Microelectrodes for the Measurement of Catecholamines in Biological

with an external fourth-order Butterworth filter (Krohn-Hite 3750, Avon, MA). ... elsewhere.40 Briefly, male Sprague-Dawley rats (Charles River In...
0 downloads 0 Views 215KB Size
Download PDF
Anal. Chem. 1996, 68, 3180-3186

Microelectrodes for the Measurement of Catecholamines in Biological Systems Paula S. Cahill, Q. David Walker, Jennifer M. Finnegan, George E. Mickelson, Eric R. Travis, and R. Mark Wightman*

Department of Chemistry, University of North Carolina, Chapel Hill, North Carolina 27599-3290

Many of the molecules involved in biological signaling processes are easily oxidized and have been monitored by electrochemical methods. Temporal response, spatial considerations, and sensitivity of the electrodes must be optimized for the specific biological application. To monitor exocytosis from single cells in culture, constant potential amperometry offers the best temporal resolution, and a low-noise picoammeter improves the detection limits. Smaller electrodes, with 1-µm diameters, provided spatial resolution sufficient to identify the locations of release sites on the surface of single cells. For the study of neurotransmitter release in vivo, larger cylindrical microelectrodes are advantageous because the secreted molecules come from multiple terminals near the electrode, and the greater amounts lead to a larger signal that emerges from the Johnson noise of the current amplifier. With this approach, dopamine release elicited by two electrical stimulus pulses at 10 Hz was detected with fastscan cyclic voltammetry in vivo. Nafion-coated elliptical electrodes have previously been shown to be incapable of detecting such concentration changes without extensive signal averaging. In addition, we demonstrate that highpass filtering (200 Hz) of cyclic voltammograms recorded at 300 V/s decreases the background current and digitization noise at these microelectrodes, leading to an improved signal. Also, high-pass filtering discriminated against ascorbic acid, DOPAC, and acidic pH changes, three common interferences in vivo. Many chemical messengers that relay information between biological cells can be electrochemically oxidized. These molecules include catecholamines,1-3 histamine,4 5-hydroxytryptamine,5-7 and insulin.8 For this reason, microsensors designed from carbon fibers have been shown to be quite useful to monitor secretion of these species in a variety of biological systems.1-8 As progress is made in these measurements, it is clear that the optimum type of electrode and electroanalytical technique depend (1) Wightman, R. M.; Schroeder, T. J.; Finnegan, J. M.; Ciolkowski, E. L.; Pihel, K. Biophys. J. 1995, 68, 383-390. (2) Garris, P. A.; Wightman, R. M. Synapse 1995, 20, 269-279. (3) Dugast, C.; Suaud-Chagny, M. F.; Gonon, F. Neuroscience 1994, 62, 647654. (4) Pihel, K.; Hsieh, S.; Jorgenson, J. W.; Wightman, R. M. Anal. Chem. 1995, 67, 4514-4521. (5) Jackson, B. P.; Wightman, R. M. Brain Res. 1995, 674, 163-166. (6) Jackson, B. P.; Dietz, S. M.; Wightman, R. M. Anal. Chem. 1995, 67, 11151120. (7) Bruns, D.; Jahn, R. Nature 1995, 377, 62-65. (8) Kennedy, R. T.; Huang, L.; Atkinson, M. A.; Dush, P. Anal. Chem. 1993, 65, 1882-1887.

