|Home | About | Journals | Submit | Contact Us | Français|
Apoptosis is essential for tissue homeostasis, particularly in the hematopoietic compartment, where its impairment can elicit neoplastic or autoimmune diseases. Whether stressed cells live or die is largely determined by interplay between opposing members of the Bcl-2 protein family. Bcl-2 and its closest homologs promote cell survival, but two other factions promote apoptosis. The BH3-only proteins sense and relay stress signals, but commitment to apoptosis requires Bax or Bak. The BH3-only proteins appear to activate Bax and Bak indirectly, by engaging and neutralizing their pro-survival relatives, which otherwise constrain Bax and Bak from permeabilizing mitochondria. The Bcl-2 family may also regulate autophagy and mitochondrial fission/fusion. Its pro-survival members are attractive therapeutic targets in cancer and perhaps autoimmunity and viral infections.
Programmed cell death plays critical roles in both development and tissue homeostasis. Indeed, in the immune system, apoptosis shapes the immune repertoire and refines and terminates immune responses [1*]. Since impaired apoptosis can lead to either autoimmunity or malignancy, there is intense interest in uncovering its control mechanisms and exploring the therapeutic options they provide.
Apoptosis is precipitated by sequential activation of cysteine proteases of the caspase family, in two distinct but converging pathways [2,3]. The extrinsic pathway activates Caspase-8 (and Caspase-10 in humans) when ligand-mediated trimerisation of ‘death receptors’ of the Tumor Necrosis Factor family on the plasma membrane recruits the adaptor protein FADD and the caspase into multi-protein complexes. The intrinsic pathway (also termed the ‘mitochondrial’ or ‘stress’ pathway) activates Caspase-9 on the scaffold protein Apaf-1 when cytochrome c is released from damaged mitochondria in response to diverse stresses, including cytokine deprivation and DNA damage. These initiator caspases can cleave and activate the effector caspases (-3, -6 and -7) that mediate cellular demolition by cleaving multiple critical cellular proteins.
The intrinsic pathway is controlled by the Bcl-2 protein family, which constitutes a tri-partite regulatory cassette [4*]. Bcl-2 itself, the oncoprotein discovered via the chromosome translocation that hallmarks human follicular lymphoma, inhibits apoptosis, as do its close homologs Bcl-xL, Bcl-w, Mcl-1, A1 and (in humans) Bcl-B. In contrast, two other sub-groups promote apoptosis: one comprises Bax and Bak (and the little-studied Bok), which also resemble Bcl-2, particularly in three “BH” (Bcl-2 Homology) domains, whereas the other group, which includes Bim, Bad, Bid, Bik, Bmf, Puma, Noxa and Hrk, shares only the BH3 domain. When activated by stress signals, these ‘BH3-only’ proteins insert that domain, an amphipathic α helix, into a hydrophobic groove on the pro-survival relatives . This coupling primes the cell for apoptosis, but commitment requires activation of Bax and/or Bak [6,7], which then form oligomers on intracellular membranes, including the mitochondrial outer membrane, and thereby perturb their integrity.
In reviewing the role of the Bcl-2 family in apoptosis, we will focus primarily on two areas receiving great attention: the vexed issue of how interactions between the warring Bcl-2 factions commit the cell to apoptosis and the exciting prospect of targeting this family as a new approach to the therapy of cancer and perhaps autoimmune and virally-induced diseases. We also briefly discuss recent evidence that the family also influences autophagy and mitochondrial morphology. Further background on the Bcl-2 family and apoptosis is provided by recent reviews [1*,4*,8,9*], including the accompanying ones.
Activated BH3-only proteins were thought to bind indiscriminately to all their pro-survival counterparts until quantitative studies revealed marked differences [10**]. Bim, Puma and tBid (the activated, truncated form of Bid) do bind avidly to all the pro-survival proteins, but the others associate only with subsets [10**]. For example, Noxa engaged only Mcl-1 and A1, and Bad only Bcl-2, Bcl-xL and Bcl-w. Importantly, the promiscuous binders killed much more potently than the selective binders, but the combination of the complementary binders Noxa and Bad readily induced cell death [10**]. These findings indicate that efficient apoptosis requires neutralization of multiple pro-survival proteins.
The BH3-only proteins clearly act upstream of Bax and Bak, because they cannot induce apoptosis in cells lacking both Bax and Bak [6,7]. Their role in the activation of Bax and Bak, however, is highly controversial. In the direct activation model (Figure 1A), a sub-group of BH3-only proteins, termed activators, are proposed to bind directly to Bax and Bak to promote their activation [11,12*,13,14*,15,16*]. The putative activators include Bim and tBid [11,12*,13,14*,15], and perhaps also Puma [16*,17], although that has been disputed [12*,13,18]. In this model, the remaining BH3-only proteins, termed sensitisers, function by binding to the pro-survival proteins and freeing any bound Bim or tBid to directly activate Bax and Bak.
