(Received for publication, November 18, 1996, and in revised form, February 19, 1997)
From the Biochemistry Section, Surgical Neurology Branch, NINDS, National Institutes of Health, Bethesda, Maryland 20892
Members of the Bcl-2 family (including Bcl-2,
Bcl-XL, and Bax) play key roles in the regulation of
apoptosis. These proteins are believed to be membrane-associated and
have been proposed to regulate apoptosis through both homodimerization
and heterodimerization. We have found that whereas Bcl-2 is
predominantly membrane-associated as previously reported, significant
amounts of Bcl-XL and most of the Bax proteins are not
membrane-associated and thus appear in the cytosolic fraction of
thymocyte and splenocyte extracts. This finding allows the study of the
dimerization properties and conformation of these proteins in the
absence of detergent perturbation. For this analysis, we have produced
monoclonal antibodies that are specific for known epitopes of Bax,
Bcl-2, and Bcl-XL. An antibody to an N-terminal epitope
( uBax 6A7) between amino acids 12 and 24 fails to bind the soluble
cytosolic form of Bax, indicating that this epitope is normally buried.
Nonionic detergents alter the Bax conformation to expose this epitope.
In the presence of nonionic detergent, the 6A7 antibody avidly binds
the monomeric form of Bax, but not Bax complexed with either
Bcl-XL or Bcl-2. In contrast, a monoclonal antibody to an
adjacent epitope of Bax (
mBax 5B7) within amino acids 3-16 binds
the soluble and detergent-altered forms of Bax and also binds the
Bax·Bcl-XL or the Bax·Bcl-2 complex. Surprisingly, in
the absence of detergent Bax fails to form homodimers or heterodimers
with Bcl-XL. These results demonstrate a novel conformational state of members of the Bcl-2 family under a
physiological condition that is distinct from the detergent-altered
state that forms dimers and is currently believed to regulate
apoptosis.
Bcl-2, Bax, and Bcl-XL are members of a family of proteins that regulate apoptosis. The gene encoding Bcl-2 was first identified at the chromosomal translocation point t(14;18) in human B cell follicular lymphoma (1-3). Overexpression of Bcl-2 enhanced cell survival by suppressing apoptosis in a number of cells subjected to a wide range of apoptosis-inducing stimuli, including nerve growth factor withdrawal, radiation, and chemotherapeutic agents (for reviews, see Refs. 4-7). The gene encoding Bcl-XL was first identified by homology cloning using the bcl-2 gene as a probe (8). Like Bcl-2, overexpression of Bcl-XL inhibited apoptosis that was induced by a variety of agents (9). Finally, Bax was identified as a protein that co-immunoprecipitated with Bcl-2 (10). In contrast to Bcl-2 and Bcl-XL, overexpression of Bax accelerated cell death (10).
By immunoprecipitation and yeast two-hybrid select studies, it appears that Bax may form homodimers or heterodimers with either Bcl-2 or Bcl-XL (10-12). Formation of Bax homodimers has been proposed to promote cell death, and this could be blocked by Bax heterodimerization with Bcl-2 or Bcl-XL. Based on Bax overexpression and dimerization studies, it has been postulated that the relative ratio of Bax homodimers to heterodimers may serve as a key sensory switch that dictates the initiation of apoptosis (10, 12, 13).
Dimerizations between Bax, Bcl-2, and Bcl-XL are believed to be mediated through three highly conserved regions within each protein known as the BH1, BH2, and BH3 domains (14, 15). BH1 and BH2 domains are required for Bcl-2 and Bcl-XL to heterodimerize with Bax. Mutations within these domains disrupt their heterodimerizations with Bax and abrogate their abilities to prevent apoptosis (12, 14). Furthermore, the BH3 domain of Bax has been shown to be essential for both its homodimerization and heterodimerization with either Bcl-2 or Bcl-XL (15).
Bcl-2, Bax, and Bcl-XL have a hydrophobic segment at their respective C-terminal ends (8, 10, 16). Hydropathy plots predict that these segments may serve as membrane anchors. Subcellular fractionation studies have shown Bcl-2 to be localized to the nuclear envelope, endoplasmic reticulum, and outer mitochondrial membranes (17, 18). Recently, immunocytochemical studies have localized Bcl-XL to outer membranes of mitochondria (19). Bax has been suggested to co-localize within the same subcellular compartments as Bcl-2 (10) through its ability to dimerize with Bcl-2.
