Anal. Chem. 1991, 63,24-28
24
T a b l e IV. Accuracy of E t h a n o l i c Potassium Analysis P e r f o r m e d with Silica Gel B o u n d Macrocvcles U s e d as Concentrator Columns4
Kt in ethanol std, M 1.02 x 10-4 5.12 X 1.28 X 10"
method concn factorn
50 50 50
K+ in concentrate, M
anal. result, Kt,
M
5.02 x 10-3 2.60 X 6.29 X
LOO x 10-4 5.20 X 1.26 X
% error in anal. -2.0 f 3.0 1.8 f 3.0 -1.6 f 5.0
O250 mL of ethanolic K+ loading solution was run through the concentrator columns. Kt was stripped in 5 mL of aqueous solution containing 0.1 M 18C6 and 0.01 M HC1. T h e aqueous solution was then analyzed by atomic absorption spectroscopy.
Within experimental uncertainty, anion type had no effect on SGBM-K+ interactions in methanol. The absence of the effect of anion type on log K was anticipated. Yu and Jing report that log K(H20)values for free macrocycle-K+ interactions vary at most by 0.4 log K unit depending on the anion used (13). Lamb et al. showed that macrocyclic-mediated cation flux through liquid membranes was dependent on anion type (14).The fluxes varied more than 2 orders of magnitude in changing from chloride to perchlorate when a chloroform membrane was used, but chloroform is much less polar than the solvents used in this study. The effect of anion type on log K values for SGBM-cation interactions might be expected to increase with decreasing solvent polarity. Unfortunately, few literature log K data are available for low-polarity solvents such as toluene or chloroform because of the low solubility of salts in these solvents. This low solubility also prevents experimentation with SGBM
when these solvents are used.
ACKNOWLEDGMENT Appreciation is expressed to Bryon Tarbet who synthesized the silica gel bound 18C6. LITERATURE CITED Duran, J.; Viswanathan, N. S. In P o ~ in ~Nectronics; s Davidson, T., Ed.; ACS Symposium Series 242; American Chemical Society: Washington, D.C., 1984; pp 239-258. Pong, C. In Polymer Materials for Electronic Applications; FeR, E. D., Wilkin, C. W., Eds.; ACS Symposium Series 184; American Chemical Society: Washington, D.C., 1982; pp 171-183. Izatt, R. M.; Bradshaw, J. S.; Nieisen, S. A.; Lamb, J. D.; Sen, D. Chem. Rev. 1985, 85, 271-339. Izatt, R. M.; Roper, D. K.; Bruening, R. L.; Lamb, J. D. J . Membrane Sci. 1989, 45, 73-64. Izatt, R. M.; Bruening, R. L.; Bruening, M. L.; Tarbet, B. J.; Krakowiak, K. E.; Bradshaw, J. S.; Christensen, J. J. Anal. Chem. 1988, 60, 1825-1826. Bradshaw, J. S.; Bruening, R. L.; Krakowiak, K. E.; Tarbet, B. J.; Bruening, M. L.; Izatt, R. M.; Christensen, J. J. J . Chem. Soc., Chem. Commun. 1988, 812-814. Izatt, R. M.; Bruening, R. L.; Tarbet, B. J.; Griffin, L. D.; Bruening, M. L.; Krakowiak, K. E.; Bradshaw, J. S. Pure Appi. Chem. 1990, 62, 1115-1118. Bradshaw,J. S.; Krakowiak, K. E.; Tarbet, 8. J.; Bruening, R. L.; Biernat, J. F.; Bochenska, M.; Izatt, R. M.; Christensen, J. J. Pure Appl. Chem. 1989, 67, 1619-1624. Weaver, M. R.; Harris, J. M. Anal. Chem. 1989, 67, 1001-1010. Bogatskii, A. V.; Luk'yanenko, N. G.; Popov, Y. A.; Zakharou, K. S.; Varava, V. M. Zh. Org. Kbim. 1981, 77, 1062-1069. Boss, R. D.; Popov, A. I. Inorg. Chem. 1988, 25, 1747-1750. Unger, K. K. Porous Siiica; Elsevier: Amsterdam, 1979; Chapter 6. Yu, L.; Jing, H. Wuli Huaxue Xuebao 16 1987, 3 , 11-15. Lamb, J. D.; Christensen, J. J.; Izatt, S. R.; Bedke, K.; Astin, M. S.; Izatt, R. M. J . Am. Chem. SOC. 1980, 702, 3399-3403.
