Cytochrome c Oxidase from Bakers’ Yeast III. PHYSICAL CHARACTERIZATION OF ISOLATED SUBUNITS AND CHEMICAL EVIDENCE FOR TWO DIFFERENT CLASSES OF POLYPEPTIDES*

Earlier studies have shown that cytochrome c oxidase from bakers’ yeast is an oligomeric enzyme which contains three polypeptides (I to III) synthesized on mitochondrial ribosomes and four polypeptides (IV to VII) synthesized on cytoplasmic ribosomes. These polypeptide subunits have now been isolated by a simple protocol which utilizes differences in polypeptide charge, solubility, and size. Their molecular weights determined by polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate, gel filtration in the presence of guanidine hydrochloride, and amino acid analysis were: I, 40,000; II, 33,000; III, 22,000; IV, 14,500; V, 12,700; VI, 12,700; and VII, 4,600. All seven polypeptide subunits exhibited acidic isoelectric points; cytoplasmically made subunits were more acidic than mitochondrially made ones. The amino acid composition of two mitochondrially made subunits and two cytoplasmically made subunits was determined. The two mitochondrial translation products, I and II, contained only 34.7% and 42.1% polar amino acids, respectively, whereas the two cytoplasmic translation products, IV and VI, contained 48.3%

Earlier studies have shown that cytochrome c oxidase from bakers' yeast is an oligomeric enzyme which contains three polypeptides (I to III) synthesized on mitochondrial ribosomes and four polypeptides (IV to VII) synthesized on cytoplasmic ribosomes. These polypeptide subunits have now been isolated by a simple protocol which utilizes differences in polypeptide charge, solubility, and size. Their molecular weights determined by polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate, gel filtration in the presence of guanidine hydrochloride, and amino acid analysis were: I, 40,000;II,33,000;III,22,000;IV,14,500;V,12,700;VI,12,700;and VII,4,600. All seven polypeptide subunits exhibited acidic isoelectric points; cytoplasmically made subunits were more acidic than mitochondrially made ones.
The amino acid composition of two mitochondrially made subunits and two cytoplasmically made subunits was determined.
The two mitochondrial translation products, I and II, contained only 34.7% and 42.1% polar amino acids, respectively, whereas the two cytoplasmic translation products, IV and VI, contained 48.3% and 49.3%, respectively. This agreed with the observation that Subunits I and II are very insoluble, requiring detergents for solubility, whereas Subunits IV and VI are water-soluble in the absence of any added detergent.
These results indicate that the cytochrome c oxidase subunits synthesized on mitochondrial and cytoplasmic ribosomes are fundamentally different in size, isoelectric properties, and hydrophobicity. They also suggest the possibility that at least some of the mitochondrially made subunits are buried in the lipid phase of the mitochondrial inner membrane.
Recent studies have shown that cytochrome c oxidase from bakers' yeast and Neurospora crassa consists of seven polypeptides (2)(3)(4), whose biosynthesis and assembly results from the coordinated functioning of cytoplasmic and mitochondrial protein synthesis (1,4,5). Although a great deal of information has been obtained concerning the sites of synthesis of these polypeptides and the interplay between mitochondrial and cytoplasmic protein synthesis in the-assembly of this enzyme (6-lo), there is still relatively little known about the function and physicochemical properties of these polypeptides and their arrangement within the mitochondrial membrane. In view of the oligomeric nature of cytochrome c oxidase, information concerning the properties and arrangement of polypeptide subunits within the membrane would be particu-* This work was supported by Grant GB-40541X from the National Science Foundation and by Hatch Project Grant NY(C)181406 from the United States Department of Agriculture. The preceding paper in this series is Ref. 1 larly pertinent for understanding the function of each subunit and the assembly of the holoenzyme in. uiuo. In this paper, we describe the isolation and characterization of the seven polypeptide subunits of yeast cytochrome c oxidase. We have found that the polypeptide subunits translated on mitochondrial ribosomes are larger, less acidic, and more hydrophobic than the subunits translated on cytoplasmic ribosomes. Subsequent communications will deal with the use of subunit-specific antisera to study the function of each subunit (11) and the use of chemical probes to study the spatial arrangements of subunits in the solubilized and membrane-bound enzyme (lla).
