Mechanism and energetics of proton translocation by the respiratory heme-copper oxidases

Biochimica et Biophysica Acta 1767 (2007) 1200 – 1214 www.elsevier.com/locate/bbabio

Review

Mechanism and energetics of proton translocation by the respiratory heme-copper oxidases☆ Mårten Wikström ⁎, Michael I. Verkhovsky Helsinki Bioenergetics Group, Structural Biology and Biophysics Programme, Institute of Biotechnology, University of Helsinki, PB 65 (Viikinkaari 1), FI-00014 University of Helsinki, Finland Received 22 May 2007; received in revised form 24 June 2007; accepted 26 June 2007 Available online 6 July 2007

Abstract Recent time-resolved optical and electrometric experiments have provided a sequence of events for the proton-translocating mechanism of cytochrome c oxidase. These data also set limits for the mechanistic, kinetic, and thermodynamic parameters of the proton pump, which are analysed here in some detail. The analysis yields limit values for the pK of the “pump site”, its modulation during the proton-pumping process, and suggests its identity in the structure. Special emphasis is made on side-reactions that may short-circuit the pump, and the means by which these may be avoided. We will also discuss the most prominent proton pumping mechanisms proposed to date in relation to these data. © 2007 Elsevier B.V. All rights reserved. Keywords: Electron transfer; Proton transfer; Proton pump; Cell respiration

1. Introduction Cell respiration in most aerobic organisms depends on one or several membrane-bound heme-copper oxidases which harness the high oxidising potential of dioxygen as the acceptor of the electrons derived from the oxidation of protein, carbohydrate and fat, thus maximising free energy release for the cells' demands. Despite substantial structural diversity among extreme members of the heme-copper oxidase enzyme family, they all share a basic structure [1–7] comprising two heme groups (hemes a and a3 in the cytochrome c oxidases considered here) within the bacterial or inner mitochondrial membrane, of which heme a3 forms a binuclear O2-reduction site with an adjacent copper ion (CuB). The other heme lies almost at van der Waals' Abbreviations: Cyt. c, cytochrome c; Em, midpoint redox potential at pH 7; eT, electron transfer; K, equilibrium constant; N-side, negatively charged aqueous side of the membrane; pK, -log10 (acid dissociation constant); P-side, positively charged aqueous side of the membrane; pmf, protonmotive force; pT, proton transfer ☆ For simplicity, we adopt the amino acid numbering of subunit I of cytochrome c oxidase from bovine heart mitochondria irrespective of the organism from which the enzyme has been isolated. ⁎ Corresponding author. E-mail address: [email protected] (M. Wikström). 0005-2728/$ - see front matter © 2007 Elsevier B.V. All rights reserved. doi:10.1016/j.bbabio.2007.06.008

contact with heme a3, and serves as an important “electronqueuing” devise [8] that donates one electron at the time to the O2 reduction site. The heme-copper oxidases are remarkable molecular machines that convert much of the free energy released in the reduction of O2 to water into an electrochemical proton gradient across the membrane [8–11]. A proton gradient is the primarily generated form of energy not only in cell respiration, but also in photosynthesis [12], and is thus a fundamental form of energy for virtually all forms of life, which is utilised for various energy-requiring purposes, most prominently for the synthesis of ATP, the energy-currency of the cell. It has been established that the energy-transducing function of the heme-copper oxidases stems from their functioning as redoxdriven proton pumps, and from the “directionality” of the chemistry where the electrons and protons required for O2 reduction to water are taken up from opposite sides of the membrane [8–11]. As will be discussed below, these two functions are intimately connected in the catalytic mechanism. Several proposals for the mechanism of proton translocation by the heme-copper oxidases have been presented (see below) during the 30 years since its discovery in mitochondria [13] and in bacteria [14–17], and this era has seen both colourful disputes and controversy. The problem has rightfully been deemed as one of the greatest challenges in bioenergetics and

M. Wikström, M.I. Verkhovsky / Biochimica et Biophysica Acta 1767 (2007) 1200–1214

molecular biology. Whilst a final conclusion has not been reached, we are today much closer to the solution due to the combined efforts of structural biology, time-resolved spectroscopy and related biophysical techniques, and more recently in computational chemistry. Here, we will attempt to describe the current knowledge, as we see it, spurred by recent experimental findings that have not been thoroughly analysed up till now. Based on time-resolved measurements of electron and charge translocation we have recently presented a scheme [18–20] of how proton translocation may occur in the hemecopper oxidases. This model is essentially one of electrostatic coupling between electron and proton transfer, a key element of which is a protonatable “pump site” (or several such sites), here called X, that is electrostatically coupled to heme a, and the heme a3 and CuB centres of the O2 reduction site. This model differs from proposals where mere oxidoreduction of heme a drives proton translocation [21–23], but is reminiscent of models proposed by Morgan et al. [24], Rich [25], Michel [26], and Siletsky et al. [27]. In these latter models the proton to be pumped across the membrane is initially transferred from the negatively charged aqueous N-side of the membrane to a “pump site” (or “trap site”, Ref. [25]) within the protein. A central element of these proposals is that this proton is expelled from the pump site towards the P-side of the membrane by electrostatic repulsion from uptake of the “chemical” proton into the binuclear site, which completes the oxygen chemistry— a notion first proposed in [24] and [25]. Morgan et al. [24] suggested the pump site to be a dissociated histidine ligand of

1201

CuB (cf. Ref. [1]), but this idea was abandoned when kinetic data revealed an H+/e− stoichiometry of unity for each pumping step in the catalytic cycle [28,29], which corrected a previous view based on equilibrium experiments with mitochondria [30]. Popovic and Stuchebrukhov [31] have proposed a related model, where the pump site is suggested to be the δN of the his291 ligand of CuB (bovine numbering)* . On this basis there seems to be a consensus on some of the major mechanistic principles of the proton pump, despite differences in details (Section 14). However, two proposed mechanisms are distinctly different. Tsukihara et al. [23] base their model on a “H-pathway” of proton transfer, which passes near the formyl group of heme a on its way from the N-to the Pside of the membrane (see Section 14). A recent model proposed by Brzezinski and Larsson [32] is also fundamentally different from those briefly outlined above. Here, initial electron transfer to the binuclear site leads to uptake of the chemical proton to that site from glu-242 at the end of the D-pathway of proton transfer (see below). Proton transfer to the binuclear heme-copper site is thus proposed to be the primary event that triggers a local conformational change (and a shift of pK) that leads to uptake of the pumped proton into the “pump site” (via glu-242). Finally, reprotonation of glu-242 causes relaxation of the conformational change with expulsion of the proton to the positively charged Pside of the membrane. In a later study, Brändén et al. [33] modified this sequence and suggested that proton transfer from glu-242 to the binuclear site, protonation of the “pump site”, and reprotonation of glu-242, in that order, precede electron transfer

Fig. 1. Reaction sequence of the proton-pumping mechanism. State 0 defines the identity of centres of electron and proton transfer. Time constants are based on the experiments in Ref. [20], except for 0 → I, which is based on Ref. [48]. Equilibrium constants (K = final/initial) are based on Ref. [20] and the text (cf. Fig. 3). Approximate pK values for the pump site X are noted (see text). Filled red areas indicate the position of the electron; filled blue areas show the position of protons.

1202

M. Wikström, M.I. Verkhovsky / Biochimica et Biophysica Acta 1767 (2007) 1200–1214

from heme a to the binuclear site. This and other proton pump models will be further discussed below (Section 14). Here, we will analyse our current model of proton translocation based on the experimental data in [20]. A related purely computational analysis was recently published by Olsson et al. [34]. Our treatment yields information on the mechanistic and energetic properties of the pump, such as estimates of the pK values of the “pump site”, and an indication of its identity. 2. The model Fig. 1 describes our model [20] of one of the four protonpumping reaction steps of the catalytic cycle of cytochrome c oxidase (see below). It is mainly based on experimental findings following electron photoinjection into the activated oxidised state of cytochrome c oxidase from P. denitrificans [20], and thus primarily describes the transition in the catalytic cycle from the oxidised state (OH) to the one-electron-reduced state (EH) of the binuclear centre (Fig. 2). We expect that also the three other proton-pumping cycles (Fig. 2) follow the same main principles (but, see Section 13), albeit the final electron (and proton) acceptor in the heme a3/CuB site varies. The numbering of states in Fig. 1 corresponds to those defined previously [20], with the addition of state 0 which is equivalent to state VI from the viewpoint of the proton-pumping cycle.

We emphasise that the rate of the reaction V → VI as deduced from [20] is less reliable than the other rates of the cycle in Fig. 1. This rate was obtained solely as a residue of the membrane potential kinetics after fitting the latter with the rates of the other transitions of the cycle, as observed spectroscopically. Also, in other electron injection experiments (see e.g. Ref. [27]) only two electrogenic phases due to pT have usually been observed (corresponding to II → III and IV → V) after the fast electrogenic phase due to eT (I → II). Consequently, state V, as drawn in Fig. 1, should be viewed with some caution, especially since it is a state where there is transient charge imbalance at the binuclear centre. The scheme in Fig. 1 shows four redox-active cofactors. CuA is the immediate electron acceptor from cytochrome c and lies at the interface between the external aqueous P-side and the membrane [35]. Heme a lies inside the membrane at a distance from this interface that corresponds to 1/3 of the total membrane dielectric [28]; next to it is heme a3, which like its associated CuB centre lies at the same level in the membrane [1–5]. The two latter redox centres constitute the binuclear site of O2 reduction; this site requires O2 binding and electron and proton transfer to it to produce water. Note that according to our current view, H2O is produced during the later stages of the catalytic cycle (Fig. 2). In addition, Fig. 1 shows a “pump site”, X, which has so far not been identified. This is the site where the proton to be pumped across the membrane is initially transferred from the N-side before it is ejected to the P-side, and the pK of which is modulated by the redox reactions. There are not many alternatives to the identity of X as it needs to be a protonatable site that exhibits strong coupling to the redox changes in heme a and the binuclear site. Plausible possibilities based on the X-ray structures [1–5] are the four heme propionates [1], the δ-nitrogen of histidine 291* [31], and the water cluster above the heme groups. Below, we will discuss these possibilities. 3. The pathways of proton uptake

Fig. 2. The catalytic cycle of cytochrome c oxidase. Only the states of the binuclear centre are shown, including heme a3, CuB and the conserved tyrosine244 (tyr-OH, tyr-O− and tyr-O, for the protonated, deprotonated and neutral radical forms, respectively). Red arrows indicate translocation of one proton across the membrane. The green arrow indicates binding of O2 into the reduced binuclear site. The blue arrow denotes the reduction of bound oxygen in the site. Not shown is the PR state, which is formed from state A when the fully reduced enzyme reacts with O2, and includes transfer of the electron from heme a to the binuclear centre. The structure of PR is equivalent to PM, except that the tyr-244 radical is reduced to tyrosinate [90], or the same as state F but with an OH− ligand instead of water on CuB. The scheme shows the activated “pulsed” states OH and EH populated during turnover (see Section 15). Note that we prefer a version of the cycle where tyrosine-244 remains deprotonated until the reductive phase ([110]; Wikström, unpublished).

