The Reactions of O₂ and NO with Mixed-Valence ba₃ Cytochrome c Oxidase from Thermus thermophilus

Abstract

Earlier CO flow-flash experiments on the fully reduced Thermus thermophilus ba₃ (Tt ba₃) cytochrome oxidase revealed that O₂ binding was slowed down by a factor of 10 in the presence of CO (Szundi et al., 2010, PNAS 107, 21010–21015). The goal of the current study is to explore whether the long apparent lifetime (∼50 ms) of the Cu_B⁺-CO complex generated upon photolysis of the CO-bound mixed-valence Tt ba₃ (Koutsoupakis et al., 2019, Acc. Chem. Res. 52, 1380–1390) affects O₂ and NO binding and the ability of Cu_B to act as an electron donor during O-O bond splitting. The CO recombination, NO binding, and the reaction of mixed-valence Tt ba₃ with O₂ were investigated by time-resolved optical absorption spectroscopy using the CO flow-flash approach and photolabile O₂ and NO carriers. No electron backflow was detected after photolysis of the mixed-valence CO-bound Tt ba₃. The rate of O₂ and NO binding was two times slower than in the fully reduced enzyme in the presence of CO and 20 times slower than in the absence of CO. The purported long-lived Cu_B⁺-CO complex did not prevent O-O bond splitting and the resulting P_M formation, which was significantly faster (5–10 times) than in the bovine heart enzyme. We propose that O₂ binding to heme a₃ in Tt ba₃ causes CO to dissociate from Cu_B⁺ in a concerted manner through steric and/or electronic effects, thus allowing Cu_B⁺ to act as an electron donor in the mixed-valence enzyme. The significantly faster O₂ binding and O-O bond cleavage in Tt ba₃ compared to analogous steps in the aa₃ oxidases could reflect evolutionary adaptation of the enzyme to the microaerobic conditions of the T. thermophilus HB8 species.

Significance

Our time-resolved optical absorption studies on the reaction of O₂ with the mixed-valence Thermus thermophilus ba₃ oxidase show that the final product is the oxyferryl P_M, indicating that Cu_B⁺ is able to provide an electron for the microsecond O-O bond splitting. This suggests that the lifetime of the Cu_B⁺-CO complex, reportedly generated upon CO photolysis, must be orders of magnitude shorter than the millisecond lifetime observed in the absence of O₂. We propose that O₂ binding to heme a₃ reduces the affinity of Cu_B⁺ for CO, allowing Cu_B⁺ to act as an electron donor for O-O bond splitting. The fast O₂ binding and O-O bond cleavage in ba₃ may be advantageous for the microaerobic conditions of the thermophilic HB8 bacterium.

Introduction

Heme-copper oxidases are classified into three subfamilies, A, B, and C, based on phylogenetic analysis (1,2) and play a key role in eukaryotic and bacterial respiration, catalyzing the reduction of dioxygen to water and conserving part of the free energy from the O₂ reduction as a transmembrane electrochemical proton gradient required for ATP synthesis (3, 4, 5, 6, 7, 8). The details of the mechanism of O₂ reduction and the exact nature of the intermediates generated during the reaction continue to be a matter of interest. The O₂ binding and reduction take place at the binuclear active site comprising Cu_B and a high-spin heme (a₃, b₃, or o₃), and these two metal centers are believed to donate one and two electrons, respectively, to the O₂ bound to the high-spin heme to ensure the breaking of the O-O bond (3,6). The fourth electron required for O-O bond cleavage is postulated to come from the low-spin heme in the fully reduced enzyme, and the cross-linked tyrosine in the case of the mixed-valence enzyme (9), in which the high-spin heme and Cu_B are reduced but the low-spin heme and Cu_A, which is the fourth metal center in many heme-copper oxidases, are oxidized.

