|Home | About | Journals | Submit | Contact Us | Français|
The prosurvival Bcl-2-family member Bfl-1/A1 is a transcriptional target of nuclear factor-κB (NF-κB) that is overexpressed in many human tumors and is a means by which NF-κB inhibits apoptosis, but its mode of action is controversial. To better understand how Bfl-1 functions, we investigated its interaction with proapoptotic multidomain proteins Bax and Bak, and the BH3-only proteins Bid and tBid. We demonstrate that in living cells Bfl-1 selectively interacts with Bak and tBid, but not with Bax or Bid. Bfl-1/Bak interaction is functional as Bfl-1 suppressed staurosporine (STS)-induced apoptosis in wild-type and Bax-deficient cells, but not in Bak−/− cells. We also show that Bfl-1 blocks tumor necrosis factor-α (TNFα)-induced activation of Bax indirectly, via association with tBid. C-terminal deletion decreased Bfl-1’s interaction with Bak and tBid and reduced its ability to suppress Bak- and tBid-mediated cell death. These data indicate that Bfl-1 utilizes different mechanisms to suppress apoptosis depending on the stimulus. Bfl-1 associates with tBid to prevent activation of proapoptotic Bax and Bak, and it also interacts directly with Bak to antagonize Bak-mediated cell death, similar to Mcl-1. Thus, part of the protective function of NF-κB is to induce Mcl-1-like activity by upregulating Bfl-1.
Interactions between Bcl-2 family proteins control cell fate in response to death-inducing signals (Cory et al., 2003). Proapoptotic BH3-only proteins (Bid, Bad, Bim, Puma, Noxa, Hrk, Bmf, Nbk/Bik) initiate apoptotic signaling in response to death-inducing stimuli. Multidomain proteins Bax and Bak are essential effectors of apoptosis and cells lacking both proteins are refractory to apoptosis induced by a broad spectrum of stimuli. In contrast anti-apoptotic Bcl-2 family members Bcl-2, Bcl-xL, Bfl-1/A1, Mcl-1 and Bcl-w safeguard cells from apoptosis. Some BH3-only proteins such as tBid and Bim trigger conformational activation of Bax and Bak, while others such as Bad, Noxa and Bik sequester anti-apoptotic Bcl-2 family members and neutralize their protective activity (Letai et al., 2002; Chen et al., 2005; Kuwana et al., 2005).
Apoptosis can be initiated through the extrinsic pathway involving activation of death receptors or via the cell-intrinsic pathway triggered by various forms of cellular stress. In the cell-intrinsic pathway, apoptotic signals converge on mitochondria to trigger the release of cytochrome C (cyto C) into the cytosol, causing caspase activation and cell death. In the extrinsic pathway, ligands like tumor necrosis factor-α (TNFα) cause death receptor-mediated activation of caspase-8, which cleaves Bid into its active form tBid. In turn, tBid induces conformational activation and oligomerization of Bax and Bak, the release of mitochondrial cyto C, caspase activation and apoptosis (Perez and White, 2000; Wei et al., 2000). Whereas anti-apoptotic Bcl-2 and Bcl-xL can associate with both Bid and tBid, Mcl-1 inhibits death receptor-induced apoptosis by selectively binding to tBid but not Bid (Cheng et al., 2001; Clohessy et al., 2006). Moreover Mcl-1 sequesters endogenous Bak but not Bax in healthy cells and blocks Bak-mediated cell death, but Bcl-2 does not (Cuconati et al., 2003; Willis et al., 2005). The adenovirus Bcl-2 homologue E1B 19K antagonizes cell death by binding directly to both Bax and Bak, but not to tBid or Bid (Oltvai et al., 1993; Perez and White, 2000; Cuconati et al., 2002; Cuconati and White, 2002), but how other anti-apoptotic Bcl-2 family members like Bfl-1 block apoptosis is not entirely clear.
Bfl-1/A1 is a transcriptional target of nuclear factor-κB (NF-κB) that suppresses apoptosis in response to several death-inducing stimuli, including TNFα and staurosporine (STS; Karsan et al., 1996; D’Sa-Eipper and Chinnadurai, 1998; Grumont et al., 1999; Lee et al., 1999; Wang et al., 1999; Zong et al., 1999; Somogyi et al., 2001). Bfl-1/A1 lacks significant homology to the BH4 domain that is necessary for the protective activity of Bcl-2 and Bcl-xL and its C-terminal region contains charged residues and is unlikely to act as a transmembrane domain like those that anchor Bcl-2 and Bcl-xL to mitochondrial membranes (Nguyen et al., 1993). This suggests that Bfl-1 might act differently from Bcl-2 and Bcl-xL to block apoptosis, and there are conflicting reports regarding its ability to associate with Bax, Bak, Bid and tBid. Human Bfl-1 was reported to interact with Bid in transiently transfected cells and with recombinant tBid in vitro, but not with Bax or Bak (Werner et al., 2002). Bfl-1/A1 was described to associate with Bax in yeast two-hybrid or by co-immunoprecipitation of in vitro translated proteins (Sedlak et al., 1995; Zhang et al., 2000). Others found no association between mouse A1 and Bax in cotransfected cells, but A1 interacted with Bak under these conditions (Holmgreen et al., 1999). Here we investigated Bfl-1’s association with endogenous Bax, Bak, Bid and tBid in living cells and examined how these interactions relate to its protective activity in response to extrinsic and intrinsic death-inducing stimuli.
