Introduction

In 1998, we published a list of caspase substrates comprising 65 different proteins that were cleaved by proteases of the caspase family.1 Most of the substrates known at that time could be categorized into a few functional groups, including proteins involved in scaffolding of the cytoplasm and nucleus, signal transduction and transcription-regulatory proteins, cell cycle controlling components and proteins involved in DNA replication and repair. Since then, the number of caspase substrates has considerably increased, more recently in particular because of a systematic proteome analysis of apoptotic cells.2,3,4 To date, more than 280 caspase targets are identified. Various methods have been employed to search for caspase substrates, including direct cDNA pool expression strategies or two-hybrid cloning approaches.5,6 By comparative two-dimensional (2D) gel electrophoresis of healthy and apoptotic cells, often a few hundred altered protein spots can be detected. Although not all of them have been confirmed as caspase targets, such proteomic approaches will certainly lead to the identification of numerous additional substrates in the near future (Table 1).

Table 1 List of known caspase substrates

Already now, a bewildering number of substrates are cleaved by caspases. However, it should be kept in mind that some proteins might be cleaved very late and less completely during apoptosis, or not in all cell types. For example, it has been reported that β-actin can be cleaved by caspases in pheochromocytoma and ovarian carcinoma cells,7,8 whereas in many other cell types no cleavage was detected.9 Thus, it is possible that certain protein cleavages are cell type-specific, which may be because of variations in the expression of individual caspases. Also, caspase cleavage sites are not always conserved in different species. For instance, cyclin A is cleaved during apoptosis of Xenopus oocytes,10 but the caspase cleavage site is not present in homologues of mammalian cells. Some proteins, such as DNase-X, contain one or more classical cleavage sites in their sequence. However, the protein is virtually not cleaved inside apoptotic cells despite massive caspase activation.11 Moreover, in some cases, a first cut by caspases unleashes additional cleavage sites for other types of proteases. Cleavage of acinus, for instance, by caspase-3 is necessary but not sufficient to activate its DNA-condensing activity. For full activation, an additional, still unknown serine protease has to intervene. Only the combined action of both proteases generates the mature fragment, which, when added to purified nuclei, causes chromatin condensation.12

For many of the identified substrates, the functional consequences of their cleavage are unknown and have only been inferred from their normal functions. In other cases, the role of caspase cleavage has been experimentally assessed by expressing substrate proteins that have mutant caspase cleavage sites or by expressing protein fragments of the caspase-cleaved products. Given the high conservation of the apoptotic phenotype, from worms to mammals, it is highly likely that a conserved group of crucial caspase substrates exist. Proteolysis of the latter substrates presumably leads to the stereotypical destructive alterations that we call apoptosis.

The search for caspase substrates has brought several major questions into focus. For instance, is there a critical death substrate or what is the minimal set of proteins that must be cleaved in order to induce the phenotypic hallmarks of apoptosis? How is caspase substrate cleavage coordinated with other cellular processes, such as removal of dead cells, or presumably unrelated events including cell proliferation and differentiation? Although the significance of cleavage is not well understood for many substrates, the intense study of caspase substrates has recently shed some light on these questions. Here, we discuss several topics that have emerged from the accumulating knowledge regarding the role of caspase substrates in different biological processes.

Key morphological alterations are determined by caspase substrate cleavage

For most proteins, the consequences of their cleavage are poorly understood. In a few cases, however, proteolysis of certain components can be linked to discrete morphological changes of cell death. A classical example is the DNase inhibitor ICAD. Cleavage of ICAD by caspase-3 liberates the active CAD nuclease that mediates apoptotic DNA fragmentation (for references, see Table 1). In addition, the cleavage of acinus and helicard, a DNA helicase, contributes to chromatin condensation and nuclear remodeling. The cleavage of several other substrates, including gelsolin as well as the kinases ROCK-1 and PAK2, has been implicated in membrane blebbing, a classical morphological feature. Gelsolin is cleaved by caspase-3 to generate a constitutively active fragment that can depolymerize F-actin. Gelsolin-deficient neutrophils exhibit greatly delayed membrane blebbing during apoptosis, implying that membrane blebbing requires actin reorganization mediated by caspase-activated gelsolin. Caspases also cleave and thereby activate ROCK-1 leading to the phosphorylation of myosin light chains, which finally results in membrane blebbing.

