Combination of Chiral Capillary Electrochromatography with

Development and Assay of Warfarin Enantiomers in. Human Plasma. Jack Zheng and Shahab A. Shamsi*. Department of Chemistry, Center of Biotechnology ...
0 downloads 0 Views 232KB Size
Download PDF
Anal. Chem. 2003, 75, 6295-6305

Combination of Chiral Capillary Electrochromatography with Electrospray Ionization Mass Spectrometry: Method Development and Assay of Warfarin Enantiomers in Human Plasma Jack Zheng and Shahab A. Shamsi*

Department of Chemistry, Center of Biotechnology and Drug Design, Georgia State University, Atlanta, Georgia 30303

The hyphenation of chiral capillary electrochromatography (CEC) with electrospray ionization mass spectrometry (ESI-MS) is very challenging but promising due to the fact that it combines sensitivity with high specificity and selectivity. In this work, CEC capillaries packed with (3R,4S)-Whelk-O1 chiral stationary phase were used for simultaneous enantioseparation of (()-warfarin and its internal standard, (()-coumachlor. Furthermore, both the chiral CEC separation and MS detection parameters were examined in detail. First, the influence of different column fabrication was investigated. Second, enantioseparation was optimized by varying CEC parameters, including acetonitrile concentration, buffer pH, and ionic strength. Under the optimum chiral CEC conditions, ESI-MS parameters such as sheath liquid pH and composition, sheath liquid flow rate, drying gas flow rate, drying gas temperature, nebulizer pressure, and fragmentor voltage were investigated to achieve maximum MS signals of the separated enantiomers. Finally, using solid-phase extraction as sample preparation method, (()-warfarin spiked in 100-µL human plasma samples were analyzed. The calibration curves showed good linearity for both (R)warfarin (R ) 0.9979) and (S)-warfarin (R ) 0.9978) enantiomers. The experimental limit of detection was ∼25 ng/mL for both enantiomers. Even though the data are still preliminary, we can state with confidence that chiral CEC-ESI-MS has the potential to establish itself as a very powerful technique for the determination of enantiomeric ratios in human body fluid. Modern electrodriven separation methods such as micellar electrokinetic chromatography (MEKC) and capillary electrochromatography (CEC) are well-suited for analysis of chiral compounds. For the separation of enantiomers in MEKC, a chiral micelle polymer (also known as polymeric surfactant) is added to the mobile phase.1-6 On the other hand, in CEC mode, racemic * Corresponding author. Phone: 404-651-1297. Fax: 404-651-2751. E-mail: [email protected]. (1) Wang, W.; Warner, I. M. Anal. Chem. 1994, 66, 3773-3776. (2) Shamsi, S. A.; Macossay, J.; Warner, I. M. Anal. Chem. 1997, 69, 29802987. 10.1021/ac030193j CCC: $25.00 Published on Web 10/18/2003

© 2003 American Chemical Society

compounds may be resolved by three different methods: (i) a capillary packed with the chiral stationary phase (CSP),7,8 (ii) an internal capillary wall coated with CSP,9 and (iii) in situ polymerization of rods or monoliths within the capillary.10 Detection in chiral MEKC or chiral CEC is commonly by UV detector, which is inherently less sensitive due to the short light path length of the capillary column. Although laser-induced fluorescence (LIF) detection may offer increase sensitivity, few chiral compounds fluoresce. Of all chiral capillary electrophoresis (CE) methods reported to-date, mass spectrometry (MS) arguably offers the greatest potential to serve as a universal detector. In addition, due to its ability to provide both molecular mass and structural information, MS has several advantages over both UV-visible and LIF detection methods when coupled to CEC or MEKC for enantiomeric analysis. For example, in clinical studies, it is imperative not only to separate all enantiomers and their enantiomeric metabolites with high efficiency and selectivity but also to identify and quantitate all eluted enantiomers and its associated enantiomeric metabolites with increase sensitivity and accuracy. The on-line combination of chiral MEKC with electrospray ionization mass spectrometry (ESI-MS) has recently been shown to be a useful step toward improving sensitivity in a MS-friendly mode. The first successful application of chiral MEKC using polysodium N-undecanoyl-L-valinate as MS-compatible surfactant was recently introduced in our laboratory.11 It was shown that the use of molecular micelles for ESI-MS is advantageous because the high molecular weight of the micelle polymers essentially provides no fragmentation and hence no spectral clutters are (3) Billiot, E.; Thibodeaux, S.; Shamsi, S. A.; Warner, I. M. Anal. Chem. 1999, 71, 4044-4049. (4) Shamsi, S. A.; Palmer, C. P.; Warner, I. M. Anal. Chem. 2001, 73, 141A149A. (5) Palmer, C. P. Electrophoresis 2002, 23, 3993-4004. (6) Yarabe, H. H.; Billiot, E.; Warner, I. M. J. Chromatogr., A 2000, 875, 179206. (7) Wolf, C.; Spence, P.; Pirkle, W.; Derrico, E.; Cavender, D.; Rozing, G. J. Chromatogr., A 1997, 782, 173-178. (8) Wolf, C.; Spence, P.; Pirkle, W.; Cavender, D.; Derrico, E. Electrophoresis 2000, 21, 917-924. (9) Wang, Y.; Zeng, Z.; Guan, N.; Cheng, J. Electrophoresis 2001, 22, 21672172. (10) La¨mmerhofer, M.; Peters, E. C.; Yu, C.; Svec, F.; Fre´chet, J. M. Anal. Chem. 2000, 72, 4614-4622. (11) Shamsi, S. A. Anal. Chem. 2001, 73, 5103-5108.

