Transforming Growth Factor-β-induced Apoptosis Is Mediated by Smad-dependent Expression of GADD45b through p38 Activation*

Transforming growth factor-β (TGF-β)-dependent apoptosis is important in the elimination of damaged or abnormal cells from normal tissues in vivo. In this report, we identify GADD45b as an effector of TGF-β-induced apoptosis. GADD45b has been shown to be a positive mediator of apoptosis induced by certain cytokines and oncogenes. We show that Gadd45b is an immediateearly response gene for TGF-β and that the proximal Gadd45b promoter is activated by TGF-β through the action of Smad2, Smad3, and Smad4. We show that ectopic expression of GADD45b in AML12 murine hepatocytes is sufficient to activate p38 and to trigger apoptotic cell death, whereas antisense inhibition of Gadd45b expression blocks TGF-β-dependent p38 activation and apoptosis. Furthermore, we also show that TGF-β can activate p38 and induce apoptosis in mouse primary hepatocytes from wild-type mice, but not from Gadd45b–/– mice. All of these findings suggest that GADD45b participates in TGF-β-induced apoptosis by acting upstream of p38 activation.

The number of cells in a particular organ is maintained through a delicate balance between cell proliferation and cell death. Precise regulation of cell division and apoptosis is required for normal morphogenesis, and alterations in these processes can lead to neoplastic transformation. Transforming growth factor-␤ (TGF-␤) 1 is a multifunctional protein with a broad spectrum of cellular activities ranging from the regulation of target gene activity to the control of cell growth and apoptosis (1,2). TGF-␤-dependent apoptosis is important in the elimination of damaged or abnormal cells from normal tissues in vivo. For example, TGF-␤ has been implicated in controlling liver size, and an intravenous injection of TGF-␤ induces atrophy and apoptosis in the normal and regressing liver (3,4).
Hepatic overexpression of TGF-␤ in transgenic mice also causes apoptosis (5), as does treatment of primary hepatocytes with TGF-␤ (6). Moreover, altered expression of TGF-␤ and its receptors in human and rodent hepatic neoplasias has been linked to apoptosis (7)(8)(9), consistent with the ability of TGF-␤ to induce apoptosis in hepatoma cell lines (10,11).
The TGF-␤ ligands signal through a heteromeric receptor complex consisting of both type I and II transmembrane receptor serine/threonine kinases. After TGF-␤ binds, the type II receptor kinase transphosphorylates and thereby activates the type I receptor kinase, which subsequently phosphorylates the cytoplasmic signal transducers Smad2 and Smad3. Receptoractivated Smad2 and Smad3 undergo a conformational change that allows heteromerization with a common partner, Smad4 (12)(13)(14)(15). This complex is subsequently translocated to the nucleus. In the nucleus, Smad complexes act as TGF-␤-sensitive transcriptional coactivators or corepressors through their interaction with a variety of transcription factors (16 -19). Current understanding of the mechanisms used by TGF-␤ to elicit its various cell biological effects is limited mostly to its effects on cell cycle arrest. Thus, although TGF-␤-induced apoptosis is a well documented phenomenon in many different cell types, the biochemical mechanisms responsible for mediating this death process are still poorly understood. Several pro-apoptotic events, including induction of oxidative stress (20), down-regulation of Bcl-2 family members (21)(22)(23)(24), up-regulation of the pro-apoptotic factor Bax (25), and activation of caspase proteases (26), have been implicated in TGF-␤-mediated apoptosis. Recently, several intracellular mediators, including apoptosisrelated protein in the TGF-␤ signaling pathway, Daxx, deathassociated protein kinase, and Src homology 2 domain-containing 5Ј-inositol phosphate, were shown to be involved in apoptosis induced by TGF-␤ (27)(28)(29)(30). However, the precise mechanisms whereby TGF-␤ signaling leads to the activation of these pro-apoptotic events are currently unknown.
