Protective Role of Transient Pore Openings in Calcium Handling by Cardiac Mitochondria

Abstract

Long-lasting mitochondrial permeability transition pore (mPTP) openings damage mitochondria, but transient mPTP openings protect against chronic cardiac stress. To probe the mechanism, we subjected isolated cardiac mitochondria to gradual Ca²⁺ loading, which, in the absence of BSA, induced long-lasting mPTP opening, causing matrix depolarization. However, with BSA present to mimic cytoplasmic fatty acid-binding proteins, the mitochondrial population remained polarized and functional, even after matrix Ca²⁺ release caused an extramitochondrial free [Ca²⁺] increase to >10 μm, unless mPTP openings were inhibited. These findings could be explained by asynchronous transient mPTP openings allowing individual mitochondria to depolarize long enough to flush accumulated matrix Ca²⁺ and then to repolarize rapidly after pore closure. Because subsequent matrix Ca²⁺ reuptake via the Ca²⁺ uniporter is estimated to be >100-fold slower than matrix Ca²⁺ release via mPTP, only a tiny fraction of mitochondria (<1%) are depolarized at any given time. Our results show that transient mPTP openings allow cardiac mitochondria to defend themselves collectively against elevated cytoplasmic Ca²⁺ levels as long as respiratory chain activity is able to balance proton influx with proton pumping. We found that transient mPTP openings also stimulated reactive oxygen species production, which may engage reactive oxygen species-dependent cardioprotective signaling.

Introduction

A key mechanism involved in cardiac injury is the mitochondrial permeability transition, due to the opening of mitochondrial permeability transition pores (mPTP)³ in the inner mitochondrial membrane (IMM). The permeability transition has been shown to occur upon reperfusion of the ischemic heart (1), and its prevention is thought to be an important component of ischemic and pharmacologic pre-conditioning and post-conditioning (PC), the most powerful forms of cardioprotection known (2–6). Supporting this idea, mPTP inhibitors such as cyclosporine A (CsA) reduce ischemia/reperfusion (I/R) injury (7), and genetic ablation of a critical mPTP component, cyclophilin D (CyPD), confers chronic protection against I/R injury, which is not further enhanced by CsA or ischemic PC (8).

Although long-lasting mPTP opening inevitably causes mitochondrial injury, evidence has been presented that mPTP can also open transiently, which may have beneficial effects, such as releasing accumulated Ca²⁺ from the matrix (9–11), ultimately delaying the onset of long-lasting, large conductance mPTP opening. Two recent findings provide strong support for a protective role of transient mPTP opening. First, CyPD knock-out (KO) mice, which are chronically protected against I/R injury (8), develop heart failure more rapidly in response to transaortic constriction or cross-breeding with some cardiomyopathic strains (12) but not all (13). These mice had elevated mitochondrial Ca²⁺ content, suggesting a defect in mitochondrial Ca²⁺ handling related to loss of mPTP function, with associated metabolic abnormalities. Second, whereas CsA given during prolonged ischemia reduces I/R injury (7), paradoxically, CsA (or sanglifehrin A) administered during ischemic or pharmacologic PC episodes blocks cardioprotection (6, 14, 15). This finding is reminiscent of reactive oxygen species (ROS), in which ROS scavengers given during prolonged ischemia reduce I/R injury, but ROS scavengers given during ischemic or pharmacologic PC episodes block cardioprotection (16). Thus, a picture is emerging that, similar to ROS, mPTP opening is a two-edged sword, with both protective and deleterious actions.

In this study, we used isolated cardiac mitochondria to explore the protective role of mPTP in mitochondrial Ca²⁺ handling. We present evidence that cardiac mitochondria can tolerate markedly elevated levels of extramitochondrial free [Ca²⁺] without collectively dissipating the average mitochondrial membrane potential (ΔΨ_m) as long as IMM leak is kept low by BSA to mimic fatty acid-binding proteins normally present in the cytoplasm. The mechanism involves mitochondria utilizing asynchronous transient mPTP openings to depolarize long enough to flush accumulated matrix Ca²⁺ without losing essential matrix metabolites such as NADH, after which they rapidly repolarize. If the subsequent rate of matrix Ca²⁺ reuptake via the Ca²⁺ uniporter is slow compared with the Ca²⁺-flushing rate, then only a small fraction (<1%) of the mitochondria are depolarized at any given point in time. We also found that when transient mPTP openings are induced by Ca²⁺ loading, mitochondrial ROS production increases markedly, which may provide a link between loss of mPTP function and loss of cardioprotection when hearts are exposed to mPTP blockers during PC episodes (6, 14, 15).

