X-linked Inhibitor of Apoptosis Protein and Its E3 Ligase Activity Promote Transforming Growth Factor-β-mediated Nuclear Factor-κB Activation during Breast Cancer Progression*

The precise sequence of events that enable mammary tumorigenesis to convert transforming growth factor-β (TGF-β) from a tumor suppressor to a tumor promoter remains incompletely understood. We show here that X-linked inhibitor of apoptosis protein (xIAP) is essential for the ability of TGF-β to stimulate nuclear factor-κB (NF-κB) in metastatic 4T1 breast cancer cells. Indeed whereas TGF-β suppressed NF-κB activity in normal mammary epithelial cells, those engineered to overexpress xIAP demonstrated activation of NF-κB when stimulated with TGF-β. Additionally up-regulated xIAP expression also potentiated the basal and TGF-β-stimulated transcriptional activities of Smad2/3 and NF-κB. Mechanistically xIAP (i) interacted physically with the TGF-β type I receptor, (ii) mediated the ubiquitination of TGF-β-activated kinase 1 (TAK1), and (iii) facilitated the formation of complexes between TAK1-binding protein 1 (TAB1) and IκB kinase β that enabled TGF-β to activate p65/RelA and to induce the expression of prometastatic (i.e. cyclooxygenase-2 and plasminogen activator inhibitor-1) and prosurvival (i.e. survivin) genes. We further observed that inhibiting the E3 ubiquitin ligase function of xIAP or expressing a mutant ubiquitin protein (i.e. K63R-ubiquitin) was capable of blocking xIAP- and TGF-β-mediated activation of NF-κB. Functionally xIAP deficiency dramatically reduced the coupling of TGF-β to Smad2/3 in NMuMG cells as well as inhibited their expression of mesenchymal markers in response to TGF-β. More importantly, xIAP deficiency also abrogated the formation of TAB1·IκB kinase β complexes in 4T1 breast cancer cells, thereby diminishing their activation of NF-κB, their expression of prosurvival/metastatic genes, their invasion through synthetic basement membranes, and their growth in soft agar. Collectively our findings have defined a novel role for xIAP in mediating oncogenic signaling by TGF-β in breast cancer cells.

uitin ligase activity (10,11). For instance, increased xIAP expression activates NF-B, whereas the expression of xIAP mutants that lack E3 ubiquitin ligase activity fails to activate the NF-B pathway (10,11). Moreover elevated xIAP expression occurs during cancer progression in a manner that correlates with the acquisition of metastatic phenotypes in cancer cells (12)(13)(14). Thus, additional investigations into the role of xIAP in mediating cancer progression appear warranted, and as such, xIAP and other E3 ubiquitin ligases currently are being interrogated as potential chemotherapeutic targets in human cancers (15,16). Along these lines, it remains to be determined whether xIAP and its E3 ubiquitin ligase activity play an essential role in coupling TGF-␤ to NF-B activation during breast cancer progression. The goal of this study was to address this important question and to determine how altered xIAP expression impacts normal and malignant MEC response to TGF-␤.
Cell Culture and Transgene Expression-Normal murine NMuMG mammary gland and malignant, metastatic 4T1 breast cancer cells were obtained from American Type Culture Collection (Manassas, VA) and cultured as described previously (6). Wild-type (WT) mouse embryonic fibroblasts (MEFs) and xIAP-deficient MEFs kindly provided by Dr. Tullia Lindsten (University of Pennsylvania, Philadelphia, PA) were cultured in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and 1% penicillin and were maintained in a constant atmosphere of 5% CO 2 at 37°C.
The following murine ecotropic retroviral constructs were used herein: (i) pMSCV-IRES-YFP (i.e. control vector) (6) and (ii) pMSCV-xIAP-YFP. Retroviral supernatants were produced by EcoPack2 retroviral packaging cells (Clontech) and used to infect NMuMG and 4T1 cells (6). Forty-eight hours postinfection, the transduced cells were analyzed and isolated on a MoFlo cell sorter (Beckman Coulter, Fullerton, CA) and subsequently were expanded to yield stable polyclonal populations of control and transgene-expressing cells that were Ն90% for expression of YFP.
The creation of NMuMG and 4T1 cells lacking xIAP was accomplished by their overnight infection with control (i.e. non-silencing shRNA) or xIAP shRNA lentiviral (pLKO.1puro) supernatants produced by 293T cells that were transiently transfected with lentiviral packaging vectors (i.e. pMD2.G, pRRE, and pRSV) according to standard protocols (17). Cells expressing non-silencing and xIAP shRNAs were isolated by puromycin selection (5 g/ml) for 14 days. Afterward puromycin-resistant NMuMG and 4T1 clones were isolated, and the extent of shRNA-mediated xIAP deficiency was monitored by immunoblotting whole cell extracts with antibodies against xIAP as described below.
