Report

Apoptosis Initiated When BH3 Ligands Engage Multiple Bcl-2 Homologs, Not Bax or Bak

See allHide authors and affiliations

Science  09 Feb 2007:
Vol. 315, Issue 5813, pp. 856-859
DOI: 10.1126/science.1133289

Abstract

A central issue in the regulation of apoptosis by the Bcl-2 family is whether its BH3-only members initiate apoptosis by directly binding to the essential cell-death mediators Bax and Bak, or whether they can act indirectly, by engaging their pro-survival Bcl-2–like relatives. Contrary to the direct-activation model, we show that Bax and Bak can mediate apoptosis without discernable association with the putative BH3-only activators (Bim, Bid, and Puma), even in cells with no Bim or Bid and reduced Puma. Our results indicate that BH3-only proteins induce apoptosis at least primarily by engaging the multiple pro-survival relatives guarding Bax and Bak.

Whether a cell dies in response to diverse developmental cues or cellular stresses is determined largely by interactions between three factions of the Bcl-2 protein family (1). Two factions promote apoptosis: The BH3-only proteins (including Bim, Bid, Puma, Bad, and Noxa) sense cellular damage, and Bax and Bak are critical downstream mediators of apoptosis because their combined absence abolishes most apoptotic responses (2, 3). When activated, Bax and Bak permeabilize the outer mitochondrial membrane, freeing proapoptogenic factors such as cytochrome c, which promote activation of the proteases (caspases) that mediate cellular demolition. Activation of Bax and Bak is opposed by the pro-survival faction: Bcl-2, Bcl-xL, Bcl-w, Mcl-1, and A1. These guardians are inactivated when BH3-only proteins insert their BH3 domain into a groove on the pro-survival proteins (1).

Awidely embraced model for the initiation of apoptosis (4) (Fig. 1A) proposes that a subset of BH3-only proteins termed “activators”—namely Bim and Bid (46) and, perhaps, Puma (7)—directly engage Bax or Bak. The other BH3-only proteins, termed “sensitizers” (for example, Bad and Noxa), purportedly act only by displacing the activators from the pro-survival proteins, allowing the activators to bind Bax and Bak. An alternative indirect model (8, 9) (Fig. 1A), supported here, posits that BH3-only proteins activate Bax or Bak not by binding either one but indirectly, by engaging the multiple pro-survival relatives that constrain them. In this scenario, Bim, Bid, and Puma are highly potent because they engage all pro-survival proteins, whereas the others are less so because they bind only selective subsets (8).

Fig. 1.

BH3 binding to Bax and Bak. (A) Models for activation of Bax or Bak. (B) Relative affinities of BH3 peptides for octylglucoside (OG)–treated Bax, determined in solution competition assays (8, 10). IC50, median inhibitory concentration. NB, no binding at 10 μM. (C) Coimmunoprecipitation tests in Triton X-100–containing lysates of healthy thymocytes (–) or 24 hours after treatment with ionomycin (+), a Bim-dependent stimulus (14). Bim-containing complexes, isolated using an antibody (3C5) that recognizes all Bim isoforms, were examined for Bax, Bcl-2, or Mcl-1 (upper three panels). Control, isotype-matched antibody immunoprecipitation; WCL, whole-cell lysates; IP, immunoprecipitation.

We first investigated whether BH3 peptides associate with Bax in vitro (10). The Bim and Bid BH3 peptides failed to bind native Bax (fig. S1A), and even a large excess of the Bim peptide did not promote Bax oligomerization (fig. S2). In non-ionic detergents that alter the Bax conformation (fig. S3) (11), perhaps mimicking an activation step, they did bind (fig. S1A). However, their affinities for Bax (in the micromolar range) (Fig. 1B) appeared to be lower than those (in the nanomolar range) for the pro-survival proteins (8) (fig. S1C). Thus, Bim and Bid peptides bound weakly and only after Bax had changed conformation. No other BH3 peptide, including that of Puma, bound even to activated Bax (Fig. 1B).

Coimmunoprecipitation studies extended these observations to full-length BH3-only proteins in cell lysates (fig. S4) (10). In lysates of human embryonic kidney 293T cells prepared with the detergent Triton X-100, BH3-only proteins bound to Bcl-2 or other pro-survival proteins (8) but not to endogenous Bak (fig. S4). Similarly, in lysates made with CHAPS, a detergent that maintains Bax as a soluble monomer (11), no BH3-only protein bound Bax. In Triton X-100 lysates, Bax did bind detectably to the truncated Bid (tBid) produced by caspases, but not to full-length Bid (fig. S4A), which has its BH3 domain buried (12). BimEL and BimL, the only Bim isoforms detectable in multiple cell types (fig. S5), failed to bind Bax (fig. S4B) (13), though both killed cells efficiently (8). Bax did bind the very minor BimS isoform (fig. S4B) and some unusual Bim variants confined to certain primates (fig. S6). However, no Bax was detectable in endogenous Bim protein complexes (Fig. 1C), even those from mouse thymocytes treated with ionomycin, which kills via a Bim-dependent pathway (14). In contrast, the binding of Bim to endogenous pro-survival proteins was evident, and this binding increased after cellular insults (Fig. 1C) (15, 16). Thus, Bim did not form physiologically relevant complexes with Bax or Bak.

