Caspase activation, inhibition, and reactivation: A mechanistic view

Abstract

Caspases, a unique family of cysteine proteases, execute programmed cell death (apoptosis). Caspases exist as inactive zymogens in cells and undergo a cascade of catalytic activation at the onset of apoptosis. The activated caspases are subject to inhibition by the inhibitor-of-apoptosis (IAP) family of proteins. This inhibition can be effectively removed by diverse proteins that share an IAP-binding tetrapeptide motif. Recent structural and biochemical studies have revealed the underlying molecular mechanisms for these processes in mammals and in Drosophila. This paper reviews these latest advances.

Apoptosis, a word coined in the 19th century (Fernandez-Flores et al. 2002) and commissioned in 1972 by Kerr, Wyllie, and Currie (Kerr et al. 1972), refers to a specific form of programmed cell death. The distinct morphological changes of cells undergoing apoptosis are sequentially characterized by shrinkage of the cell, hypercondensation of chromatin, cleavage of chromosomes into nucleosomal fragments, violent blebbing of the plasma membrane, and packaging of cellular contents into membrane-enclosed vesicles called “apoptotic bodies.” These apoptotic bodies are subsequently phagocytosed by surrounding cells. The orderly execution of apoptosis is orchestrated by an evolutionarily conserved pathway from worms to mammals (Fig. 1 ▶).

Figure 1.

A conserved apoptotic pathway in C. elegans (left), mammals (middle), and Drosophila (right). Functional homologs across species are indicated by the same color. Caspase-9 in mammals and Dronc in Drosophila are initiator caspases, whereas caspases-3 and -7 in mammals and DrICE in Drosophila belong to effector caspases. CED-3 in C. elegans is both the initiator and the effector caspase.

The executioners of apoptosis are caspases, a family of conserved cysteine proteases that usually cleave after an aspartate residue in their substrates (Thornberry and Lazebnik 1998). The critical involvement of a caspase in apoptosis was first documented in 1993, when CED3 was found to play a central role in the programmed cell death in the nematode worm Caenorhabditis elegans (Yuan et al. 1993). Since then, the conserved mechanism of apoptosis has been identified in a number of species, particularly in Drosophila melanogaster and mammals (Abrams 1999; Budihardjo et al. 1999; Earnshaw et al. 1999; Shi 2001; Fig. 1 ▶).

Caspases involved in apoptosis are classified into two groups, the initiator caspases, such as caspase-9 in mammals or its functional ortholog Dronc in Drosophila, and the effector caspases, such as caspases-3 and -7 in mammals and their homolog DrICE in Drosophila (Fig. 1 ▶). An initiator caspase invariably contains an extended N-terminal prodomain (>90 amino acids) important for its function, whereas an effector caspase frequently contains 20–30 residues in its prodomain sequence. All caspases are synthesized in cells as catalytically inactive zymogens, and must undergo an activation process. The activation of an effector caspase, such as caspase-3 or -7, is performed by an initiator caspase, such as caspase-9, through an internal cleavage to separate the large and small subunits. An initiator caspase, however, is autoactivated under apoptotic conditions, a process usually requiring and facilitated by multicomponent complexes (Adams and Cory 2002; Shi 2002b). For example, the apoptosome is responsible for the activation of caspase-9 (see later sections; Rodriguez and Lazebnik 1999).

Conserved apoptotic pathways

Genetic studies have identified four genes that act sequentially to control the onset of apoptosis in C. elegans (Horvitz 2003; Fig. 1 ▶). In contrast to the mammalian pathway, CED-3 is the only known apoptotic caspase in C. elegans, and thus is both the initiator and the effector caspase. The activation of CED-3 is mediated, at least in part, by an adaptor protein CED-4, involving the oligomerization of CED-4. In the absence of apoptotic signaling, the pro-apoptotic activity of CED-4 is constitutively suppressed by the antiapoptotic protein CED-9, through direct physical interactions. During apoptosis, the negative regulation of CED-4 by CED-9 is removed by EGL-1, which is transcriptionally activated by cell death stimuli. EGL-1 is a BH3-only member of the Bcl-2 family of proteins. During C. elegans development, the interplay among CED-3, CED-4, CED-9, and EGL-1 results in the death of 131 cells at precise times during development and in precise locations (Horvitz 2003).

