Uncoupling Proteins 2 and 3 Function in Concert to Augment Tolerance to Cardiac Ischemia *

Transient cardiac ischemia activates cell survival signaling, conferring subsequent ischemia tolerance to the heart. This biological phenomenon, termed ischemic preconditioning, results in improved clinical outcome and attenuated infarct size following myocardial infarction. To explore genomic modifications underpinning this ischemia tolerance, we delineated the regulation and functionofthecardiacenrichedmitochondrialuncouplingproteins 2 and 3 during delayed ischemic preconditioning in the rat. Cardiac transcripts of genes encoding uncoupling proteins 2 and 3 are up-regulated in parallel with infarct size reduction in preconditioned hearts. Mitochondria isolated from preconditioned hearts exhibit an augmented inducible proton leak. In parallel, following anoxia-reoxygenation these mitochondria generate less hydrogen peroxide compared with non-preconditioned mitochondria. Preconditioning in rat cardiac derived myoblasts is abolished following uncoupling protein-2 depletion by RNA-interference. RNAi of uncoupling protein-3 partially attenuates the capacity to precondition these cells. Functional characterization of anoxia and reoxygenation tolerance following uncoupling protein 2 or 3 and combined 2 and 3 RNAi shows the largest reduction in viability follows deple-tionofbothhomologues.Uncouplingprotein-2depletionalonesig- nificantly attenuates anoxia-reoxygenation

The cell survival program, termed ischemic preconditioning, evoked by transient nonlethal tissue ischemia, is evolutionarily conserved and is evident across multiple organs/tissues. The signal transduction networks engendering this phenomenon are well described (1); however, functional cellular modifications conferring ischemia tolerance are less tangible. Interestingly, the invariable metabolic signature of preconditioned tissue is enhanced capacity to restore mitochondrial function and ATP production following an ischemic insult (2). The mechanistic links between the molecular adaptations of the mitochondrion and the preconditioning cell survival program require further characterization. Further credence to the importance of mitochondrial adaptation to preconditioning-induced cell survival has been shown by improved mitochondrial respiratory and bioenergetic recovery following anoxia and reoxygenation in mitochondria extracted from ischemic, genomic, and pharmacologic preconditioned hearts (3)(4)(5). In addition, preconditioned mitochondria generate less reactive oxygen species (ROS) 3 at reoxygenation than do nonpreconditioned mitochondria (5)(6)(7). The broad molecular regulatory programs underpinning mitochondrial adaptations in preconditioning is further underscored by additional modulations in nuclear-encoded proteins modulating mitochondrial function during preconditioning (8,9).
Utilizing delayed ischemic preconditioning to identify transcriptional regulatory proteins governing mitochondrial biology and homeostasis, we previously demonstrated that the mitochondrial biogenesis regulatory protein peroxisome proliferator-activated receptor ␥ co-activator 1␣ is up-regulated in the preconditioned rat heart (3). Since peroxisome proliferator-activated receptor ␥ co-activator 1␣ up-regulates mitochondrial uncoupling proteins (10), we reasoned that this family of inner mitochondrial membrane carrier proteins may be functional components of the delayed preconditioning program.
In this report, we examine the molecular and biochemical regulation of cardiac enriched inner mitochondrial membrane uncoupling proteins (UCP2 and UCP3) in delayed preconditioning and as a component of innate cellular tolerance to anoxia-reoxygenation. We demonstrate that UCP2 and -3 are up-regulated by the delayed ischemic preconditioning program. We show that these regulatory events are ROS-inducible and appear to attenuate subsequent anoxia-reoxygenation-mediated mitochondrial ROS production. Using RNA interference in rat heart H9c2 cells, we functionally demonstrate the individual and combined requirement of UCP2 and UCP3 in the preconditioning program and in innate cellular tolerance against anoxia-reoxygenation-provoked cell death.

