INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Programmed cell death or apoptosis is a gene-directed cell suicide process that is found in virtually all metazoans to eliminate unwanted, damaged, or harmful cells (13, 19, 23). Many different stimuli can induce apoptosis, including the binding of certain ligands to cell surface death receptors, removal of extracellular survival signals, steroid hormones, DNA damage, and viral infection (1, 6). Regardless of the stimulus, however, cell killing is carried out by a family of well-conserved cysteine proteases called caspases (Cys Asp protease) (21, 29). Caspases are synthesized as inactive or weakly active proenzymes that have to be cleaved at conserved Asp residues to form the active tetrameric protease (30, 34). A combinatorial approach has been developed to determine the precise specificity of each caspase (28). It has been proposed that caspases may function in a proteolytic cascade consisting of "initiator caspases" that process and thereby activate downstream "effector caspases." According to this model, effector caspases are thought to cleave proteins that are essential for maintaining cellular structure and function (21, 28, 29).

A key question that is still incompletely understood is how caspases are activated in response to many different proapoptotic signals. One important biochemical event during caspase activation is the removal of an inhibitory N-terminal prodomain. This step involves cleavage at specific Asp residues. Many active caspases are able to autoprocess their zymogens and those of other caspases in vitro. However, in most cases it remains to be determined whether the reactions that have been observed in vitro actually occur during apoptosis in vivo.

In Drosophila melanogaster, the induction of apoptosis requires the activity of three closely linked genes, reaper (rpr), head involution defective (hid), and grim (4, 10, 32). These genes have a partially redundant function and kill by activating a caspase pathway (33, 36). Although each gene is able to induce apoptosis independently, the regulation and function of these three genes appear to be different. During development, rpr and grim are specifically expressed in the cells that are doomed to die. hid, on the other hand, is also expressed in many cells that are going to live (4, 10, 32). Significantly, the proapoptotic activity of HID, but not of Reaper and Grim, is inhibited by the Ras/mitogen-activated protein kinase (MAPK) pathway (2). Furthermore, there is evidence that cell death induced by Reaper and Grim is somewhat distinct from HID-induced death (36). Therefore, it is possible that rpr and hid activate different sets of caspases.

Two very similar Drosophila caspases, DCP-1 and drICE, have been previously identified and characterized (7, 8, 22). The two proteins have 57% amino acid identity, and they both have short prodomains, indicating that they may function as effector caspases. One difference between these two proteins is in the N-terminal portions of their p20 subunits: drICE contains a higher number of Ser and Gly residues in this region than DCP-1. DCP-1 has enzymological properties very similar to those of CED-3 (22). DCP-1 can induce apoptosis in insect and mammalian cells and apoptosis-like events in a cell-free system. Loss of zygotic dcp-1 function causes larval lethality and melanotic tumors, indicating an essential role of dcp-1 in development (22). dcp-1 function is also required for normal nurse cell death during oogenesis in Drosophila (17). Overexpression of full-length drICE sensitizes Drosophila cells to apoptotic stimuli, and an N-terminally truncated version of drICE can induce apoptosis in Drosophila SL2 cells. Moreover, treatment of SL2 cells with different death inducers results in proteolytic processing of drICE (7).

Ectopic expression of cell death genes under the control of eye-specific promoters, such as GMR, is a useful system to define genetic interactions among different components of the cell death pathway in Drosophila (4, 10, 11, 33). In this study, we have generated transgenic flies that carry either the full-length or truncated forms (without the prodomain) of DCP-1 and drICE under the control of the GMR promoter. We find that expression of a truncated dcp-1 transgene, GMR-N-dcp-1, produces small and rough eye phenotypes due to extra cell death in the developing eye. In contrast, flies carrying one copy of the full-length dcp-1 transgene, GMR-fl-dcp-1, are almost normal. Expression of either full-length or truncated drICE in the developing retina had no obvious effect. Interestingly, GMR-rpr and GMR-grim, but not GMR-hid, dramatically enhanced the eye phenotype of GMR-fl-dcp-1 flies, suggesting that dcp-1 may function downstream of rpr and grim.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Expression and purification of a truncated version of DCP-1 protein and a truncated version of drICE protein. A truncated version of DCP-1 protein, with its N-terminal 33 amino acids removed and a six-His tag linked to its C terminus, was expressed in Escherichia coli and purified with Ni²⁺ columns as described before (22). A truncated version of drICE, with its N-terminal 28-amino-acid prodomain removed and a six-His tag attached to the C terminus, was generated by PCR using drICE cDNA as the template, with the upstream primer 5'GGCAAACATATGGCCCTGGGCTCCGTGGGATCC3' and the downstream primer 5'CTCTCACATATGTCAGTGGTGGTGGTGGTGGTGAACCCGTCCGGCTGGAGCCAA3'. The PCR product was treated with NdeI and ligated into NdeI- and phosphatase-treated pET3a vector (Novagen). The clones with inserts in the correct orientation were amplified and sequenced. The six-His-tagged truncated drICE protein was expressed in E. coli BL21(ED3) by IPTG (isopropyl--D-thiogalactopyranoside) induction and purified with Ni²⁺ columns as described before (22).

