A Bax/Bak-independent mitochondrial death pathway triggered by Drosophila Grim GH3 domain in mammalian cells.

Grim encodes a protein required for programmed cell death in Drosophila, whose proapoptotic activity is conserved in mammalian cells. Two proapoptotic domains are relevant for Grim killing function; the N-terminal region, which induces apoptosis by disrupting inhibitor of apoptosis protein (IAP) blockage of caspase activity, and the internal GH3 domain, which triggers a mitochondrial pathway. We explored the role of these two domains in heterologous killing of mammalian cells by Grim. The GH3 domain is essential for Grim proapoptotic activity in mouse cells, whereas the N-terminal domain is dispensable. The GH3 domain is required and sufficient for Grim targeting to mitochondria and for cytochrome c release in a caspase- and N-terminal-independent, IAP-insensitive manner. These Grim GH3 activities do not require Bax or Bak function, revealing GH3 activity as the first proapoptotic stimulus able to trigger the mitochondrial death pathway in mammalian cells in the absence of multidomain proapoptotic Bcl-2 proteins.

Multicellular organisms eliminate unwanted or damaged cells by cell death, a process essential in shaping embryonic structures, maintaining adult tissue homeostasis, and eliminating damaged cells. Most cell death processes are executed through a defined pattern of cellular and biochemical events known as apoptosis (1)(2)(3). Genetic analysis of programmed cell death in the nematode Caenorhabditis elegans identified essential elements of the basic machinery for the regulation and execution phases of apoptotic death and revealed their conservation among metazoans (4). Thereafter, endogenous and viral molecules that promote, inhibit, or are required for apoptosis have been found conserved from nematodes to humans (5). In this scheme, mitochondria play a central role, integrating death signals and mediating apoptosis induction by releasing cytochrome c and other proteins that activate the death program (6).
Studies in the fly Drosophila melanogaster have uncovered four essential components of the genetic program controlling cell death, reaper, hid, sickle, and grim, whose transcriptional activation precedes, induces and is necessary for programmed cell death by apoptosis (7)(8)(9)(10)(11)(12). The four genes map to a single genetic complex and function as death switches that are regulated mainly at the transcriptional level, although Hid is also regulated through phosphorylation by the mitogen-activated protein kinase pathway (13,14). This genetic complex thus has a pivotal role in integrating different death stimuli that lead to insect cell death by apoptosis (15,16). Despite their central role in Drosophila cell death, no homologue for any of these genes has yet been described in any organism, although the four have been shown to activate death pathways in vertebrate cells (10,(17)(18)(19)(20)(21).
The products of these four genes show sequence similarity within the first 14 amino acids. The conserved N terminus of Reaper, Hid, Sickle, and Grim binds to members of the inhibitor of apoptosis protein (IAP) 1 family, preventing their antiapoptotic activity (8,10,(21)(22)(23)(24)(25). Flies deficient in DIAP1 suffer massive apoptosis early in development, suggesting that relief of the IAP protective effect could be sufficient for Reaper, Hid, Sickle, and Grim to trigger cell death (24,26). Whereas Hid requires the N-terminal domain to induce cell death efficiently in cultured cells (20,22), Reaper and Grim can induce apoptosis in the absence of this domain in several experimental contexts (18,19,21,27,28).
An additional Grim internal region, the GH3 domain, conserved in Reaper, Sickle, and Grim, has been identified as essential for Grim proapoptotic function in flies (19). In contrast to the N-terminal Grim domain, the GH3 domain promotes Grim mitochondrial targeting in fly cells and triggers a death pathway that involves changes in cytochrome c display. The GH3 domain is predicted to conform an amphipathic ␣-helix whose alteration leads to impaired Grim killing ability.