Using microarray analyses of mouse embryonic fibroblast cells from wild-type and Smad2 and Smad3 knockout mice, we found that the mRNA level of Gadd45b is strongly up-regulated in cells treated with TGF-␤ for 1 h (31). Previous studies have suggested that the stress-and cytokine-inducible GADD45 family proteins (GADD45a, GADD45b, and GADD45g; also referred to as GADD45, MyD118, and CR6 as well as GADD45␣, GADD45␤, and GADD45␥) function as specific activators of MTK1 (also known as MEKK4), a MAPK kinase kinase upstream in the p38 pathway (32), and induce apoptosis (33). We have also suggested that specific binding of GADD45a to p38 (but not to JNK or ERK) is a potential mechanism by which GADD45a has a direct role in p38 activation by Ha-Ras (34). In this report, we identify GADD45b as an effector of TGF-␤-induced apoptosis. * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
TGF-␤ induces GADD45b expression in a Smad-dependent manner. We show that ectopic expression of GADD45b in AML12 cells is sufficient to activate p38 and to trigger apoptotic cell death, whereas antisense inhibition of Gadd45b expression blocks TGF-␤-dependent p38 activation and apoptosis. Furthermore, we also show that TGF-␤ can activate p38 and induce apoptosis in mouse primary hepatocytes from wild-type mice, but not from Gadd45b knockout mice. All of these findings suggest that GADD45b participates in TGF-␤-induced apoptosis by acting upstream of p38 activation.
Plasmid Constructs-The mouse Gadd45b promoter fragment has been described previously (35). 5Ј-Deletion constructs of the mouse Gadd45b promoter were made by PCR using pG118Bgl (35) as a template. PCRs were carried out with the proofreading Pfu DNA polymerase. After digestion with KpnI/XhoI or MluI/XhoI, the PCR products were extracted from an agarose gel with the QIAquick gel extraction kit (QIAGEN Inc.). The purified fragments were subcloned into the pGL-3 basic vector (Promega). All constructs were verified by DNA sequencing.
Transfection and Luciferase Assays-NIH3T3 and AML12 cells were plated at 2.0 ϫ 10 5 cells/well in 6-well plates 24 h prior to transfection. Cells were transfected with the reporters in the presence or absence of expression plasmids using the pcDNA3 plasmid to normalize the total amount of transfected DNA and pCMV-␤-gal to normalize transfection efficiency. Cells were transfected with FuGENE 6 (Roche Applied Science) according to the manufacturer's instructions. After 24 h, the transfection medium was replaced with 0.2% fetal bovine serum-containing medium, and the cells were left untreated or were stimulated with 5 ng/ml TGF-␤. After 18 h, the cells were lysed, and the luciferase and ␤-galactosidase activities were determined. All assays were performed in triplicate.
Generation of Stable Cell Lines-AML12 cells were transfected with hemagglutinin-tagged and antisense Gadd45b expression plasmid. Forty-eight h after transfection and every 3-4 days thereafter, the cells were refed with fresh selective medium containing G418 (Invitrogen) at a final concentration of 500 g/ml. Neomycin-resistant clones were first visible after 7 days and were continuously cultured in selective medium for ϳ14 days. The clones were then individually transferred into 6-well plates for expansion using a cloning cylinder (Sigma). After two additional passages in selective medium, expanded independent clones were cultured in standard medium. As a mock transfection control, the pcDNA3 empty vector was used to transfect AML12 cells and selected in the presence of neomycin following the same procedure.
Immunoblotting and p38 Kinase Assay-At different time periods after treatment with TGF-␤, cells were lysed in lysis buffer (25 mM HEPES, 150 mM NaCl, 10% glycerol, 5 mM EDTA, 1% Triton X-100, 5 mM sodium orthovanadate, 50 mM NaF, 0.5 mg/ml 4-(2-aminoethyl)benzenesulfonyl fluoride, and 1 g/ml pepstatin), and protein concentrations in cell lysates were determined using the BCA reagent (Pierce). For immunoblotting, whole cell lysates were resolved by SDS-PAGE and transferred to nitrocellulose membranes. The membranes were immunoblotted with various antibodies, and the bound antibodies were visualized with horseradish peroxidase-conjugated rabbit IgG antibodies using the enhanced chemiluminescence Western blotting system (ECL, Amersham Biosciences).
