Subsecond detection of physiological adenosine concentrations using fast-scan cyclic voltammetry

Article abstract of DOI:10.1021/ac061820i

Adenosine modulates blood flow and neurotransmission and may be protective during pathological conditions such as ischemia and stroke. A real-time sensor of adenosine concentrations is needed to understand its physiological actions and the extent of receptor activation. Microelectrodes are advantageous for in vivo measurements because they are small and can make fast measurements. The goal of this study was to characterize detection of physiological adenosine concentration changes at carbon-fiber microelectrodes with subsecond temporal resolution. The oxidation potential of adenosine is +1.3 V, so fast-scan cyclic voltammetry (FSCV) was performed with an applied potential from -0.4 to 1.5 V and back at 400 V/s every 100 ms. Two oxidation peaks were detected for adenosine with T-650 carbon fibers. The second oxidation peak at 1.0 V occurs after the initial oxidation at 1.5 V and is due to a sequential oxidation step. Adsorption was maximized to obtain detection limits of 15 nM, lower than basal adenosine concentrations in the brain. The electrode was insensitive to the metabolite inosine and seven times more sensitive to adenosine than ATP. The enzymatic degradation of adenosine was monitored with FSCV. This microelectrode sensor will be valuable for biological monitoring of adenosine.

Full text of DOI:10.1021/ac061820i

Anal. Chem. 2007, 79, 744-750
Subsecond Detection of Physiological Adenosine
Concentrations Using Fast-Scan Cyclic
Voltammetry
B. E. Kumara Swamy and B. Jill Venton*
Department of Chemistry, University of Virginia, Charlottesville, Virginia 22904
Adenosine modulates blood flow and neurotransmission
and may be protective during pathological conditions such
as ischemia and stroke. A real-time sensor of adenosine
concentrations is needed to understand its physiological
actions and the extent of receptor activation. Microelec-
trodes are advantageous for in vivo measurements be-
cause they are small and can make fast measurements.
The goal of this study was to characterize detection of
physiological adenosine concentration changes at carbon-
fiber microelectrodes with subsecond temporal resolu-
tion. The oxidation potential of adenosine is +1.3 V, so
fast-scan cyclic voltammetry (FSCV) was performed with
an applied potential from -0.4 to 1.5 V and back at 400
V/s every 100 ms. Two oxidation peaks were detected for
adenosine with T-650 carbon fibers. The second oxidation
peak at 1.0 V occurs after the initial oxidation at 1.5 V
and is due to a sequential oxidation step. Adsorption was
maximized to obtain detection limits of 15 nM, lower than
basal adenosine concentrations in the brain. The electrode
was insensitive to the metabolite inosine and seven times
more sensitive to adenosine than ATP. The enzymatic
degradation of adenosine was monitored with FSCV. This
microelectrode sensor will be valuable for biological
monitoring of adenosine.
amount of adenosine available to activate receptors and the time
course of adenosine effects.
In vivo measurements are challenging because small quantities
of analyte must be rapidly detected in a chemically complex
sample. Microdialysis sampling coupled to HPLC has been the
most common method for monitoring chemical changes in vivo.
However, the method suffers from low temporal resolution and
tissue damage from implantation of the relatively large probes
(250-500 µm).7 Tissue damage causes changes in cellular
metabolism, and reported basal levels of adenosine from microdi-
alysis studies vary widely from 40 nM to 2 µM depending on the
time elapsed from probe implantation.1 Therefore, smaller probes
that cause less damage would be better for adenosine measure-
ments.
,8
Electrochemical sensors directly detect either redox-active
molecules or an electroactive product formed after an enzymatic
reaction. The direct electrochemistry of purines, such as adenine,
was characterized by Dryhurst.9 Electrochemical oxidation of
adenine proceeded in a series of 2-electron oxidations, with an
oxidation potential of 1.3 V for the first oxidation at neutral pH.
Similar mechanisms for oxidation of the nucleoside adenosine at
carbon electrodes have been elucidated.12 These studies used
traditional, large electrodes with slow scanning electrochemical
techniques that would not be amenable to making fast measure-
ments in tissue. An enzyme-based electrochemical sensor has also
been developed for adenosine that requires three enzymes to
-11
Adenosine is an important endogenous modulator, present in
all cells, that plays a role in the regulation of physiological activity
break down adenosine and produce hydrogen peroxide, which is
1
13,14
in various tissues and organs. Release of adenosine has been
detected by amperometry.
