isoindole-Amino Acids by Reverse Polarity Capillary Electrophoresis

Dec 9, 2006 - A capillary electrophoresis method with laser-induced fluorescence detection for the chiral separation of cy- anobenz[f]isoindole (CBI) ...
0 downloads 0 Views 172KB Size
Download PDF
Anal. Chem. 2007, 79, 736-743

Enantioseparation and Stacking of Cyanobenz[f]isoindole-Amino Acids by Reverse Polarity Capillary Electrophoresis and Sulfated β-Cyclodextrin Daniel L. Kirschner, Michael Jaramillo, and Thomas K. Green*

Department of Chemistry & Biochemistry, Institute of Arctic Biology, University of Alaska Fairbanks, Fairbanks, Alaska 99775

A capillary electrophoresis method with laser-induced fluorescence detection for the chiral separation of cyanobenz[f]isoindole (CBI) derivatives of amino acids was developed and optimized. The enantioseparations are accomplished with sulfated β-CD (S-β-CD) as chiral selector at low pH and reverse polarity. BGE conditions were optimized for CBI-serine and then applied to other CBIamino acids. Baseline resolution of 13 CBI-amino acids was achieved using a single BGE formulation of 2 wt % S-β-CD in 25 mM phosphate buffer at pH 2.00 and a voltage of -30 kV. pH is the most critical BGE parameter affecting resolution. At 2 wt % S-β-CD, CBI-serine enantiomers are baseline-resolved at pH 2.00 but no resolution is obtained at pH 3.00. L-Glutamate, L-aspartate and D-serine are simultaneously quantified in the microdialysate of an arctic ground squirrel to illustrate the application to biological samples. Dilute solutions of the CBIamino acids in water can be stacked by hydrodynamic injection with a 100-fold improvement in signal-to-noise ratio without loss of chiral resolution. The stacking is proposed to consist of field-amplified migration, pHmediated stacking, and sweeping by S-β-CD. The limit of detections for CBI-DL-serine and CBI-DL-glutamate are determined as 0.20 and 0.30 nM, respectively. The stacking method was not applicable to the high ionic strength microdialysates.

furoyl)quinoline-2-carboxaldehyde,8,9 and 3-(4-carboxybenzoyl)quinoline-2-carboxaldehyde (CBQCA),10 among others. Of these, NDA is the most frequently employed due to its rapid reaction kinetics,7 enhanced stability, and sensitivity compared to OPA,11 and the excitation maximums of 420 and 450 nm, which closely match those of a violet diode laser (405 nm),12 He-Cd laser (440 nm),13 or common argon laser (458 nm).14 The focus of our research was to develop an offline CE-LIF method that utilizes NDA for simultaneous monitoring of D-serine (D-ser) and L-glutamate (L-glu) in microdialysates of arctic ground squirrels. Recent studies have shown D-ser to be the primary endogeneous amino acid that binds to the glycine site of N-methylD-aspartate (NMDA) receptor,15 a highly regulated glutamatergic, ligand-gated ion channel involved in a variety of physiological functions and disorders including memory, learning, pain, and ischemia. D-Ser and L-glu are both required for activation of the NMDA receptor. Most methods for chiral separation of fluorescent amino acid derivatives rely on cyclodextrin micellar electrokinetic chromatography (CD-MEKC), in which a micellar pseudostationary phase is combined with a CD, usually under high-pH conditions, to achieve chiral resolution. Using CD-MEKC, chiral separation of D-ser from L-ser has been achieved using OPA,16 CBQCA,10 and NDA17 as labeling reagents. Recently, Zhao et al. employed a CDMEKC system consisting of a dual chiral selector of sodium deoxycholate and β-CD at high pH for the separation of cyanobenz-

Capillary electrophoresis combined with laser-induced fluorescence (CE-LIF) provides high sensitivity for amino acid analysis1,2 and is especially useful in neurochemistry studies where subnanomolar concentrations of amino acid neurotransmitters are often encountered.3,4 There are numerous dye molecules available for fluorescence labeling of amino acids including o-phthalaldehyde (OPA),5naphthalene-2,3-dicarboxaldehyde (NDA),6,7 3-(2-

(6) Shou, M.; Smith, A. D.; Shackman, J. G.; Peris, J.; Kennedy, R. T. J. Neurosci. Methods 2004, 128, 189-197. (7) Robert, F.; Bert, L.; Denoroy, L.; Renaud, B. Anal. Chem. 1995, 67, 18381844. (8) Chen, Z.; Wu, J.; Baker, G. B.; Parent, M.; Dovichi, N. J. J. Chromatogr., A 2001, 912, 293-298. (9) Velede, M. T.; de Frutos, M.; Diez-Masa, J. C. J. Chromatogr., A 2005, 1079, 335-343. (10) Thongkhao-On, K.; Kottegoda, S.; Pulido, J. S.; Shippy, S., A. Electrophoresis 2004, 25, 2978-2984. (11) Roach, M. C.; Harmony, M. D. Anal. Chem. 1987, 59, 411-415. (12) Melanson J. E.; Lucy, C. A. Analyst 2000, 125 1049-1052. (13) Orwar, O.; Fishman, H. A.; Ziv, N. E.; Scheller, R. H.; Zare, R. N. Anal. Chem. 1995, 67, 4261-4268. (14) Quan, Z.; Song, Y.; Feng, Y.; LeBlanc, M. H.; Liu, Y.-M. Anal. Chem. Acta 2005 528, 101-106. (15) Shleper, M.; Kartvelishvily, E.; Wolosker, H. J. Neurosci. 2005, 25, 94139417. (16) Ciriacks, C. M.; Bowser, M. T. Anal. Chem. 2004, 76, 6582-6587. (17) Zhao, S.; Song, Y.; Liu, Y.-M. Talanta 2005 67 212-216.

