Changes in mitochondrial superoxide and hydrogen peroxide production may contribute to various pathologies, and even aging, given that over time and in certain conditions, they damage macromolecules and disrupt normal redox signalling. Mitochondria-targeted antioxidants such as mitoQ, mitoVitE, and mitoTEMPO have opened up the study of the importance of altered mitochondrial matrix superoxide/hydrogen peroxide in disease. However, the use of such tools has caveats and they are unable to distinguish precise sites of production within the reactions of substrate oxidation and the electron transport chain. S1QELs are specific small-molecule Suppressors of site IQElectron Leak and S3QELs are specific small-molecule Suppressors of site IIIQoElectron Leak; they prevent superoxide/hydrogen production at specific sites without affecting electron transport or oxidative phosphorylation. We discuss the benefits of using S1QELs and S3QELs as opposed to mitochondria-targeted antioxidants, mitochondrial poisons, and genetic manipulation. We summarise pathologies in which site IQ in mitochondrial complex I and site IIIQo in mitochondrial complex III have been implicated using S1QELs and S3QELs.
There is an ever-increasing literature that associates mitochondrial reactive oxygen species (ROS) in either a causal or correlative manner to almost any pathology [1–5]. Long before mitochondria were shown to generate superoxide and hydrogen peroxide, Harman proposed the ‘free radical theory of aging’, citing the deleterious effects that free radicals can have on biological systems . After work by Jensen  and Chance [8,9] showed that the mitochondrial electron transport chain can generate hydrogen peroxide, Harman's theory was modified to the ‘mitochondrial theory of aging’ . Since then, the importance of superoxide and hydrogen peroxide in redox signalling pathways has increasingly been recognised [1,11]. There have been a plethora of studies on mitochondrial superoxide and hydrogen peroxide, from understanding how they drive disease pathologies to exploring their involvement in aging and longevity. This interest has generated a number of methods, probes, and genetic tools with which to measure ‘ROS’, each having its own caveats and interpretations . It is agreed that mitochondria generate ROS that can cause damage and contribute to redox signalling, and therefore drive pathologies. Here we discuss the ability to move beyond mitochondrial ROS as a collective concept, and begin to tease apart the importance of specific sites of mitochondrial superoxide and hydrogen peroxide generation and how individual sites may be driving pathologies.
Mitochondrial sites of superoxide and hydrogen peroxide generation
Mitochondria and NADPH oxidases collectively contribute to most of the measurable total cellular hydrogen peroxide production . Mitochondrial superoxide (O2−) and hydrogen peroxide (H2O2) are the ROS most commonly considered in research. They are produced by electron leak from donor redox centres in the mitochondrial substrate oxidation and electron transport systems. This leak occurs by either one-electron reduction of molecular oxygen to form O2− or two-electron reduction of molecular oxygen to form H2O2 [14,15]. At least 11 sites within mammalian mitochondria have been identified to produce O2− and/or H2O2 at a significant rate during substrate catabolism, electron transport and oxidative phosphorylation [1,16,17]. These 11 sites exhibit distinct biochemical properties and can be divided into two sub-sets: either they contain a flavin group (F), or a ubiquinone-binding site (Q). The flavin sites consist of 2-oxoglutarate (OF), branched chain 2-oxoacid (BF), 2-oxoadipate (AF) and pyruvate (PF) dehydrogenase complexes, and sites IF and IIF of mitochondrial complex I and complex II, which (except for sites IIF and EF) operate at the redox potential of the NADH/NAD+ isopotential pool. The ubiquinone-binding sites, which operate at the redox potential of the ubiquinol/ubiquinone isopotential pool, consist of the ubiquinone-binding sites of mitochondrial complex I (IQ), complex III (IIIQo), glycerol-3-phosphate dehydrogenase (GQ) and dihydroorotate dehydrogenase (DQ). The last site, EF, is a flavin site of the electron-transferring flavoprotein/electron-transferring flavoprotein-ubiquinone oxidoreductase (ETF/ETF:QOR) system [1,18]. Sites IIIQo, IQ and IIF have the largest potential capacity to produce O2−/H2O2 in non-physiological conditions in rat skeletal muscle mitochondria in vitro [1,18], and also make the greatest actual contributions to mitochondrial and cellular O2−/H2O2 production in more physiological conditions in rat skeletal muscle mitochondria, where they and site IF each contribute 15–25% of the total . Sites IIIQo and IQ together contribute at least 75% of mitochondrial H2O2 release from C2C12 mytotubes in vitro . It is, therefore, reasonable to propose that these sites play a crucial role in disease pathologies and could be cleaner and more direct therapeutic targets to decrease mitochondrial O2−/H2O2 production to re-establish ‘normal’ redox signalling and decrease pathological damage.
