IKKβ Is Required for Bcl-2-mediated NF-κB Activation in Ventricular Myocytes

The transcription factor nuclear factor κB (NF-κB) is regulated by cytoplasmic inhibitor IκBα. An integral step in the activation of NF-κB involves the phosphorylation and degradation of IκBα. We have previously reported that IκBα activity is diminished in ventricular myocytes expressing Bcl-2 (de Moissac, D., Zheng, H., and Kirshenbaum, L. A. (1999)J. Biol. Chem. 274, 29505–29509). The underlying mechanism by which Bcl-2 activates NF-κB is undefined. In view of growing evidence that the IκB kinases (IKKs), notably IKKβ, are involved in signal induced phosphorylation of IκBα, we ascertained whether IKKβ is necessary and sufficient for Bcl-2 mediated NF-κB activation. Here we demonstrate that expression of Bcl-2 in ventricular myocytes resulted in an increase in NF-κB-dependent DNA binding, NF-κB gene transcription and reduced IκBα levels. An increase in the IKKβ kinase activity was observed in cells expressing full-length Bcl-2 but not in cells expressing the BH4 deletion mutant of Bcl-2 (ΔBH4; residues 10–30). Catalytically inactive mutants of IKKβ, but not IKKα, suppressed Bcl-2-mediated IκBα phosphorylation and NF-κB activation. Transfection of human embryonic 293 cells with a kinase-defective Raf-1 or a kinase-defective mitogen-activated protein kinase/extracellular signal-regulated kinase kinase-1 (MEKK-1) suppressed Bcl-2-mediated IKKβ activity and NF-κB activation. Further, Bcl-2-mediated NF-κB activity was impaired in nullizygous mouse embryonic fibroblasts deficient for IKKβ. In this report, we provide the first direct evidence that Bcl-2 activates NF-κB by a signaling mechanism that involves Raf-1/MEKK-1 mediated activation of IKKβ.

Apoptosis is an evolutionary conserved event that permits the removal of (1) damaged or genetically unstable cells through an intrinsic cell suicide program (2). It is characterized by distinct morphological features typified by DNA fragmentation and cell shrinkage without the loss of membrane integrity (3). Despite the obvious beneficial effects, there is increasing awareness that defects in regulatory pathways that govern the apoptotic process may contribute to a number of human pathologies (reviewed in Refs. 4 and 5).
Recently, an antiapoptotic function for NF-B 1 has been described (6 -8). This is largely substantiated by studies in which genetic loss or functional inactivation of NF-B renders cells sensitive to proapoptotic signals including inflammatory cytokines such as TNF-␣ (7,8). In cells, NF-B exists as a heterodimeric complex composed of 50-and 65-kDa protein subunits bound to the inhibitory protein, inhibitor of B (IB␣) (9 -11). Ostensibly, IB␣ prevents nuclear targeting of NF-B by masking its nuclear localization motif. The mechanism by which biological signals activate NF-B in vivo is poorly defined; however, recent studies suggest a mechanism that involves the post-translational modification and degradation of IB␣ (12)(13)(14)(15). Mutagenesis studies have revealed serine residues 32 and 36 of IB␣ to be crucial for signal-induced phosphorylation and degradation of IB␣ and subsequent NF-B activation. Deletion mutations or substitution of these critical serine resides with alanine renders the IB␣ molecule defective for phosphorylation. Nonphosphorylatable IB␣ mutants remain constitutively bound to NF-B, preventing NF-B activation (16). Therefore, regulation of IB␣ by phosphorylation represents a key point in the activation of NF-B (16,17).