* To whom correspondence should be addressed. E-mail: [email protected]. (1) Poinsot, V.; Lacroix, M.; Maury, D.; Chataigne, G.; Feurer, B.; Couderc, F. Electrophoresis 2006, 27, 176-194. (2) Poinsot, V.; Bayle, C.; Couderc, F. Electrophoresis 2003, 24, 4047-4062. (3) Paez, X.; Hernandex, L. Biopharm. Drug Dispos. 2001, 22, 273-289. (4) Kennedy, R. T.; Watson, C. J.; Haskins, W. E.; Strecker, D. H. Curr. Opin. Chem. Biol. 2002, 6, 659-665. (5) Boyd, B. W.; Witowski, S. R.; Kennedy, R. T. Anal. Chem. 2000, 72, 865871.

736 Analytical Chemistry, Vol. 79, No. 2, January 15, 2007

10.1021/ac061725+ CCC: $37.00

© 2007 American Chemical Society Published on Web 12/09/2006

[f]isoindole (CBI)-ser enantiomers,17 However, their method does not report on the detection of L-glu simultaneous with D-ser. DeSilva and Kawana also separated a series of CBI-amino acids with this approach, but CBI-ser was not chirally resolved.18 Quan and Liu chirally resolved CBI-glu derivatives using CD-MEKC.19 The chiral separation of NBD-F-ser enantiomers was recently optimized using a hydroxypropyl-β-CD (no micelle) at high pH.20 An alternative for chiral resolution takes advantage of a negatively charged CD (no micelle) at low-pH conditions and reverse polarity, sometimes referred to as CD electrokinetic chromatography (CDEKC).21 As far as we are aware, there is only one report of this approach for the separation of fluorescently tagged amino acids.22 Using reverse polarity and low pH, Piehl et al. were able to baseline resolve DNS-amino acids on a microchip using sulfated γ-CD, with migration times in the subsecond time scale.22 Sulfated β-CD (S-β-CD) has also found use as a powerful chiral selector.23-26 We report here the chiral resolution of CBI-amino acid enantiomer pairs using CD-EKC with S-β-CD, including CBIser, CBI-asp, and CBI-glu. Low pH and reverse polarity conditions were employed. Conditions for separation were optimized for CBIser and then applied to other CBI-amino acids. Of 17 CBI-amino acids pairs studied, 13 pairs could be baseline resolved using a single BGE formulation. The technique is applied to brain microdialysates from the hippocampus of an arctic ground squirrel. D-Ser, L-glu, and L-asp are all quantified in the dialysates. D-Glu and D-asp are shown to be absent or at concentrations below the limit of detection (LOD). On-line preconcentration techniques such as stacking and sweeping are often employed in CE techniques to improve detection sensitivity, an inherent limitation of CE compared to HPLC.27,28 Otsuka et al. applied the technique of stacking by reverse migrating pseudostationary phase in combination with CDEKC to obtain peak enhancements in the chiral analysis of triadimenol, a fungicide.21 A 10-fold increase in signal was obtained due to stacking. More recently, Shih and Lin reported a fullcapillary sample stacking/sweeping-MEKC technique for the separation of NDA-derivatized tryptophan and isoleucine.29 In their technique, the capillary is completely filled with analyte in a lowconductivity buffer at pH 4.6. Voltage is then applied using a pH 2.0 background electrolyte (BGE) at both inlet and outlet. The anionic analyte is neutralized and stacks at the pH junction while the EOF begins to move the stacked analyte toward the outlet. Prior to ejection from the outlet, the BGE is changed to a micellar BGE, and the micelles then sweep the analyte toward the detector. (18) Desilva, K.; Kuwana, T. Biomed. Chromatogr. 1997, 11, 230-235. (19) Quan, Z.; Liu, Y.-M. Electrophoresis 2003, 24, 1092-1096. (20) Zhao, S.; Yuan, H.; Xiao, D. J. Chromatogr., B 2005, 822, 334-338. (21) Otsuka, K.; Matsummura, M.; Kim, J.-B.; Terabe, S. J. Pharm. Biomed. Anal. 2003, 30, 1861-1867. (22) Piehl, N.; Ludwig, M.; Belder, D. Electrophoresis 2004, 25, 3848-3852. (23) Stalcup, A. M.; Gahm, K. H. Anal. Chem. 1996, 68, 1360-1368. (24) Evans, C. E.; Stalcup, A. M. Chirality 2003, 15, 709-723. (25) Kavran-Belin, G.; Rudaz, S.; Veuthey, J.-L. J. Sep. Sci, 2005, 28, 21872192. (26) Daniel, C.; Lipka, E.; Bonte, J-P.; Goossens, J-F.; Vaccher, C.; Foulon, C. Electrophoresis 2005, 26, 3824-3832. (27) Beckers, J. L.; Bocek, P. Electrophoresis 2000, 21, 2747-2767. (28) Osborn, D. M.; Weiss, D. J.; Lunte, C. E. Electrophoresis 2000, 21, 27682779. (29) Shih, C.-M.; Lin, C.-H. Electrophoresis 2005, 26 (18), 3495-3499.

