Caenorhabditis elegans EGL-1 disrupts the interaction of CED-9 with CED-4 and promotes CED-3 activation.

In the nematode Caenorhabditis elegans, programmed cell death is implemented by the protease CED-3 whose activity is inhibited by CED-9 through physical associations with the regulator CED-4. The product of a recently described gene, egl-1, binds to and inhibits CED-9. In the present studies, we have addressed the molecular mechanism by which EGL-1 regulates CED-9 function and promotes cell death. Expression of CED-4 and CED-3 resulted in decreased survival and apoptosis of mammalian cells, activities that could be inhibited by CED-9. Importantly, this protective effect of CED-9 was antagonized by EGL-1. Immunoprecipitation analysis showed that EGL-1 binding to CED-9 disrupts the association between CED-4 and CED-9, an activity that required the BH3 motif of EGL-1. Consistent with these results, expression of EGL-1 promoted CED-4-dependent processing of CED-3, and this activity of EGL-1 was mediated through inhibition of CED-9. In mammalian cells, CED-9 is known to target the subcellular localization of CED-4 from the cytosol to intracellular membranes. Expression of EGL-1 resulted in redistribution of CED-4 from intracellular membranes, where it co-localized with CED-9, to the cytoplasm, providing further evidence that EGL-1 regulates CED-4 through CED-9. Finally, the levels of EGL-1 were greatly enhanced by co-expression of CED-9 in both mammalian cells and in a cell-free system, suggesting a role for CED-9 in the expression and/or stabilization of EGL-1. These studies provide a mechanism for how EGL-1 functions to antagonize pro-survival of CED-9 and to promote CED-3 activation and programmed cell death.

Programmed cell death (PCD) 1 is a conserved mechanism of cellular demise that is critical for embryonic organ development and homeostasis in adult tissues (1,2). In the nematode Caenorhabditis elegans, 131 of the 1090 somatic cells generated during development undergo PCD (3). Genetic analyses of the cell death process in the nematode have identified three genes that play critical roles in the induction and execution of PCD (4). The ced-9 gene protects cells that normally survive during worm development (5). ced-9 encodes a protein with significant homology to the mammalian Bcl-2 and Bcl-XL survival proteins (6). Two nematode genes, ced-3 and ced-4, are required for the execution of the cell death process (4). CED-3 is homologous to the mammalian interleukin-1␤-converting enzyme, which is a member of a family of cysteine proteases (designated caspases) (7). CED-3 and related caspases are thought to act as executioners of the nematode and mammalian PCD pathway (8). CED-4 also has a mammalian counterpart, Apaf-1 (9). Overexpression of ced-4 in nematode ALM neurons causes cell death that requires ced-3 activity for efficient killing, suggesting that ced-4 acts upstream of ced-3 (10).
Biochemical analyses of CED-3, CED-4, and CED-9 have provided important insight into the regulation of the central cell death machinery in the nematode. CED-9 interacts with CED-4 suggesting that CED-9 regulates cell death by binding to and inactivating CED-4 (11)(12)(13). Furthermore, CED-4 associates with CED-3 and promotes the proteolytic activation of CED-3, and this activation process is inhibited by CED-9 through a multimeric protein complex (13)(14)(15). Recent analyses of the mammalian counterparts have revealed physical associations of Bcl-XL, Apaf-1, and procaspase-9 (16,17), suggesting that the regulation of the central cell death machinery is conserved through evolution from nematodes to humans.
