Mitochondrial Complex II Prevents Hypoxic but Not Calcium- and Proapoptotic Bcl-2 Protein-induced Mitochondrial Membrane Potential Loss*

Mitochondrial membrane potential loss has severe bioenergetic consequences and contributes to many human diseases including myocardial infarction, stroke, cancer, and neurodegeneration. However, despite its prominence and importance in cellular energy production, the basic mechanism whereby the mitochondrial membrane potential is established remains unclear. Our studies elucidate that complex II-driven electron flow is the primary means by which the mitochondrial membrane is polarized under hypoxic conditions and that lack of the complex II substrate succinate resulted in reversible membrane potential loss that could be restored rapidly by succinate supplementation. Inhibition of mitochondrial complex I and F0F1-ATP synthase induced mitochondrial depolarization that was independent of the mitochondrial permeability transition pore, Bcl-2 (B-cell lymphoma 2) family proteins, or high amplitude swelling and could not be reversed by succinate. Importantly, succinate metabolism under hypoxic conditions restores membrane potential and ATP levels. Furthermore, a reliance on complex II-mediated electron flow allows cells from mitochondrial disease patients devoid of a functional complex I to maintain a mitochondrial membrane potential that conveys both a mitochondrial structure and the ability to sequester agonist-induced calcium similar to that of normal cells. This finding is important as it sets the stage for complex II functional preservation as an attractive therapy to maintain mitochondrial function during hypoxia.

Mitochondria are multifunctional organelles involved in calcium buffering (1)(2)(3), apoptosis (4 -9), necrosis (10,11), reactive oxygen species production (12)(13)(14)(15) and nuclear stress signaling (16 -18). However, despite their importance in a number of cellular processes, the primary function of mitochondria remains that of cellular energy production. Although limited energy can be derived from cytosolic enzyme systems, the vast majority of cellular ATP generation is generated by the mitochondrial electrochemical gradient and ATP synthase complex. Structurally, the inner mitochondrial membrane (IMM) 3 and the outer mitochondrial membrane divide the organelle into two well defined compartments, the matrix and the intermembrane space, respectively. Protein complexes within the highly selective IMM facilitate energetically favorable electron transfer from metabolic substrates to the terminal acceptor oxygen. These protein complexes utilize the energy derived from this electron transfer to pump protons across the IMM into the intermembrane space, effectively establishing a proton motive electrochemical gradient, known as the mitochondrial membrane potential (⌬⌿ m ). Controlled proton flux across the IMM through the F 0 F 1 -ATP synthase molecular motor then drives the generation of ATP, a process described as the chemiosmotic theory (19).
Although the chemiosmotic theory of mitochondrial energy production is widely accepted, the basic mechanism in which the mitochondria establish the ⌬⌿ m remains poorly understood. Recent evidence has suggested that stable mitochondrial supercomplexes exist in a number of organisms (20,21), including a large assembly composed of complexes I, III, and IV in bovine heart mitochondria (22). Experimentally, inhibition of complex I or complex IV leads to dissipation of the mitochondrial membrane potential and serves as an effective means to mimic the effect of hypoxia on mitochondrial function (23,24). In contrast to the organized electron flow by respiratory supercomplexes, mitochondrial complex II (succinate dehydrogenase) can function as an independent enzyme whose activity is limited only by substrate availability (25). However, inhibition of mitochondrial complex II also leads to mitochondrial depolarization (26,27) and mimics hypoxia in cells (28). As a result, the discrete roles of complex I and II in the establishment and maintenance of the ⌬⌿ m both under normoxic and hypoxic conditions are unclear. * This work was supported by American Heart Association Grant 0530087N, Here, we demonstrate that mitochondrial respiratory complex II is more efficient than complex I in establishing and maintaining ⌬⌿ m under chemical poisoning and hypoxic conditions. In support, complex II-mediated metabolism can generate a functional ⌬⌿ m in complex I-compromised mitochondrial disease patients at a decreased respiratory rate, and acute delivery of the complex II substrate succinate can maintain ⌬⌿ m and ATP levels during hypoxic conditions. This finding suggests that complex II-driven electron flow is the principle mechanism whereby the ⌬⌿ m is established during hypoxic conditions.

