Zinc Irreversibly Damages Major Enzymes of Energy Production and Antioxidant Defense Prior to Mitochondrial Permeability Transition*

Recent observations point to the role played by Zn2+ as an inducer of neuronal death. Two Zn2+ targets have been identified that result in inhibition of mitochondrial respiration: the bc1 center and, more recently, α-ketoglutarate dehydrogenase. Zn2+ is also a mediator of oxidative stress, leading to mitochondrial failure, release of apoptotic peptides, and neuronal death. We now present evidence, by means of direct biochemical assays, that Zn2+ is imported through the Ca2+ uniporter and directly targets major enzymes of energy production (lipoamide dehydrogenase) and antioxidant defense (thioredoxin reductase and glutathione reductase). We demonstrate the following. (a) These matrix enzymes are rapidly inhibited by application of Zn2+ to intact mitochondria. (b) Delayed treatment with membrane-impermeable chelators has no effect, indicating rapid transport of biologically relevant quantities of Zn2+ into the matrix. (c) Membrane-permeable chelators stop but do not reverse enzyme inactivation. (d) Enzyme inhibition is rapid and irreversible and precedes the major changes associated with the mitochondrial permeability transition (MPT). (e) The extent and rate of enzyme inactivation linearly correlates with the MPT onset and propagation. (f) The Ca2+ uniporter blocker, Ruthenium Red, protects enzyme activities and delays pore opening up to 2 μm Zn2+. An additional, unidentified import route functions at higher Zn2+ concentrations. (g) No enzyme inactivation is observed for Ca2+-induced MPT. These observations strongly suggest that, unlike Ca2+, exogenous Zn2+ interferes with mitochondrial NADH production and directly alters redox protection in the matrix, contributing to mitochondrial dysfunction. Inactivation of these enzymes by Zn2+ is irreversible, and thus only their de novo synthesis can restore function, which may underlie persistent loss of oxidative carbohydrate metabolism following transient ischemia.

Recently a growing number of reports have linked changes in intracellular free Zn 2ϩ to pathological processes, particularly in the nervous system (1). Elevated Zn 2ϩ has been implicated in neuronal death following ischemia (2) and excitotoxicity (3). The role of Zn 2ϩ in Alzheimer disease has been controversial (4), although a recent study strongly associates elevated cortical Zn 2ϩ with Alzheimer disease diagnosis and severity of dementia (5).
One mechanistic theme that has emerged is that elevated intracellular Zn 2ϩ is correlated with mitochondrial dysfunction, loss of mitochondrial defenses, and increased production of reactive oxygen species (1, 6 -10). Although Ca 2ϩ is a recognized physiological inducer of mitochondrial permeability transition (MPT) 2 pore, the role of Zn 2ϩ in mitochondria dysfunction is a newly arising question. Zn 2ϩ is a much more potent inducer of MPT than Ca 2ϩ (i.e. it works at doses at least an order of magnitude lower than those needed for Ca 2ϩ (9,11)). Early publications demonstrated Zn 2ϩ inhibition of mitochondrial respiration (12), which was attributed to the cytochrome bc 1 center (13,14). More recently, we demonstrated that Zn 2ϩ inhibits mitochondrial matrix multienzyme complexes (15,16). Other targets have been identified (11,17), including pore-forming proteins in the outer membrane (18). There is also evidence for Zn 2ϩ import through the calcium uniporter (11,19,20), and possibly, an additional unknown import route (20).
Elucidation of plausible targets for Zn 2ϩ is complicated by the fact that the mechanistic details underlying the classical phenomenon of Ca 2ϩ -induced MPT itself are unclear. The composition of the pore complex is still disputed (21,22). In this work, we addressed the question of Zn 2ϩ targets in MPT using a quantitative approach that is well established for the study of enzyme mechanism. We employed analysis of the combined effects of Ca 2ϩ and Zn 2ϩ on the time course of mitochondrial swelling, whereas varying the concentrations of both metals. This "enzyme kinetics" approach has been complemented by the direct assay of enzymatic activities of flavin-dependent thiol reductases in the mitochondrial matrix during the course of metal-induced swelling. Kinetic analysis provides evidence for both independent and competing sites of action for Ca 2ϩ and Zn 2ϩ that contribute to MPT induction. We also present direct evidence for the impairment of these major enzymes of energy production and antioxidant defense in mitochondrial *This work was supported by National Institutes of Health NINDS Grant NS38741 (to A. M. B.). 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. 1  matrix in the course of Zn 2ϩ -induced pore opening, which is distinct from the mechanism of Ca 2ϩ -induced pore opening.
