The mitochondrial KATP channel as a receptor for potassium channel openers.

The biochemical properties of the mitochondrial K channel are very similar to those of plasma membrane K channels, including inhibition by low concentrations of ATP and glyburide (Paucek, P., Mironova, G., Mahdi, F., Beavis, A. D., Woldegiorgis, G., and Garlid, K. D.(1992) J. Biol. Chem. 267, 26062-26069). Plasma membrane K channels are highly sensitive to the family of drugs known as K channel openers, raising the question whether mitochondrial K channels are similarly sensitive to these agents. We addressed this question by measuring K flux in intact rat liver mitochondria and in liposomes containing K channels purified from rat liver and beef heart mitochondria. K channel openers completely reversed ATP inhibition of K flux in both systems. In liposomes, ATP-inhibited K flux was restored by diazoxide (K = 0.4 μM), cromakalim (K = 1 μM), and two developmental cromakalim analogues, EMD60480 and EMD57970 (K = 6 nM). Similar K values were observed in intact mitochondria. These potencies are well within the range observed with plasma membrane K channels. We also compared the potencies of these K channel openers on the plasma membrane K channel purified from beef heart myocytes. The K channel from cardiac mitochondria is 2000-fold more sensitive to diazoxide than the channel from cardiac sarcolemma, indicating that two distinct receptor subtypes coexist within the myocyte. We suggest that the mitochondrial K channel is an important intracellular receptor that should be taken into account in considering the pharmacology of K channel openers.

bers of this drug family exhibit a rich and clinically important pharmacology. Thus, cell membrane K ATP channels (cellK ATP ) in different tissues are considered to mediate the hypotensive and diabetogenic effects of diazoxide (4) and the cardioprotective effects of cromakalim and its derivatives (5). It is important to determine whether these drugs also act on mitochondrial K ATP channels (mitoK ATP ) in their therapeutic range.
In the first reports of KCO actions in mitochondria, Belyaeva et al. (6) and Szewczyk et al. (7) observed stimulation of K ϩ uptake by KCOs in respiring mitochondria. RP66471 was the most potent KCO studied (K1 ⁄2 ϭ 50 M), whereas P1060 and diazoxide were only weakly active at 700 M. Because these concentrations are much higher than K1 ⁄2 values observed with cellK ATP (1), these results appear to imply that mitochondrial actions of KCOs are not pharmacologically important.
We now report that diazoxide, cromakalim, and two experimental benzopyran derivatives are very potent activators of K ϩ flux through ATP-inhibited mitoK ATP , with K1 ⁄2 values similar to those observed with cellK ATP . KCO activation of K ϩ flux was observed in both intact mitochondria and proteoliposomes containing reconstituted mitoK ATP . No effect was observed on uninhibited K ϩ flux, which likely explains the low potencies observed by previous workers (6,7) in assays that did not include Mg 2ϩ and ATP. We also found that mitoK ATP and cellK ATP from beef heart differed strongly in their sensitivity to diazoxide, indicating distinct receptor subtypes among K ATP channels from the same cell. Our results indicate that mitoK ATP may be an important intracellular receptor for K ϩ channel openers, and they raise the possibility that mitoK ATP is the site of action of cardioprotective KCOs. 2

EXPERIMENTAL PROCEDURES
Assays of K ϩ Flux in Proteoliposomes Containing Reconstituted MitoK ATP Isolated from Rat Liver Mitochondria-MitoK ATP was purified and reconstituted into proteoliposomes exactly as described previously (8,9). Internal medium contained 300 M PBFI, 0.14 mM KCl, 1 mM TEA-EDTA, 25 mM TEA-HEPES, and 100 mM TEA-SO 4 (pH 6.8). Vesicles were added in a final concentration of 0.38 mg lipid/ml to external medium containing 150 mM KCl and 25 mM TEA-HEPES (pH 7.4) at 25°C. As indicated in the text, external medium also contained 3 mM MgCl 2 or 1 mM TEA-EDTA, and 0.5 mM ATP or no ATP. Electrophoretic K ϩ flux was initiated by 1 M carbonyl cyanide-p-trifluoromethoxyphenylhydrazone, which catalyzes charge compensation. K ϩ flux was determined by linear regression of the initial changes in the K ϩdependent fluorescence of intraliposomal PBFI, which was calibrated for each preparation (8,9).