3180 Analytical Chemistry, Vol. 68, No. 18, September 15, 1996

upon the particular biological preparation investigated. Four major issues in electrochemical measurements are electrode response time and electrode size, as well as sensitivity and selectivity of the method. These issues are often intertwined. For example, early studies showed that differential pulse voltammetry at electrochemically modified electrodes had high selectivity but could not resolve the rapid concentration changes that neurotransmitters undergo.9-12 This was due to the slow response time of the electrodes as well as the slow scan rate of the technique. The two techniques most widely used today are amperometry3,14-18 and fast-scan cyclic voltammetry.5,19-22 Amperometry allows measurement of the most rapid concentration changes and is theoretically limited by the double-layer capacitance of the electrode and the resistance of the surrounding solution, but it provides no chemical identification of the species detected.3,14-16 Nevertheless, amperometry has provided unique information on the kinetics of secretion and subsequent biochemical fates of the easily oxidized chemical messengers in situations where the biological environment is well controlled. Such situations arise in cell cultures containing a single cell type,1,18,23,24 or in the brain when specific nerve terminals are activated.3,14 Fastscan cyclic voltammetry is particularly useful because the voltammogram provides information to identify the detected substance.25,26 However, cyclic voltammetry’s response time, selectivity, and sensitivity critically depend on the electrode’s surface state.6,27 (9) Gonon, F. G.; Buda, M. J.; Cespuglio, R.; Jouvet, M.; Pujol, J. F. Nature 1980, 286, 902-904. (10) Gonon, F. G.; Fombarlet, C. M.; Buda, M. J.; Pujol, J. F. Anal. Chem. 1981, 53, 1386-1389. (11) Gonon, F.; Buda, M.; Cespuglio, R.; Jouvet, M.; Pujol, J. F. Brain Res. 1981, 223, 69-80. (12) Sharp, T.; Maidment, N. T.; Brazell, M. P.; Zetterstrom, T.; Ungerstedt, U.; Bennett, G. W.; Marsden, C. A. Neuroscience 1984, 12, 1213-1221. (13) Feng, J. X.; Brazell, M.; Renner, K.; Kasser, R.; Adams, R. N. Anal. Chem. 1987, 59, 1863-1867. (14) Kawagoe, K. T.; Wightman, R. M. Talanta 1994, 41, 865-874. (15) Chergui, K.; Suaud-Chagny, M. F.; Gonon, F. Neuroscience 1994, 62, 641645. (16) Suaud-Chagny, M. F.; Dugast, C.; Chergui, K.; Msghina, M.; Gonon, F. J. Neurochem. 1995, 65, 2603-2611. (17) Cass, W. A.; Gerhardt, G. A. J. Neurochem. 1995, 65, 201-207. (18) Leszczyszyn, D. J.; Jankowski, J. A.; Viveros, O. H.; Diliberto, E. J., Jr.; Near, J. A.; Wightman, R. M. J. Neurochem. 1991, 56, 1855-1862. (19) Stamford, J. A. J. Neurosci. Methods 1990, 34, 67-72. (20) Garris, P. A.; Wightman, R. M. J. Neurosci. 1994, 14, 442-450. (21) Jones, S. R.; Garris, P. A.; Kilts, C. D.; Wightman, R. M. J. Neurochem. 1995, 64, 2581-2589. (22) Williams, J. E. G.; Wieczorek, W.; Willner, P.; Kruk, Z. L. Brain Res. 1995, 678, 225-232. (23) Wightman, R. M.; Jankowski, J. A.; Kennedy, R. T.; Kawagoe, K. T.; Schroeder, T. J.; Leszczyszyn, D. J.; Near, J. A.; Diliberto, E. J., Jr.; Viveros, O. H. Proc. Natl. Acad. Sci. U.S.A. 1991, 88, 10754-10758. (24) Chow, R. H.; von Ruden, L.; Neher, E. Nature 1992, 356, 60-63. (25) Stamford, J. A.; Kruk, Z. L. Neurosci. Lett. 1984, 51, 133-138. S0003-2700(96)00347-2 CCC: $12.00

© 1996 American Chemical Society

The optimum size for a microelectrode is largely dependent upon the size of the biological source to be studied. For the study of single cultured cells, the catecholamine source is membranebound vesicles inside the cell that have radii on the order of 100 nm.23,28 When an amperometric microelectrode was positioned so that it touched the cell surface, the highest sensitivity was achieved when the size of the microelectrode approximated the size of a single vesicle.29 In this case, any extraneous electrode area acted as a noise source.29 However, the spacing between the source and the electrode must also be minimized to prevent diffusional dispersion.1,30 Indeed, with this approach, a single secreted packet of a few thousand molecules of 5-hydroxytryptamine has been measured.7 In contrast, a voltammetric microelectrode in the rat brain monitors the concentration of catecholamine in the restricted volume of the extracellular space. The electrode does not directly measure the intrasynaptic concentration because the dimensions of a synapse (300 nm × 15 nm)31,32 are much smaller than the size of a microelectrode. Unlike the situation at single cells, the nerve terminals, i.e., the source of the chemical messenger, can be as dense as 1 × 108 terminals/mm3.33 Under these conditions, the diffusion fields of released neurotransmitters overlap within a few milliseconds.33,34 This time scale is more rapid than the repetition rate in cyclic voltammetry (1 scan every 100 ms). Therefore, it can be advantageous to use larger electrodes for in vivo monitoring. In this paper, we examine the issues of electrode size, sensitivity, selectivity, and temporal response in an attempt to optimize conditions for specific experiments. EXPERIMENTAL SECTION Apparatus. For amperometry, an EI-400 potentiostat (Ensman Instruments, Bloomington, IN), an AI-403 picoammeter (Axon Instruments, Foster City, CA), and an Axopatch 200B patch clamp (Axon Instruments) were used. For both the EI-400 potentiostat and the AI-403 picoammeter, the working electrode was held at virtual ground, while the potential (+0.65 V) was applied to the reference electrode (Ag/AgCl). The EI-400 was used in the twoelectrode configuration. The potential source for the AI-403 picoammeter was a 9-V battery. The Axopatch 200B, with a CV203BU headstage (Axon Instruments), employed resistive feedback in the whole cell configuration, and Johnson noise was reduced by Peltier cooling of the active elements in the amplifier to -15 °C. Using the voltage clamp mode with the Axopatch 200B, a potential of +0.65 V vs sodium-saturated calomel reference electrode (SSCE) was applied to the working electrode. The EI400 and Axopatch 200B currents were low-pass filtered with an external fourth-order Butterworth filter (Krohn-Hite 3750, Avon, (26) Garris, P. A.; Wightman, R. M. J. Physiol. 1994, 478, 239-249. (27) Pihel, K.; Schroeder, T. J.; Wightman, R. M. Anal. Chem. 1994, 66, 45324537. (28) Coupland, R. E. Nature 1968, 217, 384-388. (29) Schroeder, T. J.; Jankowski, J. A.; Senyshyn, J.; Holz, R. W.; Wightman, R. M. J. Biol. Chem. 1994, 269, 17215-17220. (30) Kawagoe, K. T.; Jankowski, J. A.; Wightman, R. M. Anal. Chem. 1991, 63, 1589-1594. (31) Groves, P. M.; Linder, J. C.; Young, S. J. Neuroscience 1994, 58, 593-604. (32) Pickel, V. M.; Beckley, S. C.; Joh, T. H.; Reis, D. J. Brain Res. 1981, 225, 373-385. (33) Garris, P. A.; Ciolkowski, E. L.; Pastore, P.; Wightman, R. M. J. Neurosci. 1994, 14, 6084-6093. (34) Kawagoe, K. T.; Garris, P. A.; Wiedemann, D. J.; Wightman, R. M. Neuroscience 1992, 51, 55-64.

