Biochemical Society Transactions

GlaxoSmithKline Award Lecture

Caspase activation cascades in apoptosis

Susan E. Logue, Seamus J. Martin


Apoptosis, a highly controlled mode of cell death, is utilized to eliminate superfluous, aged, injured or infected cells from the body. Caspases, a family of aspartic acid-specific proteases, are the major effectors of apoptosis. To curtail their activity, caspases are normally synthesized as inactive precursors, but become activated at the onset of apoptosis by activation signals. Once active, caspases preside over the ordered dismantling of the cell through restricted proteolysis of hundreds of substrate proteins. Over the last 10 years, intense research has focused upon the pathways that control caspase activation. Although some, such as the apoptosome and death receptor-mediated pathways to caspase activation, are well established, others are less clearly defined. In this review, we discuss current perspectives concerning the diverse pathways to caspase activation.

  • apoptosis
  • apoptosome
  • Bcl-2 family
  • caspase
  • granzyme B


Apoptosis is a highly organized mode of cell death that is used to eliminate superfluous, aged, injured or infected cells in diverse biological settings [1]. During apoptosis, cells are dismantled from within and display plasma membrane alterations that provoke their removal by phagocytic cells. Thus apoptosis represents a controlled demolition process that involves the dismantling of cellular structures and the swift removal of the dying cell so that collateral damage to surrounding tissue is minimized and release of pro-inflammatory cellular constituents is avoided. During apoptosis, the family of aspartic acid-specific proteases known as caspases are the demolition experts that are called upon to co-ordinate as well as execute the process [24].

Owing to the potentially lethal nature of these enzymes, caspases are initially expressed as inactive precursors (zymogens) and require limited proteolysis at internal aspartic acid residues for activation [3,4]. This strategy provides an important means of keeping caspase activities under control and reduces the possibility that cells will enter apoptosis inadvertently. Caspases have a substrate specificity for aspartic acid, which is relatively rare among proteases, and also require limited processing at aspartic acid residues in order to become activated. This immediately suggests that caspases either become activated through autoproteolysis or are activated by other caspases or non-caspase proteases with a similar specificity for aspartic acid residues. In fact, all three scenarios have been observed and will be discussed in detail below.

Initiators and effectors

To date, 12 human caspases have been identified, all of which share similarities in amino acid sequence, tertiary structure and substrate specificity [3]. Caspases are expressed as single-chain pro-enzymes composed of three domains: an N-terminal pro-peptide (or pro-domain), a large subunit and a small subunit. Those caspases involved in apoptosis can be divided into two functional subgroups on the basis of their known or hypothetical roles. Initiator or apical caspases (caspases 2, 8, 9 and 10) are responsible for initiating caspase activation cascades. The latter caspases have long N-terminal pro-domains that contain recognizable protein–protein interaction motifs [CARDs (caspase recruitment domains) or DEDs (death effector domains)] that are also found in molecules that promote caspase activation. The second subgroup, the downstream or effector caspases (caspases 3, 6 and 7), are thought to be responsible for the actual demolition/dismantling of the cell during apoptosis and tend to have short or absent pro-domains (Figure 1).

Figure 1 Schematic representation of the human caspases depicting their structure and proposed function

Upon activation, apical caspases propagate death signals by activating downstream effector caspases in a cascade-like manner [4]. The effector caspases orchestrate the direct dismantling of cellular structures, disruption of cellular metabolism, inactivation of cell death-inhibitory proteins and the activation of additional destructive enzymes [3,4]. To date, over 400 substrates for the effector caspases have been identified [5]. However, the cleavage of only a small subset of these substrates has been definitively associated with the characteristic features of apoptosis such as blebbing of the plasma membrane, nuclear condensation and DNA fragmentation. The serine/threonine kinase ROCK I (Rho-associated kinase I), the structural proteins vimentin, Gas2 and plectin, and the nuclear protein ICAD [inhibitor of CAD (caspase-activated DNase)] have all been linked to specific morphological changes associated with apoptosis. For example, caspase-3-mediated cleavage of ICAD, breaks the inhibitory association of ICAD with CAD, allowing CAD to initiate DNA fragmentation [6]. Targeting of the cytoskeletal proteins vimentin [7,8], Gas2 [9] and plectin [10] by caspases contributes to changes in cell shape, whereas proteolysis of ROCK I has been associated with nuclear fragmentation and plasma membrane blebbing [11,12]. However, with the exception of the above substrates, the significance of the majority of the proteolytic cleavage events seen during apoptosis remains relatively unresolved. Many of the proteins cleaved by caspases are essential for important housekeeping and structural functions, thus caspases appear to mount a comprehensive assault on most of the cell's vital systems. We have previously described apoptosis as akin to a ‘death by a thousand cuts’, where multiple fatal wounds are inflicted upon a cell due to caspase activation, with none of the specific wounds being essential for death as others will undoubtedly finish the job in their absence [2].

