Monitoring Glutamate and Ascorbate in the Extracellular Space of

Monitoring Glutamate and Ascorbate in the Extracellular Space of Brain Tissue with Electrochemical Microsensors. Nadezhda V. ... Disk-Shaped Amperomet...
0 downloads 6 Views 90KB Size
Download PDF
Anal. Chem. 1999, 71, 5093-5100

Monitoring Glutamate and Ascorbate in the Extracellular Space of Brain Tissue with Electrochemical Microsensors Nadezhda V. Kulagina, Latha Shankar, and Adrian C. Michael*

Department of Chemistry, University of Pittsburgh, Pittsburgh, Pennsylvania 15260

This paper describes electrochemical microsensors for the in vivo measurement of glutamate and ascorbate in the extracellular space of brain tissue. To prepare glutamate microsensors, carbon fiber microelectrodes (10 µm in diameter and 300-400 µm long) were modified with a cross-linked redox polymer film containing enzymes. The microsensors were coated with a thin Nafion film before use. The glutamate microsensors were both selective and sensitive toward glutamate, with detection limits in the low micromolar range. Physiologically relevant concentrations of several electroactive compounds found in brain tissue produced no response at the glutamate microsensors and also did not affect their glutamate response, the only exception being glutamine, for which a small response was observed in the absence, but not in the presence, of glutamate. The ascorbate microsensors were used in conjunction with cyclic voltammetry. They were sensitive and selective toward ascorbate, but did exhibit a small sensitivity toward the dopamine metabolite, dihydroxyphenylacetic acid. The in vivo measurements performed establish the ability of the glutamate microsensors to monitor the component of the basal extracellular glutamate level that is derived from the neuronal activity of brain tissue. Glutamate is an important neurotransmitter in the mammalian central nervous system.1 Neuronal pathways in the brain that use glutamate as a transmitter are heavily implicated in several neurological disorders, such as schizophrenia, Parkinson’s disease, epilepsy, and stroke.2-7 These pathways are also implicated in drug abuse and addiction.8 Accurate information about the level of glutamate in the extracellular space of living brain tissue would contribute significantly to the fundamental understanding of the role of glutamate in these disorders and perhaps also guide * Corresponding author. Phone: (412) 624-8560. Fax: (412) 624-8611. E-mail: [email protected]. (1) Fonnum, F. J. Neurochem. 1984, 42, 1-11. (2) Carlsson, M.; Carlsson, A. Trends Neurosci. 1990, 13, 272-276. (3) Grace, A. A. Neuroscience 1991, 41, 1-24. (4) Klockgether, T.; Turski, L. Ann Neurol. 1993, 34, 585-593. (5) Moghaddam, B. J. Neurochem. 1993, 60, 1650-1657. (6) Morrow, B. A.; Clark, W. A.; Roth, R. H. Eur. J. Phamacol. 1993, 238, 255262. (7) Xue, C.-J.; Ng, J. P.; Li, Y.; Wolf, M. E. J. Neurochem. 1996, 67, 352-363. (8) Wolf, M. E. Prog. Neurobiol. (NY) 1998, 54, 697-720. 10.1021/ac990636c CCC: $18.00 Published on Web 10/13/1999

© 1999 American Chemical Society

advances in the therapeutic strategies available for their treatment. For this reason, there is a growing body of literature on the in vivo measurement and monitoring of glutamate.9-20 This paper describes in vivo measurements carried out in the brains of anesthetized rats with a new electrochemical microsensor that is both sensitive and selective toward glutamate. Two analytical techniques are frequently used for monitoring extracellular neurotransmitter levels in the brains of living animals: microdialysis and electrochemistry. Recent microdialysisbased investigations of glutamate in the brain have revealed that under resting conditions, i.e., in the absence of pharmacological or neuronal stimulation, the glutamate in brain microdialysate is principally derived from nonneuronal sources.17-20 The classical method for establishing the neuronal origin of a suspected neurotransmitter is to administer tetrodotoxin (TTX). This toxin blocks sodium-channels and prevents the conduction of action potentials along axons, thereby inhibiting neurotransmitter release from axon terminals. In several cases, however, addition of TTX to the microdialysis perfusion fluid did not cause a decrease in microdialysate glutamate levels. This finding leads to the conclusion that under resting conditions the glutamate recovered in the microdialysate is derived from the metabolic, rather than neuronal, activity of brain tissue. Electrochemical sensors for glutamate have recently been described. Electrochemical strategies for the detection of glutamate (9) Hu, Y.; Mitchell, K.; Albahadily, F.; Michaelis, E.; Wilson, G. Brain Res. 1994, 659, 117-125. (10) Poitry, S.; Poitry-Yamate, C.; Innocent, C.; Cosnier, S.; Tsacopoulos, M. Electrochim. Acta. 1997, 42, 3217-3223. (11) Lowry, J. P.; Ryan, M. R.; O’Neill, R. D. Anal. Commun. 1998, 35, 87-89. (12) Cordek, J.; Wang, X.; Tan, W. Anal. Chem. 1999, 71, 1529-1533. (13) Berners, M.; Boutelle, M.; Fillenz, M. Anal. Chem. 1994, 66, 2017-2021. (14) Zilkha, E.; Obrenovitch, T.; Koshy, A.; Kusakabe, H.; Bennetto, H. J. Neurosci. Methods 1995, 60, 1-9. (15) Niwa, O.; Torimitsu, K.; Morita, M.; Osborne, P.; Yamamoto, K. Anal. Chem. 1996, 68, 1865-1870. (16) Lada, M. W.; Vickroy, T. W.; Kennedy, R. T. Anal. Chem. 1997, 69, 45604565. (17) Lada, M. W.; Vickroy, T. W.; Kennedy, R. T. J. Neurochem. 1998, 70, 617625. (18) Herrera-Marschitz, M.; You, Z. B.; Goiny, M.; Meana, J.; Silveira, R.; Godukhin, O. V.; Chen, Y.; Espinoza, S.; Pettersson, E.; Loidl, C. F.; Lubec, G.; Andersson, K.; Nylander, I.; Terenius, L.; Ungerstedt, U. J. Neurochem. 1996, 66, 1726-1735. (19) Semba, J.; Kito, S.; Toru, M. J. Neural Transm. 1995, 100, 39-52. (20) Shiraishi, M.; Kamiyama, Y.; Huttemeier, P. C.; Benveniste, H. Brain Res. 1997, 759, 221-227.

