Portal de Periódicos da CAPES

A License to Kill

1996; Cell Press; Volume: 85; Issue: 6 Linguagem: Inglês

10.1016/s0092-8674(00)81005-3

1097-4172

Andrew Fraser, Gérard I. Evan,

CRISPR and Genetic Engineering

Apoptosis is the descriptive name given to the process of programmed cell death in vertebrates. During apoptosis, a cell activates an intrinsic suicide mechanism that systematically trashes the cell: its surface membrane begins to bleb and express pro-phagocytic signals, the cell shrinks and severs contact with its neighbors, chromatin becomes condensed and cleaved (ensuring genetic death of the affected cell), and eventually the whole cell fragments into membrane-bound vesicles that are rapidly ingested by neighboring cells. The apoptotic process is extremely rapid (typically between a few minutes and a few hours), and the apoptotic debris is cleared with similar rapidity. Up until some 4 years ago, apoptosis held much the same status among serious scientists as aromatherapy does among physicians—an interesting and amusing concept, decorative, but definitely not mainstream. Of course, all that has now changed. Nowadays, the ferocious pace (and funding) of research into apoptosis is matched only by a deep sense of frustration that, despite elucidation of many of the important players in the cell death process, very little hangs together to give a coherent picture of how apoptosis is regulated and executed. Part of the problem is that almost everything (for example, oncoproteins, tumor suppressor proteins, growth factors, and signaling pathways) seems able to induce apoptosis and almost everything (for example, oncoproteins, tumor suppressor proteins, cytokines, and signaling pathways) seems able to suppress it. Nonetheless, despite the wide diversity and dissimilarity of factors that can modulate apoptosis, the basal apoptotic mechanism appears to be relatively simple in general structure and highly conserved throughout metazoan evolution. It is therefore difficult to see how so many different kinds of stimulus and insult, in so many diverse organisms, can be wired into one unitary suicide machine. The induction of apoptosis by the cytokine tumor necrosis factor (TNF) and by its cousin, the cytotoxic T cell ligand effector CD95L/FasL/APO-1L, have hitherto presented excellent examples of our depth of ignorance of the way apoptosis is regulated. Both TNF and CD95L are powerful agents, which upon binding their specific target receptors override much of the capacity of their target cell for survival to induce an ineluctable suicide response—the classic “cell killing” by TNF or cytotoxic T lymphocytes. However, understanding how signaling events at the plasma membrane impact on the basal machinery that executes cell suicide has been a mystery. Part of that mystery may now have been solved. It appears that there is an almost direct physical connection between the basal cell suicide machine and the TNF/CD95-derived informational signals that initiate it. Much of our knowledge of the basal mechanism underlying programmed cell death derives from the elegant studies of Horvitz and colleagues on developmental cell death in the developmentally deterministic nematode worm Caenorhabditis elegans. During development of the hermaphrodite worm, 1090 cells are born and 131 die. Mutants defective in this death program define three principal genes (ced genes) that orchestrate each cell death. ced-3 and ced-4 act to kill target cells, while ced-9 acts to suppress the killer genes. ced-9 is a structural and functional homolog of the human proto-oncogene bcl-2, one of a family of genes intimately involved in the regulation of vertebrate apoptosis. ced-4 has no known function or homolog. ced-3 encodes a cysteine protease that is related to an emerging family of vertebrate cysteine proteases of which the prototype is the eponymous interleukin-1β-converting enzyme (ICE) (21Yuan J.Y Shaham S Ledoux S Ellis H.M Horvitz H.R Cell. 