Since Bfl-1 is overexpressed in many cancers, we expressed Bfl-1 in MCF-7 cells and verified its ability to suppress mitochondrial cyto C release upon TNFα activation of the extrinsic death-signaling cascade. Bfl-1 was N-terminally tagged to green fluorescent protein (GFP), since commercial antibodies failed to successfully recognize human Bfl-1. GFP-Bfl-1 localized to mitochondria and the perinuclear region, overlapping with endogenous cyto C (Figure 1). While TNFα provoked cyto C release in GFP-expressing cells, cyto C remained mitochondrially localized in GFP-Bfl-1-positive cells, consistent with its protective activity toward TNFα (Karsan et al., 1996; Wang et al., 1999; Zong et al., 1999). A GFP-Bfl-1ΔC mutant lacking the C-terminal 24 amino acids was diffuse throughout the cell including the nucleus, similar to GFP, but still suppressed TNFα-induced cyto C release. Cyto C release and apoptosis are critically dependent on conformational activation and oligomerization of Bax and Bak, which is inhibited by anti-apoptotic Bcl-2 and Bcl-xL. Similarly, GFP-Bfl-1 and Bfl-1ΔC suppressed TNFα-induced Bax conformational activation, as seen with an antibody specific for the N terminus of Bax that is masked when Bax is inactive and is exposed upon Bax activation (Figure 2; Desagher et al., 1999; Perez and White, 2000). In contrast Bax was activated in surrounding cells lacking significant GFP-Bfl-1 expression. Thus, Bfl-1 and Bfl-1ΔC suppress Bax activation in response to TNFα.
Anti-apoptotic Bcl-2 family members block apoptosis by interacting with proapoptotic BH3-only proteins and/or with multidomain Bax and Bak. Bcl-2 and E1B 19K interact with Bax to block its oligomerization and apoptosis (Oltvai et al., 1993; Perez and White, 2000; Cuconati and White, 2002; Cuconati et al., 2002), but association of Bfl-1 with Bax has been a subject of conflicting reports (Sedlak et al., 1995; Holmgreen et al., 1999; Zhang et al., 2000; Werner et al., 2002). We asked if Bfl-1 antagonizes Bax activation via direct interaction in 3-((3-cholamidopropyl)dimethylammonio)-1-propanesulfonate (CHAPS) extracts, as other detergents can influence Bax conformation and its interaction with Bcl-2 (Hsu and Youle, 1998). Although Bfl-1 and Bfl-1ΔC co-immunoprecipitated with overexpressed FLAG-Bax (Figure 3a), neither associated with endogenous Bax activated by TNFα/cycloheximide (CHX) or with inactive Bax in CHX-treated cells, despite efficient Bax immunoprecipitation (Figure 3b). Reverse immunoprecipitation of endogenous Bax with GFP-Bfl-1 or 2×Myc-Bfl-1 similarly failed, unlike control immunoprecipitation of endogenous Bax with Bak (Figure 4a). Thus, although Bfl-1 can associate with overexpressed Bax, no complex was seen with endogenous Bax, suggesting that it antagonizes Bax activation by interacting with factors acting upstream in the TNF α cascade.
In contrast GFP-Bfl-1 and 2×Myc-Bfl-1 interacted strongly with endogenous Bak in untreated cells, similar to Mcl-1, and also with TNFα-activated Bak (Figures 4a and b). The reduced levels of Bak immunoprecipitated with 2×Myc-Bfl-1 in TNFα/CHX-treated cells agreed with lower levels of 2×Myc-Bfl-1 under these conditions, consistent with its turnover (Kucharczak et al., 2005; Herold et al., 2006). Bfl-1ΔC’s interaction with Bak was significantly weaker than that of Bfl-1 (Figure 4c). Thus, Bfl-1 associates with the inactive and active forms of endogenous Bak, whereas Bfl-1ΔC’s reduced interaction most likely results from its delocalization from mitochondria.
