THE APPLICATION OF SOLUBILITY MEASUREMENTS TO THE STUDY OF COMPLEX PROTEIN SOLUTIONS AND TO THE ISOLATION OF INDIVIDUAL PROTEINS BY J. S. FALCONER, D. J. JENDEN,* AND D. B. TAYLOR -f Division of Pharmacology and Experimental Therapeutics, University of California Medical Centre, San Francisco, and Physiology Department, Medical School, King’s College, Newcastle-upon-Tyne Received 2nd May, 1952 The influence of total protein concentration on the separability of two proteins is discussed theoretically and rules governing the use of this variable in protein separation deduced. The possible types of specific property solubility test are given, and their use in conjunction with the above rules stated. A new method for the fractionation of rat liver based on the insolubility of proteins in strong salt solutions is described, with some preliminary results of the application of solubility methods to one of these fractions.Solubility experiments on complex protein mixtures can provide valuable information about their composition, and as a result prove useful in separating individual proteins. Moreover, they can provide us with some of the most critical and sensitive criteria of purity which it is possible to obtain. While the meaning of the term “ pure ”, as it is applied to a protein is funda- mental, it has been discussed previously 1 and will not be enlarged upon. From the standpoint of this paper, it seems more useful to examine theoretically some less explored portions of the problem of protein separation and of the analysis of complex protein mixtures.It has long been known that many intracellular proteins and enzymes are not in free solution in the cell but are attached to and form a part of organized cell structures. Before a homogeneous true solution of cytoplasmic or nuclear proteins can be obtained a series of problems relating to tissue disintegration and handling have to be solved. Unlike solubility measurements, these problems cannot be covered by a general theory; rather the methods used should be based on the observance of relevant biological principles, and it will be part of the object of the second section of this paper to comment on some recent developments in this field. In general, the variables which we must recognize in determining a protein solubility are pH, temperature, total protein concentration and concentration of reagent used to alter the solubility of the protein.Experimentally the solubility may be measured as a function of any one of these variables, while the others are held constant ; studies relating solubility either to total protein concentration or to salt concentration have been most extensively used. The former, in which the protein solubility is measured as a function of total protein concentration at con- stant pH and temperature in a solvent of constant composition, can be referred to as a constant solvent test 2 , 3 and is beautifully illustrated by the classical work of Northrop, Kunitz and Herriott.4 If the composition of the solvent is used as the independent variable while pH, temperature and total protein concentration are * United States Public Health Service Fellow.t Introductory section by J. S. Falconer, D. J. Jenden and D. B. Taylor ; experimental section by D. J. Jenden and D. B. Taylor. 40J . S. FALCONER, L). J . JENDEN AND D . B . TAYLOR 41 held constant, we have the variable solvent test. The use of this is demonstrated by the very accurate measurements of Roche and Derrien.5 We cannot expect a variable solvent solubility test to do much more than indicate a lower limit to the number of proteins in solution and give some very approximate idea of the concentration of some of these. The presence of an inflection on the salting-out curve indicates the appearance of a new solid phase, but proteins present in low concentration will not necessarily be detected by gross methods such as nitrogen analyses.Moreover the adsorption of a soluble protein onto, or solution in, a solid phase already precipitated may render detection impossible. The relative positions of inflections do not provide us with a means of identifying a protein, for change in individual protein concen- tration may reverse the order of appearance of saturation points of two proteins. The specific property test was introduced because of the desirability of giving full weight to the important activities and characteristics of individual proteins. These characteristic properties are often extremely specific, and for enzymes suffi- ciently marked to enable detection and measurement to be made in very dilute solution. tion may sometimes be reduced by the sacrifice of active material, the consequences of this should be clearly recognized.If the losses during an isolation are extensive the final product will be a highly selected sample. If the molecules of the original protein are all identical this does not matter, but in many cases, we may be isolating one of a group of proteins all of which have some activity. For example, crystalline catalase 6 has been shown by Brown 7 3 8 to be a mixture of proteins with varying enzymatic activity in attacking hydrogen peroxide, and he has pointed out that some of the disagreement in the literature may have been due to different workers isolating selected samples of the original material. There is evidence that more than one pig liver esterase exists, but the