tetramethyl-p-phenylenediamine dihydrochloride. Following incuba- tion under Ar, enzyme, inhibitor (azide or cyanide), mediators, and. NADH were introduced.
THEJOURNAL
(c) 1984 by
OF BIOLOGICAL CHEMISTRY The American Society of Biological Chemists, Inc.
Vol. 259. No. 24, Issue of December 25, pp. 15094-15099,1984 Printed in U.S.A.
Metal Site Cooperativity within Cytochrome Oxidase* (Received for Publication, March 15, 1984)
Gay Goodman$ From the DeDartment of Biochemistrv and BioDhvsics. School of Medicine G4, University of Pennsylvania, Philadelphia: Pennsylvdnia 19104 ” &
.
.
I
Low temperature (9-15 K) EPR of isoIated bovine heart cytochrome oxidase titrated potentiometrically in the presence of azide reveals the formation of two distinct species of low-spin cytochrome a3(III)-azide which differ in redox properties and g values. Both species are formed with characteristic midpoint potentials during the course of oxidative titration and disappear at higher potentials. The signal appearing at lower potential has principalg values 2.88, 2.19, and 1.64;that appearing at higher potential has g values 2.77, 2.18, and 1.74. A good fit to the experimental data (per centof cytochrome present in a given paraobmagnetic state versus oxidationpotential)was tained with a model whereby the g , = 2.88 species arises from cytochromea&II)-azide with cytochrome a reduced, which is converted to theg, = 2.77 species upon oxidation of cytochrome a. Potentiometric titrationof cytochrome oxidase in the presence of cyanide produces two low-spin heme EPR signalsattributableto cytochrome a3(III)-cyanide which are incompletely resolved, but are distinguishable nonetheless. The low-potential signal has peak amplitude at g , = 3.63 and a long high-field tail; this resonance hasbeen seen by other workers in the partially reduced enzyme(DerVartanian, D. V., Lee, I. Y., Slater, E. C., and van Gelder, B. F. (1974)Biochim. Biophys. Acta 347,321-327). The high-potential signal is much more symmetric aboutits peak amplitude, which is at-10 G higher field withg, = 3.61.As with the azide complex, the titration behavior in the presence of 2 mM KCN is adequately simulatedby assuming that the appearanceof the twospecies is a function of the oxidationstate of cytochrome a. Like thea3-azide signals, the as-cyanide signals disappear upon further oxidation with some characteristic midpoint potential. If the disappearance of the a3ligand signals with increasing potentialis assumed to be the result of antiferromagnetic (or ferromagnetic) coupling of a3(III)(S = %) to CuB(II) (S = 1 4 , then cooperativity between cytochrome a and CUBis implied. The data are consistent with thehypothesis that oxidation of cytochrome a raises the midpoint potential of CUBby 55 f 10 mV.
It is well known that cytochrome oxidase (ferrocytochrome coxygen oxidoreductase, EC 1.9.3.1), the terminalenzyme of the mitochondrial electron transfer chain, is responsible for * This work was supported by Grant GM-25052from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore he hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 3 Present address: Department of Molecular Neurogenetics, E. K. Shriver Center, 200 Trapelo Rd., Waltham, MA 02254.
the concerted transfer of four electrons to molecular oxygen in an energy-yielding redox reaction coupled to ATP generation. The enzyme has been described as a redox-linked proton pump (Wikstrom and Krab, 1979). Four metal ions are involved in the electron transfer reaction: the two heme iron atoms of cytochromes a and a3 and two copper atoms, CuA and CUB(see Wikstrom et al. (1981) for review). Cytochrome a3 is the ligand-binding heme; historically, it was defined as such (Lemberg, 1969). The binding of azide to cytochrome oxidase has been studied extensively via optical absorbance spectroscopy (Wilson, 1967;Wilson et al., 1972) and, to a lesser extent, by EPR spectroscopy (van Gelder et al., 1967; van Gelder and Beinert, 1969; Wilson and Leigh, 1972; Wilson et al., 1976). The EPR studies showed that a low-spin ferric heme signal with g, = 2.9 is associated with azide binding. This signal appears and then disappears with increasing oxidation potential (Wilson and Leigh, 1972; Wilson et al., 1976), analogously to thehighspin (and/or low-spin) signals that areobserved in the absence of inhibitor ligands (See Erecinska and Wilson (1980) for a review of cytochrome oxidase inhibitor interactions.) Cyanide forms spectrally distinct compounds with both oxidized and