The BH3-only members of the Bcl-2 protein family are essential initiators of programmed cell death and are required for apoptosis induced by cytotoxic stimuli. These proteins have evolved to recognise distinct forms of cell stress. In response, they unleash the apoptotic cascade by inactivating the protective function of the pro-survival members of the Bcl-2 family and by activating the Bax/Bax-like pro-apoptotic family members.
Apoptosis is the physiological process used by an organism to selectively eliminate cells that are no longer needed, have been damaged or are dangerous ( Kerr et al., 1972). This process, critical for sculpting organs during development and ensuring homeostasis throughout life, has been conserved during evolution ( Vaux and Strasser, 1996). Defects in the control of apoptosis have been implicated as a cause or a contributing factor in a variety of diseases. For example, abnormal survival of cells that should be killed can cause cancer ( Strasser et al., 1990) or autoimmune disease ( Bouillet et al., 1999; Strasser et al., 1991b; Watanabe-Fukunaga et al., 1992), whereas premature death of normally long-lived cells may be the cause of certain degenerative disorders ( Barr and Tomei, 1994).
Genetic and biochemical studies have identified two major pathways to programmed cell death that are largely independent ( Strasser et al., 1995). On the one hand, apoptosis can be triggered by ligation of a subgroup of the tumour necrosis factor receptor (TNF-R) family of cell surface receptors, `death receptors' (e.g. CD95/Fas/APO-1 or TNF-R1). This apoptosis signalling pathway is sometimes called the `extrinsic pathway'. On the other hand, apoptosis can also be initiated by a diverse range of stress conditions. The central pathway activated by these apoptotic stimuli is sometimes called the `intrinsic pathway' or the `mitochondrial pathway'. We believe that these names are not ideally suited, firstly, because this pathway can be initiated by extrinsic signals, such as cytokine withdrawal or γ-irradiation, and secondly, because mitochondria, although affected by this process, are not required for cell execution in all cell types and in response to all stress stimuli (see below). We propose to call this the `Bcl-2 regulated pathway', because the Bcl-2 family of proteins are critical regulators of this process ( Adams and Cory, 2001). Amongst the members of the Bcl-2 family, the BH3-only proteins have now been recognised as essential initiators of programmed cell death and stress-induced apoptosis ( Huang and Strasser, 2000). This review focuses on the `Bcl-2-regulated pathway', particularly on the role of the BH3-only proteins in the apoptosis signalling cascade.
Apoptotic cell death is characterised by a series of morphological and biochemical changes such as plasma membrane blebbing, chromatin condensation, internucleosamal DNA cleavage and exposure of phospatidyl serine on the extracellular side of the plasma membrane. Genetic and biochemical studies in Caenorhabditis elegans, Drosophila melanogaster and mammals have led to the identification of the main players of the cell death machinery and have shown that this process has been conserved throughout evolution ( Strasser et al., 2000; Vaux and Strasser, 1996). The collapse of the cell is brought about by the action of aspartate-specific cysteine proteases termed caspases ( Thornberry and Lazebnik, 1998). Caspases are normally present in healthy cells as zymogens with low enzymatic activity. They become activated through proteolysis by already active caspases or through autocatalytic processing, which is mediated by aggregation of zymogens in a complex containing adapter proteins (e.g. Apaf-1, C. elegans CED-4 and FADD) and co-factors (e.g. ATP and cytochrome c) ( Thornberry and Lazebnik, 1998).
Proteins of the Bcl-2 family are critical regulators of caspase activation and apoptosis ( Adams and Cory, 1998; Gross et al., 1999). The anti-apoptotic members of the Bcl-2 family (Bcl-2, Bcl-xL, Bcl-w, Mcl-1, A1, Boo/Diva/Bcl-2-L10, Bcl-B and C. elegans CED-9) all contain three or four characteristic regions of homology (BH1-4; Bcl-2 Homology domains). According to their structure and biochemical function (see below), the pro-apoptotic Bcl-2 family members can be divided into two subgroups. Bax, Bak, Bcl-xS, Bok/Mtd and Bcl-GL contain two or three BH domains, whereas Bad, Bik/Nbk, Blk, Bid, Hrk/DP5, Bim/Bod, Bmf, Noxa, Puma/Bbc-3 and C. elegans Egl-1 share with each other and the rest of the family only the short (9-16 amino acid) BH3 domain ( Fig. 1) ( Huang et al., 2000). The BH3 domain is essential for the binding of these BH3-only proteins to the anti-apoptotic members of the family and for their ability to kill cells ( Huang et al., 2000). Hetero-dimerisation is mediated by the insertion of the BH3 domain of the pro-apoptotic molecules into a hydrophobic cleft formed by the BH1, BH2 and BH3 domains on the surface of the anti-apoptotic proteins ( Sattler et al., 1997). Many pro- as well as anti-apoptotic members of the Bcl-2 family also have a C-terminal transmembrane domain, which can target these proteins to the cytoplasmic side of intracellular membranes of the nucleus, endoplasmic reticulum and mitochondria ( Chen-Levy and Cleary, 1990; Lithgow et al., 1994). How the Bcl-2 family of proteins regulates apoptosis is still controversial ( Adams and Cory, 1998; Green and Reed, 1998; Gross et al., 1999; Strasser et al., 2000) and possible mechanisms of the biochemical action of these molecules are discussed below.
