Zinc Is a Potent Inhibitor of Thiol Oxidoreductase Activity and Stimulates Reactive Oxygen Species Production by Lipoamide Dehydrogenase*

Submicromolar zinc inhibits α-ketoglutarate-dependent mitochondrial respiration. This was attributed to inhibition of the α-ketoglutarate dehydrogenase complex (Brown, A. M., Kristal, B. S., Effron, M. S., Shestopalov, A. I., Ullucci, P. A., Sheu, K.-F. R., Blass, J. P., and Cooper, A.  J.  L. (2000) J. Biol. Chem. 275, 13441–13447). Lipoamide dehydrogenase, a component of the α-ketoglutarate dehydrogenase complex and two other mitochondrial complexes, catalyzes the transfer of reducing equivalents from the bound dihydrolipoate of the neighboring dihydrolipoamide acyltransferase subunit to NAD+. This reversible reaction involves two reaction centers: a thiol pair, which accepts electrons from dihydrolipoate, and a non-covalently bound FAD moiety, which transfers electrons to NAD+. The lipoamide dehydrogenase reaction catalyzed by the purified pig heart enzyme is strongly inhibited by Zn2+(K i ∼0.15 μm) in both directions. Steady-state kinetic studies revealed that Zn2+ competes with oxidized lipoamide for the two-electron-reduced enzyme. Interaction of Zn2+ with the two-electron-reduced enzyme was directly detected in anaerobic stopped-flow experiments. Lipoamide dehydrogenase also catalyzes NADH oxidation by oxygen, yielding hydrogen peroxide as the major product and superoxide radical as a minor product. Zn2+ accelerates the oxidase reaction up to 5-fold with an activation constant of 0.09 ± 0.02 μm. Activation is a consequence of Zn2+binding to the reduced catalytic thiols, which prevents delocalization of the reducing equivalents between catalytic disulfide and FAD. A kinetic scheme that satisfactorily describes the observed effects has been developed and applied to determine a number of enzyme kinetic parameters in the oxidase reaction. The distinct effects of Zn2+ on different LADH activities represent a novel example of a reversible switch in enzyme specificity that is modulated by metal ion binding. These results suggest that Zn2+ can interfere with mitochondrial antioxidant production and may also stimulate production of reactive oxygen species by a novel mechanism.

A number of reports suggest that mobilization of intracellular Zn 2ϩ may play a role in cellular toxicity following ischemiareperfusion injury (1)(2)(3)(4). Aberrant Zn 2ϩ regulation has also been noted in Alzheimer's disease brain (5)(6)(7). Elevated intracellular Zn 2ϩ has been associated with loss of mitochondrial membrane potential, production of reactive oxygen species, and cell death (3,8,9). Recently our laboratory reported that Zn 2ϩ is a potent inhibitor of ␣-ketoglutarate-stimulated mitochondrial respiration (10). The effect of Zn 2ϩ on respiration was attributed to inhibition of the ␣-ketoglutarate dehydrogenase complex (10). Preliminary analysis of individual subunit activities of ␣-ketoglutarate dehydrogenase complex indicated that the lipoamide dehydrogenase (LADH) 1 displayed the greatest susceptibility to Zn 2ϩ inhibition.
LADH belongs to the family of flavin-disulfide oxido-reductases (11), which include glutathione reductase and thioredoxin reductase. LADH is an essential component of the multienzyme NADH-generating complexes ␣-ketoglutarate dehydrogenase, pyruvate dehydrogenase, branched-chain ketoacid dehydrogenase, and also glycine decarboxylase in plants (12). In these multienzyme complexes, LADH catalyzes the transfer of reducing equivalents from the bound lipoate of the neighboring transferase subunit to NAD ϩ .
