The Mechanism by Which the Mitochondrial ATP-sensitive K+ Channel Opening and H2O2 Inhibit the Mitochondrial Permeability Transition*

Myocardial infarction is a manifestation of necrotic cell death as a result of opening of the mitochondrial permeability transition (MPT). Receptor-mediated cardioprotection is triggered by an intracellular signaling pathway that includes phosphatidylinositol 3-kinase, endothelial nitric-oxide synthase, guanylyl cyclase, protein kinase G (PKG), and the mitochondrial KATP channel (mitoKATP). In this study, we explored the pathway that links mitoKATP with the MPT. We confirmed previous findings that diazoxide and activators of PKG or protein kinase C (PKC) inhibited MPT opening. We extended these results and showed that other K+ channel openers as well as the K+ ionophore valinomycin also inhibited MPT opening and that this inhibition required reactive oxygen species. By using isoform-specific peptides, we found that the effects of KATP channel openers, PKG, or valinomycin were mediated by a PKCϵ. Activation of PKCϵ by phorbol 12-myristate 13-acetate or H2O2 resulted in mitoKATP-independent inhibition of MPT opening, whereas activation of PKCϵ by PKG or the specific PKCϵ agonist ψϵ receptor for activated C kinase caused mitoKATP-dependent inhibition of MPT opening. Exogenous H2O2 inhibited MPT, because of its activation of PKCϵ, with an IC50 of 0.4 (±0.1) μm. On the basis of these results, we propose that two different PKCϵ pools regulate this signaling pathway, one in association with mitoKATP and the other in association with MPT.

Myocardial infarction is a manifestation of necrotic cell death as a result of opening of the mitochondrial permeability transition (MPT). Receptor-mediated cardioprotection is triggered by an intracellular signaling pathway that includes phosphatidylinositol 3-kinase, endothelial nitric-oxide synthase, guanylyl cyclase, protein kinase G (PKG), and the mitochondrial K ATP channel (mitoK ATP ). In this study, we explored the pathway that links mitoK ATP with the MPT. We confirmed previous findings that diazoxide and activators of PKG or protein kinase C (PKC) inhibited MPT opening. We extended these results and showed that other K ؉ channel openers as well as the K ؉ ionophore valinomycin also inhibited MPT opening and that this inhibition required reactive oxygen species. By using isoform-specific peptides, we found that the effects of K ATP channel openers, PKG, or valinomycin were mediated by a PKC⑀. Activation of PKC⑀ by phorbol 12-myristate 13-acetate or H 2 O 2 resulted in mitoK ATPindependent inhibition of MPT opening, whereas activation of PKC⑀ by PKG or the specific PKC⑀ agonist ⑀ receptor for activated C kinase caused mitoK ATP -dependent inhibition of MPT opening. Exogenous H 2 O 2 inhibited MPT, because of its activation of PKC⑀, with an IC 50 of 0.4 (؎0.1) M. On the basis of these results, we propose that two different PKC⑀ pools regulate this signaling pathway, one in association with mitoK ATP and the other in association with MPT.
A major component of ischemia-reperfusion injury is necrotic cell death, manifested at the organ level as infarction. Necrotic cell death is widely thought to be the consequence of opening the mitochondrial permeability transition (MPT), 2 as proposed originally by Crompton et al. (1) and recently by Di Lisa et al. (2). Strong protection against ischemia-reperfusion injury is afforded by a brief preliminary ischemic period, a phe-nomenon known as ischemic preconditioning, which was also shown to reduce MPT opening (3,4). However, it is unclear how MPT opening is prevented when cardioprotection is triggered by ischemic preconditioning or pharmacological mitoK ATP openers such as diazoxide. In this regard, Korge et al. (5) have made important progress. They simulated ischemia in isolated mitochondria with anoxia and then exposed them to elevated Ca 2ϩ and phosphate (P i ). These conditions caused MPT opening and cytochrome c loss, which were blocked by diazoxide or phorbol 12-myristate 13-acetate (PMA). Both effects were negated by 5-hydroxydecanoate (5-HD), indicating that the protective effect of PKC in blocking MPT pore opening operates through mitoK ATP opening. Kim et al. (6) showed that MPT was inhibited by cyclic GMP in the presence of a cytosolic extract, implicating a guanylyl cyclase-dependent signaling pathway. Baines et al. (7) showed that mice with cardiac-specific expression of recombinant PKC⑀ showed inhibition of MPT opening in heart mitochondria and identified a multiprotein complex in mitochondria containing PKC⑀, adenine nucleotide translocator, the voltage-dependent anion channel, and hexokinase. Adenine nucleotide translocator, voltage-dependent anion channel, hexokinase, and cyclophilin D are hypothesized to be constituents of MPT (8).
