Microcalorimetric Studies on the Interaction ... - ACS Publications

Jul 18, 2001 - ... in cation-exchange chromatography of human monoclonal antibodies. Alexander Faude , Dörthe Zacher , Egbert Müller , Heiner Böttinge...
0 downloads 0 Views 110KB Size
Anal. Chem. 2001, 73, 3875-3883

Microcalorimetric Studies on the Interaction Mechanism between Proteins and Hydrophobic Solid Surfaces in Hydrophobic Interaction Chromatography: Effects of Salts, Hydrophobicity of the Sorbent, and Structure of the Protein Fu-Yung Lin,† Wen-Yih Chen,*,† and Milton T. W. Hearn*,‡

Department of Chemical Engineering, National Central University, Chung-Li, Taiwan 320, and Australian Centre for Research on Separation Science & Centre for Bioprocess Technology, Department of Biochemistry and Molecular Biology, Monash University, Clayton, Victoria, Australia 3800

This study examines the effects of different salts as well as the influence of the relative hydrophobicities of different sorbents on the adsorption processes of proteins in hydrophobic interaction chromatography (HIC). Comparative data acquired by the equilibrium binding analysis and by isothermal titration microcalorimetry (ITC) are presented. In particular, thermodynamic parameters, including the enthalpy changes, related to the interactions between several globular proteins and various Toyopearl 650 M sorbents under solvent conditions containing either 2.0 M ammonium sulfate or 2.0 M sodium sulfate at pH 7.0 and 298.15 K have been evaluated in terms of the molecular properties of these systems. The results reveal that the dependence of the free energy change, ∆Gads, for protein adsorption to HIC sorbents on the salt composition can be mainly attributed to the enthalpy changes associated with protein and sorbent dehydration and hydrophobic interactions. Differences in binding mechanisms between the n-butyl- and phenyl-HIC sorbents were evident. In the latter case, the participation of π-π hydrophobic interactions leads to significant differences in the associated enthalpy and entropy changes. Furthermore, an increase in the hydrophobicity of either the sorbent or the protein resulted in more negative values for the free energy change, which arose mostly from dehydration processes. Entropic effects favoring HIC adsorption increased with an increase in the exposed nonpolar surface area of the protein. Consequently, an increased contribution from the entropy change to the respective change in free energy occurs when HIC sorbents or proteins of higher hydrophobicity are employed, with these larger entropy changes consistent with a change in the interaction mechanism from a binding event dominated by adsorption to a partitioning-like process. Data extracted from the ITC measurements also provided insight into the interaction mechanisms that occur between proteins and hydrophobic solid surfaces, yielding information that can be applied to the HIC purification of 10.1021/ac0102056 CCC: $20.00 Published on Web 07/18/2001

© 2001 American Chemical Society

proteins according to the concept of critical hydrophobicity of the system and its thermodynamic consequences. Hydrophobic interaction chromatography (HIC) is a commonly used separation and purification method with proteins, permitting the resolution of these biomolecules according to their hydrophobicity differences. In recent years, many factors affecting hydrophobic interactions between a protein and immobilized nonpolar ligands have been investigated experimentally and theoretically.1,2 In this study, emphasis has been placed on the effects of two commonly used salts and the hydrophobicity of the HIC sorbents (ligand type and density) with several proteins employing equilibrium binding analysis and isothermal titration microcalorimetric (ITC) procedures. In previous investigations into the effects of salts, Horvath and co-workers3 demonstrated that, in the absence of special binding effects, an increase in the salt molality in the mobile phase or the use of a salt with a greater molal surface tension increment will result in the increased retention of proteins in HIC. The molal surface tension increment is a concentration-independent characteristic of the salt that relates to its water-structuring potential. Thus, the molal surface tension increments of salts have been reported to highly correlate with their position in the lyotropic series,4 which is an empirical relationship discovered by Hofmeister,5 linking the solution properties of salts and their effect on the solubility of proteins. * To whom correspondence should be addressed. W.-Y.C.: (tel) 886-3-4227151 ext 4222; (fax) 886-3-4225258; (e-mail) [email protected]. M.T.W.H.: (tel) 61+3+99 05 37 20: (fax) 62+3+ 99 05 58 82: (e-mail) milton.hearn@ med.monash.edu.au. † National Central University. ‡ Monash University. (1) Hearn, M. T. W. Physicochemical factors in polypetide and protein purification and analysis by high performance chromatographic techniques: Current status and challenges for the future. In Handbook of Bioseparation; Ahuja, S, Ed.; Academic Press: San Diego, 2000; pp 72235. (2) Hearn, M. T. W. Reversed phase and hydrophobic interaction chromatography of peptides and proteins. In HPLC of Biopolymers; Gooding, K., Regnier, F. E., Eds.; Marcel Dekker: New York, in press. (3) Melander, W. R. Corradini, D.; Horvath, Cs. J. Chromatogr. 1984, 317, 67. (4) Melander, W.; Horvath, Cs. Arch. Biochem. Biophys. 1977, 183, 200. (5) Hofmeister, F. Arch. Exp. Pathol. Pharmakol. 1888, 24, 247.