Proton uptake from the N-side of the membrane takes place via two well-established pathways, the so-called D-and K-channels [9–11]. Involvement of the so-called H-pathway [23] in transfer of the pumped proton will be discussed in Section 14. The logical idea that the D-and K-channels would correspond, respectively, to the uptake pathways of pumped and chemical protons [1] does not hold. Instead, it appears that whilst all four pumped protons are taken up via the D-pathway the two chemical protons taken up during the PM → F and F → OH transitions (Fig. 2) are also transferred via the Dpathway, and that the K-pathway is responsible for uptake of at least one (and probably both, see below) of the chemical protons during OH → R [10,11,36,37]. The cause of this duality is a prominent unsolved issue. pT via the K-pathway can be very fast, of the order of 50 μs [38], and a major question is therefore why this path is not used during the transitions from PM to OH (Fig. 2). The fast rate of pT via the K-pathway was measured for the direction of pT from the binuclear site to the N-side of the membrane, i.e. in the reverse direction compared to forward

M. Wikström, M.I. Verkhovsky / Biochimica et Biophysica Acta 1767 (2007) 1200–1214

function [38]. The rate of pT in the forward direction may be much slower as it may be limited by the rate by which protons from the aqueous phase enter the pathway. Proton uptake is strongly enhanced for the case of the D-pathway due to protoncollecting “antenna” groups around its entrance [37]; the apparent lack of such groups around the orifice of the Kpathway might make proton uptake much slower. However, this explanation does not appear sufficient, because mutational blockade of the D-pathway does not result in significant enzyme activity that would depend on proton uptake via the K-pathway: forward pT in the latter pathway must be possible at a rate of at least ∼ 1000 s− 1 (when Vmax = 2000 electrons × s− 1). A possible explanation is offered below. According to our model (Fig. 1), electron transfer to the binuclear site depends on prior uptake of a proton into the pump site X. The model thus predicts that in the reductive phase of the catalytic cycle (OH → EH → R; Fig. 2) the reduction of the binuclear site should be kinetically impaired by blocking the D-pathway. However, mutating the asp-91 at the entrance of the D-pathway to asparagine still allowed fast reduction of about one half of heme a3 in cytochrome c oxidase from P. denitrificans, although reduction of the other half was severely slowed down [39]. Such fast heme a3 reduction is not consistent with the model, unless there is a “stored proton” in the D-pathway above the entrance that may be recruited for pT to the X site. Glu-242 might be such a proton donor. Moreover, recent computational work [40] has suggested that a water cluster just below glu-242 may be protonated in the ground state. On this basis, the result in [39] may be reinterpreted to suggest that the fast partial reduction of heme a3 in the asp-91/asn mutant enzyme is due to transient protonation of the site X by a proton within the D-pathway. Therefore, proton transfer to the pump site X may indeed be a prerequisite for fast reduction of the binuclear site, in accordance with our model. We should also consider why the K-pathway cannot substitute for the D-pathway in overcoming mutations in the latter. The K-pathway transfers protons to be consumed at the binuclear site, but it is incapable of transferring a proton to X, whereas the D-pathway is capable of both. Electron transfer from heme a to the binuclear site without a proton in site X may be very unfavourable energetically ([41]; see below), and conversely, proton transfer to the binuclear site without an electron in the latter is also thermodynamically unfavourable. Therefore, if protonation of the group X is impaired, turnover will be slow despite an “open” K-pathway. Nevertheless, these notions do not explain the reciprocal case, viz. why uptake of chemical protons via the D-pathway cannot substitute for a blockade of the K-pathway. This is particularly odd because uptake of the chemical protons is known to occur via the D-pathway during the PM → F → OH sequence. We suspect that this may have to do with the different chemistry at the binuclear centre in the four electron-transfer reactions of the catalytic cycle. More specifically, it might be the discrete protonation state of the binuclear site in the OH → EH → R sequence that prevents its protonation by Glu242 via the D-pathway. Such preference could be achieved if tyr-244 is the proton acceptor, if it has not been reprotonated

1203

during the PM → F → OH sequence (Fig. 2), and if its reprotonation is possible only via the K-pathway. Finally, we note that proton uptake via the K-pathway during enzyme reduction is not restricted to the “resting” form of the enzyme (cf. Section 15), but is also essential in the activated “pulsed” enzyme [42]. A possible mechanism of controlling pT via the K-pathway is suggested in Section 15. 4. General thermodynamic parameters The catalytic cycle of cytochrome c oxidase (Fig. 2) may be divided into three types of reactions, (i) the binding of O2 to the reduced catalytic site (green arrow), (ii) splitting of the O\O bond by internal electron and proton transfer (blue arrow), and (iii), four one-electron transfer reactions (white arrows), each of which is coupled to uptake of a chemical proton into the active site, and to pumping (red arrow) of another proton across the membrane [42]. Each of the latter reactions constitutes a proton-pumping cycle which may be considered to be basically the same mechanistically (but see Section 13). The overall driving force for the catalytic cycle is obtained from the redox potential difference between the electron donor (cyt. c; Em 270 mV) and the acceptor (O2/H2O; Em 815 mV). Here, we use 800 mV as the potential of the O2/H2O couple (corresponding to ca. 120 μM O2) because the standard potential (815 mV, Ref. [43]) is defined at a fugacity of 1 atm O2 (ca. 1.2 mM), which is an unrealistic value for biological systems. The O2 concentrations in tissues are generally well below 20 μM, but closer to 120 μM in most experimental conditions. Analogously, since cyt. c usually runs less than 50% reduced in aerobic steady states, we use a potential of 300 mV for the electron donor. This yields a driving force of 500 mV per electron or 46.0 kcal/ mol for the overall four-electron reaction. Dioxygen binding (i) and formation of the PM state (ii) are not coupled to proton pumping (Fig. 2; [44]), and we therefore subtract the corresponding free energies to get the free energy of the proton-pumping steps. Binding of O2 to the reduced binuclear site in the step forming state A is isergonic (∼ 0 kcal/mol; [45,46]). The following reaction, where the O\O bond is broken to form the so-called PM intermediate goes to completion (confidence level ca. 1%), and is thus exergonic by at least 2.8 kcal/mol. Blomberg et al. [47] reported a value of − 3.6 kcal/mol from quantum-chemical calculations, which will be used here. Subtracting this from the total energy available yields 42.4 kcal/mol, which corresponds to an average driving force of 10.6 kcal/mol (460 mV) for each of the four proton pumping reactions (iii) in the cycle. It is interesting that this driving force is very close to the energy of transferring two electric charges against a proton-motive force of ca. 220 mV across the membrane, which shows that the proton pump of cytochrome c oxidase can operate close to thermodynamic equilibrium (see below). 5. Energetics of the pump cycle We will start the discussion from state I (Fig. 1) where an electron has been transferred to the CuA site from cytochrome

1204

M. Wikström, M.I. Verkhovsky / Biochimica et Biophysica Acta 1767 (2007) 1200–1214

c (or from a photoreductant), and return later to the transition between states 0 and I, which turns out to be important in the understanding of the process as a whole. In the next step (I → II) with the apparent time constant (τ) of 10 μs, the electron at CuA equilibrates with heme a so that the occupancy ratio between states II and I is about 70/30 [20]. Thus, the forward and reverse rate constants of electron transfer from CuA to heme a are 7 × 104 s− 1 (or 14 μs) and 3 × 104 s− 1 (or 33 μs), respectively (cf. Ref. [48]). Thermodynamic equilibrium is a reasonable approximation, because the next reaction (II → III) is much slower (τ ∼ 150 μs). Using the midpoint redox potential (Em) of CuA as a reference (250 mV, Ref. [49]), the near equilibrium position means that the Em of heme a is about 270 mV at this stage. For a system containing a single electron, this Em of heme a appears very low at first sight [43], but it is also supported by earlier kinetic data [48,50,51]. The electrometric data showed that the fast heme a reduction is not associated with vectorial proton uptake [20], nor is the electron transfer from CuA to heme a dependent on pH or on heavy water substitution [27,52,53]. We conclude, therefore, that this is the Em value alk of heme a before equilibration with protons (Em ). The Em of heme a is known to be pH-dependent in cases where the binuclear centre is oxidised [22,43,49], but on the fast time scale of the I → II transition proton equilibration has not yet taken place (see below). In states I and II it is evident that there is no reduction of either heme a3 or CuB within experimental error (∼ 1%). Yet, the heme–heme electron transfer occurs in nanoseconds [54,55]. Evidently, the Em values of heme a3 and CuB in these states are therefore ≥ 120 mV lower than the Em of heme a, i.e. ≤ 150 mV, and thus the lack of electron occupancy in the binuclear centre is of thermodynamic origin. This turns out to be a crucial property of the pump mechanism, as discussed below. The 150 μs reaction, step II → III, is of special interest, because it is accompanied by uptake of the first proton as determined electrometrically [20]. As discussed previously [20], this is the proton to be pumped across the membrane, which is first taken up into the “pump site” (contrast Refs. [32,33]), located above the heme groups. The following step, III → IV, is a fast (nanosecond) electron equilibration between the heme groups; uptake of the proton into X has raised the Em of heme a3 to allow this. In a time sequence, it would thus look as if this entire proton uptake were connected with oxidation of heme a. The IV/III equilibrium is approximately 60/40 [20], which implies that now the Em of heme a3 is about 10 mV higher than the Em of heme a. But the Em of heme a has also been raised from the situation before proton uptake, because no significant population of states I or II is observed after the 150 μs phase. The kinetic data do not allow determination of this new Em, but equilibrium redox titrations have shown that the Em of heme a is ca. 390 mV at pH 7 in the P. denitrificans enzyme [49], and we adopt this experimental value for states III and IV. From this value it follows that the Em of heme a3 has been raised to ∼400 mV in these states.

