BH3-Only Proteins and Their Roles in Programmed Cell Death

M Giam1,2, DCS Huang1 and P Bouillet1

1The Walter and Eliza Hall Institute of Medical Research, Melbourne, Australia and 2Department of Medical Biology, The University of Melbourne, Melbourne, Australia

The Bcl-2 family of proteins controls the mitochondrial space is required for the activation of the initiator pathway to apoptosis. It consists of pro-survival and pro- caspase-9 by the apoptotic protease-activating factor-1. apoptotic members, and their interactions decide whether Proteins of the Bcl-2 family control the integrity of the apoptogenic factor confined to the mitochondrial inter- mitochondrial membrane in healthy cells, and its membrane space can leak to the cytosol. Despite the permeabilization in response to apoptotic stimuli (Cory intense efforts to understand the molecular mechanisms and Adams, 2002). that lead to the permeabilization of the mitochondrial membrane, this particular issue remains a matter of The Bcl-2 family intense controversy. It is well accepted that pro-apoptotic Bax and Bak are directly responsible for the damage to All members of the Bcl-2 family share a close homology the mitochondria, but pro-survival family members pre- in up to four characteristic regions termed the BH (Bcl-2 vent them from doing so. It is also accepted that stress homology) domains (BH1–4) (Figure 1). They have signals activate selected Bcl-2 homology (BH)3-only either a pro-survival or a pro-apoptotic function. proteins. But do these BH3-only proteins bind and activate Bax and Bak directly, or do they inhibit the The Pro-survival proteins pro-survival family members? The pro-survival proteins of the Bcl-2 family include Oncogene (2009) 27, S128–S136; doi:10.1038/onc.2009.50 Bcl-2, Bcl-w, Bcl-xL, Mcl-1 and A1. They possess up to four BH domains and all have similar 3D structures Keywords: Bcl-2 family; BH3-only proteins; Bax/Bak (Strasser, 2005). The BH1, BH2 and BH3 domains activation; direct/indirect models; mitochondria; cyto- interact to form a hydrophobic groove on the surface of chrome c release the protein. This groove accommodates the BH3 domain of its pro-apoptotic partners and is critical for the pro-survival function of these proteins. Overexpres- sion of any of the pro-survival Bcl-2 family members Introduction prevents the death induced by many apoptotic stimuli, indicating a significant functional redundancy between Apoptosis is essential for the development and main- these proteins. In physiological conditions, however, it is tenance of tissue homeostasis. It allows the safe rare that a single pro-survival protein ensures the elimination of cells that are no longer needed, for survival of a cell population. As an example, T cells example, at the end of an immune response. Apoptosis successively use Bcl-2 and Mcl-1, then A1 and Bcl-xL, occurs as the result of the activation of cysteine then Mcl-1 and Bcl-xL, and finally Bcl-2 again to survive proteases called caspases, which dismantle the cells by the different stages of their development (Strasser et al., degrading essential proteins. Two apoptotic pathways 2008). This may reflect either a selective expression of converge on the activation of caspases:the ‘extrinsic these genes in response to survival signals that may be pathway’, also called the ‘death receptor pathway’, different at each stage of maturation, or an adaptation involves the activation of membrane receptors of the toward the need to counter apoptotic signals that come tumor necrosis factor receptor family and the activation in different forms at each of these stages. Genetic studies of initiator caspases, 8 and 10. The ‘intrinsic pathway’, have helped to identify the physiological role of the pro- also called ‘mitochondrial’ or ‘Bcl-2-regulated pathway’, survival proteins in mice. The various phenotypes is activated as a result of cellular stresses such as growth illustrate the different requirements for these proteins factor deprivation, DNA damage or exposure to in different cell types and at various stages of develop- cytotoxic substances. Mitochondria play a major role ment. Bcl-2-deficient mice develop polycystic kidney in the Bcl-2-regulated pathway. Indeed, release of disease and profound lymphopenia and turn gray as a cytochrome c from the mitochondrial intermembrane result of the loss of melanocyte stem cells (Veis et al., 1993; Nakayama et al., 1994). These mice fail to thrive and die early of renal failure. Loss of Bcl-xL causes Correspondence:Dr P Bouillet, The Walter and Eliza Hall Institute of Medical Research, Molecular Genetics of Cancer Division, 1G Royal massive apoptosis in the nervous and hematopoietic Parade, Parkville, 3050 Victoria, Australia. systems and embryos die at mid-gestation (Motoyama E-mail:[email protected] et al., 1995). Embryos lacking Mcl-1 die at the time of BH3-only proteins and their roles M Giam et al S129 Pro-survival members BH4 BH3 BH1 BH2 TM Bcl-2, Bcl-xL, Bcl-w, Mcl-1,A1