EXPERIMENTAL PROCEDURES
Animal Experiments-The ischemic preconditioning trigger was achieved by conferring transient regional ischemia to the left coronary artery of Wistar rats as described previously (3). Here, the heart was subjected to five episodes of 3 min of ischemia coupled to 5 min of reperfusion. Sham operations involved encircling the left coronary artery without tightening the suture. Since inhibition of antioxidant signaling abolishes the preconditioning program, 2-mercaptopropionyl glycine (2-MPG) was administered to an additional group of rats (0.42 * This work was supported by the Division of Intramural Research, National Institutes of Health (NIH) and NHLBI (NIH). 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 U.S.C. Section 1734 solely to indicate this fact. 1  mg/kg/min 2-MPG), beginning 15 min before preconditioning ischemia and continued until 1 h after the last occlusion as an additional control group. Following chest closure, animals were allowed to recover. Hearts were then extracted at 4 h for RNA extraction and at 24 h following the procedure for (i) mitochondrial isolation (11) and analysis and (ii) Western blot analysis. Animal experiments were approved by the NHLBI Animal Care and Use Committee. Quantitative Real Time PCR and Western Blot Analysis-Total RNA samples were treated with DNase I (Invitrogen) and reverse-transcribed using oligo(dT) 12-18, SuperScript TM (Invitrogen). Real time quantitative PCR analysis was performed using an MJResearch Opticon II and SYBR Green (Stratagene) as the fluorophore. Standards were prepared using serial dilutions of cDNA in the range of 1 million to 1 copy per reaction, and gene expression was quantified by comparison with the standard curve. All gene expression was normalized to 18 S rRNA and glyceraldehyde-3-phosphate dehydrogenase as internal controls.
UCP2 antibodies from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA) and Chemicon were used for immunoblot analysis. 100 g of crude whole cell protein extract was run using standard SDS-PAGE techniques, and the secondary antibody was conjugated to horseradish peroxidase. The antigen-antibody reaction was detected using enhanced chemiluminescence reagents (Amersham Biosciences), and ␤-actin levels and Ponceau staining (Sigma) were measured to confirm protein integrity and equal loading.
Mitochondrial Isolation and Respiratory Analysis-Mitochondria were isolated (11) and respiratory function evaluated using fiber-optic oxygen spectrophotometry (Instech Laboratories). Mitochondria (0.35 mg⅐ml Ϫ1 ) were resuspended in a standard sucrose-Tris assay medium (pH 7.2), and basal respiratory rates were calculated in the presence of glutamate, malate, and ADP (5, 5, and 500 nM, respectively). Basal proton leak kinetics were evaluated with the addition of oligomycin (1 g⅐ml Ϫ1 ), whereas the inducible proton leak measurements were performed in the presence of succinate (4 mM) and oligomycin (1 g⅐ml Ϫ1 ). GDP (500 M) was subsequently added to assess the response to this UCP inhibitor (12,13).
Mitochondrial Anoxia-Reoxygenation and Hydrogen Peroxide Production-Mitochondria (0.35 mg⅐ml Ϫ1 ) were resuspended in the assay buffer (containing excess ADP, 5 mM glutamate, and 5 mM malate) in a sealed, stirred chamber. Anoxia was induced in Յ3 min via consumption of available oxygen in the air-tight chamber. Anoxia was maintained for 25 min. The chamber was then bubbled with room air, and hydrogen peroxide (H 2 O 2 ) generation was subsequently determined using the p-hydroxyphenylacetic acid assay (14). The rate of H 2 O 2 production was calculated by measuring the increase in fluorescence (excitation, 320 nm; emission, 400 nm) in the presence of this ROS-sensitive probe (p-hydroxyphenylacetic acid, 50 mg⅐ml Ϫ1 ) coupled to the oxidation of horseradish peroxidase (4 units⅐ml Ϫ1 ), together with fresh substrate (14). The fluorescence was correlated against H 2 O 2 standards, which were run concurrently.

H9c2 Cell Ischemic Preconditioning and Anoxia-Reoxygenation
Protocols-RNA interference to deplete UCP2, UCP3, or both transcripts was performed in the rat H9c2 cell survival model (15). This immortalized cell line is derived from embryonic BD1X rat heart tissue and was obtained from the American Type Culture Collection. At 75-85% confluence, cells were immersed in a glucose-depleted, sodium pyruvate-free and fetal bovine serum-free medium to simulate ischemia. The cells were then subjected to total anoxia by being placed within specifically designed anoxic bags (GasPak™; BD Biosciences). The cells endured 20 h of this state before being opened and reperfused with normal growth medium. The reperfusion phase lasted 2 h within the 5% CO 2 incubator. To precondition these cells, the anoxia-reoxygenation experiments were preceded by a 2-h anoxia followed by an 18-h recovery in normal growth medium prior to the anoxia-reoxygenation protocol.
