Review

Introduction

Apoptotic stresses originating from diverse cellular loci are typically channelled into a core death pathway involving proteolytic activation of the caspase family of cysteine proteases. BCL-2 family proteins, on the other hand, localize to the membranes of various organelles and regulate the activation of caspases. For example, several members of the BH3-only subset of the BCL-2 family respond to apoptotic signals by increasing the permeability of the outer mitochondrial membrane (OMM). This releases apoptotic intermembrane space (IMS) proteins, including cytochrome c (cyt.c), which binds APAF-1 in the cytosol and triggers the activation of caspase-9 (Green and Reed, 1998). Membrane permeabilization is achieved by BH3-dependent oligomerization of a second group of multidomain proapoptotic BCL-2 proteins represented by BAX and BAK, and is inhibited by antiapoptotic BCL-2 members, including BCL-xL and BCL-2 itself. Emerging evidence suggests that the endoplasmic reticulum (ER) also regulates apoptosis both by sensitizing mitochondria to a variety of extrinsic and intrinsic death stimuli and by initiating cell death signals of its own. Members of all three classes of the BCL-2 family localize to the ER membrane and have been shown to influence ER homeostasis, perhaps by influencing membrane permeability. Calcium release from the ER, for example, has been implicated as a key signalling event in many apoptotic models and, depending on the mode of Ca²⁺ release, may either directly activate death effectors or influence the sensitivity of mitochondria to apoptotic transitions. Furthermore, a growing number of ER proteins have been shown to influence apoptosis by either interacting with BCL-2 family members or altering ER Ca²⁺ responses, whereas several ER proteins are caspase substrates that may regulate the execution phase of apoptosis. Moreover, recent studies on how stress in the ER is coupled to apoptosis have demonstrated that the ER, like mitochondria, can directly initiate pathways to caspase activation and apoptosis (Figure 1).

Figure 1.

ER stress signalling pathways. See text for details

Full figure and legend (159K)

ER stress-induced pathways

The ER is the first stop on the secretory pathway wherein chaperone-assisted polypeptide folding and modification ensures that proteins obtain their mature conformation. When the capacity of the ER to fold properly proteins is compromised or overwhelmed, a highly conserved unfolded protein response (UPR) signal transduction pathway is activated. The UPR halts general protein synthesis while upregulating ER resident chaperones and other regulatory components of the secretory pathway (Travers et al., 2000), giving the cell a chance to correct the environment within the ER (Patil and Walter, 2001). However, if the damage is too strong and homeostasis cannot be restored, the mammalian UPR ultimately initiates apoptosis. This switch, from metabolic arrest, which provides an opportunity for repair of the ER folding capacity, to cell death, which eliminates an overly damaged cell, is analogous to the p53 response to genotoxic stress. In contrast to the p53 switching mechanisms, however, the analogous processes relating to the ER remain poorly understood.

ER stress: the survival response

In mammals, three ER transmembrane proteins, Ire1, ATF6, and PERK, respond to the accumulation of unfolded proteins in the lumen (see Sidrauski et al., 1998; Kaufman, 1999; Patil and Walter, 2001 for excellent reviews). Ire1 and PERK are normally kept in an inactive state through an association between their N-terminal lumenal domains and the chaperone BiP. Under conditions of ER stress, BiP dissociates (to bind unfolded proteins) and Ire1 and PERK undergo homo-oligomerization, stimulating trans-autophosphorylation within their serine/threonine kinase domains. Ire1 also contains a C-terminal endonuclease domain that excises a short sequence from the mRNA of the X-Box binding protein (XBP-1), generating an active bZIP transcription factor that stimulates transcription of ER chaperone genes (Ma and Hendershot, 2001; Shen et al., 2001; Yoshida et al., 2001; Calfon et al., 2002). PERK, on the other hand, phosphorylates the translation initiation factor eIF2, which halts translation and prevents the continual accumulation of newly synthesized proteins into the ER when protein folding conditions are compromised (Harding et al., 1999). Upregulation of ER chaperone genes is also mediated by a second, perhaps redundant, pathway involving ATF6. In this case, ATF6 undergoes proteolytic cleavage at the ER, which releases its active bZIP transcription factor domain to the nucleus (Yoshida et al., 1998; Ye et al., 2000).

ER stress: the death response

Experimentally, ER stress is induced by pharmacological agents that inhibit N-linked glycosylation (tunicamycin, TN), block ER to Golgi transport (brefeldin A, BFA), impair disulfide bond formation (dithiothreitol, DTT), or disrupt ER Ca²⁺ stores (thapsigargin, TG, an inhibitor of the sarcoplasmic/endoplasmic reticulum Ca²⁺ ATPase (SERCA) pumps, or A-23187, a Ca²⁺ ionophore). All of these agents eventually induce apoptosis within 20–48 h depending on the cell type, suggesting that if the damage to the ER is too great, or if balance is not restored within a certain window of time, an apoptotic response is elicited (Patil and Walter, 2001). The mechanism by which ER stress is coupled to activation of caspases was for the most part a mystery until caspase-12 was characterized by Nakagawa and Yuan (2000). Caspase-12 is ubiquitously expressed and, like all caspases, is synthesized as an inactive proenzyme consisting of a regulatory prodomain and two catalytic p20 and p10 subunits (Van de Craen et al., 1997; Nakagawa and Yuan, 2000). However, unlike other caspases, caspase-12 is remarkably specific to insults that elicit ER stress and is not proteolytically activated by other death stimuli (Nakagawa et al., 2000). Accordingly, caspase-12-null mice and cells are partially resistant to apoptosis induced by ER stress but not by other apoptotic stimuli (Nakagawa et al., 2000).

Caspase-12 is localized at the cytosolic face of the ER, placing it in a position to respond to ER stress as a proximal signalling molecule (Nakagawa and Yuan, 2000). However, the mechanism of caspase-12 activation is unclear. In mouse glial cells undergoing ER stress caused by oxygen and glucose deprivation, caspase-12 was cleaved by calpain. In vitro, m-calpain cleaved caspase-12 at T132 and K158, which releases the prodomain from the catalytic subunits and increases enzymatic activity (Nakagawa and Yuan, 2000). It has also been suggested that caspase-12 activation is linked to Ire1 signalling. The cytosolic tail of Ire1 can recruit TRAF2 (Urano et al., 2000) and when overexpressed, TRAF2 can interact with caspase-12 and weakly induce its oligomerization and cleavage (Yoneda et al., 2001). Moreover, Ire1 induces apoptosis when overexpressed (Wang et al., 1998), which could be due to caspase-12 activation. The CARD in caspase-12's prodomain might mediate homotypic interaction with other CARD containing proteins, allowing its recruitment into such an activation complex (analogous to caspase-9 recruitment to Apaf-1), or mediate caspase-12 oligomerization and autoactivation at the ER. Overexpression of full-length caspase-12 induces its oligomerization and self-cleavage between the p20 and p10 subunits at D318 (Nakagawa et al., 2000; Rao et al., 2001; Fujita et al., 2002), but only caspase-12 lacking its prodomain induces apoptosis. Therefore, caspase-12 activation likely involves both calpain-dependent removal of the prodomain and self-cleavage at D318.