Insight into the signaling mechanisms that lead to IB␣ phosphorylation have identified a large molecular weight protein complex known collectively as the IB kinase (IKK) signalosome and includes IKK␣, IKK␤, and IKK␥ (18,19). IKK␣ and IKK␤ have been identified as catalytic subunits, whereas IKK␥ is a regulatory subunit (reviewed in Refs. 18 and 20). A general paradigm for the activation of NF-B by the IKK complex purports that signal-induced activation of the IB kinases phosphorylates IB␣ at serine residues 32 and 36, which triggers subsequent IB␣ degradation and NF-B activation (reviewed in Refs. 18 and 21). Biochemical and genetic ablation experiments indicate that IKK␤ may be more important in controlling NF-B activation than IKK␣ (22)(23)(24). This is supported by studies in which kinase-defective mutants of IKK␤ that fail to phosphorylate IB␣ render cells more sensitive to pro-death signals (25). Moreover, the fact that IKK␤ kinase knock-out mice die at embryonic day 14.5 from massive liver degeneration and apoptosis, similar to the p65Ϫ/Ϫ knockouts, illustrates the importance of IKK␤ for the NF-B activation pathway and the suppression of apoptosis (22,26). In contrast, disruption of IKK␣ alleles appear to only mildly affect NF-B activation despite perinatal death of the mice from severe morphogenic defects and keratinocyte abnormalities (23,24,27). The significance of NF-B activation for the suppression of apoptosis is further illustrated by the recent demonstration that life-promoting signals mediated through Ras/phosphati-dylinositol 3-kinase/Akt intersect the IKK complex to activate NF-B and prevent apoptosis (28 -30).
Although the molecular mechanisms that underlie apoptosis are poorly defined, recent evidence suggests that Bcl-2 may play a key role in this process (31,32). The fact that Bcl-2 can delay or prevent apoptosis provoked by a variety of deathpromoting signals, suggests that it probably impinges on more than one component of the death signaling pathway. Structure function studies have identified several key domains within Bcl-2 with putative antiapoptotic properties (33)(34)(35). In particular, the N terminus of Bcl-2, which encompasses an amphipathic ␣-helical loop designated the BH4 domain, has been suggested to be important in suppressing apoptosis through the formation of protein-protein interactions with key factors in the cell death pathway including calcineurin and Raf-1 (36 -39).
We have recently shown that NF-B is activated in ventricular myocytes expressing Bcl-2 (40,41). This was attributed to Bcl-2-mediated phosphorylation of IB␣ at serine residues 32 and 36 followed by IB␣ degradation by the proteasome (40), since mutations that abrogated IB␣ phosphorylation at these serine residues or interventions that blocked IB␣ degradation by the proteasome prevented Bcl-2-mediated NF-B activation. However, the signaling mechanism leading to Bcl-2-mediated IB␣ phosphorylation and NF-B activation is undetermined. In view of the growing evidence that the IB kinases, notably IKK␤, are involved in signal-induced activation of NF-B, we ascertained in the present study whether Bcl-2 impinges on IKK␤ to activate NF-B. In this report, we provide compelling new evidence that activation of NF-B by Bcl-2 occurs through a mechanism that involves the antiapoptotic BH4 domain of Bcl-2 and the Raf-1/MEKK-1 signaling pathway. We further demonstrate that the IKK␤ is necessary and sufficient for Bcl-2-mediated NF-B activation in ventricular myocytes.

MATERIALS AND METHODS
Cell Culture and Transfection-Neonatal ventricular myocytes were isolated from 2-day-old Sprague-Dawley rat hearts and submitted to primary culture under serum-free conditions as described (40). Human embryonic kidney 293 cells (HEK 293; American Tissue Type Collection) were maintained in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum (Invitrogen) as previously reported (40). Mouse embryonic fibroblasts deficient for IKK␤ were generously provided by M. Karin. For transfection experiments, cells were transfected for 3 h with Dulbecco's modified Eagle's medium containing Superfect (Qiagen) and 1-5 g of cytomegalovirus-driven eukaryotic expression vectors of either the wild type Bcl-2 (35), Bcl-2 BH4 domain deletion mutant (amino acids 10 -30) designated ⌬BH4 (kindly provided by J. Reed), IB kinase IKK ␤ wild type, or point substitutions of IKK␤ K-M , IKK␤ K-A with and without epitope Myc tag (kindly provided by W. C. Greene) (42). Epitope-FLAG-tagged wild type IB␣ and IB␣ mutant (IB␣ S32A,S36A) were kindly provided by D. Ballard (16), kinasedefective Raf-1 (Raf-1 K375M) was generously provided by D. Morrison (43), and wild type and kinase defective MEKK-1 were kindly provided by S. Gibson (44). To improve transfection efficiency in postnatal ventricular myocytes, recombinant adenoviruses encoding the cDNAs described above were generated as necessary by methods previously reported (40,45). Luciferase constructs containing NF-B response elements were previously described (40,46). Cells were transfected or infected with the cytomegalovirus-driven eukaryotic expression vector, pcDNA3 (Invitrogen) or adenovirus lacking the cDNA insert for all transfection controls. To control for potential differences in transfection efficiency among cell cultures, luciferase reporter activity was normalized to ␤-galactosidase activity and expressed as -fold increase. Following cell transfection or adenoviral infection, 293 cells or ventricular myocytes were washed and maintained in 10% fetal bovine serum-Dulbecco's modified Eagle's medium for 24 h. Data was obtained from at least n ϭ 3 independent cell cultures with replicates of 3 for each condition tested. Results were compared by Scheffe's multiple comparison test for analysis of variance and the unpaired two-tailed student t test, using a significance level of p Ͻ 0.05.
