Strategy for the isolation, derivatization, chromatographic separation, and detection of carnitine and acylcarnitines

Article abstract of DOI:10.1021/ac0487810

A strategy for detection of carnitine and acylcarnitines is introduced. This versatile system has four components: (1) isolation by protein precipitation/desalting and cation-exchange solid-phase extraction, (2) derivatization of carnitine and acylcarniti

Full text of DOI:10.1021/ac0487810

Anal. Chem. 2005, 77, 1448-1457
Strategy for the Isolation, Derivatization,
Chromatographic Separation, and Detection of
Carnitine and Acylcarnitines
†
‡
,†,‡,§
Paul E. Minkler, Stephen T. Ingalls, and Charles L. Hoppel*
Medical Research Service, Louis Stokes Department of Veterans Affairs Medical Center, Cleveland, Ohio 44106, and
Case Comprehensive Cancer Center and the Departments of Pharmacology and Medicine, Case Western Reserve
University School of Medicine, Cleveland, Ohio 44106
A strategy for detection of carnitine and acylcarnitines is
introduced. This versatile system has four components:
Other roles for carnitine include its mass action effect in establish-
ing the steady-state intramitochondrial acyl coenzyme A/coen-
zyme A ratio2 and peroxisomal fatty acid oxidation. These
functions are accomplished through the action of acyltransferases,
which produce carnitinyl esters (acylcarnitines).
3
(
1) isolation by protein precipitation/desalting and cation-
exchange solid-phase extraction, (2) derivatization of
carnitine and acylcarnitines with pentafluorophenacyl
trifluoromethanesulfonate, (3) sequential ion-exchange/
reversed-phase chromatography using a single non-end-
Methods for the detection and quantification of carnitine have
used a variety of strategies.4 Originally, Marquis and Fritz
proposed a method for the accurate quantification of carnitine that
exploited its recognition by carnitine acetyltransferase. This
approach was further refined by using radioactive acetyl coenzyme
5
8
capped C column, and (4) detection of carnitine and
acylcarnitine pentafluorophenacyl esters using an ion trap
mass spectrometer. Recovery of carnitine and acylcar-
nitines from the isolation procedure is 77-85%. Deriva-
tization is rapid and complete with no evidence of acyl-
carnitine hydrolysis. Sequential ion-exchange/reversed-
phase HPLC results in separation of reagent byproducts
from derivatized carnitine and acylcarnitines, followed by
reversed-phase separation of carnitine and acylcarnitine
pentafluorophenacyl esters. Detection by MS/MS is highly
selective, with carnitine pentafluorophenacyl ester yielding
a strong product ion at m/z 311 and acylcarnitine pen-
tafluorophenacyl ester fragmentation yielding two product
ions: (1) loss of m/z 59 and (2) generation of an ion at
m/z 293. To demonstrate this analytical strategy, phos-
phate buffered serum albumin was spiked with carnitine
and 15 acylcarnitines and analyzed using the described
protein precipitation/desalting and cation-exchange solid-
phase extraction isolation, derivatization with pentafluo-
rophenacyl trifluoromethanesulfonate, chromatography
using the sequential ion-exchange/reversed-phase chro-
matography HPLC system, and detection by MS and
MS/MS. Successful application of this strategy to the
quantification of carnitine and acetylcarnitine in rat liver
is shown.
6
A. Due to technological advances, other more recent approaches
7
have been developed employing HPLC/UV, flow injection analy-
8
9
10
sis, tandem MS, or HPLC/tandem MS.
Determination of acylcarnitines is challenging and has been
addressed using a variety of chromatographic, derivatization, or
1
1
detection schemes including radioisotopic exchange/HPLC,
12
13
14
15
16
HPLC/UV, HPLC/fluorescence, GC/MS, CE, CE/MS, and
HPLC/tandem MS.17 However, today the most common method
used to analyze acylcarnitines begins with the esterification of
acylcarnitines by means of high-temperature acylation followed
(1) McGarry, J. D.; Brown, N. F. Eur. J. Biochem. 1997, 244, 1-14.
(2) Brass, E. P.; Hoppel, C. L. Biochem. J. 1980, 190, 495-504.
(3) Jakobs, B. S.; Wanders, R. J. A. Biochem. Biophys. Res. Commun. 1995,
2
13, 1035-1041.
(4) Marzo, A.; Curti, S. J. Chromatogr. 1997, 702, 1-20.
(
(
(
5) Marquis, N. R.; Fritz, I. B. J. Lipid Res. 1964, 41, 184-187.
6) McGarry, J. D.; Foster, D. W. J. Lipid Res. 1976, 17, 277-281.
7) Minkler, P. E.; Hoppel, C. L. Clin. Chim. Acta 1992, 212, 55-64.
(8) Manjon, A.; Obon, J. M.; Iborra, J. L. Anal. Biochem. 2000, 281, 176-181.
9) Kodo, N.; Millington, D. S.; Norwood, D. L.; Roe, C. R. Clin. Chim. Acta
989, 186, 383-390.
(
1
(10) Vaz, F. M.; Melegh, B.; Bene, J.; Cuebas, D.; Gage, D. A.; Bootsma, A.;
Vreken, P.; van Gennip, A. H.; Bieber, L. L.; Wanders, R. J. A. Clin. Chem.
2
002, 48, 826-834.
11) Schmidt-Sommerfeld, E.; Zhang, L.; Bobrowski, P. J.; Penn, D. Anal. Biochem.
995, 231, 27-33.
(12) Poorthuis, B. J. H. M.; Jille-Vlckova, T.; Onkenhout, W. Clin. Chim. Acta
993, 216, 53-61.
13) Kamimori, H.; Hamashima, Y.; Konishi, M. Anal. Biochem. 1994, 218, 417-
24.
(
1
Carnitine (3-hydroxy-4-(N,N,N-trimethylammonio)butanoate)
is a low molecular weight trimethylammonio carboxylate re-
quired for the mitochondrial oxidation of long-chain fatty acids.1
1
(
(
4
14) Libert, R.; Van Hoof, F.; Thillaye, M.; Vincent, M-F.; Nassogne, M-C.;
Stroobant, V.; de Hoffmann, E.; Schanck, A. Anal. Biochem. 1997, 251,
196-205.
*
To whom correspondence should be addressed. Phone: (216) 791-3800,
×
5657. Fax: (216) 707-5973. E-mail: charles.hoppel@case.edu.
†
Louis Stokes Department of Veterans Affairs Medical Center.
(15) Vernez, L.; Thormann, W.; Krahenbuhl, S. J. Chromatogr. 2000, 895, 309-
‡
Case Comprehensive Cancer Center, Case Western Reserve University
316.
School of Medicine.
Departments of Pharmacology and Medicine, Case Western Reserve
University School of Medicine.
(16) Heinig, K.; Henion, J. J. Chromatogr. 1999, 735, 171-188.
