&NADH Decreases the Permeability of the Mitochondrial Outer Membrane to ADP by a Factor of 6*

Mitochondria with intact outer membrane (99% intact based on cytochrome c impermeability) were isolated and used to measure the permeability of their outer membrane to ADP. P-NADH reduced the permeability in a concentration-dependent manner (K, = 87 * 5 p ~ ) by a factor of 6. a-NADH and P-NAD’ cannot mimic the action of P-NADH. The mitochondrial outer membranes be- come rate-limiting in the presence of P-NADH at low, physiologically relevant, ADP concentrations (e30 p ~ ) . P-NADH has been shown to increase the voltage dependence of WAC (a major pathway for metabolite transport across the outer membrane) in a reconstituted system and this may be the way it acts on the isolated mitochondria. Inhibition of P-NADH dehydrogenases does not in- hibit the action of P-NADH indicating that it is not acting by delivering reducing equivalents. The ability of fi-NADH, produced by glycolysis, to inhibit mitochondrial function by reducing the permeability of the outer membrane may be one pathway responsible for the Crabtree effect. The mitochondrial outer membrane is permeable to small molecules (M, < 5000) while being impermeable to large poly-mers (Werkheiser Bartley, Pfaff et 1968; Wojtczak and Zaluska, This selectivity based on size is character-istic of channels and can be accounted for by the properties of VDAC (Colombini, (1.2 s/point and a total of 340 points). These were calculated as follows: V,, and P*A were allowed to vary so as to achieve the best fit (mini- mized the sum of the square of the difference between the calculated points and the data points).

The mitochondrial outer membrane is permeable to small molecules ( M , < 5000) while being impermeable to large polymers (Werkheiser and Bartley, 1957;Pfaff et al., 1968;Wojtczak and Zaluska, 1969). This selectivity based on size is characteristic of channels and can be accounted for by the properties of VDAC (Colombini, 1979(Colombini, ,1980, a highly-conserved outer membrane protein (also referred to as mitochondrial porin). Treatments that close VDAC (Colombini et al., 1987;Benz et al., 1988;Liu and Colombini, 1992) have been shown to greatly inhibit mitochondrial function, especially the flux of adenine nucleotides into the mitochondrial spaces. Experiments with isolated mitochondria (Gellerich et al., 1989) and skinned muscle fibers (Saks et al., 1993) have provided evidence that the outer membrane limits the flow of adenine nucleotides. Many have proposed that the outer membrane plays a regulatory role in mitochondrial function.
In addition to VDAC, the outer membrane contains a peptide-sensitive channel that has been implicated in protein import into mitochondria (Chich et al., 1991;Thieffry et al., 1992;Fevre et al., 1993). Unlike VDAC, this channel-former prefers cations and thus its role in the flux of metabolites (most of which are anionic) through the outer membrane is unclear. The deletion of the only known VDAC gene in yeast greatly inhibits *This work was supported by Grant N00014-90-5-1024 from the Ofice of Naval Research. 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. P Recioient of a fellowshiD from the Ministw of Education of Taiwan, Republic of China.
5 Supported by the NBS Study Bureau, Medical Corps, Belgian MOD. Current address: DeDt. of Phvsiolom K. U. Leuven Medical School, Gasthuijsberg, 3000 ieuven, Belgium: growth on glycerol but growth does occur (Guo and Lauquin, 1986;Dihanich et al., 1987) indicating the existence of another pathway for metabolite flux.
The fact that VDAC's properties are highly conserved when purified from mitochondria isolated from all eukaryotic kingdoms attests to the importance of these properties in the life of eukaryotic cells. Not only are the single channel conductance, ion selectivity, and voltage-dependence highly conserved (Colombini, 1979(Colombini, ,1989 but so are VDAC's ability to respond to NADH (Zizi et al., 1994) and to a mitochondrial intermembrane space protein referred to as the VDAC modulator Colombini, 1988, 1993). When tested, treatments that influence VDAC's ability to conduct ions in reconstituted conditions also have the expected effect on isolated mitochondria with largely intact outer membranes (Colombini et al., 1987;Benz et al., 1988;Liu and Colombini, 1992). This has led to a growing sense of the physiological importance of this regulation.
