X-linked Inhibitor of Apoptosis (XIAP) Inhibits c-Jun N-terminal Kinase 1 (JNK1) Activation by Transforming Growth Factor β1 (TGF-β1) through Ubiquitin-mediated Proteosomal Degradation of the TGF-β1-activated Kinase 1 (TAK1)*

Active NF-κB renders malignant hepatocytes refractory to the growth inhibitory and pro-apoptotic properties of transforming growth factorβ1 (TGF-β1). NF-κB counteracts TGF-β1-induced apoptosis through up-regulation of downstream target genes, such as XIAP and Bcl-XL, which in turn inhibit the intrinsic pathway of apoptosis. In addition, induction of NF-κB by TGF-β1 inhibits JNK signaling, thereby attenuating TGF-β1-induced cell death of normal hepatocytes. However, the mechanism involved in the negative cross-talk between the NF-κB and JNK pathways during TGF-β1 signaling has not been determined. In this study, we have identified the XIAP gene as one of the critical mediators of NF-κB-mediated suppression of JNK signaling. We show that NF-κB plays a role in the up-regulation of XIAP gene expression in response to TGF-β1 treatment and forms a TGF-β1-inducible complex with TAK1. Furthermore, we show that the RING domain of XIAP mediates TAK1 polyubiquitination, which then targets this molecule for proteosomal degradation. Down-regulation of TAK1 protein expression inhibits TGF-β1-mediated activation of JNK and apoptosis. Conversely, silencing of XIAP promotes persistent JNK activation and potentiates TGF-β1-induced apoptosis. Collectively, our findings identify a novel mechanism for the regulation of JNK activity by NF-κB during TGF-β1 signaling and raise the possibility that pharmacologic inhibition of the NF-κB/XIAP signaling pathway might selectively abolish the pro-oncogenic activity of TGF-β1 in advanced hepatocellular carcinomas (HCCs) without affecting the pro-apoptotic effects of TGF-β1 involved in normal liver homeostasis.

TGF-␤1 can signal through intracellular mediators known as Smads, which are localized in the cytoplasm (4). Following TGF-␤1 receptor (T␤RI)-mediated phosphorylation, Smad2 and Smad3 associate as heteromeric complexes with the common signaling mediator Smad4 and translocate into the nucleus to regulate gene expression. The inhibitory Smad7 competes for binding with Smad2 and Smad3 to the activated TGF-␤1 receptor and promotes ubiquitin-mediated receptor degradation, thereby suppressing TGF-␤1 signaling (4). TGF-␤1 promotes growth arrest through the induction of both the CDK inhibitors p15 INK4B and p21 CIP1 and the inhibition of genes that regulate cell cycle progression such as c-myc, CDC25A, and CDK6 (5). Conversely, CDK4 and CDK2 can impair TGF-␤1 signaling through direct phosphorylation of Smad3 (6).
TGF-␤1 mRNA levels are scarcely expressed in normal rodent and human livers where they appear to be localized predominantly in oval cells during early stages of hepatocytic differentiation and in nonparenchymal cells (7). However, during liver neoplastic development, liver regeneration, and hepatic fibrosis, TGF-␤1 is highly expressed in malignant hepatocytes (8) as well as in nontumorous mesenchymal cells (9). In this regard, TGF-␤1 is thought to be one of the critical mediators of the epithelial-mesenchymal interaction that regulates development, growth, and neoplastic transformation of epithelial cells (10). Furthermore, autocrine activity of TGF-␤1 can enhance the metastatic potential of advanced epithelial carcinomas (11,12).
The stage-dependent tumor-suppressing versus tumor-promoting activity of TGF-␤1 is explained, in part, by the ability of epithelial cancer cells to acquire resistance to the antiproliferative and pro-apoptotic effects of TGF-␤1 because of loss-of-function mutations within key components of the TGF-␤1 signaling pathway or impaired secretion or activation of the latent TGF-␤1 complex (1). However, several tumors are resistant to the growth inhibitory action of TGF-␤1 despite retaining an intact TGF-␤1 signaling pathway (13). Thus, a critical issue in cancer biology focuses on understanding how a cancer cell escapes from the growth inhibitory properties of TGF-␤1.
In the past, we have identified aberrant NF-B activity as one of the crucial events that renders hepatocellular carcinomas (HCCs) resistant to TGF-␤1-induced growth arrest and apoptosis (14 -16). NF-B/Rel comprises a family of dimeric factors involved in the regulation of development, regeneration, and neoplastic transformation of the liver (17). The canonical pathway of NF-B activation applies to dimers composed of RelA, c-Rel and p50 subunits that, in non-stimulated cells, are sequestered in the cytoplasm through interaction with IB-␣. In response to microbial, viral infections, DNA damage, and pro-inflammatory cytokines, the IB kinase (IKK) complex promotes NF-B activation through phosphorylation-induced ubiquitination of IB-␣, which tar-* This work was supported by National Institutes of Health Grant CA78616 (to M. A.) and, in part, by American Cancer Society Grant RSG-02-255-01-TBE (to M. A.). 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. 1  gets this molecule for proteolysis in the 26 S proteasome (18). The IKK complex is comprised of two catalytic subunits, IKK-␣ (IKK-1) and IKK-␤ (IKK-2) (19) as well as two scaffold components termed IKK-␥ and ELKS (20,21). Our previous data indicated that constitutive activation of the IKK complex through the Ras/phosphatidylinositol 3-kinasedependent pathway plays an important role in hepatic oncogenesis (15,22). Furthermore, we have found that the TAK1/IKK signaling pathway mediates the transient activation of NF-B in response to TGF-␤1 treatment of hepatocytes and fibroblasts (16). The mechanism by which NF-B contributes to liver tumor formation is, in part, caused by the transcriptional activation of prosurvival genes, which in turn oppose pro-apoptotic stimuli elicited by the host immune system. In this regard, we have shown that inhibition of NF-B activity potentiates TGF-␤1-induced apoptosis of immortalized hepatocytes (16,23). Furthermore, adenoviral-mediated inhibition of NF-B activity in murine HCCs promoted down-regulation of the prosurvival Bcl-X L and XIAP genes, thereby leading to pronounced cell killing of malignant hepatocytes by TNF-␣ and TGF-␤1 (24).