(17) Tallarico, C.; Pace, S.; Longo, A. Rapid Commun. Mass Spectrom. 1998,
12, 403-409.
§
1448 Analytical Chemistry, Vol. 77, No. 5, March 1, 2005
10.1021/ac0487810 CCC: $30.25 © 2005 American Chemical Society
Published on Web 02/03/2005

by detection using tandem MS.18,19 Primarily due to its simplicity
and rapidity, this approach has been applied with much success
to programs for newborn screening and contributed greatly to
our understanding of the role of acylcarnitines in metabolic
disorders.
Lancaster Synthesis (Windham, NH). Elemental analysis was
performed by Galbraith Laboratories (Nashville TN). NMR spectra
were obtained with a Varian (Palo Alto, CA) INOVA 600-MHz
instrument equipped with an HX probe tunable to 13C and F. All
spectra were obtained using d-chloroform solutions (20 mg/mL
2
0
19
1
The development of the benchtop electrospray ionization/ion
trap MS and its coupling to HPLC has proven to be a substantial
at 25 °C). H spectra were acquired with a 45° pulse after a delay
1
of 1 s, and eight transients were collected per spectrum. H
2
1
1
technical innovation. Previously, we reported HPLC methods
chemical shifts were calculated relative to H-chloroform, which
2
2
13
1
for the quantification of carnitine and acylcarnitines in urine,
was set to 7.24 ppm. All C spectra were H-decoupled, acquired
with a 30° pulse after a delay of 1 s, and 256 transients were
collected per spectrum. These were processed with 1-Hz line
2
3
24
plasma, and tissue. From this knowledge base, we concluded
that combining the ion trap MS with an HPLC approach would
allow the sensitive and selective detection of carnitine and acyl-
carnitines. This combination would have the ability to accom-
modate complex mixtures and distinguish isomeric acylcarnitines.
These capabilities are limited with tandem MS.
1
3
broadening and chemical shifts calculated relative to C-chloro-
form, which was set to 77.0 ppm. All 19F spectra were acquired
with a 45° pulse after a delay of 1 s, processed with 0.5-Hz line
1
9
broadening, and 16 transients were collected per spectrum.
F
This report describes and illustrates the four essential com-
ponents of this system: (1) isolation of carnitine and acylcarnitines
by protein precipitation/desalting and silica gel cation-exchange
solid-phase extraction, (2) derivatization of carnitine and acylcar-
nitines with the new reagent pentafluorophenacyl trifluoromethane-
sulfonate, (3) sequential ion-exchange/reversed-phase chroma-
tography of carnitine and acylcarnitine esters, and (4) MS and
MS/MS detection. Examples of the chromatographic behavior,
performance of the derivatization reagent, and MS characteristics
of carnitine and acylcarnitines using this procedure are shown.
chemical shifts were calculated relative to R,R,R-trifluorotoluene,
which was set to -62.80 ppm. 4′-Bromophenacyl trifluorometh-
anesulfonate was synthesized as described.29 Solid-phase extrac-
tion columns (50 mg of silica Bond Elute) were purchased from
Varian (Walnut Creek, CA) and prepared by washing with 0.5 mL
of methanol (gravity flow) just before their use. 4′-Bromophenacyl
esters of carnitine, 4-(trimethylammonio)butanoic acid, and 6-(tri-
2
6
methylammonio) hexanoic acid were synthesized as described.
Bovine albumin fraction V solution 7.5% in phosphate-buffered
saline was purchased from GIBCO BRL/Life Technologies (Grand
Island, NY).
Synthesis of Pentafluoro-2-diazoacetophenone.30,31 Cau-
tion! Syntheses involving diazomethane should be approached with
EXPERIMENTAL SECTION
Materials and Sources. Acetonitrile, methanol, acetic acid,
and triethylamine were all HPLC grade and purchased from Fisher
-
1
great care! Diazomethane was prepared from 60 g (2.8 × 10
Scientific (Cleveland, OH). Diisopropylethylamine,
chloride, O-acetyl- -carnitine chloride, and palmitoyl-DL-carnitine
chloride were purchased from Sigma (St. Louis, MO). -Carnitine
N-methyl-d , 98%) and acetyl- -carnitine (N,N-dimethyl-d , 98%)
were purchased from Cambridge Isotope Laboratories (Andover,
L
-carnitine
mol) of N-methyl-N-nitroso-p-toluenesulfonamide in a clear fit
distillation apparatus.32 The receiving flask was fitted with a
straight receiving tube and an addition funnel. A magnetic stirrer
was brought into position beneath the receiving flask. A solution
of 25 g (1.06 × 10 mol) of pentafluorobenzoyl chloride in 100
mL of diethyl ether was added dropwise. The icebath was
removed, and the reaction solution was stirred and allowed to
warm to room temperature for 1 h. This solution was evaporated
under vacuum to leave a slightly viscous yellow oil, which was
L
L
(
3
L
6
-
2
1
4
MA). Methyl[ C]-
L
-carnitine was synthesized from norcarnitine
(
[
4-N,N-(dimethylamino)-3-hydroxybutyrate) by methylation using
14
25
14
C]-methyl iodide. Palmitoyl-[ C]-
L
-carnitine was synthesized
14
26
from methyl[ C]-
carnitines used were synthesized using established methods.
L-carnitine and palmitoyl chloride. Other acyl-
27,28
dissolved in 200 mL of chloroform, treated with MgSO and
4
1
4
[
(
1- C]-Palmitic acid was purchased from New England Nuclear
Boston, MA). Radioactivity was measured using a Beckman
carbon, and filtered twice. The chloroform filtrate was concen-
trated to ∼75 mL, and 250 mL of hexane was added slowly. The
solution was cooled to room temperature and transferred to a
freezer for storage overnight at -60 °C. Crystals were collected
by vacuum filtration and allowed to dry in air for 5 min before
storage in a vacuum desiccator at room temperature. The yield
Coulter LS6500 scintillation counter (Fullerton, CA). 2,4′-Dibro-
moacetophenone, N-methyl-N-nitroso-p-toluenesulfonamide, and
trifluoromethanesulfonic acid were purchased from Aldrich (Mil-
waukee, WI). Pentafluorobenzoyl chloride was purchased from
-
2
of pale yellow needles was 18.5 g (7.8 × 10 mol, 74%), mp
(
(
(
18) Millington, D. S.; Norwood, D. L.; Kodo, N.; Roe C. R.; Inoue, F. Anal.
Biochem. 1989, 180, 331-339.
19) Rashed, M. S.; Ozand, P. T.; Bucknall, M. P.; Little, D. Pediatr. Res. 1995,
4
6-48 °C (Fisher Johns micro hot stage, atmospheric pressure,
uncorrected). Elemental analysis: Calcd, C, 40.70%; H, 0.43%; F,
3
8, 324-331.
20) Chace, D. H.; Kalas, T. A.; Naylor, E. W. Clin. Chem. 2003, 49, 1797-
817.
