Anal. Chem. 1999, 71, 4997-5002
Technical Notes
Independent Optimization of Capillary Electrophoresis Separation and Native Fluorescence Detection Conditions for Indolamine and Catecholamine Measurements Young Hun Park, Xin Zhang, Stanislav S. Rubakhin, and Jonathan V. Sweedler*
Department of Chemistry, University of Illinois, Urbana, Illinois 61801
Separation conditions in capillary electrophoresis with native fluorescence detection often represent a compromise in terms of the separation and detection figures of merit. As both the separation and fluorescence properties greatly depend on pH, the ability to independently optimize pH in the separation capillary and the detection region can improve many complex separations. When using a sheath flow cell, the pH at the detection zone can be adjusted independently of the electrophoresis buffer pH. Using capillary electrophoresis with 257-nm excitation and native fluorescence detection, more than an order of magnitude improvement in the limits of detection for dopamine (from 1400 to 120 nM) and epinephrine (from 850 to 60 nM) is achieved by maintaining the basic separation conditions and an acidified sheath buffer. The detection of dopamine in an individual Aplysia californica cerebral ganglion neuron is demonstrated. Understanding the function and interactions between even simple networks of neurons is often hampered by an incomplete knowledge of the neurotransmitters and neuromodulatory agents used within each neuron in the network. The simultaneous detection and quantitation of neurotransmitters, metabolites, and cofactors at the level of individual neurons provide background and fundamental information for neurophysiology. Although there are hundreds of transmitters, the indolamines such as 5-hydroxytryptamine (serotonin) and the catecholamines (dopamine, epinephrine, octopamine, etc.) are of particular interest because of their widespread distribution in central and peripheral tissues, broad spectrum of biological action, and critical roles in diverse pathologies.1 While the mapping of indolamine- and catecholaminecontaining neurons in invertebrate central nervous systems (CNS) reveals a number of identified cells,2,3 only a few have been * Corresponding author: Tel.: (217) 244-7359 (voice). Fax: (217) 244-8068. E-mail:
[email protected]. (1) Cooper, J. R.; Bloom, F. E.; Roth, R. H. The Biochemical Basis of Neuropharmacology, 5th ed.; Oxford University Press: New York, 1986. (2) Steinbusch, H. W. M.; Mulder, A. H. In Handbook of Chemical Neuroanatomy, Classical Transmitters and Transmitter Receptors in the CNS; Bjorklund, A., Ho¨kfelt, T., Kuhar, M., Eds.; Elsevier: New York, 1984; Vol. 3, Part II. 10.1021/ac990659r CCC: $18.00 Published on Web 09/24/1999
© 1999 American Chemical Society
characterized biochemically.4-6 In general, past assays have generally been cell-system specific and often dependent on highly molecule-specific (e.g., enzymatic) detection methods. Thus, a considerable effort has been devoted to improved analytical methods for generally applicable multicomponent quantitative single-cell assays. Capillary electrophoresis (CE) offers a number of advantages for measuring the chemical composition of individual cells. It provides rapid separations, high separation efficiencies, and small sample volumes. As the extracellular and intracellular matrixes are aqueous and many of the analytes of interest are charged, they can be separated using typical free-zone electrophoretic conditions. The theory, technique, and capabilities of CE have been extensively reviewed over the last several years.7-11 Detection in CE can be challenging owing to the small analyte band volumes and the temporally narrow peaks.12 For typical capillaries, the peak volumes are in the picoliter to nanoliter regime; as many of the molecules of interest within a cell are in the nanomolar to micromolar range, this implies that femtomole to zeptomole detection sensitivities are required. While absorbance detection has been used to measure several NO-related products in an individual neuron separated using CE,13 most analytes cannot be detected using absorbance with sufficient sensitivity for single(3) Walker, R. J. In The Mollusca, Neurobiology and Behavior; Willows, A. O. D., Ed.; Academic Press: Orlando, Florida, 1986; Vol. 3, Part 2, pp 279485. (4) Osborne, N. N. Int. J. Neurosci. 1972, 3, 215-228. (5) Goldman, J. E.; Schwartz, J. H. J. Physiol. (London) 1974, 242, 61-76. (6) Hanley, M. R.; Cottrell, G. A.; Emson, P. C.; Fonnum, F. Nature (London) 1974, 251, 631-633. (7) Ewing, A. G.; Wallingford, R. A.; Olefirowicz, T. M. Anal. Chem. 1989, 61, 292A-303A. (8) Monnig, C. A.; Kennedy, R. T. Anal. Chem. 1994, 66, 280R-314R. (9) Capillary Electrophoresis of Small Molecules and Ions; Jandik, P., Bohn, G., Eds.; VCH Publishers: New York, 1993. (10) Handbook of Capillary Electrophoresis; Landers, J. P., Ed.; CRC Press: Boca Raton, FL, 1994. (11) Capillary Electrophoresis Technology; Guzman, N. A., Ed.; Marcel Dekker: New York, 1993. (12) Cruz, L.; Shippy, S. A.; Sweedler, J. V. In High Performance Capillary Electrophoresis, Khaledi, M., Ed.; John Wiley and Sons: New York, 1998; Vol. 146, pp 303-354. (13) Cruz, L.; Moroz, L. L.; Gillette, R.; Sweedler, J. V. J. Neurochem. 1997, 69, 110-115.
