Reactive oxygen species production induced by pore opening in cardiac mitochondria: The role of complex III

Abstract

Recent evidence has implicated succinate-driven reverse electron transport (RET) through complex I as a major source of damaging reactive oxygen species (ROS) underlying reperfusion injury after prolonged cardiac ischemia. However, this explanation may be incomplete, because RET on reperfusion is self-limiting and therefore transient. RET can only generate ROS when mitochondria are well polarized, and it ceases when permeability transition pores (PTP) open during reperfusion. Because prolonged ischemia/reperfusion also damages electron transport complexes, we investigated whether such damage could lead to ROS production after PTP opening has occurred. Using isolated cardiac mitochondria, we demonstrate a novel mechanism by which antimycin-inhibited complex III generates significant amounts of ROS in the presence of Mg²⁺ and NAD⁺ and the absence of exogenous substrates upon inner membrane pore formation by alamethicin or Ca²⁺-induced PTP opening. We show that H₂O₂ production under these conditions is related to Mg²⁺-dependent NADH generation by malic enzyme. H₂O₂ production is blocked by stigmatellin, indicating its origin from complex III, and by piericidin, demonstrating the importance of NADH-related ubiquinone reduction for ROS production under these conditions. For maximal ROS production, the rate of NADH generation has to be equal or below that of NADH oxidation, as further increases in [NADH] elevate ubiquinol-related complex III reduction beyond the optimal range for ROS generation. These results suggest that if complex III is damaged during ischemia, PTP opening may result in succinate/malate-fueled ROS production from complex III due to activation of malic enzyme by increases in matrix [Mg²⁺], [NAD⁺], and [ADP].

Introduction

When the heart is reperfused after prolonged ischemia, damaging reactive oxygen species (ROS)² produced by cardiac mitochondria are thought to play a major role in ischemia/reperfusion (I/R) injury. Although at least 10 distinct sites in the electron transport chain and redox-active matrix proteins can be involved in ROS generation (1, 2), the major focus has centered on complexes I and III (3, 4), with recent evidence favoring succinate-driven reverse electron transport (RET) through complex I as the major mechanism (5, 6). However, RET is possible only when membrane potential (ΔΨ) is fully polarized (7–9). This requires mitochondria, which are depolarized during prolonged ischemia, to recover ΔΨ fully to generate succinate-driven ROS via RET. Therefore, when ΔΨ either does not recover (10, 11) or recovers only slowly and partially (12–14) upon reperfusion after prolonged ischemia as a result of permeability transition pore (PTP) opening (15) and increased inner membrane leakiness (16–18), ROS production by succinate-driven RET would be severely limited unless a sufficient subpopulation of mitochondria recovered ΔΨ fully. Given the strong evidence linking succinate accumulation to ROS-induced reperfusion injury (5), we therefore examined whether other mechanisms might come into play to promote additional ROS production by electron transport complexes after PTP opening has occurred and RET-driven ROS production has ceased.

In contrast to succinate-fueled RET-driven ROS production by complex I, ROS production by forward electron transport through complexes I or III requires the flow of electrons out of these complexes to be inhibited. When blocked using specific chemical inhibitors, the general consensus is that maximal ROS production at complex I is about half of that at complex III (19). In the absence of chemical inhibitors, mounting evidence also indicates that complexes I and III are damaged by prolonged I/R in vivo (3, 20), raising the possibility that they also contribute to the damaging burst of ROS in this setting (21–24). Nevertheless, the relative importance of complexes I and III, as well as other potential sources of ROS during reperfusion, is still debated. For forward electron transport into inhibited complex I to generate significant ROS requires a very high NADH/NAD ratio (>10) (25). Although this ratio rises during ischemia, it falls rapidly during reoxygenation (16, 17, 26) as mitochondria attempt to regenerate membrane potential (ΔΨ), leaving only a brief possible time window for significant ROS production. In addition, the complex I inhibitor rotenone has been found to decrease, rather than potentiate, ROS production by I/R (27, 28), suggesting that ROS is produced either by RET at complex I or downstream by complex III. It should be noted that recent experiments in Langendorff-perfused rat hearts subjected to 30 min of global ischemia demonstrated that increased ROS production during reperfusion occurred after, rather than before, PTP opening (29). Also, both ischemic preconditioning and cyclosporin A (CsA), despite inhibiting PTP opening during reperfusion, did not augment ROS production as might be expected if RET were prolonged during reperfusion but instead attenuated ROS production (29).

Forward electron flow into inhibited complex III can generate ROS, but the relationship between respiratory chain redox state and ROS production by complex III is less straightforward than for inhibited complex I. Experiments with isolated mitochondria have shown that ROS production by antimycin-inhibited complex III first increases and then decreases as succinate concentration is gradually increased. Likewise, the relationship between H₂O₂ production by antimycin-inhibited complex III and the supply of malate and glutamate is also bell-shaped (30). These and other observations (31, 32) suggest that in contrast to the high NADH/NAD⁺ ratio required for increased ROS production by inhibited complex I (33, 34), a relatively low NADH/NAD⁺ ratio, corresponding to a low ubiquinol (QH2)/ubiquinone (Q) ratio, is required to maximize ROS production by antimycin-inhibited complex III. Here we report the novel finding that when complex III is inhibited by antimycin in a setting in which succinate and its closely related metabolites are in limited supply, ROS production by complex III is dramatically increased after pore opening in the inner membrane by addition of NAD⁺/Mg²⁺ to activate malic enzyme (ME2). These findings may be relevant to the damaging ROS burst during reperfusion after prolonged ischemia when complex III has been damaged, and mitochondrial permeability transition pore opening leads to elevated matrix Mg²⁺, ADP, and a low NADH/NAD⁺ ratio.

Results

Oxidation of endogenous substrates and H₂O₂ production by antimycin-inhibited complex III

When isolated cardiac mitochondria were added to sucrose buffer followed by activation of the respiratory chain with ADP and carbonyl cyanide p-chlorophenylhydrazone, addition of the complex III inhibitor antimycin stimulated modest H₂O₂ production (0.1–0.15 nmol/min/mg) driven by oxidation of endogenous substrates (Fig. 1A, red trace). To simulate pore opening under these conditions, mitochondria were then exposed to alamethicin. H₂O₂ production decreased but could be markedly accelerated (6–7-fold) by the addition of Mg²⁺ and NAD⁺. The latter effect was inhibited by stigmatellin, indicating that H₂O₂ production originated from complex III. The dependence on endogenous substrates, despite inner membrane permeabilization, was revealed both by the suppression of H₂O₂ production with malonate to inhibit succinate oxidation by complex II, and by its relief by adding exogenous succinate in excess (Fig. 1A, blue trace). Fig. 1B presents summary data. The findings suggest that pore opening in the inner membrane, generated in this case by alamethicin, rapidly increased H₂O₂ production by antimycin-inhibited complex III in the presence of ADP, NAD⁺, and MgCl₂ and in the absence of exogenous substrates. This is further demonstrated in Fig. 1C, where pore opening with alamethicin induced significant acceleration of H₂O₂ production whether antimycin had been added before (Fig. 1C, red trace) or after ADP, NAD⁺, and Mg²⁺ (Fig. 1C, blue trace), indicating that the increase in matrix concentration of these substances was critical for stigmatellin-sensitive ROS production. Note also that H₂O₂ production was significantly inhibited with piericidin, which blocks Q reduction by NADH, implying that Q reduction was required for ROS production by antimycin-inhibited complex III.

Figure 1.