Determination of the substrate specificity of DCP-1. The synthesis and preparation of the positional scanning synthetic combinatorial library used in this study have been described previously, and this library has been used to determine the proteolytic specificities of nine human caspases, Caenorhabditis elegans CED-3, and cytotoxic T-lymphocyte-derived granzyme B (28). The general structure of the library, Ac-[P4]-[P3]-[P2]-Asp-AMC, permits the determination of caspase amino acid preferences in P2, P3, and P4 positions. Asp is kept invariant at P1, owing to the stringent P1 Asp specificity of all caspases, and the fluorogenic AMC moiety was incorporated in the P1 position to monitor proteolysis. The entire library contains 60 mixtures of 400 compounds each (20 × 20) thus yielding 8,000 distinct peptides, each in triplicate. Each of the 60 mixtures was prepared as a 10 mM stock in dimethyl sulfoxide. To determine protease specificity, an enzyme was added to each of the 60 reaction mixtures, which consisted of 100 µM substrate mixture (0.25 µM concentration of each of the 400 distinct peptides), 100 mM HEPES-KOH (pH 7.5), and 10 mM dithiothreitol (DTT) in a total volume of 100 µl. The liberation of AMC, as a measure of proteolytic activity, was monitored by continuous fluorescence spectroscopy using an excitation wavelength of 380 nm and an emission wavelength of 460 nm.

In vitro translation of human PARP, full-length DCP-1, and full-length drICE. ³⁵S-labeled poly(adenosine diphosphate-ribose) polymerase (PARP) protein was synthesized as described before (22). Full-length DCP-1 and full-length drICE genes were generated by PCR, blunt-ended, and cloned into SmaI-treated Bluescript. The clones with insertion downstream of the T7 promoter of Bluescript were amplified, and the plasmid DNA was isolated and sequenced. The full-length DCP-1 and full-length drICE proteins were synthesized and labeled with [³⁵S]methionine using the TNT T7 coupled reticulocyte lysate system (Promega) following the manufacturer's instructions. For generating a full-length DCP-1 gene PCR fragment, the two primers used were the upstream primer 5'CGCAAAAGATCTCATATGACCGACGAGTGCGTA3' and the downstream primer 5'GTGGTGGTCGACGGATCCCTCTTCCTAGCCAGCCTT3'. The two primers that were used for generating the full-length drICE gene PCR fragment were the upstream primer 5'GCCAAACATATGGACGCCACTAACAATGGA3' and the downstream primer 5'GCGACATCTAGATCAAACCCGTCCGGCTGGAGC3'.

Digestion of [³⁵S]methionine-labeled human PARP, full-length DCP-1, and full-length drICE with purified DCP-1 or drICE protein. One microliter of each labeled protein was incubated with 1 µg of Ni²⁺ column-purified DCP-1 or drICE in 4 µl of incubation buffer {25 mM HEPES, 5 mM EDTA, 2 mM DTT, 0.1% 3-[(3-cholamidopropyl)-dimethylammonio]-1-propanesulfonate (CHAPS), pH 7.5}. The reaction mixture was incubated at 37°C for 30 min and then subjected to sodium dodecyl sulfate-10% polyacrylamide gel electrophoresis (SDS-10% PAGE) under reducing conditions. The gels were dried and exposed to X-ray films for visualization.