Synthetic, viral and cellular inhibitors of caspases can effectively block the morphological and biochemical features of apoptosis, irrespective of the particular initiating stimulus. This evidence, combined with analysis of caspase-deficient mice or cell lines, suggests that caspases are a crucial requirement for most, if not all, forms of apoptosis in mammals.

Proximity-induced caspase activation

The common strategy used to achieve initiator caspase activation involves formation of protein complexes containing several pro-caspase molecules [1315]. This is achieved through recruitment of caspase zymogens into complexes by specific adaptor proteins [1315]. Latent caspase zymogens possess low but detectable catalytic activity which is sufficient to permit autoactivation in circumstances where sustained close proximity between several caspase zymogens is achieved.

Over the last 10 years, caspase activation pathways have been the focus of intense research. Although we do not understand precisely how caspases co-ordinate all of the events that take place during apoptosis, we do have a reasonably good understanding of how caspases become activated in some important contexts. In this review, we will briefly discuss our current understanding of how caspase activation is regulated during divergent forms of apoptosis, how activation of specific initiator caspases can result in a cascade of additional caspase activation events and how active caspases may be regulated downstream.

Caspase activation pathways

Currently, the most studied and consequently well-defined pathways leading to apoptosis are the Apaf-1 (apoptotic protease-activating factor 1) apoptosome pathway, the death receptor pathway and the granzyme B-initiated pathway.

The Apaf-1 apoptosome pathway to caspase activation

A large body of evidence suggests that mitochondria act as important conduits for signals associated with cell damage and that many key regulators of apoptosis promote or inhibit the loss of mitochondrial integrity [14,16,17]. In this pathway, divergent cellular stresses such as DNA damage, heat shock, oxidative stress and many other forms of damage, result in the release of cytochrome c and other mitochondrial intermembrane space proteins into the cytoplasm [14,16,17]. Early studies examining cell death initiated by cell damage, such as by cytotoxic drugs or ionizing radiation, found that overexpression of Bcl-2, a protein localized to mitochondria, blocked cell death [1820].

Bcl-2 is the founding member of a large family of proteins that are important in the regulation of cellular life and death decisions [21]. Over the last few years, our knowledge of this protein family has expanded dramatically, and we now know that the Bcl-2 family comprises at least 17 members, some of which promote apoptosis, whereas others suppress this form of cell death. Although functionally distinct, each Bcl-2 family member possesses at least one BH (Bcl-2 homology) domain. Pro-survival members, Bcl-2, Bcl-XL, Bcl-w, Bcl-b, Mcl-1 and A1 typically contain four BH domains, whereas pro-apoptotic members contain between one and three BH domains. The pro-apoptotic members of the Bcl-2 family can be divided further into two distinct groups: those that contain BH domains 1–3 (Bax, Bak and Bok) and those containing only a single BH-3 domain referred to as ‘BH3-only’ proteins [Noxa, PUMA (p53 up-regulated modulator of apoptosis), Bad, Bim, Bid, Bmf, Hrk and Bik] [22].

