Proc. Natl. Acad. Sci. USAVol. 90, pp. 5949-5953, July 1993Biochemistry

Proton pumping by cytochrome oxidase as studied by time-resolvedstopped-flow spectrophotometryGIOVANNI ANTONINIt, FRANCESCO MALATESTAt, PAOLO SARTIt, AND MAURIZIo BRUNORIt§tDepartment of Experimental Medicine and Biochemical Sciences, University of Rome "Tor Vergata," Department of Biochemical Sciences and ConsiglioNazionale delie Ricerche Center for Molecular Biology, University of Rome "La Sapienza," Rome, Italy; and *Institute of Biological Chemistry,University of Cagliari, Cagliari, Italy

Communicated by E. Margoliash, December 21, 1992

ABSTRACT The H+/e- stoichiometry for the protonpump of cytochrome c oxidase reportedly varies between 0 and1, depending on experimental conditions. In this paper, wereport the results obtained by a combination oftransient opticalspectroscopy with a time resolution of 10 ms and a singularvalue decomposition analysis to follow the kinetics, separate theobserved spectral components, and quantitate the stoichiom-etry of the pump. By using cytochrome oxidase reconstitutedinto small unilamellar vesicles, we show that the time coursesof ferrocytochrome c oxidation and phenol red acidification oralk tion fit a simple kinetic scheme. The fitting procedureleads to unbiased and objective determination of the H+/e-ratio under various experimental conditions. The proton-pumping stoichiometry was found to be 1.01 ± 0.10, indepen-dent ofthe number ofturnovers, proton back-leak rate, or typeof experiment (oxidant or reductant pulse).

Cytochrome c oxidase, the terminal enzyme ofthe mitochon-drial respiratory chain, is the crucial site of the energytransduction machinery in mitochondria. Evidence that cy-tochrome oxidase vectorially pumps protons was initiallyprovided by Wikstrom (1) with rat liver mitochondria, wherehe observed acidification of the external medium associatedwith the oxidation of ferrocytochrome c by oxygen. Proton-pumping evidence was substantiated by more controlledexperiments carried out with the simpler system of purifiedcytochrome oxidase reconstituted into small unilamellar ves-icles (cytochrome c oxidase vesicles, COVs), by monitoringthe external medium acidification or internal medium alka-linization either potentiometrically (pH meter) or spectro-scopically (pH indicator) (2-6).

Present structural evidence shows that three metal sites(cytochrome a, cytochrome a3, CUB) are bound to proteinligands provided by subunit I, whereas CUA is thought to bebound to subunit II (7-9). These metals must mediate theelectron-transfer processes from ferrocytochrome c leadingto the breakdown of dioxygen to water and coupled protontranslocation (9). The steady-state kinetics of cytochrome coxidation by cytochrome oxidase displays nonlinear Mi-chaelis-Menten behavior (10, 11); this nonhyperbolic steady-state behavior may be related to proton translocation if thetwo cytochrome oxidase conformational states providing thealternating access to protons on the two sides of the mem-brane are capable of accepting electrons from cytochrome c,albeit with different rates (12). On the other hand, there isgrowing evidence that the cytochrome a3-CuB binuclear siteis involved in molecular events that provide the free energyfor proton pumping (9).Data from various laboratories indicate that the H+/e-

ratio is -1; nonetheless, the stoichiometry ofthe pump underdifferent experimental conditions is under question, given

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

that (for example) values considerably less than unity havebeen observed (3, 13-15), depending on the total number ofturnovers, the absolute turnover rate, and the initial state ofthe enzyme, either pulsed or resting (for definition, see refs.16 and 17). Determination of the H+/e- stoichiometry underdifferent conditions is thought to be of great significancewhen deciding between the two alternative mechanisms ofH+ translocation-i.e., a direct or an indirect type of cou-pling. A decrease ofthe phenomenological H+/e- ratio couldbe due to a proton "leak" (enhanced proton permeability),for a direct type of coupling, or could be due to a proton" slip" (partial uncoupling ofthe H+ pumping machinery), foran indirect type of coupling (14, 18).We have shown (19) that use of stopped-flow spectropho-