EXPERIMENTAL PROCEDURES
Cell Isolation and Cell Culture-Rat pulmonary microvascular endothelial cells (RPMVECs) were cultured in DMEM supplemented with 10% FCS, nonessential amino acids and antibiotics, and endothelial cell growth supplement. Murine pulmonary microvascular endothelial cells (MPMVECs) were isolated from lungs as previously described (29) and were cultured in media identical to that of RPMVECs. Human pulmonary microvascular ECs were cultured in M199 supplemented with 15% FCS, L-glutamine, endothelial cell growth supplement, and antibiotics. Murine embryonic fibroblasts from both wild-type and bax Ϫ/Ϫ bak Ϫ/Ϫ knock-out lines were a generous gift of Dr. Craig Thompson (Abramson Family Cancer Center, University of Pennsylvania, Philadelphia, PA) and were cultured in DMEM supplemented with 10% FCS and antibiotics. C2C12 were obtained from ATCC and cultured in high glucose DMEM supplemented with 10% FCS and antibiotics. S2 cells were cultured in Schneider's Drosophila medium supplemented with 10% FBS, L-glutamine, and antibiotics. Human control (CF9), Leber Optic Atrophy (Leber hereditary optic neuropathy; LHON) and Kearns-Sayre Syndrome (KSS) untransformed fibroblasts were obtained from the Coriell Institute for Medical Research and cultured in DMEM supplemented with 15% FBS, L-glutamine, and antibiotics. Primary murine cardiomyocytes were isolated according to the method of Mitra and Morad (30) from freshly harvested hearts. Briefly, excised hearts were cannulated by the aorta and retrograde perfused with Ca 2ϩ -free Tyrode's solution plus collagenase B and D. Once the heart was flaccid and pale, they were removed from the perfusion system, and the ventricles were teased apart with forceps and gently titurated to dissociate individual myocytes. The myocytes were then strained through a 70-m mesh, allowed to settle, and re-equilibrated through increasing [Ca 2ϩ ] in Tyrode's solution up to 1.2 mM.
Mitochondrial Membrane Potential (⌬⌿ m ) Measurement-Prior to permeabilization, cells were washed in an extracellularlike Ca 2ϩ -free buffer (120 mM NaCl, 5 mM KCl, 1 mM KH 2 PO 4 , 0.2 mM MgCl 2 , 0.1 mM EGTA, and 20 mM HEPES-NaOH, pH 7.4) and stored on ice for at least 10 min. Following centrifugation, cells were transferred to an intracellular-like medium (permeabilization buffer; 120 mM KCl, 10 mM NaCl, 1 mM KH 2 PO 4 , 20 mM HEPES-Tris, pH 7.2, protease inhibitors (EDTA-free cOmplete TM tablets, Roche Applied Science), and 2 M thapsigargin). Digitonin was added to the experimental buffer to a concentration of 40 g/ml (80 g/ml digitonin was used for S2 cells and primary cardiomyocytes). In some experiments, the medium also was supplemented with either succinate (2 mM) or malate/glutamate (1 mM each) or an ATPregenerating system composed of 100 mM ATP, 5 mM phosphocreatine, 5 units/ml creatine kinase. Suspensions of cells (6 -10 ϫ 10 6 cells for RPMVECs, MPMVECs, human PMVECs, murine embryonic fibroblasts, and C2C12; 20 ϫ 10 6 for S2; 5 ϫ 10 5 for primary cardiomyocytes), equivalent to ϳ2 mg of protein, were placed in fluorimeter cuvettes and permeabilized by gentle stirring. Following 20 s (3 min for cardiomyocytes), the fluorescent dye JC-1 was added at a concentration of 800 nM for RPMVECs, MPMVECs, human PMVECs, murine embryonic fibroblasts, and C2C12 cells. S2 and cardiomyocytes required 1.2 mM JC1. Fluorescence was monitored in a temperaturecontrolled (37°C) multiwavelength-excitation dual wavelengthemission spectrofluorometer (Delta RAM, Photon Technology International, Birmingham, NJ) using 490-nm excitation/535-nm emission for the monomeric form and 570-nm excitation/595-nm emission for the J-aggregate of JC1. Recombinant mouse truncated Bid (tBid) (caspase-8-cleaved) was obtained from R&D Systems (Minneapolis, MN). ⌬⌿ m is shown as a ratio of the fluorescence of J-aggregate (aqueous phase) and monomer (membranebound) forms of JC1. Additions are noted in the figures. Tracings are indicative of the JC-1 ratio of at least three independent experiments. Hypoxia Experiments-Experimental buffers and media were made as described previously (31). Briefly, medium was saturated with a mixture of 95% N 2 , 5% CO 2 for 2 h. Cell medium was replaced with glucose-containing, serum-free hypoxic medium and the indicated mitochondrial substrates. The cationic potentiometric fluorescent dye tetramethylrhodamine ethyl ester perchlorate (TMRE; 100 nM) was added to the hypoxic medium after 4.5 h and allowed to incubate for a further 30 min. After dye loading, cells were imaged using the Bio-Rad Radiance 2000 imaging system (Carl Zeiss MicroImaging, Inc., Thornwood, NY) equipped with a krypton/argon ion laser source with excitation at 568 nm using a Nikon TE3000 inverted microscope with a 60ϫ oil objective. Upon changes in ⌬⌿ m , TMRE dissociates from the mitochondria. TMRE fluorescent changes were determined by perinuclear masking of all cells in the field and corrected to background nuclear fluorescence. Results indicate mean fluorescence of three to four independent experiments. ATP levels were assessed by ApoGlow assay kit (Cambrex Bioscience Rockland, East Rutherford, NJ) according to the manufacturer's protocol. Data are represented as mean Ϯ S.E. of three independent experiments.