Why are site-specific suppressors of superoxide and hydrogen peroxide generation preferable to non-specific mitochondria-targeted antioxidants?
Studies using mitochondria-targeted antioxidants, including mitoQ, SkQ, mitoTEMPO and mitoVitE, have made significant contributions to our understanding of the importance of mitochondrial matrix O2−/H2O2 in different disease models [20–29]. MitoQ and SkQ contain quinone moieties, whereas mitoTEMPO contains TEMPO, a superoxide dismutase and catalase mimetic , and mitoVitE contains the antioxidant vitamin E, each attached to a lipophilic triphenylphosphonium (TPP) cation . Conjugation of antioxidants to TPP drives their mitochondrial targeting, taking advantage of the highly negative-inside mitochondrial membrane potential, causing TPP and its cargo to accumulate in the mitochondrial matrix [26,31]. Major downsides to these mitochondria-targeted antioxidants are that they rely for their targeting on the high mitochondrial membrane potential, which may be altered in disease states, and even worse, they can act as prooxidants . In addition, they are unable to distinguish precise sites of O2−/H2O2 production within the reactions of substrate oxidation and the electron transport chain.
Mitochondria-targeted antioxidants elicit beneficial effects on biological systems mostly through radical scavenging (e.g. mitoTEMPO and mitoVitE) or lipid peroxidation chain-terminating activities (mitoQ) . These activities change the overall outcome of O2−/H2O2 production in mitochondria, and therefore in the cell. Such non-specific dampening of all effects of O2−/H2O2 could be problematic, given the importance of mitochondrial O2− and H2O2 in redox signalling. Specific suppression of O2−/H2O2 generation from particular mitochondrial sites may be a more effective and preferred mechanism to maintain mitochondrial redox balance without disrupting other potentially important activities.
Although it is known that several sites of O2−/H2O2 generation from mitochondria exist, progress is only recently underway in understanding how these sites may be independent drivers of pathology. In part this has been due to the lack of good investigative tools, resulting in many studies having to use mitochondrial poisons and genetic manipulations to try to assess the contributions of different sites. Unfortunately, these approaches are inevitably confounded by strong bioenergetic effects, rendering any effects hard to interpret. In order to avoid these non-desirable attributes and explore specific mitochondrial sites of O2−/H2O2 generation, Brand and colleagues identified specific suppressors of site IQ electron leak (S1QELs) and specific suppressors of site IIIQo electron leak (S3QELs), the two sites that have the largest contribution of O2−/H2O2 to the mitochondrial matrix [32,33]. These compounds were identified using isolated mitochondria in plate-based assays . They were selected for their specificity for their respective site, and for their lack of any inhibition of respiration at up to 50-fold the concentrations that suppress O2−/H2O2 generation.
There are distinct advantages of using S1QELs and S3QELs over non-specific mitochondria-targeted antioxidants. These include: (A) they inhibit electron leak selectively without inhibiting normal electron flow or oxidative phosphorylation or having any other known cellular targets (when titrated carefully), (B) they do not participate in redox reactions, are not consumed, and do not require redox recycling, so they remain highly potent regardless of the extent of an oxidative insult, (C) due to their high selectivity, other important sites of redox signalling are unaffected, and (D) published S1QELs and S3QELs are the first and currently best compounds of their kind to enable the study of potential site-specific O2−/H2O2 generation in driving and exacerbating disease pathologies (Figure 1). The physiological and pathological situations in which S1QELs and S3QELs have been shown to have effects are discussed in the following sections.
