Regulation of Citric Acid Cycle by Calcium *

The relationship of extramitochondrial Ca2+ to intramitochondrial Ca2+ and the influence of intramitochondrial free Ca2+ concentrations on various steps of the citric acid cycle were evaluated. Ca2+ was measured using the Ca2+ sensitive fluorescent dye fura-2 trapped inside the rat heart mitochondria. The rate of utilization of specific substrates and the rate of accumulation of citric acid cycle intermediates were measured at matrix free Ca2+ ranging from 0 to 1.2 pM. A change in matrix free Ca2+ from 0 to 0.3 p M caused a 135% increase in ADP stimulated oxidation of 0.6 mM aketoglutarate = 0.15 pM). In the absence of ADP and the presence of 0.6 mM a-ketoglutarate, Ca2+ (0.3 pM) increased NAD(H) reduction from 0 to 40%. On the other hand, when pyruvate (10 pM to 5 mM) was substrate, pyruvate dehydrogenase flux was insensitive to Ca2+ and isocitrate dehydrogenase was sensitive to Ca2+ only in the presence of added ADP. In separate experiments pyruvate dehydrogenase activation (dephosphorylation) was measured. Under the conditions of the present study, pyruvate dehydrogenase was found to be almost 100% activated at all levels of Ca2+, thus explaining the Ca2+ insensitivity of the flux measurements. However, if the mitochondria were incubated in the absence of pyruvate, with excess a-ketoglutarate and excess ATP, the pyruvate dehydrogenase complex was only 20% active in the absence of added Ca2+ and activity increased to 100% at 2 pM Ca”. Activation by Ca2+ required more Ca2+ ( K o . ~ = 1 pM) than for a-ketoglutarate dehydrogenase. The data suggest that in heart mitochondria a-ketoglutarate dehydrogenase may be a more physiologically relevant target of Ca2+ action than pyruvate dehydrogenase.

The relationship of extramitochondrial Ca2+ to intramitochondrial Ca2+ and the influence of intramitochondrial free Ca2+ concentrations on various steps of the citric acid cycle were evaluated. Ca2+ was measured using the Ca2+ sensitive fluorescent dye fura-2 trapped inside the rat heart mitochondria. The rate of utilization of specific substrates and the rate of accumulation of citric acid cycle intermediates were measured at matrix free Ca2+ ranging from 0 to 1.2 pM. A change in matrix free Ca2+ from 0 to 0.3 p M caused a 135% increase in ADP stimulated oxidation of 0.6 mM aketoglutarate = 0.15 pM). In the absence of ADP and the presence of 0.6 mM a-ketoglutarate, Ca2+ (0.3 p M ) increased NAD(H) reduction from 0 to 40%. On the other hand, when pyruvate (10 pM to 5 mM) was substrate, pyruvate dehydrogenase flux was insensitive to Ca2+ and isocitrate dehydrogenase was sensitive to Ca2+ only in the presence of added ADP. In separate experiments pyruvate dehydrogenase activation (dephosphorylation) was measured. Under the conditions of the present study, pyruvate dehydrogenase was found to be almost 100% activated at all levels of Ca2+, thus explaining the Ca2+ insensitivity of the flux measurements. However, if the mitochondria were incubated in the absence of pyruvate, with excess a-ketoglutarate and excess ATP, the pyruvate dehydrogenase complex was only 20% active in the absence of added Ca2+ and activity increased to 100% at 2 pM Ca".
Activation by Ca2+ required more Ca2+ ( K o .~ = 1 pM) than for a-ketoglutarate dehydrogenase. The data suggest that in heart mitochondria a-ketoglutarate dehydrogenase may be a more physiologically relevant target of Ca2+ action than pyruvate dehydrogenase.
It is generally accepted that movement of Ca2+ across the inner mitochondrial membrane occurs through separate influx and efflux pathways (1-3) and that the relationship between the concentration of intra-and extramitochondrial Ca2+ is determined by the relative activity of these two pathways. Moreover, recent observations show that the physiological concentration of matrix free calcium is in the sub-micromolar range and that the activity of three matrix enzymes, pyruvate * This work was supported in part by Grants 5P01-HL18708-13, 5R01-HL36948-02, and R01-DK29740-08 from the National Institutes of Health (to K. F. L.), research grants from the Whitaker Foundation and Juvenile Diabetes Foundation (to J. Y . C.), and a National Kidney Foundation young investigator award (to R. C. S.).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ To whom correspondence should be addressed: Dept. of Surgery, dehydrogenase, isocitrate dehydrogenase, and a-ketoglutarate dehydrogenase, are strongly activated by increases in Ca2+ in that range (4, 5). Each of these enzymes catalyzes a reaction that is displaced far from equilibrium and hence their modulation by Ca2+ could contribute significantly to the overall flux of carbon in the citric acid cycle. Recent suggestions that Caz+ rather than ADP may control generation of NADH and thus overall flux in the citric acid cycle under certain substrate conditions (6, 7) prompted this investigation of the relative potency of Ca2+ and ADP and the relative importance of the three Ca2+dependent dehydrogenases in regulating cardiac respiration.