1993; 75: 641-652Abstract Full Text PDF PubMed Scopus (2182) Google Scholar). ICE is well characterized as the activator of the inflammatory cytokine interleukin-1β through cleavage of the inactive 31 kDa interleukin-1β precursor at Asp116↓Ala117. The mammalian ICE-related protease family now comprises some nine known members. These include the Ced-3-like subfamily members CPP32β (also known as Yama and Apopain), Mch2, Mch3 (also known as ICE-LAP-3 and CMH-1), and Mch4; the ICE subfamily members ICE, ICE rel III, and ICE rel II (also known as Tx and ICH2); and the NEDD-2 family members ICH-1 and (murine) Nedd-2. Of these, ICE, CPP32β, and Mch2 have been definitely implicated in apoptosis. ICE is required for induction of apoptosis through the CD95 signaling pathway (see below) (6Enari M Hug H Nagata S Nature. 1995; 375: 78-81Crossref PubMed Scopus (789) Google Scholar, 11Kuida K Lippke J.A Ku G Harding M.W Livingston D.J Su M.-S Flavell R.A Science. 1995; 267: 2000-2003Crossref PubMed Scopus (1403) Google Scholar) and appears to be activated near the apex of an ICE-related protease cascade, one of whose critical targets is CPP32β (7Enari M Talanian R Wong W Nagata S Nature. 1996; 380: 723-726Crossref PubMed Scopus (956) Google Scholar). In addition to a role in CD95-induced cell death, CPP32β is also thought to be a general component of the apoptotic program and is a component of prICE, a proteolytic extract derived from apoptotic chick cells that induces condensation and fragmentation of interphase nuclei (12Lazebnik Y Takahashi A Poirier G Kaufmann S.H Earnshaw W J. Cell Sci. 1995; 19: 41-49Crossref Google Scholar). Mch2 cleaves nuclear lamins (18Takahashi A Alnemri E Lazebnik Y Fernades-Alnemri T Litwack G Moir R Goldman R Poirier G Kaufmann S Earnshaw W Proc. Natl. Acad. Sci. USA, in press. 1996; Google Scholar), an activity also present in prICE. All members of the ICE-related protease family share a predilection for cleavage of their substrates after an aspartate residue at the P1, usually followed by a small residue at the P1′ position, and all are synthesized as pro-enzymes that are activated by cleavage at critical aspartate residues that themselves conform to the substrate consensus for ICE family proteases. Thus, the ICE-related proteases are presumed to exist within hierarchies of auto- and trans-cleavage. For example, pro-ICE can be activated by autocleavage and pro-CPP32β can be activated by ICE as well as by the cytotoxic T cell granule serine protease, granzyme B (5Darmon A.J Nicholson D.W Bleackley R.C Nature. 1995; 377: 446-448Crossref PubMed Scopus (629) Google Scholar). ICE-related protease processing releases an N-terminal pro-domain and cleaves the remaining polypeptide to yield two subunits (p20 and p10 for human ICE) that form the active (p20:p10)2 enzyme. The larger subunit contains the catalytic cysteine, but both subunits are required for enzyme activity. ICE-related proteases are implicated in vertebrate apoptosis for a number of reasons, notwithstanding their clear structural and functional homology with the nematode Ced-3 killer protein (21Yuan J.Y Shaham S Ledoux S Ellis H.M Horvitz H.R Cell. 1993; 75: 641-652Abstract Full Text PDF PubMed Scopus (2182) Google Scholar), which is also a cysteine protease. First, ICE-like proteases comprise the prICE activity responsible for nuclear fragmentation in the in vitro system of Earnshaw and colleagues, arguably the best model for the fate of nuclei inside intact apoptotic cells (12Lazebnik Y Takahashi A Poirier G Kaufmann S.H Earnshaw W J. Cell Sci. 1995; 19: 41-49Crossref Google Scholar). Second, ectopic intracellular expression of ICE-related proteases induces apoptosis in mammalian cells (although taking this as evidence for direct involvement of ICE-related proteases in apoptosis is somewhat compromised by the fact that intracellular expression of many proteases, including the totally unphysiological trypsin and chymotrypsin, do the