Analytical Chemistry, Vol. 71, No. 22, November 15, 1999 5093

are based on one of two redox enzymes for which glutamate is the substrate: glutamate dehydrogenase or glutamate oxidase (Gox). Electrochemical sensors based on the former enzyme use the electrochemistry of the NAD/NADH redox couple,21-23 while sensors based on the latter usually monitor the production of hydrogen peroxide.9-11,24,25 The use of electrochemical glutamate sensors in the brain has been reported on several occasions,9,11 but it has not yet been demonstrated that the resting glutamate signal observed in the brain is sensitive to TTX. Thus, further development of microsensors specifically capable of monitoring the component of extracellular glutamate derived from neuronal brain activity is called for. Recent neurochemical and analytical investigations of glutamate have given rise to the concept that changes in extracellular glutamate levels are coupled, perhaps via a heteroexchange mechanism involving the glutamate transporter proteins, to fluctuations in extracellular ascorbate levels.11,26-31 Some authors, for example, have employed measurements of extracellular ascorbate levels as an index of glutamate release. This raises an important issue because ascorbate is an electroactive compound that can potentially interfere with the response of electrochemical sensors.9-11,25,32-35 Hence, in this work we also describe a new electrochemical microsensor for ascorbate. The ascorbate microsensor has been used to obtain independent, direct confirmation that signal fluctuations observed at glutamate microsensors are not related to ascorbate. EXPERIMENTAL SECTION Reagents. Glutamate oxidase from streptomyces (Gox), horseradish peroxidase type II (HRP), ascorbate oxidase from cucurbitas (AAox), l-glutamate, l-aspartate, l-ascorbate, l-glutamine, l-cysteine, dopamine, dihydroxyphenyl acetic acid (DOPAC), homovanillic acid (HVA), serotonin, and chloral hydrate were obtained from Sigma (St. Louis, MO). Quantities of enzymes mentioned below refer to the as-received solid forms, which were used without purification. Nafion (5% solution, 1100 equivalent weight), [4-(2-hydroxyethyl)-1-piprazineethane-sulfonic acid] (HEPES), and HEPES sodium salt were obtained from Aldrich (Milwaukee, WI). Poly(ethylene glycol 400 diglycidyl ether) (PEGDGE) was purchased from Polysciences (Warrington, PA). All the chemicals mentioned above were used as received without additional purification. Ultrapure water (Barnstead, Dubuque, IA) was used throughout. (21) Olson, C. R.; Kuhr, W. Anal. Chem. 1996, 68, 2164-2169. (22) Kuhr, W. G.; Barrett, V. L.; Gagnon, M. R.; Hopper, P.; Pantano, P. Anal. Chem. 1993, 65, 617-622. (23) Pantano, P.; Kuhr, W. G. Anal. Chem. 1993, 65, 623-630. (24) Cooper, J.; Foreman, P.; Glidle, A.; Ling, T.; Pritchard, D. J. Electroanal. Chem. 1995, 388, 143-149. (25) Ryan, M. R.; Lowry, J. P.; O’Neill, R. D. Analyst (Cambridge, U.K.) 1997, 122, 1419-1424. (26) O’Neill, R. D.; Fillenz, M.; Sundstrom, L.; Rawlins, N. P. Neurosci. Lett. 1984, 52, 227-233. (27) Miele, M.; Boutelle, M.; Fillenz, M. Neuroscience 1994, 62, 87-91. (28) O’Neill, R. D. In Neuromethods, Vol 27: Voltammetric methods in brain systems; Boulton, A., Baker, G., Adams, R., eds.; Humana Press: Totowa; 1995; pp 221-268. (29) Cammack, J.; Ghasemzadeh, B.; Adams, R. N. Neurochem. Res. 1992, 17, 23-27. (30) Fillenz, M.; Grunewald, R. A. J. Physiol. (London) 1983, 339, 40P-41P. (31) Walker, M. C.; Galley, P. R.; Errington, M. L.; Shorvon, S. D.; Jefferys, G. R. J. Neurochem. 1995, 65, 725-731.