It is important to know which of the three dehydrogenases are physiologically controlled by Ca2+ because the metabolic consequences of Ca2+ control over each is different. Thus, Ca2+-mediated control of pyruvate dehydrogenase would limit the oxidation of glucose, lactate, and pyruvate but not the oxidation of free fatty acids or ketone bodies. On the other hand, if isocitrate dehydrogenase were responsible for rate limitation, in the absence of Ca", feedback control via citrate inhibition of citrate synthase (8) would inhibit oxidation of all substrates which generate acetyl-coA. Since Ca2+ effects only the E(, of a-ketoglutarate dehydrogenase (9, IO), which is in the middle of the citric acid cycle, citric acid cycle flux would not necessarily decrease at low Ca2+ levels but simply operate at a higher steady-state level of a-ketoglutarate. Our recent studies indicate that the steady-state level of a-ketoglutarate can, however, modulate glutamate oxidation to aspartate, because a-ketoglutarate is a competitive inhibitor of oxalacetate for aspartate aminotransferase (11). This in turn could modulate the rate of utilization of cytosolic NADH via the malate aspartate cycle. Conclusions concerning the physiological importance of Ca2+ in control of citric acid enzyme activity was hampered initially by the inability to directly measure mitochondrial free Ca2+. This difficulty has been overcome by the introduction of the Ca2+-sensitive fluorescent probes fura-2 and indo-1 (12-17) and the realization that these indicators may gain access into the mitochondrial compartment under suitable conditions.
Although two recent studies have attempted to correlate the activity of either aketoglutarate dehydrogenase (15) or pyruvate dehydrogenase (14) with cytosolic free Ca2+, flux determinations and comparisons between the effects of ADP and Ca2+ have not been reported previously.

EXPERIMENTAL PROCEDURES
Mitochondrial Isolation-Mitochondria were isolated using a modification of the procedure originally described by Chance and Hagihara (18). The modification was described recently (19) and is specifically designed to obtain mitochondria with very low levels of endogenous Ca2+. Briefly, after anesthetizing male Sprague-Dawley rats (250-400 g) with 150 mg/kg pentabarbital, hearts were removed with a portion of the aorta intact. After cannulating the aorta, retrograde perfusion of the cardiac tissue was performed using 25 ml of an ice-cold solution of 225 mM mannitol, 75 mM sucrose, 5 mM MOPS' and 0.1 mM EGTA (MSE). The perfusate was then changed to a similar one containing 0.3 mg/ml Nagarse (a commercial mixture of proteolytic enzymes). After flushing the heart with 6 ml of the proteasecontaining solution, the tissue was minced and rinsed with MSE (no Nagarse) and mitochondria isolated by standard techniques of homogenization and differential centrifugation. Mitochondria isolated by this technique have less than 1 nmol of Ca2+/mg, are unusually stable during incubation in the range 20-37 "C and have low levels of M$+-stimulated ATPase.
Measurement of Free Caz+ in EGTA Buffers-The free Ca2+ concentrations in incubation solutions were determined using calcium electrodes constructed with the calcium ionophore ETH129 (20). This ionophore allows construction of electrodes with a linear response t o M free Ca" and is thus more sensitive to the more commonly employed ionophore ETHlOOl (21). Electrodes were constructed as described previously (21,22) and were calibrated using the method of Bers (23). Output voltage between the Ca2+ electrode and a reference calomel electrode was fed into an Orion model 811 pH meter and recorded with a Kipp & Zonen model BD41 chart recorder.
Fluorescence Measurements-Free Ca2+ in the mitochondrial matrix was determined by measuring the fluorescence of fura-2-loaded mitochondria with a SPEX dual wavelength fluorometer. The fluorescence emission (510 nm) of fura-2 loaded mitochondria was measured during excitation of the sample with light alternating between 340 and 380 nm at 25 Hz. Since changes in mitochondrial calcium occur at considerably slower rates, the observed ratio between the emission intensity obtained with 340 nm excitation relative to 380 nm excitation can be considered instantaneous. After subtracting the background fluorescence of unloaded mitochondria at each wavelength, this ratio, R, was used to calculate free Ca2+ in a manner that is independent of the degree of mitochondrial dye loading (24). R is related to free Ca2+ by the following equation: where, Kf, = the apparent dissociation constant of Ca2+ for fura-2; R = the ratio of emission intensities obtained from excitation at 340 nm relative to excitation at 380 nm; R, , = R when all of the dye is Ca2+-bound; Rmin = R when all of the dye is in the unbound (free) form; S~S S O = the fluorescence proportionality coefficient of the Ca2+bound dye at 380 nm excitation; and Sf3m = the fluorescence proportionality coefficient of the free dye at 380 nm excitation. Rmin was determined by incubation of dye loaded mitochondria with 1 mM EGTA, 3 pM bromo-A23187, 2 ng/mg nigericin, and without added calcium. Rmax was similarly determined using a 1 mM EGTA-calcium buffer containing >30 p~ free Ca2+. Since Rmax varied considerably between individual preparations of loaded mitochondria, it was measured in each batch and individual values of Rmax were used in each determination of free matrix CaZf. Autofluorescence at 340 and 380 nm excitation of unloaded mitochondria was determined in each batch and subtracted from fluorescence values at 340 and 380 nm of loaded mitochondria. When 10 mg/ml mitochondria were loaded with 10 p M fura-Z/AM (1 nmol/mg protein), autofluorescence was 8-10% of the fluorescence of fura-2-loaded mitochondria. The Kf, of the intramitochondrial dye for Ca2+ was determined using bromo-A23187 (3 p M ) and nigericin (2 ng/mg) to remove permeability restrictions to K+, H+, and Ca2+. R of ion permeable mitochondria was measured in buffers of known Ca2+ concentration, as measured by Caz+ selective electrode.

