EPR Detection of Cellular and Mitochondrial Superoxide Using Cyclic Hydroxylamines

Cell culture and mitochondria isolation

Bovine aortic endothelial cells (BAEC, passage 4 to 8) were cultured on 100 mm plates in Media 199 containing 10% fetal calf serum supplemented with 2 mM L-glutamine, and 1% vitamins. Confluent cells were used for the experiments.²⁶ Human aortic endothelial cells (HAEC) purchased from Lonza (Chicago, IL) and cultured in EGM-2 medium supplemented with 2% FBS but without antibiotics. On the day before the study, the FBS concentration was reduced to 1%. HAEC were transfected with GFP or SOD2 plasmid was described previously.³¹ Rat aortic smooth muscle cells were purchased from Cell Systems (Kirkland, Wash). Neutrophils were isolated from whole blood by Percoll density gradient centrifugation as previously described.³³

Brain mitochondria were isolated from the male Lewis rat pooled forebrain. We used the modified method of Sims³⁴ to isolate and purify brain mitochondria (RBM) in a Percoll gradient. Endothelial mitochondria were isolated using digitonin method as previously described.³⁵

All animal use complied with National Institutes of Health guidelines and was approved by the Emory University Institutional Animal Care and Use Committee.³⁶

Superoxide measurement

The cell-free xanthine oxidase superoxide-generating system contained xanthine oxidase (1 mU/ml), xanthine (50 μM), and DTPA (0.1 mM) in 50 mM sodium phosphate buffer (pH=7.4) containing 0.9% NaCl.³⁷

Measurement of mitochondrial O₂^• was performed in media containing 125 mM KCl, 10 mM MOPS, 2 mM MgSO₄, 2 mM KH₂PO₄, 10 mM NaCl, 1 mM EGTA, and 0.7 mM CaCl₂, 50 μM desferoxamine, pH=7.2. Mitochondrial O₂^• production was studied in the presence of 2 mM malate + 20 mM glutamate and 20 μg of mitochondrial protein.²⁹ Amount of detected O₂^• was calculated from inhibition of EPR signal with 50 U/ml Cu, Zn-SOD. All mitochondria measurements were carried out at 25 °C using a Bruker EMX spectrometer equipped with Temperature Controller system.

Neutrophils were incubated with 1 mM of corresponding cyclic hydroxylamine and stimulated by the addition of 1 μM PMA in Chelex-treated phosphate buffer, pH=7.4.²⁷

Production of O₂^• in bovine aortic endothelial cells (BAEC) was measured in Krebs–Hepes buffer (KHB) containing 5.786 g/L NaCl, 0.35 g/L KCl, 0.368 g/L CaCl₂, 0.296 g/L MgSO₄, 2.1 g/L NaHCO₃, 0.142 g/L K₂HPO₄, 5.206 g/L Na-Hepes, 2 g/L D-glucose, pH=7.35, in the presence of 25 μM desferoxamine and 2.5 μM diethyldithiocarbamate. Cellular production of O₂^• was stimulated by mitochondrial inhibitor antimycin A (5 μM)^16,29 or pretreatment of cells with peroxynitrite donor SIN-1.²⁶ Confluent endothelial cells were treated with 1 mM SIN-1 for 60 min at 37 °C, and then washed with PBS. Superoxide production was analyzed in cell suspensions incubated at room temperature with hydroxylamine spin probes (1 mM) or the spin trap EMPO (50 mM). The rate of O₂^• formation was in xanthine oxidase system, mitochondria, neutrophils or BAEC was measured by monitoring the amplitude of the low-field component of the EPR spectrum as previously described.¹¹ The concentration of nitroxide was determined from a calibration curve for intensity of the EPR signal of 3-carboxyproxyl at various known concentrations. The rate of O₂^• production was calculated from the accumulation of nitroxide, obtained from the EPR time scan. For this purpose, the EPR kinetics was analyzed using linear regression and WinEPR software (Bruker Biospin Corp, Billerica, MA). EPR settings were as follows: modulation amplitude, 2 G; microwave power, 20 mW; conversion time, 1.3 s; time constant, 5.2 s. Samples were scanned immediately after supplementation of spin probes unless stated otherwise.