In mammals, Bcl-2 and Bcl-xL are structurally and functionally homologous to the worm CED-9 proteins, whereas a large family of BH3-only proteins is distributed throughout the cell to sense apoptotic stress signals (Cory and Adams 2002). Upon receiving apoptotic stimuli, the BH3-only proteins transduce the signal to mitochondria. Through complex actions involving Bak and Bax, cytochrome c is released from the intermembrane space of mitochondria into the cytoplasm, where it binds to and activates Apaf-1, the mammalian ortholog of CED-4 (Li et al. 1997). Then, the binary complex of Apaf-1 and cytochrome c binds its critical cofactor, dATP or ATP, forming a multimeric complex dubbed the “apoptosome.” The only known function of the apoptosome is to recruit and to facilitate activation of caspase-9, which is the mammalian ortholog of CED-3 (Fig. 1 ▶). Once activated, caspase-9 stays associated with the apoptosome as a holoenzyme to maintain its catalytic activity, as caspase-9 in isolation is marginally active (Rodriguez and Lazebnik 1999). The primary target of the caspase-9 holoenzyme is caspase-3, one of the most deleterious effector caspases (Fig. 1 ▶).

The Inhibitor of Apoptosis (IAP) family of proteins suppresses apoptosis by interacting with and inhibiting the enzymatic activity of both initiator and effector caspases (Deveraux and Reed 1999; Salvesen and Duckett 2002; Shi 2002b; Fig. 1 ▶). Several distinct mammalian IAPs including XIAP, c-IAP1, c-IAP2, and ML-IAP, have been identified, and they all exhibit antiapoptotic activity in cell culture. The functional unit in each IAP protein is the baculoviral IAP repeat (BIR), which contains approximately 80 amino acids folded around a zinc atom. Most mammalian IAPs have more than one BIR domain, with the different BIR domains performing distinct functions (Fig. 2 ▶). For example, in XIAP, the third BIR domain (BIR3) potently inhibits the catalytic activity of caspase-9, whereas the linker sequences immediately preceding the second BIR domain (BIR2) selectively targets caspase-3 or -7 (Fig. 2 ▶).

Figure 2.

A schematic diagram of representative IAPs. Three IAPs (XIAP, c-IAP1, and c-IAP2) in mammals and DIAP1 in Drosophila are shown. The BIR1 and BIR2 domains of DIAP1 correspond to the BIR2 and BIR3 domains of a mammalian IAP, respectively. In Drosophila, DIAP1–BIR1 inhibits DrICE, whereas DIAP1–BIR2 is required for the suppression of Dronc. In mammals, the linker sequence preceding BIR2 inhibits caspases-3 and -7, whereas BIR3 inhibits caspase-9.

During apoptosis, IAP-mediated inhibition of caspases is effectively countered by a family of proteins that share an IAP-binding tetrapeptide motif (Shi 2002a). The founding member of this family in mammals is the mitochondrial protein Smac (Du et al. 2000), also known as DIABLO (Verhagen et al. 2000). Smac, synthesized in the cytosol, is targeted to the intermembrane space of mitochondria. Upon stimulation of apoptosis, Smac is released back into the cytosol, together with cytochrome c. Whereas cytochrome c directly activates Apaf-1 and caspase-9, Smac/DIABLO interacts with multiple IAPs and relieves IAP-mediated inhibition of both initiator and effector caspases (Chai et al. 2000; Fig. 1 ▶).

Homologs of most components in the mammalian pathway have been identified in fruit flies (Abrams 1999; Fig. 1 ▶). The Drosophila Apaf-1, known as Dapaf-1 (Kanuka et al. 1999), HAC-1 (Zhou et al. 1999), or Dark (Rodriguez et al. 1999), shares significant sequence similarity with its mammalian counterpart, and is critically important for the activation of the Drosophila initiator caspase Dronc. Dronc, in turn, cleaves and activates the effector caspase DrICE. The Drosophila IAP, DIAP1, binds to and inactivates both DrICE and Dronc through its BIR1 and BIR2 domains, respectively (Chai et al. 2003; Yan et al. 2004; Fig. 2 ▶). During apoptosis, the antideath function of DIAP1 is countered by at least four pro-apoptotic proteins, Reaper, Hid, Grim, and sickle, through direct physical interactions. These four proteins represent the functional homologs of the mammalian protein Smac, and they all share a conserved IAP-binding motif at their N termini (Fig. 3 ▶). The three proteins Reaper, Hid, and Grim are collectively referred to as the RHG proteins.

