Zinc ions inhibit the QP center of bovine heart mitochondrial bc1 complex by blocking a protonatable group.

Bovine heart bc1 complex is reversibly inhibited by zinc ions with an inhibition constant KI of 10(-7) M at pH > or = 7.0. Binding of zinc is at least a factor of 10 tighter than binding of any other metal ion tested. Essentially complete inhibition of ubihydroquinone:cytochrome c oxidoreductase activity is observed at concentrations of [Zn2+] > 5 microM. Zinc does not affect the Km for the substrates, ubihydroquinone or cytochrome c, but zinc inhibits reduction of the cytochromes by ubihydroquinone through the QP center. A radioactive binding assay using 65Zn revealed one high affinity binding site per bc1 complex with KD < or = 10(-7) M at pH = 7.0 and 3-4 additional low affinity binding sites (KD > 2 x 10(-6) M). Zinc binding does not depend on the redox state of the high potential chain (iron-sulfur protein and cytochrome c1). Zinc binds 3 times tighter to Fe-S-depleted bc1 complex indicating that the zinc binding site is not on the "Rieske" iron-sulfur protein in contrast to a recent report by Lorusso et al. (Lorusso, M., Cocco, T., Sardanella, A.M., Minuto, M., Bonomi, F., and Papa, S. (1991) Eur. J. Biochem. 197, 555-561). Zinc binds to a site which has the same affinity for zinc as for protons. We conclude that the zinc binding site is close to a protonatable group of the bc1 complex with pKa = 7.2 which has not been identified previously. We propose that this group is part of the proton channel at the hydroquinone oxidation center of the bc1 complex.

Inhibitors have been indispensable tools for elucidating the reactions of the ubiquitous bc 1 complexes (for a review, see Ref. 1). Almost all of these inhibitors are aromatic organic compounds where at least part of the structure has some structural relationship to the substrate, ubiquinone. The single notable exception are zinc ions which have first been reported by Skulachev et al. (2) to inhibit mitochondrial respiration in micromolar concentrations. Subsequent studies (3)(4)(5) have established that the primary site of zinc inhibition in bovine heart mitochondria is the bc 1 complex.
The knowledge about the bc 1 complex has largely increased over the past 20 years, including the structure of the redox centers and the peptide composition of the mammalian 11subunit complex. The "Q-cycle" mechanism has been established for the sequence of electron and proton transfers (see Ref. 6). In the light of this body of information, we have reinvestigated inhibition by zinc to obtain information about the specific interaction between zinc ions and the bc 1 complex and to use this information to get insight into mechanistic details of the electron and proton transfer reactions.

MATERIALS AND METHODS
Metal salts were obtained either as chloride or as nitrate salts from Fluka. Cytochrome c (horse heart) was from Sigma, prepared without trichloroacetic acid.
Bovine heart mitochondria were prepared according to Smith (7). bc 1 complex was prepared as described by Schä gger et al. (8) with the following modification: the buffer for the final Sepharose CL-6B gel filtration column was 0.05% Triton X-100, 100 mM NaNO 3 , 20 mM Pipes, 1 pH ϭ 7.2 (no NaN 3 ). Fe-S-deficient bc 1 complex ϩ antimycin was prepared following the procedure given by Schä gger et al. (9). After complete elution of the iron-sulfur protein, Fe-S-deficient bc 1 complex ϩ antimycin was eluted with 0.05% Triton X-100, 150 mM sodium phosphate, pH 7.2, and applied to a gel filtration column (Sepharose CL-6B, Pharmacia) equilibrated with 0.05% Triton X-100, 100 mM NaNO 3 , 20 mM Pipes, pH 7.2, and run with the same buffer. The fractions containing Fe-S-deficient bc 1 complex ϩ antimycin were mixed with 10% (w/v) glycerol and frozen in liquid nitrogen.
Buffer for kinetic measurements was prepared as follows: 200 mM sucrose, 50 mM Pipes (from a partially neutralized stock solution), and 1 mM NaN 3 were dissolved in tridistilled water. The solution was purified over a Chelex 100 ion exchange resin column (analytical grade, Bio-Rad). 20 mM HNO 3 , 1 mM Ca(NO 3 ) 2 , and 0.024% (0.5 mM) dodecyl maltoside (Boehringer) were added, and the pH was adjusted with 2 M NaOH (purified over Chelex 100 and stored in a plastic bottle). Calcium was added to prevent unspecific binding since it did not show any inhibition even at 10 mM concentration. For the measurement of the pH dependence, the buffer contained 25 mM Pipes and 25 mM Epps. All buffers were stored in plastic bottles.
