The Destabilization of Lipid Membranes Induced by the C-terminal Fragment of Caspase 8-cleaved Bid Is Inhibited by the N-terminal Fragment*

Bid is a proapoptotic, BH3-domain-only member of the Bcl-2 family. In Fas-induced apoptosis, Bid is activated through cleavage by caspase 8 into a 15.5-kDa C-terminal fragment (tcBid) and a 6.5 kDa N-terminal fragment (tnBid). Following the cleavage, tcBid translocates to the mitochondria and promotes the release of cytochromec into the cytosol by a mechanism that is not understood. Here we report that recombinant tcBid can act as a membrane destabilizing agent. tcBid induces destabilization and breaking of planar lipid bilayers without appearance of ionic channels; its destabilizing activity is comparable with that of Bax and at least 30-fold higher than that of full-length Bid. Consistently, tcBid, but not full-length Bid, permeabilizes liposomes at physiological pH. The destabilizing effect of tcBid on liposomes and planar bilayers is independent of the BH3 domain. In contrast, mutations in the BH3 domain impair tcBid ability to induce cytochrome c release from mitochondria. The permeabilizing effect of tcBid on planar bilayers, liposomes, and mitochondria can be inhibited by tnBid. In conclusion, our results suggest a dual role for Bid: BH3-independent membrane destabilization and BH3-dependent interaction with other proteins. Moreover, the dissociation of Bid after cleavage by caspase 8 represents an additional step at which apoptosis may be regulated.

. Recently, Bid-deficient mice have been shown to be resistant to Fas-induced hepatocellular apoptosis (6). Bid is a cytosolic protein in nonapoptotic cells, present in a variety of tissues (5). Upon induction of certain types of apoptosis, Bid is cleaved by caspase 8, and its C-terminal fragment translocates to the mitochondria (7)(8)(9). According to the x-ray structure, cleavage of Bid unmasks a large hydrophobic surface in its C-terminal fragment (10,11). The C-terminal fragment of Bid (called t c Bid) is regarded as the active form. It is far more potent in inducing apoptosis than FL Bid (7,9), and it can be inhibited by coexpression of the N-terminal fragment (t n Bid) (12).
Bid is believed to exert its proapoptotic effect by inducing the release of proapoptotic factors (cytochrome c, apoptosis-inducing factor, and procaspase 9) from mitochondria. Incubation of subnanomolar concentrations of t c Bid with isolated mitochondria results in a complete release of cytochrome c (7). More recently, it has been shown that Bid partially permeabilizes the outer mitochondrial membrane and causes the release of several proteins from the intermembrane space (13). A major controversy is whether Bid acts by itself or through modulation of other Bcl-2 family proteins.
In vitro binding and mutagenesis experiments support the hypothesis that Bid acts by regulating other Bcl-2 family proteins. FL Bid can bind Bax, Bcl-X L , and Bcl-2 (5). When the BH3 domain of Bid is mutated so that it no longer binds Bax, FL Bid fails to promote apoptosis (5). Consistently, FL Bid is over 10 times less active in releasing cytochrome c from Bax Ϫ/Ϫ mitochondria than from wild type mitochondria. Moreover, FL Bid induces a conformational change in Bax, which correlates with the release of cytochrome c (14,15). t c Bid has a higher affinity than FL Bid for Bax 2 and for Bcl-X L (7,9,12). It has been proposed that once Bid is cleaved, t c Bid binds Bcl-X L and inhibits its antiapoptotic activity.
Along with the data suggesting the importance of interaction with other proteins, there is increasing evidence for the possibility that Bid may directly permeabilize mitochondrial membranes. Recently, the channel-forming activity of Bid in synthetic lipid bilayers has been reported (16). Channel-forming activity has already been observed with other Bcl-2 family proteins (for a review see Ref. 17). Surprisingly, the channels formed by pro-and antiapoptotic proteins do not seem to differ significantly. However, in a striking paper, Basanez et al. (18) have shown that Bax, but not Bcl-X L , decreases the lifetime of planar lipid bilayers. They proposed that this effect was due to a decrease of the linear tension of the bilayers. In mitochondria, * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  this would lead to the formation of lipidic holes, allowing the release of cytochrome c. Here we describe a similar membrane destabilizing effect specific to t c Bid. We also show that t c Bid activity is inhibited by the N-terminal Bid fragment. We discuss the role of the conserved BH3 domain in Bid activity.