same). In almost all such studies, the pro-enzyme form of the ICE-related protease is expressed. Presumably, the normally tight control over ICE-related protease activation that exists in cells is compromised by the very high levels of expression obtained in such studies, which either facilitate spontaneous autocatalytic cleavage or increase the sensitivity of the pro-enzyme to activation by some preexisting low level activator. Third, inhibitors of ICE proteases, either the viral proteins p35 (baculovirus) or CrmA (poxvirus) or aldehydes or fluoromethyl ketone–derivatized synthetic peptide inhibitors based on preferred substrate sequences, suppress mammalian cell apoptosis. CrmA and the peptide inhibitor YVAD–fluoromethyl ketone are both fairly specific inhibitors of ICE, and both block vertebrate neuronal death following nerve growth factor withdrawal and suppress CD95-induced apoptosis (6Enari M Hug H Nagata S Nature. 1995; 375: 78-81Crossref PubMed Scopus (789) Google Scholar, 13Los M Van de Craen M Penning L.C Schenk H Westendorp M Baeuerle P.A Droge W Krammer P.H Fiers W Schulze Osthoff K Nature. 1995; 375: 81-83Crossref PubMed Scopus (634) Google Scholar). This indicates a specific role for ICE in the CD95 signaling pathway and is consistent with the fact that thymocytes isolated from homozygous ICE knockout mice are resistant to CD95-induced killing (11Kuida K Lippke J.A Ku G Harding M.W Livingston D.J Su M.-S Flavell R.A Science. 1995; 267: 2000-2003Crossref PubMed Scopus (1403) Google Scholar) and that ICE activation is an early event in cells following CD95 ligation (7Enari M Talanian R Wong W Nagata S Nature. 1996; 380: 723-726Crossref PubMed Scopus (956) Google Scholar). Baculovirus p35, which has a broader spectrum of ICE family protease inhibition than CrmA and inhibits ICE, ICH-1, ICH-2, CPP32β, and CED-3, is an even more generic suppressor of apoptosis and also blocks CD95- and TNF-induced cell death (1Beidler D Tewari M Friesen P Poirier G Dixit V J. Biol. Chem. 1995; 270: 16526-16528Crossref PubMed Scopus (201) Google Scholar). The essential role of Ced-3/ICE proteases in vertebrate apoptosis, together with their pivotal position in nematode cell death, is consistent with their acting as the principal effectors of apoptosis, presumably through their proteolytic action on specific targets. Indeed, a number of ICE-related protease targets have been identified, including members of the ICE-related protease family themselves. Other targets include the nuclear enzymes poly(ADP-ribose) polymerase and DNA-dependent protein kinase, both involved in aspects of DNA damage sensing and repair, U1 ribonucleoprotein and the nuclear lamins, and, in the cytoplasm, protein kinase Cδ and various components of the cytoskeleton, such as actin. However, which, if any, of these targets is responsible for the cell blebbing, condensation, and fragmentation that characterizes apoptosis is unknown. TNF receptor 1 (TNFR1 [p55]) and CD95/Fas/APO-1 are unique among members of the TNFR superfamily in that they trigger apoptosis upon binding their ligands. The ligand for TNFR1, TNF, is a widely expressed soluble cytokine. In contrast, expression of the membrane-bound CD95 ligand is restricted in the main to the surfaces of cytotoxic T lymphocytes, where it provides one of the redundant mechanisms by which such cells “kill” their targets (15Nagata S Golstein P Science. 