reduced cytochrome oxidase (Wainio, 1955), but Keilin and Hartree (1939) showed that the ferric form of a3 is stabilized by cyanide. One cyanide molecule is bound per au3 unit (van Buuren et al., 1972), as is also the case with azide (Wever et al., 1973). A broad low-spin ferric heme resonance at g = 3.6 is seen in the EPR spectrum of the partially reduced enzyme when cyanide is present(DerVartanian et al., 1974; Johnson et al., 1981); as with azide, this signal is not observed in the aerobic state, although cyanide remains bound. In thiswork, EPR observations on cyanide and azide binding are extended. For the first time, an EPR study of the redox behavior of the cyanide compound is described. MATERIALS AND METHODS
Enzyme Purification-Cytochrome oxidase was isolated from frozen beef heart mitochondria by the method of Yu et al. (1975)or by the method of Sun et al. (1968)followed by solubilization in cholate and by ammonium sulfate precipitation of impurities according to modifications of Capaldi and Hayashi (1972). The enzyme was precipitated several times frombuffer containing 1mM EDTA to remove extraneous copper. The final pellet was taken up in 0.1% polyoxyethylenesorhitan monooleate (Tween-80) and dialyzed against 0.1 M potassium phosphate (pH7.2),plus 1 mM EDTA, to remove cholate, ammonium sulfate, and copper. Beforeuse, the enzyme was redialyzed against 50 mM potassium phosphate, 50 mM potassium chloride, and 50 m M HEPES.’ The enzyme was frozen in small aliquots in liquid N2 and stored a t -80 “C. Total protein was assayed by the Bio-Rad method, and hemea was measured as the air-oxidized minus dithionite-reduced absorbancedifference a t 605 - 630 nm, assuminga value
’ The abbreviationsused are: HEPES,4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid mW, milliwatt.
15094
Cooperatiuity Site Cytochrome Metal Oxidase:
15095
of 13.5 m"' cm" for the extinction coefficient. Absolute and difference spectra in the region between 650 and 400 nm indicated that contamination was 4 % for cytochromeb and below detectability for cytochrome c. Typical values for hemcprotein were 6-9 nmol/mg. The enzyme was judged to be pure enough for the present studies when 1) enough lipid had been removed such that the enzyme was optically clear (i.e. dispersed) in0.1 M potassium phosphate (pH7.2), in the presence of 0.1% Tween-80; and 2) the EPR spectrum in tbeg = 2 region a t 10 K showed no sign of iron-sulfur proteins or any other extraneous signals when the enzyme was reduced with /3-NADH in the presence of phenazine methosulfate followed by air oxidation or anaerobic oxidation with ferricyanide to potentials between -100 and 400 mV. Sample Preparation-Redox titrations were performed under an atmosphere of Argon (Airco grade 5) in the presence of 0.2, 2, or 20 r n potassium ~ azide or 2 mM potassiumcyanide(preparedfresh daily). The calomel and platinum electrodes were calibrated before each use with 10 mM 1:l potassium ferricyanide/ferrocyanide in 0.1 M potassiumphosphate,pH 7.0. Themidpointpotential of this standard was taken to be 425 mV a t 25 "C plus 2.4 mV/"C below 25 "C (O'Reilly, 1973). Titrations were performed a t room temperature: the potentials of the redox components were not corrected for the 2-3 "C day-to-day variation. The titration medium contained 50 p M cytochrome oxidase plus 110 mM glucose, 1.7 units of glucose oxidase (Sigma, grade IV), 2.5 gg of catalase (Sigma),7 nM superoxide dismutase from yeast (gift of Dr. H. J. Forman, Physiology Department, University of Pennsylvania),andthe following concentrations of redox mediators:7 p M phenazine methosulfate, 8 p~ 1,2-napthoquinone 4-sulfonic acid, 30 p~ 2,3,5,6-tetramethyl-p-phenylenediamine, and 7 p~ N,N,N',N'tetramethyl-p-phenylenediaminedihydrochloride. Following incubation under Ar, enzyme, inhibitor (azide or cyanide), mediators, and NADH were introduced. The enzyme was titrated with microliter quantities of 0.1 M potassium ferricyanide in 0.1 M potassium phosphate, pH 7.0. Samples were transferred anaerobically to 3-mm O.D. quartz EPR tubes,frozen in 5:1 isopentane:methylcyclohexane at the freezing point, and removed rapidly to liquid nitrogen. Oriented Multilayers-Membraneous cytochrome oxidase prepared according to Sun etal. (1968) was the generous gift of Dr. T. G. Frey, Department of Biochemistry and Biophysics, University of Pennsylvania. Partially oriented multilayers were formed by drying the enzyme on strips of Mylar previously coated with nitrocellulose.