Caspases destroy several proteins involved in maintenance of the cytoskeletal architecture such as the intermediate filaments cytokeratin-18 and vimentin, or Gas2 and plectin, two proteins involved in filament organization. These cleavages may directly contribute to apoptotic changes in cell shape. Caspases attack targets of the cortical actin network such as fodrin, and several components of the focal adhesion complex which links cortical actin filaments and membrane proteins to the extracellular matrix. Examples of this kind of substrates are focal adhesion kinase, Cas or paxillin. Cleavage of these proteins presumably contributes to cell shrinkage and cell detachment and, importantly, will interrupt antiapoptotic integrin signaling. A large percentage of caspase substrates are involved in cell adhesion or mediate cell–cell commmunication in adherens and gap junctions, or in desmosomes. Examples are β-catenin, E-cadherin, plakoglobin or desmoglein.

In the course of apoptosis, disruption of the endoplasmic reticulum and Golgi apparatus also takes place. Cleavage of golgin-160 and GRASP65 was suggested to cause disassembly of the Golgi complex, and proteolysis of Bap31 disrupts the transport between the ER and the Golgi complex. During apoptosis, vesicle transport processes are also impaired, for instance by the cleavage of rabaptin-5 or kinectin.

Caspases initiate the destruction of the nucleus where a huge variety of different proteins are cleaved. By 2D gel electrophoresis it has been recently determined that approximately 70 nuclear matrix proteins are consistently degraded or translocated during apoptosis, irrespective of the cell type or apoptotic stimulus.13 Many cleavages lead to nuclear lamina disassembly, and the cleavage of several components of the nuclear pore results in impaired nuclear transport. Inhibition of DNA repair, for instance by the cleavage of PARP-1 or the kinases ATM and DNA-PK, has been long thought to promote the apoptosis process. Other targeted factors are involved in DNA synthesis and replication, such as DNA polymerase Pol ɛ, MCM3 or replication factor RFC140. In addition, various proteins that bind to chromatin, and either fulfill a transcriptional role or have structural functions in the nuclear matrix, are destroyed. In almost all cases, these cleavages result in the generation of proteins that are no longer able to bind to DNA or to stabilize chromatin in the nuclear matrix. With a few exceptions that are discussed below, virtually all pathways of macromolecular synthesis are impaired by caspases. Cleavage of RNA helicase A and multiple splicing factors, including U1 70-kDa snRNP and at least eight different heterogeneous nuclear ribonucleo-proteins (hnRNPs), leads to a general shut-off of RNA synthesis, processing and transport. Moreover, protein synthesis is blocked either by the inactivation of translation initiation factors, including eIF2α, eIF3 and eIF4G proteins, or by the activation of PKR kinase that blocks protein synthesis through eIF2-α phosphorylation.

Caspase substrates in signal transduction

A tremendous variety of proteins involved in signal transduction are cleaved by caspases. The proteolytic cleavage can either lead to the functional inhibition or to the activation of these mediators. In some cases, it has been established that caspase-mediated activation of these molecules is involved in transduction and amplification of the apoptotic signal. Caspases turn off cell-protective mechanisms and activate pathways that lead to cell destruction. Classical apoptosis inhibitors that are cleaved by caspases are Bcl-2 proteins or the caspase-8 inhibitor c-FLIP. The cleavage of Bcl-2 and Bcl-xL resulting in the removal of the N-terminal BH4 domain not only leads to a loss of their antiapoptotic function, but even converts them to proapoptotic proteins. Similarly, during death receptor-mediated apoptosis caspase-8 cleaves the Bcl-2 member Bid generating an active C-terminal fragment that induces the proapoptotic release of cytochrome c from mitochondria. The conversion of antiapoptotic into proapoptotic regulators constitutes a positive feedback loop in the terminal phase of apoptosis, removing apoptotic inhibitors and promoting caspase activation. It is interesting to note that certain viral Bcl-2 proteins can also be cleaved by caspases, but in these cases no proapoptotic fragments are generated.