Previously, the induction of cell death by TGF-␤1 has been linked to its ability to activate the JNK/AP-1 axis, which then cooperates with SMADs in inducing TGF-␤1 responsive promoters of cell death-inducing genes (2,14,(25)(26)(27). Given that the mechanism of sensitization of murine and human HCCs to TGF-␤1-induced growth arrest and apoptosis following inhibition of NF-B activity also relies on release of JNK activity (16), in this study we have examined the potential for the NF-B downstream target gene XIAP to repress JNK activity. We show that XIAP physically interacts with TAK1 in a TGF-␤1-inducible manner thereby promoting ubiquitin-mediated degradation of TAK1 and repression of the JNK/AP-1 axis.
Collectively, these studies provide the conceptual framework for the future characterization of the IKK/JNK interplay in hepatic oncogenesis and illustrate a novel role of XIAP during protection from TGF-␤1induced apoptosis.

MATERIALS AND METHODS
Cell Culture and Treatment Conditions-Normal mouse hepatocyte (NMH) cells (28) were maintained as described previously (14). Primary rat hepatocytes were isolated from livers of male Sprague-Dawley rats (Harlan Laboratories, Jerusalem, Israel) by collagenase perfusion method. Livers were removed under Nembutal anesthesia (50 mg/ml) and were perfused in a perfusion chamber through the portal vein with calcium-free Krebs bicarbonate buffer (pH 7.4) containing collagenase and glucose. The livers were perfused at 37°C with buffer gassed continuously with O 2 /CO 2 . After 30 min, the perfusate was collected in a beaker and minced with sterile scissors at 37°C for 10 min. Digested liver was filtered through sterile mesh, and cells were separated from the debris by centrifugation. Cells were suspended in Williams E medium (Invitrogen) containing 5% fetal bovine serum (Sigma-Aldrich), 0.1% antibiotic/antimycotic solution and plated in a culture dish coated with rat tail collagen (Collaborative Biochemical Products, Bedford, MA). After 4 h, non-adherent cells were removed, and adherent cells were given the indicated treatments.
Transfection Conditions-For transient transfection, cells were transfected with a solution of DNA and Lipofectamine TM reagent according to manufacturer's instructions (Invitrogen). 16 h after transfection, cells were treated with 5 ng/ml TGF-␤1 or BSA carrier solution for 5 h. Cells were harvested after treatment according to the manufacturer's protocol in the Dual-Luciferase Reporter assay system (Promega), and lysates were analyzed with a Labsystems Luminoskan 96-well plate luminometer (Thermolab System, Needham Heights, MA). Firefly luciferase activity was normalized for Renilla luciferase activity, and results were expressed as a ratio (fold induction) of normalized luciferase activities in treated cells versus vehicle-treated cells. Significance (p Ͻ 0.01) was obtained using the Student's t test.
The adenoviral vectors expressing GFP or a dominant negative form of IKK-2 (IKK-2 KϾM) have been described previously (34). Virus stocks were amplified to high titer by Quantum Biotechnologies (Montreal, Canada). The concentration of viral particles (VP) was determined as described previously (34).
RNA Interference-The SMARTpool TM siRNA specific for murine XIAP and scrambled siRNA control were purchased from Upstate Group (Lake Placid, NY) and were transfected (50 -100 nM) into NMH cells according to the manufacturer's specifications using Lipofectamine TM 2000. Sixteen hours after transfection, cells were treated with 5 ng/ml TGF-␤1 or BSA-carrier solution for the indicated times. XIAP protein expression was determined by immunoblot analysis as described below using an antibody against XIAP (AAM-050, Stressgen, San Diego, CA).
RNA Isolation and Analysis-Total cellular RNA was isolated by the guanidinium method following the manufacturer's instructions (Qiagen) and samples (1 g) subjected to semi-quantitative RT-PCR analysis. Bands were quantified by densitometric analysis and are expressed in arbitrary units as the ratio of the optical density (OD) of XIAP to that of the GAPDH. The sequence of the primers specific for murine XIAP were as follows: mXIAP forward: 5Ј-ATGACTTTTAACAGTTTT-GAAGGAACT-3Ј, mXIAP reverse: 5Ј-GAAGCACTTCACTTTAT-CGCCT-3Ј.
To amplify the mouse GAPDH gene, we used the EZ-GENXpress TM normalized gene expression kit following the manufacturer's instructions (Maxim Biotech, Inc., San Francisco, CA).
In Vivo Ubiquitination Assay-NMH cells were plated in p100 Petri dishes and were transfected using Lipofectamine TM 2000 with 7 g of pFLAG-CMV-2-TAK1 construct in the presence or absence of 7 g of pCW7-H 6 M-Ub or pCW8-H 6 M-Ub, with or without 7 g of pCS3-MT-XIAP, pCS3-XIAP-⌬BIR, or pCS3-XIAP-⌬RING. Three hours after transfection, 20 M MG132 were added to the culture medium.
