Noncaspase proteases in apoptosis

Abstract

Biochemical and genetic analysis of apoptosis has determined that intracellular proteases are key effectors of cell death pathways. In particular, early studies have pointed to the primacy of caspase proteases as mediators of execution. More recently, however, evidence has accumulated that noncaspases, including cathepsins, calpains, granzymes, and the proteasome complex, also have roles in mediating and promoting cell death. An important goal is to understand the importance of distinct noncaspases in various forms of apoptosis, and to determine whether pathways mediated by noncaspase proteases intersect with those mediated by caspases. In this review the roles of noncaspase proteases in the biochemistry of apoptosis will be discussed. Leukemia (2000) 14, 1695–1703.

Introduction

The demise of cells via apoptosis hinges on the activation of cellular proteases. Elucidation of the specific proteases involved in apoptotic execution began in the early 1990s with genetic and biochemical studies of apoptotic cell death in the nematode C. elegans. These studies revealed that apoptosis in C. elegans was dependent on an intracellular protease bearing considerable homology to the human interleukin-1β converting enzyme, or ICE.123 Subsequent studies have led to the identification of 13 mammalian proteases that are related to ICE, and these proteases, along with ICE, have been termed caspase proteases (caspase-1 to caspase-13).45 The term caspase, or c-aspase, reflects the fact that these enzymes are cysteine proteases that cleave substrate proteins after aspartate residues. Caspases are initially synthesized in the cell as inactive zymogens. However, in response to an apoptotic stimulus, initiator caspases such as caspase-8 or -9 undergo processing to active forms.5 Active initiator caspases then cleave and activate downstream executioner caspases such as caspase-3 or -7, thereby initiating a cascade of caspase activation. Once activated, executioner caspases cleave cellular substrate proteins promoting the destruction of the cell.

The role of caspases in mammalian apoptosis has been intensively investigated using a variety of approaches: (1) pharmacologic inhibition of caspases using peptide inhibitors; (2) inhibition of caspases with endogenous or viral inhibitory proteins; and (3) targeted gene disruption. These studies have shown that inhibition of caspases significantly delays apoptosis induced by a wide variety of stimuli in a variety of different cell types. Moreover, gene knockout experiments have shown that different members of the caspase family appear to be important for apoptosis execution in different cell types.

While the importance of caspases in apoptosis is clearly established, recent studies have indicated that several other types of proteases also may play a role in the execution process. The involvement of noncaspase proteases is suggested by observations that inhibition of caspase proteases generally causes delays, but does not fully block, cell death resulting from most apoptotic stimuli. In addition, implicated noncaspases frequently are upregulated or activated during apoptosis, and their inhibition, as with caspase inhibition, can serve to delay apoptosis. The noncaspase proteases that have been most closely linked with apoptosis are cathepsins, calpains, granzymes, and the proteasome. This review will focus on the current state of knowledge regarding the involvement and role of cathepsins, calpains, and granzymes in the execution process. The role of the proteasome complex, which has been strongly implicated in apoptosis, has been recently reviewed elsewhere.6

Cathepsins and apoptosis

The cathepsin protease family consists of at least 12 known members.78 Cathepsins can be subdivided into three distinct groups, based on the amino acid that comprises the active site residue: (1) serine proteases (cathepsins A and G), (2) cysteine proteases (cathepsins B, C, H, K, L, S, and T), and (3) aspartate proteases (cathepsins D and E). Like the caspases, cathepsins are synthesized as inactive zymogens, and activation involves proteolytic processing.910111213 As discussed below, the involvement of cathepsins in apoptotic cell death has primarily been studied in nonhematopoietic systems. Evidence from these systems, however, should provide the basis for investigating the role of cathepsins in the death of both normal and malignant hematopoietic cells.14

