Introduction

Cell fate decisions in the hermaphrodite nematode worm Caenorhabditis elegans are predetermined. During development, 131 particular cells (mostly neurons) die at predictable stages.1 The genes controlling this programmed cell death have been identified and cloned. The most downstream gene in the pathway encodes a cysteine aspartyl protease (caspase), CED-3.2, 3 CED-3 shares similarities in sequence and enzymatic activity with a large family of caspases, including 10 human relatives (reviewed in Salvesen4). Like other caspases, the enzymatic activity of CED-3 can be suppressed in vitro and in vivo by the baculoviral pseudosubstrate caspase inhibitor, p35.5, 6 Worms deficient in CED-3 possess an additional 131 cells,7 resulting from the abnormal survival of those cells otherwise destined to die during embryogenesis. Likewise, these 131 cells also survive in nematodes bearing loss-of-function mutations in the ced-4 gene,7 which encodes a homolog of mammalian Apaf-1. Enforced expression of CED-4 triggers apoptosis that requires CED-3 activity,8 placing ced-4 genetically upstream of ced-3. Subsequent biochemical analyses demonstrated that CED-4 binds to CED-3, promoting its aggregation and activation.9, 10, 11, 12, 13

The majority of nematode cells are protected from developmental death by CED-9.14 CED-9 exhibits limited but significant homology to the mammalian apoptosis inhibitor Bcl-2,15 which can also inhibit nematode cell death.15, 16 Conclusive evidence indicates that CED-9 can exert a prosurvival effect through direct interaction with and inhibition of CED-4.9, 10, 17, 18 However, it has also been suggested by Xue and Horvitz19 that CED-9 may habor a second antiapoptotic activity. CED-9 can be proteolytically processed by CED-319 raising the possibility that it could directly inhibit CED-3 via a p35-like pseudo-substrate mechanism. It was not possible to test whether CED-9 could inhibit CED-3 in the absence of CED-4 in worm cells, so Xue and Horvitz19 were denied the opportunity to directly test their hypothesis that CED-3 can inhibit CED-9, using their worm-based system. Uncleavable mutants of CED-9 had reduced antiapoptotic potential in transgenic worm experiments19 providing circumstantial evidence for the theory that CED-9 possesses a CED-3 inhibitory activity.

The C. elegans programmed cell death pathway is a valuable model for apoptosis pathways of other organisms, including mammals. As such, it is important that the mechanisms of action of the principal components of this pathway are fully elucidated. Although the ability of CED-9 to inhibit CED-4 has been firmly established, data verifying or refuting the proposed CED-3 inhibitory activity of CED-9 has not been published to date. This study employed a yeast-based system and biochemical analyses to investigate whether CED-9 could directly inhibit CED-3 in a manner similar to p35. Although the results of this study confirm the ability of CED-9 to inhibit CED-4, our data argue that CED-9 cannot directly suppress CED-3 activity.

Results

Active CED-3 kills Saccharomyces cerevisiae

Heterologous expression of active caspases can be lethal to the yeast S. cerevisiae.20, 21, 22, 23 For a caspase to be lethal to yeast, it must be present at sufficient levels, adopt an active conformation and its substrate specificity must permit it to cleave proteins required for yeast survival. Caspases that do not spontaneously autoactivate can be activated by coexpression of their physiological adaptors, for example Apaf-1 can activate caspase-9 in yeast.24

When expressed from the galactose inducible Gal1/10 promoter (pGALL), CED-3 caused a low but significant level of yeast cell death (Figure 1, column 2), compared to yeast transformed with empty vectors (column 1). When CED-3 was expressed with two CED-3 expression plasmids using different nutritional selections, a dramatic increase in cell death resulted (column 3), presumably because sufficient CED-3 was expressed for its autoactivation. In contrast, yeast expressing CED-3 at a lower level under the control of a crippled Gal1/10 promoter (pGALS) survived (column 4). CED-4 alone was not toxic to yeast (column 5), but yeast bearing the CED-4 plasmid together with the sublethal CED-3 construct died on inducing medium (column 6), suggesting that CED-4 could activate CED-3 in yeast. To visualize CED-3 and CED-4 expression in yeast, epitope tagged proteins were expressed, and detected by immunoblotting (Figure 1, right panels). CED-4 fused to green fluorescent protein (GFP) and myc-tagged CED-3 could cooperate to kill yeast (column 12), although the toxicity was slightly less marked than that triggered by coexpression of the untagged proteins (compare columns 6 and 12).

