Proton-transfer pathways in the mitochondrial S. cerevisiae cytochrome c oxidase

In cytochrome c oxidase (CytcO) reduction of O2 to water is linked to uptake of eight protons from the negative side of the membrane: four are substrate protons used to form water and four are pumped across the membrane. In bacterial oxidases, the substrate protons are taken up through the K and the D proton pathways, while the pumped protons are transferred through the D pathway. On the basis of studies with CytcO isolated from bovine heart mitochondria, it was suggested that in mitochondrial CytcOs the pumped protons are transferred though a third proton pathway, the H pathway, rather than through the D pathway. Here, we studied these reactions in S. cerevisiae CytcO, which serves as a model of the mammalian counterpart. We analyzed the effect of mutations in the D (Asn99Asp and Ile67Asn) and H pathways (Ser382Ala and Ser458Ala) and investigated the kinetics of electron and proton transfer during the reaction of the reduced CytcO with O2. No effects were observed with the H pathway variants while in the D pathway variants the functional effects were similar to those observed with the R. sphaeroides CytcO. The data indicate that the S. cerevisiae CytcO uses the D pathway for proton uptake and presumably also for proton pumping.


Results
We investigated two D pathway structural variants (Asn99Asp and Ile67Asn) and two H pathway variants (Ser382Ala and Ser458Ala). These structural variants were chosen from a series of modified yeast strains that displayed apparent respiratory growth defects 21 . In earlier studies, reduced minus oxidized difference spectra were recorded of samples of each mutant CytcO after affinity chromatography purification. These difference spectra of Asn99Asp and Ser382Ala were identical to that of the wild-type CytcO 21 whereas those of Ile67Asn and Ser458Ala both displayed a 2-3 nm shift of their alpha-bands to 601 nm 22 and 606 nm, respectively. A comparison of the numbering of these residues in CytcOs from S. cerevisiae, R. sphaeroides and bovine heart mitochondria is shown in Table 1.

Reaction of the reduced cytco with o 2 .
To investigate the kinetics of electron transfer in the wild-type and structural variants of the CytcO, the enzyme was fully reduced (by four electrons) and incubated under an atmosphere of carbon monoxide, which binds at heme a 3 , i.e. the ligand blocks access to O 2 . The sample was then rapidly (~2 ms) mixed with an O 2 -saturated solution in a stopped-flow apparatus. Because CO dissociation in the dark is slow (~30 s), the CytcO stays essentially fully reduced over a time scale of milliseconds. About 200 ms after mixing, the ligand was dissociated by means of a short (~10 ns) laser flash, which initiates the reaction of the rescued CytcO with O 2 simultaneously in the entire CytcO population. Figure 2 shows absorbance changes associated with reaction of the reduced wild-type CytcO with O 2 at 445 nm at pH 7.5 and pH 10. The initial increase in absorbance at t = 0 is associated with CO dissociation. The decrease in absorbance with a time constant of ~40 µs is associated with electron transfer from heme a to the catalytic site forming the peroxy state (P R ). This reaction is not associated with any proton uptake from solution and displays a very small pH dependence. In the next step a proton is taken up from solution to form the ferryl state, F with a time constant of ~100 µs at pH 7.5. This reaction is mainly seen at 580 nm where absorbance changes are small and it could not be resolved in this study. At 445 nm the reaction is observed as a small lag before the decrease in absorbance with a time constant of ~1 ms at pH 7.5. In this final step of the reaction the F state decays to form the oxidized CytcO (state O), which is associated with proton uptake from solution and proton pumping. Figure 3A shows the pH dependence of the F → O reaction. The rate constant for the wild-type enzyme dropped from ~2·10 3 s −1 at low pH to ~20 s −1 at pH 10.5. This pH dependence is qualitatively similar to that observed previously with the CytcO from R. sphaeroides or from bovine heart mitochondria 23,24 . A single kinetic component was observed in the pH range 6-9 and at pH 10.5, but at the intermediate pH values the decay was biphasic with approximately equal amplitudes of the two components. For these points an average of the two rates is plotted in Fig. 3A. With the two H pathway variants, Ser382Ala and Ser458Ala, the rates at pH 7 and pH 10 were similar to those observed with the wild-type CytcO ( Table 2).
