Cleavage of 14-3-3 protein by caspase-3 facilitates bad interaction with Bcl-x(L) during apoptosis.

The 14-3-3 epsilon protein was identified as one of the caspase-3 substrates by the modified yeast two-hybrid system. The cellular 14-3-3 epsilon protein was also cleaved in response to the treatment of apoptosis inducers in cultured mammalian cells. Asp238 of the 14-3-3 epsilon protein was determined as the site of cleavage by caspase-3. The affinity of the cleaved 14-3-3 mutant protein (D238) to Bad, a death-promoting Bcl-2 family protein, was lower than that of wild type or the uncleavable mutant 14-3-3 epsilon protein (D238A). However, Bad associated with the cellular Bcl-x(L) more effectively in human 293T cells co-expressing Bad with the truncated form of the 14-3-3 epsilon protein (D238) than in control cells co-expressing Bad with wild type or the uncleavable mutant 14-3-3 epsilon protein (D238A). The present study suggests that the cleavage of 14-3-3 protein during apoptosis promotes cell death by releasing the associated Bad from the 14-3-3 protein and facilitates Bad translocation to the mitochondria and its interaction with Bcl-x(L).

During the early events of apoptosis, mitochondria have been thought to act as central coordinators of cell death (1,2). Some apoptotic signal cascades induce mitochondrial membrane permeabilization under the control of Bcl-2-related proteins in the mitochondria. The loss of transmembrane potential in the mitochondria then induces the release of apoptotic activators such as cytochrome c, Smac/DIABLO, HtrA2/Omi, AIF, Endo G, and caspases from the mitochondria (3). Released cytochrome c activates caspase-9, binds to Apaf-1, and induces Apaf-1 oligomerization to form apoptosome (4). The apoptosome recruits and activates procaspase-9, which, in turn, activates inactive procaspase-3 into the active caspase-3, the executor molecule of apoptosis. Cytochrome c release was reported to be blocked by anti-apoptotic Bcl-2 or Bcl-x(L) and accelerated by pro-apoptotic Bax (5,6). Bcl-2 family proteins are key regulatory proteins that play critical roles in mediating the signal transduction path that leads to apoptosis. The multi-BH domain members of the Bcl-2 protein family either suppress (Bcl-2 or Bclx(L)) or promote (Bax or BAK) apoptosis, whereas "BH3-only members" (Bad or Bid) exclusively promote apoptosis (7,8).
Bad is a death-promoting BH3-only member of Bcl-2 family proteins and heterodimerizes with anti-apoptotic proteins such as Bcl-2 and Bcl-x(L). Survival kinases including Akt, protein kinase A (PKA), and ribosomal S6 kinase 1 (RSK1) phosphorylate serine residues of Bad (9 -12). Bad phosphorylation has been shown to be necessary for the release of Bad from its association with Bcl-x(L).
14-3-3 proteins are conserved dimeric regulators in eukaryotic cells. There are seven isotypes of 14-3-3 proteins in mammals and two in yeast. 14-3-3 proteins are known for their ability to bind multiple cellular protein ligands. As many as 100 cellular proteins have been found to interact with 14-3-3 proteins to date. 14-3-3 proteins require phosphorylation of their target proteins for the interaction (13,14). The large number and diversity of ligands for 14-3-3 proteins implicate a suggestion that 14-3-3 proteins are involved in many different cellular events, including signal transduction, cell cycle regulation, apoptosis, stress response, cytoskeleton organization, and malignant transformation (15). Despite the fact that the function of 14-3-3 proteins still remains unknown, there exist some general themes about the anti-apoptotic function of 14-3-3 proteins. 14-3-3 proteins have been implicated in signaling for apoptosis through interaction with pro-apoptotic molecules such as Bad (11,16,17), FKHRL1 (18), ASK1 (19), and Nur77 (20). Phosphorylated Bad by survival kinases interacts with 14-3-3 proteins resulting in the retention of Bad in the cytoplasm and the subsequent prevention of cytotoxic interaction with Bcl-x(L) at the mitochondrial membrane.
