The X-ray structure of the resting oxidized form showed three large water cluster near the P-side surface as shown in Figure 6A (Yano, et al., 2016). The structures of the interfaces between these clusters indicate that any proton exchange is impossible between these clusters. The water cluster closest to heme a ₃, the blue-colored cluster in Figure 6A, containing Mg²⁺ ion site, is designated as Mg-H₂O cluster. The cluster has a channel-like structure extending to the P-side surface as shown in Figure 6B. However, a 1.0 Å probe analysis shows that direct contact is impossible between the cluster and the P-side phase. Furthermore, five proline residues surround the channel to stiffen the channel (Figure 6B). The pathway for the O₂ inlet from the transmembrane surface of the subunit III and the H₂O (the CcO reaction product) outlet are located near the Mg-H₂O cluster as marked by the dark blue and red arrows in Figure 6C. The junction point is at heme a ₃ of the O₂ reduction site. The two propionate groups of heme a ₃ interact with the Mg-H₂O cluster by forming hydrogen bond with the water molecules in the cluster. However, the -CH₂-CH₂- moiety and the surrounding amino acid residues seem to block any proton exchange between the water cluster and the mobile waters in the O₂/H₂O pathway. One of the three imidazole groups coordinated to Cu_B is hydrogen-bonded to a fixed water (water 10) in the water cluster (Figure 6C inset).

The Mg-H₂O cluster, colored in blue in Figure 7A, is connected to the H-pathway (the blue and red arrow) with a short hydrogen bond network shaded in gray in Figure 7A (Yano, et al., 2016). The Mg²⁺ in the cluster is coordinated to the carboxyl group of Glu198, which is coordinated to the Cu_A site with the peptide C=O group (Figure 7A). Glu198 shows a redox-coupled structural changes as given in Figure 7B, suggesting significant changes in proton transfer efficiency between Glu198 and Arg439 at the one end of the short hydrogen bond network.

The improved X-ray structures of the Mg-H₂O cluster in both oxidation states (Yano et al., 2016) provide well-resolved water molecules, giving essentially identical numbers of the water molecules, 20.75 and 20.30, for the resting oxidized and fully reduced forms, respectively. This result suggests absence of water molecule exchange with those located outside of the cluster, consistent with the water-accessibility analyses of the cluster. Many hydrophilic amino acid residues and the two heme a ₃ propionates are included in the Mg-H₂O cluster as indicated in Figure 7C. Six amino acid residues and the two propionate groups of heme a ₃ in the figure are able to accept protons reversibly. The 20–21 water molecules tightly packed in the interior of the Mg-H₂O cluster, many polar but non-charged residues, and the peptide groups tightly hydrogen-bonded with each other, as shown in Figure 7C, are likely to provide a dielectric environment of the interior of the Mg-H₂O cluster closely similar to that of the N-side phase. The above structural findings indicate the sufficient capacity of storage of, at least, four protons in the Mg-H₂O cluster. Furthermore, the redox-coupled structural changes in Glu198 would facilitate the reversible and redox-coupled proton accepting capacity between the cluster and the N-side phase.

The outlet pathway for the product water was identified as described in Figure 6C. Many polarized oxygen atoms surround the pathway giving a highly hydrophilic environment. All residues surrounding the O₂-inlet pathway are hydrophobic except for Glu242. The location of the opening of the product (water) exit at the transmembrane region, as shown in Figure 6C, is critical for effective suppression of proton back leak from the P-side when the product water molecules are released. As described above, the inside of the O₂ inlet is highly hydrophobic while that of the H₂O outlet is highly hydrophilic. The D-pathway is attached to the O₂ inlet pathway at Glu242. Thus, protons, transferred from the D-pathway to the O₂ inlet pathway, would be readily transferred to the H₂O outlet pathway, unless the O₂-reduction site traps them for water formation. The D-pathway mechanism proposes that pumping protons are transferred through the D-pathway to the PLS located near the BNC. Then, the H₂O outlet pathway would readily take up the pumping protons and release them from the H₂O outlet, located in the trans-membrane surface, to dissipate the membrane potential. Therefore, this structure does not support the D-pathway mechanism.

