On the Subunit Structure of the Cold Labile Adenosine Triphosphatase of Mitochondria*

Abstract Mitochondrial ATPase has been studied by equilibrium ultracentrifugation. Under virtually all conditions examined, the oligomeric structure (mol. wt. 2.8 x 105) is in equilibrium with smaller polypeptide chains. In dilute buffer, however, the proportion of dissociated subunits become appreciaable only in the cold (5°). At this temperature, the molecular weight of the subunits can be measured directly, and a value of 4.6 x 104 is obtained. Additional experiments in solutions of guanidine hydrochloride (6 to 9 m) indicate that no further dissociation beyond the 4.6 x 104 unit occurs. The results are evaluated in terms of a probably hexameric structure for the ATPase.

Mitochondrial ATPase has been studied by equilibrium ultracentrifugation. Under virtually all conditions examined, the oligomeric structure (mol. wt. 2.8 X 105) is in equilibrium with smaller polypeptide chains. In dilute buffer, however, the proportion of dissociated subunits become appreciaable only in the cold (5"). At this temperature, the molecular weight of the subunits can be measured directly, and a value of 4.6 x lo4 is obtained. Additional experiments in solutions of guanidine hydrochloride (6 to 9 M) indicate that no further dissociation beyond the 4.6 x lo4 unit occurs. The results are evaluated in terms of a probably hexameric structure for the ATPase.
The ATPase of mitochondria is currently being studied intensively as a homogeneous coupling factor of oxidative phosphorylation and as one of the few well characterized enzymes exhibiting inactivation as a result of cold-labile subunit dissociation (l-3).
These two aspects of the ATPase molecule are intimately related since the inactivation (and presumably dissociation into subunits) in the cold is repressed when the enzyme is integrated in the mitochondrial membrane.
In addition, oligomycin inhibition observed with the enzyme in situ is not present with the isolated enzyme. In order to begin to relate the functional properties of ATPase to its quaternary interactions and conformation, we have undertaken a study to define more precisely the protein's subunit structure.
EXPERIMENTAL CONDITIONS ATPase, prepared as described by Horstman and Racker (4), was stored as the ammonium sulfate precipitate at 4" and 6 mg per ml, dry weight.
Samples were prepared by centrifuging aliquots at 16,000 rpm at 5" for 5 min and dissolving the protein in buffer to the desired concentration.
The buffer was 20 mM potassium phosphate and 2 mM EDTA neutralized with KOH to pH 7.3. For a few experiments, the pellet was dissolved in a buffer containing 0.25 M sucrose, 10 InM Tris-sulfate, 4 mM ATP, and 2 mu EDTA, pH 7.4, and passed through Sephadex as described by Penefsky and Warner (3). Sedimentation equilibrium ex-* This study was supported by National Science Foundation Grant GB-8773.
periments were performed with the use of absorption or interference optics as described previously (5, 6). Experiments with interference optics were favored especially for solutions in the buffer containing ATP, since the absorption of the nucleotide restricted measurements with the scanner in the ultraviolet range. Molecular weight calculations were based on a partial specific volume of 0.74 as estimated from the amino acid composition (7).

