(Received for publication, May 17, 1996, and in revised form, October 31, 1996)
From the Department of Biochemistry, School of Medical Sciences, University of Bristol, Bristol BS8 1TD, United Kingdom
Stimulation of the mitochondrial permeability transition (MPT) in de-energized mitochondria by phenylarsine oxide (PheArs) is greater than that by diamide and t-butylhydroperoxide (TBH), yet the increase in CyP binding to the inner mitochondrial membrane (Connern, C. P. and Halestrap, A. P. (1994) Biochem. J. 302, 321-324) is less. From a range of nucleotides tested only ADP, deoxy-ADP, and ATP inhibited the MPT. ADP inhibition involved two sites with Ki values of about 1 and 25 µM which were independent of [Ca2+] and CyP binding. Carboxyatractyloside (CAT) abolished the high affinity site. Following pretreatment of mitochondria with TBH or diamide, the Ki for ADP increased to 50-100 µM, whereas pretreatment with PheArs or eosin maleimide increased the Ki to >500 µM; only one inhibitory site was observed in both cases. Eosin maleimide is known to attack Cys159 of the adenine nucleotide translocase (ANT) in a CAT-sensitive manner (Majima, E., Shinohara, Y., Yamaguchi, N., Hong, Y. M., and Terada, H. (1994) Biochemistry 33, 9530-9536), and here we demonstrate CAT-sensitive binding of the ANT to a PheArs affinity column. In adenine nucleotide-depleted mitochondria, no stimulation of the MPT by uncoupler was observed in the presence or absence of thiol reagents, suggesting that membrane potential may inhibit the MPT by increasing adenine nucleotide binding through an effect on the ANT conformation. We conclude that CsA and ADP inhibit pore opening in distinct ways, CsA by displacing bound CyP and ADP by binding to the ANT. Both mechanisms act to decrease the Ca2+ sensitivity of the pore. Thiol reagents and oxidative stress may modify two thiol groups on the ANT and thus stimulate pore opening by both means.
Mitochondria exposed to supraphysiological concentrations of
[Ca2+] become nonselectively permeable to small molecules
(<1500 Da). The sensitivity of this mitochondrial permeability
transition (MPT)1 to [Ca2+]
is greatly increased by oxidative stress, adenine nucleotide (AdN)
depletion, elevated phosphate concentrations, low membrane potential
(), and agents that stabilize the "c" conformation of the
adenine nucleotide translocase (ANT). In contrast, protection is
afforded by low pH, high
, and agents that stabilize the "m"
conformation of the ANT (see Refs. 1-3). In addition, pore opening can
be prevented by sub-micromolar concentrations of cyclosporin A (CsA)
and its analogues (4-6). This effect appears to be mediated through
the matrix peptidyl-proline cis-trans-isomerase, since the
Ki values for inhibition of this enzyme by a number of CsA analogues correlate with their ability to inhibit pore opening
(6-8). N-terminal sequencing of the purified peptidyl-prolyl cis-trans-isomerase has shown it to be a member of the
cyclophilin (CyP) family (9) and most probably identical to the human
CyP3 gene product (10), now more usually termed CyP-D (11, 12). These
observations led us to suggest that an interaction of matrix CyP with
an integral membrane protein leads to a change in its conformation
which, when triggered by Ca2+, induces pore opening. In
view of the influence of the ANT conformation on this process, we
suggested that the integral membrane protein might be the ANT (6, 7).
Further evidence in support of this model has come from our
observations that oxidative stress, thiol reagents, NADH oxidation,
increased matrix volume, and chaotropic agents, all of which enhance
pore opening, also cause a recruitment of CyP to the inner
mitochondrial membrane (13, 14). In contrast, modulation of pore
opening by matrix pH, AdNs,
, and the conformation of the ANT
appeared to be independent of CyP binding and must involve different
regulatory sites on the pore complex (13, 14).
Matrix ADP is an important modulator of pore opening and acts by
decreasing the sensitivity of the calcium trigger site to [Ca2+]. There are thought to be two ADP binding sites,
one with high affinity associated with the ANT that is blocked by the
inhibitor carboxyatractyloside (CAT) and another lower affinity site
(15-18). Bernardi and colleagues (19-23) have provided convincing
evidence that another important modulator of the MPT is . Thus,
at a constant intramitochondrial [Ca2+], progressive
lowering of
by addition of uncoupler leads to a progressive
stimulation of pore opening. Thiol reagents shift this voltage
dependence, allowing the pore to open at higher
(19-23),
implying that the voltage sensor possesses critical thiol groups whose
modification attenuates the voltage sensitivity.
