Cardiomyocyte death during acute coronary syndromes

Transcription

1 Translocation of Connexin 43 to the Inner Mitochondrial Membrane of Cardiomyocytes Through the Heat Shock Protein 90 Dependent TOM Pathway and Its Importance for Cardioprotection Antonio Rodriguez-Sinovas,* Kerstin Boengler,* Alberto Cabestrero, Petra Gres, Miriam Morente, Marisol Ruiz-Meana, Ina Konietzka, Elisabet Miró, Andreas Totzeck, Gerd Heusch, Rainer Schulz, David Garcia-Dorado Abstract We have previously shown that connexin 43 (Cx43) is present in mitochondria, that its genetic depletion abolishes the protection of ischemia- and diazoxide-induced preconditioning, and that it is involved in reactive oxygen species (ROS) formation in response to diazoxide. Here we investigated the intramitochondrial localization of Cx43, the mechanism of Cx43 translocation to mitochondria and the effect of inhibiting translocation on the protection of preconditioning. Confocal microscopy of mitochondria devoid of the outer membrane and Western blotting on fractionated mitochondria showed that Cx43 is located at the inner mitochondrial membrane, and coimmunoprecipitation of Cx43 with Tom20 (Translocase of the outer membrane 20) and with heat shock protein 90 (Hsp90) indicated that it interacts with the regular mitochondrial protein import machinery. In isolated rat hearts, geldanamycin, a blocker of Hsp90-dependent translocation of proteins to the inner mitochondrial membrane through the TOM pathway, rapidly (15 minutes) reduced mitochondrial Cx43 content by approximately one-third in the absence or presence of diazoxide. Geldanamycin alone had no effect on infarct size, but it ablated the protection against infarction afforded by diazoxide. Geldanamycin abolished the 2-fold increase in mitochondrial Cx43 induced by 2 preconditioning cycles of ischemia/reperfusion, but this effect was not associated with reduced protection. These results demonstrate that Cx43 is transported to the inner mitochondrial membrane through translocation via the TOM complex and that a normal mitochondrial Cx43 content is important for the diazoxide-related pathway of preconditioning. (Circ Res. 2006;99: ) Key Words: mitochondria heat shock protein geldanamycin connexin 43 TOM (Translocase of the Outer Membrane) complex Cardiomyocyte death during acute coronary syndromes determines survival and quality of life of patients with coronary artery disease. 1 In the majority of these patients, cardiomyocyte death is the consequence of transient, prolonged ischemia, and there is strong evidence that a substantial part of cell death occurs at the time of reperfusion. 2,3 Preconditioning, a state of increased resistance against cell death induced by ischemia reperfusion, is elicited by brief ischemia/reperfusion episodes or by certain pharmacological stimuli and has received particular attention. 4 A wealth of information has been collected on the molecular mechanisms involved in preconditioning, but many aspects of the signaling pathways and of the end effectors of the protection remain unknown. 4,5 An intriguing and unresolved aspect is the involvement of connexin 43 (Cx43), the protein forming gap junctions connecting adjacent ventricular cardiomyocytes, 6,7 in the genesis of preconditioning. 8,9 The protection of preconditioning is abolished in Cx43-deficient mice 10 but also in isolated cardiomyocytes from Cx43- deficient hearts, 11 indicating that it cannot be explained by effects of preconditioning on gap junction mediated cell-to-cell communication 12 and cell-to-cell propagation of cell injury during ischemia and reperfusion. 8,13,14 Mitochondria play a critical role both as signal transduction elements and as end effectors of the protection afforded by preconditioning, and we have previously shown 15 that Cx43 is present in cardiomyocyte mitochondria and that the mitochondrial Cx43 content is enhanced by ischemic preconditioning (IP). Moreover, we have recently provided a mechanistic link between mitochondrial Cx43 and the protection of preconditioning by showing that in situ hearts and cardiomyocytes isolated from mice with reduced mitochondrial Original received November 9, 2005; resubmission received April 25, 2006; revised resubmission received May 19, 2006; accepted May 22, From the Servicio de Cardiologia (A.R.-S., A.C., M.M., M.R.-M., E.M., D.G.-D.), Hospital Vall d Hebron, Barcelona, Spain; and Institut für Pathophysiologie (K.B., P.G., I.K., A.T., G.H., R.S.), Universitätsklinikum Essen, Germany. *Both authors contributed equally to this study. Correspondence to David Garcia-Dorado, Servicio de Cardiología, Hospital Universitario Vall d Hebron, Pg. Vall d Hebron 119, Barcelona, Spain. dgdorado@ir.vhebron.net 2006 American Heart Association, Inc. Circulation Research is available at DOI: /01.RES de 93