Caspases play important roles in the execution of apoptosis and inflammatory responses. A large set of cellular proteins has been surveyed for their ability to be cleaved by caspases, and various signaling proteins were found to be caspase substrates (21,22). In this study, we identified the 14-3-3⑀ protein as one of the caspase-3 substrates using a modified yeast twohybrid genetic system. We explored the possibility that the cleavage of 14-3-3 proteins by caspase-3 during apoptosis might contribute to cell death by preventing the association of 14-3-3 proteins with Bad and facilitating Bad translocation to the mitochondria in which Bad associates with Bcl-x(L).
Plasmids and cDNA Library-The B42 transcriptional activation domain of pJG4-5 (Clontech) was cloned into the XhoI and XbaI sites of * This work was supported in part by grants from the Life Phenomena and Function Research Group Program and the Molecular Medicine Research Group Program, Ministry of Science and Technology, South Korea. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. pEG202 (Clontech) to generate pSub. Human PARP 1 and its mutant, PARP D214A , were cloned into the XhoI and PvuII sites of pSub to generate pSub-PARP and pSub-PARP D214A . The BALB/c mouse brain cDNA library was constructed using Universal RiboClone cDNA synthesis system (Promega) and was inserted into pSub to generate pSub-cDNAs (23). The B42 transcriptional activation domain of pJG4-5 was deleted to generate the caspase expression vector pExp. Human caspase-1 or 3 was cloned into the KpnI and XhoI sites of pExp for the expression in yeast YM4271 cells. Human Bad was cloned into a mammalian GST fusion vector, pEBG, to produce GST-fused Bad. Human 14-3-3⑀ and its mutants, D238 and D238A, were cloned into the EcoRI and XhoI sites of cloning vector pcDNA3 (Invitrogen). Hemagglutinin was inserted into the BamHI and HindIII sites of pcDNA3 to generate the N-terminal HA-tagged 14-3-3⑀ and its mutants.
Yeast Transformation and Mating-Yeast EGY48 cells were transformed with p80lacZ (Clontech) bearing the LexA-responsive lacZ gene. EGY48 cells transformed with pSub-cDNA and YM4271 cells transformed with a human caspase expression vector pExp-C3 were mated on a single plate containing YPD-rich media (24). After incubation for 6 h at 30°C, the plate was replica-plated to the plates containing the selective media consisting of 40 mg/liter X-gal either with 2% dextrose or 2% galactose plus 1% raffinose.
Caspase Activity Assay-The activities of caspase-1 or 3 were measured using a fluorogenic peptide substrate, zVAD-AMC (Bachem) as described (25). Yeast YM4271 cells transformed with human caspase expression vector encoding caspase-1 or 3 were cultured in the media containing 1% raffinose for 12 h. The expression of caspases was induced by the addition of 2% galactose to the media. After incubation for 6 h, cell pellets were placed in liquid nitrogen and lysed using glass beads (Sigma) in a disruption buffer containing 20 mM Tris, pH 7.9, 10 mM MgCl 2 , 1 mM EDTA, 5% glycerol, and protease inhibitor mix (1 g/ml leupeptin, 1 g/ml pepstatin, and 50 g/ml phenylmethylsulfonyl fluoride). Cell lysate proteins (25 g) were incubated with 50 M zVAD-AMC in a standard interleukin-1␤-converting enzyme (ICE) reaction buffer (25). Cleavage of zVAD-AMC was monitored by a fluorescence spectrophotometer at an excitation/emission wavelength pair of 355 nm/460 nm.