MA). The AI-403 signal was low-pass filtered with an external eighth-order Bessel filter (Cyberamp 320, Axon Instruments). An EI-400 potentiostat with an internal triangle generator was employed for cyclic voltammetry with a scan rate of 300 V/s. The voltammetric signal was low-pass filtered at 2 kHz using the internal filter of the potentiostat.35 The applied potential consisted of a -0.5- to +1.0-V triangle wave, which was repeated at 100-ms intervals. For high-pass filtered cyclic voltammetry, a Butterworth filter (Krohn-Hite 3750) was used with a high-pass cut-off frequency of 200 Hz and a 24 dB/octave roll-off. Backgroundsubtracted voltammograms were collected through an analog-todigital converter (DMA Labmaster, Scientific Solutions, Solon, OH) using a commercial program (CV6.EXE, Ensman Instruments) or a program created in-house (VA.EXE). Electrode Construction. Cylindrical microelectrodes were constructed from T300 carbon fibers (3.5-µm radii) without sizing (Amoco, Greenville, SC). A single carbon fiber was aspirated into a glass capillary. The electrode was pulled to micrometer dimensions on a pipet puller (Narashige, Tokyo, Japan). The fiber was then cut to the desired length, ∼10-150 µm, with a scalpel under a microscope. The electrodes were sealed with epoxy (Epon 828 with 14% m-phenylenediamine by weight, MillerStephenson Chemical Co., Inc., Danbury, CT), and excess epoxy was removed from the carbon surface by rinsing immediately with acetone. The electrodes were cured at room temperature overnight, at 100 °C for 2 h, and at 150 °C for 2 days. Electrical connection was made by back-filling with colloidal graphite (Polysciences, Inc., Warrington, PA) or mercury and inserting a wire. The electrodes were soaked in 2-propanol for at least 10 min prior to use. Elliptical microelectrodes were prepared from single 5.0-µmradius P55 carbon fibers (Amoco) as previously described.36 After the epoxy was cured, each electrode was beveled on a polishing wheel (Sutter Instruments, Novoto, CA) at an angle of 25° for in vivo experiments or 45° for cell experiments to yield an elliptical active area. The electrodes were soaked in 2-propanol for 10 min before use. Some of the electrodes were coated with Nafion as previously described.36 Briefly, after polishing, the electrodes were dipped into a 2.5% Nafion solution, dried with a heat gun for 10 min, and stored overnight at room temperature before use. Single Cell Experiments. Primary cultures of bovine adrenal chromaffin cells were prepared by enzymatic digestion of bovine adrenal medulla.18,37 A renografin density gradient was used to separate the epinephrine and norepinephrine cell fractions, and the cells were plated at a density of 6 × 105/35-mm-diameter culture plate as previously described.18,38 For secretion experiments, the culture medium was replaced with Krebs-Ringer buffer, which contained 145 mM NaCl, 5 mM KCl, 1.3 mM MgCl2, 1.2 mM NaH2PO4, 10 mM glucose, 20 mM HEPES, and 2 mM CaCl2, with the pH adjusted to 7.4. When the cells were stimulated with 60 mM KCl, the concentration of NaCl was decreased to maintain the osmotic strength. The cell culture plate was placed on the stage of an inverted microscope (Axiovert 35, Zeiss, Thornwood, NY). The micro(35) Wiedemann, D. J.; Kawagoe, K. T.; Kennedy, R. T.; Ciolkowski, E. L.; Wightman, R. M. Anal. Chem. 1991, 63, 2965-2970. (36) Kawagoe, K. T.; Zimmerman, J. B.; Wightman, R. M. J. Neurosci. Methods 1993, 48, 225-240. (37) Wilson, S. P.; Viveros, O. H. Exp. Cell Res. 1981, 133, 159-169. (38) Moro, M. A.; Lopez, M. G.; Michelena, P.; Garcia, A. G. Anal. Biochem. 1990, 185, 243-248.