~~
Mitochondrial Incubations and Flux Determinations-Pyruvate dehydrogenase flux was determined isotopically. Mitochondria (1-3 mg/ ml) were incubated at 28 "C, pH 7.0, in closed vessels containing 1 ml of medium B containing 5 mM ATP or ADP, 1 mM malate, and either with or without 0.632 p~ free Ca2+. Lactate dehydrogenase (100 units/ml), 0.1 mM NAD, and 5 mM [l-"C]lactate or 5 mM [ l -"Clpyruvate (4 pCi/ml) were added to the incubation mixture. Small plastic cups, containing hyamine hydroxide, were hung from the vessel lid to collect "COZ. Reactions were terminated at various time points by injection of 1.0 ml of 2 N sodium acetate, pH 3.4, into the incubation media. After 30 min of further incubation, the plastic cups were removed and added to scintillation vials. Corrections for counting efficiency were made and pyruvate dehydrogenase flux was calculated from the disintegrations/min l4COZ collected and the specific activity of added pyruvate or lactate.
The rates of accumulation of acetyl-coA, citrate, and a-ketoglutarate were measured in samples of the reaction mixture taken at 0, 2, 4, and 6 min. Details of this method are presented elsewhere (25). Oxygen consumption was determined polarographically in separate incubations a t 28 "C with a Clark electrode assembly.
Metabolite Assays-Acetyl-coA, citrate, a-ketoglutarate, and malate were assayed enzymatically by fluorometric techniques (26). The fluorescence of endogenous NADH was measured in mitochondria not loaded with fura-2 using a n Eppendorf fluorometer. The excitation light was transmitted through a 360 f 10 nm band-pass filter and the emission filter was 400-3000 nm.
Pyruvate Dehydrogenase Actiuity-Mitochondria (1.5-2.0 mg/ml) were incubated in medium B (containing 1 mM malate) with free Ca2+ levels ranging from 0 to 2 p~. Incubations were carried out at 28 "C for 5 min. In some cases, the ATP in medium B was replaced with 5 mM ADP. Additional substrates provided were either 0.1 or 5.0 mM pyruvate plus 1 mM malate or in the absence of pyruvate, 5 mM aketoglutarate. Following incubation, mitochondria were separated by centrifugation and the pellet was frozen in liquid Nz. Pyruvate dehydrogenase was extracted with 50 mM Tris buffer (pH 7.0) containing 5 mM EDTA, 1 mM dithiothreitol, 5 mM pyruvate, and 50% glycerol. Pyruvate dehydrogenase activity was assayed spectrophotometrically by coupling the formation of acetyl-coA from pyruvate to acylation of p-(p-aminopheny1azo)benzenesulfonic acid (27). Total activity of pyruvate dehydrogenase (active + inactive forms) was assayed after conversion of inactive complex to active complex by incubation of mitochondria for 15 min with FCCP in the absence of respiratory substrates. One unit of enzyme converts 1 pmol/min substrate into product at 30 "C. Arylamine acetyltransferase was partially purified from acetone powders of pigeon livers.
Materials-Lactate dehydrogenase from rabbit muscle, ATPase derived from potatoes (apyrase), and other enzymes were purchased from Sigma. Fura-Z/AM and fura-2 were purchased from Molecular Probes. Materials including ETH-129 used in fabrication of Ca2+sensitive electrode were purchased from Fluka. FCCP was the generous gift of Dr. Peter Heytler, Dupont de Nemours Co.

