Bcl-2 family of proteins are key regulators of apoptosis. Both proapoptotic and antiapoptotic members of this family are found in mammalian cells, but no such proteins have been described in insects. Here, we report the identification and characterization of Debcl, the first Bcl-2 homologue in Drosophila melanogaster. Structurally, Debcl is similar to Bax-like proapoptotic Bcl-2 family members. Ectopic expression of Debcl in cultured cells and in transgenic flies causes apoptosis, which is inhibited by coexpression of the baculovirus caspase inhibitor P35, indicating that Debcl is a proapoptotic protein that functions in a caspase-dependent manner. debcl expression correlates with developmental cell death in specific Drosophila tissues. We also show that debcl genetically interacts with diap1 and dark, and that debcl-mediated apoptosis is not affected by gene dosage of rpr, hid, and grim. Biochemically, Debcl can interact with several mammalian and viral prosurvival Bcl-2 family members, but not with the proapoptotic members, suggesting that it may regulate apoptosis by antagonizing prosurvival Bcl-2 proteins. RNA interference studies indicate that Debcl is required for developmental apoptosis in Drosophila embryos. These results suggest that the main components of the mammalian apoptosis machinery are conserved in insects.
Programed cell death by apoptosis is essential to remove unwanted and superfluous cells during animal development and metamorphosis to maintain tissue homeostasis (reviewed in Jacobson et al. 1997; Vaux and Korsmeyer 1999). Genetic studies in Caenorhabditis elegans have identified at least four genes, egl-1, ced-3, ced-4, and ced-9, that are essential for the regulation of all developmentally programed death of somatic cells (reviewed in Metzstein et al. 1998). EGL-1, CED-3, and CED-4 are required for cell death to occur, whereas CED-9 is essential for cell survival (Yuan and Horvitz 1992; Yuan et al. 1993; Hengartner and Horvitz 1994; Conradt and Horvitz 1998). In this developmental cell death pathway, EGL-1 functions upstream of CED-9, while CED-9 interacts with and regulates CED-4, which is required for CED-3 activation (Metzstein et al. 1998). The main apoptotic machinery has been conserved during evolution, and homologues of these C. elegans proteins are found in mammals. As expected, the pathways of cell death are considerably more complex in mammals, where EGL-1, CED-3, and CED-9 are represented by multiple family members. Although there is only one mammalian counterpart of CED-4, named Apaf-1, currently known (Zou et al. 1997), adaptor molecules that act to recruit caspases to death complexes and mediate their activation can be seen as functional homologues of CED-4 (reviewed in Kumar 1999; Kumar and Colussi 1999).
CED-3, a cysteine protease of the caspase family, is the main downstream effector of apoptosis in C. elegans (Yuan et al. 1993). There are at least 14 mammalian homologues of CED-3, some of which play key roles in apoptosis (reviewed in Cryns and Yuan 1998; Nicholson 1999). The adaptor proteins, CED-4 in C. elegans, and Apaf-1 in mammals, are essential for the activation of CED-3 and caspase-9, respectively (Li et al. 1997; Yang et al. 1998). CED-9 and its mammalian homologues, including Bcl-2, Bcl-xL, and Bcl-w, act as inhibitors of caspase activation and function upstream of CED-4/Apaf-1 (reviewed in Adams and Cory 1998; Gross et al. 1999). EGL-1 and its mammalian homologues share a small region of homology (BH3 domain) with CED-9/Bcl-2 proteins and act as proapoptotic proteins upstream of CED-9/Bcl-2 (Conradt and Horvitz 1998; Gross et al. 1999). In addition to these proteins that are distantly related to Bcl-2, mammalian cells also express a number of proapoptotic members of the Bcl-2 family, such as Bax, Bak, and Bok. These proteins, termed the Bax subclass of proteins, contain three Bcl-2 homology (BH) domains, BH1, BH2, and BH3, but lack the NH2-terminal BH4 domain present in some (e.g., CED-9, Bcl-2, Bcl-xL, and Bcl-w) prosurvival members of the Bcl-2 family (Adams and Cory 1998; Gross et al. 1999). Interestingly, no homologues of the Bax subclass of proteins have been found in C. elegans.
