A Membrane-targeted BID BCL-2 Homology 3 Peptide Is Sufficient for High Potency Activation of BAX in Vitro*

The multidomain pro-apoptotic proteins BAX and BAK constitute an essential gateway to mitochondrial dysfunction and programmed cell death. Among the “BCL-2 homology (BH) 3-only” members of pro-apoptotic proteins, truncated BID (tBID) has been implicated in direct BAX activation, although an explicit molecular mechanism remains elusive. We find that BID BH3 peptide alone at submicromolar concentrations cannot activate BAX or complement BID BH3 mutant-tBID in mitochondrial and liposomal release assays. Because tBID contains structurally defined membrane association domains, we investigated whether membrane targeting of BID BH3 peptide would be sufficient to restore its pro-apoptotic activity. We developed a Ni2+-nitrilotriacetic acid liposomal assay system that efficiently conjugates histidine-tagged peptides to a simulated outer mitochondrial membrane surface. Strikingly, nanomolar concentrations of a synthetic BID BH3 peptide that is chemically tethered to the liposomal membrane activated BAX almost as efficiently as tBID itself. These results highlight the importance of membrane targeting of the BID BH3 domain in tBID-mediated BAX activation and support a model in which tBID engages BAX to trigger its pro-apoptotic activity.

BID is a BH3-only protein that connects the extrinsic TNFR1 and Fas death signaling pathway to the mitochondrial amplification cascade of the intrinsic death pathway (5). Engagement of TNFR1 and Fas activates caspase-8, which cleaves p22 BID into N-terminal 7-kDa and C-terminal 15-kDa fragments (6,7,15). The C-terminal fragment (p15 BID or tBID) is then myristoylated at the newly generated N-terminal glycine (16). Myristoylated tBID targets mitochondrial contact sites, which are enriched with cardiolipin (17)(18)(19). Membrane-targeted tBID triggers the homo-oligomerization of BAX or BAK in the mitochondrial outer membrane (MOM) (9,11,12,20), resulting in MOM permeabilization and release of intermembrane space proteins, including cytochrome c, into the cytosol (21). Released cytochrome c forms an apoptosome complex with Apaf-1 and caspase-9, leading to activation of effector caspases and commitment to cell death (22).
The BH3 domain of BID is believed to play an essential role in BAX activation (5). Recent studies using an in vitro liposomal assay demonstrated that BAX protein, when combined with BID and BIM BH3 peptides, triggered release of liposomal contents, supporting a direct interaction between select BH3 peptides and BAX (12)(13)(14). In these studies, an excess of synthetic peptides (up to 50 M) was required to activate submicromolar concentrations of BAX, raising the question of whether this in vitro observation reflects the in vivo mechanism of BAX activation (23). We investigated whether endowing the BID BH3 peptide with membrane targeting capability could restore its biological potency, and thereby confirm its physiologic activity. Using a novel Ni 2ϩ -NTA-containing liposomal assay system, we find that tethering a BID BH3 peptide to the membrane by chemical means increases its potency in BAX activation by 2-3 orders of magnitude compared with the untargeted peptide. These data reveal the functional importance of the membrane targeting module of tBID, in combination with the BH3 domain, in triggering BAX activation.
Expression and Purification of Recombinant Proteins-Full-length human BAX protein was purified as described (27) and further purified by gel filtration chromatography using a Superdex-75 column (GE Healthcare) pre-equilibrated with buffer A containing 10 mM DTT. Two BAX mutants, one lacking helices 5 and 6 (amino acid sequence 109 -144), designated as BAX⌬⌯56, and the other with a glycine to valine substitution at residue 108, designated as BAXG108V, were engineered using the QuikChange mutagenesis kit (Stratagene). The BAX mutant proteins were prepared as described above for BAX (27). To ensure that equal concentrations of BAX and BAX mutants were employed in the biochemical experiments, BAX proteins were calibrated by Western blot analysis using the N-20 primary antibody (Santa Cruz Biotechnology) and anti-rabbit secondary IgG coupled to horseradish peroxidase (GE Healthcare). A glutathione S-transferase fusion protein of BCL-X L (residues 1-212) lacking the C-terminal transmembrane domain, designated as GST-BCL-X L ⌬C, was prepared as described (27). Caspase-8 was purified as described (28). To express N-terminally His-tagged, full-length mouse BID proteins, the cysteine-less clone p22BID30S126S was prepared in pJO vector (see supplemental Fig. S2) as described previously (29). From this