Zn2+ Inhibits α-Ketoglutarate-stimulated Mitochondrial Respiration and the Isolated α-Ketoglutarate Dehydrogenase Complex

Intracellular free Zn2+ is elevated in a variety of pathological conditions, including ischemia-reperfusion injury and Alzheimer's disease. Impairment of mitochondrial respiration is also associated with these pathological conditions. To test whether elevated Zn2+ and impaired respiration might be linked, respiration of isolated rat liver mitochondria was measured after addition of Zn2+. Zn2+ inhibition (K i app = ∼1 μm) was observed for respiration stimulated by α-ketoglutarate at concentrations well within the range of intracellular Zn2+ reported for cultured hepatocytes. Thebc 1 complex is inhibited by Zn2+(Link, T. A., and von Jagow, G. (1995) J. Biol. Chem. 270, 25001–25006). However, respiration stimulated by succinate (K i app = ∼6 μm) was less sensitive to Zn2+, indicating the existence of a mitochondrial target for Zn2+ upstream from bc 1 complex. Purified pig heart α-ketoglutarate dehydrogenase complex was strongly inhibited by Zn2+(K i app = 0.37 ± 0.05 μm). Glutamate dehydrogenase was more resistant (K i app = 6 μm), malate dehydrogenase was unaffected, and succinate dehydrogenase was stimulated by Zn2+. Zn2+inhibition of α-ketoglutarate dehydrogenase complex required enzyme cycling and was reversed by EDTA. Reversibility was inversely related to the duration of exposure and the concentration of Zn2+. Physiological free Zn2+ may modulate hepatic mitochondrial respiration by reversible inhibition of the α-ketoglutarate dehydrogenase complex. In contrast, extreme or chronic elevation of intracellular Zn2+ could contribute to persistent reductions in mitochondrial respiration that have been observed in Zn2+-rich diseased tissues.

The pool of cellular Zn 2ϩ that is not tightly bound to macromolecules or to other ligands and can be readily chelated by Zn 2ϩ -sensitive chromophores or fluorophores has been termed "chelatable zinc" (1). Interest in the biological function of chelatable Zn 2ϩ has grown steadily during the last decade. This interest was stimulated in part by the recognition that the concentration of chelatable Zn 2ϩ is elevated in some cerebral regions following episodes of transient ischemia (2,3) or excitotoxic injury (4). Elevated intracellular Zn 2ϩ has also been observed in models of cardiac ischemia and cardiac inflammation (5,6). Excess intracellular Zn 2ϩ is toxic to neurons (7)(8)(9). Various mechanisms for the toxic activity of Zn 2ϩ have been proposed including modulation of amino acid receptor activity (10 -13), alteration of nerve growth factor binding (14), induction of gene expression (15), and alterations of mitochondrial function (7,8). Elevated intracellular free Zn 2ϩ and reduced carbohydrate flux through mitochondrial energy pathways (16 -21) are prominent pathological features in some neurological and cardiac diseases.
Oxidative energy metabolism is impaired in many neurodegenerative disorders (22)(23)(24). One component of energy metabolism that has been extensively studied is the ␣-ketoglutarate dehydrogenase complex (KGDHC). 1 Reductions in KGDHC activity or protein abundance in brain occur in several neurological diseases (25)(26)(27)(28)(29)(30). The cause of this deficit has not been established. Reduced concentration and/or activity of enzymes involved in mitochondrial carbohydrate metabolism has been documented for both chronic ischemia and ischemia-reperfusion injury in heart and brain (20,21,31).
Studies on intact mitochondria have revealed that Zn 2ϩ inhibits respiration supported by combined glutamate and malate (32) or ␤-hydroxybutyrate (33). Subsequent studies identified complex III, specifically the bc 1 complex, as the site of Zn 2ϩ binding and inhibition (34,35). (Similarly, cytochrome b 562 -o complex of aerobically grown Escherichia coli K12 is inhibited by Zn 2ϩ (36).) These studies utilized substrates that enter the respiratory chain downstream from complex I such as succinate, nonylubihydroquinone, or duroquinol (32,34,35). The choice of substrates in the earlier studies precluded detection of Zn 2ϩ inhibition of upstream dehydrogenases. Evidence of disease-linked impairment of KGDHC activity (see previous paragraph), which is part of the (Krebs) tricarboxylic acid cycle that is upstream from complex I, motivated us to examine the effect of Zn 2ϩ on earlier stages of mitochondrial energy metabolism. Because the cytosolic concentration of free Zn 2ϩ in dissociated hepatocytes is in the range of 0.6 -2.7 M (37), we tested the effect of submicromolar to micromolar Zn 2ϩ upon mitochondrial respiration.
