tBID Homooligomerizes in the Mitochondrial Membrane to Induce Apoptosis*

Activation of the tumor necrosis factor R1/Fas receptor results in the cleavage of cytosolic BID to truncated tBID. tBID translocates to the mitochondria to induce the oligomerization of BAX or BAK, resulting in the release of cytochromec (Cyt c). Here we demonstrate that in tumor necrosis factor α-activated FL5.12 cells, tBID becomes part of a 45-kDa cross-linkable mitochondrial complex that does not include BAX or BAK. Using fluorescence resonance energy transfer analysis and co-immunoprecipitation, we demonstrate that tBID-tBID interactions occur in the mitochondria of living cells. Cross-linking experiments using a tBID-GST chimera indicated that tBID forms homotrimers in the mitochondrial membrane. To test the functional consequence of tBID oligomerization, we expressed a chimeric FKBP-tBID molecule. Enforced dimerization of FKBP-tBID by the bivalent ligand FK1012 resulted in Cytc release, caspase activation, and apoptosis. Surprisingly, enforced dimerization of tBID did not result in the dimerization of either BAX or BAK. Moreover, a tBID BH3 mutant (G94E), which does not interact with or induce the dimerization of either BAX or BAK, formed the 45-kDa complex and induced both Cyt c release and apoptosis. Thus, tBID oligomerization may represent an alternative mechanism for inducing mitochondrial dysfunction and apoptosis.

The BCL-2 family members are major regulators of the apoptotic process. The cell death regulatory activity of these molecules is unknown, although it is thought that their function depends mostly on their ability to modulate mitochondrial function (1). This family is comprised of both pro-apoptotic (e.g. BAX and BAK), as well as anti-apoptotic (e.g. BCL-2 and BCL-X L ) molecules. Most family members share homology in three domains (BH1-3; "multidomain" members) and carry a C-terminal membrane-anchoring domain. The BH3 domain functions as a death domain in the pro-apoptotics. In a subset of the pro-apoptotics are the BH3-only molecules (e.g. BID and BIM).
The BCL-2 family members have a strong tendency to form homodimers as well as heterodimers with each other. Using protein cross-linkers it has been demonstrated that BAX and BAK homodimerize in the mitochondrial membrane during apoptosis (2)(3)(4). Moreover, enforced dimerization of an FKBP-BAX chimera by the bivalent ligand FK1012 induces apoptosis (2). In addition, using fluorescence resonance energy transfer (FRET) 1 analysis, BCL-2 and BAX have been demonstrated to interact with each other in the mitochondrial membrane (5).
The three-dimensional structures of several family members show similarities to the structure of the pore-forming region of bacterial toxins (6). Based on this structural similarity and on in vitro electrophysiological data (7), it is suspected that BCL-2 family members may function as pore-forming proteins. Their tendency to oligomerize and their prominent subcellular location at intracellular membranes may add further support to the "pore hypothesis." Caspases are the major executioners of the apoptotic process (8). The BH3-only molecule BID serves as one of the links between BCL-2 family members and caspases. Following TNF-R1/Fas receptor activation, cytosolic p22 full-length BID is cleaved by caspase-8 (9 -11). The truncated product, p15 tBID, translocates to the mitochondria to induce the release of cytochrome c (Cyt c). Cyt c activates Apaf-1, which in turn activates caspase-9 (12). Most importantly, BID is an essential component of the TNF-R1/Fas-death receptor pathway in hepatocytes, because Bid-deficient mice are resistant to the lethal effect of anti-Fas antibody injection (13).
Full-length BID was initially identified as the only BCL-2 family member that can bind both pro-apoptotic (e.g. BAX) as well as anti-apoptotic (e.g. BCL-2) molecules (14). It was also demonstrated that full-length BID does not homodimerize. The identification of the binding partners led to the assumption that BID acts as a "death ligand": it receives a death signal in the cytosol and translocates to the mitochondria to transfer the signal to "death receptors" (e.g. BAX or BCL-2). Recently, it was indeed demonstrated that following its cleavage and translocation to the mitochondria, tBID induces the oligomerization of BAK or BAX, which results in Cyt c release (3,4). Strikingly, murine embryonic fibroblasts lacking both BAX and BAK are resistant to tBID-induced apoptosis (15).