The essential roles of the various BH3-only proteins
One thing that has become clear is that BH3-only proteins are essential initiators of programmed cell death in species as distantly related as C. elegans ( Conradt and Horvitz, 1998) and mice ( Bouillet et al., 1999). In C. elegans, a single BH3-only protein, EGL-1, is required for the initiation of all developmentally programmed deaths of somatic cells ( Conradt and Horvitz, 1998). Apoptosis initiation is clearly more complex in mammals, which have at least 8 BH3-only proteins ( Huang et al., 2000). Multiplicity can generate redundancy; so do individual BH3-only proteins have a particular role that the others do not have? The only knockout experiments of BH3-only proteins reported so far involve Bid and Bim. Bid-deficient mice appear normal, but their hepatocytes (but not their lymphocytes) are resistant to anti-Fas antibody-induced killing ( Yin et al., 1999). In contrast, Bim is required for haematopoietic cell homeostasis and as a barrier against autoimmune disease ( Bouillet et al., 1999). Interestingly, Bim-deficient lymphocytes are resistant only to certain Bcl-2-inhibitable apoptotic stimuli, including cytokine withdrawal or treatment with ionomycin or taxol, but they are almost normally sensitive to others, such as γ-radiation or treatment with phorbol ester (PMA) or dexamethasone ( Bouillet et al., 1999; Bouillet et al., 2002). We therefore speculate that in mammals, different BH3-only proteins sense different stress conditions and are essential for initiation of apoptosis in response to these stimuli. Indeed, Bim is activated by cytokine withdrawal and calcium flux ( Bouillet et al., 1999; Puthalakath et al., 1999), whereas Bmf is activated by loss of cellular attachment— anoikis ( Puthalakath et al., 2001). It appears likely that different BH3-only proteins (or combinations of BH3-only proteins) are critical for programmed cell death in specific tissues. We therefore expect that the generation and analysis of mutant mice lacking individual BH3-only proteins or combinations of these `killers' will provide interesting insight into animal development and apoptosis regulation.
Regulation of BH3-only proteins
The pro-apoptotic activity of BH3-only proteins is subject to stringent control, both at the transcriptional and post-translational level. In C. elegans, the transcriptional repressor TRA-1A regulates EGL-1 function in a group of neurons that is required for egg-laying ( Conradt and Horvitz, 1999). The TRA-1A sex determination factor is expressed in hermaphrodites where these neurons have an essential role, but TRA-1A is not expressed in males, where these cells are killed by EGL-1. It is thought that different transcription factors control Egl-1 expression in other somatic cells that are programmed to die, and it is possible that EGL-1 function can also be regulated post-translationally in certain cell types.
Evolution has led to the existence of many different BH3-only proteins in mammals and other vertebrates (e.g. frogs, fish and birds), and several mechanisms have been put in place to keep these killer proteins in check ( Huang and Strasser, 2000). As is the case for C. elegans EGL-1, the activity of some mammalian BH3-only proteins is also regulated at the transcriptional level. Noxa and puma/Bbc3 have both been identified as p53-inducible genes ( Han et al., 2001; Nakano and Wousden, 2001; Oda et al., 2000; Yu et al., 2001) and are therefore thought to be critical for DNA damage-induced apoptosis. A different stress stimulus, growth factor deprivation, causes increased hrk/dp5 and bim mRNA expression in neurons by a JNK-dependent mechanism ( Harris and Johnson, 2001; Imaizumi et al., 1997; Inohara et al., 1997; Putcha et al., 2001; Whitfield et al., 2001). In contrast, in haematopoietic cells cytokine withdrawal has been reported to augment bim expression through activation of the forkhead transcription factor FKHR-L1 ( Dijkers et al., 2000).
