Adsorption of human immunoglobulin G onto ethacrylate and histidine-linked methacrylate

Adsorption of human immunoglobulin G onto ethacrylate and histidine-linked methacrylate

L.C.L.Aquino^I; E.A.Miranda^I; I.S.Duarte^I; P.T.V.Rosa^II; S.M.A.Bueno^I*

^ISchool of Chemical Engineering

^IISchool of Food Engineering, State University of Campinas, UNICAMP, Phone: (55) (19) 3788-3919, Fax (55) (19) 3788-3890, Cx. P 6066, CEP 13083-970, Campinas - SP, Brazil, E-mail:[e-mail]

ABSTRACT

The adsorption of human IgG onto GMA (a semirigid methacrylate-based chromatography matrix) and His-GMA adsorbents was studied by chromatography and batch equilibrium binding analysis. IgG molecules adsorbed onto GMA gel by nonspecific hydrophobic interactions and the specificities were similar for both adsorbents. Adsorption data were analyzed using three isotherm models, namely the Langmuir, Freundlich and Langmuir-Freundlich models, and the adsorption parameters were computed. The experimental isotherms were best described by a combined Langmuir-Freundlich model, which indicated the presence of unequal binding sites on both adsorbents and/or positive cooperativity in the binding of the IgG molecules.

Keywords: IgG, histidine, methacrylate, protein adsorption.

INTRODUCTION

Chromatographic methods, particularly adsorption chromatography, are essential tools for the purification of most plasma proteins (Burnouf, 1995; Burnouf et al., 1998), in which target molecules bind reversibly to chemical groups on the surface of the matrix in several forms of adsorption chromatography. The molecular properties most commonly utilized in these separation methods are charge and hydrophobicity as well as biospecific or pseudobiospecific affinity interactions (Burnouf et al., 1998; Janson and Rydén, 1989; Vijayalakshmi, 1989).

Affinity adsorption is the most selective of these adsorption techniques. Biospecific ligands such as antibodies or protein A, G, or L with their specificity have been immobilized on various supports for serum immunoglobulin purification (Burnouf et al., 1998; Füglistaller, 1989; Hermanson et al., 1995; Vola et al., 1995). However, it is well known that these types of ligands are difficult to immobilize at the required correct orientation, susceptible to degradation, and often expensive. On the other hand, pseudobiospecific ligands (such as dyes, amino acids and metal ions) can be used to purifiy a wide range of biomolecules, thus offering flexibility compared with biospecific ligands. They are usually smaller and simpler molecules than biospecific ligands with higher chemical and physical stability and a lower cost (Vijayalakshmi, 1989).

The conventional affinity ligand frequently used to purify human immunoglobulin G (IgG) from serum and other sources is protein A (Ladisch, 2001). The pseudobiospecific ligand histidine, immobilized on different gels and membranes to obtain chromatographic supports, has also been applied for the purification of IgG (Bueno et al., 1995; El-Kak and Vijayalakshmi, 1991, 1992; Haupt et al., 1995; Mandjiny and Vijayalakshmi, 1993; Müller-Schulte et al., 1991). The residual charges on immobilized histidine, resulting from the coupling chemistry used, the presence or absence of a spacer arm, the type of support material, and the adsorption conditions play important roles in the efficiency of IgG purification (Bueno et al., 1995; El-Kak and Vijayalakshmi, 1991, 1992; Haupt et al., 1995; Mandjiny and Vijayalakshmi, 1993; Müller-Schulte et al., 1991). When this amino acid is linked to soft gels (aminohexyl-Sepharose), it is possible to recover the IgG₁ and IgG₂ subclasses from human plasma and to purify monoclonal antibodies from cell cultures or ascite fluids (El-Kak and Vijayalakshmi, 1991, 1992). L-histidine immobilized on poly(ethylene vinyl alcohol) hollow fiber membranes (His-PEVA) had high selectivity for IgG adsorption from untreated human serum (Bueno et al., 1995). Depending on the buffer used for serum dilution, these membranes adsorbed IgG₁, IgG₂, IgG₃ (in the case of Mops buffer) or more selectively IgG₁ and IgG₃ (in the case of Tris-HCl buffer) (Haupt et al., 1995). His-PEVA had a higher capacity for adsorption of human IgG than the capacity reported for the Sepharose-based support using the same ligand coupling strategy, thereby demonstrating the influence of the support material on IgG adsorption onto immobilized histidine (Haupt et al., 1995). Recently, a histidine linked to a methacrylate-based adsorbent, a poly(2-hydroxyethylmethacrylate) prepared by Özkara et al. (2002), was used with success in the separation of human IgG from human plasma and aqueous solution. The adsorbent showed low nonspecific adsorption and a high capacity, and allowed one step purification of IgG from human plasma.

