Verkhovsky et al. reply

Ten years ago, one of us showed that all proton translocation by cytochrome c oxidase seems to be coupled to the enzyme's oxidative catalytic half-cycle1. This led to the assumption that all proton translocation takes place during this half of the cycle. Our more recent data2 contradict that idea, as well as Michel's hypothesis3, but our original findings1 still stand.

The main reason for our proposal that proton translocation during the reductive half of the cycle may still be energetically coupled to the preceding oxidative phase was our observation that, after the energy-rich O∼ state had relaxed to O, there was no proton translocation associated with reduction2. Michel does not mention this, although it contradicts his model3.

Michel is right that our2 Fig. 2a shows the release of about 1.2 H+, but this was in a single experiment. In the text2 we reported the release of 1–2 H+. This variation is due to the difficulty of having all the enzyme molecules reduced by exactly four electrons. When more than four electrons are available (Fig. 2b of ref. 2), there is no such difficulty and the enzyme is always fully reduced. The result in Fig. 2b therefore shows much less variation, and consistently yielded four released H+. In the single trace of Fig. 2a, all enzyme molecules had not been fully reduced by four electrons. This can be seen from the small, slow tail in the corresponding absorbance trace, resulting from the slow decay of oxygen intermediates that arise when the fraction of enzyme with less than four electrons reacts with oxygen.

All the enzyme relevant to these experiments is correctly orientated relative to the membrane. We ascertained that incorrectly orientated enzyme is not reduced under our experimental conditions by the charged reductants ferrocytochrome c and hexammine ruthenium [II].

It has been shown4 that, in conditions similar to those used for our Fig. 2b (ref. 2), any additional turnovers resulting from an excess of oxygen over the enzyme would be very slow (roughly 15 seconds), as such turnovers depend on slow intermolecular reactions. Because the reaction in Fig. 2b is much faster, Michel's claim that the result is due to multiple turnovers cannot be correct. Michel's conclusion that the O2 concentration was superstoichiometric in this experiment is also flawed, because we ascertained the stoichiometry of O2versus enzyme by titrating the reduced enzyme spectroscopically with known amounts of O2. Thus, both the enzyme concentration and the number of protons released were determined on the basis of the concentration of O2 added.

Michel's comments about extinction coefficients are therefore not relevant, and they are also incorrect. He concludes that the path length was ∼1 cm in our experiments2, but we stated5 that the path length varies from experiment to experiment depending on the optical geometry, and that it must be determined separately each time. A calculation based on Fig. 2 of ref. 5 would have shown this.

Finally, assuming there would have been both an excess of O2 and an excess of reductant (cytochrome c) in the experiment2 shown in our Fig. 2b, and, as claimed by Michel, the first turnover would pump three protons and the subsequent turnovers four protons each, we can calculate that nine turnovers would be needed to yield 3.9 H+ per O2 (note that H+ per O2 is the primary quantity measured). Michel calculates 1.8 turnovers, which would give only 3.4 protons, far below the number observed. Obtaining these extra turnovers would require not only an excess of O2, but also an excess of ferrocytochrome c. Our experiment had neither of these but, even if it had both, our result could still not be explained by multiple turnovers.