The Existence of the K+ Channel in Plant Mitochondria*

In this study, evidence is given that a number of isolated coupled plant mitochondria (from durum wheat, bread wheat, spelt, rye, barley, potato, and spinach) can take up externally added K+ ions. This was observed by following mitochondrial swelling in isotonic KCl solutions and was confirmed by a novel method in which the membrane potential decrease due to externally added K+ is measured fluorimetrically by using safranine. A detailed investigation of K+ uptake by durum wheat mitochondria shows hyperbolic dependence on the ion concentration and specificity. K+ uptake electrogenicity and the non-competitive inhibition due to either ATP or NADH are also shown. In the whole, the experimental findings reported in this paper demonstrate the existence of the mitochondrial K+ ATPchannel in plants (PmitoKATP). Interestingly, Mg2+ and glyburide, which can inhibit mammalian K+ channel, have no effect on PmitoKATP. In the presence of the superoxide anion producing system (xanthine plus xanthine oxidase), PmitoKATP activation was found. Moreover, an inverse relationship was found between channel activity and mitochondrial superoxide anion formation, as measured via epinephrine photometric assay. These findings strongly suggest that mitochondrial K+ uptake could be involved in plant defense mechanism against oxidative stress due to reactive oxygen species generation.

One of most outstanding problems in mitochondria bioenergetics concerns the mitochondrial permeability to metabolites, organic compounds, including vitamins, their derived cofactors, and metal ions. The mitochondrial inner membrane contains metabolite carriers (for review, see Refs. 1 and 2), responsible for shuttling substrates between matrix and cytosol and for catabolism dependent on matrix enzymes, as well as vitamin and cofactor translocators (for review, see Refs. 3 and 4). Moreover, the inner membrane also contains the cation carriers and channels that regulate cell and mitochondrial physiology. In particular, as regards K ϩ ion, in mammalian mitochondria, the transport properties are such that net potassium flux across the mitochondrial membrane determines mitochondrial volume (Refs. 5 and 6 and references therein). It has been shown that K ϩ uptake is mediated by diffusion leak, driven by the high electric membrane potential maintained by redox-driven electrophoretic proton ejection, and that regulated K ϩ efflux is mediated by the inner membrane K ϩ /H ϩ antiporter (see Ref. 7). There is also evidence for the existence of an inner membrane protein designed to catalyze electrophoretic K ϩ uptake into mammalian (5)(6)(7)(8)(9)(10)(11)(12) and yeast (13,14) mitochondria. As far as plant mitochondria are concerned, even though mitochondrial structure and function are expected to be strictly dependent on K ϩ transport across the mitochondrial membrane, the knowledge of K ϩ permeability is not established at present. Indeed, the presence of a powerful K ϩ /H ϩ antiporter, which partially collapses ⌬pH, thereby increasing ⌬⌿, 1 has been shown (Refs. 15 and 16 and references therein). On the other hand, among a variety of compounds (including reducing sugars, proline, and Cl Ϫ ), K ϩ could play a significant role in the leaf osmotic adjustment in response to water stress (17). Thus, the purpose of this investigation was an attempt to determine whether and how K ϩ can enter per se plant mitochondria. They were isolated from a number of different plant species, and a detailed investigation was carried out using durum wheat mitochondria. The latter were chosen on the grounds of their high K ϩ permeability. The existence of a K ϩ channel, which is probably involved in plant defense against oxidative stress, is shown.

EXPERIMENTAL PROCEDURES
Chemicals and Plant Material-All reagents were purchased from Sigma. They were of purest available grade and they were used without further purification. Substrates were used as Tris salts at pH 7.20. Solution pH was adjusted with either Tris or HCl. Valinomycin, oligomycin, nigericin, and diazoxide were dissolved in ethanol.