During our studies on the effects of thiol reagents on CyP binding, we
noticed that phenylarsine oxide (PheArs) consistently produced a
greater stimulation of pore opening than diamide or t-butylhydroperoxide (TBH), and yet induced less CyP
binding. This led us to consider the possibility that thiol reagents
might antagonize ADP inhibition of the MPT through an effect on the nucleotide binding site of the ANT, independent of their effects on CyP
binding. Under the influence of a physiological , the ANT drives
ATP4
out of mitochondria in exchange for the entry of
ADP3
. This electrically driven process is brought about
by an effect of
on the conformation of the carrier and its
affinity for AdNs at the extramitochondrial binding site (24-26). If
were also able to increase the affinity of the matrix facing
nucleotide binding site for AdNs, this might explain the regulation of
the pore by
. Thus at high
AdN binding would be tighter
and the pore would be less sensitive to [Ca2+], while at
lower
binding would become weaker and the sensitivity of the
pore to [Ca2+] would increase. This effect would be
exacerbated by thiol reagents if these decrease the affinity of AdN
binding. This paper provides evidence in support of this
hypothesis.
Materials
Rat liver mitochondria were prepared as described by Halestrap and Davidson (6). For some experiments, mitochondria were depleted of their endogenous AdNs by incubation for 15 min at 25 °C in sucrose isolation buffer supplemented with 2 mM PPi as described previously (27). N-terminal antipeptide antibodies to the mature mitochondrial cyclophilin were raised and purified as described elsewhere (13). Antibodies against bovine ANT and rat mitochondrial NAD+-dependent isocitrate dehydrogenase were kindly provided by Dr. Gerard Brandolin, Department of Molecular and Structural Biology, CNRS, F-38041 Grenoble, France, and Professor R. M. Denton of this department, respectively. Eosin 5-maleimide, safranin, 4-aminophenylarsine oxide, and deoxy-ADP were from Sigma. The sources of all other chemicals and biochemicals were as described previously (13, 14, 28).
Methods
Measurements of Pore Opening Mitochondrial SwellingThis was measured by the decrease in
light scattering at 520 nm as described previously (28). Liver
mitochondria were incubated under de-energized conditions at 25 °C
and 0.5 mg of protein/ml in 3.5 ml of buffer containing 20 mM Mops, 10 mM Tris, 1 mM potassium phosphate, 2 mM nitrilotriacetic acid, 0.5 µM
rotenone, 0.5 µM antimycin, 2 µM A23187,
and 150 mM KCl, pH 7.2. The calcium ionophore A23187 was
added to ensure complete equilibration of Ca2+ across the
mitochondrial inner membrane under de-energized conditions (28). In
experiments where the effects of ATP were to be studied, 0.5 µM oligomycin was also present.
A520 was monitored continuously in a
spectrophotometer with computerized data acquisition (2 data points per
s) as described previously (6). When required, energization was
achieved by addition of 2 mM succinate in the absence of
antimycin and A23187; where indicated 150 mM KCl was
replaced by 300 mM sucrose. Other additions were made as
indicated in the figure legends. The effects of and thiol
reagents on pore opening in energized mitochondria were investigated
using a procedure similar to that described by Petronilli et
al. (22). Mitochondria (5 mg of protein) were added to 7 ml of
buffer, and aliquots (3.5 ml) of this mitochondrial suspension were
then added to both sample and reference cuvettes of a split-beam
spectrophotometer (29), and A520 was monitored.
After incubation at 25 °C for 1 min in the presence of succinate,
pore opening was initiated by addition of Ca2+ (20-70
nmol/mg of protein) followed at 50-s intervals by 0.5 mM
EGTA, the reagent of interest, and finally the uncoupler carbonyl cyanide p-trifluormethoxyphenylhydrazone (FCCP) at the
required concentration.
Mitochondria were pre-swollen as described previously (13, 14), by incubating for 20 min at 30 °C and 2 mg of protein/ml under de-energized conditions in KSCN buffer (150 mM KSCN, 20 mM Mops, 10 mM Tris, 0.5 µM rotenone, 0.5 µM antimycin at pH 7.2) containing 1 mM CaCl2. Additions of TBH, diamide, or PheArs to the buffer were made as required. Swelling was terminated by addition of 1.2 mM EGTA and centrifugation at 12,000 × g for 10 min to sediment the swollen mitochondria which were then resuspended at 20 mg of protein/ml in KSCN buffer containing 2 mM nitrilotriacetic acid and 2 µM A23187. Shrinkage of these pre-swollen mitochondria at different [Ca2+] was determined by measuring the rate of increase of A520 upon addition of 7% (w/v) polyethylene glycol (PEG) as described previously (13, 14). Briefly, pre-swollen mitochondria (approximately 1 mg of protein) were added to a cuvette containing 3 ml of KSCN buffer (with 2 mM nitrilotriacetic acid and 2 µM A23187) supplemented with CaCl2 to give the desired free [Ca2+] and other additions as required. After incubation for 1 min at 25 °C, measurement of A520 was initiated and monitored continuously (10 data points per s). Shrinkage was induced by addition of 0.5 ml of 50% (w/v) PEG 2000 through a light-sealed injection port followed by rapid mixing (<1 s). Shrinkage induced by the colloidal osmotic pressure of PEG was detected as a rapid increase in A520.
