The Nature of Zinc in Cytochrome c Oxidase*

The zinc ion in bovine heart cytochrome c oxidase can be completely depleted from the enzyme with mercuric chloride without denaturing the protein. The metal atom stoichiometry of 5Cu/4Fe/OZn/2Mg obtained for the enzyme following HgClz treatment in- dicates that this depletion is highly selective. Zinc depletion exposes one cysteine on subunit VIa and one cysteine on subunit VIb for N-iodoacetyl-N'-(5-sulfo-1-naphthy1)ethylene-diamine (1,5-I-AEDANS) label- ing, suggesting that the zinc plays a structural role in the protein by providing a bridge between these two subunits. Although the treatment of cytochrome c oxidase with mercuric chloride inhibits the steady-state activity of the enzyme, subsequent removal of the Hg2+ bound to cysteine residues by 1,B-I-AEDANS significantly reverses the inhibition. This latter result indi- cates that the removal of the zinc itself does not alter the steady-state activity of the enzyme. reduction of vectorial

The zinc ion in bovine heart cytochrome c oxidase can be completely depleted from the enzyme with mercuric chloride without denaturing the protein. The metal atom stoichiometry of 5Cu/4Fe/OZn/2Mg obtained for the enzyme following HgClz treatment indicates that this depletion is highly selective. Zinc depletion exposes one cysteine on subunit VIa and one cysteine on subunit VIb for N-iodoacetyl-N'-(5-sulfo-1-naphthy1)ethylene-diamine (1,5-I-AEDANS) labeling, suggesting that the zinc plays a structural role in the protein by providing a bridge between these two subunits. Although the treatment of cytochrome c oxidase with mercuric chloride inhibits the steady-state activity of the enzyme, subsequent removal of the Hg2+ bound to cysteine residues by 1,B-I-AEDANS significantly reverses the inhibition. This latter result indicates that the removal of the zinc itself does not alter the steady-state activity of the enzyme.
Cytochrome c oxidase is the terminal oxidase in the mitochondrial respiratory chain.
It catalyzes the reduction of molecular oxygen to water by ferrocytochrome c as well as the coupling of this exergonic reaction to the uphill vectorial translocation of protons across the inner membrane of the mitochondrion. Each functional unit of the enzyme is known to contain two heme A prosthetic groups (heme a and heme a:J and two copper ions (CuA and CUB) (Wikstrom et al., 1981) and up to 13 subunits (Downer et al., 1976;Kadenbach et al., 1983). It is generally accepted that heme a, heme as, and CuB reside in subunit I (Winter et al., 1980;Ludwig, 1980), and CUA is associated with subunit I1 (Martin et al., 1988;Hall et al., 1988). In addition, it is now established that one zinc ion and one magnesium ion as well as additional copper (Cu,) are intrinsic components of the enzyme. A stoichiometry of 5Cu/ 4Fe/2Zn/2Mg has been proposed for the dimeric protein (Einarsdottir and Caughey, 1985;Pan et al., 1991).
The zinc atom seems to be tightly bound to the enzyme. This metal ion cannot be removed by dialysis against buffers containing the chelatingagents 1,lO-phenanthroline or EDTA in the pH range 6.0-9.5. On the other hand, Moubarak et al. (1987) showed that 60% of the zinc could be removed from the enzyme following treatment of the enzyme with fluorescein mercuric acetate (FMA). ' Naqui et al. (1988) also re-* This work was supported by National Institute of General Medical Sciences Grant GM 22432. This is contribution No. 8410 from Arthur Amos Noyes Laboratory of Chemical Physics, California Institute of Technology, Pasadena, CA 91125. The costs of publication of this article were defrayed in part b) 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.
$ To whom reprint requests should he sent.
ported that zinc could be partially (up to 50%) depleted either by treating the enzyme with dipicolinic acid or by trypsin digestion. Efforts to remove this metal ion completely from the enzyme without denaturing the protein have been unsuccessful.
