2946
JULIEHILLAND DAVID J. Cox
Sedimentation Equilibrium of Ovalbumin in Concentrated Cesium Chloride1
by Julie Hill and David J. Cox Clayton Foundation Biochemical Institute and the Department of Chemistry, The University of Texas, Austin, Texas Y8Y18 (Received March 16, 1966)
Ovalbumin has been examined in the ultracentrifuge a t Sedimentation equilibrium in concentrated cesium chloride. From the position of the band of protein in the density gradient generated by the redistribution of the salt in the centrifugal field, the buoyant density of ovalbumin was found to be 1.296 f 0.003 g/cm3. This value agrees quite well with the apparent hydrodynamic density of ovalbumin obtained by plotting the viscosity-corrected sedimentation coefficients observed a t several cesium chloride concentrations against the density of the salt solutions, and extrapolating the resulting linear relation to S = 0. Sedimentation equilibrium experiments a t different rotor speeds and with solution columns of different lengths indicate that the variation of the buoyant density of ovalbumin with hydrostatic pressure is small.
Introduction A previous report has described measurements of the sedimentation velocity of ovalbumin in concentrated aqueous solutions or' five different salts.2 When the sedimentation coefficients, measured at several concentrations of a given salt, were corrected to a reference viscosity and plotted against the densities of the solvents, each set of points defined a straight line which, on extrapolation, crossed the density axis well below the reciprocal of the partial specific volume of the protein (l/Q. The discrepancy between the density intercept and l/S, could be accounted for by variation of the frictional ratio of ovalbumin with salt concentration, by the binding of water in preference to salt of the protein, or by certain combinations of the two effects. For reasons cited previously, it seems improbable that either the molecular weight or the partial specific volume of ovalbumin varies in these solvents. Cesium chloride was one of the salts in which the sedimentation velocity of ovalbumin was measured. In cesium chloride, a direct determination of the density of the sedimenting unit can be done-at one salt concentratiol'--by Of the banding method by hleselson, Stahl, and Vinograd.394 Comparison of the buoyant density of the protein measured by the banding method with the density intercept obmeasurements from the sedimentation allows the elimination of several of the possible ways The Journal of Physical Chemistry
of accounting for the way in which the sedimentation coefficient of ovalbumin varies with the concentration of cesium chloride. For example, if variation of the frictional ratio alone were responsible for the observed dependence of the sedimentation coefficient on the densit,y of the solvent, then the two densities could not agree. If the two values agree and the frictional ratio does change significantly with the composition of the solvent, then the protein must bind one component of the solvent in preference to the other and the degree of preferential solvation must also vary with salt concentration. The simplest explanation of the velocity experiments, that the protein binds water in preference to salt, and that neither the degree of preferential hydration nor the frictional ratio varies with salt concentration, would require agreement between the density intercept of the velocity data and the buoyant density measured by the banding method.
(1) This work was supported by Grant GM-11749 from the U. S. Public Health Service. Some of the work was reported at the Southwest-Southeast Regional Meeting of the Ainerican Chemical Society, Memphis, Tenn., Dec 8-10, 1965. (2) J. Hill and D. J. Cox, J . Phys. Chem., 69, 3032 (1965). (3) M. Meselson, F. W. Stahl, and J. Vinograd, Proc. Natl. Acad. Sci. U.S.,43, 581 (1957). (4) J. Vinograd and J. E. Hearst, Progr. Chem. Org. Aratall. Prod., 20, 372 (1962).
