Suppression of Tumor Necrosis Factor-mediated Apoptosis by Nuclear Factor κB-independent Bone Morphogenetic Protein/Smad Signaling*

The activation of nuclear factor κB (NF-κ B) plays a pivotal role in the regulation of tumor necrosis factor (TNF)-mediated apoptosis. However, little is known about the regulation of TNF-mediated apoptosis by other signaling pathways or growth factors. Here, unexpectedly, we found that bone morphogenetic protein (BMP)-2 and BMP-4 inhibited TNF-mediated apoptosis by inhibition of caspase-8 activation in C2C12 cells, a pluripotent mesenchymal cell line that has the potential to differentiate into osteoblasts depending on BMP stimulation. Utilizing both a trans-dominant IκBα inhibitor of NF-κB expressed in C2C12 cells and IκB kinase β-deficient embryonic mouse fibroblast, we show that BMP-mediated survival was independent of NF-κB activation. Rather, the antiapoptotic activity of BMPs functioned through the Smad signaling pathway. Thus, these findings provide the first report of a BMP/Smad signaling pathway that can inhibit TNF-mediated apoptosis, independent of the prosurvival activity of NF-κB. Our results suggest that BMPs not only stimulate osteoblast differentiation but can also promote cell survival during the induction of bone formation, offering new insight into the biological functions of BMPs.

NF-B 1 is a stress-responsive transcription factor that plays important roles in development and immunity (1)(2)(3). Classical NF-B is a heterodimer composed of p50 and p65/RelA, which is sequestered in the cytoplasm by the IB group of inhibitory proteins. Proinflammatory cytokines such as tumor necrosis factor (TNF) activate IB kinase (IKK) complex to phosphorylate the conserved N-terminal region of IB proteins (1)(2)(3). The phosphorylated IB is ubiquitinated and subsequently degraded by the 26S proteasome. This results in the nuclear translocation of NF-B and binding to NF-B-responsive elements, followed by NF-B-dependent transcriptional activation (1)(2)(3)(4)(5)(6).
We and others have demonstrated that NF-B plays a critical role in inhibition of TNF-mediated apoptosis (7)(8)(9). In the absence of NF-B activation, TNF can trigger the caspase cascade by interacting with Fas-associated death domain protein, which then recruits and activates caspase-8. (11)(12)(13). Active caspase-8 promotes cell death by either directly processing other downstream caspases or cleaving the cytosolic Bid protein, a proapoptotic family member of Bcl-2 (12,14,15). Truncated Bid translocates to mitochondria, resulting in the release of cytochrome c from mitochondria into the cytosol and the subsequent activation of apoptosis (14,15). However, in the presence of NF-B activation, the caspase-8-mediated apoptotic pathway is suppressed (4). Several important antiapoptotic molecules have been identified that are transcriptionally regulated by NF-B. These molecules include the Bcl-2 family members A1 and Bcl-x L , inhibitors of apoptosis family proteins, TNF receptor-associated factor family proteins, IEX-1L, and the recently elucidated NF-B-inducible death effector domaincontaining protein (4, 16 -19). Although several growth factors have been found to inhibit TNF-mediated apoptosis through the activation of NF-B (20), little is known regarding the regulation of TNF-mediated apoptosis by NF-B-independent signaling.
Bone morphogenetic proteins (BMPs), members of the transforming growth factor ␤ superfamily, were originally identified by their unique ability to induce bone formation in vivo (21)(22)(23). BMPs initiate a signaling cascade through the ligand-dependent activation of a complex of heteromeric transmembrane serine-threonine kinase receptors, type I and type II (24). The activated BMP type I receptor phosphorylates Smad1 and Smad5, resulting in their dissociation from the receptor complex. The phosphorylated Smad1 and Smad5 then form heterooligomeric complexes with Smad4 and translocate into the nucleus to activate the transcription of target genes (25)(26)(27). Interestingly, the BMP/Smad signaling pathway is also negatively regulated by a structurally and functionally divergent Smad protein of the subfamily of inhibitory Smads, Smad6 and Smad7 (25)(26)(27).
