Development and application of (bio)analytical methodologies in capillary electrophoresis. Enantioselectivity considerations

The development of new drugs is a long, complex and expensive process that involves the evaluation of pharmacokinetic and pharmacodynamic properties of the new molecule in different stages. In the early stages of drug development, in vitro methods for the high-throughput screening of new molecules’ properties are commonly used, with the aim of having preliminary data about the potential pharmacological activity of a molecule and also its pharmacokinetics. When chiral molecules are employed, both pharmacokinetic and pharmacodynamic properties can show a certain degree of enantioselectivity due to the interaction with optically active biomacromolecules (Brocks, 2006), so the methods applied in these preclinical early stages must allow evaluating these properties in an enantioselective way. This need comes from the growing interest of the pharmaceutical industry in the development of single-enantiomer formulations, and also from the regulations concerning racemic drugs, which indicate that enantioselective analysis should be performed also over racemates (Food and Drug Administration, 1992; European Medicines Agency, 1993). Capillary electrophoresis (CE) is a convenient analytical technique to be applied for these preclinical studies. It provides advantageous possibilities for the chiral analysis of small molecules and also for the in-line evaluation of some pharmacokinetic and pharmacodynamic processes, such as the binding of drugs to plasma proteins or their behavior in different enzymatic reactions. Its main features, which are extremely interesting for these applications, are speediness of analysis, low consumption of chemicals and reagents, ease of automation, high peak efficiencies and low environmental impact (Lin et al., 2003). The main purpose of this Thesis is the development and application of capillary electrophoretic methodologies for the in vitro enantioselective evaluation of pharmacological properties of drugs, in order to provide to the pharmaceutical industry fast and reliable alternatives for the preclinical stages of drug development. It is divided into three main parts: development of methods and theoretical studies for chiral separations of xenobiotics by CE (Papers I-IV), evaluation of the enantioselective binding of drugs to plasma proteins (Papers V-VII) and study of enzymatic reactions by electrophoretically mediated microanalysis, EMMA (Papers VIII-X). A powerful chiral selector, the highly sulfated β-cyclodextrin (HS-β-CD), has been widely used along this Thesis to perform enantioseparations. So, the first and second studies of this Thesis are theoretical; about its enantiorecognition ability. In Paper I, apparent binding constants between HS-β-CD and enantiomers of eight psychoactive drugs were estimated by electrokinetic chromatography-complete filling technique (EKC-CFT). The experimental mobility data of both enantiomers were fitted vs. the cyclodextrin concentration using a non-linear model that allows the estimation of apparent binding constants and overall limit mobilities. From these estimations, it was possible to calculate the thermodynamic (αt) and electrophoretic (αe) enantioselectivities of the separation process. The results show that this cyclodextrin strongly binds to the enantiomers of these basic drugs, with high enantioselectivity in most of the cases. The binding constants of HS-β-CD to the studied drugs are equal or higher than those found in the literature for other compounds (Vaccher et al., 2005, Lipka et al., 2007). Both thermodynamic (due to the difference between the apparent binding constants of the enantiomers) and electrophoretic (related to the difference between the overall complex enantiomer-CD mobilities) enantioselectivities were found to be responsible for the enantioseparation with the HS-β-CD, with a major contribution of αt . However, in one of the cases, where no thermodynamic enantioselectivity was found, only the electrophoretic one was enough to achieve complete separation of the enantiomers. The use of complete filling technique for the evaluation of binding constants, which is discussed in this Thesis, is presented as an alternative to conventional electrokinetic chromatography with a saving on chiral selector of around 99%. The second Paper included in this Thesis is a quantitative structure-property relationship study that models the enantiorecognition ability of the HS-β-CD for basic compounds in EKC, by relating the resolution obtained for forty xenobiotics (drugs and pesticides) with some of their structural properties. This is an attempt to provide researchers in the field of enantioseparations an alternative to the usually employed trial-and-error procedures for the development of chiral methods. A discriminant-partial least squares