PROTEIN
EXPRESSION
AND
PURIFICATION
2, 83-89 (19%)
Effect of Divalent Ions on Protein Precipitation Polyethylene G lycol: Mechanism of Action and Applications
with
Sonja L. Thrash, James C. O tto, Jr., and Thomas L. Deits Department
of Biochemistry,
Received February
15,1991,
Michigan
State University,
East Lansing,
Michigan
48824
and in revised form April 15, 1991
Polyethylene glycol (PEG) is extensively employed for protein purification by fractional precipitation. Efficiency of precipitation is highest when the solution pH is near the isoelectric point of the target protein. At pH values far from the isoelectric point of the target protein, proteins develop a net positive or negative charge and are more resistant to precipitation. W e have found that divalent cations (Ba2+, Sr2+, and Ca’+) or divalent anions (SO:-) significantly change the pattern of PEG precipitation when the ion is chosen so as to counteract the expected net charge on the target protein. At moderate (S-50 mM) concentrations of Ba2+, negatively charged proteins can be precipitated from solution at pH values as high as 10 with efficiency unchanged from precipitation at pH values near their isoelectric point values. The mechanism of PEG precipitation of protein at these high pH values appears to be unchanged from the mechanism operative at the protein isoelectric point. Precipitation is rapid and the capacity for protein precipitation is high. There is no detectable coprecipitation of small molecules (AMP, ATP, and NADH) or soluble proteins (carbonic anhydrase) induced when large quantities of protein are precipitated by this method. The purification of bovine carbonic anhydrase from erythrocyte lysate is more efficient at pH 10 in the presence of Ba2+ than is conventional PEG precipitation carried out at the isoelectric point of carbonic anhydrase. Application of these observations should broaden the utility of protein purification by fractional precipitation with PEG. o 1991 Academic POW, IIIC.
Polyethylene glycol (PEG)’ has proven to be a valuable reagent for the fractional precipitation of proteins
from solution with minimal perturbation of native structure (1). The efficiency of precipitation is sensitive to the ionic state of the protein; precipitation is usually most effective near the isoelectric point of the chosen protein, where protein net charge is minimal. Consequently, if protein or cofactor stability dictates that the protein be maintained at a pH distant from the isoelectric point value, PEG precipitation will usually be much less effective. A quantitative molecular explanation for the solubility of proteins in PEG has not yet been established; thus application of PEG precipitation remains somewhat empirical. W e have found that the addition of millimolar levels of divalent ions to a PEG precipitation mixture frequently renders PEG precipitation far less sensitive to the isoelectric point of the protein of interest, as long as the ion is chosen so as to complement the net charge on the protein at the pH chosen. Effective precipitation can be accomplished at pH values from 4 to 13 with the appropriate choice of ion. This observation extends previous studies which have used pH to modulate protein net charge (3) and emphasizes the dominant role played by protein net charge in PEG solubility. This observation in turn provides a unique means of protein precipitation at alkaline pH in which all reagents remain in solution as protein is precipitated. Applications should include protein fractionation, routine microscale deproteinization, and recovery of cofactors or metabolites with special pH sensitivities. W e demonstrate selected examples of these applications.
MATERIALS
AND
METHODS
Reagents ’ Abbreviations used: PEG, polyethylene glycol; PEG 3350, polyethylene glycol, average molecular weight 3350 (formerly referred to as PEG 4000 by the supplier (Sigma Chemical Co.); OAc, acetate ion; Hepes, 4-(2-hydroxyethyl)-1-piperazine-ethanesulfonic acid); Mes, 4-morpholine-ethanesulfonic acid. 1046-5928/91
$3.00
Copyright 0 1991 by Academic Press, Inc. All rights of reproduction in any form reserved.
Proteins and reagents were purchased from Sigma Chemical Co. unless otherwise indicated. PEG 35,000 was obtained from Fluka. The Bradford protein reagent was purchased from Bio-Rad. 83
84 Protein
THRASH,
OTTO.
Determination
Protein was determined by the Bradford method using the manufacturer’s directions (Bio-Rad) for the microassay (1 ml final volume). In control experiments, we found that PEG at concentrations above 2-3% (w/v) can interfere slightly in this assay, so all readings were compared against protein-free controls containing the same PEG concentration as the sample. PEG Precipitation Unless otherwise noted, well-mixed solutions containing all components except protein were prepared in l&ml microcentrifuge tubes and chilled on ice. Protein was added and the solutions were thoroughly mixed on a vortex mixer and allowed to stand on ice for 5 min. The precipitate was pelleted by centrifugation in a refrigerated benchtop microcentrifuge at ca. 12,000g for 3 min. In each experiment, the pH of the supernatant was checked to confirm that the indicated additions had not significantly (more than 0.1 pH unit) altered the solution pH. A 50% aqueous solution (grams PEG/100 ml solution) of PEG 3350 was routinely employed as a stock solution that could be conveniently pipetted. For those experiments where higher PEG concentrations were required, PEG was weighed directly into microcentrifuge tubes and a measured volume of water added. A calibration curve was generated to calculate the weight/volume percentage of these samples. This curve was generated by mixing known weights of PEG with known volumes of water and determining the resulting volume of the mixture. Alcohol Dehydrogenase
Actiuity
Assay
Alcohol dehydrogenase activity was assayed using (final concentration) 0.1 M Tris-Cl, pH 8.8, 5 mM MgCl,, 0.5 m M NADH, and 20 m M ethanol. Activity was monitored as the initial rate of formation of NADH at 340 nm. Bovine Erythrocyte
Lysate
Lysate preparation. Bovine blood was kindly provided by Professor R. Fogwell, Michigan State University. Blood was collected into a citrate anticoagulation buffer. Erythrocytes were washed three times with 0.9% NaCl using low-speed centrifugation. Cells were frozen at -20°C. Lysate was prepared by diluting the thawed cells with 2 vol of ice-cold Hz0 and was centrifuged at 18,000g for 15 min to remove insoluble material. Lysate was stored frozen at -20°C. The lysate was further diluted with 1.5 vol of cold H,O prior to use. PEG precipitation of lysate. Glycine buffer was added to the lysate (final concentration 25 mM), and the pH was adjusted with NaOH to 10.0 at 2°C. Ba(OAc),
AND
DEITS
(25 mM) was added, an aliquot was removed and centrifuged at 12,000g for 3 min, and the supernatant and pellet were assayed for bovine erythrocyte carbonic anhydrase activity, for hemoglobin content (estimated as absorbance at 410 nm), and for total protein. Successive aliquots of PEG 3350 were then added, the solution was allowed to stir at 2°C and the samples were removed for analysis. Studies of lysate at pH 5.9 used the same protocol except that Mes buffer was added (final concentration 50 mM), the pH was adjusted to 5.9 with HCl, and Ba(OAc), was omitted. Carbonic Anhydrase
Activity
Assay
Carbonic anhydrase was assayed using its esterase activity, with p-nitrophenyl acetate as substrate (2). Assays contained (final concentration) 50 m M Tris-acetate buffer, 0.1 m M p-nitrophenyl acetate, and 10% (VI v) acetone (added to solubilize p-nitrophenyl acetate). Activity was monitored as release of product p-nitrophenol at 370 nm. The X,, for p-nitrophenol is 400 nm, but working at 370 nm reduces interference from the Soret absorbance of hemoglobin. In order to correct for nonenzymatic ester hydrolysis and other esterase activities, assays were run in the presence and absence of 0.1 m M acetazolamide, a potent and specific inhibitor of carbonic anhydrase (2). The difference between the initial rates of the assays with and without acetazolamide was taken as the esterase activity of carbonic anhydrase. RESULTS
Effect of Divalent Cations on PEG Precipitation Bovine Serum Albumin at Alkaline pH
of
Atha and Ingham (3) reported that PEG precipitation was quite pH sensitive. They found that their test protein (human serum albumin, isoelectric point 4.9 (4)) was resistant to precipitation at alkaline pH, but that addition of 0.1 M ZnSO, facilitated protein precipitation (3). We repeated their experiment, using bovine serum albumin (isoelectric point 4.7 (4)) and found that 0.1 M ZnSO, substantially perturbed the pH of the buffered PEG solution, rendering it acidic, with concomitant precipitation of Zn(OH), and protein (data not shown). We concluded that a more alkali-soluble divalent cation was required to test the effect of divalent cations on PEG solubility of proteins. Ba2+ was the most suitable candidate, as it forms the most soluble hydroxide among the Group II elements (5). Figure 1 demonstrates that low concentrations of Ba*+ do indeed have a substantial effect on the ability of PEG 3350 to precipitate proteins even at pH 10. Neither PEG nor Ba2+ alone is able to precipitate bovine serum albumin under these conditions (Fig. 1).