Figure 1
figure 1

Killing S. cerevisiae by enforcing expression of CED-3 and CED-4. A semiquantitative assay compares the effect of transgenes on yeast growth and viability. Yeast cells were transformed with the indicated plasmids. Suspensions of each transformant were prepared at standardized concentrations. Serial dilutions were made and spotted onto solid inducing media (galactose) or repressing media (glucose) vertically down the plate. Colony size indicates growth rate and colony number cell viability. CED-3 was expressed either from a high level galactose-inducible promoter (pGALL) or from a truncated promoter, directing lower level expression (pGALS). To further manipulate expression levels, either one plasmid or two plasmids under different nutritional selections were transformed. The impact of fusion of epitope tags to CED-3 and CED-4 was assessed (right panels). Immunoblotting was performed to visualize expression and processing of the epitope-tagged CED-3. The degree of processing of CED-3 is reflected by the proportion of full-length myc-CED-3 (FL) compared to partially processed and completely processed myc-CED-3 (pro+L and pro, respectively). An asterisk denotes an endogenous protein recognized by the anti-myc antibody, which migrates slightly slower than full length myc-CED-3. Immunoblotting was also used to detect epitope-tagged CED-4. An asterisk denotes a yeast protein recognized by the anti-GFP antibody, migrating slightly faster than CED-4-GFP. Protein loading was monitored by Coomassie staining

Full size image

At least two alternative explanations could account for the yeast lethality resulting from coexpression of CED-3 and CED-4. The toxicity could either be additive (perhaps resulting from coexpression of two independent mildly lethal proteins) and/or synergistic (due to activation of CED-3 by CED-4). To differentiate between these possibilities, processing of tagged CED-3 was examined by anti-myc immunoblotting (Figure 1) and quantitated as described in the Materials and Methods. Most of the CED-3 protein (73%) in lysates from yeast transformed with the low level tagged CED-3 plasmid (lane 10) was full-length. Coexpression of CED-4 enhanced CED-3 processing: only 30% of CED-3 in lysates from a cotransformant (lane 12) was full-length. This implies that CED-4-mediated activation of CED-3 contributes to the lethality resulting from coexpression of CED-4 with low levels of CED-3.

CED-9 inhibits CED-4-dependent yeast death but does not affect lethality resulting from CED-3 autoactivation

The data presented in Figure 1 indicated that yeast could be killed in a manner either solely dependent on CED-3, or requiring both CED-3 and CED-4. We exploited this finding to explore the capacity of CED-9 to inhibit CED-4 and/or CED-3. Consistent with abundant evidence from nematode, mammalian cell culture and biochemical experiments (cited above), CED-9 greatly reduced the death of yeast provoked by the coexpression of CED-4 with low levels of CED-3 (Figure 2, compare column 2 with column 3). FLAG-tagged CED-9 had a similar effect (column 4). Coexpression of either untagged or FLAG-tagged CED-9 also decreased the proportion of CED-3 processed by CED-4 (47 or 48% respectively, compared to 79% for the CED-3/CED-4 transformant). The broad spectrum baculoviral caspase inhibitor p35 also conferred protection (column 5), as expected given the ability of p35 to inhibit CED-3 in transgenic worms and in vitro.5, 6 p35 also reduced CED-3 processing (52% processed).