KGDHC converts ␣-ketoglutarate (␣-KG), coenzyme A (CoA), and NAD ϩ to succinyl-CoA, CO 2 , and NADH in the presence of thiamin pyrophosphate (TTP) (38). The complex subunit structure of KGDHC has been extensively studied. The sequential activities catalyzed by KGDHC (39) are summarized below. KGDHC activity is feedback-inhibited by the reaction products NADH and succinyl-CoA (40). KGDHC is also sensitive to metals; activity is enhanced by millimolar Mg 2ϩ or micromolar Ca 2ϩ (41) but is inhibited by higher (ϾmM) concentrations of Ca 2ϩ (42,43).
In this paper, we report that micromolar concentrations of Zn 2ϩ in the assay buffer inhibit respiration of intact liver mitochondria. Respiration stimulated by ␣-KG is more sensitive to Zn 2ϩ inhibition than respiration stimulated by other mitochondrial energy substrates. Experiments with purified enzyme indicated that KGDHC is inhibited by submicromolar concentrations of Zn 2ϩ . The possible roles of Zn 2ϩ in the regulation of mitochondrial respiration and in mitochondrial pathology are discussed.

MATERIALS AND METHODS
Reagents-Sodium ␣-KG, sodium succinate, L-glutamate, malate, NADH, NAD ϩ , TPP, Tris, CoASH, dithiothreitol (DTT), fatty acid-free bovine serum albumin (BSA; A-6003), DL-6,8-thioctic acid amide (LipS 2 ), 2,6-dichlorophenol-indophenol, p-iodonitrotetrazolium violet, EDTA, ZnCl 2 , CaCl 2 , and MgCl 2 were purchased from Sigma. Whenever possible, ultrapure grades of reagents were used; all other reagents were of the highest quality available commercially. Stock solutions (100 mM) of lipoamide were prepared in Me 2 SO and used within a few hours or stored in frozen aliquots at Ϫ20°C until needed. Where required, lipoamide (i.e. -SS-form) was reduced to dihydrolipoamide (i.e. -(SH) 2 form) in situ by inclusion of 0.1 mM DTT in the reaction solution. Buffer solutions were prepared using water that was distilled and then deionized through alternating anion-cation exchangers (Barnstead), resulting in Ͼ15 M⍀/cm resistance. Absolute Zn 2ϩ concentrations were determined by flame atomic absorption analysis using a Hitachi 6100 Zeeman Background Correction atomic absorption spectrometer. Analysis of the buffers used in all assays indicated that total Zn 2ϩ was below the limit of detection (Ͻ0.08 M).
Isolation of Mitochondria-Rat liver mitochondria were isolated essentially as described previously (44 -46). All steps were carried out at 0 -4°C. Four-month-old Fisher 344 male rats were decapitated. Livers were rapidly removed and homogenized with a motor-driven Teflon pestle in buffer A (250 mM mannitol, 75 mM sucrose, 10 mM HEPES, adjusted to pH 7.4 with KOH) supplemented with 100 M K-EDTA and 500 M K-EGTA. After centrifugation at 1,000 ϫ g for 10 min, the supernatants were removed and subjected to centrifugation at 10,000 ϫ g for 15 min. The resulting pellets were washed in buffer A supplemented with 100 M K-EDTA and 500 M K-EGTA and 0.5% (w/v) fatty acid-free BSA and centrifuged at 10,000 ϫ g, followed by two additional washes in buffer A supplemented with 30 M EDTA and 0.5% (w/v) fatty acid-free BSA. Following the final wash, mitochondria were resuspended in buffer A and 5 M K-EDTA. Protein concentrations were estimated spectrophotometrically using BSA as a standard (47).