Here we examine the binding partners of tBID in the mitochondrial membrane following its translocation. We note that following a TNF␣ death signal, tBID becomes part of a 45-kDa cross-linkable complex in the membrane, which most likely represents a tBID homotrimer. Using a chimeric FKBP-tBID protein, we demonstrate that enforced dimerization of tBID is sufficient to induce Cyt c release, caspase activation, and apoptosis. However, these tBID dimers did not induce the dimerization of BAX or BAK. Moreover, a tBID BH3 mutant formed trimers and induced apoptosis without inducing the dimerization of BAX or BAK. Thus, tBID oligomers may act in an alternative pathway to induce mitochondrial dysfunction and apoptosis.

EXPERIMENTAL PROCEDURES
TNF␣/CHX Treatment and Subcellular Fractionation-FL5. 12, an interleukin-3-dependent murine early hematopoietic cell line, was maintained in 10% fetal bovine serum supplemented with 10% WEHI-3B conditioned medium as a source of interleukin-3. FL5.12 cells were treated with recombinant mouse TNF␣ (40 ng/ml; Sigma) and cycloheximide (1 g/ml; Sigma) for 6 h, suspended in isotonic HIM buffer (200 mM mannitol, 70 mM sucrose, 1 mM EGTA, 10 mM HEPES, pH 7.5) and homogenized either using a polytron homogenizer (Brinkmann Instruments) at setting 6.5 for 10 s or by passing the cells 20 times through a 25 G (0.5 ϫ 16) needle. Nuclei and unbroken cells were removed by centrifugation at 120 ϫ g for 5 min. The supernatant was centrifuged at 10,000 ϫ g for 10 min to collect the mitochondriaenriched fraction (mitochondria) and the supernatant (cytosol).
Transient Transfection System-293T, an embryonic kidney cell line, HeLa, a human cervical adenocarcinoma cell line, and Cos-7, a simian kidney cell line, were maintained in 10% fetal bovine serum. Transient transfections were performed by the calcium phosphate method (17) or with LipofectAMINE 2000 (Invitrogen). For the FKBP/FK1012 experiments, the cells were treated with 0.3 M FK1012E/Z alone (1 mM stock solution in 50% ethanol, 50% dimethyl sulfoxide) or in combination with 1 M FK506. FK1012E/Z and FK506 were generous gifts from P. Clemons and S. L. Schreiber (Harvard University).
For propidium iodide staining experiments, the cells were allowed to grow to ϳ60% confluence in 10-cm plates before transfection. The cells were transfected with 15 g of each of the indicated plasmids. Twentyfour hours post-transfection, the cells were collected, washed once in phosphate-buffered saline (PBS), and fixed with methanol at Ϫ20°C. The cells were recovered by centrifugation at 1,000 ϫ g for 5 min, washed once in PBS, and incubated in PBS containing 25 g/ml propidium iodide and 50 g/ml RNase A. The percentage of cells displaying a sub-G 1 DNA content was determined by FACScan (Becton Dickinson) analysis.
Cross-linking-Sulfo-BSOCOES, BS 3 , and BMH (in dimethyl sulfoxide; Pierce) were added from a 10-fold stock solution to a final concentration of 10 mM. The cross-linker was added to either the cytosolic/ soluble fraction or to the mitochondrial-enriched fraction and suspended in isotonic HIM buffer. After incubation for 30 min at room temperature, the cross-linker was quenched by the addition of 1 M Tris-HCl, pH 7.5, to a final concentration of 20 mM. After quenching, the samples were lysed and Western blot analyzed with the indicated antibody.
Expression Plasmids-Wild type p15 tBID and tBID mIII-4 were amplified by PCR from wild type p22 BID (14). Human FKBP12 (18) was ligated in-frame to the N terminus of the tBID PCR product in pcDNA3 (Invitrogen) under the cytomegalovirus immediate early promoter to create the pFKBP-tBID plasmid. pcDNA3 with a HA or a FLAG epitope tag was ligated in-frame to the tBID PCR product under the cytomegalovirus immediate early promoter to create the HA-tBID and FLAG-tBID plasmids. To create the tBID-CFP and tBID-YFP chimeras, the tBID PCR product was ligated in-frame to the C terminus of ECFP or EYFP in the pECFP or pEYFP plasmids, respectively (CLON-TECH). To create the tBID-GST chimera, the tBID PCR product was cloned into a modified pEF-BOS vector (19) in frame with the GST peptide derived from the procaryotic pGEX vector (originally from Amersham Biosciences, Inc.). 2 To create the HA-BAX plasmid, pcDNA3 with a HA epitope tag was ligated in-frame to the BAX product under the cytomegalovirus immediate early promoter.