5094

Analytical Chemistry, Vol. 71, No. 22, November 15, 1999

Scheme 1. Scheme for the Electrochemical Detection of Glutamate

Artificial cerebrospinal fluid (aCSF, 145 mM Na+, 1.2 mM Ca2+, 2.7 mM K+, 1.0 mM Mg2+, 152 mM Cl-, 2.0 mM phosphate, pH 7.4) was used for all electrochemical calibration procedures. Enzymes used for preparation of the microsensors were dissolved in a HEPES buffer, pH 8.0, prepared by the addition of the sodium salt to a 10-mM solution of the acid. The cross-linkable redox polymer used for the preparation of the glutamate microsensors was synthesized according to the procedure described in detail elsewhere.32 Electrochemical Microsensors and Techniques. Microcylinder electrodes were prepared by sealing individual 5-µm radius carbon fibers (P-55, Union Carbide, Danbury, CT) into pulled glass micropipets with a low-viscosity epoxy (Spurr, Polysciences, Warrington, PA). The fibers were trimmed with a scalpel to a length of 300-400 µm. Glutamate microsensors were prepared according to a modification of the procedure developed in this laboratory for the construction of choline microsensors.32-35 A cross-linkable redox polymer was used to immobilize Gox, HRP, and AAox onto carbon fiber microcylinder electrodes. Hydrogen peroxide is generated by the action of Gox and reduced by the action of HRP (Scheme 1). HRP is reduced by the polymer-bound redox complex, which in turn is reduced at the underlying electrode.32-37 Ascorbate interference is eliminated by operation of the microsensor at a low applied potential, by coimmobilization of AAox in the crosslinked film, and by the addition of an outer Nafion layer. The microsensors were prepared as follows. The carbon fiber was first coated with a 2-µL aliquot of an aqueous mixture containing redox polymer (1 mg/mL, 20 µL), PEGDGE (3 mg/ mL, 4 µL), HRP (3 mg/mL, 10 µL), Gox (2 mg/mL, 10 µL), and AAox (5 mg/mL, 10 µL). The microsensors were cured for 1 h at 37 °C, soaked in ultrapure water for 15 min, and dried in ambient air for 1 h. Next, the microsensors were dipped 7-10 times in a 0.5% solution of Nafion: each dip lasted for 5 s and was followed by a drying time of 20 s. The microsensors were stored in a refrigerator in a desiccator until use. Background microsensors were also prepared without glutamate oxidase and were used in vivo to obtain a glutamate-free background signal. Glutamate microsensors and background microsensors were operated in the amperometric mode at a constant potential of -100 mV vs a Ag/AgCl reference electrode. The potential was applied with a homemade potentiostat. The current signal was amplified (32) Shankar, L.; Garguilo, M. G.; Michael, A. C. In Methods in biotechnology, Vol. 6: Enzyme and microbial biosensors: techniques and protocols; Mulchandani A., Rogers K. R., eds.; Humana Press: Totowa, 1998; pp 121-132. (33) Garguilo, M. G.; Michael, A. C. J. Am..Chem. Soc. 1993, 115, 12218-12219. (34) Garguilo, M. G.; Michael, A. C. Anal. Chim. Acta 1995, 307, 291-299. (35) Garguilo, M. G.; Michael, A. C. J. Neurosci. Methods. 1996, 70, 73-82. (36) Ghobadi, S.; Csoregi, E.; Marko-Varga, G.; Gorton, L. Curr. Sep. 1996, 14, 94-102. (37) Belay, A.; Collins, A.; Ruzgas, T.; Kissinger, P. T.; Gorton, L.; Csoregi, E. J. Pharm. Biomed. Anal. 1999, 19, 93-105.

by a commercially available current amplifier (model 428, Keithley Instruments, Cleveland, OH) set to gain of 1010 V/A and a rise time of 300 ms. The amplifier output was digitized at a 20 kHz sampling rate with a 12-bit digital-to-analog converter (Labmaster PGH-DMA, Scientific Solutions, Solon, OH). The digitized points were box car averaged over 100 ms time intervals. Ascorbate microsensors were prepared by coating carbon fiber microcylinder electrodes with a 4-µL aliquot of an aqueous mixture containing redox polymer (1 mg/mL, 20 µL), PEGDGE (3 mg/ mL, 4 µL), and HRP (3 mg/mL, 10 µL). A Nafion layer was applied by dipping the microsensor into 0.5% Nafion solution 7-10 times with a 30-s drying time between dips. Ascorbate detection was carried out by slow-scan cyclic voltammetry. Nafion was used here to promote the biocompatibility of the ascorbate microsensors. Without Nafion, the microsensors were quickly passivated during in vivo experiments (unpublished observations). Between voltammetric scans, the microsensor was held at an applied potential of -100 mV vs Ag/AgCl. During individual scans, the potential was ramped to +200 mV vs Ag/AgCl at a rate of 500 mV/s, and the ascorbate signal was obtained by averaging the current that flowed between +100 and +200 mV. Scans were repeated at 2-s intervals. The signals observed during in vivo experiments were confirmed as being derived from ascorbate by examination of the background-subtracted cyclic voltammograms. All microsensors were calibrated both before and after in vivo experiments in a flow stream apparatus equipped with a loop-style sample injection valve (model 50/5701, Rheodyne, Cotati, CA). Flow of aCSF at ∼1 mL/min through the system was created by gravity feed from an elevated buffer reservoir. Enzyme-based microsensors were calibrated in air-equilibrated buffer. Ascorbate microsensors were calibrated in nitrogen-purged buffer. All calibrations were performed at 37 °C by placing the sample injector and electrochemical cell in an oven. Temperature in the electrochemical cell was monitored with a thermocouple probe positioned within a few millimeters of the microsensor. Animal and Surgical Procedures. In vivo experiments were carried out with approval of the Institutional Animal Care and Use Committee of the University of Pittsburgh. Male Sprague-Dawley rats, ∼250-350 g, anesthetized throughout the experiment with chloral hydrate, were placed in a stereotaxic frame (Kopf 1430 frame assembly, Tujunga, CA) with the incisor bar 5 mm above the intraural line.38 Body temperature was monitored and maintained at 37 °C with a homeothermic blanket (EKEG Electronics, Vancouver, BC). Small holes were drilled through the skull, and the dura was carefully removed to allow insertion of the microsensors and a micropipet unilaterally into the brain region called the striatum. Electrical contact between brain tissue and a Ag/AgCl reference electrode was established via a salt bridge fashioned from a plastic pipet tip plugged with tissue paper. Micropipets for local drug delivery by microinfusion were constructed with fused silica capillary tubes (60 cm in length, 27µm i.d., 360-µm o.d., Polymicro Technology Inc., Phoenix, AR). One end of the capillary was sealed with epoxy into the needle of a 50-µL gastight syringe (Hamilton, Reno, NV). The polymer coating was removed from the other end of the capillary with a match flame. The tip of the capillary was submerged to a depth (38) Pelligrino, L.; Pellegrino, A.; Cushman, A. A Stereotaxic Atlas of the Rat Brain, 2nd ed.; Plenum Press: New York, 1979.