These numbers may now be used to estimate the pKr values of the group X, i.e. the pK of this group when the electron resides in heme a or heme a3, respectively. pK r ¼

pH alk Em  Em þ pH 60

ð1Þ

alk pH where Em for hemes a and a3 were estimated above, and Em is the Em at pH = 7. Based on Eq. (1) the 120 mV rise in the Em of heme a on protonation of X at our “standard” pH = 7 suggests that the pKr of X is about 9.0 in state III, and the 250 mV rise in the Em of heme a3 suggests a pKr of ≥11.2 in state IV. The proton affinity of X is thus much higher when the electron resides in heme a3 (state IV) than when it resides in heme a (state III), and in this quasi-equilibrium the energy gain by the increased affinity for protons is balanced by the raised electron affinity of heme a3. We may also write the following scheme of thermodynamic interactions

where superscripts r and o stand for reduced and oxidised, 0 and H denote unprotonated and protonated states of the group X, and ΔEm is the difference in Em between the hemes. Therefore, pK a3 ¼ pK a þ

0 H DEm  DEm 60

ð2Þ

describes the relationship between the two pK values of X with hemes a and a3 reduced, respectively, and the difference in Em values between the two hemes in the unprotonated and protonated states. The stronger electrostatic interaction energy between heme a3 and X suggests that the pump site is located closer to heme a3 than to heme a (see below). 6. Energy levels The energy drop in II → III where a proton is taken up from a medium with standard pH = 7 to the group X with a pK of 9 is 2.8 kcal/mol, and state IV lies 0.25 kcal/mol (10 mV) lower, as shown in Fig. 3. Our next task is to evaluate the energy level of state V relative to the previous states. The IV → V transition goes to completion within experimental error [20], so the drop in energy is at least 2.8 kcal/mol. This reaction involves electron transfer from heme a3 to CuB and uptake of the chemical proton to a ligand of the latter. Since there is no significant

M. Wikström, M.I. Verkhovsky / Biochimica et Biophysica Acta 1767 (2007) 1200–1214

1205

potential between heme a3 and CuB in state IV (see above). Using the assessed values for x and ΔEm, pKB will attain a value of ca. 13, which reinforces the notion that it is indeed the exergonicity of the uptake of the substrate proton (reaction IV → V) that provides the main driving force in the pump mechanism. It also confirms that the “vectorial chemistry” of electron and proton uptake into the binuclear site is a key property of the proton pump [24,25], as well as being responsible for separating one of the two charges across the membrane per transferred electron. 7. The identity of X

Fig. 3. Energy profile for the reaction sequence in Fig. 1 at zero protonmotive force. The energy levels are indicated in kcal/mol. See the text for details.

electron occupancy at CuB in states III or IV [20], we may conclude that the Em of CuB in these states is ≥ 120 mV lower than that of heme a, i.e. ≤ 270 mV. In contrast, in state V the Em of CuB must be ≥ 120 mV higher than that of heme a3, which must itself be higher than the value of 400 mV in state IV due to the chemical proton taken up. It follows that the Em of CuB is at least 520 mV in state V, and that it has thus risen by more than 250 mV. The final reaction step, V → VI, constitutes release of the proton from X into the external medium. The energy drop of this reaction is related to the pKo value of X, which is the pK in the absence of an electron since state VI corresponds to state 0 with the electronic charge at the binuclear site fully compensated by the chemical proton. If y is the energy drop (in kcal/ mol) at pH 7 for reaction V → VI, it follows that y ¼ ð7  pK o Þ  1:38

ð3Þ

where 1.38 is the conversion factor between pK units and energy (in kcal/mol). We allow 2 per cent protonation of X in equilibrium with P-phase, minimising decoupling of the pump by starting a new reaction cycle with X protonated (see section 9). This yields a reasonable pKo value of 5.3. y is then ∼2.4 kcal/mol, and since the energy level for state VI is known (Fig. 3), the energy drop for IV → V is ∼5.8 kcal/mol. If x is the drop in free energy for IV → V, then x ¼ ðpKB  7Þ  1:38  ðDEm  0:023Þ

ð4Þ

where pKB is the pK of the proton acceptor in the binuclear site in state V, and ΔEm (in millivolts) is the difference in redox

As already pointed out, the protonatable pump site X appears to lie closer to heme a3 than to heme a because of its much stronger interaction with the former. From the above analysis we concluded that the pKo (no electron) of X is about 5.3, and that pKr is 9.0 and 11.2 with the electron at heme a and a3, respectively. The ΔpK values are thus 3.7 and 5.9, respectively for the effect of the electron at heme a and a3 on the proton affinity of X, which yields a ratio of 1.6. For the electrostatic interaction between either heme and the protonatable site X, we may write. DpK ¼

e2 4pee0 rkT

ð5Þ

(see Ref. [56]), where r is the distance between X and heme a or a3, the other symbols having their conventional meaning. Assuming a homogeneous dielectric, the ΔpK ratio of 1.6 should therefore be equivalent to the distance ratio (heme a to X)/(heme a3 to X), where the centre of charge change on oxidoreduction is taken to be the heme iron. We may use this ratio to explore the identity of X, and therefore the heme a propionates and the D-propionate of heme a3 may be excluded as the pump site X because the distance ratio for these is less than 1.1. The distance ratios for the δ-nitrogen of his-291 and for the A-propionate of heme a3 are 2.2 and 1.8, respectively; particularly the latter is evidently a good candidate of being identical to X. At T = 300 K Eq. (5) can be written in the form e ¼ 556=rDpK

ð6Þ

where r is the distance in Å between the heme iron (a or a3) and X. From this equation follows that if either the A-propionate of heme a3 or the δ-nitrogen of his-291 is indeed identical to X, the average dielectric constant ε is about 11–12 in this domain, which must be considered a reasonable number (see Ref. [34]), especially since it only depends on our assessment of the ΔpK values and the actual distances between the hemes and the two candidates for site X. It is important to note that this tentative identity of X by no means excludes a key role of the D-propionate of heme a3 as a transient protonatable residue in the transfer of the pumped proton to X [1,57,58]. The D-propionate of heme a3 resides about 1.5 times closer to heme a than the A-propionate and

1206

M. Wikström, M.I. Verkhovsky / Biochimica et Biophysica Acta 1767 (2007) 1200–1214

therefore reduction of heme a in state II is expected to raise its pK by some 5.6 units (the pK of X was raised by 3.7 units), which would facilitate proton transfer from glu-242 to the X-site via this residue. We note further that proton transfer from glu242 to the D-propionate may require dissociation of the ion pair between the propionate and arg-438, which we have found to occur by a spontaneous thermal fluctuation [59]. 8. Activation energies and transition states Estimates of the activation free energies for the different reactions in the pump cycle (Fig. 1) may be obtained from the observed rates. Here, we must keep in mind that these rates derive specifically from the OH → EH transition (Ref. [20]; Fig. 2), and are different in the other proton-pumping transitions of the catalytic cycle (see 13). For simplicity, we use Eyring transition state theory with a standard prefactor of 6.3 × 1012 s− 1 for the reaction steps II → III, IV → V and V → VI, and arrive at activation energies of, 12.3, 13.3, and 14.0 kcal/mol, respectively. Note here the relative uncertainty of the rate of the last reaction step, V → VI (see Section 2). For the pure electron transfer steps I → II and III → IV we have used Marcus theory [60] with a prefactor determined from the edge-to-edge distance between donor and acceptor [61]. We have added these activation barriers to the scheme of Fig. 3, but we emphasise that except for the eT steps I → II and III → IV, the barriers are likely to be too high as they assume a transmission coefficient of unity, and do not account for frictional forces typical of the dynamics of protein structures [62]. According to work by Maneg et al. [63] the activation free energy for 0 → I, i.e. the electron transfer from cytochrome c to CuA, is 14.3 kcal/mol. To this we need to add ∼ 0.7 kcal/mol to yield a barrier of ∼ 15 kcal/mol, as we are considering a situation where the potential of cyt. c is 300 mV (Section 4). This reaction may thus be one of the slowest steps in the pump cycle under normal turnover conditions, which is not due to slow eT from bound cyt. c to CuA [48], but to the activation energy for binding reduced cyt. c, or more likely, to dissociation of the oxidised form (see below). An important issue is whether or not the protonmotive force across the membrane will affect the activation barriers, in particular those of proton transfer. Proton transfer to the pump site X is expected to be initiated by pT from glu-242 to the Dpropionate of heme a3. The transition state may be envisaged as a state where the proton has been transferred to the first or second water molecule in the array between glu-242 and the D-propionate, as deduced from the analogous case of pT from asp-96 to the Schiff base in bacteriorhodopsin [64], or where the D-propionate has transiently become protonated before further pT to the pump site X. In both cases, the distance of charge separation across the dielectric is very small, which suggests that the protonmotive force will not have a significant effect on the barrier. The same is true for proton transfer from the D-propionate(a3) to either the A-propionate(a3) or the δnitrogen of his-291, both of which would occur essentially parallel to the membrane plane, presumably assisted by water molecules seen in the X-ray structures [3,23,65,66]. The height

of the barrier for pT between glu-242 and the D-propionate will be composed of at least three factors, viz. the energy of organising the intervening water molecules in a Grotthuss array, the energy of dissociating the propionate–arginine pair [59], and the barrier for the pT as such, of which the latter is presumably very small [64]. On the other hand, pT from the Nside of the membrane via the D-pathway to reprotonate glu-242, as well as proton transfer via the K-pathway to the binuclear site, take place across a substantial fraction of the membrane dielectric. In both these cases the nature of the transition state is difficult to predict. However, it seems likely that both pT reactions are rate-limited by uptake of the proton into the entrance of the pathway, in which case there will only be a very small effect of protonmotive force on the activation energy. The barrier for pT from the pump site to the P-side of the membrane is also difficult to assess since the path is unknown, but also here we assume that the transition state may not be significantly affected by the protonmotive force, because the pump site X is already located close to the outside aqueous phase. 9. Leak pathways without protonmotive force A proton pump mechanism must be effectively secured from electron and proton leaks (see also Refs. [11,34,43]). We may now evaluate such potential leaks based on Fig. 3. We will first assess such potential leaks in the situation without protonmotive force, and return to the case with protonmotive force in Section 10. In order to effectively prevent short-circuit, the leak reaction must have a rate that is at least ten times slower than the relevant slowest step in the cycle to achieve an H+/e− stoichiometry of N0.90. Thus, short-circuiting pathways initiated from intermediates I–IV must have barriers that are N1.4 kcal/mol higher than the intrinsic barrier for step IV → V (13.3 kcal/mol), i.e. N14.7 kcal/mol, because the barrier for reversion of state V to state IV is sufficiently high (19.1 kcal/mol) not to allow that reaction during turnover. Analogously, leaks initiated from state V must have barriers higher than 14.0 + 1.4 or 15.4 kcal/mol (but see below). Premature electron transfer to heme a3 in state II, before uptake of the proton to X, is the first example and can give rise to several leak pathways. With the Em values estimated above, such eT would be endergonic by ≥ 2.8 kcal/mol. Subsequent pT from the N-side to site X is allowable since it would produce intermediate IV. However, if premature eT to heme a3 were followed by pT into the binuclear site, the system would obviously become “uncoupled”. Since the barrier for such pT is 13.3 kcal/mol (see step IV → V) the barrier for this leak reaction would be at least 13.3 + 2.8 = 16.1 kcal/mol. Further, if after premature eT to the binuclear site, there is transfer of a proton from the positively charged P-side of the membrane to the pump site X, the system would reach state IV in the main cycle, but with the proton in X derived from the P-side, i.e. no protonpumping. The activation energy for this proton transfer should be equivalent to that for the reaction VI → V in the main cycle (16.4 kcal/mol, see Fig. 3), making the total barrier 2.8 + 16.4 = 19.2 kcal/mol, again preventing such a leak. We