In vertebrates, members of the Bcl-2 homology group comprising both pro-and antiapoptotic molecules play a central role in regulating mitochondrial-mediated cell death (29). Of the four homology blocks found among the Bcl-2 family members, the BH3 domain appears essential for the function of the proapoptotic members. For some of these, BH3 is the only domain shared with the rest of the family (30), and its amphipathic ␣-helical structure is an essential feature, required for its biological activity. Recent studies in cells from mice deficient in both Bax and Bak (DKO) demonstrated that BH3only proteins are unable to kill cells in the absence of Bax and Bak multidomain proapoptotic proteins (31)(32)(33). In this sce-nario, BH3-only proteins appear to function as programmed cell death initiators that bind to antiapoptotic Bcl-2 proteins in response to proapoptotic stimuli. This binding would allow conformational change, membrane integration, and aggregation of Bax/Bak proteins, responsible for release of cytochrome c and other proteins from mitochondria to the cytosol. Moreover, DKO cells are resistant to all proapoptotic stimuli tested that act through the mitochondrial death pathway (32, 34 -37).
Here we explore the relevance of the GH3 domain in the conservation of Grim proapoptotic function in vertebrate cells. Our results show that GH3 activity is conserved in evolution, triggering a mitochondrial-cytochrome c death pathway in mammalian cells, whereas the N-terminal domain appears irrelevant for Grim killing activity in this context. Furthermore, Grim-induced mitochondrial death pathway is independent of Bax and Bak function, identifying Grim as the first molecule able to trigger this pathway in mammalian cells in the absence of both multidomain proapoptotic Bcl-2 family members.
The truncated mouse Bid form (tBid) was PCR-amplified from mouse embryo cDNA using 5Ј-CTCTCTCTGCGGCCGCCATGGGCAGCCAG-GCCAGCCG-3Ј and 5Ј-GCTCGGATCCTCAGTCCATCTCGTTTCTAA-CC-3Ј primers, which introduced NotI and BamHI sites; it was cloned in the pcDNA3.1 mammalian expression vector using these sites, and the sequence was confirmed.
Cell Culture and Death Assays-NIH 3T3 fibroblasts were cultured in Dulbecco's modified Eagle's medium (Bio-Whittaker) supplemented with 10% newborn calf serum and transfected using LipofectAMINE Plus (Invitrogen) as recommended by the manufacturer. For Grim mutants death assays, cells were cotransfected with pcDNA3.1 empty vector (control) or pcDNA3.1-grim (WT or mutants) expression vector (19), together with pcDNA3.1-lacZ expression vector at a 2:1 molar ratio. After 30 h, cells were fixed in 0.2% glutaraldehyde, washed in PBS, and X-gal-stained following standard protocols. For Bad mutants death assays, cells were cotransfected with pcDNA3.1 empty vector (control) or pcDNA3.1-bad (WT or mutants) expression vector together with pcDNA3.1-lacZ expression vector at a 2:1 molar ratio. After 48 h, cells were fixed in 0.2% glutaraldehyde, washed in PBS, and X-galstained as above.
Western Blotting and Immunoprecipitation-To test the stability of the Grim protein and its mutant forms, NIH 3T3 fibroblasts were transfected with pcDNA3.1 empty vector or pcDNA3.1-grim (WT or mutants) expression vector. Cells were treated with 50 M zVAD-fmk (Bachem) and, after 20 h, they were lysed in 0.2% Nonidet P-40 isotonic lysis buffer with freshly added protease inhibitors, then proteins were eluted and analyzed in SDS-PAGE. Western blotting analyses were performed using rabbit anti-grim IgG (18) and HPR-conjugated goat anti-rabbit IgG (Dako) and developed by enhanced chemiluminescence (ECL; Amersham Biosciences). For immunoprecipitations, NIH 3T3 fibroblasts were transfected with pcDNA3.1 empty vector or pcDNA3.1-grim (WT or mutants) expression vector together with pcDNA3.1 empty vector or pcDNA-6myc-hNAIP expression vector at a 1:1 molar ratio. Cells were treated with 50 M zVAD-fmk, and, after 24 h, they were lysed in 0.2% Nonidet P-40 isotonic lysis buffer with freshly added protease inhibitors. Equal amounts of cell lysates were incubated with an affinity-purified fraction of rabbit anti-Grim IgG or with a mouse anti-c-Myc monoclonal antibody (overnight, 4°C), mixed with 30 l of a 1:1 slurry of Gammabind G-Sepharose (Amersham Biosciences) and incubated (45 min, 4°C). The Sepharose beads were washed twice in 0.2% Nonidet P-40 lysis buffer, once in 0.25 M LiCl washing buffer, and twice in 50 mM Tris-HCl, pH 7.5, washing buffer before proteins were eluted and analyzed in SDS-PAGE. Western blotting analyses were performed using rabbit anti-grim IgG and protein G-HPR conjugate (Bio-Rad) or mouse antic-Myc monoclonal antibody and HPR-conjugated goat anti-mouse IgG (Dako), then developed by ECL.