Ire1 signalling might play a role in coupling the UPR survival response with an apoptotic cascade, and the fate of the cell would be determined by an interplay between these opposing signals. This scheme is reminiscent of TNF death receptor signalling in which NFB survival pathways and a caspase-8 death pathway are simultaneously activated (Baud and Karin, 2001). At present, however, an Ire1/TRAF2/caspase-12 complex remains speculative: interactions between the endogenous proteins have not been reported and Ire1 activation and caspase-12 cleavage occur with different kinetics following ER stress (Harding et al., 2000). The dependence of caspase-12 activation on Ire1 could be tested in Ire1-deficient cells, and the dependence of Ire1-mediated apoptosis on caspase-12 could be tested in caspase-12-deficient cells. Interestingly, Zong et al., 2003 found that caspase-12 processing is abolished in BAX, BAK-1-cells, and caspase-12 cleavage could be induced by the expression of ER-targeted BAK. Of note, procaspase-12 has also been reported to be cleaved by downstream executioner caspases, such as caspase-7, but this event likely represents a downstream amplification loop (Rao et al., 2001). It raises the possibility, however, that caspase-12 might also amplify the activity of other initiator caspases.

Following its activation at the ER, caspase-12 may directly process downstream caspases in the cytosol or target other as yet unidentified substrates that influence the progression of apoptosis. Two groups recently reported that caspase-12 can directly cleave procaspase-9 in vitro, leading to caspase-9-dependent activation of caspase-3 (Morishima et al., 2002; Rao et al., 2002). Inhibition of caspase-12 by expression of MAGE-3, a protein that binds the p10 subunit and blocks catalytic activity, prevents TG- and TN-induced processing of caspase-9 and -3 (Morishima et al., 2002). In addition, ER stress-induced processing of procaspase-9 can occur in the absence of cyt.c release and in APAF-1-null fibroblasts (Rao et al., 2002). These results argue that caspase-12 can directly trigger caspase-9 activation and apoptosis independent of the mitochondrial cyt.c/Apaf-1 pathway, at least in certain cell types. Many studies, however, implicate the involvement of mitochondria in ER stress-induced apoptosis (below), which might represent a redundant pathway to caspase activation or, alternatively, might provide a pathway for accumulating Smac/Diablo and HtrA in the cytosol, where their inhibition of IAPs supports optimal activation of caspases.

The relevance of caspase-12 signalling has been somewhat clouded by the fact that a human ortholog remains elusive Fischer et al. reported that the human caspase-12 gene has acquired deleterious mutations that prevent the expression of a functional protein. In spite of this discovery, two reports showed that antibodies against murine caspase-12 detect an appropriately sized protein in human cells that is processed following ER stress (Nakagawa et al., 2000; Rao et al., 2001). Proof that a functional human caspase-12 protein does indeed exist, therefore, must await purification of the endogenous enzyme.

ER stress-induced death: contribution of mitochondria

Several lines of evidence suggest that mitochondria are an important component of the ER stress-induced apoptotic pathway. First, ER stress agents cause mitochondrial release of cyt.c and loss of mitochondrial transmembrane potential (Hacki et al., 2000; Boya et al., 2002); second, BCL-2/BCL-X_L inhibit ER stress-induced apoptosis (McCormick et al., 1997; Hacki et al., 2000; McCullough et al., 2001); and third, Bax-/-, Bak-/- MEFs are resistant to TG-, TN-, and BFA-induced apoptosis (Wei et al., 2001). While the latter two findings could be due to potential roles of BCL-2 family members at the ER (see below), Kroemer and coworkers confirmed the involvement of mitochondria by showing that the cytomegalovirus-encoded mitochondrial inhibitor of apoptosis (vMIA) potently inhibited ER stress apoptosis (Boya et al., 2002). Signalling between the ER and mitochondria presumably involves the activation of one or more BH3-only proteins. BAD is a candidate, since TG or A23187 have been shown to induce its Ca²⁺/calcineurin-dependent dephosphorylation and activation (Wang et al., 1999). Increased [Ca²⁺]_c has also been observed following other conditions of ER stress (Carlberg et al., 1996) and, therefore, BAD activation may be a conserved component of the pathway. Additionally, the UPR might transcriptionally upregulate other BH3-only proteins, analogous to the response of BH3-only genes to p53-mediated stress responses. A recent report documenting the upregulation of PHMA following ER stress suggests this is the case (Reimertz et al., 2003).

ER stress-induced cyt.c release is apparently dependent on the c-Abl tyrosine kinase, because c-Abl-/- mouse embryonic fibroblasts are resistant to A23187-, BFA-, and TN-induced cyt.c release and apoptosis (Ito et al., 2001). c-Abl redistributes from the ER to the mitochondria within several hours of TG application, which parallels an increase in its kinase activity. The mechanism by which c-Abl exerts its action at this site is unclear, but it may function in concert with JNK kinases, which are recruited and activated by Ire1 during ER stress (Urano et al., 2000), and are essential for mediating cyt.c release in other cell death pathways (Tournier et al., 2000). UPR upregulation of CHOP/GADD153, a nuclear transcription factor that represses the Bcl-2 promoter (McCullough et al., 2001), may sensitize mitochondria to the proapoptotic effects of BH3-only proteins by decreasing the cellular levels of BCL-2 protein. Consistent with this, tunicamycin-induced apoptosis is impaired in CHOP-/- MEFs, and CHOP-/- mice injected with tunicamycin show decreased apoptosis in the renal tubular epithelium (Zinszner et al., 1998).

Therefore, stress in the ER unleashes a mitochondrial-dependent apoptotic pathway and a caspase-12, mitochondrial-independent pathway (Figure 1). These two arms of the ER stress response apparently operate independent of one another, since zVAD-fmk, which effectively inhibits caspase-12 in vitro (Nakagawa and Yuan, 2000), has no effect on ER stress-induced cyt.c release (Hacki et al., 2000), and caspase-12-/- MEFs only partially resist apoptosis (Nakagawa et al., 2000). This duality in signalling may ensure complete and efficient caspase activation in physiological settings.