X-ray structural changes of the fully reduced CcO in crystals after exposure to excess amounts of O₂ were followed by static X-ray diffraction experiments in SPring-8, for determination of the intermediate forms (Shimada, et al., 2021). The data sets were analyzed by multiple structural analyses (Shimada, et al., 2021). The multiple structural analysis is indispensable, since data sets obtained from the CcO crystals during the CcO reaction are highly likely to include multiple intermediate forms. If the X-ray diffraction data were analyzed without taking the existence of the multiple components into account, the analysis would provide an unreal structure, corresponding to the weighted average of multiple component structures.

Figure 8 shows schematic representations of the BNC of the intermediate forms, R, A, P/F mixture, O and E. As shown in panel A, the R form has no ligand in both metal sites. The A form shows an oxymyoglobin-type structure in panel B. The structure suggests that the Cu_B-O distance is significantly longer than the Fe _a3-O distance (2.64 Å vs. 1.93 Å) (Shimada, et al., 2021). Panel C shows a schematic representation of the O₂-reduction site of the X-ray structure obtained from 1:1 P_m/F mixture crystals. The difference in the structures between the two forms is too small to detect at the resolution. The Fe _a3-O distance of 1.70 Å is consistent with the bond length of Fe⁴⁺ = O²⁻, not that of Fe⁴⁺-OH⁻. The distance between the two oxygen atoms of the ligands (2.54 Å) indicate the existence of a short hydrogen bond between the two ligands (Shimada, et al., 2020). In the structure of the BNC of the O-form shown in panel D, the Fe _a3-O distance of 1.82 Å is consistent with that of Fe³⁺-OH⁻, while the unusually long Cu_B-O distance, 2.70 Å, was obtained. The BNC structure of the E-form in panel D shows significantly shorter Cu_B-O distance compared with that of the O-form (2.32 Å vs. 2.70 Å). No significant structural difference was detectable between the O- and E-forms except for the Cu_B-O distance, as shown in panels D and E. The Cu_B-O distance, 2.32 Å, suggests that the Cu_B is in either Cu¹⁺-OH⁻ or Cu¹⁺-OH₂ state (Shimada, et al., 2021). The O→E transition is coupled with uptake of a proton from N-side to the O₂ reduction site. Thus, Cu_B in the E-form is likely to accept the proton to form Cu¹⁺-OH₂. Protonation of Cu¹⁺-OH⁻ would stabilizes the Cu¹⁺ state by the net positive charge increase. The water channel of the H-pathway is in the closed state in all the intermediate forms except for the R form, which has the open state (Shimada, et al., 2020; Shimada, et al., 2021).

A cryo-EM analysis for the intermediate forms of a bacterial CcO has been reported (Kolbe, et al., 2021) providing amazing structural findings, such as a peroxide dianion, intact dioxygen molecule and a superoxide anion bound at the O₂-reduction site of the O, P and F forms, respectively. The reported structure indicates that the Fe _a3-Cu_B distance of the R-form is as short as 3.7 Å. The short distance suggests a strong electromagnetic interaction between the two metal ions, while no significant anomality in the absorption spectrum is detectable in the report. However, conditions for the experiments and the structural analyses of image data have not been provided sufficiently for evaluation of their final structural results.

Based on the X-ray structure of the CO-bound fully reduced CcO, it has been proposed that uneven bridging of O₂ to the two metals in the BNC is critical for slow formation of the peroxide-bound form so that the steady state level of the peroxide intermediate is lowered down to the negligible level (Yoshikawa and Shimada, 2015). The result summarized in Figure 8B for the A-form confirms experimentally the proposal.

It has been proposed, by comparison of the X-ray structures of various inhibitor-bound forms, that the water channel of the H-pathway opens only in the R-form in the catalytic cycle (Yoshikawa and Shimada, 2015). However, it is possible that the water channel is open also in the intermediate forms. This long-standing proposal has been confirmed experimentally by these X-ray structural analyses of the intermediate forms as given above (Shimada, et al., 2020; Shimada, et al., 2021).