Luciferase Assays-Analysis of TGF-␤-stimulated luciferase activity driven by the synthetic NF-B, SBE, and 3TP promoters was performed as described previously (18). Briefly NMuMG and 4T1 MECs (25,000 -35,000 cells/well) were cultured overnight onto 24-well plates and subsequently were transfected the following morning by overnight exposure to LT1 liposomes (Mirus, Madison, WI) containing 350 ng/well individual luciferase reporter cDNA and 50 ng/well CMV-␤-gal cDNA, which was used to control for differences in transfection efficiency. Afterward the cells were washed twice with phosphate-buffered saline and stimulated overnight in serum-free medium with TGF-␤1 (5 ng/ml). The following morning, luciferase and ␤-galactosidase activities contained in detergent-solubilized cell extracts were determined.
Western Blotting Analyses-Control and TGF-␤-stimulated NMuMG and 4T1 cells were lysed and solubilized in Buffer H/Triton X-100 (23) for 30 min on ice. Clarified cell extracts were resolved by 10% SDS-PAGE gels, transferred electrophoretically to nitrocellulose membranes, and blocked in 5% milk prior to incubation with the following primary antibodies xIAP Alters TGF-␤ Signaling xIAP (1:500; Cell Signaling Technology), and (q) anti-ubiquitin (1:1000; Cell Signaling Technology). The resulting immunocomplexes were visualized by enhanced chemiluminescence. Differences in protein loading were monitored by reprobing stripped membranes with anti-␤-actin antibodies.
TAB1⅐IKK␤ Co-immunoprecipitation Assays-Detergentsolubilized NMuMG whole cell extracts (350,000 cells/tube) were prepared and incubated under continuous rotation with anti-TAB1 antibodies (2 l/tube; Santa Cruz Biotechnology) for 6 h at 4°C. The resulting immunocomplexes were collected by microcentrifugation, washed, and fractionated by 10% SDS-PAGE prior to their immobilization to nitrocellulose membranes, which subsequently were probed with anti-IKK␤ antibodies (1:500). Differences in protein loading were monitored by immunoblotting whole cell extract aliquots with antibodies against ␤-actin as above.
NF-B Biotinylated Oligonucleotide Capture Assays-The DNA binding activity of NF-B was monitored in NMuMG cells that expressed either YFP or xIAP as indicated. NF-B binding activity was determined by continuously rotating 100 g of whole cell extract with 1 g of biotinylated doublestranded DNA oligonucleotides that contained an NF-B consensus sequence site at 4°C (forward probe, 5Ј-GATCTAGG-GACTTTCCGCTGGGGACTTTCCAGTCGA; reverse probe, 5Ј-TCGACTGGAAAGTCCCCAGCGGAAAGTCCCTAG-ATC). The resulting NF-B⅐oligonucleotide complexes were captured by streptavidin-agarose beads (Pierce) and microcentrifugation. Washed complexes were fractionated by 10% SDS-PAGE prior to their immobilization to nitrocellulose membranes, which subsequently were probed with anti-p65 antibodies (1:500). Differences in extract loading were monitored by immunoblotting 25 g of resolved extract aliquots with antibodies against ␤-actin (1:1000).
Semiquantitative Real Time PCR Analyses-Total RNA from control and xIAP-deficient NMuMG and 4T1 cells was purified using the RNeasy Plus Mini kit (Qiagen, Valencia, CA) according to the manufacturer's recommendations. Afterward cDNAs were synthesized by iScript reverse transcription (Bio-Rad) that then were diluted 10-fold in H 2 O and used in semiquantitative real time PCRs (25 l) that used the SYBR Green system (Bio-Rad) supplemented with 5 l of diluted cDNA and 0.1 M oligonucleotide pairs listed below. PCRs were performed and analyzed on a Bio-Rad Mini-Opticon detection system, and differences in RNA concentrations were controlled by normalizing individual gene signals to their corresponding glyceraldehyde-3-phosphate dehydrogenase signals. The oligonucleotide primer pairs used were as follows (NCBI reference numbers in parentheses): Statistical Analyses-Statistical values were defined using an unpaired Student's t test where a p value Ͻ0.05 was considered significant.

