Apoptosis (programmed cell death) is required for the removal of infected, damaged or unwanted cells and its disrupted regulation is implicated in cancer, autoimmunity and degenerative disorders. At the molecular level, multiple signaling pathways converge on a family of cysteine proteases (caspases), which when activated cause cellular destruction by cleaving a range of vital cellular substrates. The Bcl-2 family of proteins are key regulators of many, but not all, signals leading to caspase activation. Here we highlight their features, which are extensively discussed in many notable reviews (see Borner, 2003; Cory and Adams, 2002; Cory et al., 2003; Gross et al., 1999; Strasser et al., 2000).
Evolutionary conservation of the cell death machinery from worms to man
Bcl-2 related proteins form part of the core apoptotic machinery conserved in species as diverse as Caenorhabditis elegans and mammals. Functionally, the Bcl-2-related proteins either inhibit or promote apoptosis, and interaction(s) between proteins belonging to opposing factions determines whether a cell lives or dies. Perhaps the best understood pathway is that in the worm C. elegans, where detailed genetic studies have shown that two Bcl-2 related proteins (pro-apoptotic EGL-1 and pro-survival CED-9) are essential for controlling developmentally programmed somatic cell deaths (Horvitz, 1999). Expression of EGL-1, the death trigger, is induced by damage signals. Binding of EGL-1 to CED-9, the worm Bcl-2 ortholog, releases the adapter protein CED-4 from CED-9. Once released, CED-4 binds to and activates the caspase CED-3 to cause cellular demise.⇓
(See poster insert)
Conserved domains characterize the family
Mammals bear at least five homologs of pro-survival CED-9, namely Bcl-2, Bcl-xL, Bcl-w, Mcl-1 and A1: all inhibit apoptosis during development and in response to cellular stress. They share with CED-9 at least three conserved BH (Bcl-2 homology) domains. Three of these (BH1, BH2 and BH3) fold together to form a hydrophobic groove on the pro-survival molecules (Muchmore et al., 1996). This groove is the target for ligand binding by pro-apoptotic EGL-1 or its mammalian counterparts Bik/Nbk/Blk, Bid, Bad, Hrk/DP5, Bim/Bod, Noxa, Puma/Bbc3, Bmf and Bcl-Gs. These killer proteins (BH3-only proteins) share the short BH3 domain, but no other notable sequences, with the wider Bcl-2 family (Huang and Strasser, 2000). Binding of BH3-only proteins to their cognate partners occurs through the interaction between the hydrophobic face formed by amphipathic α-helical BH3 domain and the hydrophobic groove (formed by BH1-BH3) of the pro-survival proteins (Petros et al., 2000; Sattler et al., 1997). In addition to the BH3-only proteins, the other class of pro-apoptotic Bcl-2 proteins is the multi-domain Bax-like proteins. They have remarkable sequence and structural similarity to their pro-survival cousins (Suzuki et al., 2000) but function instead to promote cell death, probably at a step distinct from the BH3-only proteins.
BH3-only proteins: sensors of cellular well-being
Biochemical and genetic experiments have suggested that the BH3-only proteins function as cell death triggers (Huang and Strasser, 2000). For example, loss of EGL-1 or Bim leads to excess cells during development, and their resistance to many damage signals. In order to prevent inappropriate cell deaths, the BH3-only proteins are held in check by a number of mechanisms until cellular insult relieves the brakes (Puthalakath and Strasser, 2002). In the hermaphrodite-specific neurons of C. elegans, transcription of EGL-1 is repressed by TRA-1A until this is removed by developmental cues (Conradt and Horvitz, 1999). Transcriptional control of mammalian BH3-only proteins has also been described for Hrk/DP5, Bim, Noxa and Puma. The latter two proteins may be key mediators of cell death induced by the tumor suppressor protein p53 in response to DNA damage (e.g. Nakano and Vousden, 2001). Post-transcriptional controls, by sequestration or protein modification, have also been described. Bim and Bmf are sequestered to the cytoskeletal structures and stress signals trigger their release, perhaps by inducing their phosphorylation (Lei and Davis, 2003). In contrast, survival factors inactivate Bad in healthy cells by phosphorylation that promotes its binding to 14-3-3 scaffold proteins. Full-length Bid appears to have little activity, but cleavage by caspases produces an active C-terminal fragment (tBid) that is a potent killer when targeted to the mitochondria by myristoylation of its N-terminus.
Activation of caspases, the cell death executioners
The different mammalian BH3-only proteins are coupled to distinct (but overlapping) stress signaling pathways. When damage signals remove the normal controls imposed on them, they are released to bind and inactivate pro-survival Bcl-2 proteins. This cellular switch may be partly regulated by the location of these proteins as BH3-only proteins target pro-survival Bcl-2 proteins located on the cytoplasmic face of the nuclear envelope, endoplasmic reticulum and outer mitochondrial membranes. Most attention has focused on the role of Bcl-2 proteins on mitochondria because when the outer mitochondrial membrane is breached, pro-apoptogenic factors such as cytochrome c, Smac/Diablo and HtrA2/Omi are released from the mitochondrial inter-membrane space (Martinou and Green, 2001). Cytochrome c triggers the formation of a holoenzyme ('apoptosome') with Apaf-1, the sole mammalian CED-4 homolog discovered to date, and the initiator caspase, caspase-9, to activate downstream caspases (Zou et al., 1997). However, mammalian pro-survival Bcl-2 proteins do not function to directly sequester Apaf-1, unlike C. elegans CED-9 sequestering CED-4 (Moriishi et al., 1999). Furthermore, the cytochrome c/Apaf-1/Caspase-9 pathway activated by stress signals does not appear to be the sole route for caspase activation in many cell types (Marsden et al., 2002). Thus, Bcl-2 must control step(s) upstream of, and in addition to, the release of mitochondrial cytochrome c (Cory and Adams, 2002).