NADPH and NADH were freshly prepared as 10 mM stock solutions in water. Oxidized glutathione and DTNB were freshly prepared as 30 and 60 mM stock solutions in water and ethanol, respectively. Lipoamide (LA) was dissolved in Me 2 SO (25 mM) and was used to make 0.2-2 mM LA solutions in 50 mM Tris-HCl buffer, pH 7.5. ZnCl 2 was made as a 10 mM stock solution in water.
Mitochondria Preparation and MPT Studies-Rat liver mitochondria were isolated as described earlier (23). Mitochondrial suspensions (50 -60 mg of protein/ml) were kept on ice for no longer than 6 h prior to MPT and/or enzymatic activity measurements.
MPT experiments were performed with 2 mg/ml mitochondria suspended in 0.25 M sucrose, 5 mM potassium HEPES, pH 7.4, and 5 mM potassium succinate as a respiration substrate, unless otherwise stated. 100-l aliquots were placed into a microplate, and the changes in absorbance at 540 nm were recorded using a plate reader, as described below. The temperature was maintained at 26°C. Concentrations of Ca 2ϩ and Zn 2ϩ were varied within the range of 1-64 and 0.3-9 M, respectively.
Enzyme activities were measured in parallel with absorbance monitoring. At varied intervals, aliquots were removed, and enzyme activities were measured as described below. In experiments with TPEN and EDTA chase, mitochondria were loaded with Zn 2ϩ , and then 1-ml aliquots were transferred into wells containing EDTA or TPEN after 1-, 2-, 3-, and 4-min incubations. In experiments with Ruthenium Red, it was added to mitochondria simultaneously with Zn 2ϩ loading.
Enzymatic Assays-All enzymatic assays were performed in a 96-well plate reader (SpectraMax Plus, GE Healthcare) with a 200-l reaction mixture per well at 20°C. The absorbance of DTNB (⑀ 412 ϭ 13.6 mM Ϫ1 cm Ϫ1 (24)) was monitored. The light path was 0.43 cm. All reactions were run in 0.2 M Tris-HCl, pH 7.4, containing 5 mM EDTA and either 0.1 mg/ml alamethicin or 0.1% Nonidet P-40 to permeabilize mitochondria. (No difference in recovered activity was observed between alamethicin and Nonidet P-40 treatments.) GR ϩ TR activity was measured with 0.2 mM NADPH, 2 mM DTNB, and 0.1 mM oxidized glutathione. LADH activity was measured in a mixture containing 0.2 mM NADH, 2 mM DTNB, and 0.1 mM lipoamide. Reactions were initiated by the addition of mitochondrial aliquots (10 l for LADH activity and 50 l for TR ϩ GR activity) to the reaction buffers. Preincubation with EDTA in the presence of alamethicin did not result in higher activity. We note that DTNBbased detection has a significant advantage over monitoring disappearance of NAD(P)H at 340 nm because it provides a 4-fold increase in the absorbance change per unit reaction. In addition, we observed that the use of DTNB as a terminal electron acceptor rescues LADH from product inhibition, i.e. removes reduced lipoamide from the reaction mixture. Discrimination between the enzymes took advantage of differences in catalytic properties of the enzymes, i.e. the absence of LADH activity with NADPH, with DTNB alone; the absence of TR and GR activity with NADH; and reduced GR activity in the absence of oxidized glutathione. In all cases, a linear slope was observed for a minimum of 20 min.

Description of Microplate-based Ca 2ϩ and Zn 2ϩ Swelling
Titrations-The MPT is commonly monitored as changes in light scattering, which is especially pronounced for liver mitochondria. It is well established that changes in mitochondrial light scattering correspond to matrix swelling associated with pore opening (25). Swelling can conveniently be recorded spectrophotometrically as a decrease in the apparent absorbance at 540 nm (i.e. increased transmittance). Our laboratory introduced the use of a microtiter plate reader to collect experimental data on mitochondria swelling (23,26). The advantage of this approach is that it allows many different experimental conditions to be monitored simultaneously and directly compared. Henceforth, we will refer to absorbance changes as the primary output signal corresponding to MPT-associated mitochondrial swelling.
