target genes. Interestingly, the dramatic increase in E2F-1 binding activity coupled with significantly decreased pRb levels in TGF-␤1treated cells suggests that the unique complexes identified in preapoptotic cells may also promote apoptosis via E2F-1 transcriptional activation (8).
Activation of E2F-1 induces cells to undergo apoptosis, and may occur through both p53-dependent and -independent mechanisms (19). These include stabilization of the tumor suppressor protein p53 via the transcription of p14 ARF , transcriptional activation of the p53 homologue p73, and inhibition of the anti-apoptotic signaling of nuclear factor B (NF-B). The ability of E2F-1 to induce apoptosis was thought to be a unique feature of this family member compared with other E2F transcription factors (20). However, it has recently been demonstrated that E2F-3 makes a major contribution to apoptotic pathways (21).
E2F-1 can also interfere with additional signaling pathways such as those mediated by NF-B (22), an inducible transcription factor that regulates expression of various proinflammatory and immune response genes (23Ϫ25). NF-B is typically a heterodimer of p65/RelA and p50 subunits, and is usually retained in the cytoplasm as an inactive form through association with the inhibitor protein IB. For NF-B activation, it is necessary that IB undergoes phosphorylation, ubiquitination, and subsequent degradation, thereby permitting the nuclear translocation of NF-B. NF-B binds to specific response elements of target genes, including proinflammatory cytokines and antiapoptotic proteins.
Although the role of p53 in suppressing cell cycle progression has been extensively described (26,27), much less is known about the mechanism by which p53 induces apoptosis. Nevertheless, p53 has been shown to regulate several apoptotic genes, including members of the Bcl-2 family, such as proapoptotic Bax or antiapoptotic Bcl-2 (28Ϫ30). During apoptosis, cytosolic Bax is translocated to the mitochondrial membrane where it induces cytochrome c release (31,32). This acts as a coactivator of the apoptosis protease-activating factor 1 (Apaf-1) in the cleavage of pro-caspase-9 and execution of programmed cell death (33). Unlike Bax, Bcl-2 and Bcl-x L exert antiapoptotic effects by heterodimerizing with Bax in the mitochondria, and preventing cytochrome c release from mitochondria into the cytosol (34,35). Thus, the cellular balance between pro-and antiapoptotic proteins of the Bcl-2 family can determine the ultimate fate of the cell. Finally, it has recently been reported that the gene Apaf-1 is a transcriptional target for both E2F and p53 (36).
We have previously shown that UDCA and its taurine-conjugated derivative (TUDCA) stabilize the mitochondrial membrane and prevent TGF-␤1-induced apoptosis in hepatocytes. They act by inhibiting mitochondrial membrane depolarization and channel formation, production of reactive oxygen species, release of cytochrome c, caspase activation, and cleavage of the nuclear enzyme poly(ADP-ribose) polymerase (4,5). Here, we further investigated the mechanism(s) by which UDCA exerts its antiapoptotic effect. Our results indicate that the bile acid specifically inhibits the E2F-1/p53 apoptotic pathway, in part, through a caspase-independent mechanism and reduces NF-B degradation induced by TGF-␤1, thus modulating the expression of Bcl-2 family elements.

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 (37). In brief, rats were anesthetized with phenobarbitol and the livers perfused with 0.05% collagenase. Hepatocyte suspensions were obtained by passing digested livers though 125 m gauze and washing cells in William's E medium (Invitrogen Corp., Grand Island, NY) supplemented with 26 mM sodium bicarbonate, 23 mM HEPES, 0.01 units/ml insulin, 2 mM L-glutamine, 10 nM dexamethasone, 5.5 mM glucose, 100 units/ml penicillin, 100 units/ml streptomycin, and 20% heat-inactivated (56°C for 30 min) fetal bovine serum (FBS; Atlanta Biologicals Inc., Norcross, GA). Cell viability was determined by trypan blue exclusion and was typically 85-90%. After isolation, hepatocytes were resuspended in complete William's E medium and plated on Primaria™ tissue culture dishes (BD Biosciences) at either 2.1 ϫ 10 4 cells/cm 2 for transfection assays, or 6.4 ϫ 10 4 cells/cm 2 for all other experiments. The cells were maintained at 37°C in a humidified atmosphere of 5% CO 2 for 6 h. Plates were then washed with medium to remove dead cells, and then incubated in William's E medium containing 10% heat-inactivated fetal bovine serum.