Analytical Chemistry, Vol. 73, No. 16, August 15, 2001 3875

An aqueous solution containing a salt with a higher value of molal surface tension increment has a higher molal surface tension, and this facilitates the expulsion of the protein from the solvent, leading to its adsorption/partition to a HIC sorbent.1-10 Other investigations with n-alkyl agarose sorbents revealed that the strength of the hydrophobic interactions increases with the hydrophobicity (ligand chain length, type, density) of the HIC sorbents.11-15 Compared to n-alkyl-based HIC sorbents, the interactions between proteins and HIC sorbents containing aromatic groups such as phenyl or benzyl ligands have followed less predictable binding behavior, and this has been attributed to the participation of π-π interactions.16-19 With aryl-based HIC sorbents, this type of interaction also involves the interplay of Lifschitz-London forces between π-electron-containing compounds such as aromatic rings, in which one moiety is electronrich (π-basic group or π-electron donor) and the other is electronpoor (π-acidic group or π-electron acceptor). In this context, two types of the forces can be defined, i.e., face-to-edge and face-toface (also called π-stacking) π-π interactions.20-22 These forces can be exploited in conjunction with hydrogen-bonding effects to separate similar biomolecules based on molecular recognition phenomena, such as those arising in HIC or with molecularly imprinted polymers.23-27 Although the contributions of hydrophobic forces involving π-π interactions were initially expected to result in higher binding strengths with proteins in HIC systems, the converse was obtained28 with the retention of proteins onto HIC sorbents such as the Toyopearl butyl sorbent stronger than that with the corresponding Toyopearl phenyl sorbent. The different interaction mechanisms evident with n-alkyl chain- and aromatic group-containing HIC sorbents have therefore been attributed to the consequences of the presence/absence of π-π interactions. Since comparative information on directly measured thermodynamic quantities, such as the enthalpy change, have not (6) El Rassi, Z.; Horvath, Cs. J. Liq. Chromatogr. 1986, 9, 3245. (7) Katti, A.; Horvath, Cs. Chromatographia 1987, 24, 646. (8) Lu, M.; Tjerneld, F. J. Chromatogr., A 1997, 766, 99. (9) Strop, P.; Cechova, D.; Tomasek, V. J. Chromatogr.1983, 259, 255. (10) Szepesy, L.; Horvath, Cs. Chromatographia 1998, 26, 13. (11) Lin, F.-Y.; Chen, W.-Y.; Ruaan, R.-C.; Huang, H.-M. J. Chromatogr., A 2000, 872, 37. (12) Haidacher, D.; Vailaya, A.; Horvath, Cs. Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 2290. (13) Gooding, D. L.; Schmuch, M. N.; Gooding, K. M. J. Chromatogr. 1984, 296, 107. (14) Schmuch, M. N.; Nowlan, M. P.; Gooding, K. M. J. Chromatogr. 1986, 371, 55. (15) Perkins, T. W.; Mak, D. S.; Root, T. W.; Lightfoot, E. N. J. Chromatogr., A 1997, 766, 1. (16) de Araujo, A. F. P.; Pochapsky, T. C.; Joughin, B. Biophys. J. 1999, 76, 2319. (17) Reubsaet, J. L. E.; Vieskar, R. J. Chromatogr., A 1999, 841, 147. (18) Reubsaet, J. L. E.; Jinno, K. Trend. Anal. Chem. 1998, 17, 157. (19) Tchapla, A.; Heron, S. J. Chromatogr., A 1994, 684, 175. (20) Brindle, R.; Albert, K. J. Chromatogr., A 1997, 757, 3. (21) Pirkle, W. H.; Liu, Y. J. Chromatogr., A 1996, 749, 19. (22) Jorgensen, W. L.; Severance, D. L. J. Am. Chem. Soc. 1990, 112, 4768. (23) Vlatakis, G.; Andersson, L. I.; Muller, R.; Mosbach, K. Nature 1993, 361, 645. (24) Whitcombe, M. J.; Rodriguez, M. E.; Villar, P.; Vulfson, E. N. J. Am. Chem. Soc. 1995, 117, 7105. (25) Kempe, M.; Mosbach, K. J. Chromatogr., A 1995, 694, 3. (26) Kriz, D.; Ramstrom, O.; Svensson, A.; Mosbach, K. Anal. Chem. 1995, 67, 2142. (27) Cong, Y.; Mosbach, K. J. Chromatogr., A 2000, 888, 63. (28) Ceccaroli, P.; Cardoni, P.; Buffalini, M.; De Bellis, R.; Piccoli, G.; Stocchi, V. J. Chromatogr., B 1997, 702, 41.

3876

Analytical Chemistry, Vol. 73, No. 16, August 15, 2001

been reported for such protein-HIC sorbent systems, an experimental strategy was developed in this investigation to further validate these considerations. By applying batch equilibrium binding procedures and directly measured adsorption enthalpies using ITC, the interaction processes between several globular proteins and various butyl- and phenyl-HIC sorbents have now been compared using the same mobile-phase conditions containing either 2.0 M (NH4)2SO4 or 2.0 M Na2SO4 at pH 7.0 and 298.15 K. The proteins used were hen egg white lysozyme (HEWL), horse heart myoglobin (HMYO), and bovine pancreas ribonuclease A (RNase A) due to their significantly different hydrophobicities yet similar molecular weights. Since protein adsorption/partition to hydrophobic sorbents is the net result of a number of processes, examination of the overall adsorption/partition as a series of additive subprocesses according to linear free energy additivity concepts was expected to facilitate the thermodynamic analysis.1,2,29-31 Consistent with other thermodynamic and microcalorimetric studies of proteins at liquid/ solid interfaces with immobilized metal ion affinity chromatography (IMAC), ion-exchange chromatography (IEC), and imprinting polymer (IMP) systems,1,2,11,32-38 the adsorption/partition of a protein to a HIC sorbent can be divided into five sequential subprocesses, e.g.: (a) water molecules or ions surrounding the protein surface are excluded, i.e., the dehydration or deionization (removing of the electrical double layer, EDL) subprocess of the protein; (b) water molecules or ions surrounding the HIC sorbent are excluded, i.e., the dehydration or deionization subprocess of the HIC sorbent; (c) the hydrophobic interactions between the protein and the hydrophobic HIC sorbent; (d) the structural rearrangement of the protein upon adsorption/partition; and (e) the structural rearrangement of the excluded water molecules or ions in the bulk solvent. This study applies information gained from equilibrium binding isotherms and directly measured adsorption enthalpies to examine the interaction thermodynamics associated with the protein adsorption/partition to several HIC sorbents of different ligand type and density under various salt conditions. The data, therefore, provide further insight into the reversible, equilibrium binding processes established between proteins and several hydrophobic sorbents commonly employed in large-scale HIC separation. EXPERIMENTAL PROCEDURES Materials. The physical and chemical properties of the Toyopearl 650 M sorbents used in this study have been characterized by a panel of biophysical procedures as part of our associated investigations. These semirigid, macroporous (g1000 Å) spherical sorbents of mean particle diameter of 65 µm were synthesized as hydrophilic vinyl copolymer composites and derivatized with the (29) Norde, W. Pure Appl. Chem. 1994, 66, 491. (30) Norde, W. J. Dispersion Sci. Technol. 1992, 13, 363. (31) Norde, W. Adv. Colloid Interface Sci. 1986, 25, 267. (32) Wu, C.-F.; Chen, W.-Y.; Lee, J.-F. J. Colloid Interface Sci. 1996, 183, 236. (33) Chen, W.-Y.; Wu, C.-F.; Tsao, H.-K. J. Colloid Interface Sci. 1997, 190, 49. (34) Lin, F.-Y.; Chen, W.-Y.; Sang, L.-C. J. Colloid Interface Sci. 1999, 214, 373. (35) Chen, W.-Y.; F.-Y. Lin; Wu, C.-F. In Interfacial Dynamics; Kallay, N., Ed.; Marcel Dekker: New York, 1999; p 419. (36) Huang, H.-M.; Lin, F.-Y.; Chen, W.-Y.; Ruaan, R.-C. J. Colloid Interface Sci. 2000, 229, 600. (37) Lin, F.-Y.; Chen, W.-Y. In The Encyclopedia of Solid Surfaces; Kellay, N., Ed.; Marcel Dekker: New York, in press. (38) Lin, F.-Y.; Chen, C.-S.; Chen, W.-Y.; Yamamoto, S. J. Chromatogr., A, in press.