Residue Asn99 is found near Asp92, which is located at the orifice of the D pathway (see Fig. 1). With the Asn99Asp mutant CytcO, we observed F → O rate constants of 1200 ± 100 s −1 and 780 ± 100 s −1 at pH 7 and pH 10 (10 measurements on each of two samples), respectively, i.e. the rate constant was larger with the Asn99Asp than with the wild-type CytcO at pH 10, consistent with earlier observations made with the R. sphaeroides CytcO 25 .
Residue Ile67 is located at a distance of ~3 Å from Glu243, a key D pathway residue, at the end of the pathway (see Fig. 1). With the Ile67Asn, the F → O rate was about a factor of 100 slower than that seen with the wild-type CytcO, which is consistent with proton uptake through the D pathway in this reactions step. To test the effect of this structural alteration in a bacterial CytcO, we introduced a mutation at the equivalent residue, Met107 in the R. sphaeroides CytcO, which is also located ~3 Å from Glu286 (equivalent of Glu243 in S. cerevisiae). In both S. cerevisiae (Ile67Asn) and R. sphaeroides (Met107Cys) mutation at this site resulted in lower CytcO activity of ~10% 22 and ~6%, respectively (see "Materials and Methods"). With the R. sphaeroides mutant CytcO we obtained www.nature.com/scientificreports www.nature.com/scientificreports/ sufficient amount of CytcO material to be able to perform measurements also at 580 nm where the P R → F reaction step is clearly observed as an increase in absorbance in the time range (Fig. 3B). As seen in the figure, the P R → F reaction rate constant was a factor of ~50 slower than that observed with the wild-type CytcO. Because the absorbance changes associated with the P R → F reaction reflect proton uptake through the D pathway 26,27 , the data indicate that in the R. sphaeroides CytcO, the amino-acid residue replacement slows proton uptake through this pathway, presumably due to an interaction with Glu286. Note that at high pH, i.e. when proton uptake from solution becomes rate limiting for the P R → F reaction, similar rate constants were observed with the wild-type and mutant CytcOs. Similar to the results obtained with the S. cerevisiae CytcO, the replacement of Met107 by a Cys in R. sphaeroides CytcO resulted in slowing the F → O reaction rate constant by a factor of 50 to ~20 s −1 at pH 7 (Fig. 3C).

Discussion
Results from earlier studies with bacterial CytcOs showed dramatic effects of mutations in the D pathway on the rate of proton transfer and/or the proton-pumping stoichiometry 28 . In the R. sphaeroides Asn139Asp variant (Asn99Asp in S. cerevisiae) proton pumping was uncoupled from O 2 reduction while the activity of the CytcO was slightly increased 29 . The maximum rate of the F → O reaction at pH < 7 was unaffected by the mutation, but the pH dependence of the reaction rate was altered resulting in an increase in the pK a from 9.4 to ~11 25 . Mutation of Asn207 to Asp (Asn164Asp in the S. cerevisiae CytcO), near Asn139, also resulted in uncoupling of proton pumping and an altered pH dependence of the F → O reaction 25 . No effects on proton pumping were observed with the equivalent mutants in the mammalian CytcO, which prompted the authors to conclude that the function of the D pathway is not conserved 8,11 and that the H pathway is involved in proton pumping 11 .
The results from the present study show that the F → O reaction rate at pH 7 was similar with the Asn99Asp CytcO as with the wild-type CytcO ( Table 2). As already noted above with the R. sphaeroides CytcO, the characteristic feature of this variant is a shift in the pH dependence of the F → O rate resulting in an elevated pK a value. While this rate constant with the wild-type R. sphaeroides CytcO decreased by a factor of five from ~1000 s −1 to ~200 s −1 upon increasing the pH from 6 to 10, in Asp139Asn variant it decreased from 1200 s −1 to 900 s −1 , i.e. by a factor of ~1.3 25 .