Analytical Chemistry, Vol. 68, No. 18, September 15, 1996

3181

electrode was positioned so that it was touching the surface of a single cell using a piezoelectric positioner (PCS-250, Burleigh Instruments, Fishers, NY). The cell was stimulated by pressure ejection (Picospritzer, General Valve Corp., Fairfield, NJ) of 60 mM K+ through a glass micropipet located ∼30 µm from the cell. Cell currents were digitized and recorded on videotape (Model PCM-2, Medical Systems Corp., Greenvale, NY). The signal was replayed, filtered at the level indicated in the text, and digitized at 1 ms/point with a digital oscilloscope (Model 310, Nicolet, Madison, WI). Mathcad 4.0 (MathSoft, Inc., Cambridge, MA) was used to evaluate equations. Electrode Calibration. Electrode sensitivity and response time were determined in a flow injection apparatus before and after in vivo use.39 Most electrodes were tested in a buffer that contained 20 mM HEPES and 150 mM NaCl, adjusted to pH 7.4. Occasionally, a phosphate buffer that contained 60 mM Na2HPO4 and 30 mM NaH2PO4 adjusted to pH 7.4 or 12.5 mM Tris buffer at pH 7.4 was used, but calibrations before and after in vivo use were always done in the same buffer. In Vivo Experiments. Surgical procedures for in vivo voltammetry have been described elsewhere.40 Briefly, male Sprague-Dawley rats (Charles River Inc., Raleigh, NC) weighing 300-500 g were deeply anesthetized with urethane (1.5 g/kg). Rats were immobilized in a stereotaxic apparatus (Narishige), and holes were drilled in the skull for placement of reference (Ag/AgCl, Bioanalytical Systems Inc., West Lafayette, IN), stimulating (Plastics One Inc., Roanoke, VA), and carbon-fiber working electrodes. The placement of working and stimulating electrodes was based on flat-skull coordinates obtained from a brain atlas.41 Stimulating electrodes were always placed in the medial forebrain bundle (MFB), while working electrodes were either placed in the caudate-putamen (CP) or nucleus accumbens (NAc). Coordinates (in mm) anteroposterior (AP) and mediolateral (ML) from bregma and dorsoventral (DV) from dura are as follows for these sites: MFB, -4.6 AP, +1.4 ML, -7.5 to -9.0 DV; CP, +1.2 to +1.4 AP, +1.2 to 3.0 ML, -4.5 to -6.0 DV; NAc, +1.4 AP, +1.4 ML, -6.5 to -6.9 DV. A triangle wave (-0.4 to 1.0 V, at a scan rate of 300 V/s) was applied every 100 ms to working electrodes upon placement in the rat brain. The carbon-fiber microelectrodes were lowered slowly to either CP or NAc. Electrical stimulation consisted of optically isolated (NL800 Neurolog, Medical Systems Corp., Great Neck, NY) biphasic pulses (300 µA, 2 ms each phase) delivered between voltammetric scans.40 Electrodes were typically in the brain for 3-5 h in these experiments. Chemicals. Nafion, 5% by weight, was obtained from Aldrich (Milwaukee, WI). Nafion was diluted with 2-propanol to achieve the desired concentration (2.5%). Cells were cultured in Dulbecco’s modified Eagles medium with Ham’s F12 (DMEM/F12) obtained from Gibco Laboratories (Ronkonkoma, NY). Collagenase, type I, was obtained from Worthington Chemicals (Freehold, NJ), and renografin-60 was from Squibb Diagnostics (New Brunswick, NJ). All other chemicals were from Sigma (St. Louis, MO) and were used as received. Solutions were made in doubly distilled water. (39) Howell, J. O.; Kuhr, W. G.; Ensman, R. E.; Wightman, R. M. J. Electroanal. Chem. 1986, 209, 77-90. (40) Garris, P. A.; Collins, L. B.; Jones, S. R.; Wightman, R. M. J. Neurochem. 1993, 61, 637-647. (41) Paxinos, G.; Watson, C. The Rat Brain in Stereotaxic Coordinates; Academic Press: New York, NY, 1986.

3182

Analytical Chemistry, Vol. 68, No. 18, September 15, 1996

Figure 1. Amperometric detection of catecholamine exocytosis from chromaffin cells using the Axopatch 200B, the AI-403 picoammeter, and the EI-400 potentiostat. Individual chromaffin cells were induced to release catecholamine by transient (3-s) application of 60 mM K+ prior to the collection of the spike data. Gain was set at 0.1 nA/V, and the applied potential was +0.65 V in all three cases. The Axopatch 200 B and the EI-400 currents were low-pass filtered with the Butterworth filter, and the AI-403 current was filtered with the Bessel filter. The low-pass filter cut-off frequency was 200 Hz for all three instruments. An expanded plot of the baseline noise measured with the Axopatch 200B, AI-403, and EI-400 is shown in the inset.