EXPERIMENTAL PROCEDURES

Mitochondrial Isolation

Mitochondria were isolated from adult 2–3-kg New Zealand White rabbit hearts or wild-type and genetically altered mice by homogenization and differential centrifugation as described previously (17) and were resuspended in EGTA-free homogenization buffer (250 mm sucrose, 10 mm HEPES (pH 7.4) with Tris base) to yield 30–50 mg of mitochondrial protein/ml. Mitochondria were incubated on ice and used within 5 h after isolation.

Isolated Mitochondrial Studies

All measurements were carried out using a customized Ocean Optics fiber optic spectrofluorometer in a continuously stirred cuvette, open to atmospheric air to permit O₂ diffusion into the buffer, and maintained at room temperature (22–24 °C). Mitochondria (0.3–0.6 mg/ml) were added to the cuvette containing standard buffer consisting of 120 mm KCl, 10 mm HEPES (pH 7.4) with Tris base. In some experiments, KCl was replaced with 250 mm sucrose. The partial pressure of oxygen (pO₂) in the buffer was continuously recorded via an Ocean Optics FOXY-AL300 fiber optic oxygen sensor inserted through the same hole. Under these conditions, buffer [O₂] depends on O₂ consumption by mitochondria relative to the rate of O₂ diffusion from air into the buffer. When consumption is balanced with diffusion, the O₂ level remains stable. The effect of stirring speed on the rate of O₂ diffusion into the buffer and how it is balanced with O₂ consumption and O₂ sensor calibration have been described in detail previously (18). Membrane potential (ΔΨ_m) was recorded using tetramethylrhodamine methyl ester (TMRM; 400 nm, excitation and emission wavelengths of 545 and 580 nm, respectively) or alternatively with a tetraphenylphosphonium electrode (World Precision Instruments). Ca²⁺ uptake and efflux were recorded with a Ca²⁺ electrode (World Precision Instruments) or alternatively with low affinity Ca²⁺ dye (Calcium Green-5N). H₂O₂ production was measured using Amplex Red (10 μm) and horseradish peroxidase (0.2 units) at excitation and emission wavelengths of 545 and 590 nm, respectively. Matrix swelling/shrinkage was measured by light scattering at 540 nm, and protein content was determine using the Lowry method. Pyruvate, malate, and glutamate were added as free acids buffered with Tris (pH 7.4). BSA was added at a final concentration of 1 or 0.5 mg/ml, and added P_i was 2.5 mm unless indicated otherwise.

RESULTS

Ca²⁺ Loading Induces mPTP-mediated Matrix Ca²⁺ Release without Collective ΔΨ_m Dissipation if IMM Leak Is Low

ΔΨ_m depolarization induced by Ca²⁺ loading is an established method for testing the susceptibility of isolated mitochondria to mPTP opening. Consistent with this classic assay, we found that when we subjected isolated energized cardiac mitochondria to successive Ca²⁺ pulses, ΔΨ_m depolarization and matrix Ca²⁺ release occurred simultaneously (Fig. 1A). Note that O₂ consumption became severely depressed during ΔΨ_m dissipation but rapidly increased with NADH addition (0.1 mm). This indicates that the IMM was permeable to NAD/NADH, indicative of long-lasting mPTP opening in the full conductance mode (M_r cutoff of ∼1500). P_i addition was also unable to promote mitochondrial Ca²⁺ re-accumulation (data not shown).

FIGURE 1.

