(Received for publication, February 27, 1997, and in revised form, May 27, 1997)
From the Department of Membrane Transport Biophysics,
Institute of Physiology, Academy of Sciences of Czech Republic,
CZ-14220 Prague 4, Czech Republic and the
Zoologisches Institut,
Biologiezentrum, Christian-Albrechts-Universität zu
Kiel, D-24118 Kiel, Germany
Single-channel and whole-mitoplast patch-clamp
recordings were employed to characterize the 108-pS
(Cl) channel in brown fat mitochondrial mitoplasts.
We demonstrated the ability of this channel to conduct di- and
trivalent anions, such as sulfate, phosphate, and
benzenetricarboxylates, and its blockage by propranolol,
1,4-dihydropyridine-type Ca2+ antagonists, and Cibacron
blue. Moreover, we have revealed its pH dependence for the first time.
As a basis for the characteristic potential dependence of the
whole-mitoplast current, we identified an open probability, increasing
with depolarizing (positive) potentials, Eh,
and being almost zero in the hyperpolarizing range. Events at negative
Eh exhibit a short flickering behavior, whereas
at positive Eh, they become much longer. This
voltage dependence is influenced by pH in such a way that, at acidic
pH, the 108-pS channel possesses a low open probability throughout the
observed potential range, whereas at alkaline pH, the channel switches to long openings, even at a negative potential. All these properties lead us to conclude that the inner membrane anion channel, which has
been characterized only by light scattering studies, and the 108-pS
inner membrane channel, which has been characterized
electrophysiologically, are one and the same process.
The 108-pS channel was the first ion channel to be discovered in mitochondria (1), nevertheless its function remains unknown, and it has not been identified with any known mitochondrial transport activity. It has been also referred as the inner mitochondrial membrane channel (2) or mitochondrial centum picosiemens channel (3). The 108-pS channel has been detected in patch-clamp experiments on mitoplasts from liver (1, 3-5), heart (5), brain (6), and brown adipose tissue (BAT; 7)1 and was characterized as pH-insensitive (1). First attempts to inhibit this channel with known channel blockers failed (5). In BAT mitochondria, the 108-pS channel displays infrequent substates, the main state having a conductance about half that of the fully open state (7), and was reported to be partially blocked by purine nucleotides (7). In mitochondria from other tissues, antimycin (8), protoporphyrin IX, and ligands of the mitochondrial benzodiazepine receptor (9) block the channel. Amiodarone and propranolol decrease the open probability, while increasing the conductance of the channel (10). Sorgato (1) and others (2) recognized the possibility that the 108-pS channel may reflect conductance through the inner membrane anion channel (11, 12); however, many observations appeared to conflict with this identification (1, 2).
Anion uniport in Mg2+-depleted mitochondria was attributed to an inner membrane anion channel (IMAC) by Garlid and Beavis in 1986 (13). The transport properties of IMAC have subsequently been characterized in detail by Beavis and co-workers (12, 14-19). IMAC is inhibited by Mg2+ and greatly activated at alkaline pH. It is nonselective for anions and conducts mono-, di-, and trivalent anions (15, 16, 20). Known inhibitors include dicyclohexylcarbodiimide (15), propranolol (17), organotin compounds (18), sulfhydryl reagents (12, 19), and dyes (12), such as Cibacron blue 3GA (21). IMAC is also inhibited by Ca2+ channel (L-type) antagonists of the DHP class such as niguldipine (22) and of phenylalkylamine and benzothiazepine classes (22). Ca2+ antagonists bind to specific binding sites in mitochondria, presumably on a putative IMAC protein (22). Although no inhibition of IMAC with nucleotides has been found, this as yet unidentified protein might also bear a structure similar to a nucleotide-binding site (21), because its inhibitor Cibacron Blue is known to bind to such sites.
Beavis (12) has proposed that the physiological function of IMAC is to contribute to the contractile phase of volume homeostasis, but this hypothesis has not been tested. Characterization of IMAC has been entirely phenomenological, relying almost exclusively on light scattering (matrix swelling) studies. The protein or gene responsible for the phenomenology has not been identified, nor has the process been reconstituted into liposomes. It is not known, in fact, whether IMAC is a channel at all, as had been proposed (13).