Assays of K ϩ Flux in Proteoliposomes Containing Reconstituted CellK ATP Isolated from Beef Heart Sarcolemmal Vesicles-Sarcolemmal vesicles were prepared from the left ventricular muscle of fresh beef heart according to a modification (10) of the method of Jones and Besch (11). The sarcolemmal K ATP channel was solubilized, purified, and * This research was supported in part by National Institutes of Health Grants GM31086 and HL336573 (to K. D. G.) and a Postdoctoral Fellowship (to P. P.) from the Oregon Affiliate of the American Heart Association. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18  reconstituted into proteoliposomes exactly as described above for mitoK ATP , except that cellK ATP activity was found in the 100 mM KCl fraction (12). Internal and external media were as described for mitoK ATP , except that Na ϩ was substituted for TEA ϩ ion, because TEA ϩ inhibits K ϩ flux through cellK ATP . Assays of cardiac cellK ATP were carried out as described for mitoK ATP .
Assays of K ϩ Flux in Intact Rat Liver Mitochondria-6 l of a stock mitochondrial suspension in 0.25 M sucrose (50 mg protein/ml) were added to 3 ml of "K ϩ medium" containing K ϩ salts of chloride (45 mM), acetate (25.4 mM), TES (5 mM), EGTA (0.1 mM), pH 7.4, plus MgCl 2 (1 mM) and cytochrome c (10 M). Substrates for respiration were either succinate (3 mM) and rotenone (2 g/mg of protein) or ascorbate (2.5 mM) and TMPD (0.25 mM), as indicated in the figure legends. For assays in "TEA ϩ medium," TEA ϩ was substituted for K ϩ . The assay cuvette was maintained at 25°C, and absorbance data at 520 nm were collected and stored digitally as described previously (13)(14)(15). Respirationdriven, electrophoretic uptake of K ϩ is accompanied by electroneutral uptake of acetate and osmotically obligated water. The light-scattering variable, ␤, is linearly dependent on matrix volume; hence, K ϩ influx (J) is proportional to d␤/dt. Chemicals and Drugs-PBFI was purchased from Molecular Probes Inc. (Eugene, OR). Two benzopyranyl derivatives, EMD57970 and EMD60480, were provided by E. Merck (Darmstadt, Germany). All other chemicals and reagents were obtained from Sigma. Fig. 1 contains dose-response curves for KCO stimulation of K ϩ flux in vesicles reconstituted with mitoK ATP . Drug assays were carried out in medium containing 3 mM Mg 2ϩ and 0.5 mM ATP, which fully inhibits mitoK ATP (K1 ⁄2 (ATP) Ϸ 22 M). The drugs tested were potent activators of K ϩ flux, restoring ATP-inhibited flux to, but not beyond, control rates measured in the absence of ATP.

Activation of K ϩ Flux through Reconstituted MitoK ATP by K ϩ Channel Openers-
Observed K1 ⁄2 values (mean and S.D.) were 1.05 Ϯ 0.06 M for cromakalim (n ϭ 5), 0.37 Ϯ 0.03 M for diazoxide (n ϭ 4), 6.1 Ϯ 1.3 nM for EMD60480 (n ϭ 2), and 6.20 Ϯ 0.02 nM for EMD57970 (n ϭ 2). As shown in the inset to Fig. 1, cromakalim and diazoxide exhibited indistinguishable Hill slopes of 2.0 Ϯ 0.5, and the benzopyranyl derivatives yielded Hill slopes of 3.5 Ϯ 0.3. Hill slopes greater than 1.0 may reflect a tetrameric structure of the channel, as observed with other K ϩ channels (16), or the existence of multiple binding sites on a regulatory ATP binding cassette, as has been proposed for the sulfonylurea receptor of the pancreatic ␤ cell (17).
We also measured KCO activation of K ϩ flux in proteoliposomes reconstituted with mitoK ATP purified from beef heart mitochondria. Observed K1 ⁄2 values from two experiments were 1 M for cromakalim and 0.4 M for diazoxide. These results extend a previous observation that cardiac and hepatic mitoK ATP behave very similarly.
We stress that these drugs stimulated K ϩ flux only when K ϩ flux was inhibited by Mg 2ϩ and ATP. Control (uninhibited) K ϩ flux is observed in media containing Mg 2ϩ alone, ATP alone, and lacking both Mg 2ϩ and ATP (8). In each of these conditions, KCOs had no effect on control K ϩ flux at doses up to 30-fold higher than their respective K1 ⁄2 values.
