Advertisement

Induction of apoptosis by Drosophila reaper, hid and grim through inhibition of IAP function

Lakshmi Goyal, Kimberly McCall, Julie Agapite, Erika Hartwieg, Hermann Steller

Author Affiliations

  1. Lakshmi Goyal1,
  2. Kimberly McCall2,
  3. Julie Agapite1,
  4. Erika Hartwieg1 and
  5. Hermann Steller*,1
  1. 1 Howard Hughes Medical Institute, Massachusetts Institute of Technology, Departments of Biology, 77 Massachusetts Avenue, Cambridge, MA, 02139, USA
  2. 2 Present address: Department of Biology, Boston University, 5 Cummington Street, Boston, MA, 02215, USA
  1. *Corresponding author. E-mail: hsteller{at}mit.edu

Abstract

Induction of apoptosis in Drosophila requires the activity of three closely linked genes, reaper, hid and grim. Here we show that the proteins encoded by reaper, hid and grim activate cell death by inhibiting the anti‐apoptotic activity of the Drosophila IAP1 (diap1) protein. In a genetic modifier screen, both loss‐of‐function and gain‐of‐function alleles in the endogenous diap1 gene were obtained, and the mutant proteins were functionally and biochemically characterized. Gain‐of‐function mutations in diap1 strongly suppressed reaper‐, hid‐ and grim‐induced apoptosis. Sequence analysis of these alleles revealed that they were caused by single amino acid changes in the baculovirus IAP repeat domains of diap1, a domain implicated in binding REAPER, HID and GRIM. Significantly, the corresponding mutant DIAP1 proteins displayed greatly reduced binding of REAPER, HID and GRIM, indicating that REAPER, HID and GRIM kill by forming a complex with DIAP1. These data provide strong in vivo evidence for a previously published model of cell death regulation in Drosophila.

Introduction

Programmed cell death or apoptosis is crucial to normal embryonic development and metamorphosis of multicellular organisms (Thompson, 1995; Jacobson et al., 1997; McCall and Steller, 1997). In Drosophila, the induction of apoptosis requires the products of reaper, hid and grim, genes encoded by the 75C1,2 region of the Drosophila third chromosome (White et al., 1994, 1996; Grether et al., 1995; Chen et al., 1996). reaper, hid and grim are transcriptionally regulated in response to diverse death‐inducing signals and appear to link distinct signaling pathways with the cell death program (Nordstrom et al., 1996; Kurada and White, 1998). In addition, HID is post‐transcriptionally regulated by the RAS‐MAP kinase pathway in response to cell survival signals (Bergmann et al., 1998a,b). Phosphorylation of HID by activated MAPK inhibits the pro‐apoptotic activity of HID (Bergmann et al., 1998a). Ectopic expression of reaper, hid and grim induces cell death in different tissues of transgenic animals and in cultured insect and mammalian cells (Grether et al., 1995; Hay et al., 1995; Chen et al., 1996; Pronk et al., 1996; White et al., 1996; Vucic et al., 1997, 1998a; McCarthy and Dixit, 1998; Haining et al., 1999). This ectopic cell death is blocked by baculovirus p35, a caspase inhibitor protein (Bump et al., 1995) and by chemical inhibitors of caspases (Hay et al., 1994; Grether et al., 1995; Chen et al., 1996; White et al., 1996; McCarthy and Dixit, 1998; Haining et al., 1999), indicating that reaper, hid and grim induce apoptosis by activating a caspase pathway. Several caspase‐like molecules have been identified in Drosophila (Fraser and Evan, 1997; Song et al., 1997; Chen et al., 1998; Dorstyn et al., 1999), but the precise mechanism of their activation is not clear.