For maximal efficiency in proton/electron coupling, each of the four transitions coupled with a proton pump in the catalytic cycle of CcO should be essentially irreversible. In other words, the intermediate forms, P_m, F, O and E, should have high electron affinity sites. The X-ray structural analyses for the intermediate forms as given in the previous section provided various insights in this point as follows: The unusually long Cu_B-O distance in the O-form suggests very weak negative charge influence of the OH⁻ to Cu_B ²⁺, indicating that Cu_B ²⁺ has an extremely high electron affinity (Shimada, et al., 2021). The X-ray structures of the F-form (as well as the P_m form) shows the Fe _a3 ⁴⁺ = O²⁻ structure, as a high electron affinity site, confirming crystallographically the structure of the F-form, proposed by a resonance Raman analysis as shown in Figure 1. The high electron affinity of Cu_B in the E-form as described above is likely to indirectly increase the effective electron affinity of the Fe_a3 site by decreasing the electron population in Fe _a3. The significant acceleration of the electron transfer to the Fe _a3 site by reduction of Cu_B ²⁺ was found by a rapid freeze EPR and stopped flow absorption spectral analyses (Jancura, et al., 2006), confirming kinetically that the Fe _a3 ³⁺ site in the E-form has high electron affinity. However, the structure of Fe _a3 itself does not suggest high electron affinity. Although Tyr244 neutral radical has been proposed as the high affinity site of the P_m-form, X-ray structural identification of the radical structure would be impossible even at the highest resolution presently available (1.3 Å) (Shinzawa-Itoh, et al., 2021). This proposal has been confirmed spectroscopically as described below.

The improved X-ray structures of bovine CcO strongly suggest that the Mg-H₂O cluster can accept at least four protons reversibly (Yano et al., 2016) as mentioned in Section 3.1.1. Thus, four protons for each catalytic cycle are collected into the Mg-H₂O cluster from the N-side in the open state (in the R-form) and pumped to the P-side from the Mg-H₂O cluster in the closed state, one by one, driven by heme a oxidation. Thus, in each of the four proton-pumping transitions (P→F, F→O, O→E, and E→R), one proton is released to the P-side coupled with uptake of one water-forming proton from the N-side. However, this is inconsistent with an experimental finding that, in each of the two transitions, P→F and F→O, two proton equivalents are collected from the N-side coupled with release of one proton to the P-side as described above, suggesting that one pumping proton in addition to one water-forming proton is collected from the N-side in each proton-pumping transition (Faxén, et al., 2005). This collected proton is unlikely to be transferred to the Mg-H₂O cluster instantaneously in each proton-pumping transition since the water channel is closed. Thus, a pool for the collected proton is necessary below the open/closed point of the water channel. In each proton pumping transition, one proton is taken up from the N-side to the proton pool. Thus, the four protons kept in the proton pool in each catalytic cycle are transferred to the Mg-H₂O cluster in the R-form. The proton pool are likely to facilitate effective proton collection from the N-side phase under extremely low proton concentration. An experimental result supports the existence of the proton pool (Zaslavsky et al., 2004). Further details are given in Supplementary Materials for the reference (Shimada et al., 2017) and in the reference (Wikström et al., 2018).