Analytical Chemistry, Vol. 71, No. 21, November 1, 1999 4997
cell assays. One problem is that highly selective detectors are required; even with the resolving power of CE, the thousands of components in a cell will not be resolved from each other without a selective detector. Hence, electrochemistry, mass spectrometry,and laser induced fluorescence (LIF) are the techniques most often applied to single-cell assays, with LIF providing the lowest limits of detection (LODs).14,15 Many of the compounds of interest are not fluorescent and so need to be derivatized before detection; tremendous progress has been made in precolumn, on-column, and postcolumn derivatization methods for cell studies.15 However, the best LODs are substantially degraded compared with the sensitivity inherent in LIF, and quantitation becomes more problematic; therefore, LIF approaches based on the native fluorescence of the analytes have been developed.16-24 When excited in the deep ultraviolet, many neurotransmitters and other molecules of interest have appreciable fluorescence; these include the catecholamines, indolamines, aromatic amino acids and peptides containing them, flavins, adenosine- and guanosine-nucleotide analogues, and others.19,21,22,24,25 While the fluorescence properties of such compounds, when excited by the appropriate ultraviolet wavelengths, are much poorer than those of common fluorophores such as rhodamine or fluorescein, a significant advantage is that no derivatization is required offering improved quantitation and ease of use in single-cell studies.15,24 In addition, native fluorescence using multiphoton excitation allows the sensitive detection of neurotransmitters26 and likely will be applied for single-cell assays soon. For complex samples such as cells, detection schemes with greater information content offer a number of advantages. Acquisition of the entire fluorescence emission spectrum at each time interval (wavelength-resolved fluorescence) provides such information.23 For wavelength-resolved native fluorescence, one advantage is that many of the analytes have different emission profiles so that they can be easily distinguished on the basis of both electrophoretic migration time and fluorescence properties. As different molecules can have vastly different fluorescence properties, the LODs can vary by orders of magnitude, ranging from 0.1 amol to 100 fmol.12,24 When injecting cells directly into a CE capillary, irreproducible migration times often result, as the surface of the capillary becomes coated with cellular proteins, but with wavelength-resolved fluorescence, misidentification is much less likely even under changing electrophoretic conditions. For (14) Swanek, F. D.; Ferris, S. S.; Ewing, A. G. In Handbook of Capillary Electrophoresis, 2nd ed.; Landers, J. P., Ed.; CRC Press: Boca Raton, FL, 1997; pp. 495-521. (15) Lillard, S. L.; Yeung, E. S. In Handbook of Capillary Electrophoresis, 2nd ed.; Landers, J. P., Ed.; CRC Press: Boca Raton, FL, 1997; pp 523-543. (16) Swaile, D. F.; Sepaniak, M. J. J. Liq. Chromatogr. 1991, 14, 869-893. (17) Lee, T. T.; Yeung, E. S. J. Chromatogr. 1992, 595, 319-325. (18) Chan, K. C.; Janini, G. M.; Muschik, G. M.; Issac, H. I. J. Liq. Chromatogr. 1993, 16, 1877-1890. (19) McGregor, D. A.; Yeung, E. S. J. Chromatogr., A 1994, 680, 491-496. (20) Timperman, A. T.; Khatib, K.; Sweedler, J. V. Anal. Chem. 1995, 67, 139144. (21) Chang, H. T.; Yeung, E. S. Anal. Chem. 1995, 67, 1079-1083. (22) Lillard, S. J.; Yeung, E. S.; Lautamo, R. M. A.; Mao, D. T. J. Chromatogr., A 1995, 718, 397-404. (23) Timperman, A. T.; Sweedler, J. V. Analyst (Cambridge, U.K.) 1996, 121, 45R-52R. (24) Fuller, R. R.; Moroz, L. L.; Gillette, R.; Sweedler, J. V. Neuron 1998, 20, 173-181. (25) Milofsky, R. E.; Yeung, E. S. Anal. Chem. 1993, 65, 153-157. (26) Gostkowski, M. L.; Shear, J. B. J. Am. Chem. Soc. 1998, 120, 12966-12967.