appropriate n-alkyl or aryl chain moieties. Proteins were purchased from Sigma Chemical Co. (St. Louis, MO). All other chemicals were analytical grade and purchased from E. Merck (Darmstadt, Germany). The equilibrium buffer consisted of 50 mM sodium dihydrogen phosphate, containing either 2.0 M (NH4)2SO4 or 2.0 M Na2SO4 at pH 7.0 and 298.15 K. Equilibrium Binding Isotherms. Equilibrium binding isotherms for the various proteins were measured in the presence of 2.0 M (NH4)2SO4 or 2.0 M Na2SO4 at pH 7.0 and 298.15 K, using the procedure developed by Hutchens et al.39 and Anspach et al.40 The protein solution (1.0 mL), prepared in the equilibrium buffer at various concentrations, was added to 500 µL of a homogeneous suspension of the HIC sorbent (containing 5 µL of sorbent) in a 1.5-mL microcentrifuge tube. The protein-sorbent suspensions were equilibrated in microcentrifuge tubes by incubating at 10 rpm (Heraeas Microfuge A, Osterode, FGR) and 298.15 K for 4 h and then spun at 4000 rpm for 3 min. The concentrations of proteins remaining in free solution were determined by UV spectrometry at 280 nm, and the equilibrium concentrations of the bound protein on the sorbent were calculated from the material balance. Heat Measurements Using Isothermal Titration Microcalorimetry. Isothermal titration calorimetry is a heat conduction type microcalorimeter containing an “insertion” vessel within a 4-mL stainless steel ampule produced by Thermometric Ltd. (Jarfalla, Sweden). For the ITC measurements, the protein solution was titrated into the dispersed sorbent suspension through a Hamilton syringe fitted with a stainless steel needle driven by a computer-controlled pump over a time interval of 30 min. The output signal was collected as a power versus time profile, integrated, and quantified by the amount of bound protein to give the enthalpy change of adsorption (∆Hads). The apparent heat generated from the titration must be corrected for the dilution heat of the protein and the HIC sorbent, to obtain the net heat of interaction between the protein and the sorbent and thus ∆Hads by calculation from the following equation:

Qads ) Vq*∆Hads

where Qads (in Joules) is the net heat attributed to the interactions between the protein and the HIC sorbent, corrected by subtracting the dilution heat of the protein and the sorbent measured under the same condition; V (mL) represents the volume of the sorbent in the ampule; and q* (mol/mL) is the amount of bound protein, which can be evaluated from the corresponding isotherm. Prior to each experiment, the ampule and the stirrer were washed with water and acetone and were then dried in air. In this investigation, the sorbent was suspended in 50 mM sodium dihydrogen phosphate buffer, pH 7.0, under various conditions. A 3-mL sorbent suspension, including 0.1 mL of sorbent, was placed in the ampule with stirring at 100 rpm. After thermal equilibrium had been reached, a 30-µL aliquot of the protein solution prepared in the same 50 mM sodium dihydrogen phosphate buffer, pH 7.0, was titrated into the dispersed sorbent suspension. All experiments were performed at 298.15 K. (39) Hutchens, T. W.; Yip, T.-T.; Porath, J. Anal. Chem. 1988, 170, 168. (40) Anspach, F. B.; Johnston, A.; Wirth, H. J.; Hearn, M. T. W. J. Chromatogr. 1989, 476, 205.

Figure 1. Isotherms measured under batch equilibrium conditions for HEWL with either the Toyopearl butyl or phenyl sorbent (with a constant ligand density) in the presence of either 2.0 M (NH4)2SO4 or 2.0 M Na2SO4, pH 7.0, and 298.15 K.

RESULTS AND DISCUSSION Equilibrium Binding Analysis. (1) Effects of Salts. All equilibrium binding isotherms arising from this study have been fitted to the Langmuir equation to evaluate the protein binding affinity and the average maximum adsorption capacity of the various HIC sorbents. The isotherms measured under batch equilibrium conditions for HEWL with either the Toyopearl butylor phenyl-HIC sorbents (with a constant ligand density), with either 2.0 M (NH4)2SO4 or 2.0 M Na2SO4 at pH 7.0 and 298.15 K, are shown in Figure 1. Higher affinities were obtained in the presence of 2.0 M Na2SO4 compared to 2.0 M (NH4)2SO4 for both types of HIC sorbents, indicating that sodium sulfate has a greater ability to enhance hydrophobic interactions between HEWL and the HIC sorbents. This observation is consistent with previously reported data4 documenting that Na2SO4 has a larger molal surface tension increment than (NH4)2SO4 (2.73 × 10-3 compared to 2.16 × 10-3 dyn‚g‚cm-1‚mol-1 respectively4,41). This result also agrees the conclusion of Horvath and co-workers3 that use of a salt with a greater molal surface tension increment will result in increased retention of proteins to HIC sorbents. As a consequence, the solutions containing 2.0 M Na2SO4 have higher surface tensions, providing solvophobically more preferred environments for protein adsorption onto the hydrophobic solid surfaces, and leading to a higher affinity constant. Notably, the average maximum binding capacities for a particular protein with the Toyopearl butyl or phenyl sorbents appear to be similar with the different salts, presumably due the similar density of binding sites on both sorbents. (2) Effects of Ligand Type. The results shown in Figure 1 demonstrate that the adsorption affinity of HEWL with the Toyopearl butyl sorbents was larger than that found with the Toyopearl phenyl sorbent. Since the Toyopearl butyl and phenyl sorbents have similar ligand densities, this observation reflects differences in the strength of the hydrophobic interactions. Although the Toyopearl phenyl sorbent has the opportunity to bind to exposed aromatic side-chain groups on the surface of HEWL through interactions based on induced dipole and π-π interactions,16-19 this does not, however, mean that a higher affinity (41) Vailaya, A.; Horvath, Cs. Ind. Eng. Chem. Res. 1996, 35, 2964.