Studies of the mechanism of O₂ reduction in the heme-copper oxidases have for decades been carried out using the CO flow-flash method, in which the reaction is initiated by photolyzing CO bound to the reduced high-spin heme in the presence of O₂ (10). The O₂ reduction has generally been interpreted in terms of a sequential “fast-slow” mechanism, with decreasing values of the apparent rates assigned to consecutive steps. However, multiwavelength time-resolved optical absorption (TROA) experiments in our laboratory have shown that during O₂ reduction in the fully reduced B-type oxidase, Thermus thermophilus ba₃ (Tt ba₃), O₂ binding is followed by faster breaking of the O-O bond (5,11,12). These studies were carried out in the absence and presence of CO using a photolabile O₂ carrier and showed that O₂ binding to heme a₃ in fully reduced Tt ba₃ in the absence of CO occurs with a superfast rate, 1 × 10⁹ M⁻¹ s⁻¹. This is 10 times faster than in the bovine enzyme (12) and was also found to be the case for NO binding using a photolabile NO carrier (11). Moreover, the O₂ and NO binding was found to be 10 times slower (1 × 10⁸ M⁻¹ s⁻¹) in fully reduced Tt ba₃ in the presence of CO, indicating that the photodissociated CO impedes the access of O₂ and NO to the active site (5,11,12). Previous infrared studies have reported that CO binds to Cu_B⁺ after photolysis of the fully reduced CO-bound Tt ba₃ (13, 14, 15) and the mixed-valence CO-bound (MVCO) enzyme (16,17), with a reported apparent lifetime of the Cu_B⁺-CO photoproduct of ∼35 ms (5,14) and ∼50 ms (17), respectively; this is in contrast to the microsecond lifetime (1.5 μs) of the Cu_B⁺-CO complex generated after photolysis of the fully reduced CO-bound bovine enzyme (18). A combined crystallographic and infrared study supports CO binding to Cu_B⁺ in Tt ba₃ after photolysis of CO from heme a₃ (15). Importantly, in the reaction of O₂ with the mixed-valence Tt ba₃, in which both heme b and Cu_A are oxidized, Cu_B⁺ is an obligatory electron donor for the breaking of the O-O bond, along with heme a₃ and the cross-linked tyrosine (9), which provide two and one electrons, respectively. Therefore, if the photodissociated CO remains on Cu_B⁺ in Tt ba₃ for milliseconds in the presence of O₂, Cu_B⁺ would be unable to act as an electron donor for rapid (microsecond) O-O bond cleavage in the mixed-valence ba₃.

In this study, we investigated the CO photodissociation and recombination of MVCO Tt ba₃ and the reaction of the mixed-valence enzyme with O₂ and NO after photolysis of the MVCO enzyme. We compare the TROA results with the corresponding data obtained for the fully reduced enzyme. The results show that the reaction of the mixed-valence Tt ba₃ with dioxygen terminates at the P_M state, indicating that Cu_B⁺ is able to donate on a microsecond timescale an electron that is required for breaking the O-O bond and forming the P_M intermediate. Thus, CO is unlikely to be bound to Cu_B⁺ on the millisecond timescale observed in the absence of O₂, and we suggest that the binding of O₂ to heme a₃²⁺ facilitates the dissociation of CO from Cu_B⁺, allowing the oxidation of Cu_B⁺.

Materials and Methods

The wild-type ba₃ enzyme was isolated from T. thermophilus HB8 as previously described (19,20). CO was purchased from Praxair (Danbury, CT) and the DDM detergent was obtained from Affymetrix (Santa Clara, CA). The standard grade (99.9%) potassium pentachloronitrosylruthenate(II) photolabile NO complex was obtained from Alfa Aesar (Haverhill, MA). The (μ-peroxo)(μ-hydroxo)bis[bis(bipyridyl)cobalt(III)] nitrate photolabile O₂ carrier was synthesized by previously described procedures (21,22). The photoproduced O₂ and NO concentrations were determined as previously described (11,12). All other chemicals were purchased from Sigma-Aldrich (St. Louis, MO) and Thermo Fisher Scientific (Waltham, MA).

Preparation of MVCO Tt ba₃ oxidase

The mixed-valence wild-type CO-bound ba₃ (heme a₃²⁺-CO, Cu_B⁺, heme b³⁺, Cu_A²⁺) in 0.1 M HEPES buffer (pH 7.5) containing 0.1% n-dodecyl-β-D-maltopyranoside was prepared by first taking the oxidized enzyme through several alternating cycles of vacuum and nitrogen. The deoxygenated sample was subsequently incubated under CO to generate the MVCO state, in which heme a₃ and Cu_B are reduced and heme b and Cu_A are oxidized. In instances in which the enzyme became slightly over-reduced, namely, with a fraction of heme b becoming partially reduced, 20 μM of deoxygenated ferricyanide was added under CO to the over-reduced enzyme.

CO flash photolysis of MVCO Tt ba₃

The flash-photolysis experiments on the MVCO Tt ba₃ were performed at room temperature in a deoxygenated quartz cuvette using a 532-nm pulse from a Q-switched DCR-11 Nd:YAG laser (∼7 ns full width at half maximum). The TROA difference spectra (post-minus prephotolysis) after photolysis of CO from heme a₃²⁺ were recorded between 350 and 760 nm at delay times logarithmically spaced in the 100 ns–800 ms time window. Each recorded spectrum was an average of 40 accumulations. The TROA spectra were analyzed using singular value decomposition (SVD) and global exponential fitting, which provided the apparent lifetimes and the associated spectral changes, the b-spectra (12,23,24).