BMPs induce osteoblast differentiation of mesenchymal cells and are involved in postnatal bone remodeling (28,29). Several studies demonstrate that BMPs control digit numbers in the limbs, possibly through the induction of apoptosis in the interdigital and anterior tissue (30,31). BMPs can inhibit proliferation and induce apoptosis of multiple myeloma cells, indicating the therapeutic potential of BMPs in cancer treatment (32).
Given that the cell death machinery can be modulated by multiple factors or signaling pathways, we hypothesized that NF-B-independent mechanisms may exist to suppress TNFmediated apoptosis. During the search for regulators of TNFmediated apoptosis, we unexpectedly found that in contrast to previous studies implicating BMPs in proapoptotic mechanisms (31,32), BMPs possessed a novel antiapoptotic activity that promoted survival of mesenchymal cells during BMPinduced osteoblast differentiation. The results reported here reveal a unique BMP/Smad signaling pathway that suppresses TNF-mediated apoptosis in a NF-B-independent manner. Given the fact that apoptosis plays an important role in osteoporosis and other inflammatory-related bone disorders (33), our results suggest that the utilization of BMP for bone regeneration and repair may have dual benefits: stimulation of osteoblast differentiation, and inhibition of apoptosis.
Trypan Blue Exclusion, Cell Death Enzyme-linked Immunosorbent Assay, Annexin V Staining, and EGFP Survival Assay-Cells (10 5 ) were plated onto 6-well plates the day before stimulation. Cells were pretreated with the indicated concentrations of BMP-2 or BMP-4 for 2 h and subsequently killed with tumor necrosis factor (20 ng/ml) for 24 h. Cell viability was determined by trypan blue exclusion. The supernatant was collected, and cell death enzyme-linked immunosorbent assay was performed as described previously (Roche Molecular Biochemicals).
For annexin V staining, 2 ϫ 10 4 cells were plated on a microscope coverslip in 24-well plates the day before stimulation. Sixteen h after stimulation, cells were gently washed once with 1ϫ binding buffer and stained with annexin V conjugated with EGFP solution (1:40; CLON-TECH) and propidium iodide (50 mg/ml, Sigma) for 15 min at room temperature in the dark. After staining, cells were washed twice with 1ϫ binding buffer and fixed in 2% formaldehyde in phosphate-buffered saline (pH 7.4) for 20 min. The coverslips were inverted on a drop of Vectashield mounting media (Fisher Scientific) on slides and examined and photographed under a fluorescence microscope using a filter set for fluorescein isothiocyanate.
For EGFP survival assay, cells were co-transfected with pCMV-GFP vector and either pCMV-Smad-7 or control vector with Superfect (Qiagen). Twenty-four h after transfection, cells were pretreated with BMP-2 or BMP-4 for 2 h and then killed with TNF for an additional 24 h. The green fluorescent protein-expressing cells were examined directly under a fluorescence microscope.
Western Blot Analysis-Cells (2 ϫ 10 6 ) were plated in a 100-mm plate the day before stimulation. Cells were pretreated with BMP-2 or BMP-4 at a concentration of 200 ng/ml for the indicated times. The detached and attached cells were collected. Whole cell extracts were prepared with radioimmune precipitation buffer containing 1% Nonidet P-40, 5% sodium deoxycholate, 1 mM phenylmethylsufonyl fluoride, 100 mM sodium orthovanadate, and a 1:100 mixture of protease inhibitors (Sigma). The proteins were resolved in by SDS-10% polyacrylamide gel electrophoresis and transferred to polyvinylidene difluoride membrane by semidry transfer apparatus (Bio-Rad). The membrane was probed with polyclonal antibodies against caspase-8 (Santa Cruz Biotechnology) and visualized using ECL reagent (Amersham Pharmacia Biotech) according to the manufacturer's recommendations (10).