The indirect activation model (Figure 1B) proposes instead that all BH3-only proteins function solely by binding to their pro-survival relatives, thereby preventing those guardians of cell survival from inhibiting Bax and Bak [10**,19**,20**,21]. In this model, Bim, tBid and Puma are the most potent BH3-only proteins simply because they can engage all the pro-survival proteins [10**].
Several recent findings strongly challenge the direct activation model [20**]. Firstly, no Bak co-immunoprecipitated with any BH3-only protein. Secondly, no Bax bound detectably to the physiologically relevant forms of Bim (BimEL and BimL), and none co-immunoprecipitated with Bim in dying cells. Thirdly, although tBid and a minor Bim isoform (BimS) bound Bax weakly, tBid and BimS bearing BH3 mutations that prevented binding to Bax but retained normal binding to pro-survival proteins remained just as potent as killers. Most tellingly, in response to several apoptotic stimuli, or enforced co-expression of Noxa and Bad, cells lacking both Bid and Bim died as readily as wild-type cells, and even down-regulation of Puma by RNAi in these cells did not impair apoptosis driven by several stimuli [20**]. Thus, none of the putative activator BH3-only proteins appears to be essential for apoptosis.
Mouse genetic studies also favor the indirect activation model. Mice lacking both Bim and Bid [20**], or both Bim and Puma , appear normal, and their cells remain normally sensitive to certain apoptotic signals. On the direct activation model, the absence of these activators should have blocked apoptosis and promoted severe developmental abnormalities, as in mice lacking both Bax and Bak .
Some viral Bcl-2 homologs, which may have been selected for resistance to inhibition by BH3-only regulators, also act by binding to and restraining Bax and Bak , further supporting the indirect activation model. Remarkably, M11L of myxoma virus, an anti-apoptotic protein lacking sequence homology to Bcl-2, has an extremely similar 3D fold [25*], and structure-based mutagenesis suggested that its critical binding targets were Bax and Bak rather than BH3-only proteins.
Collectively, these findings demonstrate that apoptosis does not rely upon direct activation of Bax or Bak by any known BH3-only protein (Figure 1A). Nevertheless, direct activation might occur in some circumstances. Protein-free liposomes can be lysed by cooperation of Bax with tBid  or with BH3 peptides from Bim or Bid at high (eg >10 micromolar) concentrations [12*], or at much lower levels if these peptides are chemically ‘stapled’ into an α helix and/or artificially targeted to membranes [14*,15]. Perhaps direct binding by Bim or tBid contributes to an amplification step in apoptosis.
In the indirect activation model, apoptosis is the default pathway and the Bcl-2 pro-survival proteins function mainly by constraining activation of Bax and/or Bak (Figure 2). The diverse phenotypes observed in mice on inactivation of individual pro-survival genes (see [4*] and the review by Opferman in this issue) presumably indicate that specific pro-survival proteins are the dominant guards on Bax/Bak in different cell types. For example, a Bcl-xL/Bak switch primarily controls the lifespan of anucleate platelets [27**]. Platelets exposed to a BH3 mimic that inactivates Bcl-xL (see below) underwent apoptosis, and the reduced platelet level found in bcl-x heterozygotes or mice bearing labile hypomorphic Bcl-xL mutants was ameliorated by concomitant Bak deficiency. The inference is that Bcl-xL is rate limiting for platelet survival, and once normal turnover drops its level below the threshold for constraining Bak, the unleashed Bak drives apoptosis [27**].
Inhibition of Bax and Bak by pro-survival Bcl-2 proteins relies, in part, upon direct association [19**, 20**]. In healthy cells, Bak can bind to either Mcl-1  or Bcl-xL (albeit, unexpectedly, not Bcl-2), and this interaction appears to require the Bak BH3 domain [19**]. To initiate apoptosis, BH3-only proteins seem to displace Bak from these pro-survival proteins. Importantly, Bak could drive cell death if and only if both Mcl-1 and Bcl-xL were inactivated by, for example, Noxa plus Bad [19**]. Bax also appears to be constrained in part by direct interaction with pro-survival proteins [20**]. The significance of their association has long been questioned, because it is often detectable only in non-ionic detergents, which change the conformation of Bax, perhaps mimicking its activation . Recent killing assays, however, have established that neutralisation of multiple pro-survival proteins, including Bcl-2, by BH3-only proteins is necessary and apparently sufficient for Bax activation (Figure 2) [20**].
