Proteolytic processing is required for the activation of most caspases. However, recent reports have suggested that the activation of the mammalian initiator caspase caspase-9 occurs during dimerization rather than after processing. Previously, we reported that, in normal living Drosophila S2 cells, the initiator caspase Dronc is continuously processed to a 40 kDa form we called Pr1 and that, during apoptosis, a second processed form of 37 kDa is also observed, which we called Pr2. In this study, we determined that Dronc Pr1 is the result of Dronc autoprocessing at amino acid E352, whereas Pr2 results from Drice cleaving full-length Dronc at amino acid D135. By using purified recombinant proteins and expressing Dronc cleavage mutants in S2 cells, we determined that autoprocessing at E352 is crucial for Dronc caspase activity, whereas Drice cleavage at D135 has little effect on Dronc activity. Suppression of the oligomerizing factor Dark by RNA interference revealed that Dark is required for Dronc autoprocessing at E352, whereas RNA interference of the effector caspase Drice revealed that Drice is also required for apoptosis in S2 cells. These results provide the first details of the mechanisms regulating initiator caspase activation in an invertebrate organism.
Apoptosis is a tightly regulated program of cell death that is used by multicellular organisms to eliminate unneeded or damaged cells. It is crucial during development as well as in the maintenance of tissue homeostasis and regulation of certain diseases. An important group of apoptotic proteins are the caspases. Caspases are cysteine proteases and are synthesized as zymogens that must usually be proteolytically cleaved to become activated (Shi, 2002). Upon reception of a death signal, upstream initiator caspases are first activated and these in turn cleave and activate downstream effector caspases, leading to apoptosis.
In mammalian cells, the initiator caspase caspase-9 associates with Apaf-1 and undergoes autoprocessing during apoptosis (Shiozaki et al., 2002; Srinivasula et al., 1998). Because proteolytic processing of caspases is a common mechanism of activation, it was initially assumed that caspase-9 activation required processing. However, recent studies reported that dimerization and cellular cofactors are sufficient to induce high levels of caspase-9 activity, whereas processing of caspase-9 only slightly enhances its activity (Rodriguez and Lazebnik, 1999; Stennicke et al., 1999).
Seven caspases have been identified from Drosophila melanogaster (Kumar and Doumanis, 2000). Of these Dronc, Dredd and Strica are predicted to be initiator caspases, whereas Drice, Dcp-1, Damm and Decay are predicted to be effector caspases. Consistent with other initiator caspases, Dronc possess a long prodomain that includes a caspase recruitment domain (CARD) and Dronc can cleave and activate various effector caspases (Chen and Wang, 2002; Dorstyn et al., 1999a; Hawkins et al., 2000). However, whereas all other caspases cleave only after aspartate residues, Dronc has been shown to cleave substrates after either aspartate or glutamate residues (Hawkins et al., 2000). This altered substrate specificity is accompanied by a unique active site in Dronc, where the active site cysteine is preceded by the amino acids PF, whereas most other known caspases contain amino acids QA at this position (Dorstyn et al., 1999a). Dronc has also been shown to associate with the Drosophila CED-4/Apaf-1 homolog Dark (Quinn et al., 2000). This interaction has been shown to be important for apoptosis and, in Drosophila S2 cells, the absence of Dark results in the accumulation of full-length Dronc protein and inhibition of apoptosis from various stimuli (Igaki et al., 2002; Muro et al., 2002; Quinn et al., 2000). In addition, dark mutations can rescue cells from death induced by reduced expression of the Drosophila inhibitor of apoptosis DIAP1 in developing Drosophila embryos (Rodriguez et al., 2002).