EXPERIMENTAL PROCEDURES
Protein Purification-Oligomeric C-terminal truncated Bax was purified as described earlier (19). Full-length His-tagged mouse Bid, wild type and mutants, were purified as described by Desagher et al. (14). VDAC was a kind gift from Dr K. Zeth. Cytochrome c and gramicidin were from Sigma. t c Bid (mouse Bid residues 60 -195) with a tag of six histidines at the N terminus was expressed in the pET23d vector in Escherichia coli. The protein was recovered in the soluble bacteria fraction and purified by chromatography on nickel-nitrilotriacetic acid-agarose followed by Q-Sepharose. The protein was stored in 25 mM Tris-HCl, 100 mM NaCl, 1% OG, 0.2 mM DTT, 30% glycerol, pH 7.5, at Ϫ80°C. t n Bid (mouse Bid residues 1-59) with a tag of six histidines at the C terminus was expressed in the pET23a vector in E. coli. The protein was recovered in the soluble bacteria fraction and purified by chromatography on nickel-nitrilotriacetic acid-agarose followed by MonoQ and gel filtration on Superdex 200. The protein was over 95% pure and was stored in 25 mM Tris-HCl, 0.2 mM DTT, 30% glycerol, pH 7.5, at Ϫ80°C.
Cutting Bid with Caspase 8 -200 l of recombinant wild type or mutated full-length (FL) Bid in 25 mM Tris, 0.2 mM DTT, 30% glycerol, pH 7.5, was diluted with 200 l of cutting buffer (50 mM Hepes, 100 mM NaCl, 10 mM DTT, 1 mM EDTA, 10% sucrose, pH 7.5) to 6.2 mg/ml. 1 l of recombinant caspase 8 at 5.6 mg/ml was added, and the sample was incubated at room temperature for 2 h. The cutting efficiency was estimated to be over 95% by SDS-polyacrylamide gel electrophoresis and Coomassie staining.
Electrophysiological Recording-Bilayers were formed from a solution of azolectin IV-S lipids (Sigma) dissolved in n-decane at 30 mg/ml across a 250-m diameter hole. Protein was added to one side of the bilayer (defined as cis). The solution bathing the membrane contained 400 mM KCl, 10 mM Tris, 1 mM NaN 3 , pH 7.4, on the cis side and 100 mM KCl, 10 mM Tris, 1 mM NaN 3 , pH 7.4, on the trans side. Current was recorded using an Axon 200B patch clamp amplifier and stored on digital audio tape (Biologic DTR 1200 recorder). Records were subsequently filtered at 1 kHz through a 4-pole Bessel filter and digitized offline at 2 kHz. The membrane potential refers to the potential of the cis side minus the potential of the trans side.
Liposome Permeabilization Assay-5,6-Carboxyfluorescein (CF) containing liposomes were prepared as described earlier (21). Briefly, 400 g of phosphatidylcholine, 400 g of phosphatidylserine, and 230 g cholesterol were solubilized in PBS, pH 7.2, containing 20 mM CF and 30 mg/ml OG through incubation at room temperature for 3 h. The liposomes were subsequently isolated by passage over a Sephadex G-25 column (1.5 ϫ 20 cm) and dialyzed in PBS. The liposomes were diluted in PBS to give a suitable fluorescence measurement, then Bid or control proteins were added as indicated in the figures, and the change in fluorescence was recorded over time with excitation at 488 nm and emission at 520 nm.