Pathologies driven by superoxide/hydrogen peroxide production from mitochondrial sites IQ and IIIQo determined using S1QELs and S3QELs.
Ischemia-reperfusion injury is caused when there is a block in blood flow to tissue for a substantial amount of time (ischemia), and then the blood flow is restored (reperfusion). Although extended ischemia will result in the death of the tissue, it is the reperfusion of oxygenated blood that results in long-lasting damage. Reperfusion injury has long been associated with a burst of mitochondrial ROS [35,36] in several different tissues including heart, brain, liver and kidney [37–40]. Inhibitors of electron transport through mitochondrial complex I protect in models of ischemia-reperfusion injury, which led to the assumption that the mitochondrial ROS burst was predominantly from complex I [41–43]. However, inhibition at one site will alter electron flow and affect O2−/H2O2 generation at other sites , so this conclusion is not warranted, and in addition, these findings are deeply confounded by the bioenergetic inhibition afforded by these inhibitors. Nonetheless, conditions that support the generation of O2−/H2O2 from site IQ of complex I, particularly a highly reduced coenzyme Q pool and high protonmotive force to drive reverse electron transport [44,45], are present during reperfusion. For reverse electron transport to occur once reperfusion begins, an accumulated electron donor must be present. Succinate was found to accumulate more than any other mitochondrial metabolite during ischemia . This is consistent with how reverse electron transport in complex I is driven in isolated mitochondria, by using succinate as a substrate . Inhibition of succinate dehydrogenase using added dimethyl malonate was protective in ischemia-reperfusion , supporting a role for succinate reoxidation. In this model of ischemia-reperfusion injury, succinate accumulates during ischemia. Upon reperfusion it is rapidly oxidised by mitochondrial succinate dehydrogenase to reduce the Q pool, forcing reverse electron transport into complex I resulting in production of O2−/H2O2 from site IQ , and subsequent injury.
The use of S1QELs has specifically implicated site IQ of complex I in ischemia-reperfusion injury in the Langendorff-perfused mouse heart model. The inclusion of S1QEL1.1 at the time of reperfusion dramatically improved post-ischemic recovery of cardiac function, and resulted in a significant decrease in infarct size . OP2113, a less potent IQ suppressor (IC50in vitro 10 µM  compared with 0.07 µM for S1QEL1.1 ) also afforded protection in the Langendorff-perfused mouse heart model . OP2113 improved post-ischemic recovery and decreased infarct size. Together these studies show that during ischemia-reperfusion, reverse electron transport drives site IQ in complex I causing a burst in O2−/H2O2 production from site IQ. Therefore, selectively suppressing O2−/H2O2 production from site IQ is a promising therapeutic target in ischemia-reperfusion injury.
Cell viability and death
Mitochondria are an integral part of the intrinsic pathway of programmed cell death . Pro-apoptotic factors drive pathways that culminate in the activation of the mitochondrial permeability transition pore (mPTP), an inner membrane protein complex that forms a non-selective channel. mPTP opening collapses the protonmotive force and prevents oxidative phosphorylation. Prolonged mPTP opening releases intermembrane proteins including cytochrome c, triggering downstream apoptotic signalling .
Treatment with rotenone or metformin can prevent mPTP opening, leading to speculation that mitochondrial O2− and/or H2O2 induce the mPTP. However, although both agents can inhibit reverse electron transport thereby inhibiting site IQ, they also inhibit oxidative phosphorylation [52,53]. More persuasively, overexpression of superoxide dismutase or treatment with mitoTEMPO, which lower matrix superoxide levels, also prevented mPTP opening, suggesting that matrix superoxide plays a role . Imeglimin is an oral glucose-lowering drug that targets type 2 diabetic symptoms and appears to be a weak S1QEL . Imeglimin can inhibit O2−/H2O2 production from complex I without inhibiting respiration, at least at 100 µM. Human endothelial cells (HMEC-1) cultured in high glucose and treated with imeglimin were protected from hyperglycemia-induced cell death, perhaps through inhibition of the mPTP by suppressing generation of O2−/H2O2 at site IQ .
