In vivo neurochemical monitoring using benzoyl chloride derivatization and liquid chromatography-mass spectrometry

Article abstract of DOI:10.1021/ac202794q

In vivo neurochemical monitoring using microdialysis sampling is important in neuroscience because it allows correlation of neurotransmission with behavior, disease state, and drug concentrations in the intact brain. A significant limitation of current practice is that different assays are utilized for measuring each class of neurotransmitter. We present a high performance liquid chromatography (HPLC)-tandem mass spectrometry method that utilizes benzoyl chloride for determination of the most common low molecular weight neurotransmitters and metabolites. In this method, 17 analytes were separated in 8 min. The limit of detection was 0.03-0.2 nM for monoamine neurotransmitters, 0.05-11 nM for monoamine metabolites, 2-250 nM for amino acids, 0.5 nM for acetylcholine, 2 nM for histamine, and 25 nM for adenosine at sample volume of 5 μL. Relative standard deviation for repeated analysis at concentrations expected in vivo averaged 7% (n = 3). Commercially available ¹³C benzoyl chloride was used to generate isotope-labeled internal standards for improved quantification. To demonstrate utility of the method for study of small brain regions, the GABA_A receptor antagonist bicuculline (50 μM) was infused into a rat ventral tegmental area while recording neurotransmitter concentration locally and in nucleus accumbens, revealing complex GABAergic control over mesolimbic processes. To demonstrate high temporal resolution monitoring, samples were collected every 60 s while neostigmine, an acetylcholine esterase inhibitor, was infused into the medial prefrontal cortex. This experiment revealed selective positive control of acetylcholine over cortical glutamate.

Full text of DOI:10.1021/ac202794q

Article
pubs.acs.org/ac
In Vivo Neurochemical Monitoring Using Benzoyl Chloride
Derivatization and Liquid Chromatography−Mass Spectrometry
Peng Song,† Omar S. Mabrouk,†,‡ Neil D. Hershey,† and Robert T. Kennedy*,†,‡
‡
†Department of Chemistry and Department of Pharmacology, University of Michigan, Ann Arbor, Michigan 48109, United States
S
* Supporting Information
ABSTRACT: In vivo neurochemical monitoring using microdialysis
sampling is important in neuroscience because it allows correlation of
neurotransmission with behavior, disease state, and drug concentrations in
the intact brain. A significant limitation of current practice is that different
assays are utilized for measuring each class of neurotransmitter. We present a
high performance liquid chromatography (HPLC)−tandem mass spectrom-
etry method that utilizes benzoyl chloride for determination of the most
common low molecular weight neurotransmitters and metabolites. In this
method, 17 analytes were separated in 8 min. The limit of detection was
0.03−0.2 nM for monoamine neurotransmitters, 0.05−11 nM for mono-
amine metabolites, 2−250 nM for amino acids, 0.5 nM for acetylcholine, 2
nM for histamine, and 25 nM for adenosine at sample volume of 5 μL.
Relative standard deviation for repeated analysis at concentrations expected
in vivo averaged 7% (n = 3). Commercially available 13C benzoyl chloride was used to generate isotope-labeled internal standards
for improved quantification. To demonstrate utility of the method for study of small brain regions, the GABAA receptor
antagonist bicuculline (50 μM) was infused into a rat ventral tegmental area while recording neurotransmitter concentration
locally and in nucleus accumbens, revealing complex GABAergic control over mesolimbic processes. To demonstrate high
temporal resolution monitoring, samples were collected every 60 s while neostigmine, an acetylcholine esterase inhibitor, was
infused into the medial prefrontal cortex. This experiment revealed selective positive control of acetylcholine over cortical
glutamate.
onitoring neurotransmitter concentration dynamics in
the living brain is a critically important tool in
(MS).18−20 Despite extensive research into methods for
chemical analysis of dialysate, all methods presently in use
can only determine a subset of common small molecule
neurotransmitters. Therefore, studies that require monitoring
different types of neurotransmitters must use multiple assays
which increases costs and time required for equipment
maintenance, method development, and analysis. Use of
multiple assays also increases sample volume requirements
and animal usage. Assays that measure only a single or few
neurotransmitters also preclude discovering involvement of
unanticipated neurotransmitter systems. A comprehensive
analytical method for neurotransmitter measurements would
be of great value to the neurosciences by revealing previously
unknown neurotransmitter interactions. Such a method could
also accelerate neurological drug development by allowing rapid
evaluation of the effect of novel compounds in the brain. Any
such method must be sensitive enough for dialysate samples
and have sufficient throughput for the many samples generated
from in vivo experiments.
