DCP-1, a Drosophila Cell Death Protease Essential for Development

See allHide authors and affiliations

Science  24 Jan 1997:
Vol. 275, Issue 5299, pp. 536-540
DOI: 10.1126/science.275.5299.536

This article has a correction. Please see:


Apoptosis, a form of cellular suicide, involves the activation of CED-3-related cysteine proteases (caspases). The regulation of caspases by apoptotic signals and the precise mechanism by which they kill the cell remain unknown. In Drosophila, different death-inducing stimuli induce the expression of the apoptotic activator reaper. Cell killing by reaper and two genetically linked apoptotic activators, hid and grim, requires caspase activity. A Drosophila caspase, named Drosophila caspase-1 (DCP-1), was identified and found to be structurally and biochemically similar to Caenorhabditis elegans CED-3. Loss of zygotic DCP-1 function in Drosophila caused larval lethality and melanotic tumors, showing that this gene is essential for normal development.

Programmed cell death, or apoptosis, is of fundamental importance for the elimination of cells that are no longer needed in an organism (1). During the past few years, there has been growing support for the idea that the basic molecular mechanism underlying apoptosis has been conserved during evolution among animals as diverse as nematodes, insects, and mammals (2). A central step in this cell suicide pathway is the activation of an unusual class of cysteine proteases, named caspases (3), that includes mammalian interleukin-1β-converting enzyme (ICE) and the ced-3 gene of nematodes (4). Caspases are synthesized as inactive zymogens that need to be processed to form active heterodimeric enzymes (4). However, the precise mechanism of caspase activation in response to apoptotic stimuli remains unknown. Likewise, with the exception of the Caenorhabditis elegans caspase CED-3, it is not clear what precise role any other caspase has in apoptosis.

The availability of many sophisticated genetic and molecular techniques makes Drosophila ideally suited for studying the questions of caspase activation and function. In Drosophila, like in mammalian systems, the onset of apoptosis is regulated by a number of distinct death-inducing stimuli (5). Genetic studies have led to the identification of three apoptotic activators, reaper (6), head involution defective (hid) (7), and grim (8), that appear to act as mediators between different signaling pathways and the cell death program. The deletion of all three genes blocks apoptosis in Drosophila (6), and overexpression of any one of them is sufficient to kill cells that would normally live (7, 8, 9). The products of these genes appear to activate one or more caspases, because cell killing by reaper, hid, and grim is blocked by the baculovirus protein p35 (7, 8, 9), a specific inhibitor of caspases (10).

To gain further insight into the function and control of caspase activity, we isolated Drosophila caspase-like sequences. Degenerate oligonucleotides corresponding to two highly conserved regions flanking the active site of the enzyme were designed and used for a polymerase chain reaction (PCR) with a Drosophila 4- to 8-hour embryo cDNA library as the template (11). We obtained several PCR products of the expected size that were subcloned and sequenced (11). One clone was highly homologous to the region containing the caspase active site, including the highly conserved QACRG (12) pentapeptide. This clone was used to isolate full-length cDNA clones and to deduce the entire amino acid sequence of this putative caspase (11). The predicted open reading frame of the full-length cDNA encodes a protein of 323 amino acids (Fig. 1A). The DNA sequence surrounding the first ATG (CAAGEmbedded ImageGACC) is in good agreement with the consensus sequence for translation initiation in Drosophila (13). The corresponding protein was named Drosophila caspase-1 (DCP-1). In comparison with other caspase family members (4, 14, 15, 16), DCP-1 is more homologous to CPP-32 and MCH-2α than to ICE. It shares 37% sequence identity with both CPP-32 and MCH-2α, 29% identity with NEDD-2 (ICH-1), 28% with CED-3, and 25% with human ICE. This sequence similarity suggests that DCP-1 may be a member of the ced-3-CPP-32 subfamily of caspases.

Fig. 1.