In this paper, we describe the generation of several epitope-specific monoclonal antibodies against Bcl-2, Bax, and Bcl-XL. Using these antibodies, we probed the subcellular localization of Bax, Bcl-2, and Bcl-XL in normal murine thymocytes and splenocytes and studied the conformational dependence of dimerization among members of the Bcl-2 family.
-Synthetic peptides were purchased from Peptide Technologies Corp. Imject® maleimide-activated keyhole limpet hemocyanin (KLH)1 was obtained from Pierce. pcDNA3 plasmid was from Invitrogen. PEG 4000, HAT medium (hypoxanthine/aminopterin/thymidine medium), and LipofectAMINE were from Life Technologies, Inc. NS-1 and HeLa cells were from the ATCC. Iscove's medium was from Biofluids, Inc. Hamster anti-murine Bcl-2 monoclonal antibody 3F11 and rabbit anti-calnexin polyclonal antibodies were from PharMingen and StressGen Biotech Corp., respectively. Sheep anti-mouse and donkey anti-rabbit immunoglobulin peroxidase conjugates and ECL Western blotting detection kit were purchased from Amersham Corp. Goat anti-hamster immunoglobulin peroxidase conjugate was from Kirkegaard and Perry Laboratories. Human thymic tissues were generous gifts from the Children's National Medical Center, Washington, D. C. All other reagents were obtained from Sigma.
Generation of Monoclonal AntibodiesPeptides corresponding
to internal sequences of Bax, Bcl-2, and Bcl-XL were
conjugated to maleimide-activated KLH through the cysteine residue
according to the protocol provided by the manufacturer. Anti-Bax
monoclonal antibodies were generated by immunizing mice with KLH
conjugated to peptides corresponding to amino acids 3-16 of murine Bax
(CGSGEQLGSGGPTSS), amino acids 3-16 of human Bax (CGSGEQPRGGGPTSS),
and amino acids 12-24 shared by Bax from both species
(CGPTSSEQIMKTGA). Anti-murine Bcl-2 antibody was generated by
immunizing mice with a KLH-conjugated peptide (amino acids 61-76) of
murine Bcl-2 (CVHREMAARTSPLRPLV). Anti-Bcl-XL antibody was
generated by immunizing mice with a KLH-conjugated peptide having a
sequence corresponding to amino acids 3-14 shared by both murine and
human Bcl-XL (CQSNRELVVDFLS). Splenocytes from immunoreactive mice were fused by PEG 4000 to murine NS-1 myeloma cells
and selected with HAT medium (20). The anti-murine, human, and
species-independent Bax antibodies were designated as mBax 5B7,
hBax 2D2 (or 1F6), and
uBax 6A7, respectively. The anti-murine Bcl-2 antibody was designated as
mBcl-2 10C4. The
anti-species-independent Bcl-XL antibody was designated as
uBcl-XL 2H12.
Murine Bax cDNA (a kind gift of Dr. Stanley Korsmeyer) was cloned into pcDNA3. Eight plates (100 × 20 mm) of HeLa cells at 80-90% confluency were each transfected with 16 µg of mBax/pcDNA3 plasmid using the LipofectAMINE (128 µg/plate) transfection protocol described in the product instructional guide. The cells were harvested at 36 h post-transfection, and soluble extract was prepared as described below.
Subcellular Localization of Bax, Bcl-2, and Bcl-XLMurine thymocytes and splenocytes were dispersed from their tissues in the presence of Iscove's medium. The cells were spun down, washed once in Iscove's medium, and resuspended in 70 ml of lysis buffer (10 mM Hepes, pH 7.4, 38 mM NaCl, 25 µg/ml phenylmethylsulfonyl fluoride, 1 µg/ml leupeptin, and 1 µg/ml aprotinin) at a cell density of 5 × 107/ml. The hypotonic lysate was Dounce-homogenized. The homogenate was spun at 900 × g to pellet the nuclei using a Sorvall SA-600 rotor. The postnuclear supernatant was then spun at 130,000 × g in a Beckman Ti 45 rotor to pellet the membranes. Both the crude membranes and the nuclear pellet were resuspended in 70 ml of lysis buffer. The extent of cross-contamination between the soluble protein and the crude membrane preparations was assessed by assaying for either the cytosolic enzyme marker lactate dehydrogenase (21) or the mitochondrial enzyme marker cytochrome c oxidase (22). Anti-calnexin polyclonal antibody was used in Western blotting to determine the level of endoplasmic reticulum membranes in the cytosolic extract. The relative levels of Bax, Bcl-2, and Bcl-XL in the cytosolic, crude membrane, and nuclear fractions were analyzed by Western blotting with the corresponding antibodies.