Western Blotting-Cytochrome c release was determined via methods used to determine ⌬⌿ m . Briefly, C2C12 cells were permeabilized by digitonin (40 g/ml). Following additions, samples were collected and centrifuged, and the supernatant (cytosolic fraction) was collected. Cell pellets were resuspended in radioimmune precipitation assay buffer (Upstate Biotechnology, Inc.) containing protease inhibitor mixture, lysed by vigorous vortexing, and respun. The supernatant (membrane fraction) was collected and gel-loading buffer was added to both the cytosolic and membrane fractions. Similarly, C2C12 cells were lysed for determination of hypoxia-inducible factor 1a (HIF-1␣) stabilization. Equal amounts of protein were separated by PAGE (4 -12% BisTris for cytochrome c and 3-8% Tris acetate for HIF-1␣), transferred to a nitrocellulose membrane, and probed with anti-HIF-1␣ (Julian Lum and Celesta Simon, Cancer Biology Department, University of Pennsylvania) and anti-cytochrome c (BD Bioscience) antibodies as described previously (8).
Specific Activity of Mitochondrial Complexes-The activity of complex I was determined by monitoring the change in absorbance of cytochrome c in the presence of NADH at 550 nm (9). Cardiomyocytes were incubated in hypoxic phosphate-buffered saline as prepared under "Hypoxic Experiments." After 3 h, cells were permeabilized with 80 g/ml digitonin in intracellular-like medium, the mitochondrial pellet resuspended in glycylglycine experimental buffer, and the complex I activity was measured as cytochrome c reduction at 550 nm. Complex II activity was determined following hypoxic exposure by determining the conversion of 2-(p-iodophenyl)-3-(p-nitropenyl)-5phenyl tetrazolium to formazan. Briefly, digitonin-permeabilized cells in phosphate-buffered saline were incubated with succinate and 2-(p-iodophenyl)-3-(p-nitropenyl)-5-phenyl tetrazolium at 37°C. After 15 min, the reaction was arrested with 10% TCA and formazan extracted with ethyl acetate. The absorbance of ethyl acetate-dissolved formazan was determined in a spectrophotometer at 490 nm.

Mitochondrial Complex II Restores ⌬⌿ m in Response to Complex I and ATP Synthase Inhibition but Not Calcium, tBid, and
Alamethicin-Mitochondrial depolarization is a key event in both apoptosis and necrosis. During both apoptotic and necrotic cell death, mitochondrial swelling via opening of the permeability transition pore (PTP) (10,11), pore formation by apoptotic Bcl-2 proteins (8,32,33), and microbial agents (34) leads to membrane rupture and collapse of membrane potential. Though antiapoptotic Bcl-2 proteins and PTP blockers prevent mitochondrial permeabilization and swelling, mitochondrial depolarization still can occur under a number of cellular conditions, including hypoxia. In fact, inhibition of mitochondrial respiratory complexes and ATP synthase are commonly employed to mimic the cellular hypoxic response during normoxia (35).
To induce chemical hypoxia, permeabilized cells were challenged with the complex I inhibitor rotenone along with the F 0 F 1 -ATPase inhibitor oligomycin to eliminate electron flow originating from complex I and prevent mitochondrial ATP consumption and proton extrusion from the matrix (36). The permeabilized system is based on the isolated mitochondrial suspension and was used previously to assess mitochondrial depolarization in response to Ca 2ϩ or after activation of proapoptotic Bcl-2 proteins (8,37,38). The permeabilized system has several advantages over the mitochondrial isolation model in that mitochondria are suspended in a cytoplasmic rather than an artificial milieu that mimics the whole cell model. In addition, this model system does not require time-intensive manipulation that may alter mitochondrial function, and eliminates the need for supplementation of various exogenous molecules. Permeabilized cells were loaded with the ratiometric mitochondrial membrane potential indicator JC-1 and then treated with the rotenone/oligomycin combination to induce ⌬⌿ m loss. Rotenone/oligomycin-induced mitochondrial depolarization was not prevented by the PTP blocker cyclosporine A (CsA), but subsequent addition of the complex II substrate succinate substantially restored ⌬⌿ m (Fig. 1A). Pretreatment with CsA also did not prevent rotenone/oligomycin-induced ⌬⌿ m loss, although it significantly blunted Ca 2ϩ -dependent mitochondrial depolarization via the PTP open even in the presence of succinate (Fig. 1B), indicating that succinate cannot prevent opening of the PTP. In parallel, succinate supplementation did not prevent ⌬⌿ m loss triggered by the mitochondrial swellinginducing agent alamethicin (Fig. 1C). Next, it was of interest to determine whether rotenone/oligomycin-induced ⌬⌿ m loss is similar to mitochondrial depolarization caused by proapoptotic Bcl-2 family proteins. A logical assumption would be that the mitochondrial proton gradient can be maintained if complex III and complex IV are intact and cytochrome c is present to enable electron transfer from complex III to IV. During apoptosis, cytochrome c is released from the intermembrane space via a Bax/Bak-dependent mechanism that precedes ⌬⌿ m loss (9, 39 -41). As shown in Fig. 1D, rotenone/oligomycin-induced ⌬⌿ m loss is a Bcl-2 independent phenomenon. ⌬⌿ m loss by the proapoptotic protein tBid, an activator of Bax and Bak, is not rescued by succinate. This observation suggests that cytochrome c is essential for normal ⌬⌿ m maintenance initiated by complex II. Similar to other cell types, wild-type and bax Ϫ/Ϫ bak Ϫ/Ϫ double knock-out fibroblasts underwent succinate-reversible ⌬⌿ m loss even in the presence of CsA (Fig. 1, E and F). Our data suggests that rotenone/oligomycin-induced mitochondrial depolarization is neither an apoptotic (Bcl-2mediated) nor a necrotic (PTP opening) phenomenon, and thus should not exhibit mitochondrial cytochrome c release. Unlike tBid (100 nM), rotenone/oligomycin-induced depolarization did not trigger cytochrome c release after 20 min exposure (Fig.  1G). These results demonstrate that rotenone/oligomycin induces ⌬⌿ m loss by mechanisms not directly dependent upon Bcl-2 or PTP opening and suggests that ⌬⌿ m might be controlled during hypoxia via the availability of specific metabolic substrates.
The Complex II Pathway Preserves the Membrane Potential in the Absence of Complex I Activity-As succinate afforded protection from ⌬⌿ m loss during pseudohypoxia, we next chose to investigate the discrete role of complex II in ⌬⌿ m maintenance. To assess the contribution of individual mitochondrial complexes to the ⌬⌿ m , permeabilized RPMVECs were supplemented with complex-specific mitochondrial sub-strates either alone or in combination with mitochondrial complex and F 0 F 1 -ATPase inhibitors. Following permeabilization, cells exhibit ⌬⌿ m loss due to lack of available substrates as the cytoplasmic contents diffuse and dilute into the experimental buffer (data not shown). Surprisingly, the complex II substrate succinate, but not the complex I substrates malate and glutamate (malate/glutamate), effectively maintained ⌬⌿ m ( Fig. 2A). Succinate also prevented ⌬⌿ m loss even in the presence of the F 0 F 1 -ATPase inhibitor oligomycin, which expectedly blocked the protective effect of ATP on ⌬⌿ m maintenance (Fig. 2B). Furthermore, the effect of succinate on ⌬⌿ m maintenance was not affected by the complex I inhibitor rotenone (Fig. 2C), excluding the possibility that succinate initiates proton pumping via the reverse flow of electrons through complex I (42), and was completely abolished by the nonmetabolizable complex II competitive inhibitor malonate (Mao) (Fig. 2D). However, succinate did not offer protection from complex III-mediated ⌬⌿ m loss in the presence of the inhibitors myxothiazol and antimycin A (Fig. 2, E and F). In particular, antimycin A induced immediate ⌬⌿ m loss even in the combined presence of succinate and ATP, which hyperpolarizes the membrane (supplemental Fig. S1A), suggesting that complex II requires the proton pumping capability of mitochondrial complex III to establish ⌬⌿ m . This initial observation raises the possibility that complex II pathway could establish ⌬⌿ m in the absence of complex I.
To further characterize the role of complex II in the ⌬⌿ m , we next assessed whether reintroduction of the complex II substrate succinate could restore ⌬⌿ m after complete dissipation of the proton gradient (⌬⌿ m loss). Following permeabilization-FIGURE 1. Complex II cannot restore mitochondrial function following irreversible ⌬⌿ m loss. A, complex I/F 0 F 1 -ATPase inhibition (rotenone/oligomycin) triggered ⌬⌿ m that was reversed by succinate (Succ; 10 mM) but not the PTP blocker CsA (5 M). B, in contrast, Ca 2ϩ -induced mitochondrial depolarization (200 M) is blocked by CsA but not succinate. C, succinate-dependent ⌬⌿ m maintenance is abolished by mitochondrial swelling triggered by the microbial antibiotic alamethacin (20 g/ml). Note: B and C were performed in the presence of 2 mM succinate. D, succinate did not restore ⌬⌿ m in response to Bcl-2 family protein tBid-induced (10 g/ml) depolarization. succinate restored ⌬⌿ m in both (E) wild type and (F) bax Ϫ/Ϫ bak Ϫ/Ϫ double-deficient murine embryonic fibroblasts ollowing ⌬⌿ m loss. G, mitochondrial (Mito) cytochrome c release from the mitochondria in response to Ca 2ϩ , tBid, and the inhibitor combination of rotenone/oligomycin; Rot, rotenone; WT, wild-type.