Pro-apoptotic members

Multidomain effectors BH3 BH1 BH2 TM Bax, Bak, Bok

BH3-only proteins BH3 TM Bim, Puma, Noxa, Bik, Bmf, Bad, Hrk, Bid Figure 1 The Bcl-2 family is made up of proteins that contain conserved functional Bcl-2 Homology domains. The family can be subdivided into pro-survival and pro-apoptotic proteins. The latter can be further divided into multidomain effectors and BH3-only proteins. The transmembrane domain (TM) is not found in some of the BH3–only proteins. implantation (Rinkenberger et al., 2000), whereas BaxÀ/ÀBakÀ/À cells, showing the strict requirement for conditional deletion of Mcl-1 in the hematopoietic Bax or Bak downstream of the BH3-only proteins. system only showed a requirement of Mcl-1 for the However, it is surprising that 10% BaxÀ/ÀBakÀ/À mice survival of hematopoietic progenitors (Opferman et al., survived to adulthood and could live for several months 2003). The loss of A1a resulted in accelerated neutrophil (Lindsten et al., 2000). This shows that some develop- apoptosis, but this mild phenotype may be because of mental cell death can still proceed in the absence of these the existence of three additional A1 genes that may have two pro-apoptotic molecules. Bok, a third multi-domain substituted for the missing A1a (Hamasaki et al., 1998). pro-apoptotic protein resembling Bax and Bak, could It has been known for a long time that overexpressed potentially fulfill the same function. Unfortunately, this pro-survival proteins prevent the release of cytochrome protein has been poorly studied so far, and it is not at all c to the cytosol and thereby also the subsequent initiator clear whether it can compensate for the absence of Bax caspase activation. However, the molecular mechanisms or Bak in certain cell types. In vivo, Bax and Bak appear that they use to protect mitochondrial integrity are still to be mostly redundant, as the presence of a single allele debated. of either of them was sufficient to prevent all the defects observed in BaxÀ/ÀBakÀ/À mice (Lindsten et al., 2000). However, recent reports suggest that this redundancy Bax, Bak and Bok may be more limited, because some apoptotic signals Bcl-2-regulated apoptosis requires Bax or Bak for seem to rely specifically on one of them for their mitochondrial damage. These two pro-apoptotic pro- execution. For example, the BH3-only protein, Bik/ teins contain BH1-3 domains, and have a 3D structure Nbk, was reported to kill in the presence of Bax only, very similar to that of the pro-survival members of but to be inert in the presence of Bak only (Theodorakis the family (Sattler et al., 1997; Suzuki et al., 2000; et al., 2002; Gillissen et al., 2003). Moldoveanu et al., 2006). Mice lacking Bax show mild Bax and Bak are the main effectors of the Bcl-2- lymphoid hyperplasia and male sterility because of regulated pathway. When activated, they permeabilize sperm-cell differentiation defects (Knudson et al., 1995). the mitochondria to allow the release of apoptogenic Mice lacking Bak have no documented defect so far. proteins into the cytosol. This is thought to occur Most mice lacking both Bax and Bak complete through the formation of pores in the outer mitochon- embryonic development, but die around birth because drial membrane. The pore theory appeared with the they fail to nurse (Lindsten et al., 2000). These mice realization that the 3D structures of Bcl-xL (and later, of show several developmental defects including the Bax) had significant similarity with the pore-forming persistence of interdigital webs, imperforated vaginal translocation domain of the diphteria toxin (Suzuki canal and accumulation of cells in the central nervous et al., 2000). Despite intense efforts to show the and hematopoietic systems (Lindsten et al., 2000; Wei existence of this pore, its molecular composition remains et al., 2001). Many cell types derived from these mice elusive. The presence of Bax or Bak is required to allow proved to be completely resistant to a large array of cytochrome c out of the mitochondria or large macro- apoptotic stimuli such as UV-irradiation, DNA-dama- molecules out of synthetic lipidic vesicles, but how they ging drugs and cytokine deprivation, but were still achieve this not completely understood. Although Bax sensitive to death by tumor necrosis factor family and Bak seem to have identical functions, one striking ligands (which trigger the extrinsic pathway) (Wei difference between them is their subcellular localization et al., 2001). Of particular interest, death induced by in healthy cells (Reed, 2006). Bak is an integral overexpression of pro-apoptotic BH3-only members of membrane protein on the outer mitochondrial mem- the Bcl-2 family seemed to be completely blocked in brane, whereas inactive Bax is a largely cytosolic protein

