22~ SHORT COMMUNICATIONS

Comme le calcul d'interaction est complet pour le monopeptide, on peut penser que toutes les configurations importantes ont 6t6 consid&~es darts l'6tude du dipeptide

Dans le Tableau II, ne figurent que les transitions de plus faible 6nergie qui, dans le dipeptide, se trouvent ~tre les transitions propres aux unit6s peptidiques.

T A B L E A U I I

TABLEAU COMPARATIF DES TRANSITIONS ~LECTRONI~UES (eV')

Monopeptide D~tffide

S.C.F l.C. S.C.F l.C

S 5.3 ° 4.98 5.41 5.22 /

5.42 5.24 T 4.9 ° 4.80 5.02 5.o5

5.o4 5.o7 ( S 7.72 7.29 7.54 7.13

* ! I 7"63 7-42 I T 5.84 5.6I 5-73 5.70

5.81 5.76 S 9.20 8.7I 9.21 8,80

i 9.22 8.89 T 7.32 7.72 7.25 7.73

7.32 7.82

Les r6sultats des transitions rendent compte comme dans le calcu] S.('.F. de la similitude des spectres pour le monopeptide et le dipeptide et restent en excellent accord avec les r6sultats exp&imentaux.

L'interaction de configurations n'apporte pas de modifications sensibles pour les transitions spectrales et permet donc de confirmer les r6sultats obtenus pr6c6demment par la m6thode S.C.F.

Universit~ de Paris, Institut de Biologie MICHP~LE SUARD Physico-Chimique, Paris (Frame)

1 i . SUARD, G. BERTHIER ET B. PULLMAN, Biochim. Bioph),.¢. dcta. 52 (196r) 254. M. SENDER ET G. BERTHIER, J. Chem. Phys., (1958) 384 .

z C. A. COOLSON ET [. FISCHER, Phil. Mag., 4 ° (1949) 386.

Requ le 3 octobre, 1961 Biochim. Biophys. Acta, 59 (r962) 227-228

Protein solubility in solvent mixtures of low dielectric constant

In the course of general studies on protein-protein interactions in solvents of various dielectric constant, we chanced to observe a monotonic decrease in protein solubility as a flmction of solvent dielectric constant. The solubility of most proteins is known to decrease with decrease in solvent dielectric constant and this is often used, empirically, for protein separations1, *. Nevertheless no quantative correlation between this behavior and physico-chemical theory appears to be available 1. We have attempted, therefore, to provide a mathematical basis for the problem by extension of the theory of small dipolar ions to proteins. It was hoped that this would yield an expression that would be analogous to the well known salting-out equation* and that it could be similarly applied to problems of protein and enzyme separation.

Theoretical. At its isoelectric point a protein bears no net charge. A number

Biochim. Biophys. Acta, 59 (1962) 228-23o

SHORT COMMUNICATIONS 229

of charged groups are invariably present, however, and proteins in solution are ionic. Further, all such proteins possess a definite dipole moment , showing tha t their ionic charges are unsymmetr ical ly distr ibuted. If we regard these molecules as prolate ellipsoids 1,*, and locate the dipole charges at the foci, we obtain, according to K IRKWOOD 3,

log / = Kf/ (DkT) 2 (t)

Here / is tile act ivi ty ratio, D tile solvent dielectric constant , K a constant charac- teristic of the protein, I the ionic s trength, k Bol tzman 's constant and T the absolute temperature. Holding I and T constant , a~ is usual in protein separations, we obtain :

x K'D'~0 log ( .' )

x0 D '~

Here x and x o are the mole-fraction solubilities, and D and D O the dielectric constants for the solvent of interest and for the reference solvent, respectively. Since x 0 and D, are constants, one may restate the expression in linear form as:

K" log x . . . . + log xo (3/

Crystalline bovine serum albumin was used in an a l b u m i n - w a t e r - e t h a n o l system to test this proposal. This system was chosen because of its similarity to the svstems used for blood-protein fractionation.