Immunofluorescence-NIH 3T3 fibroblasts were cultured in microscope slide chambers (Cultek) and transfected with pcDNA3.1-grim (WT or mutants) or pcDNA3.1-bad (WT or mutants) expression vector. When indicated, cells were treated with 100 M zVAD-fmk. After 24 h, cells were washed in PBS, fixed in 4% paraformaldehyde/PBS (15 min, room temperature), and washed three times in PBS. Cells were permeabilized in 0.05% Triton X-100/PBS for 15 min, washed three times in PBS, preincubated in 10% goat serum/0.1% Tween 20/PBS for 1 h, and incubated successively with the primary and secondary antibodies for 1 h each, with three 5-min washes in 0.1% Tween 20/PBS between incubations.
Grim was detected with rabbit anti-grim IgG, mitochondria were detected with human anti-mitochondrial serum (18), cytochrome c with mouse anti-cytochrome c monoclonal antibody (monoclonal antibody) (clone 6H2.B4, BD Pharmingen), and Bad with rabbit anti-bad Ab (New England Biolabs). Secondary antibodies used were goat anti-rabbit IgG-Alexa 488, goat anti-mouse IgG-Cy3, goat anti-human IgG-Cy3, biotinylated goat anti-human IgG, biotinylated goat anti-mouse IgG and streptavidin-Cy5 (all from Jackson ImmunoResearch). Mito-Tracker staining was performed by incubating cells with MitoTracker Red CMXRos (Molecular Probes) before fixing, as recommended by the manufacturer. Optical sections were obtained using an Ar-Kr laser and TCS-NT Leica confocal imaging systems.
WT and DKO SV40-transformed MEF were cultured in microscope slide chambers and transfected with rtTA-grim expression vector (18). After lipofection, cells were treated with 2 g/ml of doxycycline and 100 M zVAD-fmk, and after 24 h, cells were processed for immunofluorescence as described.

RESULTS
The GH3 Domain Is Essential for Grim Proapoptotic Function in Mammalian Cells-We recently described the GH3 domain as essential for Grim proapoptotic activity in flies (19); therefore, we tested whether GH3 integrity was a requisite for its conserved activity in mouse 3T3 cells. As previously described (18), wild type (WT) Grim induced cell death when overexpressed in mouse 3T3 fibroblasts (Fig. 1B). In contrast, a Grim mutant form with a 13-amino acid deletion that removes the GH3 domain (⌬86 -98, Fig. 1A) only marginally induced apoptosis in this assay (Fig. 1B). A 5Ј-shifted 11-amino acid deletion, such that the 3Ј part of the domain is preserved (⌬83-93, Fig. 1A), was less effective in eliminating the proapoptotic activity than complete GH3 deletion (Fig. 1B). The elimination of four internal amino acids common to the two larger deletions (⌬89 -92, Fig. 1A) also resulted in effective impairment of Grim killing ability (Fig. 1B). The consequences of removing these regions are similar to those observed in assays in cultured Drosophila SL2 cells and transgenic flies (19). Furthermore, point mutations in the GH3 domain that alter Grim activity in flies similarly affect Grim activity in mouse cells. A nonconservative replacement of Leu-89 by glutamic acid (L89E, Fig. 1A) impaired GH3 killing ability nearly to the level of the ⌬89 -92 mutant (Fig. 1B), whereas a semi-conservative replacement of Leu-89 by alanine (L89A, Fig. 1A) had only a mild effect on Grim proapoptotic function (Fig. 1B). These results show that, as in the fly, the GH3 domain is required for Grim proapoptotic function in mouse cells and that the residues relevant for GH3-mediated killing in flies are equally important for its conserved activity in mammalian cells.