analysis over a single response variable (DPLS1) has been performed using a categorical resolution (Rs) value as y-vector and the structural properties of the xenobiotics as X-matrix. Only four structural variables were selected as significant in the final model: logarithm of octanol-water partition coefficient estimated at pH 7.4 (lgD), polar surface area (PSA), number of hydrogen bond donors (HBD) and acceptors (HBA). From the de-scaled coefficients of the model, an explicit equation that relates the categorical Rs with these four variables was obtained: Rs = 2.354 + 0.095lgD – 0.008PSA – 0.159HBD – 0.137HBA. This model shows an explained variance of 72.4%. For those xenobiotics whose predicted Rs is not so high (i.e., expected chiral separation only under concrete experimental conditions) a multivariate optimization of the most influential experimental conditions affecting enantioseparation (cyclodextrin concentration, pH of the background electrolyte (BGE) and capillary temperature) has been proposed, via a Box-Behnken experimental design and a posterior adjust of the experimental results to a PLS2 model, which allows estimating the migration time and the Rs for each set of experimental conditions. Paper III is an application of the enantioseparation of the antidepressant fluoxetine with HS-β-CD to the chiral analysis of three pharmaceutical formulations. The separation conditions were adjusted in order to avoid a matrix interference and a suitable EKC-CFT method has been obtained, which allows the quantification of both fluoxetine enantiomers in less than 2 min with a Rs of 2.4. A partial validation of the chiral analytical method was performed following the guidance of the International Conference on Harmonization (ICH) and the US Pharmacopeia in terms of identity, linearity, precision and accuracy. Good analytical features were obtained: determination coefficients (R2) of 0.996 for the calibration curves of both enantiomers, relative standard deviations (RSD) less than 20%, obtained under intermediate precision conditions, and recoveries of 92 ± 13 and 105 ± 6 % for the first and second eluted enantiomers, respectively. For all samples, the content of racemic fluoxetine was within specifications of the US Pharmacopeia (20 mg ± 15%), and stereoisomer ratios were around 1. So, the suitability of HS-β-CD for the enantioseparation of basic drugs has been proved here, not only with theoretical studies but also with an application to chiral pharmaceutical analysis. An approach of non-aqueous capillary electrophoresis (NACE) to the chiral separation of drugs, developed during a research stay in the Katholieke Universiteit of Leuven, is described in Paper IV. It responded to the interest on developing analysis methods based on non-aqueous systems that could be able to determine new active pharmaceutical ingredients (APIs) which are non-soluble in aqueous or methanolic mediums. Many drugs in the pipeline suggested by combinatorial chemistry show low aqueous solubility due to high molecular weight. If a badly soluble lead molecule has very good potency and low toxicity, the drug might go into further development anyway, and in these cases chiral methods for their analytical control may be developed. In this work, the use of dimethyl sulfoxide (DMSO) is proposed as an alternative to the most commonly used BGEs of alcoholic nature, due to its lower volatility and hazardousness, and better solvent properties for drugs with low polarity. However, it may be taken into account that the enantioseparation is disfavored in the DMSO medium because the intermolecular forces responsible for the enantiorecognition are weaker in organic media (Chankvetadze and Blaschke, 2001). The enantioseparation of three drugs, verapamil, pindolol and fenfluramine, has been performed with this NACE methodology, using univariate optimization in order to select the best separation conditions. Finally, DMSO-based BGEs containing a variable amount of methanol (0-30%), 0.5-1 M HAc, 125 mM NH4Ac and 40 mM carboxymethyl-γ-cyclodextrin were selected, with a capillary temperature of 15 ºC and a separation voltage of 30 kV. Using these conditions, Rs values of 1.5, 2.0 and 1.2 were obtained for verapamil, pindolol and fenfluramine, respectively. The electric current was stable during all the separations performed, despite the fact that one of the most common drawbacks of NACE is current breakdowns. Papers V-VII of this Thesis deal with the enantioselective evaluation of an important pharmacokinetic property at the level of drugs distribution, the protein binding. The amount of drug that binds to plasma proteins, as well as the strength of this binding, influences considerably the availability of the drug for reaching target organs and also for being eliminated. As plasma proteins are optically active molecules, chiral drugs can show different binding properties