Multi-domain Bax-like proteins
The multi-domain Bax-like sub-family may be responsible for damaging the outer mitochondrial membrane, permitting release of factors such as cytochrome c, although how this occurs is controversial (Kuwana et al., 2002; Roucou et al., 2002). Bax and Bak are functionally redundant as loss of either protein alone causes mild abnormalities, but their combined loss causes marked accumulation of redundant tissues (Lindsten et al., 2000). Interestingly, killing by BH3-only proteins is also abrogated (Cheng et al., 2001; Zong et al., 2001), but it is unclear whether BH3-only proteins act directly on Bax/Bak to activate them. Such a model appears attractive because Bax also bears a hydrophobic groove that may be the real target for binding, and consequent killing, by BH3-only proteins. BH3 binding may promote the membrane translocation of cytosolic Bax, its conformational alteration and aggregation, events associated with Bax activation. Surprisingly little data supports direct binding of BH3-only proteins to Bax/Bak (Wang et al., 1996; Wei et al., 2000) or their presence in Bax/Bak-containing complexes (Antonsson et al., 2001; Nechushtan et al., 2001). An alternative possibility is that pro-survival Bcl-2-like proteins control Bax/Bak activation directly or indirectly: BH3 binding abolishes this activity thereby permitting Bax/Bak activation.
Sticky tails on the road to a sticky end
How might BH3-binding abrogate the activity of pro-survival Bcl-2? In addition to the conserved hydrophobic groove formed by BH1-BH3 on pro-survival Bcl-2 targeted for BH3 binding, most of the pro-survival molecules also have a C-terminal hydrophobic region. Although this region may function as a transmembrane domain allowing membrane insertion, the 3D structure of pro-survival Bcl-w reveals that the hydrophobic tail is not normally exposed but instead occupies the hydrophobic groove targeted by BH3-only ligands (Denisov et al., 2003; Hinds et al., 2003). BH3 binding would displace the tail permitting tighter membrane interaction and this step inactivates the pro-survival molecule (Wilson-Annan et al., 2003). A similar mechanism has been proposed to activate Bax as its C-terminus, which is required for biological activity and is not normally exposed (Suzuki et al., 2000). As discussed, the direct trigger for releasing the tail of Bax and its membrane translocation is unknown (Guo et al., 2003). Understanding how BH3-only proteins inactivate Bcl-2 and how this leads to the activation of Bax/Bak may be the key to understanding the function of these proteins.
Perspectives
Ever since the cloning of Bcl-2 and the discovery of its biological function (Vaux et al., 1988), the Bcl-2 family of proteins have been shown to be key regulators of apoptosis and tremendous advances have occurred in understanding the underlying molecular mechanisms of cell death control. However, many questions remain. Are the proteins of each class functionally equivalent (Nijhawan et al., 2003)? How does pro-survival Bcl-2 function? Do Bcl-2 and its pro-survival homologs, like C. elegans CED-9, directly control mammalian CED-4 orthologs, since over-expression of human Bcl-2 in worms can compensate for the loss of CED-9 (Hengartner and Horvitz, 1994)? Alternatively, does Bcl-2 directly control the integrity of mitochondrial and other intracellular membranes (Kaufmann et al., 2003; Scorrano et al., 2003)? Could this family of proteins control membrane integrity indirectly, by impinging on membrane fusion and fission (Karbowski et al., 2002)? Does Bcl-2 function only to act as a `sink' for BH3-only proteins thereby preventing Bax/Bak activation, the real targets for BH3-only ligands (Letai et al., 2002)? Alternatively, could Bcl-2 control the activation of Bax/Bak by other mechanisms and, if so, why does human Bcl-2 function in worms, which have no apparent Bax-like orthologs? Do BH3-only proteins kill solely by binding to pro-survival Bcl-2, or is there a yet to be identified target for them? The coming years offer exciting possibilities for further insight as more of this ancient, essential and ultimately fascinating cellular process is unraveled.
Acknowledgements
Work in our laboratories is supported by grants and fellowships from the NHMRC (Australia, 257502), the US NCI (CA80188 and CA43540), the Leukemia & Lymphoma Society of America (Specialized Center of Research; 7015-02), the Marsden Fund (NZ) and the Sylvia & Charles Viertel Charitable Foundation (Australia). We thank our many colleagues especially J. Adams, P. Bouillet, S. Cory, H. Puthalakath, A. Strasser and D. Vaux for their continuing input into our work. We apologize to the many scientists and colleagues who are not cited owing to space limitations.
- © The Company of Biologists Limited 2003