Western blot analysis for cleavage of Drosophila lamin Dm₀ and human lamins by DCP-1 and drICE. Drosophila SL2 cells were homogenized, and the nuclear fraction was collected by centrifugation. The nuclei were washed with phosphate-buffered saline three times and resuspended in the reaction buffer (25 mM HEPES, 5 mM EDTA, 2 mM DTT, 1% CHAPS, pH 7.5) with the ratio of 1 volume of packed nuclei to 3 volumes of reaction buffer. Fifteen microliters of this nuclear suspension was incubated with 1 µg of purified DCP-1 or drICE protein at 37°C for 1 h. Nuclei from Hela cells were isolated and treated with DCP-1 or drICE protein in the same manner. After incubation, the reaction mixtures were solubilized with 1% SDS and the proteins were separated by SDS-PAGE under reducing conditions. The separated proteins in the gel were transferred onto polyvinylidene difluoride membranes (Bio-Rad) and probed with specific monoclonal antibodies. The monoclonal antibody against the head region of Drosophila lamin Dm₀ (mAb84.12) (25) was kindly provided by P. Fisher. The monoclonal anti-lamin B antibody was purchased from Calbiochem. The monoclonal antibody against the N-terminal portions of lamins A and C was a kind gift from F. McKeon (16). The bands on the membrane recognized by the monoclonal antibodies were visualized by horseradish peroxidase-labeled rabbit anti-mouse immunoglobulin G antibody and the ECL (GIBCO/BRL) system.

DNA fragmentation assays in the cell-free systems. Cell homogenates were made from Hela cells, Schneider cells (SL2 cells) and Sf9 cells as described before (22). Active DCP-1 or drICE protein was incubated with these homogenates at 37°C for 3 h. After incubation, the DNA in these mixtures was isolated and analyzed with a 1% agarose gel.

Construction of eye expression vectors and generation of transgenic flies. The full-length dcp-1, the full-length drICE, a truncated version of dcp-1, and a truncated version of drICE were generated by PCR and cloned into the eye expression vector pGMR (11). For generating the full-length dcp-1, the 5' end primer used for PCR was GGCGGCAGATCTCAAGAACTTAAGCAAGAA. For the truncated version of dcp-1, the sequence encoding the N-terminal 33 amino acids of DCP-1 was removed by using 5'-end-specific primer GGGAAAAGATCTAACAAAATGGCCAAGGGCTGTACGCCGGAG. The 3' end primers for both dcp-1 constructs were the same: GTGGTGAGATCTCTCTTCCTAGCCAGCCTT. For generating the full-length drICE, the 5' end PCR primer was GAGACCAGATCTCACAAAATGGACGCCACTAACAAT. To generate the truncated form of drICE, the sequence encoding the N-terminal 28 amino acids of drICE was removed by using the 5' primer GAGACCAGATCTCACAAAATGGCCCTGGGCTCCGTGGGA. The 3' end primers for both drICE constructs were the same: GGCCTTAGATCTTCAAACCCGTCCGGCTGG. Each primer contained a BglII site for cloning. A Kozak consensus sequence was added upstream of the initial methionine codon in 5' end primers. All four PCR products were treated with BglII and ligated into BglII- and phosphatase-treated pGMR vector. Clones with correct insertions and sequences were used for injections. Each pGMR construct plasmid DNA was purified by Qiagen columns and mixed with the injection helper DNA p25.7wc2-3 in the ratio of 5:2. This DNA mixture (0.7 µg/µl) was microinjected into yw embryos according to standard procedures.

Drosophila stocks and genetic crosses. GMR-rpr (33) and GMR-hid (10) flies were generated in the Steller laboratory by K. White and J. Agapite, respectively. GMR-grim (4) flies were provided by J. Abrams. GMR-p35 (11) flies were provided by B. Hay. Fly culture and crosses were carried out at 25°C by standard procedures.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

DCP-1 has a substrate specificity that is very similar to those of human caspase 3 and C. elegans cell death protein CED-3. To determine the precise protease specificity of DCP-1, a six-His-tagged, truncated version of DCP-1 with its N-terminal 33 amino acids removed was expressed in E. coli and purified by Ni²⁺ columns as described before (22). The substrate specificity for DCP-1 was determined using a synthetic, positional scanning combinatorial substrate library which has been used previously to define the subsite preferences for the human caspase family and the C. elegans caspase, CED-3 (28) (Fig. 1). DCP-1 has a substrate specificity that is nearly indistinguishable from those of other death effector caspases, including caspase 3 and CED-3. It has a nearly absolute requirement for Asp in P4 and a strong preference for Glu in P3. DCP-1 is tolerant of several substitutions in P2, making it promiscuous at this subsite. The preferred recognition motif for DCP-1, DEVD, is the same as those of other group II caspases (caspases 2, 3, and 7 and CED-3) and is consistent with an important role of DCP-1 as an effector caspase during apoptosis.