In Drosophila melanogaster, six caspases have been discovered so far (Fraser and Evan 1997; Inohara et al. 1997; Song et al. 1997; Chen et al. 1998; Dorstyn et al. 1999a,Dorstyn et al. 1999b). In addition, a CED-4/Apaf-1 homologue, termed Dark/Dapaf-1/HAC-1, has recently been described (Kanuka et al. 1999; Rodriguez et al. 1999; Zou et al. 1999). Although the caspase(s) regulated by Dark is currently unknown, Dark can interact with two known Drosophila caspases, Dredd and Dronc (Kanuka et al. 1999; Rodriguez et al. 1999). So far, no CED-9/Bcl-2–like protein has been reported in the fly. Given the conservation of the cell death machinery, it is anticipated that Drosophila also has Bcl-2–like proteins. In this paper, we described the identification of two Bcl-2 homologues in Drosophila, one of which, named Debcl, was characterized in detail. Debcl is a proapoptotic member of the Bcl-2 family that contains BH1, BH2, and BH3 domains. We show that Debcl is structurally related to the mammalian proapoptotic Bcl-2 family of proteins and functions in the execution of physiological cell death in Drosophila.
Materials and Methods
Cloning of Debcl and 48A-E cDNAs
Debcl and the 48A-E Bcl-2–like proteins were identified as genomic regions encoding putative Bcl-2 family members by TBLASTN searches using Bcl-2 protein sequence (accession numbers of the genomic sequence entries are indicated below). Full-length debcl cDNA sequence of 1,535 bp was obtained from BDGP clones GH01265 and LD12719, purchased from Research Genetics. A 950-bp partial cDNA clone for the 48A-E homologue was isolated from a mixed stage Drosophila embryo cDNA library in λgt11 using a 450-bp probe derived from Drosophila genomic DNA by PCR. Sequencing of this clone confirmed that it also encoded a Bcl-2 family member (Fig. 1 C). However, since the predicted reading frame in the sequence is open at the 5′ end, it is likely that the cDNA clone is not full length.
Plasmid Vectors for Expression in Cultured Cells
The 900-bp coding region of debcl was PCR amplified by Pfu polymerase (Stratagene) with an in-frame NH2-terminal HA tag and cloned into mammalian expression vector pcDNA3 (Invitrogen) and inducible Drosophila expression vector, pRmHa.3 (Bunch et al. 1988), to generate pcDNA3-debcl and pMT-debcl, respectively. In pMT-debcl, debcl expression is under control of a metallothionein (MT) promoter (Bunch et al. 1988). A green fluorescent protein (GFP) reporter construct was generated by placing GFP coding region downstream of an actin promoter in insect vector, pPAC-5C. To generate a P35 insect expression construct, the coding region of the baculovirus p35 was PCR-amplified and cloned into pPAC-5C. Debcl BH3 domain mutants L146G and E151G were generated by Quickchange™ method (Stratagene) using pMT-debcl as a template. Vectors for the expression of FLAG-tagged Bcl-2 family members have been described previously (Huang et al. 1997; Moriishi et al. 1999).
mRNA Expression Analysis
Total RNA from various developmental stages of Drosophila or adult flies was prepared using RNAzol B according to the manufacturer's (Tel-Test Inc.) protocol. Poly A+-enriched RNA was prepared using oligo dT magnetic beads (Dynal). Northern blots were prepared and hybridized with a 900-bp debcl open reading frame (ORF) probe as described (Dorstyn et al. 1999a,Dorstyn et al. 1999b). For reverse transcriptase (RT)-PCR, 1 μg total RNA was reverse transcribed using a first strand cDNA synthesis kit (Amersham Pharmacia Biotech). Aliquots of cDNA were subjected to 30 cycle PCR using primers from debcl ORF that generate a PCR product of ∼450 bp. For in situ RNA analysis, antisense and sense digoxygenin-labeled riboprobes were prepared using T7 and SP6 RNA polymerases from linearized pcDNA3-debcl as a template. Digoxygenin labeling was performed according to the manufacturer's instructions (Roche Biochemicals). In situ hybridization to Drosophila embryos and larval tissues was essentially as described (Dorstyn et al. 1999a,Dorstyn et al. 1999b), except that hybridization signals were further amplified using Tyramide Signal Amplification (TSA™) Indirect system according to the protocol supplied by the manufacturer (New England Nuclear Life Science Products).