FRET Measurements-The cells were transiently transfected by the calcium-phosphate method. Twelve hours post-transfection, the cells were treated briefly with trypsin and seeded into eight-chamber glass slides (Lab-Tek). The cells were then allowed to settle on the slide for 5 h. Single cells were excited at 430 nm using Polychrome II (TILL Photonics, Martinsried, Germany) via an inverted microscope equipped with 40 ϫ 1.3 N.A objective, long pass dichroic mirror (455DRLP) and emission filters (450EFLP) (Omega Optical, Brattleboro, VT). An emitted fluorescence signal was passed through a MS127i spectrograph and collected using cooled, 16-bit back-illuminated CCD (Instaspec IV, Oriel, Stratford, CT) directly attached to the microscope side port (21-ms exposure with 16-s read-out times/pixel). For visualizing spectral information, dark noise was first subtracted from the emission signal, and then the spectral counts were normalized to the spectral intensity (under the curve of 1024 spectral points) of cells expressing tBIDϪCFPϩtBIDϪYFP. The average ratios of emission were calculated by first subtracting the dark noise (average of 258 counts/collection time) from the signal and then dividing the signal at 535 nm (YFP emission) by the signal at 490 nm (CFP emission).
Immunocytochemistry and Confocal Microscopy-For immunocytochemistry, the cells were grown on fibronectin-coated glass coverslips. Twelve hours post-transfection the cells were fixed with 3% paraformaldehyde in PBS for 6 min and permeabilized with 0.2% Triton X-100 in PBS for 5 min. For blocking, the cells were incubated in PBS containing 0.1% Triton and 3% BSA for 1 h at room temperature. To immunostain both BID and Cyt c, the cells were incubated overnight at 4°C with anti-BID Ab and anti-Cyt c 6H2.B4 mAb (PharMingen) diluted 1:200 in blocking solution. After three washes with PBS containing 0.1% Triton, the cells were stained for 30 min at room temperature with Cy3-labeled goat anti-mouse (dilution 1:100, Jackson ImmunoResearch) and Alexa 488-labeled goat anti-rabbit Abs (dilution 1:150, Molecular Probes), followed by 5 min of 4Ј,6-diamidino-2-phenylindole dihydrochloride staining (diluted 1:500 in PBS). For staining of mitochondria, the cells were incubated with 100 nM mitotracker red (MTR; Molecular Probes) for 30 min at 37°C prior to fixation. The coverslips were mounted with elvanol, and the cells were viewed under a Nikon fluorescence microscope at a magnification of 400ϫ. The pictures were taken with a 1310 digital camera (DVC). Confocal microscopy was performed using a Zeiss Axiovert 100 TV microscope (Oberkochen, Germany) attached to the Bio-Rad Radiance 2000 laser scanning system and operated by Laser-Sharp software.

BID Becomes Part of a 45-kDa Mitochondrial Complex in
TNF␣-activated FL5.12 Cells-To reveal the mechanism by which BID executes its function, we utilized a series of chemical cross-linkers to identify proteins that closely associate with tBID at the mitochondrial membrane following its translocation. Treatment of FL5.12 hematopoietic cells with TNF␣ in the presence of CHX induces the cleavage of cytosolic full-length BID to generate p15 tBID that translocates to the mitochondria ( It was previously demonstrated that tBID heterodimerizes with BAK or BAX to induce their oligomerization in the mitochondrial membrane (3,4). Therefore we checked whether BAX or BAK were part of the 45-kDa complex. Western blot analysis with anti-BAK antibodies did not detect any new bands (Fig.  1A, lane 9), indicating that BAK is not part of the 45-kDa BID complex. On the other hand, Western blot analysis with anti-BAX antibodies detected two additional BAX-immunoreactive bands slightly below the 46-kDa marker (Fig. 1A, lane 13). To determine whether one of these bands represents a BAX-BID heterodimer, BAX was immunoprecipitated (using the 4D2 monoclonal antibody) from mitochondria following cross-linker treatment, and the immunocomplexes were subjected to Western blot analysis using either anti-BAX (651) or anti-BID polyclonal antibodies. As shown in Fig. 1B, anti-BAX antibodies recognized these two upper bands in the BAX immunoprecipitate (lane 2), but anti-BID antibodies did not (lane 4). Based on these results, it is likely that these bands represent a BAX homodimer and not a BAX-BID heterodimer. Western blot analysis with a variety of other antibodies against BCL-X L , BCL-2, porin/VDAC, adenine nucleotide translocator, and Cyt c indicated that none of these molecules were part of the 45-kDa BID complex (not shown).
