Electron Spin Echo Studies of Cytochrome c Oxidase*

We have studied the linear electric field effect in pulsed EPR of the “EPR-detectable copper’’ signal of beef heart cytochrome c oxidase and have compared our results with those for a variety of square planar and tetrahedral Cu(I1) model compounds and with Cu(I1) proteins containing either type 1 or type 2 copper. The electric field induced g shifts (linear electric field effect) for cytochrome oxidase are comparable in magnitude to those for simple Cu(II) complexes and for some copper proteins containing type 2 sites. The shifts are smaller than those for tetrahedral copper com- plexes and for type 1 copper sites. However, the magnetic field dependence of the linear electric field effect does not resemble that observed for any Cu(I1) complex studied nor for type 1 copper. These findings cannot be reconciled with the tetrahedral Cu(II) model proposed by Greenaway, Chan, and Vincow ((1977) Biochim Biophys. Acta 490,62-78,1977) to explain the unusual EPR spectrum of cytochrome oxidase. An unusual copper binding site that gives rise to an EPR spectrum is found in cytochrome c oxidase. The spectrum for this site is peculiar for a variety of reasons; the nuclear hyperfine structure is poorly resolved at X- and Q-bands, the magnitude of gll is significantly smaller than for most known Cu(I1) compounds, and one principal axis g value is less than 2.00 (1-3). These

General Medical Sciences, National Institutes of Health, United States Public Health Service. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
' Although this electronic description is one where the copper is formally reduced, the cysteinyl sulfide is oxidized to a sulfur radical. The oxidation-reduction site would still be able to accept a single electron. a different reason must be found to explain the unusual EPR properties for this protein.
One explanation, given by Greenaway et al. ( 6 ) , is that the copper, as Cu(II), resides in a tetrahedral site and therefore has very weak hyperfine interaction. Failure to observe a well resolved hyperfine interaction at Xand Q-bands is attributed partly to the smallness of the coefficients A N and partly to line broadening caused by dipolar interaction with a close lying paramagnetic center, in this case most probably low spin ferric cytochrome a. A simulation of the EPR spectrum based on a tetrahedral model is given in the reference.
In order to examine further the validity of these models, we have performed two kinds of pulsed EPR experiments on the paramagnetic centers in cytochrome c oxidase. In the frst experiment, we measured the electric field-induced g-shifts, i.e. the linear electric field effect (7-11). This effect is sensitive to the odd component of the ligand field and can be made to yield detailed information about the symmetry of the site. In the second experiment we studied the nuclear modulation patterns in the electron spin echo decay envelope, i.e. the "nuclear modulation effect" (10-12). These patterns, which have been investigated in detail for a number of coppercontaining proteins and models (13)(14)(15)(16)(17)(18), yield information about the coupling between paramagnetic centers and,nearby nuclei.
From the LEFE2 studies it was found that the magnetic field dependence of the electric field shifts for cytochrome c oxidase resemble neither that seen for simple Cu(I1) complexes nor that seen for Cu(I1) in essentially tetrahedral sites (9). The nuclear modulation data were also unlike those obtained for any copper protein or model hitherto studied (13-17) and therefore constitute yet another property unique to the site in cytochrome c oxidase. Although they yielded little positive evidence as to the nature of the site, they showed clearly that, if indeed it is a Cu(I1) site, its ligands must differ from those commonly found in copper proteins.

MATERIALS AND METHODS
Bovine cytochrome c oxidase was prepared by the method of Hartzell and Beinert (19) and was concentrated to 0.96 mM. Cu(I1)hexaglycine, pH 6.7, was prepared as before (13). The linear electric field effect was measured at X-band and at 1.6'K by a method described previously (7,9,20). The shift parameter o was obtained by finding the value of electric field required to halve the amplitude of the electron spin echo signal for a particular setting of T, the time between the two microwave transmitter pulses of the echo-generating sequence. This parameter corresponds approximately to the average fractional g-shift (Sg/g) per unit of applied electric field, the average being taken over those Paramagnetic centers which contribute to the electron spin echo signal at the chosen magnetic field setting Ho.
The two-pulse electron spin echo decay envelopes were recorded as described by Mims and Peisach (13). Echo envelopes were also The abbreviation used is: LEFE, linear electric field effect.

6843
This is an Open Access article under the CC BY license.

Electron Spin
Echo Studies of Cytochrome c Oxidase obtained by the three-pulse method (21) but, since in the case of the copper center the modulation was shallow, the data were accumulated during repetitive sweeps over the time delay T + T in a Nicolet 1074 Instrument Computer. The resulting three-pulse envelopes were extrapolated to zero T + T and Fourier transformed to yield the superhyperfine frequency spectrum (22).