Western Blot Analysis-For immunodetection of IB␣ protein, human embryonic 293 cells and ventricular myocytes were harvested in 1.0% Triton X-100, 0.1% sodium deoxycholate, 140 mM NaCl, 10 mM Tris-HCl, pH 8.0 (radioimmune precipitation buffer). Cell lysates (50 g) were resolved on a 10% SDS-polyacrylamide gel at 140 V for 4 h and electrophoretically transferred to polyvinylidene difluoride membrane (Roche Diagnostics). For detection of IB␣ protein, the polyvinylidene difluoride filter was incubated for 3 h with a mouse monoclonal antibody directed toward human IB␣ protein clone C21 (Santa Cruz Biotechnology) in 150 mM NaCl, 50 mM Tris-HCl, pH 7.4, 0.3% Tween 20, 0.1% bovine serum albumin. Expression of wild type and ⌬BH4 deletion mutant forms of Bcl-2 proteins were detected using a hamster monoclonal antibody directed toward Bcl-2 (Pharmingen). All transfection experiments involving wild type and Bcl-2⌬BH4 mutant were subjected to Western blot analysis to verify equivalent Bcl-2 protein expression. For detection of transfected IKK␤ Myc or IB␣ FLAG-tagged proteins, cell lysates were incubated with murine anti-FLAG M2 or anti-Myc antibody, respectively (Kodak, Pharmingen) and immunoprecipitated with protein G-agarose beads (Amersham Biosciences) at 4°C for 4 h. Immunoprecipitates were washed and mixed with Laemmli loading buffer, boiled, and subjected to gel electrophoresis as described above. Bound proteins were detected by chemiluminescence reaction with horseradish peroxidase-conjugated antibody against mouse or hamster IgG using ECL reagents (Amersham Biosciences).
Assessment of Apoptosis-Nuclear morphology was assessed by Hoechst 33258 dye (Molecular Probes, Inc., Eugene, OR) as previously reported (49,50). To provoke apoptosis in 293 cells, cells were stimulated for 24 h with TNF-␣ (10 g/ml) in the presence of (5 g/ml) cycloheximide as previously reported (7,51). Nuclear morphology was visualized using an Olympus AX70 epifluorescence microscope counting Ն200 cells for each condition tested. Data are expressed as mean Ϯ S.E. percentage from control.
Electromobility Gel Shift Assay-Nuclear extracts of cells were prepared as previously described by McKinsey et al. (17). A 32 P-radiolabeled duplex oliogonucleotide probe with NF-B consensus binding sites, 5Ј-AGTTGAGGGGACTTTCGCAGGC-3Ј, was used as a template for the gel shift experiments (16). DNA binding reaction mixtures (20 g) were carried out on ice and contained 10 g of nuclear extract, 2 g of double-stranded probe, poly(dI-dC) (Amersham Biosciences), and 10 g of bovine serum albumin in 20 mM HEPES (pH 7.9), 5% glycerol, 1 mM EDTA, 5 mM dithiothreitol. Nuclear-protein complexes were resolved on a native 5% polyacrylamide gel in 1ϫ Tris borate EDTA, pH 8.0, and detected by autoradiography.
Statistical Analysis-Multiple comparisons between groups were determined by one-way analysis of variance. Unpaired two tailed Student's t test was used to compare mean differences between groups. Differences were considered to be statistically significant to a level of p Ͻ 0.05. In all cases the data were obtained from at least n ϭ 3 independent cell cultures using three replicates for each condition tested.

