Simultaneous profiling of multiple neurochemical pathways from a single cerebrospinal fluid sample using GC/MS/MS with electron capture detection

Article abstract of DOI:10.1002/jms.1376

Biogenic amines and amino acids are widely characterized in the pathways representing neurotransmission. Although several analytical methodologies have been used to detect specific target molecules in relevant fluids such as cerebrospinal fluid (CSF), multiple assays must be used to survey the primary pathways involved. This article describes the development of a GC/MS/MS method capable of analyzing up to 43 analytes (representing 20 amino acids and more than seven neurochemical pathways) from a single 50 μl CSF sample. In this procedure, a CSF sample is first treated with acetonitrile to precipitate proteins. The dried sample is then derivatized with a mixture of 2,2,3,3,3-pentafluoro-1-propanol and pentafluoropropionic acetic anhydride to replace all active hydrogen atoms with fluorine-containing groups. Due to the concentration difference between amino acids and neurotransmitters, these two compound classes are analyzed in separate injections of the same derivatized extract. The total run time for each injection is approximately 15-20 min. An essential feature of the method is the use of argon as a reagent gas for electron capture chemical ionization (ECCI), as the use of the more traditional gas (methane) lacked sufficient durability to be considered for use with the present instrumentation. This article describes the development of this method including a detailed investigation of the chemical ionization conditions used. The resultant conditions allow for the profiling of biogenic amines (e.g. serotonin, norepinephrine, and dopamine) in the low picogram per milliliter range. Copyright

Full text of DOI:10.1002/jms.1376

JOURNAL OF MASS SPECTROMETRY
J. Mass Spectrom. 2008; 43: 782–790
Published online 20 February 2008 in Wiley InterScience
(www.interscience.wiley.com) DOI: 10.1002/jms.1376
Simultaneous profiling of multiple neurochemical
pathways from a single cerebrospinal fluid sample
using GC/MS/MS with electron capture detection
James A. Eckstein,1∗ Gina M. Ammerman,2 Jessica M. Reveles3 and Bradley L. Ackermann1
1
Eli Lilly and Company, Greenfield Laboratories, Greenfield, IN 46140, USA
Department of Chemistry, Indiana University-Purdue University, Indianapolis, IN 46202, USA
Advanced Testing Management Group, Cincinnati, OH 45242, USA
2
3
Received 20 September 2007; Accepted 4 December 2007
Biogenic amines and amino acids are widely characterized in the pathways representing neurotransmission.
Although several analytical methodologies have been used to detect specific target molecules in relevant
fluids such as cerebrospinal fluid (CSF), multiple assays must be used to survey the primary pathways
involved. This article describes the development of a GC/MS/MS method capable of analyzing up to 43
analytes (representing 20 amino acids and more than seven neurochemical pathways) from a single 50 µl CSF
sample. In this procedure, a CSF sample is first treated with acetonitrile to precipitate proteins. The dried
sample is then derivatized with a mixture of 2,2,3,3,3-pentafluoro-1-propanol and pentafluoropropionic
acetic anhydride to replace all active hydrogen atoms with fluorine-containing groups. Due to the
concentration difference between amino acids and neurotransmitters, these two compound classes are
analyzed in separate injections of the same derivatized extract. The total run time for each injection is
approximately 15–20 min. An essential feature of the method is the use of argon as a reagent gas for
electron capture chemical ionization (ECCI), as the use of the more traditional gas (methane) lacked
sufficient durability to be considered for use with the present instrumentation. This article describes the
development of this method including a detailed investigation of the chemical ionization conditions used.
The resultant conditions allow for the profiling of biogenic amines (e.g. serotonin, norepinephrine, and
dopamine) in the low picogram per milliliter range. Copyright  2008 John Wiley & Sons, Ltd.
KEYWORDS: GC/MS/MS; profiling; neurotransmitters; amino acids; cerebrospinal fluid
fluid (CSF) is challenging.7 GC/MS followed by positive
detection has been used to detect neurotransmitters in
brain but would not have adequate sensitivity for CSF.8,9
One of the few techniques that exhibits the selectivity and
sensitivity to determine a wide diversity of molecules in CSF
is GC/MS/MS with electron capture chemical ionization
(ECCI). Since neither neurotransmitters nor amino acids are
innately sensitive to this method, a derivatization technique
adding electronegative groups is required.
INTRODUCTION
In the neurosciences, there is an important, expanding need
to profile targeted neurochemicalpathways. Historically, this
need has been filled with the utilization of liquid chromato-
graphy with electrochemical and fluorescence detection.1–6
Although these techniques have the sensitivity to detect
specific neurotransmitters, the scope of each method is
too limited to address the structural diversity of known
neurochemical pathways. Electrochemical detection is the
most widely used technique for neurotransmitters. While
this technique is very sensitive, column selection and
mobile phase adjustments are required to detect different
sets of metabolites. Analyzing the dopamine, epinephrine,
serotonin, and kynurenic acid pathways with a single
injection is not practical. LC/MS/MS can be used for
targeted analysis of select neurotransmitters or groups
of metabolites. Although the selectivity of LC/MS/MS is
enhanced compared to electrochemical detection, obtaining
the necessary sensitivity with small volumes of cerebrospinal
An unexploited derivatization for neurotransmitters
in CSF was utilized with detection by GC/ECCI.10 This
derivatization utilized
a mixture of pentafluoropropi-
onic acetic anhydride (PFAA) and a halogenated alcohol
(2,2,3,3,3,-pentafluoro-1-propanol). The resulting derivative
is extremely sensitive to electron capture (Fig. 1).