between enantiomers. Therefore, the development of fast and reliable procedures for the in vitro evaluation of enantioselective drug-protein binding is interesting for the pharmaceutical industry research. The most abundant protein in human plasma is albumin (HSA), and it presents a high degree of enantioselectivity in its binding with drugs. In this Thesis, the evaluation of enantioselective binding of drugs to HSA and total plasma proteins has been performed by ultrafiltration followed by chiral analysis by EKC. In ultrafiltration, a pre-equilibrated drug-protein mixture is filtered through a membrane that retains the protein and only allows the pass of the unbound drug fraction. This unbound fraction is then analyzed, here using chiral EKC. Paper V points out some weak aspects of the experimental methodologies and mathematical models previously published for the study of protein binding (Katrahalli et al., 2010; Fielding et al., 2005), and proposes a novel strategy for a reliable binding evaluation. An extended discussion concerning the different mathematical approaches to protein binding is performed, using the enantioselective binding of the antidepressant fluoxetine to HSA as an example. The analysis of fluoxetine enantiomers has been performed here using HS-β-CD as chiral selector in CFT, in a 30 mM phosphate buffer of pH 7.0. The separation temperature and voltage have been fixed at 30 ºC and 15 kV. The proposed strategy implies working close to physiological conditions, thus keeping always the ratio between the drug concentration (D) and the protein concentration (P) lower than 0.5, and P near the physiological level (around 475 μM). In this situation, a simple binding model that assumes that only one kind of binding site is occupied in the protein molecule and the stoichiometry of the binding is 1:1 can be used, always confirming these assumptions via other equations. Also, data were inspected in order to detect and eliminate outliers that could affect the reliability of the estimations. Enantiomer-protein binding constants (K1) have been estimated for fluoxetine, as well as the enantioselectivity, defined as the ratio between binding constants. Finally, the percentage of protein binding (PB%), which is an interesting parameter from a pharmacokinetic point of view, has been also estimated for both fluoxetine enantiomers, being 95.2 and 90.0% for the first and second eluted enantiomers, respectively. The correspondent enantioselectivity is around 2, in favor of the first eluted enantiomer. Still in the field of enantioselective drug-protein interactions, Papers VI and VII study the enantioselective binding of the hypnotic zopiclone and the antidepressant nomifensine to HSA and total plasma proteins, also by ultrafiltration and chiral EKC using the partial (PFT) or complete filling techniques. In Paper VI, the enantioseparation of zopiclone with carboxymethyl-β-cyclodextrin (CM-β-CD) is adjusted for the analysis of the unbound fractions after ultrafiltration. A solution containing 30 mM CM-β-CD in 50 mM Tris buffer at pH 6.0 has been used as chiral selector in PFT, injecting it at 0.5 psi during 99 s prior to samples injection. The chiral separation has been performed at 25 ºC by applying 15 kV. The mathematical approaches and experimental designs developed in Paper V have been also applied here to calculate the binding constants, enantioselectivity and protein binding. The PB% values to HSA were found to be 47 and 36% for S- and R-zopiclone, respectively, and to total plasma proteins 45 and 49%, respectively. These values were in agreement with those reported in the literature for racemic zopiclone, around 45% (Gaillot et al., 1983; Spanish Drug Agency, 2011). The enantioselective binding of nomifensine to HSA and total plasma proteins is evaluated in Paper VII, in this case using 30 mM heptakis-(2,3,6-O-methyl)-β-cyclodextrin as chiral selector in CFT for the enantioseparation of the unbound fraction, with the same BGE and voltage as in Paper VI but with a capillary temperature of 50 ºC. The experimental design of calibration curves and samples was repeated in two independent sessions in order to check the inter-day precision of the procedure. As no significant differences were found between the data of both working sessions, all data were used together for estimations. The PB% to HSA was estimated at 40 and 63% for the first and second eluted enantiomers respectively. For the binding to total plasma proteins, the estimated values were 63 and 64% respectively. By comparing these values, it can be concluded that the second eluted enantiomer of nomifensine binds mainly to HSA while the first enantiomer binds also to other plasma proteins. The experimental design and mathematical approaches proposed in this Thesis for a reliable estimation of enantioselective protein binding with ultrafiltration and chiral EKC have demonstrated