DIVALENT
IONS
IN POLYETHYLENE
GLYCOL
PRECIPITATION
85
0.55,
.-
FIG. 1. Effect of added BaCl, on the precipitation of bovine serum albumin (BSA) by PEG 3350. Values are concentrations of BSA remaining in the supernatants of solutions initially 2.5 mg/ml in BSA. (0) 25 mM glycine buffer, pH 10. (ml 25 mM glycine buffer, pH 10, plus 25 mM BaCl,. (A) 25 mM Mes buffer, pH 5.5.
Optimal Ba2+ concentration. Millimolar concentrations of Ba2+ are sufficient for precipitation of bovine serum albumin at pH 10. The concentration required is sensitive to the percentage of PEG in the precipitation mixture (Fig. 2). Large excesses of Ba(Cl), slightly reduce the ability of PEG to precipitate bovine serum albumin (Fig. 2). Precipitation of a 0.5 mg/ml bovine serum albumin solution at pH 10.0 in 25 mM Ba(Cl), was complete within 30 s of mixing. In control experiments in which protein was omitted, no inorganic salts precipitated at concentrations of
0.6
0-i 0
20
40
60
60
100
120
140
160
I 180
[Barium Chloride] (mM) FIG. 2. Effects of varying PEG 3350 concentration and BaCl, concentration on bovine serum albumin (BSA) solubility. All solutions contain 25 mM glycine, pH 10. Values are concentrations of BSA remaining in the supernatants of solutions initially 0.5 mg/ml in BSA. (0) No PEG 3350 added. (m) 10% PEG 3350. (A) 20% PEG 3350. (0) 22.5% PEG 3350. (0) 27.5% PEG 3350.
FIG. 3. Effect of added NaCl on bovine serum albumin (BSA) solubility in the presence of BaCl, at pH 10. Initial concentration of BSA was 0.5 mg/ml. Precipitation was performed in 25 mM glycine, pH 10, 22.5% PEG 3350, and 25 mM BaCl,. Values are concentrations of BSA remaining in the supernatants of solutions initially 0.5 mg/ml in BSA.
Ba(Cl), up to 300 mM and PEG 3350 concentrations as high as 50%. Previously reported methods for alkaline precipitation of proteins rely upon coprecipitation of protein with high concentrations of insoluble inorganic salts (11). Optimal PEG molecular weight. We tested PEG 400 and PEG 35,000 as alternative precipitants of bovine serum albumin. PEG 400 was generally less effective on a weight/volume basis as a precipitant. PEG 35,000, while as efficient as PEG 3350, was too viscous for convenient handling. Specific salt effects. Mg2+, Ca’+, and Sr2+ efficiently precipitated bovine serum albumin when employed with PEG 3350 at pH values near 7.0 (data not shown). These cations form substantially less soluble hydroxides than Ba2+, and so are not suitable for use at alkaline pH. BaCl,, Ba(OAc),, and Ba(Cl,CO,) were equally effective at facilitating PEG 3350 precipitation of bovine serum albumin at alkaline pH, consistent with the hypothesis that the cationic moiety is the essential factor at alkaline pH (data not shown). Elevated concentrations of NaCl (ca. 0.15 M) inhibit precipitation (Fig. 3). This accounts, we believe, for the reduced efficiency of BaCl, when it is present in excess, as seen in Fig. 2. We conclude that divalent ions are required to obtain the effects observed. Mechanism Albumin
of Ba’+IPEG Precipitation at Alkaline pH
of Bovine Serum
Concentration dependence of Ba2’IPEG precipitation of bovine serum albumin. The above results demon-
86
THRASH,
-1.51 0
OTTO,
A 5
25
ii PEG &O
30
DEITS
Effect of divalent anions on the PEG 3350precipitation of lysozyme. The above data suggest that divalent cations act by rendering the target proteins effectively electroneutral under conditions where the net charge on the protein is negative (pH > isoelectric point). On this basis, we reasoned that divalent anions might facilitate PEG precipitation at pH values below the isoelectric point of the target protein. We tested SO:- for such an effect, using lysozyme (isoelectric point 11.0-11.35 (9)) as our test protein, and carried out precipitation at pH 4. We observed significant improvement in the precipitation of lysozyme at pH 4 when SOf was added (Fig. 5). Addition of BaCl, at pH 10 had no effect on PEG precipitation of lysozyme (Fig. 5), as expected.
(w/z”,
FIG. 4. Dependence of bovine serum albumin (BSA) solubility on PEG 3350 concentration. (U) 50 mM glycine, pH 10.0, and 25 mM BaCl,. (A) 50 mM citrate, pH 4.7. Solid line: Linear least-squares fit of pH 10 data points above 10% PEG, slope = -0.182 + 0.15; ordinate intercept = 3.56 f 0.19. Dotted line: Linear least-squares fit of pH 4.7 data points above 10% PEG, slope = -0.174 + 0.004; ordinate intercept = 3.56 f 0.059. Plateau at values below 10% PEG 3350 is due to insufficient protein added to achieve saturation under the reaction conditions.