Apoptosis is negatively regulated by another conserved and important class of molecules, the inhibitor of apoptosis proteins (IAPs) (Deveraux and Reed, 1999). IAPs were first described in baculoviruses based on their ability to rescue the p35 mutant phenotype and were shown to exert anti‐apoptotic activity (Crook et al., 1993; Birnbaum et al., 1994; Clem and Miller, 1994; Miller, 1997). Since then, several cellular homologs from both vertebrates and invertebrates have been described (Clem and Duckett, 1997; Deveraux and Reed, 1999). Typically, IAP family genes have at least one and often two or three tandem baculovirus IAP repeat (BIR) motifs and may have an additional C‐terminal RING finger domain (Clem and Duckett, 1997; Deveraux and Reed, 1999). Some mammalian IAPs have been implicated in human diseases, including spinal muscular atrophy (SMA) (Roy et al., 1995) and cancer (Ambrosini et al., 1997). In Drosophila, two IAP homologs have been reported: diap1 and diap2 (Hay et al., 1995; Duckett et al., 1996; Uren et al., 1996). diap1 is encoded by the thread (th) locus (Lindsley and Zimm, 1992), and loss‐of‐function mutations in diap1 are lethal and enhance reaper‐, hid‐ and grim‐induced cell death (Hay et al., 1995). Ectopic expression of either diap1 or diap2 suppresses apoptosis (Hay et al., 1995; Vucic et al., 1997, 1998a). Baculovirus, Drosophila and mammalian IAPs can physically interact with REAPER, HID and GRIM and antagonize their pro‐apoptotic properties (Vucic et al., 1997, 1998a; McCarthy and Dixit, 1998). Human XIAP, cIAP1 and cIAP2 can bind to and inhibit caspase 3, 7 and 9 in vitro (Deveraux et al., 1997, 1998; Roy et al., 1997). These observations have led to the proposal that IAPs are apoptosis antagonists exerting their inhibitory effect at several levels in the apoptotic cascade and, in Drosophila, inhibit REAPER, HID and GRIM from activating effectors downstream in the death program (McCarthy and Dixit, 1998; Vucic et al., 1998b). More recently, however, results obtained primarily from a yeast expression system suggest that REAPER, HID and GRIM block DIAP1 from inhibiting caspase activity (Wang et al., 1999). Consistent with this observation, loss‐of‐function diap1 mutants are embryonic lethal, with the widespread induction of apoptosis (Wang et al., 1999; this study). We now report the isolation and characterization of several gain‐of‐function and loss‐of‐function mutants in the endogenous diap1 gene. The gain‐of‐function mutant DIAP1 proteins are impaired for binding to REAPER and HID. This suggests that the association of REAPER and HID with DIAP1 is critical for the induction of apoptosis by these pro‐apoptotic genes in vivo. Collectively, these data provide strong support for the idea that REAPER, HID and GRIM kill by inhibiting DIAP1's ability to antagonize caspase function.

Results

Mutations in diap1 modify reaper‐ and hid‐induced cell death phenotypes

Ectopic expression of reaper, hid and grim under the control of an eye‐specific promoter (GMR) results in ectopic cell death in the developing retina and causes rough and reduced eyes (Grether et al., 1995; Hay et al., 1995; Chen et al., 1996; White et al., 1996). The phenotypes caused by GMR‐reaper and GMR‐hid transgenes are dosage dependent and very sensitive to the dosage of other cell death genes acting downstream or in parallel pathways. This was the basis of a mutagenesis screen to isolate modifiers of reaper‐ and hid‐induced cell death (J.Agapite, K.McCall and H.Steller, unpublished) (Bergmann et al., 1998a). Among a large number of dominant modifiers of apoptosis obtained in this screen, at least 10 were mutations in diap1. As expected, loss‐of‐function diap1 alleles were strong enhancers of both reaper‐ and hid‐induced eye phenotypes (Hay et al., 1995) (Figure 1; Table I). However, we also obtained several gain‐of‐function mutants of diap1 that potently inhibited cell death and restored the eye morphology to near wild‐type levels (Figure 1; Table I). Another class of mutants that enhance reaper‐induced death but suppress the hid phenotype was also identified.

Figure 1.

Mutations in diap1 modify reaper‐ and hid‐induced cell death. A gain‐of‐function mutation in diap1, th6−3s, suppresses the GMR‐reaper (B) (GMR‐rpr/th6‐3s) and GMR‐hid (E) (GMR‐hid/th6‐3s) induced eye phenotypes. Reduction of diap1 activity as in th11‐3e enhances the eye phenotypes associated with GMR‐rpr (C) (GMR‐rpr/th11‐3e) and GMR‐hid (F) (GMR‐hid/th11‐3e). The unmodified phenotypes are shown in (A) (GMR‐rpr/+) and (D) (GMR‐hid/+). The wild‐type eye (not shown) is similar to that in (B).