Transient mPTP openings allow energized isolated mitochondria to tolerate elevated extramitochondrial [Ca²⁺] without collective ΔΨ_m dissipation when BSA is present. Mitochondria were added to KCl buffer (without P_i) at the beginning of the trace, followed by Complex I substrates pyruvate (P), malate (M), and glutamate (G) (3.5 mm each), while recording TMRM fluorescence (upper), extramitochondrial [Ca²⁺] ([Ca]_e; middle), and O₂ (lower). A, during seven successive 2 μm Ca²⁺ additions (between arrows), extramitochondrial [Ca²⁺] initially returned to the base-line level after each Ca²⁺ pulse until mitochondria then released matrix Ca²⁺ (dashed vertical line), which rose to >10 μm. In the absence of BSA, matrix Ca²⁺ release was accompanied by ΔΨ_m dissipation and decreased O₂ consumption due to mPTP opening in the large conductance mode. B, in the presence of BSA (1 mg/ml), however, the average ΔΨ_m was maintained with no decrease in O₂ consumption during matrix Ca²⁺ release, consistent with transient mPTP openings. Calibration was obtained using alamethicin (Ala) to dissipate ΔΨ_m fully and release all matrix Ca²⁺.

However, if BSA was included in the buffer to mimic cytoplasmic proteins that normally bind fatty acids and therefore decrease IMM proton leak associated with fatty acid cycling (19), matrix Ca²⁺ release occurred without significant dissipation in the average ΔΨ_m of the mitochondrial population (Fig. 1B). Each Ca²⁺ pulse caused a transient elevation of extramitochondrial Ca²⁺ (measured with a Ca²⁺-sensitive electrode), which was taken up by the mitochondria in association with a transient small ΔΨ_m depolarization, apparent as small oscillations in the TMRM tracing (which, for larger Ca²⁺ pulses, were correspondingly larger). However, in the presence of BSA, the average ΔΨ_m then re-equilibrated. At a critical Ca²⁺ load, mitochondria began to release their accumulated matrix Ca²⁺, but the average ΔΨ_m of the mitochondrial population as a whole remained polarized. Unlike the results shown in Fig. 1A, in which O₂ consumption decreased after matrix Ca²⁺ release, O₂ consumption remained high after matrix Ca²⁺ release in the presence of BSA (indicated by the stable non-zero level reflecting O₂ consumption in balance with O₂ diffusion into the open cuvette), consistent with adequate matrix NAD/NADH. These findings were confirmed when extramitochondrial free Ca²⁺ was monitored using Calcium Green-5N salt instead of the Ca²⁺-sensitive electrode, and ΔΨ_m was measured with a tetraphenylphosphonium electrode (supplemental Fig. S1). Fig. 2 summarizes the changes in ΔΨ_m after matrix Ca²⁺ release in the absence and presence of BSA and also shows that similar findings were obtained with exogenous P_i present, although a significantly larger total Ca²⁺ load was required to release matrix Ca²⁺ (204 ± 20 nmol/mg of protein (n = 6) with P_i and 66 ± 5 nmol/mg of protein (n = 9) without P_i; p < 0.01). The higher Ca²⁺ load is consistent with the known ability of P_i to enhance matrix Ca²⁺ buffering. Fig. 2 also shows that similar findings were obtained when mitochondria were suspended in sucrose buffer in place of KCl buffer. In this case, the Ca²⁺ loads required to induce matrix Ca²⁺ release averaged 187 ± 22 nmol/mg of protein with P_i (n = 4) and 108 ± 26 nmol/mg of protein without P_i (n = 4; p < 0.01).

FIGURE 2.

Bar graph summary of percent ΔΨ_m dissipation with or without BSA (1 mg/ml) measured at near-peak matrix Ca²⁺ release when isolated rabbit cardiac mitochondria were incubated in KCl buffer, KCl + P_i (2. 5 mm) buffer with and without FCCP (15–25 nm), and sucrose (250 mm) buffer. White and black bars indicate that BSA was absent (−) or present (+), respectively. The rightmost bars compare cardiac mitochondria isolated from WT and CyPD KO mice, respectively, incubated in KCl buffer + P_i (2.5 mm) with BSA present. Suc, sucrose buffer. *, p < 0.05.