p38 kinase activity was determined using a p38 MAPK assay kit (Cell Signaling Technology) as recommended by the manufacturer. Briefly, endogenous phospho-p38 was immunoprecipitated from 200 g of cell lysates with immobilized anti-phospho-p38 antibody overnight at 4°C. The precipitates were washed with lysis buffer and kinase buffer (25 mM Tris (pH 7.5), 5 mM ␤-glycerophosphate, 2 mM dithiothreitol, 0.1 mM Na 3 VO 4 , and 10 mM MgCl 2 ). p38 kinase activity was measured using 2 g of ATF-2 fusion protein as substrate, and the reaction was initiated by the addition of 200 M ATP. After the cells were incubated for 30 min at 30°C, the reactions were stopped with Laemmli sample buffer. The proteins were resolved by 12% SDS-PAGE and transferred to nitrocellulose membrane. The membrane was immunoblotted with anti-phospho-ATF-2 antibody.
Northern Blot Analysis and RT-PCR-Total RNA was extracted from AML12 cells at the indicated times using the RNeasy kit (QIAGEN Inc.) as directed by the manufacturer. RNAs were separated on a formaldehyde-agarose gel, transferred to a nylon filter. and then hybridized with 32 P-labeled cDNA probe for Gadd45b. The blot was washed with SSC/ SDS solutions before autoradiography. To assess the presence of sense transcripts in antisense transfectants, 2 g of total RNA was reversetranscribed using Superscript (Invitrogen) and random hexamer at 42°C for 50 min and at 70°C for 15 min. Following reverse transcription, the samples were incubated with RNase H for 20 min at 37°C. Subsequently, PCR amplification was performed as follows: 94°C for 1 min, 55°C for 1 min, and 72°C for 1 min for 30 cycles, followed by a 10-min incubation at 72°C. The Gadd45b 5Ј-primer (cttctggtcgcacgggaagg) and 3Ј-primer (gctccaccgcggcagtcacc) and the glyceraldehyde-3phosphate dehydrogenase 5Ј-primer (ctcatgaccacagtccatgc) and 3Јprimer (ccctgttgctgtagccaaat) were used. The RT-PCR products were separated by electrophoresis on 2% agarose and stained with ethidium bromide.
Apoptosis Detection-For detection of apoptosis using the TUNEL assay, cells were fixed and stained using the in situ cell death detection POD kit (Roche Applied Science) and developed with 3,3Ј-diaminobenzidine substrate solution (Roche Applied Science). The cell death detection ELISA kit (Roche Applied Science), which detects internucleosomal fragmentation of DNA, was used with lysates of transfected cells according to the manufacturers' protocol. Results were read using an FL-600 microplate fluorescence reader (Bio-Tek).
Preparation of Primary Hepatocytes-Hepatocytes were isolated by two-step collagenase perfusion of the liver, followed by isodensity centrifugation in Percoll. Liver was perfused for 5 min with Hanks' balanced solution without calcium and magnesium and containing 10 mM HEPES and 0.2 mM EGTA, followed by a 10 -15-min perfusion with Williams' medium E containing 10 mM HEPES and 0.03% collagenase H (0.19 units/mg; Roche Applied Science). Viable hepatocytes were selected by isodensity centrifugation in Percoll and plated in a 1:1 mixture of Dulbecco's modified Eagle's medium and Ham's F-12 medium supplemented with 6.25 g/ml insulin, 6.25 g/ml transferrin, 6.25 ng/ml selenium, 1.25 mg/ml bovine serum albumin, 5.25 g/ml linoleic acid, 2 mM glutamine, 30 g/ml proline, 1 mg/ml galactose, 18 mM HEPES, 1 mM sodium pyruvate, 14 mM sodium bicarbonate, and 10% fetal bovine serum. After a 4-h attachment, the medium was changed to 0.2% fetal bovine serum-containing medium.