2 94 Circulation Research July 7, 2006 Cx43 content cannot be preconditioned by exposure to diazoxide, 16 a pharmacological agent inducing preconditioning through a mechanism involving the generation of reactive oxygen species (ROS). 17 Indeed, Cx43-deficient cardiomyocytes show a markedly attenuated production of ROS in response to diazoxide but not to menadione or valinomycin. 16 Although these observations strongly support the participation of mitochondrial Cx43 in the genesis of the protective state induced by preconditioning, a direct cause effect relationship between mitochondrial Cx43 and the induction of the protective state is missing. Moreover, although Cx43 is detected in the mitochondria, 15 its exact localization there has not been definitely established. In the present study, we therefore aimed to demonstrate the submitochondrial localization of Cx43, analyze the mechanism of translocation of Cx43 to the mitochondria, and investigate the effect of interfering with translocation on the ability of cardiomyocytes to be preconditioned. Materials and Methods The present study conforms to the NIH Guide for the Care and Use of Laboratory Animals (NIH publication No , revised 1996). A complete description of the methods used is in the online data supplement, available at Isolated Rat Heart Hearts of adult male Sprague Dawley rats (250 to 350 g) were excised and retrogradely perfused with an oxygenated Krebs solution at 37 C. 18,19 Infarct size was measured by 2,3,5-triphenyltetrazolium chloride (TTC) staining. 13,19 The extent of contraction band necrosis was quantified in histological sections. 18 Isolation of Mitochondria Mitochondria were isolated from whole rat ventricles by differential centrifugation and FACS (fluorescence-activated cell sorting [FACS]) as previously described. 15 For positive control, intercalated disk-enriched membrane fractions from whole rat ventricles were also obtained. 20 Mitochondria were also isolated from left ventricular biopsies obtained from Göttinger minipigs using differential centrifugation and percoll gradient ultracentrifugation. 15 Functionality of isolated mitochondria was confirmed by measuring a normal mitochondrial membrane potential and oxygen consumption (see the online data supplement). Analysis of Cx43 Localization Within Mitochondria Mitochondrial Subfractionation Freshly isolated pig mitochondria were subfractionated into their outer membrane, their inner membrane/matrix components, and in the fraction containing the inner membrane associated proteins using 0.6% digitonin and differential centrifugation. Mitochondria were further separated into their membrane constituents and matrix components by adding sunflower oil and 0.6% digitonin to frozen and thawed mitochondria, followed by centrifugation. Western Blot Analysis Western blot analysis on mitochondrial proteins was performed according to standard procedures. 15 Immunohistochemistry and Confocal Laser Scan Microscopy in Pig Cardiac Mitochondria Purified pig mitochondria (intact or treated with 0.6% digitonin) were incubated with anti-cx43 and anti adenine nucleotide transporter (anti-ant) or anti-voltage-dependent anion channel (anti- VDAC) and anti-ant antibodies, for 1 hour at 25 C, and with the respective secondary antibodies (n 4). For negative control, the primary antibodies were omitted. Mechanism of Cx43 Translocation to Mitochondria Immunoprecipitation of Cx43 With Tom20 and Heat Shock Protein 90 Samples (250 g of pig mitochondrial protein or 800 g of rat total protein) were incubated with anti-tom20, anti-cx43 antibody (rabbit or goat-polyclonal), anti-heat shock protein 90 (Hsp90) / antibodies, or with anti-rabbit IgGs before binding to Protein A/G Plus Agarose. After centrifugation, the beads were washed three times with 0.5 ml of 1 cell lysis buffer or 50 mmol/l Tris/HCl ph 7.4. Figure 1. Intact and digitonin-treated pig mitochondria were stained with antibodies against VDAC (red) and ANT (green) and analyzed by confocal laser scan microscopy. Colocalization pixels are shown in yellow.

3 Rodriguez-Sinovas et al Mitochondrial Connexin 43 and Cardioprotection 95 Subsequently, proteins were subjected to Western blot analysis. Each immunoprecipitation was performed in triplicate. Effect of Geldanamycin on Translocation of Cx43 to Mitochondria The effect of geldanamycin, a blocker that inhibits Hsp90-mediated mitochondrial import, 21 on mitochondrial Cx43 content, expressed as the Cx43/Ox-Phos Complex II (succinate-ubiquinol oxidoreductase) ratio, was analyzed by Western blot. Seventeen hearts were perfused for 15 minutes with normoxic Krebs buffer containing geldanamycin (5.4 mol/l) and compared with 17 control hearts receiving only vehicle (DMSO, 0.15%). Effect of Geldanamycin on Pharmacological and Ischemic Preconditioning s Protection The effect of pretreatment with geldanamycin (5.4 mol/l) or its vehicle for 15 minutes on mitochondrial Cx43 content was measured in 16 isolated rat hearts submitted to preconditioning with diazoxide (50 mol/l for 10 minutes), in 18 hearts submitted to IP (2 cycles of 5 minutes of global ischemia followed by 5 minutes of reperfusion, IP), in 8 hearts submitted to isoprenaline preconditioning (0.1 mol/l for 5 minutes), and in 38 normoxic hearts. For