Chiral Discrimination of Inhalation Anesthetics and Methyl Propionates

Anal. Chem. 1997, 69, 4017-4031

Chiral Discrimination of Inhalation Anesthetics and Methyl Propionates by Thickness Shear Mode Resonators: New Insights into the Mechanisms of Enantioselectivity by Cyclodextrins K. Bodenho 1 fer,† A. Hierlemann,*,†,§ M. Juza,‡ V. Schurig,‡ and W. Go 1 pel†

Institute of Physical and Theoretical Chemistry, Center of Interface Analysis and Sensors, and Institute of Organic Chemistry, University of Tu¨ bingen, Auf der Morgenstelle 8, D-72076 Tu¨ bingen, Germany

The discrimination of the enantiomers of methyl lactate, methyl 2-chloropropionate, and the inhalation anesthetics enflurane, isoflurane, and desflurane in the gas phase has been performed using thickness shear mode resonators. The selective coating was a modified perpentylated γ-cyclodextrin derivative dissolved in a polysiloxane matrix. A new model for the sorption of the chiral compounds into the cyclodextrin cavities and into the polymer matrix was established for the purpose of characterizing the sensor responses. This characterization included the fitting of the sensor responses (preferential and nonpreferential sorption) according to the model and extracting the characteristic parameters. In particular we attempted to explain the observed variation of the chiral discrimination factor r with changing analyte or cyclodextrin concentrations and search for an invariable parameter, characteristic for a certain analyte-cyclodextrin combination. The process of chiral or “molecular” recognition was thoroughly investigated. Chiral discrimination is considered a very important principle in biological systems: the human nose is able to discriminate the different odors of, for example, the enantiomeric pairs of limonene1 or carvone.2 The enantiomers of the vast majority of chiral compounds, e.g., the lactates, however, cannot be recognized by olfaction. Enantioselectivity plays an important role in pharmacology as well: in the case of the chiral inhalation anesthetics, an enantiomeric bias has been proposed.3 Consequently, modern methods providing a chiral on-line analysis are desirable. Chiral discrimination constitutes one of the most difficult tasks in analytical chemistry, since the physical and chemical properties of enantiomers are identical in a nonchiral environment. Discrimination of enantiomers is achieved in gas chromatography (GC) by using different types of chiral recognition structures in the stationary phases. Besides the discrimination by chiral amide stationary phases based on hydrogen bonding (as described for sensors in ref 4 and GC in refs 5 and 6) and the coordination of the optically active compound to metal chelates, which are attached to polysiloxanes,7 one of the most popular methods †

Institute of Physical and Theoretical Chemistry. Institute of Organic Chemistry. § Present address: Sandia National Laboratories, Microsensor R&D, MS 1425, P.O. Box 5800, Albuquerque, NM 87185-1425. (1) Ide, J.; Nakamoto, T.; Moriizumi, T. Sens. Actuators A 1995, 49, 73-78. (2) Friedman, L.; Miller, J. G. Science 1971, 172, 1044. (3) Franks, N. P.; Lieb, W. R. Science 1991, 254, 427-430; Nature 1994, 367, 607-613. ‡

