Enhancement of the LC/MS analysis of fatty acids through derivatization and stable isotope coding

Article abstract of DOI:10.1021/ac070311t

This paper focuses on the development of an enhanced LC/ESI-MS method for the identification and quantification of fatty acids through derivatization. Fatty acids were derivatized with 2-bromo-1-methylpyridinium iodide and 3-carbinol-1-methylpyridinium iodide, forming 3-acyloxymethyl-1-methylpyridinium iodide (AMMP). This process attaches a quaternary amine to analytes and enabled ESI-MS in the positive mode of ionization with common LC mobile phases. Moreover, detection sensitivity was generally 2500-fold higher than in the negative mode of ionization used with underivatized fatty acids. The limits of detection were roughly 1.0-4.0 nM (or 10 pg/injection) for standard fatty acids from C10 to C24 and spanned ～2 orders of magnitude in linearity. AMMP derivatives had unique tandem mass spectra characterized by common ions at m/z 107.0, 124.0, and 178.0. Individual fatty acids also had unique fingerprint regions that allowed identification of their carbon skeleton number, number of double bonds, and double bond position. The derivatization method also allowed coding of analytes as a means of recognizing derivatives and enhancing quantification. ²H-Coding was achieved through derivatization with deuterated 3-carbinol-1-methyl-d₃-pyridinium iodide. The ²H-coded derivatization reagent, 3-acyloxymethyl-1-methyl-d ₃-pyridinium iodide, was used in two ways. One was to differentially label equal fractions of a sample such that after being recombined and analyzed by ESI-MS all fatty acids appeared as doublet clusters of ions separated by roughly 3 amu. This greatly facilitated identification of fatty acids in complex mixtures. Another use of stable isotope coding was in comparative quantification. Control and experimental samples were differentially labeled with nondeuterated and deuterated isotopomers of CPM, respectively. After mixing the two samples, they were analyzed by ESI-MS. The abundance of a fatty acid in an experimental sample relative to the control was established by the isotope ratio of the isotopomeric fatty acids. Absolute quantification was achieved by adding differentially labeled fatty acid standards to experimental samples containing unknown quantities of fatty acids. Utility of the method was examined in the analysis of human serum samples.

Full text of DOI:10.1021/ac070311t

Anal. Chem. 2007, 79, 5150-5157
Enhancement of the LC/MS Analysis of Fatty Acids
through Derivatization and Stable Isotope Coding
Wen-Chu Yang, Jiri Adamec, and Fred E. Regnier*
Department of Chemistry, Purdue University, West Lafayette, Indiana 47907
This paper focuses on the development of an enhanced
LC/ESI-MS method for the identification and quantifica-
tion of fatty acids through derivatization. Fatty acids were
derivatized with 2-bromo-1-methylpyridinium iodide and
The analysis of fatty acids and metabolites containing a
carboxyl group is of major interest in metabolomics because of
1
their prominent role in metabolic pathways and importance in
2
food. Moreover, ∼8% of organic compounds of pharmaceutical
3
3
-carbinol-1-methylpyridinium iodide, forming 3-acyl-
and biomedical interest possess a carboxyl group. Although it
oxymethyl-1-methylpyridinium iodide (AMMP). This pro-
cess attaches a quaternary amine to analytes and enabled
ESI-MS in the positive mode of ionization with common
LC mobile phases. Moreover, detection sensitivity was
generally 2500-fold higher than in the negative mode of
ionization used with underivatized fatty acids. The limits
of detection were roughly 1.0-4.0 nM (or 10 pg/injec-
tion) for standard fatty acids from C10 to C24 and
spanned ∼2 orders of magnitude in linearity. AMMP
derivatives had unique tandem mass spectra character-
ized by common ions at m/z 107.0, 124.0, and 178.0.
Individual fatty acids also had unique fingerprint regions
that allowed identification of their carbon skeleton num-
ber, number of double bonds, and double bond position.
The derivatization method also allowed coding of analytes
as a means of recognizing derivatives and enhancing
would seem that their analysis by gas chromatography (GC), high-
performance liquid chromatography (HPLC), or capillary electro-
phoresis would be straightforward,4 extracts from biological
fluids are of such great complexity and difference in concentration
there is a problem with resolving and detecting all the compo-
nents. One of the ways to deal with this problem is through mass
spectrometry (MS). MS has the great advantage of increasing
detection selectivity and sensitivity of unresolved components
-6
7
while providing structural information at the same time. Tradi-
tionally, GC/MS methods employed derivatization by sialylation
8
-12
or methylation with detection by electron impact ionization,
but HPLC/MS has become a powerful alternative with the advent
of atmospheric pressure ionization (API) and “soft” ionization
techniques such as electrospray ionization (ESI) and atmospheric
pressure chemical ionization.1
3-15
A major issue in LC/MS is relative ionization efficiency and
the ease with which analytes are ionized. In the case of carboxyl-
containing species, ionization is best achieved in the negative ion
2
quantification. H-Coding was achieved through deriva-
tization with deuterated 3-carbinol-1-methyl-d
um iodide. The H-coded derivatization reagent, 3-acyl-
3
-pyridini-
2
current mode using a basic mobile phase in which carboxyl groups
are ionized.1
6-20
Unfortunately, best chromatographic resolution
3
oxymethyl-1-methyl-d -pyridinium iodide, was used in two
ways. One was to differentially label equal fractions of a
sample such that after being recombined and analyzed by
ESI-MS all fatty acids appeared as doublet clusters of ions
separated by roughly 3 amu. This greatly facilitated
identification of fatty acids in complex mixtures. Another
use of stable isotope coding was in comparative quanti-
fication. Control and experimental samples were differ-
entially labeled with nondeuterated and deuterated iso-
topomers of CPM, respectively. After mixing the two
samples, they were analyzed by ESI-MS. The abundance
of a fatty acid in an experimental sample relative to the
control was established by the isotope ratio of the isoto-
pomeric fatty acids. Absolute quantification was achieved
by adding differentially labeled fatty acid standards to
experimental samples containing unknown quantities of
fatty acids. Utility of the method was examined in the
analysis of human serum samples.