Figure 2
figure 2

CED-9 inhibits CED-4-dependent yeast death but not lethality due to CED-3 autoactivation. Yeast transformed with the indicated plasmids were spotted onto inducing and repressing medium as detailed in Figure 1. A FLAG-tagged CED-9 construct was used to visualize CED-9 expression. The impact of coexpression of CED-9 or p35 on processing of myc-CED-3 was monitored by anti-myc immunoblotting, as described in Figure 1. Expression of GFP-tagged CED-4, FLAG-tagged CED-9 and p35 were verified by immunoblotting. Asterisks indicate endogenous proteins detected by the anti-GFP and anti-myc antibodies. Coomassie staining indicates loading

Full size image

To test whether CED-9 could also directly inhibit CED-3, we transformed yeast bearing the high-level CED-3 expression plasmid with plasmids encoding untagged or tagged CED-9. Empty vector and a p35 plasmid were used as controls. Yeast transformed with pGALL-CED-3 were killed upon induction of the caspase (Figure 2, column 7). This lethality was completely suppressed by p35 (column 10). In contrast to its efficient inhibition of CED-4-dependent death, neither tagged nor untagged CED-9 afforded any protection against yeast death induced by overexpression of CED-3 (columns 8 and 9). CED-9 also had no effect on CED-3 autoprocessing. In total, 55% of CED-3 was processed in yeast cotransformed with pGALL-(LEU2)-CED-3 and empty vector, compared to 50% in yeast coexpressing CED-3 together with untagged CED-9 (or 59% for FLAG-CED-9). However, coexpression of p35 did suppress CED-3 autocleavage (25% processed). Expression of FLAG-CED-9 and p35 was verified by immunoblotting.

Purified CED-9 does not inhibit the substrate cleavage activity of recombinant CED-3

The results presented above indicate that CED-9 is incapable of directly inhibiting CED-3, even though it can function in yeast, as demonstrated by its ability to inhibit CED-4-dependent yeast death. This argues against the possibility that in these assays yeast proteins were somehow inactivating CED-9. To completely preclude any possibility that the inability of CED-9 to inhibit CED-3 was due to interference from yeast proteins, a biochemical analysis was performed using purified proteins. Recombinant CED-3 was generated, and its ability to cleave the synthetic fluorogenic substrate Ac-DEVD-AFC was assayed in the presence and absence of CED-9. We, like others,25 were unable to produce full-length CED-9 from bacteria with sufficient purity for biochemical characterization. A truncation mutant lacking the hydrophobic carboxyl terminal domain was therefore used. To confirm that this deletion did not affect stability or activity of the protein, a FLAG-tagged deletion mutant was tested in the yeast death assays and its expression verified by immunoblotting. As for full-length CED-9, the deletion mutant (CED-91–251) could inhibit CED-4-dependent yeast death but not death triggered by autoactivated CED-3 (Figure 3a). Bacterially generated, purified CED-3 required its active site cysteine to cleave the synthetic substrate Ac-DEVD-AFC, as an active site mutant (CED-3C358S) was inactive in this assay (data not shown). p35 significantly reduced the substrate-cleavage activity of recombinant CED-3 (Figure 3b). Like the negative control, glutathione-S-transferase (GST), CED-9 and a cleavage site mutant of p35 were ineffectual in this assay (Figure 3b). CED-9 was unable to inhibit CED-3, regardless of whether the substrate was present at 250-fold excess relative to CED-9, or at equimolar concentration.

Figure 3
figure 3

Purified CED-9 does not inhibit the enzymatic activity of recombinant CED-3. (a) Yeast transformed with the indicated plasmids were spotted onto inducing and repressing medium as detailed in Figure 1. Expression of FLAG-CED-9 and FLAG-CED-91–251 were verified by immunoblotting. Coomassie staining indicates loading. (b) Recombinant CED-3 was incubated with the indicated purified proteins and the caspase substrate Ac-DEVD-AFC. The substrate was used at either a 250-fold molar excess over CED-9 or control proteins, or at equimolar concentration. Substrate cleavage was measured fluorometrically; the maximal rate of cleavage is shown

Full size image

Analysis of the cleavage by CED-3 of CED-9

To determine the significance of cleavage of CED-9 by CED-3, we mutated the previously identified cleavage site aspartates DAQD47 and ES/D67. Wild-type CED-9 and cleavage site mutants were in vitro translated in the presence of 35S-methionine and incubated with various amounts of CED-3 or caspase-3. In vitro translated p35 was used as a control (Figure 4a). Both caspase-3 and CED-3 cleaved CED-9 (Figure 4b). Replacing the cleavage site DAQD47 with a consensus Caspase-3 cleavage site (DEVD47) increased the ability of both Caspase-3 and CED-3 to cleave CED-9 (Figure 4c). In contrast, replacing both proposed P1 aspartate residues with alanines prevented cleavage of CED-9 by CED-3 or caspase-3 (Figure 4d).