These findings take on greater significance with our recent demonstration that protein kinase G (PKG) opens mitoK ATP and that this effect is mediated via an endogenous mitochondrial PKC⑀ (9). We showed that addition of PKG ϩ cGMP to isolated mitochondria from rat heart, brain, or liver resulted in increased K ϩ influx comparable with that induced by diazoxide or cromakalin and that these effects were blocked by PKC⑀specific inhibitors.
We now show that exogenous PKG ϩ cGMP inhibits MPT and that this protective effect is blocked by inhibitors of PKG, PKC⑀, and mitoK ATP . Thus, mitoK ATP is an obligatory intermediate in the PKG-dependent inhibition of MPT. We show that valinomycin mimics the effects of PKG, PKC⑀, or mitoK ATP activators, establishing that this protective effect of mitoK ATP opening is due specifically to K ϩ influx into the mitochondrion. Moreover, inhibition of MPT by mitoK ATP opening or valinomycin is blocked by an ROS scavenger and by PKC⑀ inhibitors, indicating that a mitoK ATP -dependent increase in mitochondrial ROS production is the signal transmitted to the PKC⑀-MPT interaction. Activation of endogenous PKC⑀ by PMA or H 2 O 2 also inhibits MPT opening, and this effect does not involve mitoK ATP opening. The latter results suggest that two different PKC⑀-containing subcomplexes modify these mito-chondrial functions. These and other recent findings from our group lead to a preliminary model of the mitochondrial segment of the cardioprotective signaling pathway that inhibits MPT opening and cell necrosis.

MATERIALS AND METHODS
Mitochondrial Isolation-Heart, liver, and brain mitochondria were isolated by differential centrifugation from male Sprague-Dawley rats (220 -240 g) exactly as described previously (9). Mitoplasts were prepared by digitonin permeabilization of the outer membrane (10).
Measurement of Light Scattering-Matrix swelling is the standard technique for assaying the MPT. As described previously (11,12), light scattering changes of mitochondrial suspensions (0.1 mg/ml) were followed at 520 nm and 30°C and are reported as ␤, which is inverse absorbance normalized for protein concentration, Ps, as shown in Equation 1, Rates of light scattering change, reflecting rates of matrix swelling, were obtained by taking the linear term of a second-order polynomial fit of the light scattering trace, calculated over the initial 2 min following MPT induction by the uncoupler (from 60 -180 s after mitochondrial addition). Data in this paper are summarized in bar graphs as "Ca 2ϩ -induced swelling rate (%)," which is calculated by taking Ca 2ϩ -induced swelling rates in the absence and the presence of 1 M cyclosporin A as 100 and 0%, respectively. Statistical significance of the difference of the means was assessed using unpaired Student's t test. A value of p Ͻ 0.05 was considered significant and is indicated in each figure by an asterisk. (All asterisks indicate comparison of that particular mean with Fig. 1B, column Ca 2ϩ .) Except where noted, assay medium contained K ϩ salts of Cl Ϫ (120 mM), HEPES, pH 7.2 (10 mM), succinate (10 mM), and phosphate (5 mM). Where indicated, tetraethylammonium (TEA ϩ ) or Li ϩ salts replaced K ϩ salts. All traces were also supplemented with MgCl 2 (0.5 mM), 5 M rotenone, 0.67 M oligomycin, and 50 M Na ϩ /K ϩ -ATP. In K ϩ -free media, Tris salts of ATP were used. Osmolality of all media ranged between 275 and 280 mosM.