This electrode is sensitive, and a
1
5
shown to protect heart cells from damage during an ischemic
small, 25-µm microelectrode version has been fabricated, but it
suffers from interferences of adenosine metabolites, has a slower
temporal response (2 s to 1 min), and is complicated to fabricate
due to the necessity of three enzymes. An ideal sensor for
2
attack, where there is lack of oxygen delivery. In the central
3
nervous system, adenosine can regulate cerebral blood flow and
4
modulate neurotransmission. Because adenosine is a product of
ATP degradation, its release from cells can also be a sign of a
high metabolic rate or metabolic stress. Four adenosine receptors
(
6) Collis, M. G.; Hourani, S. M. O. Trends Pharmacol. Sci. 1993, 14, 360-
5
366.
regulate a variety of adenosine functions with a range of affinities
from low nanomolar to tens of micromolar. Therefore, rapid
measurements of adenosine in vivo are needed to understand the
(7) Clapp-Lilly, K. L.; Roberts, R. C.; Duffy, L. K.; Irons, K. P.; Hu, Y.; Drew,
6
K. L. J. Neurosci. Methods 1999, 90, 129-142.
(8) Grabb, M. C.; Sciotti, V. M.; Gidday, J. M.; Cohen, S. A.; van Wylen, D. G.
J. Neurosci. Methods 1998, 82, 25-34.
(9) Dryhurst, G.; Elving, P. J. J. Electrochem. Soc. 1968, 115, 1014-1020.
*
To whom correspondence should be addressed. Phone: (434) 243-2132.
(10) Dryhurst, G. J. Electrochem. Soc. 1969, 116, 1411-1412.
(11) Dryhurst, G. Electrochemistry of biological molecules; Academic Press: New
York, 1977; pp 71-185.
Fax: (434) 924-3710. E-mail: bjv2n@virginia.edu.
1) Latini, S.; Pedata, F. J. Neurochem. 2001, 79, 463-484.
(
(
2) Villarreal, F.; Zimmermann, S.; Makhsudova, L.; Montag, A. C.; Erion,
M. D.; Bullough, D. A.; Ito, B. R. Mol. Cell Biochem. 2003, 251, 17-26.
3) Phillis, J. W. Cerebrovasc. Brain Metab. Rev. 1989, 1, 26-54.
4) Brundege, J. M.; Dunwiddie, T. V. Adv. Pharmacol. 1997, 39, 353-391.
5) Masino, S. A.; Dulla, C. G. Neurol. Res. 2005, 27, 149-152.
(12) Goyal, R. N.; Sangal, A. J. Electroanal. Chem. 2002, 521, 72-80.
(13) Dale, N. J. Physiol. 1998, 511, 265-272.
(14) Dale, N.; Pearson, T.; Frenguelli, B. G. J. Physiol. 2000, 526, -143.
(15) Llaudet, E.; Botting, N. P.; Crayston, J. A.; Dale, N. Biosens. Bioelectron.
2003, 18, 43-52.
(
(
(
744 Analytical Chemistry, Vol. 79, No. 2, January 15, 2007
10.1021/ac061820i CCC: $37.00 © 2007 American Chemical Society
Published on Web 12/15/2006

adenosine would be sensitive, small, easy to fabricate, and allow
rapid detection of adenosine changes.
were prepared in 0.1 M perchloric acid and diluted to the desired
concentration in tris buffer on the day of the experiment. The
tris buffer solution was (in mM) 15 tris(hydroxymethyl)ami-
Carbon-fiber microelectrodes have been used extensively as
in vivo sensors because of their small size, good electron-transfer
nomethane, 140 NaCl, 3.25 KCl, 1.2 CaCl
2 2 4
, 1.25 NaH PO , 1.2
16
properties, and ease of fabrication. Research has focused on rapid
detection of dopamine neurotransmission using fast-scan cyclic
MgCl , and 2.0 Na SO with the pH adjusted to 7.4. All aqueous
2
2
4
solutions were made by using deionized water (Milli-Q Biocel,
Millipore, Billerica, MA).
1
7
voltammetry. Brajter-Toth and colleagues have studied the
detection of adenosine at carbon-fiber microelectrodes.1 How-
ever, their peak oxidation potentials, between 0.9 and 1.2 V, were
less than expected for adenosine oxidation and their limits of
detection were only 2-5 µM, not sufficient for in vivo use. Limits
of detection under 40 nM will be necessary to monitor changes
in adenosine from basal levels.
