Caspases: Intracellular Signaling by Proteolysis

The discovery of a novel proteolytic network in the cytosol of metazoan cells comes as a surprise to a field that previously considered intracellular proteolysis to mainly be a mechanism to process antigen or remove effete proteins. The network, supported by caspases, a family of cysteine proteases that specifically cleave proteins after Asp residues, involves limited proteolysis of a growing number of cellular substrates. It is not surprising that a signaling pathway utilizes proteolysis, for this is the essence of the blood coagulation cascade. What is surprising is that the caspase pathway takes place in the interior of a cell. Caspase precursors are usually activated at internal conserved Asp residues, with the result that most activated caspases can process their own and other caspase zymogens given sufficient time and a high enough concentration in vitro (see for example,12Srinivasula S.M Ahmad M Fernandes-Alnemri T Litwack G Alnemri E.S Proc. Natl. Acad. Sci. USA. 1996; 93: 14486-14491Crossref PubMed Scopus (474) Google Scholar). This suggests a cascade mechanism for transmission of signals, but the extent to which this happens in vivo is currently an enigma, and a fertile area of research. The caspases seem to be a development of multicellular organisms, and in humans at least seven of the ten currently known family members participate in one of two distinct signaling pathways: (1) activation of proinflammatory cytokines, and (2) promotion of apoptotic cell death. Recent publications have reviewed the discovery of the caspases and some of their substrates (see for example,11Nicholson D.W Nat. Biotechnol. 1996; 14: 297-301Crossref PubMed Scopus (231) Google Scholar, 19Zhivotovsky B Burgess D.H Vanags D.M Orrenius S Biochem. Biophys. Res. Commun. 1997; 230: 481-488Crossref PubMed Scopus (173) Google Scholar), and here we provide an update that focuses on the initiation, transmission, and regulation of caspase activity. In common with other protease zymogens, generation of an active form requires limited proteolysis (Figure 1). For the caspases, this results from cleavage in an interdomain linker segment to give a heterodimeric enzyme, with both chains containing essential components of the catalytic machinery. Sometimes an N-terminal peptide, not required for enzyme activity, is released. The role of the N peptide is only known for caspases 1 and 8, where it behaves as a protein interaction domain to modulate activation. The 3-D structures of caspases 1 and 3 reveal two heterodimers interacting via the small chains, providing each molecule with two active sites. The primary recognition pocket (S1) is well adapted to accept an Asp side chain of the substrate, and additional pockets (S2–S4) distinguish the caspases from each other (Figure 2). Caspases cleave a number of cellular proteins, and the process is one of limited proteolysis where a small number of cuts, usually only one, are made in interdomain regions. Sometimes cleavage results in activation of the protein, sometimes in inactivation, but never in degradation since their substrate specificity distinguishes the caspases as among the most restricted of endopeptidases. Caspase 1 is remarkably specific for the precursors of IL-1β and IL-18 (interferon-γ-inducing factor), making a single initial cut in each procytokine that activates them and allows exit from the cytosol. Ectopic expression of caspase 1 in some cells can result in apoptosis, but a role in developmentally programmed cell death is unlikely given the normal phenotype of knockout mice (see for example,5Ghayur T Banerjee S Hugunin M Butler D Herzog L Carter A Quintal L Sekut L Talanian R Paskind M et al.Nature. 1997; 386: 619-623Crossref PubMed Scopus (985) Google Scholar). Processing of the caspase 1 precursor, required for activation, is rarely seen in experimental models of apoptosis, and thus, though it is clearly required for activation of the cytokines mentioned above, the role of this caspase in apoptosis, if any, is still uncertain. The close relationship in sequence identity and predicted substrate specificity of caspases 4 and 5 implicate them as effectors of cytokine activation, possibly as upstream activators of caspase 1 itself. Whereas caspase 1 (and possibly 4 and 5) is primarily involved in procytokine activation, other caspases, notably 2, 3, 6, 7, 8, and 10, are considered to promote pathways to apoptosis. This conclusion is based largely on the following observations: (1) the zymogens are seen to be processed during apoptosis, or in vitro models of apoptosis, and (2) at least in vitro, they cut proteins whose cleavage is associated with apoptotic cell death. So far, about a dozen proteins have been shown to be specifically cleaved during apoptosis. It