The most extensive evidence linking cathepsins with apoptosis has come from studies of the cysteine protease, cathepsin B, and the aspartate protease, cathepsin D. Therefore, this review will focus on these two members of the cathepsin family. Both cathepsin B and cathepsin D are found primarily in lysosomes or endosomes. Historically, it has been presumed that these proteases are mainly involved in the terminal degradation of proteins within the lysosomal compartment. However, both proteases can also be secreted, and once secreted can degrade collagen, fibronectin, laminin, and proteoglycans.915 The degradation of extracellular matrix components by cathepsins B and D accounts for the ability of these proteases to promote cell migration or malignant invasion.16171819 Indeed, overexpression and secretion of cathepsin D, which can be induced by estrogen, closely correlates with high metastatic potential in breast cancer.20212223 Additional evidence (described below) suggests that cathepsins B and D are translocated to the cytoplasm during apoptosis. This is significant, since most of the known apoptosis execution pathways occur in the cytoplasm. Thus, the translocation of cathepsins to the cytoplasm may allow these proteases to intersect with and augment other apoptosis signaling mechanisms.

Early evidence suggesting a role for cathepsins B and D in apoptosis came from studies of apoptotic cell death in rat prostate and mammary tissues. Following androgen withdrawal in prostate, or following lactation in breast, massive apoptosis occurs in the epithelial cells of these tissues.2425 Concurrent with this apoptosis, there is upregulation of mRNA and protein for both cathepsin B and D.2627 While these initial findings suggested a role for cathepsins in regressing prostate and mammary gland, they did not directly demonstrate cathepsin involvement in the execution pathways. It remained possible that cathepsins were simply needed for ‘cleanup’ in the dying cells.

More definitive experiments showing a role for cathepsins in apoptotic execution have come from studies of bile salt-induced apoptosis in cultured hepatocytes. This in vitro model mimics apoptosis that occurs in human hepatocytes in diseases where normal bile flow is impaired. In cell culture, bile salt-induced apoptosis was found to be markedly inhibited by CA-074-Me, a specific inhibitor of cathepsin B, and by pepstatin A, an inhibitor of cathepsin D.2829 Antisense-mediated downregulation of cathepsin B also markedly inhibited hepatocyte apoptosis following treatment with bile salts.28 Together, these findings showed that, similar to caspases, cathepsins B and D, can be important components of apoptosis execution pathways. Interestingly, pepstatin A was found to inhibit bile salt-induced cathepsin B activation, while CA-074-Me failed to impact cathepsin D activation.29 This revealed that during bile salt-induced apoptosis, cathepsin D is activated upstream of cathepsin B. Thus, like the caspases, cathepsins may be activated in a cascade-like fashion during apoptosis. In support of this idea, cathepsin D has been shown to directly cleave and activate cathepsin B.3031

Cathepsins B and D exert opposing effects on apoptosis in serum- and neurotrophic factor-deprived neuronal cells. During apoptosis of serum-deprived rat PC12 cells, the levels of cathepsin B protein decline while those of cathepsin D increase.32 Interestingly, cathepsin D levels are also elevated in the pyramidal neurons of Alzheimers patients.33 In PC12 cells, inhibition of cathepsin B with CA-074, or downregulation of cathepsin B with antisense oligonucleotides resulted in enhanced apoptosis.34 By contrast, inhibition of cathepsin D with pepstatin A significantly inhibited PC12 cell apoptosis. Similar results were seen using dorsal root ganglion neurons deprived of nerve growth factor.34 It will be interesting to see whether these findings can be extended to hematopoietic cells. Many hematopoietic cells are dependent on specific cytokines for survival, and undergo apoptosis when deprived of these essential factors.3536 Future studies are needed to define the role and involvement of cathepsins during cytokine withdrawal-induced apoptosis.3738

The involvement of cathepsin D in the execution pathways triggered by a variety of other stimuli has been demonstrated by Deiss et al.39 This group isolated antisense cathepsin D cDNA while screening an antisense library for clones that could inhibit IFN-γ-induced apoptosis in HeLa cells. In addition to apoptosis caused by IFN-γ, antisense cathepsin D also markedly inhibited Fas-induced apoptosis. Moreover, pepstatin A was found to inhibit both forms of cell death, as well as apoptosis induced by TNF-α treatment of U937 histiocytic lymphoma cells. IFN-γ- and TNF-α-induced apoptosis were also associated with induction and processing of cathepsin D protein.