Because residual proapoptotic function was observed even in the most extreme case of Grim function impairment, we tested whether the conserved N-terminal domain was responsible for the residual killing activity. As reported (18), the N-terminal domain was unnecessary for Grim killing in mouse 3T3 fibroblasts (⌬2-14, Fig. 1B). Combinations of the N-terminal deletion with either the complete GH3 deletion (⌬2-14, ⌬86 -98) or the internal four-amino acid deletion (⌬2-14, ⌬89 -92) did not further reduce Grim function (Fig. 1B). We therefore could not demonstrate a proapoptotic function for the N-terminal domain in this context.
Deletion of the GH3 domain, alone or in combination with the N-terminal domain deletion, did not compromise Grim ability to bind mammalian, viral, or insect IAPs ( Fig. 2 and not shown). Furthermore, IAPs only show a residual rescue capacity of Grim killing activity in mouse cells (Ref. 18 and not shown), suggesting an IAP-independent mechanism of action for the GH3 domain. This conclusion is also validated by the fact that GH3-induced proapoptotic events are independent of caspase activity (see below), which would not be the case for an IAP-dependent pathway. We conclude that the proapoptotic activity of the GH3 domain is required for Grim function in mammalian cells and is not related to an IAP-inhibitory pathway, whereas N-terminal domain function does not appear to be essential in this context.
Grim Targets Mitochondria and Induces Cytochrome c Release in Mammalian Cells-We previously showed that Grim localizes diffusely in the cytoplasm of preapoptotic mouse 3T3 fibroblasts and progressively colonizes mitochondria as apoptosis progresses (18). To identify the mechanism involved in Grim cell killing in mammalian cells, we tested for cytochrome c release and mitochondrial membrane potential (MMP) state in 3T3 fibroblasts overexpressing Grim. We found that cytochrome c was undetectable in mitochondria of the large majority of cells expressing Grim (Fig. 3, A-D). This Grim-induced cytochrome c release is unaffected by the presence of the broadspectrum caspase inhibitor zVAD-fmk (Fig. 3, E-H MMP, as measured by MitoTracker incorporation (Fig. 3, I-L). Conservation of MMP extends into apoptosis progression, when changes in cell shape and organelle distribution have already taken place (Fig. 3, I-L). MMP loss was only observed in cells in which the apoptotic process was very advanced (not shown). To confirm that cytochrome c release preceded MMP loss, we stained simultaneously with the cytochrome c antibody and MitoTracker and observed that most Grim-expressing cells retained MMP but had released cytochrome c, whereas the reverse was never found (Fig. 3, M-P). These results show that in mammalian cells cytochrome c release is the first Grim-induced proapoptotic event and that it does not require the previous activation of cytosolic caspases; MMP loss occurs later as a consequence of mitochondrial dysfunction or advanced apoptosis.

