Anal. Chem. 1995,67, 1115-1120
Fast-Scan Cyclic Voltammetry of 5-Hydroxytryptamine Brad P. Jackson, Susanne M. Dietz, and R. Mark Wightman* Department of Chemistry, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina 27599-3290
Fast-scan cyclic voltammehy, a demonstrated analytical method for the in vivo detection of dopamine, is extended to the detection of in vitro and in vivo 5-hydroxyt1yptamine (5-HT) with the use of a specific potential wave form applied at 1000 V/s. The wave form, 0.2 to 1.0 to -0.1 to 0.2 V, is employed to accelerate electrode response limes which are sigdicantly slower with other wave forms due to the adsorption of 5-H.". The scan rate of 1000V/s enables follow-up reactions which lead to the formation of strongly adsorptive products to be outrun. The peak current at a carbon fiber disk microelectrode exposed to 1 pM 5-HTin flow injection experiments is 1 nA, with a half-rise time of less than 200 ms. The peak current of Nafion-coated electrodes exposed to the same concentration of 5-HT is 5 nA,with a half-rise time on the order of 400 ms. The rate of adsorption of 5-HTwas determined to be 4.22 f 0.33 s-l. Several compounds present in brain tissue as well as the pharmacological agents used to elicit 5-HT release in the caudate of the rat were evaluated. Those which gave a response could be differentiated from 5-HTon the basis of respective oxidative and reductive peak potentials. Nafion-coatedelectrodes were used to monitor transient increases in both dopamine and exogenous 5-HTin the caudate of the anesthetized rat in response to electrical stimulation. The rate of cellular uptake of 5-HTwas shown to be %fold slower than dopamine uptake. 5Hydroxytryptamine (5W,also known as serotonin, is an easily oxidized molecule which serves as a neurotransmitter in the mammalian brain. Although its precise neurochemical role is still the subject of research, an imbalance in its normal role has been associated with various mental disorders such as depression.' Indeed, the major action of the recently introduced antidepressant Prozac is to blockade 5HT reuptake into the nerve terminals from which it was secreted.2 Because 5HT can be oxidized electrochemically,3carbon fiber microelectrodes should be able to serve as in vivo sensors for this molecule in much the same way that these electrodes are used for the detection of another neurotransmitter, d ~ p a m i n e .However, ~ the oxidation products of 5HT and other hydroxyindoles tend to be very reactive. At conventional scan rates, an insulating film forms on (1)Aghajanian, G. IC; Wang, R Y. In Psychophamacologv: A Generation of Progrw; Lipton, M. A, DiMascio, A, Killam, K F., Eds.; Raven Press: New York, 1978;pp 171-183. (2) Lemberger, L.; Rowe, H.; Carmichael, R; Oldham, S.;Homg,J. S.; Bymaster, F. P.; Wong, D. T. Science 1978,199,436-437. (3)Stamford, J. A 1.Neurosci. Methods 1986,17,1-29. (4)Adams, R. N.Prog. Neurobiol. 1990,35, 297-311. 0003-2700/95/0367-1115$9.00/0 0 1995 American Chemical Society
the electrode after electrooxidation,thus preventing detection on subsequent scans.5 Large-scale electrolysis of hydroxyindoles in acidic solution reveals that a large number of products are formed, including hydroxylated products, dimers, and other species derived from radicals.6-8 Previously we have shown that cyclic voltammetry at 300 V/s is sufiiciently fast to prevent extensive follow-up reactions, thus avoiding the formation of an insulating film on the electrode? Nevertheless, the primary electroactive products and reactants also appear to adsorb and subsequently slow the response time, preventing the study of rapid concentration changes that occur during neurotransmission.*0The goal of this work is to eliminate the slow response time. We have reexamined the electrochemistry of 5HT with fastscan cyclic voltammetry at carbon fiber microelectrodes. We demonstrate that deleterious side reactions can be outrun at lo00 V/s and that the slow response times associated with adsorption can be avoided by judicious selection of wave form. The large background that exists at this scan rate is removed via digital background subtraction? In addition, we examine the effects of various potentidly interfering substances at microelectrodes which have been coated with a thin film of Nafion, a perfluorinated cationexchange polymer. The Nafion acts as a chemical filter that prevents access to the electrode for most anionic species while simultaneously allowing free access for amines such as 5HT, which is protonated at the physiological pH value of 7.4. Additionally, when used with the fast-scan technique, the Nafion restricts the diffusion layer to the interior of the film so that