Cathepsin D also participates in the execution of cells treated with chemotherapy drugs or radiation. Wu et al,40 using a subtractive hybridization strategy, identified cathepsin D mRNA as an upregulated transcript in doxorubicin-treated cells. Induction of cathepsin D protein was observed in ML1 leukemic cells treated with doxorubicin, etoposide, or γ-irradiation. Upregulation of cathepsin D may be impacted by p53, as two p53 binding sites were found in the promoter region of cathepsin D. Wu et al40 further showed that drug-induced apoptosis was inhibited by pepstatin A. Moreover, fibroblasts derived from cathepsin D^−/− gene knockout mice displayed enhanced resistance to adriamycin and etoposide relative to fibroblasts from wild-type mice.40 Similar studies by Roberts et al29 have suggested the involvement of both cathepsin B and cathepsin D in camptothecin-induced apoptosis of a hepatocyte cell line (Hep3B). Following treatment of Hep3B cells with camptothecin, an increase in the cellular activity of cathepsin B was observed, as measured by cleavage of z-ValLeuLys-7-amino-4-chloromethylcoumarin. Increased activity of cathepsin D, as measured with the substrate D-PheSerPhePheAlaAla-p-aminobenzoate, was also detected. Inhibition of cathepsins B and D with CA-074-Me and pepstatin A, respectively, markedly diminished camptothecin-induced apoptosis. Also, as observed in bile salt-induced apoptosis of hepatocytes, cathepsin D is upstream of cathepsin B in the camptothecin-induced cell death pathway.

While inhibition of cathepsins clearly delays some forms of apoptosis, any discussion of the role of cathepsins in apoptotic execution should consider the issue of subcellular localization. Apoptosis signal transduction pathways have been found to occur in the cytoplasm, on the inner surface of the plasma membrane, in mitochondria, and in the nucleus.5414243444546 In particular, caspase-mediated signaling occurs predominantly in the cytoplasm.54748 By contrast, cathepsins B and D are localized in lysosomes, or are secreted by the cell. This raises the question as to how these enzymes can facilitate apoptosis signaling. An emerging line of evidence suggests that during apoptosis, cathepsins B and D are translocated from lysosomes to other subcellular localizations. Three studies have utilized gold-labeled antibodies followed by electron microscopic immunocytochemistry to examine cathepsin D localization during apoptosis.495051 Li et al50 found that treatment of the mouse macrophage cell line J-774 or human macrophages with apoptosis-inducing oxidized LDL changed the pattern of cathepsin D staining from granular lysosomal to diffuse cytoplasmic. Similar results were seen in rat myocytes treated with the quinone naphthazarin,51 and in cells treated with hydrogen peroxide.49 In the case of bile salt-induced apoptosis of hepatocytes, Roberts et al28 performed subcellular fractionation, followed by Western blotting, and found that preexisting cathepsin B redistributed to the nucleus in treated cells. This group also expressed a cathepsin B-green fluorescent protein chimeric molecule, and observed translocation of the fluorescent signal to the hepatocyte nucleus following treatment with bile salts. Another group has examined the localization of cathepsin B in response to the agent atractyloside.52 Atractyloside is known to induce mitochondrial permeability transition, and subsequent apoptosis. Surprisingly, it was shown that atractyloside caused the release of cathepsin B from highly purified lysosomes. Thus, current evidence indicates that cathepsins may be released from lysosomal compartments during some forms of apoptosis. It remains unclear, however, whether this is a general apoptosis phenomenon, or is restricted to execution induced by only some apoptotic stimuli. In addition, it is unclear whether release of cathepsins from lysosomes is controlled by pores or translocators, or is simply the consequence of damage to lysosomal membranes during the apoptotic process.

