Definition of the Interaction Domain for Cytochrome con Cytochrome c Oxidase

The reaction between cytochrome c(Cc) and Rhodobacter sphaeroides cytochrome coxidase (CcO) was studied using a cytochrome c derivative labeled with ruthenium trisbipyridine at lysine 55 (Ru-55-Cc). Flash photolysis of a 1:1 complex between Ru-55-Cc and CcO at low ionic strength results in electron transfer from photoreduced heme c to CuA with an intracomplex rate constant ofk a = 4 × 104 s−1, followed by electron transfer from CuA to heme a with a rate constant of k b = 9 × 104s−1. The effects of CcO surface mutations on the kinetics follow the order D214N > E157Q > E148Q > D195N > D151N/E152Q ≈ D188N/E189Q ≈ wild type, indicating that the acidic residues Asp214, Glu157, Glu148, and Asp195 on subunit II interact electrostatically with the lysines surrounding the heme crevice of Cc. Mutating the highly conserved tryptophan residue, Trp143, to Phe or Ala decreased the intracomplex electron transfer rate constant k a by 450- and 1200-fold, respectively, without affecting the dissociation constant K D . It therefore appears that the indole ring of Trp143 mediates electron transfer from the heme group of Cc to CuA. These results are consistent with steady-state kinetic results (Zhen, Y., Hoganson, C. W., Babcock, G. T., and Ferguson-Miller, S. (1999) J. Biol. Chem. 274, 38032–38041) and a computational docking analysis (Roberts, V. A., and Pique, M. E. (1999) J. Biol. Chem.274, 38051–38060).

termediate ionic strength, and was strongly inhibited at high ionic strength, indicating a significant electrostatic interaction between the two proteins (1)(2)(3)(4)(5)(6). Extensive chemical modification studies have shown that six or seven highly conserved lysine amino groups surrounding the heme crevice of cytochrome c are involved in the electrostatic complex with cytochrome oxidase (7)(8)(9)(10)(11)(12)(13). An early carbodiimide modification study provided evidence that four carboxylate groups on subunit II of cytochrome oxidase might be involved in the electrostatic interaction with cytochrome c (14). Three of these carboxylates are located in an unusual sequence (positions 104 -115) consisting of alternating acidic and aromatic residues that were proposed to be involved in electron transfer to Cu A and/or heme a (14). Recent site-directed mutagenesis studies have also implicated acidic residues on subunit II in the interaction (15)(16)(17)(18).
While steady-state kinetics studies have provided a great deal of insight into the reaction between Cc and CcO (1-6), it has not been possible to measure the rate constants of individual steps in the overall mechanism. A new ruthenium photoreduction technique has allowed the resolution of a key step in the mechanism, intracomplex electron transfer from Cc to the initial acceptor in CcO (19,20). In this technique, the 1:1 electrostatic complex between CcO and a ruthenium-Cc derivative labeled with ruthenium trisbipyridine is formed at low ionic strength. The Ru(II) group is photoexcited to a metal-toligand charge transfer state, Ru(II*), which is a strong reducing agent and rapidly transfers an electron to heme c (21). The photoreduced heme c in yeast Ru-39-Cc transfers an electron to Cu A in beef CcO with a rate constant of 6 ϫ 10 4 s Ϫ1 , followed by electron transfer from Cu A to heme a with a rate constant of 2 ϫ 10 4 s Ϫ1 (20). Cu A was thus identified to be the electron entry site in cytochrome oxidase, in agreement with a previous study using a CO photodissociation technique (22). The x-ray crystal structures of bovine (23) and Paracoccus denitrificans (24) CcO have revealed a prominent cluster of acidic residues on subunit II near the binuclear Cu A center. The unusual sequence of acidic and aromatic residues terminates with Trp 143 , which is in van der Waals contact with the Cu A center and could form an electron transfer pathway from the surface of subunit II to Cu A . In the present study, the involvement of acidic residues on subunit II of R. sphaeroides CcO in the interaction with Cc has been tested by mutating them to neutral Asn or Gln residues, while the role of Trp 143 has been studied by mutating it to Phe or Ala (25). The electron transfer reaction between Cc and R. sphaeroides CcO was measured using the derivative Ru-55-Cc, which contains a ruthenium trisbipyridine complex covalently attached to lysine 55 on the bottom of horse cytochrome c. The ruthenium pho-toreduction technique allows measurement of the rate constant for intracomplex electron transfer between Cc and Cu A , as well as the rate constants for complex formation and dissociation as a function of ionic strength. The equilibrium dissociation constant of the complex was also measured by an ultracentrifuge technique.

EXPERIMENTAL PROCEDURES
Materials-Horse heart cytochrome c (type VI) and lauryl maltoside were obtained from Sigma. The Ru-55-Cc derivative was prepared as described by Liu et al. (26). The R. sphaeroides cytochrome oxidase wild type and mutants were prepared as described by Zhen et al. in the accompanying paper (25).