M
neuroscience. In vivo measurements enable study of the
relationship between neurotransmitter concentrations in
relevant brain nuclei and behavior, drug effects, or disease
states. Since its inception, microdialysis sampling has been the
preeminent tool for making such measurements.1−3 In this
approach, a semipermeable membrane probe is inserted into
the brain and perfused with artificial cerebral spinal fluid
(aCSF). Molecules in the extracellular space diffuse across the
membrane according to their concentration gradient and are
collected into fractions which are analyzed for neurotransmitter
or metabolite content. This tool has been invaluable for
neuroscience; e.g., it has been used to demonstrate that all
drugs of abuse activate the mesolimbic dopamine (DA)
system.4 Glutamate (Glu) sustains drug seeking behavior,5,6and
adenosine (Ado) is a modulator of sleep.7 The technique is also
used clinically for studying epilepsy8 and brain trauma9,10 and
plays a prominent role in the pharmaceutical industry when
screening novel neurological and psychiatric therapeutics.
A key to using microdialysis is analysis of sample fractions.11
Many assays for neurotransmitters have been developed using
high performance liquid chromatography (HPLC)-electro-
chemical detection,12,13 HPLC-fluorescence detection,14 capil-
lary electrophoresis-laser induced fluorescence,15,16 immuno-
assay,17 and more recently, HPLC-mass spectrometry
Here, we report a HPLC-MS method for the measurement
of 12 of the most commonly studied neurotransmitters or
Received: October 20, 2011
Accepted: November 27, 2011
Published: November 27, 2011
© 2011 American Chemical Society
412
dx.doi.org/10.1021/ac202794q | Anal. Chem. 2012, 84, 412−419

Analytical Chemistry
Article
Figure 1. Chemical structure of targeted neurotransmitters and metabolites (A). Reaction scheme of benzoylation using benzoyl chloride (B).
neuromodulators (Figure 1A) including ACh, Ado, DA,
norepinephrine (NE), serotonin (5-HT), histamine (Hist),
Glu, glycine (Gly), aspartate (Asp), γ-aminobutyric acid
(GABA), serine (Ser), and taurine (Tau). The method also
assays the metabolites homovanillic acid (HVA), 5-hydrox-
yindole-3-acetic acid (5-HIAA), 3,4-dihydroxyphenylacetic acid
(DOPAC), normetanephrine (NM), and 3-methoxytyramine
(3-MT). The method is compatible with challenging experi-
ments which generate low concentration samples such as using
small microdialysis probes for high spatial resolution and fast
sampling rates (60 s/sample) for high temporal resolution in
vivo monitoring.
A major difficulty to overcome in developing such an assay is
identifying chromatographic conditions that can resolve the
highly polar neurochemicals while remaining compatible with
MS detection. We discovered that derivatization with benzoyl
chloride renders the compounds more hydrophobic so that
they can be separated by reversed phase chromatography.
Derivatization also increases sensitivity and provides a
convenient way to improve quantification by stable-isotope
labeled internal standards generated using 13C6 benzoyl
chloride. Importantly, benzoyl chloride reacts with primary
and secondary amines, phenols, and ribose-hydroxyl groups
(Figure 1B) allowing nearly all small organic molecule
neurotransmitters to be labeled. ACh, which cannot be labeled
413
dx.doi.org/10.1021/ac202794q | Anal. Chem. 2012, 84, 412−419

Analytical Chemistry
Article
by benzoyl chloride, is directly detected in this method. Rapid
derivatization and an 8 min separation time give the method
sufficient throughput for the large number of samples generated
by microdialysis experiments. The method uses commercial
instrumentation and readily available reagents; therefore, it can
easily be adopted by other laboratories.
v). The mixture was vortexed, and 2.5 μL of internal standard
was added before LC-MS analysis.
HPLC-MS Analysis. The HPLC system was a Waters
(Milford, MA) nanoAcquity HPLC. A Waters BEH C18
column (1 mm × 100 mm, 1.7 μm, 130 Å pore size) was used
for separation. Mobile phase A was 10 mM ammonium formate
and 0.15% (v/v) formic acid in water. Mobile phase B was
acetonitrile. The mobile phase gradient for all 17 analytes was
as follows: initial, 0% B; 0.1 min, 15% B; 2 min, 20% B; 2.3 min,
25% B; 2.31 min, 50% B; 5.31 min, 50% B; 5.57 min, 65% B;
6.57 min, 65% B; 6.58 min, 0% B; 8.0 min, 0% B.
The flow rate was 100 μL/min, and sample injection volume
was 9 μL in partial loop injection mode. An autosampler was
kept at ambient temperature, and column temperature was
maintained at 27 °C. A Waters/Micromass Quattro Ultima
triple quadrupole mass spectrometer was used for detection. An
atmospheric pressure ionization source was operated in positive
ESI mode at 3 kV. Source temperature was 140 °C, and
desolvation temperature was 400 °C. Cone gas and desolvation
gas flowed at 150 L/h and 500 L/h, respectively.
MRM conditions are listed in Table S1, Supporting
Information. Interchannel delay and intercycle delay were
both 10 ms. Automated peak integration was performed using
Waters Masslynx version 4.1. All peaks were visually inspected
to ensure proper integration. Calibration curves were
constructed on the basis of peak area ratio (Panalyte/PI.S.) versus
concentrations of internal standard by linear regression.
