14-3-3 Interacts Directly with and Negatively Regulates Pro-apoptotic Bax*

The Bcl-2 family of proteins comprises well characterized regulators of apoptosis, consisting of anti-apoptotic members and pro-apoptotic members. Pro-apoptotic members possessing BH1, BH2, and BH3 domains (such as Bax and Bak) act as a gateway for a variety of apoptotic signals. Bax is normally localized to the cytoplasm in an inactive form. In response to apoptotic stimuli, Bax translocates to the mitochondria and undergoes oligomerization to induce the release of apoptogenic factors such as cytochromec, but it is still largely unknown how the mitochondrial translocation and pro-apoptotic activity of Bax is regulated. Here we report that cytoplasmic protein 14-3-3θ binds to Bax and, upon apoptotic stimulation, releases Bax by a caspase-independent mechanism, as well as through direct cleavage of 14-3-3θ by caspases. Unlike Bad, the interaction with 14-3-3θ is not dependent on the phosphorylation of Bax. In isolated mitochondria, we found that 14-3-3θ inhibited the integration of Bax and Bax-induced cytochromec release. Bax-induced apoptosis was inhibited by overexpression of either 14-3-3θ or its mutant (which lacked the ability to bind to various phosphorylated targets but still bound to Bax), whereas overexpression of 14-3-3θ was unable to inhibit apoptosis induced by a Bax mutant that did not bind to 14-3-3θ. These findings indicate that 14-3-3θ plays a crucial role in negatively regulating the activity of Bax.

Regulation of programmed cell death, or apoptosis, is essential for normal development and for the maintenance of homeostasis in most metazoans. Various apoptotic signals eventually converge into a common death mechanism, in which members of the cysteine protease family (known as caspases) are activated and cleave various cellular proteins. In mammals, the mitochondria play an essential role in apoptosis by releasing apoptogenic factors including cytochrome c, Smac/ Diablo, and Omi/HtrA2 from the intermembrane space into the cytoplasm (1)(2)(3). Once in the cytoplasm, cytochrome c binds to Apaf-1, a mammalian homologue of Ced-4, that recruits and activates initiator caspase-9, which subsequently activates effector caspase-3/caspase -7 (4,5), whereas Smac/Diablo and Omi/HtrA2 facilitate caspase activation by interacting with and inhibiting IAPs, the endogenous caspase inhibitor family (3,6,7).
The Bcl-2 family of proteins includes the best characterized regulators of apoptosis, comprising anti-apoptotic members, including Bcl-2 and Bcl-x L , and pro-apoptotic members that include multi-domain Bax and Bak and various single-domain BH3-only proteins (1,8). Proteins of this family directly regulate the release of mitochondrial apoptogenic factors. Many of the pro-apoptotic family members, such as Bax, Bid, Bad, Bim, and Bmf, are localized in the cytoplasm, and apoptotic stimulation results in their translocation to the mitochondria and induction of the release of apoptogenic factors, probably by inactivating anti-apoptotic members of the family and activating multi-domain members like Bax and Bak (9 -12). Translocation of the BH3-only proteins appears to involve various post-translational modifications. For instance, cytoplasmic Bid is cleaved by caspase-8 and then undergoes translocation to the mitochondria (13,14). Dephosphorylation by calcineurin frees Bad from cytosolic 14-3-3 and allows it undergo translocation to the mitochondria (15). Although the mechanism involved is still unknown, Bim and Bmf are freed from microtubular dynein motor complexes and myosin V actin motor complexes, respectively, during certain forms of apoptosis (16,17). Bax has also been shown to undergo translocation and integration into the mitochondrial membrane during apoptosis (18 -22), and the translocation process has been suggested to involve a conformational change of the Bax molecule, especially exposure of the C terminus (20,21,23). It has also been reported that translocation of Bax to the mitochondria is enhanced by caspases (20) or by intracellular alkalization (24) and is negatively regulated by Bcl-2 through a still unidentified mechanism (19,21,22). The 14-3-3 proteins (seven isomers in mammals: ␤, ␥, ⑀, , , /, and ) are highly conserved cytoplasmic molecules that form homodimers and heterodimers and interact with various cellular proteins. These proteins seem to control various cellular processes by sequestering regulatory molecules (25). The 14-3-3 proteins have also been implicated in signaling for apoptosis through interaction with apoptotic molecules such as Bad (26), ASK1 (27), and FKHRL1 (28). Furthermore, 14-3-3⑀ and are known to act as mitochondrial import stimulation factors (29) and appear to play a crucial role in intracellular protein trafficking, although the precise mechanism by which these isomers of 14-3-3 participate in protein translocation is not yet understood.
