Preparation of Barbiturate Optical Isomers and Their Effects on GABA (^A) Receptors

We prepared milligram quantities of the enantiomers of hexobarbital, pentobarbital, and thiopental using high-performance liquid chromatography (HPLC). We used a column (250 mm long, 10 mm ID, about 20 ml capacity) prepared by Macherey-Nagel, which consisted of a stationary chiral phase of permethylated [Greek small letter beta]-cyclodextrin covalently bonded to silica (5 [micro sign]m particle size, Nucleodex [Greek small letter beta]-PM). The column was temperature-controlled and held at 20-22 [degree sign]C. The mobile phase was a volatile buffer solution composed of 65% (v/v) methanol (HPLC Hipersolv grade, BDH Laboratory Supplies, Poole, Dorset, UK) and 35% (v/v) 0.1% or 1.0% triethylammonium acetate (HPLC grade, Applied Biosystems, Foster City, CA) in water titrated to either pH 7.0 or pH 4.0 with acetic acid. The mobile phase was filtered through a 0.2-[micro sign]m nylon 66 filter (Anachem, Luton, Bedfordshire, UK). The racemic barbiturates (Sigma Chemical, Poole, Dorset, UK) were obtained as sodium salts, in the cases of pentobarbital and thiopental, or as the free acid in the case of hexobarbital. Further details of the HPLC procedures are given in Table 1. Barbiturate solutions were filtered prior to injection with a 0.22-[micro sign]m Millex-GV4 filter (Millipore, Bedford, MA), and the flow rate through the column was 2 ml/min. After column separation, the enantiomers were dried by either a rotary or centrifugal evaporator. Both thiopental enantiomers, as well as the R(+)-pentobarbital, required a second column purification to achieve the highest levels of optical purity (>or= to 99.0%).

Mouse fibroblast L cells stably transfected [16]with bovine [Greek small letter alpha]¹[Greek small letter beta]¹[Greek small letter gamma]^2Lsubunits of the GABA^Areceptor were kindly supplied by P. Whiting (Merck Sharp and Dohme Research Laboratories, Harlow, Essex, UK). These PA3 cells were grown for electrophysiological experiments using methods exactly as described previously. [17] 

Inward currents, caused by the efflux of chloride ions, evoked by the application of GABA to PA3 cells were recorded using the standard whole-cell patch-clamp technique with an Axopath 200 amplifier (Axon Instruments, Foster City, CA). Recording pipettes were fabricated from thin-walled filamented borosilicate glass capillary tubes (GC150TF, Clark Electromedical Instruments, Reading, Berkshire, UK) using a two-stage pull (Narishige PB-7 micropipette puller, Tokyo, Japan) and lightly fire-polished to give typical electrode resistances of about 5 M Omega. Pipettes filled with intracellular recording solution containing 130 mM CsCl, 1 mM MgCl², 10 mM HEPES, and 11 mM EGTA (titrated to pH 7.2 with CsOH) easily formed "gigaohm" seals with the cells. Once the whole-cell configuration was established, the cells were voltage-clamped at -40 mV. During recording, the PA3 cells were bathed in an extracellular solution containing 124 mM NaCl, 5 mM KCl, 2 mM CaCl², 1 mM MgCl², 5 mM HEPES, and 11 mM D-glucose (titrated to pH 7.4 with NaOH). Membrane currents were filtered at 50 Hz (-3 dB; eight-pole Bessel Filter, FE-301-SF, Fylde Electronic Laboratories, Preston, Lancashire, UK). Membrane voltages were compensated for series resistance by more than 80% for all recordings. All drugs were applied rapidly via a double-barrelled glass capillary tube as described previously. [18]Experiments were performed at room temperature (20-23 [degree sign]C).