After 24 h, cells were harvested in lysis buffer and WCEs (300 g) were immunoprecipitated with an anti-FLAG antibody (F-7425) (Sigma-Aldrich) or mouse IgG control (Upstate Group) and subjected to immunoblotting using a polyclonal anti-ubiquitin antibody (a kind gift from Dr. Min Wu, Novartis). Equal immunoprecipitated aliquots were also subjected to immunoblotting using an anti-FLAG antibody (F-4042) (Sigma-Aldrich).
TUNEL and Caspase-3 Assays-TUNEL assay was performed on cultures of live cells using the DeadEnd TM Colorimetric TUNEL System (Promega) following the manufacturer's instructions as we have described previously (24). For the caspase-3 assay, NMH cells were seeded in p35 Petri dishes and were given a pretreatment with 10 M SP600125 for 30 min or with 5 M JIP-1 for 3 h followed by treatment with TGF-␤1 (5 ng/ml) for 48 h. After 24 -48 h, the cells were washed with cold 1ϫ PBS and lysed with cell lysis buffer. The lysates were sheared through sonication and centrifuged to obtain the supernatants. Cell extracts (100 g) were subjected to caspase-3 assay using the CaspAce colorimetric assay kit (Promega).

JNK Activity Is Required for TGF-␤1-induced Apoptosis of Immortal-
ized NMH-To assess the role of JNKs in TGF-␤1 cell killing of hepatocytes, we treated NMH cells, which undergo apoptosis after 16 -24 h of TGF-␤1 treatment (14) with various doses of the small molecule inhibitor of JNK SP600125 and assessed c-Jun phosphorylation and cell death following TGF-␤1 treatment. At a concentration of 10 M, SP600125 significantly reduced c-Jun phosphorylation in NMH cells treated for 3 h with TGF-␤1 (Fig. 1A). Furthermore, incubation of NMH FIGURE 1. JNK signaling is required for TGF-␤1-mediated apoptosis of NMH cells. A, NMH cells were treated for 3 h with 5 ng/ml TGF-␤1 in the presence of the indicated doses of the JNK inhibitor SP600125 or Me 2 SO control (D). WCEs were prepared, and samples were subjected to immunoblotting using an antibody specific for phospho-Jun, c-Jun, or actin. B, NMH cells were treated for 24 or 48 h with 5 ng/ml TGF-␤1 in the presence or absence of 10 M SP600125. Cell death was determined by TUNEL assay and expressed as percent cell viability relative to total cell number. Means and S.D. represent two independent experiments, each carried out in triplicate (**, p Ͻ 0.01). C, WCEs were prepared from NMH cells treated with either Me 2 SO (ϪSP600125) or 10 M SP600125 with or without 5 ng/ml TGF-␤1 for 48 h. WCEs (50 g) were subjected to caspase-3 activity assay using the CaspACE TM assay system. Caspase-3 specific activity was measured as pmol of p-nitroaniline liberated/h and expressed as fold induction relative to that of BSA carrier solution-treated cells, which was set at 1. Means and standard deviations are representative of three independent experiments carried out in duplicate. D, NMH cells were transfected by lipofection with vectors directing expression of an N-terminally HA-tagged constitutively active JNKK2-JNK1 fusion protein (3HA-JNKK2-JNK-1) or the backbone vector pSR␣. Transfected cells were then treated for 24 h with TGF-␤1, and cell death was determined by TUNEL assay as described above. WCEs (40 g) were subjected to immunoblot analysis using an antibody against HA, phospho-Jun, or actin. IB, immunoblot. NOVEMBER 18, 2005 • VOLUME 280 • NUMBER 46 JOURNAL OF BIOLOGICAL CHEMISTRY 38601 cells with 10 M SP600125 reduced significantly cell death following 24 or 48 h of TGF-␤1 treatment, as measured by TUNEL staining and enzymatic activity of caspase-3 (Fig. 1, B and C). In addition, ectopic expression of a constitutively active JNKK2-JNK1 fusion protein (36) (Fig. 1D) promoted enhanced phosphorylation of c-Jun (bottom panel) and apoptotic cell death (top panel) in the absence of TGF-␤1 stimulation. Lastly, ectopic expression of the JNKK2-JNK1 fusion protein potentiated TGF-␤1-driven apoptosis (Fig. 1D, top panel).

TAK1 Ubiquitination by XIAP
To further assess the role of JNK activity in TGF-␤1 cell killing, we employed the JNK inhibitor 1 (D-stereoisomer) JIP-1 (37). Incubation of NMH cells with 5 M JIP-1, led to inhibition of c-Jun phosphorylation but not Smad2 in TGF-␤1-treated NMH cells ( Fig. 2A). Furthermore JIP-1-treated NMH cells failed to undergo apoptosis, as judged by TUNEL staining and enzymatic activity of caspase-3 (Fig. 2, B and C). Thus, active JNKs play a role in TGF-␤1 cell killing of hepatocytes.
Inhibition of NF-B Activity Potentiates TGF-␤1 Cell Killing of NMH Cells through the JNK-AP-1 Axis-Our previous findings indicated that the early transient activation of NF-B seen in TGF-␤1-treated hepatocytes was opposing TGF-␤1-induced apoptosis through inhibition of JNK signaling (16). Indeed, we observed a more pronounced and persistent phosphorylation of c-Jun in IKK-2 KϾM than GFP-expressing NMH cells (Fig. 3A). This effect was specific for the JNK pathway, because ectopic expression of the IKK-2 KϾM did not affect the levels of Smad2 phosphorylation following TGF-␤1 stimulation (Fig. 3B). Furthermore, IKK-2 KϾM-expressing cells displayed a marked increase in cell death compared with that of GFP-transfected cells, which was blunted in SP600125-treated hepatocytes (Fig. 3C). Thus, JNK signaling is required for the potentiation of TGF-␤1 cell killing via NF-B inhibition.