8,19
Instrumentation and Electrochemistry. Fast-scan cyclic
voltammograms were collected using a GeneClamp 500B poten-
tiostat (Molecular Devices, Union City, CA with a custom-modified
headstage). The data acquisition software and hardware were the
same as described by Heien et al.20 Briefly, two computer interface
boards (National Instruments PCI 6052 and PCI 6711, Austin, TX),
were used to apply the triangular waveform and collect the
resultant current data through a breakout box. For detection of
adenosine, the electrode was scanned from -0.4 to 1.5 V and back
at 400 V/s every 100 ms. The reference electrode was a silver-
silver chloride electrode.
Flow Injection Analysis. The carbon-fiber electrode was
positioned at the output of a flow injection apparatus consisting
of a six-port, stainless steel HPLC loop injector mounted on a two-
position air actuator (Valco, Houston, TX).21 The sample was
loaded into a 500-µL sample loop before the experiment, and then
the air actuator would turn the valve to allow sample to flow by
the electrode. The air actuator was controlled by a solenoid valve
mounted to a digital voltage interface (Valco), which received
digital signals from the breakout box. The buffer was pumped
through the flow cell at 2 mL/min using a syringe pump (Low
RFI Syringe Pump 22, Harvard Apparatus, Holliston, MA). Three-
second injections of the compounds were made to mimic fast
concentration changes that occur in the brain.
The objective of this study was to develop a fast-scan cyclic
voltammetry method for detection of physiological levels of
adenosine at carbon-fiber microelectrodes. We found that, depend-
ing on the type of carbon fiber used, two or three successive
oxidation steps could be detected for adenosine. In addition, we
show that the kinetics of adenosine detection are primarily
adsorption controlled and optimizing adsorption is the key to low
nanomolar limits of detection. Finally, we demonstrate that the
negative holding potential associated with fast-scan cyclic volta-
mmetry allows higher sensitivities for adenosine than for adenos-
ine nucleotides such as adenosine monophosphate (AMP) and
adenosine triphosphate (ATP). The sensor was used to monitor
an enzymatic reaction degrading adenosine.
EXPERIMENTAL SECTION
Electrode Construction. Carbon-fiber microelectrodes were
fabricated by aspirating a single carbon fiber into a glass capillary
(1.2 mm × 0.68 mm, A-M systems, Carlsburg, WA). Either 6-µm-
diameter T-650 fibers or 10-µm P-55 fibers were used (Cytec
Engineering Materials, West Patterson, NJ). The capillary was
pulled to form two electrodes on a vertical pipet puller (Narishige,
model PE-21). The extended fiber was trimmed with a scalpel at
the glass/fiber interface under the microscope. The electrodes
were epoxied to obtain a seal between the fiber and glass. The
epoxy, Epon Resin 828 (Miller-Stephenson, Danbury, CT), was
mixed with 14% (w/w) m-phenylenediamine hardener (Fluka,
Milwaukee, WI) and heated to ∼80 °C. Electrodes were dipped
for 30 s in the epoxy and left overnight at room temperature. They
were cured in an oven at 100 °C for 2 h and then at 150 °C
overnight. The electrode surface was polished on a beveling wheel
Electrodes were cycled using the experimental waveform (-0.4
to 1.5 V, 400 V/s, 10 Hz) for 15 min before collecting any cyclic
voltammograms to stabilize the background current. Carbon-fiber
electrodes were first tested by injecting dopamine, a compound
22
with known electrochemistry. Electrodes were discarded if the
signal was less than 2nA for 5 µM dopamine or if the oxidation
peak was shifted more than 100 mV from the normal oxidation
potential (600 mV). Current versus time traces were obtained by
integrating the current in a 100-mV window centered around the
oxidation peak for each cyclic voltammogram. Cyclic voltammo-
grams were background-subtracted by averaging 10 background
scans taken directly before the compound was injected into the
flow cell.
Monitoring Enzyme Activity. Adenosine deaminase in 3.2 M
ammonium sulfate (Sigma) had an activity of 1.5 units/µL. The
stock enzyme was diluted 1:100 in tris buffer, and then 1 µL of
the diluted enzyme was added to 5 mL of 10 µM adenosine in
buffer. Aliquots were analyzed by flow injection analysis every
minute.
(K.T. Brown type; Sutter Instrument Co. model BV-10, Novoto,
CA) at an angle of 30°, producing an elliptical electroactive surface.
Electrical connection was made by backfilling the capillary with
a high ionic strength solution (4 M potassium acetate, 150 mM
potassium chloride). The electrodes were soaked in 2-propanol
for at least 10 min prior to use.