For molecular characterization of IMAC, it would be very useful to identify its corresponding channel activity. To this end, we decided to re-examine the possibility that the 108-pS channel reflects IMAC activity by applying patch-clamp to mitoplasts of BAT mitochondria, which have been shown to contain IMAC (23). We found that the 108-pS channel exhibits the salient properties of IMAC, including substrate specificity, inhibitor specificity, and pH dependence. We conclude that the 108-pS channel is responsible for IMAC activity.2
Chemicals purchased from Merck (Darmstadt, Germany) were: CaCl2, NaCl, KCl, K2SO4, and sucrose; from Sigma (Deisenhofen, Germany): Cibacron blue 3GA, 1,2,3-benzenetricarboxylic acid, 1,3,5-benzenetricarboxylic acid, EGTA, GDP, HEPES, and propranolol; and from Research Biochemicals International (Natick, MA): nifedipine and niguldipine.
Mitochondria were prepared from interscapular brown adipose tissue from 7-21-week-old male hamsters (Mesocricetus auratus), cold-adapted at 6 °C for 4 weeks. Mitochondria from 1 g of BAT were prepared (24) in a medium containing 250 mM sucrose, 5 mM K-HEPES, 1 mM K-EGTA, and 0.1% bovine serum albumin (pH 7.2). The final two washes were performed in 150 mM KCl, 20 mM K-HEPES, 1 mM K-EGTA, and 0.1% bovine serum albumin at pH 7.2. Mitochondria were resuspended in 1 ml of hypertonic medium (750 mM KCl, 100 mM K-HEPES, 1 mM K-EGTA (pH 7.2)) and stored on ice for up to 48 h. Mitoplasts were obtained by hyposmotic shock: removal of the outer membranes of the shrunken ice-cold mitochondria was carried out by sonication (5 × 2-s intervals). Subsequently, the ionic strength was lowered 10-fold by addition of a hypotonic medium (5 mM K-HEPES, 1 mM K-EGTA (pH 7.2)). After a further 1-2-min incubation at room temperature, isotonicity was restored by the addition of the hypertonic medium described above. Mitoplasts were kept on ice for a maximum of 12 h.
Patch-clamp experiments followed the method of Hamill et al.
(25). Borosilicate glass pipettes were polished to give a resistance of
9-15 M and were filled by a pipette solution of 150 mM
KCl, 20 mM K-HEPES, 1 mM K-EGTA (pH 7.2). Bath
solutions were applied by inserting the pipette tip into a pipe of a
peristaltic pump-driven "sewer-pipe flow system." Free-floating
mitoplasts were "chased" by means of an electrically driven
micromanipulator and moved to their final position on the pipette tip
by gentle suction. Gigaseals of 5-20 G
formed spontaneously in
about 80% of the cases, thus improving the method previously reported
(7). Signals were recorded by a L/M-EPC 7 patch-clamp amplifier,
filtered by a 4-pole Bessel filter at a corner frequency 1 kHz, and
sampled at 2.5 kHz. Data recorded during the slow voltage ramps were
sampled at 16 Hz. Processing of the data was carried out by means of
the pCLAMP6 program package (Axon Instruments, Foster City, CA).
Experiments were performed at room temperature (22-27 °C). Various
blockers were added by using the described flow system, which gave a
constant flow of the blocker around the mitoplasts, and records were
repeated every 2-3 min, thus monitoring the approach to the
steady-state equilibrium, which was reached usually after 10-13 min.
Measurements in symmetrical sulfate were made by using an
Ag|AgCl|KCl salt bridge (150 mM KCl in 1% agar) in the
patch-clamp pipette interior and with mitoplasts washed solely in
sucrose medium.