comparison, the effects of preconditioning protocols on Cx43 content in intercalated disk-enriched membrane fractions were analyzed (n 4 to5). The effect of geldanamycin on the protection afforded by diazoxide (n 19), IP (n 23), or isoprenaline (n 13) was analyzed in 66 isolated rat hearts submitted to 60 minutes of global ischemia followed by reperfusion. Statistical Analysis Parametric or nonparametric measures analysis of variance (ANOVA) test for single or repeated measures, followed by post hoc tests, were used as adequate. Data are expressed as mean SEM. A probability value less than 0.05 was considered to be significant. Results Cx43 Localizes at the Inner Membrane of Pig Mitochondria Intact mitochondria were stained with antibodies against the outer mitochondrial membrane protein VDAC and the inner mitochondrial membrane protein ANT. The mitochondria were positive for both proteins as shown by confocal laser scan microscopy. Digitonin-treated mitochondria displayed immunoreactivity mainly for ANT and only trace amounts of VDAC were observed (Figure 1). Intact mitochondria and mitochondria devoid of the outer membrane were positive for Cx43 and ANT (Figure 2A). The quantification of ANT and Cx43 immunoreactivity showed no significant difference in ANT and Cx43 signal intensities between intact and digitonin-treated mitochondria (Figure 2A) (see also colocalization in Figure I in the online data supplement). We further compared the mitochondrial Cx43 to the cytosolic Cx43 content (both expressed as a percentage of total Cx43 content). Western blot analysis indicated that % of total Cx43 was found in the cytosol and % of total Cx43 was detected in isolated mitochondria (n 4, P NS; Figure 2B). Within mitochondria, Cx43 was localized mainly at the inner membrane. Using marker proteins of the different mitochondrial compartments (VDAC for the outer membrane [ %], ANT for the inner membrane [ %], cytochrome c for the inner membrane associated proteins [ %], cyclophilin D for the matrix [ %], Figure 2. A, Intact and digitonin-treated pig mitochondria stained with antibodies against ANT (green) and Cx43 (red) and analyzed by confocal laser scan microscopy (left). Quantification of ANT and Cx43 immunoreactivities showed no significant difference between groups (right). B, Quantification of mitochondrial and cytosolic Cx43 in pig left ventricular myocardium. Cx43 content in 20 g of mitochondrial and cytosolic as well as in a dilution series of total (7.5 to 0.5 g) pig left ventricular myocardial protein extracts was determined by Western blot analysis (top). Bar graphs showing mitochondrial and cytosolic Cx43 level as percentage of total protein content (bottom). Mitochondrial vs cytosolic Cx43: P NS. n 4; Figure 3), Cx43 was mainly found in the protein preparation containing inner and outer membrane (95.4 1%), the inner mitochondrial membrane, and the matrix ( %), whereas only trace amounts of Cx43 were detected in the matrix fraction (4.6 1%) or the outer membrane preparation ( %) (n 4). In accordance with the confocal laser scan microscopy, some VDAC was resistant to digitonin treatment: % of VDAC was detected in the inner membrane and matrix fraction. Cx43 Translocates to the Inner Mitochondrial Membrane Through the Hsp90-Dependent TOM Protein Import System In pig cardiac mitochondria, Cx43 coprecipitated with Tom20 (Translocase of the outer membrane 20) and ANT but not with VDAC or cytochrome c. Using the anti-tom20 antibody for precipitating mitochondrial proteins revealed coprecipitation of Cx43 and ANT but not of VDAC or cytochrome c. As both Cx43 and Tom20 antibodies are derived from rabbits, rabbit IgGs were used as control immunoprecipitation, thereby allowing to differentiate between IgG signals and

4 96 Circulation Research July 7, 2006 Figure 3. A, Western blot analysis for Cx43 and marker proteins of different mitochondrial compartments (VDAC, outer membrane; ANT, inner membrane; cytochrome c [Cyt C], inner membrane associated proteins; cyclophilin D [CyP D], matrix) was performed on pig cardiomyocyte mitochondria fractionated in the outer membrane, the inner membrane, matrix, and inner membrane associated proteins (left). Western blot analysis for Cx43, ANT, VDAC, or CyP D was performed on mitochondria fractionated into the matrix and the membrane components (right). B, Bar graphs representing the percentage of protein in mitochondrial compartments. The total amount of pixels from each subfractionation protocol was set as 100%. *P 0.05 for both Cx43 and VDAC immunoreactivity in the outer membrane preparation vs the inner membrane and matrix. coprecipitating proteins. The use of anti-rabbit IgGs for immunoprecipitation did not indicate a precipitation of one of the analyzed proteins (Figure 4A). To further strengthen the above results, the coprecipitation of Tom20 with Cx43 was confirmed using goat polyclonal Cx43 antibodies (Figure 4B). In rat myocardial protein extract, using the anti-hsp90 antibody for precipitating myocardial proteins revealed