A detection limit of ∼10-9 M is achieved with a 75-µm capillary filled with 60 cm of analyte (2.65 µL). Here we also report a new technique for stacking of CBI-amino acids but which also provides chiral resolution. The technique is proposed to be a combination field-amplified and pH-mediated stacking, followed by sweeping by the highly anionic S-β-CD and stacking at a dynamic pH junction. Dilute water solutions of racemic CBI-ser and CBI-glu are employed to illustrate the technique. Enhanced signal-to-noise ratios of ∼100 are achieved for each enantiomer without loss of chiral resolution, with detection limits in the subnanomolar range for each enantiomer using a 25-µm capillary filled with 19.5 cm of analyte (0.096 µL). The technique does not require the switching of BGE during analysis. EXPERIMENTAL SECTION Chemicals and Reagents. All chemicals were analytical reagent grade unless otherwise stated. NDA and sodium cyanide were purchased from Sigma-Aldrich. All amino acids were purchased from Sigma-Aldrich. Water was prepared from a Millipore Milli Q ultrapure water purification system with 18 MΩ‚ cm resistivity. S-β-CD was purchased from Fluka. Phosphoric acid and sodium monohydrogen phosphate were obtained from Fisher. BGEs were typically 25 mM phosphate adjusted to a desired pH by mixing equal concentrations of aqueous 25 mM sodium monophosphate and 25 mM phosphoric acid solutions in the appropriate ratios. Capillary Electrophoresis Apparatus. The capillary electrophoresis instrument was in-house built using an Olympus BH2 microscope. A high-voltage power supply (0-30 kV, Spellman CZE 1000R) was used to drive the electrophoresis. The bare silica capillary (Polymicro Technologies) was focused with an oil immersion 100× 1.3 na, 0.17 working distance Olympus UVFL fluorite microscope objective. A 420-nm blue diode laser (Power Technology), adjusted to 5-mW power, was used for excitation. The laser beam passed initially through a bandpass exciter filter (D425, 40× Chroma Technology) followed by reflection of 90° by a dichroic filter (460DCLP, Chroma Technology) prior to focusing on the capillary. Fluorescence emission passed back through the dichroic filter and then through a bandpass emission filter (D490/40m) to a PMT (R1527, Hamamatsu) housed in a Photon Technology Instrument DB104 photometer. The analog signal was converted to digital signal, stored, and processed using LabView software (National Instruments). The signal was collected at 200 Hz and signal averaged to 20 Hz. Injection was accomplished by computer-controlled application of vacuum at the outlet side of the capillary for a specified time. PeakFit software was used to fit the peaks to a Gaussian line shape to determine resolution values, Rs, and noise levels to the 95% confidence interval. Derivatization Procedure for Separation and Stacking Studies. CBI-amino acids were prepared for separation and stacking studies by reacting equal volumes of amino acid (1 mM), NDA (1 mM in methanol), and NaCN (1 mM in 60 mM borate buffer, pH 9.3) and allowed to react for a minimum of 15 min. CBI-amino acids were diluted to 10 µM with water. The use of equal molar quantites of amino acid, NDA, and NaCN were used to minimize NDA side products for the separation and stacking Analytical Chemistry, Vol. 79, No. 2, January 15, 2007

737

studies. For derivatization of microdialysates and standards for calibration curves, see below. Capillary Electrophoresis and Enantioseparation Studies. All electrophoresis experiments were operated with reverse polarity (anode at detector). For separation studies, the capillary was 25-µm i.d., 70-cm total length, and 45 cm to the detector. Injection was accomplished by vacuum injection at the detector end. Typical currents ranged from 20 to 60 µA. Typical injection times were 1 s, but ranged from 2 to 180 s for stacking experiments. Injection vacuum was 380 mbar, which gave an injection plug of 19.5 cm for 180-s injection. New capillaries were conditioned once for 8 min with 1 M NaOH, then for 2 min with 0.1 M NaOH, and 2 min with BGE between each experiment. Construction of Standard Curves and Analysis of Microdialysates. All biological samples and standards were analyzed using a bare fused-silica capillary of 25 µm i.d. × 48 cm (45 cm effective). Voltage was reduced to 21 kV to allow for similar applied field conditions as described for separation studies. Five standards were prepared for CBI-D-ser, CBI-DL-asp, and CBI-DL-glu with concentrations of 0.3, 0.8, 1.4, 2.4, and 5.0 µM by dilution of a 2 mM stock amino acid solution with artificial cerebral spinal fluid (124 mM NaCl, 2.7 mM KCl, 1 mM CaCl2, 0.85 mM MgCl2, 24 mM NaHCO3, 1.4 mM D-glucose, pH 7.4). Standards and samples were reacted by equal volume dilutions with NDA (1 mM in methanol) and NaCN (1 mM in 60 mM tetraborate with 9 µM L-homoarginine as the internal standard). Capillary was preconditioned and the standards/samples were injected using 1-s vacuum injection (380 mbar). Using peak areas, the CBI-amino acid/internal standard ratio was plotted versus [CBI-amino acid]. A five-point calibration curve was constructed for each amino acid (n g 4 at each concentration) using a weighted least-squares analysis. For microdialysis, typically 1-2 µL of dialysate was reacted. Samples and standards were reacted a minimum of 15 min prior to analysis. Identities of CBI-amino acids in the microdialysates were confirmed by spiking with known CBI-amino acids. Triplicate analyses were conducted on all microdialysates. Concentrations of amino acids were determined from calibration curves using a weighted least-squares analysis, with uncertainty reported as (s, where s is the standard deviation. The average and standard deviation in peak area ratios (CBI-amino acid/ internal standard) for nine injections of 2.4 µM CBI-amino acid were as follows; CBI-L-glu, 0.461 ( 0.070; CBI-D-ser, 0.400 ( 0.085, CBI-L-asp, 0.358 ( 0.036. 13C NMR Titration of Ser and CBI-Ser. All 13C NMR measurements were made with a Varian Mercury FT-NMR at 75 MHz. Ser titration: 15 mL of 60 mM borate buffer, pH 9.3, was mixed with 5 mL of methanol-d4 and then 0.50 g of L-ser in 15 mL of D2O. The pH was 8.70 using a H2O-calibrated pH meter. The solution was titrated incrementally with 1.0 mM HCl followed by measurement of the 13C NMR chemical shift of the carboxyl carbon referenced to methanol-d4 (49.9 ppm). CBI-ser titration: The same procedure was used except 15 mL of 60 mM borate buffer, pH 9.3, containing 15 mM NaCN was mixed with 15 mM NDA in 15 mL of methanol and then with 15 mL of (C-1)-13Cenriched L-ser in D2O. The pKaH*, which determined in D2O using an H2O-calibrated pH meter, was determined by plotting the first derivative of the chemical shift in ppm against pH*. The peak 738 Analytical Chemistry, Vol. 79, No. 2, January 15, 2007

Figure 1. Influence of BGE pH on chiral resolution of CBI-ser. BGE: 2 wt % S-β-CD, 25 mM phosphate.

maximum was taken as the pKa*. The pKa was determined from pKa ) 0.929pKaH* + 0.42.30 RESULTS AND DISCUSSION Reverse Polarity CE. Reverse polarity was utilized in all experiments, where the anode is at the detector end. Under the conditions of low pH (2.00-3.50), electroosmotic flow is minimized, and the negatively charged S-β-CD will migrate toward the detector. The hydrophobic naphthalene ring of the CBI-amino acid is expected to form a strong inclusion complex with S-β-CD, causing the CBI-amino acid to migrate toward the detector. The migration time, to a first approximation, is a measure of the strength of interaction of the CBI-amino acid with S-β-CD; a CBIderivative that interacts strongly will migrate faster. Chiral Separation of CBI Derivatives of Serine, Glutamate, Alanine, and Arginine. Because of the importance of D-ser in the mammalian brain,31 chiral separation of CBI-ser was optimized prior to applying the BGE conditions to other CBI-amino acids. The literature reveals that CBI-ser enantioseparation presents one of the more challenging chiral separations among the CBI-amino acids.17,18 In the range of 2-6 wt % S-β-CD, we found pH of the BGE to be the most critical factor controlling resolution, as shown in Figure 1. The enantiomers are baseline resolved at pH 2.00 (Rs 1.72), but resolution completely vanishes at pH 3.00. pH has only a minimal effect on migration time, not unexpectedly given the minimized EOF in this pH range. pH studies were also conducted on CBI-ala, CBI-glu, and CBI-arg. The results are shown in Figure 2, where resolution is plotted against pH of the BGE. Resolution of CBI-ser, CBI-glu, and CBI-ala enantiomers increases linearly with decreasing pH, with similar negative slopes, suggesting a common mechanism for chiral recognition by S-β-CD among these derivatives. In contrast, CBI-arg resolution decreases with decreasing pH (positive slope). CBI-amino acids contain carboxyl groups, which are increasingly ionized over the pH range investigated in this study. Not only will degree of ionization affect electromigration, but it may also change the nature of the interaction with S-β-CD, and thus affect chiral resolution. To our knowledge, no pKa values of CBI(30) Krezel, A.; Wojciech, B. J. Inorg. Biochem. 2004, 98 (2004) 161-166. (31) Mustafa, A. K.; Kim, P. M.; Snyder, S. H. Neuron Glia Biol. 2004, 1, 275281.

Figure 2. Effect of BGE pH on chiral resolution of CBI-amino acids. BGE: 2 wt % S-β-CD, 25 mM phosphate. Each Rs value represents an average (n ) 3).

