Eur. J. Biochem. 268, 48±55 (2001) q FEBS 2001 Leakage and aggregation of phospholipid vesicles induced by the BH3-only Bcl-2 family member, BID Dayong Zhai, Qi Miao, Xiaofeng Xin and Fuyu Yang National Laboratory of Biomacromolecules, Chinese Academy of Sciences, Beijing, China BID is a BH3 domain-only member of the Bcl-2 family that acts as an apoptotic agonist in programmed cell death. After cleavage by caspase-8, the N-terminal of BID (N-BID) stays in the cytosol while the C-terminal of BID (C-BID) translocates to mitochondria, leading to cytochrome c release in vivo and in vitro. We have previously reported that BID or truncated BID (tBID) can induce the release of entrapped trypsin and cytochrome c from large unilamellar vesicles (LUVs). Further studies have been performed and are presented here; the results demonstrate that C-BID, like BID and tBID, induces vesicle leakage, whereas N-BID or the BID mutants BID (D59A) and BID (G94E) fail to have any significant effects. The affinity of the above-mentioned proteins for soybean phospholipid LUVs (SLUVs) decreased in an order similar to their leakage-inducing capability: tBID . BID . BID (D59A), while N-BID and BID (G94E) were unable to bind to the vesicles at all. BID-induced leakage was dependent on the lipid composition of vesicles. Acidic phospholipid (e.g. phosphatidic acid or phosphatidylglycerol) was necessary for BID-induced leakage while the presence of phosphatidylethano- lamine or cholesterol reduced the leakage. It was also found C-BID is better able to penetrate the soybean phospholipid monolayer than BID or tBID. A further finding was that tBID, but not full-length BID, could stimulate the aggregation of SLUVs. Finally, Bcl-xL, an apoptotic antagonist in programmed cell death, can prevent the aggregation of LUVs induced by tBID, but not the release of entrapped trypsin. It is postulated that two separate domains of tBID are responsible for inducing leakage and aggregation of phospholipid vesicles. Keywords: aggregation; apoptosis; BID; large unilamellar vesicles; leakage. Apoptosis is an evolutionarily conserved process critical in and BAD, can lead to mitochondrial damage. The ratio of these various biological events, such as embryonic development, two groups determines the fate of mitochondria and the whole maintenance of tissue homeostasis, removal of noninstructed, cell [4,5]. The sequences of Bcl-2 family proteins share several misinstructed and damaged cells, and immunological defense conserved motifs designated BH1±4 (Bcl-2 homology) . Various stimuli, including developmental and environ- domains . Among them, the BH3 domain from pro-apoptotic mental ones, deliver complex signals to promote apoptosis or members of the Bcl-2 family is essential for the `killer' function survival. A large number of pro-apoptotic and anti-apoptotic . NMR analysis of Bcl-xL ±BAK BH3 peptide complex has molecules have been identified as playing principal roles in revealed both hydrophobic and electrostatic interactions apoptosis, in which members of the Bcl-2 family act as between the Bcl-xL pocket and a BH3 amphipathic a-helical regulators . peptide from BAK . Deletions in BAK and an extensive Mitochondria, which were once thought simply to generate mutation analysis of BAK suggest that the BH3 domain serves energy for a cell, have been implicated as important sensors and as a minimal `death domain' critical for both dimerization and amplifiers in intracellular death signaling pathways by releas- killing . ing proteins such as cytochrome c and apoptosis-inducing The three-dimensional structure of Bcl-xL contains a bundle factor . Bcl-2 family proteins mainly located in the of seven a helices with two central, predominantly hydro- mitochondrial outer membrane constitute a pivotal checkpoint phobic, helices forming the core of the molecule while BID also within the common part of the apoptotic pathway. The anti- has two central hydrophobic helices surrounded by six apoptotic members of Bcl-2 family, such as Bcl-2 and Bcl-xL, amphipathic helices [9±11]. Both of these resemble pore- play important roles as protectors of mitochondrial integrity forming bacterial toxins although their homologous sequence is while various pro-apoptotic members, such as BAX, BAK, BID limited, suggesting they may have channel-forming potential. In fact, BAX, Bcl-xL and Bcl-2 have already been shown to Correspondence to F. Yang, National Laboratory of Biomacromolecules, have channel activity in artificial lipid membranes [12±14], and Institute of Biophysics, Academia Sinica, 15 Datun Road, Beijing, 100101, this has also been recently reported for BID . The ability of China. Fax: 1 86 10 64872026, Tel.: 1 86 10 64888574, these proteins to form ion channels has fostered the idea that E-mail: firstname.lastname@example.org they may open pores or produce breaks in the mitochondrial Abbreviations: DOPA, dioleoylglycerophosphatidic acid; DOPC, outer membranes, allowing exit of cytochrome c . More- dioleoylglycerophosphocholine; DOPE, over, BAX and BAK have been shown to stimulate the opening dioleoylglycerophosphoethanolamine; DOPG, of the voltage-dependent anion channel, a mitochondrial dioleoylglycerophosphoglycerol; LUVs, large unilamellar vesicles; SLUVs, channel through which cytochrome c permeates . soybean phospholipid LUVs; tBID, truncated BID. BID is a BH3 domain-only protein that lacks transmembrane (Received 24 July 2000, revised 16 October 2000, accepted domains and is predominantly localized in the cytosol . It 20 October 2000) demonstrates unique and important properties after cleavage by q FEBS 2001 BID-inducing leakage and aggregation of vesicles (Eur. J. Biochem. 