Confirmation of Mitochondrial Energization Using SafraninMitochondria (1 mg of protein/ml) were incubated in the cuvette
of a double-beam spectrophotometer at 25 °C in buffer containing 125 mM KCl, 20 mM Mops, 10 mM Tris, 0.5 mM EGTA, 0.5 µM rotenone, and 10 µM safranin. Energization was determined as the
difference in absorbance at 554 and 524 nm
(A554-524) (30) and was monitored continuously.
Energization was induced either by addition of 2 mM
succinate or 0.2 mM ATP. Other additions were made as described in the figure legend. No attempt was made to calibrate
measurements.
CyP binding to mitochondrial membranes was measured in parallel incubations as described previously (14). Mitochondrial membranes were prepared by freeze-thawing mitochondrial incubations following addition of KSCN buffer containing the protease inhibitors antipain, pepstatin, and leupeptin at 1 µg/ml and phenylmethylsulfonyl fluoride at 100 µM. Detection of CyP binding to the membranes was performed using SDS-PAGE (10 µg of protein per track determined by Bradford assay using bovine serum albumin as standard) and Western blotting with N-terminal antipeptide antibodies to mitochondrial CyP using an enhanced chemiluminescence detection kit (Amersham Corp., Bucks., UK). Western blotting with antibody to NAD+-dependent isocitrate dehydrogenase was performed in parallel to correct for changes in nonspecific CyP binding.
Binding of the ANT to a Phenylarsine Oxide Column4-Aminophenylarsine oxide was conjugated to Affi-Gel 10 (Bio-Rad, prepared for reaction as described by the manufacturer). The 4-aminophenylarsine oxide was dissolved in H2O at 5 mg/ml by vigorous stirring and addition of 1 M HCl until the pH was 1.5. HEPES (50 mM) was then added and the pH brought to 7.2 with 1 M NaOH. An aliquot equivalent to 3 mg of the reagent was reacted with 3 ml of Affi-Gel 10 overnight at room temperature, followed by blocking of unreacted groups by incubation for 1 h at pH 8.0 with 0.3 M ethanolamine. Columns (usually 0.5 ml) were poured and washed with 50 mM HEPES, pH 7.2, containing 150 mM Na2SO4, 1 mM EDTA, 0.25% (w/v) Triton X-100, and the protease inhibitors antipain, leupeptin, and pepstatin at 1 µg/ml and phenylmethylsulfonyl fluoride at 100 µM (Buffer A). Coupling efficiency of the phenylarsine oxide was not determined. Mitochondria were solubilized at 0 °C and 10 mg of protein per ml in Buffer A containing additional Triton X-100 (3%). Insoluble material was removed by centrifugation at 180,000 × g for 1 min in a Beckman Airfuge, and 1 ml of sample was added to a 0.5-ml column. Unbound material was washed through the column with Buffer A until no protein was detected in the effluent. Bound protein was then eluted with Buffer A containing 10 mM dithiothreitol, and the protein concentration in each fraction (0.6 ml) was determined by Bradford assay using bovine serum albumin as standard. Samples (20 µl) of both flow-through and eluted fractions were separated by SDS-PAGE on 15% acrylamide gels, and Western blotting was performed with anti-ANT antibodies.
Expression of ResultsAll light scattering traces shown are representative of at least three separate experiments. Initial rates were determined by differentiation of traces (ignoring the response during the first second which is associated with mixing artifacts) and used to provide a relative measure of the extent of mitochondrial pore opening. The resulting data were analyzed using FigP and PFit software (Biosoft, Cambridge, UK). The dependence of pore opening on [Ca2+] is shown as smooth curves generated by least squares regression analysis to Equation 1 for cooperative kinetics:
![]() |
(Eq. 1) |
![]() |
(Eq. 2) |
![]() |
(Eq. 3) |
TBH, diamide, and PheArs all increase CyP binding to the
inner membrane of de-energized mitochondria, in parallel with their ability to increase the Ca2+ sensitivity of the MPT (8, 13,
14, 31). However, as illustrated by the data in Fig. 1,
20 µM PheArs consistently gave a greater stimulation of
the MPT than diamide, while the increase in CyP binding was
substantially less. These data suggested that PheArs must exert an
effect on the MPT independent of CyP binding. One such site of action
could be on the ability of matrix ADP to inhibit the MPT, because ADP,
in contrast to CsA, does not appear to displace membrane-bound CyP
(Fig. 2).