Flash Photolysis Experiments-Transient absorbance measurements were carried out as described by Geren et al. (20) by flash photolysis of 300-l solutions contained in a 1-cm glass semimicrocuvette. The excitation pulse was provided by a Phase R model DL1400 flash lamppumped dye laser using coumarin 490 to produce a 480-nm light flash of Ͻ0.5 s duration. The reaction of cytochrome c was monitored at 550 nm using an extinction coefficient of ⌬⑀ 550 ϭ 18.5 mM Ϫ1 cm Ϫ1 (27). The reduction of heme a was monitored at 605 nm using ⌬⑀ 605 ϭ 16 mM Ϫ1 cm Ϫ1 and at 444 nm using ⌬⑀ 444 ϭ 59 mM Ϫ1 cm Ϫ1 (28). The reaction of Cu A was monitored at 830 nm using ⌬⑀ 830 ϭ 2.0 mM Ϫ1 cm Ϫ1 (29). The extinction coefficients for the heme in Ru-55-Cc at 605 and 830 nm were ⌬⑀ 605 ϭ 1.2 mM Ϫ1 cm Ϫ1 and ⌬⑀ 830 ϭ 0.15 mM Ϫ1 cm Ϫ1 . The transient absorbance changes for heme a and Cu A at 605 nm and 830 nm, respectively, were corrected for the small contribution from heme c at these wavelengths. Reaction solutions typically contained 3-10 M ruthenium-Cc, 5-20 M CcO, 10 mM aniline, 1 mM 3CP in 5 mM Tris-Cl, pH 8.0, at 22°C. The aniline and 3CP functioned as sacrificial electron donors to reduce Ru(III) and prevent the back reaction k 2 shown in Scheme 1. The ionic strength was adjusted by adding sodium chloride. The transients were fitted to appropriate theoretical equations as described by Geren et al. (20), and the reported errors are the estimated S.D. values.
Ultracentrifuge Experiments-The equilibrium dissociation constant of the high affinity complex between horse cytochrome c and R. sphaeroides CcO was measured using the Beckman XL-A analytical ultracentrifuge. Samples typically contained 5 M horse Cc, 6 M CcO in 5 mM Tris-Cl, pH 8, 0.1% lauryl maltoside, and 0 -250 mM NaCl. The ultracentrifuge was run in the sedimentation velocity mode at a speed of 48,000 rpm. Absorbance scans were recorded every 15 min at a wavelength of 410 nm. Free Cc sedimented at a sedimentation velocity of 1.74 S, while CcO sedimented at 9.7 S. The concentration of free Cc in the sample was obtained from the absorbance in the free Cc plateau region, using an extinction coefficient of 106 mM Ϫ1 cm Ϫ1 . The concentration of the Cc⅐CcO complex was calculated from the difference between total Cc and free Cc.

RESULTS
The Reaction of Ru-55-Cc with Wild-type R. sphaeroides CcO-The electron transfer reaction between Cc and R. sphaeroides CcO was studied using the ruthenium photoreduction technique previously described (19,20). Ru-55-Cc was chosen because it is labeled with ruthenium trisbipyridine at lysine 55 on the bottom of Cc, remote from the binding domain, and should not significantly affect the interaction with CcO (see below). Laser flash photolysis of Ru-55-Cc resulted in rapid electron transfer from Ru(II*) to heme c Fe(III) with a rate constant of 4 ϫ 10 5 s Ϫ1 (26), as shown in Scheme I. The yield of heme c photoreduced by a single laser flash was approximately 2%. Flash photolysis of a 1:1 complex between Ru-55-Cc and CcO at low ionic strength led to reoxidation of photoreduced heme c Fe(II) with a rate constant of (4 Ϯ 0.6) ϫ 10 4 s Ϫ1 , as indicated by the 550 nm absorbance transient (Fig. 1). The 605-nm absorbance transient indicated that heme a was reduced with approximately the same rate constant, (4 Ϯ 0.6) ϫ 10 4 s Ϫ1 (Fig. 1). The 830-nm transient showed that Cu A was reduced with approximately the same rate constant as the reduction of heme a. However, the 830-nm transient was not resolved in the 0 -25-s time scale due to interference from the ruthenium luminescence, so the rate constant could not be determined very accurately (data not shown). The total amount of Cu A and heme a reduced was the same as the amount of heme c reoxidized, indicating that there was no direct reduction of Cu A or heme a by Ru(II*). These results are consistent with a mechanism involving initial electron transfer from Ru-55-Cc heme c to either Cu A or heme a, followed by rapid electron transfer equilibrium between Cu A and heme a. Although the present kinetic results cannot distinguish whether Cu A or heme a is the initial electron acceptor, Cu A is assumed to be the initial acceptor by analogy with beef CcO, where that assignment was made definitively using yeast Ru-39-Cc (20). Yeast Cc reacts very poorly with R. sphaeroides CcO (data not shown), so yeast Ru-39-Cc could not be used in the present studies.
The kinetic results are consistent with electron transfer according to the mechanism shown in Scheme 2.