The purpose of this work was to study IgG adsorption onto histidine immobilized on a support also based on methacrylate, the semirigid GMA HW-65F gel (in this paper referred to as GMA). This support was obtained by the copolymerization of glycydyl methacrylate, polyethylene glycol, and the cross-linking agent pentaerythritol dimethacrylate (Hermanson et al., 1992). We were interested to know if this support, also methacrylate based but different from the matrix produced by Özkara et al. (2002), would also have a good performance as an IgG adsorbent. The advantages of this matrix, besides commercially availability, are two characteristics that made it appropriate from large-scale processing. First, it is capable of withstanding pressure drops of up to 100 psi without collapsing. In addition, the unusual chemical strength of the ester bonds in methacrylate-based polymers gives the support excellent resistance to extremes in pH (Hermanson et al., 1992).

In this paper we report on the human IgG adsorption on GMA and histidine coupled to GMA aiming at purification of this antibody from human serum or plasma. IgG adsorption isotherms determined at 25 and 37ºC were analyzed using three models, namely, the Langmuir, Freundlich, and Langmuir-Freundlich models. The best fit to the data was obtained using the Langmuir-Freundlich equation, which allowed detection of cooperativity in the protein adsorption. Protein adsorption-desorption performance and selectivity under identical conditions were compared for GMA gel and histidine coupled to GMA. Parameters pertinent to the adsorption process such as the dissociation constant, maximum capacity, and cooperativity were analyzed and discussed.

EXPERIMENTAL

Materials

The methacrylate matrix GMA (GMA HW-65F), sodium borohydride, L-histidine, 1,4 butanediol diglycidyl ether, crystalline bovine serum albumin, and human IgG (prepared from Cohn Fraction II, III, electrophoretic purity of approximately 99%) were purchased from Sigma (St. Louis, USA). The GMA had a pore diameter of 100 nm and a surface area of 105 m²/g, according to the supplier. All other chemicals were of analytical grade. Ultrapure water was obtained using the Millipore Milli-Q system (Millipore, Bedford, USA).

Immobilization of L-Histidine on GMA Gel

The GMA matrix was activated with oxirane 1,4 butanediol diglycidyl ether as in the protocols described by Sundberg and Porath (1974). Reactive oxiranes were introduced into the gel and subsequently opened and coupled with the amino group of histidine (Bueno et al., 1995; Haupt et al., 1995; Mandjiny and Vijayalakshmi, 1993). The derivatized gel is referred to as His-GMA in this work.

Determination of the Amount of Immobilized Histidine

A His-GMA column (10 cm ´ 1 cm i.d., packed bed volume of 5 mL) was equilibrated with copper ions by percolating 80 mL (16 column volumes) of 50 mM copper sulfate solution through it, followed by washing with 100 mL (20 column volumes) of water. The elution was done with 50 mM EDTA in 50 mM phosphate buffer containing 1 M NaCl at pH 7.0. The total amount of copper ions in the eluate was determined by spectrophotometry at 730 nm (DU 650 UV/vis spectrophotometer by Beckman, USA) after proper dilution using the EDTA solution as a blank. The experiments were performed in triplicate at a volumetric flow rate of 0.5 mL/min at 25° C. The metal-loading capacity of His-GMA was 44 mol/mL gel, assuming an ideal situation in which each copper ion would bind to only one histidine molecule.

Preparation of Human Serum

Human blood samples from a healthy person were collected without anticoagulant. These samples were kept in glass tubes at room temperature for about 2 hours to allow agglutination. Then they were centrifuged for 5 min at 200 g and the supernatant (serum) was used without further treatment in the adsorption experiments.

Determination of Protein Concentration

Protein concentration in samples from experiments carried out with high purity human IgG (99% pure IgG preparation) was determined by measuring absorbance at 280 nm, assuming a molar absorptivity of 14 for a 1% solution of IgG. For the experiments with human serum, protein concentration was determined by the dye binding method (Bradford, 1976) using crystalline bovine serum albumin as reference.

SDS-Polyacrylamide Gel Electrophoresis (SDS-PAGE)

Chromatographic fractions were analyzed by SDS-PAGE (4 to 15% gradient gels) under nonreducing conditions, using a Pharmacia PhastTM system (Pharmacia, Uppsala, Sweden). The gels were silver-stained according to the method provided by the manufacturer.