1995; 267: 1449-1456Crossref PubMed Scopus (3901) Google Scholar). Both TNFR1 and CD95 share a related intracellular “death domain” (DD) of some 70 amino acids (19Tartaglia L Ayres T Wong G Goeddel D Cell. 1993; 74: 845-853Abstract Full Text PDF PubMed Scopus (1136) Google Scholar) that plays a pivotal role in the abilities of these receptors to trigger apoptosis. The DDs of TNFR1 and CD95 act as interfaces that mediate ligand-dependent recruitment of other DD-containing proteins: Fas-associated protein with death domain (FADD)/MORT1 and receptor-interacting protein (RIP) in the case of CD95 (3Chinnaiyan A.M O'Rourke K Tewari M Dixit V.M Cell. 1995; 81: 505-512Abstract Full Text PDF PubMed Scopus (2085) Google Scholar, 17Stanger B.Z Leder P Lee T.H Kim E Seed B Cell. 1995; 81: 513-523Abstract Full Text PDF PubMed Scopus (830) Google Scholar) and TNFR1-associated death domain protein (TRADD) and RIP in the case of TNFR1 (9Hsu H Xiong J Goeddel D.V Cell. 1995; 81: 495-504Abstract Full Text PDF PubMed Scopus (1676) Google Scholar, 17Stanger B.Z Leder P Lee T.H Kim E Seed B Cell. 1995; 81: 513-523Abstract Full Text PDF PubMed Scopus (830) Google Scholar). FADD, TRADD, and RIP all induce apoptosis when ectopically expressed, consistent with their having roles in the death signaling pathway. The DD has been conserved through much of metazoan evolution. It shares limited homology with the protein product of the Drosophila reaper gene (8Golstein P Marguet D Depraetere V Cell. 1995; 81: 185-186Abstract Full Text PDF PubMed Scopus (88) Google Scholar), originally identified through an elegant gross chromosome deletion mapping study as a gene on the third chromosome (75C1,2) that was essential for all of the many cell deaths that occur during normal Drosophila embryogenesis. Deletion of reaper not only forestalls all developmental cell death, but also suppresses cell death in response to X-irradiation or mutation-induced errors in developmental programming. In all cases of cell death, reaper is induced some 1–2 hr prior to the demise of the actual cell. The naturally occurring lpr (lymphoproliferation) mouse mutant (which suffers from lymphadenopathy, splenomegaly, and overproduction of IgG and IgM antibodies) has a mutation at a critical valine residue (Val-238) in the CD95 DD that inactivates its ability to form a heterodimer with the FADD/MORT1 DD. Similarly, experimental mutation of the equivalent residue (Val-121) in FADD/MORT1 destroys its ability to bind CD95 (3Chinnaiyan A.M O'Rourke K Tewari M Dixit V.M Cell. 1995; 81: 505-512Abstract Full Text PDF PubMed Scopus (2085) Google Scholar) and might therefore be expected to inactivate its ability to induce apoptosis. However, deletion analysis of FADD/MORT1 indicates that, although its DD is required for association with CD95, it is not required for induction of apoptosis. Instead, a discrete N-terminal region is necessary—this has been dubbed the “death effector domain” (DED) (3Chinnaiyan A.M O'Rourke K Tewari M Dixit V.M Cell. 1995; 81: 505-512Abstract Full Text PDF PubMed Scopus (2085) Google Scholar). The DDs on TNFR1 and CD95 therefore act as ligand-dependent seeds that promote more extensive oligomerization. Whereas CD95 directly recruits FADD/MORT1, TNFR1 binds TRADD, which then acts as an adaptor protein to recruit FADD/MORT1 (10Hsu H Shu H.-B Pan M.-G Goeddel D Cell. 1996; 84: 299-308Abstract Full Text Full Text PDF PubMed Scopus (1685) Google Scholar). Thus, FADD/MORT1 seems to be the point of convergence between the CD95 and TNFR1 death pathways. Consistent with this, a dominant-negative truncation mutant of FADD/MORT1 lacking the N-terminal DED blocks both CD95- and TNFR1-induced apoptosis (4Chinnaiyan A Tepper C Seldin M O'Rourke K Kischkel F Hellbrardt S Krammer P Peter M Dixit V J. Biol. Chem. 1996; 271: 4961-4965Crossref PubMed Scopus (695) Google Scholar). The formation of CD95–FADD or TNFR1–TRADD–FADD complexes following ligand binding is clearly important for the induction of apoptosis by these signals. How these complexes actually activate the cell death machinery including the ICE-related proteases has until now been completely unknown. Two papers (14Muzio M Chinnaiyan A.M Kischkel F.C O'Rourke K Shevchenko A Scaffidi C Bretz J.D Zhang M Ni J Gentz R Mann M Krammer P.H Peter M.E Dixit V.M Cell. 1996; 85 (this issue)Abstract Full Text Full Text PDF PubMed Scopus (2656) Google Scholar [this issue of Cell ]; 2Boldin M.P Goncharov T.M Goltsev Y.V Wallach D Cell. 1996; 85 (this issue)Abstract Full Text Full