Chemicals. Adenosine, dopamine, ATP, AMP, guanine, and
inosine were purchased from Sigma-Aldrich (Milwaukee, WI), and
all other chemicals were purchased from Fisher Scientific (Su-
wanee, GA) and used as received. The 10 mM stock solutions
Statistics. All values are given as the mean ( standard error
of the mean (SEM). For each electrode, three replicates were
collected and averaged for each data point. Experiments were
(
16) Michael, A. C.; Wightman, R. M. In Laboratory Techniques in Electroanalytical
Chemistry, 2 ed.; Heineman, W. R., Kissinger, P. T., Eds.; Marcel Dekker:
New York, 1996; pp 367-402.
(20) Heien, M. L. A. V.; Phillips, P. E. M.; Stuber, G. D.; Seipel, A. T.; Wightman,
R. M. Analyst 2003, 128, 1413-1419.
(17) Venton, B. J.; Wightman, R. M. Anal. Chem. 2003, 75, 414A-421A.
(18) El-Nour, K. A.; Brajter-Toth, A. Electroanalysis 2000, 12, 305-310.
(19) Brajter-Toth, A.; El-Nour, K. A.; Cavalheiro, E. T.; Bravo, R. Anal. Chem.
(21) Kristensen, E. W.; Wilson, R. W.; Wightman, R. M. Anal. Chem. 1986, 58,
986-988.
(22) Venton, B. J.; Troyer, K. P.; Wightman, R. M. Anal. Chem. 2002, 74, 539-
2000, 72, 1576-1584.
546.
Analytical Chemistry, Vol. 79, No. 2, January 15, 2007 745

at higher potentials) in fast-scan cyclic voltammetry because of
the time necessary for electron transfer.23 Therefore, the peak at
1
.5 V is likely to correspond to the literature peak for adenosine
at 1.3 V.
To investigate the mechanism of formation of two peaks, the
current versus time traces were examined for a flow injection
analysis experiment (Figure 2A). The current at each oxidation
peak is plotted in the large graph and normalized current in the
inset. The current from the first oxidation peak at 1.5 V increased
before the current from the second oxidation peak at 1.0 V. For
example, when adenosine was first introduced to the electrode,
the current from the first peak (solid triangles) was clearly
elevated at 3.5 s while the current from the second oxidation peak
was still at baseline. This indicates that the second oxidation peak
is a secondary reaction and involves oxidation of a product made
on the first scan. Indeed, the cyclic voltammograms confirm this
because the first cyclic voltammogram after adenosine was
introduced to the electrode contained no secondary oxidation peak
since the precursor oxidized species was not present. However,
both oxidation peaks were present at 4 s, when the current was
near its maximum.
Figure 1. Example cyclic voltammetry of adenosine. (A) The applied
potential was scanned from -0.4 to 1.5 V and back at 400 V/s every
100 ms. (B) This rapid scanning produced a large background
charging current at the electrode (solid line). The current after 5 µM
adenosine was added (dashed line) changed only slightly. (C) An
example backround-subtracted cyclic voltammogram of 5 µM ad-
enosine at physiological pH 7.4, an average of 10 cyclic voltammo-
grams. Two oxidation peaks were detected: peak 1 at the switching
potential, 1.5 V, and peak 2 around 1.0 V. (D) An example cyclic
voltammogram of a low, physiological concentration of adenosine,
2
00 nM. The main oxidation peak at the switching potential was still
The switching potential was also varied to demonstrate that
the peak at 1.0 V is a secondary oxidation product (Figure 2B).
When the potential was extended to 1.5, two peaks were detected.
But when the potential was swept to only 1.3 V, neither peak was
observed, even though the switching potential was higher than
easily observable.
repeated using multiple electrodes, and error bars are plotted as
the SEM for the values from different electrodes (generally n )
1
.0 V where peak 2 appeared. This indicates that the second
8
). All statistics were performed in GraphPad Prism. Limits of
oxidation peak cannot be detected unless the primary oxidation
at 1.5 V occurs.
detection were defined as a S/N ratio of 3.
Adenine is known to undergo a series of 2-electron oxidations.11
Each product formed is easier to oxidize, so oxidation potentials
should be less positive for each successive step. These data
suggest the following scheme (Scheme 1). The first 2-electron
oxidation of adenosine (I) results in the peak at the switching
potential, 1.5 V. On subsequent scans, that oxidation product (II)
can be oxidized again by another 2-electron oxidation, which
results in the peak at 1.0 V. Some of II may also be oxidized on
the scan back down when the potential is still sufficient for
oxidation. Note that the current does not come back to 0 until
after 1.0 on the back scan in Figure 1C, indicating a species is
still being oxidized.