the measurements obtained in vivo can be converted to concentration on the basis of in vitro calibration curves.11 The utility of this method is demonstrated with the measurement of dynamic concentration changes of 5HT in the brain of an anesthetized rat. EXPERIMENTAL PROCEDURES Electrodes. A single carbon fiber (Thornell P55,5pm radius; Amoco, Greenville, SC) was sealed with an epoxy resin (Epon 828 plus 14%(w/w) m-phenylenediamine hardener; Miller-Steven(5) Ewing, A G.; Dayton, M. A; Wightman, R M. Anal. Chem. 1981,53,18421847. (6)Wrona, M. Z.;Lemordant. D.; Lin, L.; Blank, L.;Dryhurst, G.J. Med. Chem. 1986,29,499-505. (7)Dryhurst, G.;Anne, A; Wrona, M. Z.; Lemordant, D. /. Am. Chem. Soc. 1989,I l l , 719-726. (8) Verbiese-Genard, J. C.; Kauffman, J. M.; Hanocq, M.; Molle, L.]. Electmanal. Chem. Interfacial Electrochem. 1984,170,243-254. (9)Baur, J. E.; Kristensen, E. W.; May, L. J.; Wiedemann, D. J.; Wightman, R M. Anal. Chem. 1988,60, 1268-1272. (10)Ganis,P. A;Wightman, R M.J. Neurosci. 1994,14, 442-450. (11) Wiedemann, D. J.; Tomusk, A. B.; Wilson, R L.; Rebec, G. V.; Wightman, R. M.]. Neurosci. Methods 1990,35, 9-19.
Analytical Chemistv, Vol. 67, No. 6, March 15, 1995 1115
son, Danbury, 0in the tip of a tapered micropipet12prepared from glass capillaries (AM Systems,Everett, WA). The electrode surface was polished on a beveling wheel (K. T. Brown type;Sutter Instrument Co., Novato, CA) at a 25" angle, producing an elliptical electroactive surface with a minor radius defined by the fiber and a major radius defined by the polishing angle. As noted in the text, some working electrodes were dip coated with a thii layer of 2.5%Nation (Aldrich, Milwaukee, WI). In all instances, the reference electrode was a sodium-saturated calomel electrode (SSCE), and the auxiliary electrode was a nichrome wire. Cyclic Voltammetzy. Cyclic voltammograms were repetitively recorded every 100 ms at various scan rates utilizing an E1400 potentiostat (Ensman Instrumentation, Bloomington, IN). Locally written software controls the triggering and acquisition parameters via a commercial interface board (Labmaster; Scientik Solution, Solon, OH). Applied wave forms were produced by the triangle wave form generator internal to the EI-400 or by a function generator (Model 52004 Krohn-Hite, Avon, MA). The software also provides for background subtraction of the nonfaradaic residual current from the cyclic voltammograms and for the averaging of multiple scans.13 Background subtraction was accomplished with the use of a specific number of scans (set in the program) collected immediately prior to exposure to the analyte in the flow cell or prior to the initiation of the stimulus in vivo. Usually 50 scans were averaged to provide the background, and a time window (normally 2 s or 20 scans) during concentration changes was used. All voltammograms shown in this work are background subtracted. Electrode characterization and calibration were conducted in a flow injection system with the electrode positioned in the exit tube of a loop injector. The buffer, 150 mM sodium chloride and 20 mM HEPES at pH 7.4, was delivered at a rate of 2 mL/min. The loop injector is activated by a pneumatic actuator under software control so that the time and duration of the delivery of the electroactive compound can be reg~lated.~ Cyclic voltammograms were recorded continuously at 10@ms intervals for 5 s before and 10 s during and after the injection of compound into the flowing buffer. In Vivo Techniques. Adult male Sprague-Dawley rats (300550 g) were purchased from Charles River (Wilmigton, MA). Food and water were provided ad libitum, and animal care was in accordance with the Guide for Care and Use of Laboratory Animals (NIH Publication No. 86523; Bethesda, MD). The surgical procedure is described in detail e1se~here.l~ Rats were anesthetized with urethane (1.5 g/kg ip) and immobilized in a stereotaxic apparatus (Narishige,Tokyo, Japan). Body temperature was maintained by isothermal heating pads (Deltaphase isothermal pads; Braintree Scientific Inc., Braintree, MA). Holes were drilled in each rat3 skull for placement of working, reference, and stimulating electrodes. A salt bridge was used for electrical contact between the reference electrode and dura mater. The carbon fiber working electrode was placed in the caudate putamen (coordinates according to the atlas of Paxinos and (12) Kawagoe, K T.; Zimmerman, J. B.; Wightman, R M. J. Neurosci. Methods
1993,48,225-240. (13) Wiedemann. D. J.; Kawagoe, IC T.: Kennedy, R T.; Ciolkowski, E. L.; Wightman, R. M. Anal. Chem. 1991,63, 2965-2970. (14) Garris, P. A: Collins, L. B.: Jones, S. R; Wightman, R M. J. Neurochem. 1993,61,637-647.