Figure 1. The response of glutamate microsensors in the flow injection system with (A) and without (B) immobilized AAox and Nafion overlayer. The horizontal bars represent the time of injection.

of ∼4 mm in 48% HF for 80 min. While submerged, water was pumped though the capillary by means of a syringe driver (NA 1, Sutter Instruments, Novatto, CA) to prevent etching of the inner wall. This etching procedure caused the outer diameter of the capillary tip to decrease to ∼30-35 µm. Despite their small dimensions, micropipets prepared in this way could be reused for multiple in vivo experiments. During in vivo experiments, a micropipet filled with an infusion solution of interest was mounted vertically on a stereotaxic micromanipulator and lowered into the striatum at a point 2.4 mm anterior to bregma, 2.5 mm lateral from midline, and 4.5 mm below dura.38 Either one or two microsensors were mounted onto individual stereotaxic micromanipulators angled 10° from vertical and were implanted into the striatum in the same coronal plane as the micropipet and at a distance of ∼100 µm from the micropipet tip. All local microinfusions were performed in the striatum of the rat brain at 12.5 nL/s. Local microinfusion experiments were initiated about 1 h after implantation of the microsensors, by which time stable in vivo baseline signals were well-established. RESULTS AND DISCUSSION Performance Characteristics of Glutamate Microsensors Prior to In Vivo Experiments. Figure 1A shows the time course of the typical response of these new glutamate microsensors to the introduction of standard solutions containing 100-µM glutamate with and without 400-µM ascorbate to the flow stream calibration apparatus. This figure demonstrates the effectiveness of the steps taken in this work to eliminate electrochemical interference from Analytical Chemistry, Vol. 71, No. 22, November 15, 1999

5095

Table 1. Impact of Some Compounds that May Be Encountered in the Brain on Glutamate Microsensors test mixture

Figure 2. Calibration curves for the glutamate microsensor with (b) and without (2) 400 µM ascorbate. Each point is the mean of three observations, but error bars were omitted because the standard deviations are smaller than the symbols.

ascorbate, even at ascorbate concentrations that exceed those found in the extracellular space of brain tissue (200-400 µM).39-43 These glutamate microsensors do not detect physiologically relevant concentrations of ascorbate, and physiologically relevant concentrations of ascorbate have no effect on the glutamate signal. For the sake of comparison, Figure 1B shows the result of a similar experiment carried out with a microsensor lacking the coimmobilized ascorbate oxidase and the Nafion over layer. Without AAox and Nafion, the response of the microsensor to a 100-µM glutamate standard is almost eliminated in the presence of 400-µM ascorbate, due to the ability of ascorbate to reduce the redox mediator.33-35 Even without AAox and Nafion, however, the microsensor still does not detect ascorbate because of the low value of the applied potential. Figure 1 also shows that the response time of the glutamate microsensors is in the range of 20-40 s (defined as time required for the signal to increase from 10 to 90% of the steady-state signal). This response time is similar to that provided by microdialysis probes44,45 but is considerably slower than that reported for some other enzyme-modified electrodes.9-11,21-23,25 We have shown before that the response time of these redox-polymer-based sensors is dependent on the thickness of the polymer films deposited on the electrodes.32-35 Thinner films provide faster response times but at the expense of sensitivity. For this initial investigation of the in vivo behavior of these glutamate microsensors, we have not attempted any experiments that involve rapid fluctuations in the extracellular level of glutamate, and therefore, we have not yet optimized the response time of the microsensors. Figure 2 shows typical calibration curves obtained with the glutamate microsensors over a glutamate concentration range of 0-100 µM both in the presence and in the absence of 400 µM ascorbate. The calibration experiment was performed at 37 °C. Before exposure to brain tissue, the microsensors exhibited a glutamate sensitivity of 3.4 ( 0.94 pA/µM (mean ( s.d., n ) 6) (39) Ghasemzedah, B.; Cammack, J.; Adams, R. Brain Res. 1991, 547, 162166. (40) Schenk, J.; Miller, E.; Gaddis, R.; Adams, R. Brain Res. 1982, 253, 353356. (41) Stamford, J. A.; Kruk, Z. L.; Millar, J. Brain Res. 1984, 299, 289-295. (42) Grunwald, R. A. Brain Res. Rev. 1993, 18, 123-133. (43) Lada, M.; Kennedy, R. T. J. Neurosci. Methods 1995, 63, 147-152. (44) Lu, Y.; Peters, J. L.; Michael, A. C. J. Neurochem. 1998, 70, 584-593. (45) Newton, A. P.; Justice, J. B., Jr. Anal. Chem. 1994, 66, 1468-1472.