with reversed-phase columns is achieved at acidic pH where the
(
1) Claudio, G., Artemis, P., Simopoulos, E. T., Eds. Effects of fatty acids and
lipids in health and disease; World Review of Nutrition and Dietetics 76;
Karger: Basel, 1994.
(2) Ching, K. C. Fatty acids in foods and their health implications; Marcel Dekker,
Inc.: New York 1992.
(
(
3) Dunges, W. Anal. Chem. 1977, 49, 442-445.
4) Eder, K. J. Chromatogr., B 1995, 671, 113-131.
(5) Chen, S.-H.; Chuang, Y.-J. Anal. Chim. Acta 2002, 465, 145-155.
(6) Mofaddel, N.; Desbene-Monvernay, A. Analusis 1999, 27, 120-124.
(7) Vekey, K. J. Chromatogr., A 2001, 921, 227-236.
(8) Brooks, C. J. W.; Edmonds, C. G.; Gaskell, S. J.; Smith, A. G. Chem. Phys.
Lipids 1978, 21, 403-16.
(9) Christie, W. W. Lipids 1998, 33, 343-353.
(
(
10) Griffiths, W. J. Mass Spectrom. Rev. 2003, 22, 81-152.
11) Moldovan, Z.; Jover, E.; Bayona, J. M. Anal. Chim. Acta 2002, 465, 359-
378.
(12) Ng, L.-K. Anal. Chim. Acta 2002, 465, 309-318.
(13) Bogusz, M. J. J. Chromatogr., B 1999, 733, 65-91.
(14) Lachatre, G.; Marquet, P. J. Chromatogr., B 1999, 733, 93-118.
(15) Niessen, W. M. A. J. Chromatogr., A 1998, 794, 407.
(
16) Nagy, K.; Jakab, A.; Fekete, J.; Vekey, K. Anal. Chem. 2004, 76, 1935-
941.
17) Sajiki, J.; Yonekubo, J. Anal. Chim. Acta 2002, 465, 417-426.
1
(
*
To whom correspondence should be addressed. Phone: (765)494-9390.
(18) Carrier, A.; Parent, J. J. Liq. Chromatogr., Relat. Technol. 2001, 24, 97-
Fax: (765)494-0239. E-mail: fregnier@purdue.edu.
107.
5150 Analytical Chemistry, Vol. 79, No. 14, July 15, 2007
10.1021/ac070311t CCC: $37.00 © 2007 American Chemical Society
Published on Web 05/11/2007

ionization of carboxyl groups is suppressed.21 This problem could
be circumvented by using a postcolumn pH adjustment with
Scheme 1. Fatty Acid Derivatization Reactions
18
ammonia, but would increase the complexity of the analysis and
dilute analytes. Another possibility would be to derivatize carboxyl
groups with a reagent that causes them to have a positive charge
in acidic mobile phases. This strategy has been very successful
22,23
in the HPLC/ESI-MS analysis of amino acids and peptides
along with neutral analytes.24,25
Quantification in LC/MS is also related to ionization efficiency.
Matrix suppression of ionization can greatly reduce sensitivity and
complicate quantification. The problem is that ionization efficiency
of analytes is variable, being related to the concentration of other
analytes in the mixture that suppress their ionization. One way
to deal with this problem has been to increase ionization efficiency
through the introduction of positive charge into analytes, as
discussed above. Another is to add an internal standard that
experiences the same suppression of ionization. When it is the
objective to quantify large numbers of analytes as in proteomics
and metabolomics, large numbers of isotopically coded internal
standards of structure identical to analytes are added to a mixture.
sion of ionization.19,30,31 The advantage of this approach is that
internal standards have the same chromatographic properties as
the analytes but can still be differentiated from analytes on the
basis of mass. The disadvantage of this strategy is that a different
internal standard has to be synthesized for each analyte. Also,
this can only be done with known analytes with known structure.
An alternative approach is to attach a stable isotope coded
derivatizing agent to analytes.32 The attractive feature of this
strategy is that it allows all analytes in a sample to be coded with
one isotopomer of the derivatizing agent while those from another
sample can be coded with a second isotopomer.33 Moreover, both
relative and absolute comparisons of concentration between
samples are easily achieved with this method.