activities of the various enzymes isolated seem to be very similar.2 A clearer understanding of the factors influencing yield in fractionation procedures therefore seems desirable.This problem is bound up with the influence of total protein concentration on the separability of saturation points of individual proteins in a variable solvent test and will be considered in the next section. property solubility test enables the optimum pH, temperature, salt and protein concentration for purification to be estimated. The methods for estimating the first three have already been discussed,2s 3 but the influence of total protein con- centration requires further quantitative calculation. In general successive steps in an isolation tend to increase in difficulty because the remaining impurities are those whose solubility characteristics resemble the protein to be isolated.At the same time, however, the system becomes simpler and therefore more susceptible to analysis. Ultimately we shall be concerned with the separation of two proteins, and we can therefore simplify our calculations to the influence of total protein concentration on the separability of two proteins. The relation between the solubilities of the great majority of proteins and the salt concentration is exponential or nearly so; it follows that in salting-out, the precipitation range of every protein in solution overlaps that of every other protein present. In separating proteins by salting-out, therefore, the problem to be solved is to determine the experimental conditions which reduce precipitation range over- lap to a minimum. Moreover, if two proteins occur in such concentrations that their saturation points occur at the same salt strength, no single fractionation procedure, without alteration in protein concentration, can yield any pure protein.To obtain the maximum yield of pure protein the conditions for maximum separa- tion of saturation points should be determined. The following calculation is useful in that it provides information on the influence of total protein concentration on the ratio of the amounts of the two proteins present as a function of salt strength. THE PROBLEM OF YIELD IN PROTEIN ISOLATlON.-~hile the difficulties Of a11 iSOla- THE USE OF THE SPECIFIC PROPERTY TEST IN PROTEIN ISoLATION.-The SpeCifiC €342 S o L u B I L i r y MEASUKEMENI s Cohn's equation relates solubility S to salt strength I when /3 and k are con- stants as follows : S = exp (p - - Ad).From this the ratio R of impurity to enzyme in a solution saturated with respect where the suffixes E and I refer to enzyme and impurity respectively. On differentiation, dR/dI =;: (kE - kj)R, so that if ki > kE, dR/dl is negative and R decreases with increase in salt strength; on the other hand, if k E > ki, R must decrease with increase in salt strength. For given amounts of two proteins alteration in R must result in separation of saturation points which occurred at the same salt strength, reduced overlap of precipitation ranges and absolute separ- ability of the two proteins. Even when the saturation points are already separate, they may be further parted by altering the total protein concentration in the direction indicated by the relative values of the constants k .It is clear, therefore, that alteration in total protein concentration can be used in separating proteins. It to both is given by R - exp (pi - ,BE -- I(ki -- k ~ } ) , FIG. 1 .-Possible types of specific property solubility test. Vertical axes represent protein in solution ; horizontal axes represent activities. also follows that if ki = kE there is nothing to be gained by altering the total pro- tein concentration, and also that the conditions which make the difference between ki and k~ a maximum enable the greatest benefit to be derived from alterations in protein concentration. Since there will always be some overlap of precipitation ranges the yield of pure protein in the final steps can never be theoretical; its actual value will depend on the limits of the range of protein concentration over which we can work and on the inevitable losses which occur during fractionation.The experimental limits to the range of protein concentration we can use are worth considering. It is usually not practicable to work with protein solutions more concentrated than can be prepared by dissolving a well-pressed moist filter cake in an equal weight of water; such solutions contain about 10-30 % protein. On the other hand, the general procedures of protein purification such as fractionation, etc., are not par- ticularly satisfactory when the total protein concentration goes below 0.5 mg protein per ml.A several hundredfold concentration difference is therefore about the maximum range available for use in fractionation procedures. proteins we can reduce the possible types of solubility tests to four which are illustrated in fig. 1. The linear parts of curves I1 and IV represent salting out of PO§§IBLE VARIATIONS OF SPECIFIC PROPERTY TEST.-Again considering Only twoJ . S. FALCONER, I ) . J . JENDEN A N D D. B. 