Reactive oxygen species are generated by the respiratory complex II – evidence for lack of contribution of the reverse electron flow in complex I



R. Moreno-Sánchez, Instituto Nacional de Cardiología, Departamento de Bioquímica, Juan Badiano No. 1, Sección XVI, Tlalpan DF 14080, Mexico

Fax: + 52 55 5573 0994

Tel: +52 55 5573 2911 ext 1298



Succinate-driven oxidation via complex II (CII) may have a significant contribution towards the high rates of production of reactive oxygen species (ROS) by mitochondria. Here, we show that the CII Q site inhibitor thenoyltrifluoroacetone (TTFA) blocks succinate + rotenone-driven ROS production, whereas the complex III (CIII) Qo inhibitor stigmatellin has no effect, indicating that CII, not CIII, is the ROS-producing site. The complex I (CI) inhibitor rotenone partially reduces the ROS production driven by high succinate levels (5 mm), which is commonly interpreted as being due to inhibition of a reverse electron flow from CII to CI. However, experimental evidence presented here contradicts the model of reverse electron flow. First, ROS levels produced using succinate + rotenone were significantly higher than those produced using glutamate + malate + rotenone. Second, in tumor mitochondria, succinate-driven ROS production was significantly increased (not decreased) by rotenone. Third, in liver mitochondria, rotenone had no effects on succinate-driven ROS production. Fourth, using isolated heart or hepatoma (AS-30D) mitochondria, the CII Qp anti-cancer drug mitochondrially targeted vitamin E succinate (MitoVES) induced elevated ROS production in the presence of low levels of succinate(0.5 mm), but rotenone had no effect. Using sub-mitochondrial particles, the Cu-based anti-cancer drug Casiopeina II-gly enhanced succinate-driven ROS production. Thus, the present results are inconsistent with and question the interpretation of reverse electron flow from CII to CI and the rotenone effect on ROS production supported by succinate oxidation. Instead, a thermodynamically more favorable explanation is that, in the absence of CIII or complex IV (CIV) inhibitors (which, when added, facilitate reverse electron flow by inducing accumulation of ubiquinol, the CI product), the CII redox centers are the major source of succinate-driven ROS production.


Casiopeina II-gly


carbonylcyanide m-chlorophenylhydrazone


respiratory complex I


respiratory complex II


respiratory complex III


respiratory complex IV


diphenylene iodonium


mitochondrially targeted vitamin E succinate


2-oxoglutarate dehydrogenase




reactive oxygen species


sub-mitochondrial particles




It is generally thought that the principal intracellular producer of ROS is the mitochondrial electron transport chain [1]. However, the quantitative contribution of mitochondria to total cellular production of reactive oxygen species (ROS) is still not fully resolved. A study by Boveris et al. [2] in 1972 showed that mitochondria are a major but not the only source of ROS.

The initial ROS generated by the mitochondrial electron transport chain is superoxide O2, and the respiratory complexes I (CI) and III (CIII) are generally considered the main ROS producers under both basal and stress conditions, releasing O2 towards both the mitochondrial matrix and the inter-membrane space. Mitochondrial superoxide is further transformed into other ROS such as H2O2 and OH, catalyzed by superoxide dismutases and Fe2+ in a non-enzymatic reaction, respectively [3].

A common observation is that the rate of H2O2 production is remarkably higher under succinate oxidation than under other conditions where NADH-linked (glutamate/pyruvate) oxidation prevails in isolated rat brain, heart and skeletal muscle mitochondria [4-9] and plant mitochondria [10]. Because H2O2 production driven by succinate oxidation is strongly inhibited by the CI inhibitor rotenone, it is generally believed that succinate oxidation produces ROS via a ‘reverse electron flow’ from complex II (CII) to CI, where O2 and H2O2, are then generated [4-15]. However, the situation whereby oxidation of sub-millimolar (physiological) concentrations of succinate still results in the production of ROS, even in the presence of high concentrations of NAD-dependent substrates, questions the possibility of reverse electron flow. In this last case, NADH-producing substrates drive net downhill flow of electrons through CI to the ubiquinone (Q) pool [12-15], i.e. in the forward direction, such that, under these conditions, the presumed CII→CI reverse electron flow is much more thermodynamically unfavorable.

Moreover, although rotenone strongly blocks ROS production driven by succinate oxidation, ~ 10–15% of the initial ROS remains, which is unlikely to result from catalysis via reverse electron flow at the CI FMNH2 redox center, although ROS may still be derived from the CI flavin domain via the forward reaction in the presence of sufficient NADH (and low NAD+). However, the rotenone-insensitive ROS fraction may be significantly greater (1.3–15-fold) than the level of ROS produced under conditions that maximize the CI forward reaction [4-15]. A plausible explanation for rotenone-insensitive ROS production driven by succinate oxidation is that they emanate from and are directly produced by the CII redox components themselves [16]. It is worth recalling that the CII redox centers are similar to those found in CI. Furthermore, the rotenone-sensitive ROS production driven by succinate oxidation may also be dependent on CII redox components, in that rotenone binding to CI may also indirectly affect (i) Q binding and the electron transfer to and from CII, and/or (ii) the redox state of the Q pool.