As XIAP is a known downstream target of NF-B (38) and because our previous data showed that ectopic expression of XIAP rescued NMH cells from TGF-␤1-induced apoptosis (16), we sought to determine the regulation and role of XIAP gene expression during TGF-␤1 signaling. We observed induction of XIAP mRNA expression after 1 h of TGF-␤1 treatment that remained elevated for up to 6 h, as judged by semi-quantitative RT-PCR analyses (Fig. 4A). Furthermore, ectopic expression of IKK-2 KϾM inhibited TGF-␤1-mediated up-regulation of XIAP gene expression (Fig. 4B). Likewise, we observed up-regulation of XIAP protein expression in response to TGF-␤1 treatment (Fig. 4C) that was blocked by ectopic expression of IKK-2 KϾM (Fig. 4D). Furthermore, ectopic expression of a constitutively active mutant IKK-2 S176E/S180E (IKK-2 EE) led to up-regulation of XIAP protein levels (ϳ3-fold), indicating that XIAP is a downstream target of NF-B (Fig.  4E). Given that JNKs are required for TGF-␤1-mediated apoptosis (Figs. 1 and 2), we next asked whether XIAP was counteracting TGF-␤1induced cell death, in part, through inhibition of JNK activity. To address this issue, we silenced XIAP protein expression using SMARTpool TM siRNAs targeted against murine XIAP (siXIAP). In si-control (siC)-transfected cells, we observed increased expression levels of XIAP protein after 1 h of TGF-␤1 treatment, which remained elevated throughout the entire time course (Fig. 5A). In contrast, XIAP protein expression was virtually undetectable in siXIAP-expressing cells, indicating successful knockdown of the XIAP gene (Fig. 5A). In siC-transfected cells, we observed a transient up-regulation of c-Jun phosphorylation levels after 3 h of TGF-␤1 treatment (Fig. 5A). In siXIAPtransfected NMH cells, we observed enhanced and persistent c-Jun phosphorylation in response to TGF-␤1 treatment (Fig. 5A), indicating that XIAP plays a role in NF-B-mediated suppression of the JNK/AP-1 axis. Likewise, we noticed persistent JNK-mediated phosphorylation of a GST-Jun substrate in TGF-␤1-treated XIAP-null NMH cells (Fig. 5B).
Because phosphorylation of c-Jun enhances AP-1 transcriptional activity (39), we sought to determine the effect of XIAP and its functional domains on the transcriptional activation by TGF-␤1 of a luciferase reporter construct driven by four tandem AP-1-binding sites from the collagenase promoter (4XTRE-Lux) (40). We observed a ϳ6-fold induction of 4XTRE-driven luciferase activity in TGF-␤1-treated cells that had been transfected with the empty vector (Fig. 5C). In contrast, ectopic expression of Myc-tagged wild-type XIAP or a deleted mutant of XIAP lacking the BIR domain (XIAP-⌬BIR), which is required for inhibition of caspase-3 (41), reduced significantly the activation of the 4XTRE-Lux construct following TGF-␤1 stimulation (Fig. 5C). Intriguingly, ectopic expression of a deleted mutant XIAP lacking the C-terminal RING domain, which does not display E3 ligase activity (42), did not affect the induction of 4XTRE luciferase activity by TGF-␤1 (Fig. 5C). A, NMH cells were incubated with 1 or 5 M JIP-1 or phosphatebuffered saline control for 3 h and then treated with 5 ng/ml TGF-␤1 for 3 h. WCEs were subjected to immunoblotting using an antibody specific for phospho-Jun, c-Jun, phospho-Smad2, Smad2, or actin. B, NMH cells were treated for 24 or 48 h with 5 ng/ml TGF-␤1 in the presence or absence of 5 M JIP-1. Cell death was determined by TUNEL assay and expressed as percent of cell viability relative to total cell number. Means and standard deviations represent three independent experiments, each carried out in triplicate. Significance was calculated using the Student's t test (**, p Ͻ 0.01). C, WCEs were prepared from NMH cells treated with either PBS (ϪJIP-1) or 5 M JIP-1 with or without 5 ng/ml TGF-␤1 for 48 h. WCEs (50 g) were assayed for caspase-3 activity using the CaspACE ® assay system. Caspase-3 specific activity was measured as pmol p-nitroaniline liberated per hour and expressed as fold induction relative to that of BSA carrier solution-treated cells, which was set at 1. Means and standard deviations are representative of three independent experiments carried out in duplicate. IB, immunoblot.
Furthermore, ectopic expression of wild-type XIAP or XIAP-⌬BIR but not XIAP-⌬RING reduced significantly TGF-␤1-induced activation of the 3TP-Lux construct (Fig. 5D), which contains the Smad-regulated plasminogen activator inhibitor (PAI-1) promoter element and three AP-1 binding sites (25). Thus, ectopic expression of XIAP with an intact ubiquitin ligase activity is sufficient to inhibit TGF-␤1 signaling, presumably, through down-regulation of JNK-mediated phosphorylation of c-Jun. Conversely, inhibition of XIAP gene expression through siRNA-directed knockdown or via inhibition of NF-B activity potentiates TGF-␤1-mediated activation of the JNK-AP-1 signaling pathway and apoptosis.
