Examination of the superoxide/hydrogen peroxide forming and quenching potential of mouse liver mitochondria

Pyruvate dehydrogenase (PDHC) and α-ketoglutarate dehydrogenase complex (KGDHC) are important sources of reactive oxygen species (ROS). In addition, it has been found that mitochondria can also serve as sinks for cellular hydrogen peroxide (H2O2). However, the ROS forming and quenching capacity of liver mitochondria has never been thoroughly examined. Here, we show that mouse liver mitochondria use catalase, glutathione (GSH), and peroxiredoxin (PRX) systems to quench ROS. Incubation of mitochondria with catalase inhibitor 3-amino-1,2,4- triazole (triazole) induced a significant increase in pyruvate or α-ketoglutarate driven O2-/H2O2 formation. 1-choro-2,4-dinitrobenzene (CDNB), which depletes glutathione (GSH), elicited a similar effect. Auranofin (AF), a thioredoxin reductase-2 (TR2) inhibitor which disables the PRX system, did not significantly change O2-/H2O2 formation. By contrast catalase, GSH, and PRX were all required to scavenging extramitochondrial H2O2. In this study, the ROS forming potential of PDHC, KGDHC, Complex I, and Complex III was also profiled. Titration of mitochondria with 3-methyl-2-oxovaleric acid (KMV), a specific inhibitor for O2-/H2O2 production by KGDHC, induced a ~86% and ~84% decrease in ROS production during α-ketoglutarate and pyruvate oxidation. Titration of myxothiazol, a Complex III inhibitor, decreased O2-/H2O2 formation by ~45%. Rotenone also lowered ROS production in mitochondria metabolizing pyruvate or α-ketoglutarate indicating Complex I does not contribute to ROS production during forward electron transfer from NADH. Taken together, our results indicate that KGDHC and Complex III are high capacity sites for O2●-/H2O2 production in mouse liver mitochondria. We also confirm that catalase plays a role in quenching either exogenous or intramitochondrial H2O2.

Mitochondria are quantifiably the most important sources of ROS in mammalian cells. Once considered unfortunate by-products of respiration, it is now well known that ROS can serve as important secondary messengers [1]. Superoxide (O2-) and hydrogen peroxide (H2O2) are the principle ROS formed by mitochondria. Superoxide is often considered to be the proximal ROS produced by mitochondria which is then rapidly dismutated to H2O2. However, recent studies have revealed that ROS forming sites in mitochondria can produce a mixture of O2- and H2O2 [2-4]. For instance, H2O2 accounts for ~75% of the total ROS formed by dihydrolipoamide dehydrogenase subunit (E3) of pyruvate dehydrogenase (PDHC) and α-ketoglutarate dehydrogenase (KGDHC) [3]. Complex I and III of the respiratory chain are often considered to be the only sites for ROS formation in mitochondria. However, evidence collected over the past few years has shown that skeletal muscle mitochondria can contain up to 11 potential sources of O2- and/or H2O2 [5]. PDHC and KGDHC have been found to be important sources of O2-
/H2O2 in liver, muscle, and brain mitochondria [3, 6]. PDHC and KGDHC have also been shown to form ~4x and ~8x more ROS than Complex I in skeletal muscle mitochondria [6]. Other high capacity ROS forming sites in muscle mitochondria include Complex II, Complex III, and glycerol-3-phosphate dehydrogenase [5]. Complex I can also be a high capacity ROS forming site but only during reverse electron flow from the ubiquinone pool [5]. In addition, purified PDHC and KGDHC have the capacity to generate ROS via reverse electron transfer from NADH [3, 4, 7]. PDHC and KGDHC are also targets for oxidative modification by H2O2 where the reversible redox modification of either enzyme complex controls ROS emission, enzyme activity, and protects from oxidative damage [8, 9].

At high enough quantities O2- and H2O2 can induce oxidative distress leading cell damage and death [10]. Thus, mitochondria must tightly control ROS levels to take advantage of its signaling properties whilst simultaneously avoiding toxicity. Mitochondria can control production from individual sites either through redox signaling or induction of proton leaks [11]. Mitochondria are also enriched with antioxidant defense enzymes which play a vital role in controlling ROS levels. The importance of mitochondrial antioxidant defense systems in clearing O2- and H2O2 is underscored by the consequences of the targeted deletion or overexpression of ROS degrading enzymes. Deletion of SOD2 is neonatally lethal and loss of SOD1 is associated with the development of cardiovascular abnormalities [12]. Mitochondria also contain high levels of glutathione (GSH) which utilizes glutathione peroxidase (GPX) to clear H2O2. Mitochondria contain two GPX isozymes; GPX1 is required to eliminate H2O2 and GPX4 quenches lipid hydroperoxides [13]. Elimination of GPX1 accelerates cardiac disease and deletion of GPX4 is embryonically lethal [14, 15]. The peroxiredoxin (PRX) system also plays a vital role in removing mitochondrial H2O2 [16]. Following a round of H2O2 elimination, PRX is oxidized resulting in its deactivation. PRX is reactivated by mitochondrial thioredoxin-2 (TRX2) which is then reduced by thioredoxin reductase-2 (TR2) in the presence of NADPH. Disabling the PRX system by knocking out TRX2 or TR2 is also embryonically lethal illustrating its importance in removing H2O2 [12]. By contrast, overexpression of mitochondria-targeted catalase has been associated with enhanced longevity attributed to increased resistance to oxidative damage [12]. The GSH and PRX systems are often considered the chief H2O2 quenching systems in mitochondria [17]. However, catalase has also been documented to occur in liver and heart mitochondria [18-20]. Overall, O2-/H2O2 removal, much like regulating its production, is vital for harnessing the redox signaling properties of ROS.