clone, four BID mutant genes were generated by alanine substitution mutations using the QuikChange mutagenesis kit (Stratagene) and designated as follows: BID 97A98A (with M97A/D98A substitutions), BID 90A95A (L90A/D95A substitutions), BID (93-96)A (with I93A/G94A/D95A/E96A substitutions), and BID 94A (with G94A substitution). All the "wild type" and BH3 "mutant" BID proteins have C30S and C126S substitutions (29). BH3 BID mutant proteins were purified from the periplasmic extracts by Ni 2ϩ affinity and anion-exchange chromatography using an AKTA FPLC system (GE Healthcare), as described (29). Proteins were stored at Ϫ80°C in 20 mM Tris buffer (pH 7.5) containing 16% (v/v) glycerol, 10 mM DTT, and ϳ200 mM NaCl. Protein concentrations were determined by the Bradford assay with Coomassie Brilliant Blue G-250 dye (Bio-Rad) using bovine serum albumin as a standard (30). tBID or p15 BID samples were prepared from the corresponding p22 BID proteins as follows. p22 BID proteins (ϳ5 mg) were first cleaved by incubation with caspase-8 in the presence of 10 mM DTT, resulting in p7/p15 BID. DTT was then removed by eluting the reaction mixture through a PD-10 desalting column (GE Healthcare) after pre-equilibration with buffer A. The purified p7/p15 BID was then loaded onto a column containing 0.5 ml of Ni 2ϩ -NTA-agarose beads (Qiagen) and washed with 10 ml of buffer A. p15 BID was eluted with 40 ml of buffer A containing 2% n-octyl-␤-D-glucopyranoside (Calbiochem). The eluted p15 BID solution was then concentrated using a centrifugal concentrator (5 kDa; Millipore) to a final volume of ϳ2.5 ml. To remove the excessive detergent, the resultant p15 BID concentrate was passed through a PD-10 column pre-equilibrated with 2% n-octyl-␤-D-glucopyranoside. Protein concentration was determined using bicinchoninic acid protein assay kit (Pierce) using bovine serum albumin as a standard (31).
Peptide Synthesis-Peptides corresponding to the BH3 domains of wild type and mutant BID or BAD proteins (see Table 1) were synthesized at the Tufts Peptide Synthesis Core facility. The peptides were purified by reverse-phase high pressure liquid chromatography and masses confirmed by mass The peptides listed above were further modified by addition of an N-terminal His 6 tag. The N and C termini of the peptides were otherwise derivatized with acetyl and amide groups, respectively.

Peptide
Sequence Species NLWAAQRYGRELRRMSDEFVDSFKK Human spectrometry. His-tagged synthetic BH3 peptides were similarly prepared and included six N-terminal histidines. Lyophilized peptides were reconstituted in dimethyl sulfoxide (Me 2 SO) to generate stock solutions for biochemical experiments. Hydrocarbon-stapled BID BH3 peptides were synthesized as described previously (32). Peptide dry weight was measured directly or, when not feasible, determined by amino acid analysis (Molecular Biology Core Facility, Dana-Farber Cancer Institute).
In Vitro Cytochrome c Release Assay-The biological activity of tBID proteins was measured by cytochrome c release assay (6,29). Mouse liver mitochondria were prepared from Bak Ϫ/Ϫ mice at 4°C as described previously (6,29,33). BAX with or without tBID samples were added to 100 l of the mitochondrial suspension (0.5 mg of protein/ml). The reaction mixture was incubated at room temperature for 40 min and then centrifuged at 12,000 ϫ g for 5 min at 4°C to pellet the mitochondria. The supernatants were quickly removed, and the pellets were resuspended in 100 l of phosphate-buffered saline containing 0.5% Triton X-100. The amount of cytochrome c in the supernatants and pellet fractions was determined by ELISA (Quantikine M kit; R & D Systems, Minneapolis, MN).
Liposomal Release Assay-We adapted the fluorescence dequenching assay described by Terrones et al. (14) and Kuwana et al. (12,13) to assess the biological activity of BAX and tBID proteins and select BH3 peptides. The release of the FITC-labeled Dextran 10 from the LUVs was monitored by fluorometry (FluoroMAX-2, ISA Inc.) using a thermostatted 1-cm path length quartz cuvette with constant stirring at 37°C. Excitation and emission wavelengths were 488 and 525 nm, respectively (slits, 2 nm). The extent of marker release was quantified on a percentage basis according to the following equation: ((F t Ϫ F 0 )/(F 100 Ϫ F 0 ) ϫ 100), where F t is the measured fluorescence of reagent-treated LUVs at time t; F 0 is the average fluorescence of the LUV suspension for the initial 1-2 min before reagent addition, and F 100 is the average fluorescence value of the final 1-2 min after complete disruption of LUVs by addition of Triton X-100 (final concentration, 0.66 mM). Lipid concentration was 10 g/ml unless otherwise stated.