Active TAK1 Mediates IKK and JNK Activation in TGF-␤1-treated NMH Cells-Our previous studies identified the TGF-␤1-activated kinase 1 (TAK1) as the crucial upstream component bridging the IKK complex to the T␤RI (16). Given the ability of TAK1 to induce the MKK4/JNK axis in other cell types (43)(44)(45), we sought to determine whether TAK1 played a role in the activation of both JNK and IKK pathways in response to TGF-␤1 treatment. We observed an early induction of TAK1-mediated phosphorylation of a GST-MKK4/7 fusion protein following TGF-␤1 treatment of NMH cells (Fig. 6A). Furthermore, ectopic expression of the dominant negative TAK1 (TAK1-K63W) in TGF-␤1-treated NMH cells blocked TAK1-mediated phosphorylation of the GST-MKK4/7 substrate but did not affect Smad2 phosphorylation levels (Fig. 6A). Likewise, ectopic expression of TAK1-K63W blocked TGF-␤1-mediated phosphorylation of the IKK complex (Fig. 6B). To determine the functional impact of TAK1 activity, we measured the transcriptional activation of TGF-␤1-responsive promoters following ectopic expression of the TAK1-K63W mutant. In TAK1-K63W-expressing NMH cells we observed attenuation of TGF-␤1-mediated induction of luciferase activity of both the 4XTRE-Lux and 3TP-Lux reporter genes following TGF-␤1 treatment (Fig. 5, C and  D). Furthermore, we observed that ectopic expression of the TAK1-K63W mutant inhibited the transcriptional induction of both a b-Lux and a 4XTRE-Lux construct following TGF-␤1 treatment or co-expression of a constitutively active form of the T␤RI (T204D) (Fig. 6C). Thus, TAK1 is an upstream kinase that plays a role in the activation of both the JNK and IKK pathways in response to TGF-␤1 signaling.
XIAP Promotes Ubiquitination and Degradation of TAK1 in Response to TGF-␤1 Treatment-Previously, XIAP was found to interact with the type I BMP receptor ALK-3 through association with the TAK1⅐TAB1 complex (30). Thus, we sought to determine whether XIAP associates with TAK1 following TGF-␤1 stimulation, and whether this interaction

. Inhibition of NF-B activity potentiates TGF-␤1-induced apoptosis through persistent activation of the JNK/AP-1 signaling pathway.
A, NMH cells were infected with 10 pfu/ml Ad viral constructs directing expression of the dominant negative IKK-2 KϾM (AdIKK-2 KϾM), or GFP (Ad-GFP). After 24 h, NMH cells were treated for the indicated times with 5 ng/ml TGF-␤1. WCEs were subjected to immunoblotting using an anti-IKK-2, phospho-Jun, c-Jun, or actin antibody. B, WCEs from IKK-2 KϾM or GFP-expressing NMH cells treated with 5 ng/ml TGF-␤1 for the indicated time points were subjected to immunoblotting using antibodies against phospho-Smad2 (pSmad2), total Smad2, or actin. C, NMH cells were infected as described above and treated for 24 h with 5 ng/ml TGF-␤1 in the presence of 10 M SP600125 (SP) or Me 2 SO control (D). Cell death was determined by TUNEL assay as described under "Materials and Methods." IB, immunoblot. , or GFP (Ad-GFP). After 24 h of infection, NMH cells were treated for the indicated times with 5 ng/ml TGF-␤1. XIAP gene expression was determined by semi-quantitative RT-PCR as described above. C, NMH cells were treated for the indicated times with 5 ng/ml TGF-␤1. Samples were subjected to immunoblotting using an antibody specific for either XIAP or actin. D, NMH cells were infected as above described with AdIKK-2 KϾM, or adGFP control. Infected cells were treated for the indicated times with 5 ng/ml TGF-␤1. WCEs were subjected to immunoblotting using antibodies against XIAP or actin. E, NMH cells were transfected with the indicated amount of the pCMV-IKKEE expression vector or with 4 g of the pCMV-Neo control vector. After 24 h, WCEs (40 g) were subjected to immunoblotting using antibodies specific for IKK-2, XIAP, or actin. IB, immunoblot. NOVEMBER 18, 2005 • VOLUME 280 • NUMBER 46 affects TGF-␤1-mediated activation of the JNK pathway. Indeed, the endogenous XIAP formed a transient TGF-␤1-inducible complex with the endogenous TAK1 in NMH cells (Fig. 7A). This observation led us to hypothesize that XIAP might suppress JNK activity through disruption of TAK1 signaling. Given that XIAP has been shown to display ubiquitin-protein ligase activity (E3) (46), we assessed whether XIAP was promoting TAK1 degradation in the proteasome through a ubiquitin-dependent pathway. Indeed, we observed down-regulation of TAK1 protein expression after 6 -12 h of TGF-␤1 treatment of NMH cells and primary rat hepatocytes (Fig. 7, B and C). Moreover, pretreat- FIGURE 5. XIAP mediates the negative cross-talk between the NF-B and JNK pathways. A, NMH cells were transfected with 50 nM siRNA specific for murine XIAP (siXIAP) or scrambled siRNA control (siC). After 24 h, siRNA-transfected cells were treated for the indicated times with 5 ng/ml TGF-␤1. WCEs were subjected to immunoblotting using antibodies against XIAP, phospho-Jun, c-Jun, or actin. B, NMH cells were transfected with siXIAP or siC as described above and WCEs isolated in kinase buffer. WCEs (300 g) were immunoprecipitated (IP) with an antibody against JNK1 and an aliquot was subjected to a kinase assay using GST-Jun as substrate. An equal aliquot of each immunoprecipitant was subjected to immunoblotting, as indicated. C and D, NMH cells were plated in triplicate in 96-well plates and transfected by lipofection with 50 ng of 4XTRE-Lux (C) or 3TP-Lux (D) constructs in the presence or absence of wild-type XIAP (wtXIAP), XIAP-⌬BIR, XIAP-⌬RING, or TAK1-K63W and an internal