Introduction

Top Abstract Introduction Materials and Methods Results Discussion Appendix References

Understanding the interaction of local anesthetics with plasma proteins is essential to predicting their systemic pharmacology and toxicology. Many basic drugs, including local anesthetics (LAs) and class I antiarrythmics, are bound in blood principally or in part by AGP (Routledge, 1986; Stanski and Watkins, 1986; Wood, 1986; Kremer et al., 1988). The binding of a drug to a plasma protein reduces its free fraction in serum, thus reducing its availability for active uptake or diffusion into surrounding tissue and significantly influencing its pharmacokinetics, toxicity, etc. (Wood, 1986). For example, the local anesthetic bupivacaine binds tightly to AGP (with K_D = 1 µM) (Essassi et al., 1989), and this binding may mitigate the systemic toxicity of bupivacaine (Wulf et al., 1991; Mazoit et al., 1996).

Indeed, altered pharmacokinetics have been observed following the binding of AGP to the LAs lidocaine and cocaine, and the same holds true with other clinically relevant amphipathic amines: amitriptyline, chlorpromazine, nicardipine, and verapamil (Benet and Hoener, 2002). For example, AGP concentration has been shown to correlate inversely with the free fraction of the LA ropivacaine 60 min after epidural injection (Porter et al., 2001). Also, one of the AGP variants has been shown to limit the blood-to-brain transfer of the amphipathic amines imipramine, disopyramide, and methadone (Jolliet-Riant et al., 1998).

Therefore, there has been much interest in determining the affinities of drugs (or potential drugs) for ₁-acid glycoprotein (Belaiba et al., 1989; Belpaire and Bogaert, 1989). Indeed, the ability to predict both effective doses for drugs and their potential for toxicity hinges on our knowledge of these drugs' affinities for AGP. The ability to predict affinities in turn depends on an understanding of the molecular determinants of their binding.

Here we report the affinities of three clinically used local anesthetics for the major (F1*S) variant of AGP (bupivacaine, lidocaine, and mepivacaine), along with the affinity of a racemic mixture of ropivacaine (LEA-103) and its enantiomer, LEA-104. We also report the affinities of several lidocaine homologs, including the ionizable compounds RAD-240, RAD-241, RAD 242, L-30, and W-6603, and the quaternized lidocaine derivative QX-314. These molecules were chosen to assess the contributions of both hydrophobicity and electrostatics to the binding of LAs to AGP. Hydrophobic contributions were examined at physiologic pH (7.40) by correlating the binding affinity of the LAs with their partition coefficients between octanol solvents and aqueous buffer. Ionic contributions were examined both by repeating the binding analyses with lidocaine at a higher pH (8.40), where the majority of the local anesthetic is unprotonated (pK_a = 8.19 at 25°C; Strichartz et al., 1990), and by using the permanently positively charged lidocaine analog QX-314. We contrast our results for the representative LAs examined here with the prevailing pharmacophoric model developed using a series of structurally related antihistamines (Kaliszan et al., 1996).

Affinities were determined in competition experiments with the probe molecule 2-hydroxy-3,5-diiodo-N-[2(diethylamino)ethyl]benzamide (DEDIC), recently developed in our laboratory (Cogswell et al., 2001). Addition of DEDIC, an iodinated local anesthetic analog, to AGP results in quenching of the protein's native tryptophan fluorescence by binding with a known K_D. Subsequent addition of a LA, with [DEDIC] kept constant, results in fluorescence recovery from which the concentration dependence permits determination of its affinity. The local anesthetics used in this study were themselves not efficient quenchers, and therefore, it was not possible to determine their affinities directly.

A major advantage to using DEDIC, as opposed to other probes of drug binding, is that it binds selectively to a high-affinity site on the major (F1*S) variant of AGP. This selectivity permits the determination of drug affinities to the F1*S variant in naturally occurring heterogeneous AGP (as is available commercially) and therefore obviates the need to first separate the variants chromatographically (Cogswell et al., 2001). The importance of determining affinities to the variants individually is underscored by observations that the major and minor variants bind to many drugs with differing affinities (Hervé et al., 1998). The failure to account for this heterogeneity would preclude the determination of individual binding constants and, therefore, the development of more accurate pharmacokinetic models, and greatly complicates the consideration of structure-activity relationships for protein binding.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion Appendix References

Materials. Human ₁-acid glycoprotein, lot 128H7606 (AGP), bupivacaine, and lidocaine, all >99% pure, were obtained from Sigma-Aldrich (St. Louis, MO). Ethanol, 100%, anhydrous, in glass bottles was obtained from Aldrich Chemical Co. (Milwaukee, WI). Mepivacaine, LEA-103, LEA-104, RAD-240, RAD-241, RAD-242, L-30, W-6603, and QX-314 were obtained from Dr. Rune Sandberg, Astra Pharmaceuticals (Sodertalje, Sweden); and bupivacaine enantiomers were obtained from CellTech Group PLC (London, UK). The probe molecule DEDIC was previously developed and synthesized in our laboratory (Cogswell et al., 2001). Water was doubly deionized to >18.2 M with a Milli-Q ultra pure water system (Millipore Corp., Bedford, MA) and was free of fluorescent contaminants. The structures of DEDIC and the drugs used in this study are given in Fig. 1.

View larger version (21K): [in this window] [in a new window]   Fig. 1.   Structures of the ligands used in this study. All are tertiary amines, except for quaternary QX-314.

Preparation of biochemical solutions. All solutions were prepared in standard aqueous medium (SAM), containing 0.15 M NaCl and 5 mM MOPS buffer adjusted to pH 7.40. Standard practices for the preparation and storage of solutions have been described previously (Cogswell et al., 2001).

Fluorescence Assays. Emission measurements were taken at 22°C with an SLM-Aminco SPF-500 spectrofluorometer operating in ratio mode, using rhodamine B (3 g/l in ethylene glycol) as a standard for longitudinal stability. Repeated measures of the fluorescence spectrum of a stable solution of AGP demonstrated a precision better than ±2% for recordings collected over 2 to 3 h. The excitation and emission bandpasses were set at 2.5 nm and 20 nm, respectively. As previously reported (Cogswell et al., 2001), the shape of the emission spectrum of AGP was essentially independent of excitation wavelength over the range 265 to 295 nm, indicating that tyrosine residues did not contribute significantly to the observed fluorescence. Therefore, to yield the maximal fluorescence signal, the protein was excited at either 278 nm or 285 nm, near its wavelength of maximum absorbance (278 nm). However, in experiments with QX-314, the protein was excited at 295 nm to avoid large, inaccurate (i.e., greater than a factor of 2) inner filter corrections due to absorbance by drug at shorter wavelengths.

Fluorescence Quenching Experiments with DEDIC. The use of DEDIC as a variant F1*S-selective antagonist of drug binding to whole AGP has already been described in detail (Cogswell et al., 2001), and only essential details are repeated here. Although DEDIC is structurally similar to local anesthetics, it contains two iodine atoms, which are believed to be responsible for its efficient quenching of proteins' intrinsic fluorescence. One of the noteworthy physical properties of DEDIC is that it is zwitterionic at physiologic pH, with its phenolic and amino hydrogens having pK_a values of 5.0 and 9.8, respectively (Fig. 1). Therefore, DEDIC does not undergo major changes in protonation at pH levels near the pK_a values of most local anesthetics (ca. 7.5-8.5). In the present work, this property of DEDIC greatly facilitated its use as a probe of drug binding at more than one pH.

(1)

(2)

Fluorescence Recovery Experiments with LAs. The affinities of the test compounds were determined in displacement experiments with DEDIC at either pH 7.4 or 8.4, as noted. Again, a titration method was used, taking the same precautions to prevent photodegradation. Two hundred fifty microliters of a 5 µM AGP solution were added to 2.19 ml of SAM in a cuvette and equilibrated for 5 min before an initial fluorescence reading was taken. Then, 62.5 µl of either 20 µM, 40 µM, or 80 µM DEDIC solution was pipetted into the cuvette to make a nominally 0.5 µM, 1 µM, or 2 µM solution of DEDIC, respectively. After stirring for 4 min to allow for equilibration, a second fluorescence reading was taken. Subsequent readings were taken 15 s after addition of successive aliquots of local anesthetic directly to the cuvette. Again, the volume of each aliquot was variable, ranging from 5 to 200 µl.