Oncogene BH3-only proteins and their roles M Giam et al S130 (Wolter et al., 1997). This is because the C-terminal How does it work? membrane-anchoring domain of Bax is tucked into its hydrophobic pocket and an additional step is required How the proteins of the Bcl-2 family interact to to release it, allowing it to translocate to the mitochon- permabilize to the mitochondrial outer membrane is dria where it undergoes oligomerization (Nechushtan still a matter of intense controversy. To date, it is not et al., 1999). clear what the role of the different factions are, or even whether each faction might have more than one role. Two main models, known as the ‘indirect’ or ‘direct’ The BH3-only proteins model of activation of Bax and Bak, have been proposed The mammalian BH3-only proteins currently known to take into account the numerous findings in several include Bad, Bid, Bik/NBK, Bim/Bod, Bmf, Hrk/DP5, model systems. Noxa and Puma/BBC3. Depending on the definition of Everybody agrees on the fact that either Bax or Bak what constitutes a BH3 domain, several other proteins are necessary for mitochondrial outer membrane per- may also be included in this subfamily. The BH3-only meabilization, because their combined absence renders proteins are pro-apoptotic and the only region that they cells resistant to multiple apoptotic stimuli. It is also well share with each other and their relatives of the Bcl-2 accepted that, to trigger mitochondrial outer membrane family is the BH3 domain (Willis and Adams, 2005). permeabilization, Bax or Bak must be in the mitochon- Overexpression of BH3-only proteins promotes apopto- drial outer membrane, or at least close to it (Nechushtan tic death in many cell types, but requires the presence of et al., 2001). This is the first hurdle for Bax, which has to either Bax or Bak for the death to occur (Zong et al., translocate from the cytoplasm to this membrane. The 2001). This places them upstream of Bax/Bak in the mechanism of this translocation is not known, but it chain of events that lead to caspase activation. involves a structural rearrangement that exposes an N- Genetic inactivation of Bim provided valuable in- terminal epitope (detected by the 6A7 monoclonal formation about the role of BH3-only proteins (Bouillet antibody) in the monomeric form found in healthy cells et al., 1999). Similar to mice that overexpress Bcl-2 in (Nechushtan et al., 1999). Pore formation involves their B and T cells (Strasser et al., 1991a, b), Bim- altered conformation and oligomerization for both deficient mice had massive numbers of B cells, CD4 þ 8À Bax and Bak (Eskes et al., 2000; Wei et al., 2000). and CD4À8 þ T cells. It is interesting that, although Bim- Recent results about Bak suggest that, upon receiving an deficient thymocytes became resistant to apoptotic apoptotic stimulus, the inactive Bak molecules present stimuli that could be inhibited by Bcl-2 overexpression, on the mitochondrial membrane become activated and such as growth factor withdrawal or ionomycin, they expose their BH3 domain (Dewson et al., 2008). This remained sensitive to other Bcl-2-inhibitable stimuli, allows dimerization of these activated forms by the such as exposure to the phorbol ester phorbol insertion of the BH3 domain of one Bak molecule into 12-myristate 13-acetate (PMA). This showed that the hydrophobic groove of another. This BH3:groove individual BH3-only proteins are activated only in interaction is postulated to produce symmetric Bak response to certain apoptotic stimuli and not to others. dimers that presumably contain additional interaction The physiological roles of the different BH3-only interfaces to produce higher order oligomers, resulting proteins have been reviewed elsewhere (Strasser, 2005; in homo-oligomerization and permeabilization of the van Delft and Huang, 2006). outer mitochondrial membrane (Dewson et al., 2008). All BH3-only proteins bind strongly to at least some Bax is also suggested to undergo similar steps of pro-survival members of the family. The BH3 domain of activation and oligomerization. BH3-only proteins inserts in the hydrophobic groove formed by the BH1, BH2 and BH3 domains at the surface of the pro-survival molecules. Some BH3-only Direct activation model proteins (Bim, Puma and Bid) bind to all the pro- Indications that certain BH3-only proteins might trigger survival molecules, whereas others have a more limited the activation of Bax and Bak came from several repertoire (Chen et al., 2005). Bad, for example, binds observations that the activated form of Bid, truncated Bcl-2, Bcl-xL and Bcl-w, but not Mcl-1 or A1. Noxa, by Bid, could induce conformational changes in either Bax contrast, binds Mcl-1 and A1, but not the others. The or Bak (Desagher et al., 1999; Wei et al., 2000). Using a BH3 domain seems to be entirely responsible for this large array of synthetic BH3 peptides, two groups later interaction, and is absolutely required for the killing showed that some of these peptides (Bid, Bim and activity of the BH3-only proteins. In addition, Bim, Bid Puma) could directly induce oligomerization of Bax and and Puma were also found to interact with Bax (Marani Bak to release cytochrome c, whereas the others could et al., 2002; Cartron et al., 2004; Willis et al., 2007). Bim, not (Letai et al., 2002; Kuwana et al., 2005; Kim et al., Bad and Bmf are unstructured in the absence of a 2006b). The term ‘activator’ was chosen to describe this binding partner, and only their BH3 domain becomes particular ability. The remaining BH3-only proteins structured upon binding a pro-survival protein (Hinds were termed ‘sensitizers’, as they could synergize with et al., 2007). In contrast, Bid shows a structure strikingly the activators in mitochondrial permeabilization, but similar to that of Bax and Bcl-xL, even though it lacks could not induce it by themselves. The direct model recognizable BH1 and BH2 domains (Chou et al., 1999; (Figure 2) thus proposes that the ‘activator’ BH3-only McDonnell et al., 1999). proteins are able to activate Bax/Bak directly, and are

Oncogene BH3-only proteins and their roles M Giam et al S131 prevented from doing so in healthy cells by sequestra- proteins need to bind at least some pro-survival tion by the pro-survival proteins. When an apoptotic members of the family to have some pro-apoptotic stimulus is received, the ‘sensitizer’ BH3-only proteins activity. Mutations that disrupt the binding also impair are activated, which bind to the pro-survival proteins to the activity of these proteins. The indirect activation liberate the ‘activator’ BH3-only proteins that are now model (also known as the displacement model) differs able to interact with Bax/Bak to induce apoptosis (Letai from the direct activation model on the role of the BH3- et al., 2002). Various groups have reported evidence to only proteins. In this model, Bax/Bak are kept in check support the existence of the two proposed classes of by the pro-survival proteins and induce apoptosis upon BH3-only proteins. Kuwana et al. (2005) used an release from these repressors. The BH3-only proteins innovative Bax-dependent membrane permeabilization activate apoptosis by binding and neutralizing the pro- assay to show that recombinant Bim and truncated Bid survival proteins, allowing Bax/Bak to homo-oligomer- BH3 peptides were able to activate Bax directly whereas ize and permeabilize the mitochondria. This displace- other BH3-only proteins required the presence of pro- ment model does not require any interaction between survival proteins to induce permeabilization. To cir- the BH3-only proteins and Bax/Bak (Willis et al., 2007). cumvent the shortcomings of using short BH3 peptides, Pro-survival proteins protect cells from apoptosis by another group used full-length proteins to assess the sequestering Bax/Bak rather than the ‘activator’ BH3- ‘two class’ model (Kim et al., 2006b). They found that only proteins, as suggested by the direct model. The Bim and Puma (‘activators’) were able to activate Bax/ differences in the pro-apoptotic potential between BH3- Bak regardless of whether the pro-survival proteins were only proteins are explained by the fact that Bim, Puma present, and that these ‘activators’ were required for and Bid are capable of binding to all pro-survival Bad- or Noxa-mediated apoptosis. In addition, they proteins, whereas the others (Bad, Noxa, Hrk, Bik and reported that Bax/Bak mutants, which were unable to Bmf) only bind to a subset of these (Chen et al., 2005; bind Bcl-2, Bcl-xL and Mcl-1, were not constitutively Willis et al., 2007). According to this model, apoptosis activated, suggesting that they are not directly repressed occurs when all the pro-survival proteins are effectively by the pro-survival proteins and require additional steps neutralized, explaining why Bad (which only binds Bcl- to activate them (Kim et al., 2006b). A few weeks ago, 2, Bcl-xL and Bclw) and Noxa (which, in contrast, only the ﬁrst NMR structure of a constrained Bim BH3 binds Mcl-1 and A1) are poor inducers of apoptosis on peptide bound to Bax was published, providing one their own, but efﬁcient inducers as a combination (Willis more argument that Bax may be activated by the direct et al., 2007). binding of Bim (Gavathiotis et al., 2008). Arguments Indirect activation model Because the direct interactions observed between mem- The indirect activation of Bax and Bak model bers of the Bcl-2 family seemed to correlate with their (Figure 2) is derived from the fact that all BH3-only pro- or anti-apoptotic function, it is widely expected