An example of a two-dimensional titration is shown in Fig.  1A, where two different MPT effectors are titrated simultaneously. In Fig. 1A, the far left column presents the titration against increasing concentrations of Ca 2ϩ in the absence of added Zn 2ϩ , whereas the other columns contain different fixed concentrations of Zn 2ϩ . Conversely, the highlighted row presents the titration against Zn 2ϩ in the absence of Ca 2ϩ , whereas the other rows contain either EDTA or a different fixed concentration of Ca 2ϩ . This experimental format allows simultaneous generation of a complete set of titration data for two effectors. Quantitative relationships between these effectors can therefore be determined from analysis of a single experimental run, eliminating the confounding effects of variations in reagent concentration or aging of the mitochondria.
The data presented in Fig. 1B (Zn 2ϩ only) and Fig. 1C (Ca 2ϩ only) appear superficially similar and might lead one to conclude (erroneously) that Zn 2ϩ and Ca 2ϩ are qualitatively similar, except that Zn 2ϩ is ϳ20-fold more potent Ca 2ϩ . It is true that increasing concentrations of either Ca 2ϩ or Zn 2ϩ accelerate pore opening by shortening the lag period, and both agents increase the maximum rate of swelling. However, this first impression is incomplete, as we will show below.
Kinetic Model of Swelling Curves-The sigmoidal shape of the swelling time course is characterized by the lag period and maximum slope (Fig. 1, inset). The swelling time course resembles the shape of a sigmoid (Fig. 1, inset), particularly for faster transitions. The simplest reaction model (27) for sigmoidal kinetics is a series of two consecutive steps for mitochondria transformation via at least one intermediate, which has the same absorbance characteristics as the original state. In this model, only the final state manifests itself by the changes in absorbance. This reaction model is described in Scheme 1.
The first step corresponds to the binding or uptake of Ca 2ϩ by mitochondria, presumably by the Ca 2ϩ uniporter, and may be reversible. The second step corresponds to the rate-limiting step in the unknown intermediate transformation(s) that eventually lead(s) to pore opening. Similarly, a recent study proposes that MPT is at least a two-step mechanism culminating in pore opening (28). The effector (e.g. Ca 2ϩ or Zn 2ϩ ) may alter any or all of these characteristics by influencing different steps in the transformation chain presented in Scheme 1. The lag period may correspond principally to a "pore-forming event" (28) or a "cristae remodeling event" (29). In contrast, the maximum rate of swelling will reflect contributions from both the processes leading to the initial pore-opening event and the propagation of pore opening through the mitochondrial population. This is because pore opening results in Ca 2ϩ release, which increases the concentration of effector in the first step of Scheme 1.
Large-amplitude swelling is a consequence of the pore-opening event that is conjectured to be "allor-none" in nature (30), i.e. the pore is either open or closed. Once the pore opens, the released calcium is consumed by the remaining coupled mitochondria, leading to progressively greater Ca 2ϩ overload to the coupled fraction of mitochondria. Moreover, swollen mitochondria are known to release more calcium than was added initially, further contributing to the overall Ca 2ϩ load. In the terminology of chemical kinetics, the Ca 2ϩ -induced swelling reaction is autocatalytic or product-catalyzed. No analytical solution exists for the complete system of equations describing autocatalytic reactions in chemical kinetics. Nevertheless, kinetic analysis can be applied to separate the "events" represented by different characteristics of the swelling curves. Analysis of Maximum Rates of Swelling, Mixed Competition of Zn 2ϩ and Ca 2ϩ Sites in MPT-The maximum rate of swelling, measured as the maximal rate of change of mitochondrial absorbance, has been recognized as an integral characteristic of the process and analyzed in double-reciprocal plots (31). This approach is based on the assumption that the maximum rate corresponds to the establishment of a pseudo-steady state where the concentration of M intermediate reaches the maximum, and for a short time period, its concentration can be considered constant. During this interval, the maximum rate corresponds to the maximum concentration of M intermediate multiplied by the rate constant for its final conversion (k 2 ).
