The use of Amplex Red or Amplex UltraRed (AmR) at concentrations up to 50 µM is suggested by commercial suppliers for the determination of H₂O₂ production, but much lower concentrations down to 1 µM have been applied successfully [12,13,14,20,21]. Since some fluorescence probes, e.g., mitochondrial membrane potential sensitive dyes (safranin, rhodamine), inhibit mitochondrial respiration [22,23], we considered it advisable to check for such undesired effects and to evaluate the optimal concentration of AmR prior to the actual experimental series. The dose-dependent effect of AmR on mitochondrial respiration is shown in Figure 1 and Figure 2 for intact and permeabilized HEK 293T cells.

In the experiments with intact cells titrations with AmR were conducted in the ROUTINE state of respiration. ROUTINE respiration was not completely stable in controls for the duration of the carrier titrations (DMSO) which lasted approximately 35 min. Titration of AmR caused a slight but non-significant further reduction of ROUTINE respiration by up to 8% at 30 µM compared to time-matched controls (Figure 1B). Subsequent addition of oligomycin induced an immediate inhibition of respiration, and LEAK respiration was indistinguishable between both groups. To obtain a measure for ETS, i.e., the maximal capacity of the electron transfer system under the conditions examined, a step-by-step titration with uncoupler CCCP was performed showing that AmR caused a slight but variable, on average insignificant reduction of ETS capacity. Residual oxygen consumption, ROX, obtained after inhibition of Complexes I and III by rotenone (Rot) and antimycin A (Ama), respectively, was identical in controls and AmR-treated cells. Taken together, these results suggest that AmR may be used at concentrations up to 30 µM to determine H₂O₂ production in intact HEK 293T cells with minor side effects on respiration. Exposure of these cells to 20 µM AmR for more than 45 min, however, caused about 13% inhibition of ROUTINE respiration [24], suggesting that AmR concentrations and exposure time should be limited. The inhibitory effect depends on the medium used. Since the sensitivity of the AmR assay was rather low in DMEM, we applied Dulbecco’s phosphate-buffered saline, in which case excellent assay sensitivity was associated with seriously compromised respiration, both with (up to 50% inhibition) and without AmR (up to 20% inhibition). A brief literature survey indicates that AmR concentrations and media applied are highly variable, ranging from 1 µM AmR in PBS for the study of permeabilized C2C12 myoblasts and myotubes [25], 10 µM AmR in MiR05 for primary human skeletal myotubes [26], to 50 µM AmR in various phosphate-buffered or bicarbonate-buffered saline media investigating N27 cells [27], A549 lung epithelial cells [20], HaCaT keratinocytes [28], or HEK 293T cells [29], or mitochondria prepared from H9c2 rat cardiac myocytes [29] or from HepG2 cells [30]. In the absence of adequate controls it is not clear to which extent these treatments might have affected the results on H₂O₂ production. Therefore, careful optimization is required both with regard to AmR concentrations and the incubation media for quality control of measurements on H₂O₂ production in cells.

Experiments with permeabilized cells were performed in MiR05 [31] and AmR was titrated in the CI&II-linked OXPHOS state, i.e., in the presence of pyruvate, malate and succinate at saturating concentration of ADP (Figure 2). Respiration in controls was not affected by the carrier DMSO. In contrast, AmR inhibited CI&II-linked OXPHOS capacity in a dose-dependent manner, resulting in 15% ± 7% inhibition at 30 µM (mean ± SD, N = 5). Similarly, CI&II-linked ETS capacity was reduced by 19% ± 8%, with a shift to a lower optimum uncoupler concentration compared to controls. CII-linked ETS capacity was not affected, possibly indicating that AmR inhibition occurred at CI, which is highly sensitive to agents damaging mitochondria [22,32,33]. In order to minimize such side effects, the AmR concentration was reduced to 10 µM for experiments with permeabilized cells.

O₂ and H₂O₂ fluxes were determined in permeabilized HEK 293T cells in a sequence of respiratory substrate and coupling states using pyruvate&malate (PM; CI), pyruvate&malate&succinate (PMS; CI&II) or succinate with Rot (S(Rot)) as respiratory substrates (Figure 3). Results are summarized in Figure 4 and Table 2. The H₂O₂/O flux ratio is frequently applied to evaluate the relative importance of H₂O₂ production at different respiratory states [34,35].