OVALBUMIN IN CESIUMCHLORIDE
Experimental Section Materials. The ovalbumin was twicecrystallized material purchased from Worthington Biochemical Corp. The protein was taken from the same lot used in previous work2 Cesium chloride was obtained from the Gallard-Schlesinger Chemical Manufacturing Corp. Salt solutions for the ultracentrifuge experiments were prepared by dissolving the appropriate amounts of cesium chloride in potassium phosphate buffer at pH 6.8. The buffer was 0.01 M with respect to phosphate and was prepared using glass-distilled water. Solvent densities were measured using l-ml vented-cap pycnometers, and the results were reproducible within *0.001 g/cm3. A sample of the dense and insoluble fluorocarbon FC-43, produced by Minnesota Mining and Manufacturing Corp., was obtained from the Spinco Division of Beckman Instruments, Inc. Ultracentrifugation. Solutions of ovalbumin in concentrated cesium chloride were brought to sedimentation equilibrium a t 20" in a Spjnco Model E analytical ultracentrifuge, and the experiments were observed using schlieren optics. The cesium chloride concentrations used were so selected that the position of the band of protein at equilibrium lay near the middle of the solution column. Ovalbumin, with a molecular weight of 45,000, formed relatively wide bands at sedimentation equilibrium and, in order to obtain bands that were sharp enough to be located accurately, it was necessary to operate the centrifuge a t relatively high rotor speeds. Double-sector cells could not be used with dense solutions a t high velocities, since these generally leaked across the partition between the sectors. The protein solutions and the solvents were, therefore, run separately in single sector cells equipped with 1O negative wedge windows. I n order to c:alculate the density gradient obtained a t sedimentation equilibrium in any particular experiment, the schlieren photograph of water a t the same rotor speed was required. I n principle, the appropriate base line for the density gradient calculation would be derived from a photograph of the schlieren pattern of the salt solution taken as the rotor reaches the operating speed, but before the salt gradient has d e v e l o ~ e d . ~I,n~ practice, it was found that the gradient began to develop rapidly from the ends of the solution column during acceleration and, by the time the rotor reached top speed, no more than the central third of the schlieren pattern remained unperturbed by the redistribution of the salt. Experiments were done in which the developing schlieren pattern of a cesium chloride solution was photographed a t several angular velocities as the rotor was accelerated to 59,780 rpm. I n separate experi-
2947
ments, the cell was filled to the same level with water, and its schlieren pattern was photographed at each of the rotor speeds at which the cesium chloride solution had been observed. The water and cesium chloride base lines both rose appreciably as the centrifugal field was increased. However, a t a given angular velocity, the water pattern coincided quite closely with whatever portion of the cesium chloride base line was still unaffected by the development of the salt gradient. It was concluded that the schlieren pattern of water a t the appropriate rotor speed could be used as the base line for the density gradient calculation in these particular measurements. This procedure might not be correct for other solvents, and its use would have to be justified experimentally in each case. All three of the runs required for a given buoyant density measurement-protein and salt, salt alone, and water-were done in the same cell, using the same windows. The rotor speed, the schlieren bar angle, and the length of the solution column were carefully matched among the three runs. It was found that the column lengths could be reproduced most precisely by weighing the amounts of solutions or solvents introduced into the cell. The relation between the density gradient at sedimentation equilibrium and the area under the schlieren pattern was determined in separate experiments using a valve-type synthetic boundary cell. A cesium chloride solution of known density (PI) was introduced into the centerpiece of the cell, and a solution of lower density ( p z ) was placed in the cup. The schlieren bar was set a t the same position as was used for the equilibrium runs. The rotor was accelerated until the cup emptied, and the boundary was allowed to diffuse until the entire schlieren pattern was visible. The pattern was photographed, the area under the boundary (A1,2) was measured, and the quantity K = (PI P ~ ) / Awas ~ , ~calculated. The measurements were most conveniently done using pairs of solutions that differed in density by about 0.05 g/cm3, and seven pairs were chosen which covered the range of densities encountered in the cell a t sedimentation equilibrium. The precision (f2-3%) of the individual measure ments of K was not sufficient to allow the detection of a significant trend in the values with density; a mean value was used in subsequent calculations. In a few cases, after the schlieren pattern had been photographed a t a relatively low rotor speed, the rotor was accelerated to 59,780 rpm and the pattern was photo( 5 ) J. E. Hearst, J. B. Ifft,and J. Vinograd, PTOC. Natl. Acad. Sei. U . S.,47, 1015 (1961).