In Vitro Assay for Cytochrome c Release-The subcellular fractionation was performed as described earlier (15). Mitochondria (100 g) were incubated in the presence or absence of various recombinant proteins in 100 l of MBC buffer (210 mM mannitol, 70 mM sucrose, 10 mM Hepes, 0.5 mM EGTA, 4 mM MgCl 2 , 5 mM Na 2 HPO 4 , 5 mM succinate,

FIG. 2. Effect of Bid and control proteins on the lifetime of planar lipid bilayers.
We measured the lifetime of azolectin bilayers after setting a 250 mV membrane potential with different proteins in the bilayer chamber at pH 7.5. Before addition of each protein the average lifetime without protein was measured. The lifetimes with t c Bid and control proteins are expressed as percentages of the lifetime without proteins. Data represent the means and S.E. of 12-24 experiments. cut Bid, Bid cut with caspase 8. 5 M rotenone, pH 7.5) for 15 min at 30°C and then centrifuged for 5 min at 13,000 ϫ g at 4°C. The supernatants were separated by SDSpolyacrylamide gel electrophoresis on 4 -20% Tris-Gly gels (NOVEX), and the amount of cytochrome c was estimated by Western blotting using a polyclonal anti-cytochrome c antibody (raised in-house, dilution 1:1000) and a horseradish peroxidase-conjugated goat anti-rabbit IgG (from Dako, Denmark, dilution 1:1000) and developed with ECL from Amersham Pharmacia Biotech.

Destabilization of Planar Lipid Bilayers by Bid-
The permeabilizing properties of t c Bid were investigated using planar lipid bilayers. Addition of recombinant t c Bid (10 -300 nM) to the cis compartment of a bilayer chamber invariably resulted in the rupture of the bilayer within minutes (Fig. 1, curve a). Membrane ruptures occurred at low membrane potentials, from 20 to 80 mV, at which the bilayer was normally stable without addition of t c Bid. Membrane breaking could only be induced by t c Bid; the dialysis buffer and FL Bid had no effect. However, Bid gained the ability to break lipid bilayers at nanomolar concentrations after cutting with recombinant caspase 8.
When using the highest concentrations of t c Bid, we occasionally saw voltage-dependent channel activity, which had all the characteristics of bacterial porins. However, this was also observed after adding FL Bid to the bilayer chamber at similar concentrations and seems to be an inevitable consequence of working with high concentrations of bacteria-expressed proteins. Aside from the occasional porin activity, no other channel activity could be recorded prior to the rupture of the membrane. The bilayer ruptures induced by t c Bid were not preceded by any discernible increases in conductance, suggesting that they were not simply due to a rapid insertion of a large number of channels. In contrast, addition of channel-forming proteins like VDAC or gramicidin to the bilayer chamber caused gradual increases of membrane conductance that did not lead to the rupture of the bilayer (Fig. 1, curves b and c). Lowering the concentration of t c Bid to 1-3 nM decreased the frequency of the bilayer breaking events without the appearance of channel activity. At still lower protein concentrations, the bilayers were electrically silent. Varying salt concentration between 50 and 400 mM or doping the bilayer with cholesterol did not induce formation of channels by t c Bid. In studying Bax channel properties, we observed that the insertion of channels was often followed by rupture of the bilayer, an effect that has been recently analyzed in detail by Basanez et al. (18). We found that preincubation of Bax with azolectin liposomes and fusion of these liposomes to the planar bilayer under asymmetrical conditions, as described by Schlesinger et al. (22), allowed the recording of sustained channel activity at low Bax concentrations without rupture of the bilayer (data not shown). Using this procedure, we were unable to detect t c Bid channel activity.