is relatively simple to demonstrate which clips are caused by caspases (it turns out to be almost all of them) through analysis of the cleavage site, but more difficult to determine which caspases are responsible in vivo. For example, the first protein discovered to be cleaved by caspases during apoptosis, poly(ADP-ribose) polymerase, is a substrate for most caspases under somewhat extreme conditions in vitro, but in vivo is probably targeted by caspases 3 and 7. Thus, by combining in vivo observations with in vitro tests on suspect proteins, one can recognize distinctions in substrate specificity within the caspase family that fit remarkably closely to the S4−S1 subsite preferences deduced from synthetic peptidic substrates (13Talanian R.V Quinlan C Trautz S Hackett M.C Mankovich J.A Banach D Ghayur T Brady K.D Wong W.W J. Biol. Chem. 1997; 272: 9677-9682Crossref PubMed Scopus (752) Google Scholar, 16Thornberry N.A Rano T.A Peterson E.P Rasper D.M Timkey T Garcia-Calvo M Houtzager V.M Nordstrom P.A Roy S Vaillancourt J.P et al.J. Biol. Chem. 1997; 272: 17907-17911Crossref PubMed Scopus (1786) Google Scholar) (Figure 2). Singularly important in this context is that caspase zymogens are themselves substrates for caspases, and inspection of the individual interdomain linker in each zymogen reveals target sites that confirm the preference of some caspases to activate others in a hierarchical relationship. Thus, pathways exist to transmit signals via sequential caspase activations, and this has been most extensively examined in apoptosis. A key to apoptosis is the discovery in many laboratories that, irrespective of the lethal stimulus, death results in the same apoptotic morphology that includes cell and organelle dismantling and packaging, DNA cleavage to nucleosome-sized fragments, and caspase-mediated cleavage of the same cellular proteins. A fascinating demonstration of the importance of caspases was provided by engaging the death pathway through expression of the pro-apoptotic factor Bax in cells bathed in the presumptive cell-permeable caspase inhibitor Z-VAD-fluoromethyl ketone (18Xiang Q Chao D.T Korsmeyer S.J Proc. Natl. Acad. Sci. USA. 1996; 93: 14559-14563Crossref PubMed Scopus (853) Google Scholar). Though death was delayed, it was not abolished despite complete inhibition of caspase activity. However, the resulting death morphology was not apoptotic but more reminiscent of necrosis. In contrast, ligation of the death receptor Fas on cells expressing the viral caspase inhibitor CrmA had no effect—the cells were completely vibrant in otherwise lethal levels of Fas antibody and exhibited no signs of apoptosis or other forms of death (14Tewari M Quan L.T O'Rourke K Desnoyers S Zeng Z Beidler D.R Poirier G.G Salvesen G.S Dixit V.M Cell. 1995; 81: 801-809Abstract Full Text PDF PubMed Scopus (2215) Google Scholar). The explanation for the different outcomes cuts to the heart of the death pathway. In initiation of death via Fas, the first signal is caspase 8 activation at the cytosolic face of the receptor, but in Bax-initiated death, the signal is integrated inside the cell, resulting in a death pathway that is caspase-independent, and a parallel apoptosis pathway that is completely dependent on caspases for its accurate outcome. Caspases are not the only enzymes that participate in apoptosis, since nucleases and protein kinases may also participate, but they are absolutely required for the accurate and limited proteolytic events that typify this type of programmed cell death. The caspases implicated in apoptosis are currently divided into initiators and executioners (Figure 3). The exact order of the executioners and the place of other caspases in the pathway are still controversial, but at least in the Fas pathway, signaling of death is transmitted in part by sequential caspase activations. In regard to the order of caspases in apoptotic pathways, there is a valuable lesson to be learned from the caspase 3 knockout mouse, which has a profound developmental defect (8Kuida K Zheng T.S Na S Kuan C.-y Yang D Karasuyama H Rakic P Flavell R.A Nature. 