Quantitative optical measurement of mitochondrial superoxide dynamics in pulmonary artery endothelial cells

Reactive oxygen species (ROS) play a vital role in cell signaling and redox regulation, but when present in excess, lead to numerous pathologies. Detailed quantitative characterization of mitochondrial superoxide anion ( O2•-) production in fetal pulmonary artery endothelia cells (PAECs) has never been reported. The aim of this study is to assess mitochondrial O2•- production in cultured PAECs over time using a novel quantitative optical approach. The rate, the sources, and the dynamics of O2•- production were assessed using targeted metabolic modulators of the mitochondrial electron transport chain (ETC) complexes, specifically an uncoupler and inhibitors of the various ETC complexes, and inhibitors of extra-mitochondrial sources of O2•-. After stabilization, the cells were loaded with nanomolar mitochondrial-targeted hydroethidine (Mito-HE, MitoSOX) online during the experiment without washout of the residual dye. Time-lapse fluorescence microscopy was used to monitor the dynamic changes in O2•- fluorescence intensity over time in PAECs. The transient behaviors of the fluorescence time course showed exponential increases in the rate of O2•- production in the presence of the ETC uncoupler or inhibitors. The most dramatic and the fastest increase in O2•- production was observed when the cells were treated with the uncoupling agent, PCP. We also showed that only the complex IV inhibitor, KCN, attenuated the marked surge in O2•- production induced by PCP. The results showed that mitochondrial respiratory complexes I, III and IV are sources of O2•- production in PAECs, and a new observation that ROS production during uncoupling of mitochondrial respiration is mediated in part via complex IV. This novel method can be applied in other studies that examine ROS production under stress condition and during ROS-mediated injuries in vitro.

• − production in cultured PAECs over time using a novel quantitative optical approach. The rate, the sources, and the dynamics of

Introduction
Reactive oxygen species (ROS) are biologically important molecules. They are involved in signaling, but when present in excess (oxidative stress), they exert deleterious effects on cell structure and function. Cellular ROS are produced in mitochondria and from nonmitochondrial sources, including the NADPH oxidase (NOX) system and uncoupled endothelial nitric oxide synthase (eNOS). Mitochondrial ROS are generated primarily during electron transfer along the electron transport chain (ETC) complex proteins. Of all the cellular sources of ROS, electron leakage from the ETC to O 2 (dioxygen) is responsible for a steady flux of superoxide ( O 2 • − ) anions, which makes mitochondria the major site of the primordial ROS production. [1][2][3][4] Under physiological conditions, small amounts of ROS are generated due to partial reduction of O 2 into O 2 • − anion. [4][5][6] The major areas for electron leak leading to O 2 • − production includes flavins and quinones of the ETC complexes, and this is more prominent under conditions that decrease electron transfer to complex IV, the terminal electron acceptor. 4 Physiological ROS are maintained at acceptable levels by endogenous matrix of antioxidant defense mechanisms. In this case, the O 2 • − anions are readily dismutated to other radical species, such as hydrogen peroxide (H 2 O 2 ) and sometimes to the more reactive hydroxyl radicals (OH • ) 4 by mitochondrial manganese superoxide dismutase (MnSOD) and other enzymatic reactions. 4,6 Therefore, since O 2 • − is the precursor of most of the downstream ROS and it is also involved in the propagation of oxidative stress-mediated damages, it has become an important biomarker to assess oxidative damage of key macromolecules. 6,7 There is an expanding body of evidence that ROS participate in numerous cell signaling mechanisms, and are widely implicated in the pathologies of cellular and tissue injury. [8][9][10][11][12][13][14][15] Understanding and discerning the role of ROS depends on the ability to measure and quantify the dynamics of mitochondrial ROS production under normal and stress conditions. A variety of metabolic modulators of mitochondria, such as ETC uncouplers and inhibitors, as well as metabolic stress conditions, for example, hyperoxia and hypoxia (O 2 stress), lead to altered mitochondrial ROS production. [16][17][18][19] MitoSOX Red, a fluorescence probe, and a derivative of Hydroethidine (HE), is a widely used probe for mitochondrial-targeted O 2 charge on the phosphonium group in MitoSOX Red selectively targets this cell-permeant HE derivative to mitochondria. 20  production during simulated metabolic/hypoxic stress. Therefore, additional experiments were conducted using apocynin to assess potential ROS production from NOX sources. Overall, the use of time-lapse microscopy provides an ideal approach to study the spatial and temporal changes in mitochondrial O 2 • − production in real-time during metabolic stress in live cells. This method not only assesses the O 2 • − production from the ETC, but also localizes the source of ROS in mitochondria and extra-mitochondrial sources. Lastly, the application of this approach is not limited to studying dynamic ROS production in real-time in PAECs, it can be used in other types of cells under normal and pathophysiological conditions.