Pro-apoptotic activity of BH3-only proteins can also be regulated post-transcriptionally. For example, BimEL and BimL, the two most abundantly expressed isoforms of the Bim gene ( O'Connor et al., 1998; O'Reilly et al., 2000), are sequestered to the microtubular dynein motor complex by binding to the dynein light chain DLC-1/LC8 ( Puthalakath et al., 1999). Certain apoptotic stimuli cause the release of Bim (still associated with DLC-1) from the cytoskeleton and allow it to translocate to mitochondria and the nuclear envelope, where it can bind to and antagonise the function of pro-survival Bcl-2 molecules ( Puthalakath et al., 1999). Interestingly, Bmf is regulated in a similar way by binding to another dynein light chain molecule, DLC-2, and sequestration to myosin V motors on the actin cytoskeleton ( Puthalakath et al., 2001). This difference in subcellular localization may account for the fact that Bmf is activated by cellular detachment (anoikis), whereas Bim senses the effects of cytokine deprivation, abnormal calcium flux and treatment with taxol.
Bid can be cleaved by caspase-8 and certain other caspases ( Li et al., 1998; Luo et al., 1998). The truncated p15 tBid polypeptide is thought to trigger apoptosis more efficiently than full-length Bid because the proteolytic fragment can be myristoylated, which promotes its translocation to intracellular membranes (e.g. on mitochondria) where anti-apoptotic Bcl-2 relatives reside ( Zha et al., 2000). Moreover, the cleavage of Bid by caspase-8 has been reported to be attenuated by phosphorylation by casein kinase I and casein kinase II ( Desagher et al., 2001). Bad, on the other hand, is phosphorylated in response to cytokine signalling by Akt/PKB ( Datta et al., 1997; del Peso et al., 1997) or by the mitochondrial kinase PKA ( Harada et al., 1999). Phosphorylated Bad is sequestered in the cytosol by binding to 14-3-3τ scaffold proteins, and cytokine withdrawal causes de-phosphorylation of Bad, thereby allowing it to break away from 14-3-3τ proteins and to translocate and bind to pro-survival Bcl-2 molecules ( Zha et al., 1996). Bik has recently also been shown to be regulated by phosphorylation, but, unlike in the case of Bid and Bad, phosphorylation of Bik somehow increases its pro-apoptotic activity ( Verma et al., 2001). Collectively, these data indicate that the subcellular localisation of BH3-only proteins plays a critical role in cell death control, enabling the cell to react rapidly and efficiently to external and internal death signals, but keeping pro- and anti-apoptotic Bcl-2 molecules apart in healthy cells.
Functional interactions between BH3-only proteins and anti-apoptotic Bcl-2 family members
The multiplicity and probable redundancy of pro- as well as anti-apoptotic Bcl-2 family members in mammals has made it difficult to discern their physiologically relevant interactions. Theoretically, specificity could be generated by preferential binding of particular BH3-only proteins to particular anti-apoptotic Bcl-2-like molecules. There is so far no evidence for or against this theory and we believe that clarification of this important issue requires careful measurements of the affinities of these protein-protein interactions. Alternatively, specificity of interaction of particular BH3-only proteins with particular anti-apoptotic Bcl-2 family members could simply be a reflection of their expression patterns. Interestingly, the haematopoietic features of Bim-deficient mice are opposite to those found in mice lacking Bcl-2 ( Matsuzaki et al., 1997; Veis et al., 1993), perhaps indicating that, at least in those cells, Bim might be the critical initiator for apoptosis and Bcl-2 the essential guardian. Apart from a fragile lymphoid and myeloid system, Bcl-2-deficient mice also have defects in survival of stem cells for renal epithelial cells and melanocytes. Consequently, bcl-2-/- mice become runted, turn gray and succumb to polycystic kidney disease ( Kamada et al., 1995; Nakayama et al., 1994; Veis et al., 1993). Remarkably, all the deficiencies caused by the absence of Bcl-2 could be efficiently rescued by the concomitant absence of Bim ( Bouillet et al., 2001). Gene dosage was shown to be a main feature of Bim function, as removal of a single Bim allele in bcl-2-/- mice was sufficient to restore normal kidney development, normal weight gain, normal lifespan and also improved lymphocyte survival ( Bouillet et al., 2001). These observations demonstrate that in several cell types — kidney epithelial cells, melanocytes, lymphocytes and myeloid cells — Bcl-2 is the critical guardian of cell survival, and in its absence Bim causes these cells to die. These data may also indicate that BH3-only proteins are involved in the induction of degenerative diseases. If this can be proved, these molecules may be interesting targets for new drugs aimed at inhibiting their function.