coprecipitation of Cx43. When Cx43 was used for immunoprecipitation, a coprecipitation of Hsp90 was detected. The use of anti-rabbit IgGs for immunoprecipitation did not indicate a precipitation of 1 of the analyzed proteins (Figure 4C). Western blot analysis detected Cx43 immunoreactivity in mitochondrial fractions purified by FACS from normoxic rat hearts (Figures 5 and 6). Most of the Cx43 detected in the mitochondrial fraction was phosphorylated. Treatment with geldanamycin for 15 minutes significantly reduced the mitochondrial Cx43 content (expressed as fraction of the Ox-Phos Complex II) by 25% to 40% (P 0.05), without changes in the Cx43 phosphorylation status (Figures 5 and 6). Geldanamycin did not modify Ox-Phos Complex II immunoreactivity. Cx43 Translocation to Mitochondria by Preconditioning: Effects of Geldanamycin Preconditioning with diazoxide did not cause any significant change in mitochondrial Cx43/Ox-Phos Complex II ratio compared with normoxic hearts, whereas pretreatment with geldanamycin before diazoxide reduced mitochondrial Cx43/ Ox-Phos Complex II ratio by 30% (Figure 5), comparable to that observed in normoxic hearts treated with the Hsp90 blocker. Neither with nor without pretreatment with geldanamycin, preconditioning with diazoxide did not modify the amount of mitochondrial Ox-Phos Complex II. Cx43 in the intercalated disk-enriched membrane fraction was not modified by any treatment (Figure 5C). In isolated rat hearts, IP increased mitochondrial Cx43/Ox- Phos Complex II ratio compared with normoxic hearts to 2-fold (P 0.05). Pretreatment with geldanamycin abolished this increase (Figure 6B) but did not decrease Cx43 below normoxic values. IP did not modify the amount of mitochondrial Ox-Phos Complex II. Cx43 expression in the intercalated disk-enriched membrane fraction was not modified by any treatment (Figure 6C). Preconditioning with isoprenaline did not modify mitochondrial Cx43 (ratio Cx43/Ox-Phos Complex II: in normoxic hearts versus in hearts preconditioned with isoprenaline; n 4 in each group), and pretreatment with geldanamycin before isoprenaline was associated with a trend toward a reduction in mitochondrial Cx43/Ox-Phos Complex II ratio ( , n 4, P ). Effects of Preconditioning and Geldanamycin on Cell Death Induced by Ischemia/Reperfusion Ischemia induced rapid contractile arrest followed by rigor contracture, detected as an abrupt increase in end-diastolic

5 Rodriguez-Sinovas et al Mitochondrial Connexin 43 and Cardioprotection 97 Figure 4. A, Pig mitochondrial proteins were immunoprecipitated for Cx43 (rabbit polyclonal antibodies) or Tom20. Western blot analysis for Cx43, Tom20, VDAC, ANT, or cytochrome c was performed on the precipitated proteins and on the supernatant of the precipitation. Precipitation of mitochondrial proteins with anti-rabbit IgGs was used as negative control. B, Pig mitochondrial proteins were immunoprecipitated for Cx43 (goat polyclonal antibodies). Western blot analysis for Cx43 and Tom20 (both rabbit polyclonal antibodies) was performed on the precipitated proteins and on the supernatant of the precipitation. As positive control, total pig myocardial proteins were analyzed. C, Rat total proteins were precipitated with Cx43 and Hsp90. Subsequent Western blot analysis was performed on the precipitated proteins and on the supernatants of the precipitations for Cx43 and Hsp90. A rat total protein extract was analyzed as positive control. left ventricular pressure. In the absence of treatments, there was only minor functional recovery during reperfusion. In contrast, hearts submitted to preconditioning with diazoxide or IP had a trend toward an improved functional recovery compared with control hearts (see supplemental Figure II). Marked lactate dehydrogenase (LDH) release occurred at the time of reperfusion. Preconditioning with diazoxide markedly attenuated LDH release during reperfusion, but this protective effect was abolished by pretreatment with geldanamycin (Figure 7B and 7C). IP was associated with a significant attenuation in LDH release during reperfusion independently of the pretreatment with geldanamycin (Figure 7B and 7C). A higher geldanamycin concentration (10.8 mol/l) also failed to abolish protection induced by IP (accumulated Figure 5. A, Protocols followed to obtain cardiac mitochondria from rat hearts preconditioned with diazoxide with or without pretreatment with geldanamycin. B, Representative Western blots (10% polyacrylamide gels) showing 1-Na /K - ATPase, Cx43, and Ox-Phos Complex II immunoreactivity in mitochondrial fractions obtained from normoxic control hearts, from normoxic hearts treated with geldanamycin (5.4 mol/l), and from hearts submitted to pharmacological preconditioning with diazoxide (50 mol/l) without and with pretreatment with geldanamycin (5.4 mol/l). C, Representative Western blots showing 1-Na /K - ATPase and Cx43 immunoreactivity in intercalated disk-rich membrane fractions. *P 0.05, significant differences vs normoxic hearts. GA indicates geldanamycin; Dzx, diazoxide.