tometry to follow the time course of electron transfer and H+translocation allows us to kinetically resolve proton pumpingfrom back-leak, which may jeopardize the analysis of thekinetics and the stoichiometry of the pump; this is likely tohave happened for the early experiments on subunit III-depleted cytochrome oxidase, which was incorrectly thoughtto have lost the proton-pumping activity (ref. 20, but see alsoref. 6). Most of the previous experiments (6) have beencarried out by measuring external acidification by potentio-metry; with this method, it is very difficult to follow the timecourse of proton pumping, especially during the initial turn-overs. On the other hand, overlap of the absorption bands ofthe cytochromes and the pH indicator phenol red (pKa = 7.7)makes it very difficult to objectively analyze the time courseof electron transfer and H+ translocation; the outcome relieson calibrations to set proper isosbestic points for a point-by-point correction of the time-dependent spectral overlaps.Therefore, application of stopped-flow spectrophotometryhas been very limited, and the relatively slow-respondingpotentiometric method has been almost universally em-ployed.

In this study, we have employed transient spectroscopyand singular value decomposition (SVD) analysis to followthe time course of the various spectral components and toassess quantitatively the stoichiometry of the proton pump inCOVs, under selected conditions. A simple kinetic schemefits the data at all wavelengths (500-650 nm) and allows us toprove (i) that the H+/e- stoichiometry (1.01 ± 0.10) isindependent of the total number of turnovers (from 2 to 10turnovers) and of the rate of proton back-leak [increased byvarious concentrations of carbonylcyanide m-chlorophenyl-hydrazone (CCCP)] and (ii) that experiments initiated eitherwith a pulse of reductant (cytochrome c2+) or with a pulse ofoxidant (oxygen) yield the same stoichiometry, proving that

Abbreviations: COV, cytochrome c oxidase vesicle; SVD, singularvalue decomposition; CCCP, carbonylcyanide m-chlorophenylhy-drazone.§To whom reprint requests should be addressed at: Department ofBiochemical Sciences, University of Rome "La Sapienza," Piaz-zale Aldo Moro 5, 00185 Rome, Italy.

5949

Dow

nloa

ded

by g

uest

on

Dec

embe

r 31

, 202

1

5950 Biochemistry: Antonini et al.

resting and pulsed states of cytochrome oxidase are indis-tinguishable in this respect.

MATERIALS AND METHODSCytochrome c oxidase was purified from ox heart as de-scribed by Yonetani (21). The final pellet was dissolved in 0.1M potassium phosphate (pH 7.4), containing 0.5% dodecylmaltoside, at a protein concentration of -0.3 mM (totalheme). Samples were frozen in liquid nitrogen, stored at-70°C, and used within 2 months of preparation. Proteinconcentration was determined spectrophotometrically usinga Ae (reduced-oxidized, at 605 nm) of 22 mM1-cm-l perfunctional unit (aa3).Soybean phospholipids (L-a-phosphatidylcholine type II S

from Sigma) were repurified by acetone/ether fractionationcycles (2). Reconstitution of COVs was achieved by thecholate dialysis method (22); the internal buffer was 0.1 MHepes-KOH (pH 7.3); the external medium was brought to 50AM Hepes-KOH containing 43.6 mM KCI and 46.1 mMsucrose (pH 7.3) by passage through a Sephadex G-25 columnimmediately before the experiment. The phospholipid/protein weight ratio was 50:1 or 70:1. The orientation ofcytochrome oxidase in the reconstituted system was =85%cytochrome a facing outside, as described by Wrigglesworth(23). The maximal stimulation ofcytochrome oxidase activityby ionophores (respiratory control ratio) was 5-8, as mea-sured spectroscopically (24).Cytochrome c type VI was from horse heart, and valino-

mycin and CCCP were from Sigma.The experiments were carried out with two different pro-

tocols:Reductant pulse. COV [1.25 uM (aa3)] in air with 10 ,uM

valinomycin and various concentrations of CCCP (from 0 to2.5 uM) was mixed in the photodiode array stopped-flowapparatus (see below) with phenol red (60 pM) and variousconcentrations of ferrocytochrome c (from 5 to 40 FLM). Inthis type ofexperiment, cytochrome oxidase is (by definition)in the resting state.