RESULTS AND DISCUSSION
Adenosine Cyclic Voltammetry. With fast-scan cyclic volta-
mmetry, concentrations are sampled by repeating scans at regular
intervals, and the current arises from analyte oxidation or
2
3
reduction. For adenosine detection, the electrode was ramped
from a resting potential of -0.4 to 1.5 V versus Ag/AgCl and back
at 400 V/s every 100 ms as shown in Figure 1A. These fast scan
rates produced a background charging current that was much
larger than faradaic currents. The background current of a T-650
disk electrode with (dashed) and without (solid) 5 µM adenosine
is shown in Figure 1B. The background current is relatively stable
over time and can be subtracted out, resulting in a background-
subtracted cyclic voltammogram for 5 µM adenosine (Figure 1C).
Two oxidation peaks were observed, the main oxidation near the
switching potential at 1.5 V (peak 1) and a secondary oxidation at
Our fast scan data are different from slow-scan cyclic voltam-
mograms that contain only one oxidation peak, present at 1.3 V,
-
for a 6-e oxidation of adenosine. However, with slow-scan rates
of 100 mV/s, the electrode would be at a potential sufficient to
oxidize adenosine for several seconds. With fast-scan cyclic
voltammetry, the electrode is at a potential sufficient for the first
step of oxidation for only 0.5 ms each scan. Insufficient time for
the sequential reactions to occur on that scan would allow
detection of a separate peak on the subsequent scan. Indeed,
complete coulometry of adenosine at a pyrolitic graphite electrode
took nearly 2 days to complete, indicating that full oxidation can
be slow.12
∼
1.0 V (peak 2). The oxidation appeared irreversible with no
detectable reduction peaks. Detection of a low, physiological
concentration, 200 nM, is shown in Figure 1D. The main oxidation
peak at the switching potential is still easily observable.
Mechanism of Adenosine Oxidation. The adenine moiety
undergoes oxidation and reduction in the adenosine molecule.
Previous studies using larger electrodes and slower electrochemi-
cal methods have detected one peak for adenosine at a potential
1
2
of 1.3 V versus Ag/AgCl. However, we observed two oxidation
peaks for adenosine at T-650 electrodes, one at 1.5 V and one at
A peak representing the oxidation of product III to IV (Scheme
1
) was not detected using T-650 disk electrodes. However, when
1
.0 V. Oxidation potentials are often shifted in time (i.e., appear
using a different type of carbon fiber, P-55, a third peak was
detected around 0.5 V (Figure 2C). The second oxidation peak at
the P-55 carbon fiber also appears to be shifted to a slightly lower
(
23) Baur, J. E.; Kristensen, E. W.; May, L. J.; Wiedemann, D. J.; Wightman,
R. M. Anal. Chem. 1988, 60, 1268-1272.
746 Analytical Chemistry, Vol. 79, No. 2, January 15, 2007

Figure 2. Mechanism of adenosine oxidation. (A) Current versus time traces for the peak oxidative currents (triangles, peak 1; circles, peak
) for 5 µM adenosine. In this flow injection analysis experiment, the 3-s exposure of the electrode to adenosine began at 3.5 s. The inset shows
2
normalized current for each peak (maximum value is 100 for each curve). The left cyclic voltammogram, at the 3.5-s time point, demonstrates
only the first oxidation peak was detected on the first scan adenosine was present. The right cyclic voltammogram was taken at 4.0 s when both
peaks were nearly at the maximum. (B) Cyclic voltammograms of 5 µM adenosine when the switching potential was 1.3 (dashed line) and 1.5
(solid line). Only the forward, oxidative scan is shown for clarity. No oxidation peaks are detected when the potential is only scanned to 1.3 V.
(C) Cyclic voltammogram of 5 µM adenosine at a P-55 disk electrode. Three oxidation peaks and two reduction peaks were detected.
Scheme 1. Sequential Oxidation Steps for Adenosinea
a
R is a ribose unit.
potential (0.9 V) than with T-650 electrodes. Two reduction peaks
were detected at 0.25 and -0.05 V. The reduction peaks are
believed to be due to the second and third redox reactions in the
scheme, because no reduction peak is present in the very first
cyclic voltammogram collected when adenosine is introduced into
the flow cell. The shifts in peak potentials and appearance of new
oxidation and reduction peaks are likely due to differences in
electron-transfer properties between the two types of carbon fibers.