How do the pro-survival proteins restrain Bax and Bak, given that the known structures of Bax  and Bak  have the BH3 domain buried? One proposal is that, in healthy cells, a small proportion of Bax and Bak (see below) is in a ‘primed’ state, with its BH3 domain exposed but restrained by binding to pro-survival proteins [19**] (Figure 2). Any primed Bax is likely to reside on the mitochondrial membrane, because the cytosolic form is monomeric . Rather than pre-existing, the primed conformers of Bax and Bak may instead form early in apoptotic signalling, either spontaneously or in response to an independent cytotoxic signal [20**]. For example, phosphorylation of Bax may regulate its translocation to the membrane [33,34].
The pro-survival proteins may also constrain Bax and Bak by a second, unknown mechanism not requiring their association via the Bax/Bak BH3 domain. BH3 mutants of Bak and Bax have been identified that cannot bind detectably to their pro-survival relatives but are still functional upon cytotoxic stimulation [16*] (J Fletcher, DSC Huang and JMA, unpublished results).
Since most tumors have defects in the p53 pathway or over-express a Bcl-2 homolog, interest is surging in the potential of anti-cancer drugs that, like the BH3 domain, bind one or more Bcl-2 homologs and trigger apoptosis [4*,9*]. Such ‘BH3 mimetics’ should be more effective than anti-tumor agents acting far upstream. Furthermore, BH3 selectivity [5,10**] heralds the prospect of specifically targeting the Bcl-2 homolog(s) required to maintain a particular tumor type, sparing more of the normal cells. Such drugs may also be applicable to certain autoimmune diseases and perhaps to infections with viruses that express responsive vBcl-2 homologs.
Although few drugs are peptides, BH3 peptides constrained as an α-helix by a chemical staple [14*,35], or by inclusion of unnatural amino acids [36,37], may have potential. The ability of a stapled Bid BH3 peptide to activate Bax, at least in vitro [14*], may mean that Bax (or Bak) could also be targeted, although that should kill more normal cells than a compound engaging selected pro-survival proteins.
Small organic BH3 mimics appear more promising. A structure-based approach led to ABT-737, which binds strongly (low nM affinity) to Bcl-2, Bcl-xL and Bcl-w but not to Mcl-1 or A1 [38**]. ABT-737 proved cytotoxic as a single agent for many samples of B lymphoid tumors and chronic lymphocytic leukemia (CLL), as well as small cell lung cancer cell lines. Significantly, it induced stable regression of lung cancer in mouse xenografts, with minimal side effects [38**]. It also killed AML cells but not normal hematopoietic progenitor cells in vitro and delayed leukemia in xenografts without collateral damage .
Unlike six other putative BH3 mimetics, ABT-737 behaved in the mechanism-based fashion, since it spared cells lacking both Bax and Bak [40*]. Notably, cells with high levels of Mcl-1, which the drug does not bind, proved much more refractory than those with little Mcl-1 (compare Figure 3A and B) [39,40*,41,42*,43], and the less frequently expressed A1 also promoted some resistance [40*]. However, down regulation of the short-lived Mcl-1 by different strategies, some potentially clinically applicable, conferred sensitivity [39,40*,41,42*,43]. Interestingly, the drug readily overcame high levels of Bcl-2 [40*]. Indeed, abundant Bcl-2 may actually render neoplasms such as CLL more sensitive, by sequestering an elevated level of a potent killer like Bim – presumably induced by oncogenic stress - that the drug can release (Figure 3C) . Hence, ABT-737 should be effective on its own in tumors with low Mcl-1 and A1, even in the face of high Bcl-2, and may prove more widely efficacious if combined with agents that prevent Mcl-1 synthesis, promote its degradation or inactivate it.
BH3 mimetics may also prove to be important adjuvants in conventional therapy. ABT-737 and the Bcl-xL-specific A-385358  markedly sensitised cells to diverse chemotherapy agents, in part due to Mcl-1 degradation [38**,39,40*,41,42,46,47*]. For example, Bim and Bad mediate the killing of CML cells by imatinib, and co-treatment with ABT-737 might augment responses in some patients with refractory disease [47*]. Thus, a BH3 mimetic might allow lower doses of a conventional agent or more durable responses, unless substantial adverse effects arise. Phase I clinical trials have begun for an orally available ABT-737 derivative, and Phase II trials for the less well-characterized GeminX broad-spectrum BH3 mimetic GX15-070 .