40.23%; N, 11.87%. Found, C, 40.31%; H, <0.5%; F, 41.69; N, 12.61%.
1
13
H NMR: singlet, 5.66 ppm. (CHN
2
). C NMR: singlet 60.23 ppm
1
(
CHN ), complex aromatic carbon resonances between 136.8 and
2
(21) Jonscher, K. R.; Yates, J. R. Anal. Biochem. 1997, 244, 1-15.
(22) Minkler, P. E.; Hoppel, C. L. J. Chromatogr. 1993, 613, 203-221.
(23) Minkler, P. E.; Hoppel, C. L. Anal. Biochem. 1993, 212, 510-518.
(24) Minkler, P. E.; Brass, E. P.; Hiatt, W. H.; Ingalls, S. T.; Hoppel, C. L. Anal.
Biochem. 1995, 231, 315-322.
1
9
145.7 ppm (JCF ) 255 Hz), singlet 176.70 ppm (CO). F NMR:
doublet -140.81 ppm (J ) 18.2 Hz, relative intensity 2, o-F), triplet
(
25) Ingalls, S. T.; Hoppel, C. L.; Turkaly, J. S. J. Labelled Compd. Radiopharm.
(29) Ingalls, S. T.; Minkler, P. E.; Hoppel, C. L.; Nordlander, J. E. J. Chromatogr.
1984, 299, 365-376.
1982, IX, 535-541.
(
26) Minkler, P. E.; Ingalls, S. T.; Kormos, L. S.; Weir, D. E.; Hoppel, C. L. J
(30) Erickson, J. L. E.; Dechary, J. M.; Kesling, M. R. J. Am. Chem. Soc. 1951,
73, 5301-5302.
Chromatogr. 1984, 336, 271-283.
(
(
27) Ziegler, H. J.; Bruckner, P.; Binon, F. J. Org. Chem. 1967, 32, 3989-3995.
28) Brendel, K.; Bressler, R. Biochim. Biophys. Acta 1967, 137, 98-106.
(31) Berenbom, M.; Fones, W. S. J. Am. Chem. Soc. 1949, 71, 1629.
(32) Technical Bulletin AL-180, Aldrich Chemical Co., Milwaukee, WI.
Analytical Chemistry, Vol. 77, No. 5, March 1, 2005 1449

-
149.76 ppm (J ) 17.96 Hz, relative intensity 1, p-F), complex
presented are the (-) minimum and (+) maximum values
determined.
multiplet -160.40 ppm (relative intensity 2, m-F).
Synthesis of Pentafluorophenacyl Trifluoromethane-
sulfonate. The procedure is modeled on the synthesis of 4′-
HPLC/UV. The HPLC/UV system consisted of an HP 1100
Series quaternary pump, autosampler, and diode array detector
purchased from Agilent (Wilmington, DE). The analytical column
29,33
bromophenacyl trifluoromethanesulfonate.
A clean and dry 500-
mL two-necked round-bottom flask was fitted with a stirring bar,
Claisen head, gas inlet tube, and drying tube. Sulfur dioxide was
admitted to the flask through the gas inlet tube, and a dry ice/
acetone bath was raised into position beneath the reaction vessel.
Sulfur dioxide (300 mL) was condensed into the flask. The flask
was a 4.6 × 100 mm i.d. Hypersil (MOS-1) C
8
(3-µm particle size)
column purchased from Alltech Associates, Inc. (Deerfield, IL).
Four chromatographic eluents were employed to formulate the
gradients described; eluent-1, acetonitrile/water (800+200, v+v);
eluent-2, acetonitrile/water (200+800, v+v); eluent-3, acetoni-
trile/water/acetic acid/triethylamine (200+800+5+5, v+v+v+v);
and eluent-4, acetonitrile/water/acetic acid/triethylamine
(900+100+5+5, v+v+v+v). The pump flow rate was 1.5
mL/min, and the diode array detector collected UV data at 240
and 260 nm. The quaternary gradient was formulated as follows:
initially, 100% eluent-1 was pumped. At 2 min after sample injection,
eluent-1 was replaced by 100% eluent-2. At 4 min after sample
injection, eluent-2 was replaced by 100% eluent-3 and then
proceeded to 100% eluent-4 using a 16-min linear gradient. The
system maintained 100% eluent-4 for 6 min. At 26 min, the system
returned to 100% eluent-1 for an additional 4 min before it began
its initiation sequence for the next injection (total run time of 30
min/injection).
-
2
was opened briefly, and 15.8 g (6.7 × 10 mol) of pentafluoro-
-diazoacetophenone was added in one portion to the stirred sulfur
dioxide. The flask was closed, and the contents were stirred for
2
-
2
5
min. The flask again was opened, and 10 g (6.7 × 10 mol) of
trifluoromethanesulfonic acid was added rapidly from a Pasteur
pipet. The drying tube was replaced, and the flask contents were
stirred for 5 min. The dry ice/acetone bath was removed, and
the reaction solution was allowed to warm. After 1 h, an ice/water
bath was used to further warm the reaction solution and hasten
solvent evaporation, leaving a pale green product mass. This was
dissolved in 200 mL of methylene chloride, treated with MgSO
4
and carbon, and filtered twice. The nearly colorless solution was
evaporated slowly on hot plate to ∼75 mL. Pentane (300 mL) was
added slowly with continuous stirring. The beaker was removed
from the hot plate and allowed to cool to room temperature before
storage overnight in a freezer at -60 °C. The product was
collected by vacuum filtration, washed with 50 mL of pentane,
and allowed to dry in air for 10 min before storage in a vacuum
desiccator at room temperature. The yield of colorless, transpar-
Reactivity Studies. Reaction progress curves using the
reagents 2,4′-dibromoacetophenone, 4′-bromophenacyl trifluo-
romethanesulfonate, and pentafluorophenacyl trifluoromethane-
sulfonate were generated as follows: solutions of carnitine,
acetylcarnitine, and palmitoylcarnitine (10 mM in methanol) were
prepared, and 10-µL aliquots were dispensed into vials. Derivati-
zation was performed by the addition of 20-µL of diisopropylethyl-
amine solution (0.1% in methanol; to neutralize any acid present)
and 50 µL of 2,4′-dibromoacetophenone, 4′-bromophenacyl tri-
fluoromethanesulfonate, or pentafluorophenacyl trifluoromethane-
sulfonate (100 mM in acetonitrile). Samples then were immediately
vortex-mixed, and at four different time points (0, 5, 15, and 30
min), triplicate samples were quenched by the addition of 50 µL
of 80:20 acetonitrile/water, evaporated to dryness, and reconsti-
tuted in 50 µL of 80:20 acetonitrile/water. An aliquot (5 µL) from
each sample was injected into the HPLC/UV system.