View larger version (31K): [in this window] [in a new window]   FIG. 1.   Substrate specificity of DCP-1. The substrate specificity for recombinant DCP-1 was determined using an Ac-[P4]-[P3]-[P2]-Asp-AMC positional scanning synthetic combinatorial library. The y axis represents the rate of proteolytic release of AMC for each of the indicated positionally defined amino acids. DCP-1 is a group II (DEVD-selective) caspase and has a substrate specificity profile that is remarkably similar to those of human caspase 3 and C. elegans CED-3. Data from the latter two enzymes are derived and reproduced, with permission, from reference 28.

drICE and DCP-1 have similar yet different protease specificities. A truncated version of drICE, with its N-terminal 28 amino acids removed and a six-His tag at its C terminus, was also expressed in E. coli and purified by Ni²⁺ columns. In the original report on drICE, the authors made a truncation by removing the N-terminal 80 amino acids (7). Based on the conserved cleavage site for removal of the prodomains, the amino acid sequence ²⁵DHTDA²⁹ in the drICE protein matches the conserved cleavage site sequence DXXDA (where X can be any amino acid), whereas the sequence surrounding Asp⁸⁰, ⁷⁷MVTDR⁸¹, does not seem to match any conserved sequence. Therefore, we made the truncation by removing the N-terminal 28 amino acids as the prodomain. The purified drICE and DCP-1 were incubated with in vitro-translated [³⁵S]methionine-labeled human PARP. The cleaved products were analyzed by SDS-10% PAGE. As shown in Fig. 2A, both DCP-1 and drICE cleaved PARP and the cleaved fragments migrate the same distance in the SDS-PAGE gel, indicating that they both recognize the same cleavage site in the PARP protein.

View larger version (59K): [in this window] [in a new window]   FIG. 2.   A comparison of protease specificities between DCP-1 and drICE. (A) Both DCP-1 and drICE cleave human PARP with similar patterns. 35S-labeled human PARP was incubated with 1 µg of purified DCP-1 or drICE at 37°C for 30 min. The cleavage of PARP was analyzed by SDS-PAGE under reducing conditions. (B) DCP-1 and drICE cleave Drosophila nuclear lamin Dm0 with similar patterns. SL2 cell nuclear fractions were incubated with purified DCP-1 or drICE. The treated nuclei were solubilized with SDS, and the proteins were separated by SDS-PAGE under reducing conditions. The lamin Dm0 molecules and the cleaved fragments were probed by an N-terminal-specific monoclonal antibody and were visualized by standard Western blotting. (C) DCP-1, not drICE, cleaves human lamins. Hela cell nuclear fractions were incubated with DCP-1 or drICE. The treated nuclei were solubilized with SDS, and the proteins were separated by SDS-PAGE. The lamin cleavage was analyzed by Western blotting using monoclonal antibodies specific to human lamin B and lamins A and C.

Neither DCP-1 nor drICE can induce DNA fragmentation in a Drosophila cell-free system. Chromosomal DNA fragmentation is commonly observed during apoptosis, and the activation of specific nucleases that degrade chromosomal DNA is mediated by caspase activity (5, 15). We have previously reported that DCP-1 can induce DNA fragmentation in a Hela cell-free system (22). We show now that DCP-1 can also induce DNA fragmentation in Sf9 cells, a lepidopteran cell line. On the other hand, drICE failed to do so in either system (Fig. 3). This provides additional evidence that these two caspases are biochemically distinct. Surprisingly, neither DCP-1 nor drICE was able to induce DNA fragmentation in the Drosophila SL2 cell-free system (Fig. 3). Likewise, both caspases failed to trigger DNA fragmentation in a cell-free system made from Drosophila embryos (data not shown). Therefore, it is possible that DCP-1 and drICE alone are unable to activate the nuclease responsible for DNA fragmentation in Drosophila cells. When DCP-1 treated Sf9 cell or Hela cell cytosol was incubated with purified SL2 cell nuclei, we observed the classical DNA fragmentation (data not shown). Treatment of SL2 cells with cycloheximide induced apoptosis with typical DNA fragmentation (data not shown). These observations suggest that a caspase(s) in Drosophila other than DCP-1 and drICE may be responsible for DNA fragmentation. Another possibility is that there is a very potent inhibitor of DCP-1 in Drosophila cells, whereas the lysates made from Hela and Sf9 cells may lack such an inhibitor.

View larger version (109K): [in this window] [in a new window]   FIG. 3.   Neither DCP-1 nor drICE induces DNA fragmentation in Drosophila SL2 cells. DCP-1, not drICE, induces DNA fragmentation in Hela and Sf9 cell-free systems. Cell homogenates made from Hela cells, Sf9 cells, and SL2 cells were incubated with DCP-1 or drICE. After incubation, DNA from each reaction mixture was isolated and analyzed with 1% agarose gel.

