Anal. Chem. 2002, 74, 282-287
A Framework Based on the Extended Wyman Concept for Analyzing the Salt Effects on the Solute Retention in High-Performance Affinity Chromatography Ines Slama,† Corinne Ravelet,† Catherine Grosset,† Anne Ravel,† Annick Villet,† Edwige Nicolle,‡ and Eric Peyrin*,†
Laboratoire de Chimie Analytique and Laboratoire de Chimie Organique, UFR de Pharmacie de Grenoble, UJF, Domaine de la Merci, 38700 La Tronche, France.
The analysis of binding data of a ligand to a macromolecule in the presence of an additive can be classically formulated in terms of the linked functions of Wyman. In the case of a salt, this approach has been extended by Tanford such that the contributions of both salt and water are taken into account. In this paper, the extended Wyman theory was applied to high-performance affinity chromatography (HPAC) in order to define a general model describing the effects of the mobile-phase salts on the ligand binding. Various HPAC literature data, as well as our data concerning dansyl amino acid retention on a vancomycin stationary phase, were examined in relation to this model. From the results, this theoretical approach was considered to be adequate to describe accurately the salt dependence on solute retention. This work shows the importance of taking into account the effects of both ionic species and water in the investigation of relative contributions of the interactions involved in the ligand binding to immobilized receptor.
Various experimental approaches have been proposed to analyze the retention mechanisms in high-performance affinity chromatography (HPAC). For example, some studies have examined the effects of temperature on retention.1-5 The change in enthalpy and entropy associated with the transfer of the solute can be extracted from the linear van’t Hoff plots and analyzed in order to obtain information about the driving forces implied in the association process. Another approach for studying the interactions between a ligand and a receptor involves the variation * Corresponding author. E-mail:
[email protected]. † Laboratoire de Chimie Analytique. ‡ Laboratoire de Chimie Organique. (1) Peter, A.; Torok, G.; Armstrong, D. W.; Toth, G.; Tourwe´, D. J. Chromatogr. A 1998, 828, 177. (2) Armstrong, D. W.; Rundlett, K.; Reid, G. L. Anal. Chem. 1994, 66, 1690. (3) Gilpin, R. K.; Ehtesham, S. B.; Gilpin, C. S.; Liao, S. T. J. Liq. Chromatogr. Relat. Technol. 1996, 19, 3023. (4) Peyrin, E.; Guillaume, Y. C.; Guinchard, C. Anal. Chem. 1997, 69, 4979. (5) Peyrin, E.; Guillaume, Y. C.; Morin, N.; Guinchard, C. Anal. Chem. 1998, 808, 113.
282 Analytical Chemistry, Vol. 74, No. 1, January 1, 2002
of the mobile phase composition. Classically, it may be apprehended by varying the proportion of organic modifier, pH, or ionic strength of the mobile phase. The effects of the variation of the eluent ionic strength have been examined by several authors for various immobilized receptors, such as proteins,6-14 cyclodextrins15,16 or macrocyclic antibiotics,17 in order to distinguish the relative contributions of ionic and hydrophobic interactions in the association process; however, in most cases, no theoretical treatment has been reported concerning the salt effects on the ligand-receptor association in HPAC. Our group previously reported that the Wyman linkage relations constitute a valuable tool to describe the salt dependence on the ligand binding in HPAC.18,19 The major advantage of the Wyman functions consists of the fact that the variation of the association constant between a ligand and a receptor can be described whatever the chemical properties of the mobile phase additives. The theory of linked functions allows following the change in the ligand binding when different additives, such as a competing