Direct Activation Model

Bim Bcl-2 Bad Bcl-2

Bad Bax Bim Bax

Apoptosis Indirect Activation Model Bax Bax dimer

Bim Bax Bcl-2

Bim Bcl-2 Figure 2 Competing models for Bax/Bak activation by BH3-only proteins. In the direct activation model, activator BH3-only proteins, such as Bim, are kept in check by pro-survival proteins until displaced by sensitizer BH3-only proteins such as Bad. Free ‘activators’ can then bind to and activate the effector proteins (Bax) and trigger their oligomerization. In the indirect model, activated effector proteins are prevented from oligomerizing by the pro-survival proteins, until BH3-only proteins neutralize the pro-survival proteins.

Oncogene BH3-only proteins and their roles M Giam et al S132 free Bcl-2

Survival

Bax Bcl-2

Bax Bim BH3 BH3 Bax no free Bcl-2

Apoptosis Bad Bax Dimer Figure 3 Possible consequences of Bim binding to Bax. In reference to Dewson et al. (2008) and Green and Chipuk (2008), we speculate that the binding of stapled Bim BH3 peptide might expose Bax BH3. Such activated Bax molecules could either be captured and inactivated by free pro-survival proteins (represented here by Bcl-2), or, if the pro-survival proteins are saturated with BH3-only proteins, homo-dimerize as a ﬁrst step toward the formation of ‘pores’ in the outer mitochondrial membrane.

that activator BH3-only proteins should interact with buffer devoid of lipids. It is noteworthy that none of Bax and Bak in order to activate them. While binding of these major rearrangements were visible in the chemical Bid to Bax is relatively easy to observe by co- shifts associated with the binding of the Bim peptide to immunoprecipitation in the presence of non-ionic Bax. An accompanying News and Views article (Green detergents (Wei et al., 2000; Walensky et al., 2006), and Chipuk, 2008) drew a parallel with the recently binding of Bim to Bax could be shown only for two described activation of Bak (Dewson et al., 2008) and minor isoforms, BimS and BimAD (Marani et al., 2002), speculated that binding of Bim to Bax might ultimately whose physiological significance is also a matter of expose the Bax’s BH3 domain and allow ‘nose-to-nose’ debate. No direct binding of the major isoform, BimEL, dimerization of two activated Bax molecules. This is an could be observed under the same conditions (Willis interesting possibility indeed, to which we would like to et al., 2007), and no binding of any Bim isoform to Bak add another:the activated Bax molecule could also be was ever reported. To account for this lack of stable captured by any free pro-survival protein, until such interaction and for the need for non-ionic detergents, it time when free pro-survival proteins are rendered has been proposed that direct activation may involve unavailable by the accumulation of BH3-only proteins transient interactions within a lipidic membrane envir- (Figure 3). Does the binding of a constrained Bim-BH3 onment, in a hit-and-run mechanism (Kuwana et al., peptide to a recombinant Bax molecule in vitro prove 2002). The presence of a membrane was shown to be that all isoforms of Bim bind and activate Bax in vivo, necessary for Bid to induce the conformational change when BimS and BimAD bind Bax in a way that can be associated with Bax activation (Yethon et al., 2003), and shown by co-immunoprecipitation, whereas BimL and a membrane-tethered Bid BH3 peptide was more potent BimEL do not? Clearly, sequences outside the BH3 at activating Bax than its soluble form (Oh et al., 2006). domain influence these interactions. Remembering only The hit-and-run mechanism does not explain how Bax is that Bim binds to Bax directly will remain untrue for the induced to open its structure and translocate to the two main isoforms of Bim, until otherwise proven. mitochondrial outer membrane. The direct activation model defines Bim and Bid as The recent determination of the structure of a Bax/ the sole BH3-only direct activators of Bax and Bak BimBH3 peptide complex reinforces the direct activa- (Letai et al., 2002; Kuwana et al., 2005). However, Willis tion theory (Gavathiotis et al., 2008). It is interesting et al. showed that Bax and Bak can mediate apoptosis that the constrained BH3 peptide was found to interact without discernable association with the putative BH3- with a site involving helices a1 and a6, different from the only activators, even in cells with no Bim or Bid (Willis canonical binding groove characterized for pro-survival et al., 2007). Accordingly, BimÀ/ÀBidÀ/À mice were born proteins. This finding supported an earlier observation at a normal frequency and appeared healthy, dismissing that the first a-helix of Bax participates in an interaction these two proteins as the sole activators of Bax and Bak with the BH3 domains of Bid and Puma (Cartron et al., (Willis et al., 2007). The list of possible activator BH3- 2004). Binding of the stapled peptide was reported to only proteins may not be complete as yet, and Puma is open the conformation of Bax, as judged by the sometimes considered as one (Cartron et al., 2004). Why exposure of the 6A7 epitope, and oligomerization in a the Puma BH3 peptide did not prove to be an activator