The control of mitochondrial respiration has been reviewed (for example, Tager et al. (1983) and Brown (1992)). In the early studies on isolated mitochondria, the rate of mitochondrial respiration and ATP synthesis were thought to change with metabolic demand by responding to changes in extramitochondrial ADP and Pi (Lardy and Wellman, 1952;Chance and Williams, 1955). However, the physiological relevance of this concept to adult cardiac muscle has been questioned by the finding that oxygen consumption responded to alterations in cardiac work without any significant changes in ADP and Pi concentration in both in. vivo (Chance et al., 1986;Katz et al., 1989;Balaban and Heineman, 1989) and in vitro (From et al., 1990;Katz et al., 1987) conditions. Recently, NADH was shown to increase the voltage-dependence of W A C (Zizi et al., 19941, indicating that NADH might regulate the permeability of outer membrane and thus control mitochondrial function. We now report that NADH reduces the permeability of the mitochondrial outer membrane to adenine nucleotides. EXPERIMENTAL PROCEDURES Measurement of Mitochondrial Respiration and Intactness-Intact mitochondria from potato tubers were isolated as described previously (Schwitzguebel and Siegenthaler, 1984). The oxygen consumption of the mitochondria (50 pg -100 pg/ml protein) was measured in a 3-ml stirred cell (modified to achieve more even stirring) containing respiration buffer at 24 "C (0.3 M mannitol, 10 m~ NaH,PO,, 5 mM MgCl,, and 10 mM KC1 (pH 7.2)) by using a Clark oxygen electrode (Yellow Springs Instrument Co.). The mitochondria were pretreated with 0.1 m~ EGTA for 10 min to block NADH dehydrogenases. Five mM succinate was used as the substrate for respiration. ADP was added to trigger the state 111. Four state III-state l V respiration cycles were generated by 4 consecutive ADP (90 p~) additions. NADH (or other dinucleotides) was added between the first and second dose of ADP. The P/O ratio was estimated from the ADP-dependent oxygen consumption. The oxygen concentration in the air-saturated medium was taken as 250 p~.
The intactness of the mitochondrial outer membrane was quantitated by measuring cytochrome c-dependent oxygen consumption (Douce et al., 1987). Exogenously added reduced cytochrome c must pass through outer membrane to be oxidized by the cytochrome c oxidase on 30974 the outer surface of the inner membrane. The percent intactness of the mitochondrial outer membrane for each preparation was taken as 100 times 1 minus the ratio of KCN-sensitive cytochrome c-dependent oxygen consumption of untreated and osmotically shocked mitochondria (30 pl of mitochondrial suspension to 1.5 ml of water for 3 min followed by addition of 1.5 ml of 2 x respiration buffer). For these experiments, we used a mitochondrial protein concentration of 25 -50 pg/ml. The Assay of Protein Content and NADH Oxidation-Mitochondrial protein was measured using the BCAmethod (Pierce) following addition of Triton X-100 to the mitochondrial suspension (1% w/v final). Bovine serum albumin was the standard. The oxidation of exogenous NADH was measured by following the decrease in absorbance at 340 nm at room temperature (Arron and Edwards, 1980).
Calculation of the Free ATP Concentration-Several components of the experimental solution can complex Mg2'. Known values (Critical Stability Constants, Smith and Martell (1976)) of the association constants were used in a spreadsheet to calculate the concentration of all the complexes (MgEGTA'-, MgHEGTA", MgHPO,, MgEDTA2-, MgHEDTA", MgADP", MgHADP, M g m P ' , MgATP'-, MgHATP", MgfiTP) as a function of the I M P ] from 1 to 5 mM. The other constituents were held constant: 16.67 p total EDTA, 100 p total EGTA, 10 mM Na', 10 mM total phosphate, 90 PM total ADP, and either 90 or 270 p total ATP. The two total ATP concentrations bracketed the conditions present in all experiments. We report the percentage of free ATP concentration as a function of the total magnesium concentration present in solution (all forms containing magnesium).