Nephelometric Quantification of IgA, IgG, and IgM

Concentration of IgA, IgG, and IgM in samples from chromatographies carried out with His-GMA and GMA were quantified using a Beckman Array Protein System with Beckman reagents (Beckman Instruments, USA).

Evaluation of IgG Adsorption Kinetics

IgG adsorption kinetics was evaluted with His-GMA gels at 25° C using batch experiments. These experiments, carried out in triplicate, were to determine the time required for adsorption equilibrium to be reached. The adsorbents (10.0 mg dry mass) were weighed in 1 mL plastic syringes to which filters were attached to retain adsorbent particles. The gel was then equilibrated with degassed 25 mM Tris-HCl buffer at pH 7.4 by drawing fresh solution into and pushing it out of the syringes twice. The syringes were then slowly filled with 1.0 mL of purified IgG solution at protein concentrations of 0.26, 3.0, and 9.0 mg/mL and rotated end-over-end at 6 rpm. Stirring was stopped at different time intervals and the IgG solution was pushed out of three syringes for each initial concentration. The unbound protein concentration in the solution was determined using UV spectrophotometry (280 nm) and the amount of bound protein was calculated by mass balance.

Protein Adsorption Studies

The protein adsorption experiments were carried out in triplicate at 25 and 37° C with GMA and His-GMA adsorbents. Ten milligrams (dry mass) of these adsorbents were weighed in 1 mL plastic syringes and equilibrated with buffer as described for the adsorption kinetics. Then 1.0 mL of IgG solution was slowly drawn into the syringes (concentration from 0.5 to 15 mg/mL). The syringes were rotated end-over-end at 6 rpm for 2 h to allow equilibrium to be established. After pushing the IgG solution out of the syringes, the unbound protein concentration in the liquid phase was determined using UV spectrophotometry at 280 nm and the amount of bound protein was calculated by mass balance. The adsorbed protein mass, Q, was determined by the difference between the amount of protein injected and that present in the supernatant C after equilibrium, multiplied by the feed volume (1.0 mL). Plotting Q against C yielded the equilibrium isotherm.

Analysis of Equilibrium Data

a) Langmuir Model

The Langmuir model assumes that the adsorption process takes place on a surface composed of a fixed number of adsorption sites of equal energy, with one molecule adsorbed per adsorption site until a monolayer coverage is obtained (Adamson, 1990). The Langmuir model can be described by the equation

where C is the adsorbate liquid-phase equilibrium concentration, Q is the adsorbate surface concentration, Q_m is the maximum adsorbate binding capacity and K_d is the apparent dissociation constant, which represents the affinity between adsorbate and adsorbent. These parameters were evaluated by fitting the Langmuir model to the experimental data, employing the iterative fitting method of Levenberg-Marquardt.

Some of the assumptions made in this model are not necessarily realistic because of the large difference between macromolecules and small molecules in the adsorption mechanisms at interfaces. The differences result mainly from (a) multiple-site binding for proteins, which often results in irreversible adsorption, (b) the heterogeneous nature of most solid surfaces, and (3) lateral and other interactions referred to as cooperative interactions (Luo and Andrade, 1998).

To establish the validity of the assumption that independent interaction sites of equivalent affinity are present on the adsorbent surface, several linear transformations of equation [1], including the Scatchard plot (Adamson, 1990), can be employed as follows:

The shapes of the plots of Q/C vs Q are particularly sensitive to whether independent, dependent nonidentical, or cooperative interaction processes participate in the binding of the protein to the immobilized ligands. If the Scatchard plot is convex, negative cooperativity is observed. If it is linear, independent interaction between protein and the ligand-binding sites exists, and if the Scatchard plot is asymptotic (or in some cases concave), positive cooperativity is observed.

Adsorbent surfaces are rarely homogeneous. There are a number of classical isotherm models for heterogeneous surfaces with continuous energy distribution: the Freundlich (Adamson, 1990) and Langmuir-Freundlich (Andrade, 1985) models are two examples.

b) Freundlich Model

In the Freundlich model (Adamson, 1990) the adsorbed mass per mass of adsorbent is a power law function of the solute concentration as follows:

where K and h are the Freundlich equilibrium constant and the Freundlich isotherm power term, respectively. Unlike the Langmuir, the Freundlich equation does not become linear at low concentrations but remains convex not does it show a saturation or limiting value. Although usually applied in a strictly empirical sense, this can be of theoretical interest in terms of adsorption onto an energetically heterogeneous surface (Adamson, 1990).

c) Langmuir-Freundlich Model

The single-component Langmuir-Freundlich model is

where K_d* is the apparent dissociation constant that includes contributions from ligand binding to monomer, monomer-dimer, and more highly associated forms of proteins; Q_mLF is the maximum binding capacity; and n is the Langmuir-Freundlich coefficient.