Induction of Apoptosis-Freshly isolated hepatocytes were cultured for 6 h as described above, washed and then incubated in 10% fetal bovine serum William's E medium supplemented with either 100 M UDCA or TUDCA (Sigma), or no addition (control) for 12 h. Cells were then exposed to 1 nM recombinant human TGF-␤1 (R & D Systems Inc., Minneapolis, MN) for 6, 12, 24, 36, or 48 h. In parallel studies, hepatocytes were incubated with 50 M of either caspase-3 (z-DEVD.fmk), caspase-9 (z-LEHD.fmk), or caspase-8 inhibitor (z-IETD.fmk) (EMD Biosciences, San Diego, CA) for 1 h prior to TGF-␤1 incubation. Attached and floating cells were combined and total, cytosolic and nuclear proteins extracted for immunoblotting and caspase activity assays; and total RNA for RT-PCR. Attached cells were fixed for Hoechst staining and immunocytochemical analysis.
Morphologic Evaluation of Apoptosis and Caspase Activation-Hoechst labeling of cells was used to detect apoptotic nuclei. In brief, the medium was gently removed at the indicated times 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 mg/ml in PBS for 5 min, washed with PBS and mounted using PBS:glycerol (3:1, v/v). Fluorescent nuclei were scored blindly by laboratory personnel 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. Three random microscopic fields per sample of ϳ250 nuclei were counted and mean values expressed as the percentage of apoptotic nuclei. In addition, caspase activation 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 KAc, 2 mM dithiothreitol, and protease inhibitor mixture tablets (Complete; Roche Applied Science, Mannheim, Germany). 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.). The proteolytic reaction was carried out in isolation buffer containing 50 g cytosolic protein and 50 M DEVD-pNA. The reaction mixtures were incubated at 37°C for 1 h, and the formation of pNA was measured at 405 nm using a 96-well plate reader.
Transfections and CAT Assays-Transfections were performed using reporter constructs E2F-1CAT and 4xE2FCAT, and four expression constructs, pCMVE2F-1, pCMVE2F-1⌬53, pCMVp53, and pCMVp53-(143Ala). E2F-1CAT consisted of the entire human E2F-1 promoter fused to the chloramphenicol acetyltransferase (CAT) gene (38); 4xE2FCAT was generated by insertion of a synthetic promoter containing four E2F consensus binding sites (39) upstream of the CAT reporter gene. Overexpression plasmids were generated by cloning either wildtype E2F-1 (pCMVE2F-1) and p53 (pCMVp53) or mutant E2F-1 (pCMVE2F-1⌬53) and p53 (pCMVp53(143Ala)), all under CMV enhancer/promoter control (40). Twelve hours after plating, hepatocytes at 40% confluence were transfected with 4 g of each plasmid using conjugated polyethylenimine (PEI) as previously described (41). To assess transfection efficiency, hepatocytes were cotransfected with the luciferase reporter construct, PGL3-Control vector (Promega Corp., Madison, WI). Based on this control, transfection efficiencies were ϳ70% and did not differ between wild-type and dominant negative plasmids. Twelve hours after E2F-1CAT or 4xE2FCAT transfection, vehicle or 100 M of either UDCA or TUDCA was added to cells. After an additional 12 h, 1 nM TGF-␤1 was included in the cultures. The cells were incubated with TGF-␤1 for 36 h after which all floating and attached cells were harvested for CAT ELISA (Roche Applied Science) and luciferase assays (Promega Corp.), according to the manufacturers' instructions. Twelve hours prior to transfection with the expression plasmids, hepatocytes were treated with vehicle or 100 M of either UDCA or TUDCA. At 48 and 60 h post-transfection, all cells were harvested, and total protein extracts were analyzed for p53 or Bax expression, respectively. In addition, attached cells were also fixed for morphologic detection of apoptosis.