Data Collection-In a typical mitochondrial-respiration experiment, four sequential ADP additions were performed resulting in the recording of four state 111-state IV segments. The first of these was never used and considered to be a treatment that would prime the mitochondria for further testing. Each of the subsequent three state 111-state IV recordings following the addition of an aliquot of ADP were digitized using Ungraph The CN--sensitive cytochrome c oxidase activity of potato mitochondria was assayed in the presence of added ascorbic acid and cytochrome c for freshly-isolated (control) and hypotonically shocked mitochondria. A comparison of the two measurements yields the degree of outer membrane intactness. To mitochondria in 3 ml of respiration buffer (see "Experimental Procedures") were added 50 pl of 1.8 mg/ml cytochrome c (cyt c) and 50 p1 of 0.48 M sodium ascorbate (uit c). The final KCN concentration was 0.2 mM. The numbers on the records are the oxygen consumption rates (nanomoles/min).
where P is the permeability of the outer membrane to ADP; A is the total area of the outer membrane of all the mitochondria present; C, and C, are the ADP concentrations' in the intermembrane space and the medium, respectively; and V, , and K, are enzyme kinetics parameters of the adenine nucleotide translocator (K, was taken to be 4.3 PM, see "Results"). Solving for C, in terms of C,, yields: during the state IV phase was subtracted from the previous state I11 recording in order to obtain just ADP-dependent respiration. The oxygen concentration scale was converted to an ADP concentration scale by knowing the amount of ADP added to the chamber and assuming that, during state IV the [ADP] = 0. Such isomorphism between the two scales is valid only if the P/O ratio of the respiring mitochondria remains unchanged.
Theoretical Model Development and Calculation ofpermeability-We considered that ADP consumption involves: 1) translocation across the outer membrane through W A C (non-saturable); 2) translocation across the inner membrane through the adenine nucleotide translocator; 3) phosphorylation by the mitochondrial ATP synthetase. We assumed that step 3 is fast and thus not rate-limiting. At high ADP concentrations (soon after ADP addition), step 1 is also fast and step 2 is saturated and operating at maximal velocity. This results in a constant rate of oxygen consumption. However, as the ADP in the medium is depleted, the flux through the outer membrane becomes limiting and the flux through the inner membrane is reduced as a consequence of the diminished [ADPI concentrations in the intermembrane space. At steady state: flux of ADP across outer membrane = flux of ADP across inner membrane,' Therefore: ' This ignores the activity of adenylate kinase in the intermembrane space. The maximal velocity of potato adenylate kinase (75 nmoY (min*mg mitochondrial protein), Day et al. (1979)) is one-ninth of the mitochondrial ADP consumption rate we measured (675 nmol/min*mg mitochondrial protein). Similar results were reported for brain (BeltrandelRio and Wilson, 1991). In addition, the K, of adenylate kinase was reported to be 270 (yeast, Su and Russel (1967)) and 180 (liver, Sapico et al. (1972)) p~. Furthermore, sequential additions of ADP to mitochondria that would raise the ATP concentration and affect the consumption or production ofADP by the kinase, had no measurable effect on the respiration curves. Thus, in our conditions, this activity is negligible.
The theoretical curve of the decline of the medium [ADP] as a function of time was generated by calculating the [ADP] at time (t) at intervals (dt) corresponding to the collected data points (1.2 s/point and a total of 340 points). These were calculated as follows: V, , and P*A were allowed to vary so as to achieve the best fit (minimized the sum of the square of the difference between the calculated points and the data points).