BCL-X L Binding Assays-BID BH3 peptides were incubated with lipid vesicles for 10 -20 min at room temperature in 50 l of buffer A with or without EDTA. An equal volume of GST-BCL-X L ⌬C, preincubated with or without a competing peptide (BAD BH3), was added to the reaction mixture at 10-fold excess and co-incubated for 60 min. Final concentrations of BID BH3 peptides, GST-BCL-X L ⌬C, EDTA, and lipids were 20 M, 10 M, 10 mM, and 4.23 g/l, respectively. After incubation, the vesicles were pelleted by 1 h of centrifugation in a Beckman A-110 Airfuge rotor at 104,000 rpm and then successively washed (buffer A) and pelleted again. The pellets were then resuspended in 100 l of buffer A and an equal volume of reaction supernatant (first centrifugation), and resuspended vesicles were analyzed by gel electrophoresis using pre-cast NuPAGE (Invitrogen) gels. Proteins were transferred to the Immobilon-P membrane (Millipore), and GST-BCL-X L ⌬C was detected using a primary mouse anti-glutathione S-transferase antibody (Pharmingen) and an anti-mouse secondary IgG coupled to horseradish peroxidase (GE Healthcare).
Circular Dichroism-Mixtures of His-BID BH3 peptides (13.4 M) and MOM vesicles (13 nM LUVs, 1 mg/ml) containing 0 or 5% DOGS-NTA-Ni were prepared in buffer A. BID BH3 peptides and their hydrocarbon-stapled derivatives were dissolved in water to a concentration of 25-50 M. CD spectra were obtained in a quartz cell with a 1-mm path length using an Aviv (Lakewood, NJ) CD spectrometer at 20°C. A bandwidth of 1.5 nm was used for all measurements. Signals were averaged for 5 s for every half nanometer measurement between 215 and 260 nm. Signals below 215 nm were not recorded because of light scattering induced by the vesicles. Each spectrum represents the average of three distinct spectral recordings. To correct for the spectral contribution of MOM vesicles, the CD spectra of the vesicles alone were subtracted from those of the peptide/vesicle mixtures. Similarly, the CD spectra of the peptides alone were corrected by subtraction of the solvent blank (buffer A). Peptide concentrations were calibrated by amino acid analysis. The helical content was estimated from the mean residue molar ellipticity value at 222 nm ([] 222 ), according to the following empirical Equation 1 (34,35).
where N is the number of peptide bonds in the helix.
EPR Spectroscopy-EPR spectroscopy was used to investigate potential conformational changes of the spin-labeled peptides upon membrane binding. Peptides bound to MOM vesicles containing 5% DOGS-NTA-Ni were prepared as follows. Peptides (184 M) were incubated with 0.20 M vesicles (15.6 mg/ml) in a final volume of 90 l in 20% Me 2 SO/buffer A for 30 min at room temperature. Membrane-bound peptides were further washed with 100 l of 20% Me 2 SO/buffer A by two successive centrifugations (30 min, 100,000 rpm) using an Airfuge (Beckman). Pelleted vesicles were resuspended in 5 l of 20% Me 2 SO/buffer A. For low temperature measurements (at 232 K), the pelleted vesicles were resuspended in 100 l of buffer A containing 16% (v/v) glycerol and pelleted by centrifugation. The resulting pellets were resuspended in 5 l of buffer Membrane-targeted BID BH3 Peptide Activates BAX DECEMBER 1, 2006 • VOLUME 281 • NUMBER 48

JOURNAL OF BIOLOGICAL CHEMISTRY 37001
A, 16% glycerol. The mixtures of peptides (184 M) and 0.20 M MOM vesicles (15.6 mg/ml) without DOGS-NTA-Ni were also prepared in 20% Me 2 SO/buffer A and used without further treatment. The peptides bound to Ni-NTA-agarose resins (Qiagen) were prepared by mixing 50 l of 50% (v/v) Ni-NTAagarose slurry pre-equilibrated in buffer A with peptides (final concentration 110 ⌴) to a final volume of 90 l 20% Me 2 SO/ buffer A. After incubation for 5 min at 37°C, the gels were spun down by centrifugation for 5 min at 4,000 rpm on a tabletop Beckman microcentrifuge. The resulting gel pellet was washed five times by resuspension and pelleting in 100 l of 20% Me 2 SO/buffer A. The resulting pellet was resuspended in 20% Me 2 SO/buffer A in a volume of 50 l. For low temperature experiments, the peptide/gel resuspension was prepared similarly and washed three times with 20% Me 2 SO/buffer A, followed by two washes with 3ϫ volumes of 16% glycerol in buffer A. The samples were pipetted into capillary tubes (Vitro Com Inc, NJ; inner diameter ϭ 0.5 mm) and further compacted after sealing with wax. EPR spectra were obtained on a Bruker EMX spectrometer using a Bruker high sensitivity resonator. Spectra were recorded at a 1-2-milliwatt incident microwave power using a field modulation of 1.0 -2.0 G at 100 kHz at room temperature. To determine the number of spin labels attached to peptides, EPR spectra were taken after liberating the spin labels from the protein molecules by incubating the labeled proteins with 50 mM tris-(2-carboxyethyl)phosphine (Molecular Probes, Inc.). The amount of spin label was calculated by double integration of the EPR spectra using 3-carboxyproxyl (Sigma) as a standard. The concentrations of the spin-labeled peptides were determined by amino acid analysis. The distance between two R1 residues were estimated by the deconvolution method as described (36,37). Briefly, the spectra of singly labeled samples and the spectrum of the doubly labeled samples were obtained for a pair of R1 residues at 232 K (Ϫ41°C). EPR spectra were normalized to the same area by double integration after baseline corrections. Depending on the magnitude of the line broadening, various scan widths such as 100, 150, or 250 G were used to avoid the base-line distortions of the spectra. Using the normalized spectrum of the doubly labeled sample and the sum of the two normalized spectra of the singly labeled samples, the distance distributions between the paired nitroxide residues were calculated by the method developed by Altenbach et al. (36) and Rabenstein et al. (37).