In previous work, we showed that, in normal living S2 cells, Dronc is continuously processed to a 40 kDa form we called Pr1 (Muro et al., 2002). In addition, we observed that, upon stimulation of apoptosis, Pr1 levels initially increased and then steadily decreased as a second 37 kDa processed form of Dronc (which we called Pr2) accumulated. We were therefore interested in investigating the mechanism of Dronc processing and its effect on Dronc activity. In this study, we show that Dronc autoprocessing at amino acid E352 produces Pr1, whereas Drice cleavage of full-length Dronc at amino acid D135 results in Pr2. We further show both in vitro and in vivo that autoprocessing of Dronc at E352 is required for Dronc caspase activity. Using RNA interference (RNAi) in S2 cells, we also determined that Dark is required for Dronc autoprocessing at E352 and that the effector caspase Drice is required for the execution of apoptosis. These results indicate that, similar to caspase-9, Dronc requires an oligomerizing factor (Apaf-1 and Dark, respectively) for activation but, in contrast to caspase-9, Dronc also requires autoprocessing for complete activation. In addition these results support an earlier hypothesis that, in S2 cells, Drice might be the only effector caspase required for apoptosis (Fraser et al., 1997). Taken together, these results provide further evidence that apoptosis is regulated somewhat differently in flies than in mammals.
Materials and Methods
Drosophila S2 cells (Invitrogen) were maintained in Schneider's medium (Invitrogen) supplemented with 10% fetal bovine serum (FBS) (Invitrogen).
Dronc mutants were generated using the QuikChange Site-Directed Mutagenesis kit (Stratagene) and plasmids were sequenced to verify mutations and to ensure the absence of inadvertent mutations. All residues mutated were converted to alanine.
pET23b plasmids expressing C-terminally His-tagged Dronc (obtained from B. Hay, California Institute of Technology, Pasadena, CA) and Drice (Drice-81, lacking the first 81 amino acids) were transformed into Escherichia coli strain BL21pLysS(DE)3. Overnight cultures were grown at 37°C and then diluted 1:20 in LB broth and grown at 25°C to an optical density at 600 nm of 0.4, at which time they were induced with 0.1 M IPTG for 1 hour. After harvesting, bacteria were sonicated in lysis buffer A (200 mM Tris HCl pH 8.0, 0.4 M ammonium sulfate, 10 mM MgCl2, 10% glycerol) and lysates were then centrifuged at 12,000 g for 10 minutes at 4°C. The soluble fraction was then used to purify recombinant proteins using Talon Metal Affinity Resin (Clontech) according to the manufacturer's instructions.
S2 cells were treated with double-stranded RNA (dsRNA) as previously described (Muro et al., 2002).
2.5×106 S2 cells per well were plated for 2 hours in TC-100 medium (Invitrogen) with 10% FBS. After 2 hours, medium was removed and replaced with 0.8 ml TC-100 medium with no FBS. Cells were transfected with 3 μg of each plasmid per well using Cellfectin (Invitrogen) according to the manufacturer's instructions. Dronc and chloramphenicol acetyl transferase (CAT) were expressed from an hsp70 promoter in the pHsp70PLVI+ vector (Clem and Miller, 1994). After 5.5 hours, the reaction mixture was replaced with TC-100 plus 10% FBS. For the caspase-inhibitor experiments, cells were treated with 100 μM ZVAD-FMK (Enzyme Systems Products) during the transfection and after removal of the transfection mixture. Cells treated with Dark dsRNA were transfected 20 hours after dsRNA addition and an additional dose of dsRNA was included in the DNA-Cellfectin mixture.
Radiolabeled Dronc was made using the TNT T7/SP6 Coupled Reticulocyte Lysate System (Promega). 4 μl reticulocyte lysate were then incubated with 2.5 μg recombinant Dronc, 500 ng recombinant Drice or both, and the reaction volume was brought to 50 μl with caspase buffer A (25 mM Tris Cl, pH 8.0, 50 mM NaCl, 10 mM dithiothreitol). Reactions were incubated for 1 hour at 30°C and subjected to SDS-PAGE. After electrophoresis, gels were fixed in 30% methanol plus 10% acetic acid for 2 hours and then soaked in 16% salicylic acid for 30 minutes. After drying, gels were exposed to film at -80°C.