Reactions of the mixed-valence ba₃ with photoproduced O₂ and NO: double-laser flow-flash approach

The reactions of O₂ and NO with the mixed-valence ba₃ were investigated using a photolabile O₂ or NO carrier, respectively, in combination with the double-laser flow-flash technique (11,12). In this approach, the reaction was initiated by simultaneously photodissociating the O₂ or NO complex with a 355-nm laser pulse and the MVCO ba₃ with a 532 nm laser flash. The TROA spectra were recorded between 350 and 760 nm at 12 logarithmically spaced delay times between 1 μs and 5 ms after CO photolysis for the double-laser O₂ study and at nine delay times between 5 μs and 2 ms for the double-laser NO study. To resolve additional steps during the reaction of the mixed-valence Tt ba₃ with O₂, we carried out TROA studies using a modified CO flow-flash method, described in more detail below.

Results

CO flash photolysis and recombination of MVCO Tt ba₃

The CO photolysis and rebinding dynamics of MVCO Tt ba₃ and potential intramolecular electron transfer after CO photolysis were investigated. The TROA difference spectra (post-minus prephotolysis) are shown in Fig. 1 A. The spectra were analyzed with SVD and global exponential fitting. Two exponentials appear to satisfactorily fit the TROA data. The spectral changes (b-spectra) corresponding to each apparent lifetime are represented in Fig. 1 B. The first b-spectrum (bs₁), a minor contribution, corresponds to an apparent lifetime of 49 μs and could arise from a less populated CO conformer (25, 26, 27). The bs₀ b-spectrum represents the difference spectrum extrapolated to infinite time. The large amplitude b-spectrum (bs₂) corresponds to an apparent lifetime of 330 ms and represents the rebinding of CO to heme a₃²⁺ (Fig. 1 B). This rate is similar to that reported for the fully reduced Tt ba₃ using transient absorption spectroscopy (5,28,29) but somewhat slower than reported by time-resolved infrared spectroscopy (5) and by time-resolved step-scan Fourier transform infrared (FTIR) measurements on the fully reduced and mixed-valence enzyme (14,17). As discussed below, we observed no backflow of electrons from heme a₃ to heme b after CO photolysis.

Figure 1.

(A) TROA difference spectra (post-minus prephotolysis) recorded after flash photolysis of the MVCO Tt ba₃. The arrows represent the direction of the absorption changes with time. The spectra were recorded at logarithmically spaced time delays between 100 ns and 800 ms. (B) The b-spectra resulting from a two-exponential fit of the time-resolved data are shown. The bs₁ corresponds to an apparent lifetime of 49 μs, and bs₂ represents an apparent lifetime of 330 ms; the bs₀ along the baseline represents the time-independent b-spectrum, the spectrum of the end product within the experimental timescale.

Double-laser approach using photolabile O₂ and NO carriers

The significantly faster thermal dissociation of CO from the ba₃-CO complex (0.8 s⁻¹) (19) compared to the CO-bound bovine enzyme (0.023 s⁻¹) (30) renders the traditional CO flow-flash method unsuitable for ba₃ kinetic studies (except by very rapid-mixing instruments) as reported previously for the fully reduced enzyme (11,12). To circumvent this problem, we used photolabile carriers of O₂ and NO in the current study and in our prior investigations (11,12). Because the photolysis of the carriers yields O₂ and NO concentrations far below the solubility levels of the gases, the rate of the respective ligand binding is relatively slow and, for O₂, sets an upper limit to additional observable reaction rates. To monitor faster reaction rates during the O₂ reaction, we modified our CO flow-flash technique. The details and results of the modified approach will be discussed later.

Reaction of mixed-valence Tt ba₃ with photoproduced NO

The binding of NO to the mixed-valence Tt ba₃ was investigated to explore ligand binding in the absence of fast redox chemistry. Although ba₃ has NO reductase activity (31), it is minimal and does not occur on the microsecond timescale during which NO binding to heme a₃²⁺ takes place. The reaction of the mixed-valence Tt ba₃ with NO was investigated using the double-laser flow-flash technique, described previously to investigate the reactions of NO and O₂ with the fully reduced enzyme in the presence of CO (11,12). This approach involves simultaneously photolyzing the MVCO enzyme with a 532-nm laser pulse and the photolabile NO carrier with a 355-nm laser pulse. The TROA difference spectra (post-minus prephotolysis) recorded between 5 μs and 2 ms are shown in Fig. 2 A. The spectrum at each delay time is an average of 20 accumulations. The TROA difference spectra are those obtained after the subtraction of the spectral contribution of the photolyzed NO complex.

Figure 2.

(A) TROA difference spectra (post-minus prephotolysis) of the reaction of the mixed-valence Tt ba₃ with photoproduced NO in the presence of CO. The arrows represent the direction of the absorption changes with time. The spectra are referenced against the MVCO enzyme present before photolysis and were obtained by subtracting the spectral contribution of the photolyzed NO complex. The spectra were recorded at nine logarithmically spaced delay times between 5 μs and 2 ms. (B) The b-spectra resulting from a one-exponential fit to the time-resolved data are shown. The bs₁ (dashed line) corresponds to an apparent lifetime of 170 μs (at 130 μM NO concentration), and the bs₀ (solid line) corresponds to the time-independent b₀ spectrum.