Notwithstanding the above, it is surprising to find that the exact role for the contribution of CII to ROS production has been mostly disregarded, and hence has not been experimentally assessed. A relatively low, almost negligible, production of ROS by purified CII has been described [17]. In the present study, experiments were designed to more closely examine the possibility of active ROS production catalyzed by the reverse electron flow from CII to CI. The results rule out this possibility, and are in agreement with a major role for succinate-driven CII-mediated ROS production and the role of CII as a direct contributor to the generation of ROS.

Results and Discussion

Determination of the rate of H2O2 production in isolated mitochondria using Amplex Red involves determination of the rate of H2O2 efflux from the mitochondrial matrix as well as the rate of H2O2 production by the cytosolic phase of the respiratory chain. In both mitochondrial compartments, superoxide is initially formed and immediately transformed to H2O2 by Mn and Cu/Zn superoxide dismutases. In turn, the rate of H2O2 efflux may be affected by the activity of H2O2-consuming mitochondrial matrix enzymes [18]. Nevertheless, quasi-linear relationships of respiratory rate versus ROS production rate were obtained for heart and AS-30D (hepatoma) mitochondria incubated under basal conditions (state 4) and ATP synthesis conditions (state 3), and with various oxidizable substrates (Fig. 1). These results indicated that Amplex Red is a reliable probe for determination of H2O2 production by mitochondria under a wide range of experimental conditions.

Figure 1.

O2 uptake/ROS production relationships in isolated mitochondria. The rates of respiration and ROS production were determined as described in Experimental procedures for heart (A), liver (B) and AS-30D (C) mitochondria. Each experimental point shown represents parallel determination of both respiration and ROS production in the same mitochondrial preparation. The solid lines represent fitting of the experimental points to a straight line. For heart mitochondria, the parameters of the straight lines were slope = 0.00187 ± 0.00019 (mean ± SEM) and correlation coefficient = 0.91 ± 0.09 (mean ± SD; < 0.0001), excluding the two outliers. For liver mitochondria, the fitting parameters were slope = 0.00033 ± 0.00023 and correlation coefficient = 0.28 ± 0.15 (< 0.164). For AS-30D mitochondria, the parameters were slope = 0.00018 ± 0.00014 and correlation coefficient = 0.36 ± 0.02 (< 0.217) for experiments with no rotenone, and slope = 0.00144 ± 0.0002 and correlation coefficient = 0.94 ± 0.04 (< 0.0003), excluding the two outliers, for experiments with rotenone. Abbreviations: GM, 5 mm glutamate plus 5 mm malate for heart and liver mitochondria, or 5 mm glutamate plus 0.1 mm malate for AS-30D mitochondria under state 4 conditions; SR, 5 mm succinate plus 1 μm rotenone for the three types of mitochondria under state 4 conditions; GMS, 5 mm glutamate plus 5 mm malate plus 5 mm succinate for heart and liver mitochondria, or 5 mm glutamate plus 0.1 mm malate plus 5 mm succinate for AS-30D mitochondria under state 4 conditions; SRSt3, 5 mm succinate plus 1 μm rotenone under state 3 conditions, i.e. in the presence of exogenous ADP. The rates of ROS production and respiration are expressed as nmol per min·mg−1 protein and ng atom oxygen per min·mg−1 protein, respectively. The rates of respiration under state 4 conditions for heart and AS-30D mitochondria incubated with succinate without rotenone were similar to those in the presence of the CI inhibitor (data not shown); as determination of the rates of respiration and ROS production under these conditions were not performed in parallel with the same mitochondrial preparations, these data are omitted from the figure.

The data in Fig. 1 also indicate that ROS production accounted for ~ 0.1–0.2% of the total O2 consumption by isolated mitochondria at 37 °C and pH 7.2 in high-potassium medium. A unique linear relationship was obtained for rat heart mitochondria (Fig. 1A). Although there was undoubtedly a direct correlation for heart mitochondria, this relationship was more complex for liver mitochondria, as the scattering of the experimental points did not indicate a straight line (see Fig. 1 legend for statistical analyses). For tumor mitochondria, two relationships were observed, with a much steeper curve being obtained for succinate + rotenone under state 4 and state 3 conditions compared with glutamate + malate, or glutamate + malate + succinate (Fig. 1C; P < 0.001). Thus, the presence of the CI inhibitor induces a shift in the respiration/ROS relationship in AS-30D mitochondria (see below).

Effect of rotenone on succinate-dependent ROS production

The rate of H2O2 production by rat heart mitochondria was markedly increased by adding high levels of succinate (5 mm), whereas ROS production was ~ 10–17 times lower at tenfold lower succinate levels (Fig. 2A,B and Table 1), and was very similar to that induced by oxidation of the NAD-linked substrates glutamate + malate (Fig. 2C,D). Rotenone was a potent inhibitor of ROS production driven by high succinate (Fig. 2B). In contrast, rotenone induced a slight increase in ROS production driven by glutamate + malate (Fig. 2C,D) and did not affect that driven by low succinate (Fig. 2A). Similar effects for rotenone have been reported previously [4, 6, 9, 14]. However, few studies have addressed the rotenone-insensitive ROS production mediated by high succinate, which is similar to or even higher than that displayed in response to either glutamate + malate + rotenone or low succinate + rotenone (Fig. 2B versus Fig. 2A,D, and Table 1, condition 1 versus condition 2).