RESULTS
Calibration of Fura-2 Signals and Kf, Determination-sev-era1 studies have demonstrated that isolated heart and liver mitochondria can be loaded with the Ca2+-sensitive dyes, fura-2, o r indo-1 by incubation with the membrane-permeant acetoxymethyl esters (12)(13)(14)(15)(16)(17). These studies demonstrate that the trapped dyes are localized within the matrix space of the mitochondria. Our experience with this technique confirms these observations. Fura-2-loaded mitochondria were incubated in medium B except that the EGTA concentration was 20 PM and sodium was omitted. Fig. lA shows that Ca2+ addition caused a large increase in the measured R value, reflecting uptake of Ca2+ by the mitochondria. Subsequent addition of EGTA (5 mM) to the medium had little effect on R, indicating that most of the fura-2 was trapped within the mitochondria. Addition of Na+ slowly decreased R, consistent with activation of Na+-Ca2+ exchanger. Addition of the ionophore bromo-A23187 (3 p~) in the presence of excess extramitochondrial EGTA (5 mM) decreased R rapidly to values below base line, since the ionophore equilibrated intra-and extramitochondrial free Ca2+. On the other hand, if bromo-A23187 was added prior to Na+ and EGTA ( Fig. l B ) , R did not decrease in response to Na' addition but showed the expected rapid decline when EGTA was subsequently added. Taken together, these data demonstrate that fura-2 was localized within the mitochondrial matrix. An ) yields a straight line with a slope equal to l/Kb. Fig. 2 illustrates the data obtained at pH 7.7 using mitochondria containing either 0.6 or 5 nmol/mg fura-2. The alkaline pH was used because the ionophores also collapse the mitochondrial pH gradient, and this is the likely intramitochondrial pH under these conditions. The data were linear with respect to free Ca2+ and the Kh was independent of the degree of loading and equal to 0.379 k 0.019 pM. This value was used in subsequent experiments and agrees well with the value (0.312 p~) obtained by Reers et al. (13) for intramitochondrial fura-2 in heart mitochondria and differs only slightly from the value reported in the literature for free fura-2 in solution fura-2/AM for 20 min. Mitochondria were then washed and resuspended in medium B (pH adjusted to 7.70) containing various amounts of CaC12 to achieve media free Ca2+ levels (measured with Ca2+-selective electrode) as indicated in the figure. Bromo-A23187 (3 p~) and nigericin (2 ng/mg) were added to equilibrate Ca2+ and pH gradients across the mitochondrial membrane. After 5 min of equilibration, R at every free Ca2' concentration was measured. Kb was obtained from the plot of [ ( R -Rmd/(Rmax -R)].(SfdSbz) uersus free Ca2+ (see "Experimental Procedures"). Inset shows the same plot at low free Ca2+ range. By monitoring fluorescence emissions (510 nm) at excitation wavelength 365 nm, it was found that mitochondria exposed to 100 p~ fura-2/AM contained 5.5 times the amount of intramitochondrial fura-2 than those exposed to 10 p M fura-2/AM. Values represent the mean f S.E. of three separate experiments. at pH 7.0 (24) (0.224 PM) and from that reported by Gunter et al. (17) for fura-2 in liver mitochondria.
Determination of Ca2+ Gradient across the Mitochondrial Membrane as a Function of Nu+, Mg2+, and ADP-Recent studies (13,14) of matrix free Ca2+ in heart mitochondria using the trapped dye technique have indicated large (10-fold) inverse gradients (oukin) of Ca2+ across the mitochondrial inner membrane especially when the external Ca2+ is in the physiological submicromolar range. The gradient diminishes sharply with increasing external Ca2+. Although previous studies have documented the inhibitory effect of M e on Ca2+ entry ( K i = 30 p~) (28) and the stimulatory effect of Na+ on Ca2+ efflux (KO = 4-5 mM) (29), no previous studies have investigated in a systemic way the effect of Na+ and Mg2+ on the Ca2+ gradient across intact mitochondrial membranes. Moreover, recent studies of Azzone et al. (30) suggest that ADP may inhibit efflux of Ca2+ from mitochondria more than it inhibits Ca2+ entry, thus potentially altering the gradient as a function of external ATP/ADP ratio. Therefore, the effect of ADP on the ea2+ gradient was also monitored under approximately physiological conditions. The minimal effect of ADP is shown in Fig. 3. The effect of varying Na+ is shown in Fig. 4A. As in Fig. 2, the relationship of internal to external Ca2+ is sigmoidal and Na+ decreases matrix Ca2+ and its effect saturates at 5 mM Na'. The sigmoidicity of these curves might have been anticipated, because the ea2+ uniporter kinetics are sigmoidal (second order) with respect to Ca2+. At 5 mM Na+ the inverse gradient diminishes to nearly one when external free Ca2+ is 1 pM. Fig. 4B shows the effect of M$+. M e as opposed to Na+, changes the sigmoidal nature of the internal/external relationship. Indeed, in the absence of added external Mg2' the relationship appeared hyperbolic and an inverse gradient was not observed. It is important to note that the concentration of added M e is not equivalent to free Mg2' due to the presence of 5 mM ATP which binds M$+. Thus, the data are in approximate agreement with previous kinetic determinations (28)  In further studies of the effect of matrix Ca2+ on citric acid cycle flux, data were gathered using medium B which contains 5 mM Na+, 5 mM M$+, and 5 mM ATP. In some cases (State

3) ATP was replaced with ADP.