Cell Death Assays
Schneider L2 (SL2) cells were maintained and transfected using Cellfectin (Life Technology) as described (Chen et al. 1996). For death assays, 1.5 × 106 SL2 cells were cotransfected with 1.6 μg vector or pMT-debcl (wild-type or mutants) and 0.4 μg pPAC-GFP reporter. 24 h later, cells were split into two halves, one of which was treated with 0.7 mM CuSO4 for 8 h (for immunoblots) or 16 h (for death assays). Where indicated, 50 μM zVAD-fmk (Enzyme Systems Inc.) was added to cultures at the time of addition of CuSO4. After fixation with 4% paraformaldehyde, GFP positive cells were counted by fluorescence microscopy. Cell survival was calculated as the percentage of GFP positive cells in CuSO4-treated cells, relative to the percent of GFP positive cells in untreated cells. The results, shown as average percentages ± SEM, were derived from three independent experiments. To check copper-induced Debcl expression, after an 8-h CuSO4 treatment, cells were lysed in SDS-PAGE buffer and lysates subjected to immunoblotting using an αHA antibody (Roche Biochemicals). NIH 3T3 cells were transfected using Fugene6 (Roche Biochemicals) with 1.0 μg of pcDNA3-debcl, and 0.2 μg of a β-galactosidase expression plasmid (pEF-βgal; Kumar et al. 1994). Where indicated, cells were cotransfected with P35, Bcl-2, Bcl-xL, MIHA, and Bcl-xLexpression constructs (described in Uren et al. 1996; Dorstyn and Kumar 1997; Huang et al. 1997). In these experiments, we used 3 μg of the specific inhibitor expression construct mixed with 1 μg of pcDNA3-debcl and 0.2 μg of pEF-βgal. Cells were fixed and stained with X-gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside) 48 h after transfection, and β-galactosidase positive cells were scored for apoptotic morphology (Kumar et al. 1994; Dorstyn and Kumar 1997). Data, presented as percent apoptotic cells as a fraction of total β-galactosidase positive cells ± SEM, were derived from three independent experiments.
Analysis of Apoptotic Cells in Embryos
Transgenic Flies and Genetic Interaction Studies
The PCR-amplified 900-bp debcl coding region tagged with HA (see above) was cloned into the pUAST plasmid (Brand and Perrimon 1993). Transgenic flies were generated and maintained essentially as described (Richardson et al. 1995). A homozygous line on the 2nd chromosome was used for the analysis of cell death in larval tissues. For genetic interaction studies, a line on the 3rd chromosome, UAS-debcl#26, was used to generate the strain GMR-GAL4/CyO; UAS-debcl#26/TM6B, which gave rise to adults with severely ablated eyes. This strain was crossed to strains containing GMR-p35; a deficiency of rpr, hid, and grim (Df(3L)H99); a deficiency of diap1 (Df(3L)brm11 or Df(3L)stf-13); a deficiency of diap2 (Df(2R)Jp1); or a P allele mutation of dark, darkCD8 (Rodriguez et al. 1999), and the eye phenotypes of the progeny were compared with GMR-GAL4; UAS-debcl#26 adult eyes using light microscopy or scanning EM as described previously (Richardson et al. 1995). darkCD8 flies were kindly provided by J. Abrams (University of Texas, Southwestern Medical Center, Dallas, TX). All other fly stocks were obtained from the Bloomington stock center.
Debcl/Bcl-2 Interaction Studies
293T cells were cotransfected with HA-tagged Debcl in pcDNA3 and FLAG-tagged control vector or various Bcl-2 family proteins in the presence of a baculoviral P35 expression construct by lipofection as described (Huang et al. 1997). 36 h after transfection, cells were labeled with 100–200 μCi/ml 35S-methionine (New England Nuclear) and cell lysates prepared in lysis buffer (20 mM Tris-HCl, pH 7.4, 135 mM NaCl, 1% Triton X-100, 10% glycerol, supplemented with 0.5 μg/ml Pefabloc, 100 μg/ml soybean trypsin inhibitor, and 1 μg/ml each of leupeptin, aprotinin, and pepstatin). In some experiments, unlabeled cell extracts from transfected cells were used for immunoprecipitation/immunoblotting analyses. Equivalent trichloro-acetic acid precipitable counts (5 × 108 cpm) or 1 mg of cell lysates were used for each immunoprecipitation with 5 μg of a control isotype-matched αEE (Berkeley Antibody Co.), αFLAG M2 (Sigma Chemical Co.) or αHA 11 (Berkeley Antibody Co. or Roche Biochemicals) antibodies. Immunoprecipitations were performed according to a previous protocol (Moriishi et al. 1999) and the immunoprecipitated material fractionated by SDS-PAGE. For 35S-methionine–labeled proteins, the signals were detected by fluorography after signal enhancement using AMPLIFY (Amersham Pharmacia Biotech). For unlabeled proteins, signals were detected by immunoblotting using the ECL detection system (Amersham Pharmacia Biotech).