-
2
ent, flat needles was 20.8 g (5.8 × 10 mol, 86%), which melted
sharply at 63 °C. Elemental analysis: Calcd, C, 30.18%; H, 0.56%;
F, 42.44%; S, 8.95%. Found, C, 29.67%; H, 0.66%; F, 41.33%; S, 9.17%.
1
13
H NMR: singlet 5.31 ppm (CH
4.2 Hz, CH ), quartet centered at 118.68 ppm (JCF ) 319.01 Hz,
CF ), 3 complex aromatic doublets centered at 138.2, 144.8, and
45.7 ppm (JCF ) 255 Hz), singlet 183.21 ppm (CO). 19F NMR:
singlet -74.8 ppm (CF ), broad multiplets centered at -138.51
2
). C NMR: triplet 76.39 ppm (J
)
2
3
1
3
ppm (relative intensity 2, o-F), -143.98 (relative intensity 1, p-F)
and -158.66 ppm (relative intensity 2, m-F).
Protein Precipitation/Desalting and Cation-Exchange Solid-
Phase Extraction Recovery Study. Replicate human plasma
aliquots (50 µL) from a normal volunteer were mixed with 10-µL
HPLC/MS. The HPLC system consisted of an HP 1050 Series
quaternary pump and autosampler (Agilent). The analytical
column and chromatographic eluents were the same as those
employed in the HPLC/UV system. The MS was an LCQ ion trap
(ThermoFinnigan), using source parameters of sheath gas (90
units), auxiliary gas (30 units), spray voltage (5.2 kV), and capillary
temperature (245 °C). The entire HPLC effluent stream, when
not diverted to waste, was directed into the MS interface. Initially,
the pump flow rate was 1.75 mL/min using 100% eluent-1 and the
divert valve was set to waste. At 2 min after sample injection,
eluent-1 was replaced with 100% eluent-2. At 4 min after sample
injection, eluent-2 was replaced with 100% eluent-3 and then pro-
ceeded to 80% eluent-3/20% eluent-4 using a 3-min linear gradient.
At 7 min after injection, the eluent flow rate was switched to 0.50
mL/min, the divert valve was triggered from waste to the detector,
and a linear gradient of 80% eluent-3/20% eluent-4 to 50% eluent-
3/50% eluent-4 over 25 min was formulated. At 32 min after injec-
tion, a linear gradient from 50% eluent-3/50% eluent-4 to 100%
eluent-4 was formulated over 20 min. The system then was
1
4
aliquots of [ C]-carnitine, (0.491 mM, specific radioactivity 2374
14
dpm/nmol), palmitoyl-[ C]-carnitine (1.75 mM, specific radioactiv-
1
4
ity 886 dpm/nmol), or [ C]-palmitic acid (0.104 mM, specific
radioactivity 11988 dpm/nmol), and 1 mL of acetonitrile/methanol
(3+1, v+v). These samples were centrifuged (5 min at 13200g),
and the entire supernatants were either removed and their
radioactivity content determined or added to solid-phase extraction
columns. The resulting column flow-through fractions were
collected. Then the columns were washed with 0.5 mL of
methanol, followed by elution with 1 mL of water/methanol/acetic
acid (10+9+1, v+v+v). These fractions were also collected. The
radioactivity content in all these samples was then determined,
and the results were compared with the radioactivity content in
14
14
14
1
0-µL aliquots of [ C]-carnitine, palmitoyl-[ C]-carnitine, or [ C]-
palmitic acid, which served as external standards. The values
reported are calculated from triplicate analyses, and the ranges
1450 Analytical Chemistry, Vol. 77, No. 5, March 1, 2005

maintained at 100% eluent-4 for 11 min. At 63 min after injection,
the divert valve was returned to waste, the eluent flow rate was
increased to 1.75 mL/min, and the system returned to 100%
eluent-1 for an additional 7 min before the HPLC system began
its initiation sequence for the next injection (total run time of 70
min/injection).
prepared in water and methanol from standard solutions of
carnitine (0, 5, 10, 20, 30, 50, 70, and 100 µmol/L) and acetylcar-
nitine (0, 5, 10, 20, 35, 50, 75, and 100 µmol/L). Samples and
standards were isolated and derivatized as described above.
For carnitine, the mass spectrometer was programmed to
generate data from two sequential scan events: selected reaction
monitoring of the transition of m/z 370 to 311 (for carnitine
An abbreviated version of this eluent system was used for the
analysis of carnitine in rat liver. The initial pump flow rate was
3
detection) and 373 to 311 (for carnitine-d detection) with 30%
1
.75 mL/min using 100% eluent-1. At 1.2 min after sample injection,
relative collision energy. For the acetylcarnitine analysis, two scan
events were also used: (1) full-scan MS/MS (m/z 110-450) on
m/z 412 (for acetylcarnitine) and (2) full-scan MS/MS (m/z
eluent-1 was replaced with 100% eluent-2. At 2 min after sample
injection, eluent-2 was replaced with 100% eluent-3 and then
proceeded to 85% eluent-3/15% eluent-4 using a 2.75-min linear
gradient. At 4.75 min after injection, the eluent flow rate was
switched to 0.50 mL/min, the divert valve was triggered from
waste to the detector, and a linear gradient of 85% eluent-3/15%
eluent-4 to 50% eluent-3/50% eluent-4 over 4.75 min was formu-
lated. At 9.5 min after injection, the divert valve was triggered to
waste, the flow rate increased to 1.75 mL/min, and the system
was maintained 100% eluent-4 for 2 min. At 11.5 min after injection,
the system returned to 100% eluent-1 for an additional 2 min before
the HPLC system began its initiation sequence for the next
injection (total run time of 13.5 min/injection).
Analysis of Spiked Phosphate-Buffered Serum Albumin.
Solutions containing 1 nmol each of carnitine and 15 different
acylcarnitines were added to microcentrifuge tubes along with 50
µL of phosphate-buffered serum albumin. In addition, a single
microcentrifuge tube containing combined aliquots from all these
solutions, along with 50 µL of phosphate-buffered serum albumin,
was prepared. The microcentrifuge tubes were vortex-mixed for
1
6
15-450) on m/z 418 (for acetylcarnitine-d ) both using 30%
relative collision energy.
RESULTS AND DISCUSSION
Isolation of Carnitine and Acylcarnitines by Protein
Precipitation/Desalting and Cation-Exchange Solid-Phase
Extraction. The sample isolation procedure for carnitine and
acylcarnitines has three requirements: (1) Because of our varied
research interests, it is imperative that the procedure be able to
accommodate diverse specimen types without adjustments. (2)
Within a biological system, there are many acylcarnitine species
exhibiting a wide range of structures and polarities. Despite this
diversity, we would like to isolate all of these in a single fraction.
(3) Since we plan to follow the isolation step with a derivatization,
the sample isolation must be compatible with the requirements
of the subsequent derivatization procedure.