induced ⌬⌿ m loss, cells were supplemented with either a combination of complex I substrates (malate/glutamate or malate/ pyruvate) or succinate. In contrast to malate/glutamate or malate/pyruvate, addition of succinate resulted in dramatic ⌬⌿ m restoration (Fig. 3A). Because complex I substrates were ineffectual at maintaining ⌬⌿ m , we developed a rapid chemical means to further test the efficacy of complex II in mitochondrial function and mimic hypoxia in normoxic cells. Cells were supplemented with malate/glutamate, and then ⌬⌿ m loss was induced by a combination of rotenone/oligomycin, completely eliminating the contribution of complex I (via secondary ATP production as in Fig. 2A) as well as intermembrane space proton loss through F 0 F 1 -ATPase. Cells were then treated with the complex II substrate succinate and extramitochondrial ATP alone or in combination. This approach (succinate versus ATP) allowed us to distinguish between proton gradient maintenance by succinate or glycolytic ATP and to directly monitor ⌬⌿ m recovery. The chemical combination of rotenone/oligomycin triggered rapid ⌬⌿ m loss, which could be reversed by supplementation with succinate in several cell types that we examined (mouse, rat, and human pulmonary microvascular endothelial cells (MPMVEC, RPMVEC, and human PMVEC, respectively), C2C12 myoblast, and S2 Drosophila cells) (Fig. 3, B, C, and F and supplemental Fig. S1C and supplemental Table 1) in a dose-dependent fashion (supplemental Fig. S2). As in Fig. 1, complex III inhibition resulted in ⌬⌿ m loss that could not be rescued by the proton-pumping capabilities of complex I alone (supplemental Fig. S1B). Whereas ATP in combination with succinate produced an additive effect on ⌬⌿ m , ATP alone did not reverse ⌬⌿ m loss (Fig. 3B). The fatty acid ␤-oxidation end-product acetyl CoA is a major fuel in some tissues and ultimately enters the citric acid cycle where it will increase all TCA cycle metabolites, including succinate and malate. However, rotenone/oligomycin-induced ⌬⌿ m loss was not affected by acetyl CoA supplementation (Fig. 3C). Addition of the citric acid cycle metabolite ␣-ketoglutarate also did not restore ⌬⌿ m , despite being a precursor of succinate in the TCA cycle (Fig. 3C). In addition to rotenone/oligomycin, ⌬⌿ m loss also occurred following addition of the complex II competitive inhibitor malonate (Mao). Subsequent addition of the complex I substrates malate/pyruvate and their essential cofactor NAD ϩ did not restore ⌬⌿ m even at a concentration of 10 mM (Fig. 3D), indicating a rapid effect of succinate in the establishment of the ⌬⌿ m .
Although our data demonstrate the restoration of ⌬⌿ m by succinate following ⌬⌿ m loss, there is a possibility that restoration of ⌬⌿ m by succinate is actually due to either a difference in substrate delivery or to the secondary generation of NADH from mitochondria. To address this issue, we simultaneously measured both ⌬⌿ m and NADH generation in permeabilized cells. Inhibition of complex I by rotenone rapidly increased cellular NADH levels but did not affect ⌬⌿ m as detected by TMRE fluorescence. However, subsequent addition of succinate rapidly increased ⌬⌿ m (as indicated by a decrease in TMRE as more dye sequesters into the mitochondria) without an appreciable effect on NADH generation (Fig. 3E). Conversely, malate/glutamate did not increase ⌬⌿ m when complex II was inhibited by Mao despite a robust increase in cellular NADH levels. The rapid increase in NADH levels following addition of complex I substrates to permeabilized cells indicates sufficient delivery to the mitochondria, excluding the possibility that are findings are due to differences in mitochondrial loading of complex I versus II substrates. Because succinate-mediated electron transfer relies on complex III (Fig. 2, E and F), we next assessed whether bypass of complex II/III by delivery of the complex IV substrates tetramethyl-p-phenylenediamine (TMPD) and ascorbate could preserve ⌬⌿ m in response to a combined complex I/F 0 F 1 -ATPase inhibition (rotenone/oligomycin). TMPD/ascorbate supplementation only partially restored ⌬⌿ m (Fig. 3F). However, unlike the dose-dependent response seen with succinate (supplemental Fig. S2), addition of a higher concentration of TMPD/ascorbate was toxic to mitochondria and ⌬⌿ m maintenance.