The two primary roles of the reagent gas in ECCI are to
produce thermal electrons to facilitate electron capture and to
stabilize the radical anion produced in the ionization process.
Methane is the most frequently used reagent gas for negative
ionization chemical ionization (NICI) and ECCI techniques,
and variables related to its use have previously been
investigated.11–13 Initial efforts utilizing methane with the
current instrumentation resulted in rapid signal deterioration
ŁCorrespondence to: James A. Eckstein, Eli Lilly and Company,
Greenfield Laboratories, Greenfield, IN 46140, USA.
E-mail: eckstein jim a@lilly.com
Copyright  2008 John Wiley & Sons, Ltd.

Detection of multiple neurochemicals 783
F
F
F
2,2,3,3,3 Pentafluoro-1-Propanol
Pentafluoroproprionic Anhydride
F
F
O
F
O
N O
80 °C
Pentafluoroproprionic Anhydride
80 °C
N O
F
F
O
F
O
O
O
F
O
F
F
F
O
F
F
F
O
F
F
L-DOPA
F
F
Figure 1. Derivatization procedure for GC/MS with ECCI detection of neurotransmitters and amino acids.
°
°
with complete failure of both detection and the autotune
process. To ameliorate this problem, argon was utilized
as the buffer gas. Utilization of this inert gas significantly
increased the durability of the method and allowed for
excellent sensitivity under optimized conditions.
60 to 300 C at 13 C/min. The temperature was then held at
°
300 C for 2.0 min, and the total run time was 21.46 min. The
oven temperature program for the amino acid assay began
at 60 C and was held for 1 min. The temperature was then
ramped in a linear fashion from 60 to 300 C at 20 C/min
°
°
°
°
This manuscript describes the detection of neurotrans-
mitters and amino acids in CSF using GC/MS/MS with ECCI
detection. The derivatization technique utilized converts
functional groups to fluorinated products that are sensitive
to ECCI detection. Argon, which had rarely been used in neg-
ative CI, was developed as the reagent gas. The conditions
necessary to utilize argon are also described. The sensitivity,
durability, and linearity of the technique are enhanced using
argon as the reagent gas with the current instrumentation.
The ability to profile a set of 43 analytes consisting of amino
acids and biogenic amines from a single 50 µl aliquot of CSF
is demonstrated.
and was held at 300 C for 2.0 min. The total run time was
°
15 min. The transfer line temperature was 280 C. The CI
source pressure was programmed at 3.2 Torr using argon as
the reagent gas, and the collision pressure was 1.3 mTorr of
°
argon. The source temperature was 120 C, and the manifold
°
temperature was 40 C. The mass spectrometer was tuned
and calibrated using FC-43 on m/z 283, 452, 595, and 633.
Derivatization procedure
The compounds were derivatized using a procedure orig-
inally developed by Watson in 1973.10 The compounds of
interest were dissolved in 1% formic acid with 0.1% ascorbic
acid used as an antioxidant. Compounds (20 µg) were dried
EXPERIMENTAL
°
in a silanized glass tube under nitrogen at 70 C. A total
of 240 µl of 2,2,3,3,3-pentafluoro-1-propanol and 270 µl of
PFAA was added to the dry sample. The sample was mixed
Materials
The instrumentation utilized was the 3800 Varian GC with
a Varian 1200 MS/MS (Varian Inc, Palo Alto, CA). Injec-
tions were performed by a Leap CTS Pal autosampler
(Leap Technologies, Carrboro, NC). The GC column was a
DB-5MS, 30 m ð 0.25 mm I.D. ð 0.25 µm film thickness from
Agilent Technologies (Folsom, CA). Formic acid was certi-
fied ACS from Thermo Fisher Scientific (Waltham, MA). The
derivatizing reagent PFAA was obtained from Pierce (Rock-
ford, IL), and 2,2,3,3,3-pentafluoro-1-propanol was acquired
from Sigma (St. Louis, MO). All glassware was deactivated
°
and heated at 80 C for 30 min. The mixture was then dried
°
at 70 C under nitrogen. Another 270 µl of PFAA was then
°
added, and the solution was heated at 80 C for 5 min. The
sample was dried and reconstituted in 2 ml of MTBE. This
derivatization process added up to 20 fluorines to molecules
such as L-DOPA (Fig. 1). For CSF samples, 50 µl of CSF was
added to an eppendorf tube or 96-well plate. A total of 150 µl
of cold 50% methanol/acetonitrile was added, and the sam-
ple was vortexed. The mixture was centrifuged to remove
protein, and the supernatant was transferred to a silanized
16 ð 100 mm glass tube. The solution was dried under nitro-
°
at 150 C in a vacuum oven using hexamethyldisilazane
(HMDS) obtained from Pierce Chemical Co. Ascorbic acid,
sodium dichromate, and all amino acids and neurotrans-
mitters were from Sigma (St. Louis, MO). Methanol and
acetonitrile were optima grade (Thermo Fisher Scientific),
and methyl tert-butyl ether (MTBE) chromsolv 99.9% was
from Sigma (St Louis, MO). Helium (99.9999% purity) and
argon (99.999% purity) were obtained from AGAnLinde gas
(Independence, OH).