Bfl-1’s selective interaction with endogenous Bak is functional, as it suppressed intrinsic STS-induced apoptosis in wild-type and Bax-deficient immortal baby mouse kidney (iBMK) epithelial cells, but not in Bak-deficient iBMKs that express Bax but not Bak (Figure 4d; Degenhardt et al., 2002). Two independent iBMK clones of each genotype gave similar results (not shown). Bfl-1’s behavior was strikingly similar to Mcl-1’s, which also interacts with Bak but not Bax (Figures 4b and c; Cuconati et al., 2003; Gélinas and White, 2005; Willis et al., 2005). As anticipated, double-knockout iBMKs lacking Bax and Bak were highly resistant to apoptosis. Bfl-1ΔC was less protective than Bfl-1 in Bax−/− and wild-type iBMKs, consistent with its reduced association with Bak. As with Bfl-1, Bfl-1ΔC failed to suppress apoptosis in Bak−/− cells. This showed that Bfl-1’s selective interaction with Bak but not Bax correlates with suppression of Bak- but not Bax-mediated apoptosis, similar to Mcl-1.
TNFα-induced processing of Bid into tBid leads to activation of Bax and Bak. Since simultaneous inactivation of Bax and Bak is necessary to antagonize TNFα-induced death (Degenhardt et al., 2002) and Bfl-1 fails to interact with endogenous Bax, we asked if its ability to suppress Bax activation involved interaction with Bid and/or tBid. Overexpressed Bid-Myc or tBid-Myc co-immunoprecipitated GFP-Bfl-1 and Bfl-1ΔC (Figure 5a), as previously observed (Werner et al., 2002). However both failed to associate with endogenous Bid, despite efficient immunoprecipitation of 2×Myc-Bfl-1 and Bfl-1ΔC (Figure 5b). The lower amounts of Bfl-1 vs Bfl-1ΔC in input lanes is consistent with Bfl-1’s constitutive turnover that involves its C terminus (Kucharczak et al., 2005; Herold et al., 2006). The results were notably different when we probed Bfl-1’s association with endogenous tBid in cells treated with TNFα/CHX to promote processing of Bid to tBid (Figure 5b). Although endogenous tBid is difficult to detect due to its short half-life (Breitschopf et al., 2000), an interaction was clearly seen with 2×Myc-Bfl-1 or Bfl-1ΔC, particularly upon longer exposure. In support of a functional interaction, Bfl-1 and Bfl-1ΔC suppressed tBid-induced apoptosis, although Bfl-1ΔC was somewhat less effective than Bfl-1 (Figure 5c). This illustrates that, similar to its differential interaction with endogenous Bak but not Bax, Bfl-1 selectively associates with endogenous tBid but not Bid, and suggests that this interaction is important for suppressing the extrinsic apoptotic cascade initiated by TNFα.
The sequence divergence between Bfl-1 and other anti-apoptotic Bcl-2 family members has raised questions regarding its mode of action. We demonstrate that in the functional context of living cells where Bfl-1 inhibits apoptosis, Bfl-1 selectively associates with endogenous tBid and Bak, but not Bid or Bax. While Bfl-1 antagonized Bax activation in response to extrinsic death signaling by TNFα, it failed to suppress Bax-mediated cell death in response to intrinsic death signaling that is independent of tBid in Bak−/− cells, but effectively blocked STS-induced apoptosis in Bax−/− cells that express endogenous Bak. This indicates that Bfl-1’s constitutive interaction with Bak is important for suppressing Bak-mediated apoptosis, whereas its association with tBid is most likely responsible for blocking Bax activation in response to death receptor stimulation.
Conversion of Bid to tBid amplifies death receptor signaling and leads to activation of Bax and Bak. Some studies suggested that tBid acts directly upon mitochondria to promote cyto C release in absence of Bax and/or Bak (Epand et al., 2002; Grinberg et al., 2002). However knockout studies showed that tBid most likely functions by activating Bax and Bak, since cells lacking both factors are resistant to apoptosis (Wei et al., 2001; Degenhardt et al., 2002). Our data support the idea that Bfl-1’s association with tBid is most likely responsible for suppressing Bax activation, as Bfl-1 failed to bind endogenous Bax in TNF-treated cells. This agrees with reports showing that although Bcl-2 and Bcl-xL can directly associate with Bax, their interaction with tBid is most important for their protective function in the TNF cascade (Cheng et al., 2001; Werner et al., 2002; Yi et al., 2003).
Importantly, like Mcl-1, Bfl-1 interacts with and suppresses Bak-mediated apoptosis and this interaction does not depend upon Bak activation (Cuconati et al., 2003; Willis et al., 2005). Recent work suggests that they also show strikingly similar affinities for BH3 peptides from certain BH3-only factors that differed from those of Bcl-2, Bcl-xL and Bcl-w (Chen et al., 2005), although some found otherwise (Certo et al., 2006). Our finding that Bfl-1 physically and functionally interacts with Bak in living cells differs from others who failed to see an interaction between in vitro translated Bfl-1 and Bak in presence of mouse liver mitochondria (Werner et al., 2002). Since recombinant tBid was included in these assays and association of tBid with Mcl-1 can displace Bak from Mcl-1–Bak complexes (Clohessy et al., 2006), tBid might likewise dissociate Bfl-1 from Bak, thereby precluding detection of Bfl-1–Bak complexes in these assays.
