Regulation of the Apaf-1/caspase-9 apoptosome by caspase-3 and XIAP.

The apoptosome is a multiprotein complex comprising Apaf-1, cytochrome c, and caspase-9 that functions to activate caspase-3 downstream of mitochondria in response to apoptotic signals. Binding of cytochrome c and dATP to Apaf-1 in the cytosol leads to the assembly of a heptameric complex in which each Apaf-1 subunit is bound noncovalently to a procaspase-9 subunit via their respective CARD domains. Assembly of the apoptosome results in the proteolytic cleavage of procaspase-9 at the cleavage site PEPD(315) to yield the large (p35) and small (p12) caspase-9 subunits. In addition to the PEPD site, caspase-9 contains a caspase-3 cleavage site (DQLD(330)), which when cleaved, produces a smaller p10 subunit in which the NH(2)-terminal 15 amino acids of p12, including the XIAP BIR3 binding motif, are removed. Using purified proteins in a reconstituted reaction in vitro, we have assessed the relative impact of Asp(315) and Asp(330) cleavage on caspase-9 activity within the apoptosome. In addition, we characterized the effect of caspase-3 feedback cleavage of caspase-9 on the rate of caspase-3 activation, and the potential ramifications of Asp(330) cleavage on XIAP-mediated inhibition of the apoptosome. We have found that cleavage of procaspase-9 at Asp(330) to generate p35, p10 or p37, p10 forms resulted in a significant increase (up to 8-fold) in apoptosome activity compared with p35/p12. The significance of this increase was demonstrated by the near complete loss of apoptosome-mediated caspase-3 activity when a point mutant (D330A) of procaspase-9 was substituted for wild-type procaspase-9 in the apoptosome. In addition, cleavage at Asp(330) exposed a novel p10 NH(2)-terminal peptide motif (AISS) that retained the ability to mediate XIAP inhibition of caspase-9. Thus, whereas feedback cleavage of caspase-9 by caspase-3 significantly increases the activity of the apoptosome, it does little to attenuate its sensitivity to inhibition by XIAP.

The apoptosome is a multiprotein complex comprising Apaf-1, cytochrome c, and caspase-9 that functions to activate caspase-3 downstream of mitochondria in response to apoptotic signals. Binding of cytochrome c and dATP to Apaf-1 in the cytosol leads to the assembly of a heptameric complex in which each Apaf-1 subunit is bound noncovalently to a procaspase-9 subunit via their respective CARD domains. Assembly of the apoptosome results in the proteolytic cleavage of procaspase-9 at the cleavage site PEPD 315 to yield the large (p35) and small (p12) caspase-9 subunits. In addition to the PEPD site, caspase-9 contains a caspase-3 cleavage site (DQLD 330 ), which when cleaved, produces a smaller p10 subunit in which the NH 2 -terminal 15 amino acids of p12, including the XIAP BIR3 binding motif, are removed. Using purified proteins in a reconstituted reaction in vitro, we have assessed the relative impact of Asp 315 and Asp 330 cleavage on caspase-9 activity within the apoptosome. In addition, we characterized the effect of caspase-3 feedback cleavage of caspase-9 on the rate of caspase-3 activation, and the potential ramifications of Asp 330 cleavage on XIAP-mediated inhibition of the apoptosome. We have found that cleavage of procaspase-9 at Asp 330 to generate p35, p10 or p37, p10 forms resulted in a significant increase (up to 8-fold) in apoptosome activity compared with p35/p12. The significance of this increase was demonstrated by the near complete loss of apoptosome-mediated caspase-3 activity when a point mutant (D330A) of procaspase-9 was substituted for wild-type procaspase-9 in the apoptosome. In addition, cleavage at Asp 330 exposed a novel p10 NH 2 -terminal peptide motif (AISS) that retained the ability to mediate XIAP inhibition of caspase-9. Thus, whereas feedback cleavage of caspase-9 by caspase-3 significantly increases the activity of the apoptosome, it does little to attenuate its sensitivity to inhibition by XIAP.
Caspases are a unique class of cysteine proteases that function as effectors of apoptosis, or programmed cell death (1,2). Caspases are expressed in virtually all metazoan cells as catalytically inactive zymogens known as procaspases. Two mechanisms have been described for activating caspases: noncovalent association with caspase activating proteins (e.g. Fadd, Apaf-1) leading to autocatalytic cleavage of the procaspase polypeptide at specific aspartic acid residues, or cleavage at specific aspartic acid residues within the zymogen by other activated caspases. Caspases have been divided into initiators (caspases-2, -8, -9, and -10) and effectors (caspases-3, -6, and -7) based on their relative position within a caspase cascade (3)(4)(5)(6)(7). Initiator caspases are activated by the former mechanism, effector caspases by the latter. The combined proteolytic activities of initiator and effector caspases cleave a variety of vital protein substrates, including DFF45/ICAD, lamin B, gelsolin, Bid, and PAK2, leading to the morphological and biochemical characteristics of apoptosis (8 -11).