After permeabilization of the cell membranes with digitonin, addition of PM as substrates supporting CI-linked LEAK respiration (CI_L) induced a moderate increase in respiration (Figure 3A). ADP added at a saturating concentration stimulated respiration about 5 times (CI-linked OXPHOS capacity, CI_P). Succinate induced convergent CI&II-linked OXPHOS (CI&II_P), at a 2.2-fold higher level compared to CI_P. CCCP at optimum concentration elevated respiration further, showing a significant apparent excess ETS capacity in these cells. Subsequently Rot and Ama were added to obtain ROX. Despite these dramatic differences in oxygen fluxes in different respiratory states, alteration in H₂O₂ fluxes were comparatively small. When total observed H₂O₂ flux was corrected for the background chemical flux obtained in the absence of cells, the net fluxes were negative. This may indicate the ROS scavenging capacity introduced with the cells, as will be discussed below. Thus, even the most pronounced relative change in H₂O₂ flux induced by inhibition of CIII (Ama) after ETS was close to the chemical background. As an alternative approach, the lowest flux detected in each experimental run was subtracted from H₂O₂ flux in each respiratory state, expressing H₂O₂ flux not as an absolute metabolic flux but as a difference, ΔH₂O₂ flux (Figure 4C). The lowest H₂O₂ flux was observed in fully uncoupled cells in almost all cases.

LEAK respiration supported by succinate (CII_L) was about three times higher than L supported by PM (CI_L) and was associated with much higher H₂O₂ flux in the absence of rotenone. ADP induced a 6.7-fold increase in respiration (CII_P) and a concurrent pronounced reduction in H₂O₂ flux (Figure 3B). Surprisingly, addition of pyruvate did not stimulate respiration, and malate caused a slight inhibition, such that CI&II_P respiration was slightly reduced compared to CII_P, whereas H₂O₂ flux remained constant. The fact that rotenone was not required to obtain a high CII-linked OXPHOS capacity is consistent with a high malic enzyme activity in these cells [36]. Thus, oxaloacetate does not accumulate to concentrations which inhibit succinate dehydrogenase, but pyruvate and further acetyl-CoA are formed from malate, supporting the utilization of oxaloacetate in the citrate synthase reaction [19,37]. Formation of NADH and its utilization by CI, therefore, proceeded in the CII-linked OXPHOS state, and CI&II-linked and CII-linked OXPHOS capacities were not different (Figure 4). However, CI&II_P was limited by the capacity of the phosphorylation system, as shown by the increased ETS capacity with PMS (CI&II_E) which was higher than CII_E, indicating the additive effect of convergent electron flow from CI&II to the Q-junction in these cells [38]. CCCP titration reduced the H₂O₂ flux to a level slightly below baseline. As observed in the previous protocol, Ama induced an increase in H₂O₂ flux (Figure 4).

Respiration and H₂O₂ flux in the LEAK state were similar with S(Rot) and S alone (Figure 4). In many other cells and tissues (e.g., rat brain homogenate as shown below), Rot causes a significant reduction of H₂O₂ production observed in the CII-linked LEAK state due to inhibition of reversed electron flow to CI [39]. The absence of such an effect is consistent with the effect of malic enzyme on respiration [19]. In contrast to the presence of S alone, addition of ADP to S(Rot) exposed cells did not decrease the H₂O₂ flux. The effects of PM and CCCP were relatively small, and the addition of Ama did not stimulate ROS production, in contrast to results obtained with the other protocols (Figure 4).

Taken together, in permeabilized HEK 293T cells the highest H₂O₂ flux was observed in the CII-linked LEAK state, amounting to about 0.8% of oxygen flux (Table 2). In comparison, CI-linked H₂O₂ flux appears negligible, while inhibition of CI increased H₂O₂ production independent of respiratory state. H₂O₂ flux in permeabilized HEK 293T cells was extremely low under the presently investigated conditions, accounting for less than 0.2% of O₂ flux in the OXPHOS and ETS states (Table 2).