2
6
Isotope coded affinity tagging and global internal standard
2
7
technology are examples of the large-scale use of internal
standards for relative quantification in proteomics. Very similar
differential coding methods have been described for metabolom-
2
8,29
ics.
Functional groups in a control sample are derivatized in
vitro with an isotopically coded reagent to target a particular
functional group or molecular feature of analytes in a mixture.
The experimental sample in contrast is derivatized with an isotopic
isoform of the reagent. After the differentially coded samples are
mixed, they are analyzed and as a last step the isotope ratio of
the isotopomers is determined by mass spectrometry. In this way,
differences in concentration between control and experimental
samples are easily determined. Although there is substantial
similarity between global internal standard coding methods in
proteomics and metabolomics, there are differences. The major
one is that almost all peptides contain an amino group or carboxyl
group that can be isotopically coded and used in relative
quantification studies. Metabolites on the other hand have no
single, common functional group that can be used for isotope
coding. This means that a number of reagents will have to be
used to achieve truly global internal standard quantification.
In the analysis of carboxylic acids, stable isotope coded fatty
acids are often used as internal standards due to matrix suppres-
This paper describes a strategy in which carboxyl-containing
analytes are derivatized with a reagent that both increases their
ionization efficiency in HPLC/ESI-MS analysis and isotopically
codes them for internal standard-based quantification.
EXPERIMENTAL SECTION
Materials and Reagents. Fatty acids (FAs), human serum,
2
-bromopyridine, 3-carbinolpyridine, triethylamine (TEA), iodo-
methane, and anhydrous acetonitrile (ACN), were purchased from
Sigma-Aldrich (St. Louis, MO). HPLC grade acetonitrile, acetone,
and formic acid were obtained from Mallinckrodt Baker (Phil-
3
lipsburg, NJ). Iodomethane-d was a product of Cambridge Isotope
Laboratories (Andover, MA). Double-deionized water was pro-
duced by a Milli-Q gradient A10 system from Millipore (Bedford,
MA).
(
(
(
19) Valianpour, F.; Selhorst, J. J. M.; van Lint, L. E. M.; van Gennip, A. H.;
Wanders, R. J. A.; Kemp, S. Mol. Genet. Metab. 2003, 79, 189-196.
20) Perret, D.; Gentili, A.; Marchese S.; Sergi M.; Caporossi, L. Rapid Commun.
Mass Spectrom. 2004, 18, 1989-1994.
Synthesis of 2-Bromo-1-methylpyridinium Iodide (BMP),
3
-Carbinol-1-methylpyridinium Iodide (CMP), and 3-Carbinol-
1-methyl-d -pyridinium Iodide (CMP-d ). Five-fold excess
iodomethane-d or -d was added to 2-bromopyridine (10 mmol,
.97 mL) or 3-carbinolpyridine (10 mmol, 0.96 mL). The solution
21) Jemal, M.; Zheng, O.; Teitz, D. Rapid Commun. Mass Spectrom. 1998, 12,
3
3
4
29-434.
0
3
(
(
22) Yang, W.-C.; Mirzaei, H.; Liu, X.; Regnier, F. E. Anal. Chem. Submitted.
23) Ren, D.; Julka, S.; Inerowicz, H.; Regnier, F. E. Anal. Chem. 2004, 76, 4522-
0
4
530.
24) Quirke, J. M. E.; Adams, C. L.; Van Berkel, G. J. Anal. Chem. 1994, 66,
302-1315.
was stirred at room temperature for 1 h. The crystals were washed
with cold acetone and dried in vacuum. H NMR properties were
as follows: (200 MHz, ACN-d ) for BMP, δ 9.12 (d, 1H, C(3)H),
3
(
1
1
(
(
25) Quirke, J. M. E.; van Berkel, G. J. J. Mass Spectrom. 2001, 36, 1294-1300.
26) Gygi, S. P.; Rist, B.; Gerber, S. A.; Turecek, F.; Gelb, M. H.; Aebersold, R.
Nat. Biotechnol. 1999, 17, 994-999.
(30) Lagerstedt, S. A.; Hinrichs, D. R.; Batt, S. M.; Magera, M. J.; Rinaldo, P.;
McConnell, J. P. Mol. Genet. Metab. 2001, 73, 38-45.
(
(
(
27) Ji, J.; Chakraborty, A.; Geng, M.; Zhang, X.; Amini, A.; Bina, M.; Regnier, F.
J. Chromatogr., B 2000, 745, 197-210.
28) Yang, W.-C.; Mirzaei, H.; Liu, X.; Regnier, F. E. Anal. Chem. 2006, 78, 4702-
(31) Yang, Y. J.; Choi, M. H.; Paik, M.-J.; Yoon, H.-R.; Chung, B. C. J. Chromatogr.,
B 2000, 742, 37-46.
4
708.
(32) Cowen, A. E.; Hofmann, A. F.; Hachey, D. L.; Thomas, P. J.; Belobaba, D.
T.; Klein, P. D.; Tokes, L. J. Lipid Res. 1976, 17, 231-238.
(33) Julka, S.; Regnier, F. E. J. Proteome Res. 2004, 3, 350-363.
29) Shortreed, M. R.; Lamos, S. M.; Frey, B. L.; Phillips, M. F.; Patel, M.;
Belshaw, P. J.; Smith, L. M. Anal. Chem. 2006, 78, 6398-6403.