Caspase-9 is the initiator caspase in the "intrinsic" or mitochrondrial caspase pathway. Interaction of caspase-9 with Apaf-1, a human homologue of the Caenorhabditis elegans CED4 protein, to form the apoptosome occurs in response to cytochrome c release from the mitochondria of pre-apoptotic cells (12). The apoptosome is a multiprotein complex comprised of Apaf-1, cytochrome c, and caspase-9 in a 1:1:1 molar ratio (12,13). The function of the apoptosome is to cleave and activate the apoptosis effector caspases-3, -6, and -7 (14,29). The apoptosome is assembled when seven Apaf-1:cytochrome c heterodimers oligomerize to form a symmetrical "wheel" and procaspase-9 molecules become associated noncovalently to Apaf-1 via caspase-9 CARD/Apaf-1 CARD heterophilic interaction (15,16). Binding of procaspase-9 to Apaf-1 is important for two reasons. First, it increases the intrinsic catalytic activity of the caspase-9 protease leading to the autolytic cleavage of procaspase-9 at Asp 315 to yield a large (p35) and a small (p12) subunit (14,17,18). And second, cleavage exposes a neoepitope comprising the NH 2 -terminal four amino acids (ATPF) of the small p12 subunit that has been shown to be both necessary and sufficient for binding to the BIR3 domain of XIAP, leading to inhibition of caspase-9 (19). Once activated in the apoptosome, caspase-9 cleaves procaspase-3 at Asp 175 and activates caspase-3. Studies in vitro (14,29) and in intact cells (20) have shown that caspase-3 is capable of feedback cleavage of caspase-9 at Asp 330 , and that this cleavage is associated with an increase in apoptosome activity. Caspase-3-directed feedback cleavage of caspase-9 p35/p12 at Asp 330 would remove the BIR3 recognition motif of caspase-9 (19), creating a caspase-9 species, p35/p10, which may be insensitive to XIAP inhibition.
Using purified, recombinant Apaf-1, caspase-9, and caspase-3 the present study has addressed the regulation of the apoptosome at three levels. 1) Does caspase-9 cleavage affect apoptosome activity? 2) What is the impact of caspase-3mediated feedback cleavage of caspase-9 on apoptosome activity? And 3) following feedback cleavage of caspase-9, what impact does the loss of the BIR3 binding motif from the linker region have on XIAP inhibition. Our results show that recombinant proteins can combine in a dATP-and cytochrome cdependent manner to yield a catalytically active apoptosome * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. capable of cleaving and activating recombinant procaspase-3. The data further demonstrate that association of procaspase-9 with Apaf-1 leads to a partial active apoptosome containing p35/p12 caspase-9. Activation of caspase-3 by the p35/p12containing apoptosomes leads to the initiation of a feedback loop whereby caspase-3 cleaves caspase-9 at Asp 330 . Our data further demonstrate that p10-containing apoptosomes have enhanced catalytic properties relative to p12-containing species. Finally, our study identifies a novel motif at the NH 2 terminus of the p10 subunit capable of mediating XIAP inhibition. Thus, whereas feedback cleavage of caspase-9 by caspase-3 significantly increases the proteolytic activity of the apoptosome, it does little to attenuate its sensitivity to inhibition by the endogenous caspase-9 inhibitor, XIAP.

EXPERIMENTAL PROCEDURES
General Methods and Materials-Tris glycine gels, molecular weight standards for SDS-PAGE, and 10ϫ Tris glycine gel running buffer were from Invitrogen, dATP was obtained from Amersham Biosciences. Horse heart cytochrome c was purchased from Sigma and further purified by passing through the ion exchange column (Mono S). The fluorogenic tetrapeptide substrates DEVD-AMC and LEHD-AMC were synthesized at Idun Pharmaceuticals as previously described (21). General molecular biology methods were used as described in Sambrook et al. (22).