Analytical Chemistry, Vol. 73, No. 16, August 15, 2001

3877

Table 1. Ligand Densities and Surface Microdensities of the Various Toyopearl Butyl Sorbents Used in These Investigations with, e.g., Ribonuclease Aa Toyopearl butyl sorbents

surface microdensity (µmol/m2)

average ligand density (mmol/m2)

1 2 3

12.96 ( 1.44 5.29 ( 0.38 2.67 ( 0.21

3.16 ( 0.35 3.78 ( 0.27 2.54 ( 0.2

aData

adapted from ref 59.

will necessarily be obtained for the Toyopearl phenyl sorbent compared to the Toyopearl butyl sorbents with HEWL, as shown in Figure 1. This result is in agreement with the observation of Stocchi and co-workers,28 who demonstrated that the Toyopearl 650 M butyl sorbent has a higher retention for proteins than the Toyopearl 650 M phenyl sorbent. Consequently, the Toyopearl 650 M phenyl sorbent may be a better choice for the separation of conformationally promiscuous proteins due to its intermediate hydrophobicity, which allows milder elution conditions to be employed. The binding affinity associated with the adsorption process, i.e., the free energy change, can be attributed to the interplay of both enthalpic and entropic effects. The higher affinity of HEWL with the Toyopearl butyl sorbents cannot be accounted for solely in terms of the enthalpy change of the subprocess due to hydrophobic interactions, but that the entropy change associated with the hydrophobic interactions as well as enthalpy/entropy changes arising from desolvation must also be included. Further support for this conclusion was obtained from the ITC studies discussed later. (3) Effect of Hydrophobicity of the HIC Sorbents and Proteins. The effects of the hydrophobicity of the HIC sorbents and proteins on the isothermal adsorption behavior between proteins and hydrophobic solid surfaces were investigated with RNase A either by varying the sorbent ligand density (surface microdensity) or by using proteins of different exposed hydrophobic surface areas with the Toyopearl butyl sorbents in the presence of 2.0 M (NH4)2SO4, pH 7.0, and 298.15 K. Table 1 lists the relevant data on the surface microdensity and ligand density of the Toyopearl butyl sorbents used with RNase A as the protein probe. Since earlier results obtained by Hearn and co-workers have shown that the contact areas established between proteins and HIC sorbents increase with the surface microdensity, this quantity provides a reliable and direct parameter to assess changes in the molecular basis of interactions between hydrophobic sorbents and proteins. Accordingly, we examined the effects of the relative hydrophobicities of the various Toyopearl butyl sorbents on the interaction behavior of RNase A in terms of equilibrium binding isotherms. As shown in Figure 2, the result demonstrates that the adsorption affinity of RNase A increased with surface microdensity of the Toyopearl butyl sorbents, while the average maximum binding capacity of RNase A was found to be dependent on the ligand density of the HIC sorbent. This finding can be explained by the fact that, at a higher surface microdensity, a stronger multisite interaction occurs between the HIC sorbent and RNase A, thereby resulting in a larger adsorption affinity. Moreover, the increase in the binding capacity of RNase A with 3878 Analytical Chemistry, Vol. 73, No. 16, August 15, 2001

Figure 2. Isotherms measured under batch equilibrium conditions for RNase A with Toyopearl butyl sorbents of different ligand densities in the presence of either 2.0 M (NH4)2SO4 or 2.0 M Na2SO4, pH 7.0, and 298.15 K. Table 2. Physicochemical Properties of Proteins Used in This Studya protein

code

mol wt

∆A (Å2)

pI

ribonuclease A RNase A 13 674 5802 8.8 chick lysozyme HEWL 14 295 7024 11 horse muscle myoglobin HMYO 16 935 8884 7.3 a

net charge at pH 7.0 4.2 8.0 1.0

Data adapted from ref 59.

HIC sorbents of higher ligand density probably reflects a surface reorganization of the adsorbed protein, analogous to that observed in IMAC42,43 or in the biospecific affinity chromatography44-46 of proteins with sorbents of different ligand densities. Support for this conclusion can also be found in our previous studies with HIC and RPC sorbents.1,2,11,36 Since an increase in the magnitude of the exposed hydrophobic surface area of the protein is anticipated to cause a similar result in terms of enhancement of the binding strength as increasing the surface microdensity of the HIC sorbent, investigations were carried out using proteins with different exposed hydrophobic surface areas (Table 2). Figure 3 shows the influence of hydrophobicity of proteins on the adsorption affinity and averaged maximum binding capacity with the Toyopearl butyl sorbents using 2.0 M (NH4)2SO4, pH 7.0, and 298.15 K. The results reveal that the binding affinity increased with the exposed hydrophobic surface area of the proteins and correlated with the size of the molecule as shown in Table 3, whereas the binding capacity was found to be dependent on the molecular weight of the proteins. These observations agree with other data7,10,12,41 showing that proteins with larger hydrophobicities tend to have longer retention times with hydrophobic interaction sorbents under the same elution conditions. Thus, the isothermal measurement data indicate that an increase in the adsorption affinity of proteins with hydrophobic sorbents can be (42) Wirth, H. J.; Unger, K. K.; Hearn, M. T. W. Anal. Biochem. 1993, 208, 16. (43) Anspach, F. B.; Wirth, H. J.; Unger, K. K.; Stanton, P. G.; Davies, J. D.; Hearn, M. T. W. Anal. Biochem. 1989, 179, 171. (44) Andrade, J. D. In Surface and Interfacial Aspects of Biomedical Polymers; Andrade, J. D., Ed.; Plenum: New York, 1985; Vol. 2, pp 1-62. (45) Janzen, R.; Unger, K. K.; Geische, H.; Kinchel, J. N.; Hearn, M. T. W. J. Chromatogr. 1987, 397, 91. (46) Wirth, H. J.; Unger, K. K.; Hearn, M. T. W. J. Chromatogr. 1991, 550, 383.