Increased xIAP Expression Alters MEC Response to TGF-␤-
Elevated xIAP expression has been linked to the enhanced survival of cancer cells confronted with chemotherapeutic or apoptotic signals (15,24). Accordingly we found that culturing non-tumorigenic NMuMG cells in suspension for 24 h over poly(2-hydroxyethyl methacrylate) reduced their expression of xIAP while simultaneously inducing their cleavage of poly-(ADP-ribose) polymerase and caspase-3 (Fig. 1A). In stark contrast, overnight suspension of malignant, metastatic 4T1 cells actually elicited a dramatic increase in xIAP expression as well as a significant reduction in the cleavage of poly(ADP-ribose) polymerase and caspase-3 (Fig. 1A). Thus, elevated xIAP FIGURE 1. Increased xIAP expression alters MEC response to TGF-␤. A, cell detachment activated apoptotic signaling and reduced xIAP expression in NMuMG cells. In contrast, this same cellular condition suppressed apoptotic stimuli and elevated xIAP expression in metastatic 4T cells. Data are from a representative experiment that was performed three times with similar results. B, stable xIAP expression in NMuMG cells significantly potentiated basal and TGF-␤1 (5 ng/ml)-stimulated pSBE-luciferase activity. Data are the mean (n ϭ 3) luciferase activities relative to untreated NMuMG-YFP cells (*, p Ͻ 0.05; Student's t test). C, stable xIAP expression in NMuMG cells significantly enhanced basal NF-B-driven luciferase activity. In contrast to YFPexpressing cells, TGF-␤1 (5 ng/ml) treatment of xIAP-expressing cells increased their expression of luciferase driven by NF-B. Data are the mean (n ϭ 3) luciferase activities relative to untreated NMuMG-YFP cells (*, p Ͻ 0.05; Student's t test). D, detergent-solubilized whole-cell extracts prepared from NMuMG-YFP and NMuMG-xIAP cells were immunoprecipitated (IP) with anti-TAB1 antibodies and immunoblotted for IKK␤ as shown, and whole-cell extracts were immunoblotted with antibodies against TAK1, TAB1, xIAP, IKK␤, and ␤-actin as indicated (left panels). Additionally whole-cell extracts from YFP-or xIAP-expressing cells were incubated with biotinylated NF-B oligonucleotides to isolate active p65 (p65 oligo; right panels). Images are from a representative experiment that was performed three times with identical results. PARP, poly(ADP-ribose) polymerase. The error bars indicate Ϯ S.E. xIAP Alters TGF-␤ Signaling AUGUST 7, 2009 • VOLUME 284 • NUMBER 32 expression in malignant MECs correlates with their acquisition of resistance to anoikis-induced cell death.
In addition, elevated xIAP expression also has been associated to the activation of both the TGF-␤ and NF-B signaling systems; however, the molecular mechanisms that mediate these functions of xIAP remain relatively undefined. Consistent with previous reports (10, 11), we too observed elevated xIAP expression in NMuMG cells to increase both basal and TGF-␤stimulated Smad2/3-driven luciferase reporter activity (Fig.  1B). In addition, whereas TGF-␤ stimulation repressed NF-Bdriven luciferase activity in control (i.e. YFP) NMuMG cells, applying this same treatment condition to their xIAP-expressing counterparts enabled TGF-␤ to activate NF-B in NMuMG cells (Fig. 1C). We previously established the essential role of TAB1⅐IKK␤ complexes to promote the activation of NF-B in MECs stimulated with TGF-␤ (6). We now show that elevating xIAP expression was sufficient to induce the formation of these TAB1⅐IKK␤ complexes (Fig. 1D, left panels) and consequently their activation of NF-B (Fig. 1D, right panels). Thus, increased expression of xIAP alters TGF-␤ signaling and results in its conversion from an inhibitor to a stimulator of NF-B in MECs in part via the formation of TAB1⅐IKK␤ complexes.
xIAP Interacts with T␤R-I and Enables Autocrine TGF-␤ Signaling to NF-B in Malignant MECs-We (6) and others (25) have shown that TGF-␤ signaling typically represses NF-B activity in non-tumorigenic MECs but readily activates this transcription factor in their malignant, metastatic counterparts. Accordingly transient expression of a constitutively active T␤R-I (i.e. T204D-T␤R-I) in 4T1 cells significantly induced their expression of luciferase driven by the synthetic NF-B promoter ( Fig. 2A). Thus, in addition to mediating activation of canonical Smad2/3 signaling, T␤R-I also couples TGF-␤ to the stimulation of NF-B in malignant MECs. Because T␤R-I has been shown to interact physically with xIAP (10), we also wished to determine whether T204D-T␤R-I could form stable complexes with xIAP. Fig. 2B shows that T204D-T␤R-I did indeed interact physically with co-expressed xIAP. Moreover TGF-␤ treatment of 4T1 cells readily elicited xIAP binding to T␤R-I, an interaction that was reduced significantly by expression of an shRNA against xIAP in 4T1 cells (Fig. 2C). In addition, administration of a T␤R-I antagonist to NMuMG or 4T1 cells revealed that both cell lines are subjected to extensive autocrine TGF-␤ signaling that governs regulation of the NF-B pathway, namely suppression of NF-B in normal MECs and activation of this transcription factor in their malignant counterparts (Fig. 2D). Along these lines, this same T␤R-I antagonist also decreased the ability of autocrine TGF-␤ signaling to stimulate TAK1 phosphorylation in 4T1 cells, suggesting that the activation of NF-B by TGF-␤ in malignant MECs proceeds through a T␤R-I-and TAK1-dependent mechanism (Fig. 2E). Based on the established role of NF-B to promote the survival of breast cancer cells, we predicted that uncoupling TGF-␤ signaling through TAK1 would diminish the growth and survival of breast cancer cells in soft agar. Accordingly we observed stable