Interaction and Regulation of the Caenorhabditis elegans Death Protease CED-3 by CED-4 and CED-9*

In the nematode Caenorhabditis elegans, three genes, ced-3, ced-4, andced-9, play critical roles in the induction and execution of the death pathway. Genetic studies have suggested thatced-9 controls programmed cell death by regulatingced-4 and ced-3. However, the mechanism by which CED-9 controls the activities of CED-4 and the cysteine protease CED-3, the effector arm of the cell-death pathway, remains poorly understood. Immunoprecipitation analysis demonstrates that CED-9 forms a multimeric protein complex with CED-4 and CED-3 in vivo. Expression of wild-type CED-4 promotes the ability of CED-3 to induce apoptosis in mammalian cells, which is inhibited by CED-9. The pro-apoptotic activity of CED-4 requires the expression of a functional CED-3 protease. Significantly, loss-of-function CED-4 mutants are impaired in their ability to promote CED-3-mediated apoptosis. Expression of CED-4 enhances the proteolytic activation of CED-3. We also show that CED-9 inhibits the formation of p13 and p15, two cleavage products of CED-3 associated with its proteolytic activationin vivo. Moreover, CED-9 inhibits the enzymatic activity of CED-3 promoted by CED-4. Thus, these results provide evidence that CED-4 and CED-9 regulate the activity of CED-3 through physical interactions, which may provide a molecular basis for the control of programmed cell death in C. elegans.

Programmed cell death (PCD) 1 is critical during organ development and tissue homeostasis (1). In the nematode Caenorhabditis elegans, 131 of the 1090 somatic cells generated during development undergo PCD (2). Three genes have been identified in the nematode that play critical roles in the induction and execution of PCD (3). The ced-9 gene protects cells that normally survive from PCD during worm development (4,5). ced-9 encodes a protein with significant homology to the mammalian Bcl-2 and Bcl-X L survival proteins (4,5). Furthermore, bcl-2 can partially substitute for ced-9 in C. elegans, suggesting that bcl-2 is a homolog of ced-9 (6,7). In contrast, two nematode genes, ced-3 and ced-4, are required for the execution of the cell death program. Thus, loss-of-function mutations of ced-3 and ced-4 cause all 131 somatic cells that normally die to survive (3). The ced-3 product is homologous to the mammalian interleukin-1␤-converting enzyme (ICE), which is a member of a family of cysteine proteases (designated caspases) (7,8). CED-3 and related caspases are thought to act as effectors of the nematode and mammalian PCD pathway (7,8).
Genetic experiments have suggested that ced-9 protects cells from undergoing PCD by preventing the death-promoting activity of ced-3 and ced-4 (4,9). Consistent with genetic experiments in C. elegans, Bcl-2 and Bcl-X L , two mammalian homologs of CED-9, can inhibit the activation of ICE-like proteases and therefore appear to act upstream of the death proteases in the mammalian apoptotic pathway (10,11). Overexpression of ced-4 in the nematode ALM neurons causes cell death that requires ced-3 activity for efficient killing, suggesting that ced-4 acts upstream of ced-3 (9). Furthermore, protection against ced-3-induced cell death by ced-9 requires ced-4 activity, suggesting that ced-9 controls ced-3 by acting at least in part through ced-4 (9). Although genetic analysis has been essential for the identification and initial characterization of the nematode PCD pathway, the biochemical basis by which CED-3, CED-4, and CED-9 regulate cell death has remained elusive. Recent studies, however, indicate that CED-9 interacts with CED-4 suggesting that CED-9 regulates cell death by binding to and inactivating CED-4 (12)(13)(14). In the present studies, we performed further functional and biochemical analyses of CED-3, CED-4, and CED-9 interactions using mammalian cells as a model system to gain insight into the regulation and molecular basis of the PCD pathway.

EXPERIMENTAL PROCEDURES
Plasmid Constructions-The expression plasmids producing epitopetagged CED-4 and CED-9 have been described (13). Plasmids encoding Myc-tagged CED-4 mutants were generated by PCR amplification of ced-4S template DNA using 3Ј primers that included the natural translation termination sequences as described (15). Subsequently, the ced-4 constructs were ligated into the KpnI site of pcDNA3 (Invitrogen). An HA-or Flag-tagged ced-3 insert was constructed by introducing each epitope tag at the COOH terminus of CED-3 by PCR. Inserts to express mutant CED-3 proteins were generated by PCR amplification of ced-3 template using a 3Ј primer that included a translation termination codon (D220 CED-3) or by two-step mutagenesis to generate G360S CED-3. The HA-, Flag-, or Myc-tagged inserts were ligated into the BamHI site of pcDNA3. Orientation of the inserts was determined by restriction mapping. Authenticity of all tagged constructs was confirmed by dideoxy sequencing.