268) 49 caspase-8; the 15-kDa BH3-containing C-terminal translocates containing 10 mm KCl, 2 mm MgCl2, 1 mm EDTA, 1 mm from cytosol to the mitochondrial membrane and leads to EGTA, 1 mm dithiothreitol, 1 mm phenylmethanesulfonyl cytochrome c release, while N-BID remains localized in the fluoride (buffer A). After re-equilibration with buffer A, the cytosol in vivo [19,20]. The data from the structure of BID and pH was raised to 12 to obtain C-BID. the yeast two hybrid experiments suggest that N-BID has an inhibitory action associated with the C-terminal BH3 domain Liposome preparation and detection of liposome leakage [10,11,21], which may represent an important means of regulating the activity of the whole BID molecule. LUVs were prepared as in Rietveld et al. . Briefly, 2.5 mL It is still unclear how BID leads to cytochrome c release diethyl ether and 0.5 mL Pipes buffer (10 mm Pipes, pH 7.0, from mitochondria. Some authors speculate that BID may cause 50 mm NaCl, 0.2 mm EDTA) containing trypsin (2 mg´mL21) cytochrome c release via formation of ion channels or pores were added to a dry lipid film of 5 mm phospholipid. After . Others, however, argue that BID may directly perturb the sonication for 20 min at 4 8C with a bath sonicator, and membrane integrity by interacting with membrane lipids . evaporation of ether under reduced pressure, the vesicles were Another putative function of BID is that tBID can trigger the dialyzed overnight at 4 8C in Pipes buffer. The vesicles clustering of mitochondria in cells during the process of formed were then fractionated by centrifugation in a apoptosis . Li et al. suggest that two independent pathways Beckman Optima TLX table-top ultracentrifuge (TLA-100.3 for the BID-mediated destruction of mitochondria in cell death rotor) at low speed (15 min, 5564 g, 4 8C) to discard the may operate, based on analysis of the structure of BID . multilamellar vesicles, and high speed (30 min, 42 000 g, 4 8C) The relative simplicity of the model system based on pure to collect the LUVs. The LUVs were washed at least three lipid bilayers invites its use in the study of the molecular times to remove the nonenclosed trypsin. The phospholipid mechanisms of membrane±protein interactions. We have concentration was determined by perchloric acid destruction. previously reported that BID can induce trypsin as well as Various concentrations of the test proteins were added to cytochrome c release from the internal medium of LUVs . 100 mL volumes of the LUV suspension, and pre-incubated at Here, we used this model system to study the molecular 30 8C for 15 min. Ph-CO-Arg-OEt was introduced as substrate mechanism underlying the actions of BID and other Bcl-2 to measure the degradation rate by the released trypsin at proteins. Data show that in contrast to C-BID, the mutants of 253 nm by using a Shimadzu UV-2101PC spectrophotometer. BID, BID (D59A), BID (G94E), and N-BID have almost no leakage-inducing ability. Moreover, tBID rather than BID can BID binding to LUVs lead to the aggregation of LUVs. The leakage of vesicles by LUVs were prepared as above without entrapped trypsin. BID, BID and the aggregation-inducing effect of tBID may be due BID (D59A), BID (G94E), or N-BID (0.1 mg´mL21) was to the different nature of the two hydrophobic domains, incubated at different lipid:protein molar ratios in 100 mL Pipes mimicking the functions of BID in vivo. buffer for 1 h at 30 8C. The proteins binding to the LUVs were separated from the free parts by centrifugation at 96 000 g in a M AT E R I A L S A N D M E T H O D S Beckman Optima TLX table-top ultracentrifuge (TLA-100.3 rotor) for 30 min at 30 8C. The protein concentration in the Materials supernatant was determined by the Bradford method . Human BID, BID (D59A), BID (G94E) and caspase-8 expression plasmids were kindly provided by X. Wang Penetrating of the phospholipid monolayer by BID (Department of Biochemistry, University of Texas South- A film balance, type Han-2000, designed and made in our Western Medical Center, Dallas, TX, USA). The Bcl-xL-GST laboratory was used to study the membrane penetrating ability expression plasmid was provided by J. Yuan (Department of of proteins. Briefly, 3 mL of Pipes buffer as a subphase was Cell Biology, Harvard Medical School, Boston, MA, USA). added into the mini-trough, which has been described The BAK-GST expression plasmid was obtained from Y. previously . The monomolecular lipid layer was spread to Tsujinoto (Biomedical Research Center, Osaka University, give the desired initial surface pressure by dropping aliquots of Japan). Trypsin, proteinase K, cholesterol, dioleoylglycero- lipid dissolved in chloroform on the