Production of Recombinant Proteins-Recombinant
Apaf-1, procaspase-9, and procaspase-3 proteins were produced in a baculovirus expression system. Expression plasmids for Apaf-1 and wild-type procaspase-9 were constructed as described previously (12) and kindly provided by Dr. Xiaodong Wang (University of Texas, Dallas, TX). Caspase-3 with an in-frame His 9 coding sequence at the 3Ј-end was subcloned into pFastBacI vector at BamHI and NotI sites. Caspase-9 active site (C287A) mutant, cleavage site mutants (D315A, D330A, and D315A/D330A), and the p12 and p10 NH 2 -terminal mutants were generated by the PCR-SOEing method (23). These mutants, engineered to express a His 9 tag at the COOH termini, were subcloned into pFastBacI vector at BamHI and EcoRI sites. Expression plasmids were transformed into DH10Bac Escherichia coli cells (Invitrogen), recombinant bacmids were purified as recommended by the manufacturer (Invitrogen), and their identity were confirmed by PCR amplification analysis. The DNA was then used to transfect Sf21 cells, and virus was amplified as described (12). The virus stocks were amplified to 200 ml and used to infect 1 liter of Sf21 cells at a density of 1 ϫ 10 6 cells/ml. The infected cells were harvested after 38 h for Apaf-1, 20 h for procaspase-9 and procaspase-3, and 24 h for the procaspase-9 mutants and processed caspase-9 (p35/p12). Recombinant proteins were purified by nickel affinity chromatography as described (12), followed by ion exchange chromotography (Mono Q). The eluted protein was dialyzed with buffer A (20 mM Hepes-KOH, pH 7.4, 10 mM KCl, 1.5 mM MgCl 2 , 1 mM NaEDTA, 1 mM NaEGTA, 1 mM dithiothreitol, and 0.1 mM phenylmethylsulfonyl fluoride) and stored in multiple aliquots with 20% glycerol at Ϫ80°C. To obtain the p35/p10 and p37/p10 forms of caspase-9, 2 ϫ 10 4 units of recombinant human caspase-3 was incubated with 17 mg of either caspase-9 p35/p12 or the D315A mutant of caspase-9 at 30°C for 60 min in a total of 5 ml, pH 7.5, ICE buffer. The caspase-3 was separated from the caspase-9 by ion exchange chromatography (Mono Q). The active site mutant of procaspase-3 (C163A) was expressed from pET 15b in BL21(DE3) cells (Novagen), and purified through the nickel affinity column. Active caspase-3 was expressed and purified as described previously (24). Recombinant full-length XIAP with a COOH-terminal His 6 tag was cloned into the pFastBacI vector and expression was achieved as described above in Sf21 cells. The BIR3 domain (amino acids 241-356) of XIAP was expressed and purified as previously described (30).
Time Course Analysis of Procaspase-9 and Procaspase-3 Cleavage-Aliquots of 2.5 g of Apaf-1, 0.75 g of procaspase-9, and 0.375 g of wild-type procaspase-3 or active site mutant procaspase-3 (C163A) were combined, brought to a final volume of 300 l in buffer A with 200 M dATP, 600 nM cytochrome c, and incubated at 30°C. Following incubation for the indicated times, aliquots of 20 l (each containing 50 ng of caspase-9 and 25 ng of caspase-3) were mixed with SDS loading buffer, boiled, and subjected to 16% SDS-PAGE. Detection of immunoreactive caspases was achieved as described above.
Procaspase-3 Cleavage Analysis-60 ng of wild-type procaspase-3, or its active site mutant (C163A) were combined with increasing amounts of a 1:1 molar ratio of Apaf-1 and caspase-9 (p35/p12), 200 M dATP, and 600 nM cytochrome c in buffer A to a total volume of 40 l. After incubation at 30°C for 30 min, 15-l aliquots of each reaction mixture (containing 22.5 ng of caspase-3) were mixed with SDS loading buffer, boiled, and subjected to 16% SDS-PAGE. Detection of immunoreactive caspases was performed as described above.
Detection of Caspase-9 and Caspase-3 Activity with Fluorogenic Tetrapeptide Substrates-Typically, Apaf-1 was mixed with procaspase-9 and procaspase-3 at the desired concentration in a 30-or 40-l reaction mixture in buffer A containing both dATP and cytochrome c. At the end of 30 min, the reaction was stopped by 5-fold dilution with ICE buffer (25 mM Hepes, 1 mM EDTA, 0.1% Chaps, 1 and 10% sucrose) containing either DEVD-AMC (50 M, pH 7.5, for caspase-3) or LEHD-AMC (100 M, pH 6.5, for caspase-9). Caspase activity was monitored as the release of the AMC product over 60 min at room temperature using a Cytofluor fluorescence photometer. Caspase activity was expressed as the change in fluorescence over time derived from the linear phase of the reaction.
Active Site Titration of Caspase-9 -To determine accurately the relative kinetic properties of the various caspase-9 species, two irreversible caspase-9 inhibitors were used to determine the concentration of caspase-9 active sites after association with Apaf-1. Briefly, serial dilutions of the inhibitors were incubated with Apaf-1, dATP, cytochrome c, and the various caspase-9 species for 2 h at room temperature to allow complete inactivation. Aliquots were removed, mixed with 60 nM procaspase-3, and subsequently incubated for an additional 30-min prior to the addition of 50 M DEVD-AMC. Caspase-3 activity assay was monitored as above. The active enzyme concentration was defined as the minimal inhibitor concentration that completely depleted the caspase-9 activities. Titration with either the two inhibitors resulted in identical results. Caspase-9 and caspase-9 mutant concentrations used in Figs. 3 and 4 and Tables I and II, refer to active site concentrations.