Transfection, Immunoprecipitation, and Western Blot Analysis-Culture dishes containing 2-5 ϫ 10 6 human embryonic kidney 293T cells were transfected with the indicated amount (see figure legends) of plasmid DNA by the calcium phosphate method. The expression of HA-CED-9, Myc-CED-4, and Flag-tagged CED-3 was determined in total lysates by immunoblotting as described previously (16). For immunoprecipitations, cells from each dish were lysed in 1 ml of Nonidet * This work was supported by Grant CA-64556 from the National Institutes of Health. 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.
Cleavage of Fluorogenic Substrate-293T cells (5 ϫ 10 6 ) in 100-mm plates were transiently co-transfected with 5 g of pcDNA3-ced-4-Myc and/or pcDNA3-ced-3-Flag in the presence or absence of 5 g of pcDNA3-HA-ced-9 by calcium phosphate method. The total amount of transfected plasmid DNA was always 15 g/well and was adjusted by adding vector pcDNA3 DNA. At 14 h, cells were lysed in buffer A (25 mM HEPES, 1 mM EGTA, 5 mM MgCl 2 , pH 7.6, containing protease inhibitors and 0.2% Nonidet P-40). 150 g of protein lysate was immunoprecipitated with 10 g of anti-Flag antibody or control in 200 l of buffer A and incubated at room temperature for 1 h. CED-3 complexes were immunoprecipitated with 20 l of protein A-Sepharose and washed with buffer A four times. 100 M of the fluorogenic substrate Ac-DEVD-AMC (Alexis, San Diego, CA) was added in buffer A and incubated at 37°C for 1, 3, and 6 h in triplicate. Cleavage of substrate emitted a fluorescent signal that was quantified in a fluorometer at excitation 340 nm and emission 460 nm. Statistical significance was determined by 1-way ANOVA followed by Student-Neuman-Keuls post-hoc comparisons.
Apoptosis Assay-293T cells (5 ϫ 10 5 ) in each well of 6-well plate were transiently co-transfected in triplicate with 0.33 g of a reporter plasmid pcDNA3-␤-gal plus 0.66 g of pcDNA3-ced-4-Myc and/or pcDNA3-ced-3-Flag or mutant tagged-ced-3 constructs in the presence or absence of 0.66 g of pcDNA3-HA-ced-9 by calcium phosphate method. The total amount of transfected plasmid DNA was always 2 g/well and was adjusted by adding vector pcDNA3 DNA. Apoptosis was determined at 24 h after transfection by analysis of at least 300 cells expressing ␤-galactosidase as described (8).
Interaction of CED-4 Mutants with CED-3 and CED-9 -To further characterize the specificity and functional significance of the interaction between CED-3 and CED-4, we determined the ability of CED-3 to associate with wild-type CED-4 and three natural CED-4 mutants that exhibit a loss of function phenotype in C. elegans ( Fig. 2A). Two point mutations, n1948 and n1894, introduce a single amino acid change (Ile to Asn) at

FIG. 1. CED-4 interacts with CED-3 and CED-9.
