CO impedes superfast O₂ binding in ba₃ cytochrome oxidase from Thermus thermophilus

Abstract

Kinetic studies of heme-copper terminal oxidases using the CO flow-flash method are potentially compromised by the fate of the photodissociated CO. In this time-resolved optical absorption study, we compared the kinetics of dioxygen reduction by ba₃ cytochrome c oxidase from Thermus thermophilus in the absence and presence of CO using a photolabile O₂-carrier. A novel double-laser excitation is introduced in which dioxygen is generated by photolyzing the O₂-carrier with a 355 nm laser pulse and the fully reduced CO-bound ba₃ simultaneously with a second 532-nm laser pulse. A kinetic analysis reveals a sequential mechanism in which O₂ binding to heme a₃ at 90 μM O₂ occurs with lifetimes of 9.3 and 110 μs in the absence and presence of CO, respectively, followed by a faster cleavage of the dioxygen bond (4.8 μs), which generates the P intermediate with the concomitant oxidation of heme b. The second-order rate constant of 1 × 10⁹ M^-1 s^-1 for O₂ binding to ba₃ in the absence of CO is 10 times greater than observed in the presence of CO as well as for the bovine heart enzyme. The O₂ bond cleavage in ba₃ of 4.8 μs is also approximately 10 times faster than in the bovine enzyme. These results suggest important structural differences between the accessibility of O₂ to the active site in ba₃ and the bovine enzyme, and they demonstrate that the photodissociated CO impedes access of dioxygen to the heme a₃ site in ba₃, making the CO flow-flash method inapplicable.

The reduction of dioxygen to water in the heme-copper oxidases takes place at the high-spin heme a₃ and Cu_B heterodinuclear center (for review, see refs. 1 and 2). The reaction has been extensively investigated in several aa₃-oxidases by time-resolved spectroscopic techniques in combination with the CO flow-flash technique (1, 2), in which the reaction is initiated by photolyzing CO bound to heme in the presence of O₂ (3). The O₂ reduction has commonly been interpreted in terms of a unidirectional sequential mechanism (Scheme 1).

Scheme 1.

The conventional sequential unidirectional mechanism for the reduction of dioxygen to water.

The O₂ reduction in Thermus thermophilus ba₃, a B-type oxidase with distant sequence homology to the A-type oxidases (4), has received much less attention (5–7). The enzyme contains the four redox-active metal centers (8–10) and functions as a terminal oxidase for aerobic metabolism under limited oxygen concentration (8–11). It also possesses NO reductase activity (12) suggesting shared evolutionary lineage of O₂/NO reduction in this enzyme. In ba₃, the thermal dissociation of CO from heme in the dark is significantly faster (0.8 s^-1) (5) than in the bovine aa₃ (0.023 s^-1) (3, 13), and therefore CO flow-flash experiments on ba₃ require fast mixing; such experiments have recently been reported (6, 7). Moreover, the complex formed following CO photolysis from heme in ba₃ decays with a lifetime of approximately 30 ms (14), a rate much slower than that of O₂ binding to heme in the aa₃-oxidases (approximately 10 μs at 1 mM O₂). This raises the question whether the O₂ binding in ba₃ is compromised by the fate of the photodissociated CO.

In this study, we investigated the reaction kinetics of fully reduced T. thermophilus ba₃ with photoproduced O₂ in the absence and presence of CO. Our results show a superfast rate of O₂ binding to heme a₃ in the absence of CO and that the photodissociated CO directly or indirectly impedes dioxygen access to the active site. Both the O₂ binding and O─O bond cleavage rates are 10 times faster in reduced ba₃ than in the reduced bovine enzyme.

Results and Discussion

Time-Resolved Optical Absorption Spectra and Global Exponential Fitting.

Fig. 1 A and B shows the time-resolved difference spectra (post- minus pre-photolysis) recorded during the reaction of fully reduced ba₃ with photoproduced O₂ in the absence and presence of CO, respectively; we will refer to these forms of the enzyme as “ba₃-without-CO” and “ba₃-with-CO.” The spectra for the latter were recorded following the simultaneous photolysis of the O₂-carrier with 355 nm light and the fully reduced CO-bound ba₃ with a second 532-nm laser pulse for efficient photolysis. The spectra in the absence and presence of CO are referenced against the spectrum of the reduced enzyme (A) and the spectrum of the CO-bound reduced ba₃ (B), respectively. Fig. 2 shows a comparison of the O₂ reduction kinetics at 560 and 444 nm for ba₃-with-CO (open and closed circles, respectively) and ba₃-without-CO (open and closed triangles, respectively). It is clear that the early microsecond kinetics of the O₂ reduction are significantly slower for ba₃-with-CO compared to ba₃-without-CO, and the characteristic absorbance changes due to the oxidation and subsequent reduction of heme b observed at 560 nm in the absence of CO (open triangles) are much less pronounced in the presence of CO (open circles). The solid and broken lines are the absorbance traces at 560 and 444 nm, respectively, calculated on the basis of the global exponential fits discussed below.

Fig. 1.

Time-resolved optical absorption difference spectra (post-minus pre-photolysis) recorded during the reaction of the fully reduced ba₃ with dioxygen in the absence of CO (A) and the presence of CO (B). The spectra are those obtained after subtracting the spectral contribution of the O₂ complex, determined in a separate experiment. The spectra (SVD-filtered) were recorded at 15 delay times, equally spaced on a logarithmic scale, between 500 ns and 20 ms. The arrows (left to right) represent the direction of the absorbance change with time. Conditions: 0.1 M HEPES (pH 7.5), 0.1% DM; effective enzyme concentration: 2.6 μM (A) and 2.0 μM (B); optical path, 0.5 cm.

Fig. 2.

Comparison of the transient absorbance changes taking place during the reduction of dioxygen to water in ba₃ at 560 nm and 444 nm in the presence of CO (open and closed circles, respectively) and in the absence of CO (open and closed triangles, respectively). The kinetic traces are from the time-resolved data in Fig. 1 and are normalized to the total absorbance change. The solid and broken lines represent the absorbance traces at 560 and 444 nm, respectively, calculated on the basis of the global exponential fits (see text for details).

Singular value decomposition (SVD)-based global exponential fitting (15–18) resolved four apparent lifetimes (τ), 4.8 μs, 9.3 μs, 55 μs, and 1.0 ms, for ba₃-without-CO (Fig. S1). Three apparent lifetimes, 59 μs, 110 μs, and 0.82 ms, were obtained for ba₃-with-CO with no early microsecond lifetimes being observed.

Kinetic Analysis Based upon the Conventional Unidirectional Sequential Mechanism.

The apparent lifetimes and the associated spectral changes, the b-spectra (Fig. S2), obtained in the global exponential fitting do not represent real physical processes or different phases of a reaction sequence. Rather, they reflect the fact that all the steps involved in the mechanism are first-order or pseudo-first-order reactions. It is the proposed mechanism that connects the apparent rates to the microscopic rates and the b-spectra to the spectra of the intermediates in a real physical mechanism (17, 18). The simplest and mathematically defined mechanism is the conventional unidirectional sequential mechanism in which the experimental apparent rate constants are assigned to microscopic rate constants of the consecutive steps. We first analyzed the ba₃-without-CO data using a five-intermediate unidirectional sequential mechanism (Scheme 2). Because only three lifetimes (four intermediates) were obtained for ba₃-with-CO, we first calculated its experimental intermediate spectra on the basis of a four-intermediate unidirectional sequential mechanism (Scheme 3).

Scheme 2.

The conventional unidirectional sequential mechanism of O₂ reduction for ba₃-without-CO.

Scheme 3.

A four-intermediate unidirectional sequential mechanism of O₂ reduction for ba₃-with-CO.