RESULTS
Our earlier work (40,41) demonstrated that NF-B activity was increased in ventricular myocytes expressing Bcl-2, which prompted us to examine whether the IB kinases are involved in Bcl-2-mediated NF-B activation. For these experiments, cardiac myocytes were transfected with catalytically inactive forms of the IB kinases, IKK␣ K-M and IKK␤ K-M , previously shown to impair NF-B activation (19,42,48). Expression of Bcl-2 in cells resulted in a significant 1.3-fold increase (p Ͻ 0.05) in NF-B gene activation compared with control cells (Fig. 1). Interestingly, Bcl-2-mediated NF-B gene transcription was suppressed to basal levels by the kinase-defective IKK␤ K-M . Importantly, similar effects were observed with an alternate IKK␤ mutant, IKK␤ K-A , 2 indicating that the inhibitory effects observed on Bcl-2-mediated NF-B gene activation were not restricted to the IKK␤ K-M point mutation utilized. In contrast, Bcl-2-mediated NF-B gene transcription was impaired to a lesser extent with the kinase-defective IKK␣. The finding that the kinase-defective IKK␤ had a greater inhibitory effect on Bcl-2-mediated NF-B activation is consistent with reports documenting that IKK␤ may be more important than IKK␣ for NF-B activation (19,22,42). Therefore, we focused on the role played by IKK␤ in Bcl-2-mediated NF-B activation.
Electromobility shift analysis of nuclear extract from ventricular myocytes expressing Bcl-2 revealed an increase in NF-B-dependent DNA binding ( Fig. 2A, lane 2) compared with control cells (lane 1), a finding consistent with the transcription data. Importantly, Bcl-2-mediated NF-B-dependent DNA binding was impaired in cells expressing the kinase-defective IKK␤⌬KD (Fig. 2, lane 4 versus lane 2). In contrast, mutations outside the kinase domain that render IKK␤ defective for dimerization (IKK␤⌬LZ) minimally impaired NF-B DNA binding (Fig. 2, lane 3), indicating that this region of IKK␤ may not be important for Bcl-2-mediated NF-B activation. Further, Bcl-2-mediated NF-B-dependent DNA binding was impaired by the nonphosphorylatable IB␣ S32A,S36A mutant.
Since NF-B activity is largely governed by IKK␤-dependent phosphorylation and degradation of IB␣, we next tested whether IKK␤ activity is increased in cells expressing Bcl-2.
For these experiments, cell lysate derived from HEK 293 cells expressing Bcl-2 was immunoprecipitated with a murine antibody directed toward IKK␤ and subjected to in vitro kinase assay. As shown in Fig. 2B, transfection of 293 cells with wild type IKK␤ resulted in an 18.5-fold increase (p Ͻ 0.05) in GST-IB␣ phosphorylation compared with vector-transfected cells or cells expressing the IKK␤ K-M kinase mutant, a finding consistent with our IKK␤ transcription data. Moreover, a 2.9-fold (p Ͻ 0.05) increase in GST-IB␣ phosphorylation was observed in cells expressing Bcl-2 compared with vector transfected cells. Transfection of 293 cells with the kinase-defective mutant IKK␤ K-M suppressed Bcl-2-mediated phosphorylation of GST-IB␣. Importantly, Bcl-2 had no effect on the GST-IB␣ S32A,S36A defective for phosphorylation at serine residues 32 and 36, 2 concordant with our previous data indicating that these residues are crucial for Bcl-2-mediated phosphorylation of IB␣ (40).
Western blot analysis of cardiac cell lysate derived from Bcl-2-expressing cells displayed a marked reduction in IB␣ protein levels compared with vector-transfected control cells. Importantly, the Bcl-2-mediated loss of IB␣ could be prevented with the kinase-defective IKK␤ K-M , Fig. 3, suggesting its involvement in Bcl-2-mediated IB␣ phosphorylation.
To verify that IKK␤ is necessary and sufficient for Bcl-2mediated effects on NF-B activation, we next tested whether Bcl-2 could activate NF-B in nullizygous mouse embryonic fibroblasts deficient for IKK␤. In contrast to wild type cells, IKK␤Ϫ/Ϫ cells failed to activate NF-B gene transcription in the presence of Bcl-2 (Fig 4). Importantly, transfection of IKK␤Ϫ/Ϫ cells with a eukaryotic expression vector encoding IKK␤ restored the ability of Bcl-2 to activate NF-B gene transcription, substantiating the requirement of IKK␤ for Bcl-2-mediated NF-B activation.
