Spectroscopic identification of the catalytic intermediates of cytochrome c oxidase in respiring heart mitochondria

Abstract

The catalytic cycle of cytochrome c oxidase (COX) couples the reduction of oxygen to the translocation of protons across the inner mitochondrial membrane and involves several intermediate states of the heme a₃-Cu_B binuclear center with distinct absorbance properties. The absorbance maximum close to 605 nm observed during respiration is commonly assigned to the fully reduced species of hemes a or a₃ (R). However, by analyzing the absorbance of isolated enzyme and mitochondria in the Soret (420-450 nm), alpha (560-630 nm) and red (630-700 nm) spectral regions, we demonstrate that the Peroxy (P) and Ferryl (F) intermediates of the binuclear center are observed during respiration, while the R form is only detectable under nearly anoxic conditions in which electrons also accumulate in the higher extinction coefficient low spin a heme. This implies that a large fraction of COX (>50%) is active, in contrast with assumptions that assign spectral changes only to R and/or reduced heme a. The concentration dependence of the COX chromophores and reduced c-type cytochromes on the transmembrane potential (ΔΨ_m) was determined in isolated mitochondria during substrate or apyrase titration to hydrolyze ATP. The cytochrome c-type redox levels indicated that soluble cytochrome c is out of equilibrium with respect to both Complex III and COX. Thermodynamic analyses confirmed that reactions involving the chromophores we assign as the P and F species of COX are ΔΨ_m-dependent, out of equilibrium, and therefore much slower than the ΔΨ_m-insensitive oxidation of the R intermediate, which is undetectable due to rapid oxygen binding.¹

1. Introduction

The respiratory chain of mammalian mitochondria consists of three multi-subunit protein complexes that generate an electrochemical proton gradient across the inner membrane by coupling electron transfer between redox cofactors (and finally to oxygen) to net positive charge translocation, either by direct proton pumping or by the net movement of positive charges from the matrix to the intermembrane space. This gradient, mostly comprised of an electrical potential (ΔΨ_m), is used to synthesize ATP by Complex V. Many of the electron transferring groups in the three energy transducing complexes exhibit absorbance changes in the visible range depending on their redox state: flavin mononucleotide (FMN) in NADH dehydrogenase (Complex I), the b_L, b_H and c₁ hemes in ubiquinol:cytochrome c oxidoreductase (Complex III), and the a and a₃ hemes in cytochrome c oxidase (COX), in addition to cytochrome c, which serves as a diffusible electron carrier between the two latter complexes. These redox active mitochondrial chromophores have been extensively studied since the pioneering work reported by Keilin [1] and Chance [2, 3]. However, most studies (with few exceptions [4, 5]) have relied on a very restrictive and potentially misleading [6] dual wavelength analysis in which peak wavelengths for each chromophore are used relative to absorbance minima or allegedly fixed isosbestic points [7, 8], either in isolated mitochondria [9] or in the intact heart system (both perfused [10, 11] and in vivo [12]).

We have recently reported the advantages of using linear regression multicomponent analysis that includes all spectral data in an accessible wavelength range to accurately determine changes in each mitochondrial chromophore, even in the presence of overlapping absorption bands. These changes were monitored during metabolic transitions using integrating sphere spectroscopy in the case of isolated mitochondria [13–15], or transmural absorbance spectroscopy in the case of the perfused heart [6, 16, 17]. In attempting to translate this approach to the intact heart, we have considered the optical properties of myoglobin, which is present in high concentrations in the heart and is sensitive to cytosolic oxygen concentration [18]. The inclusion of myoglobin to either difference [6, 16, 17] or absolute spectra [19] is necessary to separate its absorbance contribution from mitochondrial chromophores in the perfused heart.

The optically observable Complex I and III chromophores, together with soluble cytochrome c, transition between only two states (oxidized and reduced) which allows their redox state to be assessed easily by visible spectroscopy. In contrast, the COX catalytic cycle includes several optically active intermediates in which the binuclear a₃-Cu_B center (where oxygen reduction takes place) is not simply fully oxidized (O) or reduced (R) [20, 21]. The catalytic cycle of COX has been elucidated mostly from experiments that start with the fully reduced (including the binuclear center in the R state) enzyme that is then allowed to combine with oxygen to determine the rate constants and properties of the subsequently formed remaining catalytic intermediates during a single turnover. Only in a few cases, such as the studies performed by Wikström [22, 23], have mitochondria been used to study some of the COX intermediate species, although in the absence of forward electron transfer to oxygen. An unexpected result we have obtained from spectral fittings of metabolic transitions in isolated mitochondria and intact hearts is the required inclusion of an optical component with a peak at 580 nm, which is strongly sensitive to changes in ΔΨ_m [13, 15], as well as a 607 nm peak that responded linearly with respiratory rate [13, 15]. Although we proposed that these components correspond to the so called Ferryl (F) and Peroxy (P) intermediates in the catalytic cycle of COX [20, 21], these species have only been reported under non-physiological and strongly oxidizing conditions, such as titrating isolated COX with H₂O₂ at alkaline pH [24, 25], or by inducing reverse electron transfer in mitochondria with ATP and ferricyanide to remove electrons from cytochrome c and the a-type hemes of COX [22, 23]. To our knowledge, the detection of the F and P species has never been reported in intact mitochondria under conditions of steady state respiration.

In this study, we confirmed that the 580 nm and 607 nm absorbance peaks observed during respiration correspond to the F and P intermediates of COX, respectively, by expanding our optical detection analysis to include the Soret and red bands in isolated mitochondria. This approach also permitted the determination of a negligible contribution of the R species of the binuclear center and/or the reduced a heme to the normoxic mitochondrial spectrum and generated estimates of the steady state concentration of other COX intermediates. Thermodynamic analyses of the resulting data were applied to estimate which transitions between spectroscopically detectable intermediates of the COX reaction cycle are out of equilibrium by inducing changes in ΔΨ_m. This provided further confirmation that the absorbance changes observed in COX-related chromophores during mitochondrial respiration are caused mostly by variations in P and F, with negligible contribution from the R state of the binuclear center or the reduced a heme.

2. Materials and methods

2.1. Heart excision

All animal protocols were approved by the National Heart, Lung, and Blood Institute Animal Care and Use Committee and performed in accordance with the guidelines described in the Animal Care and Welfare Act (7 USC 2142 § 13). New Zealand male white rabbits were preanesthetized via an intramuscular injection of 1.5 mL ketamine/acepromazine mixture (10:1). Approximately 15 minutes later, 3% isoflurane was administered via inhalation for complete anesthetic effect. A line was placed in the marginal ear vein for administration of subsequent drugs, after confirming proper depth of anesthesia by toe pinching. 1,500 U of Heparin were administered and after 3 minutes of circulation and proper depth of anesthesia confirmed again, the animal was euthanized with 6 mEq KCl. The chest was then opened and the hearts were rapidly excised and further processed.