Induction of Smad-dependent Gadd45b Expression during TGF-␤-induced Apoptosis-Microarray analyses of mouse embryonic fibroblast cells from wild-type and Smad2 and Smad3
knockout mice demonstrated that Gadd45b is an immediateearly response gene for TGF-␤ treatment because its expression is up-regulated in 30 min, peaks at 1 h, and decreases to basal levels in 4 h (31). Northern blot analysis of RNA isolated from AML12 murine hepatocytes, which underwent apoptosis as early as 8 h after exposure to 5 ng/ml TGF-␤ (Fig. 1, A and  B), demonstrated that Gadd45b mRNA was present in these cells and was rapidly induced after treatment with TGF-␤ (Fig.  1C). To investigate whether the induction of Gadd45b expression by TGF-␤ requires de novo protein synthesis, we assessed the effect of the protein synthesis inhibitor cycloheximide. Although cycloheximide itself caused an increase in Gadd45b mRNA levels, a significant superinduction was seen in cells treated with cycloheximide and TGF-␤ (Fig. 1D), suggesting that Gadd45b is a primary target of TGF-␤ signaling. The increase in Gadd45b mRNA levels in the presence of cycloheximide was probably caused by an increase in mRNA stability or loss of transcriptional repressors by cycloheximide. The addition of actinomycin D, an inhibitor of transcription, inhibited the TGF-␤-induced Gadd45b expression (Fig. 1D), indicating that up-regulation of Gadd45b expression is transcriptionally dependent. Thus, we conclude that Gadd45b is an immediateearly gene product of TGF-␤ in cells responsive to TGF-␤induced apoptosis.
To determine whether TGF-␤-induced expression of Gadd45b involves direct transcriptional activation of the Gadd45b promoter, the 5Ј-flanking region of the Gadd45b gene was isolated and characterized. A DNA fragment corresponding to region Ϫ4730 to ϩ235 of the Gadd45b promoter (35) was inserted upstream of a luciferase reporter gene to determine the induction of transcription from the Gadd45b promoter by TGF-␤. When transfected into AML12 cells, this reporter conferred promoter activity and inducibility by TGF-␤ ( Fig. 2A). To identify the minimum promoter region required for induction by TGF-␤, a series of 5Ј-promoter deletion constructs was generated, and promoter activity was assessed in NIH3T3 cells because the promoter inducibility by TGF-␤ was higher in these cells than in AML12 cells ( Fig. 2A). Deletion up to position Ϫ220 still conferred responsiveness to TGF-␤. However, deletion of sequence Ϫ220 to Ϫ100 reduced TGF-␤ induction, and deletion up to position Ϫ1 totally abolished TGF-␤ induction (Fig. 2B). These findings suggest the existence of a TGF-␤-responsive sequence in promoter region Ϫ220 to Ϫ1.
Next, to confirm the involvement of Smad proteins in TGF-␤-induced Gadd45b promoter activation, we examined the effect of overexpression of Smad proteins on the Gadd45b promoter activity in the presence or absence of exogenous TGF-␤. Cotransfection of the Ϫ220/ϩ235 reporter construct, which contains the TGF-␤-responsive sequence (Fig. 2B), with Smad2, Smad3, and Smad4 independently resulted in an ϳ3-fold increase in luciferase activity, whereas combinations resulted in an ϳ5-fold increase (Fig. 2C). TGF-␤ enhanced the level of transactivation of the Gadd45b promoter by all overexpressed Smad proteins. Furthermore, expression of dominant-negative Smad2 and Smad3 completely abolished inducibility by TGF-␤ (Fig. 2C). Altogether, our results indicate that Smad proteins are required for TGF-␤-induced Gadd45b expression.