Figure 3.

A conserved family of IAP-binding motifs. The tetrapeptide motif has the consensus sequence A-(V/T/I)-(P/A)-(F/Y/I/V). Conserved residues are colored green and yellow. The invariant Ala, highlighted by an arrow, is necessitated by the mode of binding. The Drosophila proteins have an additional binding component (conserved 6th–8th residues, colored blue). It should be noted that the IAP-binding tetrapeptide motif has nothing to do with the well-characterized tetrapeptidic substrate specificity of caspases.

The purpose of this review is not to cover all areas of advances in caspase regulation. Rather, I would like to focus on the underlying molecular mechanisms of caspase regulation, most of which have been revealed in the last three to four years.

How is an effector caspase activated?

Caspases are synthesized as single-chain zymogens. An effector caspase is known to exist constitutively as a homodimer, both before and after the intrachain activation cleavage. However, as a consequence of the intrachain cleavage, the catalytic activity of an effector caspase is increased by several orders of magnitude (Salvesen and Dixit 1999). The mechanism of activation for a representative effector caspase, caspase-7, is revealed by the conforma tional changes of the active site following the activation cleavage (Chai et al. 2001b; Riedl et al. 2001a).

The “activated” active site conformation is trapped in the structure of the processed caspase-7 bound to a covalent peptide inhibitor (Wei et al. 2000; Fig. 4A ▶). The active site comprises four surface loops, L1 through L4, all from the same monomer. L1 and L4 constitute two sides of the substrate-binding groove; L3 forms the base. L2 lies across the groove and harbors the catalytic residue Cys186, poised for catalysis (Fig. 4A ▶). Importantly, the L2′ loop, which comes from the adjacent monomer, plays an essential role in stabilizing the “activated” conformation of the active site through intimate interactions with loops L2 and L4.

Figure 4.

Molecular mechanism of caspase-7 activation. (A) Structure of an activated and inhibitor-bound caspase-7 (Wei et al. 2000). Active site loops are labeled, with the catalytic cysteine highlighted in red. The covalently bound inhibitors are shown in orange. (B) Structure of a procaspase-7 zymogen (Chai et al. 2001b). Compared to that of the inhibitor-bound caspase-7, the conformation of the active site loops does not support substrate-binding or catalysis. The L2′ loop, locked in a closed conformation by covalent linkage, is occluded from adopting its productive and open conformation. Although surface loops of the two caspase-7 monomers adopt different conformations, the structural feature essential to the explanation of its activation, namely the orientation of the L2′ loop, is identical. (C) Comparison of the conformation of the active site loops. Compared to the procaspase-7 zymogen, the L2′ loop is flipped 180° in the inhibitor-bound caspase-7 to stabilize loops L2 and L4. The broken connection in loops L3 and L4 indicates high mobility of these regions, as reflected by their high temperature factors from crystallographic refinement.

The active site conformation prior to the intrachain activation cleavage is revealed by the structure of the unprocessed procaspase-7 zymogen (Chai et al. 2001b; Riedl et al. 2001a; Fig. 4B ▶). Compared to the inhibitor-bound caspase-7, the core structural elements of the procaspase-7 zymogen remain unchanged. However, the active site undergoes drastic conformational changes. Loop L2, which contains the catalytic cysteine, is rotated to a large extent, making this residue inaccessible to solvent (Fig. 4C ▶). Remarkably, the L2′ loop, which provides critical support to the formation of an active site, is flipped by nearly 180 degrees, adopting a “closed” conformation. In addition, loops L3 and L4, which form the base and one side of the catalytic groove in the active caspase-7, respectively, become highly flexible (Fig. 4C ▶). These conformational rearrangements in the pro-caspase-7 zymogen do not allow formation of a substrate-binding groove, thus explaining why the procaspase-7 zymogen does not possess detectable catalytic activity.