5096 Analytical Chemistry, Vol. 71, No. 22, November 15, 1999

100 µM glutamate 100 µM glutamate + 400 µM ascorbic acid 400 µM ascorbic acid 100 µM glutamate + 100 µM uric acid 100 µM uric acid 100 µM glutamate + 20 µM DOPAC 20 µM DOPAC 100 µM glutamate + 5 µM dopamine 5 µM dopamine 100 µM glutamate + 1 µM serotonin 1 µM serotonin 100 µM glutamate + 4 µM norepinophrine 4 µM norepinophrine 100 µM glutamate + 20 µM homovanillic acid 20 µM homovanillic acid 100 µM glutamate + 100 µM aspartate 100 µM aspartate 100 µM glutamate + 100 µM glutamine 100 µM glutamine 100 µM glutamate + 4 µM cysteine 4 µM cysteineb 100 µM glutamate + 100 µM TTX 100 µM TTX

response relative to 100 µM glutamatea 1 0.92 ( 0.043 0 1 0 0.96 ( 0.026 0 1 0 1 0 1 0 1 0 1 0 0.96 ( 0.02 0.04 ( 0.01 0.97 ( 0.02 -0.10 ( 0.01 1 0

a Values listed are the mean ( s.d. of at least 3 observations. Where 0 or 1 is listed, the test compound had no discernible effect on the microsensor response. b High concentrations of cysteine cause a reversible decrease in the sensitivity to glutamate.

and a detection limit of 1 ( 3 µM. Here, the detection limit is reported as the glutamate concentration that corresponds to a signal three times larger than the noise. The lowest concentration of glutamate routinely tested during calibration was 10 µM. The sensitivity of these microsensors toward glutamate, however, did not significantly change when 400 µM ascorbate was included in the glutamate standard solutions (2.9 ( 0.89 pA/µM, mean ( s.d., n ) 6). Ascorbate is the primary potential source of electrochemical interference during experiments in brain tissue because it is a powerful reducing agent and is present in high extracellular concentration.26-29,31,39-43 Other electroactive compounds are also present, albeit at lower levels than ascorbate. Table 1 reports the response of the glutamate microsensors to the several other compounds present in the brain and the interference by those compounds of the response to a 100 µM glutamate standard. Because they are present in lower concentration and because they are less powerful reducing agents, the majority of compounds tested could not be detected by, and did not interfere with, the glutamate microsensors. Two of the compounds tested, aspartate and glutamine, have been reported to serve as substrates for glutamate oxidase.9,46,47 Even so, aspartate exhibited no activity at these microsensors. In the absence of glutamate, the microsensors exhibited a slight sensitivity to a 100 µM glutamine standard (extracellular concentration ∼100 µM),47 although this concentration of glutamine could not be detected in the presence of 100 (46) Kusakabe, H.; Midorikawa, Y.; Fujishima, T.; Kuninaka, A.; Yoshimo, H. Agric. Biol. Chem. 1983, 47, 1323-1328. (47) Espey, M. G.; Kustova, Y.; Sei, Y.; Basile, A. S. J. Neurochem. 1998, 71, 2079-2087.

Figure 3. Signals recorded in vivo with glutamate (Glu) and background (Bkg) microsensors during the microinfusion of (A) aCSF (200 nL), (B) glutamate (200 nL, 10 mM), and (C) Gox (800 nL, 2 mg/mL) in the striatum of three different rats. The horizontal bars represent the duration of the microinfusion. The vertical scale bars were obtained by postcalibration of the glutamate microsensors following their removal from the brain.

µM glutamate. In fact, the glutamate signal was slightly decreased in the presence of glutamine (Table 1), suggesting the latter may inhibit the enzymatic oxidation of glutamate. Exposure of the microsensors to high micromolar cysteine caused a slight reversible fouling of the electrode; the extracellular concentration of cysteine in the brain, however, is reported to be in the nanomolar to low micromolar range.48 These lower concentrations did not affect the microsensors. The In Vivo Response of Glutamate and Background Microsensors During Local Microinfusion Experiments. Figure 3 shows typical responses recorded simultaneously with a glutamate microsensor and a background microsensor implanted side by side in the striatum of an anesthetized rat during the local infusion of aCSF (200 nL), exogenous glutamate (10 mM, 200 nL), and glutamate oxidase (2 mg/mL, 800 nL). Neither type of microsensor responded significantly to microinfusions of aCSF performed over a range of volumes (200-800 nL). Only the signal at the glutamate microsensor responded to the local microinfusion of glutamate or glutamate oxidase. The signal at the background microsensors did not respond to any of these microinfusions. The various responses reported in Figure 3 are representative of at least three infusions repeated in at least three different animals with different microsensors. All concentration scale bars were obtained by postcalibration performed following removal of the microsensors from the brain. The in vivo baseline signals at the glutamate microsensors were consistently higher than those at the background microsensors. This can be seen in all three panels in Figure 3. On the basis of sensitivity data collected during postcalibration of the microsensors after their removal from the brain, the consistent difference in the baseline signals was found to correspond with a resting glutamate concentration of 29 ( 9.0 µM (mean ( s.d. of observations in 26 rats). This value for the basal glutamate concentration is noticeably higher than values previously reported by others using microdialysis (1-4 µM).14-20 The higher glutamate levels found with the microsensors are attributed to the small dimensions of the microsensors as compared with microdialysis probes. We have demonstrated elsewhere that implantation of the larger microdialysis probes inflicts trauma on the brain tissue that (48) Landolt, H.; Lutz, H.; Langemann, H.; Stauble, D.; Mendelowitsch, A.; Gratzl, O.; Honegger, C. G. J. Cereb. Blood Flow Metab. 1992, 12, 96-102.