LADH also catalyzes the reduction of free lipoic acid by NADH and thus helps to maintain the reducing environment of mitochondria required for defense against reactive oxygen species and formation of electron-rich structures such as ironsulfur clusters. Dihydrolipoic acid, the reduced form of lipoic acid, is a potent antioxidant that scavenges reactive oxygen species such as superoxide, peroxyl radicals, hypochlorous acid, and nitric oxide. Dihydrolipoic acid may also regenerate other antioxidants such as vitamins C and E and glutathione through redox cycling (13). Preliminary human studies indicate the efficacy of lipoic acid in treatment of cerebral ischemiareperfusion, excitotoxic amino acid brain injury, mitochondrial dysfunction, and other disorders involving free radical processes (13,14). LADH (EC 1.8.1.4) is a homodimeric molecule, each subunit (ϳ52 kDa) contains two catalytic centers. FAD is responsible for NAD ϩ /NADH reduction/oxidation and the redox-active disulfide interacts with the dihydrolipoyl cofactor that is covalently linked to the acetyl transferase component of each dehydrogenase complex. The reaction with lipoic acid/lipoamide is reversible. When the reaction occurs in the same direc-tion as within the multienzyme complex, it is referred to as the forward reaction. When it proceeds in the direction of lipoic acid reduction by NADH, it is referred to as the reverse reaction.
NAD ϩ ϩ L(SH) 2 where LS 2 and L(SH) 2 are lipoic acid and dihydrolipoic acid, respectively. LADH also exhibits catalytic activity in the reduction of oneand two-electron organic and inorganic acceptors with NADH and in the reduction of molecular oxygen as shown below.
The lipoamide dehydrogenase reaction follows a ping-pong mechanism with the intermediate formation of the two-electron-reduced enzyme (15) (Scheme 1). The latter can exist in forms containing both reducing equivalents on FAD (E FADH2 ), both reducing equivalents on thiols (E FAD(SH)2 ), or in the form, referred to as the charge-transfer complex, with one electron on a reduced thiol and the other electron shared between FAD and thiolate (E FADHSSH ). The charge-transfer complex is the major form of two-electron-reduced enzyme and has a characteristic absorption shoulder at 530 nm distinguishing it from other enzyme forms (16). Early studies concluded that under anaerobic conditions of equimolar reduction with NADH the form with two reduced thiols contributes 22% and the charge-transfer complex contributes 34% to the total two-electron-reduced pig heart enzyme. Thus, the major form (44%) contains fully reduced FAD (17). More recently it has been proposed that NAD ϩ release directly yields the charge-transfer complex. The enzyme form with fully reduced FAD is not thought to be significantly populated (18).
Based upon the current understanding of the enzymatic mechanism (11) (Scheme 1) and structure of LADH (19 -21), our working hypothesis was that the target for Zn 2ϩ inhibition is the reduced catalytic disulfide (see Fig. 1). To test this hypothesis, we have undertaken detailed studies on the effect of Zn 2ϩ on LADH-catalyzed reactions. The present paper demonstrates that submicromolar levels of free Zn 2ϩ strongly inhibit lipoamide dehydrogenase activity by interaction with the catalytic disulfide. Zn 2ϩ binding also stimulates production of reactive oxygen species by LADH. Steady-state and anaerobic transient kinetics experiments are presented that establish the mechanism of these phenomena.

Steady-state Kinetics
Lipoamide Dehydrogenase Activity-All spectrophotometric measurements were performed in a 96-well plate reader (SpectraMax Plus, Molecular Dynamics, Sunnyvale, CA) with a 200-l reaction mixture per well at 20°C. The absorbance of NADH (⑀ 340 ϭ 6,220 M Ϫ1 cm Ϫ1 (23)) was monitored. The light path was 0.43 cm. All reactions were run in 50 mM Tris-HCl, pH 7.5. Each set of experiments was performed in triplicate. All data reported are as the means Ϯ S.E.
Reverse reaction: the final concentrations of reagents were 30 nM LADH, 0.05-0.2 mM LA, 0.01-0.1 mM NADH, 0.1-10 M ZnCl 2 . To avoid severe substrate inhibition by NADH (17) the concentration range for NADH was chosen to be lower than that for LA.
Forward reaction: the final concentrations were 30 nM LADH, 0.02-0.1 mM reduced LA, 0.2-2 mM NAD ϩ , 1-12 M ZnCl 2 (final Me 2 SO concentration in individual wells was Ͻ1%). The study of this reaction in the presence of zinc is complicated because Zn 2ϩ can be coordinated by thiols, such as reduced LA and DTT. In the presence of excess DTT (1 mM) no inhibition by Zn 2ϩ was observed. To avoid chelating Zn 2ϩ with excess of thiol reagents, LA in Me 2 SO was reduced with equimolar DTT dissolved in water and then diluted into a buffer solution saturated with argon in order to avoid oxidation. Aliquots of NAD ϩ and zinc solutions were placed into the wells, and the reaction was initiated by the addition of argon-saturated buffer containing reduced lipoamide and enzyme.