Volume 70,Number 9 September 1966
2948
graphed again. The areas under the boundary at the two speeds were compared and were found to be the same. In each of the centrifuge runs to be described, 0.02 ml of FC-43 was introduced into the cell, along with the solution to be examined, in order to ensure that the entire length of the solution column could be observed.5 It was determined that neither water nor cesium chloride was soluble in FC-43 to any measurable extent, and no water-extractable impurities could be detected in the fluorocarbon. I n a few of the experiments, the ovalbumin sample used contained a small amount of insoluble material, presumably denatured protein. During centrifugation to sedimentation equilibrium, this material formed a sharp band which migrated rapidly to a point somewhat below the position eventually occupied by the center of the band of soluble protein. The amount of insoluble protein did not increase visibly during experiments prolonged for several days. Sedimentation velocity experiments carried out with similar, very slightly turbid solutions indicated that the protein remaining in solution was entirely homogeneous. The presence of the precipitated material did not interfere with the measurement of the buoyant density of the soluble protein. When the ultracentrifuge cell was completely filled, giving a solution column about 1.2 cm long, and the centrifuge was operated at top speed (59,780 rpm), the salt reached its equilibrium distribution in about 8 hr and the ovalbumin was a t equilibrium within 20 hr. A few experiments were continued for an additional 12 to 24 hr, but no further change in the solute distribution could be detected. In cases where the effect of rotor speed on the apparent buoyant density of ovalbumin was to be examined, a solution of the protein in concentrated cesium chloride was run to sedimentation equilibrium at 59,780 rpm. The rotor speed was then decreased and the experiment was continued for another 20 hr a t a lower speed. The rotor speed was again decreased and the experiment was continued in the same manner. The complete series of measurements was thus done over the course of several days without emptying and refilling the cell. Calculations. Conservation of mass requires that
where r is the distance from the axis of rotation, PO is the density everywhere in the cell at the beginning of the run, p, is the density at sedimentation equilibrium at position r , and the integration is carried out between the upper and lower boundaries of the solution column. The Journal of Physical Chemistry
JULIEHILLAND DAVIDJ. Cox
Subtracting, from both sides of eq 1, the quantity a".
-'b
d(rz), where
pmJ
pm
is the density a t the meniscus
Tm
sedimentation equilibrium
L"P~ -
Pm
)d(r2) = (PO -
fm)(rb2
- Tm2) (1%)
Twenty to thirty areas between the meniscus and various positions r and between the equilibrium schlieren pattern for the cesium chloride and the water base line were measured. The areas were multiplied by the factor K , whose measurement is described above, to obtain a series of values of ( p r - pm). The integral on the left side of eq l a was evaluated numerically or graphically, and the equation was solved for pm. With the density a t the meniscus known, pr was calculated for each T . The position in the gradient a t which the diphasic schlieren pattern corresponding to the band of ovalbumin crossed the salt base line was taken as the buoyant density of the protein. Since no correction was made for the compressibility of the salt solutions, the values obtained were, more properly, apparent buoyant den~ities.~The purpose of the experiments was to compare the buoyant density with the apparent hydrodynamic density derived from sedimentation velocity measurements. Since the latter values were not corrected for the compressibility of the solvent, the uncorrected buoyant densities were the appropriate ones to use for comparison. As will be seen, the pressure corrections are probably small in any case.
Results The apparent buoyant density of ovalbumin in cesium chloride was measured by the banding method at four different protein concentrations. The centrifuge cell was filled with about 0.7 ml of the proteincesium chloride solution in each case, and the length of the solution column was about 1.2 em. The initial density of each solvent was about 1.30, and each measurement was carried out at a rotor speed of 59,780 rpm. The results are shown in Table I. The table records the ovalbumin concentration at the beginning of the run; a t sedimentation equilibrium, when the protein had collected in a band near the middle of the solution column, the concentration a t the center of the band was between four and five times the initial value. The data in Table I show that, in a given solvent and a t a given rotor speed, the apparent buoyant density of ovalbumin is essentially independent of protein concentration. All subsequent experiments were done using an initial protein concentration of 3 mg/ml.