1116 Analytical Chemistry, Vol. 67, No. 6, March 75,7995
Watson15were 1.2 mm anterior, 2.0 mm lateral, and 4.5 mm ventral to bregma). A stainless-steel, twisted, bipolar stimulating electrode (Plastics One, Roanoke, VA) was placed in the medial forebrain bundle (4.6 mm posterior, 1.6 mm lateral, and 7.5-9.0 mm ventral to bregma). Dopamine neurons were selectively ~timulated'~ by an electrical pulse train that was optically isolated (NL800stimulus isolators, Neurolog; Medical Systems Corp., Great Neck, NY) and timesynchronizedto the cyclic voltammetric data acquisition. The stimulus was a biphasic square wave (2 ms each phase) of 300-350 pA amplitude in trains of 120-180 pulses at a frequency of 60 Hz. Determination of Tissue Content. After in vivo measurements, the animal was decapitated and the brain removed. Microdisks were cut from 2-mm-thick brain slices chilled in iced buffer. To coincide with the locations of the cyclic voltammetric measurements,punches of striatum 3 mm in diameter were taken from the caudate putamen from 0 to 2 mm anterior to bregma according to the atlas of Paxinos and Watson.15 The tissue was homogenized and centrifuged, and 100 pL of supernatant was injected onto a reverse-phase HPLC column with amperometric detection at +0.6 V vs SSCE. The mobile phase consisted of 27 mM citrate, 94 mM sodium phosphate, 1.9 mM sodium octyl sulfate, 0.14 mM EDTA, and 20%methanol at a pH of 4.02. Tissue proteins were determined using a Bio-Rad assay kit (Richmond, CA) . Reagents. R04-4602, a peripheral decarboxylase inhibitor,was a gift from Val Wagner of Hoffman-LaRoche. All other chemicals were purchased from Sigma Chemical Co. (St. Louis, MO) and were reagent grade and used as received. Solutions were prepared using doubly distilled, deionized water (Mega Pure system; Coming Glasswork, Coming, NY) . 10 mM catecholamine and indolamine stock solutions were prepared in 0.1 N perchloric acid and diluted to the desired concentration with physiological buffer. All drugs, including urethane, were dissolved in physiological saline. RESULTS AND DISCUSSION
characterization of 5-HTCyclic Voltammeby. Backgroundsubtracted cyclic voltammograms of 1pM 5HTrecorded at the same electrode at three different scan rates in the flow injection system are shown in Figure 1. The applied potential wave form consisted of an anodic sweep from the rest potential of -0.4 V to 1.0 V, followed by a return to the rest potential. The scans were repeated at 10@msintervals, and the cyclic voltammograms are the average of 20 scans obtained during exposure to 5HT. This averaging protocol has been successfully used for the detection of dopamine in v i ~ 0 . IAt ~ 100 V/s, two oxidative waves and a reductive wave and a portion of a second wave can be observed. The data suggest two electrochemical couples, one with an oxidative peak at approximately 0.5 V and a reductive peak at approximately 0.0 V, and the other with an oxidation at 0.0 V and its corresponding reductive peak at potentials more negative than -0.4 V. At 1000 V/s, the more negative couple is less apparent, with the remaining oxidative wave shifted to approximately 0.6 V and its reductive wave shifted to approximately -0.1 V. The current amplitude for the oxidative wave at 1000 V/s is l@fold greater than that at 100 V/s, which suggests that the peak amplitudes are controlled by adsorption. (15) Paxinos. G.; Watson, C. The Rut Brain in Stereotaxic Coordinates;Academic Press: New York. 1986.