2.2. Mitochondrial isolation

The mitochondrial preparation procedure previously reported for dog and pig heart [26] was modified as follows: A total of four rabbit hearts were excised for each mitochondrial isolation procedure. Each heart was perfused in situ with ice-cold Krebs-Hanseleit (KH) buffer (137mM NaCl, 5.4mM KCl, 10mM HEPES, 1.8mM CaCl₂, 10mM glucose, 0.5mM MgCl₂, 0.5mM Na₂HPO₄ and 1mM lactate, adjusted to pH 7.4 with NaOH) to prevent warm ischemia, remove blood, and supply the heart with Ca²⁺. The left ventricles and septa were dissected of all fat and connective tissue on ice, weighed and minced with scissors in the appropriate volume of ice-cold buffer A (0.28 M sucrose, 10 mM HEPES, 1 mM EDTA, 1 mM EGTA and 5 mM K₂HPO₄ adjusted to pH 7.1 with KOH) to obtain a 20% w/v suspension in 50 ml conical tubes. The minced fragments from two hearts were pooled and diluted to yield a ~10% w/v suspension before disrupting for 20 s in a 100 ml beaker on ice at 40% power using a tissue homogenizer (Virtis, Gardiner, NY). Each suspension was treated with 2.5 mg/5 g tissue wet weight of trypsin and incubated at 4°C for 15 mins while mixing. This was followed by the addition of 13 mg/5 g tissue wet weight of trypsin inhibitor and 100 mg/10 ml of bovine serum albumin. The trypsin-treated homogenate was then centrifuged at 600g at 4°C for 10 min. The supernatant was collected and the pellet was resuspended in buffer A. The suspension was further homogenized with 40 ml Teflon homogenizers; two passes of a 1 mm (loose) clearance and five passes of a 0.2 mm (tight) clearance (Thomas Scientific, Swedesboro, NJ). The heart homogenate was centrifuged again at 600g at 4°C for 10 min. The supernatant that contained the mitochondria was pooled with the previous supernatant. The pelleted homogenate was then resuspended in 40 ml of cold Buffer A and centrifuged at 600g at 4°C for 10 min. This procedure was repeated at least twice. The mitochondria in the pooled supernatants were pelleted at 8,000g at 4°C for 10 min and the resulting supernatant was discarded. The pelleted mitochondria were resuspended in buffer A followed by another 8,000g centrifugation. Pelleted mitochondria were washed and resuspended once with buffer B (125 mM KCl, 15 mM NaCl, 20 mM Hepes, 1 mM EGTA, 1 mM EDTA, 5 mM MgCl₂, 5mM potassium phosphate pH 7.1), centrifuged at 8,000g and resuspended in 1 ml of buffer B. Mitochondria were further purified by centrifugation in a Percoll gradient, as was described before for skeletal muscle mitochondria [27] with modifications. Briefly, 1 to 2 ml of mitochondrial suspension was placed on top of 8 ml of a 30% v/v Percoll solution in buffer B and centrifuged at 68,000g at 4°C for 40 min. After carefully removing the top and middle layers containing broken mitochondria and cell debris, the lower brown layer containing heavier mitochondria was collected and transferred to four 2 ml centrifuge tubes, which were filled with buffer B and gently inverted to mix. After centrifuging at 13,000g at 4°C for 10 min and discarding the Percoll containing supernatant, the mitochondrial pellets were resuspended in 4 ml of buffer B and centrifuged at 10,000g at 4°C for 10 min. The final mitochondrial pellets were resuspended by adding 1 ml of buffer B to each. Cytochrome a content was quantified spectroscopically using an extinction coefficient of 12 mM⁻¹ cm⁻¹ for the reduced minus oxidized cytochrome a as previously reported [28], which was validated by quantifying the copper content by atomic absorption spectroscopy in a sample of known absorbance of COX isolated as described below. Mitochondrial preparations typically resulting in concentrations of 50–60 nmoles cytochrome a/ml corresponding to a total yield of 140–180 nmoles of cytochrome a. Using this modified mitochondrial isolation protocol, no traces of contaminating hemoglobin or myoglobin were detected spectroscopically. The average respiratory control ratio obtained for 11 mitochondrial preparations was 21.4 ± 1.7 at 30°C, indicating a high degree of coupling.

2.3. COX isolation, quantitation and spectroscopy

COX was isolated by anionic exchange chromatography using a procedure originally reported for Complex III purification [29] with modifications. Frozen isolated rabbit heart mitochondria (corresponding to 200 nmoles of cytochrome a) were thawed and diluted to 10 nmol cytochrome a/ml in ice cold buffer C (50 mM Tris, 1 mM MgSO4, adjusted to pH 8.45 at 4°C) before dropwise addition of 10% w/v n-β-D-dodecyl maltoside (DM) to a final detergent concentration of 1% on ice with gentle mixing. This extract was centrifuged at 40,000g for 40 min at 4°C and the supernatant was applied to a 2.5x20 cm glass column (Bio Rad) previously packed with DEAE Sepharose Fast Flow anion exchange resin (GE Healthcare) equilibrated with 5 column volumes buffer C containing 0.02% w/v DM. After washing with 2 column volumes of buffer C with 0.02% DM, elution of mitochondrial complexes was achieved by applying a linear 0–400 mM NaCl gradient in the same buffer, and collecting the eluate in 4 ml fractions. Green fractions containing COX eluted close to the middle of the gradient (~180 mM NaCl) before the reddish fractions containing Complex III. The pooled COX containing fractions were concentrated to 20-30 nmoles cytochrome a/ml in an Amicon Ultra-15 centrifugal filter (Millipore) with a 10 kDa membrane cutoff. Sodium ascorbate- or hydrosulfite-reduced minus air- or ferricyanide-oxidized spectra showed no contamination by b or c-type cytochromes. This rapid one-step chromatographic purification procedure yielded a homogeneous ‘fast’ COX sample in which electron transfer between heme a and a₃ is unimpeded, based on an oxidized Soret maximum peak at 424 nm and complete reduction of both hemes by hydrosulfite within 1 min [20, 30].

Validation of the extinction coefficient for the fully reduced cytochrome a was done by quantifying copper in this spectroscopically pure COX preparation by atomic absorption spectroscopy (AAS) using a PinAAcle 900Z instrument (Perkin Elmer) equipped with a graphite furnace and a hollow cathode lamp for copper and calibrated with a 10-320 μg/l copper standard solution. A COX sample of known absorbance was digested in 6% HNO₃ for 20 min in boiling water and centrifuged at 16,000g for 10 min; the supernatant was then placed in the AAS instrument for copper quantitation. After subtracting for the background AAS absorbance using an equivalent volume of buffer C + 6% HNO₃, the copper amount was divided by the optical absorbance considering 3 Cu/cytochrome a yielding an extinction coefficient of 12.6 ± 0.12 mM⁻¹ cm⁻¹ (n=3), almost identical to the value of 12 mM⁻¹ cm⁻¹ reported previously [28, 31].

Generation of the F and P species [20, 21] using this spectroscopically pure COX preparation was achieved by the addition of H₂O₂ as described before [13, 24] except that buffer B with 0.02% DM (pH 7.1) was used to dilute the isolated COX to 1.5 μM before addition of either 5 (7.5 μM) or several 600 (0.9 mM) equivalents of H₂O₂. Spectra were collected every 0.5 s in an integrating sphere using the settings described below for mitochondria. Alternatively, buffer B (pH 7.1) supplemented with 0.02% DM was used to dilute the isolated COX to 2 μM and mix it with 5 or 10 mM sodium ascorbate in the presence or absence of 3 μM oxidized horse heart cytochrome c (from Sigma Chemical). In some experiments, catalase (1100 U/ml) was added to quench any H₂O₂ that could be formed during ascorbate oxidation. Ascorbate was omitted from some assays in which 2 μM COX was mixed with 8 μM reduced cytochrome c that was prepared by reaction with hydrosulfite and separation through a Sephadex G-25 column equilibrated with buffer B.

2.4. Mitochondrial absorption spectroscopy

All mitochondrial optical measurements were made in an integrating sphere as described earlier [13] with some modifications. A custom made cylindrical (15 mm in diameter x 50 mm of height) glass cuvette (Chemglass) was center mounted inside an integrating sphere (Labsphere) on a magnetic stirring plate. A nozzle was inserted through the cap at the top of the integrating sphere to allow blowing of a stream of pure oxygen, which was heated through a NeoPod T humidification system (Westmed) set to 38°C. This oxygen flow allowed the temperature of the liquid inside the integrating sphere cuvette constant at ~30°C, as determined using a temperature probe. This warm oxygen flow obviated the need for a glass water jacket system used in a prior study [13], which diverted incident light from the sample decreasing sensitivity. In most experiments, the appropriate volume of buffer B (supplemented with 0.6 mM CaCl₂ to ensure optimal mitochondrial activity) was added to the center mounted cuvette so that the addition of mitochondria to a final concentration of 2 nmol cytochrome a/ml resulted in a total volume of 5 ml. Another port in the integrating sphere cap allowed addition of reagents (Glu + malate, ATP, apyrase, inhibitors) to the mitochondrial suspension using a Hamilton syringe. The LLC-10 Fiber Optic Tungsten-Bromine Light Source (Lambda Scientific Systems) was chosen as the light source because it provided adequate light intensity in most of the Soret spectral region (down to ç410 nm). A camera (Potensic^® Digital Endoscope-5M), was attached to one of the ports of the integrating sphere to enable visualization and recording of reagent additions to the interior of the cuvette and adjusting of the mixing speed and oxygen flow. Spectra were collected every 0.5 s using a custom LabVIEW-based program for spectral acquisition and analysis installed on a PC using LabVIEW drivers provided by Ocean Optics and analyzed as previously described [6]. Steady state regions were averaged in the time domain with variable durations to obtain a characteristic spectrum and to maximize the signal to noise ratio.

2.5. Determination of ΔΨ_m in isolated mitochondria

In order to measure ΔΨ_m simultaneously with the absorption spectra of the mitochondrial cytochromes, 2 μM tetraphenylphosphonium (TPP⁺) was added to buffer B before addition of mitochondria in the glass cuvette of the integrating sphere. The corresponding relative voltage was recorded using a TPP⁺-selective (World Precision Instruments) and a reference (Microelectrodes) electrode inserted through holes on the sphere’s lid. Upon addition of mitochondria to the TPP⁺-containing buffer B, the voltage decreased and then slowly recovered during a starvation period of 15–20 min until stabilizing at a value slightly lower than that recorded before adding mitochondria. The difference in this voltage relative to its initial value without mitochondria was used to correct for TPP⁺ binding, and was found to be smaller and more reproducible than the usual correction performed by adding mitochondrial uncouplers at the end of the experiment. External [TPP⁺] was quantified during subsequent addition of reagents to mitochondria by interpolating the electrode’s measured voltage to a standard [TPP⁺] curve collected separately in buffer B in the absence of mitochondria. Intramitochondrial [TPP⁺] was calculated from the missing TPP⁺ (corrected for binding) assuming 2 μL/nmol cytochrome a for mitochondrial volume, and the corresponding ΔΨ_m was derived using the Nernst equation[32]. A new TPP⁺-selective electrode was used every week due to decay in the quality of the signal after more than a few days of experiments.

2.6. COX thermodynamic modeling

The catalytic cycle of oxygen reduction was analyzed with a simplified model of COX including four states: P, F, OE and R. Within the OE state we grouped together the optically invisible O_H. and E_H states, and the A intermediate between the R and P species was omitted due to its very low occupancy at room temperature [20, 21]. The concentration of P and F were determined from their respective fitted absorbance values at 607 and 580 nm. The P species was reported to have a >2-fold higher extinction coefficient than the F intermediate in isolated liver mitochondria under conditions of reverse electron transfer [23]. However, a very similar maximal absorbance between the two species in isolated COX when converted sequentially to the F and P forms with different H₂O₂ concentrations was found for the bacterial enzyme [24]. We determined a maximal absorbance of 16% for F and 20% for P relative to the fully reduced COX (see Results). Since reduced heme a₃ in the binuclear center has been reported to have an extinction coefficient of 20% relative to the fully reduced cytochrome a hemes [33, 34], the absorbance measured under anoxic conditions was multiplied by 0.2 to calculate the maximal absorbance attributed to the P + F + OE + R forms, all of which correspond to heme a₃-Cu_B intermediates. Because we could not detect any appreciable absorbance corresponding to the R form (see Results section below), its reaction with oxygen was assumed to be rapid compared to the other reactions and its concentration was fixed to an arbitrarily low value of 0.1% of the total COX concentration. The concentration of the OE form was inferred by subtraction of the other known forms such that OE = 1 − P − F − R. The redox reactions and net movement of charges across the membrane between these four states are given by:

$$c_{red}+P+H^{+}\leftrightarrow c_{ox}+F,2H_N^{+}\to 2H_P^{+}$$ Eq. 1a

$$c_{red}+F+H^{+}\leftrightarrow c_{ox}+OE,2H_N^{+}\to 2H_P^{+}$$ Eq. 1b

$$2c_{red}+OE+2H^{+}\leftrightarrow 2c_{ox}+R+2H_{2}O,4H_N^{+}\to 4H_P^{+}$$ Eq. 1c

$$R+O_2\leftrightarrow P$$ Eq. 1d

where c_red and c_ox refer to the reduced and oxidized forms of cytochrome c, cyt c²⁺ and cyt c³⁺, respectively. The redox reactions 1a–1d involve the exchange of n electrons, with n equal to 1, 1, 2 and 4 respectively, which are coupled to the translocation of p positive charges (expressed as protons for simplicity) from the matrix (N-phase) to the inter membrane space (IMS) (P-phase), with p equal to 2, 2, 4 and 0, respectively [21]. The conversions from the O_H to the E_H intermediate and from E_H to the R form, each of which involves the transfer of one electron and the translocation of two positive charges, were grouped as a single reaction (1c) because of the lack of optical information to distinguish between the O_H and E_H forms [21]. The Nernst equations for the four reactions are:

$$E_1=E_{F/P}^0-E_{c_{red}/c_{ox}}^0-\frac{ℛT}{F_a}\mathrm{Ln}\left(\frac{F{c}_{ox}}{P{c}_{red}}\right)$$ Eq. 2a

$$E_2=E_{OE/F}^0-E_{c_{red}/c_{ox}}^0-\frac{ℛT}{F_a}\mathrm{Ln}\left(\frac{OE{c}_{ox}}{F{c}_{red}}\right)$$ Eq. 2b

$$E_3=E_{R/OE}^0-E_{c_{red}/c_{ox}}^0-\frac{ℛT}{2F_a}\mathrm{Ln}\left(\frac{R{c_{ox}}^2}{OE{c_{red}}^2}\right)$$ Eq. 2c

$$E_4=-E_{R/P}^0-\frac{ℛT}{4F_a}\mathrm{Ln}\left(\frac{P}{R}\right)$$ Eq. 2d

where F, P, OE, R, c and c_ox refer to mole fractions, so that P + F + OE + R = 1 and c + c_ox = 1, ℛ is the gas constant, T the temperature in °K, F_a is the Faraday constant and the activity of water and oxygen was assumed equal to unity. Neglecting small differences in pH between the matrix and the IMS [9], protonmotive force was assumed to be equal to ΔΨ_m. To calculate the mid-point potential (E⁰) for each of the first three reactions above encompassing the conversion from P to R, it is necessary to account for the effect of ΔΨ_m because of the coupling of the electron transfer to the movement of positive charges. The R to P conversion does not involve the net gain or loss of electrons, but instead an internal transfer of four electrons inside the binuclear center upon oxygen binding. Assuming the reactions are at thermodynamic equilibrium, i.e. the free energy of reaction i ΔG_i satisfies ΔG_i − p_iF_aΔΨ_m = 0, the mid-point potentials are calculated as follows:

$$E_{F/P}^0=E_{c_{red}/c_{ox}}^0+\frac{ℛT}{F_a}\mathrm{Ln}\left(\frac{F{c}_{ox}}{P{c}_{red}}\right)-2{\text{ΔΨ}}_{\text{m}}$$ Eq. 3a

$$E_{OE/F}^0=E_{c_{red}/c_{ox}}^0+\frac{ℛT}{F_a}\mathrm{Ln}\left(\frac{OE{c}_{ox}}{F{c}_{red}}\right)-2{\text{ΔΨ}}_{\text{m}}$$ Eq. 3b

$$E_{R/OE}^0=E_{c_{red}/c_{ox}}^0+\frac{ℛT}{2F_a}\mathrm{Ln}\left(\frac{R}{OE}{\left(\frac{c_{ox}}{c_{red}}\right)}^2\right)-2{\text{ΔΨ}}_{\text{m}}$$ Eq. 3c

$$E_{R/P}^0=-\frac{ℛT}{4F_a}\mathrm{Ln}\left(\frac{P}{R}\right)$$ Eq. 3d

Under the equilibrium assumption for the reactions above, free energies for the complete cycle add up to 8F_aΔΨ_m. This energy can be compared to the overall free energy ΔG_c/O2 = −4F_aE_c/O2 from the reduction of oxygen by cyt c, where ${\text{E}}_{c/O_2}=E_{H_{2}O/O_2}^0-E_{c_{red}/c_{ox}}^0-RT/F_a\mathrm{Ln}\left(c_{ox}/c_{red}\right)$, $E_{H_{2}O/O_2}^0=0.802\text{V}$ at pH = 7.1, T = 303.15 °K, and $E_{c_{red}/c_{ox}}^0=0.25\text{V}$ [21].

3. Results

3.1. Identification and assignment of COX chromophores in the isolated enzyme

As we have reported before [6, 13], multi-wavelength analysis of isolated and in situ heart mitochondrial spectra allows a reliable quantitation of all the individual chromophores that contribute to optical changes upon metabolic transitions. Using integrating sphere spectroscopy, we previously identified in isolated mitochondria two species which we named a580 and a607, based on their absorbance maxima, which could be well fitted using spectral references for the P and F catalytic intermediates of COX and collected in the isolated enzyme [13, 15]. Therefore, we sought to further validate the assignment of the a580 and a607 spectral components to the P and F catalytic intermediates of COX by performing the experiments shown in Figure 1 with spectroscopically pure COX in solution. The commonly used procedure to obtain the P and F species using H₂O₂ is reported to generate a stable P form at low H₂O₂ concentrations at alkaline pH, converting to the F intermediate upon addition of higher H₂O₂ or by acidification to pH 6.0-6.5 [24, 25]. As shown in Fig. 1A and B, the P species generated by adding a few equivalents of H₂O₂ relative to COX was stable for at least 5 min at the physiological, nearly neutral pH of 7.1. Furthermore, both this and the F species (generated after adding a higher H₂O₂ concentration) exhibited the reported bleaching of the 655-660 nm band [35, 36] (appearing as a trough in the difference spectra), which is sharper than the broader decrease in absorbance observed in the fully reduced form of COX (Fig. 1B). The absorbance maximum of the P (607 nm) and F (580 nm) species was 21% and 16%, respectively, relative to that the fully reduced COX. The Soret spectral region showed the expected ~435 nm maximum for the P and F species of COX [24, 25] (Fig. 1C), which is very different from the ~444 nm peak characteristic of the reduced forms of both hemes a and a₃ [30, 33, 34] that was generated by reduction with hydrosulfite (Fig. 1D). Complete reduction of both hemes was attained in less than 1 min, confirming that our isolated COX preparation was able to transfer electrons very rapidly between hemes a and a₃, in contrast with the ‘slow’ (likely non-physiological [37]) form of isolated COX that exhibits a very slow reduction of the a₃ heme by hydrosulfite due to an uncharacterized impediment of electron transfer into the a₃-Cu_B binuclear center [20, 30].

Figure 1:

Generation of the P and F species in isolated COX. A: Time course of absorbance at 607 nm upon addition of H₂O₂ to isolated COX (1.5 μM) at pH 7.1. H₂O₂ additions of 7.5 μM (a), 0.9 mM (b, c) and 1.8 mM (d, e) are indicated by vertical arrows. Horizontal bars show the time regions averaged to generate the difference spectra shown in B and C using the oxidized COX (O) as a reference. B: Difference spectra in the alpha spectral region of the P (1-O, black), and increasingly pure F species (2-O, blue; 3-O, green; 4-O, orange; 5-O, red) of COX (left y-axis) obtained by adding the H₂O₂ concentrations shown in A. For comparison the fully reduced COX difference spectrum is shown in cyan (right y-axis) obtained after hydrosulfite addition. C: Same as in B, but in the Soret spectral region. D: Difference spectra of isolated COX (1.5 μM) relative to the oxidized state collected every 2.5 s after addition of a single dose of hydrosulfite (7.2 mM) to reduce both the a and a₃ hemes. Notice that the absorbance maximum for the P and F species (435 nm, indicated by the thick line) is very different from the maximum at 444 nm (indicated by the thin line) observed as the a hemes are reduced.

Interestingly, we were able to obtain a mixture of the 607 and 580 nm species at neutral pH when reducing COX with ascorbate alone (Fig. 2A) or in combination with cytochrome c in the presence of oxygen to allow enzymatic turnover (Fig. 2B). The Soret absorbance peak position (at 436 nm) at pH 7.1 was identical to that obtained with H₂O₂, and clearly different from the 444 nm peak obtained when oxygen was depleted (Fig. 2B), and which results from full reduction of both the a and a₃ hemes [34]. These results indicate that it is possible to obtain and spectroscopically measure the P and F species of COX at physiological pH values during steady state electron transfer from the a hemes to oxygen without the use of peroxide. The possibility that ascorbate was generating peroxide and thus generating the P and F species was discarded by performing the experiments shown in Fig. 2C and 2D, in which a high concentration of catalase was used to eliminate any H₂O₂ derived from ascorbate oxidation (Fig. 2C) or, alternatively, by entirely omitting ascorbate from the assay using purified reduced cytochrome c (Fig. 2D). In both cases, the COX spectral features in the presence of oxygen (Soret absorption peak ~435 nm and alpha peak and shoulder at 607 and 580 nm, respectively) corresponded to the P and F species. In contrast, the R form of the binuclear center, as well as the reduced a heme, with their distinctive Soret peak at ~44 nm were only detected at anoxia when no electron transfer to oxygen was possible and both heme a and a₃ became reduced.

Figure 2:

Generation of the P and F species in respiring isolated COX in the absence of H₂O₂ at pH 7.1. A: Difference spectrum obtained by adding ascorbate (10 mM) to isolated COX (2 μM) in the presence of oxygen at neutral pH (7.1). B: Difference spectrum of COX obtained by adding ascorbate (5 mM) to isolated COX (2 μM) and cytochrome c (3 μM) in the presence of oxygen at pH 7.1 (black trace, y-axis on the left). The red spectrum shows the anoxic spectrum of isolated COX and cytochrome c (y-axis on the right). The fully reduced state (R plus reduced a heme) of COX has a Soret peak at 444 nm whereas the P and F species have a peak at 436 nm (identical to that observed in the presence of H₂O₂). The reference spectrum in A corresponded to oxidized COX, whereas in B it was collected with the mixture of oxidized COX and cytochrome c before addition of ascorbate. C: Difference spectrum of a mixture of isolated COX (2 μM) and cytochrome c (8 μM) after adding ascorbate (10 mM) in the presence of catalase (1100 U/ml) during steady state oxygen consumption (black trace). The reference spectrum was collected with the mixture of oxidized COX and cytochrome c before addition of ascorbate or catalase. The anoxic difference spectrum (red trace) was obtained after adding a few grains of hydrosulfite. D: Difference spectrum obtained by adding isolated COX (2 μM) to reduced cytochrome c (8 μM) in the presence of oxygen. The reference spectrum was collected after COX had completely transferred all electrons from cytochrome c to oxygen. The anoxic spectrum (red trace) was obtained after adding a few grains of hydrosulfite. The Soret and alpha peaks of COX during oxygen consumption correspond to a mixture of P and F species. The Soret peak at 444 nm corresponding to the fully reduced state of both a and a₃ hemes was only observed at anoxia.

3.2. Identification and assignment of COX chromophores in isolated heart mitochondria

Further evidence indicating that the ~607 nm species observed in respiring COX corresponds to the P catalytic intermediate, and not to the R form of the binuclear center (or the reduced a heme), was obtained by comparing the mitochondrial difference spectra shown in Figs. 3 and 4. When respiring heart mitochondria were allowed to reach anoxia (Fig. 3), the most prominent absorbance peak in the Soret region was located at 445 nm (Fig. 3A), corresponding to the R form of the binuclear center together with the reduced a heme, which also absorb at 605 nm (Fig. 3B), but with a ~3-fold lower extinction coefficient compared to the Soret peak. However, as shown in Figure 4, under conditions in which ΔΨ_m was decreased while compensating the electron input from the substrates in order to cancel the contribution of the b_L and b_H hemes from Complex III in the Soret region, no significant peak was observed at 445 nm (Fig. 4A) even though a strong absorbance was clearly detected at 607 nm, indicative of the P species (Fig. 4B). The 436 nm peak observed in the Soret in isolated COX (see Figs. 1 and 2) was absent in this experiment due to the negative change of the b_L heme absorbance, which cancelled both the b_H and the P absorbance peaks in this spectral region, as indicated by the combinations of spectral components used by the fit (Fig. 4C). More importantly, the fit used very little of the a605 reference (coming from the binuclear center R form together with reduced a heme) in the Soret region to account for the weak and broad absorbance around 445 nm using instead the a607 reference that corresponds to the P intermediate at a level several-fold higher than the absorbance of the same species in the alpha region (compare Figs. 4C and D), as expected from the higher extinction coefficient in the Soret. These results strengthened the conclusion that the reduced state of the a hemes is virtually undetectable in isolated mitochondria in the presence of oxygen.

Figure 3:

Spectral characteristics of the fully reduced (R in addition to the reduced a heme) form of COX in isolated mitochondria at anoxia. A: Soret region raw, fit, and residual spectra of the anoxia (after addition of hydrosulfite) minus oxidized difference absorbance. The mitochondrial concentration used was equivalent to 5 nmol COX/ml. B: Same as A but in the alpha region. C and D: Contribution of each spectral reference used to obtain the fitted data in A and B. Notice that the absorbance of the a605 (R form of the binuclear center together with reduced a heme) reference is ~3-fold higher in the Soret region (peak at 445 nm) compared to the alpha region.

Figure 4:

Absence of the reduced form of COX in isolated respiring mitochondria. A: The reference averaged spectrum of mitochondria respiring with a small concentration (0.1 mM) of Glu + malate was subtracted from the averaged spectrum after increasing the concentration of Glu + malate to 0.5 mM and adding 2 mM ADP in the Soret spectral region. The raw, fit, and residual spectra are plotted. B: Same as A but in the alpha region. C and D: Contribution of each cytochrome reference used to obtain the fitted data in A and B. Notice the absence of a strong 445 nm absorbance in the Soret region (see A) despite the absorbance in the 607 nm region (see B), which could be well fitted without a significant contribution of the a605 species corresponding to the R form of the COX binuclear center in addition to the reduced a heme.

As shown in Figure 5, fitting of the difference average spectrum resulting from adding saturating substrate (Glu + malate) to oxidized isolated mitochondria (after substrate depletion) required the inclusion of the a580 spectral reference, which was obtained using isolated COX under conditions reported to generate the F intermediate of the a₃-Cu_B binuclear center [24]. Even though the a580 component is relatively broader than other cytochrome references, it did not correspond to a simple baseline distortion given that integrating sphere spectroscopy greatly minimizes light scattering, which is the main factor affecting the spectral baseline shape. Also, it would be highly unlikely that any baseline distortion induced by increased light scattering would have a maximum at 580 nm.

Figure 5:

Requirement of a580 for spectral fitting of the effect of substrate addition in isolated mitochondria. A: Raw, fitted, and residual spectra of the absorbance difference between the steady states after and before Glu + malate addition. The mitochondrial concentration used was equivalent to 2 nmol COX/ml. B: Same as A but without using a580 as a reference in the fit. C and D: Contribution of each mitochondrial chromophore reference used to obtain the fitted data in A and B. In A, the residuals indicated a good fit of the raw spectrum. However, in B where the reference spectrum of a580 was removed, the fit was poor.

Inspection of our difference spectral references revealed that in the red spectral band (620-700 nm) the b and c-type hemes are essentially flat without any peaks or troughs, whereas the fully reduced form of hemes a + a₃ is also flat between 630 and 660 nm (see Fig. 1B). In contrast, the a580 and a607 references possess a relatively broad trough centered at ~660 and ~655 nm, respectively. In addition, the a580 species presents a small positive shoulder between 630 and 635 nm that is absent in the a607 reference. All these features have been previously reported for the P and F intermediates of the isolated COX [35, 36, 38]. As shown in Figure 6, the difference spectrum of isolated mitochondria respiring with Glu + malate showed spectral features in the red band that could be well fitted with a combination of a580 and a smaller amount of a607, consistent with the relative contribution of the two species in the alpha band (see Fig. 5). Using the a605 (fully reduced form of COX) did not improve the quality of the fit (not shown).

Figure 6:

Fitting of the spectral response in the red band upon addition of substrate to isolated mitochondria. A: Raw, fitted, and residual spectra of the absorbance difference between the steady states after and before 0.5 mM Glu + malate addition in the red region. The mitochondrial concentration used was equivalent to 2 nmol COX/ml. B: Contribution of each mitochondrial chromophore reference used to obtain the fitted data in A. The a580 reference was the largest contributor to the fit in this spectral region.

3.3. Changes in the absorbance of COX intermediates and c-type cytochromes as a function of ΔΨ_m

Since the interconversion of the COX catalytic intermediates involves proton pumping steps across the membrane and movements of charges within it, we sought to characterize the dependence of the changes in the P and F species on ΔΨ_m in isolated mitochondria, as shown in the representative experiment of Fig. 7. The depolarization induced by ATP hydrolysis by apyrase caused a nearly linear increase in a607 absorbance (corresponding to the P form) accompanied by a nonlinear decrease in a580 (F species) that was steeper at high ΔΨ_m with less variation as depolarization continued (Fig. 7A). This was similar to our previously reported observations in pig heart mitochondria [13]. In contrast, titration of Glu + malate after the addition of 5 mM ATP and a low concentration of apyrase to sustain a slow ATP turnover rate, the increase in ΔΨ_m generated upon higher substrate concentrations induced little change in the amount of the P form, whereas the F species increased almost linearly as a function of ΔΨ_m. We also simultaneously determined the electron availability for the COX reactions by deconvoluting the absorbance contribution from cytochromes c₁ (a membrane-bound subunit of Complex III) and that of cytochrome c, the direct soluble electron donor for COX. As shown in Figure 7B, reduction of cytochrome c and c₁ changed in the same direction as the P species in the case of the apyrase titration, but followed the change in the F species when substrate was gradually increased. In both cases, the reduction level of cytochrome c was significantly lower than that of cytochrome c₁. Given that the total absorbance of cytochrome c in the fully reduced mitochondria was ~1.7 higher than that of cytochrome c₁, the absorbance data shown in Fig. 7B implies that the reduction level of cytochrome c was of only 10-20%, less than half of cytochrome c₁ (25-50%). Since the E^o of these two cytochromes is very similar [39], this surprising result suggests a faster electron transfer rate from cytochrome c to COX than from cytochrome c₁ in Complex III to cytochrome c.

Figure 7:

Dependence of the absorbance of the P and F species of COX (A) and of cytochromes c and c₁ (B) on ΔΨ_m in isolated mitochondria. Two protocols were followed to obtain redox changes as a function of ΔΨ_m. For the data points labeled as Apyrase, a high ΔΨ_m was attained initially by adding 10 mM Glu + malate followed by addition of 5 mM ATP, after which ΔΨ_m was progressively decreased by titration with apyrase. For the data points labeled as GM, a subsaturating concentration of Glu + malate (0.5 mM) was added before the addition of 5 mM ATP and a low concentration of apyrase (0.1 U) to maintain a slow hydrolysis rate back to ADP. Then, higher concentrations of Glu + malate were added progressively to increase ΔΨ_m until substrate saturation was reached. The concentration of mitochondria was equivalent to 2 nmol COX /ml. The absorbance value of the fully reduced cytochrome c and cytochrome c₁ in these experiments was 0.017 and 0.0105, respectively, as determined by adding hydrosulfite at the end of the titrations.

Assuming that the extinction coefficients of the P and F species in the alpha spectral region are 16-20 % relative to that of the fully reduced (a + a₃ hemes) COX (based on the maximal absorbances shown in Fig. 1B), the minimum fraction of COX that participates in electron transfer at steady state can be calculated from experiments similar to those shown in Fig. 7 by converting absorbance to concentration and normalizing to the total COX in the mitochondrial preparation. As shown in Fig. 8, the sum of COX in the P and F states was between 40 and 60% at all ΔΨ_m values with saturating substrate in the case of the titration with apyrase, dropping slightly below 40% only at the lower Glu + malate titration points when substrate was insufficient to sustain a highly polarized ΔΨ_m. This implies that at least ~50% of the COX present in the heart mitochondria was in the reaction cycle, with the remainder either in the O_H, E_H, or fully oxidized (O) state, which are all indistinguishable from each other by visible spectroscopy. However, since the O state is only expected to exist in the absence of electron flow [20, 21], the fraction of COX involved in the oxygen reduction cycle steps that conserve energy (P → F → O_H → E_H → R) was probably close to 100% throughout the apyrase titration and at the higher concentrations of the Glu + malate titration that resulted in a larger ΔΨ_m generation.

Figure 8:

Fraction of COX present in either the P or F species in isolated mitochondria as a function of ΔΨ_m. Glu + malate (blue circles) and apyrase (red squares) titration data points from six or five independent titration experiments, respectively, are shown for each steady state. The concentration of COX present in the P or F state was calculated by converting the absorbance obtained in experiments similar to that shown in Fig. 7 to molar fraction as explained in the text. The total COX present was calculated from the hydrosulfite minus fully oxidized spectra of each mitochondrial experiment, and corresponded to 2 nmol/ml.