View this table:
Table 1. Phenotypes and molecular changes associated with the diap1 alleles

The gain‐of‐function mutants were also tested for suppression of reaper‐ and hid‐induced cell death in the wing (J.Agapite and H.Steller, unpublished) and the suppression of larval lethality induced by the expression of hid under a heat shock promoter (Bergmann et al., 1998a). In all assays tested, the diap1 gain‐of‐function mutants suppressed reaper‐ and hid‐induced cell death very efficiently (data not shown) and the diap1 mutant phenotype was recapitulated outside the context of the developing eye.

Gain‐of‐function mutations in diap1 map to the BIR domains

Molecular analysis revealed that all five gain‐of‐function alleles were associated with single amino acid changes in the BIR domain of DIAP1 (Table I; Figure 2). Significantly, this domain of DIAP1 has been implicated in the binding of REAPER, HID and GRIM (Vucic et al., 1998a,b). The proline residue mutated to serine in th21‐2s is highly conserved among different BIR domains (Uren et al., 1998). In th6‐3s and th45‐2s as well as in th23‐4s and th23‐8s, amino acid substitution of a specific glycine residue in corresponding locations in BIR1 and BIR2, respectively, was found to cause the gain‐of‐function phenotype. The flanking amino acid of the mutated glycine is a conserved non‐polar residue (valine, leucine or isoleucine) in 100% of all BIR domains (Uren et al., 1998). Two previously reported loss‐of‐function thread alleles, th4 and th5 (Hay et al., 1995), were also included in the molecular analysis. th4 is a change in a highly conserved histidine residue (Uren et al., 1998), and the th5 mutation introduces a premature truncation resulting in the deletion of several amino acids at the C‐terminus of DIAP1 (Figure 2).

Figure 2.

Schematic representation of the diap1 mutants. The DIAP1 protein is boxed, with the BIR domains indicated as cross‐hatched regions and the RING domain in black. The gain‐of‐function mutants are indicated in bold and the loss‐of‐function mutants are in normal face. The figure is drawn to scale.

To confirm that the missense mutation in the primary protein sequence of diap1 was responsible for the observed phenotype, we employed a cell death assay in transfected culture cells. Expression of reaper and hid can induce apoptosis in several cultured cell types (Pronk et al., 1996; Bergmann et al., 1998a; McCarthy and Dixit, 1998; Vucic et al., 1998a; Haining et al., 1999), which is inhibited efficiently by co‐transfecting IAPs (McCarthy and Dixit, 1998; Vucic et al., 1998a; Haining et al., 1999). Representative gain‐of‐function and loss‐of‐function diap1 alleles were tested in Drosophila Schneider (S2) cells for their ability to overcome reaper‐ and hid‐induced cell death. All diap1 mutants tested expressed protein at levels comparable to wild‐type, indicating that these mutations did not significantly affect the expression and stability of the protein (Figure 3). Gain‐of‐function mutants showed an ability to inhibit reaper‐ and hid‐induced cell death by 40–50% above wild‐type levels. All the mutants exhibited the same phenotypes as observed in vivo (Table I), except for the th4 allele, which showed a near wild‐type level of protection against hid‐induced death (Figure 3, see below).

Figure 3.

diap1 mutant alleles maintain their phenotype when expressed in S2 cells. (A) The wild‐type and mutants of diap1 were expressed in S2 cells under the control of a viral promoter with a FLAG epitope tag at the N‐terminus of the protein. Protein expression levels were analyzed by immunoblotting equal amounts of protein extracts using an epitope‐specific antibody. Epitope‐tagged versions of wild‐type and mutant proteins are expressed at comparable levels and are indicated with arrows. The th5 mutant is a truncated protein due to the introduction of a premature stop codon. (B) The wild‐type and mutant diap1 constructs were co‐transfected with plasmids for the expression of REAPER and HID in a ratio of 3:4. A β‐galactosidase plasmid was also co‐transfected as a marker. The numbers of viable cells with β‐galactosidase activity were counted and expressed as a percentage of the number of viable blue cells in the wild‐type diap1‐transfected wells, to assess the ability of the mutants of diap1 to inhibit reaper‐ and hid‐mediated apoptosis. Except for th4, all the mutants maintained the phenotypes observed in vivo with respect to reaper‐ and hid‐induced eye ablation (see Table I and text).