Production of intracellular O₂^• in intact human aortic endothelial cells (HAEC) was stimulated with PMA (5 μM, 30 min). Generation of mitochondrial O₂^• was induced by rotenone (5 μM, 30 min). Following treatment with PMA, rotenone or vehicle, HAEC were incubated with 0.5 mM spin probes for 20 minutes at 37 °C. Then cells were collected and placed into a 1 ml syringe with 0.6 ml buffer and the suspension was snap-frozen in liquid nitrogen. Samples were analyzed in finger Dewar vessel filled with liquid nitrogen.⁶ ESR spectra were recorded using the following ESR settings: field sweep, 80 G; microwave frequency, 9.39 GHz; microwave power, 2 mW; modulation amplitude, 5 G; conversion time, 327.68 ms; time constant, 5242.88 ms; 512 points resolution and receiver gain, 1×10⁴.

Analysis of cell permeability of hydroxylamine probes

Cell permeability of hydroxylamine probes was determined in cellular lysates. For this purpose confluent rat aortic smooth muscle cells were subjected to 20-minute incubation with corresponding hydroxylamine (1 mM) at 37 °C. Then cells were centrifuged and supernatant discarded. Cells were resuspended in PBS and lysated by ultrasound. Spin probe cellular content was determined by oxidation of cell lysates with NaIO₄ (50 mM) to convert cyclic hydroxylamines into nitroxides detected by EPR.³⁸

To determine partition coefficients, solutions of the hydroxylamines (1 mM) in 50 mM phosphate buffer saline (pH 7.4) were prepared. The portions of the solutions were extracted with equal amount of octanol. The concentrations of hydroxylamine probes in water phase were calculated from EPR intensities of samples treated with NaIO₄ (50 mM) to convert cyclic hydroxylamines into nitroxides. The partition coefficients were determined from a ratio of hydroxylamine concentrations in the phosphate buffer before (I₀) and after (I) a single extraction with equal amount of octanol, K_p =(I₀/I)−1. EPR measurements were performed in 50 μL glass capillary tubes at room temperature using the Bruker EMX spectrometer. Spectrometer settings were as follows: field sweep, 60 G; microwave frequency, 9.82 GHz; microwave power, 20 mW; modulation amplitude, 1 G; conversion time, 164 ms; time constant 328 ms; sweep time, 168 s; receiver gain, 1×10⁵; number of scans, 4.

Reaction of hydroxylamine spin probes with O₂^•

All the hydroxylamine spin probes readily reacted with O₂^• produced in a cell-free xanthine oxidase system.³⁷ Figure 2 demonstrates comparison of O₂^• detection using the spin trap EMPO (50 mM) and the spin probe TMT-H (1 mM). Spin trapping of O₂^• by EMPO produced the relatively stable EMPO/^•OOH radical adduct with the characteristic six-line EPR spectrum (Figure 2A, insert). The accumulation of EMPO/^•OOH radical adduct was monitored by following the intensity of the low-field component of its EPR spectrum. The rate of accumulation of the radical adduct (130 nM/min) decreased by more than 60% after 6-minute data acquisition (Figure 2A, XO+X). Addition of Cu, Zn-SOD (50 U/ml) to xanthine/xanthine oxidase completely blocked the accumulation of the radical adduct (Figure 2A, XO+X+SOD).

Reaction of TMT-H with O₂^• produced the stable nitroxide radical 4-isobutyramido-TEMPO (TMT), which has a three-line EPR spectrum (Figure 2B, insert). Addition of Cu, Zn-SOD (50 U/ml) to the reaction mixture blocked nitroxide accumulation (Figure 2B, XO+X+SOD). Interestingly, there was no significant difference in the initial rates of accumulation of the EMPO/^•OOH and TMT, but after 6-minutes the difference in the rates of product accumulation was more than two-fold (Figure 2). As a result, concentration of TMT was more than 60% higher than concentration of EMPO/^•OOH after 11 minute scan, which is likely due to the lower stability of the EMPO/^•OOH radical adduct compared to TMT nitroxide.

