AU7241294A - High salt binding buffer for immunoglobulin purification with a synthetic affinity ligand - Google Patents

Description

HIGH SALT BINDING BUFFER FOR IMMUNOGLOBULIN PURIFICATION WITH A SYNTHETIC AFFINITY LIGAND

TECHNOLOGICAL FIELD

This invention relates to the affinity chromatographic purification of immunoglobulins. More particularly, the invention relates to a high salt buffer which increases the binding capacity of synthetic affinity chromatographic gels of the kind described in United States patent 4,582,875. BACKGROUND OF THE INVENTION

Recovery of immunoglobins present in cell culture supernatants is routinely accomplished by the use of affinity gels such as protein A, protein G and AVID-AI®, a synthetic gel available from UniSyn Technologies, 14272 Franklin Avenue, Tustin, CA. Such gels are used as a chromatographic matrix, as a particulate slurry and in other ways known to or within the skill of the art.

AVID-AL has been found to bind antibodies whose immunoglobins do not bind with high affinity to protein A or protein G. See,e.g., Ngo (1990); Narinesingh (1991) and Khatter (1991) . Moreover, AVID-AL has a lower mass than protein A or protein G with consequent greater resistance to acid, base, organic solvent, proteolytic enzyme and autoclaving treatment.

Various specialized and proprietary buffers are said to enhance the binding capacity of affinity gels. A need exists, however, for buffers effective to significantly increase the affinity with which immunoglobins bind to synthetic gels such as AVID-AL. SUMMARY OF THE INVENTION

Hydrophobic interaction impacts inter se repulsion which adversely impacts in two ways the affinity gel absorption of IgG molecules dissolved in an aqueous medium. First, aggregation of unbound molecules is constrained or inhibited. Second, bound molecules constrain or inhibit attraction by an affinity gel of as yet unbound molecules.

Pursuant to this invention, the efficiency of affinity gel immunoglobin purification procedures is mediated by the establishment within an aqueous medium of a dissolved salt concentration defined by two parameters. A first parameter defines a lower limit of said concentration effective to modify the hydrophobic interaction of immunoglobin molecules and thus reduce inter se repulsion. In general, any concentration of dissolved salt higher than that of conventional PBS is useful in the invention.

A second parameter defines the upper limit of salt concentration which must be lower than that which would result in indiscriminate precipitation of the desired immunoglobin and other proteins as a consequence of the known "salting out" effect.

Pursuant to this invention affinity gel purification yields high purity immunoglobins in good yield from contaminated aqueous media such as cell culture supernatants. The salt concentration of such media is adjusted to a value in excess of that conventional with PBS and below that which would result in the precipitation or salting out of the desired IgG molecules with consequent mediation of the hydrophobic interaction between the IgG molecules and reduction of inter se repulsion. When combined with the unique specificity of synthetic affinity to ligands, the unexpected high yields of immunoglobins of high purity are obtained from supernatants and comparable impure solutions of immunoglobins. DEFINITION OF PHYSIOLOGICAL SALINE SOLUTION (PBS)

As used herein, PBS means an aqueous solution 0.01 molar in KH2P04; 0.01 molar in Na2HP04 and 0.14 molar in NaCl having an ionic strength of 0.26. BRIEF DESCRIPTION OF THE FIGURES

Figure 1 depicts generalized chemistry for the production of affinity gels more fully described in U.S. patent 4,582,875.

Figure 2 depicts the effect of various salts on the binding of purified human IgG to AVID-AL and rProtein A gels. The binding capacity is shown as a percentage of the binding capacity obtained with KCl and assayed at a total buffer ionic strength of 1.28 M at pH 7.0.

Figure 3 depicts the relationship between the amount of IgG bound on AVID-AL gel and the concentration of purified human IgG in physiological saline solution (PBS) containing different concentrations of Na2Sθ₄ (the saturation curves) .

Figure 4(a) depicts the isolation of IgG from human serum by AVID-AL gel. PBS loading and washing conditions are indicated.