Oxidant pulse. COV [1.25 ,uM (aa3)J with 10 pM valino-mycin and various concentrations of CCCP (from 0 to 2.5,uM) was reduced under anaerobic conditions by addition offerrocytochrome c (from 10 to 30 pM) and 0.5 mM sodiumascorbate; after 20 or 30 min to dissipate any transmembranegradient, the solution was mixed with an air-equilibratedsolution of 60 AM phenol red. In this protocol, cytochromeoxidase is in the pulsed state (as confirmed by an enhancedcytochrome c oxidation rate, see Results).

Spectra were collected by using a rapid scanning photo-diode array spectrophotometer (model TN6500, TracorNorthern, Madison, WI) adapted to a Durrum-Gibsonstopped-flow apparatus with a 2-cm light path. The Tracorrapid scanning spectrophotometer acquires 512 or 1024 pho-todiode elements in 5-10 ms and can store up to 62 spectra.The absorption spectra were recorded over a 150-nm range atvarious times after mixing; the time recording mode waslogarithmic to properly cover several decades. Sometimes weobserved a mixing artifact, seen also with COVs plus phenolred in the absence ofcytochrome c, as reported and discussed(19), where necessary corrections were made for this artifact.Data were always analyzed in terms of difference spectra bytaking as the baseline the fully oxidized sample (see Fig. 1).The TN6500 data sets were transferred to a MicroVax 3500

for spectral analysis, by using MATLAB (Math Works, SouthNatick, MA) for computation and graphic procedures. Thedata matrices were built by linking together the spectral timecourses obtained at CCCP concentrations from 0 to 1.25 .M(62 spectra and three to five conditions), adding the spectralacid-base transition of phenol red (recorded in the appara-tus), and reducing the observed wavelengths to 256 by

averaging every four-diode data point in the spectral range of500-650 nm. The resulting arrays (256 x 248 or 256 x 372)were analyzed by an SVD algorithm (25, 26) that transformsthe spectral matrix OD (OD = m diodes x n times) into theproduct of three matrices; i.e., OD = U x S x VT, where Uis a matrix of basis spectra, S is a positive diagonal matrixwith decreasing values, and VT is the time course of thecorresponding Uspectra. The S matrix contains the "singularvalues" of the OD matrix, thus representing the contributionof the corresponding U spectra to the observed OD spectralmatrix (see Fig. 2). The OD changes of the various chro-mophores were thus separated by this procedure, which isessentially based on resolving processes with different tem-poral and spectral behaviors.

Reconstruction of the separate spectral components ispossible by taking the U, S, and V columns of interest. Thespectral change of cytochrome c with time, separated bythose ofphenol red and cytochrome oxidase, can be obtainedby taking only the U column where the cytochrome ccontribution is largest (U2), multiplied by its singular value(S2,2) and its time course (V2). The optical transition moni-toring ApH was calibrated internally by (i) quantitation oftheextent ofalkalinization observed in the presence of saturatingCCCP and assuming a H+/e- ratio of 1.0 for the scalarprocess:

02+4e-+4H+-2H20,

and (ii) correction for the ratio between the external and theinternal plus external (19, 27) buffer power.Data fitting was carried out with the MATLAB variable

minimization procedure and the time courses of the kineticmechanism shown in Scheme I (see below) were calculatedby numerical integration of the ordinary differential equa-tions that describe the mechanism.