The 45-kDa Mitochondrial Complex Appears Immediately following the Mitochondrial Targeting of tBID-To confirm that tBID is part of the 45-kDa complex and to assess the time course of its formation in the mitochondrial membrane, we performed a standard protein import reaction using tBID. 35 S-Labeled in vitro transcribed and translated tBID was incubated with purified intact mouse liver mitochondria, followed by centrifugation to separate the mitochondria from the soluble fraction. Both fractions were treated with the sulfo-BSOCOES cross-linker and analyzed by autoradiography. Ten minutes after adding tBID to mitochondria, a substantial amount of tBID was incorporated into the mitochondrial membrane ( Fig.  2A, lane 4). As soon as tBID appeared in the mitochondria, the 45-kDa cross-linked complex also appeared. Moreover, as more tBID incorporated into the membrane, the intensity of the 45-kDa band increased ( Fig. 2A, lane 6). Western blot analysis with anti-BID antibodies verified that tBID was part of this complex ( Fig. 2A, lanes 8 and 9). The ϳ40-kDa bands that appear in the mitochondrial and soluble fractions do not seem to include tBID because they are not detected with anti-BID antibodies ( Fig. 2A, lane 9). We have also probed the blot with antibodies against BAX, BAK, BCL-X L , BCL-2, porin/VDAC, adenine nucleotide translocator, and Cyt c and found that none of these molecules were part of the mitochondrial 45-kDa complex assembled from 35 S-labeled tBID.
Recombinant tBID Homooligomerizes in Solution-We have performed a similar protein import assay using recombinant tBID (rtBID). rtBID also became part of a 45-kDa protein complex in the mitochondrial membrane following cross-linker  1 and 3). The samples were analyzed by Western blot with anti-BID Ab. treatment (not shown). However, in the absence of both purified mitochondria and cross-linker, rtBID appeared as two bands in SDS gels (ϳ15 and ϳ30 kDa) (Fig. 2B, lane 1). Moreover, in the absence of a reducing agent (dithiothreitol) or without boiling the sample prior to loading, additional ϳ45and ϳ60-kDa bands appeared (Fig. 2B, lanes 2-4). Thus, rtBID forms homooligomers in solution that are sensitive to reducing agents and to high temperature but resistant to SDS. Based on these findings, we analyzed whether cytosolic tBID (from FL5.12 cells) could also oligomerize on SDS gels in the absence of a reducing agent and without prior boiling. In contrast to rtBID, cytosolic tBID did not oligomerize under these conditions (not shown).
tBID-tBID Interactions Occur in the Mitochondrial Membrane but Not in the Cytosol-Based on our previous findings with rtBID, we suspected that the 45-kDa complex might represent a tBID homotrimer. To assess whether tBID-tBID interactions occur in the mitochondrial membrane, we have performed co-immunoprecipitation experiments in cells transiently transfected with tBID tagged with either an HA or a FLAG epitope. It was previously demonstrated that non-ionic detergents (e.g. Nonidet P-40) artificially induce the dimerization of BCL-2 family members, whereas zwitterionic detergents (e.g. CHAPS) do not (20). Therefore, we have used CHAPS in these experiments to extract tBID from the mitochondrial membrane. Mitochondria-enriched fractions purified from transfected cells were incubated in isotonic buffer containing 0.1% CHAPS for 1 h, followed by centrifugation and separation of the mitochondrial pellet from the soluble fraction. FLAG-tBID was immunoprecipitated from the soluble fraction by anti-FLAG antibodies followed by Western blot with either anti-FLAG antibodies (Fig. 3A, left panel) or anti-HA antibodies (right panel). The results indicate that HA-tBID interacts with FLAG-tBID in the mitochondrial membrane. Next, we addressed the question of whether tBID-tBID interactions occur only in the mitochondrial membrane or can also occur in the cytosol. Using differential centrifugation, we have purified the cytosolic and mitochondria-enriched fractions from transfected cells and immunoprecipitated FLAG-tBID from each fraction. As shown in Fig. 3B, the amount of HA-tBID that was co-immunoprecipitated from the mitochondria-enriched frac-tion, was ϳ10-fold higher than the amount co-immunoprecipitated from the cytosolic fraction (right panel). Similar levels of FLAG-tBID were immunoprecipitated from both fractions (Fig.  3B, left panel). Thus, the major site for tBID-tBID interactions is the mitochondrial membrane.