Mitochondria contribute significantly to cellular ROS handling [21]. Hydrogen peroxide formed in the cytosol can be quenched by mitochondrial antioxidant defense systems [18, 22, 23]. Substrate catabolism is required to support this function through NADPH formation which provides the necessary reducing equivalents to maintain antioxidant defense systems in an active state [21]. Liver mitochondria are significant sources of ROS in hepatocytes. Detoxification reactions in the cytosol and fatty acid catabolism in peroxisomes can also result in the over production of cellular ROS. Thus, liver mitochondria are likely to be exposed to high concentrations of O2- and H2O2 due to substrate oxidation in mitochondria and the natural physiological functions of hepatocytes. Here, we confirm that catalase forms an important arm of the antioxidant defense system in liver mitochondria which plays a role in quenching exogenous H2O2 or ROS formed by substrate catabolism. We also show that KGDHC is a major source of O2-/H2O2 in liver mitochondria forming almost as much ROS as Complex III.Chemicals: 1-chloro-2,4-dinitrobenzene (CDNB), 3-amino-1,3,4-triazole (triazole), auranofin (AF), H2O2 (30% solution), pyruvate, malate, 2-oxoglutarate, mannitol, Hepes, sucrose, EGTA, fatty acid free bovine serum albumin, Bradford reagent, superoxide dismutase (SOD), myxothiazol, 3-methyl-2-oxovaleric acid (KMV), and horse radish peroxidase (HRP) were purchased from Sigma. Amplex Ultra Red (AUR) reagent was acquired from Invitrogen. Anti-acyl-CoA oxidase-1 and anti-KGDHC (E1 subunit) were purchased from Abcam. Anti-catalase, anti-glutathione peroxidase-1, anti-Sod2, anti- thioredoxin-2, anti-mouse and anti-rabbit horseradish peroxidase secondary antibodies and CPI-613 were purchased from Santa Cruz.

Preparation of mitochondria: All experiments were approved by Memorial University’s Animal Care and Use committee and conducted according to institutional and Federal animal care guidelines. All steps were performed on ice or 4°C. Male C57BL/6N mice were purchased from Charles River Laboratories at 8 weeks of age. Mice were euthanized by cervical dislocation under isoflurane anesthesia at 9 weeks of age. Livers were then removed and placed in buffer composed of mannitol (220 mM), EGTA (1 mM), sucrose (70 mM), and Hepes (10 mM) (pH 7.4, abbreviated MESH buffer)supplemented with 0.5% fatty acid free BSA (MESH-B). Livers were cut into small pieces, washed twice, and homogenized in 15 mL MESH-B using the Potter-Elvejham method. Homogenates were centrifuged at 800 xg for 9 min. Fat was skimmed from the top and the supernatant was centrifuged at 10,000 xg for 9 min to pellet mitochondria. The supernatant was decanted, excess fat was wiped clean from the sides of the tubes, and then mitochondrial pellets were washed in 15 mL MESH-B and centrifuged at 10,000 xg for 9 min. The final mitochondrial pellet was resuspended in 500 µL MESH giving a final concentration of 15-20 mg/mL. Protein content was quantified by Bradford Assay using BSA as a standard.Assessment of mitochondrial O2-/H2O2 production: Mitochondrial O2-/H2O2 production was examined using the Amplex Ultra Red (AUR) assay as described previously [24].

Mitochondria were first diluted to 3 mg/mL in MESH and then stored on ice. Samples were then diluted to 0.3 mg/mL in the wells of a black 96-well plate containing MESH-B (note that for CPI-613 assays BSA was excluded). Mitochondria were allowed to equilibrate for a few minutes and were then treated with CDNB (0-10 µM), triazole (0-10 mM), AF (0-10 µM), KMV (0-10 mM) with or without rotenone (4 µM), CPI-613 (0-150 µM), or myxothiazol (0-10 µM) and incubated for 15 min at 25 °C. Reaction mixtures were then supplemented with HRP (3 U/mL), SOD (25 U/mL), and AUR (10 µM). Reactions were initiated by the addition of pyruvate (50 µM) and malate (50 µM) or α- ketoglutarate (50 µM) and malate (50 µM). We have shown in previous studies that 50 µM pyruvate or α-ketoglutarate with 50 µM malate can induce measurable rates in mitochondrial O2-/H2O2 production [3, 11]. This concentration is slightly below the reported Km for pyruvate carrier (~150 µM) [25]. In addition, malate transporter and α- ketoglutarate carrier were reported to have a Km of ~200 µM [26, 27]. However, it should be noted that in a previous study we observed that adding pyruvate to a final concentration of 150 µM only induces a small increase in O2-/H2O2 production [3]. In addition, it has been found by others that low µM concentrations of other substrates like succinate can induce measurable rates of ROS production by submitochondrial particles [28]. Considering that the reported levels of Krebs cycle metabolites in mitochondria is in the low to mid-µM range we decided to stimulate ROS production with 50 µM substrate [29]. O2-/H2O2 production was tracked at the excitation:emission wavelengths 565 nm: 600 nm using a Synergi Mx2 microplate reader and Gen5 software (BioTek).