Statisical Analysis. Data were transformed to percent of
baseline measurement to normalize pretreatment levels to
100%. All analyses were performed in Graphpad (La Jolla, CA)
Prism 5. The measurements were all continuous variables, and
the Kolmogorov−Smirnov test was used to assess normality of
the residuals for each individual repeated measurement; this
assumption was met. A two-tailed repeated measures analysis of
variance (RM ANOVA) was performed on all microdialysis
data followed by a post-hoc Tukey test to test the pairwise
difference between every time point and baseline (i.e., 100%).
Although animal numbers were relatively small (n = 8 for the
bicuculline study and n = 5 for the neostigmine study),
following baseline, we collected 7 (for bicuculline) and 60 (for
neostigmine) measures over time on each animal, enabling
analyses of variation in the outcomes both between and within
animals. In addition, with 8 animals in one group and 5 in the
other and 8 repeated measures on each animal, a RM ANOVA
will have more than 90% power to detect a mean difference of
50% points between the two groups at a 0.05 level of
significance (assuming correlations of 0.25 of the repeated
measures and standard deviations of 10% points in each group).
RM ANOVA is the primary statistical test used for such
continuous measurements. Since the two experiments (i.e.,
bicuculline and neostigmine) were performed independently
and did not contain control groups since basal levels were taken
as control, no randomization was necessary. However, the
individual who processed the data was blind to any expected
outcomes for each measurement.
METHODS
■
Chemicals and Reagents. All chemicals, drugs, and
reagents were purchased from Sigma Aldrich (St. Louis, MO)
unless otherwise noted. Water and acetonitrile are Burdick &
Jackson HPLC grade purchased from VWR (Radnor, PA). A 10
mM stock solution of each analyte was made in HPLC water
and kept at −80 °C. A standard mixture was diluted from stock
using aCSF consisting of 145 mM NaCl, 2.68 mM KCl, 1.10
mM MgSO4, 1.22 mM CaCl2, 0.50 mM NaH2PO4, and 1.55
mM Na2HPO4, adjusted pH to 7.4 with 0.1 M NaOH.
Calibration curves were made using standards at 50 nM, 500
nM, 1000 nM, 5000 nM, and 10000 nM for Gly, Ser, and Tau;
5 nM, 50 nM, 100 nM, 500 nM, and 1000 nM for Asp, Glu,
GABA, Hist, Ado, HVA, 5-HIAA, and DOPAC; 0.5 nM, 5 nM,
10 nM, 50 nM, and 100 nM for ACh, 5-HT, NE, DA, NM, and
3-MT. Internal standard was 1 mM Gly, Ser, and Tau; 100 μM
Asp, Glu, GABA, Hist, Ado, HVA, 5-HIAA, and DOPAC; and
10 μM 5-HT, NE, DA, NM, and 3-MT derivatized with 13C6
benzoyl chloride using the same procedure as 12C reagent, then
diluted 100-fold in DMSO containing 1% formic acid. d4-ACh
(C/D/N isotopes, Pointe-Claire, Canada) was spiked into the
reaction mixture to a final concentration of 100 nM.
Microdialysis. Adult male Sprague-Dawley rats (Harlan,
Indianapolis, IN) weighing between 295 and 355 g were used.
Ketamine (65 mg/kg i.p.) and dexdormitor (0.25 mg/kg i.p.)
were used for operative anesthesia. For dual probe experiments,
concentric microdialysis probes (250 μm diameter) were
implanted unilaterally in both the VTA (1 mm long probe)
and NAc (1.5 mm long probe) according to following
coordinates from bregma and dura: anterior−posterior (AP)
−5.3, medial−lateral (ML) 0.5, and dorsal−ventral (DV) −
8.0 mm and AP +1.2, ML 1.4, and DV −7.8 mm, respectively.
For single probe experiments, probes (3 mm long) were
implanted into the mPFC from bregma and dura: AP +3.0, ML
0.5, and DV − 4.0 (Paxinos and Watson 2007). Probes were
secured to the skull by acrylic dental cement and metallic
screws. Following surgery, rats were allowed to recover and
experiments were performed later. Animals were awake and
freely moving with access to food and water throughout the
experiment. Microdialysis probes were flushed at 1.5 μL/min
with aCSF for 3 h using a Chemyx (Stafford, TX) Fusion 400
syringe pump. Perfusion flow rate was reduced to 0.6 μL/min,
and samples were collected every 20 min for VTA-NAc
experiments. For mPFC experiments, perfusion flow rates were
reduced to 1 μL/min to generate 1 μL samples. One μL
fractions were diluted with 4 μL of aCSF, and then, treated the
same way as a 5 μL sample described below. Following
collection of basal fractions, VTA lines were switched to aCSF
containing 50 μM bicuculline for the duration (2 h) of
experiments. Though 12 μL of dialysate were collected per
fraction, only 5 μL of total volume was used for analysis. For
mPFC, following 20 min of basal fraction collections, 1 μM
neostigmine was perfused through the probe for 5 min.
