1817
NOTES solvation, owing to the four CHs groups which will induce some hydrophobic solvation.28 The existence of some close-packed structure formation in the vicinity of the i\/IerN+ ion is indicated by the larger values of dVo/dt for 1\Iie4sIz9solutions as compared to that for pure water (Figure 2). Acknotuledgment. Financial assistance from the Society of Sigma Xi and the RESA for buying R4NI salts is gratefully acknowledged. (28) Compare R. Gopal and A. K. Rastogi, J. I n d i a n Chem. Soc., 43, 269 (1966). If the Me4N+ ion is to fall in line with other R I N + ions, the contraction in volume because of “hydrophobic” solvation should be about 5 ml rather than 16.3 ml, which indicates some other solvation as well. (29) Compare ref 26. The structure-breaking nature of the Iion necessitates the Mea” ion t o be a better structure promoter than is indicated by dVoldt values, which give the net effect of the MeaNI salt. The structure-breaking nature, as indicated by nonequilibrium properties like conductance and viscosity, must be due to the “thawed region”27 around the ion, due to electrostatic solvation.
Temperature Coefficients of Protein Partial Volumes
by Henry B. Bull and Keith Breese Biochemistry Department, University of I o w a , Iowa City, I o w a (Received August 1:7,1067)
6,9340
Partial specific volumes of proteins have been measured on numerous occasions largely in connection with sedimentation studies usually by one of three techniques : pycnometry, magnetic floats, and density gradient columns. More recently, Hunter1 has devised a diver method suitable for small volumes of solution. Information concerning the temperature coefficients of the partial volumes of proteins is limited, and we have attempted to extend our knowledge of this subject. We have adapted the specific gravity balance for this purpose.
Methods and Materials A brass rod about 1.3 cm in diameter and about 9.5 cm long was rounded at the ends, a small hook embedded in one end, and the whole gold plated. One pan of a Sartorius anatytical balance was removed and the plummet suspended from the unoccupied arm by means of a constantan wire 0.0071 cm in diameter. The balance could be read to 0.1 mg and estimated to 0.01 mg. The balance sat on a sturdy low table, the bottom of which was enclosed except for a front door. The constantan wire extended through the floor of the balance and through the top of the table. The solution whose density was to be measured was held in a 60-ml test tube sealed in a jacket through which water a t the desired temperature was circulated at a brisk rate. The
Figure 1. Plot of the apparent partial specific volume of KC1 against temperature for a 1.0004 w t % solution of KCI.
temperature was held within 0.01” of a given setting and was read with a set of Brooklyn thermometers calibrated by the National Bureau of Standards. The test tube and jacket were in the cabinet beneath the table. The plummet was weighed in air and, without disturbing the plummet, about 40 ml of the solution was run into the test tube such that the plummet was completely immersed and weighed at a series of temperatures extending from 5 to 45”. The volume of the plummet as a function of temperature was calculated from the measured buoyancy and the density of freshly boiled distilled watern2 The volume of the plummet increased linearly with temperature. The average of five determinations of the density of a KC1 solution containing 0.900034 g/100 g of solution was 1.001948 at 28.00” with a standard deviation of the mean of 0.93 X Hunter1 gave 1.001954 for the density of a KCl solution of this concentration and temperature, and the value she interpolated from the “International Critical Tables” was 1.001952. The apparent partial specific volume of the solute has been calculated by the expression @ =
1 - WlVl w2
where W1 and W zare the weights of solvent and solute, respectively, in 1 ml of solution and Til is the specific volume of water all at the experimental temperature. Figure 1 shows a plot of the partial specific volume of KCl. The specific gravity balance requires, in principle, a surface tension correction if the calibrating liquid and the experimental solutions have different surface tensions: the suspending wire passes through the surface. The film pressures of the protein solutions used would be expected to be about 28 dyn/cmj3corresponding to a (1) M. J. Hunter, J.P h y s . Chem., 70, 3285 (1966). (2) H. S. Harned and B. B. Owen, “The Physical Chemistry of Electrolytic Solutions,” 3rd ed, Reinhold Publishing Corp., New York, N. Y., 1958. (3) S. Ghosh and H. B. Bull, Biochim. Biophys. A c t a , 66, 150 (1963). V o l u m e 72, Number 6
M a g 1968
NOTES
1818 buoyancy correction of about 0.6 mg for a wire 0.0071 cm in diameter. The constantan wire was, however, poorly wet both by mater and by the protein solutions. We have attempted to estimate the surface tension effect by suspending 10 short sections of the suspension wire from a Lucite platform. The wires were well separated from each other and extended vertically in water. The water was replaced by 5% solutions of the proteins and the platform again reweighed. There was a gain of 0.87 mg per wire in going from air to water. As compared with the weight with the wires in water, there was a gain of 0.08 mg per wire in the methemoglobin solution, 0.34 mg per wire in the bovine serum albumin solution, and a loss of 0.03 mg per wire of the egg albumin solution. Thus it appears both the methemoglobin and the bovine serum albumin solutions