Other cyclic hydroxylamine probes (CAT1-H, PP-H, CP-H and CM-H) showed similar accumulation of nitroxides in the xanthine oxidase system and supplementation of SOD blocked accumulation of nitroxides in a dose-dependent manner (Figure 3) while catalase did not affect kinetics of nitroxide production (data not shown). Competition of SOD and hydroxylamines was used to measure the rate constant of the spin probe reaction with O₂^•. Figure 3 shows the linear graphs of V/V_o-1 (where V and V_o represent the rates of nitroxide accumulation in the presence or absence of SOD, respectively) as a function of SOD activity as was previously described.^23,39 The slope of the graph reflects the ratio of k_SOD/k_Probe, therefore, higher slope corresponds to lower rate constant. Among the hydroxylamines studied, the positively charged mitoTEMPO-H and CAT1-H show the highest rate constants of reaction with O₂^•, while the negatively charged PP-H shows the smallest rate constant. The ratio of the slopes equals to the ratio of the apparent reaction rate constants of corresponding hydroxylamine spin probes with superoxide. Taking into account the value of the rate constant for the reaction of CP-H with O₂^• reported earlier (3.2×10³ M⁻¹s⁻¹)²³ the corresponding rate constants for TM-H, TMT-H, CM-H, mitoTEMPO-H and CAT1-H were estimated (Table 1). The reactivity of HO₂^• is known to be higher than that of O₂^•.39 Moreover, cyclic hydroxylamines are known to be bases, with pK varying in a broad range (5÷8), depending on electronic and steric effects of the substituents, and the reactivity of free hydroxylamine base may differ strongly from that of hydroxylammonium cation.⁴⁰ Therefore, the apparent reaction rates of the hydroxylamine probes with O₂^•/HO₂^• are expected to demonstrate complex pH-dependence, which deserves a separate study. Finally, it is important to note that correct determination of “absolute values” for the reaction rate constants from competition kinetics using biological sources of superoxide may be complicated. Therefore, the apparent rate constants presented in the table 1 should be taken primarily for the ranking purpose and considered as a semi-quantitative.

Lipophilicity and cell permeability of spin probes

Cell permeability of the spin probes may affect the outcome of the cellular O₂^• detection. Lipophilicity is an important factor modulating cellular permeability of the cyclic hydroxylamines, and the partition coefficient in a water/octanol mixture is a standard measure of lipophilicity. The partition coefficients of the spin probes were determined in octanol/phosphate buffer saline at pH=7.4 mixture as octanol/aqueous phase concentration ratio. Relatively high lipophilicity of CM-H, TMT-H and TM-H (Table 1) is an indication of their ability to cross cell membranes. Low lipophilicity of CP-H most likely results from ionization of the carboxylic group (pK_a is ca. 5) at physiological pH values. However, ionization of the carboxylic group only attenuates the cell permeability but does not abolish it completely. Ionic PP-H and CAT1-H showed the lowest lipophilicity.

We studied cell permeability of the hydroxylamine spin probes in cultured rat aortic smooth muscle cells. For this purpose rat aortic smooth muscle cells were subjected to 20-minute incubation with corresponding hydroxylamine probe. Then cells were centrifuged and cellular content of the probe was determined by oxidation of cell lysates with NaIO₄ (50 mM) to convert cyclic hydroxylamines into nitroxides detected by EPR.³⁸ We found that CM-H, PP-H, TM-H and TMT-H accumulated in the cellular cytoplasm (Figure 4A-E). No significant accumulation of CAT1-H in the cells was observed (Figure 4F).

Lipophylic triphenylphosphonium cation is known to be cell-permeable and provide mitochondria-targeted accumulation.⁴¹ Indeed, incubation of cells with 50 μM mitochondria-targeted spin probe mitoTEMPO-H resulted in significant cellular accumulation of mitoTEMPO (Figure 4C). The concentration of mitoTEMPO in cellular lysate was close to nitroxide concentration in lysates of cells incubated with 1 mM TM-H or TMT-H, thereby supporting active cellular accumulation of mitoTEMPO-H. Indeed, mitochondria isolated from mitoTEMPO-H treated cells showed strong EPR spectra (data not shown).