RECEIVED for review July 20,1990. Accepted October 8,1990. This work was supported by the State of Utah Centers of Excellence Program.
Simultaneous Electrochemical Measurements of Oxygen and Dopamine in Vivo Jayne B. Zimmerman and R. Mark Wightman* Department of Chemistry, C B No. 3290, Venable Hall, T h e University of North Carolina at Chapel Hill, Chapel Hill, North Carolina 27599-3290 Fast-scan cyclic voltammetry, a demonstrated analytical method for the in vivo detection of catecholamine neurotransmitters, Is extended to the simultaneous determination of molecular oxygen (0,). Cyclic voltammograms were recorded at a scan rate of 400 V I S at carbon-fiber disk electrodes coated wlth a perfluorinated ion-exchange materlai. The peak current for O2 occurs near -1.2 V under these condltlons. I n flow-injectlon experiments, these electrodes respond to step changes In dopamine and 0, wlth a half-rise time of less than 200 ms. The voltammetrlc peak current Is Independent of flow rate, lndicatlng a diffusion-llmlted response unaffected by convectlon. Several compounds present in the In vivo matrlx (adenosine, glutathione, and NAD and glutamic, lactic, and uric acids) were tested and shown not to Interfere wlth the voltammetrlc slgnai for 0,. These electrodes malntaln a stable response in vivo for at least 6 h. They have been used to measure translent increases in both dopamlne and 0, In the extracellular fluid of the caudate nucleus of an anesthetlzed rat In response to an electrical stlmulus.
* T o whom correspondence should be addressed. 0003-2700/9 110363-0024$02.50/0
INTRODUCTION Amperometry provides a particularly convenient way to measure dissolved O2 The type of electrode most commonly used, the Clark electrode, contains a platinum working electrode that is coated with electrolyte and covered with a gaspermeable membrane such as polyethylene (1). Typically, these electrodes are operated with a dc applied potential, and the measured current is taken to be proportional to the concentration of dissolved 02.To enable implantation in tissue, microelectrodes have been employed (2-4). Usually, these employ a cathode that is recessed inside the surrounding insulation. The recess is filled with a hydrogel plug, which prevents fouling of the electrode surface and establishes a diffusion layer boundary that is constant in vitro and in vivo, thus permitting calibration curves to be used (5). In addition to reducing tissue damage, miniaturization of the cathode area makes the electrode less sensitive to convection, which is important when the electrode is implanted in blood vessels. The time response of these types of electrodes is typically 1 s because of the time for O2diffusion inside the recessed tip. In previous work we have demonstrated that beveled carbon-fiber microelectrodes can be used for chemical detection deep inside the brain (6). When coated with a perfluorinated 0 1990 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 63, NO. 1, JANUARY 1, 1991
cation-exchange polymer, these electrodes are capable of subsecond time resolution for the selective detection of dopamine, a neurotransmitter, in brain tissue (7). The polymer film acts as an effective screen t o prevent interference from other compounds, while a t the same time providing an electrolytic environment adjacent to the electrode. While these electrodes are larger than some used for in vivo O2 measurements, the beveled tip allows facile insertion into tissue. The rapid response time is possible because thin films are employed and because the concentration is measured with fast-scan cyclic voltammetry. The fast-scan technique has three additional advantages. First, the size of the diffusion layer is small and, thus, the regions sampled by the in vivo measurements and those in vitro are more comparable, which also enables the use of calibration curves (8). A second advantage of fast-scan techniques is that the amount of electrogenerated products (which may be toxic) generated during the transient measurement is much smaller than that generated with a dc applied potential (9). A third advantage is that the voltammogram can serve as an identifier for the detected substance (9). A significant disadvantage of the fast-scan technique is the presence of the large background due to the charging current, but this background is remarkably stable for carbon electrodes, making straightforward the digital background subtraction to obtain the faradaic signal. In this work we show that this approach can be extended to provide a technique that is suitable for simultaneous in vivo detection of O2and dopamine. The oxidation of glucose by O2is the energy source for the brain. Since O2 availability and neurotransmitter release are intimately linked, such a sensor should be particularly useful to investigate these events (10). The advantages described above for dopamine detection with transient techniques are also pertinent in the