Experiments were carried out using a low concentration (1 [micro sign]M) of GABA, which under control conditions induced a nondesensitizing current with a magnitude of approximately 1% of the maximal current. Pairs of stable GABA-activated control currents were recorded before and after application of the anesthetic and averaged to give a control response. At concentrations at which the barbiturate was found to have a significant effect on the baseline current, it was pre-applied before the application of GABA (in order to correct for direct barbiturate-gated currents by establishing an accurate baseline for GABA-dependent potentiation). Only in the case of thiopental at concentrations greater than or equal to 25 [micro sign]M was this found to be necessary, and even then the direct effects were very small-for example, 40 [micro sign]M thiopental activated a current only 13 +/- 2% (n = 7) of that induced by 1 [micro sign]M GABA. In all other cases, GABA and the barbiturate were coapplied. The order of application of the isomers was varied from cell to cell. Unless otherwise stated, all chemicals were obtained from Sigma Chemical.

Anesthetic potentiation was defined as 100 x (I-I⁰)/I⁰, where I⁰is the peak of the control GABA-induced current and I is the peak of the GABA-induced current in the presence of anesthetic.

Values throughout this article are given as mean +/- SEM. Statistical significance was assessed using the Student t test.

The HPLC protocol described in the Materials and Methods section gave close to baseline separation for the three pairs of anesthetic barbiturates we studied, and milligram quantities of the enantiomers could be prepared relatively easily. Typical column separations are shown in Figure 2. For the cases of hexobarbital and pentobarbital the preparation of the enantiomers was particularly straightforward because of the excellent separation and relatively short column retention times. For the thiopental enantiomers, the appearance of a contaminating peak (peak 1 in Figure 2C) and the longer retention times made the separation slightly more difficult. The first small peak, which consists of about 3-4% of the total material, is almost certainly the structural isomer 5-ethyl-5-(1-ethylpropyl)-2-thiobarbituric acid. [19]For all three of the barbiturates, the enantiomers could be prepared to high purities (99.0-99.8%). The identities of the major peaks in the chromatograms could be readily determined by measuring the optical rotations with a polarimeter. Values for the specific optical rotations (degrees of rotation of plane-polarized light at a concentration of 1 g/cm³and for a pathlength of 10 cm) we observed for the enantiomers were -13.0 and +13.2 for hexobarbital, -14.5 and +14.0 for pentobarbital, and -9.8 and +9.8 for thiopental. Figure 3shows chromatograms for the purified thiopental enantiomers.

All of the barbiturates markedly potentiated the current induced in PA3 cells by the application of a low concentration (1 [micro sign]M) of GABA. EC⁵⁰concentrations for general anesthesia in mammals have been estimated [15]for both pentobarbital (50 [micro sign]M) and thiopental (25 [micro sign]M), and at these concentrations the racemic mixtures of the barbiturates potentiated the current (data not shown) to roughly the same extent (940 +/- 50% and 1,200 +/- 140%, respectively). For all three barbiturates, however, the S isomer was substantially more effective than the R isomer. These data are illustrated in Figure 4, which shows the concentration-dependence of the potentiation for all six enantiomers over the range of concentrations that is likely to be of pharmacological relevance. Because the observed potentiation for a given barbiturate concentration varied substantially between different cells, maximum precision in estimating the degree of stereoselectivity for the barbiturate enantiomers was obtained by measuring the ratios of potentiations for individual cells exposed to fixed concentrations of both enantiomers (Figure 5). The degree of stereoselectivity (defined as the ratio of the potentiations of the S and the R isomers) was, averaged over the concentration ranges investigated, 2.3 for hexobarbital, 3.2 for pentobarbital, and 1.8 for thiopental, although for pentobarbital there was a small but statistically significant (P < 0.05) decrease in stereoselectivity with increasing concentration (see Figure 5B).