'TAYLOR 43 pure enzyme and the slopes of these lines provide an estimate of its specific activity, while in I and 111, the linear portions are due to the precipitation of impurity with- out any concomitant precipitation of enzyme. As the curved portions of I and I11 approach thc origin, the ratio of enzyme to impurity steadily increases, the constant k of Cohn's equation for the impurity being greater than the constant k for the enzyme.In I11 and IV the reverse is true and ke > k'. The slope of the tangents to I and 11 at the origin also provide an estimate of the specific activity of the enzyme. No such estimate of the specific activity can be deduced directly from 111. In this case, removal of non-overlapping impurity followed by dilution should give a mixture from which pure enzyme should salt out first as in IV. EXPERIMENTAL It is clearly desirable for studies on the fractionation of cell proteins to start with the separation of the particulate cell components. Each fraction can then be extracted separ- ately and the soluble proteins produced subjected to the available analytical procedures.The methods available for the initial particulate fractionation are broadly divisible into two classes : (i) those employing aqueous medias-14 all of which will dissolve proteins, and cannot be relied upon to leave all proteins in their natural cell components ; (ii) those employing non-aqueous media, viz., the Behrens procedure and its modi- fications.15-17 Mixtures of carbon tetrachloride and cyciohexane or other organic solvents are used to suspend the material for differential centrifugation, after preliminary lyophilization and homogenization in a ball-mill. These procedures are lengthy (1-2 days), and again leave room for doubt as to possible protein losses ; for example, lipo- proteins will almost certainly be dissociated, and catalase has been shown to lose much of its activity after freeze-drying.18 Before a cell fractionation process can be accepted as completely adequate for the systematic examination of cell proteins it must satisfy at least four criteria : (i) Each cell fraction must contain all the proteins in the proportion in which they occur in association with it in the living cell.(ii) The yield must be high. In addition to the desirability of a high yield from the standpoint of reducing initial bulk, it is essential in order to avoid the possibility of selection.19 (iii) Each fraction should be pure as regards freedom from extraneous particulate matter and dissolved or adsorbed proteins from plasma or other cell fractions. (iv) Proteins must be stable in the medium and under the conditions used.This point has already been commented upon in connection with the Behrens procedure, but pre- vention of autolysis and degradation resulting from natural instability must be prevented. This may be facilitated by speed and by working at low temperatures. THE AMMONIUM SWLPHATE TECHNIQUE FOR THE SEPARATION OF RAT LIVER NUCLEI.-- Method-A rat which has been fasted overnight is killed and bled. The liver is removed as fast as possible and forced through a perforated steel plate (holes, 0-015 in. diam.) by means of a press into three times its volume of saturated ammonium sulphate at 0" C . The suspension is then homogenized for 3-4 min in a precision bore glass tube (int. diam. 1 in.) with a Plexiglas rotor (diam. 0-998 in.) at a speed of 780 rev/min.Cooling is effected by immersing the tube in a sodium chloride+ice mixture. After dilution of the homogenate with about half its volume of 0.8 saturated ammonium sulphate it is filtered through one layer of Irish linen (63 threads to 1 in.) and transferred to a refriger- ated centrifuge to be spun at 20,000 g for 10 min. This procedure separates the homo- genate into nuclei, which are packed in the bottom of the tube, cells, mitochondria and precipitated proteins which are packed at the top, and a brownish fluid of varying clarity between. The exact concentration of ammonium sulphate necessary to effect this separa- tion is almost always between 0.8 and 0.9 saturated provided that a fasted rat was used, but must be adjusted to suit each batch if necessary.If the rat has not been fasted, it has been found that the entire homogenate generally sink, even in saturated ammonium sulphate, so that fractionation is impossible. If the two fractions obtained are resuspended and centrifuged under the same con- ditions, the sediments and floating layers separately recombined and the procedure re- peated, further cleaning of the nuclei can be achieved. The rapidity and completion of this process can be facilitated at some sacrifice in yield by increasing or decreasing the44 salt concentration in cleaning the lower or upper fractions respectively, and by reducing the time of centrifugation. If the number of intact cells in the upper fraction is large, this is removed and rehomogenized ; the number of nuclei destroyed in this procedure is small since most of them have already been separated.The entire process as described can be completed in about 1 h. Preliminary experiments show that other salting out agents, e.g. potassium phosphate, 3-5-45 M, containing equal molar proportions of the monobasic and dibasic salts, may also be used. DISCUSSION OF METHoD.