3.4. Free energy and relative electron flow rates between COX catalytic intermediates

We compared the free energy from the reduction of oxygen by cytochrome c to the free energy of the complete COX catalytic cycle assuming each intermediate reaction is at thermodynamic equilibrium. These energies are shown in Figure 8 as a function of ΔΨ_m for Glu + malate (GM) (Fig. 9A) and apyrase (Fig. 9B) titration and show in both cases an excess of energy from cytochrome c after proton pumping. This indicates that the overall COX catalytic cycle is thermodynamically favorable but not at equilibrium. We observe that the free energy from cytochrome c increases slightly with membrane polarization with added substrate (GM titration) whereas it remains constant with apyrase additions.

Figure 9:

Free energy from the reduction of oxygen by cytochrome c (4E_c/O2, open symbols) and free energy of the complete COX catalytic cycle assuming thermodynamic equilibrium (E₁ + E₂ + 2E₃ + E₄ = −8ΔΨ_m, filled symbols) as a function of ΔΨ_m for Glu + malate (GM) (A) and apyrase (B) titration. Data points from six (A) or five (B) independent titration experiments are shown. The excess energy from cytochrome c after proton pumping shows that the complete cycle is thermodynamically favorable and not at equilibrium.

Using the information on the relative abundance of COX intermediates with respect to cytochrome c redox state and ΔΨ_m, we attempted to calculate the mid-point potential values E⁰ for the reactions in the catalytic cycle, assuming thermodynamic equilibrium between each member of the redox couples. However, if a reaction is slow relative to the next one in the cycle, this condition is violated, and a range of apparent E⁰ values will be obtained as ΔΨ_m (and therefore conversion rate) is varied. Hence, the constancy (or lack thereof) of E⁰ for each step can be used to determine which steps are slower or faster relative to others in the COX catalytic cycle. This approach was followed for a simplified four reaction model of the catalytic cycle that treats the O_H and E_H optically invisible intermediates as a single species (OE). E⁰ values calculated via Equation 3a–d (see Materials and Methods) are shown in Figure 10 as a function of ΔΨ_m. Calculations for the step P to F in the reaction cycle (Fig. 10A) showed the greatest variation in E⁰ as a function of ΔΨ_m, especially in the case of apyrase titration, implying that this reaction is slow and consequently far away from equilibrium. Conversion of F to the OE species (Fig. 10B) was also away from equilibrium, especially when changing ΔΨ_m with apyrase. The OE to R transition (Fig. 10C) showed a similar pattern as that of the P to F conversion, although with a slightly smaller deviation from the ideal Nernst behavior. As expected, the R to P conversion (Fig. 10D), which requires only the binding of oxygen, exhibited a nearly constant E⁰ at all ΔΨ_m values assayed, consistent with the lack of proton pumping or electron movement from cytochrome c on the positive side of the membrane into the binuclear center during this step of the catalytic cycle. Varying the arbitrary low R fraction between 10⁻⁴ to 10⁻² did not significantly impact the slopes of the plots shown in Fig 9D and only increased $E_{R/OE}^0$ and $E_{R/P}^0$ by ~20 mV for every 10-fold increase in R concentration (not shown).

Figure 10:

Mid-point potential (E⁰) for the steps involved in the catalytic cycle of COX as a function of ΔΨ_m in isolated mitochondria titrated with GM (blue circles) or apyrase (red squares). Data points represent six (GM) or five (apyrase) independent titration experiments. E⁰ values were calculated according to Equations 3a–3d (see Materials and Methods) for the conversion of P to F (A), F to OE (B), OE to R (C) and R to P (D). The dashed horizontal line on each panel corresponds to the ideal Nernst behavior in which each redox couple is at equilibrium. Deviation of the data from this behavior suggests that the reaction shown is away from equilibrium.

4. Discussion

The chromophores of oxidative phosphorylation provided one of the earliest glimpses [40] of how redox potentials are used to convert energy via oxidative phosphorylation in intact systems. This was extensively exploited by Chance [41–46] and others to provide unique non-invasive exploration of the redox interactions in the cytochrome chain in isolated systems as well as within intact tissues. With the further development of rapid scanning optical methods and their application to intact biological systems, as well as a better understanding of the sources [23] and behavior [47] of the chromophores of oxidative phosphorylation, it has become clear that more information on the control and function of the oxidative phosphorylation in intact tissues may be available from this non-destructive optical approach. The development of robust optical transmission data and complete spectral analysis in the perfused heart is one example of these technical advances [6, 48]. In the present study we sought to characterize the relative distribution of the COX intermediate states under normal oxygen and physiological ΔΨ_m values. To our knowledge, the detection of appreciable concentrations the P and F intermediates under steady state turnover conditions has only been reported under highly oxidizing conditions by inducing reverse electron transport from the a hemes to cytochrome c in the case of mitochondria [22, 23], or at alkaline (pH 8.5-9) when using isolated COX [24, 25]. Many of the studies that have characterized the COX catalytic cycle (reviewed in [20, 21]) have used the stable CO-bound fully reduced COX as starting material, which after flash photolysis to induce the dissociation of CO and the binding of oxygen to the binuclear center, is initially in the R form. However, during normal respiring conditions in the presence of oxygen, we found that no R form of the binuclear center (nor the reduced form of the a heme, which has very similar spectral features with respect to R in the Soret region [30, 33–36]) can be appreciably detected in isolated enzyme (see Figs. 1 and 2) and mitochondria (compare Figs. 3 and 4), with the clear changes detected instead in the P and F states (see Figs. 5 and 6). The high affinity of COX for oxygen (K_m <10 μM even under energized conditions [49, 50]) is also consistent with the R form being negligible under normal oxygen concentrations.

Previous studies using either liver mitochondria [51] or isolated COX [36, 52] have interpreted the low absorbance observed close to 605 nm under steady state respiration to correspond to reduction of the a heme that is not within the binuclear center. These observations are complicated by one or more of the following factors: the use of TMPD as redox mediator, which is known to absorb at similar wavelengths as the COX chromophores [51], not monitoring the Soret spectral region where the P and reduced heme are clearly different (see in contrast Figs. 1–4), or the use of harsh methods for COX isolation involving ammonium sulfate precipitation that result in very slow turnover rates due to a blockage of electron transfer into the binuclear center [30]. The isolated COX we have used in the present study does not have such an impediment in the electron transfer between the a hemes (see Fig. 1D and [35]), which is known to be very rapid based on the short distance between the hemes [53]. Thus, our results in the presence of oxygen, both in isolated enzyme and in heart mitochondria, support the conclusion that in uninhibited ‘fast’ COX electrons do not reside in either the a or a₃ hemes at detectable levels under steady state respiration as judged by the lack of the ~445 nm peak that in difference spectra is characteristic of the reduced a and a₃ hemes [30, 33–36] (see Figs 2 and 4). Only when oxygen is removed will electrons coming from cytochrome c quickly accumulate in all redox centers of COX to almost simultaneously generate the R form of the binuclear center and reduce the a heme.