°
gen at 70 C and then derivatized in the same manner as
described above. The final derivative was reconstituted in
100 µl of MTBE.
Determination of selected reaction monitoring
conditions
Full scan and product ion spectra were collected for
23 neurotransmitters including dopamine, epinephrine,
serotonin, and kynurenic acid pathway metabolites and 20
amino acids including GABA. Representative spectra for
derivatized neurotransmitters and amino acids are shown in
Figs 2 and 3. The product ion spectrum of the Mꢀž, [M ꢀ 20]
(loss of HF), or fragment ion was collected at collision
energies of 10, 20, and 30 eV. Selected reaction monitoring
transitions were chosen based on the most intense fragment
(Tables 1 and 2).
Instrument conditions
A volume of 1 µl was injected into a deactivated glass inlet
programmed for splitless injection at 280 C. The gas flow of
helium was programmed to pulse at 50 psi for 0.5 min and
then to maintain a flow of 1.2 ml/min for the remainder of
the chromatographic run. The oven temperature program for
the neurotransmitters began at 60 C and was held for 1 min.
The temperature was then ramped in a linear fashion from
°
°
Copyright  2008 John Wiley & Sons, Ltd.
J. Mass Spectrom. 2008; 43: 782–790
DOI: 10.1002/jms

784
J. A. Eckstein et al.
(A)
Norepinephrine MW = 753
(B)
100%
Serotonin MW = 614
753.3
594.0
100%
75%
50%
25%
0%
75%
50%
25%
0%
466.2
448.1 574.3
163.1
441.3
128.1
175.0 291.1
606.1
144.2
308.1
100 200 300 400 500 600 700 800
m/z
100 200 300 400 500 600
m/z
(C)
L-DOPA MW = 767
747.1
(D)
5HTP MW = 790
770.0
100%
75%
50%
25%
0%
100%
75%
50%
25%
0%
620.7
128.2
351.1
291.1
351.1
128.2
767
433.2
643.3
100 200 300 400 500 600 700 800
m/z
100 200 300 400 500 600 700 800
m/z
Figure 2. Representative spectra of neurotransmitters following derivatization and GC/MS with ECCI detection. The y axis
represents abundance.
(A)
Valine MW = 395
(B)
Methionine MW = 427
407.2
375.1
100%
100%
75%
50%
25%
0%
75%
50%
25%
0%
162.2
162.1
217.2
262.4
249.1
128.2
182.8
100
200
m/z
300
400
100
200
300
m/z
400
500
(C)
Tryptophan MW = 628
(D)
Phenylalanine MW = 443
423.1
100%
608.3
100%
75%
50%
25%
0%
75%
50%
25%
0%
351.0
371.2
128.0
280.1
189.1
200
284.2
458.3
588.2
100 200 300 400 500 600
m/z
100
300
m/z
400
500
Figure 3. Representative spectra of amino acids following derivatization and GC/MS with ECCI detection. The y axis represents
abundance.
Copyright  2008 John Wiley & Sons, Ltd.