Figure 3. (A) 13C NMR titration of L-ser and CBI-L-ser. 13C chemical shift of carboxyl group monitored. (B) First derivative of titration curve. Peak maximums yield pKa*.

amino acid derivatives have been experimentally determined. Utilizing 13C NMR spectroscopy, the chemical shift of the carboxyl carbon of 13C-enriched CBI-L-ser (99.0 atom % at C-1) was determined over the pH* range 1-6. pH* is the direct reading of the D2O/methanol solution used in the titration with an H2Ocalibrated pH meter. The titration curves are shown in Figure 3A, along with results on L-ser. The first-derivative curves for these titrations are shown in Figure 3B. The pKaH* values for L-ser and CBI-L-ser are determined to be 1.9 and 2.8, respectively, according to the peak maximums. A recently published equation allows for converting these pKaH* values determined in D2O into an H2O

Figure 4. Chiral separation of a mixture of racemic CBI-arg, CBIala, CBI-ser, and CBI-glu. D enantiomer migrates faster for each pair, except for CBI-glu. BGE: pH 2.00, 25 mM phosphate with 2% S-βCD.

equivalent, thus yielding pKaH values of 2.2 for L-ser and 3.0 for CBI-L-ser.30 The accepted pKa of ser is 2.2.32 These results demonstrate that incorporation of the amino function into the NDA molecule substantially decreases the acidity of the carboxyl group of serine. A pKa of 3.0 for CBI-L-ser means that it is 50% ionized at pH 3.0. Thus, chiral resolution may be lost at pH 3.0 due to electrostatic repulsions with the anionic S-β-CD. The anionic form of CBI-ser can migrate with its own mobility at pH 3.0, but the migration time is 24.5 min in the absence of S-β-CD, much longer compared to 10.0 min using 2 wt % S-β-CD at pH 3.0. Thus, CBIser strongly interacts with S-β-CD in the pH range where chiral resolution is lost. The loss of chiral resolution is therefore more probably related to the change in the nature of this interaction as pH increases, not a total loss of interaction. At pH 3.00, the anionic carboxylate group may be positioned above the CD cavity due to repulsions with the anionic sulfate groups, which may prevent the naphthyl group from full inclusion into the CD cavity. At pH 2.00, CBI-ser is 90% protonated at the carboxyl group, which may allow deeper penetration of the naphthyl group into the CD cavity and, thus, allow stereoselective binding in the vicinity of the R-carbon of the amino acid structure. It is also possible that hydrogen bonding between the CD and carboxyl group becomes important in the fully protonated form of the CBI derivative. In contrast to CBI-ser, CBI-glu, and CBI-ala, CBI-arg resolution decreases with decreasing pH (positive slope) with nearly 2-fold resolution improvement when pH is increased from 2.0 to 3.5 as shown in Figure 2. The guanidine functional group of CBI-arg is fully protonated over the pH range investigated here, and thus, electrostatic attraction between the cationic CBI-arg and the anionic S-β-CD is expected at low pH. This electrostatic attraction is likely the primary cause of the relatively fast electromigration of CBI-arg relative to other amino acids at pH 2.0 as shown in Figure 4. Evidence further implicating an electrostatic interaction for CBI-arg is shown when plotting concentration of S-β-CD as a function of the average chiral pair migration time as shown in Figure 5. CBI-arg electromigration does not appreciably change over the S-β-CD concentration investigated whereas CBI-ser, CBIglu, and CBI-ala all show appreciable change. (32) Nelson, D. L.; Cox, M. M. Lehninger Principles of Biochemistry, 3rd ed.; W. H. Freeman: New York, 2004.

Analytical Chemistry, Vol. 79, No. 2, January 15, 2007

739

Figure 5. Effect of S-β-CD concentration on the migration time of CBI-amino acid pairs. BGE: pH 2.00, 25 mM phosphate. Each chiral pair migration time represents an average (n ) 3).

Figure 6. CBI-arg electromigration time as a function of BGE pH. BGE: 2 wt % S-β-CD, 25 mM phosphate. Each data point is an average (n ) 3).

Finally, as pH is increased, CBI-arg migration time dramatically increases, in contrast to the nonbasic CBI-amino acids (Figure 6). This increase in migration time with pH can be attributed to changes in the degree of ionization of the CBI-arg carboxylic acid group with increasing pH. Indeed, at pH 3.5, CBI-arg is approaching a neutral, zwitterionic form with decreased electrostatic interaction with S-β-CD, and its migration time approaches that of nonbasic CBI-amino acids at pH 2.00. The impact of phosphate buffer concentration on resolution is shown in Figure 7. Increasing phosphate concentration beyond 25 mM is detrimental to resolution for three of the four CBI-amino acid pairs shown. In summary, we chose a BGE of 2 wt % S-β-CD, 25 mM phosphate, pH 2.00 for application to other CBI-amino acid pairs. Using a fused-silica capillary of 25 µm i.d. × 70 cm (45 cm effective) and a separation voltage of -30 kV, currents were typically ∼30 µA. An Ohm’s law plot revealed a linear relationship between current (µA) and voltage over the range of 5-30 kV (Correlation coefficient, R2 ) 0.991) demonstrating that Joule heating should not be a significant factor at an operating current of 30 µA.33 (33) Weinberger, R. Practical Capillary Electrophoresis, 2nd ed., Academic Press: San Diego, 2000.

740 Analytical Chemistry, Vol. 79, No. 2, January 15, 2007

Figure 7. Influence of phosphate buffer concentration on chiral resolution of CBI-amino acid pairs. BGE: 2 wt % S-β-CD, pH 2.00. Each resolution value is an average (n ) 3).