control Renilla luciferase expression plasmid. Following 6 h of treatment with 5 ng/ml TGF-␤1, luciferase activity was measured and expressed as fold induction relative to that of BSA-treated cells that was set at 1. Means and standard deviations are representative of two independent experiments carried out in triplicate. Inset, NMH cells were transfected for 24 h with Myc-tagged wtXIAP, XIAP-⌬BIR, XIAP-⌬RING. WCEs were subjected to immunoblot analysis using an anti-Myc antibody. IB, immunoblot. FIGURE 6. TGF-␤1 induces both the IKK and JNK signaling pathways. A, NMH cells were transfected with vectors directing expression of N-terminally FLAG-tagged TAK1 (TAK1) or the dominant negative FLAG-TAK1-K63W (TAK1-K63W). After 24 h, transfected cells were treated for the indicated times with 5 ng/ml TGF-␤1. WCEs (300 g) were immunoprecipitated (IP) with an antibody against TAK1, and an aliquot was subjected to a kinase assay using GST-MKK4/7 as substrate. An equal aliquot of each immunoprecipitant was subjected to immunoblotting using antibodies specific for TAK1, FLAG, phospho-Smad2 (pSmad2), Smad2, and actin as indicated. B, NMH cells were transfected and treated as described above. WCEs (40 g) were subjected to immunoblotting using antibodies raised against phospho-IKK1/2 (pIKK1/2), IKK1/2, Flag, or actin, as indicated. C, NMH cells transfected with 50 ng of b-Lux or 4XTRE-Lux construct in the presence or absence of the constitutive active form of the T␤RI T204D and an internal control Renilla luciferase expression plasmid. Cells not expressing the T␤RI T204D construct were stimulated for 6 h with 5 ng/ml TGF-␤1. Luciferase activity was measured and expressed as fold induction relative to that of BSA-carrier solution-treated cells that was set at 1. Means and standard deviations are representative of two independent experiments carried out in triplicate. IB, immunoblot. ment of NMH cells with the proteasome inhibitor MG132 reverted TGF-␤1-mediated down-regulation of TAK1 protein expression (Fig.  7D) indicating that TAK1 is degraded in the proteasome. To determine whether XIAP played a role in TAK1 degradation, we analyzed TAK1 protein expression in siXIAP-treated cells. After 6 -12 h of TGF-␤1 treatment, we observed down-regulation of TAK1 protein in siC-transfected cells (Fig. 7E). In contrast, in siXIAP-treated cells, TAK1 protein expression levels were not reduced by TGF-␤1 treatment (Fig. 7E), indicating that XIAP is required for TAK1 degradation during TGF-␤1 signaling. To determine whether the ubiquitin ligase activity of XIAP played a role during TAK1 degradation in response to TGF-␤1 treatment, we transfected NMH cells with mammalian vectors directing expression of FLAG-tagged TAK1 (pCMV2-Flag-TAK1), His-tagged ubiquitin (pCW7-H 6 M-Ub), in the presence of Myc-tagged XIAP (pCS3-XIAP) or the pCS3 backbone plasmid. Cells were then incubated with the proteasome inhibitor MG132 to inhibit TAK1 degradation. Upon immunoprecipitation of TAK1 with the anti-FLAG antibody and immunoblotting with an anti-ubiquitin antibody, we observed enhanced levels of higher molecular weight species of TAK1 in the FLAG immunoprecipitants of pCS3-XIAP-expressing NMH cells compared with pCS3-transfected cells, which was indicative of covalent ubiquitination of TAK1 protein (Fig. 8A). Furthermore, XIAP-mediated ubiquitination of TAK1 was reduced in cells expressing a mutant version of ubiquitin that contains a K48R mutation (pCW8-H 6 M-Ub) (data not shown), which exerts a dominant negative chain-terminating effect on multiubiquitin chains (47). To determine the domain/s of XIAP involved in TAK1 ubiquitination, we compared levels of covalent ubiquitination of ectopic HA-tagged TAK1 in NMH cells expressing exogenous wild-type XIAP to that of cells transfected with deleted mutants of XIAP lacking either the RING or BIR domain. We observed basal level of ubiquitination of TAK1, presumably because of the E3 ligase activity of the endogenous XIAP (Fig. 8B, lane 1). As expected, ectopic expression of wild-type XIAP led to a significant enhancement of higher molecular species recognized by the anti-ubiquitin antibody (lane 2). As predicted by our previous data, we observed further accumulation of ubiquitinated TAK1 species in cells expressing the XIAP-⌬BIR deleted mutant (lane 4) but not in cells transfected with the XIAP-⌬RING constructs (lane 3). Thus, the RING domain of XIAP plays a potential role in Lys 48 polyubiquitination of TAK1, which then targets this molecule for proteosomal degradation.

TAK1 Ubiquitination by XIAP
XIAP Protects Hepatocytes from TGF-␤1-induced Apoptosis through Inhibition of Both Caspase and JNK Activities-To determine the contribution of XIAP to cell survival during TGF-␤1-induced cell death, we assessed cell viability in TGF-␤1-treated NMH cells following silencing of XIAP gene expression. After 16 h of TGF-␤1 treatment of siC-treated cells, we observed a modest induction of cell death (Fig. 9A). In contrast, siXIAP-transfected cells displayed enhanced basal levels of cell death (ϳ15%), which was dramatically induced (ϳ50%) at the 16 h time point (Fig. 9A). Enhanced cell death in siXIAP-treated cells was accompanied by a faster and more pronounced accumulation of the 89-kDa cleaved fragment of the poly (ADP-ribose) polymerase (PARP) gene product compared with siC-transfected cells (Fig. 9B). Thus, silencing of the XIAP gene product accelerates TGF-␤1-induced apoptosis, presumably, through release of JNK activity.