NBH:cytochrome c reductase activity was measured at ambient temperature in 1-cm cuvettes, stored in EDTA solution, washed extensively with NaOH, and finally rinsed with water. 10 l of cytochrome c stock solution (5 mM in 50 mM Pipes, pH 7.0) were diluted into 1 ml of buffer. 5 l of bc 1 complex (diluted to 1 M cytochrome b in buffer containing 30% glycerol) and metal ion solution were added. Reductase reaction was started by addition of NBH in dimethyl sulfoxide (final concentration 45 M).
Michaelis-Menten kinetics was performed as above in stirred 2-ml cuvettes thermostatted at T ϭ 24°C. For cytochrome reduction, NBH was added to bc 1 complex (2 M cytochrome b in kinetics buffer) in a stirred 2-ml cuvette to a final concentration of 27 M. Reduction kinetics was measured at pH 6.2 where electron transfer from hydroquinone to bc 1 complex is slower than at pH 7.
The radioactive binding assay was performed in 4.5-ml centrifuge tubes, washed first with EDTA solution, extensively with NaOH, and finally with water. A 65 ZnCl 2 stock solution (50 M, specific activity 195 kBq/ml, diluted from Amersham ZAS.1 batch 63) was diluted into 2.5 * This work was supported by Grant Li 474/2 from the Deutsche Forschungsgemeinschaft (DFG), Priority Programme "Transition Metals in Biology and their Coordination Chemistry." The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed: Universitä tsklinikum Frankfurt, ZBC, Therapeutische Biochemie, Theodor-Stern- Kai  The bc 1 complex was sedimented by centrifugation (Rotor Ti 70.1, 35,000 rpm, 3 h). 2 ml of the supernatant were withdrawn to measure free 65 Zn. The remaining supernatant was discarded, and the tubes were centrifuged again (20,000 rpm, 30 min). The supernatant was removed and the pellet was incubated overnight in 50 l of 10% Triton X-100, 1 M NaCl. The dissolved pellet was diluted with 625 l of buffer containing 0.5% Triton X-100, 200 mM NaCl, 20 mM Mops and finally with another 625 l of water. Thus, the final volume was exactly one-half of the original assay solution. 1 ml (water added to a final volume of 2 ml) was used to measure bound 65 Zn in a Beckman Gamma 5500 counter. bc 1 concentration was determined from reduced-oxidized difference spectra using ⑀ 562-575 ϭ 28.5 mM Ϫ1 cm Ϫ1 . EPR spectroscopy was done with a Bruker 200 D spectrometer equipped with cryogenics, peripheral equipment, and data acquisition/manipulation facilities as described previously (10).

Metal Ion Inhibition of Bovine Heart
Mitochondria-Inhibition of succinate:O 2 respiration was measured in uncoupled mitochondria washed with Chelex-purified buffer. Complex inhibition curves were obtained, indicating multiple metal binding sites. About 40% of the succinate:O 2 activity was inhibited at low zinc concentration (Ͻ2 M), but a concentration of Ͼ400 M (500 Zn 2ϩ /bc 1 ) was required for 90% inhibition. Hg 2ϩ , Ag ϩ , Cu 2ϩ , and Cd 2ϩ were found to be less effective in this order.
Inhibition of Isolated bc 1 Complex-In contrast to whole mitochondria, simple inhibition curves were obtained when isolated bc 1 complex was titrated with various metal ions. The strongest inhibition was observed by Zn 2ϩ ions, followed by Ag ϩ , Hg 2ϩ , and Cu 2ϩ ( Table I). The K I for Cd 2ϩ was 100 times higher than the K I for Zn 2ϩ . Inhibition by Hg 2ϩ and Cu 2ϩ was not completely reversed by addition of the metal chelator, diethylenetriaminepentaacetic acid, while inhibition by Zn 2ϩ and Cd 2ϩ was completely reversible. Fig. 1 shows inhibition of NBH:cytochrome c oxidoreductase activity of isolated bc 1 complex by Zn 2ϩ ions. The K I could be obtained by fitting the Zn 2ϩ dependence with a single homogeneous inhibition site: At pH 7.0, a K I of 1 ϫ 10 Ϫ7 was obtained. A single inhibition site is also observed by plotting the linearized form At [Zn 2ϩ ] Ͼ 5 M, no steady state electron transfer activity was detectable.