Mg²⁺ accelerates H₂O₂ production by antimycin-inhibited complex III in isolated mitochondria respiring on endogenous substrates. A, when isolated mitochondria (Mito) treated with carbonyl cyanide p-chlorophenylhydrazone (CCCP) (1 μm) and ADP (0.25 mm) were exposed to antimycin (Ant, 1 μm), H₂O₂ production increased. Permeabilization with alamethicin (Ala, 20 μg/ml) reduced H₂O₂ production, but when MgCl₂ (2.5 mm) and NAD⁺ (0.5 mm) were added, H₂O₂ production increased markedly and was inhibited by stigmatellin (Stigm, 0.1 μm), indicating the ROS originated from complex III (red trace). During Mg²⁺/NAD⁺-stimulated H₂O₂ production, addition of malonate (2.5 mm) to inhibit endogenous succinate oxidation decreased H₂O₂ production that was reversed by adding exogenous succinate (Succ) (5 mm) (blue trace). In this and other figures, additions to all traces are indicated by black arrows; additions to one trace only are indicated by the arrows of the same color as the trace. B, summary data (median ± 95% confidence intervals, *, p < 0.05) for antimycin-dependent H₂O₂ production in intact and permeabilized mitochondria in the absence of exogenous substrates/co-factors. C, complex III was inhibited with antimycin (1 μm) added before (red trace) or after ADP (0.25 mm), NAD⁺ (0.5 mm), and MgCl₂ (2.5 mm) (blue trace). Pore opening with alamethicin rapidly increased H₂O₂ production that was inhibited by stigmatellin (0.5 μm) or piericidin (pier, 1.5 μm). D, in permeabilized mitochondria, addition of ADP (0.25 mm) and NAD⁺ (0.5 mm) did not stimulate O₂ consumption (upper traces) until Mg²⁺ (2.5 mm) was added, indicating that Mg²⁺ activated oxidation of endogenous substrates. Note that the modest increase in O₂ consumption was maintained for 4 min (blue trace) and 11 min (red trace), and even after antimycin addition (1 μm) rapidly accelerated H₂O₂ production (lower traces). Both O₂ consumption and H₂O₂ production were inhibited by stigmatellin (0.1 μm).

The relationship between H₂O₂ production and O₂ consumption in the absence of exogenous substrates is shown in Fig. 1D. When cardiac mitochondria were permeabilized with alamethicin, sequential addition of ADP, NAD⁺, and MgCl₂ caused O₂ consumption to increase without significant H₂O₂ production, consistent with slow oxidation of endogenous substrates. This low oxidation rate was maintained for 4 min (Fig. 1D, blue trace) or 11 min (red trace) before antimycin was added, suggesting that inner membrane permeabilization does not result in rapid depletion of endogenous substrates to a level inhibiting complex III reduction over this time period. Antimycin addition did not affect O₂ consumption, but promoted rapid H₂O₂ production that was not much different for the 4- or 11-min incubation periods (Fig. 1D, lower red and blue traces). Under these conditions, stigmatellin inhibited both mitochondrial O₂ consumption and H₂O₂ production, indicating that (a) H₂O₂ was produced by complex III and (b) O₂ consumption after inhibition of the respiratory chain with antimycin was diverted to H₂O₂ generation.

Retention of substrates is required for NAD⁺ reduction during Mg²⁺-stimulated H₂O₂ generation in alamethicin-permeabilized mitochondria

A surprising finding in Fig. 1D is that neither NAD⁺/Mg²⁺-stimulated O₂ consumption nor antimycin-induced ROS production were decreased in alamethicin-permeabilized mitochondria despite the relatively long incubation in the absence of exogenous substrates. However, previous studies have demonstrated that mitochondria isolated from various tissues (35), including rabbit heart (36), contain endogenous substrates in low millimolar concentrations. In this context, the experiments in Fig. 2 were designed to study how quickly efflux of substrates following alamethicin administration begins to limit ADP/NAD⁺/Mg²⁺-induced H₂O₂ production by antimycin-inhibited complex III. Mitochondria were permeabilized with alamethicin and incubated for 1, 3, 6, and 9 min before antimycin was added, followed by ADP, NAD⁺, and MgCl₂ addition. It is evident that stigmatellin-sensitive NAD⁺/Mg²⁺-dependent H₂O₂ production was not significantly reduced even after the longest delay. Thus, even after 9 min, NADH-related reduction of complex III had not yet decreased below the level required for NAD⁺/Mg²⁺-induced ROS production by antimycin-inhibited complex III.

Figure 2.

Endogenous substrates supporting NAD⁺/Mg²⁺-induced complex III reduction and H₂O₂ production are retained for >9 min after alamethicin addition. Mitochondria (Mito) were permeabilized with alamethicin (Ala), and antimycin (Ant) was added 1 (purple trace), 3 (green trace), 6 (red trace), and 9 min later (blue trace) as indicated, followed by ADP (0.25 mm), NAD⁺ (0.5 mm), MgCl₂ (2.5 mm), and stigmatellin (0.5 μm). Additions to all traces are indicated by black arrows; additions to one trace only are indicated by the arrows of the same color as the trace.

Mg²⁺ or Mn²⁺-stimulated NAD⁺-dependent malic enzyme (ME2) activity in cardiac mitochondria

Because malic enzyme (ME2), which converts malate + NAD(P)⁺ to pyruvate + NAD(P)H + CO₂, is to our knowledge the only NAD(P)H-generating enzyme in the matrix that has an absolute requirement for Mg²⁺ or Mn²⁺ (37), we next examined whether ME2 might be the source of NADH production fueling NAD⁺/Mg²⁺-induced ROS production by antimycin-inhibited complex III. ME2 activity in our preparations was assessed as illustrated in Fig. 3A. NAD⁺ addition to permeabilized mitochondria incubated in the presence of piericidin and malate resulted in a small increase in NADH followed by rapid inhibition. As shown previously (33), the inhibition is due to the increase in [NADH] and [oxaloacetate] that easily reverses the malate dehydrogenase (MDH) reaction in the direction of NADH oxidation and malate generation. Consistent with this mechanism, glutamate addition to promote oxaloacetate removal via aspartate aminotransferase resulted in rapid acceleration of NADH generation (Fig. 3A, green trace). Despite strong inhibition of the MDH reaction in forward direction in the absence of glutamate, however, MgCl₂ addition increased NADH generation, indicating Mg²⁺-induced activation of malate oxidation and NAD⁺ reduction. This concentration-dependent effect of MgCl₂ was inhibited and reversed with EDTA added in excess of MgCl₂. Under these conditions, ME2 should be fully inhibited such that only MDH is available for oxidation of the NADH generated by ME2 before EDTA addition. As expected, NADH oxidation by MDH was rapidly reversed into NADH generation by addition of glutamate. Note, that in the presence of 5 mm malate, addition of CoA to accelerate the ME2 reaction by removing pyruvate also accelerated NADH production.

Figure 3.

Stimulation of ME2 activity in cardiac mitochondria by [Mg²⁺] and [Mn²⁺]. A, NAD⁺ (0.5 mm) addition to alamethicin (Ala)- permeabilized mitochondria (Mito), incubated in the presence of piericidin (Pier) (1.5 μm) and malate (Mal) (5 mm), resulted in a limited and transient increase of NADH generation that was rapidly accelerated by glutamate (5 mm) addition (green trace). In the absence of glutamate, MgCl₂ addition caused a much slower increase in NADH that was a concentration-dependent effect, slightly accelerated by CoA (0.15 mm) and rapidly inhibited and reversed by adding EDTA > MgCl₂ (purple, orange, red, and blue traces). The decrease in [NADH] after EGTA was due to MDH working in the reverse direction as shown by a rapid increase in [NADH] after adding glutamate. B, similar results were obtained using MnCl₂ in place of MgCl₂ to activate ME2. C, acceleration of H₂O₂ production by antimycin (Ant)-inhibited complex III after MnCl₂ or MgCl₂ addition. H₂O₂ production by alamethicin-permeabilized, antimycin-inhibited mitochondria was stimulated by sequential addition of ADP (0.25 mm), NAD⁺ (0.5 mm), and MnCl₂ in the concentrations indicated at the end of traces. For comparison, an increase in H₂O₂ production after 2.5 mm MgCl₂ is also presented (green trace). In all cases H₂O₂ production was inhibited with piericidin (1 μm).

The experiments in Fig. 3A were also repeated with MnCl₂ addition in place of MgCl₂ (Fig. 3B). Compared with Mg²⁺ addition, the Mn²⁺-induced increase in NADH production was about 4–5 times faster. In addition, significantly lower [Mn²⁺] compared with [Mg²⁺] was required to increase NADH production.