In order to investigate the interaction of thiol reagents with ADP inhibition of the MPT, we used the PEG-induced shrinkage assay of the MPT first introduced by Haworth and Hunter (16). The advantage of this assay is that it allows the extent of pore opening to be determined after equilibration with defined matrix [Ca2+] and [ADP]. The data of Fig. 2 investigate the interaction between ADP and CsA on the Ca2+ sensitivity of pore opening in parallel with measurements of CyP binding. The data confirm that activation by Ca2+ is cooperative (16, 17) and that ADP and CsA both decrease the Ca2+ sensitivity by shifting the activation curve to the right. Comparison of the effect of 20 and 100 µM ADP shows that higher concentrations of ADP shift the curve further to the right suggesting that ADP acts as a competitive inhibitor of the Ca2+ trigger site in agreement with the conclusions of others (17, 18). In contrast, increasing the concentration of CsA from 0.2 to 1 µM had almost no additional effect (data not shown). Furthermore, the effects of ADP and CsA were additive, confirming that they act on different sites as implied by the CyP binding experiments. It should be noted that complete removal of bound CyP by addition of CsA was never obtained. Although this may account for the CsA-insensitive pore opening seen at high [Ca2+], it is more likely that bound CyP is not essential for pore opening but rather enhances its Ca2+ sensitivity (8, 13, 14, 31).
Oxidative Stress and Thiol Reagents Decrease the Affinity of ADP for Its Inhibitory SiteIn Fig. 3 we investigate
the effects of pretreatment of mitochondria with TBH, diamide, and
PheArs on the inhibition of the MPT by ADP at a fixed
[Ca2+]. Since such pretreatment increases the sensitivity
of the MPT to [Ca2+] (13, 14), we chose to fix the
[Ca2+] at different concentrations for the control and
pretreated mitochondria in order to give similar rates of shrinkage in
the absence of ADP. However, we performed ADP titrations on control
mitochondria at two different [Ca2+] in order to confirm
that [Ca2+] did not influence the Ki
for ADP. Data are shown in Fig. 3A. The
Ki values derived using a single site model were
6.9 ± 1.0 and 8.1 ± 1.1 µM at 50 and 100 µM [Ca2+], respectively. Thus
Ca2+ does not appear to compete with ADP for a common
binding site. In contast, pretreatment of mitochondria with TBH,
diamide, and PheArs all greatly increased the Ki for
ADP, as shown in Fig. 3. The data of Fig. 3A show that 1 mM TBH and 0.1 mM diamide gave a large increase
in Ki from 8.1 ± 1.1 to 42.9 ± 2.9 µM and 91.4 ± 11.1 µM, respectively,
while the data of Fig. 3B, obtained with a different
mitochondrial preparation, show that pretreatment with 20 µM PheArs or 1 mM TBH increased the Ki from 1.7 ± 0.4 to 491 ± 27 µM and 67.0 ± 9.7 µM, respectively. The higher Ki values obtained in the control
experiments of Fig. 3A than those of Fig. 3B
reflect a variability in this parameter which we believe to reflect
oxidative stress during the pre-swelling of the mitochondria. The
presence of 1 µM CsA significantly reduced the
[Ca2+] sensitivity of the MTP, requiring a higher
[Ca2+] to induce the same rate of shrinkage. However, CsA
had no effect on the Ki for ADP in either control or
pretreated mitochondria as shown by the dashed traces of
Fig. 3B. This is consistent with the lack of any interaction
between CsA and ADP on CyP binding (Fig. 2).
In order to ascertain whether the effects of the thiol reagents were
dependent on the presence of any matrix component such as glutathione,
the effects of adding TBH, diamide, and PheArs to pre-swollen
mitochondria, which will have lost their matrix glutathione (32), were
investigated. The data of Fig. 4 show that addition of
0.1 mM diamide produced a small increase in
Ki from 9.1 ± 0.4 to 13.0 ± 0.8 µM, whereas 50 µM PheArs added directly to
the shrinking assay gave a considerable increase in the
Ki for ADP (131 ± 14 µM). TBH
was without any effect under these conditions (data not shown). These
data are in agreement with the observations of Chernyak and Bernardi
(33) who showed that TBH exerts its effect on the PheArs-sensitive
thiol group through oxidized glutathione, and it is likely that diamide
acts largely through the same mechanism. Oxidized glutathione may
oxidize two closely associated protein thiols to form a disulfide as
argued by Petronilli et al. (22) but is also capable of
forming adducts with membrane proteins under conditions of TBH-induced
oxidative stress (34, 35). An additional effect of diamide and TBH
might be to oxidize matrix thioredoxin and lipoamide which have been proposed to interact with membrane protein thiols (36). In contrast, PheArs, which attacks vicinal thiols on proteins (37), appears to be
able to attack the pore component directly, without involvement of
matrix components such as glutathione.