strate that divalent cations can dramatically alter the solubility of bovine serum albumin in alkaline PEG solutions. Our next concern was to determine the mechanism by which PEG acts under these conditions. Atha and Ingham (3) had previously demonstrated that many proteins share a common log-linear dependence of solubility on PEG concentration, consistent with a common mechanism of action (1). We examined the solubility of bovine serum albumin over a wide concentration range at pH 4.7, the isoelectric point of bovine serum albumin, and at pH 10.0 in the presence of 25 m M BaCl, (Fig. 4). In both cases, bovine serum albumin solubility exhibits log-linear behavior. The solubility equation governing this log-linear behavior is (3) 1ogs=pc+s,,
AND
Representative
Applications
of Ba’+IPEG
Precipitation
Precipitation of yeast alcohol dehydrogenase at pH 7.0 with retention of activity. Yeast alcohol dehydrogenase (isoelectric point 5.4 (8)) can be efficiently precipitated by the Ba2+/PEG system at pH 7; in the absence of Ba2+, no precipitation is observed (Fig. 6). We were able to recover more than 90% of the enzyme activity from the pellet precipitated under these conditions (Fig. 6). This shows that the Ba2+/PEG precipitation method retains the ability of PEG to precipitate proteins in their native conformation. Removal of protein from solutions containing metabolites. Neither AMP, NADH, nor ATP was coprecipitated when present during the precipitation of high concentrations of bovine serum albumin at alkaline pH in the presence of Ba2+ (Table 1). The recovery of ATP was not high, but the data show that ATP was precipitated in the presence of Ba2+ and PEG alone and that its
111
where S represents the solubility of bovine serum albumin, C is the weight/volume percentage concentration of PEG, 6 is the slope of the best fit line to the S versus C curve, and S, is the ordinate intercept of the best fit line. S, can be interpreted as a reasonable approximation to the solubility of the test protein in the absence of PEG (3). When the data of Fig. 4 are fitted to this equation, the resulting values are /3 = -0.18, S, = 3.57 at pH 4.7 and p = -0.17, S, = 3.56 at pH 10 in the presence of BaCl,. These results are essentially identical and are comparable to the values previously reported for precipitation of human serum albumin at pH 5 (p = -0.21, S, = 3.67) (3).
0
0
IO
50
; PEG-&
(w,;
FIG. 6. Effect of added divalent ions on the solubility of lysozyme. Values are concentrations of lysozyme remaining in the supernatants of solutions initially 6 mg/ml in lysozyme. (0) 25 mM glycine, pH 10. (m) 25 mM glycine, pH 10, and 25 mM BaCl,. (A) 25 mM acetate buffer, pH 4.0. (0) 25 mM acetate buffer, pH 4.0, and 25 mM Na,SO,.
DIVALENT
0
0
4
2
PEG-&O
IONS
(w/$
IN POLYETHYLENE
20
24
GLYCOL
0.2-1 0
87
PRECIPITATION
5
IO
15
20
25
30
35
4( 1
% PEG-3350 (w/v) FIG. 6. Effect of added BaCl, on the recovery of purified yeast alcohol dehydrogenase from PEG precipitation mixtures at pH 7. Precipitation mixtures contain 25 mM Hepes buffer, pH 7.0, with 26 mM BaCl,. (0) Enzyme activity recovered in supernatant. (0) Enzyme activity recovered in pellet. (m) Protein recovered in supernatant. (0) Protein recovered in pellet. (A) Protein recovered in supernatant when BaCl, is omitted.
precipitation was not further enhanced by the concomitant precipitation of bovine serum albumin. Fractional precipitation of proteins from crude extracts with Ba’+IPEG: (a) Precipitation of purified carbonic anhydrase. Carbonic anhydrase (isoelectric point 5.9 (7)) was chosen for study using alkaline Ba2+/PEG precipitation because of its stability at high pH; it can withstand incubation at pH 12 for hours at room temperature without loss of activity (6). Carbonic anhydrase was TABLE Recovery
(X)
1
of Small Molecules from Precipitation Mixtures
Compound tested
PEG 4000 (25%) + Ba*+ (25 mM)
AMP ATP NADH
98 23 102
PEG 4000 (25%) + Ba2+ (25 mM) + BSA (1.8 mg/ml) 100 23 101
PEG/BA2+
Bar+ (25 mM) only nd. 88 n.d.
Note. Solutions containing 25 mM glycine buffer (final concentration), pH 10.0, with the indicated additions, except for bovine serum albumin (BSA) and the test compound, were prepared. The mixture was allowed to stand on ice for 10 min and then a solution containing both BSA and the test compound was added. After an additional 10 min on ice, the mixture was centrifuged at 12,OOOg for 3 min. Values are percentage recovery in the supernatant of the indicated mixtures, compared to controls lacking PEG and BA*+. Concentrations were determined spectrophotometrically, using 260 nm for AMP and ATP, and 340 nm for NADH. Initial concentrations of the test compounds were [AMP] = 0.56 mM, [ATP] = 0.36 mM, [NADH] = 0.20 mM. Values are averages of four determinations, with a typical variance of 5%. n.d., not determined.
FIG. 7. Effect of added BaCl, on the precipitation of bovine carbonic anhydrase (BCA) by PEG 3350. All solutions contain 25 mM glycine, pH 10. Values are concentrations of BCA remaining in the supernatants of solutions initially 2.0 mg/ml in BCA. (0) No addition. (0) 25 mM BaCl, added. (A) 25 mM Hepes buffer, pH 7.0 (no BaCl,). (Cl) BCA was denatured by treatment at pH 13 for 3 h at room temperature (6), then precipitated with no additions. (m) BCA, treated at pH 13 as above, precipitated in the presence of 25 mM BaCl,.
resistant to precipitation by Ba2+/PEG at pH 10 (Fig. 7). Carbonic anhydrase could be precipitated only by first denaturing the protein by incubation at pH 13 for 3 h (6) and then carrying out Ba2+/PEG precipitation (Fig. 7). In contrast, carbonic anhydrase exhibits the expected log-linear precipitation curve seen for bovine serum albumin and for other proteins (3) when precipitated at its isoelectric point value. When fitted to Eq. [l], carbonic anhydrase gives values of /3 = -0.099 and S, = 2.35 for PEG 3350 precipitation at pH 5.9 (data not shown). (b) Fractionation of erythrocyte lysate. The resistance of carbonic anhydrase to precipitation by Ba2+/ PEG at high pH (Fig. 7) suggested a protocol for carbonic anhydrase purification by Ba2+/PEG precipitation at alkaline pH. If, as we hypothesized, most proteins were susceptible to Ba2+/PEG precipitation at high pH, then we should be able to effect substantial purification of carbonic anhydrase by precipitating away most other erythrocyte proteins. When bovine erythrocyte lysate was subjected to alkaline Ba2+/PEG precipitation, 90% of the hemoglobin and 88% of the total protein present precipitated, while 76% of the carbonic anhydrase activity was retained in the supernatant (Fig. 8A). This represents a 6.3-fold purification of carbonic anhydrase and confirms that most proteins in the erythrocyte are efficiently precipitated by this method. This result can be compared with a similar fractionation experiment carried out at a pH near the isoelectric point of carbonic anhydrase (Fig. 8B). Acceptable frac-
88
THRASH.
t 8
OTTO.
AND
‘E
60
is a’
t II
DEITS
60.
-
60.
40 40.