aqueous surface. The phosphatidic acid (DOPA), dioleoylglycerophosphoglycerol surface pressure of the monolayer was measured by the (DOPG), dioleoylglycerophosphoethanolamine (DOPE) and Wilhelmy plate method using plates cut from filter paper and dioleoylglycerophosphocholine (DOPC) were obtained from rinsed with methanol prior to use. After the initial surface Sigma. Ph-CO-Arg-OEt was from the Shanghai Institute of pressure had stabilized to a plateau value, the desired Biochemistry (Chinese Academy of Sciences, Shanghai, China). amount of protein was injected to the mixing chamber (which contained a magnetic stir bar) through a 0.7-cm2 Expression and purification of recombinant proteins hole in the edge. The samples were then rapidly mixed with bulk solution, and radically diffused to the upper monolayer- Human BID, BID (D59A), BID (G94E), BAK, Bcl-xL and spreading disk. All measurements were performed at room caspase-8 were expressed recombinantly as has been described temperature. Usually, an increase of the surface pressure Dp previously [19,20]. The plasmids for expression of the proteins is measured as a function of the initial surface pressure p. were transformed into bacteria BL21 (DE3) cells; proteins were A plot of Dp versus p yields a straight line with negative purified from the cell lysate using nickel affinity (Qiagen) or slope that intersects the abscissa at the value named as the GST affinity chromatography. A Bio-scale Q5 column (Bio- limiting surface pressure. Rad) was used for further purification. Truncated BID was obtained by adding 1 : 50 (vol/vol) caspase-8, and incubating CD measurements overnight at 4 8C. After dilution, the sample was loaded on to a Bio-scale S5 column (Bio-Rad). Elution was performed with a CD measurements were carried out on a JASCO J-720 linear gradient of 0 to 100 mm NaCl in 20 mm Hepes, pH 7.4, instrument. Measurements were taken using a 1-mm path 50 D. Zhai et al. (Eur. J. Biochem. 268) q FEBS 2001 Fig. 1. SDS/PAGE analysis of purified BID, and its mutants and fragments. Purified BID, BID (G94E), BID (D59A), tBID, C-BID and N-BID were analyzed by SDS/PAGE in 15% gel and stained with Coomassie blue. Molecular mass markers are indicated. length. All spectra were recorded in 1.0-nm wavelength increments with a 1-s time constant and a full-scale sensitivity of 20 millidegrees. Each spectrum was the average of six scans corrected for background solvent effects by subtraction of the appropriate buffer blank. BID, BID (D59A), BID (G94E), and N-BID were diluted to a concentration of 20 mm in NaCl/Pi buffer; C-BID was diluted in 10 mm Tris buffer, pH 12. The lipid concentration was 1 mm. Spectra were scanned in the far UV from 250 to 190 nm. Fig. 2. Effect of BID and lipid concentration on the leakage efficiency of LUVs. (A) O Shows the percentage of leakage after 15 min as a function of lipid concentration. The amount of lipid and BID was varied, a constant mole:mole ratio of 2000:1 being kept. X Shows the percentage of leakage Turbidity measurement after 15 min as a function of lipid concentration when BID was held The aggregation of SLUVs after the proteins had been constant at 0.1 mm. The lipid concentration was varied to obtain the desired introduced was monitored spectrophotometrically at 600 nm lipid:BID ratio. (B) Percentage of leakage after 15 min as a function of BID in a cuvette with 1-cm path length using a Shimadzu UV-2101 concentration. The lipid concentration was kept constant at 300 mm while PC spectrophotometer. Aliquots of BID, tBID or tBID (G94E) the BID concentration was varied to obtain the desired ratio. In all cases, the were added successively to the suspension of SLUVs (1 mm extent of trypsin leakage was calculated as the slope of a fitted line drawn to lipid concentration) in Pipes buffer. Truncated BID or tBID the individual leakage curves. (G94E) pre-incubated with Bcl-xL at 4 8C for 10 min (1:1, mol:mol) was introduced and aggregation detected as above. 95% as determined by SDS/PAGE analysis with Coomassie The molar ratios of lipid to protein ranged from 2000:1 to blue staining (Fig. 1.). 100:1. Comparison of trypsin release from LUVs induced by BID, R E S U LT S C-BID, N-BID and its mutants Trypsin-containing LUVs have previously been used to monitor Purification of BID, and its fragments and mutants protein translocation across the model membranes . Here Recombinant BID, BID (D59A), BID (G94E), BAK, Bcl-xL, this system was used to assay the BID-induced leakage of and caspase-8 were purified in accordance with previously contents from the internal medium of the liposomes. Leakage is published methods [19,20]. Truncated BID was prepared as the studied as a function of a constant lipid:BID ratio, BID product of BID cleaved by caspase-8 following overnight concentration (lipid concentration kept constant), and lipid incubation at 4 8C; N-BID was still tightly associated with concentration (BID concentration kept constant). At a constant C-BID in the buffer. After dilution, the tBID sample was loaded lipid:BID ratio of 2000:1, the extent of leakage after 15 min to the Bio-scale S5 column. The N-BID was easily washed out reduced as the lipid concentrations decreased (Fig. 2A). This is with 0.1 m NaCl, while C-BID could only be eluted when the expected on the basis of the partitioning of the proteins between pH value was above 12. The purity of the proteins was above the bilayer and the aqueous phase; as the lipid concentration is q FEBS 2001 BID-inducing leakage and aggregation of vesicles (Eur. J. Biochem. 