Cell preparation
PAECs from normotensive fetal lambs (NFL) were isolated and characterized using techniques we described previously. 32 . 35,36 For the time lapse monitoring of the O 2 • − production, MitoSOX loading was performed online while the experiment was running in the microscope chamber environment. To maintain a high intra-mitochondrial concentration of MitoSOX, the loading process was not followed by washout of residual dyes. This approach is to ensure that the MitoSOX concentration (nM) in mitochondria is high enough to compete with MnSOD for O 2 • − .
Ongoing binding to mitochondrial O 2 • − allows for real-time monitoring of the MitoSOX oxidation rate, which reflects O 2 • − concentration and rate of production over time.

Fluorescent microscopy-Live cells were imaged using a Nikon
Ti-E inverted microscope, with four fluorescent interchangeable filter cubes in addition to the standard DIC and bright-field channels. The bright field images were acquired using an overhead halogen lamp, whereas the fluorescent images used a mercury arc lamp, to take advantage of its intense peaks in the ultraviolet range. Each image was acquired at a magnification of 20 × with a scale of 0.32 μm per pixel. The images were captured using a charge-coupled device camera (Q-imaging, Aqua Exi, 14 bit, 6.45 μm per pixel) with exposure time (0.68 μs/pixel in red channel) set to ensure proper use of the dynamic range of the camera, while avoiding saturation and photo bleaching. The filter set in the blue and red channels filters excitation spectra at 340-380nm and 528-553 nm, respectively with emission spectra at 435-485nm and 590-650 nm, respectively. 2 Time-lapse images were obtained in the blue (Hoechst), red (MitoSOX), and bright field (BF) channels to monitor nuclei, mitochondrial ROS levels and the structure of the cells, respectively. Four fields of view (FOV) of cells were imaged (one FOV in each chamber of the bottom-glass dish) under the aforementioned settings. The microscope is surrounded by a custom-made chamber (Okolab) housed around the stage, providing gas exchange and controlled temperature for time-lapse imaging over several hours. During the experiments, the level of the CO 2 inside the chamber was maintained at 5% by mixing CO 2 with room air at the proper ratio and chamber temperature was kept at 37 ± 1 (SE)°C. O 2 and CO 2 levels were continuously monitored with an O2-BTA model O 2 sensor and CO 2 -BTA model CO 2 probe (Vernier Co., Beaverton, OR).

2.2.3.
Co-localization-Z-stacks of green and red images of 20 randomly selected PAECs previously stained with Mitotracker green (50 nM) and loaded online with MitoSOX red (500 nM) were acquired. The nuclear region of each cell image in z-stacks was excluded using the nuclear mask obtained in blue channel. Co-localization analysis was performed by pseudocoloring and merging green and red fluorescence images together. Both fluorophores resided within the same 3D volume whose minimum size is defined by the resolution limits of the microscope (0.32 μm at a magnification of 20 ×). Quantitative statistical analyses of both the spatial distribution and the correlation between the intensities of the green and red fluorescence images were performed to measure co-localization. Co-localization was determined by quantification of overlapping channels, performed by using the "object-based methods" algorithm of JACoP plugin V2.0. 37 Viability of PAECs treated with MitoSOX only, MitoSOX and PCP, MitoSOX and PCP and KCN was assessed by trypan blue exclusion assay. 38 Furthermore, additional experiments revealed that none of the mitochondrial modulators at the used concentrations interfered with MitoSOX fluorescent emission signal. 39 was used to detect the border of the cells in the bright field images. The obtained mask (cells contours) was applied to the time-lapse image stack in the red fluorescent channel. The nuclei were also identified in the blue images and the resulting blue binary image was used as a mask for the stack of the red images to exclude the nuclei contribution to the red intensity profiles. The mean intensity of the mitochondria in red channel images was calculated as a raw intensity profile of the live PAECs over time. This profile helps to monitor the dynamics of the mitochondrial ROS production before, during and after altering ETC function.