Specific small interfering RNAs aligned to rat UCP2 and UCP3 were designed by Ambion. H9c2 cells were transfected with these interfering RNAs using Lipofectamine 2000 (Invitrogen). The efficacy of gene knockdown was assessed using real time quantitative PCR for UCP2 and UCP3 and by Western blot analysis for UCP2. The control RNAi is designed by Ambion and has no homology with any rat mRNA or DNA sequence.
The cells underwent the transfection procedure 48 h prior to being inserted into the anoxic bags for exposure to the index anoxia and 2-h reperfusion protocol. For the preconditioning protocols, a preconditioning 2-h anoxic trigger was applied ϳ28 h following transfection so that all index anoxia-reoxygenation studies were uniformly performed 48 h following transfection. The index ischemia for the preconditioning studies was 20 h and reduced to 16 h for nonpreconditioned experiments.
The cells were analyzed in a FACSCalibur (BD Biosciences) for either cellular viability (by 7-AAD exclusion) or ROS for the amount of production (using chloromethyl-2Ј,7Ј-dichlorodihydrofluorescein diacetate).
Statistical Methods-Differences between data groups were evaluated for significance using Student's t test or one-way analysis of variance. p Ͻ 0.05 was considered statistically significant, and data are expressed as mean Ϯ S.E.

RESULTS
Using the rat heart delayed preconditioning protocol (3), we show that UCP2 and UCP3 gene transcripts are up-regulated (255 Ϯ 21 and 194 Ϯ 30%, respectively, p Ͻ 0.001) 4 h following ischemic preconditioning versus sham-operated controls (Fig. 1A). Interestingly, this gene regulatory event is abrogated by inhibition of delayed preconditioning with the antioxidant and preconditioning antagonist 2-MPG (16). In parallel with gene expression, UCP2 steady-state protein levels are upregulated 24 h following the ischemic trigger (Fig. 1B). The UCP3 transcript is 10-fold less abundant than UCP2 in the rat heart. This probably accounts for our inability to measure steady-state protein levels of UCP3.

Delayed preconditioned mitochondria exhibit a GDP-sensitive proton leak
Oxygen consumption is measured in nmol⅐O 2 Ϫ1 ⅐mg of protein Ϫ1 ⅐min Ϫ1 . Succinate and oligomycin respiration represents steady-state oxygen consumption under these conditions. The addition of GDP reflects the respiratory rate after blockade of UCP-sensitive oxygen consumption. The change in respiration reflects the percentage change in these two rates. Values are reflected as mean Ϯ S.E. n Ն 6 experiments in both groups. To establish whether the up-regulation of these inner mitochondrial carrier proteins modulate respiratory function, mitochondria were isolated from rat hearts 24 h following ischemic preconditioning and sham procedures. Basal glutamate/malate-driven mitochondrial oxygen consumption (State 3 respiration) and oligomycin-insensitive respiration (basal leak) are similar between preconditioned and nonpreconditioned mitochondria, and no differences in basal ROS levels were measured (data not shown). Interestingly, UCP2 and -3 have recently been shown to be quiescent during basal respiration but demonstrate a capacity to induce a mitochondrial proton leak in response to ROS signaling (17,18). Mitochondrial ROS was generated via succinate-facilitated reverse electron flow, in conjunction with F 1 F 0 -ATPase inhibition using oligomycin (19,20). Here, preconditioned mitochondria display a 15 Ϯ 3.9% greater inducible proton leak compared with sham control mitochondria (TABLE ONE). Notably, this enhanced inducible leak is inhibited by the UCP antagonist GDP (12,13) (TABLE ONE). Since up-regulation of an inducible proton leak following delayed preconditioning would be compatible with an attenuation of mitochondrial ROS generation in response to anoxia-reoxygenation, we evaluated the generation of H 2 O 2 at postanoxic reoxygenation. The delayed preconditioned mitochondria exhibited a 51.4 Ϯ 7.6% reduction in H 2 O 2 production at reoxygenation following 25 min of in vitro anoxia compared with sham control mitochondria as measured fluoroscopically (Fig. 2).