Several kinases and transcription factors with antiapoptotic activity are inactivated during apoptosis. Akt and Raf-1 provide two examples of antiapoptotic kinases that are cleaved by caspase-3. As both kinases can inactivate proapoptotic molecules such as Bad, their degradation presumably constitutes a positive feedback loop in apoptosis. Antiapoptotic transcription factors inhibited by caspases include the cAMP-responsive factor CREB, heat-shock factor HSF-1 and NF-κB. The NF-κB pathway is a paradigm of how caspase cleavage may result in a complete loss of the transcription factor's antiapoptotic function: (i) Cleavage of NF-κB subunit p65 (RelA) generates a dominant-negative fragment that is still able to bind to DNA but looses its transactivating activity, and therefore functions as a dominant-negative inhibitor. (ii) The NF-κB inhibitor IκB-α is normally inducibly degraded by the proteasome. The N-terminal cleavage of IκB-α by caspases generates a constitutive super-repressor that can no longer be removed by the proteasome. (iii) The cleavage of the adapter proteins TRAF-1 and RIP-1 that are involved in receptor-mediated pathways also contributes to impaired NF-κB activation and antiapoptotic capacity. Thus, cells have elaborate mechanisms in order to interrupt antiapoptotic signaling efficiently.

While some substrates are functionally inactivated upon caspase-mediated cleavage, other proteins and enzymes can be activated, mostly by removing an inhibitory or regulatory domain within the caspase target. The physiological consequence of this gain-of-function cleavage for apoptosis remains mostly unclear. Several members of the PKC family and MAP kinase pathway are constitutively activated by the separation of an N-terminal regulatory and the C-terminal catalytic domain. Examples are the p21-activated kinase PAK2 as well as ROCK-1. As described above, activation of PAK2 and ROCK-1 is important for cytoskeletal reorganization and plasma membrane blebbing. In the case of MEKK1, expression of the caspase-cleaved kinase fragment induces caspase activation, thereby providing a positive feedback loop for apoptosis. Epithelial cells undergo apoptosis if they are detached from the basement membrane, a process called anoikis. MEKK1 is activated following cell detachment, and blockade of either MEKK1 or caspase activity blocks anoikis. Cleavage of several MST kinases by caspase-3 also yields constitutively active molecules and potent inducers of apoptosis. Apoptosis induction by all these upstream kinases in the SAPK/JNK pathway may be explained in part by their ability to activate JNK, which then phosphorylates and inactivates Bcl-2.

Most kinase pathways exert antiapoptotic functions. It is thus not unexpected that a major cellular protein phosphatase, PP2A, which counteracts the survival function of kinases, is activated by caspases. Protein phosphorylation can also protect caspase substrates from proteolysis. This has been convincingly demonstrated for Bid that is protected from caspase-8 cleavage through phosphorylation by casein kinases I and II.14 Another example is Max, a transcription factor in the c-Myc network, which can be cleaved only if dephosphorylated. A very intriguing finding has been recently made for C/EBPβ. The transcription factor itself is not cleaved by caspases, but curiously acts as caspase inhibitor upon phosphorylation.15 Threonine phosphorylation of C/EBPβ within a KTVD sequence creates a noncleavable mimic of an XEXD cleavage site, which binds caspases and thereby inhibits caspase action. Hence, such dummies of caspase substrates may represent a novel survival mechanism.

Some peculiarities of substrate cleavage

Caspase cleavage can also result in the cellular redistribution and dislocation of signaling mediators. In some cases, such as the Grb2 adapter protein GrpL or the phosphodiesterase PDE4A5, an SH3-domain within the substrate is removed causing its inability to bind to physiological interaction partners. A change of subcellular localization following caspase cleavage has also been observed for the kinases Fyn and MEKK1. Another notable example is Bid. Upon cleavage by caspase-8, the proapoptotic p15 fragment of Bid undergoes post-translational rather than the classical cotranslational N-myristoylation at a glycine residue that becomes newly exposed by the cleavage.16 This postproteolytic N-myristoylation then enables Bid to target mitochondria and serves as an activating switch, which strongly enhances cytochrome c release.