With the S. cerevisiae wild-type CytcO, the F → O rate constant decreased from ~2200 s −1 at pH 7 to ~330 s −1 at pH 10, i.e. by a factor of ~7, while with the Asn99Asp CytcO the rate decreased from 1200 s −1 at pH 7 to ~780 s −1 at pH 10, i.e. by a factor of ~1.6 ( Table 2). Hence, we observed a similar behaviour with the S. cerevisiae enzyme as with the R. sphaeroides CytcO.
The current study shows that structural alterations close to the D pathway, at the site of residue Ile67 in S. cerevisiae or Met107 in R. sphaeroides CytcO, result in a dramatically slower F → O reaction rate (by factors of ~100 and ~50, respectively). This residue is located at hydrogen-bonding distance from the Glu243 (or Glu286 in the R. sphaeroides CytcO). Results from earlier studies have shown that replacement of Glu286 by other residues results in a dramatic decrease or impaired proton transfer through the D pathway 30 . Hence, this segment of the D pathway is particularly sensitive to structural alterations. Because with the R. sphaeroides CytcO we observed a ~50-fold slower P R → F reaction rate with the Met107Cys variant (Fig. 3), the effect of the mutation is www.nature.com/scientificreports www.nature.com/scientificreports/ a dramatically slowed proton transfer through the D pathway. Hence, we conclude that also with the S. cerevisiae CytcO the slowed F → O reaction rate is caused by slowed proton transfer through the D pathway. This interpretation is also consistent with earlier FTIR studies which showed that the low activity of the Ile67Asn variant was linked to perturbations of Glu243 and impairment of its proposed role in proton coupled electron transfer 22,31 .
The observations discussed above are consistent with data obtained from studies of the Glu243Asp variant 32 , and, collectively, they indicate that the F → O reaction is associated with proton transfer through the D pathway also in the S. cerevisiae CytcO. We did not observe any effects on the F → O rate constants at pH 7 or pH 10, measured with the H pathway variants (see Table 2). Because proton pumping in CytcO is driven by electron transfer to O 2 (and the associated uptake of substrate protons for water formation), if proton pumping would be impaired in the H-pathway mutants, we would expect to observe an effect on the O 2 -reduction reaction. Because the oxidation kinetics was unperturbed, we conclude that the functionality of the mutants investigated here is unperturbed.
Data from earlier studies showed that the Ser458 to Ala replacement resulted in a decreased respiratory growth 21 . However, this observation does not imply that proton-pumping would be impaired in the Ser458Ala variant. For example, the mutation could instead yield a fraction CytcO that is not assembled correctly, resulting in slowed overall respiration. In the present study we studied purified CytcO and the observed signals originate from CytcO that bind CO to an intact catalytic site. Results from a study with the R. sphaeroides CytcO in which Ser425, the equivalent of the S. cerevisiae CytcO Ser382 residue, was modified, showed that this residue is not important for proton conduction, but only results in a small increase in the midpoint potential of heme a 33 .
The data discussed above indicate that the mitochondrial CytcO from S. cerevisiae uses the D pathway for proton transfer from solution to the catalytic site in the F → O reaction step. Even though here we studied electron transfer to O 2 at the catalytic site, this electron transfer is linked to proton pumping. Thus, as indicated above, any effect on proton pumping is expected to result in an effect on the intramolecular electron transfer. Mutations of residues in the D pathway displayed qualitatively the same functional characteristics in the S. cerevisiae as with the R. sphaeroides CytcO and the F → O rate was insensitive to changes in the proposed H pathway. Hence, the data suggest that there are no major functional differences between the mitochondrial S. cerevisiae and the bacterial R. sphaeroides CytcOs.
Methods preparation of mutants. The construction of the modified yeast strains is described in 21 .