In both the absence (Fig. 3A) and presence (Fig. 3B) of BSA, matrix Ca²⁺ release was reversed by CsA if added early, consistent with matrix Ca²⁺ release being mediated by mPTP openings. To investigate the role of mPTP further, we also studied cardiac mitochondria isolated from CyPD KO (Ppif ^−/−) mice, in which mPTP formation is defective but not abolished (8, 12). Cardiac mitochondria from wild-type mice maintained ΔΨ_m during matrix Ca²⁺ release when BSA was present (averaging 99 ± 13 nmol/mg of protein in KCl buffer with P_i, n = 6). In contrast, mitochondria from CyPD KO mitochondria tolerated higher Ca²⁺ loads before matrix Ca²⁺ release occurred (averaging 280 ± 24 nmol/mg of protein in KCl buffer with P_i, n = 4), as reported previously (8), but developed ΔΨ_m dissipation coincident with matrix Ca²⁺ release (Fig. 2). CsA did not reverse matrix Ca²⁺ release or restore ΔΨ_m. Thus, CyPD KO mitochondria were unable to maintain ΔΨ_m during matrix Ca²⁺ release in the presence of BSA, consistent with a defect in transient mPTP opening.

FIGURE 3.

Matrix Ca²⁺ release is inhibited by CsA, and transient mPTP openings occur in a low conductance mode when BSA is present. Mitochondria were added to 250 mm sucrose buffer containing 0.2 mm P_i at the beginning of the trace, followed by Complex I substrates pyruvate (P), malate (M), and glutamate (G) (3.5 mm each), while simultaneously recording extramitochondrial [Ca²⁺] ([Ca]_e; upper) and light scattering at 540 nm (LS_450nm; lower) to monitor matrix swelling (downward deflection = increased swelling). A, in the absence of BSA, Ca²⁺ pulses (2.5 μm each) were added until matrix Ca²⁺ release occurred (asterisk), at which time the matrix swelling rate also increased markedly. CsA caused Ca²⁺ reuptake and immediately stopped further swelling. B, with 1 mg/ml BSA present, a larger amount of Ca²⁺ was required to induce matrix Ca²⁺ release (asterisk), which was reversed by addition of CsA. The swelling rate did not change appreciably before or after CsA addition, indicating that the IMM remained impermeant to sucrose. In both cases, alamethicin (Ala; 10 μg) was added at the end for calibration purposes to show maximum Ca²⁺ release and swelling.

Why is BSA required to observe released matrix Ca²⁺ without detectable dissipation of the average ΔΨ_m? We hypothesized that when a mPTP stochastically opens and immediately dissipates ΔΨ_m, it will close stochastically while the mitochondria are still depolarized. The mitochondria can then regenerate ΔΨ_m, which they can do much more rapidly if the IMM proton leak is low due to BSA chelating fatty acids. Because mPTP open probability is voltage-dependent, the probability that the mPTP will reopen remains high as long as the ΔΨ_m is depolarized but becomes much lower once ΔΨ_m has been re-established. Thus, the time to repolarize ΔΨ_m is a critical determinant of whether the mPTP will reopen or not. We postulated that in the absence of BSA, the longer time required to repolarize ΔΨ_m due to the greater IMM proton leakiness makes the closed mPTP more likely to reopen, creating a positive feedback in which loss of essential matrix metabolites such as NAD/NADH further impedes the ability of respiration to restore ΔΨ_m, such that mPTP opening becomes long-lasting and eventually irreversible. To test this hypothesis, we examined the effects of increasing the IMM proton leak when BSA was present by adding a low concentration of the protonophore carbonyl cyanide p-trifluoromethoxyphenylhydrazone (FCCP). With BSA present, 15–25 nm FCCP caused only a mild ΔΨ_m dissipation before Ca²⁺ addition but eliminated the ability of mitochondria to maintain ΔΨ_m (Fig. 2) and robust O₂ consumption during matrix Ca²⁺ release (supplemental Fig. S2), consistent with long-lasting mPTP opening. Thus, FCCP abrogated the effects of BSA.