Gain‐of‐function mutants of DIAP1 show impaired binding to REAPER and HID

REAPER, HID and GRIM can bind to DIAP1 via the BIR domains (Vucic et al., 1998b), and all the strong diap1 gain‐of‐function suppressor mutations that we obtained in our modifier screen map to a specific region in the BIR domains (Figure 2). Therefore, we hypothesized that these mutations may have altered the binding of DIAP1 to REAPER, HID and GRIM. Based on previous findings that REAPER, HID and GRIM can interact functionally and physically with IAPs (Vucic et al., 1997, 1998a), two simple models can be proposed to explain the role of diap1 in regulating apoptosis. In the first model, diap1 acts upstream of reaper, hid and grim to prevent them from activating the cell death program (McCarthy and Dixit, 1998; Vucic et al., 1998b). Alternatively, reaper, hid and grim may induce apoptosis by inhibiting diap1 from blocking cell death (Wang et al., 1999). According to the later model, mutations in diap1 that inhibit binding of REAPER, HID and GRIM would be less susceptible to inactivation by the pro‐apoptotic proteins and hence show increased anti‐apoptotic activity. To test this hypothesis directly, we performed in vitro binding assays of wild‐type and mutant DIAP1 protein to GST–REAPER and GST–HID. As predicted, gain‐of‐function mutants of diap1 (th6‐3s and th23‐4s) showed greatly diminished binding to both REAPER and HID as compared with wild‐type DIAP1 (Figure 4). The effect of these single amino acid changes appears to be quite specific, since the mutant DIAP1 proteins retain potent anti‐apoptotic activity (Figures 1 and 3; Table I), presumably by direct interaction with caspases (Deveraux et al., 1997, 1998; Roy et al., 1997; Hawkins et al., 1999). Since the anti‐apoptotic activity of IAPs requires a functional BIR domain, it is highly unlikely that these diap1 gain‐of‐function mutations have a general effect on the overall structure of the BIR domains.

Figure 4.

Gain‐of‐function mutations in diap1 diminish binding to REAPER and HID. Equal amounts of 35S‐labeled wild‐type and mutant DIAP1 proteins (IVT input) were tested for association with GST, GST–REAPER and GST–HID in an in vitro binding assay. Wild‐type DIAP1 protein shows specific binding to GST–REAPER and GST–HID. Mutant DIAP1 proteins encoded by th6‐3s and th23‐4s show significantly reduced binding of GST–REAPER and GST–HID as compared with wild‐type DIAP1. In contrast, a protein corresponding to th4, a loss‐of‐function allele of diap1, retains essentially normal affinity for REAPER and HID. The lower band visible in most lanes is due to initiation of translation at the second methionine in DIAP1 during in vitro synthesis. The th5 mutant translates into a truncated protein due to the introduction of a premature stop codon. The IVT input represents 10% of the in vitro translated protein used in each pull‐down reaction.

The gain‐of‐function mutations in diap1 reduce binding to REAPER and HID up to 14‐fold (Figure 4). The residual association with REAPER and HID may be due to the presence of one intact BIR domain in both the mutants (Figure 2). However, it is notable that mutations in either the first or the second BIR domain can reduce binding of HID and REAPER by >2‐fold, suggesting that both BIR domains may cooperate for the efficient binding of REAPER and HID. On the other hand, we cannot exclude the possibility that other regions of DIAP1 are also involved in binding of REAPER, HID and GRIM, perhaps involving domains distinct from the N‐terminal region of these proteins. In any event, our results show that gain‐of‐function alleles of diap1 with increased anti‐apoptotic activity are associated with reduced in vitro binding activity to REAPER and HID. These findings are not reconciled easily with the model that DIAP1 protects against cell death by inhibiting the pro‐apoptotic activity of REAPER and HID. Rather, they suggest that REAPER and HID activate apoptosis by binding to and inactivating DIAP1.

Not all BIR domain mutations have impaired binding to REAPER, HID and GRIM. The th4 mutant, which harbors a single amino acid change in BIR2 (Figure 2), binds to REAPER and HID in vitro (Figure 4). Although this is a loss‐of‐function allele in Drosophila, presumably by lacking the ability to inhibit caspases, overexpression of th4 in cultured cells can block hid‐induced cell death (Figure 3). In order to explain this apparent paradox, we propose that overexpression of the mutant protein at unphysiologically high levels in cultured cells protects against hid‐induced apoptosis by binding to and sequestering HID. While this proposed mechanism is not reflective of the normal function of DIAP1, it explains results from previous cell transfection studies suggesting that IAPs may function upstream of REAPER, HID and GRIM to inhibit apoptosis (McCarthy and Dixit, 1998; Vucic et al., 1998b).

