Partial Resolution of the Enzymes Catalyzing Oxidative Phosphorylation

1. The ability to phosphorylate ADP during oxidation of NADH by ubiquinone-1 was restored to the NADH-ubiquinone reductase complex by combining the latter with phospholipids and a hydrophobic protein fraction derived from bovine heart mitochondria. 2. Phosphorylation was abolished by rotenone, uncoupling agents, or rutamycin. The efficiency of ATP formation was as high as 0.5 mole per mole of NADH oxidized under optimal conditions. 3. Reconstitution of phosphorylation had an absolute requirement for phosphatidylethanolamine and a partial requirement for phosphatidylcholine, a molar ratio of approximately 4: 1 being optimal. A much more marked requirement for phosphatidylcholine was observed in the presence of low concentrations of cardiolipin (0.05 to 1.5% of the total phospholipid). In the presence of cardiolipin, an equal molar ratio of phosphatidylethanolamine to phosphatidylcholine gave the highest phosphorylation efficiency. 4. The NADH-ubiquinone reductase complex is oriented in the reconstituted vesicles such that approximately 50% of the molecules can react with added NADH. Reaction of all the molecules with NADH occurs in the presence of 0.5% deoxycholate. 5. Phosphorylation efficiency can be significantly improved by purification of the vesicles on sucrose density gradients.


SUMMARY
1. The ability to phosphorylate ADP during oxidation of NADH by ubiquinone-1 was restored to the NADH-ubiquinone reductase complex by combining the latter with phospholipids and a hydrophobic protein fraction derived from bovine heart mitochondria.
The efficiency of ATP formation was as high as 0.5 mole per mole of NADH oxidized under optimal conditions.
3. Reconstitution of phosphorylation had an absolute requirement for phosphatidylethanolamine and a partial requirement for phosphatidylcholine, a molar ratio of approximately 4: 1 being optimal.
A much more marked requirement for phosphatidylcholine was observed in the presence of low concentrations of cardiolipin (0.05 to 1.5% of the total phospholipid).
In the presence of cardiolipin, an equal molar ratio of phosphatidylethanolamine to phosphatidylcholine gave the highest phosphorylation efficiency. 4. The NADH-ubiquinone reductase complex is oriented in the reconstituted vesicles such that approximately 50% of the molecules can react with added NADH.
Reaction of all the molecules with NADH occurs in the presence of 0.5% deoxycholate. 5. Phosphorylation efficiency can be significantly improved by purification of the vesicles on sucrose density gradients.
Hydrophobic protein, a heterogeneous fraction solubilized from bovine heart submitochondrial particles with cholate and ammonium sulfate, has been reconstituted with phospholipids and coupling factors to form vesicular structures which catalyzed rutamycirl-and uncoupler-sensitive "Pi-ATP exchange (1). Incorporation of cytochrome c and cytochrome oxidase into these vesicles resulted in a preparation which coupled ATP * This work was supported by United States Public Health Service Grant CA-08964. C. I. R. was in receipt of a Sir Henry Welcome Travelling Fellowship from the United Kingdom Medical Itesearch Council.
The support from the Albert Einstein Award to one of (1s (11:. I<.) is gratefully acknowledged.
formation from ADP and Pi to the osidation of ascorbate in the presence of N-methylphenazinium methyl sulfate at the third site of energy conservation (2). In this paper, we report the reconstitution of the first site of energy conservation by combining NL4DH-ubiquinone reductase (3) with hydrophobic proteins and phospholipids.
EXPERIMENTaL PROCEDURE lMateri&--Cholic and deoxycholic acids were obtained from Matheson Colemen and Bell, East Rutherford, N. J., and purified by treatment with charcoal and recrystallization from 70% ethanol (4). Asolectin was obtained from Associated Concentrates, Woodside, N. Y., and cardiolipin was from General Biochemicals.
Preparation of Soybean Phospholipids-Phosphatidylethanolamine was purified as described by Kagawa el al. (9). Further purification was achieved by chromat'ography on a second column of silicic acid under identical conditions. Phosphatidylethanolamine from the second column contained only two minor contaminants in thin layer chromatography, one of which gave a purple color with ninhydrin and had an Rp value expected for lysophosphatidylethanolamine. No phosphatidylcholine could be detected.