TGF-␤ Specifically Activates p38 in AML12 Cells, Resulting in Apoptosis-To address the role of GADD45b in TGF-␤-induced apoptosis in AML12 cells, we assessed the activation of p38 and JNK because GADD45b has been shown to activate p38 and JNK and to induce apoptosis (33). For this purpose, phospho-specific antibodies for the activated forms of p38, SAPK/JNK, and ERK1/2 were used in Western blot analyses of cell lysates. Anti-phospho-p38 antibody, which recognizes specifically the activated form of p38 phosphorylated at Thr 180 and Tyr 182 , revealed TGF-␤-induced p38 phosphorylation 8 h after stimulation of AML12 cells, which persisted for up to 24 h (Fig.  3A). The total amount of p38 did not change during the period of observation. In contrast, TGF-␤ did not lead to phosphorylation of SAPK/JNK in AML12 cells, and the level of phospho-ERK did not change significantly (Fig. 3B). We also assessed the activation of JNK in the FaO rat hepatoma cell line to confirm the sensitivity of anti-phospho-JNK antibody (data not shown). These results indicate that p38 MAPK (but not JNK or ERK) was activated in TGF-␤-induced apoptosis of AML12 cells.
To confirm the involvement of the p38 signaling pathway in TGF-␤-induced apoptosis, we investigated the role of SB203580, a selective inhibitor of p38, in TGF-␤-induced apoptosis in AML12 cells. Pretreatment of cells with SB203580 (1 M) blocked the apoptotic response induced by TGF-␤ (Fig. 3, C and D). Consistent with the lack of induction of JNK or ERK by TGF-␤ in these cells, pretreatment of cells with PD98059 and SP600125, selective inhibitors of ERK and JNK, respectively, had no effect on the apoptotic response induced by TGF-␤ (data not shown). All of these results suggest that p38 activation is important to induce TGF-␤-mediated apoptosis.
GADD45b Activates p38 MAPK and Enhances TGF-␤-mediated Apoptosis-Having demonstrated that Gadd45b expression is induced by TGF-␤ signaling through the action of Smad proteins and that p38 is activated by TGF-␤, we next investigated whether ectopic overexpression of GADD45b could activate p38 and lead to apoptosis in the absence of TGF-␤ stimulation. To test this prediction, AML12 cells were stably transfected with a hemagglutinin-tagged Gadd45b expression plasmid or empty vector used as a control. In GADD45b-expressing cells (Gadd#2 and Gadd#4, detected by anti-hemagglutinin antibody), p38 was activated in the absence of TGF-␤ treatment (Fig. 4A). To assure that the enhanced levels of activated p38 represented increased p38 kinase activity, an in vitro kinase assay was performed using ATF-2 as substrate. The results show that the basal level of p38 kinase activity in Gadd#2 cells exceeded that in empty vector control cells that had been treated with TGF-␤ and that this activity could be enhanced slightly by treatment of the cells with TGF-␤ ( Fig.   FIG. 1.

Up-regulation of Gadd45b expression during TGF-␤-induced apoptosis.
A and B, induction of apoptosis in AML12 cells by TGF-␤. Cells were treated with TGF-␤ (5 ng/ml) for various times, and DNA fragmentation was detected by ELISA (A) and TUNEL assay (B). *, p Ͻ 0.001 versus control. C, Northern blot analysis of Gadd45b RNA expression in AML12 cells stimulated with TGF-␤ for the indicated times. D, effect of cycloheximide (CHX; 5 g/ml) or actinomycin D (ActD; 5 g/ml) on TGF-␤-induced Gadd45b expression in AML12 cells. Cycloheximide and actinomycin D were added 1 h before TGF-␤ treatment. Equal loading of RNA samples (20 g/ml) in C and D was assessed by the ethidium bromide staining pattern of gels before Northern blotting. 4B). To confirm that the induction of GADD45b is a determining factor in the sensitivity of cells to the apoptotic effect of TGF-␤, we tested the ability of these cells to undergo apoptosis in the absence or presence of TGF-␤. In GADD45b-expressing cells, both the basal level of apoptosis and the sensitivity to TGF-␤-induced apoptosis were increased (Fig. 4, C and D). Thus, our results indicate that the induction of Gadd45b expression by TGF-␤ is an important step in the apoptotic pathway of TGF-␤.