expression of kinase-dead K63M-TAK1 in malignant, metastatic human MCF10A-CA1a breast cancer cells to reduce their growth in soft agar by 53 Ϯ 9% (n ϭ 3; p ϭ 0.005), whereas similar inactivation of TAK1 in 4T1 cells also tended to decrease their anchorage-independent growth by 56 Ϯ 20% (n ϭ 2; p ϭ 0.105). Thus, the formation of T␤R-I⅐xIAP complexes promotes autocrine TGF-␤ activation of NF-B in malignant MECs and presumably enhances their ability to grow and survive under adverse adhesive conditions. xIAP-deficient Murine Embryonic Fibroblasts Exhibit Reduced TGF-␤ Signaling-To further explore the functional association of xIAP to TGF-␤ signaling in responsive cells, we obtained MEFs prepared from xIAP-deficient mouse embryos  ). B, human 293T cells were transiently transfected either with xIAP or with xIAP together with T204D-T␤R-I as indicated. Afterward detergent-solubilized whole-cell extracts were prepared and immunoprecipitated (IP) with anti-HA antibodies followed by immunoblotting with antibodies against xIAP or T␤R-I as shown. Differences in protein loading were monitored by reprobing stripped membranes with ␤-actin antibodies. Images are from a representative experiment that was performed three times with identical results. C, control (i.e. non-silencing (Non-Sil)) or xIAP-deficient (i.e. xIAP shRNA) 4T1 cells were stimulated with TGF-␤1 (5 ng/ml) for 0 -15 min as indicated and subsequently immunoprecipitated with anti-xIAP antibodies followed by immunoblotting for T␤R-I as shown. Additionally whole-cell extracts also were immunoblotted for xIAP, T␤R-I, and ␤-actin as indicated. Data are from a representative experiment that was performed at least three times with similar results. D, NMuMG and 4T1 cells were transiently transfected overnight with pNF-B-luciferase and pCMV-␤-gal prior to administration of a T␤R-I antagonist (100 ng/ml) for 24 h as indicated. Data are the mean (n ϭ 3) luciferase activities relative to untreated NMuMG cells (*, p Ͻ 0.05; Student's t test). E, administration of a T␤R-I antagonist (100 ng/ml for 24 h) to 4T1 cells decreased the phosphorylation and activation of TAK1 as determined by immunoblotting for phospho-TAK1 (pTAK1). Differences in protein loading were monitored by reprobing stripped membranes with TAK1 and ␤-actin antibodies as shown. Data are from a representative experiment that was performed three times with similar results. Inh, inhibitor. The error bars indicate Ϯ S.E. (26,27) and monitored their ability to activate Smad2/3, NF-B, and TAK1. These analyses showed that MEFs lacking expression of xIAP have significantly reduced capacity to induce the transcriptional activity (Fig. 3A) and phosphorylation (Fig. 3B) of Smad2/3, whose overall rate of degradation appeared to be enhanced by xIAP deficiency (Fig. 3B). In addition, the magnitude of TGF-␤-mediated stimulation of NF-B transcriptional activity was severely impaired in xIAP-deficient MEFs as compared with their normal counterparts (Fig. 3C) presumably because of the inability of TGF-␤ to activate TAK1 in xIAP-deficient MEFs (Fig. 3D). Collectively these findings complement those of Fig. 1, which demonstrated that up-regulated xIAP expression enhanced NMuMG cell activation of Smad2/3 and NF-B. Extending these findings to MECs suggests that rendering MECs deficient in xIAP may reduce the oncogenic activities of TGF-␤ (see below).
xIAP Ubiquitinates TAK1 and Facilitates Its Interaction with IKK␥/NEMO and Activation of NF-B-We showed recently that TAB1⅐IKK␤ complexes require TAK1 activity to mediate NF-B activation in response to TGF-␤ (6). Based on these and our above findings, we evaluated the effects of expressing WT or mutant xIAP that lacked E3 ligase activity (i.e. H467A-xIAP) on the ability of TAK1 to interact with IKK␥/NEMO. To do so, 293T cells were transiently transfected with FLAG-tagged NEMO and HAtagged TAK1 together with either the aforementioned WT or mutant xIAP. Afterward TAK1 immunocomplexes were isolated and probed for the presence of NEMO. As shown in Fig. 4A, WT xIAP expression, but not that of its E3 ligasedeficient counterpart, facilitated the interaction between TAK1 and IKK␥/NEMO. Moreover Fig. 4B shows that the expression of WT xIAP induced the ubiquitination of endogenous TAK1, a post-translational modification that was not recapitulated by expression of H467A-xIAP. Interestingly recent evidence indicates that Lys-48linked ubiquitin moieties promote protein degradation, whereas Lys-63-linked ubiquitin moieties mediate protein-protein interactions by molecules housing ubiquitin-binding domains (i.e. IKK␥/NEMO (28)). Because TRAF signaling ubiquitinates TAK1 and facilitates its interaction with IKK (29), we speculated that xIAP-mediated ubiquitination of TAK1 (Fig. 4B) also would facilitate a similar interaction within the oncogenic TGF-␤ signaling pathway, leading to its activation of NF-B. As such, we observed xIAP expression to clearly enhance the activation of NF-B in 293T cells, a signaling event that was abrogated by co-expression of K63R-ubiquitin mutants (Fig. 4C). More importantly, expression of K63Rubiquitin mutants completely converted T204D-T␤R-I from a stimulator to an inhibitor of NF-B activity in malignant, metastatic 4T1 cells (Fig. 4D). Collectively these findings suggest that the ability of TGF-␤ to couple to NF-B is dependent upon xIAP-mediated ubiquitination of TAK1, which enables its association with TGF-␤ receptors and components of the IKK complex. Moreover our findings also identify xIAP and its E3 ligase activity as an essential switch that is operant in converting TGF-␤ from an inhibitor to a stimulator of NF-B activity.