Immunoblotting-Steady-state levels of pRb, p53, Mdm-2, Bcl-2, Bclx L , and Bax proteins, as well as NF-B and IB and their intracellular location were determined by Western blot. Briefly, 200 g of total, cytosolic, or nuclear protein extracts were separated on 6 or 12% SDSpolyacrylamide electrophoresis gels. Following electrophoretic transfer onto nitrocellulose membranes, immunoblots were incubated with 15% H 2 O 2 for 15 min at room temperature. After blocking with 5% milk solution, the blots were incubated overnight at 4°C with primary mouse monoclonal antibodies reactive to p53 and Bax or primary rabbit polyclonal antibodies to pRb, Bcl-2, Bcl-x L , Mdm-2, NF-B p65 subunit and IB (Santa Cruz Biotechnology, Santa Cruz, CA); and finally with secondary antibodies conjugated with horseradish peroxidase (Bio-Rad Laboratories) for 3 h at room temperature. The membranes were processed for protein detection using Super Signal TM substrate (Pierce). ␤-actin was used as a loading control. Protein concentrations were determined using the Bio-Rad protein assay kit according to the manufacturer's specifications.
Immunocytochemistry-Hepatocytes were fixed with 1% paraformaldehyde in PBS, pH 7.4 at room temperature for 10 min, washed three times with a mixture containing PBS, 0.5% bovine serum albumin, and 0.01% sodium azide, and permeabilized with 0.1% Triton-100 in PBS for 10 min. After blocking with PBS, 0.5% bovine serum albumin, and 0.01% sodium azide for 15 min, cells were incubated with an activationdependent anti-NF-B antibody at a dilution of 1:50 in PBS, 0.5% bovine serum albumin, and 0.01% sodium azide for 1 h at room temperature. After washing, hepatocytes were incubated with secondary donkey anti-goat IgG antibody conjugated with Cy5 at a dilution of 1:100 for 1 h at room temperature in the dark. Cells were washed for 5 min and mounted using Fluoromount-G. Fluorescence was visualized using a MRC1000 confocal microscope (Bio-Rad).
Densitometry and Statistical Analysis-The relative intensities of protein and nucleic acid bands were analyzed using the ImageMaster 1D Elite densitometric analysis program (Amersham Biosciences). Statistical analysis was performed using GraphPad InStat version 3.00 for Windows 95 (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.

RESULTS
UDCA Inhibits E2F-1-mediated Transcription Induced by TGF-␤1-It has been established that unbound E2F mediates apoptosis by TGF-␤1 in human hepatoma HuH-7 cells (14). In addition, UDCA inhibits TGF-␤1-induced apoptosis in HuH-7 cells and in primary rat hepatocytes (4,5), suggesting that E2F could be an important regulatory factor targeted by UDCA. In this study, apoptosis was assessed by changes in nuclear morphology and by caspase activation (Fig. 1). Significant levels of apoptosis occurred in cultured primary rat hepatocytes after incubation with TGF-␤1, with a maximum apoptotic response at 36 h (p Ͻ 0.01). UDCA and TUDCA protected against the TGF-␤1-induced nuclear fragmentation and caspase-3-like activation by 50 -70%. In contrast, the caspase-3 inhibitor significantly prevented apoptosis at 12 h (p Ͻ 0.05), but not at later time points, suggesting that UDCA protection is not simply due to caspase inhibition. Both re-supplementation with z-DEVDfmk at 12 h and caspase activity assays confirmed that the weak inhibitory effect on apoptosis after 12 h did not result from loss of activity of the caspase inhibitor. In addition, similar results were obtained with caspase-9 and -8 inhibitors, as well as with the z-VAD-fmk pan-caspase inhibitor, thus minimizing the possibility that other caspases could induce apoptosis at 24 h or later, in the absence of caspase-3 activation (data not shown).