Isolation of DWM-Durum wheat seeds, cv. Ofanto (250 g) were sowed on a distilled water-saturated polyurethane foam sheet, and they were covered with a Whatman filter paper. They were then grown in the dark at 25°C and 95% relative humidity for 72 h in an Heraeus HPS 1500 incubator.
About 200 g of etiolated shoots (1-2 cm long) were removed from seedlings, and mitochondria were isolated essentially as in Ref. 18 with minor modifications; the grinding and washing buffers were 0.5 M sucrose, 4 mM cysteine, 1 mM EDTA, 30 mM Tris-HCl, pH 7.50, 0.1% (w/v) defatted BSA, 0.6% (w/v) PVP; and 0.5 M sucrose, 1 mM EDTA, 10 mM Tris-HCl, pH 7.40, 0.1% (w/v) defatted BSA, respectively. Purification of washed mitochondria was performed by isopycnic centrifugation in a self-generating density gradient containing 0.5 M sucrose, 10 mM Tris-HCl, pH 7.20, and 28% (v/v) Percoll (colloidal PVP-coated silica, Amersham Pharmacia Biotech) in combination with a linear gradient of 0% (top) to 10% (bottom) (w/v) PVP-40 (19). The final mitochondrial suspension was diluted with an appropriate volume of sucrose-free washing medium, in order to obtain a 0.3 M sucrose concentration. Mitochondrial protein content was determined by the method of Lowry modified as in Ref. 20, using BSA as a standard.
The purified mitochondria showed 95% and 90% intactness of inner and outer membrane, respectively, determined as in Ref. 18. They were tightly coupled: oxidation of 2-oxoglutarate occurs with a respiratory control ratio equal to 6.7 Ϯ 1.4 (S.E., four experiments) and with an ADP/O ratio, i.e. the ratio between the phosphorylated ADP and reduced oxygen, equal to 3.6 Ϯ 0.15 (S.E., four experiments). Oxygen uptake was measured at 25°C by means of a Gilson Oxygraph model 5/6-servo channel, pH 5, equipped with a Clark-type electrode (5331; YSI, Yellow Spring, OH) in a medium consisting of 0.3 M mannitol, 5 mM MgCl 2 , 10 mM KCl, 0.1% (w/v) defatted BSA, 10 mM potassium phosphate buffer, pH 7.20.
Isolation of Mitochondria from Other Plant Species-Mitochondria from etiolated seedlings of bread wheat, spelt, rye, and barley were obtained as above reported for DWM isolation. Mitochondria from potato and spinach were obtained from 100 g of tubers and 200 g of leaves, respectively; homogenization was carried out by using a Waring Blendor homogenizer (speed setting 3 ϫ 3 s ϫ 3 times). Purification was performed as for DWM; in order to prevent chlorophyll contamination, washed spinach mitochondria were passed two consecutive times through the Percoll gradient.
In any experiment, coupled mitochondria were used showing osmotic response, as monitored in swelling experiments, as well as the capability to generate membrane potential following succinate addition, as monitored via safranine response.
Swelling Experiments-Swelling experiments were performed at Calibration of the safranine fluorescence decrease as a function of K ϩ diffusion potential was performed according to Ref. 22 by using rat liver mitochondria isolated as in Ref. 23; the K ϩ diffusion potential in rat liver mitochondria was induced by adding 0.05 g/ml valinomycin (24).
Superoxide Anion Assay-Production of superoxide anion was determined essentially as in Ref. 25 by monitoring photometrically (480 nm, ⑀ 480 ϭ 4.00 mM Ϫ1 cm Ϫ1 ) the rate of epinephrine oxidation to adrenochrome. The assay was carried out in the presence of 5 mM succinate in 2 ml of a medium consisting of 100 mM KCl, 110 mM mannitol, 5 mM MgCl 2 , 2 EU of catalase, 10 mM Tris-HCl, pH 7.20; 10 M cytochrome c was also present to account for the swelling-dependent endogenous cytochrome c release (26). Mitochondria (0.1 mg of protein) were incubated for 10 min during which any mitochondrial swelling was completed, then the reaction was started by adding 1 mM epinephrine. In certain experiments the osmoticum was 0.5 mM KCl plus 309 mM mannitol instead of 100 mM KCl plus 110 mM mannitol.