% PEG-3350
(w/v)
% PEG-3350
(w/v)
FIG. 8. Precipitation of bovine carbonic anhydrase, hemoglobin, and total protein from erythrocyte lysate. Lysate preparation and PEG precipitation are described under Materials and Methods. (A) pH 10.0. (B) pH 5.9. (0) Carbonic anhydrase activity recovered in supernatant. (0) Carbonic anhydrase activity recovered in pellet. (m) Hemoglobin recovered in supernatant. (0) Hemoglobin recovered in pellet. (A) Total protein recovered in supernatant. (A) Total protein recovered in pellet.
tionation of carbonic anhydrase can be achieved over a narrow range of PEG concentrations (between 25 and 30% PEG). However, the degree of purification is not as high (ca. 2- to 3-fold). In addition, no more than ca. 65% of erythrocyte proteins can be precipitated at pH 5.9 with PEG 3350 alone. DISCUSSION
These data demonstrate that the solubility of proteins in PEG solutions can be significantly altered by relatively low concentrations (5-50 mM) of divalent cations or anions, chosen so as to counteract the net charge on the protein of interest. The effect can be dramatic; a protein can be precipitated at a pH 6-7 pH units away from its isoelectric point with unimpaired efficiency. This effect is seen with a variety of proteins with a wide range of isoelectric point values; bovine serum albumin (isoelectric point 4.7 (4)), yeast alcohol dehydrogenase (isoelectric point 5.4 (S)), hemoglobin (isoelectric point ca. 7.4), lysozyme (isoelectric point ca. 11.2 (9)), and the majority of erythrocyte proteins exhibit this effect. Only carbonic anhydrase (isoelectric point 5.9 (7)) was found to be resistant to Ba2+-facilitated precipitation at alkaline pH. It is not necessary to work at extremes of pH to employ this method. For example, yeast alcohol dehydrogenase can be efficiently precipitated at pH 7.0 only when Ba2+ is present (Fig. 6); only the difference between solution pH and protein isoelectric point is relevant. Mechanism
of Action
The effect of divalent ions is most simply explained by a mechanism in which the chosen counterion is able to shield net charge on the protein molecule, either by
interacting with negative charges on a single protein molecule or by bridging ionic sites between proteins. This is essentially equivalent to reducing net charge on the protein by adjusting the solution pH to near the isoelectric point value of the protein. This explanation is supported by the data of Fig. 3 which show that bovine serum albumin precipitation at pH 4.7 and at pH 10 in the presence of Ba2+ have identical log-linear solubility behavior. The ability of divalent anions to facilitate precipitation of lysozyme at a pH value well below its isoelectric point (Fig. 5) also supports this mechanism. When mixtures containing proteins of differing net charge are subjected to PEG precipitation, electrostatic interactions between the two proteins can cause coprecipitation. This phenomenon, termed heteroassociation, can sometimes be a useful adjunct to conventional precipitation (lo), but can also complicate precipitation of protein mixtures. If charge quenching is largely responsible for the effect of divalent ions, a protein mixture brought to alkaline pH and treated with Ba2+ will consist essentially entirely of proteins which are either electroneutral or negatively charged. This circumstance should effectively eliminate heteroassociation. Figure 8 bears this out, as even when high concentrations of hemoglobin and other erythrocyte proteins are precipitated from crude lysate, carbonic anhydrase is not brought down by coprecipitation. The effect of divalent ions on PEG solubility might conceivably be due to an effect on the solvation properties of PEG itself, rather than on the target proteins. The parameter S, represents the apparent intrinsic solubility of the target protein in the absence of PEG (3). If divalent ions were affecting PEG alone, we would expect the apparent intrinsic solubility of bovine serum albumin to be considerably different at its isoelectric
DIVALENT
IONS
IN POLYETHYLENE
point and at a pH value 5 units higher. This explanation is, however, not supported by the data of Fig. 3; S, is identical for bovine serum albumin at pH 4.7 and pH 10 in the presence of BaCl,. Divalent ions may be more effective than singly charged ions (which tend to render proteins more soluble in PEG solutions (Fig. 4)) because they can bind more tightly due to their higher net charge, or because nearby charged groups can mutually interact with a divalent ion, enhancing binding by a proximity effect. These possibilities are not exclusive, and a distinction between them is beyond the scope of the present data. A converse consequence of these observations should be borne in mind. Relatively low concentrations of divalent ions, either endogenous or added exogenously as with a phosphate buffer, may significantly perturb the pattern of PEG fractionation. In general, minimizing divalent ion concentration should maximize differences in protein solubility and enhance fractionation, unless a special feature of a given protein, such as carbonic anhydrase above, is being exploited.
Applications Our observations demonstrate that manipulation of divalent ion concentration can be a useful method for modulating protein solubility in PEG solutions over the full range of pH values employed in protein purification. This method retains the speed, capacity, and retention of native state of PEG precipitation. Other reagents for protein precipitation, including PEG without divalent ions, NH,SO,, and organic solvents, are ineffective at high pH. The only previously reported method of protein precipitation at alkaline pH relies on high concentrations of inorganic salts which can cause substantial coprecipitation by adsorption on the resulting insoluble inorganic precipitate (11). In the present method, no insoluble products are formed other than the protein itself; interference from bulk coprecipitation and contamination of precipitated proteins is eliminated. This method should prove valuable for protein purification by fractional precipitation, isolation of proteins
GLYCOL
PRECIPITATION
89
containing labile cofactors or modified amino acids, deproteinization of solutions containing pH-sensitive metabolites, and alteration of solubility for the purpose of PEG-induced enzyme crystallization (12). ACKNOWLEDGMENTS This research was supported by United States Department of Agriculture, Competitive Research Grants Office, Award 89-37120-4757, and by the Michigan State University Department of Biochemistry Summer Undergraduate Research Program.
REFERENCES of proteins with polyethylene 1. Ingham, K. C. (1990) Precipitation glycol, in. “Methods in Enzymology” 182, pp. 301-306, Academic Press, San Diego. 2. Packer, Y., and Sarkanen, S. (1978) Carbonic anhydrase: Structure, catalytic versatility, and inhibition. Adv. Enzymol. 47,149273. 3. Atha, D. H., and Ingham, K. C. (1981) Mechanism of precipitation of proteins by polyethylene glycols: Analysis in terms of excluded volume. J. Biol. Chem. 256, 12,10812,117. of Biochemistry and Mo4. Fasman, G. D. (Ed.) (1976) “Handbook lecular Biology,” 3rd ed., p. 174, CRC Press, Boca Raton, FL. 5. Weast, R. C. (Ed.) (1979) “Handbook of Chemistry and Physics,” 60th ed, pp. B-51-B-144, CRC Press, Boca Raton, FL. 6. Nilsson, A., and Lindskog, S. (1967) Hydrogen ion equilibria and the chemical modification of lysine and tyrosine residues in bovine carbonic anhydrase B. Eur. J. Biochem. 2,309-317. 7. Lindskog, S., Henderson, L. E., Kannan, K. K., Liljas, A., Nyman, P. O., and Strandberg, B. (1971) Carbonic anhydrase, in “The Enzymes” (Boyer, P. D., Ed.), 3rd ed., Vol. 5, p. 600, Academic Press, New York. in 8. Sund, H., and Theorell, H. (1963) Alcohol dehydrogenases, “The Enzymes” (Boyer, P. D., Lardy, H., and Myrback, K., Eds.) 2nd ed., Vol. 8, p. 58, Academic Press, New York. 9. Imoto, T., Johnson, L. N., Phillips, D. C., and Ripley, J. A. (1971) Lysozyme, in “The Enzymes” (Boyer, P. D., Ed.), 3rd ed., Vol. 7, p. 732, Academic Press, New York. 10. Miekka, S. I., and Ingham, K. C. (1980) Influence of hetero-association on the precipitation of proteins by poly(ethylene glycol). Arch. Biochm. Biophys. 203,630-641. 11. Bergmeyer, H. U., Bernt, E., Gawehn, K., and Michal, G. (1974) Handling of biochemical reagents and samples, in “Methods of Enzymatic Analysis,” 2nd English ed., Vol. 1, p. 178 Verlag Chemie, Weinhein-Bergstra, Berlin. 12. Ollis, D., and White, S. (1990) Protein crystallization, in “Methods in Enzymology” (Deutscher, M. P., Ed.), Vol. 182, pp. 646659, Academic Press, San Diego.