RESULTS AND DISCUSSION
The EPR spectrum of cytochrome c oxidase arises from cytochrome a and from a second site associated with the socalled EPR-detectable copper, which for brevity we refer to here as the copper site, while reserving judgment as to its exact nature. These two paramagnetic species are present in equal concentrations. However, the cytochrome a echo signal is observed over a much wider range of Ho settings than the copper signal and is correspondingly weaker. This is illustrated in Fig. 1 where we show the electron spin echo signal for cytochrome c oxidase as a function of Ho. Within the range of field settings covered by the EPR signal of the copper site, the contribution due to cytochrome a is relatively small. LEFE and nuclear modulation effect data taken within this range will therefore reflect the properties of the copper center, subject to a small error due to the underlying cytochrome a signal. Some additional discrimination against the cytochrome a signal was achieved by adjusting the microwave transmitter power level and, in the case of the LEFE measurements, by choosing longer times 7 at which the cytochrome a echo signal had undergone a proportionately greater decay.
In the upper curve of Fig as a function of magnetic field for the copper site. As can be seen, the amplitude of the spin echo increases with increasing magnetic field so as to be maximal near gI. The apparent peaks in the curve are not due to hyperfine structure, but appear in the spectrum because of the nuclear modulation effect (12). By using a different value of 7 they can be made to change position. The spectrum of electron spin echo amplitudes is approximately equivalent to the integral of the spectrum obtained in a typical EPR spectrometer with field modulation, but it tends to be less useful, except for purely illustrative purposes, because of intervention by this effect.
In the lower portion of Fig. 2, we show the effect of electric field on spin echo intensity as a function of magnetic field for the copper resonance, and for two orientations of the electric field with respect to the magnetic field. The magnitude of shifts observed is comparable to that seen for simple copper complexes (9) and for some copper proteins having type 2 copper sites (18). However, the shifts are smaller than those seen for tetrahedral Cu(I1) complexes (9) and the form of the curves is not the same as that formed in any previously studied Cu(I1) complex or copper protein (9, 14,18). Cu(I1) complexes and type 2 Cu(I1) sites in proteins yield curves which fall off at the low HO end of the spectrum and dip near gL. The cytochrome c oxidase LEFE data are also quite unlike the LEFE data obtained for type 1 Cu(I1) which yield large shift parameters and are characterized by curves in which the E& shifts rise well above the E Ho shifts at g 11. We conclude therefore that the ligand field for the copper site in cytochrome c oxidase is not tetrahedral and that it differs radically from the ligand fields associated with type 1 and type 2 Cu(I1) in proteins.
Although the signal-to-noise ratio for the spin echo signal arising from cytochrome a was much poorer than that for copper (Fig. 2), we were able to make LEFE measurements over a restricted range of magnetic fields excluding those HO settings where the copper signal is paramount. The shifts obtained were about an order of magnitude larger than for  . 3 (left). Two-pulse electron spin echo decay envelope for the EPR-detectable copper signal of cytochrome c oxidase. The spectrometer frequency was 9.548 GHz and the magnetic field setting was 3396 G.

Electron Spin
FIG . 4 (center). Three-pulse electron spin echo decay envelope for the EPR-detectable copper signal of cytochrome c oxidase. The spectrometer frequency was 9.202 GHz and the magnetic field setting, 3240 G. The modulation attributed to I4N is shallow the copper site3 and were comparable with those for low spin ferric hemoproteins and models (8). However, the form of the curves was unusual. In low spin heme compounds and hemoproteins it is generally found that the shift seen for the E 11 Hn setting is larger than that seen for the E& setting at g,,,, whereas at g,,, the opposite is observed. This can be explained by the fact that the g,,, principal axis and the axis of the two asymmetric bonds responsible for the odd ligand field are both oriented in approximately the same direction along the heme normal. In the one exceptional case of myoglobin hydroxide the LEFE curves suggest that the gmax principal axis is perpendicular to the asymmetric bond axis and thus lies, presumably, in the heme plane rather than normal to it. Although the LEFE data for cytochrome a are incomplete, they point to a similar conclusion. The data obtained at the ends of the EPR spectrum, where there is no contribution from the cytochrome c oxidase copper center, show clearly that the shift for the E II Ho setting is less than that for the E,& setting at g,,,, the roles being reversed at g,,,. This result suggests that the g,,, axis and the odd field asymmetry axis are more or less parallel to one another, thus casting further doubt on the common assumption that the g,,, axis is normal to the heme plane in a l l low spin heme proteins.
In Fig. 3, we show the two-pulse echo envelope for the copper signal of cytochrome c oxidase. The major features of the trace consist of 35-and 70-ns periods that arise from weakly coupled protons. There is, in addition, a shallow component with a longer period similar to one which was seen earlier for a Cu(I1). hexaglycine complex and attributed to coupling between the Cu(I1) spin and peptide "N nuclei (13).