Figure 4
figure 4

Caspase cleavage of wild-type CED-9 and CED-9 mutants. The indicated 35S-labeled in vitro translated proteins were incubated with the stated concentrations of caspase-3 or CED-3 for 90 min. The cleavage was visualized by SDS-PAGE and autoradiography. Arrows denote full-length proteins, asterisks indicate nonspecific bands and circles show cleavage products

Full size image

The ability of CED-9 to be cleaved by CED-3 may be important for its function in inhibiting nematode programmed cell death.19 To explore this possibility, we tested the abilities of FLAG-tagged CED-9 cleavage site mutants to inhibit CED-4-dependent yeast death. Mutations in CED-9 that either enhanced or abolished its ability to be cleaved by CED-3 did not affect the ability of CED-9 to inhibit CED-4-dependent yeast death (Figure 5, compare column 3 with columns 4 and 5). Like wild-type CED-9, the mutants were unable to inhibit yeast death induced by CED-3 autoactivation (Figure 5, columns 10–12). The product of CED-3 cleavage of CED-9 (CED-968–280) resembles Bcl-2 more closely than does full-length CED-9. It has been previously reported that a truncation protein engineered to mimic this cleavage product protected mammalian cells from apoptosis more efficiently than wild-type CED-9.26 When expressed in yeast, this truncation mutant was functionally indistinguishable from the wild type CED-9 protein, inhibiting yeast death triggered by coexpression of CED-3 and CED-4 but not by CED-3 autoactivation (Figure 5, compare columns 6 and 13). Approximately equivalent expression of FLAG-tagged wild-type and mutant CED-9 proteins was verified by immunoblotting (Figure 5). These results imply that the amino terminal portion of CED-9 is dispensable for inhibition of the core nematode cell death pathway.

Figure 5
figure 5

Altering the susceptibility of CED-9 to caspase cleavage does not affect its ability to inhibit CED-4-dependent yeast death. Yeast transformed with the indicated plasmids were spotted onto inducing and repressing medium as detailed in Figure 1. Expression of either wild-type FLAG-CED-9 or the indicated mutants was monitored by anti-FLAG immunoblotting. Coomassie staining illustrates loading

Full size image

Discussion

The pathway controlling nematode developmental programmed cell death was previously genetically elucidated. Molecular biology and biochemistry were subsequently used to define the intermolecular relationships of the various pathway components. This pathway, which represents an important model for describing mammalian apoptotic signal transduction, is now generally well understood. However, two mechanisms have been proposed to explain the prosurvival function of CED-9. Inhibition of CED-4 has been firmly demonstrated, but direct inhibition of CED-3 has also been suggested. CED-9 is a homolog of the mammalian antiapoptotic protein Bcl-2, whose mechanism of action is still hotly debated. It was important, therefore, to experimentally test the second potential mechanism of action (that of direct CED-3 inhibition by CED-9). By manipulating expression levels of CED-3, we have developed a system in which yeast death can be induced by CED-3 autoactivation, or dependent on CED-4-mediated activation of CED-3. CED-9 could inhibit yeast death resulting from CED-4 activation of CED-3, but could not block lethality due to CED-3 autoactivation. Verification of the inability of CED-9 to directly inhibit CED-3 was obtained using a fluorometric assay in which cleavage of a substrate by CED-3 was assayed in the presence or absence of CED-9. Unlike p35, CED-9 could not suppress CED-3 cleavage of the fluorogenic substrate Ac-DEVD-AFC (present at 250-fold excess over CED-9), implying that CED-9 is not suicide substrate inhibitor of CED-3. Even when the substrate was present at the same concentration as CED-9, CED-9 had no effect in the fluorometric assay, arguing against a competitive inhibition function. Together, these data refute the proposition that CED-9 can directly inhibit CED-3.