pH 7.2
g=2.88
9 *6 I
I
Electron Paramagnetic Resonance-EPR spectra were taken on a Varian E-109 spectrometer equipped with a Varian E-231 T E 102 rectangular cavity;modulationfrequency and amplitude were 100 kHz and 16 G, unless stated otherwise. An Air Products LTD-3-110 flowing helium cryostat cooled the quartz Dewar; the temperature was monitored continuously with a calibrated carbon resistor placed directly below the sample. Temperature was regulated with Johnson a Foundation thermostat. Simulation of the EPR powder line shapes utilized the transition probability correction of Aasa and Vanngird (1975). The method involves computation of a stick spectrum weighted average from the transition probabilities for a spherical distribution of g values. The derivation of theoretical Nernstian curves for multi-component one-electron systems followed the principals of Clark (1960); the E,,, or midpoint potentialof an oxido/reductive titration isdefined analogously to the pK €or an acidbase titration. RESULTS AND DISCUSSION
Azide Complex-EPR spectra of cytochromeoxidase titrated oxidatively in thepresence of 2 mM azide are shown in Fig. 1. The spin concentration represented by the high-spin heme signal a t g = 6 is insignificant at all potentials, indicating that azide binding to a33+is essentially complete. Earlier optical absorbance studies, when interpreted within the context of the heme-hemeredox interaction model (Nicholls and Petersen, 1974; Wikstrom et al., 1976) provided evidence that azide binds to oxidized cytochrome u3 with dissociation conof stant KO = 0.25 mM at pH 7.2 (based on the concentration free azide ion), and that reduced cytochrome a3 has a much lower affinity for azide (Wilson et al., 1972). The active inhibitory form for both cyanide and azide is the acid (i.e. HCN, HN3) rather than the anion (Stannard and Horecker, 1948). Since hydrazoic acid has pK = 4.7, the actual KO for HN3 binding is computed to be 6 ~ L Mat pH7.2 from the data of Wilson et al. (1972). Even though the reaction mechanism favors the acid, the ionized form is more likely to be liganded to ferriciron. Forthisreason,the ligand complexes
+ 2 m M oxide A
g'2.19 I
It)
\
04 mV
131mV
20BmV
I 991.64
1
242 m V
g 9 3.a
"
-
3
4
3
m
"
' 9 * 2.77
273mV
305 mV
I
911.74
800
I000
2800
3800
Goum
A
I g.1.45
4000
2000
2400
Gourr
FIG. 1. Low-temperature EPR spectra of cytochrome oxidase at pH 7.2, with 2 mM azide present. Redox titration performed anaerobically with ferricyanide in the presence of redox mediators; enzyme reduced initially with NADH/phenazine methosulfate. (See "Materials and Methods.") The signals a t g = 6 and g = 4 correspond to high-spin ferric a3 and non-heme iron, respectively; each represents insignificant intensity with respect t,o heme a concentration. The strong,nearly axially symmetric resonance a t g = 2 is due to CUA.The three low-spin ferric heme signals are discussed in the text. EPR conditions: temperature, 15 K; microwave power, 5 mW; modulation frequency, 100 kHz; modulation amplitude, 16 G; time constant, 0.25 s; scanning rate, 33 G/s; microwave frequency, 9.134 GHz.
395mV
Cytochrome Oxidase: Metal Site Cooperativity
15096
discussed here will be referred to arbitrarily as a33+-N3 and a:i'+-CN. In Fig. 1, three distinct low-spin heme signals appear at different redox potentials. At 208 mV, the major heme signal present has principal g values 2.88, 2.19, and 1.64; this is an azide-bound form of ferric cytochrome a3. At higher potentials, thissignal diminishes in intensity as the other two heme signals increase in apparent conjunction with one another. These are due to cytochrome a, with g values 3.0,2.2, and 1.5, -N~ with g values and a second cytochrome u ~ ~ + resonance, 2.77, 2.18, and 1.74. Previous EPR measurements of cytochrome oxidase partially oxidized in the presence of azide discerned only one a:>'+-azideresonance at g = 2.9 (van Gelderet al., 1967; Wilson and Leigh, 1972; Wilson et al., 1976; Erecinska et al., 1979). The earlier spectra were taken atliquid nitrogen temperature (van Gelder et al., 1967). At 15 K, the lines arenarrower, and hence, the resolutiongreater. However, van Gelder and Beinert (1969), working at 80 K with concentrated (0.5 mM) enzyme, noted the presence of a minor signal at g, = 1.76 with its low-field counterpart (gJ visible only at the beginning of reductive titration. Of the spectra taken near liquid helium temperature, two of three studies were conducted with mitochondria or submitochondrial particles (Wilson and