There have been speculations on the possible role of zinc in the structure and function of cytochrome c oxidase. Einarsdottir and Caughey (1984) have proposed that the zinc is involved in the proton pumping activity of the enzyme. Naqui et al. (1988) have suggested that the zinc ion may play a structural role. Since the subunit location of the zinc in the enzyme is still not known, these ideas cannot be tested directly. However, possible candidates for the zinc binding site have been suggested, including subunits VIa, VII, and VIIIc (Vb, VIb, and VIIa, respectively, in the nomenclature of Kadenbach'), which contain 4, 4, and 2 cysteine residues, respectively (Buse et al., 1985). Yewey and Caughey (1987) have found that subunit 111, subunit VII, polypeptide a, and polypeptide b (111, VIIa, V b , and VIa, respectively, in the nomenclature of Kadenbach) do not contain zinc as well as other metal ions. On the basis of extended x-ray absorption fine structure (EXAFS) measurements, Naqui et al. (1988) suggested that the zinc might reside in subunit VIa (Vb in nomenclature of Kadenbach) coordinated to two sulfur ligands. However, in a different EXAFS experiment, Scott (1989) obtained evidence for three or four sulfur ligands and one nitrogen ligand to the zinc.
In the present study we have completely removed the zinc from cytochrome c oxidase with mercuric chloride without affecting the other metal ions in the protein. A metal atom stoichiometry of 5Cu/4Fe/OZn/2Mg is obtained for the zincdepleted enzyme. This experiment represents the first successful attempt to deplete the zinc totally from the oxidase. Zinc depletion appears not to inhibit the steady-state electron transfer activity of the enzyme. Finally, the removal of the zinc exposes cysteines on both subunit VIa and VIb for fluorescent labelling by N-iodoacetyl-N'-(5-sulfo-l-naph-thy1)ethylene-diamine (1.5-I-AEDANS). This observation implicates these cysteines on subunit VIa and VIb as the location of the zinc ion. We propose that the zinc plays a structural role in the protein by providing a bridge between these two subunits.

EXPERIMENTAL PROCEDURES
Materials-Beef heart cytochrome c oxidase was isolated by the method of Hartzell and Beinert (1974). Special precaution was taken to avoid contamination of the preparation from exogenous metal ions during the isolation and purification. Enzyme concentrations were determined spectrophotometrically by using Ac (reduced minus oxidized = 24 mM" cm") at 605 nm. The enzyme preparation was stored a t -80 "C until used. Tris, sodium deoxycholate, and lauryl maltoside were obtained from Sigma and used without further purlfication.
' The subunit nomenclature of Kadenbach (Kadenbach et al., 1983) is used throughout this manuscript.

19109
Nature of Zinc in Cytochrome c Oxidase FMA, dipicolinic acid, and 1,5-I-AEDANS were also obtained from Sigma. Depletion of Zinc Using Mercuric Chloride-A 50 p~ solution of cytochrome c oxidase in 100 mM sodium phosphate, pH 7.4, containing 0.3% Tween 80 was treated with 200 p~ HgCI,. The incubation was carried out for 10 min at 37 "C. The HgC1,-treated sample was then passed through a 1.5 X 20 cm Sephadex G-25 column equilibrated with 100 mM sodium phosphate, pH 7.4, containing 0.1% sodium cholate and 10 mM EDTA to remove the zinc and excess HgC1,.
Depletion of Zinc Using Dipiolinic Acid-Cytochrome c oxidase (100 p M ) in 100 mM sodium phosphate, 1 mM EDTA, 1% Tween 20, 9% glycerol, pH 6.0 was treated with 40 mM dipicolinic acid at 4 "C for 50 min as described by Naqui et al. (1988). After incubation, the sample was passed through a Sephadex G-25 column that was previously equilibrated with the same buffer at 4 "C.