Biochemical analyses confirmed the earlier demonstration19 that CED-3 could cleave CED-9. (No cleavage of CED-9 by CED-3 was detected in yeast lysates, presumably because CED-3 levels in yeast were much lower, relative to CED-9, than those used in vitro.) Both an uncleavable mutant and a truncation mutant designed to recapitulate the CED-3 cleavage product of CED-9 (CED-968–280) were able to inhibit CED-4 activity (but not CED-3 activity) efficiently in yeast. Although these mutants were previously reported to exhibit protective activity in worms, they were significantly less potent suppressors of programmed cell death than wild-type CED-9; the cleavage site mutant rescued around a third as many cells as wild-type CED-9 and the truncation mutant rescued around half.19 If CED-9 cannot inhibit CED-3, as suggested by our data, why would mutations affecting the ability of CED-3 to cleave CED-9 affect CED-9's prosurvival capacity in nematode cells? It is possible that these cleavage site mutations diminish the functional interaction between CED-9 and CED-4 in nematode cells, but that this reduction is too subtle to be detected using the yeast system. Alternatively, residues D47 and D67 may be required for maximal stability of the CED-9 protein, such that mutation of these residues or their removal by an amino-terminal deletion may lead to a degree of protein instability in nematode cells, but not in yeast.

In summary, we have used yeast-based and biochemical assays to explore two proposed mechanisms of action of the nematode-programmed cell death inhibitor, CED-9. Our data confirm the ability of CED-9 to inhibit CED-4, but are not consistent with direct inhibition of CED-3 by CED-9.