It has been previously suggested that hydrophilic nitroxides do not diffuse through cellular lipid membranes. The absence of cell permeability for CAT1-H is in line with this assumption. However, PP-H unexpectedly demonstrates high capacity of cellular accumulation similar to neutral CM-H. Our data (Figure 4) suggest that cells actively import PP-H through some unknown mechanism. The mechanism of PP-H accumulation by cells could be similar to the active transport of phosphate.⁴² This process is ATP-dependent and mitochondrial uncoupling should inhibit accumulation of PP-H. Indeed, we found that accumulation of PP-nitroxide in vascular cells was inhibited by antimycin A and PP-nitroxide was not accumulated in red blood cells which are deficient of mitochondria (data not shown).

Detection of extracellular O₂^• in human neutrophils

Previously, we have shown that human neutrophils produce predominantly extracellular O₂^• both in basal and PMA-stimulated conditions.²⁷ In this work we compared O₂^• detection by spin probes and the spin trap EMPO in human neutrophil suspensions. Production of O₂^• was monitored by accumulation of nitroxides or EMPO radical adducts which were followed by the increase in the low-field component of their EPR spectra. The rate of nitroxide accumulation significantly increased upon stimulation of neutrophils with PMA. Specificity of O₂^• detection by spin probes was confirmed by inhibition of EPR signal with SOD supplementation (Figure 5). The rate of CM-H oxidation into nitroxide was 20% higher than that of spin probes PP-H, TM-H and CAT1-H, while accumulation of the EMPO radical adduct was 2 fold lower than nitroxide formation from the spin probes (Figure 5, table). These differences in O₂^• detection in PMA-stimulated neutrophils correlated with the cell-free experiments in xanthine oxidase O₂^• generating system (Figures 3 and 5).

Detection of O₂^• in intact endothelial cells treated with the peroxynitrite donor SIN-1

It has been previously reported that that exposure of endothelial cells to peroxynitrite leads to oxidation of tetrahydrobiopterin (BH₄), which is a crucial intracellular cofactor of eNOS (Figure 6, scheme).⁴³ Depletion of cellular BH₄ results in uncoupling of endothelial NO synthase (eNOS), which then produce O₂^• instead of NO.^25,43 Nitric oxide synthase inhibitor L-NAME blocks O₂^• production by uncoupled eNOS,^25,43,44 which is likely due to inhibition of heme domain of eNOS.⁴⁵ Uncoupling of eNOS by peroxynitrite donor SIN-1 is a well-known model of endothelial dysfunction which increases both intra- and extracellular O₂^• production in endothelial cells.^25,26 We have investigated production of endothelial O₂^• after eNOS uncoupling using hydroxylamine probes and the spin trap EMPO.

Both EMPO and the hydroxylamine probes detected O₂^• produced by BAEC (Figure 6, table1 ). Interestingly, CAT1-H and EMPO formed similar amounts of nitroxides. The nitroxides formation from CAT1-H and EMPO was completely suppressed by Cu, Zn-SOD supplementation, thus, it corresponds to basal extracellular production of O₂^•. Reactions with other spin probes gave stronger EPR signals, and supplementation with extracellular SOD did not inhibit nitroxide accumulation completely. Presumably, CM-H, PP-H, TM-H and TMT-H react with intracellular O₂^•. CM-H showed highest reactivity among hydroxylamine spin probes producing the highest amount of nitroxide (Figure 6, table1 ).

Uncoupling of eNOS was induced by 1-hour incubation of endothelial cells with 1 mM peroxynitrite donor SIN-1, the later was then washed out with PBS to avoid interference with cellular O₂^• detection. Cells were centrifuged and O₂^• production was analyzed in cell suspension.²⁶ Treatment of endothelial cells with the peroxynitrite donor SIN-1 significantly increased the production of cellular O₂^• detected both by hydroxylamine probes and EMPO. The contribution of the eNOS uncoupling in O₂^• production was estimated by decrease of nitroxide accumulation in the presence of eNOS inhibitor L-NAME (Figure 6).²⁶ Treatment with SIN-1 resulted in increase of the intensity of the EPR signal: 2-fold for CAT1-H and EMPO, 2.5-fold for TM-H and TMT-H, and 3-fold for PP-H (Figure 6, table 1). Supplementation of SIN-1-treated cells with Cu, Zn-SOD significantly reduced nitroxide accumulation, while addition of DMSO or uric acid did not affect oxidation of cyclic hydroxylamines. These data confirm that reaction of hydroxylamines with O₂^• is a major source of accumulated nitroxidein endothelial cells.