6 98 Circulation Research July 7, 2006 Figure 6. A, Protocols followed to obtain cardiac mitochondria from preconditioned rat hearts treated or not with geldanamycin. B, Representative Western blots (10% polyacrylamide gels) showing 1-Na /K -ATPase, Cx43, and Ox-Phos Complex II immunoreactivity in mitochondrial fractions obtained from normoxic control hearts, from normoxic hearts treated with geldanamycin (5.4 mol/l), and from hearts submitted to IP without and with pretreatment with 5.4 mol/l geldanamycin. C, Representative Western blots showing 1-Na /K - ATPase and Cx43 immunoreactivity in intercalated disk-rich membrane fractions. *P 0.05, significant differences vs normoxic hearts. GA indicates geldanamycin. LDH release was U/g dry tissue per 10 minutes, similar to the other 2 IP groups; n 3). Geldanamycin did also not modify either the protection induced by preconditioning with isoprenaline (accumulated LDH release was U/g dry tissue per 10 minutes in hearts pretreated with geldanamycin and in hearts receiving vehicle, as compared with in controls, P 0.05). Myocardial necrosis (TTC) was 57 6% of ventricular mass in control rat hearts (n 4) (Figure 7D). Infarct size was significantly reduced by preconditioning with diazoxide, IP, or isoprenaline (29 8%). Pretreatment with geldanamycin (5.4 mol/l) abolished protection afforded by diazoxide but not by IP (Figure 7D) or isoprenaline (33 12%). Geldanamycin at 10.8 mol/l also failed to abolish the protection induced by IP (34 7%, n 3; P NS). These effects were associated with reduced extent of contraction band necrosis in histological sections, as compared with control hearts, in all groups but in hearts Figure 7. Protocols followed to assess the effects of geldanamycin on protection by ischemic and pharmacological preconditioning (A), time course of LDH release during reperfusion (B), cumulative LDH release in the first 10 minutes of reperfusion (C), and infarct size (TTC) (D) in control rat hearts submitted to 60 minutes of ischemia/reperfusion, in control hearts pretreated with geldanamycin, and in hearts submitted to IP or preconditioning with diazoxide without or with 5.4 mol/l geldanamycin. *P 0.05, **P 0.01, and ***P 0.01 indicate significant differences between control and treated hearts. Abbreviations are as in Figures 6 and 7.

7 Rodriguez-Sinovas et al Mitochondrial Connexin 43 and Cardioprotection 99 receiving geldanamycin before preconditioning with diazoxide (see supplemental Figure III). Discussion In the present study, we identified Cx43 at the inner mitochondrial membrane, and provided evidence that Cx43 translocation involves the Hsp90-dependent TOM complex pathway, as demonstrated by the coimmunoprecipitation of Cx43 with Tom20 and Hsp90, and by the inhibition of translocation by geldanamycin. Reducing mitochondrial Cx43 content with geldanamycin abolished the protection afforded by diazoxide-induced preconditioning but had no effect on the protection against cell death afforded by IP or preconditioning with isoprenaline. Localization of Cx43 at the Inner Mitochondrial Membrane Previous studies with immunoelectron microscopy suggested the localization of Cx43 at the inner mitochondrial membrane. 15 However, the total size of the complex formed by the primary, the secondary antibody, and the gold microbead limits the spatial resolution of the technique and prevented us to determine the exact localization of Cx43. In the present study, treatment of isolated mitochondria with digitonin resulted in a marked reduction of VDAC immunoreactivity, whereas an ANT-specific signal was detected by confocal laser scan microscopy. Both intact mitochondria and mitoplasts (mitochondria devoid of the outer membrane) were positive for ANT and Cx43, thereby showing that Cx43 is not primarily located at the outer mitochondrial membrane. Western blot analysis confirmed the detection of Cx43 in the preparation containing the inner membrane and matrix or the inner and outer membrane and not the matrix. Taken together, our data indicate that Cx43 is located at the inner membrane of cardiomyocyte mitochondria. Mechanism of Translocation of Cx43 to the Inner Mitochondrial Membrane Coimmunoprecipitation experiments indicated an interaction of Cx43 with Tom20. Besides Tom5, -6, -7, -22, -40, and -70, Tom20 is part of the translocase of the outer mitochondrial membrane, which is the only known protein complex involved in the entering of nuclear-encoded proteins into mitochondria. 22 Mitochondrial matrix proteins and also some inner mitochondrial membrane proteins are synthesized as precursor proteins, characterized by having a cleavable amino-terminal targeting signal named presequence. These proteins bind TOM through Tom20, which acts as the presequence receptor. However, most of the internal membrane proteins lack this presequence and are targeted to mitochondria through not well defined internal sequences. These proteins form a complex with the cytosolic chaperones Hsp70 and Hsp90 that are recognized by Tom This mechanism is consistent with the observed coimmunoprecipitation of Cx43 and Tom20, as all TOM proteins form a complex, 23 the composition of which modifies its structural organization. 24 In addition, proteins lacking the aminoterminal presequence, as the mitochondrial phosphate carrier, have multiple recognition sequences and interact with both Tom70 and Tom After the recognition step, the precursor proteins are translocated to the intermembrane space through the Tom40 pore. Thereafter, proteins reach the internal membrane or the matrix through the TIM (Translocase of the Inner Membrane) complexes, mainly through the Tim23 subunit for presequence proteins and through Tim22 for internal sequence proteins. 22,23 As antibodies against Tim22 are not commercially available, we were not able to investigate an interaction of Cx43 with Tim22. Indeed, our data show that Cx43 interacts with Hsp90, making it likely that Cx43 is delivered to the TOM complex and subsequently imported into mitochondria via targeting by the chaperone Hsp90. To further asses the involvement of the Cx43 Hsp90 Hsp70 complex formation in the translocation of Cx43 to mitochondria, we analyzed the effect of geldanamycin on mitochondrial Cx43 content, both in normoxic rat hearts and in hearts submitted to ischemic or pharmacological preconditioning with diazoxide or isoprenaline. Geldanamycin, a benzoquinone ansamycin with cytostatic properties, binds to the N-terminal ATP-binding site of Hsp90 in vitro, destabilizing heterocomplexes of Hsp90 and its target proteins, 26 and thereby blocking with high affinity the ATP-driven chaperone cycle of Hsp Disruption of the chaperone/protein complex formation, as with the use of geldanamycin, inhibits the import of several mitochondrial proteins, such as the mitochondrial peptide transporter or the mitochondrial phosphate carrier. 