Preparation and characterization of cytochrome c oxidase vesicles with high respiratory control.

1. Isolation of cytochrome c oxidase which yields high respiratory control ratios in reconstituted vesicles is reported. Fusion of reconstituted vesicles with low and high respiratory control ratio suggests that lack of respiratory control is caused by faulty incorporation of protein rather than by the presence of a proton leak. 2. Digestion of cytochrome c oxidase with chymotrypsin removed some contaminating polypeptides without damaging reconstitution of cytochrome c oxidase vesicles. Ureadodecyl sulfate gel electrophoresis revealed the presence of six subunits in the digested protein. 3. A method is described by which the orientation of cytochrome c oxidase in the membrane was evaluated. With cytochrome c oxidase as the only protein, the enzyme was assembled in the mitochondrial orientation (right-side-out). In the presence of cytochrome c there was considerable scrambling with about 50% of the enzyme in the right-sideout orientation. 4. Phospholipid requirements for the preparation of vesicles with high respiratory control differed considerably with the method of recoatitution. Marked variation in the sensitivity to detergents was observed dependent on the reconstitution procedure. Oxidation of reduced cytochrome c by vesicles reconstituted by the incorporation procedure was markedly stimulated by about 0.03% Tween 80. This respiration was blocked by 0.2 mM dicyclohexylcarbodiimide but not by rutamycin and was completely restored by uncouplers of oxidative phosphorylation. This phenomenon was not seen with vesicles reconstituted by cholate dialysis. 5. Fractionation of cytochrome c oxidase vesicles on Ficoll gradients revealed that cytochrome c oxidase.reconstituted by incorporation was associated with only a small fraction of the phospholipid population. The isolated protein-free liposomes were incapable of incorporating cytochrome c oxidase when exposed to a second incubation with the enzyme.


Isolation
of cytochrome c oxidase which yields high respiratory control ratios in reconstituted vesicles is reported. Fusion of reconstituted vesicles with low and high respiratory control ratio suggests that lack of respiratory control is caused by faulty incorporation of protein rather than by the presence of a proton leak.
2. Digestion of cytochrome c oxidase with chymotrypsin removed some contaminating polypeptides without damaging reconstitution of cytochrome c oxidase vesicles. Ureadodecyl sulfate gel electrophoresis revealed the presence of six subunits in the digested protein.
3. A method is described by which the orientation of cytochrome c oxidase in the membrane was evaluated. With cytochrome c oxidase as the only protein, the enzyme was assembled in the mitochondrial orientation (right-side-out). In the presence of cytochrome c there was considerable scrambling with about 50% of the enzyme in the right-sideout orientation.
4. Phospholipid requirements for the preparation of vesicles with high respiratory control differed considerably with the method of recoatitution.
Marked variation in the sensitivity to detergents was observed dependent on the reconstitution procedure. Oxidation of reduced cytochrome c by vesicles reconstituted by the incorporation procedure was markedly stimulated by about 0.03% Tween 80. This respiration was blocked by 0.2 mM dicyclohexylcarbodiimide but not by rutamycin and was completely restored by uncouplers of oxidative phosphorylation.
This phenomenon was not seen with vesicles reconstituted by cholate dialysis. 5. Fractionation of cytochrome c oxidase vesicles on Ficoll gradients revealed that cytochrome c oxidase.reconstituted by incorporation was associated with only a small fraction of the phospholipid population.
The isolated protein-free liposomes were incapable of incorporating cytochrome c oxidase when exposed to a second incubation with the enzyme. (1) as required by the chemiosmotic hypothesis (2). The enzyme was reconstituted into liposomes (3)(4)(5)  Preparations Cytochrome c oxidase from bovine heart mitochondria was isolated by several procedures (5,(11)(12)(13)(14)(15)(16)(17). Three of these preparations (11,14,15)  The reconstitutive properties of several enzyme preparations are summarized in Table I. Enzymes exposed to deoxycholate or to 3% cholate at 30" (13) gave low respiratory control ratios (1 to 2) upon reconstitution into liposomes. Enzymes prepared with cholate gave ratios of 3 to 6. This includes preparations delipidated by relatively mild procedures (5,16). All preparations showed similar major subunit patterns on dodecyl sulfate-polyacrylamide gels (23) with varying amounts of high molecular weight contaminants ( Fig. 1). Refractionation (see "Experimental Procedures") of the enzyme yielded vesicles with high control ratios (6 to 12) as shown in Table II. Even enzymes that had been exposed to deoxycholate (11) gave vesicles with control ratios of 4 to 6 after refractionation.
To determine whether the high control ratios are a result of the removal of material interfering either with the incorporation of the enzyme into liposomes or of material giving rise to a proton leak, fusion experiments with liposomes with high and low control ratios were performed.