Fig. 3 shows the experimental intermediate spectra for ba₃-without-CO (A) and ba₃-with-CO (B) calculated for Schemes 2 and 3, respectively. For easier comparison, the datasets are normalized to the concentration of ba₃-with-CO, and both are referenced versus the reduced enzyme, R. The spectrum of Int 2 for ba₃-without-CO (Fig. 3A, green curve) has no equivalent in the spectra for ba₃-with-CO. Importantly, this spectrum has the same shape as that of compound A, observed for the reduced bovine aa₃ using the traditional CO flow-flash method (19) but with a significantly smaller amplitude (approximately 50%); similar amplitude discrepancy was also noted in a recent fast-mixing CO flow-flash experiment on ba₃ (6). The spectrum of Int 3 for ba₃-without-CO (Fig. 3A, red curve) is well modeled by the heme a₃ spectrum of P (referenced versus the reduced heme a₃) and the oxidized-minus-reduced spectrum of heme b (Fig. S3) and will be referred to as P_I. The spectrum of Int 2 for ba₃-with-CO (Fig. 3B, red curve) has very similar shape to that of Int 3 for ba₃-without-CO (Fig. 3A, red curve) but smaller amplitude. The origin of this amplitude difference and that of Int 2 for ba₃-without-CO and the model compound A spectrum will be discussed in detail below. The observed spectral shape and amplitude of Int 4 for ba₃-without-CO (Fig. 3A, cyan) were found to be in good agreement with those of Int 3 for ba₃-with-CO (Fig. 3B, cyan). These spectra represent P with reduced heme b; we designate this intermediate as P_II. Similarly, good agreement is observed between the final intermediates (the oxidized enzyme) in the absence of CO (Fig. 3A, magenta) and presence of CO (Fig. 3B, magenta). While the spectra of the last two intermediates in the absence and presence of CO are in agreement and consistent with expected model spectra of the respective intermediates, the amplitude of Int 2 calculated from the ba₃-without-CO data using Scheme 2 is significantly smaller than that of the bovine compound A model spectrum. There is no physical reason that would justify such a large amplitude difference, and therefore we must conclude that the traditional unidirectional sequential scheme does not describe the experimental data.

Fig. 3.

The experimental intermediate spectra for the reaction of reduced ba₃-without-CO (A) and ba₃-with-CO (B) with dioxygen, extracted on the basis of Schemes 2 and 3, respectively. The spectra in A are normalized to the enzyme concentration in B. The intermediate spectra are referenced versus the reduced enzyme. A (green) Int 2, (red) Int 3, (cyan) Int 4 and (magenta) Int 5 of Scheme 2; the spectrum of Int 2 has the shape of compound A of the bovine heart oxidase but significantly lower amplitude. B (red) Int 2, (cyan) Int 3 and (magenta) Int 4 of Scheme 3.

Analysis of the ba₃-Without-CO Data Using a Reversible O₂-Binding Scheme.

The most obvious explanation for the reduced amplitude of Int 2 in Scheme 2 and the one proposed in a recent CO flow-flash study on ba₃ (6) is that the O₂ binding is reversible in ba₃. This mechanism is represented in Scheme 4. It should be noted that having only four experimental apparent rates and five microscopic rates in Scheme 4, the spectra of the intermediates can no longer be calculated directly. However, they can be obtained for a presumed degree of reversibility using the algebraic kinetic analysis method outlined below.

Scheme 4.

The sequential scheme involving a reversible first step (O₂ binding) during the reaction of ba₃ with photoproduced O₂ in the absence of CO. Note that k₁₂ is a pseudo-first-order rate constant: k₁₂ = k[O₂].