Given that estimated H₂O₂ fluxes were close to or even below background observed in the absence of cells, the sensitivity of the AmR assay may present a limiting factor at the experimental cell density. By increasing the cell density in the respiration chamber, however, the slopes of the fluorescence signal corrected for the background determined in the absence of cells actually displayed an inverse relation to cell density (Figure 5A). In contrast, when corrected for the slope observed in the presence of digitonin permeabilized cells without substrates added, net H₂O₂ fluxes were largely independent of cell density (Figure 5B). Importantly, cell-specific respiration and net H₂O₂ flow (per million cells) were independent of cell density (Figure 5C,D). A tentative explanation for these observations is that by increasing the number of cells in the chamber we also enhance the total ROS scavenging capacity associated with the cellular antioxidant systems. In addition, the optical properties are affected by cell density, as shown by the step change of the fluorescence signal upon injection of cells, and by the change of sensitivity when comparing calibrations before and after addition of cells, and after titration of digitonin. For example, sensitivity declined from 0.282 ± 0.019 V/µM to 0.260 ± 0.019 V/µM after addition of cells (n = 8 experiments), from 0.249 ± 0.033 V/µM to 0.179 ± 0.018 V/µM after adding brain tissue homogenate (n = 8 experiments), and from 0.262 V/µM to 0.250 V/µM after injection of mitochondria (means of two experiments). Whereas calibration of the fluorescence signal in the absence of cells is required to obtain the apparent background flux, H₂O₂ calibrations are required in the presence of cells and under various respiratory states (Figure 3).

Figure 6 shows representative experiments of respiration and H₂O₂ fluxes in mouse brain homogenate with identical SUIT protocols as applied with permeabilized HEK 293T cells (Figure 3). In contrast to permeabilized HEK 293T cells (Figure 4), CII_L respiration was lower but CII_P was higher with S(Rot) compared to S (Figure 7A). Addition of PM to S induced an increase of respiratory OXPHOS capacity (CI&II_P), despite of the inhibitory effect of malate (Figure 6B). Consistent with results in permeabilized cells, malate caused a significant decrease of CII_P respiration with S(Rot) (Figure 6C). The inhibitory effect of malate on CII-linked respiration is a general feature of TCA cycle control [40].

In contrast to results with permeabilized cells, H₂O₂ flux in state CI_L was well above background in brain homogenate (Figure 6A). As in permeabilized HEK 293T cells, the highest H₂O₂ flux was observed in the LEAK state with S alone (Figure 6B), and H₂O₂ flux was significantly reduced in the OXPHOS and ETS state compared to LEAK (Figure 7B). However, addition of Rot caused a significant reduction of H₂O₂ flux, as did addition of ADP (Figure 7B). Compared to CI_L, H₂O₂ flux in CII(Rot)_L was 2.5 times higher (Figure 7B). No significant difference in H₂O₂ flux was noted between LEAK and OXPHOS states with S(Rot). Surprisingly, addition of PM in state CII(Rot)_P caused an increase of H₂O₂ flux, although Complex I was inhibited by rotenone (Figure 6A and Figure 7B). As expected [41], inhibition of CIII by Ama (ROX state) caused an increase in H₂O₂ flux in all SUIT protocols (Figure 6 and Figure 7B). H₂O₂ flux resulted in similar patterns when corrected for background without homogenate (Figure 7B) or when presented as the difference of flux after subtraction of the minimal observed slope (Figure 7C).

The extraordinarily high H₂O₂/O flux ratio of 6% in CII_L without Rot reflects maximal reversed electron transfer under an artificial substrate condition (Table 2). H₂O₂/O flux ratios between 0.05% and 1.35% in OXPHOS and ETS states (Table 2) are consistent with values reported for isolated mitochondria [1,11,42]. It is thought that isolation procedures may impact on mitochondrial function (e.g., [43,44]). Isolated mitochondrial preparations may represent a selection for particular mitochondrial subpopulations. This is not the case in tissue homogenate or permeabilized tissue preparations. Tissue homogenization can be achieved without injury of the outer mitochondrial membrane using specifically dedicated instruments such as the PBI tissue shredder [45,46]. A more simple glass Potter homogenizer can be used similarly for homogenate preparation of soft tissues.