In order to establish whether superoxide anion formation can affect PmitoK ATP activity, superoxide anion was generated by using the properly developed superoxide anion-producing system consisting of 0.1 mM xanthine plus 0.038 EU of xanthine oxidase. This system was added to 2 ml of 0.18 M KCl, 2 mM Tris-HCl, pH 7.00. 30 s later mitochondria were added, and the swelling was continuously monitored. In ⌬⌿ experiments, the superoxide anion producing system was added to 2 ml of the safranine medium containing mitochondria; after 30 s of incubation, 5 mM succinate was added, then the ⌬⌿ decrease was induced by KCl (25 mM) addition. Swelling and ⌬⌿ experiments were carried out as described in the respective sections.

RESULTS
DWM Swelling-In order to gain a first insight into the K ϩ permeability of purified DWM, swelling experiments were carried out, using isolated mitochondria in the absence of externally added respiratory substrates. In control experiments, the absorbance of DWM, suspended in 0.36 M sucrose solution, was found to remain constant during the time, since sucrose cannot enter mitochondria (Fig. 1A, trace a); conversely, DWM were found to swell in 0.14 M NH 4 P i solution (Fig. 1A, trace b). The mitochondrial K ϩ permeability was checked by suspending DWM in potassium acetate and in KCl isotonic solutions. In the 14 M ammonium phosphate (NH 4 P i ) (A), added with 2 mM Tris-HCl, pH 7.00, in the absence of externally added respiratory substrates. Swellings were continuously monitored at 546 nm and 25°C. Where indicated, the medium contained 0.1 g of valinomycin (Val), 1 mM ATP, 2 g of oligomycin (Oligo), or 10 M atractyloside (Atr); at the arrows either 0.1 g of valinomycin (Val) or 10 nM nigericin (Nig) were added. SCHEME 1. DWM permeability in KCl solution. The processes catalyzed by the already reported PIMAC and K ϩ /H ϩ antiporter, by the novel PmitoK ATP and the K ϩ diffusive leak are shown. K ϩ -induced ⌬⌿ decrease of mitochondria can be explained either as a result of the H ϩ entry via the combined work of PmitoK ATP and K ϩ /H ϩ antiport or as a result of K ϩ uniport via PmitoK ATP . I.M.M., inner mitochondrial membrane. ATP and NADH inhibit PmitoK ATP , but not the K ϩ /H ϩ antiporter. For details, see the "Results." light of the existence of K ϩ /H ϩ antiport, previously reported in plant mitochondria (Refs. 15 and 16 and references therein), spontaneous swelling was found in potassium acetate with initial swelling rate equal to 0.08 ⌬A/min⅐mg protein (Fig. 1A, trace c). A further addition of nigericin, which allows K ϩ /H ϩ exchange, has been proven to increase swelling rate up to 0.16 ⌬A/min⅐mg protein.
Surprisingly, DWM were found to swell rapidly in KCl (initial rate equal to 0.12 ⌬A/min⅐mg protein) (Fig. 1A, trace d); since Cl Ϫ can enter plant mitochondria (27), such a swelling indicates that K ϩ can penetrate per se mitochondria in the absence of valinomycin. When valinomycin, which allows K ϩ permeability, was added at the end of the spontaneous swelling, little additional absorbance decrease was observed. No significant changes in both rate and extent of swelling was found in the presence of the respiratory substrate succinate or of the uncoupler FCCP or of the inhibitor pair cyanide plus salicylhydroxamic acid, which can inhibit cytochrome c oxidase and plant alternative oxidase, respectively (data not shown).