J. Mass Spectrom. 2008; 43: 782–790
DOI: 10.1002/jms

Detection of multiple neurochemicals 785
Table 1. Conditions for NCI detection of neurotransmitter derivatives by GC/MS
MW
Typea (amu) Q1 (m/z)
Product
(m/z)
Collision Retention
Compound
Common name
(V)
(min)
Pyridine-2-carboxylic acid
2,3-Pyridinedicarboxylic acid
3,4, Dihydroxyphenyl glycol
3,4,-Dihydroxyphenylacetic acid
4-Hydroxy-3- methoxymandelic acid
4-Hydroxy-3-methoxyphenylacetic acid
4-Hydroxy-3-methoxyphenylglycol
Norepinephrine
3,4-Dihydroxymandelic acid
Epinephrine
L-DOPA
Dopamine
Normetanephrine
2-Picolinic acid
Quinolinic acid
DHPG
K
K
D
D
E, NE
D
255
431
740
592
622
460
608
753
740
767
767
591
621
635
459
615
453
467
794
632
790
614
337
255.1
431.1
593.3
444.3
459.3
312.2
608.2
753.3
593.2
767.1
747.2
571.2
621.2
615.6
439.0
595.0
453.1
467.1
630.1
612.2
770.0
594.2
337.1
122.1
298.0
162.9
175.0
309.0
162.1
147.1
163.0
442.9
162.9
375.9
375.9
163.0
163.1
308.0
467.0
320.0
320.0
218.9
218.0
351.1
464.8
160.1
10
10
10
20
15
20
10
10
10
10
10
10
10
10
10
15
10
20
10
10
15
10
20
7.30
9.17
9.41
DOPAC
VMA
HVA
9.51
10.09
10.18
10.18
10.50
10.68
10.71
10.90
11.12
11.29
11.67
11.78
12.15
12.33
12.61
12.64
12.82
13.09
13.65
15.70
MHPG
D
NE
NE
E
D
D
NE
E
D
S
K
K
K
K
S
S
K
Metaneprhine
3-Methoxy-4-hydroxyphenethylamine
5-Hydroxyindole-3-acetic acid
Kynurenic acid
8-Hydroxyquinoline-2-carboxylic acid
3-Hydroxy-kynurenine
Kynurenine
5-Hydroxy-tryptophan
Serotonin
4,8-Dihydroxyquinoline-2-carboxylic acid
3-O-Methyldopamine
5 HIAA
5 HTP
Xanthrenic acid
a K, Kynurenine pathway; E, epinephrine; NE, Norepinephrine; D, Dopamine; S, Serotonin
Table 2. Conditions for electron capture detection of amino acid derivatives by GC/MS
MW
(amu)
Q1
(m/z)
Product
(m/z)
Collision
Retention
(min)
Compound
Abbreviation
(V)
Alanine
Glycine
Threonine
Valine
Serine
Isoleucine
Leucine
Aspartate
Cysteine
ꢀ-Aminobutyric acid
Proline
Glutamine
Asparagine
Glutamate
Methionine
Phenylalanine
Tyrosine
Lysine
Arginine
Tryptophan
Ala
Gly
Thr
Val
Ser
Ile
Leu
Asp
Cys
GABA
Pro
Gln
Asn
Glu
Met
Phe
Tyr
Lys
Arg
Trp
367
353
543
395
529
409
409
543
545
381
393
570
556
557
427
443
605
570
744
628
347.1
333
523.1
375.1
509
197
313
377
225.1
145
10
10
10
10
10
10
10
20
10
10
10
10
20
20
20
20
20
20
20
20
3.94
4.45
4.84
4.86
5.15
5.60
5.68
6.53
6.55
6.61
6.87
7.38
7.63
7.69
8.07
8.98
10.17
10.22
11.06
11.93
389
389
239
239
523.3
525.1
341.1
393
550.3
536.2
537.2
407.2
423.2
585.2
550.3
704.3
608.2
353.3
364.3
208.1
352.9
422
211
313
190
273
268
340
643.9
350.9
Copyright  2008 John Wiley & Sons, Ltd.
J. Mass Spectrom. 2008; 43: 782–790
DOI: 10.1002/jms

786
J. A. Eckstein et al.
Argon pressure optimization
Argon source pressure was optimized for maximum sensi-
tivity for ECCI detection. The source temperature was main-
(A)
Argon m/z = 40
3e+7
3e+7
2e+7
2e+7
1e+7
5e+6
0
°
tained at 150 C for these experiments. Initially, the argon
source pressure was optimized on the response of the FC-43
tune ion at m/z 452. Further experience indicated that this
was not an appropriate method; therefore, pressure was opti-
mized on representative compounds for neurotransmitters
and amino acids. These experiments were performed in both
selected reaction monitoring mode (SRM) and full scan mode.
In SRM mode, nine analytes including GABA, 3-hydroxy-4-
methoxyphenethylamine, dopamine, serotonin, tryptophan,
kynurenine, arginine, L-DOPA, and epinephrine were used
to optimize the argon pressure. For full scan optimiza-
tion, GABA, methionine, valine, phenylalanine, tryptophan,
dopamine, L-DOPA, serotonin, epinephrine, and DOPAC
were used. The full scan of the derivatized compounds was
collected scanning from m/z 50 to 800. For both experiments,
the argon source pressure was adjusted in increments of
0.2 Torr from 1.95 to 5.2 Torr and a 1-µl (20 pg or 200 pg)
injection was made with the compounds of interest. The area
of the peak in SRM representing each compound was plotted
against source pressure (Fig. 4).
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
Source Pressure (torr)
(B)
Argon Dimer m/z = 80
8e+5
7e+5
6e+5
5e+5
4e+5
3e+5
2e+5
1e+5
0
Optimization of argon pressure utilizing the ratio
of Ar to Ar2
Source pressure may be monitored by various techniques
with different mass spectrometers. Therefore, argon pressure
was optimized with a more universal method utilizing the
ratio of Ar to the argon dimer (Ar2). For this experiment, a
full scan method was used monitoring from m/z 10 to 100 in
profile mode. The argon pressure was altered by increments
of 0.2 Torr from 1 to 4.5 Torr. The peaks for argon (m/z D 40)
and the argon dimer (m/z D 80) were integrated and plotted
against source pressure (n D 4 for determination at each
pressure) (Fig. 5).
