Immunoprecipitation of Mcl-1, Bcl-xL, and Bcl-2 in Bim/Bid DKO MEFs

Following incubation with BH3 peptides, an aliquot of the pellet of digitonin-permeabilized cells was dissolved in lysis buffer (10 mm HEPES, pH 7.4, 10 mm KCl, 0.1 mm EGTA, 200 mm NaCl, 1% CHAPS, or 0.5–1% Nonidet P-40) containing Complete Protease Inhibitor (Roche, 11873580001) and 1 mm PMSF. After incubation for 30 min on ice, cells were spun down at 13,000 × g for 15 min and the supernatant was collected. Antibodies (0.5–1 μg) (anti-Bcl-2; Santa Cruz Biotechnology, sc-23960, anti-Mcl-1; Rockland, 600-401-394, anti-Bcl-xL; Cell Signaling, 2762) and 20 μl of protein A/G PLUS-agarose (Santa Cruz Biotechnology) were added to this cell lysate (450 μg of protein) and incubated overnight at 4 °C. The beads were washed in lysis buffer and added directly to the loading buffer for SDS-PAGE, together with the depleted lysate. Mcl-1, Bcl-2, and Bak were detected by immunoblotting.

Transient Overexpression in Bim/Bid DKO MEFs using Retrovirus

293T cells were transfected with pMSCV-IRES-GFP with/without the gene of interest together with an equal amount of AdVAntage^TM Vector (Promega, Part TB207) and ectopic helper virus DNA using Lipofectamine 2000 (Invitrogen, 11668-019) according to the manufacturer's instructions. The medium containing retroviral pseudotypes were collected at 16, 24, 40, and 48 h post-tansfection individually and added to growing MEFs in 6-well plates after each viral harvest in the presence of 8 μg/ml polybrene (Sigma, H9268). Separately prepared viruses containing Bad or Noxa genes were used for simultaneous overexpression of both genes. MEFs were infected with the first virus for 2–3 h and then with the second virus overnight. For both single and double gene overexpression, MEFs were harvested after 50 h post-transfection, and cell death was measured by FACS-based analysis of Annexin V staining, using the AnnexinV-Cy5 apoptosis detection kit (Biovision Inc. K103-100). Transfection efficiency was determined as FACS-based measurement of GFP expression, and was >80% in all transfectants.

Recombinant Proteins and BH3 Peptides

Recombinant full-length human Bax, N/C-Bid (cleaved Bid), Bcl-xL, Mcl-1, and cBak were produced in Escherichia coli according to the published procedures (13, 22, 25). All these proteins showed more than 90% purity, as assessed by Coomassie Blue staining. BH3 peptides were obtained from AnaSpec at >95% purity. We used human sequences for the BH3 peptides. We did not detect significant difference in activity between mouse and human Bid or Bmf in terms of cytochrome c release and liposome permeabilization.

Liposome Assays

Liposomes containing 7% cardiolipin were generated by extrusion, and the end-point assay was performed at room temperature for 2.5 h, as previously reported (13, 22). The release was normalized to Bax and N/C-Bid induced release as 100%. We adjusted the concentrations of Bax, cBak, and N/C-Bid in order to achieve ~20% release by the effector alone and ~70%, the effector plus N/C-Bid between different liposome and protein preparations. The concentrations of Bcl-xL or Mcl-1used in Fig. 4 were expected to interact with Bax or cBak (see Fig. 5A). We averaged 3–5 independent experiments for each graph, or chose representative of 4–6 independent experiments.

Open in a separate window

FIGURE 4.

BH3 ligands show direct activation when anti-apoptotic proteins are inhibited. A, in the presence of Bcl-xL, Bax was activated by the same set of BH3 peptides as in the absence of Bcl-xL, with the exception that Noxa BH3 was ineffective. B denotes the fact that this situation was addressed in the following panel, B. B, bad BH3 peptide presumably removed Bcl-xL inhibition, revealing direct activation of Bax by Noxa BH3. C, in the presence of Bcl-xL, cBak was only activated by Bim and Bmf BH3. D and E denote the fact that these situations were addressed in the following panels, D and E, respectively. D, bad BH3 interfered with Bcl-xL, revealing direct activation of cBak by Bid BH3. E, Bad BH3 inhibited Bcl-xL, uncovering direct activation of cBak by Noxa BH3. F and G, in the presence of Mcl-1, Bax and cBak were directly activated by the same peptides that directly activated them in the absence of Mcl-1, except for Puma BH3 exhibiting more potent direct activation in both cases and Noxa BH3 failing to activate cBak.

Open in a separate window

FIGURE 5.