Phosphatidylcholine was prepared from crude soybean phospholipids enriched in phosphatidylcholine by ethanol fractionation (10). This fraction (18 mmoles of phospholipid phosphorus) was dissolved in 100 ml of chloroform and applied to a column (5 X 50 cm) of Bio-Rad silicic acid HA minus 325 mesh, previously washed with Fischer pesticide grade "hexanes" and equilibrated with chloroform. -Colored material was eluted first with chloroform and second with chloroform-methanol (4 : 1, v/v). Phosphatidylethanolamine was then eluted with chloroform-methanol (3:2, v/v) and, when no more phospholipid phosphorus could be detected, elution was started with a linear gradient of methanol from 40 y0 by volume in chloroform to 100 '% in a t'otal volume of 3 liters. Following this, elution was continued with methanol.
Phosphatidylcholine was eluted either during the gradient or with methanol, and the peak was located by spot-testing with Dragendorf reagent (11) on filter paper. The phosphatidylcholine preparation gave one major spot in thin layer chromatography.
Phospholipids eluted from the column were evaporated to dryness, dissolved in chloroformmethanol (4:1, v/v), and stored at -20" under nitrogen.  (14). Rotenone-insensitive NADH oxidation was determined under the same conditions but in the presence of 1 pg of rotenonc. A sample lacking the vesicles was used to correct for nonextractable radioactivity in the "Pi.

RESULTS
dnalytical Melhods-Protein was estimated by the method of Lowry et al. (12), and total phosphorus was determined as described by Ames and Dubin (13). NADH-ferricyanide reductase was measured at 420 nm in a final volume of 1.0 ml containing 40 pmoles of Tris-HCI (pH 7.8), 1.5 pmoles of K3Fe(CN)6, and 0.15 pmole of NADH.
Oxidative phosphorylation at Site I was measured as follows. Aliquots (0.25 ml) of the reconstituted vesicles (containing 0.4 to 0.8 mg of protein) were incubated at 23" for 10 to 15 min with 25 kmoles of KPi (pH SO), 3 mg of bovine serum albumin, 2 pmoles of MgSO+ 2 pmoles of ATP, 160 pg of F1, and 120 pg of OSCP in a final volume of 0.45 ml. Of this mixture, 0.2 ml was then incubated at 30" in a cuvette with a l-cm light path, in a final volume of 1.13 ml containing 85 units of hexokinase, 32 pmoles of glucose, 0.2 pmole of EDTA, 3 pmoles of ATP, 4 pmoles of MgS04, 2 mg of bovine serum albumin, 5 pmoles of Tris-SO1 (pH 7.3), 0.125 pmole of NADH, 1.6 pmoles of KCN (added as a freshly prepared 40 mu solution in 0.1 M Tris-SO1 (pH 7.4)), 1.5 pg of antimycin, and "Pi (1 X lo6 cpm).
After 30 set at 30", the reaction was started by the addition of 10 ~1 of a 15 mM solution of ubiquinone-1 in 95% ethanol, and the reaction was followed by measuring the decrease in extinction of NADH at 340 nm in a Gilford multiple sample spectrophotometer or, for highly turbid samples, the decrease at 340 nm minus 430 nm in an Aminco DW2 spectrophotometer.
An extinction coefficient of 6.81 m&I-i cm-i (14) was used at 340 nm, and 6.41 rnM-i cm-' was used at 340 nm minus 430 nm. When the reaction had neared completion (2 to 6 min), the cuvette contents were mixed with 0.1 ml of 50% trichloroacetic acid and cooled at 4". After removal of denatured protein by centrifugation, 0.5-ml aliquots of the supernatant were assayed for nonextractable 32P as described by Schatz and Racker Calculation of P/ge Ratios-ATP formation associated with the oxidation of NADH by ubiquinone-1 was completely inhibited by rotenone; NADH oxidation was only partly inhibited ( Table I). The rotenone-insensitive extinction decrease at 340 nm is caused by two factors: (a) the nonphosphorylating pathway of ubiquinone-1 reduction described in detail by Schatz and Racker (14), and (b) nonenzymic reduction of ubiquinone-1 by residual dithiothreitol, which is also not associated with ATP formation.
Oxidized ubiquinone absorbs at 340 nm (E,llM = 0.59 111~~~ cm-l (14) compared with 6.22 rnM+ cm-' for NADH), and its reduction by dithiothreitol was seen as an initial rapid decrease in extinction which decayed to a steady rate (due to rotenone-insensitive NADH-ubiquinone-1 reductase) after about 1 min. For simplicity, the total rotenone-insensitive extinction change has been given as micromoles of NADII (using an E,,,, of 6.81 rnM+ cm-l) but, since the nonenzymic reduction of ubiquinone-1 by dithiothreitol made a relatively larger contribution to the rotenone insensitive extinction change than to the extinction change in the absence of rotenone, this underestimated somewhat the true inhibition by rotenone, which was generally 70 to 80%.