Assay of MPT-In vitro studies of MPT have traditionally been performed in media containing sucrose, to support irreversible swelling, and lacking Mg 2ϩ and ATP, which inhibit MPT (13). However, in order to study interactions among mitoK ATP , PKC, and MPT, it was necessary to employ salt media containing Mg 2ϩ and ATP. By using these media, we observed that high [Ca 2ϩ ] caused a robust MPT opening after a variable time lag of 1-3 min duration. Bernardi and co-workers (14) found that the time lag is primarily a population effect, reflecting the fraction of mitochondria that have been recruited at a given time, and they showed that it is possible to synchronize MPT opening by sequential additions of Ca 2ϩ , ruthenium red (RR), and CCCP. This protocol was followed in all of the experiments reported here. Mitochondria (0.1 mg of mitochondrial protein/ml) were added to medium at 30°C (t ϭ 0). CaCl 2 was added at 20 s; RR (0.5 M, to block further Ca 2ϩ uptake) was added at 40 s; and CCCP (250 nM, to initiate MPT) was added at 60 s. Free Ca 2ϩ concentrations were calculated using the Freeware computer program BAD4 (15).
Immunodetection of PKC⑀-Samples (80 g) from crude mitochondria, Percoll-purified mitochondria, and mitoplasts made from Percoll-purified mitochondria were precipitated using methanol/chloroform (Ref. 16 with minor modifications) and resolved on 10% SDS-polyacrylamide gels. The gels were electrophoretically transferred onto polyvinylidene difluoride membranes. After blocking with 3% gelatin in Tris-buffered saline, immunoblots were exposed to anti-PKC⑀ antibody (BD Transduction Laboratories) in a 1:500 dilution followed by alkaline phosphatase-conjugated secondary antibody (immunoblot assay kit goat anti-mouse IgG alkaline phosphatase; Bio-Rad). A colorimetric assay was used to visualize antigen-antibody reactions following the manufacturer's instructions.
Chemicals-Protein kinase G isoform I␣, cGMP, KT5823, Gö6983, and Ro318220 were purchased from Calbiochem. PKC isoform-specific peptides were synthesized by EZ Biolabs (Westfield, IN), according to the published amino acid sequences (17). All other chemicals were from Sigma. The PKG1␣ concentration and activity used in this study was comparable with that used in our previous paper, and the concentration present in cells (see Ref. 9 and references therein). In the present experiments the enzyme had a specific activity of 10 units/g (1 unit is the amount of enzyme required to transfer 1 pmol of phosphate from ATP to the synthetic substrate GRT-GRRNSI per min at 30°C). We used 25 ng/ml, corresponding to 1.5 ϫ 10 Ϫ10 mol/liter. PKG in smooth muscle cells is ϳ1 ϫ 10 Ϫ7 mol/liter and slightly lower in cardiomyocytes (9).