The choice of bovine serum albumin was also dictated by its apparent homo- geneity, the relative cer ta inty of its molecular weight and its known stabil i ty and solubility at the isoelectric point (pH 5-5, sec ref. 2).

Known masses of 5% bovine serum albumin in distilled, deionized water were distr ibuted into tared centrifuge tubes. Conduct ivi ty measurements on these solutions showed their mole-fraction ionic s trengths to be less than ro -4. Known masses of absolute ethanol were added to each of the tubes and the protein precipitate removed by centrifugation at 5ooo × g or more. Protein concentrat ion in the supernatant fluid was determined by spec t rophotometry at 28o m~ after suitable dilution. Under these conditions we found the extinction coefficient of bovine serum albumin to be 43700 1/mole/cm, identical to tha t observed in water. The temperature was held at 2 ° .

A second set of tubes was similarly prepared except for the addition of a neutral salt. LiC1 was used because of its solubility in cold ethanol. Before mixing, the ethanol and serum albumin solutions were each brought to o.oo867 mole fraction ionic strength with LiC1. In this way the ionic s trength of the protein solution remained constant during the addition of the ethanol--LiC1 solution.

The results are presented in Fig. I. Dielectric constants of the mixtures were calculated according to AKERLOF 4. The mole fraction of the serum albumin was calculated using a molecular weight of 69000 (ref. 2).

The fit of the data to the predictions of Eqn. (3) appears to be quite satisfactory. This indicates that , at least at its isoelectric point, the solubility of a globular protein is inversely proportional to solvent dielectric constant . Addition of neutral salt pro- duces a not unexpected salting-in effect. This is reflected by a change in K". Since K " is primarily a function of protein shape and charge distribution, this effect agrees with current proposals as to the effects of ions on protein molecules1, *.

Biochim. Biophys. Acta, 59 (1962), "z8-23o

2.30 SHORT COMMUNICATIONS

We have also examined the glycolic acid oxidase-water-acetone system 5. Except for somewhat different slopes, this system, too, gave excellent straight-line fits to Eqn. (3). As with the serum albumin, salting in was evident in the presence of LiCl.

log X

0

*00 Ol.6 70.7 63.2 57.7 535 50 [ \ 1 I I

-50

-55

zero ionic s t r e n g t l

-6.0

-65

-7.0

I I 0 " 1 .5

mole \ f ~ o c i i o n LiCl 0.00867

I I =1 I I 20 2.5 3.0 3.5 40

¢04102

Fig. i . Mole-fract ion so lubi l i ty in w a t e r - e t h a n o l m i x t u r e s . F ive - t imes recrysta l l ized bovine s e r u m a l b u m i n dissolved in distilled, deionized water , and t r ea ted wi th abso lu te e thanol . T e m p e r a t u r e ,

2°; p H 5.5.

Thus Eqn. (3) and the salting-in effect appear to be of general applicability and should serve as a convenient guide to the dielectric separation of proteins and enzymes. Further, the applicability of Eqn, (3) in practice argues that at least some proteins may be successfully treated as prolate ellipsoid dipolar ions in the derivation of physico-chemical parameters and relationships.

This work was performed in part under the auspices of the U,S. Atomic Energy Commission.

Division of Biological and Medical Research, Argonne National Laboratory, Argonne, Ill., and

St. Procopius College, Lisle, Ill. (U.S.A.)

NORMAN A. FRIGERIO

THOMAS P . HETTINGER

I j . T. EDSALL AND J. WYMAN, Biophysical Chemistry, Academic Press, Inc. , New York, I95~S. pp. 241-385, 591-659.

2 H. NEURATH AND N. BAXLEV, The Proteins, Academic Press, Inc. , New York, I953, pp. 36-57, 549-8o6.

3 j . G. KIRKWOOD, J. Chem. Phys., 2 (1934) 351. 4 G. AKI~RL6F, J. Am. Chem. Soc., 54 (I932) 4125 •

N. A. FRIG~RIO AND H. A. HARBURY, J. Biol. Chem., 231 (1958) 135.

Received November 27th, I96I

Biochim. Biophys. Acta, 59 (I962) 228-230