Improvements in technique have made it possible for us to
It should be noted here that the measured LEFE in the region where both cytochrome a and copper signals contribute to the echo is not given by the weighted average of the LEFE's of the two component species. As may be seen from Figs. 1 and 2 the LEFE shifts for cytochrome a are large compared with the shifts for the copper center. The initial effect of the applied electric field is therefore to eliminate entirely the cytochrome a contribution to the echo signal leaving the remaining signal (in most places 1 9 0 % of the total amplitude) due solely to the copper center. The rest of the LEFE measurements (i.e. the reduction to 50% echo amplitude) then represents the effect of the applied field on the copper center. In principle it should be possible to resolve the two contributions in cases like this where one shift is much larger than the other. We were unable to do so here because of the weakness of the cytochrome a signal in relation to the copper signal.

r + r(psec)
MHz and the data were accumulated during repetitive sweeps. The baseline shown here is fictitious; the actual modulation depth being 1 5 times less than in the figure. The dotted line is an extrapolation of the data FIG . 5 (right). Fourier transform of the three-pulse spin echo decay envelope for the EPR-detectable copper signal of cytochrome c oxidase. The data used for the transform are shown in Fig. 4. undertake a more detailed study of this shallow component in both cytochrome c oxidase and Cu(I1) hexaglycine using three-pulse methods. The three-pulse echo envelope for the former is shown in Fig. 4. Frequencies of 0.9, 1.5, 1.9, and 3. 1 MHz are clearly visible in the Fourier transform of these data (Fig. 5) and in Fourier transforms of data taken at other T settings (T is the time between the fist two pulses of the three-pulse sequences (21)). Some of the smaller peaks may be artifacts, which are difficult to avoid here since the shallow nature of the modulation tends to exaggerate echo overlap "glitches" and other defects of the data (see Ref. 21, Fig. 1 ) .
The low frequency modulation pattern for Cu(I1) hexaglycine was less than half as deep as that obtained for cytochrome c oxidase thus placing it at the very limit of our present detection capability. It was nevertheless possible to see that the three-pulse envelope contained frequencies different from those seen in the three-pulse envelope for cytochrome c oxidase. This does not necessarily exclude peptide 14N as the origin of the pattern in cytochrome c oxidase but merely indicates that if the pattern is due to peptide I4N the electron nuclear coupling would have to be different in the two cases. 4 Perhaps more significant is the utter lack of resemblance between the cytochrome c oxidase nuclear modulation data and the data obtained for every other Cu(I1) protein which we have examined (13, 14, 16-18). In all these other cases we have observed a deep modulation which could be identified as being due to I4N in one or more imidazole ligands coordinated with Cu(I1). As demonstrated earlier (13, 23) 14N nuclei directly coordinated with Cu(I1) yield no modulation pattern in X-band experiments, the observed pattern being due to the remote N-3 nitrogen belonging to the imidazole ligand (15,17). The present result enables us to exclude any possibility of imidazole coordinated to Cu(I1) for the EPR detectable copper site in cytochrome c oxidase, a fact which gains significance from the apparent generality of this type of coordination in The modulation data in Figs. 3 and 4 were obtained at a field setting where the echo signal due to the copper center was at a maximum. It is therefore unlikely that the modulation was due to cytochrome a whose signal is 220 times weaker at this field setting (Fig. 1). However, to investigate this possibility further we studied the nuclear modulation pattern at a higher field setting where the copper center makes no contribution. Although the data were less good than for the copper center we were able to recognize frequencies different from those seen in Figs. 3 and 4 and typical of I4N in low spin heme compounds (21).

Electron Spin
Echo Studies of Cytochrome c Oxidase copper proteins. It does not however exclude imidazole coordination if the unpaired electron is assumed to reside mainly on a sulfur ligand and only to a smaller extent on copper as suggested by Peisach and Blumberg (4). In this case the coupling between the electron spin and the I4N nucleus would be weakened and might conceivably give rise to modulation effects such as those we have observed.
In conclusion, the pulsed EPR experiments reported here have tended to emphasize rather than to explain away the atypical nature of the EPR detectable copper center in cytochrome c oxidase. The results indicate that the symmetry is not tetrahedral and that the ligation pattern is unlike that found for either type 1 or type 2 copper in other copper proteins. The smallness of the electric field-induced g-shifts, the unusual symmetry, and the lack of a deep nuclear modulation pattern might indeed all be explained if an otherwise typical Cu(I1) center were to receive the donation of an electron from an RS-ligand, thus giving a center in which the unpaired electron spin resided mainly on the radical and only to a lesser extent on the copper. Lack of a suitable model makes it impossible, however, to pursue this hypothesis further by the methods employed here.