Zn 2ϩ Binding Assay Using 65 Zn-In the binding assay, multiple binding sites with different stoichiometry and affinity were observed (Fig. 2). The binding curve could be fitted by the standard binding equation assuming two independent types of binding sites (pH 7.2). High affinity site: n, 1.1 Ϯ 0.1 Zn 2ϩ /c 1 ; K I , 0.13 M; low affinity sites: n, 3-4 Zn 2ϩ /c 1 ; The K D of the high affinity binding site was essentially identical with the inhibition constant K I under all conditions tested. Therefore, we conclude that inhibition is caused by binding of a single zinc ion per cytochrome c 1 with a dissociation constant of approximately 10 Ϫ7 M. The additional multiple low affinity binding sites are not related to the inhibition of catalytic activity.
Note that the high affinity binding constant cannot be obtained by using a linear fit of the Scatchard plot (see Fig. 2, inset). A line representing only high affinity binding does not match the data points even at low zinc concentrations since approximately 15% of the first zinc ions bind to low affinity sites.
Inhibition of Single Turnover Cytochrome Reduction-No zinc inhibition of cytochrome b reduction through the Q N center was observed in the presence of Q P inhibitors (not shown). Antimycin blocks electron transfer through the Q N center so that only the Q P center is active (11). In the presence of antimycin and 100 M Zn 2ϩ , both cytochrome c 1 reduction and cytochrome b reduction were significantly slower compared to the antimycin control (Fig. 3). The potent Q P center inhibitor, MOA-stilbene, blocked both cytochrome c 1 and cytochrome b reduction completely (12). The data show that zinc affects the Q P but not the Q N center.
Kinetics of Partially Inhibited bc 1 Complex-In order to determine the inhibition mechanism, we tested whether zinc interferes with binding of either substrate, ubihydroquinone or cytochrome c. Michaelis-Menten enzyme kinetics of bc 1 complex for NBH and cytochrome c were measured at zinc concentrations of 0, 0.075, and 0.2 M corresponding to 0, 35, and 55% inhibition, respectively (Fig. 4). With both hydroquinone and cytochrome c, V max decreased while the K m values were unchanged as indicated by the parallel lines in the Eadie-Hofstee plot (Fig. 4). This shows that zinc does not interfere with the interaction of substrate and enzyme; inhibition occurs by a mechanism which is noncompetitive for both substrates.

Interaction between Zinc and Organic Q P Center Inhibitors-
The interaction between zinc and the Q P inhibitor, MOA-stilbene, was tested in two different ways. The binding constant for MOA-stilbene measured by fluorescence quench titration (13) was identical in the absence and presence of 100 M Zn 2ϩ . The K D for zinc binding was identical for bc 1 complex with or without bound MOA-stilbene (data not shown).
Binding of the Q P center inhibitor, stigmatellin, which interacts with both the iron-sulfur protein and cytochrome b (14), was determined by red shift titration of reduced bc 1 complex (15). The K D could be determined exactly due to the high affinity; for bc 1 complex both in the absence and presence of 100 M Zn 2ϩ , K D values below 20 nM were obtained. We conclude that zinc does not affect the binding of other Q P center inhibitors.
pH Dependence of Zinc Binding and Inhibition-The pH dependence of zinc inhibition was tested by determining both the inhibition constant K I and the high affinity dissociation constant K D at different pH values. Both K I (q) and K D (å) are less than 0.1 M at pH Ͼ 7, while both values increased (i.e. affinity decreased) at lower pH values (Fig. 5). Since the values obtained by both methods were identical, we will use only the term K D during the following discussion for values obtained by both inhibition and binding experiments.