In Vitro Caspase-3 and Caspase-8 Activity Assay-Cells were treated as described above for Western blot. The detached and attached cells were collected, washed with phosphate-buffered saline, and lysed in 200 l of ice-cold hypotonic lysis buffer provided by the manufacturers (R&D Systems, CLONTECH, or Promega). The cell extracts were centrifuged, and supernatants were collected. Protein extracts (200 -300 g) were incubated in reaction buffer containing IETD-pNA (colorimetric caspase-8 substrate; R&D Systems) or DEVD-pNA (colorimetric caspase-3 substrate; Promega) at 37°C for 2 to 3 h. The samples were analyzed with a plate reader by measuring the optical density at a wavelength of 405 nm.
Electrophoretic Mobility Shift Assays and NF-B Luciferase Reporter Assays-Cells were treated with TNF (10 ng/ml) for the indicated time points. Nuclear extracts were prepared for electrophoretic mobility shift assays as described previously (15,22). Aliquots (5 g) of nuclear extracts were preincubated with 1 g of poly(dI-dc) in binding buffer (10 mM Tris, pH 7.7, 50 mM NaCl, 20% glycerol, 1 mM dithiothreitol, and 0.5 mM EDTA) for 10 min at room temperature. Approximately 20,000 cpm of 32 P-labeled DNA probe containing the class I major histocompatibility complex NF-B site (5Ј-CAGGGCTGGGGATTCCCCATCTCCACA-GTTTCACTTC-3Ј) was then added, and binding proceeded for 15 min. The complexes were separated on a 5% polyacrylamide gel and exposed for autoradiography (5,17).
To determine NF-B transcription activity, cells were transfected with 2ϫ B-dependent luciferase reporter constructs using the Superfect reagent (Qiagen) according to the manufacturer's instructions (20). PRL-TK Renilla luciferase reporter was co-transfected to normalize for transfection efficiency. Luciferase activities were measured using a dual luciferase system (Promega).

BMP-2 and BMP-4 Inhibit TNF-mediated Apoptosis-C2C12
cells are pluripotent mesenchymal cells that are widely used for the study of cell differentiation in vitro (6,28). To study the role of NF-B in cell growth and apoptosis, we have established a C2C12I cell line stably expressing a modified form of the NF-B inhibitor, super-repressor-IB␣ (SR-IB␣; Fig. 1A). SR-IB␣ contains serine-to-alanine mutations at residues 32 and 36, which prevent signal-induced phosphorylation and subsequent proteasome-mediated degradation of IB␣. As shown in Fig. 1A, TNF rapidly induced the phosphorylation and degradation of endogenous IB␣ in C2C12 control cells (C2C12V) but

FIG. 1. SR-I B␣ suppresses the activation of NF-B in C2C12 cells.
A, SR-IB␣ inhibited the phosphorylation and degradation of IB␣. C2C12 cells were transduced with retroviruses expressing human SR-IB␣ or control vector and selected with neomycin (400 g/ml). The resistant clones were pooled and designated as C2C12I cells or control cells (C2C12V), respectively. Of note, the molecular weight of human SR-IB␣ was slightly higher than that of endogenous mouse IB␣. C2C12I cells or control cells were treated with TNF (20 ng/ml) for the indicated time periods. The whole extracts were probed with polyclonal antibody against IB␣ or phosphorylated IB␣ (p-IB␣). For the loading control, the blot was stripped and reprobed with monoclonal antibodies against ␣-tubulin. B, SR-IB␣ inhibited the nuclear translocation of NF-B. The nuclear extracts were prepared as described under "Experimental Procedures." Aliquots (5 g) of protein were incubated with 32 P-labeled NF-B probe. The reaction was resolved by 5% polyacrylamide gel electrophoresis and exposed to film. C, BMP-4 did not activate NF-B in C2C12I cells. Cells were transfected with 2ϫ B-dependent luciferase reporter constructs. Twenty-four h after transfection, cells were treated with BMP-4 for 16 h. Luciferase activities were measured as described under "Experimental Procedures." did not induce phosphorylation and degradation of SR-IB␣ in C2C12I cells. As predicted, SR-IB␣ inhibited the nuclear translocation of NF-B induced by TNF (Fig. 1B). Consistent with our previous studies (10), C2C12I cells were sensitive to TNF-mediated killing, confirming that NF-B inhibits TNFmediated apoptosis in cells ( Fig. 2A).