Cell segmentation-Our previously reported cell segmentation algorithm
Intensity profile: Figure 1 represents the overall methodology to obtain final intensity profiles of the cells from the input fluorescent images. Intensity profile of the red fluorescence images shows the dynamic of O 2 • − production in response to MitoSOX (t = 10 min) and metabolic agents (t = 30 min). In order to quantify the dynamic changes in mitochondrial O 2 • − levels and compare the changes in O 2 • − production between the control group (PAECs with MitoSOX, i.e., no treatment) and treated groups (PAECs treated with inhibitors or/and the uncoupler), the raw intensity profiles were first background subtracted and then slope calibrated. Day-to-day variation of light intensity and illumination pattern led to variations in the basal level intensity, which was accounted for background subtraction. A linear scaling, slope calibration method, which preserve the slope ratio (SR), was employed for better demonstration and comparison. The slope calibration was done in the single agent/ modulator experiments, using the linear property of the MitoSOX-induced intensity rate (for 20 min interval after administration of MitoSOX and before addition of the metabolic agent). The slope of the linear fit of the intensity profiles was calculated in the MitoSOX interval (t = 11-29 min) for both control and treated groups. The MitoSOX-interval slope of the cell intensity profile in both control (no treatment) and treated groups were calibrated based on the MitoSOX-interval slope of a control in such a way that their slopes at this interval and before adding a metabolic agent were the same; the difference in the slopes of the control and treated groups were distinguishable after adding the agent. Therefore, the resulting intensity profiles are visually comparable for control and treated groups in the time interval of t = 30-80 min, which is helpful to evaluate the effect of the metabolic agent on the intensity profile. As mentioned earlier, for the dual agent experiments, the second agent, rotenone, antimycin A or KCN was administrated 5 mins after addition of PCP. Since the objective of the dual agent experiments is the evaluation of the inhibitor effect in the presence of the PCP, the slope calibration was applied based on the PCP-interval slope (t = 33-35 min) in such a way that their slopes at this interval and therefore before adding the second agent were the same. The resulting intensity profiles of the dual agent groups were compared to the fluorescent intensity profiles of the PCP only group in the time interval of t = 35-80 min, to evaluate the effects of ETC inhibitors (ROT, AA and KCN) on the rate of fluorescent change after maximally oxidizing the ETC with PCP.

Quantification of the kinetics of superoxide production
Exponential and sigmoidal empirical functions were used for quantitative description of the red fluorescence time-intensity curves measured in treated and control cells, respectively.
For the exponential function (Eq. (1) (2) to the red fluorescence timeintensity profiles from treated, and control groups, respectively.
The estimated values of the above parameters can be used to quantify the initial surge in • − production in the presence and absence of the metabolic agents, respectively.

Statistical analysis
Data are shown as means ± SE. Student's t-test was used for normally distributed data. A p value < 0.05 was considered significant.

Results
Co-localization of MitoTracker green signal with the MitoSOX red signal in PAECs confirmed that O 2 • − anions were produced from mitochondria in cells exposed to metabolic stress conditions. The degree of co-localization of these two fluorescent signals in the PAECs was 0.91 ± 0.06, indicating significant mitochondrial localization of the O 2 • − production.
Viability of PEACs treated with MitoSOX only, MitoSOX plus PCP, and MitoSOX plus PCP and KCN was assessed at the end of the 80 min experimental protocol. The percentages of cells that were viable following the aforementioned treatments were. 88 ± 2%, 76 ± 3%, and 80 ± 3%, respectively.
The top panel in Fig. 2  production when compared to the KCN treated cells. Enhanced fluorescence intensity was evident right after addition of the metabolic agents, and the intensity increased continuously in an exponential manner over time, demonstrating a time-dependent amplification of ROS production.
To quantify the rate of O 2 • − production in the absence and presence of inhibitors of the ETC or the uncoupler, PCP, (t = 30-80 min), sigmoidal and exponential empirical equations were fitted to the experimental data (control and treated cells). Figure 3 shows the initial slope of the intensity profile right after addition of the uncoupler/inhibitor. These nonlinear fits demonstrate that treating the cells with an oxidizing agent ( To confirm our method for localizing the mitochondrion as the source of O 2 • − production under metabolic stress conditions, a dual agent exposure protocol was designed. This method was devised to partition the O 2 • − production from different parts of the ETC. After oxidizing the ETC with PCP, the ETC was reduced with the different ETC complex inhibitors (complexes I, III and IV) to tease out the complex/s responsible for the surge in O 2 • − production when PCP fully oxidized the ETC. The curves in Fig. 5 represent the profiles of mean intensity in each time point for the corresponding groups. The green curve displays PCP-induced intensity profile compared to the orange, red, and purple curves, which demonstrate the effects of the addition of ROT, AA, and KCN to the PCP treated cells, respectively. The SR was calculated for all four groups, with n = 6 per group. Bar graphs in Fig. 5 show the means and SEM of the SR of the red fluorescence intensity for each of the four groups of cells. When compared to the addition of ROT and AA (orange and red bars, respectively), the addition of KCN to PCP-treated PAECs (purple bar) resulted in a significant reduction (p = 9.8026e-5), 62.7%, in the rate of the fluorescence intensity as compared to the fluorescence intensity of PAECs treated with PCP alone (green bar). This result shows that only KCN was able to significantly attenuate the O 2 • − production by the uncoupling agent PCP.
To provide further evidence that the surge in O 2 • − production by PCP was mainly from mitochondria, the mitochondrial-targeted ROS scavenger, MitoTempol was used. The dual agent exposure protocol was used to validate mitochondria as the primary source of O 2 • − production during real-time O 2 • − oxidation of MitoSOX, using the uncoupler PCP followed by MitoTempol. Figure 6 shows fluorescence intensity profiles in untreated cells (control; blue curve) and PCP treated cells in the absence and presence of MitoTempol over time (green and red curves, respectively). As observed previously, adding PCP at t = 30 min