It was shown previously (21) that a proton leak as measured by an increase in oxidase activity can be induced into cytochrome c oxidase vesicles by fusion with liposomes containing the hydrophobic protein fraction of mitochondria (Table III). When cytochrome c oxidase vesicles with no respiratory control were fused with vesicles with high respiratory control, the rate of oxidation in the absence of uncouplers could be mostly accounted for by the activity of the uncontrolled vesicles, while the net stimulation by uncouplers was the same as with the controlled vesicles alone. This indicates that the observed low control ratio is primarily due to the lack of incorporation of the enzyme rather than to the introduction of a proton leak. A similar conclusion was reached when instead of fusion a mixture of the two types of cytochrome c oxidase were co-reconstituted (data not shown).

Minimum Subunit Composition
-In the course of attempts to define the minimum subunit composition required for the reconstitution of cytochrome c oxidase, we attempted to repeat preparations (12,(25)(26)(27) reported to be lacking some of the major subunits. In our hands these procedures either yielded enzyme preparations with an unaltered subunit composition or with such low enzyme activity as to preclude significant assays. However, exposure of the enzyme to proteolytic enzymes (17, 26) gave rise to much cleaner preparations ( Fig.  2B) devoid of some of the polypeptide bands that appear on urea sodium dodecyl sulfate gels (24,28,29). It can be seen that Band IIb was proportionately reduced by >50% as determined by relative staining intensity and Band Va was completely eliminated. Time courses of chymotrypsin digestion of the enzyme revealed (data not shown) that Band Va was rapidly digested with greater than 90% reduction in staining intensity in about 30 min. Initially Band IIb was also rapidly digested with reduction in staining intensity leveling off after 15 min. Since the specific activity and heme a content increased proportionately with digestion and since the digested and reconstituted enzyme showed no loss of respiratory control, we conclude that Band Va is not an essential subunit of the enzyme. Since Band IIb appears to be more than 50% digested it seems also unlikely that this polypeptide is a subunit. However, we cannot exclude the possibility that Band IIb or the other minor components remaining are either poorly staining subunits or function at substoichiometric levels. Moreover, it is conceivable that other methods of resolution might reveal addition components.
For the time being we conclude that a protein with six subunits is reconstitutively active.
Phospholipid Requirements for Reconstitution-Five methods of reconstitution of membrane proteins into liposomes have been developed in this laboratory. All are effective for cytochrome c oxidase and yield respiratory control ratios greater than 4. The method of sonication (20) was not used frequently since losses of cytochrome c oxidase activity up to 50% were observed. Table IV gives a summary of the observed phospholipid requirements for the four methods of reconstitution.
Reconstitution of cytochrome c oxidase with respiratory control was obtained with either phosphatidylcholine or phosphatidylethanolamine in combination with an acidic phospholipid or diacetylphosphate.
The greatest dependency on acidic phospholipids was observed with the incorporation procedure (5) or when phosphatidylcholine was used alone by any method of reconstitution.
Exceptions to dependency on acidic phospholipids are liposomes with phosphatidylethanolamine:phosphatidylcholine (4:l) which gave high control ratios by either cholate dialysis or dilution.
Pure phosphatidylethanolamine also appears to yield vesicles with high control ratios when prepared by cholate dilution. The phosphatidylethanolamine used was better than 98% pure as analyzed by thin layer chromatography (see "Experimental Procedures") in two different solvent systems. That phosphatidylethanolamine:phosphatidylcholine (4:l) mixtures eliminated the need for acidic phospholipid has previously been reported for the reconstitution of site III oxidative phosphorylation (6). There is evidence that phosphatidylethanolamine prepara-  (30). It was postulated that phosphatidylethanolamine was labile giving as hydrolytic products either free fatty acids or phosphatidic acid, but it was not reported that these breakdown components were actually observed. We have, in fact, oRen detected phosphatidic acid in both synthetic and highly puritied natural phosphatidylethanolamine preparations.
It is therefore possible that some cleavage of this phospholipid takes place during sonication and subsequent incubation for reconstitution.
Additional evidence of a requirement for a net negative charge in the reconstitution of cytochrome c oxidase vesicles comes from experiments with stearylamine.
As shown in Fig.  3, at a fixed level of added acidic phospholipid, the presence of stearylamine during reconstitution inhibited respiratory control at low concentrations and oxidase activity at high concen-trations. It can be seen in Fig. 4 that the inhibition by stearylamine of both respiratory control and oxidase activity was reversed by dicetylphosphate.
The loss of oxidase activity with excess stearylamine seems to be a function of reconstitution rather than a direct inhibitory effect since enzyme added to preformed liposomes with the same amounts of stearylamine gave full activity.