EXPRESSION
AND
PURIFICATION
2, 83-89 (19%)
Effect of Divalent Ions on Protein Precipitation Polyethylene G lycol: Mechanism of Action and Applications
with
Sonja L. Thrash, James C. O tto, Jr., and Thomas L. Deits Department
of Biochemistry,
Received February
15,1991,
Michigan
State University,
East Lansing,
Michigan
48824
and in revised form April 15, 1991
Polyethylene glycol (PEG) is extensively employed for protein purification by fractional precipitation. Efficiency of precipitation is highest when the solution pH is near the isoelectric point of the target protein. At pH values far from the isoelectric point of the target protein, proteins develop a net positive or negative charge and are more resistant to precipitation. W e have found that divalent cations (Ba2+, Sr2+, and Ca’+) or divalent anions (SO:-) significantly change the pattern of PEG precipitation when the ion is chosen so as to counteract the expected net charge on the target protein. At moderate (S-50 mM) concentrations of Ba2+, negatively charged proteins can be precipitated from solution at pH values as high as 10 with efficiency unchanged from precipitation at pH values near their isoelectric point values. The mechanism of PEG precipitation of protein at these high pH values appears to be unchanged from the mechanism operative at the protein isoelectric point. Precipitation is rapid and the capacity for protein precipitation is high. There is no detectable coprecipitation of small molecules (AMP, ATP, and NADH) or soluble proteins (carbonic anhydrase) induced when large quantities of protein are precipitated by this method. The purification of bovine carbonic anhydrase from erythrocyte lysate is more efficient at pH 10 in the presence of Ba2+ than is conventional PEG precipitation carried out at the isoelectric point of carbonic anhydrase. Application of these observations should broaden the utility of protein purification by fractional precipitation with PEG. o 1991 Academic POW, IIIC.
Polyethylene glycol (PEG)’ has proven to be a valuable reagent for the fractional precipitation of proteins
from solution with minimal perturbation of native structure (1). The efficiency of precipitation is sensitive to the ionic state of the protein; precipitation is usually most effective near the isoelectric point of the chosen protein, where protein net charge is minimal. Consequently, if protein or cofactor stability dictates that the protein be maintained at a pH distant from the isoelectric point value, PEG precipitation will usually be much less effective. A quantitative molecular explanation for the solubility of proteins in PEG has not yet been established; thus application of PEG precipitation remains somewhat empirical. W e have found that the addition of millimolar levels of divalent ions to a PEG precipitation mixture frequently renders PEG precipitation far less sensitive to the isoelectric point of the protein of interest, as long as the ion is chosen so as to complement the net charge on the protein at the pH chosen. Effective precipitation can be accomplished at pH values from 4 to 13 with the appropriate choice of ion. This observation extends previous studies which have used pH to modulate protein net charge (3) and emphasizes the dominant role played by protein net charge in PEG solubility. This observation in turn provides a unique means of protein precipitation at alkaline pH in which all reagents remain in solution as protein is precipitated. Applications should include protein fractionation, routine microscale deproteinization, and recovery of cofactors or metabolites with special pH sensitivities. W e demonstrate selected examples of these applications.
MATERIALS
AND
METHODS
Reagents ’ Abbreviations used: PEG, polyethylene glycol; PEG 3350, polyethylene glycol, average molecular weight 3350 (formerly referred to as PEG 4000 by the supplier (Sigma Chemical Co.); OAc, acetate ion; Hepes, 4-(2-hydroxyethyl)-1-piperazine-ethanesulfonic acid); Mes, 4-morpholine-ethanesulfonic acid. 1046-5928/91
$3.00
Copyright 0 1991 by Academic Press, Inc. All rights of reproduction in any form reserved.
Proteins and reagents were purchased from Sigma Chemical Co. unless otherwise indicated. PEG 35,000 was obtained from Fluka. The Bradford protein reagent was purchased from Bio-Rad. 83
84 Protein
THRASH,
OTTO.
Determination
Protein was determined by the Bradford method using the manufacturer’s directions (Bio-Rad) for the microassay (1 ml final volume). In control experiments, we found that PEG at concentrations above 2-3% (w/v) can interfere slightly in this assay, so all readings were compared against protein-free controls containing the same PEG concentration as the sample. PEG Precipitation Unless otherwise noted, well-mixed solutions containing all components except protein were prepared in l&ml microcentrifuge tubes and chilled on ice. Protein was added and the solutions were thoroughly mixed on a vortex mixer and allowed to stand on ice for 5 min. The precipitate was pelleted by centrifugation in a refrigerated benchtop microcentrifuge at ca. 12,000g for 3 min. In each experiment, the pH of the supernatant was checked to confirm that the indicated additions had not significantly (more than 0.1 pH unit) altered the solution pH. A 50% aqueous solution (grams PEG/100 ml solution) of PEG 3350 was routinely employed as a stock solution that could be conveniently pipetted. For those experiments where higher PEG concentrations were required, PEG was weighed directly into microcentrifuge tubes and a measured volume of water added. A calibration curve was generated to calculate the weight/volume percentage of these samples. This curve was generated by mixing known weights of PEG with known volumes of water and determining the resulting volume of the mixture. Alcohol Dehydrogenase
Actiuity
Assay
Alcohol dehydrogenase activity was assayed using (final concentration) 0.1 M Tris-Cl, pH 8.8, 5 mM MgCl,, 0.5 m M NADH, and 20 m M ethanol. Activity was monitored as the initial rate of formation of NADH at 340 nm. Bovine Erythrocyte
Lysate
Lysate preparation. Bovine blood was kindly provided by Professor R. Fogwell, Michigan State University. Blood was collected into a citrate anticoagulation buffer. Erythrocytes were washed three times with 0.9% NaCl using low-speed centrifugation. Cells were frozen at -20°C. Lysate was prepared by diluting the thawed cells with 2 vol of ice-cold Hz0 and was centrifuged at 18,000g for 15 min to remove insoluble material. Lysate was stored frozen at -20°C. The lysate was further diluted with 1.5 vol of cold H,O prior to use. PEG precipitation of lysate. Glycine buffer was added to the lysate (final concentration 25 mM), and the pH was adjusted with NaOH to 10.0 at 2°C. Ba(OAc),
AND
DEITS
(25 mM) was added, an aliquot was removed and centrifuged at 12,000g for 3 min, and the supernatant and pellet were assayed for bovine erythrocyte carbonic anhydrase activity, for hemoglobin content (estimated as absorbance at 410 nm), and for total protein. Successive aliquots of PEG 3350 were then added, the solution was allowed to stir at 2°C and the samples were removed for analysis. Studies of lysate at pH 5.9 used the same protocol except that Mes buffer was added (final concentration 50 mM), the pH was adjusted to 5.9 with HCl, and Ba(OAc), was omitted. Carbonic Anhydrase
Activity
Assay
Carbonic anhydrase was assayed using its esterase activity, with p-nitrophenyl acetate as substrate (2). Assays contained (final concentration) 50 m M Tris-acetate buffer, 0.1 m M p-nitrophenyl acetate, and 10% (VI v) acetone (added to solubilize p-nitrophenyl acetate). Activity was monitored as release of product p-nitrophenol at 370 nm. The X,, for p-nitrophenol is 400 nm, but working at 370 nm reduces interference from the Soret absorbance of hemoglobin. In order to correct for nonenzymatic ester hydrolysis and other esterase activities, assays were run in the presence and absence of 0.1 m M acetazolamide, a potent and specific inhibitor of carbonic anhydrase (2). The difference between the initial rates of the assays with and without acetazolamide was taken as the esterase activity of carbonic anhydrase. RESULTS
Effect of Divalent Cations on PEG Precipitation Bovine Serum Albumin at Alkaline pH
of
Atha and Ingham (3) reported that PEG precipitation was quite pH sensitive. They found that their test protein (human serum albumin, isoelectric point 4.9 (4)) was resistant to precipitation at alkaline pH, but that addition of 0.1 M ZnSO, facilitated protein precipitation (3). We repeated their experiment, using bovine serum albumin (isoelectric point 4.7 (4)) and found that 0.1 M ZnSO, substantially perturbed the pH of the buffered PEG solution, rendering it acidic, with concomitant precipitation of Zn(OH), and protein (data not shown). We concluded that a more alkali-soluble divalent cation was required to test the effect of divalent cations on PEG solubility of proteins. Ba2+ was the most suitable candidate, as it forms the most soluble hydroxide among the Group II elements (5). Figure 1 demonstrates that low concentrations of Ba*+ do indeed have a substantial effect on the ability of PEG 3350 to precipitate proteins even at pH 10. Neither PEG nor Ba2+ alone is able to precipitate bovine serum albumin under these conditions (Fig. 1).