In mammals, a family of proteins that belong to the Bcl-2 family including Bax (18), Bak (19 -21), Bad (22), Bik/Nbk (23,24), Bid (25), Hrk/DP5 (26,27), Bim (28), Bok/Mtd (29,30), and BNIP3 (31) activate apoptosis. Structural and functional analyses have revealed that these pro-apoptotic proteins require the conserved BH3 region to interact with pro-survival Bcl-2/ Bcl-XL/Mcl-1 and to activate apoptosis in transient assays (18 -28, 31). Moreover, NMR studies have demonstrated that the BH3 domain of Bak interacts with a hydrophobic cleft formed by the conserved BH3 and BH1 regions of Bcl-XL (32). To date, all death-promoting Bcl-2-related proteins have been shown to heterodimerize with Bcl-2, Bcl-XL, or Mcl-1 suggesting that these molecules promote cell death, at least in part, by interacting with and antagonizing Bcl-2, Bcl-XL, and Mcl-1 (18 -29, 31). Recent analyses in C. elegans have identified EGL-1, a nematode counterpart of pro-apoptotic Bcl-2 family members. Gain-of-function mutations of egl-1 caused hermaphrodite-specific neurons to undergo PCD, whereas a loss-of-function egl-1 mutation prevented most, if not all, somatic cell deaths in the worm (33). Genetic analyses revealed that egl-1 acts upstream of or in parallel to ced-4 and ced-3 and requires ced-9 to exert its effect on PCD (33). Sequence analysis revealed that EGL-1 is a BH3-containing protein most homologous to mammalian Bad, Bik/Nbk, Bid, and Hrk/DP5 (33). EGL-1 physically associates with CED-9, as determined by in vitro interaction analysis, and requires its BH3 domain for efficient binding to CED-9 (33), suggesting that EGL-1 might act by binding to and inhibiting CED-9. However, the mechanism by which EGL-1 antagonizes the activity CED-9 remains unknown. In the present study, we have performed biochemical and functional analyses to assess the mechanism by which EGL-1 inhibits the activity of CED-9 and regulates CED-3 and CED-4.

EXPERIMENTAL PROCEDURES
Plasmid Constructions-The expression plasmids producing epitopetagged CED-4, CED-3, and CED-9 have been described (14). The egl-1 gene was cloned by polymerase chain reaction using specific primers (5Ј-GTTGCTAGCGCGGCCGCATCTAGAATGTCCAACGTTTTTGAC-GTTCAATCTTCCG-3Ј, 5Ј-GTTGAATTCGGGCCCGAGCTCTTAAAAA-GCGAAAAAGTCCAGAAGACGATGG-3Ј) and C. elegans cDNA from mixed stage embryos as template. The polymerase chain reaction product was digested with XbaI and ApaI and cloned in frame into pcDNA3 vectors engineered to encode N-terminal FLAG or AU1 epitope tags. The BH3 deletion mutant (⌬BH3) was constructed by replacing NheI/ ApaI fragment of the gene with a double strand oligonucleotide containing a deletion that correspond to amino acids 60 -67 of EGL-1 (5Ј-GGCTCCAAGCTAGCTCAGATGATGTCCTACTCGGCCCATGCT-TCCGACAGAAGCCTCTTCCATCGTCTTCTGGACTTTTTCGCTTTT-TAAGGGCCCGAATTC-3Ј). Authenticity of each construct was confirmed by dideoxy sequencing.
Transfection, Immunoprecipitation, and Immunoblotting-Human embryonic kidney cells (HEK293T) were cultured in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) supplemented with 10% fetal bovine serum (HyClone). Culture dishes were transfected by the calcium phosphate method or lipofection with the indicated amounts of plasmids (see figure legends). Transfected cells were harvested by centrifugation at 4°C and lysed with 0.2% Nonidet P-40 lysis buffer (10 mM HEPES, pH 7.2, 142.5 mM KCl, 5 mM MgCl 2 , 1 mM EGTA, 0.2% Nonidet P-40, 2 g/ml aprotinin, 2 g/ml leupeptin, 1 mM phenylmethylsulfonyl fluoride, 2.5 mM sodium pyrophosphate, and 10 mM sodium fluoride). About 10% of each lysate was saved to assay for expression of transfected plasmids, and the remainder of the lysate was immunoprecipi-tated with 5 g/ml anti-epitope tag antibodies (anti-FLAG M2 monoclonal antibody (mAb) from Sigma; anti-Flag D-8 polyclonal antibody from Santa Cruz; anti-HA 12CA5 mAb from Boehringer Mannheim; anti-HA Y-11 polyclonal antibody from Santa Cruz Biotechnology; anti-AU1 mAb from Babco, and anti-c-Myc A-14 polyclonal antibody from Santa Cruz Biotechnology). Proteins were immunoprecipitated with protein A/G-Sepharose 4B (Zymed Laboratories Inc.), subjected to SDS-polyacrylamide gel electrophoresis, and immunoblotted with indicated antibodies. Proteins were detected by ECL (Amersham Pharmacia Biotech).