RESULTS
■
HPLC-MS of Benzoylated Neurotransmitters. All ana-
lytes of interest were benzoylated (except ACh), resolved by
reversed phase HPLC, and detected by MS/MS (Figure 2).
Amino acids were singly benzoylated while monoamines
containing phenolic groups were doubly or triply labeled as
determined by the mass detected. Protonated benzoylation
Benzoylation Reaction. Five μL of standard or sample
was mixed with 2.5 μL of borate buffer (sodium tetraborate,
100 mM) and 2.5 μL of benzoyl chloride (2% in acetonitrile, v/
414
dx.doi.org/10.1021/ac202794q | Anal. Chem. 2012, 84, 412−419

Analytical Chemistry
Article
two peaks corresponding to N6-benzoyl-Ado and 3′-O-benzoyl-
Ado. The latter product is a result of activity of the 3′-hydroxy
group on the ribose ring in nucleophilic substitutions.21,22
Besides m/z 105, both N6-benzoyl-Ado and 3′-O-benzoyl-Ado
generated a characteristic m/z 136 fragment corresponding to
an adenine moiety under MS/MS conditions. The 3′-O-
benzoyl-Ado produced more abundant fragment ions and was
chosen for MRM quantification. For ACh, the quanternary
ammonium residue (m/z 87) after loss of acetyl ester was used
for quantification.
Besides enhancing retention on reversed phase columns,
benzoylation also enhanced sensitivity for ESI-MS. For
example, signal (peak area) was increased 22-fold for 5-HT,
23-fold for Glu, 37-fold for DA, 460-fold for 3-MT, and 1400-
fold for GABA compared to their respective native form.
Analytical performance of benzoylation HPLC-MS method was
satisfactory, with average relative standard deviation (RSD) of
7% for analysis of samples made in triplicate, linear response
(R2 > 0.99), biologically relevant limits of detection (LOD),
and adequate analyte stability (Table 1). Use of isotopically
labeled internal standard improved measurement precision with
reduced RSDs for repeated analysis of the same sample (Figure
S1, Supporting Information).
Figure 2. Ion chromatogram for all 17 analytes. The concentration of
each analyte was: ACh (10 nM); Tau (1000 nM); Hist (100 nM); Ser
(1000 nM); Asp (100 nM); Gly (1000 nM); Glu (100 nM); GABA
(100 nM); Ado (100 nM); 5-HIAA (100 nM); HVA (100 nM); NM
(10 nM); 5-HT (10 nM); DOPAC (100 nM); 3-MT (10 nM); NE
(10 nM); and DA (10 nM).
products (MW + 1) were observed by positive electrospray
ionization (ESI) for primary and secondary amines and ribose
hydroxy groups. For analytes containing only phenol groups
(HVA, DOPAC, 5-HIAA), the ammonium adduct of their
products (MW + 18) was detected. For ACh, the unlabeled
molecular ion was used for detection.
Analytes were detected by MS/MS under collision activated
dissociation (CAD) conditions; therefore, the fragmentation of
each analyte was examined to determine the best product ions
to use for quantification (Table S1; Supporting Information).
For most benzoylated analytes, the benzoyl fragment (m/z
105) was the most abundant product ion (Figure 3, top) and
was subsequently used for multiple reaction monitoring
(MRM). 5-HT generated additional fragment ions (m/z 160,
m/z 264) besides the benzoyl fragment (Figure 3, bottom).
The unique fragment (m/z 264) was selected for MRM of 5-
HT. Most analytes were fully labeled, yielding a single
chromatographic peak; however, benzoylated Ado produced
Microdialysis. To demonstrate utility of the method under
conditions requiring good spatial resolution, we performed dual
probe microdialysis experiments in the mesolimbic pathway,
i.e., the ventral tegmental area (VTA) and the nucleus
accumbens (NAc). Though probes were only 1 mm and 1.5
mm in length, respectively, we detected all expected neuro-
chemicals in these samples (Figure S2, Supporting Informa-
tion). Basal neurotransmitter concentrations in VTA and NAc
dialysate, reported in Table 1, are within the expected
range.23,24
Perfusion of the GABAA receptor antagonist bicuculline (50
μM) into the VTA (n = 8) yielded a complex set of changes in
Figure 3. MS/MS spectra of benzoylated DA (top) and 5-HT (bottom). Benzoylated 5-HT gave a unique fragment of 264 m/z besides the
characteristic 105 m/z (benzoyl group). Most other benzoylated analytes like DA produced only 105 m/z in CAD.
415
dx.doi.org/10.1021/ac202794q | Anal. Chem. 2012, 84, 412−419

Analytical Chemistry
Article
a
Table 1. Figures of Merit for Benzoylation-HPLC-MS Method
basal in vivo dialysate concentration
RSD
LOD
(nM)
% remaining after storage
analyte
Gly
(n = 3)
R2
(n = 3)
VTA (n = 8)
NAc (n = 8)
mPFC (n = 5)
5
5
500
2
0.9933
0.9989
0.9975
0.9995
0.9987
0.9995
0.9981
0.9997
0.9998
0.9961
0.9974
0.9999
0.9973
0.9973
0.9996
0.9999
0.9963
80 18
114 10
94 21
4
2 μM
7
2 μM
7
1 μM
GABA
Ser
0.04 0.01 μM
6.4 0.8 μM
12 4 μM
0.03 0.01 μM
20 6 μM
44 8 nM
10 4 μM
4.3 0.6 μM
n.d.