RESULTSFig. 1 shows 62 difference spectra (baseline, fully oxidized atinfinite time) recorded after mixing oxidized COVs contain-ing 10 AM valinomycin and 0.1 I&M CCCP with a solutioncontaining reduced cytochrome c and phenol red in thepresence of air; the corresponding SVD analysis is depictedin Fig. 2. The time courses of absorbance change at 550 nm(representing mainly cytochrome c oxidation) and at 556.6nm (representing mainly phenol red acidification and close tobeing an isosbestic point for cytochrome c) are shown in Fig.3 Upper, traces *. The other traces in Fig. 3 are describedbelow. It is clear that quantitation of the extent of cy-tochrome c oxidation and phenol red protonation, whichallows calculation of the vectorial H+/e- stoichiometry, isdependent on how the partially overlapping spectral transi-tions are separated. For a two-component system, this maybe achieved if (i) an isosbestic point for either opticaltransition can be found (i.e., 556.6 nm) and (ii) an appropriateextinction coefficient at a selected wavelength is available. Ingeneral, and specifically here, the system is much morecomplex, containing more than two chromophores with non-synchronous time dependencies. Thus, the separation ofchromophores must rely on some bias. The SVD technique,on the other hand, is specially suited for chromophoreseparation if the behavior of each component follows adifferent time course (see ref. 26). Therefore, unbiased sep-aration of the various spectral components was obtained bySVD.The SVD analysis of the experiment, depicted in Fig. 1,

with the separation of the spectral contribution of phenol red(UW) and cytochrome c (U2), is shown in Fig. 2. Notice thatthe two spectral components are almost completely repre-sented in U1 and U2; U3, though containing an extremely

Proc. Natl. Acad. Sci. USA 90 (1993)

Dow

nloa

ded

by g

uest

on

Dec

embe

r 31

, 202

1

Proc. Natl. Acad. Sci. USA 90 (1993) 5951

0.2

00

A~~~~~~

-0.2 L 5 6480 540 600 660

A (nm)

FIG. 1. Sixty-two time-resolved difference spectra obtained aftermixing COVs (1.25 p,M aa3) containing valinomycin (10 AM) andCCCP (0.1 ,uM) with phenol red (60 ,uM) and ferrocytochrome c (20,AM), in air at 20°C (2-cm light path). Buffer internal to COVs was 0.1M Hepes-KOH (pH 7.3); buffer external to COVs was 50 juMHepes-KOH/43.6 mM KCl/46.1 mM sucrose, pH 7.3. (A) Spectraare superimposed. (B) Spectra are plotted on a logarithmic timescale.

small residual contribution of cytochrome c and phenol red,brings out the spectral contribution of cytochrome oxidase,the third component of this system (see below); U4 containsinformation not interpretable at present as well as noise.Examination of the V columns that represent the time de-

A 0.1| U1 U2

-0.2

0

450 650 450 650

U3 U0 0

'n -0.2 -0.2X -2450 650 450 650

A (nm)

B °t0-<

Time (s)

FiG. 2. SVD analysis of the spectra reported in Fig. 1. (A) Ucolumns 1-4 (S values: Si, 129; S2, 30.7; S3, 0.64; S4, 0.32). (B) Vcolumns 1-4.

0.1

0 40

4 0Time (s)

FIG. 3. Time course of cytochrome c oxidation (Left) and protonpumping (Right) by COVs containing 5 ,uM valinomycin and 0, 0.05,0.15, 0.35, and 1.25 ,uM CCCP (after mixing). The asterisk-labeledtraces are taken from the experiment shown in Fig. 1. The SVDanalyses of these traces are shown in Fig. 2. Other conditions are asin Fig. 1. (Upper) Observed time courses at 550 and 556.6 nm areplotted from the data of Fig. 1. (Lower) Same data after SVD analysisand spectral reconstruction. Data are from Figs. 1 and 2.

pendencies of the corresponding U columns (Fig. 2B) yieldsthe following results. In V1 we observe the transient acidifi-cation and subsequent alkalinization of phenol red (U1) seenwhen CCCP, added to increase the rate constant ofthe protonback-leak, was not saturating. In V2 the time course ofcytochrome c oxidation is seen. These experiments (repeatedat several concentrations ofCCCP) allow efficient separationof the chromophores by the SVD algorithm, given that in theabsence of the protonophore cytochrome c oxidation andphenol red protonation are synchronous (19) and thus notseparated by SVD (results not shown). V3 shows the timecourse of the corresponding U3 column, which representsmainly cytochrome oxidase but also contains small contri-butions from cytochrome c and phenol red; the time coursecorresponds to oxidation of cytochrome a and reflects elec-tron redistribution in a fraction of the enzyme molecules(s20% of the total) when oxidation ofcytochrome c is almostcomplete (28). This process is "catalyzed" by cytochrome cacting as an electron shuttle and seems to be associated withproton consumption. Thus, cytochrome c (-1.5% of thetotal) and phenol red (<1%) signals share the same timecourse in V3 as cytochrome oxidase and are represented inthe same U3 column. The slow phase in V4 is not interpretableat present, but accounts for at most 0.2% of the totalabsorbance change of cytochrome c.