This higher sensitivity of ME2 to Mn²⁺ was accompanied by greater Mn²⁺-induced H₂O₂ production by antimycin-inhibited complex III in the presence of NAD⁺ and absence of exogenous substrates (Fig. 3C). Under the same conditions in the same preparation, addition of 0.5 mm Mn²⁺ resulted in 2.8 times higher H₂O₂ production compared with addition of 2.5 mm Mg²⁺ (1.99 and 0.70 nmol/min/mg correspondingly). NAD⁺-dependent Mn²⁺- or Mg²⁺-activated H₂O₂ production by antimycin-inhibited complex III was significantly inhibited with piericidin (Fig. 3C), indicating that ROS production was largely dependent on NADH-related Q reduction.

Collectively, these results demonstrate that cardiac mitochondria contain ME2 which, when activated by Mg²⁺- or Mn²⁺, reduces NAD⁺ in the presence of malate and piericidin. Although this activity is low compared with MDH activity (about 115 versus 25 nmol/min/mg for Mn²⁺-activated versus Mg²⁺-activated enzyme), it is more than sufficient to reduce antimycin-inhibited complex III to an extent promoting significant H₂O₂ production in the presence of only endogenous substrates. In fact, ME2-mediated ROS production by complex III seems to work most efficiently in the presence of endogenous substrates, and it is easily inhibited by increasing [malate] to increase enzyme activity, as described below.

Functional analysis of key endogenous substrates/co-factors supporting Mg²⁺/NAD⁺-induced ROS production by antimycin-inhibited complex III, effects of complex I and II inhibitors

The simplest explanation for why small amounts of substrates are retained in alamethicin-permeabilized mitochondria may be binding to matrix structures and constituents, possibly enzymes. There is some evidence that ME2 binds NAD⁺ first, followed by malate (38). Perhaps this binding can take place before oxidative decarboxylation is stimulated by Mg²⁺ or Mn²⁺. Whatever the mechanism for retaining malate and related substrates in the vicinity of ME2, the expected significant efflux of substrates after pore opening lowers their level in the matrix, most likely below the detection limit. Therefore, the functional approach, although indirect, seems to be the most practical method available to clarify the importance of NAD⁺-related substrates for reduction and ROS production by antimycin-inhibited complex III in the absence of exogenous substrates.

In Fig. 4A, H₂O₂ production by antimycin-inhibited permeabilized mitochondria was evaluated after sequential addition of ADP, NAD⁺, and Mg²⁺ in the presence or absence of respiratory chain inhibitors. Mg²⁺ addition promoted significant H₂O₂ production that was inhibited with stigmatellin (orange trace, control). Stigmatellin addition to permeabilized mitochondria before ADP, NAD⁺, and Mg²⁺ also inhibited the NAD⁺- and Mg²⁺-induced increase in H₂O₂ production (Fig. 4A, green trace). Importantly, increased H₂O₂ production was also absent when NADH-related QH2 generation was blocked with piericidin (2 μm, Fig. 4A, red trace). Because piericidin is not expected to inhibit oxidation of endogenous succinate leading to QH2 generation, we conclude that in the absence of exogenous substrates, NADH-related QH2 generation and complex III reduction are mostly responsible for ROS production by antimycin-inhibited complex III. The NAD⁺/Mg²⁺-induced increase in H₂O₂ production was also blocked in the combined presence of piericidin to inhibit complex I and malonate (5 mm) to inhibit succinate oxidation by complex II (Fig. 4A, blue trace). Under those conditions, however, malonate inhibition was overcome by excess succinate (5 mm), which accelerated H₂O₂ production (Fig. 4A, blue trace). Malonate alone in the absence of piericidin, however, only partially inhibited H₂O₂ production, which could be rapidly accelerated by addition of malate (150 μm, Fig. 4A, purple trace). Note that H₂O₂ production rates are similar (about 1.1 nmol/min/mg) for all conditions, suggesting that perhaps the maximum rate was achieved during oxidation of endogenous substrates. These results suggest that cardiac mitochondria after pore opening in the inner membrane still contain NAD⁺-related substrates that can be utilized (oxidized) for Mg²⁺-stimulated NAD⁺ reduction by ME2 to support ROS production by antimycin-inhibited complex III. Fig. 4B summarizes inhibitory effects of stigmatellin, piericidin, malonate, and piericidin plus malonate on Mg²⁺/NAD⁺-related H₂O₂ production by antimycin-inhibited complex III.

Figure 4.

Relative effect of NADH-related and succinate-related Q reduction on NAD⁺/Mg²⁺-stimulated H₂O₂ generation by antimycin-inhibited complex III in the absence of exogenous substrates. A, alamethicin (Ala)-permeabilized, antimycin (Ant)-inhibited mitochondria (Mito) were subjected to sequential addition of ADP (0.25 mm), NAD⁺ (0.5 mm), and MgCl₂ (2.5 mm) that resulted in rapid H₂O₂ production suppressed by stigmatellin (Stigm, 0.5 μm) (orange trace, control). Stigmatellin (green trace) and piericidin (Pier, 1.5 μm) (red trace) added to permeabilized mitochondria before ADP, NAD⁺, and MgCl₂ inhibited H₂O₂ production. Piericidin inhibition of H₂O₂ production was alleviated by addition of malate/glutamate (5 mm each) to enhance the matrix NADH/NAD⁺ ratio sufficiently to support ROS production by complex I (red trace). Preincubation with malonate (Malon, 5 mm) resulted only in partial inhibition of H₂O₂ production that was enhanced to the control level by addition of malate (Mal, 0.15 mm) (purple trace). As expected, H₂O₂ production was inhibited in the presence of both piericidin and malonate and significantly activated after succinate (Succ) (5 mm) addition (blue trace). As in other figures, additions to all traces are indicated by black arrows; additions to one trace only are indicated by the arrows of the same color as the trace. B, summary data for inhibition of NAD^+/Mg²⁺-induced antimycin-dependent H₂O₂ production by complex III in the absence of exogenous substrates. H₂O₂ production in the absence of inhibitors other than antimycin averaged 0.99 nmol/min/mg, which was taken as 100%.

Functional analysis: order of substrate/co-factor addition

Although the best way to mimic the effects of pore opening at the start of reperfusion is to apply alamethicin in the presence of ADP, NAD⁺, and Mg²⁺ (as in Fig. 1), these conditions are not suitable for studying the regulation of H₂O₂ production by ADP, NAD⁺, and Mg²⁺ individually. Experiments therefore were performed with sequential addition of substrates, cofactors, and inhibitors after mitochondria were permeabilized with alamethicin to permit the relative importance of each factor on ROS production by antimycin-inhibited complex III to be assessed.

In the absence of exogenous substrates, antimycin had no significant effect on H₂O₂ production when added to permeabilized mitochondria before NAD⁺ and Mg²⁺ (Fig. 5A, as well as Figs. 2 and 3C). Addition of NAD⁺ alone slightly increased H₂O₂ production (Fig. 5A, red trace), whereas addition of ADP alone had no effect (Fig. 5A, blue trace). In both cases, however, Mg²⁺ stimulated robust H₂O₂ production that was rapidly inhibited with stigmatellin. These results show that H₂O₂ production by antimycin-inhibited complex III in the absence of exogenous substrates depends critically on Mg²⁺ and NAD⁺, although for maximum stimulation ADP is also required (Fig. 5B). Fig. 5C summarizes average stigmatellin-sensitive H₂O₂ production by antimycin-inhibited complex III in permeabilized mitochondria in the presence of ADP (0.25 mm), NAD⁺ (0.5 mm), and/or MgCl₂ (2.5 mm).

Figure 5.

Relative importance of Mg²⁺, NAD⁺, and ADP in stimulating H₂O₂ production by antimycin-inhibited complex III. A, in alamethicin (Ala)-permeabilized mitochondria (Mito) without exogenous substrates, antimycin (Ant, 1 μm) followed by addition of NAD⁺ (0.5 mm, red trace) and/or ADP (0.25 mm, blue trace) had little effect on H₂O₂ production until Mg²⁺ (2.5 mm) was also added. B, similarly, after antimycin, Mg²⁺ had only a modest effect on H₂O₂ production until NAD⁺ was also added (red trace), which was maximized when ADP was also present (blue trace). In all cases, stigmatellin (Stigm, 0.1 μm) was suppressed on H₂O₂ production. C, summary data (median ± 95% confidence intervals, *, p < 0.05) under the various conditions indicated. As in other figures, additions to all traces are indicated by black arrows; additions to one trace only are indicated by the arrows of the same color as the trace.