Mitochondrial and Liposomal Release Assays Similarly
Reflect the Specificity of tBID-induced BAX Activation-To validate the liposomal release assay as a high fidelity experimental correlate to the mitochondrion, we probed the specificity of tBID-induced BAX activation in both assay systems using a comprehensive panel of tBID, BCL-X L , and BAX mutants (supplemental Figs. S3-S6). In mitochondrial and liposomal assays, wild type tBID and tBID 97A98A, tBID 94A, tBID(93-96)A and tBID 90A95A similarly demonstrated the differential activities of tBID constructs in triggering BAX pore formation (supplemental Fig. S4). Wild type tBID displayed the highest BAX activation activity in both systems, whereas mutant tBID 90A95A was completely ineffective at submicromolar concentrations (supplemental Fig. S4, A and C). The specificity of tBID-induced BAX activation was similarly confirmed in both assay systems using BCL-X L blockade (supplemental Fig. S5). BCL-X L inhibition experiments using mitochondria demonstrated that wild type tBID and tBID94A maintained high affinity BCL-X L binding, whereas 97A98A and (93-96)A mutagenesis abrogated BCL-X L binding (supplemental Fig. S5A). Strikingly similar findings were observed when BCL-X L blockade experiments were performed using the liposomal system (supplemental Fig. S5, B-E). As a further measure of the specificity of tBID-induced BAX activation, we evaluated BAX mutants in both in vitro assays (supplemental Fig. S6). In the mitochondrial assay, only wild type BAX in combination with tBID induced cytochrome c release (supplemental Fig. S6, A  and B). Consistent with a previous report (38), a BAX mutant missing a critical helical hairpin structure (BAX⌬H56) was unable to induce cytochrome c release. BAXG108V, which lacks the ability to oligomerize (39), also showed no significant activity, either alone or in combination with tBID. Consistent with the mitochondrial data, the liposomal assays demonstrated dextran release only in response to co-incubation with tBID and wild type BAX but not with either of the two BAX mutants (supplemental Fig. S6, C and D).
BID BH3 Peptide Alone Does Not Activate BAX Efficiently-The explicit correlation between mitochondrial and liposomal release data confirm that tBID-induced BAX activation can be reconstituted with three essential components, BAX, tBID, and membrane. We employed this validated experimental system to further investigate the functional roles of tBID subdomains in direct BAX activation. Contrary to tBID, submicromolar concentration of BID BH3 peptide did not trigger BAX activation (13,29) (see also supplemental Fig. S7). Furthermore, the BID BH3 peptide was unable to complement the BH3 defect in mutant tBID L90A/D95A (supplemental Fig. S7). These results suggested that the isolated BID BH3 domain is an inefficient ligand for BAX activation unless maintained in the proper context via collaboration with other domains of tBID. For example, in the absence of membrane targeting capability (17,19,29,40), a synthetic BH3 peptide may not trigger BAX activation at physiologic dosing. We thus adapted the liposomal system to facilitate artificial membrane targeting of BH3 peptides.
Hexahistidine-tagged BH3 Peptides Target to Artificial Membranes Decorated with Ni 2ϩ -NTA Moieties and Specifically Bind BCL-X L -We employed the polyhisitidine-Ni 2ϩ -NTA affinity system to target histidine-tagged peptides to the membranes derivatized with the nickel chelating lipid, DOGS-NTA-Ni (41). When histidine-tagged BID BH3 peptide was preincubated with liposomes containing increasing amounts of DOGS-NTA-Ni, GST-BCL-X L ⌬C partitioned with the liposomal pellet in a dose-dependent manner (Fig. 1A). BID BH3 peptide lacking the hexahistidine tag did not induce liposomal GST-BCL-X L ⌬C binding (Fig. 1B, lanes 5 and 6). Addition of EDTA, which strips Ni 2ϩ ion from the NTA moieties of DOGS-NTA-Ni, abrogated GST-BCL-X L ⌬C partitioning with the liposomal pellet, confirming the mechanism of His-BID BH3based targeting of GST-BCL-X L ⌬C to the liposomes (Fig. 1B,  lanes 7 and 8). BAD BH3 peptide, which also binds GST-BCL-X L ⌬C, effectively competes with His-tagged BID BH3, enabling GST-BCL-X L ⌬C to distribute to both the pellet and supernatant (Fig. 1B, lanes 9 and 10). In addition, an L90A/D95A mutant version of His-BID BH3 significantly reduced GST-BCL-X L ⌬C binding to the liposomes, confirming the specificity of His-BID BH3 activity (Fig. 1C). These results demonstrated that the hexahistidine N-terminal tag facilitated BID BH3 targeting to the DOGS-NTA-Ni liposomes through Ni 2ϩ -hexahistidine complex formation and that the membrane-tethered BID BH3 peptide could then selectively bind GST-BCL-X L ⌬C.