xIAP Deficiency Alters TGF-␤ Signaling and Its Induction of Invasion and Mesenchymal Gene Expression in MECs-Altered expression of various scaffolding proteins has been associated with defects in the TGF-␤ signaling system (1). In addition to its E3 ligase activity, xIAP also serves as a molecular scaffold for TAB1 and TAK1 (8); however, whether the expression and scaffolding function of xIAP are essential for TGF-␤ signaling in MECs is unknown. To address this question and to test the  test). B, quiescent WT and xIAPdeficient (xIAP Ϫ/Ϫ ) MEFs were incubated in the absence or presence of TGF-␤1 (5 ng/ml) as indicated and subsequently immunoblotted with anti-phospho-Smad3 (p-Smad3) antibodies as shown. Differences in protein loading were monitored by reprobing the stripped membranes with antibodies against Smad3 and ␤-actin as indicated. Data are from a representative experiment that was performed at least three times with similar results. C, WT and xIAP-deficient (xIAP Ϫ/Ϫ ) MEFs were transiently transfected with pNF-B-luciferase and ␤-galactosidase and subsequently processed for determination of luciferase and ␤-galactosidase activities as above. Data are the mean (n ϭ 3) luciferase activities relative to untreated WT MEFs (*, p Ͻ 0.05; Student's t test). D, quiescent WT and xIAP-deficient (xIAP Ϫ/Ϫ ) MEFs were stimulated with TGF-␤1 (5 ng/ml) as indicated and subsequently immunoprecipitated (IP) with anti-TAK1 antibodies followed by immunoblotting for phospho-TAK1 (p-TAK1) as shown. Additionally whole-cell extracts also were immunoblotted for TAK1 and ␤-actin as indicated. Data are from a representative experiment that was performed at least three times with similar results. The error bars indicate Ϯ S.E. AUGUST 7, 2009 • VOLUME 284 • NUMBER 32

JOURNAL OF BIOLOGICAL CHEMISTRY 21213
hypothesis that rendering MECs deficient in xIAP may reduce the oncogenic activities of TGF-␤, we infected NMuMG cells with lentivirus encoding for either control (i.e. non-silencing) or xIAP shRNA. In accord with our hypothesis, xIAP deficiency significantly inhibited the ability of TGF-␤ to induce Smad2/3driven luciferase expression in NMuMG cells (Fig. 5A) but simultaneously enhanced their repression of NF-B activity when stimulated by TGF-␤ (Fig. 5B). It is interesting to note that although elevated xIAP expression had no effect on the ability of TGF-␤ to down-regulate E-cadherin expression in NMuMG cells this same cellular condition significantly enhanced the stimulation of Cox-2 expression by TGF-␤ (Fig.  5C). Because we recently established Cox-2 as an essential mediator of EMT induced by TGF-␤ (30), we predicted that xIAP deficiency would diminish the ability of TGF-␤ to induce EMT in NMuMG cells. Contrary to our expectations, xIAP deficiency had little effect on the acquisition of fibroblastoid morphologies (Fig. 5D) and down-regulation of E-cadherin expression (Fig. 5E) in NMuMG cells stimulated with TGF-␤. Interestingly the ability of TGF-␤ to induce the expression of the mesenchymal markers Cox-2 and N-cadherin was significantly impaired in NMuMG cells lacking xIAP as compared with their control counterparts (Fig. 5E). Taken together, these findings point to a potentially important bifurcation in the TGF-␤ signaling system that dissociates its regulation of epithelial gene expression profiles from its ability to activate xIAP, which selectively promotes the acquisition of mesenchymal gene expression profiles.