In the present study, we showed that 14-3-3 protein was bound to Bax in the cytoplasm of living cells and that Bax underwent dissociation from this protein by caspase-independent and -dependent mechanisms during apoptosis to induce apoptotic changes of the mitochondria, indicating that 14-3-3 plays a crucial role in the negative regulation of Bax activity in living cells.
Protein Purification-Recombinant His-tagged human Bax was expressed and purified as described previously (31). Recombinant human 14-3-3⑀, , and were expressed as GST fusion proteins in Escherichia coli (strain DH5␣) and were purified on a glutathione-Sepharose column. 14-3-3 was released from GST by cleavage with thrombin and purified to homogeneity by MonoQ chromatography. Mock proteins were produced by the same method using empty plasmids. In some experiments, GST-14-3-3 was used without cleavage of GST. Recombinant caspase-3, -7, and -8 were expressed as His-tagged proteins in E. coli (strain DH5␣) and purified on a Ni ϩ -nitrilotriacetic acid column. One unit was defined as the amount of enzyme that released 1 nmol of 7-amino-4-methylcoumarin in a buffer (50 mM Tris/HCl, pH 7.4, 1 mM EDTA, and 1 mM EGTA) containing 100 M of substrate (Ac-DEVD-MCA, Ac-IETD-MCA, or Ac-VEID-MCA) over 10 min at 30°C. All of the proteins were dissolved in a buffer composed of 20 mM Tris/HCl (pH 7.4), 2 mM MgCl 2 , and 1 mM dithiothreitol. Mouse Bax and its mutants were produced using an in vitro translation method. Briefly, DNAs were transcribed/translated with a TNT T7 transcription/translation kit (Promega, Japan) in the presence of [ 35 S]methionine according to the manufacturer's instructions.
Analysis of Protein-Protein Interaction-For immunoprecipitation experiments, HeLa and NIH3T3 cells were incubated with 2 mM DTBP (a protein cross-linker) for 30 min. Then the cells were lysed and sonicated in lysis buffer (50 mM Tris/HCl, pH 7.4, 142.5 mM KCl, 5 mM MgCl 2 , 1 mM EGTA, and 0.2% Nonidet P-40) containing proteinase inhibitors. To investigate the interaction with exogenous Bax and Bad, 293T cells were transiently transfected with the expression plasmids using LipofectAMINE in the presence of zVAD-fmk (100 M) to prevent Bax-or Bad-induced apoptosis. Then the cells were lysed, and the lysates were subjected to immunoprecipitation with the indicated antibodies, and the precipitates were analyzed by Western blotting. To detect binding between purified proteins, recombinant proteins were incubated for 8 h with either GST-14-3-3 proteins or GST alone in 100 l of the lysis buffer, and then these proteins were incubated with glutathione-Sepharose for 3 h. After brief centrifugation, the beads were washed and resuspended in the SDS-PAGE sample buffer, as described elsewhere (32). The proteins were analyzed by Western blotting and autoradiography.
Surface Plasmon Resonance-The affinity between Bax and 14-3-3 proteins was measured by surface plasmon resonance using a Bia-core2000 (Biacore). Equivalent molar amounts of GST-mock and GST-14-3-3 proteins were immobilized on the sensor chip (CM5; Pharmacia Corp.) by the amine-coupling method. Bax was added as the analite, and the affinity was calculated from the difference between the resonance units with GST-14-3-3 proteins and those with GST-mock.