Depletion of Zinc Using FMA-Cytochrome c oxidase (50 p~) in 100 mM sodium phosphate, 0.3% Tween 80, pH 7.4, was incubated with 200 p M FMA at 37 "C for 5 min as described by Moubarak et al. (1987). The solution was then passed through a Sephadex G-25 column that had previously been equilibrated with 100 mM sodium phosphate, 0.1% sodium cholate, 10 mM EDTA, pH 7.4.
Dissociation of Cysteine-bound Hg2+ from the HgC1,-treated Enzyme by 2-Mercaptoethanol-Sodium deoxycholate was added to a HgC1,treated oxidase sample (20 p~) to a final concentration of 1%. The sample then was incubated with 120 mM 2-mercaptoethanol at 27 "C for 20 min. The excess 2-mercaptoethanol was removed by passing the sample through a Sephadex G-25 column that had previously been equilibrated with 100 mM Tris-CI, 0.1% lauryl maltoside, pH 8.0.
Assay of Electron Transfer Activity-The activities of the native and zinc-depleted enzymes were determined spectrophotometrically using a Beckman DU-7400 diode array spectrophotometer. Assays were performed by following the oxidation of 20 p~ ferrocytochrome c at 550 nm in 50 mM sodium phosphate, 0.1% lauryl maltoside, pH 6.0. The concentration of enzyme was 5-20 nM. The turnover number of the native enzyme was typically about 500 s-'.
Metal Analysis-Metal contents of the enzyme preparations were determined by direct current plasma atomic emission spectrometry as described by Pan et al. (1991). These experiments were performed in the Laboratory of Dr. Stolper, Division of Geological and Planetary Science at Caltech. SDS-Polyacrylamide Gel Electrophoresis-Cytochrome c oxidase samples were dissociated into subunits for 1 h at 25 "C in 8 M urea, 5% SDS. Slab gels were run on a LKB 2001 vertical electrophoresis unit as described by Darley-Usmar et al. (1981) using a 7% polyacrylamide stacking gel and a 14% running gel, both containing 6 M urea. The gels were illuminated with UV light and photographed to observe the AEDANS and FMA fluorescence. Finally, the gels were stained with Coomassie Blue.

Metal Contents of Native and Zinc-depleted Cytochrome c
Oxidase-The average metal atom ratios of native and three zinc-depleted samples are presented in Table I. Three determinations were made for each sample. The experimental error of an individual measurement is within k5%. The stoichiometry of 5Cu/4Fe/2Zn/2Mg per dimer obtained for the native enzyme, is consistent with the determination of Yoshikawa et al. (1988) and the data recently reported from this laboratory (Pan et al., 1991). The metal contents of the dipicolinic acid-treated and FMA-treated samples confirm that about 43 and 51% of the zinc are depleted by these treatments, respectively, as determined earlier (Moubarak et al., 1987;Naqui et al., 1988). On the other hand, a metal stoichiometry of 5Cu/4Fe/OZn/2Mg was obtained for the HgC1,-treated sample. Thus, it is possible to completely deplete the zinc from the enzyme without affecting the other metal centers.
Metal Contents of 1,5-I-AEDANS-labeled Cytochrome c Oxidme-The metal contents of 1,5-I-AEDANS-labeled cytochrome c oxidase samples are also summarized in Table I. From these results, we see that the metal contents of native, HgC1,-treated and dipicolinic acid-treated samples remain unchanged after the labeling. For the FMA-treated enzyme, however, we find that the residual zinc is also removed after 1,5-I-AEDANS-labeling. As shown in Table I, the Zn/Fe ratio decreases from 0.23 to 0.08, indicating that the remaining 50% of the total zinc is removed by 1,5-I-AEDANS labeling.
Labeling of Cytochrome c Oxidase by 1,5-I-AEDANS-l,5-I-AEDANS labels free cysteines in the enzyme. Since cysteines have been implicated as ligands to the zinc ion, removal of the zinc might possibly expose these cysteine residues for labeling by 1,5-I-AEDANS. As shown in lune 1 of Fig. 1, 1,5-I-AEDANS exclusively labels subunit TI1 in the native enzyme, presumably one of the exposed cysteines (Hall et al., 1988). In the case of the dipicolinic acid-treated sample, subunit VIa and/or VIb (subunits VIa and VIb cannot be discerned) are labeled in addition to subunit I11 (lune 3 of Fig.  1). This result suggests exposure of a cysteine(s) on the subunits upon removal of 43% of the zinc from the enzyme by dipicolinic acid treatment.