RNA Interference (RNAi) Methods
RNAi methods were essentially as described (Bhat et al. 1999; Misquitta and Paterson 1999). debcl sense and antisense RNA transcripts were synthesized using the Ambion Megascript kit using linearized pcDNA3-debcl as a template. After purification and annealing, double-stranded RNA was dissolved in injection buffer (5 mM KCl in 0.1 mM phosphate buffer, pH 7.8) at 0.75 mg/ml. Precellularized embryos were injected at 50% egg length as described (Bhat et al. 1999; Misquitta and Paterson 1999). Embryos were aged at 18°C to stage 16 before TUNEL staining.
debcl sequence has been deposited in GenBank/EMBL/DDBJ under accession number AF178430. The EST clones (BDGP clones GH01265 and LD12719) from which debcl sequence was derived have the accession numbers AI513093 and AI062455, respectively. debcl genomic sequence is contained in AC007624. Coding region for 48A-E Bcl-2–like protein is contained in AC007473.
Two Bcl-2 Homologues in Drosophila
To identify Bcl-2–like proteins, we searched the Drosophila sequence database using the TBLASTN program and identified two putative Bcl-2 homologues in the genomic regions 42E-43A and 48A-E. The 42E-43A region was represented in two EST clones, which were sequenced in full. The 1,535-bp full-length cDNA for the putative Bcl-2 homologue on 42E-43A contains a 410-bp 5′ untranslated region, a 900-bp coding sequence, and a 225-bp 3′ untranslated region. A comparison of the cDNA and genomic sequences identified three exons in this gene (Fig. 1 A). The cDNA encoded a Bcl-2–like protein consisting of 300 amino acids with an estimated molecular mass of ∼33 kD. This was confirmed by in vitro translation of transcribed RNA (data not shown). We have named this protein Debcl (pronounced debacle) for death executioner Bcl-2 homologue. Debcl contains three of the BH domains, BH1, BH2, and BH3 (Fig. 1B and Fig. C), but lacks the NH2-terminal BH4 domain found in some antiapoptotic members of the Bcl-2 family. The COOH terminus of Debcl contains a putative hydrophobic membrane anchor, similar to that found in many Bcl-2–like proteins (Fig. 1C and Fig. D).
Debcl is most similar to the other putative Drosophila Bcl-2–like protein on region 48A-E, sharing 42% identity and 62% similarity in a 169 amino acid stretch (Fig. 1 C). Among the published mammalian Bcl-2 family members, the homology is mostly limited to the regions that comprise the three BH domains. In this region, Debcl shares the highest degree of homology (35% identity, 52% similarity) with Bok, a proapoptotic Bax subfamily member (Adams and Cory 1998; Gross et al. 1999). Debcl shares 20–30% identity with various prosurvival members of the Bcl-2 family, including A1 (30% identity, 49% similarity), Bcl-2 (25% identity, 41% similarity), Bcl-xL (25% identity, 46% similarity), Mcl-1 (26% identity, 46% similarity), and Bcl-w (21% identity, 41% similarity). The overall structure of Debcl is similar to Bax, Bak, and Bok, all of which contain BH1, BH2, and BH3 domains, a membrane anchor region, and a relatively long NH2-terminal region (Fig. 1, B–D).
debcl Expression Correlates with Cell Death during Drosophila Development
debcl expression was determined by RNA blotting, RT-PCR, and in situ hybridization to Drosophila embryos and tissues. In most cases low levels of debcl mRNA expression was detected. In RNA blots, a 1.5-kb transcript was evident in most developmental stages, but expression was relatively high in 0–4-h embryos and adult female flies (Fig. 2 A). In late embryos, larvae and pupae debcl expression was somewhat reduced and barely detectable by Northern blotting. However, RT-PCR analysis indicated that debcl mRNA was present in all developmental stages examined (Fig. 2 B).