To quantify the destabilizing effect of t c Bid, we compared the lifetime of lipid bilayers in the presence and absence of t c Bid or control proteins in the bilayer chamber. At low membrane potential, lipid bilayers are very stable in the absence of t c Bid, and their lifetime could not be meaningfully measured. We therefore used the procedure described by Basanez et al. (18) and worked at a very high membrane potential (250 mV) at which the bilayers ruptured spontaneously within tens of seconds. Under these conditions, addition of 1 nM t c Bid decreased the lifetime more than 2-fold, and 10 nM t c Bid decreased the lifetime approximately 30-fold (Fig. 2). The activity of Bid cut with caspase 8 was 10-fold weaker, whereas FL Bid and lipid-interacting proteins (bovine serum albumin, cytochrome c, and mitochondrial channel VDAC) had no effect on the bilayers' lifetime at concentrations comparable to t c Bid. In general, at 160 or 250 mV membrane potential, t c Bid was over 30 times more active in inducing bilayer rupture than FL Bid ( Fig. 2

and data not shown). Bax had an activity comparable with that of t c Bid. t c Bid Permeabilizes Liposomes and Mitochondria, and t n Bid Inhibits This Effect-Another method used to investigate protein-lipid interactions is based on the release of a fluorescent dye from liposomes. Addition of 100 nM t c Bid to a solution containing
phosphatydylcholine-phosphatydylserine-cholesterol (39%:39%:22%) liposomes filled with CF resulted in a strong release of CF at pH 7.2. Bid cut with caspase 8 was much less efficient in inducing CF release, and FL Bid did not induce any release at all (Fig. 3A). Fig. 3B shows the release of CF from liposomes after 3 min of incubation with t c Bid, Bid cut with caspase 8 and FL Bid as a function of protein concentration.
The finding that Bid cut with caspase 8 was much less active than t c Bid in permeabilizing planar bilayers and liposomes suggested that t n Bid may have an inhibitory function. To test this hypothesis, we produced recombinant t n Bid and tested its activity on liposomes. As expected, t n Bid did not induce any CF release by itself (data not shown). However, when incubated with t c Bid, it inhibited its activity in a concentration-depend-ent manner (Fig. 3C). The activity of 100 nM t c Bid was reduced by half with concentrations of t n Bid as low as 50 nM. The effect of t n Bid was specific for t c Bid, because it did not inhibit the permeabilization of liposomes induced by Bax (data not shown).
We examined whether t n Bid is capable of inhibiting t c Bidinduced release of cytochrome c from mitochondria. It has been shown earlier that addition of 1 nM t c Bid results in a complete release of cytochrome c. Our experiments confirm this result (Fig. 3D, lane 2). Co-addition of 1 nM t n Bid markedly decreased t c Bid effect (lane 3), whereas 100 nM t n Bid suppressed t c Bid effect almost completely (lane 5). t n Bid alone did not induce any release of cytochrome c at concentrations up to 100 nM (lanes 6  and 7).
OG Dissociates Bid Cut with Caspase 8 and Increases Its Permeabilizing Activity-The release of CF from liposomes induced by Bid cut with caspase 8 (but not by FL Bid) could be potentiated by preincubating Bid with OG, a nonionic detergent. Cut Bid was activated after incubation with 2% but not with 0.2% OG (Fig. 4A). The detergent alone did not induce CF release at the concentrations used. The addition of cut Bid and OG directly to the liposomes without preincubation did not induce more CF release than addition of cut Bid alone (data not shown). Because the effect of OG depended on its concentration during the preincubation with cut Bid and not on the final FIG. 4. OG dissociates Bid cut with caspase 8 and increases its CF-releasing activity. A, Bid was cut with caspase 8 and incubated for 30 min at RT with 0.2 or 2% OG. A suitable amount was then added to 900 l of liposomes (at the time indicated by an arrow) and adjusted to 1 ml with PBS to obtain 240 nM cut Bid and 0.002% OG final concentration in all assays. Curve a, cut Bid incubated with 2% OG; curve b, cut Bid incubated with 0.2% OG; curve c, cut Bid alone (240 nM); curve d, OG alone (0.002%). B-E, the Superdex 200 column was equilibrated in 25 mM Hepes/NaOH/300 mM NaCl/0.2 mM DTT, pH 7.5, with or without 2% OG as indicated below. The column was run at a flow rate of 1 ml/min, and 750 g of Bid cut with caspase 8 was loaded in 500 l. The eluate was monitored at 280 nm, and fractions of 2 ml were collected. The fractions were analyzed by silver staining. B and C, elution without OG. D and E, elution with 2% OG. concentration with the liposomes, we concluded that the OG effect was due to its direct action on cut Bid rather than on the liposomes.