Molecular Weight Cutoff of Transient mPTP Openings

If mitochondria can utilize transient mPTP openings to flush matrix Ca²⁺ rapidly, they must retain key matrix metabolites such as NADH during the brief time when an mPTP is open to have sufficient respiratory power to repolarize once the mPTP closes. However, the free diffusion of NADH in water is only ∼2-fold slower compared with Ca²⁺, implying that another factor must be present to restrict NADH diffusion relative to Ca²⁺ diffusion through the open mPTP. Possible explanations are that transient mPTP openings are too brief or occur in a low conductance mode that permits Ca²⁺ efflux but restricts efflux of larger molecules such as NADH (9–11, 20, 21). To investigate this possibility, we compared the extent of matrix swelling during Ca²⁺ loading in the absence and presence of BSA (Figs. 3 and 4). mPTP opening causes matrix swelling due to unrestricted entry of osmotically active molecules into the matrix when the pore is open. When mitochondria were suspended in KCl buffer, swelling after the onset of matrix Ca²⁺ release was comparable without or with BSA present (Fig. 4). In sucrose buffer without BSA, matrix swelling was also similarly large and terminated once CsA was added (Fig. 3A). However, with BSA present in sucrose buffer, matrix swelling was markedly blunted (Fig. 3B), indicating that the IMM remained impermeant to sucrose (M_r 340) during matrix Ca²⁺ release. This indicates that with BSA present, mPTP openings allowed small ions such as potassium and calcium with M_r <340 to pass freely through the pore but were either too brief or exhibited a reduced conductance state that restricted permeation of larger molecules such as sucrose (and NAD/NADH). In contrast, in the absence of BSA, the similar magnitude of swelling as in KCl buffer implies that sucrose was able to permeate the pore, consistent with classic large conductance mPTP openings, whose molecular weight cutoff has been estimated at 1500 (22).

FIGURE 4.

Summary of effects of BSA on matrix swelling in sucrose versus KCl buffer. Left, using the Ca²⁺-loading protocol in Fig. 3, mitochondrial swelling in sucrose buffer with 0.2 mm P_i, recorded from the onset of matrix Ca²⁺ release (0%), relative to maximal swelling after addition of alamethicin (100%) was much greater without BSA (●) than with BSA (○). Right, comparison with KCl buffer with 0.2 mm P_i in the presence and absence of BSA. Values are the mean ± S.D. of three to five preparations.

ROS Production Is Increased during Matrix Ca²⁺ Release via Transient mPTP Openings

Transient mPTP openings have been shown to increase mitochondrial ROS production (23) by inducing “superoxide flashes” (24). Consistent with these observations, we found that in the presence of BSA, matrix Ca²⁺ release with ΔΨ_m maintained was accompanied by a marked increase in mitochondrial ROS generation, detected using the H₂O₂ indicator Amplex Red (Fig. 5). Note that in Fig. 5, O₂ consumption transiently increased during matrix Ca²⁺ release and then remained stable, indicating that respiratory chain activity was not compromised by loss of essential matrix metabolites such as NADH during transient mPTP openings. Also, subsequent addition of P_i stimulated reuptake of Ca²⁺, which can occur only if mPTP can close to regenerate ΔΨ_m. The ability of P_i to reverse matrix Ca²⁺ efflux under these conditions was also confirmed using the extramitochondrial Ca²⁺-sensitive dye Calcium Green-5N in place of the Ca²⁺-sensitive electrode (supplemental Fig. S3). Finally, increased H₂O₂ production during matrix Ca²⁺ release was also observed when P_i was present prior to the onset of Ca²⁺ loading (Fig. 5, lower panel).

FIGURE 5.

Increased ROS production during matrix Ca²⁺ release. Upper panel, traces of Amplex Red (resorufin) fluorescence measuring mitochondrial H₂O₂ production (F_H2O2; upper), O₂ consumption (lower), and extramitochondrial free [Ca²⁺] ([Ca]_e; middle) during addition of Ca²⁺ pulses to induce matrix Ca²⁺ release in the absence of P_i as in Fig. 1A. O₂ consumption and H₂O₂ production rates increased markedly coincident with matrix Ca²⁺ release and slowed after addition of P_i (2.5 mm). Note that mitochondrial Ca²⁺ reuptake promoted by P_i is possible to explain only if mPTP openings during matrix Ca²⁺ release were transient. Ala, alamethicin; P, pyruvate; M, malate; G, glutamate. B, bar graph summarizing the increase in H₂O₂ production rate after the onset of matrix Ca²⁺ release (MCR) without (−) or with (+) 2.5 mm P_i present from the start. The increase in H₂O₂ production rate was measured as the percent increase in the slope of the F_H2O2 curve after matrix Ca²⁺ release (as in A). Values are the mean ± S.D. for the number of preparations indicated, reflecting the slope of F_H2O2. a.u., arbitrary units.