To test whether cell killing by tBid or BimS requires their binding to Bax, we used a mutation, glycine to alanine [Gly94 → Ala94 (G94A)], within their BH3 domains that abrogated binding to Bax but not to the pro-survival proteins (Fig. 2 and figs. S7 to S9). Despite its inability to bind Bax, tBid G94A killed mouse embryo fibroblasts (MEFs) and mouse myeloid cells as effectively as the wild-type protein, in both short-term viability (figs. S7 and S9) and clonogenic (Fig. 2B) assays. Our studies with MEFs lacking Bax or Bak (fig. S9) showed that either Bak or Bax could mediate the killing. Furthermore, the Bid G94A BH3 peptide, like the wild-type peptide, promoted cytochrome c release from mitochondria in vitro (Fig. 2C) (10). The equivalent BimS mutant gave similar results (figs. S8 and S9).

Fig. 2.

Binding to Bax is not required for killing by tBid. (A) In Triton X-100–containing 293T lysates, the binding of overexpressed hemagglutinin (HA)–tagged tBid [wild type (WT)] or its G94 mutants to endogenous Bax, or of overexpressed pro-survival proteins, was tested by coimmunoprecipitation with an antibody to HA and immunoblotting with the indicated antibodies. Lower panels, protein expression in WCLs. Control, immunoprecipitation from untransfected cells. (B) Colony formation 6 days after plating of equivalent numbers of retrovirally infected MEFs (10), normalized to the number of colonies formed with the parental retrovirus. Data represent means ± SD from three independent experiments. (C) Cytochrome c release elicited by BH3 peptides. Permeabilized MEFs incubated with each BH3 peptide (10 μM) were separated into soluble (s) and sedimented (p, pellet) fractions that were probed with antibodies to cytochrome c, Bak (membrane marker), or HSP70 (cytosolic marker).

Thus, tBid and Bim can induce apoptosis without binding Bax or Bak. We inferred that these proteins and their G-to-A mutants caused cell death by binding to and neutralizing diverse pro-survival proteins (8). Indeed, the corresponding glycine-to-glutamate mutants, which do not bind most (or any) pro-survival family members, did not kill (Fig. 2 and figs. S7 to S9).

If Bim and Bid [the major candidate activator BH3-only proteins (46)] were essential for apoptosis, as the direct-activation model implies, mice lacking both proteins (10) should exhibit developmental abnormalities, like Bax-Bak–deficient animals, which usually die as neonates (2). Instead, mice lacking both Bim and Bid were born at the expected Mendelian frequency (fig. S10), appeared normal (e.g., no webbing between the toes), and remained healthy for over 12 months. Moreover, comparison of the sensitivity to DNA damage of immortalized MEFs from Bim-Bid double-knockout (DKO), wild-type, or Bax-Bak DKO embryos (fig. S11) revealed that, whereas Bax-Bak deficiency conferred resistance to ultraviolet (UV) irradiation and etoposide, the absence of both Bim and Bid neither abrogated apoptosis nor prevented Bax activation (Fig. 3, A and B).

Fig. 3.

Apoptosis in the absence of both Bim and Bid. (A) Viability of cells lacking Bim and Bid after UV irradiation or etoposide treatment. (B) Bax activation without Bim-Bid. Immunoprecipitation with monoclonal antibody 6A7, which recognizes only activated Bax (11), from WT or bim–/–bid–/– MEFs treated for 24 hours with etoposide. (C and D) Killing by sensitizer BH3-only proteins. (C) BH3 specificity. (D) Colony formation after Bad retroviral infection of MEFs stably expressing Noxa (to target Mcl-1) or the inert Noxa mutant 3E (9). (E) Cytochrome c release without Bim or Bid. Digitonin-permeabilized MEFs expressing Noxa or Noxa 3E were incubated with 10 μM Bad BH3 peptide before separation into soluble (s) and pellet (p) fractions. Data in (A) and (D) represent means ± SD from three independent experiments.

Because the direct-activation model (Fig. 1A) proposes that only an activator BH3-only protein can kill cells, we tested whether the sensitizers Bad and Noxa (4) could kill cells lacking both Bim and Bid. Although overexpression of Bad and Noxa together killed wild-type MEFs (8) and neither bound Bax or Bak (Fig. 1B and fig. S4C), we could not exclude the possibility that they killed by displacing endogenous Bim or tBid. However, even in the absence of both Bim and Bid, the Noxa-Bad combination (Fig. 3, C to E), which neutralizes all four pro-survival proteins expressed in MEFs (Bcl-2, Bcl-xL, Bcl-w, and Mcl-1), still induced apoptosis and cytochrome c release in vitro, as did the Bad-like BH3 mimetic ABT-737 (17) plus Noxa (fig. S12). Mouse liver mitochondria yielded similar results (fig. S13).