For the study of Bax homodimerization, human and murine thymocytes were hypotonically lysed at a cell density of 1 × 108/ml, whereas HeLa cells transfected with mouse Bax were lysed at a cell density of 5 × 106/ml. Soluble proteins from these cells were prepared as described above. The soluble extracts were used for the immunoprecipitation studies as described below.
Immunoprecipitation of BaxMonoclonal antibodies mBax
5B7,
hBax 1F6, and
uBax 6A7 were purified from ascites fluids
by ammonium sulfate precipitation and DEAE fractionation (23, 24). The
purified antibodies were immobilized onto CNBr-activated Sepharose 4B
at 2.5 mg of protein/ml of packed beads (23, 24). For the
immunoprecipitation of Bax from detergent-solubilized samples, murine
splenocytes or thymocytes were solubilized in 10 ml of 10 mM Hepes, pH 7.4, 150 mM NaCl, and 1% Triton
X-100 or 0.2% Nonidet P-40 at a cell density of 5 × 107/ml in the presence of proteolytic inhibitors (25 µg/ml phenylmethylsulfonyl fluoride, 1 µg/ml leupeptin, and 1 µg/ml aprotinin). The lysate was spun at 14,000 rpm in a Sorvall SA
600 rotor for 15 min to pellet the nuclei and the unsolubilized
material. 4.5 ml of the detergent-solubilized lysate was then mixed
with 150 µl of the
mBax 5B7 or
uBax 6A7 antibody beads and
allowed to incubate for 2 h at 4 °C. The beads were then washed
twice with 10 ml of solubilization buffer containing either 0.2%
Triton X-100 or 0.2% Nonidet P-40. The bound proteins were then eluted
from the beads by the addition of 180 µl of 0.1 M acetic
acid containing either 0.2% Triton X-100 or Nonidet P-40. The acid
eluants were neutralized by the addition of 30 µl of 1 M
Tris, pH 8.0.
For the immunoprecipitation of murine Bax from the cytosolic extract of
murine thymocytes, the NaCl concentration of the extract was adjusted
to 150 mM, and the extract was then filtered through a
0.45-µm filter (Millipore Corp.). 4.5 ml of the soluble extract was
mixed with 150 µl of mBax 5B7 or
uBax 6A7 antibody beads either in the absence or presence of 0.2% Triton X-100 or Nonidet P-40
and allowed to incubate for 2 h. The unbound proteins were removed
by washing the beads twice with 10 ml of 10 mM Hepes, pH
7.4, and 150 mM NaCl or with the same buffer containing
either 0.2% Triton X-100 or Nonidet P-40. The bound proteins were then eluted from the beads with 180 µl of 0.1 M acetic acid
containing either 0.2% Triton X-100 or Nonidet P-40. The acid eluants
were neutralized with 30 µl of 1 M Tris, pH 8.0.
For Bax homodimerization studies, 4.5 ml of the mixed human and murine
cytosolic extracts (with the NaCl concentration adjusted to 150 mM) that had been filtered were allowed to incubate at 4 °C for 1 h either in the absence or presence of 0.2% Triton X-100 or Nonidet P-40. The three different extracts were then mixed
with 150 µl of hBax 1F6 antibody beads, and immunoprecipitation was carried out as described above. Immunoprecipitation of murine Bax
from transfected HeLa cells was carried out by mixing 150 µl of
mBax 5B7 antibody beads with 4.5 ml of the filtered soluble extract
from transfected HeLa cells either in the absence or presence of 0.2%
Triton X-100 or Nonidet P-40. The washing and elution steps were
performed as described above.