100 v/s
h
a
300 V/s
1000 v/s
0.2
2
0.1
1
0.0
0
T
v
0.5 t o 0.7 V 0.15 t o 0.25
v 1
C v
2
t
E
5
-0.1
-1
-0.2
-2
:fr ,
,
-3 1.0
0.5 0.0
0.5 0.0 E (V vs. SSCE)
0.5 0.0 -0.5
Figure 1. Background-subtracted cyclic voltammograms of 1 pM 5-HT recorded at the same carbon fiber microelectrode at three different scan rates (100, 300, and 1000 V/s). Each cyclic voltammogram is the average of 20 scans recorded every 100 ms. The
applied potential wave form was a triangle wave with a rest potential of -0.4 V and a maximum potential of 1.O V. Crosses designate the beginning of the scan. The amplitude of the peak current (ip) of a reversible cyclic voltammetric wave is given by
ip = (2.69
1 0 5 ) n 3 / 2 ~ ; / 2 v 10/*2 ~
(1)
where n is the number of electrons per mole of species electrclyzed, A is the electroactive surface area of the electrode, DOis the diffusion coefficient of 5HT, v is the scan rate, and CO*is the concentration of 5HT in bulk solution. The peak current calculated for the oxidative peak determined at 300 V/s, assuming a twoelectron oxidation, is only 3%of the measured current. This further supports the conclusion that the faradaic response is dominated by adsorption. Temporal Resolution of the Cyclic Voltammetric Couples. To clarify the origin of these peaks, cyclic voltammograms recorded at 300 V/s were examined at different times after introduction of 5HT to the electrode surface by flow injection analysis (Figure 2). The average currents from 0.15 to 0.25 V and from 0.5 to 0.7 V on the initial positive going scan from each successive cyclic voltammogram during exposure to 1,uM5HT are plotted. The current increases more rapidly at the more positive potential window at initial times after exposure (see inset of Figure 2). Indeed, a cyclic voltammogram F i r e 2A) recorded 1.7 s after the opening of the loop (the transport time to the electrode is approximately 1.5 s) shows a single oxidative wave. Neither current trace reaches a steady-statelevel, although other, more well-behaved compounds such as ascorbate exhibit a rectangular response reaching and returning from steady-state levels instantaneously in this system? At the time when the current at both potential windows of the cyclic voltammograms is at a maximum (10 s), the cyclic voltammogram (Figure 2B) clearly indicates a second, more negative, oxidative wave. When the 5HT is removed by the flowing buffer, the current at the more positive potential window declines more quickly than the other trace, but at a slower rate than well-behaved species would exhibit. A cyclic voltammogram Figure 2C) recorded 20 s after introduction, 10 s after removal, of 5HT shows that both couples are of similar amplitude. Note that the current from the more negative potential window remains at 60%of its maximal response. These data suggest the following scheme. At pH 7.4, the oxidation of 5HT results in the wave which appears at 0.6 V at
,
,
11.3 1.5 1.7 1.9 I
I
I
U
CI
\ IV
‘7 r;:
Figure 2. Comparison of normalized current vs time responses of the two oxidative peaks, 0.15-0.25 V and 0.5-0.7 V, obtained during the introduction of a bolus of 1 pM 5-HT in the flow injection system (-0.4 to 1.0 V, 300 V/s). Each point on the trace is from a single cyclic voltammogram repetitively recorded every 100 ms. The inset expands the 1.2-2.0 s time scale of the current response. Cyclic voltammograms A, B, and C are single scans recorded at 1.7, 10, and 20 s, respectively, after introduction of 5-HT into the flow system.