Analytical Chemistry, Vol. 75, No. 22, November 15, 2003 6295

Figure 1. Chemical structure, pKa, and log P of warfarin and coumachlor. Scifinder Scholar Version: 2001 (©2001 American Chemical Society, Columbus, OH).

observed in the mass domain of the chiral analytes. In addition, micelle polymers have some unique advantages for interfacing to a mass spectrometer. These benefits include low surface activity, low volatility, and zero critical micelle concentration. Thus, polar (charged or uncharged) chiral analytes can be conveniently separated and detected by MEKC-MS with high efficiency and selectivity with minimal loss in detection sensitivity. The technique of chiral CEC in which the chiral selector is immobilized on the column bed simply avoids the introduction of the chiral selector into the ESI-MS interface. Therefore, the use of packed enantioselective columns in CEC provides obvious advantages over other chiral CZE and MEKC techniques. This is because the latter two techniques require the use of a nonvolatile chiral selector in the mobile phase resulting in suppression of analyte signal in ESI-MS and lower sample loading capacity.12 Despite these important advantages of chiral CEC-MS, it is surprising that there are only two reports to-date on this hyphenation. The first demonstration of coupling a chiral CEC technique with MS was in 2001 by Schurig and Mayer,13 who used open tubular capillary electrochromatography for the separation and MS detection of hexobarbital enantiomers. Very recently, von Brocke and co-workers coupled packed column CEC with ESIMS and coordination ion spray mass spectrometry.14 The authors described an approach based on pressure-supported CEC in capillaries packed with permethylated β-CD-modified silica (ChiralDex-silica) for the enantiomeric analysis of barbiturates and chlorinated phenoxypropanoates. In this paper, we report on the hyphenation of a chiral CEC column tapered at the outlet end and coupled to ESI-MS for simultaneous analysis of (()-warfarin and (()-coumachlor. The chiral CEC-MS of these enantiomers was performed in capillaries packed with 5.0-µm (3R,4S)-Whelk-O1 CSP.7,8 To our knowledge, this is the first report in which a commercially available stationary phase is packed and tested with respect to mobile phase, sheath (12) La¨mmerhofer, M.; Svec, F.; Fre´chet, J. M. J.; Lindner, W. Trends Anal. Chem. 2000, 19, 676-698. (13) Schurig, V.; Mayer, S. J. Biochem. Biophys. Methods 2001, 48, 117-141. (14) Von Brocke, A.; Wistuba, D.; Gfro ¨rer, P.; Stahl, M.; Schurig, V.; Bayer, E. Electrophoresis 2002, 23, 2963-2972.

6296 Analytical Chemistry, Vol. 75, No. 22, November 15, 2003

liquid, and MS parameters to obtain high-efficiency chiral CEC with sensitive MS detection of chiral compounds. The reason that we chose (()-warfarin is because this racemic mixture is a widely used oral anticoagulant.15 This chiral drug has a narrow therapeutic index and complex pharmacology. In addition, the (R)- and (S)-warfarin exhibit marked differences in pharmacokinetics and pharmacodynamics.16 For example, it is important to monitor the concentration ratio of (()-warfarin in patients with thromboembolic disease. However, the development of a (()-warfarin assay is very challenging since the drug strongly binds to plasma protein and only a very low concentration of (()-warfarin exists in human plasma.16 Thus, a sample preparation process including deproteinization and solid-phase extraction (SPE) has allowed us to quantitate the warfarin enantiomers at adequate sensitivity in human plasma samples. EXPERIMENTAL SECTION Reagents and Chemicals. The 5-µm (3R,4S) Whelk-O1 CSP (100-Å pore size) was a gift from Regis Technologies, Inc. (Morton Grove, IL). Racemic mixtures of warfarin and coumachlor (Figure 1) were purchased from Aldrich (Milwaukee, WI). (R)-warfarin (>99%) and (S)-warfarin (>99%) were obtained from Cedra Co. (Austin, TX). Acetonitrile (ACN) and methanol, both HPLC grade, were purchased from Burdick & Jackson (Muskegon, MI). Ammonium hydroxide, acetic acid (HOAc), hydrochloric acid, and sodium chloride were supplied by Fisher Scientific (Springfield, NJ). Perchloric acid was obtained from J. T. Baker (Phillipsburg, NJ) as a 70% (v/v) solution. Ammonium acetate (NH4OAc) was purchased from Sigma (St. Louis, MO) as a 7.5 M solution. Water used in all of the experiments was purified by a Barnstead Nanopure II water system (Barnstead International, Dubuque, IA). CEC Column Fabrication. A Knauer pneumatic pump (Wissenschaftliche Geraˇtebau, Dr. Ing. Herbert Knauer GmbH, Berlin, Germany), capable of operating up to 700 bar, was used to pack capillary columns.17 Production of retaining frits was carried out using a homemade frit burner. (15) Gage, B. F.; Fihn, S. D.; White, R. H. Am J. Med. 2000, 109, 481-488. (16) Holbrook, A. M.; Wells, P. S.; Crowther N. R. In Oral Anticoagulants; Poller, L., Hirsh, J., Eds.; Arnold: London, 1996; pp 30-42.

Figure 2. Configuration of the untapered (a) and the tapered (b) CEC-MS columns.

In this study, two different fabrications for the CEC-MS column (untapered and tapered column18,19) were investigated. The untapered column is generally fabricated using a common squaretip capillary. In the untapered column, two frits (one at the inlet side and the other at the outlet side) are fabricated for retaining the packing material; a short unpacked section (∼1 cm) is served as the electrospray tip. On the other hand, the tapered column is prepared with a capillary with either reduced outer (external tapered) or reduced inner (internal tapered) diameter. Thus the packing material can be retained by means of the taper, and no outlet frit is constructed. In addition, such capillary is fully packed and the pack tapered end is directly exposed to the ESI source. For untapered column fabrication (Figure 2a), a fused-silica capillary (o.d. 363 µm, i.d. 75 µm), obtained from Polymicro Technologies Inc. (Phoenix, AZ) was used to slurry pack with (3R,4S)-Whelk-O1 CSP according to the method reported previously.17 Briefly, the slurry was prepared by dispersing ∼15 mg of CSP in 0.3 mL of ACN followed by sonication for 10 min and subsequent transfer into a stainless steel packing reservoir. After fitting the capillary outlet with a 0.5-µm metal screening as a temporary frit, the capillary inlet was connected to the outlet of the reservoir. Using the pneumatic pump, the packing solvent, ACN, was delivered through the reservoir and capillary at 300 bar for a period of 12 h. During this entire period, the capillary was immersed in an ultrasonication bath to achieve homogeneous packing. Prior to the frits fabrication, the capillary was flushed with 10 mM NaCl for 3 h at 300 bar. Under the same pressure, the inlet and outlet frits were prepared by applying heat to a small segment of the packed capillary with the frit burner for 14 s. After pumping ACN through the column, the residual particles after the outlet frit were flushed out. The total packed bed length was ∼59 cm, and a 1-cm-long unpacked capillary tube was kept after the outlet frit. The tip of unpacked section is exposed to the atmospheric pressure ionization source of the electrospray. (17) Zheng, J.; Shamsi, S. A. J. Chromatogr., A 2003, 1005, 177-187. (18) Lord, G. A.; Gordon, D. B.; Myers, P.; King, B. W. J. Chromatogr., A 1997, 768, 9-16. (19) Choudhary, G.; Horva´th C.; Banks, J. F. J. Chromatogr., A 1998, 828, 469480.