A, CED-3 co-immunoprecipitates with CED-4 and with CED-9 in the presence of CED-4. 293T cells (5 ϫ 10 6 / 100-mm plate) were transiently transfected with 5 g of the indicated HAtagged, Flag-tagged, or Myc-tagged expression plasmids or empty vector. In the case of transfection with one or two plasmids, cells were co-transfected with 5-10 g of empty vector so that the total amount of transfected plasmid DNA was always 15 g. Rabbit anti-Myc or anti-HA immunoprecipitates were immunoblotted with anti-Flag antibody to detect Flagtagged CED-3. In the bottom panel, immunoprecipitates were immunoblotted with anti-HA to detect CED-9 or anti-Myc to detect CED-4. B, CED-9 is immunoprecipitated with CED-3 through CED-4. Lysates were immunoprecipitated with rabbit anti-Flag, and immunoprecipitates were immunoblotted with anti-HA to detect the HA-CED-9 protein. In the bottom panel, immunoprecipitates were immunoblotted with anti-Flag to detect CED-3. The amount of plasmid DNA and number of transfected cells was identical to that described for panel A. Size markers are in kDa. IP, immunoprecipitation. position 258 and a stop codon at residue 401 of the CED-4 protein, respectively (15). We engineered another mutation, G328, to insert a stop codon at amino acid 328 to recapitulate the loss of function mutation n1416, which results from a Tc4 insertion at residue 328 of CED-4 (15). To assess the interaction of CED-3 with CED-4 proteins, 293T cells were transiently co-transfected with plasmids producing Flag-tagged CED-3 and Myc-tagged wild-type or mutant CED-4 proteins. Immunoprecipitation analysis revealed that CED-3 co-immunoprecipitated wild-type CED-4 (Fig. 2B), confirming our results presented in Fig. 1. Significantly, W401 and G328, two loss of function CED-4 mutants resulting from C-terminal truncations retained their ability to associate with CED-3 (Fig. 2B). Similarly, the missense, loss of function I258N mutant resulting from a single amino acid substitution at position 258 of CED-4 retained its ability to interact with CED-3 (Fig. 2C). Because CED-4 also interacts with CED-9 (12-14), we next determined the ability of the CED-4 mutants to associate with CED-9. Immunoprecipitation analysis revealed that all three loss of function CED-4 mutants retained their capacity to interact with CED-9 although the binding of the CED-4 mutants, par-ticularly G328 and I258N, was reduced when compared with wild-type CED-4 Fig. 2D).

CED-4 Associates with the Prodomain of CED-3 and with a Loss of Function CED-3 Mutant-
To further define the interaction of CED-4 with CED-3, the ability of CED-4 to interact with two CED-3 mutants was investigated. We engineered CED-3 D220 to express a truncated CED-3 mutant (residues 1-220) containing the prodomain of CED-3 and CED-3 G360S, a natural loss of function CED-3 mutant (17) with a single amino acid change (Gly to Ser) at position 360 located in the conserved pentapeptide QACRG of CED-3 ( Fig. 2A). CED-4 coimmuprecipitated with both CED-3 mutants (Fig. 2E) The results indicate that CED-4 can interact with the prodomain of CED-3 and with CED-3 containing a single amino acid change in its catalytic domain, which is in agreement with a recent report (12). CED-3 D220 is a truncated mutant (residues 1-220) of CED-3 containing the prodomain of CED-3 (solid box). CED-3 G360S represents a natural loss of function mutation (n2433) of CED-3 with a single amino acid change (Gly to Ser) at position 360 located in the conserved pentapeptide QACRG. p15 represents the major processed product of CED-3 associated wuth activation. The position of point mutations n1948 and n1894 of ced-4 with loss of function phenotypes are shown. Another CED-4 mutant, G328, that recapitulates the n1416 mutation is also shown. B, CED-3 interacts with wild-type CED-4, and W401 and G328 CED-4 mutants. 293T cells (2 ϫ 10 6 /60-mm plate) were transiently transfected with 3 g of the indicated Myc or Flag-tagged expression plasmids. In the case of transfection with one plasmid, cells were cotransfected with 3 g of empty vector so that the total amount of transfected plasmid DNA was always 6 g. Anti-Flag or control antibody immunoprecipitates were immunoblotted with anti-Myc antibody to detect Myc-tagged CED-4 proteins. C, interaction of CED-3 with mutant I258N CED-4. Lysates were immunoprecipitated with anti-Flag or control antibody and immunoblotted with anti-Myc to detect the Myc-tagged CED-4 proteins. D, interaction of CED-9 with loss of function CED-4 mutants. Lysates were immunoprecipitated with anti-Myc or control antibody and immunoblotted with anti-HA antibody to detect HA-tagged CED-9 proteins. A nonspecific protein band representing immunoglobulin heavy chain is indicated by an asterisk. In bottom panels, immunoprecipitates were immunoblotted with anti-Myc, or total lysates were immunoblotted with anti-HA. E, interaction of CED-4 with mutant CED-3 proteins. Lysates were immunoprecipitated with anti-Flag, anti-AU1, or control antibody and immunoblotted with anti-Myc antibody to detect Myc-tagged CED-4. Expression of Flag-CED-3-G360S and AU1-CED-3-D220 in total lysates is shown at the bottom. Size markers are in kDa. IP, immunoprecipitation.