It is important to point out that all calculations regardingthe H+/e- stoichiometry and the fit to the kinetic schemehave been carried out by reconstruction ofthe USVr columns1 and 2, which account for >98% of the total spectraltransition of phenol red and cytochrome c, respectively.Experiments similar to those depicted in Fig. 1 were

carried out at various CCCP concentrations and are shown inFig. 3 at selected wavelengths. Fig. 3 Upper shows actual rawdata, as obtained from the experiment before chromophoreseparation. Fig. 3 Lower shows the results obtained by theSVD analysis as shown in Fig. 2. The reconstructed timecourses shown in Fig. 3 Lower were obtained by multiplyingany given basis spectrum (Ucolumn) by its singular value andby its corresponding transposed V column. This yields amatrix containing the time dependence of the single spectralcomponent of choice. When cytochrome c is separated fromphenol red in the wavelength and time domains, it is possibleto evaluate the apparent H+/e- stoichiometry. Notice thatincreasing the CCCP concentration has a dramatic effect onthe H+ signal (Fig. 3 Lower right) since the proton perme-ability is severely increased. On the contrary, the oxidationofcytochrome c (Fig. 3 Lower left) is only slightly affected by

Biochemistry: Antonini et aL

Dow

nloa

ded

by g

uest

on

Dec

embe

r 31

, 202

1

5952 Biochemistry: Antonini et al.

a a0-1

0.2 \

c

-2 ()

0.4

-2 0

log tinme (s)

FIG. 4. Best fit of the kinetics of cytochrome c oxidation andproton pumping by COVs, according to Scheme I. The figure showsthe time courses at 556.6 nm (Upper) (phenol red) and 550 nm(Lower) (cytochrome c) carried out with COVs (0.6 .M aa3), in thepresence of valinomycin (5 ,uM) and 0, 0.05, 0.15, 0.35, and 1.25 uMCCCP (all concentrations after mixing) at 20°C (2-cm light path). Thedots are the experimental points; the continuous lines are the best fitto Scheme I, see Results, with the parameters shown in Table 1. (Aand C) Ferrocytochrome c is 9.5 ,uM. (B and D) Ferrocytochrome c

is 17.5 ,uM.

.2 J0. 2

-2 0 -2 0log time (s)

FIG. 5. Time courses at 556.6 nm (Upper) (phenol red) and 550nm (Lower) (cytochrome c) for oxidant and reductant pulse exper-iments, carried out with COVs (0.6 ,uM aa3) in the presence ofvalinomycin (5 ,uM) and 0, 0.05, and 1.05 ,uM CCCP (all concentra-tions after mixing) at 20°C (2-cm light path). The time courses areobtained after SVD analysis and spectral reconstruction. The dotsare the experimental points; the continuous lines are the best fit toScheme I, with the parameters shown in Table 2. (A and C)Reductant pulse experiment. Ferrocytochrome c is 8 ,uM. (B and D)Oxidant pulse experiment. Ferrocytochrome c is 7 ,uM.

the addition of the protonophore (as also deduced from thefitting procedures shown below), since as reported (29, 30),valinomycin alone in this condition accounts for release of70-80%o of the respiratory control ratio. Fig. 4 and Table 1show the reconstructed time courses at 550 nm and 556.6 nmfor experiments carried out at two cytochrome c concentra-tions and various CCCP concentrations. The time scale islogarithmic for the best representation of both the early (H+pumping) and late (H+ back-leak) events. The continuouslines are the best fit to the following simple scheme:

cyt c2 + cyt c3+ (electron transfer)