The kinetics of the reaction at 5 mM ionic strength were independent of protein concentration above 1 M, provided that the concentration of CcO was equal to or greater than that of Ru-55-Cc. This indicates that the reaction is due to electron transfer within a 1:1 complex with an equilibrium dissociation constant K D ϭ k d /k f less than 1 M (e.g. the top line of Scheme 2). The transients at 550, 605, and 830 nm were fitted to the kinetic equations for the top line of Scheme 2 with k a ϭ 4 ϫ 10 4 s Ϫ1 and (k b ϩ k c ) Ͼ 10 5 s Ϫ1 . Kinetic simulations revealed that (k b ϩ k c ) has to be larger than 10 5 s Ϫ1 to explain the similarity of the rate constants for reoxidation of cytochrome c 2ϩ and reduction of Cu A and heme a 3ϩ . The relative amounts of reduced heme a 2ϩ and Cu A 1ϩ formed upon completion of the transients after 1 ms were in a ratio of 6.1:1, indicating that the equilibrium constant for electron transfer between Cu A and heme a was K ϭ k b /k c ϭ 6.1. The rate constant for reoxidation of heme a in one-electron reduced CcO is very slow, approximately 0.3 s Ϫ1 , consistent with previous observations using stopped-flow spectroscopy (30).
The kinetics observed for a solution containing 10 M Ru-55-Cc and 15 M CcO remained independent of ionic strength from 5 to 40 mM, and then at higher ionic strength the amplitude of the fast intracomplex phase decreased, and a second, slow phase appeared due to the reaction of uncomplexed Ru-55-Cc 2ϩ with CcO according to Scheme 2 ( Fig. 2). At 55 mM ionic strength, the rate constant of the slow phase was k s ϭ 6900 s Ϫ1 , and its relative amplitude was 47%, while at 75 mM ionic strength, the rate constant of the slow phase was 7900 s Ϫ1 , and its relative amplitude was 68%. The relative amplitudes of the two phases were used to estimate the equilibrium dissociation constant K D for the high affinity 1:1 complex to be 7 M at 55 mM ionic strength and 27 M at 75 mM ionic strength. The rate constant k a of the fast phase remained constant at (4 Ϯ 1) ϫ 10 4 s Ϫ1 from 5 to 80 mM ionic strength. As the ionic strength was increased above 90 mM, the fast phase disappeared entirely, and the rate constant of the slow phase decreased (Fig. 2). The rate constant of the slow phase was the same for the 550-, 830-, and 605-nm transients, consistent with initial reduction of Cu A according to Scheme 2. The relative amounts of reduced cytochrome a 2ϩ and Cu A 1ϩ formed remained in the ratio of 6.1:1 at all ionic strengths, indicating that the equilibrium constant for electron transfer between Cu A and cytochrome a remained constant at K ϭ k b /k c ϭ 6.1. The rate constant of the slow phase was a linear function of the CcO concentration above 90 mM ionic strength, from which the second-order rate constant, k 2nd , could be determined for the bimolecular reaction between solution phase Ru-55-Cc 2ϩ and CcO. k 2nd decreased as the ionic strength increased above 90 mM, consistent with the reaction between oppositely charged proteins (Fig. 3).
Three different experimental methods were used to determine whether Ru-55-Cc interacts with CcO in the same way as native horse Cc. In the first method, a solution containing 5.0 M Ru-55-Cc and 5.0 M CcO in 5 mM Tris-Cl, pH 8.0, was found to have a single fast phase of electron transfer with a rate constant of 4 ϫ 10 4 s Ϫ1 . The addition of 5.0 M horse Cc to this solution decreased the amplitude of the fast phase by 50%, and a new slow phase appeared with a rate constant of 140 s Ϫ1 and the same amplitude as the fast phase. This indicates that horse Cc binds to the high affinity site on CcO with the same affinity as Ru-55-Cc and displaces the latter from this site. The slow phase is due to the reaction of solution phase Ru-55-Cc with the 1:1 complex between horse Cc and CcO, which is rate-limited by dissociation of the complex (20). It is concluded that Ru-55-Cc has the same K D value as native horse Cc at 5 mM ionic strength, to within an error limit of Ϯ20%. In the second method, the K D values for both Ru-55-Cc and horse Cc were found to be less than 0.2 M at 5 mM ionic strength using an ultracentrifuge technique (see below). With this method, the K D value of Ru-55-Cc was found to be 3 Ϯ 0.7 M at 65 mM ionic strength, which is somewhat smaller than the K D value of 5.6 Ϯ 1.2 for horse Cc (see below). In the third method, the secondorder rate constants for the reactions of horse Cc and Ru-55-Cc with CcO were measured at high ionic strength using the lumiflavin flash photolysis technique described in Ref. 20. The rate constants at 155 mM ionic strength were (8 Ϯ 2) ϫ 10 7 M Ϫ1 s Ϫ1 for Ru-55-Cc and (7 Ϯ 2) ϫ 10 7 M Ϫ1 s Ϫ1 for horse Cc.
