(A) Two-dimensional map of superoxide flashes in a cardiac cell. Yellow boxes mark locations of superoxide flashes detected during a 100 s-scan after 6 hr of anoxia and red boxes mark active sites ~5 min following reoxygenation. (B) Temporal diaries of superoxide flash incidence in three representative cells during anoxia (left panel) and reoxygenation (right panel). Data in the top rows correspond to the cell depicted in (A). (C) Averaged superoxide flash frequency during anoxia, 5 min after reoxygenation and 1 hr after reoxygenation, in the absence or presence of adenosine (100 µM, added 1 hr prior to anoxia). Data are mean ± SEM. n = 6–16 cells. *, P<0.05 versus all other groups; #, P<0.01 versus adenosine. (D) Time course of superoxide flash frequency during hypoxia (1% O₂ + 99% N₂) and early reoxygenation. Data are mean ± SEM. n = 10–51 cells. *, P<0.05 normoxia (N) versus hypoxia.

(A) Schematic of the cardiac-specific mt-cpYFP expressing vector used to generate mt-cpYFP transgenic mice. (B) Images of Langendorff-perfused hearts of wild type (WT) and mt-cpYFP transgenic mice (TG) under UV illumination. Note the green fluorescence of the TG heart. (C) Visualization of superoxide flashes in a freshly isolated ventricular myocyte from a TG mouse. Upper panel: Image of a representative myocyte. Lower panel: Time course of the mt-cpYFP signals (488 nm excitation) from mitochondria indicated in the image, showing two active and one quiescent mitochondria. Similar results were obtained in 12 myocytes from 3 TG mice. (D) Imaging of superoxide flashes in the Langendorff-perfused heart from a TG mouse. Upper panel: Illustration of experimental setting. Middle panel: 2-D image of cardiac myocytes in the myocardium of the beating heart from a TG mouse. Lower panel: time course of mt-cpYFP signals from two active and one quiescent mitochondria (box 3, illustrating absence of motion artifact). Similar results were obtained in 4 other hearts.

In this model, the mPTP opens stochastically in response to physiological ROS levels set by constitutive ROS production by the ETC. Opening of the mPTP causes depolarization of ΔΨ_m, dissipation of chemical gradients across the inner membrane, and mitochondrial swelling due to water movement as well as changes in inner membrane fluidity and rigidity, which diverts electrons of the ETC to ROS generation. This simple model explains many salient features of superoxide flash activity (e.g., requiring the activities of mPTP, ETC and ATP synthase, all-or-none behavior, sensitive to SOD mimetic and superoxide scavenger) and frequency-dependent modulation by maneuvers that alter mPTP and ETC activities. The model also predicts that superoxide flash incidence is not only an optical readout of physiological mPTP activity in living cells, but also serves as a biomarker of oxidative stress in individual mitochondria. OMM: outer mitochondrial membrane; IMS: intermembrane space; IMM: inner mitochondrial membrane.

(A) Confocal visualization of a single-mitochondrion superoxide flash in a rat cardiac myocyte. Upper panel: Confocal image of a cardiac myocyte expressing mt-cpYFP. The enlarged view shows dual excitation (488 and 405 nm) imaging of the superoxide flash at 2 s intervals. The area shown has dimensions of 2.2×1.7 µm². (B) Time course of the superoxide flash shown in A. (C and D) Depression of superoxide flash frequency (C) and amplitude (ΔF/F₀, D) by an SOD mimetic, MnTMPyP (50 µM), and a superoxide scavenger, tiron (1 mM). Data are mean ± SEM. n = 12–64 flashes from 10–16 cells. *, P<0.05; #, P<0.01; †, P<0.001 versus control. (E) Superoxide flashes in primary cultured hippocampal neurons. Arrows mark two ends of a spaghetti-shaped mitochondrion undergoing repetitive superoxide flashes. Images correspond to the designated time points (1–5). (F–I) Properties of superoxide flashes in different cell types. Frequency (F), amplitude (ΔF/F₀, where F₀ refers to basal fluorescence intensity, G), time to peak (T_p, H), and 50% decay time after the peak (T₅₀, I). Data are mean ± SEM. n = 57–409 events from 21–53 cells.