Functional interactions between BH3-only proteins and Bax/Bak-like multi-domain pro-apoptotic Bcl-2 family members
Until recently it was unclear whether the multi-BH domain Bax/Bak-like proteins function, similarly to BH3-only proteins, as stress-activated intracellular death ligands or have a different role in the apoptosis machinery. Experiments with cells from mice lacking both Bax and Bak (bax-/-bak-/-) have demonstrated that Bax and Bak have a role distinct from that of the BH3-only proteins. Although mice lacking either Bax or Bak alone had little or no abnormality, those lacking both proteins displayed multiple developmental defects and only a few survived to adulthood ( Lindsten et al., 2000). Interestingly, the progressive lymphoid hyperplasia in bax-/-bak-/- mice is very similar to that observed in mice lacking Bim ( Bouillet et al., 1999) or those overexpressing Bcl-2 in the lymphoid ( McDonnell et al., 1989; Sentman et al., 1991; Strasser et al., 1991a; Strasser et al., 1991b) or in the entire haematopoietic compartment ( Ogilvy et al., 1999). Interestingly, Bim and other BH3-only proteins, such as Bid, Bad and Noxa, were unable to kill embryonic fibroblasts lacking both Bax and Bak but could efficiently kill cells having only one copy of either of these two genes ( Cheng et al., 2001; Zong et al., 2001). Cultured neurons lacking only Bax cannot be killed by transfection with expression constructs encoding Bim, Hrk or Bid ( Putcha et al., 2001). This may indicate that Bak is not expressed in neurons or that it cannot compensate for Bax in this cell type, at least in these experiments in tissue culture. Regardless, these findings demonstrate that BH3-only proteins require the presence of Bax/Bak-like proteins for their ability to kill cells. Conversely, Bax/Bak-like proteins require for cell killing a signal from BH3-only proteins, which sense and are activated by cell stress. For example, cytokine-withdrawal-induced apoptosis is greatly diminished in bim-/- lymphocytes although they have a full complement of Bax and Bak ( Bouillet et al., 1999). Interestingly, the death rate of the bim-/- B cells is not even increased by concomitant loss of Bcl-2, although Bcl-2 is the critical guardian against apoptosis in these cells ( Bouillet et al., 2001).
Collectively, these results demonstrate that BH3-only proteins and Bax/Bak-like proteins have distinct but interdependent functions that are both essential for initiation of apoptosis. Whether the two types of pro-apoptotic proteins are part of the same linear pathway or act in parallel, both impinging on the Bcl-2-like pro-survival proteins, is presently not clear ( Fig. 2).
Regulation of Bax/Bak-like pro-apoptotic Bcl-2 family members
In contrast to the dramatic effects caused by loss of a single allele of the BH3-only gene bim ( Bouillet et al., 2001; Bouillet et al., 1999), the gene dosage of bax and bak does not appear to be limiting, since mice that have only a single copy of either of these genes (bax+/-bak-/- or bax-/-bak+/- mice) appear normal ( Lindsten et al., 2000). This may indicate that transcriptional upregulation of Bax and Bak does not play a critical role in cell death control. However, in common with many of the BH3-only proteins, Bax and Bak are stringently regulated at the post-translocational level. In healthy cells, Bax is normally found as a monomer in the cytoplasm or loosely attached to the outer mitochondrial membrane ( Desagher et al., 1999; Hsu et al., 1997; Nechushtan et al., 1999; Wolter et al., 1997), whereas most Bak molecules appear to be attached (probably loosely) to mitochondria ( Krajewski et al., 1996). In response to apoptotic stimuli, Bax changes its conformation and translocates from the cytoplasm to mitochondria where it undergoes oligomerisation ( Hsu et al., 1997; Wolter et al., 1997). Although less is known about Bak, it appears that it undergoes similar conformational changes and oligomerisation in stressed cells ( Griffiths et al., 1999; Nechushtan et al., 2001; Wei et al., 2000). The C-terminal α-helix appears to play a critical role in controlling subcellular localisation of Bax. Structural analysis has shown that this region occupies the hydrophobic pocket formed by the BH1, BH2 and BH3 domains ( Suzuki et al., 2000). This contributes to the solubility of the molecule by reducing the exposed hydrophobic residues, which is consistent with the localisation of monomeric Bax in the cytoplasm. Apoptotic stimuli also elicit conformational changes at the N-terminus of Bax, because antibodies specific for an epitope in this region bind to Bax only in stressed but not in healthy cells ( Desagher et al., 1999). Whether release of the C-terminal α-helix, exposure of the N-terminal epitope, translocation and aggregation of Bax occur concomitantly or sequentially is still controversial. These steps are thought to be critical for the initiation of stress-induced apoptosis and, according to this theory, retention of Bax in the cytoplasm may represent a safeguard against inadvertent activation of the cell death effector machinery. Consistent with this idea, mutants of Bax that are constitutively localised at the outer mitochondrial membrane are more potent killers than wild-type Bax ( Suzuki et al., 2000), and enforced Bax dimerisation induces apoptosis ( Gross et al., 1998). Since dimerisation and translocation of Bax to mitochondria has been observed in stressed cells that survive this damage ( Makin et al., 2001), these changes must, however, represent steps that occur prior to commitment to apoptosis and are not sufficient to kill cells.