Caspase activity assay
Recombinant Dronc (0.225 μg μl-1) was incubated in caspase buffer A for 30 minutes at 30°C. 0.2 μM IETD-afc (Enzyme Systems Products) was added and the reactions were incubated at 37°C for 30 minutes before fluorimetric analysis (excitation at 405 nm, emission at 535 nm). Activity was expressed in relative arbitrary fluorescence units. For caspase assays using cell lysate, 1.5×108 S2 cells were suspended in 800 μl lysis buffer [20 mM Hepes KOH, pH 7.5, 50 mM KCl, 1.5 mM MgCl2, 1 mM EDTA, 1 mM EGTA, 1 mM dithiothreitol, 250 mM sucrose, 1 complete mini EDTA-free protease inhibitor tablet (Roche Molecular Biochemicals) per 50 ml lysis buffer] and subjected to 300 strokes with a Dounce homogenizer. Unbroken cells and nuclei were removed by centrifugation at 500 g for 5 minutes. Lysates were snap frozen in liquid nitrogen after adding glycerol to 10% and were stored at -80°C. 50 μl non-apoptotic cell lysate was added to each reaction and total reaction volume was brought to 100 μl using caspase buffer A.
S2 cells were co-transfected with plasmids expressing various tagless Dronc constructs under the control of the baculovirus ie-1 promoter, which were constructed by insertion into plasmid vector pIE1hr/PA (obtained from P. Friesen, University of Wisconsin-Madison, Madison, WI), and a plasmid expressing enhanced green fluorescent protein (eGFP) from the hsp70 promoter (Clarke and Clem, 2002). Fluorescent cells were counted at 48 hours after transfection and viability was determined as a proportion of control transfected cells at the same time. Experiments were done in triplicate and three fields of view were counted for each well. For Drice viability, cells were plated overnight and then were mock treated or treated with drice dsRNA or cat dsRNA for 20 hours before the addition of diap1 dsRNA. The number of viable cells was determined by using a hemocytometer to count non-apoptotic cells at each time point and the untreated cell count at 6 hours was set at 100%.
Immunoblots for endogenous Dronc and DIAP1 were performed as previously described (Muro et al., 2002). For detection of overexpressed Dronc, rabbit anti-Dronc was diluted 1:7000.
Dronc is processed at amino acids E352 and D135
To investigate the mechanism of Dronc processing in S2 cells, we searched for potential caspase cleavage sites that would result in cleavage products corresponding to Pr1 and Pr2. Because Dronc has previously been reported to autoprocess at residue E352 when expressed in E. coli (Hawkins et al., 2000), we mutated E352 and the aspartate residues D113 and D135, both of which are in approximately the correct region for processing between the prodomain and the large subunit, to alanines (Fig. 1A). We next expressed recombinant versions of these Dronc mutants as well as wild-type Dronc in E. coli. Coomassie-blue staining of purified wild-type Dronc showed that Dronc was processed into two main products of approximately 28 kDa and 10 kDa (Fig. 1B). These products appeared to correspond to the fully processed large and small subunits of Dronc and suggest that similar to several other caspases made in E. coli, recombinant Dronc is active and autoprocesses when overexpressed at high levels. The observation that the alanine mutation at residue E352 dramatically reduced Dronc autoprocessing (Fig. 1B) suggested that this residue was important for Dronc activity and confirms a similar observation in an earlier report (Hawkins et al., 2000). Mutations at D113 and D135 had no effect on Dronc autoprocessing (Fig. 1B), indicating that the cleavage between the prodomain and the large subunit observed in E. coli might not occur at an aspartate residue and might be due to non-specific cleavage by a bacterial protease.
As further evidence that Dronc can cleave itself at residue E352, we used active recombinant wild-type Dronc to cleave various radiolabeled Dronc proteins produced by in vitro translation (Fig. 2A). Analysis of wild-type 35S-Dronc showed that in vitro translated Dronc was full length and that, when incubated with active Dronc, a 40 kDa product (similar to Pr1) was observed. In addition, although other Dronc mutants tested were cleaved normally by active Dronc, the E352A mutant was not cleaved and remained full length. These results provide further evidence that Dronc undergoes autoprocessing at E352, which (according to the amino acid sequence) would be predicted to result in a product of approximately 40 kDa, similar to Pr1.