When DWM were suspended in KCl in the presence of valinomycin, a very high rate swelling was found to occur (initial rate was 1.9 ⌬A/min⅐mg protein) (Fig. 1A, trace e), indicating that K ϩ rather than Cl Ϫ uptake via PIMAC (27) represents the rate-limiting step of DWM swelling in KCl.
In the light of the already reported ATP inhibition of K ϩ channel in mammalian mitochondria (6,7,9,10,12), we investigated whether externally added ATP can prevent DWM swelling in KCl (Fig. 1B). First, in a control experiment it was observed that ATP (1 mM) did not affect DWM absorbance in sucrose medium (Fig. 1B, traces f and g); however, ATP inhibited both the initial rate (about 25%) and the extent of the KCl swelling (Fig. 1B, traces k and l). Inhibition was insensitive to oligomycin and atractyloside, powerful inhibitors of ATP synthase and of ADP/ATP translocator, respectively (Fig. 1B, trace  j), which have no effect on the KCl swelling (data not shown).
Valinomycin, added at the end of the swelling, induced an additional significant swelling (Fig. 1B, traces j and k). This finding indicates that K ϩ uptake rather than Cl Ϫ uptake was inhibited by ATP under these experimental conditions (1 mM ATP, absence of valinomycin). Swelling experiments in KCl plus valinomycin occurred at very fast rates, and these rates were inhibited by ATP (Fig. 1B, traces m and n), raising the possibility of ATP inhibition of Cl Ϫ transport. In the same experiment, swelling in TEA ϩ medium was also carried out either in the absence or in the presence of ATP (see Ref. 10); ATP was found to cause no significant change in TEACl swelling, which occurs with a low rate (Fig. 1B, traces h and i).
These qualitative findings are consistent with the existence of a mitochondrial K ATP channel; however, two findings render them insufficient to establish such a process with certainty. ATP inhibition of Cl Ϫ transport could explain the data; moreover, swelling in KCl plus succinate is unexpected if mitochondria maintain their membrane potential during respiration.
The Effect of Externally Added K ϩ on Mitochondrial ⌬⌿-The observed K ϩ uptake as well as the possible potassium cycle in Garlid's terms (6) are expected to cause membrane potential decrease, which could occur as a result of K ϩ and/or H ϩ transport by DWM (see Scheme 1). We monitored ⌬⌿ using a safranine probe, and observed that succinate addition (5 mM) to DWM in the absence of K ϩ caused a rapid increase of ⌬⌿ to about 185 mV ( Fig. 2A). Phosphate caused a slight increase of ⌬⌿, as expected if it enters via proton-compensated symport (21,24), whereas ADP caused a decrease of ⌬⌿, as expected from the increased proton current secondary to ATP synthesis. The ATP synthase inhibitor, oligomycin, partially restored ⌬⌿, and the uncoupler, FCCP, abolished ⌬⌿ ( Fig. 2A). Taken together, these findings indicate that DWM exhibit normal functionality with respect to ⌬⌿ in K ϩ -free medium.
Entirely different results were obtained when KCl was added to the medium (Fig. 2, B-D). When KCl was added after ener- gization and hyperpolarization, ⌬⌿ was collapsed (Fig. 2B) similarly to the effect of FCCP. This depolarization was prevented by both ATP and NADH. Sequential additions of 1 mM KCl caused progressive depolarization, whereas NaCl had no effect (Fig. 2C). Indeed, ⌬⌿ was generated at a rate and to an extent inversely relating to KCl concentration (Fig. 2D).
When valinomycin was added to DWM in the absence of added K ϩ , a small depolarization was observed (Fig. 2, C and  D). This is consistent with the contribution of the K ϩ /H ϩ antiporter to extramitochondrial K ϩ in such media, resulting in K ϩ cycling and some uncoupling when valinomycin is added.