S0003-2700(97)00367-3 CCC: $14.00

© 1997 American Chemical Society

includes chiral recognition by intramolecular entrapment into cavities such as cyclodextrins (sensors;1 GC8,9). Here we report on the recognition of enantiomers by a sensor array, which takes advantage of such entrapment mechanisms. The chiral model receptor was a very versatile (the enantiomers of more than 300 compounds could be separated in GC using this receptor molecule10) modified γ-cyclodextrin derivative [octakis(3-O-butanoyl-2,6-di-O-n-pentyl)-γ-cyclodextrin, CD],11 which had been dissolved in a polysiloxane (SE-54) matrix (Figure 1). The first objective included the unambiguous discrimination of the enantiomers of two derivatives of propionic acid, methyl lactate (methyl 2-hydroxypropionate) and methyl 2-chloropropionate, by the chiral sensors. These analytes are commercially available; hence we could perform a comprehensive set of measurements. Following previous results on the chiral amide coatings,4 we performed the measurements at a ratio of recognition sites to analyte molecules of 10:1 [occupation ratio (OR)4 below 10%]. Since the cyclodextrin units have a larger molecular weight, the required analyte concentrations had to be appreciably lower. But even in this low concentration range we noticed, in contrast to the previously studied chiral amides,4 strong nonlinearities in the sensor responses within a confined concentration range and a pronounced concentration dependence of the chiral discrimination factor, R, as determined from the sensor responses. Following our definition of the Rsensor value as the ratio of ∆fS analyte/∆fR analyte in ref 4 we use, in the following, R values for methyl chloropropionate, where the S enantiomer is sorbed to a larger extent into the chiral matrix than the R enantiomer, and R′ values for methyl lactate and the anesthetics, where the reverse holds. Adjustment of different concentrations of the chiral recognition sites in the sorption matrix [10, 50, 100% (w/w)] and systematic variation of the analyte concentration in the gas phase within a broad range, (4) Bodenho ¨fer, K.; Hierlemann, A.; Seemann, J.; Gauglitz, G.; Go ¨pel, W. Anal. Chem. 1997, 69, 3058-3068. (5) Koppenhoefer, B.; Bayer, E. Chromatographia 1984, 19, 123. (6) Schreier, P.; Bernreuther, A.; Huffer, M. Analysis of chiral organic molecules: methodology and applications; Walter de Gruyter: Berlin, New York, 1995; Chapter 3.5.2.1, pp 151-175. (7) Schurig, V. In Chromatographic separations based on molecular recognition; Jinno, K., Ed.; Wiley: New York, 1996; Chapter 7, pp 371-418. (8) Meinwald, J.; Thompson, W. R.; Pearson, D. L.; Ko¨nig, W. A.; Runge, T.; Francke, W. Science 1991, 251, 560. (9) Snopek, J.; Smolkova-Keulemansova, E.; Cserhati, T.; Gahm, K.; Stalcup, A. In Comprehensive supramolecular chemistry; Szejtli, T., Osa, T., Eds.; Pergamon: New York, 1996; Vol. 3, Chapter 18, pp 515-571. (10) Ko ¨nig, W. A. J. High Resolut. Chromatogr. 1993, 16, 569-586. (11) Ko ¨nig, W. A.; Krebber, R.; Mischnick, P. J. High Resolut. Chromatogr. 1989, 11, 732-738.

Analytical Chemistry, Vol. 69, No. 19, October 1, 1997 4017

Figure 2. Analytes under investigation in this study. Only the respective (S)-enantiomers are shown.

Figure 1. Coating compounds used for chiral discrimination: nonchiral poly(etherurethane), and polysiloxane bearing phenyl, vinyl, and methyl groups (SE-54) as well as enantioselective octakis(3-Obutanoyl-2,6-di-O-n-pentyl)-γ-cyclodextrin.

which cannot be performed in GC measurements, allowed a thorough characterization of the occurring sorption and recognition processes. The application of gas sensors instead of GC methods here offered the unique advantage of defined sorption conditions: exactly adjustable analyte concentrations, steady state signals, and only one “theoretical” plate. The distinct concentration dependence of the chiral discrimination factors (R) was the starting point for more revealing considerations. In the Institute of Physical and Theoretical Chemistry we tried to set up a model for the sorption and chiral recognition processes by fitting the sensor responses over a broad concentration range on the basis of simple assumptions (see Results section). With the experience (cyclodextrin to analyte molecule ratio below or approximately 10:1) obtained from investigating the propionic acid derivatives, we tried to discriminate the enantiomers of the common inhalation anesthetics (Figure 2): enflurane [2-chloro-1-(difluoromethoxy)-1,1,2-trifluoroethane], isoflurane [2-chloro-2-(difluoromethoxy)-1,1,1-trifluoroethane], and desflurane [2-(difluoromethoxy)-1,1,1,2-tetrafluoroethane]. These chiral inhalation anesthetics are still produced and clinically administered as racemic mixtures, although the consequences of the (unintended) introduction of a chiral center into these haloethers and its impact on enantioselective biological reactions is still debated.3,12 As this situation may be subject to reconsideration in the future, the on-line recognition and monitoring of chiral inhalation anesthetics, as provided by gas sensors, could be of (12) Halpern, D. F. In Organofluorine Compounds in Medicinal Chemistry and Biomedical Applications; Filler, R., Kobayashi, Y., Yagupolskii, L. M., Eds.; Studies in Organic Chemistry 48; Elsevier: Amsterdam, 1993; p 125.