To determine which domain of XIAP was involved in the rescue from TGF-␤1 cell killing, we compared cell survival of TGF-␤1-treated NMH cells expressing ectopic wild-type XIAP to that of cells transfected with deleted mutants of XIAP lacking either the RING or BIR domain. The levels of protein expression of the various transgenes were determined by immunoblotting using an anti-Myc antibody (data not shown) and were found to be similar to those depicted in Fig. 4C. In backbone vector-transfected cells, we observed enhanced PARP cleavage following 16 h of TGF-␤1 treatment (Fig. 9C). In contrast, ectopic expression of wild-type XIAP prevented the accumulation of the 89-kDa PARP fragment (Fig. 9C). Likewise, ectopic expression of either the XIAP-⌬BIR or the XIAP-⌬RING mutants inhibited PARP cleavage, indicating that both domains are involved in rescue from TGF-␤1-induced cell death presumably through inhibition of either JNK or caspase activities.

DISCUSSION
Here, we have identified a novel signaling pathway that plays a role in the regulation of the negative cross-talk between NF-B and JNKs dur- FIGURE 7. XIAP promotes TAK1 degradation in response to TGF-␤1 treatment. A, NMH cells were treated for the indicated times with 5 ng/ml TGF-␤1. WCEs (300 g) were immunoprecipitated (IP) with a monoclonal anti-XIAP antibody or IgG control. Immunoblotting of equal immunoprecipitated aliquots was performed using an anti-TAK1 or an anti-XIAP polyclonal antibody. B and C, WCEs of TGF-␤1-treated NMH cells (B) or primary cultures of rat hepatocytes (C) were subjected to immunoblotting using an antibody specific for either TAK1 or actin. D, NMH cells were treated for 12 h with 5 ng/ml TGF-␤1 alone or in combination with 20 M proteasome inhibitor MG132. WCEs were subjected to immunoblotting as described above. E, NMH cells were transfected with 50 nM siRNA specific for murine XIAP (siXIAP) or 50 nM nonspecific siRNA control (siC). Si-transfected cells were then treated with TGF-␤1 for the indicated times. WCEs were subjected to immunoblotting using antibodies against TAK1, XIAP, or actin. IB, immunoblot.  NOVEMBER 18, 2005 • VOLUME 280 • NUMBER 46 ing TGF-␤1-induced apoptosis in murine hepatocytes. We show that active NF-B plays a role during induction of XIAP gene expression following TGF-␤1 treatment of NMH cells. This event results in enhanced association of XIAP with the upstream TAK1 kinase. Furthermore, we find that XIAP promotes ubiquitin-mediated degradation of TAK1 thereby leading to the shutdown of JNK signaling (Fig. 10). Conversely, silencing of the XIAP gene blocks TAK1 degradation by TGF-␤1 thereby leading to persistent and enhanced JNK activation. Consistent with the requirement of JNK signaling for TGF-␤1-induced apoptosis, siXIAP-transfected cells displayed enhanced apoptosis. Last, both the XIAP-⌬BIR and XIAP-⌬RING mutants were able to protect from TGF-␤1-induced apoptosis, suggesting that the anti-apoptotic activity of XIAP is not restricted solely to the inhibition of caspase activity but also involves attenuation of JNK signaling through ubiquitininduced degradation of TAK1.

TAK1 Ubiquitination by XIAP
The negative cross-talk between the NF-B and JNK pathway was first described in the TNF-␣ signaling pathway where two NF-B target genes, GADD45␤/MyD118 and XIAP, appeared to play a role in the shutdown of JNK signaling after TNF-␣ stimulation (48). However, the exact mechanism of JNK suppression by these two genes has remained highly controversial (49). Indeed, genetic disruption of XIAP or GADD45␤/MyD118 did not alter activation of JNK or cell death in response to TNF-␣ stimulation (50,51), suggesting that other factors might be compensating for the lack of one or both genes. In addition, inhibition of NF-B during TNF-␣ signaling has been shown to decrease expression levels of radical scavengers thereby promoting accumulation of reactive oxygen species (ROS) (52). Increased presence of ROS in these TNF-␣-treated cells has been recently shown to promote sustained JNK activation and necrotic cell death through reversible oxidation of mitogen-activated protein kinase phosphatases (53).
With regard to the TGF-␤1 signaling pathway, we have shown that inhibition of NF-B activity in TGF-␤1-treated NMH cells led to pronounced activation of JNK-mediated phosphorylation of Jun and to enhancement of AP-1/Smads-mediated apoptosis (16). Whereas we ruled out the MyD118 gene because of previously reported pro-apoptotic effects during TGF-␤1 signaling (54,55), we proposed XIAP as one of the candidate factors involved in JNK suppression. Here, we show that the XIAP gene product is up-regulated following TGF-␤1 treatment and plays a role in the shutdown of JNK signaling through ubiquitin-mediated degradation of TAK1.