M. Wikström, M.I. Verkhovsky / Biochimica et Biophysica Acta 1767 (2007) 1200–1214

conclude that prevention of premature eT into the binuclear site by making this reaction unfavourable by ≥ 2.8 kcal/mol is an important detail of the mechanism. Premature pT from the N-side into the binuclear site in state II (and I) must also be prevented (see also Section 3 for the role of the K-pathway). The activation energy for this reaction is 13.3 kcal/mol as discussed above, but the reaction is expected to be strongly endergonic, raising the energy level prior to eT (activation energy 2.8 kcal/mol) to an unknown value E. Thus, if E + 2.8 kcal/mol significantly exceeds the barrier of the slowest step in the normal sequence this leak would be prevented. However, premature proton transfer to the binuclear site with the electron residing in heme a may be prevented by an additional kinetic barrier due to the orientation of the water molecules in the cavity above glu-242, which favours pT from glu-242 to X at the expense of pT to the binuclear site [67]. The energetics and the mechanistic basis of this effect are currently being evaluated (V. Kaila et al., in preparation). Preliminary results suggest that the energy barrier against a “wrongly” orientated water chain is considerable. Next, we consider the situation in state II where the model predicts the subsequent step to be uptake of the proton to be pumped from the N-side (primarily from glu-242) into the pump site X. The question arises why transfer of a proton to X could not instead occur from the P-side of the membrane as the pK of X has been raised due to reduction of heme a. Ideally, at this stage the proton exit pathway between the P-side and X should be closed by some kind of a gating mechanism (see e.g. Ref. [11]), but there is no experimental evidence for that at this time. However, pT from glu-242 to X via the D-propionate of heme a3 may be much faster than pT from the P-side into X. We have suggested that the electron at heme a will tend to arrange the water molecules in an array from glu-242 to the D-propionate on a picosecond time scale [67]. Such an arrangement will increase the probability of dissociation of the propionate/ arginine ion pair [59], which is a prerequisite for the pT. Moreover, the hydrogen-bonded water chain from glu-242 to the propionate will impose a partial positive charge on the Dpropionate before the actual pT [67], which will transiently oppose proton transfer into X from the P-side of the membrane. In state V the protonated site X has a lowered pK value (see above) that helps to eject the proton towards the P-side of the membrane in V → VI. Transfer of this proton instead to the opposite side of the membrane would obviously constitute a serious leak. This leak reaction is reminiscent of the backward III → II reaction in the main cycle with a barrier of 15.1 kcal/ mol (Fig. 3), except that heme a is reduced in the latter case. Here we are reminded of the proposal discussed above that pT between site X and the N-side takes place via the D-propionate of heme a3. The pK of this latter residue is expected to be raised considerably (by ∼ 3 pK units, see above) by the electron in heme a, favouring its protonation. It may therefore be argued that the barrier against backward pT from X towards the N-side in state V is higher by some 3 * 1.38 ∼4 kcal/mol relative to the case where heme a is reduced, i.e. of the order of 19 kcal/mol. It is noteworthy that this case also emphasises the risk of leakage if heme a is reduced prematurely in state V (see below).

1207

The final leak to be considered is of special interest. The slowest step of the pump cycle (disregarding 0 → I) is the step V → VI, where the proton in X is ejected. This is reminiscent of the conclusion by Salomonsson et al. [68] that proton ejection from the pump site may normally be the rate-limiting step when electron transfer is not limited by electron input from cyt. c (single turnover experiments; but see Section 13). Clearly, if the next electron were to arrive from cyt. c to heme a faster than state V is dissipated, a situation would arise where the next pump cycle would be started with the “old” proton at X, i.e. a short-circuit of the pump. As already discussed, the activation free energy for electron transfer into the enzyme (step 0 → I) is ∼15 kcal/mol. Interestingly, the barrier corresponding to the rate of dissociation of the product ferricytochrome c, sometimes considered rate-limiting for the entire process, is of a similar magnitude [69]. This barrier is higher than that for step V → VI (14.0 kcal/mol), but does not appear sufficiently high to prevent this short-circuit. However, as mentioned above (Section 2) the rate of this reaction is less certain than those of the others in the scheme of Fig. 1, and may be faster. We also emphasise that the barrier of electron input from cyt. c must not be too high either; if it is higher than the lowest barrier of short-circuit, decoupling of the pump will also take place (see Section 12). We have estimated the pK of X to be ca. 5.3 in the oxidised enzyme, or when reduction of the binuclear site has been charge-compensated (see above). This implies that X would be significantly protonated at thermodynamic equilibrium when pH ≤6, and thus decoupling of the proton pump should be observed in such conditions. Yet, the experimental observation is a virtually constant H+/e− ratio of proton-pumping over a wide pH range [70], albeit measurements well below pH 6 are beset with technical problems. However, it should be borne in mind that nearly all measurements of the H+/e− ratio are done applying multiple turnovers, whereas proton equilibration between X and the aqueous P-phase may only affect the very first turnover. In subsequent turnovers there may not be sufficient time for such proton equilibration, implying that proton transfer from the P-phase to site X (∼7 s−1 with the current pKo) should be significantly slower than the turnover of the enzyme. Our current estimate is that the barrier for this pT is ∼16.4 kcal/mol (Fig. 3), i.e. not far from the estimated ratelimiting barrier of electron input into the enzyme. However, it is also known that the turnover rate increases with lowered pH [69], which might provide the solution for this problem. Our analysis emphasises the delicate balance between rates of allowed and disallowed reactions in the pump mechanism, and on this basis it may be surprising that the mechanism is very robust allowing full stoichiometry of pumping even at greatly decreased rates of electron and proton transfer (e.g. [33]), and at high protonmotive force (Section 10), even though mutations that compromise the pump have been found (see Section 12). This discussion leads to predictions that can be tested experimentally. If, for example, the rate of electron transfer into the enzyme could be made sufficiently fast during turnover conditions, a short-circuit of the proton pump would be expected. Also, the very first turnover after equilibration at

1208

M. Wikström, M.I. Verkhovsky / Biochimica et Biophysica Acta 1767 (2007) 1200–1214

pH b6 would be expected to yield a lowered H+/e− ratio of proton translocation. 10. Leak pathways in the presence of protonmotive force The pumping mechanism is particularly vulnerable to leaks when there is an electrochemical proton gradient across the membrane. Remarkably, such leaks are insignificant in mitochondria catalysing net ATP synthesis (“State 3”) where the electrochemical proton gradient is of the order of 170 mV [71], because the proton pump stoichiometry of cytochrome c oxidase remains close to unity in such conditions [72]. In “State 4” where the gradient is maximal (ca. 220 mV, Ref. [71]), and where there is minimal flux through the respiratory chain, there is no net synthesis of ATP and the rate of generation of electrochemical potential is equal to the rate of its dissipation. In this state the cytochrome oxidase proton pump might “leak”, although this situation is very difficult to assess experimentally. In Fig. 4 we show the energy diagram for the case with a membrane potential (ΔΨ) of 170 mV (the ΔpH term of the pmf is neglected for simplicity), under which conditions the proton pump stoichiometry has been found to be close to unity [72]. Local activation barriers are assumed to be the same as at ΔΨ = 0 (Section 8). Electron transfer to heme a is assumed to be opposed by 1/3 of ΔΨ, proton transfer from the N-side to the binuclear site by 2/3 of ΔΨ, pT from the N-side to X by 80% and release of the X proton to the P-side by 20% of ΔΨ. The latter values are rough estimates from the electrometric data in [20]. In comparison with Fig. 3, it is seen from Fig. 4 that some activation barriers are raised at 170 mV of membrane potential due to the higher energy (lower occupancy) of intermediate states. As anticipated, this puts further restrictions on the rates of short-circuit pathways, the barriers of which must be N 1.4 kcal/mol (for H+/e− N0.9) to 2.8 kcal/ mol (for H+/e− N 0.99) higher than the barrier of the slowest step (∼15.3 kcal/mol, for IV → V). We emphasise again that the scheme of Fig. 1 and the data in [20] are based on the OH → EH transition (Fig. 2), where some of the rates in the pump cycle may be slower than for other transitions in the catalytic cycle (Section 13). Based on this analysis, the proton pump would clearly be prone to short-circuits at the maximal pmf of 220–230 mV [71], which may be the reason why equilibrium experiments with mitochondria at high pmf resulted in too high stoichiometries of the proton pump, and the erroneous conclusion that proton translocation would be restricted to the PM → F and F → OH transitions of the catalytic cycle [30]. 11. The role of glutamic acid 242 Glu-242 lies at the bottom of the D-pathway through which all pumped protons are transferred (but, see Ref. [23] and Section 13). Assuming that the D-pathway functions kinetically as a pure proton conductor, the pump mechanism must assure

Fig. 4. Energy profile for the reaction sequence in Fig. 1 at a membrane potential of 170 mV. See the text for details.

that glu-242 is protonated also at high protonmotive force in order to be an effective proton donor to the pump site X, but also to prevent back-leakage of protons (see Section 9). We propose that this is the reason for the high pK of glu-242, a situation reminiscent of the role of asp-96 in bacteriorhodopsin [73]. Moreover, it is likely that the carboxylic side chain of glu-242 will have to rotate from its crystal structure position at about the middle of the membrane to a donor position about 2/3 of the membrane from the N-side [74,75]. At a protonmotive force (pmf) of 220 mV (5.1 kcal/mol) its pK in the donor position will therefore be lowered by ∼ 2.5 pK units, which makes it imperative that the pK at pmf = 0 is ≥ 10.5–11.5 to assure a protonation state of at least 90% at high pmf. This range is 1–2 units higher than the pK of glu-242 estimated from work by Brzezinski et al. [76], but consistent with recent FTIR data [77]. Interestingly, it is also close to our estimate of the pK of X in state IV (11.2). Our model of proton translocation does not give any other specific mechanistic role for glu-242, and this residue is indeed replaced with full function in some members of the hemecopper oxidases [78], even though its importance in the oxidases where it exists is without question [79,80]. 12. Mutations that decouple the proton pump Using the quinol-oxidising cytochrome bo3 from Escherichia coli, Thomas et al. [81] showed that blocking the input of what is now known as the D-pathway of proton transfer by mutating the equivalent of asp-91 to asparagine leads to decoupling of the proton pump, but also to diminished turnover of the enzyme. This observation was confirmed in cytochrome c oxidase from Rhodobacter sphaeroides [82], and provided the first evidence for the role of the D-pathway in transferring the pumped protons. However, even more compelling evidence was presented with the finding that mutation of conserved asparagines in the D-pathway (asn-98, asn-163) to aspartate