With NAD⁺ present, the rapid stimulation of H₂O₂ production by Mg²⁺ was concentration-dependent (Fig. 6A). EDTA almost completely inhibited Mg²⁺-induced H₂O₂ production, which was reactivated by adding Mg²⁺ in excess of EDTA. The small increase in H₂O₂ production after NAD⁺ alone may have been due to residual endogenous Mg²⁺, as it was significantly suppressed in the presence of 250 μm EDTA (Fig. 6A, blue trace). If Mg²⁺ was added before NAD⁺, however, then the increase in H₂O₂ production was modest (Fig. 6B). We attributed this modest H₂O₂ production to the presence of endogenous NAD⁺, because significant amounts of matrix NAD(H) are bound to matrix proteins (39, 40), which slow NAD(H) efflux after membrane permeabilization. In the absence of exogenous substrates, H₂O₂ production was significantly potentiated by exogenous NAD⁺ (0.5 mm).

Figure 6.

Dependence of H₂O₂ production by antimycin-inhibited complex III on [Mg²⁺] and exogenous [malate]. A, when alamethicin (Ala)-permeabilized mitochondria (Mito) without exogenous substrates were exposed to antimycin (Ant) to inhibit complex III, NAD⁺ (0.5 mm) caused a modest increase in H₂O₂ production that was progressively accelerated by increasing [Mg²⁺] from 0.05 to 2.5 mm (traces color-labeled as indicated) and then stopped by adding EDTA in excess, restarted by adding Mg²⁺ in excess, and terminated by stigmatellin (Stigm, 0.1 μm). Note that with 0.25 mm EDTA present from the start to chelate endogenous Mg²⁺, NAD⁺ alone failed to increase H₂O₂ production until Mg²⁺ was added in excess (blue trace). B, in the absence of NAD⁺, Mg²⁺ added at the various concentrations increased H₂O₂ production progressively that was markedly accelerated by NAD⁺ (0.5 mm). Addition of malate (Mal) at low concentration (0.1 mm) further accelerated H₂O₂ production at submillimolar [Mg²⁺], which was suppressed by adding EGTA in excess. Addition of malate at a high concentration (5 mm), however, suppressed H₂O₂ production in the presence of NAD⁺ (0.5 mm) and Mg²⁺ (2.5 mm) (pink trace). As in other figures, additions to all traces are indicated by black arrows; additions to one trace only are indicated by the arrows of the same color as the trace.

Addition of exogenous malate significantly influenced ROS production under these conditions. At a concentration of 0.1 mm, malate potentiated H₂O₂ production, especially at Mg²⁺ <0.5 mm (Fig. 6B). However, 5 mm malate suppressed H₂O₂ production (Fig. 6B, pink trace), consistent with the bell-shaped relationship reported previously (30).

These results show that in the absence of exogenous substrates, H₂O₂ production by antimycin-inhibited complex III depends on matrix [Mg²⁺] and requires NAD⁺. In addition, exogenous malate at low concentrations under these conditions further stimulates H₂O₂ production, whereas high malate concentrations inhibit H₂O₂ production.

Functional analysis: substrate/co-factor concentration dependence

To account for the ability of Mg²⁺ and NAD⁺ to stimulate robust activation of H₂O₂ production from antimycin-inhibited complex III when only endogenous or low concentrations of exogenous malate were present, we postulated that ME2 activation by Mg²⁺ and NAD⁺ generated a small increase in the NADH/NAD⁺ ratio, which optimized the QH2/Q ratio for maximal H₂O₂ production by inhibited complex III (30–32). With high concentrations of malate (5 mm), however, the NADH/NAD⁺ ratio, and consequently the QH2/Q ratio, becomes too high for maximal H₂O₂ production. We tested this hypothesis by first inhibiting endogenous succinate oxidation to malate by complex II inhibitors (malonate and AA5) and then adding malate at various concentrations to activate ME2. As expected, Mg²⁺-dependent generation of NADH increased progressively as malate concentration was increased (Fig. 7A). In contrast, H₂O₂ production rapidly increased after addition of low concentrations of malate but then decreased at concentrations >0.5 mm (Fig. 7B). Under these conditions, Q reduction depends on NADH generation, and the QH2/Q ratio is expected to increase progressively as malate concentration increases to support high NADH levels. Thus, ROS production was greatest when the NADH/NAD⁺ (and presumably QH2/Q) ratio was low at low malate concentrations, and subsequently decreased as malate concentration increased, with half-maximal inhibition of H₂O₂ production at about 0.7 mm malate (Fig. 7C). The ability of mitochondria to oxidize NADH was also evaluated by adding NADH after NADH generation/oxidation was close to steady state at different malate concentrations (Fig. 7A). When the rate of NADH oxidation (i.e. the rate of NADH fluorescence decrease after NADH addition) was plotted against H₂O₂ production, H₂O₂ production increased as NADH oxidation rate increased (Fig. 7D). Thus, when NADH oxidation was inhibited due to the high NADH/NAD⁺ ratio, which is also expected to increase the QH2/Q ratio in the presence of antimycin, Mg²⁺, NAD⁺ and high [malate], ROS production by complex III was low. In contrast, when NADH oxidation was rapid, as expected to occur at a low QH2/Q ratio, ROS production by antimycin-inhibited complex III was high.

Figure 7.

Relationship between [malate]-dependent increases in NADH and H₂O₂ production by antimycin-inhibited complex III. A, in the presence of AA5 (1 μm) and malonate (Malon, 2.5 mm) to inhibit succinate oxidation by complex II, alamethicin (Ala)-permeabilized mitochondria (Mito) were exposed to NAD⁺ (0.5 mm), MgCl₂ (2.5 mm), and antimycin (Ant, 1 μm), after which malate (Mal) was added in the concentrations indicated at the end of each trace. NADH fluorescence (F_NADH) increased progressively with increasing [malate]. After F_NADH reached near steady state, NADH boluses (25 μm three times) were then added to assess NADH oxidation rates relative to NADH generation. B, stigmatellin (Stigm, 0.1 μm)-sensitive H₂O₂ production under the same conditions. C, rate of H₂O₂ production versus [malate], calculated from the traces in B. H₂O₂ production peaked at 0.25 mm malate and thereafter rapidly decreased as [malate] increased. D, rate of H₂O₂ production versus NADH oxidation rates at different [malate], calculated from the traces in A and B, illustrating that decreased ability to lower [NADH] at high [malate] results in inhibition of malate-stimulated H₂O₂ production in the presence of NAD⁺ and Mg²⁺. The rate of NADH oxidation was calculated from the slope of NADH fluorescence decrease over 1 min, measured after last 25 μm NADH bolus. A.U., arbitrary unit.

Endogenous malate can either be present in the matrix from the start or generated by endogenous succinate/fumarate oxidation. Consistent with the latter pathway, Fig. 8 shows that under conditions where respiration was inhibited due to lack of O₂, the addition of succinate resulted in NADH generation that was Mg²⁺-dependent. Because NAD⁺ reduction by RET is excluded after membrane permeabilization, the most likely mechanism for the Mg²⁺-dependent increase in NADH is malate generation due to succinate oxidation, with subsequent malate oxidation by ME2 generating NADH, pyruvate, and CO_2. This interpretation is supported by the finding that adding CoA to metabolize the pyruvate generated by ME2 further potentiated Mg²⁺-induced NADH generation. Under those conditions, NADH generation was clearly demonstrated by adding O₂ (air) transiently to oxidize the accumulated NADH and observing its regeneration after O₂ was again removed. The NADH increase in the presence of 2.5 mm Mg²⁺ under these conditions was inhibited by 3 mm EDTA and subsequently stimulated by adding Mg²⁺ in excess of EDTA, demonstrating the Mg²⁺ dependence of this reaction.