Oncogene BH3-only proteins and their roles M Giam et al S133 in Kuwana’s or Letai’s studies remains to be explained. and not through Bak (Gillissen et al., 2003); these are It is possible that more BH3-only proteins with an only two examples of a long list of exceptions to the activator function may exist. Other proteins capable of rules. It may well be that some BH3-only proteins have triggering mitochondrial damage have been reported. evolved in such a manner that they can both activate Some, like BCL2L12 (Scorilas et al., 2001), belong to the Bax and/or Bak and inhibit their binding to the pro- Bcl-2 family, whereas others, such as p53 (Chipuk et al., survival proteins. A recent report using inducible 2004), do not contain any BH domain. It is also overexpression of Bim proposed that mitochondrial proposed that the activators are released from pro- translocation and insertion of BimS were enough to survival proteins upon displacement by the sensitizers. recruit and activate cytosolic Bax. It was suggested that However, measurements of the affinity of the different insertion of the protein changes the membrane composi- BH3-only proteins to their pro-survival counterparts tion and topology, conveying an activation signal to have indicated that Bim binds more strongly than the Bax. BimEL and BimL did not induce Bax activation in ‘sensitizers’. It is thus difficult to imagine how the this study (Weber et al., 2007). Could it be that BimS is sensitizers could efficiently release Bim from its guar- an activator whereas BimEL and BimL are sensitizers? dians. These measurements were, however, performed Alternatively, Bax and Bak may be activated com- with BH3 peptides and with truncated pro-survival pletely independently of the other members of the proteins, in conditions that may not recapitulate the family, by events such as phosphorylation (Kim et al., lipid environment of a membrane. It is thus possible that 2006a) or by any other protein modification as an early the true values corresponding to the affinities between event after an apoptotic stimulus. That both BH3-only- full-length proteins may differ slightly from the reported dependent and BH3-only-independent activation of Bax ones. may occur is actually supported by the observation that The indirect model implies that pro-survival proteins overexpression of Bax in yeast does not require the keep Bax and Bak in check to prevent them from presence of any BH3-only protein to delay growth damaging mitochondria. It is unclear, however, whether (Khoury and Greenwood, 2008), but can be accelerated healthy cells contain activated Bax and Bak in complex by the presence of BimS (Weber et al., 2007). Although with the pro-survival proteins. Bak can be found in BH3-only proteins are often considered as the first step complex with either Bcl-xL or Mcl-1, but not with Bcl-2, in the apoptosis signaling cascade, mainly because their in healthy cells (Cuconati et al., 2003; Willis et al., 2005). overexpression or BH3 peptides are sufficient to trigger These results were corroborated by measuring the cell death in vitro, they are activated in vivo by a affinity of a Bak BH3 peptide for recombinant Bcl-xL, signaling cascade that is unlikely to have, as sole target, Mcl-1, Bcl-2 and Bcl-w on a Biacore instrument (Willis a particular BH3 protein. Let’s take an example: et al., 2005). The study of Bax has been hampered by the Sodium arsenite kills PC12 cells by activating p38 fact that non-ionic detergents trigger the conformational MAP kinase, which in turn causes the translocation of change that is considered to indicate its activation. This FoxO3A into the nucleus, which induces the expression property has been conveniently used to either show the of BimEL (Cai and Xia, 2008). How many targets other binding of Bax to Bim or to Bcl-2, or discount it as non- than FoxO3A may have p38 activated, how many genes physiological. In healthy cells, Bax appears to be a other than Bim may have FoxO3A turned on, or off ? A monomeric protein. A recent paper also proposes that quick look at the MAP kinase pathways gives an idea of almost all Bak would also be in the non-activated the number of events that could happen between the conformation in healthy cells (Dewson et al., 2008). This time PC12 cells are exposed to arsenite and the moment is actually not in conflict with the indirect model, when Bim is induced. It is well possible that one of these because the activation of Bax and Bak could occur soon unrecognized events could lead to the activation of Bax after the apoptotic stimulus, and the activated proteins or Bak, which could be then prevented from damaging could then be captured by pro-survival proteins when mitochondria for as long as BH3-only proteins have not they expose their BH3 domain. If the amount of overwhelmed the pro-survival proteins. activated Bax/Bak molecules is greater than the amount As neither the indirect nor the direct model of Bax/ of pro-survival proteins, or if the pro-survival molecules Bak activation can explain all the results that have been are already engaged with BH3-only proteins and are described, it seems reasonable to assume that both of thus not capable of inhibiting activated Bax and Bak, them are wrong, at least partially, or that perhaps both then these could proceed to form dimers and oligomers of them are partially right. Following that sort of logic, and apoptosis would ensue (Dewson et al., 2008). The David Andrews recently proposed the ‘embedding main caveat of the indirect model is that it does not together’ model, which tries to incorporate elements provide any explanation as to how Bax and Bak become from both the direct and indirect activation model to activated. Ten years ago, we believed that Bax and Bak form a new hypothesis in which the pro-survival were redundant, that all pro-survival proteins inhibited proteins promote survival both by inhibiting the BH3- Bax and Bak equally, and that all BH3-only proteins only proteins (direct model) as well as Bax/Bak (indirect inhibited all the pro-survival proteins in a similar model) (Leber et al., 2007). In this model, membrane manner. Many results of the past few years have permeabilization does not occur until Bax/Bak insert highlighted the differences between the members of multiple sequences into the mitochondrial outer each faction of the Bcl-2 family. Bcl-2 does not inhibit membrane lipid bilayer (Leber et al., 2007; Billen Bak any more (Willis et al., 2005), Bik kills through Bax et al., 2008). Bax undergoes a conformational change