DISCUSSION

Transient mPTP openings have been proposed to serve as a Ca²⁺ release mechanism by which mitochondria avoid matrix Ca²⁺ overload (9–11, 20, 21). We reasoned that this mechanism could explain our experimental finding of matrix Ca²⁺ release without dissipation of the average ΔΨ_m of the mitochondrial population if (i) transient mPTP openings occur asynchronously in the mitochondrial population, and (ii) the release of accumulated matrix Ca²⁺ during a transient mPTP opening is rapid relative to the rate of Ca²⁺ reuptake into the matrix. For example, if an mPTP opening of 100 ms were sufficient to re-equilibrate matrix-free Ca²⁺, but subsequent re-accumulation of matrix Ca²⁺ reuptake took 10 s (consistent with the half-time of Ca²⁺ uptake after a Ca²⁺ pulse in Fig. 1B), then the 100-to-1 difference between the rates of release and reuptake would require each mitochondrion to be depolarized only 1% of the time to flush Ca²⁺ from the matrix. Thus, as long as transient mPTP openings occur asynchronously due to their stochastic properties, only 1% of the mitochondria in the cuvette would be depolarized at any given moment, a fraction too small to be detected using TMRM (i.e. if ΔΨ_m remained at −180 mV in 99% of mitochondria, the 1% depolarized fraction would decrease the average ΔΨ_m by 1% to −178 mV). Moreover, because increased matrix Ca²⁺ stimulates respiration, promoting ΔΨ_m hyperpolarization, then if 99% of mitochondria modestly increased ΔΨ_m (e.g. from −180 to −190 mV), the average ΔΨ_m of the population would appear to hyperpolarize (e.g. −190 mV × 0.99 = −188 mV), completely obscuring the 1% of completely depolarized mitochondria with open mPTP. (The feasibility of this scenario was substantiated quantitatively in a mitochondrial model incorporating transient mPTP openings (supplemental Fig. S4).)

Fig. 6 summarizes schematically the proposed interplay between transient and long-lasting mPTP openings. When isolated mitochondria are subjected to Ca²⁺ loading under conditions in which IMM proton leak is allowed to increase (presumably caused by increased fatty acid cycling (19)), matrix Ca²⁺ release occurs concurrently with ΔΨ_m dissipation and swelling due to triggering of long-lasting mPTP openings. This is the basis of the classic assay for the Ca²⁺-induced mitochondrial permeability transition. However, when IMM proton leak during Ca²⁺ loading is minimized by including BSA to bind fatty acids, cardiac mitochondria are able to utilize transient mPTP openings to release accumulated matrix Ca²⁺ and then repolarize rapidly enough after the mPTP stochastically closes to inhibit subsequent mPTP reopening because mPTP open probability is suppressed by negative ΔΨ_m. If the repolarization rate is slow due to proton leak (i.e. without BSA or with BSA + a low FCCP concentration), then the mPTP reopens, further compromising respiratory power. Thus, the response depends on the ability of the respiratory chain to balance proton influx with proton pumping, which is possible when IMM proton leak is low. Because the reuptake of Ca²⁺ into the matrix via the mitochondrial Ca²⁺ uniporter is a slow process relative to mPTP-mediated Ca²⁺ release from the matrix, a mitochondrion can tolerate markedly elevated extramitochondrial free Ca²⁺ levels by depolarizing intermittently to flush Ca²⁺ while remaining polarized >99% of the time. Because mitochondria in situ are in a highly proteinaceous environment in which endogenous cytoplasmic proteins bind fatty acids similar to BSA, this mechanism may be physiologically important in allowing cardiac mitochondria to avoid matrix Ca²⁺ overload and maintain ATP production in the face of cyclical elevations in cytoplasmic free [Ca²⁺] to the micromolar level in the beating heart. This may also explain why large conductance mPTP opening is difficult to induce by Ca²⁺ loading alone in intact cardiac myocytes unless other factors such as high ROS levels are also present (25).

FIGURE 6.

Schema of matrix Ca²⁺ regulation by transient mPTP openings. As long as mitochondria have sufficient electron transport/proton pumping power, they can use transient mPTP openings to flush matrix Ca²⁺ and then quickly repolarize while only slowly re-accumulating matrix Ca²⁺, as outlined by the cycle inside the dashed box. This allows a population of asynchronously cycling mitochondria to tolerate markedly elevated extramitochondrial free Ca²⁺ levels without collectively depolarizing, whereas increased ROS production due to transient mPTP opening engages cardioprotective signaling. Eventually, however, depletion of critical metabolites from the matrix, together with a gradual decrease in respiratory chain activity, compromises the ability of mitochondria to repolarize quickly after transient mPTP openings, promoting the transition to long-lasting mPTP opening and its destructive consequences. See “Discussion” for further details. ETC, electron transport chain.