measurement of O2 (11, 12). O2 is readily able to permeate Nafion (13) and thus reach the electrode surface for reduction. Carbon is infrequently used as a cathode for O2because the heterogeneous kinetics for the reduction of O2are slow at unactivated carbon surfaces, leading to drawn out, irreversible voltammetric waves (14,15). However, in our studies of easily oxidized substances in brain tissue, we have found that carbon electrodes are advantageous because they are less prone to surface changes over the long times required for an in vivo sensor. We find that the voltammetric wave observed for O2 reduction a t carbon electrodes is stable and reproducible and provides a characteristic identifier for 02. EXPERIMENTAL SECTION Electrochemistry. The carbon-disk working electrodes were fabricated from 5-pm radius carbon fibers that were sealed with epoxy into glass capillaries (6). The tips were beveled on a polishing wheel to give an elliptical surface (major radius = 35 pm) and dip-coated with the perfluorinated cation-exchange polymer Nafion (Aldrich,Milwaukee,WI), as described previously (7). A saturated sodium calomel electrode (SSCE) was used as the reference in the two-electrode cell. The current transducer had a 10-ps time constant, and the signal was subsequently lowpass filtered at 4.2 kHz (EI-400, Ensman Instrumentation, Bloomington, IN). Data acquisition was under software and hardware control from an IBM-PC with a commercial interface board (Labmaster, Scientific Solution, Solon, OH). The potential waveform for cyclic-staircase voltammetry was computer generated with a step potential, AE,of 24.4 mV. Current measurements were made 8 ps prior to the end of each step (7 = 60 p s ) . Background subtraction of the voltammograms was accomplished with the use of data obtained immediately prior to exposure to the analyte or prior to initiation of the stimulus in the in vivo experiment. Other signal-processing procedures are as described in ref 9. Electrode characterization and calibration were conducted with a flow-injectionsystem (9). Use of Teflon tubing and components was avoided because they are permeable to 02.Glass and stainless
25
steel components were used instead. The working electrode was centered in the outlet tube (1/16 in. glass-lined stainless steel) of a loop injector (Rheodyne 7010, Cotati, CA). The flow rate, controlled by a syringe pump (Harvard Apparatus, South Natick, MA) was normally 1.91 mL/min. The reference electrode was positioned in a reservoir downstream from the carbon-fiber electrode. The buffer normally employed in the flow stream was 150.0 mM NaCl and 20 mM HEPES at pH 7.4. Dopamine solutions were prepared fresh daily in this buffer by diluting aliquots from a stock solution of 10 mM dopamine in 0.1 N HC104. Calibrations for dissolved O2used three solutions: an N2-purged NaCl solution, an air-saturated solution, and an 02-saturated solution. The solution temperature and atmospheric pressure were used to calculate the concentration of O2in each solution, according to Chapter 2 of ref 1: 1 mol pT - pl [O,] (mol/L) = a02(frac)760 Torr 22.414 L where a is the Bunsen absorption coefficient,which includes the effect of temperature and salt concentration, 02(frac) is the proportion of O2 in the dry gas phase with which the solution is in equilibrium (0.2095 for air), PT is the atmospheric pressure, and Pl is a correction for PT for water vapor in the gas phase above the solution. In Vivo Techniques. The in vivo procedures were as described previously (16). Briefly, an adult male Sprague-Dawley rat was anaesthetized with urethane (ethyl carbamate, 1500 mg/kg, ip) and positioned into a stereotaxic frame. The skull was surgically exposed, and small holes were drilled for electrode implantation. After the carbon-fiber electrode was positioned in the caudate nucleus, a twisted bipolar stimulating electrode was lowered to the medial forebrain bundle. (Coordinates used were AP +1.2, ML +2.0, DV -4.9, and AP -3.8, ML +1.6, DV -7.8 for the caudate and MFB, respectively. All dimensions are in millimeters acThe reference cording to the atlas of Paxinos and Watson electrode was placed in electrical contact with the brain tissue through a salt bridge. The stimulus employed was a train of biphasic current pulses, each 2 ms wide and 300 pA in amplitude, timed so that no stimulus pulses occur during acquisition of a voltammogram. Reagents. All chemicals were used as received from commercial sources. Dopamine was obtained from Sigma Chemical Co. (St. Louis, MO) as 3-hydroxytyramine hydrochloride. Compounds used in the interference study were adenosine (Sigma), L-glutamic acid monosodium salt (Sigma), glutathione (oxidized form) disodium salt (Sigma), DL-lactic acid, 85% syrup (Fisher Scientific, Fair Lawn, NJ), @-nicotinamideadenine dinucleotide (NAD) sodium salt (Sigma), and uric acid (MC and B, Norwood, OH). Compressed O2 was USP grade (Sunox, Charlotte, NC).