DIVALENT
IONS
IN POLYETHYLENE
GLYCOL
PRECIPITATION
85
0.55,
.-
FIG. 1. Effect of added BaCl, on the precipitation of bovine serum albumin (BSA) by PEG 3350. Values are concentrations of BSA remaining in the supernatants of solutions initially 2.5 mg/ml in BSA. (0) 25 mM glycine buffer, pH 10. (ml 25 mM glycine buffer, pH 10, plus 25 mM BaCl,. (A) 25 mM Mes buffer, pH 5.5.
Optimal Ba2+ concentration. Millimolar concentrations of Ba2+ are sufficient for precipitation of bovine serum albumin at pH 10. The concentration required is sensitive to the percentage of PEG in the precipitation mixture (Fig. 2). Large excesses of Ba(Cl), slightly reduce the ability of PEG to precipitate bovine serum albumin (Fig. 2). Precipitation of a 0.5 mg/ml bovine serum albumin solution at pH 10.0 in 25 mM Ba(Cl), was complete within 30 s of mixing. In control experiments in which protein was omitted, no inorganic salts precipitated at concentrations of
0.6
0-i 0
20
40
60
60
100
120
140
160
I 180
[Barium Chloride] (mM) FIG. 2. Effects of varying PEG 3350 concentration and BaCl, concentration on bovine serum albumin (BSA) solubility. All solutions contain 25 mM glycine, pH 10. Values are concentrations of BSA remaining in the supernatants of solutions initially 0.5 mg/ml in BSA. (0) No PEG 3350 added. (m) 10% PEG 3350. (A) 20% PEG 3350. (0) 22.5% PEG 3350. (0) 27.5% PEG 3350.
FIG. 3. Effect of added NaCl on bovine serum albumin (BSA) solubility in the presence of BaCl, at pH 10. Initial concentration of BSA was 0.5 mg/ml. Precipitation was performed in 25 mM glycine, pH 10, 22.5% PEG 3350, and 25 mM BaCl,. Values are concentrations of BSA remaining in the supernatants of solutions initially 0.5 mg/ml in BSA.
Ba(Cl), up to 300 mM and PEG 3350 concentrations as high as 50%. Previously reported methods for alkaline precipitation of proteins rely upon coprecipitation of protein with high concentrations of insoluble inorganic salts (11). Optimal PEG molecular weight. We tested PEG 400 and PEG 35,000 as alternative precipitants of bovine serum albumin. PEG 400 was generally less effective on a weight/volume basis as a precipitant. PEG 35,000, while as efficient as PEG 3350, was too viscous for convenient handling. Specific salt effects. Mg2+, Ca’+, and Sr2+ efficiently precipitated bovine serum albumin when employed with PEG 3350 at pH values near 7.0 (data not shown). These cations form substantially less soluble hydroxides than Ba2+, and so are not suitable for use at alkaline pH. BaCl,, Ba(OAc),, and Ba(Cl,CO,) were equally effective at facilitating PEG 3350 precipitation of bovine serum albumin at alkaline pH, consistent with the hypothesis that the cationic moiety is the essential factor at alkaline pH (data not shown). Elevated concentrations of NaCl (ca. 0.15 M) inhibit precipitation (Fig. 3). This accounts, we believe, for the reduced efficiency of BaCl, when it is present in excess, as seen in Fig. 2. We conclude that divalent ions are required to obtain the effects observed. Mechanism Albumin
of Ba’+IPEG Precipitation at Alkaline pH
of Bovine Serum
Concentration dependence of Ba2’IPEG precipitation of bovine serum albumin. The above results demon-
86
THRASH,
-1.51 0
OTTO,
A 5
25
ii PEG &O
30
DEITS
Effect of divalent anions on the PEG 3350precipitation of lysozyme. The above data suggest that divalent cations act by rendering the target proteins effectively electroneutral under conditions where the net charge on the protein is negative (pH > isoelectric point). On this basis, we reasoned that divalent anions might facilitate PEG precipitation at pH values below the isoelectric point of the target protein. We tested SO:- for such an effect, using lysozyme (isoelectric point 11.0-11.35 (9)) as our test protein, and carried out precipitation at pH 4. We observed significant improvement in the precipitation of lysozyme at pH 4 when SOf was added (Fig. 5). Addition of BaCl, at pH 10 had no effect on PEG precipitation of lysozyme (Fig. 5), as expected.
(w/z”,
FIG. 4. Dependence of bovine serum albumin (BSA) solubility on PEG 3350 concentration. (U) 50 mM glycine, pH 10.0, and 25 mM BaCl,. (A) 50 mM citrate, pH 4.7. Solid line: Linear least-squares fit of pH 10 data points above 10% PEG, slope = -0.182 + 0.15; ordinate intercept = 3.56 f 0.19. Dotted line: Linear least-squares fit of pH 4.7 data points above 10% PEG, slope = -0.174 + 0.004; ordinate intercept = 3.56 f 0.059. Plateau at values below 10% PEG 3350 is due to insufficient protein added to achieve saturation under the reaction conditions.