Regulation of apoptosis by ER Ca²⁺ signals

The release of Ca²⁺ from the ER is primarily achieved by two well-characterized types of channels, the inositol 1,4,5-triphosphate receptor (IP₃R) and the Ryanodine receptor (RyR) families (Pozzan et al., 1994; Berridge et al., 2000). Also, other Ca²⁺ release channels likely exist (Cavalli et al., 2002). Release of calcium from the ER has been observed in many forms of apoptosis, and regulated oscillations of Ca²⁺ levels by IP₃R-mediated spikes may directly regulate the cell death machinery. Jurkat cells deficient in type 1 IP₃R (IP₃R1) do not increase [Ca²⁺]_c in response to dexamethasone, ionizing radiation, and T-cell receptor or Fas stimulation, and resist subsequent apoptosis (Jayaraman and Marks, 1997). Moreover, targeted disruption of all three IP₃R isoforms in the chick DT40 B-cell line blocked Ca²⁺ mobilization and apoptosis by B-cell receptor crosslinking, and the degree of resistance increased with the number of IP₃R genes deleted, suggesting that all three isoforms can contribute to apoptotic signalling (Sugawara et al., 1997). How IP₃R activation is coupled to such diverse death signals, however, remains to be determined.

The way in which Ca²⁺ signals engage cell death pathways may largely be decided by the spatio-temporal pattern and intensity of ER Ca²⁺ release. IP₃R/RyR-dependent global increases in [Ca²⁺]_c can influence apoptosis by several mechanisms. For example, Ca²⁺-dependent activation of calpains has been implicated in activating caspases in several forms of apoptosis (Wang, 2000), including ER stress- (Nakagawa and Yuan, 2000), B-cell receptor- (Ruiz-Vela et al., 1999, 2001), and radiation-induced apoptosis (Waterhouse et al., 1998). In addition, calpain cleavage of BAX has been reported to increase its proapoptotic activity (Wood et al., 1998), and in some cases calpain may cooperate with caspases in the execution phase of apoptosis (Wood and Newcomb, 1999). As discussed above, Ca²⁺ signals can lead to activation of BAD, and Ca²⁺-dependent activation of the transcription factor MEF2 leads to upregulation of Nur77/TR3 (Youn et al., 1999), which can bind to mitochondria and induce cyt.c release (Li et al., 2000). In contrast, privileged transport of Ca²⁺ between juxtaposed ER and mitochondrial membranes may sensitize mitochondria to the effects of proapoptotic BCL-2 family members (Hajnoczky et al., 2000), perhaps through Ca²⁺-induced opening of the mitochondrial permeability transition pore. For an in depth description of the mechanisms by which ER Ca²⁺ signals can influence apoptosis, the reader is referred to chapter XXX in this issue.

BCL-2 family proteins and the ER: antiapoptotic members

Although the function of BCL-2 proteins is best characterized at the mitochondrion, these proteins also localize to other intracellular membranes such as the ER/nuclear envelope. Studies employing BCL-2 selectively targeted to the ER with the transmembrane sequence of cytochrome b5 (b5-BCL-2) have demonstrated that BCL-2 has antiapoptotic activity at this location. For example, b5-BCL-2 prevents apoptosis induced by ER stress agents, ceramide, Myc, ionizing radiation, or BAX overexpression (Hacki et al., 2000; Annis et al., 2001; Rudner et al., 2001; Wang et al., 2001). Annis et al. observed that b5-BCL-2 conferred protection against apoptotic stimuli that caused an obvious mitochondrial depolarization prior to the release of cyt.c (Myc, C2 ceramide), but not against stimuli that induced cyt.c release in the absence of large mitochondrial depolarizations (etoposide). Wild-type (wt) BCL-2, in contrast, could inhibit both types of stimuli. The fact that ER-localized BCL-2 can only inhibit certain apoptotic pathways suggests that its antiapoptotic effect at the ER may not be the result of simple sequestration of endogenous proapoptotic BCL-2 proteins, in which case its membrane location would be irrelevant. This assumes, however, that the various pathways tested are indeed coupled to proapoptotic BCL-2 binding partners.

Both pro- and antiapoptotic BCL-2 proteins have predicted pore forming properties (Antonsson and Martinou, 2000) and in theory could influence ion flux across cellular membranes. Not surprisingly, therefore, many studies implicate BCL-2 in regulating ER Ca²⁺ homeostasis. However, the mechanism by which it does so is controversial and may depend on the cell type tested. For example, several groups have reported that BCL-2 increases the ER Ca²⁺ content and/or prevents ER Ca²⁺ release during apoptosis (Lam et al., 1994; Distelhorst et al., 1996; He et al., 1997; Ichimiya et al., 1998; Nutt et al., 2002). In contrast, studies by Rizzuto and co-workers convincingly demonstrated that overexpression of BCL-2 reduces the steady-state level of [Ca²⁺]_ER by increasing the permeability of the ER membrane to Ca²⁺ (Pinton et al., 2000). Consistent with this latter finding, there are now several examples in the literature where manipulations that lead to decreased [Ca²⁺]_ER protected cells against apoptosis, whereas manipulations that increased [Ca²⁺]_ER sensitized cells to apoptosis (Nakamura et al., 2000, Pinton et al., 2001; Lilliehook et al., 2002). Changes in [Ca²⁺]_ER affect the intensity IP₃R/RyR Ca²⁺ spikes, which are known to sensitize mitochondria to BH3-dependent cyt.c release during apoptosis (Csordas et al., 2002). It has also been reported that BCL-2 can increase the capacity of mitochondria to store Ca²⁺ (Murphy et al., 1996; Ichimiya et al., 1998; Zhu et al., 1999), presumably by preventing the opening of the PTP, which releases matrix Ca²⁺. Thus, by localizing to both ER and mitochondria, BCL-2 might prevent apoptotic crosstalk between the two compartments by lowering the amount of free [Ca²⁺]_ER for IP₃R/RyR release and by increasing the tolerance of mitochondria to high Ca²⁺ loads.

BCL-2 family proteins and the ER: proapoptotic BAX and BAK

The proapoptotic multidomain BCL-2 family members BAX and BAK have also been reported to localize to the ER in some cell types (Nutt et al., 2001) and these proteins may induce apoptosis from this location. For example, when overexpressed in human PC-3 cells, BAX and BAK localize to both the mitochondria and ER, and induce caspase-independent emptying of ER Ca²⁺ pools concomitant with an increase in [Ca²⁺]_m (Nutt et al., 2001; Pan et al., 2001). Coexpression of BCL-2/BCL-xL inhibits this Ca²⁺ mobilization, and an inhibitor of mitochondrial Ca²⁺ uptake, RU360, blocked BAX/BAK- and staurosporin-induced increase in [Ca²⁺]_m, and subsequent cyt.c release and apoptosis (Nutt et al., 2001). Moreover, BAX-null DU-145 cells resist staurosporin-induced ER Ca²⁺ release, mitochondrial Ca²⁺ uptake and cyt.c release, all of which are overcome by re-expression of BAX (Nutt et al., 2002). Consistent with a role for these proteins at the ER, BAK and BAX interact with the cytosolic tail of the ER chaperone calnexin in yeast, and BAK-induced lethality in S. pombe is dependent upon this interaction (Torgler et al., 1997). Moreover, b5-BCL-2 can inhibit BAX-induced apoptosis (Wang et al., 2001) as can BAX inhibitor-1, an antiapoptotic transmembrane protein of the ER (Xu and Reed, 1998). BAX, BAK double-null cells, which are variant to diverse apoptosis stimuli; have reduced [Ca²⁺]_ER that result in decreased Ca²⁺ uptake by the mitochondria following release of Ca²⁺ from the ER by apoptotic agents (Scorrano et al., 2003). Over expression of SERCA corrected the Ca²⁺ imbalance in these cells and fully or partially restored apoptosis in response to several intrinsic cell death stimuli. Thus BAX and BAK have dual roles at the ER and mitochondria and regulate Ca²⁺-dependent cross link between the two organelles.