tBID-CFP Interacts with tBID-YFP in Living Cells-We also wanted to test whether tBID-tBID interactions occur in living cells. For this purpose we have used cyan and yellow fluorescence proteins (CFP and YFP) and performed FRET analysis (21). To perform these experiments with tBID, we constructed chimeras of tBID with either CFP (tBID-CFP) or YFP (tBID-YFP), and transfected cells with both tBID-CFP and tBID-YFP, only tBID-CFP, only tBID-YFP, or both unfused CFP and YFP as a control. Western blot analysis of whole cell lysates with anti-BID antibodies indicated that both fusion proteins were expressed at the expected sizes (not shown). To assess the cellular localization of both chimeras, transfected cells were incubated with MTR (to label mitochondria) and were analyzed by confocal microscopy. These studies demonstrated that a major part of both chimeric molecules co-localized with MTR, suggesting mitochondrial localization (Fig. 4A). To assess whether both chimeric proteins were inducing apoptosis, we measured the percentage of cells displaying a sub-G 1 DNA content. Both chimeras were found to be fully functional, because transfection with either of the chimeras induced apoptosis to levels that were similar to the levels induced with wild type tBID (Fig. 4B). Next, we performed FRET analysis on cells 14 h post-transfection. Single intact cells were excited at 430 nm, and the emission spectrum between 450 and 700 nm was recorded. As expected, cells transfected with only tBID-CFP showed an emission peak at 490 nm, whereas cells transfected with only tBID-YFP showed an emission peak at 535 nm (Fig.  4C). A significant difference in the emission ratio (535/495 nm) was seen between cells co-transfected with both tBID-CFP and tBID-YFP compared with cells co-transfected with CFP and YFP, with emission ratios of 1.39 Ϯ 0.04 (n ϭ 20) and 0.98 Ϯ 0.04 (n ϭ 17) (p Ͻ 0.005), respectively (Fig. 4D). These results indicate that there is a significant FRET between the tBID-CFP and tBID-YFP in living cells, suggesting physical interaction between the tBID molecules.
tBID Forms Homotrimers in the Mitochondrial Membrane-The results presented in Figs. 3 and 4 indicate that tBID-tBID interactions occur in mitochondria of living cells. To assess whether tBID forms homotrimers in mitochondria, we have constructed a tBID-human GST chimera and performed crosslinking experiments. We have transiently transfected tBID, tBID-GST, or unfused GST into cells. Mitochondria-enriched fractions prepared from these cells were treated with crosslinker followed by Western blot analysis with anti-BID or anti-GST antibodies. Cross-linking treatment of mitochondria prepared from cells transfected with tBID resulted in the appearance of the 45-kDa complex (Fig. 5A, lane 4). Crosslinking treatment of mitochondria from these cells also resulted in the appearance of an additional ϳ30-kDa band, which may represent a tBID homodimer. Next we analyzed the mitochondria-enriched fractions prepared from cells transfected with either tBID-GST or unfused GST. Because tBID-GST is a ϳ40-kDa protein, a tBID-GST homotrimer would appear as a ϳ120-kDa protein band. However, if the 45-kDa original complex represented a complex between a tBID monomer and a 30-kDa protein or a complex between a tBID homodimer and a 15-kDa protein, cross-linking of tBID-GST would result in the appearance of a 72-kDa protein band or a 99-kDa protein band, respectively. Expression of tBID-GST in cells and treatment of the mitochondrial-enriched fraction with the sulfo-BSOCOES cross-linker resulted in the appearance of one new band, above the 100-kDa marker, that correlates with a tBID-GST homotrimer (Fig. 5B, lane 2). Stripping and reprobing the blot with an anti-GST antibody confirmed that tBID-GST is part of this cross-linked complex (lane 4). Analysis of the mitochondrialenriched fraction from cells transfected with unfused GST indicated that most of the GST localized to the cytosolic fraction (Fig. 5B, lanes 5-8). Cross-linking treatment of the cytosolic fraction resulted in the appearance of one new ϳ50-kDa band that correlates with a GST dimer (Fig. 5B, lane 8). Thus, the ability of tBID-GST to trimerize in the mitochondrial membrane relies on tBID and not on GST.  ment of the mitochondrial-enriched fraction prepared from cells transfected with FKBP-tBID resulted in the appearance of several bands that probably represent FKBP-tBID dimers and trimers (Fig. 6A, lane 2). Treatment with FK1012 resulted in a clear enhancement in FKBP-tBID dimers and possibly trimers (lane 4), and co treatment with FK506 significantly inhibited the effect of FK1012 (lane 6).