Predicted amino acid sequence of DCP-1 and its homology to other caspase family members (12). (A) Predicted amino acid sequence of DCP-1. The putative prodomain contains the NH2-terminal 33 amino acids that appear to be removed by proteolytic cleavage at DNTDA (12), between Asp33 and Ala34. The cleavage site that generates the large and small subunits is at TETDG, between Asp215 and Gly216. Another possible cleavage site, DRLDG, is between Asp202 and Gly203. The active site QACQG is in bold face, and putative cleavage sites are underlined. (B) Sequence alignment of the most conserved regions among several caspases. The active site pentapeptide is boxed. The cleavage site that generates the large (p20) and small (p10) subunit in DCP-1 is between Asp215 and Gly216. Dotted lines indicate gaps in the sequence to allow optimal alignment. MACH (FLICE) shares identical sequence with Mch5 in the listed regions, therefore it is not shown here. The crystal structure of ICE indicates that Cys285, His237, and Gly238 (□) are involved in the catalysis, and Arg179, Gln283, Arg341, and Ser347 (ˆ) are involved in the recognition of the P1 Asp. All these residues are conserved among these proteins except one close substitution of Ser to Thr in Mch5.

Caspases are synthesized as inactive proenzymes that are proteolytically processed to form the active heterodimer consisting of a p10 (10 kD) and a p20 (20 kD) subunit (4). The consensus sequence for proteolysis of many ced-3-like caspases is (D/E)XXD-Y (12), where X can be any amino acid and Y is a small amino acid, such as Ala, Gly, or Ser. Cleavage occurs between Asp and Y. A good match to this consensus, DNTD-A, is found in the expected region of DCP-1, indicating that cleavage may occur between Asp33 and Ala34. This prediction was supported by the biochemical properties of a truncated version of DCP-1 (discussed below). Therefore, like CPP-32 and MCH-2α, DCP-1 appears to have a short prodomain of only 33 amino acids. On the basis of the x-ray crystal structure of ICE (17), it is thought that Cys285, His237, and Gly238 in this molecule are involved in the catalysis of the peptide cleavage, whereas Arg179, Gln283, Arg341, and Ser347 are involved in the recognition of the Asp at the P1 position. All the corresponding amino acid residues are conserved in DCP-1 (Fig. 1B) and in other family members. DCP-1 contains a slightly modified pentapeptide, QACQG, instead of the more common QACRG, which is also found in mammalian Mch4 and Mch5 (also called MACH or FLICE) (16).

To show that DCP-1 protein has protease activity, we expressed two different versions in Escherichia coli and tested the biochemical activity of the recombinant proteins on known substrates for caspases (18). Full-length DCP-1 gave no or very little activity, whereas a truncated protein lacking the putative prodomain had very strong protease activity. This form of DCP-1 cleaved both poly(adenosine diphosphate-ribose) polymerase (PARP) (Fig. 2A) and p35 into fragments of the predicted size. Direct comparison with fragments generated by CED-3 cleavage indicated that both proteases cleave at identical sites (Fig. 2A). DCP-1 protease activity was completely abolished by iodoacetamide; thus cysteine is critical for enzyme activity. The CPP-32-specific inhibitor Ac-DEVD-CHO (14) completely inhibited PARP cleavage by DCP-1, whereas the ICE-specific inhibitor Ac-YVAD-CHO was ineffective (Fig. 2A). Therefore, DCP-1 is biochemically more closely related to CED-3 and CPP-32 than to ICE. DCP-1 also cleaved p35 in a manner identical to that of CED-3. Finally, the composition of the autoprocessed mature DCP-1 enzyme was determined. After purification of the truncated form of DCP-1, two bands of about 22 kD (corresponding to p20) and 13 kD (corresponding to p10) were detected with SDS-polyacrylamide gel electrophoresis (PAGE). Microsequencing analysis of the small subunit (19) demonstrated that the cleavage site producing the two subunits was, as expected, between Asp215 and Gly216 (Fig. 1B). These results show that DCP-1 is a cysteine protease and has biochemical properties that are similar to that of the C. elegans cell death protease CED-3.

Fig. 2.

DCP-1 has caspase activity and induces DNA fragmentation in HeLa cell nuclei. (A) DCP-1 cleaves PARP, and this activity can be inhibited by iodoacetamide (Iodoac.) and Ac-DEVD-CHO (ADC), but not by the ICE inhibitor Ac-YVAD-CHO (AYC). 35S-labeled human PARP (hPARP) was used as the substrate for protease activity analysis. For inhibition, 10 mM iodoacetamide, Ac-DEVD-CHO, or Ac-YVAD-CHO, as indicated, was used as an inhibitor. They were mixed and incubated with the enzyme at 37°C for 10 min. Then 35S-labeled hPARP was added and incubated at 37°C for 30 min. CED-3 protein was used as control and PARP cleavage was analyzed by 10% SDS-PAGE. (B) DCP-1 induces DNA fragmentation in HeLa cell nuclei. Methods are described in (22).