Succinate
In order to establish the functional requirement of UCP2 and UCP3 in modulating tolerance to preconditioning and anoxia-reoxygenation, we employed RNA interference to deplete UCP2 and -3 in rat H9c2 cells. In parallel with the rat heart, UCP2 is the predominant uncoupling protein in H9c2 cells and is expressed in excess of 20-fold greater abundance than the UCP3 transcript (data not shown). RNAi resulted in a 75% reduction in UCP2 transcript and an 85% reduction in UCP3 and a similar combined depletion in response to RNAi of both constructs as measured by real time quantitative PCR (Fig. 3A). UCP2 steady-state protein levels were reduced to 39 Ϯ 11% of base-line expression (Fig.  3B). Gene silencing was maintained from 24 to 72 h following transfection. RNAi and scrambled control RNAi constructs were maintained in equal molar amounts in all experimental groups, and cell viability was assessed by flow cytometry using the nuclear stain 7-AAD. Small interfering RNA treatment did not modulate cellular viability under nonstressed conditions over a 72-h period (data not shown).
The H9c2 cells exhibit anoxia-induced preconditioning with 84 Ϯ 4% viability versus 22 Ϯ 9% viability in nonpreconditioned controls in response to the index anoxia-reoxygenation insult (p Ͻ 0.001; Fig. 4). Prior depletion of UCP2 completely abolished preconditioning-mediated protection, whereas UCP3 depletion partially abolished the protective phenotype. Viability, however, was significantly greater than nonpreconditioned cells but yet significantly less than that seen in control RNAi preconditioned cells (Fig. 4).
As the preconditioning trigger induces modifications in cellular tolerance to subsequent ischemia-reperfusion injury, we then evaluated whether genomic depletion of the cardiac enriched UCPs directly modulates anoxia-reoxygenation injury following 16 h of anoxia and 2 h of  re-oxygenation. Viability was incrementally diminished from modest but nonsignificant effects on viability with UCP3 depletion to increasingly significant loss of viability following UCP2 and combined UCP2 and UCP3 depletion compared with scrambled RNAi transfected controls (Fig. 5, A and B). The cell viability diminished from 60 Ϯ 4% in the scrambled RNAi controls to 47 Ϯ 2% viability with UCP3 depletion (p ϭ not significant) to 31 Ϯ 7% following UCP2 RNAi (p Ͻ 0.05) to 19 Ϯ 5% (p Ͻ 0.001) following the UCP2 and UCP3 depletion (Fig. 5, A and B). A putative mechanism whereby depletion of UCP2 and the combined UCP2 and UCP3 depletion would augment cell death in response to anoxia-reoxygenation is via increased ROS production. To assay ROS generation, the amount of dichlorofluorescein (DCF) fluorescence was measured post-anoxia-reoxygenation, using cells gated for viability (7-AAD-impermeable). The amount of DCF fluorescence inversely correlated with cell viability. The least ROS was present in cells transfected with scrambled RNAi, and the most ROS was found in cells depleted in both UCP2 and UCP3 (Fig. 5, C and D). Interestingly, in the knockdown experiments, viability was rescued when the cells were incubated with 1 mM 2-MPG throughout the anoxia-reoxygenation protocol (Fig. 6A). In concert, DCF fluorescence and hence ROS production was also attenuated with 2-MPG (Fig. 6B).