It has been well-established that, in both P_m and P_r forms, O₂ bound at the A-form is completely reduced to give Fe _a3 ⁴⁺ = O²⁻, and Cu_B ²⁻-OH⁻. Thus, no intermediate species during the O₂ reduction process during CcO reaction was identified experimentally, although formation of a peroxide-type intermediate has been expected by various synthetic-chemical analyses (Yoshikawa and Shimada, 2015). Perhaps, the first successful experimental identification of an intermediate between the A and P_m forms was accomplished by a flash photolysis analysis for the O₂ reduction by the fully reduced Thermus thermophilus CcO in which heme a is replaced with heme b (Poiana, et al., 2017). At 10°C and pH 7, after photolysis of the CO-bound ba ₃ CcO, absorbance changes at 610 nm showed O₂-binding at a time constant of 3.6 µs, followed by P_r formation at 110 µs, while at 560 nm, heme b oxidation was observed at 11 µs. This result indicates that the heme b oxidation is 10 times faster than P_r appearance. With an increase in pH up to 10, the heme b oxidation was slowed down to a time constant of 38 μs (3.5 times slowed) without any significant pH effect on the O₂ binding and on the P_r formation. As described above, in bovine and bacterial aa ₃ CcO, P_r is formed by electron donation from reduced heme a (Fe_a ²⁺) to the A-form as follows, A F e a 3 2 + − O 2 , C u B 1 + , T y r 244 O H + F e a 2 + → Pr F e a 3 4 + = O 2 + , C u B 2 + − O H − , T y r 244 O − + F e a 3 + , where Tyr244O⁻ denotes deprotonated Tyr244. The above result for the ba ₃ CcO, that is, the faster heme b oxidation than the Pr appearance indicates that the Fe _b ²⁺ oxidation provides an intermediate form before the Pr formation. The pH sensitivity of the heme b oxidation suggests that the new intermediate is protonated. Thus, a hydroperoxo-bound form, Fe _a3 ³⁺-O-OH⁻, has been proposed for the new intermediate form, including Cu_B ¹⁺and Tyr244O⁻, assuming that Fe _a3 ²⁺-O₂ has absorption spectrum closely similar to that of Fe _a3 ³⁺-O-OH⁻ at least at 610 nm, since no significant spectral change is detectable corresponding to the heme b oxidation at 610 nm (Poiana, et al., 2017). However, the assumption is in conflict with a consensus in the field of the hemoprotein chemistry. That is, upon the transition from Fe²⁺-O₂ (or Fe³⁺-O₂ ⁻) to Fe³⁺-O-OH⁻, the hemochrome type spectrum in the α-band region (550–600 nm) is eliminated giving significant absorbance decrease in the region. The consensus has been confirmed recently by a synthetic chemical approach (Kim, et al., 2020). Extensive spectroscopic (especially resonance Raman) analyses are desirable for identification of the structure of the intermediate.

As mentioned above, in the P_m form, the bound O₂ has been completely reduced to provide Fe _a3 ⁴⁺ = O²⁻ and Cu_B ²⁺-OH⁻. The X-ray structure of the BNC strongly suggests that one of the four electron equivalents and one proton equivalent required for formation of the P_m form are from Tyr244OH, leaving a neutral radical (Tyr244O•). However, in spite of various experimental trials (Jose, et al., 2021 and references therein), the existence of the radical in the P_m form had not been experimentally proven. The Fe⁴⁺ = O²⁻ site in the P_m form was examined extensively by variable-temperature, variable-field magnetic circular dichroism (VTVH-MCD) spectroscopy using E. coli ubiquinol oxidase which has a ubiquinol site instead of Cu_A site, which obscures the MCD feature of the Fe⁴⁺ = O²⁻ site (Jose, et al., 2021). The VTVH-MCD results showed that the Fe _O ⁴⁺ = O²⁻ is magnetically coupled to additional paramagnetic sites assignable to the Cu_B ²⁺ and Tyr244O• sites, as schematically shown in Figure 9. The antiferromagnetic coupling between the Fe _O ⁴⁺ = O²⁻ and Cu_B ²⁺-OH⁻ is induced by the super-exchange pathway through the hydrogen bond between the two oxygen atoms indicated by a dotted line in the Figure. The antiferromagnetic coupling between the Cu_B ²⁺ and Tyr244O• is facilitated by His240 which is covalently linked to Tyr244 and coordinated to Cu_B. The role of His240 is described by a black curve for the sake of simplicity. The ferromagnetic coupling between Tyr244O• and the Fe _O ⁴⁺ = O²⁻ is promoted by the hydrogen bond between the OH group of the hydroxyfarnesyl ethyl group and O• group of Tyr244. These magnetic couplings and spin density values prove that the neutral radical in P_m is located at Tyr244 (Jose, et al., 2021). The neutral radical is reduced to TyrO⁻ in the Pm→Pr transition (Jose, et al., 2021).

It has long been known that mitochondrial respiration is controlled by the proton motive force (pmf). However, its mechanism is still essentially unknown. The effect of the pmf on the CcO function was examined by using an artificial vesicle preparation from the mitochondrial inner membrane, designated as SMP (submitochondrial membrane particle), which is a preparation of vesicles inside-out relative to intact mitochondria (Bjorck, and Brzezinski, 2018). By addition of ATP to the outside of SMPs, the pH in the inside of the vesicle can be lowered to create the pmf to the SMP membrane. The effects of the pmf on the oxidative phase of the CcO reaction were examined by following the absorption spectral changes of the CcO in the SMPs after flash-photolysis of the CO-bound fully reduced CcO at 445 nm and 605 nm in the presence of ATP (Figure 10). Both hemes a and a ₃ contribute to the absorbance changes at 445 nm roughly equally, while the 605 nm changes are mainly induced by heme a. Following large changes due to the R→A and A→P_r transitions coupled with heme a oxidation within the initial 100 μs and a much smaller and slower change for about 1 ms, due to P_r→F, without a significant redox state change in heme a, the F→O transition with the time constant of ∼5 ms coupled with heme a oxidation was observed. The effects of the pmf were examined by the addition of valinomycin and FCCP, which collapse completely the pmf (the red traces). The red traces at 605 nm are essentially identical to those of the black traces (Figure 11B), indicating that heme a oxidation is insensitive to the membrane potential. However, significant decrease (14%) in the amplitude of the 5 ms phase without changing the time course is detectable at 445 nm (Figure 10A).

Absorbance spectrometric and electrometric analyses previously suggested that during the F→O transition, F receives one electron equivalent without significant influence on its absorbance spectrum giving an intermediate form, Fe _a3 ⁴⁺ = O²⁻, Cu_B ¹⁺, designated as Fr and that only one of the two transitions, F→F_r and F_r→O, is coupled with a proton-pump. However, because the F_r formation is much slower than the F_r→O transition, the F_r is not detectable in the F→O transition (Zaslavsky, et al., 1993; Zaslavsky, 1999; Brändén, et al., 2006).

The third phase at 445 nm in the presence of the pmf (the black trace) corresponds to the absorbance decrease only due to oxidation of heme a (Figure 10A). In other words, the F→O transition is suppressed by the pmf. In fact, the decrease in the amplitude of the 5 ms process upon application of the pmf is essentially identical to the absorbance difference between the F- and O-forms at 445 nm. However, heme a oxidation was insensitive to the pmf, as shown by the 605 nm traces in Figure 10B. This result indicates that heme a oxidation induces the transition from the F-form to the F_r-form, which has the absorbance spectrum identical to the F-form. Perhaps, this is the first conclusive experimental evidence for the involvement of the F_r-form in the normal catalytic process. The F→Fr transition is insensitive to the membrane potential, since this transition is driven by the electron transfer from heme a, which is parallel to the membrane surface. However, the Fr→O transition, which is coupled with proton uptake from N-side and proton pumping, is suppressed by the pmf. Thus, these results strongly suggest that the activity of CcO under turnover conditions is controlled at the Fr→O transition, which produces the initial intermediate of the catalytic cycle of the CcO reaction (i.e., the O-form). Thus, this site is reasonable for the feedback regulation of CcO function. Important future works would be experimental trials for identification of the structural basis for avoiding the membrane potential influence on the Pr-F transition.

Atomic molecular dynamics simulations (Sharma et al., 2017) indicated that protonation of His413 in the water channel of the H-pathway was necessary for incorporation of water molecules into the water channel to form a proton transfer pathway from the N-side to the hydrogen bond network of the H-pathway. The X-ray structure at 1.8 Å resolution (PDB ID: 1V54) shows that the main part of the water channel is not sufficiently hydrated for proton transfer through it, indicating that proton-transfer through the water channel is impossible. However, the improved X-ray structures at 1.65 Å or better (PDB ID: 5B1A, 5B1B, 5ZCQ, 5ZCP, 7COH, 7VUW, 7VVR, and 7YPY) show that the water channel is fully hydrated to provide the structure identical to that predicted by the above simulations, assuming that His413 is protonated. Thus, these improved X-ray structures demonstrate that the proton transfer through the water channel is possible. It seems that the 1.8 Å resolution (PDB ID: 1V54) is not sufficiently high for resolving these water molecules. This point has been mentioned previously (Shimada, et al., 2018).