In the most general case, the maximum concentration of M intermediate will be a function of both rate constants (formation and consumption) in Scheme 1. Therefore, if an inducer (such as Zn 2ϩ ) influences both rate constants, the maximum rate may be a complex function of the inducer concentration. Nevertheless, double reciprocal treatment may point to the relationships between two co-varied inducers. Fig. 2 presents double-reciprocal plots for the maximum rate of swelling. The plots indicate that both Zn 2ϩ and Ca 2ϩ influence the maximum rate (intercept on the y axis) and apparent "binding constant" (intercept on the x axis) for the site(s) that regulate the swelling process. Stated in other words, the two metals compete in stimulating the swelling process but do not fully substitute for each other. The activation constants for the competitive Zn 2ϩ and Ca 2ϩ site, determined from the intersection points, are equal to 0.25 and 5 M, respectively. This corresponds to the 20-fold more potent effect of Zn 2ϩ when com- pared with Ca 2ϩ in MPT induction seen by inspection of Fig. 1, B and C.
It is well established that MPT induction involves the Ca 2ϩ uniporter (32,33). As we will show later, Zn 2ϩ also enters the mitochondrial matrix through the Ca 2ϩ uniporter. The possible identity of the competitive Zn 2ϩ and Ca 2ϩ site will be commented on in the "Discussion." Analysis of Swelling at Saturating Concentrations, Identification of Kinetically Independent Zn 2ϩ and Ca 2ϩ Sites-An enzyme kinetics approach that can help to untangle the relationship between two activators or inhibitors is to study the effect of varying one in the presence of the saturating concentration of the other. If the effectors (i.e. Ca 2ϩ and Zn 2ϩ ) compete for a single "physical site" (which in the case of MPT may be the Ca 2ϩ uniporter, a component of the permeability pore, or an unknown matrix enzyme/protein), this site can be completely occupied by saturating concentrations of either metal. No effect will be observed upon the addition of the second effector if the concentration of the first one was saturating. In contrast, if the effector sites are independent, the addition of the second effector at the saturating concentration of the first one will still result in a pronounced change on the time course curve.
In the case of MPT, the concentration of the effector is considered saturating if its subsequent addition causes no major changes in the time course of swelling. In the case of Ca 2ϩ , the saturated concentration range starts at ϳ30 M, whereas for Zn 2ϩ , it is above 3 M.
The existence of two distinct sites in MPT for Ca 2ϩ and Zn 2ϩ was visualized by the following approach. Varying Zn 2ϩ in the presence of saturating Ca 2ϩ (Fig. 3A) results in lag period shortening with the maximum slope lines that are nearly parallel. In contrast, varying Ca 2ϩ in the presence of saturating Zn 2ϩ (Fig.  3B) results in the increase of the maximum slope with no changes in the lag period duration (see the slope lines intersecting at the same time point). This qualitative difference in the concentration-dependent behavior is evidence for the presence of distinct, kinetically independent effector sites Ca 2ϩ and Zn 2ϩ in MPT. We interpret this qualitative difference in the  following manner. Zn 2ϩ speeds up step(s) that precede largeamplitude swelling with a resulting shortening of the lag period. On the other hand, Ca 2ϩ speeds up propagation of the poreopening event through the mitochondrial population, resulting in a faster swelling rate.
Irreversible Inactivation of LADH and GR ϩ TR Pools in Zn 2ϩ , but not Ca 2ϩ , MPT-An important goal for understanding the nature of the Zn 2ϩ -induced shortening of lag is the identification of molecular targets that are altered by Zn 2ϩ . We have previously observed that LADH is a target of Zn 2ϩ (16), an NADH-dependent thiol reductase. Therefore, we investigated the impact of Zn 2ϩ treatment upon NAD(P)H-dependent thiol reductase enzyme activities inside mitochondria, whereas simultaneously monitoring the time course of MPT pore opening.