Table 1. H2O2 production driven by high levels of succinate (5 mm), and glutamate + malate, in mitochondria isolated from various sources. Superscript letters indicate values that are significantly different from each other at < 0.005 using Student's t test for non-paired samples. Values are means ± SD. The number of independent samples assayed is indicated in parentheses. Abbreviations: Ant, antimycin; G, glutamate; M, malate; Mit, mitochondria; Rot, rotenone; Succ, succinate; 5 mm glutamate plus 5 mm malate were used for heart and liver mitochondria, and 5 mm glutamate plus 0.1 mm malate were used for AS-30D mitochondria
 Rate of H2O2 production (nmol min−1 mg−1)
Condition 1
+ Mit0.00 (10)0.04 ± 0.02 (7)0.01 ± 0.02 (4)
+ 5 Succ3.2 ± 0.7 (10)a0.2 ± 0.1 (7)0.07 ± 0.03 (4)i
+ Rot0.8 ± 0.1 (10)a,b,d0.2 ± 0.2 (7)0.3 ± 0.1 (4)i,j
+ Ant1.9 ± 0.4 (7)b0.4 ± 0.3 (7)0.4 ± 0.1 (4)
Condition 2
M + Mit0.00 (11)0.00 (5)0.00 (6)
+ 5 Glut0.15 ± 0.04 (11)c0.1 ± 0.05 (5)0.00 (6)
+ Rot0.5 ± 0.1 (7)c,d0.2 ± 0.1 (5)0.5 ± 0.2 →0.1 ± 0.06j (6)
+ Ant1 ± 0.2 (4)0.3 ± 0.2 (5)0.2 ± 0.04 (5)
Condition 3
MG + Mit0.1 ± 0.1 (12)0.1 ± 0.05 (6)0.03 ± 0.01 (5)
+ 5 Succ2.4 (2)0.4 ± 0.1 (6)0.05 ± 0.02 (5)k
+ Rot1.0 (2)0.5 ± 0.1 (3)0.4 ± 0.2 (5)k
+ Ant2.1 (2)0.6 ± 0.1 (3)0.5 ± 0.3 (4)
Condition 4
MG + Mit0.1 ± 0.01 (5)0.30.0
+ Rot0.4 ± 0.05 (5)e0.60.35 → 0.09
+ 5 Succ1.2 ± 0.1 (4)e,f,h0.80.36
+ Ant2.4 ± 0.1 (4)f0.90.82
Condition 5
Rot + Mit0.1 ± 0.02 (4)g0.1 ± 0.06 (6)0.02 ± 0.03 (4)l
+ 5 Succ0.5 ± 0.1 (4)g,h0.2 ± 0.1 (5)0.3 ± 0.07 (4)l
+ Ant1.7 (2)0.3 ± 0.2 (4)0.4 ± 0.1 (4)
Figure 2.

Rotenone inhibition of H2O2 production driven by succinate oxidation in rat heart mitochondria. Mitochondria (0.25 mg protein·mL−1) were incubated in KMEPi buffer (5 mm potassium phosphate), 4 μm Amplex Red and 5 units·mL−1 horseradish peroxidase at 37 °C and pH 7.2. Cyanide was omitted from all these experiments. The rate of H2O2 production was determined from the change in the absorbance at 573 nm minus 595 nm. (A) Mitochondria were added to KMEPi buffer containing 0.5 mm succinate (0.5 mm S) and 2 μm rotenone (R) and then 2 μm antimycin (A) was added as indicated. (B) Mitochondria were added to KMEPi buffer lacking oxidizable substrate and then 5 mm succinate (5 mm S), rotenone and antimycin were added as indicated. (C) Mitochondria were added to KMEPi buffer containing 5 mm malate (M), and then 5 mm glutamate (G), rotenone, 5 mm succinate and antimycin were added as indicated. (D) Mitochondria were added to KMEPi buffer containing glutamate plus malate (GM). The numbers on the traces indicate the rates of H2O2 production in nmol min−1 mg protein−1, and the bar on the x axis shows the time scale.

Addition of a high succinate concentration after glutamate + malate + rotenone (Fig. 2C), or after addition of rotenone (in the absence of NAD+-linked substrates), significantly increased the rate of H2O2 production by heart and AS-30D mitochondria but not by liver mitochondria (Table 1, condition 2 versus conditions 1, 4 or 5). Mechanistically, it follows that the elevated ROS production induced by succinate ± rotenone may not be attributed to the CI FMNH2 redox site, which is reduced as a result of the reverse electron flow from CII to CI [4-15], but that it more likely involves CIII indirectly and /or CII directly, as recently noted by Pryde & Hirst [19] in experiments with sub-mitochondrial particles (SMPs).

An alternative explanation for succinate-driven ROS production involving the reverse 2-oxoglutarate dehydrogenase (2OGDH) reaction has also been proposed [16]. This requires NADH oxidation and FMN reduction in the flavin site of the dihydrolipoamide dehydrogenase moiety promoted by high succinate, (causing backflow via a build-up of succinyl CoA), and rotenone (inducing accumulation of NADH). However, the results obtained with CII inhibitors and described below rule out this mechanism.

