Physicochemical Characterization of Citrate Synthase and Its Subunits*

and also a of The and coefficients found to and X 10 see-I, respectively, and the intrinsic viscosity was 3.95 Optical rotatory dispersion showed a 233 rnp trough and a 198 mp peak with [m’l2~ = -7,150 and = +28,500 Circular dichroism a double These indicate the of a of a-helical segments in the protein

the enzyme. Since the model of Monod, Wyman, and Changeux (3) for allosteric proteins requires the participation of subunits within the enzyme molecule, one of our main objectives was then the identification and characterization of these subunits if they were present.
We first studied the size and conformation of native citrate synthase by a combination of hydrodynamic and optical measurements.
Next, we attempted to dissociate the protein molecule into subunits with various treatments, and succeeded with succinylation and with 6 M guanidine hydrochloride. This was then followed by studying the size and conformation of the two physically indistinguishable subunits (succinylated) .

EXPERIMENTAL PROCEDURE
M&r&--Citrate synthase from pig heart was purchased from Boehringer Mannheim and also prepared in this laboratory according to the method of Srere and Kosicki (4). The commercial sample was a crystalline suspension in 2.2 M ammonium sulfate (pH about 7). Its enzymatic activity was approximately 70 units per mg; the contaminants, based on the specific activity, were less than 0.1 y0 aconitase and isocitrate dehydrogenase and less than 0.1 y0 malate dehydrogenase.
The enzyme was assayed at pH 8.2 with the method of Srere,Brazil,and Gonen (5).
The citrate synthase solution was prepared as follows. The protein suspensions were dialyzed against 0.1 M Tris buffer (pH 8.2) or 0.1 M phosphate buffer (pH 7.4) at 4' for at least 24 hours. (The dialysis tubings had been washed with 1% EDTA to remove any heavy metal ions, then with boiling distilled water, hot 5% sodium bicarbonate, and finally with ethanol.) The dialyzed solution was centrifuged at 12,000 rpm with a Sorvall RC 2-B centrifuge for 20 min to remove any precipitate.
For chemical modifications, the enzyme suspensions were dialyzed against the buffer containing appropriate reagents.
Succinylation of citrate synthase was carried out according to the method of Habeeb,Cassidy,and Slinger (6), with 12 moles of succinic anhydride for each mole of free amino group (assuming 10 moles/100 moles of lysine in the protein molecule).
The pH of the reaction mixture was maintained at 8.2 by adding 1 M NaOH; the reaction proceeded at 4" for 3 hours and excess succinic anhydride was removed through dialysis against 0.1 M Tris buffer for 24 hours.
The concentration of the enzyme solution was determined by the Folin-Ciocalteu method and checked against a standard solution of crystalline bovine plasma albumin (7). For routine analysis, the concentration was also measured spectrophoto- Sedimentation velocity at 20" was measured at 59,780 rpm for native protein and 56,100 rpm for succinylated protein, with the use of schlieren optics. Diffusion was run in a synthetic boundary double sector cell at 9,945 rpm at 20". Sedimentation equilibrium was carried out at 4" or 20" in  and eight-channel circular (9)) with the use of Rayleigh interference optics.
Diffusion coefficient, D, was calculated from the slope of a plot of (A2/H2) against t according to the equation: Viscometry-Viscosity was measured in an Ubbelhode type viscometer with solvent flow time of more than 800 set at 20 f O.lO. The kinematic viscosity, [Y], thus determined, was converted into intrinsic viscosity, [q], according to the suggestion of Tanford (11) : where A is the area under the diffusion peak in centimeters, H is the maximum height of the peak in centimeters, and t is the time in seconds (10). The area was calculated by trapezoidal integration. All areas were converted to zero time according to the formula A0 = At(Xt/Xo)2, where Xt and X0 refer to the boundary position at time, t, and zero, respectively. The mean value of Ao, &, was calculated and then a smoothed value of A, was obtained from A, = &,(X0/Xt)2 and used in Equation 1. Both the sedimentation and diffusion coefficients at infinite dilution, so and Do, were obtained by linear extrapolation of the s-C and D-C plots to zero concentration.
where w is the angular velocity and C is the concentration (expressed in terms of the Rayleigh fringe displacements). The plot of (l/M"PP) against C was linear and its intercept at zero concentration gave the reciprocal of the weight average molecular weight of the protein. Edelstein and Schachman (12) proposed that the molecular weight and partial specific volume, 8, of a macromolecule can be determined simultaneously from sedimentation equilibrium in both Hz0 and D20. This method is based on the changes produced in the concentration distribution at equilibrium due to increase in the density of solution when D20 (or D&*0) replaces Hz0 as the solvent. Thus, B can be calculated according to the equation : k -[d ln C/drz)n&(d ln C/dr2)H201 ' = PD%O -pd(d In C/dr2)~ol(d ln C/dr2)n,01 Here the p's are the densities of the solvents and k is the ratio of the molecular weight of the deuterated protein to that of the undeuterated.