Reaction between Ru-55-Cc and CcO Subunit II Surface Mutants-The reactions of Ru-55-Cc with a series of R. sphaeroides CcO mutants containing acidic to neutral mutations at surface residues on subunit II were studied in order to determine the location of the binding domain for Cc. The largest effect on the kinetics was observed for the D214N mutant, which had no fast intracomplex phase at low ionic strength corresponding to the one observed for wild-type CcO. Instead, slow biphasic transients were observed for D214N at 550, 830, and 605 nm, with rate constants of k a1 ϭ 700 s Ϫ1 and k a2 ϭ 85 s Ϫ1 at low ionic strength (Table I). These rate constants were independent of protein concentration, indicating that they were due to intracomplex electron transfer. The rate constants increased to k a1 ϭ 2800 s Ϫ1 and k a2 ϭ 600 s Ϫ1 as the ionic strength was increased to 45 mM ( Fig. 4). At ionic strengths above 60 mM, only a single phase was observed, which had second-order kinetics (Fig. 4). The second-order rate constant was only 37 M Ϫ1 s Ϫ1 at 95 mM ionic strength, compared with 310 M Ϫ1 s Ϫ1 for wild-type CcO (Table I) and decreased with increasing ionic strength (Fig. 3). The equilibrium constant for electron transfer between Cu A and heme a was the same as wild type, K ϭ 6.2, supporting spectral analysis that shows no alteration in the Cu A or heme a, a 3 centers in these surface mutants (25). The E157Q mutant also had biphasic kinetics at low ionic strength, with rate constants of 12,000 s Ϫ1 and 260 s Ϫ1 , which were independent of protein concentration (Table I).
The rate constant of the fast phase did not change as the ionic strength was increased to 35 mM, but the fast phase disappeared at higher ionic strength (Fig. 5). The slow phase increased to a maximum of 2900 s Ϫ1 at 35 mM ionic strength and then decreased (Fig. 5). The kinetics of the E148Q and D195N mutants were intermediate between those of wild type and E157Q, while the D151N/E152Q and D188N/E189Q double mutants had similar kinetics to wild type (Table I; Figs. 3-5). The E254A mutant had a single intracomplex electron transfer phase at low ionic strength, with a rate constant of k a ϭ 10,500 s Ϫ1 . The rate constant of this phase did not change as the ionic strength was increased to 55 mM, and then this phase disappeared at higher ionic strength (Fig. 6). A slow phase first appeared at 25 mM ionic strength with a rate constant of 320 s Ϫ1 . The rate constant of this phase increased to a maximum of 3000 s Ϫ1 at 75 mM ionic strength and then decreased with further increases in ionic strength (Fig. 6).
The potential role of Trp 143 in mediating electron transfer between cytochrome c and Cu A was investigated using the mutants W143F and W143A. The reaction between Ru-55-Cc and the W143F mutant had a rate constant of 85 s Ϫ1 at low ionic strength, which was independent of protein concentration and therefore due to intracomplex electron transfer. The rate constant remained the same as the ionic strength was increased to 75 mM and then decreased with further increases in the ionic strength, indicating dissociation of the complex and a transition to second-order kinetics (Fig. 6). The ionic strength at which intracomplex electron transfer changes to secondorder kinetics is the same as for wild type, suggesting that the The intracomplex rate constants k a1 and k a2 for electron transfer from heme c to Cu A in 1:1 complexes between Ru-55-Cc and CcO mutants were measured in 5 mM Tris-Cl, pH 8, at 23°C. The second order rate constant k 2nd was measured in 5 mM Tris-Cl, pH 8, 90 mM NaCl, at 23°C. The equilibrium constant for electron transfer between Cu A and heme a, K ϭ k b /k c , was independent of ionic strength. The error limits in k a , K, and k 2nd are Ϯ20%.   Table I).
Ultracentrifuge Measurement of Dissociation Constant for Cc⅐CcO Complex-The formation of a complex between horse Cc and R. sphaeroides CcO was studied using the Beckman XL-A analytical ultracentrifuge operating in the sedimentation velocity mode as described by Cann (31). At 5 mM ionic strength, only a single band was observed with a sedimentation velocity of 11.4 S, indicating that all of the Cc was complexed to CcO (Fig. 8). As the ionic strength was increased above 40 mM, a second band appeared with a sedimentation velocity of 1.74 S, which is due to uncomplexed Cc. The concentration of uncomplexed Cc was measured from the 410-nm absorbance at the plateau of the Cc band, as described by Cann (31). The concentration of uncomplexed Cc increased as the ionic strength was increased, until it reached the total concentration of Cc in the cell at an ionic strength of 255 mM. The equilibrium dissociation constant of the complex was calculated from Equation 1.

K D ϭ ͓Cc][CcO]/[Cc⅐CcO]
(Eq. 1) This equation assumes the formation of only a 1:1 complex between Cc and CcO. Experiments using higher Cc concentrations than that of CcO demonstrated that Cc could also bind at a second low affinity site to form a 2:1 complex (data not shown). However, the dissociation constant of the second low affinity binding site was found to be more than 10-fold larger than that of the high affinity 1:1 complex at all ionic strengths. Thus, under conditions of excess CcO used in Fig. 8 and Table  II, Cc will bind predominantly to the high affinity site, and the K D measured using Equation 1 will be the dissociation constant of that site. The value of K D increases from less than 0.2 M at ionic strengths below 30 mM to 1.0 M at 45 mM, 5.6 M at 65 mM, and 18 M at 85 mM (Table II). K D becomes too large to measure at ionic strengths above 100 mM.