(3)

(4)

Physicochemical Properties of LA Homologs. Octanol:buffer partition coefficients were determined at 25°C by an optical method, as described in Sanchez et al. (1987) and Strichartz et al. (1990), using well buffered aqueous solutions of 0.15 M NaCl. These data, obtained by G. Strichartz and V. Sanchez, have not been previously reported. pK_a values of the LAs and homologs are taken from Strichartz et al. (1990) and from Bokesch et al. (1986).

Calculation of LA Surface Areas and Volumes. To probe the relation between the affinity of the LA homologs (Fig. 1) and several indices of hydrophobicity, the van der Waals surface areas and the volumes of the LAs' amino termini were calculated using Alchemy 2000 molecular modeling software (Tripos Inc., St. Louis, MO). The amino terminus was defined as the amine nitrogen plus any appended aliphatic chains, including the main chain up to, but not including, the carbonyl group (Fig. 1). These terminal fragments were energy minimized prior to calculation of surface area.

Control Experiments. To investigate the combined effects of LA fluorescence and LA quenching of AGP, each LA was added to 0.5 µM AGP without DEDIC over the range of the concentrations used in this study. In cases where the magnitude of apparent quenching was greater than the signal variance (ca. 2% of the protein's unquenched fluorescence), the data were corrected for this effect. Although no correction was necessary for bupivacaine, L-30, or RAD-242, corrections were made for mepivacaine, LEA-103/104 ("racemic ropivacaine"), QX-314, (R)-bupivacaine, (S)-bupivacaine, RAD-241, RAD-240, and lidocaine, which at the highest concentrations encountered in these experiments quenched approximately 15, 10, 8, 8, 8, 7, 6, and 5% of the original AGP signal, respectively. These curves served as DEDIC-independent baselines for the titration experiments with LAs and were subtracted from the corresponding recovery curves prior to their analysis.

Calculating the Affinity of Neutral and Protonated Lidocaine to AGP at pH 7.4. Lidocaine, an ionizable LA with a pK_a of 8.19 ± 0.03 (mean ± S.D.) at 25°C (Strichartz et al., 1990), exists as a mixture of positive and neutral species, each of which may bind to AGP with different affinities. By determining the binding affinity of a LA at two different pH levels, one can calculate the affinity of each species to AGP at pH 7.4: K_D⁰ and K_D⁺ for the neutral and protonated species, respectively. However, in this analysis, pH dependent changes of the binding site must be properly accounted for. The nonionizable, positively charged QX-314 was therefore used to assess the influence on LA binding of possible pH dependent changes in the protein-binding site. The effect of protein ionization on LA affinity, if present, could then be factored into the calculation of affinities of neutral and protonated lidocaine for AGP at pH 7.4. The derivation of equations that describe these behaviors is given in the Appendix.

	Results

Top Abstract Introduction Materials and Methods Results Discussion Appendix References

Determination of K_D (F1*S) for DEDIC at pH 7.4 and 8.4 . The addition of DEDIC to 0.5 µM AGP results in the quenching of up to 75 to 80% of the protein's intrinsic fluorescence. These data for pH 7.4, shown in Fig. 2, are well fitted by a two-site model [eq. 1 to yield a K_D (F1*S/DEDIC) of 0.057 ± 0.013 µM (mean ± S.D.)] with P_T¹ = 0.21 ± 0.02 µM, and a K_D of 9.82 ± 2.60 µM, with P_T² indeterminate, for the higher-affinity and lower-affinity site, respectively. (Here P_T¹ and P_T² refer to the apparent molar concentrations of binding sites.) The normalized amplitudes of the fluorescence decreases for the high- and the low-affinity processes are 0.53 ± 0.02 (mean ± S.D.) and 0.25 ± 0.03, respectively, with residual, unquenchable fluorescence of 0.22 ± 0.03. At pH 8.4, the affinity of DEDIC for either binding site is approximately halved; the model yields K_D values (F1*S/DEDIC) of 0.109 ± 0.025 µM (mean ± S.D.) with P_T¹ = 0.22 ± 0.02 µM, and 20.03 ± 5.94 µM, with P_T² indeterminate for the high- and low-affinity sites, respectively (Fig. 2). The relative amplitudes of the fluorescence changes for the high- and the low-affinity processes are unchanged by this elevation of pH: 0.49 ± 0.02 and 0.27 ± 0.01, respectively (mean ± S.D.). The average coefficient of determination (R²) for all fitted curves in Fig. 2 was 0.999. 