Loss of diap1 results in the induction of apoptosis

Loss‐of‐function diap1 mutants are embryonic lethal (Hay et al., 1995) (Table I). Since diap1 has anti‐apoptotic activity, we investigated whether zygotic loss of diap1 function may lead to excessive apoptosis during embryogenesis. Using the TUNEL technique to label apoptotic nuclei, essentially all nuclei of homozygous th11‐3e embryos stained positive by 7–8 h after egg laying (AEL) (Figure 5). In contrast, only a few TUNEL‐positive cells were observed in correspondingly aged wild‐type embryos (stage 11), the first time point at which cell deaths are normally seen in Drosophila embryos (Figure 5A) (Abrams et al., 1993). Due to the extraordinary amount of apoptosis, thread loss‐of‐function embryos appeared highly disorganized at this time (Figures 5B and E, and 6B–D). However, time course analyses revealed that these embryos develop normally until the cellular blastoderm stage and begin to gastrulate, but become abnormal shortly after the formation of the ventral furrow (J.Tittel and H.Steller, unpublished observations). Analysis of th11‐3e embryos by electron microscopy revealed many nuclei with features characteristic of apoptosis, including condensed chromatin and electron‐dense nuclei (Figure 6B–D). These experiments reveal a striking dependence of cell survival on zygotic diap1 function. We conclude that diap1 is required to prevent apoptosis in most, if not all cells of the Drosophila embryo.

Figure 5.

diap1 function is essential to prevent apoptosis. Embryos were labeled using the TUNEL procedure (A, B and C) to detect apoptotic nuclei (White et al., 1994) or with Oligreen (D, E and F) to stain DNA. (A) In wild‐type embryos, the first evidence of apoptosis is seen at stage 11; the arrows point to clusters of apoptotic cells. (B) A th11‐3e embryo aged to a physiological age of 7 h (7 h AEL). In diap1 loss‐of‐function mutants, such as th11‐3e and th5, essentially all nuclei of embryos are TUNEL positive 7–8 h AEL. (C) th11‐3e Df(3L)H99 double mutant embryos. As with th11‐3e embryos, essentially all nuclei of the early H99 th11‐3e double mutant embryos are TUNEL positive. Therefore, induction of apoptosis in th11‐3e embryos does not require the activity of reaper, hid and grim, indicating that diap1 functions downstream from these genes. (D) Stage 5 (cellular blastoderm) wild‐type embryo. (E) th11‐3e embryo aged to 7 h AEL. At this stage, mutant nuclei have condensed chromatin characteristic of apoptosis, and the embryo is organized abnormally. (F) th11‐3e Df(3L)H99 double mutant embryos. The absence of reaper, hid and grim does not attenuate the th11‐3e phenotype, since the phenotype of th11‐3e and the double mutant embryo is virtually indistinguishable (compare E and F).

Figure 6.

Cell deaths in the diap1 mutant embryo display the ultrastructural features of apoptosis. (A) Electron micrograph of a stage 6 wild‐type embryo. All nuclei have normal ‘healthy’ morphology, and no signs of condensed chromatin can be detected until ∼7 h AEL (stage 11) (Abrams et al., 1993). (BD) Homozygous th11‐3e mutant embryos were selected by morphology 4 h after egg laying and processed for electron microscopy. Representative micrographs of electron‐dense nuclei from th11‐3e homozygous mutant embryos showing condensed chromatin characteristic of apoptosis. The scale bar as represented in (D) is 1 μm.

diap1 functions downstream of reaper, hid and grim

If diap1 functions upstream of reaper, hid and grim, the excessive apoptosis that we observed in diap1 loss‐of‐function embryos should be blocked in the absence of reaper, hid and grim. In contrast, if reaper, hid and grim function upstream of diap1, apoptosis in diap1 loss‐of‐function embryos should be independent of these pro‐apoptotic genes. To distinguish between these two models, we constructed double mutant embryos that are homo– zygous for both H99 and th11‐3e, a deletion that removes reaper, hid and grim. Wild‐type, th11‐3e and H99th11‐3e embryos were analyzed for their phenotypic morphology and by TUNEL labeling as before. We found that the H99th11‐3e phenotype was indistinguishable from the th11‐3e phenotype observed earlier (Figure 5). Similar results were reported recently by Wang et al. (1999). The failure of the H99 deletion to inhibit cell death in diap1 loss‐of‐function embryos is not consistent with the idea that DIAP1 blocks apoptosis by inhibiting REAPER, HID and GRIM.