11
7
250
250
50
5
Tau
86 21
4
1 μM
Asp
10
8
109 19
107 14
109 14
0.6 0.1 μM
0.9 0.2 μM
0.7 0.1 μM
1.9 0.3 μM
Glu
0.7 0.3 μM
2.2 0.4 nM
2.0 0.4 nM
0.36 0.08 nM
1.0 0.4 nM
n.d.
DA
4
0.03
0.2
0.1
2
3
1 nM
8
2 nM
NE
10
4
97
7
1.4 0.2 nM
0.36 0.06 nM
1.2 0.5 nM
18 3 nM
2.0 0.5 nM
0.32 0.08 nM
0.8 0.2 nM
28 5 nM
5-HT
Hist
118 19
118 14
102 15
116 18
13
4
Ado
25
0.05
2
3-MT
DOPAC
HVA
NM
1
2.2 0.3 nM
0.8 0.1 μM
0.12 0.01 μM
0.8 0.1 nM
1.1 0.1 μM
1.7 0.4 nM
n.d.
5
103
108
102
5
3
2
5
1 μM
110 9 nM
49 9 nM
2.7 0.9 nM
0.26 0.04 μM
3.8 0.6 nM
4
0.5
0.1
5
0.7 0.2 μM
0.46 0.03 nM
0.67 0.07 μM
15 2 nM
2
5-HIAA
ACh
7
115 20
7
0.5
nonlabeled
8
2 nM
a
RSDs were calculated by comparing peak area of 3 individually derivatized standards at approximate in vivo concentrations, thus evaluating both
reaction and analytical variability. Dynamic range tested for linearity was 500−10 000 nM for Gly, Ser, and Tau; 5−100 nM for HVA, 5-HIAA, NM,
3-MT, and DOPAC; 50−1000 nM for GABA, Asp, Glu, Hist, and Ado; 0.5−100 nM for ACh, 5-HT, NE, and DA. Percent remaining was calculated
by comparing peak area of stored standard solution (room temperature for a week) to a freshly derivatized standard solution. Basal concentration of
each analyte in dialysate from different brain regions is listed. These values are not corrected for microdialysis probe recovery.
Figure 4. Dual probe microdialysis of the mesolimbic pathway. Bicuculline (50 μM) was perfused in the VTA while monitoring neurotransmitters
both locally and distally in the NAc. The heat map shows all changes of neurochemicals monitored where colors correlate with changes expressed as
percentage of baseline. Times in the heat map are referenced to infusion of biculline. RM ANOVA and a posthoc Tukey test were performed to
compare basal levels against post drug levels. *p < 0.05, **p < 0.01, and ***p < 0.001 compared to basal level in the respective brain region. A line
graph of these data is presented in Figure S3, Supporting Information.
the neurochemicals measured (Figure 4 and Figure S3,
Supporting Information). This treatment stimulated local
dendritic (153%; F(7,7) = 10.15, p < 0.0001) and limbic
(106%; F(7,7) = 5.835, p < 0.0001) DA release, as anticipated.25
DA release also correlated with an increase in the DA
metabolite DOPAC in the VTA (60%; F(7,5) = 9.209, p <
0.0001) and the NAc (49%; F(7,5) = 9.641, p < 0.0001). We
(153%; F(7,5) = 3.809, p = 0.0036), Ado (53%; F(7,5) = 5.086, p
= 0.0005), and Gly (66%; F(7,6) = 3.588, p = 0.0041).
In the NAc, we observed unanticipated increases in 5-HT
(106%; F(7,6) = 2.450, p = 0.0335), NE (132%; F(7,7) = 4.843, p
= 0.0003), Glu (73%; F(7,6) = 6.759, p < 0.0001), ACh (66%;
F(7,5) = 2.782, p = 0.0207), and Hist (95%; F(7,5) = 3.137, p =
0.0111). We also observed an unexpected reduction in GABA
(34%; F(7,6) = 2.291, p = 0.0452) in the NAc (Figure 4 and
Figure S3, Supporting Information).
observed unanticipated increases in VTA 5-HT (146%; F(7,7)
=
5.00, p = 0.003), NE (97%; F(7,7) = 3.570, p = 0.0035), ACh
416
dx.doi.org/10.1021/ac202794q | Anal. Chem. 2012, 84, 412−419

Analytical Chemistry
Article
To demonstrate the utility of the method under conditions
of high temporal resolution, we used 60 s sampling in the
medial prefrontal cortex (mPFC) to observe changes in ACh
release following a 5 min local perfusion of the ACh esterase
inhibitor neostigmine (1 μM; n = 5). Neostigmine perfusion
caused a prompt increase in ACh (518%; F(60,4) = 4.193, p <
0.0001) levels in the mPFC as anticipated (Figure 5A). We also
protects them against oxidation under the high pH conditions
used for labeling.29 Another issue is that metabolites labeled at
multiple sites became hydrophobic and insoluble in 100%
aqueous solution; therefore, 25% organic solvent content in the
reaction mixture was used to keep these compounds in
solution.