Given that cardiac mitochondria face varying cytoplasmic free [Ca²⁺] under physiological and pathological conditions, the importance of a mechanism to protect against excessive matrix Ca²⁺ overload and thereby avoid long-lasting mPTP opening and its destructive consequences is obvious. When ΔΨ_m is fully polarized, the energetic cost of removing excess Ca²⁺ from the matrix against a −180 mV driving force is huge, and both sodium-calcium exchange and hydrogen-calcium exchange are kinetically slower than Ca²⁺ uptake via the Ca²⁺ uniporter (26). Complete ΔΨ_m dissipation by a transient mPTP opening allows accumulated matrix Ca²⁺ to flow rapidly out of the matrix down its concentration gradient and equilibrate with cytoplasmic free [Ca²⁺] (27) at a much faster rate than possible via sodium-calcium or hydrogen-calcium exchange. As long as the mPTP reliably closes again (promoted by reduction of matrix-free [Ca²⁺], proton influx acidifying the matrix pH, and magnesium released during ATP hydrolysis (22)) and respiratory chain activity is not compromised by loss of critical matrix metabolites such as NADH, the mitochondrion can rapidly regenerate ΔΨ_m and return to its normal function of synthesizing ATP. The average ΔΨ_m of the mitochondrial population was sometimes even observed to hyperpolarize modestly during matrix Ca²⁺ release, which we speculate may have resulted from the combination of Ca²⁺-stimulated respiration and matrix acidification accompanying transient mPTP opening, which converts the chemical pH gradient into ΔΨ_m.

For the mechanism in Fig. 6 to work, transient mPTP openings must promote rapid Ca²⁺ efflux from the matrix without allowing loss of key metabolites such as NAD/NADH, which would compromise the ability of electron transport to regenerate ΔΨ_m rapidly once the pore closed. One possibility is that the transient openings occur in a low conductance mode, as reported previously (9, 20, 21), perhaps corresponding to a subconductance state of the fully open pore as recorded in mitoplasts (28). This possibility is consistent with the results using sucrose buffer (Figs. 3 and 4) showing that mitochondrial swelling during matrix Ca²⁺ release was minimal with BSA present compared with no BSA or with KCl buffer. This could be interpreted to indicate that molecules with M_r 340 or greater (including both sucrose and NAD/NADH) remain impermeant to the IMM during transient mPTP openings. On the other hand, it has been shown previously that calcein (M_r 623) can be released by transient mPTP openings without any detectable loss of ΔΨ_m (29), suggesting that calcein and NAD/NADH may have different permeation properties despite their similar molecular weights. Thus, it is possible that transient openings occur in the full conductance mode, but other factors, such as the brevity of openings or binding properties in the matrix, restrict the diffusion of specific molecules such as NAD/NADH (M_r ∼663) and sucrose relative to calcein and small ions like calcium and potassium. With long-lasting mPTP openings sufficient to depolarize the average ΔΨ_m of the mitochondrial population, however, NADH, like sucrose, freely permeated the pore, as shown by the increase in O₂ consumption upon NADH addition in Fig. 1A. It seems unlikely that matrix Ca²⁺ efflux could occur via the mitochondrial Ca²⁺ uniporter instead of mPTP because the available evidence suggests that the uniporter is an inwardly rectifying Ca²⁺ channel with very small unitary conductance compared with the mPTP (30, 31). In addition, the rate of matrix Ca²⁺ uptake via the uniporter had a half-time exceeding 10 s (Fig. 1), yet matrix Ca²⁺ efflux through the uniporter (which is slower than influx due to the inward rectification) would have to occur much more rapidly during the brief (<1 s) depolarization while the mPTP is open.