Figure 4(b) depicts the isolation of IgG from human serum by AVID-AL gel with PBS 0.75 molar of sodium sulfate. Loading and washing conditions are indicated. For each of the experiments illustrated by Figures 4(a) and 4(b), gel volume was 0.5 mL. Human serum was diluted 8.9-fold with PBS binding buffer per se or 0.75 M in Na2Sθ4 to provide an IgG concentration of about 2 mg/mL in solution and applied at the flow rate of 0.25 mL/min at room temperature. Fractions of 1.0 mL were collected. Elution was carried out with neutral elution buffer, pH 7.4. Figure 5 depicts the polyacrylamide gel electrophoresis (PAGE) sodium dodecyl sulfate (SDS) gradient (8-25%) in non-reducing conditions of fractions obtained from human serum in Figures 4(a) and 4(b). Lane 1) molecular weight marker reference proteins; 2) unfractionated human serum; 3) and 6) unbound, flow-through fractions; 4) and 7) unbound, after column wash; 5) and 8) fractions eluted with neutral elution buffer, pH 7.4. Lanes 3-5 correspond to Figure 4(a) , lanes 6-8 to Figure 4(b) .

Figure 6 depicts the SDS-PAGE electrophoretic analyses for estimating the purity of IgG from human serum as a function of different sodium sulfate concentration.

DETAILED DESCRIPTION OF THE INVENTION

Affinity gels useful in this invention are described and claimed in and by U.S. patent 4,582,875. Figure 1 schematically illustrates some of the chemistry relevant to the synthesis of these gels. As shown in Figure 1, X~ is a nucleophile, PFP is pentafluoropyridine, DMAP is

4-dimethylaminopyridine, and DMF is N,N'-dimethyl formamide.

The reaction of Sepharose C1-4B with PFP and DMAP in organic solvent yields an intermediate product upon which further reacts in aqueous solution with nucleophiles (ethylene glycol or glycine) to form gels with unique selectivity toward immunoglobulins and albumins from several animal species. The synthetic routes and possible structures of the ligand of immunoglobin-binding gels are shown in Figure 1. Elemental analysis of the gel gave an N:F ratio of 3:2 which is consistent with both structures. The ligand density of the gel was calculated to be 1-13 mmol per gram of dry gel. The actual structure of these gels is undefined. As claimed by patent 4,852,875, these gels are activated polymers capable of forming covalent leakages with a nucleophilic ligand.

AVID-AL (UniSyn) binds a wide range of immunoglobins under PBS conditions. The bound IgG can be eluted by a neutral elution buffer at physiological pH (pH 7.4).

MATERIALS

The AVID-AL^R (UniSyn) used in the following examples was a commercial product from manufacturing Lot #3-4-1. A commercial rProtein A gel was provided by RepliGen (Cambridge, MA) . Solutions of purified human IgG containing a concentration of 12.5 mg/mL were purchased from The Binding Site (San Diego, CA) . Human serum was obtained from Sigma (St. Louis, MO) . Cell culture media (RPMI-1640) and fetal bovine serum (FBS) were obtained from Mediatech (Herndon, VA) and Summit Biotechnology (Greeley, CO) , respectively. All other chemicals were of analytical grade.

EXAMPLE I

Human IgG Binding to AVID-AL Gel

Human IgG was bound to AVID-AL gel by placing 18.0 μL of purified human IgG in 972.0 μL PBS (0.02 M phosphate, 0.14 M sodium chloride (pH 7.0) with 10.0 μL of AVID-AL gel taken from a bulk of 20 mL which was washed first with regeneration buffer (20% methanol, 1% acetic acid, pH 3.4) and then with PBS for equilibration of the gel before using to bind IgG. The binding adsorption was run in a test tube overnight at 4°C with rotating on a Fisher hematology/chemistry mixer (Fisher Scientific, Pittsburgh, PA) . Unbound human IgG in solution, after binding was complete, was determined by using a radial immunodiffusion assay and a spectrophotometer at 750 nm with Dc protein assay (Bio-Rad, Hercules, CA) . The centrifuged supernatant of 5.0 μL was applied to a radial immunodiffusion plate (The Binding Site, San Diego, CA) following standard operating protocol. The binding capacity of AVID-AL gel was determined by calculation of the corresponding amount of IgG for mass balance. The same was done with a rProtein A gel as a reference under the same conditions as used to determine the binding capacity of AVID-AL gel.

Chemical Dependence of Human IgG Binding

Seven PBS solutions, one for each of seven inorganic salts—NaCl, Na2Sθ4, (NH4)2S04, MgCl2, ammonium acetate, NH4CI and KCl were prepared. Solution concentrations of 0.25 M, 0.50 M, and 0.75 M in each salt were prepared by diluting a 1.0 M solution. Since MgCl2 has a solubility of less than 1.0 M, concentrations of 0.25 M, 0.50 M, and 0.75 M were used in the experiment. All of the solutions had a pH value of 7.0.