agent, osmolytes or denaturants, are added to the medium. The initial relations of the Wyman theory are limited by the fact that the experimental data are evaluated only through the assumption that the additive is bound to a unique class of sites. In addition, these data do not take into account that the additive can indirectly affect the water activity at rather high concentrations. Therefore, Tanford and coworkers20,21 have reevaluated the Wyman relations with regard (6) Wainer, I. W. J. Chromatogr. A 1994, 666, 221. (7) Sly, L. A.; Reynolds, D. L.; Walker, T. A. J. Chromatogr. 1993, 641, 249. (8) Orn, G.; Lahtonen, K.; Jalonen, H. J. Chromatogr. 1990, 506, 627. (9) Hermansson, J.; Hermansson, I. J. Chromatogr. A 1994, 666, 181. (10) Hermansson, J.; Grahn, A. J. Chromatogr. A 1995, 694, 57. (11) Allenmark, S.; Bomgren, B.; Boren, H. J. Chromatogr. 1984, 316, 617. (12) Yang, J.; Hage, D. S. J. Chromatogr. A 1996, 725, 273. (13) Loun, B.; Hage, D. S. Anal. Chem. 1994, 66, 3814. (14) Hedeland, M.; Holmin, S.; Nygard, M.; Pettersson, C. J. Chromatogr. A 1999, 864, 1. (15) Paleologou, M.; Li, S.; Purdy, W. C. J. Chromatogr. Sci. 1990, 28, 311. (16) Stalcup, A. M.; Gahm, K. H. Anal. Chem. 1996, 68, 1369. (17) D’acquarica, I.; Gasparrini, F.; Misiti, D.; Villani, C.; Carotti, A.; Cellamare, S.; Muck, S. J. Chromatogr. A 1999, 857, 145. (18) Peyrin, E.; Guillaume, Y. C.; Guinchard, C. Biophys. J. 1999, 77, 1206. (19) Guillaume, Y. C.; Peyrin, E.; Villet, A.; Nicolas, A.; Guinchard, C.; Millet, J.; Robert, J. F. Chromatographia 2001, 52, 753. (20) Tanford, C. J. Mol. Biol. 1969, 39, 539. 10.1021/ac010696u CCC: $22.00
© 2002 American Chemical Society Published on Web 12/01/2001
to the additive binding to multiple classes of sites and molecular hydration. Applications of the extended linkage relations that involve the change in water activity have been reported for the binding of small ligands to protein.22,23 In this paper, the complete extended concept was applied specifically to the investigation of the effects of a salt on analyte retention in HPAC. A general equation was derived from the analysis of solute binding to an immobilized receptor. Various literature data that report the solute retention factor against salt concentration were reexamined in relation to the model equations. In addition, the effects of eluent salt concentration on the dansyl amino acids (used as test solutes) retention on immobilized vancomycin were studied using this general model. The results were discussed in order to provide information about the possible mechanistic contributions of both salt and water on the ligandreceptor affinity. THEORY In HPAC, the retention factor, k, is classically expressed as follows24 p
k)
∑β j
() ()
ii
i)1
m
Vm
)K
m
Vm
(1)
where βi is the equilibrium constants at the i individual sites; ji, the fraction of each type of site; m, the total number of moles of binding sites; Vm, the void volume; and K, the apparent equilibrium constant for the solute-receptor association. The salt dependence on the apparent equilibrium constant, K, can be described using a simplified model derived from the linkage relations of Wyman. K for both nonspecific and specific interactions can be directly linked to the change in salt concentration, c, using the following equation as previously reported by Aune et al.21
ln K ) ln K0 + ∆n ln(1 + κc)
ln k ) ln k0 + ∆n ln(1 + κc) - ∆χc
(4)