M. Wikström, M.I. Verkhovsky / Biochimica et Biophysica Acta 1767 (2007) 1200–1214

[83–85] likewise leads to decoupling of the pump, but now with unchanged or even increased overall turnover. Based on kinetic experiments, the reason for this short-circuit has been attributed to an electrostatic effect raising the pK of glu-242 [33], despite the fact that the distance from the mutated asparagine domain to glu-242 is ≥ 20Å. We offer an alternative explanation, in which it is assumed that the rate of pT through the D-pathway to glu242 is slowed down in these mutants, albeit not so much as to depress overall activity. We assume further that very fast reprotonation of glu-242 must be assured to uphold the pump mechanism. If pT in the D-pathway is slowed down, the steady state population of the glu-242 anion will increase at the expense of the protonated form. Such an effect will lower the barrier for proton transfer from the outside (e.g. from the site X) back to glu-242 (see Section 9), which will decouple the pump mechanism. In such a case, the slowest step of the protonpumping cycle may be by-passed, which may be the reason for the overall enhancement of steady state activity in one of the asn/asp mutants. Interestingly, Brändén et al. [33] showed that when asp-91 was mutated to asparagine in addition to the mutation of asn-98 to asp, proton-pumping in the steady state was regained even though turnover was decreased to ∼ 20% of wild type enzyme. At first sight this finding seems to contradict our proposal that loss of proton-pumping in the asn-98/asp mutant is due to a partial block of pT within the D-pathway. We suggest that the rate-limiting step of pT is different in these two mutant enzymes. The double mutant is comparable to the wild type enzyme, except that the single aspartic acid is more deeply buried at position 98 within the D-pathway. Consequently, the rate-limiting step is the uptake of protons from the N-side into asp-98. In the asn-98/asp single mutant, where position 91 at the channel entrance is also occupied by an aspartate, the carboxylic group of asp-98 may attain a very high pK value due to the negatively charged asp-91. The pT from asp-98 to glu-242 may therefore be unfavourable. In that case the pump site X will be rapidly protonated by glu-242, but ineffective reprotonation by asp-98 will favour pT back from X to glu-242, followed by transfer of the proton to the binuclear site and decoupling the pump. In contrast, in the double mutant, glu-242 will be readily reprotonated by asp-98, which prevents proton back-leakage. Whilst the asn-98-asp variant was reported not to exhibit any proton-pumping activity in stopped-flow assays of cytochrome oxidase vesicles with reduced cytochrome c as substrate [83,84], some pumping activity was recorded for this mutant using the O2-pulse technique in intact P. denitrificans cells [84]. We have confirmed this observation, and moreover we find a low but significant proton-pumping stoichiometry (∼ 0.2 H+/e−) in vesicles reconstituted with this mutant enzyme as determined by the O2 pulse method, but no pumping using stopped-flow, in agreement with [83,84]; C. Ribacka et al., unpublished]. The O2 pulse method differs from the stoppedflow technique in that in the former the donor, cytochrome c, is kept highly reduced by an excess of reductant, whereas the redox potential of cytochrome c is much higher on the average during a stopped flow experiment where cytochrome c is the ultimate electron donor. These observations suggest that whilst

1209

the asn-98-asp mutation lowers the barrier for a leak pathway (Section 9), this effect may be partially overcome by lowering the barrier for electron input (reaction 0 → I, Fig. 1), accomplished by keeping cytochrome c highly reduced. These results deserve further attention and confirmation, because if correct, they may be an interesting example of the balance between pumping and decoupled pathways of the mechanism, which under normal conditions is robustly controlled, but where failure may be observed with site-specific mutants. 13. The reaction of the fully reduced enzyme with O2 We note that our model (Fig. 1) is incomplete at least in one respect. Consider state R in the catalytic cycle, where the binuclear site is reduced and the electrons are chargecompensated by protons (Fig. 2). Further reduction of heme a in this state is predicted to cause protonation of X. Yet, it is known experimentally that heme a oxidoreduction is only weakly pH-dependent when the binuclear site is fully reduced [86,87]. More importantly, this weak pH-dependence is not exerted from the N-side of the membrane as predicted according to the model in Fig. 1. It is thus clear experimentally that the fully reduced and protonically charge-compensated binuclear site profoundly changes the proton-dependence of heme a and brings its Em to a very low value (ca. 220 mV). Whether this is associated with lowering of the pK value of the X site will have to be assessed by future work. The initial steps in the reaction of the fully reduced enzyme with O2 is another aberration from the scheme of Fig. 1. Here, protonation of X by glu-242 appears to be the first step of the pump mechanism after binding of O2 to the binuclear site, occurring synchronously with electron transfer from heme a to the binuclear site [18], which supports the view that X is not protonated in the fully reduced enzyme. Although the fully reduced state is unlikely to be relevant in physiological conditions, its reaction with O2 is nevertheless associated with proton-pumping [44,88] in a sequence of events that uniquely shows kinetic distinction between electron transfer from heme a to the binuclear site (in the so-called A → PR transition), and proton uptake from the aqueous N-side of the membrane plus release of the pumped proton to the P-side in the subsequent PR → F transition. This sequence is different from Fig. 1 in several respects: First, state III is not observed as an intermediate, and secondly, although the PR state corresponds to state IV, the donor of the proton in X (i.e. glu-242) has not been reprotonated from the N-side of the membrane, which occurs only in the next PR → F transition [44,88]. The PR state also differs from state IV in that the electron acceptor is the covalently bonded tyrosine-244 in the binuclear site [89,90], and not heme a3. Finally, the rates in the reaction sequence A → PR → F, which yields a full pumping cycle, correspond to time constants of ca. 25 and 100 μs. These differences are of importance since they demonstrate that the proton pump mechanism is robust and may operate at velocities much faster than those in Fig. 1, which is based on the OH → EH transition, and even in a different order. The A → PR → F sequence also demonstrates the very high proton transfer rate of the D-

1210

M. Wikström, M.I. Verkhovsky / Biochimica et Biophysica Acta 1767 (2007) 1200–1214

pathway, where two consecutive pT reactions (transfer of one pumped and one chemical proton) occur within the ∼100 μs of the PR → F transition. There certainly are plausible reasons for these differences relative to the events described in Fig. 1 and their rates. As already pointed out, the Em of heme a is exceptionally low in the fully reduced enzyme, but in addition the Em of the electron acceptor (tyr-244) is expected [91] to be much higher than the Em of heme a3 in state IV (400 mV). These differences may explain why the early reaction with O2 shows no measurable intermediate analogous to state III. However, the reason for the very fast proton uptake from the N-side and proton release to the P-side during PR → F, compared to the reactions IV → VI in Fig. 1, is more obscure. One possibility is the known engagement of the D-pathway for uptake of both protons during PR → F, whereas uptake of the chemical proton in OH → E (which is the basis of Fig. 1) takes place via the Kpathway, which may be considerably slower (see Section 3). This might also explain why proton release to the P-side of the membrane is much slower in V → VI than during PR → F. 14. Comparison between models As already briefly outlined (Section 1), some of the currently proposed models for the proton translocation mechanism of cytochrome c oxidase are similar in general outline but differ in key details. The model by Brzezinski and Larsson [32] is an outstanding exception in that it proposes electron transfer to the binuclear site and transfer of the chemical proton from glu-242 to that site to initiate the mechanism. This notion is likely to have be driven by the results from the exceptional reaction of the fully reduced enzyme with O2 (Section 13), where early eT from prereduced heme a to the binuclear site (during the step A → PR; see Fig. 2) is known to occur. However, it was recently shown that this reaction is accompanied by an internal proton transfer [18] that cannot be attributed to pT to the binuclear site, which would yield the F state, and which had been previously overlooked in electrometric experiments due to its low amplitude in comparison with membrane potential formation during the subsequent step, PR → F [44]. In elegant experiments, Faxén et al. [88] showed that both proton uptake from the N-side and release on the P-side occur with kinetics comparable to the PR → F reaction. It was concluded that proton-pumping is not mechanistically coupled to electron transfer, suggestive of a conformational mechanism, which was the basis of proposing a new mechanistic principle of the pump. We propose that this conclusion may be premature because internal proton transfer was not measured in [88], and since such internal pT, presumably from glu-242 to the pump site X, was detected during the prior electron transfer step [18]. Finally, we note that in the Brzezinski and Larsson mechanism [32] the deprotonated glu-242, formed by pT to the binuclear site, is an essential feature that gives rise to conformational changes which initiate proton pumping. This mechanism does not therefore account for proton-pumping during the OH → EH → R reactions, where transfer of the “chemical” protons to the binuclear site takes place via the K-pathway and glu-242 is not involved.

Interestingly, the mechanisms of Brzezinski and Larsson [32], Siletsky et al. [27] and Popovic and Stuchebrukhov [31] all propose eT to the binuclear site as the initial step in the sequence of proton translocation. In [27] and [31] this electron transfer leads to proton transfer to the pump site, primarily from the D-propionate/arg-438 ion pair, which is then reprotonated by glu-242. As these authors point out, along with Tsukihara et al. [23], this ion pair is hardly protonatable as such (but, see Section 7 and Ref. [58]); we would claim that it is hardly deprotonatable either. These three proposals put little or no emphasis on the role of heme a, which is in contrast to the proposals in [21–23,92] as well as to our suggestion here (Fig. 1; cf. 18–20,64,93) of a key role in proton pumping of this completely conserved redox centre. Mathematical modelling of proton translocation indeed suggests that an electron donor site in close proximity to the binuclear centre can greatly enhance the proton-pumping efficiency, especially in the presence of a protonmotive force across the membrane [93]. Our analysis of potential leak reactions (Section 9) also indicates that premature eT to the binuclear site may compromise the mechanism, and experimentally such eT has been shown to be limited by proton uptake [94]. Brändén et al. [33] significantly modified their earlier mechanism ([32]; see Section 1) to suggest that pT to the binuclear site may initiate the sequence of events prior to electron transfer to that site, which occurs only after protonation of the pump site. We have already discussed the notion of pT to the binuclear site at the stage where the electron still resides at heme a and we consider that to be an improbable event both thermodynamically and kinetically (Section 9). We also do not see why, in this mechanism, initial protonation of the binuclear site would not cause eT to that site prior to subsequent events. Tsukihara et al. [23] and most recently Shimokata et al. [95] have forcefully brought forward a model where a “H-pathway” of proton transfer deduced from the crystal structure of the mitochondrial cytochrome c oxidase from bovine heart is uniquely involved in transfer of the pumped proton, whilst the D-and K-pathways are used for transfer of the “chemical” proton to the binuclear centre. The major structural features of the “H-pathway” are conserved in the best studied bacterial enzymes (cytochromes aa3 from P. denitrificans and Rh. sphaeroides and cytochrome bo3 from E. coli), except for aspartic acid 51, which is found only in the enzymes from higher eukaryotes. Extensive site-directed mutations of residues in the H-pathway of bacterial oxidases have dismissed its importance in the function of these enzymes [96,97]. Nevertheless, data with isolated mitoplasts (i.e. mitochondria lacking the outer membrane) from HeLa cells, into which subunit I from the bovine enzyme has been inserted by truly amazing technology [23,95] have been taken to support the importance of the H-pathway in proton translocation. Three mutations in the H-pathway of subunit I, using this hybrid system, were reported to abolish proton-pumping without a significant effect on turnover, in diametrical contrast to the results in the bacterial enzymes [96,97]. Many questions remain unanswered, however. A blockade of a proton-pumping pathway should a priori also block the electron transfer reaction which – by definition – is tightly coupled to proton-pumping.