Cell Fractionation-Cell fractionation was performed using digitonin, as described previously (22). Briefly, after washing twice with phosphate-buffered saline, the cultured cells were collected and treated with 10 M digitonin for 5 min at 37°C. The cytosolic and organellar fractions were separated by centrifugation and lysed with RIPA buffer. As a result, more than 92% of cytosolic protein was recovered in the supernatant, and more than 95% of mitochondrial protein was localized to the pellet. The heavy membrane fraction enriched for mitochondria was prepared as follows. The cells were washed twice in phosphatebuffered saline, resuspended in isotonic buffer (20 mM potassium Hepes, pH 7.4, 1.5 mM MgCl 2 , 10 mM KCl, and 250 mM sucrose), and then homogenized with a Dounce homogenizer. After separation of nuclei and unbroken cells by centrifugation at 600 ϫ g for 10 min, the post-crude nuclear supernatant was centrifuged at 10,000 ϫ g for 10 min to collect the heavy membrane fraction.
Assessment of the Integration of Bax into the Mitochondrial Membrane-Heavy membrane fractions enriched for mitochondria from cells or isolated rat mitochondria were incubated in 0.1 M Na 2 CO 3 (pH 11.5), and then were centrifuged at 200,000 g for 45 min to separate the supernatant and pellet as described elsewhere (21).
Analysis of Bax Translocation and Cytochrome c Release in Vitro-Mitochondria were prepared from the livers of male Donryu rats in Mt-A buffer (0.3 M mannitol, 10 mM potassium Hepes, pH 7.4, 0.1% fatty acid-free bovine serum albumin), as described previously (33). Recombinant 14-3-3 (the indicated amounts) and rBax (1 g) were preincubated for 30 min at 25°C and then were added to the mitochondria (100 g) and incubated for a further 3 min at 25°C in 100 l of Mt-B buffer (Mt-A buffer plus 100 M potassium phosphate and 4.3 mM succinate). Next, the mixture was then centrifuged to collect the mitochondria, and aliquots of mitochondria or supernatant were analyzed by Western blotting using anti-Bax antibodies. To detect the release of cytochrome c, mitochondria were centrifuged, and the supernatant was analyzed by Western blotting with an anti-cytochrome c antibody.
In Vitro Assay of 14-3-3 Cleavage by Caspases-Cytosolic fractions from healthy HeLa cells or recombinant 14-3-3 were incubated for 5 h at 37°C with or without caspases and in the presence or absence of 200 M zVAD-fmk. Then the cleavage of 14-3-3 was detected by Western blotting.
Analysis of Cell Death-293T cells were transiently transfected with human Bax DNA (0.2 g) with or without DNA expressing human 14-3-3 or its mutants (0.5 g), plus 0.1 g of the green fluorescent protein (GFP) expression construct (pEGFP-N1; Clontech). Transfected cells were incubated for 24 h at 37°C and stained with 1 M Hoechst 33342, after which the extent of apoptosis was calculated as the percentage of GFP-positive cells showing nuclear fragmentation relative to all GFP-positive cells.

RESULTS
Bax Interacts with 14-3-3-Although translocation and integration of cytoplasmic Bax into the mitochondrial membrane is a critical step for its pro-apoptotic activity, the mechanism of action is poorly understood. To improve our understanding of the regulation of Bax, we searched for a molecule that interacted with Bax, modulated its activity, and found that Bax was bound to protein 14-3-3 in HeLa cells (Fig. 1a). The same interaction between Bax and 14-3-3 was also observed in NIH3T3 cells (Fig. 1b). Because there are several isoforms of 14-3-3 (25), we next tested the interaction of Bax with other isoforms. As shown in Fig. 1a, Bax was also bound to 14-3-3 and 14-3-3⑀ in HeLa cells, whereas there was no interaction with 14-3-3␤ or 14-3-3␥ (data not shown). Furthermore, it has recently been reported that Bax binds to 14-3-3 (34). Although the interaction between Bax and 14-3-3 was initially detected in the presence of a protein cross-linker (Fig. 1a, left panel), a similar level of binding was observed in the absence of the cross-linker (Fig. 1a, right panel). Recombinant His-tagged Bax (rBax) showed binding to GST-fused 14-3-3, , and ⑀ but not to GST (Fig. 1c), indicating that Bax directly interacted with these 14-3-3 isoforms. Furthermore, surface plasmon resonance analysis revealed that 14-3-3, , and ⑀ all had a comparable affinity for rBax (Fig. 1d). Estimation of the amount of each endogenous 14-3-3 isoform in HeLa cells by comparison with recombinant isoforms on Western blots revealed that 10 g of HeLa cell lysate contained ϳ35, 22, and 10 ng of 14-3-3, , and ⑀, respectively (Fig. 1e). According to these findings, although Bax interacted with 14-3-3, , and ⑀, 14-3-3 was the major isoform in HeLa cells, so we studied its role further. Although Bad is known to bind to 14-3-3 (26), we were unable to detect any interaction between 14-3-3 and the other proapoptotic Bcl-2 family members Bid or Bak, either by immunoprecipitation or by the surface plasmon resonance method (data not shown).