Materials and Methods

Plasmids

Yeast expression constructs

Yeast expression plasmids based on the pRS31X series27 directed expression of the various transgenes from the Adh promoter (pADH) or either an intact (pGALL) or truncated (pGALS) Gal 1/10 promoter. The following constructs have been previously described: pADH-(TRP1),28 pGALL-(LEU2) and pGALS-(LEU2),20 pADH-(TRP1)-p3522 and pGALL-(URA).24 The HIS3 gene containing PvuI fragment of pRS31327 was ligated to the PvuI fragment of pADH-(TRP1), which contained the promoter, polylinker and terminator,28 yielding pADH-(HIS3). The CED-4-coding sequence was amplified with primers 1 and 2. A portion of the product was cut with EcoRI, BamHI and NotI. The EcoRI/NotI fragment, encoding the carboxyl-terminal half of CED-4, was ligated into pADH-(HIS3) cut with EcoRI and NotI, to give pADH-(HIS3)-CED-4C. Another portion of the PCR product was cut with EcoRI and NdeI and the EcoRI/EcoRI fragment, encoding the amino-terminal half of CED-4, was then ligated into pADH-(HIS3)-CED-4C cut with EcoRI and treated with alkaline phosphatase, yielding pADH-(HIS3)-CED-4 (PCR was used to determine insert orientation). To make pADH-(HIS3)-CED-4-GFP, GFPS65 T was amplified from p416-MET25-GFPS65 T (a kind gift from Trevor Lithgow) with primers 3 and 4 and cut with NotI and XhoI and cloned into pADH-(HIS3)-CED-4 cut with NotI and SalI. To remove the stop codon of CED-4 to produce pADH-(HIS3)-CED-4-GFP, the XbaI/NotI fragment was replaced with a fragment amplified with primers 5 and 6 and cut with XbaI and NotI. The coding region of CED-3 was amplified with primers 7 and 8 and the product cut with BamHI and XbaI and cloned into pGALL-(LEU2) cut with BamHI and XbaI to produce pGALL-(LEU2)-CED-3. CED-3-coding sequence was excised from pGALL-(LEU2)-CED-3 with BamHI and XbaI and inserted both into pGALS-(LEU2) cut with BamHI and XbaI to make pGALS-(LEU2)-CED-3 and into pGALL-(URA) cut with BamHI and XbaI to generate pGALL-(URA)-CED-3. To express myc-tagged CED-3, the coding region of CED-3 was amplified with primers 9 and 10, cut with BamHI, XhoI and NotI, and the 720 base pair fragment was ligated into pGALL-(LEU2)-CED-3 cut with BamHI and XhoI, yielding pGALL-(LEU2)-myc-CED-3. The BamHI/XbaI fragment was then ligated into pGALS-(LEU2) to give pGALS-(LEU2)-myc-CED-3. The CED-9 coding sequence was amplified with primers 11 and 12, cut with EcoRI and NotI and ligated into pADH-(TRP1) cut with EcoRI and NotI to make pADH-(TRP1)-CED-9. pADH-(TRP1)-FLAG-CED-9 was made by amplifying CED-9 with primers 13 and 14, cutting with EcoRI and NotI and ligating into pADH-(TRP1). pADH-(TRP1)-CED-91–251 was generated by amplifying CED-91–251 with primers 15 and 16, cutting the product with EcoRI and NotI and ligating it into pADH-(TRP1) cut with EcoRI and NotI. pADH-(TRP1)-FLAG-CED-91–251 was made by replacing the PstI fragment of pADH-(TRP1)-FLAG-CED-9, encoding the carboxyl portion of the protein, with the PstI fragment from pADH-(TRP1)-CED-91–251. To generate pADH-(TRP1)-CED-9D47,67A, DNA encoding the amino terminus of CED-9 was amplified from pADH-(TRP1)-CED-9 with primers 13 and 17 and cut with EcoRI and HindIII. DNA encoding the carboxyl terminal part of CED-9 was amplified with primers 18 and 14, cut with HindIII and NotI and inserted into a Bluescript vector (modified such that its polylinker bore EcoRI, HindII and NotI sites in that order and lacked a ClaI site), as a three-way ligation, yielding pBluescript-FLAG-CED-9N+C. Oligonucleotides 19 and 20 were annealed together, then ligated into pBluescript-FLAG-CED-9N+C cut with HindIII and ClaI to produce pBluescript-FLAG-CED-9D47,67A. The EcoRI/NotI fragment encoding the mutated FLAG-CED-9 was then excised and ligated into pADH-(TRP) to give pADH-(TRP1)-FLAG-CED-9D47,67A. To make pADH-(TRP1)-CED-9DEVD, DNA encoding the amino-terminal portion of CED-9 was amplified with oligonucleotides 13 and 21 and the product was cut with EcoRI and BglII. DNA encoding the carboxyl-terminal portion of CED-9 was amplified the primers 22 and 14 and the product was cut with BglII and NotI. The two fragments were inserted into pADH-(TRP1) as a three-way ligation to give pADH-(TRP1)-FLAG-CED-9DEVD. CED-968–280 was amplified with primers 23 and 14, the product cut with EcoRI and NotI and ligated into pADH-(TRP1) cut with EcoRI and NotI, to give pADH-(TRP1)-FLAG-CED-968–280.

Bacterial expression constructs

CED-3 was amplified with primers 24 and 25 and the product cut with NdeI and NotI, giving two fragments. The 5′ NdeI/NotI fragment was cloned into pET23a (Invitrogen) to produce pET23a-CED-3N. pET23a-CED-3N was then cut with NotI and alkaline phosphatase treated and the 5′ NotI/NotI fragment was then cloned into it to give pET23a-CED-3. To produce enzymatically inactive recombinant CED-3, DNA encoding a mutated amino terminal fragment of CED-3 was amplified using primers 24 and 26 and the product cut with NdeI and SalI, then ligated into pET23a-CED-3 cut with NdeI and SalI, to give pET23a-CED-3C385S. DNA encoding residues 1–251 inclusive of Ced-9 was amplified with primers 27 and 28, cut with NdeI and XhoI and cloned into pET23a cut with NdeI and XhoI to produce pET23a-CED-91–251. The coding region of p35, lacking the stop codon, was amplified with primers 29 and 30 and cloned into pET23a cut with NdeI and XhoI to give pET23a-p35.