ORDl and CD-ORD was measured with a Cary model 60 recording spectropolarimeter and CD with a Jasco ORD/UV-5 with a CD attachment, both under a constant nitrogen flush. Specially designed thermostatable cell holders and jackets were installed in both instruments. The data were expressed in terms of reduced mean residue rotation, [m'], and mean residue ellipticity, [0] (without reflective index correction), respectively (15). The rotation data in the visible region were analyzed with the Moffitt equation (16) : The extrema of the Cotton effects (see "Results") have now also been used for the estimation of helical content in a protein molecule. With synthetic poly-L-glutamic acid as a reference compound (helical at pH 4.75 and unordered at pH above 6),2 we can have the following expressions: with XO preset at 212 rnp. Adopting the current practice, the bo value is taken as -630 for a 100% a-helix and zero for the unordered form. Amino Acid AnaZysis-The amino acid composition of citrate synthase was analyzed by the method of Moore and Stein (17). One milliliter of 0.1% protein solution was mixed with 1 ml of concentrated HCI in the ignition tube. The mixture was frozen in Dry Ice and degassed exhaustively under vacuum. After sealing, the tube was heated at 110" for 24,48, or 72 hours. The hydrolyzed products were then transferred to a 50-ml round bottom flask and evaporated to dryness. The dried hydrolysates were redissolved in 2 ml of citrate buffer (pH 2.2). Aliquots of this solution were then analyzed in an automatic amino acid analyzer and the results, after heating for 24, 48, or 72 hours, were extrapolated to zero time. These analyses were carried out in the laboratory of Professor F. H. Carpenter of the University of California at Berkeley.
Trytophan was determined spectroscopically according to the method of Edelhoch (18)   of Native Citrate Synthase-We determined the molecular weight of the protein by sedimentation equilibrium. Fig. 1 plots the logarithm of concentration, C, in terms of the Rayleigh fringe displacements, 1-1 in microns, against the square of the distance, r in centimenters, from the center of rotation in both Hz0 and DzO. We then calculate the 0 in Equation 4 from the slopes in the figure.
Our density measurements indicated a 93% DzO in our experiment, which gave k = 1.0144 through interpolation (k = 1.0155 in a 100% DzO solution). This, in turn, gave a B of 0.733 for citrate synthase.
From Equation 3 we then determined the Mapi', which was virtually concentration-independent in the range of concentrations used (Table I). Thus, we obtained a weight average molecular weight of 1.0 x lo5 for citrate synthase at infinite dilution. a Apparent M, at concentration of 2.2 mg per ml. 6 Dimension of 7 to 13: deg cm2 per dmole (mean residue). symmetrical peak suggesting the homogeneity of the preparations (figures not shown). Fig. 2 illustrates a representative plot for the diffusion runs. From the slope we can calculate the diffusion coefficient, D, according to Equation 1 (see "Experimental Procedure").
The diffusion coefficient, Do, was then determined by extrapolating D to zero concentration.

Hydrodynamic
Properties of Citrate Synthase- Table   II lists all of the pertinent results of the hydrodynamic and optical properties of citrate synthase and its subunits (see below). In all cases the sedimentation velocity patterns showed a single The molecular weight of citrate synthase as determined from the combination of so and DO (the Svedberg equation) is very close to that obtained directly from sedimentation equilibrium (Table II). This is also true for the estimated molecular weight based on the Scheraga-Mandelkern equation (Equation  5). In this case we assumed @ = 2.1 x lo6 and used [v] = 0.0395 dl per g. The @ value was so chosen because of the low intrinsic viscosity of the native protein, suggesting a very compact globular molecule.  Table II. The characteristic 233 rnp trough and 198 rnE.1 peak of ORD (Fig. 3) are indicative of the presence of a-helical segments in the protein molecule as are the typical double minimum at 222 and 210 rnp and the maximum at 191 ml from the CD spectrum (Fig. 5). The ORD in the visible region obeys the Moffitt equation (16) ; its negative bo again suggests the presence of cr-helices.
Amino Acid Composition of Citrate Synthase- Table III lists the amino acid composition of the enzyme. The contents of asparagine and glutamine were not determined in our analysis. Our estimate of the free sulfhydryl groups was about eight, suggesting that citrate synthase contains at most one cystine group and very likely none at all.