Our results contradict other findings published in the literature showing that XIAP acts as a cofactor of BMP and TGF-␤1 signaling in different cell types (30,56). In addition, others have found that overexpression of XIAP activates the JNK1 signaling pathway through TAK1 thereby protecting 293 or MCF7 cells from apoptosis in response to TNF-␣, ICE, or FAS-induced apoptosis (57,58). Although these studies lacked a rigorous genetic analysis of the role of XIAP in each specific signaling pathway, the differences between our results and these findings could be potentially reconciled by a cell type/signal-dependent function of XIAP (56,59). Furthermore, it is noted that in those cell models JNK activity was shown to counteract TNF-␣or Fas-induced apoptosis, whereas in the TGF-␤1 signaling pathway, JNK activity is required for TGF-␤1-induced apoptosis (16,25,60,61). Intriguingly, XIAP has been shown to induce NF-B activity in endothelial cells through TAK1 in what appeared to be a positive regulatory loop (59). However, despite several technical efforts, we were unable to observe activation of a b-driven luciferase reporter construct upon ectopic expression of XIAP in NMH cells (data not shown). Thus, it appears that XIAP displays context-specific functions depending on its ability to affect the kinase activity of TAK1. If XIAP can indeed work as a cofactor of TGF-␤1 signaling in other cell types or in response to other stimuli, it will be interesting to assess whether the ubiquitin ligase activity of XIAP can promote the activating Lys 63 versus the inhibitory Lys 48 polyubiquitination at the level of upstream regulatory kinases (e.g. TAK1) in a context-and signal-dependent fashion. Nonetheless, our data clearly demonstrate a novel antiapoptotic role of XIAP through attenuation of the TAK1/JNK signaling pathway in rodent hepatocytes. This observation is in line with a previous report by Park et al. (45) showing that the FIGURE 9. XIAP protects from TGF-␤1-induced apoptosis through both its RING and BIR domains. A, NMH cells were transfected with 50 nM siRNA specific for murine XIAP (siXIAP) or nonspecific siRNA control (siC). After 24 h, siRNA-transfected cells were treated for 6 h with 5 ng/ml TGF-␤1 or BSA-carrier solution as control, and cell death was determined by colorimetric TUNEL assay. Cell death was expressed as the percent of TUNELpositive cells relative to the total number of cells in each representative field. Means and standard deviations are representative of three independent experiments carried out in triplicate. B, NMH cells were transfected with siXIAP or siC as described above. After 24 h, siRNA-transfected cells were treated for the indicated times with 5 ng/ml TGF-␤1. WCEs were subjected to immunoblotting using antibodies specific for XIAP, the 89 kDa cleaved form of PARP, or actin. C, NMH cells were transfected for 24 h with Myc-XIAP, Myc-XIAP-⌬BIR, and Myc-XIAP-⌬RING. WCEs were subjected to immunoblot analysis using an antibody against the 89-kDa cleaved form of PARP or actin. IB, immunoblot. FIGURE 10. Model for the negative cross-talk between the NF-B and JNK signaling pathways during TGF-␤1-induced apoptosis. In response to T␤RI signaling, TAK1 activates both IKKs and JNKs. Activation of JNKs promotes AP-1/Smads-mediated apoptosis. In contrast, activation of the IKK/NF-B axis results in protection from TGF-␤1 cell killing through regulation of XIAP gene expression. The proposed mechanism of this negative cross-talk between the NF-B and JNK pathways relies on the ability of XIAP to promote ubiquitin-mediated degradation of TAK1 in the 26 S proteasome thereby leading to the shutdown of JNK activity and rescue from cell death.
NF-B protein Relish, which mediates the transcriptional induction of antimicrobial peptides during the innate immune response in Drosophila (62), plays a role in the shutdown of JNK signaling through proteosomal-mediated degradation of TAK1 (45). Thus, it is tempting to speculate that attenuation of JNK signaling by NF-B/Rel family members through inactivation of TAK1 might constitute an evolutionary conserved mechanism involved in the regulation of inflammation, immune response, and homeostasis. In this regard, it will be crucial to determine whether the Drosophila homolog of XIAP or other E3 ligases mediate degradation of TAK1 in response to Gram-negative infections in Drosophila, and whether XIAP plays a role in the initiation of the innate immune response in mammals.
The role of c-Jun during cell survival of hepatocytes appears to be signal-specific. Deletion of c-Jun in the hepatic compartment has been shown to sensitize hepatocytes to TNF-␣-induced apoptosis thereby impairing chemically-induced tumor formation at early stages (63,64). The mechanism of the anti-apoptotic role of c-Jun during liver neoplastic progression was shown to be directly linked to its ability to antagonize p53 activity (63). Intriguingly, inactivation of c-Jun in advanced HCCs did not affect tumor formation probably because of the lack of a functional p53 protein (63). In contrast, we and others have shown that c-Jun is required for TGF-␤1-mediated apoptosis of hepatocytes (16,25). Indeed, ablation of c-Jun rendered primary hepatocytes resistant to TGF-␤1 cell killing (63). Thus, it is tempting to speculate that constitutive activation of NF-B might render malignant hepatocytes that retain both a functional p53 and an intact TGF-␤1 signaling pathway more resistant to TGF-␤1-induced apoptosis through suppression of the JNK/AP-1 signaling pathway. Consistent with this hypothesis, chronic hepatic overexpression of TGF-␤1 in mice promotes the development of hepatocellular carcinomas (HCCs) (27), which are typified by constitutive NF-B activity and reduced levels of JNK activity (16).
Collectively, our findings indicate that XIAP along with the Smad ubiquitination-related factor (Smurf1), Smurf2, and SCF/Roc1 (65) constitutes a novel class of molecules typified by E3 ligase activity that confers resistance to the antiproliferative and pro-apoptotic action of TGF-␤1 through ubiquitin-mediated proteosomal degradation of crucial components of the TGF-␤1 signaling pathway. Consequently, the use of polyphenylurea-based compounds that have been proven effective in inhibiting XIAP activity and promoting cell death of myeloid and epithelial cancers (66, 67) could potentially constitute a valid approach to sensitize non-resectable human HCCs to the growth inhibitory properties of TGF-␤1 without affecting normal liver homeostasis.