4998
Analytical Chemistry, Vol. 71, No. 21, November 1, 1999
native fluorescence, the excitation and emission properties of each analyte are highly variable as opposed to derivatized analytes in which all analytes have the same fluorophore and hence similar fluorescence characteristics. In the latter case, the detection conditions can be optimized for the fluorescent reagent used. For native fluorescence, each analyte has different chemical and electronic properties, so that for complex mixtures, the separation and detection conditions tend to be a compromise. For example, the fluorescence emission and electrophoretic migration rates of tyrosine (Tyr)-containing peptides are highly pH-dependent. If the buffer composition is optimized for resolution of a complex mixture, significant degradation of LODs may result. Pioneered by Dovichi, sheath flow cells for use in CE provide improved performance for LIF detection by reducing background contributions from scattered excitation light and the fluorescence from impurities in fused silica.27,28 In almost all cases, the composition of the sheath buffer and the CE buffer are similar; this minimizes possible density differences between the sheath and core and reduces light scattering from index-of-refraction changes at the sheath-core interface. The use of sheath flow cells also provides the ability to change the chemical composition of the capillary effluent prior to detection. Chemiluminescence detection of amino acids was accomplished by Dovichi and coworkers by exciting isoluminol isothiocarbamyl derivatives with compounds in the sheath flow which catalyze light emission.29 We have described the use of a sheath flow cell as a postcapillary reactor for fluorescence derivatization, by adding OPA and β-mercaptoethanol into the sheath solution, resulting in fluorescent products through the formation of covalent bonds.30 McGregor and Yeung reported preliminary results using a sheath buffer of different pH than that which was used for separation in an effort to improve sensitivity for native fluorescence detection of DNA fragments. In this case, the acidification/improvement in detection sensitivity was not successfully implemented so that alternative approaches were used.19 We report here the use of a sheath flow cell to enable independent optimization of separation and detection conditions to greatly increase the detection sensitivity for catecholamines. We demonstrate an order of magnitude improvement in detection sensitivity for dopamine and the ability to detect dopamine in single cells using this approach. EXPERIMENTAL SECTION Reagents. The electrophoresis running buffer was 80% borate (50 mM, pH 8.7) and 20% acetonitrile. The buffer was prepared using 3.0 g of boric acid (H3BO3; EM Science, Gibbstown, NJ, or Sigma, St. Louis, MO), 9.2 g of sodium borate (Na2B4O7‚10 H2O; EM Science, or Sigma) in 1.00 L of ultrapure water (Milli-Q filtration system; Millipore, Bedford, MA). The sheath fluid was either the 50 mM pH 8.7 borate buffer or 50 mM citrate buffer (1.0 g of citric acid; EM Science, 9.8 g of sodium citrate; EM Science in 1.00 L ultrapure water) at pH 3.0. All standards were obtained from Sigma and were reagent-quality or better. (27) Chen, D. Y.; Adelhelm, K.; Cheng, X. L.; Dovichi, N. J. Analyst (Cambridge, U.K.) 1994, 119, 349-352. (28) Wu, S.; Dovichi, N. J. J. Chromatogr. 1989, 480, 141-155. (29) Zhao, J. Y.; Labbe, J.; Dovichi, N. J. J. Microcolumn Sep. 1993, 5, 331339. (30) Oldenburg, K. E.; Xi, X.; Sweedler, J. V. Analyst (Cambridge, U.K.) 1997, 122, 1581-1585.