To date, relatively little is known about potential intersections of cathepsin- and caspase-mediated pathways, although it seems likely that these pathways will be integrated in some fashion. One report has shown that cathepsin G, which is abundantly expressed in neutrophils, can cleave and activate procaspase-7 in vitro.53 Cathepsin G cleaves procaspase-7 between the large and small subunits in the zymogen molecule. Another report has demonstrated that cathepsin B can readily cleave caspase-1 and procaspase-11.52 In addition, cathepsin B showed weak cleavage activity towards procaspase-2, -6, -7 and -14. It was not determined, however, whether cathepsin B-mediated cleavage of procaspases resulted in the activation of these enzymes. Also, while cathepsin G and cathepsin B clearly can cleave certain procaspases in vitro, it is not known whether cathepsin-mediated cleavage of caspases occurs in whole cells. Thus, it is difficult to establish whether caspase activation may be upstream or downstream of cathepsin activation during apoptosis. Perhaps the use of highly selective inhibitors will help to address these issues. In this regard, however, a caveat should be raised. Peptides based on caspase cleavage sequences, and derivatized with fluoromethyl ketone or chloromethyl ketone have commonly been used as specific inhibitors of caspases in vitro and in whole cells. Strikingly, a recent study found that z-VAD-FMK, z-DEVD-FMK, and Ac-YVAD-CMK potently inhibited cathepsin B activity both in vitro and in cells.54 Therefore, the impact of these inhibitors on biochemical events may be due to inhibition of either caspases or cathepsins, or both.

Calpains and apoptosis

Calpains comprise a family of cytoplasmic neutral cysteine proteases.5556575859 Both ubiquitously expressed and tissue-specific calpains have been identified.586061 The most ubiquitous calpains are μ-calpain (or calpain I) and m-calpain (calpain II). The μ- and m-calpains have been the focus of most studies involving calpains, and this review will focus exclusively on these two isoforms. It should be kept in mind, however, that tissue-specific calpains may have essential in vivo roles in apoptosis execution. In addition, calpain homologs have been identified in organisms such as C. elegans, where genetic analyses may provide clues regarding the importance of these proteases.62

Both μ-calpain and m-calpain are composed of two subunits, a large catalytic subunit of approximately 80 kDa, and a smaller subunit of approximately 30 kDa.5863 The 80 kDa subunits specific for μ-calpain and m-calpain are encoded by distinct genes, while the 30 kDa subunit is shared between the two isoforms. Both enzymes bind and require Ca⁺⁺ for optimal activity.555657596465 Although μ- and m-calpains appear to have the same substrate specificity, they are functionally distinquished on the basis of their Ca⁺⁺ sensitivity. μ-Calpain requires micromolar concentrations of Ca⁺⁺ for optimal activity, while m-calpain requires millimolar concentrations.

The activation and activity of calpains is influenced by several factors. As mentioned, calpains must bind Ca⁺⁺ in order to be active, and sequences resembling EF-hand Ca⁺⁺ binding motifs are found in both the 80 and 30 kDa subunits.66 Another interesting feature of the calpains is their ability to associate with membrane phospholipids. Membrane association occurs in response to certain cellular stimuli, and this association may lower the intrinsic Ca⁺⁺ requirements of these enzymes.67 A third factor influencing calpain activity may be autolysis. Following the binding of Ca⁺⁺, calpains undergo autolysis, with cleavages occurring near the amino terminii of the large and small subunits.686970717273 Autolysis appears to increase activity and lower the requirement for Ca⁺⁺. Although it remains controversial as to whether autolysis is absolutely required for activation, Western blotting, looking for conversion of calpains to lower molecular weight forms, is a convenient and frequently used means of assessing calpain activation in cells. Finally, a fourth factor influencing calpain activity is the presence of calpastatin, a 110 kDa endogenous protein inhibitor.7475 Calpastatin is commonly expressed and is highly specific for calpains; it is not known to inhibit any other proteases. The expression and function of calpastatin may be regulated during apoptosis by other proteases, including caspases (discussed below).