BCL-2 family proteins and the ER: BH3-only members

Recent models hold that the BH3-only subset of proapoptotic BCL-2 family members sense diverse death signals and respond by initiating caspase activation (Puthalakath and Strasser, 2002). BH3-only molecules may achieve this by binding and inhibiting antiapoptotic BCL-2 family members or by directly activating proapoptotic BAX and BAX (Letai et al., 2002). Certain BH3-only molecules, after activation, translocate to mitochondria and induce the oligomerization of BAX and BAX into predicted pores in the OMM, facilitating egress of the protein contents of the intermembrane space. Consistent with this model, most BH3-only proteins localize to mitochondria, at least in part, and trigger cyt.c release, which in all cases is inhibited by BCL-2 overexpression.

Given that representatives from all three groups of the BCL-2 family have been found at the ER, however, it is likely that the ratiometric interactions between antiapoptotic and proapoptotic members that regulate the mitochondrial apoptotic pathway likely also regulate apoptotic pathways at the ER (Figure 2). The BH3-only protein BIK, for example, is located primarily at the ER (Mathai et al., 2002). BIK mRNA and protein are induced by p53 in response to DNA damage or oncogenic stress (Mathai et al., 2002; unpublished), and ER BIK induces cyt.c release independent of an association with mitochondria or zVAD-sensitive caspases (Germain et al., 2002). An S9 fraction containing both microsomes and cytosol, but not either fraction alone, isolated from BIK-treated H1299 cells (in the presence of caspase inhibitors) can induce mitochondrial release of cyt.c in vitro. This suggests that BIK activates factors in both the ER and cytosol to induce mitochondrial transformations (Germain et al., 2002). This activity of BIK is influenced by BCL-2 at the ER since BIK/BCL-2 heterodimers can be crosslinked at the surface of the ER and high BIK : BCL-2 ratios change the set of proteins BCL-2 interacts with at the ER, resulting in Ca²⁺ release, egress of cyt.c from mitochondria, and apoptosis (unpublished). When in excess, however, ER-localized BCL-2 is able to protect against BIK-induced apoptosis. The ratio of BH3-only and antiapoptotic BCL-2 family members at the ER, therefore, may regulate downstream signals emitted from the organelle, including Ca²⁺ signals and/or proximal zVAD-insensitive caspase activation, which in turn regulate mitochondrial release of cyt.c. Other BH3-only proteins, such as BIM and NOXA (Puthalakath et al., 1999; Oda et al., 2000), also partially locate to the ER and, therefore, may elicit similar effects.

Figure 2.

BH3-driven apoptotic pathways at the ER. Diverse apoptotic stresses may increase the concentration of BH3-only proteins, such as BIK, at the ER. When in excess, BH3-only proteins may mobilize ER Ca²⁺ stores and sensitize mitochondria to undergo apoptotic transitions in response to a direct BH3 hit, or activate other Ca²⁺-dependent processes in the cytosol. BH3-only proteins might also stimulate other Ca²⁺-indpendent pathways leading to cyt.c release from the mitochondria or facilitate local procaspase-12, procaspase-8L, or procaspase-2 activation at the ER surface. To accomplish these tasks, BH3-only proteins may simply relieve BCL-2 repression of these processes or, additionally, they may play a second role in activating these pathways after BCL-2 is repressed. It remains to be determined if the ratiometric influence of anti- and proapoptotic members also includes BAX,BAK

Full figure and legend (19K)

By analogy to their known functions at mitochondria, it is also possible that certain BH3-only proteins induce BAX/BAK oligomerization in the ER membrane and mediate release of lumenal proteins that regulate caspase activation. Additionally, BAX/BAK pores might control ER Ca²⁺ stores, influencing local [Ca²⁺]_c and/or facilitating mitochondrial Ca²⁺ uptake and sensitization for cyt.c release. This model is consistent with pore forming properties of BCL-2 proteins, their ability to modulate movement of Ca²⁺ across the ER membrane, and the dependence of most BH3-only proteins on BAX/BAK to induce apoptosis (Cheng et al., 2001). Moreover, the ER undergoes profound dilation and unfolding of cisternae during apoptosis (Weller et al., 1995; Wu et al., 1999; Chang et al., 2000; Herrera et al., 2000; Johnson et al., 2000; Muriel et al., 2000; Zeng and Xu, 2000) suggesting that ER homeostasis and volume control is lost, perhaps due to membrane permeabilization.

Regulation of initiator caspases at the ER by BCL-2 proteins

Other apoptotic models posit that of BCL-2 proteins regulate caspase activation upstream or in parallel to the mitochondrial pathway (Huang and Stasser, 2000) and that mitochondria amplify weak initiator caspase signals. As discussed above, caspase-12 activation at the ER membrane clearly represents one mechanism of mitochondria-independent initiator caspase activation. The ability of ER-localized BCL-2 family members to directly regulate this caspase certainly deserves testing. Additionally, the novel procaspase-8 isoform, procaspase-8L, like procaspase-12, is peripherally associated with the cytosolic face of the ER (Breckenridge et al., 2002). However, procaspase-8L processing seems to be regulated through an association with the BAP31 complex. BAP31 is a ubiquitously expressed polytopic integral membrane protein of the ER that associates with related BAP29, components of the actomyosin network, and BCL-2/BCL-xL (Adachi et al., 1996; Ng et al., 1997; Nguyen et al., 2000). Procaspase-8L, but not procaspase-8, is selectively recruited to the BAP31 complex during oncogenic E1A-induced apoptosis, which coincides with its proteolytic processing. This recruitment is dependent on procaspase-8L's Nex domain, a 59 amino acid stretch that extends canonical procaspase-8/a at the N-terminus. E1A-induced processing (and presumably activation) of procaspase-8L is regulated by its association with BAP proteins since cells doubly deficient in Bap31 and Bap29 do not process procaspase-8L in response to E1A and show reduced caspase-8 and -3 activity and subsequent apoptosis. It is possible that the cytosolic tails of BAP31 and BAP29 recruit factors that facilitate procaspase-8L activation. Importantly, BCL-2 overexpression does not inhibit recruitment of procaspase-8L to the BAP31 complex but abrogates subsequent enzyme processing and apoptosis. Moreover, E1A-induced processing of procaspase-8L is subject to ratiometric regulation by BIK and BCL-2 located at the ER (unpublished), suggesting that it might be a target of BH3-initiated caspase activation. At present, it is not known whether other initiator caspases respond to signals at the ER. Given that initiator procaspase-2 has been implicated upstream of mitochondrial dysfunction during drug-induced apoptosis (Lassus et al., 2002) and that this caspase is located at Golgi and ER membranes in some cell types (Mancini et al., 2000), however, it is conceivable that procaspase-2 is also regulated by BCL-2 proteins at the ER.