Subsequent addition of antimycin was found to further enhance ROS production to a similar extent when using either low or high succinate, whereas with the CI substrates glutamate + malate, the levels of antimycin-increased ROS produced by heart or AS-30D mitochondria were halved (Fig. 2 and Table 1). This enhanced ROS production induced by antimycin with either CII or CI substrates is most probably derived from the quinone in (Qi) CIII redox site to which antimycin binds. At this stage, the question remained as to which respiratory complex, CII or CIII, was responsible for the ROS produced in the presence of rotenone and the absence of antimycin (and cyanide). To address this last question, stigmatellin was used as a second CIII inhibitor, as it blocks the quinone out (Qo) site to prevent binding of ubiquinol and its initial reduction to semiquinone [20, 21]. No effect of stigmatellin was found for the ROS production induced when high succinate + rotenone or high succinate alone were used, but the increased production of ROS induced by antimycin was inhibited (Fig. 3A,B). This last observation indicates that the ROS production induced by high succinate in the presence of rotenone did not emanate from CIII, but rather originated directly from the CII redox components. It also demonstrates that ROS production by CIII becomes apparent only after adding antimycin, a specific and potent CIII inhibitor of electron transfer and semiquinone oxidation reactions occurring between the two heme groups in cytochrome b, leading to an increase in the half-life of semiquinone at the Qo site on CIII [17, 21, 22].

Figure 3.

Effect of TTFA and stigmatellin on rotenone-insensitive ROS production driven by high succinate levels. Rat heart mitochondria (0.19 mg protein·mL−1) were incubated as described in the legend to Fig. 2. (A) Mitochondria were added to KMEPi buffer lacking oxidizable substrate, and then 5 mm succinate (5 mm S), rotenone and antimycin, or alternatively 5 mm succinate, rotenone, 2 μm stigmatellin (Stig.) and antimycin were added as indicated. (B–D) Mitochondria were added to KMEPi buffer containing 5 mm malate (M), and then 5 mm glutamate (G), rotenone, 5 mm succinate and antimycin were added (black trace, B), or glutamate (G), rotenone plus succinate, stigmatellin and antimycin (green trace, C), or glutamate (G), rotenone plus 25 μm TTFA (TT), succinate and antimycin were added (red trace, D) as indicated.

The rotenone-insensitive ROS production driven by high succinate was completely inhibited by adding either 25 μm TTFA (Fig. 3C), the CII Q redox site inhibitor [23, 24], or 30 μm diphenylene iodonium (DPI), a non-specific flavin-dehydrogenase inhibitor [25, 26] (data not shown). By comparison, 250 μm diazoxide, a vasorelaxing and hypoglycemic drug, only partially inhibited the rotenone-insensitive ROS production (~ 30%) and the rates of state 3 and state 4 respiration (~ 55%) driven by high succinate (data not shown), identifying diazoxide as a weak inhibitor of CII. TTFA at 25 μm and DPI at 30 μm also completely inhibited the ROS production induced by high succinate in the absence of rotenone.

The results for the two CII inhibitors, TTFA (Fig. 3D and data not shown) and malonate (data not shown), which completely inhibited succinate-driven ROS production, do not support involvement of the dihydrolipoamide dehydrogenase flavin in the 2OGDH enzyme complex in this process via a possible reverse reaction. This is because inhibiting CII activity and inducing succinate accumulation would be predicted to stimulate rather than inhibit ROS generation via the reverse 2OGDH-dihydrolipoamide dehydrogenase reaction. DPI may also inhibit the flavin in dihydrolipoamide dehydrogenase, and this is the reason why the DPI effect on CII cannot be grouped together with the TTFA and malonate effects.

Furthermore, at the low concentrations used here, TTFA and DPI did not inhibit the ROS production induced by glutamate + malate + rotenone (data not shown), confirming the specificity of the two drugs for CII under these conditions. In addition, TTFA, DPI and the competitive and potent CII inhibitor malonate at 10.5 mm also blocked the antimycin-induced ROS production when glutamate + malate + rotenone were used (data not shown). This result for malonate inhibition indicates that the source of electrons for CIII was probably CII, fed by succinate formed in the Krebs cycle from glutamate via 2-oxoglutarate [27, 28].

In marked contrast to the results with heart mitochondria, rotenone did not abolish the ROS production driven by succinate oxidation in AS-30D mitochondria, but instead did the opposite, inducing a significant stimulation (Table 1, conditions 1 and 5; see also Fig. 1C). This result casts considerable doubt on the possibility of a reverse electron flow from CII to CI as the principal cause for the elevated ROS production that occurs when succinate is added to AS-30D mitochondria. Furthermore, adding rotenone to AS-30D mitochondria in the presence of glutamate + malate also initially promoted significant ROS production, which was attenuated after a few minutes (Table 1, condition 2). Although these findings imply a possible use for therapeutic targeting, the nature of the dual pattern of ROS production (activation then inhibition) induced by rotenone in AS-30D mitochondria requires further study. In this regard, it was previously shown that CI exerts strong flux control over oxidative phosphorylation in intact AS-30D cells [29], further supporting its potential therapeutic value as a target for anti-cancer treatment.