Calpains have been implicated in apoptosis based on two types of observations: (1) the activation of calpains during cell death and (2) the inhibition of apoptotic execution by various calpain inhibitors. Calpain activation has been assessed using several different assays including, Western blotting to detect calpain autolysis, Western blotting to detect cleavage of known calpain substrate proteins, and in vitro and whole cell assays to detect cleavage of the fluorogenic calpain substrate N-succinyl-leu-leu-val-tyr-7-amido-4-methylcourmarin (N-s-LLVY-AMC). Using these and a few less common assays, calpain activation has been detected in response to a variety of apoptotic stimuli, in a variety of different cell types. For example, in murine thymocytes treated with dexamethasone, calpain activity increases rapidly, peaking within 1 h of treatment.76 Also, in freshly isolated, apoptosis-prone, human neutrophils, calpain enzymes are found to be constitutively active.77 Treatment of BL30A Burkitt’s lymphoma cells with radiation induces rapid calpain activation (15 min), as measured by cleavage of the calpain substrate fodrin.78 By contrast, incubation of HL-60 cells with 9-amino-20(S)-camptothecin causes slow autolysis/activation of calpain (8–10 h) compared with rapid activation of caspase-3 (2 h;79). Activation of calpains has also been observed in nonhematopoietic cells. In neuronally differentiated PC12 cells, induction of apoptosis by ceramide analogs stimulates calpain-mediated cleavage of N-s-LLVY-AMC.80 Similar activation of N-s-LLVY-AMC cleavage activites are seen during apoptosis caused by reovirus infection of cultured fibroblasts (81) or in vivo ischemia-reperfusion injury in rat hepatocytes.82

In addition to observations of calpain activation in experimental apoptosis, activation or overexpression of calpains has also been seen in human diseases marked by excessive cell death. For example, considerable activation of calpains is seen in brain tissue from patients with Alzheimer’s disease when compared with normal brain tissue.8384 Moreover, Alzheimer’s brains also exhibit significantly reduced expression of the calpain inhibitor calpastatin.84 In Parkinson’s disease, elevated expression of m-calpain is found in the mesencephalon region of the brain.85 Thus, it is possible that overzealous expression and/or activation of calpains may contribute to inappropriate cell loss in some important pathological conditions.

While the activation of calpains during apoptosis suggests an involvement of these enzymes in the execution process, it does not directly demonstrate an essential role. An understanding of the importance of calpains in apoptotic execution has come from studies using calpain inhibitors. A number of different calpain inhibitors have been developed. Unfortunately, many of the earlier inhibitors are somewhat nonspecific. Thus, experiments using these inhibitors must be interpreted with caution; it is possible that actions ascribed to calpains may actually be due to other cross-inhibited proteases (eg the proteasome or cysteine proteases in the cathepsin family). The first generation of calpain inhibitors included the active site inhibitors leupeptin and E64 (E64d is the cell permeable derivative of E64).86878889 These inhibitors also exhibit considerable activity against the proteasome and lysosomal cysteine proteases. More recent active site inhibitors include calpain inhibitor I (N-acetyl-leu-leu-norleucinal (aLLnL)), calpain inhibitor II (N-acetyl-leu-leu-methioninal (aLLM)), and calpeptin (benzyloxycarbonyl-leu-norleucinal).888990 Although more specific than leupeptin and E64, these inhibitors also inhibit to some degree lysosomal cysteine proteases and the proteasome. An even more specific calpain inhibitor is the compound PD150606 [3-(4-iodophenyl)-2-mercapto-(Z)-2-propenoic acid], which interacts with Ca⁺⁺ binding sites in the calpain enzymes.91 Finally, the most specific calpain inhibitor known to exist is the protein calpastatin, which directly binds to calpains and potently inhibits their activities.7475 To date, however, relatively few studies have employed calpastatin to investigate the role of calpains in apoptotic cell death.