strate that divalent cations can dramatically alter the solubility of bovine serum albumin in alkaline PEG solutions. Our next concern was to determine the mechanism by which PEG acts under these conditions. Atha and Ingham (3) had previously demonstrated that many proteins share a common log-linear dependence of solubility on PEG concentration, consistent with a common mechanism of action (1). We examined the solubility of bovine serum albumin over a wide concentration range at pH 4.7, the isoelectric point of bovine serum albumin, and at pH 10.0 in the presence of 25 m M BaCl, (Fig. 4). In both cases, bovine serum albumin solubility exhibits log-linear behavior. The solubility equation governing this log-linear behavior is (3) 1ogs=pc+s,,
AND
Representative
Applications
of Ba’+IPEG
Precipitation
Precipitation of yeast alcohol dehydrogenase at pH 7.0 with retention of activity. Yeast alcohol dehydrogenase (isoelectric point 5.4 (8)) can be efficiently precipitated by the Ba2+/PEG system at pH 7; in the absence of Ba2+, no precipitation is observed (Fig. 6). We were able to recover more than 90% of the enzyme activity from the pellet precipitated under these conditions (Fig. 6). This shows that the Ba2+/PEG precipitation method retains the ability of PEG to precipitate proteins in their native conformation. Removal of protein from solutions containing metabolites. Neither AMP, NADH, nor ATP was coprecipitated when present during the precipitation of high concentrations of bovine serum albumin at alkaline pH in the presence of Ba2+ (Table 1). The recovery of ATP was not high, but the data show that ATP was precipitated in the presence of Ba2+ and PEG alone and that its
111
where S represents the solubility of bovine serum albumin, C is the weight/volume percentage concentration of PEG, 6 is the slope of the best fit line to the S versus C curve, and S, is the ordinate intercept of the best fit line. S, can be interpreted as a reasonable approximation to the solubility of the test protein in the absence of PEG (3). When the data of Fig. 4 are fitted to this equation, the resulting values are /3 = -0.18, S, = 3.57 at pH 4.7 and p = -0.17, S, = 3.56 at pH 10 in the presence of BaCl,. These results are essentially identical and are comparable to the values previously reported for precipitation of human serum albumin at pH 5 (p = -0.21, S, = 3.67) (3).
0
0
IO
50
; PEG-&
(w,;
FIG. 6. Effect of added divalent ions on the solubility of lysozyme. Values are concentrations of lysozyme remaining in the supernatants of solutions initially 6 mg/ml in lysozyme. (0) 25 mM glycine, pH 10. (m) 25 mM glycine, pH 10, and 25 mM BaCl,. (A) 25 mM acetate buffer, pH 4.0. (0) 25 mM acetate buffer, pH 4.0, and 25 mM Na,SO,.
DIVALENT
0
0
4
2
PEG-&O
IONS
(w/$
IN POLYETHYLENE
20
24
GLYCOL
0.2-1 0
87
PRECIPITATION
5
IO
15
20
25
30
35
4( 1
% PEG-3350 (w/v) FIG. 6. Effect of added BaCl, on the recovery of purified yeast alcohol dehydrogenase from PEG precipitation mixtures at pH 7. Precipitation mixtures contain 25 mM Hepes buffer, pH 7.0, with 26 mM BaCl,. (0) Enzyme activity recovered in supernatant. (0) Enzyme activity recovered in pellet. (m) Protein recovered in supernatant. (0) Protein recovered in pellet. (A) Protein recovered in supernatant when BaCl, is omitted.
precipitation was not further enhanced by the concomitant precipitation of bovine serum albumin. Fractional precipitation of proteins from crude extracts with Ba’+IPEG: (a) Precipitation of purified carbonic anhydrase. Carbonic anhydrase (isoelectric point 5.9 (7)) was chosen for study using alkaline Ba2+/PEG precipitation because of its stability at high pH; it can withstand incubation at pH 12 for hours at room temperature without loss of activity (6). Carbonic anhydrase was TABLE Recovery
(X)
1
of Small Molecules from Precipitation Mixtures
Compound tested
PEG 4000 (25%) + Ba*+ (25 mM)
AMP ATP NADH
98 23 102
PEG 4000 (25%) + Ba2+ (25 mM) + BSA (1.8 mg/ml) 100 23 101
PEG/BA2+
Bar+ (25 mM) only nd. 88 n.d.
Note. Solutions containing 25 mM glycine buffer (final concentration), pH 10.0, with the indicated additions, except for bovine serum albumin (BSA) and the test compound, were prepared. The mixture was allowed to stand on ice for 10 min and then a solution containing both BSA and the test compound was added. After an additional 10 min on ice, the mixture was centrifuged at 12,OOOg for 3 min. Values are percentage recovery in the supernatant of the indicated mixtures, compared to controls lacking PEG and BA*+. Concentrations were determined spectrophotometrically, using 260 nm for AMP and ATP, and 340 nm for NADH. Initial concentrations of the test compounds were [AMP] = 0.56 mM, [ATP] = 0.36 mM, [NADH] = 0.20 mM. Values are averages of four determinations, with a typical variance of 5%. n.d., not determined.
FIG. 7. Effect of added BaCl, on the precipitation of bovine carbonic anhydrase (BCA) by PEG 3350. All solutions contain 25 mM glycine, pH 10. Values are concentrations of BCA remaining in the supernatants of solutions initially 2.0 mg/ml in BCA. (0) No addition. (0) 25 mM BaCl, added. (A) 25 mM Hepes buffer, pH 7.0 (no BaCl,). (Cl) BCA was denatured by treatment at pH 13 for 3 h at room temperature (6), then precipitated with no additions. (m) BCA, treated at pH 13 as above, precipitated in the presence of 25 mM BaCl,.
resistant to precipitation by Ba2+/PEG at pH 10 (Fig. 7). Carbonic anhydrase could be precipitated only by first denaturing the protein by incubation at pH 13 for 3 h (6) and then carrying out Ba2+/PEG precipitation (Fig. 7). In contrast, carbonic anhydrase exhibits the expected log-linear precipitation curve seen for bovine serum albumin and for other proteins (3) when precipitated at its isoelectric point value. When fitted to Eq. [l], carbonic anhydrase gives values of /3 = -0.099 and S, = 2.35 for PEG 3350 precipitation at pH 5.9 (data not shown). (b) Fractionation of erythrocyte lysate. The resistance of carbonic anhydrase to precipitation by Ba2+/ PEG at high pH (Fig. 7) suggested a protocol for carbonic anhydrase purification by Ba2+/PEG precipitation at alkaline pH. If, as we hypothesized, most proteins were susceptible to Ba2+/PEG precipitation at high pH, then we should be able to effect substantial purification of carbonic anhydrase by precipitating away most other erythrocyte proteins. When bovine erythrocyte lysate was subjected to alkaline Ba2+/PEG precipitation, 90% of the hemoglobin and 88% of the total protein present precipitated, while 76% of the carbonic anhydrase activity was retained in the supernatant (Fig. 8A). This represents a 6.3-fold purification of carbonic anhydrase and confirms that most proteins in the erythrocyte are efficiently precipitated by this method. This result can be compared with a similar fractionation experiment carried out at a pH near the isoelectric point of carbonic anhydrase (Fig. 8B). Acceptable frac-
88
THRASH.
t 8
OTTO.
AND
‘E
60
is a’
t II
DEITS
60.
-
60.
40 40.