Much investigation has revealed a complex network of interactions between family members in which the ratio of pro- to anti-apoptotic Bcl-2 family members controls the release of cytochrome c from mitochondria. Pro-apoptotic members Bax and Bak have essential and partially redundant functions in regulating cytochrome c release. Normally, Bax is localized to the cytoplasm, where it is maintained in an inactive conformation, possibly via interactions with pro-survival proteins Bcl-2, Bcl-XL and Mcl-1 [23,24]. Similarly, Bak, an integral membrane protein localized to the outer mitochondrial membrane, is restrained through binding to anti-apoptotic Bcl-2 proteins. Following receipt of pro-apoptotic signals, levels of active BH3-only proteins increase by a mixture of transcriptional up-regulation (PUMA and Noxa) and post-translational modification (Bim, Bad and Bid), depending upon the initiating stimulus [25]. Activation of the BH3-only cohort of proteins shifts the balance in favour of apoptosis by relieving the inhibition placed upon Bak and Bax by anti-apoptotic Bcl-2 proteins (Figure 2). As a consequence, Bax and Bak undergo conformational changes that permit oligomerization of these proteins within the mitochondrial outer membrane and results in release of mitochondrial intermembrane space proteins, the most important of which is cytochrome c.

Figure 2 The major routes to caspase activation that have been defined to date

See the main text for details of each pathway. (I) The extrinsic or death receptor pathway, (II) the intrinsic or mitochondrial pathway, (III) the granzyme B pathway to caspase activation.

BH3-mediated repression of pro-survival Bcl-2 family members was thought, until recently, to be a relatively non-selective process. The use of peptides mimicking the α-helical BH3 domain permitted studies examining interactions between BH3-only proteins and other members of this family and found that the BH3-only subfamily can be divided into direct activators and derepressors of Bax/Bak channel opening. Direct activators, such as Bid and Bim, have the ability to trigger directly Bax and Bak oligomerization and efflux of mitochondrial proteins [26]. Other BH3-only proteins such as Bad, Bik and PUMA, while not directly activating Bax or Bak, do so indirectly by neutralizing pro-survival Bcl-2 proteins. Furthermore, there is significant selectively amongst the interaction of derepressors with pro-survival Bcl-2 proteins. For instance, Bad has been demonstrated to interact with Bcl-2 and Bcl-XL, but not Mcl-1, whereas Noxa interacts with Mcl-1, but not with Bcl-2 or Bcl-XL [27].

Ultimately, the balance of pro- and anti-apoptotic Bcl-2 family proteins controls permeabilization of the outer mitochondrial membrane and release of intermembrane space proteins, most notably cytochrome c. Efflux of cytochrome c from mitochondria drives the assembly of a high-molecular-mass caspase-activating complex in the cytoplasm, termed the mitochondrial apoptosome [14,16,28]. In the presence of cytochrome c and dATP, Apaf-1, the scaffold around which the apoptosome is built, recruits and activates caspase 9, which then propagates a cascade of further caspase activation events downstream [2931]. Thus Apaf-1 functions as a sensor for cell damage by detecting the release of a common mitochondrial protein that is not found in the cytosol of healthy cells. In the absence of cytochrome c, Apaf-1 exists as a monomeric protein that is incapable of interacting with caspase 9. However, cytochrome c and dATP promote Apaf-1 oligomerization into a wheel-like structure consisting of seven Apaf-1 molecules and a similar number of caspase 9 dimers (Figure 2) [32]. Upon recruitment to the apoptosome, caspase 9 is thought to become activated owing to the increase in local concentration of caspase 9 zymogens, and also because of the fact that association of caspase 9 with Apaf-1 may induce the caspase active site into an active configuration [33].

Evidence from several groups suggests that there are several tiers of caspase activation events (see Figure 2) resulting from caspase 9 activation within the apoptosome complex [2931]. Using cell-free extracts devoid of specific caspases, work performed in our laboratory has revealed that caspase 9 is directly responsible for the activation of caspases 3 and 7 downstream (Figure 1) [31,34]. This is consistent with gel-filtration chromatography experiments which have shown that these enzymes, particularly caspase 3, co-elute with the apoptosome [28,35]. Although caspase 7 does not appear to drive any further caspase activation event, caspase 3 propagates the cascade further by proteolytic processing and activation of caspases 2 and 6 downstream [31,34]. In addition, caspase 3 participates in a positive-feedback amplification loop to promote further processing of caspase 9 [30,31]. In the final stage of this cascade, caspase 6 catalyses the activation of caspase 8 and caspase 10 [31,36]. Knockout mouse studies have confirmed the importance of caspase 9 and Apaf-1 in the intrinsic pathway to caspase activation. Cells derived from Casp-9-null animals demonstrated resistance to internal stress agents, such as cytotoxic drugs and radiation [37,38]. A similar resistance to apoptotic stimuli was also evident in Apaf-1-knockout animals, reinforcing the importance of apoptosome formation to intrinsic caspase activation [39].