268) 51 lowered, the fraction of the peptide that becomes vesicle- associated decreases. When BID concentration was held constant, increasing the lipid concentration led to a decrease in the extent of leakage (Fig. 2B). As the number of vesicles in suspension grows with increasing lipid concentration, the results are in agreement with the pore-forming hypothesis. In other words, a defined number of peptides is required for the formation of a channel or pore in the bilayer. These data also indicate that once in bilayer, BID does not rapidly redistribute between different vesicles. Finally, at constant lipid, increasing BID concentration induced a greater extent of leakage until a maximum value was reached (Fig. 2C). This result is consistent with a monomer±multimer assembly pore-forming process being involved in the mechanism of leakage. In all cases, the lipid:BID ratios were determined from the amount of vesicles and BID initially added to the cuvette. The effective ratio depends on the lipid:water partition coefficient of BID. When tBID is added to trypsin-encapsulated LUVs, it triggered release more efficiently than did the full-length BID. To investigate how BID destabilizes the LUVs, BID mutants and its fragments were examined. Both BID (D59A), a mutant that cannot be cleaved by caspase-8, and BID (G94E), which contains a point mutation in the BH3 domain, were unable to induce cytochrome c release from mitochondria. Strikingly, BID (D59A) and BID (G94E) were shown to lose the ability to induce trypsin release from LUVs at the same concentrations as BID did. A partial leakage of trypsin could only be observed when excessive amounts of these mutants were added (Fig. 3A). After cleavage by caspase-8, C-BID translocates to mito- chondria to induce the release of cytochrome c, whereas N-BID remains in the cytosol [19,20]. It is difficult to separate C-BID from N-BID in the experiments carried out in vitro, but a portion of C-BID could be precipitated by adjusting the pH of Fig. 3. BID-induced release of entrapped trypsin from LUVs compris- the medium to 7.0. The partially aggregated C-BID could ing 20% DOPG and 80% 1,2-dioleoylphosphatidylcholine. LUVs trigger the release of entrapped trypsin from the liposomes. In (300 mm lipid) were incubated for 15 min at 30 8C in the presence of test contrast, N-BID was incapable of inducing leakage even at protein in a 100-mL reaction mixture. After incubation, the substrate higher concentrations (Fig. 3B). Ph-CO-Arg-OEt was introduced to the supernatant to measure degradation BAK, caspase-8 alone or BID proteolyzed with proteinase K by the released trypsin at 253 nm. The same amount of buffer was used as a were used as controls. None of these showed leakage-inducing control. (A) BID (BID:lipid, 1:3000), BID (G94E) (protein:lipid, 1:100), activity. To investigate further whether BID is able to induce the and BID (D59A) (protein:lipid, 1:100). (B) BID (protein:lipid, 1:4000), release of other molecules, fluorescein sulfonate (FS) or C-BID (protein:lipid, 1:4000), and N-BID (protein:lipid, 1:50). cytochrome c were entrapped in LUVs. A similar releasing effect could also be observed following BID treatment (data not shown). BID-induced leakage of vesicles with different lipid compositions To test whether the charge of the membrane constitutes an important factor in destabilizing liposomes, the effect of BID or tBID on leakage of LUVs composed of DOPA and 1,2-dioleoylphosphatidylcholine, or DOPG and 1,2-dioleoyl- phosphatidylcholine, at ratios of 2 : 8, 3 : 7, 4 : 6, and 5 : 5 (mol/mol), were studied. The results indicated that none of the lipid matrices tested changed the ability of the proteins to induce leakage (data not shown). This confirms that membranes containing varying amounts of negatively charged phospho- lipids allow the proteins to act similarly. Regarding Fig. 4. Effect of lipid composition on the efficiency of leakage induced 1,2-dioleoylphosphatidylcholine-only LUVs entrapped with by BID or tBID. Curves are shown for LUV-entrapped trypsin with lipid trypsin, the efficiency of induction decreased drastically composition: 20% DOPG, 80% DOPC (X); 20% DOPG, 10% DOPE, 70% (Fig. 4). It may deduced that a negative surface potential of a DOPC (W); 20% DOPG, 30% DOPE, 50% DOPC(K); 20% DOPG, 30% lipid bilayer is necessary for BID or tBID to bind to membranes cholesterol, 50% DOPC(L); and 100% DOPC(A). during the initial step of the interaction, albeit that the motif 52 D. Zhai et al. (Eur. J. Biochem. 