Cell Isolation. Aplysia californica were obtained from the Aplysia Research Facility (Miami, FL) and Marinus (Long Beach,CA) and were stored at 15° C. The procedure for cell isolation has been described previously;24,31 briefly, individual ganglia from the CNS were dissected from A. californica under cold anesthesia in molluscan physiological saline and incubated in proteolytic enzyme mixture (1% protease, Type IX, Sigma) at 37 °C. Next, the ganglionic connective tissues were removed with tungsten needles, and the identifiable neuron somas were isolated under a stereomicroscope using a microsyringe or micropipet. Within a minute, the neuron was placed in the microvial and homogenized by the combined action of tungsten-needle manipulation and hypoosmotic buffer damage. The samples were either diluted to 300 nL and injected directly into the capillary electrophoresis system for analysis or immediately frozen on dry ice for storage. Electrophoresis system and data processing. The in-house constructed CE system has been described in detail previously.24,32 In this case an 800-mm-long, 50-µm i.d./150-µm o.d., untreated fused-silica capillary (Polymicro Technologies, Tucson, AZ) was used for all experiments. A custom-made nano injector was used for all sample injections; the system consists of a servo-type linear actuator (model 850B-HS; Newport Corp., Irvine, CA) to move the inlet end of the capillary and position it vertically over a rotary stage (model 495; Newport) which was used to move samples and buffers into position under the capillary. Both stages were controlled by a programmable motion controller (model PMC200P; Newport). Injections were made from a stainless steel 316 disk which contained both 110-µL wells and supports for the removable microvials. For the cell injections, the microvial sample well was a 0.4-mm-deep cone, ∼300 nL, and was used for singlecell manipulation and injections. For all separations, the applied voltage was 21.0 kV, and the injections were performed electrokinetically with a 2.1 kV applied voltage for 10 s, for a total volume injected of ∼2 nL. The detection end of the capillary was directed into a laboratory-assembled sheath flow cell.32 The capillary was centered into a 1.0 mm × 1.0 mm × 20 mm quartz cuvette (NSG Precision Cells, Farmingdale, NY) serving as a sheath flow chamber. Sheath flow was generated by hydrodynamic pressure from a buffer-filled reservoir elevated above the height-matched injection wells and sheath outlet reservoir. Linear sheath flow velocity was 0.5 mm/ s. The sheath outlet reservoir was electrically connected to a 30kV power supply (Glassman, Whitehouse Station, NJ) for electrophoresis between the grounded injection disk and the sheath flow cell. Excitation of the core stream in the sheath flow cell was provided by a frequency-doubled, liquid-cooled argon-ion laser (Innova 300 FReD; Coherent, Palo Alto, CA) operating at 257 nm. Approximately 0.8 mW was directed into the sheath flow cell and focused to a spot either 0.5 or 1.5 mm below the capillary outlet with a 20-mm focal length quartz spherical concave lens (Spindler and Hoyer Inc., Medford, MA). Collection optics were orthogonal to the excitation beam and consisted of a 15× all-reflective microscope objective (Opticon, Billerica, MA) and a 30-mm focal length quartz spherical concave lens focusing the fluorescence (31) Floyd, P. D.; Moroz, L. L.; Gillette, R.; Sweedler, J. V. Anal. Chem. 1998, 70, 2243-2247. (32) Timperman, A. T.; Oldenburg, K. E.; Sweedler, J. V. Anal. Chem. 1995, 67, 3421-3426.
emission onto a 300-µm slit attached to an f/2.2 CP140 imagingspectrograph (Instruments SA, Edison, NJ). A 1024 × 256 detector-array, liquid nitrogen-cooled, scientific CCD (EEV15-11; Essex, U.K.) was mounted at the focal plane of the imaging spectrograph and was controlled by an AT200 controller card (Photometrics Ltd., Tucson, AZ). This CCD is over-coated with a phosphor down-converter (Metachrome II; Photometrics Ltd.) and has a nearly constant wavelength response from 200 to 900 nm. The CCD operated without a shutter and was oriented to permit subarray readout and binning while preserving the wavelength information over the 260-710 nm wavelength range focused across the CCD face; for all measurements presented here, the data were acquired with 2 × 256 binning (∼500 wavelength elements). CCD data were 16-bit and read out at 50 kHz/pixel. The CCD spectrograph was wavelength-calibrated with the 10 most intense lines from a Hg(Ne) pen lamp (Oriel Corp., Stratford, CT). Computer software written in C with both LabWindows v. 3.1 (National Instruments, Austin, TX) and Microsoft C++ V. 7.00 (Microsoft Corp., Redmond, VA) controlled all aspects of the injection, separation, data acquisition, and data storage. Briefly, programming code was installed into the CCD’s digital signal processor which, when triggered, would transfer a predefined readout array to the AT200 controller card. The trigger was programmed automatically and was generated by an A/D timing card (AT-MIO-16F; National Instruments). Spectral data returned from the CCD after each read were briefly manipulated and stored directly onto hard disk. Data acquisition was set to record and store complete spectra at 2 Hz. The wavelength-resolved CE data were processed first with custom despiking and backgroundsubtraction algorithms then processed and viewed in MATLAB (The Mathworks, Inc., Natick, MA) on an IBM RS/6000 workstation. To generate conventional electropherograms from the wavelength-resolved data, the appropriate wavelength range was selected, and the intensity integrated over this range (∫idλ).24 In essence, the one-dimensional electropherograms involved creating an optimally designed “optical filter” to extract analytical information after the wavelength-resolved data have been acquired. Signalto-noise ratios, quantitation, and reproducibility were determined according to standard techniques.24 RESULTS AND DISCUSSION While excellent performance (