Immunoblotting and Co-immunoprecipitation-COS cells were harvested 10 h after staurosporine treatment or 24 h after UVC irradiation. Cells were lysed in lysis buffer A containing 50 mM Tris, pH 7.5, 120 mM NaCl, 0.5% Nonidet P-40, 100 mM NaF, 200 M sodium orthovanadate, and 1 mM phenylmethylsulfonyl fluoride. Cellular 14-3-3⑀ protein was resolved by 12% SDS-PAGE, and cleavage of the 14-3-3⑀ protein and PARP was analyzed by immunoblotting using antibodies directed against the 14-3-3⑀ protein (Santa Cruz Biotechnology) and PARP (Santa Cruz Biotechnology). Effects of 14-3-3⑀ protein cleavage on Bad association with cellular Bcl-x(L) were examined in human 293T cells. Cells were co-transfected with GST-Bad and constructs of HA-14-3-3⑀ by the calcium phosphate method as described (26). After incubation for 24 h, cells were lysed in lysis buffer A. Bad and its associated proteins were then "pulled down" using glutathione-Sepharose 4B (Amersham Biosciences). HA-14-3-3⑀ protein and Bcl-x(L), which were associated with Bad, were analyzed by immunoblotting using antibodies directed against HA and Bcl-x(L) (Transduction Laboratories), respectively.
Apoptosis-Apoptosis was measured by fluorescence-activated cell sorter (FACS) analysis. Cells were harvested 24 h after transfection and fixed with cold ethanol. DNA contents were determined by FACS after propidium iodide staining. DNA contents of cells at sub-G 0 /G 1 were used as the index for the degree of apoptosis. Fig. 1A illustrates the strategy for identifying caspase substrates in the yeast genetic system. The cDNAs encoding caspase substrates were fused between the LexA DNA-binding domain and the B42 activation domain. Yeast EGY48 cells expressing the fusion products produced the blue ␤-galactosidase positive colonies under the LexA operator on plates containing the indicator X-gal. The transformants failed to express lacZ and grew as white ␤-galactosidase negative colonies when plasmids harboring the cDNA that encodes caspases were introduced by mating, and the fusion products were cleaved by caspases. In Fig. 1B, a known caspase-3 substrate, poly(ADP-ribose)polymerase, was examined for its ability to be cleaved by caspase-3 in the modified yeast two-hybrid system. Yeast cells expressing the fusion product containing PARP transcribed lacZ under the LexA operator and grew as white ␤-galactosidase negative colonies after the introduction of caspase-3 by mating. However, cells FIG. 1. Modified yeast two-hybrid system to identify caspase substrates. A, yeast cells were created that express the fusion products in which the LexA DNA-binding domain and the B42 transcriptional activation domain were linked by the mouse brain cDNA library. The bacterial lacZ gene was used as a LexA-dependent transcriptional reporter. Yeast cells expressing the fusion products grew as the blue ␤-galactosidase positive colonies on the plate containing the indicator X-gal, whereas the transformants that failed to express lacZ grew as the white ␤-galactosidase negative colonies. B, a plasmid carrying a fusion product in which PARP or caspase-resistant PARP D214A links the LexA DNA-binding domain and the B42 transcriptional activation domain was transformed into yeast cells. The expression of human caspases was induced by adding 2% galactose to the culture media. Note that yeast cells expressing PARP grew as the white ␤-galactosidase negative colonies after the induction of caspase-3, whereas the transformants expressing caspase-resistant PARP D214A grew as the blue ␤-galactosidase positive colonies after the induction of caspase-3. C, the expression of human caspase-1 or 3 under the GAL1 inducible promoter was induced in yeast cells by adding 2% galactose to the culture media. Each yeast cell lysate was incubated with a fluorogenic peptide substrate, zVAD-AMC (0.5 M), and the activities of caspases in YM4271 cells were fluorometrically measured (excitation/emission of 355 nm/ 460 nm). Ac-YVAD-CHO (0.5 M) and Ac-DEVD-CHO (0.5 M) were used as specific inhibitors of caspase-1 and 3, respectively. expressing the fusion product containing caspase-resistant PARP (PARP D214A ) in which Asp 214 at the P1 position was substituted with Ala (27) remained ␤-galactosidase positive, although caspase-3 was introduced by mating. These results suggest the feasibility of the modified yeast two-hybrid system as a genetic system for identifying caspase substrates. Both caspase-1 and 3 were active in yeast YM4271 cells, and their activities were inhibited by the specific caspase inhibitors, supporting the idea that yeast can be used as a cell-based reporter system for caspase activity (Fig. 1C). Screening the mouse brain cDNA fusion library (ϳ10 8 colony-forming units/ml) resulted in the isolation of 16 clones. 