Weak Polar Interactions Confer Albumin Binding Site Selectivity for Haloether Anesthetics

Human serum albumin (essentially fatty acid free) was purchased from Sigma Chemical Company (St. Louis, MO) and used without further purification. Isoflurane and enflurane were obtained from Halocarbon Laboratories (Liberty Corner, NJ). All other chemicals were reagent grade or better and were obtained from Sigma.

Isothermal titration calorimetry can measure the full thermodynamic profile of a bimolecular interaction (binding) without modification of the ligand or the protein. The method consists of an ultrasensitive thermometer to measure the heat changes that occur when ligand and protein are mixed, and by fitting the heat signal of multiple, systematic injections of ligand into protein solutions, various binding models can be used to derive the underlying thermodynamics (enthalpy and entropy changes), including the association constant (K^A) and stoichiometry (n). We and others have demonstrated good agreement with data derived from other techniques.5,6 

Briefly, titrations were performed at 20°C using a Microcal, Inc. VP ITC (Northampton, MA). The sample cell contained 0.21 mm HSA, and the reference cell contained water. Ligand, 15 μl (injector stock concentrations of 12 mm for isoflurane and 10 mm for enflurane), was injected at 5-min intervals into the HSA sample solution. Sequential titrations were performed to ensure full occupancy of the binding sites by loading and titrating with the same ligand without removing the samples from the cell until the titration signal was essentially constant. After the full titration, final concentrations of anesthetics in the HSA cell were approximately 4 mm. The titrations were linked together for data analysis using ConCat32 software distributed from Microcal, Inc. Four separate titrations were performed, including ligand into buffer, buffer into protein, buffer into buffer, and ligand into protein, and the titration corrected accordingly. To ensure data reliability, at least three experiments were performed for each ligand.

Origin 5.0 (Microcal Software, Inc., Northampton, MA) was used to fit thermodynamic parameters to the heat profiles. The following formulawas used for data fitting:

where K is the binding constant, n is the number of sites, V^ois the active volume involved in interaction, M^tis the total concentration of protein in V^o, X^tis the total concentration of ligand, ΔH is the molar heat of ligand binding, and Q is the total heat content of the solution contained in V^o(determined relative to zero for the unliganded species) at fractional saturation. The process of fitting experimental data then involves iterative improvement of initial values of n, K, and ΔH by standard Marquardt methods to minimum chi-square values.

Electrostatic interactions in protein binding sites can be influenced by salt concentration. Increased concentrations of charged ions compete with the ligand for charged residues in the cavity and thus reduce apparent ligand affinity if dependent on that interaction. Therefore, ITC experiments were performed at 130 mm NaCl and at 500 mm NaCl with 20 mm NaHPO⁴and a pH of 7.0.

To test for overlapping binding sites, competition experiments were performed using ITC in a buffer condition of 130 mm NaCl and 20 mm NaHPO⁴, with a pH of 7.0. For competition between isoflurane and enflurane, 0.075 mm HSA in the sample cell was titrated with isoflurane or enflurane followed by enflurane or isoflurane, respectively.

Because the HSA crystallographic binding sites for propofol have been recently reported7and confirmed under solution conditions,5propofol was used as a probe of haloether location by using competition experiments. HSA (0.015 mm) that had been preequilibrated with 0.5 mm propofol was titrated with propofol, isoflurane, or enflurane. For comparison, we repeated the same titrations into HSA without propofol preequilibration.

The molecular properties of isoflurane and enflurane were calculated using Molecular Analysis Pro (ChemSW, Inc., Fairfield, CA). Dipole moment and partial charges were calculated using the modified partial equalization of orbital electronegativity method.8,9The sterics of protein pockets or cavities and their lining residues in HSA in the absence (1AO6)10and presence of propofol (1E7A)7were calculated using CASTp (a Web-based program to determine cavity information‡).11Protein coordinates were obtained from the Protein Data Bank (§).