Tunicamycin can be applied as an extrinsic insult to induce endoplasmic reticulum (ER) stress and apoptosis. S3QELs protected against tunicamycin-induced cleavage of caspase 3 and 7 in a rat INS-1 insulinoma cell line , consistent with a role for superoxide generation at site IIIQo of complex III in driving oxidative stress. S3QEL treatment also increased survival and function of isolated primary pancreatic islets more strongly than did EUK-134, a superoxide dismutase mimetic, indicating that site-specific suppression may be more beneficial than non-specific antioxidant protection . Corroborating these results, another group used S1QELs and S3QELs to show that culturing INS-1 cells in high glucose caused a significant increase in superoxide generation at site IIIQo, but not at site IQ , although this interpretation has caveats due to the use of mitoSOX as an indicator of mitochondrial oxidant levels and the reliability of this probe . Both S1QELs and S3QELs protected embryonic cardiomyocyte H9C2 cells treated with tunicamycin against caspase 3 and 7 cleavage . Taken together, these results suggest that O2−/H2O2 production from sites IQ and IIIQo play a role in determining apoptosis, and that the involvement of each site may be different in different cell types.
Airborne particulates are a complex heterogeneous mixture with properties that can vary between sources, time of year, and atmospheric conditions. It is accepted that the most harmful particles are those with a diameter <2.5 µm. At this size, the particulates are able to enter the lung, deposit in the alveolae and enter the systemic circulation . Inhalation of particulate matter can increase the incidence of chronic obstructive pulmonary disease, asthma, pneumonia, bronchitis, and emphysema [58,59]. There is an accepted association of particulate matter with oxidative stress in the lungs and in other organs. It is thought to be due to disruption in endogenous redox signalling and/or exacerbation of endogenous sources of ROS. Although there is no direct evidence in vivo showing an increase in mitochondrial O2−/H2O2, as this is difficult to measure, changes in mitochondrial morphology and number have been documented .
In a murine alveolar macrophage cell line (MHS) incubated with particulate matter, there was an increase in mitochondrial oxidants measured using mitoSOX, and S3QELs significantly ameliorated this increase. Likewise, S3QEL treatment inhibited IL-6 production. These effects were not seen with S1QEL treatment, suggesting that particulate matter specifically induces superoxide generation from site IIIQo in these cells to induce an inflammatory response .
Antigen presentation to CD8+ T-cells is crucial for their activation and for defence against intracellular pathogens such as bacteria and viruses, as well as defence against tumour formation . The antigen-presenting cells most effective at cross-presenting antigens to CD8+ T-cells are the conventional dendritic cells (cDC1) . Plasmacytoid dendritic cells (pDCs) also have the capability to present antigen, though this function is highly regulated in this cell type and requires activation by Toll-like receptors (TLR activation). The intracellular pathways that support the ability of pDCs to present antigens are not fully understood, however, superoxide generation from site IIIQo has been implicated as an important factor [64,65]. Activation of pDCs caused an increase in mitochondrial oxidants measured by mitoSOX, and this increase was inhibited by incubation with a S3QEL. Concurrently, treatment with S3QEL suppressed the cross-presentation capacity of activated pDCs to elicit a clonal CD8+ T cell expansion response . These results suggest an important role of superoxide generation by site IIIQo in the activation of antigen presentation by pDCs. Such information about the site-specificity of superoxide production is lost when using non-specific antioxidants.
Cellular signalling pathways
There is a wealth of literature describing how mitochondrial superoxide and hydrogen peroxide act as messengers in redox signalling through different pathways. The ability of mitochondrial ROS to act as messengers is accepted in most cases, however, there are some instances where it is controversial. For example, there is disagreement over whether the stabilisation of hypoxia-inducible transcription factor α (HIF-1α) and subsequent signalling is caused by mitochondria-derived ROS or by succinate or other metabolites. There is evidence for both, and both could be involved in some congruent mechanism [5,66]. The effects of complex III poisons and genetic manipulations have been taken as strong evidence that superoxide generated from site IIIQo is essential for the stabilisation of HIF-1α [67–69], but such interpretations are limited by the secondary effects of inhibiting electron flow, namely interruption of ATP supply and alterations in levels of metabolites such as succinate. The use of S3QELs enables such site-specific arguments to be addressed. Treatment of HEK-293 cells with S3QELs in hypoxic conditions inhibited the accumulation of HIF-1α, and the same effect was seen when a transcriptional HIF-1α luciferase reporter cell line was used . The effects of S3QELs suggest that superoxide from site IIIQo is required for HIF-1α stabilisation in hypoxia.