In Vitro Transcription and Translation of Proteins-Coupled in vitro transcription and translation was carried out with the TNT system from Promega according to the manufacturer's instructions. Briefly, pcDNA3 plasmids producing FLAG-Egl-1 (wt and ⌬BH3 mutant) and HA-CED-9 were transcribed with T7 polymerase and translated in the presence of [ 35 S]methionine using a rabbit reticulocyte lysate.
Cell Immunostaining and Confocal Microscopy-HEK293 cells were cultured on poly-L-lysine-coated cover slides and transfected with the indicated plasmids (see figure legends) by the calcium phosphate method. After 24 h, cells were fixed with cold methanol for 20 min at Ϫ20°C. To prevent nonspecific hybridization, fixed cells were blocked with 30% goat serum, 1% bovine serum albumin in phosphate-buffered saline. Cells were stained with 5 g/ml indicated mouse mAb (see figure legends) for 30 min, washed, and incubated with fluorescein isothiocyanate-conjugated goat anti-mouse IgG-Fc antibody (Sigma). Subcellular localization was assessed by laser-confocal microscopy (Bio-Rad) as described (11).
Survival and Apoptosis Assays-HEK293T cells or MEF were cultured in 24-well plates (5-7.5 ϫ 10 4 cells/well) and transfected with the indicated plasmids (see figure legends) with LipofectAMINE (Life Technologies, Inc.) in triplicate. Each transfection mixture contained 20 ng of pcDNA3-␤-galactosidase plasmid as reporter. Cell survival of transfected cells was determined by quantification of reporter gene activity as described (34). Briefly, 48 h after transfection cells were lysed in 200 l of 1ϫ Reporter lysis buffer (Promega), and 25-50 l of each lysate was assayed for ␤-galactosidase activity in a reaction mixture containing 1 mg/ml o-nitrophenyl-␤-D-galactopyranoside, 670 mM sodium phosphate, pH 7.5, 1 mM magnesium chloride, 45 mM ␤-mercaptoethanol. Reactions were incubated at 37°C for 30 -120 min, and product forma-

FIG. 1. EGL-1 inhibits the interaction between CED-9 and CED-4.
A, HEK293T cells (2 ϫ 10 6 /60-mm plate) were transiently transfected by lipofection with plasmids producing FLAG-EGL-1 (0.66 g), Myc-CED-4 (0.66 g in A or 0.33 g in B), and HA-CED-9 (0.33 g in A or 0.66 g in B). In all cases 0.2 g of pcDNA3-p35 was included, and pcDNA3 empty plasmid was added so that the total amount of DNA was 2 g. 24 h after transfection, cell lysates were immunoprecipitated (IP) with anti-HA mAb and immunoblotted with anti-c-Myc, anti-HA, or anti-FLAG rabbit polyclonal antibody. Expression of proteins in total lysate is shown in the lower three panels. B, HEK293T cells were transfected as in A and lysates immunoprecipitated with anti-Myc mAb. Immunoprecipitates were immunoblotted as in A. Size markers are in kDa. The results shown are representative of at least three independent experiments. tion was monitored by reading optical density at a wavelength of 420 nm. Apoptotic morphology of transfected cells was determined 24 h after transfection by analysis of at least 100 cells expressing ␤-galactosidase as described (14).