268) q FEBS 2001 Fig. 6. Comparison of the initial surface pressure and the surface pressure increase of phospholipid monolayer following injection of different BID proteins. Test proteins (0.5 mg´mL21) were injected Fig. 5. Binding of different BID to LUVs. BID preparations were underneath a monolayer of asolectin at different initial pressures. The incubated with LUVs for 1 h at 30 8C. Vesicles were collected by limiting surface pressure of BID, tBID and C-BID were 36.4, 42 and 44 centrifugation, and protein associated with the liposome fraction was mN/m, respectively. (A) Kinetic curves showing the change in surface detected by the Bradford method. The continuous line represents the best fit pressure. (B) p-Dp plots of surface pressure change. of the data. (A) BID, tBID, BID (D59A), BID (G94E) and N-BID were incubated with SLUVs. (B) BID and tBID were incubated with LUVs comprising either 20% DOPA and 80% DOPC, or 20% DOPA, 30% DOPE Comparison of the LUV binding of BID, and its fragments and and 50% DOPC. mutants To study whether the discrepancy in the competency of these proteins to induce leakage was attributed to the difference in directing leakage may be insensitive to the charge density on their binding activity, the affinities of the BID proteins to the the membrane surface. same LUVs were determined and compared. The SLUVs were DOPE, the second most abundant phospholipid of mitochon- incubated with BID, BID (D59A), BID (G94E), tBID, and dria, or cholesterol was introduced into vesicles to investigate N-BID. After centrifugation, free proteins in the supernatant whether lipid composition would alter the interaction of BID were analyzed. BID proteins associated with SLUVs were with the vesicles. When the percentage of DOPE was about plotted as a function of the lipid concentration (Fig. 5A); it can 10%, the efficiency of BID induction was not obvious. Adding be observed that tBID has the highest affinity for SLUVs, BID 30% DOPE to bilayers containing 20% DOPG and 50% DOPC (D59A) has a lower affinity for the membrane than does the resulted in a steep decrease in the leakage efficiency (Fig. 4), wild type BID, and BID (G94E) and N-BID hardly bind to indicated by an increase in the protein:lipid ratio needed to get SLUVs. the same final leakage (about ten-fold more protein was Binding of BID to vesicles comprising: 20% DOPG and 80% required). Similar results were obtained in the case of SLUVS 1,2-dioleoylphosphatidylcholine; 20% DOPG, 30% DOPE and (data not shown). Vesicles comprising 30% cholesterol, 20% 50% 1,2-dioleoylphosphatidylcholine; and DOPC only, was DOPG and 50% DOPC also caused a clear drop in the also compared. As shown in Fig. 5B, BID had a higher affinity induction efficiency of BID (Fig. 4). These vesicle leakage for DOPG/DOPC vesicles than for DOPG/DOPE/DOPC results resemble those obtained using the pore-forming peptide vesicles, which was consistent with the disparity in the synthetic GALA, which is similarly affected by DOPE and leakage-inducing activities with the two vesicle types. How- cholesterol [27,28]. ever, the discrepancy in binding affinity of tBID to these q FEBS 2001 BID-inducing leakage and aggregation of vesicles (Eur. J. Biochem. 268) 53 increase when BID was pretreated with proteinase K, indicating that the rise in surface pressure reflected the association of BID with the monolayer. When using the BID protein-induced increase in surface pressure as a function of the initial surface pressure, an inverse linear relationship appears (Fig. 6B). The lipid-packing density of the liposomes used was equivalent to a monolayer surface pressure of 32±35 mN´m21 . Our results indicate that BID could, conceivably, penetrate monolayers of soybean phospholipid. That the limiting surface pressure for C-BID, 44 mN´m21, was much higher than that for BID, 36 mN´m21, or tBID, 41 mN´m21, (Fig. 6B) implies that BID exposes more hydrophobic residues after cleavage with caspase-8. Secondary structural changes in BID on interaction with phospholipid The secondary structures of BID, BID (D59A), BID (G94E), tBID, N-BID and C-BID were determined and compared using CD spectra. The only minor difference was found in the shape or amplitude of the far-UV spectrum among BID, BID (D59A), BID (G94E), tBID and C-BID, while N-BID showed little a helix but an abundance of b structure (data not shown). No obvious differences occurred in any of the above proteins after lipid was introduced (data not shown). The high a-helix content of BID is reminiscent of the channel-forming Bcl-xL, Bcl-2 and BAX proteins, and the structurally related bacterial toxins such as the pore-forming colicins and diphtheria toxin . Aggregation of SLUVs induced by tBID In addition to inducing leakage from liposomes, it was interesting to note that tBID, but not full-length BID, could stimulate the aggregation of SLUVs, and that such aggregation was dependent on the protein concentration. Significant aggregation was observed only at tBID concentrations greater than 1 mm (tBID : lipid of approximately 1 : 1000, mol/mol). The line in Fig. 7A shows the time course of the tBID-induced aggregation. Neither proteinase K-treated tBID, caspase-8 nor BAK caused aggregation. The turbidity was unaffected even Fig. 7. Aggregation of SLUVs induced by BID proteins. Aggregation was when these proteins were tested at ratios up to 1 : 25 monitored spectrophotometrically at 600 nm at 30 8C (A) Aggregation of (protein : lipid, mol/mol). SLUVs (1 mm) on