Membrane-targeted BID BH3 Peptide Efficiently Activates BAX-To assess the role of membrane targeting on BID BH3 function, we compared the effects of submicromolar concentrations of tBID, BID BH3 peptide, and His-tagged BID BH3 peptide on BAX-induced FITC-dextran release using the Ni 2ϩ -NTA-derivatized liposomal system ( Fig. 2 and supplemental  Fig. S8). In combination with 10 nM BAX, His-tagged BID BH3 achieved near-maximal liposomal release at a concentration of only 50 nM ( Fig. 2A, red trace), exhibiting strikingly similar release potency to tBID ( Fig. 2A, black trace) except for delayed release kinetics. The corresponding His-tagged L90A/D95A mutant peptide had markedly diminished activity ( Fig. 2A, blue  trace), and the untagged BID BH3 peptides had no effect at the same concentration ( Fig. 2A, purple trace). None of the reagents alone caused membrane permeabilization (Fig. 2B), confirming that release activity was triggered by the combined action of His-tagged BID BH3 and BAX, as likewise demonstrated for native tBID and BAX. The activity of His-tagged BH3 peptide is abrogated by EDTA treatment, which chelates Ni 2ϩ ion from the NTA groups ( Fig. 2A, green trace), thereby confirming that membrane targeting of the BH3 peptide is essential for BAX activation at nanomolar concentrations. The integrity of the DOGS-NTA-Ni-derivatized liposomal system was confirmed by the observation that, in contrast to Histagged BID BH3, the stimulatory effect of tBID on BAX-induced liposomal release occurred with or without added DOGS-NTA-Ni (Fig. 2D). As a further measure of specificity, we found that GST-BCL-X L ⌬C treatment completely abrogated His-tagged BID BH3 activity (Fig. 3A, cyan trace), and this inhibition was reversed by the addition of BAD BH3, which in turn blocks the effect of GST-BCL-X L ⌬C (Fig. 3A, green trace).  DECEMBER 1, 2006 • VOLUME 281 • NUMBER 48 BAD BH3 itself had no effect on His-tagged BID BH3-mediated BAX activation ( Fig. 2A, red trace), and when administered alone (Fig. 3B, black trace), or in combination with other reagents (Fig. 3, C and D), it likewise had no membrane permeabilization effects. Finally, we explored the potency of His-tagged BID BH3 in triggering BAX activation (supplemental Fig. S8). As reported previously (13), we find that BID BH3 triggers BAX activation at high micromolar concentrations (supplemental Fig. S8, A and  C). In contrast, His-BID BH3 peptide demonstrated a dose-dependent increase in BAX activation within the nanomolar range (supplemental Fig. S8, A and B). At a saturating concentration of His-BID BH3, BAX also showed a dose-dependent increase in FITC-dextran release activity (supplemental Fig. S8,  D and E). We find that the degree of BAX-induced release triggered by 5-10 M BID BH3 peptide can be achieved with only 2.5-50 nM His-BID BH3 peptide. Thus, the activity of BID BH3 is increased by 2-3 orders of magnitude as a direct result of membrane targeting.