Both the N-and C-terminal Regions of Bax Are Required for Interaction with 14-3-3-We then attempted to determine the regions of Bax involved in binding to 14-3-3 by employing an in vitro interaction assay using 35 S-labeled, HA-tagged mouse Bax mutants and GST-14-3-3, because the level of expression of the Bax mutants varied considerably in transfection experiments. Wild-type Bax and three of its deletion mutants (⌬␣1 (lacking ␣-helix 1), ⌬BH3 (lacking the BH3 region), and ⌬␣5/6 (lacking the channel-forming ␣-helices 5 and 6)) showed binding to GST-14-3-3, whereas Bax⌬N and ⌬C (lacking the Nterminal 20 amino acids and C-terminal 22 amino acids, respectively) did not bind to GST-14-3-3 (Fig. 2), suggesting that both the N-and C-terminal regions of Bax were involved in this binding process. Although it was reported that some detergents, such as Nonidet P-40, could enhance the conformational changes of Bax and increase its homodimerization and heterodimerization with other Bcl-2 family members (35), the interaction between Bax and 14-3-3 was decreased rather than enhanced by addition of Nonidet P-40 (data not shown).
Interaction of Bax with 14-3-3 Is Independent of Bax Phosphorylation-The 14-3-3 proteins bind to various phosphorylated proteins, such as Raf-1 and Bad, via phosphorylated serine residues (26, 36), but these proteins are also known to bind to several nonphosphorylated proteins (37). Therefore, we tested whether phosphorylation of Bax was involved in its were preincubated with 2 mM DTBP (a cleavable protein cross-linker), and the lysates were immunoprecipitated (IP) with anti-Bax polyclonal antibody (N20) (␣Bax) or normal rabbit IgG. The immune complexes were analyzed by Western blotting using anti-14-3-3 antibodies specific for the indicated isoform. The same experiment was also performed without DTBP in HeLa cells (a, right panel). lysate indicates the portion ( 1 ⁄10) of the total lysate that was subjected to immunoprecipitation. c and d, direct interaction between Bax and several 14-3-3 isoforms (, , and ⑀) with comparable affinities. c, recombinant Bax (rBax, 2 g) was incubated with 2 g of the indicated GST-14-3-3 proteins or the equivalent amount of GST-mock protein for 8 h. Then GSH-Sepharose was added for 3 h and collected by centrifugation, after which bound rBax was analyzed by Western blotting. total indicates the total amount of rBax used. d, the indicated amount (15 g or 30 g) of rBax or bovine serum albumin was run over a chip containing immobilized GST-mock protein, GST-14-3-3, , or ⑀, and protein interactions were measured by surface plasmon resonance as described under "Experimental Procedures." e, the amount of each 14-3-3 isoform in HeLa cells. Lysates from healthy HeLa cells (10 g) and the indicated GST-14-3-3 isoforms (25 ng each) were analyzed by Western blotting using antibodies specific for 14-3-3⑀ (left panel), 14-3-3 (middle panel), and 14-3-3 (right panel). The amount of each of the 14-3-3 isoforms in 10 g of lysate was estimated by comparison with the GST-14-3-3 proteins using densitometric analysis and is shown below the blots (in nanograms).

FIG. 2. Regions of Bax essential for interaction with 14-3-3.