For tapered column18,19 fabrication, the external tip was fabricated by first heating a 3-4-cm-long segment at the middle of a 1.5-m-long, 75-µm-i.d. capillary with a burner for 20 s, followed by pulling the capillary quickly and cooling to room temperature. Then, this external capillary tip was slurry packed with (3R,4S)Whelk-O1 CSP using the same procedure described for untapered column fabrication, except that no outlet frit was necessary due to the tip partly serving as an outlet frit. The total pack length was 60 cm, and the column effluent from the external tip of ∼10 µm (i.d.) at the exit end was electrosprayed directly into the atmospheric region of the ionization source. However, extreme cautions were taken to avoid damage on the capillary tip during installation. CEC-ESI-MS Instrumentation. All chiral CEC-ESI-MS experiments were carried out with an Agilent capillary electrophoresis system (Agilent Technologies, Waldbronn, Germany) interfaced to a quadrupole mass spectrometer, Agilent 1100 series MSD, a G1603A CE-MS adapter kit, and a G1607 CE-ESI-MS sprayer kit, all from Agilent Technologies (Palo Alto, CA). An Agilent 1100 series HPLC pump equipped with 1:100 splitter was used to deliver the sheath liquid. The Agilent ChemStation and CE-MS add-on software were used for instrument control and data analysis. CEC-ESI-MS Conditions. After installation of the CEC column on the cartridge and mounting in the nebulizer, the column was preconditioned for 1 h with the desired mobile phase by connecting the inlet end of column to a manual syringe pump (Unimicro Technologies Inc., Pleasanton, CA). Further conditioning was done with 12 bar pressure at the inlet side of the column slowly increasing the voltage by 10 kV each 20 min until reaching the maximum operating voltage and stable baseline. Injection was made electrokinetically at 6 kV for 8 s in all cases, unless mentioned. The separation voltage was set at 30 kV, employing a voltage ramp of 3 kV/s. During the separation, an external pressure of 12 bar was applied to the inlet buffer vial. Unless otherwise stated, the following ESI-MS conditions were used: sheath liquid, CH3OH/H2O (50:50, v/v) containing 5 mM NH4OAc, pH 6.8; sheath liquid flow rate, 5.0 µL/min; capillary voltage, -2500 V; fragmentor voltage, 80 V; drying gas flow rate, 5 L/min; drying gas temperature, 150 °C; nebulizer pressure, 0.276 bar (4 psi). The MS detection was performed in the selective ion monitoring (SIM) mode. Since both (()-warfarin and (()coumachlor exist as anions in basic solutions, negative [M - H]ions were monitored at 307.0 and 341.0 (m/z), respectively. MS spray chamber parameters were optimized by a procedure similar to LC-MS manual tune, which involves continuously flushing 1 mg/mL (()-warfarin through a 60-cm-long, 75-µm-i.d. unpacked capillary at 50 mbar to the ESI interface and monitoring the SIM abundance at 307.0 (m/z). Mobile-Phase and Sheath Liquid Preparation. All mobile phases or sheath liquids were obtained by first adjusting the pH of NH4OAc buffer to the desired value with HOAc or NH3‚H2O. The pH of the aqueous buffer was checked and adjusted with an Orion 420A pH meter (Beverly, MA) before addition of an appropriate volume ratio of ACN or CH3OH. The final mobile phase or sheath liquid was degassed for 30 min and filtered with a 0.45-µm PTFE membrane before use. Analytical Chemistry, Vol. 75, No. 22, November 15, 2003

6297

Standard Analytes and Plasma Sample Preparation. Both (()-warfarin and (()-coumachlor were dissolved in 60% (v/v) ACN at a concentration ∼1 mg/mL. This 60% (v/v) ACN was also used to dilute the stock sample and reconstituted the plasma residue. The elution order of enantiomers was determined by spiking racemic warfarin sample solution with (S)-warfarin enantiomer. For (()-warfarin, a 500 µg/mL stock solution was prepared and diluted daily to prepare subsequent solution at concentrations of 10 and 100 µg/mL. For (()-coumachlor, a 300 µg/mL stock aqueous stock solution was prepared and diluted daily to prepare a 30 µg/mL internal standard solution. Blank human plasma samples were obtained from Mercer University Southern School of Pharmacy (Atlanta, GA) and stored under -78 °C until analysis. For each plasma sample, a 100.0-µL aliquot of plasma was spiked with the desired volume of (()warfarin solution at levels of 0.2, 0.5, 1.0, 2.5, 5.0, and 10.0 µg/ mL and an internal standard of (()-coumachlor was added at a concentration of 2.0 µg/mL in each of the six microcentrifuge tubes. The content of each tube was acidified with 0.2 mL of 10% HClO4 and vortexed for 30 s. All six tubes were centrifuged at 10 000 rpm for 10 min to precipitate the protein. The supernatant of each tube was then transferred to a C18 solid-phase extraction column (J. T. Baker, Phillipsburg, NJ) which had been previously primed with 2 mL of CH3OH and 2 mL of 1 M HCl. The SPE column was washed again with 2 mL of 1 M HCl prior to elution of (()-warfarin and (()-coumachlor with 2 mL of CH3OH. The eluate from the SPE column was evaporated to dryness in a water bath at 60 °C under a stream of N2. The inner wall of each tube was rinsed twice with 100 µL of CH3OH followed by evaporation to obtain a dry residue. Finally, this residue was reconstituted with 60 µL of 60% (v/v) ACN and injected hydrodynamically into the CEC column using 12 bar for 3 min. Calculations. Chiral resolution (Rs), selectivity (R), and separation efficiency of enantiomers were calculated with Agilent Chemstation software (V 9.0) as described elsewhere.17 All chromatograms shown were smoothed with a factor of 0.1 min. The noise level was determined using 6 times the standard deviation of the linear regression of the baseline drift for a selected time range between 10 and 20 min. The signal-to-noise ratio (S/ N) was obtained as the ratio of peak height over noise level. (S)Warfarin and (R)-warfarin calibration curves were obtained by plotting the peak area ratio of the respective enantiomer to the internal standard ((S)-coumachlor) versus concentration. To assess linearity, the line of best fit was determined by least-squares regression. RESULTS AND DISCUSSION Column Fabrication for CEC-MS and Retention Time Reproducibility Study. For most of the current CEC column configurations,20-22 the packed column consists of two segments. A packed segment is usually followed by an open tubular segment to achieve on-column UV detection. The advantages of this configuration are the ease of fabrication and good reproducibility. (20) Colo´n, L. A.; Maloney, T. D.; Fermier, A. M. J. Chromatogr., A 2000, 887, 43-53. (21) Boughtflower, R. J.; Paterson, C. J.; Knox, J. H. J. Chromatogr., A 2000, 887, 409-420. (22) Byrne, C. D.; Smith, N. W.; Dearie, H. S.; Moffatt, F.; Wren, S. A. C.; Evans, K. P. J. Chromatogr., A 2001, 927, 169-177.