A further effect of the nonenzymic reduction of ubiquinone-1 was to decrease the total amount of NADH that was oxidized.
Thus, following addition of 150 PM ubiquinone-1, the reaction terminated after about 70 PM NADH had been oxidized.
In calculating P/2e ratios, the rotenone-insensitive extinction decrease was subtracted from the total.
Table I also shows that NADH oxidation and ATP formation occurred to some extent in the absence of ubiquinone-1 and were completely inhibited by rotenone.
The acceptor for this NADH oxidation is not known, but it may be a slowly autoxidizable component of the electron transport chain. Since the reaction occurred in the presence of both antimycin and cyanide, it is presumably a component on the substrate side of cytochrome b. with an efficiency similar to that obtained with added Complex I. Phosphorylation was again completely inhibited by rotenone, but it was only partially dependent on added ubiquinone-1 and therefore accounted for a considerable part of the ATP formation that took place in the absence of ubiquinone-1 and in the presence of Complex I (Table I). As shown in Table II, lowering the concentration of the hydrophobic proteins had little effect on the overall phosphorylation efficiency in the presence of Complex I, but it decreased the phosphorylation efficiency obtained in the absence of Complex I. Thus, at a concentration of 2 mg of hydrophobic protein per ml, the contribution to the total phosphorylation by the hydrophobic protein was only 9%. No phosphorylation was obtained in the absence of hydrophobic protein.
Dependence on Phospholipids-32Pi-ATP eschange and the third site of oxidative phosphorylation were successfully reconstitutcd (1, 2) with soybean phospholipids obtained from crude asolectin by removing neutral lipids with acetone and then initially extracting the residue with ether and finally extracting the ether-soluble material with ethanol (10). This ethanolsoluble fraction was found to be unsuitable for the reconstitution of the first site, and better results were obtained by using mixtures of purified soybean phospholipids. compared with a mixture of purified phosphatidylethanolamine and phosphatidylcholine, the ethanol-soluble lipids yielded a preparation with a somewhat lower rotenone sensitivity of ubiquinone-1 reduction and low P/2e ratio.
Either a constituent of the ethanol-soluble lipids interfered with the reconstitution of the Complex I or the ratio of the phospholipid components was not optimal.
The effect on reconstitution of varying the ratio of phosphatidylethanolamine to phosphatidylchloline is shown in Fig. 1. In this experiment, the dependence on phosphatidylethanolamine was absolute, but only a partial dependence on phosphatidylcholine was observed. The phosphorylation efficiency of vesicles reconstituted with phosphatidylethanolamine alone was rather variable, and occasionally no phosphatidylcholine requirement could be demonstrated. Since purification of phosphatidylethanolamine on a second silicic acid column, which removed all contaminating phosphatidylcholine, did not cause a consistent decrease in the phosphorylation efficiency obtained with phosphatidylethanolamine alone, the variability remains unexplained.
The optimal phosghatidylcholine requirement for reconstitution varied therefore between 0 and 20% of the total phospholipid. Cardiolipin had a very marked effect on the dependency on phosphatidylcholine as shown in Fig. 2. In the presence of cardiolipin, the P/2e ratio was lower at high phosphatidylethanolamine concentrations, the effect being to move the optimal molar ratio of phosphatidylethanolamine to phosphatidylcholine to 1: 1. Thus a clear stimulation by both phosphatidylcholine and cardiolipin was observed under these conditions. Cardiolipin also stimulated the rate of reduction of ubiquinone-1 by NADH by as much as 100% at a molar ratio of phosphatidylethanolamine to phosphatidylcholine of 1: I, but it gave only a 20 to 30 y0 stimulation with phosphatidylethanolamine alone. Very low concentrations of cardiolipin u-ere required for these effects (Fig. 3). The P/2e ratio was stimulated 70% by 0.1 to 0.3 pmole of cardiolipin per ml in the presence of phosphatidylethanolamine and phosphatidylcholine, both at 10 pmoles per ml. The stimulation of NADH-ubiquinone-1 reductase did not follow the same pattern, the greatest increase in rate being at the highest concentration used, namely, 2 pmoles per ml. The effect of the total  I  I  I  I  I  I  I  I  18  16  14  12  IO  8  concentration of phospholipid on reconstitution is shown in Fig. 4. With a molar ratio of phosphatidylethanolamine to phosphatidylcholine of 4: 1 or 3 :2, the optimal total concentration was 20 pmoles of phospholipid phosphorus per ml in either case. The small decline in P/2e ratio at 25 pmoles per ml was quite reproducible and often more marked than in this experiment. This is of particular interest since it was shown in Table I that, at a fixed phospholipid concentration of 25 pmoles per ml, decreasing the concentration of hydrophobic protein caused a slight increase in P/2e ratio.