RESULTS AND DISCUSSION
EGL-1 Binding to CED-9 Inhibits the Interaction of CED-4 with CED-9 -Genetic studies have shown that egl-1 acts upstream of ced-9 and that the egl-1 product physically interacts with CED-9 in vitro (33). Because CED-9 binds to the adaptor molecule CED-4 and this activity appears critical for the antiapoptotic function of CED-9 (11-15), we tested whether EGL-1 could affect the CED-9/CED-4 interaction. In these experiments, we transiently co-transfected HEK293T cells with expression plasmids producing HA-tagged CED-9, Flag-tagged EGL-1, and Myc-tagged CED-4. The interactions between these proteins were analyzed by immunoprecipitation and immunoblotting. As shown in Fig. 1A, immunoprecipitation of CED-9 co-immunoprecipitated CED-4 and expression of EGL-1 inhibited the association of CED-9 with CED-4 (Fig. 1A). Immunoblotting of CED-9 immunoprecipitates with anti-FLAG antibody revealed that EGL-1 was bound to CED-9 (Fig. 1A), suggesting that the interaction of EGL-1 with CED-9 prevents the binding of CED-9 to CED-4. In contrast to wt EGL-1, transfection of a plasmid producing a mutant form of EGL-1 deficient in CED-9 binding due to deletion of 5 amino acids in the BH3 motif (EGL-1 ⌬BH3) did not inhibit the association of CED-9 with CED-4 (Fig. 1A). However, the results with the EGL-1 ⌬BH3 mutant are difficult to interpret since the mutant protein was consistently expressed at lower levels than wt EGL-1 ( Fig. 1 and below). To verify further these results, we performed reciprocal experiments in which CED-4 was first immunoprecipitated with anti-Myc followed by immunoblotting with anti-HA to detect CED-9. CED-9 co-immunoprecipitated with CED-4, and as in the reciprocal experiment, EGL-1 inhibited the binding of CED-9 with CED-4 (Fig. 1B). Although EGL-1 binds to CED-9, EGL-1 might also inhibit the CED-4-CED-9 association by direct binding to CED-4. To test this possibility, we blotted CED-4 immunoprecipitates with anti-FLAG to determine whether EGL-1 was bound to CED-4. EGL-1 did not co-immunoprecipitate with CED-4 (Fig. 1B), confirming that CED-4 displacement from CED-9 complexes was due to CED-9 sequestration by EGL-1. These results indicate that EGL-1 binds to CED-9, an interaction that interferes with the ability of CED-9 to bind CED-4, releasing CED-4 from the death inhibitor.
CED-9 Regulates the Expression Levels of EGL-1-We have consistently found that EGL-1 expression in HEK293T cells is very low in the absence of CED-9. Furthermore, EGL-1 mu- tants deficient in CED-9 binding are expressed at lower levels than wt EGL-1 in the presence of CED-9. These observations suggested that EGL-1 expression might be regulated by its association CED-9. To test this hypothesis, we co-transfected HEK293T cells with constructs producing wt EGL-1 and CED-9 or control plasmids, and we analyzed EGL-1 expression by immunoblotting. In the absence of CED-9, no detectable expression of EGL-1 was observed ( Fig. 2A). In contrast, EGL-1 was expressed in HEK293T cells when co-transfected with a CED-9 plasmid (Fig. 2A). The effect of CED-9 on EGL-1 expression was specific in that co-expression of CED-3 or CED-4 did not promote EGL-1 expression (Fig. 2B). Moreover, these results could not be explained by poor viability of cells transfected with EGL-1, as expression of EGL-1 did not induce cell death in the absence or presence of CED-9 (data not shown and Fig. 4). To investigate further the role of CED-9 in the expression of EGL-1, we synthesized EGL-1 and CED-9 in a cell-free system. EGL-1 was synthesized and translated in reticulocyte lysate in the presence or absence of similarly translated CED-9. EGL-1 protein was detected when co-translated with CED-9 but not when EGL-1 mRNA was translated in the absence of CED-9 (Fig. 2C). Immunoprecipitation experiments confirmed that EGL-1 was expressed only in the presence of CED-9 (Fig. 2C). In contrast, the EGL-1 ⌬BH3 mutant was not expressed even in the presence of in vitro translated CED-9 (Fig. 2C), suggesting that the interaction with CED-9 is critical for the synthesis and/or stabilization of EGL-1. This CED-9-dependent regulation of its inhibitor EGL-1 is reminiscent to that of the apoptotic CAD/DFF40 nuclease and its inhibitor DFF45/ICAD (35). CED-9 may promote the translation of egl-1 mRNA and/or stabilize the nascent EGL-1 polypeptide, allowing its accumulation. In the worm, this mechanism may serve to ensure coordinated expression of EGL-1 and CED-9 during development.