Since the hydrophobic C terminus of multidomain Bcl-2 family proteins is implicated in membrane localization and charged residues are present in Bfl-1’s C-terminus, Bfl-1 may be a peripheral membrane protein that is loosely associated with mitochondria and may depend on interaction with mitochondrial proteins for localization to this organelle. While we do not rule out the possibility that Bfl-1ΔC retains some localization at mitochondria, its altered localization is consistent with its decreased association with endogenous Bak that is constitutively found at mitochondria, and its reduced ability to efficiently suppress Bak-dependent apoptosis. This may explain why recent work, using a Bfl-1ΔC-like form of A1 lacking 20 C-terminal amino acids, failed to detect an interaction with Bak (Chen et al., 2005; Willis et al., 2005). The weakened association of Bfl-1ΔC with Bak may also clarify why deletion of Bfl-1’s C terminus reduced its protective activity toward p53-induced apoptosis, in which Bak plays an important role (D’Sa-Eipper and Chinnadurai, 1998). Together, these findings indicate that efficient Bfl-1/Bak interaction may be an important means for Bfl-1 to antagonize intrinsic death signaling that relies on Bak and does not signal through tBid.
Overall our data indicate that Bfl-1’s mode of action is similar to Mcl-1, which sequesters Bak in inactive complexes and interacts with tBid to suppress death receptor-mediated apoptosis (Figure 5d; Oltvai et al., 1993; Perez and White, 2000; Cuconati and White, 2002; Cuconati et al., 2002, 2003; Clohessy et al., 2006). Although Bfl-1 and Mcl-1 utilize similar means to suppress apoptosis, they are likely to act in a non-redundant manner in different cells and in response to different stimuli, since they are under different transcriptional control and Mcl-1 is not an NF-κB target (Michels et al., 2005). Bfl-1 is overexpressed in many human tumors, including several in which constitutive NF-κB activation is necessary for survival and chemoresistance. Bfl-1’s functional interaction with tBid and Bak may thus contribute to its role in tumorigenesis and therapy resistance. Collectively these studies provide new insights into the mechanisms by which Bfl-1 counteracts the extrinsic and intrinsic death-signaling cascades and suggest that approaches to specifically block Bfl-1’s interaction with tBid and/or Bak might improve the response of Bfl-1-expressing tumor cells to anticancer treatment. This may be important, since inhibition of Mcl-1 is necessary to sensitize some tumors to the small-molecule BH3 mimetic ABT-737 that antagonizes Bcl-xL, Bcl-2 and Bcl-w (Konopleva et al., 2006; Letai, 2006; van Delft et al., 2006). If Bfl-1 is an NF-κB-induced substitute for Mcl-1, antagonizing Bfl-1 function may be required for cancer therapy and is perhaps one of the means by which NF-κB inhibitors provoke cancer cell apoptosis.
Transfected MCF-7 cells treated with TNFα for 8 h were analysed with anti-cyto C or -BaxNT antibodies.
HeLa cells cotransfected with FLAG-Bax, Bid-Myc or tBid-Myc were supplemented with caspase inhibitor z-Val-Ala-Asp(OMe)-FMK (zVAD-fmk). Interaction with endogenous factors was analysed in CHAPS extracts from HeLa cells treated with TNFα (10 ng ml−1) plus CHX (30 µg ml−1), or CHX alone for 6 h using anti-FLAG, -Myc, -Bak or -Bax-loop antibodies and immunoblotting for Myc, GFP, BaxNT, BakNT, BaxN20 or Bid.
Immortalized BMK cell lines transfected with enhanced green fluorescent protein (EGFP), Bfl-1, Bfl-1ΔC or Mcl-1 were left untreated or treated with STS (1.5 µM) for 24 h. Survival was determined by Trypan blue exclusion and normalized to transfection efficiency. Survival of β-galactosidase-positive MCF-7 cells cotransfected with GFP-Bfl-1 or -Bfl-1ΔC, and tBid-FLAG or pcDNA3 along with pCMV-β-gal was determined.
Additional experimental details are available in the Supplementary Information.
We thank R Sundararajan, D Perez and N Gupta for discussions. This work was supported by NIH Grant CA083937, the Charlotte Geyer Foundation and the Foundation of UMDNJ. MJS was partially supported by NIH predoctoral training Grant GM08360.
Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc).