Rapid mitochondrial repolarization upon reperfusion after cardiac ischemia

Abstract

The mitochondrial membrane potential (ΔΨ_m) drives oxidative phosphorylation and alterations contribute to cardiac pathologies, but real-time assessment of ΔΨ_m has not been possible. Here we describe noninvasive measurements using mitochondrial heme b_L and b_H absorbances, which rapidly respond to ΔΨ_m. Multi-wavelength absorbance spectroscopy enabled their continuous monitoring in isolated mitochondria and the perfused heart. Calibration of heme b absorbance in isolated mitochondria revealed that reduced heme b_L relative to total reduced heme b (fb_L = b_L/(b_L + b_H)) exhibits a sigmoidal relationship with ΔΨ_m. Extrapolating this relationship to the heart enabled estimation of ΔΨ_m as 166 ± 18 mV (n = 25, mean ± s.d.). We used this approach to assess how ΔΨ_m changes during ischemia-reperfusion injury, an unknown limiting the understanding of ischemia-reperfusion injury. In perfused hearts, ΔΨ_m declined during ischemia and rapidly reestablished upon reperfusion, supported by oxidation of the succinate accumulated during ischemia. These findings expand our understanding of ischemia-reperfusion injury.

Fig. 1. Multi-wavelength absorbance spectroscopy in isolated mitochondria.

a, Mitochondrial oxidative phosphorylation. b, Upon reperfusion, succinate accumulated during ischemia is proposed to be oxidized by SDH, driving RET and superoxide production initiating ischemia–reperfusion injury. IMS, intermembrane mitochondrial space; MIM, mitochondrial inner membrane. c, Electron transfer within the cytochrome bc₁ complex and electron distribution between hemes b_L and b_H in response to ΔΨ_m. d, Structure of the bovine cytochrome bc₁ complex from PDB 1BE3. The distinct coordination environments of the heme groups in the bc₁ complex give rise to unique absorption spectra. e, Reference spectra including for reduced hemes b_L and b_H and cytochromes c₁ and c and for oxygenated (MbO) and deoxygenated (MbD) purified mouse myoglobin, as well as oxidized cytochrome bc₁ complex, were collected using a combination of biochemical techniques. Thus, relative optical density (OD) is arbitrary. f, Schematic of the integrating sphere system used to measure the optical absorbance of mitochondrial suspensions. TPMP⁺ distribution was measured simultaneously by a TPMP⁺-selective electrode. Substrates, inhibitors and uncouplers (S) and gases (G) were introduced via separate ports. g, Spectrum of incident light used to analyze mitochondrial suspensions. a.u., arbitrary units. h, Isolated mitochondria were incubated with succinate/rotenone, and a multi-wavelength nonlinear least squares regression was performed on the experimental absorbance spectrum using reference spectra for mitochondrial hemes and cytochromes and a line. i, The best fit and individual chromophore contributions. j,k, The linear component, incorporated into the fit to account for absorption by the integrating sphere and oxidized chromophores, was subtracted from the experimental spectrum (j) and best fit (k). l, The residuals of the best fit. Source data

Fig. 2. In vitro calibration of b heme absorbance against ΔΨ_m.

a–d, Experimental absorbance spectra and best fits of mitochondria incubated with succinate/rotenone and 30 mM malonate (a), ADP (b), BAM15 (c) or nigericin (d). After spectral fitting, the linear component was subtracted from all spectra and best fits. e, Mitochondria were incubated with succinate/rotenone (n = 14) and nigericin (n = 5), 30 mM malonate (n = 5) or ADP (n = 4). BAM15 was added at the conclusion of all experiments (n = 14). Heme b_L and b_H absorbance was quantified from corresponding spectral fits. f, fb_L was calculated from b heme absorbance. g, Representative time course of mitochondria incubated with succinate/rotenone and titrated with malonate (M) and BAM15 (B). A rolling average of five consecutive spectra was taken before multi-wavelength nonlinear least squares regression analysis and quantification of heme b_L and b_H. ΔΨ_m, calculated from the Nernstian distribution of TPMP⁺, and fb_L are plotted over time. h, fb_L and ΔΨ_m under each condition (n = 5 biological replicates). i–l, Mitochondria were incubated sequentially with succinate, nigericin, malonate and BAM15 (i); succinate and nigericin, malonate and BAM15 (j); succinate, ADP, malonate and BAM15 (k); and GM ADP, rotenone and BAM15 (l). fb_L and ΔΨ_m were determined under each condition shown. A line connects mean values. m, fb_L plotted as a function of ΔΨ_m for all data points derived from h–l. n, Average fb_L plotted as a function of average ΔΨ_m for all conditions shown in h–l (n = 23). Data were fit using nonlinear regression with a four-parameter sigmoidal model with variable slope (solid line), and 95% confidence intervals were calculated (dotted lines). o, Mitochondria were incubated with GM and deoxygenated by blowing N₂ over the suspension before reoxygenating 5 min later with O₂. A line connects mean values. p, fb_L and ΔΨ_m were determined at each steady state shown (n = 5). fb_L is plotted as a function of ΔΨ_m and superimposed onto the sigmoidal curve generated in n. q, Mouse heart mitochondria were incubated with GM followed by ADP, rotenone and BAM15, and fb_L and ΔΨ_m were determined under each condition (n = 5). These data are superimposed onto the sigmoidal curve generated in n. Unique colors are used to represent biological replicates. All data are shown as the mean ± s.d. Source data