Few comparable data are published on permeabilized cells. From data by Kwak et al. [26] on permeabilized human skeletal myotubes it can be calculated that about 0.9% of LEAK respiration measured in the presence of a complex substrate mixture was diverted towards H₂O₂ production, in close agreement with our result on 0.8% for HEK 293T cells in the presence of succinate (Table 2). In permeabilized yeast cells the H₂O₂/O ratio varied between 0.25% and 2% depending on the glucose level supplied and the growth phase investigated [47]. The H₂O₂/O flux ratio determined in a mutant E. coli strain was 0.35% to 0.6% in different substrate regimes [48].

Since the most pronounced changes of H₂O₂ flux in permeabilized cells and brain homogenate were observed in protocols using S and S(Rot), we performed additional experiments applying CII-linked protocols with mitochondria isolated from mouse hearts. Like in mouse brain homogenate, the H₂O₂ flux was extremely high in the succinate-supported LEAK state in the absence of Rot, resulting in a H₂O₂/O flux ratio of nearly 10%, compared to 1.5% in the presence of Rot (Figure 8). The addition of ADP to S(Rot) stimulated respiration and concurrently reduced H₂O₂ flux (Figure 8B). In contrast, addition of ADP to S did not increase respiration, but dramatically diminished H₂O₂ flux (Figure 8A). The subsequent addition of P caused a pronounced increase of respiratory OXPHOS capacity, by removing the inhibitory oxaloacetate and restoring CI&II-linked TCA cycle activity [19]. Malate exerted an inhibitory effect in the presence and absence of Rot, similar to results with brain homogenate (Figure 6B,C). Uncoupler did not stimulate respiration beyond OXPHOS capacity, indicating that there is no apparent ETS excess capacity in mouse heart in contrast to mouse brain mitochondria. H₂O₂ fluxes were slightly elevated by P in both protocols, largely unresponsive to M, and slightly diminished by uncoupler, while Rot and Ama caused a substantial increase of H₂O₂ flux (Figure 8A,B). H₂O₂/O flux ratios ranged from 0.04% in ETS in the absence of Rot to 0.9% in OXPHOS and ETS in the presence of Rot, consistent with data reported in the literature [5,49].

Dulbecco’s modified eagle medium (DMEM-low glucose, with l-glutamine) was from PAA Laboratories GmbH, Pasching, Austria, fetal bovine serum from Biowest, Nuaillé, France, and penicillin and streptomycin stocks were from Gibco, Vienna, Austria.

Amplex^® UltraRed was obtained from Life Technologies. H₂O₂, HRP, SOD, substrates, inhibitors and other chemicals were from Sigma-Aldrich, Acros Organics or Invitrogen [50].

The Oxygraph-2k (O2k, OROBOROS Instruments, Innsbruck, Austria) was used for measurements of respiration [50] and combined with the Fluorescence-Sensor Green of the O2k-Fluo LED2-Module for H₂O₂ measurement. Up to four O2k instruments (eight chambers) were used in parallel. Experiments using tissue homogenate and permeabilized cells were performed in MiR05 (110 mM sucrose, 60 mM K-lactobionate, 0.5 mM EGTA, 3 mM MgCl₂, 20 mM taurine, 10 mM KH₂PO₄, 20 mM HEPES, pH 7.1 at 30 °C, and 0.1% BSA essentially fatty acid free; [31]). Dulbecco’s modified eagle medium supplemented with 10% fetal bovine serum and 50 units/mL penicillin and 50 μg/mL streptomycin was used for measurements on intact cells. All experiments were performed at 37 °C. The medium was reoxygenated when oxygen concentrations reached 80 µM unless otherwise indicated.

Respiration of permeabilized cells and tissue homogenate was determined using substrate-uncoupler-inhibitor titration (SUIT) protocols [50] with modifications. Pyruvate and malate (5 mM and 0.5 mM, respectively) or succinate (10 mM) with or without 0.5 µM Complex I inhibitor rotenone (Rot) were used to determine Complex I (CI) or Complex II (CII) linked LEAK respiration. ADP was added at 2.5 mM final concentration, which was saturating for oxygen flux to obtain OXPHOS capacity. Additional substrates were added sequentially to reconstitute convergent CI&II-linked respiration. Titrations with the uncoupler CCCP (0.5–1 µM steps) were performed to determine electron transfer system (ETS) capacity. Rot, if not already present, and Ama (2.5 µM to inhibit Complex III) were added for determination of residual oxygen consumption (ROX).