Analytical Chemistry, Vol. 79, No. 14, July 15, 2007 5151

Table 1. Relative Mass Spectral Response of Fatty Acids (%) Affected by Formic Acid (FA) and Acetonitrile (ACN)
Concentration in Mobile Phases
mobile phase
0
5
mM FA,
0% ACN,
5 mM FA,
50% ACN,
pH 2.66
20 mM FA,
50% ACN,
pH 2.36
50% ACN,
5 mM FA,
pH 2.66
70% ACN,
5 mM FA,
pH 2.82
95% ACN,
5 mM FA,
pH 4.02
FA
pKaa
pH 4.30
C10:0
C12:0
C14:0
C16:0
C18:0
C20:0
C22:0
C24:0
7.1-7.3
∼7.5
100
100
100
100
100
100
100
100
12.3
16.8
1.8
3.4
100
100
100
100
100
100
100
100
50.7
59.0
58.7
48.0
36.6
55.5
57.1
85.7
7.8
13.1
8.1-8.2
8.6-8.8
10.5
25.5
8.1
20.0
41.6
25.0
57.7
80.0
125.0
161.3
22.6
62.8
22.7
90.0
58.3
150.0
225.8
162.3
375.0
a
pKa values for C10:0 to C16:0 were from ref 53, which were determined in water solution. The value for C18:0 was from ref 54, which was
determined in water solution.
8
(
.72-8.76 (d, 1H, C(6)H), 8.49-8.53 (d, 1H, C(4)H), 8.12-8.16
m, 1H, C(5)H), 4.44 (s, 3H, NCH ). For CMP, δ 8.90 (s, 1H,
C(2)H), 8.70-8.76 (d, 1H, C(6)H), 8.51-8.56 (d, 1H, C(4)H),
.10-8.15 (m, 1H, C(5)H), 4.43 (s, 3H, NCH ).
Derivatization of Fatty Acids. Ten microliters of standard
range of 150-400 or in the positive mode for fatty acid derivatives
within the m/z range of 250-500.
3
Analysis of Human Serum Fatty Acids. Total fatty acid
extraction was carried out by following a protocol suggested by
Valianpour et al.19 with minor modifications. Briefly, 10 mL of
ACN/HCl (37%) (4:1, v/v) and 1 mL of serum were added to a
glass vial. The vial was capped with a Teflon/silicone disk in a
closed screw cap and incubated at 90 °C for 2 h. After cooling to
room temperature, 7.5 mL of hexane was added. The vial was
vortex-mixed for 20 s and centrifuged at 1500g for 1 min. A 1.0-
mL aliquot of upper phase was placed in a 15-mL centrifuge tube
and flushed with nitrogen to dryness. Meanwhile, 100 µL of the
standard fatty acid mixture (even carbon number only from C10:0
to C24:0 along with C16:1, C18:1, and C18:2; 0.2 µM concentration)
8
3
fatty acid mixture (0.5 µmol/mL each, ∼5 µmol/mL total FA, in
acetone) was mixed with 20 µL of BMP (15.0 mg/mL or 50 µmol/
mL, in ACN) and 20 µL of CMP (20.0 mg/mL or 200 µmol/mL,
in ACN). After mixing, 1.0 µL of TEA was added, and the solution
was heated in water at 50 °C for 30 min. The derivatizing reaction
is presented in Scheme 1. The resulting solution was diluted 7:3
with DI water (solution/water, v/v) before it was subjected to
analysis. The derivatizing reagents and derivatives were stable in
ACN at ambient temperature for at least one week.
HPLC/ESI-MS. The HPLC/ESI system consisted of a capil-
lary HPLC system (1100 Series LC, Agilent) and an ESI source
of time-of flight (TOF) mass spectrometer (MSD TOF, Agilent).
The system was controlled by ChemStation software (Agilent).
The instruments were set up at nontemperature control. The
separations were performed on a low TFA Vydac analytical column
3
were processed and derivatized with CMP-d in parallel. These
were used as internal standards for quantitative analysis. For
qualitative analysis, dried sample was reconstituted with 0.4 mL
of acetone. Two 0.1-mL aliquots were derivatized with CMP and
CMP-d
then mixed at a ratio of 1:1. The mixture was subjected to HPLC/
MS analysis after 7:3 dilution with H O (solution/H O, v/v). The
3
separately, following the procedure described above, and
(
C4 Mass Spec, 4.6 mm × 250 mm). Typical elution condition
were to begin by maintaining 70% mobile phase A (H O + 0.05%
formic acid) and 30% mobile phase B (95% ACN + 5% H O + 0.05%
formic acid) for 0-5 min and then increase solvent B from 30 to
0% in 5 min followed by an increase in solvent B from 40 to 100%
in 35 min with a 5-min hold at 100% B. The flow rate was set at
.75 mL/min, and the injection volume was 40.0 µL. The column
2
2
2
derivatization for quantitative analysis was performed as follows:
20 µL of the above CMP-labeled serum was added into the above
dried internal standard vial; the mixture was diluted to 1.0 mL by
ACN, further diluted to 70% ACN with water, and then analyzed
by HPLC/MS.