RESULTS AND DISCUSSION Instrumental Noise in Amperometry. The current noise is proportional to capacitance, which is dependent upon the electrode’s area.35 With very small electrodes, the capacitance becomes sufficiently small that the Johnson noise of the electronic instrument predominates. The Johnson noise (Nrms) in volts for a current transducer or follower can be calculated from42

Nrms ) (4kBTRf f )1/2

where kB is the Boltzmann constant (1.37 × 10-23 V/K), Rf is the feedback resistor, T is the temperature (298 K), and f is the instrument’s bandpass in hertz. Since the signal is proportional to Rf, the signal-to-noise ratio increases with Rf1/2. However, the time constant caused by stray capacitance in the instrument limits the upper values of Rf. Very high values of Rf require frequency compensation. Low-noise current amplifiers using this approach have been developed by patch-clamp physiologists, and we tested two, the Axopatch 200B and the AI-403 picoammeter, in an application requiring very small electrodes. The Axopatch 200B also has Peltier cooling to lower the temperature of the headstage and thereby to further decrease the impact of Johnson noise on the measurement. As a result, the noise is mainly a factor of uncompensated resistance in solution, which is a property of the electrochemical cell. Catecholamine release from a single bovine adrenal chromaffin cell was monitored with the Axopatch 200B, the EI-400 potentiostat, and the AI-403 picoammeter (Figure 1). The cell was exposed to a 3-s application of 60 mM K+ to cause secretion of (42) Horowitz, P.; Hill, W. The Art of Electronics; Cambridge University Press: Cambridge, UK, 1989; Chapter 7.

Figure 2. Voltammetric response to dopamine measured in a flow injection apparatus. Average current (b) from a 100-mV window centered at the potential for dopamine oxidation. The valve was switched on at 5 s and off at 15 s. The insets are the backgroundsubtracted cyclic voltammograms at 300 V/s. The voltammograms were constructed by averaging the current of over 30 scans before and during the DA exposure and then subtracting the currents: (A) 1 µM DA at a bare elliptical microelectrode, (B) 1 µM DA at a bare cylindrical electrode, 60 µm in length, (C) 100 nM DA at the same elliptical microelectrode, and (D) 100 nM DA at the same cylindrical microelectrode.

the catecholamines, which was observed as a series of sharp current spikes resulting from the release of the contents of individual vesicles.23 As shown in the inset, both the Axopatch 200B and the AI-403 picoammeter permit the measurement of cell data with a wider band pass, without the constraints of large background instrument noise. The EI-400 potentiostat was designed to be used with repetitive fast-scan cyclic voltammetry, with the repetition rate of the voltammograms synchronized with line frequency. Thus, special precautions concerning line noise were not employed to remove it. When examined in the frequency domain, white noise is present, but also noise at line frequency and its harmonics occurs with the EI-400. In contrast, the AI-403 and Axopatch 200B are much closer to the theoretical expectations. (Note, in a previous report, we attributed the noise of the EI-400 to Johnson noise alone. This was erroneous, and the theoretical line in Figure 2 of ref 35 includes the computed amplifier noise as well.) Characterization of Cylindrical Microelectrodes. When the source of the chemical messenger is large, electrodes with larger surface areas can be beneficial. Increasing microelectrode size is not detrimental to the overall sensitivity in this case because both signal and noise are proportional to electrode area. The electrode size can be increased without producing any additional tissue damage in the brain if cylindrical microelectrodes are used. In a cylindrical microelectrode, the carbon fiber extends 10-150 µM beyond the end of the glass micropipet. A comparison of the peak currents and cyclic voltammograms measured with an elliptical electrode and a cylindrical electrode, 60 µm in length, with the EI-400 potentiostat is shown in Figure 2. The surface area of the cylindrical electrode was about 11 times that of the elliptical electrode. Both types of electrodes produced signals for 1 µM dopamine that had reasonable signal-to-noise ratios (S/N), although the larger electrode is clearly superior (S/N ratios were 29 and 74 for elliptical and cylindrical electrodes, respectively). Both background-subtracted voltammograms had oxidation and

Figure 3. Two-pulse, 10-Hz stimulation in vivo using a 60-µm bare cylindrical electrode. The dopamine concentration change as a function of time is shown. The squares represent the start and end of the stimulation. A background-subtracted cyclic voltammogram, constructed by averaging five scans after the stimulation onset and subtracting the background current before the stimulation, is shown in the inset. The cyclic voltammogram was three-point smoothed.