Death receptor-initiated pathways to caspase activation

Certain members of the TNF (tumour necrosis factor) receptor superfamily share a distinct domain within their cytoplasmic tails that has been termed the ‘death domain’ [15]. These receptors have been implicated in transducing signals for apoptosis in diverse contexts, but appear to be particularly important for the regulation of lymphocyte numbers within the immune system. For example, peripheral T-cells undergo clonal expansion in response to antigenic challenge, but the majority of these cells subsequently die when their presence is no longer required. Elimination of these cells is a natural part of immune system homoeostasis and ensures that lymphocyte cell numbers do not continually increase throughout life (if they did, we would soon be overwhelmed by their numbers). Signalling through the CD95 (Fas/APO-1) death receptor appears to be important for the ongoing elimination of recently stimulated peripheral T- and B-cells.

Upon engagement of death receptors with their extracellular ligands [such as FasL/CD95L, TNF and TRAIL (TNF-related apoptosis-inducing ligand)], the receptor death domains recruit adaptor proteins that can recruit caspases directly into the receptor complex (Figure 2) [15]. In most cases, the initiator caspase in death receptor signalling pathways appears to be caspase 8; however, a close relative of the latter (caspase 10) may also become recruited to death receptor complexes in certain instances [40]. As in the apoptosome context, recruitment of caspase 8 or 10 into death receptor complexes is thought to drive caspase activation through increasing the proximity of caspase 8/10 zymogens, which facilitates complete processing of these proteases. In this scenario, the pro-apoptotic stimulus (in the form of extracellular death ligands) is transduced by a bipartite caspase adaptor protein called FADD (Fas-associated death domain), which acts as a caspase-8-aggregating scaffold within intracellular death receptor complexes [15].

Downstream of caspase 8 activation within the death receptor complex, there are potentially two alternative pathways to destruction that a cell may embark upon [41]. Both pathways ultimately make use of the same repertoire of caspases to trigger demolition of the cell. However, the ability of caspase 8 to generate sufficient quantities of active caspase 3 to propagate the caspase cascade determines the route to death that a cell will embark upon.

In many cell types, stimulation of death receptors results in the activation of sufficient caspase 8 to propagate the death signal through direct activation of downstream caspases [41]. In these cells, death receptor-induced death cannot be blocked by the anti-apoptotic molecule Bcl-2 or its close relatives (Bcl-XL, Mcl-1, A1 and Bcl-w). As illustrated in Figure 2, in this scenario, mature caspase 8 cleaves sufficient caspase 3 to engage directly the full repertoire of caspases required to destroy the cell.

However, certain cell types, such as hepatocytes, fail to activate sufficient amounts of caspase 8 within death receptor complexes (for reasons that are currently unclear) to properly guarantee the destruction of the cell [41]. In these cell types, the pro-apoptotic signal must be amplified via engagement of the cytochrome c/Apaf-1 pathway, with the result that death receptor-dependent death is Bcl-2-sensitive in these cells [41,42]. As illustrated in Figure 2, cells that engage the apoptosome pathway as a means of enhancing the death signal appear to do so via the BH3-only protein, Bid [43,44]. In this pathway, the reduced level of caspase 8 activated at death receptor complexes is nonetheless sufficient to fuel the proteolytic processing of Bid [43,44]. Upon proteolysis by caspase 8, the active ∼15 kDa C-terminal fragment of Bid (tBid) provokes the release of cytochrome c from mitochondria by stimulating the oligomerization of Bax and/or Bak to form channels in mitochondrial outer membranes (Figure 2) [45,46]. The cytochrome c that is released as a result of the actions of tBid on mitochondria triggers apoptosome assembly and amplification of caspase activity via the Apaf-1/caspase 9 cascade (Figure 2). Furthermore, the release of another mitochondrial factor, Smac (second mitochondrial-derived activator of caspase)/DIABLO [direct IAP (inhibitor of apoptosis protein)-binding protein with low pI], boosts caspase activation further by overcoming the ability of XIAP (X-linked IAP) and other IAP molecules to inhibit active caspases [47,48].