To prepare the Met107Cys CytcO variant in R. sphaeroides a Quik-Change II site-directed mutagenesis kit (Agilent technologies) was used. The resulting amino-acid replacement was verified by sequencing. The PJS3-SH plasmid was used as a template plasmid for making mutations while pRK415-1 plasmid 34 was used for expression 35 . preparation of the S. cerevisiae cytco. Yeast cells were grown aerobically in YPGal medium at 28 °C and harvested in late log phase as described in 21 . Mitochondrial membranes were prepared and the S. cerevisiae CytcO was purified as described by Meunier et al. 21 with some modifications as outlined here. During solubilization the membranes were diluted to 2 mg/ml in buffer (50 mM KPi, 100 mM KCl and 1.5% n-dodecyl β -D-maltoside (DDM (w/v)) and solubilized for 1 hour. To remove cyt. c the supernatant from the centrifugation step after solubilization was run over a column loaded with a cation-exchanger (Bio-Rex 70, Bio-Rad) equilibrated with 50 mM KPi, 100 mM KCl and 0.035% DDM. After addition of 5 mM imidazole to the flow-through from the ion-exchanger it was loaded on a column filled with 25 ml Ni-resin (NI Sepharose 6 Fast Flow from GE healthcare), pre-equilibrated with (20 mM KPi, 150 mM KCl and 0.035% DDM). The column was washed with four column volumes of wash buffer (20 mM KPi, 150 mM KCl, 10 mM imidazole and 0.035% DDM). This was followed in time by a two-step elution of four column volumes at each imidazole concentration (20 mM KPi, 150 mM KCl, 40 or 100 mM imidazole and 0.035% DDM). The eluted fractions were pooled and concentrated in a 100 kDa cut-off filter (Merck Millipore). Buffer exchange (elution buffer without imidazole) was performed to lower the imidazole content to sub-µM concentrations.
preparation of the R. sphaeroides cytco. Expression of the R. sphaeroides CytcO was achieved by growing bacteria aerobically (Sistrom medium) in the dark at 30 °C. After harvesting the cells (at OD 550 ≅ 1.5) they were re-suspended (50 mM Tris-buffer, pH 8.0) at 4 °C in the presence of DNase I (0.05 mg/ml final concentration) and passed twice through a continuous-flow cell disruptor (Constant Systems LTD) operating at 170 MPa. The inner membrane fraction was collected by ultracentrifugation (138 000 g for 90 minutes at 4 °C) and 1.5% DDM was added to solubilize the membrane fraction. The histidine-tagged CytcO was purified using Ni 2+ -NTA affinity chromatography, essentially as described in 35,36 .  flow-flash measurements. The sample was prepared by exchanging the CytcO buffer to 10 mM (wild-type, Ile67Asn and Met107Cys variants) or 3 mM (Asn99Asp, Ser382Ala and Ser458Ala)) Bis-tris propane pH 7, 150 mM KCl and 0.035%DDM. The CytcO concentration after buffer exchange was 3-10 µM depending on mutant (see figure legends). The sample was transferred to a Thunberg cuvette and air was exchanged for nitrogen. The sample was then reduced with 4 mM ascorbate and 1 µM PMS (electron mediator) (S. cerevisiae CytcO) or 2 mM ascorbate and 1 µM ruthenium hexachloride (electron mediator) (R. sphaeroides CytcO). Nitrogen was then exchanged for CO. The absorbance spectra were recorded in the range 400-700 nm at each step to ensure complete reduction and ligation with CO. The sample was loaded into one syringe of a modified stopped-flow apparatus (Applied Photophysics). The other syringe was loaded with an oxygen-saturated buffer. The buffer composition for measurements with the S. cerevisiae CytcO: 100 mM (for 1:5 mixing of CytcO:O 2 solutions) or 200 mM buffer (for 1:1 mixing) MES for pH 6 and 6.5; Bis-tris propane pH 7, 7.5 and 9; CHES for pH 9.5 and CAPS for pH 10 and 10.5. In addition, the buffer was supplemented with 150 mM KCl and 0.035% DDM. For measurements with R. sphaeroides CytcO: 100 mM Bis-tris propane buffer for pH 6.5, 7.5, 8 and 9; CAPS for pH 10.5 with 0.05% DDM and 0.1 mM EDTA. The CO ligand was dissociated at 0.2 s after mixing, which initiated the reaction. The reaction was monitored by following in time absorbance at specific wavelengths (see figure legends). An amplifier (C11184, Hamamatsu) was used to amplify the signal before recording using a digital oscilloscope.