reduction peaks for dopamine at the locations found at higher concentrations. In contrast, the background-subtracted cyclic voltammogram for 100 nM dopamine at the elliptical electrode did not show well-defined peaks for dopamine, and the S/N was only 4.9. The lower concentration was easily detected at the cylindrical electrode with a S/N of 14. Since the currents are larger with the cylinder, the potentiostat is no longer working in the Johnson noise limited regime, which reduces the relative noise, and sensitivity is increased. The dopamine peak current was found to be linear with concentration for both elliptical and cylindrical electrodes up to 1 µM dopamine. At concentrations greater than 1 µM dopamine, the current at cylindrical electrodes is nonlinear with concentration because of adsorption.43 The cyclic voltammogram for 100 nM dopamine at a cylindrical electrode (Figure 2D) shows an unusual feature. On the forward scan, the current is greater than zero except near the oxidative peak potential, and on the reverse scan, the current is less than zero except near the reductive peak potential. As a result, the oxidative and reductive scans in the voltammogram cross each other. This effect can also be seen at cylindrical electrodes in vivo (Figure 3). The size of this artifact was independent of the dopamine concentration but occurred only when dopamine was present. It may reflect a capacitance decrease resulting from adsorption of dopamine to active sites on the electrode surface. Consistent with this hypothesis, the effect is not seen with Nafioncoated elliptical electrodes because the Nafion layer blocks the adsorption sites. Cylindrical Microelectrodes in Vivo. To determine whether exposure to the in vivo environment caused decreased electrode sensitivity, the electrodes were calibrated with dopamine in a flow injection system, implanted in the brain for at least 3 h, removed from the brain, rinsed with saline, and then tested in the flow injection system again as before. Table 1 compares bare cylindrical and Nafion-coated elliptical electrode performance after in vivo use. As expected, the background current, faradaic current, and noise are larger for cylindrical electrodes because of their larger surface area. However, the signal-to-noise ratio is about a factor of 3 better for cylindrical electrodes than elliptical electrodes. The time required for the electrode to reach half maximum (t1/2) is also less for cylindrical electrodes. The increase in speed is not the result of the larger surface areas but is instead because the (43) Michael, A. C.; Justice, J. B., Jr. Anal. Chem. 1987, 59, 405-410.

Analytical Chemistry, Vol. 68, No. 18, September 15, 1996

3183

Table 1. Cyclic Voltammetry Response to 1 µM DA after in Vivo Usea

P55 elliptical T300 cylindrical

area × 106 (cm2)

ibkd (nA)

iDA (nA)

Nrms (nA)

S/N

t1/2 (s)

n

2.22 19.9

24.8 ( 2.2 286 ( 59.6

0.157 ( 0.019 3.71 ( 0.63

0.035 ( 0.01 0.16 ( 0.02

6.04 ( 1.34 22.7 ( 1.5

0.54 ( 0.09 0.31 ( 0.01

13 7

a The P55 elliptical electrodes were coated with Nafion, and the T300 cylindrical electrodes were bare. The mean length of the T300 cylindrical electrodes was 88.6 ( 14.4 µm. From the nominal carbon fiber radii and the optically measured lengths, the electrode area was calculated. Background current (ibkd), peak current due to dopamine (iDA), and the root-mean-squared noise (Nrms) were calculated from the cyclic voltammograms. S/N is the signal-to-noise ratio. The time required for the current to reach half of its maximum (t1/2) was taken from a plot of current as a function of time. The number of electrodes tested is reported as n. Values are reported as the mean ( the standard error of the mean.

analytes did not have to diffuse through a polymer layer to be detected. Comparison of data from before and after in vivo use shows the extent that electrode performance has been degraded by exposure to the brain environment. Previous studies have shown that Nafion-coated electrodes had a smaller decline in sensitivity during implantation than bare electrodes.36,44,45 In the present study, Nafion protected the electrode’s surface, and the Nafioncoated elliptical electrodes retained greater than 90% (postcalibration/precalibration values) of their original background and dopamine current response. The electrode response during postcalibration is dependent upon the complete removal of blood vessels from the brain surface and the method of rinsing the electrode. In contrast, bare cylindrical electrodes retained 73% and 33% of their original background and dopamine signals, respectively. The decrease in sensitivity at bare elliptical electrodes has been shown previously to occur during implantation of the electrode since repeated electrical stimulations elicit the same dopamine responses for hours.46 Similar experiments with repeated stimulations after implantation in the brain showed that the mean change in evoked release of dopamine detected, expressed as a percentage, was 95.6 ( 12.9% for bare cylindrical electrodes (n ) 5) implanted for 2.2 h. In comparison, the mean percent change was 107 ( 28.7% for Nafion-coated elliptical electrodes (n ) 8) implanted for 3.0 h. This result confirms that the decrease in response to dopamine occurs early during implantation and that, once decreased, the sensitivity does not change with time. Therefore, the postcalibration response was used to convert the measured currents into the in vivo concentrations. The loss of dopamine signal at cylindrical electrodes was compensated by the fact that the noise (Nrms) was unchanged by the implantation in the brain, but it increased by a factor of 3 at elliptical electrodes. It is believed that the increased noise at the elliptical electrodes was caused by degradation of the epoxy seals between the carbon fiber and the glass insulation. This region is a significant factor of the total area. Less of the current density is focused near the epoxy seal in cylindrical electrodes. The degradation in overall sensitivity is similar at the two types of electrodes, with the S/N ratios decreasing to 55% and 56% of their values before implantation for elliptical and cylindrical electrodes, respectively. The half-rise time (t1/2) of the elliptical electrodes increased to 467% of its value prior to in vivo use, while the (44) Ewing, A. G.; Dayton, M. A.; Wightman, R. M. Anal. Chem. 1981, 53, 18421847. (45) Kristensen, E. W.; Kuhr, W. G.; Wightman, R. M. Anal. Chem. 1987, 59, 1752-1757. (46) Ewing, A. G.; Bigelow, J. C.; Wightman, R. M. Science 1983, 221, 169170.