(A) Colocalization of the ΔΨ_m indicator TMRM and the superoxide indicator mt-cpYFP in cardiac mitochondria revealed by tri-wavelength excitation imaging. (B and C) Two types of ΔΨ_m depolarization were distinguished by the presence and absence of concurrent superoxide flashes. n = 83 events from 19 cells. (D) Permanent reduction in mitochondrial rhod-2 fluorescence coincides with the onset of a superoxide flash (n = 39 events in 8 cells). Scale bars in (B) apply to (B)–(D). (E) Opposing effects of mPTP activation by atractyloside (20 µM, n = 5 cells) and inhibition by either bongkrekic acid (BA, 100 µM, n = 16 cells) or cyclosporin A (1 µM, n = 15 cells) on superoxide flash properties. T_p, time to peak; T₅₀, time from peak to 50% decay. Data are mean ± SEM. *, P<0.05; †, P<0.001 versus control. (F) Knockdown of cyclophilin D (insert, CypD KD, n = 4) by two sets of siRNA constructs (siRNA1 and siRNA2) both significantly decreased superoxide flash incidence in neonatal cardiac myocytes. n = 53–57 cells. Data are mean ± SEM. *, P<0.05 versus control.

(A) Excitation and emission spectra for fully reduced (blue lines, 10 mM reduced DTT) and fully oxidized (red lines, 1 mM aldrithiol) cpYFP. Ex: Excitation spectra obtained at 515 nm emission; Em: Emission spectra at 488 nm excitation. An isosbestic point was identified near 405 nm excitation, permitting ratiometric measurement via dual wavelength excitation (488 nm/405 nm). (B) The increase of cpYFP fluorescence emission (at 488 nm excitation) when reduced cpYFP was exposed to xanthine (2 mM) plus xanthine oxidase (20 mU) under aerobic conditions and its complete reversal to the aerobic control level by subsequent addition of Cu/Zn-SOD (600 U/ml). (C) cpYFP emission was insensitive to H₂O₂ (0.1 and 10 mM) and slightly decreased by hydroxyl radical (produced by the Fenton reaction: 1 mM H₂O₂ plus 0.1 mM FeSO₄ under anaerobic conditions). (D) cpYFP emission was insensitive to physiological concentrations of Ca²⁺. (E) cpYFP emission was insensitive to NAD⁺ (10 mM) and NADH (1 mM). (F) cpYFP emission was insensitive to a broad range of redox potentials (−7.5 ~ −319 mV). Solutions of variable redox potentials were made by mixing reduced and oxidized DTT in different proportions (total DTT concentration 10 mM). The exact redox potential for each mixture was monitored by a redox sensitive electrode (Orion 91-80, Thermo Electron) at pH 8.0 and 20–23°C. The same solutions were then used for cpYFP spectral measurement.

(A–F), Absence of superoxide flashes in 143B cells completely devoid of mitochondrial DNA (ρ° 143B). Superoxide flashes exhibiting a reduction in TMRM fluorescence were readily observed in wild type 143B TK⁻ human osteosarcoma cells (WT 143B) as shown by fluorescence traces (A) and three representative diaries of superoxide flash incidence (each vertical tick denotes a flash event, C), but not in ρ° 143B cells (B and D) in spite of the presence of ΔΨ_m fluctuations (B). Atractyloside (Atra, 20 µM) did not rescue superoxide flash activity in ρ° 143B cells (E). Scale bars in (B) also apply to (A). (F) Statistics of superoxide flash frequency in WT and ρ° 143B cells. Data are mean ± SEM. n = 5–21 cells. (G) Attenuated superoxide flash activity in ETC-deficient cells. Upper panel: Inhibition of mitochondrial DNA replication by treatment of PC12 cells with ethidium bromide (EB, 200 ng/ml) for up to 60 days (referred to as ρ⁻ PC12 cells) resulted in a time-dependent decrease in the expression of the mitochondrial DNA-encoded cytochrome C oxidase subunit I (COX-1). Lower panel: a significant decrease in the frequency of superoxide flashes was observed in ρ⁻ PC12 cells. R&A: rotenone (5 µM) and antimycin A (5 µg/ml). Data are mean ± SEM. n = 16–46 cells. *, P<0.05 versus wild type (WT PC12) cells. #, P<0.01 versus –R&A. (H) Inhibition of superoxide flash activity in rat adult cardiac myocytes by rotenone (Rot, 5 µM), antimycin A (AA, 5 µg/ml), myxothiazol (Myx, 5 µM), sodium cyanide (CN, 5 mM), oligomycin (5 µM), and FCCP (300 nM). Data are mean ± SEM. n = 7–18 cells. †, P<0.001 versus control. Label of Y axis in (F) applies to (F)–(H).