To investigate the effect of OG on Bid quaternary structure, we measured the elution times of Bid from a Superdex 200 column. In the absence of detergent in the migration buffer, FL Bid migrated at a molecular mass of 24 kDa, which corresponds to its calculated monomeric mass (22,844 Da) (data not shown). After cutting with caspase 8, Bid still migrated in one peak at 24 kDa (Fig. 4B). The proportions of C-and N-terminal fragments were identical in all fractions forming the peak (Fig. 4C). This suggests that Bid is not dissociated in solution after cleavage by caspase 8. Moreover, when recombinant t n Bid and t c Bid are preincubated without detergent, they also migrate together on gel filtration, showing their ability to associate (data not shown). However, when the migration buffer contained 2% OG, the C-and N-terminal fragments of Bid dissociated. A second peak can be detected in the elution profile as a shoulder running before the main peak (Fig. 4D), and analysis of the eluted fractions shows that the shorter N-terminal fragment migrated at a lower molecular mass than t c Bid (Fig. 4E). Recombinant t c Bid migrated in the presence of OG at the same elution time as the C-terminal fragment of Bid cut with caspase 8 (data not shown). It should be noted that the elution time of the proteins in OG did not correspond to their calculated molecular mass, probably because of oligomer formation.
Permeabilizing Properties of Bid BH3 Mutants-The BH3 domain is the only fragment of Bid showing sequence similarity to other Bcl-2 family proteins. It has been shown to be important for the proapoptotic activity of Bid as well as for its binding to Bax, Bcl-X L , and Bcl-2 (5). To test whether the BH3 domain is necessary for the permeabilizing activity of Bid, we produced two BH3 mutants cut by caspase 8: cut Bid mIII-2 ( 93 IGDE 96 3 AAAA) and cut Bid mIII-3 (G 94 3 A). Both mutants were less efficient in inducing cytochrome c release from mitochondria than wild type cut Bid. (Fig. 5A). The mIII-2 mutation decreased Bid activity by a factor of 20. These results support the importance of the BH3 domain for Bid proapoptotic activity. Surprisingly, both mutants had an activity similar to the wild type protein in releasing CF from liposomes (Fig. 5B) and in decreasing the lifetime of planar lipid bilayers (Fig. 5C).

DISCUSSION
Intrinsic Activity of t c Bid-In the present study, we describe a membrane destabilizing activity of t c Bid. Addition of nanomolar concentrations (up to 300 nM) of t c Bid to the bilayer chamber results in the rupture of lipid bilayers without appearance of ionic channels. t c Bid also induces a vigorous release of CF from liposomes. Interestingly, both events occur at physiological pH, and both are specific to t c Bid, because FL Bid is neither active in destabilizing planar bilayers nor in permeabilizing liposomes.
Recently, Schendel et al. (16) have described channel-forming activity of t c Bid in planar lipid bilayers at acidic and neutral pH, with 1.8 -3.6 M (30 -60 g/ml) t c Bid in the bilayer chamber. The channels were voltage-dependent and displayed multiple conductance levels ranging from 7.4 pS to 100 pS in 150 mM KCl. Channels were also detected with high concentrations of FL Bid but only at acidic pH. In contrast to our results, Schendel et al. (16) did not describe any lytic effect of t c Bid in planar lipid bilayers. The channel-forming activity was observed only at micromolar concentrations, whereas in our experiments t c Bid had a strong membrane destabilizing activity already in the low nanomolar range. Whether differences in experimental procedures or in protein preparation account for these differences is unclear.