Figure 3. Isotherms measured in batch equilibrium conditions for various proteins with the Toyopearl butyl sorbent in the presence of either 2.0 M (NH4)2SO4 or 2.0 M Na2SO4, pH 7.0, and 298.15 K.

Figure 4. Adsorption enthalpy (∆Hads) of HEWL with either the Toyopearl butyl or phenyl sorbent in the presence of either 2.0 M (NH4)2SO4 or 2.0 M Na2SO4, pH 7.0, and 298.15 K.

Table 3. ∆Gads, δHads, and δSads Values of the Toyopearl Butyl or Phenyl Sorbent with HEWL in the Presence of Either 2.0 M (NH4)2SO4 or 2.0 M Na2SO4, pH 7.0, Determined from the Equilibrium Binding Isotherms and Initial Point of the Enthalpy Change Data ∆Gads (kJ/mol) salt butyl phenyl

∆Hads (kJ/mol)

T∆Sads (kJ/mol)

(NH4)2SO4 Na2SO4 (NH4)2SO4 Na2SO4 (NH4)2SO4 Na2SO4 -20.34 -20.04

-23.39 -21.16

-3.70 -8.21

-7.56 -9.69

16.64 11.83

15.83 11.47

achieved either by increasing the hydrophobicity of the proteins, by varying the ligand type, or by increasing the ligand concentration (surface microdensity) of the HIC sorbent. On the other hand, the average maximum binding capacity of proteins was dependent on the density of the HIC sorbent and the size of the protein molecule. Microcalorimetric Measurements. (1) Effects of Salts. As noted above, the measured thermodynamic quantities, such as enthalpy changes, for a reversible, equilibrium interaction between proteins and hydrophobic solid surfaces can be described in terms of at least five sequential subprocesses.1,2,11,32-38 In this study, the effect of salts on the change of enthalpy was found to be mainly associated with the dehydration subprocesses (i.e., the a and b subprocesses described above) and hydrophobic interactions (i.e., the c subprocess). The adsorption enthalpy (∆Hads) of HEWL with the Toyopearl butyl and phenyl sorbents with either 2.0 M (NH 4)2SO4 or 2.0 M Na2SO4 at pH 7.0 and 298.15 K are presented in Figure 4. The results show that lower values of ∆Hads were obtained in the presence of 2.0 M Na2SO4. Two phenomena contribute to this observation. First, sodium sulfate has a greater ability to decrease the endothermic component of the dehydration of the protein and HIC sorbent as well as to reduce the EDL (i.e., the a and b subprocesses) when compared to ammonium sulfate. Since sodium sulfate has a higher molal surface tension increment,47 a more solvophobic environment will occur for HEWL (and the other proteins) in an aqueous solution containing 2.0 M Na2(47) International Critical Tables; McGraw-Hill: New York, 1929; Vol. 4.

Figure 5. Adsorption entropy (∆Sads) of HEWL with either the Toyopearl butyl or phenyl sorbent in the presence of either 2.0 M (NH4)2SO4 or 2.0 M Na2SO4, pH 7.0, and 298.15 K, determined by fitting the equilibrium Langmuir isotherms.

SO4 rather than 2.0 M (NH4)2SO4. A 2.0 M Na2SO4 condition will favor a reduction in the energy (endothermic) required for protein dehydration prior to adsorption because of greater exposure of hydrophobic segment(s) of the protein. The initial condition of state for HEWL (or the other proteins) will differ with different salts at the same concentration in buffers of the same composition, pH value, and temperature, and this state of matter difference will be reflected as different values of the enthalpy of dehydration, ∆Hdehydr. Second, compared to the 2.0 M (NH4)2SO4 condition, the interaction between HEWL and the HIC sorbent was enhanced by the use of 2.0 M Na2SO4 with an increase in the exothermic amount of heat (i.e., the c subprocess). Although the heat required for structural rearrangement, i.e., ∆Hstruct, for HEWL will not be the same for different salts, this contribution to the overall free energy change was small. This conclusion is consistent with the fact that proteins generally show lower binding strengths and less denaturation with HIC sorbents compared to reversed-phase or some affinity chromatographic systems. As shown in Figure 5, changes in the adsorption entropy, ∆Sads, of HEWL with the Toyopearl butyl or phenyl sorbents in the presence of either 2.0 M (NH4)2SO4 or 2.0 M Na2SO4 at pH 7.0 and 298.15 K were relatively small, indicating that these two Analytical Chemistry, Vol. 73, No. 16, August 15, 2001