SVD-based global exponential fitting of the TROA difference spectra resolved a single lifetime of 170 μs, which is attributed to NO binding to heme a₃²⁺ based on the shape of the corresponding b-spectrum (Fig. 2 B). The bs₁ (dashed line) represents the apparent lifetime of 170 μs, whereas the bs₀ (solid line) is the nonzero time-independent spectrum, namely, the spectrum of the end product. The lifetime of 170 μs at 130 μM NO corresponds to 4.3 × 10⁷ M⁻¹ s⁻¹ second-order rate constant of NO binding in the mixed-valence Tt ba₃ in the presence of CO, which is roughly half of the rate found for NO binding to the fully reduced enzyme in the presence of CO (1 × 10⁸ M⁻¹s⁻¹) and 20 times slower than the rate observed for NO binding to the reduced Tt ba₃ in the absence of CO (1 × 10⁹ M⁻¹s⁻¹) (11).

Reaction of the mixed-valence Tt ba₃ with photoproduced O₂

The investigation of the mixed-valence Tt ba₃ with O₂ in the presence of CO is critical for determining whether prolonged (millisecond) binding of the photodissociated CO to Cu_B⁺, reported after photolysis of the CO-bound fully reduced Tt ba₃ (5,14) and the mixed-valence enzyme (17) in the absence of O₂, prevents Cu_B⁺ from donating an electron required for the cleavage of the O-O bond and the presumed formation of the P_M intermediate. For this experiment, the double-laser flow-flash technique was utilized to simultaneously photolyze the MVCO enzyme with a 532-nm laser pulse and the photolabile O₂ carrier with a 355-nm laser flash. The resulting TROA difference spectra (post-minus prephotolysis) are shown in Fig. 3 A. The spectra are referenced against the MVCO enzyme present before photolysis, and the spectral contribution of the O₂ carrier has been subtracted.

Figure 3.

(A) TROA difference spectra (post-minus prephotolysis) of the reaction of the mixed-valence Tt ba₃ with photoproduced O₂ in the presence of CO. The arrows represent the direction of the absorption changes with time. The spectra were obtained by subtracting the spectral contribution of the photolyzed O₂ complex. The spectra were recorded at 12 logarithmically spaced delay times between 1 μs and 5 ms. (B) The b-spectra resulting from a one-exponential fit to the time-resolved data are shown. The bs₁ (dashed line) corresponds to an apparent lifetime of 380 μs (at 50 μM O₂ concentration), and the bs₀ (solid line) corresponds to the time-independent b₀ spectrum.

SVD and global exponential fitting applied to the TROA difference spectra resolved one apparent lifetime of 380 μs, which corresponds to a bimolecular rate constant of 5.3 × 10⁷ M⁻¹ s⁻¹ at 50 μM O₂. This rate constant is again roughly half of the rate constant observed for O₂ binding to the fully reduced Tt ba₃ in the presence of CO and 20 times slower than observed for O₂ binding to the reduced enzyme in the absence of CO (12). The b-spectrum associated with the 380-μs apparent lifetime, bs₁, is shown by the dashed line in Fig. 3 B; bs₀ represents the time-independent b₀ spectrum. At the low O₂ concentration (50 μM) produced by photolyzing the O₂ carrier, the O₂ binding becomes rate limiting, preventing any faster kinetic step from being detected.

Detection of faster rates in the reaction of the mixed-valence Tt ba₃ with O₂

In our combined conventional CO flow-flash (single laser pulse) and TROA spectroscopic experiments, in which CO-bound cytochrome oxidase is mixed with a saturated O₂ buffer solution, the light intensity profiles of the pulsed probe beam are recorded before and after the sample is photolyzed by a laser pulse. This yields the post-minus prephotolysis absorption difference spectra. For technical reasons, the time duration between the two pulsed probe beam recordings is long enough to allow a significant fraction of CO to dissociate from the MVCO Tt ba₃ and react with O₂ (19).