LADH Oxidase Activity-Oxygen consumption was measured in a "Strathkelvin Instruments" (Glasgow, UK) multichannel apparatus model SI-928 with a Clark type electrode. Reactions were monitored in thermostatted (28°C) glass chambers fitted with special glass stoppers that contained a small injection hole (Gilson) and equipped with a magnetic stirrer ("Instech" 2060, Plymouth Meeting, PA). The electrode was calibrated in the range 0 -0.240 mM oxygen immediately prior to the experiments. Oxygen concentration was varied in the range 20 -240 M by bubbling buffer with argon to displace oxygen. The final concentrations of reagents were in the range 0.1-0.2 M enzyme, 0.01-0.5 mM NADH, 0.01-0.25 NAD ϩ , and 1-40 M ZnCl 2 . Data analyses were performed using "Strathkelvin Instruments" software SI version 2.1. Each experiment with the oxygen electrode was performed in triplicate. The rate constants were calculated as an average from three independent sets of experiments.

Transient Kinetic Studies
Anaerobic stopped-flow measurements were performed in duplicate using a High-Tech SF-61 stopped-flow rapid scan spectrophotometer (High Tech Scientific, Salisbury, Wiltshire, UK) installed in an anaerobic glove box operating under N 2 with less than 1 ppm O 2 . The temperature was controlled at 22°C with a Techne-400 circulator (Techne (Cambridge) Ltd., Duxford, Cambridgeshire, UK) with an external cooler. The enzyme solution and substrate powder were placed into the sealed serum vials and deoxygenated 1 h before being placed into the glove box. The 0.05 M Tris-HCl buffer, pH 7.5, used in all experiments was deoxygenated overnight in the glove box. NADH (0.75 mM) and ZnCl 2 (1.875 mM) stock solutions were prepared anaerobically under N 2 in the same buffer.
Initial experiments were performed in rapid-scan mode with a xenon lamp. The enzyme stock solution (15 M) was shot against rising concentrations of NADH (7.5-60 M) to determine the NADH concentration exhibiting the highest absorbance at 530 nm, which is characteristic of the two-electron-reduced LADH in the form of a charge-transfer complex (16). Based on this determination, a 1.5 molar excess of NADH over the enzyme was used to generate the maximum content of a charge transfer complex, which then was shot against varying concentrations of ZnCl 2 (7.5-75 M). The absorbance time-curves at 530 nm were fitted to exponential functions using a least-squares minimization program supplied by High-Tech Scientific.

Calculation of Free Zn 2ϩ Concentration in Tris Buffer Solutions
Since Tris is known to bind Zn 2ϩ with a binding constant of ϳ10 Ϫ4 M (24), the 50 mM Tris buffer used in our experiments is expected to sequester most of Zn 2ϩ added. The thermodynamic binding constant reported for Tris binding of Zn 2ϩ must be corrected for pH and ionic strength before free Zn 2ϩ can be calculated. The concentration of free Zn 2ϩ available for enzyme binding was determined experimentally. In calculation of Zn 2ϩ inhibition constants, a denominator of 20 was used to recalculate the available zinc concentration from the total added one (see Appendix data for details).

Zn 2ϩ Inhibition of the Reverse Lipoamide Dehydrogenase
Reaction-No changes in the mechanism of LADH reaction were found within the range of 0.05-1 M free Zn 2ϩ concentrations. The reaction continues to obey a ping-pong mechanism. In the reverse reaction, Zn 2ϩ and LA compete for one and the same enzyme form. The addition of rising concentrations of zinc results in increased slopes and an unchanged intercept in double-reciprocal plots of 1/v versus 1/[LA] (not shown). Competitive inhibition is confirmed by a clear single intersection point in the Dixon plot (see Fig. 2A), corresponding to the inhibition constant of ϳ0.15 M.