The first step in describing a reversible mechanism is constructing the kinetic matrix containing the microscopic rate constants and deriving its eigenvalues, λ_i (see SI Text for a detailed analysis). The eigenvalues with the sign reversed are the apparent rate constants r_i, which are the inverse of the apparent lifetimes τ_i(r_i = 1/τ_i). The last two apparent rates, r₃ and r₄, are equal to the microscopic rates of the last two steps, k₃₄ and k₄₅ in Scheme 4, respectively. By introducing the equilibrium constant, Q = k₁₂/k₂₁, we can relate the microscopic rate constants, k₁₂, k₂₁, and k₂₃, to the first and second apparent rates, r₁ and r₂, and express them as a function of Q (see SI Text for a detailed analysis):

[1]

[2]

[3]

Considering only real number solutions for k₂₃ places the following limits on Q:

[4]

This interesting and unexpected result means that the degree of reversibility cannot be set to any arbitrary value. Instead, its range is restricted by the values of the apparent lifetimes, and the closer in value they are, the narrower this range becomes. In the final step of the analysis, the intermediate spectra were obtained from the experimental b-spectra (b_exp) and the eigenvectors (W_rev) of the kinetic matrix constructed from the microscopic rates (18):

[5]

The above kinetic analysis was applied to ba₃-without-CO. From the two shortest apparent lifetimes, 4.8 μs and 9.3 μs, we calculated a minimum value for the equilibrium constant, Q_min = k₁₂/k₂₁ = 7.8, which indicates that the binding of O₂ to reduced ba₃ is practically irreversible. As Eq. 3 predicts and is shown in Fig. 4A, there are two values for each of the microscopic rates, k₁₂, k₂₁, and k₂₃, at every Q > Q_min. The dashed lines represent the case in which k₂₃ (circles) decreases and k₁₂ (triangles) increases, approaching r₂ and r₁ at Q = ∞, respectively, a situation reflecting the traditional mechanism in which the fastest process comes first. The solid lines represent the slow-to-fast order of the kinetic steps in which k₂₃ increases and reaches r₁ at Q = ∞ while k₁₂ decreases, approaching r₂ at Q = ∞.

Fig. 4.

(A) The dependence of the rate constants, k₁₂ (triangles), k₂₁ (squares), and k₂₃ (circles), in Scheme 4 on the equilibrium constant Q above its minimum value (7.8) for the ba₃-without-CO; at any Q > Q_min all three microscopic rates have two values (Eqs. 1, 2, and 3). Dashed lines: decreasing k₂₃ and increasing k₁₂, approaching r₂ and r₁ at Q = ∞, respectively. Solid lines: increasing k₂₃ and decreasing k₁₂, approaching r₁ and r₂ at Q = ∞, respectively. (B) The spectra of Int 2 for ba₃-without-CO calculated at the three Q values marked on the curves in A; the bold blue spectrum is at Q = Q_min, the dashed (green and red) and solid (green and red) curves represent the fast-slow and slow-fast rate combinations, respectively. The spectrum of compound A of the bovine aa₃ (CO flow-flash method) is the model spectrum (black curve).

Fig. 4B shows the spectra of Int 2 in Scheme 4 for ba₃-without-CO (blue, green, and red curves) calculated at the three Q values marked on the curves in Fig. 4A; the dashed curves (green and red) represent the fast-slow order of the microscopic rates, while the solid curves (green and red) represent the slow-fast rate combination. The spectrum of compound A of the bovine aa₃ is plotted for comparison (black curve). The amplitude of the Int 2 spectrum at Q_min (bold, blue trace) is only a fraction of the model compound A bovine spectrum. When the first step (O₂ binding) is slower than the second one (Fig. 4A, solid lines), the amplitude of Int 2 increases and approaches that of the model compound A spectrum at high values of Q (Fig. 4B, solid green and red curves). However, when the first (equilibrium) step is faster than the second (irreversible) step (Fig. 4A, dashed lines), the amplitude of Int 2 decreases further at higher Q (Fig. 4B, dashed, green and red curves, respectively). These results show unambiguously that only the mechanism with the slow-fast step combination (Scheme 5) produces an acceptable compound A spectrum and that the 9.3 μs and not the shorter 4.8 μs lifetime should be assigned to O₂ binding to heme . This assignment is confirmed by the dependence of the apparent lifetimes on the O₂ concentration. At half the O₂ concentration (45 μM), the 9.3-μs lifetime was increased by a factor of two to 18 μs, which is expected if there is no reversibility, while the 4.8-μs lifetime was unaffected (see SI Text and Figs. S4 and S5 for details). The 9.3-μs lifetime observed at 90-μM O₂ concentration corresponds to a lifetime of approximately 1 μs at 1 mM O₂ or a second-order rate constant for O₂ binding of 1 × 10⁹ M^-1 s^-1. This rate is 10 times faster than observed in our laboratory for the bovine enzyme in the absence or presence of CO.