Establishment of ⌬⌿ m in Human Fibroblasts from Patients with Mitochondrial Disorders-Mitochondria are the only organelles outside the nucleus that contain DNA (mtDNA) and the machinery for synthesizing RNA and proteins. Mitochondrial complexes I, III, and IV and F 0 F 1 -ATPase are multisubunit protein complexes derived from both the nuclear and mitochondrial genome. In contrast, complex II polypeptides are derived solely from nuclear DNA (43). Many maternally inherited mitochondrial diseases arise from mutations in mitochondrial complexes I, III, and IV and F 0 F 1 -ATPase, including KSS, which results from a 1.9-kb mtDNA deletion spanning both complex I (ND1, ND2, ND4, NDL, and ND6 gene products) and F 0 F 1 -ATPase (ATP6 and ATP8 genes) (44) and LHON, which affects subunit 4 of complex I (43). Primary fibroblasts from these patients therefore constitute an ideal model to identify the unique role of complex II in ⌬⌿ m maintenance. Cells derived from LHON (Fig. 2B) and KSS (Fig. 2C) patients exhibited a similar ⌬⌿ m response to rotenone/oligomycin as control fibroblasts ( Fig. 2A) that was reversed by succinate. Intact CF9, LHON, and KSS cells also exhibited a similar phenotype with a predominance of filamentous mitochondria (Fig. 4, D, E, and F)  and displayed a similar ⌬⌿ m that was established principally by mitochondrial complex II (Fig. 4, G-J). Because Mao addition resulted in ⌬⌿ m loss in all three cell types, this evidence also suggests that LHON and KSS cells do not utilize a compensatory mechanism to generate basal ⌬⌿ m , but rather, all three cells establish ⌬⌿ m predominately through a mitochondrial complex II-dependent pathway. An intact membrane potential generated by complex II also allows for LHON and KSS cells to sequester cytosolic calcium in an identical pattern to that of CF9 control cells (Fig. 4K), indicating that ⌬⌿ m -dependent mitochondrial function is largely intact and dependent upon establishment of the IMM proton gradient by complex II. However, despite the ability to generate an intact membrane potential, complex I deficiencies lead to a compensatory glycolytic shift (45) and a decrease in ATP generation (46). Accordingly, LHON cells consumed significantly less oxygen than both KSS and CF9 cells (Fig. 4L). The fact that the LHON phenotype exhibited less oxygen utilization than KSS cells was surprising considering only complex I was affected and led us to consider the possibility that establishment of the ⌬⌿ m would be possible by diverting cells into succinate-driven metabolism during reduced oxygen availability.
Complex II Substrate Supplementation Restores Mitochondrial Function during Hypoxia-Our results demonstrate that complex IIdependent ⌬⌿ m maintenance is dependent upon electron flow through mitochondrial complexes II-IV in the permeabilized system. Furthermore, our studies with mitochondrial disease patients demonstrate that electron flow initiated by complex II can establish an intact ⌬⌿ m even when complex I is nonfunctional. However, once electrons pass through the mitochondrial respiratory chain to complex IV, molecular oxygen is required as a terminal acceptor. Because LHON cells can generate a ⌬⌿ m using less oxygen, we next focused on the idea that under basal conditions, complex II-driven electron flow may predominate as it consumes less oxygen than electron flow originating at complex I. Sequential addition of specific mitochondrial complex substrates (10 mM total substrate for each addition) and their respective inhibitors revealed that succinate reestablished a higher ⌬⌿ m than either complex I or complex IV (Fig. 5A). However, whereas ⌬⌿ m restoration was higher following supplementation of succinate, complex II-specific oxygen consumption was less than that of both complex I and complex IV (Fig. 5B). These data confirm that complex II-driven electron flow requires less oxygen than electron flow originating from complex I and led us to hypothesize that ⌬⌿ m could be maintained during hypoxic conditions via succinate metabolism.   (n ϭ 4). B, oxygen measured via a Clark-type electrode following successive additions as indicated in permeabilized cells. Rate of oxygen consumption for complex II was less than both complexes I and IV. Rate of oxygen consumption was measured and is displayed as nmol oxygen/min/6 ϫ 10 6 cells (mean Ϯ S.E.; n ϭ 12). The trace is a single experiment that is representative of mean oxygen consumption. Rot, rotenone.