The Langmuir-Freundlich model was analyzed by Sips who found that the energy distribution function corresponds to a symmetrical quasi-Gaussian function (Adamson, 1990; Quinones and Guiochon, 1998). At low concentrations, the model reduces to the Freundlich model and in the case of a homogeneous surface, it reduces to the Langmuir model (Quiones and Guiochon, 1998).

By analogy with protein-multiple ligand interactions it has been suggested that equation (4) works well to model adsorption cooperativity (Jiang and Hearn, 1996; Sharma and Agarwal, 2001). For purely independent noninteracting sites, the value of n is 1. For positive cooperativity of the protein binding sites, n is greater than 1, while when 0 < n < 1 negative cooperativity in the binding process is indicated. The value of n can thus be employed as an empirical coefficient, representing the type and the extent of cooperativity present in the binding interaction.

Since the Langmuir-Freundlich model has more than two adjustable parameters, it is not easily fitted to the experimental data by linear regression or graphical means. In this case, it is necessary to apply nonlinear least-squares analysis.

Chromatographic Procedures

All chromatographic procedures were carried out with an automated Econo Liquid Chromatography System (Bio-Rad, USA) at 25ºC at a linear velocity of 38.2 cm/h (30.0 mL/h). GMA or His-GMA particles were suspended in water, degassed, and packed into a column (10 cm x 1 cm I.D.) to give a bed volume of 1 mL or 8 mL. The packed column was then equilibrated with 25 mM Tris-HCl buffer at pH 7.4 (equilibration buffer).

a) Frontal Analysis

A 2.0 mg/mL IgG solution in equilibration buffer was fed continuously into the column (bed volume of 1 mL) until the available capacity of the adsorber was exhausted and the IgG concentration in the effluent reached the value of the IgG feed concentration. Then the column was washed with equilibration buffer until the absorbance of the effluent at 280 nm reached zero. The capacity of the gel at this particular concentration of the protein was determined by eluting the column with the equilibration buffer containing 0.2 M NaCl. After each experiment, the gel was washed with three column volumes of 50 mM NaOH solution, followed by washing with water and finally with the equilibration buffer. The experiment was repeated with an IgG feed concentration of 4.0 mg/mL.

b) Purification of IgG from Human Serum

Human serum (2.0 mL, about 120 mg of total protein) was diluted fivefold with equilibration buffer and loaded into the column (bed volume of 8 mL), which had been pre-equilibrated with the equilibration buffer. The column was then washed with equilibration buffer until the absorbance of the effluent at 280 nm reached zero. Elution was performed with the same buffer containing 0.2 M NaCl. The absorbance of the eluate was monitored at 280 nm. After each experiment, the column was washed with 50 mM NaOH, followed by washing with water and finally with the equilibration buffer.

RESULTS AND DISCUSSION

Adsorption Isotherms

The adsorption of IgG onto His-GMA was relatively fast. In all cases (three different initial protein concentrations), most of the adsorption occurred in the first 2 min and the unbound protein concentration in solution remained approximately constant 20 min after the addition of IgG solution to the adsorbent (data not shown). In the isotherm determination experiments that followed, all the syringes were kept under stirring for 2 hours in order to give sufficient time to establish equilibrium.

The adsorption isotherms were plotted from the experimental data obtained from the batch adsorption experiments at 25° C for His-GMA and GMA adsorbents (Figure 1). The results showed deviation from the Langmuir form for both adsorbents. For both adsorbents a plateau followed by a second stage of protein adsorption at an equilibrium concentration higher than 7 mg/mL was observed, suggesting a BET type II isotherm (Adamson, 1990).

When the equilibrium IgG concentration was in the range of 4 to 8 mg/mL, the capacity for His-GMA was about 60% lower (despite its high ligand density, 44 m mol/mL) than the capacity obtained for GMA. The GMA polymer is rich in hydroxyl groups and ether bonds. The surface environment formed by this type of polymer network is only mildly hydrophilic and may even display some hydrophobic characteristics under certain conditions (Hermanson et al., 1992). Therefore, the IgG retention could be explained by assuming that, for GMA, the IgG molecules are adsorbed by nonspecific hydrophobic interactions. This is in contrast to the results obtained for adsorption of IgG from human and other sources onto histidine immobilized on several supports, such as gel beads (agarose) and membranes (flat and hollow fiber), where nonspecific interactions were not detected or were not significant (the IgG adsorbed onto PEVA hollow fiber membranes was equivalent to 12% of the total adsorbed protein on PEVA hollow fiber-linked histidine) (Bueno et al., 1995; El-Kak and Vijayalakshmi, 1991; Mandjiny and Vijayalakshmi, 1993; Müller-Schulte et al., 1991). Despite the fact that in pseudobioaffinity chromatography using histidine-linked matrix the adsorption mechanism between ligand and proteins is based on multiple interactions such as electrostatic, hydrophobic, charge transfer and possibly a nucleophilic attack due to the imidazole ring of histidine (Vijayalakshmi, 1989), some specificity is possible. For example, the interaction of human IgG with histidine immobilized on PEVA hollow fiber membranes by an ether linkage (the same coupling mode as that used in this work) was studied by Haupt and colleagues, who found that this IgG adsorbed onto this pseudobioaffinity support via their Fab part (Haupt et al., 1995).

When histidine was immobilized on GMA, the number of hydrophobic sites available on the matrix probably decreased due to reaction or to steric effects, thus decreasing the nonspecific IgG bonds. Moreover, some of the histidine molecules may not be accessible to IgG when coupled onto GMA because activation with bisoxirane may result in cross-linking of the methacrylate gel, reducing effective pore size.

Histidine has been coupled to gels (polyamide and agarose) and membranes (including PEVA) in different ways (Bueno et al., 1995; El-Kak et al., 1992; Mandjiny and Vijayalakshmi, 1993; Müller-Schulte et al., 1991). The coupling mode, the functional groups remaining free on the ligand, and the polymer support largely influence adsorption capacity and selectivity for IgG. In this study, we also observed the influence of the matrix (GMA) on adsorption of IgG onto immobilized histidine.

The data on equilibrium adsorption of IgG onto His-GMA and GMA for equilibrium protein concentrations lower than 7 mg/mL were analyzed using the Langmuir, Freundlich, Langmuir-Freundlich, Temkin, Bilangmuir, and extended Langmuir (multilayer) isotherms models. Only the former three models gave satisfactory fits for the data obtained with both adsorbents. The latter three models were not applicable to the data due to unrealistic (extremely high or negative) values for adsorption parameters. The isotherm model parameters for the Langmuir, Freundlich, and Langmuir-Freundlich models were evaluated by nonlinear regression using the Levenberg-Marquart method for GMA and His-GMA (Table 1). A comparison between experimental and theoretical profiles for adsorption of IgG onto both adsorbents is shown in Figure 2.

The Langmuir model was not useful to explain the behavior of protein adsorption satisfactorily. This was manifested in the poor agreement between the experimental data and simulated curves for both adsorbents (Figure 2a). This deviation was more evident in the Scatchard plots (Q/C versus Q), which were nonlinear (data not shown).

These results are in contrast to the results obtained for adsorption of human IgG onto PEVA hollow fiber membrane-linked histidine, where the same coupling mode method was used by Bueno et al. (1995). In their work, the Langmuir model could be used to explain the adsorption pattern of the equilibrium data, and the Scatchard plots indicated that the adsorption of human IgG onto this PEVA support obeys the Langmuir isotherm at 25° C (linear plot). The value obtained for n (1.05) from the Hill plot proved that there was no cooperativity due to protein-protein interaction for this system.

Analysis of adsorption using the Freundlich model also indicates that the fit was inadequate to describe adsorption equilibrium data for IgG on both adsorbents studied (correlation coefficient of 0.964 and 0.962 for GMA and His-GMA, respectively). To overcome this inadequacy the Langmuir and Freundlich models may be combined to give the composite Langmuir-Freundlich equation (Andrade, 1985; James and Do, 1991; Jiang and Hearn, 1996; Sharma and Agarwal, 2001). Good fits were obtained for IgG adsorption using this model (variance equal to 2.82 and 1.15 and correlation coefficient equal to 0.993 and 0.992 for GMA and His-GMA, respectively) (Figure 2c and Table 1). Furthermore, the model was able to account for the sigmoidal shape of the experimental profile, which was probably due to cooperative interactions (Sharma and Agarwal, 2001). The contribution of hydrophobic and protein-protein interactions is manifested in the alteration of the isotherm shape from Langmuirean to sigmoidal. Values of n > 1 for both adsorbents (n = 3.24 and n = 2.00 for GMA and His-GMA, respectively) indicate positive cooperativity in binding (attractive force due to lateral interactions) and the heterogeneous nature of the adsorption. Cooperativity depends on the macromolecular nature and the multiple functional groups, which usually result in multiple interactions.

The values for the apparent dissociation constant for the Langmuir-Freundlich model were of the order of magnitude of 10^-4 and 10^-5 M. The K_d for human IgG reported previously for PEVA hollow fiber membranes (Bueno et al., 1995) and for different flat sheet membranes (Mandjiny and Vijayalakshmi, 1993) with immobilized histidine were of the same order of magnitude (10^-5 M). These values for K_d indicate a medium affinity between IgG and adsorbents (Vijayalakshmi, 1989).

In order to verify whether the amount of IgG adsorbed onto GMA and His-GMA would be different with an increase in temperature (probing the role of hydrophobic interaction in the adsorption, since they are favored at high temperatures), adsorption isotherms were also determined at 37° C. Adsorption for both adsorbents at 37° C was higher than that observed at 25° C: Q increased approximately from 35 to 40-45 mg/mL and from 20-25 to 30 mg/mL for GMA and His-GMA, respectively, for C equals to 7 mg/mL (Table 2 and Figure 3). This corroborates the idea that hydrophobic interactions have a role in adsorption. Similar results were reported by Haupt et al. (1995) for adsorption of the human IgG onto PEVA hollow fiber membrane-linked histidine. The Langmuir, Freundlich, and Langmuir-Freundlich models were not appropriate to satisfactorily explain the protein adsorption behavior at 37° C on both adsorbents, as indicated by the poor relationship between the experimental data and theoretical profiles.

Desorption Studies

In order to verify the general observation that nonspecific adsorption of IgG onto chromatographic support media is often of weak affinity, the actual adsorbed IgG mass was measured after desorption. The experiments were carried out in dynamic mode (frontal analysis), using chromatographic conditions similar to those reported by Bueno et al. (1995) for IgG adsorption, elution, and regeneration of histidine immobilized on PEVA hollow fiber membranes (Table 3).

Under the first set of operational conditions (initial IgG concentration of 2.0 mg/mL), approximately the same amount of IgG was removed from the adsorbents during washing with equilibrium buffer (7.54 and 7.75 mg for GMA and His-GMA, respectively). This relatively large amount of protein (92 and 91% of total protein fed into the columns) suggests a weak interaction between IgG and the adsorbents. Also, since the particles are highly porous, some IgG may have remained inside the pores without being adsorbed at all. About the same amount of adsorbed IgG molecules were eluted with 0.2 M NaCl for both adsorbents (0.39 and 0.38 mg equivalent to 60 and 50% for GMA and His-GMA, respectively). The remaining protein was desorbed only with 50 mM NaOH. The results obtained under the second set of operational conditions (initial IgG concentration of 4.0 mg/mL) were similar. This desorption with low salt concentration allows us to assign partial responsibility for the adsorption onto His-GMA to electrostatic interaction, since the histidine molecule has a dipole at the adsorption pH used (pH 7.4). However, these desorption results differ from those obtained by Bueno et al. (1995), when all the human IgG adsorbed onto PEVA membrane-linked histidine was eluted with 0.2 M NaCl.

The low IgG capacity obtained for His-GMA (0.76 and 0.77 mg/mL for initial IgG concentration of 2.0 and 4.0 mg/mL, respectively) may be explained based on the buffer system used (Tris-HCl buffer). Haupt et al. (1995) reported that IgG adsorption onto histidine immobilized on PEVA hollow fiber membranes is strongly influenced by the nature and protonation state of the buffer ions. In their work, an apparently small amount of IgG eluted when using Tris-HCl buffer was due to the fact that the IgG fraction was not entirely adsorbed (only IgG₁ and IgG₃ were adsorbed) (Bueno et al., 1995). This may have occurred in the present work since when we fed the column (bed volume of 1 mL) with a saturating amount of IgG at a concentration of 20 mg/mL, higher adsorption capacities were obtained with the zwitterionic buffer hydroxyethylpiperazineethanesulfonic acid (Hepes) (up to 2.0 mg of the total IgG adsorbed) than with the Tris-HCl buffer (1.2 mg of the total IgG adsorbed).

IgG Adsorption from Human Serum

Experiments were run to evaluate the adsorption selectivity of GMA and His-GMA for serum immunoglobulins. Untreated human serum (2.0 mL, about 120 mg of total protein) was chromatographed on both adsorbents with elution performed with 0.2 M NaCl in the equilibration buffer. The amount of adsorbed protein for His-GMA (1.63 mg) was double the amount for GMA (0.85 mg). However, nephelometric analysis of the eluted fractions revealed similar IgG:IgA:IgM mass ratios (40:12:48 and 41:10:49 for GMA and His-GMA, respectively). Both supports were also similar in protein profile given by SDS-PAGE electrophoresis (Figure 4). The nonretained proteins contained albumin, immunoglobulins and other serum proteins. The adsorbed proteins contained immunoglobulins with albumin and other protein impurities. The high-molecular-mass protein present in the eluted fractions that only poorly penetrate the gel probably correspond to IgM (high molecular mass, 935 kDa) since these bands were not detected when electrophoresis was carried out under reducing conditions (data not shown).

Neither one of the adsorbents allowed for IgG separation at high resolution. This result may seem to contradict that reported by Bueno et al. (1995) (histidine immobilized on poly(ethylenevinyl alcohol hollow fiber membranes), but we have to note that the support was different in each case. Despite this low resolution, the chromatographies with GMA gel indicated that there is a potential for its use in IgG separation when high resolution is not required, since the major impurity, albumin, could be efficiently separated from the fraction containing IgG. Surprisingly, the immobilization of histidine did not improve the performance of the GMA in this specific separation.

CONCLUSIONS

We studied the adsorption of human IgG onto GMA and His-GMA adsorbents under identical conditions. IgG molecules were adsorbed onto GMA by nonspecific hydrophobic interactions and the specificities were similar for both adsorbents. The adsorption data were analyzed using three different isotherm models, namely the Langmuir, Freundlich, and Langmuir-Freundlich models. Although empirical, the Langmuir-Freundlich model provided a good fit for equilibrium isotherm data. This model described the protein binding to GMA and His-GMA gels and enabled an explanation of binding cooperativity and heterogeneity in quantitative terms for both protein-gel systems studied. The desorption studies also indicated that IgG adsorption seemed to be governed by hydrophobic interactions and that other interactions (van der Waals interactions and electrostatic forces (charge-induced dipole)) have some role in the adsorption. Data indicate that there is a potential for use of GMA gel in IgG separation when high resolution is not required.

ACKNOWLEDGMENTS

We acknowledge the financial support received from FAPESP and FAEP-UNICAMP, Brazil. The authors thank Dr. Cesar Costapinto Santana at the School of Chemical Engineering, State University of Campinas, Brazil, in whose laboratories this work was developed.

Received: December 2, 2002

Accepted: May 8, 2003

• Adamson, A.W., Physical Chemistry of Surfaces, fifth edition, John Wiley and Sons, Inc., New York (1990).
• Andrade, J.D., in J.D. Andrade (Editor), Surface and Interfacial Aspects of Biomedical Polymers, Plenum Press, New York, Vol. 2, pp. 1-80 (1985).
• Bradford, M.M., A Rapid and Sensitive Method for the Quantitation of Microgram Quantities of Protein Utilizing the Principle of Protein-Dye Binding, Analytical Biochemistry, 72, 248-254 (1976).
• Bueno, S.M.A., Haupt, K. and Vijayalakshmi, M.A., Separation of Immunoglobulin G from Human Serum by Pseudobioaffinity Chromatography Using Immobilized L-Histidine in Hollow Fibre Membranes, Journal of Chromatography B, 667, 57-67 (1995).
• Burnouf, T., Chromatography in Plasma Fractionation: Benefits and Future Trends, Journal of Chromatography B, 664, 3-15 (1995).
• Burnouf, T., Goubran, H. and Radosevich, M., Application of Bioaffinity Technology in Therapeutic Extracorporeal Plasmapheresis and Large-Scale Fractionation of Human Plasma, Journal of Chromatography B, 715, 65-80 (1998).
• El-Kak, A. and Vijayalakshmi, M.A., Study of the Separation of Mouse Monoclonal Antibodies by Pseudobioaffinity Chromatography Using Matrix-Linked Histidine and Histamine, Journal of Chromatography, 570, 29-41 (1991).
• El-Kak, A. and Vijayalakshmi, M.A., Separation and Purification of IgG₁ and IgG₂ from IgG-Cohn-II Fraction of Human Plasma by Pseudobioaffinity Chromatography on Histidyl-AH-Sepharose, Bioseparation, 3, 47-53 (1992).
• El-Kak, A., Manjini, S. and Vijayalakshmi, M.A., Interaction of Immunoglobulin G with Immobilized Histidine: Mechanistic and Kinetic Aspects, Journal of Chromatography, 604, 29-37 (1992).
• Füglistaller, P., Comparison of Immunoglobulin Binding Capacities and Ligand Leakage Using Eight Different Protein A Affinity Chromatography Matrices, Journal of Immunological Methods, 124, 171-177 (1989).
• Haupt, K., Bueno, S.M.A. and Vijayalakshmi, M.A., Interaction of Human Immunoglobulin G with L-Histidine Immobilized onto Poly(ethylene vinyl alcohol) Hollow-Fiber Membranes , Journal of Chromatography B, 674, 13-21 (1995).
• Hermanson, G.T., Mallia, A.K. and Smith, P.K., Immobilized Affinity Ligand Techniques. Academic Press, Inc., San Diego (1992).
• Hermanson, G.T., Mattson, G.R. and Krohn, R.I., Preparation and Use of Immunoglobulin-Binding Affinity Supports on Emphaze Beads, Journal of Chromatography A, 691, 113-122 (1995).
• James, E.A. and Do, D.D., Equilibria of Biomolecules on Ion-exchange Adsorbents, Journal of Chromatography, 542, 19-28 (1991).
• Janson, J.C. and Rydén, L., Protein Purification. Principles, High Resolution Methods and Applications. VCH Publishers, New York (1989).
• Jiang, W. and Hearn, M.T.W., Protein Interaction with Immobilized Metal Ion Affinity Ligands under High Ionic Strength Conditions, Analytical Biochemistry, 242, 45-54 (1996).
• Ladisch, M.R., Bioseparations Engineering, Wiley-Interscience, New York, p. 673 (2001).
• Luo, Q. and Andrade, J.D., Cooperative Adsorption of Proteins onto Hydroxyapatite, Journal of Colloid Interface Science, 200, 104-113 (1998).
• Mandjiny, S. and Vijayalakshmi, M.A., in C. Rivat and J.-F. Stoltz (Editors), Biotechnology of Blood Proteins, Colloque INSERM, John Libbey Eurotext, Montrouge, Vol. 227, pp.189-195 (1993).
• Müller-Schulte, D., Manjini, S. and Vijayalakshmi, M.A., Comparative Affinity Chromatographic Studies Using Novel Grafted Polyamide and Poly(Vinyl Alcohol) Media, Journal of Chromatography, 539, 307-314 (1991).
• Özkara, S., Yavuz, H., Patir, S., Arica, M.Y. and Denizli, A., Separation of Human Immunoglobulin G from Human Plasma with L-Histidine Immobilized Pseudo-specific Bioaffinity Adsorbents, Separation Science and Technology, 37, 717-731 (2002).
• Quiñones, I. and Guiochon, G., Extension of a Jovanovic-Freundlich Isotherm Model to Multicomponent Adsorption on Heterogeneous Surfaces, Journal of Chromatography A, 796, 15-40 (1998).
• Sharma, S. and Agarwal, G.P., Interactions of Proteins with Immobilized Metal Ions: a Comparative Analysis Using Various Isotherm Models, Analytical Biochemistry, 288, 126-140 (2001).
• Sundberg, L. and Porath, J., Preparation of Adsorbents for Biospecific Affinity Chromatography. I. Attachment of Group-containing Ligands to Insoluble Polymers by Means of Bifunctional Oxiranes, Journal of Chromatography, 90, 87-98 (1974).
• Vijayalakshmi, M.A., Pseudobiospecific Ligand Affinity Chromatography, Trends in Biotechnology, 7, 71-76 (1989).
• Vola, R., Lombardi, A., Tarditi, L., Björck, L. and Mariani, M., Recombinant Proteins L and LG: Efficient Tools for Purification of Murine Immunoglobulin G Fragments, Journal of Chromatography B, 668, 209-218 (1995).

*

To whom correspondence should be addressed

Publication Dates

• Publication in this collection
  02 Sept 2003
• Date of issue
  Sept 2003

History

• Received
  02 Dec 2002
• Accepted
  08 May 2003