The GH3 Domain Is Required for Grim Targeting to Mitochondria and Cytochrome c Release in Mammalian Cells-
Because the GH3 domain triggers a mitochondrial death pathway in flies, we tested the involvement of the GH3 domain in the mitochondrial changes induced by Grim in mammalian cells. Complete or partial elimination of the GH3 domain, or even the L89E substitution, abolished mitochondrial localization (Fig. 4, C and E and not shown) and cytochrome c release (Fig. 4, D and F, and not shown). In all cases, however, the subcellular distribution of the mutant proteins was not diffuse, but concentrated in prominent cytoplasmic dots (Fig. 4, C-F). In contrast, the ⌬2-14 mutant, which did not appreciably impair Grim function in cultured cells, retained mitochondrial localization and provoked cytochrome c release (Fig. 4, G and  H). Furthermore, NAIP or OpIAP coexpression does not impair the ability of Grim or any of its mutant proteins to induce apoptosis or to release cytochrome c in 3T3 cells (not shown). The L89A Grim mutant, which showed slightly reduced proapoptotic activity, displayed mitochondrial staining and in-duced cytochrome c release, although in a lower proportion of transfected cells than that observed for WT Grim, suggesting that greater concentrations of the mutant protein were required to achieve functional levels (not shown). When a combination of the ⌬2-14 and ⌬89 -92 mutations was tested, not only was the mitochondrial localization lost, but the cytoplasmic dots also disappeared, resulting in uniform cytosolic staining (Fig. 4I). Double ⌬2-14 and ⌬89 -92 mutant Grim did not induce cytochrome c release (Fig. 4J). These results indicate that an intact GH3 domain is required for Grim targeting to mitochondria and cytochrome c release, whereas the N-terminal conserved domain and IAPs inhibition are not involved in these proapoptotic functions in mammalian cells.
The GH3 Domain Is Sufficient for Apoptosis Induction, Mitochondrial Targeting, and Cytochrome c Release in Mammalian Cells-We tested whether the GH3 domain alone was able to confer proapoptotic functions to a heterologous protein. The GH3 domain is predicted to form an amphipathic ␣-helix, in similarity with the proapoptotic BH3 domain, present in many pro-and antiapoptotic Bcl-2 family members (38 -40). Given the structural similarities between the two domains, replacement of the BH3 with the GH3 domain would probably allow correct folding and exposure of the GH3 domain within the chimeric protein. Deletion of the BH3 domain of Bad, a BH3only Bcl-2 family member, abolishes its proapoptotic functions, making it a candidate for use as a carrier protein. We thus replaced the Bad BH3 domain with the 13 amino acids most reliably predicted to conform an ␣-helical structure within the Grim GH3 domain (BadGH3). We then assayed for killing ability and mitochondrial events induced, compared with WT Bad and BH3-deleted Bad (Bad⌬BH3).
Deletion of the BH3 domain abolished most of the Bad killing function (Fig. 5A), mitochondrial localization (Fig.   FIG. 3. Grim induces caspase- 5D), and cytochrome c release ability (Fig. 5E). In contrast, replacement of the Bad BH3 domain with a GH3 domain greatly increased Bad⌬BH3 proapoptotic function (Fig. 5A), and completely restored mitochondrial targeting (Fig. 5F) as well as cytochrome c release activity (Fig. 5G). These effects were specific to the GH3 domain activity, because an Leu to Glu substitution in the residue equivalent to Grim Leu-89 (BadGH3LE) impaired both killing and cytochrome c release (Fig. 5, A and I), although it did not affect the mitochondrial localization (Fig. 5H). These results show that the GH3 domain, when placed in a heterologous protein, is sufficient to induce cell killing, mitochondrial targeting, and cytochrome c release in mammalian cells.