The ultracentrifuge method was used to measure dissociation constants of the complexes between Cc and the CcO surface mutants. All of the mutants formed a strong 1:1 complex with horse Cc at 5 mM ionic strength with a K D value less than 0.2 M. The greatest effect on the dissociation constant was observed for the E157Q and D214N mutants, which had K D values between 3 and 4 times larger than that of wild-type CcO in the ionic strength range of 45-85 mM (Table II). The D195N and E148Q mutants had K D values about 2-fold larger than that of wild type, while the D151N/E152Q, D188N/E189Q, and E254A mutants were similar to wild type. The W143F and W143A mutants also had nearly the same K D values as wild type throughout the measurable ionic strength range.
The dissociation constants of complexes between R. sphaeroides cytochrome c 2 and CcO mutants were also measured using the ultracentrifuge method (Table III). The dissociation constant for the complex of cytochrome c 2 with wild-type CcO was larger than that with horse Cc at all ionic strengths (Table  III). The effects of the CcO mutations on K D were somewhat different from the effects on the complex with horse Cc, as observed in the steady-state analyses (25). For example, the D195N mutation had a larger effect than the D214N mutation on K D for the complex with cytochrome c 2 , whereas the opposite was true for the complex with horse Cc (Tables II and III). These differences are most likely caused by differences in the charge distribution on the two cytochromes, as discussed in the accompanying paper (25). DISCUSSION

Intracomplex Electron Transfer between Ru-55-Cc and CcO-
The ruthenium photoreduction technique has made it possible  to resolve a key step in the reaction between Cc and CcO. The rate constant for intracomplex electron transfer from Ru-55-Cc to the initial acceptor in R. sphaeroides CcO is k a ϭ 4 ϫ 10 4 s Ϫ1 .
The kinetic results are consistent with a mechanism involving initial electron transfer from Ru-55-Cc to either Cu A or heme a, followed by rapid electron transfer equilibrium between Cu A and heme a. Since Cu A has been definitively assigned as the initial acceptor in beef CcO (20,22), and the x-ray crystal structures of beef and P. denitrificans CcO show essentially identical structure in the region of Cu A (23,24), the initial electron acceptor in R. sphaeroides CcO is undoubtedly Cu A . Since the rate constants for reduction of Cu A and heme a appear to be the same as that for reoxidation of heme c, the apparent rate constant for electron transfer between Cu A and heme a, k b ϩ k c , must be greater than 10 5 s Ϫ1 . The equilibrium constant for electron transfer between Cu A and heme a, K ϭ k b /k c ϭ 6.1, indicates that the redox potential of heme a is 46 mV more positive than that of Cu A in the state where heme a 3 and Cu B are both oxidized. A photoreduction technique utilizing a dimer of ruthenium trisbipyridine has been used to measure the rate constants for electron transfer between Cu A and heme a to be k b ϭ 9.3 ϫ 10 4 s Ϫ1 and k c ϭ 1.7 ϫ 10 4 s Ϫ1 (32). The value of k b is significantly larger than the corresponding rate constant for beef CcO, k b ϭ 2 ϫ 10 4 s Ϫ1 (19, 20, 33-35) but is comparable with the value of k b ϭ 6.7 ϫ 10 4 s Ϫ1 measured for R. sphaeroides CcO using a time-resolved electrogenic method (36).

Ionic Strength Dependence of the Reaction between Ru-55-Cc and
CcO-Although the physiological conditions for the reaction of Cc with CcO in R. sphaeroides are not known, it is likely that it occurs at relatively high ionic strength (37,38). We have, therefore, studied the reaction of Ru-55-Cc with R. sphaeroides CcO as the conditions were changed continuously from low ionic strength, where intracomplex kinetics are observed, to high ionic strength, where bimolecular kinetics occur. The rate constant of the fast phase of the reaction between Ru-55-Cc and CcO remains constant at (4 Ϯ 1) ϫ 10 4 s Ϫ1 as the ionic strength is increased from 5 to 80 mM. This indicates that the orientation of the 1:1 complex is optimal for electron transfer at low ionic strength and does not change as the ionic strength is increased. Similar behavior is also observed in the reactions of Cc with beef CcO (19, 20), cytochrome c 1 (39), cytochrome c peroxidase (40 -44), and cytochrome b 5 (45,46). Experiments using soluble reducing agents led to the suggestion that the Cc⅐CcO complex may not be optimized for rapid electron transfer at low ionic strength (47). However, it was not possible to resolve the rapid electron transfer reaction within the high affinity Cc⅐CcO complex using these techniques (19, 20).