wet the wires better than did water. If the surface tension correction be applied to the calculated partial specific volume of bovine serum albumin at 25", the calculated value of 0.7376 is reduced to 0.7371. We have refrained from applying the surface tension corrections. Since brass weights were used, there is no air buoyancy correction to be made to the air weight of the plummet. Upon immersion, the air buoyancy due to the plummet disappears and a corresponding air buoyancy of the weights appears. Since the correction applies both to the water calibration as well as to the experimental solutions, the correction tends to cancel, but not quite; it becomes a second-order correction. Air buoyancy corrections have been applied to all of our results. The three proteins investigated were egg albumin, bovine methemoglobin, and bovine serum albumin. Egg albumin was crystallized and recrystallized twice from fresh hen eggs by the method of Kekwick and CannanO4 Bovine oxyhemoglobin was crystallized and
0.76
0
Q)
0
0
I
I
I
I
I
I
I
I
I
6
&Q
I5
20
25
30
35
40
45
TEMPERATURE
Figure 2. Plots of the apparent partial specific volumes of proteins against temperature: curve 1, bovine methemoglobin; curve 2 egg albumin; curve 3, bovine serum albumin.
The Journal of Physical Chemistw
recrystallized in the cold from red cell hemolysates using cold ethanol. Crystalline bovine serum albumin was from Sigma Chemical Co. The proteins were exhaustively dialyzed against water and then passed through mixed-ion-exchange columns. The oxyhemoglobin and serum albumin were lyophilized. During lyophilization, the oxyhemoglobin was quantitatively converted into methemoglobin as shown by its absorption spectrum from 500 to 600 mp. Dissolved air was removed from the protein solutions by applying a vacuum after they had been warmed to about 35". Protein concentrations were determined on the solutions after the completion of the density measurements, by drying weighed aliquots of the solutions in a vacuum oven at 105" for 24 hr. About 2, 5 , and 10% solutions of egg albumin, about 5 and 10% solutions of methemoglobin, and about 5% solution of bovine serum albumin were used. In agreement with the conclusions of Dayhoff, Perlmann, and A l a ~ I n n e s ,no ~ dependence of the apparent partial specific volumes on protein concentration could be detected ; accordingly, we have averaged our results from the different concentrations, and these are reported in Figure 2 . The fact that the apparent partial specific volumes are independent of protein concentration means that the values given are actually partial specific volumes.
Results and Conclusions The apparent partial specific volume of KC1 as a function of temperature is not an essential part of this report, but since me have this information on hand, we are including it. Figure 1shows a volume-temperature plot for a solution containing 0.010004 g of KCl/g of solution. As can be seen, the relation is not linear, and over part of the plot, the slope is large. At 20" the slope is 11.3 X ml/deg. Shown in Figure 2 are plots of the apparent partial specific volumes of bovine serum albumin, of egg albumin, and of bovine methemoglobin as a function of the temperature. The plots are linear up to 30 or 35", but above these temperatures the slopes decrease significantly. For the linear portions, the slope of bovine serum albumin is 4.70 X ml/deg, for egg albumin 4.27 X ml/deg, and for bovine methemoglobin ml/deg. The apparent partial specific 4.07 X volumes at 25" are: bovine serum albumin, 0.7376; egg albumin, 0.7477 ; methemoglobin, 0.7583. Our value for the apparent partial specific volume of bovine serum albumin a t 25" is significantly higher than that given by Hunter (0.7348).' It is also larger than that reported by Dayhoff, Perlmann, and MacInnes (0.7343).s We notice, however, that Charlwood6 R. A. Kekwick and R. K. Cannan, Biochem. J., 30, 227 (1936). (5) M. 0. Dayhoff, G. E. Perlmann, and D. A. hiaclnnes, J . Am. Chem. SOC.,74, 2515 (1952). (6) P. A. Charlwood, ibid., 79, 776 (1957). (4)
NOTES
1819
finds the partial specific volume of this protein depends both on the sample lot and on the environment of the protein. For example, he reports a value of 0.7381 obtained after dialysis against water at pH 4.9 and one of 0.7357 for dialysis at pH 5.1. Failure to dry the protein leads, as pointed out by Hunter,l to a larger calculated partial specific volume. It does not appear, however, that our differences can be reconciled on the basis of water removal from the protein. The situation in respect to the temperature coefficient of the apparent partial specific volume of bovine serum albumin is even more disturbing. Our coefficient of ml/deg is indeed significantly different 4.70 X from the one reported by Hunter of 3.65 X As Hunter remarks, the temperature coefficient does not depend on protein concentration, and, accordingly, the extent of moisture removal from the protein should be unimportant. We are unable to account for the difference betweein our thermal coefficient and hers. Our value of 0.7477 for the apparent partial specific volume of egg albumin at 25’ compares favorably with that reported by Dayhoff, Perlmann, and &.lacInnes5of 0.7481. We have been unable to find a literature value for the apparent partial specific volume of bovine methemoglobin. It appears likely that the thermal coefficient of the apparent partial specific volume reflects the thermal expansion of the protein as well as the release of water of hydration with increasing temperature.