M. Wikström, M.I. Verkhovsky / Biochimica et Biophysica Acta 1767 (2007) 1200–1214

1211

Although purely “decoupling” mutations have been found in the D-pathway of the bacterial enzymes (Section 12), it is remarkable that all three mutation protocols in the H-pathway of the bovine enzyme [23,95] block proton pumping without effect on turnover. Unfortunately, the authors provide little information on the statistics of their proton pump measurements, making it difficult to scrutinise them. Another difficulty is that the reported rates of proton consumption with or without the uncoupler FCCP show poor and variable respiratory control of the mitoplasts [95]. Muramoto et al. [98] recently reported that Zn2+ (and Cd2+) binds to a site close to his-503, located close to asp-91 at the entrance of the D-pathway (cf. Ref. [99]). The location of Zn2+ was found in conditions comparable to those in experiments with the enzyme from P. denitrificans, where proton pumping was shown to be abolished by Zn2+ with only a modest effect on the turnover [100]. The work by Aagaard and Brzezinski [101] had already suggested that comparable concentrations of Zn2+ block proton uptake via the D-pathway. Thus, the binding of Zn2+ to a site at the entrance of the D-pathway causes an effect very similar to those of the mutations in the D-pathway discussed above (Section 12). In our view, the recent identification of the binding site for Zn2+ in the crystal structure [98,99] provides further strong support for the notion that the Dpathway indeed conducts the pumped protons. For reasons not understood, the authors in [98] question the possibility that proton pumping may be lost by partial blockade of proton uptake via the D-pathway, a scenario specifically discussed above (Section 12). Of the mechanisms discussed, those proposed by Rich et al. [25,92] and Michel [26] are the ones most closely related to our current view. Both emphasise reduction of heme a in initiating the pump sequence, although this eT is coupled to protonation of glu-242 in [92], and proton translocation during O → E is absent in [26], occurring instead spontaneously on decay of the PM state (contrast Ref. [44]).

functions and to form the OH (“pulsed”) state [28,42,104,105], could mean that the nonpolar cavity is devoid of water molecules in state O, or contains a suboptimal number to allow pT. As already discussed (Section 3), an “open” K-pathway may not be capable of efficiently restoring such a deficit because transfer of an uncompensated electron, or a proton, to the binuclear site are each strongly endergonic processes (see also below). Water production on reduction of the binuclear centre (Fig. 2) may restore the situation, enabling protonation of the pump site X via the D-pathway and the water molecules in the cavity, thus lowering the thermodynamic barrier for eT to the binuclear site. Liu and Hill [106] recently showed that formamide blocks electron transfer between heme a and the binuclear site, presumably by replacing water molecules in the structure, in accordance with earlier observations of water dynamics in cytochrome oxidase by Kornblatt and coworkers [107,108]. Build-up of the formamide blockade required several turnovers of the enzyme, which was then stalled essentially in the O state since no P or F intermediates accumulated [106]. In our view, this finding suggests water dynamics in the K-pathway, such that it may be filled with water molecules during the oxidative phase (PM → OH; Fig. 2) to become proton-conductive during the reductive phase (OH → R), after which it is again emptied, except for the very few water molecules seen in the static crystal structures [3,23,62,63]. The trigger for filling the K-pathway with water might be formation of the tyrosinate-244 anion at the bottom of this pathway (Fig. 2). This might be sufficient since Hummer et al. [109] have demonstrated that very small changes in the attraction between the walls of a nanotube and water molecules within the tube can have a dramatic “all or none” effect of filling or emptying the tube. This idea may be tested by molecular dynamics simulations of the K-pathway in the two protonation states of tyr-244.

15. The dynamics of water molecules

The described mechanism may be briefly summarised as follows. Electron transfer into heme a increases the proton affinity (pK) of the pump site X, which can now accept a proton from the N-side of the membrane. The protonation of X, which is much slower than heme a reduction, raises the redox potential of the binuclear site domain, which now attracts the electron and further stabilises the proton at X. Electron transfer to the binuclear site, in turn, increases the proton affinity for that site, which results in uptake of the chemical proton. The latter event decreases the proton affinity of the pump site back to its original low value, which drives the expulsion of the pumped proton towards the P-side of the membrane [24,25]. The mechanism, as described, is one of pure electrostatic coupling, where dissipative leak reactions are avoided in part due to kinetic gating, the necessity of which has been anticipated [31,64,93]. In view of the delicate balance between rates of productive proton-pumping and short-circuit pathways, as analysed here, it is entirely possible that the mechanism requires further gating that has not been identified so far. It is also clear that the described sequence of the pump mechanism (Fig. 1) need not be

Each catalytic cycle produces two water molecules (Fig. 2), which are probably ejected into the nonpolar cavity above glu-242. Isotopically labelled product water is found near the Mg/Mn site above the heme groups within some 6 ms after mixing the enzyme with reduced cyt. c and 17O2 [102]. Molecular dynamics simulations have suggested that water molecules may reach the Mg/Mn site from the cavity via a spontaneous dissociation of the D-propionate(a3)/arginine-438 ion pair [59]. Water molecules are constantly produced into the cavity during activity and expelled from it. The dynamics of this process is not fully understood, but may be of crucial importance, because the water molecules in the cavity are essential for proton transfer from glu-242 to the pump site X, as well as to the oxygen reduction site [31,34,57,59,64,71, 72,103]. The fact that the fully oxidised enzyme, as isolated (“resting” O state), is incapable of fast electron transfer from heme a to the binuclear site and of proton translocation, but requires reduction and reoxidation by O2 to achieve these

16. Conclusion

1212

M. Wikström, M.I. Verkhovsky / Biochimica et Biophysica Acta 1767 (2007) 1200–1214

unique, but that both the sequence and the intrinsic rates may vary in the four proton-pumping steps of the catalytic cycle, even though the basic coupling principles prevail. Finally, we have given evidence for the tentative identity of the pump site X. The A-propionate group of heme a3 resides in a very wellconserved domain of the heme-copper oxidase structures, and histidine-291 is also fully conserved. Acknowledgments MW gratefully acknowledges useful discussions with Margareta Blomberg, Leslie Dutton, Adrian Goldman, Peter Rich, Per Siegbahn, and Alexei Stuchebruchov. This work was supported by grants from the Sigrid Jusélius Foundation, Biocentrum Helsinki, the University of Helsinki, and the Academy of Finland. References [1] S. Iwata, C. Ostermeier, B. Ludwig, H. Michel, Structure at 2.8Å resolution of cytochrome c oxidase from Paracoccus denitrificans, Nature 376 (1995) 660–669. [2] T. Tsukihara, H. Aoyama, E. Yamashita, T. Tomizaki, H. Yamaguchi, K. Shinzawa-Itoh, R. Nakashima, R. Yaono, S. Yoshikawa, The whole structure of the 13-subunit oxidized cytochrome c oxidase at 2.8 Å, Science 272 (1996) 1136–1144. [3] M. Svensson-Ek, J. Abramson, G. Larsson, S. Törnroth, P. Brzezinski, S. Iwata, The X-ray crystal structures of wild-type and EQ(I-286) mutant cytochrome c oxidase from Rhodobacter sphaeroides, J. Mol. Biol. 321 (2002) 329–339. [4] J. Abramson, S. Riistama, G. Larsson, A. Jasaitis, M. Svensson-Ek, L. Laakkonen, A. Puustinen, S. Iwata, M. Wikström, The structure of the ubiquinol oxidase from Escherichia coli and its ubiquinone binding site, Nat. Struct. Biol. 7 (2000) 910–917. [5] T. Soulimane, G. Buse, G.P. Bourenkov, H.D. Bartunik, R. Huber, M.E. Than, Structure and mechanism of the aberrant ba(3) cytochrome c oxidase from Thermus thermophilus, EMBO J. 19 (2000) 1766–1776. [6] J. Hemp, C. Christian, B. Barquera, R.B. Gennis, T.J. Martinez, Helix switching of a key active-site residue in the cytochrome cbb3 oxidases, Biochemistry 44 (2005) 10766–10775. [7] V. Sharma, A. Puustinen, M. Wikström, L. Laakkonen, Sequence analysis of the cbb3 oxidases and an atomic model for the Rhodobacter sphaeroides enzyme, Biochemistry 45 (2006) 5754–5765. [8] G.T. Babcock, M. Wikström, Oxygen activation and the conservation of energy in cell respiration, Nature 356 (1992) 301–309. [9] S. Ferguson-Miller, G.T. Babcock, Heme/copper terminal oxidases, Chem. Rev. 96 (1996) 2889–2907. [10] R.B. Gennis, Coupled proton and electron transfer reactions in cytochrome oxidase, Front. Biosci. 9 (2004) 581–591. [11] P. Brzezinski, Redox-driven membrane-bound proton pumps, TIBS 29 (2004) 380–387. [12] J. Barber, Photosystem II: an enzyme of global significance, Biochem. Soc. Trans. 34 (2006) 619–631. [13] M.K.F. Wikström, Proton pump coupled to cytochrome c oxidase in mitochondria, Nature 266 (1977) 271–273. [14] H.W. van Verseveld, K. Krab, A.H. Stouthamer, Proton pump coupled to cytochrome c oxidase in Paracoccus denitrificans, Biochim. Biophys. Acta 635 (1981) 525–534. [15] M. Solioz, E. Carafoli, B. Ludwig, The cytochrome c oxidase of Paracoccus denitrificans pumps protons in a reconstituted system, J. Biol. Chem. 257 (1982) 1579–1582. [16] A. Puustinen, M. Finel, M. Virkki, M. Wikström, Cytochrome o (bo) is a proton pump in Paracoccus denitrificans and Escherichia coli, FEBS Lett. 249 (1989) 163–167.