To address the question of whether endogenous GADD45b might be essential for TGF-␤-mediated p38 activation and apoptosis, AML12 cells were stably transfected with antisense Gadd45b cDNA, thereby blocking expression of the endogenous Gadd45b gene. In each transfectant, Gadd45b mRNA levels were assayed by RT-PCR before and after TGF-␤ stimulation.
Antisense Gadd45b effectively suppressed endogenous Gadd45b expression (Fig. 5A, AS#1 and AS#9). We then assessed the p38 activity in these cells using anti-phospho-p38 immunoblot analysis. Fig. 5B shows that antisense Gadd45b RNA significantly suppressed p38 activation in response to TGF-␤ stimulation, indicating a requirement for GADD45b expression in TGF-␤-induced p38 activation. We further tested the effect of reduced GADD45b expression on apoptosis. Both the ELISA (Fig. 5C) and TUNEL stain (Fig. 5D) showed that AML12 cells transfected with antisense Gadd45b exhibited a marked reduction in their apoptotic response to TGF-␤, indicating that expression of endogenous GADD45b is necessary to direct TGF-␤ signaling toward apoptotic end points.
Finally, to determine whether GADD45b is also important for the TGF-␤-induced apoptosis of primary hepatocytes, we  A-B). B, serial deletion mutants of Gadd45b promoter constructs were transfected into NIH3T3 cells, untreated or treated with TGF-␤, and assessed for luciferase activity. C, NIH3T3 cells were transfected with the Ϫ220/ϩ235 Gadd45b promoter construct and various Smad expression plasmids as indicated. Cells were then stimulated with TGF-␤ and assessed for luciferase activity. Total cell lysates were prepared from untreated or TGF-␤-treated AML12 cells and used for immunoblotting. The phosphorylated (P) and the non-phosphorylated forms of p38, JNK, and ERK are indicated in the upper and lower panels, respectively. C and D, cell death detection ELISA analysis and TUNEL analysis, respectively, of cell lysates isolated from AML12 cells, TGF-␤-treated or not for 8 h in the presence or absence of SB203580. *, p Ͻ 0.001 versus control. examined the effects of TGF-␤ on apoptosis of primary hepatocytes from wild-type and Gadd45b-deficient mice (36). 2 TGF-␤ treatment of wild-type primary hepatocytes induced apoptosis, with ϳ50% of the cells undergoing cell death in wild-type primary hepatocytes compared with 6% in Gadd45b-deficient primary hepatocytes (Fig. 6, A and B). p38 was also activated by TGF-␤ in wild-type cells, but not in Gadd45b-deficient cells (Fig. 6C). Our results demonstrate that induction of GADD45b by TGF-␤ is a critical and necessary prerequisite for its p38dependent apoptotic effects in normal primary hepatocytes. DISCUSSION In this study, we explored the mechanisms of TGF-␤-induced apoptosis in hepatocytes. TGF-␤ regulates cell growth during embryonic development and has been suggested to be a potent inducer of apoptosis in hepatocytes in a wide variety of human liver diseases (37,38). Loss of TGF-␤-induced apoptosis has been implicated in the development of hepatocellular carcinoma (39). Our results provide strong evidence suggesting that GADD45b is a critical upstream component in the apoptotic pathway of TGF-␤ in the liver. First, TGF-␤ rapidly induced expression of GADD45b in cells sensitive to TGF-␤-induced apoptosis. Second, ectopic expression of GADD45b was sufficient to trigger apoptotic death in the absence of TGF-␤ stimulation. Third, down-regulation of GADD45b expression with an antisense construct inhibited TGF-␤-induced apoptosis. Finally, TGF-␤-mediated apoptosis was strikingly reduced in Gadd45b-deficient primary hepatocytes. Altogether, these data indicate that GADD45b functions as an effector of TGF-␤-dependent apoptosis. However, this does not rule out the possibility that other factors may cooperate with GADD45b in mediating TGF-␤-induced cell death or, alternatively, that other cell-specific pathways might also be operative.
In addition to demonstrating that GADD45b is essential for TGF-␤-induced apoptosis, we found that TGF-␤-dependent induction of GADD45b requires Smad proteins. Gadd45b has 2 A. G. Balliet, B. Hoffman, and D. A. Liebermann, unpublished data.