Because of low expression of debcl, we used tyramide amplification after hybridization to detect debcl mRNA expression in situ (Fig. 3). In early embryos, debcl mRNA was present uniformly, but became more concentrated in the tissues of the gut later in embryogenesis (Fig. 3, A–D). The relatively high levels of debcl RNA in early embryos (Fig. 3 A; data not shown for precellularized embryos) are likely to be derived maternally, as zygotic transcription does not begin until stage 5. From stage 14 embryos, debcl mRNA was detected in regions in the head that correspond to the pharynx and clypeolabrum, where many TUNEL positive cells are detected (Fig. 3 G). Expression could also be detected in a segmentally reiterated pattern in stage 14 embryos (Fig. 3 C) that may correlate with the TUNEL positive cells that are detected in the nervous system at this stage (Fig. 3 G and 4 C). During 3rd instar larval development, debcl expression was detected in the brain lobes in the outer proliferative center (Fig. 3 J), in the posterior part of the eye imaginal disc (Fig. 3 K), and in the gut (Fig. 3 N) where TUNEL positive cells were clearly seen (Fig. 3 O and 4, E and G). debcl expression was also clearly evident in the salivary glands, particularly in the ducts (Fig. 3 L). Because of background staining problems in salivary glands, acridine orange staining instead of TUNEL was used to detect apoptotic cells in this tissue. Using this technique, no apoptotic cells were detected in 3rd instar salivary glands (Fig. 4 K), suggesting that debcl expression may precede cell death in this tissue. High levels of debcl expression was detected in the nurse cell compartment of stage 10a ovaries (Fig. 3 P), which undergo apoptosis at stage 10b (Foley and Cooley 1998). Thus, debcl expression late in embryogenesis, during larval development, and during oogenesis significantly correlates with tissues undergoing apoptosis.
Debcl Is a Proapoptotic Protein
To investigate whether Debcl is a pro- or antiapoptotic protein in vivo, we generated transgenic flies with the debcl cDNA under control of the yeast UAS-GAL4 promoter. Ectopic expression was then achieved by crossing these flies to various GAL4 drivers. To express debcl in all tissues at various developmental stages, UAS-debcl flies were crossed to hsp70-GAL4 flies and embryos or larvae were heat shocked. Heat shock-induced expression of debcl resulted in enhanced levels of TUNEL positive cells in the embryo (Fig. 4B and Fig. D) and in larval tissues (Fig. 4F and Fig. H; data not shown).
Tissue specific drivers were then used to express debcl during larval development. Ectopic expression of debcl in the posterior region of the eye imaginal disc using the GMR-GAL4 driver (Ellis et al. 1993) resulted in increased acridine orange staining cells in the posterior region of the eye (Fig. 4 I). Similarly, expression throughout the eye imaginal disc of 2nd instar larvae using the eyeless-GAL4 driver (Hauck et al. 1999) resulted in increased TUNEL positive cells in the anterior and posterior regions of the eye (Fig. 4 J). We predicted that expression of debcl from eye specific drivers would result in adults with ablated eyes, as does expression of rpr, hid, and grim from the GMR enhancer (White et al. 1994; Grether et al. 1995; Chen et al. 1996). Surprisingly, despite the increase in apoptotic cells seen in the imaginal discs, the adult flies from these crosses exhibited only a mild rough eye phenotype (data not shown), possibly because of the excess number of cells that are normally generated during eye development. However, other UAS-debcl lines, which presumably have a much higher level of expression, resulted in adults with severely ablated eyes when crossed to GMR-GAL4 (see below). We also expressed debcl in the larval salivary gland using a salivary gland specific driver, 109-88-GAL4, which resulted in a massive increase in acridine orange staining cells (Fig. 4 L) and a reduction in the size of the salivary glands (not shown). Thus, debcl induces cell death when ectopically expressed in a number of different tissue types during Drosophila development, indicating that Debcl is a proapoptotic protein of the Bcl-2 family.
Debcl Function Requires the BH3 Domain and Is Caspase-dependent
To characterize further the biological activity of Debcl, we expressed debcl in Drosophila SL2 cells under the control of an inducible insect promoter. Within 16 h of transfection, Debcl induced apoptosis in a majority of the transfected SL2 cells (Fig. 5 A). By 48 h, all debcl transfected cells had been lost (not shown). This cell death was partially inhibited by the cell permeable peptide caspase inhibitor zVAD-fmk and much more effectively by baculovirus caspase inhibitor P35, indicating that Debcl-induced apoptosis is, at least in part, mediated by caspases. While zVAD-fmk is an efficient inhibitor of many mammalian caspases, it is not known whether it can inhibit Drosophila caspases as effectively. Therefore, the partial inhibition of Debcl-induced cell death by zVAD-fmk may reflect its inability to efficiently inhibit all Drosophila caspases. To confirm that Debcl's cell killing function is dependent on caspase activity, we crossed debcl transgenic flies with GMR-p35 flies. As discussed below and shown in Fig. 6, in the resulting flies the effect of Debcl in eye ablation was significantly reduced.