Apoptosis is generally associated with a shut-down of cap-dependent protein translation, which is mediated by caspase cleavage of several translation factors. Interestingly, it has been recently observed that during apoptosis, translation of a subset of mRNAs prevails. The reason for this is presumably a switch from cap-dependent to internal ribosome entry site (IRES)-mediated protein translation. DAP-5, a member of the eIF4G family, is activated by caspases and stimulates translation from the IRES sites of c-Myc, Apaf-1, and its own mRNA.17 Thus, DAP-5 is a rather unique caspase-activated factor that supports cap-independent translation of apoptosis-related proteins and thereby may amplify the apoptosis cascade.

Most caspase substrates identified so far are cleaved by caspase-3. This has been convincingly shown in the system of MCF-7 breast carcinoma cells that lack caspase-3, and caspase-3 re-expressing derivatives.18 Nevertheless, several substrates that are efficiently cleaved by caspase-3 can also be targeted by caspase-7, suggesting an at least partial redundance of both caspases. Caspase-7 activity is upregulated in cells of caspase-3-deficient mice, where it might compensate for the loss of caspase-3. Caspase-7 and -5, but not caspase-3, cleave transcription factor Max. Interestingly, in this case Max is not cleaved at the classical aspartate residue in the P1 position, but at an unusual glutamate residue.19 Cleavage of the cytoplasmic tail of TNF-R1, the cardiac myosin light chain vMLC and connexin 45.6 at a glutamate instead of an aspartate residue are further examples. Cleavage at these noncanonical sites suggests that the specificity of caspases may in fact be broader than generally thought. Also, the Drosophila caspase DRONC can cleave substrates following glutamate residues.20 Caspase-7 not only cleaves substrates at atypical motifs, but can be activated itself by a rather unusual processing event. It has been reported that various serine proteases can trigger the proteolytic activity of the caspase-7 zymogen.21 For instance, cathepsin G activates caspase-7 by cleaving at a glutamate bond, indicating that the cleavage specificity at aspartic acid is not strictly required for caspase activation.

The interaction of caspases with other classes of proteases, including calpains, cathepsins or the proteasome, is poorly understood. When searching for caspase substrates, it must be considered that high concentrations of caspase inhibitors, such as the fluoromethylketone zVAD-fmk, are less specific than often anticipated, because calpains are inhibited as well. Several substrates of caspases are also cleaved by calpains including structural proteins, such as fodrin, keratins and β-actin, and proteins involved in signal transduction, such as Bid, Bax, focal adhesion kinase and many others. It has been found that caspases and calpains interfere with each other, resulting in mutual protease activation. Caspases can indirectly activate calpain by cleavage and inactivation of its inhibitor calpastatin, and thereby turn on downstream events leading to cellular destruction. However, it is still controversial as to whether calpains function upstream or downstream of caspases. It has also been reported that calpains cleave procaspases to generate proteolytically inactive caspase fragments.22

Caspases are not only involved in apoptosis but also in the induction of inflammation. In fact, the former notion that apoptosis and inflammation are exclusive processes should be replaced, as both processes are linked at various levels. Caspase-1 processes and maturates the cytokine precursors pro-IL-1β and pro-IL-18, also known as IFN-γ-inducing factor. Although caspase-1 is required mainly for induction of inflammation, it can process the effector caspases-3, -6 and -7 and may initiate apoptosis under certain conditions. Effector caspases can also activate pro-IL-16 and pro-EMAP-II, an endothelial-monocyte-activating polypeptide. This precursor of EMAP-II is an intriguing substrate, because it exerts a dual function:23 Pro-EMAP-II is identical to the p43 cofactor of the aminoacyl-tRNA synthetase complex. After cleavage, preferentially by caspase-7, its t-RNA binding capacity is lost and protein translation is blocked. The translation arrest is accompanied by the release of the EMAP-II cytokine that may play a role in the engulfment of apoptotic cells by phagocytes. Caspase-mediated substrate cleavage therefore has multiple effects summarized as (i) a halt of cell cycle progression, (ii) disabling of repair mechanisms, (iii) disassembly of molecular structures, (iv) cell detachment, and (v) maturation of cytokine precursors.