DISCUSSION
Our results show that the delayed preconditioning program up-regulates UCP2 and UCP3 in the rat heart. This genomic regulation is associated with an inducible mitochondrial GDP-dependent proton leak and an augmented capacity to ameliorate postanoxic H 2 O 2 production. In parallel, the preconditioning phenotype is abolished following UCP2 RNAi depletion and partially blunted by UCP3 depletion in H9c2 cells. Functional characterization of these inner mitochondrial membrane proteins in the rat heart-derived myoblasts demonstrates that depletion of UCP2 or both UCP2 and UCP3 is associated with a diminished cellular tolerance to anoxia-reoxygenation injury and is associated with increased production of postanoxic ROS. Conversely, this susceptibility to anoxia-reoxygenation injury is annulled by the concurrent administration of the antioxidant 2-MPG. Together, these data strongly support a pivotal role of the cardiac enriched uncoupling proteins in innate cardiomyoblast ischemia tolerance and demonstrate that these  following preconditioning in H9c2 cells with control RNAi (PC) and in preconditioned cells that had been previously transfected with RNAi to UCP2 and UCP3, respectively. The index anoxic insult in these experiments was of 20-h duration to profoundly reduce viability in control cells to demonstrate the cytoprotection of preconditioning. B, representative flow cytometric overlaid images, displaying impermeable (viable) and nucleusincorporated (nonviable) 7-AAD peaks. Black histogram, anoxia-reoxygenation with control RNAi; green histogram, preconditioning with control RNAi; orange histogram, preconditioning with UCP2 RNAi; purple histogram, preconditioning with UCP3 RNAi. All of these experiments were performed at least four times in triplicate. *, p Ͻ 0.001 versus control RNAi cells; **, p Ͻ 0.01 versus preconditioning with control RNAi; ***, p value Ͻ 0.01 versus PC with control RNAi versus control RNAi cells and versus preconditioning with UCP2 RNAi.
inner mitochondrial membrane carrier proteins are modulated as putative adaptive components of the delayed ischemic preconditioning cell survival program.
Twenty-four hours following the ischemic preconditioning trigger, the intact beating heart exhibits hemodynamic parity with nonpreconditioned hearts (3). This observation probably reflects similar energetic demand of preconditioned and nonpreconditioned hearts in the absence of additional biomechanical stress. Here, the observation that basal state 3 respiration in delayed preconditioned mitochondria is unaltered, despite the up-regulation of UCP2 and UCP3, is compatible with this hemodynamic profile. Furthermore, these data suggest that the preconditioning-mediated up-regulation of UCP2 and UCP3 are functionally quiescent during basal mitochondrial function. This quiescent nature of increased levels of uncoupling proteins aligns with the observations in transgenic murine hearts overexpressing UCP1 (21). Here Bouillaud and co-workers demonstrate that under normoxic conditions, the UCP1 transgenic mice have similar oxygen consumption, contractility, and energetic yield compared with littermate control mice (21).
Interestingly, recent studies demonstrate that UCP2 is activated/induced via ROS signaling (17,18). Thus, we would postulate that a subsequent index ischemia-reperfusion-mediated oxidative stress could function as a trigger to mediate uncoupling protein activation. This concurs with observations that UCP-mediated proton leak is unmasked by ROS generation via reverse electron transfer through complex I of the electron transfer chain (19,20). This modus operandi of UCPs as ROS detoxifying proteins supports their role as end effectors in the preconditioning-induced cell survival program. Moreover, it is feasible that this protein may function in an autoregulatory feed-forward manner to protect the mitochondria and cell from future excessive ROS generation (22).
The role of excessive ROS-mediated damage in cellular injury following ischemia-reperfusion is subject of an ongoing debate. Nevertheless, a significant body of evidence implicates a substantial role for ROS suppression in the attenuation of ischemia-reperfusion injury (23). Furthermore, previous studies have demonstrated that ROS scavenging proteins, such as SOD2, are up-regulated following delayed preconditioning (8). SOD2 overexpression in turn confers protection against cardiac ischemia reperfusion injury (24). Our data provide additional evidence to support the control of ROS as a mechanism to sustain cellular viability in response to anoxia-reoxygenation. The reduction in postanoxic ROS in the isolated preconditioned mitochondria probably encompasses both the up-regulation of the uncoupling proteins and other known antioxidant defenses, such as SOD2, which are activated by the preconditioning program. However, our gene depletion studies do support the involvement of the UCP homologues in postanoxic ROS generation. Taken together, this study suggests an endogenous regulatory program whereby these inner mitochondrial membrane transport proteins are up-regulated and functionally activated by ROS to depress subsequent mitochondrial ROS production. These findings are compatible with the known evolutionarily conserved function of the uncoupling proteins to modulate mitochondrial ROS generation (25,26) as evidenced in potatoes, birds, and mammals (12,20,27,28). A postulated mechanism of ROS reduction by the mitochondrial uncoupling proteins is based on changes in inner mitochondrial membrane potential. When the mitochondrial membrane potential is high, superoxide production increases as a result on nonspecific single-electron reductions of molecular oxygen at complexes I and III of the electron transfer chain. Activation of uncoupling proteins results in mild proton leak across the inner mitochondrial membrane with a modest reduction in mitochondrial membrane potential, with a limit in superoxide production (22).