Because autocatalytic cleavage of Dronc did not account for the Pr2 band observed in cells, we hypothesized that another caspase could cleave Dronc to produce Pr2. Because Dronc is an initiator caspase and Pr2 is observed only in dying cells, we predicted that, during apoptosis, Dronc is subsequently processed by a downstream effector caspase. Drice is an effector caspase expressed in S2 cells and becomes processed (active) 6 hours after ultraviolet treatment (Muro et al., 2002), corresponding to the appearance of Dronc Pr2. These observations suggest that, in apoptotic cells, Drice might cleave Dronc to produce Pr2. We thus used active recombinant Drice to cleave various 35S-Dronc proteins. Drice cleavage of wild-type 35S-Dronc resulted in a product of 37 kDa, similar in size to Pr2 (Fig. 2A). In addition, active Drice was able to cleave E352A Dronc to produce Pr2 but was not able to cleave the D113A/D135A Dronc mutant, which remained full length (Fig. 2A,B). Although Drice was able to cleave both the D113A and D135A single mutants, only cleavage of the D113A mutant resulted in the production of Pr2; cleavage of the D135A mutant resulted in a cleavage product larger than either Pr1 or Pr2 (Fig. 2B). These results indicate that Drice normally cleaves Dronc at position D135, but that cleavage at D113A can also occur.
Drice cleavage at D135 would result in two different forms of Dronc depending on the substrate. If Drice cleaved full-length Dronc, a product similar to Pr2 would be produced. However, if Drice cleaved Dronc Pr1, a 28 kDa product would be produced, corresponding to the large subunit of Dronc cleaved at D135 and E352 (Fig. 2C). Having shown that Drice could cleave full-length Dronc, we wanted to determine whether Drice could cleave Pr1, and so active Dronc and active Drice were used together to cleave 35S-Dronc (Fig. 2A,B). Cleavage of wild-type 35S-Dronc by active Dronc and Drice resulted in a cleavage product of 28 kDa, corresponding to fully processed Dronc. When both caspases were used to cleave the E352A 35S-Dronc mutant, only Pr2 was detected and, when both caspases were used to cleave the D113/D135 Dronc mutant, only Pr1 was detected. In addition, when S35-Dronc was mutant at all three sites, all cleavage was eliminated. These results further verify the specificity of Dronc and Drice cleavage, and indicate that Drice is able to cleave both full-length Dronc and Pr1, suggesting that three forms of Dronc probably exist in apoptotic cells (Fig. 2C).
Processing at E352 is required for Dronc activity
Previous caspase-9 studies have suggested that dimerization and cellular cofactors are sufficient to induce activation of caspase-9, whereas processing only slightly enhances its activity (Rodriguez and Lazebnik, 1999; Stennicke et al., 1999). In addition, a non-cleavable mutant of caspase-9 was shown to induce wild-type levels of apoptosis when overexpressed in 293T cells (Stennicke et al., 1999). We therefore wanted to determine the effect of Dronc processing at E352 and D135 on Dronc activity. Using recombinant protein, the activities of wild-type Dronc and various mutants were tested in a fluorimetric assay using the fluorogenic substrate IETD-afc. Whereas mutation of D135 or D113 had no effect on recombinant Dronc activity, mutation of E352 reduced caspase activity to background levels (Fig. 3 and data not shown).
Previously, it has been shown that, although non-cleavable caspase-9 also had little activity in vitro, its activity was near wild type when incubated with cell lysate, suggesting that cellular cofactors stimulate caspase-9 activity (Stennicke et al., 1999). To determine whether cellular cofactors were able to stimulate Dronc activity, various recombinant Dronc proteins were incubated with non-apoptotic S2 cell lysate (Fig. 3). Although wild-type Dronc activity greatly increased, presumably as a result of recombinant Dronc activating caspases found in the lysate, E352A Dronc activity was only slightly enhanced. These results suggest that autoprocessing at E352 is important for Dronc activation and that cellular cofactors are not sufficient to activate Dronc in the absence of autoprocessing.
Dronc processing to Pr1 requires Dark and is sufficient to induce apoptosis
To examine Dronc processing in vivo, we expressed various Dronc constructs in S2 cells. An anti-Dronc western blot of cells transfected with wild-type Dronc showed that overexpressed Dronc was processed into three forms: Pr1, Pr2 and fully processed Dronc (Fig. 4A). Because Drice activity is required to produce both Pr2 and fully processed Dronc, and Dronc is required to activate Drice, the logical conclusion would be that Pr1 is sufficient to activate Drice. Therefore, to examine further the role of Pr1 in apoptosis, we transfected S2 cells with plasmids expressing various cleavage-site mutants of Dronc. In agreement with our in vitro studies, an anti-Dronc western blot indicated that the D113A/D135A Dronc mutant was processed only to Pr1, whereas the E352A mutant remained full length (Fig. 4A). The D135A mutant was also processed, but it was processed to an approximately 32 kDa form, which would be expected if cleavage occurred at E352 and D113. To determine whether these processing events affected Dronc activity, cell viability was determined at 24 hours after transfection and the results indicated that wild-type Dronc and the D135A and D113A/D135 mutants all induced apoptosis with similar efficiency (Fig. 4B). By contrast, when cells were transfected with the E352A mutant, only minimal levels of apoptosis were observed (Fig. 4B).