Activation of K ϩ Flux through Reconstituted Cardiac Plasma Membrane K ATP by K ϩ Channel Openers-The high potency of diazoxide for cardiac mitoK ATP was somewhat surprising and suggested a pharmacological distinction from cardiac cellK ATP , which is relatively insensitive to diazoxide (18). Accordingly, we evaluated the effects of the same set of KCOs on cellK ATP reconstituted from cardiac sarcolemmal vesicles (12, 19). Fig. 2 contains dose-response curves for KCO stimulation of K ϩ flux in proteoliposomes reconstituted with cardiac cellK ATP . These assays were carried out in media containing 2 mM ATP (K1 ⁄2 ϭ 0.5 mM). Each of the drugs restored ATP-inhibited K ϩ flux to control rates measured in the absence of ATP. As was the case with mitoK ATP , these KCOs had no effect on the control rate in the absence of ATP. Observed K1 ⁄2 values and Hill slopes (in parentheses) for opening the plasma membrane K ATP channel were 3.7 nM (1.1) for EMD57970, 22 nM (1.2) for EMD60480, 17 M (1.1) for cromakalim, and 855 M (0.9) for diazoxide. Similar values were obtained in a separate preparation. The K1 ⁄2 values for the benzopyranyl derivatives are reasonably similar to those observed for mitoK ATP ; however, the K1 ⁄2 value for diazoxide is about 2000-fold higher. In contradistinction to the finding with mitoK ATP , the Hill slopes for activation of cellK ATP were indistinguishable from 1 for all of the KCOs tested.
Activation of ATP-Sensitive K ϩ Flux in Intact Mitochondria by K ϩ Channel Openers-The preceding results appear to conflict with previous work (6,7) showing very low potencies for KCO activation of K ϩ flux in mitochondria. Accordingly, it was important to determine whether the high potencies observed in Fig. 1 are also observed in situ. Fig. 3 contains representative light-scattering traces from rat liver mitochondria respiring on ascorbate-TMPD. Swelling in K ϩ salts (trace a) was sharply inhibited by addition of 100 M , and cromakalim (f). "⌬J max " is the maximum ATP-sensitive K ϩ flux, i.e. the difference between control fluxes in the absence and presence of saturating ATP (0.5 mM). "⌬J" is the difference between fluxes in the presence or absence of the drug, both measured in the presence of 0.5 mM ATP. In nine separate preparations, ⌬J max ranged between 400 and 600 M/s, similar to values previously reported using these protocols (8). Observed K1 ⁄2 values and Hill slopes are given in the text.
FIG. 2. Activation of K ؉ flux by K ؉ channel openers in liposomes reconstituted with cardiac sarcolemmal K ATP channels (cellK ATP ). Dose-response curves and Hill plots (inset) for activation of K ϩ flux through cellK ATP by four KCOs: EMD57970 (q), EMD60480 (E), cromakalim (f), and diazoxide (Ⅺ). "⌬J max " and "⌬J" are defined as in Fig. 1, but with reference to rates in 2 mM ATP. Observed K1 ⁄2 values and Hill slopes are given in the text. Duplicate experiments on an independent preparation yielded similar results. ATP (down arrow to trace b) to levels close to those observed in TEA ϩ salts (trace c). In agreement with previous results (10), higher [ATP] had no further effect in K ϩ medium, and ATP had no effect on TEA ϩ flux (not shown). When 20 M cromakalim was included in the assay medium containing 100 M ATP, ATP inhibition was prevented (up arrow to trace d, Fig. 3). In the absence of ATP, cromakalim had no effect on the control rate up to 100 M, the highest dose tested. When cromakalim was added during the inhibited state, ATP inhibition was reversed (not shown). Fig. 4 contains dose-response curves for activation of K ϩ flux by diazoxide, cromakalim, and EMD60480 in mitochondria. Activation was measured relative to rates in the presence of 100 M ATP and 1 mM Mg 2ϩ , conditions in which the K1 ⁄2 for ATP inhibition is 2-3 M (13). The estimated K1 ⁄2 values were 2.3 M for diazoxide, 6.3 M for cromakalim, and 5.4 nM for EMD60480.
As in proteoliposomes, these drugs stimulated K ϩ flux only when K ϩ flux was inhibited by Mg 2ϩ and ATP. In doses 20-fold higher than their respective K1 ⁄2 values, these KCOs had no effect on flux through the uninhibited channel. This effect was verified in media containing Mg 2ϩ alone, ATP alone, and lacking both Mg 2ϩ and ATP. DISCUSSION This is the first report showing that KCOs activate mitoK ATP over the same dose range as they activate cellK ATP . This finding was observed in mitochondria and in proteoliposomes reconstituted with mitoK ATP and raises the possibility that mitoK ATP may be activated by KCOs in vivo. Kinetic parameters differed between intact mitochondria and the reconstituted preparations. As previously reported (13), the K1 ⁄2 for ATP inhibition is lower in mitochondria (2-3 M) than in proteoliposomes (20 -25 M). We now show that the K1 ⁄2 values for diazoxide and cromakalim are about 6-fold higher in mitochondria than in liposomes. On the other hand, the K1 ⁄2 for EMD60480 is about the same in the two preparations. These differences may reflect regulatory complexity in intact mitochondria, which is lost upon extraction and reconstitution.