RESULTS
The Intactness of Mitochondria-The isolation of t r u l y intact mitochondria is essential for measuring the permeability of their outer membranes. The degree of intactness was iissayed by measuring the KCN-dependent cytochrome c oxidase activity (see "Experimental Procedures" for details). Typically, freshly-isolated mitochondria showed only KCN-insensitive respiration ( Fig. 1) indicating that it was not the result of cytochrome oxidase activity. After hypotonic shock, the cytochrome e-dependent respiration was increased 4-fold. The increase was totally KCN-inhibitable leaving the same baseline level of KCN-insensitive respiration. Mitochondrial intactness in all preparations ranged from 98 to 100%.
NADH Inhibits ADP-dependent Respiration-NADH increases the voltage dependence of VDAC channels reconstituted into planar phospholipid membranes (Zizi et al., 1994). Consequently, NADH may change the permeability of outer membrane by regulating the gating of W A C channels. When 2 min the mitochondria were treated with NADH, a pronounced curvature was observed (Fig. 2) between state I11 and IV respiration. The NADH effect was clearly visible when the concentration ofADP was very low. Other regions of the respiration trace were virtually superimposable. Thus the initial stage of state I11 respiration were not affected by NADH treatment. The P/O ratio was also unaffected. NADH may decrease ADP flux through the outer membrane and this would have its greatest impact at low external ADP concentration.
NADH Oxidation Is Inhibited by EGTA-The possibility that NADH oxidation was responsible for the observation was largely eliminated by inhibiting NADH oxidation. At the 0.02 mM NADH level, NADH would be completely oxidized in 20 min by mitochondria under control conditions (Fig. 3). However, in the presence of 0.1 mM EGTA, only 17% was oxidized in 30 min. The further addition of succinate and ATP reduced the amount oxidized to only 4%. At higher initial NADH concentrations, the percentage of oxidation was less but again a combination of the three reagents essentially eliminated oxidation of NADH. Thus, a very strong inhibition of NADH oxidation could be obtained when mitochondria were treated with EGTA and incubated with succinate and ATP.
Estimation of Permeability of the Outer Membrane-From the respiration measurements one can obtain the rate of decline of the ADP concentration with time. This was fit to Equation l to obtain values for P, V,, and K,,, (see "Experimental Procedures" for details). An example of the fit is shown in Fig. 4.
In our analysis, the value of the K , is assumed to be constant. By definition, the K,,, value should be a constant (at constant temperature and pressure) for any particular enzyme, but the apparent K,,, of the adenine nucleotide translocator is known to change in response to the free ATP concentration outside the mitochondrial matrix. By keeping the free ATP concentration outside the mitochondrion low, this effect can essentially be eliminated (Kramer and Klingenberg, 1982). Therefore we added MgC1, to convert most of the ATP to MgATP2-. Data calculated from published binding constants (Smith and Martell, 1976) are shown in Fig. 5. In the presence of 5 mM MgCI,, the free ATP concentration is less than 2% for a total of either 0.09 or 0.3 mM ATP (these span the concentration range present in our experiments). Changing the ADP concen-tration did not significantly change the free ATP concentration (data not shown). Broken mitochondria (intactness 60%) were used to estimate this K,. We set P to a large value (e.g. 1000) and obtained a fitted value for the K, of 4.3 p~.~ This K, value was then used to analyze the data obtained with intact mitochondria. NADH Decreases the Permeability of Outer Membrane-That NADH decreased the permeability of outer membrane is shown in Fig. 6. Although we can estimate the surface area of mitochondria from published data, we prefer t o combine permeability (P) with the total area in the preparation (A) to yield the total permeability. The figure shows the results of four experiments. The total permeability (i.e. permeability*area) increased with the amount of mitochondrial protein present reflecting the increase in total area of outer membrane of all mitochondria. NADH decreased the permeability of the outer membrane by a factor of 6.