causes a decrease in neuronal activity in the immediate vicinity of the probes themselves.44,49,50 Implantation of microsensors, which are ∼10 000 times smaller than the microdialysis probes, causes less trauma of the adjacent tissue. Hence, the higher level of glutamate reported here is attributed to the greater neuronal activity in the vicinity of the microsensors (see also Figure 9, below). Figure 3B shows that upon microinfusion of exogenous glutamate the signal at the glutamate microsensor increased rapidly. After reaching a peak, the signal returned to baseline in about 1 min. The rapid decrease in the signal compared with that observed following microinfusion of choline33-35 or ascorbate (see Figure 5, below) reflects the actions of the glutamate uptake mechanism that provides for the clearance of glutamate from the extracellular space.51 Typically, three glutamate infusions were performed in each rat, and the responses between infusions were reproducible, with less than a 20% variation in the amplitude. During similar experiments in five rats with five different pairs of microsensors the maximum change in the signal from the baseline value corresponded to a glutamate concentration of 73 ( 50 µM (mean, ( s.d., n ) 5) according to the postcalibration sensitivities of the individual microsensors: the standard deviation mainly reflects variability between experiments. The reproducibility of the response during consecutive glutamate microinfusions illustrates the stability of the glutamate microsensors under in vivo experimental conditions. The decrease in the signal at the glutamate microsensors after the microinfusion of glutamate oxidase (Figure 3C) confirms that the resting signal was due to glutamate present in the extracellular space. Typically, microinfusion of glutamate oxidase caused a decrease in the signal that corresponded to ∼50% of the resting signal, although, as in Figure 3C, a larger decrease was occasionally observed. Glutamate oxidase in the presence of oxygen breaks down glutamate with the formation of hydrogen peroxide. Nevertheless, no response was observed at the background microsensors, which are highly sensitive to peroxide. This result suggests that any peroxide formed in the tissue during these experiments is rapidly destroyed. Evaluation of Glutamate Microsensors after In Vivo Experiments. During in vivo experiments, the sensitivity of the microsensors toward glutamate typically decreased by 50-65% to 1.1 ( 0.25 pA/µM (mean ( s.d, n ) 6). This is often observed with implanted microelectrodes52 and is attributed to fouling of the sensor surface, which causes a decrease in the permeability of the polymer layers of the microsensor. The decrease in sensitivity of the microsensors appeared to be essentially complete approximately 1 h after implantation into brain tissue. The baseline signals routinely required about 1 h following implantation to stabilize, and microsensors removed from the brain after 2 h exhibited similar postcalibration sensitivity to microsensors removed from the brain after 7 h. Upon postcalibration, the microsensors were again found to be immune to interference by physiological concentrations of ascorbate. In the presence of 400 µM ascorbate the sensitivity toward glutamate decreased by ∼10% (49) Peters, J. L.; Michael, A. C. J. Neurochem. 1998, 70, 594-603. (50) Yang, H.; Peters, J. L.; Michael, A. C. J. Neurochem. 1998, 71, 684-692. (51) Kanai, Y.; Smith, G. P.; Hediger, M. A. FASEB J. 1994, 8, 1450-1459. (52) Stamford, J. A. J. Neurosci. Methods 1986, 17, 1-29.

Analytical Chemistry, Vol. 71, No. 22, November 15, 1999

5097

to 0.92 ( 0.19 pA/µM (not significantly different than in the absence of ascorbate). At a concentration of 200 µM, ascorbate had no effect on the sensitivity toward glutamate.

detection limit for ascorbate was routinely less than 2 µM. These microsensors were highly selective for ascorbate over dopamine and HVA but their sensitivity toward ascorbate was only approximately 2-3 times larger than their sensitivity toward DOPAC, which is present in striatum at much lower levels (5-20 µM)53 than ascorbate (200-400 µM).39-43 Hence, it is possible that signals recorded in vivo with these microsensors might contain a small contribution from DOPAC. The low sensitivity toward dopamine, a cationic neurotransmitter, can be attributed to the perm-selectivity of the redox polymer gel, which is also cationic.

Figure 4. Cyclic voltammograms recorded at ascorbate microsensors: (A) in aCSF buffer and in 200 µM ascorbate and (B) in aCSF buffer and in rat brain. The large voltammetric wave is due to the Os complex of the cross-linked redox polymer while the shoulder in the 100-200 mV range is the ascorbate response.

Figure 6. Signals recorded in vivo with ascorbate microsensors during the local microinfusion of (A) aCSF (200 nL), (B) ascorbate (200 nL, 10 mM), and (C), AAox (800 nL, 2 mg/mL) in the striatum of three different rats. The horizontal bars indicate the duration of the microinfusion. The vertical scale bars were obtained by postcalibration of the microsensors following their removal from the brain.

Figure 5. Pre- and postcalibration of an ascorbate microsensor in 500 µM ascorbate in the flow injection system. The horizontal bar represents the injection time. The insert: The background-subtracted cyclic voltammogram of ascorbate during pre- and postcalibration.