Experiments involving the assessment of the ATP-mediated inhibition of ROS production by KGDHC and PDHC were carried out as described in [6]. Briefly, mitochondria were diluted to 3 mg/mL in 1 mL of ice cold MESH-B containing MgCl2 (2.5 mM) and alamethicin (40 µg/mL) , vortexed, and then incubated on ice for 10 minutes [11]. Samples were then centrifuged at 30,000 xg for 10 min and the resulting pellet was resuspended in 1 mL of MESH-B. Mitochondria were then diluted in a reaction buffer containing HRP, SOD, and AUR as described above. TPP (0.3 mM), NAD+ (1 mM), and CoASH (0.1 mM) were then added to facilitate pyruvate or α- ketoglutarate metabolism. Rotenone (4 µM) was then added to prevent electron movement through the respiratory chain and then reaction mixtures were supplemented with ATP (2.5 mM), pyruvate or α-ketoglutarate (50 µM). Changes in O2-/H2O2 production were measured as described above.Mitochondrial H2O2 clearance: The quenching of exogenous H2O2 was assayed as described in [22]. First, pyruvate (50 µM) and malate (50 µMwere added to each well followed by CDNB (10M), triazole (10 mM), and/or AF (10 µM). Next, MESH-B was added to each well along with H2O2 (2.5 µM). Reactions were initiated by the addition of mitochondria (0.3 mg/mL). Reactions were then tracked for different time intervals ranging from 30 seconds to 5 minutes. At the different time intervals HRP, SOD, and AUR were added and H2O2 measured as described above.

Immunoblot: Mitochondria were diluted to 1 mg/mL in Laemmli buffer and then heated at 100 °C for 10 minutes. Twenty micrograms of protein was loaded in each well and then samples were electrophoresed through either a 10% isocratic SDS gel (for detection of ACOX1 or catalase) or a 12% isocratic SDS gel (for GPX1 and TRX2). Upon completion proteins were transferred to nitrocellulose membranes by tank transfer. Successful transfer was ascertained by Ponceau S staining. Membranes were then rinsed once with tris-buffered saline (TBS) containing 0.1% (v/v) Tween-20 (TBS-T) and then blocked for 1 hour under constant agitation at room temperature with TBS-T containing 5% (w/v) non-fat skim milk (blocking solution). Membranes were then washed twice with TBS-T and probed with anti-ACOX1 (1/3000), anti-KGDHC (1/3000), anti-GPX1 (1/2000), anti-TRX2 (1/2000), anti-Sod2 (1/3000), or anti-catalase (1/2000) diluted in TBS-T containing 5% (w/v) BSA and 0.02% NaN3. Membranes were incubated overnight under constant agitation at 4 °C and then washed twice with TBS-T. Membranes were then probed with anti-mouse and anti-rabbit horseradish peroxidase conjugate secondary antibodies for 1 hour at room temperature under constant agitation. Bands were visualized using WestPico Super Signal Chemiluminescent substrate and ImageQuant LAS 4000. Band intensities were quantified using ImageJ software.Data analysis: Amplex Ultra Red assays were performed 4-6 times and in duplicate. Immunoblots were conducted in triplicate or quadruplicate and densitometry normalized to loading controls. All results were analyzed using GraphPad Prism 6 software using 1- way and 2-way ANOVA with Fisher’s LSD posthoc test. Results were expressed as a percentage of the control. * or #; p ≤ 0.05, ** or ##; p ≤ 0.01, *** or ###; p ≤ 0.001.

Mitochondria were treated with different selective inhibitors that deactivate GSH, PRX, or catalase to discern the individual contribution of each antioxidant defense system towards scavenging H2O2 formed during substrate oxidation. The GSH system was disabled with CDNB and AF was utilized to inhibit TR2 which would prevent the reactivation of PRX. Substrate oxidation was initiated with pyruvate (50 µM) and malate (50 µM). Treatment with 0.5 µM CDNB induced a significant increase in O2/H2O2 production (Fig 1A). However, no further increases in O2●-/H2O2 emission were observed at concentrations greater than 0.5 µM (Fig 1A). It has been documented that 1 µM CDNB can deplete a majority of the GSH pool [8]. Thus, it is anticipated that no further increase in ROS production was observed since 0.5 µM CDNB depleted the mitochondrial GSH pool. Next, we examined the effect of disabling the PRX system on substrate-driven mitochondrial ROS production. For this AF was utilized which is a documented inhibitor for TR2. Inclusion of AF did not alter O2-/H2O2 emission even when added to a final concentration of 10 µM (Fig 1B). It has been shown that 1 µM AF induces a robust increase in ROS production in brain mitochondria via the inhibition of TR2 [18]. Moreover, it has been previously documented that low µM amounts of AF can strongly inhibit TR2 in rat liver mitochondria [30]. Next, we tested if catalase inhibitor triazole could affect mitochondrial O2-/H2O2 production. Inclusion of 0.1 mM triazole induced a significant increase in pyruvate-driven ROS formation (Fig 1C). Higher concentrations of triazole did not increase O2-/H2O2 production further indicating that 0.1 mM triazole was sufficient to fully deactivate catalase. Next, the impact of CDNB, AF, and triazole on mitochondrial O2-/H2O2 production was tested with α-ketoglutarate. An almost 2-fold increase in ROS production was observed in mitochondria treated with 0.5 µM CDNB (Fig 2A). However, the rate of O2-/H2O2 production did not increase further at higher CDNB concentrations. Next, the effect of AF on the scavenging of H2O2 formed by α-ketoglutarate/malate metabolism was tested. No change in O2●-/H2O2 production was observed in mitochondria oxidizing α-ketoglutarate even when 10 µM AF was included (Fig 2B). Next, the effect of triazole on α-ketoglutarate driven ROS production was tested. Incubation of mitochondria in triazole (0.1 mM) induced a robust increase in ROS production (Fig 2C). However, no further increases in O2●-/H2O2 production were observed at higher triazole concentrations (Fig 2C). Collectively our results demonstrate that GSH and catalase but not the PRX system is required to quench H2O2 formed during the oxidation of pyruvate or 2-oxoglutarate.