DCP-1 cleaves full-length DCP-1 and full-length drICE in vitro. We investigated whether DCP-1 and drICE can cleave each other in vitro. Full-length DCP-1 and drICE were synthesized by in vitro translation and were radioactively labeled with [³⁵S]methionine. The labeled full-length DCP-1 and full-length drICE were incubated with purified active DCP-1 or drICE. After digestion, the reaction mixtures were analyzed by SDS-PAGE. As shown in Fig. 4, DCP-1 cleaved full-length DCP-1 and full-length drICE. drICE showed weak activity towards full-length DCP-1 but completely failed to cleave full-length drICE. These data suggest that DCP-1 has the potential to autoactivate, whereas drICE cannot activate itself. Therefore, it is possible that DCP-1 is required to activate drICE in vivo.

View larger version (48K): [in this window] [in a new window]   FIG. 4.   DCP-1 cleaves full-length DCP-1 and full-length drICE, but drICE does not cleave full-length drICE. 35S-labeled full-length DCP-1 and full-length drICE were incubated with the purified active form of DCP-1 or drICE at 37°C for 30 min. After incubation the proteins were separated by SDS-10% PAGE under reducing conditions. The gels were dried and exposed to X-ray films for visualization. (A) 35S-labeled full-length DCP-1 was treated with active DCP-1 or active drICE; (B) 35S-labeled full-length drICE was treated with active DCP-1 or active drICE.

Ectopic expression of the truncated form of dcp-1 induces apoptosis in the Drosophila eye. In order to study the genetic interactions between caspases and other Drosophila cell death genes in vivo, we generated transgenic flies that carry either the full-length or truncated dcp-1 and drICE under the control of the GMR promoter. In this way, the associated proteins can be specifically expressed in the developing Drosophila retina. Two different dcp-1 constructs, one with the full length dcp-1 (GMR-fl-dcp-1) and the other encoding a protein with the N-terminal prodomain removed (GMR-N-dcp-1), were microinjected into yw embryos. Fourteen different GMR-N-dcp-1 lines and seven GMR-fl-dcp-1 lines were obtained from these experiments. As shown in Fig. 5, GMR-N-dcp-1 flies showed a small and rough eye phenotype which was dosage dependent. Flies carrying two copies of GMR-N-dcp-1 (Fig. 5B) have a much stronger eye phenotype than flies that carry only one copy of the same transgene (Fig. 5A). The dcp-1-induced eye phenotype was suppressed by crossing these flies to GMR-p35 flies (11) (Fig. 5C). Since p35 has been shown to be a specific inhibitor of caspases (3, 35), these results confirm that the phenotypes observed in the dcp-1 transgenic flies are indeed the result of caspase activity. Flies carrying one copy of GMR-fl-dcp-1 had almost-normal eye morphology. However, with two copies of GMR-fl-dcp-1, a weak but reproducible eye phenotype was obtained (Fig. 6H). This indicates that the full-length DCP-1 may have weak protease activity.

View larger version (102K): [in this window] [in a new window]   FIG. 5.   dcp-1 can induce apoptosis in vivo. Ectopic expression of N-dcp-1 in the Drosophila eye causes partial eye ablation, and this phenotype is completely suppressed by coexpression of a caspase inhibitor, the viral protein p35. (A) Flies carrying one copy of GMR-N-dcp-1 have a small and rough eye phenotype. (B) Flies carrying two copies of GMR-N-dcp-1 have a much stronger eye phenotype. (C) Eye of a fly carrying one copy of GMR-N-dcp-1 and one copy of GMR-p35, showing that the dcp-1-induced eye phenotype was suppressed. (D) Wild-type fly eye. Bar, 100 µm.