+kiH+ -. H+0t (vectorial process)

Hi- H20 (scalar process)

kblH+ - H,+(back-leak)Scheme I

This mechanism contains the following features: (i) Theoxidation of cytochrome c follows exponential behavior, k1representing the rate-determining step of the catalytic cycleand being faster in an oxidant pulse experiment (leading topulsed cytochrome oxidase). (ii) The time course of H+pumping is synchronous with cytochrome c oxidation, asshown by Sarti et al. (19). (iii) The H+ back-leak, very slowin the presence of valinomycin alone, is increased by adding

Table 1. Parameters for Fig. 4

cyt c, ,uM H+/e ratio CCCP, ,M kl, s-1 kbl, s-

9.5 0.95 0 3.71 0.0320.05 3.37 0.550.15 3.01 1.60.35 3.08 5.021.25 3.38 >100

17.5 0.95 0 1.23 0.0420.05 1.27 0.760.15 1.41 2.870.35 1.54 10.051.25 1.93 >100

cyt c, ferrocytochrome c.

CCCP at various concentrations and is assumed to be thesame at any given CCCP concentration in the reductant andoxidant pulse experiments. (iv) In the presence of saturatingCCCP, the total buffer power (inside + outside) is higher thanin the absence of CCCP (i.e., only outside) (27). Therefore,for a given H+ pulse, the spectral transition of phenol red issmaller in the presence ofCCCP (80% of that observed in theabsence of CCCP); (v) the net consumption ofH+ inside thevesicles (scalar reaction) is 1.0 per electron.The variables used in the fit ofthe experimental data are as

follows: (i) the stoichiometry of the pump (H+/e- ratio), (ii)the rate ofcytochrome c oxidation (kl), and (iii) the rate ofH+back-leak (kbl). The fit of the data shown in Fig. 4 and Table1 to the kinetic scheme illustrated above is satisfactory (withthe parameters given). It leads to an unequivocal determina-tion of the H+/e- stoichiometry of the pump as a function ofthe number ofturnovers and ofthe actual value ofApH (giventhat A4i = 0 because valinomycin is saturating). The H+/e-stoichiometry in the presence of nonsaturating CCCP isunequivocally determined by the constraints used in the fit(i.e., initial and final absorbance values and kl, as determinedby the corresponding rate of cytochrome c oxidation). Ofcourse, with saturating CCCP, when no transient acidifica-tion could be observed, the H+/e- stoichiometry is essen-tially undetermined, and in the latter case, a H+/e- of 1 is oneofthe possible solutions ofthe fit. However, we rely on fittingthe family of curves in Fig. 4, where only the time course atthe highest CCCP concentration displays no significant spec-tral contribution of the vectorial process.

Fig. 5 and Table 2 show the observed time courses offerrocytochrome c oxidation and phenol red protonation/deprotonation (after SVD analysis and reconstruction) for anoxidant pulse and a reductant pulse experiment. The best fitof data in Fig. 5 shows that (i) the rate of cytochrome c

Table 2. Parameters for Fig. 5cyt c, ,uM H+/e- ratio CCCP, ,uM kl, s-1 kbl, S-'

8 1.07 0 2.10 0.010.05 2.00 0.391.05 2.05 >300

7 0.96 0 3.79 0.010.05 3.89 0.391.05 4.01 >300

cyt c, ferrocytochrome c.

Proc. Natl. Acad. Sci. USA 90 (1993)

Dow

nloa

ded

by g

uest

on

Dec

embe

r 31

, 202

1

Proc. Natl. Acad. Sci. USA 90 (1993) 5953

oxidation is 2-fold higher in the oxidant pulse experiment (asexpected since cytochrome oxidase is in the pulsed state) andincreases slightly with increasing concentration ofCCCP andthat (ii) the H+/e- ratio is, however, 0.96 and 1.07 for thepulsed and resting enzyme, respectively, thus showing iden-tical stoichiometry.