The ease with which we have been able to prepare milligram quantities of highly pure enantiomers of hexobarbital, pentobarbital, and thiopental suggests that this preparative HPLC method could be readily extended to other barbiturates. Although the separation on an analytical scale of the enantiomers of barbiturates with a chiral carbon atom in the pyrimidinetrione ring (such as hexobarbital, see Figure 1) has been reported several times, [10-13]the difficulties in separating enantiomers in which the chiral carbon atom is in the butyl or pentyl aliphatic chain are well known, and good separations have not often been achieved. [11,12]However, the separations we obtained with pentobarbital and thiopental were sufficient for rapid purification on a preparative scale.

Before considering the effects of the barbiturate enantiomers on the GABA^Areceptor, or indeed on any other putative anesthetic target, it is important to make some reference to those concentrations that are pharmacologically relevant. For the cases of thiopental and pentobarbital, good estimates are available for the concentrations that are needed to achieve a variety of anesthetic endpoints in mammals. Of most relevance to surgical anesthesia are the EC⁵⁰concentrations required to abolish a response to a painful stimulus, and these concentrations (in free aqueous solution) are 25 [micro sign]M for racemic thiopental and 50 [micro sign]M for racemic pentobarbital. [15]No directly equivalent estimate exists, however, for racemic hexobarbital. Nonetheless, for many general anesthetics there appears to be a reasonable agreement between the EC⁵⁰concentrations needed to abolish a painful stimulus in mammals, and those for a loss of righting reflex in tadpoles. [15,20-24]For hexobarbital, this latter value is about 100 [micro sign]M, [22]so we might take this concentration to be a reasonable benchmark, in lieu of more definitive measurements in mammals.

As can be seen from the data in Figure 4, at the relevant EC⁵⁰concentrations for general anesthesia, all six of the barbiturate enantiomers caused substantial potentiations of the current induced by a low concentration of GABA. It is well known that barbiturates act at the GABA^Areceptor by increasing the apparent affinity of GABA for its receptor (see, for example, [25]), so it is only fair to point out that our experimental protocol (using a low concentration of agonist) is designed to maximize the anesthetic effect. In addition, our protocol differs from the physiologic situation at inhibitory GABAergic synapses, where high agonist concentrations are encountered for very short times and the anesthetic effect of increasing apparent affinity for GABA manifests itself as a prolongation of inhibitory postsynaptic current (see, for example, [26]). It therefore seems likely that the potentiations we have observed with our protocol at pharmacologically relevant concentrations of barbiturates would translate into substantial increases in charge transfer at inhibitory GABAergic synapses.

For each of the barbiturates, the S isomer was more effective than the R isomer by a factor of roughly two or three (see Figure 5). This may reflect differences in potency, efficacy, or both. The larger effects observed with the S isomers of hexobarbital and pentobarbital are consistent with the results of the earliest studies, [27,28]as well as the comprehensive analysis on the allosteric effects of barbiturates performed by Olsen et al. [29]Surprisingly, however, no data exist, at least to our knowledge, on the effects of the thiopental enantiomers on GABA^Areceptor function. The data in Figure 4C and Figure 5C show that, as with the other barbiturates, the S isomer of thiopental is also the more effective, again by a factor of roughly two.

How can these in vitro data be compared with potency measurements of barbiturate enantiomers in mammals? The most obvious problem that must be addressed is the extent to which the pharmacokinetics of drug delivery and elimination affect the observed stereoselectivity in animal potencies. This is a complex issue, and it is difficult to be definitive, particularly for drugs that are almost entirely eliminated by hepatic biotransformation. Nonetheless, although redistribution and metabolism greatly affect the duration of anesthesia (i.e., sleeping times), they should have relatively smaller effects on the rapid loss of righting reflex following a bolus injection. The main pharmacokinetic factors that affect this measurement of anesthetic potency are the rate of transfer across the blood-brain barrier, which is largely determined by lipid solubility, and binding to plasma proteins. An overall assessment of the importance of these factors on the observed stereoselectivity of animal potencies can be made by comparing the plasma and brain concentrations of the enantiomers achieved following an equal bolus dose. The available evidence strongly suggests that under these conditions the concentrations of the S and R isomers of a particular barbiturate achieve very similar concentrations in the brain, [9,30-36]yet produce different anesthetic responses.