In several proapoptotic Bcl-2 members, the BH3 domain is essential for their killing function (Adams and Cory 1998; Gross et al. 1999). To determine whether the BH3 domain in Debcl was required for its proapoptotic function, we generated two substitution mutants (L146G and E151G) of the Debcl BH3 domain and analyzed their killing activity in SL2 cells. The 146L residue is conserved in the BH3 domains of most proapoptotic Bcl-2 members, whereas 151E corresponds to an acidic residue in most BH3 domains. Whereas the L146G mutation partially inhibited apoptosis induction by Debcl, E151G mutation completely abrogated Debcl-mediated cell killing (Fig. 5 A). Drosophila proteins Grim, Reaper, and Hid are able to induce apoptosis in mammalian cells (McCarthy and Dixit 1998; Claveria et al. 1998; Haining et al. 1999), despite the fact that mammalian homologues of these proteins have not been found. To determine whether Debcl can also induce apoptosis in mammalian cells, we cloned debcl cDNA in a mammalian expression vector and transfected it into NIH 3T3 cells. Most of the debcl-transfected cells underwent apoptosis (Fig. 5 B). When Debcl was cotransfected with expression vectors carrying caspase inhibitors P35, MIHA, or IAP (reviewed in Ekert et al. 1999), a substantial decrease in apoptosis was evident. These results indicate that Debcl-induced killing is dependent upon its BH3 domain and requires caspase function. In addition to caspase inhibitors, coexpression of prosurvival Bcl-2 and Bcl-xL proteins also significantly inhibited Debcl-induced apoptosis (Fig. 5 B).
Genetic Interactions of debcl with p35, H99, diap1, and dark
After screening a number of UAS-debcl lines, two lines (UAS-debcl#26 on chromosome III and UAS-debcl#18 on chromosome II) were found that, when crossed to GMR-GAL4, gave rise to adults with severely ablated eyes (Fig. 6B and Fig. D). To use this phenotype to examine genetic interactions, a stock was generated containing GMR-GAL4 (2nd chromosome) and UAS-debcl#26. To examine whether the rough eye phenotype was due to the activity of caspases, we crossed GMR-p35 to these flies and examined the eye phenotype of the progeny. As shown in Fig. 6 E, GMR-p35 significantly improved the severe rough eye phenotype of GMR-GAL4; UAS-debcl#26 eyes (Fig. 6 D). These results confirm that Debcl functions in a caspase-dependent fashion upstream of caspase activation.
To determine the involvement of rpr, hid, or grim in the GMR-GAL4; UAS-debcl#26 eye phenotype, we crossed these flies to a deficiency that removes all three genes (Df(3L)H99). If rpr, hid, or grim are rate limiting for Debcl function, then suppression of the GMR-GAL4; UAS-debcl#26 eye phenotype would be expected. However, no significant suppression of this phenotype was observed (Fig. 6 F), suggesting that the GMR-GAL4; UAS-debcl#26 eye phenotype is not dependent on the gene dosage of rpr, hid, or grim.
Next, we tested whether the inhibitor of apoptosis (IAP) homologue, diap1, genetically interacted with debcl, by examining the GMR-GAL4; UAS-debcl#26 eye phenotype when the dosage of diap1 was halved. Halving the dosage of diap1, using two different deficiencies, resulted in a strong enhancement of the GMR-GAL4; UAS-debcl#26 eye phenotype (Fig. 6 G). Furthermore, there was a significant reduction in the number of flies expected containing either of the diap1 deficiencies and GMR-GAL4; UAS-debcl#26. This was possibly due to leaky expression of the GMR-GAL4; UAS-debcl#26 construct in other tissues during development and the enhancement of this effect by reducing the dose of diap1. Thus, diap1 genetically interacts with debcl. We did not observe any genetic interaction between debcl and diap2 when a diap2 deficiency was crossed with GMR-GAL4; UAS-debcl#26 (data not shown).
Recently, a mutation in the Drosophila Apaf1/ced4 homologue, dark, has been described (Rodriguez et al. 1999). To assess the effect of reducing the dosage of dark on the GMR-GAL4; UAS-debcl#26 eye phenotype, a hypomorphic allele of dark (darkCD8) was crossed to GMR-GAL4/CyO; UAS-debcl#26/TM6B flies. As shown in Fig. 6 H, reducing the dosage of dark suppressed the rough eye phenotype of GMR-GAL4; UAS-debcl#26 flies. Therefore, dark genetically interacts with debcl.