6298 Analytical Chemistry, Vol. 75, No. 22, November 15, 2003

Table 1. Comparison of (()-Warfarin Retention Time Reproducibility for Different Column Fabricationa untapered column CEC run

tapered column

tR1 (min)

tR1′ (min)

tR1 (min)

tR1′ (min)

first second third fourth fifth

27.4 56.0 57.8 67.7 71.1

28.0 58.6 61.7 71.4 75.6

28.1 30.8 32.2 30.9 32.0

29.8 33.2 34.3 33.2 34.2

RSD (%)

31

32

5

6

a Condition: 60-cm-long untapered or tapered capillary column packed with 5-µm (3R,4S)-Whelk-O1 CSP; mobile phase, 80% (v/v) ACN containing 10 mM NH4OAc, pH 3.0; separation voltage, 20 kV; analyte, (()-warfarin. For other parameters, see Experimental Section.

However, when this untapered CEC column (Figure 2a) was employed for CEC-MS study, it seemed to have a critical drawback of irreproducible retention time. Table 1 compares the reproducibility in retention times of two columns. Using an untapered column, the retention time of warfarin enantiomers showed an increase from ∼28 (first run) to ∼76 min (fifth run). The value of relative standard deviation (RSD) for retention time was higher than 30% for the five consecutive injections. In addition, microscopic inspection verified the existence of bubbles in the open segment of the column. Since the pressurization in CEC-MS is only carried out from the inlet side of the buffer reservoir, one possible reason for the irreproducible retention behavior is associated with the retaining frit located between the packed and open segments of the capillary. Therefore, the discontinuity of the ζ potential can lead the development of flow equalizing intersegmental pressure at such frits, which results in bubble formation. This bubble formation decreases the CEC current and EOF. Consequently, retention time increased from run to run. To overcome this problem of irreproducible retention time, tapered column (Figure 2b) was packed with 5-µm (3R,4S)-WhelkO1 CSP (see Experimental Section) and tested for five consecutive separations of warfarin enantiomers. As shown in Table 1, compared to the untapered column, the tapered column demonstrated reproducible retention time (RSD ) 5-6%). In fact, if the first run is considered as a precolumn equilibration, the retention time was very consistent (RSD ∼2%). This indicates that the tapered column end acts as a back-pressure resistor, which suppressed the bubble formation.18,19 In addition, the decreased tip dimension assisted to stabilize the electrospray.23 Optimization of CEC and ESI-MS Parameters. In this study, we have investigated more details of the experimental parameters that influence both chiral CEC and ESI-MS of (()warfarin and (()-coumachlor. The factors include CEC parameters, sheath liquid parameters, and MS spray chamber parameters. The influence of all of the aforementioned parameters is discussed below. A. Optimization of CEC Parameters. Considerations regarding the appropriate choice of mobile phase for CEC-MS are important to achieve the highest resolution and ESI-MS sensitivity. In general, the eluents for CEC-MS should comprise a binary (23) Covey, T. R.; Pinto, D. In Applied Electrospray Mass Spectrometry; Pramanik, B. N., Ganguly, A. K., Gross, M. L., Eds.; Marcel Dekker: New York, 2002; pp 105-148.

Figure 3. Electropherograms (a-c) and plots (d-f) showing effects of percent ACN (v/v) for simultaneous enantioseparation of warfarin (peak 1, (R)-warfarin; 1′, (S)-warfarin) and coumachlor (peak 2, (R)-coumachlor; 2′, (S)-coumachlor), (d) Rs, (e) R, and (f) N. Conditions: 60-cm-long, 75-µm-i.d. tapered capillary packed with 5-µm (3R,4S)-Whelk-O 1 CSP; mobile phase, (a) 80, (b) 70, and (c) 60% (v/v) ACN, 10 mM NH4OAc, pH 3.0; sheath liquid, 5 mM NH4OAc in CH3OH/H2O (50:50, v/v), 5 µL/min. For the other CEC-ESI-MS conditions, see Experimental Section.

mixture of aqueous-organic in which a volatile buffer is added to obtain a desired pH. The first set of experiments to be discussed is related to the role of the mobile-phase parameters that included percent ACN, ionic strength, and pH of the mobile phase. (1) Influence of Mobile-Phase ACN Content. In this set of experiments, the ionic strength (10 mM NH4OAc) and pH (pH 3.0) of the mobile phase were fixed and the volume fraction of ACN was varied from 80 to 60% (v/v). As shown in the electropherograms (Figure 3a-c) and plots (Figure 3d-f), 80% (v/v) ACN provided the shortest analysis time, but lowest selectivity (R) and Rs for warfarin and coumachlor enantiomers. When the ACN concentration decreased from 80 (v/v) to 60% (v/v), not only the chiral Rs of (()-warfarin (1,1′) and (()-coumachlor (2,2′) increased but also the Rs values between (S)-warfarin (peak 1′) and (R)-coumachlor (peak 2) increased even more significantly (Figure 3d). This suggests that both chiral and achiral R of (3R,4S)-Whelk-O1 CSP could be enhanced by increasing the polarity of the mobile phase (Figure 3e). In addition, the decreased ratio of permittivity to viscosity (r/η) of the eluent and the variation on the packing surface charge may have contributed to lower electroosmotic flow (EOF) 24,25 that in turn increased Rs. However, one should note that the analysis time for simultaneous enantioseparation of (()-warfarin and (()-coumachlor increased (24) Crego, A. L.; Martı´nez, J.; Marina, M. L. J. Chromatogr., A 2000, 869, 329337. (25) Colo´n, L. A.; Burgos, G.; Maloney, T. D.; Cintro´n, J. M.; Rodrı´guez, R. L. Electrophoresis 2000, 21, 3965-3993.