To determine if DCP-1 can induce cell death, we expressed the gene in several mammalian cell lines (20). Cells expressing DCP-1 displayed the typical apoptotic morphology, such as condensed, rounded cell morphology and severe membrane blebbing. These observations indicate that expression of DCP-1 is sufficient to induce apoptosis. However, because expression of several proteases, including proteinase K, trypsin, and chymotrypsin, can induce apoptosis (21), DCP-1 may kill by causing cellular damage that subsequently triggers an apoptotic response. To eliminate this possibility, we used a cell-free apoptosis system that permits the investigation of apoptosis-like nuclear events (22). In this system, cellular structures have been destroyed and therefore are no longer capable of sensing physiological signals. Purified DCP-1 or proteinase K as a control was added to such a cell-free preparation from HeLa cells (22). Upon treatment with DCP-1, the chromosomal DNA was fragmented and displayed the characteristic apoptotic DNA ladder (Fig. 2B). In contrast, proteinase K failed to induce DNA fragmentation in this system (Fig. 2B). Thus, DCP-1 acts far downstream to induce apoptosis, either by directly cleaving apoptotic targets or by activating other procaspases that may be present in the cell-free system. In either case, the fact that a Drosophila protein, DCP-1, can engage at least part of the apoptotic program in mammalian cells suggests that its targets have been conserved in evolution.

A Drosophila cell death caspase should be expressed in all cells that have the ability to undergo apoptosis. We determined the distribution of dcp-1 mRNA during Drosophila embryogenesis by in situ hybridization (23). Preblastoderm embryos, a stage before the onset of zygotic transcription, contained large and uniform amounts of dcp-1 RNA (Fig. 3A). Therefore, dcp-1 is maternally expressed. In later stages, dcp-1 transcripts continued to be present throughout the embryo (Fig. 3B). This uniform pattern of RNA distribution is consistent with a role of dcp-1 as an apoptotic effector. Toward the end of embryogenesis, dcp-1 expression became more restricted (Fig. 3C). The reduction of dcp-1 transcript correlated well with the increased resistance of late embryos to the induction of apoptosis by x-ray irradiation and ectopic expression of reaper (9).

Fig. 3.

dcp-1 mRNA expression in Drosophila embryos. Whole mount wild-type embryos were hybridized with a dcp-1 digoxigenin-labeled probe (23). (A) Stage 4 embryo (31). Uniform mRNA distribution was observed in all embryos before cell formation and the onset of zygotic transcription (stages 1 to 5), demonstrating that dcp-1 is maternally expressed. (B) Stage 10 embryo (31). dcp-1 transcripts were found in essentially all cells during germband extension. The weakly stained central region of the embryo contains mainly yolk. (C) Stage 16 embryo (31). The expression of dcp-1 became nonuniform in advanced stages of embryogenesis. Although low levels of transcript appeared to be still present throughout the embryo, some regions of the embryo, including the head, some cells within the central nervous system, the developing gonads, and a portion of the gut, were strongly labeled. (D) Lateral view of the central nervous system of a stage 17 embryo (31). Strong expression of dcp-1 was seen in cells along the midline of the central nervous system. Scale bars are 50 μm.

To begin investigating the function of DCP-1, we obtained loss-of-function mutations in the gene. dcp-1 was mapped by in situ hybridization to the cytological position 59F on the right arm of chromosome II, and chromosomal deletions for this locus were identified (24) (Fig. 4). In addition, two preexisting lethal P element insertions, l(2)01862 and l(2)02132, were found to be inserted at different positions in the first exon of dcp-1 (Fig. 4) (24). These P element mutants behaved genetically as null alleles and will be referred to as the dcp-11862 and dcp-12132 alleles. Viable revertants of these alleles were generated and were associated with P element excisions, demonstrating that the phenotypes described below are indeed caused by the transposon insertions into the dcp-1 gene (25). To eliminate possible contributions of other mutations in the genetic background of the dcp-1 P element alleles, we conducted phenotypic analyses in trans to a deletion for dcp-1 (26).