Prior to the activation cleavage, as is the case in the procaspase-7 zymogen, the L2 and L2′ loops are covalently linked to their C-terminal and N-terminal fragments, respectively. Thus, the unproductive conformation of the active site is a direct consequence of the uncleaved nature of the procaspase-7 zymogen, which locks the L2′ loop in the closed conformation and occludes it from stabilizing the active site through interactions with the L2 and L4 loops. The intrachain cleavage allows the L2′ and L2 loops to switch to their open conformation as observed in the inhibitor-bound caspase-7. As the interactions between loop L2′ and loops L2/L4 are generally conserved, this mechanism is likely to be general for other effector caspases.

In this mechanism, the ability of loop L2′ to move freely in response to inhibitor/substrate binding is a decisive feature. In procaspase-7, movement of the L2′ loop is severely restricted due to the covalent linkage. In caspase-7, this ability is acquired through proteolytic cleavage after Asp198. Because L2′ is at the N terminus of the small subunit, inverting the order of the large and small subunits at the primary sequence level effectively frees the L2′ loop, and hence, is predicted to activate caspases. Indeed, this prediction was confirmed for caspase-3 and -6 (Srinivasula et al. 1998) as well as for the Drosophila caspase DrICE (Wang et al. 1999).

How is an initiator caspase activated?

The activation of an effector caspase zymogen is defined as the intrachain cleavage mediated by a specific initiator caspase. For the initiator caspases, however, the definition of activation carries an entirely different meaning. Although an initiator caspase undergoes an autocatalytic intrachain cleavage, this cleavage appears to have only modest effect on its catalytic activity (Stennicke et al. 1999; Srinivasula et al. 2001). For example, the fully processed caspase-9 in isolation is only marginally active, much the same way as the unprocessed caspase-9 zymogen. In sharp contrast, association with the apoptosome leads to an enhancement of three orders of magnitude in the catalytic activity for the processed as well as the unprocessed caspase-9 (Rodriguez and Lazebnik 1999; Srinivasula et al. 2001). Thus, for caspase-9, activation has little to do with the intrachain cleavage; rather, it refers to the apopto-some-mediated enhancement of the catalytic activity of caspase-9.

At present, we do not understand the molecular mechanism for the activation of any initiator caspase. Nonetheless, two models have been proposed. Based on results using heterologous fusion proteins, an Induced Proximity model was proposed to provide a general explanation for the activation of initiator caspases (Salvesen and Dixit 1999). It states that the initiator caspases autoprocess themselves when they are brought into close proximity of each other. However, this model merely summarizes what have been observed experimentally in laboratories, and does not reveal the molecular mechanisms for the activation of initiator caspases.

More recently, dimerization of the initiator caspases, such as caspases-8 and -9, was proposed to be the driving force for their activation (Renatus et al. 2001; Boatright et al. 2003; Donepudi et al. 2003). This hypothesis provides a mechanism-based explanation for initiator caspase activation, and thus represents a qualitative advance over the previous Induced Proximity model. Based on this model, the function of the apoptosome is to promote the homodimerization of caspase-9 due to its increased local concentrations in the apoptosome. Hence, this model is also known as Proximity-Induced Dimerization, and was proposed to be the unified mechanism for the activation of initiator caspases (Boatright and Salvesen 2003). The validity of this model remains to be examined experimentally, as the supporting evidence is inclusive.

The common theme in the activation of initiator caspases appears to the critical involvement of multicomponent protein complexes. Similar to the caspase-9–activating apoptosome, the assembly of a death-inducing signaling complex (DISC) is indispensable for the activation of caspase-8 (Peter and Krammer 2003). Recent studies have led to the identification of inflammasome (Tschopp et al. 2003), which facilitates the activation of pro-inflammatory caspases involved in the processing of cytokines, and PID-Dosome (Tinel and Tschopp 2004), that underlies the activation of initiator caspase-2. However, except apoptosome, none of these multicomponent protein complexes has been reconstituted in vitro using purified proteins. In some cases, the full components remain to be identified.

How does an IAP inhibit an effector caspase?