Figure 8.

Mg²⁺-dependent NADH generation in succinate-energized mitochondria. Mitochondria (Mito) were energized with 5 mm succinate (Succ), which maximized NADH fluorescence (F_NADH, upper trace) while depleting buffer O₂ (lower trace). Mitochondria were then permeabilized with alamethicin (Ala), causing a decrease in F_NADH which recovered only slightly when NAD⁺ (0.5 mm) was added, until Mg²⁺ (2.5 mm) and CoA (0.15 mm) were also added to stimulate succinate/malate oxidation and NAD⁺ reduction. The generated NADH was rapidly oxidized by allowing a brief infusion of O₂ (air) into the buffer. NADH generation was stopped with excess EDTA (3 mm) and restarted with excess Mg²⁺ (5 mm). A.U., arbitrary unit.

Similar to malate, succinate's ability to support H₂O₂ production by antimycin-inhibited complex III was also concentration-dependent (Fig. 9A). In the absence of exogenous malate, H₂O₂ production increased when succinate concentration was increased from 25 to 100 μm but then decreased as concentration was further elevated to 0.5 or 2 mm. Addition of ADP and NAD⁺ had little effect in the absence of Mg²⁺. Most importantly, addition of MgCl₂ (2.5 mm) rapidly accelerated H₂O₂ production in the presence of low concentrations of succinate (25–100 μm), but it had no effect at 0.5 or 2 mm succinate. In the presence of 2 mm succinate, inhibition of succinate oxidation with malonate (2.5 mm) rapidly increased H₂O₂ production to the level recorded in the presence of low [succinate]. Note that Mg²⁺ added to mitochondria incubated in the absence of exogenous succinate also induced significant H₂O₂ production (Fig. 9A, red trace) that was only slightly lower than the Mg²⁺-stimulated production rates in the presence of 25–100 μm succinate. Changes in O₂ consumption recorded simultaneously with ROS production show that O₂ consumption slightly increased after acceleration of ROS production induced by MgCl₂ or malonate. Both H₂O₂ production and O₂ consumption were inhibited with stigmatellin, indicating that both were related to ROS production by complex III.

Figure 9.

Dependence of H₂O₂ production by antimycin-inhibited complex III on exogenous [succinate] and [fumarate]. A, alamethicin (Ala)-permeabilized mitochondria were energized with exogenous succinate (Succ) at the concentrations indicated, which induced O₂ consumption (lower trace) but not H₂O₂ production (upper trace) until antimycin (Ant, 1 μm) was added. H₂O₂ production was not affected by ADP (0.25 mm) and NAD⁺ (0.25 mm) addition, but was significantly accelerated by Mg²⁺ (2.5 mm) when [succinate] was 0.1 mm or less but not 0.5 mm or greater (brown and orange traces). With 2 mm [succinate], malonate (Malon, 2.5 mm) administered to inhibit succinate oxidation caused H₂O₂ production to increase markedly (orange trace). Stigmatellin (Stigm, 0.1 μm) inhibited H₂O₂ production in all cases. B, ROS production increased after sequential addition of ADP (0.25 mm), NAD⁺ (0.5 mm), and MgCl₂, depending on [MgCl₂] indicated at the end of trace. Fumarate (1 mm) accelerated H₂O₂ production, especially at low [Mg²⁺]. As in other figures, additions to all traces are indicated by black arrows; additions to one trace only are indicated by the arrows of the same color as the trace.

Fumarate in low millimolar concentrations is known to allosterically activate purified human recombinant ME2 (41). Fig. 9B demonstrates that 1 mm fumarate increased NAD⁺/Mg²⁺-activated H₂O₂ production by antimycin-inhibited complex III. Consistent with the more pronounced effect of fumarate on enzyme activity at lower substrate concentrations (41), the fumarate-induced increase in H₂O₂ production was more significant at lower [Mg²⁺]. However, besides allosteric activation of ME2, fumarate is also expected to increase malate production in permeabilized mitochondria, and we cannot exclude the possibility that this effect predominated under our conditions.

In the absence of exogenous substrates, NAD⁺/Mg²⁺ addition is expected to activate not only oxidation of malate available in the matrix, but all tricarboxylic acid segments involving succinate → fumarate → malate. Inhibition of succinate oxidation by malonate and AA5 cuts off malate production from succinate, such that subsequent activation of malate oxidation by adding NAD⁺/Mg²⁺ before inhibition of complex III should decrease endogenous malate levels available for NADH generation after antimycin is added. In line with these predictions, complex II inhibition depressed NAD⁺/Mg²⁺-stimulated H₂O₂ production by antimycin-inhibited complex III that varied with the timing of malonate + AA5 addition (Fig. 10A). The effect was least expressed when antimycin was added before (Fig. 10A, green trace) or together with NAD⁺/Mg²⁺ (orange trace). Adding antimycin 1 min (Fig. 10A, purple trace) or 3 min (Fig. 10A, red trace) after NAD⁺/Mg²⁺ resulted in further progressive decreases in H₂O₂ production. These results contrast with the much greater NAD⁺/Mg²⁺-stimulated ROS production by antimycin-inhibited complex III in the absence of complex II inhibitors (Fig. 10A, blue trace). Note that decreased H₂O₂ production after complex II inhibition was rapidly activated by low concentrations of malate and inhibited with stigmatellin.

Figure 10.

A, complex II inhibition significantly depresses NAD⁺/Mg²⁺-induced H₂O₂ production by antimycin-inhibited complex III in the absence of exogenous substrates. In the presence of AA5 (1 μm) and malonate (Malon, 2.5 mm) to inhibit succinate oxidation by complex II, alamethicin (Ala)-permeabilized mitochondria (Mito) were exposed to NAD⁺ (0.5 mm), MgCl₂ (2.5 mm), and antimycin (Ant, 1 μm) at the times indicated by the same-colored arrows for each trace. Whether antimycin was added before (green trace) or together with NAD⁺/Mg²⁺ (orange trace), H₂O₂ production was greater than when antimycin was added later (lilac and red traces), presumably because less endogenous malate had been oxidized by ME2 before complex III was inhibited. Consistent with this prediction, in all conditions subsequent addition of 0.25 mm malate (Mal) rapidly activated H₂O₂ production to a similar rate. In the absence of complex II inhibitors, NAD⁺/Mg²⁺-stimulated H₂O₂ production after antimycin was greater because oxidation of endogenous succinate was available to generate malate (blue trace). Stigmatellin (Stigm, 0.1 μm) suppressed H₂O₂ production in all cases. B, relative effectiveness of MDH versus ME2 at stimulating H₂O₂ production by antimycin-inhibited complex III. In alamethicin-permeabilized mitochondria exposed to antimycin (1 μm), addition of 1 or 5 mm malate (orange and brown traces, respectively) stimulated robust H₂O₂ production that was suppressed by addition of NAD⁺ and Mg²⁺, presumably by increasing the NADH/NAD⁺ ratio beyond that optimal for H₂O₂ production by antimycin-inhibited complex III (see text). Lower [malate] (green, purple, and blue traces) stimulated less H₂O₂ production initially until NAD⁺ and Mg²⁺ were added, after which H₂O₂ production markedly accelerated. Note that with no exogenous malate (red trace), oxidation of endogenous malate was insufficient to cause any H₂O₂ production by MDH until NAD⁺ was added, whereas subsequent activation of ME2 by adding Mg²⁺ led to a much more robust increase in H₂O₂ production. Stigmatellin (0.1 μm) suppressed H₂O₂ production in all cases. C, dependence of H₂O₂ production by antimycin-inhibited complex III on exogenous [NAD⁺]. In alamethicin (Ala)-permeabilized mitochondria exposed to antimycin (0.5 μm) in the absence of exogenous substrates, addition of NAD⁺ at the concentrations indicated stimulated progressive H₂O₂ production that was markedly accelerated by Mg²⁺ (2.5 mm) and inhibited by stigmatellin (0.1 μm). As in other figures, additions to all traces are indicated by black arrows; additions to one trace only are indicated by the arrows of the same color as the trace.