Given that Bcl-2 possesses no kinase activity, we reasoned that Bcl-2 probably impinges on cellular kinases that lead to IKK␤ activation. In this regard, the BH4 domain of Bcl-2 can reportedly interact with Raf-1, a known activator of the extracellular signal-regulated kinase kinase-1, MEKK-1, which can stimulate the kinase activity of IKK␤ (52). Since the BH4 domain of Bcl-2 has been shown to be important for Bcl-2mediated NF-B activation (41), it raises the interesting possibility that this region of Bcl-2 signals through Raf-1/MEKK-1/IKK␤ pathway to activate NF-B (52)(53)(54). To formally test this notion, we first assessed whether the BH4 domain of Bcl-2 was required to activate IKK␤. As shown in Fig. 5A (top), a 1.7-fold increase (p Ͻ 0.05) in GST-IB␣ phosphorylation was observed in HEK 293 cells expressing full-length Bcl-2, compared with vector-transfected control cells or cells expressing the Bcl-2 ⌬BH4 mutant. Importantly, densitometric analysis confirmed that both the wild type Bcl-2 and Bcl-2⌬BH4 deletion mutant were expressed to comparable levels in cells, indicating the observed differences in IKK␤ activity probably were not due to discrepancies in Bcl-2 protein expression or protein loading (Fig. 5A, bottom). These results support our contention that the BH4 domain of Bcl-2 may be required for IKK␤ activation. We next assessed whether Raf-1 was involved in the signaling pathway, leading to NF-B activation by Bcl-2, by testing whether a kinase-defective Raf-1 would suppress Bcl-2-mediated NF-B activation. As shown in Fig. 5B, Bcl-2-mediated NF-B-dependent DNA binding was impaired by the kinase-defective Raf-1. Moreover, consistent with these findings was the observation that the Bcl-2-mediated loss of IB␣ could be prevented by the kinase-defective Raf-1 (Fig. 3). Together, our data strongly suggest that Raf may be part of the signaling pathway involved in Bcl-2-mediated NF-B activa-tion. To more formally address this notion, we next assessed whether the Raf-1 target MEKK-1 is sufficient to activate NF-B in ventricular myocytes. For these experiments, myocytes were transfected with wild type and/or kinase-defective versions of MEKK-1 and assessed for NF-B gene transcription. As shown in Fig 6A, an increase in NF-B gene transcription was observed in cells transfected with MEKK-1 compared with vector-transfected cells, establishing that MEKK-1 can activate NF-B in ventricular myocytes. To assess whether Bcl-2 signals through MEKK-1 to activate IKK␤, ventricular myocytes were infected with a replication-defective adenovirus encoding a kinase-defective version of MEKK-1 and assessed for IKK␤ kinase activity. In contrast to cells expressing Bcl-2 alone, cells expressing Bcl-2 in the presence of the kinasedefective MEKK-1 displayed a marked reduction in IKK␤ activity ( Fig 6B). Importantly, Bcl-2-mediated loss of IB␣ activity could be prevented with the kinase-defective MEKK-1 (Fig.  6C). To verify that MEKK-1 signals through IKK␤ to activate NF-B, cells were transfected with either the wild type MEKK-1 or MEKK-1 kinase-defective mutant in the presence

Bcl-2-and IKK␤-mediated NF-B Activation
and absence of wild type or mutant IKK␤ and assessed for NF-B activity. As shown in Fig. 6A, MEKK-1-dependent NF-B activation was suppressed to basal levels by the kinaseinactive IKK␤ K-M. In contrast, however, the kinase-defective MEKK-1 had no apparent effect on IKK␤-dependent activation of NF-B, indicating that IKK␤ is probably downstream of MEKK-1 in the NF-B activation pathway.