N-dcp-1 does not enhance the eye phenotype of GMR-fl-drICE or GMR-fl-dcp-1 flies. Transgenic flies that each carry a full-length drICE transgene (GMR-fl-drICE) or the truncated version of drICE encoding a protein with its N-terminal 28 amino acids removed (GMR-N-drICE) were also generated. Flies carrying either one or two copies of the GMR-fl-drICE or GMR-N-drICE transgene had no detectable abnormalities (Fig. 6A, B, D, and E). We generated 14 GMR-fl-drICE lines and 5 GMR-N-drICE lines, but none of them showed any eye phenotype. Flies carrying a different drICE transgene encoding a protein with its N-terminal 80 amino acids truncated, as described by Fraser and Evan (7) did not show any eye phenotype either (data not shown). One explanation for these observations is that drICE cannot cleave the drICE protein itself. If so, both the truncated and full-length drICE proteins would remain inactive upon overexpression in the eye. Since we have found that active DCP-1 can cleave full-length DCP-1 and drICE in vitro (Fig. 4), we examined whether GMR-N-dcp-1 would enhance the eye phenotype of GMR-N-drICE or GMR-fl-drICE flies. As shown in Fig. 6C and F, the truncated form of dcp-1 did not enhance the eye phenotype of either GMR-N-drICE or GMR-fl-drICE flies (compare to Fig. 5A). One copy of GMR-fl-dcp-1 had essentially no effect on eye morphology (Fig. 6G), but two copies of the same transgene produced a weak but significant eye defect (Fig. 6H). The coexpression of GMR-N-dcp-1 and GMR-fl-dcp-1 in flies did not significantly enhance the eye phenotype either (Fig. 6I; compare to Fig. 5B). These data indicate that results obtained from caspase cleavage in vitro cannot be simply extrapolated to the situation in vivo. It is possible that both caspases do not have full access to each other because of, for example, different subcellular localizations. Also, the protein concentrations in vitro may be much higher than those in vivo, and this situation may be further complicated due to the presence of specific inhibitors in vivo. In any event, the results from our genetic interaction studies illustrate the limitations of in vitro experiments to establish a cascade of caspase function during apoptosis.

View larger version (120K): [in this window] [in a new window]   FIG. 6.   Flies carrying GMR-fl-drICE or GMR-N-drICE do not have obvious eye phenotypes. GMR-N-dcp-1 does not enhance the eye phenotype of GMR-fl-drICE or GMR-fl-dcp-1 flies. Flies carrying one copy (A) or two copies (B) of GMR-N-drICE have normal eyes. Flies carrying one copy of GMR-N-drICE and one copy of GMR-N-dcp-1 (C) have the same eye phenotype as those carrying only one copy of GMR-N-dcp-1. Flies carrying one copy (D) or two copies (E) of GMR-fl-drICE have normal eyes. Flies carrying one copy of GMR-fl-drICE and one copy of GMR-N-dcp-1 (F) have the same eye phenotype as those carrying only one copy of GMR-N-dcp-1. Flies carrying one copy of GMR-fl-dcp-1 show an almost-normal eye phenotype (G). Two copies of GMR-fl-dcp-1 show a clear effect on the eye (H). Flies with one copy of GMR-fl-dcp-1 and one copy of GMR-N-dcp-1 (I) show the additive effect of the two transgenes. Bar, 100 µm.

rpr and grim, but not hid, activate full-length DCP-1 in the Drosophila eye. The proapoptotic proteins encoded by rpr, hid, and grim all require caspase activity to kill cells. We wanted to know whether dcp-1 or drICE or both are involved in the death pathways activated by any of these genes. In particular, we investigated whether coexpression of caspases and these proapoptotic genes could lead to significantly enhanced cell killing. For this purpose, flies carrying GMR-rpr, GMR-hid, and GMR-grim were crossed to GMR-fl-dcp-1 and GMR-fl-drICE flies. Two different GMR-fl-dcp-1 transgenic fly lines were crossed to GMR-rpr46, GMR-hid1M, and GMR-grim flies, with identical results. Likewise, two GMR-fl-drICE transgenic fly lines were crossed to GMR-rpr46, GMR-hid1M, and GMR-grim flies, again with identical results. As shown in Fig. 7, flies carrying one copy of GMR-fl-dcp-1 or GMR-fl-drICE have almost normal eye morphology (Fig. 7B and C). Flies transgenic for GMR-hid1M, GMR-grim, or GMR-rpr46 have a mild but easily detectable eye phenotype (Fig. 7D, G, and J). The coexpression of hid and full-length drICE produced no obvious enhancement of the eye phenotype, but rather an additive effect of the two transgenes (Fig. 7F). Also, the expression of hid together with full-length dcp-1 enhanced the eye phenotype only weakly (Fig. 7E), comparable to what is seen for coexpression of many other proapoptotic gene combinations (data not shown). In stark contrast, expression of either rpr or grim together with GMR-fl-dcp-1 yielded a dramatically enhanced eye phenotype that cannot be simply explained by additive effects (Fig. 7H and K). In our hands, rpr produced a stronger effect than grim. The expression of rpr also enhanced the eye phenotype of GMR-fl-drICE flies (Fig. 7L), whereas grim was not very effective (Fig. 7I). This finding is consistent with the observation that drICE is activated in rpr-transfected S2 cells (8). Among the different cell types of the Drosophila retina, the pigment cells appear to be particularly sensitive to DCP-1. Both the truncated and the full-length DCP-1 caused pigment cell death (Fig. 5A and B and 6H). Judging by the complete loss of eye color, all pigment cells were eliminated in flies that coexpressed DCP-1 with either rpr and grim (Fig. 7H and K). In order to further investigate the specificity of this interaction, we also crossed GMR-fl-dcp-1 flies to a transgenic line with strong hid expression, GMR-hid 10 flies. Again, the eye phenotype observed for this combination was not significantly enhanced. Overall, we found that rpr and grim interact with dcp-1 much more strongly than hid and interact more effectively with dcp-1 than with drICE. Taken together, these observations suggest that dcp-1 is rate limiting for cell killing by rpr and grim, but not hid. Therefore, we propose that rpr and grim function upstream of dcp-1 in vivo.