where the constant ∆χ is linked to the water molecule release at the ligand-receptor interface and corresponds to the classical hydrophobic effect. In this approach, no allowance has been made for the possible antagonist effect of the salt on the association equilibrium, although it has been shown that a salt can affect the ligand binding in opposite directions. For example, the interaction between a cationic solute and a receptor with negatively charged sites can be modulated when the salt concentration increases via two antagonist phenomena: (i) a reduction of the ion-exchange process between the opposite charges of the ligand and the receptor and (ii) an enhancement of the ion-pairing process favoring the solute binding to uncharged binding sites. Such a behavior has been previously reported by Hermansson and coworkers for the salt dependence on the retention of cationic or anionic drugs on immobilized protein.9,10 This duality can be expressed in terms of salt binding to multiple classes of sites at the interface between receptor and ligand. Using the simplifying assumptions that the salt can react at n1or2 identical and independent sites, the following equation can be obtained from eq 3,
ln k ) ln k0 + ∆n1 ln(1 + κ1c) + ∆n2 ln(1 + κ2c)
(5)
where κ1or2 is the single average binding constant between the salt and the site classes, 1 or 2; ∆n1or2 is related to the difference (between the two states) in the number of salt molecules bound to site classes 1 or 2. Taking into account the contribution of water, eqs 4 and 5 can be rearranged as follows:
ln k ) ln k0 + ∆n1 ln(1 + κ1c) + ∆n2 ln(1 + κ2c) - ∆χc (6)
(2)
where the release parameter ∆n is related to the difference in the number of salt molecules bound in the receptor-ligand interface between the two states in equilibrium (choice of the simplest case that involves n identical and independent sites for salt binding), with the average additive binding constant, κ, and K0 is the theoretical K value for c ) 0. Combining eqs 1 and 2 gives the following,
ln k ) ln k0 + ∆n ln(1 + κc)
incorporated into the previous equation so that the following relation is obtained,
(3)
where k0 ) K0m/Vm. As stated above, this latter equation takes into account only the direct salt effect; however, in a wide enough salt concentration range, the notion of change in the water activity must be introduced, as first anticipated by Tanford.20 This concept can be (21) Aune, K. C.; Tanford, C. Biochemistry 1969, 8, 4586. (22) Haire, R. N.; Hedlund, B. E. Biochemistry 1983, 22, 327. (23) Colombo, M. F.; Rau, D. C.; Parsegian, V. A. Proc. Natl. Acad. Sci. U.S.A. 1994, 91, 10517. (24) Hage, D. S.; Tweed, S. A. J. Chromatogr. B 1997, 699, 499.
Eq 6 describes the complete contributions of salt and water to the retention factor when two classes of salt sites are involved in the ligand affinity. These various model equations 3 to 6 were used to analyze different literature HPAC data as well as the retention behavior of dansyl amino acids on immobilized vancomycin. EXPERIMENTAL SECTION Apparatus. The HPLC system consisted of a LC Shimadzu pump 10AT (Touzart et Matignon, Courtaboeuf, France); a Rheodyne injection valve, model 7125 (Interchim, Montluc¸ on, France) fitted with a 20-µL sample loop; and a Shimadzu SPD10A UV-vis detector. An Astec 150 × 4.6 mm Chirobiotic V HPLC column (packed with a stationary phase produced by chemically bonding the macrocyclic glycopeptide vancomycin to a 5-µm silica gel) was used with controlled temperature in an oven Igloocil (Interchim). The mobile phase flow rate was 0.8 mL min-1. Reagents and Operating Conditions. D-Dansyl valine, Ddansyl tryptophan, and D-dansyl phenylalanine were obtained from Sigma Aldrich (Saint-Quentin, France). HPLC grade methanol, tri-sodium citrate, and citric acid were supplied by Prolabo (Paris, France). Water was obtained from an Elgastat option water Analytical Chemistry, Vol. 74, No. 1, January 1, 2002
283
Figure 1. Plots of ln k against ln c (c, sodium phosphate concentration) for two HPAC data: aspartame-R-chymotrypsin6 (9), medetomide-R-1-glycoprotein8 (b). The theoretical curves are recreated from eq 3 (s).