Granzyme B-induced caspase activation

The third major pathway to caspase activation that has been characterized in detail is initiated by the constituents of cytotoxic granules that are released by CTLs (cytotoxic T-cells) and NK (natural killer) cells upon encounter with transformed or virally infected target cells [49]. CTLs and NK cells induce apoptosis through the concerted action of the effector molecules contained within their cytotoxic granules that trigger caspase activation in the target cell [49]. These granules contain components such as perforin, a pore-forming protein that facilitates the delivery of the other granule components into target cells, and granzyme B, a serine protease that cleaves after aspartic acid residues. Perforin and granzyme B alone are sufficient for the induction of apoptosis [50]. Granzyme B, similarly to the caspases, cleaves its substrates after aspartic acid residues [51,52], suggesting that this protease has the ability to activate members of the caspase family directly. Indeed, caspase 3 was the first substrate identified for granzyme B [53,54]. The cohort and activation cascade of caspases activated in the granzyme B-mediated pathway has been elucidated [55]. Caspases 3, 7, 8 and 10 are directly processed by granzyme B whereas caspases 2, 6 and 9 are processed in a second, caspase-3-dependent, wave of processing [55]. Because granzyme B cleaves and activates caspases directly, addition of caspase-specific inhibitors might be expected to inhibit granzyme B-mediated cell death. However, although incubation of Jurkat cells with the broad-spectrum caspase inhibitor zVAD-fmk (benzyloxycarbonyl-Val-Ala-DL-Asp-fluoromethylketone) reduced DNA fragmentation in response to granzyme B and perforin, no long-term protection was evident, indicating that granzyme B has other non-caspase targets [56]. This is not entirely surprising when we consider that many viruses encode caspase inhibitors, such as Crm A or p35, as a means to aid their survival and replication in host cells. Therefore, by incorporating a caspase-independent route to cell death, granzyme B has the ability to overcome these viral defences.

Subsequent studies demonstrated that overexpression of Bcl-2 clonogenically protected cells against granzyme B/perforin-mediated apoptosis [5658]. Protection by Bcl-2 implied that granzyme B may also act upstream of mitochondria. Indeed, the BH3-only Bcl-2 family member Bid was subsequently identified as a target of granzyme B [59]. Cleavage of Bid by granzyme B generates a truncated fragment which translocates to mitochondria, initiating the release of intermembrane space proteins such as cytochrome c (Figure 2). Both granzyme B and caspase 8 target and cleave Bid, albeit at distinct sites. Overexpression studies with a mutant, granzyme B-resistant, Bid (D75E) demonstrated inhibition of granzyme B and perforin-induced apoptotic features, whereas overexpression of D59E Bid (mutated at the caspase 8 site) failed to abolish the apoptotic phenotype [60]. The latter result illustrates that granzyme B cleaves Bid directly rather than doing so indirectly by caspase-8-mediated cleavage.

Several studies indicate that Bid, rather than caspases, is the preferential substrate for human granzyme B, with Bid cleavage being evident within minutes of granzyme B entry to the target cell. Upon entry into the target, granzyme B rapidly induces mitochondrial permeabilization, via Bid, and release of intermembrane space proteins leading to apoptosome assembly and caspase activation (Figure 2). Concurrently, granzyme B can also target and cleave caspases directly, thereby amplifying the level of caspase activation. However, recent studies have shown that mouse Bid is a poor substrate for murine granzyme B. Therefore, in the mouse context, granzyme B may kill in a more caspase-dependent way than in humans [61,62].

Enigmatic caspase activation pathways

Although the caspase activation cascades discussed above have been confirmed by many laboratories, additional caspase-dependent pathways to apoptosis have been proposed, but have yet to be widely substantiated.