␣-Helicity of Membrane-targeted BID BH3 Increases Upon
Membrane Binding-To investigate the mechanism underlying enhanced BAX activation by a membrane-targeted BID BH3 peptide, we probed the conformation of His-tagged BH3 peptides in the presence and absence of liposomal membrane using CD and EPR spectroscopy. The CD spectrum of the His-tagged BID BH3 peptide in the presence of MOM vesicles containing 5% DOGS-NTA-Ni (open circles) shows a dramatic increase in [] compared with the spectrum measured in the absence of membrane (filled circles) (Fig. 4A). The mean residue molar ellipticity at 222 nm, [] 222 , a measure of ␣-helicity, more than doubled upon exposure to the vesicles, reflecting an increase in percent ␣-helicity from 12 to 29% (34,35). Of note, this induction of ␣-helical structure was not observed when the peptide was mixed with MOM vesicles that lacked DOGS-NTA-Ni (Fig. 4B), confirming that the increase in ␣-helicity resulted from membrane targeting. The EPR spectra of the R1 residues labeled at position 92, 96, or both are consistent with the CD data (Fig. 5). Both in solution (Fig. 5B, 1st row) and in the presence of MOM vesicles without DOGS-NTA-Ni (Fig. 5B, 2nd  row), the EPR spectra of His-BID 92R1 and His-BID 96R1 show extremely sharp lines (at the corresponding positions indicated by the open arrowhead), indicating that the peptides were tumbling rapidly without motional restrictions. When the peptides were bound to membranes containing 5% DOGS-NTA-Ni, a new motionally restricted spin population appeared, as indicated by the filled arrowheads (Fig. 5B, 3rd row). The line shapes are consistent with R1 residues positioned on ␣-helical surface (42). Anchoring the peptides to the Ni-NTA-agarose resin resulted in line broadening compared with the peptides in solution, but the motionally restricted spin populations did not appear. These data confirm that membrane targeting of His-BID BH3 peptide resulted in enhanced ␣-helix formation, causing motional restriction of the spin labels at positions 92 and 96. The inter-residue distance between the two R1 residues would be expected to decrease upon ␣-helix formation, resulting in stronger spin-spin interactions. Indeed, strong spin-spin interactions are observed for the pair of spin labels in solution alone (Fig. 5B, top spectrum on the 3rd column), but even stronger spin-spin interactions are observed in the line shapes for samples in MOM vesicles with 5% DOGS-NTA-Ni (Fig. 5B, 3rd  spectrum from the top on the 3rd column). To compare the strength of the spin-spin interactions more accurately without motional contributions to the line shape, we froze the samples bound to the Ni-NTA moiety and obtained powder spectra (Fig. 5C). Compared with the spectra of the 92R1/96R1 pair bound to the Ni-NTA-agarose resin (red trace), the R1 pair in the MOM vesicles containing 5% DOGS-NTA-Ni (green trace) showed broader line width and decreased amplitude of the EPR spectrum (Fig. 5C, see the overlaid spectra, bottom), classic features of strong spin-spin interactions at close proximity (less than 10 Å) (36,37). The estimated distance between the 92R1 and 96R1 pair in the Ni-NTA-agarosebound His-BID BH3 peptide is greater than 10 Å (10 -15 Å) (36,37). These results demonstrate that the inter-residue distance between the i (92R1) and i ϩ 4 (96R1) positions decreases upon membrane binding, consistent with a random coil to ␣-helix transition upon membrane binding. To assess the relative contributions of BH3 ␣-helicity and membrane targeting to BAX activation, we compared the effect of an untargeted but chemically reinforced BID BH3 helix (32,43) to that of the His-targeted BID BH3 peptide, which was not structurally stabilized (supplemental Fig.  S9). A stabilized ␣-helix of BCL-2 domain (SAHB) corresponding to the murine BID BH3 domain (supplemental Fig. S9A) exhibited 85% ␣-helicity in water (supplemental Fig. S9B), and at 75 nM triggered 25% liposomal release in the presence of 7.5 nM BAX (supplemental Fig. S9C). As a measure of specificity, murine BID SAHB alone has no effect on the liposomes (43), and a double mutant BID SAHB with 65% ␣-helicity caused no BAX-induced release (supplemental Fig. S9, B and C). Although BID BH3 peptide with 28% ␣-helicity had no effect on BAX, the Histagged BID BH3 peptide (19% ␣-helicity in aqueous solution, 29% ␣-helicity upon membrane binding) triggered 50% BAX-induced liposomal release (supplemental Fig. S9C). Thus, helical reinforcement alone facilitates BAX-induced liposomal release (32), yet a membrane-targeted BID BH3 with at least 3-fold lower helical content triggers double the BAX-induced liposomal release of murine BID SAHB. Taken together, the CD, EPR, and comparative liposomal release data highlight the importance of both ␣-helicity and the membrane targeting functionality itself in enabling potent BAX activation by His-tagged BID BH3.