Discussion

In this study, we report the isolation of novel gain‐of‐function and loss‐of‐function mutants in one of the Drosophila IAP genes, diap1. Gain‐of‐function mutants of diap1 strongly suppressed reaper‐ and hid‐induced cell death and correspond to single amino acid changes in the BIR domain of the DIAP1 protein. The gain‐of‐function mutant DIAP1 proteins displayed impaired binding to REAPER and HID, indicating that the ability of REAPER and HID to bind DIAP1 is critical for the regulation of DIAP1 in vivo. Loss‐of‐function mutations in diap1 caused widespread induction of cell death resulting in embryonic lethality, and a deletion in reaper, hid and grim was unable to rescue the diap1 loss‐of‐function phenotype (Wang et al., 1999; this study). These results indicate that diap1 is required for cell survival and is genetically downstream of reaper, hid and grim.

Based on overexpression studies in cultured cells, it was initially proposed that IAPs function upstream of reaper, hid and grim to prevent them from activating caspases (Vucic et al., 1997, 1998a). Subsequently, IAPs were also reported to bind to and inhibit caspases (Deveraux et al., 1997, 1998; Roy et al., 1997; Kaiser et al., 1998). In order to reconcile this apparent paradox, it was proposed that IAPs act on multiple targets and at different steps in the apoptotic pathway (Kaiser et al., 1998; McCarthy and Dixit, 1998; Vucic et al., 1998b). Based on data obtained primarily from a yeast expression system, Wang et al. (1999) have suggested a more simple linear model for the induction of apoptosis by reaper, hid and grim. According to this ‘double inhibition model’, REAPER, HID and GRIM induce apoptosis by binding directly to and preventing DIAP1 from inhibiting caspase activation (Figure 7). The diap1 mutants analyzed in this study provide evidence for the function of reaper and hid in vivo and concur with the conclusions obtained by Wang et al. (1999) from their study in yeast. The decreased binding of REAPER and HID to DIAP1 correlates with a gain‐of‐function diap1 phenotype, suggesting that the association with DIAP1 is important for the pro‐apoptotic activity of REAPER, HID and GRIM. These data strongly support the idea that in Drosophila, REAPER and HID induce apoptosis by binding to and inhibiting DIAP1 function (Figure 7; Wang et al., 1999). We believe that the previously proposed function of IAPs upstream of reaper, hid and grim is simply an artifact of unphysiologically high levels of protein expression in heterologous systems. When IAP expression constructs are introduced into cultured cells under the control of strong promoters and at high copy numbers, the levels of proteins expressed far exceed those of the endogenous cellular IAP proteins. Under these unphysiological conditions, cellular IAPs can display properties that do not reflect their normal mechanism of action. In particular, our results with th4 and other diap1 alleles (data not shown) demonstrate that mutant proteins that completely lack anti‐apoptotic activity in vivo can still inhibit cell death in vitro as long as they can bind to REAPER, HID and GRIM (Figures 3 and 4). Conversely, gain‐of‐function diap1 alleles that display reduced binding to REAPER, HID and GRIM have strongly increased anti‐apoptotic function in vivo (Figures 1 and 4), but show reduced protection in heterologous cell transfection assays (data not shown). These results clearly reveal the limitations of overexpression studies in cultured cells for determining the normal mechanism of action of these proteins in the cell death pathway.

Figure 7.

diap1 functions downstream of reaper, hid and grim to inhibit caspase activation. The ‘double inhibition’ model presented here and by Wang et al. (1999) indicates that diap1 is required to keep the caspases in check. The binding of reaper, hid and grim to diap1 results in the release of inhibition and therefore the activation of caspases and apoptosis.