[17] M. Lauraeus, M. Wikström, The terminal quinol oxidases of Bacillus subtilis have different energy conservation properties, J. Biol. Chem. 268 (1993) 11470–11473. [18] I. Belevich, M.I. Verkhovsky, M. Wikström, Proton-coupled electron transfer drives the proton pump of cytochrome c oxidase, Nature 440 (2006) 829–832. [19] M.I. Verkhovsky, I. Belevich, D.A. Bloch, M. Wikström, Elementary steps of proton translocation in the catalytic cycle of cytochrome oxidase, Biochim. Biophys. Acta 1157 (2006) 401–407. [20] I. Belevich, D. Bloch, N. Belevich, M. Wikström, M.I. Verkhovsky, Exploring the proton pump mechanism of cytochrome c oxidase in real time, Proc. Natl. Acad. Sci. U. S. A. 104 (2007) 2685–2690. [21] S. Papa, N. Capitanio, G. Capitanio, A cooperative model for proton pumping in cytochrome c oxidase, Biochim. Biophys. Acta 1655 (2004) 353–364. [22] V. Yu. Artzatbanov, A.A. Konstantinov, V.P. Skulachev, Involvement of intramitochondrial protons in redox reactions of cytochrome a, FEBS Lett. 87 (1978) 180–185. [23] T. Tsukihara, K. Shimokata, Y. Katayama, H. Shimada, K. Muramoto, H. Aoyama, M. Mochizuki, K. Shinzawa-Itoh, E. Yamashita, M. Yao, Y. Ishimura, S. Yoshikawa, The low-spin heme of cytochrome c oxidase as the driving element of the proton-pumping process, Proc. Natl. Acad. Sci. U. S. A. 100 (2003) 15304–15309. [24] J.E. Morgan, M.I. Verkhovsky, M. Wikström, The histidine cycle: a new model for proton translocation in the respiratory heme-copper oxidases, J. Bioenerg. Biomembranes 26 (1994) 599–608. [25] P.R. Rich, Towards an understanding of the chemistry of oxygen reduction and proton translocation in the iron-copper respiratory oxidases, Aust. J. Plant Physiol. 22 (1995) 479–486. [26] H. Michel, Cytochrome c oxidase: catalytic cycle and mechanisms of proton pumping—a discussion, Biochemistry 38 (1999) 15129–15140. [27] S.A. Siletsky, A.S. Pawate, K. Weiss, R.B. Gennis, A.A. Konstantinov, Transmembrane charge separation during the ferryl-oxo → oxidized transition in a nonpumping mutant of cytochrome c oxidase, J. Biol. Chem. 279 (2004) 52558–52565. [28] M.I. Verkhovsky, A. Jasaitis, M.L. Verkhovskaya, J.E. Morgan, M. Wikström, Proton translocation by cytochrome c oxidase, Nature 400 (1999) 480–483. [29] D. Bloch, I. Belevich, A. Jasaitis, C. Ribacka, A. Puustinen, M.I. Verkhovsky, M. Wikström, The catalytic cycle of cytochrome c oxidase is not the sum of its two halves, Proc. Natl. Acad. Sci. U. S. A. 101 (2004) 529–533. [30] M. Wikström, Identification of the electron transfers in cytochrome oxidase that are coupled to proton-pumping, Nature 338 (1989) 776–778. [31] D.M. Popovic, A.A. Stuchebrukhov, Proton pumping mechanism and catalytic cycle of cytochrome c oxidase: Coulomb pump model with kinetic gating, FEBS Lett. 566 (2004) 126–130. [32] P. Brzezinski, G. Larsson, Redox-driven proton pumping by heme-copper oxidases, Biochim. Biophys. Acta 1605 (2003) 1–13. [33] G. Brändén, A.S. Pawate, R.B. Gennis, P. Brzezinski, Controlled uncoupling and recoupling of proton pumping in cytochrome c oxidase, Proc. Natl. Acad. Sci. U. S. A. 103 (2006) 317–322. [34] M.H.M. Olsson, P.E.M. Siegbahn, M.R.A. Blomberg, A. Warshel, Exploring pathways and barriers for coupled ET/PT in cytochrome c oxidase: a general framework for examining energetics and mechanistic alternatives, Biochim. Biophys. Acta 1767 (2007) 244–260. [35] P.R. Rich, I.C. West, P. Mitchell, The location of CuA in mammalian cytochrome c oxidase, FEBS Lett. 233 (1988) 25–30. [36] A.A. Konstantinov, S. Siletsky, D. Mitchell, A. Kaulen, R.B. Gennis, The roles of the two proton input channels in cytochrome c oxidase from Rhodobacter sphaeroides probed by the effects of site-directed mutations on time-resolved electrogenic intraprotein proton transfer, Proc. Natl. Acad. Sci. U. S. A. 94 (1997) 9085–9090. [37] M. Karpefors, P. Ädelroth, A. Aagaard, H. Sigurdson, M. Svensson-Ek, P. Brzezinski, Electron–proton interactions in terminal oxidases, Biochim. Biophys. Acta 1365 (1998) 159–169. [38] I. Belevich, A. Tuukkanen, M. Wikström, M.I. Verkhovsky, Proton coupled electron equilibrium in soluble and membrane bound

M. Wikström, M.I. Verkhovsky / Biochimica et Biophysica Acta 1767 (2007) 1200–1214

[39]

[40]

[41]

[42]

[43] [44]

[45]

[46]

[47]

[48]

[49]

[50] [51]

[52]

[53]

[54]

[55]

[56] [57] [58]

[59]

cytochrome c oxidase from Paracoccus denitrificans, Biochemistry 45 (2006) 4000–4006. M. Wikström, A. Jasaitis, C. Backgren, A. Puustinen, M.I. Verkhovsky, The role of the D-and K-pathways of proton transfer in the function of the haem-copper oxidases, Biochim. Biophys. Acta 1459 (2000) 514–520. J. Xu, M.A. Sharpe, L. Qin, S. Ferguson-Miller, G.A. Voth, Storage of an excess proton in the hydrogen-bonded network of the D-pathway of cytochrome c oxidase: identification of a protonated water cluster, J. Am. Chem. Soc. 129 (2007) 2910–2913. M.I. Verkhovsky, J.E. Morgan, M. Wikström, Control of electron delivery to the oxygen reduction site of cytochrome c oxidase: a role for protons, Biochemistry 34 (1995) 7483–7491. D. Bloch, I. Belevich, A. Jasaitis, C. Ribacka, A. Puustinen, M. M.I. Verkhovsky, M. Wikström, The catalytic cycle of cytochrome c oxidase is not the sum of its two halves, Proc. Natl. Acad. Sci. U. S. A. 101 (2004) 529–533. M. Wikström, K. Krab, M. Saraste, Cytochrome Oxidase—A Synthesis, Academic Press, London, 1981. A. Jasaitis, M. Verkhovskaya, J.E. Morgan, M.I. Verkhovsky, M. Wikström, Assignment and charge translocation stoichiometries of the major electrogenic phases in the reaction of cytochrome c oxidase with dioxygen, Biochemistry 38 (1999) 2697–2706. B. Chance, C. Saronio, J.S. Leigh Jr., Functional intermediates in the reaction of membrane-bound cytochrome oxidase with oxygen, J. Biol. Chem. 250 (1975) 9226–9237. M.I. Verkhovsky, J.E. Morgan, M. Wikström, Oxygen binding and activation: early steps in the reaction of oxygen with cytochrome c oxidase, Biochemistry 33 (1994) 3079–3086. M.R.A. Blomberg, P.E.M. Siegbahn, G.T. Babcock, M. Wikström, Modeling cytochrome c oxidase—a quantum chemical study of the O\O bond cleavage mechanism, J. Am. Chem. Soc. 122 (2000) 12848–12858. K. Wang, L. Geren, Y. Zhen, L. Ma, S. Ferguson-Miller, B. Durham, F. Millett, Mutants of the CuA site in cytochrome c oxidase of Rhodobacter sphaeroides: II. Rapid kinetic analysis of electron transfer, Biochemistry 41 (2002) 2298–2304. E.A. Gorbikova, K. Vuorilehto, M. Wikström, M.I. Verkhovsky, Redox titrations of all electron carriers of cytochrome c oxidase by FTIR spectroscopy, Biochemistry 45 (2006) 5641–5649. O. Farver, O. Einarsdóttir, I. Pecht, Electron transfer rates and equilibrium within cytochrome c oxidase, Eur. J. Biochem. 267 (2000) 950–954. O. Farver, E. Grell, B. Ludwig, H. Michel, I. Pecht, Rates and equilibrium of CuA to heme a electron transfer in Paracoccus denitrificans cytochrome c oxidase, Biophys. J. 90 (2006) 2131–2137. M. Oliveberg, P. Brzezinski, B.G. Malmström, The effect of pH and temperature on the reaction of fully reduced and mixed valence cytochrome c oxidase with dioxygen, Biochim. Biophys. Acta 977 (1989) 322–328. M. Ruitenberg, A. Kannt, E. Bamberg, B. Ludwig, H. Michel, K. Fendler, Single-electron reduction of the oxidized state is coupled to proton uptake via the K pathway in Paracoccus denitrificans cytochrome c oxidase, Proc. Natl. Acad. Sci. U. S. A. 97 (2000) 4632–4636. M.I. Verkhovsky, A. Jasaitis, M. Wikström, Ultrafast haem–haem electron transfer in cytochrome c oxidase, Biochim. Biophys. Acta 1506 (2001) 143–146. E. Pilet, A. Jasaitis, U. Liebl, M.H. Vos, Electron transfer between hemes in mammalian cytochrome c oxidase, Proc. Natl. Acad. Sci. U. S. A. 101 (2004) 16198–16203. R.P. Bell, The Proton in Chemistry, Chapman and Hall, London, 1973. A. Puustinen, M. Wikström, Proton exit from the heme–copper oxidase of coli, Proc. Natl. Acad. Sci. U. S. A. 96 (1999) 35–37. G. Brändén, M. Brändén, B. Schmidt, D.A. Mills, S. Ferguson-Miller, P. Brzezinski, The protonation state of a heme propionate controls electron transfer in cytochrome c oxidase, Biochemistry 44 (2005) 10466–10474. M. Wikström, C. Ribacka, M. Molin, L. Laakkonen, M.I. Verkhovsky, A. Puustinen, Gating of water and proton transfer in the respiratory enzyme cytochrome c oxidase, Proc. Natl. Acad. Sci. U. S. A. 102 (2005) 10478–10481.