The isolation of mitochondria from liver tissue by differential centrifuation can be accompanied by the unwanted contamination of samples with peroxisomes and other organelles. Peroxisomes contain high levels of catalase which are required to eliminate H2O2 formed during fatty acid oxidation. Thus, it is entirely plausible that the triazole effect we observed above was associated with peroxisomal contamination of our mitochondrial preparations. We decided test our mitochondrial preparations for acyl- CoA oxidase 1 (ACOX1), an enzyme that oxidizes fatty acids in the lumen of peroxisomes. Immunoblot analysis revealed the almost complete absence of ACOX1 in four different mitochondrial samples suggesting that peroxisomal contamination was minimal (Fig 3A). Next, the relative protein levels for GPX1, TRX2, and catalase were tested. Both GPX1 and TRX2 displayed variable expression between the three separate mitochondrial preparations. However, this was not associated with a variation in total amount of protein loaded per well (Fig 3B). Catalase displayed strong immunoreactive bands between individual mitochondrial samples (Fig 3B). Band quantification also revealed that catalase protein levels were significantly higher than GPX1 and TRX2 (Fig 3B). Taken together, liver mitochondria contain catalase which plays a part in quenching H2O2 formed during substrate oxidation.

Liver mitochondria utilize all three antioxidant systems to quench H2O2 in the surrounding medium.To examine the capacity of the different antioxidant systems in clearing exogenous H2O2 mitochondria were treated with CDNB, AF, or triazole. Figure 4A demonstrates that all three systems play a role in quenching exogenous H2O2. After a 2.5 minute incubation H2O2 levels in control samples were ~20% relative to time zero (Fig 4A). Inclusion of either CDNB, AF, or triazole significantly diminished mitochondrial H2O2 clearance. However, triazole was the most effective at inhibiting H2O2 elimination. Indeed, after 2.5 min only ~55% of the external H2O2 was removed by mitochondria treated with triazole (Fig 4A). By contrast ~67%, and ~70% of the external H2O2 was scavenged after an incubation period of 2.5 minutes by mitochondria treated with CDNB or AF. At 3.5 minutes, only ~28% of the H2O2 remained in the external medium following triazole treatment (Figure 4). By contrast, after 3.5 minutes ~41% and ~33% external H2O2 remained in the surrounding medium in CDNB and AF treated mitochondria. Longer incubations resulted in the quenching of H2O2 to ~20% of its original concentration in all samples. Next mitochondria were treated with a combination of different inhibitors. As shown in Figure 4B, the catalase system plays a central role in the immediate clearance of H2O2 from the surrounding medium. Triazole in combination with either CDNB or AF diminished mitochondrial H2O2 consumption by ~70% (Fig 4B). By contrast mitochondria were still able to clear ~60% of the H2O2 following CDNB + AF treatment. Inclusion of all three inhibitors (Triazole + AF + CDNB) induced a ~90% inhibition of mitochondrial H2O2 scavenging. Collectively these results demonstrate that all three antioxidant systems eliminate extramitochondrial H2O2. Catalase is required to for immediate clearance of H2O2 at higher than normal concentrations whereas GSH and PRX systems are utilized to maintain steady state H2O2 levels. Assessment of the effect of 3-methyl-2-oxovaleric acid on O2●-/H2O2 production