Benzoyl chloride labeling is similar to dansyl chloride labeling
previously reported for metabolomics30 but has several
advantages due to its structural simplicity. A preliminary
study revealed that monoamines which were multiply labeled
with the dansyl group had low CAD efficiency on our triple
quadrupole mass spectrometer, rendering MRM of these
analytes unsuitable for high sensitivity analysis. In contrast,
multiply labeled benzoylated monoamines are easily frag-
mented. 13C dansyl chloride must be synthesized for differential
labeling using light and heavy isotope reagent30 while 13C6
benzoyl chloride is commercially available at relatively low cost.
Dansylation requires 30 min reaction time at elevated
temperatures while the benzoylation reaction was instant at
room temperature. Unlike dansyl chloride, excess benzoyl
chloride is completely hydrolyzed so there is no need to
scavenge reagent after the reaction. Importantly, neither
benzoyl chloride nor the benzoylated product is light sensitive
in contrast to dansyl chloride and its derivatives.
The HPLC separation time is 8 min, and total analysis time
for each sample is around 12 min, taking into account
autosampler injection, elution, and column re-equilibration.
Over 80 samples plus standards are routinely analyzed per day
in our laboratory using this method. Such throughput is
important for microdialysis since hundreds of samples are
typically generated per study.
Figure 5. Microdialysis in the mPFC. Neostigmine (1 μM) was
perfused (black bar) for 5 min in the mPFC while neurotransmitters
were monitored locally. Dialysate was collected every 60 s. Error bar is
1 SEM (n = 5). Data was expressed as percent of baseline levels. RM
ANOVA and a posthoc Tukey test to compare basal levels against post
drug levels. *p < 0.05, **p < 0.01, and ***p < 0.001 compared to
basal mPFC level.
Mesolimbic Regulation by GABA. To demonstrate the
utility of this system, we investigated GABAergic control of the
mesolimbic pathway. The mesolimbic pathway is made up of
DA containing cells which originate in the VTA and extend to
the NAc. This pathway is intensively studied for its role in
reward processing, motivation, and addiction since all drugs of
abuse activate this pathway.4,31,32 GABA, primarily from VTA
interneurons, is an important regulator of the mesolimbic
pathway33,34 and blockade of GABAA receptors stimulates
mesolimbic DA release.25 Evidence suggests that the rewarding
effects of opiates, cannabinoids, and benzodiazepines rely on
their ability to activate the mesolimbic DA pathway by
inhibiting GABAergic interneruons in the VTA.34−36 Neuro-
transmitters other than dopamine regulated by GABA have
received less attention. Here, our method has uncovered
previously unknown GABAergic involvement in several neuro-
transmitter systems apart from DA as bicuculline perfusion into
the VTA evoked not only the expected increase in NAc DA23
(and DA metabolite, DOPAC) but also 5-HT, NE, Glu, ACh,
and Hist while reducing GABA levels. Locally in the VTA,
bicuculline not only enhanced dendritic DA (and DOPAC) but
also evoked local 5-HT, NE, ACh, Ado, and Gly release (Figure
4). These results show that GABAergic control of the VTA is
not limited to limbic DA release and that a complex interplay
between several neuronal substrates is involved in mesolimbic
activation and possibly motivation. These measurements also
demonstrate compatibility with high spatial resolution experi-
ments. The VTA is ∼1 mm in height × ∼1 mm in width,37 so a
1 mm long × 250 μm diameter dialysis probe was used. A small
probe produces low recovery (∼10%) requiring high sensitivity,
a criterion met by the current method.
observed a large (229%; F(60,4 = 4.443, p < 0.0001) transient
increase in Glu concentration (Figure 5B). Neostigmine
perfusion had no effect on DA, GABA (Figure 5C,D), or
other measured neurotransmitters (data not shown). Basal
neurotransmitter concentrations in mPFC dialysate are
reported in Table 1.
DISCUSSION
■
HPLC-MS of Neurotransmitters. This study demon-
strates that HPLC-MS is suitable for measuring all of the
most commonly studied low molecular weight neurotransmit-
ters, neuromodulators, and metabolites in a single analysis
based on both direct detection (ACh) and benzoylation. Unlike
reagents that target only primary amines, benzoyl chloride also
labels secondary amines, phenols, and ribose hydroxyl groups.
This allows several neurochemicals which do not have primary
amine groups, such as DOPAC, 5-HIAA, and HVA, to be
labeled. Even though many molecules can be detected directly
by MS without derivatization, labeling of polar neuro-
transmitters has several distinct advantages. Benzoylation
enhances reversed phase LC retention, improves ESI-MS
sensitivity, and allows for a low-cost stable isotope internal
standard.