In an attempt to reduce the extent of the dark reaction contribution to the recorded spectra and resolve rapid steps during the reaction of the mixed-valence Tt ba₃ with O₂, we modified the sequence of the data collection in our CO flow-flash method. Instead of recording the reference probe light intensity immediately after mixing the MVCO enzyme with the saturated O₂ buffer solution, the photolyzing laser pulse was delivered first. This was followed by recording the probe light intensity at a selected delay time; the reference signal was recorded several hundred milliseconds later, shortly before the sample was pushed out of the flow cell. Because the reactions of interest take place in the first millisecond after the laser pulse, their time course is practically unaffected by the much slower dark reaction of the unphotolyzed fraction of the enzyme. The difference spectra recorded in this manner were collected in the 350–750 nm wavelength interval at nine delay times ranging from 2 μs to 1 ms. The spectra obtained after subtracting the dark reaction contribution are shown in Fig. 4 A. Because of the relatively slow mixing, a significant fraction of the enzyme is still “lost” in the dark reaction, and, accordingly, the size of the recorded signal and the signal/noise ratio are significantly smaller than in the double-laser study reported above. However, the time resolution is improved, and by applying SVD and global exponential fitting to the time-resolved spectral data, we were able to resolve two apparent lifetimes, 18 and 38 μs. The corresponding b-spectra, bs₁ and bs₂, are shown in Fig. 4 B. The accuracy of the respective lifetimes is reflected in the significance values of the v-vectors generated by the SVD analysis and used in the exponential fit. The shorter lifetime is based mainly on the time dependence of the v₃ vector with a small significance value (0.016) (Fig. 4 A, inset), and thus, it must be regarded as less accurate than the longer lifetime, which is supported by the v₂ and v₁ vectors with higher significance values (0.027 and 0.27, respectively). Mechanistic analysis and the extraction of the spectra of the intermediates will be addressed in the Discussion.

Figure 4.

(A) TROA difference spectra of the reaction of the mixed-valence Tt ba₃ with O₂ obtained using the modified CO flow-flash method; the spectra are those obtained after subracting the dark reaction contribution (see text for details). The arrows represent the direction of the absorption changes with time. Inset: the circles and triangles represent the v-vectors, plotted on a logarithmic timescale (v₁: filled circles, v₂: open circles, v₃: filled triangles) from the SVD analysis of the time-resolved data. The solid lines represent the reproduced v-vectors using a two-exponential fit to the data. (B) The b-spectra resulting from a two-exponential fit to the time-resolved data are shown. The bs₁ (solid line) corresponds to an apparent lifetime of 18 μs, and the bs₂ (dashed line) corresponds to an apparent lifetime of 38 μs. The bs₀ along the baseline corresponds to the time-independent b₀ spectrum.

Discussion

CO photolysis of MVCO Tt ba₃ shows no electron backflow

Photolysis of the MVCO Tt ba₃ was carried out to determine if there was backflow of electrons from heme a_3²⁺ to heme b³⁺. Any electron backflow would result in a distinct absorption at 560 nm because of the reduced heme b, which was not observed (Fig. 1); previous transient absorption and FTIR studies on MVCO Tt ba₃ reported minimal (4.2%) or no electron backflow (16,32). Although our CO photolysis data were fitted with two exponentials, the b-spectrum associated with the 49-μs lifetime, bs₁, is very small (Fig. 1 B) and does not have the characteristic spectral changes expected upon the reduction of heme b. In contrast, previous flash-photolysis and recombination studies of the MVCO aa₃ oxidases showed significant electron backflow, ∼75% in Rhodobacter sphaeroides aa₃ (33,34) and ∼25% in the bovine enzyme (35, 36, 37) on early microsecond timescale. The lack of electron backflow observed in previous FTIR studies on the mixed-valence Tt ba₃ was attributed to a larger difference in the midpoint redox potentials of the metal centers in Tt ba₃ compared to the aa₃ oxidases (16). A high driving force for electron transfer from heme b to heme a₃ was suggested based on electrochemical titration of the two hemes, making electron backflow from heme a₃²⁺ to heme b³⁺ unlikely (16,38,39).

Reactions of photoproduced NO and O₂ with mixed-valence Tt ba₃

Previous TROA studies in our laboratory have shown that the rate of O-O bond splitting during the reaction of O₂ with the fully reduced Tt ba₃ was 5 μs, significantly faster than the ∼35-ms lifetime reported for the Cu_B⁺-CO complex formed after CO photolysis from heme a₃²⁺ in the fully reduced enzyme in the absence of O₂ (5,14). Therefore, if CO remains on Cu_B⁺ for milliseconds in the presence of O₂, Cu_B⁺ would be unable to act as an electron donor for the rapid O-O bond cleavage in the fully reduced ba₃ (12). However, Cu_B⁺ is not an obligatory electron donor in the fully reduced enzyme because the four electrons required for O-O bond splitting could theoretically come from heme a₃ (two electrons), heme b (one electron), and possibly the cross-linked tyrosine (one electron) (9). In contrast, during the reaction of O₂ with mixed-valence ba₃, in which both heme b and Cu_A are oxidized, the breaking of the O-O bond requires the oxidation of Cu_B⁺. Hence, if CO remains bound to Cu_B⁺ for milliseconds in the mixed-valence enzyme in the presence of O₂, then the O-O bond would remain intact, and the reaction would terminate at the heme a₃-O₂ state, A_M. However, if the reaction proceeds to the P_M state on a fast (microsecond) timescale, CO would not remain bound to Cu_B⁺ for milliseconds during the breaking of the O-O bond.