Substrate cleavage at the balance between necrosis and apoptosis

Although caspases are presumably not essential for necrotic cell death, recent evidence suggests that the cleavage of certain substrates may determine the form of cell death. One of the first death substrates found to be cleaved by caspases was PARP-1, which catalyzes the transfer of ADP-ribose polymers to nuclear proteins and thus presumably facilitates DNA repair.24 Owing to its role in DNA repair, it was originally hypothesized that the cleavage of PARP may lead to lethal DNA damage and compromise most of its DNA repair activity, and thus may contribute to the demise of the cell. However, PARP(−/−) mice neither reveal a phenotype which would indicate a crucial role in apoptosis nor is the sensitivity towards CD95- and TNF-R1-mediated apoptosis affected.25 Thus, cleavage of PARP may be a characteristic event, but is presumably dispensable for most apoptotic pathways.

New evidence, however, suggests that PARP inactivation by caspase-3 is important for turning off an energetically expensive DNA repair pathway and for maintaining ATP levels that are required for the execution of apoptosis. PARP is rapidly activated during oxidative stress and DNA damage. Activated PARP then transfers more than 100 ADP-ribose moieties to each acceptor site in target proteins, and each cycle of ADP-ribosylation is coupled with consumption of one NAD molecule, which is metabolically equivalent to four ATP molecules. Hence, it can be imagined that excessive activation of PARP will quickly deplete cellular energy stores. In the absence of an energy pool sufficient to execute apoptosis or to maintain ionic homeostasis, cells can die quickly by necrosis. Indeed, when cells engineered to express caspase-resistant PARP are treated with apoptotic stimuli, they undergo extensive necrosis instead of apoptosis.26 Consistent with the requirement of maintaining cellular energy during apoptosis, cells artificially depleted of ATP undergo necrosis instead of apoptosis under conditions that would normally trigger caspase activation.27 Thus, cleavage of PARP prevents depletion of the cellular energy needed for apoptosis and thus may function as a molecular switch between apoptotic and necrotic cell death. Similar to PARP, also the cleavage of other substrates may provide a link between apoptosis and necrosis. For instance, cleavage and inactivation of the plasma membrane calcium ATPase PMCA-4, which removes calcium from the cytosol, disturbs ion homeostasis.28 The subsequent cellular calcium overload may be responsible for the secondary necrosis that is observed in the late stages of apoptosis.

Role of caspase substrates in disease progression

Increased caspase activation has been recently demonstrated in various diseases. However, the cleavage of several substrates may not only contribute to increased tissue damage, but may also play an active role in disease progression. Such a direct role of substrate cleavage has been most intensively studied in neurodegeneration and autoimmune diseases. Autoimmunity to intracellular proteins has been identified as an important factor in autoimmune diseases. Massive apoptosis or defective clearance may lead to an accumulation of apoptotic cells that concentrate caspase-cleaved proteins in their apoptotic bodies and membrane blebs. The presence of autoantibodies against caspase substrates, such as lamins, fodrin, DNA-PK, PARP or NuMA, has been demonstrated in several autoimmune diseases.29 Cleavage of these autoantigens presumably enhances their immunogenicity by exposing cryptic neoepitopes. The cleaved proteins are then processed and presented by dendritic cells to circulating autoreactive T cells, triggering an autoimmune response.

The cleavage of specific substrates can be directly linked to the pathogenesis of certain neurodegenerative disorders. Huntington's disease, a genetically determined neurodegenerative disease, results from the expansion of CAG triplets at the 5′-primed end of the gene encoding huntingtin, a protein with a long polyglutamine stretch. Huntingtin is cleaved by caspase-3 and results in an N-terminal fragment, which aggregates and forms nuclear inclusions that are directly cytotoxic for neurons.30 Huntington's disease manifests only if huntingtin exceeds 35 glutamine residues. Because the rate of caspase cleavage of huntingtin correlates with the length of the polyglutamine stretch, accumulation of the fragment may cause a vicious cycle. A pathogenic role of caspase cleavage has also been implicated in other neurodegenerative disorders. Similar to huntingtin, the polyglutamine tract proteins atrophin-1, androgen receptor and ataxin-3 are caspase substrates. Indeed, mutations of the caspase recognition sites in atrophin-1 and androgen receptor abrogate their cytotoxicity in vitro.