Fig. 3. Measurement of ΔΨ_m in the heart.

a, Photograph of an isolated, perfused mouse heart with an optical light catheter (200 µm) inserted through the mitral valve and positioned in the left ventricle. b, Schematic of the integrating sphere system used to measure light transmittance across the left ventricular wall in real time. c, Incident light spectrum. d, The isolated perfused Mb^−/− mouse heart was perfused for 20 min, and a multi-wavelength nonlinear least squares regression was performed on the experimental spectrum using reference spectra for mitochondrial hemes and cytochromes and a line. e, The best fit and individual chromophore contributions. f, The residuals of the best fit. g, Data in d–f with the linear component subtracted from the experimental spectrum and best fit. h–j, Isolated perfused hearts from Mb^−/− mice were treated in sequence with adenosine (h), KCl (i) and FCCP (j) to alter ΔΨ_m. Absorbance spectra were fit by multi-wavelength nonlinear least squares regression as described above. After spectral fitting, the linear component was subtracted from all experimental spectra and best fits. k,l, Heme b_L and b_H (k) and cytochrome c (l) absorbances were quantified from the spectral fits (n = 9). m, fb_L was quantified from the absorbance of hemes b_L and b_H under each condition. All data are shown as the mean ± s.d. Due to one missing data point, a mixed-effects model with restricted maximum likelihood (REML) estimation was used, and fixed effects were analyzed using Dunnett’s multiple comparisons test to compare treatment to control. Source data

Fig. 4. Changes in ΔΨ_m within the heart during ischemia and reperfusion.

a, Schematic depicting the experimental protocol for global ischemia–reperfusion in the perfused heart. b,c, Hearts from Mb^−/− mice were subjected to global ischemia–reperfusion. Transmural absorbance was measured every second, and a rolling average of four consecutive spectra was taken before multi-wavelength nonlinear least squares regression analysis and quantification of hemes b_L and b_H (b) and cytochrome c (c) over time (n = 7). d, fb_L was quantified from the absorbances of hemes b_L and b_H, and the average fb_L was plotted over time. e, ΔΨ_m was calculated from fb_L using the calibration curve generated in isolated mitochondria, and average ΔΨ_m was plotted over time. f,g, fb_L (f) and ΔΨ_m (g) were determined at control, 10 min or 20 min of ischemia and 1, 2, 5, 10 or 20 min of reperfusion in isolated hearts from Mb^−/− mice (n = 7). NS, not significant. h, Isolated perfused Mb^−/− hearts were perfused normally (n = 7) or with glucose-free KH buffer supplemented with sodium acetate and glucagon (n = 5). ΔΨ_m was determined just before ischemia (control) or after 10 min of ischemia. i, Panel e replotted to highlight ΔΨ_m over the first 2 min of reperfusion. j–l, Isolated hearts from Mb^−/− were subjected to global ischemia–reperfusion and treated with the uncoupler BAM15 (n = 6) at the onset of reperfusion or the SDH inhibitor AA5 before ischemia (n = 5) or after ischemia (n = 5). fb_L (j) and ΔΨ_m (k) were calculated after 1 min of reperfusion, and fb_L was again calculated at 2 min of reperfusion (l). All data are shown as the mean ± s.d. Statistical comparisons in f,g,j–l were made using one-way ANOVA with Dunnett’s multiple comparisons test. Multiple unpaired two-tailed t tests were performed with Holm-Šídák correction for multiple comparisons to compare substrate conditions in h. Source data

Fig. 5. Changes in ΔΨ_m within the WT heart during ischemia and reperfusion.