Succinate-dependent ROS production in the absence of rotenone

Succinate at concentrations of 5 mm has been typically used in the majority of bioenergetic studies in which isolated mitochondria have been used. In contrast, physiological succinate levels have been shown to oscillate in the 0.1–1 mm range in blood during exercise, in brain tissue and in fibroblast cells [26, 27, 30, 31]. To more closely approximate physiological ROS production levels driven by succinate, and hence by CII, a lower succinate concentration is required. However, when using heart mitochondria (Fig. 2A) or AS-30D mitochondria (Table 2), 0.5 mm succinate was unable to promote the increased levels of ROS production obtained when 5 mm succinate was used.

Table 2. Effect of the triphenylphosphonium-tagged vitamin E analog MitoVES (MV11) on the rate of H2O2 production driven by succinate and glutamate + malate in rat heart, liver and AS-30D mitochondria. Superscript letters indicate values that are significantly different from each other at < 0.005 using Student's t test for non-paired samples. Values are means ± SD. The number of independent samples assayed is indicated in parentheses. Abbreviations: Ant, antimycin; G, glutamate; M, malate; Mit, mitochondria; Rot, rotenone; Succ, succinate; MV11, MitoVES; 5 mm glutamate plus 5 mm malate were used for heart and liver mitochondria, and 5 mm glutamate plus 0.1 mm malate were used for AS-30D mitochondria. ND, not determined.
 Rate of H2O2 production (nmol min−1 mg−1)
Condition 1
0.5 Succ + Mit0.3 ± 0.08 (4)a0.3 ± 0.2 (4)0.03 (1)
+ 50 MV113.0 ± 0.8 (4)a,b,d0.6 ± 0.2 (4)0.3 (1)
+ Rot2.6 ± 0.6 (4)0.5 ± 0.3 (3)0.4 (1)
+ Ant5.4 ± 0.6 (4)b0.7 ± 0.5 (3)0.5 (1)
Condition 2
5 Succ + Mit2.8 ± 0.4 (3)0.2 ± 0.1 (5)0.09 ± 0.04 (4)i
+ 50 MV112.3 ± 0.6 (3)0.3 ± 0.1 (5)0.5 ± 0.1 (4)i,k
+ Rot2.3 ± 0.6 (3)0.4 ± 0.1 (5)0.5 ± 0.1 (4)
+ Ant4.2 ± 0.8 (3)0.5 ± 0.2 (5)0.4 (2)
Condition 3
GM + Mit0.1 ± 0.1 (5)c0.2 ± 0.1 (4)0.00 (4)j
+ 50 MV111.1 ± 0.2 (5)c,d,f0.2 ± 0.04 (4)0.2 ± 0.06 (4)j,k
+ Rot1.1 ± 0.3 (5)0.2 ± 0.07 (4)0.2 ± 0.1 (4)
+ Ant1.2 ± 0.3 (5)0.3 ± 0.1 (4)0.1 (2)
Condition 4
GM, 0.5 Succ + Mit0.2 ± 0.04 (4)eNDND
+ 50 MV112.8 ± 0.4 (4)e,f  
+ Rot2.9 ± 0.5 (4)  
+ Ant5.3 ± 0.8 (4)  
Condition 5
0.5 Succ + Mit0.3 ± 0.2 (3)g0.3 ± 0.2 (5)0.05 (2)
+ 10 MV111.0 ± 0.1 (3)g0.5 ± 0.2 (3)0.27 (2)
Condition 6
GM, 0.5 Succ + Mit0.3 ± 0.03 (4)h0.35 (2)0.04
+ 10 MV111.5 ± 0.3 (4)h0.39 (2)0.39
Condition 7
5 Succ + Mit2.00.18 (2)0.08 ± 0.03 (4)
+ 10 MV111.10.26 (2)0.2 ± 0.07 (4)

Although it may appear that mitochondrial ROS production induced by 5 mm succinate is not physiologically relevant, situations do exist where such levels might be reached. For example, in Cowden syndrome (a pathology that predisposes to breast and thyroid cancers) and in several neoplastic pathologies such as paragangliomas and phaechromocytomas, mutations in succinate dehydrogenase (SDHB–D) occur [32-34], promoting a significant increase (11–14-fold) in the levels of succinate in the affected tissues [35] or plasma [36]. In all these diseases, the higher succinate levels are correlated with the presence of higher lipid peroxidation and a substantial increase in ROS production [33, 36].

Mitochondrially targeted vitamin E succinate, MitoVES, an anti-cancer drug generated by tagging vitamin E succinate with a triphenylphosphonium group, behaves as a classical uncoupler; it also preferentially locates and binds to the inner mitochodrial membrane near the matrix face, interacting with the CII quinone positive (QP) site to block electron transfer, and induces apoptosis of cancer cells by increasing oxidative stress [37-39]. Hence, to further test the proposal that succinate-driven ROS production does not derive from the reverse electron flow from CII to CI, but rather stems from CII, MitoVES was assayed for its effects on ROS production in the three types of isolated mitochondria used in the present study. Addition of MitoVES to heart mitochondria in the presence of 0.5 mm succinate promoted a marked (tenfold) increase in ROS production, significantly lower that that obtained for glutamate + malate (Fig. 4 and Table 2). A similar MitoVES response pattern was obtained in AS-30D mitochondria in the presence of 5 mm succinate, but liver mitochondria were much less susceptible to the agent (Table 2). In heart mitochondria in the presence of 5 mm succinate, MitoVES was not able to increase ROS production but instead induced a slight decrease (Table 2).