Photoinduced electron transfer from Ru-55-Cc to CcO in-volves both a fast intracomplex phase and a slow bimolecular phase at intermediate ionic strengths (Fig. 2). This observation provides strong evidence that the complete bimolecular reaction between solution Cc and CcO involves formation of a 1:1 complex followed by intracomplex electron transfer according to Scheme 2. The dissociation rate constant k d must be much smaller than k a , since if k d were larger than k a , then rapid equilibrium conditions would apply, and separate slow and fast phases would not be observed (44). Assuming that the bimolecular reaction obeys Scheme 2 and k d Ͻ Ͻ k a , the observed rate constant is given by Equation 2 (44), where E o is the concentration of CcO and C o is the concentration of Ru-55-Cc. From this equation, the formation and dissociation rate constants can be estimated to be k f ϭ 6.1 ϫ 10 8 M Ϫ1 s Ϫ1 and k d ϭ 1.6 ϫ 10 4 s Ϫ1 at 75 mM ionic strength.
The second-order rate constant k 2nd for the reaction between Ru-55-Cc and R. sphaeroides CcO decreases rapidly with increasing ionic strength (Fig. 3), providing another indication that there is a strong electrostatic interaction between Cc and CcO (4,48,49). The ionic strength dependence of k 2nd indicates that the electrostatic interaction between Ru-55-Cc and R. sphaeroides CcO is essentially the same as that between horse Cc and beef CcO (4). k 2nd is most likely a function of all three rate constants, k f , k a , and k d , at high ionic strength.
The turnover number for the steady-state reaction between wild-type R. sphaeroides CcO and 20 M horse Cc has a bellshaped dependence on ionic strength, with a maximum at 75 mM ionic strength (25). A bell-shaped ionic strength dependence of the turnover number was also observed for the reaction between P. denitrificans CcO and 20 M horse Cc (17,18). The decreased turnover number at low ionic strength was suggested by these authors to be due to an electrostatically stabilized reactant complex that is locked into a configuration that is poorly oriented for electron transfer. The ruthenium rapid kinetic experiments clearly indicate that the intracomplex electron transfer rate constant k a is very large and independent of ionic strength and cannot limit turnover at low ionic strength. Therefore, enzyme turnover must be limited by the rate of product dissociation, k d , at low ionic strength. This indicates an increase in k d with increasing ionic strength, leading to faster turnover until the optimum ionic strength is reached. Above the optimum ionic strength, the complex will be dissociated, and turnover will be limited by the rate of complex formation, k f , which is expected to decrease with increasing ionic strength.
Definition of the Cc Binding Domain by Studies of CcO Surface Mutants-In order to investigate the role of specific residues on CcO in the interaction with Cc, kinetic and equilibrium binding studies were carried out on a series of mutants in which acidic residues on subunit II were replaced with neutral Asn or Gln residues (Fig. 9). All of the mutants formed high affinity 1:1 complexes with Ru-55-Cc at low ionic strength, allowing measurement of intracomplex electron transfer. The largest effect was observed for the D214N mutant, which has an intracomplex electron transfer rate constant of only 700 s Ϫ1 at 5 mM ionic strength. Asp 214 is located close to Trp 143 , the proposed electron entry site to Cu A , and is hydrogen-bonded to the Cu A ligand His 217 . The D214N mutation causes Ru-55-Cc to bind in a configuration that is very poorly oriented for electron transfer to Cu A at low ionic strength. The intracomplex rate constant increases to 2800 s Ϫ1 at 45 mM ionic strength, suggesting that a decrease in the strength of the electrostatic interactions allows Cc to undergo rotational diffusion to an orientation that is more efficient for electron transfer. The D214N mutation leads to only a 3-fold increase in the equilibrium dissociation constant, K D , indicating that Asp 214 plays a more important role in maintaining the optimal configuration for rapid electron transfer from Cc to Cu A than in simply contributing to the binding energy. The E157Q, E148Q, and D195N mutants have intracomplex electron transfer rate constants of 1.2 ϫ 10 4 s Ϫ1 , 1.5 ϫ 10 4 s Ϫ1 , and 2.5 ϫ 10 4 s Ϫ1 , respectively, indicating that the binding orientation of Ru-55-Cc to these mutants is not quite as favorable for electron transfer as that of wild-type CcO. The intracomplex rate constants for these mutants do not change with increasing ionic strength, demonstrating that the orientations of the complexes are independent of ionic strength. The effects of the E157Q, E148Q, and D195N mutations on K D and k 2nd are similar to their effects on intracomplex electron transfer, indicating that Glu 157 , Glu 148 , and Asp 195 make comparable contributions to binding strength and orientation. The D151N/ E152Q and D188N/E189Q double mutants have essentially the same k a , K D , and k 2nd values as wild-type CcO, indicating no change in binding strength or orientation.