Concordantly, previous gain-of-function genomic studies support the induction of uncoupling proteins as a mechanism to attenuate ischemic injury in the heart, in brain endothelial cells, and in H9c2 cells (15,21,26,29,30) in parallel with a diminution of ROS production (15,21,29,30). UCP2 and UCP3 depletion by RNA interference in this study supports this paradigm in that the viable UCP2 or combined UCP2-and UCP3-depleted cells show excess ROS generation following anoxiareoxygenation compared with the H9c2 cells with maintained UCP levels. Interestingly, the effects of uncoupling proteins may be tissue-spe-cific, as evident by a lack of mitochondrial uncoupling effects following UCP2 ablation in spleen and lung mitochondria (31). Furthermore, a contrary result was observed where UCP2 knockout mice demonstrate resistance to cerebral ischemic injury (32). In these mice, the persistent depletion of UCP2 is associated with the induction of alternative antioxidant defense mechanisms such as SOD2 and glutathione (32). We speculate that this induction of adaptive and compensatory ROS-controlling enzymes may indirectly support the role of UCP2, in that its chronic depletion results in activation of complementary programs to control ROS biology. In data not shown, we demonstrate no change in SOD2, catalase, or mitochondrial glutathione S-transferase gene expression in H9c2 cells depleted in UCP2 and/or UCP3. A possible explanation for the lack of up-regulation of alternate antioxidant defense encoding genes in our study may reflect the partial reduction in the UCP proteins in these experiments versus the complete lack of UCP2 in the knockout mouse.
Mild uncoupling of oxidative phosphorylation is suggested to regulate the intrinsic capacity of mitochondria to control energy conservation and simultaneously manage reactive oxygen species disposal (33) above and beyond its classic role in thermogenesis. The cytoprotective response to the nonthermogenic effects of mild mitochondrial uncoupling is further supported by the cardioprotective and neuroprotective effects of pharmacologic mitochondrial uncouplers (34 -36). The data presented in this study suggest that intrinsic uncoupling proteins in the mitochondria are modulated by innate cell survival signaling as "triggered" by preconditioning. The RNA interference studies demonstrate that the mitochondrial uncoupling proteins are necessary components of cellular tolerance to oxidative stress and furthermore mediate the delayed preconditioning program. Finally, we propose that these data enhance our understanding of how the mitochondria function as a cellular rheostat balancing cellular prosurvival and cell death homeostatic programs. However, it should be noted that the roles of the various mitochondrial uncoupling protein homologues are still being investigated and have a multitude of metabolic and signaling effects, (22,37,38) that have not been investigated in this study but may contribute to their role in cellular protection.
In conclusion, we previously demonstrated that isolated mitochondria following activation of the delayed preconditioning cell survival program are more tolerant of ex vivo anoxia-reoxygenation (3). Here we demonstrate that this resilient mitochondrial phenotype is associated with an up-regulation of the cardiac enriched UCP2 and UCP3 genes. In addition, preconditioned mitochondria exhibit a ROS-inducible modest uncoupling of oxidative phosphorylation as quantified by oligomycin-insensitive respiration in response to GDP inhibition. These preconditioned mitochondria also shown enhanced capacity to attenuate ROS generation following anoxia-reoxygenation. The genetic depletion of the UCP2 and UCP3 in mitochondria in situ further demonstrates that these inner mitochondrial membrane proteins are a necessary component of cellular antioxidant defense mechanisms, whose effects can be rescued by exogenous ROS scavenging. These findings strongly support UCP2 and UCP3 as endogenous cytoprotective proteins that are both regulated and induced in response to ischemic stress. Their cytoprotective effects parallel the relative expression of these two mitochondrial uncoupling protein homologues in that UCP2 depletion has greater susceptibility to anoxia than does UCP3 depletion. Finally, UCP2 and UCP3 are shown to be necessary components of innate cellular antioxidant defense. These data support a functional role of the uncoupling proteins in nonthermogenic tissue and implicate the mitochondrial uncoupling proteins as potential therapeutic targets for the management of ischemic diseases.