3-methyl-2-oxovaleric acid (KMV) is a structural analog of α-ketoglutarate and a competitive inhibitor for KGDHC. KMV has been documented to serve as a strong inhibitor for O2●-/H2O2 production by KGDHC [6]. It has also been shown that 10 mM KMV can decrease ROS production by KGDHC by ~90% in skeletal muscle and liver mitochondria [3, 6]. However, to our knowledge the amount of KMV required to lower ROS emission from KGDHC by ~90% has not been tested in liver mitochondria. Figure 5A demonstrates that 0.1 mM KMV induced a ~40% decrease in α-ketoglutarate driven ROS production. At 0.5 mM KMV, O2●-/H2O2 production was inhibited by ~70%. At higher concentrations KMV diminished O2●-/H2O2 production further reaching saturation at 1 mM (Fig 5A). Our group has shown that 10 mM KMV can also lower pyruvate/malate-driven ROS production by ~90% [3]. Thus, we decided to examine how much KMV is required to fully inhibit ROS production in liver mitochondria oxidizing pyruvate and malate. At 0.1 mM KMV elicited no significant change in O2-/H2O2 formation (Fig 5B). However, increasing the concentration of KMV to 0.5 mM induced an ~60% decrease in ROS production. At a concentration of 10 mM, KMV was able to inhibit pyruvate driven ROS production by ~90% (Fig 5B). We decided to compare the KMV effect between mitochondria metabolizing α-ketoglutarate versus pyruvate. KMV was slightly more effective at inhibiting ROS production when α-ketoglutarate was the substrate (Fig 5C). In the presence of pyruvate slightly more ROS was formed in comparison to α-ketoglutarate up to 1 mM KMV. However, at 5 mM and 10 mM KMV no differences were apparent between mitochondria metabolizing α-ketoglutarate or pyruvate (Fig 5C). Collectively these results indicate that KGDHC is a more significant source of ROS than PDHC even when pyruvate is the substrate.

CPI-613 is a lipoic acid analog and putative anti-cancer agent [31]. Lipoic acid is found in the E2 subunit of PDHC and KGDHC and is required to drive the catalytic activity of both enzymes. CPI-613 has been shown to inhibit PDHC and KGDHC likely through the replacement of lipoic acid in the E2 subunit. We decided to follow up the observations in Figure 5 by testing the effect of CPI-613 on α-ketoglutarate and pyruvate-induced ROS production. It had been previously shown that 150 µM CPI-613 is effective at inhibiting PDHC and KGDHC [31, 32]. It is important to emphasize that albumin needs to be excluded from reaction mixtures since it can interfere with the CPI- 613 mediated inhibition of PDHC or KGDHC. Five micromolar CPI-613 induced a significant decrease in O2●-/H2O2 production by α-ketoglutarate (Fig 6A). At higher concentrations, CPI-613 decreased ROS production further reaching ~87% inhibition at 150 µM. Similar observations were made with pyruvate. Five micromolar CPI-613 induced a ~40% decrease in O2●-/H2O2 production (Fig 6B). In addition, CPI-613 mediated inhibition of ROS production reached saturation at 50 µM (Fig 6B). Next, we compared the effect of CPI-613 on α-ketoglutarate and pyruvate-induced ROS production. Overall, no differences were observed between the two different substrates (Fig 6C). To our knowledge this is the first study to examine the effectiveness of CPI- 613 at limiting mitochondrial ROS formation. Taken together, CPI-613 is a good pan- inhibitor for O2●-/H2O2 production by lipoic acid containing enzymes.

Several studies have demonstrated that myxothiazol is an effective inhibitor for ROS production by Complex III [33, 34]. Our group has found that myxothiazol can induce a significant decrease in pyruvate and succinate driven ROS production in liver mitochondria [3, 11]. Myxothiazol selectively binds to the ubiquinone binding pocket that is close to the positive side of the inner mitochondrial membrane (Qp site) which inhibits ROS emission from Complex III [33]. The results collected above indicate that KGDHC can be a significant source of ROS. However, inhibition of KGDHC (or PDHC) also curtails NADH production which would also limit O2●-/H2O2 production by Complex I or Complex III. Thus, we decided to study the impact of myxothiazol on ROS production. Figure 7A shows that 0.1 µM myxothiazol was sufficient to induce a significant decrease in O2●-/H2O2 production induced by α-ketoglutarate oxidation. In addition, 5 µM myxothiazol was found to be the point for full inhibition of ROS production by Complex III (Fig 7A). Similar observations were made with pyruvate. At
0.1 µM, myxothiazol significantly lowered mitochondrial O2●-/H2O2 formation reaching saturation at 5 µM (Fig 7B). In addition, the percent inhibition at 5 µM was ~48% (Fig 7B). No differences in the inhibition of O2●-/H2O2 production by myxothiazol between mitochondria metabolizing either pyruvate or α-ketoglutarate were observed (Fig 7C). Complex III and KGDHC are high capacity ROS forming sites in liver mitochondri.The maximum inhibition of ROS production by myxothiazol in liver mitochondria was recorded to be ~45% for α-ketoglutarate ant ~48% for pyruvate. This would mean that ~52-55% of the ROS produced in a system exclusively metabolizing pyruvate or α- ketoglutarate is originating from KGDHC and PDHC and potentially Complex I.