The ER has an array of diverse functions such as protein folding and quality control, lipid synthesis, and calcium homeostasis. ER stress can be caused by different factors (including nutrient deprivation, hypoxia, and calcium depletion) that lead to the misfolding of proteins and engagement of the unfolded protein response (UPR) . The UPR is beneficial, as it works to bring the ER back to homeostasis, however chronic ER stress is deleterious and results in extensive damage in many pathologies [71–73]. Tunicamycin induces ER stress by inhibiting glycosylation of proteins and therefore preventing them from folding . There is evidence to suggest that mitochondrial ROS are increased during ER stress, although the mechanism is not fully defined . S3QELs protect INS-1 cells, and both S1QELs and S3QELs protect H9C2 cells, against tunicamycin-induced ER stress [32,33]. Both S1QELs and S3QELs also protected against tunicamycin-induced intestinal stem cell hyperplasia in vivo. S1QELs and S3QELs even protected against intestinal stem cell hyperplasia when the mutated oncogenic Rasv12 protein was overexpressed in only the intestinal stem cells . It is most likely that S1QELs and S3QELs, by protecting against ER stress-induced mitochondrial O2−/H2O2 production, prevent downstream cell death, thereby indirectly inhibiting the release of the UPD3 cytokine that signals to the intestinal stem cells to proliferate .
Signatures of cancer development and growth include loss of succinate dehydrogenase activity resulting in accumulation of succinate , HIF-1α stabilisation resulting in hypoxia signalling , and Ras mutations resulting in hyperplasia and altered Ras signalling . There is an extensive literature implicating a change in mitochondrial O2−/H2O2 production that affects redox signalling pathways and contributes to cancer development and persistence [80,81]. However, the roles of specific mitochondrial sites of O2−/H2O2 production are underexplored. Because S1QELs and S3QELs can change key signalling pathways that drive cancer, they have the potential to be good therapeutic tools in the cancer field.
It is generally agreed that mitochondria generate O2−/H2O2 that can disrupt normal redox signalling and cause damage, thereby driving numerous pathologies, but good tools to interrogate and manipulate mitochondrial O2−/H2O2 production have been lacking.
Newly discovered small molecules that suppress O2−/H2O2 production from different sites in the mitochondrial electron transport chain without directly impacting oxidative phosphorylation or carbon metabolism (S1QELs and S3QELs) enable us to begin to tease apart the importance of superoxide and hydrogen peroxide generation by mitochondrial sites IQ and IIIQo and how these individual sites may be contributing to overall O2−/H2O2 production and driving pathologies (Figure 1).
Current S1QEL and S3QEL data suggest that in some circumstances O2−/H2O2 production by different mitochondrial sites has very different effects, and in some instances perhaps it does not. We finally have excellent tools to better address and control these sites in pathology, opening new pathways to better therapeutics.
hypoxia-inducible transcription factor α
mitochondrial permeability transition pore
plasmacytoid dendritic cells.
reactive oxygen species
suppressor of site IQ electron leak
suppressor of site IIIQo electron leak
- site IIIQo
the site of O2− production in complex III, the outer ubiquinone-binding site
- site IQ
the site in mitochondrial complex I active in O2− and/or H2O2 production during reverse electron transport, nominally the ubiquinone-binding site
unfolded protein response
M.A.W. and H.S.W. wrote the first draft of the manuscript, all authors edited the final version.
M.A.W. and H.S.W. were supported by funding from Calico LLC. M.D.B. was supported by funding from Calico LLC and the Buck Institute.
M.A.W., H.S.W. and M.D.B. received funding from Calico LLC.