We monitored the activity changes of mitochondrial LADH and GR ϩ TR activities in the course of Zn 2ϩ -and Ca 2ϩ -induced MPT. This was accomplished by exposing coupled mitochondria to Ca 2ϩ or Zn 2ϩ and then assaying residual enzyme activities at different times after the addition of effector. Rapid assay of these matrix enzymes was accomplished by adding mitochondrial aliquots directly to assay mixtures that included detergent or alamethicin, a pore-forming antibiotic peptide, to rapidly expose the matrix contents to enzyme substrates. The assay mixture also included a large excess of EDTA (0.5 mM) to chelate all free Zn 2ϩ or Ca 2ϩ and prevent inhibition by Zn 2ϩ or Ca 2ϩ that had not entered into the mitochondria prior to lysis. Because this procedure also relieved any reversible metal-dependent enzyme inhibition, the changes detected represent irreversible enzyme inactivation. Previous studies demonstrated that reversal of Zn 2ϩ inhibition of ␣-ketoglutarate dehydrogenase complex is gradual, as demonstrated by upward curvature of the reaction curves (15). Delayed recovery of activity has been avoided in the current experiments by the use of higher EDTA concentrations (0.5 mM versus 10 M). The resulting activity measurements show no curvature (see "Experimental Procedures"), in contrast to our previous observations. Inactivation of LADH (Fig. 4A) and GR ϩ TR (Fig. 4B) occurs largely prior to pore opening (Fig. 4C) and is proportional to Zn 2ϩ concentration. No consistent drop in these enzyme activities was observed (Fig. 4D) upon the addition of sufficient Ca 2ϩ to induce MPT in a similar time scale. Furthermore, inactivation of these enzymes requires that Zn 2ϩ treatment be applied to intact, coupled mitochondria. Fig. 4E illustrates that even high doses of Zn 2ϩ applied to permeabilized mitochondria result in only modest enzyme inhibition when compared with low doses applied to intact mitochondria (Fig. 4F). Taken together, these results suggest that Zn 2ϩ enters the mitochondrial matrix prior to pore opening, resulting in irreversible inactivation of LADH and GR ϩ TR. Enzyme inactivation was observed in coupled mitochondria for all respiration substrates tested (succinate, succinate/rotenone, glutamate/ malate, ␣-ketoglutarate, isocitrate). LADH inactivation is well fit by a single exponential (Fig. 4A), whereas GR ϩ TR inactivation is biphasic (Fig. 4B). The initial phase of the GR ϩ TR pool inactivates more rapidly and is more sensitive to Zn 2ϩ than LADH, although the residual GR ϩ TR activity remains higher than LADH (Fig. 4F, compare curves). The biphasic response reflects the complex composition of the GR ϩ TR pool. The order of sensitivity to Zn 2ϩ is TR Ͼ Ͼ LADH Ͼ GR. 3 Therefore, we suggest that Zn 2ϩ rapidly inactivates TR, whereas GR is less sensitive to Zn 2ϩ than LADH.
TPEN, but Not EDTA, Rescues the Intramitochondrial Enzymes from Zn 2ϩ -The enzyme inactivation observed after the addition of Zn 2ϩ to intact mitochondria might be due to direct or indirect effects of Zn 2ϩ . An example of the indirect effects of Zn 2ϩ is inhibition of the electron transport chain (12)(13)(14), which would increase matrix NADH levels and result in over-reduction of LADH. EDTA was previously shown to reverse inhibition of mitochondrial respiration resulting from Zn 2ϩ binding to Complex III (17). However, EDTA has almost no effect on the enzyme inactivation time course (Fig.  5, A and B). This suggests that the Zn 2ϩ binding site responsible for inactivation thiol enzymes is not on the mitochondrial membrane.
Direct inhibition of matrix enzymes requires Zn 2ϩ entry into the matrix. If this is the cause of inactivation of the enzymatic activities, then the enzymes should be rescued from inactiva-tion by a membrane-permeable Zn 2ϩ chelator, such as TPEN. In contrast to EDTA, TPEN stabilizes the enzyme activity to the level at the time of its addition (Fig. 5, C and  D). These results are consistent with rapid Zn 2ϩ entry into the matrix followed by direct Zn 2ϩ action on these matrix enzymes. Once Zn 2ϩ has entered the matrix, subsequent EDTA treatment is unable to prevent enzyme inactivation. However, TPEN penetrates the matrix and chelates Zn 2ϩ , thereby arresting the inactivation process.
The enzymological evidence of Zn 2ϩ entry into the matrix is in agreement with previous studies where intramitochondrial Zn 2ϩ was visualized with fluorescent indicators (8,20). The data presented above make it clear that sufficient Zn 2ϩ enters the matrix to alter critical enzyme activities.