As shown above, H₂O₂ production by antimycin-inhibited complex III was maximal at low succinate or malate concentrations, which limited NADH and QH2 generation, consistent with previous evidence (30, 32, 42) that H₂O₂ production is maximized by a low QH2/Q ratio and suppressed by a high ratio. In addition to restricting substrate availability, NADH and QH2 generation can also be limited by restricting NAD⁺ availability. In Fig. 10B, malate was added to antimycin-inhibited permeabilized mitochondria in the absence of exogenous NAD⁺ and Mg²⁺. Without Mg²⁺ to activate ME2, malate can still be oxidized by MDH, but NADH generation is limited because the increase in oxaloacetate and NADH inhibits the reversible MDH reaction, unless oxaloacetate is removed by also adding glutamate (see Fig. 3A) (33). Under these conditions without exogenous NAD⁺ and Mg²⁺, malate concentrations of >0.2 mm were required to produce enough NADH by MDH to stimulate H₂O₂ production by antimycin-inhibited complex III (Fig. 10B). However, if NADH and QH2 generation were then further boosted by adding exogenous NAD⁺, H₂O₂ production was rapidly inhibited. In the same experiments, addition of Mg²⁺ to activate ME2 only stimulated robust H₂O₂ production at low malate concentrations (<0.2 mm). The rates of Mg²⁺-stimulated H₂O₂ production were similar in the absence of malate as for malate concentrations up to 0.2 mm, suggesting that the endogenous malate concentration in the matrix is submillimolar under these conditions.

We also studied other permutations of substrates NAD⁺ and Mg²⁺. Addition of 0.1 mm malate to antimycin-inhibited mitochondria slightly increased H₂O₂ production that was further significantly enhanced by Mg²⁺. Incubation of mitochondria with 250 μm EDTA suppressed malate-induced H₂O₂ production, which was then reversed by adding 2.5 mm Mg²⁺. MgCl₂ addition significantly activated H₂O₂ production at low concentrations of malate/glutamate (5–100 μm) in the absence of exogenous NAD⁺, but no activation was observed at 1 mm malate/glutamate. Because no NAD⁺ was added in these experiments, the findings indicate the following: (a) small amount of NAD(H) is still available for Q reduction after inner membrane permeabilization; (b) substrate-dependent activation of NADH generation can compensate for low NAD⁺ availability; and (c) NADH production above a certain level inhibits H₂O₂ generation, even when only endogenous NAD(H) is available (results not shown).

However, in the absence of exogenous substrates, Mg²⁺-stimulated H₂O₂ production by antimycin-inhibited complex III depended on exogenous [NAD⁺] in a concentration-dependent manner (Fig. 10C). Note that the NAD⁺-dependent increase in H₂O₂ production started at very low concentrations of added NAD⁺ and was already significant at 50 μm NAD⁺.

Collectively, our results show that the rate of ROS production by antimycin-inhibited complex III is strongly modulated by electrons derived from NADH and depends on a low NADH/NAD⁺ and presumably low QH2/Q ratio for significant ROS production, as dually determined by substrate and NAD⁺ concentrations. Accordingly, the level of H₂O₂ production can be regulated by increasing NAD⁺ when substrate availability is limited or by increasing substrate concentrations when NAD⁺ availability is limited.

Ca²⁺/P_i-induced PTP opening reproduces alamethicin's effects on Mg²⁺/NAD⁺-stimulated ROS production by antimycin-inhibited complex III

The above experiments were performed in mitochondria permeabilized by alamethicin to create nonselective pores in the inner mitochondrial membrane. To determine whether PTPs have similar effects as alamethicin pores, PTP opening was induced by sequential addition of P_i and Ca²⁺ before MgCl₂ and NAD⁺ were added (Fig. 11A). Similar to alamethicin, addition of NAD⁺/MgCl₂ induced an increase in H₂O₂ production when complex III was inhibited by antimycin that was suppressed by EGTA or CsA (Fig. 11B). In general, the population of mitochondria with open pores allowing NAD⁺/MgCl₂ access to the matrix induced with 100 μm Ca²⁺ was about half of that induced with alamethicin (20 μg/ml) (Fig. 11C).

Figure 11.

Ca²⁺-induced PTP opening accelerates H₂O₂ production by antimycin-inhibited complex III in the presence of NAD⁺ and MgCl₂. A, mitochondria (Mito) incubated in the presence of ADP (0.5 mm) and antimycin (Ant, 0.2 μm) were challenged with P_i (2.5 mm) and CaCl₂ (concentration indicated at the end of tracing). NAD⁺(0.5 mm) and MgCl₂ (2.5 mm) addition resulted in H₂O₂ production that was dependent on [Ca²⁺], inhibited in the presence of CsA (1 μm), and rapidly inhibited with stigmatellin (Stigm, 0.2 μm) or EDTA (3 mm). B, mitochondria, incubated under the same condition as in A and challenged with 100 μm Ca²⁺ in the presence of CsA (1 μm, red trace) or EGTA (250 μm, blue trace), followed by alamethicin and stigmatellin (0.2 μm). Green trace represents control without inhibition of Ca²⁺/P_i-induced PTP opening. C, summary data (median ± 95% confidence intervals) of NAD⁺/Mg²⁺-induced H₂O₂ production triggered by Ca²⁺-induced PTP opening as demonstrated in A. Values for H₂O₂ production in the presence of CsA (1 μm) and EGTA (0.25 mm) were obtained after adding 100 μm Ca²⁺.

Discussion

The major finding of this study is that Mg²⁺, NAD⁺, and ADP induce marked acceleration of ROS production when complex III is inhibited by antimycin in permeabilized cardiac mitochondria. This novel mechanism is able to increase mitochondrial ROS production significantly when formation of alamethicin pores or PTP in the inner membrane allows Mg²⁺, NAD⁺, and ADP to equilibrate between the matrix and extramitochondrial space. Under these conditions, Mg²⁺-dependent activation of ME2 generates reducing equivalents from malate that can markedly stimulate ROS production by inhibited complex III, but only when NADH production/oxidation rates are in an optimal range to maintain a low NADH/NAD⁺ ratio. A reasonable assumption is that QH2/Q ratio is also low under those conditions (Fig. 12). As discussed later, we hypothesize that if such conditions are present when PTP opening occurs following prolonged I/R, this mechanism could contribute to continued ROS production after succinate-fueled RET-driven ROS production has ceased.

Figure 12.

Schematic of ROS production when ME2 is activated by increases in matrix Mg²⁺, NAD⁺, and ADP after inner membrane permeabilization by alamethicin or PTP opening. Malate oxidation by ME2 generates NADH that donates electrons (red arrows) to the flavoprotein site (FMN) in complex I (C I, yellow), reducing Q to QH2. Piericidin (Pier) blocks Q reduction. QH2 is oxidized by complex III (C III, tan), donating electrons to cytochrome c (cyt c) and complex IV in the absence of complex III inhibitors. Antimycin (Ant) disrupts electron flow from cytochrome b (bL and bH) in complex III to Q, such that electrons accumulate in FeS clusters and are donated to O₂ to form superoxide or H₂O₂ (dashed red arrow), but only when the QH2/Q ratio is low (inset). Stigmatellin (Stigm) blocks electron transfer to the FeS clusters with which O₂ interacts to prevent ROS generation. By oxidizing succinate to fumarate (Fum), complex II (C II, blue) provides fumarate to ME2 to generate malate (Mal). Malonate (Malon) inhibits succinate oxidation by complex II, limiting the supply of malate to ME2. Pyr, pyruvate.