Evaluation of Ascorbate Microsensors. Figure 4 shows typical cyclic voltammograms recorded at ascorbate microsensors in both aCSF buffer and 200 µM ascorbate (A) and in aCSF buffer and rat brain (B) at a 0.1 V/s scan rate. In both cases, a voltammetric wave for an irreversible oxidative process can be observed in the 100-200 mV potential range. Because of the similarity between the wave observed in ascorbate solutions and that in the brain, we supposed that the wave observed in the brain might be due to brain ascorbate, and therefore, we examined this wave in some detail. Figure 5 shows the response at an ascorbate microsensor to a 500 µM ascorbate standard during pre- and postcalibration in aCSF buffer. The symbols in the main part of the figure were calculated by averaging the oxidation current between 100 and 200 mV on the positive potential sweep of consecutive cyclic voltammograms. The inset shows the features of the background-subtracted voltammograms. Even though the ascorbate microsensor lost some sensitivity during the in vivo experiment, the shapes of the background-subtracted voltammograms remained unchanged and, therefore, provide a valuable voltammetric fingerprint for the identification of ascorbate during in vivo experiments. Ascorbate microsensors exhibited a pre-in vivo sensitivity toward ascorbate of 20 ( 8.6 pA/µM and a post-in vivo sensitivity of 10 ( 5.4 pA/µM (mean ( s.d., n ) 7 in both cases). The 5098 Analytical Chemistry, Vol. 71, No. 22, November 15, 1999

The In Vivo Response of Ascorbate Microsensors during Local Microinfusion Experiments. Figure 6 shows examples of the signals recorded in vivo with ascorbate microsensors during local infusion of aCSF (200 nL), ascorbate (10 mM, 200 nL), and ascorbate oxidase (2 mg/mL, 800 nL). Figure 6 clearly shows that the ascorbate signal remained stable during the microinfusion of aCSF, increased during the microinfusion of exogenous ascorbate, and decreased during the microinfusion of the degradative enzyme. The oxidation current at the ascorbate microsensors rapidly reached a maximum upon microinfusion of exogenous ascorbate but returned slowly to baseline over the next 20 min (Figure 6B). This prolonged decay reflects the time course of the diffusion of ascorbate from the microinfusion site. The average increase in current following ascorbate microinfusion corresponded to 740 ( 80 µM (mean ( s.d., n ) 4) ascorbate according to postcalibration of the microsensors following their removal from the brain. Three such microinfusions were performed in each rat with variability between response amplitudes of less than 10%. The reproducibility of the response to consecutive microinfusions demonstrates the stability of the sensitivity of the ascorbate microsensors under in vivo conditions. Figure 7 shows the cyclic voltammograms recorded at the ascorbate microsensors before and after the microinfusion of 10 mM ascorbate (A) and 2 mg/ mL AAO (B). The difference between the two voltammograms (insets in Figure 7), which represents the background-subtracted voltammogram of ascorbate under in vivo conditions, is identical to those obtained during calibration of ascorbate microsensor in aCSF buffer (Figure 5). (53) Gonon, F.; Buda, M.; Cespuglio, R.; Jouvet, M.; Pujol, J. F. Brain Res. 1981, 223, 69-80.

Figure 7. Cyclic voltammograms recorded in vivo at ascorbate microsensors before and after the local microinfusion of ascorbate (200 nL, 10 mM) (A) and before and after the local microinfusion of AAox (800 nL, 2 mg/mL) (B) in the striatum of rat brain. Inset panels: The background-subtracted cyclic voltammograms after the microinfusion of ascorbate (A) and AAox (B), i.e., the difference between the voltammograms in the main panels.

Ascorbate is oxidized in the presence of ascorbate oxidase and oxygen. Infusion of ascorbate oxidase has been used before to confirm both the in vivo and in vitro voltammetric detection and estimation of basal levels of ascorbate.39,41,54-56 On the basis of sensitivity data collected after the in vivo experiments, the decrease in signal following the first ascorbate oxidase microinfusion (Figure 6C) corresponded to 240 ( 25 µM (mean ( s.d., n ) 4). This is very similar to other estimates of the extracellular ascorbate level in the brain of anesthetized rats.43 Subsequent microinfusions of ascorbate oxidase at 40-min intervals produced only small responses, suggesting that the first microinfusion was sufficient to completely eliminate extracellular ascorbate.

of a few percent of the baseline level, and is not substantially different from the response to aCSF microinfusion. The reason for examining the response of ascorbate microsensors during the microinfusion of glutamate stems from previous reports that suggest a coupling between extracellular glutamate and ascorbate levels.26-31,42 The apparent coupling, which has been attributed to a heteroexchange mechanism at the glutamate uptake site, has led to the suggestion that ascorbate might provide a useful index of changes in extracellular glutamate levels. The heteroexchange mechanism, which is hypothesized to involve the release of ascorbate as glutamate is taken up by neurons and glia, suggests that an increase in ascorbate should be observed following the microinfusion of glutamate. Figure 8, however, does not reveal evidence for such a process. From this result it appears that monitoring extracellular ascorbate may not be a reliable approach to obtaining an index of extracellular glutamate levels, since Figure 8 demonstrates that under some conditions the expected coupling between extracellular ascorbate and glutamate levels is not evident.

Figure 9. Signals recorded in vivo with glutamate (glu) and background (bkg) microsensors (A) and with ascorbate microsensors (B) during the local microinfusion of TTX (200 nL, 100 µM). The horizontal bars represent the duration of the microinfusion. The vertical scale bars were obtained by postcalibration of the microsensors following their removal from the brain. Inset: The backgroundsubtracted cyclic voltammogram obtained at the ascorbate microsensor, i.e., the difference between voltammograms obtained after and before microinfusion of TTX.