A mutant of HID that fails to bind IAPs has no pro‐apoptotic activity (Vucic et al., 1998a; Haining et al., 1999; Wang et al., 1999), implying that HID functions primarily, if not exclusively, by inhibiting IAPs. Although the precise mechanism by which IAPs inhibit apoptosis is not known, there are strong reasons to believe that the primary mode of action is a direct inhibition of caspases and/or their activation (Hawkins et al., 1999; Wang et al., 1999; Meier et al., 2000). Mammalian IAPs can inhibit caspases in vitro (Deveraux et al., 1997, 1998; Roy et al., 1997), and DIAP1 has been shown to interact with and inhibit Drosophila caspases DrICE and DCP1 in heterologous systems (Kaiser et al., 1998; Hawkins et al., 1999). Although these inhibitory activities have been detected with active caspases, we prefer the model that DIAP1 interacts with pro‐caspases to prevent caspase activation (Meier et al., 2000; Song et al., 2000). In order to explain the diap1 loss‐of‐function phenotype, we propose that in the absence of inhibition by DIAP1, caspases are activated either by a weak autocatalytic activity or by the presence of DARK/HAC‐1/dAPAF‐1, the Drosophila APAF1/CED4‐like activator (Kanuka et al., 1999; Rodriguez et al., 1999; Zhou et al., 1999).

How general is the proposed model of IAP regulation and function? In mammals, as in Caenorhabditis elegans and Drosophila, caspases are widely expressed as weakly active zymogens in virtually all cells. This includes cells that survive for the lifetime of the animal, such as neurons. Therefore, powerful control mechanisms must be in place to prevent the inappropriate activation of caspases in cells that should live, and it appears that some IAPs play a critical role in this regard. On the other hand, these ‘brakes on death’ must somehow be removed in cells that are selected to die. In Drosophila, this appears to be mediated by REAPER, HID and GRIM. Although no mammalian homologs of reaper, hid and grim have been reported yet, we expect that a similar regulation must occur in mammalian cells. Consistent with this idea, reaper, hid and grim can activate apoptosis in mammalian cells, and REAPER has been shown to activate caspases and apoptosis‐like events in a vertebrate cell‐free system (Evans et al., 1997; McCarthy and Dixit, 1998; Haining et al., 1999). Many of the residues involved in the diap1 gain‐of‐function mutants have been highly conserved during evolution, indicating that the domain that is necessary for binding to REAPER/HID/GRIM is under considerable selective pressure. Therefore, we consider it likely that the inhibition of IAP function is a general mechanism of cell death induction. If so, gain‐of‐function mutations similar to those described here may contribute to aberrant regulation of apoptosis in human diseases, such as cancer.

Materials and methods

Genetic screen and fly stocks

The genetic screen performed to isolate modifiers of GMR‐hid has been described previously (Bergmann et al., 1998a). In a similar screen for dominant modifiers of GMR‐reaper‐induced eye ablation, 180 000 mutagenized F1 progeny were screened and 46 dominant modifiers were identified (J.Agapite, K.McCall and H.Steller, unpublished). The following mutant and transgenic fly strains were used for the genetic analysis: GMR‐hid and GMR‐reaper (Bergmann et al., 1998a), th11‐3e and th6‐3s (this study). H99th11‐3e recombinants were generated by standard Drosophila genetic techniques. All crosses were performed at 25°C.

Molecular analysis and plasmid construction

Single embryos of the diap1 mutants over a balancer with a LacZ marker were selected 4 h AEL and squashed in 10 μl of Gloor and Engel's buffer (10 mM Tris pH 8.2, 1 mM EDTA, 25 mM NaCl, 200 μg/ml proteinase K). The squashed single embryo was incubated at 37°C for 30 min and at 95°C for 2 min to inactivate proteinase K. A 4 μl aliquot of the single embryo DNA was used in each PCR with primers specific for the open reading frame (ORF) of diap1. Primers specific for LacZ were used in an independent PCR to identify homozygous diap1 mutant embryos. The PCR product from homozygous mutant embryos was either sequenced directly or cloned into pIE1–3 vector (Novagen) with a FLAG epitope tag at the N‐terminus using restriction sites engineered into the primers. Cleaned PCR product or plasmid DNA carrying the mutant diap1 coding sequence was subjected to automated sequencing to determine the molecular nature of the mutation. Mutations were confirmed by sequencing the diap1 ORF from multiple independent homozygous mutant single embryos. Similar procedures were used to clone and sequence the ORF of wild‐type diap1 from the background strain as control.