Assuming a one to one interaction between W A C and NADH, the following equation can be derived: where P is the permeability, Po is the maximal permeability In order to examine the specificity of the effect of P-NADH, we also treated mitochondria with &-NADH and P-NAD' (Fig.  8). The a-NADH had a small effect on the permeability at low concentrations (around 20 p~) but no further change between 50 p~ and 1 mM. At higher concentrations (2 mM), a-NADH has a similar effect to P-NADH. However, since a-NADH is contaminated with 3% P-NADH, then at least some of the effect at high concentrations can be accounted for by the contaminant. P-NAD' has some effect on the permeability at low concentra-We obtained essentially the same value if we used intact mitochondria and fit all three parameters (&'*A, V,,,, Km), but such fits yielded less precise results because of the added degree of freedom. tions and no further effect at higher concentrations. Since P-NAD' can enter the matrix space, this result may be due to metabolic changes resulting from p-NAD' entry. DISCUSSION ADP-dependent respiration requires the exchange of extramitochondrial ADP for ATP produced in the mitochondrial matrix. These nucleotides must then flow across both the inner and the outer membranes. At steady state, the net flow of ADP across the outer membrane must equal that across the inner membrane. Thus, the steady-state level of ADP in the intermembrane space has to adjust automatically maintaining the same net flow across both membranes. When the ADP level in the intermembrane space is much higher than the K,,, of the adenine nucleotide translocator, a constant state I11 respiration is observed because the flux through the adenine nucleotide translocator is maximal. As the [ADP] in the intermembrane space approaches the K, of the translocator, the ADP flux declines and the trace of oxygen consumption begins to curve. A reduction in the permeability of the outer membrane t o ADP results in an increase in the ADP concentration gradient across the outer membrane. Thus, the same intermembrane space [ADP] is achieved at a higher extramitochondrial [ADP]. Therefore a reduction in the outer membrane permeability results in the curvature in the oxygen consumption trace beginning at a higher extramitochondrial [ADP]. Under these conditions, the curvature is also more pronounced. By analyzing this curvature, we have estimated the permeability of the outer membrane and demonstrated a reduction in this permeability Mitochondrial Zntactness Affects ADP-dependent Respiration-With the basic model of mitochondrial structure in mind, one might expect that damaging the outer membrane would be a good way of confirming that NADH was acting on the outer membrane. However, when we tried to break the mitochondrial outer membrane by hypotonic shock or digitonin, we could only observe reasonable levels of ADP-dependent respiration when the degree of breakage (as measured by KCN-sensitive cytochrome c-stimulated respiration) was 40% or less (data not shown). Thus the mitochondrial outer membrane seems to be important in maintaining ADP-dependent respiration. The reason for this is unclear. The possibility that this was due to the loss of cytochrome c upon breakage of the outer membrane was tested by adding cytochrome c (60 pglml final) to the suspension of broken mitochondria. While this did increase oxygen consumption, it did not restore ADP-dependent respiration. An alternate hypothesis is that hypotonic shock or digitonin treatment changes the mitochondrial ultrastructure and resulting in loss of ADP-dependent respiration. NADH Oxidation Is Inhibited by Chelators-The possibility that the observed effects of NADH are due to its enzymatic oxidation and the delivery of reducing equivalents to the mitochondria was virtually eliminated by taking steps to inhibit NADH oxidation. There are four NADH dehydrogenases in plant mitochondria (M~ller, 1986;Palmer and M~ller, 1982). Two of the NADH dehydrogenases are located on the outer surface of outer and inner membrane. These two dehydrogenases can oxidize externally added NADH (Arron and Edwards, 1980) while the others can only access matrix NADH. NADH dehydrogenases are Ca'+-dependent (Moiler et al., 1981) and therefore should be inhibited by Ca2+ chelators (Arron and Edwards, 1980;M~l l e r and Palmer, 1981). In order to maintain a high Mg2' concentration in the solution, EGTA was chosen. It was also reported that succinate can inhibit the oxidation of externally added NADH by the interaction between NADH dehydrogenase and succinate dehydrogenase, but the presence of external NADH does not appear to alter the rate of succinate oxidation (Cowley and Palmer, 1980). We also observed that the ATP has a small inhibitory effect on the oxidation of added NADH (reason unknown). Therefore, by using a combination of EGTA, succinate, and ATP, we reduced NADH oxidation to less than 4% in 30 min. By inhibiting NADH oxidation we achieved 2 goals: 1) we avoided a variable effect of NADH during the experiment, and 2) we essentially eliminated the possibility that NADH oxidation was somehow responsible for the changes in respiration that we are interpreting as changes in outermembrane permeability. As to the latter, experiments performed without added EGTA, under conditions where the NADH oxidation rate was much higher (data not shown), produced very similar changes in permeability at the elevated NADH concentrations where NADH consumption during the experiment did not change the medium NADH concentrations appreciably. Thus, while the respiration rates were somewhat elevated due to the oxidation of NADH, the shapes of the curves that reflect the permeability of the outer membrane were essentially the same.