The effects of the surface charge mutations on steady-state kinetics with horse Cc, reported in the accompanying paper (25), are fully consistent with the effects on the rapid kinetics and the equilibrium dissociation constant. The effect on the turnover number at 75 mM ionic strength follows the order D214N Ͼ E157Q Ͼ E148Q Ͼ D195N Ͼ D151N/E152Q Ͼ D188N/E189Q Ϸ wild type (25), the same as observed for the second-order rate constant (Table I). The mutations E157Q, D214N, E148Q, and D195N each shift the ionic strength for optimal turnover number to a lower value, and the extent of this shift follows the same order as the effect of the mutations on the equilibrium dissociation constant K D . Thus, the optimum in the bell-shaped turnover number profile represents the ionic strength at which the complex dissociates, and mutations that increase K D lead to complex dissociation at a lower ionic strength. Steady-state kinetics studies have also been carried out on acidic-to-neutral subunit II mutants of P. denitrificans CcO (17). The effect on the turnover number with horse Cc at 56 mM ionic strength follows the order E157N Ͼ D214N Ͼ E148Q Ͼ D195N Ͼ wild type (R. sphaeroides numbering). This is similar to the results with the R. sphaeroides mutants, except that the effect of the D214N mutant is smaller than that of E157N. The second-order rate constant of the D214N P. denitrificans mutant is only decreased by 17% compared with wild-type CcO, whereas the D214N R. sphaeroides mutant is decreased by a factor of 10. It appears that Asp 214 does not play as important a role in the interaction of Cc with P. denitrificans CcO as it does with R. sphaeroides CcO.
Previous chemical modification (14) and mutagenesis (15,50) studies have suggested that Glu 254 might be involved in binding Cc. However, the x-ray crystal structures of both bovine and Paracoccus CcO revealed that the backbone carbonyl of Glu 254 is a ligand to the Cu A binuclear center, while its carboxylate group is a ligand to magnesium in subunit I (Fig. 9) (23,24). It therefore appears unlikely that this internal residue could interact directly with Cc. Supporting this conclusion, the kinetic results for the E254A mutant are quite different from the other charge mutants. The value of k a was decreased to 1.0 ϫ 10 4 s Ϫ1 at low ionic strength, but the ionic strength at which intracomplex kinetics changed to bimolecular kinetics was the same as for wild-type CcO. This indicates that the equilibrium dissociation constant K D is the same as that of wild-type CcO, in agreement with the ultracentrifuge results. In contrast to the other mutants examined, E254A has an altered heme spectrum, does not bind manganese, and has a lower stability leading to some loss of subunit II (25). Therefore, the decreased value of k a for the E254A mutant probably arises from a secondary change in structure rather than a direct effect of the altered residue on Cc binding orientation. The earlier finding that Cc binding protected Glu 254 from chemical modification probably results from its location at the aqueous interface between subunits I and II (14).
Identification of the Electron Entry Site into Cu A -The x-ray crystal structures of bovine and P. denitrificans CcO have revealed that a highly conserved sequence of acidic and aromatic residues terminates with Trp 143 , which is located on the surface of subunit II and is in van der Waals contact with the Cu A center (Fig. 9). In order to test the involvement of the aromatic indole group of Trp 143 in electron transfer from Cc to Cu A , the mutants W143F and W143A were prepared and found to have the same visible, Cu A EPR, and manganese EPR spectral properties as wild-type CcO (25). The small changes in K (Table I) indicate that the redox potential of Cu A changes by only Ϫ18 Ϯ 7 mV and ϩ7 Ϯ 7 mV for W143F and W143A, respectively, assuming that the redox potential of heme a does not change in these mutants. The intracomplex electron transfer rate constants for W143F and W143A are 85 and 32 s Ϫ1 , respectively, which are 450-and 1200-fold smaller than that of wild-type CcO. This very large decrease in k a strongly suggests that a pathway for electron transfer between heme c and Cu A has been substantially disrupted. Since there is essentially no change in the equilibrium dissociation constant K D for either mutant, it is unlikely that the decrease in k a is due to a change in the binding configuration of Ru-55-Cc on the surface of the mutant. This conclusion is supported by the finding that the rate constant remains the same as the ionic strength is increased to 75 mM and then decreases continuously with further increases in ionic strength. In contrast to wild-type CcO and the other mutants, only a single phase is observed for W143F and W143A during the transition from intracomplex to bimolecular kinetics as the ionic strength is increased from 60 to 100 mM. This indicates that k d is large compared with k a , and rapid equilibrium binding conditions are satisfied. The second-order rate constant measured at high ionic strength is affected by these mutations nearly as much as the intracomplex rate constant k a measured at low ionic strength, providing further evidence that the mutations do not cause a change in binding configuration. The steady-state turnover numbers of W143F and W143A (40 and 20 s Ϫ1 , respectively) are similar to the values of k a , indicating that intracomplex electron transfer from Cc to Cu A becomes rate-limiting in these mutants (25). Witt et al. (18) have reported that the mutation W143Q in P. denitrificans CcO also leads to a large loss in steady-state electron transfer activity, further supporting the identification of Trp 143 as the electron entry site to Cu A .