Data were analyzed by employing plots of integral conductance
G = I/Eh
versus holding potential Eh. With BAT
mitoplasts, these plots showed a sigmoidal shape. The curves
represented the transition from an ensemble of mostly closed channels
to an ensemble in which the open channels dominated. This allowed the
conductance to be determined around 0 mV (which is inaccessible in
I-Eh plots) and, the upper limit of
the leak conductance GL determined as
G near 40 mV. The plots were fitted by using the modified
equation of Bräu et al. (26),
![]() |
(Eq. 1) |
Typical current-voltage
(I-Eh) characteristics as obtained in
the "whole-mitoplast mode" (by analogy to the "whole-cell
mode") in symmetrical 150 mM KCl medium (pH 7.2) with
voltage ramps between 40 and 40 mV are illustrated in Fig.
1a. To avoid time hysteresis, the ramps had to be very slow (2 min). A typical feature of the I-Eh characteristics was their
asymmetry in the voltage range. The current was one to 2 orders of
magnitude lower at negative potentials than at positive potentials. The
transition between these domains was observed in the region of
10 to
35 mV, with inflection occurring between
10 and 16 mV
(n = 30) as apparent from plots of an integral
conductance G versus Eh
(I/Eh versus Eh plots, Fig. 1b). A low G in
the range below
20 mV reflected a state in which most of the channels
were closed, whereas above 35 mV nearly all channels were open as
indicated by saturation of the integral conductance.
The reason for the observed asymmetry was a changing pattern of channel
fluctuations as detected by the single-channel recordings in the
mitoplast-attached mode (Fig. 2,
a and c) and a changing pattern of the observable
channel events in the whole-mitoplast mode (Fig. 2, b and
d). The former yielded statistically reproducible conductances of 109.6 ± 0.7 pS (S.D., n = 5, 24 °C) in 150 mM KCl. The derived temperature dependence
(Fig. 3) yielded a value of 1.2 pS·K1, explaining the scatter in values of the 108-pS
channel conductances reported in the literature (2). A distinct marker
property of the 108-pS channel (Figs. 2, a and b,
and 4a) was a flickering behavior, i.e. bursts of
frequent short openings, at more negative (hyperpolarizing) potentials
(clearly visible in Fig. 4a
and between
30 and
10 mV in Fig. 2c) and long time
openings at positive (depolarizing) potentials. This behavior seemed to
be the main cause for the observed asymmetry of the
I-Eh characteristics. Consequently, the Cl
uptake into the mitoplast was much
higher in comparison with the Cl
efflux. This asymmetry
was a distinct electrophysiological marker of the 108-pS
channel.3 Those
G-Eh plots, in which the
single-channel conductance was visible (Fig. 2, c and
d), demonstrated the constancy of the apparent channel
conductance at all voltages applied. Moreover, infrequent substates
were observed, corresponding to channel openings at about
(Fig. 2c), 1/2, and
of the main full-open
amplitude.
Propranolol, Cibacron Blue, and DHP Ca2+ Antagonist Blockage of the 108-pS Channel
The single-channel events after
transferring the patches into a medium containing 50 µM
propranolol, an amphiphilic amine better known as a -receptor
antagonist, are illustrated in Fig. 4a. At all voltages
applied, propranolol blocked the 108-pS channel completely. Moreover,
the block was reversible, as apparent when flickering behavior (at
negative voltages) or long openings (at positive voltages) were
restored after washing. Unlike Antonenko et al. (10), we did
not observe an increase in the single-channel conductance with
propranolol.
In the whole-mitoplast mode and after 10-13-min incubation in the
presence of saturating amounts (constant flow) of propranolol (Fig.
4b), Cibacron blue 3GA (Fig.
5a), or the DHP
Ca2+ antagonist nifedipine (Fig. 5b), the
typical current in depolarizing potentials was suppressed close to the
magnitude of the leak current (Fig. 4b and insets
in Fig. 5). Another DHP Ca2+-antagonist, niguldipine (35 µM, Fig. 5c), caused only partial inhibition
(43%). Moreover, intermediate concentrations of propranolol (Fig.
4b) lead to a partial suppression of the integral
conductance G, whereas the half-maximum parameter
EM of the theoretical fit (cf. Eq. 1)
increased with increasing propranolol concentration. With 50 µM propranolol, the threshold was shifted from 20 mV in
controls to 10 mV, and EM increased to around 63 mV (Fig.