Chiral Separation of CBI-Amino Acids Using pH 2.00 and 2 or 6 wt % S-β-CD. S-β-CD concentration in the BGE has a less dramatic affect than pH on chiral resolution, as shown in Table 1. Using a separation voltage of -30 kV and a single BGE formulation of 2 wt % S-β-CD, 25 mM phosphate buffer, pH 2.00, 13 CBI-amino acid pairs were baseline resolved (Rs g 1.5). As expected, the use of 6 wt % S-β-CD concentration dramatically decreases migration times for most CBI-amino acids but, in most cases, results in lower resolution. Only valine, leucine, and lysine were not well resolved in this study. Both valine and leucine have symmetrical alkyl side chains, and alternative BGE formulations were not successful in providing resolution. Lysine probably reacts preferentially with NDA at the primary amine of the side chain, placing the chiral R-carbon four bonds away from the naphthalene unit, where interaction with S-β-CD is expected. Further attempts to resolve CBI-lysine were also unsuccessful. The migration times for CBI-arg, CBI-his, and CBI-lys, all with basic side chains, do not change appreciably at high S-β-CD concentrations, indicative of a strong electrostatic interaction of these cationic CBI-amino acids with S-β-CD, as discussed above for CBI-arg. Analysis of D-Ser, DL-Asp, and DL-Glu in the Hippocampus of Arctic Ground Squirrel. To demonstrate application of this new method, we chose to quantify levels of CBI-DL-amino acids sampled from the brain of live arctic ground squirrels (AGS). Using a standard mixture of 22 primary amines commonly found in brain microdialysates, no interfering species were identified for CBI-D-ser, CBI-DL-asp, and CBI-DL-glu (data not shown). Calibration curves were constructed for low-micromolar detection. An internal standard of L-homoarginine was utilized to minimize error in quantitative analysis. A typical electropherogram of a microdialysate from an AGS is illustrated in Figure 8. Appreciable levels of D-ser, L-asp, and L-glu were observed, while D-asp and D-glu are found only as trace components below the LOD. Quantitative analysis of CBI-AA concentrations is shown in Table 2, along with equations of the standard curves used for analysis. Stacking of CBI-Ser and CBI-Glu. CBI-ser and CBI-glu standards diluted in water could be stacked by vacuum injection as shown in Figure 9, where the injection time ranges from 2 to 120 s (380 mbar) for a 0.05 µM solution of each enantiomer. We observe enhancements of S/N of ∼100 for a 120-s injection

Table 1. Migration Times and Resolution Values for 17 CBI-DL-Amino Acids at Both 2 and 6% S-β-CDa migration time (s) 6 wt % S-β-CD CBI-amino acid tyrosine threonine asparagine phenylalanine histidine glutamic acid methionine alanine arginine aspartic acid isoleucine serine tryptophan glutamine valine leucine lysine

resolution

2 wt % S-β-CD

D/L 411/361 402/428 358/374 347/328 306/313 421/413 367/360 353/358 287/290 403/408 331/341 392/397 392/366 371 334/335 344 291

D/L 636/509 639/704 512/549 504/460 325/336 690/667 ( 39/37 545/527 525/539 ( 24/25 307/312 ( 8/8 637/648 485/506 621/632 ( 31/32 576/513 580/571 492 462 289

6 wt % S-β-CD

2 wt % S-β-CD

Rs 17.66 8.2 6.02 5.17 3.22 2.67 2.55 1.92 1.89 1.86 1.77 1.58 1.91 0 0.51 0 0

Rs 15.74 8.89 7.15 3.95 4.88 3.29 ( 0.08 3.02 2.79 ( 0.11 2.37 ( 0.15 1.79 3.16 1.69 ( 0.05 2.74 1.45 0 0 0

a BGE: 25 mM phosphate, pH 2.00. All CBI-amino acids pairs were 10 µM total (1:2 D/L). 6 wt % S-β-CD, n ) 1; 2 wt %, n ) 3 except for CBI-glu, CBI-arg, CBI-ala, and CBI-ser, n ) 9.

Figure 8. Typical electropherogram of a microdialysate from the hippocampus of an AGS. (A) Standard amino acid mixture including 1, CBI-D-ser; 2, CBI-L-ser + byproduct; 3, CBI-D-asp; 4, CBI-L-asp; 5, CBI-L-glu; 6, CBI-D-glu. (B) Microdialysate obtained 40 min after probe insertion at flow rate of 0.6 µL/min. The internal standard is labeled as peak 0. Identities of CBI-amino acids in microdialysates were determined by spiking with known CBI-amino acids. Table 2. Quantitative Analysis of CBI-AA Levels in Dialysates from Hippocampus of an AGS CBI-AA

standard curvea

R2

CLOD, nMb

[dialysate], µMc

[dialysate], µMd

CBI-D-ser CBI-L-asp CBI-L-glu CBI-D-asp CBI-D-glu

Y ) 0.171x - 0.005 Y ) 0.145x + 0.005 Y ) 0.203x - 0.013 Y ) 0.130x - 0.004 Y ) 0.203x + 0.004