The experiments in Fig. 2 appear to resolve uncertainties arising from the data in Fig. 1. Thus, high KCl concentrations are able to uncouple DWM completely, thus accounting for the observed swelling in the presence of succinate. Moreover, the failure of NaCl to mimic the uncoupling effects of KCl identifies K ϩ uniport as the uncoupling modality.
The y intercepts of the Dixon plots in Fig. 3 coincide with the experimental values measured in the absence of inhibitor. The x intercepts indicate K 1/2 values for ATP and NADH under these conditions of 290 and 390 M, respectively.
Mitochondrial K ϩ Permeability in Other Plant Species-In order to ascertain whether other plant mitochondria are permeable to externally added K ϩ , swelling and safranine experiments were carried out by using mitochondria from monocotyledonous (cereals) and dicotyledonous (potato and spinach) species, including etiolated and green tissues (Table I). Inter-estingly, all the investigated mitochondria were found to show a K ϩ -dependent membrane potential decrease and to swell in 0.18 M KCl in the absence of valinomycin. The K ϩ permeability was markedly evident in cereals; the highly permeable durum wheat species was used for further investigations. Potato mitochondria proved to be the least permeable to K ϩ among the investigated species. It should be noted that, in all cases, the swelling in KCl was enhanced by valinomycin, thus indicating that the K ϩ permeability rather than Cl Ϫ permeability was investigated (data not shown). We also studied swelling in 0.18 M potassium acetate in a medium containing 1% BSA in order to inhibit plant uncoupling mitochondrial protein activity (28,29). As expected, all plant species were found to possess the K ϩ /H ϩ antiporter.
Certain Features of K ϩ Uptake by DWM-In order to gain some insight into the mechanism by which K ϩ enters DWM, the dependence of the K ϩ uptake rate was investigated as a function of 0.005-25 mM KCl concentration (Fig. 4). Hyperbolic dependence on the concentration was found, suggesting that K ϩ uptake takes place in a protein-dependent manner, probably via PmitoK ATP . In three different experiments, K m and V max were 2.2 Ϯ 0.78 mM (S.E.) and 12.5 Ϯ 1.96 mV/s (S.E.), respectively. The rate of ⌬⌿ decrease due to FCCP addition was found to be greater than that observed with all KCl additions (Fig. 4, inset), confirming that this rate is not limited by the rate of safranine response.
To gain an insight into the specificity of PmitoK ATP , we compared the swelling response to various monovalent chloride salts (Fig. 5A). The rate and amplitude of mitochondrial swelling varied in the order Cs ϩ Ͼ K ϩ ϭ Rb ϩ Ͼ Na ϩ ϭ Li ϩ . We also examined their ability to collapse ⌬⌿ in respiring mitochondria (Fig. 5B). Na ϩ and Li ϩ had no significant effect, whereas K ϩ , Cs ϩ , and Rb ϩ caused ⌬⌿ decrease at a significant rate.
In order to ascertain whether and how the K ϩ channel is ⌬⌿-dependent, DWM, previously energized by succinate, were treated with FCCP in the nanomolar concentration range, designed to partially collapse membrane potential. Then the dependence of the rate of ⌬⌿ decrease on the actual mitochondrial ⌬⌿ was investigated, using 25 mM KCl (Fig. 6). A fast decrease of K ϩ -dependent depolarization, i.e. of the K ϩ channel   Fig. 2B (Control trace), and it was expressed as the percentage of the rate of ⌬⌿ decrease, caused by externally added FCCP (1 M). The rates of passive swelling in both KCl and potassium acetate (Kac), carried out as reported in Fig.  1A (traces c and d), are expressed as the percentage of the rate of mitochondrial swelling in ammonium phosphate. The swelling in Kac was performed in presence of 1% (w/v) BSA except for Triticum durum.
activity, was found in ⌬⌿ range varying from about 175 to 140 mV. The rate was found to remain rather constant in the range 140 -95 mV.