To resolve this issue, we performed a mechanistic analysis of the TROA data obtained for the reactions of the mixed-valence ba₃ with the photoproduced NO and O₂ after photolysis of the CO-bound enzyme. The experimental intermediate spectra were calculated based on a single-step unidirectional mechanism using the b-spectra and assigning the apparent rate constant calculated from the lifetime to the microscopic rate constant. Fig. 5, A and B shows spectra of intermediate 1 (Int 1) (dashed lines) and intermediate 2 (Int 2) (solid lines) referenced against the MVCO enzyme, for the NO and O₂ reaction, respectively. The spectra of Int 1 for NO (Fig. 5 A, dashed line) and O₂ (Fig. 5 B, dashed line) are similar and reflect the typical difference spectrum between the reduced heme a₃ and its CO-bound form, also observed in the CO recombination experiment (Fig. 1). The Int 2 difference spectrum for NO (Fig. 5 A, solid line) is the difference between the NO-bound and CO-bound ba₃, and the spectrum is analogous to the corresponding spectrum for the fully reduced enzyme reported earlier (11). The Int 2 spectrum for the O₂ reaction (Fig. 5 B, solid line) shows the characteristic peak of a P intermediate at ∼610 nm, which was previously observed in the reaction of O₂ with the fully reduced Tt ba₃, both in the presence and absence of CO (12). Although the rate of the formation of Int 2 is determined by the rate of O₂ binding, its spectral features are not that of A_M but of P_M. This is because the rate of the A_M-to-P_M conversion is significantly faster than the O₂ binding rate at the low (50 μM) O₂ concentration.

Figure 5.

The experimental intermediate spectra for the reaction of the mixed-valence Tt ba₃ with photoproduced NO (A) and O₂ (B). Int 1 is represented by the dashed lines, and Int 2 is represented by the solid lines. The intermediate spectra are referenced against the MVCO-bound enzyme. The spectrum of Int 2 for the O₂ reaction is that of P_M.

Generation of A_M and its conversion to P_M

Although the reaction of mixed-valence Tt ba₃ with the photoproduced O₂ clearly shows that the final intermediate is P_M, this approach does not allow us to gain insight into the A_M-to-P_M conversion because of the rate-limiting O₂ binding at the 50 μM O₂ concentration. However, by using the CO flow-flash technique with a modified data collection sequence and ∼10 times higher O₂ concentration, we were able to resolve two closely spaced apparent lifetimes, 18 and 38 μs. Because the two lifetimes have close values, theoretically, either of them could be associated with the rate of O₂ binding at the elevated O₂ concentration. As discussed in detail in our previous study of the reaction of O₂ with the fully reduced ba₃ (12), two unidirectional sequential schemes having a fast-slow mechanism (the faster 18 μs step is followed by the slower 38 μs step) or a slow-fast rate combination (the 38 μs step precedes the 18 μs step) can be considered for analysis. By examining the shapes and amplitudes of the intermediate spectra produced by each of the schemes, we can establish the order of the lifetimes in the reaction sequence (12).

For easier spectral recognition and comparison to the fully reduced enzyme, the intermediate spectra of the mixed-valence enzyme that were generated using either scheme were referenced against the reduced heme a₃ spectrum; it should be noted that heme b does not contribute to the recorded difference spectra. As expected, both the fast-slow and slow-fast rate combinations produced the same first and third (last) intermediate spectra. The spectrum of the first intermediate with reduced heme a₃ (created upon the immediate photolysis of the MVCO enzyme) has zero amplitude, as expected, because the spectrum is referenced against reduced heme a₃. It is the shape and amplitude of the second intermediate spectrum for the two schemes that determines the details of the kinetics. Similar to that observed earlier for the kinetics of the reaction of O₂ with the fully reduced Tt ba₃ (12), the fast-slow order of rates did not give a meaningful amplitude for the second intermediate (data not shown). However, the slow-fast combination of lifetimes shown in Scheme 1 produced a spectrum for the second intermediate (Fig. 6 A, black curve), referenced versus that of reduced heme a₃, which has both the shape and the relative amplitude expected for compound A based on the corresponding intermediate of the bovine enzyme.

Scheme 1.

Slow-fast mechanism for the reaction of the mixed-valence Tt ba₃ with O₂.

Figure 6.

Comparison of A and P intermediate spectra between the mixed-valence and fully-reduced forms of the ba₃ enzyme. (A) (Black curve) Shown is the experimental spectrum of Int 2 in Scheme 1, the O₂-bound heme a₃ in the mixed-valence enzyme, A_M, obtained in the reaction with O₂ (∼625 μM) using the modified CO flow-flash method (see text for details). (Gray curve) The spectrum of compound A of the fully reduced ba₃, A_R, was generated during the reaction with photoproduced O₂. (B) (Black curve) Shown is the experimental spectrum of intermediate 3 in Scheme 1, P_M. (Gray curve) The spectrum of P_II, with heme b rereduced (12), was generated during the reaction of the fully reduced ba₃ with photoproduced O₂; the heme b contribution is not present.