With respect to NADH, Zn 2ϩ is an uncompetitive inhibitor showing parallel lines both in double-reciprocal plots and in the Dixon plot (not shown). Thus, zinc is unable to react with oxidized enzyme. The data fit the equation Lipoamide and Zn 2ϩ compete for the same enzyme form. It is possible that Zn 2ϩ either attacks the charge-transfer complex, which then rearranges into a form with both thiols participating in Zn 2ϩ binding or directly binds to the form with reduced thiols, shifting the equilibrium toward the above enzyme form. We can speculate that the catalytic His residue paired with a highly conserved Glu residue also participate in Zn 2ϩ coordination (Fig. 1). The inhibition mechanism can be described by Scheme 1 with Zn 2ϩ binding the two-electron-reduced enzyme in the form of a charge-transfer complex or in the form with reduced thiols.
Zn 2ϩ Inhibition of the Forward Lipoamide Dehydrogenase Reaction-In the forward reaction of NAD ϩ reduction by reduced lipoamide, Zn 2ϩ demonstrates uncompetitive inhibition with respect to lipoamide (not shown). This type of inhibition suggests that Zn 2ϩ does not react with oxidized enzyme, which is consistent with the mechanism proposed in Scheme 1. However, double-reciprocal plots of 1/v versus 1/[NAD ϩ ] indicate that Zn 2ϩ affects both slope and intercept (Fig. 2B).
If the reaction follows the mechanism presented in Scheme 1, the dependence of the forward reaction rate in double-reciprocal plots should display pure competitive inhibition with respect to NAD ϩ in accordance with Equation 1.
Even if one accounts for the possible presence of oxidized lipoamide, the effect of Zn 2ϩ in accordance with Scheme 1 will be seen only on the apparent rate constant for NAD ϩ (slope), but not on intercepts as shown in Equation 2.
Thus, the model presented in Scheme 1 is not consistent with the experimental data. The effect of Zn 2ϩ on both the rate constant for NAD ϩ and the rate constant for a unimolecular step suggests that Zn 2ϩ binds to an enzyme form other than the one directly interacting with NAD ϩ . Taking into account the existence of different forms of the two-electron-reduced enzyme (18), we can introduce a unimolecular interconversion step into Scheme 1 to get Scheme 2, which satisfactorily describes the inhibition by Zn 2ϩ for both reverse and forward reactions (see also Table I). The reverse and forward reactions exhibit equal values for the inhibition constant K Zn ϭ K i (1 ϩ k Ϫ3 /k 3 ), within experimental error (see Table I). The trend toward a larger inhibition constant determined for the forward reaction from Dixon plot (Fig. 2C, 0.22 Ϯ 0.05 M) compared with that determined for the reverse reaction ( Fig. 2A, 0.15 Ϯ 0.05 M) may originate from Zn 2ϩ chelation by reduced lipoamide.
Zn 2ϩ Interaction with the Charge Transfer Complex-The charge transfer complex is clearly distinguishable from the oxidized enzyme and two-electron-reduced enzyme with reducing equivalents on both thiols (11) due to a characteristic shoulder at 530 nm. The effect of Zn 2ϩ on the spectrum of the two-electron-reduced enzyme was studied by the stopped-flow technique under anaerobic conditions (Fig. 3). ZnCl 2 addition to the charge transfer complex leads to rapid disappearance of the 530-nm shoulder and an increase in absorption at 445 nm (see inset A in Fig. 3). Thus, the Zn 2ϩ ⅐LADH complex is spectrally equivalent to the oxidized enzyme form.
The rate of the spectral change at 530 nm displays linear dependence on Zn 2ϩ concentration (inset B in Fig. 3). This indicates that the experimentally measurable step is Zn 2ϩ binding not interconversion between the charge-transfer complex and the dithiol form (Scheme 3). The determined rate constant for Zn 2ϩ interaction with the two-electron-reduced enzyme is (3.7 Ϯ 0.7) ϫ 10 5 M Ϫ1 s Ϫ1 . This value is comparable with the rate constant for the two-electron-reduced form interaction with lipoamide determined from the steady-state kinetics (3.0 Ϯ 0.1 ϫ 10 5 M Ϫ1 s Ϫ1 ). Thus, Zn 2ϩ successfully competes with lipoamide for this enzyme form.