The fitted parameters from ITC enthalpograms are presented as mean ± SD, and the mean values were compared with unpaired t  tests using InStat v 3.06 (San Diego, CA). A P  value less than 0.05 was considered significant.

The titration of HSA with isoflurane at 130 mm NaCl released heat (fig. 2), and subsequent fitting of the enthalpogram with a single-class, variable n binding site model produced a single site with a K^Aof 1,400 m⁻¹(K^D= 0.7 mm; table 1). At 500 mm NaCl, the affinity for isoflurane decreased significantly (P < 0.05; table 1), but mostly through a less favorable entropy term.

The enflurane-HSA interaction was also exothermic, with a derived K^Aof 1,000 m⁻¹(K^D= 1 mm; table 1) for about two identical binding sites. In contrast to isoflurane, 500 mm NaCl did not significantly alter the enflurane-HSA affinity (table 1).

The heat release diminished to a constant but nonzero level after a full enflurane titration into 0.075 mm HSA. Further titration of this sample with isoflurane was not accompanied by any additional heat release (fig. 3, left). Similarly, heat release diminished to a constant level after full isoflurane titration into 0.075 mm HSA, but in this case, further titration with enflurane was accompanied by further heat release (fig. 3, right).

Preincubation of HSA with 0.5 mm propofol inhibited subsequent heat release with either isoflurane or enflurane titration (fig. 4, left) as compared with HSA without propofol preincubation (fig. 4, right).

Isoflurane and enflurane have the same atomic composition, the same molecular weight, and essentially the same molecular volume of approximately 110 ų, but isoflurane has an almost twofold larger dipole moment than enflurane, 0.7 debye for enflurane versus  2.0 debye for isoflurane.

Isoflurane and enflurane binding was inhibited by propofol, implicating the two propofol binding sites in domain 3 of HSA as haloether binding sites. Both propofol cavities contain charged residues; the one containing tyrosine-411 has charged atoms from four lining residues (NE and NH2 of R410; CE and NZ of K414; CG of R445; and CA, O, CG, and CD of R485; fig. 5), whereas the other pocket has no charged atoms in the lining (fig. 5). Using CASTp and a 1.4-Å radius probe, we determined the cavity volume in the presence and absence of propofol and found that the site containing tyrosine-411 increased from 340 ųin 1AO6 to 510 ųin 1E7A on propofol binding. The second propofol binding pocket initially consisted of small pockets of only 20–200 ų(fig. 5) in the unliganded 1AO6 but coalesced to a larger cavity of 810 ųin the propofol-bound state (1E7A).

The principal finding of this study is that structural isomers of the haloether anesthetics retain sufficient electrostatic identity to select different protein binding sites. Therefore, it is entirely feasible that the differences in the various actions of isoflurane versus  enflurane are due to occupancy of different binding sites, perhaps on different molecular targets.

In contrast to the large number of haloalkane binding sites on HSA demonstrated with many approaches,5,7,12including ITC,5our current ITC experiments indicate that isoflurane has only a single energetically significant binding site. Using mutagenesis, we have previously demonstrated that isoflurane binds at the Y411 binding site of HSA,12indicating that this site is the dominant isoflurane binding site. Competition between isoflurane and enflurane indicated that enflurane also binds to the Y411 site. Further confirmation is provided by the propofol competition experiments, because the Y411 site is also a known propofol binding site.5,7Enflurane seems to have an additional site that excludes isoflurane in the concentration range achieved here (up to 4 mm). The observation that propofol binding fully inhibits haloether binding suggests that the additional enflurane binding site is the second site for propofol, also in domain three.