Functions of the K-, D-, and H-pathways of a mitochondrial CcO from the yeast, Saccharomyces cerevisiae, were examined by a recently developed method for mitochondrial DNA mutagenesis (Maréchal, et al., 2020). The mutants prepared include Asn99Asp, Glu243Asp, and Ile67Asn in the D-pathway, Gln411Leu, Gln413Leu, Ser382Ala, Ser458Ala, Ser455Ala, and Ser52Asp in the H-pathway, and Thr316Lys in the K-pathway (The yeast numbering is used in this section.) The effects of these mutations were examined by the cell growth rates, the oxygen consumption rates of the mitochondrial membrane fragments, and the energy coupling efficiency (H⁺/e⁻ ratio) in intact isolated mitochondria respiring a respiratory substrate, α-ketoglutarate. The H⁺/e⁻ ratio was estimated from the ADP/O ratios in state 3 subtracting the background O₂ consumption in state 4. No significant effect of these H-pathway mutations is detectable on these CcO activities. Furthermore, the kinetics of electron and proton transfer during the reaction of the fully reduced forms of the Ser382Ala and Ser458Ala mutant CcOs with O₂ were investigated by a flow flash method to show that these mutant CcOs were as active as the wild-type CcO (Bjorck, et al., 2019). Thus, it has been concluded that H-pathway is not involved in the proton pumping function of the S. cerevisiae CcO.

However, the structural modifications introduced by the H-pathway mutations, as mentioned above, do not seem to influence significantly the proton pumping function of the H-pathway. The five mutation sites for Gln441, Gln413, Ser382, and Ser458 are located in or near the water channel of the H-pathway through which mobile water molecules transfer protons to the hydrogen bond network starting from Arg38 hydrogen-bonded to the formyl group of heme a. In order to block the water accessibility, mutation to the bulkier residue is necessary as in the case of the double mutation for bovine CcO, Val386Leu/Met390Trp (Shimokata et al., 2007). The structural changes in these Gln411Leu, Gln413Leu, Ser458Ala, and Ser455Ala mutations are too small to effectively block the water channel. Ser382 is hydrogen-bonded to the OH group of the hydroxyfarnesyl ethyl group of heme a to eliminate a water cavity in the water channel to block the water access to the hydrogen bond network. However, an energy minimization analysis suggests that the structural change in the Ser382Ala mutation is too small to eliminate the closed/open transition as detectable upon complete reduction of the resting oxidized CcO. Ser52Asp mutation, located in a fairly hydrophilic environment in the P-side end of the H-pathway, is highly unlikely to affect the function of the H-pathway. In general, a mutational result that provides no significant effect on its function is not conclusive evidence that the mutated residue has no critical role in its function, as mentioned in Section 2.4. Furthermore, the various structural diversities in the H-pathway of mitochondrial CcO, such as in Asp51, suggest significant diversity in the proton pumping function of the H-pathway. It is possible that the H-pathway pumps protons in cow, but not in yeast, as discussed in Section 2.5.

In the S. cerevisiae CcO, His413 in the mammalian CcO is replaced with Gln413. If the H-pathway in the S. cerevisiae CcO has a proton-pumping function identical to that of the mammalian CcO, the exchange of His413 to Gln413 suggests that the protonation of His413 is not necessary for the proton transport function of the water channel of the mammalian CcO. High-resolution structural analysis for the yeast CcO is desirable.

The H-pathway mechanism proposes that proton pump of CcO is driven by the electrostatic repulsions between the pumping protons on the hydrogen bond network of the H-pathway and the net positive charges created upon oxidation of the heme a. In fact, X-ray structure shows that heme a is hydrogen-bonded to the hydrogen bond network of the H-pathway (Figure 3). In order to evaluate whether the electrostatic interactions are strong enough for the proton pumping, azide-induced structural perturbations of the H-pathway were examined crystallographyically. The two-azide binding to the O₂-reduction site of the resting oxidized CcO (Figure 11A) induced significant structural changes in the side chains of Ser382 and Met383, as shown by a red circle and a blue oval, respectively, as described in Figures 11B, C. These side chains are in the water channel of the H-pathway near its upper end close to the heme a formyl group, which is at the bottom end of the hydrogen bond network of the H-pathway. The brown and gray dotted surfaces denote water cavities determined using probes with a radius of 1.2 and 0.8 Å, respectively. A possible water pathway to the formyl group from the N-side is marked by a dotted thick arrow in each panel. In the resting oxidized form structure (Figure 11C), although a possible water pathway is drawn by a dotted arrow, water molecules are not accessible to the formyl group, at least, in the physiological time scale. The continuous gray surfaces of the water channel structure of the two azide-bound form (Figure 11B), indicate that water molecules are accessible to the formyl groups within the physiological time scale. On the other hand, it has been reported that, the redox potential of heme a was significantly decreased by azide (Wilson, et al., 1972). The water channel opened by the azide binding is highly likely to decrease the proton level in the hydrogen-bond network of the H-pathway. This electron affinity decrease in heme a upon azide binding strongly suggests the existence of significant electrostatic interactions between the protons on the hydrogen-bond network of the H-pathway and heme a.