Determination of Caspase-9 Kinetic Parameters-The K m for LEHD-AMC was determined for each form of Apaf-1/caspase-9 apoptosome. Aliquots containing 100 nM Apaf-1 and 100 nM caspase-9 (concentration determined by active site titration) were incubated at 30°C for 30 min in the presence of 200 M dATP and 600 nM cytochrome c in a total volume of 10 l in buffer A to allow the formation of Apaf-1/caspase-9 apoptosome. The reaction mixture was subsequently assayed at various LEHD-AMC concentrations in a total volume of 100 l. Rate of catalysis was calculated using the initial slope. K m and V max were determined using a nonlinear regression method to fit the Michaelis-Menten equa- where V ϭ initial catalytic rate, in nanomole of AMC/h, [S] ϭ concentration of the substrate in micromolar; V max ϭ a limiting value of V at sufficiently high or saturating [S]; n ϭ the hill coefficient.

Recombinant Apaf-1, Procaspase-9, and Procaspase-3 Reconstitute dATP/Cytochrome c-regulated Caspase-3 Activation-To
characterize the biochemical events leading to Apaf-1/caspase-9 apoptosome-mediated cleavage of procaspase-3, we expressed recombinant Apaf-1, procaspase-9, and procaspase-3 using a baculovirus expression system as described under "Experimental Procedures." Each protein was purified to apparent homogeneity using nickel affinity chromotography followed by ion exchange chromatography (Fig. 1B). To demonstrate that these recombinant proteins could support dATP/cytochrome c-dependent caspase activation, we monitored the cleavage of procaspase-9 and -3 after incubation with Apaf-1, dATP, and cytochrome c. As expected, procaspase-9 and procaspase-3 were processed in a time-dependent manner yielding products that were consistent with previously published reports (Fig. 1, A and C) (26,27). To confirm that the cleaved caspase-3 products were catalytically active, we incubated aliquots with DEVD-AMC. The time-dependent appearance of DEVD-AMC cleaving activity parallels the degree of caspase-3 processing in each sample (Fig. 1C, lower panel).
Cleavage Requirements for Procaspase-9 in the Apoptosome-Several publications have addressed the relationship between caspase-9 cleavage status and the activity of the apoptosome (17,19). These studies, employing cell-free extracts with and without immunodepletion of endogenous caspase-9, disagree on the impact of the cleavage state of caspase-9 on apoptosome activity. The establishment of a completely recombinant system to study Apaf-1-mediated caspase activation allowed us to address this issue under more defined conditions. To determine the relationship between sites of procaspase-9 cleavage and apoptosome activity, we first expressed and purified several procaspase-9 proteins containing point mutations, including D315A, D330A, D315A/D330A, and C287A ( Fig. 2A). Each of the mutant proteins, and wild-type procaspase-9, were tested for their ability to be cleaved, in vitro, by recombinant active caspase-3. Consistent with previous reports (14), procaspase-9 mutants D330A and D315A/D330A were not processed by caspase-3, whereas the cleavage site mutant (D315A) and the procaspase-9 active site mutant (C287A) yielded the expected cleavage product, p37. Incubation of wild-type caspase-9 with caspase-3 led to the formation of a p35 product, suggesting that caspase-3 directed cleavage of caspase-9 yields an active caspase-9 (p37/p10) capable of cleaving p37 subunits at Asp 315 to generate p35. The lack of p35 production in reactions con-taining procaspase-9 (C287A) supports this proposal ( Fig. 2A).
Caspase-3-directed Cleavage of Caspase-9 Is Required for Full Activation of the Apoptosome-To directly demonstrate that full activation of the apoptosome requires caspase-3 feedback cleavage of caspase-9, we compared the processing of procaspase-3 in two separate reconstituted reactions. One reaction contained wild-type procaspase-3, whereas the other contained procaspase-3 with an active-site cysteine mutation (C163A). As shown in Fig. 3A, the presence of a catalytically active procaspase-3 in these reactions leads to a change in both the procaspase-9 cleavage products and the rate of procaspase-3 cleavage. In the reactions containing wild-type procaspase-3, caspase-9 cleavage products of p37, p35 (note the p35/p37 doublet at 5 min), p12, and p10 were generated, whereas the majority of procaspase-3 was cleaved within 5 min (Fig. 3A). In the procaspase-3 (C163A) mutant reaction, the caspase-9 cleavage products were exclusively p35 and p12, and no processing of procaspase-3 was observed within the first 5 min (Fig. 3A, lower right panel). To rule out the possibility that the difference in procaspase-3 cleavage observed in Fig. 3A was because of cleavage of procaspase-3 by activated caspase-3 in the wild-type caspase-3 reactions, we incubated procaspase-3 with increasing amounts of active caspase-3. Fig. 3B demon-strates that up to 10 units of caspase-3 is insufficient to cleave procaspase-3, whereas as little as 2 units efficiently cleaves procaspase-9, indicating that all the procaspase-3 cleavage observed in Fig. 3A is apoptosome-mediated. In addition to the differences in caspase-3 cleavage rate, the rates for caspase-9 cleavage in each reaction were also different. In reactions containing wild-type procaspase-3, procaspase-9 cleavage was observed within 2 min and its cleavage was nearly complete by 10 min. In the reactions with mutant procaspase-3 the rate of procaspase-9 cleavage was slower, such that no cleavage was observed at 2 min and the reaction had not reached completion by 20 min.