View larger version (19K): [in this window] [in a new window]   Fig. 2.   Quenching of the intrinsic (tryptophan) fluorescence of 0.5 µM AGP following addition of the probe molecule, DEDIC, to the nominal concentrations shown on the abscissa. The lines are the best fits of eq. 1 to data from three separate experiments, each of whose individual results are shown. Open symbols, pH 7.4; and closed symbols, pH 8.4.

Fluorescence Recovery on Addition of LAs (pH 7.4). Addition of increasing concentrations of lidocaine, from 0.5 µM to 10 mM, to pre-equilibrated mixtures of DEDIC and AGP restores the original protein fluorescence in a concentration-dependent manner (Fig. 3). At each of the three DEDIC concentrations used (0.5, 1, and 2 µM), primarily the major (F1*S) variant of AGP is bound by the quencher. Also note that the independent variable in the plot is the nominal concentration of LA and not the free LA concentration. The increase of IC₅₀ for lidocaine (Fig. 3, inset) with increasing nominal DEDIC concentration confirms the competitive nature of the interaction between DEDIC and lidocaine, a behavior that held for all the compounds studied here. From these data we calculated the average K_I for lidocaine binding to the F1*S variant at each DEDIC concentration using eq. 3. The averaged K_I values at 0.5, 1, and 2 µM [DEDIC] were 30.2 ± 4.9, 20.1 ± 2.9, and 26.0 ± 2.3 µM, and the overall average of all these determinations was K_I = 25.4 ± 5.3 µM (mean ± S.D.), n = 6. These calculated K_I values agree well with values in the literature, further validating this method (see Table 1). The average coefficient of determination (R²) for all fitted curves in Fig. 3 was 0.991. 

View larger version (25K): [in this window] [in a new window]   Fig. 3.   Fluorescence recovery on addition of lidocaine to premixed solutions of 0.5 µM AGP and DEDIC at pH 7.4. Three sets of recovery curves (each with two replicates) are shown, corresponding to a different [DEDIC]: 0.5 µM, open symbols; 1.0 µM, cross symbols; and 2.0 µM, closed symbols. Best-fit lines for eq. 3 are shown for each set of replicates. Inset, IC50 versus each [DEDIC].

View this table: [in this window] [in a new window]   TABLE 1 Local anesthetic binding to the major variant of AGP (F1*S) at pH 7.4 and pH 8.4  The first two columns give the KD values determined for ligand binding to the major (F1*S) variant of human  1-acid glycoprotein at pH 7.4 and pH 8.4, respectively. For comparison with our data at pH 7.4, columns three and four give previously reported KD values (where available) for drug binding to F1*S and whole AGP, respectively. With the exception of DEDIC, previous determinations were made in buffers of varying compositions, and different from our own. The number of independent determinations is given in parentheses.

Local Anesthetics: Ionic Factors in Binding to AGP. To examine the effects of charge on the drug-protein interaction, we determined the affinity of lidocaine for the major AGP variant at pH 8.4. This increase in pH changes the fraction of lidocaine in the neutral form from 0.14 to 0.63, and the K_I to 6.75 ± 0.68 µM (mean ± S.D.), about a quarter of the value at pH 7.4 (Table 1). To probe the effects on binding from changes in the ionization of the protein between pH 7.4 and 8.4, the affinity of the positive, nonionizable lidocaine homolog QX-314 was determined at both pH levels (Table 1): 90.3 ± 7.6 µM and 52.2 ± 14.8 µM (mean ± S.D.) at pH 7.4 and 8.4, respectively. Raising the pH over this range approximately halves the K_D for QX-314, showing that the charged compound has a higher affinity at the higher pH.