from less than 30 min at 80% (v/v) ACN to about 70 min for 60% (v/v) ACN. As a result, 70% (v/v) ACN was selected as the optimum ACN concentration because it provided a good compromise between Rs and analysis time. (2) Influence of Mobile-Phase Ionic Strength. The ionic strength effect on simultaneous enantioseparation of (()-warfarin and (()-coumachlor was studied by using optimum 70% (v/v) ACN containing 2.5, 5.0, and 10.0 mM NH4OAc (pH 3.0). Previous study in our laboratory17 has demonstrated that ionic strength influences the separation efficiency (N) and hence chiral Rs in a fashion similar to that of applied voltage. As shown in Figure 4a, the use of 2.5 mM NH4OAc (pH 3.0) provided the shortest analysis time but lowest chiral Rs for (()-warfarin and (()-coumachlor. As the NH4OAc concentration increased to 5.0 mM and finally to 10.0 mM, the simultaneous enantioseparation of (()-warfarin and (()-coumachlor enhanced (Figure 4b,c) primarily due to increase in N (Figure 4f). Hence, both higher chiral and achiral Rs values of (()-warfarin and (()-coumachlor were achieved (Figure 4d) although R between each pair of enantiomers was more or less the same (Figure 4e). However, it should be noted that higher ionic strength had an undesired effect by decreasing the EOF primarily because of the decrease of double-layer thickness and increase of mobile-phase viscosity.26 Therefore, retention time of (()-warfarin and (()-coumachlor became longer (Figure 4a-c). (26) Rathore, A. S.; Reynolds, K. J.; Colo´n, L. A. Electrophoresis 2002, 23, 29182928.

Analytical Chemistry, Vol. 75, No. 22, November 15, 2003

6299

Figure 4. Electropherograms (a-c) and plots (d-f) showing effects of mobile-phase ionic strength upon (d) Rs, (e) R, and (f) N for simultaneous enantioseparation of (()-warfarin and (()-coumachlor. Conditions: mobile phase, 70% (v/v) ACN containing (a) 2.5, (b) 5.0, and (c) 10.0 mM NH4OAc, pH 3.0. Peak identifications and other conditions are the same as Figure 3 or described in the Experimental Section.

For example, the retention time of the later eluting (S)-coumachlor increased by nearly 50% at 10.0 mM NH4OAc compared to 2.5 mM NH4OAc. A good compromise seems to be obtained at 5.0 mM NH4OAc because retention times were significantly reduced while simultaneous enantioseparation of (()-warfarin and (()coumachlor was maintained. (3) Influence of Mobile-Phase Buffer pH. Previous study in our laboratory17 has shown that mobile-phase pH was critical for the separations of acidic enantiomer and a monobasic sodium phosphate adjusted with phosphoric acid at pH 3.0 was the optimum buffer. However, due to the volatility consideration, phosphate buffer is not suitable for the CEC-ESI-MS study. Thus, in this study, the pH of volatile aqueous mobile-phase 70% (v/v) ACN containing 5 mM NH4OAc was adjusted using HOAc to pH 3.0, 4.0, and 5.0. The electropherogram (Figure 5a) and plot (Figure 5d) respectively revealed that the lowest chiral Rs for (()-warfarin and (()-coumachlor is obtained at pH 5.0. In addition, as the buffer pH decreased from 5.0 to 3.0, the N increased significantly (Figure 5c); therefore, higher Rs for (()-warfarin and (()-coumachlor was achieved although the R between each pair of enantiomers was more or less the same (Figure 5b). On the other hand, the Rs between (S)-warfarin and (R)-coumachlor (Rs1′,2) kept fairly constant (Figure 5d). This is because the efficiency (N1′, N2) tends to increase and selectivity (R1′,2) to decrease with decreasing pH. Hence, the net result is no significant change in Rs value between the two peaks (1′, 2). 6300

Analytical Chemistry, Vol. 75, No. 22, November 15, 2003

As expected, decreasing the buffer pH also affects the retention time of (()-warfarin and (()-coumachlor by decreasing the EOF velocity and electrophoretic mobility of these pH-dependent acidic analytes.17,27,28 When buffer pH decreased from 5.0 to 4.0, the retention times of both (()-warfarin and (()-coumachlor decreased (Figure 5a,b). This can be explained by the suppressed ionization of warfarin and coumachlor (both compounds have pKa ) 4.5). Therefore, electrophoretic mobility of these two partially anionic analytes toward to the anodic end (injection end) is decreased and the analytes are rapidly swept toward the cathodic end (detection end) by the EOF. Although a decrease of mobile-phase pH also suppresses the EOF, this suppression appears to be less significant than ion suppression of (()-warfarin and (()-coumachlor. Consequently, faster analysis was obtained at pH 4.0. Further decrease in mobile-phase pH, from 4.0 to 3.0, reduces the EOF velocity and increases analysis time (Figure 5c). As a result, pH 4.0 was chosen as the optimum because it provided a reasonable tradeoff between Rs and analysis time. B. Optimization of the Sheath Liquid Parameters. Normally, the flow rate through the packed CEC column, approximately 10-100 nL/min, is too low for supporting a stable electrospray, which require a flow rate typically a few microliters per minute.29 Hence, a makeup liquid known as sheath liquid is typically added as postcolumn to increase the total flow rate up (27) Smith, N. W.; Carter-Finch, A. S. J. Chromatogr., A 2000, 892, 219-255. (28) Altria, K. D.; Smith, N. W.; Turnbull C. H. J. Chromatogr., B 1998, 717, 341-353.