Using S2 cells, we have previously found that Dark (the Drosophila Apaf-1/CED-4 homolog) was required for Dronc processing to Pr1, suggesting that Dark is involved in the activation of Dronc, possibly by inducing its dimerization of Dronc (Muro et al., 2002). To determine whether Dark was also required for activation of overexpressed Dronc, we treated cells with dark dsRNA before transfection with wild-type Dronc. When Dark protein levels were reduced, transfected wild-type Dronc was no longer processed and was unable to induce apoptosis (Fig. 4A and data not shown). Because, in control cells, wild-type Dronc was processed and ∼70% of transfected cells died, it appears that Dark must be present for Dronc activation even under these overexpression conditions. However, the observation that, even in the presence of Dark, overexpressed E352A Dronc does not induce apoptosis suggests that, although Dark has a role in Dronc activation, Dronc must still autoprocess at E352 to become active.
Drice is required for apoptosis in S2 cells
Previous studies showed that, when the effector caspase Drice was immunodepleted from S2 cell lysate, most of the measurable apoptotic activity was removed (Fraser et al., 1997), suggesting that Drice was the main, if not the only, effector caspase expressed in S2 cells. To examine this hypothesis further, S2 cells were treated with drice dsRNA 20 hours before inducing apoptosis with diap1 RNAi. 24 hours after the addition of diap1 dsRNA, cell viability was determined and silencing of Drice was found to significantly inhibit apoptosis induced by loss of DIAP1 (Fig. 5A). Because Drice cleaves Dronc at D135, we wanted to examine the processing of endogenous Dronc in the absence of Drice after the addition of diap1 dsRNA. Immunoblots of cells treated with diap1 dsRNA and pretreated with drice dsRNA showed that DIAP1 protein was efficiently removed (Fig. 5B) and that endogenous Dronc was processed only to Pr1 (Fig. 5C). These results verify that endogenous Drice cleaves endogenous Dronc and support the hypothesis that Drice is the only effector caspase active in S2 cells.
Dimerization by Apaf-1 is sufficient (along with the activities of other cellular cofactors) to induce activation of mammalian caspase-9 and, although caspase-9 undergoes autoprocessing upon dimerization, caspase-9 activity is only slightly decreased if autoprocessing is blocked (Rodriguez and Lazebnik, 1999; Stennicke et al., 1999). By contrast, we observed that Dronc autoprocessing is vital for its activation in Drosophila S2 cells and that cellular cofactors including Dark are not sufficient to induce activation of Dronc in the absence of cleavage at E352. The importance of Dronc processing to Pr1 is highlighted by several observations. First, when overexpressed in S2 cells, wild-type Dronc was processed to Pr1 and induced high levels of apoptosis, and, although the D113A/D135A mutant was only processed to Pr1, it was still able to induce wild-type levels of apoptosis (Fig. 4B,C). By contrast, the E352A mutant remained full length and did not induce apoptosis when overexpressed in cells. Knockdown of Dark by RNAi inhibited autoprocessing of overexpressed wild-type Dronc and protected cells from apoptosis, suggesting that, even though Dronc was expressed at abnormally high levels, interaction with Dark was still required for its activation. Because wild-type overexpressed Dronc still requires endogenous Dark for activation, the observation that E352A Dronc does not induce apoptosis even in the presence of Dark suggests that this is due to E352A being unable to autoprocess and not because it cannot interact with Dark. Together, these results suggest that, in S2 cells, Dronc autoprocessing to Pr1 is necessary and sufficient to induce apoptosis and that Dark is required to facilitate autoprocessing of Dronc to Pr1.