The indicated mutants of mouse Bax were produced by in vitro translation in the presence of [ 35 S]methionine and were incubated with GST-14-3-3 or GST-mock protein. After GST-14-3-3 and GST-mock protein were precipitated with GSH-Sepharose as described under "Experimental Procedures," Bax bound to GST-14-3-3 or GST-mock was analyzed by SDS-PAGE followed by autoradiography. A diagram of the Bax deletion mutants is also shown. These deletion mutants retained the regions shown by horizontal lines. ␣1-␣7 indicate the possible helices retained by Bax.
interaction with 14-3-3. As shown in Fig. 3a, although Bax was bound to 14-3-3, we could not detect any phosphorylation of Bax in 293T cells when overexpressed Bax was labeled with [ 32 P]orthophosphate, a result consistent with previous reports (38). Under the same experimental conditions, we readily detected phosphorylation of Bad (Fig. 3a), which is known to be phosphorylated before binding to 14-3-3 (26). Furthermore, the immunoprecipitated Bax did not react with antibodies specific for phosphoserine or phosphothreonine (data not shown). These results indicated that phosphorylation of Bax was not necessary for interaction with 14-3-3. To further confirm that phosphorylation of Bax did not play an essential role in the interaction with 14-3-3, we examined the binding of Bax to a mutant of 14-3-3 (⌲49⌭/V176D) that had lost the ability to bind to various target phosphoproteins, including Raf-1 and ASK1 (27). As shown in Fig. 3b, whereas wild-type 14-3-3 was co-immunoprecipitated with both Bax and FLAG-Bad, 14-3-3 K49E/V176D was co-immunoprecipitated with Bax but not with FLAG-Bad, supporting our hypothesis that phosphorylation of Bax was not necessary for interaction with 14-3-3.
Bax Is Negatively Regulated by 14-3-3 and Dissociates during Apoptosis in Both a Caspase-dependent and Caspase-independent Manner-To obtain some insight into the biological significance of the interaction of Bax with 14-3-3, we next examined whether this interaction was altered during the apoptotic process. As shown in Fig. 4a, treatment with VP16 (etoposide) caused the amount of 14-3-3 interacting with Bax to decrease markedly, and a large fraction of Bax was translo-cated to the mitochondria (Fig. 4b) with the release of cytochrome c (Fig. 4a). To test whether Bax in the mitochondrial fraction was stably integrated into the mitochondrial membrane, mitochondrial fractions were treated with an alkaline solution (pH 11.5) that only saved Bax, showing stable integration into the membrane. Before incubation with VP16, a very small amount of Bax was found in the heavy membrane fraction (Fig. 4b), half of which was stably integrated into the mitochondrial membrane (Fig. 4c). Note that a much larger amount of ϪVP16 sample was analyzed than that of ϩVP16 in Fig. 4c. On the other hand, the majority of Bax was found in the heavy membrane fraction after VP16 treatment (Fig. 4b), most of which was stably integrated into the membrane (Fig. 4c). In the presence of zVAD-fmk, which completely inhibited caspase Heavy membrane (HM) and cytosolic fractions were prepared by differential centrifugation without alkaline treatment. Then the fractions were analyzed by Western blotting using anti-Bax antibody. c, integration of Bax, but not 14-3-3, into the mitochondrial membrane. HeLa cells were incubated for 24 h with 200 M VP16 in the absence or presence of 200 M zVAD-fmk. Heavy membrane (HM) fractions were prepared by differential centrifugation, incubated in pH 7.5 buffer (Ϫ) or in 0.1 M Na 2 CO 3 (pH 11.5) (ϩ) on ice for 30 min, and then centrifuged at 200,000 g for 45 min to yield a supernatant (S) and a pellet containing heavy membranes (P). The fractions were analyzed by Western blotting using anti-Bax and anti-14-3-3 antibodies. Equivalent amounts of the HM fraction from ϩVP16 and ϩVP16 ϩzVAD Ϫfmk cultures were analyzed, but a much larger amount of the fraction from ϪVP16 cultures was assessed. d, dissociation of Bax from 14-3-3 induced by the cytosol of apoptotic cells. GST-14-3-3 (2 g) and rBax (5 g) were preincubated for 8 h at 4°C, and then the mixture was incubated with glutathione-Sepharose for 3 h. After brief centrifugation, the beads retaining 14-3-3-Bax complex were incubated with the indicated cytosol (10 g) for 12 h at 25°C. After brief centrifugation, the beads were washed. Then the combined supernatants (sup) and the beads resuspended in sample buffer were analyzed by Western blotting using anti-Bax and anti-14-3-3 antibodies. total indicates the amount of rBax in the pellet after incubation with glutathione-Sepharose. activation, dissociation of Bax from 14-3-3 was only partly inhibited (Fig. 4a), indicating that dissociation occurred in both a caspase-independent and caspase-dependent manner. As shown in Fig. 4c, integration of Bax into the mitochondrial membrane was also partly inhibited by zVAD-fmk, whereas translocation of Bax to the mitochondria was not affected by this caspase inhibitor (Fig. 4b). Note that the amount of caspase-independent dissociation of Bax-14-3-3 was well correlated with the caspase-independent mitochondrial integration of Bax (Fig. 4, a and c).