Succinylation of Citrate Synthase: Number of Subuni+One of our main objectives was an attempt to split this regulatory enzyme into its subunits, if any, by various treatments. Therefore, we measured the sedimentation coefficient and also the optical properties of citrate synthase after treatment with the following reagents: salt of high concentration, organic solvents, urea, guanidine hydrochloride, anionic detergent, p-chloromercuribenzoate, P-mercaptoethanol, and, finally, succinic anhydride. Our first criterion was to detect any marked change in the sedimentation coefficient of citrate synthase in a dilute solution. In all cases we observed a single sedimentation peak. Judging from the ,920,~ values in Table IV, only treatment with 5 M guanidme hydrochloride, 1% sodium dodecyl sulfate, or succinylation might have split citrate synthase into two or more ultracentrifugally indistinguishable subunits. The use of salt as high as 2 M KC1 did not significantly change either the hydrodynamic or optical properties of citrate synthase (Table IV) and the enzyme was fully active in a standard assay, nor did treatment with dioxane, p-chloromercuribenzoate, or ,&mercaptoethanol seem to dissociate the protein into subunits or to effect its optical properties very much (Table IV).
With p-chloromercuribenzoate, the protein solution became turbid and precise measurements of its optical properties were very difficult to obtain.
The enzyme was still partially active in the presence of these organic reagents.
We concentrated our effort on the modification of citrate synthase by succinylation.
Sedimentation velocity experiments of succinylated citrate synthase showed a single symmetrical peak. The corresponding So is about one-half that of the native protein, whereas the diffusion coefficient, Do, only alters slightly upon succinylation (Table II). These observations suggest that the protein molecule was split into two physically indistinguishable subunits after treatment with succinic anhydride. This is fur-ther substantiated by the molecular weight determination which gave a value of about 54,000 from sedimentation equilibrium at one finite concentration (Fig. 6, Curve C) and 55,000 from the combination of so and Do values. Although the number of succinyl groups covalently bound to the protein molecule was not analyzed, the e-NH2 groups of the lysine residues (see Table III) were expected to be completely succinylated in the presence of excess succinic anhydride.
The fully succinylated protein will increase its molecular weight by about 4,600 because of the succinyl groups which have a residue molecular weight of 100. Thus, each subunit should have a molecular weight of 52,000 to 53,000, which is in accord with the values listed in Table II.
The succinylated citrate synthase was no longer enzymatically active.
Its optical properties were also altered; the ORD still  6. Sedimentation equilibrium plots. Curve A, 0.44 mg per ml of protein in 6 M guanidine hydrochloride plus 1.7 X lo-$ M EDTA, centrifuged at 40,000 rpm at 20' for 17 hours; Curve B, 0.5 mg per ml of protein in 6 M guanidine hydrochloride plus 0.1 M P-mercaptoethanol, centrifuged at 40,000 rpm at 20" for 44 hours; Curve C, 2.2 mg per ml of succinylated protein, centrifuged at 35,600 rpm at 4" for 51 hours. retained the 233 rnp trough (Fig. 3, Curve 2) and the CD retained from the sedimentation and diffusion coefficients (14) leads to the double minimum (Fig. 5, Curve 2), but their magnitudes were the same conclusion. On the other hand, analysis of the hydroreduced (Table II), suggesting a decrease in the helical content dynamic measurements of the succinylated subunits suggests after succinylation.
The same was true for the rotations in the that each subunit is more asymmetrical than the native protein. visible region, where the levorotations of the protein increased Conceivably, the two subunits could be associated side by side, upon succinylation and the -bo of the Moffitt equation became thus reducing the asymmetry of the native protein molecule. small as compared with that of the native protein.
To summa-Since the optical properties indicate some significant changes in rize: both the hydrodynamic and optical properties of citrate conformation upon succinylation, the shape of the dissociated synthase are altered significantly when the protein molecule is subunits (succinylated) may differ from that of the native protein dissociated upon succinylation into two physically indistinguish-and any unfolding of the molecule will further contribute to an able subunits.
increase in the asymmetry of the subunits. Subunits in Presence of Denaturing Agent-Since the sedimentation coefficient of citrate synthase was markedly reduced in guanidine hydrochloride solution (Table II), we determined the molecular weight of this protein in 6 M guanidine hydrochloride with and without 0.1 M P-mercaptoethanol by Yphantis' high speed sedimentation equilibrium method (Fig. 6, Curves A and B). Assuming that the partial specific volume of the protein remains the same as that of the native protein, we obtained a weight average molecular weight of 49,000 in both cases. This suggests that citrate synthase can also be dissociated into two physically indistinguishable subunits by this denaturing agent. Furthermore, since the presence of a reducing agent does not affect the results, such a dissociation does not involve the cleavage of the disulfide bond, if any, between the subunits.