Acknowledgment. We wish to thank Dr. Carl S. Vestling for the loan of the Sartorius balance. This research was supported by a grant from the Division of lllolecular Biology, Xational Science Foundation.
Studies of the Solvent Effects on the Chemical Shifts in Nuclear Magnetic Resonance Spectroscopy. V.
A Model for the
Benzene Solutions of Polar Molecules’
by Taku RIatsuo Department of Organic Synthesis, Faculty of Engineering, K y u s h u University, Hakozaki, Fukuoka-Shi, J a p a n (Received September IS, 1967)
Anomalously large solvent shifts of proton signals in nmr spectra of polar molecules in the benzene solutions have been reported by many authors. Ronayne and Williams recently proposed general principles to be used in predicting the directions and the relative magnitudes of the benzene-induced solvent shifts.2 It is
suggested in that paper that benzene solvent molecule, an able r-electron donor, will solvate the electrondeficient site of each local dipole in a solute molecule, probably in a transient 1: 1 association. The concept of single 1: 1 association with a specific conformation is very attractive to organic chemists. It should be noted, however, that the interaction energy is only a few kilocalories per mole,3-6 as Ronayne and Williams admitted. This is close to the translational energy of molecules at the room temperature. The interactions in this energy region have been generally considered as due to van der Waals forces. Since van der Waals forces are known to be additive, one should expect that a solute molecule may be interacting with several surrounding molecules at the same time. Then it is rather difficult to believe that the time-averaged environments are always represented by the single 1: 1 interaction between the solvent benzene molecule and each electron-deficient site of the solute. I n fact, some of the observed values for the benzene-induced shifts are too large to be explained as being due to the formation of the l : l complex with a reasonable conformation at the electron-deficient site of each local dipole. For example, the benzene-induced shifts for the completely complexed species have frequently been observed to exceed 2 ~ p m . ~ - ’On the basis of Johnson and Bovey’s calculation,* the values are considered to be reasonable only if the solute protons are at about 2.6 A directly above (or below) the plane of a benzene ring. In order to satisfy this condition, the C-H bonds of the solute molecules are required to be approximately perpendicular to the plane of the benzene molecule. Ronayne and Williams suggest that noncoplanar conformations are expected of the 1:1 association where the simple coulombic interactions between the 7r electrons and the local dipole of the solute molecule play major roles. The present author believes that such conformations are very unlikely in the association between benzene and rather strong r acceptors, like maleic anhydride6 and p-nitroben~aldehyde.~Even though the chargetransfer forces are not as important, the interacting molecules are very likely to take almost parallel conformation where the attractions due to the dipoleinduced dipole interaction and the dispersion interaction will be certainly larger than otherwise. In this case, the solute protons may be at least 3.5 8 apart from the plane of the benzene molecule. The shielding (1) Contribution No. 140 from the Department of Organic Synthesis, Faculty of Engineering, Kyushu University, Fukuoka, Japan. (2) J. Ronayne and D. H. Williams, J . Chem. Sac., B, 540 (1967). (3) J. H. Bowie, J. Ronayne, and D. H. Williams, ibid., 785 (1966). (4) Y. Kakayama and T. Matsuo, J . Chem. Sac. J a p . , I n d . Chem. Sect., 69, 1925 (1966). (5) R. E. Klinck and J. B. Stothers, Can. J . Chem., 44, 37 (1966). (6) T. Matsuo, ibid., 45, 1829 (1967). (7) T. Matsuo and Y. Kodera, J . P h y s . Chem., 70, 4087 (1966). (8) C. E. Johnson, Jr., and F. A. Bovey, J . Chem. Phys., 2 9 , 1012 (1958).
V o l u m e 79, X u m b e r 6
M a y 1968