Scheme 1
300 V/s. The product of the oxidation, based on prior work,6s7 appears to be the pquinone imine, which can be reduced at 0.0 V at this scan rate (Scheme 1). The large amplitude of the 5HT oxidation peak and its failure to return completely to baseline when 5HT is removed from the buffer indicates that it is strongly adsorbed. The other couple appears to be a product from a subsequent chemical reaction since it is less apparent at the highest scan rate tested. At lower scan rates, this product appears to arise from material accumulated on the electrode since it grows in amplitude with time. Furthermore, the reduced form of this product is strongly adsorbed because its time for removal from the electrode is much slower than that of 5HT itself. The product may arise from the addition of water to the phenylene ring. Indeed, such a reaction mechanism has been proposed for the electrooxidation of 5HT in acidic solution.6-8 In any case, the slow adsorption of 5HT and the formation of a more strongly adsorbed product which slowly desorbs are both undesirable features of an analytical scheme for the detection of rapid concentration changes of the molecule. Alternative Wave Forms. To avoid complications of the follow-up chemical reaction, all of the following experiments employed a scan rate of 1000 V/s. Alternative wave forms were investigated in an attempt to increase the rate of change in the oxidation current seen with 5HT. Figure 3 shows the temporal response of the current at 0.5-0.7 V on the positive scan for 1 pM 5HT utilizing three different applied potential wave forms. When the wave form used in Figures 1 and 2 is used at the faster scan rate, the averaged cyclic voltammogram (Figure 3B) shows only one couple, but the current from successive cyclic voltammograms still indicates a slowed response due to adsorption Figure 3A). The current shows a signilicantly improved response when the initial (and rest) potential of the applied wave form is altered to f0.2 V (Figure 3C). However, the reductive wave, Analytical Chernistty, Vol. 67, No. 6,March 15, 1995
1117
10
0
0
10 Time (.5 .)
0
10
20
+
+
-2.0+ 1.0
0.5
0.0
100
I 0.8 0.6 0.4
0
0
0.8 0.4 0.0
E (V v s . SSCE)
Figure 3. Temporal response of the oxidation current (A, C, and E) and the corresponding averaged cyclic voltammograms (e, D, and F) of 1 pM 5-HT utilizing three different potential wave forms applied at 1000 VIS at a carbon fiber microelectrode (A,B, -0.4 to 1.O V; C,D, 0.2 to 1.0 V; E,F, 0.2 to 1.0 to -0.1 to 0.2 V). Small squares on the temporal current response indicate opening (0 s) and closing (10 s) of the loop injector.
which aids in the identification of the detected substance, is not present (Figure 3D). The results in parts E and F of Figure 3 were obtained with a wave form with a rest (and initial) potential of 0.2 V as well, but in this case the negative going scan was extended to -0.1 V before its return to the rest potential. With this wave form, the response to a concentration change is rectangular, as found with other nonadsorbing compounds such as ascorbate, and the cyclic voltammogram exhibits both oxidative and reductive waves for 5HT. The more positive rest potential appears to accelerate adsorption and desorption of 5HT from the carbon surface. With the more positive rest potential, the 5HT is driven to the oxidized form, suggesting that the accumulation of the initial reduced form is responsible for the behavior seen in Figure 3A. Note, however, that adsorption of 5HT still occurs since the peak current for the oxidative wave predicted by eq 1is only 5.7%of the measured current. Rates of adsorption of 5HT when the two m o a e d wave forms are used were calculated. For a stationary surface in a stirred solution, equivalent to the conditions in a flow cell, mass transportcontrolled adsorption can be described by
where r(t) is the time-dependent surface excess and k is the overall rate of adsorption.16 Therefore, if the measured peak adsorption current (i(f)) is assumed to be proportional to the surface excess, the data should be linear when plotted according to a rearranged form of eq 2: ln(1 - i(t)/imw)= -kt
(3)
Data collected with both wave forms were plotted in logarithmic fashion according to eq 3 and were found to be linear (+ = 0.99 for both). The rate of adsorption of 5HT when the 0.2-1.0 V (16) Stamford, J. A; Hurst, P. R; Kuhr, W. G.; Wight", Chem. 1989,265, 291-296.