Mitochondrial Respiration Assays-Mitochondrial oxygen consumption was measured at 28°C using a Clark electrode in a computer controlled system (Hansatech, PP Systems, Haverhill, MA) as described previously with minor modification (44). Mitochondria were added to 1.7-ml chambers containing respiration buffer (270 mM sucrose, 10 mM KH 2 PO 4 , pH 7.4, 3 mM MgCl 2 ; final mitochondrial concentration, 0.8 Ϯ 0.2 mg protein/ml). Samples were preincubated with or without added ZnCl 2 for 2 min. Following preincubation, state 4 respiration (respiration in the absence of ADP) was initiated by addition of substrate (5 mM ␣-KG, 7.5 mM succinate, or 6.25 mM glutamate/malate). These substrate concentrations were saturating for state 3 respiration under the conditions used (data not shown). After an additional 3 min, state 3 respiration rates (ADP-stimulated) were determined by adding ADP to a final concentration of 0.5 mM (48). Stabilized respiration rates (i.e. 1 min after ADP addition) were determined by least square regression analysis using Hansatech software. The intervals used for determination of the rate of oxygen consumption were 1.5 min for state 4 and 2-3 min for state 3.
Preparation of KGDHC-The enzyme is sensitive to inhibition by low concentrations of Zn 2ϩ (see "Results"). Therefore, as a prerequisite for the study of the effect of Zn 2ϩ on KGDHC, the purified enzyme complex was subjected to Sephadex chromatography to remove EGTA, EDTA, and BSA (components of the storage buffer that chelate Zn 2ϩ and other divalent ions). An aliquot (100 l) of the commercial KGDHC preparation was loaded onto a 0.5 ϫ 8 cm Sephadex G-200 (Amersham Pharmacia Biotech) column equilibrated with 50 mM Tris-HCl, pH 7.4, and 0.1 mM DTT. The enzyme was eluted with 100-l aliquots of the same buffer, and fractions containing enzyme activity (fractions 6 -8) were used. Activity of the most-concentrated fractions was stable at 4°C for 3-7 days.
Enzyme and Subunit Reaction Schemes-KGDHC is composed of multiple copies of three different subunits that transfer intermediate enzyme products in an ordered fashion (39). The combined E1, E2, and E3 subunits of KGDHC catalyze the following reaction sequence in the presence of ␣-KG, TPP, CoA, and NAD ϩ .
where [Lip(SH) 2 ] and [LipS 2 ] represent the dihydrolipoamide (reduced form) and lipopamide (oxidized form) of the tethered lipoic acid prosthetic group of E2, respectively (39). The overall reaction is as follows.
Assay of the combined E1-E2 activity (Equations 1-3) can be accomplished by introducing free LipS 2 to replace E3-FAD as an electron acceptor (Equation 7). The reduced lipoamide prosthetic group is oxidized by disulfide exchange, allowing E2 to recycle while generating a pool of Lip(SH) 2 . [ Subsequent quantitation of Lip(SH) 2 is accomplished in an end point assay that measures the burst of NADH produced upon addition of the accumulated Lip[SH] 2 to a mixture of purified E3 and excess NAD ϩ .
Reduced E3-FAD then converts NAD ϩ to NADH (Equation 5) giving the following overall reaction.
Assay of E3 activity (independent of E1 and E2) is accomplished by providing exogenous Lip(SH) 2 and NAD ϩ to either KGDHC or isolated E3, as in Equation 9. Reaction 9 is freely reversible. Therefore, the reverse (diaphorase) reaction catalyzed by E3 can be monitored in the presence of NADH and LipS 2 (see below).
Activity Measurements of KGDHC and Its Components-The volume of the reaction mixtures was 200 l, except where noted. Activities were determined spectrophotometrically at 23°C using a 96-well spectrophotometric plate reader (SpectraMax 250, Molecular Dynamics, Sunnyvale, CA). NADH appearance or disappearance was measured at 340 nm (⑀ 340 nm ϭ 6.23 ϫ 10 3 ). Typically, the absorbance values of individual wells were determined at 5-s intervals. In all cases where Zn 2ϩ was present the cation was added as the chloride salt, prior to initiation of the reaction. Standard assay mixtures were as follows: (i) for KGDHC: suspension, and 50 mM Tris-HCl, pH 7.4. Zn 2ϩ was added to mixture 1 before initiating the reaction with ␣-KG. After 15 min, 100 l of mixture 1 was added to an equal volume of Mixture 2 in a microtiter well. The optical density was recorded before and after mixing. Mixture 1 in the presence of ␣-KG supports the combined E1-E2 activity (Equations 1-3 and 7) generating Lip(SH) 2 as the final product. Addition of mixture 2 promotes the consumption of Lip(SH) 2 and production of NADH (Equation 9). The rapid increase in absorbance, because of reduction of NAD ϩ after combining mixtures 1 and 2, was interpreted as a quantitative measure of the pool of Lip(SH) 2 formed by the combined E1-E2 activities. Note that EDTA was included in mixture 2 to prevent possible inhibition of E3 by chelating Zn 2ϩ carried over from Mixture 1; (iii) for E3 forward reaction: 2 mM NAD ϩ , 0.25 mM LipS 2 , 0.1 mM DTT, and 100 mM Tris-HCl, pH 7.4; and (iv) for E3 reverse reaction: 0.1 mM NADH, 0.1 mM LipS 2 , and 100 mM Tris-HCl, pH 7.4.