0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5
Source Pressure (torr)
(C)
Argon Ratio m/z-40/m/z-80 Dimer
100
80
60
40
20
0
Optimized ratio for Argon ECCI = 9.5
Optimization for source temperature, filament eV,
and amperage
It is well known that ECCI can be sensitive to source
temperature, electron filament voltage, and amperage. These
parameters can alter spectral ratios and electron capture
0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5
Source Pressure (torr)
GABA
Figure 5. Argon optimum source pressure determined by
relative amounts of Ar (m/z D 40), Ar2 (m/z D 80), and the ratio
of argon to its dimer (n D 4 for each determination).
3 OH 4 methoxy phenethylamine
Dopamine
Serotonin
1e+10
1e+9
Tryptophan
Kynurenine
1e+8
1e+7
1e+6
1e+5
1e+4
1e+3
1e+2
1e+1
1e+0
1e-1
Arginine
L-DOPA
efficiency by altering reaction rate and availability of
electrons for the electron capture process. These parameters
were tested using 200-pg injections of five amino acids
(GABA, methionine, valine, phenylalanine, and tryptophan)
and five neurotransmitters (dopamine, serotonin, L-DOPA,
epinephrine, and DOPAC). The source temperature was set
Epinephrine
Optimization with ion 452 (Cal gas)
Normal Range for NCI (Methane)
10 11 12
°
at 100, 150, 200, and 250 C. Four injections were made at each
1e-2
0
1
2
3
4
5
6
7
8
9
temperature. Compounds were monitored in full scan from
m/z 50 to 800 to detect both the intensity of ions and spectral
changes. The effect of electron voltage was determined by
monitoring ion intensities at 20, 70, and 150 eV. The effect of
Source Pressure (torr)
Figure 4. Optimization of argon in SRM analysis following
derivatization and GC/MS with ECCI detection.
Copyright  2008 John Wiley & Sons, Ltd.
J. Mass Spectrom. 2008; 43: 782–790
DOI: 10.1002/jms

Detection of multiple neurochemicals 787
filament amperage was determined in a similar manner by
varying the amperage from 50 to 500 µA.
cleaning the ion source. The procedure classically used for
cleaning the ion volume involves removal of contamination
with a paste of wenol (metal polishing compound). Although
this process is followed by multiple washes in water,
methanol, and hexane, it exposes the surface to free iron
that is reactive. In the new process termed passivation, the
stainless undergoes a chemical neutralization of the surface
using 50% nitric acid with 0.01% sodium dichromate. In this
process, the ion volumes are incubated in the passivation
solution for 2 days at room temperature, and the iron is
replaced with a layer of chromium that is more inert. Figure 8
represents the durability obtained from the use of argon
and the passivation method for cleaning the ion volume.
After a total of 125 injections of representative amino acids
and neurotransmitters, a successful tune was obtained to
illustrate instrument performance.
Linearity of response
The linearity of response was determined in rat CSF by
serial dilution of stable labeled amino acids aspartate d3
and phenylalanine d8 and the neurotransmitters L-DOPA
d3 and dopamine d4. Linearity for amino acids was deter-
mined from 1 to 125 ng/ml. Linearity for neurotransmitters
was determined from 0.05 to 65 ng/ml. Standard curves
were analyzed as per the instrument parameter methods
previously discussed. Standard curves were plotted as area
of analyte with no internal standard versus concentration
(Fig. 6).
Durability of response versus methane
Methane is the most extensively used reagent gas for
NCI GC/MS; however, this reagent gas caused rapid
failure in response to NCI GC/MS on the instrument
utilized. This response failure was documented with the
amino acid phenylalanine, which was found to be a very
stable derivative using the method presented (Fig. 7). The
durability of the method presented is due to the utilization
of argon as the reagent gas as well as a revised method for
Chromatographic resolution and selectivity
To determine chromatographic/mass spectral resolution
and selectivity, a sample of rat CSF was analyzed using
the derivatization and optimum argon conditions for ECCI
detection. Chromatograms of representative neurotransmit-
ters and amino acids found in CSF are shown in Figs 9
and 10.