100
200
300
400
1 5 - m (M Figure 4. Calibration curves for 5-HT determined using the three applied potential wave forms given in Figure 3 (0,-0.4 to 1.O V; v, 0.2to 1.O to -0.1 to 0.2 V; v, 0.2 to 1.O V). The dashed line at 600 amol represents monolayer coverage on the electrode surface.
triangle wave was used was 3.36 f 0.22 s-l. The use of the 0.2 to 1.0 to -0.1 to 0.2 V wave form resulted in a k of 4.22 f 0.33 S-1.
Response of Carbon Fiber Microelectrodes. Calibration curves for 5HT are linear (+ > 0.99) for concentrationsup to 10 pM when the applied potential wave form utilized in Figure 3E,F is employed. Figure 4 shows calibration curves obtained at higher concentrationsusing the three previously discussed wave forms scanned at 1000 V/s. The dashed line at 600 amol represents monolayer coverage on the electrode surface, a value approximated using the dimensions of the minimized planar structure of 5HT obtained with Chem 3D (software package from Cambridge Scientific) and the surface area of the carbon fiber. The value is also consistent with literature values for similar compounds adsorbed on platinum surface^.'^ At concentrations 1400 pM, monolayer coverage is not achieved using any of the wave forms. At concentrations > 400 pM, the current due to oxidation actually decreases compared to that at lower concentrations. This suggests a polymerization of the oxidized products, which causes blocked surface sites on the carbon electrode. Calibration curves for 5HT are linear at low concentrations (up to 10 pm) with the wave form employed in Figure 3E,F. Typical sensitivities of the electrodes for 5HT are 1 nA/pM compared to 0.1 nA/pM for dopamine, which also adsorbs? Average in vitro response times, the time required to reach half the maximal current, range from 75 to 175 ms. Sensitivity and Selectivity of Nation-Coated Electrodes. Microelectrodesused in vivo are often coated with a thin film of the cation exchanger Nafion. Calibration curves obtained with Nafion-coated electrodes are also linear (7.2 > 0.99) for 5HT concentrationsup to 10 pM with the applied potential wave form employed in Figure 3E,F. Typical sensitivities obtained from calibration curves are greater for 5HT with Nafion and are approximately 5 nA/pM. The Nafion-coated electrodes are, on average, 20 times more sensitive to 5HT than dopamine. This is especially important when considering that the signal-tenoiseratio is decreased between 1.4 and 2.7 times when scanning at 1000 V/s as compared to scanning at 300 V/s. This is in part due to the >J-fold increase in background current from 300 to 1000V/s
R M. J. Electroanal.
1118 Analytical Chemistry, Vol. 67, No. 6, March 15, 1995
(17) Soriaga. M. P.; Hubbard, A. T. J. Am. Chem. SOC.1982, 104, 3937-3945.
/ \
++.lo 1 .oo
F==T!=7
_ - -5-HT
$1
1.00
l.@$eIO
'
, -AMPT
J
-DA
0.45 -0.10 Ro4-4602
-
d 1.00
,45
1.0
-
-0.101.00
-5-HIAA
l e 5-HTP ! ol*.
E (V vs. SSCE)
1.00
,g; ., post-cal
1 .OO 0.45 -0.1 0 E (U vs. SSCE)
I I 1 .OO 0.45 -0.10
0.45 -0.10 Uric Acid
0.45
-0.10
NADH
Figure 5. Background-subtracted,averaged cyclic voltammograms of compounds with the potential for interfering with the in vivo 5-HT signal recorded at a single Nafion-coated electrode. All compounds are lOOpM, except fordopamine (DA, 10pM) and 5-HT (1 pM)..Cyclic voltammograms were recorded with the 0.2 to 1.0 to -0.1 to 0.2 V potential wave form applied at 1000 V/s. All cyclic voltammograms are on the same current scale. Dashed cyclic voltammograms are of 5-HT.