Although the signal/noise ratio of the spectrum for compound A_M (Fig. 6 A, black curve) is modest because of limited accumulation of this intermediate, the spectrum is in good agreement with the spectrum of compound A generated during the reaction of the fully reduced ba₃ with photoproduced O₂ (Fig. 6 A, gray curve). The decay of A_M to P_M of ∼18 μs is significantly faster (5–10 times) than that recently reported by Brzezinski and co-workers for the mixed-valence enzyme under similar conditions (40); the origin of this difference between the two studies is unknown.

The spectrum of the third (last) intermediate, referenced against the reduced heme a₃, is shown in Fig. 6 B (black curve). The spectrum is consistent with the difference spectrum between P_M and the reduced form of heme a₃, thus confirming that P_M is the final state. The experimental difference spectrum of P_M is in good agreement with that of P_II (in which heme b is rereduced) generated during the reaction of the fully reduced ba₃ with photoproduced O₂ in the absence of CO (Fig. 6 B, gray curve); P_II was chosen for comparison because it has no heme b contribution when referenced against the fully reduced enzyme. The above analysis shows that the data from the modified CO flow-flash experiment is able to resolve the mechanism for the mixed-valence enzyme. Thus, not only does the fully reduced Tt ba₃ have the peculiar slow-fast rate combination for O₂ binding and P formation, the mixed-valence enzyme does as well.

The binding of O₂ and NO to heme a₃²⁺ in the mixed-valence Tt ba₃ appears to be a factor of two slower (∼5 × 10⁷ M⁻¹ s⁻¹) than observed in the fully reduced enzyme in the presence of CO (∼1 × 10⁸ M⁻¹ s⁻¹). Slower O₂ binding in mixed-valence bovine heart aa₃ and R. sphaeroides aa₃ compared to the fully reduced enzymes was attributed to electron backflow in the mixed-valence enzymes after photolysis of CO from heme a₃ (35,41). However, no electron backflow was observed in our CO flash-photolysis studies on the MVCO Tt ba₃, in agreement with previous studies (16,32). The twofold differences in the rates of O₂ and NO binding between the mixed-valence and fully reduced Tt ba₃ could arise from conformational differences at the binuclear site in the two states after CO photolysis.

P_M versus P_R

The exact nature of P_M in the mixed-valence state compared to that of the P_R intermediate in the reduced form of the different heme-copper oxidases, including their spectroscopic properties, continues to be a subject of discussion (35,40, 41, 42, 43, 44). Previous studies in our laboratory on the reaction of O₂ with the fully reduced and mixed-valence bovine heart aa₃ oxidase reported spectral differences between P_R and P_M (35). The latter form will be referred to here as P. Our earlier studies showed that the spectrum of P_R in the bovine enzyme was best modeled by a pH-dependent mixture of the spectra of compound A, P, and F, which was analyzed with a kinetic model consisting of branched electron transfer pathways linked by proton transfer (42,45). However, in the study reported here on Tt ba₃, no distinct spectral differences were observed between P_R generated during the reaction of the fully reduced ba₃ with O₂ (12) and P_M generated in the current study on the mixed-valence ba₃ (Fig. 6 B). These results show that the P_R intermediate observed during the reaction of O₂ with the fully reduced ba₃ is a “pure” P form and not a mixture as appears to be the case in the bovine aa₃ enzyme. Interestingly, the P_R intermediate was not observed in our TROA studies of the reaction of O₂ with the fully reduced R. sphaeroides aa₃ at neutral pH. Instead, compound A was converted to F (46); this observation is in contrast to observations by Brzezinski and co-workers (47), which could be related to different enzyme preparations. Notably, the 580-nm F form is not observed during the reaction of the fully reduced Tt ba₃ with O₂ (12,32,48). We have attributed the lack of observation of F in fully reduced Tt ba₃, P_R in fully reduced R. sphaeroides aa₃, and the presence of both P and F in the bovine enzyme to differences in the relative rates of internal electron transfer during the conversion of compound A to P_R and proton movement during the conversion of P_R to F (46). It should be noted that the lack of the detection of P or F forms during O₂ reduction in the different oxidases, and their absence in experimental schemes, does not mean they cannot be part of a hypothetical molecular mechanism; the relative rates of respective steps could simply result in a population of a particular intermediate too small to be detected.