the addition of different amounts of tBID. The control It has been reported that Bcl-xL can counteract the function was tBID (10 nmol) pre-incubated with proteinase K for 10 min at 30 8C. of BID, particularly tBID, due to the interaction of the (B) Proteins were mixed with SLUVs (protein:lipid, 1:300, mol:mol) hydrophobic cleft of the former with the BH3 domain of the following pre-incubation of Bcl-xL with tBID or tBID (G94E) (1:1, latter . This is supported by the results described in Fig. 7B; mol:mol) for 10 min at 4 8C. Bcl-xL could prevent the aggregation of LUVs induced by tBID, but had no obvious influence on tBID (G94E), in which the BH3 domain is changed by point mutation. Li et al. have vesicles was negligible. Compared with BID, tBID always described similar phenomena where overexpression of tBID has showed higher affinity to the vesicles. Almost all of the BID led to the clustering together of mitochondria in vivo . proteins tested showed very weak binding affinity for the DOPC vesicles (data not shown). These results showed that the DISCUSSION affinity of individual BID proteins to a particular membrane parallels their capacity to induce leakage. BH3 domain-only proteins have been previously viewed as transdominant inhibitors that rely exclusively on dimerization with other Bcl-2 family proteins to exert effects on cell life and Penetrance of the phospholipid monolayer death . In this work, we have further demonstrated that BID The experiments on vesicles were complemented by studying can directly induce the release of entrapped trypsin and BID protein binding to a phospholipid monolayer. An increase cytochrome c release from LUVs, as has been previously in the surface pressure will be observed where the proteins have reported . The experiments described here were designed to penetrated the phospholipid monolayer. As shown in Fig. 6A, at understand the intrinsic activity of BID as a membrane- an initial surface pressure of about 20 mN´m21, C-BID had the integrating or pore-forming entity in the model membrane highest penetrating ability among these proteins. The order was system. then tBID, BID and BID (D59A); BID (G94E) and N-BID had Firstly, we tested the hypothesis that BID causes leakage almost no penetrating ability. The surface pressure did not by a detergent-like mechanism. In our experiments with BID 54 D. Zhai et al. (Eur. J. Biochem. 268) q FEBS 2001 concentration kept constant, increasing the lipid concentration experiments also show that C-BID has the highest potential resulted in a decrease in the extent of leakage, while decreasing ability to penetrate into membranes compared with BID or the concentration of the lipid led to a reduction in leakage. This tBID. finding is in disagreement with the detergent-like hypothesis, The N-terminal segment itself has no leakage-inducing but is consistent with a monomer±multimer assembly pore- ability, but may serve as an inhibitor of pro-apoptotic activity. forming process . Decreasing the lipid concentration lowers The N-terminal domain remains intact at approximately the the extent of leakage at a constant lipid:BID ratio, which would same position and orientation after caspase-8 cleavage [10,11], confirm this. but even in the uncleaved BID, the N-terminal helices may It has been reported that lipid composition may affect the temporarily detach from the protein core, leading to some efficiency of leakage induction by a pore-forming peptide. For activity of the uncleaved protein . Our results showed that example, cholesterol significantly reduces the efficiency of full-length BID contains much of its intrinsic function leading leakage by the synthetic peptide GALA . Our results also to leakage of liposomes, which may be due to the temporary showed that cholesterol can lead to a clear drop in the detachment of N-terminal segment. efficiency of BID-induced leakage. It can be envisaged that In addition to inducing the passage of cytochrome c, tBID penetration of BID into more rigid membranes is less favorable can trigger the clustering of mitochondria around the nucleus to for pore forming. 1,2-Dioleoylphosphatidylethanolamine also form a ring in the early stage of apoptosis after transient has an obvious effect on BID-induced leakage. Such a transfection of tBID-GFP into an ecdysone-inducible system reduction is partially due to the decrease in the binding affinity . In the present paper, it is interesting to note that tBID, but of BID protein for the liposomes, though it seems that for the not full-length BID, shows the ability to aggregate SLUVs. The same number of membrane-bound BIDs per vesicle, the amount extent of aggregation depends on the amount of tBID added. of leakage from DOPE-containing liposomes is still several- These results mimic the function of tBID in vivo . Full- fold less than that from the DOPA/DOPC or DOPG/DOPC length BID can induce the leakage but is unable to aggregate vesicles. During the preparation of the present paper, Szoka LUVs. et al.  reported that GALA-induced 1-aminonaphthalene- Recently, the three-dimensional structures of both BID and 3,6,8-trisulfonic acid/p-xylene-bis-pyridinium bromide (ANTS/ tBID have been