Measurement of Mitochondrial SDH Activity and Purified GDH and MDH Activities-For SDH activity, isolated mitochondria (ϳ20 mg/ml protein) were fractured by three freeze-thaw cycles and mixed with an equal volume of reaction buffer (50 mM sodium phosphate, pH 7.4). The standard reaction mixture contained ϳ0.5 mg/ml mitochondrial protein, 7 mM sodium succinate, 0.5 mM p-iodonitrotetrazolium violet in reaction buffer. Reaction progress was monitored as an increase of absorbance at 490 nm. For GDH, 10 mM ␣-KG, 0.5 mM NADH, 50 mM ammonium sulfate, 0.1 mM ADP, and 100 mM Tris-HCl, pH 7.6, were used. For MDH, 10 mM oxaloacetate (added as a solid immediately before assay), 1 mM NADH, and 100 mM Tris-HCl, pH 7.6, were used.
Data Analyses-All data are reported as the means Ϯ S.E. Enzyme velocities (V) were determined by regression analysis of the change in NADH absorbance with time. Apparent inhibition constants (K i app ) for enzyme data were determined by Dixon plots of 1/V plotted against the concentration of added Zn 2ϩ (49). K i app was the negative of the x intercept calculated from the linear regression coefficients determined using SIGMAPLOT (SPSS, Inc.). Regression analysis for Dixon plots of respiration data was performed using the Robust Regression routine in the NCSS 97 statistics package (Kaysville, UT). Standard errors for K i app were calculated from the product of the estimated K i app and the estimated confidence value (the root of the sum of the squares of the confidence values for the two regression coefficients). Significance between means was tested using Student's t test (two-tailed).

Effect of Zn 2ϩ on Mitochondrial Respiration-Respiration
was measured in isolated, intact liver mitochondria by monitoring oxygen consumption using succinate, glutamate/malate, or ␣-KG as substrates. Note that 3 mM Mg 2ϩ was included in the respiration buffer to protect mitochondria from induction of the permeability transition (46,50,51). Under these conditions Zn 2ϩ did not induce mitochondrial swelling, as assessed by absorbance at 540 nm. 2 Mitochondria were preincubated for 2 min in the presence of 0 -6.4 M Zn 2ϩ . Substrate was added, and 3 min of state 4 respiration was monitored prior to ADP addition (state 3 respiration). Fig. 1A illustrates that in the presence of succinate state 3 respiration was at least partly maintained at 6.4 M Zn 2ϩ . All five mitochondrial preparations tested in the presence of succinate failed to recover state 4 respiration when exposed to 6.4 M Zn 2ϩ , which may be due to uncoupling and/or ATPase activation. Glutamate/malate-stimulated state 3 respiration was somewhat inhibited by 0.4 or 1.6 M Zn 2ϩ and completely inhibited by 6.4 M Zn 2ϩ (Fig. 1B). In contrast, partial inhibition of ␣-KG-stimulated State 3 respiration was observed at 0.4 M, and complete inhibition required only 3.2 M Zn 2ϩ (Fig. 1C). The IC 50 values for each substrate (Table I) were statistically different from each other (p Ͻ 0.05). The K i app for Zn 2ϩ inhibition of respiration determined analytically (Fig. 2) also differed significantly for each substrate and was smallest for ␣-KG (Table I).
Effect of Zn 2ϩ on SDH, GDH, and MDH-The above findings suggested that there might be previously unidentified sites in the respiratory chain that are upstream from and more sensitive to Zn 2ϩ inhibition than the bc 1 complex. Possible sites of Zn 2ϩ inhibition include mitochondrial dehydrogenases associated with the oxidation of the various metabolites tested above. Therefore, the Zn 2ϩ sensitivities of the relevant mitochondrial dehydrogenases were directly tested.