The ANT Is the Site of Action of Matrix ADP and Its Antagonism by Thiol Modification
The suggestion that the ANT might be involved
in pore formation (6, 38, 39) has recently been given further strong support by the demonstration that the ANT reconstituted into
proteoliposomes exhibits Ca2+-dependent channel
activity that is blocked by ADP in the presence of bongrekic acid (BKA)
(31). Previous data investigating the effects of carboxyatractyloside
(CAT) on the ADP inhibition of the MPT implied that ADP acts at two
inhibitory sites, one on the ANT and the other an additional matrix
facing site with lower affinity (15, 17, 18). We have investigated
these binding sites further using the PEG-induced shrinkage assay to
measure the effects of BKA and CAT on Ki for ADP
inhibition of the MPT at fixed [Ca2+] (65 µM). These two inhibitors trap the ANT in the c and m
conformations, respectively (40). BKA was found to have no effect on
the Ki for ADP nor on the sensitivity of the MPT to
matrix [Ca2+] under these conditions (data not shown). In
contrast CAT was found to antagonize the effect of ADP as described by
others (15-18), and this effect was analyzed more carefully in the
data of Fig. 5. The effect of CAT could not be increased
by using higher concentrations of the inhibitor and was equally
effective at 1 µM, implying very tight binding to its
inhibitory site. When data were fitted by nonlinear least squares
regression analysis assuming a single inhibitory site, a good fit was
obtained only when CAT was present, with a Ki value
(± S.E. for the fit shown) of 17.8 ± 1.4 µM.
However, in the absence of CAT the fit was poor at higher ADP
concentrations (dashed line), and a better fit was obtained when the data were refitted using a two-site model (solid
line). Derived values for Ki1 and
Ki2 were 0.63 ± 0.17 and 17.2 ± 12.8 µM, respectively, and although the
Ki2 could not be determined with great
accuracy by such a regression analysis, it would appear to be similar
to the single Ki value determined in the presence of
CAT. Indeed in three similar experiments we found that when the
two-site equation was used with the high Ki value
fixed to that obtained in the presence of CAT, a good fit to the data
was consistently obtained. The mean values (± S.E.) of the single site
Ki for ADP in the absence and presence of CAT were
1.9 ± 0.4 and 27.2 ± 5.2 µM, respectively,
whereas the high affinity Ki for the two-site model
(assuming the low affinity site to have the same Ki
as that in the presence of CAT) was 1.2 ± 0.32 µM. Thus our data are consistent with there being two sites for ADP inhibition, a high affinity site on the ANT (Ki
about 1 µM) which is only present in the m but not the c
conformation of the carrier and a lower affinity site
(Ki about 25 µM) that is independent
of carrier conformation. Neither site appears to be sensitive to CsA
since the inhibition curve was not affected by the presence or absence
of 1 µM CsA (Fig. 3), although the concentration of
[Ca2+] had to be raised from 65 to 100 µM
to give the same rate of shrinking.
The ANT has two ADP binding sites of high and low affinity, respectively (40, 41). In liver mitochondria ADP binding to the high affinity site has a Kd of 0.5 µM and is displaced by CAT (40, 42). The Kd of the low affinity site for ADP has not been reported for liver mitochondria, but in heart mitochondria it is 10-40 times greater than the Kd of the high affinity site. This would suggest that it might be about 20 µM, similar to the value measured for inhibition of pore opening. Thus ADP might be exerting its effect entirely through an interaction with the ANT but on two distinct sites. The presence of two such AdN binding sites is also consistent with the emerging model of the mechanism of the ANT, which proposes a sequential model with two internal and two external substrate binding sites, one high and one low affinity site on each side of the membrane (41, 43, 44).