Discussion
This study aimed to quantify mitochondrial O 2 • − production dynamics in cultured endothelial cells undergoing metabolic stress, utilizing fluorescent time-lapse microscopy.
Experimental protocols were designed to (1) localize the source of the mitochondrial O 2 production is favored in general by high ΔΨm and large NADH(H + ), or when electron transfer is impeded by alteration in the ETC complexes. 6,47 In this scenario, a decrease in ROS production would portend ΔΨm depolarization due to enhanced electron transfer, as observed with uncoupling agents. Paradoxically, though, conditions have been reported in which mitochondrial uncoupling and ΔΨm dissipation are associated with increased production of ROS. 48 This observation is consistent with our results which show PCP, an by the hypothesis that physiological signaling ROS occurs within an optimized redox state, and oxidative stress can happen at the extreme of either reduction or oxidation of the ETC.
Consistent with this hypothesis, our results demonstrated higher O 2 • − levels and rates of production in both the reduced (inhibited) and oxidized (uncoupled) ETC. A plausible explanation is that the elevated levels of O 2 • − initially overwhelm the scavenging potential of mitochondrial antioxidant system and leads to excess ROS emission (oxidative stress).
Complexes I and III are fully reduced when they are blocked with ROT and AA, respectively. The highly reduced redox state creates a buildup of electrons along the ETC, complexes I and III. [64][65][66] Pentachlorophenol (PCP) is a powerful uncoupler of oxidative phosphorylation and also induces oxidative stress to cause mitochondrial damage. [67][68][69][70] PCP dissipates the proton gradient by consuming the proton motive force via increased proton leak back into the matrix. A recent study reported that PCP oxidizes the ETC and increase the activity of complex I, 71 leading to the production of more protons and transfer of more electrons along the ETC. Lack of proton gradient for phosphorylation and polarized ΔΨm, activates a mechanism to compensate for the uncoupling effect of PCP, by increasing proton pumping and respiration, in an attempt to reestablish the proton gradient. Ironically, the increase in activity of the complexes in the uncoupled chain, increases electron transfer along the ETC, and as a result, increases electron leak to O 2 leading to O 2 • − production. 49,[72][73][74][75] This notion is supported by our observation that uncoupling respiration with PCP led to the fastest and greatest increase (Fig. 3) in the mitochondrial O 2 • − production. The finding also suggests that the increased protonophoric effect of the uncoupler (electron leak) leads to a feedforward mechanism that exacerbates ROS production.
Administration of the metabolic stressor/modulators (PCP, KCN, AA, or ROT) in PAECs induces two phases of O 2 • − production (Figs. 2 and 3). The initial phase is marked by immediate increase in O 2 • − production following the addition of metabolic agents (surge phase); this is followed by smaller rate of O 2 • − production (decrease in slope intensity) and eventually reaches the steady-state phase. This steady-state phase is consistent with the view that increased O 2 • − level may induce MnSOD activity as a negative feedback mechanism to dismutate the excess O 2 • − and therefore decrease the rate of O 2 • − production over time 76   Consistent with the redox-optimized ROS balance hypothesis, our method has the potential to model redox as ROS modulator and confirm the important role that redox modulation plays in controlling ROS production and potentially, ROS-mediated mitochondrial dysfunction and concomitant cellular injury. The single agent experiments showed that reduction (inhibition) or oxidation (uncoupling) of ETC leads to exponential increase in O 2 • − production, with more pronounced effect in the presence of PCP (Fig. 3(a)). Therefore, any shift toward oxidation or reduction leads to an increase in the rate of O 2 • − production.
As Eq. (3) shows, this ROS production is proportional to the difference of the redox state (R) from its optimal value (Ropt).