Incubation of mitochondrial lysates with 100 M Zn 2ϩ resulted in a ϳ2-fold activation of SDH activity (data not shown). Purified mitochondrial MDH was unaffected by 100 M Zn 2ϩ (Fig. 3A), whereas purified GDH activity was inhibited by Zn 2ϩ (Fig. 3B) with K i app ϭ 6.1 Ϯ 0.6 M. In contrast, as detailed below, KGDHC activity was inhibited by submicromolar concentrations of Zn 2ϩ . The high Zn 2ϩ sensitivities observed for both ␣-KG-stimulated mitochondrial respiration and purified KGDHC activity (relative to the other substrates and dehydrogenases) supports the hypothesis that KGDHC is a target for intracellular Zn 2ϩ . To obtain additional evidence in support of this conclusion, detailed experiments delineating the effect of Zn 2ϩ on purified KGDHC were carried out.
Preparation of Chelator-free KGDHC-Preliminary experiments with purified KGDHC indicated that Zn 2ϩ in the range of 2-5 M inhibited enzyme activity. However, the value of the determined K i app progressively declined as the concentration of enzyme in the reaction mixture was reduced (data not shown). The enzyme storage buffer contains EDTA and EGTA, both highly avid chelators of Zn 2ϩ (52), and BSA, which has been reported to bind Zn 2ϩ with submicromolar affinity (53). Direct dilution of KGDHC in storage buffer into assays results in concentrations of 2-5 M for both EDTA and EGTA and 0.1-0.3 M BSA in the final assay mixture. Therefore, the residual concentration of chelators was high enough to interfere with the determination of inhibition constants for metal ions. EDTA, EGTA, and BSA were simultaneously removed from the enzyme preparation by rapid gel filtration chromatography on a Sephadex 200 column. As expected, the apparent K i app for Zn 2ϩ decreased after these chelators were removed from the enzyme stock. Denaturing gel electrophoretograms stained with Coomassie Blue confirmed that G-150 chromatography removed Ն97% of the BSA initially found in the enzyme preparation (data not shown). KGDHC subjected to gel filtration was used in all of the experiments presented below.
Inhibition of KGDHC Activity by Zn 2ϩ Is Concentration-de-pendent- Fig. 4A illustrates the inhibition of the KGDHC reaction by Zn 2ϩ in the presence of saturating substrates and co-factors. Zn 2ϩ addition to KGDHC results in a dose-dependent slowing in the rate of reaction product formation (Fig. 4A). The slow onset of inhibition causes a notable curvature in the reaction progress curves, presenting a dilemma in the assignment of velocity values used in the determination of K i app . In the experiments described below a standard interval of 15 min was arbitrarily chosen for calculating velocity (V av ). Dixon plots of 1/V av versus the Zn 2ϩ concentration for three different enzyme concentrations (Fig. 4B) resulted in similar intercepts on the x axis, providing an estimate of ϪK i app (49). The lack of dependence of K i app on enzyme concentration indicates the successful removal of chelators. The average K i app for Zn 2ϩ for four preparations of chelator and BSA-free KGDHC was 0.37 Ϯ 0.05 M.
The sensitivity of KGDHC to inhibition by low concentrations of Zn 2ϩ raised the possibility that KGDHC might be inhibited by residual Zn 2ϩ or another metal in an assay mixture that is free of chelators. Fig. 4C demonstrates that V av increases up to 2-fold upon addition of EDTA to the assay reaching a plateau at ϳ0.3 M EDTA. The source of this metal has not been determined.
Inhibition of KGDHC Activity by Zn 2ϩ Requires Enzyme Cycling-The gradual onset of inhibition by Zn 2ϩ , reflected by curvature during the first minutes of the reaction (Fig. 4A), is consistent with a slow binding mechanism (54). Preincubation of KGDHC with Zn 2ϩ might be expected to allow the slow binding step to take place prior to initiation of the enzyme reaction. However, preincubation of KGDHC with Zn 2ϩ in the presence of all substrates except ␣-KG for 78 min did not abolish the curvature of the reaction progress curves (data not shown). This observation indicates that substrate cycling in the presence of Zn 2ϩ is required for the development of inhibition of KGDHC.
The requirement for substrate cycling in the inhibition of KGDHC is further illustrated by the data in Table II. Preincubation of KGDHC with Zn 2ϩ for up to 78 min did not change the value of K i app when V av was determined from the initial 15-min interval that followed addition of ␣-KG. In contrast, K i app declined steadily when V av was determined for intervals that were sampled after increasing periods of substrate cycling (Table II). These data indicate that after about 1 h of cycling the calculated K i app was reduced by about one half. Partial Reversibility of Inhibition of KGDHC by Zn 2ϩ -The KGDHC-catalyzed reactions represented in Fig. 4A were allowed to proceed in the presence of 0 -5 M Zn 2ϩ for 110 min. At the end of this period, each reaction mixture was adjusted to a final concentration of 10 M EDTA, and the reaction progress was again monitored (Fig. 5A). A gradual increase in enzyme activity was apparent. Control experiments indicated that 10 M ETDA was sufficient to attain maximal reversal of KGDHC inhibition for Zn 2ϩ concentrations up to 5 M (data not shown). However, the extent of reversal of KGDHC inhibition was dependent upon the prior concentration of free Zn 2ϩ . This effect was manifested in two ways: the time to reach a linear rate of product formation was greater, and the maximal rate of product formation was lower for samples that were previously exposed to higher Zn 2ϩ (e.g. compare 0.5 and 5 M traces in Fig. 5A).
The extent of KGDHC recovery was also dependent upon the duration of exposure to Zn 2ϩ . Fig. 5B shows the recovery for reaction mixtures that were exposed to the same concentration of Zn 2ϩ (1 M) for different reaction durations prior to reversal by EDTA. The time to reach maximum reaction velocity and the magnitude of this velocity after EDTA reversal diminished as the duration of exposure to a fixed Zn 2ϩ increased.

DISCUSSION
Initial observations that Zn 2ϩ -inhibited mitochondrial respiration (32,33) led to a series of studies that culminated in the proposal that the bc 1 complex of the electron transport chain is a target for Zn 2ϩ (34,35). However, Zn 2ϩ -induced inhibition of upstream dehydrogenases was not systematically examined (see the Introduction). The coincidence of elevated intracellular free Zn 2ϩ and reduced carbohydrate flux that was reported in some disease states (16 -21) led us to hypothesize that elevated intracellular Zn 2ϩ directly inhibits NADH-producing dehydrogenases, in addition to inhibiting electron transport. This hypothesis was investigated by varying the concentration of Zn 2ϩ within the physiological range (37) while monitoring respiration of isolated mitochondria. The effect of Zn 2ϩ on the activity of selected mitochondrial dehydrogenases was also determined.
The changes in mitochondrial respiration that were induced by Zn 2ϩ are unlikely to be due to changes in mitochondrial integrity. Zn 2ϩ has been implicated in several mitochondrial changes, such as ion transport (55-57) and membrane swelling (33,56,58,59). Membrane swelling has been associated with the opening of the mitochondrial permeability transition pore, loss of membrane potential, and uncoupling of respiration (60). However, Mg 2ϩ was included in the respiration buffer. Control experiments have established that 3 mM Mg 2ϩ prevents induction of the permeability transition by Zn 2ϩ . 2 The absence of any absorbance change in mitochondrial suspensions also argues against Zn 2ϩ -induced disruption of mitochondrial membrane integrity.
Zn 2ϩ Inhibits Mitochondrial Respiration Supported by ␣-KG, Glutamate/Malate, or Succinate-The order of sensitivity to Zn 2ϩ inhibition among the various substrates tested in intact mitochondria was ␣-KG Ͼ glutamate/malate Ͼ succinate (Figs. 1 and 2; Table I). Electrons (reducing equivalents) from glutamate/malate or ␣-KG enter the electron transport chain as NADH at complex I. Electrons from succinate enter the electron transport chain as FADH 2 at complex II. Electrons from both complexes I and II then feed into the Q cycle portion of   complex III (cytochrome bc 1 complex). It was reported that purified cytochrome bc 1 is inhibited by submicromolar concentrations of Zn 2ϩ (34,35). However, the greater Zn 2ϩ sensitivity observed for intact mitochondria respiring on complex I substrates (␣-KG or glutamate/malate) than on complex II substrate (succinate) (Fig. 2) indicates that a site upstream from complex III is more sensitive to Zn 2ϩ -mediated inhibition, at least for intact mitochondria. The greater sensitivity of ␣-KG stimulated respiration suggested that KGDHC might be the most Zn 2ϩ sensitive upstream factor. We therefore compared the Zn 2ϩ sensitivity of KGDHC to other dehydrogenases that oxidize mitochondrial substrates used in this study.