Text PDF PubMed Scopus (2055) Google Scholar [this issue of Cell ]) have now identified a protein that provides the missing link both functionally and physically. Muzio et al. have directed their attention to CAP4, a protein known to associate with the CD95 death-inducing signaling complex, DISC, in a ligand-dependent manner. Using nanoelectrospray tandem mass spectrometry, an emerging technique that permits the direct sequencing of femtomolar quantities of proteins analytically separated by polyacrylamide gel electrophoresis, they obtained CAP4 peptide sequence and used this to isolate a cDNA. This cDNA encodes a novel member of the ICE protease family, which they call FLICE. Boldin et al. identify the same protein, which they call MACH (for MORT1-associated CED-3 homolog), by a yeast two-hybrid screen using FADD/MORT1 as bait. Moreover, dominant-negative isoforms of MACH block both CD95- and TNF-induced apoptosis, implying a role for MACH in both signaling pathways. At its N-terminus, FLICE/MACH possesses two tandem regions of homology with the N-terminal DED of FADD/MORT1. These mediate the interaction with FADD/MORT1. However, at its C-terminus, FLICE/MACH possesses a 260 amino acid region with strong homology to the ICE-related cysteine proteases, including a large pro-domain and all known residues absolutely required for protease activity. Identification of multiple MACH isoforms that have varying effects on apoptosis, from activation of cell death to dominant repression of the CD95 and TNFR signaling pathways, adds the usual fractal complexity expected in biology, but the overall picture is clear. Ligand-activated CD95 and TNFR1s induce apoptosis by the recruitment (and presumably concomitant activation) of an ICE-related protease to the receptor signaling complex. Although FLICE/MACH1 makes a striking connection between CD95 and TNFR complexes and the cell death machinery including the ICE proteases, many questions remain unanswered even within these pathways. Perhaps the most pressing is how recruitment of FLICE/MACH1 to the TNFR1/CD95–FADD/MORT1 complex results in its activation. ICE-related protease activity could in principle be controlled at two points—by modulating activation of the pro-enzyme or by regulating the availability of downstream inhibitors (like p35 and CrmA) that block the active proteases. No cellular counterparts of CrmA or p35 have yet been found, and most evidence indicates that activation of the proteases is the major regulatory mechanism. Unfortunately, very little is known about how ICE-related protease activation is regulated. The only clear example of an upstream activator is the cleavage and activation of CPP32β by the cytotoxic T cell serine protease, granzyme B (5Darmon A.J Nicholson D.W Bleackley R.C Nature. 1995; 377: 446-448Crossref PubMed Scopus (629) Google Scholar), which circumvents the problem of infinite proteolytic regress by initiating the proteolytic cascade with a protease that is kept physically isolated from its substrate until required. ICE-related proteases exist as (p20:p10)2 homodimers that appear to associate first as two pro-ICE p45 molecules. Autoactivation probably involves trans-cleavage between the two p45 molecules, so that ICE protease processing could, in principle, be controlled by regulating association between the two pro-enzyme monomers. One model for ligand-dependent FLICE/MACH activation at the TNFR1 and CD95 receptors is that FLICE/MACH autoactivates via a two-stage process. First, FLICE/MACH1 is recruited by FADD/MORT1 to the CD95 or TNFR1 signaling complex via its DED domain. Abstraction of one of its two DED domains into the receptor complex might relieve autoinhibition of FLICE/MACH1 caused by intramolecular interaction between its two DED domains, which are effectively extensions of the FLICE/MACH1 pro-domain. This would prime FLICE/MACH1 for a second trans-cleavage activation step brought about by the close proximity of other recruited