SDS-polyacrylamide gel
electrophoresis was carried out on 12% polyacrylamide gels according
to the method of Laemmli (25). 23 µl of the samples from subcellular
fractionation and immunoprecipitation studies were analyzed by
SDS-PAGE. The gels were either stained with Coomassie Blue or
electroblotted onto Immobilon membranes (20). For immunoblotting
analysis, the blots were blocked in phosphate-buffered saline, 0.05%
Tween 20 containing 5% nonfat milk or fetal bovine serum and incubated
with either hBax 2D2 (1:10 diluted culture fluid),
mBax 5B7
(1:10 diluted culture fluid),
uBcl-XL 2H12 (1:10
diluted culture fluid),
mBcl-2 10C4 (1:10 diluted culture fluid),
hamster anti-murine Bcl-2 monoclonal antibody 3F11 (1 µg/ml), or
rabbit anti-calnexin polyclonal antibody (1:1000 dilution) diluted in
the blocking buffer for 45 min. The blots were then washed in
phosphate-buffered saline, 0.05% Tween 20 and incubated in blocking
buffer containing either 1:5000 diluted sheep anti-mouse, donkey
anti-rabbit, or goat anti-hamster immunoglobulin peroxidase for an
additional 30 min. The blots were washed again in phosphate-buffered
saline, 0.05% Tween 20 and visualized with ECL Western blotting
detection kit.
Several monoclonal
antibodies were raised against Bax, Bcl-2, and Bcl-XL for
the purpose of Western blotting detection and immunoprecipitation
analysis. The hBax 2D2 and hBax 1F6 monoclonal antibodies were
directed against an N-terminal peptide (amino acids 3-16) of human
Bax, and they do not cross-react with either murine or rat Bax. On the
other hand, the
mBax 5B7 monoclonal antibody was raised against an
N-terminal peptide (amino acids 3-16) of murine Bax and is specific
for mouse. The
uBax 6A7 monoclonal antibody was produced against a
peptide sequence (amino acids 12-24) common to human, murine, and rat
Bax, and this antibody binds to Bax from all three species. The
mBcl-2 10C4 monoclonal antibody was raised against a murine Bcl-2
peptide (amino acids 61-76), and this antibody cross-reacts with rat
Bcl-2. Lastly, the
uBcl-XL 2H12 monoclonal antibody was
generated against an N-terminal peptide (amino acids 3-14) common to
human and murine Bcl-XL, and this antibody binds to human,
murine, and rat Bcl-XL.
We have recently found a differential
compartmentalization of Bax, Bcl-2, and Bcl-XL (30). To
further confirm this finding, murine thymocytes and splenocytes were
hypotonically lysed and Dounce-homogenized. The cell lysates were
separated into soluble, crude membrane, and nuclear fractions by
differential centrifugation. As shown on the Coomassie Blue-stained
SDS-polyacrylamide gel (Fig. 1), the total proteins from
these three different subcellular compartments (lanes c,
d, and e) have distinct banding patterns in both
thymus and spleen. Contamination from mitochondria and endoplasmic
reticulum was not detected in the soluble protein fractions as
determined by cytochrome c oxidase activity assay and
Western blotting analysis with an anti-calnexin polyclonal antibody,
respectively (data not shown). Western blotting analyses of these
samples revealed Bax as a predominantly soluble protein in both
thymocytes (Fig. 1A, lane c, arrow)
and splenocytes (Fig. 1B, lane c,
arrow). Bcl-2, on the other hand, was found in both crude
membrane and nuclear fractions but was not detected in the soluble
fraction (Fig. 1, A and B, lanes d and
e). Finally, Bcl-XL was found to be present in
all three compartments of thymocytes (Fig. 1A, lanes
c, d, and e) and in soluble and nuclear
fractions of splenocytes (Fig. 1B, lanes c and
e). These results indicate that Bax, Bcl-2, and
Bcl-XL are differentially compartmentalized with Bax being
predominantly soluble, Bcl-2 being exclusively membrane-bound, and
Bcl-XL being present in both soluble and membrane-bound forms.