14-3-3⑀ (amino acid residues 144 -255) was chosen as one of the caspase-3 substrates isolated by the modified yeast genetic system. 35 S-labeled 14-3-3⑀ protein was prepared by in vitro transcription/translation procedures and partially cleaved by the recombinant human caspase-3, while PARP was completely cleaved under the same condition. The cleavage was completely inhibited by a specific caspase-3 inhibitor, Ac-DEVD-CHO ( Fig. 2A). The cleavage of the 14-3-3⑀ protein in COS cells in response to apoptosis inducers was examined in Fig. 2B. Data revealed that apoptosis inducers such as UVC or staurosporine induced the cleavage of the endogenous 14-3-3⑀ protein in COS cells. The concomitant cleavage of PARP implicates the activation of caspase-3 in COS cells after UVC irradiation or staurosporine treatment. Indeed, the activities of caspase-3 were elevated in parallel with the cleavage of cellular 14-3-3⑀ in COS cells after UVC irradiation (data not shown). Amino acid sequence analysis of the cleavage product revealed that caspase-3 cleaves the 14-3-3⑀ protein at Asp 238 of MQGD. A mutant 14-3-3⑀ protein in which Asp 238 was substituted with Ala (D238A) was not cleavable by caspase-3 (Fig. 2C).

Identification of 14-3-3⑀ Protein as Caspase Substrate by the Modified Yeast Two-hybrid System-
Cleavage of 14-3-3⑀ Protein Impairs Its Binding Ability with Bad and Facilitates Bad-mediated Apoptosis-Bad and 14-3-3⑀ constructs were over-expressed in cultured 293T cells. The cleaved 14-3-3⑀ protein fragment, D238, interacted with Bad with lower affinity than its wild type or the uncleavable counterpart, D238A. Bad interacted with the endogenous Bcl-x(L) in cells co-expressing Bad either with the wild type 14-3-3⑀ protein or the uncleavable D238A. Interestingly, co-expression of the truncated 14-3-3⑀ protein D238 with Bad significantly increased Bad association with cellular Bcl-x(L) (Fig. 3), and the increased Bad association with Bcl-x(L) appeared to contribute to Bad-mediated apoptosis. Bad-mediated apoptosis was not affected by the ectopic expression of the wild type 14-3-3⑀ protein of the uncleavable D238A (Fig. 4A). The expression of the truncated D238 also enhanced staurosporine-induced apoptosis, whereas the expression of the uncleavable D238A did not increase apoptosis in 293T cells after staurosporine treatment (data not shown). Cleavage of the 14-3-3⑀ protein by caspase-3 did not seem to change the pattern of 14-3-3⑀ protein dimerization (Fig. 4B). GST-tagged wild type S-labeled 14-3-3⑀ protein and PARP were produced by in vitro transcription/translation procedures. 35 S-labeled 14-3-3⑀ protein or 35 S-labeled PARP was cleaved by the recombinant human caspase-3, and the cleavages were analyzed by 12% SDS-PAGE and subsequent autoradiography. Ac-DEVD-CHO was used as a specific caspase-3 inhibitor. B, COS cells (10 7 cells) were treated with staurosporine or UVC and incubated for 10 h or 24 h respectively. Cell lysate proteins (50 g) were separated on 12% SDS-PAGE. Cleavages of the cellular 14-3-3⑀ protein and PARP were examined by immunoblot analysis using antibodies against the 14-3-3⑀ protein and PARP, respectively. C, the 35 S-labeled 14-3-3⑀ protein and the uncleavable D238A were produced by in vitro transcription/translation procedures. The cleavage of the 35 S-labeled proteins by the recombinant human caspase-3 was analyzed by 12% SDS-PAGE and subsequent autoradiography. Ac-DEVD-CHO was used as a specific caspase-3 inhibitor.

FIG. 3. Co-expression of the truncated 14-3-3⑀ protein-enhanced Bad association with Bcl-x(L) in 293T cells.