O₂ binding and O-O bond cleavage

Although the rate of O₂ and NO binding to reduced heme a₃ in the mixed-valence Tt ba₃ is not very different from that observed for the fully reduced Tt ba₃ in the presence of CO, both rates are approximately an order of magnitude slower than the rate of O₂ binding to the fully reduced Tt ba₃ in the absence of CO, 1 × 10⁹ M⁻¹ s⁻¹ (12). Hence, the current study indicates that the photodissociated CO, presumably through its binding to Cu_B⁺, impedes the binding of O₂ to heme a₃ in the mixed-valence enzyme as it does in the fully reduced enzyme (12). The reaction of the mixed-valence Tt ba₃ with O₂ inherently has to be investigated in the presence of CO because the preferential reduction of heme a₃ and Cu_B and the oxidation of heme b and Cu_A are possible only in the presence of CO.

The conversion of A_M to P_M in the mixed-valence Tt ba₃ (∼18 μs) is slower than the conversion of A_R to P_R in the fully reduced enzyme (∼5 μs) (12). This trend was also observed in the bovine and R. sphaeroides aa₃, although both rates in the aa₃ oxidases are significantly slower than in Tt ba₃. In the aa₃ oxidases, the slower formation of P_M, ∼300 μs in R. sphaeroides aa₃ (35,41) and ∼160 μs in the bovine enzyme (35,41), compared to that of P_R, ∼35 and ∼55 μs in the bovine and R. sphaeroides aa₃, respectively (41,45), has been attributed to slower internal proton transfer in the mixed-valence enzyme or a higher activation barrier in forming a true ferric-peroxy transition state in P_M before the breaking of the O-O bond (40,41,44); this may also be the case for Tt ba₃. The significantly faster rates of P_M and P_R (P_I in (12)) formation in the mixed-valence and fully reduced Tt ba₃ compared with those of P_M and P_R in the aa₃ oxidases, respectively, could arise from a shorter distance between heme a₃ and Cu_B in ba₃ (49,50). Alternatively (or additionally), electron transfer and/or proton transfer from the cross-linked tyrosine could be significantly faster in ba₃ than in the aa₃ oxidases.

Conclusions

Our results show that in mixed-valence Tt ba₃, Cu_B⁺ is able to transfer an electron to the dioxygen-bound heme a₃, which is required to break the O-O bond on a microsecond timescale. This indicates that CO does not remain on Cu_B for a prolonged period of time (milliseconds) in the presence of O₂ but still long enough to decrease the rate of O₂ binding by an order of magnitude as observed in the fully reduced enzyme (12). In light of the fast rate of O₂ binding to heme a₃ in Tt ba₃ in the absence and presence of CO and the high affinity of Cu_B⁺ for CO in this enzyme (19,28), it is unlikely that O₂ would replace CO on Cu_B⁺. Therefore, the path of O₂ and NO to and from the high-spin heme in ba₃ may not involve obligatory binding to Cu_B as has been suggested for the bovine enzyme based on time-resolved infrared measurements (18,51) and transient optical absorption studies performed as a function of CO pressure (52). We propose that during the CO flow-flash method or when using the double-laser approach involving CO, the O₂ binding to heme a₃²⁺ in Tt ba₃ causes CO to dissociate from Cu_B⁺ in a concerted manner through steric and/or electronic effects, thereby allowing Cu_B⁺ to act as an electron donor during O-O bond cleavage. As articulated previously (11), a change in the geometry around Cu_B⁺ from tetrahedral to more square planar could facilitate the dissociation of CO from Cu_B⁺ and its subsequent oxidation. In contrast, the Cu_B⁺-CO complex photoproduct in the bovine enzyme decays on early microsecond timescale (1.5 μs) (18), significantly faster than the 10 μs O₂ binding to heme a₃ observed in the presence of CO at ∼625 μM O₂. This indicates that the CO dissociation from Cu_B⁺ does not rate limit O₂ and NO binding to Cu_B or heme a₃ in the bovine enzyme, and therefore, these ligands could potentially bind to Cu_B⁺ before the high-spin heme in this enzyme.

The transient presence of O₂ at heme a₃ and CO on Cu_B in Tt ba₃ is supported by the NO reductase activity of this enzyme, which requires, at least transiently, the simultaneous binding of two NO molecules at the active site. Importantly, the reactions of O₂ with the fully reduced and mixed-valence Tt ba₃ both follow a slow-fast mechanism, in which compound A generation is followed by two times faster P formation. The more open O₂ channel in Tt ba₃ compared to the aa₃ oxidases and other structural differences at the active site of the two oxidase families could reflect evolutionary adaptation of the thermophilic enzyme to increase both the rate of O₂ diffusion to the binuclear active site and subsequent O-O bond splitting, permitting physiologically significant reaction rates under the microaerobic conditions of the T. thermophilus HB8 bacterium.

Author Contributions

Ó.E., I.S., and C.F. designed research. I.S. and C.F. performed research. I.S. analyzed the data. T.S. provided the T. thermophilus ba₃ for all the measurements. Ó.E. and I.S. wrote the manuscript.

Editor: Jane Dyson.