BH3 ligands overcome effector inhibition by anti-apoptotics in liposomes. A, high levels of Bax (left panel) or cBak (right panel) led to significant permeabilization of liposomes, and this permeabilization was inhibited in a dose-dependent manner by Bcl-xL or Mcl-1. B, both Bad BH3 and its mimetic, ABT 737, were capable of interfering with the inhibition of Bax or cBak by Bcl-xL or Mcl-1.

The time course assay measured the released fluorescein-dextrans through the quenching by anti-fluorescein antibodies (Invitrogen A889) added to the external solution (26). N/C-Bid (45 nm) and BH3 peptides (25 μm) were added 2–3 min prior to the addition of Bax (120 nm). The fluorescence measurement was taken every 30 s in a 96-well plate by a microplate fluorescence reader, FL600 (Bio-Tek). The experiments were performed more than 5 times to ensure the reproducibility.

cBak Association with the Liposomes and Oligomerization in the Liposome

cBak (5 μm), N/C-Bid (190 nm), and Bcl-xL (4.6 μm) were incubated with the liposomes (~4 times more than the regular liposome assay) at room temperature for 2.5 h. At the end of the incubation, the sample was divided to into two aliquots to measure dextran release and to carry out the association assay as described previously (22). The liposome suspension in the running buffer was loaded on a Superdex 200HR(10/30)(GE) as previously described (13). Each fraction was concentrated using a microcon filter and loaded on an SDS-PAGE gel. cBak was detected by immunoblotting.

Cytochrome c Release Assay with Mouse Liver Mitochondria

Mouse liver mitochondria were prepared and the cytochrome c release assay was performed as described previously (27). Cytochrome c was detected by immunoblotting. Mouse liver mitochondria deficient for both Bax and Bak were prepared as described in Cassidy-Stone et al. (28).

Immunoblotting for Bcl-2 Family Proteins

MEFs or digitonin-permeabilized cell pellets (25 μg protein) were loaded onto 12% SDS-PAGE gels and transferred to nitrocellulose membrane for immunoblotting using antibodies against Bcl-2 family members; anti-Bcl-2 (Santa Cruz Biotechnology, sc-23960 and BD, 554087), anti-Bcl-xL (Santa Cruz Biotechnology, sc-8392), anti-Bcl-w (Calbiochem, 197209), anti-Mcl-1 (Rockland, 600-401-394), anti-Bax (Santa Cruz Biotechnology, sc-493), anti-Bak (Upstate, 06–536), anti-Bid (R & D, AF860), anti-Bmf (Lifespan, LS-C4126), and anti-Bim (Sigma, B7929 or Stressgen, AAP330). An anti-actin antibody was obtained from Sigma (ac-40). Secondary antibodies used were; anti-mouse IgG (Bio-Rad, 170-6516), anti-rabbit IgG (Bio-Rad, 170-6515), anti-rat IgG (Santa Cruz Biotechnology, sc-2006) and anti-goat IgG (Santa Cruz Biotechnology, sc-2350).

Approximately 10% of Bak Was Bound to Mcl-1 in Bim/Bid DKO MEFs

The neutralization model hypothesizes that Bax and Bak are sequestered by anti-apoptotic Bcl-2 proteins until BH3-only proteins liberate them. We immunoprecipitated Bcl-2, Bcl-xL, and Mcl-1 from non-apoptotic Bim/Bid DKO MEFs to the extent that the amount left in the cytosol was minimal. The immunoprecipitation was performed in CHAPS, a detergent known not to induce effector conformational changes associated with activation and oligomerization, although the results using NP-40 turned out identical. We found that Bax was not co-immunoprecipitated with any of the anti-apoptotic proteins and only ~10–15% Bak was found associated with Mcl-1, but not at all with Bcl-2 or Bcl-xL (Fig. 1D). It appears that most Bax and Bak molecules were not sequestered by binding to anti-apoptotic Bcl-2 proteins. These cells do not express A1 (29) and only trace amounts of Bcl-w (Fig. 1E), making it unlikely that the rest of the effectors are bound to these anti-apoptotic proteins. We conclude that the majority of Bax and Bak are not bound to anti-apoptotic Bcl-2 proteins and, yet, not inducing MOMP in healthy cells.

cBak Associates with the Membrane, Oligomerizes, and Permeabilizes the Membrane

We were also interested to know whether Bak is activated by a different or the same set of BH3-only proteins as with Bax. To investigate the activation of Bak by BH3 peptides in the liposome system, we used recombinant cBak. cBak was generated by calpain cleavage from a C-terminally truncated species, and lacks 15 amino acids at the N terminus and 21 at the C terminus (Fig. 2A) (25). cBak was soluble and monomeric in solution, became associated with the liposome membrane and formed oligomers in the membrane in the presence of cleaved Bid (Fig. 2, B and C). Furthermore, it was capable of synergizing with BH3 peptides to promote the release of cytochrome c from mouse liver mitochondria devoid of Bax and Bak. The mitochondrial permeabilization correlated with cBak oligomer formation (Fig. 2D). Cytochrome c release in the presence of BH3 peptides was also tested in mitochondria isolated from wild-type mouse livers that are known to contain Bak only. The release profile with native Bak and that with cBak were in good agreement (Fig. 2, E and D, upper panel). We conclude that cBak mimics native Bak, although ~10 times more cBak than Bax was required to produce the same level of membrane permeabilization.