The first step in our method for isolation of carnitine and
acylcarnitines is protein precipitation/desalting using organic
solvents, followed by centrifugation and collection of the super-
natant. We found that a solution of acetonitrile/methanol (3+1,
v+v), added to the specimen in a volume ratio of 10:1, to be
advantageous. Methanol alone left a viscous, sticky precipitate,
whereas acetonitrile alone created a flocculent suspension in the
supernatant. By combining these two solvents, the resulting
precipitate adhered strongly to the wall of the microcentrifuge
tube. Additionally, because of the more limited solubility of many
inorganic salts in acetonitrile, specimens are substantially desalted
2
s, and 1 mL of acetonitrile/methanol (3+1, v+v) was added to
each tube. The tubes were then capped, vortex-mixed for 2 s, and
centrifuged for 5 min at 13200g. The resulting supernatants were
decanted into solid-phase extraction columns, and the flow-
throughs were discarded. The columns were rinsed with 0.5 mL
of methanol and then placed above 12 × 75 mm glass tubes into
which were eluted carnitine and acylcarnitines using a single 1-mL
aliquot of water/methanol/acetic acid (10+9+1, v+v+v). The
contents of each tube was evaporated to dryness with nitrogen
and derivatized with pentafluorophenacyl trifluoromethanesul-
fonate as described above. The mass spectrometer was pro-
grammed to generate data from two sequential scan events: (1)
full-scan MS, 250-1000 m/z, and (2) data-dependent full-scan
MS/MS on the most intense MS ion with a relative collision
energy setting of 30%.
Analysis of Rat Liver. All procedures were approved by the
Veterans Administration Institutional Care and Use Committee
and in accordance with National Institutes of Health Guildlines
for Care and Use of Animals in Research. A male Sprague-Dawley
rat (420 g) was obtained from Zivic-Miller (Allison Park, PA) and
had free access to food and water. The animal was killed, and the
liver was rapidly excised and frozen in liquid nitrogen. A small
amount (40 mg) was homogenized in 1 mL of water, and 10
(inorganic anions react with our reagents; therefore minimizing
the content of inorganic salts is desirable).
The one common structural feature of all acylcarnitines is car-
nitine, and its quaternary ammonio functional group is a candidate
for ion-exchange isolation schemes. Although silica gel is most
often employed in normal-phase chromatography, it also has been
3
4,35
shown to be effective for cation-exchange chromatography.
have demonstrated, using 300-mg columns of silica gel, that the
We
22
recovery of carnitine from urine and the recovery of acetyl- and
23
octanoylcarnitine from plasma are ∼90%. This strategy was
further developed by Poorthuis et al.,12 who employed commer-
cially available 500-mg silica gel solid-phase extraction columns.
Recovery studies using human plasma and radiolabeled
compounds were performed for the protein precipitation/desalting
step (using acetonitrile/methanol) and cation-exchange isolation
5
0-µL aliquots were placed into microcentrifuge tubes and
(using solid-phase extraction columns containing 50 mg of silica
analyzed for carnitine and acetylcarnitine (5 × each). Internal
standards were added; carnitine samples received 50-µL aliquots
(33) Vedejs, E.; Engler, D. A.; Mullins, M. J. J. Org. Chem. 1977, 42, 3109-
113.
34) Golkiewicz,W.; Kuczynski, J.; Markowski, W.; Jusiak, L. J. Chromatogr.
994, 686, 85-91.
3
of carnitine-d
0-µL aliquots containing acetylcarnitine-d
quantification, eight-point standard curves were concurrently
3
(25 µmol/L), and acetylcarnitine samples received
(
5
6
(25 µmol/L). To permit
1
(35) Ohta, K.; Tanaka, K.; Haddad, P. R. J. Chromatogr. 1996, 739, 359-365.
Analytical Chemistry, Vol. 77, No. 5, March 1, 2005 1451

progress curves using 4′-bromophenacyl trifluoromethanesulfonate,
with virtual reaction completion at time point zero. This is in
contrast to Figure 2A, where 2,4′-dibromoacetophenone required
at least 30 min before approaching this amount of ester formation.
We demonstrated that the reaction of carnitine with 4′-bro-
mophenacyl trifluoromethanesulfonate in a urine specimen, using
14
[
C]-carnitine, resulted in complete conversion of carnitine to
22
carnitine 4′-bromophenacyl ester. Figure 2B is consistent with
that result. Because carnitine 4′-bromophenacyl ester can be
detected simultaneously with acetyl- and palmitoylcarnitine 4′-
bromophenacyl esters, we examined the acylcarnitine chromato-
grams for the presence of carnitine 4′-bromophenacyl ester and
found none. We concluded that acylcarnitines remain intact and
are not converted to carnitine under the derivatization reaction
conditions. This is in contrast to the popular procedure for butyl
ester formation, where 30% of the acetylcarnitine is hydrolyzed
during 15 min of reaction.36
Reaction progress curves using pentafluorophenacyl trifluo-
romethanesulfonate are shown in Figure 2C. As with 4′-bro-
mophenacyl trifluoromethanesulfonate, pentafluorophenacyl tri-
fluoromethanesulfonate reacted rapidly at room temperature. Both
of these reagents led to maximum ester formation at time point
zero, with product chromatographic peaks remaining unchanged
over the 30-min reaction duration. As with the 4′-bromophenacyl
trifluoromethanesulfonate studies described, we examined the
acylcarnitine chromatograms for the presence of carnitine pen-
tafluorophenacyl ester and found none. However, the UV absor-
bance of pentafluorophenacyl esters at 260 nm is substantially
lower than for 4′-bromophenacyl esters. Therefore, the peak areas
for pentafluorophenacyl esters were measured at 240 nm rather
than at 260 nm. This aside, the performance of pentafluorophena-
cyl trifluoromethanesulfonate is comparable to 4′-bromophenacyl
trifluoromethanesulfonate.
Figure 1. Isolation of carnitine and acylcarnitines by cation-
exchange solid-phase extraction. This bar graph shows the recovery
14
of radioactivity from human plasma spiked with [ C]-carnitine,
palmitoyl-[14C]-carnitine, or [14C]-palmitic acid using the cation-
exchange solid-phase extraction procedure. The results are the
averages from triplicate analyses, and the ( error bars represent the
maximum (+) and minimum (-) values determined. Experimental
conditions are described in the text.
gel). The results from the protein precipitation/desalting step are
that recoveries of carnitine, palmitoylcarnitine, and palmitic acid
are 88-89, 87-90, and 92-95%, respectively. The elution pattern
using the solid-phase extraction column is shown in Figure 1.