Validity of the Estimated K,,, Value for the Adenine Nucleotide Danslocator-The K,,, of the adenine nucleotide translocator for ADP was estimated to be 4.3 p~ (from measurements on broken mit~chondria).~ It is within the range of reported values: i.e. between 1 and 10 p~ (Tyler, 1992). This value depends on the [ M e ] because the adenine nucleotide translocator only translocates free ADP and ATP rather than ADP.Mg" and ATP.Mg" complexes (Kramer, 1980). Pfaff and co-workers (1969) determined a value of 4 p~ in the presence of 4 mM Mg2' (1.3 p~, in the absence). Thus the value of 4 p~ is close to our estimation of 4.3 p~ in the presence of 5 mM M e .
The free ATP concentration was kept low to avoid changes in K,,, during our experimental runs. In experiments with recon-stituted adenine nucleotide translocator, Kramer and Klingenberg (1982) showed that the V,, varied with transmembrane potential while the K, varied with the free ATP concentration. In the presence of 5 mM MgCl,, the free ATP and free ADP concentrations are less than 2 and 17%, respectively. At this low free [ATP], the K, of adenine nucleotide translocator should not change significantly during the experiments.
The Free Cytosolic ADP Concentration Is Very Low-In our experiments, the permeability of the outer membrane only becomes significant at low ADP concentrations (<30 PM). Although total cytosolic [ADP] is between 0.25 and 0.7 mM as measured and calculated (Soboll et al., 1978;Sies, 1982;Geisbuhler et al., 1984), freely moving forms of [ADP] could not be directly detected in vivo by 31P NMR (Brindle et al., 1989;Roth and Weiner, 1991) and thus should have been less than 0.1 m~.~ The amount of protein-bound ADP has been reported to be more than 50% (Morikofer-Zwez and Walter, 1989) or to be more than 90% (Brindle et al., 1989;Brown, 1992). The free cytosolic [ADP] has only been determined indirectly. It is generally assumed that the creatine kinase-catalyzed reaction is near or at equilibrium (Roth and Weiner, 1991 (Brindle et al., 1989;Roth and Weiner, 1991;Wan et al., 1993). A free intracellular [ M e ] of 1 mM has been measured (Vink et al., 1988). According to our calculations, 6% of ATP and 40% ofADP are free in the presence of 1 mM free M e . Thus, if a cell contained 10 PM ADP not bound to protein, then only 4 PM ADP would be able to interact with the adenine nucleotide translocator and 6 PM ADP.Mg-would interact with creatine kinase. In cardiac muscle the free [ADP] has been estimated to be 6 p at rest and 6.7 PM after heavy exercise (Roth and Weiner, 1991). Thus the effects of NADH occur at physiologically relevant cytosolic [ADP].