Figure 5. Electropherograms (a-c) and plots (d-f) showing effects of mobile-phase pH upon (d) Rs, (e) R, and (f) N for simultaneous enantioseparation of (()-warfarin and (()-coumachlor. Conditions: mobile phase, 70% (v/v) ACN containing 5.0 mM NH4OAc and pH (a) 5.0, (b) 4.0, and (c) 3.0. Peak identifications and other conditions are the same as Figure 3 or described in the Experimental Section.

to a normal rate of 1-10 µL/min. Similar to CZE-ESI-MS, sheath liquid for CEC-ESI-MS serves not only to establish an electrical connection between the outlet end of the CEC column and electrosprayer but also to assist the electrospray ion source and serve as a terminal electrolytic reservoir.19,29 Therefore, it is important to optimize the sheath liquid parameters (composition, pH, flow rate) for high ESI-MS sensitivity. In this section, the composition of CH3OH was first studied, followed by pH and flow rate of the sheath liquid. (1) Influence of Sheath Liquid CH3OH Composition. The effect of sheath liquid CH3OH composition was investigated from 50 to 90% CH3OH (v/v) with 5 mM NH4OAc at pH 6.8. It is clear that sheath liquid composition has little to no effect on simultaneous enantioseparation of (()-warfarin and (()-coumachlor (data not shown). For example, there are no significant differences between the retention time and resolution. This result is in contrast to the report by Choudhary et al.,19 who reported that sheath liquid composition has pronounced effects on the achiral separation of phenylthiohydantoin-amino acids. Evaluation of MS signal intensity was carried on two parameters, i.e., the abundance and S/N ratio. The abundance of (R)warfarin increased to almost 3-fold (Figure 6a) upon increasing the CH3OH composition from 50 (v/v) to 90% (v/v). One possible interpretation of this result supports the hypothesis that for solution containing a large content of CH3OH decreasing the (29) Choudhary, G.; Apffel, A.; Yin, H.; Hancock, W. J. Chromatogr., A 2000, 887, 85-101.

droplet size results in higher efficiency for desolvation.30 Unlike abundance, S/N first increases from 50 (v/v) to 70% (v/v) and then decreases at 90% (v/v) CH3OH (Figure 6b). This latter decrease could be due to much higher noise level (data not shown), which is possibly caused by too low a liquid surface tension resulting in unstable spray at such high content of CH3OH in the sheath liquid.30 The observed ghost peak at 90% (v/v) CH3OH sheath liquid (data not shown) could be generated for a similar reason. (2) Influence of Sheath Liquid pH. Recent work in our laboratory using CEC-UV detection17 and our earlier discussion in this paper suggest that chiral Rs of (()-warfarin and (()coumachlor is best seen at pH 3-4. However, to achieve higher MS sensitivity, basic sheath liquid is required to facilitate deprotonation of acidic analytes. Thus, in this experiment, sheath liquids containing 70% (v/v) CH3OH, 5 mM NH4OAc at pH 6.8, 8.5, and 10.0 were investigated. Similar to the study on sheath liquid composition, sheath liquid pH had no significant impact on retention time and chiral Rs (data not shown). Figure 7a,b shows that mass abundance remains more or less the same with increasing pH whereas the trends in S/N are complicated by substantial differences in the amplitude of error bars. This suggests that there must be a source of random noise in which at several instances a fraction of ions may reach the detector without going through the mass analyzer. Therefore, the random noise is possibly the limiting factor in determining the S/N. (30) Cech, N. B.; Enke, C. G. Mass Spectrom Rev. 2001, 20, 362-387.

Analytical Chemistry, Vol. 75, No. 22, November 15, 2003

6301

Figure 6. Effects of sheath liquid CH3OH content upon (a) abundance and (b) S/N for (R)-warfarin. The error bar represents one standard deviation (SD) of three measurements. Conditions: mobile phase, 70% (v/v) ACN containing 5.0 mM NH4OAc (pH 4.0); sheath liquid, 5 mM NH4OAc in 90, 70, and 50% (v/v) CH3OH, 5 µL/min. Other conditions are the same as Figure 3 or described in the Experimental Section.

Nevertheless, due to relatively low background noise at pH 8.5, this pH was chosen in the continuation of the optimization. (3) Influence of Sheath Liquid Flow Rate. The effect of sheath liquid flow rate was studied using 70% (v/v) CH3OH containing 5 mM NH4OAc, pH 8.5, at five different flow rates that ranged from 1.0 to 10.0 µL/min. In general, the sheath liquid flow rate had no significant impact on the separation (data not shown); however, both abundance and S/N increased (Figure 8). Initially, at a flow rate of 1.0 µL/min, the flow is too low to establish the electrical contact necessary for CEC separation.23 By increasing the sheath liquid flow rate from 1.0 to 7.5 µL/min, both abundance and the S/N tend to increase. In general, the sheath liquid flow rate over the range of 5.0-7.5 µL/min seems to be effective in establishing a stable electrospray. Again, analyte abundances were relatively not sensitive to the flow rate increase from 7.5 to 10.0 µL/min, but the S/N began to decrease probably because of an increase in noise (Figure 8). A flow rate of 7.5 µL/min was chosen since it provided both higher abundance and higher S/N. C. Optimization of Electrospray Chamber Parameters. In our third set of experiments, we examined the role of electrospray chamber parameters that included capillary voltage, fragmentor voltage, drying gas flow rate, drying gas temperature, and nebulizer pressure. The effect of each of the aforementioned parameters is discussed below. (1) Capillary Voltage and Fragmentor Voltage. Capillary voltage and fragmentor voltage are critical MS parameters for maximizing signal intensity. Both of these parameters were 6302 Analytical Chemistry, Vol. 75, No. 22, November 15, 2003

Figure 7. Plots showing effects of sheath liquid pH upon (a) abundance and (b) S/N of (R)-warfarin. The error bar represents one SD of three measurements. Conditions: sheath liquid, 70% (v/v) CH3OH containing 5 mM NH4OAc at pH 6.8, 8.5, and 10.0, 5 µL/min. The other conditions are same as Figure 6 or described in the Experimental Section.