4018 Analytical Chemistry, Vol. 69, No. 19, October 1, 1997

great importance. For mechanistic studies, we additionally used the nonchiral inhalation anesthetic sevoflurane [fluoromethyl 2,2,2trifluoro-1-(trifluoromethyl)ethyl ether]. Since only very small quantities of the anesthetics in enantiomeric purity were available (preparative GC, around 100-200 µL), we just could establish the ability of our sensors to discriminate the enantiomers. For further systematic investigations (especially for the fits, see Results section), we therefore used racemic enflurane and isoflurane and the nonchiral sevoflurane. Signal transduction (chemical into electrical signal) was achieved by use of thickness shear mode resonators (TSMRs). The setup consisted of discrete TSMRs operating at a fundamental frequency of 30 MHz. As shown by Sauerbrey,13 the vibrating frequency of a quartz crystal changes to a first approximation proportionally to the mass deposited onto or removed from the surface. When the polymer-coated TSMRs are exposed to analyte gas, sorption of molecules into the polymer generates a change of the vibrating mass which causes a shift of the operating frequency. The sorption strength depends on the interaction mechanisms and forces between matrix and analyte molecule. Since the applied analyte concentrations and the TSMR fundamental frequency are rather low (30 MHz), viscosity effects are assumed to contribute only to a small extent to the frequency change due to gas sorption.14 In any case, the TSMR signal was linear proportional to the analyte content in the gas phase. Changes in conductivity do not effect the device response since the electrodes are located on the opposite faces of the quartz disk. In addition, such effects should not affect the chiral discrimination since they should be identical at comparable analyte concentrations of R and S enantiomer. For details about resonant mass sensitive devices, see refs 15-17. In addition to the various sets of chiral sensors [four of each cyclodextrin concentration: 10, 50, 100% (w/w) to provide statistical evidence (parallel approach)], four reference devices coated with the nonenantioselective poly(etherurethane) (PEUT) and SE54 were included into the array to recognize incidential artifacts caused by fluctuating gas phase concentrations or contaminations of the S and R enantiomers. EXPERIMENTAL SECTION Thickness Shear Mode Resonators. The TSMR array consisted of discrete piezoelectric quartz crystals (AT cut) with (13) Sauerbrey, G. Z. Phys. 1959, 155, 206-222. (14) Martin, S. J.; Frye, G. C.; Senturia, S. D. Anal. Chem. 1994, 66, 2201. (15) Grate, J. W.; Martin, S. J.; White, R. M. Anal. Chem. 1993, 65, 940A948A, 987A-996A. (16) Nieuwenhuizen, M. S.; Venema, A. Sens. Mater. 1989, 5, 261-300. (17) Bodenho ¨fer, K.; Hierlemann, A.; Noetzel, G.; Weimar, U.; Go ¨pel, W. Anal. Chem. 1996, 68, 2210-2218.