We have investigated the specificity of the nucleotide binding site for inhibition of pore opening by investigating the concentration dependence of inhibition by a variety of nucleotides. Additional CaCl2 was added where necessary to correct for the chelation of Ca2+ by some nucleotides such as ATP. The derived Ki values for ADP, deoxy-ADP, and ATP (± S.E. for a single site model) were 1.7 ± 0.4, 23.2 ± 2.4, and 544 ± 75 µM, respectively. The ability of deoxy-ADP to inhibit the pore with a 10-15-fold lower affinity than ADP is similar to the difference in Km of the ANT for ADP and deoxy-ADP (40). At 1 mM AMP, GDP, GTP, NAD+, NADP+, NADH, and NADPH were all without significant effect on the pore, correlating with their lack of binding to the ANT (40). CAT did not appear to have a significant effect on the inhibition caused by deoxy-ADP (data not shown), in contrast to its ability to raise the apparent Ki for ADP by more than 10-fold. This may indicate that deoxy-ADP is binding only at the low affinity ADP binding site.
We have confirmed that the large difference in the inhibitory potency
of matrix ADP and ATP is not an artifact of using pre-swollen mitochondria in KSCN medium. Mitochondria were incubated in KCl medium
under de-energized conditions in the presence of oligomycin, and either
100 µM ADP or ATP was added before initiation of pore opening with 200 µM Ca2+. In the absence of
ATP will enter in exchange for ADP, both nucleotides moving down
their concentration gradients, and matrix [ATP] will be high relative
to [ADP]. In contrast, after addition of ADP matrix [ATP] will be
very low relative to [ADP]. In the presence of ADP, little or no pore
opening was observed, whereas rapid swelling occurred when ATP was
present (data not shown). This confirms that ATP is a relatively
ineffective inhibitor of the pore in comparison to ADP.
Our data suggest that PheArs decreases binding of ADP to the
ANT, implying that this reagent must bind to specific thiol groups on
the carrier. We have sought to demonstrate this directly by synthesizing a phenylarsine oxide affinity column (see "Methods") and investigating the binding of the ANT to it. The data of Fig. 6 confirm that such binding occurs and is prevented by
pretreatment of mitochondria with 20 µM CAT or PheArs
before membrane solubilization. Binding was also greatly reduced by
pretreatment with 0.1 mM diamide or 1 mM TBH
(data not shown), confirming that these two reagents can also attack
the thiol group modified by PheArs. This would be predicted from their
ability to lower the affinity of ADP for its inhibitory site.
The ANT is known to have three reactive cysteines (Cys56, Cys159, and Cys256) that are all on putative matrix-facing loops and whose thiol groups show differential reactivity to various thiol reagents in a conformation-dependent manner (45-47). Cys159 can be selectively labeled in submitochondrial particles with the membrane-impermeant thiol reagent, eosin 5-maleimide, which also inhibits transport. Labeling is decreased in the presence of ADP (45), suggesting that this residue is at or close to an ADP binding site, consistent with previous studies using lysine-specific labeling reagents that identified Lys162 and Lys165 as being involved in substrate binding (48). Thus, Cys159 would seem to be a possible locus for the effects of thiol reagents that antagonize ADP inhibition of the mitochondrial pore. We have tested this possibility by investigating the ADP sensitivity of the MPT in mitochondria incubated with 40 µM eosin maleimide during the pre-swelling. The Ca2+-dependent pore opening in these mitochondria became extremely insensitive to ADP inhibition, with only about 20% inhibition being observed at 0.5 mM ADP (data not shown). Thus it would seem possible that Cys159 is the locus of action of thiol reagents that affect the ADP sensitivity of the MPT. In parallel experiments we attempted to investigate whether eosin maleimide had any effect on CyP binding to the inner membrane but found that this reagent led to the formation of protein aggregates during SDS-PAGE that prevented consistent quantitative Western blots.
Another thiol group is probably involved in CyP binding to the inner membrane, and this might be Cys56 of the ANT. Thus at low concentrations (<50 µM) N-ethylmaleimide (NEM) binds to Cys56 only when the carrier is in the m conformation (40, 49), whereas LeQuoc and LeQuoc (50) have demonstrated that oxidative stress decreases NEM binding to the ANT, implying that such treatment modifies Cys56. Indeed, Zwizinski and Schmid (51) have reported that exposure of cardiac mitochondria to oxidative stress leads to an inhibition of ANT activity and a slight decrease in its mobility on SDS-PAGE. More striking is the demonstration by Majima et al. (47) that treatment of sub-mitochondrial particles with copper o-phenanthroline cross-links the ANT dimer by forming a disulfide bridge between the Cys56 residues of the two monomers. This reagent causes the formation of a disulfide bridge between two adjacent thiols and thus may well reflect the action of TBH and diamide. Indeed, these reagents have been shown to cross-link proteins in mitochondrial membranes (52).