1213

[60] H.B. Gray, J.R. Winkler, Electron tunneling through proteins, Q. Rev. Biophys. 36 (2003) 341–372. [61] C.C. Page, C.C. Moser, P.L. Dutton, Natural engineering principles of electron tunnelling in biological oxidation–reduction, Nature 402 (1999) 47–52. [62] H. Frauenfelder, P.G. Wolynes, R.H. Austin, Biological Physics, Rev. Mod. Phys. 71 (1999) 419–430. [63] O. Maneg, B. Ludwig, F. Malatesta, Different interaction modes of two cytochrome-c oxidase soluble CuA fragments with their substrates, J. Biol. Chem. 278 (2003) 46734–46740. [64] Y.-S. Lee, M. Krauss, Dynamics of proton transfer in bacteriorhodopsin, J. Am. Chem Soc. 126 (2004) 2225–2230. [65] C. Ostermeier, A. Harrenga, U. Ermler, H. Michel, Structure at 2.7 Å resolution of the Paracoccus denitrificans two-subunit cytochrome c oxidase complexed with an antibody Fv fragment, Proc. Natl. Acad. Sci. U. S. A. 94 (1997) 10547–10553. [66] L. Qin, C. Hiser, A. Mulichak, R.M. Garavito, S. Ferguson-Miller, Identification of conserved lipid/detergent-binding sites in a highresolution structure of the membrane protein cytochrome c oxidase, Proc. Natl. Acad. Sci. U. S. A. 103 (2006) 16117–16122. [67] M. Wikström, M.I. Verkhovsky, G. Hummer, Water-gated mechanism of proton translocation by cytochrome c oxidase, Biochim. Biophys. Acta 1604 (2003) 61–65. [68] L. Salomonsson, K. Faxén, P. Ädelroth, P. Brzezinski, The timing of proton migration in membrane-reconstituted cytochrome c oxidase, Proc. Natl. Acad. Sci. U. S. A. 102 (2005) 17624–17629. [69] B.G. Malmström, L.-E. Andréasson, The steady state rate equation for cytochrome c oxidase based on a minimal kinetic scheme, J. Inorg. Biochem. 23 (1985) 233–242. [70] K. Krab, M. Wikström, On the stoichiometry and thermodynamics of proton-pumping cytochrome c oxidase in mitochondria, Biochim. Biophys. Acta 548 (1979) 1–15. [71] D. Nicholls, The influence of respiration and ATP hydrolysis on the proton-electrochemical gradient across the inner membrane of rat-liver mitochondria as determined by ion distribution, Eur. J. Biochem. 50 (1974) 305–315. [72] M. Wikström, M.I. Verkhovsky, Proton translocation by cytochrome c oxidase in different phases of the catalytic cycle, Biochim. Biophys. Acta 1555 (2002) 128–132. [73] M. Wikström, Proton translocation by bacteriorhodopsin and hemecopper oxidases, Curr. Opin. Struct. Biol. 8 (1998) 480–488. [74] I. Hofacker, K. Schulten, Oxygen and proton pathways in cytochrome c oxidase, Proteins: Struct., Funct., Bioinf. 30 (1998) 100–107. [75] R. Pomès, G. Hummer, M. Wikström, Structure and dynamics of a proton shuttle in cytochrome c oxidase, Biochim. Biophys. Acta 1365 (1998) 255–260. [76] A. Namslauer, A. Aagaard, A. Katsonouri, P. Brzezinski, Intramolecular proton-transfer reactions in a membrane-bound proton pump: the effect of pH on the peroxy to ferryl transition in cytochrome c oxidase, Biochemistry 42 (2003) 1488–1498. [77] E.A. Gorbikova, N.P. Belevich, M. Wikström, M.I. Verkhovsky, Protolytic reactions on reduction of cytochrome c oxidase studied by ATR-FTIR spectroscopy, Biochemistry 46 (2007) 4177–4183. [78] M.M. Pereira, M.L. Verkhovskaya, M. Texeira, M.I. Verkhovsky, The caa3 terminal oxidase of Rhodothermus marinus lacking the key glutamate of the D-channel is a proton pump, Biochemistry 39 (2000) 6336–6340. [79] M.L Verkhovskaya, A. Garcìa-Horsman, A. Puustinen, J.-L. Rigaud, J.E. Morgan, M.I. Verkhovsky, M. Wikström, Glutamic acid 286 in subunit I of cytochrome bo3 is involved in proton translocation, Proc. Natl. Acad. Sci. U. S. A. 94 (1997) 10128–10131. [80] P. Ädelroth, M. Svensson-Ek, D.M. Mitchell, R.B. Gennis, P. Brzezinski, Glutamate 286 in cytochrome aa3 from Rhodobacter sphaeroides is involved in proton uptake during the reaction of the fully reduced enzyme with dioxygen, Biochemistry 36 (1997) 13824–13829. [81] J.W. Thomas, A. Puustinen, J.O. Alben, R.B. Gennis, M. Wikström, Substitution of asparagine for aspartate-135 in subunit I of the cytochrome bo ubiquinol oxidase of Escherichia coli eliminates protonpumping activity, Biochemistry 32 (1993) 10923–10928.

1214

M. Wikström, M.I. Verkhovsky / Biochimica et Biophysica Acta 1767 (2007) 1200–1214

[82] J.R. Fetter, J. Qian, J. Shapley, J.W. Thomas, A. Garcia-Horsman, E. Schmidt, J. Hosler, G.T. Babcock, R.B. Gennis, S. Ferguson-Miller, Possible proton relay pathways in cytochrome c oxidase, Proc. Natl. Acad. Sci. U. S. A. 92 (1995) 1604–1608. [83] U. Pfitzner, K. Hoffmeier, A. Harrenga, A. Kannt, H. Michel, E. Bamberg, O.M. Richter, B. Ludwig, Tracing the D-pathway in reconstituted sitedirected mutants of cytochrome c oxidase from Paracoccus denitrificans, Biochemistry 39 (2000) 6756–6762. [84] A.S. Pawate, J. Morgan, A. Namslauer, D. Mills, P. Brzezinski, S. Ferguson-Miller, R.B. Gennis, A mutation in subunit I of cytochrome oxidase from Rhodobacter sphaeroides results in an increase in steadystate activity but completely eliminates proton pumping, Biochemistry 41 (2002) 13417–13423. [85] D. Han, A. Namslauer, A. Pawate, J.E. Morgan, S. Nagy, A.S. Vakkasoglu, P. Brzezinski, R.B. Gennis, Replacing asn207 by aspartate at the neck of the D channel in the aa3 type cytochrome c oxidase from Rhodobacter sphaeroides results in decoupling the proton pump, Biochemistry 45 (2006) 14064–14074. [86] M.I. Verkhovsky, N. Belevich, J.E. Morgan, M. Wikström, Proton linkage of cytochrome a oxidoreduction in carbon monoxide-treated cytochrome c oxidase, Biochim. Biophys. Acta 1412 (1999) 184–189. [87] R. Mitchell, The nature and significance of the pH dependence of electron equilibration in the cytochrome c- cytochrome c oxidase system, PhD Thesis, University of London (1991). [88] K. Faxén, G. Gilderson, P. Ädelroth, P. Brzezinski, A mechanistic principle for proton pumping by cytochrome c oxidase, Nature 437 (2005) 286–289. [89] G.T. Babcock, How oxygen is activated and reduced in respiration, Proc. Natl. Acad. Sci. U. S. A. 96 (1999) 12971–12973. [90] J.E. Morgan, M.I. Verkhovsky, G. Palmer, M. Wikström, The role of the PR intermediate in the reaction of cytochrome c oxidase with O2, Biochemistry 40 (2001) 6882–6892. [91] C. Tommos, G.T. Babcock, Proton and hydrogen currents in photosynthetic water oxidation, Biochim. Biophys. Acta 1458 (2000) 199–219. [92] B. Meunier, C. Ortwein, U. Brandt, P.R. Rich, Effect of mutation of residue I67 on redox-linked protonation processes in yeast cytochrome c oxidase, Biochem. J. 330 (1998) 1197–1200. [93] Y.C. Kim, M. Wikström, G. Hummer, Kinetic models of redox-coupled proton pumping, Proc. Natl. Acad. Sci. U. S. A. 104 (2007) 2169–2174. [94] M.I. Verkhovsky, J.E. Morgan, M. Wikström, Control of electron delivery to the oxygen reduction site of cytochrome c oxidase: a role for protons, Biochemistry 34 (1995) 7483–7491. [95] K. Shimokata, Y. Katayama, H. Murayama, M. Suematsu, T. Tsukihara, K. Muramoto, H. Aoyama, S. Yoshikawa, H. Shimada, The proton pumping pathway of bovine heart cytochrome c oxidase, Proc. Natl. Acad. Sci. U. S. A. 104 (2007) 4200–4205. [96] U. Pfitzer, A. Odenwald, T. Ostermann, L. Weingard, B. Ludwig, O.-M.

[97]

[98]

[99]

[100]

[101] [102]

[103]

[104]

[105] [106] [107] [108] [109] [110]

H. Richter, Cytochrome c oxidase (heme aa3) from Paracoccus denitrificans: analysis of mutations in putative proton channels of subunit I, J. Bioenerg. Biomembranes 30 (1998) 89–97. H.-M. Lee, T.K. Das, D.L. Rousseau, D. Mills, S. Ferguson-Miller, R.B. Gennis, Mutations in the putative H-channel in the cytochrome c oxidase from Rhodobacter sphaeroides show that this channel is not important for proton conduction but reveal modulation of the properties of heme a, Biochemistry 39 (2000) 2989–2996. K. Muramoto, K. Hirata, K. Shinzawa-Itoh, S. Yoko-o, E. Yamashita, H. Aoyama, K. Tsukihara, S. Yoshikawa, A histidine residue acting as a controlling site for dioxygen reduction and proton pumping by cytochrome c oxidase, Proc. Natl. Acad. Sci. U. S. A. 104 (2007) 7881–7886. L. Qin, D.A. Mills, C. Hiser, A. Murphree, R.M. Garavito, S. FergusonMiller, J. Hosler, Crystallographic location and mutational analysis of Zn and Cd inhibitory sites and role of lipidic carboxylates in rescuing proton path mutants in cytochrome c oxidase, Biochemistry 46 (2007) 6239–6248. A. Kannt, T. Ostermann, H. Müller, M. Ruitenberg, Zn2+ binding to the cytoplasmic side of Paracoccus denitrificans cytochrome c oxidase selectively uncouples electron transfer and proton translocation, FEBS Lett. 503 (2001) 142–146. A. Aagaard, P. Brzezinski, Zinc ions inhibit oxidation of cytochrome c oxidase by oxygen, FEBS Lett. 494 (2001) 157–160. B. Schmidt, J. McCracken, S. Ferguson-Miller, A discrete water exit pathway in the membrane protein cytochrome c oxidase, Proc. Natl. Acad. Sci. U. S. A. 100 (2003) 15539–15542. S.A. Seibold, D.A. Mills, S. Ferguson-Miller, R.I. Cukier, Water chain formation and possible proton pumping routes in Rhodobacter sphaeroides cytochrome c oxidase: A molecular dynamics comparison of the wild type and R481K mutant, Biochemistry 44 (2005) 10475–10485. E. Antonini, M. Brunori, A. Colosimo, C. Greenwood, M.T. Wilson, Oxygen “pulsed” cytochrome c oxidase: Functional properties and catalytic relevance, Proc. Natl. Acad. Sci. U. S. A. 74 (1977) 3128–3132. M. Brunori, G. Antonini, F. Malatesta, P. Sarti, M.T. Wilson, Cytochrome-c oxidase, Eur. J. Biochem. 169 (1987) 1–8. Y. Liu, B.C. Hill, Formamide probes a role for water in the catalytic cycle of cytochrome c oxidase, Biochim. Biophys. Acta 1767 (2007) 45–55. J.A. Kornblatt, G.H. Hoa, A nontraditional role for water in the cytochrome c oxidase reaction, Biochemistry 29 (1990) 9370–9376. J.A. Kornblatt, The water channel of cytochrome c oxidase: inferences from inhibitor studies, Biophys. J. 75 (1998) 3127–3134. G. Hummer, J.C. Rasaiah, J.P. Noworyta, Water conduction through the hydrophobic channel of a carbon nanotube, Nature 414 (2001) 188–190. M. Iwaki, A. Puustinen, M. Wikström, P.R. Rich, Structural and chemical changes of the PM intermediate of Paracoccus denitrificans cytochrome c oxidase revealed by IR spectroscopy with labelled tyrosines and histidine, Biochemistry 45 (2006) 10873–10885.