Although benzoyl chloride has been used for polyamine and
steroid analysis,26,27 we found that some specific conditions
were necessary to achieve reproducible results for neuro-
transmitter detection. Borate buffer was preferred over other
basic buffers for derivatization because it improved precision in
catecholamine measurements. We believe that this is because
borate forms a reversible complex with catechol groups28 and
417
dx.doi.org/10.1021/ac202794q | Anal. Chem. 2012, 84, 412−419

Analytical Chemistry
Article
Prefrontal ACh and Glu Interaction. We also measured
prefrontocortical signaling to test the method under the
challenging condition of a high sampling rate. The mPFC is a
known center of emotional and cognitive function and is often
studied in psychiatric and attentional disorders.38,39 To uncover
physiologically and behaviorally relevant neurotransmitter
interactions in the mPFC, common microdialysis sampling
rates (i.e., 10- 30 min/sample) remain inadequate. Here, we
have made such measurements every 60 s to better capture
concentration dynamics on a more physiologically relevant time
scale.39 Measurement at short intervals requires high sensitivity
(1 μL samples are collected containing femtomoles of analyte)
and good throughput because of the large number of samples
generated (e.g., 80 samples per session in Figure 5).
The ACh esterase inhibitor neostigmine (5 min perfusion)
elevated ACh concentrations in the mPFC as expected but also
correlated with a transient increase in mPFC Glu concen-
tration. This result suggests that muscarinic or nicotinic ACh
receptors in the mPFC are responsible for stimulating Glu
release. Past studies have shown that exogenous ACh receptor
agonists such as nicotine enhance cognition and stimulate Glu
activity in the mPFC.40 Our data show that stimulated
endogenous ACh is capable of enhancing Glu release in this
brain area. This ACh−Glu interaction may play a role in
cognition and the beneficial effects of drugs which stimulate
ACh receptors.40 Interestingly, other neurotransmitter systems,
including the monoamines, were left unaffected by neostigmine
perfusion. These results show how a comprehensive view of
neuropharmacological effects can be assayed by the method.
They also show the value of the temporal resolution as the
transient rise of Glu would likely be undetected at longer
sampling intervals.
ACKNOWLEDGMENTS
■
We thank the Center for Statistical Consultation and Research
(CSCAR) at the University of Michigan for assistance with
statistical analyses. This work was supported by NIH grant R37
EB003320 (R.T.K.) and National Institute on Drug Abuse T32
training grant DA07268 (O.S.M.).
REFERENCES
■
14.
(1) Tossman, U.; Ungerstedt, U. Acta Physiol. Scand. 1986, 128, 9−
(2) Robinson, T. E.; Justice, J. B. J. Techniques in the behavioral and
neural sciences; Elsevier: Amsterdam, Netherlands, 1991.
(3) Westerink, B. H. C.; Cremers, T. I. F. H. Handbook of
microdialysis; Elsevier: Amsterdam, Netherlands, 2007.
(4) Di Chiara, G.; Imperato, A. Proc. Natl. Acad. Sci. U. S. A. 1988, 85,
5274−5278.
(5) Cornish, J. L.; Kalivas, P. W. J. Neurosci. 2000, 20, RC89.
(6) LaLumiere, R. T.; Kalivas, P. W. J. Neurosci. 2008, 28, 3170−
3177.
(7) Porkka-Heiskanen, T.; Strecker, R. E.; Thakkar, M.; Bjorkum, A.
A.; Greene, R. W.; McCarley, R. W. Science 1997, 276, 1265−1268.
(8) During, M. J.; Ryder, K. M.; Spencer, D. D. Nature 1995, 376,
174−177.
(9) Feuerstein, D.; Manning, A.; Hashemi, P.; Bhatia, R.; Fabricius,
M.; Tolias, C.; Pahl, C.; Ervine, M.; Strong, A. J.; Boutelle, M. G. J.
Cereb. Blood Flow Metab. 2010, 30, 1343−1355.
(10) Bullock, R.; Zauner, A.; Woodward, J.; Young, H. F. Stroke
1995, 26, 2187−2189.
(11) Perry, M.; Li, Q.; Kennedy, R. T. Anal. Chim. Acta 2009, 653,
1−22.
(12) Felice, L. J.; Felice, J. D.; Kissinger, P. T. J. Neurochem. 1978, 31,
1461−1465.
(13) Liu, Y.; Zhang, J.; Xu, X.; Zhao, M. K.; Andrews, A. M.; Weber,
S. G. Anal. Chem. 2010, 82, 9611−9616.
(14) Yoshitake, T.; Yoshitake, S.; Fujino, K.; Nohta, H.; Yamaguchi,
M.; Kehr, J. J. Neurosci. Methods 2004, 140, 163−168.
(15) Zhou, S. Y.; Zuo, H.; Stobaugh, J. F.; Lunte, C. E.; Lunte, S. M.
Anal. Chem. 1995, 67, 594−599.