2
4
0
was preconditioned by running a blank with the above elution
profile and then pumping mobile phase A for an additional 15 min.
LC/ESI-MS chromatograms were acquired in the positive ion
mode with the capillary voltage set at 4000 V and fragmentor at
RESULTS AND DISCUSSION
Enhancing Electrospray Ionization by Derivatization. It has
been reported21 that increasing mobile-phase concentration of
formic acid in reversed-phase chromatography (RPC) from 0 to
10 mM decreased the detection sensitivity of carboxylic acid
compounds 5.9-fold in the negative ion current mode of ESI-MS.
Similar effects were noted with fatty acids in this work (Table 1).
The problem is larger with smaller fatty acids as noted in Table
1. This is an important issue because it is necessary to use acidic
mobile phases in RPC to eliminate the impact of residual silanol
groups on separations.
1
75 V while the dry temperature was set at 350 °C and dry gas
flow was maintained at 13 L/min. The mass acquisition range was
from m/z 150 to 1299.
ESI-MS/MS. ESI-MS or ESI-MS/MS analyses were per-
formed on an API QSTAR (Applied Biosystems, Framingham,
MA) equipped with an ESI ion source. Free fatty acids or FA
2
derivatives were prepared in various ACN/H O ratio solvents
containing various concentrations of formic acid and injected into
the ESI source by infusion at 10 µL/min. Both single-dimension
mass spectra and tandem mass spectra were acquired typically
for 2 min in the negative mode for free fatty acids within an m/z
It is worth noting that the change in response seen in Table 1
is not proportional to the change in carboxyl group dissociation
as a function of pH. For example, it can be calculated that the
degree of dissociation of C10:0 in going from zero to 5 mM formic
5152 Analytical Chemistry, Vol. 79, No. 14, July 15, 2007

Figure 1. Mass spectra of derivatized fatty acids in various matrixes: (A) 50% ACN + 0 mM formic acid; (B) 50% ACN + 5 mM formic acid;
C) 50% ACN + 20 mM formic acid; (D) 95% ACN + 5 mM formic acid.
(
Figure 2. Comparison of detection limits by mass spectrometry for (A) free fatty acids (25 µM, in 80% ACN + 5 mM formic acid) and (B)
derivatized fatty acids (10 nM, in 80% ACN + 5 mM formic acid).
acid would be ∼43-fold. Going to 20 mM would produce another
-fold decrease, i.e., an 86-fold change compared to no formic acid.
However, as shown in Table 1, the actual changes were 8.1- and
5-fold, respectively. This is below what would be expected,
suggesting that formic acid might be making some additional
contribution to ionization.
The fatty acid mixture was then derivatized with BMP and
CMP. Esterification of carboxylic acids with this reaction has been
previously reported37 as proceeding via an intermediate 2-acyloxy-
1-methylpyridinum iodide that undergoes nucleophilic attack by
the alcohol in the presence of a tertiary amine. The amine
sequesters the hydrogen iodide produced in the reaction. The
fact that activation of carboxylic acids and esterification takes place
in a single step is an obvious advantage that has been exploited
in the preparation of fluorescent derivatives of fatty acids for HPLC
analysis.3
2
5
Ionization efficiency in the negative ion current mode of
ionization is also impacted by the concentration of organic solvent
(Table 1). Again the effect is greater with the smaller C10 than
8-40
the larger C20 fatty acid in the range of 50-95% acetonitrile. In
fact, detection sensitivity increased with larger fatty acids. Clearly,
(
34) Rahman, M. A.; Ghosh, A. K.; Bose, R. N. J. Chem. Technol. Biotechnol.
3
4-36
fatty acid dissociation is suppressed in acetonitrile
as seen in
(1979-1982) 1979, 29, 158-162.
(
(
(
35) Grunwald, E.; Berkowitz, B. J. Am. Chem. Soc. 1951, 73, 4939-4944.
36) Lynch, C. C.; Lamer, V. J. J. Am. Chem. Soc. 1938, 60, 1252-1259.
37) Mukaiyama, T.; Usui, M.; Shimada, E.; Saigo, K. Chem. Lett. 1975, 1045-
1048.
the cases of the C10, C12, and C14 fatty acids. But beyond C20,
the impact of carboxyl dissociation on ionization in 95% acetonitrile
is not a factor.
Analytical Chemistry, Vol. 79, No. 14, July 15, 2007 5153

Figure 3. ESI tandem mass spectra showing fragmentation of derivatized C18:0, C18:1, and C18:2 fatty acids. The zoom ratio of the inset
spectra are 18:1 in (A), 18:1 in (B), 11:1 in (C), and 7:1 in (D). The boldface and italicized numbers in (C) and (D) indicate the unique masses
for unsaturated, derivatized fatty acids.
Quaternization of fatty acids according to Scheme 1 greatly
increased the ease with which they were ionized under acidic
conditions (Figure 1). No significant change in ESI response in
the positive ion mode was observed (Figure 1A-C) with increas-
ing concentrations of formic acid. The fatty acid detection limit
before and after quaternization can be seen in Figure 2. Signal
intensity with 25 µM free fatty acids was close to that of 10 nM
derivatized fatty acids. Detection sensitivity was enhanced roughly
2
500-fold by derivatization. Even though acetonitrile still reduced
detection sensitivity by 2-3-fold (Figure 1B, D), the detection
limits were far below those of free fatty acids. As with free fatty
acids, the improvement of detection sensitivity was greater with
the larger (>C18) than smaller (<C18) fatty acids.