Zn 2ϩ Competes with Ca 2ϩ for the Uniporter-As shown in Fig. 4D, inactivation of these enzymes occurs only with Zn 2ϩ but not Ca 2ϩ treatment, indicating that LADH and GR ϩ TR are exclusively targets of Zn 2ϩ . In contrast, analytical treatment of swelling curves presented above (Fig. 2) revealed a kinetic step where Zn 2ϩ and Ca 2ϩ compete. Candidates for this competitive site include ion transporters and pore components. Several studies have suggested a role for the calcium uniporter in mitochondrial uptake of Zn 2ϩ (11,19,20), based upon the inhibition of Zn 2ϩ uptake by ruthenium red.
Ruthenium Red, a Ca 2ϩ blocker (34), protected GR ϩ TR and LADH from Zn 2ϩ -induced inactivation (Fig. 6). The time course of enzyme inactivation in the presence of Ruthenium Red shows almost complete protection for LADH and partial protection for GR ϩ TR pool (Fig. 6B). Ruthenium Red also prolonged the lag period for Zn 2ϩ -treated mitochondria (Fig.  6B). Ruthenium Red protects LADH within the whole range of Zn 2ϩ concentrations studied, whereas the GR ϩ TR pool appears to be more sensitive to Zn 2ϩ concentrations above 1 M (not shown). This difference probably reflects the greater sensitivity of TR to Zn 2ϩ . 3 These results provide direct evidence for Zn 2ϩ import through the uniporter.
Relationship between Enzyme Inactivation and MPT Induction-There was a linear relationship between the characteristic time of pore opening (lag period duration) and the residual LADH or GR ϩ TR activity preserved by the mitochondria (Fig. 7). The different slopes for GR ϩ TR and LADH probably reflect the presence of easily inactivated TR in the GR ϩ TR pool. The observed linear correlation does not prove that enzyme inactivation results in pore opening but suggests that perturbation of the reducing capacity may make mitochondria  more susceptible to the sequence of events, such as oxidative damage, resulting in pore opening. Alternately, inactivation of flavin-dependent thiol-disulfide oxidoreductases is a marker of Zn 2ϩ penetration into the mitochondrial matrix. Once Zn 2ϩ has entered the matrix, it may interact with other targets that are directly relevant to the mechanism of pore opening.
Conclusions-We have shown that Zn 2ϩ can be imported through the Ca 2ϩ uniporter into the mitochondrial matrix where it inactivates LADH, GR, and TR. Inactivation of LADH will prevent NADH (energy) production because it is the terminal component of the pyruvate, ketoglutarate, and branched chained dehydrogenase complexes. Antioxidant defense will be severely affected because LADH, TR, and GR are the major generators of reduced thiols in the matrix.
Inhibition of LADH may be the mechanism underlying the loss of ␣-ketoglutarate-stimulated respiration in the presence of Zn 2ϩ that we previously reported (15). This is supported by our preliminary observation that matrix ␣-ketoglutarate dehydrogenase complex activity is inhibited following treatment of intact mitochondria with Zn 2ϩ , with kinetics that appear to parallel the inactivation of LADH (data not shown). Zn 2ϩ is mobilized following brain ischemia/reperfusion. Therefore, Zn 2ϩ inhibition of LADH may underlie the persistent loss of flux through pyruvate dehydrogenase complex observed after transient ischemia in either heart or brain (35,36). Since pyruvate dehydrogenase complex is the entry point for carbohydrate oxidative metabolism into the Krebs cycle, Zn 2ϩ inhibition may be the cause of the well documented depression of post-infarct oxygen utilization (37).
In addition to the inactivation of LADH, Zn 2ϩ -induced irreversible inactivation of glutathione reductase and thioredoxin reductase will likely contribute to gradual oxidation of matrix thiols, consistent with the data that the drop in enzyme activities parallels the delay in MPT induction. This raises the question of whether the thioredoxin reductase/thioredoxin system is directly involved in controlling the thiol ("S"-) site in MPT (38). Future work will be directed toward the elucidation of a molecular link between the inactivation of these important enzymes by Zn 2ϩ and the opening of MPT.