Figure 2. Plot of ln k against ln c (c, sodium phosphate concentration) for the beraprost (two stereoisomers)-R-1-glycoprotein association.7 The theoretical curves are recreated from eq 3 (s).
purification system (Odil, Talant, France) fitted with a reverse osmosis cartridge. The mobile phase consisted of citrate buffer pH 7.0/methanol 90/10 (v/v). The variation range of the sodium citrate and sodium chloride was 5 × 10-4-3 × 10-1 M. To examine the concentration dependencies of solute retention corresponding to the binding capacity of the immobilized vancomycin, retention measurements were related to varying amounts of injected solute. Solute samples were prepared at different concentrations in the mobile phase from 0.125 to 10 µg mL-1. A 20-µL portion of each solute (at a concentration of 0.25 µg mL-1) was injected in triplicate, and the retention times were measured. RESULTS AND DISCUSSION Analysis of Literature HPAC Data. To test the validity of the equations of the proposed model, various experimental data from the HPAC literature giving the retention factor as a function of the concentration of the buffer salt were reevaluated. Figures 1 and 2 show the experimental retention data as a function of sodium phosphate concentration for aspartame on an R-chymotrypsin stationary phase (ACHT)6 and beraprost and medetomide on immobilized R-glycoprotein (AGP).7,8 It was observed that the salt release hindered the aspartame-ACHT and medetomide-AGP associations when the salt concentration increased (Figure 1), but the binding of beraprost on immobilized AGP increased with salt concentration (Figure 2). The use of a simplified relation of the Wyman theory (eq 3) was sufficient to fit these experimental data. It corresponds to the case in which the influence of the change in water activity was neglected in 284
Analytical Chemistry, Vol. 74, No. 1, January 1, 2002
relation to the direct salt effect and only one class of additive site was taken into account. The parameters ∆n (corresponding to the slope of the ln-ln plots) were calculated using eq 1 for the four association processes evaluated. They are presented in Table 1 with the corresponding regression coefficients. The κ values can be calculated, but with an insufficient accuracy, in the absence of data at low enough concentration of salt. The high R2 values indicate that the simplified model can be considered adequate to describe the retention behavior. The ∆n values agree well with the values of the salt release (ranging from -0.2 to -2.8 in relation to the salt type) for the polyphosphate binding to exopolyphosphatase.25 The decrease in the binding of anionic aspartame and cationic medetomide on ACHT and AGP, respectively, over the phosphate buffer concentration range suggests an ion-exchange retention phenomenon between charged solutes and protein surface sites of opposite charges (salt competition for binding at the aspartame and medetomide sites). In the case of beraprostimmobilized AGP association, the results were interpreted as an ion-pairing process that involves the Na+ of the buffer as counterion of the anionic solute. Illustrations of the antagonist effects of various salts (eq 5) for different solute-protein associations are presented in Figures 3 and 4. Two retention behavior types were analyzed. In Figure 3, an initial increase in the solute retention was observed at low additive concentration, followed by a decrease in the ligandreceptor affinity at high salt concentration. This was attained for the retention of ibuprofen and cationic drugs on immobilized AGP9,10 using sodium phosphate or sodium acetate as buffer. It was possible to fit accurately theoretical binding curves to the experimental data using eq 5, as shown in Figure 3; however, the data were not large enough (small number of experimental points) to extract values of the equation parameters. It has been previously proposed9,10 that two antagonist retention phenomena could occur for both anionic and cationic compounds on immobilized AGP: an ion exchange with oppositely charged sites at the protein surface and an ion pairing with binding of solutes to uncharged locations, such as hydrophobic sites. Thus, the initial retention increase was attributed to the enhancement of the ion pairing between charged solutes and buffer a counterion of opposite charge (Na+ for anionic ibuprofen and acetate for cationic compounds). At higher salt concentrations, the affinity decrease was possibly due to the reduction of the ion exchange between the charged solutes and the binding sites of opposite charge. In Figure 4, a reverse effect is described for amino acid derivatives and L-tryptophan on immobilized bovine serum albumin (BSA)11 and human serum albumin (HSA)12, with the concentration of sodium phosphate buffer varying. Using eq 5, the ∆n1 parameters related to the decrease in the ligand affinity were calculated. These values and the nonlinear regression coefficients are shown in Table 1. Values of ∆n2 (corresponding to the solute retention increase) are not shown, because they were not sufficiently accurate as a result of the limited range of high salt concentration. The retention decrease that occurred at low salt concentrations was probably due to a direct competition of salt for the ligand binding sites (ion-exchange process). At a high additive amount, a curvature was observed that may be due to an increase in the (25) Bolesch, D. G.; Keasling, J. D. Biochem. Biophys. Res. Commun. 2000, 274, 236.