Inc.. New York, N.Y.) as described previously (2)  and sucrose gradient centrifugation were performed as previ-described by Moss and Rosenblum (12). The column was poured over a ously described (2) to establish the number of polypeptide subunits in yeast cytochrome c oxidase by more rigorous means.
Reliability of Gel Electrophoresis in Sodium Dodecyl Sulfate-The aggregation and incomplete dissociation that has been observed for some proteins in sodium dodecyl sulfate gels can be prevented by heating the sample to 100" for a few minutes in the presence of 1% 2-mercaptoethanol (36). To exclude the reformation of disulfide bonds during electrophoresis, cytochrome c oxidase was first dissociated in 8 M urea and either reduced with dithiothreitol, carboxamidomethylated with iodoacetamide, or oxidized with performic acid before it was denatured in sodium dodecyl sulfate and subjected to electrophoresis. Fig. 1, A to D shows that the gel electrophoretie patterns of untreated enzyme, reduced enzyme, and enzyme reacted with iodoacetamide or performic acid are identical5 This result makes it very unlikely that the polypeptide subunits of this enzyme are linked to each other by disulfide bridges or that the polypeptide bands observed on sodium dodecyl sulfate gels are aggregates or incompletely dissociated polypeptide subunits. Although most proteins migrate according to t,he logarithm of their molecular weight in sodium dodecyl sulfate gels, some proteins with unusual charge (32), conformation (34), unreduced sulfhydryl groups (35), and carbohydrate moieties (33) do not. As a prelude to an accurate estimate of molecular weight by sodium dodecyl sulfate gel electrophoresis and an approach to detecting anomalous migration, the enzyme was dissociated in sodium dodecyl sulfate and run at different acrylamide concentrations to determine free electrophoretic mobilities and retardation coefficients (37). A plot of log HF uersus acrylamide concentration (38) for the sodium dodecyl SThe relative amount of protein stain associated with each of the subunit bands was also identical as shown by densitometry at 600 nm (data not shown). The arrow indicates the intercept at the ordinate for five standard proteins (cytochrome c, RNAse, n-amino acid oxidase, ovalbumin, and bovine serum albumin). sulfate-polypeptide complexes is shown in Fig. 2. For all polypeptides, the plot is essentially linear. The free electrophoretie mobilities (intercept at acrylamide concentration = 0) for Subunits II to VII are similar and identical with those of a number of standard proteins subjected to co-electrophoresis on the same gels. However, the free electrophoretic mobility of Subunit I is extremely high (about 50% higher than the average). This anomalous migration makes it virtually impossible tn determine the molecular weight of Subunit I by sodium dodecyl sulfate gel electrophoresis and may have contributed to the variance in subunit number noted above. Because Subunits I and II have identical relative mobilities on 8.5% acrylamide gels (see Fig. 2 focusing of a urea-dissociated enzyme in a gel containing urea and Triton X-100 and ampholytes ranging from pH 3 to 10 revealed seven major bands (Fig. 3). Initial attempts to identify these bands combined sodium dodecyl sulfate gel electrophoresis with gel isoelectric focusing. After electrofocusing, the cylindrical gels were laid across a sodium dodecyl sulfate-polyacrylamide slab gel containing 12% acrylamide and subjected to electrophoresis in a second dimension. Unequivocal results could not be obtained from this system because Subunits I, II, and III precipitated at their isoelectric points and were thus retained by the electrofocusing gel. Isoelectric focusing of yeast cytochrome c oxidase in a urea-polyacrylamide gel. Gels (4%) with pH 3 to 10 ampholytes were used. The gels were stained, scanned, and sliced for pH determination as described under "Methods." However, Subunits IV to VII did enter the second dimension where they formed four discrete spots. Unequivocal results for all seven subunits were obtained by slicing an isoelectric focusing gel, extracting each slice with boiling sodium dodecyl sulfate dissociation buffer and subjecting to electrophoresis each extract on a separate sodium dodecyl sulfate gel. Each protein peak obtained by isoelectric focusing yielded only a single protein band by sodium dodecyl sulfate gel electrophoresis. A two-dimensional analysis using size fractionation by guanidine HCl gel filtration in the first dimension and charge fractionation by gel isoelectric focusing in the second dimension gave similar results. Each polypeptide peak from the gel filtration column (cf. below) gave only one major polypeptide band upon gel isoelectric focusing.
Thus, by the criteria of charge and size, applied separately in one-dimensional systems and together in two-dimensional systems, purified yeast cytochrome c oxidase appears to contain seven polypeptide subunits.