Primary rat hepatocytes were then transfected with one of two CAT transcription reporter constructs under E2F-dependent promoters, E2F-1CAT and 4xE2FCAT, and exposed to TGF-␤1 for up to 36 h. The observed cell death was associated with a marked increase in transcriptional activation of E2F-1, which was detectable at 6 h but significant only at 12 h (p Ͻ 0.05) through 36 h (p Ͻ 0.01) of incubation with TGF-␤1 ( Fig.  2A). Interestingly, UDCA significantly abrogated the TGF-␤1induced E2F-1 expression as early as 12 h (p Ͻ 0.05) (Fig. 2B). Thus, we determined whether UDCA reduces E2F-1 activation by inhibition of caspase activity. Our results indicated that inhibitors to caspase-3, -9, or -8 did not prevent TGF-␤1-induced E2F-1 expression at 6 or 12 h, suggesting that E2F-1 is a caspase-independent target of UDCA. Curiously, all caspase inhibitors slightly increased TGF-␤1-induced E2F-1 activation, perhaps due to their inherent toxicity. However, because of their downstream role in the apoptotic cascade, it is not surprising that caspase inhibitors prevented apoptosis with TGF-␤1 at 12 h but not E2F-1 activation. In addition, CAT activity was Ͼ2-fold higher with 4xE2F than with the E2F-1 promoter at 36 h (Fig. 2C). UDCA reduced CAT activity by ϳ 70% in cells transfected with E2F-1CAT (p Ͻ 0.01) and 4xE2FCAT (p Ͻ 0.05) reporter plasmids. TUDCA also significantly inhibited TGF-␤1-mediated E2F transcription at 36 h Cells were fixed and stained for morphological evaluation of apoptosis and cytosolic proteins were extracted for caspase activity assays as described under "Experimental Procedures." Top, percentage of apoptosis in cells exposed to TGF-␤1 Ϯ bile acids or caspase-3 inhibitor for the indicated time points. Bottom, DEVD-specific caspase activity in cytosolic fractions after incubation with TGF-␤1 Ϯ bile acids or caspase-3 inhibitor for the indicated times. Cells treated with z-DEVD.fmk alone did not show significant differences compared with controls. The results are expressed as mean Ϯ S.E. of at least three different experiments. *, p Ͻ 0.01 from all others at the same time point.
(p Ͻ 0.01). Based on the observation that TGF-␤1 induced significant hepatocyte apoptosis, we determined whether the activation of E2F transcription was due to a caspase-mediated loss of pRb. The results showed that TGF-␤1-induced transcriptional activation of E2F-1 was indeed associated with a loss of pRb at 12 h through 48 h (Fig. 3A). However, the inhibition of the E2F-mediated transcription by UDCA and TUDCA was not entirely dependent on changes in pRb expression. In fact, both bile acids significantly prevented TGF-␤1 associated loss of pRb at 12 and 24 h, but not beyond these time points (Fig. 3B). Moreover, coincubation with caspase inhibitors did not prevent TGF-␤1-induced pRb degradation at 6 or 12 h, and further increased pRb loss at 6 h (Fig. 3C). Thus, these data suggest that TGF-␤1-induced E2F activation is not merely a secondary event due to decreased levels of pRb, but rather directly modulated by UDCA.
UDCA Prevents TGF-␤1-induced p53 Stabilization by the p53/Mdm-2 Pathway-Many cellular genes have been identified that contain E2F sites contributing to transcriptional regulation (42,43). The p53/Mdm-2 pathway is a potential downstream target of TGF-␤1 via the E2F-1 transcription factor, which can stabilize p53 through inhibition of the Mdm-2 protein (44,45). Therefore, protein extracts were prepared from primary rat hepatocytes incubated with TGF-␤1 in the pres- FIG. 2. Effects of UDCA on E2F-1mediated transcription in primary rat hepatocytes incubated with TGF-␤1. Cells were cotransfected with CAT transcription reporter plasmids under E2Fdependent promoters, E2F-1CAT or 4xE2FCAT, and luciferase control constructs as described in "Experimental Procedures." Twelve hours later, vehicle or 100 M of either UDCA or TUDCA was added to cells. After an additional 12 h, 1 nM TGF-␤1 was included in the cultures and cells harvested for the CAT ELISA and luciferase assays. Cells were also incubated with 50 M of either caspase-3 (z-DEVD.fmk), caspase-9 (z-LEHD.fmk), or caspase-8 inhibitor (z-IETD.fmk) 1 h prior to adding TGF-␤1. A, E2F-1-mediated transcription in cells exposed to TGF-␤1 for 6, 12, 24, and 36 h. B, CAT activity in cells incubated with TGF-␤1 Ϯ UDCA or caspase inhibitors for 6 and 12 h. Cells treated with caspase inhibitor alone did not show significant differences compared with controls. C, E2F-1-and 4xE2F-mediated transcription in cells exposed to TGF-␤1 for 36 h Ϯ bile acids. CAT activity (absorbancy/mg protein) was normalized to control luciferase expression, and the results are expressed as mean Ϯ S.E. arbitrary units for 3-5 different experiments. §, p Ͻ 0.05 and †, p Ͻ 0.01 from control; ‡, p Ͻ 0.05 and *, p Ͻ 0.01 from TGF-␤1. ence or absence of bile acid. Immunoblot analysis of p53 and Mdm-2 protein expression with TGF-␤1 showed a 5-fold increase in p53 levels (p Ͻ 0.05) and a 2-fold decrease of the p53 protein inhibitor Mdm-2 (p Ͻ 0.01) (Fig. 4). Incubation with UDCA or TUDCA alone produced no significant changes in p53 and Mdm-2 protein levels. However, coincubation with UDCA reduced TGF-␤1 increase in p53 levels by 98% (p Ͻ 0.05) and inhibited the decrease in Mdm-2 by 50% (p Ͻ 0.01), while coincubation with TUDCA was similarly protective.