CED-4 Promotes the Ability of CED-3 to Kill Mammalian
death-promoting ability of CED-3 in mammalian cells. To do this, 293T cells were transiently transfected with expression constructs producing CED-4, CED-3, and empty vector. Expression of CED-4 did not induce apoptosis in 293T cells above the levels observed with control plasmid (Fig. 3A). Transfection of CED-3 induced a modest but significant level of apoptosis in 293T cells (Fig. 3A). More importantly, the great majority of the cells underwent apoptosis when co-transfected with CED-3 and CED-4 (Fig. 3A), indicating that CED-4 potentiates the ability of CED-3 to induce apoptosis. Significantly, expression of CED-9 blocked the ability of CED-4 and CED-3 to induce apoptosis (Fig. 3A). Together with genetic experiments in C. elegans (3,9), these results suggest that CED-4 interacts with and activates the death-promoting ability of CED-3, which can be inhibited by the interaction of CED-9 with CED-4. To fur-ther assess the function of CED-4, we tested the ability of loss of function CED-4 mutants to promote CED-3-mediated apoptosis. Functional analysis revealed that the loss of function CED-4 mutants were defective in their ability to potentiate CED-3-mediated apoptosis when compared with wild-type CED-4 (Fig. 3A). The levels of wild-type and mutant W401 and G328 CED-4 proteins expressed in 293T cells were similar, ruling out the trivial explanation that the differential ability of these proteins to promote apoptosis is due to differences in protein expression (Fig. 2 and data not shown). However, the expression of the I258N CED-4 mutant is reduced and appears unstable, which agrees with previous results (14).
A Functional CED-3 Protein Is Required for CED-4 to Promote Apoptosis-To further determine the mechanism by which CED-4 potentiates CED-3-mediated apoptosis, we exam- ined whether the pro-apoptotic activity of CED-4 requires a "functional" CED-3 protease. In contrast to the killing observed with wild-type CED-3, CED-4 failed to promote apoptosis mediated by either a truncated CED-3 mutant containing the prodomain of CED-3 or G360S CED-3, a natural loss of function CED-3 mutant with a single amino acid change (Gly to Ser) at position 360 located in the conserved active pentapeptide QACRG of CED-3 (Fig. 3B). Because the CED-3 (G360S) mutant protein is known to be inactive as a cysteine protease in vitro (17), we conclude that the protease activity of CED-3 is required for CED-4 to promote apoptosis in mammalian cells. These results are consistent with genetic studies which have shown that the CED-3 protease activity is crucial for the deathpromoting function of CED-4 in C. elegans (9).

CED-4 Enhances the Proteolytic Activation of CED-3 and CED-9 Inhibits the Processing of CED-3 in Mammalian
Cells-To assess if the survival protein CED-9 regulates the activation of the CED-3 protease in vivo, cellular lysates from 293T cells transiently transfected with plasmids expressing CED-3, CED-4, and CED-9 were analyzed for the presence of p13 and p15, two cleavage products of CED-3 associated with its proteolytic activation (17,18). As shown in Fig. 4A, expression of CED-3 resulted in partial processing of the immature 56-kDa CED-3-FLAG protein as determined by the detection of the mature p13 and p15 products of CED-3 (17). Significantly, processing of CED-3 increased dramatically in the presence of CED-4 when compared with cells expressing CED-3 alone (Fig.  4A). Moreover, expression of CED-9 inhibited the processing of CED-3 in cells transiently transfected with CED-3 plus CED-4 as determined by the absence of the p13 and p15 proteolytic products of CED-3 (Fig. 4A). To verify these results, we determined the caspase enzymatic activity of CED-3 in lysates from cells transiently transfected with CED-3, CED-3 plus CED-4, and CED-3 plus CED-4 plus CED-9. To measure the protease activity associated with CED-3, aliquots of the same cellular lysates used in Fig. 4B were incubated with anti-Flag antibody to immunoprecipitate CED-3 protein complexes, and the immunoprecipitates were assayed for enzymatic activity using the Ac-DEVD-AMC fluorogenic substrate (19). In agreement with results shown in Fig. 4A, lysates from cells expressing CED-3 or CED-3 plus CED-4 exhibited significant enzymatic activity when compared with control lysates (Fig. 4B). Importantly, the caspase activity of CED-3 from cells expressing CED-3 and CED-4 was inhibited by CED-9 (Fig. 4B). In control experiments, there was no significant protease activity when the same cellular lysates were immunoprecipitated with control antibody (data not shown). Because the cellular expression of CED-3 was diminished in the presence of CED-4 and enhanced in the presence of CED-9 when compared with cells expressing CED-3 alone (Fig. 4), we adjusted the values of enzymatic activity obtained with the Ac-DEVD-AMC substrate to normalize for the amount of CED-3 protein (immature plus mature). After normalization, we found that the enzymatic activity of CED-3 was enhanced significantly by co-expression of CED-4 (Fig. 4C). In contrast, CED-9 inhibited the catalytic activity of CED-3 in the presence of CED-4 (Fig. 4C).