The massive surface area of the ER membrane may provide a platform for the assembly of caspase regulatory complexes. This model parallels apoptotic regulation in C. elegans, wherein a BCL-2 protein (Ced-9) binds an adapter molecule (Ced-4) preventing it from activating a caspase (Ced-3). By binding Ced-9, BH3-only Egl-1 initiates apoptosis by displacing Ced-4 (Metzstein et al., 1998). BCL-2 can compensate for Ced-9 loss-of-function mutations in nematodes, indicating that BCL-2 may be able to function in a similar manner in mammalian cells (Hengartner and Horvitz, 1994). For example, upregulation of BIK following oncogenic stress increases the BIK : BCL-2 ratio at the ER, which might relieve the inhibitory effect of BCL-2 on procaspase-8L processing. ER proteins such as the reticulon family members NSP-C and RTN-x_s, which bind and inhibit BCL-2 and BCL-xL (Tagami et al., 2000), or p53-inducible genes that target the ER, including PIDD and Scotin (Lin et al., 2000; Bourdon et al., 2002), might also play a role in the regulation of caspase activation at the ER. Alternatively, high BH3/BCL-2 ratios may relieve BCL-2 inhibition of some other apoptotic process at the ER, notably calcium homeostasis.

The various models for the function of BCL-2 family proteins at the ER are not mutually exclusive and one could imagine that, depending on the apoptotic stimulus, multiple signalling pathways could emerge from the ER to either directly activate the mitochondrial pathway (as with BIK) via caspase-dependent and -independent means, or to sensitize mitochondria to a secondary direct BH3 hit at this organelle (as with ER Ca²⁺ release), or to synergize with mitochondrial cyt.c release to activate downstream caspases (as with the caspase-12/ER stress pathway) (Figure 2).

Caspase substrates located at the ER

Several ER proteins have now been identified as caspase substrates. The majority seem to be caspase-3 targets and, therefore, their cleavage likely contributes to the coordinated shutdown of normal cellular processes during the execution phase of apoptosis.

BAP31

An exception, however, is BAP31, which in addition to being a regulator of procaspase-8L, is also a caspase-8 substrate itself (Ng et al., 1997). The cytosolic tail of human BAP31 contains two identical caspase recognition sequences that are rapidly cleaved by caspase-8 following stimulation of the Fas death receptor (Nguyen et al., 2000; Wang et al., 2003), and presumably following procaspase-8L activation at the ER (Breckenridge et al., 2002). Fas initiates apoptosis by directly recruiting and activating procaspase-8 at the plasma membrane. Caspase-8 in turn cleaves the BH3-only molecule, BID, generating tBID, which induces BAX/BAK-dependent cyt.c release from mitochondria and subsequent activation of downstream caspases (Korsmeyer et al., 2000). In some cell types, Fas signalling does not require the participation of mitochondria for apoptotic execution because sufficient caspase-8 is activated to permit direct cleavage and activation of downstream caspases (Scaffidi et al., 1998). Stable expression of a caspase-resistant BAP31 (crBAP31) mutant, in which the caspase-8 recognition asp residues have been substituted with ala, strongly inhibits Fas-induced apoptosis in KB epithelial cells (Nguyen et al., 2000). While crBAP31 had a very weak influence on Fas-induced caspase activation or cleavage of effector caspase substrates, it strongly inhibited cyt.c release, mitochondrial depolarization, cytoskeletal reorganization, and membrane blebbing. The fact that crBAP31 inhibits these events in the face of caspase activation suggests that, at least in these intact epithelial cells, the ER exerts a restraint on apoptotic mitochondrial transition and cytoplasmic restructuring that is overcome by BAP31 cleavage. Interestingly, crBAP31 did not inhibit Fas-induced cleavage of BID or downstream insertion of BAX into mitochondria, but strongly inhibited subsequent activation and oligomerization of BAX and BAK (Wang et al., 2003). To exert this antiapoptotic effect on mitochondria, BAP31 requires an association with the putative ion channel protein of the ER, A4. Although speculative at present, it is possible that BAP31–A4 complex might regulate Ca²⁺ flux between the ER and mitochondria during Fas-mediated apoptosis, which in turn sensitizes mitochondria to tBID (see below). Indeed, dynamic ER Ca²⁺ spikes have been implicated in mediating TNF--induced cyt.c release (Pu et al., 2002).

Caspase-8 cleavage of BAP31 generates a membrane embedded p20 fragment that induces cell death when expressed ectopically for prolonged periods (Ng et al., 1997). Adenoviral expression of p20 induces an immediate release of Ca²⁺ from the ER, which is followed by mitochondrial recruitment of Drp1, a critical mediator of OMM fission, and dramatic fragmentation and fission of the mitochondrial network into small punctiform organelles (Breckenridge et al., 2003). p20 seems to regulate the fission machinery through Ca²⁺ signals because inhibiting ER–mitochondria Ca²⁺ transmission prevents Drp1 redistribution and mitochondrial fission. Interestingly, ER Ca²⁺ signals have also been shown to regulate mitochondrial restructuring during ceramide-induced apoptosis (Pinton et al., 2001). Given that Drp1-dependent mitochondrial fission is a requirement for cyt.c release during apoptosis (Frank et al., 2001), p20 might assist other caspase-8 substrates, such as tBID, to induce mitochondrial cristae restructuring (Scorrano et al., 2002) and cyt.c release during apoptosis. Accordingly, mitochondria that had undergone p20-induced fission were strongly sensitized to caspase-8-induced cyt.c release. Collectively these data are consistent with the 'two-hit' model described by Hajnoczky and co-workers (Szalai et al., 1999), in which an ER Ca²⁺ signal cooperates with a direct BH3-only protein hit on mitochondria to efficiently promote the mitochondrial release of cyt.c. Of note, crBAP31 inhibits Fas-induced ER Ca²⁺ release and mitochondrial fission (our unpublished data). These opposing effects of crBAP31 and p20 on ER Ca²⁺ homeostasis and mitochondrial integrity are independent of each other since reconstitution of crBAP31 into Bap31-null cells (which cannot generate p20) still prevents caspase-8-induced cyt.c release, and p20 still induces its proapoptotic effects in the absence of Bap31. Thus, caspase cleavage of BAP31, like other antiapoptotic proteins (Cheng et al., 1997; Clem et al., 1998), converts it from an inhibitor to activator of apoptosis (Figure 3).