Figure 8. Signals recorded in vivo with ascorbate microsensors during the local microinfusion of 200 nL of 10 mM glutamate (main panel) and 200 nL of aCSF (inset) into rat striatum. The horizontal bars represent the duration of the microinfusion. The vertical scale bars were obtained by postcalibration of the microsensors following their removal from the brain.

The In Vivo Response of Ascorbate Microsensors during the Microinfusion of Glutamate. Figure 8 shows an example of the signals observed at ascorbate microsensors during the local microinfusion of glutamate (10 mM, 200 nL) and aCSF (200 nL). Similar results were obtained in three rats. The response to a glutamate infusion is very slight, representing a transient decrease (54) Schenk, J. O.; Miller, E.; Adams, R. N. Anal. Chem. 1982, 54, 1452-1454. (55) Schenk, J. O.; Miller, E.; Rice, M. E.; Adams, R. N. Brain Res. 1983, 277, 1-8. (56) Brazel, M. P.; Marsden, C. A. Brain Res. 1982, 249, 167-172.

The Dependence of the Resting Glutamate and Ascorbate Signals on Neuronal Brain Activity. Figure 9A shows the signals recorded simultaneously in vivo at a glutamate and background microsensor during the microinfusion of TTX (100 µM, 200 nL). Figure 9B shows the signal recorded in a different animal with an ascorbate microsensor during a similar microinfusion of TTX. The microinfusion of TTX caused a substantial change in the resting signal observed at the glutamate microsensors but caused no response at the background microsensors. The signals at the glutamate microsensor decreased by an amount that corresponded to 25-85% of the original baseline signal (n ) 9). This is attributed to a decrease in the extracellular glutamate level upon the TTX-induced cessation of neuronal activity. As reported in Table 1, TTX itself is not detected by the glutamate microsensors and does not interfere with their response to glutamate. We have demonstrated, above, that physiologically relevant concentrations of ascorbate, i.e., concentrations as high as 400 µM, cause negligible interference with the glutamate microsensors. Nevertheless, higher concentrations could cause interference Analytical Chemistry, Vol. 71, No. 22, November 15, 1999

5099

and decrease the response to glutamate. It was necessary, therefore, to confirm that the decrease in signal observed in Figure 9A was not a consequence of a massive increase in extracellular ascorbate induced by the microinfusion of TTX (although no evidence for such an effect of TTX exists). In fact, Figure 9B shows that the microinfusion of TTX induced a slight decrease in the signal at the ascorbate microsensors. Although the exact reason for this small decrease is unclear since ascorbate is not generally viewed as a classical neurotransmitter, the result confirms that the TTX-induced decrease in the resting signal at the glutamate microsensor is not attributable to an increase in extracellular ascorbate during these experiments. Hence, Figure 9 provides the first direct evidence that glutamate microsensors are useful for probing the contribution to the resting basal extracellular glutamate level in the brain that is derived from the neuronal, as opposed to metabolic, activity of living brain tissue. CONCLUSION Over the past several years, efforts to measure and monitor neurotransmitter levels in the extracellular space of living brain tissue have relied mainly on microdialysis14-20 and electrochemical microelectrodes44,48,49 or microsensors.9,11,33-35 The quantitative interpretation of the results has been a contentious issue because of the lack of suitable methods for the in vivo calibration of the analytical devices that are used. This problematic issue has been exacerbated by the tendency for the techniques to be utilized in isolation of one another so that, in many cases, either microdialysis or electrochemical results are available, but not both. Recently, we used microdialysis probes and carbon fiber microelectrodes simultaneously to monitor dopamine in the extracellular space of living brain tissue.44,49,50 Specifically, we compared the responses at carbon fiber electrodes positioned about 1 mm away from a microdialysis probe to responses observed at electrodes positioned immediately adjacent to the microdialysis probe. On the basis of

5100

Analytical Chemistry, Vol. 71, No. 22, November 15, 1999

that comparison we demonstrated that (A) the brain tissue immediately adjacent to a microdialysis probe is virtually devoid of neuronal activity, presumably as a consequence of probeinduced trauma, and (B) that the dopamine uptake mechanism decreases the probe recovery value, rather than increases it as had been previously believed. These observations lead to the conclusion that microdialysate levels of dopamine underestimate the levels of dopamine in the extracellular space of the brain tissue. The effects of probe-induced tissue trauma and the existence of neurotransmitter uptake mechanisms do not uniquely impact investigations of the dopaminergic systems of the brain. Glutamate terminals in the vicinity of microdialysis probes are unlikely to be spared from probe-induced trauma, and an uptake mechanism that removes glutamate from the extracellular space also exists.51 In analogy with the conclusions reached about dopamine, therefore, we expected that microdialysate glutamate levels would underestimate extracellular levels of glutamate. The results of this study appear to confirm that expectation, since the resting signal at the glutamate microsensors corresponds to a glutamate concentration that is 10-100 times higher than values reported on the basis of various microdialysis results. Furthermore, the resting signal at the glutamate microsensors was found to be dependent on the neuronal activity of the brain tissue, which has generally not been the case for microdialysis-based measures of resting glutamate levels. Hence, the results reported here confirm that the glutamate microsensors, by virtue of their small dimensions, inflict less trauma on the brain tissue and thereby gain greater access to neuronally active brain tissue. ACKNOWLEDGMENT This work was financially supported by NIH (Grant NS 31442). Received for review June 15, 1999. Accepted September 7, 1999. AC990636C