OVALBUMIN IN CESIUM CHLORIDE
2949
Table I: Apparent Buoyant Density of Ovalbumin as a Function of Protein Concentration Ovalbumin concn, mg/ml
Buoyant density, g/cmS
4
1.297 1.296 1.296 1.300
3 2 1
The measurement, by the banding method, of the buoyant density of a protein of relatively low molecular weight is subject to considerably larger errors than those commonly encountered when the method is applied to proteins of very high molecular weight or to nucleic acids4t6 When the cell is completely filled with a cesium chloride solution whose density is 1.30 and sedimentation equilibrium is established a t 59,780 rpm, the density increases by about 0.21 g/cm3 between the meniscus and the bottom of the cell. Even in the steepest gradients obtainable in the ultrai centrifuge, the protein band is rather wide, and the location of the point a t which it crosses the base line is subject to small errors. I n addition, there is a slight uncertainty in the calculation of the gradient itself, primarily in the reproducible leveling of the schlieren photographs of the salt and water base lines. The relative importance of these difficulties varies somewhat from one run to the next, depending on the rotor speed and on the length of the solution column. Taking these sources of error into account, along with the possible uncertainty of the measurement of pol errors as large as *0.003 g/cm3 could be encountered in the measured values of the buoyant density of ovalbumin. Duplicate experiments gave results that were reproducible within *0.002 g/cm3. A series of experiments was undertaken in an attempt to find conditions under which the precision of the buoyant density measurement could be improved. The two experimental variables tested were the rotor speed and the length of the solution column. Decreasing the rotor speed, for example, reduced the steepness of the density gradient, so that a given absolute error in the location of the band of protein produced a smaller error in the buoyant density. On the other hand, the band itself was less sharp a t the lower rotor speed, and the precision with which its center could be located was correspondingly decreased. With shorter solution columns, errors in the calculation of the density gradient were reduced, but the correct matching of the protein and salt patterns was more difficult. No great improvement in the precision of the results could
be obtained by varying either the rotor speed or the length of the solution column. There seemed to be some slight advantage in the use of short columns and high rotor speeds. The experiments in which the angular velocity or the length of the solution column was varied also provided an opportunity to observe any substantial effect of hydrostatic pressure on the apparent buoyant density of ovalbumin in cesium chloride. Measurements were done with solution columns of three different lengths, each at four or five rotor speeds. The results are collected in Table 11. The initial densities of the cesium chloride solutions were so selected that the protein bands formed near the middle of the solution columns a t each rotor speed. The variation of the angular velocity from 42,040 to 59,780 rpm therefore represents, at a given column length, a twofold change in the pressure at the protein band, since the pressure is roughly proportional to the square of the rotor speed. The length of the solution column was varied over a threefold range. At any given column length, there was no significant variation of the apparent buoyant density with rotor speed. When the results using different column lengths are compared, however, a trend is observed. The buoyant density appears to be slightly higher with longer columns and thus at higher pressures. The apparent variation of the buoyant density with column length is, however, of doubtful significance, since the precision of the individual data cannot be assumed to be better than *0.003 g/cm3. It can be concluded that, if the buoyant density does vary with hydrostatic pressure, the variation is small. Table 11: Apparent Buoyant Density (g/cm3) of Ovalbumin in Cesium Chloride as a Function of Column Length and Rotor speed Rotor speed, rpm
1 . 2 1 cm
59,780 56,100 52,600 47,660 42 040
1.297 1.299 1.297 1.300 1,300
~
Column length---0.80 cm
1.295
0.39 cm
1.293
...
...
1.295 1.295 1.296
1.293 1.293 1.293
Hearst, et u Z . , ~ have measured the buoyant density of bacteriophage T-4 DNA and of tobacco mosaic virus as a function of hydrostatic pressure. They (6) D. (1961).
J. Cox and V. N. Schumaker, J . A m . Chem. SOC.,8 3 , 2439
Volume 70, Number 9 September 1966
2950
report that the apparent buoyant densities of these materials decrease slightly as the pressure is increased. An equally moderate trend of the buoyant density of ovalbumin with pressure would not be outside the experimental uncertainty of the present data.