Tandem Mass Spectrometry. Collision-induced fragment ion
spectra of the C18:0, C18:1, and C18:2 fatty acid derivatives were
found to correlate with their structure as seen in Figure 3. Ions
common to all the fatty acid spectra were at m/z 107, 124, and
Figure 4. Reconstructed ion chromatogram of CMP derivatives of
standard fatty acid mixture. The concentration was 200 nM for each,
and the injection volume was 40 µL. Elution protocol is given in the
Experimental Section.
1
78. The ions at m/z 107 and 124 were derived from the
N-pyridylcarbinol moiety. The first lacks the oxygen, and the
second is from a rearranged proton-transfer fragmentation with
retention of the oxygen. The fragment ion at m/z 165 is the result
of fragmentation with a McLafferty rearrangement. An ion at
m/z 178 was also seen in all the spectra that results from
fragmentation of the fatty acid ester γ to the C-terminus of the
fatty acids. The series of ions at m/z 178, 192, 206, 220, 234, 248,
indicated this is the spectrum of a saturated fatty acid. Comparing
spectra derived from the d
, it is seen that all the ions in the spectrum of the d
Figure 3B) are 3 amu higher than those in the d
0 3
and d isotopomers of derivatized C18:
4
1
0
(
(
3
isotopomer
0
isotopomer
Figure 3A). This confirms that the N-pyridylcarbinol group is
attached to all the fragments ions in the fatty acid spectra. Panels
C and D in Figure 3 show that it is possible to locate the position
of double bonds in the C18:1 and C18:2 acids examined. The
spectrum of C18:1 (Figure 3C) shows an interruption of the
aliphatic series noted above after m/z 248. The difference be-
tween m/z 248 and 274 in the spectrum is 26 amu, equivalent to
a -CHdCH- group. The prominent ion at m/z 288 is from
2
62, 276, 290, 304, 318, 332, 346, and 360, all differing by 14 amu,
(
38) Lingeman, H.; Hulshoff, A.; Underberg, W. J. M.; Offermann, F. B. J. M. J.
Chromatogr. 290, 184, 215-222.
(
39) Baty, J. D.; Willis, R. G.; Travendale, R. Biomed. Mass Spectrom. 1985, 12,
5
65-569.
(
(
40) Baty, J. D.; Pazouki, S. J. Chromatogr. 1987, 395, 403-411.
41) McLafferty, F. W. Anal. Chem. 1959, 31, 82-87.
5154 Analytical Chemistry, Vol. 79, No. 14, July 15, 2007

Table 2. Mass, Retention Time, and Detection Sensitivity of Fatty Acid Derivatives
dynamical
detection limit
(nM/pg)
derivative
C10:0
m/z
tr (min)
linearitya
range (µM)
278.21
22.1
y ) 6064.2x + 336.5
0.02-1.0
4.0/27.6
2.0/16.0
2.0/18.3
2.0/20.3
2.0/22.4
2.0/20.5
2.0/22.6
1.0/11.3
1.0/12.5
1.0/13.6
1.0/14.7
2
r ) 0.9959
C12:0
C14:0
C16:1
C18:2
C16:0
C18:1
C18:0
C20:0
C22:0
C24:0
306.24
334.27
360.28
386.30
362.30
388.31
390.33
418.36
446.39
474.42
26.4
30.1
31.2
42.3
33.4
34.3
36.5
39.7
42.5
45.0
y ) 34382x + 488.4
0.004-0.4
0.004-0.2
0.004-1.0
0.004-0.2
0.004-0.2
0.002-0.2
0.002-0.2
0.002-0.2
0.002-0.2
0.002-0.2
2
r ) 0.9968
y ) 73297x + 383.3
2
r ) 0.9993
y ) 61084x + 1946.1
2
r ) 0.9954
y ) 79985x + 1093
2
r ) 0.9962
y ) 147025x + 677.6
2
r ) 0.9993
y ) 93147x + 550.0
2
r ) 0.9946
y ) 97140x + 788.9
2
r ) 0.9937
y ) 34445x + 339.3
2
r ) 0.9959
y ) 32115x + 629.4
2
r ) 0.9926
y ) 38064x + 1188.4
2
r ) 0.9966
a
In the linearity equation, y is instrument response and x is concentration (µM).