Open in a separate window

FIGURE 2.

Recombinant cBak permeabilizes the liposome membrane and mitochondria. A, schematic diagrams of native Bak and cBak. Recombinant cBak lacks the N-terminal 15 amino acids and the C-terminal 25. B, cBak, N/C-Bid and Bcl-xL were incubated with fluorescein (F)-dextran-loaded liposomes, and dextran release was measured as reported previously (13). The liposomes were recovered by float-up centrifugation and Bak associated with the membranes was detected by immunoblotting. cBak association with the membrane was promoted by N/C-Bid, and this effect was blocked by Bcl-xL. Liposome permeabilization correlated with the amount of cBak in the membrane. Blots shown are representative of five experiments. C, N/C-Bid induced oligomerization of cBak in the membrane. cBak and N/C-Bid-incubated liposomes from the middle lane in Fig. 2B were solubilized in 2% CHAPS and fractionated by gel-filtration (N/C-Bid treated). As a control, total cBak that was incubated with liposomes in the absence of cleaved Bid was fractionated and found to be monomeric (Non treated). D, cBak was incubated with mouse liver mitochondria devoid of Bax and Bak. The Bim, Bid, and Bmf BH3 peptides stimulated robust cytochrome c release, and the Hrk, Puma, and Bik BH3 showed a modest release (upper panel). cBak oligomers were detected (lower panel). Note that cytochrome c release correlated well with the formation of cBak oligomers. E, BH3 peptides (25 μm) were added to mitochondria isolated from wild type mouse livers. The profiles of cytochrome c release in response to native Bak in combination with BH3 peptides was similar to those for cBak with BH3 peptides; Bim and Bid BH3 were highly potent, and Bmf and Puma BH3 also triggered moderate release consistently. We used mouse BH3 peptides for Bid and Bmf and the rest were human BH3s.

A Number of BH3 Peptides Other than Bim and Bid BH3 Directly Activate Bax and cBak

Liposome permeabilization is influenced by differences in recombinant protein and liposome preparations. To remedy this problem and obtain representative data, we selected and averaged the experiments in which we saw that the release by Bax alone was 20–30% that of Bax plus N/C-Bid (that typically gives the maximum release in our system). BH3 peptides alone did not permeabilize liposomes at 25 μm (Fig. 3A). We examined direct activation of Bax or cBak by each BH3 peptide and found that the direct activation profile of cBak in response to various BH3 peptides was similar to that for Bax; Bim BH3 exhibited the highest degree of activation, Bid, Bmf, and Noxa BH3 constituted the next highest group, then, the group of Bik, Hrk and Puma BH3, and Bad BH3 showed the lowest activation (Fig. 3, B and C). Most peptides exhibited some degrees of direct activation. Particularly, the previously unrecognized Bmf and Noxa BH3 peptides were notable. We confirmed that there was no significant false direct activation over a range of Bax-alone release by Bad BH3 (Fig. 3D). To characterize direct activation by BH3 peptides other than Bim and Bid BH3, we performed a time course assay. In this assay, as the membrane permeabilizes and the dextrans are released, anti-fluorescein antibodies in the external milieu quench the fluorescence instantaneously (26). Two representative data are shown in Fig. 3E. The permeabilization induced by the weak direct activator BH3 peptides was slower than the strong direct activators.

Open in a separate window

FIGURE 3.

Bax and cBak activation by BH3 ligands reveal low-grade direct activation. A, BH3 peptides alone do not trigger liposome permeabilization. Bax (120 nm), N/C-Bid (45 nm), or BH3 peptides (25 μm) were added to liposomes containing fluorescein (F)-dextran (10 kDa). B and C, direct activation of Bax or cBak by a panel of BH3 peptides. The potency of direct Bax activation ranked Bim BH3 as the strongest activator, followed by Bid, Bmf, and Noxa BH3 as moderate activators, Bik, Hrk, and Puma BH3 as weak activators and Bad BH3 as non-activating. D, difference in background release (Bax-alone release; black bars) does not lead to false direct activation. Bad BH3 peptide was consistently unable to directly activate Bax (gray bars) over a range (8–42%) of Bax alone-induced permeabilization. E, a time course of dextran release demonstrates low-grade direct activation by Bmf and Noxa BH3 (upper panel) and Bik, Hrk, and Puma BH3 (lower panel). Bim and Bmf BH3 peptides were included in both assays to show the relative potency between the peptides. Permeabilization induced by weak direct activators was slow, but notably higher than the Bax-alone background, which is consistent with the end-point assay. F, CD confirms that all the BH3 peptides used in this study were intrinsically disordered at 10 mm phosphate buffer of pH 7.0. The poor solubility of Bik BH3 in aqueous solutions required the use of 0.1% DMSO in the CD sample, which introduced significant noise in the data below 200 nm. Although the Bim and Noxa BH3 peptides apparently dissolved in water, they may be less soluble in the CD buffer, illustrated by a less pronounced ellipticity drop at 200 nm.