The premise of our current findings indicates that Bcl-2mediated NF-B activation may be important for suppressing apoptosis. Therefore, to verify the physiological significance of our observations and the interrelationship between the Bcl-2 BH4 domain and IKK␤ for the suppression of apoptosis, we tested whether ⌬BH4Bcl-2 defective for IKK␤ activation would impair the ability of Bcl-2 ability to suppress apoptosis trig- Experiments were repeated at least n ϭ 3 with independent culture conditions using replicates of 3 for each condition. B, top, in vitro kinase assay for IKK␤. Neonatal myocytes were infected with a recombinant adenovirus encoding Bcl-2 in the absence and presence of a kinase-defective MEKK-1MT and assessed for IKK␤ activity. Bottom, Western blot analysis of cardiac cell lysate to demonstrate equivalent IKK␤ content. C, top, Western blot analysis for IB␣. Bcl-2-mediated loss of IB␣ activity is suppressed by the kinase-defective MEKK-1MT. Bottom, actin to demonstrate equivalent protein loading. D, histogram represents apoptosis in human embryonic 293 cells determined by Hoechst 33258 nuclear staining. Since TNF-␣ does not appreciably provoke apoptosis of cells in the absence of a protein synthesis inhibitor (7, 41), we stimulated 293 cells with 10 ng/ml TNF-␣ in the presence of 5 g/ml cycloheximide (TNF-␣) to trigger apoptosis as previously reported (7,51). As shown in Fig. 6D

Bcl-2-and IKK␤-mediated NF-B Activation
gered by TNF-␣ in HEK 239 cells (41). As shown in Fig. 6D, a 4.5-fold increase (p Ͻ 0.001) in apoptotic nuclei was observed in cells by Hoechst 33258 nuclear staining in the presence of TNF-␣ compared with vehicle-treated control cells (Fig. 6D). Importantly, the cell death triggered by TNF-␣ was suppressed to basal levels in cells expressing wild type Bcl-2 (p Ͻ 0.001). In contrast, however, the Bcl-2⌬BH4 mutant defective for IKK␤ activation failed to suppress TNF-␣-induced apoptosis. Notably, transfection of IKK␤ into 293 cells expressing the Bcl-2⌬BH4 restored the ability of Bcl-2⌬BH4 to suppress TNF-␣induced apoptosis in a dose-dependent manner. DISCUSSION In the present report, we provide evidence for the regulation of IB␣ activity by Bcl-2 through a mechanism that involves the Raf-1/MEKK-1 signaling pathway. Furthermore, our data suggest that IKK␤ is necessary and sufficient for Bcl-2-mediated NF-B activation. Another important feature of our study is that the antiapoptotic BH4 domain of Bcl-2 is required for IKK␤ activity and suppression of apoptosis.
Since Bcl-2 possesses no inherent kinase activity, we reasoned that Bcl-2 probably impinges on cellular factors that directly or indirectly lead to NF-B activation. Because the IB kinase complex was identified as being involved in the signalinduced activation of NF-B, we ascertained whether Bcl-2 signals through the IKK complex to activate NF-B. Indeed, several key observations suggest that Bcl-2-mediated IB␣ phosphorylation requires IKK␤. First, in the presence of Bcl-2, IB␣ phosphorylation and degradation was inhibited by the dominant-negative IKK␤. Second, Bcl-2-mediated NF-B-dependent DNA binding and NF-B gene transcription was suppressed by the kinase-defective form of IKK␤. Third, Bcl-2mediated NF-B activation was impaired in cells deficient for IKK␤. Taken together, our data strongly suggest that Bcl-2mediated IB␣ inactivation and NF-B activation requires IKK␤. The contribution of IKK␣ to Bcl-2-mediated NF-B activation in the present study was less apparent. Since the kinase-defective IKK␣ suppressed the induction of NF-B to a lesser extent than IKK␤ suggests that IKK␤ may play a more dominant role in Bcl-2-mediated NF-B activation. This finding is concordant with earlier reports documenting a greater importance of IKK␤ for NF-B activation (18,19,22,55). However, we cannot exclude the possibility that IKK␣ may contribute to IKK␤ complex formation or act peripherally to activate NF-B. This point awaits further investigation.