The findings obtained using cell lysates were similar to those obtained with living cells. As shown in Fig. 4d, lysates from VP16-treated cells were more efficient at causing Bax to dissociate from 14-3-3 than lysates from normal cells. Interestingly, this dissociation was partially inhibited in the presence of the caspase inhibitor zVAD-fmk, indicating that dissociation of Bax from 14-3-3 occurred via both caspase-dependent and -independent mechanisms in the cell lysates (Fig. 4d), as it did in living cells (Fig. 4a). All of these results suggested that 14-3-3 had a role in the sequesteration of Bax.
The dissociation of Bax from 14-3-3 during apoptosis suggested that 14-3-3 was a negative regulator of Bax. To test this possibility, we examined whether 14-3-3 affected the mitochondrial translocation of Bax using isolated mitochondria. As shown in Fig. 5, the addition of recombinant 14-3-3 (or ) protein inhibited the integration of Bax into the mitochondrial membrane (Fig. 5a) as well as Bax-induced release of cytochrome c (Fig. 5b).
Bax Dissociates from 14-3-3 under Basic and Acidic Conditions-It has been demonstrated that in the early phase of apoptosis induced by a variety of stimuli, including cytokine deprivation, cytoplasmic alkalization occurs and induces a conformational change of Bax that results in its integration into the mitochondria (24). Therefore, we examined whether alkal-

FIG. 5. Inhibition of the mitochondrial integration of Bax by 14-3-3.
GST-14-3-3, GST-14-3-3 (50 g), and the equivalent amount of GST-mock were incubated with or without rBax (25 g) for 3 h and then were incubated with the mitochondria (1 mg) for 20 min at 25°C. Supernatants and pellets were obtained by centrifugation. The pellets were incubated in 0.1 M Na 2 CO 3 (pH 11.5) on ice for 30 min, followed by centrifugation, and then each pellet was analyzed by Western blotting using anti-Bax antibody (a). Cytochrome c in the supernatant was also analyzed by Western blotting using an anti-cytochrome c antibody (b). total represents the total amount of rBax used (a) or an equivalent aliquot of mitochondria (b).
ization had an influence on the interaction of Bax with 14-3-3. As shown in Fig. 7a, the interaction of Bax with 14-3-3 was weaker at pH 8.0 than at pH 7.5. Treatment of Bax, but not 14-3-3, with an alkaline solution (pH 8.0) decreased their affinity (Fig. 7a), supporting the earlier finding that alkalization induced a conformational change of Bax (24). Dissociation of Bax from 14-3-3 at an alkaline pH (Fig. 7b) implies that cytoplasmic alkalization during apoptosis may be one of the initial caspase-independent mechanisms promoting dissociation of the complex between Bax and 14-3-3. Dissociation of Bax from 14-3-3 was also observed at an acidic pH of 6.5 (Fig.  7c). It has been reported that cells show cytoplasmic acidification in the early phase of apoptosis induced by staurosporine and anti-Fas antibodies (39), so acidification may also be a trigger for the dissociation of Bax from 14-3-3.