Orientation of Cytochrome c Oxidase in Reconstituted Vesicles -In order to determine the orientation of cytochrome c oxidase reconstituted into liposomes, it was necessary to compare the uncoupled rate of respiration driven by external ferrocytochrome c to the activity obtained with ferrocytochrome c after disruption of the liposomes by a detergent which did not inhibit enzyme activity. Of several detergents tested, Tween 80 was the only one that opened the vesicles without inhibiting oxidase activity.  and in the presence of 3% Tween 80. Also shown are the activities obtained with enzyme added to preformed liposomes. Since the observed activities were in good agreement with each other we conclude that reconstitution of the enzyme is unidirectional with full accessibility to external ferrocytochrome c. This result has been obtained with all the methods of cytochrome c oxidase reconstitution.
On the other hand, reconstitution of oxidative phosphorylation (6) suggested that some of the cytochrome c oxidase must have been oriented inside-out.
Tests with the individual components used in these reconstitution experiments revealed that cytochrome c was responsible for the scrambling of oxidase. When 80 PM or more ferricytochrome c was present during reconstitution the rates of oxidation obtained in the presence of 3% Tween 80 were from 30 to 60 ng atoms of oxygenlmin higher than the uncoupled rates (Fig. 5A). There was also a loss of total activity approaching 70% at 140 pM ferricytochrome c and at concentrations above 100 pM of ferricytochrome c uncouplers no longer stimulated. Reconstitution with ferrocytochrome c did not induce random orientation or loss of activity as shown in Fig. 5B. The vesicles that were reconstituted with ferricytochrome c were capable of setting up a membrane potential as measured by anion uptake by the procedure of Jasaitis et al. (    This possibility was also unlikely since a high phospholipid:protein ratio was required with a large variety of phospholipid mixtures (see Table IV).
Other possibilities that were considered are selection of asymmetrically assembled liposomes, differences in membrane curvature, or phospholipid packing density. Reconstitution of Vesicles at Low Phospholipid:Protein Ratios -In the course of experiments exploring these possibilities it was noted that low phospholipid:protein ratios can be used in the cholate dilution procedure provided exposure to 1% cholate was limited to 10 min as shown in Fig. 8. These observations led us to modify the cholate dialysis technique in order to remove cholate as quickly as possible. This was done by working rapidly, dialyzing immediately after the addition of enzyme, and by reducing the sample volume from 0.6 to 0.2 ml, thereby increasing the surface during dialysis. As shown in Fig. 9, good reconstitution efficiency was obtained at phospholipid:protein ratios of 2.5 with both cholate dialysis and cholate dilution procedures. Similar results were obtained with several other phospholipid mixtures. However, the incorporation procedure still required phospholipid:protein ratios of 20 (Fig. 9). We therefore pursued the problem of the properties of vesicles prepared by the incorporation procedure.

Characterization of Subpopulations of Cytochrome c Oxidase Liposomes Prepared by Incorporation
Procedure -Fkconstitution by the incorporation procedure revealed that only a 10% subpopulation of liposomes allowed incorporation of the enzyme. It can be seen from Table VII that fractionation  of  vesicles reconstituted at phospholipid:protein ratios between 40 to 5 yielded liposome bands (5% Ficoll layers) containing 87 to 90% of the phospholipids and less than 12% of the activity. Most of the activity was associated with about 10% of the original phospholipids used for reconstitution.
This was also the case when larger amounts of enzyme were used. Whereas always 100% of the phospholipid was recovered, activity recoveries were only 80% for samples with phospho-1ipid:protein ratios of 15 or 10 and only 35% with a ratio of 5. The reconstituted vesicles at a phospholipid:protein ratio of 5 gave highly aggregated enzyme fractions in the 15 and 30% Ficoll layers. These fractions had low activities but were reactivated by incubation with 4% phospholipid plus 2% cholate for 15 min at 20%. Thus the loss of activity was due to aggregation of enzyme and phospholipid deficiency.
Addition of fresh cytochrome c oxidase to liposomes isolated from the gradient which had not incorporated the enzyme, yielded vesicles that had little or no respiratory control (Table vesicles and that about 90% of the liposomes were not suitable for incorporation. This is not due to the presence or absence of some component in the liposome band since after extraction and resonication these phospholipids were again suitable for incorporation of cytochrome c oxidase. We were therefore left with two alternatives, either a difference in the size of the competent vesicles or in the packing of the phospholipids.
Thus far we have been unable to detect reproducible differences in the properties of the competent and incompetent liposome. The answer to this puzzling question must await the development of physicochemical methods that are as sensitive as cytochrome c oxidase in discerning between the different populations of liposomes.