21 No previous data have related short-term effects of treatment with geldanamycin with preconditioning. Our present data suggest that mitochondrial Cx43 is imported to the inner mitochondrial membrane by a chaperonemediated mechanism that is active during control normoxic conditions and that is involved in the enhanced translocation of Cx43 to the mitochondria induced by IP. The rapid reduction of mitochondrial Cx43 content after exposure to geldanamycin in the absence or presence of diazoxide indicates a rapid cycling of Cx43 between the cell compartments and mitochondria. In addition to the mitochondria, 8% of Cx43 were detected in the cytosol. Western blot analysis in the different cell fractions demonstrated that the amount of Cx43 in the mitochondrial compartment is very small as compared with total Cx43 protein. This explains why changes in the mitochondrial Cx43 content are not necessarily accompanied by detectable changes in other cell fractions. Cx43 Translocation to the Inner Mitochondrial Membrane and Cardioprotection of Preconditioning In the present study, a moderate reduction (20% to 40%) of mitochondrial Cx43 content by inhibiting Hsp90 with geldanamycin abolished the protection induced by diazoxide. There is strong evidence that diazoxide induces the protection of preconditioning through a signaling pathway involving ROS generation. 17,28 We have shown in previous studies that hearts from Cx43-deficient mice, and myocytes isolated from these hearts showing a marked reduction in total and even more so in mitochondrial Cx43 content, cannot be preconditioned by diazoxide, 16 because of a blunted generation of ROS after administration of the drug. Moreover, because

8 100 Circulation Research July 7, 2006 Figure 8. Proposed role of mitochondrial Cx43 in preconditioning protection based on the present results and previous studies. Mitochondrial Cx43 is required for K ATP -mediated ROS generation and, hence, its reduction blocks diazoxide protection. IP and isoprenaline induced protection through K ATP -mediated ROS production, but also by other redundant pathways, and were not abolished by subthreshold reductions in mitochondrial Cx43 content. In addition, IP increased Cx43 translocation of Cx43 to mitochondria via Hsp90 (HSP) and prevented reduction of mitochondrial Cx43 content by geldanamycin (but not by genetic ablation). -AR indicates -adrenoreceptor; 5-HD, 5-hydroxydecanoate; MAPK, mitogen-activated protein kinase; MPG, mercaptopropionyl glycine; MPTP, mitochondrial permeability transition pore; PKs, protein kinases; RISK, reperfusion injury salvage kinase. ROS formation with menadione was not affected by Cx43 deficiency, it was suggested that Cx43 deficiency was associated with a specific defect in diazoxide-induced ROS formation. The present results provide evidence directly linking mitochondrial Cx43 content to ROS-dependent, diazoxide-induced protection. Reduction of mitochondrial Cx43 content with geldanamycin before application of either IP or isoprenaline did not abolish the protection against cell death afforded by these treatments. The different effect of geldanamycin on protection induced by diazoxide (abolition) or by IP or isoprenaline (no effect) may be explained by the redundancy of signal transduction pathways in IP. IP elicits protection by activating several signal transduction pathways that act in parallel and have, at least to some extent, additive effects on the magnitude of protection. 4,5,29 Mitochondrial ATP dependent, potassium channel dependent ROS generation represents only 1 of these pathways, and several studies have shown that protection may still be elicited by brief ischemic episodes when mitochondrial ATP dependent potassium channels (mitok ATP) are blocked For example, adenosineinduced preconditioning is independent of mitok ATP in rabbits. 33 The discrepant effect of geldanamycin on the protective effects induced by ischemia/isoprenaline and by diazoxide could thus reflect the fact that ablation of 1 of the signaling pathways involved in IP, ie, ROS generation, is not sufficient to abolish protection, whereas it may abolish protection induced by a drug, such as diazoxide, specifically stimulating that pathway. 17,28 A scheme of the role of mitochondrial Cx43 in the framework of the protection of preconditioning is shown in Figure 8. The lack of effect of geldanamycin on the protection of IP may appear to be at odds with previous studies showing that reduced Cx43 expression in Cx43-deficient mice is associated with abrogation of the protection of preconditioning. 10 The explanation for this apparent discrepancy may be related to the fact that whereas geldanamycin induces a moderate reduction in mitochondrial Cx43 without changes in total Cx43, in Cx43-deficient mice, mitochondrial Cx43 content was markedly reduced to approximately 20% of control values, 16 and this reduction was not confined to the mitochondrial compartment. Considering the results of this study, together with previous studies, it appears that mitochondrial Cx43 is essential for mitok ATP-mediated protection (even mild reductions in mitochondrial Cx43 abolish this pathway), whereas only severe and/or global reductions in Cx43 abolish mitok ATP-independent protection. Unanswered Questions: The Molecular Interactions and Functions of Mitochondrial Cx43 The present study does not provide direct information on the molecular mechanism by which mitochondrial Cx43 participates in mitok ATP-mediated ROS generation and cardioprotection. Cx43 could be part of the multiprotein complexes forming mitok ATP channels, the molecular composition of which has not been elucidated, 34,35 or could somehow modify/gate the mitok ATP-induced ROS formation. Cx43 hexamers could also form hemichannels in the inner mitochondrial membrane that could contribute to K and water fluxes under certain circumstances or, as it occurs in membrane hemichannels, 36,37 be the origin of signal transduction cascades involving mitok ATP. Previous studies have shown that Cx43 is part of multiprotein protein kinase C (PKC)-signaling complexes potentially important in preconditioning. 38 The role of mitochondrial Cx43 in the protection of preconditioning could also be explained by interactions with the mitochondrial permeability transition pore, a multiprotein channel, the molecular composition of which has not yet been fully established. 39 Acknowledgments We appreciate the technical assistance of M. Angeles Garcia.