Alzheimer's disease is characterized by brain lesions of neurofibrillary tangles, and senile plaques built of aggregates of the β-amyloid peptide. Aggregates of β-amyloid peptide induce neuronal apoptosis, and increased production of β-amyloid peptide has been postulated as an important pathologic mechanism in early-onset familial Alzheimer's disease. Effector caspases presumably increase β-amyloid production by several mechanisms. Loss-of-function mutations in the presenilin-1 and -2 genes are responsible for the majority of familial Alzheimer's disease and are thought to increase β-amyloid production. Caspase-3 can cleave and inactivate presenilins, which may mimic the effect of pathologic presenilin mutations. The 40- to 42-amino-acid β-amyloid peptide is derived from proteolytic processing of the amyloid precursor protein (APP) at two sites by the β- and γ-secretase. Caspase-3 cleaves APP at a site different from the γ-secretase site.31 Nevertheless, the N-terminal caspase cleavage product of APP strongly facilitates the production of β-amyloid peptide, and appears itself to be a component of senile plaques found in Alzheimer patients. Because caspase-3 activation and APP cleavage are also induced in vitro after ischemic brain injury, a risk factor for Alzheimer's disease, these results provide another example of a positive feedback loop between caspase substrate cleavage and neurodegeneration. Neuronal apoptosis from ischemia or other causes activates caspase-3 and stimulates APP cleavage, which increases the propensity for β-amyloid peptide production. In turn, increased extracellular β-amyloid peptide production may induce neuronal apoptosis, leading to further deposition of senile plaques. The cytotoxic properties of their cleavage products illustrate that specific caspase substrates are not only involved in cell destruction, but also fulfill an active role in the exacerbation of disease processes.

Caspases: more than just killers?

A strikingly large number of caspase targets are involved in cell cycle regulation. This has led to speculations that caspases are not only involved in cell death but also in proliferative events.32 Supportive, yet indirect evidence for a role of caspases in cell growth is the observation that proliferation of primary T cells is inhibited by cell-permeable caspase inhibitors.33,34 Moreover, interference with pathways leading to caspase processing, as in FADD-deficient or Bcl-2-transgenic mice, also results in impaired mature T-cell proliferation.

Several negative regulators including Wee1, an inhibitor of the cell cycle-regulatory kinases CDK2 and CDC2, as well as CDC27, a component of the anaphase-promoting complex, are cleaved by caspases. Wee1 is a critical component of the G2/M cell cycle checkpoint machinery and mediates cell cycle arrest by phosphorylation of CDC2. Therefore, cleavage of Wee1 in proliferating lymphocytes could lead to its inactivation, thus allowing cell cycle progression. Of note, Wee1 processing by caspases during apoptosis in Jurkat T cells correlated with a strong decrease in Wee1 activity and an increase in CDC2 activity.35 Moreover, the cyclin inhibitors p21Waf1 and p27Kip1 are targeted by caspases resulting in increased CDK2 activity that could allow cell cycle progression.

If caspases are activated during mitosis, a critical question is then, how could caspase cleavage be restricted to those cell cycle regulators, while leaving other vital proteins intact? The answer could lie in a specific subcellular compartmentalization of caspases, the existence of scaffold proteins or a different accessibility of cleavable substrates. Some caspases are translocated to a certain organelle during activation, and in some cell types certain caspases have been localized in the nucleus. Interestingly, it has been found that, although caspases were activated and Wee1 was cleaved after mitogenic T-cell stimulation, neither DNA replication factor RFC140 nor ICAD were cleaved in proliferating T cells.33 Cleavage of RFC140 and ICAD would lead to inhibition of DNA replication and fragmentation of genomic DNA, events that are not compatible with cell proliferation. Thus, selective substrate processing could explain why nonapoptotic cells survive and proliferate despite caspases being activated.