Once activated, caspases are subject to inhibition by IAPs. The precise mechanism of IAP-mediated inhibition of caspase-3 and -7 was reveled by a combination of structural and biochemical analyses (Shi 2002b). XIAP is the most potent inhibitor of caspases-3 and -7 in vitro, while c-IAP1 or c-IAP2 has an inhibitory constant that is approximately 100-fold weaker than XIAP (Deveraux and Reed 1999). The structures of caspase-3 and caspase-7, each bound to an inhibitory XIAP fragment, are both available, and reveal identical mechanisms (Chai et al. 2001a; Huang et al. 2001; Riedl et al. 2001b). In the structures, the linker sequences immediately preceding the BIR2 domain of XIAP forms highly similar interactions with both caspases-3 and -7 (Fig. 5A ▶). Compared to the covalent peptide inhibitors, the linker segment of XIAP occupies the active site of caspases in a reverse orientation in terms of peptide orientation, resulting in a blockade of substrate entry (Fig. 5B ▶). Asp148 of XIAP, which was shown to be critical for the inhibition of caspase-3 (Sun et al. 1999), binds the S4 pocket in the same manner as the P4 residue of the covalent peptide inhibitors (Fig. 5B ▶). In addition, Val146 makes a similar set of van der Waals contacts to surrounding caspase residues as does the P2 residue. In this respect, the way the natural inhibitor XIAP binds to caspase-3 or -7 resembles that of the covalent inhibitor, despite a reversal of orientation. In contrast to the covalent peptide inhibitors, Gly144 is located close to the S1 pocket, within van der Waals contact distance of the catalytic Cys in the caspases. Nonetheless, the S1 pocket, which determines the most important selectivity for caspase substrates, remains largely unoccupied.

Figure 5.

Molecular mechanism of IAP-mediated inhibition of effector caspases. (A) Superposition of the structures of caspase-3 (magenta; Riedl et al. 2001b) and -7 (green; Chai et al. 2001a; Huang et al. 2001) together with their bound XIAP fragments colored yellow and gold, respectively. The interactions primarily occur between a linker segment N-terminal to the BIR2 domain of XIAP and the active site of caspase-3 or -7. The small white circle on BIR2 marks where the Smac N-terminal tetrapeptide binds. (B) A close-up view of how the linker sequence of XIAP binds to the active site of caspase-7. Caspase-7 is shown in a surface representation. XIAP residues important for the recognition of the caspase-7 active site are labeled.

Although the linker sequence between BIR1 and BIR2 of XIAP plays a dominant role in inhibiting caspase-3 and -7, this fragment in isolation does not inhibit caspase-3 or -7 (Sun et al. 1999; Chai et al. 2001a). Intriguingly, an engineered protein with the linker peptide fused either N- or C-terminal to BIR1 was fully able to bind to and inhibit caspase-3, while neither BIR1 nor BIR2 in isolation exhibited any effect (Sun et al. 1999). These observations suggest that the linker peptide needs to be presented in a “competent” conformation by a surrounding folded domain. In support of this hypothesis, the linker peptide fused to glutathi-one S-transferase (GST) was able to inhibit caspase-3 and -7 (Chai et al. 2001a; Huang et al. 2001). Nevertheless, the BIR domains also contribute to the inhibition of caspases, as XIAP exhibits about 20-fold higher potency than the GST-linker peptide fusion (Chai et al. 2001a; Huang et al. 2001). Consistent with this observation, XIAP-BIR2 also makes direct contacts to caspase-3 in the crystal structure (Riedl et al. 2001b).

How does an IAP inhibit caspase-9?

Only XIAP has been shown to inhibit caspase-9 in vitro. The required determinants for this inhibition are the conserved surface groove on the BIR3 domain of XIAP and the conserved IAP-binding tetrapeptide motif in caspase-9 (Fig. 3 ▶). Despite the major contribution from the binding of the caspase-9 tetrapeptide to XIAP, other weaker interactions appear to be essential for the inhibition of caspase-9. For example, mutation of His343 in the BIR3 domain of XIAP results in complete loss of inhibition to caspase-9, suggesting that His343 makes important contacts to caspase-9 (Sun et al. 2000).

A mechanistic explanation on the inhibition of caspase-9 by XIAP was revealed by the crystal structure of caspase-9 bound to the BIR3 domain of XIAP (Shiozaki et al. 2003; Fig. 6A ▶). In the uninhibited state, the processed caspase-9 exists exclusively as a monomer (Shiozaki et al. 2003). This caspase-9 monomer has both the potential to be activated by apoptosome as well as the possibility to be inhibited by XIAP. XIAP potently inhibits the catalytic activity of caspase-9 by using the BIR3 domain to heterodimerize with a caspase-9 monomer through the same interface that is required for the homodimerization of caspase-9 (Fig. 6B ▶). This trapped caspase-9 is catalytically inactive due to the absence of a supporting sequence element (L2′ loop) that could be provided by homodimerization. Thus, XIAP inhibits caspase-9 by sequestering it in a monomeric state, which serves to prevent catalytic activity.