Model for the Cc/CcO Interaction Domain and Comparison of the Rate of Electron Transfer from Heme c to Cu A with Theory-The kinetic and binding studies of the surface mutants indicate that Cc binds to a highly conserved acidic domain on the surface of subunit II (Fig. 9). It is concluded that the acidic residues Asp 214 , Glu 157 , Glu 148 , and Asp 195 interact electrostatically with the lysines surrounding the heme crevice of Cc, while Trp 143 at the center of the interaction domain mediates electron transfer from heme c to Cu A . Asp 151 and Glu 152 are located at the edge of the binding domain, while Asp 188 and Glu 189 are completely outside it. The absence of any effect due to substitution of these residues demonstrates that the charge effects are specific, not just a function of overall charge on subunit II. Extensive chemical modification studies have implicated lysines 8, 13, 72, 86, and 87 surrounding the heme crevice of Cc in the interaction with CcO (7-13). In the accompanying paper, Roberts and Pique (51) have used the computational docking program DOT and the crystal structures of horse Cc and bovine CcO to characterize the interaction between Cc and CcO. The complex between Cc and CcO determined from this theoretical approach is fully consistent with the kinetic and binding studies (Fig. 9). The interaction consists of a central hydrophobic domain, surrounded by complementary electrostatic interactions between Cc Lys 8 , Lys 13 , Lys 86/87 , and Lys 72 and CcO Asp 195 , Glu 157 , Glu 148 , and Asp 214 , respectively. At the center of the hydrophobic domain, the indole ring of Trp 143 is in van der Waals contact with the heme CBC methyl group.
According to a semiclassical theory developed by Marcus (52,53), the rate constant for electron transfer is controlled by the driving force ⌬G 0 Ј, the reorganization energy , and the electronic coupling H AB between the two redox centers. In this equation, r represents the distance between the closest macrocycle atoms in the two redox centers, the van der Waals contact distance r o ϭ 3.6 Å, ␤ is taken to be 1.4 Å Ϫ1 , and the nuclear frequency k o is 10 13 s Ϫ1 (54). The driving force ⌬G 0 Ј for the reaction between Cc and Cu A has been found to be Ϫ0.030 eV (19), consistent with the difference in redox potential between bound Cc and Cu A (55,56). Measurements for the intrinsic reorganization energy c of Cc have ranged from 1.04 eV (57) to 1.5 eV (58). Binding Cc to CcO is expected to lower the solvent contribution to the reorganization energy by excluding water from the interaction domain. A decrease in of 0.09 eV was observed for the complex between Cc and cytochrome b 5 due to this effect (59). Taking this into account, the intrinsic reorganization energy of bound Cc will be assumed to be c ϭ 1.1 eV. The intrinsic reorganization energy of Cu A has been calculated to be A ϭ 0.2 eV (60), which is consistent with kinetic studies of the reaction between Cu A and heme a (61). The reorganization energy for the reaction between Cc and Cu A can be calculated from the Marcus cross-relation to be cA ϭ ( c ϩ A )/2 ϭ 0.65 eV. Using this value of cA and ⌬G 0 Ј ϭ Ϫ0.03 eV, the Frank-Condon term in Equation 2 is 0.0032. The reaction between Cc and Cu A is thus significantly limited by the low value of the driving force relative to the reorganization energy.
In the Cc⅐CcO complex determined by Roberts and Pique (51), the closest distance between the heme macrocycle and a Cu A ligand is 12.5-13.0 Å, which is between heme C2C and the sulfur atom of Met 263 (Fig. 9). Using this distance and the values of and ⌬G 0 Ј given above, k et is calculated to be (6 -12) ϫ 10 4 s Ϫ1 from Equation 4. This is in good agreement with the experimental values of k et ϭ 6 ϫ 10 4 s Ϫ1 for the reaction between Ru-39-Cc and Cu A in beef CcO (20), and 4 ϫ 10 4 for the reaction between Ru-55-Cc and R. sphaeroides CcO. Application of the dominant pathway method developed by Beratan et al. (62) is problematic in this case, because it does not take into account the orbitals of the indole ring. Nevertheless, electronic coupling should be extremely good because the indole ring of Trp 143 is in direct van der Waals contact with both the Cc heme and the Cu A ligands Cys 256 and Met 263 (Fig. 9).
The large decrease in k a observed in the W143F and W143A mutants is most likely due to a decrease in the electronic coupling. The small changes in the redox potential of Cu A indicate that the driving force for electron transfer from Cc to Cu A changes by only ϩ18 Ϯ 7 mV and Ϫ7 Ϯ 7 mV for W143F and W143A, respectively. Since there is no change in the Cu A EPR spectrum or 830-nm absorption band of these mutants, it is also unlikely that the reorganization energy has been altered. In addition, since the W143F and W143A mutations do not affect the binding energy, it is reasonable to assume that the configuration of the Cc⅐CcO complex, including the distance between heme c and Cu A , is not affected. Equation 4 would thus predict no change in rate constant for these mutants. The decreased rate constant may be due to the presence of an additional through-space gap between the edge of heme c and the Phe 143 or Ala 143 side chain. Using the Beratan theory, an additional gap of 1.8 Å would account for the 450-fold decrease in k a for the W143F mutant, while a gap of 2.1 Å would account for the 1200-fold decrease for W143A. The size of this additional gap is consistent with the size difference between a methyl or phenyl group and an indole group. Other possible interpretations of these findings would require more drastic changes in the structure of the protein interface and cannot be ruled out without atomic level structural information about the complex. Computational docking analysis may give some preliminary clues (51).