Given the fact that the cell death machinery could be modulated by different factors and stimuli, we hypothesized that a NF-B-independent pathway might be able to regulate TNFmediated apoptosis. Thus, we utilized C2C12I cells to screen new growth factors or signaling molecules that could regulate TNF-mediated apoptosis in a NF-B-independent fashion. Because C2C12 cells are a well-established model system to study BMP-induced osteoblast differentiation, we sought to determine whether BMPs would modulate TNF-mediated apoptosis. Previously, Lopez-Rovira et al. (34) have found that transforming growth factor ␤ activates NF-B that is independent of IB degradation. Thus, to confirm the observation that BMPs did not activate NF-B in C2C12I cells, NF-B-dependent luciferase reporter assays were performed. As shown in Fig. 1C, BMP-4 did not induce NF-B transcriptional activities in C2C12I cells. To test whether BMPs promoted cell survival, C2C12I cells were pretreated with BMP-4 or vehicle control for 2 h and then treated with TNF for 24 h. As shown in Fig. 2A, trypan blue exclusion analysis found that TNF stimulation caused 70 -80% cell death in C2C12I cells. In contrast, only approximately 20% of cells were killed by TNF when cells were pretreated with BMP-4. BMP-4-mediated protection against TNF killing was dose-dependent, as shown in Fig. 2B. Furthermore, we also examined whether another BMP family member could inhibit TNF killing. As shown in Fig. 2, C and D, BMP-2, like BMP-4, also potently inhibited TNF killing in a dosedependent manner.
To further confirm the results described above, DNA fragmentation and histone release from cell culture supernatant were also measured with cell death enzyme-linked immunosorbent assay (4). As shown in Fig. 3, A and B, pretreatment with BMP-2 and BMP-4 significantly inhibited DNA fragmentation induced by TNF. To determine whether BMP-mediated survival was due to the modification of apoptosis, EGFP-conjugate annexin V staining in conjunction with propidium iodide was performed. Changes in the plasma membrane of the cell surface are one of the earliest features of cells undergoing apoptosis. After initiating apoptosis, phosphatidylserine is translocated from the inner face of the plasma membrane to the cell surface (11)(12)(13). Because annexin V has a high affinity for phosphatidylserine and can bind to cells with exposed phosphatidylserine, it has been used to detect the early stage of apoptosis (4,17). As shown in Fig. 3C, compared with numerous annexin V-positive-stained cells in the TNF-treated cells, only limited staining was detected in cells pretreated with BMP-2 or BMP-4. Annexin V-positive cells failed to take up propidium iodide (data not shown), indicating cell death via apoptosis, but not necrosis. Taken together, these results suggest that BMPs activate a NF-B-independent signaling pathway to suppress TNF-induced apoptosis.
BMP Signaling Blocks Caspase-8 Activation and the Subsequent Cleavage of Bid-Next, we examined the molecular mechanism by which BMP signaling inhibited TNF-mediated apoptosis. To initiate apoptosis, TNF binds its receptor to recruit TNF receptor-associated death domain protein and Fasassociated death domain protein to activate caspase-8 (4,8,11). To determine whether BMPs inhibited TNF-induced caspase-8 activation, cells were pretreated with BMP-2 and subsequently killed by TNF. As shown in Fig. 4A, Western blot analysis

FIG. 2. BMP-2 and BMP-4 inhibit TNF-mediated apoptosis through a NF-B-independent mechanism.
A, BMP-4 inhibited TNF killing. Cells were pretreated or not pretreated with BMP-4 for 2 h (100 ng/ml) and then treated with TNF (20 ng/ml) for 24 h. Cell viability was determined by trypan blue exclusion. The assays were performed in triplicate, and the results represent the mean value from three independent experiments. Statistical differences between each group were determined by Student's t test. ‫,ء‬ p Ͻ 0.01. B, BMP-4 inhibited TNF killing in a dose-dependent fashion. Cells were pretreated with the indicated concentrations of BMP-4 and then killed by TNF. C and D, BMP-2 inhibited TNF killing. The experiments were performed as described in A and B, respectively. Statistical differences between each group were determined by Student's t test. ‫,ء‬ p Ͻ 0.01.