3879

kosmotropic salts influenced the enthalpy change more significantly than the entropy change. The entropy term associated with the adsorption/partition process with HEWL was not significantly affected by the choice of the salts used in this study. However, it can be noted from the ∆Hads and T∆Sads values determined from the initial point of the ITC data (Table 3) that the T∆Sads term overall does play an important role even under exothermic binding conditions (negative enthalpy change values), since the T∆Sads term contributed more than half of the total free energy change of the process. This result provides evidence to support the conclusion12,46,48,49 that chromatographic processes based on the hydrophobic interactions are mostly driven by the entropy gain. (2) Effects of Ligand Type. The results shown in Figure 4 and Table 3 demonstrate that the ∆Hads values for HEWL with the Toyopearl butyl sorbent were higher than that with the Toyopearl phenyl sorbent in the presence of either 2.0 M (NH4)2SO4 or 2.0 M Na2SO4 at pH 7.0 and 298.15 K. This finding can be explained by the interplay of two effects. First, the endothermic amount of the heat associated with the dehydration and the reduction in the EDL (i.e., the a and b subprocesses) is smaller when the Toyopearl butyl sorbent is used in comparison to the Toyopearl phenyl sorbent. This conclusion is consistent with reported data,50,51 showing that the enthalpy change for the hydration of butane measured by calorimetry from the gas to liquid phase was -23.63 kJ/mol, while that value for phenol was -54.65 kJ/mol. Second, the binding strength associated with the hydrophobic interactions of HEWL with the Toyopearl phenyl sorbent was stronger than that with the Toyopearl butyl sorbent, leading to an increase in the exothermic amount of heat (i.e., the c subprocess). This outcome is probably a consequence of the ability of the Toyopearl phenyl sorbent to provide additional π-π interactions with HEWL, resulting in a larger binding strength. van’t Hoff analysis has also revealed16 that the enthalpy changes for the binding of tryptophan and phenylalanine derivatives to a phenylalanine immobilized sorbent were -37.64 and -30.11 kJ/ mol, respectively, while aliphatic-aliphatic interactions, such as the binding of valine, leucine, and isoleucine derivatives with isoleucine immobilized sorbents, were found to have less negative values for the enthalpy change.16 As evident from Figure 5 and Table 3, the adsorption entropy (∆Sads) of HEWL with the Toyopearl butyl sorbents with either 2.0 M (NH4)2SO4 or 2.0 M Na2SO4 was larger than that with the Toyopearl phenyl sorbent. This result is consistent with the observations of de Araujo and co-workers,16 who concluded that the interactions of aromatic amino acid residues with immobilized phenylalanine sorbents are entropically less favorable than the interactions of aliphatic residues but that the longer retention of compounds containing aromatic residues is due to more favorable enthalpy change. As a consequence, the data obtained in the present study imply that the adsorption processes occurring for HEWL with the Toyopearl butyl and phenyl sorbents are mediated by significant differences in hydration/dehydration and the nature of the hydrophobic interaction processes. This finding is similar (48) Cole, L. A.; Dorsey, J. G. Anal. Chem. 1992, 64, 1317. (49) Dill, K. A. J. Phys. Chem. 1987, 91, 1980. (50) Dec, S. F.; Gill, S. J. J. Solution Chem. 1984, 13, 27. (51) Cabani, S.; Gianni, P.; Mollica, V.; Lepori, L. J. Solution Chem. 1981, 10, 563.

3880 Analytical Chemistry, Vol. 73, No. 16, August 15, 2001

Figure 6. Adsorption enthalpy (∆Hads) of RNase A with Toyopearl butyl sorbents containing different surface microdensities in the presence of 2.0 M (NH4)2SO4, pH 7.0, and 298.15 K. Table 4. δGads, ∆Hads, and ∆Sads Values for Several Toyopearl Butyl Sorbents with Ribonuclease A in the Presence of Either 2.0 M (NH4)2SO4 or 2.0 M Na2SO4, pH 7.0, Determined from the Equilibrium Binding Isotherms and Initial Point of the Enthalpy Change Data Toyopearl butyl sorbent

∆Gads (kJ/mol)

∆Hads (kJ/mol)

T∆Sads (kJ/mol)

1 2 3

-20.86 -19.72 -18.83

0.41 -5.19 -8.64

20.45 14.54 9.19

to the results obtained by Makhatadze and Privalov52 for the solvation of aromatic hydrocarbons, whereby the enthalpic contribution predominates over the entropic contribution in the Gibbs energy of hydration. Furthermore, the different sign for the hydration Gibbs energies determined by Makhatadze and Privalov indicates that the processes mediated by the hydrophobicity of aromatic hydrocarbons has a nature different from those mediated by the hydrophobicity of aliphatic hydrocarbons. (3) Effects of Surface Microdensity of the HIC Sorbents. As discussed above, the affinity of HEWL increased with the surface microdensity of the HIC sorbents; that is, a larger reduction in the overall free energy occurred with HIC sorbents of higher hydrophobicity. With the available range of different HIC sorbents, it was possible to delineate from the measured thermodynamic data the nature of the contribution of hydrophobic interaction processes established between the protein and nonpolar solid surfaces. This analysis of the relationship between ∆Hads and the surface microdensity of the HIC sorbents emphasized again the importance of the dehydration processes for the hydrophobic ligands of the sorbent and the protein (i.e., the a and b subprocesses) and the hydrophobic interactions between proteins and the HIC sorbent (i.e., the c subprocess). Figure 6 and Table 4 show the ∆Hads values for the adsorption of RNase A to Toyopearl butyl sorbents of different surface microdensity in the presence of 2.0 M (NH4)2SO4, pH 7.0, and 298.15 K. The trend for the ∆Hads values of proteins to increase with the surface microdensity of the HIC sorbent appears to have its origin in one (52) Makhatadze, G. I.; Privalove, P. I. J. Biophys. Chem. 1994, 50, 285.

Table 5. δGads, δHads, and ∆Sads Values for Several Different Proteins with the Toyopearl Butyl Sorbent in the Presence of Either 2.0 M (NH4)2SO4 or 2.0 M Na2SO4, pH 7.0, Determined from the Equilibrium Binding Isotherms and Initial Point of the Enthalpy Change Data proteins

∆Gads (kJ/mol)

∆Hads (kJ/mol)

T∆Sads (kJ/mol)

RNase A HEWL HMYO

-19.72 -20.34 -22.61

-5.19 -3.70 -1.32

14.54 16.64 20.29

Figure 7. Adsorption entropy (∆Sads) of RNase A with Toyopearl butyl sorbents containing different surface microdensities in the presence of 2.0 M (NH4)2SO4, pH 7.0, and 298.15 K, determined by fitting the equilibrium Langmuir isotherms.