RESULTS
Regulation of MPT by mitoK ATP -A typical set of experiments carried out in K ϩ medium (see "Materials and Methods") is shown in Fig. 1A. That the Ca 2ϩ -dependent swelling (Fig. 1A, trace marked none) is because of MPT opening is confirmed by the fact that it was inhibited by cyclosporin A. The K ATP channel opener diazoxide (Dzx) also inhibited MPT opening, and 5-HD (Fig. 1A, Dzx ϩ 5HD) blocked this inhibition. In this and the experiments that follow, we added mitoK ATP openers and blockers either before or immediately after the mitochondria and before adding Ca 2ϩ . Fig. 1B summarizes the results of experiments carried out as in Fig. 1A. It can be seen that both selective (diazoxide) and nonselective (cromakalim and nicorandil) mitoK ATP channel openers inhibited MPT opening and that this inhibition was blocked by the mitoK ATP blocker 5-HD (18) as well as by the nonselective mitoK ATP blocker tetraphenylphosphonium (TPP ϩ ) (19). Importantly, valinomycin also inhibited MPT opening at a concentration that increases K ϩ flux to the same extent as a K ATP channel opener (20). As expected, this effect of valinomycin was not blocked by 5-HD. We conclude from these results that mitoK ATP opening inhibits MPT in isolated heart mitochondria and that the inhibition is a consequence of mitoK ATP -dependent K ϩ influx. mitoK ATP Opening Inhibits MPT via ROS Activation of PKC⑀-The preceding results raised the following question How does increased K ϩ uptake into the matrix cause inhibition of MPT? We have shown that net K ϩ influx leads to matrix alkalinization and thereby to increased ROS production (20 -22). It is known that increased ROS activates protein kinases (23); and it has been reported that PKC⑀ inhibits MPT opening in heart mitochondria (5,7). Accordingly, we hypothesized that a PKC may be an intermediate in this process. We investigated the effects of ROS scavengers and PKC inhibitors on mitoK ATP -mediated MPT inhibition, with the results shown in Fig. 2. These data show that diazoxide-mediated inhibition of MPT opening was blocked by the free radical scavenger MPG, by the PKC inhibitors chererythrine and Ro318220, and by the PKC⑀specific inhibitor peptide ⑀V 1-2 (17). Moreover, MPT inhibition mediated by valinomycin was also inhibited by MPG or ⑀V 1-2 . MPT inhibition by diazoxide was not blocked by the PKC␦ inhibitor Gö6983, by the PKC␦-specific inhibitor peptide ␦V 1-1 , or by the scrambled peptide analogue of ⑀V 1-2 (not shown). On the basis of these findings, we conclude that mitoK ATP opening inhibits MPT via PKC⑀ which is activated by valinomycin-or mitoK ATP -dependent ROS production.
PKG-mediated MPT Inhibition Occurs via mitoK ATP , ROS, and PKC⑀-We showed recently that addition of PKG ϩ cGMP to isolated rat heart mitochondria causes PKC⑀-dependent opening of mitoK ATP (9). This finding, together with the results contained in Figs. 1B and 2, suggests that PKG ϩ cGMP should inhibit MPT in a mitoK ATP -and ROS-dependent manner. This is indeed the case, as shown in Fig. 3. Inhibition of MPT by PKG ϩ cGMP was blocked by the PKG-specific inhibitor KT5823, by the mitoK ATP blockers 5-HD, glibenclamide, and TPP ϩ , by the ROS scavenger MPG, and by the PKC⑀ inhibitors chelerythrine and peptide ⑀V 1-2 . PKG-induced MPT inhibition was not blocked by the PKC␦ inhibitor peptide ␦V 1-1 . Moreover, neither heat-inactivated PKG nor PKG in the absence of cGMP had any effect on MPT (data not shown). From these experiments, we conclude that PKG inhibits MPT via mitoK ATP opening and that this effect is mediated by PKC⑀ and ROS.
Evidence for Two PKC⑀s, One Upstream and the Other Downstream of mitoK ATP -The finding in Fig. 2 that the peptide ⑀V 1-2 blocks mitoK ATP inhibition of MPT leads us to conclude that PKC⑀ is downstream of mitoK ATP ; however, the finding that PKC⑀ is required for mitoK ATP opening by PKG ϩ cGMP   (9) implies that PKC⑀ is upstream of mitoK ATP . Fig. 4 contains data that address the question of how PKC⑀ can be both upstream and downstream of mitoK ATP . In both Fig. 4A (K ϩ medium) and Fig. 4B (TEA ϩ medium), we see that MPT was inhibited by the PKC⑀ activators PMA and H 2 O 2 . MPT was not inhibited by the inactive ␣PMA (data not shown). Inhibition by PMA and H 2 O 2 was prevented by the PKC⑀-specific peptide inhibitor ⑀V 1-2 but not by glibenclamide (Fig. 4, A and B) or 5-HD (data not shown). Thus, PKC⑀ activation by PMA or H 2 O 2 can inhibit MPT directly, without intervention of mitoK ATP . This conclusion is further supported by the fact that PMA and H 2 O 2 have identical effects in K ϩ medium (Fig. 4A) and TEA ϩ medium (Fig. 4B). Note in particular that the peptide ⑀V 1-2 is still able to block PMA or H 2 O 2 inhibition of MPT in K ϩ -free medium, showing that a PKC⑀ exists that can modulate MPT without the intervention of mitoK ATP .