Table 1. Analysis of Various HPAC Data Using Model Equations with the Determination of the Salt Release Parameters ∆na models used eq (3); salt effects ligandb
receptorb
ref
c range
∆n
R2
ASP MDT BRPf BRPg AAh AAi TRP
ACHTc
6 8 7 7 11 11 12
0.05-0.5 0.005-0.05 0.005-0.06 0.005-0.06 0.005-0.5 0.005-0.5 0.0012-0.25
-0.7 -0.3 +0.6 +0.5
0.989 0.987 0.997 0.997
AGPc AGPd AGPd BSAe BSAe HSAe
eq (4); salt and water effects
eq (5); dual salt effects
∆n
R2
∆n1
R2
-0.4 -0.3 -1.5
0.999 0.987 0.999
-0.4 -0.3 -1.5
0.999 0.987 0.999
a Nonlinear regression coefficient R2. c See Figure 1. d See Figure 2. e See Figure 4. f Stereoisomer 1. g Stereoisomer 2. h N-benzoyl valine. i Nbenzoyl phenylalanine. b ASP, aspartame; MDT, medetomide; BRP, beraprost; AA, amino acid derivative; TRP, L-tryptophan; ACHT, R-chymotrypsin; AGP, R-1-glycoprotein; BSA, bovine serum albumin; HSA, human serum albumin.
Figure 3. Plots of ln k against ln c (c, sodium phosphate or acetate concentration) for various HPAC data: ibuprofen-R-1-glycoprotein9 (9), cationic drugs (propranolol) ([), diperodon (b), bupranolol-R1-glycoprotein10 (2). The theoretical curves are recreated from eq 5 (s).
Figure 4. Plot of ln k against ln c (c, sodium phosphate concentration) for two HPAC data: amino acid derivative: N-benzoyl valine (2), N-benzoyl phenylalanine-bovine serum albumin11 (9), L-tryptophanhuman serum albumin12 (b). The theoretical curves are recreated from eq 5 (s).
ion-pairing process. Alternatively, this curvature can be attributed to a change in the water activity, that is, the enhancement of the hydrophobic effect at high enough buffer concentration (>0.2 M). Thus, eq 4 was fitted to the data for the association processes. The fits obtained were not detectably different from the curves reported in Figure 4 using eq 5. The parameters ∆n, presented in Table 1, were identical to the ∆n1 values calculated using eq 5. As evoked above for the analysis using eq 5, ∆χ cannot be determined with a sufficient accuracy from the data and as a result, its value is not presented. It could be also hypothesized that salt effects and water contribution were involved concomitantly in the solute retention increase at high salt concentration, as theoretically described by eq 6.
Figure 5. Plots of ln k against ln c (c, sodium citrate concentration) for dansyl amino acids using immobilized vancomycin: D-dansyl tryptophan (9), D-dansyl phenylalanine (b), D-dansyl valine (2). The theoretical curves are recreated from eq 6 (s).
Figure 6. Plots of ln k against ln c (c, sodium chloride concentration) for dansyl amino acids using immobilized vancomycin: D-dansyl tryptophan (9), D-dansyl phenylalanine (b), D-dansyl valine (2). The theoretical curves are recreated from eq 6 (s).
Effects of Salt on the Solute-Vancomycin Association. To gain further insight into the validity of the proposed complete model, the salt effects on another association type were analyzed. The retention factor values for D-dansyl amino acids on immobilized vancomycin were determined in relation to the concentration of sodium citrate and sodium chloride in the mobile phase (5 × 10-4-3 × 10-1 M). The coefficients of variation of the k values were