Our results also indicate that the artificial oxidized heme a:xreduced heme a₃ species obtained in the presence of cyanide [34] that has been used as a fitting parameter in several studies [5, 54] does not exist in the presence of oxygen, but has probably being confused with a mixture of the P and F intermediates. The prevalence of the P and F states in COX in mitochondria under normal experimental conditions confirms that oxygen binding to the R form of the binuclear center proceeds very rapidly, driving the enzyme to the highly unstable A intermediate (undetectable at physiological temperature) that quickly decays into the more stable P intermediate [20, 21]. The presence of detectable P and F intermediates was also confirmed by the appearance of the sharp trough in the ~655 nm region in respiring mitochondria (see Fig. 6) that could be well fitted using a combination of the P and F difference spectra obtained with H₂O₂ (see Fig. 1B) as references. The bleaching of the 655-660 nm absorbance visible in the absolute spectrum of oxidized COX is also known to occur when Cu_B in the binuclear center is reduced while maintaining the a₃ heme in the oxidized state [36, 38]. Thus, although we cannot discard that other species within the catalytic cycle also contribute to the observed trough in this spectral region, it is clear that the broad spectral shape observed upon reduction of the binuclear center (see Fig. 1B) does not contribute to the observed shape in respiring mitochondria. Furthermore, our P and F references were sufficient to fit the mitochondrial difference spectra, arguing against the need to include some other species such as an oxidized a₃-reduced Cu_B species that probably exhibits a somewhat different spectral shape in this region.

Taking into account the lower maximal absorbance (16-20%) for the P (607 nm) and F (580 nm) intermediates relative to the fully reduced form of COX (see Fig. 1B), we estimate that a large fraction of COX (~50% in the P and F states plus an undetermined amount in the O_H and E_H states) is involved in electron transfer in the presence of oxygen (see Fig. 8). This conclusion is in contrast with interpretations which assumed that the small absorbance detected under respiring conditions relative to the total absorbance of the R form [55] implied that only a few percent of COX is active at a given time. These results are relevant with regards to the proposals that mitochondrial supercomplexes are the functional units in which electron transfer occurs [56]. We can estimate the ratio of COX to complex III to be in the order of 5:1 based on reduced minus oxidized spectra similar to that shown on Fig. 3 and taking into account the extinction coefficients for cytochromes c₁ [57] and a + a₃ (see Materials and Methods). Given that the observed stoichiometry in supercomplexes Complex III and COX in a 2:1 ratio [58], at least 90% of the total COX in our heart mitochondria preparation would be unavailable to form a supercomplex. Therefore, the association of COX in a supercomplex is not necessary for it to receive electrons from cytochrome c.

Previously, we showed that the putative P intermediate increased linearly with respiratory rate and membrane depolarization in isolated mitochondria from pig muscle and heart [13, 15], while the F intermediate peak decreased steeply upon slight depolarization from the maximum ΔΨ_m when decreasing the ATP/ADP ratio with apyrase [13], which was reproduced in this study in rabbit heart mitochondria (see Fig. 7). However, the non-physiological range of ATP/ADP values attained in isolated mitochondria along with numerous other regulatory factors and isolation artifacts are very likely not representative of the mitochondria in a working heart cell. Thus, even with its inherent limitations in terms of experimental control, the heart data, when available, should be considered more representative of physiological mitochondrial function and regulation. An unexpected observation was that electron transfer between the membrane-bound cytochrome c₁ in Complex III and the soluble cytochrome c (which does not involve proton pumping) appears not to be in rapid equilibrium in mitochondria (see Fig. 7B). This differs from what has been observed in isolated Complex III [59], sub-mitochondrial particles [60], and cultured cells [47], but agrees with what was observed in isolated mitochondria when ATP was added [61]. Thus, it could be speculated that structural changes in the inner mitochondrial membrane related to hyperpolarization could decrease the relative fraction of cytochrome c that is able to rapidly access Complexes III and/or IV. The importance of cytochrome c availability in determining respiration makes the analysis of mitochondria in intact tissue of paramount importance, since partial depletion of cytochrome c during the isolation procedure, especially in hearts from smaller animals such as the rabbit, could potentially result in a different behavior of COX in isolated versus in situ mitochondria in an intact tissue.

It could be argued that the lack of rapid equilibrium between cytochromes c₁ and c is simply a consequence of the ~5-fold excess of COX relative to Complex III we estimate in heart mitochondria (see above). Although this could be an important factor in maintaining cytochrome c more oxidized than expected from the cytochrome c₁ redox state, our free energy calculations showing a higher reduction of cytochrome c than what would be needed to pump protons while transferring electrons to oxygen (see Fig. 9) suggest that the overall reaction catalyzed by COX is also out of equilibrium. Thus, the availability of reducing equivalents from cytochrome c does not seem to be the main controlling factor in determining COX activity, as has been proposed before for bacterial systems [62]. As expected for an overall out of equilibrium reaction, most of the steps in the COX catalytic cycle were found to exhibit a variable E⁰ as a function of ΔΨ_m (see Fig. 10), with the notable exception of the R to P transition, which does not involve electron input nor proton translocation. This analysis also confirms our assignment of the peaks at 607 and 580 nm to the P and F intermediates, respectively, and agrees with a negligible steady state concentration of the R form in the presence of oxygen. Given that the O_H to E_H transition, which also involves electron transfer from cytochrome c coupled to net charge translocation across the membrane [20, 21], could not be evaluated due to the lack of distinguishing optical absorbance signals between the two forms, future work will aim at quantifying the O_H species by monitoring the reported resonance Raman vibration coming from the Fe³⁺-OH bond [63] and that is thought to be absent in the E_H and the non-catalytical relaxed O forms [21].

In conclusion, our present studies reveal that the COX redox state in respiring mitochondria is much different than commonly assumed. In the presence of oxygen, only the P and F states of the heme a₃-Cu_B binuclear reaction center can be detected, with the R state of the binuclear center and the reduced a heme only accumulating under oxygen insufficiency. This assignment of absorbance features to the P and F intermediates was confirmed by examination of different spectral bandwidths under normoxic and hypoxic conditions. Thermodynamic analyses demonstrated that the steady state concentrations of the species we assign to the P and F intermediates are dependent on ΔΨ_m and the availability of electrons from cytochrome c in a manner consistent with out of equilibrium reactions. This information will prove to be fundamental in future studies aiming at understanding the control of respiration and oxidative phosphorylation in intact tissues and organs such as the heart using the optical monitoring of mitochondrial chromophores.

Highlights.

• Reduction of mitochondrial cytochrome c oxidase hemes not detected during respiration

• Spectral fitting identifies absorbance peaks as species formed after oxygen binding

• Cytochrome c redox state shows electron transfer to oxygen is out of equilibrium

• Membrane potential slows conversion rate of intermediates relative to oxygen binding