The Effects of the MitochondrialThe mitochondrial greatly
increases the efflux of matrix ATP on the ANT in exchange for ADP. This
electrophoretic transport must entail an effect of
on either AdN
binding to the ANT or on the translocation cycle (24, 25, 53). An
increase in ATP binding to the matrix nucleotide binding site of the
ANT caused by increased
is likely (24), and this would be
capable of explaining the inhibitory effect of
on the MPT. Such
an effect might be reflected by the ability of ATP to inhibit the MPT
more effectively in energized than de-energized mitochondria. In order to investigate this, manipulation of the matrix ATP/ADP ratio in
energized mitochondria was necessary. In the absence of oligomycin, succinate is well known to establish a high
and an ATP/ADP ratio
of about 3 (24). However, if oligomycin is added before addition of
succinate, oxidative phosphorylation is prevented and ATP/ADP ratios
are an order of magnitude lower (24). In Fig. 7 we show
that when energized mitochondria were loaded with moderate amounts of
Ca2+ (50 µM), in the absence of oligomycin
(i.e. low [ADP]), a small amount of pore opening
(swelling) was apparent which was enhanced by addition of increasing
concentrations of uncoupler. In the presence of oligomycin
(i.e. high [ADP]) less pore opening was observed, and the
process was less sensitive to uncoupler. These differences were
exaggerated at higher Ca2+ (80 µM) as shown
in Fig. 7B. Our data could be explained if
enhances
the binding of AdNs to the inhibitory site of the MPT pore, ATP binding
being influenced more than that of ADP. This would be consistent with
the conformation-dependent binding of AdNs to the ANT (24,
25, 53).
If the explanation above is correct, the MPT in mitochondria depleted
of AdNs by treatment with PPi (27) should be insensitive to
. The data of Figs. 8 and 9
demonstrate that this is the case. It was necessary, first, to
establish that PPi-treated mitochondria are able to
maintain a high
in the presence of succinate. This was confirmed
by monitoring
with safranin as shown in Fig. 8. Addition of
succinate to the de-energized mitochondria gave a similar change in
safranin signal for both control and PPi-treated mitochondria, implying the generation of a similar
in both. In
contrast, addition of ATP to energize the mitochondria produced a much
smaller change in safranin signal in the PPi-treated
mitochondria. This would be predicted since ATP is unable to enter the
mitochondria on the ANT except in exchange for ADP, whose concentration
is greatly reduced when the mitochondria are depleted of endogenous AdNs.
In Fig. 9A we confirm the effects of thiol reagents on pore opening in control mitochondria induced by membrane depolarization with uncoupler. Mitochondria incubated with 70 µM Ca2+ alone did not demonstrate significant pore opening. However, addition of EGTA after 50 s followed by addition of uncoupler did initiate swelling, the extent of which was increased as the concentration of uncoupler was increased. As described by Bernardi and colleagues (19-23), the sensitivity toward uncoupler was greatly increased by the presence of diamide, TBH, or PheArs. In contrast, mitochondria that had been pretreated with PPi to deplete them of endogenous AdNs were insensitive to uncoupler whether or not they had been treated with thiol reagents (Fig. 9B). Furthermore, as expected for mitochondria whose matrix has been depleted of AdNs, sensitivity to [Ca2+] was very much greater. Thus swelling was apparent 30 s after addition of 30 µM Ca2+ and was not further stimulated by additions of EGTA and FCCP, implying that depolarization had no effect on the MPT under these conditions. Higher concentrations gave faster rates of swelling, again unstimulated by uncoupler (data not shown). In fact, in contrast to control mitochondria, it can be seen from the data of Fig. 9B that addition of EGTA and uncoupler to the PPi-treated mitochondria after calcium loading actually inhibited swelling. The data of Fig. 9B also demonstrate that TBH, diamide, and PheArs had little or no effect on the swelling of energized PPi-treated mitochondria. This is consistent with these reagents sensitizing the MPT to depolarization through an effect on the binding of matrix AdNs to the ANT; with little or no AdNs present in these mitochondria no sensitizing effect would be possible.
ConclusionsOur data imply that thiol reagents can increase
the Ca2+ sensitivity of the MTP by two mechanisms: first
through an increase in CyP binding to its membrane target protein on
the pore complex, and second through a decrease in ADP binding to its
inhibitory site(s). Both of these actions could be exerted on the same
target protein, and the evidence discussed above continues to make the ANT an attractive candidate as suggested in our original model of pore
opening (6, 7). The recent demonstration that the purified and
reconstituted ANT can form Ca2+-dependent
channels that are blocked by ADP and BKA provides additional strong
evidence for this hypothesis (31). Our data also provide an explanation
for the observed voltage dependence of the MPT and the ability of thiol
reagents to allow pore opening at higher (22, 23). We suggest
that the ANT itself may be the voltage sensor, because its binding of
matrix AdNs is
-dependent and antagonized by thiol
group modification. We believe that our data allow the regulation of
the MPT by a variety of effectors to be grouped into three broad
categories: those that modify AdN binding to the ANT, those that affect
CyP binding, and those that interact directly with the Ca2+
trigger site. These are summarized in Table I.