y = 698003x - 3E+06
(A)
RESULTS AND DISCUSSION
2
9.00E+07
8.00E+07
7.00E+07
6.00E+07
5.00E+07
4.00E+07
3.00E+07
2.00E+07
1.00E+07
1.00E-01
R
= 0.9979
Aspartate d
3
Preliminary experience with ECCI
Phenylalanine d
8
A technique to detect amino acids and neurotransmitters
in CSF has been developed by GC/MS/MS with ECCI
detection. Preliminary detection of the analytes was achieved
with methane as the reagent gas at 7–10 Torr. With
the current instrumentation (Varian 1200 MS/MS), using
methane as the reagent gas caused contamination of the ion
source and pre-octapole lenses. This contamination led to
complete signal failure and inability to autotune with FC-
43. Upon cleaning the ion source and lenses 1, 2, and 3,
the response returned but deteriorated in the same fashion
upon reanalysis. This deterioration occurred after as few as 10
y = 218507x - 2E+06
2
R
= 0.9856
0
20
40
60
80
100
120
140
Conc. ng/ml
(B)
y = 1E+06x + 3E+06
2
Phenylalanine (Run 1)
1.00E+08
9.00E+07
8.00E+07
7.00E+07
6.00E+07
5.00E+07
4.00E+07
3.00E+07
2.00E+07
1.00E+07
0.00E+00
R
= 0.9928
Dopamine d
Phenylalanine (Run 2)
1e+10
4
L-DOPA d
3
1e+9
1e+8
1e+7
1e+6
y = 708601x + 582335
2
R
= 0.997
0
10
20
30
40
50
60
70
0
10
20
30
40
50
60
Conc. (ng/ml)
Injection
Figure 6. Representative linearity of amino acids and
neurotransmitters in rat CSF using ECCI with argon as the
reagent gas (n D 2 at each concentration). The data shown
were not corrected using an internal standard.
Figure 7. Durability of signal at two different concentrations
with the use of methane as the reagent gas. (The instrument
had to be cleaned and retuned after each run).
Copyright  2008 John Wiley & Sons, Ltd.
J. Mass Spectrom. 2008; 43: 782–790
DOI: 10.1002/jms

788
J. A. Eckstein et al.
shape. Examination of individual full scan mass spectra
across chromatographic peaks revealed changes in mass
spectralcomposition, suggestingthatcomplexreactionswere
occurring in the CI source ion volume. This phenomenon
prevented the use of ammonia as a reagent gas.
GABA
Valine
Methionine
Phenylalanine
Tryptophan
(A)
1e+8
1e+7
1e+6
1e+5
1e+4
1e+3
Utilization of argon as reagent gas
Few investigators have used argon as the reagent gas in
negative detection mode. Some found that utilizing argon as
the reagent gas resulted in a 30-fold reduction in sensitivity
versus methane.14,15 Other investigators found a striking
enhancement in negative ions when argon pressure was
increased in an EI source.16–19 With the instrumentation used
for current analysis, the inert gas argon gave no response
when used at the same pressure as methane (7–10 Torr). The
argon pressure was initially optimized with the response of
the m/z 452 ion (FC-43 calibration mix). However, optimum
pressure for argon was further refined by using a selection
of compounds detected in SRM (Fig. 4) and full scan mode.
Such analysis indicated an optimum pressure of argon at
3.2 Torr. Since most instruments use different methods to
measure source pressure, optimization was also determined
by measuring the ratio of Ar to Ar2 (Fig. 5). The optimized
ratio (Ar/Ar2) for ECCI was 9.5. This ratio was reproduced
several times when the source pressure was reset to its
optimum pressure of 3.2 Torr.
0
20
40
60
80
100 120 140
Number of Injections
DOPAC
Dopamine
Serotonin
Epinephrine
L-dopa
(B)
1e+9
1e+8
1e+7
1e+6
1e+5
1e+4
1e+3
Determination of SRM conditions
0
20
40
60
80
100 120 140
Conditions for SRM of amino acids and neurotransmitters are
shown in Tables 1 and 2. Representative spectra are shown
in Figs 2 and 3. The full scan data for the neurotransmitters
(Table 1, Fig. 2) demonstrate that nine compounds gave the
Mꢀž radical ion for the Q1 scan, eight compounds gave
the [M ꢀ 20] ion, and six compounds gave fragment ions
related to the derivatized structure. In contrast, for amino
acids (Table 2, Fig. 3), 17 compounds gave [M ꢀ 20] ions
for the Q1 scan, 2 compounds gave the [M ꢀ 40] ion, and
1 compound gave the Mꢀž radical ion. The spectra for all
representative compounds are relatively ‘soft’ spectra with
minimal fragmentation (Figs 2 and 3). The retention time of
each analyte is included in this data.
Number of Injections
Figure 8. Durability of signal with the use of argon as the
reagent gas for amino acids (A) and neurotransmitters (B). The
total number of injections was 125.