xIAP Deficiency Impairs the Activation of NF-B and Tumorigenic Behavior of Malignant MECs-We showed previously that expression of a truncated TAB1 mutant (i.e. TAB1(411)) uncoupled TGF-␤ from regulating NF-B activity in NMuMG cells and inhibited TGF-␤ stimulation of NF-B in 4T1 cells (6). Because elevated expression of xIAP was sufficient to drive the formation of TAB1⅐IKK␤ complexes and their activation of NF-B in NMuMG cells (Fig. 1), we reasoned that depleting xIAP expression in 4T1 cells would prevent their activation of NF-B as well as diminish their tumorigenic behavior. To explore the merits of this supposition, we again utilized a lentivirus-based shRNA expression system that significantly depleted the expression of xIAP in 4T1 cells (Fig. 6A). Functionally xIAP deficiency did indeed prevent TAB1 from interacting physically with IKK␤ (Fig. 6A), resulting in a significant loss of NF-B activity (Fig. 6B) and Cox-2 expression (Fig. 6C) in 4T1 cells. We also investigated the connection between xIAP expression and that of (i) plasminogen activator inhibitor-1, which correlates with increased carcinoma cell invasion and diminished patient survival (31) and with the induction of EMT stimulated by TGF-␤ (32), and (ii) survivin, which is associated with breast cancer cell development and progression and with their detection in the circulation of breast cancer patients (33). As shown in Fig. 6D, xIAP deficiency significantly inhibited 4T1 cell expression of plasminogen activator inhibitor-1 and survivin as well as their ability to grow in soft agar (Fig. 6E) and invade synthetic basement membranes (Fig. 6F). Along these lines, xIAP-deficient MEFs possessed little to no ability to invade synthetic basement membranes as compared with their WT counterparts (Fig. 6G), further emphasizing the general importance of xIAP in mediating cell invasion. Collectively these findings show that xIAP is essential for regulating the activation of NF-B in breast cancer cells as well as promoting their expression of proinflammatory and metastatic genes necessary for breast cancer growth and invasion. Afterward detergent-solubilized wholecell extracts were prepared and immunoprecipitated (IP) with TAK1 antibodies. The resulting immunocomplexes were probed with antibodies against IKK␥/NEMO or HA as shown. Direct immunoblot analysis of an aliquot of the total cell extract was performed to monitor the levels of xIAP and IKK␥/NEMO expression and differences in protein loading. Images are from a representative experiment that was performed three times with identical results. B, human 293T cells were transiently transfected with ubiquitin (HA-tagged) together with either WT or mutant (H467A) xIAP as indicated. Afterward TAK1 immunocomplexes were isolated and immunoblotted with antibodies against HA or TAK1 as shown. Direct immunoblot analysis of an aliquot of the total cell extract was performed to monitor the levels of HA, ubiquitin, xIAP, TAK1, and ␤-actin expression and differences in protein loading. Images are from a representative experiment that was performed three times with identical results. C, human 293T cells were transiently transfected with pNF-Bluciferase and ␤-galactosidase together with either empty vector (E.V.), xIAP, or K63R-ubiquitin (Ub) as indicated. Afterward luciferase and ␤-galactosidase activities present in whole-cell extracts were measured. Data are the mean Ϯ S.E. (n ϭ 3) luciferase activities relative to 293T cells transfected with empty vector (*, p Ͻ 0.05; Student's t test). D, 4T1 cells were transiently transfected with pNF-B-luciferase and ␤-galactosidase together with either empty vector (E.V.), T204D-T␤R-I, and H467A-xIAP as indicated. Afterward luciferase and ␤-galactosidase activities present in whole-cell extracts were measured. Data are the mean (n ϭ 3) luciferase activities relative to 4T1 cells transfected with empty vector (*, p Ͻ 0.05; Student's t test). The error bars indicate Ϯ S.E.

DISCUSSION
An essential step during mammary tumorigenesis is the conversion of TGF-␤ from a suppressor to a promoter of breast cancer development and progression (1). Unfortunately the nature and sequence of events needed to transform the biological actions of TGF-␤ during mammary tumorigenesis remain inadequately defined, a fact that has hampered the development of chemotherapeutics capable of specifically targeting the oncogenic activities of TGF-␤ (34,35). The development of targeted TGF-␤ chemotherapies also has been hindered by the extensive roles TGF-␤ plays during the evolution of newly formed neoplasms, including their acquisition of EMT, angiogenic, and immunoevasive phenotypes. TGF-␤ also plays a prominent role in mediating breast cancer cell invasion and metastatic dissemination, which are the primary factors responsible for elevating cancer mortality rates (36). Thus, pharmacological targeting of breast cancer metastasis remains an attractive treatment modality, which traditionally has involved the targeting of antiapoptotic proteins, including xIAP, to inhibit the survival of metastatic cells (37). Although xIAP antisense therapeutics have been developed and demon-strated to sensitize tumors to tumor necrosis factor-related apoptosisinducing ligand (TRAIL)-induced apoptosis (38,39), the vast majority of xIAP antagonists typically target the BIR2 and BIR3 domains in xIAP that bind endogenous Smac/Diablo and neutralize the antiapoptotic activities of xIAP (40). At present, pharmacological agents capable of targeting the BIR1 domain of xIAP that binds TAB1 and mediates NF-B activation (8) have yet to be developed. Interestingly endogenous Smac proteins bind xIAP and prevent its binding of TAB1 via steric hindrance, a phenomenon that is not recapitulated by administration of small molecule xIAP inhibitors (8). Thus, the antitumor activities of small molecule xIAP inhibitors may be compromised by their inability to neutralize the coupling of xIAP to NF-B and its tumor-promoting functions. Indeed our findings herein establish xIAP as a mediator of oncogenic signaling by TGF-␤ and as such suggest that chemotherapeutic targeting of xIAP, particularly its coupling to NF-B activation, may diminish the ability of TGF-␤ to promote the development and progression of breast cancers.