Critical importance of increased matrix [Mg²⁺] for NADH generation by ME2

In antimycin-inhibited permeabilized mitochondria, Mg²⁺-induced H₂O₂ production was rapidly and almost completely inhibited with EDTA > Mg²⁺ and reactivated with Mg²⁺ > EDTA, demonstrating that H₂O₂ production was strongly dependent on Mg²⁺. To the best of our knowledge, ME2 is the only NADH-generating enzyme in the matrix that has absolute requirement for Mg²⁺ or Mn²⁺ (37). In catalyzing oxidative decarboxylation of malate to generate NAD(P)H, pyruvate and CO₂, ME2 can reduce either NAD⁺ and NADP⁺, but under most conditions prefers NAD⁺ (37, 43). ME2 is allosterically activated by fumarate and inhibited by ATP (41, 43). Fumarate's effect is mainly to decrease the K_m value for malate, Mg²⁺, and NAD⁺, such that the greatest activation occurs at lower concentrations of malate, Mg²⁺, and NAD⁺ (41). We demonstrated that fumarate activates NAD⁺/Mg²⁺-induced ROS production by antimycin-inhibited complex III, the effect being most pronounced at low [Mg²⁺] (Fig. 9B). Because fumarate addition is expected to increase [malate] in the matrix as well, this effect may dominate in our conditions.

In permeabilized cardiac mitochondria, both Mg²⁺ and Mn²⁺ are known to have concentration-dependent effects on NADH generation in the presence of saturating concentrations of malate, NAD⁺, and complex I inhibitors, which are rapidly inhibited with EDTA (33). Mn²⁺ activates ME2 more effectively as demonstrated by the increase in NADH production at significantly lower [Mn²⁺] compared with [Mg²⁺] (33, 37) In this study we confirmed higher activation of NADH production by ME2 with Mn²⁺ compared with Mg²⁺ in permeabilized cardiac mitochondria. In addition, Mn²⁺ activated NADH production at lower concentrations compared with Mg²⁺. Although ME2 activity is relatively low in cardiac mitochondria, it is sufficient to generate a small increase in the NADH/NAD⁺ ratio required for significant ROS production by antimycin-inhibited complex III in the absence of exogenous substrates. In this regard, the higher rates of NADH production stimulated by Mn²⁺ also resulted in higher H₂O₂ production rates compared with Mg²⁺ in the absence of exogenous substrates. We did not study the effect of Mn²⁺ on H₂O₂ production in detail, mainly because the effect of I/R on [Mn²⁺] in cardiac cells is not known. Mg²⁺ alone had modest effects on H₂O₂ production that can be attributed to residual NAD(H) bound to matrix proteins (39, 40). Conversely, NAD⁺ addition alone slightly increased H₂O₂ production that was suppressed by 250 μm EDTA to chelate residual endogenous matrix Mg²⁺. Importantly, for low [Mg²⁺] at which H₂O₂ production after NAD⁺ addition was relatively modest, exogenous malate (0.1 mm) significantly accelerated H₂O₂ production. For higher [Mg²⁺] at which NAD⁺ addition stimulated more robust H₂O₂ production, however, exogenous malate (0.1 mm) had little additive effect on H₂O₂ production. Adding exogenous malate in relatively high (millimolar) concentrations during NAD⁺/Mg²⁺-stimulated H₂O₂ production, however, rapidly inhibited H₂O₂ production. These results suggest that the critical level of NAD⁺-related reduction of complex III, accomplished via Q reduction, is required to maximize stigmatellin-sensitive ROS production (30, 32, 42), and once this level is exceeded, ROS production rapidly decreases (Fig. 6).

Importance of limited NADH generation

Rather surprisingly, alamethicin-permeabilized mitochondria still contained enough endogenous substrates to reduce antimycin-inhibited complex III and markedly increase ROS production when Mg²⁺ and NAD⁺ were added to activate ME2. Because piericidin, known to inhibit Q reduction by NADH, effectively inhibited H₂O₂ production, oxidation of NAD⁺-related substrates was most likely responsible for NAD⁺/Mg²⁺-induced reduction of antimycin-inhibited complex III. However, because complex II inhibition significantly decreased NAD⁺/Mg²⁺-stimulated H₂O₂ production, we conclude that the sequence from succinate to fumarate was involved in malate production, thereby fueling ME2-induced NADH generation. The stimulating effect of ADP on Mg²⁺/NAD⁺-induced ROS production (Fig. 5) could be related to succinate generation from succinyl-CoA. In addition to GTP/GDP-specific succinyl-CoA synthetase, cardiac mitochondria also contain an ATP/ADP-specific enzyme (44) that catalyzes the reversible conversion of succinyl-CoA and ADP to CoA, ATP, and succinate. Succinate oxidation can also reduce the Q pool that connects complexes II and III. In intact mitochondria, an increase in added succinate has been shown to increase H₂O₂ production, reaching a maximum at about 1 mm, followed by a decrease at higher concentrations (30). However, because reported K_m values of dicarboxylate transporter for succinate are around 1 mm (45), the rate of succinate influx into the matrix in intact mitochondria may not be sufficient to maintain matrix [succinate] at the buffer level. This could explain why in permeabilized mitochondria, stigmatellin-sensitive H₂O₂ production reached a maximum at succinate concentration significantly lower than 1 mm. Most importantly, the increased ROS production at low [succinate] was significantly accelerated by adding 2.5 mm Mg²⁺ in the presence of ADP and NAD⁺. The ability of Mg²⁺ to enhance H₂O₂ production to an almost equal extent in the absence or presence of exogenous succinate (Fig. 9A) suggests that enough malate is available or generated from endogenous succinate under those conditions to optimize ROS production by antimycin-inhibited permeabilized mitochondria. The importance of endogenous substrate oxidation for limited NADH generation and NAD⁺/Mg²⁺-dependent ROS production was also demonstrated by the finding that adding exogenous malate only slightly increased H₂O₂ production when malate was increased from 0 to 0.2 mm, followed by significant inhibition at malate concentrations >1 mm (Fig. 10B). When complex II was inhibited to prevent succinate oxidation, higher exogenous malate concentrations were required to increase NAD⁺/Mg²⁺-dependent ROS production by complex III, but ROS production was still significantly inhibited at malate concentrations of >1 mm (Fig. 7B). Here, an important finding is that the decrease in H₂O₂ production occurred at malate concentrations that stimulated a robust increase in NADH fluorescence in the mitochondrial suspension (Fig. 7A).

At first sight these results may seem confusing because NADH-related reduction of complex III is definitely required for the Mg²⁺/NAD⁺-stimulated ROS production by complex III. However, these results might be explained by efficient electron transfer from ME2 to complex I and further from NADH to Q within the I–III supercomplex. It has been proposed that a certain fraction of Q, not utilized for succinate oxidation, is channeled within the I–III supercomplex in a compartmentalized manner. Some complex III units interact with complex I to form a supercomplex exclusively dedicated to NADH oxidation, where other complex III units not as tightly bound to complex I are mainly responsible for oxidation of QH2 generated by SDH using the free Q pool (46). If redox carriers are indeed compartmentalized and can accomplish efficient electron transfer from NADH to complex III by using only Q available within the I–III supercomplex, the corresponding changes in NADH fluorescence that are responsible for ROS production are likely to be small. In contrast, the increase in NADH production beyond the limit required to maximize stigmatellin-sensitive ROS production was readily detected (Fig. 7A). Nevertheless, based on recent kinetic evidence, caution should be exercised regarding substrate channeling between the enzymes in supercomplexes. If metabolic pathways for NADH and succinate interact and catalyze via a universally accessible Q pool as suggested (47), H₂O₂ production by antimycin-inhibited complex III is still dependent on the QH2/Q ratio.

Implications for ROS production in the setting of ischemia/reperfusion

Based on the work of Chouchani et al. (5, 6), a major source of damaging ROS upon reperfusion after prolonged ischemia is related to RET driven by succinate that has accumulated during ischemia. Because mitochondria become depolarized during ischemia, this mechanism requires that a significant subpopulation of mitochondria is able to fully recover ΔΨ during early reperfusion (7–9). As soon as RET-related ROS production or other factors trigger PTP opening, however, matrix depolarization will immediately terminate ROS production by RET (48). This negative feedback feature tends to limit sustained ROS production by RET. Indeed, after prolonged ischemia, ΔΨ may either recover only briefly before PTP opening occurs, may not recover at all (10, 11), or may recover only slowly and partially (12–14). In this scenario, our findings provide a novel mechanism that could sustain ROS generation from inhibited complex III when PTP opening causes RET to cease, assuming matrix succinate, fumarate, malate, and NAD⁺ concentrations are in the appropriate range for optimizing complex III reduction for ROS production (30–32) when ME2 is activated by elevated matrix Mg²⁺ and ADP (33, 37).