UDCA Modulates TGF-␤1-induced Expression of Bcl-2 Family Members-Through its activity as a sequence-specific transcriptional activator, p53 has been shown to regulate the expression of Bcl-2 family proteins, including Bax and Bcl-2 (28,29). Our data suggested that exposure to TGF-␤1 in primary rat hepatocytes resulted in E2F-1-mediated p53 stabilization, which in turn was inhibited by UDCA and TUDCA. Thus, we investigated whether Bcl-2 family proteins might also be transcriptionally modulated under the same experimental conditions. At 36 h of incubation, RNA was extracted from TGF-␤1treated cells for RT-PCR analysis (Fig. 5A). The results indicated a significant decrease in bcl-2 mRNA levels (p Ͻ 0.01), together with a Ͼ 30% increase in bax transcript expression (p Ͻ 0.01). TGF-␤1-induced changes were abrogated by coincubation with either UDCA or TUDCA. In fact, UDCA completely inhibited both the decrease of bcl-2 and the increase of bax mRNA (p Ͻ 0.05), while TUDCA was slightly more efficient in preventing bax transcript accumulation than the decrease in bcl-2 mRNA.
Modulation of Bcl-2 family proteins was confirmed by immunoblot analysis (Fig. 5B). Bcl-2 protein levels were decreased by 50 -90% throughout the time course of 12-48 h incubation with TGF-␤1 (p Ͻ 0.01). In contrast, although Bax protein levels were markedly lower than the controls at earlier time points, there was a significant induction of Bax protein by TGF-␤1 at 48 h compared with 36 h (p Ͻ 0.01). The anti-apoptotic Bcl-x L protein and mRNA levels remained relatively unchanged throughout the time course (data not shown). Coincubation with either UDCA or TUDCA inhibited Bcl-2 protein alterations. UDCA reduced TGF-␤1 associated Bcl-2 and Bax protein changes at 36 h by 70 and 65%, respectively (p Ͻ 0.05), while TUDCA was slightly less protective (Fig. 5B, lower  panel). Similarly, both UDCA and TUDCA abrogated changes in Bcl-2 and Bax protein levels induced by TGF-␤1 at 12 and 24 h of incubation (p Ͻ 0.05). Finally, the change of Bax protein observed at 48 h was also entirely prevented by both bile acids (p Ͻ 0.05). The E2F-1/p53/Bax Pathway Is Modulated by UDCA-To further characterize the mechanism by which UDCA modulates the p53-regulated apoptosis pathway, we investigated its effects within the E2F-1/p53/Bax pathway. Hepatocytes were transfected with plasmids to overexpress wild-type or mutant E2F-1, or p53. More than 40% of cells with wild-type E2F-1 showed morphologic signs of apoptosis after Hoechst staining (p Ͻ 0.01) compared with only 10% of cells transfected with the E2F-1 mutant (Fig. 6A). Significant nuclear fragmentation was also observed in hepatocytes that overexpressed p53 (p Ͻ 0.01). Interestingly, both UDCA and TUDCA markedly reduced apoptosis after transfection with wild-type plasmids to control levels. Hepatocytes that overexpressed wild-type E2F-1 showed an almost 50% direct increase in p53 protein levels (p Ͻ 0.01) relative to cells expressing the corresponding E2F-1 mutant (Fig. 6B). Interestingly, pretreatment with either UDCA or TUDCA completely abolished E2F-1-driven p53 expression (p Ͻ 0.05). Bax protein was also increased after transfection of hepatocytes with wild-type E2F-1, and both UDCA and TUDCA reduced expression to control levels (data not shown). Hepatocytes overexpressing wild-type p53 showed a 3-fold direct increase in Bax protein compared with cells expressing mutant p53 (p Ͻ 0.05). In addition, both bile acids inhibited the p53-induced Bax levels by Ͼ 60% (p Ͻ 0.05). Finally, E2F-1 and p53 control transfections with the mutant version of each construct showed low levels of p53 and Bax protein, respectively (data not shown).