Detection of O₂^• in intact endothelial cells treated with antimycin A

Antimycin A stimulates the release of O₂^• into mitochondrial matrix and cellular cytoplasm.¹⁶ It has been recently reported that nitrone spin traps fail to detect intracellular O₂^•.16 In this work we have investigated antimycin A - induced O₂^• production in endothelial cells using hydroxylamine spin probes (Figure 7, table 1).

It was found that all spin probes detected O₂^• in BAEC under basal conditions. Incubation of BAEC with CM-H and PP-H produced the strongest EPR signal, and formation of corresponding nitroxides was only partially inhibited by extracellularly added Mn-SOD (50 U/ml). Amount of nitroxides formed due to reaction of O₂^• with TM-H and TMT-H was 3 times less compared to CMHor PP-H.

Treatment of endothelial cells with antimycin A led to major increase in CM-H oxidation rate, which was only partially suppressed by supplementation with extracellular Mn-SOD. Antimycin A caused moderate increase in O₂^•-induced oxidation of TM-H and TMT-H. In contrast, oxidation of PP-H decreased in antimycin A-treated cells (Figure 7). Antimycin A inhibits ATP synthesis, thereby blocking all energy-dependent processes including active transport. This may prevent accumulation of PP-H inside antimycin A-treated cells and decrease the rate of oxidation of PP-H. A passive cellular diffusion of CM-H, TM-H and TMT-H was not likely affected by antimycin A supplementation and, thereby, antimycin A did not decrease the ability to scavenge intracellular O₂^• in by these probes.

Interestingly, detection of cellular O₂^• in BAEC by cell impermeable CAT1-H was completely blocked by supplementation with extracellular Mn-SOD both in basal conditions and upon stimulation with antimycin A (Figure 7, table 1) indicating that antimycin A increased not only intracellular but also extracellular O₂^• levels.

Detection of mitochondrial O₂^• in isolated mitochondria

Small fractions of electrons in the mitochondrial respiratory chain can leak out and reduce oxygen, producing O₂^• radicals.⁴⁶ Superoxide radicals formed on the inner membrane of the matrix are quickly dismutated by Mn-SOD, and therefore, cannot be detected outside of mitochondria, but can be estimated indirectly by leakage of H₂O₂from mitochondria.

Superoxide produced on the outer membrane of the matrix can be released by mitochondria and detected as extramitochondrial O₂^•.29 Previously, succinate-dependent mitochondrial production of O₂^• has been studied using DMPO and DEPMPO. In these experiments formation of the radical adducts was completely prevented by supplementation with extramitochondrial SOD.^14,47 However, identification of mitochondrial O₂^• using spin trapping was complicated because of conversion of primary OOH-adduct into secondary OH-radical adduct by glutathione peroxidase and reduction of radical adducts.^13,14,15 Hydroxylamine spin probes can react with mitochondrial O₂^• producing stable nitroxides which can be resistant to mitochondrial reduction. In this work we have investigated production of O₂^• in isolated mitochondria under basal condition and stimulated with antimycin A or rotenone using various hydroxylamine spin probes and EMPO.

No EMPO radical adducts formation was observed in rat brain mitochondria under basal conditions in the presence of malate+glutamate (Figure 8A). However, stimulation of mitochondrial O₂^• production with antimycin A resulted in significant accumulation of EMPO/^•OH radical adduct, which was completely blocked by supplementation with extramitochondrial SOD.

In contrast, TMT-H has been able to detect basal production of O₂^• in rat brain mitochondria with malate+glutamate (Figure 8B, RBM vs background). Antimycin A further stimulated SOD-inhabitable accumulation of nitroxide (Figure 8B, RBM+AA vs RBM+AA+SOD). Interestingly, in antimycin A stimulated mitochondria TMT-H gave twice higher EPR signal compared to EMPO (Figure 8A, B).