0.996 0.998 0.998 0.991 0.997

43.8 ( 0.4 40.3 ( 0.4 32.3 ( 0.3 57.1 ( 0.1 41.2 ( 0.4

0.25 ( 0.09 1.54 ( 0.23 0.86 ( 0.06 trace trace

0.23 ( 0.09 0.43 ( 0.11 0.57 ( 0.15 -

a y ) mx + b. (CBI-AA/internal std) ) m[CBI-AA] + b. b Concentration limits of detection (CLOD) calculated from response of low-concentration standard of 0.3 µM (n ) 4) assuming S/N ) 3. c 20 min after probe insertion. d 110 min after probe insertion.

compared to 2-s injection, with only a moderate loss of chiral resolution (Table 3). A 180-s injection of 1 nM of each enantiomer of CBI-ser and CBI-glu is shown in Figure 10. A detection limit of 0.3 nM of CBI-ser and 0.5 nM CBI-glu is calculated (S/N ) 3) from this electropherogram. Using long injections of concentrated blue dye, an injection plug of ∼19.5 cm was visualized for a 180-s injection, corresponding to an injection volume of 96 nL. The stacking technique can also be applied to more complex mixtures of amino acids without loss of resolution, as shown in

Figure 11. In this figure, the electropherograms of 1- and 60-s injections are compared using a mixture of 11 CBI-amino acids (D and L). For all racemic CBI-amino acids present in the mixture (DL-arg, DL-his, DL-ser, and DL-glu), resolution is maintained under stacking conditions. The proposed stacking mechanism is illustrated in Figure 12. It involves a combination of field-amplified stacking, pH-mediated stacking, and sweeping by S-β-CD. A dilute solution of the analyte in water (pH ∼6) is injected hydrodynamically, filling up to 1/3 Analytical Chemistry, Vol. 79, No. 2, January 15, 2007

741

Figure 9. Stacking of CBI-ser and CBI-glu by hydrodynamic vacuum injection. Analyte concentration: 0.05 µM each enantiomer in water. BGE: 25 mM phosphate pH 2.00 with 3 wt % S-β-CD. Table 3. Signal to Noise (S/N) Ratio, Resolution (Rs), and Estimated Concentration Limits of Detection (CLOD) for CBI-Ser and CBI-Glu

injection, 2 5 10 20 60 120 180

sa

S/Nb CBI-ser, CBI-glu 2.6, 2.3 17.0, 15.8 31.4, 27.2 60.9, 51.9 135, 130 226, 186 14.9, 10.2

CLOD (nM)c CBI-ser, CBI-glu 57.4, 65.7 8.8, 9.5 4.8, 5.5 2.5, 2.9 1.1, 1.2 0.66, 0.81 0.20, 0.30

Figure 11. Stacking of a complex dilute sample of 11 CBI-amino acids (∼0.5 uM each). Injection time corresponds to 380 mbar vacuum. Peak identities: 1, CBI-D-arg; 2, CBI-L-arg; 3, CBI-D-his; 4, CBI-L-his; 5, CBI-gly; 6, CBI-L-tyr; 7, CBI-L-gln; 8, CBI-D-ser; 9, CBIL-ser; 10, CBI-L-glu; 11, CBI-D-glu. BGE: 25 mM phosphate, pH 2.00 with 2 wt % S-β-CD.

Rs CBI-ser, CBI-glu 1.92, 3.26 1.77, 3.32 1.78, 3.41 1.91, 3.38 1.57, 2.82 1.34, 2.46 1.72, 2.20

a 50 nM of each enantiomer hydrodynamically vacuum injected except 180-s injection, which was 1 nM each enantiomer (See Figure 9). b Noise was estimated at the 95% CI by PeakFit. c CLOD were estimated at a S/N ) 3.

Figure 12. Schematic diagram of the pH-mediated stacking/ sweeping-S-β-CD of racemic CBI-amino acids.

Figure 10. Electropherogram showing typical low detection limits obtained for low ionic strength solutions of racemic CBI-ser and CBIglu using CDEKC. Injection was 180 s (19.5 cm). BGE: 25 mM phosphate pH 2.00 with 3 wt % S-β-CD. Estimated LOD; CBI-ser, 0.20 nM; CBI-glu, 0.30 nM (S/N ) 3).

of the capillary length (A). Reverse polarity is then applied. The CBI-amino acids are anionic at this pH (except for the ones with basic side chains, see above) and will thus migrate toward the pH junction at the outlet side of the injection plug (B), where they will be substantially neutralized and stacked by the low pH buffer. The migration of the analyte anions should be enhanced due to field amplification in the dilute water. Simultaneously, EOF 742 Analytical Chemistry, Vol. 79, No. 2, January 15, 2007

will begin to pump water out of the capillary and move the stacked analyte band toward the inlet (C). During this period, the current is near zero or a few microamperes. Once the analyte is nearly ejected from the capillary, the current rises to nearly 30 µA as the highly anionic S-β-CD begins to migrate rapidly through the stacked band of analyte, sweeping it to the outlet (D). S-β-CD must be migrating into the capillary prior to expulsion of the stacked analyte by EOF; otherwise, the EOF would push out the analyte into the inlet vial and it would not be detected. Thus, we envision a combination of field-amplified migration, pH-mediated stacking, followed by sweeping by S-β-CD. Support of this mechanism comes from two experiments for which the detection window is placed 3 cm from the inlet, shown in Figure 13. In the first experiment (lower trace), 0.1 µM CBIL-ser in water was injected for 180 s (19.5 cm), followed by application of -30 kV in the presence of 25 mM phosphate buffer at pH 2.00 (no S-β-CD). A peak is observed at ∼60 s, which