Isolated perfused C57BL/6N mouse hearts were subjected to global ischemia–reperfusion (n = 9). a, Average transmural absorbance of isolated perfused hearts from WT or Mb^−/− mice under control conditions (n = 3 each). b, Transmural absorbance was determined under normoxic conditions, and a multi-wavelength nonlinear least squares regression was performed using reference spectra for hemes b_L and b_H and cytochromes c₁ and c, as well as MbO and MbD, and a line. c, The best fit and individual chromophore contributions. d, The residual of the best fit. e–g, The absorbances of hemes b_L and b_H (e), cytochrome c (f) and MbO and MbD (g) were quantified from the spectral fits at selected time points. Four data points from three biological replicates were excluded due to low spectral signal-to-noise ratio. h, fb_L was calculated as previously described. i, Comparison of ΔΨ_m at selected time points in isolated hearts from WT (n = 9) or Mb^−/− (n = 7) mice subjected to global ischemia–reperfusion. j,k, Isolated hearts from WT mice were perfused with KH buffer containing glucose only (n = 6) or glucose, lactate, pyruvate and butyrate (complex buffer) (n = 10) and subjected to global ischemia–reperfusion. Multi-wavelength nonlinear least squares regression was performed on the experimental spectra collected at selected time points, and fb_L (j) and ΔΨ_m (k) were calculated as previously described. Due to missing data points, a mixed-effects model with REML estimation was used in h, and fixed effects were analyzed using Dunnett’s multiple comparisons test to compare selected time points to control. Multiple unpaired two-tailed t tests with a Holm-Šídák correction were used to compare genotypes in i and substrate conditions in k. All data are shown as the mean ± s.d. Source data

Fig. 6. Metabolic changes during ischemia and reperfusion.

Isolated perfused C57BL/6N or Mb^−/− mouse hearts (n = 6) were subjected to global ischemia–reperfusion. Whole hearts were rapidly frozen at control, after 20 min of ischemia or the indicated times after reperfusion. a, Succinate was extracted from frozen tissue samples and measured by LC–MS/MS against a [¹³C₄]succinate internal standard. b, CoQH₂ and CoQ were extracted from frozen tissue, and the percent reduction was determined by LC–MS/MS by determining the proportion of CoQH₂ relative to total CoQ and CoQH₂. c, Succinate and CoQH₂ data from a and b were fit to interpolated curves and are plotted over time along with cytochrome c absorbance (Fig. 4c) and ΔΨ_m (Fig. 4i). d, Curves in c replotted to highlight reperfusion. The y axes of c and d represent the range of values observed in the heart in this study. Data presented in a and b are the mean ± s.d. Source data

Extended Data Fig. 1. Measurement of ΔΨm in the Mb^−/− heart during IR injury.

a–d, The isolated perfused Mb^−/− heart was subjected to global ischemia-reperfusion, and a multi-wavelength nonlinear least squares regression was performed on the experimental spectra collected under normoxic conditions (a), after 20 min of ischemia (b) and at 1 (c) and 20 (d) minutes of reperfusion using reference spectra for mitochondrial cytochromes and a line. After spectral fitting, the linear component was subtracted from all experimental spectra and best fits. e, Isolated perfused Mb^−/− hearts were perfused with glucose-free KH buffer supplemented with sodium acetate and administered glucagon (indicated by Glu) 3 min prior to ischemia (n = 5). Transmural absorbance was measured every second, and a rolling average of five consecutive spectra was taken prior to multi-wavelength nonlinear least squares regression analysis. Heme b_L and b_H absorbances were quantified, and fb_L was calculated over time. f–h, Isolated perfused Mb^−/− hearts received 2 µM BAM (n = 5) (indicated by Un) (f) at the onset of reperfusion or 20 nM AA5 (n = 5) (indicated by AA5) (g) 3 min prior to 20 min of global ischemia or 200 nM (n = 5) (indicated by AA5) (h) at the onset of reperfusion. fb_L was calculated over time as described above. All data are means ± s.d. Source data

Extended Data Fig. 2. Multi-wavelength absorbance spectroscopy in the isolated Wt heart.

a–d, Isolated perfused Wt mouse hearts (n = 9) were subjected to global ischemia-reperfusion, and a multi-wavelength nonlinear least squares regression was performed on experimental spectra collected under normoxic conditions (a), after 20 min of ischemia (b) and at 1 (c) and 20 (d) minutes of reperfusion. After spectral fitting, MbO, MbD and the linear component were subtracted from all experimental spectra and best fits. Source data