or both of the following effects: (i) a reduction in the endothermic heat of dehydration occurs with HIC sorbents of lower surface microdensity, with the heat per unit mass required for dehydration of the sorbent increased as the surface microdensity is raised; and (ii) a decrease in the exothermic heat of hydrophobic interaction occurs with HIC sorbents of higher surface microdensity. Since the ∆Hads values increased with surface microdensity of the HIC sorbent, the incremental free energy changes associated with hydrophobic binding and surface-mediated reorganization of RNase A per se may not play a dominant role in the adsorption process. Rather, the major contributor to the free energy change appears to be the heat required for protein and sorbent dehydration. Furthermore, the ∆Sads values of RNase A increased with surface microdensity of the HIC sorbent as shown in Figure 7 and Table 4, indicating that the contribution of the dehydration subprocess to the entropy gain increased with hydrophobicity of the HIC sorbents. This result agrees with the study of Cole and Dorsey,48 who found that the enthalpy and entropy values associated with the interaction of benzene with immobilized C18 ligands calculated from van’t Hoff plots increased with ligand density. These observations led the investigators to the conclusion that the interaction behavior of solutes with more hydrophobic sorbents is dominated by a partition mechanism. By analogy, an increasing contribution from the entropy change to the corresponding free energy change when HIC sorbents of higher surface microdensity are employed can be considered as evidence for a change in the interaction mechanism from a binding process dominated by an adsorption phenomenon to a partitioninglike process. As can be seen from Table 4, the ∆Hads and T∆Sads values increased with the surface microdensity of the HIC sorbents, with the Toyopearl butyl-1 sorbent exhibiting a positive value of ∆Hads and a much larger entropy change, consistent with a partition-dominated HIC process. This conclusion has parallels in our previous studies11,53 that showed that the interaction mechanism of proteins with n-octyl- or n-octadecyl-containing sorbents is more partition-like while with n-butyl immobilized sorbents adsorption process dominated. (4) Effects of the Exposed Hydrophobic Surface Area of the Protein. Several proteins (Table 5) were used to examine the effect of increases in protein hydrophobicity on the interaction

Figure 8. Adsorption enthalpy (∆Hads) of various proteins with the Toyopearl butyl sorbent in the presence of 2.0 M (NH4)2SO4, pH 7.0, and 298.15 K.

thermodynamics with HIC sorbents. The results confirmed that as the hydrophobicity of the protein increased, more negative values for the change in free energy occurred, attributable mainly to the entropy gain. As shown in Figure 8 and Table 5, the ∆Hads values also increased with the hydrophobicity of the proteins with 2.0 M (NH4)2SO4, pH 7.0, and 298.15 K. This result is similar to the findings described above whereby the ∆Hads values increased with the surface microdensity of the HIC sorbent. Thus, these results indicate that the major contributor to the free energy changes with different proteins and the same HIC sorbent was the heat required for the dehydration of the protein. This effect was associated with an increase in ∆Sads to more positive values when the hydrophobicity of proteins was increased as shown in Figure 9 and Table 5. This observation also agrees with the findings of Horvath and co-workers,7,10,12,41 who demonstrated that protein retention to silica-based hydrophobic interaction chromatographic columns increased with increased exposure of the hydrophobic surface area of the protein. Furthermore, results obtained from van’t Hoff plot analysis of the retention data of dansyl amino acids on n-alkyl silica sorbents revealed41 that the values of enthalpy and entropy change become more positive with the hydrophobicity of the amino acid. The results from the present investigation, as well as these earlier studies, collectively lead to the important conclusion that the entropic effect favoring the HIC process increases with the magnitude of the nonpolar surface area of the protein. Since the interaction of a protein with immobilized nonpolar ligands involves a loss of entropy, the protein(53) Purcell, A. W.; Aguilar, M. I.; Hearn, M. T. W. Anal. Chem. 1999, 71, 2440.

Analytical Chemistry, Vol. 73, No. 16, August 15, 2001

3881

Figure 9. Adsorption entropy (∆Sads) of various proteins with the Toyopearl butyl sorbent in the presence of 2.0 M (NH4)2SO4, pH 7.0, and 298.15 K, determined by fitting the equilibrium Langmuir isotherms.

hydrophobic ligand complex will only survive if its formation is favored on enthalpic grounds. This loss of entropy is associated with the conversion of free translational and rotational degree of freedom into constrained motions. Many biomolecular interactions involving proteins and small drug ligands, or alternatively proteins with membrane-bound receptors or even with the phospholipid bilayer itself, have negative values for the standard entropy change, with values typically54 in the range of -60 kJ/mol e T∆Sads e -5 kJ/mol. Since the T∆Sads values for the adsorption of the various proteins onto the different HIC sorbents were in each case positive, i.e., a gain in entropy occurred, this finding indicates that the system overall became more disorganized with increased degrees of freedom but was compensated by the corresponding changes in ∆Hads, resulting overall in a favorable negative value for the free energy change, ∆Gads, for proteinnonpolar ligand complex formation. The role of the entropy/ enthalpy compensation effect in reversible equilibrium interactions between proteins and hydrophobic interaction and reversed-phase sorbents has been the discussion of detailed examination recently.55-58 In particular, these studies have documented the interplay between the entropy changes associated with protein and these studies have documented the interplay between the entropy changes associated with protein and ligand desolvation, ∆Sdesolv; protein and ligand structural reorganization, ∆Sstruct; and protein and ligand interaction, ∆Sint. According to the linear free energy additivity concepts, this entropic dependency can overall be expressed as

∆Sads ) ∆Sdesolv + ∆Sstruct + ∆Sint Because the interaction of proteins with HIC sorbents and solvent conditions tends to be “softer” than with reversed-phase (54) Ben-Tal, N.; Honig, B.; Bagdassarian, C. K.; Ben-Shaul, A. Biophys. J. 2000, 79, 1180. (55) Dill, K. A. Biochemistry 1990, 29, 7133. (56) Boysen, R. I.; Jong, A. J. O.; Hearn, M. T. W. Biochemistry, in press. (57) Gallicchio, E.; Kubo, M. M.; Levy, R. M. J. Am. Chem. Soc. 1998, 120, 4526. (58) Nichol, L. W.; Winzor, D. J. J. Theor. Biol. 1985, 117, 597. (59) Wang, Y.; Boysen, R. I.; Hearn, M. T. W., unpublished observations, 2001.