These results demonstrate that CED-4 interacts with both CED-3 and CED-9 in mammalian cells and presumably in C. elegans cells. Our studies confirmed and extended the reported interaction of CED-4 with CED-3 and CED-9 (14). However,

FIG. 4. CED-4 enhances the activation of CED-3, which is inhibited by CED-9 in vivo.
A, 293T cells (5 ϫ 10 6 /100-mm plate) were transiently transfected with 5 g of plasmids producing Flag-tagged CED-3, Myc-tagged CED-4, and HA-tagged CED-9. In the case of transfection with one or two plasmids, cells were cotransfected with 5-10 g of empty vector so that the total amount of transfected plasmid DNA was always 15 g. At 14 h, CED-3 complexes were immunoprecipitated and concentrated with rabbit anti-Flag antibody and immunoblotted with anti-Flag antibody to detect Flag-tagged CED-3. Total amount of CED-3 protein was calculated by densitometry and the percent of mature CED-3 was calculated by comparing the values from immature and processed CED-3. A longer exposure of the film to depict the p13 and p15 mature CED-3 proteins is shown at the bottom. The p32 band is an intermediate processed fragment of CED-3 consistent with a previous report (17). our functional results differ from those recently reported (14), which demonstrated that CED-4 was capable of inducing apoptosis in 293T cells even in the absence of CED-3. In addition to 293T cells, we found that expression of CED-4 failed to kill human MCF-7 breast carcinoma cells and COS-7 monkey cells. 2 In agreement with our results, another group has found that expression of CED-4 fails to kill Rat-1 and HeLa cells. 3 Thus, we do not have an explanation to account for the discrepancy in the results. In our studies, CED-4 functioned only to promote apoptosis in mammalian cells, which was dependent upon the expression of enzymatically active CED-3. These observations support previous genetic experiments in C. elegans in which efficient killing by CED-4 was shown to be dependent upon CED-3 activity (9). Together with the results from other investigators (14), these studies argue that CED-4 promotes cell death by interacting with and activating CED-3. However, the mechanism by which CED-4 activates CED-3 is unclear and needs to be further investigated. Because three loss of function CED-4 mutants retained their ability to associate with CED-3, our studies suggest that the interaction between CED-4 and CED-3 is not sufficient for death-promoting function of CED-4. Consistent with the analysis in C. elegans, the loss of function CED-4 mutants were impaired in their ability to potentiate CED-3-mediated apoptosis in mammalian cells. Expression of CED-9 inhibited the killing activity of CED-4 plus CED-3, and in another study, it inhibited killing induced by CED-4 alone (14). Because CED-9 can form a protein complex with CED-3 through CED-4, the protective function of CED-9 could be mediated by preventing the activation of CED-3 through CED-4. Indeed, we show for the first time that CED-4 enhances the processing and conversion of CED-3 into the active CED-3 form in vivo. Furthermore, we show that CED-9 inhibits the proteolytic activation of CED-3 in cells that co-express CED-4. Since a C. elegans cell culture is not available, the biochemical analysis of the death regulators CED-3, CED-4, and CED-9 was performed in mammalian cells. The biochemical results that we have obtained with CED-3, CED-4 and CED-9 are consistent with the genetic analysis in C. elegans (9). However, we cannot formally rule out that these worm proteins behave differently in C. elegans cells. These studies predict that Bcl-2 and Bcl-X L , the mammalian CED-9 homologs, may regulate apoptosis by interacting with caspases, the cysteine proteases of the ICE/CED-3 family through a mammalian CED-4 counterpart.