Thus, the absolute concentration of phospholipid rather than ratio of phospholipid to protein seems to be important in reconstitution of phosphorylation at Site 1.
Effect of Inhibitors of Oxidative Phosphorylation- PC, phosphatidylcholine. There was no effect on the rate of NAliH oxidation by either of these reagents, i.e. no respiratory control could be demonstrated.
Dependency on Coupling Factors-Hydrophobic proteins prepared by extracting submitochondrial particles with cholate and 100 mM ammonium sulfate contain coupling factors (I), and the dependency of oxidative phosphorylation on added F1 and OSCP was not very marked (Table V). Extraction with 20 nl~ ammonium sulfate gave a preparation of hydrophobic proteins which showed a marked dependency on added coupling factors.
Omission of either Fi or OSCP caused a significant decrease in the P/2e ratio. Omission of both factors led to negligible rates of ATP formation.
The experiments of Table V were not performed in parallel, and the higher P/2e ratios observed with low salt hydrophobic proteins do not imply that, in general, low salt hydrophobic proteins give superior phosphorylation efficiency. Dependence on Cholate-Reconstitution was optimal at an initial cholate concentration of about 10 mg per ml, and higher concentrations were inhibitory (Fig. 5). This concentration of cholate is lower than that found to be optimal for the reconstitution of 32Pi-ATP exchange (9).
Dependence on Complex I-Although ATI formation was virtually completely dependent on added Complex I, phoxphorylation efficiency was not (Tables I and II)  tivity of the lightest fraction (Fraction I) reached a maximum at the lowest concentration of Complex I used and did not change in activity as ihe Complex I concentration was raised, indicating that this fraction had become saturated at low levels of Complex I. Fraction II increased in activity with increasing Complex I concentration but showed some tendency to reach a maximum. Fraction III (the pellet) increased in activity linearly with added Complex I. Phosphorylation efficiency of the fractions was measured in a separate experiment at tFo different concentrations of Complex I (Table VI).
Before sucrose density gradient fractionation, the P/2e ratio at the higher Complex I concentration was lower, as expected.
After centrifugation, the P/2e ratios of Fractions I and II were very high, and those of Fraction TIT WPTC? negligible!.
The P/~A ra,tios xwF: largely independent of the initial Complex I concentration.
Thus it appeared that only a limited concentration of Complex I could be incorporated into phosphorylating vesicles; therefore the progressive "uncoupling" with increasing concentrations of Complex I can be explained by the increasing proportion of nonphosphorylating NADH oxidation. Orientation of Complex I in the Membrane-NADH dehydrogenase is located on the matrix side of the inner membrane of intact mitochondria and on the outer surface of submitochondrial particles obtained by sonication (15). For evaluation of the orientation of this enzyme in the reconstituted vesicles, a number of the properties of Complex I were examined before reconstitution, after reconstitution, and after dissolution of the reconstituted vesicles by deoxycholate. NADH-ubiquinone reductase and NADH-K3Fe(CN)G reductase activities were found to be unsuitable for evaluation, as the rates of both activities in the mixture of Complex I, hydrophobic proteins, and phospholipids were stimulated by deoxycholate even before dialysis by 100 and 5Oa/ respectively.
On the other hand, the bleaching by NADH of the flavin and non-heme iron of Com- plex I measured at 460 nm (3) was not affect,ed by deoxycholate before reconstitution (Table VII). After reconstitution, the bleaching was incomplete and could be restored to the original level with 0.5% deoxycholate.
In this experiment, the proportion of the enzyme inaccessible to NADH in the intact vesicles was 317& of the total.
In other experiments, values of 24 and 29% were obtained.
Although these exT,eriments gave values for the proportion of the NADH site of Complex I located on the inner surface of the vesicles, they did not distinguish between Complex I accessible at the outer side of the vesicles and unincorporated Complex I. Fig. 8a shows that with increasing Complex I concentrations, the amount of Complex I inaccessible to NADH in the intact vesicles remained approximately constant.
Sucrose density gradient fractionation of the vesicles showed that only the two lightest fractions (Fig. 8, b and c),