Enforced Dimerization of Chimeric FKBP-tBID by FK1012 Induces Cyt c Release, Caspase Activation, and Apoptosis but Does Not Induce the Dimerization of Either BAX or BAK-We
To assess whether FKBP-tBID dimerization was inducing apoptosis of 293T cells, we measured the percent of cells displaying a sub-G 1 DNA content. Treatment of cells with FK1012 for 12 h induced a 2-fold increase in apoptosis compared with untreated cells, and co-addition of FK506 significantly reduced this enhanced death (Fig. 6B). The addition of FK1012 to cells transfected with the empty vector had no effect on the viability of cells (not shown). The death of HeLa and Cos-7 cells expressing FKBP-tBID was also enhanced by FK1012 (not shown), indicating that this effect is not lineage-specific.
To assess whether caspases were activated following tBID dimerization, the cleavage of a specific fluorogenic peptide substrate DEVD-AMC for the caspase-3-like subset was measured. Enforced dimerization of FKBP-tBID induced a 2-fold increase in caspase-3 activity compared with untreated transfected cells (Fig. 6C). This enzymatic activity was blocked by pretreatment with the pan-caspase inhibitor, zVAD-fmk. Pretreatment of FK1012-treated cells with zVAD-fmk also completely blocked apoptosis (not shown), indicating that caspases were essential for tBID dimer-induced death.
To determine whether FKBP-tBID dimerization induces Cyt c release from mitochondria, the cells were treated with or without FK1012, and Cyt c release was monitored by immunocytochemistry 12 h post-transfection. For the immunofluorescence experiments, the cells were co-stained with an anti-BID and an anti-Cyt c antibody. The nuclei were stained by 4Ј,6diamidino-2-phenylindole dihydrochloride. Only cells that were positive for BID and showed intact nuclei were counted. As shown in Fig. 6D, FK1012 treatment induced a ϳ40% increase in the number of cells releasing Cyt c. Of note, the nuclear staining of FKBP-tBID (upper panel) does not represent its pattern of staining in all cells and does not seem to occur as a result of enforced dimerization because a similar pattern of staining is also seen in cells that were not treated with FK1012 (not shown).
It was previously demonstrated that BID induces the oligomerization of either BAK or BAX, resulting in the release of Cyt c (3,4). Based on these studies and on the fact that enforced dimerization of tBID induces Cyt c release, we expected that enforced dimerization of tBID would induce the oligomerization of either BAX or BAK. To analyze this point, FKBP-tBID transfected cells were treated with either FK1012 alone or a combination of FK1012 together with FK506, and the oligomerization of BAX or BAK was assessed using specific cross-linkers. Mitochondria-enriched fractions were treated with either the BS 3 or the BMH cross-linker, which were previously demonstrated to cross-link BAX or BAK oligomers, respectively (2,4). At the end of the reaction, mitochondria were lysed and analyzed by Western blot using anti-BAX or anti-BAK antibodies. It was previously demonstrated that the majority of BAK displays a faster migrating inactive form (ϳ21 kDa) following BMH cross-linking and that the expression of tBID eliminates this band and induces the dimerization of BAK (15). BMH treatment of mitochondria prepared from transfected, FK1012untreated cells indicated that FKBP-tBID does not induce the dimerization of BAK, nor does it eliminate the faster inactive form (Fig. 6E, top panel, lane 4). The addition of FK1012 had no additional effect (lane 6). On the other hand, BS 3 treatment of mitochondria prepared from transfected, FK1012-untreated cells indicated that expression of FKBP-tBID induces the dimerization of BAX (Fig. 6E, bottom panel, lane 4). Surprisingly, the addition of FK1012 did not increase the amount of BAX homodimers but rather decreased it (lane 6). Because the cross-linked endogenous BAX dimer showed a very faint signal, we performed the experiment again using mitochondria prepared from cells co-transfected with FKBP-tBID together with HA-BAX. Western blot analysis with anti-HA antibodies clearly demonstrated that FKBP-tBID induces the dimerization and perhaps trimerization of BAX (lane 10) and that the addition of FK1012 decreased the amount of BAX homodimers to basal levels (lane 12). Co-addition of FK506 increased back the levels of BAX dimers (lane 14), indicating that the FK1012induced decrease in BAX dimers was due to dimerization of tBID.
A tBID BH3 Mutant, Which Does Not Induce the Dimerization of Either BAX or BAK, Forms Homotrimers and Induces Cyt c Release, Caspase Activation, and Apoptosis-The results presented above suggest that the BH3 domain of tBID may not be involved in tBID dimer-induced apoptosis, because the interaction of BID with either BAX or BAK is mediated through this domain (4,14,22). To analyze this possibility we determined whether a BH3 mutant of tBID (mIII-4/G94E) that does not interact with either BAX or BAK forms homotrimers and induces apoptosis.