Effects of Free Ca2+ on Mitochondrial Metabolism-Using fura-2 as an indicator for matrix free Ca2+ it is possible to assess the effects of intramitochondrial free Ca2+ on mitochondrial metabolism. Control (unloaded) mitochondria and mitochondria loaded with 0.6 nmol of fura-2/mg were incubated in media B with various amounts of added ea2+ plus 5 mM ADP. Free Ca2+ ranged from 0 to 1.2 p~ and the matrix free ea2+ was measured in each buffer. In order to identify Ca2+-sensitive enzymes in functioning mitochondria, O2 consumption by fura-2 loaded mitochondria was measured when incubated with different substrate combinations at each level -.  of free Ca2+. Unloaded and fura-2 loaded mitochondria displayed the same respiration rates under similar incubation conditions (but see Ref. 16). This indicated that fura-2 does not inhibit citric acid cycle enzymes, does not affect Ca2+ gradient across the membranes or coupling of oxidation to phosphorylation. When mitochondria were incubated with various concentrations of pyruvate, altering matrix free Ca2+ appeared to have little effect on pyruvate oxidation (Fig. 5A). Doublereciprocal plots indicated that increasing matrix free Ca2+ primarily elevated Vmax with little to no effect on the K, of pyruvate dehydrogenase for pyruvate (Fig. 5B). Since Ca" promotes the conversion of pyruvate dehydrogenase from the inactive (phosphorylated) to active (dephosphorylated) form (31,32), the increase in V,,, by Ca2+ is an expected finding. What is unexpected, however, is the small magnitude of the observed effect. On the other hand, when the same experiment was performed using a-ketoglutarate as substrate, a slight increase in intramitochondrial free Ca2+ from 0 to 0.3 PM had a dramatic effect on substrate oxidation (Fig. 6 A ) . The effect of Ca2+ is more clearly shown in Lineweaver-Burk plots ( 6B) which indicate that the K,, rather than the V, . , , of aketoglutarate dehydrogenase was affected. Previous workers using other techniques and broken mitochondria have also shown that the effect of Ca2+ on a-ketoglutarate is on the substrate K , (9,10). Increasing matrix free Ca2+ from 0 to 0.64 p~ decreased the apparent K,,, for a-ketoglutarate from 2.5 to 0.6 mM (Fig. 6B). Stated in another way, at 0.6 mM aketoglutarate, an increase of matrix free Ca2+ from 0 to 0.3 pM caused a 135% increase in O2 consumption (Fig. 6A).
The above experiments were performed in the presence of excess ADP and phosphate, conditions in which mitochondrial NADH and ATP levels are low. Since these metabolites may affect the ability of Ca2+ to interact with Ca2+-sensitive dehydrogenases (31), the effects of Ca2+ on enzyme activity was also measured in the presence of ATP and absence of added ADP. This was achieved by measuring endogenous NADH fluorescence of intact mitochondria exposed to different concentrations of extramitochondrial free Ca", a technique previously employed (15,34) to assess effects of Ca2+ on a-ketoglutarate dehydrogenase. Mitochondrial dehydrogenases produce NADH when they oxidize substrates and thus the steady-state level of NADH provides an index of dehydrogenase activity. To achieve a steady state with respect to NADH, mitochondria were incubated for 4 min in the presence of the specific substrate prior to measuring NADH fluorescence. If Ca2+ activates a particular dehydrogenase in the presence of its specific substrate, the effect should be observed as an increase in NADH fluorescence. These experiments were performed in mitochondria not loaded with fura-2, since fura-2 interferes with measurement of endogenous NADH. The results, illustrated in Fig. 7, again demonstrate that pyruvate dehydrogenase was relatively insensitive to increases in matrix free Ca2+, whereas a-ketoglutarate dehydrogenase flux (at low substrate concentration of 0.6 mM) was highly sensitive. With 0.2 mM pyruvate plus 1 mM malate as substrates, increasing intramitochondrial free Ca2+ from 0 to 0.8 p~ increased the percent reduction of NAD from 42 to 65 (Fig. 7B). By contrast, in the presence of 0.6 mM a-ketoglutarate, similar increases in matrix free Ca2+ increased the percent reduction of NAD from 0 to 40 (Fig. 7A). With pyruvate and malate as substrates, the half-maximal effective concentration of matrix free Ca2+ (K&) was estimated to be 0.21 p~ in the presence of ADP and phosphate (Fig. 5A) and 0.2 p~ in the presence of ATP (Fig. 78). With a-ketoglutarate as substrate the K&, for Ca2+ was approximately 0.12 PM, both in the presence of ADP (Fig. 6 A ) or ATP (Fig. 7 A ) . The difference between the K&6 for Ca2+ when pyruvate was substrate compared to when a-ketoglutarate was substrate may not be significantly different. In fact, data to be discussed below suggest that Ca2+ stimulation observed when pyruvate was the substrate relates not to pyruvate dehydrogenase activation but to the utilization of a-ketoglutarate produced by pyruvate oxidation in the citric acid cycle.