(In.)
RESULTS AND DISCUSSION Voltammetry of Dopamine and O2 The potential waveform for cyclic-staircase voltammetry consists of an initial anodic excursion to +0.8 V from the rest potential of 0.0 V and a cathodic sweep to -1.4 V followed by a return t o the rest potential. At 400 V/s, the complete scan occurs in 10.5 ms. Shown in Figure 1A is the background current obtained during a single scan in N2-purged solution and a scan in which both dopamine (10 pM) and O2(100 pM) are present in the solution. As can be seen, the background current a t this scan is large compared to the faradaic signals for the two analytes. However, digital subtraction of the background current gives a cyclic voltammogram which contains only the faradaic components, as shown in Figure 1B. The peak potentials for the oxidation and rereduction of dopamine are, respectively, +0.61 and -0.15 V. The O2 wave is chemically irreversible, with a peak potential near -1.3 V. Similar voltammetric behavior was found when the initial voltammetric scan was to negative potentials. The small reverse wave seen following O2reduction was present in these scans as well. However, in some cases the time for dopamine to permeate the ionomer film was found to deteriorate with prolonged use when O2 was detected first in the scan. This behavior may be a consequence of the presence of O2reduction
26
ANALYTICAL CHEMISTRY, VOL. 63,NO. 1, JANUARY 1, 1991
7 1 IC””””””””””’””””
I50 nA
#--
I
:
0.5
:
:
:
0.0 -0.5 -1.0
!
0.5
:
:
:
0.0 -0.5 -1.0
:
/
E (V vs. SSCE) Flgure 1. Cyclic voltammetry of dopamine and dissolved 0, at a Nafoncoated carbon-fiber electrode (scan rate 407 V/s). (A) Dashed line: background current obtained during a single scan in N,saturated buffered saline containing no electroactive species. The scan begins at 0.0 V and is initially swept anodically. Solid line: total current following injection of a solution containing 10.0 pM dopamine and 100 pM dissolved 02.(B) Background-corrected cyclic voltammogram obtained by digital subtraction of the two signals shown in A.
products in the Nafion film when the cathodic portion of the scan is performed first. In any case, an initial anodic scan was used in all subsequent work. The voltammogram of O2 and dopamine is consistent with their known redox behavior. Dopamine is oxidized in a two-electron process to the corresponding o-quinone, which is stable on the time scale of the voltammograms, and can be rereduced (18). At this scan rate the rate-limiting step in the oxidation is the initial one-electron oxidation of the catechol anion (19). O2 is reduced in a two-electron process at carbon electrodes to give predominantly hydrogen peroxide (14), which is not oxidized at carbon electrodes, leading to the observed, irreversible behavior. The peak potential is shifted from that observed in prior work a t unactivated carbon electrodes (15, 20) because the reduction is kinetically limited and, thus, the peak potential is a t more negative potentials a t the high scan rates used in this work. The rate-limiting step in O2 reduction at carbon is the initial one-electron reduction followed by protonation of the superoxide ion and further reduction (14). The wave height for dopamine is larger than expected because of accumulation in the film (9). Continuous Monitoring by Cyclic Voltammetry. When voltammograms are recorded continuously at 100-msintervals, the oxidation current for dopamine and reduction current for O2 from successive voltammograms can be used to measure simultaneous changes in their concentration. Figure 2 shows such data for an injection of a solution containing dopamine and O2 into a solution of deaerated buffer. The time before concentration changes are seen represents the time for transport from the loop injector to the electrode. While the currents change in opposite directions for the two compounds (oxidation current for dopamine detection, reduction current for 02), we have chosen to show these curves in the direction indicative of the concentration changes. The upper curve shows the increase in relative current due to O2 reduction, measured in the potential window from -1.20 to -1.40 V on the cathodic scan. The potential window is chosen to be a 200-mV window centered at E,. The lowest curve shows the increase in anodic current in the potential window from 0.55 to 0.75 V recorded during the initial oxidative scan, indicating the presence of dopamine. Note that the two current plots are not parallel: the O2 signal responds more quickly than the dopamine signal, in terms of both the time of initial response and the time to rise to full response. Since both analytes are transported by convective flow a t the same rate, the difference in response times is a direct reflection of the time to cross the Nafion membrane. (The initial current deflection is simultaneous at an uncoated electrode.) The
0
.............
i ................