Although one study reported that Puma bound Bax (7), its putative activator role is disputed (5, 6). We and others (18) have detected no association of Puma with Bax or Bak (Fig. 1B and fig. S4C). Nevertheless, to assess whether Puma has any role as an activator, we decreased Puma expression with RNA interference in bim–/–bid–/– MEFs (fig. S14). As expected (19), its reduction impaired etoposide-induced death. Nevertheless, in cells retaining only a small amount of Puma and no Bim or Bid, the Noxa–ABT-737 combination still efficiently provoked apoptosis and cytochrome c release (fig. S14).

We conclude that BH3-only proteins can initiate apoptosis without binding Bax or Bak, although the cooperativity with Bax of tBid (or of a stapled Bid BH3 peptide) in lysing protein-free liposomes (5, 20, 21) may indicate that direct activation contributes under some circumstances. We (8, 9) and others (22) have suggested that all BH3-only proteins function by engaging their pro-survival relatives and overcoming their block to Bax or Bak activation (Fig. 1A). Mcl-1 and Bcl-xL (but not Bcl-2) can constrain Bak by sequestering it, because enforced coexpression of Bad (to target Bcl-xL) and Noxa (to target Mcl-1) elicited Bak-dependent apoptosis without any additional stimulus (9).

Which of the multiple pro-survival proteins that can bind Bax (fig. S15A) can functionally restrain it? Mcl-1 must, because neutralizing Mcl-1 by enforced Noxa expression rendered MEFs containing only Bax (Bak KO cells) sensitive to the Bad BH3 mimetic ABT-737 (Fig. 4A), which inactivates Bcl-2, Bcl-xL, and Bcl-w (17). To identify other pro-survival antagonists of Bax (10), we used Noxa mutant m3, which binds Mcl-1, Bcl-xL, and Bcl-w, but not Bcl-2 (8) (Fig. 4B). It kills cells expressing only Bak (9), but not Bax-expressing ones (i.e., Bak KO cells) (Fig. 4B). Because Noxa m3 spares Bcl-2, Bcl-2 is implicated as another check on Bax. We tested this hypothesis using a functional Bax mutant [Lys64 → Ala64 (K64A)] (fig. S15, C and D) that loses Bcl-2 binding but still binds the other pro-survival proteins (fig. S15B). The ability of Noxa m3 to kill MEFs that express K64A but not wild-type Bax (Fig. 4C) confirms that Bcl-2 can prevent Bax activation. Hence, Bax is held in check by Mcl-1, Bcl-2, and either Bcl-xL or Bcl-w, or by all four (Fig. 4D). Genetically, both Bcl-2 and Bcl-xL can constrain Bax, because its concomitant loss precludes the lymphoid hypoplasia (23) or neuronal apoptosis (24) caused by their respective deficiencies. Thus, whereas only certain pro-survival proteins keep Bak in check (9), all of these proteins probably inhibit Bax (Fig. 4D). Presumably, they bind a “primed” conformer of Bak or Bax—a form that may normally be present at low levels or may be formed early in apoptotic signaling.

Fig. 4.

Control of Bax by multiple pro-survival relatives. (A) Colony formation by WT or Bax-expressing (bak–/–) MEFs stably expressing Noxa (or the inert Noxa 3E) (9) and treated with ABT-737 (Bad BH3 mimetic) to inactivate diverse pro-survival proteins. (B) Short-term viability or colony formation for bax–/– or bak–/– MEFs infected with Noxa m3. The model depicts Bak (9) or Bax activation upon neutralization of pro-survival proteins by Noxa m3. (C) Short-term or clonogenic assays after Noxa m3 infection of bax–/–bak–/– MEFs reconstituted with Bax or the K64A mutant. (D) Regulation of Bak or Bax by multiple pro-survival relatives. Until inhibited by BH3 ligands, the indicated pro-survival proteins can sequester primed Bak or Bax (possibly a form with its BH3 exposed) on intracellular membranes. Data in (A) to (C) represent means ± SD from three independent experiments.

Our findings emphasize that the central role of the mammalian pro-survival Bcl-2 proteins, like their virally encoded orthologs (25), is to inhibit Bax and Bak (Fig. 4D) (8, 9). Thus, BH3-only proteins trigger apoptosis indirectly through Bax or Bak by neutralizing all of the relevant pro-survival proteins, allowing the activation of Bax and Bak to proceed.

Supporting Online Material

www.sciencemag.org/cgi/content/full/315/5813/856/DC1

Materials and Methods

Figs. S1 to S15

References

References and Notes

View Abstract

Navigate This Article