Column Configuration

Approximately 0.5 mL of AVID-AL gel was packed into a 1.0 mL column. A Rainin peristaltic pump with an eight-channel head (Rainin, Woburn, MA) was used for buffer delivery. The column was washed with regeneration buffer until a baseline was obtained according to the absorbance, A₂₈₀nm given by LKB UV detector. The column was then pre-equilibrated with PBS. A sample of 1.35 mL of prefiltered human serum dissolved in 10.35 mL of binding buffer was pumped through the column with a flow rate of 0.25 mL/min, and 1.0 mL fractions were collected from the start of the loading step. The amount of fed human IgG was about 22 mg. Bound IgG was eluated using neutral elution buffer (composition 20% glycerol and 1.0 M Tris.HCl in DI water, pH adjusted to 7.4±0.2 by 6N NaOH) provided by UniSyn Technologies (Tustin, CA) with a flow rate of 1.0 mL/min. Protein output was measured and the absorbance of each fraction at 280 ran was used to determine its IgG concentration. Between runs the column was washed with regeneration buffer and PBS with minimal to no protein elution seen. Electrophoretic analyses were performed on 8-25% polyacrylamide gradient gels by using the Phast system from Pharmacia. The amount of purified IgG in fractions was confirmed using radial immunodiffusion. Effect in IgG Purification With AVID-AL The effect of different types of salts on the binding capacity of AVID-AL gel is summarized in Figure 2. By adjusting the salt concentration to the same ionic strength in the assay (1.28 M) , sodium sulfate, ammonium sulfate, ammonium acetate, ammonium chloride,and sodium chloride gave an increase of the binding ability up to 44%, 38%, 33%, 30% and 13%, respectively, as compared to potassium chloride. The same ionic strength (1.28 M) of buffer containing MgCl2 showed a decreased binding capacity corresponding to 91.2% as compared to the same ionic strength of KCl. From the data, the salt effect on the binding capacity with respect to the various anions and cations can be deduced to be as follows: S0₄ ²">CH₃COO~>Cl~, and NH4⁺>Na⁺>K+>Mg²⁺. This represents the Hofmeister series of the salting out effect of ions (Scopes (1984)). For studies of the binding characteristics of AVID-AL gel, rProtein A gel was used as a reference. The results shown in Figure 2 indicate that use of the salts, ammonium acetate, ammonium chloride, sodium sulfate, and sodium chloride results in an increase of the binding ability to 18%, 15%, 12%, and 8%, respectively, as compared to potassium chloride with the same ionic strength (1.28 M) in the buffer. However, for any given salt, no significant relationship has been observed between the binding capacity and the ionic strength for rProtein A gel.

Table 1 summarizes the effect of salt type and concentration on the partition coefficient, a, which is defined as the fraction of the total solute (purified human IgG) adsorbed at any instant because the affinity adsorption of purified human IgG to AVID-AL gel represents a partitioning of solute between two phases, liquid and solid. Table 1 shows that the binding capacity of AVID-AL gel improves at higher ionic strengths of Na2S0₄ and (NH4)2S04 with increasing α. By increasing the ionic strength form 0.75 to 3.0 M, a 1.8-2.2-fold increase in a was observed with Na2Sθ4 and (NH4)₂S0₄« However, as expected for cations, only a weak effect on a was observed with increasing ionic strength from 0.25 to 1.0 M (NH₄ ⁺>Na⁺>K+>Mg²+) .

TABLE 1

Salt effect on the partition coefficient of purified human IgG binding to AVID-AL and rProtein A gels

l a² α3 α4

AVID rProtein AVID rProtein AVID rProtein AVID rProtein

Salt AL A AL A AL A AL A

Na2Sθ4 0.45 0.49 0.63 0.47 0.76 1.00 0.83 0.97

(NH₄)₂ so₄ 0.32 0.45 0.60 0.34 0.69 0.61 0.71 0.56

NH Ac 0.42 0.45 0.57 0.55 0.37 0.46 0.49 0.51

NH C1 0.41 0.42 0.40 0.49 0.45 0.55 0.46 0.50

NaCl 0.40 0.48 0.45 0.51 0.41 0.47 0.40 0.47

KCl 0.40 0.42 0.39 0.46 0.36 0.38 0.36 0.43

MgCl₂ 0.31 0.32 0.35 0.40 0.38 0.40 n.d. n.d.

1,2,3, and 4—tested at a salt concentration of 0.25 M, 0.50 M, 0.75 M, and 1.00 M, respectively; n.d.—not determined. The parameters for the binding of purified human IgG to AVID-AL gel, where the 9.4 to 27.5 mg IgG were added per mL gel, were determined using PBS containing different concentrations of Na2S04 (from 0.25 to 1.00 M) . The corresponding binding capacity is shown in Figure 3. The apparent dissociation constant K_a (mg/mL in solution) , and the theoretical binding capacity of AVID-AL gel, Ct (mg/mL gel), under the given experimental conditions are described by the equation:

1/P = Ct/(K_aP_b) - 1/K_a where P is the IgG concentration in free solution at equilibrium with AVID-AL gel, and Pfc is the IgG bound onto the AVID-AL gel (mg/ml gel) . Both K_a and C^ can be estimated from double reciprocal plots of the respective binding saturation curves and the values are summarized in Table 2. The dissociation constants for purified human IgG in PBS obtained under equilibrium conditions were 0.048, 0.022, 0.0058, 0.0052, and 0.0046 mg/mL, for Na2≤θ4 concentrations of 0, 0.25, 0.50, 0.75, and 1.00 M, respectively.

TABLE 2

Binding parameters of AVID-AL gel for purified human IgG in PBS containing different concentrations of Na2≤θ4

a2Sθ4 Concentration Dissociation constant Binding Ability (M) Ka, mg/mL Ct, mg/mL gel

PBS 0.0480 13.5

PBS + Na2Sθ4, 0.25 0.0220 16.1

PBS + Na2S0 _/ 0.50 0.0058 22.7

PBS + Na2Sθ4, 0.75 0.0052 27.1

PBS + Na2S04, 1.00 0.0046 29.9 It is apparent from Figure 3 and Table 2 that the value of the dissociation constant estimated for purified human IgG binding to AVID-AL gel decreased with increasing concentration of Na2Sθ4 in PBS (pH 7.0). The difference between K_a values for purified human IgG in PBS with no added Na2Sθ4 and in PBS plus 1.00 M of Na2Sθ4 was as much as ten fold. The theoretical binding capacity of AVID-AL gel for purified human IgG was 13.5 mg/mL gel in PBS, which is similar to the results reported by Ngo et al. (1992), who used a packed AVID-AL column (0.4 mL) and tested the binding capacity under equilibrium conditions.

The salting out effect of anions, such as sulfate, rather than the ionic strength is the major factor responsible for the observed increase in the binding capacity of AVID-AL gel. It is assumed that this is consequent from a lowering of the dissociation constant for the affinity adsorption of purified human IgG onto the gel. This suggests that the high ionic strength causes aggregation by strengthening hydrophobic interaction between proteins. Consequently, in the presence of sulfate the specific affinity of IgG for AVID-AL may be improved by reducing the repulsion between IgG molecules. In general, the repulsion of bound IgG molecule on the gel prevents the affinity attraction of additional unbound IgG molecules to an available adjacent ligand.

EXAMPLE II

The results from Example I were confirmed by performing experiments in which 0.5 L of AVID-AL gel was packed in a 1.0 L chromatographic column. The chromatograms shown in Figures 4(a) and 4(b) are for such columns, wherein human serum was diluted by either PBS (regular binding buffer) or PBS containing 0.75 M of Na2Sθ₄ as an improved binding buffer. This operation yielded 8.0 mg and 14.6 mg bound IgG on the column, respectively. The bound IgG can be eluted using neutral elution buffer with pH of 7.4. The binding capacity for AVID-AL in the high salt buffer was 26.2 mg IgG/mL gel, which was 2.1-fold higher than those of the run using PBS as a binding buffer. A comparison of the characteristics for IgG purification using AVID-AL gel has been listed in Table 3 for the two binding buffer systems. In both cases, the IgG was recovered at greater than 90% purity as judged by SDS-PAGE (8-25%) analysis, as shown in Figure 5. In particular, albumin and human serum cannot be strongly adsorbed onto the AVID-AL gel in the presence of high salt buffer solution when the amount of IgG being isolated is more concentrated than other proteins in solution (1) . Therefore, non-specific adsorption, as a major source of contamination in IgG purification by AVID-AL gel, has not been significantly observed when sodium sulfate was added to increase the ability to isolate IgG from human serum (see lane 8 in Figure 5) .

TABLE 3

Comparison of properties of purifying IgG from human serum using AVID-AL gel in two different binding buffer systems

Terms Exp 1 Exp 2

Volume of column (mL) 0.5 0. 5 Binding Buffer PBS PBS + 0.75

M Na2S04

Fed IgG (mg) 21.2 21. 2

Bound IgG (mg) 8. 0 14 . 6

Eluted IgG (mg) 6.2 13 . 1

Regenerated IgG (mg) 1. 0 0. 9

Recovery yield (%) 77. 5 89 . 7

Binding Capacity (mg/mL gel) 12 . 4 26 . 2 Table 4 is a comparison of the binding capacity for IgG between AVID-AL gel when used with PBS buffer and a high salt binding buffer of the invention. A comparison with Protein A gel is also set forth. The composition of the high salt binding buffer (HSBB) was KH₂P0₄ 0.01 M; Na₂HP04 0.01 M, NaCl 0.14 M and Na2≤θ4 0.5 M. The buffer is adjusted to physiological pH 7.4 with IN NaOH as necessary.

TABLE 4

l

**; A,-BPS; B,-HSBB (High Salt Binding Buffer)

*; The binding capacity and the recovery for IgG are the function of either IgG concentration in species or loading speed of sample passed through the gel column. The data in Table 4 unambiguously demonstrates that the immunoglobin hydrophobicity modifying effect of the buffer compositions of this invention combines with the specificity of the AVID-AL affinity ligand in a very unique way. Specifically, both high yield and high purity product is obtained. This result contrasts sharply with the indiscriminate salting out of all proteins present in a cell culture supernatant or the like.

EXAMPLE III

This example shows the effect of Na2Sθ4 concentration in high salt binding buffer on binding capacity of human IgG and the purity of IgG eluted from 0.5 mL Avid AL gel.

As Table 5 shows, reducing Na2Sθ4 from 0.75 M to 0.5 M in PBS lowers the binding capacity of AVID AL gel from human IgG. However, the purity of eluted from gel column using neutral elution buffer is increased.

. TABLE 5

Comparison on IgG purification from human serum using AVID AL gel with different Na2S04 concentrations in PBS as the binding buffer

Binding Buffer Protein in Elution Pooled Purity* Bindin

(mg/mL) volume (%) Capacit

Concentra- Concentra¬ of (mg/mL tion based tion based Elution gel) on A28O human IgG (mL)

RID kit

PBS 0.560 0.498 10.2 88.9 10.1

PBS/0.75M Na₂S0₄ 1.298 1.214 10.0 93.6 24.3

PBS/0.50M Na₂S0₄ 0.913 0.885 10.7 97.0 18.9

* Purity was calculated with the eluted human IgG concentratio based on human IgG RID kit divided by eluted protein concentration based on A280 nm. The date in Table 5 shows an optimum concentration range of salt in the high salt buffer of this invention in which the combined effect described with reference to Table 4 in Example III is observed. The limits of this optimum concentration range will vary according to the nature of the isotype and relative hydrophobicity of the immunoglobin recovery. When the upper limit of salt concentration useful in this invention is exceeded, extraneous proteins such as albumin, begin to co-purify contemplating the desired product.

In order to confirm the results obtained by use of radial immunodiffusion assay (RID) , the electrophoretic analysis for estimating the purity of IgG purified by AVID AL were performed.

Figure 6 depicts the SDS-polyacrylamide gradient gels on 8-25% in the Phast Gel System were shown below:

Lane 1) and 9) Molecular weight marker reference proteins Lane 2) Unbound protein after human serum binding by using PBS Lane 3) Unadsorbed protein after column wash using PBS Lane 4) Eluted human IgG by using Neutral

Elution Buffer Lane 5) Pooled fractions from gel regeneration Lane 6) Unbound protein after human serum binding by using PBS/O.75 M Na2Sθ4 Lane 7) Unadsorbed protein after column wash using PBS/O.75 M Na2Sθ4 Lane 8) Eluted human IgG by using Neutral

Elution Buffer Lane 10) Pooled fractions from gel regeneration Lane 11) Unbound protein after human serum binding by using PBS/O.5 M Na2Sθ4

Lane 12) Unadsorbed protein after columr wash using PBS/O.5 M Na2Sθ4

Lane 13) Eluted human IgG by using Neutral Elution Buffer

Lane 14) Pooled fractions from gel regeneration

Lane 15) Undiluted ran human serum EXAMPLE 4

12 mL of cell culture supernatant (#8000, batch #C3U281) containing human IgM was determined to contain 0.5 mg/mL. This was added with 0.5 mL AVID AL gel in a tube (sample 1) . A second tube was prepared similarly, with 0.85g Na2S0₄ also added (sample 2) . Both tubes were shaken overnight at room temperature. The gel from each tube was packed in a 1.0 mL syringe column, washed: sample 1 was washed with PBS, and sample 2 was washed with PBS/0.5 M Na2Sθ4. The human IgM was eluted from each column in ten fractions. Recovery was determined using human IgM specific RID plates, purchased from The Binding Site Company, San Diego, California. Purity was determined from the RID data and spectraphotometric measure of total protein concentration at 280 nm. The results of this experiment are reported in Table 6.

TABLE 6

Effect of Adding Salt Directly to Cell culture Supernatant on Human IgM Recovery

Items Sample 1 Sample 2 No Salt Added Sodium Sulfate Added

Volume of Column 0.5 mL 0.5 mL

Starting 12 mL of Cell Culture 12 mL of Cell Culture Material Supernatant Supernatant

Recovery 19.3% 54.8%

Purity 88.6% 85.7%

Binding Capacity 2.8 mg/mL 8.3 mg/mL This invention provides a novel and significant improvement in techniques for the purification of immunoglobin solutions, particularly cell culture supernatants by affinity chromatography in which a synthetic ligand or gel is utilized. Pursuant to this invention, aqueous buffer solutions containing sulfate, acetate, chloride potassium or sodium ions in a concentration sufficient to enhance the binding of a synthetic affinity gel for an immunoglobin for an immunoglobin are provided. The buffers of the invention preferably comprise aqueous solutions which are at least 0.1 molar, preferably 0.5 to 2.0 molar in sodium, potassium or ammonium sulfate, acetate or chloride. A particularly preferred buffer is 0.25 to 0.5 molar in sodium sulfate. Another preferred buffer is PBS which is 0.1 to 2.0 molar in sodium sulfate.

BIBLIOGRAPHY

1. Ngo, T.T. , et al., J. Chromatogr. 510:281-291 (1990)

2. Narinesingh, D. , et al., Analytical Lett. 24^2005-2015 (1991)

3. Khatter, N., et al. Amer. Lab. May (1991)

4. Ngo, T.T., et al. J. Chromatogr. 597:101-109 (1992)

5. Ngo, T.T. , et al. Appl. Biochem. and Biotechnol. 20:111-119 (1991)

6. Rosenberg, J., Amer. Biotech. Lab. In Press (1993)

7. Scopes, R. , In C.R. cantor (Editor), Protein Purification, Principles and Practice, Springer-Verlag, New York, pp. 47-50 (1984)

Claims (4)

CLAIMS :

1. A process which comprises:

(i) providing immunoglobin molecules dissolved in an aqueous medium also containing dissolved impurities, said medium containing sulfate, acetate, chloride, sodium, potassium or ammonium ions or combinations thereof in a concentration effective to reduce the inter se repulsion of said immunoglobin molecules in solution in said medium;

(ii) providing a synthetic affinity gel to which said immunoglobin molecules bind;

(iii) contacting said medium as defined in step (i) with said affinity gel to cause said immunoglobin molecules to bind to said gel; and

(iv) thereafter recovering said immunoglobin molecules which bind to said gel; wherein said immunoglobin molecules recovered in step (iv) are in high yield based upon the concentration thereof in said medium defined by step (i) and are substantially free of contaminants present in said medium as defined by step (i) .

2. The process of claim 1 in which said aqueous medium specified in part (i) is from .01 molar to 2 molar in said ions.

3. A process of claim 1 or claim 2 in which said aqueous medium also containing dissolved impurities, is a cell culture supernatant.

4. A process which comprises

(i) providing a cell culture supernatant containing a dissolved immunoglobin;

(ii) dissolving sodium sulfate in said supernatant in an amount sufficient to render said supernatant from 0.01 molar to 2 molar in said sodium sulfate;

(iii) providing a synthetic affinity gel to which said immunoglobin molecules in said supernatant bind;

(iv) contacting said supernatant as defined in step (i) with said affinity gel to cause said immunoglobin molecules in said supernatant to bind to said gel; and

(v) thereafter recovering the immunoglobin molecules which bind to said gel.