So, as a first approximation, we can interpret the stereoselectivity observed for the barbiturates acting on animals to cause a loss of righting reflex as reflecting their intrinsic effectiveness at their relevant target sites. Reliable values of the relative potencies of the S(-) and R(+) isomers of thiopental [6,7]and pentobarbital [7,8]in causing a loss of righting reflex in mice are available and are 1.7 and 1.9, respectively. For hexobarbital, the S(+) isomer has been shown to be over 2.5 times more effective than the R(-) isomer in anesthetizing rats, [1,37]as measured using an electroencephalographic criterion (the dose needed to produce a burst suppression of at least 1 s). These degrees of stereoselectivity in animals should provide an important guide as to which molecular targets are most likely to be relevant for barbiturate-induced general anesthesia. A comparison with the data shown in Figure 4and Figure 5supports the already strong case [14,15]that the GABA^Areceptor is very likely to be an important target, because the rank order and approximate degree of stereoselectivity observed for all three barbiturates in animals can be largely accounted for by actions at this receptor. In contrast, the extreme insensitivity and complete lack of stereoselectivity observed for pentobarbital acting on voltage-gated sodium channels [38]makes it very implausible that these are relevant sites of action. This does not, of course, rule out contributions from other targets, and there is good evidence that certain non-NMDA glutamate receptors are sensitive to reasonable levels of pentobarbital. [39]Barbiturate enantiomers should be extremely useful in assessing the possible roles of these and other putative targets in barbiturate anesthesia, and this will be the subject of a future study.

What can we learn from our results about the molecular mechanisms involved in the actions of the barbiturates? Although it is possible, in principle, that the barbiturate enantiomers could partition stereoselectively into lipid bilayers, when this has been studied with other anesthetics (isoflurane and etomidate) no measurable stereoselectivity has been observed. [17,40,41]The degree of stereoselectivity that we have observed can, we believe, only plausibly be interpreted in terms of these anesthetics binding directly to the GABA^Areceptor protein.

In previous studies on the interactions of isoflurane [42]and etomidate [17]enantiomers with GABA^Areceptors, we found that the degree of stereoselectivity increased significantly with anesthetic concentration. We interpreted this as reflecting a change in the structure of the anesthetic binding sites, as increasing anesthetic concentrations favored on allosteric conformational change in the receptor (this conformational change can safely be presumed to occur in order to account for the increased apparent affinity of the receptor for its agonist). For the barbiturate enantiomers, a slightly different picture emerges because the data in Figure 5show that there is little change in the degree of stereoselectivity with anesthetic concentration over the range of concentrations tested (for pentobarbital the change is significant, but small). If this interpretation is correct, then one must conclude that the binding environment of the barbiturates remains unchanged during the allosteric conformational change, which in turn suggests that the barbiturates bind to a site distinct from that of either isoflurane or etomidate. However, we do not consider this to be a particularly strong argument, and more direct information on the extent to which anesthetics occupy unique or overlapping sites is likely to come from mutagenesis studies. [43,44]Furthermore, it must not be forgotten that the chiral effects we have observed may be caused partly by gating effects, such as stereoselective differences in the degree to which bound barbiturates favor opening of the GABA-complexed receptor channel. Whatever the details, there seems to be little doubt that the barbiturates exert their effects by binding to specific sites on the GABA^Areceptor, and that these effects play a major role in the anesthetic properties of these agents.

The authors thank the Medical Research Council for support. Ms. Tomlin was the recipient of an MRC Studentship. The authors also thank Dr. H. Riering and Mr. J. Akins for help and advice with the chromatography, Mr. T. Mayer for performing the polarimeter measurements, and Dr. P. Whiting for the stably transfected PA3 cells.