Respiration of intact cells was measured applying a coupling control protocol [19]. Up to 300 µL of suspended cells were added to the respiration medium. After stabilization of ROUTINE respiration, the ATP-synthase inhibitor oligomycin (Omy, 2 µg/mL) was added to obtain a measure of LEAK respiration, followed by titration of CCCP to maximum oxygen flux (ETS capacity). Finally, Rot and Ama were added to obtain ROX.

H₂O₂ flux was measured simultaneously with respirometry in the O2k-Fluorometer using the H₂O₂-sensitive probe Amplex® UltraRed [15]. 10 µM Amplex® UltraRed (AmR), 1 U/mL horse radish peroxidase (HRP) and 5 U/mL superoxide dismutase (SOD) were added to the chamber. The reaction product between AmR and H₂O₂, catalyzed by HRP, is fluorescent, similar to resorufin. Calibrations were performed with H₂O₂ repeatedly added at 0.1 µM steps as indicated (H₂O₂*). Volume-specific H₂O₂ fluxes were calculated real-time by the DatLab software (OROBOROS INSTRUMENTS, Innsbruck, Austria) from the positive time derivative of the resorufin signal over time (converted to H₂O₂ concentration based on the calibrations with H₂O₂). Only the stable portions of the apparent fluxes were selected and artifacts induced by additions of chemicals or re-oxygenations were excluded.

Human embryonic kidney cells (HEK 293T, ATCC collection code CRL-1573) were cultured in 10 cm² culture dishes in DMEM high glucose medium supplemented with additions as indicated above until approximately 90% confluence was reached. Immediately prior to respirometric assays the cells were washed with PBS, trypsinized and resuspended in MiR05 or DMEM. The final concentration of permeabilized cells in the O2k-chamber was 1.75·10⁶/mL or 1.5–2·10⁶/mL when intact cells were examined. Cells were permeabilized after addition to the respirometer chambers using digitonin at a final concentration of 10 µg/mL. This concentration was evaluated in preliminary experiments to achieve full permeabilization of cells to allow for uninhibited access of substrates and ADP to mitochondria without compromising mitochondrial function [50].

Wild-type C57BL/6 mice (age 2–3 months) were housed under standard conditions (21–23 °C, 12 h light/dark cycle, relative humidity 45%–65%) with unlimited access to food and water. The experimental procedures were performed in accordance with the guidelines of the European Community as well as local laws and policies.

Animals were sacrificed by cervical dislocation, the skull opened with scissors and the brain removed. Brain cortex was dissected and washed in ice-cold MiR05Cr (MiR05 supplemented with 20 mM creatine). The tissue was transferred to a pre-cooled glass Potter homogenizer and homogenized with 10–15 strokes at medium speed. The resulting homogenate was then kept on ice and used for respirometry without further processing. The final concentration of tissue in the O2k-chamber was 1 mg/mL.

Wild-type C57BL/6 mice were sacrificed by cervical dislocation and the heart was excised and weighed. The heart was washed in ice-cold BIOPS and minced in 1 mL of BIOPS. The tissue was transferred to a pre-cooled glass Potter homogenizer with 2 mL of isolation buffer (225 mM mannitol, 75 mM sucrose, 1 mM EGTA, 2.5 mg/mL BSA) supplemented with Subtilisin (0.5 mg/mL). The tissue was homogenized with 6–8 strokes at medium speed. The resulting homogenate was centrifuged for 10 min at 800× g, 4 °C. Then, the supernatant was transferred to a new tube and centrifuged for 10 min at 10,000× g, 4 °C. After centrifugation, the supernatant was carefully discarded, the mitochondrial pellet was washed in 2 mL isolation buffer and resuspended in 100 µL of isolation buffer. Isolated heart mitochondria were stored on ice until use. 5 µL of mitochondrial suspension per chamber were used for each measurement.