Figure 4.

Effect of MitoVES on succinate-driven ROS production. Rat heart mitochondria [0.21 mg protein·mL−1, except for (B), which contained 0.26 mg protein·mL−1] were incubated as described in the legend to Fig. 2. Other additions were 10 or 50 μm MitoVES (MV11).

MitoVES-induced ROS production was not affected by rotenone when either 0.5 or 5 mm succinate was used with heart or AS-30D mitochondria (Table 2), again indicating that reverse electron flow from CII to CI is not involved. Addition of antimycin further enhanced the ROS levels induced by MitoVES (Fig. 4 and Table 2), whereas stigmatellin did not alter the rate of ROS production induced by MitoVES but blocked the antimycin effect (data not shown). These results implicate CII itself as the main site for succinate-induced ROS production in the presence of MitoVES. TTFA at 25 μm only partially attenuated the MitoVES-induced ROS production (data not shown). As MitoVES increased ROS levels and TTFA did not affect the MitoVES-induced ROS production driven by succinate, it would be interesting to analyze the nature of the interaction of these two drugs with the CII Q binding site in terms of probable competitive binding against each other and/or the natural substrate (ubiquinol).

The high levels of ROS production driven by high levels of succinate (5 mm) in heart mitochondria were similarly decreased when either rotenone (see Fig. 2B) or the uncoupler carbonylcyanide m-chlorophenylhydrazone (CCCP) were added (data not shown). In the latter reaction, as expected, the rate of respiration was also immediately and drastically decreased by CCCP (data not shown). These observations are explained by the severe inhibition of CII activity induced by rapid accumulation of oxaloacetate, which occurs under uncoupling conditions [40, 41]. The Krebs cycle intermediate metabolite oxaloacetate is a potent tightly bound competitive inhibitor of CII succinate dehydrogenase activity, and therefore inhibits succinate binding to CII. On the other hand, in the presence of rotenone, oxidation of succinate is not inhibited because oxaloacetate is not accumulated when electron transfer and H+ pumping by the respiratory chain is accelerated by CCCP.

The ROS production driven by high levels of succinate and inhibited by CCCP in the absence of rotenone was significantly stimulated by stigmatellin (2.9 ± 0.6-fold, = 3, < 0.001). However, with succinate plus rotenone, CCCP addition had no effect on the levels of ROS production, which remained significant, and further addition of stigmatellin slightly stimulated ROS production (30 ± 5% increase, = 3, < 0.01). In turn, ROS production driven by glutamate + malate + rotenone was slightly stimulated by CCCP (~ 40%), and was not affected by stigmatellin. Thus, for succinate ± rotenone in the presence of an uncoupler, ROS production stimulated by stigmatellin cannot be derived from the reverse electron transfer from CII to CI, because the H+ gradient has collapsed. Under such conditions, it is very unlikely that the energy barrier required to favor the reverse reaction, reducing NAD+ to form NADH, may be overcome. Therefore, the data indicate that CII is the site of succinate-induced ROS production as well as the increased ROS production that occurs when stigmatellin inhibits ubiquinol oxidation by CIII in uncoupled heart mitochondria.

In experiments with bovine heart SMPs, it was also observed that subsequent addition of rotenone after succinate blocked succinate-driven ROS production by 34 ± 14% (= 5) (Fig. 5A). It should be noted that no pyridine nucleotides were present in this experimental model, and no reverse electron flow was achieved under these conditions (well-coupled SMPs with respiratory ratios of 1.6–2.2, sucrose-based medium, 10 mm succinate, 1 mm NAD+, 10 mm MgCl2, 37 °C). Addition of ATP did not prompt the reverse electron flow nor did it increase ROS production. Only in the presence of cyanide (plus succinate and MgATP [42-44]) was a reverse electron flow from succinate to NAD+ of 10–30 nmol·min−1·mg−1 detectable, together with increased ROS production of ~ 0.1 nmol·min−1·mg−1.

Figure 5.

Effect of Casiopeina II-gly on succinate-driven ROS production in bovine heart sub-mitochondrial particles. SMPs (0.41 mg protein·mL−1) were incubated in a buffer containing 250 mm sucrose, 10 mm Hepes, 1 mm EGTA, pH 7.3, 5 mm potassium phosphate, 10 mm MgCl2, 5 units·mL−1 horseradish peroxidase and 4 μm Amplex Red at 37 °C. Other additions were 10 mm succinate (10 mm S), 2 μm rotenone (R), 2 μm antimycin (A), and the indicated micromolar concentrations of Casiopeina II-gly (Cas). The addition of rotenone or Casiopeina is indicated by the larger third arrow, whereas addition of antimycin is indicated by the fourth arrow. For clarity, further additions of rotenone and antimycin after Cas are omitted.