Endoplasmic reticulum stress-associated caspase activation

Accumulation of misfolded proteins and alterations in calcium homoeostasis in the endoplasmic reticulum provokes endoplasmic reticulum stress and commonly results in cell death. However, the signal-transducing events that connect endoplasmic reticulum stress to known caspase activation pathways are incompletely understood. Early work proposed caspase 12 as the initiator caspase in endoplasmic reticulum stress-induced apoptosis as mouse embryonic fibroblasts from Casp-12-deficient animals displayed modest resistance (approx. 20%) to the endoplasmic reticulum stress-inducing agents brefeldin A and tunicamycin [63]. However, recent studies using mouse Casp-12 knockout cells, from a different source, have cast doubt upon this claim. Saleh et al. [64] reported that caspase 12, rather than being implicated in apoptosis, functions in pro-inflammatory responses as a negative regulator of caspase 1 activation within the inflammasome. To date, no substrates for mouse caspase 12, aside from caspase 12 itself, have been reported. Indeed the importance of caspase 12 for endoplasmic reticulum stress-induced apoptosis has been undermined further by the discovery that the majority of humans lack full-length caspase 12. A frameshift mutation, producing a premature stop codon, is present in most human populations, resulting in the production of a short CARD-only protein [65]. Certain individuals of African descent lack this mutation and are able to produce full-length and presumably active caspase 12 [66]. Studies examining the outcome of caspase 12 expression in this subset of the population determined that expression of full-length caspase 12 correlated with increased susceptibility to severe sepsis [66]. Collectively, these data argue that caspase 12 processing does not trigger endoplasmic reticulum stress-induced caspase activation, but is involved as a negative regulator of inflammatory responses.

Additional data indicate that endoplasmic reticulum stress-induced apoptosis is dependent upon mitochondrial-mediated processes to promote caspase activation. Processing of pro-caspases 9, 3, 6 and 7 in response to endoplasmic reticulum stress has been reported, indicative of the Apaf-1 pathway to caspase activation. Moreover, studies using cells devoid of a functional mitochondrial pathway (Bax/Bak-null or Casp-9-null mouse embryonic fibroblasts), or overexpressing anti-apoptotic Bcl-2, fail to activate caspases in response to endoplasmic reticulum stress signals [67]. These observations indicate that endoplasmic reticulum stress-induced caspase activation is dependent on the intrinsic or mitochondrial pathway. As yet, the signalling pathways employed by the endoplasmic reticulum to trigger cytochrome c release have not been delineated, but are most likely to occur by regulation of BH3-only members of the Bcl-2 family. Indeed, the BH3-only protein, Bim, has recently been implicated in endoplasmic reticulum stress-induced apoptosis: abolition of Bim expression correlated with decreased endoplasmic reticulum stress-induced cell death [68]. Further work is required to clarify the role of Bcl-2 family members in regulating endoplasmic reticulum stress-induced apoptosis.

Caspase 2 as an initiator caspase

Although one of the first apoptotic caspases to be discovered, the precise role of caspase 2 remains enigmatic. Structurally, owing to the presence of an N-terminal CARD within its pro-domain, caspase 2 is classed as an initiator caspase. Studies profiling the proteolytic activity of caspase 2 suggest that this caspase cannot process any other member of the caspase family, but can cleave the Bcl-2 family member Bid, presumably to instigate cytochrome c release [69]. Indeed, several groups have reported caspase 2 activation before cytochrome c release in response to genotoxic stress, while others have only observed that this caspase is activated in the context of heat-shock-induced cell death [6973]. Although unable to cleave other caspases directly, caspase 2 may initiate apoptosis by harnessing the mitochondrial pathway. If caspase 2 can function in this manner, what mechanism facilitates its initial activation? On the basis of our knowledge of caspases 8 and 9, an activation platform akin to the apoptosome or DISC (death-inducing signalling complex) would be expected. The formation of high-molecular-mass caspase 2 complexes was first observed in 2002, and subsequent studies have reported that RAIDD {RIP (receptor-interacting protein)-associated ICH-1 [ICE (interleukin-1β-converting enzyme)/CED-3 (cell-death determining 3) homologue 1] protein with a death domain}, an adaptor protein, and PIDD (p53-inducible protein with a death domain), a p53-inducible scaffold protein, are components of a caspase-2-activation platform commonly referred to as the ‘PIDDosome’ [72,74]. Direct death domain interactions between RAIDD and PIDD and CARD–CARD interactions between RAIDD and caspase 2 are thought to enable PIDDosome formation [72]. Recent elucidation of the crystal structure of the PIDDosome suggest that it comprises five PIDD and seven RAIDD molecules, enabling the recruitment and presumably the activation of seven caspase 2 molecules [75].