It was not possible to use either NADH fluorescence or O2 consumption measurements to evaluate isocitrate dehydrogenase activity since citrate and isocitrate are not readily transported across the inner membrane of cardiac mitochondria (35). A different approach was used to evaluate the effect of Ca" on isocitrate dehydrogenase flux. Mitochondria were incubated with [l-'4C]pyruvate and the rate of I4CO2 production was used as an estimate of pyruvate oxidation. In parallel experiments, samples were taken for determination of acetyl-CoA, citrate, and a-ketoglutarate. In the presence of 5 mM pyruvate and 1 mM malate, increases in matrix free ca2+ from (0.04 to 0.62 WM appeared to have little effect on pyruvate oxidation, regardless of whether the mitochondria were in State 3 (ADP) or State 4 (ATP) (Fig. 8). Metabolite determinations indicated that levels of acetyl-coA (data not shown) and citrate (Fig. 9) were not affected by changes in matrix free Ca", during State 4 respiration. Since pyruvate dehydrogenase flux was not significantly affected by Ca2+ (Figs. 5A, 7B, and 8), the lack of effect of Ca2+ on citrate levels in State 4 (Fig. 9B) suggests that Ca2+ in the concentration range studied has little effect on isocitrate dehydrogenase flux. By contrast, in the absence of ADP, a-ketoglutarate levels were low and constant in the presence of Ca2+ (0.62 FM), but accumulated a t a rate about half that of [l-"C] pyruvate oxidation when matrix free Ca2+ levels were extremely low ((0.04 FM) (Fig. 10). These results suggest that in State 4 (no ADP), the most significant effects of Ca2+ were exerted via stimulation of a-ketoglutarate dehydrogenase flux. Similar results were obtained when extramitochondrial pyruvate concentrations were kept low (10 FM rather than 5 mM) by inclusion of [l-'4C]lactate (5 mM), lactate dehydrogenase (10 units) and NAD' (100 p M ) in the incubation medium. Rates of I4CO2 production were much lower (25 nmol/min.mg) in the presence and absence of ADP, but increasing matrix free Ca2+ had no affect on the rates. Acetyl-CoA levels could not be detected and much more a-ketoglutarate accumulated in the absence than in the presence of Ca2+ (data not shown).
When mitochondria were incubated under State 3 condi- tions (+ADP), citrate levels were about 2-fold higher in the absence of matrix free Ca2+ (t0.04 p~) than in its presence (0.62 p~) (Fig. 9A). This difference cannot be due to increased citrate formation from pyruvate since, as shown in Fig. 8, Ca2+ did not alter pyruvate oxidation. Rather, the calculated flux through isocitrate dehydrogenase was 10% higher in the presence of Ca2+ than in its absence. These data differ markedly from those obtained under State 4 conditions and suggest that isocitrate dehydrogenase flux was stimulated by Ca2+, but only in the presence of ADP. Since citrate is a competitive inhibitor of citrate synthase (8), Ca2+ regulation of isocitrate dehydrogenase could influence flux in the citric acid cycle. Flux in the first committed step of the citric acid cycle could be decreased if citrate levels became high enough. If this were an important consideration in the present experiment, acetyl-coA would have accumulated when citrate levels were high. However, this was not observed.
Since previous studies (31,32) indicate that conversion of pyruvate dehydrogenase from inactive to active form is modulated by increases in matrix free Ca2*, and since our data did not demonstrate a significant effect of Ca2+ on pyruvate dehydrogenase flux, we performed further experiments to reconcile these apparently contradictory findings. With pyruvate and malate as substrate (in the absence of added aketoglutarate), mitochondrial pyruvate dehydrogenase was nearly completely activated regardless of the external Ca2+ concentration or the external ATP/ADP ratio (Fig. 11). On the other hand, in the presence of a-ketoglutarate (in the absence of pyruvate), activation of pyruvate dehydrogenase in mitochondria incubated under State 4 conditions was very low when external Ca2+ was low, but progressively increased with increasing Ca2+ (Fig. 12), confirming previous findings that pyruvate dehydrogenase activation can be in part mediated by Ca2+. The lack of effect of Ca" on pyruvate dehydrogenase activation in mitochondria incubated with a-ketoglutarate in the presence of ADP may be due to low ATP levels in the matrix. This would result in depressed pyruvate dehydrogenase kinase activity and thus most of the pyruvate dehydrogenase would remain in the active form.