mitoK ATP Channel Opening and H 2 O 2 Inhibit MPT
The data in Fig. 4A are in apparent disagreement with the findings of Korge et al. (5), who found that 5-HD blocked PMA protection just as it did diazoxide protection. A possible explanation for this result is that mitoK ATP was not sufficiently blocked before addition of PMA, and therefore MPT was blocked via the mitoK ATP -dependent pathway. We can obtain the same result if we add PMA simultaneously with 5-HD. In our protocols, PMA was added 4 s after 5-HD. Under these conditions, glibenclamide (or 5-HD; data not shown) consistently fails to inhibit the protective effect of PMA or H 2 O 2 , as shown in Fig. 4A. Again, this ability of PMA or H 2 O 2 to inhibit MPT by a mitoK ATP -independent pathway is confirmed by the findings in TEA ϩ medium (Fig. 4B).
In contrast to the results with PMA or H 2 O 2 , PKC⑀ activation by PKG ϩ cGMP (9) or by the pseudo-RACK peptide ⑀RACK cannot inhibit MPT directly and requires mitoK ATP opening as an intermediate. This is evidenced by the fact that the effects of these agents were abolished by glibenclamide (Fig. 4A) or 5-HD (not shown). (The K1 ⁄ 2 for ⌿⑀RACK in opening mitoK ATP was about 0.2 M, and 2 mM ⌿⑀RACK was also ineffective in blocking MPT in the presence of glibenclamide.) These findings suggest that two different PKC⑀ pools are involved in this part of the mitochondrial signaling pathway, one in association with mitoK ATP and the other in association with MPT. This conclusion is confirmed by experiments performed in K ϩ -free medium (Fig. 4B), in which diazoxide, or PKG ϩ cGMP, or ⑀RACK was ineffective in inhibiting MPT.

H 2 O 2 -induced Inhibition of H 2 O 2 -induced MPT
Opening-MPT onset in vivo is thought to be an interplay between Ca 2ϩ , ROS, and the anti-oxidant system (glutathione peroxidase in  Glibenclamide (glib; 10 M) was added immediately 1 s after mitochondria, whereas PMA, H 2 O 2 , PKG ϩ cGMP, or ⑀RACK were added 5 s after mitochondria. Ca 2ϩ -induced swelling rate is expressed as % of MPT-dependent rate as described in Fig. 1A. Note that the effect of diazoxide is observed solely in K ϩ -based medium. The data are shown as average Ϯ S.D. of at least five independent experiments.

mitoK ATP Channel Opening and H 2 O 2 Inhibit MPT
heart mitochondria) (24,25). In the presence of phosphate, MPT can be elicited with high concentrations of Ca 2ϩ and the low amounts of H 2 O 2 that are normally produced by the respiratory chain. In the 1st two bars of Fig. 5 is mediated by activation of the MPT-associated PKC⑀, as shown by reversal of the inhibition by the specific inhibitor peptide ⑀V 1-2 . Furthermore, this protection is independent of mitoK ATP activity, as evidenced by the inability of 5-HD to inhibit it (Fig. 5). In further experiments not shown, we have estimated the concentration dependence of H 2 O 2 inhibition of MPT and found that the IC 50 is 400 Ϯ 100 nM (n ϭ 4).
mitoK ATP -dependent Inhibition of MPT via ROS and PKC⑀ in Isolated Liver and Brain Mitochondria-The pathway that transmits the signal from extra-mitochondrial PKG to mitoK ATP is also present in rat brain and rat liver mitochondria (9). The data in Fig. 6 show that the pathways in rat liver mitochondria are exactly the same as those described for rat heart mitochondria in Figs. 1-4. Similarly, in experiments with n ϭ 2-4 for each condition, we observed identical behavior in isolated rat brain mitochondria (not shown). Therefore, this signaling pathway appears to be a general phenomenon in mammalian mitochondria.