.............
1
1
2
Time
I 3
4
5
I
(0)
Figure 2. Response of a Nafion-coated carbon-fiber electrode to a simultaneous step change in dopamine and 02.Cyclic voltammograms were collected at 100-ms intervals; each point represents the result from a single voltammogram. Solution and concentrations are the same as in Figure 1. The O2profile (A) was obtained by plotting the current measured in a window from -1.20 to -1.40 V on the cathodic scan. The dopamine plot (C) was constructed similarly with data from 0.55 to 0.75 V on the initial anodic sweep of each cyclic voltammogram. The middle trace (B)is plotted as a control based on data from 0.00 to 0.20 V on the initial anodic scan. The dashed line is the time at which 0, first deviates from the baseline. Triangle = 90% response.
anionic sites in the polymer, while promoting the partitioning of the hydrophobic cation dopamine into the film, also impede its progress through the film (D= 1 X cm2/s; see refs 7 and 21). In contrast, the polymer is relatively permeable to O2 (D = 2 X lo4 cm2/s; see ref 13). The current in Figure 2B, obtained over the potential range 0 . 0 . 2 V on the initial anodic scan, indicates the stability of the background and the specific nature of the changes seen at the other two potential ranges. The peak current for O2 was independent of flow rate in a range from 0 to 3.82 mL/min. The variation in i, was 5.8% (rsd) and did not correlate with the flow rate (n = 10). Sensitivity and Selectivity. Calibration curves for dopamine and O2 are linear with r > 0.99 for concentrations in a range of physiological interest: up to 10 pM for dopamine and up to 250 pM for 02.Typical sensitivities are, for dopamine, 0.2 nA/pM and, for 02,0.04 nA/pM. The greater sensitivity for dopamine is a consequence of the large value of its partition coefficient into Ndion (21),as well as possible adsorption of dopamine at carbon (9). In vitro detection limits are approximately 200 nM for dopamine and 5 pM for 02, determined from a single current vs time curve, and indicate the concentration of analyte necessary to produce a current change whose amplitude is 3 times the rms variation in the background signal. Detection limits can be improved by signal averaging of repetitive events (22). Due to the wide separation between the values of E, used for quantitation (see Figure l), interference of dopamine with O2 detection or vice versa is not normally encountered, except when low micromolar levels of dopamine are assayed in a solution nearly saturated (1mM) with 02.This interference is apparent in the cyclic voltammogram, preventing mistaken assignment of the signal to a change in dopamine. Several compounds known to be present in unbound form in the mammalian brain were tested for interference in the range 0.0 to -1.6 V a t 400 VIS. Most are anions not expected to penetrate the Ndion film. None of the following produced a detectable change from background current a t concentrations 2100 pM: adenosine, glutamic acid, glutathione, lactic acid, nicotinamide adenine dinucleotide, and uric acid. Although most of these are not reducible substances and would not interfere faradaically, they could interfere by altering the large background current. Changes in pH may occur in physiological experiments, and these could affect both of the redox reactions of interest, since both are associated with proton transfer. In particular, the
ANALYTICAL CHEMISTRY, VOL. 63, NO. 1, JANUARY 1, 1991
deprotonation of dopamine that occurs prior to the rate-limiting, one-electron oxidation should lead to a shift in the peak potential of 60 mV/pH unit. T o test for the pH sensitivity of the voltammograms, the buffer pH was changed from 7.4 to 7.1. No change was observed in the potential of the peak current for any of the three peaks considered here within the 25-mV resolution of the voltammetric scan. Changes in background current due to reduction of H+occur at potentials negative of -1.4 V for this range of pH shift, so this potential sets the negative limit for O2 measurements. Simultaneous Detection in Vivo. Electrodes were implanted in the rat brain in a region known as the caudate nucleus, an area that contains a large concentration of dopamine nerve terminals (23). Stable voltammetric responses were obtained, but the current at the potential for O2 reduction was found to vary with electrode position. O2 concentration in the extracellular fluid of the brain is a dynamic balance between supply of O2 via blood flow in capillaries and O2 consumption associated with respiration and metabolism in cells (24). Dissolved O2 concentrations form a gradient which is highest at the capillary wall and lowest a t the site of metabolic consumption. The range measured with implanted microelectrodes is from