Figure 3.

Cleavage of BAP31 at the ER sensitizes mitochondria to caspase-8-initiated release of cyt.c to the cytosol in intact cells. Activation of caspase-8 leads to simultaneous processing of BID and BAP31. tBID activates cristae remodelling and BAK/BAX oligomerization in the OMM providing a conduit for cyt.c efflux across, enhancing cytic release the membrane. Cleavage of BAP31 generates p20, which triggers ER Ca²⁺ release and Drp1-dependent mitochondrial fission. In contrast, full-length BAP31 prevents Fas-induced mitochondrial fission and cyt.c release. ER-driven Ca²⁺ signals are likely to be modulators of apoptotic mitochondrial transition rather than obligatory events

Full figure and legend (18K)

SREBPs

Cleavage of other ER proteins by downstream executioner caspases might contribute to morphological progression of apoptosis or in some cases simply disable cellular processes at the ER during cell suicide. For example, caspase-3 cleavage of the sterol-regulatory element binding proteins (SREBPs) SREBP-1 and SREBP-2 upregulates sterol response genes, which may affect lipid rearrangements and/or morphological changes of the plasma membrane during apoptosis (Higgins and Ioannou, 2001). SREBPs are ER membrane-bound transcription factors that are normally transported to the Golgi in sterol-depleted cells, where they undergo sequential cleavage by Site-1 and -2 proteases (DeBose-Boyd et al., 1999). This cleavage releases an N-terminal basic helix–loop–helix leucine zipper transcription factor domain to the nucleus that activates transcription of the genes involved in sterol metabolism, such as the LDL receptor and HMG-CoA reductase. Caspase cleavage of SREBPs occurs at the ER during apoptosis, independent of cellular sterol levels, and liberates an active transcription factor to the nucleus (Wang et al., 1996).

SRP 72

Caspase cleavage of the 72-kDa component of the signal recognition particle (SRP 72) might shutdown or alter the translation of secretory proteins during apoptosis. SRP is composed of six polypeptides and a 7S structural RNA, which binds the signal sequence of newly synthesized polypeptides as they surface from the ribosome, arrests translation elongation until the complex has bound the SRP receptor at the ER membrane, and reinitiates translation elongation and translocation into the ER lumen or membrane (Keenan et al., 2001). SRP 72 is thought to help facilitate binding of SRP to the SRP receptor and translocation of polypeptides across the ER membrane. Caspase-3 cleavage of SRP-72 removes a short conserved, and highly serine phosphorylated C-terminal fragment that may be important for regulation in vivo, although the truncated protein can still function in vitro (Utz et al., 1998).

IP₃ receptor and GRP94

IP₃R1 and 2 are cleaved by caspase-3, and IP₃R3 undergoes calpain proteolysis during apoptosis, which ablates IP₃-induced Ca²⁺ flux (Hirota et al., 1999; Diaz and Bourguignon, 2000). In the light of the important role of IP₃R signalling during the early stages of apoptosis (Jayaraman and Marks, 1997), caspase-mediated proteolysis might represent functional downregulation of this channel during the execution phase of apoptosis. Similarly, calpain proteolysis of the Ca²⁺-binding ER chaperone GRP94 may also affect Ca²⁺ signalling during apoptosis (Reddy et al., 1999). Overexpression of GRP94 confers resistance to apoptosis, and antisense knockdown of GRP94 sensitizes cells to etoposide-induced apoptosis.

Presenilins and APP

Caspases cleave presenilins and the amyloid precursor protein (APP), which are involved in the pathogenesis of Alzheimer's disesase (AD). Presenilin-1 (PS1) and -2 (PS2), are highly homologous integral membrane proteins of the ER and Golgi that are mutated in most aggressive early-onset forms of familial AD (Wellington and Hayden, 2000). Both presenilins have eight predicted transmembrane domains, and a cytosolic hydrophilic loop between transmembrane domains 6 and 7. Cleavage within the cytosolic loop by endoproteases generates stable C-terminal and N-terminal fragments, which are the functionally active form of the protein. Although the exact function of the presenilins is not known, PS1 and PS2 modulate APP cleavage by -secretase and the sensitivity of cells to apoptosis (Selkoe, 1999). Overexpression of full-length wt PS1 and PS2 or their AD mutants sensitizes cells to apoptosis (Wolozin et al., 1996; Wellington and Hayden, 2000) by increasing the ER Ca²⁺ load and agonist-induced ER Ca²⁺ release (Mattson et al., 2001) or by binding BCL-2 (Alberici et al., 1999; Passer et al., 1999). Conversely, antisense knockdown of PS2 or expression of the mature C-terminal fragments of PS2 confers protection to apoptosis (Wellington and Hayden, 2000). PS1 and PS2 are cleaved within their C-terminal fragments by caspases, which abrogates their antiapoptotic activity. Interestingly, caspase cleavage of PS2 is regulated by phosphorylation of two Ser residues overlapping the cleavage sites at Asp₃₂₆ and Asp₃₂₉. Phosphorylation of these residues inhibits caspase-3 cleavage in vitro, and mutation of the Ser residues to Asp or Glu (which mimic the phosphorylated state) blocks caspase cleavage in vivo and enhances the antiapoptotic function of PS2 (Walter et al., 1999). AD-causing mutations in presenilins shift the normal processing pattern of APP by secretases toward the production of the amyloidogenic and toxic A-42 peptide, which is linked to AD. In addition, caspase-3 cleavage of APP at VEVD⁷²⁰ and caspase-6 cleavage at VNLD⁶⁵³ in patients harboring the Swedish APP mutation increases the amount of APP that is processed to A-42 (Gervais et al., 1999; Wellington and Hayden, 2000). Lu et al. (2000) reported that the short intracellular C-terminal fragment arising from caspase cleavage at VEVD⁷²⁰ is also proapoptotic and might contribute to AD pathogenesis. Therefore, caspase cleavage of presenilins and APP creates an apoptotic amplification cycle, accelerating cell death in the neurons of AD patients: caspase cleavage of APP increases the generation of A-42, which is proapoptotic and in turn leads to further caspase activation and APP processing, and caspase cleavage of presenilins increases the sensitivity of neurons to apoptosis induced by A-42 (Wellington and Hayden, 2000). Interestingly, A-42-induced apoptosis seems to be dependent on caspase-12 activation at the ER (Nakagawa et al., 2000), although it is not clear how A-42 exerts this effect.