14-3-3 Inhibits Bax-and Fas-induced Apoptosis-Finally, we examined the physiological role of the interaction between 14-3-3 and Bax in the regulation of apoptosis. If 14-3-3 negatively regulates Bax, overexpression of 14-3-3 could be expected to inhibit Bax-induced apoptosis. As shown in Fig. 8, (a,  left panel, and b), apoptosis induced by transfection of Bax DNA was significantly reduced by co-transfection of 14-3-3 DNA, and integration of Bax into the mitochondrial membrane was also inhibited. Importantly, a mutant form of 14-3-3 (K49E/V176D) that bound to Bax but not to phosphorylated targets, including Bad (Fig. 3b), also inhibited Bax-induced apoptosis (Fig. 8a, left panel), suggesting that the inhibition was due to direct association with Bax and not to the influence of various other 14-3-3-binding proteins, including Bad, Raf-1, and forkhead protein. Furthermore, a caspase-cleaved mutant of 14-3-3 (1-239) with a weak affinity for Bax (Fig. 6f) caused less inhibition of Bax-induced apoptosis, whereas another mutant (14-3-3-␣1-6) that did not bind to Bax (data not shown) could not inhibit such apoptosis (Fig. 8a, middle panel). Incomplete suppression of Bax-induced apoptosis by overexpression of 14-3-3 was probably due to the abundance of endogenous 14-3-3. Moreover, as shown in Fig. 8a (right panel), 14-3-3 did not inhibit apoptosis induced by Bax⌬N, to which it did not bind (Fig. 2). Consistent with the inability of 14-3-3 to sequester Bax⌬N, we found that Bax⌬N showed efficient translocation to the mitochondria (Fig. 8c) and induced more apoptosis than wild-type Bax (Fig. 8a). These results suggested that 14-3-3 inhibits Bax-induced apoptosis in an interaction-dependent manner.
Next, we examined whether overexpression of 14-3-3 could inhibit apoptosis induced by an anti-Fas antibody. Because we found that caspase-8 cleaved 14-3-3 to release Bax, we also tested a caspase-resistant (noncleavable) mutant of 14-3-3 (14-3-3D239A). As shown in Fig. 8d, overexpression of 14-3-3D239A significantly inhibited apoptosis induced by anti-Fas antibody, suggesting that the cleavage of 14-3-3 by caspases could facilitate Fas-induced apoptosis. DISCUSSION Bax is mainly found in cytoplasmic and/or peri-mitochondrial locations in living cells, and apoptotic stimulation causes its stable integration into the mitochondrial membrane, along with the induction of cytochrome c release (18 -22, 31, 40, 41). However, it is still poorly understood how Bax remains inactive in healthy cells. Although it has been suggested that Bax exists as a monomer in the cytoplasm of healthy cells and forms dimers or oligomers on the mitochondrial membrane during apoptosis (10,21), the present study clearly showed that a significant fraction of Bax interacts with 14-3-3 in living cells and that this interaction negatively regulates Bax by seques- FIG. 7. Caspase-independent dissociation of Bax from 14-3-3. a, decrease of Bax binding to 14-3-3 by treatment at pH 8.0. Upper panel, the interaction between rBax and GST-14-3-3 at pH 7.5 or 8.0 was analyzed using surface plasmon resonance as in Fig. 1d. Lower panel, GST-14-3-3 (2 g) and rBax (5 g) were preincubated for 30 min at 4°C at the indicated initial pH and then were mixed for 8 h at 4°C in buffer at the indicated terminal pH. The mixture was then incubated with glutathione-Sepharose for 3 h. After a brief centrifugation, the beads were washed, and the amount of rBax and 14-3-3 on the beads was analyzed by Western blotting. total indicates the total amount of rBax used. b and c, the pH-dependent dissociation of Bax from 14-3-3. GST-14-3-3 (2 g) and rBax (5 g) were preincubated at 4°C for 8 h, and then the mixture was incubated with glutathione-Sepharose for 3 h. After brief centrifugation, the beads were treated in buffer (pH 7.5, pH 8.0 (b), or pH 6.5 (c)) at 4°C for 12 h. The amount of rHis-Bax and 14-3-3 on the beads was then analyzed by Western blotting. tering it to the cytoplasm. Among seven isoforms, 14-3-3⑀, 14-3-3 (this study), and 14-3-3 (34) also bind to Bax and probably play a redundant role.