The non-zero intercept at the ordinate axis clearly points to the reversibility of the reaction and corresponds to the rate constant for the Zn 2ϩ -complex dissociation (Scheme 3). The dissociation rate constant k Ϫi is equal to 0.040 Ϯ 0.005 s Ϫ1 . The equilibrium constant for Zn 2ϩ binding calculated on the basis of stopped-flow measurements K i ϭ k Ϫi /k i ϭ 0.11 Ϯ 0.04 M, which is close to the inhibition constants determined from steady-state kinetics ( Table I). The ratio between the inhibition constants determined from steady-state data and the binding constant determined from transient kinetics allows to estimate the ratio k Ϫ3 /k 3 as 1.0 Ϯ 0.5. Thus, the contribution of the charge-transfer complex and the form with reduced two thiols to the total pool of two-electron-reduced enzyme are comparable within experimental error.
Zn 2ϩ binding is reversible. This has been demonstrated both by dilution of assay mixtures and by EDTA addition, which completely reversed all inhibitory effects (data not shown). In addition, gel-filtration of the Zn 2ϩ -inhibited assay mixture restored the initial catalytic activity (results not shown).
Zn 2ϩ Stimulation of ROS Production by LADH-Activation of molecular oxygen by flavoproteins, yielding hydrogen peroxide (dehydrogenases/transhydrogenases, oxidases) and superoxide radical (electron transferases such as flavodoxin), is a well known process (25). After the flavoenzymes of the respiratory chain, the most abundant flavoprotein in the mitochondrial matrix (ϳ0.5% of matrix protein as calculated from Ref. 26) is lipoamide dehydrogenase. Production of hydrogen peroxide and superoxide radical catalyzed by lipoamide dehydrogenase was first reported in 1955 (27) and then confirmed in 1969 (28). More recently, identification of reaction products revealed the ratio between superoxide and hydrogen peroxide to be 1:9 (29). However, no quantitative studies to determine the oxidase reaction mechanism have been reported.
In the present work the formation of hydrogen peroxide as a major product was demonstrated by addition of catalase (Fig.  4A), which resulted in recovery of ϳ50% of the consumed oxygen via disproportionation to O 2 and H 2 O. Prior addition of catalase reduced the rate of oxygen consumption by half, providing further support that hydrogen peroxide is the major product (Fig. 4B). Addition of SOD at the end of the reaction resulted in detectable O 2 recovery (Fig. 4, A and B) indicating the generation of superoxide anion radical, which is consistent with the published report (29).
Zn 2ϩ addition increased the rate of oxygen consumption, but did not change the character or distribution of the products formed (Fig. 4A). The stimulatory effect of Zn 2ϩ was observed in the presence of NAD ϩ , although the overall reaction showed strong product inhibition (Fig. 4A). At lower oxygen concentration product inhibition by NAD ϩ was even more pronounced, SCHEME 2 but the stimulatory effect of Zn 2ϩ remained clearly evident (Fig. 4C). The dependence of the reaction rate on oxygen concentration in double-reciprocal plots shows parallel lines both in the absence and in the presence of Zn 2ϩ (Fig. 5A). It is also evident that Zn 2ϩ affects the rate constants toward NADH and oxygen in a different manner. There is little or no increase in the rate constant toward oxygen in the presence of Zn 2ϩ (Fig. 5A, compare slopes), while changes in the rate constant for NADH are more pronounced (Fig. 5A, see intercepts).
The study of the reaction in air-saturated buffer shows that the dependence of the initial reaction rate on NADH concentration in double-reciprocal plots is not linear (Fig. 5B). The break in slope at high concentrations of NADH clearly demonstrates a switch from one reaction mechanism to another at elevated NADH concentrations. As Zn 2ϩ concentration increases, the dependence on NADH concentration approaches linearity (Fig. 5B). The best fit of the data was obtained using a combination of linear and hyperbolic dependence y ϭ y o ϩ ax/(b ϩ x) ϩ cx. As Zn 2ϩ concentrations increased, no significant changes were observed in the parameters y o , b, and c, but a gradually decreases as it shown in the inset in Fig. 5B. Parameter a characterizes the increase in an intercept cut by an asymptote y o ϩ a ϩ cx with rising [Zn 2ϩ ].