Furthermore, between the two enzymes it is likely that KGDHC is the more potent ROS generator (Figure 5). We decided to utilize the results collected in Figures 5-7 to graph the mean of the percent inhibition of ROS formation for each inhibitor (Figure 8). Based on our calculations, regardless of whether pyruvate or α-ketoglutarate served as the substrate, Complex III accounted for around 46% of the ROS formed. In the presence of α-ketoglutarate, KMV and CPI-613 induced ~87% inhibition in ROS formation at saturation (Figure 8). In pyruvate treated mitochondria, the highest amounts of KMV and CPI-613 elicited a similar effect. Taken together this would mean that PDHC and KGDHC account for ~41% of the ROS formed by mitochondria. The other ~13% can be attributed to other sources. Based on our experimental model, it is difficult to ascertain precisely how much ROS is being emitted from PDHC and KGDHC. However, based on the effect of KMV in mitochondria metabolizing pyruvate or α-ketoglutarate it can be inferred that KGDHC is a far more potent site for production than PDHC. Figure 8 demonstrates that 10 mM KMV suppressed pyruvate-induced O2-/H2O2 production by ~89%. At lower concentrations, KMV was not as effective as CPI-613 at suppressing ROS formed by pyruvate metabolism. This is in contrast to mitochondria oxidizing α- ketoglutarate where KMV and CPI-613 elicited similar effects (Fig 8). Although KMV was not as effective at CPI-613 at inhibiting O2●-/H2O2 production, the difference between the two was small during pyruvate catabolism (at point 3 in the pyruvate graph, which corresponds to 500 µM KMV and 10 µM CPI-613 the difference was ~7% whereas at point 4 it was ~13%). Therefore, KGDHC is a higher capacity site for ROS generation than PDHC and Complex III produces the most O2-/H2O2 in liver mitochondria.

The results collected above indicate that Complex III and KGDHC are high capacity ROS forming sites in liver mitochondria. However, it is also important to discern if Complex I also generates ROS during pyruvate or α-ketoglutarate oxidation. Thus, we decided to examine if Complex I can be a significant source ROS under our experimental conditions by testing the effect of rotenone in combination with ATP or KMV on overall O2-/H2O2 emission during pyruvate and α-ketoglutarate oxidation. First, we permeabilized mitochondria with alamethicin to test the effect of ATP in combination with rotenone on ROS emission. For these experimental conditions malate was omitted to avoid the priming of the Krebs cycle and the condensation of acetyl-CoA with oxaloacetate. ATP has been documented to impede ROS production by KGDHC and is a known inhibitor for PDHC activity [6]. Alpha-ketoglutarate was more effective than pyruvate at driving O2●-/H2O2 production confirming observations made in a previous study (Fig 9A) [3]. Surprisingly the inclusion of rotenone induced a significant decrease in ROS production in mitochondria oxidizing α-ketoglutarate or pyruvate (Fig 9A). ATP also inhibited ROS production in mitochondria oxidizing pyruvate (Fig 9A). Next, to establish that KGDHC and Complex III are high capacity ROS forming sites and Complex I is not significant source of O2●-/H2O2, intact mitochondria oxidizing pyruvate or α-ketoglutarate with malate were treated with KMV and rotenone. As expected KMV (10 mM) decreased mitochondrial ROS production by ~88% in mitochondria oxidizing pyruvate or α-ketoglutarate in the presence of malate (Fig 9B). The inclusion of rotenone also decreased ROS production by ~32% and ~40% in mitochondria oxidizing either substrate. KMV in combination with rotenone decreased ROS production by almost 90% regardless of which substrate was being oxidized (Fig 9B). These results demonstrate that Complex I is not a significant source of ROS during forward electron flow in liver mitochondria, KGDHC produces more O2●-/H2O2 than PDHC, and that Complex III is likely the highest capacity ROS forming site. Collectively, it can be inferred that KGDHC is a more potent source of ROS than PDHC in liver mitochondria. Based on this the hierarchy for ROS production in the presence of pyruvate or α-ketoglutarate in liver mitochondria is Complex III > KGDHC >> PDHC > other sites (e.g. Complex I).

Mitochondrial O2●-/H2O2 production was first documented half a century ago and is now being intensely researched more than ever due to its relationship with health and disease. However, considerable gaps in our knowledge still exist in terms of how mitochondria maintain overall cellular ROS balance. For instance, it has now become evident that mitochondria can contain up to 11 potential ROS forming sites with some enzymes producing more than others [5]. Identification of which metabolic enzymes in mitochondria serve as high capacity sites has mostly been carried out in skeletal muscle [5]. These studies have found that high capacity sites for ROS production include KGDHC, PDHC, Complex I (during reverse electron flow from succinate or fatty acids), Complex II, and III [5]. Recent work has also shown that KGDHC and PDHC can serve as important sources of ROS and that Complex III is likely the highest capacity O2●-/H2O2 producing site in cardiac and liver mitochondria [3]. Formation of supercomplexes, mitochondrial shape, respiratory complex assembly, post-translational modifications, and redox signals have been implicated in controlling mitochondrial ROS formation [35-37]. However, since mitochondria can contain up to 11 potential sites for production it complicates our capacity to understand 1) how mitochondrial O2●-/H2O2 production is regulated, 2) which sites are the most potent sources in different tissues, and 3) which sites make contributions to intramitochondrial and cellular signaling. The complexities surrounding understanding mitochondrial ROS homeostasis are compounded by observations that mitochondria also quench exogenous H2O2. Indeed, muscle, brain, heart, and liver mitochondria play a significant role in the decomposition of exogenous H2O2 [17, 21, 22]. Moreover, Dey et al recently demonstrated that cytosolic ROS handling is dependent on respiring mitochondria [17]. Thus, the capacity of mitochondria to control cellular H2O2 levels depends on several factors including 1) mitochondrial number, 2) the metabolic and bioenergetic state of a population of mitochondria, 3) H2O2 concentration outside of the mitochondrion, 4) the redox state of the mitochondrial matrix (e.g. mitochondrial redox buffering networks and NADPH pools), and 5) the ROS forming potential of different sites of production in the presence of different O2●-/H2O2 forming substrates. Collectively, mitochondria play a dynamic role in controlling overall cellular ROS levels via the production and degradation of O2●- and H2O2.