gold electrodes operating at a fundamental frequency of 30 MHz (quartz plate thickness, 55.6 µm) purchased from Kristallverarbeitung (KVG), Neckarbischofsheim, Germany. Each crystal was powered by an oscillator circuit (bipolar, parallel resonance) constructed in our laboratory. Only one single coaxial cable is required for voltage supply and signal transmission. A selfdeveloped scanner (up to 16 channels) operating at frequencies between 100 kHz and 100 MHz was controlled by a PCL 726 interface card (Labtech, Wilmington, MA) in an IBM-compatible PC-AT and allowed for the sequential monitoring of each TSMR output using a Hewlett-Packard 5334 B frequency counter. The computer acquired the frequency values via an IEEE 488 interface bus. The first monitored frequency value of each device was set equal to zero; hence the frequency differences were monitored in reference to this first value. The frequency outputs of the TSMRs were recorded every 30 s at 0.1 Hz resolution. The gate time of the counter (HP 5334 B) was set to 1 s for all the devices. The absolute sensor responses were given by the frequency difference between gas exposure and purging. Sensor Coatings. The adsorptive films consisted of a cyclodextrin derivative diluted in a polysiloxane matrix, which previously has been applied to enantioseparation in capillary gas chromatography.18 The chosen γ-cyclodextrin derivative was octakis(3-O-butanoyl-2,6-di-O-n-pentyl)-γ-cyclodextrin (Figure 1).11 The matrix was a polysiloxane backbone bearing phenyl, vinyl, and methyl groups, which is commercially available from Macherey and Nagel, Du¨ren, Germany (SE-54, Figure 1). Besides the pure compounds (CD, SE-54) mixtures with 50 and 10% (w/w) cyclodextrin contents were used in this study. Since the molar weight of the cyclodextrins is high compared to that of the polymer repeating units, the molar ratio of cyclodextrin molecules to polymer segments is quite small. Besides these coatings, the sensor array included the nonchiral poly(etherurethane) (Thermedics Inc., Woburn, MA) to check the reproducibility of the gas flow and to additionally monitor the exact analyte concentration. We tried to use poly(dimethylsiloxane) (SE-30 and SE-54) as a reference; however, by applying these coatings, the signals of all the analytes were very small compared to that of the chiral coating. Hence it was impossible to use SE30 (SE-54) as a reference in the low-concentration range, which was crucial for the investigations described in the following sections. For coating TSMRs, the polymers (or polymer solutions) were dissolved in dichloromethane (concentrations, ∼1 mg/mL). The solutions were sprayed onto the cleaned devices with an air brush using pure nitrogen as a carrier gas. On-line monitoring of the frequency decrease allowed us to determine the frequency shifts due to coating and to calculate the layer thickness by assuming a uniform and homogenous distribution of the polymer over the sensitive surface. The obtained frequency shifts due to coating were about 60-100 kHz (d ) 300-500 nm) for the different polymers. Gas Mixing. The test vapors were generated from specially developed thermocontrolled (T ) 223-293 K) vaporizers (details in ref 19) using dry synthetic air as carrier gas and then diluted to known concentrations by the computer-driven mass flow (18) Schurig, V.; Juza, M. J. Chromatogr., A 1997, 757, 119-135. (19) Bodenho ¨fer, K.; Hierlemann, A.; Schlunk, R.; Go ¨pel, W. Sens. Actuators, submitted.

controllers. The internal volume of these vaporizers was drastically reduced, and hence, very small quantities of the expensive chiral analytes (minimum, 50 µL) could be released at constant concentration. The noise in the sensor signals caused by concentration fluctuations or aerosol formation of the liquid analytes was reduced, and the reproducibility of the adjusted gas phase concentrations was significantly enhanced by using these vaporizers. The vapor phase concentrations at the respective temperatures were calculated according to the Antoine equation.20 All vapors were mixed and temperature stabilized before entering the thermoregulated chamber. The gas tubing was made of stainless steel. The sensors (all types of coatings were measured simultaneously) were mounted inside a flow-through brass cell, and the measurements were performed at a constant temperature of 303 K. The thermostat used for the measuring chamber was a microprocessor-controlled Julabo FP 30 MH (Julabo, Seelbach, Germany, precision of 0.01 K guaranteed). To achieve the necessary very low analyte concentrations especially in the case of the inhalation anesthetics, the temperature of the vaporizers had to be fixed at ∼223 K. The thermostat used for this purpose was an Ultra Kryomat RUK 60 from Lauda Dr. Worbser GmbH, Lauda-Ko¨nigshofen, Germany. The gas flow rate to the sensors was 200 mL/min at a total pressure of 105 Pa. The response time of the sensors is in the order of seconds (98%, bp 417-418 K, Aldrich-Chemie, Steinheim, Germany) and both enantiomers of methyl 2-chloropropionate [purum; > 97% (GC), bp 405-407 K, Fluka Chemie AG, Buchs, Switzerland]. In addition we investigated the nonchiral anesthetic sevoflurane and the chiral anesthetics isoflurane, enflurane, and desflurane purchased from Abbott GmbH, Wiesbaden, Germany. Caution: Inhalation of these anesthetics causes immediate inconsciousness and dizziness. Therefore, all the experiments have to be performed in a hood or in closed gas-tight setups. The enantiomers were separated by preparative GC on cyclodextrin-containing stationary phases. A racemic mixture of the gaseous anesthetics was fed to a preparative column containing 80% (w/w) ceramic support [diatomacious earth, Chromosorb, P(AW-DMCS) 80-100 mesh, Macherey-Nagel, Du¨ren, Germany] and 20% (w/w) of a mixture of 90% (w/w) SE-54 and 10% (w/w) modified γ-cyclodextrin (Figure 1). The pure enantiomers were condensed at the bottom of the column and collected in cooling traps by using liquid nitrogen (details of the preparative enantioseparation procedure are given in ref 21). The purity was checked by GC analysis: (R)-(-)-isoflurane 88.5%, (S)-(+)isoflurane >97%, (R)-(-)-enflurane >99%, (S)-(+)-enflurane >99%,(R)-(-)-desflurane 69%, and (S)-(+)-desflurane 94%. (20) Riddick, J.; Bunger, A. Organic Solvents. In Techniques of Chemistry; Weissberger, A., Ed.; Wiley Interscience: New York, 1970; Vol. II. (21) Grosenick, H.; Juza, M.; Schurig, V.; Klein, J. GIT Fachz. Lab. 1995, 39, 1039-1041.