and the higher filter frequency required at 1000 V/s. To prevent overfiltering of the data, the 2.WkHz filtering frequency used at 300 V/s must be increased to at least 6.00 kHz when scanning at 1000 V/s. In vitro response times typically range from 275 to 400 ms as a result of permeation through the Nafion film. Detection limits in vitro are approximately 1 nM, with signal averaging (20 scans) of repetitive events. Several compounds found in the mammalian brain as well as the pharmacological agents used to evoke 5HT during in vivo experimentation were tested for interference with 5HT at Nationcoated electrodes. The results are shown in Figure 5. These cyclic voltammograms were recorded utilizing the 0.2 to 1.0 to -0.1 to 0.2 V wave form applied at a rate of 1000 V/s. The peripheral decarboxylase inhibitor R04-4602, uric acid, and the reduced form of nicotinamide adenine dinucleotide produce no detectable change from background current over the potential range -0.1 to 1.0 V at 100 pM. a-Methyl-ptyrosine (AMIT), an inhibitor of dopamine synthesis at the level of tyrosine hydroxylase,ls penetrates the Nafion film and produces a cyclic voltammetric signal at a concentration of 100pM. However, the AMFT signal can be easily distinguished from that of 5HT by its significantly more positive E,,,,, and the lack of a reductive peak. 5Hydroxytryptophan (5H",100pM), the immediate precursor to 5HT, is, like AMFT, a zwitterion and also elicits a signal which could possibly interfere with 5HT. The E,,,,, of 5H" is very close to that of 5HT, leaving the difference in E,,& of the two compounds as the best means of differentiation. Surprisingly, 100 pM Shydroxyindoleacetic acid (SHIAA), an anion at physiological pH and the major in vivo metabolite of 5HT, also produces a signal at a Nation-coated electrode. However, the amplitude of the signal is only 10%of the amplitude generated by the same concentration of 5HIAA at an uncoated electrode. Since 5HT and its metabolite have the same E,,o,, the only means of differentiationis on the basis of their respective Ep,red values. The E p , r e d of 5HIAA is signilicantly more negative than that of 5HT. Similarly, dopamine and 5-HT must be d~erentiatedby their respective reductive peaks since their E,,o, values overlap. (18) Brodie, B. B.; Costa, E.; Dlabac, A; Neff, N. H.: Smookler, H. H.J. Phanacol. Ex$. mer. 1966,154, 493-498.
i\ Figure 6. Response of the same Nafion-coated electrode to the stimulated release of dopamine (A) and 5-HT (B) after pharmacological manipulation (see text) at the same coordinates in the caudate putamen of the rat. Stimulations (180 pulses, 60 Hz, 3 s) were applied to the medial forebrain bundle. Background-subtracted cyclic voltammograms are shown in comparison to cyclic voltammograms obtained during postcalibration of the electrode. The cyclic voltammograms are an average of multiple single scans (scans designated by large boxes). Small boxes mark the beginning and end of the applied stimulation.
Thus, while concentration changes measured at the potential for oxidation of 5HT could be due to several substances, the voltammogram gives a clear indication of the substance(s) detected. In Vivo Detection. The detection of in vivo 5HT in this work is modeled after an experiment reported by Stamford et In their experiment, striatal dopaminergic terminals are forced to release 5HT after loading with the precursor SHTP. First, they used electrical stimulation of the medial forebrain bundle to evoke release of dopamine from dopaminergic terminals in the caudate nucleus, the site of the working electrode. The caudate nucleus, a major dopaminergic terminal field, contains only one-tenth as many 5HT terminals.2O AMPT was then administered to deplete the releasable stores of dopamine. Subsequent administration of 5HTP lead to the stimulated release of 5HT. The 5HT is presumably formed from the decarboxylation of 5H" by nonspecific amino acid decarboxylase contained in dopamine neurons. To ensure that a majority of the 5HTP reached the brain, Ro4-4602 was administered to inhibit peripheral decarboxylase enzymes. This remarkable result suggests that dopamine neurons can be pharmacologically altered to release 5HT. We replicated the results of Stamford et a1.6 using the new cyclic voltammetric scheme designed for detecting 5HT. Figure 6A shows the time course and the averaged cyclic voltammogram obtained during electrical stimulation of the medial forebrain bundle before administration of pharmacological agents. The voltammogram clearly identifies the detected substance as dopamine, and the duration of the concentration change is similar to that observed in previous studies.21 The transient increase in dopamine concentration returns to prestimulus levels due to cellular uptake after termination of the stimulation. The initial rate of disappearance of dopamine was taken as the Vmaxfor (19) Stamford, J. A; Kmk, 2. L.; Millar, J. Brain Res. 1990,515, 173-180. (20)Saavedra, J. M. Fed. PYOC.1977,36, 2134-2141. (21) Wightman, R. M.; Zimmerman, J. B. Brain Res. Rev. 1990,15, 135-144.