In the present study, we characterized the O2●-/H2O2 degrading properties of mouse liver mitochondria. First, we focused on examining the importance of different antioxidant systems in quenching H2O2 formed during substrate oxidation. Several important observations were made in this regard; 1) GSH and catalase are integral for quenching H2O2 formed during carbon metabolism, 2) the PRX system does not degrade H2O2 formed during carbon oxidation, 3) all three antioxidant systems are required to quench H2O2 from the surrounding medium. Treberg et al showed that selective inhibition of TR2 with AF induced a robust increase in O2●-/H2O2 production in mitochondria metabolizing glutamate/malate or succinate [23]. The capacity of AF to increase H2O2 efflux from mitochondria oxidizing glutamate or succinate was confirmed in a later study by the same group [22]. In addition, it has been documented in several other studies that TR2 is vital for H2O2 quenching in brain mitochondria [18, 38]. However, in our hands we found no evidence that the PRX system is required to remove O2●-/H2O2 formed during substrate oxidation. Indeed, Figures 1 and 2 demonstrate that inclusion of AF (up to 10 µM) did not alter mitochondrial O2●-/H2O2 production. Notably this concentration is 5-10 times higher than what is normally used to deactivate TR2 [18, 23]. Although this result may seem surprising, similar observations had been made with intact liver mitochondria in a previous study. Rigobello et al found that AF does not alter O2●-/H2O2 emission from rat liver mitochondria oxidizing several different substrates [38]. However, mitochondria treated with AF displayed a much higher ROS forming profile after antimycin A treatment in comparison to samples treated with antimycin A alone [38]. Intriguingly we also found that triazole, which selectively disables catalase, induced a robust increase in O2●-/H2O2 production. Initially we were skeptical of this observation since the presence of catalase in mitochondria is often attributed to peroxisomal contamination. However, immunoblot analysis revealed very little peroxisomal contamination and that catalase displayed a higher protein expression profile than GPX1 or TRX2. Liver mitochondria can be exposed to high amounts of ROS due to the physiological function of hepatocytes (e.g. detoxification reactions which form ROS).

In addition, during substrate oxidation liver mitochondria can form high amounts of O2●-/H2O2. Thus, catalase likely serves as an additional line of defense to ensure mitochondria are not exposed to high ROS formed during normal liver function. It is also possible that skeletal muscle and brain do not express mitochondrial catalase since both tissues do form as much ROS as hepatocytes. Thus, only the GSH and PRX systems are required in these two tissues.Hepatocytes are armed with a battery of O2●- and H2O2-forming cytosolic enzymes that are required to eliminate toxins, metabolic waste products, and activate different signaling pathways. Overproduction of ROS by these cytosolic sites of formation is also associated with mitochondrial dysfunction and oxidative stress, which has been suggested to cause a number of liver disorders. Thus, it is likely that liver mitochondria serve as cellular ROS stabilizers. In the present study we found that control samples quenched~80% of the external H2O2 within 2.5 minutes which based on our knowledge is a faster rate of clearance than brain or muscle mitochondria [21, 22]. This rapid quenching is likely associated with the high enrichment of H2O2 clearing enzymes in liver mitochondria which includes catalase. Hydrogen peroxide levels remained at ~20% relative to time 0 for the rest of the experiment which may represent the steady-state level for H2O2 in liver mitochondria. We found that all three enzyme systems were involved in quenching H2O2 from the surrounding medium. This is in contrast to a previous study that showed that only catalase was required to quench extramitochondrial H2O2 [18]. Here, inclusion of triazole resulted in the scavenging of ~55% of the external H2O2.

CDNB and AF were less effective after a 2.5 minute incubation but still induced a significant decrease in H2O2 clearance. At longer incubations (e.g. 3.5 minutes) external H2O2 was still significantly higher in mitochondrial preparations treated with CDNB or AF but not triazole. Considering that H2O2 quenching ceased when external levels were at ~20% it can be concluded that mitochondria reached a steady-state in terms of clearance and production. It important to point out that we utilized 2.5 µM H2O2 in our experiments which is high in terms concentrations normally found in the liver (usually in the low to mid-nanomolar range) [39]. Catalase was instrumental for the initial removal of H2O2 from the surrounding medium. However, once levels were low enough GSH and PRX systems were responsible for quenching any remaining H2O2 until a steady-state was reached (~20% remaining in solution). It was also found that inclusion of all three inhibitors (CDNB + AF + Triazole) induced a ~90% decrease in the clearance of exogenous H2O2. By contrast adding triazole in combination with CDNB or AF induced a~65% decrease in H2O2 and AF with CDNB only resulted in about a 35% inhibition of clearance. First, this illustrates that catalase is the most important system for the initial clearance of H2O2 when it is at a higher than normal concentration. Second, since we obtained ~90% inhibition with all three inhibitors this would suggest there is a fourth H2O2 clearing system in mitochondria. One possibility is α-keto acids like pyruvate or α- ketoglutarate which have been implicated in H2O2 degradation [40]. However, both molecules exhibit very slow kinetics in terms of H2O2 clearance when compared to enzymatic systems (GSH, PRX, and catalase) and thus the role of α-keto acids in antioxidant defense is still questionable [41]. Overall these results demonstrate there may
be a “division of labor” in mitochondrial antioxidant defense where catalase is required to quench H2O2 when it is at a higher than normal concentration. This could be vital for situations when hepatocytes are forming higher than normal ROS levels during fat or toxin elimination. By contrast, GSH and PRX most likely serve as a buffer system maintaining H2O2 at low levels which may be required for mitochondrial redox signaling.