Certainly, there exist many links, also at the morphological level, between the processes of cell death and proliferation. However, it must be emphasized that the view of a potential involvement of caspases in proliferation is largely based on indirect evidence and therefore remains highly speculative. Because cleavage of cell cycle regulators occurs late in apoptosis by caspase-3-like activities in parallel with the dismantling of the transcription and translation machinery, caspase activation cannot trigger the normal mitotic program. For example, mitotic spindles do not form in apoptotic cells, distinguishing apoptosis from a mitotic catastrophe.

Limited substrate cleavage in terminal differentiation and hematopoiesis

In contrast to the rather speculative involvement of caspases in proliferation, there is an increasing body of evidence suggesting that caspases might act in cellular differentiation. A physiological role of caspases in this process has first been suggested for keratinocytes and lens fiber cells, in which the characteristic enucleation of the cells could be regarded somehow as a caspase-mediated incomplete apoptotic process.36,37 Caspases have also been implicated in erythropoiesis, because caspase inhibitors suppressed the nuclear extrusion process and consequent erythrocyte formation.38 Furthermore, caspase activation can be detected during thrombopoiesis and the fragmentation of proplatelets from megakaryocytes, without a concomitant induction of cell death.39 Both the incubation with peptide caspase inhibitors and the overexpression of Bcl-2 blocked proplatelet formation. Interestingly, in transgenic mice overexpressing Bcl-2 under the control of a hematopoietic cell-specific promoter, also a reduction in platelet formation is found, whereas the number of megakaryocytes remains unchanged. Finally, caspases might be required for differentiation processes also of nucleated cells such as macrophages and muscle cells. Elevated caspase activation is detectable in monocytes when they undergo M-CSF-stimulated macrophage differentiation.40 This is not only prevented by pharmacological caspase inhibitors, but also by the overexpression of Bcl-2 and p35. In myoblasts, homologous deletion of caspase-3 leads to a dramatic reduction in myofiber formation and decreased expression of muscle-specific proteins.41 Thus, all these lines of evidence suggest that caspases are not only required for cell death processes, but might also be capable of regulating nonapoptotic functions in certain cell types.

It is obvious that differentiation-related caspase activation must be tightly regulated to prevent cells from dying by apoptosis. During cellular differentiation, caspase activation is apparently either very limited, transient or localized. For instance, during megakaryocyte differentiation, the limited caspase activation is confined to dot-like structures.39 When senescent megakaryocytes die, however, caspase activation switches from a localized to a diffused and largely increased cytosolic activation. Also, little is known about the proteins cleaved by caspases during differentiation processes. Only a limited number of distinct substrates seem to be cleaved. For instance, in erythroblasts cleavage of PARP, lamin B and acinus was found, while the ICAD and GATA-1, a transcription factor essential for erythrocyte formation, remained intact. Interestingly, MST1 kinase was identified as a crucial caspase-3 effector in myoblast differentiation.41 As mentioned above, MST1 is cleaved and activated by caspase-3, and serves to enhance the activity of downstream MAP kinases that promote skeletal muscle differentiation. Expression of the truncated active kinase restored the differentiation phenotype in caspase-3 deficient myoblasts.

As discussed above, it remains currently unexplained as to how caspases could selectively cleave some targets without cleaving others. The compartmentalization of caspases, the duration of the caspase signal, or the coordinated expression of antiapoptotic molecules might play a role in the selectivity of caspase cleavage. It is also conceivable that low levels of caspase activity, such as those observed in differentiating cells, are associated with protective mechanisms. For instance, it was reported that the partial cleavage of Ras-GAP, a GTPase in the Ras signaling pathway, owing to low caspase activity first generates an N-terminal fragment that is antiapoptotic by activating the PI3K pathway.42 Increased caspase levels, in contrast, result in the further cleavage of Ras-GAP into two proapoptotic fragments. Thus, caspase cleavage of intracellular target proteins may strongly depend on the cellular context including the differentiation status. Clearly, much remains to be learned about a potential dual role of caspases in apoptosis and cellular differentiation. Characterization of the molecules that regulate this limited caspase activation and the relevant substrates will certainly provide exciting new insights into processes that, beyond cell death, might link caspase cleavage to important nonapoptotic biological processes.