Glassy carbon and pyrolitic graphite have also been shown to have
different electron-transfer properties for adenosine at traditional
millimeter-sized electrodes. The two carbon fibers we used are
made of different precursor materials, T-650 fibers from polyacry-
lonitrile carbon and P-55 fibers from pitch. The geometry of the
basal and edge planes may be different between the two fibers,
or the extent of surface oxide functionalization may differ.24 The
adenosine oxidation products might be more stably adsorbed to
the P-55 electrodes, or the geometry of the carbon might facilitate
electron transfer and detection of this third peak.
estimated from the 200 nM cyclic voltammograms. Assuming a
detection limit of three times the noise, the limit of detection was
14 ( 5 nM (n ) 8). This is 100 times lower than the previous
1
9
report of adenosine detection at carbon-fiber microelectrodes
and similar to studies performed with diamond microelectrodes
and enzyme electrodes. Basal adenosine concentrations are
estimated to be 50-200 nM in the brain,1 so these limits of
detection should be sufficient to measure changes from basal
concentrations.
25
15
To test whether the detection of adenosine was adsorption or
diffusion controlled, the scan rate was varied between 200 and
1200 V/s, typical scan rates for in vivo measurements. The peak
oxidation current for 5 µM adenosine was linear with scan rate
1
2
2
(r ) 0.99 for peak 1) indicating the kinetics are adsorption
controlled (Figure 3B). Plots of oxidation current versus the
2
square root of scan rate have lower correlation coefficients (r )
0.92, peak 1, data not shown). While there will be a diffusional
component of the kinetics, these scan rate studies as well as the
holding potential and time between scan experiments described
below indicate that the kinetics are largely adsorption controlled.
As shown in Figure 3A, the adsorption-controlled behavior extends
until 20 µM. At higher concentrations, adsorption sites would be
saturated and the behavior more diffusion-controlled. This is
consistent with the detection of other electroactive molecules at
Characterization of Adenosine Electrochemistry. Concen-
trations of adenosine were varied between 0.2 and 40 µM to
determine the linear range of detection. The plot of peak current
versus concentration (Figure 3A) shows linearity up to 20 µM,
with similar behavior for both oxidation peaks. The limits of
detection for adenosine at carbon-fiber microelectrodes were
(
24) McCreery, R. L. In Electroanalytical Chemistry, 17 ed.; Bard, A. J., Ed.; Marcel
(25) Xie, S.; Shafer, G.; Wilson, C. G.; Martin, H. B. Diamond Relat. Mater. 2006,
15, 225-228.
Dekker: New York, 1991; pp 221-374.
Analytical Chemistry, Vol. 79, No. 2, January 15, 2007 747

Figure 3. Characterization of adenosine electrochemistry. (A) Effect
of adenosine concentration on current. Peak oxidation currents are
plotted for the first (triangles) and second (circles) oxidation peaks
Figure 4. Evidence for adsorption. (A) The holding potential was
varied between 0 and -0.6 V, and peak oxidative current for 5 µM
adenosine is plotted for peak 1 (triangles) and peak 2 (circles) (n )
(at 1.5 and 1.0 V in the cyclic voltammogram, respectively). Error
bars are the SEM for n ) 8 electrodes. The isotherms are linear up
8). (B) The frequency of repeating the cyclic voltammetry waveform
2
to 20 µM, and the regression values for the linear portion were r )
was varied from 5 to 100 Hz. The peak current ( SEM is plotted for
oxidative peaks 1 (triangles) and 2 (circles) (n ) 8).
2
0
.98 for the first peak and r ) 0.95 for the second peak. (B) Scan
rate dependence. Scan rate was varied from 200 to 1200 V/s, and
the resultant oxidative currents were measured for detection of 5 µM
2
adenosine. Current was linear with scan rate. For peak 1, r was 0.99,
2
while for peak 2, r was 0.98. Error bars are the SEM (n ) 8
electrodes).
carbon-fiber microelectrodes, such as dopamine, which are also
Figure 5. Repeated exposure of a carbon-fiber microelectrode to
5 µM adenosine. Stable oxidation currents were observed for the main
oxidation peak for 10 consecutive flow injection exposures of the
electrode to adenosine. The relative standard deviation between peak
heights was 3%.
adsorption-controlled at low concentrations.26
Polished carbon surfaces have a significant amount of oxides,27
and these sites are generally favorable for adsorption. Adenosine
has been previously shown to adsorb well to conventional pyrolitic
graphite electrodes.28 Electrochemical pretreatment of carbon
surfaces, by scanning from 0 to +3.0 V or scanning from -1 to
affect the amount of adsorption and the current measured. Indeed,
the current increased as the holding potential was lowered from
1
.5 V repeatedly, has been shown to increase surface oxide
0
to -0.4 V (Figure 4A). However, lowering the holding potential
1
9,29
groups.
Similarly, for dopamine, an increased sensitivity is
to -0.6 did not increase the current for the first oxidation peak.