In the second part of our study, we set out to characterize the O2●-/H2O2 forming potential of three potent sources of mitochondrial ROS; Complex III, KGDHC, and PDHC. Our choice to characterize these three sites further is rooted in observations we have made in previous studies. First, our group recently showed that KGDHC and PDHC are significant sources of ROS in liver mitochondria with the former producing more O2/H2O2 than the latter [3]. Second, we found in the same study that Complex III is a significant source of ROS in liver mitochondria metabolizing pyruvate or succinate and that Complexes I and II were not high capacity sites for production [3]. We have also found that KGDHC and PDHC are sites for S-glutathionylation, a redox modification which alters the amount of ROS formed by either complex which may play a critical role in controlling H2O2 signaling [11]. Thus, in the present study we conducted a detailed assessment of the ROS forming potential of Complex III, KGDHC, PDHC, and ComplexI. Based on our analysis we found that the hierarchy for O2●-/H2O2 production was Complex III > KGDHC >> PDHC > Complex I from highest to lowest. This arrangement is in agreement with Quinlan et al where it was found that 1) KGDHC and PDHC form ~8x and ~4x more ROS than Complex I and 2) Complex III, is the most significant source overall [6]. Other sources such as, proline dehydrogenase or dihydroorotate dehydrogenase form negligible amounts. The intriguing aspect about these observations is the role of Complex I in forming ROS. Traditionally, Complex I is often viewed as one of the most important ROS forming sites in mitochondria. However, this concept is now currently being challenged since enzymes like KGDHC, PDHC, and branched chain keto acid dehydrogenase have been found to produce more than Complex I in the presence of NADH [6]. To our surprise, it was found that Complex I is not a high capacity ROS forming site even when rotenone was added.

Indeed, the addition of rotenone actually decreased O2●-/H2O2 production indicating that the principle site for ROS formation in the electron transport chain is Complex III. This conclusion is supported by the observation that myxothiazol lowered ROS production by ~45%. It should be emphasized that Complex I can produce high amounts of ROS during reverse electron flow from succinate [42]. However, it would appear that during forward electron flow Complex I is not a significant source of ROS. In the present study, we employed three different inhibitors, KMV, CPI-613, and ATP, to investigate the ROS forming potential of KGDHC and PDHC. To our knowledge this is the first study to use the lipoic acid analog, CPI-613, to examine mitochondrial ROS emission. Based on our findings, CPI- 613 is a good pan-inhibitor for ROS production from lipoic acid containing enzymes. KMV is a specific inhibitor for KGDHC and was highly effective at inhibiting O2●-/H2O2 production in mitochondria metabolizing pyruvate. Moreover, it was found that ATP is effective at limiting ROS production during pyruvate or α-ketoglutarate oxidation in permeabilized liver mitochondria. Collectively, we can conclude KGDHC is the chief site for ROS production even when pyruvate is being oxidized in the presence of malate. This could have implications for groups invested in deciphering which mitochondrial sites produce the most ROS in cells metabolizing glucose or other carbon sources like amino acids or fatty acids in different physiological and pathological models (e.g. cancer).

The emergence of new and sensitive technologies has allowed the identification of new sites for ROS production in mitochondria. This has strong implications for the signaling function of H2O2 and how it is used by mitochondria to communicate to the rest of the cell. Hydrogen peroxide is often referred to as a mitokine, a secondary messenger that serves as a superimposed co-signal that controls mitochondrial function(s) and cell physiology. However, it is also clear that mitochondria serve as a sink for cellular ROS. Thus, it is likely that the secondary messenger function of H2O2 is influenced by production and clearance by mitochondria [43]. The concept that mitochondria can serve as a “ROS stabilizing device” was proposed by Starkov et al and since then several groups have shown that mitochondria are required to manage the cell’s ROS budget – forming and quenching H2O2 in response to mitochondrial and cellular changes in redox state and ROS levels [21, 43]. Here, we have shown that this concept applies to liver mitochondria. It was found that liver mitochondria rely on the GSH, peroxiredoxin, and catalase systems to quench exogenous or H2O2 formed during substrate oxidation. We also show that KGDHC is an important ROS generator which could have strong implications for studies tracking the effect of glucose or other sources of carbon on mitochondrial O2-/H2O2 production. We propose that ROS source/sink function in liver cells is CPI-613 crucial for 1) maintaining cellular redox signaling networks and 2) buffering cellular ROS when hepatocytes are engaged in detoxification reactions or other activities, which can potentially cause cellular and mitochondrial damage.