Scheme 5.

The slow-fast sequential mechanism for the reaction of ba₃ with photoproduced O₂ in the absence of CO.

The slow-fast mechanism in Scheme 5 is in contrast to recent fast-mixing CO flow-flash studies on ba₃ (O₂ concentration of approximately 1 mM) in which apparent lifetimes of 5.3 μs and 13 μs were assigned to O₂ binding to heme a₃ and the subsequent formation of P_R, respectively (6). The reduced yield of compound A was attributed to the reversibility of O₂ binding based on a model assuming established coupled equilibria (6). This assumption is, however, not valid because at least one of the equilibria, that of CO leaving , has a reported lifetime of approximately 30 ms (14), more than 1,000 times longer than the measured rate of O₂ binding to heme a₃. However, the reported apparent rates and reduced yield of compound A (6) are in agreement with the slow-fast rate combination discussed above, suggesting that the 13 μs lifetime represents O₂ binding to heme a₃ and the 5.3-μs lifetime the subsequent P formation accompanied by heme b oxidation.

Analysis of the ba₃-With-CO Data Using a Slow-Fast Rate Combination.

For ba₃-with-CO, the 110 μs lifetime is assigned to the O₂ binding based on the following arguments. The spectrum of Int 2 in Scheme 3 for ba₃-with-CO (Fig. 3B, red curve) has the same shape as that of intermediate 3 (P_I) in Scheme 2 in the absence of CO (Fig. 3A, red curve) but with a significantly smaller amplitude, a situation reminiscent of the smaller-than-expected amplitude of Int 2 (compound A) for ba₃-without-CO discussed above. Therefore, we extracted the spectra of the intermediates for ba₃-with-CO using the slow-fast step combination in Scheme 6. Using this mechanism, the spectrum of Int 2 for ba₃-with-CO was found to be in good agreement with that of Int 3 (P_I) for ba₃-without-CO (Fig. S6), showing the 110 μs lifetime should be assigned to O₂ binding.

Scheme 6.

The slow-fast sequential mechanism for the reaction of ba₃ with photoproduced O₂ in the presence of CO.

The 110 μs O₂ binding corresponds to a lifetime of approximately 10 μs at 1 mM O₂, which is practically identical to the 13 μs lifetime (our assignment) observed in the recent fast-mixing CO flow-flash experiment on ba₃ (1 mM O₂ ) (6). The 110-μs step is followed by a faster process, 59 μs, which is attributed to the subsequent heme b rereduction. This rate is very similar to the rate of electron transfer between heme b and Cu_A for ba₃-without-CO (55 μs) and in good agreement with the analogous rate recently reported using the CO flow-flash method (7).

Consistent with our results on ba₃-without-CO and the recent fast-mixing CO flow-flash results on ba₃ (6, 7), the heme a₃ in ba₃ is not converted to the 580-nm oxyferryl form (F) but rather maintains the spectral characteristics of P. This suggests that the proton acceptor proposed to be responsible for the spectral difference between the P and F forms in the aa₃ enzyme, possibly the cross-linked tyrosine or the hydroxide bound to Cu_B, may be different from that in ba₃ (6, 7). It should also be noted that it is the spectral difference related to heme a₃ that distinguishes the P and F forms and not their relative positions in the order of the sequential intermediates. Accordingly, we designate the second 610 nm absorbing species in the ba₃ kinetics as a P state (P_II) regardless of its place in the scheme.