The mode by which Bcl-2 confers resistance to pro-death signals is poorly defined; however, the fact that Bcl-2 can suppress apoptosis provoked by a variety of pro-death signals suggests that it probably impinges on more than one component of the cell death pathway. In this regard, the BH4 domain of Bcl-2 has been suggested to be important for cell survival (31, 56). For example, the BH4 domain was identified to be crucial for disrupting Apaf-1/cytochrome c/caspase 9-dependent processing of caspase 3 (38,39). The fact that the BH4 domain of Bcl-2 was necessary for induction of IKK␤ activity is intriguing and raises the possibility that this region of Bcl-2 modulates cellular factors that lead to NF-B activation. The BH4 domain can reportedly interact with Raf-1 (36), a known activator of MEKK-1 (52,58). Indeed, our studies showed that dominant-negative Raf-1 suppressed Bcl-2-mediated NF-B dependent DNA binding and degradation of IB␣ activity, suggesting its involvement in Bcl-2-mediated NF-B activation. The fact that MEKK-1 is activated by Raf-1 and can stimulate the kinase activity of IKK␤ (30,52,53,59) prompted us to ascertain whether MEKK-1 was involved in Bcl-2-mediated effects on IB␣. Interestingly, the kinase-inactive form of MEKK-1 prevented Bcl-2-mediated phosphorylation of IB␣, consistent with the notion that MEKK-1 is part of the signaling pathway leading to NF-B activation by Bcl-2. Furthermore, we demonstrated that MEKK-1 resides upstream of IKK␤ in the Bcl-2-NF-B activation pathway, given that the kinaseinactive MEKK-1 suppressed Bcl-2-mediated IKK␤ kinase activity. Although protein-protein interactions were not studied here, it is tempting to speculate that Bcl-2, through its BH4 domain, associates with Raf-1, leading to the downstream activation of MEKK-1 and subsequent IKK␤-dependent NF-B activation (Fig. 7).
The significance of Bcl-2-mediated NF-B activation is unknown but may be related to the suppression of cell death. In this regard, a recent report documented the caspase 3-dependent cleavage of IB␣, resulting in a peptide fragment that inhibits NF-B activation (47,60). In addition, the caspase 3-dependent cleavage of IKK␤ was proposed as a putative mechanism for TNF-␣-induced apoptosis. The fact that the cellular caspase inhibitors c-IAP-1 and c-IAP-2 (61) as well as other antiapoptotic factors (62,63,64) are known transcriptional targets of NF-B supports the notion that Bcl-2 may suppress apoptosis at least in part by activating NF-B-regulated factors. This is consistent with our earlier work demonstrating a requirement for the induction of NF-B-regulated antiapoptotic genes for Bcl-2 to suppress TNF-␣-induced apoptosis of ventricular myocytes (40,41,50).
The relationship between Bcl-2 and the NF-B signaling pathway becomes even more profound, given that IB␣ has been found localized with the adenine nucleotide translocator (65), a key component of the mitochondrial permeability transition pore (66 -68). The opening of the permeability transition pore by pro-death factors such as Bax and Bad has been linked to mitochondrial swelling, loss of ⌬⌿ m , cytochrome c release, caspase activation, and cell death (68). Interestingly, the BH4 domain of Bcl-2 was found to be crucial for preventing mitochondrial permeability transition pore changes and apoptosis (57). The fact that deletion of the BH4 domain rendered Bcl-2 defective for activating NF-B as well as preventing TNF-␣induced apoptosis strongly suggests the importance of the BH4 domain for suppressing apoptosis. Moreover, the fact that IKK␤ could functionally rescue cells from apoptosis in the presence of the Bcl-2⌬BH4 deletion mutant supports our hypothesis that Bcl-2, through its BH4 domain, impinges on NF-B to suppress apoptosis (41). Whether Bcl-2 prevents cell death by regulating permeability transition pore changes through modulation of the IB␣⅐NF-B complexes at the level of the mitochondria is unknown and has not been formally FIG. 7. An operational model for the activation of NF-B by Bcl-2. The N-terminal BH4 domain of Bcl-2 (black) signals through Raf-1/MEKK-1 signaling complex to activate IKK␤ followed by the subsequent phosphorylation-induced degradation of IB␣ by the proteasome. This leads to the activation of NF-B and NF-B-dependent genes to suppress apoptosis.
Bcl-2-and IKK␤-mediated NF-B Activation tested. Moreover, it must also be stated that the significance of the Raf-1/MEKK-1 signaling pathway in the activation of NF-B by Bcl-2 for the suppression of apoptosis is equally unknown and remains a point of active investigation.
Nevertheless, under the conditions tested, our data provide the first evidence for the regulation of IB␣ by Bcl-2 through a mechanism that involves Raf-1/MEKK-1 and the IKK␤ kinase. Moreover, our data indicate that the ␣-helical BH4 domain of Bcl-2 is required for the observed actions of Bcl-2 on IB␣ activity, further linking Bcl-2 to the NF-B signaling pathway for the suppression of apoptosis.