Debcl Interacts with Bcl-2 and its Prosurvival Homologues
Since Debcl induces cell death, which is partly inhibited by the overexpression of Bcl-2 and Bcl-xL, an attractive hypothesis is that Debcl binds to and neutralizes prosurvival Bcl-2 homologues. Because no prosurvival Bcl-2–like proteins have been identified so far in Drosophila, we tested if Debcl can bind to any of the known mammalian or viral prosurvival homologues of Bcl-2. In transient overexpression experiments in mammalian cells, Debcl associated with Bcl-2 and most of its functional homologues, although the binding to Bcl-xL, Mcl-1, and adenovirus E1B19K protein was relatively weaker (Fig. 7). For these interaction studies, we used a method involving radiolabeled cell extracts that allows the simultaneous detection of two interacting proteins in the same sample (Fig. 7 A). We also used conventional immunoblotting of the immunoprecipitated proteins and obtained similar results (Fig. 7 B). These data clearly show that Debcl can interact with most of the known prosurvival Bcl-2 proteins and is likely to induce cell death by the same molecular mechanisms as other proapoptotic Bcl-2–related proteins. These results further provide evidence for the functional conservation of Bax-like proteins in mammals and flies. We also tested whether Debcl interacts with the proapoptotic members of Bcl-2 family. In coimmunoprecipitation experiments, Debcl did not associate with any of the BH3-only proteins (Bik, Bid, Bad, and Bim) or the BH1-, BH2-, and BH3- containing proteins (Bax and Bak; data not shown).
Debcl Is Required for Embryonic Cell Death
Currently, no specific debcl mutants are available. Therefore, to examine the in vivo function of Debcl, we carried out RNAi studies to inhibit debcl gene function. RNAi is a powerful technique to disrupt the function of specific genes (Hunter 1999). Additionally, RNAi has the advantage of ablating maternally contributed mRNA that is difficult to achieve genetically. It was originally used in C. elegans, but recently has been successfully applied to Drosophila (Bhat et al. 1999; Misquitta and Paterson 1999). Precellularized embryos were injected with debcl double-stranded RNA, aged until stage 16, and analyzed by TUNEL staining. As shown in Fig. 8B–E, debcl RNAi resulted in a large reduction in TUNEL positive cells in the embryos. We analyzed at least 200 embryos injected with the debcl double-stranded RNA and, in all cases, a 50–90% reduction in TUNEL positive cells was evident. In control experiments, embryos injected with the injection buffer did not show any inhibition of apoptosis (as in Fig. 8 F). In fact, most of the buffer-injected embryos showed slightly higher numbers of TUNEL positive cells, as compared with uninjected embryos (compare Fig. 8A and Fig. F). These RNAi results indicate that debcl gene function is required for programed cell death in the embryos. Additionally, these data suggest that maternally deposited debcl mRNA may be required for normal cell death in fly embryos.
In this paper, we have described the identification of two Bcl-2 homologues in Drosophila. We have shown that one of these, Debcl, is a proapoptotic protein. Given that the existence of a proapoptotic Bcl-2 protein in Drosophila is now established, it can be envisaged that antiapoptotic Bcl-2 proteins are also present in insects. Unlike in C. elegans, which has a single Bcl-2 homologue, CED-9, Drosophila contains at least two such proteins. Interestingly, Debcl-like proteins, which are structurally similar to the mammalian Bax subclass, are not found in C. elegans. Also, considering that Drosophila contains at least six caspases, most of which have been implicated in apoptosis execution, it is likely that the degree of complexity of apoptotic pathways in the fly is much closer to that in mammals than in the worm.
Although debcl expression is low during embryonic and larval development, the expression appears to correlate with cell death in various tissues. With the exception of early embryos, our TUNEL data suggest that debcl may be expressed mainly in the cells that are destined to die. The function of higher debcl mRNA levels in early embryos is currently not known. Transgenic experiments show that ectopic Debcl expression is accompanied by potent apoptosis induction. This may explain why debcl expression is mostly limited to cells that are committed to undergo programed cell death during development. The closest mammalian relative of Debcl, Bok, is mainly expressed in adult reproductive tissues (Hsu et al. 1997). In adult female flies, debcl expression is relatively high and may be mainly contributed by the ovaries. Thus, debcl may function throughout embryonic and larval development, and also in the apoptosis of nurse cells in the adult ovaries. Our in vitro and in vivo data with P35 clearly shows that debcl-induced cell death is caspase-dependent and that Debcl lies upstream of caspases in the death pathway. Additionally, an inhibition of debcl eye phenotype in dark mutant flies further substantiates the finding that Debcl lies upstream of the Dark-mediated caspase activation pathway (Fig. 9). Given that debcl RNAi suppresses most of the programed cell death in embryos, debcl is likely to be a critical regulator of cell death during embryogenesis. Since embryos injected with the debcl double-stranded RNA do not progress into larval development, it was not possible to study the function of debcl later in development using RNAi technique.