By evaluating the response of normal and malignant MECs to TGF-␤, we discovered a novel and unexpected function of xIAP to selectively mediate TGF-␤ stimulation of mesenchymal gene expression in MECs undergoing EMT (Fig. 5). It is interesting to note that although mice engineered to lack xIAP expression are viable (27), presumably because of a high degree of functional redundancy exhibited by inhibitor of apoptosis protein family members, these animals do display developmental mammary gland defects (41). Our results showing that xIAP mediates MEC mesenchymal gene expression and invasion stimulated by TGF-␤ may underlie some of these developmental abnormalities. Along these lines, we observed elevated xIAP expression to potentiate TGF-␤ stimulation of N-cadherin and Cox-2 expression (i.e. mesenchymal markers) in MECs in a manner that is independent of their ability to acquire fibroblastoid morphologies. Clinically aberrant up-regulation of N-cadherin (42) and Cox-2 (43) expression is associated with increasing disease severity and the acquisition of metastatic phenotypes in breast cancers. Thus, chemotherapeutic targeting of xIAP may mitigate the metastatic functions of N-cadherin as well as the inflammatory, invasive, and metastatic activities of Cox-2 in breast cancer cells. In addition to invoking changes in epithelial and mesen- A and B, control (i.e. non-silencing (Non-Sil)) and xIAP-deficient (i.e. xIAP shRNA) NMuMG cells were transiently transfected with ␤-galactosidase and either pSBE-luciferase (A) or NF-B-luciferase (B) followed by overnight stimulation with TGF-␤1 (5 ng/ml) prior to measuring luciferase and ␤-galactosidase activities. Data are the mean (n ϭ 3) luciferase activities relative to untreated control cells (*, p Ͻ 0.05; Student's t test). C, parental (i.e. empty vector (E.V.)) and xIAP-expressing NMuMG cells were stimulated with TGF-␤1 (5 ng/ml) for 24 h at which point detergent-solubilized whole-cell extracts were prepared and immunoblotted with antibodies against E-cadherin, Cox-2, or xIAP as indicated. Differences in protein loading were monitored by reprobing stripped membranes with ␤-actin antibodies. Data are from a representative experiment that was performed three times with similar results. D, bright field images of control (i.e. non-silencing (Non-Sil)) and xIAP-deficient (i.e. xIAP shRNA) NMuMG cells before and after their stimulation with TGF-␤1 (5 ng/ml for 24 h) were captured from a representative experiment that was performed three times with identical results. E, control (i.e. non-silencing (Non-Sil)) and xIAP-deficient (i.e. xIAP shRNA) NMuMG cells were stimulated with TGF-␤1 (5 ng/ml) for 24 h at which point detergent-solubilized whole-cell extracts were prepared and immunoblotted with antibodies against Cox-2, N-and E-cadherins, or xIAP as indicated. Differences in protein loading were monitored by reprobing stripped membranes with ␤-actin antibodies. Data are from a representative experiment that was performed three times with similar results. The error bars indicate Ϯ S.E. chymal gene expression, EMT induced by TGF-␤ also induces diminished MEC proliferation and resistance to apoptotic stimuli (44). We too observed TGF-␤ and xIAP to suppress apoptotic signaling in malignant MECs in a manner consistent with the formation of xIAP⅐TAB1⅐TAK1⅐IKK␤ complexes and their activation of NF-B. Recently Derynck and co-worker (45) described the dissociation of mTOR signaling and its induction of increased MEC cell size and invasion from the ability of these same cells to acquire fibroblastoid morphologies in response to TGF-␤. Interestingly we too observed a bifurcation in the EMT program initiated by TGF-␤ that separates epithelial versus mesenchymal gene expression profiles in part via activation of the xIAP⅐TAB1⅐TAK1⅐IKK␤ signaling axis. Thus, the extent to which xIAP and mTOR signaling intersect and regulate MEC response to TGF-␤ is not yet known and needs to be explored in future studies.
Although the p38 MAPK and AKT pathways can participate in activating NF-B, our work defines a novel pathway that enables TGF-␤ to stimulate NF-B in response to xIAP-mediated ubiquitination of TAK1 and, consequently, to the formation of xIAP⅐TAB1⅐TAK1⅐IKK␤ complexes. The molecular determinants in xIAP that mediate its coupling to NF-B have been mapped to amino acids located within and adjacent to its C-terminal RING domain, sequences distinct from those that are operant in binding and inhibiting caspases (11). More recently, the BIR1 domain of xIAP was shown to interact physically with TAB1, thereby enabling its activation of TAK1 and subsequent stimulation of NF-B (8). Additionally the ability of TAK1 to activate IKK␤ depends upon ubiquitin-mediated protein-protein interactions particularly with respect to its recruitment of IKK␥/NEMO that recognizes and binds ubiquitin conjugates via its ubiquitin-binding domains (28,29). Thus, T␤R-I-mediated activation of xIAP facilitates the ubiquitina-tionofTAK1andthesubsequentformationofxIAP⅐TAB1⅐TAK1⅐ IKK␤ complexes that promote oncogenic signaling by TGF-␤. In addition, we further observed Lys-63-linked ubiquitin conjugates to promote TAK1-mediated formation of this oncogenic signaling axis and its ability to activate NF-B, findings reminiscent of other ubiquitin-dependent pathways that are operant in stimulating NF-B (46). Interestingly two recent reports identified TRAF6 as a novel T␤R-I-interacting protein that ubiquitinates TAK1, leading to its activation of the noncanonical TGF-␤ effectors JNK and p38 MAPK (47,48). Whether this pathway also couples TGF-␤ to NF-B activation was not addressed and clearly needs to be explored in future studies as does the interplay between xIAP and TRAF6 in mediating activation of these TGF-␤ effectors in breast cancer cells. Indeed given the ability of xIAP and TFAF6 to bind T␤R-I, future studies also need to investigate whether both ubiquitin ligases function cooperatively, independently, or redundantly in coupling TGF-␤ to the activation of NF-B activation and to the stimulation of JNK and p38 MAPK in developing and progressing mammary tumors.