Future perspective

Although many studies implicate a regulatory role for the ER in apoptosis, the signalling pathways that emerge from this organelle remain largely obscure. In the case of ER stress, where the apoptotic signal originates within the ER itself, it is clear that the ER has its own complement of apoptotic accessories that independently activate caspases and mitochondrial dysfunction. It remains to be determined whether some of these ER signalling pathways function in other apoptotic programs. In contrast, ER Ca²⁺ signals that affect the ability of mitochondria to release IMS proteins in intact cells may be a general component of many apoptosis pathways. It will be important to understand the nature of proapoptotic Ca²⁺ signals, how apoptotic signals converge on Ca²⁺ release channels, and how BCL-2 proteins regulate this process. Most models of cyt.c release stem from studies testing the effect of recombinant BCL-2 family members on isolated mitochondria (or liposomes), where the dynamic nature of the mitochondrial network and its contacts with the ER and cytoskeleton are lost. These may well impact important optimal steps in apoptotic transformations of mitochondria in intact cells, including remodelling of cristae (Scorrano et al., 2002) and organelle fission (Breckenridge et al., 2003; Frank et al., 2001). In physiological settings, the function of proapoptotic BH3-only proteins may be suboptimal and their ability to transform mitochondria into a state competent for efficient cyt.c release may largely depend on the alliance with costimulatory signals from the ER.

A better understanding of the ER in apoptosis may have to await the clarification of BCL-2 function at this location. Do BCL-2 proteins operate at the ER as they do at mitochondria and influence membrane permeability? If so, and if proapoptotic BAX and BAK are also involved, is the mechanism different from that at the OMM (Kuwana et al., Cell, 2002)? BCL-2 proteins may also play novel roles at the ER, for example by regulating initiator caspase activation complex(es) analogous to Ced-9 in C. elegans. Since deregulation of BCL-2 family members at the ER affects cell survival outcomes, an understanding of their functions at this site could be important for developing new therapeutics for treating diseases such as cancer. To date, however, little has emerged in the way of elucidating the critical ER transitions during apoptosis initiation that might impact the immediate downstream pathways. In retrospect this leads one to contemplate on what our current vision of mitochondria in apoptosis would be had cyt.c not been discovered as a cytosolic factor required for caspase-3 activation in vitro.