CONCLUSION
■
The results show that benzoylation with HPLC-MS is a robust
method for determination of small molecule neurotransmitters
and metabolites. The method allows 17 neurotransmitters and
metabolites to be detected with 8 min separation at
concentrations from picomolar to micromolar. Internal stand-
ard is conveniently generated using commercial 13C6 reagent,
enabling improved quantification. The comprehensiveness, low
sample volume requirement, sensitivity, and throughput make
this method valuable for neurochemical monitoring under
challenging conditions such as small brain regions (VTA) and
at short intervals. These properties will make the method
widely applicable for pharmacological and brain disorder
studies. The good temporal resolution will be important for
correlating neurotransmitter concentrations to behaviors which
change on similar time scales.
(16) Robert, F.; Bert, L.; Denoroy, L.; Renaud, B. Anal. Chem. 1995,
67, 1838−1844.
(17) Maidment, N. T.; Brumbaugh, D. R.; Rudolph, V. D.; Erdelyi,
E.; Evans, C. J. Neuroscience 1989, 33, 549−557.
(18) Zhang, X. Z.; Rauch, A.; Lee, H.; Xiao, H. B.; Rainer, G.;
Logothetis, N. K. Rapid Commun. Mass Spectrom. 2007, 21, 3621−
3628.
(19) Eckstein, J. A.; Ammerman, G. M.; Reveles, J. M.; Ackermann,
B. L. J. Neurosci. Methods 2008, 171, 190−196.
(20) Ji, C. J.; Li, W. L.; Ren, X. D.; El-Kattan, A. F.; Kozak, R.;
Fountain, S.; Lepsy, C. Anal. Chem. 2008, 80, 9195−9203.
(21) Hong, S.; Pedersen, P. L. Microbiol. Mol. Biol. Rev. 2008, 72,
590−641.
(22) Skorka, G.; Shuker, P.; Gill, D.; Zabicky, J.; Parola, A. H.
Biochemistry 1981, 20, 3103−3109.
(23) You, Z. B.; Wang, B.; Zitzman, D.; Azari, S.; Wise, R. A. J.
Neurosci. 2007, 27, 10546−10555.
(24) Parsons, L. H.; Justice, J. B. J. Neurochem. 1993, 61, 1611−1619.
(25) Westerink, B. H. C.; Kwint, H. F.; deVries, J. B. J. Neurosci.
1996, 16, 2605−2611.
ASSOCIATED CONTENT
■
S
* Supporting Information
(26) Redmond, J. W.; Tseng, A. J. Chromatogr. 1979, 170, 479−481.
(27) Novotny, M.; Alasandro, M.; Konishi, M. Anal. Chem. 1983, 55,
2375−2377.
Additional information as noted in text. This material is
available free of charge via the Internet at http://pubs.acs.org.
(28) Higa, S.; Suzuki, T.; Hayashi, A.; Tsuge, I.; Yamamura, Y. Anal.
Biochem. 1977, 77, 18−24.
AUTHOR INFORMATION
■
(29) Miki, K.; Sudo, A. Clin. Chem. 1998, 44, 1759−1762.
(30) Guo, K.; Li, L. Anal. Chem. 2009, 81, 3919−3932.
(31) Nestler, E. J. Nat. Neurosci. 2005, 8, 1445−1449.
(32) Koob, G. F.; Le Moal, M. Neurobiology of addiction; Elsevier:
Amsterdam, Netherlands, 2006.
Corresponding Author
*Address: Department of Chemistry, University of Michigan,
930 N. University Ave., Ann Arbor, MI 48109-1055. Phone:
734-615-4363. E-mail: rtkenn@umich.edu.
418
dx.doi.org/10.1021/ac202794q | Anal. Chem. 2012, 84, 412−419

Analytical Chemistry
Article
(33) Dobi, A.; Margolis, E. B.; Wang, H. L.; Harvey, B. K.; Morales,
M. J. Neurosci. 2010, 30, 218−229.
(34) Johnson, S. W.; North, R. A. J. Neurosci. 1992, 12, 483−488.
(35) Szabo, B.; Siemes, S.; Wallmichrath, I. Eur. J. Neurosci. 2002, 15,
2057−2061.
(36) Tan, K. R.; Brown, M.; Labouebe, G.; Yvon, C.; Creton, C.;
Fritschy, J. M.; Rudolph, U.; Luscher, C. Nature 2010, 463, 769−U78.
(37) Paxinos, G.; Watson, C. The rat brain in stereotaxic coordinates, 6
ed.; Elsevier: Amsterdam, Netherlands, 2007.
(38) Lipska, B. K.; Weinberger, D. R. Neuropsychopharmacology 2000,
23, 223−239.
(39) Parikh, V.; Kozak, R.; Martinez, V.; Sarter, M. Neuron 2007, 56,
141−154.
(40) Hasselmo, M. E.; Sarter, M. Neuropsychopharmacology 2010, 36,
52−73.
419
dx.doi.org/10.1021/ac202794q | Anal. Chem. 2012, 84, 412−419