To further establish that caspase-9 cleaved at Asp 330 to generate a p10 subunit is critical for full activation, we added increasing amounts of caspase-9 p35/p12 apoptosomes (at a constant 1:1 molar ratio of Apaf-1:caspase-9) to either wild-type procaspase-3 or the C163A mutant. If caspase-3-mediated feedback can enhance the ability of caspase-9 to cleave procaspase-3, then we would expect procaspase-3 cleavage to be detected at lower apoptosome concentrations in reactions containing catalytically competent caspase-3. As show in Fig. 3C, cleavage of wild-type procaspase-3 is observed when the apoptosome concentration reaches 3.1 nM (Fig. 3C, upper panel), with substantial cleavage occurring at 50 nM. In reactions with the C163A procaspase-3 mutant, cleavage is not observed until the apoptosome concentration reaches 25 nM with only marginal cleavage at 50 nM (Fig. 3C, lower panel). The data presented in Fig. 3 therefore demonstrate that cleavage of caspase-9 at Asp 330 by caspase-3 can significantly enhance the activity of the apoptosome. The magnitude of the enhancement was calculated to be up to 8-fold.
Activities of Different Forms of Caspase-9 in the Apoptosome-To directly demonstrate that apoptosomes containing fully processed caspase-9 are more active than those containing partially processed caspase-9, we produced the three distinct processed forms of caspase-9. Each of the three purified proteins, p35/p10, p37/p10, and p35/p12, migrated as predicted in SDS-PAGE (Fig. 4A), and the molecular masses of each individual subunit were confirmed using mass spectrometry (not shown). We first compared the ability of apoptosomes containing either p35/p12 or p37/p10 to process recombinant procaspase-3 (C163A) by titration of the apoptosome added to the reaction. Caspase-9 p37/p10 apoptosomes were able to cleave procaspase-3 at concentrations of 0.4 to 0.8 nM, whereas caspase-9 p35/p12 apoptosome required 3.1 to 6.3 nM to cleave procaspase-3 during this period (Fig. 4B). To assess the differences between apoptosomes in a quantitative manner, we used the caspase-9 tetrapeptide substrate, LEHD-AMC. After preincubation of Apaf-1 with p35/p12, p37/p10, or p35/p10 to allow the formation of Apaf-1/csp-9 apoptosomes, LEHD-AMC was added to the reactions and apoptosome activity was assessed by following the liberation of AMC. The activities of p35/p10-and p37/p10-containing apoptosomes were significantly greater than that of p35/p12 over the entire range of apoptosome concentration (Fig. 4C). Together these results suggest that removal of the NH 2 -terminal linker region from the p12 small subunit of caspase-9 yields an apoptosome with greater specific activity.
To establish that the observed effects on LEHD-AMC cleavage were because of changes in the intrinsic catalytic properties of the apoptosomes, we determine the K m and k cat for each of the apoptosome species. We found that processing of procaspase-9 at Asp 330 , to form the p10 subunit, was associated with a modest decrease in K m for LEHD-AMC (Table I). This effect was most pronounced with the p37/p10 species yielding a K m 2-fold lower than that for p35/p12. The first-order rate FIG. 3. Caspase-3 feedback cleavage of caspase-9 is critical for full activation of Apaf-1/caspase-9 apoptosome. A, time course analysis of procaspase-9 and procaspase-3 cleavage in the Apaf-1 reconstituted system determined by Western blot analysis. Wild-type procaspase-3 (left panels) or its active site mutant (C163A, right panels) were used as substrates in the reactions. B, Coomassie Blue-stained gel of procaspase-9 and procaspase-3 following incubation with varying amounts of active caspase-3. In C, Western blot analysis was used to follow the processing of wild-type procaspase-3 (upper panel) or its active site mutant (C163A) (lower panel) with increasing amounts of the Apaf-1/caspase-9 (p35/p12) apoptosome. The molar ratio of Apaf-1 to caspase-9 in the apoptosome was maintained at 1:1 in each reaction and the concentration ranged from 0.4 to 50 nM in 2-fold increments. WT, wild type.
constants, k cat , of Apaf-1/caspase-9 (p35/p10) and Apaf-1/ caspase-9 (p37/p10) were 2.8-and 3.6-fold greater than that of Apaf-1/(p35/p12), respectively (Table I). An apparent catalytic rate can be derived from k cat /K m , and is a measure of the ability of the enzyme to turn over substrate. The combined effects of processing at Asp 330 in procaspase-9 to yield a p10 subunit is reflected in a nearly 8-fold increase in the catalytic rate for apoptosomes containing the p37/p10 form of caspase-9.