References

1. Adachi T, Schamel WW, Kim KM, Watanabe T, Becker B, Nielsen PJ and Reth M. (1996). EMBO J., 15, 1534–1541. | PubMed | ISI | ChemPort |
2. Alberici A, Moratto D, Benussi L, Gasparini L, Ghidoni R, Gatta LB, Finazzi D, Frisoni GB, Trabucchi M, Growdon JH, Nitsch RM and Binetti G. (1999). J. Biol. Chem., 274, 30764–30769. | PubMed |
3. Annis MG, Zamzami N, Zhu W, Penn LZ, Kroemer G, Leber B and Andrews DW. (2001). Oncogene, 20, 1939–1952. | Article | PubMed | ISI | ChemPort |
4. Antonsson B and Martinou JC. (2000). Exp. Cell Res., 256, 50–57. | Article | PubMed | ISI | ChemPort |
5. Baud V and Karin M. (2001). Trends Cell Biol., 11, 372–377. | Article | PubMed | ISI | ChemPort |
6. Berridge MJ, Lipp P and Bootman MD. (2000). Nat. Rev. Mol. Cell Biol., 1, 11–21. | Article | PubMed | ISI | ChemPort |
7. Bourdon JC, Renzing J, Robertson PL, Fernandes KN and Lane DP. (2002). J. Cell Biol., 158, 235–246. | Article | PubMed | ISI | ChemPort |
8. Boya P, Cohen I, Zamzami N, Vieira HL and Kroemer G. (2002). Cell Death Differ., 9, 465–467. | Article | PubMed | ISI | ChemPort |
9. Breckenridge DG, Nguyen M, Kuppig S, Reth M and Shore GC. (2002). Proc. Natl. Acad. Sci. USA, 99, 4331–4336. | Article | PubMed | ChemPort |
10. Breckenridge DG, Stoovic M, Marcellus R and Shore GC. (2003). J. Cell. Biol., 160, 1115–1127. | Article | PubMed | ISI | ChemPort |
11. Calfon M, Zeng H, Urano F, Till JH, Hubbard SR, Harding HP, Clark SG and Ron D. (2002). Nature, 415, 92–96. | Article | PubMed | ISI | ChemPort |
12. Carlberg M, Dricu A, Blegen H, Kass GE, Orrenius S and Larsson O. (1996). Carcinogenesis, 17, 2589–2596. | Article | PubMed | ISI | ChemPort |
13. Cavalli AL, O'Brien NW, Barlow SB, Betto R, Glembotski CC, Palade PT and Sabbadini RA. (2002). Am. J. Physiol. Cell Physiol., 6, 6.
14. Chang SH, Phelps PC, Berezesky IK, Ebersberger Jr ML and Trump BF. (2000). Am. J. Pathol., 156, 637–649. | PubMed | ISI | ChemPort |
15. Cheng EH, Kirsch DG, Clem RJ, Ravi R, Kastan MB, Bedi A, Ueno K and Hardwick JM. (1997). Science, 278, 1966–1968. | Article | PubMed | ISI | ChemPort |
16. Cheng EH, Wei MC, Weiler S, Flavell RA, Mak TW, Lindsten T and Korsmeyer SJ. (2001). Mol. Cell, 8, 705–711. | Article | PubMed | ISI | ChemPort |
17. Clem RJ, Cheng EH, Karp CL, Kirsch DG, Ueno K, Takahashi A, Kastan MB, Griffin DE, Earnshaw WC, Veliuona MA and Hardwick JM. (1998). Proc. Natl. Acad. Sci. USA, 95, 554–559. | Article | PubMed | ChemPort |
18. Csordas G, Madesh M, Antonsson B and Hajnoczky G. (2002). EMBO J., 21, 2198–2206. | Article | PubMed | ISI | ChemPort |
19. DeBose-Boyd RA, Brown MS, Li WP, Nohturfft A, Goldstein JL and Espenshade PJ. (1999). Cell, 99, 703–712. | Article | PubMed | ISI | ChemPort |
20. Diaz F and Bourguignon LY. (2000). Cell Calcium, 27, 315–328. | Article | PubMed | ISI | ChemPort |
21. Distelhorst CW, Lam M and McCormick TS. (1996). Oncogene, 12, 2051–2055. | PubMed | ISI | ChemPort |
22. Fischer H, Koenig U, Eckhart L and Tschachler E. (2002). Biochem. Biophys. Res. Commun., 293, 722–726. | Article | PubMed | ISI | ChemPort |
23. Frank S, Gaume B, Bergmann-Leitrar ES, Laitmer WW, Robert EG, Catez F, Smith CL and Youle RS. (2001). Dev. Cell, 1, 515–525. | Article | PubMed | ISI | ChemPort |
24. Fujita E, Kouroku Y, Jimbo A, Isoai A, Maruyama K and Momoi T. (2002). Cell Death Differ., 9, 1108–1114. | Article | PubMed | ISI | ChemPort |
25. Germain M, Mathai JP and Shore GC. (2002). J. Biol. Chem., 277, 18053–18060. | Article | PubMed | ISI | ChemPort |
26. Gervais FG, Xu D, Robertson GS, Vaillancourt JP, Zhu Y, Huang J, LeBlanc A, Smith D, Rigby M, Shearman MS, Clarke EE, Zheng H, Van Der Ploeg LH, Ruffolo SC, Thornberry NA, Xanthoudakis S, Zamboni RJ, Roy S and Nicholson DW. (1999). Cell, 97, 395–406. | Article | PubMed | ISI | ChemPort |
27. Green DR and Reed JC. (1998). Science, 281, 1309–1312. | Article | PubMed | ISI | ChemPort |
28. Hacki J, Egger L, Monney L, Conus S, Rosse T, Fellay I and Borner C. (2000). Oncogene, 19, 2286–2295. | Article | PubMed | ISI | ChemPort |
29. Hajnoczky G, Csordas G, Madesh M and Pacher P. (2000). Cell Calcium, 28, 349–363. | Article | PubMed | ISI | ChemPort |
30. Harding HP, Zhang Y, Bertolotti A, Zeng H and Ron D. (2000). Mol. Cell, 5, 897–904. | Article | PubMed | ISI | ChemPort |
31. Harding HP, Zhang Y and Ron D. (1999). Nature, 397, 271–274. | Article | PubMed | ISI | ChemPort |
32. He H, Lam M, McCormick TS and Distelhorst CW. (1997). J. Cell Biol., 138, 1219–1228. | Article | PubMed | ISI | ChemPort |
33. Hengartner MO and Horvitz HR. (1994). Cell, 76, 665–676. | Article | PubMed | ISI | ChemPort |
34. Herrera PL, Harlan DM and Vassalli P. (2000). Proc. Natl. Acad. Sci. USA, 97, 279–284. | Article | PubMed | ChemPort |
35. Higgins ME and Ioannou YA. (2001). J. Lipid Res., 42, 1939–1946. | PubMed |
36. Hirota J, Furuichi T and Mikoshiba K. (1999). J. Biol. Chem., 274, 34433–34437. | Article | PubMed | ISI | ChemPort |
37. Huang DC and Shamer A. (2000). Cell, 103, 839–842. | Article | PubMed | ISI | ChemPort |
38. Ichimiya M, Chang SH, Liu H, Berezesky IK, Trump BF and Amstad PA. (1998). Am. J. Physiol., 275, C832–C839. | PubMed | ISI | ChemPort |
39. Ito Y, Pandey P, Mishra N, Kumar S, Narula N, Kharbanda S, Saxena S and Kufe D. (2001). Mol. Cell. Biol., 21, 6233–6242. | Article | PubMed | ISI | ChemPort |
40. Jayaraman T and Marks AR. (1997). Mol. Cell. Biol., 17, 3005–3012. | PubMed | ISI | ChemPort |
41. Johnson VL, Ko SC, Holmstrom TH, Eriksson JF and Chew SC. (2000). J. Cell. Biol., 113, 2941–2953.
42. Kaufman RJ. (1999). Genes Dev., 13, 1211–1233. | PubMed | ISI | ChemPort |
43. Keenan RJ, Freymann DM, Stroud RM and Walter P. (2001). Annu. Rev. Biochem., 70, 755–775. | Article | PubMed | ISI | ChemPort |
44. Korsmeyer SJ, Wei MC, Saito M, Weiler S, Oh KJ and Schlesinger PH. (2000). Cell Death Differ., 7, 1166–1173. | Article | PubMed | ISI | ChemPort |
45. Kuwana T, Mackey MR, Perkins G, Ellison MH, Latterich M, Schmeiter R, Green DR and Newmeyer DD. (2002). Cell, 111, 331–342. | Article | PubMed | ISI | ChemPort |
46. Lam M, Dubyak G, Chen L, Nunez G, Miesfeld RL and Distelhorst CW. (1994). Proc. Natl. Acad. Sci. USA, 91, 6569–6573. | Article | PubMed | ChemPort |
47. Lassus P, Opitz-Araya X and Lazebnik Y. (2002). Science, 297, 1352–1354. | Article | PubMed | ISI | ChemPort |
48. Letai A, Bassik M, Walensky L, Sorcinelli M, Weiler S and Korsmeyer S. (2002). Cancer Cell, 2, 183. | Article | PubMed | ISI | ChemPort |
49. Li C, Fox CJ, Master SR, Bindokas VP, Chodosh LA and Thompson CB. (2002). Proc. Natl. Acad. Sci. USA, 99, 9830–9835. | Article | PubMed | ChemPort |
50. Li H, Kolluri SK, Gu J, Dawson MI, Cao X, Hobbs PD, Lin B, Chen G, Lu J, Lin F, Xie Z, Fontana JA, Reed JC and Zhang X. (2000). Science, 289, 1159–1164. | Article | PubMed | ISI | ChemPort |
51. Lilliehook C, Chan S, Choi EK, Zaidi NF, Wasco W, Mattson MP and Buxbaum JD. (2002). Mol. Cell. Neurosci., 19, 552–559. | Article | PubMed | ISI | ChemPort |
52. Lin Y, Ma W and Benchimol S. (2000). Nat. Genet., 26, 122–127. | Article | PubMed | ISI | ChemPort |
53. Lu DC, Rabizadeh S, Chandra S, Shayya RF, Ellerby LM, Ye X, Salvesen GS, Koo EH and Bredesen DE. (2000). Nat. Med., 6, 397–404. | Article | PubMed | ISI | ChemPort |
54. Ma Y and Hendershot LM. (2001). Cell, 107, 827–830. | Article | PubMed | ISI | ChemPort |
55. Mancini M, Machamer CE, Roy S, Nicholson DW, Thornberry NA, Casciola-Rosen LA and Rosen A. (2000). J. Cell Biol., 149, 603–612. | Article | PubMed | ISI | ChemPort |
56. Mathai JP, Germain M, Marcellus RC and Shore GC. (2002). Oncogene, 21, 2534–2544. | Article | PubMed | ISI | ChemPort |
57. Mattson MP, Chan SL and Camandola S. (2001). BioEssays, 23, 733–744. | Article | PubMed | ISI | ChemPort |
58. McCormick TS, McColl KS and Distelhorst CW. (1997). J. Biol. Chem., 272, 6087–6092. | PubMed | ChemPort |
59. McCullough KD, Martindale JL, Klotz LO, Aw TY and Holbrook NJ. (2001). Mol. Cell. Biol., 21, 1249–1259. | Article | PubMed | ISI | ChemPort |
60. Metzstein MM, Stanfield GM and Horvitz HR. (1998). Trends Genet., 14, 410–416. | Article | PubMed | ISI |