Figure 6.

Molecular mechanism of XIAP-mediated inhibition of caspase-9. (A) Structure of the caspase-9/XIAP–BIR3 complex (Shiozaki et al. 2003). XIAP–BIR3 binds to a large caspase-9 surface that is normally required for its homodimerization. Caspase-9 and XIAP–BIR3 are shown in blue and in green, respectively. The catalytic residue, Cys287 on loop L2, and the zinc atom in XIAP–BIR3 are shown in red. (B) Superposition of the caspase-9/XIAP–BIR3 complex (Shiozaki et al. 2003) with the caspase-9 homodimer (Renatus et al. 2001). The coloring scheme for the caspase-9/XIAP–BIR3 complex is the same as in A. The two monomers of the caspase-9 dimer are shown in pink and in orange. Note that XIAP–BIR3 completely overlaps with one caspase-9 monomer.

The specific recognition of the caspase-9 dimerization interface requires four amino acids in the BIR3 domain of XIAP, which are not conserved in the BIR3 domain of c-IAP1 or c-IAP2. This observation indicates that, although c-IAP1 and c-IAP2 can bind to caspase-9 involving interactions with its IAP-binding tetrapeptide, neither IAP can occupy the dimerization interface of caspase-9. Hence, caspase-9 should not be subject to inhibition by c-IAP1 or c-IAP2. Indeed, the BIR3 domain of c-IAP1 or c-IAP2 exhibits no effect on the catalytic activity of caspase-9 (Shiozaki et al. 2003).

How does Smac remove caspase inhibition?

The newly synthesized Smac protein contains 239 amino acids. Its N-terminal 55 residues encode the mitochondrial targeting sequence, and are proteolytically removed in the mature Smac protein (Du et al. 2000; Verhagen et al. 2000). This cleavage results in the exposure of four hydrophobic amino acids, Ala-Val-Pro-Ile, at the N-terminus of the mature Smac (Fig. 3 ▶). This tetrapeptide represents the founding member of a family of IAP-binding motifs in mammals as well as in fruit flies (Shi 2002a). Structural analysis revealed that this tetrapeptide motif binds to a highly conserved surface groove on the BIR3 domain of XIAP, with the first amino acid Ala playing an essential role (Liu et al. 2000; Wu et al. 2000; Fig. 7A ▶). The interactions between Smac and IAPs require an exposed N terminus in this tetrapeptide, hence explaining why only the mature form of Smac is functional in cells. Prior to apoptosis, accidental activation of caspase-9 or caspases-3 and -7 does not lead to cell death because of the inhibitory effect of IAPs. During apoptosis, Smac is released from mitochondria and reactivates the processed initiator as well as effector caspases by relieving IAP inhibition.

Figure 7.

A central role of the IAP-binding motif in the regulation of caspases. (A) Molecular mechanism of Smac-mediated removal of caspase-9 inhibition by XIAP. The IAP-binding tetrapeptide motifs from Smac and from caspase-9 both bind to the same conserved surface groove on XIAP–BIR3. It is this mutual exclusion that allows Smac to remove XIAP-mediated inhibition of caspase-9. (B) Molecular mechanism of RHG-mediated removal of Dronc suppression by DIAP1. The IAP-binding motifs of the RHG proteins bind to the conserved surface groove on DIAP1–BIR2. The binding of an internal peptide fragment of Dronc to the same groove is required for DIAP1-mediated Dronc suppression. (C) The same surface groove is also conserved on the BIR1 domain of DIAP1.

Caspase-9 contains an internal IAP-binding tetrapeptide motif, Ala-Thr-Pro-Phe (Srinivasula et al. 2001; Fig. 3 ▶). In the absence of proteolytic processing, procaspase-9 is unable to form a stable complex with IAPs, because the tetrapeptide motif does not have an exposed N terminus. Proteolytic processing of procaspase-9 at Asp315 leads to the exposure of this internal tetrapeptide motif, which recruits XIAP to inhibit caspase-9 (Figs. 6A ▶, 7A ▶). The mature Smac protein uses a similar N-terminal tetrapeptide to bind to XIAP, and hence competitively relieves XIAP-mediated inhibition of caspase-9. Thus, a conserved IAP-binding motif in caspase-9 and Smac mediates opposing effects on caspase activity. It should be noted that the IAP-binding tetrapeptide motif has nothing to do with the well-characterized tetrapeptidic substrate specificity of caspases.