Fig. 4.

Map of the 59F region. The hatched bar represents the wild-type chromosome, and the cytological divisions are indicated. The deletion strains used for in situ analysis are drawn above the hatched bar, with the dashed lines representing the deleted regions. They are labeled with their reported breakpoints (24). Genetic analysis with a number of lethals that map to the region revealed that Df(2R)bw5 and Df(2R)G10-BR27 may overlap, because both deletions fail to complement at least one lethal complementation group. The approximate position of the dcp-1 gene is indicated below the hatched bar; the orientation is drawn arbitrarily as it has not yet been determined. The genomic DNA 3′ of the P elements has been partially sequenced, revealing a 430-base pair intron. The P elements are inserted in exonic sequences 182 and 291 base pairs downstream of the start of the cDNA and 179 and 70 base pairs upstream of the initiation codon.

Because ced-3 mutants of C. elegans are defective in programmed cell death, we investigated the pattern of apoptosis in embryos lacking zygotic dcp-1 function. Using TUNEL labeling and ENGRAILED antibody staining, we detected no significant abnormalities in the pattern of cell death (27). These data indicate that zygotic DCP-1 function is not required for most embryonic cell deaths in Drosophila, perhaps because of the existence of additional caspases (28). However, because DCP-1 has significant maternal expression, it is also possible that sufficient DCP-1 protein is present during embryogenesis for cell death to occur.

Both alleles of dcp-1 caused lethality during larval stages, showing that dcp-1 is an essential gene. This lethality was associated with the transposon insertions, because it was seen in transheterozygotes of the two different P insertions and because it could be reverted by excision of the P elements (25). Although most of the dcp-1 homozygotes died before the third instar larval stage, some of the dcp-1 homozygotes reached that stage and displayed several abnormalities. Larvae mutant for dcp-1 had an overall normal central nervous system but lacked imaginal discs and gonads. In addition, they had fragile trachea. However, the most prominent phenotype of these larvae was the presence of melanotic tumors, located in various parts of the body (Fig. 5). Melanotic tumors can result from either the overproliferation of blood cells or from an immune response toward abnormal cells and tissues in the larva (29). In dcp-1 mutants, no evidence for hyperplasia of the lymph glands or overproliferation of blood cells was found. This suggests an immune reaction toward abnormal tissues or cells, possibly resulting from a defect in cell death. According to this model, cells that would have normally been eliminated by apoptosis persist in DCP-1-deficient animals but are eventually recognized by the fly's immune system. Although mammalian caspases have not yet been implicated in tumor suppression, this scenario would be consistent with the known role of apoptosis in preventing tumorigenesis in mammals (30). Alternatively, the lack of dcp-1 function may lead to the aberrant differentiation of certain cells to a state where they become recognized as “nonself.” In this case, dcp-1 would have a novel function that is independent and distinct from a role in cell death. This possibility is supported by the tracheal and imaginal disc phenotypes that are not easily explained by defects in programmed cell death.

Fig. 5.

Melanotic tumor phenotype of dcp-1 mutants. (A) A wild-type (Canton S) larva at the wandering third instar stage (∼5 days). (B and C) dcp-11862/Df(2R)bwDRa transheterozygous larvae (∼11 to 12 days). The conspicuous dark masses, indicated by arrows, are the melanotic tumors that occur in various locations in the larvae. Scale bars in (A), (B), and (C) are 500 μm. (D and E) Higher magnification view of two melanotic tumors from a dcp-11862/Df(2R)bwDRa larva. Scale bars are 50 μm.

The existence of prominent, fully penetrant phenotypes in dcp-1 mutants should facilitate future investigations on whether caspases may indeed have important and currently unknown developmental functions. Because Drosophila contains multiple caspases (28), it will also be possible to investigate whether these proteases function in a cascade or in parallel pathways or have redundant functions. Additionally, it should now be possible to identify functionally relevant targets of these proteases by the use of genetic as well as biochemical means. Finally, the identification of Drosophila caspases should help elucidate the mechanism by which they are regulated in response to apoptotic activators, such as reaper, hid, and grim. Because these genes provide a crucial link between different death-inducing signaling pathways and caspase activation, it should eventually be possible to deduce the precise mechanism by which defined apoptotic stimuli activate the cell death program.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
View Abstract

Navigate This Article