Cell culture assays

SL2 cell transient transfection experiments were performed as described (Bergmann et al., 1998a). Briefly, 0.15 μg of the FLAG epitope‐tagged wild‐type or mutant diap1 cDNA in the pIE1–3 vector (Novagen) were co‐transfected with 0.2 μg of reaper or hid cDNA in the pIE1–3 vector into each well of a 24–well plate of SL2 cells. pIE1–3 LacZ (18 ng) was included in all transfections as a reporter. Transfections were performed using Cellfectin (Gibco‐BRL) according to the manufacturer's instructions. SL2 cells adhere poorly to culture dishes and apoptotic cells float up into the culture medium. At 24 h post‐transfection, cells still adhering to the dish are viable. Therefore, 24 h post‐transfection, the cells still adhering to the dish were fixed and stained for LacZ activity and the number of blue cells was counted as an indicator of cell survival. All plasmid combinations were transfected in triplicate and the mean and standard deviation calculated. For Western blots, extracts were made from SL2 cells transfected with FLAG‐tagged wild‐type or mutant pIE1–3 diap1 plasmids and resolved by SDS–PAGE. After transfer of the proteins to nitrocellulose membranes, the blots were probed with a monoclonal antibody directed to the FLAG epitope tag (Kodak). An anti‐mouse horseradish peroxidase (HRP)‐conjugated secondary antibody was used and the immune complexes were detected by enhanced chemiluminescence (Amersham).

TUNEL labeling of embryos

For TUNEL and Oligreen staining, Canton S, th11‐3e or H99th11‐3e embryos were aged for 16 h at 18°C before processing. TUNEL was carried out as previously described (White et al., 1994) with some modifications. Briefly, embryos were fixed in 4% paraformaldehyde, in phosphate‐buffered saline (PBS) with heptane, and devitellinized with methanol. After several washes with PBS plus 0.1% Tween‐20 (PBT), the embryos were treated with proteinase K (10 μg/ml), washed in PBT and treated with reagents from the Apotag Plus kit (Oncorr). For detection, an alkaline phosphatase‐conjugated antibody to digoxigenin and reagents from the genius kit (both from Boehringer Mannheim) were used. For Oligreen (Molecular Probes) staining to visualize the nuclei, fixed and washed embryos were incubated with a 1:5000 dilution of Oligreen in PBS. Embryos were examined and photographed on a Zeiss Axiophot microscope.

Electron microscopy

th11‐3e homozygous mutant embryos were aged for 4–5 h AEL and selected under a dissection microscope. Embryos were dechorionated and fixed with 1.5% glutaraldehyde, 1% paraformaldehyde and 0.75% acrolein in 0.1 M cacodylate buffer (pH 7.4). Fixed embryos were devitellinized by hand peeling and post‐fixed in 1% osmium tetroxide. Thin sections were cut after embedding in an eponate mixture and photographed on a Jeol 1200EX II microscope.

In vitro binding assays

Full‐length coding sequences of reaper and hid in the pGEX vector (Amersham Pharmacia Biotech) (kindly provided by Vivian Hua and Andreas Bergmann) or the pGEX vector alone were used to generate recombinant GST, GST–REAPER and GST–HID proteins in Escherichia coli. [35S]Methionine‐labeled wild‐type and mutant DIAP1 proteins were generated in an in vitro transcription–translation system (Promega) using T7 expression plasmids. Equivalent amounts of the labeled proteins, as determined by SDS–PAGE analysis of one‐tenth of the in vitro translation mix, were incubated with 2 μg of the GST or GST fusion proteins immobilized on glutathione–Sepharose beads. The binding was performed for 2 h at 4°C in the buffer described previously (Vucic et al., 1997) containing 1% NP–40 and washed four times with the same buffer. Protein‐bound Sepharose beads were boiled in SDS‐containing sample buffer and resolved on 10% SDS–PAGE gels. Gels were fixed, dried and autoradiographed.

Acknowledgements

We thank A.Bergmann, N.Haining, J.Tittel and other members of the laboratory for useful discussions and suggestions, E.Seling for technical help with scanning electron microscopy, and J.Ginsburg for help with sequencing. K.M. was supported by a post‐doctoral fellowship from the American Cancer Society and L.G. is a fellow of The Jane Coffin Childs Memorial Fund for Medical Research. L.G. was an associate and H.S. is an investigator of the Howard Hughes Medical Institute.

References