We also performed circular dichroism on our peptides to test whether the peptide's helicity determines the potency of direct activation. The spectra show that all the peptides were equally unstructured (Fig. 3F), therefore, eliminating the possibility of the helical propensity of free peptides exclusively correlating with the observed degree of direct activation.

Low-grade Direct Activation Influences Membrane Permeabilization

To confirm direct activation by BH3 peptides other than Bim and Bid, we performed the liposome assay in the presence of anti-apoptotic proteins. In our previous work, de-repression was examined in the presence of a cleaved form of Bid protein (N/C-Bid) that was ready to activate Bax. This obscured the BH3 peptides' ability to directly activate Bax. Therefore we performed the assay without N/C-Bid. BH3 peptides were consistently used at 25 μm, except for titrations. As shown in Fig. 4A, Bim, Bid, Bmf, Hrk, and Puma BH3 were capable of removing Bcl-xL inhibition as well as directly activating Bax. Hrk and Puma BH3 peptides appear to have more pronounced Bax activation in the presence of Bcl-xL (compare with Fig. 3B). Bcl-xL might somehow change the conformation of the peptides, which enhanced their activities, as discussed below in the context of Mcl-1. Noxa BH3 can activate Bax directly (Fig. 4A), but in the presence of Bcl-xL, it could not do so. We hypothesized that there was direct inhibition of Bax by Bcl-xL, which Noxa BH3 could not remove. When we added Bad BH3 to the system to inhibit Bcl-xL, now direct activation of Bax by Noxa BH3 was revealed (Fig. 4B). This data also demonstrates that neutralization is necessary, but not sufficient, and direct activation is required to achieve full membrane permeabilization. We performed the same experiment with cBak and found that Bim, Bmf and Puma BH3 could directly activate cBak, but not Bid and Noxa BH3 (Fig. 4C). We reasoned that Bid and Noxa BH3 could not remove Bcl-xL inhibition, so added Bad BH3. As shown in Fig. 4, D and E, their direct activation capabilities manifested when Bcl-xL was inhibited. Bid BH3 is known to inhibit Bcl-xL, however, we suspect that the interaction between cBak and Bcl-xL is stronger than Bid BH3 and Bcl-xL at the concentrations used. We performed the same experiments with Mcl-1 instead of Bcl-xL and the results show that Bim, Bid, Bmf, Puma, and Noxa BH3 directly activated Bax or cBak in the presence of Mcl-1 (Fig. 4, F and G). It is somewhat surprising that we did not see significant levels of permeabilization by Noxa BH3 in the presence of Mcl-1 (Fig. 4G, the last bar). On the other hand, it appears inconsistent that Puma BH3 very modestly activates Bax or cBak (Fig. 3, B and C), and yet, in the presence of Mcl-1, its direct activation was more pronounced than would have been anticipated (Fig. 4, F and G). We do not have good explanations for these discrepancies regarding Puma and Noxa BH3s, however, the presence of Mcl-1 might somehow alter the folding state of some BH3 peptides and their activities.

Direct Inhibition of Bax or cBak by Anti-apoptotic Proteins and Its Removal by BH3 Domain

Because neutralization of Bax/cBak by anti-apoptotic Bcl-xL was suspected in the experiments described above, we investigated the neutralization in the liposome system. We increased Bax or cBak to the levels at which they significantly permeabilize liposomes on their own, and found that they were directly inhibited by Bcl-xL or Mcl-1, in a dose dependent manner (Fig. 5A). cBak was more easily inhibited than full-length Bax, possibly because its truncation renders it less potent. The inhibition of Bax or cBak by the anti-apoptotic Bcl-xL was reversed by applications of Bad BH3 peptide (Fig. 5B) or a Bad-like BH3 mimetic compound, ABT-737 (30). Direct demonstration of binding between the proteins by immunoprecipitation was impossible, as the proteins bound to the beads non-specifically even in the presence of carrier proteins. We conclude that BH3-only proteins can dissociate Bax/Bak from anti-apoptotic proteins, much like they displace direct activator BH3-only proteins from the anti-apoptotic proteins (20, 21).