3184 Analytical Chemistry, Vol. 68, No. 18, September 15, 1996

cylindrical electrode’s half-rise time increased to only 162% of its value prior to in vivo use. It has been proposed that, during implantation of cylindrical electrodes in the brain, damage to the electrode caused by adsorption of proteins and tissue occurs only on the end of the cylinder, while the sides are relatively undamaged by exposure to the brain.47 This theory could explain why the response time for cylindrical electrodes is less affected by implantation than the response time for elliptical electrodes. Thus, the cylindrical electrodes are superior because they have higher sensitivity (S/N) and faster temporal response (t1/2) since the analyte does not have to diffuse through the Nafion layer. Fast temporal response is especially important with short-duration stimulations, like a single-pulse stimulation, because the small dopamine concentration change is very short-lived.3 A two-pulse, 10-Hz stimulation in vivo using a 60-µm, bare cylindrical electrode is shown in Figure 3. In previous work with Nafion-coated elliptical electrodes, the smallest stimulation that was reliably recorded was four stimulation pulses delivered in a 30-ms interval.33 The background-subtracted cyclic voltammogram is shown in the inset. The characteristic oxidation and reduction peaks for dopamine are present, which indicates that dopamine is the substance being monitored. Selective Filtering To Improve Sensitivity. The largeamplitude background currents in fast-scan cyclic voltammetry require a less sensitive amplifier setting to prevent the saturation of the operational amplifiers in the potentiostat. Because the dopamine signal is usually smaller than the background current, the amplifier setting is not ideal and reduces sensitivity. Highpass filtering of the signal was employed to reduce the size of the background current. The background current occurs mainly at the fundamental frequency of the cyclic scan, while faradaic current occurs at higher frequencies. When the scan rate was 300 V/s, the fundamental frequency was approximately 100 Hz. The high-pass filter cut-off frequency was chosen to be 200 Hz, allowing the faradaic signals to pass while screening out the background current. A comparison of cyclic voltammetry and high-pass filtered cyclic voltammetry of 10 µM dopamine is shown in Figure 4. Without high-pass filtering, the amplitude of the background current was ∼400 nA, but when high-pass filtering was used, the amplitude was reduced to ∼100 nA. The reduction in charging current allowed a more sensitive setting to be used on the analog-to-digital converter; thus, there was less digitization noise in the measurement. The amplitude of the dopamine signal was decreased by high-pass filtering as well, but there was an improvement in the signal-to-noise ratio at low concentrations. The background-subtracted voltammogram was changed as a result (47) Gonon, F. In Voltammetric Methods in Brain Systems; Boulton, A. A., Baker, G. B., Adams, R. N., Eds.; Humana Press: Totowa, NJ, 1995; Chapter 5.

Figure 4. Comparison of cyclic voltammetry and high-pass filtered cyclic voltammetry of 10 µM DA at a cylindrical electrode, 70 µm in length. The current was high-pass filtered at 200 Hz with a 24 dB/ octave roll-off. (A) Background current in cyclic voltammetry, (B) sum of background and dopamine faradaic current, and (C) resulting background-subtracted cyclic voltammogram. (D) Background current in high-pass filtered cyclic voltammetry, (E) sum of background and dopamine faradaic current, and (F) resulting background-subtracted cyclic voltammogram with high-pass filtering. The oxidative scans are shown by solid lines, and the reductive scans are shown by dashed lines.