The form of caspase-9 within the apoptosome can clearly impact cleavage of the synthetic substrate LEHD-AMC. To ask whether this was also true of full-length endogenous substrates, we assessed the ability of each of the distinctly processed caspase-9 species, in association with Apaf-1, to cleave and activate recombinant procaspase-3. Apaf-1 was preincubated with p35/p12, p37/p10, or p35/p10 in the presence of dATP and cytochrome c to allow the formation of apoptosomes. Following preincubation, wild-type procaspase-3 and DEVD-AMC were added simultaneously and the time-dependent activation of caspase-3 was monitored. Apoptosomes containing either p35/p10 or p37/p10 demonstrated similar activity in this experiment, reaching a steady state level of caspase-3 activity within 10 -15 min, with a rate of DEVD-AMC cleavage during this phase of ϳ1500 fluorescence units/h (Fig. 4D). In contrast, the activity of the reaction containing p35/p12 during this time period was ϳ10-fold lower at 150 fluorescence units/h (Fig. 4D). None of these caspase-9 species demonstrated appreciable activity in the absence of Apaf-1 under these conditions (Fig. 4D). At later time points in the p35/p12 reaction the rate of DEVD-AMC cleavage begins to approach that of the other two reactions. This change in the rate of DEVD-AMC cleavage in the Apaf-1/p35/p12 containing reaction (between 30 and 50 min) is a result of feedback cleavage of p12 by active caspase-3 to yield a p10 subunit, and thus p35/p10-containing apoptosomes (not shown). The lag time required for this change in rate to occur is inversely proportional to the amount of input procaspase-3 (not shown).
Fully Active Caspase-9 Is a Target for XIAP-mediated Inhibition through the BIR3 Domain-Srinivasula et al. (19) have reported that the 316 ATPF 319 motif at the NH 2 terminus of the p12 domain, a motif similar to the AVPI motif responsible for SMAC-mediated BIR3 binding, is critical for inhibition of caspase-9 by the BIR3 domain of XIAP (19). Because feedback cleavage of caspase-9 by caspase-3 removes the ATPF motif as part of the linker, we were interested in determining whether p35/p10 or p37/p10 caspase-9 were less sensitive to inhibition by XIAP. Accordingly, we reconstituted the apoptosome with various forms of caspase-9 and tested the ability of BIR3 to inhibit caspase-9-mediated activation of procaspase-3. As expected, BIR3 inhibited caspase-3 activation by apoptosomes containing procaspase-9 (not shown) or caspase-9 p35/p12 (Table II, procaspase-3 as substrate, Fig. 5A). Surprisingly, apoptosomes containing the p35/p10 form of caspase-9 were inhibited with nearly the same IC 50 values as procaspase-9-and p35/p12-containing apoptosomes (Table II and Fig. 5A). These data implied that other regions within the caspase-9 protein are able to mediate inhibition by BIR3. The NH 2 terminus of the fully cleaved p10 small subunit of caspase-9 also contains a tetrapeptide motif, 331 AISS 334 , with similarity to the SMAC and caspase-9 p12 motifs (Fig. 5B). Although Srinivasula et al. (19) found little binding of BIR3 to this motif, we expressed point mutants targeting both of the potential BIR3 binding motifs within caspase-9 (Fig. 5, C and D) to test whether the p10 NH 2 -terminal motif is responsible for BIR3-mediated inhibition of fully processed caspase-9. Apoptosomes containing caspase-9 p35/p10 A331G/I332G (M4 in Table II and Fig. 5C), or caspase-9 p35/p12 A316G/T317G/A331G/I332G (M3) were no longer sensitive to BIR3-mediated inhibition in the fully reconstituted system (Fig. 5A). This result is consistent with a model where the AISS motif present at the NH 2 terminus of p10 is responsible for the inhibition by BIR3 of caspase-9 p35/ p10 observed in Fig. 5A. Interestingly, caspase-9 p35/p12 A331G/I332G (M2) was also insensitive to inhibition by BIR3, even though the input caspase-9 had a p12 subunit with the FIG. 4. Activities of different forms of caspase-9 in the apoptosome. A, Coomassie Blue-stained gel of the various forms of recombinant caspase-9. In B, Western blot analysis was used to follow the processing of procaspase-3 (C163A) when increasing amounts of Apaf-1/caspase-9 (p35/p12) apoptosome (upper panel), or Apaf-1/caspase-9 (p37/p10) apoptosome (lower panel) were added to the reaction. The molar ratio of Apaf-1 to caspase-9 in the apoptosome was maintained at 1:1 in each reaction and the concentration ranged from 0.4 to 50 nM in 2-fold increments. C, increasing amounts of caspase-9 p35/p12 (circles), p35/p10 (squares), or p37/p10 (triangles) were incubated with Apaf-1 at a constant molar ratio in the presence of dATP and