The Crabtree Effect Might be Induced by Cytosolic NADH-The Crabtree effect is a manifestation of respiratory inhibition after the addition of glucose or another hexose that is capable of being phosphorylated by hexokinase (Crabtree, 1929;Koobs, 1972;Rimmer and Linsenmeier 1993;Guppy et al., 1993). It is generally accepted that the respiratory inhibition is due to two factors: 1) reduction in [ADP] due to ADP phosphorylation by glycolysis; and 2) reduction in [Pi] due to phosphorylation of a n added hexose (Koobs, 1972). The addition of Pi (Brin and Mckee, 1956) or iodoacetate (Coe, 1964) could release the respiratory inhibition presumably by supplying the needed Pi or by blocking ADP consumption by glycolysis, respectively. However, NADH levels may also play a role and the added Pi and iodoacetate may decrease NADH levels. The added Pi would be transported into matrix and thus stimulate the state I11 respiration which decreases cellular NADH fluorescence (Jobsis and Duffield, 1967). Iodoacetate inhibits glyceraldehyde-3-phosphate dehydrogenase, an enzyme responsible for the production of NADH by the glycolytic pathway. Moreover, the addition of glucose and 2-ketoisocaproate to the isolated islet P-cell was found to increase NAD(P)H fluorescence (Pralong et al., 1990). The addition of glucose to tumor cells both inhibits cellular respiration and increases the fluorescence corresponding NAD(P)H (Ibsen and Schiller, 1971). In cells where the addition of glucose causes a stimulation of respiration (rabbit cortical tubules), this addition also causes a decrease in cellular NAD(P)H fluorescence (Gullans et al., 1984). However, 2-deoxyglucose, a non-metabolizable sugar, can also induce the Crabtree effect (Ibsen et al., 1958). Thus the situation is complex. 2-Deoxyglucose may have other effects that simply act as R. S. Balaban personal communication

Membrane
Permeability 30979 a sink for phosphorus. Alternatively, the phenomenon referred to as the Crabtree effect may result from various processes.

Relationship between the Cellular NADH Concentrations and the Estimated K,-It
is difficult to determine whether the estimated KD for NADH of 87 PM is in the physiological range. The estimates of the cytoplasmic [NADH] are indirect and require that assumptions be made. The direct measurement of total cytoplasmic NAD(H) in adult rat heart yields 0.7 mM (Geisbuhler et al., 1984). The estimates of the NADWNAD' ratio in the cytosol vary considerably (Heber and Santarius, 1965;Williamson et al., 1967;Veech et al., 1969;Hampp et al., 1984;Kromer and Heldt, 1991) from 0.001 to 0.24. In rat liver, the extramitochondrial NADH was estimated to be 24 nmoVg wet weight which translates to approximately 50 PM (Sies, 1982). In oat mesophyll protoplasts the average [NADH] in the cytoplasm and vacuole was estimated at 6.6 PM (Hampp et al., 1984). To our knowledge, no estimates of the cytoplasmic [NADH] are available for potato cells.
Another way of assessing the physiological range of cytoplasmic NADH levels is to examine the K, values (as rough estimates of K,) of cytoplasmic enzymes that act on NADH. NADH dehydrogenases on the outer surface of the inner membrane of plant mitochondria have K, values between 20 and 100 PM (Palmer and Moller, 1982). Bovine lactate dehydrogenase has a value of 24 PM. The cytoplasmic form of bovine heart malate dehydrogenase has K, of 38 PM (Barman, 1969). Thus the estimated KD of 87 p~ may be in the physiological range.
The K, of the VDAC.NADH complex was estimated for HV-DACl (one of the forms of WAC in humans), in a reconstituted system, to be 16 p (Zizi et al., 1994). It is unclear at this point why this value is smaller than that estimated for potato mitochondria. Besides the obvious differences: different species, different techniques, different forms of expressed VDAC, there is also the effect ofthe electrical potential across the mitochondrial outer membrane. Zizi et al. (1994) reported that NADH increased the voltage dependence of WAC. It is possible that the effectiveness of NADH depends on the membrane potential. In the isolated intact mitochondria, the intermembrane space proteins should contribute to a Donnan potential across the outer membrane (value unknown). Thus, this potential (different from that in vivo due to the presence of cytoplasmic proteins and different ionic strength) may have altered the KD for NADH.