GH3-mediated Grim Proapoptotic Activity Is Independent of Bax and Bak Function in Mammalian
Cells-Multidomain proapoptotic Bcl-2 family members Bax and Bak are essential mediators of the mitochondrial death pathway in mammals (31)(32)(33)(34). Following apoptotic stimulation, conformational changes at the N terminus of Bax can be detected specifically with antibodies directed against an epitope in this region (41). To test whether the Grim-induced mitochondrial death pathway was mediated by Bax, we analyzed Bax conformation in Grim-transfected mammalian cells. Whereas tBid expression always induced Bax conformational change prior to cytochrome c release (Fig. 6, E-H), no conformational change in Bax was observed in Grim-expressing cells, even when cytochrome c had been completely released (Fig. 6, I-L). These results show that Grim-induced cytochrome c release and cell death do not require Bax conformational change, and suggest that the Grim mechanism of action is Bax-independent.
To further study the implication of Bax and Bak in the Grim-induced death pathway, we tested Grim ability to induce cell death and mitochondrial events in DKO MEF (deficient in Bax and Bak). Bad induced extensive cell death in WT MEF, whereas it was only able to induce residual cell killing in DKO MEF (Fig. 7A), showing the Bax/Bak requirement in Bad proapoptotic function. In contrast, Grim induced apoptosis equally in both WT and DKO MEF (Fig. 7A), showing that Griminduced cell death is independent of Bax and Bak function. We also tested the implication of N-terminal and GH3 Grim domains in DKO MEF killing. Whereas the N-terminal domain was unnecessary for Grim-induced apoptosis in DKO MEF, the GH3 domain was essential for Grim proapoptotic activity in these cells (Fig. 7A). We conclude that Grim-induced apoptosis in mammalian cells is mediated by the GH3 domain and is independent of Bax and Bak function.
We then analyzed whether Grim mitochondrial localization and Grim-induced cytochrome c release were preserved in DKO MEF. No difference in Grim subcellular localization was observed between WT and DKO MEF (not shown). Furthermore, Grim was equally able to induce cytochrome c release in both WT and DKO MEF (Fig. 7, B-G). Grim-induced cytochrome c release was unaffected by the presence of the broad-spectrum caspase inhibitor zVAD-fmk (Fig. 7, B-G) or p35 coexpression (not shown). These results show that Bax and Bak function is not required for Grim-induced cytochrome c release and place Grim as the first proapoptotic stimulus known to trigger the mitochondrial death pathway in the absence of both Bax and Bak in mammalian cells.

DISCUSSION
Two distinct Grim proapoptotic domains have been identified, the N-terminal domain, involved in IAP inhibition, and the GH3 domain, involved in a mitochondrial pathway. The proapoptotic activity of Drosophila Grim is conserved in mammalian cells, although no homologues have yet been isolated. We show here that the GH3 domain is responsible for Grim killing activity in mammalian cells. Not only was this domain essential for Grim induction of death in mammalian cells, but the residues critical for GH3 activity in fly cells were found essential for Grim activity in mouse cells. The effects of Grim expression in mouse fibroblasts indicate a direct mitochondrial pathway as the target of Grim action. Grim localizes to mitochondria and induces cytochrome c release prior to any other proapoptotic symptom, in a caspase-and IAP-independent manner.
The consequences of Grim expression are nonetheless not identical in insect and mammalian cells. In fly cells, cyto- chrome c does not diffuse freely to the cytoplasm and is retained by mitochondria, showing altered display (19,42,43). In mouse cells, in contrast, Grim induces rapid cytochrome c release to the cytosol. This difference resembles that reported for Debcl overexpression, which in Drosophila cells does not induce cytochrome c release (43), whereas it does in mammalian cells (44). The differences observed between insect and mammalian cells in cytochrome c dynamics after BH3/GH3 overexpression may therefore be specific to the responding system, rather than reflecting distinct activities of the inducing factors. In addition, these differences do not appear to represent a fundamental change in the proapoptotic pathway triggered, which is dependent in both cases on the activity of Apaf-1, the molecule responsible for transducing cytochrome c-primed apoptosis (45), and which activates similar downstream caspases (43).