SDS-polyacrylamide gel electrophoresis of the FMAtreated enzyme reveals FMA fluorescence from subunits I, 11, Va, and VIIa (lune 2 of Fig. l ) , consistent with the earlier results of Stonehuerner et al. (1985). Unfortunately, these results cannot be used to infer the cysteine(s) that have been modified in situ by the FMA treatment since it is well known that the FMA mercurial reagent migrates during the denaturing conditions of the gel. However, if the FMA treatment is followed by covalent labeling by 1,5-I-AEDANS prior to gel electrophoresis to dissociate the subunits, the cysteine(s) exposed by the zinc depletion may be inferred. The results of such a "double" labeling experiment show extensive FMA or 1,5-I-AEDANS labeling of subunits I, 11, 111, Va, VIa, VIb, and VIIa, as judged by the FMA and/or 1,5-I-AEDANS fluorescence (data not shown). It is clear from this experiment that cysteines of subunits VIa and VIb have become labeled by 1,5-I-AEDANS upon the total depletion of the zinc.
A similar 1,5-I-AEDANS-labeling pattern is obtained for the HgC12-treated sample (lune 2 of Fig. 2), where we observe extensive labeling of subunits I, 11,111, Va, VIa, VIb, and VIIa. It is clear that the HgC1, treatment loosens up the protein and exposes a number of cysteine residues for 1,5-I-AEDANS labeling. However, if the HgC1,-treated oxidase were incubated with 2-mercaptoethanol to dissociate the bound H g + prior to 1,5-I-AEDANS labeling, we observe labeling of only subunits 11,111, VIa and VIb (lane 1 of Fig. 2). 1,5-I-AEDANS labels subunit I11 even in the native enzyme. The cysteine(s) modified on subunit I1 have previously been identified as cysteines 196 and/or 200, which become exposed upon removal of CuA by 2-mercaptoethanol (Hall et al., 1988). By elimination then, we can conclude that total zinc depletion by the HgC1,-treatment exposes cysteines on subunits VIa and VIb to 1,5-I-ADEANS labeling. We obtain a 1:l fluorescence intensity ratio for the 1,5-I-AEDANS-labeled subunits VIa and VIb. This result implicates cysteine residues on subunits VIa and VIb as ligands to the zinc ion in the native enzyme.  were treated with 4 mM dithiothreitol and 5 mM 1,5-I-AEDANS for 12 h in 100 mM Tris-CI, 0.1% lauryl maltoside, pH 8.0, and passed through a Sephadex G-25 column to remove excess 1,5-I-AEDANS. Lane 2, 50 p~ native cytochrome c oxidase in 100 mM sodium phosphate, 0.3% Tween 80, pH 7.4, was incubated with 200 p~ FMA at 37 "C and passed through a Sephadex G-25 column to remove excess FMA. 15-p1 aliquots of each sample were then dissociated for 1 h at 25 "C in 3% SDS and 6 M urea and run on a 14% polyacrylamide Laemmli gel system. The slab gel was photographed with UV illumination to detect the fluorescence and then stained with Coomassie Blue to locate the subunits as indicated.

Electron Transfer Activity of the Zinc-depleted Oxidme-
The steady-state electron transfer activities of the various zinc-depleted cytochrome c oxidases are compared with that of the native oxidase in Table 11. Labeling of the native oxidase with 1,5-I-AEDANS does not alter the steady-state electron transfer activity. The electron transfer activity of the dipicolinic acid-treated sample is also essentially the same as that of the native sample. Both the HgClz treated and FMAtreated samples display a 50% decrease in the electron transfer activity, but the inhibitions are reversed significantly after subsequent incubation of the samples with 1,5-I-AEDANS. This result indicates that the inhibition of the activity is caused by the binding of the mercuric compounds rather than the result of zinc depletion.