Oncogene BH3-only proteins and their roles M Giam et al S134 upon membrane interaction and then a second change proteins. In the indirect model, all the pro-survival triggered by truncated Bid or other activators. This proteins must be inhibited to allow the activated Bax activated integral form of Bax then recruits more and Bak to oligomerize and damage mitochondria. Both cytosolic Bax in the process of homo-oligomerization. models agree that the excess of pro-survival proteins at Pro-survival proteins, such as Bcl-xL, are proposed to the root of many malignancies must be the primary perform two functions:sequestering the activators and target of our efforts to develop therapeutic drugs. The also neutralizing the membrane-bound Bax to prevent development of small molecules mimicking the activity them from homo-oligomerizing (functioning as a of BH3-only proteins has already shown an enormous dominant-negative Bax molecule) (Billen et al.,2008).More therapeutic potential (Oltersdorf et al., 2005; Tse et al., work on these hypotheses will for sure test their merit. 2008). Understanding the intricacies of the Bcl-2 regulation system will certainly help the design of better strategies for tailored cancer treatment. Conclusions

The ongoing battle between the ‘indirect’ and the ‘direct’ Conflict of interest models is the result of the imperfections of the model systems that we use to try to understand these M Giam has declared no conflict of interest. DCS Huang mechanisms. As most cell types express more than one is currently receiving grant support from Genentech member of each faction of the Bcl-2 family, their study (San Francisco, CA, USA), the Leukemia & Lymphoma in the context of the whole animal has been rather Society (White Plains, NY, USA) (7015-12) and the difficult because of functional redundancy. For this Cancer Council Victoria (Carlton, Australia) (461239), reason, simpler experimental settings have been adopted and is also in receipt of an program grant (461221) and a to try and sort out the role of individual components of Research Fellowship grant (576701) from the National the Bcl-2 regulation system. These include the use of Health and Medical Research Council (NHMRC, isolated mitochondria, lipid vesicles, recombinant pro- Canberra, Australia). DCS Huang holds patents and teins and, of course, massive overexpression and RNA co-ordinates the drug discovery and development interference in all sorts of cell lines. All these approaches collaboration between Genentech, Abbott (Abbott have generated a wealth of invaluable data, but certainly Park, IL, USA) and the Walter and Eliza Hall Institute also some artifacts and misinterpretations. Using BH3 (WEHI, Melbourne, Australia). P Bouillet is currently peptides instead of the corresponding full-length pro- receiving a program grant (461221) and a Career teins made certain studies possible, but relating the Development Award from the NHMRC (516706). results to the function of an entire protein within a physiological context will require the use of more physiological experimental settings. Acknowledgements The differences between the direct and indirect models may not be that important after all, because both This work was supported by the NHMRC (Program Grant, Fellowship, Career Development Award and Project Grant) theories are actually interpretations of the same data. In and the Charles and Sylvia Viertel Charitable Foundation. We the direct model, all BH3-only proteins can be are grateful to our colleagues at WEHI for stimulating sensitizers and a few of them can be activators as well. discussions and insightful comments. We apologize to the For apoptosis to occur, there is a need for enough authors whose contributions have not been cited owing to sensitizers to overpower the resistance of pro-survival space limitations.

References

Billen LP, Kokoski CL, Lovell JF, Leber B, Andrews DW. (2008). Chipuk JE, Kuwana T, Bouchier-Hayes L, Droin NM, Newmeyer DD, Bcl-XL inhibits membrane permeabilization by competing with Schuler M et al. (2004). Direct activation of Bax by p53 mediates Bax. PLoS Biol 6:e147. mitochondrial membrane permeabilization and apoptosis. Science Bouillet P, Metcalf D, Huang DCS, Tarlinton DM, Kay TWH, 303:1010–1014. Ko¨ ntgen F et al. (1999). Proapoptotic Bcl-2 relative Bim required Chou JJ, Li H, Salvesen GS, Yuan J, Wagner G. (1999). Solution for certain apoptotic responses, leukocyte homeostasis, and to structure of BID, an intracellular ampliﬁer of apoptotic signaling. preclude autoimmunity. Science 286:1735–1738. Cell 96:615–624. Cai B, Xia Z. (2008). p38 MAP kinase mediates arsenite-induced Cory S, Adams JM. (2002). The Bcl2 family:regulators of the cellular apoptosis through FOXO3a activation and induction of Bim life-or-death switch. Nat Rev Cancer 2:647–656. transcription. Apoptosis 13:803–810. Cuconati A, Mukherjee C, Perez D, White E. (2003). DNA damage Cartron PF, Gallenne T, Bougras G, Gautier F, Manero F, Vusio P response and MCL-1 destruction initiate apoptosis in adenovirus- et al. (2004). The ﬁrst alpha helix of Bax plays a necessary role in its infected cells. Genes Dev 17:2922–2932. ligand-induced activation by the BH3-only proteins Bid and Desagher S, Osen-Sand A, Nichols A, Eskes R, Montessuit S, PUMA. Mol Cell 16:807–818. Lauper S et al. (1999). Bid-induced conformational change of Bax Chen L, Willis SN, Wei A, Smith BJ, Fletcher JI, Hinds MG et al. is responsible for mitochondrial cytochrome c release during (2005). Differential targeting of pro-survival Bcl-2 proteins by their apoptosis. J Cell Biol 144:891–901. BH3-only ligands allows complementary apoptotic function. Mol Dewson G, Kratina T, Sim HW, Puthalakath H, Adams JM, Cell 17:393–403. Colman PM et al. (2008). To trigger apoptosis Bak exposes its