Here, carnitine and palmitoylcarnitine are almost completely
retained during the flow-through and wash steps, whereas palmitic
acid is not retained and is recovered in the flow-through and
washing fractions. The elution solvent, containing water and acetic
acid, disrupted the ion-exchange interaction between the silica
gel and both carnitine and palmitoylcarnitine. This resulted in the
elution of carnitine and palmitoylcarnitine in the same 1-mL elution
fraction, with recoveries from the combined protein precipitation/
desalting and cation-exchange steps of 77-80% for carnitine and
Sequential Ion-Exchange/Reversed-Phase Chromatogra-
phy. Reports of the use of bare silica for ion-exchange HPLC
37
38
separations of organic amines and basic drugs have shown the
feasibility of this chromatographic mode. Reversed-phase chro-
matography, accomplished not by bonded silica phases but by
dynamically modifying bare silica surfaces using quaternary
ammonium salts (thus demonstrating the ion-exchange interaction
of the quaternary ammonium functional group with silica gel),
83-85% for palmitoylcarnitine. That essentially 100% of the palmitic
acid radioactivity is found in the flow-through and wash fractions
shows complete separation of fatty acids from carnitine and
acylcarnitine using this isolation scheme. We have successfully
applied this procedure, documented by recovery of added internal
standards, to urine, tissues, dialysate fluid, bile, pet food, diet
preparations, isolated sperm, and specimens generated from cell
culture without any modification.
39
also has been reported. Furthermore, this type of ion-exchange
40
interaction is present on bonded-phase silica and is observed in
the strong adsorption of amine solutes. Considerable efforts have
Derivatization of Carnitine and Acylcarnitines. Derivatiza-
tion of carnitine and acylcarnitines for sensitivity enhancement is
41
been devoted to control or eliminate these effects with, for
example, the inclusion of amine modifiers as HPLC mobile-phase
7
-20
common.
We studied the carboxyl-O-alkylation of carnitine
42,43
components.
However, we saw the presence of cation-exchange
and acylcarnitines with the derivatization reagent 2,4′-dibromo-
acetophenone. Unfortunately, this reaction does not proceed at
the low concentrations of carnitine and acylcarnitines present in
interactions with bonded-phase silica as providing a means for
sequential ion-exchange/reversed-phase chromatography. We
chose to develop this technique using Hypersil (MOS-1) C as
8
2
6
biological specimens. Therefore, we developed a far more
(36) Johnson, D. W. J. Inherited Metab. Dis. 1999, 22, 201-202.
reactive derivatization reagent for 4′-bromophenacyl ester deriva-
(37) Bidlingmeyer, B. A.; Del Rios, J. K.; Korpi, J. Anal. Chem. 1982, 54, 442-
447.
29
tive formation: 4′-bromophenacyl trifluoromethanesulfonate. This
(
(
38) Law, B. Trends Anal. Chem., 1990, 9, 31-36.
39) Honore Hanson, S.; Helboe, P.; Thomsen, M. J. J. Pharm. Biomed. Anal.
reagent allowed mild, rapid, and quantitative derivatization of
2
2,23,26
carnitine and acylcarnitines.
1
984, 2, 165-172.
To illustrate the reactivity differences between 2,4′-dibromo-
acetophenone and 4′-bromophenacyl trifluoromethanesulfonate,
reaction progress curves were generated with carnitine, acetyl-
carnitine, and palmitoylcarnitine. Figure 2B shows the reaction
(40) Weber, S. G.; Tramposch, W. G. Anal. Chem. 1983, 55, 1771-1775.
(41) Stadalius, M. A.; Berus, J. S.; Snyder L. R. LC-GC 1988, 6, 494-500.
(42) Gill, R.; Alexander, S. P.; Moffat, A. C. J. Chromatogr. 1982, 247, 39-45.
(43) Kiel, J. S.; Morgan, S. L.; Abramson, R. K. J. Chromatogr. 1985, 320, 313-
323.
1452 Analytical Chemistry, Vol. 77, No. 5, March 1, 2005

Figure 2. Derivatization of carnitine and acylcarnitines. Reaction curves for carnitine, acetylcarnitine, and palmitoylcarnitine using three different
reagents. (A) 2,4′-dibromoacetophenone, (B) 4′-bromophenacyl trifluoromethanesulfonate, and (C) pentafluorophenacyl trifluoromethanesulfonate.
The results are the averages from triplicate analyses, and the ( error bars represent the maximum (+) and minimum (-) values determined.
Reaction and chromatographic conditions are described in the text.
the chromatographic column medium, since this material is non
end capped and therefore contains many free silanols available
for ion-exchange chromatography.
and 6-(trimethylammonio)hexanoic 4′-bromophenacyl esters. In
contrast, Figure 3B displays overlaid chromatograms generated
using the same column, but with the first two chromatographic
eluents replaced with 100% eluent-3 (reversed-phase only separa-
tion). Because the ion-exchange capability of the column is not
utilized, the elution order of reagent and derivatives is reversed,
with carnitine, 4-(trimethylammonio)butanoic, and 6-(trimethyl-
ammonio)hexanoic 4′-bromophenacyl esters all eluting before 2,4′-
dibromoacetophenone.
We synthesized carnitine, 4-(trimethylammonio)butanoic, and
6
-(trimethylammonio)hexanoic 4′-bromophenacyl esters and in-
vestigated the retention of these compounds and 2,4′-dibromo-
acetophenone using sequential ion-exchange/reversed-phase chro-
matography. These results were compared to the separations
using reversed-phase conditions alone. Compound solutions were
formulated (1 mM in 80% acetonitrile), and 5 µL injections of each
compound were made into the HPLC/UV system with detection
at 260 nm (Figure 3A). The first eluent, (80% acetonitrile) caused
rapid elution of 2,4′-dibromoacetophenone but not of carnitine,
22,23
We have applied this procedure to 4′-bromophenacyl,
2-(2,3-
24
naphthalimino)ethyl, and pentafluorophenacyl esters of carnitine
and acylcarnitines and have obtained similar results. We also have
observed that other columns are capable of sequential ion-
exchange/reversed-phase chromatographic behavior, but their use
for this purpose should be examined on a case-by-case basis. The
amine modifier combination of triethylamine and acetic acid was
chosen because these materials are inexpensive and commercially
available in high-purity HPLC grades. This combination also is
appropriate because of its volatility, a necessary property for
detection with electrospray ionization MS.
4-(trimethylammonio)butanoic, or 6-(trimethylammonio)hexanoic
4
′-bromophenacyl esters. Despite the high organic content of the
eluent, these compounds were retained owing to the ion-exchange
interactions of their positively charged trimethylammonio func-
tional groups with the free silanols on the chromatographic
column. The second eluent (20% acetonitrile) was employed to
lower the acetonitrile content before reversed-phase gradient
elution using eluent-3 and eluent-4 containing acidified triethyl-
amine as the amine modifier. This amine modifier component
disrupted the ion-exchange interactions and allowed reversed-
phase order elution of carnitine, 4-(trimethylammonio)butanoic,
HPLC/MS. Detection of 4′-bromophenacyl and pentafluo-
rophenacyl esters of carnitine and acylcarnitines is feasible by UV,
but the selectivity possible with mass spectrometry is clearly
superior. Zero time point reaction samples were injected into the
Analytical Chemistry, Vol. 77, No. 5, March 1, 2005 1453

and detection by MS. Figure 5A shows 16 overlaid chromatograms
generated from 16 individual injections. Individual injections of
single-component samples were performed to unambiguously
establish the identity of the isomeric acylcarnitines based on their
retention time. These chromatograms were reconstructed from
the full-scan MS spectra by selection of the molecular ion mass
of the compound injected. The resulting chromatographic peaks
signals for the carnitine and acylcarnitine pentafluorophenacyl
esters (100 pmol of each injected) are very large, and reversed-
phase separation of all compounds is accomplished within the
7
5
0-min chromatographic run time. The chromatogram in Figure
B was generated from two overlaid reconstructed ion chromato-
grams from a single injection of a mixture of these 16 compounds.