As perhaps may have been expected, numerous studies have shown that calpain inhibitors inhibit apoptosis in response to some, but not all, apoptotic stimuli. Studies by Squier et al7692 showed that dexamethasone-induced apoptosis of mouse thymocytes was inhibited by E64d and calpain inhibitor I. It was also found that calpain inhibitor I could inhibit thymocyte apoptosis caused by low-dose radiation, A23187 ionophore, ionomycin, and forskolin; each of these forms of apoptosis requires new RNA synthesis in the thymocyte.92 By contrast, calpain inhibitor I could not prevent apoptosis that is not dependent on RNA synthesis, namely apoptosis caused by heat shock or valinomycin. In human neutrophils, calpain inhibitor I and PD150606 were shown to inhibit apoptosis caused by cycloheximide treatment, but not apoptosis resulting from Fas stimulation.93 During spontaneous neutrophil apoptosis, inhibition of calpains by calpeptin alone, or the proteasome-specific inhibitor, lactacystin, alone had no effect on the rate of cell death.77 However, when used in combination, these inhibitors displayed a synergistic inhibitory effect, indicating synergy between calpains and the proteasome in spontaneous neutrophil apoptosis. Related studies have revealed that antisense-mediated downregulation of calpastatin inhibitor protein significantly accelerates neutrophil apoptosis.93 Studies with HL-60 cells have shown that calpain inhibition by calpeptin fails to block DNA fragmentation and loss of viability caused by treatment with camptothecin.79 However, calpeptin was able to block TNF-induced U937 cell apoptosis, while calpain inhibitor I delayed U937 apoptosis resulting from calphostin treatment.9495 In human platelets, calpeptin blocked A23187-induced caspase fragmentation, cleavage of apoptosis substrate proteins, and microvesiculation.96 Studies with nonhematopoietic cells have shown that calpain inhibitor I and PD150606 blocked reovirus-induced apoptosis in murine fibroblasts, and calpain inhibitors I and II blocked TGF-β-induced DNA fragmentation and loss of viability in rat hepatocytes.8197 In addition, calpain inhibitor I blocked apoptosis in chick ciliary neurons deprived of CNTF, and the calpain inhibitor MDL 28,170 inhibited the death of rat hippocampal pyramidal neurons treated with β-amyloid peptide or staurosporin.9899 Taken together, the studies described above show that calpains can play important roles in the apoptotic execution of both hematopoietic and nonhematopoietic cell types.

The ability of calpains to promote apoptosis in a variety of systems raises questions regarding the molecular mechanism(s) of calpain action. Undoubtedly, the cleavage of specific substrate proteins accounts for the proapoptotic activity of calpain, since the catalytic activity of the enzyme is required. A large variety of proteins have been shown to be calpain substrates, including actin, alpha-actinin, talin, filamin,77100 fodrin,94101 gelsolin,96 FAK,102 integrin β₃,103 PKC α/β/γ,104 calcium/calmodulin-dependent kinase IV,105 c-mos,106 c-fos, c-jun,107108 cyclin D1,109 p53,110 procaspase-3 and -9,96 and Bax.111 This group of substrate proteins is clearly very diverse. However, it should be noted that several of these proteins are cytoskeletal proteins or proteins that can associate with cell membranes. This has led to speculation that calpains may be particularly important in destruction of cellular architecture during apoptosis. The impact of calpain-mediated cleavage of the apoptosis regulatory proteins p53, Bax, and procaspase-3 and -9 is presently unclear. In the case of caspase-3 and -9, calpain cleavage neither activates nor inactivates the caspase enzyme.96 Unfortunately, many of the proteins identified as calpain substrates have only been shown to be cleaved using in vitro experiments. Thus, it remains unclear how many of these proteins are directly cleaved by calpains in vivo. Also, it is currently unknown which substrates may be priority substrates, whose cleavage is critical to the proapoptotic action of calpain.