DISCUSSION
We document the close correlation between mitochondrial and liposomal assay systems in measuring tBID-induced BAX activation, and thereby further validate that reconstitution of the proapoptotic machinery in artificial membranes is a physiologically relevant approach for mechanistic studies of BAX activation. We first examined tBID activity using both   DECEMBER 1, 2006 • VOLUME 281 • NUMBER 48 in vitro assays (supplemental Figs. S4 -S6) to evaluate and confirm previously defined roles of specific BID BH3 residues in BCL-X L binding and BAX activation (5,8,20,44). For example, mutant tBID M97A/D98A (previously designated as BID mIII-1 by Wang et al. (5)) triggered BAX activation that was not inhibited by BCL-X L (supplemental Figs. S4, A and C, and S5, A and C), underscoring the requirement of M97/D98 for BCL-X L binding but not for BAX activation. Conversely, a tBID G94A mutant (designated as BID mIII-3 mutant (5)) bound to BCL-X L (supplemental Fig. S5, A and E) but lost significant BAX activating capability (supplemental Fig. S4, A and C), indicating that residue Gly-94 is important for BAX activation but not required for BCL-X L binding. Mutant tBID (93-96)A (designated BID mIII-2 (5)) exhibited both compromised BCL-X L binding (supplemental Fig. S5, A and D) and markedly reduced BAX activation (supplemental Fig. S4, A and C). Mutant tBID 90A95A displayed no BAX activation in either assay (supplemental Fig.  S4, A and C), confirming the critical roles of Leu-90 and Asp-95. These results suggest that tBID-induced BAX activation is indeed dependent upon the BH3 domain of tBID, and therefore, a BH3-independent mechanism need not be invoked (14). The impact of BAX ⌬H56 and BAX G108V mutations in abrogating BAX activity is also consistent between liposomal and mitochondrial assays systems and confirms previous observations (supplemental Fig. S6) (38,39). For example, we find that BAX G108V, identified in Burkitt's lymphoma (39), is incapable of pore formation in either in vitro assay. This is consistent with the inability of BAX G108V to homodimerize as measured by yeast two-hybrid assay (39). Thus, we find that BID BH3-dependent BAX activation can be effectively reconstituted in an artificial membrane system that closely reflects the in vitro mitochondrial release assay.

Membrane-targeted BID BH3 Peptide Activates BAX
tBID engages the mitochondrial membrane in vivo via critical interactions between its H4 -H6 helical bundle and membrane lipids, including cardiolipin (17,45). Helices 6 and 7 have been reported to be the minimal structural subunit required for the mitochondrial interaction of tBID (40). Our recent site-directed spin labeling study demonstrated that membrane association of tBID helices 6 -8 triggers its structural reorganization and exposure of the BH3 domain (29), consistent with the model proposed by van Mau et al. (46). N-terminal myristoylation of tBID further facilitates its targeting to the mitochondrial membrane (16). More recently, tBID has been reported to form a complex with the mitochondrial homolog 2 (Mtch2) protein, which localizes to the mitochondrial membrane (47). We sought to determine and simulate the essential components within tBID required for in vitro BAX activation at physiologic dosing. Although a synthetic BID BH3 peptide is unable to trigger BAX activation at submicromolar dosing, we demonstrate that membrane targeting of BID BH3 via a liposomal hexahistidine Ni 2ϩ -NTA affinity system dramatically improves its BAX activation capability (Fig. 6A). Liposomal targeted BID BH3 peptide triggered up to 90% of BAXinduced FITC-dextran release after 1-2 h of incubation, recapitulating nearly the full activity of native tBID-induced BAX activation at similar dosing (Figs. 2 and supplemental Fig. 8). Thus, tBID activity can be effectively reconstituted by combining its pro-apoptotic BH3 death domain with a membrane targeting module. These in vitro data highlight the essential roles of BID BH3 and the membrane interaction domain of tBID in activating BAX (Fig. 6B).
Interestingly, the His-tagged peptides demonstrate somewhat delayed responses in triggering BAX-induced FITC-dextran release compared with tBID ( Fig. 2A). For example, the time required to achieve 50% release is 17 min longer for the targeted peptide than for tBID. Whereas the maximal rate of release induced by tBID is ϳ0.2%/s at ϳ450 s, the corresponding value for His-tagged BID BH3 is ϳ0.05%/s at ϳ1000 s. Thus, FIGURE 6. Schematic representation of BID BH3-induced BAX activation. A, N-terminal hexahistidinetagged BID BH3 domain (blue sphere) binds to the Ni-NTA moiety (pink sphere) on the surface of the DOGS-NTA-Ni containing lipid vesicles, resulting in membrane targeting of BID BH3 and induction of ␣-helicity. BAX is recruited and activated by the juxta-membrane, ␣-helical BH3 domain, triggering BAX oligomerization and FITC-dextran (green balls) release through putative pores. B, the membrane interaction module of tBID, consisting of core amphipathic helices (blue cylinders) engages the mitochondrial membrane, facilitated by targeting properties of the N-terminal myristoyl moiety. The native orientation of tBID at the mitochondrial membrane optimally activates BAX via BH3 domain engagement, resulting in BAX oligomerization and mitochondrial cytochrome c (red balls) release.
the maximal rate of release triggered by tBID is 4-fold greater than that of His-tagged BID BH3 under the same conditions. By binding mitochondrial membrane lipids or other protein components (12,14,47), tBID is likely more effective than the synthetic peptide at orienting within the membrane to achieve optimal presentation of the BH3 domain for BAX activation. Despite the kinetic difference, membrane tethering increased the efficacy of BID BH3 peptide by 2-3 orders of magnitude compared with the untagged peptide, resulting in BAX activation at nanomolar potency (Figs. 2 and supplemental Fig. S8).