Figure 13. Lower trace: Electropherogram illustrating stacking when operating BGE without S-β-CD at low-pH reversed polarity. The capillary (70 cm × 25 µm, 3 cm effective) was filled 19.5 cm with 0.10 µM CBI-L-ser in water (pH 5.2), and a voltage (-30 kV) was applied. The dilute CBI-ser is stacked at the pH boundary and driven toward the inlet vial and detected prior to exiting the capillary. Upper trace: electropherogram illustrating stacking/sweeping when operating BGE with S-β-CD. Once the stacked zone nears the inlet (peak 1), it reverses direction as S-β-CD enters the capillary and sweeps the stacked CBI-ser toward the outlet vial (peak 2). Table 4. Comparison of Full-Capillary Sample Stacking/ Sweeping-MEKC29 and pH-Mediated Stacking/ Sweeping-S-β-CD of CBI-Amino Acids

method analyte injection length, cm capillary i.d., µm injection volume, µL concn LOD, M mole detection limit, fmol switching of BGE migration time, min chiral separation plate number, N

FCSS/sweeping MEKCa

pH stacking/ sweeping S-β-CDb

CBI-Ile 60 (full capillary) 75 2.65 1.1 × 10 -9 2900

CBI-DL-ser 19.5 (28% of length) 25 0.096 0.3 × 10-9 29

Yes ∼30 no 1.2 × 105

No ∼11 yes 1.8 × 105

a Total and effective length of capillary 60/54 cm, violet LED 410 nm, 2 mW. b Total and effective length of capillary 75/45 cm, violet diode laser, 420 nm, 5 mW.

represents the pH-mediated stacked analyte being pushed back toward the inlet. This stacked analyte is ejected from the inlet since no S-β-CD is present. To illustrate sweeping, a second experiment is performed in the presence of S-β-CD. An additional peak is observed at ∼110 s, which represents the pH-mediated stacked analyte being swept back toward the detector by S-β-CD. Interestingly, the pH-mediated stacked peak at 60 s is considerably narrower than the sweeping peak, indicative of a destacking of (34) Gillogly, J. A.; Lunte, C. E. Electrophoresis 2005, 26, 633-639.

analyte upon its return from the inlet side of the capillary. The cause of the destacking is currently under investigation. The results suggest that pH-mediated stacking is principally responsible for the overall stacking effect observed. This proposed mechanism is similar to that proposed by Lin and Shih for CBI-amino acids using anionic SDS and low pH BGE,29 except their stacking/sweeping-MEKC apparently requires that initial BGE contain no SDS. After migration of the stacked analyte to nearly the inlet, the vials at both inlet and outlet are replaced to contain BGE-SDS. SDS then sweeps the stacked analyte toward the detector. Comparison of the techniques can be made and is shown in Table 4. The concentration detection limits are similar for dramatically different injection volumes, with the molar detection limit for pH-mediated stacking/sweeping-Sβ-CD being ∼100-fold more sensitive. Given the similar separation efficiencies (N) of the two methods, it is difficult to reconcile the different molar detection limits for the two methods. The application of this stacking technique to brain microdialysates discussed above was not possible due to the high ionic strength of the sample, as discussed by Gillogly and Lunte.34 Injections of microdialysates greater than 5 s resulted in significantly broader peaks with loss of resolution. We believe pretreatment of the sample, such as desalting, will be necessary for successful application of this stacking technique to microdialysates. The promise of the technique may lie in its ability to provide low detection limits for evaluation of enantiopurity of amino acids and other chiral molecules, using either UV or fluorescence detection. CONCLUSIONS Commercially available S-β-CD is found to be an excellent chiral selector for CBI-amino acids using low-pH BGE and reverse polarity. The strength of this CE technique lies in its ability to baseline resolve a large number of CBI-DL-amino acids with a single buffer formulation. D-Asp, D-ser, and D-glu can all be potentially quantified in neuronal samples, if present. In our study, basal levels of D-ser, L-glu, and L-asp, all important neurotransmitters/neuromodulators, were quantified in microdialysates from the hippocampus of an arctic ground squirrel. Stacking with enantioseparation of low ionic strength solutions of CBI-amino acids was demonstrated with detection limits of 0.20 and 0.30 nM for CBI-ser and CBI-glu, respectively. ACKNOWLEDGMENT We gratefully acknowledge the support of a grant from the U.S. Army Research Office (W911NF-05-1-0280) and a grant from the U.S. Army Medical Research and Materiel Command (05178001). M.J. acknowledges a summer fellowship from the INBRE Program, UAF. We thank Kelly Drew of UAF for microdialysate samples and valuable discussions. Received for review September 13, 2006. Accepted November 5, 2006. AC061725+

Analytical Chemistry, Vol. 79, No. 2, January 15, 2007

743