To assess establishment of ⌬⌿ m during hypoxic conditions, we chose to utilize freshly isolated murine cardiomyocytes (supplemental Fig. S3) as these cells are a primary target of hypoxic damage in humans during myocardial infarction. Chemical inhibition of complexes I and IV (rotenone/oligomycin) induced rapid ⌬⌿ m loss in cardiomyocytes that was similar to other cell types (Figs. 1, 2, and 3) that could be reversed by succinate supplementation (Fig. 6A). To directly assess the relationship between ⌬⌿ m loss and hypoxia, ⌬⌿ m was assessed following cell permeabilization in hypoxic experimental buffer. Marked ⌬⌿ m loss was observed when cells were permeabilized in this buffer (Fig.  6B), the hypoxic status of which was verified by probing for HIF-1␣ stabilization after 5 h (Fig. 6C). Succinate but not malate/pyruvate supplementation restored ⌬⌿ m in response to hypoxia (Fig. 6B). Although it is possible that the nominal effect of complex I substrates on ⌬⌿ m maintenance during hypoxia could be due the inactivation or enhancement of specific mitochondrial complexes, no differences in the activity of either complex I or complex II were noted (Fig.  6D). Similar to the permeabilized cell model, intact C2C12 and freshly isolated cardiomyocytes subjected to hypoxic conditions (Fig. 6E) exhibited a significant reduction in ⌬⌿ m that was effectively restored by supplementation with complex II but not complex I substrates in the hypoxic medium (Fig. 6, E-H). Active mitochondrial complexes support the generation of an electrochemical gradient across the IMM, which is vital for the efficient production of ATP (33). We therefore analyzed the role of complex II-dependent ⌬⌿ m restoration in the maintenance of ATP levels under hypoxia. Although less than that during normoxia, succinate supplementation resulted in significant ATP level maintenance, consistent with the restoration of ⌬⌿ m by succinate in response to hypoxia (Fig. 6I). loss is reversed by succinate (Succ; 10 mM) but not malate/pyruvate (5 mM/5 mM) in freshly isolated permeabilized, mature murine cardiomyocytes and C2C12 cells, respectively. C, hypoxic status was confirmed by HIF-1␣ stabilization in C2C12 cells following 5 h of hypoxia (NS; nonspecific). D, primary adult cardiomyocyte mitochondrial complex I and complex II activity were unaltered during hypoxia. E, intact C2C12 or freshly isolated (F) adult cardiomyocytes were subjected to hypoxic conditions for 5 h, and cells were loaded with the ⌬⌿ m indicator TMRE. Normoxic and succinate-supplemented hypoxic cells retained ⌬⌿ m. In contrast, hypoxia alone or supplemented with malate/pyruvate (Mal/Pyr) caused mitochondrial depolarization. F and G, average fluorescence intensity of mitochondrial TMRE retention as described in F and G following normoxic and hypoxic conditions. H, ATP level maintenance by succinate supplementation in response to hypoxia. All values in D, G, H, and I indicate mean Ϯ S.E. (n ϭ 3). Rot, rotenone. f.a.u., fluorescence arbitrary units.

DISCUSSION
Mitochondrial depolarization via Bcl-2-mediated membrane permeabilization, activation of the PTP complex, and toxininduced mitochondrial pore formation are features of not only apoptotic but also necrotic cell death (33) and are intimately involved in organ dysfunction during ischemia/reperfusion injury (47). Subsequently, antiapoptotic Bcl-2 family proteins and the PTP component cyclophilin D tightly regulate mitochondrial permeabilization and preserve the ⌬⌿ m . Although apoptotic and necrotic mitochondrial depolarization has been rescued by synthetic small molecules such as cyclosporine A, ⌬⌿ m loss following mitochondrial substrate deprivation was not preventable (Fig. 1). Therefore, our present investigation demonstrates another mode of mitochondrial depolarization that is independent of outer mitochondrial membrane permeabilization and PTP opening. It is clear from Fig. 1 that rotenone plus oligomycin causes mitochondrial depolarization that is prevented by the complex II substrate succinate. This mitochondrial depolarization does not require complex I activity, because it is abolished when succinate is present (Fig. 2). ⌬⌿ m preservation is dependent on complex II, III, and IV, because it is eliminated by inhibitors of both complex II and III (Fig. 2). ⌬⌿ m is also partially preserved when complex IV substrate is present, which directly donates protons. Rotenone/oligomycin-induced mitochondrial depolarization is furthermore unrelated to outer mitochondrial membrane permeabilization or PTP opening, because cytochrome c is retained in the mitochondria even after mitochondrial depolarization (Fig. 1G).
In an attempt to decipher the differential roles of complex I and II in mitochondrial function during hypoxia, we chose to optimize our permeabilized cell experiments. Surprisingly, we found that ⌬⌿ m could be maintained using complex II rather than complex I substrates and that blockade of complex II would lead to membrane depolarization. This observation intrigued us and led to the development of additional studies that supported the idea that basal ⌬⌿ m was a complex II-driven process. Indeed, supplementation of mitochondrial membrane-permeable NAD ϩ or an increase in the cellular availability of NADH did not result in repolarization of the mitochondrial membrane following complex II inhibition (Fig. 3, D and  E), suggesting that electron flow initiating at complex II can more readily polarize the mitochondrial membrane than electron transfer originating from complex I. As complex I substrates can partially restore membrane potential when all mitochondrial respiratory chain enzymes are functioning properly (Fig. 3A), our findings suggest that complex I may contribute to overall ⌬⌿ m. However, complex I substrates malate/glutamate did not retain ⌬⌿ m when F 0 F 1 -ATPase was inhibited by oligomycin (Fig. 1B), making it unlikely that our results are simply due to a decreased ⌬⌿ m during ATP production. Furthermore, proton pumping by complex I alone is not sufficient to reestablish the ⌬⌿ m when the respiratory chain is stalled downstream at complex III (supplemental Fig. S1B). succinate was effective in establishing and maintaining ⌬⌿ m in fibroblasts from patients with dysfunctional complex I and IV, indicating that a polarized mitochondrial membrane (Fig. 4H) can be preserved even when ATP production is compromised (48).