pro-FLICE/MACH molecules. Active FLICE/MACH would then be released from the signaling complex. Whether the remaining DED-containing pro-domain of the FLICE/MACH molecule left in the receptor complex has any distinct role in signaling or regulation of cell death is unclear. One further outstanding problem with placing FLICE/MACH at the apex of the TNFR1/CD95 protease cascade is the substantial evidence implicating ICE in CD95- and TNF-induced apoptosis. Not only is ICE required for CD95 killing—lymphocytes from ICE knockout mice show resistance to CD95-induced apoptosis (11Kuida K Lippke J.A Ku G Harding M.W Livingston D.J Su M.-S Flavell R.A Science. 1995; 267: 2000-2003Crossref PubMed Scopus (1403) Google Scholar)—but following CD95 activation there appears to be an ordered sequence of two distinct ICE-like protease activities. The initial activity has a preference for a YVAD tetrapeptide substrate (as ICE itself does), whereas the later activity preferentially cleaves a DEVD tetrapeptide substrate (CPP32β-like) (7Enari M Talanian R Wong W Nagata S Nature. 1996; 380: 723-726Crossref PubMed Scopus (956) Google Scholar). Both activities are required for CD95-induced apoptosis (7Enari M Talanian R Wong W Nagata S Nature. 1996; 380: 723-726Crossref PubMed Scopus (956) Google Scholar). However, MACH/FLICE exhibits no activity against a synthetic YVAD-derived substrate but cleaves DEVD substrates instead. It remains possible that a further ICE family protease exists as an intermediate between FLICE and ICE or that ICE can be directly activated by FLICE under physiological conditions obtaining within the cell. Either way, it is likely that CD95 ligand binding results in activation of only a small number of FLICE molecules, which then directly or indirectly activate ICE as part of an amplification step: ICE then in turn activates the downstream DEVD-specific proteases, which are the final effectors of the proteolytic cascade (see Figure 1). The direct link between ICE-related proteases and TNFR1/CD95 signaling underscores the critical role that the pro-domains may play in ICE-related protease regulation—it is the pro-domain of FLICE/MACH containing two DED modules that links the protease with the receptors. ICE-related proteases are typically categorized on the basis of their substrate specificities, but they also differ in the size of their pro-domains. Some, like ICH-1, ICE, FLICE/MACH, and CED-3, have extensive pro-domains, while others, such as CPP32β, Tx, MCH-2, and MCH-3, have almost none. The fact that one strong inactivating mutant of CED-3 lies within its pro-domain indicates that the pro-domain plays a vital regulatory role (16Shaham S Horvitz H Genes Dev. 1996; 10: 578-591Crossref PubMed Scopus (195) Google Scholar). It seems likely that the extensive pro-domains of some ICE-related proteases act as interfaces with regulatory machinery: ICE-related proteases with very short pro-peptides may simply execute the death program. Pro-domains and their interactions are likely to be an area of intense future interest. The pathway from stimulus to activation of cell death machinery exemplified by the CD95/TNFR1 pathway is attractive in its neatness and relative simplicity. It remains to be seen how generally it is employed by other triggers of apoptosis such as DNA damage, oncogene expression, survival factor withdrawal, and alterations in the levels of BCL-2 family proteins. However, if Drosophila biology is any guide to vertebrate cell physiology (a perennial chestnut in the great debate about “relevance”), then one of the most intriguing observations may be that radiation-induced cell death in Drosophila embryos depends largely on expression of the DD protein Reaper (20White K Grether M.E Abrams J.M Young L Farrell K Steller H Science. 1994; 264: 677-683Crossref PubMed Scopus (872) Google Scholar). The CD95/TNFR1 signaling pathway may be a more ancient and generic conduit of lethality than is commonly supposed.