Rather, it is caused by the disruption of noncovalent interactions that hold the two subunits together. This is also consistent with the observation that succinylation of citrate synthase only splits the molecule into two subunits.
Sodium dodecyl sulfate also seems promising for the dissociation of citrate synthase (Table II).
However, experimental difficulties made accurate molecular weight determination difhcult.

DISCUSSION
The Cotton effects of citrate synthase and its succinylated subunits (Figs. 3 to 5) display the extrema that are characteristic of a-helix.
The ORD in the visible region shows a levorotation of the native protein less than that expected of a protein molecule containing a moderate amount of the helix. This is more obvious in comparing the Moffitt parameters of citrate synthase with other globular proteins.
Our results show a positive a0 of 30 as contrasted with a negative one that is common for most globular proteins having a bo in the range of our observation ( -240). (Srere (22) reported a0 = -4 and bo = -230.) One other known exception is glutamate dehydrogenase, which has an 00 of +120 and ba of -200 to -270 (23). The origin of this "anomaly" is not clear at present, but it could be due to the presence of other secondary structures such as the P-form, which may alter the ao, but not the 60 (24). Since the contribution of the P-form to the optical activity of a protein molecule is overshadowed when a-helix and p-form coexist, there is presently no satisfactory method of analysis that will overcome this problem, numerous attempts described in the literature notwithstanding.
Succinylation of citrate synthase does reverse a0 from positive to negative and reduce the magnitude of bo (Table II), suggesting some significant change in the secondary and even tertiary structures of the protein molecule.
The molecular weight of citrate synthase as reported in the literature is rather unsettling.
Srere and Kosicki (4) found the enzyme from pig heart to have a sedimentation and diffusion coefficient of 5.0 S and 7.5 x 10m7 cm2 per set, which gave a molecular weight of 56,000, assuming 27 = 0.70 ml per g. Later, Srere et al. (5) reported an so value of 4.5 S for this enzyme from moth, 6.2 S from pigeon, and 6.1 S from pig heart.
More recently, Srere (21) revised the molecular weight to 67,000 based on sedimentation and diffusion measurements and to 80,000 using the Archibald's approach to equilibrium method. Srere and Senkin3 also reported a value of 85,000 for citrate synthase from pig heart.
Our sedimentation coefficient of 6.0 S agrees with the recent value reported by Srere and associates, but our diffusion coefficient of 5.8 X lo-' cm2 per set is much smaller than their 7.5 x lo-' cm2 per sec. The discrepancy between our molecular weight results and theirs could be partly attributed to the difference in the partial specific volume used. If ii = 0.733 ml per g instead of 0.70 ml per g is used, their molecular weight results would have been raised by about 12%. For instance, their value of 85,000 would become 95,000, which is close to our value of 1.0 X 106.
Using experimental values listed in Table II, we estimated for native citrate synthase a helical content of about 40% from the 233 rnp trough method (Equation 7a) as well as the bo method (see Equation 6). On the other hand, the CD data gave an estimate of about 55% and 50% from the 222 and 210 rnp minimum (Equations 7b and 7c), respectively. Of course, such analysis is subjected to several uncertainties (for a review, see Reference 24), and it will be rather premature to attempt to resolve these discrepancies at this stage of development.
Nevertheless, our results indicate that native citrate synthase does contain a moderate amount of a-helix.
ORD and CD are especially powerful in detecting any conformational changes that accompany physical and chemical treatments.
Thus, we can reasonably conclude from the data in Table II that succinylated citrate synthase subunits contain relatively less amount of a-helix than the native enzyme, although succinylation does not destroy the secondary structure of the protein molecule completely.
On the other hand, citrate synthase becomes unordered in 6 M guanidine hydrochloride or 8 M urea.
The low intrinsic viscosity of native citrate synthase clearly indicates that this globular protein is very compact and not highly asymmetrical.
Calculation of the frictional ratio, cf/jo), Succinylation has been used for the dissociation of proteins such as methemerythrin and oxyhemerythrin (25) and catalase (26). Habeeb et al. (6) reported that succinic anhydride attacked specifically and exhaustively the free amino groups in a protein molecule.
Based on our amino acid analysis (Table III), citrate synthase contains about 46 lysine residues in a molecule of about 860 amino acid residues.
Thus, succinylation results in a conversion of about 23 positive charges into negative ones for each of the