Isolation of Subunits
The seven subunits were isolated by the protocol shown in Fig. 4. This scheme permits the isolation of all seven polypeptides from both crude (3 to 8 nmol of heme a per mg of protein) and purified (9 to 10.4 nmol of heme a per mg of protein) preparations of yeast cytochrome c oxidase. A partially purified enzyme preparation which has not been carried through the final DEAE-cellulose column step is the usual starting material. The key step in this scheme is the use of guanidine-HCl to partially dissociate the enzyme. Denaturation of yeast cytochrome c oxidase in 6 M guanidine HCl and reducing agent at room temperature selectively removes Subunits IV and VI from an aggregate of the other five subunits. Subunits IV and VI can be separated from the aggregate by either of two methods: (a) gel filtration on a 6% agarose column in 6 M guanidine HCl; (b) rapid removal of guanidine HCl by dialysis followed by centrifugation.
Method a reveals a large peak of absorbance at 280 nm in the excluded volume and a smaller double peak (corresponding to a mol wt of 14,000) which contains Subunits IV and VI. Dissociation of the enzyme by higher concentrations of guanidine HCl and at a higher temperature completely dissociates all seven subunits as discrete peaks (Fig. 5). The large peak of absorbance in the excluded volume of the gel filtration column probably represents solely Triton X-100. Method b leads to the quantitative precipitation of an aggregate containing Subunits I, II, III, V, and VII within 1 to 2 hours. Subunits IV and VI remain in solution and can be conveniently recovered by centrifugation.
Because Method b proved to be reproducible and easy, it was adopted as the first step in our purification procedure.
Purification of Subunits Wand VI-The supernatant resulting from the centrifugation of the flocculent precipitate (Method b) contains subunits IV and VI and 75% of the Triton X-100 present in the starting material (as assessed by using [3H]Triton X-100). Subunits IV and VI are separated from each other by DEAE-cellulose chromatography in the presence of 6 M urea (Fig. 6). Omission of urea leads to aggregation of Subunit IV and VI and incomplete separation. Typical recoveries of Subunits IV and VI are 50% and 30%, respectively, of the amount of each subunit loaded on the column. More than 95% of the [3H]Triton X-100 applied to the column emerges in the flow-through (Fig. 6). The remaining [3H]'l'riton X-100 is more or less randomly distributed over all of the column fractions. Counts which appear in fractions containing Subunits IV and VI do not increase across the peak and amounts to less than 0.4 mol of Triton X-100 per 1 mol of protein, indicating that there is little, if any, Triton X-100 bound to these subunits. Because these subunits appeared to be free of bound detergent, we tested their solubility in water by dialyzing them extensively against distilled water and centrifuging them at 209,000 x g for 3 hours. Under these conditions, insoluble proteins or large aggregates of subunits would be expected to form a measurable turbidity or even a pellet. There was no turbidity, no pellet, and no loss of protein for either purified subunit. This finding, in combination with the observation that these subunits do not bind detergent, shows that they are water-soluble under these conditions.
Purification of Subunits I and II-Preliminary fractionation of the flocculent-precipitate remaining after removal of guanidine HCl and dithiothreitol was achieved by denaturation in sodium dodecyl sulfate and gel filtration on 6% agarose in the anomalous migration on sodium dodecyl sulfate gels, the use of high acrylamide concentrations maximizes differences in mobilities of these two subunits and thus facilitates their separation. Between 60% and 80% of Subunits I and II found in the flocculent precipitate are recovered as purified subunits. Purification of Subunits ZZZ, V, and VII-Peak 2 from the 6% agarose-0.5% sodium dodecyl sulfate column is essentially free of non-cytochrome c oxidase protein. Subunits III, V, and VII present in this peak are purified by preparative electrophoresis on 10% or 12% polyacrylamide gels. Recoveries for these subunits vary from 60% to 90% of the amount of each subunit estimated to be in the flocculent precipitate.

Characterization of Isolated Subunits
Homogeneity-The purified subunits move as single bands of protein on sodium dodecyl sulfate-polyacrylamide gel electrophoresis ( Fig. 9) and gel isoelectric focusing (not shown).
Molecular Weights-The molecular weights of the isolated subunits were determined by polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate, by gel filtration in Distance FIG. 9. Densitometric tracings of yeast cytochrome c oxidase (top) and purified subunits (I to VII) analyzed by electrophoresis on polyacrylamide gels (12%) containing sodium dodecyl sulfate. The gels were stained with Coomassie blue R-250 and scanned as described under "Methods." The arrow marks the position of the tracking dye I bromphenol blue). The holoenzyme used for the top tracing contained 10.0 nmol of heme a per mg of protein, and the subunits displayed in each of the lower tracings were isolated as described in Fig. 4. the presence of 6 M guanidine HCl, and by determination of the minimal molecular weight derived from amino analysis.
In order to improve the accuracy of the molecular weights determined by gel electrophoresis, all of the samples were run on a slab gel in slots adjacent to a mixture of standard proteins. In addition, gels of three different compositions were used. The values presented in Table I are the average of duplicate samples of three enzyme preparations (9.5 to 10.3 nmol of heme a per mg of protein) which had been obtained by using three different procedures for the final step in the purification protocol.
As mentioned earlier, Subunits I and V exhibit anomalous migration on sodium dodecyl sulfate gels. Any anomalies resulting from conformation can be assessed by determining molecular weights by gel filtration in 6 M guanidine HCl because protein conformation in this denaturant is quite different from that in sodium dodecyl sulfate (39. 40). As can be seen from Fig. 5 and Table I, there is remarkably good agreement between the molecular weights determined by these two procedures for all seven subunits. The over-all agreement between the values for molecular weight obtained by amino acid analysis and gel filtration in guanidine HCl with those determined by sodium dodecyl sulfate gel electrophoresis suggests that the high per cent acrylamide gels chosen for these studies give reasonably accurate estimates of molecular weight even for Subunits I and V.