A number of studies have indicated cross-talk between calpain-mediated pathways and caspase-mediated pathways. In some cases, caspase activation has been reported to be upstream of calpain activation, while in other cases the opposite has been found. Wood and Newcomb79 determined that calpain activation in camptothecin-treated HL-60 cells can be blocked by the pancaspase inhibitor z-VAD-FMK. By contrast, the calpain inhibitor calpeptin failed to block activation of caspases. Interestingly, two groups have reported that activated caspases cleave and inactivate calpastatin inhibitor.112113 Caspase-mediated calpastatin cleavage was seen in Jurkat and U937 cells following stimulation of Fas or TNF receptors, or following treatment with staurosporin. It remains undetermined, however, whether cleavage of calpastatin inhibitor is entirely responsible for the calpain activation in these cells. An opposite relationship between calpains and caspases has been reported during radiation-induced apoptosis. Waterhouse et al78 observed that following treatment of BL30A lymphoma cells with radiation, calpain was activated early, and caspases were activated significantly later. Moreover, inhibition of calpain activity abrogated downstream activation of caspase-3. In yet another study, calpain activation was found to occur independently of caspase activation during activation of platelets.96 Given these different findings, it seems likely that the relationship between caspases and calpains may vary depending on the cell type and apoptotic stimulus.

Granzyme B and apoptosis

Cytotoxic T lymphocytes (CTLs) and natural killer (NK) cells are important for the host defense against viruses, parasitic agents, and transformed cells.114115 CTLs and NK cells induce apoptosis in target cells using at least two distinct mechanisms. One mechanism involves stimulation of cell surface death receptors (such as Fas) on the target cells by death ligands expressed on the surface of the effector cell.116117 This leads to activation of caspase cascades in the target cell. Another mechanism, termed ‘granule exocytosis’, involves the vectoral transfer of the contents of effector cell cytoplasmic granules into the target cell.118119120 Key components of these granules are perforin and the granzyme family of serine proteases.

Perforin is a 70 kDa protein that binds to membrane phosphorylcholine groups in a calcium-dependent manner.121122123 Following binding, perforin inserts into the membrane and oligomerizes, forming pores. This permeabilization of the membrane may facilitate the entry of other molecules, including granzymes, into the target cell.

Within the granules of CTLs and NK cells, granzymes A and B are particularly abundant.124 Granzyme B (also called fragmentin or cytotoxic T cell protease (CCP)) shares with caspases the unique characteristic of cleaving substrate proteins after aspartate residues.125126127128129 An important role for granzyme B in the induction of target cell apoptosis has been demonstrated using gene knockout mice. CTLs and NK cells derived from granzyme B^−/− mice exhibit greatly reduced capacity to induce apoptotic DNA fragmentation in target cells.119130 Earlier complementary studies showed that purified granzyme B alone did not promote apoptosis when added to target cells. However, cotreatment with purified granzyme B and perforin proteins induced marked DNA fragmentation and apoptotic features in four lymphoma target cell lines.129 Based on the ability of perforin to form membrane pores, it has been the general consensus that granzyme B gains entry into target cells through perforin pores. This mode of entry, however, is currently being called into question. Several studies have shown that granzyme B is internalized by target cells in the absence of added perforin.131132133134135 The internalized granzyme B has been reported to reside in the cytoplasm,132133 or in a novel vesicular compartment.134 The triggering of apoptosis in cells that have internalized granzyme B requires further addition of perforin to the cells.131132133134135 It may be that perforin is required for the release of granzyme B from internal vesicles in the target cell. Other studies have indicated that perforin facilitates translocation of granzyme B to the nucleus, and that nuclear localization is critical to the ability of granzyme B to cause apoptosis.132133134135136137

While the importance of granzyme B subcellular localization remains controversial, it is quite clear that granzyme B has the ability to impact the caspase pathway of apoptosis. Studies done in vitro, have shown that granzyme B is capable of cleaving procaspase-3, -6, -7, -8, -9 and -10.138139140141142143144145146147148149150151152153 In the case of procaspases-3, -7 and -9, granzyme B-mediated processing has been shown to generate active caspase enzymes.138141146149 More importantly, studies with whole cells have shown that caspases are activated in target cells following coincubation with granzyme B and perforin.139153154 It remains to be determined, however, which caspases are the preferred in vivo substrates for granzyme B. In any event, it is reasonable to propose that granzyme B may promote apoptosis simply by cleaving and activating endogenous caspases in the target cell.