3882

Analytical Chemistry, Vol. 73, No. 16, August 15, 2001

sorbents and the typical low-pH, aquo-organic solvent mobilephase conditions, i.e., the binding tends to have a shallower binding potential that perturbs the native structure of the protein to a smaller degree, the ∆Sstruct term is expected to be either negative or only slightly positive for many protein-HIC sorbent complexes. The dominant entropy term leading to the observed large positive values of ∆Sads must be associated with the entropy changes for protein and ligand desolvation, ∆Sdesolv, with the interaction entropy, ∆Sint, playing a minor, but not negligible, role in the adsorption process. For desorption of the protein from the HIC sorbent, the converse situation must occur, with a larger negative value achieved for the entropy of structured solvation, ∆Ssolv, whereby the translational and rotational degrees of freedom of the surrounding solvent molecules are reduced on binding to the protein and the hydrophobicity ligands of the HIC sorbent. Since the magnitude of the ∆Sads increases as more degrees of freedom are restricted in the protein-nonpolar ligand complex, as the interaction becomes less “soft” and tighter binding occurs, then larger values for the interaction entropy will arise, resulting in a critical transition value beyond which desorption of the protein in its native state cannot be achieved. This transition point corresponds to the critical hydrophobicity criterion of the particular protein-HIC sorbent system.1,2,57 The above ITC measurements also provide a means to analyze the thermodynamic quantities as a function of the amount of protein bound. As shown in Figure 8, the ∆Hads values were found to increase with the amount of bound protein using 2.0 M (NH4)2SO4, pH 7.0, and 298.15 K. At least three possibilities need to be considered to account for these observations. First, adjacent protein molecules adsorbed on the surface of the HIC sorbent are anticipated to progressively interact more significantly with each other due to their reduced separation distance as the amount of bound protein increases. Protein-protein interaction will thus become enhanced due to self-association or isodesmic phenomena. Additional heat will be required to overcome this unfavorable effect. Second, the number of hydrophobic binding sites on the HIC sorbent will be stochastically reduced by the progressive increase in the number of bound protein molecules. This effect will promote steric hindrance between the protein and the remaining binding sites on the HIC sorbent. Third, the heat required for dehydration of the HIC sorbent (i.e., the b subprocess) will reduce as the amount of bound protein increases. A possible cause of this latter effect is associated with the initial stabilization by the hydrogen bonding of the water molecules that partition around the hydrophobic surface of the HIC sorbent. Since the water structure surrounding the surface of the sorbent will become more disordered during the dehydration process, the heat required for the dehydration of the HIC sorbent will reduce as the amount of bound protein increases. From the experimental data it is, however, apparent that the ∆Hads values increased with the amount of bound protein. Thus, the energy terms associated with the stochastic loss of binding sites or the dehydration of the HIC sorbent do not represent the thermodynamically dominant process with these proteins. In the case of HEWL, in particular, protein-protein interaction appears to be the significant contributor to the observed results. Previously, HEWL was shown to readily undergo isodesmic self-association in bulk solution under “salting out” conditions,58 and a similar phenomenon appears to

prevail in the case of interactions with HIC sorbents. The variation in ∆Hads for HEWL with the HIC sorbents was larger than for RNase A or HMYO as the amount of bound protein was increased, implying that the extent of protein-protein interaction is different due to the intrinsic binding characteristics of these proteins. The dilution heat of HEWL obtained by the above ITC measurements also supports this conclusion, demonstrating that the dilution heat of HEWL was more exothermal than that of RNase A or HMYO. This result implies that the proteinprotein repulsion between HEWL molecules due to electrostatic effects is much stronger than with the other proteins but that selfself association can still dominate due to hydrophobic interactions. This conclusion is also supported by our previous study11 that stronger protein-protein repulsion, as measured by the dilution heats, leads to a steeper increment in the values of the enthalpy change. Consequently, the heat required to overcome the repulsive interaction of molecules becomes greater for HEWL molecules as the amount of bound protein increases in comparison to RNase A or HMYO.

sorbents was characterized by the presence/absence of π-π hydrophobic interactions, leading to significantly different values of the associated enthalpy and entropy changes. Furthermore, an increment in hydrophobicity of either the HIC sorbent or the protein results in lower values of free energy change, which is contributed mostly from the dehydration processes, indicating that entropic effect favoring the HIC processes increases with the magnitude of the nonpolar surface area of the protein. As a consequence, an increased contribution of the entropy change to the respective free energy change with HIC sorbents of higher surface microdensity or alternatively with proteins containing larger exposed surface areas provides evidence for a change in the interaction process from an adsorption-dominated binding event to a partitioning-like process. Thus, the isothermal binding and ITC results obtained in this study not only support the conclusion that the hydrophobic interaction-based chromatographic processes are mostly driven by entropy gain but also provided an alternative way to quantitatively evaluate protein interactions with HIC systems.

CONCLUSIONS This investigation has demonstrated that an increase in the adsorption affinity of proteins with hydrophobic interaction chromatographic sorbents arises with an increase in the hydrophobicity of the proteins and can be manipulated by varying the ligand type or density (surface microdensity) of the HIC sorbents. However, the average maximum binding capacity of proteins was dependent on the ligand density of the HIC sorbent and the size of the protein molecule. The effects of salts on the free energy change were found to mostly attribute to the enthalpy change that is associated with the dehydration and hydrophobic interaction between proteins and the hydrophobic surfaces. The difference in binding processes between the Toyopearl butyl and phenyl

ACKNOWLEDGMENT These studies were supported by the Australian Research Council and the National Science Council of R.O.C. under Contract NSC 87-2214-E-008-019. These investigations were carried out during the tenure of a Visiting Professorship in the Centre for Bioprocess Technology, Monash University, awarded to W.-Y.C. and supported by the Australian Academy of Science and the Australian Academy of Technological Sciences and Engineering.

Received for review February 19, 2001. Accepted May 9, 2001. AC0102056

Analytical Chemistry, Vol. 73, No. 16, August 15, 2001

3883