Time-Evolution of the Intermediates.

The relative concentration profiles of the intermediates of ba₃ in the absence and presence of CO as a function of time are shown in Fig. 5 (dashed and solid lines, respectively). The time profiles are calculated based on Scheme 5, established for the kinetic description of ba₃-without-CO and supported by the previously reported CO flow-flash experiment (6). For ba₃-with-CO, we have replaced the 9.3-μs O₂ binding in Scheme 5 with the observed 110-μs O₂ binding. The time profiles for ba₃-without-CO show that the 9.3-μs O₂ binding and the subsequent 4.8-μs P_I formation result in a limited, although observable, accumulation of compound A (Fig. 5, dashed green line). When O₂ binding slows down to 110 μs in the ba₃-with-CO reaction, there are potentially two faster steps, the 4.8-μs P_I formation and the subsequent 55-μs P_II formation. In this case, the predicted accumulation of compound A (Fig. 5, solid green curve) is too small to be detected; however, P_I reaches high enough concentration to be resolved by the exponential fitting (Fig. 5, solid red curve).

Fig. 5.

The time-dependent concentration profiles of the sequential intermediates for the reaction of reduced ba₃ with dioxygen in the absence and presence of CO (broken and solid lines, respectively). The time-dependent concentration profiles in the absence of CO were calculated based on the slow-fast unidirectional sequential model (Scheme 5) with apparent lifetimes of 9.3 μs, 4.8 μs, 55 μs, and 1.0 ms. The time profiles in the presence of CO were calculated based on the same scheme except the 9.3-μs O₂ binding was replaced with the experimental 110-μs lifetime.

The Superfast Steps in the ba₃ O₂ Kinetics. The Effect of CO on O₂ Access to Heme a₃.

The above results demonstrate unequivocally that the binding of photoproduced dioxygen to reduced ba₃ in the absence of CO is 10 times faster than the O₂ binding observed in the presence of CO in our double-laser flow-flash experiment and that reported using the CO flow-flash method (6) (our interpretation). These results suggest that the photodissociated CO, either bound to and/or trapped at a ligand docking site, impedes access of O₂ to heme a₃ in ba₃. Time-resolved step-scan FTIR studies have suggested that 15–20% of CO may be trapped at a ligand docking site following photolysis of CO from heme in ba₃ (20). The CO at a site close to Cu_B may interact with the heme methyl group and the propionate side chain of ring A, thus impeding access of O₂ to the binuclear center (20, 21). An alternative explanation for the slower O₂ binding rate in the presence of CO is that the reduced enzyme generated after CO photodissociation has a conformation less favorable for O₂ access to heme a₃ than the normal reduced enzyme.

The other remarkable feature of the O₂ reduction mechanism in ba₃ is the fast P_I formation and heme b oxidation (approximately 5 μs) in both the absence of CO (this study) and the presence of CO (our assignment) (6). This rate is approximately 10 times faster than the corresponding rate reported for the bovine enzyme (2, 22) and may arise from a shorter distance between heme a₃ and Cu_B in ba₃ (23, 24) or a redox potential difference that favors electron transfer from heme b²⁺ to heme .

Is Cu_B an Obligatory Stop for O₂ Entry to Heme a₃ in ba₃?

The obligatory path of CO to and from heme a₃ in the bovine enzyme has been proposed to involve the binding of CO to (13, 25), and an analogous mechanism has been postulated for ba₃ (14, 26); the O₂ ligand has been proposed to follow the same path (27). However, crystallographic studies using Xe and Kr to map out the O₂ pathway within ba₃ have indicated that there is no straight-access line from Xe1, the site closest to the binuclear cavity, to Cu_B, and that Trp229 and His283 appear to form a barrier that O₂ needs to circumvent to access the dinuclear center (21). These observations, together with the long lifetime of the transient complex in ba₃ (14), suggest that Cu_B may not necessarily act as an obligatory stop for O₂ entry to heme a₃ in ba₃-with-CO. In the absence of CO, the entry of O₂ to heme a₃ in ba₃ may or may not involve prior binding to Cu_B.