Inhibitory Effect of t n Bid-Previous experiments have shown that Bid is activated through cleavage by caspase 8 (7,9). t c Bid has cytochrome c releasing and proapoptotic activities much stronger than those of FL Bid, whereas t n Bid is inactive. Furthermore, it has been shown that cleavage of Bid by caspase 8 does not result in its immediate dissociation in vitro (11). This suggested that t n Bid may be an inhibitor of t c Bid activity. Consistently, coexpression of t n Bid with t c Bid has been shown to reduce the proapoptotic activity of t c Bid in MCF-7 cells (12).
It has been believed that t n Bid inhibits t c Bid by masking its BH3 domain, thereby inhibiting its interaction with other Bcl-2 family proteins (10). Here we report that t n Bid inhibits t c Bid permeabilizing activity even in the absence of additional proteins. First, Bid cut with caspase 8 is much less active in liposomes and planar bilayers than t c Bid. Second, addition of t n Bid potently inhibits the CF releasing activity of t c Bid in liposomes. Third, gel filtration experiments confirm that the Cand N-terminal fragments of Bid remain associated after cleavage by caspase 8. Incubation of cut Bid with OG both dissociates the C-and N-terminal fragments and increases Bid activity. Importantly, t n Bid is also able to inhibit t c Bid-induced release of cytochrome c from mitochondria. We suggest that t n Bid acts by masking the large hydrophobic domain of t c Bid and inhibiting its association with mitochondrial lipids.
The dissociation of t c Bid and t n Bid is an additional step at which Bid activity can be regulated. Several mechanisms for this regulation are possible: specific protein-protein or proteinlipid interactions may promote the dissociation; t n Bid may be proteolysed after cleavage; or phosphorylation of Bid may promote or inhibit dissociation. The hypothesis that a specific interaction with lipids may promote the dissociation of cleaved Bid merits special attention. We have shown that cleaved Bid dissociates in the presence of a nonionic detergent, a condition that could mimic the membrane environment. Selective targeting of Bid to mitochondria could be the consequence of a specific ability of mitochondrial lipids to dissociate the two subunits.
Mechanism of Bid Action-What is the physiological relevance of our findings? Is the bilayer destabilizing activity of t c Bid sufficient, or necessary, for its proapoptotic function? The proapoptotic effect of Bid seems to rely on its ability to induce cytochrome c release from mitochondria into the cytosol. The mechanism of cytochrome c release is the subject of much controversy. It has been proposed to depend upon PTP opening and mitochondrial swelling (23,24) or channels formed by Bax (20 -22). Here we suggest that the membrane destabilizing activity of t c Bid could reflect its ability to directly permeabilize mitochondrial membranes in vivo.
Recently, Basanez et al. (18) described a destabilizing effect of Bax on planar lipid bilayers. They proposed that Bax decreases the linear tension of the membranes, which promotes the formation of lipidic holes in mitochondria. Cytochrome c and other proteins could then be released through those holes. The activity of t c Bid described here strongly suggests that it could play a role similar to Bax. It is of interest to note that whereas Bax induces both channel activity and bilayer rupture, in our experiments with Bid, only the destabilizing effect was observed. This effect could therefore reflect an important physiological role of these proteins.
One important question to be addressed is the physiological role of the conserved BH3 domain of Bid. Some mutations in the BH3 domain (mIII-2 and mIII-3) do not alter the membrane destabilizing activity of cleaved Bid in artificial models, but they decrease the cytochrome c releasing activity. The difference may be explained by the impaired binding of the mutants to Bax (or another protein) (7). The activity of Bid in vivo would then rely on two mechanisms: first, a BH3-independent permeabilization of mitochondrial membranes, and second, a BH3dependent activation of Bax or inhibition of Bcl-X L . Both activities of Bid are regulated by cleavage with caspase 8. We suggest that the dissociation of the inhibitory N-terminal fragment following the cleavage is an additional step necessary for the activation of Bid. The physiological mechanism regulating this dissociation needs further investigation.