solved [10,11]. The surface electrostatic DPX) leakage is also decreased when the vesicles contain potential of BID does not reveal any unusually charged regions. DOPE. They attribute the reduction to the lowering of bilayer However, two hydrophobic patches appear on the surface. One deformation energy caused by GALA aggregations that are is the BH3 domain, where four partially conserved hydrophobic adsorbed on the membrane surface in the presence of DOPE. resides are exposed (I82, I86, L90, M97), the other is a large Our results also show the influence of DOPE on the efficiency hydrophobic cleft formed by L105, Y140, V150, L151 and of leakage induced by BID. We postulate that the presence of L154. The biological implication of the BID hydrophobic cleft DOPE, which is well known as a nonbilayer-forming phos- remains to be investigated. Inasmuch as cleavage by caspase-8 pholipid, may be unfavorable for the formation of pore can greatly enhance the pro-apoptotic activity of BID, it has structures, as in the case of GALA. It was reported previously been hypothesized that such a dramatic activation is accom- that proteins such as Bcl-2 and Bcl-xL are almost unable to panied by a conformational change after cleavage. However, form ion channels in neutral vesicles [12±15]. In our case, a the overall structural integrity of BID is preserved on caspase-8 small amount of BID was found to bind with DOPC vesicles; cleavage. Minor conformational change predominantly hence only a small amount of entrapped molecules could be occurred in the loop region near the cleavage site , but released. The results obtained with DOPE, cholesterol and the results from McDonnell et al.  suggest that a higher 1,2-dioleoylphosphatidylcholine-containing vesicles further extent of the BH3 domain may be exposed following cleavage. indicate that the pore-forming rather than the micellization Considering the difference between BID and tBID in inducing mechanism may account for the BID-induced leakage. the aggregation of LUVs, we postulate that different domains of The results of binding experiments with the monolayer may tBID molecule may be responsible for the aggregation and indicate that part of the differences between leakage-induction leakage of LUVs. In other words, the BH3 domain of tBID efficiencies of the different BID proteins is due to differences in plays its main role in the clustering of liposomes, the BH3- binding affinities for liposomes, particularly in the case of BID independent hydrophobic cleft is more important for inducing mutants and fragments. Compared with wild-type BID, an leakage. Insertion of the hydrophobic H6 and H7 helical hairpin excessive amount of BID (G94E) is required for the mutant to perpendicularly into the lipid bilayer may form putative pores, induce partial leakage. This result is in line with the lower structurally similar to the pore-forming domain of bacterial affinity of BID (G94E) to vesicles. Hence, we can conclude that toxins . This is further supported by the observation that the BH3 domain of BID is necessary for its binding to Bcl-xL can counteract tBID-induced aggregation as the BH3 membranes. amphipathic helix of tBID binds to the hydrophobic cleft of The difference in the leakage activity of BID (D59A) and Bcl-xL . Adding Bcl-xL to tBID (G94E), which has a point wild-type BID is unexpected, because the mutation site of BID mutation in the BH3 domain, fails to prevent LUV aggregation. (D59A) is not in the BH3 domain or other helices. Although its Furthermore, it is interesting to note that, in contrary to the binding activity with liposomes partially remains, it is no longer results obtained with mitochondria in vivo, Bcl-xL cannot capable of inducing leakage of LUVs. As a consequence of the inhibit the activity of BID or tBID-induced leakage of LUVs. point mutation, a suitable conformation of BID for pore Possibly, in the former case, a BH3-dependent pathway for formation may not be attained. inducing leakage of mitochondrial membrane may co-exist Fragments of BID were used to identify which part may be with a BH3-independent pathway . responsible for inducing LUV leakage. The results showed that The in vitro assays using model membrane systems have C-BID keeps the leakage-inducing activity, while N-BID loses facilitated the study of the mechanism by which BID damages both the ability to bind to liposomes and leakage-inducing mitochondria during apoptosis. Results from different activity. We can propose that C-BID is responsible for inducing approaches showed that the pore-forming rather than the leakage of vesicles. Moreover, the results from monolayer micellization mechanism may account for the BID-induced q FEBS 2001 BID-inducing leakage and aggregation of vesicles (Eur. J. Biochem. 268) 55 leakage of liposomes. Based on the results obtained, we 14. Antonsson, B., Conti, F., Ciavatta, A., Montessuit, S., Lewis, S., tentatively suggest that the BH3 domain and the hydrophobic Martinou, I., Bermasconi, L., Bernard, A., Mermond, J.J., Mazzei, G., cleft of tBID may be responsible for the aggregation and Maundrell, K., Gambale, F., Sadoul, R. & Martinou, J.C. (1997) leakage of phospholipid vesicles, respectively. Inhibition of Bax channel-forming activity by Bcl-2. Science 277, 370±372. 