Effect of Detergent on the Heterodimerization of Bax
The
differential subcellular localization of Bax and Bcl-2 in both
thymocytes and splenocytes raises uncertainty as to whether the
formation of Bax·Bcl-2 heterodimer complexes exist in healthy living
cells. The finding of the soluble cytosolic localization of Bax and a
significant fraction of Bcl-XL (Fig. 1) eliminates the
necessity of detergent solubilization to analyze Bax heterodimer formation. Therefore, immunoprecipitation of Bax from the cytosolic extract of murine thymocytes in the absence of detergent was compared with that in the presence of the detergents Triton X-100 or Nonidet P-40. These nonionic detergents do not disrupt Bax heterodimer formation. Anti-Bax monoclonal antibody 5B7 was used to
immunoprecipitate Bax. The immunoprecipitated samples were analyzed by
Western blotting with uBcl-XL 2H12 monoclonal antibody
for the detection of murine Bcl-XL. In the absence of
detergent, immunoprecipitation of Bax by the
mBax 5B7 antibody
failed to co-precipitate Bcl-XL (Fig. 2,
lower left panel, lane c). However, if either
Triton X-100 or Nonidet P-40 was present during the
immunoprecipitation, a significant amount of Bcl-XL was
co-precipitated with Bax (Fig. 2, lower middle and
right panels, lanes c). Immunoprecipitation analysis has also been carried out using murine splenocyte cytosolic extract and has yielded similar detergent-dependent binding
of Bax with Bcl-XL (data not shown). Since there is no
evidence of Bax association with Bcl-XL in the absence of
detergent, it appears that detergent induces a conformational change in
Bax that facilitates the heterodimerization process. The lack of
Bax/Bcl-XL heterodimer formation in the absence of
detergent together with the mutually exclusive localization of Bax and
Bcl-2 stimulates a reconsideration of the possible roles that dimer
formation may play in the regulation of apoptosis.
Detergent Exposure of an N-terminal Epitope Correlates with Heterodimer Formation
Immunoprecipitation of Bax from the
cytosolic extract of murine thymocytes with mBax 5B7 antibody was
compared with that of the
uBax 6A7 antibody, which is an antibody
whose epitope (amino acids 12-24) partially overlaps or is very near
to that of the 5B7 antibody (amino acids 3-16). The immunoprecipitated samples were analyzed by Western blotting with
mBax 5B7 monoclonal antibody for the detection of murine Bax. Whereas the binding of murine
Bax to
mBax 5B7 antibody occurred in the presence or absence of
detergents (Fig. 2, upper panels, lanes c), the
uBax 6A7 antibody required the presence of either Triton X-100 or
Nonidet P-40 detergent during immunoprecipitation to bind Bax (Fig. 2,
upper panels, lanes e). A similar
detergent-dependent binding of Bax to the
uBax 6A7
antibody was observed with murine splenocyte cytosolic extract (data
not shown). Thus, the amino acid 12-24 segment in the N terminus of
Bax is normally inaccessible to 6A7 antibody binding and becomes
exposed in detergent. This is particularly interesting because the
heterodimerization of Bax also depends on the presence of detergent
(Fig. 2), suggesting that a conformational change exposing this
N-terminal epitope region may be involved in heterodimer formation.
To test this hypothesis, we examined heterodimer formation by
immunoprecipitation with the uBax 6A7 antibody. In the presence of
detergent, the
uBax 6A7 monoclonal antibody bound only to Bax
species that were not associated with Bcl-XL (Fig. 2,
lower middle and right panels, lanes e
in each panel). We further examined the competition between
the 6A7 antibody binding to Bax and Bax heterodimer formation with
Bcl-XL and Bcl-2. Anti-Bax monoclonal antibodies
mBax
5B7 and
uBax 6A7 were used for the immunoprecipitation of Bax from
the total cell lysates of murine thymocytes (Figs. 3A and 4A) and splenocytes (Figs.
3B and 4B). Immunoprecipitations were carried out
in the presence of either 1% Triton X-100 or 0.2% Nonidet P-40 (10).
The immunoprecipitated samples were analyzed by Western blotting with
either
mBax 5B7, hamster
mouse Bcl-2 3F11, or
uBcl-XL 2H12 monoclonal antibodies. As shown in Fig. 3,
immunoprecipitation with the
mBax 5B7 antibody from either
thymocytes or splenocytes resulted in the co-precipitation of Bcl-2 and
Bcl-XL with Bax in the presence of either Triton X-100 or
Nonidet P-40. The immunoprecipitation patterns with the
uBax 6A7
monoclonal antibody as shown in Fig. 4, however,
appeared to be quite different. Neither Bcl-2 nor Bcl-XL
from thymocytes or splenocytes was found to associate with the
uBax
6A7 antibody-immunoprecipitated Bax in the presence of either Triton
X-100 or Nonidet P-40. This suggests that the epitope for the 6A7
antibody (amino acids 12-24) may be in the vicinity of the
dimerization domain of Bax so that the antibody directly competes with
either Bcl-2 or Bcl-XL for binding to Bax. This is
interesting because the 6A7 antibody epitope partially overlaps or is
adjacent to the 5B7 antibody epitope (amino acids 3-16) and yet the
latter antibody does not cross-compete with Bcl-2 and
Bcl-XL for binding to Bax.