cleavage allylic to the double bond. In the case of the C18:2 fatty
acid (Figure 3D), the dominant ion in the cluster of ions associated
with the double bonds is again at m/z 288, allylic to the first double
bond. Cleavage allylic to the second double bond at m/z 328 is
also apparent in the spectrum. The fragmentation pattern of the
fatty acid derivatives in Figure 3 is similar to the extensive descrip-
tion of the fragmentation of picolinyl esters of fatty acids.4
Absolute and Relative HPLC/ESI-MS Quantitation by
Table 3. Linearity and Dynamical Range of Light and
Heavy Isotope Labeled Mixtures of Fatty Acid
Derivatives
m/z
fatty
acid
dynamical
ratio range
light
heavy
linearitya
2
C10:0
C12:0
C14:0
C16:1
C18:2
C16:0
C18:1
C18:0
C20:0
C22:0
C24:0
278.21
306.23
334.23
360.28
386.30
362.30
388.31
390.33
418.36
446.39
474.42
281.23
309.26
334.26
363.30
386.32
365.32
391.33
393.35
421.38
449.41
477.44
y ) 1.003x; r ) 0.9992
1:1-1:50
1:1-1:100
1:1-1:100
1:1-1:100
1:1-1:100
1:1-1:100
1:1-1:100
1:1-1:100
1:1-1:100
1:1-1:50
1:1-1:10
2-46
2
y ) 0.8902x; r ) 0.9972
2
y ) 1.0934x; r ) 0.9984
2
y ) 1.0745x; r2 ) 0.9991
H-Coded Derivatization. Long-chain fatty acids and their esters
2
y ) 0.9185x; r ) 0.9998
are strongly retained on reversed-phase columns. Mobile phases
containing up to 95% organic solvent are often required to effect
elution. Even with this, the retention is still so strong on C18
2
y ) 1.0247x; r ) 0.9995
y ) 1.0745x; r2 ) 0.9991
2
y ) 0.9774x; r ) 0.9998
2
y ) 1.0179x; r ) 0.9986
1
8,41,47
columns that investigators often go to C8 columns
or C6
2
y ) 0.7967x; r ) 0.9997
4
8
columns. A C4 reversed-phase column was use in this work.
Figure 4 shows the reconstructed ion chromatogram of 3-acyl-
oxymethyl-1-methylpyridinium iodide (AMMP) derivatives of a
standard mixture of eight saturated and three unsaturated fatty
y ) 0.8727x; r2 ) 0.9954
a
x, theoretic concentration ratio; y, experimental mass spectrometric
peak intensity ratio.
4
1
acids. Although difficult to achieve in the past, complete
separation of all the physiological significant fatty acids is seen in
Figure 4. The retention times and m/z values identifying these
fatty acids are summarized in Table 2.
tion. Linearity spanning roughly 2 orders of magnitude for most
fatty acids was observed with regression coefficients greater than
0
.99. The limit of detection (LOD) for standard AMMP mixtures
Quantitative aspects of the method were examined in three
trials using a sample of 11 fatty acids in the concentration range
from 2.0 µM to 1.0 nM that had been derivatized with CMP. Peak
height and peak area were derived from extracted ion chromato-
grams using characteristic ions. Instrument response for individual
fatty acids was fitted to a linear regression analysis of concentra-
was determined from mass spectra using a signal-to-noise ratio
limit of 3. The LOD varied from 1.0 to 4.0 nM, i.e., 10-pg level/
inject (40 µL), for the fatty acids studied here. Detailed data are
summarized in Table 2.
In vitro stable isotope coding for comparative quantification
has been very useful in proteomics.33 This strategy has generally
been performed in the following manner. Sample and control are
derivatized separately with isotope coded or non isotope coded
derivatizing reagents, respectively. When these differentially coded
samples are then mixed at a certain ratio, e.g., 1:1, and analyzed
by ESI-MS, doublet clusters of ions will be seen that are separated
by the difference in mass between the isotopically coded deriva-
tizing reagents. The MS peak height (area) ratio of this pair of
(
(
(
(
42) Dobson, G.; Christie, W. W. Trends Anal. Chem. 1996, 15, 130.
43) Christie, W. W. Lipids 1998, 33, 343-353.
44) Jensen, N. J.; Gross, M. L. Mass Spectrom. Rev. 1987, 6, 497-536.
45) Christie, W. W.; Brenchany, E. Y.; Holman, R. T. Lipids 1987, 22, 224-
2
28.
46) Lie Ken Jie, M. S. F.; Choi, Y. C. J. Am. Oil Chem. Soc. 1992, 69, 1245-
247.
(
1
(
47) Kargas, G.; Rudy, T. J. Chromatogr. 1990, 526, 331-340.
(48) Mehta, A.; Oeser, A. M.; Carlson, M. G. J. Chromatogr., B 1998, 719, 9-23.
Analytical Chemistry, Vol. 79, No. 14, July 15, 2007 5155

Figure 5. Mass spectra of light (d0) and heavy (d3) isotope-labeled CMP derivatives of stearic acid mixed at different ratios.
Figure 6. (A) Total ion chromatogram of human serum extract derivatized with light (d0) and heavy (d3) isotope-labeled CMP at the isotope
ratio of 1:1; (B) accumulated mass spectrum within the elution period from 20 to 55 min in (A) with the insets of extracted spectra of d0/d3-
labeled C16:0, C18:1, and C18:0; (C) reconstructed ion chromatogram of identified fatty acids from (B).
isotopomers is proportional to their concentration ratio in the
sample and control.
method exhibits linearity for concentration ratios ranging from
1:10 to 1:100 for the various fatty acids studied with high
correlation coefficients (Table 3). Figure 5 illustrates a series of
spectra from isotopically derivatized stearic acid. No chromato-
graphic isotope effect was observed with the CMP reagent.