Oncogene BH3-only proteins and their roles M Giam et al S135 BH3 domain and homo-dimerizes via BH3:grooove interactions. hematopoietic cells and neurons in Bcl-x deficient mice. Science Mol Cell 30:369–380. 267:1506–1510. Eskes R, Desagher S, Antonsson B, Martinou JC. (2000). Bid induces Nakayama K, Nakayama K-I, Negishi I, Kuida K, Sawa H, Loh DY. the oligomerization and insertion of Bax into the outer mitochon- (1994). Targeted disruption of bcl-2ab in mice:occurrence of gray drial membrane. Mol Cell Biol 20:929–935. hair, polycystic kidney disease, and lymphocytopenia. Proc Natl Gavathiotis E, Suzuki M, Davis ML, Pitter K, Bird GH, Katz SG Acad Sci USA 91:3700–3704. et al. (2008). BAX activation is initiated at a novel interaction site. Nechushtan A, Smith CL, Hsu YT, Youle RJ. (1999). Conformation Nature 455:1076–1081. of the Bax C-terminus regulates subcellular location and cell death. Gillissen B, Essmann F, Graupner V, Starck L, Radetzki S, Dorken B EMBO J 18:2330–2341. et al. (2003). Induction of cell death by the BH3-only Bcl-2 homolog Nechushtan A, Smith CL, Lamensdorf I, Yoon SH, Youle RJ. (2001). Nbk/Bik is mediated by an entirely Bax-dependent mitochondrial Bax and Bak coalesce into novel mitochondria-associated clusters pathway. EMBO J 22:3580–3590. during apoptosis. J Cell Biol 153:1265–1276. Green DR, Chipuk JE. (2008). Apoptosis:Stabbed in the BAX. Nature Oh KJ, Barbuto S, Pitter K, Morash J, Walensky LD, Korsmeyer SJ. 455:1047–1049. (2006). A membrane-targeted BID BCL-2 homology 3 peptide is Hamasaki A, Sendo F, Nakayama K, Ishida N, Negishi I, Nakayama sufficient for high potency activation of BAX in vitro. J Biol Chem K-I et al. (1998). Accelerated neutrophil apoptosis in mice 281:36999–37008. lacking A1-a, a subtype of the bcl-2-related A1 gene. J Exp Med Oltersdorf T, Elmore SW, Shoemaker AR, Armstrong RC, Augeri DJ, 188:1985–1992. Belli BA et al. (2005). An inhibitor of Bcl-2 family proteins induces Hinds MG, Smits C, Fredericks-Short R, Risk JM, Bailey M, regression of solid tumours. Nature 435:677–681. Huang DC et al. (2007). Bim, Bad and Bmf:intrinsically Opferman JT, Letai A, Beard C, Sorcinelli MD, Ong CC, Korsmeyer unstructured BH3-only proteins that undergo a localized conforma- SJ. (2003). Development and maintenance of B and T lymphocytes tional change upon binding to prosurvival Bcl-2 targets. Cell Death requires antiapoptotic MCL-1. Nature 426:671–676. Differ 14:128–136. Reed JC. (2006). Proapoptotic multidomain Bcl-2/Bax-family proteins: Khoury CM, Greenwood MT. (2008). The pleiotropic effects of mechanisms, physiological roles, and therapeutic opportunities. Cell heterologous Bax expression in yeast. Biochim Biophys Acta 1783: Death Differ 13:1378–1386. 1449–1465. Rinkenberger JL, Horning S, Klocke B, Roth K, Korsmeyer SJ. Kim BJ, Ryu SW, Song BJ. (2006a). JNK- and p38 kinase-mediated (2000). Mcl-1 deficiency results in peri-implantation embryonic phosphorylation of Bax leads to its activation, mitochondrial lethality. Genes Dev 14:23–27. translocation and to apoptosis of human hepatoma HepG2 cells. Sattler M, Liang H, Nettesheim D, Meadows RP, Harlan JE, J Biol Chem 281:21256–21265. Eberstadt M et al. (1997). Structure of Bcl-xL-Bak peptide Kim H, Rafiuddin-Shah M, Tu HC, Jeffers JR, Zambetti GP, complex:recognition between regulators of apoptosis. Science 275: Hsieh JJ et al. (2006b). Hierarchical regulation of mitochondrion- 983–986. dependent apoptosis by BCL-2 subfamilies. Nat Cell Biol 8: Scorilas A, Kyriakopoulou L, Yousef GM, Ashworth LK, Kwamie A, 1348–1358. Diamandis EP. (2001). Molecular cloning, physical mapping, and Knudson CM, Tung KSK, Tourtellotte WG, Brown GAJ, Korsmeyer expression analysis of a novel gene, BCL2L12, encoding a proline- SJ. (1995). Bax-deficient mice with lymphoid hyperplasia and male rich protein with a highly conserved BH2 domain of the Bcl-2 germ cell death. Science 270:96–99. family. Genomics 72:217–221. Kuwana T, Bouchier-Hayes L, Chipuk JE, Bonzon C, Sullivan BA, Strasser A. (2005). The role of BH3-only proteins in the immune Green DR et al. (2005). BH3 Domains of BH3-only system. Nat Rev Immunol 5:189–200. proteins differentially regulate Bax-mediated mitochondrial mem- Strasser A, Harris AW, Cory S. (1991a). Bcl-2 transgene inhibits T cell brane permeabilization both directly and indirectly. Mol Cell 17: death and perturbs thymic self-censorship. Cell 67:889–899. 525–535. Strasser A, Puthalakath H, O0Reilly LA, Bouillet P. (2008). What do Kuwana T, Mackey MR, Perkins G, Ellisman MH, Latterich M, we know about the mechanisms of elimination of autoreactive T and Schneiter R et al. (2002). Bid, Bax, and lipids cooperate to form B cells and what challenges remain. Immunol Cell Biol 86:57–66. supramolecular openings in the outer mitochondrial membrane. Strasser A, Whittingham S, Vaux DL, Bath ML, Adams JM, Cory S Cell 111:331–342. et al. (1991b). Enforced BCL2 expression in B-lymphoid cells Leber B, Lin J, Andrews DW. (2007). Embedded together:the life and prolongs antibody responses and elicits autoimmune disease. Proc death consequences of interaction of the Bcl-2 family with Natl Acad Sci USA 88:8661–8665. membranes. Apoptosis 12:897–911. Suzuki M, Youle RJ, Tjandra N. (2000). Structure of Bax: Letai A, Bassik M, Walensky L, Sorcinelli M, Weiler S, Korsmeyer S. coregulation of dimer formation and intracellular localization. Cell (2002). Distinct BH3 domains either sensitize or activate mitochon- 103:645–654. drial apoptosis, serving as prototype cancer therapeutics. Cancer Theodorakis P, Lomonosova E, Chinnadurai G. (2002). Critical Cell 2:183–192. requirement of BAX for manifestation of apoptosis induced by Lindsten T, Ross AJ, King A, Zong W, Rathmell JC, Shiels HA et al. multiple stimuli in human epithelial cancer cells. Cancer Res 62: (2000). The combined functions of proapoptotic Bcl-2 family 3373–3376. members Bak and Bax are essential for normal development of Tse C, Shoemaker AR, Adickes J, Anderson MG, Chen J, Jin S et al. multiple tissues. Mol Cell 6:1389–1399. (2008). ABT-263:a potent and orally bioavailable Bcl-2 family Marani M, Tenev T, Hancock D, Downward J, Lemoine NR. (2002). inhibitor. Cancer Res 68:3421–3428. Identification of novel isoforms of the BH3 domain protein Bim, van Delft MF, Huang DCS. (2006). How the Bcl-2 family of proteins which directly activate Bax to trigger apoptosis. Mol Cell Biol 22: interact to regulate apoptosis. Cell Res 16:203–213. 3577–3589. Veis DJ, Sorenson CM, Shutter JR, Korsmeyer SJ. (1993). Bcl-2- McDonnell JM, Fushman D, Milliman CL, Korsmeyer SJ, deficient mice demonstrate fulminant lymphoid apoptosis, poly- Cowburn D. (1999). Solution structure of the proapoptotic molecule cystic kidneys, and hypopigmented hair. Cell 75:229–240. BID:a structural basis for apoptotic agonists and antagonists. Cell Walensky LD, Pitter K, Morash J, Oh KJ, Barbuto S, Fisher J et al. 96:625–634. (2006). A Stapled BID BH3 Helix Directly Binds and Activates Moldoveanu T, Liu Q, Tocilj A, Watson MH, Shore G, Gehring K. BAX. Mol Cell 24:199–210. (2006). The x-ray structure of a BAK homodimer reveals an Weber A, Paschen SA, Heger K, Wilfling F, Frankenberg T, inhibitory zinc binding site. Mol Cell 24:677–688. Bauerschmitt H et al. (2007). BimS-induced apoptosis requires Motoyama N, Wang FP, Roth KA, Sawa H, Nakayama K, mitochondrial localization but not interaction with anti-apoptotic Nakayama K et al. (1995). Massive cell death of immature Bcl-2 proteins. J Cell Biol 177:625–636.