of high-pass filtering of the current (Figure 4F). It appears similar to the derivative of a dopamine cyclic voltammogram (Figure 4C) because a high-pass filter acts similar to a differentiator.48 The curve does not contain all of the low-frequency information; therefore, the integration of the high-pass filtered signal does not yield a cyclic voltammogram. Discriminating against Interferences with High-Pass Filtering. The extracellular fluid of the rat brain contains many compounds in addition to dopamine. Fortunately, most of the species present are not electroactive and therefore do not interfere with the dopamine signal. Ascorbic acid can reach concentrations as high as 400 µM in the brain, and it is oxidized at potentials near that for dopamine oxidation. 3,4-Dihydroxyphenylacetic acid (DOPAC), a dopamine metabolite, is also found at higher concentrations and can interfere with dopamine electrochemically. In the past, a cation exchange polymer, Nafion, has been coated onto carbon microelectrodes to screen out the ascorbate and DOPAC anions.49-51 This method has been effective, but the presence of the additional layer on the electrode has decreased the temporal response of the electrode.34,45 Also, because Nafion is dip-coated onto the electrode, it is difficult to obtain a uniform layer on the electroactive surface of a cylindrical electrode. A previous report discussed electrodepositing polypyrrole on the surface of the electrode to increase catecholamine sensitivity.52 Since overoxidized polypyrrole is electrodeposited, these coatings (48) Malmstadt, H. V.; Enke, C. G.; Crouch, S. R. Electronics and Instrumentation for Scientists; Benjamin/Cummings Publishing: Menlo Park, CA, 1981; Chapter 5. (49) Brazell, M. P.; Kasser, R. J.; Renner, K. J.; Feng, J.; Moghaddam, B.; Adams, R. N. J. Neurosci. Methods 1987, 22, 167-172. (50) Garris, P. A.; Ciolkowski, E. L.; Wightman, R. M. Neuroscience 1994, 59, 417-427. (51) Kawagoe, K. T.; Garris, P. A.; Wightman, R. M. J. Electroanal. Chem. 1993, 359, 193-207.

Figure 5. Discrimination against ascrobate and acidic pH changes with high-pass filtering. Cyclic voltammograms recorded with cylindrical electrodes for (A) 200 nM DA, (B) 200 µM AA, and (C) an acidic change in HEPES, NaCl buffer from pH 7.4 to 7.3. Cyclic voltammograms that were high-pass filtered at 200 Hz of (D) 200 nM DA, (E) 200 µM AA, and (F) an acidic change from pH 7.4 to 7.3. The voltammograms in (A) and (D) were recorded with a 170-µm cylindrical electrode, and those in (B), (C), (E) and (F) were recorded with a 100-µm cylindrical electrode. The oxidative scans are shown by solid lines, and the reductive scans are shown by dashed lines.

are useful for cylindrical electrodes. However, decreased temporal response was observed because of the time required for diffusion through the polymer layer and dopamine adsorption into the polypyrrole film. Therefore, bare cylindrical electrodes are an improvement over cylindrical microelectrodes coated with overoxidized polypyrrole. It has also been shown that pH changes can also interfere with the measured dopamine signal, even at Nafion-coated electrodes.51,53 The cyclic voltammograms and high-pass filtered voltammograms recorded with a bare cylindrical electrode for 200 nM dopamine, 200 µM ascorbate, and an acidic pH change from 7.4 to 7.3 are shown in Figure 5. The cyclic voltammograms for ascorbate and acidic pH changes are of lower frequency than the faradaic processes for dopamine (Figure 5A-C). High-pass filtering can be used to discriminate against less reversible electrochemistry, like that of ascorbate or an acidic pH change (Figure 5D-F). A 50 µM DOPAC sample was also tested with cyclic voltammetry and high-pass filtering at bare cylindrical electrodes, and the voltammograms had shapes similar to those for ascorbate (data not shown). The ratios of the signal for 10 µM dopamine to the ascorbate signal were 20 and 99 for cyclic voltammetry and high-pass filtered cyclic voltammetry, respectively. For an acidic pH change, the ratios were 21 for cyclic voltammetry and 78 for cyclic voltammetry with high-pass filtering. The ratios of the dopamine signal to the DOPAC signal were 18 and 387 for cyclic voltammetry and high-pass filtered cyclic voltammetry, respectively. High-pass filtering permits the measurement of dopamine in vivo with at least 4 times less interference from ascorbate, DOPAC, or acidic pH changes. (52) Pihel, K.; Walker, Q. D.; Wightman, R. M. Anal. Chem. 1996, 68, 20842089. (53) Jones, S. R.; Mickelson, G. E.; Collins, L. B.; Kawagoe, K. T.; Wightman, R. M. J. Neurosci. Methods 1994, 52, 1-10.

Analytical Chemistry, Vol. 68, No. 18, September 15, 1996

3185

CONCLUSIONS

ACKNOWLEDGMENT

Although small electrodes are advantageous at single cells, cylindrical electrodes with greater areas have been shown to increase the sensitivity to dopamine release in the brain. A factor of 3 improvement in the signal-to-noise ratio was observed for bare, cylindrical electrodes after in vivo use compared to Nafion-coated elliptical electrodes. High-pass filtering was combined with fastscan cyclic voltammetry at 300 V/s to decrease the impact of charging current on the measurement and to discriminate against ascorbic acid and acidic pH changes.

This work was funded by a grant from the National Institutes of Health.

3186

Analytical Chemistry, Vol. 68, No. 18, September 15, 1996

Received for review April 10, 1996. Accepted July 10, 1996.X AC960347D X

Abstract published in Advance ACS Abstracts, August 15, 1996.