JULIE HILLAND DAVIDJ. Cox
axis at a point different from the buoyant density measured by the banding method. If the frictional ratio does vary with salt concentration, then the protein must bind water in preference to salt and, moreover, the degree of preferential hydration and the frictional coefficient must vary together in a rather special way in order to produce a linear relation between the sedimentation coefficient and the density of the solvent which, on extrapolation, crosses the density axis a t the equilibrium buoyant density of the protein. Even if the frictional ratio does not vary with salt concentration, a linear relation between the sedimentation coefficient and the density of the solvent which yields the “correct” buoyant density on extrapolation need not imply that the preferential solvation of the macromolecule is independent of salt concentration. It has recently been shown’ that results of the kind described here for ovalbumin will be obtained if the variation of the preferential hydration with density satisfies the equation (Bruner and Vinograd’s eq 5 )
Discussion All of the values of the buoyant density of ovalbumin in cesium chloride measured by the banding method fall between 1.293 and 1.300 g/cm3. These values agree very well with the apparent buoyant density calculated by plotting the viscosity-corrected sedimentation coefficients of ovalbumin in cesium chloride solutions of various concentrations against the solvent densities and extrapolating the plot to S = 0. Three separate determinations, by the plotting method, of the apparent buoyant density of ovalbumin in cesium chloride gave values of 1.293, 1.298, and 1.302 g/cm3. Bruner and Vinograd’ have reported experiments with bacteriophage T-4 DNA in cesium chloride which allow a similar comparison to be made. As is the case with ovalbumin, the viscosity-corrected sedi1 - 6,p mentation coefficient of T-4 DNA varies linearly with = solvent density, and the S us. p plot crosses the density axis at the buoyant density measured by the banding In eq 2, r’ is the preferential hydration in grams of method. preferentially bound water per gram of protein and If the protein did not bind water in preference to &. is the partial specific volume of water. salt and if the frictional ratio did not vary with salt There are, therefore, three possible ways to explain concentration, a plot of the viscosity-corrected sedithe linear relation between the sedimentation coefmentation coefficient against the density of the solficient of ovalbumin in cesium chloride and the density vent mould be linear and would cross the density axis of the solvent. The preferential solvation and the at the reciprocal of the partial specific v ~ l u m e , * ~frictional ~ ratio may vary together with salt concentrawhich is 1.335 for ovalbumin.1° It is necessary to tion, the preferential solvation niay vary alone, in the account for the fact that the experimental plots are way implied by eq 2, or both the frictional ratio and the linear but cross the density axis significantly below preferential solvation may be independent of salt con1/6. centration. The range of possibilities would be further The results are consistent with a situation in which narrowed if it could be determined u-hether or not the the frictional ratio of the protein is independent of degree of preferential hydration varies with salt concesium chloride concentration and in which the protein centration. Hearst and Vinograd” have measured binds water in preference to the salt to a degree which the buoyant density of bacteriophage T-4 DKA in a is also independent of salt concentration. There are, series of cesium salts and have found that the preferenhowever, other possible models, consistent with the tial hydration of the nucleic acid varies from one salt information available, which involve the possible variato another. The activity of mater in the salt solutions tion with salt concentration of the frictional ratio or at the respective buoyant densities of the nucleic acid the degree of’preferential solvation. The agreement between the buoyant densities de(7) R. Bruner and J. Vinograd, Bioehim. Biophys. A c t a , 108, 18 (1965). rived from the plotting and banding methods serves (8) H. K. Schachman and 11. A. Lauffer, J . Am. Chem. Soc., 71, to eliminate one possible explanation of the behavior 536 (1949). of ovalbumin in sedimentation velocity experiments in (9) S. Katr and H. K. Schachman, Biochim. B w p h y s . Acta, 18, 28 (1955). cesium chloride. If the frictional ratio varied and the (10) 11. 0. Dayhoff, G. E. Perlmann, and D. A. MacInnes, J . Am. protein did not bind water in preference to salt, then Chem. Soc., 74, 2515 (1952). either the sedimentation coefficient vs. density plots (11) J. E. Hearst and J. Vinograd, Proc. Natl. Acad. Sci. U. S., 47, would be nonlinear or they would cross the density 1005 (1961).