4b). In addition, the G versus Eh
plots indicated that the current at negative (hyperpolarizing)
potentials was also reduced. The dose-response curves constructed from
G values at 35 mV (Eh at which
G is saturated in controls without blocker) and corrected for the leak conductance are illustrated in Fig.
6. Thus, the 108-pS channel in the
whole-mitoplast mode was inhibited by propranolol (Fig. 6a)
and Cibacron blue (Fig. 6b) with apparent
Ki values of 2.8 µM and 0.15 µM, respectively.
Sulfate, Phosphate, and Benzenetricarboxylate Conduction by the 108-pS Channel
Single-channel recordings in symmetrical 100 mM (Fig. 7a) and
150 mM potassium sulfate media demonstrated channel events
with a similar pattern of fluctuation in negative and positive
Eh as those found in KCl medium; they exhibited
unit conductances of 34 and 56 pS, respectively. In the whole-mitoplast
mode (Fig. 7b), sulfate currents also exhibited the same
asymmetrical I-Eh relationship and
were sensitive to propranolol, which inhibited at higher concentration
than that in KCl medium (Fig. 7b). Therefore, we concluded
that the same channel was responsible for both sulfate and
Cl conductances.
Only noise was observed at positive Eh when a
potassium gluconate medium replaced the previous anions (7). Small
(10% with regard to Cl at 20 mV) and intermediate (20%
of Cl
) currents were observed when the external KCl was
substituted by potassium salts of 1,3,5-benzenetricarboxylate (BTC,
Fig. 8a) and phosphate (Fig.
8b), respectively. However, a different behavior was found
in the presence of external 1,2,3-BTC, when the
I-Eh characteristics had completely
lost their asymmetry in the whole-mitoplast mode (Fig. 8a, trace
1). At negative voltages, where the Cl
efflux mainly
contributed to the total current, one could observe an increase in leak
current between
40 and
25 mV. Between
20 and 0 mV, smaller
absolute values of current than those in symmetrical KCl medium were
apparent, indicating a possible trans-inhibition of the
Cl
efflux by the external 1,2,3-BTC. At positive
voltages, the I-Eh plot showed the
true characteristics of the 1,2,3-BTC current. The single-channel
conductance carried by 1,2,3-BTC was apparently lower than that carried
by sulfate, if one assumed the same number of channels to be open.
Increase in 108-pS Channel Activity at Alkaline Extramitoplast pH
Transference of mitoplast-attached patches to KCl media of
variable pH during single-channel recording (Fig.
9) clearly demonstrated that pH affected
the fluctuation pattern of the 108-pS channel. At pH 6.0, the long
channel openings were rare, even at high positive potentials, whereas
at pH 7.2, shorter events at negative Eh and long channel openings (or clusters, as they were interrupted by short
closures) at positive Eh were predominant, as
illustrated above (Fig. 2a). At pH 8.5, the channel was so
active that, at hyperpolarizing (negative) potentials, the open
probability was high, and activity appeared as a dense flickering. At
positive Eh, the clusters of channel openings
were slightly prolongated. Thus, contrary to the report of Sorgato
et al. (1), we observed that the 108-pS channel was
pH-dependent. The pH effect was however reversible, since
on transferring patches back to the original medium at pH 7.2, the
characteristic fluctuation pattern described for pH 7.2 (Figs.
4a and 9) was completely restored.
Moreover, whole-mitoplast recordings at various pH (Fig.
10, a and b)
confirmed that the asymmetry of the
I-Eh characteristics was
pH-dependent. At pH 6.0, the whole-mitoplast current was
lower than at neutral pH within the complete range of applied voltages up to 32 mV (Fig. 10b). At pH 8.5, the absolute value of
current at negative Eh was higher than at pH
7.2, whereas at positive Eh, the current was
slightly lower. However, when the mitoplast was transferred back to pH
7.2, the current at positive Eh did not return
to its original value. Hence, the marginal decrease in current at pH
8.5 was not attributable to the function of the 108-pS channel. Plots
of integral conductance versus potential (G versus
Eh plots) illustrated the pH effect more clearly
(insets in Fig. 10). At pH 6.0, both the threshold of
transition and EM were shifted by about 10 mV to
positive voltages, whereas at pH 8.5, the G versus
Eh characteristic was shallower and began at much
higher G, and G gradually increased to saturation
over the whole Eh range. This corresponded to
single-channel recordings, where 108-pS channels were predominantly
open (Fig. 9). One could assume saturation of G at
Eh higher than 80 mV.
