Affinity electrophoresis in gels containing hydrophobic substituents.

The electrophoresis of a number of proteins was studied in poly(N,N-dimethylacrylamide) gels containing entrapped linear copolymers of N,N-dimethylacrylamide with N-alkyl-substituted acrylamides. The mobility of soybean trypsin inhibitor, carbonic anhydrase, ovalbumin, and myoglobin were unaffected by the hydrophobic residues in the gel. The mobilities of bovine serum albumin and beta-lactoglobulin A were sharply reduced, with the effect increasing as the alkyl side chain on the acrylamide residue was extended from dodecyl to octadecyl. The effect of the octadecyl ligand could be further increased by interposing a spacer between it and the polymer chain backbone. The retardation of the mobilities was used to obtain protein association constants with the alkyl residues. Interaction with the hydrophobic ligand produced a separation of beta-lactoglobin A into two fractions.

Gel electrophoresis has found a number of important applications in biochemistry. It is a powerful tool for the characterization of mixtures of native proteins, being particularly useful in distinguishing between closely related species (1,2). The electrophoretic mobility depends here on the charge and the size of the particle. In the presence of anionic detergents, proteins unfold, and since the charge of the adsorbed anions seems to be proportional to the length of the chain molecule, the electrophoretic mobility may be used to estimate the molecular weight (3, 4). Such estimates seem to be generally fairly reliable, although lactose permease exhibits a mobility which is much higher than expected, presumably because of unusually strong binding of the detergent to hydrophobic sequences of the polypeptide chain (5). Electrophoresis carried out in two mutually perpendicular directions under different conditions constitutes a powerful method to characterize complex protein mixtures (6). When the electrophoresis of a protein is observed in a gel with a gradient of urea concentration perpendicular to the electrical field, the result may be interpreted in terms of the kinetics of protein unfolding (7). Finally, gel electrophoresis has been of inestimable value in the development of procedures for the sequencing of DNA (8,

91,
Relatively few studies have attempted to introduce groups into the gel which would interact specifically with a protein to be studied by gel electrophoresis. Takeo and Nakamura (10) observed that glycogen entrapped in polyacrylamide gel re-* This study was supported by Grant GM-05811 from the National Institutes of Health, A preliminary report is given in Polymer Preprints 22, (1981), Division of Polymer Chemistry, American Chemical Society. The paper will form part of the Ph.D. dissertation to be submitted by J.-L. C . to the Graduate School of the Polytechnic Institute of New York. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. duced the electrophoretic mobility of glucan phosphorylase, and they showed how the data may be interpreted in terms of the association constant of the enzyme with its substrate. Most of the work utilizing this technique was concerned with the study of lectins. It used either agarose gels with covalently attached or entrapped lectins which interacted specifically with glycoproteins (11) or polyacrylamide gels with entrapped linear copolymers of acrylamide with allyl glycosides (12,13) which slowed the electrophoresis of lectins.
It has long been known that serum albumin exhibits a high affinity for long chain fatty acids (14). This has been interpreted in terms of clefts in the surface of the globular protein which have hydrophobic character and contain a cationic site (15). In this investigation, we used the technique of affinity electrophoresis to study the interaction of bovine serum albumin with paraffhic chains, More limited data were obtained for p-lactoglobulin A, which is also known to bind aliphatic hydrocarbons (16).
The monomers were characterized by NMR spectra recorded on a Varian model A-60 spectrometer, and IR spectra recorded on a Perkin-Elmer model 457 spectrometer. N-Tetradecyl-, N-hexadecyl-, and N-octadecylacrylarnide were also characterized by chemical ionization mass spectra and found essentially free of their homologs.
Lineer Copolymers-N,N-Dimethylacrylamide was copolymerized ' The abbreviations used are: BSA, bovine serum albumin; P-LG- for 24 h with acrylamide derivatives carrying hydrophobic side chains in methanol solution (containing 25 %, w/w, of the mixed monomers) at 60 "C using 0.1 mol % (based on monomer concentration) of azobis-isobutyronitrile initiator. Complete monomer conversion was obtained so that the copolymers had the same composition as the mixed monomers. The copolymers were three times dissolved in acetone and reprecipitated in anhydrous ether before drying in a vacuum oven at 50 "C. Electrophoresis-The disc electrophoresis procedure of Davis (17) was used with the multiphasic buffer systems A of Rodhard and Chrambach (18). Separation gels were prepared by dissolving linear copolymers, N,N-dimethylacrylamide and methylene-bis-acrylamide (3 %, w/w, based on monomers and linear copolymer) in pH 8.79 buffer so as to obtain a 7% (w/v) gel. After exposing the solution at room temperature to reduced pressure for 5 min to eliminate atmospheric oxygen, 0.02 ml of 1% ammonium persulfate and 1 pl of N,N,N',N'-tetramethylethylenediamine were added per ml of the solution. Polymerization to a gel was carried out in glass tubes (0.5cm inner diameter) at 25 "C for 1 h. The spacer gel and the sample gel (pH 7.18, containing 30 pg of protein/ml) were prepared as described by Davis (17). Electrophoresis was performed at 25 "C and a current of 3 mA/tube. After electrophoresis, the gel was stained with Coomassie blue, rinsed with water, and destained. The destaining was found to be much less efficient in gels containing hydrophobic substituents. The position of the protein zone was generally determined by inspection with a Manostat caliper (k0.05 mm). For some gels, a densitometer trace was obtained.

Polyacrylamide is insoluble in organic solvents,
and it is therefore difficult to prepare acrylamide copolymers with hydrophobic monomers. We employed, therefore, N,N-dimethylacrylamide as the main monomer constituent, since its polymers are soluble both in water and in organic media.
Thus, it was possible to copolymerize in methanol or in benzene dimethylacrylamide with various acrylamides N-substituted with long chain alkyl groups, purify the copolymer, and entrap it in a poly(N,N-dimethylacrylamide) gel prepared from an aqueous solution of the monomer and a cross-linking reagent. Table I lists   Am-n stands for an acrylamide substituted with C,H2,,+ Other abbreviations used are: Am-10-COOH, N-acrylyl-11-amidoundecanoic acid; Am-05-10-COOH, N-acrylyl-6-amidocaproyl-11-amidoundecanoic acid; Am-05-18, N-acrylyl-6-amidocaproic acid N-octadecylamide.  the presence and the absence of the hydrophobic ligand. It may be seen that the protein peak is greatly broadened by the introduction of associating groups into the gel. In addition, lactoglobulin is split into two peaks.
In Fig. 3, we have plotted the ratio of the BSA mobilities uo and u, in the absence and presence of the hydrophobic groups in the gel, against the concentration c of the hydrophobe.
Theory predicts (10) uo/u = 1 + CK (1) where K is the association constant of the protein with the ligand group. It should be noted that this formulation is based on the assumption that the protein has a single binding site for the ligand; in the case of BSA, we are dealing with a protein with a number of sites exhibiting varying affinities for, for example, fatty acids. Nevertheless, we have obtained the linear plots consistent with the simple model, and the apparent association constants derived from their slopes characterize, presumably, the strongest binding site. Table I lists association constants obtained in this manner for BSA with our various gel substituents. The following points may be noted. 1) When comparing the binding to alkyl substituents of varying length, using entrapped linear copolymers with a similar content of the hydrophobic comonomer, K is found to increase sharply when the alkyl group is extended from tetradecyl to octadecyl. By contrast, Reynolds et al. (15) reported that the binding of alkyl sulfates to BSA is very similar for all ligands whose alkyl group is longer than decyl, while Spector et al. (14) reported binding of a 16-carbon fatty acid to be stronger than that of either a 14-or 18-carbon homolog. In our case, steric hindrance may restrict the approach of the protein to the backbone of the polymer chain. To check on  Table I. this possibility, we prepared a copolymer of N-acrylyl-&amidocaproic acid N-octadecylamide in which a spacer group separates the octadecyl group from the chain backbone. Table  I shows that the use of this spacer increased the apparent association constant 7-fold, approaching association constants observed for the binding of long alkyl chain anions (14, 15).
2) When linear copolymers of dimethylacrylamide with different contents of the same hydrophobic comonomer, Ntetradecylacrylamide, were entrapped in the cross-linked gel, the efficiency with which the hydrophobe reduced the serum albumin mobility was found to decrease with an increase of the hydrophobe content in the linear copolymer. We believe that this effect is due to an increasing tendency of the hydrophobic side chains to associate with each other, so that they become unavailable for protein binding.
3) When the alkyl side chain was terminated by a carboxyl group in copolymers of N-acrylyl-11-aminoundecanoic acid and N-acrylyl-6-amidocaproyl-11-amidoundecanoic acid, no significant retardation of BSA was observed. This is in striking contrast with the binding of fatty acids. It may be concluded that in fatty acid binding, the alkyl group is placed at the bottom of the cleft in the globular protein while the carboxyl which the carboxyl is at the end cannot be accommodated in the binding site.
The most striking result in our study of P-lactoglobulin A was the observation that this protein, which migrated as a single sharp peak in the control gel, separated in the presence of hydrophobes into a very slowly migrating fraction with a relatively sharp peak and a fraction with a higher mobility characterized by a diffuse peak. We found no previous record of such a fractionation. A f f i t y constants for the faster fraction were slightly higher than those for BSA when a tetradecyl group was the ligand while the longer ligands bound more weakly to P-LG-A. This protein was reported to have a single hydrophobic binding site which can accommodate only relatively small ligands (16).
The effect of hydrophobic substituents on the electrophoretic mobility of proteins is analogous to the effect of such substituents on protein behavior in chromatography. Phenomena of this type have been studied using agarose modified with a variety of hydrophobic substituents (19, 20). Proteinligand association constants have also been obtained from the retardation of a protein on a chromatographic column with the ligand attached to the stationary phase (21), and spacers are widely used to facilitate protein-ligand interaction in affinity chromatography (22). Both affinity electrophoresis (23) and affinity chromatography (24, 25) can resolve mixtures of similar proteins with slightly different ligand affkities. We see then that the two techniques have many similar characteristics. The advantage of affinity electrophoresis is in the very small sample size required and the wide variation of well defied continuous media which may be employed.