Isoelectric
Properties-The isoelectric points of the subunits were determined in the presence of 8 M urea and reducing agent. Under these conditions the value obtained for each subunit should approach that of a completely denatured polypeptide in which all charged groups are accessible and capable of contributing to the migration of the polypeptide in a pH gradient. The total charge exhibited by these denatured polypeptides would be expected to mirror that which could be predicted from their amino acid composition.
As seen in Table  II, all of the subunits of yeast cytochrome c oxidase have acidic isoelectric points. The three large subunits which are synthesized on mitochondrial ribosomes have less acidic isoelectric points than the four smaller subunits which are synthesized on cytoplasmic ribosomes. With those four subunits for which the amino acid composition is known (ct. below), the isoelectric point decreases with an increase in the ratio of acidic to basic amino acid residues.

Amino Acid
Composition-Amino acid analyses of four subunits of yeast cytochrome c oxidase and of the native enzyme were performed on samples of the preparations shown in Fig. 9. The amino acid composition of the holoenzyme (Table III) is very similar to that of cytochrome c oxidase from beef heart or Neurospora crassa (3,41). All three enzymes have a low content of hydrophilic amino acids. The polarities (42) of the holoenzymes from yeast, Neurospora, and beef heart are 39.20/c,, X%5%, and 40.4%1, respectively.
Two subunits (I and II) synthesized on mitochondrial ribosomes and two subunits (IV and VI) synthesized on cytoplasmic ribosomes were chosen for a comparison in order to see if any major differences in amino acid composition could be detected between the two types of translation products. The data of Table III indicate that each of the four subunits possesses a distinct amino acid composition.
However, common to the two mitochondrially made subunits is a high content of apolar amino acids. Subunits I and II have polarities (42) of 34.7% and 42.1%, respectively. The two cytoplasmically made subunits have a low content of apolar amino acids and a polarity of approximately 49%,, characteristic of soluble cytoplasmic proteins. Although all four subunits analyzed exhibit a high content of acidic amino acids, the cytoplasmically made subunits are distinctly more acidic than the mitochondrially made ones. Thus, although the holoenzyme itself has a low polarity typical of many membrane proteins, it is composed of Three different preparations of cytochrome c oxidase isolated by filtration in the presence of 6 M guanidine HCl as described under using either DEAE-cellulose chromatography, hydroxylapatite chro-"Methods". The minimal molecular weights given for Subunits I, II, matography, or sucrose gradient centrifugation as the final purification IV, and VI are taken from Isoelectric points were determined by gel isoelectric focusing in the presence of urea (see "Methods").
All values given represent the average of at least four independent determinations. Minimal molecular weights were calculated with cysteine set equal to 12 for the holoenzyme, with cysteine set equal to 4 for Subunit I, with tryptophan set equal to 4 for Subunit II, with tryptophan set equal to 1 for Subunit IV, and with methionine set equal to 1 for Subunit VI.

DISCUSSION
The results of this study show that the apoprotein component of yeast cytochrome c oxidase is composed of seven polypeptides which differ in molecular weight, isoelectric point, and hydrophobicity.
It has been shown earlier that the three largest polypeptides are synthesized on mitochondrial ribosomes whereas the four smallest polypeptides are synthesized on cytoplasmic ribosomes (1,5,9). We have now shown that this intriguing duality of origin is reflected in the chemical properties of these polypeptides. The mitochondrially made polypeptides are larger, less acidic, and more hydrophobic than the cytoplasmically made polypeptides. Throughout this paper we have referred to these polypeptides as subunits of cytochrome c oxidase. In the absence of decisive biochemical data on the function of a polypeptide in any enzyme, it is difficult to state with certainty that a given polypeptide is a bona fide subunit and not merely a contaminant. This difficulty is amplified for membrane-bound enzymes for which one must also decide "where the enzyme ends and the membrane begins" and for which one can envision 118,000 33,*40 10.26" been aware of this difficulty since the beginning of our studies on yeast cytochrome c oxidase (2). However, on the basis of the following recent observations we are inclined to consider it likely that all seven polypeptides not only are physically associated but also are probably subunits of the enzyme. (a) All seven polypeptides copurify through a series of preparative procedures including DEAE-cellulose chromatography, hydroxylapatite chromatography, and sucrose gradient centrifugation. (b) An antibody which cross-reacts with only one polypeptide precipitates all seven polypeptides (11). (c) Antisubunit functions other than catalysis and regulation. We have bodies specific for either polypeptide II, VI, or V plus VII