Effect of Detergent on the Homodimerization of Bax
Bax has
been previously reported to homodimerize in the presence of Nonidet
P-40 (10). Considering that Bax did not heterodimerize with
Bcl-XL in the soluble detergent-free state, we examined
whether Bax homodimerization required detergent. Bax was
immunoprecipitated from a mixture of the soluble extracts of human and
murine thymocytes either in the absence of detergent or in the presence
of Triton X-100 or Nonidet P-40. Human Bax was immunoprecipitated from
the mixture by hBax 1F6 monoclonal antibody to determine whether human Bax can dimerize with murine Bax. This was made possible because
the N-terminal epitope recognized by the anti-human Bax antibody 1F6 is
distinct from murine Bax. The immunoprecipitated samples were analyzed
by Western blotting with species-specific anti-human Bax 2D2 and
anti-murine Bax 5B7 monoclonal antibodies (Fig.
5A). As revealed by Western blotting with
hBax 2D2 antibody, the affinity of anti-human Bax antibody 1F6 to Bax
was not affected by detergent as evidenced by the equal amounts of
human Bax being immunoprecipitated in the absence and presence of
detergent (Fig. 5A, top, lanes f-h).
Homodimerization of human Bax to murine Bax as determined by Western
blotting with
mBax 5B7 antibody occurred readily in the presence of
detergent (Fig. 5A, bottom, lanes g and h). However, in the absence of detergent dimerization
between human and murine Bax failed to occur (Fig. 5A,
bottom, lane f).
Nevertheless, it is possible that the endogenous Bax present in murine
and human thymocytes is already in a tight homodimer state, and that
the addition of detergent merely allowed the exchange of partners. To
examine this possibility, we transiently expressed murine Bax in HeLa,
a human cervical carcinoma cell line that expresses endogenous human
Bax. Immunoprecipitation of murine Bax was carried out from the soluble
extract of transfected HeLa cells by mBax 5B7 antibody beads either
in the absence of detergent or in the presence of detergent Triton
X-100 or Nonidet P-40. This allows us to determine whether murine and
human Bax forms homodimers when co-expressed or whether the formation
of homodimers is induced artificially by detergents. As shown by
Western blotting analysis with
mBax 5B7 antibody, murine Bax is
expressed in transfected HeLa cells (Fig. 5B,
top, lanes b-d). The affinity of
mBax 5B7
antibody to Bax was again shown to be unaffected by detergent as
evidenced by the equal amounts of murine Bax being immunoprecipitated
in the absence and presence of detergent (Fig. 5B,
top, lanes e-g). Western blotting analysis with
hBax 2D2 antibody revealed that whereas human Bax readily dimerizes
with murine Bax in the presence of detergent Triton X-100 (Fig.
5B, bottom, lane f) or Nonidet P-40
(Fig. 5B, bottom, lane g), no dimerization was observed in the absence of detergent (Fig.
5B, bottom, lane e). These results
suggest that Bax homodimerization, as was seen for Bax
heterodimerization (Fig. 2), is a detergent-induced process.
Bcl-2, Bax, and Bcl-XL are believed to be a group of
membrane proteins that regulate apoptosis through in vivo
dimerizations (10, 12, 13). In healthy cells, the formation of
pro-apoptotic Bax homodimers is thought to be countermanded by Bax
heterodimerization with pro-survival factors Bcl-2 or
Bcl-XL. Our finding that both Bax and Bcl-XL
are soluble (30) allowed us to explore the dimerization process without
perturbing the protein structures with detergent. Surprisingly, in the
absence of detergent Bcl-XL failed to co-immunoprecipitate with Bax by the mBax 5B7 antibody, whereas these two proteins readily co-immunoprecipitated in the presence of detergent. In addition, Bax homodimerization also was found to be dependent upon the
presence of detergent. The differential localization of Bax and Bcl-2
suggests that these two proteins do not tightly associate in healthy
thymocytes and splenocytes.