Immunodetection of PKC⑀ in Rat Heart Mitochondria-
Crude and Percoll-purified heart mitochondria and mitoplasts were resolved by SDS-PAGE and transferred to polyvinylidene difluoride membranes for immunodetection with anti-PKC⑀ antibodies (1:500). Percoll purification removes contaminants, including microsomes and peroxisomes (26). As shown in Fig.  7, similar amounts of PKC were detected in Percoll-purified mitochondrial and mitoplasts, whereas a higher amount was found in non-Percoll-purified mitochondria, perhaps reflecting contamination with sarcolemmal membranes and cytoplasmic proteins (26). Therefore, we conclude that heart mitochondria contain endogenous PKC⑀. The finding that PKC⑀ is retained in mitoplasts indicates that this PKC⑀ is either associated with the outer surface of the inner membrane or in the matrix or both.

DISCUSSION
In the biology of ischemia, mitoK ATP opening has come to be seen as protective and MPT opening as harmful. The relationships between these two phenomena have been explored, but previously there has been no clear picture of the overall process. Korge et al. (5) have shown that activation of mitoK ATP results in inhibition of MPT. Korge et al. (5) and also Baines et al. (7) have shown that PKC activation is able to inhibit MPT. Kim et al. (6) have shown that activation of a cytosolic PKG resulted    Fig. 1A. The low "conditioning" dose of H 2 O 2 (2 M) was added immediately before the mitochondria. Other compounds tested were 5-HD (0.3 mM) and the PKC⑀-specific inhibitor peptide ⑀V 1-2 (0.5 M), added as indicated. Note that the low conditioning dose of H 2 O 2 inhibited MPT-dependent swelling in both high Ca 2ϩ and low Ca 2ϩ ϩ H 2 O 2 conditions. Ca 2ϩinduced swelling rate is expressed as % of MPT-dependent rate as described in Fig. 1A. The data are shown as average Ϯ S.D. of at least five independent experiments. JULY 28, 2006 • VOLUME 281 • NUMBER 30 in inhibition of MPT. In view of our recent results showing that PKG opens mitoK ATP via PKC⑀ (9) and that mitoK ATP activation stimulates ROS production (22), we can now propose a complete picture of a signaling pathway present in mitochondria that extends the abovementioned findings.

mitoK ATP Channel Opening and H 2 O 2 Inhibit MPT
The data in this paper show that mitoK ATP opening inhibits MPT opening in heart, liver, and brain mitochondria and that it does so via K ϩ flux, increased ROS, and PKC⑀ (Figs. 1-6). We confirm and extend the findings of Korge et al. (5) and Baines et al. (7) that PKC⑀ is constitutively expressed in mitochondria (Fig. 7). The rather complex sequence of reactions is contained in Fig. 8, which also displays the agonists and antagonists on which the sequence is based. Fig. 9 contains a descriptive diagram of the interrelationships among mitoK ATP , PKC⑀, and MPT. Our data indicate that two different PKC⑀ pools are involved. PKC⑀1 is upstream of mitoK ATP and is required for opening of mitoK ATP by PKG ϩ cGMP (see Ref. 9 and the diagrams in Figs. 8 and 9). PKC⑀2 is downstream of mitoK ATP and is required for mitoK ATP -mediated inhibition of MPT (Figs. 2 and 3 and the diagrams in Figs. 8 and 9). The remainder of the "Discussion" will explore the basis of these conclusions and the proposed mechanisms.