(3)
T h e Journal of Physical Chemistry
THERADIOLYSIS OF ETHYL MERCAPTAN
also varies from salt to salt, and there is a smooth correlation between water activity and preferential hydration. On t,he basis of these data, each obtained in a different salt, Bruner and Vinograd7 suggest that the preferential hydration of a macromolecule in a given salt should vary with salt concentration since the activity of water does so. Their interpretation would argue against the possibility that the sedjmentation behavior of ovalbumin may be accounted for by the constancy with salt concentration of preferential
2951
hydration and the frictional ratio. It should be possible to approach the problem more directly by measuring the buoyant density of ovalbumin in cesium chloride and deuterium oxide. The salt concentration and thus the water activity a t the protein band should be different in water and in deuterium oxide, and it should be possible, in this way, to determine whether or not the preferential solvation of the protein is the same a t two different concentrations of the same salt.
The Radiolysis of Ethyl Mercaptan1
by J. J. J. Myron and R. H.Johnsen Department of Chemistry and the Institute of Molecular Biophysics, Florida State University, Tallahassee, Florida (Received March 81,1966)
A study of the radiolysis of liquid ethyl mercaptan has been undertaken. The data strongly suggest that the radiolytic behavior of the mercaptan differs significantly from that of the corresponding alcohol. A comparison of the two radiolyses in the light of the present results is presented.
Introduction I n the past, radiolytic studies of compounds containing sulfhydryl groups have generally been carried out on dilute aqueous solutions of polyfunctional thiols. Except for data on the esr spectra of irradiated methyl and ethyl mercaptan2 taken at 77”K, no studies on the radiolytic behavior of these compounds appear to have been made. Investigation of simple mercaptans in the “pure” state should be of interest as a comparison to that of the corresponding aliphatic alcohols which have been extensively studied by various workers. 3--6 The possible importance of thiol radiolysis studies is evidenced by the number of publications dealing with biological systems in which mercaptans were present cysteas additives. Compounds such as aminello and glutathione11*12 have been widely used as “protectors” of biologically significant systems against radiation damage. Oxidation of the thiol group by radiation-produced radicals from the other compo-
nents of the system is probably an important mode of “protection” or “inhibition” afforded by the sulfur compound. The corresponding disulfide and small amounts of hydrogen sulfide are the usual products of oxidation. ~
~~
(1) Research supported in part by A.E.C. Contract AT-(40-1)-2001 and in part by A.E.C. Contract AT-(40-1)-2690. (2) C. L. Luck and W. Gordy, J. A m . Chem. SOC.,78, 3240 (1956). (3) (a) W. McDonnell and A. S. Newton, ibid., 76, 4651 (1954) ; (b) I. A. Taub and L. M. Dorfman. ibid., 84, 4053 (1962). (4) (a) J. G. Burr, J. Phys. Chem., 61,1477 (1957); (b) G. E. Adams, J. H. Baxendale, and R. D . Sedgwick, ibid., 63,854 (1959). ( 5 ) E. M. Hayon and J. Weiss, J. Chem. SOC.,3962 (1961). (6) R. H. Johnsen, J. Phys. Chem., 65, 2144 (1961). (7) S. L. Whitcher, M. Rotheram, and N. Todd, NucEeonics, 1 1 , 30 (1953). (8) P. Rieez and B. E. Burr, Radiation Res., 16, 661 (1962). (9) C. N. Trumbore, et al., J . Am. Chem. SOC.,86, 3177 (1964). (10) B. Shaprio and L. Eldjar, Radiation Res., 3 , 225 (1955). (11) G . E. Woodward, Biochem. J.,27, 1411 (1933). (12) E. S. G. Barron and V. Flood, J. Gen. Physiol., 33, 229 (1950).
Volume 70, Number 9 September 1966