Analytical Chemistry, Vol. 67,No. 6,March 15, 1995
1119
uptake. Uptake of dopamine in the caudate nucleus has been shown to follow Michaelis-Menten kinetics, with a K, of 0.2 , U M . ~ ~ Since the concentration of dopamine in the extracellular fluid immediately after stimulation exceeds the K, by Dfold, this is a reasonable approximation. The V , , of dopamine was determined to be 3.53 i0.28 yMIs. After administration of AMPT, the evoked dopamine signal decreased below background noise levels.23 Postmortem tissue content was determined with high-performance liquid chromatography after treatment with the pharmacological agents (Table 1). The amount of dopamine in the caudate nucleus after administration of AMPT and subsequent stimulation is 17%of predrug levels.'O This amount represents a slightly greater reduction from control than reported in the literat~re?~ which may be due to the longer time scales of the experiments described here. The amount of 5HT in the caudate nucleus after administration of 5-HTP increases to 297%of predrug levels.20 This result is consistent with increases in extracellular concentrations of 5HT measured with microdialysi~.~~ Following administration of 5HTP, the time course and averaged cyclic voltammogram shown in Figure 6B were obtained. The stimulating and working electrodes were kept at the same stereotaxic coordinates. The cyclic voltammogram identiiles the released compound as 5HT, which is consistent with the chromatographic results. From the temporal concentration response, the rate of 5HT uptake appears to be significantly slower than the uptake of dopamine. An initial uptake rate of 1.32 f 0.11 pMls was determined from the data. This rate is not distorted by adsorption since adsorption and desorption occur on a much faster time scale (vide supra). The approximately 3-fold slower uptake rate of 5HT into dopamine neurons as compared (22) Hom, A S. Prog. Neurobiol. 1990,34, 387-400. (23) Ewing, A G.; Bigelow, J. C.; Wightman, R. M. Science 1983,221, 169171. (24) Westerink, B. H. C.; De Vries, J. B. J. Neurochem. 1991,56, 228-233. (25) Shaskan, E. G.; Snyder, S. H. J. Pharmacol. E@. mer. 1970,175, 404418. (26) Zimmerman, J. B.; Wightman, R M.Anal. Chem. 1991,63, 24-28.
1120 Analytical Chemisrfy, Vol. 67, No. 6, March 75, 7995
to dopamine is in close agreement with the diflerence in V,, values for uptake into isolated rat caudate tissue.25 As can be seen from the chromatographicdata in Table 1,all 5HTP is not converted into 5HT on the time scale of this experiment. This raises the possibility that stimulation is causing the release of 5HTP instead of 5HT. However, Stamford et al. showed that this was not the case by administering NSD 1015, a peripheral and central decarboxylase inhibitor.lg NSD 1015 inhibits the central nonspecific amino acid decarboxylase used by the dopamine neurons to form 5HT from SHTP. Upon inhibition, release of 5HT was not observed, indicating that 5HTP was neither released nor converted into 5HT. Cyclic voltammetric responses in vivo using the wave form employed in Figure 3E,F were stable for at least 6 h with little loss of sensitivity. Most decreases in sensitivity typically occurred immediately upon implantation. Following removal from tissue, current amplitudes are, on average, 50-60% of the preimplant response for dopamine26 and 65-75% for 5HT. CONCLUSIONS
Through the appropriate selection of applied potential wave form and scan rate, interferences from follow-up chemical reactions and adsorption can be avoided in the detection of 5HT employing fast-scan cyclic voltammetry. The technique provides a tool which allows for the determination of rates of adsorption in vitro and rates of uptake in vivo. ACKNOWLEDGMENT
Discussions with Jim Anderson are gratefully acknowledged.
This work was supported by the National Institutes of Health (NS 15841). Received for review October 19, 1994. Accepted January
8, 1995.@ AC9410225 @
Abstract published in Aduance ACS Abstracts, February 15, 1995.