The dependence of parameter a on free [Zn 2ϩ ] is perfectly fitted to a hyperbolic dependence a ϭ K a /(K a ϩ [Zn 2ϩ ]) with K a ϭ 0.09 Ϯ 0.02 M (Fig. 5B, inset). Qualitatively, the addition of Zn 2ϩ switches the reaction from a complex hyperbolic dependence to a simple linear dependence in double-reciprocal plots. The half-activation concentration (K a ) is ϳ0.09 M, which is close to the Zn 2ϩ binding constant determined from kinetic analysis of the LADH reaction. This similarity indicates the common nature of oxidase reaction activation and LADH reaction inhibition by Zn 2ϩ .

Mechanism of Oxidase Reaction Catalyzed by LADH-Let us
consider the NADH-oxidase reaction mechanism in the absence of Zn 2ϩ . Upon enzyme reduction with NADH, electrons are distributed between the FAD center, which reduces oxygen to hydrogen peroxide, and the disulfide center, which does not; thus, only a portion of the two-electron-reduced enzyme is active in the oxidase reaction. To accelerate the reaction rate further, the FAD moiety must be fully reduced. A second reduction step with NADH converts the enzyme into the form containing both reduced FAD and thiols (four-electron-reduced enzyme; Scheme 4). However, the rate constant for the enzyme reduction by the second NADH molecule (k 5 ) should be much lower than that for the first one (k 1 ) because this process is less thermodynamically favorable.
To simplify the kinetic equations the rate constants k 2, k Ϫ2 , and k 4 are assumed to be equal for the left and right cycles, respectively, in Scheme 4. Assuming k 3 /k Ϫ3 ϭ [E FAD(SH)2 ]/ [E FADHSSH ] ϭ 1, based upon the anaerobic stopped-flow results, the dependence of initial rate on NADH, NAD ϩ , and O 2 is presented by Equation 3 which after simple algebraic manipulations, can be presented in the form equivalent to that used for fitting the experimental data in Fig. 5B (Equation 4), with x ϭ 1/[NADH] and the following expressions for the fitting parameters (Equations 5-8).
Mechanism of Oxidase Reaction in the Presence of Zn 2ϩ -It can be assumed that all enzyme forms containing catalytic dithiols are capable of binding Zn 2ϩ ion (Scheme 5). Zn 2ϩ binding switches the rate for the NADH interaction from the slow two-electron-reduced enzyme rate (k 5 ) to a much faster rate (k 1 Ј). We again assume that k 2 , k Ϫ2 , and k 4 are equal for all enzyme forms, respectively. The equilibrium between Zn 2ϩbound and free thiol-reduced enzyme forms is characterized by an apparent activation constant K a . Under the above conditions, a simplified analytical treatment of the above scheme under the conditions of k 1 Ј[NADH]Ͼ Ͼk 4 [O 2 ] gives the following expression for the initial reaction rate (Equation 9).
The above equation corresponds to Equation 4 with the addition of a Zn 2ϩ -dependent factor 1/(1ϩ[Zn]/K a ) ϭ K a /(K a ϩ[Zn]) to the hyperbolic term. This exactly describes the experimental dependence obtained for parameter a Zn ϭ {(k 1 Ϫ k 5 Fig. 5B, inset) converts into a linear one (Equation 10), which is consistent with the experimental data.
The mechanistic approach developed to describe LADH behavior in the oxidase reaction is consistent with the experimental data. A number of kinetic parameters, characteristic of individual catalytic steps, can be determined from fitting parameters. The apparent binding constant characterizing the interaction of Zn 2ϩ with the pool of thiol-reduced enzyme forms  Table II. Thus, the activation effect on the LADH-catalyzed oxidase reaction by Zn 2ϩ is mainly due to a switch from the rate constant for the reduction of two-electron-reduced enzyme by the second NADH molecule (k 5 ϭ (2.9 Ϯ 0.5) ϫ 10 3 M Ϫ1 s Ϫ1 ) to the rate constant for the reduction of Zn 2ϩ -modified enzyme with NADH (k 1 Ј ϭ (4.4 Ϯ 0.5) ϫ 10 4 M Ϫ1 s Ϫ1 ). This change results in a 4-to 5-fold increase of the steady-state LADH catalytic activity in the oxidase reaction in the presence of saturating Zn 2ϩ and buffer equilibrated with atmospheric oxygen.