optimized in the off-line mode. Because (()-warfarin can exist as a negatively charged ion ([M - H]-), capillary voltage was optimized in negative ion mode. It was concluded that a capillary voltage of -2500 V was suitable since it provides relatively high sensitivity without excessive noise caused by arcing at high capillary voltage in negative ion mode. Fragmentor voltage was optimized by ramping the fragmentor voltage within the range of 60-150 V at m/z 307.0. In a typical procedure, a solution of warfarin was introduced continuously at 50 mbar from a sample vial into the spray chamber. From this manual tune procedure, the fragmentor voltage 91 V was found to produce a maximum response for warfarin. (2) Influence of Drying Gas Flow Rate and Drying Gas Temperature. To study the effect of drying gas flow rate and temperature, simultaneous enantioseparation of (()-warfarin and (()-coumachlor was studied under different drying gas flow rates and temperatures. However, no significant impact was found for these two parameters on the CEC separation (data not shown). By using the manual tune procedure described earlier in this paper, the MS response for warfarin enantiomers was optimized by varying drying gas flow rate and drying gas temperature. The respective curves are shown in Figure 9a and b. The drying gas flow rate was varied from 3 to 13 L/min. Figure 9a shows that initially by increasing the drying gas flow rate from 3 to 5 L/min, the response increased ∼2-fold. This is because at higher flow rate of drying gas an increased number of ions from the bulk solution come closer to the liquid-gas interface and evaporate thus increasing the desolvation velocity.31 However, at a flow rate

Figure 8. Effects of sheath flow rate upon (a) abundance and (b) S/N of (R)-warfarin. The error bar represents one SD of three measurements. Conditions: sheath liquid, 70% (v/v) CH3OH containing 5 mM NH4OAc (pH 8.5), 1.0, 2.5, 5.0, 7.5, and 10.0 µL/min. The other conditions are the same as Figure 6 or described in the Experimental Section.

higher than 5 L/min, the abundance began to decrease, possibly from the combined effects of dispersion and decrease in the residence time of the sample droplet.31 The influence of drying gas temperature was studied at 150, 200, 250, and 350 °C with a fixed drying gas flow rate at 5 L/min. Figure 9b shows that initially increasing the drying gas temperature from 150 to 250 °C, the response decreased by ∼20%. Similar decrease in abundance of solute with increasing temperature was observed during the MEKC-ESI-MS study for 1,1′-binaphthol.11 One possible explanation for this trend is that the temperature of the drying gas is inversely related to residence time of the sample droplet.31 Thus, droplet size reduction at higher temperature increases the mobility and consequently decreases the residence time of the droplet.31 However, at temperatures higher than 250 °C, the abundance began to increase. In fact, a close inspection shows that warfarin abundance reached almost the same level at 350 °C as it was as initially at 150 °C. The increase in abundance at 350 °C could be analyte dependent. As a result, drying gas temperature of 350 °C was considered reasonable. (3) Nebulizer Pressure. Shamsi11 observed that nebulizer pressure could influence the retention and chiral Rs in MEKCESI-MS because it generates a suction force at the outlet of the open tublar capillary.32 To study the effect of nebulizer pressure (31) Sjoeberg, Per J. R.; Boekman, C. Fredrik; Bylund, Dan; Markides, K. E. J. Am. Soc. Mass Spectrom. 2001, 12, 1002-1010. (32) Lazar, I. M.; Lee, M. L. J. Am. Soc. Mass Spectrom. 1999, 10, 261-264.

Figure 9. Effect of (a) drying gas flow rate, (b) drying gas temperature, and (c) nebulizer pressure upon abundance of warfarin. The error bar represents one standard deviation (SD) of 30 measurements. Conditions: 60-cm-long, 75-µm-i.d. unpacked capillary continuously flushed at 50 mbar with 1 mg/mL (()-warfarin. Sheath liquid, 5 mM NH4OAc (pH 8.5) in 70% (v/v) CH3OH, 7.5 µL/min. MS spray chamber parameters: drying gas temperature rate in (a) was varied from 3 to 13 L/min, at a drying gas temperature of 350 °C, and nebulizer pressure of 0.689 bar (10 psi); drying gas temperature in (b) was varied from 150 to 350 °C, at a drying gas flow rate of 5 L/min, and nebulizer pressure of 0.689 bar (10 psi); nebulizer pressure in (c) was varied from 0.069 to 0.896 bar (1 to 13 psi), at a drying gas flow rate of 5 L/min, and drying gas temperature of 350 °C. Capillary voltage, -2500 V; fragmentor voltage, 91 V; SIM 307.0 (m/ z).

on separation, simultaneous enantioseparations of (()-warfarin and (()-coumachlor were studied on-line under different nebulizer pressures. In contrast to the open tubular capillary, both migration and enantioseparations remain unchanged (data not shown) in the packed bed with tapered column, suggesting no suction effects are produced by variations in nebulizer pressure. In Figure 9c, the mass abundances are plotted against nebulizer pressure, which was varied from 0.069 to 0.896 bar (1 to 13 psi) with a fixed drying gas flow rate of 5 L/min and a temperature of 350 °C. When the nebulizer pressure was initially increased from 0.069 (1 psi) to 0.276 bar (4 psi), the response increased by Analytical Chemistry, Vol. 75, No. 22, November 15, 2003

6303

Figure 10. Analysis of human plasma spiked with warfarin enantiomers. (a) Calibration curves of (a) (R)-warfarin and (b) (S)-warfarin. (c) The electropherogram [negative SIM 307.0 and 341.0 (m/z)] of human plasma spiked with (()-warfarin (1, 1′) at LOD and (()-coumachlor as the IS (2, 2′). The inset of (c) shows the extract ion chromatogram (EIC) for (()-warfarin at LOD. (d) The electropherogram [negative SIM 307.0 (m/z)] of a plasma sample spiked with 0.1 µg/mL (R)-warfarin and 10.0 µg/mL (S)-warfarin. The inset of (d) shows the expanded version of enantiomeric peaks. Conditions: mobile phase, ACN/H2O (70:30, v/v) containing 5 mM NH4OAc, pH 4.0; injection, 12 bar for 3 min; sheath liquid, 70% (v/v) CH3OH containing 5 mM NH4OAc (pH 8.5), flow rate 7.5 µL/min. MS spray chamber parameters: drying gas flow rate 5 L/min; drying gas temperature, 350 °C; nebulizer pressure, 0.276 bar (4 psi); capillary voltage, -2500 V; fragmentor voltage, 91 V.