KGDHC Sensitivity to Zn 2ϩ Inhibition-The activity of purified KGDHC is potently inhibited by Zn 2ϩ (Fig. 4), consistent with the high Zn 2ϩ sensitivity of ␣-KG-stimulated mitochondrial respiration. Purified GDH is ϳ15-fold less sensitive to Zn 2ϩ , and Zn 2ϩ does not inhibit SDH and MDH activities. The weaker but appreciable inhibition of mitochondrial respiration observed in the presence of substrates other than ␣-KG suggests that there may be other mitochondrial targets for Zn 2ϩ . One such possible target is the succinate transporter, which is inhibited by Zn 2ϩ in bacteria (61). Also, some inhibition of complex I by Zn 2ϩ cannot be excluded by our data.
Zn 2ϩ Inhibition of KGDHC Is Time-and Activity-dependent-Zn 2ϩ inhibits KGDHC activity with a K i app of ϳ0.4 M for the intact, purified enzyme (Fig. 4B) and K i app of ϳ1 M for intact mitochondria respiring on ␣-KG (Fig. 2C). Slow onset of Zn 2ϩ inhibition was observed for the isolated enzyme in the presence of Zn 2ϩ and ␣-KG (Fig. 4A). Slow onset and slow reversal (Fig. 5A) of inhibition is often observed with tightly binding inhibitors (54). The values of K i app reported here represent an upper limit for the true K i because of two considerations. First, Zn 2ϩ inhibition increases with time as illustrated in Table II, but the determination of K i app was based on substrate formed during the first 15 min after initiation of the reaction. A second factor that may contribute to an excessively high estimate of the K i for Zn 2ϩ is the presence of ϳ0.3 M endogenous Zn 2ϩ or other chelatable divalent ion inhibitors within the KGDHC reaction mixture (Fig. 4C). Taken together, these considerations suggest that the actual K i may be 0.1 M or less.
The K i app for KGDHC and Zn 2ϩ progressively diminishes with continued substrate cycling in the presence of ␣-KG. In contrast, preincubation in the absence ␣-KG has no effect on the value of K i app (Table II). Substrate cycling may be required because Zn 2ϩ binds to a site on the enzyme that becomes available only during turnover. For example, Zn 2ϩ might inhibit KGDHC activity by binding to the lipoyl prosthetic group of E2. The lipoyl group (i.e. [LipS 2 ]-E2) is oxidized in the resting state of the enzyme. Therefore, strong (bidentate) binding is only possible after complete reduction of the lipoyl group to a di-thiol (i.e. [Lip(SH) 2 ]-E2), which requires the presence of substrates (Equations 2 and 3). Similarly, E3 in the resting state contains a disulfide that is reduced to a dithiol by Lip(SH) 2 -E2 (62), which is available only during enzyme cycling. If Zn 2ϩ inhibition of E3 involves dithiol binding, then it necessarily requires enzyme cycling. Reduced E3 is only likely to exist when [Lip(SH) 2 ]-E2 is available, which is substrate-dependent. Other explanations for slow onset of inhibition are possible. For example, Zn 2ϩ may form an inhibitory complex with one of the enzyme products (e.g. NADH; see below) generated during enzyme cycling. Preliminary experiments of the subunit reactions suggest that both E1/E2 and E3 activities are sensitive to Zn 2ϩ inhibition. Further studies are needed to clarify the precise mechanism by which Zn 2ϩ inhibits intact KGDHC.