NAD ϩ Inhibition of the Oxidase Reaction-The mechanistic approach developed above provides a framework to study the effect of product, NAD ϩ , on the reaction mechanism. Earlier work (29) noted in passing that NAD ϩ strongly inhibited the oxidase reaction. We observed that in the presence of saturating Zn 2ϩ and atmospheric oxygen concentrations the enzyme shows no inhibition by NAD ϩ up to 0.02 mM concentrations, while in the absence of Zn 2ϩ the enzyme is strongly inhibited (compare Fig. 6, A and B). Above this threshold NAD ϩ inhibits the reaction in the presence of Zn 2ϩ , as well (Fig. 6B). Fitting of the data to Equations 4 and 9 allows determination of k 1 and k 1 Ј, respectively. While the value of k 1 Ј is the same as the one extracted from the data presented in Fig. 6A, the value of k 1 is smaller not by a factor of 2, (compare k 1 Ј ϭ (4.4 Ϯ 0.5) ϫ 10 4 M Ϫ1 s Ϫ1 and k 1 ϭ (2.1 Ϯ 0.5) ϫ 10 4 M Ϫ1 s Ϫ1 determined from Fig. 5B) but of 3.5 and is equal to (1.3 Ϯ 0.2) ϫ 10 4 M Ϫ1 s Ϫ1 . The confidence in the second determination is higher because it relies upon fitting to a larger data set (six titration curves in Fig. 6A) than the original determination (Fig. 6A, no Zn 2ϩ ).
As predicted from Equations 4 and 9, and actually observed, the presence of NAD ϩ affects parameter y 0 ϭ 1/k 2 (1 ϩ k Ϫ2 [NAD]/k 4 [O 2 ]) in a linear manner. Dixon plots (Fig. 6, C and D) of the data presented in Fig. 6, A and B confirm the linear character of NAD ϩ -inhibition with an apparent inhibition constant of ϳ70 M. Based on this value and the value for k 4 , the value k Ϫ2 can be calculated as ϳ10 4 M Ϫ1 s Ϫ1 (Table II). NAD ϩ inhibition in the absence of Zn 2ϩ (Fig. 6C) is nearly competitive with respect to NADH. This suggests competition of NAD ϩ and NADH for the same enzyme form, i.e. the twoelectron-reduced enzyme. This result is consistent with the mechanism presented in Scheme 5. In the presence of saturating [Zn 2ϩ ] and low [NADH], NAD ϩ inhibits the reaction in an uncompetitive manner (Fig. 6D) and thus, NADH and NAD ϩ compete for different enzyme forms. In other words, under these conditions the enzyme is distributed between the left and top right cycles in Scheme 5. At higher [NADH], NAD ϩ inhibits the reaction in a competitive manner indicating the involvement of the bottom right cycle in Scheme 5.
The observed degree of NAD ϩ inhibition strongly depends on available oxygen (the ratio between k Ϫ2 [NAD] and k 4 [O 2 ], see Equations 4 and 9). Under the conditions of low oxygen, the inhibition by NAD ϩ should be more severe, which is consistent with the results obtained at 30 M oxygen (see Fig. 4C). Thus, in the presence of NAD ϩ and low oxygen, ROS production by LADH will be completely suppressed, while the presence of Zn 2ϩ will switch the reaction on. DISCUSSION The current study of LADH-catalyzed lipoamide dehydrogenase and oxidase reactions provides evidence for strong effects of submicromolar free Zn 2ϩ on the catalytic activity of this important enzyme of energy metabolism. Internalization of cytosolic Zn 2ϩ into the mitochondrial matrix has not been unequivocally demonstrated. However, indirect evidence was provided for isolated mitochondria by monitoring disappearance of extra-mitochondrial Zn 2ϩ and competition with Ca 2ϩ transport   (30). Zn 2ϩ inhibition of respiration in isolated, intact mitochondria in a substrate-dependent manner (10) provides further indirect evidence of Zn 2ϩ internalization by mitochondria. The inhibition of the forward lipoamide dehydrogenase reaction suggests that elevated Zn 2ϩ concentrations inside the mitochondrial matrix will inhibit NADH production by the ketoglutarate, pyruvate, and branched-chain dehydrogenases, interrupting the Krebs cycle and mitochondrial energy metabolism. LADH inhibition may be sufficient to explain the previously observed selectivity of Zn 2ϩ inhibition for ␣-ketoglutaratedependent respiration over glutamate/malate or succinatedependent respiration in intact mitochondria (10).