UDCA Reduces the Degradation of NF-B and Its Inhibitor IB by TGF-␤1-It has also been suggested that E2F-1-induced apoptosis can occur independently of p53, through inhibition of the anti-apoptotic signaling of NF-B. The regulation of Bcl-2 family proteins with TGF-␤1 and its modulation by UDCA or TUDCA, suggest that NF-B might play a functional role. Thus, we investigated NF-B translocation from the cytosol to the nucleus in hepatocytes treated with TGF-␤1. Immunoblotting and cytochemistry demonstrated that TGF-␤1 does not induce NF-B translocation to the nucleus; rather it is reduced simultaneously in the nucleus and the cytoplasm. In fact, TGF-␤1 reduced nuclear NF-B protein levels ϳ 40 and 90% at 12 h (p Ͻ 0.05) and 36 h (p Ͻ 0.01), respectively. Surprisingly, cytosolic levels of the transcription factor were similarly decreased in the TGF-␤1-treated cells compared with control hepatocytes (p Ͻ 0.01). Moreover, the overall decrease in total NF-B was not accompanied by an increase in IB, as the protein repressor was also markedly degraded during TGF-␤1 exposure (p Ͻ 0.01). Both UDCA and TUDCA alone caused no significant changes in total cellular levels of NF-B and IB. However, coincubation with UDCA or TUDCA partially inhibited TGF-␤1-induced NF-B degradation (p Ͻ 0.05), although more efficiently at 12 than at 36 h of incubation. The bile acids were less effective at inhibiting IB degradation. Finally, im-munocytochemistry confirmed that TGF-␤1 induces NF-B degradation in rat hepatocytes (data not shown). Both 12 and 36 h of TGF-␤1 exposure resulted in a marked simultaneous decrease in fluorescence signal in both nucleus and cytosol. Moreover, UDCA and TUDCA alone produced no detectable changes in NF-B immunostaining, whereas pretreatment with bile acids clearly prevented TGF-␤1-induced signal decrease. 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 prevents TGF-␤1-induced hepatocyte apoptosis by inhibiting the mitochondrial pathway of cell death (4,5). The results presented here provide additional mechanisms of action for UDCA, where the bile acid is shown to interfere with alternate and upstream molecular targets.
TGF-␤1 promotes the growth of fibroblasts and other cell types, inhibits epithelial cells and induces apoptosis of hepatocytes and various cancer cells. TGF-␤ functions by initially binding to its membrane-bound receptors followed by phosphorylation and dimerization of the receptor-activated Smad proteins. These molecules then translocate to the nucleus and target the transcriptional machinery (46). The cytokine is quite remarkable for its ability to affect other cellular signaling pathways that regulate either proliferation or apoptosis in different cell types. TGF-␤1-induced apoptosis in primary rat hepatocytes appears to be dependent on both protein synthesis and transcriptional activity (47). Caspase-3 activation, but not caspase-1, was shown to be involved in this process. In addition, we have reported that TGF-␤1 induces reactive oxygen species and the loss of the mitochondrial membrane potential prior to the apoptotic cell death (4,5). Both the mitochondrial permeability transition and the translocation of pro-apoptotic Bax resulted in the efflux of cytochrome c, which clearly preceded and was required for subsequent caspase activation, poly(ADP-ribose) polymerase cleavage, and nuclear fragmentation induced by TGF-␤1. Notably, UDCA antagonized the pro-apoptotic effect of TGF-␤1, as pretreatment of primary rat hepatocytes or HuH-7 hepatoma cells with the bile acid suppressed cell death as well as the above apoptotic events by ϳ 80%.