Effector Sensitivity of PmitoK ATP -We examined inhibitors and activators of mammalian mitoK ATP to determine whether PmitoK ATP shared any or all of the ligands previously reported for mitoK ATP (10,12,30,31). Results are reported in Table II. Like mammalian mitoK ATP , PmitoK ATP is inhibited by ATP ( Figs. 2A and 3A) and ADP, and ATP inhibition is prevented or reversed by GTP and diazoxide. PmitoK ATP is also stimulated by the sulfydryl group reagents mersalyl and N-ethylmaleimide. In contrast to mammalian mitoK ATP , PmitoK ATP is activated, rather than inhibited, by palmitoyl-CoA, and it is not inhibited by glyburide. The glyburide effect was examined in both the absence or presence of Mg 2ϩ (0.1 mM) plus ATP (1 mM) plus either diazoxide (10 M) or GTP (1 mM), which proved to be essential for inhibition due to glyburide of mitochondrial K ϩ channel in mammalian (31); however, PmitoK ATP was not inhibited by glyburide under any condition. A difference was also observed in the effect of Mg 2ϩ ion. Mg 2ϩ ion is required for inhibition of mammalian mitoK ATP by ATP and CoA esters; however, it has no effect on its own (6). In contrast, Mg 2ϩ had no effect on PmitoK ATP , and ATP inhibition did not require Mg 2ϩ .
PmitoK ATP also differs from the plant inward rectifying K ϩ of nonmitochondrial membranes because it is not inhibited by Al 3ϩ , Ba 2ϩ , and TEA ϩ . Inhibition by NADH ( Figs. 2A and 3B) and Zn 2ϩ and stimulation by coenzyme A are also distinctive of the plant mitochondrial channel.
PmitoK ATP Activity and Superoxide Anion Generation-Since electron flow via the respiratory chain can produce oxygen radical species, which are assumed to damage severely both animal and plant cells (32,33), and given that a number of energy dissipating processes have been proposed to decrease the mitochondrial reactive oxygen species generation (34 -41), investigation was made in order to establish whether K ϩ uptake and superoxide anion generation are somehow related. Superoxide anion formation was monitored as in Ref. 25 by measuring the absorbance increase due to the epinephrine to adrenochrome conversion. Measurements were done in me-dium containing low (0.5 mM) and high (100 mM) K ϩ , in which PmitoK ATP is less and more active, respectively (Fig. 7). DWM respiring in 0.5 mM KCl produced superoxide anion at a rate equal to 40 nmol/min⅐mg protein. This rate was decreased to 19 nmol/min⅐mg protein in 100 mM KCl. Control was made in  6. Dependence of the rate of K ؉ -induced ⌬⌿ decrease in DWM on imposed membrane potential. In a series of measurements, mitochondria were added to the safranine medium and energized with 5 mM succinate as in Fig. 2B (Control trace). Then, in order to partially decrease ⌬⌿, FCCP was added at nM concentrations (1-10 nM); when a stable imposed ⌬⌿ was reached, 25 mM KCl was added; the obtained rate of K ϩ -induced ⌬⌿ decrease was reported as a function of the imposed ⌬⌿. Bars represent the S.E. relative to three different experiments. order to check that the different KCl concentrations had no effect on epinephrine response. In five different experiments, the rate of superoxide anion generation was 42 Ϯ 8.8 (S.E.) nmol/min⅐mg protein and 22 Ϯ 5.6 (S.E.) nmol/min⅐mg protein in 0.5 and 100 mM KCl medium, respectively. SOD prevented the epinephrine response in both 0.5 and 100 mM KCl media. Moreover, ATP and mersalyl, an inhibitor and an activator, respectively, of PmitoK ATP , proved to enhance and to prevent, respectively, the superoxide anion generation by DWM in 100 mM KCl medium (data not shown).