Full-scan MS spectra were again used, and the two ions selected
for this figure were the secondary ion for carnitine (m/z 311) for
the peak labeled 1 and the ion common to all acylcarnitine
pentafluorophenacyl esters (m/z 293) for peaks labeled 2-16. In
addition, MS/MS full-scan spectra were automatically collected
for carnitine and each acylcarnitine pentafluorophenacyl ester
using data-dependent sampling (full-scan MS/MS on the most
intense ion from each full-scan MS spectrum collected), and
these spectra were all consistent with the fragmentation pat-
terns depicted in Figure 4. Chromatographic resolution of several
groups of isomers is shown (isobutyryl and butyryl; 2-methylbu-
tyryl-, isovaleryl, and valeryl; valproyl and octanoyl). Below the
chromatogram in Figure 5B are the MS and MS/MS spectra for
Figure 3. Sequential ion-exchange/reversed-phase chromatogra-
phy. (A) Chromatographic separation using sequential ion-exchange/
reversed-phase separation. (B) Chromatographic separation using
reversed-phase separation alone. Four chromatograms from four
different injections were overlaid to produce (A) and (B). Chromato-
graphic conditions are described in the text. Peak identities: (1) 2,4′-
dibromoacetophenone, (2) carnitine 4′-bromophenacyl ester, (3)
4-(trimethylammonio)butanoic 4′-bromophenacyl ester, and (4) 6-(tri-
methylammonio)hexanoic 4′-bromophenacyl ester.
8
the two isomeric C acylcarnitines contained in this chromato-
gram: valproylcarnitine and octanoylcarnitine. For both com-
pounds, the full-scan MS spectra contained the molecular ion (m/z
HPLC/MS system, and the MS and MS/MS spectra of 4′-
bromophenacyl and pentaflourophenacyl esters of carnitine, ace-
tylcarnitine, and palmitoylcarnitine were examined. 4′-Bromophen-
acyl esters have molecular ions with two signals of almost equal
4
96) as the base peak. Also present are signals resulting from
the loss of trimethylamine (m/z 437) and from the acylcarnitine
pentafluorophenacyl ester common ion (m/z 293). Below the full-
scan MS spectra are the corresponding full-scan MS/MS spectra
generated from the selected ion m/z 496. The same three ions
are present in the full-scan MS/MS spectra, but with increased
proportions of m/z 437 and 293, as expected. From the data shown
in Figure 5, we conclude that the electrospray MS and collision-
induced dissociation MS/MS spectra of valproylcarnitine and
octanoylcarnitine contain identical features and are essentially
indistinguishable. Therefore, the means to identify the specific
isomeric form of these acylcarnitine rests with the chromato-
graphic system rather than the mass spectrometer.
7
9
intensity, consistent with the natural abundance of Br at 50.7%
8
1
and Br at 49.3%. This results in a division of the MS signal
intensities and lower relative abundances versus pentafluorophen-
acyl esters. Based on this observation, pentafluorophenacyl esters
of carnitine and acylcarnitines were chosen for mass spectral de-
tection of carnitine and acylcarnitines. The mass spectrum of the
carnitine pentafluorophenacyl ester yields a strong signal for the
molecular ion at m/z 370, and collision-induced dissociation yields
a product ion at m/z 311, consistent with loss of trimethylamine.
Acetylcarnitine and palmitoylcarnitine pentafluorophenacyl esters
fragment with collision-induced dissociation to yield two product
ions: (1) loss of m/z 59 and (2) generation of an ion at m/z 293.
This is a general pattern of fragmentation observed in all 15
acylcarnitines examined below and allowed qualitative detection
of acylcarnitine pentafluorophenacyl ester species based on the
generation of a m/z 293 ion (see Figure 5B). This fragmentation
The chromatographic separation of carnitine and these acyl-
carnitines required an analysis time of 70 min, due to the sampling
frequency limitations of the LCQ ion trap. This is a disadvantage
of this HPLC/MS method in relation to rapid tandem MS analysis.
However, the advantages of this chromatography strategy are as
follows: (1) The ability to distinguish isomers. No method that
does not include a chromatographic step has this capability. Using
high-energy FAB, isomeric acylcarnitines were purported to yield
10
pattern is entirely consistent with other reported carnitine and
1
6
acylcarnitine MS transitions. These findings are illustrated with
example spectra and the proposed structures shown in Figure 4.
Spiked Phosphate-Buffered Serum Albumin. To demon-
strate this analytical strategy, phosphate-buffered serum albumin
was spiked with carnitine and 15 acylcarnitines. This was analyzed
using the described protein precipitation/desalting and cation-
exchange solid-phase extraction isolation, derivatization with
pentafluorophenacyl trifluoromethanesulfonate, chromatography
using the sequential ion-exchange/reversed-phase HPLC system,
44
distinguishable MS fragments. However, this finding has not
been reported with electrospray MS. (2) The inherent selectivity
afforded by chromatography. Relying exclusively on MS data for
compound identification and quantification can be problematic
45
because of MS interferences. In the specific case of carnitine
(44) Gaskell, S. J.; Guenat, C; Millington, D. S.; Maltby, D. A.; Roe, C. R. Anal.
Chem. 1986, 58, 2801-2805.
(45) Jemal, M.; Ouyang, Z. Rapid Commun. Mass Spectrom. 2000, 14, 1757-
1765.
1454 Analytical Chemistry, Vol. 77, No. 5, March 1, 2005

Figure 4. MS/MS of carnitine and acylcarnitine pentafluorophenacyl esters. (A) Full-scan MS/MS spectra of carnitine, with transition structures
shown below. (B) Full-scan MS/MS spectra of acetylcarnitine and palmitoylcarnitine with general acylcarnitine transition structures shown below.