Possible models for the biochemical function of the Bcl-2 family members
In C. elegans, the process of cell death control appears relatively simple ( Fig. 2). In healthy cells, the caspase activator CED-4 appears to be restrained by direct binding to the Bcl-2 homologue CED-9 ( Chinnaiyan et al., 1997; Spector et al., 1997; Wu et al., 1997), and expression of the BH3-only protein EGL-1 is repressed ( Conradt and Horvitz, 1999). In cells programmed to die, EGL-1 levels increase, EGL-1 binds to CED-9 and thereby liberates CED-4 to promote aggregation-induced autocatalytic activation of the caspase CED-3 ( Chen et al., 2000). Although such a sequestration model was initially reported to be valid as well in mammalian cells, subsequent studies failed to detect direct binding of the caspase adapter Apaf-1 to any of the pro-survival Bcl-2 family members ( Conus et al., 2000; Hausmann et al., 2000; Moriishi et al., 1999).
The studies in the bax-/-bak-/- and bim-/- mice strongly suggest that Bax/Bax-like multi-BH domain and the BH3-only pro-apoptotic proteins function at different levels of a linear pathway to death. It is commonly accepted that some mechanism of Bax/Bak aggregation is critical for cell killing, but how this signal is translated into cell destruction is still hotly debated. According to one model, the principal function of multi-BH domain pro-apoptotic proteins is to disrupt mitochondrial membrane integrity and allow the release of cytochrome c (and other pro-apoptotic molecules). It has been proposed that Bax/Bak-like proteins mediate this process either by binding to and modifying mitochondrial channel proteins (VDAC or ANT) ( Marzo et al., 1998; Narita et al., 1998) or by direct pore formation ( Antonsson et al., 1997; Eskes et al., 1998; Nechushtan et al., 2001; Pavlov et al., 2001). How Bcl-2-like pro-survival molecules would block this activity of the Bax/Bak-like proteins is unclear. Inhibition may happen by direct binding of Bcl-2 to Bax ( Oltvai and Korsmeyer, 1994), but several reports suggest that such an interaction does not occur physiologically within cells but may only be detected as a by-product of cell lysis when certain detergents are used ( Antonsson et al., 2001; Hsu and Youle, 1998; Nechushtan et al., 2001). In this regard, it is also noteworthy that certain mutants of Bcl-xL that are unable to bind to Bax/Bak (even in detergent lysates) can protect cells against apoptotic stimuli that were shown to be Bax/Bak-dependent ( Cheng et al., 1996). Although Bcl-2 is unable to prevent dimerisation of Bax or its translocation to the mitochondrial surface, it appears to block integration and aggregation of Bax/Bak in the outer mitochondrial membrane ( Antonsson et al., 2001; Nechushtan et al., 2001). The following scenario appears plausible. Upon an apoptotic signal, cohorts of BH3-only proteins, transcriptionally induced or waiting in different places in the cytoplasm, are unleashed and move to the surface of the mitochondria and probably also to other intracellular membranes. These BH3-only proteins will bind to the prosurvival members of the Bcl-2 family, and this will somehow facilitate membrane integration and aggregation of Bax/Bak-like proteins ( Fig. 3). It has been proposed that Bax/Bak aggregation is induced by their binding to some of the BH3-only proteins, as has been reported for Bid ( Eskes et al., 2000). However, there is no evidence so far that binding of any of the BH3-only proteins to Bax or Bak occurs with an affinity that is biologically meaningful. Collectively, these observations indicate that Bcl-2-like pro-survival molecules can prevent membrane integration and aggregation of Bax/Bak-like proteins until they are antagonised by BH3-only proteins.