Signal/Noise
1.3e4/1
4.7e1/1
DOPAC
VMA
MHPG
2.4e4/1
1.1e6/1
Effect of source temperature, filament eV, and
amperage
The effect of source temperature on the ECCI detection
of amino acids and neurotransmitters was investigated. In
general, the response of amino acids increased with an
increase in temperature. However, the lowest molecular
weight amino acids (glycine and valine) were not detected
Dopamine
1.3e3/1
1.2e7/1
DHPG
°
at higher temperatures of 200 and 250 C. In contrast, the
L-DOPA
10.5
response for neurotransmitters tended to decrease with an
increase in temperature. Because sensitivity is most vital for
neurotransmitters, the optimum temperature was found to
9.5
10.0
11.0
Minutes
°
be 120 C. The effect of filament (eV) was determined at 20,
Figure 9. GC/MS/MS ECCI detection of representative
70, and 150 eV. In general, extremely low or no response
was seen at 20 and 150 eV. The filament set to 70 eV was
capable of producing the thermalized electrons needed for
ECCI with argon pressure at 3.2 torr. Consistently, 50 µA
gave the optimum response with most analytes.
neurotransmitters in rat CSF following derivatization.
injections. Ammonia gas was initially used to circumvent this
problem; however, this reagent gas gave very broad peak
Copyright  2008 John Wiley & Sons, Ltd.
J. Mass Spectrom. 2008; 43: 782–790
DOI: 10.1002/jms

Detection of multiple neurochemicals 789
Signal/Noise
Signal/Noise
Glycine
1.5e5/1
2.9e6/1
1.4e6/1
4.8e3/1
2.6e6/1
2.3e5/1
7.3e5/1
6.1e5/1
3
9.2e5/1
8.8e6/1
1.5e7/1
1.5e6/1
4.7e5/1
6.9e5/1
2.6e5/1
Methionine
Phenylalanine
Glutamate
Alanine
Leucine
Isoleucine
Serine
Lysine
Threonine
Tyrosine
Proline
Tryptophan
Aspartate
Cysteine
Arginine
10 11 12
Minutes
4
5
6
7
7
8
9
Minutes
Figure 10. GC/MS/MS ECCI detection of amino acids following derivatization in rat CSF.
same CSF sample because the preparation and derivatization
are identical for both amino acids and neurotransmitters.
Linearity of response
Linearity of response was determined for several amino
acids and neurotransmitters using argon as the reagent
gas. Representative linearity in rat CSF using stable labeled
compounds is shown in Fig. 6. The compounds L-DOPA d3
and dopamine d4 show a linear response from 50 pg/ml
to 62.5 ng/ml, and the amino acids phenylalanine d8 and
aspartate d3 show linearity from 1 to 125 ng/ml.
CONCLUSIONS
Neurotransmitters and amino acids are important as
biomarkers for the neurosciences. In many cases, the matrix
that is available to analyze, such as microdialysate or CSF,
contains only low concentrations of these compounds and
is available in only small volumes. Therefore, sensitive and
selective techniques are required to determine the concen-
tration of these compounds. Advantages to the technique
outlined in this manuscript include the low volume required
for analysis, the resolution power of GC, the selectivity of
MS/MS, and the sensitivity of ECCI. In addition, the deriva-
tization technique used for this method was successful for
a wide class of molecules. It allowed for conversion of phe-
nols, amines, carboxylic acids, and in some cases, aliphatic
hydroxyl groups (epinephrine, norepinephrine) to highly
fluorinated derivatives, which allowed for a robust ECCI
signal.
Methane, which is the most widely used reagent gas,
caused failure in the ECCI detection due to contamination
of the ion volume and pre-octapole lenses. However, argon
was capable of thermalizing electrons for the ECCI detection
process. Its utilization gave an ECCI process that was almost
an entirely dissociative electron capture for amino acids and
a mixture of resonance and dissociative electron capture for
neurotransmitters. The detection of both amino acids and
neurotransmitters was sensitive, durable, and linear over
a suitable range for negative ion detection. Argon pressure
optimization was accomplished by using the ratio of argon to
its dimer in a manner that is transferable to instruments that
may have different design. However, the effects of source
design on the utilization of this inert gas is not known.
Although targeted profiling is not strictly a quantitative
method, a validation was completed utilizing stable labeled
internal standards in CSF. This data, along with information
on the procedure for running study samples, quality
controls, and semiquantitation, will be presented in a future
Durability of response
The durability of response previously determined with
methane is shown in Fig. 8. On two separate occasions,
the analytical sequence was started and found to fail after
20 and 50 injections of derivatized phenylalanine. The
signal decreased drastically as the run progressed, and
the instrument had to be cleaned, as unsymmetrical burn
in the ion volume and pre-octapole lenses prevented ion
transmission. In addition, the instrument would not tune
with FC-43. In contrast, with argon as the reagent gas, after
100–200 injections the ion burn was minimal and symmetric,
and the instrument successfully tuned with FC-43. Figure 8
shows the response of amino acids and neurotransmitters
following 125 injections with argon as the reagent gas.
Although there were some fluctuations of the signal, the
instrument remained operational at the end of the run and
was successfully tuned. The cleaning process with nitric acid
and chromium is known to remove iron from the surface
of stainless steel and replace it with a more inert chromium
layer. When this method was utilized, there was significantly
less corrosion and/or material deposits in the ion volume
following several hundred injections. In addition, the spectra
were consistently ‘softer’ with less fragmentation following
this treatment.