The copper-based anti-cancer drug Casiopeina II-gly (CasII-gly), a lipophilic cation similar to MitoVES, targets mitochondria by virtue of its delocalized net positive charge and hydrophobic chemical nature [45]. In turn, the transition metal in CasII-gly preferentially interacts with the pyruvate dehydrogenase and 2OGDH complexes and the CII sulfhydryl groups, which are its main targets in mitochondria [46, 47]. Therefore, it was judged relevant to the aim of the present study to also test this anti-cancer drug for its effects on CII and ROS production. CasII-gly had negligible effects on succinate-driven ROS production by heart and AS-30D mitochondria. However, when SMPs were treated with 50 μm CasII-gly, a significant transient stimulatory effect on succinate-driven ROS production resulted, increasing from 0.21 ± 0.09 (= 6) to 0.84 ± 0.2 nmol min−1·mg−1 (= 3) at 50 μm (< 0.005) (Fig. 5). This transient increase in ROS was followed by a steady decline in ROS production, presumably due to accumulated formation of adducts between the drug CasII-gly and reactive cysteines in target sites. MitoVES at 50 μm also increased ROS production but by a lower magnitude (~ 0.1 nmol·min−1·mg−1, = 3).

The CasII-gly increased ROS production was only partially inhibited by rotenone (by 21%, data not shown). CasII-gly was probably able to more specifically interact with CII to accelerate ROS production in SMPs because of the absence of NAD+-linked enzymes under these conditions, as they are removed during preparation, and the absence of the inner membrane barrier. CasII-gly did not affect the increase in ROS production induced by antimycin (Fig. 5A), indicating that this anti-cancer drug does not affect CIII, but may preclude the flow of electrons from CII to CIII.

Concluding remarks

The inhibitory effects of rotenone on succinate-driven ROS production described here for heart mitochondria and SMPs, but not for liver and AS-30D mitochondria, are consistent with data from skeletal muscle and brain mitochondria [4-9]. These observations have been the main support for the claim that reverse electron flow from CII to CI exists and is responsible for generation of ROS in the mitochondrial electron transport chain. Such an interpretation does not take into account the highly unfavorable thermodynamic barrier for electron transfer from ubiquinol to NAD+G°′, ~ +54–63.5 kJ·mol−1 [44, 48]) and the fact that rotenone-sensitive succinate-driven ROS production has generally been examined only in the absence of inhibitors of complexes III and IV. Indeed, it is possible for the H+ electrochemical gradient (inline image) across the inner membrane to offset the thermodynamic requirement to reverse the CI electron transfer reaction, reaching near equilibrium. For a total H+ gradient of 0.13–0.18 V (i.e. ΔG°′ = ~ 12.5–17.3 kJ mol−1), the inline image may actually supply the energy required to take the reaction of the CI reverse electron transfer to near equilibrium, assuming a H+/2e stoichiometry of 4 for CI [49]. However, for a inline image of lower magnitude and/or a lower H+/2e stoichiometry of 3 for CI [50, 51], inline image is not sufficient to allow reverse electron transfer to occur.

To overcome the energy barrier required to favor the CI reverse reaction and attain net reverse electron transfer from CII to CI, high concentrations of succinate and NAD+, low concentrations of fumarate and NADH, and the presence of cyanide [or any other complex IV (CIV) or CIII blocker] to induce high ubiquinol accumulation, are all required, in addition to ATP for H+ gradient generation [42-44]. These physicochemical conditions have no physiological relevance, i.e. succinate is usually low and the NADH level is not negligible. Moreover, there are no ‘respiratory inhibitors’ such as cyanide present, and hence the H+ gradient is generated by the respiratory chain forward reaction and ATP is also unlikely to be hydrolyzed by the ATP synthase. It therefore follows that simple addition of high succinate levels to intact mitochondria is not suffice to drive the CI reverse reaction because of the significant NADH and fumarate levels, and the low ubiquinol/ubiquinone ratio will instead favor the CI forward reaction. Therefore, a more thermodynamically favorable explanation for succinate-driven ROS production when no inhibitors of CIII or CIV are present is that it is the CII redox sites themselves that actively generate ROS directly. The overall results of the present study using rotenone, antimycin, TTFA, stigmatellin, CCCP, MitoVES and CasII-gly with heart mitochondria, AS-30D mitochondria and SMPs clearly validate this proposal. Consequently, it follows that the rotenone inhibitory effect on the succinate-driven ROS production may be due to either (i) rotenone directly affecting and modifying the binding properties of the CII Q site to decrease ROS production, or (ii) the ensuing accumulation of ubiquinone caused by rotenone binding near the CI Q site, leading to increased oxidized levels of the Q pool to which CII is also linked; this situation may promote faster release of ubiquinol from the CII Q site, thereby decreasing the levels of ROS production from the CII Q sites.