to be a robust tool which allows obtaining good in vitro estimates in near-physiological conditions, thus providing valuable data from a pharmacokinetic point of view. Enantioselective protein binding data for fluoxetine, zopiclone and nomifensine have been estimated here, thus completing the information available in the literature about these drugs. The third part of this Thesis is related to the development of in-capillary enzymatic reactions by electrophoretically mediated microanalysis (EMMA). In EMMA the reagents are sequentially introduced in an electrophoretic capillary and left to mix and react for a fixed time. The separation of substrates, enzymes and products is then performed by electrophoresis. The main advantage of EMMA is that the reaction takes place in a nanoliter scale, so the consumption of reagents is extremely low. Furthermore, it is fully automated since the reaction and the separation take part in a unique step inside the capillary, which is another important advantage for screening procedures. In Paper VIII an EMMA screening methodology for the evaluation of acetylcholinesterase (AChE) inhibitors has been developed and applied to the study of five drugs with known inhibitory activity. Tacrine has been used as model compound in order to optimize the method conditions. The progress of the enzymatic reaction of the hydrolysis of acetylthiocholine in presence of AChE and the inhibitors has been determined by measuring the peak area of the reaction product thiocholine. A “sandwich” injection has been performed by injecting the substrate plug between two enzyme plugs. The plugs’ mixing has been performed by simple diffusion, with an incubation time of 2 min. The BGE for the reaction and separation was 30 mM borate-phosphate buffer at pH 8.0, the capillary temperature has been set at 37 ºC and the separation voltage at 15 kV. Due to the dilution of solutes in the EMMA methodology, a correction may be performed in calculations to achieve accurate results. Here, a dilution factor is estimated by comparing the calculated inhibitory potency of tacrine with that reported with the reference method (Andrisano et al., 2001), and applied then to all calculations. The proposed screening methodology has proved to be useful for the estimation of inhibitory constants, inhibitory potency, a qualitative approach to the inhibition mechanism and the percentage of inhibition of tacrine, edrophonium, neostigmine, pyridostigmine and eserine. The screening results show that according to their percentage of inhibition under fixed conditions, the order of inhibitor potency for these compounds is tacrine > edrophonium > neostigmine > eserine > pyridostigmine. This screening procedure, with an analysis time shorter than 5 min, can be employed for the evaluation of new potential AChE inhibitors. Continuing with the use of EMMA methodologies for the evaluation of drug-enzyme interactions, Papers IX and X are studies of enantioselective metabolism by different isoforms of the cytochrome P450. In these studies, the capillary acts simultaneously as reactor and chiral separation system, thus providing enantioselective information with a fast, fully automated assay. The chiral separation is performed in both cases using HS-β-CD and the partial filling technique. Paper IX deals with the enantioselective metabolism of the Ca+2-channel blocker verapamil by CYP3A4, the most important isoform of cytochrome P450 in humans. Different possibilities for the EMMA assay have been considered and variables have been optimized prior to calculation of the Michaelis-Menten parameters for the enantioselective metabolism kinetics. The selected experimental conditions for the EMMA assay included a “sandwich” injection of the substrate plug between two enzyme plugs, and two plugs of incubation buffer before and after the “sandwich” in order to give to the enzyme its adequate reaction medium. The mixing of plugs has been performed by simple diffusion during an incubation time of 5 min. After that, the separation of all the reaction mixture, including the enantiomers of verapamil and its major metabolite norverapamil, has been performed in the same capillary with the following conditions: 50 mM phosphate at pH 8.8 as BGE, 2.5% HS-β-CD injected at 10 psi/2 min, 25ºC and 20 kV with a little pressure of 0.2 psi in order to shorter the migration times. The result is a fully automated methodology that allows the enantioselective evaluation of verapamil metabolism by CYP3A4 in less than 35 min including capillary preconditioning, injection, incubation and separation. Michaelis-Menten parameters (Km and Vmax) have been calculated for the overall enantioselective metabolism of verapamil by CYP3A4. Km values have been estimated as 51 ± 9 and 47 ± 9 μM for S- and R-verapamil, respectively, while Vmax estimates were 22 ± 2 and 21 ± 2 pmol•min-1•(pmol CYP)-1 for S- and