Spectra were generated using the HPLC/MS system. Conditions are described in the text.
detection, chromatographic separation prior to MS/MS detection
removed ionization suppressing matrix components, increasing
MS/MS signal response for serum specimens 16-35-fold.46
Quantification of Carnitine and Acetylcarnitine in Rat
Liver. Application of this strategy to a biological specimen is
illustrated in Figure 6. Rat liver was analyzed for carnitine and
acetylcarnitine, and these compounds were quantified using heavy
labeled internal standards and multiple-point standard curves. The
eight-point carnitine standard curve was constructed from the total
ion current selected reaction monitoring chromatograms of
3
carnitine standard peak areas versus carnitine-d internal standard
peak areas and was linear (r2 ) 0.9988). The acetylcarnitine
standard curve was constructed from MS/MS chromatograms
generated from the sum of the responses of the m/z 293, 353,
6
and 412 ions versus the acetylcarnitine-d internal standard peak
areas, generated using the sum of the m/z 293, 353, and 418 ions;
2
this curve was also linear (r ) 0.9989). The rat liver sample
quantification results for tissue content of the five replicate samples
(46) Ho, C. S.; Cheng, B. S. S.; Lam, C. W. K. Clin. Chem. 2003, 49, 1189-
1191.
Analytical Chemistry, Vol. 77, No. 5, March 1, 2005 1455

Figure 5. HPLC/MS of carnitine and acylcarnitine pentafluorophenacyl esters. (A) Solutions containing 1 nmol each of carnitine or an acylcarnitine
were added to 50-µL aliquots of phosphate-buffered serum albumin. Carnitine and acylcarnitines were isolated using the described protein
precipitation/desalting and cation-exchange solid-phase extraction isolation, derivatized with pentafluorophenacyl trifluoromethanesulfonate,
chromatographed using the sequential ion-exchange/reversed-phase chromatography HPLC system, and detected by MS (16 injections, 100
pmol each compound, 1 compound/injection). From the full-scan MS spectra (250-1000 m/z), 16 ion chromatograms were extracted and overlaid
to produce this figure. Peak identities and the extracted molecular mass for each chromatogram: (1) carnitine m/z 370, (2) acetylcarnitine m/z
4
12, (3) propionylcarnitine m/z 426, (4) isobutyrylcarnitine m/z 440, (5) butyrylcarnitine m/z 440, (6) 2-methyl-butyrylcarnitine m/z 454, (7)
isovalerylcarnitine m/z 454, (8) valerylcarnitine m/z 454, (9) hexanoylcarnitine m/z 468, (10) valproylcarnitine m/z 496, (11) octanoylcarnitine
m/z 496, (13) decanoylcarnitine m/z 524, (14) lauroylcarnitine m/z 552, (15) myristoylcarnitine - m/z 580, (16) palmitoylcarnitine m/z 608, and
(18) stearoylcarnitine m/z 636. (B) The same solutions in (A) were combined, added to 50 µL of phosphate-buffered serum albumin, and prepared
as described above (1 injection). From the full-scan MS spectra (250-1000 m/z), two ion chromatograms were generated by extracting the
secondary ion for carnitine (m/z 311) and the ion common to acylcarnitine pentafluorophenacyl esters (m/z 293). These two chromatograms
were overlaid to produce this figure. Peak identities are the same as in (A). Below the chromatogram are the full-scan MS and full-scan MS/MS
spectra for the C8 isomeric acylcarnitines valproylcarnitine (left) and octanoylcarnitine (right). Chromatographic and MS conditions are described
in the text.
were for carnitine (mean ( SD): 347.6 ( 17.4 nmol/g wet weight
ww) and for acetylcarnitine 75.4 ( 1.7 nmol/g ww. Figure 6A
shows superimposed carnitine (1) and carnitine-d (1a) chromato-
grams from one of the rat liver replicates, while Figure 6B shows
superimposed acetylcarnitine (2) and acetylcarnitine-d (2a) chro-
matograms. The variability (RSD) of these determinations was
% for carnitine and 2% for acetylcarnitine, demonstrating the
strategy for carnitine and acylcarnitine analysis consisted of sample
isolation, derivatization, chromatography, and detection. Selectivity
first came from protein precipitation/desalting and cation-
exchange solid-phase extraction, which isolated both carnitine and
acylcarnitines with high recoveries from biological matrixes. The
sample simplification was shown to exclude palmitic acid while
recovering both carnitine and palmitoylcarnitine in a single
fraction. This isolation was demonstrated in radioactively labeled
spiked human plasma, spiked phosphate-buffered serum albumin,
and rat liver. The synthesis of pentafluorophenacyl trifluo-
romethanesulfonate, a new derivatization reagent, was described.
Derivatization under mild conditions using this reagent not only
(
3
6
5
reproducibility and precision of this approach.
SUMMARY
Analytical techniques usually rely on a combination of selectiv-
ity and detection for compound analysis. Our four-component
1456 Analytical Chemistry, Vol. 77, No. 5, March 1, 2005

Figure 6. Analysis of carnitine and acetylcarnitine in rat liver. The 50-µL aliquots of a 40 mg/mL homogenate of rat liver were analyzed.
Quantification was performed using multiple-point standard curves. (A) Carnitine + carnitine-d3 internal standard selected reaction monitoring
MS/MS chromatograms. Peak identities: (1) carnitine (quantified at 13.97 µmol/L) and (1a) carnitine-d3 (added at a concentration of 25 µmol/L).
(B) Acetylcarnitine + acetylcarnitine-d6 internal standard MS/MS chromatograms generated from the sum of the m/z 412 MS/MS fragment
responses at masses m/z 293, 353, and 412 (acetylcarnitine) or the sum of the m/z 418 MS/MS fragment responses at masses m/z 293, 353,
and 418 (acetylcarnitine-d6). Peak identities: (2) acetylcarnitine (quantified at 3.02 µmol/L) and (2a) acetylcarnitine-d6 (added at a concentration
of 25 µmol/L).
increased detector response without unintended hydrolysis of
acylcarnitines but also imparted desirable chromatographic prop-
erties to carnitine and acylcarnitines. These qualities were
exploited by sequential ion-exchange/reversed-phase HPLC. Here,
carnitine and acylcarnitine pentafluorophenacyl esters were iso-
lated from the derivatization reaction mixture by ion-exchange
chromatography and then separated by reversed-phase HPLC.
This approach provided highly pure carnitine and acylcarnitine
pentafluorophenacyl esters for detection, with unambiguous
identification of isomeric acylcarnitines due to their chromato-
graphic resolution. While sacrificing the rapidity of tandem MS
analysis, this report shows that the combination of HPLC and MS
can provide isomeric compound identification not obtainable by
tandem MS analysis alone, with a new derivatization reagent
usable under conditions that do not hydrolyze acylcarnitines, and
an isolation technology that removes previous limitations on
sample matrix selection.
ACKNOWLEDGMENT
This work was supported in part by the Office of Research
and Development, Medical Research Service, Department of
Veterans Affairs, and Grants DK36069, 1POAG15885, and NIH
Road Map Grant 1R33DK70291 from the National Institute of
Health.
Received for review August 16, 2004. Accepted December
6, 2004.
AC0487810
Analytical Chemistry, Vol. 77, No. 5, March 1, 2005 1457