Our genetic data show that gene dosage of rpr, hid, and grim have no effect on Debcl-induced apoptosis. This suggests that debcl either lies downstream of the H99 genes or on a separate pathway (summarized in Fig. 9). Our data also show that diap1, but not diap2, genetically interacts with debcl. Since diap1's function as a prosurvival gene is antagonized by the genes of H99 complex (Wang et al. 1999), it is possible that debcl is a component of this pathway (Fig. 9). On the other hand, as diap1 is known to inhibit caspases (Wang et al. 1999), the enhancement of debcl transgenic phenotype when diap1 dosage is halved simply may be due to reduced caspase inhibition in these flies. Future studies involving generation of debcl mutant flies and crossing these flies with rpr, hid, and grim transgenic flies should address whether the H99 genes and debcl indeed lie in the same death pathway.
Debcl binds to many of the mammalian prosurvival Bcl-2 homologues. Thus, it is likely that Debcl functions in a manner analogous to the mammalian or worm proapoptotic Bcl-2–like proteins (Adams and Cory 1998; Gross et al. 1999). These proteins probably act by binding to, and neutralizing the activity of, Bcl-2 or its closest relatives. Bcl-2 and its putative Drosophila homologue(s) probably control the activation of initiator caspases by targeting the adaptor molecules, such as Apaf-1/CED-4/Dark, by regulating the release of apoptogenic factors, such as cytochrome c from the mitochondria (Green and Reed 1998). Bcl-2 or its homologues may perform this function by maintaining organelle integrity, possibly by regulating the mitochondrial membrane pores, such as those containing voltage-dependent anion channel (VDAC; Shimizu et al. 1999), or by controlling a CED-4/CED-3 containing “apoptosome” complex (Hengartner 1997). Most proapoptotic proteins are normally sequestered away from Bcl-2 until receipt of a death stimulus (Adams and Cory 1998; Gross et al. 1999).
An additional level of complexity in mammals is provided by observations that some proapoptotic Bcl-2 relatives, such as Bax and Bak that contain the BH1, BH2, and BH3 regions, may induce caspase-independent death (Xiang et al. 1996; Antonsson et al. 1997), possibly by forming pores in membranes (Antonsson et al. 1997). Although Debcl belongs to this subclass of proteins, as discussed above, our studies indicate that Debcl-induced apoptosis is caspase-dependent. Mutation data indicate that the BH3 domain of Debcl is essential for Debcl-mediated cell killing. It is possible that BH3-mediated interactions with prosurvival members of the Bcl-2 family are necessary for the proapoptotic function of Debcl, in a manner similar to BH3-only proteins (Adams and Cory 1998; Gross et al. 1999).
The prosurvival Bcl-2 proteins have not been reported in Drosophila to date. One possible candidate may be the putative product of the gene at 48A-E locus (Fig. 1). The available cDNA and genomic sequence can encode a putative protein with BH1, BH2, and BH3 domains (Fig. 1 D) similar to Debcl. As we do not currently have information about the NH2-terminal region of this protein, we do not know if the 48A-E protein contains a BH4-like region. Further analysis will determine whether the 48A-E protein is a proapoptotic or a prosurvival Bcl-2 homologue in the fly. Given that Drosophila homologues of CED-3, CED-4, and CED-9 have now been discovered, it is likely that flies also contain BH3-only proteins, such as EGL-1.
We thank J. Beaumont for technical assistance, J. Abrams for dark mutant flies, D. Vaux for MIHA cDNA, Y.N. Jan for 109-88-GAL4 fly stocks, K. White for TUNEL protocol, U. Theopold for help with SL2 cells, and W. Sullivan for protocols on RNAi.
This work was supported by the Wellcome Trust and the National Health and Medical Research Council. S. Kumar and H. Richardson are Wellcome Senior Fellows in Medical Science. D.C.S. Huang is a Special Fellow of the Leukemia Society of America.
Paul A. Colussi and Leonie M. Quinn contributed equally to this work.
Abbreviations used in this paper: BH domain, Bcl-2 homology domain; GFP, Aequorea victoria green fluorescent protein; RNAi, RNA interference; RT, reverse transcriptase; SL2 cells, Schneider L2 cells.