FIG. 3. BMP-2 and BMP-4 inhibit TNF-mediated apoptosis.
A and B, BMP-2 and BMP-4 inhibited DNA fragmentation. Cell treatment was performed as described in the Fig. 1 legend. Cell supernatant (20 l) from each cell group was incubated with anti-histone and anti-DNA antibody at room temperature for 2 h. The reaction was measured with a microplate reader at a wavelength of 405 nm. The assays were performed in triplicate, and the results represent the mean value from three independent experiments. Statistical differences between each group were determined by Student's t test. ‫,ء‬ p Ͻ 0.01. C, annexin V staining of apoptotic cells. Cells were untreated or pretreated with BMP-2 (200 ng/ml) or BMP-4 (100 ng/ml) for 2 h and then killed by TNF for 14 -16 h. Cells were washed and incubated with EGFP-annexin V (1:40) and propidium iodide. The cells were examined by fluorescence microscopy.
found that the processing of caspase-8 induced by TNF was strongly inhibited by BMP-2. Because we could not detect the active subunit of caspase-8 by Western blot analysis, we performed the caspase-8 enzymatic assay to determine whether BMP-2 suppressed caspase-8 activity induced by TNF. As shown in Fig. 4B, caspase-8 enzymatic activity induced by TNF was inhibited by BMP-2. A proapoptotic family member of Bcl-2, Bid, has been found to be a specific substrate of caspase-8 during death receptor-mediated apoptosis. The truncated Bid promotes and/or amplifies apoptosis by inducing the release of cytochrome c from mitochondria to the cytosol (14,15). To confirm that caspase-8 activity was inhibited by BMP-2, we examined the cleavage of Bid. As shown in Fig. 4A, a substantial amount of truncated Bid was detected in cells after TNF stimulation, but not in cells pretreated with BMP-2. In addition to the cleavage of Bid, caspase-8 can also directly activate executing caspases such as caspase-3 to induce apoptosis (11,13). Thus, we also determined whether caspase-3 activity was affected by the inhibition of caspase-8. As predicted, significantly lower DEVDase activity was present in the cells pretreated with BMP-2 compared with cells without BMP-2 pretreatment after TNF stimulation (Fig. 4C). These results demonstrated that BMP-2 suppressed TNF-mediated apoptosis by inhibiting caspase-8 activation.
BMP-mediated Cell Survival Is Dependent on SMAD Signaling-Because the results presented above were from C2C12 cells in which NF-B activities were suppressed, they indicated that BMP-2-or BMP-4-mediated survival was independent of NF-B. Biochemical and genetic studies have demonstrated that BMPs exhibit their biological functions through the Smad signaling pathway (35)(36)(37). Thus, we examined whether BMPmediated cell survival was dependent on the Smad signaling pathway. To block BMP-mediated Smad signaling, an inhibitory Smad, Smad7, was utilized. Smad7 interacts with activated BMP type I receptors and inhibits BMP-mediated signaling by inhibiting Smad1 and Smad5 phosphorylation (37). Cells were co-transfected with pEGFP expression vector and pCMV-Smad7 expression vector or a control vector. Twenty-four h after transfection, cells were pretreated with BMP-2 for 2 h, and TNF was subsequently added to induce cell death. As shown in Fig. 5, A and B, cells transfected with control vector had no effect on BMP-mediated protection against TNF killing. In contrast, cells transfected with Smad7 were sensitive to TNF killing regardless of BMP pretreatment, indicating that BMP-mediated survival was dependent on Smad signaling.