R-verapamil, respectively. Intrinsic clearances, defined as the amount of plasma from which a xenobiotic is removed in a certain time, and estimated in vitro as Vmax/Km (Kroemer et al., 1992), have been also calculated for both enantiomers. The enantioselective metabolism of fluoxetine by other CYP450 isoform, CYP2D6, is estimated in Paper X following the same methodology as in the previous work. The EMMA conditions from Paper IX were directly applied, while the separation conditions were adjusted in order to separate the enantiomers of fluoxetine and its major metabolite norfluoxetine. The influence of the most important variable on the enantioseparation, the concentration of HS-β-CD, was studied and a final concentration of 1.25% was selected for the separation of the reaction mixture. The other separation conditions were kept as in Paper IX. Kinetic parameters have been estimated by non-linear fitting of experimental data to the Michaelis-Menten equation. Km values were found to be 30 ± 3 μM for S-fluoxetine and 39 ± 5 μM for R-fluoxetine. Vmax estimates were 28.6 ± 1.2 and 34 ± 2 pmol•min-1•(pmol CYP)-1 for S- and R-fluoxetine, respectively. The kinetic assay was repeated using a single enantiomer (R-fluoxetine) instead of the racemic mixture. Similar results were obtained indicating that the use of the racemate is a good option for obtaining enantioselective estimations. The usefulness of EMMA as screening methodology for the in vitro evaluation, enantioselective or not, of drug-enzyme interactions has been proved in this Thesis. The developed methods allow to estimate kinetic and inhibition parameters in fast assays with low consumption of reagents and samples. The coupling of EMMA with the enantioseparation of substrates and products in a unique step carried out in the CE system provides valuable enantioselective information about the pharmacokinetic or pharmacodynamic processes evaluated. All the methods proposed and applied in this Thesis (chiral separations, prediction models, enantioselective protein binding evaluation and drug-enzyme interaction) can be applied in the pharmaceutical industry to obtain useful in vitro information in the preclinical phases of drug development and for the enantioselective quality control of raw materials, synthesis intermediates and final products. References:- Andrisano, V., Bartolini, M., Gotti, R., Cavrini, V., Felix, G. (2001) Determination of inhibitors¿ potency (IC50) by a direct high-performance liquid chromatographic method on an immobilised acetylcholinesterase column. Journal of Chromatography B 753, 375-383. - Brocks, D.R. (2006) Drug disposition in three dimensions: an update on stereoselectivity in pharmacokinetics. Biopharmaceutics and Drug Disposition 27, 387-406. - Chankvetadze, B., Blaschke, G. (2001) Enantioseparations in capillary electromigration techniques: Recent developments and future trends. Journal of Chromatography A 906, 309-363. - European Medicines Agency. (1993) Investigation of Chiral Active Substances. Available in: http://www.ema.europa.eu/docs/en_GB/document_library/Scientific_guideline/2009/09/WC500002816.pdf [accessed February 2013] - Fielding, L., Rutherford, S., Fletcher, D. (2005) Determination of protein-ligand binding affinity by NMR: observations from serum albumin model systems. Magnetic Resonance in Chemistry 43, 463-470. - Food and Drug Administration. (1992) Development of New Stereoisomeric Drugs. Available in: http://www.fda.gov/drugs/guidancecomplianceregulatoryinformation/guidances/ucm122883.htm [accessed February 2013] - Gaillot, J., Houghton, G.W., Marc Aurele, J., Dreyfus, J.F. (1983) Pharmacokinetics and metabolism of zopiclone. Pharmacology 27, 76-91. - Katrahalli, U., Jaldappagari, S., Kalanur, S.S. (2010) Probing the binding of fluoxetine hydrochloride to human serum albumin by multispectroscopic techniques. Spectrochimica Acta Part A 75, 314-319. - Kroemer, H.K., Echizen, H., Heidemann, H., Eichelbaum, M. (1992) Predictability of the in vivo metabolism of verapamil from in vitro data: contribution of individual metabolic pathways and stereoselective aspects. Journal of Pharmacology and Experimental Therapeutics 260, 1052-1057. - Lin, C.C., Li, Y.T., Chen, S.H. (2003) Recent progress in pharmacokinetic applications of capillary electrophoresis. Electrophoresis 24, 4106-4115. - Lipka, E., Danel, C., Orhan, H., Bonte, J.P., Vaccher, C. (2007) Chiral resolution of melatoninergic ligands by EKC using highly sulfated CDs. Electrophoresis 228, 3915-3921. - Spanish Drug Agency (Agencia Española del Medicamento). Available in: http://www.aemps.es [accessed March 2011] - Vaccher, M.P., Lipka, E., Bonte, J.P., Foulon, C., Gossens, J.F., Vaccher C. (2005) Enantiomeric analysis of baclofen analogs by capillary zone electrophoresis, using highly sulfated cyclodextrins: inclusion ionization constant pKa determination. Electrophoresis 26, 2974-2983.