Leigh, 1972; Wilson et al., 1976) in which the limited a a 3 concentration ensured poor signal-to-noise. It is significant that in the remaining study, on isolated cytochrome oxidase in oriented multilayers, a single, narrower resonance line appears at g = 2.9 when cytochrome a is reduced (Erecinska et al., 1979); no spectrum is shown for the case when cytochrome a is in the ferric state. In order to simulate the redox behavior of the various metal ion centers, itwas necessary to quantitate the a3-azide signals relative to cytochrome a. Fig. 2A shows the low-spin heme region of an EPR spectrumof cytochrome oxidase titrated to 320 mV in the presence of 20 mM azide. The microwave power (0.8 mW) was chosen such that none of the signals were saturated at the measuring temperature(10 K). The cytochrome a g, signal was simulated as the superposition of two resonance lineshaving principal g values of 3.00, 2.21, and 1.48 and of 3.05,2.21, and 1.45, in the ratio 0.35:0.65, both with line width 42 G (Fig. 2B). These two components contribute to the total intensity in approximately the same
proportion throughout the course of the titration, enabling one to treat the peak amplitude as though wereitarising from a single resonance line. The existence of multiple forms of cytochrome a has been notedmanytimes before (e.g. see Wever et al. (1977)), butnever adequately explained. In Fig. 3, amplitudes of the g = 3, g = 2.88, and g = 2.77 signals are plottedversus oxidation potential for cytochrome oxidase titrated at pH 7.2 in the presence of 2 mM (A) or 20 mM ( B ) azide. The lines are theoreticalcurves obtained with the model depicted in Fig. 4. The curves through theg = 2.88 and g = 2.77 points are the simulated u * + u ~ ~ +and - L a3+a33+-L species, respectively, when CUBis reduced. The g = 3 curve is the sum of the three species in which a is in theferric state. The E , parameters used in the simulations are given in the A
L
2100
2300
100%
-
2500
Gauss
FIG. 2 . Simulation of low-spin aa3+-azideand as+EPR signals. A, EPR spectrum of cytochrome oxidase near g = 3 at pH 7.2, with 20 mM azide present; redox potential, 320 mV. EPR conditions: temperature, 10 K; microwave power, 0.8 mW; modulation frequency, 100 kHz; modulation amplitude, 20 G; time constant, 0.25 s; scanning rate, 8.3 C/s; microwave frequency, 9.321 GHz. B, simulated EPR spectrum of the g, region of four low-spin heme signals in the ratio 1.00:0.53:0.96:0.99 with g,, g,, g, values 3.05,2.21,1.45;3.00,2.21, 1.48; 2.88,2.19,1.64; and 2.77,2.18, 1.74, respectively; all g, line widths, 42 G.
E h (mV)
.,
FIG. 3. Amplitudes of cytochrome a,cytochrome ag3+-azide, and CUAEPR signals versus redox potential at pH 7.2. Samples taken during redox titrations in the presence of 2 mM ( A and C) or 20 mM azide ( B ) ;0, g = 3.0; 0,g = 2.88, 0,g = 2.77, CUA.A and B, theoretical Nernst curves for one-electron oxidation derived using the model of Fig. 4. E,,, (la, lb, 2, 3a, 3b) (mV): 200, 220, 285, 325, 375 ( A ) ;115, 220, 305, 330, 390 ( B ) .C, noninteracting Nernst curve for n = 1with E, = 240 mV.
Cytochrome Oxidase: MetalSite Cooperativity
15097
zuki and Wilson, 1971) have been measured for isolated cytochrome oxidase titrated oxidoreductively at room temperL L ature with 0.5 atm of carbon monoxide present. Bothmethods yield the same n = 1Nernstian dependence on redox potential, with E , = 245 k 10 mV at pH 7.2. The dependence of the midpoint potential of a,-azide on azide concentration as determined by low-temperature EPR spectroscopy is also in L L good agreement with that found by room-temperature optical FIG. 4. Model of redox relationships in the presence of an inhibitor ligand. For simplicity, ligand L is depicted as binding only methods. Wilson et al. (1972) measured a heme signal comto ferric as. This feature is not essential to the arguments developed prising 12-26%of the total 605 - 630 nm transition and below. Only a, as, and CUBare considered. See text for details. The found the midpoint potential to be 185, 155, and 105 mV in species u'+u~~+-L Cug and a3+aS3+-L Cug are plotted separately in Fig. the presence of 1.0, 2.7, and 20 mM azide, respectively. (See 3, A and B, and in the dashed curues of Fig. 6. also Wikstrom et al. (1976) for interpretation of these data.) By comparison, the low-temperature EPR dataof the present legend to Fig. 3. In the 20 mM azide simulation, the curves study are fit with E,,, values of 200 mV for 2 mM azide and are insensitive to the choice of E m l b , for Emla< E m l b