We examined the possibility that superoxide anion might modulate PmitoK ATP . DWM were incubated with xanthine plus xanthine oxidase, which generate superoxide anion, then the swelling in KCl (Fig. 8A) and ⌬⌿ response to KCl (Fig. 8B) were monitored. Xanthine oxidase per se had no effect on both swelling and ⌬⌿ decrease rate; on the contrary, xanthine was found to cause 35% increase in the rates. When added together, xanthine and xanthine oxidase were found to cause approximately 100% increase in PmitoK ATP activity. Such an increase was found to be partially prevented by SOD. Under our experimental conditions, mitochondria treated with superoxide anion show the same polarization rate and ⌬⌿ of the control (data not shown), thus indicating that the increase of K ϩ uptake is not due to gross membrane damage. DISCUSSION It is central in the understanding of ion membrane permeability and transport to recognize that there is a vectorial component whereby ions are translocated through a protein embedded in a lipid bilayer membrane. Plant cells contain K ϩ channels in the plasma membrane, in the tonoplast (Refs. 42 and 43 and references therein), in the chloroplast envelope (44,45) and probably in the Golgi (45). To the best of our knowledge, K ϩ channels have not previously been reported in plant mitochondria. Our first observation was that DWM swell spontaneously in isotonic KCl (Fig. 1). The inner membranes of plant mitochondria transport Cl Ϫ ions via the pH-regulated PIMAC (27). Since mitochondria swell rapidly in KCl, it follows that TABLE II Effect of different compounds including cations, nucleotides, and -SH group reagents on the DWM K ϩ channel A series of compounds that have known effects on mammalian mitoK ATP channel and plant plasma membrane K ϩ channel were investigated with respect to their effect on DWM K ϩ channel activity, evaluated by ⌬⌿ and/or passive swelling experiments. ⌬⌿ experiments were carried out as reported in Fig. 2B. The investigated compounds were added to mitochondria; after 30 s of incubation, succinate (5 mM) was added to the sample; then, after 120 s, depolarization was induced by adding KCl (5 mM). Swelling experiments were carried out as in Fig. 1A (trace d) in a medium containing the listed compounds. Activation and inhibition are reported as the percentage of the control rate of ⌬⌿ decrease or swelling. Compounds that differently affect DWM K ϩ channel and other K ϩ channels are made evident.

Compound
Effect on DWM K ϩ channel Effect on other K ϩ channels Ref.
plant mitochondria also contain a K ϩ conductance pathway. KCl swelling was inhibited by external ATP, and ATP inhibition was insensitive to oligomycin and atractyloside. Moreover, 1 mM ATP inhibited swelling in KCl plus valinomycin only slightly and had no effect on the rate of swelling in TEACl (see Ref. 10). These findings strongly indicate that ATP specifically inhibits swelling by inhibiting K ϩ uniport; however, the question whether ATP inhibits K ϩ or Cl Ϫ influx is not entirely clear in these protocols. Thus, we observed that 5 mM ATP inhibited swelling in KCl plus valinomycin (data not shown).
A clearer picture emerged from the measurements in respiring mitochondria, in which we monitored ⌬⌿ as a measure of K ϩ cycling across the membrane. In particular, these experiments (Fig. 2) demonstrate that endogenous K ϩ influx uncouples mitochondria to a remarkable degree. DWM and mammalian mitochondria differ strongly in this respect. The K ϩ cycle in mammalian mitochondria cannot uncouple completely, because the maximal rate of the cycle, given by the V max of the K ϩ /H ϩ antiporter, is only about 20% of the maximal rate of proton ejection (46). The existence of a very active K ϩ /H ϩ antiporter (15,16) in plant mitochondria implies that a matching K ϩ channel activity should also exist in these membranes, as has been found. Evidently the activity of each pathway is approximately equal to the proton-ejecting capacity of the electron transport chain. This quantitative difference between mammalian and plant mitochondria is also reflected in the energy dependence of swelling in K ϩ salts; swelling of respiring rat liver mitochondria is due to electroneutral uptake of acetate and/or phosphate in combination with electrophoretic uptake of K ϩ , driven by ⌬⌿. Cl Ϫ does not penetrate significantly, because the endogenous K ϩ cycle flux pathways are insufficient to collapse ⌬⌿. In contrast, respiring plant mitochondria swell in both respiring and non-respiring states, because the active K ϩ cycle completely collapses ⌬⌿, permitting electrophoretic Cl Ϫ uptake.