An explanation for this is that the background current is less stable
when lower holding potentials are used. Background subtraction
errors are especially apparent at the switching potential where
the change in current is steepest. Previous studies for dopamine
detection have used -0.6 V holding potentials, but erroneous
peaks due to background subtraction errors around the switching
potential were present.20 Those switching potential errors do not
affect the potentials where dopamine is detected but do interfere
with adenosine observation.
The amount of adsorption is also controlled by the time spent
at the holding potential. Using a holding potential of -0.4 V, the
frequency of waveform application was varied (Figure 4B). The
higher frequencies resulted in lower currents because less time
was available for adsorption at the holding potential. The 100-Hz
frequency results in nearly continuous application of our wave-
form, and the detection limits were substantially less than the
normal 10-Hz application. The previous reports of cyclic voltam-
metry detection of adenosine at a carbon-fiber microelectrode
achieved only 5 µM detection limits.19 Their method applied a
observed using an upper scanning potential of 1.4 V instead of
the traditional 1.0 V.20 This increased current for dopamine was
attributed to an increase in oxide groups that promote adsorption
and increase electron-transfer kinetics. The waveform used in our
adenosine studies is similar to that previously used to increase
2
0
dopamine sensitivity. Adsorption of adenosine is likely due to
molecular interactions of the adenine group with surface oxide
groups. Electrochemical pretreatment of T-650 disk electrodes by
1
9
repeatedly scanning from -1 to 1.5 V at 10 V/s leads to a 2.5-
fold increase in the adenosine oxidation current (data not shown),
confirming that increased sensitivity for adenosine is observed at
electrodes with increased oxide groups.
Adsorption of adenosine to the electrode is due to the negative
holding potential. Therefore, varying the holding potential should
(
26) Bath, B. D.; Michael, D. J.; Trafton, B. J.; Joseph, J. D.; Runnels, P. L.;
Wightman, R. M. Anal. Chem. 2000, 72, 5994-6002.
(
(
(
27) Duvall, S. H.; McCreery, R. L. Anal. Chem. 1999, 71, 4594-4602.
28) Dryhurst, G. Talanta 1972, 19, 769-787.
29) Gonon, F. G.; Fombariet, C. M.; Buda, M. J.; Pujol, J. F. Anal. Chem. 1981,
5
3, 1386-1389.
748 Analytical Chemistry, Vol. 79, No. 2, January 15, 2007

Figure 8. Monitoring an enzymatic reaction. Adenosine deaminase
0.015 unit) was added to 5 mL of 10 µM adenosine. Every minute,
(
an aliquot was analyzed using flow injection analysis and the
concentration of adenosine measured by monitoring the peak current
at 1.5 V. The results are a plot of the normalized current before
addition of enzyme. Adenosine was completely degraded within
4
min.
adenosine. Figure 6B and C show that the currents for 5 µM AMP
and ATP are much smaller than for adenosine (Figure 6A). The
addition of phosphate groups gives these molecules a negative
charge, which would cause repulsion at the negative holding
potential of the electrode. ATP, with three phosphate groups, is
more negatively charged and therefore is least likely to be
adsorbed to the electrode. On average, the oxidation current for
AMP was 36 ( 10% and for ATP was 14 ( 5% of the current for
adenosine. Therefore, the interference from ATP in vivo is
expected to be minimal.
Figure 6. Cyclic voltammograms of possible interferents. Cyclic
voltammograms are plotted for (A) 5 µM adenosine, (B) 5 µM
adenosine monophopshate, (C) 5 µM adenosine triphosphate, (D) 5
µM dopamine, (E) 5 µM guanine, and (F) 5 µM inosine.
Dopamine is well characterized at carbon-fiber microelectrodes,
and the oxidation peak at 0.6 V occurs at much lower potentials
than the oxidation of adenosine (Figure 6D), Similarly, guanine,
another purine, had an oxidation potential of 1.1 V, which is 0.4
V lower than the oxidation potential for the main adenosine
oxidation peak (Figure 6E). The last two oxidation steps for
adenosine are very similar to the oxidation of guanine. Guanine
(
2-amino-6-hydroxypurine) has an oxy group at position 6 so
Figure 7. Codetection of dopamine and possible interferents. (A)
Cyclic voltammogram for a mixture of 5 µM dopamine and 5 µM
adenosine. (B) Cyclic voltammogram for a mixture of 5 µM guanine
and 5 µM adenosine.
oxidation gives a dioxyproduct, similar to product III for adenosine
oxidation (Scheme 1). This product can then be reversibly
oxidized to a diimine. For guanine, we also observe two peaks, at
11
0
.6 and 1.1 V (Figure 6E), similar to peaks for the second and
third oxidation steps for adenosine at a P-55 electrode (Figure
C). Inosine, a product of adenosine metabolism, was not elec-
cyclic voltammetry waveform from -1.0 to 1.5 V continuously,
which would not allow time for adsorption. In contrast, applying
the waveform only 5 or 10 times a second allows time for
adsorption. Maximizing adsorption is the key to achieving low
limits of detection.