In the present study, we investigated additional effects of UDCA on TGF-␤1-induced apoptosis in primary rat hepatocytes. The bile acid either directly or indirectly was a potent inhibitor of TGF-␤1-induced transcriptional activation of E2F-1-mediated apoptosis. In fact, UDCA prevented the changes in p53 levels by E2F-1 overexpression as well as the increased Bax expression induced by p53. However, it did not significantly reduce the loss of pRb by TGF-␤1. UDCA also inhibited FIG. 4. Effects of UDCA on TGF-␤1induced modulation of p53 and Mdm-2. After plating, primary rat hepatocytes were incubated with 1 nM TGF-␤1 or no addition (control) Ϯ 100 M UDCA or TUDCA for 36 h. In coincubation experiments, UDCA or TUDCA was added to hepatocytes 12 h prior to incubation with TGF-␤1. Total proteins were extracted and subjected to Western blot analysis as described under "Experimental Procedures." The results are expressed as mean Ϯ S.E. arbitrary units of at least four different experiments. §, p Ͻ 0.05 and †, p Ͻ 0.01 from control; *, p Ͻ 0.01 from TGF-␤1. the down-regulation of Bcl-2 by TGF-␤1, which is consistent with decreased p53 stabilization and/or reduced degradation of NF-B.
p53 appears to play a significant role in E2F-1 associated apoptosis. The regulation of p53 is thought to occur mainly through modulation of protein stability. In normal cells, p53 is tightly regulated by a negative feedback loop in which the tumor suppressor induces Mdm-2 transcription, which in turn binds to p53 and mediates its degradation (44,45). In cells under stress, p53 is stabilized through inhibition of its degradation by Mdm-2 (48). The transcription factor E2F-1 has also been shown to stabilize p53 through induction of the human tumor suppressor protein p14 ARF (49), which directly binds to Mdm-2 and prevents the degradation of p53 (50). Our model supports a p53-dependent mechanism of E2F-1 mediated apoptosis in that TGF-␤1 induced E2F-1 transactivation, which then resulted in Mdm-2 degradation and p53 stabilization. The 3-fold greater induction of the 4xE2F construct relative to the E2F-1 promoter by TGF-␤1 suggests that TGF-␤1-induced apoptosis likely involves other transduction pathways.
Moreover, TGF-␤1 appears to also modulate mRNA and protein levels of Bcl-2 family members in primary rat hepatocytes resulting in down-regulation of antiapoptotic Bcl-2 and increased expression of proapoptotic Bax. This is not surprising since p53 is a known modulator of bcl-2 and bax gene expression (28Ϫ30). Further, pRb was recently shown to specifically interact with the promoter of bcl-2 and regulate its transcrip- FIG. 5. Modulation of Bcl-2 family members in primary rat hepatocytes. Cells were incubated with 1 nM TGF-␤1 or no addition (control) Ϯ 100 M UDCA, or TUDCA. In coincubation experiments, UDCA or TUDCA was added to hepatocytes 12 h prior to incubation with TGF-␤1. Total mRNA was obtained for RT-PCR analysis, and total proteins were extracted for Western blot analysis as described under "Experimental Procedures." A, representative RT-PCR of bax and bcl-2 in cells exposed to TGF-␤1 for 36 h Ϯ bile acids. B, immunoblots of Bax and Bcl-2 proteins, and corresponding histograms, in cells exposed to TGF-␤1 for 12, 24, 36, and 48 h. The results are expressed as mean Ϯ S.E. arbitrary units of at least four different experiments. §, p Ͻ 0.05 and †, p Ͻ 0.01 from control. C, immunoblots of Bax and Bcl-2 proteins in cells exposed to TGF-␤1 for 12, 24, 36, and 48 h, Ϯ bile acids. The blots were normalized to either endogenous ␤-actin mRNA or ␤-actin protein levels.