A number of problems need to be addressed before the precise role of caspase 2 in apoptosis can be determined. The most troubling of these is the fact that Casp-2-null mice are essentially normal, suggesting that caspase 2 is unlikely to be essential for stress-induced apoptosis [76,77]. Indeed, lymphocytes and neurons derived from Casp-2-null mice retain normal sensitivity to pro-apoptotic stimuli in vitro [76,77]. The latter results contrast sharply with the cell death-defective pheonotypes of Apaf-1- and Casp-9-null animals. This anomaly may be a consequence of redundancy between caspase 2 and another caspase, with caspase 9 being the most likely candidate. However, the generation of Casp-2/Casp-9-double knockouts suggests that this is not the case, as these animals exhibited a phenotype strikingly similar to those lacking caspase 9 alone [78]. Secondly, caspase 2 and caspase 9 do not exhibit the same enzymatic activity towards substrates, rendering a functional overlap between these caspases unlikely. Thirdly, in the majority of stress-induced apoptosis pathways, pro-caspase 2 processing appears to be dependent on caspase 3, which would place caspase 2 activation downstream of Apaf-1/caspase 9 (Figure 2) [31]. As further evidence of this, cells from Apaf-1- or Casp-9-null animals also fail to process caspase 2, again arguing that caspase 2 activation occurs downstream of apoptosome assembly. It is possible that caspase 2 functions as an apical caspase under very specific circumstances or in particular cell types. Obviously, further studies are required to reconcile the disparities concerning the role of caspase 2 in apoptosis.


Apoptosis is essential for the removal of aged, damaged or infected cells from the body. Too much or too little apoptosis is associated with disease, therefore activation of caspases must be a tightly regulated process. To date, the Apaf-1 apoptosome, granzyme B and death receptor-initiated pathways are the best understood routes to caspase activation. We currently have a sound understanding of the basic mechanisms controlling these pathways, and through the development of new techniques are starting to unravel the fine details of these processes. A major unresolved challenge in the field is understanding how caspases co-ordinate the terminal events of apoptosis that culminate in the death of the cell and recognition by phagocytes. In particular, how active caspases promote the membrane alterations that trigger recognition of apoptotic cells by phagocytes remains totally obscure. The recent identification of numerous additional caspase substrates using various proteomic approaches should hopefully provide useful clues in this regard.

GlaxoSmithKline Award Lecture GlaxoSmithKline Award Lecture Seamus Martin


We acknowledge Science Foundation Ireland (PI1/B038), the Wellcome Trust, the Health Research Board of Ireland and IRCSET (Irish Research Council for Science, Engineering and Technology) for financial support of ongoing work in our laboratory.


  • GlaxoSmithKline Award Lecture:

Abbreviations: Apaf-1, apoptotic protease-activating factor 1; BH, Bcl-2 homology; CAD, caspase-activated DNase; CARD, caspase recruitment domain; CTL, cytotoxic T-cell; IAP, inhibitor of apoptosis protein; ICAD, inhibitor of CAD; NK, natural killer; PIDD, p53-inducible protein with a death domain; PUMA, p53 up-regulated modulator of apoptosis; RAIDD, RIP (receptor-interacting protein)-associated ICH-1 [ICE (interleukin-1β-converting enzyme)/CED-3 (cell-death determining 3) homologue 1] protein with a death domain; ROCK, I, Rho-associated kinase I; tBid, C-terminal fragment of Bid; TNF, tumour necrosis factor


View Abstract