Although the Smac tetrapeptide alone can counter XIAP-mediated caspase-9 inhibition, it plays a less direct role in the removal of IAP-mediated inhibition of effector caspases. The binding site for the tetrapeptide IAP-binding motif of Smac maps to a conserved surface groove of the BIR domain; however, the fragment responsible for inhibiting caspase-3 or -7 is the flexible sequences preceding the BIR2 domain of XIAP (Fig. 5A ▶). In this respect, binding of a Smac tetrapeptide to the BIR2 surface groove does not impact on and cannot relieve the IAP-mediated inhibition of caspase-3 or -7. Nonetheless, the mature Smac protein is capable of removing caspase-3 inhibition by XIAP. What is the underlying molecular mechanism here? Although a conclusive mechanism remains elusive, modeling studies indicated that, once the BIR2 domain is bound to the mature Smac protein, the linker sequence required for caspase-3 inhibition is partially shielded by the structural scaffold of Smac (Chai et al. 2001a). This steric clash precludes XIAP-BIR2 from simultaneously binding to caspase-3 and Smac. In this model, binding to the BIR2 domain requires not only the N-terminal tetrapeptide of Smac but also an extensive surface available only in the wild-type dimeric Smac protein. This model is consistent with the observation that monomeric Smac mutants only weakly interacted with BIR2, and were unable to remove IAP-mediated inhibition of caspase-3 (Chai et al. 2000).

How does DIAP1 suppress Dronc in fruit flies?

Given the extraordinary conservation of apoptotic pathway between mammals and fruit flies (Fig. 1 ▶), it was fully anticipated that the underlying mechanisms of caspase inhibition and reactivation should be conserved as well. Supporting this prediction, the N-terminal IAP-binding motifs from the pro-apoptotic proteins Reaper, Hid, Grim, and Sickle bind to the conserved surface groove of the BIR2 domain of DIAP1 in a mode that is nearly identical to the Smac/BIR3 interactions (Wu et al. 2001). Surprisingly, however, recent biochemical and structural studies have revealed very different mechanisms for the inhibition and reactivation of caspases in Drosophila (Chai et al. 2003; Yan et al. 2004).

In contrast to XIAP in mammals, DIAP1 in Drosophila only contains two BIR domains (Fig. 2 ▶). Sequence comparison revealed that the BIR1 and BIR2 domains of DIAP1 correspond to the BIR2 and BIR3 domains of XIAP, respectively. Both XIAP and DIAP1 contain a RING domain at their C termini, and can act as an E3 ubiquitin ligase. Indeed, both XIAP and DIAP1 have been shown to promote self-ubiquitination and degradation as well as to negatively regulate the target caspases . Nonetheless, important differences exist between XIAP and DIAP1. The primary function of XIAP is thought to inhibit the catalytic activities of caspases; to what extent the ubiquitinating activity of XIAP contributes to its function remains unclear. For DIAP1, however, the ubiquitinating activity appears to be essential for its function (Hays et al. 2002; Holley et al. 2002; Ryoo et al. 2002; Wing et al. 2002; Yoo et al. 2002).

The caspase-9 ortholog in Drosophila is Dronc. Similar to the mammalian case, DIAP1 directly binds to and suppresses Dronc (Wilson et al. 2002), whereas the RHG proteins can relieve DIAP1-mediated suppression of Dronc. The N-terminal IAP-binding motifs of the RHG proteins bind to a conserved surface groove on the BIR2 domain of DIAP1 (Wu et al. 2001), suggesting that Dronc may bind to DIAP1 in a way similar to the caspase-9/XIAP interactions. However, Dronc does not contain any sequence motif that is remotely similar to the known IAP-binding motifs. How does Dronc bind to DIAP1 then? Biochemical analyses revealed that the BIR2 domain of DIAP1 recognizes a five-amino acid sequence in the linker region between the prodomain and the caspase unit of Dronc (Chai et al. 2003). This recognition is essential for DIAP1-mediated negative regulation of Dronc. Strikingly, structural analysis revealed that the Dronc-binding surface on DIAP1–BIR2 coincides with that required for binding to the N-terminal sequences of the RHG proteins (Chai et al. 2003; Fig. 7B ▶), thus explaining how the RHG proteins competitively eliminate DIAP1-mediated negative regulation of Dronc.