FIG. 5. GADD45b is necessary for TGF-␤-induced phosphorylation of p38 and apoptosis.
A, sense Gadd45b transcripts in AML12 cells stably transfected with either an empty vector (#1) or an antisense Gadd45b expression vector (AS#1 and AS#9) were subjected to RT-PCR detection. Following a 1-h incubation of the cells with or without TGF-␤, total RNA was prepared and subjected to RT-PCR using primers specific to the sense Gadd45b transcript. GAPDH, glyceraldehyde-3-phosphate dehydrogenase. B, phosphorylation of endogenous p38 in total cell lysates prepared from same cells in A, untreated or treated with TGF-␤ for 8 h, was analyzed by immunoblotting. The phosphorylated (P) and non-phosphorylated forms of p38 are shown in the upper and lower panels, respectively. C and D, AML12 cells stably transfected with either an empty vector (#1) or an antisense Gadd45b expression vector (AS#1 and AS#9) were treated with or without TGF-␤ for 8 h. DNA fragmentation was detected by ELISA (C) and TUNEL assay (D), respectively. *, p Ͻ 0.01 versus control. FIG. 6. GADD45b is necessary for TGF-␤-induced phosphorylation of p38 and apoptosis in primary hepatocytes. A, p38 was activated in primary hepatocytes isolated from wild-type mice (but not Gadd45b knockout mice) by TGF-␤ treatment. Phosphorylation of endogenous p38 in primary hepatocyte cell lysates purified from wild-type (WT) and Gadd45b knockout (KO) mice, untreated or treated with TGF-␤ for 8 and 24 h, was analyzed by immunoblotting. The phosphorylated (P) and non-phosphorylated forms of p38 are shown in the upper and lower panels, respectively. B, primary hepatocytes purified from wild-type and Gadd45b knockout mice were treated with or without TGF-␤ for 48 h, and DNA fragmentation was detected by TUNEL assays. C, the TUNEL assay was performed; positive apoptotic cells at each time were counted, and the percentage was determined. *, p Ͻ 0.001 versus control. been shown to be the only target gene for TGF-␤ among the Gadd45 family genes (Gadd45a, Gadd45b, and Gadd45g) (40). The rapid induction of Gadd45b expression by TGF-␤ and the finding that this induction does not require new protein synthesis indicate that Gadd45b is a direct target of TGF-␤ and Smad signaling. We also showed that cotransfection of a Smad expression plasmid with a Gadd45b promoter construct resulted in an increase in promoter activity, whereas expression of dominant-negative Smad proteins completely abolished inducibility by TGF-␤, suggesting that Smad proteins are required for Gadd45b expression by TGF-␤. We also identified the minimum promoter region required for induction by TGF-␤ using 5Ј-promoter deletion analysis. However, there are no identifiable Smad-binding elements (GTCT or CAGA) in this region (positions Ϫ220 to Ϫ1), and we failed to detect any inducible protein binding to this region by TGF-␤ (data not shown). Biochemical and overexpression studies have demonstrated that Smad proteins are capable of functional interaction with other transcription factors such as Sp1 and AP-1 (41)(42)(43). All of our promoter constructs have 235 bp of 5Јuntranslated region, and this region contains four consensus Smad-binding elements. We are now investigating the significance of these Smad-binding elements in the 5Ј-untranslated region and their possible collaboration with other transcription factor-binding element(s) in region Ϫ220 to Ϫ1 (44).