1996; 384: 368-372Crossref PubMed Scopus (1655) Google Scholar). Mice frequently die in utero, and those that survive to birth live only a few days with brains twice the normal volume, due to a large number of supernumerary cells that presumably failed to die during neuronal development. One way to interpret this result is that some forms of developmentally programmed cell death may utilize executioners (in this case caspase-3) as initiators of an apoptotic pathway. This adds a cautionary note to the common impression that all apoptotic pathways utilize caspases in the same order. Why have such a complex network in place to kill a cell when all that is really needed is to turn off a crucial housekeeping gene? Presumably, having necrotic cells around is dangerous, and the apoptotic response is thus a mechanism to dismantle cells for disposal in a way that does not compromise the rest of the organism. It therefore becomes important to determine whether the observed caspase-mediated cleavages are a vital part of the apoptotic program, or just bystander events of no direct significance. The answer to this question is still open, but it is instructive to note that proteins targeted for cleavage by caspases during apoptosis are involved in RNA splicing, DNA repair, and scaffolding of the cytosol and nucleus. Once the decision to die is made, specific mechanisms must be terminated to prevent futile attempts at repair. Before the dying cell can be packaged for endocytosis, it must be dismantled. Superimposed on this dismantling issue are the crucial observations that executioner caspases activate cellular proteins, with a recent example being the description of a protein whose cleavage is required for induction of DNA fragmentation (9Liu X Zou H Slaughter C Wang X Cell. 1997; 89: 175-184Abstract Full Text Full Text PDF PubMed Scopus (1593) Google Scholar). Consequently, some of the most informative advances in future cell death research will undoubtedly be the discovery of the key proteins activated by caspases during execution. The point of commitment in pathways initiated by cell surface death receptors is evident. Receptor clustering results in apical caspase activation to generate the first proteolytic signal, though the mechanism is far from clear. The point of commitment in other pathways is still uncertain, but it has been suggested to encompass a cusp at which the rate of damage outstrips the ability of the cell to repair itself (10Martin S.J Green D.R Cell. 1995; 82: 349-352Abstract Full Text PDF PubMed Scopus (1241) Google Scholar). Thus, low levels of damage beyond the repair ability of the cell result in post-mitotic arrest, and the cell lives. More severe damage, integrated by the apostat (Figure 3), results in transduction of a death signal that likely activates the executioners. The identity of the activator is unknown, but clues to its identity come from genetic and biochemical dissection of cell death in the nematode Caenorhabditis elegans, which serves as an elegant model of apoptosis in higher organisms. In this primitive metazoan, the cell death promoter Ced4 is required to activate the effector caspase Ced3 (2Chinnaiyan A.M O'Rourke K Lane B.R Dixit V.M Science. 1997; 275: 1122-1126Crossref PubMed Scopus (543) Google Scholar), and so the human counterpart of Ced4 would presumably be a required component of the apostat. Since apical caspases transmit signals to executioners by cleaving zymogen inter-domain linkers, it is reasonable that other proteases would be able to activate death by the same mechanism. Indeed, this strategy has been adopted by the cytotoxic cell protease granzyme B, which is the only other mammalian protease that shares the caspase primary specificity for Asp. This protease is able to activate most caspase zymogens, though significantly not caspase 1, by cleaving at the interdomain linker to engage the endogenous cascade following its delivery from cytotoxic cells to sensitive targets (reviewed by19Zhivotovsky B Burgess D.H Vanags D.M Orrenius S Biochem. Biophys. Res. Commun. 1997; 230: 481-488Crossref PubMed Scopus (173) Google Scholar). More remarkable is the recent demonstration that proteases without specificity for Asp are able to activate caspase zymogens at alternate sites in the linker region, at least in vitro (20Zhou Q Salvesen G.S Biochem. J. 1997; 324: 361-364Crossref PubMed Scopus (120) Google Scholar). It seems that the linker segment between the large and small subunits is an unusually susceptible interdomain connector designed to be utilized for rapid proteolysis. Normal caspase zymogen activation takes place at conserved Asp residues, but the proteolytic sensitivity of the caspase interdomain link may allow non-Asp-specific proteases such as those from lysosomes or viruses to engage the apoptotic apparatus under pathological conditions. Perhaps this explains the observation that injection of seemingly nonspecific proteases into cells can result in apoptosis (17Williams M.S Henkart P.A J. Immunol. 