With several exceptions (2, 27), no channel phenomena have been
correlated up to date with the known ion transport proteins in
mitochondria, i.e. the described uniporters. On the
contrary, when inserted into giant liposomes and patch-clamped, the
known purified anion carriers, such as the ADP/ATP carrier (28), the Pi carrier (29, 30), and uncoupling protein (31), display completely different properties, namely switching to the uniport mode
and translocating Cl and other nonphysiological
substrates (28-30). In this work, we provide the evidence that the
108-pS channel is identical with the phenomenon of IMAC.
We have further characterized the 108-pS channel in BAT mitoplasts and demonstrated clearly its ability to conduct di- and trivalent anions and its blockage by DHP-type Ca2+ antagonists and by Cibacron blue; we have also shown its pH dependence for the first time. Moreover, we have identified a basis for the asymmetry of "whole-mitoplast" currents to which it mostly contributes, i.e. the voltage sensitivity of channel openings, which are almost absent at hyperpolarizing (negative) potentials. Their frequency increases, however, with increasing voltage, so that at positive voltages, the open states are "packed" into single long openings. Moreover, this voltage dependence is influenced by pH, as we found contrary to Sorgato et al. (1) in the mitoplast-attached mode: at acidic pH, the 108-pS channel exhibits very little activity with a fluctuation pattern identical to that found at negative potentials at neutral pH. At neutral pH and positive potentials, the channel switches to long openings as it does over the whole potential range at alkaline pH. Such activation of the 108-pS channel at alkaline pH resembles the phenomenon of alkaline pH-dependent swelling of mitochondria first described by Azzi and Azzone (32) and involves a dicyclohexylcarbodiimide-sensitive anion pore (15, 33, 34). This has subsequently been attributed to the phenomenon of the IMAC, which has been studied biochemically by Beavis and co-workers (12, 14-19). They have pointed out the dependence of the IMAC on internal pH. Nevertheless, in the mitoplast-attached mode (single-channel recording), external pH could influence a putative internal protonation site via a leak pathway. In the whole-mitoplast mode, when intramitoplast (matrix) pH is controlled by the buffer of the pipette solution, the external pH effect is analogous, but much less pronounced, than that in the mitoplast-attached mode. Thus, the sensitivity of the channel to changes in the matrix pH is the first important property that the 108-pS channel and IMAC have in common, suggesting their possible identity.
Respiring mitochondria possess a negative potential, which will tend to
drive anions outward through IMAC. On the other hand, most measurements
on IMAC were made at nearly zero potential, at which anions move inward
(16). To compare ion selectivities with those of the 108-pS channel, we
must compare currents of our patch-clamp study taken as the positive
voltages approach zero. Under these conditions, the estimated order of
conductances Cl > SO42
> Pi
1,2,3-BTC > 1,3,5-BTC for the 108-pS
channel is almost identical to the order of rates reported for IMAC
(16). Thus, in addition to Cl
, the 108-pS channel
conducts divalent anions, such as sulfate and
HPO42
, and trivalent 1,2,3-BTC. The
low currents observed in 1,2,3-BTC at positive potentials could partly
result from trans-inhibition by Cl
from the
mitoplast interior, as Cl
was replaced only externally in
this case. On the other hand, Cl
efflux (current at
negative Eh) was trans-inhibited by
external 1,2,3-BTC. In conclusion, the observed anion pattern seems to be similar for the 108-pS channel and IMAC, thus again suggesting their
identity.