FIG. 6. UDCA specifically inhibits the E2F-1/p53/Bax pathway. Twelve hours after incubation with either 100 M UDCA or TUDCA, cells were transfected with the constructs pCMVE2F-1 and pCMVE2F-1⌬53, or with pCMVp53 and pCMVp53(143Ala) plasmids. At 48 and 60 h post-transfection, cells were fixed and stained for morphological detection of apoptosis. In addition, total proteins were extracted and subjected to Western blot analysis of p53 and Bax as described in "Experimental Procedures." A, percentage of apoptosis in cells exposed to bile acids and transfected with the constructs pCMVE2F-1 and pCMVE2F-1⌬53 (left), or with pCMVp53 and pCMVp53(143Ala) plasmids (right). The results are expressed as mean Ϯ S.E. of at least 3 different experiments. B, p53 expression in cells transfected with the constructs pCMVE2F-1 and pCMVE2F-1⌬53 (left) and Bax expression in cells transfected with pCMVp53 and pCMVp53(143Ala) plasmids (right). The results were normalized to luciferase expression and expressed as mean Ϯ S.E. arbitrary units relative to mutant E2F-1 or p53 of at least three different experiments. §, p Ͻ 0.05 and †, p Ͻ 0.01 from respective mutant; *, p Ͻ 0.01 from respective wild-type. tion (51). Whereas Bcl-2 has been considered necessary to block TGF-␤1-induced apoptosis (52), Bax is required for radiationinduced TGF-␤1 enhanced apoptosis (53). UDCA, by decreasing E2F-1 transcriptional activation, prevented the downstream events of TGF-␤1-induced cell death associated with Mdm-2, p53, bcl-2, and bax. The protective role of UDCA could also result from its direct effect on mitochondria, thus inhibiting caspase-3 activation, and the downstream degradation of both pRb and Mdm-2 (54). Previous studies have indeed shown that pretreatment of TGF-␤1-exposed cells with caspase inhibitors prevented both apoptosis and the loss of pRb (55). However, our results from transgene overexpression and caspase inhibition suggest that UDCA can specifically modulate the E2F-1/p53/ Bax pathway, abrogating E2F-1-induced p53 and p53-associated Bax expression, independently of its effect on mitochondria and/or caspases. Further, cleavage of pRb was not markedly affected by this bile acid. Thus, the ability of UDCA to inhibit TGF-␤1-induced apoptosis appears to involve both stabilization of the mitochondrial membrane and inhibition of Bax translocation (4,5), as well as modulation of the E2F-1 apoptotic pathway.
The identification of p14 ARF as an E2F target gene links E2F-1 activation to elevation of p53, but does not fully explain the role of E2F-1 in cell death. Indeed, several studies have reported E2F-1-induced apoptosis in p53-null cells (56,57), while others confirmed that E2F-1 promotes cell death by inhibiting anti-apoptotic factors such as NF-B (22). Our results also demonstrated that increased E2F-1 expression induced by TGF-␤1 was associated with a rapid and marked decline of NF-B and IB levels, without detectable nuclear translocation of NF-B. Thus, the NF-B survival pathway is not activated during TGF-␤1-induced apoptosis in primary rat hepatocytes. In fact, it was recently shown that TGF-␤1 resulted in a rapid decrease of cytosolic IB in mink lung Mv1Lu epithelial cells, suggesting that IB participates in TGF-␤1-mediated growth arrest (58). In addition, IB expression also appears to be regulated by the nuclear translocation of p53. IB normally interacts with p53, and the IB-p53 complex is perturbed in response to TGF-␤1, hypoxia, or DNA damage. Interestingly, our results also show that UDCA partially reduces degradation of NF-B and IB, which may explain, in part, the restoration of Bcl-2 levels in the presence of UDCA. In fact, Bcl-2 is also an activator of the NF-B transcription factor (59), and its downregulation may also result from NF-B degradation during TGF-␤1-induced apoptosis.
Our studies suggest that UDCA specifically inhibits the E2F-1/p53 apoptotic pathway and reduces NF-B degradation induced by TGF-␤1, thus modulating the expression of Bcl-2 family elements. However, it remains to be determined whether the activation of apoptosis by E2F-1 is modulated directly by UDCA or the result of a downstream effect, independent of caspase inhibition. Nevertheless, further identification of cellular targets of this bile acid may evolve into pharmacological approaches that can perhaps more efficiently regulate death and survival pathways.