DISCUSSIONThe use ofrapid scanning spectrophotometry and analysis bythe SVD procedure have been employed to reinvestigate theH+ pumping properties of COVs. The remarkable advantageof this technique to separate the time courses and amplitudesof the three chromophores (cytochrome c, phenol red, andcytochrome oxidase) is shown in Figs. 2-4. Analysis allowsobservation and quantitation, in a single shot, of the kineticsof cytochrome c oxidation and medium acidification (oralkalinization). Fits of the data with a kinetic scheme forcytochrome c oxidation and H+ pumping, taking into accountthe H+ back-leak rate, allow objective quantitation of thetime courses of electron transfer, H+ translocation, and theH+/e- ratio under various conditions.

In these experiments, the stoichiometry of the pump wasdetermined to be 1.01 ± 0.10 under all conditions explored,and the optical changes related to ferrocytochrome c oxida-tion and external medium acidification are synchronous at theresolution level used (5-10 ms), in agreement with Sarti et al.(19). The independence of the H+/e- ratio on CCCP con-centration, and thus on the rate of back-leak, emerges fromthe fit. This implies that the absolute value of the protonchemical gradient does not affect the proton-pumping stoi-chiometry, provided the electrical component of the gradientis collapsed by valinomycin; this should be contrasted withthe fact that in the presence of a fully developed membranepotential, proton translocation by cytochrome oxidase hasnot been observed. The cubic model we proposed (30) for thecontrol of cytochrome oxidase activities predicts that theelectrical component of the gradient controls the rate ofelectron entry and of the internal electron transfer and theproton-pumping stoichiometry, by freezing the oxidase in a"slipping state." Since in the present experiments, theH+/e- is, within the errors, the same for pulsed and restingCOVs, it follows that neither of these states of the enzyme isa slipping state.The best fit of the time-resolved spectroscopic data to

Scheme I shows that proton pumping by cytochrome oxidasehas a constant stoichiometry over the whole time range andat various concentrations of cytochrome c (from 5 to 25 ttM;i.e., from 2 to 10 turnovers). This indicates that the H+/e-ratio is the same from the first turnover onward; i.e., theproton-pumping machinery is fully active after the initialfour-electron reduction of the enzyme.A variable-pump stoichiometry has been reported by sev-

eral groups under different experimental conditions. As anexample, the H+/e- stoichiometry was reported to be 0.8 ina reductant pulse experiment but dropped to 0.3 in an oxidantpulse experiment (13). Given that the two experimentalprotocols yield, respectively, resting and pulsed oxidase (16,17), it has been argued that only the resting state of theenzyme is a competent proton pump with high efficiency.Because pulsed oxidase is probably the only functional stateof the enzyme in vivo, this result may have had someconsequence in assessing the significance and the stoichiom-etry of the pump in vivo. The evidence presented aboveshows convincingly that the H+/e- ratio is the same for theresting and pulsed enzyme and implies that the variablestoichiometry must be attributed to other presently unknownexperimental variables.

In conclusion, we have demonstrated that using transientspectroscopy and SVD analysis the proton-pumping activityof cytochrome oxidase reconstituted into vescicles is wellresolved and, thereby, amenable to quantitative kinetic anal-ysis. Accurate quantitation of the kinetics and stoichiometryof proton pumping by cytochrome oxidase can thus resolvesome experimental discrepancies related to variable H+/e-ratios; in particular we have shown that pulsed and restingoxidase are both able to pump protons with the same stoi-chiometry.

We express our appreciation to Drs. W. A. Eaton and E. Henry(National Institutes of Health) for introducing us to SVD analysis,and to E. D'Itri, A. Giuffre, and F. Nicoletti for their skillfulcontribution in preparation and characterization ofCOVs. This workwas partially supported by Ministero dell'Universita e della RicercciScientifica e Tecnologica (40%o Live protein), by Consiglio Nazionaledelle Ricerche (Target Project on Biotechnology and Bioinstrumen-tation) of Italy, and by Commission of the European CommunityGrant SC1-CT91-0698 (CAESAR Project).