Several experimental observations challenge the idea that outer mitochondrial membrane disruption and release of cytochrome c are absolutely required for apoptosis initiation ( Huang and Strasser, 2000; Strasser et al., 2000). Indeed, a number of developmental processes, such as cavitation or organ morphogenesis, which require cell death for their completion, still occur normally in embryos lacking either Apaf-1 ( Cecconi et al., 1998; Yoshida et al., 1998), caspase-9 ( Hakem et al., 1998; Kuida et al., 1998), cytochrome c ( Li et al., 2000), caspase-3 ( Kuida et al., 1996) or Bax/Bak ( Lindsten et al., 2000). Moreover, we have shown that apoptosis occurs normally in a number of haematopoietic cell types lacking either Apaf-1 or caspase 9 (V. Marsden and A.S., unpublished). We have therefore postulated that cell death is initiated by one or several CED-4-like caspase adapters that are directly regulated by interaction with members of the Bcl-2 family ( Strasser et al., 2000) ( Fig. 3). According to this model, cytochrome c and Apaf-1-mediated caspase-9 activation form part of an amplification loop in apoptosis that is essential in some cell types (e.g. developing neurons) but dispensable in others (e.g. haematopoietic cells). Further genetic and biochemical experiments are needed to determine which of these two mechanisms is the critical initiator of programmed cell death and whether they act in parallel or in a linear pathway.
Another puzzling conundrum is that C. elegans has only a pro-survival Bcl-2 homologue, CED-9, and a BH3-only protein, EGL-1, but, in contrast to mammals and flies, apparently lacks Bax/Bak-like multi-BH domain pro-apoptotic Bcl-2 family members. CED-9 is not only essential for cell survival but has also been reported to promote apoptosis in nematodes with a certain genetic make-up (hypomorphic ced-3 mutation) ( Hengartner and Horvitz, 1994). It is therefore interesting to consider the possibility that CED-9 has features of both pro-survival Bcl-2-like and pro-apoptotic Bax/Bak-like molecules. Because the structure of Bax ( Suzuki et al., 2000) is remarkably similar to that of Bcl-xL ( Muchmore et al., 1996), it is possible that they represent the two different conformational states that CED-9 can assume, one when it is free and the other when it is bound to EGL-1. If one accepts this idea, it is possible that mammalian CED-4 homologues may interact with Bax/Bak-like molecules rather than with the pro-survival members of the Bcl-2 family. We therefore believe that it may be informative to solve the structures of free CED-9, EGL-1-CED-9 and CED-9-CED-4 complexes and compare them to those of free Bcl-2, free Bax and Bcl-2 bound to a BH3-only protein.
Proteins of the Bcl-2 family are major regulators of developmentally programmed cell death and stress-induced apoptosis. Recent genetic and biochemical studies indicate that members from both pro-apoptotic subgroups of the Bcl-2 family act in a concerted and mutually interdependent manner in cell destruction. The BH3-only proteins are unable to kill cells in the absence of the Bax/Bak-like multi-BH domain proapoptotic Bcl-2 family members, but it also appears that Bax/Bak-like proteins can only execute cells in which BH3-only proteins have been activated in response to stress stimuli. Mammals have at least eight BH3-only proteins and it appears that they have evolved to recognise different stress stimuli. In response to these death stimuli, they initiate the death cascade by antagonising the pro-survival Bcl-2-like molecules and by activating the Bax/Bak-like pro-apoptotic Bcl-2 family members. Further biochemical, structural and genetic analyses are needed to determine the molecular mechanisms that then lead to caspase activation and cell collapse.
We wish to thank all of our past and present colleagues, particularly D. Huang, H. Puthalakath, J. Adams, L. O'Reilly, A. Villunger, S. Cory, L. O'Connor, L.-C. Zhang, L. Cullen, K. Scalzo, L. Tai, M. Pellegrini, L. Coultas, K. Newton, V. Marsden, A. Harris, S. Bath, A. Egle, Y Laabi and D. Vaux for their help with our research and for discussions. Work in our laboratory is supported by grants and fellowships from the NHMRC (Canberra), the NIH, the Dr Josef Steiner Cancer Research Foundation (Bern, Switzerland), the Leukemia and Lymphoma Society of America, the Cancer Research Institute (New York) and the Anti-Cancer Council of Victoria.
- © The Company of Biologists Limited 2002