9 Rodriguez-Sinovas et al Mitochondrial Connexin 43 and Cardioprotection 101 Sources of Funding Partially supported by the Spanish Ministries of Education and Science (grant CICYT-SAF/ ) and Health (RECAVA). A.R.-S. (FIS 99/3142) and A.C. have a personal contract grant from the Instituto de Salud Carlos III. R.S. was the recipient of a grant from the Deutsche Forschungsgemeinschaft (Schu 843/7-1). None. Disclosures References 1. Garcia-Dorado D. Myocardial cell protection in acute coronary syndromes. In: Theroux P, ed. Acute Coronary Syndromes. A Companion to Braunwald s Heart Disease. Philadelphia: Saunders; Piper HM, Garcia-Dorado D, Ovize M. A fresh look at reperfusion injury. Cardiovasc Res. 1998;38: Piper HM, Abdallah Y, Schafer C. The first minutes of reperfusion: a window of opportunity for cardioprotection. Cardiovasc Res. 2004;61: Schulz R, Cohen MV, Behrends M, Downey JM, Heusch G. Signal transduction of ischemic preconditioning. Cardiovasc Res. 2001;52: Yellon DM, Downey JM. Preconditioning the myocardium: from cellular physiology to clinical cardiology. Physiol Rev. 2003;83: Davis LM, Rodefeld ME, Green K, Beyer EC, Saffitz JE. Gap junction protein phenotypes of the human heart and conduction system. J Cardiovasc Electrophysiol. 1995;6: Jalife J, Morley GE, Vaidya D. Connexins and impulse propagation in the mouse heart. J Cardiovasc Electrophysiol. 1999;10: Garcia-Dorado D, Rodriguez-Sinovas A, Ruiz-Meana M. Gap junctionmediated spread of cell injury and death during myocardial ischemiareperfusion. Cardiovasc Res. 2004;61: Schulz R, Heusch G. Connexin 43 and ischemic preconditioning. Cardiovasc Res. 2004;62: Schwanke U, Konietzka I, Duschin A, Li X, Schulz R, Heusch G. No ischemic preconditioning in heterozygous connexin43-deficient mice. Am J Physiol Heart Circ Physiol. 2002;283:H1740 H Li X, Heinzel FR, Boengler K, Schulz R, Heusch G. Role of connexin 43 in ischemic preconditioning does not involve intercellular communication through gap junctions. J Mol Cell Cardiol. 2004;36: Padilla F, Garcia-Dorado D, Rodriguez-Sinovas A, Ruiz-Meana M, Inserte J, Soler-Soler J. Protection afforded by ischemic preconditioning is not mediated by effects on cell-to-cell electrical coupling during myocardial ischemia-reperfusion. Am J Physiol Heart Circ Physiol. 2003;285: H1909 H Garcia-Dorado D, Inserte J, Ruiz-Meana M, Gonzalez MA, Solares J, Julia M, Barrabes JA, Soler-Soler J. Gap junction uncoupler heptanol prevents cell-to-cell progression of hypercontracture and limits necrosis during myocardial reperfusion. Circulation. 1997;96: Rodriguez-Sinovas A, Garcia-Dorado D, Ruiz-Meana M, Soler-Soler J. Enhanced effect of gap junction uncouplers on macroscopic electrical properties of reperfused myocardium. J Physiol. 2004;559: Boengler K, Dodoni G, Rodriguez-Sinovas A, Cabestrero A, Ruiz-Meana M, Gres P, Konietzka I, Lopez-Iglesias C, Garcia-Dorado D, Di Lisa F, Heusch G, Schulz R. Connexin 43 in cardiomyocyte mitochondria and its increase by ischemic preconditioning. Cardiovasc Res. 2005;67: Heinzel FR, Luo Y, Li X, Boengler K, Buechert A, Garcia-Dorado D, Di Lisa F, Schulz R, Heusch G. Impairment of diazoxide-induced formation of reactive oxygen species and loss of cardioprotection in connexin 43 deficient mice. Circ Res. 2005;97: Pain T, Yang XM, Critz SD, Yue Y, Nakano A, Liu GS, Heusch G, Cohen MV, Downey JM. Opening of mitochondrial K(ATP) channels triggers the preconditioned state by generating free radicals. Circ Res. 2000;87: Inserte J, Garcia-Dorado D, Ruiz-Meana M, Padilla F, Barrabes JA, Pina P, Agullo L, Piper HM, Soler-Soler J. Effect of inhibition of Na( )/ Ca(2 ) exchanger at the time of myocardial reperfusion on hypercontracture and cell death. Cardiovasc Res. 2002;55: Inserte J, Garcia-Dorado D, Hernando V, Soler-Soler J. Calpain-mediated impairment of Na /K -ATPase activity during early reperfusion contributes to cell death after myocardial ischemia. Circ Res. 2005;97: Miura T, Ohnuma Y, Kuno A, Tanno M, Ichikawa Y, Nakamura Y, Yano T, Miki T, Sakamoto J, Shimamoto K. Protective role of gap junctions in preconditioning against myocardial infarction. Am J Physiol Heart Circ Physiol. 