The above observations raised questions as to the extent to which Bax
heterodimer and homodimer formation occur under physiological conditions. Fig. 6 proposes a model of the
detergent-induced dimerization of Bax. The observations that Bax did
not homodimerize or heterodimerize with Bcl-XL in the
absence of detergent indicate that the BH3 domain of Bax, which is
crucial for Bax homodimerization and heterodimerization (15), may be
inaccessible under physiological conditions. Immunoprecipitation studies with uBax 6A7 antibody also indicate that the amino acid
residues 12-24 of Bax are not exposed in the cytosol in the absence of
detergent. The addition of detergent may thus cause a conformational
change of Bax and reveal both the BH3 domain for Bax homo- and
heterodimerization and the 6A7 epitope for antibody binding. Detergent
is generally known to disrupt protein-protein interactions. However, in
the case of Bax, it seems to promote heterodimeric and homodimeric
complexes. At present, without the equivalent conformation-sensitive
antibodies to Bcl-2 and Bcl-XL, it is not yet known whether
such detergent-induced conformational changes also occur in Bcl-2 and
Bcl-XL to allow for the exposure of BH1 and BH2
dimerization domains.
In addition to immunoprecipitation analyses, dimerization of Bax also has been observed in both the yeast two-hybrid select systems and the recombinant glutathione S-transferase-fusion protein binding studies (11, 12, 15, 26). In both cases, Bax was present as a fusion protein. It is possible that Bax, as part of a fusion protein, displayed an alternate conformation that is similar to the detergent state of this protein with the BH3 domain readily accessible for dimerization. Moreover, by site-directed mutagenesis analyses of Bcl-XL and Bax, it was reported that heterodimerization of Bcl-XL with Bax is not necessary for death-repressing activity (27, 28). Those results can be explained by our findings that soluble Bax and Bcl-XL do not form dimers in the cytoplasm in the first place. Mutations described in these two studies appeared to have eliminated the "detergent-induced" Bax heterodimer formation.
The uBax 6A7 antibody, which recognizes the detergent-induced
exposure of the amino acid 12-24 epitope, also appears to compete with
the heterodimeric partners of Bax. The
uBax 6A7 antibody-immunoprecipitated Bax was not associated with either Bcl-2 or
Bcl-XL. This suggests that the dimerization of
Bcl-XL and Bcl-2 to Bax may cover up the 6A7 antibody
epitope on Bax. Alternatively, the binding of
uBax 6A7 antibody to
Bax may block the heterodimerization of Bax to Bcl-2 and
Bcl-XL by steric hindrance. Thus, under
detergent-solubilized conditions the epitope for the 6A7 antibody may
be in close proximity to the BH3 dimerization domain of Bax (15) at the
tertiary structural level. Recently, the N-terminal region of the Bax
homolog Bcl-XL has been also shown to be in proximity to
its BH3 domain by x-ray crystallography (29).
Bax homodimers have been proposed to accelerate cell death, and this can be blocked by Bax heterodimerization with Bcl-2 and Bcl-XL (10, 12). In our studies, we detected no Bax dimers in the cytosol. It is possible, however, that the Bax molecules that migrate into membranes during apoptosis (30) may acquire the detergent-induced dimer-promoting state of Bax. Alternatively, it is possible that Bax, Bcl-2, and Bcl-XL, which induce opposite effects, may be involved in novel binding with and regulation of other cellular proteins that may function in opposing biochemical pathways. Overexpression of either Bax or Bcl-2 and Bcl-XL may then shift the equilibrium that favors either cell death or survival. If dimerizations of these proteins do not occur in vivo, then it would be crucial to determine whether these proteins compete for the binding of a common partner. Binding to a common partner could explain why mutations in BH1 and BH2 domains of Bcl-2 destroy its function. Resolving these issues will undoubtedly bring us closer to understanding the biochemical functions of members of the Bcl-2 family.
We thank Shu-Chan Hsu at Stanford University for technical assistance and critical review of the manuscript, Pat Johnson for invaluable technical assistance and transport of human thymus, Frank Midgley, Jeffrey Sell, and Roma Chandra at the Children's National Medical Center in Washington, D. C. for the continuing supply of human thymus, Stanley Korsmeyer for murine Bax cDNA, Steven Rafferty for helpful suggestions on the manuscript, Katherine Wood and Shailendra Saxena for technical discussion, Veena Vasandani, Ester Boix, Joon Lee, JoAnn Castelli, and Jamal Warner for assistance in immunization, and John Harrington for animal care.