DISCUSSION
Most studies of the control of respiration have focused on the role of ADP, the ratio of ATP/ADP or the phosphorylation potential ATP/ADP.Pi as potential modulators of respiration. The classical studies of Chance and Williams (36) demonstrated large increases in respiration of isolated mito- chondria on addition of ADP. With few exceptions, respiration and thus mitochondrial ATP synthesis adjusts to the needs of intact tissue for ATP utilization. Thus, ATP utilization for muscle contraction is accompanied by increased O2 consumption. In a similar way, increased Na' reabsorption by the kidney causes increased renal respiration, and stimulation of ATP-consuming gluconeogenesis in the liver increases hepatic O2 consumption. Since ADP is the product of such ATP consumption, it seemed reasonable to assume that ADP, a potent in vitro stimulant of O2 consumption by isolated mitochondria, might be responsible for control of respiration in intact tissue. In the past, debate centered not on whether ADP controls respiration but on how (19,37,38). More recently, studies of cardiac respiration have questioned whether ADP does control respiration (7, 39) since changes in the concentrations of ADP and phosphate do not appear to adequately account for changes of 0 2 consumption. Moreover, changes in NADH during increased cardiac work are opposite to those expected in the presence of increased ADP (6, 7). Therefore, other modes of stimulation of cardiac respiration in response to increases in contraction have been sought. Ca2+ is a potential regulator of cardiac metabolism since it modulates muscle contraction, and simultaneous stimulation of contraction and respiration would result in the observed increases in respiration without large changes in tissue free ADP.
Certain mitochondrial dehydrogenases are known to be sensitive to Ca2+ (40), but the physiological significance of these observations were not immediately appreciated since the Ca2+ sensitivity of the dehydrogenases was in the low micromolar range, whereas early estimates based on equilibrium thermodynamic considerations (41) suggested values of 0.7 to 1.9 mM for free intramitochondrial Ca2+. Total matrix Ca2+ contents of mitochondria as usually isolated averaged 25-39 nmol/mg protein (42). Later estimates based on mitochondrial Ca2+ activity coefficient and liver mitochondrial Ca2+ content of at least 16 nmol/mg protein gave values of about 16 p~ (43). This derived value was still more than 1 order of magnitude above the range of free Ca2+ concentrations required to activate mitochondrial Ca2+-sensitive enzymes. These high matrix free Ca2+ estimates may relate to the fact that the mitochondrial total Ca2+ contents (42, 43) are much higher than the 1 nmol of Ca2+/mg mitochondrial protein measured by electron probe x-ray microanalysis of tissues frozen rapidly in vivo (44). It is known that matrix free Ca2+ increases directly with total mitochondrial Ca2+ content (45) and that total Ca2+ content in mitochondria can vary manyfold depending on isolation conditions (46).
The first direct measurement of matrix free Ca2+ using the so-called null-point titration method gave values of 0.5 and 1.5 p~, corresponding to mitochondrial total Ca2+ contents of 1 and 2 nmol/mg protein, respectively (45). Despite the critical importance of these observations on the physiological relevance of Ca2+ in regulating mitochondrial metabolism, the null-point titration technique is cumbersome, measures only steady-state values, and is inaccurate at low matrix free Ca2+ levels. The recent introduction of the Ca2+-sensitive fluorescent probe fura-2 (12, 13,[15][16][17] and indo-1 (14) and the observation that fura-2 may distribute in mitochondria in single cells (47, 48) have made it possible to monitor matrix free Ca2+ changes in response to experimental manipulations.
Since matrix free Ca2+ levels are critically dependent on total Ca2+ content of isolated mitochondria (45), our method for mitochondrial isolation (19) was specifically designed to yield mitochondria containing <1 nmol of Ca2+/mg protein. These Ca2+ content values have been previously shown to reflect those present in vivo (44,46). Particular attention was paid to the fura-2 loading conditions since overexposure of cardiac mitochondria to fura-Z/AM may result in a decrease in the dynamic range (ratio of R,,, to R,in) of the trapped dye. The optimal loading conditions established in our study (10 pM fura-Z/AM with 10 mg/ml mitochondrial protein for 20 min or 1 nmol/mg protein) resulted in Rmax/Rmin ratio of 11.1 which is still somewhat lower than that determined for fura-2 free acid in solution (19.60 f 3.20, n = 5). This is most likely due to incomplete ester hydrolysis in the matrix (49). For this reason, individual R, , , and Rmin values of each batch of mitochondria were determined and used in the calculation of free matrix Ca2+ levels.
Previous studies of matrix free Ca2+ using trapped dye techniques are in general agreement with the present studies though none of them have simultaneously compared and examined all three Ca2+-sensitive matrix dehydrogenases nor compared the effect of ADP with the effect of Ca2+ on fluxes. The recently published studies of Reers et al. (13) are in excellent agreement with those reported here, especially with respect to the Kf, of intramitochondrial fura-2 (0.37 p~) .
These authors find a somewhat steeper gradient of Ca2+ (outin) than our current results, especially in the higher range (>1.0 p~ Ca"). This is not surprising since they used a higher ratio of MF/ATP. Their conditions are different in other respects as well, since they employ perfused mitochondria attached to glass coverslips. On the other hand, Lukacs and Kapus (17) and Lukacs et al. (15) have reported the relationship of matrix Caz+ levels to a-ketoglutarate dehydrogenase activity measured at low concentrations of a-ketoglutarate in State 3 (using the NADH fluorescence method). They report a value of Ca2+ for a-ketoglutarate dehydrogenase of 0.8 p~ but use a different Kf, value (0.135 p~, assumed) for fura-2 in calculating matrix free Ca2+. Their results would be in good agreement with ours using a fura-2 KL of 0.37 pM.