|
An involvement of the unique mitochondrial CyP-D (9, 10) is now widely accepted and accounts for the ability of CsA analogues to inhibit pore opening with the same affinity as they inhibit the peptidyl-prolyl cis-trans-isomerase activity of CyP-D (7, 8, 54). We have shown that CyP-D binding to its membrane target protein is increased by a variety of procedures that also sensitize pore opening to [Ca2+] including oxidation of matrix NADH, oxidative stress induced by TBH, thiol reagents, increased matrix volume, and chaotropic agents (13, 14). As outlined above, we suggest that Cys56 may be the thiol group whose modification is responsible for the increased CyP binding by thiol reagents. This is close to Pro61 that we suggested in our original model of pore opening might bind CyP (6, 7). In this context, it may be significant that using a wide range of conditions we have been unable to demonstrate the MPT is mitochondria from yeast (Saccharomyces cerevisiae), and the three ANT isoforms of these mitochondria all lack Cys56 and Pro61 both of which are conserved in mammalian mitochondrial ANT isoforms (55, 56).
Our original model for the mechanism of pore formation suggested that mitochondrial CyP binding to the ANT in the presence of Ca2+ might induce a conformational change sufficient to form the pore (6, 7). However, on the current balance of data, it seems likely that the pore may be able to open in the absence of CyP binding but that this requires substantially higher [Ca2+]. Thus studies on the megachannel of patched clamped mitochondria, which probably reflects pore opening, have shown that the inhibitory effect of CsA is overcome at higher [Ca2+] (57, 58). Studies of Novgorodov et al. (17), Crompton and Andreeva (59), and ourselves (Fig. 2 and Ref. 17), using different techniques, have shown that at high matrix [Ca2+] inhibition of the MPT by CsA is overcome, especially in the absence of matrix AdNs. Finally, there is the recent demonstration that the purified and reconstituted ANT can form Ca2+-dependent channels resembling the permeability transition pore (31). Thus the evidence favors a modulatory role of CyP on pore opening rather than an essential role.
The regulation of the MPT by matrix AdNs, the conformation of the ANT,
and were all without effect on CyP-D recruitment to the membrane
(Fig. 2 and Ref. 14), implying an alternative mode of action for these
effectors. The data we present here suggest that the ANT provides a
common locus for these effects. Binding of matrix AdNs to the ANT
inhibits pore formation, and this binding is antagonized by thiol
reagents, the c conformation of the carrier, and membrane
depolarization. In contrast the m conformation and increased
enhance binding and inhibit the pore. As oulined above, we suggest that
Cys159 may be the thiol group responsible for the
antagonistic effects of thiol reagents on ADP inhibition of the pore.
The binding of ADP to its regulatory site(s) does not appear to be
influenced by [Ca2+] (Fig. 3A) suggesting that
there is not direct competition between ADP and Ca2+ but
rather that ADP causes the interconversion of the
Ca2+-binding component of the pore from a high affinity to
a low affinity conformation. In contrast to Novgorodov et
al. (17), we find no evidence for CsA influencing the affinity of
ADP binding to its inhibitory site(s) (Fig. 3B). Rather its
effect to desensitize the pore to [Ca2+] appears to be
additive with respect to ADP (Fig. 2) suggesting an alternative site of
action. Thus our data are compatible with ADP and CyP binding to
distinct sites on the ANT and inducing, respectively, a decrease or
increase in the affinity of Ca2+ binding to the site which
triggers pore opening. We suggest that the inhibitory effects of
Mg2+ and H+ (low pH) are through direct
competition with Ca2+ at its binding site (14, 16, 28, 60).
Calmodulin does not appear to be associated with the
Ca2+-binding site since the calmodulin antagonist
trifluoperazine (1-100 µM) gave no inhibition of the
rate of shrinkage of pre-swollen mitochondria (data not shown).
An association of the mitochondrial benzodiazepine receptor with porin and the ANT has been reported (61), and the possibility that this might represent the pore complex or some part of it is attractive (see Ref. 3). However, using the shrinkage assay we failed to detect any effect of diazepam or protopophyrin IX, both ligands for the mitochondrial benzodiazepine receptor (62), over a range of concentrations (50 nM to 50 µM).
We thank Dr. Gerard Brandolin for the generous gift of antibodies against bovine ANT.