View larger version (86K): [in this window] [in a new window]   FIG. 7.   rpr and grim, but not hid, enhance the eye phenotype of GMR-fl-dcp-1 flies. The effect of rpr and grim on dcp-1 is much stronger than their effect on drICE. (A) Wild-type fly eye. Flies carrying one copy of GMR-fl-dcp-1 (B) or one copy of GMR-fl-drICE (C) have an almost-normal eye phenotype. (D) Eyes from flies carrying one copy of GMR-hid1M; (G) eyes from flies carrying one copy of GMR-grim; (J) eyes from flies carrying one copy of GMR-rpr46. GMR-hid1M does not enhance the eye phenotype of either GMR-fl-dcp-1 (E) or GMR-fl-drICE flies (F). GMR-grim strongly enhances the eye phenotype of GMR-fl-dcp-1 flies (H), whereas it has little effect on that of GMR-fl-drICE flies (I). GMR-rpr enhances the phenotype of both GMR-fl-dcp-1 (K) and GMR-fl-drICE flies (L). Clearly, rpr interacts with dcp-1 much more strongly than with drICE. Bar, 100 µm.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

A key biochemical event during apoptosis is the activation of caspases. At least in some cases, proteolysis of specific target proteins can explain some of the discrete morphological changes that are commonly observed during apoptosis (24). Both DCP-1 and drICE appear to be effector caspases, since their enzymological and structural properties, including a short prodomain, are shared with mammalian effector caspases. In this report we show that DCP-1 has a specificity almost identical to those of caspase 3 and C. elegans cell death protein CED-3. drICE and DCP-1 have similar yet different enzymatic specificities. DCP-1 can cleave itself efficiently in vitro, but drICE cannot. This may explain why expression of DCP-1 in the Drosophila retina can cause cell death, whereas drICE failed to do so. Finally, we found that rpr and grim, but not hid, dramatically enhanced the eye phenotype of GMR-fl-dcp-1 flies. This finding is consistent with the idea that dcp-1 is a downstream target for cell killing by rpr and grim.

The substrate specificity for DCP-1 determined by the combinatorial library revealed that DCP-1 specifically recognizes the DEVD motif, indicating that, like caspase 3 and CED-3, DCP-1 plays an important role as a death effector during apoptosis by cleaving key protein targets. Like caspase 3, DCP-1 may also function downstream of different pathways activated by distinct death signals. drICE is very similar to DCP-1, both in structure and protease specificity. Both DCP-1 and drICE cleave PARP and Drosophila nuclear lamin Dm₀ with the same patterns. However, only DCP-1, not drICE, was able to cleave human lamins. Although this finding has no direct physiological relevance, it indicates that DCP-1 and drICE have somewhat distinct enzymological properties.

Chromosomal DNA fragmentation is a hallmark of apoptosis. In nonapoptotic cells, the nuclease responsible for DNA fragmentation forms an inactive complex with its inhibitor. During apoptosis, this inhibitor is cleaved by a caspase and the nuclease is released from the complex. The free nuclease then enters the nucleus and causes DNA fragmentation (5, 15). DCP-1 was able to induce DNA fragmentation in a Hela and Sf9 cell-free systems. On the other hand, drICE was unable to induce DNA fragmentation in either cell-free system. Surprisingly, neither DCP-1 nor drICE was able to induce DNA fragmentation in Drosophila cell-free systems, including one made from Drosophila embryos. However, DCP-1 could induce DNA laddering of Drosophila nuclei in the presence of HeLa cell cytosol. This eliminates the possibility that Drosophila nuclei are somehow resistant to DNA fragmentation. Rather, it appears that DCP-1 alone is insufficient to promote DNA fragmentation in Drosophila cells. However, a relevant DNase activity can be induced in Drosophila SL2, since we did observe DNA fragmentation upon treatment with cycloheximide.