We were able to show that uncoupling is specific for K ϩ (Cs ϩ , Rb ϩ ) ion and does not occur with Na ϩ or Li ϩ . Moreover, the K ϩ -specific uncoupling is inhibited by ATP. Taking the rate of ⌬⌿ change to be proportional to the rate of electrophoretic K ϩ influx, we observed hyperbolic dependence on K ϩ concentration. The apparent K m for K ϩ uptake was about 2 mM, which is lower than the value of 32 mM for the purified mitoK ATP from rat liver mitochondria (7). ATP and NADH inhibition is noncompetitive with K i values equal to 290 and 390 M, respectively; moreover, ATP shows a K i higher than the one observed in rat liver and beef heart mitochondria (7). ATP inhibition of PmitoK ATP is independent of the presence of Mg 2ϩ ions, whereas ATP inhibition of mammalian mitoK ATP exhibits an absolute requirement for Mg 2ϩ (6). Moreover, PmitoK ATP was activated, rather than inhibited, by palmitoyl-CoA, which is a potent inhibitor of mammalian mitoK ATP . In both of these respects, PmitoK ATP more closely resembles mammalian plasma membrane K ATP channels (47). The K ϩ channel opener, diazoxide, reversed inhibition by ATP; however, PmitoK ATP was insensitive to glyburide under all conditions tested. This property raises the possibility that PmitoK ATP , unlike mammalian K ATP channels, may not be regulated by a sulfonylurea receptor.
Since electrophoretic K ϩ uptake dissipates energy and uncouples oxidative phosphorylation, the mitochondrial K ϩ channel must be regulated in vivo. The observed inhibition by ATP and NADH meets this requirement, but it remains to be determined how the channel is opened under physiological conditions. The possible role of ATP and other ligands in regulating mammalian mitoK ATP has been reviewed (6,12). The physiological role of PmitoK ATP is also unknown. In vitro results in a KCl medium cannot be extrapolated to behavior in vivo; however, it seems clear from the agreement between rates of K ϩ /H ϩ antiport and K ϩ uniport via PmitoK ATP that plant mitochondria are capable of complete uncoupling. The demonstration that operation of PmitoK ATP reduces superoxide anion formation is not surprising, because most modes of uncoupling have this effect (34, 35, 40, 41, 48 -50). Of potentially greater interest is the finding that superoxide anion formation stimulated PmitoK ATP , suggesting a possible feedback mechanism to protect against reactive oxygen species.
FIG. 8. The superoxide anion effect on PmitoK ATP activity. The superoxide anion effect was evaluated on both the DWM swelling in KCl (A) and the rate of K ϩ -induced ⌬⌿ decrease (B). Swelling experiments were carried out as in Fig. 1A, trace d. ⌬⌿ measurements were carried out as in Fig. 2B (Control trace). The first part of the trace, regarding hyperpolarization of mitochondria due to 5 mM succinate addition, was omitted. For further details, see "Experimental Procedures." The superoxide anion producing system consisted in xanthine plus xanthine oxidase (XAN/XOX). In both A and B, the reaction media contained, where indicated: (i) 0.038 EU of xanthine oxidase (XOX), (ii) 0.1 mM xanthine (XAN), or (iii) xanthine plus xanthine oxidase either in the absence or presence of SOD (10 g/ml). Control was made in order to verify that xanthine plus xanthine oxidase had no effect on DWM absorbance in sucrose medium (data not shown).