Analytical Chemistry, Vol. 69, No. 19, October 1, 1997

4019

Figure 3. Sensor response of an array of four sensors coated with 50% (w/w) CD in SE-54 and two nonchiral reference sensors (PEUT) upon exposure to different concentrations of (R)- and (S)-isoflurane at 303 K.

Since the data for applying the Antoine equation were not available for some of the analytes, we determined the saturation vapor pressures at the respective temperatures by the procedure described in detail in ref 19. GC Measurements. The chiral separation factors, RGC, were determined at 303 K using a fused-silica capillary column (length, 25 m; inner diameter, 0.25 mm) coated with 50% (w/w) modified γ-cyclodextrin dissolved in SE-54. The average film thickness was 0.25 µm. Helium was used as carrier gas at an inlet pressure of 1.1 × 105 Pa. RESULTS AND DISCUSSION In contrast to the original order as described in the introduction, we want to first give the results of the anesthetics, continue with the derivatives of propionic acid, and finally present the fits and theoretical considerations. Chiral Discrimination of Inhalation Anesthetics. Since the chiral discrimination of inhalation anesthetics and the investigation of their narcotic potency is of great current interest,3,8 we tried to develop a sensor system, which in contrast to GC measurements offers the possibility of on-line operation. We investigated the haloethers enflurane and isoflurane (Figure 2), which are the most frequently administered inhalation anesthetics, and additionally, the highly volatile narcotic gas desflurane (Figure 2), which is in its clinical test phase. Based on the previous results from the chiral amides,4 we chose an analyte concentration range where the ratio of recognition units to analyte molecules was larger than 10:1, in order to observe unambiguous chiral recognition. Other aspects concerned the very small available quantities of the analytes in enantiomeric purity (between 150 and 50 µL of liquid) and the enormous volatility of the anesthetics. Consequently, we cooled the vaporizers to 223 K and injected the liquids through septa. The analyte liquids and the syringes were kept in dry ice to limit the vaporization losses during the transfer into the vaporizers. Water condensation was minimized by filling the syringes in a closed box under a CO2 atmosphere placed in the immediate vicinity of the vaporizers. The first tested anesthetic was isoflurane. The S enantiomer is eluted in front of the R enantiomer in GC. Figure 3 shows the sensor response of an array of four sensors coated with 50% (w/ w) CD in SE-54 and two nonchiral reference sensors (PEUT) upon 4020 Analytical Chemistry, Vol. 69, No. 19, October 1, 1997

exposure to different concentrations of (R)- and (S)-isoflurane. The layer thicknesses of the four chiral sensors and the two reference sensors are nearly identical. Switching from the (R)-enantiomer to the (S)-enantiomer decreases the sensor response in the case of the chiral sensors, whereas the responses of the nonchiral sensors remain unchanged. Since the nonchiral reference polymer PEUT is very sensitive to humidity, the water content by condensation into the precooled analytes (R and S) and the syringe is negligible or identical. The chiral discrimination is under no circumstances affected. The interaction between the CD recognition unit and the (R)-enantiomer is considerably stronger than that between the CD cage and the (S)-enantiomer. Both CDcontaining coatings absorb the isoflurane to a much larger extent than the reference polymer PEUT. The scattering of the four sensor responses to the respective enantiomer is slight; the chiral recognition is distinct. The fast sensor response (high volatility of the anesthetics), the stability of the base line, and the excellent reversibility and reproducibility of the signals are displayed for an arbitrarily chosen chiral sensor (50% (w/w) CD in SE-54) in the upper part of Figure 4. The solid line marks the sensor response upon exposure to (S)-isoflurane, the dashed line the response to (R)-isoflurane. In this case (analyte quantity,