The marked enhancement of BAX activation by membranetethered BID BH3 peptide may be attributed to the following: (i) increased surface density of the BH3 peptide, (ii) induction of peptide ␣-helicity by the membrane surface with resultant increase in biological potency, and (iii) lowered activation barrier of BAX conformational change and oligomerization because of either facilitated recruitment of BAX by the membrane-tethered BH3 domain or catalytic action of the tethered BH3 domain on membrane-resident Bax. Because of the extremely high affinity of the hexahistidine tag to NTA-Ni (K d ϭ (1-5) ϫ 10 Ϫ9 ; Qiagen), the binding of His-tagged peptide (50 nM) to DOGS-NTA-Ni (ϳ4.75 M in Figs. 2 and 3) is expected to be nearly complete at equilibrium, thereby increasing the local concentration of peptide on the membrane surface. In spherical liposomes of 100 nm in diameter (supplemental Fig. S1), we estimate that the equilibrium concentration of the peptide tethered to the liposome surface within a 50-Å-thick shell would be ϳ2.3 mM, which is ϳ4.6 ϫ 10 5 -fold more condensed than the total bulk concentration of 50 nM in the presence of ϳ0.13 nM liposomes (see supplemental Fig. S1B), the conditions used in Figs. 2 and 3. We expect that such a large increase in local peptide concentration near the membrane surface would enhance the probability of BID BH3-BAX interactions, thereby accelerating the rate of Bax activation and liposomal release of FITC-dextran.
Our CD and EPR data confirm that membrane tethering of the BID BH3 peptide also induces ␣-helix formation, suggesting that the observed enhancement of activity is at least partially attributed to reinforcement of the bioactive secondary structure of BID BH3 at the membrane surface. Indeed, chemical reinforcement of BID BH3 ␣-helicity by hydrocarbon stapling (32) independently improves the activity of BID BH3 peptide-induced BAX activation (supplemental Fig. S9) (43). However, the His-tagged BID BH3 peptide, which exhibits markedly lower ␣-helicity in solution compared with BID SAHB, exhibits comparatively higher potency in triggering BAX-induced liposomal release. Thus, the enhanced activity of His-tagged BID BH3 in BAX activation compared with unmodified BID BH3 derives from both membrane surface enrichment and increased ␣-helicity derived from membrane partitioning-induced folding (48).
BAX activation is a tightly regulated, multistep process in which the sequence of BAX derepression, conformational change, mitochondrial membrane association, tBID interaction, membrane insertion, and pore formation remains to be fully clarified. Recent work has explored the biochemical and biophysical prerequisites for BAX-tBID-membrane interactions. For example, based on the surface pressure increase that results from adsorption of tBID or BAX onto a lipidic monolayer resembling the composition of mitochondrial membrane contact sites (POPC/POPE/cardiolipin 1:1:1, mol/mol), van Mau et al. (46) observed that interaction of the monolayer with tBID was 12-fold stronger than that with BAX. Although tBID interacted weakly with a 1,2-dioleoylsn-glycero-3-phosphocholine monolayer that simulates the outer leaflet of the mitochondrial outer membrane, BAX showed no interaction, suggesting that BAX does not spontaneously insert into the mitochondrial membrane. Thus, the authors propose that membrane-inserted tBID initiates BAX insertion into the membrane (46). However, Yethon et al. (49) indicate that the lipid-induced BAX conformational change, as measured by the conformation-specific antibody 6A7 (50), is a prerequisite for the tBID-BAX interaction. Upon incubation with lipid vesicles resembling the mitochondrial outer membrane, BAX was shown to undergo a reversible conformational change in the absence of tBID. Although tBID did not induce a BAX conformational change in the absence of liposomes, the liposome-induced BAX conformer did not insert into membranes, become oligomeric, or form pores unless tBID was present (49). We find that a juxtamembrane BID BH3 peptide is almost as effective as tBID in triggering the functional activation of BAX. Our findings emphasize the essential roles of the BID BH3 domain and lipid exposure in optimally triggering BAX-induced mitochondrial and liposomal release. Indeed, BID BH3-induced BAX activation may be a catalytic process, in which activated BAX propagates its pro-apoptotic activity by subsequent BAX BH3-mediated auto-activation, as recently described by Tan et al. (51).
Because chemical manipulation of BAX-mediated pro-apoptotic activity may have significant therapeutic consequences, the molecular mechanism of BAX activation continues to be an active area of investigation. Small molecules and peptidic compounds have been developed to target BCL-2 family proteins in vivo in an effort to reactivate mitochondrial apoptosis in human disease, such as cancer (32,(52)(53)(54)(55)(56)(57)(58). Our results suggest that synthetic efforts to target such compounds to the mitochondria may enhance therapeutic potency by increasing their effective concentration at the mitochondrial membrane and optimizing molecular presentation of the simulated BH3 interface to multidomain BCL-2 family proteins.