We have previously described the detection of O₂^• in mitochondria using PP-H.²⁹ Indeed, supplementation of PP-H to isolated mitochondria resulted in significant accumulation of nitroxide which was further increased by antimycin A (Figure 9A). However, only a small fraction of the EPR signal was inhibited by SOD under basal conditions and PP-H showed a rather high background oxidation (Figure 9A).

We have recently found that CAT1-H is a very stable probe with minimal background oxidation.³⁰ Supplementation of CAT1-H to mitochondria resulted in accumulation of CAT1 nitroxide both in basal and antimycin A stimulated conditions (Figure 9B). These experiments showed that CAT1-H can be successfully used for detection of mitochondrial O₂^• under basal conditions as well as upon stimulation by antimycin A. The most important difference between PP-H and CAT1-H was observed in the detection of O₂^• under basal conditions. Supplementation of SOD decreased the accumulation of CAT1 by the factor of four due to low background oxidation of CAT1-H, while PP-nitroxide signal was reduced only by 25 %. It is interesting to note, that antimycin A did not influence PP-nitroxide accumulation in presence of SOD, indicating that intramitochondrial O₂^• does not participate in PP-H oxidation. Thus, similarly to CAT1-H, PP-H and does not penetrate mitochondria and detects only extramitochondrial O₂^•.

In order to measure O₂^• production in mitochondrial matrix we used mitochondria-targeted probe mitoTEMPO-H. Addition of mitochondria isolated from BAEC to mitoTEMPO-H significantly increased EPR signal (Figure 9C). Both rotenone and antimycin A further increased accumulation of nitroxide. As expected, supplementation of SOD did not affect rotenone-induced nitroxide accumulation and only partially attenuated EPR signal indicating that reaction with O₂^• proceeds inside of mitochondrial matrix.

Detection of intracellular and mitochondrial O₂^• in intact cells

The above data indicate that that mitoTEMPO-H and CM-H can penetrate the cells and react with mitochondrial O₂^• (Figure 4, 7, 9). In contrast to CMH, mitochondria-targeted mitoTEMPO-H is actively accumulated in the mitochondrial matrix similar to other triphenylphosphonium cations⁴¹ while CM-H may only passively diffuse across lipid membrane. Another probe, CP-H, can permeate the cells but it is mitochondria impermeable.⁴⁸ It is conceivable that specific intracellular localization of these spin probes will allow detection of O₂^• in distinct cellular compartments: CP-H in the cytoplasm, mitoTEMPO-H in mitochondria and CM-H in both compartments. In this work we have stimulated O₂^• production in the cytoplasm with NADPH oxidases activator PMA while generation of O₂^• in the mitochondrial matrix was induced by complex I inhibitor rotenone.²⁹ The site specific production of O₂^• in cytoplasm and mitochondria was investigated using 0.5 mM CP-H, 0.5 mM CM-H or 50 μM mitoTEMPO-H in intact human aortic endothelial cells (Figure 10) by EPR analysis of frozen samples.⁶ In order to confirm detection of mitochondrial O₂^• we used specific overexpression of mitochondrial isoform of superoxide dismutase SOD2 while transfection of HAEC with GFP plasmid was used as a control.

Treatment of HAEC with PMA increased nitroxide accumulation in cells incubated with CP-H or CM-H but not with mitoTEMPO-H (Figure 10 A,B,C). Stimulation of mitochondrial O₂^• by rotenone increased nitroxide accumulation in cells incubated with CM-H or mitoTEMPO-H but not with CP-H (Figure 10 A,B,C). Contribution of mitochondrial O₂^• in EPR measurements of cellular O₂^• was determined by decrease of EPR signal in HAEC transfected with SOD2 plasmid compared to cells transfected with control GFP plasmid. EPR analysis of transfected HAEC showed that SOD2 overexpression did not affect measurements with CP-H (Figure 10D) but significantly attenuated measurements with CM-H and mitoTEMPO-H. Overexpression of mitochondrial SOD2 has significantly decreased detection of basal and rotenone-induced O₂^• by CM-H or mitoTEMPO-H (Figure 10 E,F).