It has been suggested that Bax undergoes a conformational change during apoptosis, on the basis of its increased susceptibility to proteolytic cleavage (20) and binding with some antibodies (42). These changes can be explained by our proposal that Bax dissociates from 14-3-3 during apoptosis. Our finding that both N-and C-terminal regions of Bax were required for its interaction with 14-3-3 is also consistent with previous observations that translocation of Bax to the mitochondria is stimulated by N-terminal deletion (Fig. 8c and Ref. 20) as well as by mutation of charged residues in the N-and C-terminal regions (23,24). We also showed that Bax dissociates from 14-3-3 under both alkaline and acidic conditions, which is consistent with the previous observation that translocation of Bax to the mitochondria is enhanced at an alkaline pH (24). Taken together, it seems likely that a significant fraction of Bax is sequestered by cytoplasmic 14-3-3 in healthy cells, whereas apoptotic stimuli cause the dissociation of Bax from 14-3-3 and translocation to the mitochondria. The 14-3-3 proteins are highly conserved cytoplasmic molecules that interact with various cellular proteins and that are thought to be involved in the regulation of various cellular processes, including apoptotic signal transduction (25). It has been reported that different isoforms of 14-3-3 sequester different pro-apoptotic molecules through a phosphorylated serine residue on the target molecule and thus inhibit apoptosis (e.g. for ASK1 (27), for Bad (26), and for FKHRL1 (28)). The different 14-3-3 proteins therefore appear to protect cells from apoptosis at various steps of the death signaling pathway by sequestering different pro-apoptotic proteins (37). Most of the 14-3-3-binding proteins interact with 14-3-3 via phosphorylated serine or threonine residue (25,36). In contrast, we showed that the interaction of Bax with 14-3-3 occurs in a phosphorylation-independent manner, based on the observations that phosphorylation of Bax was undetectable (Fig. 3a) and that Bax still bound to a mutant form of 14-3-3 (⌲49⌭/ V176D) lacking the ability to bind to various phosphoprotein targets (Fig. 3b). It has been reported that 14-3-3 also interacts with nonphosphorylated proteins such as ADP-ribosyltransferase Exoenzyme S (ExoS) from Pseudomonas aeruginosa (43). Our preliminary study suggested that a region from ␣-helix 7 to the C terminus of 14-3-3, the three-dimensional structure of which could not be identified (possibly because of its high flexibility) (44,45), was crucial for the interaction with Bax (data not shown). ␣-Helix 7 and the more C-terminal ␣-helix 8 comprise the box-1 region, where phosphorylated target proteins mainly bind by hydrophobic interaction (36,46,47), and this region probably undergoes a conformational change upon binding of a phosphoprotein to 14-3-3. It is therefore conceivable that Bax binds to 14-3-3 in healthy cells, and unidentified phosphoprotein(s) may interact with 14-3-3 to release Bax after an apoptotic stimulus is delivered.
We showed that Bax dissociates from 14-3-3 by caspaseindependent and -dependent mechanisms. For the caspaseindependent process, one possible trigger is alteration of cytosolic pH (acidification or alkalization), which has been shown to occur in the early phase of apoptosis (24,39), and indeed we found that this induced the dissociation of Bax from 14-3-3. Bax also underwent dissociation after the direct cleavage of 14-3-3 by caspases. The caspase-dependent dissociation of Bax from 14-3-3 and subsequent integration of Bax into the mitochondrial membrane probably represents a positive feedback loop for death signal transduction. In death receptor-mediated apoptotic signaling, however, casapase-8 (which cleaves 14-3-3) is activated upstream of the mitochondria, so the caspasedependent dissociation of Bax from 14-3-3 acts as an initial trigger for apoptotic mitochondrial changes. When HeLa cells were treated with anti-Fas antibody, cleavage of 14-3-3, dissociation of Bax from 14-3-3, and integration of Bax into the mitochondria were observed simultaneously (data not shown). Furthermore, overexpression of caspase-uncleavable 14-3-3 mutant (D239A) conferred stronger resistance to Fas-mediated apoptosis than overexpression of wild-type 14-3-3 (Fig. 8d), suggesting that cleavage of 14-3-3 by caspase-8 is one of the crucial steps in Fas/TNF-mediated apoptosis. An essential role of Bax in Fas-mediated apoptosis has been shown by genetargeting studies, because hepatocytes of Bax/Bak-deficient mice are resistant to Fas-mediated apoptosis, whereas hepatocytes from Bax-deficient, Bak-deficient, and wild-type mice are all equally sensitive to Fas-mediated apoptosis (11).
In summary, we investigated the mechanisms by which translocation of Bax into the mitochondrial membrane is regulated and found that 14-3-3 plays a crucial role in sequestering Bax to the cytoplasm, where apoptotic stimulation causes it to release Bax in both a caspase-independent and -dependent manner. Further studies are required to identify the trigger that induces caspase-independent dissociation of Bax from 14-3-3 and the signals that enhance translocation of Bax to the mitochondria.