Chromatographic resolution and selectivity
Chromatography of amino acids and neurotransmitters
in CSF yielded sharp symmetrical peaks with excellent
separation (Figs 9 and 10). No interferences were apparent
in the SRM channels. Although the GC and MS methods are
different, both sets of compounds can be determined from the
Copyright  2008 John Wiley & Sons, Ltd.
J. Mass Spectrom. 2008; 43: 782–790
DOI: 10.1002/jms

790
J. A. Eckstein et al.
quantitation of tBDMS derivatives using ammonia positive
chemical ionization. Journal of Chromatography B 2006; 831: 313.
10. Watson E, Wilk S. Derivatization and gas chromatographic
determination of some biologically important acids in
cerebrospinal fluid. Analytical Biochemistry 1974; 59: 441.
11. Stemmler EA, Hites RA. A systematic study of instrumental
parameters affecting electron capture negative ion mass spectra.
Biomedical and Environmental Mass Spectrometry 1988; 15: 659.
12. Stemmler EA, Hites RA, Arbogast B, Budde WL, Deinzer ML,
Dougherty RC, Eichelberger JW, Foltz RL, Grimm C, Grim-
srud EP, Sakashita C, Sears LJ. Interlaboratory comparison of
methane electron capture negative ion mass spectra. Analytical
Chemistry 1988; 60: 781.
13. Leis HJ, Fauler G, Rechberger GN, Windischhofer W. Electron-
Capture Mass Spectrometry: A powerful tool in biomedical trace
level analysis. Current Medicinal Chemistry 2004; 11: 1585.
14. Chernetsova ES, Revelsky AI, Revelsky IA, Mikhasenko IA,
Sobolevsky TG. Determination of polychlorinated dibenzo-P-
dioxins, dibenzofurans, and biphenyls by gas chromatogra-
phy/mass spectrometry in the negative chemical ionization
mode with different reagent gases. Mass Spectrometry Reviews
2002; 21: 373.
15. Harrison AG. Chemical Ionization Mass Spectrometry. CRC Press:
New York, 1992; 49.
16. Sears LJ, Grimsrud EP. Elimination of unexpected ions in
electron capture mass spectrometry using carbon dioxide buffer
gas. Analytical Chemistry 1989; 61: 2523.
17. Szulejko JE, Howe I, Beynon JH, Schlunegger UP. Negative ion
mass spectrometry: the generation of high concentrations of low
energy molecular negative ions at high source pressures. Organic
Mass Spectrometry 1980; 15: 263.
18. Lauber R, Schlunegger UP. Enhancement of negative ion
currents by argon/CO2 mixtures in an electron impact ion
source. Organic Mass Spectrometry 1986; 21: 401.
manuscript. The method will be used as a general profiling
procedure for biomarker assessment of pharmacology and
toxicology in the neurosciences.
REFERENCES
1. Kissinger PT, Bruntlett CS, Shoup RE. Neurochemical applica-
tions of liquid chromatography with electrochemical detection.
Life Sciences 1981; 28: 455.
2. Mefford IN. Application of high performance liquid chromatog-
raphy with electrochemical detection to neurochemical analysis:
measurement of catecholamines, serotonin and metabolites in
rat brain. Journal of Neuroscience Methods 1981; 3: 207.
3. Mefford IN. Biomedical uses of high-performance liquid
chromatography with electrochemical detection. Methods of
Biochemical Analysis 1985; 31: 221.
4. Yi Z, Brown PR. Chromatographic methods for the analysis of
basic neurotransmitters and their acidic metabolites. Biomedical
Chromatography 1991; 3: 101.
5. Kema IP, de Vries EG, Muskiet FA. Clinical chemistry of
serotonin and metabolites. Journal of Chromatography. B,
Biomedical Sciences and Applications 2000; 747: 33.
6. Peaston RT, Weinkove C. Measurement of catecholamines and
their metabolites. Annals of Clinical Biochemistry 2004; 41: 17.
7. Hows EP, Lacroix ML, Heidbreder C, Organ AJ, Shah AJ.
High-performance liquid chromatography/tandem mass
spectrometric assay for the simultaneous measurement of
dopamine, norepinephrine, 5-hydroxytryptamine and cocaine
in biological samples. Journal of Neuroscience Methods 2004; 138:
123.
8. Loutelier B, Legros CH, Bonnet JJ, Costentin J, Lange CM.
Gas chromatography/mass spectrometric identification of
dopaminergic metabolites in striata of rats treated with L-DOPA.
Rapid Communications in Mass Spectrometry 2004; 18: 571.
9. Wood PL, Khan MA, Moskal JR. Neurochemical analysis of
amino acids, polyamines and carboxylic acids: GC-MS
19. Laramee JA, Arbogast BC, Deinzer ML. Electron capture
negative ion chemical ionization mass spectrometry of 1,2,3,4-
tetrachlorodibenzo-p-dioxin. Analytical Chemistry 1986; 58: 2907.
Copyright  2008 John Wiley & Sons, Ltd.
J. Mass Spectrom. 2008; 43: 782–790
DOI: 10.1002/jms