The third piece of evidence for the identity of the 108-pS channel and IMAC comes from the finding of a similar pattern of blockers. Pharmacological determination of channels is a classic paradigm, but previously, only propranolol, amiodarone (10), and benzodiazepines have been found to be blockers that the 108-pS channel and IMAC have in common (2). Now, we have discovered two new types of common blockers of the 108-pS channel and IMAC and have demonstrated more clearly the effect of propranolol. For each drug tested, its potency in inhibiting the 108-pS channel was found to be in the same concentration range as its reported potency for inhibiting IMAC (2, 12). Thus, the DHP-type Ca2+ antagonist nifedipine (cf. Fig. 5b and Ref. 22) and the amphiphilic amine propranolol (Figs. 4, a and b, 6a, 7b and Ref. 10 versus Ref. 12) inhibit in a micromolar range, whereas the antraquinone dye Cibacron blue 3GA (Fig. 5a versus Ref. 34) inhibits even in the submicromolar range. Unlike previous studies (10), our measurements show a complete block of the 108-pS channel activity by propranolol, besides a complete block by Cibacron blue and nifedipine. The nature of the blocker action corresponds to noncompetitive inhibition of transport, since we have observed shifts in the EM of G transition, i.e. shifts in the saturation of the integral conductance in the whole-mitoplast mode toward higher voltages with all blockers tested. Hence, with a blocker at low or intermediate concentration, a low opening probability is manifested only at high positive voltage. Thus, the blockers act in a similar way as acidic pH. The effect resembles the reported antimycin-induced shift of curves relating the open probability to Eh (8).
The only (as yet unexplained) discrepancy concerns the block by nucleotide di- and triphosphates (7), which have never been shown to inhibit IMAC in biochemical studies (16) or BAT mitochondria (34). In patch-clamp studies, a partial block of 44-84% was reported by Klitsch and Siemen (7) with BAT mitoplasts from warm-adapted rats, whereas there is only an occasional block in our recent experiments with BAT mitoplasts from cold-adapted hamster.4 Nevertheless, since nucleotides prevent the noncompetitive binding of DHP Ca2+-antagonists (22), their effects have to be studied further.
The three lines of evidence described above lead us to the conclusion that the 108-pS channel is identical to the biochemically observed phenomenon of the IMAC. This is not only because mitochondriologists have called it "a channel" (13), but because the similarities shown above are pronounced. To our knowledge, this is the first positive correlation of the single-channel measurements of the 108-pS channel with biochemical IMAC studies. Moreover, when single-channel measurements were attempted with mitoplasts loaded with 1 mM MgCl2, no channels with 108-pS characteristics were found.4 This is in accordance with IMAC inhibition by matrix Mg2+.
One may compare the conductance of the 108-pS channel in KCl and the
maximum rate found for Cl uniport by the IMAC as follows.
Vmax of 1.35 µmol·min
1·(mg
of protein)
1 was reported at [H+] limiting
to zero (12). To reach 108 pS at 10 mV, i.e. to reach the
current of 1.08 pA that accounts for a turnover of
6.75·106 s
1. Taking these data, the channel
content is equal to 3.3 fmol/mg of mitochondrial protein. Further
adjustment of the open probability and different potentials would yield
the estimated content of 10-100 fmol/mg of protein, which appears
reasonable for a channel abundance.
Identification of the 108-pS channel with IMAC does not exclude the
possibility that this activity represents an altered state of a known
carrier.5 Each integral
membrane transport protein, biochemically defined as a carrier and
physiologically mediating the electroneutral symport or antiport, can
exhibit the behavior of an anion channel, while switching to a
nonselective uniport mode. Such behavior has been reported for the
ADP/ATP carrier (28), Pi carrier, glutamate/aspartate
carrier (29, 30), and the uncoupling protein (31). Although all but the
uncoupling protein do not physiologically allow for the uniport mode,
they exhibit channel behavior in giant liposomes, conducting
nonstandard substrates such as Cl. The uniport mode of
these carriers, which can be inhibited from the trans-side
by other anions, has also been induced by mercurials, namely by
HgCl2 (29, 30). For a definitive answer of this question,
the protein responsible for both the biochemical IMAC phenomenon and
the 108-pS channel must be found.
Stimulating discussions with Prof. Dr. Reinhard Krämer (Institute of Biotechnology, Jülich, Germany) are gratefully acknowledged.