∼50%. However, at nebulizer pressures higher than 0.276 bar (4 psi), the abundance began to decrease. The trend is in agreement with Sjoberg et al.,31 who explained that the flow rate of nebulizer gas had complex effects on the instrumental response and the droplet surface partitioning coefficients for some analytes. A nebulizer pressure of 0.276 bar (4 psi) was chosen since it gave the highest abundance. Application of Optimized CEC-MS Conditions for (()Warfarin Assay in Human Plasma Sample. In view of the capabilities of CEC-MS to efficiently separate and detect enantiomers of (()-warfarin, it seemed desirable to complete this study by developing a quantitative chiral assay for human plasma sample. To obtain standard calibration curves, a 100-µL aliquot of plasma was spiked with a racemic (()-warfarin solution at levels 6304 Analytical Chemistry, Vol. 75, No. 22, November 15, 2003

from 0.2 to 10.0 µg/mL and internal standard of (()-coumachlor was added at a concentration of 2.0 µg/mL. (()-Coumachlor was chosen as the internal standard (IS) since it has a chemical structure and physicochemical properties similar to that of (()warfarin (Figure 1). The spiked plasma samples were then treated with a SPE procedure33 followed by CEC-ESI-MS analysis. The injection is carried by hydrodynamically applying 12 bar for 3 min instead of electrokinetic injection at 6 kV for 8 s, because the poorer injection volume led to low sensitivity for the latter injection method. The calibration curves of blank plasma spiked with (()-warfarin and IS (()-coumachlor are shown in Figure 10a and b, respec(33) Henne, K. R.; Gaedigk, A.; Gupta, G.; Leeder, J. S.; Rettie, A. E. J. Chromatogr., B 1998, 710, 143-148.

tively. The value of the ordinate shows the ratio of the peak area of (R)-warfarin (A1) or (S)-warfarin (A1′) to that for the (R)coumachlor (A2). Good linearity for both (R)-warfarin (R ) 0.9979) and (S)-warfarin (R ) 0.9978) in the concentration range from 0.05 to 5 µg/mL was obtained. Several analytical methods have been employed for the determination of (R)- and (S)-warfarin in human plasma. These methods include HPLC with UV33-35 or tandem MS detection,36 and more recently cyclodextrin-modified capillary zone electrophoresis using sulfated β-cyclodextrin.37 While HPLC with UV or tandem MS detection provided good LODs of 12.5 and 1 ng/mL, respectively, it should be noted that these methods provided lower chromatographic efficiency and required large sample volume (0.3-1 mL). Higher efficiency is particularly desirable for simultaneous enantioseparation of chiral drugs and their metabolites in a single run. As illustrated in Figure 10c, for a 100-µL plasma sample spiked with racemic warfarin, a limit of detection as low as 25 ng/mL (for single enantiomer) was possible. In addition, using a plasma sample spiked with a known concentration of 0.1 µg/mL (R)-warfarin and 10.0 µg/mL (S)warfarin, the enantiomeric ratio test of warfarin enantiomers was successfully performed. The electropherogram in Figure 10d reveals that the enantioseparation was achieved and the minor enantiomer ((R)-warfarin) at 1% level can be easily detected along with 99% major enantiomer ((S)-warfarin). The enantiomeric excess 38 of (S)-warfarin over (R)-warfarin (calculated from the ratio of peak area) was found to be ∼98.8%, a value that compared well after spiking the plasma sample with a known concentration of (R)-warfarin and (S)-warfarin. CONCLUSIONS For the first time, chiral CEC coupling with ESI-MS for simultaneous enantioseparation and detection of structurally similar enantiomers (()-warfarin and (()-coumachlor using a commercially available chiral stationary phase has been achieved and optimized. First, column fabrication was found to be important for separation and to obtain a stable electrospray. The tapered (34) Prangle, A. S.; Noctor, T. A. G.; Lough, W. J. J. Pharm. Biomed. Anal. 1998, 16, 1205-1212. (35) Ring, P. R.; Bostick, J. M. J. Pharm. Biomed. Anal. 2000, 22, 573-581. (36) Naidong, W.; Ring, P. R.; Midtlien, C.; Jiang, X. J. Pharm. Biomed. Anal. 2001, 25, 219-226. (37) Yau, W.; Chan, E. J. Pharm. Biomed. Anal. 2002, 28, 107-123. (38) March, J. Advanced Organic Chemistry, 3rd ed.; John Wiley and Sons: New York, 1985; pp 126-127.

column showed much more reproducible retention time compared to the untapered column. Second, the mobile phase used for CECMS showed trends similar to those observed in a previous CECUV study with the same CSP.14 The ACN content in the mobile phase influenced selectivity and retention of enantiomers. Ionic strength affected both chiral and achiral resolution by affecting the EOF velocity and separation efficiency. Mobile-phase pH influenced the retention of these acidic analytes by affecting the EOF velocity and analyte electrophoretic mobility. Third, the sheath liquid and electrospray chamber parameters were found to have no impact on the simultaneous enantioseparation, but mainly influenced the detection sensitivity. Under the optimum conditions [60-cm-long, 75-µm-i.d. tapered capillary packed with 5-µm (3R,4S)-Whelk-O1 CSP; ACN/H2O (70:30, v/v) containing 5 mM NH4OAc, pH 4.0; injection, 12 bar for 3 min; 70% (v/v) CH3OH sheath liquid containing 5mM NH4OAc (pH 8.5); sheath liquid flow rate, 7.5 µL/min; drying gas flow rate, 5 L/min; drying gas temperature, 350°C; nebulizer pressure, 0.276 bar (4 psi); capillary voltage, -2500 V; fragmentor voltage, 91 V], the chiral CEC-ESI-MS technique was applied to assay (()warfarin in human plasma samples. The results showed good linearity and limit of detection comparable with the other published reports,33-37 along with the added benefits of higher chromatographic efficiency and minimal sample as well as mobilephase consumption, although slightly longer analysis time is required.33-37 In addition, this technique is capable of distinguishing minor enantiomer (as little as 1%) in excess of major enantiomer (99%). Studies are currently under investigation to enhance the robustness of this technique and application of CECESI-MS to monitor the enantiomeric ratio of (()-warfarin in plasma samples of patients with thromboembolic disease. ACKNOWLEDGMENT Financial support for this project was provided by the National Institutes of Health (Grant GM 62314-02). The authors thank Regis Technologies Inc. (Morton Grove, IL) for donation of (3R,4S)Whelk-O1 CSP. J.Z. is grateful to Dat Phan (Agilent Technologies) and Christine Hon (Mercer University) for helpful discussions.

Received for review May 13, 2003. Accepted September 5, 2003. AC030193J

Analytical Chemistry, Vol. 75, No. 22, November 15, 2003

6305