In the original report, the N-terminal 80 amino acids of drICE were considered the prodomain and were truncated for generating active drICE (7). While that truncated version of drICE showed protease activity and induced apoptosis in transfected cells, we believe that the DHTDA sequence from amino acids 25 to 29 in drICE protein is the cleavage site for removal of the prodomain, because it matches the DXXDA conserved sequence. It is also very similar to the sequence of the cleavage site in DCP-1, which is DNTDA. Therefore, it appears that both DCP-1 and drICE have short prodomains with only 33 and 28 amino acids, respectively. Although DCP-1 cleaves full-length DCP-1 and full-length drICE in vitro, GMR-N-dcp-1 did not enhance the eye phenotype of GMR-fl-drICE flies. This indicates that DCP-1 did not activate drICE under these conditions in vivo. On the other hand, expression of rpr did enhance the eye phenotype of GMR-fl-drICE flies, demonstrating that this construct was functional. We conclude that dcp-1 activity is insufficient to activate drICE in vivo.

A major difference between DCP-1 and drICE in our experiments was in the proteolytic processing of both proenzymes. At low concentrations, active DCP-1 cleaved full-length DCP-1, whereas active drICE did not cleave full-length drICE. Apparently, autoprocessing of truncated drICE requires very high concentrations. Therefore, it appears that processing of pro-drICE into the active enzyme does not occur readily in vitro. This property may explain why expression of drICE in the Drosophila retina fails to induce apoptosis, whereas dcp-1 is capable of doing so. At this point it is unclear whether DCP-1 and drICE act in a proteolytic cascade. If they do, our results would place drICE downstream from dcp-1 in this cascade.

The eye phenotype of GMR-fl-dcp-1 flies was strongly enhanced by rpr and grim, but not by hid. Among the different cell types in the Drosophila eye, pigment cells appear to be particularly sensitive to DCP-1-mediated cell killing. Flies with one copy of the GMR-N-dcp-1 transgene lacked most pigment cells (Fig. 5A), and two copies of GMR-N-dcp-1 eliminated them completely (Fig. 5B). Likewise, expression of either rpr or grim with fl-dcp-1 completely eliminated all pigment cells (Fig. 7H and K), whereas hid was unable to do so. These results indicate that Reaper and Grim, but not Hid, can lead to DCP-1 activation. Several other observations also indicate that rpr and grim have cell killing properties that are distinct from those of hid. For example, the Ras/MAPK pathway inhibits hid-induced cell death but has no effect on rpr- or grim-induced death (2). In addition, we have isolated mutations in the diap1 gene of Drosophila that enhance rpr- and grim-induced cell killing but suppress hid-induced cell killing (J. Agapite, K. McCall, and H. Steller, unpublished data). The easiest interpretation of all these observations is that rpr and grim kill cells by activating the same (set of) caspases and that hid activates a distinct caspase (Fig. 8). Since it has been recently shown that rpr, hid, and grim induce cell death by inhibiting the antiapoptotic activity of diap1 (9, 31), diap1 must control at least two distinct caspase pathways. According to this model, Reaper and Grim and HID would interact selectively with specific DIAP1-(pro)caspase complexes. The binding of Reaper, Grim, or HID to the relevant DIAP1-(pro)caspase complex is thought to result in caspase activation. This model is consistent with a variety of findings from both invertebrate and vertebrate systems. However, the possibility that rpr and grim may also activate DCP-1 through a DIAP1-independent pathway cannot be ruled out. Although several Drosophila caspases have been described, our results indicate that additional caspases, in particular ones activated by HID, remain to be identified.

View larger version (7K): [in this window] [in a new window]   FIG. 8.   Model of caspase activation and apoptosis in Drosophila. In living cells, caspases are inhibited by forming complexes with DIAP1 protein. In response to various apoptotic stimuli, the binding of Reaper, Grim, and HID to DIAP1-(pro)caspase complexes leads to caspase activation. Reaper and Grim, but not Hid, appear to specifically activate DCP-1, perhaps through directly binding to DIAP1-DCP-1 complexes. In addition, the Ras/MAPK pathway inhibits the function of HID, but not Reaper or Grim. The caspase that is activated by HID, caspase X, remains unknown. Likewise, details of the presumptive proteolytic cascade(s) that may result from the activation of an apical caspase remain to be determined.