We first wished to confirm that tBID mIII-4 does not induce the dimerization of either BAX or BAK in cells. Mitochondriaenriched fractions prepared from cells transfected with either wild type (wt) tBID or tBID mIII-4 were treated with either the BS 3 or the BMH cross-linker, lysed, and analyzed by Western blot. BMH treatment of mitochondria-enriched fractions indicated that neither wt tBID nor tBID mIII-4 induce the dimerization of BAK or eliminate the faster migrating inactive form (Fig. 7A, right panel). For the BAX cross-linking experiments, the cells were transfected with HA-BAX alone or in combination with either wt tBID or tBID mIII-4. Western blot analysis with anti-HA antibodies clearly demonstrated that wt tBID induces a significant increase in BAX homodimeres, whereas tBID mIII-4 does not (Fig. 7A, left panel).
To assess whether tBID mIII-4 forms the 45-kDa complex in the mitochondrial membrane, the cells were transfected with either wt tBID or tBID mIII-4, and mitochondria-enriched fractions prepared from these cells were treated with the sulfo-BSOCOES cross-linker. Western blot analysis demonstrated that tBID mIII-4 formed the 45-kDa complex as efficiently as wt tBID, indicating that an intact BH3 domain was not essential for forming this complex (Fig. 7B). The ability of tBID mIII-4 to form the 45-kDa complex was also observed in HeLa and Cos-7 cells (not shown), indicating that this phenomena is not lineage-specific.
To assess whether tBID mIII-4 can induce apoptosis, we measured the percentage of 293T, HeLa, and Cos-7 cells displaying a sub-G 1 DNA content following transfection. In all three cell lines, tBID mIII-4 was capable of inducing apoptosis ( Fig. 7C and data not shown). As shown in Fig. 7 (D and E), tBID mIII-4 was also capable of inducing caspase-3 activation and Cyt c release. DISCUSSION In this study we demonstrate that tBID forms a 45-kDa complex in the mitochondrial membrane following a death signal. Our studies using FRET analysis, co-immunoprecipitation, and cross-linkers strongly suggest that the 45-kDa complex represents a tBID homotrimer. In addition, enforced dimerization of a chimeric FKBP-tBID molecule by the bivalent ligand FK1012 induces Cyt c release, caspase activation, and apoptosis. Strikingly, tBID dimers did not induce the dimerization of BAX or BAK, and a tBID BH3 mutant was capable of forming the 45-kDa complex and inducing apoptosis.
It is currently believed that tBID acts as a monomer because yeast two-hybrid and in vitro binding assays showed that fulllength BID does not homodimerize (14). Our results indicate that recombinant tBID forms homooligomers in solution (Fig.  2). To assess whether tBID-tBID interactions occur in cells, we have performed co-immunoprecipitation experiments using cells transfected with tBID tagged with either a FLAG or an HA epitope. These experiments led to the conclusion that tBID-tBID interactions occur in cells and that the major site of these interactions is the mitochondrial membrane (Fig. 3). To further confirm that tBID-tBID interactions occur in cells, we analyzed these interactions in living cells using FRET. Out data strongly suggest that tBID-CFP and tBID-YFP co-assemble to allow FRET (Fig. 4). The fact that both of these molecules localize to the mitochondria suggests that FRET between tBID-CFP and tBID-YFP occurs in the mitochondrial membrane.
TNF␣ treatment of FL5.12 cells, targeting of tBID to mouse liver mitochondria, or transfection of cells with tBID induces the formation of a 45-kDa cross-linkable mitochondrial complex that includes tBID (Figs. 1, 2, and 5). A tBID BH3 mutant, mIII-4, that does not interact with BAX, BAK, BCL-2, or BCL-X L (14) could still form this 45-kDa mitochondrial com-plex (Fig. 7). Moreover, Western blot analysis with antibodies against the BCL-2 family members mentioned above or with antibodies against several mitochondrial proteins indicated that none of them were part of this complex ( Fig. 1 and data not shown). Based on these data and on the fact that recombinant tBID was capable of forming homooligomers, we suspected that the 45-kDa complex represented a tBID homotrimer. To address this possibility, we transfected cells with a ϳ40-kDa tBID-GST chimera and performed cross-linking experiments with the mitochondrial-enriched fraction. Cross-linking treatment resulted in the appearance of a single ϳ120-kDa band, which most likely represents a tBID-GST homotrimer (Fig. 5). Similar experiments using unfused GST indicated that GST forms dimers but not trimers in the cytosolic fraction. Taken together, these data strongly suggest that the tBID 45-kDa complex represents a tBID homotrimer.