It is predicted that the relation between cavity volume and ligand molecular volume plays a role in binding site selectivity. Using the short dimension of these molecules of approximately 5.8 Å and using CASTp with this probe size (instead of the normal 1.4 Å), six cavities in HSA (1E7B) are large enough for isoflurane or enflurane. Therefore, binding site selectivity must rely on features other than volume. Although the same atoms comprise the two haloether molecules, the different arrangement produces a different shape (fig. 1), and so the corresponding shape of the cavity might also contribute; few are expected to match that of the anesthetic perfectly. However, this may be less important than supposed because of the dynamic nature of proteins and their cavities—in part reflected by the substantial change in cavity volume noted in the analysis of x-ray diffraction data. A clear difference between enflurane and isoflurane is the permanent dipole moment; therefore, dipole-dipole interactions might contribute to selectivity. Consistent with this possibility, there are four positively charged residues lining the isoflurane binding site with the charged side chain atoms forming the pocket surface (fig. 5). The enhanced enthalpy per site for isoflurane as compared with enflurane also supports this idea. Even further support comes from the observation that increased salt concentration reduces isoflurane affinity, presumably via  charge “screening.” Although two positively charged residues line the additional enflurane binding site, the charged atoms are not part of the pocket surface, suggesting that this cavity is not as polar as the other. Propofol, being intermediate in dipole moment (1.6 debye), binds both cavities, although consistent with the above, the crystal data suggests higher occupancy of the Y411 cavity.

Using ¹⁹F nuclear magnetic resonance spectroscopy13,14and competitive photoaffinity labeling,15isoflurane has been shown to bind to bovine serum albumin with K^Dvalues of 1.36–1.5 mm. Using a very different method, ITC, we found comparable overall affinity for isoflurane (0.7 mm). It is important to note that although ITC can determine the full thermodynamic profile for bimolecular interactions, it is difficult to unambiguously derive all parameters when the Wiseman c parameter16(the product of the protein concentration and K^A) is less than approximately 10. Nevertheless, recent studies have indicated reliable parameter estimation using ITC in low-affinity systems.17ITC has the distinct advantage of being unbiased with respect to any particular “reporter” feature of either ligand or target—heat change is a feature of all energetically significant interactions. Thus, heat is released (exothermic; negative ΔH) or absorbed (endothermic; positive ΔH) in direct proportion to the extent and strength of interaction that occurs. For example, the interactions of inhaled anesthetics with a designed anesthetic-binding, four-helix bundle protein were exothermic.6Further, the anesthetic-firefly luciferase interaction is exothermic,18as is the chloroform-bovine serum albumin interaction.19Finally, we have shown that both halothane and propofol binding to HSA is exothermic.5However, the Overton-Meyer relation (positive correlation between hydrophobicity and potency) and what limited structural information exists indicate that the dominant interactions are hydrophobic in nature. Hydrophobic interactions are generally accompanied by a very small ΔH—most of the binding energetics being driven by favorable changes in entropy. However, the consistently negative ΔH in this and past studies suggests that some polar interactions must contribute to anesthetic binding. It is perhaps surprising that the HSA interaction with both haloethers has a considerably negative ΔH value, suggesting that even with the smaller dipole moment, other weak electrostatics, like van der Waals interactions, must play a role in enflurane binding. Nevertheless, in the case presented here, it is highly probable that the permanent dipole of the anesthetic molecule provides the force responsible for site selectivity.

The observed selectivity may have pharmacologic significance. The ability of several general anesthetics (halothane, isoflurane, enflurane, and propofol) to bind at the same site in domain 3 of HSA suggests the possibility that this site bears resemblance to sites in the central nervous system that contribute to the effect that all of these drugs have in common: anesthesia. The additional site for enflurane might resemble those responsible for its different effects.20–24For example, enflurane is associated with epileptiform activity in some patients,25and other inhaled drugs with this property have a tendency to be even more apolar. It is tempting to speculate that the lower potency of enflurane as compared with the less hydrophobic isoflurane, a violation of the Overton-Meyer relation, is in part due to occupancy of functionally opposed sites—in either the same or different targets.

In summary, two binding sites of different character exist in HSA for the haloether anesthetics. One site is more polar and prefers isoflurane, presumably because of its larger dipole. The second site is less polar and binds only enflurane. Therefore, in addition to molecular volume and hydrophobic surface area, weak polar interactions confer considerable binding selectivity, which may underlie differences in drug action.