2003;286:H214 H Young JC, Hoogenraad NJ, Hartl FU. Molecular chaperones Hsp90 and Hsp70 deliver preproteins to the mitochondrial import receptor Tom70. Cell. 2003;112: Truscott KN, Brandner K, Pfanner N. Mechanisms of protein import into mitochondria. Curr Biol. 2003;13:R326 R Rehling P, Brandner K, Pfanner N. Mitochondrial import and the twin-pore translocase. Nat Rev Mol Cell Biol. 2004;5: Model K, Prinz T, Ruiz T, Radermacher M, Krimmer T, Kuhlbrandt W, Pfanner N, Meisinger C. Protein translocase of the outer mitochondrial membrane: role of import receptors in the structural organization of the TOM complex. J Mol Biol. 2002;316: Brix J, Rudiger S, Bukau B, Schneider-Mergener J, Pfanner N. Distribution of binding sequences for the mitochondrial import receptors Tom20, Tom22, and Tom70 in a presequence-carrying preprotein and a non-cleavable preprotein. J Biol Chem. 1999;274: Stebbins CE, Russo AA, Schneider C, Rosen N, Hartl FU, Pavletich NP. Crystal structure of an Hsp90-geldanamycin complex: targeting of a protein chaperone by an antitumor agent. Cell. 1997;89: Young JC, Hartl FU. Polypeptide release by Hsp90 involves ATP hydrolysis and is enhanced by the co-chaperone p23. EMBO J. 2000;19: Krenz M, Oldenburg O, Wimpee H, Cohen MV, Garlid KD, Critz SD, Downey JM, Benoit JN. Opening of ATP-sensitive potassium channels causes generation of free radicals in vascular smooth muscle cells. Basic Res Cardiol. 2002;97: Inserte J, Garcia-Dorado D, Ruiz-Meana M, Agullo L, Pina P, Soler-Soler J. Ischemic preconditioning attenuates calpain-mediated degradation of structural proteins through a protein kinase A-dependent mechanism. Cardiovasc Res. 2004;64: Grover GJ, Murray HN, Baird AJ, Dzwonczyk S. The KATP blocker sodium 5-hydroxydecanoate does not abolish preconditioning in isolated rat hearts. Eur J Pharmacol. 1995;277: Lim KH, Javadov SA, Das M, Clarke SJ, Suleiman MS, Halestrap AP. The effects of ischaemic preconditioning, diazoxide and 5-hydroxydecanoate on rat heart mitochondrial volume and respiration. J Physiol. 2002;545: Hanley PJ, Mickel M, Loffler M, Brandt U, Daut J. K(ATP) channelindependent targets of diazoxide and 5-hydroxydecanoate in the heart. J Physiol. 2002;542: Cohen MV, Yang XM, Liu GS, Heusch G, Downey JM. Acetylcholine, bradykinin, opioids, and phenylephrine, but not adenosine, trigger preconditioning by generating free radicals and opening mitochondrial K(ATP) channels. Circ Res. 2001;89: Ardehali H, O Rourke B. Mitochondrial K(ATP) channels in cell survival and death. J Mol Cell Cardiol. 2005;39: Halestrap AP, McStay GP, Clarke SJ. The permeability transition pore complex: another view. Biochimie. 2002;84: Plotkin LI, Aguirre JI, Kousteni S, Manolagas SC, Bellido T. Bisphosphonates and estrogens inhibit osteocyte apoptosis via distinct molecular mechanisms downstream of extracellular signal-regulated kinase activation. J Biol Chem. 2005;280: Przyklenk K, Maynard M, Darling CE, Whittaker P. Pretreatment with D-myo-inositol trisphosphate reduces infarct size in rabbit hearts: role of inositol trisphosphate receptors and gap junctions in triggering protection. J Pharmacol Exp Ther. 2005;314: Ping P, Song C, Zhang J, Guo Y, Cao X, Li RC, Wu W, Vondriska TM, Pass JM, Tang XL, Pierce WM, Bolli R. Formation of protein kinase C(epsilon)-Lck signaling modules confers cardioprotection. J Clin Invest. 2002;109: Halestrap AP, Clarke SJ, Javadov SA. Mitochondrial permeability transition pore opening during myocardial reperfusion a target for cardioprotection. Cardiovasc Res. 2004;61: