Protein±Protein Interactions in Concentrated Electrolyte Solutions
Total Page:16
File Type:pdf, Size:1020Kb
Protein±Protein Interactions in Concentrated Electrolyte Solutions Hofmeister-Series Effects R. A. Curtis,1 J. Ulrich,3 A. Montaser,1 J. M. Prausnitz,1,2 H. W. Blanch1* 1Chemical Engineering Department, University of California, Berkeley, California 94720; telephone:(510) 642-1387; fax:(510) 643-1228; e-mail:[email protected] 2Chemical Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, California 94720 3Automatic Control Laboratory, Swiss Federal Institute of Technology, ETH-Z, ETL I 22, CH-8092 Zurich, Switzerland Received 12 November 2000; accepted 4 July 2001 DOI:10.1002/bit.10342 Abstract: Protein±protein interactions were measured precipitate a target protein is dicult. Protein solubility for ovalbumin and for lysozyme in aqueous salt solu- is governed by many factors, including pH, surface tions. Protein±protein interactions are correlated with a proposed potential of mean force equal to the free en- hydrophobicity, surface-charge distribution, size, salt- ergy to desolvate the protein surface that is made inac- type, and salt concentration (Rothstein, 1994). The goal cessible to the solvent due to the protein±protein of this work is to understand how these factors in¯uence interaction. This energy is calculated from the surface protein solubility. The ®rst step is to determine the in- free energy of the protein that is determined from pro- teractions between the protein molecules, salt ions, and tein±salt preferential-interaction parameter measure- ments. In classical salting-out behavior, the protein±salt water; these interactions, described quantitatively, are preferential interaction is unfavorable. Because addition then used to predict conditions for optimal separation of of salt raises the surface free energy of the protein ac- a mixture of proteins. However, generating a phase cording to the surface-tension increment of the salt, diagram from knowledge of the intermolecular potentials protein±protein attraction increases, leading to a reduc- between the proteins, salt, and water is impossible with- tion in solubility. When the surface chemistry of proteins is altered by binding of a speci®c ion, salting-in is ob- out adopting physically realistic simplifying assump- served when the interactions between (kosmotrope) ion± tions. In general, most models of protein precipitation protein complexes are more repulsive than those be- are based on an eective protein±protein interaction tween the uncomplexed proteins. However, salting-out (Chiew et al., 1995; Fornaseiro et al., 1999; Malfois et al., is observed when interactions between (chaotrope) ion± 1996; Piazza, 1999; Poon, 1997; Rosenbaum et al., 1996) protein complexes are more attractive than those of the that is mediated by salt ions and water. At the present uncomplexed proteins. ã 2002 Wiley Periodicals, Inc. Bio- technol Bioeng 79: 367±380, 2002. time, eective protein±protein interactions in solutions Keywords: Hofmeister-series; protein interactions; hy- of concentrated electrolytes are not well understood. drophobic effect; protein salting-out Because protein solubility usually decreases with rising salt concentration (salting-out) (Cohn, 1943), we simi- larly expect that as salt concentration rises the eective INTRODUCTION protein±protein interactions become more attractive. However, it is not clear whether these interactions are Salt-induced precipitation/crystallization provides an related to solvation forces between proteins or possibly extensively used method in biotechnology for obtaining to speci®c short-range forces that stabilize protein crys- high-quality crystals and for separating target proteins tals, such as van der Waals contacts, salt bridges, or from multicomponent protein solutions as the ®rst pu- hydrogen bonds (Neal et al., 1999; Asthagiri et al., 1999). ri®cation step. However, because protein phase behavior In this work, we focus on determining these eective is not well understood, selecting optimum conditions to protein±protein interactions; from these we develop speci®c criteria for choosing conditions favorable for *Correspondence to: H. W. Blanch selective protein precipitation or protein crystallization. Contract grant sponsors: the Oce for Basic Energy Sciences of the U.S. Department of Energy, the National Science Foundation In concentrated salt solutions, protein solubility de- Contract grant number: CTS-9530793 pends on the anion's or cation's position in the lyotropic ã 2002 Wiley Periodicals, Inc. series (Hofmeister, 1888), but this dependence cannot be work to transfer a protein molecule from a hypothetical explained by considering salt ions as charged hard ideal gas into the aqueous salt solution. This work is the spheres. It is well known that salting-out eectiveness is protein solvation free energy; it is given by the product related to the water-structure-making or water-struc- of the surface free energy, r, and the solvent-accessible ture-breaking ability of the salt (Collins and Washa- surface area of the protein molecule A: baugh, 1985). To predict protein solubility, this eect needs to be incorporated into models for eective pro- h ig l2 l2 NavAr 1b tein±protein interactions. Presently, the only theories that account for these ion-speci®c eects are based on where Nav is Avogadro's number (Ben-Naim, 1978). the approach originally proposed by Melander and Substituting Eq. (1a) into (1b) and taking the derivative Horvath (1977), where protein solubility is determined with respect to molality of salt, m3, gives: by the preferential interactions between the protein and salt (Arakawa and Timashe, 1985). To incorporate d ln S2 dr RT NavA : 1c these theories into a statistical-mechanical model based dm3 dm3 on eective protein±protein interactions, we need to The problem of determining protein solubility as a formulate eective protein±protein interactions in terms function of salt concentration is reduced to solving for of protein±salt interactions. the surface free energy of the protein molecule. The For this reason, we summarize the salting-out theory dependence of the surface free energy of the protein on of Melander and Horvath and present a brief discussion salt molality can be determined from preferential inter- of protein±salt interactions. The following section dis- action-parameter measurements (Casassa and Eisen- cusses current models for eective protein±protein in- berg, 1964): teractions and proposes a simple form for the potential of mean force that can be used in protein phase equi- or o om 1 ol o A À 3 3 2 librium models for predicting protein solubility and for om3 T;p om2 T;l ;l om3 T;p determining conditions favorable for protein crystalli- o 1 3 o zation. Ion-speci®c eects are included in the model by Here, m3 and m2 are salt and protein molality and l1 relating the protein±protein potential of mean force to and l3 are the chemical potential of water and salt, re- the experimentally determinable protein±salt preferen- spectively. Superscript o denotes that the property is tial interaction parameter. PROTEIN±SALT INTERACTIONS Salting-Out Theory The important result of Melander and Horvath (1977) is that protein solubility can be expressed in terms of the solvation free energy of the protein molecule in the equilibrated ¯uid phase. When the solubility is small, we can neglect ¯uid-phase protein±protein interactions. The phase-equilibrium criterion (the chemical potential of the protein in the crystal is equal to that in the ¯uid) at a given salt concentration reduces to: S ls À lh RT ln 2 1a 2 2 h S2 s h where l2 and l2 are the chemical potential of the protein in the solid phase and the in®nite-dilution standard-state chemical potential of the protein, respectively, and S2/ h S2 is protein solubility relative to a standard state sol- ubility. The major assumption of the salting-out theory is that the protein crystal is a pure phase; in that event, the salting-out behavior is determined from the depen- dence of the standard-state chemical potential on salt concentration. The standard-state chemical potential can be further decomposed into a hypothetical ideal-gas contribution and a contribution from the reversible vorable interactions between the salt and the nonpolar water due to interaction with the peptide group, whereas surface of the protein and by favorable weak ion-bind- kosmotropic anions were not retarded due to unfavor- ing interactions between the salt and either the charged able interaction with the nonpolar backbone of poly- surface groups or the surface peptide groups of the acrylamide. protein. In most cases, because the unfavorable hydro- Anion binding to the positively charged surface phobic interactions are greater than the attractive weak groups of protein molecules has been observed in studies ion-binding interactions, salting-out is observed. concerning stabilization of folded structures of protein molecules at low pH (Goto et al., 1990). The strength of this interaction is related to the ion's position in the Surface-Tension-Increment Effect electroselectivity series (Gjerde et al., 1980), as measured The surface free energy of a nonpolar surface in contact by the anity of an ion for an anion-exchange resin. The with a solvent is related to the surface tension of the series depends on the resin employed, but the general 2) ) ) ) ) solvent. Because all salts increase the surface tension of trend is in the order SO4 > SCN >I >Br >Cl . water, they similarly increase the surface free energy of For monovalent anions the electroselectivity series is the