In cells undergoing granzyme B/perforin-mediated apoptosis, cleavage of the caspase substrate proteins PARP, lamin B, and U1–70 kDa is also observed.151153154155 These cleavage events appear to be due to caspases activated by granzyme B, and not the result of direct granzyme B cleavage, since cleavage of all three proteins is inhibited by 100 μM DEVD- or VAD-containing peptides (100 μM DEVD-CHO or VAD peptides inhibit caspases, but not granzyme B).139151153154155 Two additional caspase substrate proteins, DNA-PK_cs and NuMA, are also cleaved in granzyme B/perforin-treated cells, but cleavage of these proteins is insensitive to DEVD or VAD peptide inhibitors.155 Moreover, the sizes of the DNA-PK_cs and NuMA proteolytic fragments generated by granzyme B differ from those resulting from caspase cleavage. This indicates that during granzyme B-mediated apoptosis, important cellular substrates are cleaved in a caspase-independent fashion. The importance of these caspase-independent cleavage events remains to be determined. However, the fact that granzyme B/perforin-mediated DNA fragmentation and apoptotic death is significantly delayed by 100 μM DEVD/VAD,139153154 underscores the necessity for caspase activation during this form of apoptosis.

Granzyme A and apoptosis

Granzyme A is the most abundant protease found in the granules of CTL cells. The mature granzyme A enzyme is a disulphide cross-linked homodimer of 50 kDa that cleaves substrate proteins following lysine or arginine residues.127156157 Although granzyme A is capable of inducing apoptosis after loading into target cells, the mechanism of action of this protease differs significantly from that of granzyme B. In addition, based on the experimental systems that have been employed thus far, it appears that the role of granzyme A in CTL-induced apoptosis is far more subtle than that of granzyme B. Mice which are deficient in granzyme A expression (granzyme A^−/− mice), exhibit relatively normal CTL-mediated cytotoxicity.158 The only reported defect for granzyme A^−/− mice is an inability to clear the mouse pox virus Ectromelia.159 By contrast, CTLs from granzyme B^−/− mice are capable of inducing target cell death only after prolonged coincubation.130 Thus, granzyme B is critically important for rapid CTL killing. The possibility that granzyme A does have some role in CTL-mediated killing, has been suggested by recent experiments using mice that are deficient in both granzyme A and granzyme B. CTLs from granzyme A^−/−/granzyme B^−/− mice are unable to induce target cell DNA fragmentation, even after prolonged coincubation.160 This indicates that granzyme A activity accounts for the ability of granzyme B^−/− CTLs to induce target cell apoptosis after prolonged exposure. Therefore, granzyme A may allow CTLs to kill target cells under conditions where granzyme B activity is inhibited (eg target cells that express granzyme B inhibitors).

Studies using recombinant proteins have shown that coincubation of granzyme A and perforin with target cells leads to rapid (within 2 h) accumulation of DNA single-strand breaks.161162 This contrasts with the rapid degradation of DNA to oligonucleosomal-length fragments seen in cells treated with granzyme B and perforin. Granzyme A/perforin treatment also leads to nuclear condensation.162 The DNA single-strand breakage and nuclear condensation that occurs in response to granzyme A is insensitive to caspase inhibitors, indicating that these actions of granzyme A are caspase-independent.162 Consistent with this, granzyme A/perforin treatment does not result in processing/activation of procaspase-3 or cleavage of the caspase substrate proteins PARP, lamin B, or rho-GTPase.162 By contrast, granzyme B-induced DNA fragmentation is strictly dependent on the activation of caspases. Both granzyme A and granzyme B (in conjunction with perforin) also induce target cell cytolysis, and in both cases this is a caspase-independent event. Taken together, current evidence indicates that granzyme B is the primary CTL mediator of target cell DNA fragmentation and apoptotic death, and that the apoptotic effects of this protease are mediated primarily through the activation of caspases. Granzyme A, on the other hand, may be more of a default or specialized mediator of target cell apoptosis, with the pathways initiated by granzyme A being distinctly different from those initiated by granzyme B.