Constructs for in vitro transcription/translation

CED-9 was cut out of pADH-(TRP)-CED-9 with EcoRI and NotI and ligated into pBluescript II SK+ (Stratagene) cut with EcoRI and NotI, yielding pBluescript-CED-9. CED-9D47, 67A was cut out of pADH-(TRP)-CED-9D47, 67A with EcoRI and NotI and ligated into pBluescript II SK+ cut with EcoRI and NotI to generate pBluescript-CED-9D47, 67A. CED-9DEVD was cut out of pADH-(TRP)-CED-9DEVD with EcoRI and NotI and ligated into pBluescript II SK+ cut with EcoRI and NotI, yielding pBluescript-CED-9DEVD. p35 was amplified with primers 31 and 32, cut with BamHI and XbaI and cloned into pBluescript II SK+ to produce pBluescript-p35.

All PCR-derived stretches of DNA were sequenced to verify the presence of intended mutations and the absence of unintended ones. See Table 1

Table 1 Oligonucleotides
Full size table

Yeast transformation and death assays

The S. cerevisiae yeast stain W303α was transformed and analyzed in survival assays as previously described.23 Briefly, single transformants bearing each desired combination of expression plasmids or empty vectors were grown until stationary phase in selective repressing medium. After washing three times in Tris HCl 10 mM pH 8, EDTA 1 mM, 5 μl of serial five-fold dilutions of each transformant suspension (from OD550=0.086) were spotted onto inducing and repressing selective minimal media plates.

Immunoblotting

Transformants were grown and expression induced as previously described.22 Immunoblotting was performed as previously reported.22 Lysates were also run on gels and stained with Coomassie brilliant blue (BioRad) to visualize protein loading. The antibodies used in this study were anti-GFP (a kind gift from Trevor Lithgow and Lena Burri), anti-p35 (BD Biosciences), anti-myc (clone 9B11; Cell Signaling), anti-FLAG (clone M2; Sigma), anti-rabbit-HRP (Amersham) and anti-mouse-HRP (Sigma). Enhanced chemoluminescence (ECL) anti-myc signals were quantitated using a Chemi-doc apparatus (BioRad). Processing of myc-CED-3 was calculated by adding the specific ECL signals corresponding to partially and fully processed bands and expressing that as a percentage of the total myc-CED-3 signal in each lane.

Generation of recombinant proteins

The pET23a constructs described above were introduced into BL21(DE3)LysS bacteria (Novagen). Transformants were grown overnight at 37°C, then the cultures were expanded 1 : 5 into fresh medium including IPTG (1 mM) and grown at 30°C for a further 4 h (6 h for CED-9). The bacteria were harvested and lysed by freeze-thawing then resuspended in binding buffer (50 mM NaH2PO4, 300 mM NaCl) with 5 μM imidazole and sonicated. The sonicate was bound to NiNTA resin (Qiagen), then the resin was washed and the His-tagged proteins eluted with binding buffer containing 250 μM imidazole.

Fluorogenic CED-3 activity assays

The cleavage of Ac-DEVD-AFC by CED-3 (10 nM) was measured fluorometrically in the presence of GST, p35, p35D87A or CED-91–251. To achieve a substrate : inhibitor ratio of 250 : 1, 50 μM Ac-DEVD-AFC and 200 nM recombinant proteins were used. For equal molar ratios of substrate and inhibitor, 5 μM Ac-DEVD-AFC and 5 μM recombinant proteins were used. The maximal cleavage rate was used to quantitate caspase activity, expressed as the change in concentration of free AFC per minute.

Caspase cleavage of in vitro translated substrates

35S-labeled proteins were generated using the TNT coupled in vitro transcription/translation kit (Promega). The substrates were incubated with recombinant CED-3 (generated as described above) or caspase-3 (BioMol) for 90 min at 20°C (CED-3) or 37°C (caspase-3).