Mammalian succinate dehydrogenase is an enzyme complex composed of four nuclear-encoded subunits: two membranespanning anchoring proteins and two matrix subunits that constitute the catalytic subunit of the enzyme complex (49). Functionally, oxidation of succinate to fumarate allows for the entry of electrons into the mitochondrial respiratory chain through coenzyme Q. However, unlike complexes I, III, and IV, succinate dehydrogenase does not directly couple electron transfer with proton pumping into the intermembrane space. Rather, complex II utilizes the labile properties of coenzyme Q within the IMM to transfer electrons to complex III, which can then act as a proton pump to contribute to the ⌬⌿ m. Therefore, it is interesting that our studies uncovered a greater reliance on complex II to generate a membrane potential, as it does not directly contribute to membrane potential as a proton pump. The solid-state model proposes that ordered electron flow occurs through static enzyme complexes (50) and that the interaction between complex I and III allows for the efficient transfer of electrons through coenzyme Q, a process known as channeling (25). In support, isolated respiratory supercomplexes composed of complex I, III, and IV can form functional respiratory units (21), indicating that coenzyme Q is fixed within the supercomplex. In contrast to the Solid-State Mode, the Random Collision Model describes electron transfer facilitated by the interaction between small diffusible molecules (cytochrome c and coenzyme Q) and independent enzymes (51). As the majority of coenzyme Q is not bound to complex I but rather is found free in the membrane bilayer (52) where it may not participate in supercomplex-mediated electron flow, the IMM coenzyme Q pool has a much greater capacity to store electrons under succinate rather than an NADH-dependent metabolism. Our findings suggest that increasing the cellular availability of succinate may essentially saturate the inner mitochondrial membrane with partially reduced (ubisemiquinone) or fully reduced (ubiquinol) coenzyme Q. Although it is possible that succinate-mediated reduction of the coenzyme Q pool may displace oxidized coenzyme Q (ubiquinone) within the complex I-III supercomplex, our studies conducted under either chemical (rotenone) or genetic (LHON) complex I inhibition make it unlikely that succinate-mediated metabolism would alter ubiquinone channeling between complex I and III via reverse electron flow. Rather, as complex III and IV can exist either as a supercomplex with complex I or as separate units (21), our results suggest that complex II-mediated saturation of the mitochondrial inner membrane with ubisemiquinone or ubiquinol establishes the ⌬⌿ m through the labile movement of coenzyme Q to complex III and IV. This work therefore supports the hypothesis that basal ⌬⌿ m is established primarily through the random collision model as opposed to the solidstate model under both normal and reduced oxygen tensions. However, the presence of stable mitochondrial supercomplexes indicates that solid-state model electron flow is important for mitochondrial function. We postulate that it is the establishment of the ⌬⌿ m by complex II that allows the efficient electron transport between complex I and III and enhances ATP generation.
An important finding of our work was that supplementation of the complex II substrate succinate could maintain ⌬⌿ m dur-ing hypoxia. During normoxic glucose metabolism, pyruvate enters the mitochondria and generates ATP through the TCA cycle with molecular oxygen as the terminal electron acceptor. As oxygen tension decreases, pyruvate is converted to lactate rather than enter the TCA cycle, reducing mitochondrial substrate availability and the ability to generate ⌬⌿ m (53). However, we found that supplementing cells with exogenous succinate could restore ⌬⌿ m at low oxygen tensions. Lacking an innate proton pumping capacity, complex II requires complex III to transfer electrons and initiate establishment of the ⌬⌿ m , and previous studies have described intact complex II/III function during hypoxia (54). Indeed, the ability to continue mitochondrial enzymatic activity during hypoxia is a hallmark of cancer cell metabolism (55), and alterations in succinate dehydrognase have been implicated in certain forms of cancer (56). Our findings demonstrate that, unlike the aberrant metabolism observed during cancer, acute succinate metabolism in normal cells can maintain a ⌬⌿ m and acute ATP production during hypoxia.
In summary, our study demonstrates that mitochondrial depolarization either by inhibition of mitochondrial complex I and ATP synthase or during hypoxia was prevented by the complex II substrate succinate. However, irreversible ⌬⌿ m loss through PTP opening or activation of pro-apoptotic Bcl-2 family proteins could not be prevented by succinate. This indicates an important role for establishing the ⌬⌿ m via electron flow initiated by complex II. Strategies employing complex II-dependent preservation of mitochondrial function during hypoxia might therefore constitute a promising tool for the treatment of ischemia-associated organ dysfunction.