Stability and Selectivity. An ideal in vivo sensor responds
identically to a concentration change every time. To test the
stability, we repeatedly exposed an electrode to boluses of
adenosine and measured the resultant currents (Figure 5). The
stable response indicates that the oxidation products were not
fouling the electrode and the electrode is suitable for repeated
measurements. The relative standard deviation (rsd) for 10
injections was only 3%, indicating good reproducibility. Similar rsd
values were observed for repeated injections at other electrodes,
with an average rsd of 5% for eight electrodes.
2
troactive at these potentials and therefore would not be an
interferent (Figure 6F). The inability of carbon-fiber microelec-
trodes to detect inosine is an advantage over previously con-
structed enzyme sensors that were equally sensitive to adenosine
and inosine.15
To demonstrate that adenosine could be discriminated from
dopamine and guanine, we also studied mixtures of compounds.
Figure 7 shows mixtures of 5 µM adenosine with 5 µM dopamine
(Figure 7A) and 5 µM guanine (Figure 7B). Three peaks are
visible for the mixture of dopamine and adenosine, corresponding
to oxidation of dopamine and the two adenosine oxidation peaks.
For guanine and adenosine, the main oxidation peak for guanine
and the secondary peak for adenosine overlap, but are distinguish-
able from the main oxidation peak of adenosine at 1.5 V. While
the electrode is more sensitive to dopamine and guanine than to
adenosine, the peak heights observed in the mixtures were the
same as observed for single injections of the compounds with that
Other compounds could act as interferents for adenosine
detection in biological samples. For example, the adenosine
nucleotides AMP and ATP are energetic molecules found in
abundance in biological systems. ATP is the major energy unit of
cells, and in the brain, it can be released from vesicles as a
3
0
neurotransmitter. Catabolism of ATP yields AMP and then
(30) Zimmermann, H. Trends Neurosci. 1994, 16, 420-426.
Analytical Chemistry, Vol. 79, No. 2, January 15, 2007 749

electrode. Therefore, dopamine and guanine are not expected to
interfere with adenosine detection in biological experiments.
Monitoring an Enzymatic Reaction. The utility of the carbon-
fiber microelectrode to monitor a biological reaction is shown in
Figure 8. Adenosine deaminase, an endogenous enzyme that
metabolizes adenosine to inosine, was added to a solution of 10
µM adenosine. Every minute, an aliquot of the mixture was
analyzed with flow injection analysis. Inosine is not electroactive
low-nanomolar detection limits are less than basal adenosine levels
in tissue and should enable monitoring of biological functions. In
the brain, adenosine is reported to increase after a stroke but exact
concentration changes and the potential effects for adenosine
receptors are not known. An adenosine sensor would give insight
into the mechanism of adenosine release and the neuroprotective
functions of adenosine. Adenosine also has cardioprotective
properties that may protect heart tissue during ischemia, a lack
of blood flow. The release of adenosine from myocardial cells
could be monitored under conditions of oxygen depletion to
determine the actual levels of adenosine. The applicability of this
sensor for physiological experiments will allow real-time measure-
ments of adenosine concentrations in normal and disease states.
(Figure 6F), so the extent of reaction is observed by a decrease
in the current for oxidation of adenosine. Adenosine was com-
pletely degraded within 4 min. This is in good agreement with
the stated enzyme activity that predicted the adenosine should
be metabolized in 3.3 min. After the experiment, the stock 10 µM
adenosine solution was analyzed again and the current was 97 (
ACKNOWLEDGMENT
5
% of the initial adenosine signal, indicating that the electrode
This work was partially supported by a grant from the Jeffress
was not fouled and could still detect adenosine. This experiment
shows that fast changes in adenosine produced in biological
environments should be able to be monitored and that adenosine
metabolites do not interfere with its detection.
Conclusions. A rapid, sensitive method has been developed
for detection of physiological concentrations of adenosine. The
Memorial Trust to B.J.V.
Received for review September 27, 2006. Accepted
November 14, 2006.
AC061820I
750 Analytical Chemistry, Vol. 79, No. 2, January 15, 2007