The Permeability Changes Are
Specific to the P Form of NADH-The physiological form of NADH is P-NADH. However, a-NADH has been shown to be recognized by some enzymes (Douce et al., 1973). In our experiments, P-NADH is more potent than a-NADH but the latter seemed to induce a permeability change at very low concentrations. However, &-NADH had no additional effect in the concentration range of 50 p~ to 1 mM. At least some of the permeability reduction observed in the presence of 2 mM a-NADH must be due to the 3% contamination of P-NADH (contamination in the product purchased from Sigma). P-NAD', the oxidized form of P-NADH, has an effect at low concentrations, but no effect on reconstituted VDAC channels (Zizi et al., 1994). This difference may be due to the fact that the inner membrane is impermeable to NADH, but permeable to NAD+ (Neuburger et al., 1985;Tobin et al., 1980;Neuburger and Douce, 1983). The entry NAD+ may change the metabolism of mitochondria and yield a spurious results. A progressive inhibition of succinate oxidation in state I11 caused by added NAD+ has been observed (Douce et al., 1986). Estimation of the Permeability of the Outer Membrane and Its Physiological Implications-While the theory allows us to extract the total permeability (cm3/s) from the records (i.e. permeability per unit area x the total area of the outer membrane), we can estimate the surface area of the outer membrane and calculate the permeability per unit area (cds). Knowing the amount of mitochondrial protein used in the experiments, taking the mitochondrial water as 1 pVmg protein, and assuming the mitochondria are spheres 1 pm in diameter, we calculate a permeability of the outer membrane for ADP (without NADH added) of 0.056 c d s . This value is half of the value calculated for unrestricted uptake (0.1 c d s ) which was calculated from the following equation (Hille, 1992).
Maximal flux = 4*?r*r*D*C = P*A*C (Eq. 5 ) where r is 0.5 pm (the radius of the ideal spherical mitochondrion), D was taken as 5 x lo4 cm2/s (the diffusion coefficient for ADP), P is the permeability to ADP ( c d s ) , A is 3.1 x cm2 (the outer membrane surface area calculated for a single mitochondrion), and C is the ADP concentration in the medium. Note that the ratio of the permeability calculated from our measurements to the maximal value calculated from Equation 5 is independent of the radius assumed for the sphere representing the mitochondrion. Thus mitochondria must consume ADP at a rate comparable to the rate of diffusion for the outer membrane to become limiting. This occurs a t low, physiologically relevant, ADP concentrations.
The conditions within the cell differ markedly from those in the isolated organelle. The Donnan potential across the outer membrane must differ because the cytoplasmic macromolecules are absent in the isolated system. In addition, crowding phenomena in the cytoplasm should substantially reduce the effective value of the KD for NADH (Berg, 1990;Zimmerman and Trach, 1991). It must also be pointed out that the colloidal osmotic pressure, shown to favor VDAC closure (Zimmerberg and Parsegian, 19861, has also been shown to reduce the permeability of the outer membrane presumably by closing VDAC channels (Gellerich et al., 1993). Thus, while it is difficult to extrapolate to the in vivo situation, all factors favor the conclusion that the outer membrane is a limiting barrier. Cytoplasmic NADH thus regulates the permeability of an important barrier to the flow of metabolites.
In conclusion, we present evidence that the flux of ADP from the cytoplasm to the mitochondrial matrix is a two-step process: flow through VDAC channels in the outer membrane (step 1) and flow through the adenine nucleotide translocator (step 2). At physiological [ADP] in the cytoplasm, both processes influence the rate of ADP uptake. This rate is close to being diffusion limited. Cytoplasmic NADH can reduce the permeability of the outer membrane t o ADP by a factor of 6 under our experimental conditions, but this may be quite different in vivo.
These results indicate a new pathway for cross-talk between mitochondrial and cytoplasmic energy production and may be responsible to an aspect of the Crabtree effect.