To rule out a nonspecific effect of SR-IB␣, we also determined whether BMP inhibited TNF-mediated apoptosis in mouse embryonic fibroblasts from IB kinase ␤-deficient mice (IKK␤Ϫ/Ϫ cells). IKK␤ is a key component of IKK complex for phosphorylation of IBs and has been found to play an essential role in NF-B activation (38). TNF is unable to activate NF-B transcription in IKK␤Ϫ/Ϫ cells due to lack of IKK activity and the subsequent absence of IB degradation. Therefore, similar to SR-IB␣-expressing cells, IKK␤Ϫ/Ϫ fibroblasts are also capable of undergoing TNF-mediated apoptosis (38). To test this effect, IKK␤Ϫ/Ϫ cells were pretreated with or without BMP-2 for 2 h and then killed with TNF for 24 h. As shown in Fig. 6A, BMP-2 also significantly inhibited TNFmediated apoptosis in IKK␤Ϫ/Ϫ cells. Similarly, the ectopic expression of Smad7 also rendered BMP-2-pretreated cells sensitive to TNF-mediated apoptosis (Fig. 6B), confirming that BMP-mediated survival was dependent on Smad signaling, but not on NF-B activation. DISCUSSION Our results reveal a unique property of BMPs that inhibit TNF-mediated apoptosis in a NF-B-independent manner. Furthermore, we found that BMP-mediated survival was dependent on Smad signaling, indicating that other survival pathways can replace NF-B antiapoptotic function to suppress TNF-mediated apoptosis. BMPs suppressed TNF-mediated apoptosis by inhibiting caspase-8 activation. Because NF-Binducible genes including TRAF-1, TRAF-2, c-IAP-1, c-IAP-2, c-Flip, and NDED play a critical role in inhibition of TNFmediated caspase-8 activation (4), we examined whether the expression of those genes was regulated by BMPs. Western blot or Northern blot analysis found that none of these genes were induced by BMPs. Additionally, we also found that the expression of Bcl-2 family proteins, including Bcl-2, Bcl-x L , Bax, Bad, and Bik, was not modulated by BMPs. 2 Thus, these results indicate that BMP/Smad signaling may regulate other antiapoptotic molecules that have yet to be identified. However, we cannot rule out the possibility that BMP/Smad signaling may posttranslationally modify antiapoptotic proteins.
BMPs are multifunctional cytokines that are widely expressed and play important role in morphogenesis and development (23)(24). Contrary to our results, several reports indicate that BMPs function in a proapoptotic manner perhaps required for interdigit cell death (30,31). BMPs have been found to induce apoptosis in multiple myeloma (32). These seemingly contradictory findings may be due to cell type-spe-cific effects. However, it is of interest that both C2C12 cells and embryonic fibroblasts used for our studies have the potential to differentiate into osteoblasts upon BMP stimulation (28,29). The process of osteoblast differentiation may be a stress stimulus that requires a survival signal to prevent the cell from undergoing apoptosis (33). Consistent with our results, Yang et al. (39) have found that deletion of Smad5 led to apoptosis in mesenchymal cells and angiogenesis defects during early mouse development by unknown mechanisms. Given the fact that Smad5 is a key component of BMP signaling (35,36), the result implies that BMP/Smad signaling may play a role in cell survival during embryogenesis and development. In conclusion, our results suggest that BMPs can prevent apoptosis as well as stimulate osteoblast differentiation and provide new insight into biological functions of BMPs.
FIG. 6. BMP/Smad signaling inhibits TNF-mediated apoptosis in IKK؊/؊ mouse embryonic fibroblasts. A, IKKϪ/Ϫ cells were pretreated or not pretreated with BMP-2 for 2 h and then killed by TNF for 24 h. The cell viability was determined as described in the Fig. 1 legend. The assays were performed in triplicate. The results represent the mean value from three independent experiments. Statistical differences between each group were determined by Student's t test. ‫,ء‬ p Ͻ 0.01. B, Smad7 abolished BMP-mediated survival. Cell transfection was performed as described in the Fig. 4 legend. Twenty-four h after transfection, cells were not pretreated or pretreated with BMP-2 (200 ng/ml) and then killed by TNF (20 ng/ml) for 24 h. The EGFP-positive cells were counted by fluorescence microscopy. The assays were performed in triplicate, and the results represent the mean value from three independent experiments. Statistical differences between each group were determined by Student's t test. ‫,ء‬ p Ͻ 0.01.