The molecular mechanisms between IAP-mediated regulation of caspase-9 and Dronc are different not only in their mutual recognition, but more importantly, in the way the caspases are regulated. In mammals, XIAP potently inhibits the catalytic activity of caspase-9. In Drosophila, however, DIAP1 exhibits absolutely no effect on the catalytic activity of Dronc (Yan et al. 2004). Rather, DIAP1 acts as an E3 ubiquitin ligase to recognize and to ubiquitinate Dronc (Wilson et al. 2002), hence targeting Dronc for the 26S proteasome-mediated degradation.

How does DIAP1 suppress DrICE in fruit flies?

The caspase-3 ortholog in Drosophila is DrICE. In mammals, the inhibition of caspase-3 or -7 entails a linker segment immediately preceding the BIR2 domain of XIAP (Fig. 5 ▶). However, this linker sequence is not conserved in DIAP1, suggesting a different mode of inhibition. Biochemical analyses demonstrated that DIAP1 directly inhibits the catalytic activity of DrICE through its BIR1 domain, and this inhibition can be countered effectively by the RHG proteins (Yan et al. 2004). Interestingly, DIAP1 binds to and inhibits DrICE only after the cleavage of its N-terminal 20 amino acids, which serve as an autoinhibitory sequence for DIAP1 function (Yan et al. 2004). DIAP1-mediated inhibition of DrICE involves a highly conserved surface groove on BIR1. Crystal structures of BIR1 bound to the RHG peptides show that the RHG proteins use their N-terminal IAP-binding motif to bind to the same surface groove, hence removing DIAP1-mediated inhibition of DrICE (Yan et al. 2004; Fig. 7C ▶). Despite these biochemical advances, we do not yet understand the exact molecular mechanism of how DIAP1 inhibits DrICE.

Perspectives

Our knowledge on programmed cell death primarily comes from investigations on the three model organisms, C. elegans, Drosophila, and mammals. The apoptotic pathway is conserved among these three species (Fig. 1 ▶); however, the mechanisms governing the regulation of caspases are not. In particular, how the Drosophila IAP (DIAP1) exerts its negative regulation on Dronc and DrICE serves as an intriguing contrast to the mammalian XIAP and caspases. Nonetheless, a conserved family of IAP-binding tetrapeptide motifs has apparently survived the harsh selection of evolution, and play a central role in the regulation, both positive and negative, of caspases in both Drosophila and mammals (Figs. 3 ▶, 7 ▶). It should be noted that, in addition to Smac, there are other mammalian proteins that contain the IAP-binding tetrapeptide motif, such as HtrA2/Omi (Fig. 3 ▶). However, the molecular mechanism by which HtrA2/Omi functions in cells remains unclear.

Our understanding on caspase inhibition and reactivation is far from complete. At present, we have absolutely no knowledge on how CED-3, the founding member of the apoptotic caspase family, is activated. It is not clear whether CED-4 alone is sufficient for the activation of CED3. Neither do we understand whether CED-4 acts as part of a CED-3-containing holoenzyme in a way similar to the apoptosome or it merely facilitates the autoactivation of CED-3. In Drosophila, we have gained some insights into the molecular mechanisms by which Dronc and DrICE are regulated by DIAP1 and by the RHG family of proteins. Although we have a structural basis for the recognition of Dronc by DIAP1 and its subsequent implication on Dronc regulation, we do not yet understand how DIAP1 interacts with DrICE. In addition, we have little information on the activation of Dronc, the caspase-9 ortholog in fruit flies. Our past experience does not allow us to predict how Dronc is activated in any convincing manner. But it is certain that the Drosophila Apaf-1 plays a major role in this process (Rodriguez et al. 2002). Finally, even in the mammalian pathway, which has been subject to intense investigation for years, the mechanistic picture remains incomplete. We simply do not understand the molecular mechanisms by which the apoptosome activates procaspase-9. Nonetheless, recent studies of caspase activation, inhibition, and reactivation have galvanized the apoptosis field and will certainly spur more systematic studies on these processes.