In the dose ranges studied, KCOs had no effect on K ϩ flux when Mg 2ϩ and/or ATP were omitted from the assay medium. The lack of effect of KCOs on the open channel is also charac-teristic of cellK ATP (20). The finding that KCOs in low doses have no effect on the uninhibited channel is also consistent with the results of Belyaeva et al. (6) and Szewczyk et al. (7), who did not include Mg 2ϩ and ATP in the assay medium used for their studies.
Physiological Consequences of Opening and Closing MitoK ATP -Opening of mitoK ATP will shift the balance between K ϩ uniport and K ϩ /H ϩ antiport, causing transient net K ϩ uptake and matrix swelling to a higher steady-state volume (21). Halestrap (22) has established that increasing matrix volume over a fairly narrow range greatly activates electron transport at the point where electrons feed into ubiquinone, and he has suggested (23) that this sequence may be triggered by opening of mitoK ATP . Thus, opening of mitoK ATP may be a necessary component of the cellular signals calling, for example, for higher ATP production to support increased work in heart or for faster ␤ oxidation of fatty acids to support thermogenesis in brown adipose tissue. Conversely, blocking mi-toK ATP may interfere with the cell's response to these signals.
MitoK ATP as a Pharmacological Receptor-Recognition of mitoK ATP as an intracellular receptor for KCOs adds a new dimension to the KCO pharmacology, which has heretofore focused exclusively on plasma membrane K ATP channels. Pharmacological regulation of K ATP channels has many important, tissue-dependent consequences (1-3); however, the receptors for these effects have not yet been identified, and a mitochondrial contribution cannot be excluded. The role of K ATP channels in pancreatic ␤ cells is a case in point. Flatt et al. (24) have recently shown that Ca 2ϩ -dependent insulin release from electopermeabilized ␤ cells is stimulated by glyburide and inhibited by diazoxide. Because plasma membrane K ATP channels are inoperative in the permeabilized cell, these effects point to an intracellular receptor for these agents (24).
A particularly exciting development in heart is the finding by Grover and colleagues (5,25) and others (26,27) that KCOs are cardioprotective during experimental ischemia. KCO-treated hearts maintained higher ATP levels and exhibited reduced infarct size and enhanced post-ischemic recovery upon reperfusion. All of these effects were blocked by glyburide, which is contraindicated in patients susceptible to cardiac ischemia. Preconditioning, in which a period of brief ischemia reperfusion Relative ATP-sensitive K ϩ uptake into respiring rat liver mitochondria; ⌬J/⌬J max , is plotted versus drug concentration. "⌬J max " and "⌬J" are defined as in Fig. 1. The dose-response curves reflect activation by EMD60480 (f), diazoxide (q), and cromakalim (E). Assay medium for ⌬J contained 1 mM Mg 2ϩ and 0.1 mM ATP, which maximally inhibited K ϩ uniport in mitochondria (10), and ascorbate-TMPD as respiratory substrates. The K1 ⁄2 values reported in the text are means of two independent experiments. For each drug, duplicate K1 ⁄2 values were within 5% of each other.
protects the heart against subsequent ischemic damage (28), was also blocked by glyburide (29). These pharmacological effects point to a role of K ATP channels in myocardial protection; but, again, the receptor for these effects has not been identified, and a mitochondrial site of action cannot be excluded (25).
Exploration of this possibility is aided by the existence of receptor subtypes among K ATP channels (1). For example, cromakalim is a potent activator of cellK ATP from heart and vascular smooth muscle (29) but has a minimal effect on insulin secretion (4,30). Diazoxide is a potent vasodilator (4) and also reduces insulin secretion (31) but has little effect on cardiac cellK ATP (18). This raises the question whether mitoK ATP and cellK ATP from the same cell differ pharmacologically. Accordingly, we have compared drug sensitivities of cardiac mitoK ATP and cellK ATP reconstituted from beef heart. These experiments yielded the following preliminary results: (i) mitoK ATP from heart and liver do not differ significantly in their drug sensitivities (K1 ⁄2 values); (ii) cardiac mitoK ATP and cardiac cellK ATP exhibit similar sensitivities to benzopyran derivatives; however, (iii) cardiac mitoK ATP is about 2000 times more sensitive to diazoxide than cardiac cellK ATP . The low sensitivity of reconstituted cardiac cellK ATP to diazoxide is entirely consistent with previous reports (18). The inferred existence of receptor subtypes within the cardiac myocyte may provide a means to determine the site of cardioprotective action of K ϩ channel openers.