Drice cleavage of Dronc at D135 can produce two different forms depending on the substrate, with Drice cleavage of full-length Dronc resulting in Pr2 and Drice cleavage of Dronc Pr1 resulting in fully processed Dronc. However, as stated above, Drice cleavage had little effect on Dronc activity, so Drice cleavage of Dronc during apoptosis is advantageous in ways other than by directly affecting Dronc activity. Drice cleavage of Dronc at D135 removes the recently identified Dronc IAP-binding domain (Chai et al., 2003) and would result in a processed form of Dronc that can no longer bind (and therefore be inhibited by) DIAP1. Because DIAP1 has been suggested both to inhibit and to degrade Dronc (Hawkins et al., 2000; Meier et al., 2000; Muro et al., 2002; Wilson et al., 2002), loss of DIAP1 binding would also block degradation of Dronc. The question then arises of whether Pr2 or fully processed Dronc is relevant to carrying out apoptosis. During the early stages of apoptosis, DIAP1 is removed rapidly by various mechanisms (Martin, 2002; Wang et al., 1999; Yoo et al., 2002), but it can still be detected 3-6 hours after inducing apoptosis (Muro et al., 2002) and any DIAP1 still remaining might still bind, inhibit and degrade Dronc Pr1. This suggests that removal of the Dronc IAP-binding site by Drice cleavage during the early stages of apoptosis might increase the rate of Drice activation. Although both full-length Dronc and Dronc Pr1 are present at the early stages of apoptosis, Dronc Pr1 levels increase and appear to be the main form during this time. This suggests that, in the early stages of apoptosis, Dronc Pr1 might be the most abundant substrate for Drice cleavage that would result in fully processed Dronc. Therefore, Dronc Pr1 might activate Drice and be immediately cleaved to the fully processed form by the newly activated Drice molecule. Fully processed Dronc would then have no IAP-binding domain, which might help it to amplify the apoptotic signal. Although we have been unable to detect endogenous fully processed Dronc in S2 cells, it is easily detected when Dronc is overexpressed, suggesting that endogenous levels of fully processed Dronc might simply be too low to detect by immunoblotting. The production of Pr2 in the later stages of apoptosis might be an artefact of high Drice activity cleaving full-length Dronc before autoprocessing (activating) at E352 and might not have a role in apoptosis, because we predict that Pr2 would be inactive without cleavage at E352.
In addition to limiting Dronc processing to Pr1, we also determined that reducing endogenous Drice levels by RNAi resulted in inhibition of apoptosis induced by diap1 RNAi. In this experiment, DIAP1 levels were reduced normally and endogenous Dronc was processed to Pr1, suggesting that the inhibition of apoptosis was due to Drice silencing. Because Drice RNAi resulted in Dronc only being processed to Pr1, inhibition of apoptosis might be due to lack of further Dronc processing. However, the observation that the D113A/D135A-mutant Dronc induced wild-type levels of apoptosis even though it was only processed to Pr1 suggests that Drice cleavage of Dronc is not required for apoptosis. Therefore, the requirement for Drice in apoptosis must be due to its cleavage of other substrates required for death. Although several effector caspases are known to be expressed in Drosophila (Dorstyn et al., 1999b; Harvey et al., 2001; Kumar and Doumanis, 2000), the observation that silencing of Drice is sufficient to inhibit apoptosis suggests that Drice is the major, if not the only, active effector caspase in S2 cells, which supports earlier results (Fraser et al., 1997).
These data provide insights into the initiation of apoptosis in Drosophila and also provide a framework for understanding the mechanism of Dronc activation. In addition, these results indicate that, although apoptosis is similar in both mammals and flies, there appear to be some significant differences. The relevance of these differences for human health is illustrated by a recent observations that some mammalian tumour cells appear to have a deregulated apoptotic pathway that more closely resembles the fly death pathway (Huang et al., 2004). Therefore, further understanding of the Drosophila apoptotic pathway might provide a better understanding of how apoptosis is regulated in mammalian tumour cells.
We thank B. Hay and P. Friesen for providing reagents. This work was supported by NIH grants RR107686 from the National Center for Research Resources (NCRR) and P20 RR16475 from the BRIN Program of the NCRR, and by the Kansas Agricultural Experiment Station. This is contribution number 04-365-J from the Kansas Agricultural Experiment Station.