Local Anesthetics: Relation between Affinity and Hydrophobicity. In Table 2, the Gibbs free energy of dissociation (G) of the local anesthetics (calculated as RT ln[K_eq]) and LA derivatives is listed along with several measures of hydrophobicity: number of amino-terminal carbons, surface area and volume of the amino terminus, octanol:buffer distribution coefficient of each compound at pH 7.4 (Q_7.4); octanol:buffer partition coefficients for the neutral (P⁰) and charged (P⁺) drug species, and pK_a. Graphs of several of these molecular properties versus the apparent free energy of binding reveal the features of LAs that systematically correlate with AGP affinity, and those features that do not.

View larger version (16K): [in this window] [in a new window]   Fig. 4.   Relationship between G (dissociation) of structurally homologous LAs from the F1*S variant of AGP and the experimentally determined octanol:buffer partition coefficient of the neutral species (G. R. Strichartz and V. Sanchez, unpublished observation) (r2 = 0.82).

View larger version (19K): [in this window] [in a new window]   Fig. 5.   A, relationship between G (dissociation) of structurally homologous LAs from the F1*S variant of AGP and the number of alkyl-amino carbons (r2 = 0.22). B, analogous relationship between G (partitioning) of neutral species between octanol and buffer (G. R. Strichartz and V. Sanchez, unpublished observation) (r2 = 0.34).

View larger version (12K): [in this window] [in a new window]   Fig. 6.   Relationship between G (dissociation) of structurally homologous LAs from the F1*S variant of AGP and the LA pKa (Bokesch et al., 1986) (r2 = 0.26).

View larger version (14K): [in this window] [in a new window]   Fig. 7.   Relationship between G (dissociation) of structurally homologous LAs from the F1*S variant of AGP and the molecules' octanol:buffer distribution coefficient at pH 7.4 (G. R. Strichartz and V Sanchez, unpublished observation) (r2 = 0.08).

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion Appendix References

The affinities of the major AGP variant for the clinically important drug mepivacaine, racemic ropivacaine, the quaternary lidocaine derivative QX-314, and the tertiary amine lidocaine homologs RAD-240, RAD-241, RAD-242, W-6603, and L-30 are reported here for the first time. Although values for bupivacaine and lidocaine were previously reported (Denson et al., 1984; Kaliszan et al., 1996; Cogswell et al., 2001), we repeated these determinations using our current lot of AGP and buffer system to allow valid comparison with the present determinations of the other compounds. Not only can the buffer affect the K_D (Ravis et al., 1988), complicating comparisons of drug affinities under different conditions, but the composition and quality of commercial AGP are also known to vary significantly, even from lot to lot from the same supplier (Lunde et al., 1986; Morin et al., 1986; Hervé et al., 1997). Nevertheless, our results for bupivacaine and lidocaine binding to the major AGP variant are consistent with those reported in the literature (Table 1).

Hydrophobic Effects on LA Binding. For the two series of local anesthetics, the linear alkyl amino homologs of lidocaine and the piperidine ring-containing homologs of mepivacaine, we found a linear relationship (r² = 0.82) between G (dissociation) for AGP binding and the octanol:buffer partition coefficient of the neutral drug species. The strong correlation between hydrophobicity and the affinity for the LAs is in accord with studies showing that generalized hydrophobicity is a major determinant of binding for other drug classes, including basic drugs such as antihistamines and antihypertensives (Kaliszan et al., 1996), diazepines (Maruyama et al., 1992), and phenothiazines (Miyoshi et al., 1992), as well as acidic drugs such as coumarin anticoagulants (Maruyama et al., 1990). That the same linear correlation holds for both homologous series, each with a different conformation of carbons around the amino nitrogen, implies that this hydrophobic interaction is relatively forgiving, suggesting that the binding site is a broad hydrophobic surface or a flexible pocket that can accommodate ligands with a range of sizes.

Electrostatic Effects on LA Binding. The effects of pH on local anesthetic pharmacokinetics are complex and poorly understood, yet clearly important (Porter et al., 2000). Given that LAs, particularly the notoriously cardiotoxic bupivacaine, are reported to bind tightly to AGP (Essassi et al., 1989), it seemed plausible that pH dependent changes in LA pharmacology/toxicology could be mediated through changes in their affinity for the protein.