PTP opening has been demonstrated to occur at the start of reperfusion after prolonged ischemia (29, 49) and results in a collapse of protonmotive force, reversal of the F₀F₁-ATPase, and direct access to cytoplasmic ATP. As F₀F₁-ATPase hydrolyzes MgATP, both ADP/P_i and free Mg²⁺ increase, because the affinity of ATP for Mg²⁺ is about 10-fold higher than that of ADP (50). Alternatively, membrane potential depolarization may increase matrix-free Mg²⁺ and its release (51). Free Mg²⁺ elevation after ischemia/reperfusion (52–54) or anoxia/reoxygenation has been well documented (52, 53). Thus, if complex III has become damaged during prolonged ischemia, reperfusion sets the stage for robust continued ROS generation when PTP is open and allows the matrix to equilibrate with elevated cytoplasmic Mg²⁺ and to activate ME2 at the same time that the NADH/NAD⁺ ratio is low and ADP is elevated. Although increases in matrix [ADP] and [Mg²⁺] are known to inhibit PTP opening in settings that do not involve ROS production (55, 56), these inhibitors act from the matrix side where they also simultaneously increase ROS generation by the mechanism proposed in this study. Therefore, it seems unlikely that ROS generation will be significantly and rapidly inhibited even if ADP and Mg²⁺ induce PTP closure. In addition, increased ROS production favors PTP opening that may overpower inhibitory effect of ADP and Mg²⁺.

Inhibition of complex III has been reported to require ischemia durations of >30 min, whereas complex I inhibition occurs early after 15–20 min of ischemia (3, 57). However, ROS generation by inhibited complex I requires a very high NADH/NAD⁺ ratio (>10) (25), whereas the NADH/NAD⁺ ratio declines very rapidly upon reperfusion when O₂ again becomes available for ROS production (16, 17, 26, 29), agreeing with other evidence that forward electron flow through damaged complex I is not the major source of ROS during reperfusion (27–29). In contrast, for damaged complex III to produce ROS when ME2 is activated by increases in matrix Mg²⁺, NAD⁺, and ADP, the level of endogenous/exogenous substrates generating malate has to be at appropriate concentrations such that the NADH/NAD⁺ ratio remains low, expected to maintain a low Q/QH2 ratio required to maximize ROS production by antimycin-inhibited complex III (30–32).

Our proposed mechanism is consistent with the observation that inhibition of succinate oxidation by malonate (5, 58, 59) is cardioprotective, because malonate would not only limit RET through complex I, but also inhibit succinate conversion to fumarate and malate, thereby limiting the availability of malate for oxidation by ME2. Likewise, by preventing RET-derived ROS generation, complex I inhibition with rotenone or amobarbital during I/R should also abolish NADH-related reduction of complex III with a resulting decrease in ROS production and (by reducing PTP opening) reduced mitochondrial damage during cardiac ischemia, consistent with the cardioprotective effect of rotenone (27) or amobarbital (60). However, despite existing evidence that complex III plays a central role in ROS production (27, 61), the extent of damage to complex III by ischemia/reperfusion is not well defined. Ischemia/reperfusion may induce changes in the structural integrity of complex III, which in association with decreased electron transfer may result in higher ROS production. It has been demonstrated that proteinase K or heat treatment can induce increases in superoxide production by complex III similar to antimycin treatment. Increased ROS production has been explained by changes in the hydrophobic QH2 oxidation pocket that may also lead to increased accessibility of O₂ to the pocket to generate more superoxide or facilitate its release (62). Some useful information may come from comparative studies with the bacterial (Rhodobacter capsulatus) mutant complex III and human complex III mutant Y278C (homolog of R. capsulatus cytochrome b Y302C mutant). In both cases, mutation of a highly conserved tyrosine residue of cytochrome b, in the quinol oxidation site, results in surprisingly similar decreases in catalytic activity of complex III, accompanied by increased ROS production and cellular damage (63). However, the precise mechanism of complex III damage during ischemia/reperfusion remains elusive.

Limitations

Extrapolating our findings in alamethicin-permeabilized mitochondria to I/R in intact hearts is highly speculative, given that the intracellular milieu during I/R is constantly changing and is not precisely defined. Nevertheless, the finding that calcium-induced PTP opening reproduced the effects of alamethicin demonstrates that Mg²⁺/NAD⁺-stimulated ROS production by complex III is not an artifact of alamethicin pores, but is a general response to inner membrane permeabilization. Whether complex III is comparably inhibited during genuine ischemia as with antimycin, and whether the relative concentrations of Mg²⁺, NAD⁺, succinate, fumarate, malate, ADP, and other metabolites during reperfusion approximate those optimizing ROS production by inhibited complex III when ME2 is activated remain open questions. Although the definitive contribution to cardiac injury of ME2 and complex III, relative to RET via complex I, remains to be established, this is potentially testable by knocking down cardiac ME2 in mice to determine whether they demonstrate a smaller ROS burst and less cardiac injury following ischemia/reperfusion compared with wild-type mice. As shown in the accompanying article (73), complex II can also become a significant source of ROS production after PTP opening when either complex II or III is inhibited, and its quantitative contribution, relative to complex III-derived ROS, remains to be characterized.

Experimental procedures

Ethical approval

This study was approved by the UCLA Chancellor's Animal Research Committee (ARC 2003-063-23B) and was performed in accordance with the Guide for the Care and Use of Laboratory Animals published by the National Institutes of Health (Publication No. 85-23, revised 1996) and with UCLA Policy 990 on the Use of Laboratory Animal Subjects in Research (revised 2010).

Experimental techniques

All measurements were carried out using customized Fiber Optic Spectrofluorometer (Ocean Optics) in a partially open continuously stirred cuvette at room temperature (22–24 °C) (64). Mitochondria were isolated from rabbit hearts as described previously (65). Intact mitochondria were added (0.6 to 1.0 mg/ml) to incubation buffer containing 250 mm sucrose, 10 mm Hepes, pH 7.4, with Tris. Mitochondria were permeabilized with addition of alamethicin (20 μg/ml). Alamethicin creates pores that allow equilibration of low-molecular-weight components across the inner membrane, whereas high-molecular-weight proteins are retained in the matrix and intermembrane space (66). Electron microscopy of alamethicin-treated mitochondria reveals no major disruption of membranes (66), and a consensus exists that enzymes, including those involved in ROS production, remain fully functional and have enhanced access to exogenously delivered substrates (33, 66–69). All chemicals were obtained from Sigma.

H₂O₂ release from mitochondria was measured using 5 μm Amplex Red and 0.2 units of HRP in the buffer (excitation/emission, 540:590 nm). The increase in resorufin fluorescence was calibrated by adding H₂O₂ in known concentrations to the buffer containing permeabilized mitochondria. Mitochondrial O₂ consumption was measured continuously by monitoring buffer O₂ content using a fiber optic oxygen sensor FOXY-AL300 (Ocean Optics) (64).

NADH fluorescence was recorded at 366/460 nm excitation/emission wavelengths. NAD⁺-dependent ME2 activity was determined in permeabilized mitochondria by measuring the ability of Mg²⁺ or Mn²⁺ to increase [NADH] in the presence of piericidin, malate, and NAD⁺. The increase in NADH fluorescence was calibrated by adding NADH in known concentration to permeabilized mitochondria in the presence of piericidin.

Statistical analysis

For each data set, the mean and accompanying 95% confidence intervals are reported. The conventional percentile bootstrap-resampling approach with 10,000 replications was used for estimating 95% confidence intervals as well as examining the significant difference between groups (effect size statistics) (70–72). A p value <0.05 was considered statistically significant. All analyses were performed by subroutines for bootstrapping developed in the Python programming language (72).

Author contributions

P. K. and J. N. W. conceived and coordinated the study. P. K., G. C., and S. J. A. performed and analyzed the experiments. P. K., J. N. W., G. C., and S. J. A. wrote the paper. All authors reviewed the results and approved the final version of the manuscript.