Dual Roles of Mitochondrial PKC⑀-Our data show that PMA, H 2 O 2 , and ⌿⑀RACK cause inhibition of MPT opening and that this inhibition is prevented by ⑀V 1-2 but not by ␦V 1-2 (Fig. 4A). Exactly the same sensitivities have been observed for mitoK ATP opening (9). 3 Therefore, we conclude that PKC⑀ has the dual role of regulating the activities of both mitoK ATP and MPT. Because the outer membrane constitutes a barrier to the enzyme, we conclude that these PKC⑀s are endogenous to mitochondria. These results clarify those obtained by Baines et al. (7), who observed inhibition of MPT onset when heart mitochondria were incubated with exogenous PKC⑀. These authors also included PMA in the assay medium, which, as shown in Fig. 4A, can open mitoK ATP and inhibit MPT per se.
PKC is subject to complex regulation. In the presence of an anionic phospholipid, such as cardiolipin or phosphatidyl-Lserine, diacylglycerol or PMA is thought to activate the enzyme by promoting flexure at the hinge region causing removal of the pseudosubstrate from the catalytic site (27). ROS stimulate PKC⑀ by causing thiol oxidation to disulfide and the loss of Zn 2ϩ (28). Korichneva et al. (23) presented evidence that Zn 2ϩ is bound to two zinc fingers in PKC and that PMA/diacylglycerol binding to one or oxidation of the other leads to Zn 2ϩ release and activation.
PKC activation may also be associated with translocation to a site where it is anchored to a specific RACK (29). Mochly-Rosen and co-workers (30 -32) have identified isozyme-specific activator and inhibitor peptides and hypothesized specific binding sites and consequences to explain their function. We have used ⑀V 1-2 , a specific inhibitor of PKC⑀ translocation activation, and ⑀RACK, a specific activator of PKC⑀ translocation and kinase function (30). These peptides were designed to interact with specific regions of PKC⑀ and alter their function. Thus, the ⑀RACK peptide acts as an allosteric agonist by preventing intramolecular autoinhibitory interaction within PKC⑀, and the ⑀V 1-2 peptide blocks substrate access to the catalytic site (31,32). has been shown to be a signaling molecule (33)(34)(35)(36)(37)(38), but its relationship with MPT has previously been that of an inducer, rather than an inhibitor, of MPT opening (24,25,39). The important physiological (signaling) role of H 2 O 2 has always been difficult to distinguish from its pathological role. Here we show H 2 O 2 acting as a signaling molecule, protecting against MPT pore opening in isolated mitochondria, at a concentration FIGURE 9. The cardioprotective signaling pathway in mitochondria. In step 1, activated PKG, which cannot cross the outer mitochondrial membrane, phosphorylates a protein at the external surface of the outer membrane. This leads by an unknown mechanism to activation of PKC⑀1, which is presumably bound to the outer surface of the inner mitochondrial membrane. In step 2, PKC⑀1 phosphorylates mitoK ATP , causing it to open and catalyze net K ϩ influx (step 3) (9). This leads to matrix alkalinization (20), which causes a modest increase in matrix H 2 O 2 (steps 4 and 5) (22). Increased H 2 O 2 activates PKC⑀2 (step 6), causing inhibition of MPT opening (step 7) (this paper). We have found that increased H 2 O 2 also activates PKC⑀1, causing mitoK ATP opening (step 8) and constituting a positive feedback loop. Because this work is not yet published, step 8 is labeled ?. Finally, it is generally accepted that H 2 O 2 acts as a second messenger in cell signaling by activating a variety of kinases (step 9) (33-38).

mitoK ATP Channel Opening and H 2 O 2 Inhibit MPT
at least 50ϫ smaller than that shown to induce MPT pore opening (Fig. 5). The demonstration that prior exposure to a low concentration of H 2 O 2 protects against MPT opening caused by higher levels of H 2 O 2 may be of pathophysiological importance for cardioprotection. Indeed, cardiomyocytes were shown to be protected from simulated ischemia via a H 2 O 2 induction of a pathway that contains PKC⑀ and mitoK ATP (37).