cytochrome c at 30°C for 30 min. The activity of the apoptosomes formed was detected with LEHD-AMC. D, caspase-9 p35/p12 (circles), p35/p10 (squares), or p37/p10 (triangles) were incubated with Apaf-1 in the presence of dATP and cytochrome c at 30°C for 30 min. Procaspase-3 and DEVD-AMC were then added to the reactions and the liberation of AMC was continuously monitored over 60 min. Control reactions lacking Apaf-1 yielded no activity (open circles). Fu, fluorescence units. ATPF BIR3 recognition motif. These data can be accommodated by at least two models, either ATPF-mediated inhibition by BIR3 is reversible by removal of the linker region by caspase-3 cleavage of Asp 330 , or caspase-3 cleavage of Asp 330 occurs more rapidly than binding of BIR3 to the NH 2 terminus of p12. Finally, caspase-9 p35/p12 A316G/T317G (M1) was inhibited by BIR3 with nearly the same IC 50 as caspase-9 p35/p10, suggesting this the M1 mutant was inhibited following cleavage at Asp 330 and exposure of the AISS motif.
To demonstrate more directly that each of the NH 2 termini of the small subunits of caspase-9 can mediate BIR3 inhibition, we tested the various apoptosomes in the absence of caspase-3 (and thus without feedback cleavage at Asp 330 ) using the LEHD-AMC substrate. We observed potent BIR3 inhibition of p35/p12 WT and the p35/p12 M2 mutant, whereas the p35/p12 M1 mutant was not sensitive to BIR3 (Table II, LEHD as substrate, and Fig. 5E). As expected from Fig. 5A, we also observed inhibition by BIR3 of caspase-9 p35/p10 WT but no inhibition of the p35/p10 M4 mutant or the p35/p12 M3 mutant. Finally, to demonstrate that the above observations are relevant to BIR3-mediated inhibition in the context of fulllength XIAP, we repeated the experiment using purified fulllength XIAP. As shown in Fig. 5F, the ability of full-length XIAP to inhibit the apoptosome is dependent on the presence of  (nM) of XIAP or BIR3 mediated inhibition of caspase-9 and caspase-9 mutants in apoptosome XIAP-or BIR3-mediated inhibition of wild type and mutant Apaf-1/caspase-9 apoptosome was measured in Fig. 5 5. Either of the two conserved XIAP binding motifs within caspase-9, ATPF or AISS, are sufficient for inhibition by XIAP or its BIR3 domain. A, caspase-9 or its mutant were incubated with Apaf-1, procaspase-3, dATP, cytochrome c, and different amounts of BIR3. The activity of processed caspase-3 was measured by the cleavage of DEVD-AMC. Open circles, p35/p12 (WT); open diamonds, p35/p10 (WT); open squares, p35/p12 (M1); filled diamonds, p35/p12 (M2); filled triangle, P35/p12 (M3); filled square, P35/p10 (M4). Symbols are the same for A, E, and F. B, alignment of the NH 2 -terminal amino acids from mouse, Xenopus, and human caspase-9 p10, along with the NH 2 terminus of human caspase-p12 and human SMAC/Diablo. C, details of caspase-9 p35/p12 p35/p10 point mutants used to assess amino acid requirements for BIR mediated inhibition. D, Coomassie Blue-stained gel of caspase-9 small subunit NH 2 -terminal mutants. In E and F, caspase-9 or the caspase-9 mutants were incubated with Apaf-1 dATP, cytochrome c, and either BIR3 or full-length XIAP. The activity of Apaf-1/caspase-9 holoenzyme was directly measured by the cleavage of LEHD-AMC. either ATPF or AISS exposed at the NH 2 terminus of the small subunit. The IC 50 values for each of these reactions is shown in Table II (XIAP inhibitor, LEHD as substrate). These data support a model where XIAP is able to inhibit caspase-9 at all levels of activation by virtue of a conserved motif at the NH 2 terminus of the less active p12 subunit as well as the more active p10 subunit. DISCUSSION In the present study, we have addressed the potential physiological role(s) of caspase-3-mediated feedback cleavage of caspase-9 using purified, recombinant Apaf-1, caspase-9, and caspase-3 proteins in a reconstituted apoptosome in vitro. Because caspase-9 can be cleaved by itself (Asp 315 ) or by caspase-3 (Asp 330 ), the combined action of the two proteases on procaspase-9 can produce three different caspase-9 enzymes: p35/ p12, p37/p10, or p35/p10 (Fig. 1). We attempted to discriminate the role of each cleavage site by generating various recombinant caspase-9 species. First we generated procaspase-9 species in which each cleavage site had been eliminated alone or in combination and assessed their ability to activate procaspase-3. Second we produced all three processed forms of caspase-9, p35/p12, p37/p10, and p35/p10, and reconstituted them separately into the apoptosome to compare their relative catalytic activities and sensitivities to inhibition by the BIR3 domain of XIAP.