HA-14-3-3⑀, HA-D238, or HA-D238A was cloned into the pcDNA3 cloning vector. GST-fused human Bad was co-expressed with each HA-tagged 14-3-3⑀ derivative in 293T cells. Cell lysates (50 g) were resolved by 12% SDS-PAGE, and the expression levels of GST-Bad, HA-14-3-3⑀, and cellular Bcl-x(L) were examined by immunoblot analysis using antibodies against GST (top panel), HA (second panel from the top), and Bcl-x(L) (fourth panel from the top). GST-Bad was pulled down using glutathion-Sepharose 4B beads, and the bound proteins were analyzed by immunoblotting using monoclonal antibodies directed against HA (third panel from the top) or Bcl-x(L) (bottom panel). IP, immunoprecipitation; WB, Western blot; WT, wild type.
The highly conserved 14-3-3 proteins have been known to be present as multiple isotypes in many eukaryotic organisms (31,32). The primary function of mammalian 14-3-3 protein isotypes is to inhibit apoptosis. 14-3-3 proteins have been thought to inhibit the pro-apoptotic function of their binding proteins, of which Bad is a typical example (17). Recently, Xing et al. (33) depicted 14-3-3 proteins as general anti-apoptotic factors. They reported that the dominant negative mutant forms of the 14-3-3 protein disrupted the anti-apoptotic function of the wild type 14-3-3 protein and increased apoptosis in response to UV irradiation. In the present study, we suggest that the 14-3-3⑀ protein is cleaved by the activated caspase-3 to promote cell death during exposure to apoptotic signals. The cleaved 14-3-3⑀ protein impaired its binding affinity to Bad and released Bad to translocate to the mitochondrial outer membrane where Bad heterodimerizes with Bcl-x(L). 14-3-3 proteins have been shown to inhibit Bad-induced apoptosis through interaction at the phosphorylation sites of Bad (11,16). A number of studies have described the dimeric array of 14-3-3 proteins that interacts with 14-3-3 binding proteins at two 14-3-3 consensus binding motifs. The dominant-positive lethal effect of the truncated 14-3-3⑀ protein (D238) on Bad-induced apoptosis in 293T cells (Fig. 4A) cannot be explained by the failure of the intermolecular association of D238 with Bad (Fig. 3), because the endogenous cellular 14-3-3⑀ protein in 293T cells expressing D238 should also have interacted with Bad. The results in Fig. 4B offered an explanation that the truncated D238 heterodimerized with the endogenous 14-3-3⑀ protein and lowered the intracellular level of self-associated wild type 14-3-3⑀ dimers that could sequester Bad in co-transfected cells. 14-3-3 proteins may form heterodimers as well as homodimers; therefore the truncated D238 may also dimerize with isoforms of 14-3-3 proteins in 293T cells. It is unlikely that cleavage of the 14-3-3⑀ protein by caspase-3 changes the pattern of dimerization during Badinduced apoptosis, because the residues of 14-3-3 isoforms involved in the dimer formation are largely conserved among mammalian isoforms (34). The fact that D238 forms heterodimers with isoforms of 14-3-3 proteins may not affect the interaction of D238 with Bad, because the different isoforms of the 14-3-3 protein were reported to interact equally well with Bad (14). The binding ability of HA-D238 with Bad was compared with that of HA-14-3-3⑀ or the uncleavable HA-D238A in an experiment in which HA-14-3-3⑀ or HA-D238A were coexpressed with or without wild type 14-3-3⑀ protein in 293T cells. Over-expression of the 14-3-3⑀ protein did not interfere with the binding affinity of HA-D238 with Bad, suggesting that cleavage of the 14-3-3⑀ protein by casapse-3 lowers its binding affinity with Bad regardless of the state of the intermolecular association of the cleaved 14-3-3⑀ protein (data not shown). The released death promoting Bad then heterodimerizes with Bclx(L) located in the outer membrane of the mitochondria. The binding of Bad to Bcl-x(L) might cause Bcl-x(L) to interact with pro-apoptotic Bax to release cytochrome c or regulate other Bcl-x(L) activities (3), leading to the progression of cell death.