We speculate that the transition from transient to long-lasting mPTP openings is gradual, eventually leading to depletion of key metabolites for electron transport. This, together with increased IMM leak, could be a critical factor converting transient openings to long-lasting openings with their destructive consequences (Fig. 6). Irrespective of the precise mechanism regulating pore permeation, for mPTP openings to remain transient requires that mitochondria have a well functioning respiratory chain with low IMM proton leak. Although we speculate that the major role of BSA is to prevent IMM proton leakiness by chelating fatty acids (mimicking cytoplasmic fatty acid-binding proteins normally present in situ), we cannot absolutely exclude the possibility that BSA also has direct effects on mitochondrial Ca²⁺-handling proteins or the mPTP.

Implications for Cardioprotection

The scenario outlined in Fig. 6 could explain the accelerated development of heart failure in CyPD KO mice exposed to transaortic constriction or cross-bred with cardiomyopathy-susceptible mice overexpressing Ca²⁺/calmodulin-dependent protein kinase IIδc, consistent with the impaired mitochondrial Ca²⁺ handling observed in these animals (12) due to defective transient mPTP openings. In addition, this scenario may account for several observations relevant to cardioprotection by ischemic and pharmacologic PC. The induction of cardioprotection is known to be dependent on signaling cascades triggered by ROS production during the PC period (4, 5). Pre-conditioning I/R episodes are likely to expose mitochondria to modest Ca²⁺ overload conditions, which, if sufficient to trigger transient mPTP openings, would secondarily increase ROS production by generating superoxide flashes (23, 24), as observed in Fig. 5. This ROS generation may synergistically summate with other sources of mitochondrial ROS production, e.g. from activation of mitochondrial K_ATP channels, to initiate ROS-dependent cardioprotective signaling. A requirement for multiple sources to generate adequate ROS to engage cardioprotective signaling would explain why ROS scavengers, CsA, or mitochondrial K_ATP antagonists are all individually effective at blocking cardioprotection when administered during PC episodes (4, 5). The specific source generating ROS seems to be less important than the absolute amount of ROS because exogenously applied ROS are also effective at triggering cardioprotective signaling (4, 5).

The activation of this protective role of transient mPTP openings is, however, a two-edged sword. To tolerate extramitochondrial free [Ca²⁺] >5 μm without depolarizing, isolated cardiac mitochondria had to have well functioning electron transport with efficient proton pumping coupled to a low level of IMM proton leak. If electron transport/proton pumping was compromised and/or the IMM was leaky, matrix Ca²⁺ release was always accompanied by significant ΔΨ_m dissipation. As the duration of ischemia is prolonged, the progressive accumulation of fatty acids (32) increases IMM leakiness, and electron transport power weakens. Under these conditions, the likelihood increases that transient mPTP openings will be converted into long-lasting full conductance openings, leading to irreversible mitochondrial injury.

Limitations

Several limitations of this study should be recognized. First, our studies were performed in isolated mitochondria, in which the physiological environment of mitochondria is perturbed, and normal signaling may be partially or completely disrupted. BSA may have different properties than the fatty acid-binding proteins normally present in the cytoplasm or could regulate mitochondrial Ca²⁺-cycling proteins or mPTP directly. However, the BSA concentration used in our study was much lower than that used in other studies to mimic the oncotic pressure of the cytoplasm (33), and the rate of Ca²⁺ uptake during the initial Ca²⁺ pulses was similar in the presence and absence of BSA (e.g. Fig. 1). We confirmed that BSA did not quench TMRM fluorescence or bind tetraphenylphosphonium in a manner that could artifactually explain the maintenance of ΔΨ_m during Ca²⁺ loading. We inferred transient mPTP openings but could not directly observe them in individual mitochondria because only the average ΔΨ_m of the total mitochondrial population can be measured in a spectrofluorometer cuvette. However, the plausibility of the proposed mechanism is supported by theoretical considerations (22) and reproduced quantitatively by the mathematical modeling (see supplemental “Experimental Procedures”). In future studies, it may be possible to observe transient mPTP openings directly by imaging single isolated (34) or in situ (24) mitochondria. Although the open times may be too short to detect with voltage indicators such as TMRM, our findings predict that transient mPTP openings in single mitochondria should be accompanied by rapid matrix Ca²⁺ depletion, and recover slowly over a time course of minutes and be suppressed by CsA. On the other hand, if transient mPTP openings occur as infrequently as once every 25 min, as predicted in the quantitative model, they may be difficult to detect.