1. Wikstrom, M. K. F. (1977) Nature (London) 266, 271-273.2. Krab, K. & Wikstr6m, M. K. F. (1978) Biochim. Biophys. Acta

504, 200-214.3. Casey, R. P., Chappel, J. B. & Azzi, A. (1979) Biochem. J. 182,

149-156.4. Sigel, E. & Carafoli, E. (1980) Eur. J. Biochem. 111, 299-306.5. Thelen, M., O'Shea, P., Petrone, G. & Azzi, A. (1985) J. Biol.

Chem. 260, 3626-3631.6. Brunori, M., Antonini, G., Malatesta, F., Sarti, P. & Wilson,

M. T. (1987) Eur. J. Biochem. 169, 1-8.7. Shapleigh, J. P., Hosler, J. P., Tecklenburg, M. M. J., Kim,

Y., Babcock, G. T., Gennis, R. B. & Ferguson-Miller, S.(1992) Proc. Natl. Acad. Sci. USA 89, 4786-4790.

8. Steffens, G. J. & Buse, G. (1979) Hoppe-Seyler's Z. Physiol.Chem. 360, 613-619.

9. Babcock, G. T. & Wikstrom, M. (1992) Nature (London) 356,301-309.

10. Ferguson-Miller, S., Brautigan, D. L. & Margoliash, E. (1976)J. Biol. Chem. 251, 1104-1115.

11. Garber, E. A. E. & Margoliash, E. (1990) Biochim. Biophys.Acta 1015, 279-287.

12. Brzezinski, P. & Malmstrom, B. G. (1986) Proc. Natl. Acad.Sci. USA 83, 4282-4286.

13. Papa, S., Capitanio, N. & De Nitto, E. (1987) Eur. J. Biochem.164, 507-516.

14. Capitanio, N., Capitanio, G., De Nitto, E., Villani, G. & Papa,S. (1991) FEBS Lett. 288, 179-182.

15. Solioz, M., Carafoli, E. & Ludwig, B. (1982) J. Biol. Chem. 257,1579-1582.

16. Antonini, E., Brunori, M., Colosimo, A., Greenwood, C. &Wilson, M. T. (1977) Proc. Natl. Acad. Sci. USA 74, 3128-3132.

17. Brunori, M., Colosimo, A., Rainoni, G., Wilson, M. T. &Antonini, E. (1979) J. Biol. Chem. 254, 10769-10774.

18. Murphy, M. P. (1989) Biochim. Biophys. Acta 977, 123-141.19. Sarti, P., Jones, M. G., Antonini, G., Malatesta, F., Colosimo,

A., Wilson, M. T. & Brunori, M. (1985) Proc. Natl. Acad. Sci.USA 82, 4876-4880.

20. Pentilla, T. (1983) Eur. J. Biochem. 133, 355-361.21. Yonetani, T. (1961) J. Biol. Chem. 236, 1680-1688.22. Racker, E. (1979) Methods Enzymol. 55, 699-711.23. Wrigglesworth, J. M. (1985) J. Inorg. Biochem. 23, 311-316.24. Sarti, P., Colosimo, A., Brunori, M., Wilson, M. T. & Anto-

nini, E. (1983) Biochem. J. 209, 81-89.25. Golub, G. H. & Reinsch, C. (1970) Numer. Math. 14, 403-420.26. Henry, E. R. & Hofrickter, J. (1992) Methods Enzymol. 210,

129-192.27. Sarti, P., Malatesta, F., Antonini, G., Colosimo, A. & Brunori,

M. (1985) Biochim. Biophys. Acta 809, 39-43.28. Antonini, E., Brunori, M., Greenwood, C. & Malmstrom,

B. G. (1970) Nature (London) 228, 936-937.29. Brunori, M., Sarti, P., Colosimo, A., Antonini, G., Malatesta,

F., Jones, M. G. & Wilson, M. T. (1985) EMBO J. 4, 2365-2368.

30. Antonini, G., Malatesta, F., Sarti, P. & Brunori, M. (1991) J.Biol. Chem. 266, 13193-13202.

Biochemistry: Antonini et al.

Dow

nloa

ded

by g

uest

on

Dec

embe

r 31

, 202

1