Zn 2ϩ inhibition of lipoamide dehydrogenase activity in the reverse reaction may also have important physiological consequences. Elevated intra-mitochondrial Zn 2ϩ concentrations would interfere with dihydrolipoate recycling catalyzed by LADH. Interference with reduced thiol recycling could underlie the mechanism of Zn 2ϩ induction of permeability transition (31), which is sensitive to thiol redox status (32).
Zn 2ϩ is known to stimulate a burst of mitochondrial ROS production (9). The conventional model is that this is due to a blockade of the electron transport chain at the level of Complex III. It has been demonstrated that cytochrome bc 1 is inhibited by Zn 2ϩ , and two Zn 2ϩ binding sites have been recently localized in the crystal structure (33). Inhibition of bc 1 by antimycin A results in increased generation of superoxide, which is converted to H 2 O 2 (Refs. 34, 35 and references therein). As demonstrated above, LADH oxidase activity produces H 2 O 2 and superoxide directly and is strongly induced by Zn 2ϩ . A preliminary estimate of LADH-catalyzed oxidase activity in mitochondria can be made using the reported value of LADH activity in the rat liver mitochondrial matrix fraction (26) and the rate constants determined in this work (Table II). In the presence of 0.1 mM NADH, saturating oxygen, saturating Zn 2ϩ , and at 37°C, mitochondrial LADH could produce ϳ10 nmol of H 2 O 2 /min per mg of matrix protein. This corresponds to ϳ1 nmol/min per mg of total mitochondrial protein (compared with antimycin A-induced mitochondrial production of 2.5 nmol of H 2 O 2 /min/mg of protein at 37°C (36)). Thus, it is possible that under some pathological conditions LADH may contribute significantly to the overall ROS load of mitochondria.
Acknowledgments-We thank Profs. John P. Blass, Arthur J. Cooper, and Gary E. Gibson for critical reading and comments on the manuscript.

APPENDIX
Background-Under the equilibrium conditions with Tris and enzyme competing for Zn 2ϩ , the system is described by the following equilibria, To determine the constant for Tris-Zn binding in the presence of dye we consider the same equations as above. Replacing the enzyme, E, with the dye, B, we have the following expression for the ratio of the binding constants. Determination of the Zn 2ϩ -APTRA Binding Constant-The measurement of free zinc concentrations available for APTRA binding in Tris buffers of different molarity was performed varying Tris and ZnCl 2 concentrations at fixed non-saturating concentrations of APTRA. Fluorescence intensity of the mixtures of Zn 2ϩ and APTRA in Tris buffer solutions, pH 7.5, was measured using a fluorescence plate reader (Spectramax Gemini, Molecular Devices, Inc.) with excitation at 365 nm and emission at 510 nm. The final concentrations in a 200-l well volume were varied in the range of 2.5-200 mM for Tris, 2-70 M for ZnCl 2 , and 0.5-2 M for APTRA.
The use of the fluorescent dye concentrations lower than that of added Zn 2ϩ leads to saturation of the fluorescent signal with rising Zn 2ϩ concentrations. The saturation limit linearly corresponds to the dye concentration added. The dependence for different Tris concentrations was fitted to a hyperbolic function y Ϫ y o ϭ a[Zn]/(b ϩ [Zn]) (Fig. 1S, A and B).
To estimate the binding constant independent of Tris concentration, the calculated parameter b was extrapolated to [Tris] ϭ 0 (Fig. 1S, C) The extrapolated K B is independent of dye concentration used within the range of 0.5-2 M and is equal to 1.85 Ϯ 0.15 M (Fig. 1S, C). This value is within the range reported by Combining the above value together with the estimated K B and the known concentration of Tris, we calculate K Tris as 2.3 Ϯ 0.2 mM. The obtained value for K Tris results in a denominator of 23 Ϯ 3 for the recalculation of free zinc concentration from the total Zn 2ϩ concentration added to 50 mM Tris-HCl buffer, pH 7.5.
For simplicity, we have employed a factor of 20 to calculate free Zn 2ϩ concentrations.