Experiments in which procaspase-9 with cleavage mutations were used suggest that cleavage at Asp 330 is required for full activation of the apoptosome (Fig. 2C). This observation is inconsistent with previously published reports (17)(18)(19) in which little, if any, difference was observed in the ability of unprocessed and processed caspase-9 to cleave and activate procaspase-3. Although no definitive explanation for this discrepancy can be presented, there were significant differences in methodology between these previous reports and the present work. In general, the previous reports employed cellular extracts as the context for assessing relative activities of caspase-9, whereas the present work utilized recombinant proteins at defined concentrations and molar ratios. The use of cellular extracts may introduce additional cytosolic factors, including but not limited to endogenous procaspase-9 that may modulate apoptosome activity and obscure the potential impact of cleavage at Asp 330 .
Cleavage of caspase-9 at Asp 330 by caspase-3 increases the catalytic activity of caspase-9 in the apoptosome up to 8-fold. This is because of an increase in k cat and a decrease in K m of p10-containing forms of the apoptosome as measured by cleavage of the caspase-9 substrate LEHD-AMC (Table I). Thus, caspase-3 feedback cleavage of either p35/p12 (to generate p35/ p10) or uncleaved procaspase-9 (to generate p37/p10) has the potential to amplify the proteolytic activity of the apoptosome. If operational in cells, this would create a feed-forward mechanism to accelerate the apoptotic destruction of the cell once the downstream effector caspase-3 has been activated. At least two recent studies support this possibility. Fujita et al. (20), using caspase-9 cleavage site-specific antibodies, demonstrated in living mouse cells a sequential cleavage of procaspase-9, first by caspase-9 and then by caspase-3. These authors also noted a correlation between cleavage at Asp 330 and increased cellular caspase-9 activity. In a second study, Slee et al. (29) using cell-free extracts, have demonstrated a requirement for caspase-3 activity to achieve maximum apoptosome activity. The optimum activity in this study also correlated with the presence of an Asp 330 cleavage product. Our data are consistent with these observations and provide biochemical support for the potential impact of such a feedback loop.
Recent work by Srinivasula et al. (19) would predict that feedback cleavage of caspase-9 by caspase-3, and loss of the BIR3 binding motif, would be accompanied by a loss of sensitivity to inhibition by XIAP, representing a potential point of no return for the cell. Thus, we were surprised to observe that either full-length XIAP or the isolated BIR3 domain is equally effective at inhibiting p10-containing forms of caspase-9 (p35/ p10, p37/p10) as the p12-containing caspase-9 (p35/p12, Table  II and Fig. 5). Cleavage by caspase-3 at Asp 330 exposes a tetrapeptide motif AISS that appears to be responsible for XIAP inhibition (Fig. 5). Two lines of evidence support this conclusion: 1) the similarity of the AISS sequence to the SMAC/ reaper tetrapeptide motif (Fig. 5B), and 2) the nearly complete loss of function when the NH 2 -terminal alanine of p10 is mutated to glycine (Fig. 5, A, E, and F). This p10 NH 2 terminus is well conserved (Fig. 5B), indicating that it is important that caspase-3 feedback-cleaved caspase-9 still be subject to regulation by IAPs.
The catalytic activity of caspase-9 in the apoptosome is regulated at multiple levels. First, noncovalent association with Apaf-1 via CARD/CARD interaction causes an increase in protease activity (18,28). The precise activation mechanism is unknown but may involve allosteric changes in procaspase-9 induced upon binding to Apaf-1. Second, autolytic cleavage at Asp 315 , perhaps because of the induced proximity (14) of caspase-9 molecules in the multivalent apoptosome, further increases caspase-9 protease activity. Third, feedback cleavage at Asp 330 by caspase-3 increases caspase-9 activity still further, providing for an amplification of the cascade. How the loss of the 15-amino acid linker region leads to enhancement of the catalytic activity is not clear. In the recently published structure of caspase-9 the NH 2 terminus of the p12 subunit was not resolved (28). However, the NH 2 terminus of the p12 subunit was adjacent to the catalytic site, suggesting that removal of the NH 2 terminus could affect either substrate recognition or catalysis by caspase-9. Finally, all forms of activated caspase-9 are subject to inhibition by XIAP. Thus, even after full activation of the apoptosome, it appears that XIAP may be capable of setting a threshold over which sufficient active caspase-9 must be generated before a cell can complete the apoptotic program.