15. Schendel, S.L., Azimov, R., Pawlowski, K., Godzik, A., Kagan, B.L. & ACKNOWLEDGEMENTS Reed, J.C. (1999) Ion channel activity of the BH3 only Bcl-2 family member, BID. J. Biol. Chem. 274, 21932±21936. This work was financially supported by the National Natural Science 16. Green, D.R. (1998) Apoptotic pathways: the roads to ruin. Cell 94, Foundation of China (39730130) and the Chinese Academy of Sciences. We 695±698. thank X. Wang, J. Yuan and Y. Tsujinoto for their generous gifts. We are 17. Shimizu, S., Narita, M. & Tsujlmoto, Y. (1999) Bcl-2 family proteins also grateful to X. Han and X. Hang for their interesting discussions. regulate the release of apoptogenic cytochrome c by the mitochon- drial channel VDAC. Nature 399, 483±487. REFERENCES 18. Wang, K., Yin, X.-M., Chao, D.T., Milliman, C.L. & Korsmeyer, S.J. (1996) BID: a novel BH3 domain-only death agonist. Genes Dev. 10, 1. Steller, H. (1995) Mechanisms and genes of cellular suicide. 2859±2869. Science 267, 1445±1449. 19. Luo, X., Budihardjo, I., Zou, H., Slaughter, C. & Wang, X. (1998) BID, 2. Adams, J.M. & Cory, S. (1998) The Bcl-2 protein family: arbiters of a Bcl2 interacting protein, mediates cytochrome c release form cell survival. Science 281, 1322±1326. mitochondria in response to activation of cell surface death receptors. 3. Martinou, J.C. (1999) Key to the mitochondria gate. Nature 399, Cell 94, 481±490. 411±412. 20. Li, H., Zhu, H., Xu, C.-J. & Yuan, J. (1998) Cleavage of BID by 4. Li, H. & Yuan, J. (1999) Deciphering the pathways of life and death. caspase-8 mediates the mitochondrial damage in the Fas pathway of Curr. Opin. Cell Biol. 11, 261±266. apoptosis. Cell 94, 491±501. 5. Chao, D.T. & Korsmeyer, S.J. (1998) Bcl-2 family: regulators of cell 21. Tan, K.O., Tan, K.M.L. & Yu, V.C. (1999) A novel BH3-like domain in death. Annu. Rev. Immunol. 16, 395±419. BID is required for intramolecular interaction and autoinhibition of 6. Muchmore, S.W., Sattler, M., Liang, H., Meadow, R.P., Harlan, J.E., pro-apoptotic activity. J. Biol. Chem. 274, 23697±23690. Yoon, H.S., Nettesheim, D., Chang, B.S., Thompson, C.B., Wong, 22. Zhai, D.-Y., Huang, X.-X., Han, X.-H. & Yang, F. (2000) Character- S.L., Ng, S.L. & Fesik, S.W. (1996) X-ray and NMR structure of ization of tBID-induced cytochrome c release from mitochondria and human Bcl-xL, an inhibitor of programmed cell death. Nature 381, liposomes. FEBS Lett. 472, 293±296. 335±341. 23. Rietveld, A., Jordi, W. & Ben, D.K. (1985) Studies on the lipid 7. Sattler, M., Liang, H., Nettesheim, D., Meadow, R.P., Harlan, J.E., dependency and mechanism of the translocation of the mitochondrial Eberstadt, M., Yoon, H.S., Shuker, S.B., Chang, B.S., Minn, A.J., precursor protein apocytochrome c across model membrane. J. Biol. Thompson, C.B. & Fesik, S.W. (1997) Structure of Bcl-xL-Bak Chem. 261, 3846±3856. peptide complex: recognition between regulators of apoptosis. 24. Bradford, M.M. (1976) A rapid and sensitive method for the Science 275, 983±986. quantitation of microgram quantities of protein utilizing the principle 8. Chittenden, T., Flemington, C., Houghton, A.B., Ebb, R.G., Gallo, G.J., of protein-dye binding. Anal. Biochem. 162, 11±17. Elangovan, B., Chinnadurai, G. & Lutz, R.J. (1995) A conserved 25. Han, X.-H., Sui, S.-F. & Yang, F. (1996) A mini-trough for the study of domain in Bak, distinct from BH1 and BH2, mediates cell death and membrane insertion ability of proteins. Thin Solid Films 284/285, protein binding functions. EMBO J. 14, 5589±5596. 789±792. 9. Green, D.R. & Reed, J.C. (1998) Mitochondria and apoptosis. 26. Marsh, D. (1996) Lateral pressure in membranes. Biochim. Biophys. Science 281, 1309±1312. Acta 1286, 183±223. 10. Chou, J.J., Li, H., Salvesen, G.S., Yuan, J. & Wagner, G. (1999) 27. Nicol, F., Nir, S. & Szoma, F.C. Jr (1996) Effect of cholesterol and Solution structure of BID, an intracellular amplifier of apoptotic charge on pore formation in bilayer vesicles by a pH-sensitive signaling. Cell 96, 615±624. peptide. Biophys. J. 71, 3288±3301. 11. McDonnell, J.M., Fushman, D., Milliman, C.L., Korsmeyer, S.J. & 28. Nicol, F., Nir, S. & Szoma, F.C. Jr (2000) Effect of phospholipid Cowburn, D. (1999) Solution structure of the proapoptotic molecule composition on an amphipathic peptide-mediated pore-formation in BID: a structural basis for apoptotic agonists and antagonists. Cell 96, bilayer vesicles. Biophys. J. 78, 818±829. 625±634. 29. Parente, R.A., Nir, S. & Szoma, F.C. Jr (1990) Mechanism of leakage 12. Minn, A.J., Velez, P., Schendel, S.L., Liang, H., Muchmore, S.W., of phospholipid vesicle contents induced by the peptide GALA. Fesik, S.W. & Fill, M. (1997) Bcl-xL forms an ion channel in Biochemistry 29, 8720±8728. synthetic lipid membranes. Nature 385, 353±357. 30. Xie, Z.H., Schendel, S., Matsuyama, S. & Reed, J.C. (1998) Acidic pH 13. Schendel, S.L., Xie, Z., Montal, M.O., Matsuyama, S., Montal, M. & promotes dimerization of Bcl-2 family proteins. Biochemistry 37, Reed, J.C. (1997) Channel formation by antiapoptotic protein Bcl-2. 6410±6418.31. Proc. Natl Acad. Sci. USA 93, 5113±5118.