Evidence for Two Distinct Mitochondrial PKC⑀s-PKC⑀ participates in cardioprotection (40 -42), and it is known that signaling by this PKC⑀ depends strongly on its location (43)(44)(45)(46). In the model of Fig. 8, PKC⑀1 is associated with and regulates mitoK ATP , whereas PKC⑀2 is associated with and regulates MPT. The hypothesis that two separate PKCs are involved is suggested by the following studies. 1) Activation of PKC⑀ by either PMA or H 2 O 2 opens mitoK ATP and inhibits MPT. ⑀V 1-2 blocks both of these effects, whereas 5-HD and glibenclamide block only mitoK ATP opening and have no effect on MPT inhibition by PMA or H 2 O 2 (Fig. 4A). This shows that PKC⑀2 can inhibit MPT directly, without the intervention of mitoK ATP (see Fig. 8).
2) The demonstration that PMA or H 2 O 2 inhibits MPT in the complete absence of mitoK ATP activity (TEA ϩ medium, see Fig. 4B) further demonstrates that PKC⑀2 acts independently of mitoK ATP . 3) Conversely, PKG inhibition of MPT requires mitoK ATP opening (Fig. 4, A and B). Because it is known that PKG opens mitoK ATP by activating a mitochondrial PKC⑀ (9), it follows that this PKC (PKC⑀1) cannot open MPT directly, i.e. if there were only one PKC⑀ in this pathway, PKG should be able to inhibit MPT without intervention of mitoK ATP . These findings are consistent with the conclusion of Ping (46) that PKC⑀ does not function in isolation during cardioprotection, but rather forms close alliances with a variety of other proteins.
Differential Behavior of ⌿⑀RACK-Based on the proposed mechanism by which ⌿⑀RACK activates PKC⑀ (29,30), both PKCs are expected to respond to this peptide. Indeed, both PKCs respond in the predicted manner to the peptide inhibitor ⑀V 1-2 . Why then is ⌿⑀RACK able to activate PKC⑀1 but not PKC⑀2? A plausible explanation is based on the chemical structure of the two peptides as follows: the ionic state of the ⌿⑀RACK is negative (Ϫ1), whereas ⑀V 1-2 is electroneutral. Therefore, it is likely that ⑀V 1-2 can cross the inner membrane and interact with a PKC⑀ located on its inner face, whereas it is virtually impossible for negatively charged ⌿⑀RACK to diffuse across the inner membrane in the face of a large negative membrane potential. Thus, the findings are consistent with the possibility that PKC⑀1 is located on the exterior face of the inner membrane, whereas PKC⑀2 is located on the interior face of the inner membrane. The validity of this explanation must await further experiments. mitoK ATP Effect on Ca 2ϩ Uptake-The cardioprotective effect of mitoK ATP opening has been shown to be upstream of the onset of MPT (47). The inhibitory effect of mitoK ATP activity on MPT has usually been attributed to decreased Ca 2ϩ uptake derived from the slight uncoupling induced by K ϩ flux into mitochondria (48 -50). However, this conclusion has been controversial, and the results of such studies have been variable, perhaps due to the different experimental conditions used. Under the conditions in which we study isolated mitochondria, we showed that pharmacological concentrations of diazoxide induced opening of mitoK ATP without affecting Ca 2ϩ uptake or membrane potential (20,51). Therefore, the results shown herein cannot be attributed to decreased Ca 2ϩ uptake into the mitochondrial matrix.
Summary-We show that mitoK ATP opening, either by intracellular signaling (9) or by K ATP channel openers, causes inhibition of MPT opening. This may be an in vitro manifestation of a cardioprotective effect, because MPT opening is considered to be the primary cause of necrotic cell death after ischemiareperfusion (1,2). Identification of PKC⑀ within these pathways was inferred from the actions of a variety of agonists and antagonists, including peptides that are specific for PKC⑀. An interesting outcome of these studies is the suggestion of two separate pools of mitoPKC⑀, emphasizing the additional point that the same PKC isoform may exert effects at multiple locations in the cytosolic and mitochondrial compartments.