ROS | R − Ropt | . (3)
In addition, the dual agent experiments demonstrated the shift in the redox state towards oxid ation by PCP, and the reversal of the O 2 • − by KCN suggests that the redox optimized balance could be modulated by targeting specific ETC complexes. In this scenario, the redox state (R) of the uncoupled ETC becomes more reduced after adding KCN and moves toward optimal redox state and attenuates ROS production. This is a conundrum! The relationship between ROS and redox state provided in Eq. (3) attempts to resolve this conundrum. That is, even though KCN potentially increases ROS production, paradoxically, in the presence of PCP, it decreases the uncoupling effect on increased ROS production.
Since mitochondria are the major source of O 2 • − production during the uncoupling of respiration with PCP, we also examined whether the initial surge in O 2 • − production was amenable to the mitochondria-targeted ROS scavenger, Mito-Tempol. This agent is hitched to the cationic agent TPP+, which pulls the scavenger into mitochondria where it acts as an effective SOD mimetic. The dynamic of mitochondrial O 2 • − scavenging in the presence of MitoTempol (the growing space between the green and red profiles in Fig. 6) confirms that the initial surge in mitochondrial O 2 • − production following uncoupling of respiration with PCP is primarily mediated by deranged electron transfer in the ETC. This is also consistent with the observation that complex IV might be the source of electron leak that contributes to the surge in O 2 • − generation in the initial phase, i.e., when production exceeds the scavenging potential of MnSOD. We further verified that the primary source of the O 2 • − anion during simulated metabolic stress is mitochondria, because the NOX inhibitor, apocynin, did not alter the cellular ROS production (data not shown).
In conclusion, we used a novel approach that enables us to, for the first time to the best of our knowledge, partition and quantify the dynamics of mitochondrial O 2 • − production from ETC complexes under different simulated metabolic stress conditions in intact live PAECs. Nanomolar Mito-SOX, to minimize its toxic effects, was used to monitor the rate of mitochondrial O 2 • − production in the PAECs, which was attenuated by the mitochondriaspecific ROS scavenger (Mito-Tempol). We believe this approach could have far-reaching implications for assessment of ROS in physiology and pathophysiology. Therefore, accurate detection of mitochondrial ROS would allow us to establish a diagnostic tool for assessing the role of mitochondrial oxidative stress in the pathogenesis of diseases.

Limitations
We relied on the concept that MitoSOX is oxidized by O 2 • − and the oxidation product becomes highly fluorescent upon subsequent binding to mitochondrial DNA over time. 21 It is possible that as a cationic molecule, MitoSOX uptake into mitochondria can also contribute to direct ROS production by depolarizing ΔΨm. However, the ROS generated in this case was miniscule when compared to the ROS produced from modulating ETC complexes. While nanomolar MitoSOX (used in this study) had no adverse effects, 42  It should also be noted that MitoSOX uptake increases 10 fold for every 60mV increase in ΔΨm. 21,77 Since ΔΨm decreases in the presence of uncoupler 78,79 and ETC inhibitors, 78,80 the decrease in ΔΨm could impede the uptake of MitoSOX. Therefore, the fluorescence intensities presented could be an underestimate for the levels of the ROS in the inhibited and uncoupled ETC.
Interaction of O 2 • − and MitoSOX is also affected by the pharmacokinetics of MitoSOX and its binding properties. The initial phase of the intensity profile shows a greatly enhanced fluorescence due to binding of the oxidized MitoSOX to mitochondrial DNA. In the steady state phase of the fluorescence recording, the steady-state signal profile can be possibly due to binding of oxidized MitoSOX to nuclei, which results in apparent nuclear and nucleolar localization ( Fig. 2(a)). Due to nuclei binding of MitoSOX and thus, the subsequent diminished mitochondrial uptake of the fluorescent probe, the fluorescence intensity of the mitochondrial compartment becomes saturated and after a while reaches the final level.