1995; 153: 4247-4255Google Scholar). Not surprisingly, inhibition of caspases is a strategy adopted by viruses in their attempt to elude the response of the cell to the infectious insult (Figure 4). Thus, much of our grasp of the pivotal role played by proteases in apoptosis and cytokine activation comes from understanding that the pox virus protein CrmA and the baculovirus protein p35 specifically target caspases. The main target for CrmA in context of a cowpox virus infection is probably caspase 1, enabling the virus to abrogate IL-1β and IL-18 production in infected cells, and also possibly caspase 8 for which the inhibitor demonstrates high potency. CrmA binds other caspases too weakly to directly affect the execution phase of apoptosis. In contrast, p35 has a much broader range, and although its natural targets are not in mammalian cells, it has the ability to terminate activity of almost all caspases. Interestingly, no homologs of p35 are known in mammals, and though mammals possess many homologs of the serpin CrmA, none of them seem to be targeted against caspases. So far, the only demonstrated caspase inhibitor endogenous to mammals is human X-linked IAP (XIAP), a member of a family of similar proteins found throughout metazoans, and first identified in baculovirus. It is far from clear why cells that have committed to apoptosis would need an inhibitor, so presumably XIAP, which seems to be targeted against caspases 3 and 7 (3Deveraux Q Takahashi R Salvesen G.S Reed J.C Nature. 1997; 388: 300-303Crossref PubMed Scopus (1666) Google Scholar), serves to regulate adventitious proteolysis before it has reached the catastrophic threshold. The discovery of caspases has proceeded well ahead of our understanding of their place in the transmission of cellular inflammatory and apoptotic signals. One of the greatest challenges facing us now is to determine how the regulation of caspase activity and activation is handled, and how cleavage of proteins leads to apoptosis. Given the restricted substrate specificity, it is almost certain that caspases and their substrates coevolved during the course of metazoan radiation. In this context, a close investigation of the recognition of known protein substrates by caspases will likely stimulate the discovery of additional key substrates in both the proinflammatory and apoptotic branches of the network. The other major challenge is to apply knowledge of the caspase network to therapy. Much of the early work on caspase 1 was conducted in pharmaceutical companies that reasoned that prevention of proinflammatory cytokine activation would be therapeutic for inflammatory disease. In this context, it is heartening to see that the caspase 1 knockout mouse has a complete absence of IL-1 and IL-18 activity (5Ghayur T Banerjee S Hugunin M Butler D Herzog L Carter A Quintal L Sekut L Talanian R Paskind M et al.Nature. 1997; 386: 619-623Crossref PubMed Scopus (985) Google Scholar, 6Gu Y Kuida K Tsutsui H Ku G Hsiao K Fleming M.A Hayashi N Higashino K Okamura H Nakanishi K et al.Science. 1997; 275: 206-209Crossref PubMed Scopus (970) Google Scholar) but develops normally (8Kuida K Zheng T.S Na S Kuan C.-y Yang D Karasuyama H Rakic P Flavell R.A Nature. 1996; 384: 368-372Crossref PubMed Scopus (1655) Google Scholar). Taken together with the somewhat restricted extended substrate specificity of the caspase 1 branch of the family (Figure 2), this bodes well for targeting by synthetic protease inhibitors without interfering with essential apoptotic functions, at least in the short term. Therapeutic intervention in apoptosis would also be beneficial since inappropriate increases in cell death have been reported in AIDS, neurodegenerative disorders, and ischemic injury; and a decrease in normal cell deaths is a characteristic of cancer (15Thompson C.B Science. 1995; 267: 1456-1462Crossref PubMed Scopus (6009) Google Scholar). A significant problem in the development of antidegenerative drugs is the potential toxic effect due to abrogation of normal (homeostatic) apoptosis in the human adult, which accounts for over 1011 cell deaths per day. Given this, the first generation of caspase inhibitors are likely to be used only in acute settings such as the salvage of cells destined to die by apoptosis following stroke or myocardial ischemia. While the goal of treating acute degenerative diseases is to prevent unwanted death, the major challenge in treating cancer is to kill cells that have become resistant to available chemotherapy. Since the failure to undergo apoptosis is associated with drug resistance, direct activation of caspases in cancer cells may be an effective strategy to kill the resistant cells. Obviously, for such an approach to be practical the caspase activator therapeutics should be selective for cancer cells, utilizing specific properties of the oncogenic state (4Fearnhead H.O McCurrach M.E O'Neill J Zhang K Lowe S.W Lazebnik Y.A Genes Dev. 1997; 11: 1266-1276Crossref PubMed Scopus (61) Google Scholar). Thus, the long-term challenge in basic research is to understand the course of caspase activity during animal development, homeostasis, and pathology; and in therapy to develop drugs targeted to specific cells and tissues to block acute and chronic disease through intervention in the caspase network.