Oncogene BH3-only proteins and their roles M Giam et al S136 Wei MC, Lindsten T, Mootha VK, Weiler S, Gross A, Ashiya M et al. engage multiple Bcl-2 homologs, not Bax or Bak. Science 315: (2000). tBID, a membrane-targeted death ligand, oligomerizes BAK 856–859. to release cytochrome c. Genes Dev 14:2060–2071. Wolter KG, Hsu YT, Smith CL, Nechushtan A, Xi XG, Youle RJ. Wei MC, Zong WX, Cheng EH, Lindsten T, Panoutsakopoulou V, Ross (1997). Movement of Bax from the cytosol to mitochondria during AJ et al. (2001). Proapoptotic BAX and BAK:a requisite gateway to apoptosis. J Cell Biol 139:1281–1292. mitochondrial dysfunction and death. Science 292:727–730. Yethon JA, Epand RF, Leber B, Epand RM, Andrews DW. (2003). Willis SN, Adams JM. (2005). Life in the balance:how BH3-only Interaction with a membrane surface triggers a reversible con- proteins induce apoptosis. Curr Opin Cell Biol 17:617–625. formational change in Bax normally associated with induction of Willis SN, Chen L, Dewson G, Wei A, Naik E, Fletcher JI et al. (2005). apoptosis. J Biol Chem 278:48935–48941. Pro-apoptotic Bak is sequestered by Mc1-1 and Bcl-xL, but not Bcl- Zong WX, Lindsten T, Ross AJ, MacGregor GR, Thompson CB. (2001). 2, until displaced by BH3-only proteins. Genes Dev 19:1294–1305. BH3-only proteins that bind pro-survival Bcl-2 family members fail to Willis SN, Fletcher JI, Kaufmann T, van Delft MF, Chen L, induce apoptosis in the absence of Bax and Bak. Genes Dev 15:1481– Czabotar PE et al. (2007). Apoptosis initiated when BH3 ligands 1486.

Oncogene