Partial Reversibility of Zn 2ϩ Inhibition of KGDHC-The re-covery of KGDHC activity is independent of EDTA concentration beyond that which is necessary to stoichiometrically bind Zn 2ϩ . This finding indicates that dissociation of Zn 2ϩ from the inhibited complex is a unimolecular process followed by sequestration of free Zn 2ϩ by EDTA. The dependence of recovery of KGDHC activity on both the dose and duration of exposure to Zn 2ϩ suggests that initially the inhibited enzyme complex is fully reversible but that it is subsequently converted to a second irreversible or slowly reversible form. In the terminology of Morrison and Walsh (54), the first complex exhibits "slow" inhibition, whereas the second is characteristic of "slow-tight" inhibition. The gradual onset of Zn 2ϩ inhibition and gradual recovery after addition of EDTA is characteristic of a slow binding complex. The development of stable inhibition characterized by a gradual loss of EDTA reversibility reflects the conversion from a slow to a slow-tight binding complex. The gradual development of inhibition is not simply a consequence of substrate consumption, because the fraction of substrate consumed is 5% or less. Physiological Implications-Our data demonstrate that concentrations of Zn 2ϩ reported to occur in cultured hepatocytes (37) inhibit KGDHC activity. KGDHC activity within intact liver mitochondria is partly inhibited by 0.4 M and completely inhibited by 3.2 M Zn 2ϩ (Fig. 1); an even lower range of Zn 2ϩ concentrations (Fig. 4) inhibits purified KGDHC. The estimated range of cytosolic Zn 2ϩ concentration in cultured hepatocytes (0.6 -2.7 M) (37) overlaps with the range of mitochondrial sensitivity observed in this report. The correspondence between the availability in cells and the sensitivity of mitochondria leads us to propose that modest fluctuations of intracellular Zn 2ϩ may play a physiological role in the regulation of mitochondrial energy metabolism. Inhibition of KGDHC by chronic low doses of Zn 2ϩ or high doses of short duration is readily reversible (Fig. 5). Therefore, mild or transient elevation of cellular free Zn 2ϩ levels would be predicted to transiently reduce KGDHC-dependent respiration. More severe or prolonged exposure to elevated Zn 2ϩ would result in irreversible inactivation of some or all of the available KGDHC and could lead to persistent inhibition of mitochondrial respiration. Preliminary experiments indicate that the mitochondrial pyruvate dehydrogenase complex can also be inhibited by micromolar concentrations of Zn 2ϩ . More precise definition of the potential role of Zn 2ϩ as a physiological regulator of mitochondrial oxidative metabolism will require further studies.
Pathological Implications-Changes in the distribution of free Zn 2ϩ have been described for several disorders, including hypoxia/ischemia (2, 3) and Alzheimer's disease (63)(64)(65). Zn 2ϩ influx into cells is associated with neuronal death (2, 3, 10). Moreover, direct involvement of Zn 2ϩ in cytotoxic mitochondrial changes, including free radical formation, have been suggested (7,8,66). 2 Oxidative energy metabolism is impaired in many neurodegenerative disorders (22)(23)(24). In particular, KGDHC activity in brain is reduced in Alzheimer's disease and a number of other neurodegenerative disorders (25,26,29,67), but the possible relationship of the enzyme deficiency to Zn 2ϩ levels in these conditions has not yet been explored. The decrease in KGDHC activity in Alzheimer's disease brain has been reported to exceed the decrease in KGDHC protein (67). Imbalances in the distribution of Zn 2ϩ (63)(64)(65) in Alzheimer's disease brain may contribute to the loss of KDGHC activity.
The data in the present study indicate that damage because of pathological elevations of Zn 2ϩ may occur, at least in part, through inhibition of ␣-KG oxidation at the KGDHC-catalyzed step. As discussed above, shorter exposure to relatively lower levels of Zn 2ϩ is associated with reversible inhibition of KG-DHC, and low levels of Zn 2ϩ may thus play a role in metabolic regulation. However, prolonged exposure to higher concentrations of Zn 2ϩ may lead to slow and incomplete recovery from inhibition of KGDHC activity and mitochondrial oxidative function (Fig. 5). The present findings raise the possibility that extreme elevations of Zn 2ϩ that may exist under pathological conditions lead to such severe and long-lasting inhibition of ␣-KG oxidation in mitochondria that they contribute to cell death. This possibility also needs to be tested by further experiments.
Conclusion-The data reported here raise the possibility of a previously unsuspected role of Zn 2ϩ in the normal regulation of metabolism, namely, at the KGDHC-catalyzed step of mitochondrial oxidation through the Krebs tricarboxylic acid cycle. Oxidative energy metabolism is impaired in many neurodegenerative disorders and heart disease, particularly at the step catalyzed by KGDHC. The current studies indicate that Zn 2ϩ may well play a critical role in these disorders. Further studies of the role of Zn 2ϩ in the regulation of energy metabolism in health and disease are warranted.