In a recent report, reverse electron flow within the CII neck region between the proximal Qp binding site and the FAD group was proposed to mediate ROS production by CII [50]. Thus, when CI and CIII were inhibited using rotenone and myxothiazol, respectively, and the succinate concentrations used were low, CII in rat skeletal muscle mitochondria was shown to generate ROS at high rates, approaching or exceeding the maximum rates achieved by CI or CIII. In addition, it was shown that CII generates these ROS by both the forward reaction, using electrons supplied by succinate, and a reverse reaction to fumarate, using electrons supplied from the reduced Q pool. ROS production in the reverse direction was prevented by inhibition of CII at either the Q binding sites (using atpenin A5) or the flavin (using malonate), whereas ROS production in the forward reaction was prevented by malonate but not by atpenin A5. Based on their results, the authors proposed that the ROS from CII was derived preferentially from the flavin site (referred to as site IIF) [50]. Furthermore, they proposed a mechanism for this ROS production by CII based upon occupancy of the substrate oxidation site and the reduction state of the enzyme.

As indicated by the results presented here using the anti-cancer drugs CaslI-gly and MitoVES, the actions of TTFA and atpenin A5 on the CII Q sites as potent inhibitors of ROS production may be clearly distinguished from those of α-tocopheryl succinate and MitoVES, which bind to the same CII Q sites but promote ROS production, particularly in cancer cells [37, 38, 51-53]. The precise mechanism of CasII-gly-induced ROS production is not yet clear, but, given its promotion of succinate-driven ROS production, this drug may also interact with the CII Q sites. In this regard, the synthetic retinoid N- (4-hydroxyphenyl) retinamide also induces apoptosis of human cervical carcinoma C33A cells by a ROS-mediated process that is partially rotenone-inhibited and fully blocked by TTFA [54], suggesting interaction with CII.

In contrast, carboxins such as TTFA and atpenins, which, with their high binding affinities, are potent inhibitors of the Q sites in CII, surprisingly all show lethality in rats and mice, with LD50 values in the range 10–100 mg·kg−1 [55]. This probably relates to the mechanism of action of carboxins as strong mixed-type inhibitors of the Q sites in CII (suppressing the succinate ubiquinone oxidoreductase activity), although atpenins are not capable of fully blocking electron transfer from the iron–sulfur centers in the succinate dehydrogenase B subunit of CII to water-soluble dyes such as 2,6-dichlorophenol-indophenol, a process that is considered to occur via a non-physiological electron transfer pathway [[56, 57],]. On the other hand, drugs such as α-tocopheryl succinate, MitoVES and CasII-gly are selective for cancer cells and do not show the same level of toxicity, having been successfully used to treat murine models of cancer [37, 38, 43, 45, 52]. Further studies into the mechanisms of ROS production at the CII redox sites at the molecular level are required to characterize the activities of the various drugs described here, and their particular suitability and efficacy as beneficial anti-cancer chemotherapeutic agents.

Experimental procedures

SMPs and mitochondria

Tightly coupled rat heart [57], liver [58] and AS-30D mitochondria [59] were obtained as previously described. Tightly coupled bovine heart SMPs were prepared by sonicating mitochondrial suspensions in the presence of 20 mm succinate [60].

Experimental procedures for determination of energy-dependent mitochondrial functions

The rate of respiration of isolated mitochondria (1 mg protein·mL−1) was determined using a Clark-type O2 electrode in an air-saturated medium containing 120 mm KCl, 20 mm MOPS, 0.5 mm EGTA, 5 mm (heart) or 2 mm (liver/AS-30D) potassium phosphate, pH 7.2 (KMEPi buffer) and the indicated oxidizable substrate at 37 °C. Respiration of SMPs was assessed using 250 mm sucrose, 10 mm HEPES, 1 mm EGTA (SHE buffer) containing 10 mm MgCl2, pH 7.5, and the indicated oxidizable substrate at 37 °C. The O2 solubility at 2240 m altitude (Mexico City) was 400 and 390 ng oxygen atoms per ml at 37 °C in the KMEPi and SHE buffers, respectively.

The rate of H2O2 production was assessed using Amplex Red (Invitrogen, Grand Island, NY, USA) in the presence of an excess of peroxidase [61, 62]. Mitochondria at 0.2–0.45 mg protein·mL−1 were incubated under gentle O2 gassing at 37 °C in KMEPi buffer with 4–5 μm Amplex Red and 5 units (μmol min−1)·mL−1 horseradish peroxidase (Sigma, St Louis, MO, USA); the absorbance difference at 573 minus 595 nm was determined using a dual-beam spectrophotometer (2501PC; Shimadzu, Kyoto, Japan) over a range of wavelengths, at which the probe absorbance response is close to linear and ROS production is not under-estimated. Under these conditions, the oxidation of Amplex Red to resorufin yielded an absorbance difference signal that was a linear function of the amount of H2O2 added from 0.1–5 nmol. Due to direct perturbation of the Amplex Red absorbance difference signal by a number of compounds such as cyanide and NADH (and the non-enzymatic H2O2 production induced by strong reductants such as NADH and glutathione [61]), freshly prepared concentrated H2O2 stock solutions, calibrated at 248 nm using an extinction coefficient of 43.6 m−1·cm−1, were routinely used to calibrate the probe signal.


The present work was partially supported by grant numbers 80534, 107183, 123636 and 180322 from the Consejo Nacional de Ciencia y Tecnología México and grant number PICS08-5 from the Instituto de Ciencia y Tecnología del Distrito Federal.