Finally the biological and pathological outcomes of elevated xIAP expression and function, particularly those linked to oncogenic TGF-␤ signaling, may manifest themselves in a celland context-specific manner. Indeed potential mechanisms that may account for altered xIAP behavior include: (i) Aktmediated phosphorylation and stabilization of xIAP (49), (ii) Che-1-mediated induction of xIAP expression in response to DNA damage (50), (iii) up-regulated expression of the E2 ubiq-FIGURE 6. xIAP deficiency impairs the activation of NF-B and tumorigenic behavior of malignant MECs. A, detergent-solubilized whole-cell extracts prepared from control (i.e. non-silencing (Non-Sil)) or xIAP-deficient (i.e. xIAP shRNA) 4T1 cells were immunoprecipitated (IP) with anti-TAB1 antibodies followed by immunoblotting for IKK␤ or TAB1 as shown. Additionally whole-cell extracts also were immunoblotted for IKK␤, xIAP, TAB1, and ␤-actin as indicated. Data are from a representative experiment that was performed three times with identical results. B, control (i.e. non-silencing (Non-Sil)) or xIAP-deficient (i.e. xIAP shRNA) 4T1 cells were transiently transfected with pNF-B-luciferase and ␤-galactosidase and subsequently stimulated overnight with TGF-␤1 (5 ng/ml) prior to measuring luciferase and ␤-galactosidase activities. Data are the mean (n ϭ 2) luciferase activities relative to unstimulated control NMuMG cells (*, p Ͻ 0.05; Student's t test). C, detergent-solubilized whole-cell extracts prepared from control (i.e. non-silencing (Non-Sil)) or xIAP-deficient (i.e. xIAP shRNA) NMuMG and 4T1 cells were immunoblotted with antibodies against Cox-2, xIAP, and ␤-actin as indicated. Images are from a representative experiment that was performed three times with identical results. D, total RNA was isolated from control (i.e. non-silencing (Non-Sil)) and xIAP-deficient (i.e. xIAP shRNA) 4T1 cells and subjected to semiquantitative real time PCR to monitor expression of xIAP, plasminogen activator inhibitor-1 (PAI-1), and survivin (Surv) as indicated. Data are the mean (n ϭ 3) -fold changes in gene expression relative to control cells. E, the growth of control (i.e. non-silencing (Non-Sil)) and xIAP-deficient (i.e. xIAP shRNA) 4T1 cells in soft agar was quantified after 14 days in culture. Data and images are from a single experiment that was performed three times with identical results (*, p Ͻ 0.05; Student's t test). F, control (i.e. non-silencing (Non-Sil)) and xIAP-deficient (i.e. xIAP shRNA) 4T1 cells were induced by 2% serum to invade through synthetic basement membranes. Data are the mean (n ϭ 3) invasion relative to that of control 4T1 cells (*, p Ͻ 0.05; Student's t test). G, WT and xIAP-deficient (xIAP Ϫ/Ϫ ) MEFs were induced by 2% serum to invade through synthetic basement membranes. Data are the mean (n ϭ 3) invasion relative to that of WT MEFs (*, p Ͻ 0.05; Student's t test). The error bars indicate Ϯ S.E.

xIAP Alters TGF-␤ Signaling
uitin-conjugating enzyme Uev1A during cell immortalization and/or transformation (51,52), and (iv) up-regulated xIAP expression by NF-B inducers (i.e. IL-1) following cell detachment (53). It is interesting to note that xIAP deficiency failed to provide a significant antitumor activity in a mouse model of prostate cancer progression and instead tended to promote the development of aggressive tumor formation (54). This dichotomy in xIAP function may represent global differences in how normal and malignant MECs sense and respond to signals propagated by xIAP, a situation that mirrors the duality of TGF-␤ during cancer progression. Thus, similar to TGF-␤, xIAP may function initially as a tumor suppressor whose biological benefits are lost during mammary tumorigenesis, leading to its promotion of oncogenic signaling by TGF-␤ in late stage breast cancers. Should the merits of this supposition prove to be correct, our findings herein indicate that pharmacological targeting of xIAP may one day improve the ability of science and medicine to negate the oncogenic activities of TGF-␤ in late stage breast cancers.