AND DISCUSSION
Experiments with ATPase under conditions that maintain the oligomeric structure (moderate temperature, low salt, and added ATP) confirmed the finding of Penefsky and Warner (3) of a 12 S species with a molecular weight of about 2.8 x 106. The results, as presented in Table I, however, were somewhat more variable (10 to 15%) than is generally found with homogeneous proteins.
Two factors are likely to be responsible for this variation. First, the preparation may have contained small amounts of the inhibitor associated with mitochondrial ATPase (4, 8). The inhibitor, which has a molecular weight on the order of 4 x lO*,l would tend to raise the observed molecular weight to the extent that it is present.
Second, a close examination of the sedimentation equilibrium data for 25", particularly with absorption optics, revealed that, at very low concentrations of protein, some dissociation occurs, especially when no ATP is added. The dissociation is revealed by upward curvature of log c verms r2 near the meniscus (Fig. 1). The observed molecular weight for oligomeric ATFase will tend to be lowered by the dissociation into subunits.
The observation that dissociation of ATPase is noticeable at 25" and pronounced at 5" (Fig. 1) even in the absence of added excess salt permitted the subunit molecular weight to be determined directly.
Since the lighter species is predominant at 5", its molecular weight can be determined from the sedimentation equilibrium data of log c versus r2 from the region of the solution column near the meniscus.
The line obtained from these points can then be extrapolated to the bottom of the cell and subtracted from the observed curve to provide an estimate of the heavier species present as well (6). This procedure is illustrated in Fig.  2. The results for the lighter species in a number of such experiments are presented in Table II. Considering that data are obtained from only a limited portion of the centrifuge cell, the data from different runs are in good agreement and suggest a molecular weight of 4.6 x lo4 for the subunit formed in the cold.   1. Sedimentation equilibrium of ATPase at 5'. The experiment was performed at 15,000 rpm in buffer without ATP. Data was recorded with the absorption optical system scanner and light of 280 rnp. The samples were centrifuged overnight and equilibrium w&s verified by the invariance of traces recorded at an interval of 1 hour.
The molecular weight of the heavier species was also estimated from the bottom of the curve, although in this case only data at the lower speed (15,000 rpm) were suitable for the analysis.
The results, which show some variation largely attributable to the nature of the estimation by subtracting the lighter species, nevertheless indicate a molecular weight of 2.8 X lo5 i 10%.
The   (3), from experiments with sodium dodecyl sulfate, reported a much lower subunit molecular weight of 29,000 and suggested a lo-subunit structure, the possibility of further dissociation beyond the 4.6 X 104 level was considered.
To determine whether even smaller units are present in the molecule, sedimentation equilibrium experiments Equilibrium was achieved after 40 hours of centrifugation at 40,000 rpm, 25'. Data were collected with the absorption optical system at 280 rnp.
were conducted with solutions of guanidine hydrochloride (6 M). As seen in Fig. 3, a linear dependence of log c versus r2 was observed, as expected for a homogeneous system. Since the solution contains three components, the ATPase, guanidine, and water, the partial specific volume (ii) alone cannot be used in calculating molecular weight.
Rather, it must be replaced by the term 4' to correct for possible preferential interactions between the protein and either guanidine or water (9). Assuming that 4' bears the same relationship to ii as is found for aldolase in 6 M guanidine hydrochloride, i.e. 5 -4' = 0.01 to 0.02 (9), values of 4' = 0.72 to 0.73 must be employed for the ATPase. These values correspond to molecular weights of 4.5 X lo4 to 4.9 to 104. Molecular weights in this range, which are consistent with the value 4.6 x lo4 obtained in the absence of guanidine (Table II), provide strong evidence that the 4.6 X lo4 unit is the smallest size polypeptide chain present. Subsequent experiments at 9 M guanidine also revealed no further dissociation.
The simplest interpretation of the results of these sedimentation equilibrium experiments is that mitochondrial ATPase is an oligomer built up of six polypeptide chains of molecular weight 4.6 x 104. No evidence is available at the present time concerning whether or not these chains are identical.
If they are 2 J. 0. Thomas and S. J. Edelstein. manuscrint in arenaration. different in size, the differences would have to be relatively small to account for the homogeneity observed in guanidine hydrochloride (Fig. 3). The chains could be of similar size but quite different composition, however.
In this regard, more precise chemical analysis and the determination of the number of binding sites for ATP would be of interest.
The results presented here also permit an interpretation of the 3 S, 9 S, and 12 S components of ATPase observed by Penefsky and Warner (3). The 12 S species is the oligomer of molecular weight 2.8 x lo5 present at 25". We observed that, at low temperature and in dilute buffer, only the 9 S and 3 S species are present, in roughly comparable amounts.
Since only material with molecular weights of 2.8 x lo5 and 4.6 X lo4 is observed at sedimentation equilibrium under these conditions (Fig. 2), it is likely that the 9 S species is identical in molecular weight with the 12 S macromolecule but has undergone a conformational change to a more extended structure with a higher frictional coefficient.
In effect, the 9 S species would be an intermediate on the way to the 3 S unit.
These relationships may be summarized as: 12 S (mol wt = 2.8 X 105) 11 9 S (mol wt = 2.8 X 105) 11 6 @ 3 S (mol wt = 4.6 X 10') The equilibrium may be shifted, by cold or salt (3), progressively, first to the 9 S species, then to the 3 S species. Although cold lability is often assumed to reflect hydrophobic bonding, the salt lability of ATPase (3) suggests that ionic interactions, which, like hydrophobic interactions, are predominantly entropic (lo), may be the major factor involved.