% PEG-3350
(w/v)
% PEG-3350
(w/v)
FIG. 8. Precipitation of bovine carbonic anhydrase, hemoglobin, and total protein from erythrocyte lysate. Lysate preparation and PEG precipitation are described under Materials and Methods. (A) pH 10.0. (B) pH 5.9. (0) Carbonic anhydrase activity recovered in supernatant. (0) Carbonic anhydrase activity recovered in pellet. (m) Hemoglobin recovered in supernatant. (0) Hemoglobin recovered in pellet. (A) Total protein recovered in supernatant. (A) Total protein recovered in pellet.
tionation of carbonic anhydrase can be achieved over a narrow range of PEG concentrations (between 25 and 30% PEG). However, the degree of purification is not as high (ca. 2- to 3-fold). In addition, no more than ca. 65% of erythrocyte proteins can be precipitated at pH 5.9 with PEG 3350 alone. DISCUSSION
These data demonstrate that the solubility of proteins in PEG solutions can be significantly altered by relatively low concentrations (5-50 mM) of divalent cations or anions, chosen so as to counteract the net charge on the protein of interest. The effect can be dramatic; a protein can be precipitated at a pH 6-7 pH units away from its isoelectric point with unimpaired efficiency. This effect is seen with a variety of proteins with a wide range of isoelectric point values; bovine serum albumin (isoelectric point 4.7 (4)), yeast alcohol dehydrogenase (isoelectric point 5.4 (S)), hemoglobin (isoelectric point ca. 7.4), lysozyme (isoelectric point ca. 11.2 (9)), and the majority of erythrocyte proteins exhibit this effect. Only carbonic anhydrase (isoelectric point 5.9 (7)) was found to be resistant to Ba2+-facilitated precipitation at alkaline pH. It is not necessary to work at extremes of pH to employ this method. For example, yeast alcohol dehydrogenase can be efficiently precipitated at pH 7.0 only when Ba2+ is present (Fig. 6); only the difference between solution pH and protein isoelectric point is relevant. Mechanism
of Action
The effect of divalent ions is most simply explained by a mechanism in which the chosen counterion is able to shield net charge on the protein molecule, either by
interacting with negative charges on a single protein molecule or by bridging ionic sites between proteins. This is essentially equivalent to reducing net charge on the protein by adjusting the solution pH to near the isoelectric point value of the protein. This explanation is supported by the data of Fig. 3 which show that bovine serum albumin precipitation at pH 4.7 and at pH 10 in the presence of Ba2+ have identical log-linear solubility behavior. The ability of divalent anions to facilitate precipitation of lysozyme at a pH value well below its isoelectric point (Fig. 5) also supports this mechanism. When mixtures containing proteins of differing net charge are subjected to PEG precipitation, electrostatic interactions between the two proteins can cause coprecipitation. This phenomenon, termed heteroassociation, can sometimes be a useful adjunct to conventional precipitation (lo), but can also complicate precipitation of protein mixtures. If charge quenching is largely responsible for the effect of divalent ions, a protein mixture brought to alkaline pH and treated with Ba2+ will consist essentially entirely of proteins which are either electroneutral or negatively charged. This circumstance should effectively eliminate heteroassociation. Figure 8 bears this out, as even when high concentrations of hemoglobin and other erythrocyte proteins are precipitated from crude lysate, carbonic anhydrase is not brought down by coprecipitation. The effect of divalent ions on PEG solubility might conceivably be due to an effect on the solvation properties of PEG itself, rather than on the target proteins. The parameter S, represents the apparent intrinsic solubility of the target protein in the absence of PEG (3). If divalent ions were affecting PEG alone, we would expect the apparent intrinsic solubility of bovine serum albumin to be considerably different at its isoelectric
DIVALENT
IONS
IN POLYETHYLENE
point and at a pH value 5 units higher. This explanation is, however, not supported by the data of Fig. 3; S, is identical for bovine serum albumin at pH 4.7 and pH 10 in the presence of BaCl,. Divalent ions may be more effective than singly charged ions (which tend to render proteins more soluble in PEG solutions (Fig. 4)) because they can bind more tightly due to their higher net charge, or because nearby charged groups can mutually interact with a divalent ion, enhancing binding by a proximity effect. These possibilities are not exclusive, and a distinction between them is beyond the scope of the present data. A converse consequence of these observations should be borne in mind. Relatively low concentrations of divalent ions, either endogenous or added exogenously as with a phosphate buffer, may significantly perturb the pattern of PEG fractionation. In general, minimizing divalent ion concentration should maximize differences in protein solubility and enhance fractionation, unless a special feature of a given protein, such as carbonic anhydrase above, is being exploited.
Applications Our observations demonstrate that manipulation of divalent ion concentration can be a useful method for modulating protein solubility in PEG solutions over the full range of pH values employed in protein purification. This method retains the speed, capacity, and retention of native state of PEG precipitation. Other reagents for protein precipitation, including PEG without divalent ions, NH,SO,, and organic solvents, are ineffective at high pH. The only previously reported method of protein precipitation at alkaline pH relies on high concentrations of inorganic salts which can cause substantial coprecipitation by adsorption on the resulting insoluble inorganic precipitate (11). In the present method, no insoluble products are formed other than the protein itself; interference from bulk coprecipitation and contamination of precipitated proteins is eliminated. This method should prove valuable for protein purification by fractional precipitation, isolation of proteins
GLYCOL
PRECIPITATION
89
containing labile cofactors or modified amino acids, deproteinization of solutions containing pH-sensitive metabolites, and alteration of solubility for the purpose of PEG-induced enzyme crystallization (12). ACKNOWLEDGMENTS This research was supported by United States Department of Agriculture, Competitive Research Grants Office, Award 89-37120-4757, and by the Michigan State University Department of Biochemistry Summer Undergraduate Research Program.
REFERENCES of proteins with polyethylene 1. Ingham, K. C. (1990) Precipitation glycol, in. “Methods in Enzymology” 182, pp. 301-306, Academic Press, San Diego. 2. Packer, Y., and Sarkanen, S. (1978) Carbonic anhydrase: Structure, catalytic versatility, and inhibition. Adv. Enzymol. 47,149273. 3. Atha, D. H., and Ingham, K. C. (1981) Mechanism of precipitation of proteins by polyethylene glycols: Analysis in terms of excluded volume. J. Biol. Chem. 256, 12,10812,117. of Biochemistry and Mo4. Fasman, G. D. (Ed.) (1976) “Handbook lecular Biology,” 3rd ed., p. 174, CRC Press, Boca Raton, FL. 5. Weast, R. C. (Ed.) (1979) “Handbook of Chemistry and Physics,” 60th ed, pp. B-51-B-144, CRC Press, Boca Raton, FL. 6. Nilsson, A., and Lindskog, S. (1967) Hydrogen ion equilibria and the chemical modification of lysine and tyrosine residues in bovine carbonic anhydrase B. Eur. J. Biochem. 2,309-317. 7. Lindskog, S., Henderson, L. E., Kannan, K. K., Liljas, A., Nyman, P. O., and Strandberg, B. (1971) Carbonic anhydrase, in “The Enzymes” (Boyer, P. D., Ed.), 3rd ed., Vol. 5, p. 600, Academic Press, New York. in 8. Sund, H., and Theorell, H. (1963) Alcohol dehydrogenases, “The Enzymes” (Boyer, P. D., Lardy, H., and Myrback, K., Eds.) 2nd ed., Vol. 8, p. 58, Academic Press, New York. 9. Imoto, T., Johnson, L. N., Phillips, D. C., and Ripley, J. A. (1971) Lysozyme, in “The Enzymes” (Boyer, P. D., Ed.), 3rd ed., Vol. 7, p. 732, Academic Press, New York. 10. Miekka, S. I., and Ingham, K. C. (1980) Influence of hetero-association on the precipitation of proteins by poly(ethylene glycol). Arch. Biochm. Biophys. 203,630-641. 11. Bergmeyer, H. U., Bernt, E., Gawehn, K., and Michal, G. (1974) Handling of biochemical reagents and samples, in “Methods of Enzymatic Analysis,” 2nd English ed., Vol. 1, p. 178 Verlag Chemie, Weinhein-Bergstra, Berlin. 12. Ollis, D., and White, S. (1990) Protein crystallization, in “Methods in Enzymology” (Deutscher, M. P., Ed.), Vol. 182, pp. 646659, Academic Press, San Diego.