DISCUSSION
The nature of the zinc in cytochrome c oxidase, including its subunit location and the role of the zinc in the structure and function of the enzyme, has attracted considerable attention in recent years. The ideal approach for ascertaining the role of the zinc is to prepare a zinc-depleted oxidase. However, a number of laboratories have reported that it is very difficult to remove the zinc completely from the enzyme; and typically, the zinc is released only when the protein is denatured. The experiment we report here represents the first successful attempt to totally deplete the zinc without denaturing the enzyme. HgC12-treated samples give a metal stoichiometry of 5Cu/4Fe/OZn/2Mg, indicating that the method that we have developed here to deplete the zinc is highly selective. Although It is also noteworthy that incubation of the FMA-treated enzyme with 1,5-I-AEDANS removed the remaining zinc. This protocol provides an alternate approach to prepare the zincdepleted oxidase. The steady-state electron transfer activities of the zincdepleted samples indicate that the zinc does not have a profound influence on the enzymatic activity. Although partial inhibition of the enzyme is observed after HgClz and FMA treatments, this inhibition can be largely reversed by subsequent 1,5-I-AEDANS labeling, in accordance with the earlier observations of Mann and Auer (1980). The partial inactivation of the enzyme by mercuric compounds has been attributed to the binding of mercuric compound to a cysteine in subunit I that is crucial for electron transfer (Mann and Auer, 1980;Stonehuerner et al., 1985). Since H$+ ions bound to the cysteines can be removed by thiol-exchange, incubation of HgC12-or FMA-treated oxidase with 1,5-I-AEDANS should reverse the inactivation, as observed. In contrast to the inhibition of the steady-state activity noted here, the pre-steadystate reduction of cytochrome c oxidase by ferrocytochrome c is not affected by FMA treatment (Moubarak et al., 1987). It will be of interest to compare the pre-steady-state kinetics and proton pumping activities of the zinc-depleted oxidase with these same properties of the native enzyme. These studies are currently in progress.
It is very interesting that both subunits VIa and VIb are labeled after the zinc is removed (and after any substituted Hg'+ had been displaced by 1,5-I-AEDANS). This result directly implicates these two subunits as the zinc binding domain. On the other hand, Yewey and Caughey (1987) found that removal of subunits 111, Vb, VIa, and VIIa did not affect the zinc content in the enzyme. Since the removal of subunit VIb was not confirmed in this experiment, there is a possibility that the zinc remained associated with the subunit VIb after the subunit VIa removal. If the subunit VIb was also removed by the treatment (Penttila, 1983), then our present findings would be in contradiction to the earlier results of Yewey and Caughey. However, subunits Vb, VIb, and VIIa have also been suggested as the possible subunit location of the zinc (Buse et al., 1985). On the basis of EXAFS studies, Naqui et al. (1988) and Scott (1989) have indicated that the zinc is ligated by two or three cysteines. Since subunit VIa alone can provide for only one of the cysteines of the zinc, one of the other cysteine rich subunits must be involved as well. The results of the present study are consistent with a ligand structure of the zinc wherein the zinc is ligated to one cysteine on subunit VIa and one cysteine on subunit VIb. If the two cysteines are derived from two subunits, namely VIa and VIb, then we may conclude that the zinc plays a structural role in the enzyme by providing a bridge between these two subunits. In the dipicolinic acid-treated sample, 1,5-I-AE-DANS labeling occurs on subunit VIa and/or VIb (subunit VIa and VIb cannot be discerned), providing further evidence in support of the subunit location of the zinc.
In conclusion, we show in this work that: 1) the zinc ion of cytochrome c oxidase can be completely dissociated from the enzyme without denaturing the protein; 2) the zinc does not appear to be involved in the electron transfer activity of the enzyme; and 3) the zinc serves a structural role in the enzyme by providing a bridge between subunit VIa and VIb.