The data have been analyzed to obtain the dissociation constants for both Zn 2ϩ and protons; the dissociation constant for protons, K a , is generally expressed as pK a value (Ϫlog K a ). Two different models have been considered: competitive and noncompetitive binding. Competitive binding occurs if zinc and protons displace each other from the same binding site, while noncompetitive binding requires different interacting binding sites so that proton binding to one site will influence the K D for zinc at a site nearby. For competitive binding, the apparent KЈ D for zinc ions as a function of pH is obtained as i.e. KЈ D is independent of pH at pH Ͼ Ͼ pK a and depends linearly on [H ϩ ] for pH Ͻ Ͻ pK a . For noncompetitive binding, the apparent KЈ D is where K D1 and K D2 are the zinc dissociation constants at high and low pH, respectively (see Appendix). Negative cooperativity becomes indistinguishable from direct competition at high ratios of K D1 /K D2 and the equation reduces to Equation 3.
The equations for both competitive and noncompetitive binding fit the data points if the zinc dissociation constant at low pH (K D2 ) is at least 3 orders of magnitude higher than that at high pH (K D1 ), i.e. K D2 /K D1 Ն 1000 (Fig. 5A). From these fits, a K D1 (high pH) of 0.07 M and a pK a of 7.2 were obtained. In Fig. 5A, we have also constructed the curve for noncompetitive binding using K D2 /K D1 ϭ 100 and the same values for K D1 and pK a in order to show that the curve for noncompetitive binding does not match the data points if the ratio K D2 /K D1 is 100 (or even less).
Since it was not possible to measure either zinc binding or zinc inhibition at pH Ͻ 5.5, we cannot distinguish between competitive and noncompetitive binding models, if K D2 /K D1 Ն 1000. However, in both cases, there is strong negative cooperativity between zinc and proton binding; on protonation, the zinc dissociation constant increases by more than 2 orders of magnitude and upon zinc binding, the pK a of the proton binding site is lowered by more than 2 pH units from pK a ϭ 7.2 to below 5.
The activity of the bc 1 complex follows a different pH dependence (Fig. 6). Activity increases with pH up to a pH optimum of 8.5 and decreases again. The increase of the activity can be modelled with a single pK a value of 6.6 where the protonated form is assumed to be completely inactive. The pH dependence of zinc binding cannot be modelled using the pK a of 6.6 (Fig. 5B).
Redox Dependence of Zinc Binding-Since the bc 1 complex has several groups with redox dependent pK values, we tested the effect of the redox state of the "high potential chain" of the bc 1 complex (iron-sulfur protein and cytochrome c 1 ) on zinc binding. The redox state was maintained during the 65 Zn binding assay by addition of a 250-fold excess of oxidized or reduced cytochrome c (50 M). At the end of the experiment, cytochrome c was still Ͼ90% in its original redox state. Electrons equilibrate rapidly between cytochrome c, cytochrome c 1 , and the iron-sulfur protein. The redox state of cytochrome b could not be determined in the presence of the high excess of cytochrome c.
The same dissociation constant K D for zinc was obtained in all three experiments: without cytochrome c added, with oxi-dized and with reduced cytochrome c added (data not shown). This shows that zinc binding is redox independent. Therefore, the group where protons bind with negative cooperativity to zinc must have a redox independent pK a value of 7.2.
Does Zinc Bind to the "Rieske" Iron-Sulfur Protein?-In order to test whether zinc binds to the iron-sulfur protein as reported previously (16), we measured zinc binding to Fe-S-depleted bc 1 complex. This preparation is inactive due to loss of the ironsulfur protein and has also lost the smallest subunit (6.4 kDa). Fe-S-depleted bc 1 complex can be reactivated by addition of isolated iron-sulfur protein (17,18) and still binds Q P center inhibitors (15).
Since Fe-S-depleted bc 1 complex is prepared in the presence of antimycin (9), we first tested whether antimycin affects zinc binding. The K D observed with and without saturating concentrations of antimycin was identical (Fig. 7). The K D observed for the Fe-S-depleted bc 1 complex was 3 times lower than for whole bc 1 complex, i.e. binding is three 3 times tighter (Fig. 7). The K D showed the same pH dependence as whole bc 1 complex (data not shown). We conclude that zinc does not bind to the ironsulfur protein but to residues in the vicinity of the iron-sulfur protein.