Possible Implications of CO on Intramolecular Electron Transfer.

Whether ba₃ can simultaneously accommodate two ligands at the active site is unclear (28–30). If simultaneous binding of CO to and O₂ to heme does occur, the CO-bound Cu_B may not be able to act as an electron donor during dioxygen reduction. Because the rate of P_I formation is the same in the absence of CO (this study) and presence of CO (our assignment) (6), the donor of the fourth electron required for dioxygen bond cleavage would presumably be the same in both cases, perhaps the cross-linked tyrosine (31, 32), as suggested for the reaction of O₂ with the mixed-valence aa₃ enzyme (32). Alternatively, the binding of O₂ to heme may reduce the affinity of for CO, causing CO to dissociate, thus allowing to act as an electron donor to the bound dioxygen.

Conclusions.

Our results demonstrate superfast O₂ binding in ba₃ in the absence of CO and that the presence of CO directly or indirectly impedes dioxygen access to the active site thus making the CO flow-flash method inapplicable. The O₂ binding in ba₃-without-CO is also 10 times faster than its binding to the bovine enzyme under similar conditions. Comparative studies on the bovine enzyme in our laboratory have shown that the presence of CO does not affect the O₂ binding rate. This suggests that O₂ binding in the bovine enzyme and ba₃ are inherently different. Crystallographic studies of Xe binding within the internal cavity of ba₃ show that the dimensions of the ba₃ O₂ channel pose no obstruction for O₂ access to the a₃-Cu_B center (21). In contrast, in Rhodobacter sphaeroides, Paracoccus denitrificans, and bovine aa₃ enzymes, there is a constriction point that reduces the channel diameter (21); therefore in these oxidases a conformational change would be required for dioxygen access to the binuclear site. We propose that the structural differences in the dioxygen channel between these two enzymes give rise to the 10-times faster O₂ binding in the ba₃ enzyme. The elevated O₂ transport and cleavage rates may be advantageous to the organism considering the low oxygen solubility under atmospheric pressure at 70 °C, the optimum growth temperature of the T. thermophilus HB8. Whether the CO binding to Cu_B interferes with the access of other ligands to the active site, including that of NO, will be determined in due course.

Materials and Methods

Enzyme and O₂-Complex Preparation.

The wild-type ba₃ was isolated from T. thermophilus HB8 cells according to previously published procedures (5, 33). The fully reduced CO-bound enzyme was generated upon addition of anaerobic solutions of sodium ascorbate and ruthenium (II) hexammine chloride to deoxygenated oxidized enzyme solutions under nitrogen atmosphere, followed by exposure to CO atmosphere. The photolabile O₂ carrier, (μ-peroxo)(μ-hydroxo)bis[bis(bipyridyl)-cobalt (III)] complex, was synthesized as previously described (34).

Time-Resolved Optical Absorption Measurements.

The reactions of the reduced ba₃ with O₂ were monitored by time-resolved optical absorption spectroscopy using a CCD detector (Andor Technology). The spectra were recorded between 360–760 nm at delay times from 500 ns to 20 ms after laser photolysis. The enzyme solutions with or without CO were mixed in a 1∶1 volume ratio with a solution of the photolabile O₂-carrier. The O₂ was generated upon photolysis with a 355-nm laser pulse (Nd∶YAG, 7 ns pulse). In the presence of CO, the reaction was initiated by simultaneously photolyzing the CO-bound enzyme with 532-nm laser light because of the high absorbance of the O₂ carrier at 355 nm. The time-resolved difference spectra were analyzed by SVD and global exponential fitting using Matlab (Mathworks) (15–18). To test the validity of the proposed mechanism, the extracted experimental intermediate spectra were compared to the respective model spectra of the proposed intermediates.