Autophagy is an ancient mechanism by which starved cells produce energy and stave off death by gradually targeting their organelles and cytoplasmic elements to lysosomes for digestion. Excessive self-cannibalisation may represent a second form of programmed cell death [49*,50]. Significantly, Bcl-2 and Bcl-xL associate with the evolutionarily conserved autophagy inducer Beclin-1, a haplo-insufficient tumor suppressor , and inhibit autophagy [52**]. The inhibition may require Bcl-2 localized on the endoplasmic reticulum (ER) [52**,53]. Notably, a BH3 domain within Beclin-1 mediates their association .
Since the Beclin-1:Bcl-xL association is of low affinity (1–2 μM versus low nM for BH3-only proteins) , presumably the BH3-only proteins induced by stress (or some BH3 mimetic agents) can displace Beclin-1, rendering the cell susceptible to autophagy as well as apoptosis (Figure 4). Curiously, if apoptosis is blocked, e.g. by the absence of Bax and Bak, the autophagic response to nutrient/cytokine deprivation is prolonged, albeit not indefinite, survival [49*,52**], whereas that to cytotoxic compounds (e.g. etoposide) is primarily autophagic cell death [55**] (Figure 4). Therefore, manipulating autophagy may well affect therapy, but whether its promotion or its inhibition  will be more effective remains to be established.
Independent of its apoptotic role, the Bcl-2 family seems to influence mitochondrial shape, which reflects the balance between continual fission and fusion processes . Paradoxically, however, whereas either loss of both Bax and Bak or expression of pro-survival relatives curtails apoptosis, the former promotes fission  and the latter fusion [59,60]. In any case, most evidence suggests that the mitochondrial fission machinery is not required for mitochondrial permeabilization or apoptosis [57,61].
Surprisingly, most Bax molecules probably translocate from the cytosol to membranes after caspases are activated [62*,63,64]. Bax translocation and cytochrome c release were slower in cells lacking both Caspase-3 and -7, implicating a proteolytic amplification loop [62*]. Conceivably, these activated caspases augment the initial response by cleaving Bcl-2 pro-survival proteins or by generating tBid from full-length Bid. In any case, such findings imply that commitment to apoptosis requires only a small proportion of Bax, even in the absence of Bak .
Several Bcl-2 family members, including Bcl-2 and Bak, reside not only on mitochondria but also on the ER/nuclear envelope, where they may regulate cytosolic Ca2+ levels. Association of Bcl-2/Bcl-xL with the inositol triphosphate receptor type 1 (InsP3R) reportedly increases leakage of Ca2+ ions into the cytosol [65,66]. Bax and Bak may also modulate the ER unfolded protein response by binding to the cytosolic domain of the inositol-requiring endoribonuclease 1alpha (IRE1alpha) . Whether apoptotic signals from the ER must be routed through the mitochondria is not certain.
Bid loss primarily affects the extrinsic pathway, but conflicting recent reports suggested that Bid either countered or enhanced apoptosis due to replicative stress or genotoxic damage [68,69], Contrary to those reports, Bid appears to be dispensable for both DNA damage-induced apoptosis and cell cycle arrest .
Both biochemical and genetic evidence now strongly supports the hypothesis that the BH3-only proteins trigger apoptosis primarily by engaging their pro-survival relatives (Figure 1B), freeing Bax and Bak (Figure 2) [10**,19**,20**,21]. Nevertheless, in vitro data from liposome disruption [12*,14*] and from some mutant proteins [16*], argue that certain BH3-only proteins, such as tBid, can directly activate Bax/Bak (Figure 1A). Hence, direct activation may still have a role, perhaps in amplifying the apoptotic signal once mitochondrial permeation has commenced. Precisely how the pro-survival proteins restrain Bax and Bak remains unclear, as more evidence is needed for the presumptive ‘primed’ Bax/Bak (Figure 2), and to clarify the mechanism of the ‘association-independent’ constraint hypothesized above.
Pre-clinical studies with BH3-mimetics, particularly ABT-737, suggest that these new agents will prove valuable additions to the oncologist’s armamentarium (Figure 3). Eventually, we envision a panel of BH3 mimetics that target different Bcl-2 homologs or combinations thereof. It remains unclear whether strategies targeting multiple or individual family members will prove most efficacious, but both may well find important niches. The possibility that BH3 mimetics may also trigger autophagy (Figure 4) raises new questions and prospects.
We thank our colleagues, particularly David Huang, Andreas Strasser, Peter Colman, Mark Hinds, Philippe Bouillet, Ruth Kluck, Jamie Fletcher, Simon Willis and Mark van Delft for valuable discussions on the issues addressed here. Our research is supported by a National Health and Medical Research Council Program Grant (257502), a Specialized Center of Research Grant from the Leukemia and Lymphoma Society and project grants from the US National Cancer Institute (CA80188 and CA43540).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
* of special interest
** of outstanding interest