However, previous studies of partially purified a-ketoglutarate dehydrogenase and of a-ketoglutarate dehydrogenase in intact mitochondria made permeable to Ca2+ reported significantly higher values (1 p~) of than those found in the present study.
The relationship of free mitochondrial Ca2+ to pyruvate dehydrogenase was recently explored by Moreno-Sanchez and Hansford (14). These authors use indo-1 as the trapped Ca2+ indicator dye and measure the relationship of free mitochondrial Caz+ to the activation state of pyruvate dehydrogenase. They find that activated pyruvate dehydrogenase in State 4 increases with matrix Ca2+ level and that the of Ca2+ is 0.3 p~. This value is lower than the value of approximately 1 p~ which we found. However, again the disagreement may lie in the measurement of the Kf, of indo-1, since the value used by these workers is almost one-third lower than that reported for the free dye in solution and was moreover determined a t neutral pH rather than the more alkaline values likely present in the intact mitochondrial matrix.
Moreno-Sanchez and Hansford (14) have also measured Ca2+ gradients across the mitochondrial membranes and, in agreement with the present study and that of Reers et al. (131, find lower values in the mitochondrial matrix than outside in the presence of 1 mM M P (but not in its absence).
Our data demonstrated that a-ketoglutarate dehydrogenase in the mitochondrial matrix is more sensitive to lower levels of Ca2+ than pyruvate dehydrogenase. The physiological significance of this is of course uncertain until measurements of mitochondrial Ca2+ can be made in in situ heart mitochondria. Since the cytosolic Ca2+ of heart cells oscillates from below 0.1 to perhaps over 1 p~ from beat to beat, matrix free Ca2+ is difficult to predict even from the present steady-state measurements of Ca2+ gradients. Moreover, numerous previous studies have demonstrated the fact that pyruvate dehydrogenase is subject to Ca2+ activation due to the Ca2+sensitive pyruvate dehydrogenase phosphatase. Nevertheless it is also clear that other modes of control of pyruvate dehydrogenase may override that of Ca2+ (50, 51). For example, although conversion of pyruvate dehydrogenase to its active dephosphorylated form is readily observed on addition of epinephrine to perfused hearts (50, 52), Hiraoka et al. (50) have shown that this occurs only when low concentrations of pyruvate are available. In the absence of additional exogenous substrates (P-hydroxybutyrate), oxidation of pyruvate may actually decrease, despite the hormone-dependent enzyme activation. However, since the normal physiological condition in vivo is one in which pyruvate concentrations are relatively low and other substrates are present, Caz+-dependent activation/deactivation of pyruvate dehydrogenase cannot be dismissed as physiologically insignificant. It is apparent from the present study, as well as those of others, that Ca2+ does not provide overriding control of pyruvate dehydrogenase (50, 51) and moreover Ca2+ control of pyruvate dehydrogenase is unlikely to control overall rates of cardiac respiration. Likewise, it seems unlikely from the present study that isocitrate dehydrogenase would provide a severe limitation for citric acid cycle flux in the absence of Caz+.
On the other hand, under some physiological circumstances Ca2+ control of a-ketoglutarate may contribute significantly to the rate of generation of ATP, which cannot be overcome even by increases in ADP. Numerous studies of liver metabolism demonstrate that Ca2+ levels may control the flux through the malate aspartate cycle (53-55) and therefore transport of reducing equivalents from the cytosol to the mitochondria. Our recent studies (11) indicate that this Ca2+ effect is mediated via the Ca2+ sensitivity of a-ketoglutarate dehydrogenase. Ca2+-mobilizing hormones cause large decreases in hepatic a-ketoglutarate levels. The decrease in aketoglutarate in turn causes significant stimulation of aspartate formation because a-ketoglutarate is a competitive inhibitor of aspartate aminotransferase (11). Since the mitochondrial enzymes (aspartate aminotransferase and a-ketoglutarate dehydrogenase) involved in Ca2+ control of the malate/ aspartate shuttle in the liver are very similar to those of the heart, a Ca2+-linked control of the malate/aspartate shuttle may occur in cardiac tissue. This would explain anomalous increases in mitochondrial NADH fluorescences following an increase in work when hearts are perfused with glucose as substrate (7). When hearts are perfused with pyruvate, whose oxidation is not dependent on oxidation of cytosolic NADH, the expected decrease in NADH fluorescence occurs (56, 57). Clearly Ca2+ stimulation of the malate/aspartate shuttle would lower lactate/pyruvate ratios in the heart and thus promote pyruvate and glucose oxidation. Ca2+ activation of pyruvate dehydrogenase at higher levels of Ca2+ would augment the effect of Ca2+ on reducing equivalent transport. Thus, these two Ca2+-sensitive dehydrogenases may work synergistically providing a two-tiered system of activation of glucose and lactate oxidation as cardiac contractility increases.