-Although it is not claimed by any means that the above technique satisfies all the criteria mentioned earlier, it has some advantages over previously described methods. Differential floating on a dense medium has been used in Behrens type procedures, and achieves a very high degree of separation, while the present method is shorter. The stability of proteins in strong ammonium sulphate solutions is well known,4 and filter cakes prepared from such suspensions can be kept for long periods of so L u B I L 1.r Y M E A s u R EM E N T s FIG.2.Variable solvent test on extract of some soluble rat liver proteins. = optical density at 280 mp, + = optical density at 405 mp, and 0 = pg nitrogen per ml. time at low temperatures. The rapidity with which the tissue is brought into intimate contact with saturated ammonium sulphate ensures that protein loss from all fractions is minimized, since practically all proteins are insoluble in the 0.8 saturated salt. The proportion of three volumes of saturated salt solution to one of liver was chosen to give a final concentration of 0.8 or larger after dilution with the liver fluid. A further advantage of this method is that it allows temperatures below 0" C to be maintained during the pro- cessing owing to the depression of freezing point by the strong salt ; in this way autolysis may be considerably slowed or stopped, especially since it is probable that the enzymes responsible for it have been precipitated.Preliminary perfusion of the liver with saline has been used by several groups of investigators in order to remove red blood cells and plasma proteins ; this has not been used in the present method, partly because the red cells are lysed by the strong salt solution, and partly because perfusion almost certainly removes some protein from the liver cells, while at the same time it increases autolysis owing to the delay.J . S. FALCONER, D. J . JENDEN A N D D . B . TAYLOR 45 Certain disadvantages must be admitted ; adsorption of the soluble cytoplasmic pro- teins on to particulate fractions almost certainly occurs, and some plasma protein remains to contaminate the fractions.PROTEIN FRACTIONATION ExPEmMmTs.-In a protein isolation the repeated fractiona- tions used tend both to reduce the complexity of the system and to leave the impurities behind which are harder to separate. The increasing difficulty of eliminating residual impurities is offset by the fact that specific property test analysis is greatly facilitated by the increasing simplicity of the mixture. For complex solutions, however, different methods will have to be used and the present work was designed to study methods by which useful information about such solutions might be obtained. The existence of well-defined points of inflection on salting- out curves, corresponding to the appearance of new solid phase, has been demonstrated.5 If in conjunction with the determination of such a curve, any characteristics of a protein or group of proteins were measured we could obtain information about the distribution FIG.3.-Relation between protein in solution expressed in pg nitrogen per ml and absorption at 405 mp. or pattern of these characteristics with respect to the relative solubilities of the various proteins. In the present study a variable solvent test was carried out on an aqueous extract of the filter cake prepared from the floating layer described in the previous section. The salting-out was studied by estimating the protein nitrogen and absorption at 265, 280 and 405 mp.The curves show a few well-defined inflections and it is possible that increase in the number of points would provide evidence for many more. Those well-defined inflections which occur provide suitable points for fractionation and it seems reasonable that separation and concentration of a fraction followed by further solubility tests would be a better method of study than simply increasing the number of points. Comparison of the material absorbing at 405 mp with nitrogen left in solution shows some of the complexity of the mixture. Catalase will clearly be responsible for part of the absorption at 405 mp but it is probable that adsorbed bile pigments and other haem proteins also contribute. That catalase itself is not a single protein seems to be estab- lished 20 and its salting-out curve has been experimentally investigated.7, 8 Comparison of protein nitrogen and of material absorbing at 405 mp indicates that the latter is chiefly associated with protein salted-out at the beginning and end of the curve.Investigations of this type, designed to develop analytical methods for complex protein solutions, are proceeding.46 GENERAL DISCUSSION This work was supported by a grant from the Medical Research Committee of the University of California. 1 Eyering, Anal. Chem., 1948, 20, 96. 2 Falconer and Taylor, Biochem. J., 1946, 40, 855. 3 Falconer and Taylor, Proc. llth Int. Cong. Pure Appl. Chem., 1947. 4 Northrop, Kunitz and Herriott, Crystalline Enzymes (Columbia University Press, 5 Roche and Derrien, Proc. llth Int. Congr. Pure Appl. Chem., 1947. 6 Sumner and Dounce, J. Biol. Chem., 1937, 121,417. 7 Brown, Ph. D. Thesis (University of London, 1950). 8 Randall, Proc. Roy. SOC. B, 1951, 138, 301. 9 Douuce, J. Biol. Chem., 1943, 147, 685. 10 Claude, J. Expt. Med., 1944, 80, 19. 11 Schneider, J. Biol. Chem., 1946, 165, 585. 12 Mirsky and Pollister, J. Gen. Physiol., 1947, 30, 117. 13 Schneider and Peterman, Cancer Res., 1950, 10, 751. 14 Wilbur and Anderson, Expf. Cell. Res., 1951, 2, 47. 15 Behrens, 2. Physiol. Chem., 1932, 209, 59. 16 Mayer and Gulick, J. Biol. Chem., 1942, 146, 433. 17 Allfrey, Stern, Mirsky and Saetren, J. Gen. Physiol., 1952, 35, 529. 18 Dounce and Howland, Science, 1943,97,21. 19 Dounce, Ann. N. Y. Acad. Sci., 1950, 50, 982. 20 Sumner, Adv. Eng., 1941, 1, 163. New York, 1948), 2nd ed.