Effect of Zinc Binding on the EPR Spectrum of the Rieske Iron-Sulfur Protein-EPR spectra of bc 1 complex in Chelex 100-purified nitrate buffer and reduced with dithionite were recorded with and without addition of saturating zinc concentrations ([bc 1 ] ϭ 76 M; [Zn 2ϩ ] ϭ 173 M; T ϭ 17 K). The spectra without and with zinc added were indistinguishable (not shown). 1 Complex-Of the metal ions tested, only Zn 2ϩ and Ag ϩ bind with higher affinity to the bc 1 complex than to other sites in isolated mitochondria. Binding of Zn 2ϩ to the bc 1 complex is more than 4 orders of magnitude tighter than binding of Co 2ϩ . This prevents spectroscopic investigation of the binding site as zinc ions are not amenable to most spectroscopic or magnetic techniques and is surprising since Co 2ϩ (ionic radius 0.72 Å) generally is an excellent probe for natural zinc sites (ionic radius of zinc: 0.74 Å). The highest differences between the stability of zinc and cobalt complexes are observed for tetrahedral sites with nitrogen or sulfur ligands (19).

Effect of Different Metal Ions on the bc
Binding of zinc is completely reversible. Studies using ophenanthroline showed that reactivation requires 3 mol of o-phenanthroline/mol of Zn 2ϩ added. Therefore, reactivation involves removal of zinc from the bc 1 complex and complexation as Zn(phen) 3 2ϩ . Binding of Hg 2ϩ and Cu 2ϩ was not fully reversible, probably due to the oxidation of disulfide groups by both ions.
In isolated bovine mitochondria, the zinc content has been FIG. 5. pH dependence of inhibition of bc 1 complex by zinc ions. A: q, K I values obtained from inhibition kinetics; å, K D values obtained from binding assays. The bold line was fitted to the data points using the competitive binding model (see text) giving K D1 (high pH) ϭ 0.07 M; pK a ϭ 7.2. The dotted and the dashed lines represent calculated curves for the noncompetitive binding model using K D2 ϭ 0.07 M; pK a ϭ 7.2; K D2 /K D1 ϭ 100 and 1000, respectively. B, same data as before. The curves were calculated for fully competitive inhibition using three different pK values of 6.6, 7.2, and 7.6 (from left to right).
FIG. 6. pH dependence of electron transfer activity of bc 1 complex. The line was fitted giving pK a ϭ 6.6. Above pH 8.5, activity decreases. determined as 76 -104 g of Zn 2ϩ /g of protein ϭ 1.2-1.6 mol of Zn 2ϩ /g of protein (20). Using the bc 1 concentration in bovinemitochondria isolated according to Smith (7), 0.25 mol of bc 1 /g of protein, the zinc content is obtained as 5-6 Zn 2ϩ /bc 1 . At this ratio, approximately 25% inhibition of the activity of the bc 1 complex was observed. We therefore conclude that zinc inhibition may occur at physiological zinc concentrations.
One Zn 2ϩ Binds with High Affinity to the Q P Center-The inhibition and binding data show that binding of one Zn 2ϩ ion per bc 1 monomer inhibits electron transfer. The dissociation constant, 10 Ϫ7 at pH 7.0, is 1 order of magnitude lower than reported previously (5,16). In order to obtain this value, every care had to be taken to clean the glassware and to protect buffers and solutions from adventitious metal contamination. In addition, 3-4 low affinity sites per bc 1 monomer were observed.
If zinc ions were added together with organic Q P center inhibitors, no additional effect was observed while zinc plus antimycin lead to a reduction of the rate of cytochrome c 1 and b reduction through the Q P center. This clearly shows that zinc inhibits at the Q P center. However, it is less effective than organic Q P center inhibitors since it does not block single turnover through the Q P center completely even if added at high concentration (1000 times K D ). A residual activity of less than 1 s Ϫ1 is sufficient for the reduction kinetics shown in Fig. 3 but is not detected under standard activity assay conditions.
Site of Zinc Binding-Zinc binds to Fe-S-depleted bc 1 complex with 3 times higher affinity and with the same pH dependence as compared to whole bc 1 complex. This clearly shows that the high affinity binding site is not located on the iron-sulfur protein but on domains in the vicinity of the iron-sulfur protein.