It has been shown that CO binding to the fully reduced form induces the open to closed transition of the water channel essentially identical to that detectable upon complete oxidation of the fully reduced form (Yano, et al., 2016). In order to investigate the dynamic aspects of the closed/open structural transition in the water channel of the H-pathway, the structural changes after photolysis of the CO-bound fully reduced CcO were examined by using time-resolved XFEL and infrared (IR) techniques with a pump-probe method (Shimada, et al., 2017b). Figure 12 illustrates schematically the structures of the BNC before CO-photolysis and 20 ns and 100 μs after CO-photolysis. At 20 ns, CO was migrated stoichiometrically to Cu_B ¹⁺, concomitantly with a decrease in the Fe _a3-Cu_B distance from 5.31 Å to 5.05 Å without influencing the structure of the water channel of the H-pathway (Shimada, et al., 2017b). At 100 ms approximately 75% of CO has been released from Cu_B ¹⁺. The residual CO at Cu_B ¹⁺ at 100 μs is not given in the figure for the sake of simplicity. The CO release from Cu_B ¹⁺ induced the closed to open transition in the water channel, giving 45% open and 55% closed structures. The coexistence of the open and closed structures in the X-ray structural results obtained at 100 μs was found by examination of the possibility of multiple structures. Ignoring the existence of the multiple structures would provide an unreal structure as in the case reported (Ishigam et al., 2017). At 100 μs, although 75% of the bound CO was released from Cu_B ¹⁺, only 45% of the channels are in the open state, indicating that the response of the water channel against the O₂ reduction site is not instantaneous, perhaps due to the slow structural changes in the 380-385 segment of the helix X. It is noteworthy that these dynamic aspects of the transition are undetectable without this time-resolved technique.

These results suggest critical roles of the Cu_B site in preventing back-leak of the pumping protons. In analogy to the CO binding to Cu_B ¹⁺, CO binding to Fe _a3 ²⁺ is unlikely to close the water channel instantaneously so that heme a oxidation, which triggers the proton pump, occurs before the water channel closure, giving back-leaks of pumping protons. However, the stoichiometric CO migration at 20 nsec as mentioned above indicates the absence of spontaneous reverse CO migration from Cu_B ¹⁺ to Fe _a3 ²⁺. In other words, in the CO binding process, direct CO binding to Fe _a3 ²⁺ without forming Cu_B ¹⁺-CO is highly unlikely. Furthermore, the stoichiometric CO migration from Fe _a3 ²⁺ to Cu_B ¹⁺ without the structural changes in the water channel suggests that the CO migration from Cu_B ¹⁺ to Fe _a3 ²⁺ occurs only when the water channel is closed. These structural findings suggest a mechanism of the water channel closure upon O₂ binding as summarized in Figure 12B. One of the water molecules in the Mg-H₂O cluster (water 10 in Figure 7) is hydrogen-bonded to one of the three histidine imidazole groups coordinated to Cu_B (Figure 12Ba). When the cluster is fully protonated (or has received four proton equivalents) (Figure 12Bb), the hydronium ion triggers a coordination state change in Cu_B to stimulate the O₂ binding. Thus, O₂ at Cu_B (Figure 12Bc) induces migration of the heme a ₃ plane to close the water channel through a structural relay system including the heme a ₃ vinyl group, Leu381, and Ser382. After the channel closure, the O₂ is migrated to Fe _a3 ²⁺ (Figure 12Bd). Then, the electron transfer from Fe _a ²⁺ is induced by the O₂ bound at Fe _a3 ²⁺, triggering the proton pump.