The Grim mode of action resembles that of the proapoptotic members of the Bcl-2 family, in which BH3 domain integrity is required for proapoptotic function, localization to mitochondria, and cytochrome c release (46,47). Similar to the BH3, the GH3 domain is predicted to form an amphipathic ␣-helix; the mitochondrial targeting and alterations triggered by these two domains may thus rely on a similar mechanism. Multidomain proapoptotic Bcl-2 family members are essential mediators of the mitochondrial death pathway in mammalian cells. Nonetheless, the Grim GH3 domain is able to induce mitochondrial targeting, cytochrome c release, and cell killing in cells deficient for both Bax and Bak. This shows that, unlike the BH3 proteins, Grim does not induce mitochondrial permeabilization through activating Bax/Bak, but uses some other parallel mechanism triggered by its GH3 domain. In agreement with this, one of the main properties of a BH3 domain, its ability to bind antiapoptotic Bcl-2 family members, is not exhibited by the GH3 domain, either in wild type Grim or in the chimeric protein. 2 In addition, inhibition of Grim-induced apoptosis and cytochrome c release by antiapoptotic Bcl-2 family members is quite limited (18), 2 and probably due to a general mitochondrial protective function instead of a direct inhibitory effect. Different models have been proposed to explain the Bax/Bakinduced release of proapoptotic proteins from mitochondria to the cytosol. One is the interaction of multidomain proapoptotic Bcl-2 family members with resident mitochondrial proteins, such as the voltage-dependent anion channel and the adenine nucleotide exchanger, proposed components of the permeability transition pore (PTP). PTP opening leads to mitochondrial depolarization and swelling, followed by mitochondrial outer membrane rupture and release of the intermembrane content. Transient openings of the PTP could nonetheless allow protein release without mitochondrial depolarization and swelling. Another proposed mechanism for Bax/Bak action is the formation of pores in the outer mitochondrial membrane by oligomerization. Bax has been shown to form channels in lipid membranes, suggesting pore-forming activity (48,49). In this scenario, Grim could induce cytochrome c release by any of these mechanisms or by alternative means. Mitochondrial membrane potential loss is a late event in Grim-induced apoptosis, subsequent to cytochrome c release. Any putative involvement of the PTP in Grim-induced mitochondrial events would therefore consist at most of transient PTP openings. On the other hand, no oligomerization or pore-forming ability has been explored 2 C. Clavería and M. Torres, unpublished data.
forms. Survival ratios were calculated as in Fig. 1. Images show colocalization of transfected WT or mutant Bad forms, as indicated (green), with either a mitochondrial antigen (Mitoch in B, D, F, and H) or cytochrome c (Cyt c in C, E, G, and I), shown in red.  for Grim, and thus, further studies will be required to determine this issue. Rather than representing homologous proapoptotic pathways, the BH3 and GH3 domains may represent parallel pathways for inducing apoptosis by mitochondrial permeabilization. The fact that genuine components of the Bcl-2 proapoptotic family are present in Drosophila (44, 50 -52) indicates that both proapoptotic pathways can coexist and might thus also coexist in other organisms.
While displaying limited structural similarity to Reaper, Hid, Sickle, and Grim, the vertebrate Smac/Diablo protein can bind to and block the protective effect of IAPs (53,54); it may thus represent a functional homologue of the fly IAP inhibitory pathway. We nonetheless found no evidence of a relevant role for Grim IAP inhibitory function in mammalian cells. The role of Smac/Diablo in regulating endogenous apoptosis has not yet been demonstrated, because mice deficient in this factor do not show altered cell death (55). Smac/Diablo is therefore unlikely to represent a homologue of the mitochondrial-cytochrome c pathway triggered by Grim GH3. Although Reaper, Hid, Sickle, and Grim induce specific apoptotic pathways in vertebrate cells and participate in highly conserved routes in the fly (13,14,56), no direct homologue of these proteins has yet been identified in any other organism. It thus remains to be determined whether there are vertebrate proteins with proapoptotic activities homologous to those of the GH3 domain.