Our results demonstrate that TGF-␤ activates p38 at least in part through induction of GADD45b and that this activation of p38 is critical for TGF-␤-induced apoptosis in hepatocytes. p38 MAPK belongs to a subfamily of MAPKs and is activated primarily in response to stress as well as cytokine stimulation (45,46). Earlier studies suggest that different subfamilies of MAPKs play distinct roles in regulating apoptosis, with the ERK MAPK serving as a pro-survival signal and p38 and JNK as pro-death signals (47)(48)(49). In this regard, we found no significant effects of TGF-␤ on ERK and JNK MAPKs in AML12 cells, suggesting that, in these cells, p38 MAPK activation plays a predominant role in mediating TGF-␤-induced cell death. Further support for this hypothesis comes from our data demonstrating that the strong reduction in TGF-␤-mediated apoptosis in primary hepatocytes derived from Gadd45b-null mice is associated with an absence of TGF-␤-induced activation of p38. It is thought that these MAPK signal transduction pathways may mediate expression of pro-apoptotic genes (50). JNK phosphorylation and activation of the transcription factor c-Jun during cell death are well established (50). However, little is known about potential downstream pro-apoptotic targets of p38 MAPK. Monoamine oxidase has been identified as a gene that is regulated by the p38 MAPK pathway during nerve growth factor withdrawal-induced apoptosis (51), and p53 is also a target for p38 in apoptotic cells induced by UV radiation (52). Studies are ongoing in our laboratory to identify the downstream target genes of p38 MAPK during apoptosis in hepatocytes.
Our data suggest that delayed and sustained (rather than rapid and transient) activation of p38 may be important in inducing apoptosis. It has been reported that TGF-␤ activates p38 through both Smad-independent and Smad-dependent pathways. The kinetics of these two pathways are strikingly different. Smad-independent activation of p38 by TGF-␤ is dependent on TAK1, and TGF-␤ induces TAK1 activity rapidly, but only transiently (53,54). Rapid activation of p38 by TGF-␤, consistent with the kinetics of TAK1 activity, has been reported in some cell lines (55)(56)(57). In many other cell types, however, maximal p38 activation occurs only 1 or more hours following TGF-␤ stimulation, and the activity may persist for several hours (58 -62). The timing of this Smad-dependent p38 activa-tion can vary in different tissues or cells. Thus, we have shown that activation of p38 by TGF-␤ in primary hepatocytes is delayed compared with that seen in AML12 cells. This could be related to different culture conditions of the AML12 versus primary cells or to the nature of the cells themselves. It is noteworthy that, in both AML12 cells and primary hepatocytes, the induction of p38 by TGF-␤ was dependent on GADD45b, suggesting that the delay might represent the time required for accumulation of a permissive level of cellular GADD45b. Consistent with our results, it has recently been shown that GADD45b is responsible for the delayed activation of p38 by TGF-␤ in pancreatic cells (63) and that another member of the GADD45 family, GADD45a, functions to sustain activation of p38 after UV radiation (64). The close concordance of the response of AML12 cells stably overexpressing antisense Gadd45b RNA and the primary hepatocytes derived from Gadd45b-null mice, in terms of both the lack of p38 induction in response to TGF-␤ and protection from TGF-␤-induced apoptosis, provides strong evidence that GADD45b is also an important mediator for delayed and sustained activation of p38 by TGF-␤ in cells in which it is critical for TGF-␤-induced apoptosis such as hepatocytes.
Last, it must be noted that previous studies with hepatoma cell lines have suggested a more complex involvement of MAPK pathways in TGF-␤-mediated apoptosis. Thus, in FaO rat hepatoma cells, TGF-␤ treatment activates not only p38, but also JNK and ERK, in contrast to the untransformed hepatocyte cell line AML12, which we have used here and in which only p38 is activated. Inhibition of either JNK or p38 interferes with TGF-␤-dependent apoptosis in FaO cells, whereas inhibition of ERK potentiates the apoptosis, suggesting that a shift in the balance between the various MAPK pathways might be important in controlling the commitment to TGF-␤-induced apoptosis (48). Other hepatoma cell lines such as HepG2 are resistant to TGF-␤-induced apoptosis, suggesting that alterations in the pathway have occurred in the acquisition of a tumorigenic phenotype, resulting in escape from homeostatic control by TGF-␤ (65). However, the close mechanistic overlap between our studies in AML12 cells and primary hepatocytes suggests that the pathway we have described, TGF-␤ 3 GADD45b 3 p38 3 apoptosis, is representative of that in primary cells and likely in the liver in vivo.