The Subcellular Localization of CED-4 Is Altered in the Presence of EGL-1-In mammalian cells, CED-9 binds CED-4 resulting in the redistribution of CED-4 from the cytosol to intracellular membranes (11). We reasoned based on the results shown in Fig. 1 that EGL-1 might interfere with the ability of CED-9 to regulate the subcellular localization of CED-4. To test this hypothesis, we transiently transfected HEK293 cells with plasmids producing tagged EGL-1, CED-4, and CED-9, and we assessed the subcellular localization of the proteins by immunostaining and confocal microscopy. As previously reported (11), CED-9 displayed a granular, extra-nuclear staining pattern that is consistent with a localization to membranes of intracellular organelles such as mitochondria (Fig. 3B). In the majority of the cells, EGL-1 labeling was undetectable in the absence of CED-9 which is consistent with the results shown in Fig. 2. In cells in which EGL-1 expression was detected, the labeling was weak and cytosolic in the absence of CED-9 (Fig.  3A) which is consistent with the structure of EGL-1 that lacks a hydrophobic transmembrane tail (33). As previously reported (11), the labeling pattern of CED-4 was diffuse and cytoplasmic consistent with a cytosolic localization, but it was altered to a granular, cytoplasmic pattern in the presence of CED-9 (Fig. 3,  D and E). Significantly, expression of EGL-1 altered the distribution of CED-4 in cells that co-express CED-9 from an intracellular membrane pattern to a cytosolic pattern (Fig. 3F, compare E and F). Furthermore, the labeling pattern of EGL-1 changed from cytosolic to granular in the presence of CED-9 FIG. 5. EGL-1 promotes CED-3 processing by inhibiting CED-9. A, HEK293T cells (4 ϫ 10 6 /100-mm plate) were transiently transfected with 2 g of plasmid producing AU1-EGL-1, Myc-CED-4, FLAG-CED-3, and HA-CED-9. pcDNA3 empty plasmid was added so that the total amount of DNA was always 10 g. 12 h after transfection, cell lysates were immunoprecipitated with anti-FLAG polyclonal antibody. Immunocomplexes were resolved by SDS-polyacrylamide gel electrophoresis, transferred to a nitrocellulose membrane, and analyzed by immunoblotting with anti-FLAG antibody. Total amount of CED-3 protein was estimated by densitometry, and the percent of mature CED-3 was calculated by comparing the values from pro-form and mature CED-3. A longer exposure of the film to depict the mature p13 and p15 CED-3 products is shown in the bottom panel. Size markers are in kDa. The asterisk indicates immunoglobulin heavy chain. The results shown are representative of at least three independent experiments. B, molecular model to explain the regulation of CED-3 by EGL-1 through CED-4 and CED-9. The regulation of EGL-1 expression during worm development is unknown but in the model is induced or activated in cells undergoing PCD by death signals. Binding of EGL-1 to CED-9 displaces CED-4 from CED-9 that is localized at intracellular membranes of organelles such as mitochondria. The release of CED-4 from its inhibitor CED-9 enables CED-4 to activate CED-3 and downstream events required for programmed cell death. (Fig. 3C), which is consistent with the interaction between CED-9 with EGL-1 observed in the same cells (Fig. 1). These results support the biochemical analysis shown in Fig. 1 that showed that EGL-1 inhibits the association of CED-9 with CED-4. Furthermore, they indicate that EGL-1 inhibits the ability of CED-9 to target CED-4 to intracellular membranes. In surviving cells of the worm, CED-4 might be sequestered at intracellular membranes and inactivated by CED-9, particularly if CED-9 is expressed in excess relative to CED-4. In cells that undergo PCD, EGL-1 might be induced or activated leading to the release of pro-apoptotic CED-4 from its inhibitor CED-9.