The iron-sulfur cluster and cytochrome b are close to each other and are both involved in the formation of the Q P reaction center where zinc inhibition occurs (15). Therefore, it is likely that the zinc binding site resides on cytochrome b.
While this work was in progress, Lorusso et al. (16) reported reversible inhibition of the Q P center, binding of zinc to the iron-sulfur protein, 42% decrease of the Rieske EPR signal upon zinc binding, and 50% inhibition of electron flow at saturating zinc concentrations. However, there are some problems with the experimental conditions used in the work of Lorusso et al. (16) that are likely to cause artifacts.
Lorusso et al. (16) have measured cytochrome reductase activity and its inhibition in buffer containing 1 mM KCN. Cyanide is a strong complexing agent and forms the complex Zn(CN) 4 2Ϫ with a cumulative binding constant ␤ 4 ϭ 5 ϫ 10 16 M

Ϫ4
. Thus, at a cyanide concentration of 1 mM, a total zinc concentration in the micromolar range would lead to a free zinc concentration below the nanomolar level. The zinc effect observed in Ref. 16 can therefore be explained only if a major fraction of cyanide is complexed with other metal ions. It is likely that the half-inhibition observed by Lorusso et al. (16) is due to a redistribution of metal ions between various binding sites, including cyanide and bc 1 complex. We have used azide which has very low metal complex formation constants. The uninhibited reductase activity given in Ref. 16 corresponds to a turnover number of 5-8 s Ϫ1 for durohydroquinone as substrate using the specific bc 1 content reported in Ref. 21. 50% inhibited activity corresponds to a turnover number of 2-4 s Ϫ1 . We have obtained turnover numbers in excess of 350 s Ϫ1 using nonylubihydroquinone as substrate.
Zinc does not bind to the iron-sulfur protein since Fe-Sdepleted bc 1 complex, which can be reconstituted with ironsulfur protein and does still bind MOA inhibitors to the Q P center, binds zinc with 3 times higher affinity than bc 1 complex. In addition, we have not observed any effect of saturating zinc concentrations on the EPR of the iron-sulfur cluster.
Zinc Does Not Compete with Either Cytochrome, Quinone, or Inhibitors, but Shows Negative Cooperativity with Proton Binding-Despite showing noncompetitive inhibition kinetics, E-␤methoxyacrylates, known Q P center inhibitors, increase the K m for hydroquinone (12). In contrast, zinc does not affect binding of any of the substrates, cytochrome c or quinone; this is emphasized by the fact that zinc also does not compete with organic Q P center inhibitors for a common binding site. This suggests that zinc has a mechanism of inhibition completely different from any other known bc 1 inhibitor. bc 1 complex will bind protons and zinc with approximately equal affinity (K D (Zn 2ϩ )/K a (H ϩ ) Ϸ 1). Due to the strong negative cooperativity, zinc and protons displace each other from their binding site(s). On zinc binding, the pK a of the proton binding site drops from 7.2 to below 5. However, this proton binding site is not related to the group determining the pH dependence of the activity of the bc 1 complex, which depends on deprotonation of a group with pK a ϭ 6.6 ( Fig. 6) (21, 22). The pK a of the zinc binding site is also not identical with the redox dependent pK a of the oxidized Rieske iron-sulfur protein (pK a ϭ 7.7) (23)(24)(25)(26). Therefore, we conclude that we have identified another functionally important protonatable group of the bc 1 complex.
Mechanism of Zinc Inhibition-Binding of zinc does not interfere with the binding of either of the substrates, hydroquinone or cytochrome c, but interferes with the binding of protons. Therefore, the most likely mechanism for inhibition by zinc is interference of zinc with proton transfer reactions at the Q P center of the bc 1 complex.
Hydroquinone oxidation requires the release of two protons from the Q P center to the aqueous environment. The quinone reaction pocket itself must be shielded from the environment in order to prevent side reactions of the reaction intermediates with water or oxygen. Therefore, proton release will require a proton-conducting pathway to the aqueous phase (see Ref. 6). The pK a ϭ 7.2 of the newly identified protonatable group is optimally suited for proton conduction at neutral pH. Since binding of zinc shifts the pK a value of the proton binding site from 7.2 to below 5, channelling of protons is inhibited by binding of zinc. Electron transfer is then inhibited by preventing release of the protons liberated in the redox reaction.