The 45-kDa complex seems to represent a very small percentage of the total tBID present in the mitochondrial fraction (Fig. 1). Similarly, the BAX dimer seems to represent a very small percentage of the total BAX present in the mitochondrial fraction (Fig. 1). Because the dimerization of BAX is a well established phenomenon, the faint cross-linked bands of BAX and tBID may not represent the actual concentrations of these complexes in the mitochondrial membrane.
Enforced dimerization of tBID seems to enhance its trimerization (appearance of a ϳ100-kDa band; Fig. 6), suggesting that dimer formation may lead to trimer formation and that trimers may contribute to the cellular effects observed with the FKBP system. One possible explanation for this result is that dimerization of tBID is the rate-limiting step in the process of trimerization, and once dimerization is accelerated, then much more trimers can be formed. Nevertheless, the FKBP-tBID/ FK1012 strategy argues that tBID dimers are active, because enforced dimerization of FKBP-tBID induces Cyt c release, caspase activation, and apoptosis (Fig. 6). A similar approach using FKBP-BAX demonstrated that enforced dimerization of BAX resulted in mitochondrial dysfunction and apoptosis (2). Based on these similar effects, we suspected that tBID dimers were acting through BAX dimers to induce apoptosis. Moreover, it was previously demonstrated that tBID translocates to the mitochondria to induce the oligomerization of BAX or BAK (3,4) and that both molecules are essential for tBID to induce apoptosis in MEFs (15). Our results with the FKBP-tBID/ FK1012 system indicate that tBID dimers are inducing Cyt c release and apoptosis without inducing the dimerization of BAX or BAK (Fig. 6). The fact that enforced dimerization of FKBP-tBID by FK1012 reduced the amount of BAX dimers further suggests that tBID dimers are inducing apoptosis in certain cells by an alternative mechanism/pathway that does not include the dimerization of either BAX or BAK. Supporting evidence for the existence of two separate pathways are the findings that in certain settings BID induces Cyt c release in the absence of mitochondria depolarization, whereas BAX and BAK induce apoptosis with mitochondria depolarization (23).
To further assess the involvement of multidomain pro-apoptotic molecules in the tBID oligomer death pathway, we have analyzed tBID mIII-4 in three different cell lines. This BH3 mutant was previously shown to be incapable of inducing the dimerization of BAK or of inducing the release of Cyt c from purified mouse liver mitochondria (4). In our experiments performed in intact cells, tBID mIII-4 did not induce the dimerization of either BAK or BAX but was capable of inducing Cyt c release, caspase activation, and apoptosis (Fig. 7). Thus, it seems that a BH3-independent pathway for inducing Cyt c release exists in intact cells. The fact that tBID mIII-4 was capable of inducing apoptosis and of forming the 45-kDa com- plex in a variety of cell lines (Fig. 7) suggests that homotrimers of this mutant might be the active components responsible for Cyt c release. These results, together with the results obtained with the FKBP-tBID/FK1012 system, suggest that two separate pathways may exist downstream of the mitochondrial targeting of tBID. One pathway is BH3-dependent and leads to BAX/BAK oligomerization and Cyt c release, and the other is BH3-independent and leads to tBID oligomerization and Cyt c release (Fig. 8).
The results presented above suggest that insertion of tBID into the mitochondrial membrane may cause a conformational change that enables oligomerization of tBID. This conformational change may include the exposure of hydrophobic domains that were previously buried. One or more domains that become exposed as a result of membrane insertion are likely to be involved in the oligomerization process. tBID oligomers may create new interfaces that will enable interactions with new molecules. On the other hand, the three-dimensional structure and the in vitro electrophysiological data suggest that tBID is a pore-forming protein (6,16). Thus, tBID oligomers may form pores in the mitochondrial membrane. In addition to its ability to form pores, tBID is capable of destabilizing lipid membranes in vitro (24). Interestingly, this destabilizing effect of tBID was demonstrated to be independent of its BH3 domain.
This study indicates that tBID itself can oligomerize in the mitochondrial membrane in addition to its ability to induce the oligomerization of multidomain pro-apoptotic molecules. Enforced dimerization of tBID results in Cyt c release, caspase activation, and apoptosis but not the dimerization of BAX or BAK. Thus, tBID oligomerization in response to a death signal appears to represent an alternative/additional mechanism for inducing mitochondrial dysfunction.