EGL-1 Blocks the Protective Activity of CED-9 Against CED-4/CED-3-mediated Apoptosis-In HEK293T cells, CED-4 enhances apoptosis induced by CED-3, an activity that is inhibited by CED-9 (14,15). These results are in agreement with genetic and functional analyses of ced-3, ced-4, and ced-9 in C. elegans (4). Therefore, we used HEK293T cells to assess the ability of EGL-1 to regulate the protective activity of CED-9 against apoptosis promoted by CED-3 and CED-4. To test the function of EGL-1 in the mammalian model, HEK293T cells were co-transfected with a reporter plasmid expressing ␤-galactosidase with constructs producing EGL-1, CED-3, CED-4, CED-9, and control plasmid at different combinations. The cell killing activity was measured by a reduction in ␤-galactosidase activity (34) or by assessing the apoptotic morphology of ␤-galactosidase-positive cells (14). The results of six or three such independent experiments in which we measured cell survival or apoptosis are shown in Fig. 4, B and C. At the low concentrations of CED-3 and CED-4 plasmids used in these experiments, CED-3 or CED-4 had little effect on HEK293T survival. 2 In contrast, co-expression of CED-3 and CED-4 dramatically reduced HEK293T cell survival, an effect that was reversed by CED-9 (Fig. 4B). Expression of EGL-1 had little effect by itself on HEK293T cell survival (Fig. 4A) and did not significantly alter cell survival when co-expressed with CED-3, CED-4, CED-9, or CED-3 plus CED-4 ( Fig. 4B and data not shown). However, the protective effect of CED-9 against CED-3/CED-4-mediated killing was completely abrogated by EGL-1 (Fig. 4B). The ability of EGL-1 to inhibit the protective activity of CED-9 was also observed when HEK293T cells were scored for apoptosis in three independent experiments (Fig.  4C). Furthermore, similar results were also observed when the effect of EGL-1 was assessed in MEF cells (Fig. 4D).
The results presented herein provide a mechanism to explain how EGL-1 promotes cell death through its binding to CED-9. The interaction of EGL-1 with CED-9 results in the release of CED-4 from CED-9⅐CED-4 complexes and translocation of CED-4 from intracellular membranes to the cytoplasm. In this model, EGL-1 promotes the release of CED-4 from its inhibitor CED-9, allowing the activation of CED-3 by CED-4 (Fig. 5B). Consistently, EGL-1 abrogated the CED-9-protective effect against apoptosis, but it did not have any significant effect on ced-3/ced-4-induced killing when assessed in the absence of CED-9. These results are in agreement with the genetic studies in C. elegans that showed that egl-1 acts upstream of ced-9 and that its pro-apoptotic activity is dependent on ced-9 function (33). The model of action of EGL-1 suggested by our results resembles that of its mammalian homologues such as Bad, Hrk, Bik, or Bid, whose pro-apoptotic effect is mediated at least in part through direct binding and inhibition of Bcl-2 or Bcl-XL (22)(23)(24)(25)(26)(27). Moreover, it has been recently reported that binding of Bak, a pro-apoptotic member of the Bcl-2 family, to Bcl-XL prevents Bcl-XL binding to Apaf-1 (17), suggesting that the mechanism of action of these pro-apoptotic proteins has been conserved during evolution.