Although the presence of deuterium in a coding agent can cause
chromatographic isotope effects, placing the deuterium on a
A similar strategy will be of value in metabolomics and other
fields, where it is the objective to compare the concentration of
analytes between samples, especially when the stable isotope
coded isoforms of analytes coelute from chromatographic systems.
Coelution of coded isoforms is necessary to minimize several
problems. One is that ionization of isotopomers may be differen-
tially suppressed if they do not coelute. A second is that the isotope
ratio of isotopomers will change across a peak if they are partially
2
2
quaternary amine as was done here negates the isotope effect.
Applications. The utility of the method was examined in the
analysis of fatty acids in human serum. Human serum or plasma
contains several physiologically important fatty acids, among
49
resolved. Both of these phenomena can impact linearity and the
accuracy of quantification. These issues were examined with the
CMP and CMP-d
3
derivatizing agent. The results show that this
(49) Julka, S.; Regnier, F. J. Proteome Res. 2004, 3, 350-363.
5156 Analytical Chemistry, Vol. 79, No. 14, July 15, 2007

Table 4. Fatty Acids Found in Human Seruma
m/z
concentration (µM)
fatty acid
found
light
heavy
retention
determined
ref 21
ref 53b
labeled
labeled
time (min)
(mean, % RSD, n ) 4)
(mean, % RSD, n ) 8)
(mean ( SD, n ) 23)
C10:0
C12:0
C14:0
C20:5
C18:3
C16:1
C18:2
C16:0
C18:1
C20:2
C18:0
C20:1
C20:0
C22:0
C24:0
278.21
306.23
334.27
408.25
384.29
360.28
386.30
362.30
388.31
414.30
390.33
416.32
418.36
446.39
474.42
281.23
309.26
337.29
413.28
387.31
363.30
389.32
365.32
391.33
417.33
393.35
419.34
421.38
449.41
477.44
22.1
26.4
30.1
30.8
31.0
31.2
32.3
33.4
34.3
36.2
36.5
38.5
39.7
42.5
45.0
21.8, 6.8
32.1, 4.7
411.4, 4.5
13.9 ( 17.2
37.6 ( 24.7
156.8 ( 84.8
256.7 ( 73.8
88.0 ( 52.6
c
nq
nq
123.7, 5.6
1822.5, 6.2
2280.0, 5.1
1485.0, 5.9
nq
204 ( 62.3
3990.9 ( 646.5
2255.4 ( 452.1
1929.9 ( 586.4
1166, 3.8
1175, 4.1
2445.0, 6.3
nq
72.0, 4.2
73.1, 6.5
42.2, 6.9
683.0 ( 204.6
74, 4.0
34, 4.3
67, 4.6
a
Parallel experiments were performed in quadruplicate from the identical derivatized serum sample. The results were the average of four runs.
The results represent the concentration range of total fatty acid in normal serum. The original data were reported in the total concentration of
b
total fatty acid and percentage of individual fatty acid in serum. These values were converted to the present form by the authors. c Not quantified.
which six are most noteworthy: myristic (C14:0), palmitic
C16:0), palmitoleic (C16:1), stearic (C18:0), oleic (C18:1), and
2H-coded derivatizing reagents (CMP-d
inexpensively prepared by the methylation of 3- carbinolpyridine
with iodomethane-d . In order to obtain reliable results, the
standard fatty acids were processed and derivatized with CMP-d
3 3
). CMP-d is easily and
(
5
0
linoleic (C18:2). The analysis of fatty acids in human serum/
plasma has been the object of numerous HPLC development
3
3
39,40,47,48
studies.
Because of the complexity of serum, identification
simultaneously with the serum sample and then put into deriva-
tized serum as a set of internal standards to perform HPLC/ESI-
MS quantitation analysis. The results are reported in Table 4.
Some fatty acids were found in the sample but not quantified
because they were not included in our standard mixture of fatty
of fatty acids directly by HPLC/MS/MS may be time-consuming.
We have developed a simple method for this objective by first
tagging and sorting fatty acids through derivatization and then
performing MS/MS for the confirmation of structure. Serum
1
6,52
samples were first derivatized with CMP and CMP-d
3
separately,
acids. The results were close to those previously reported.
and the two resulting solutions were then mixed at a 1:1 ratio
and analyzed by HPLC/ESI-MS. From the total ion chromatogram
CONCLUSIONS
It can be concluded from this work that in vitro derivatization
of fatty acids affords the opportunity to incorporate features that
can substantially enhance their detection and quantification by
HPLC/ESI-MS/MS. One of these features is the introduction of
a quaternary amine. Clearly this greatly increases detection
sensitivity in the positive ion mode of detection by avoiding
suppression of ionization in acidic mobile phases. A second feature
of the derivatized fatty acid esters was that they showed unique
fragmentation in ESI-MS/MS that made it very easy to identify
their position in chromatograms along with the position of double
bonds in their carbon skeleton. A third feature is the introduction
of isotopic coding that facilitates both qualitative and quantitative
analysis. Comparative quantitation and absolute quantitation were
(scan range 200.0-550.0) in Figure 6A, the accumulated mass
spectrum in Figure 6B was obtained. From this mass spectrum,
fatty acids were recognized by looking for the doublet ion clusters
with a 3.01-3.04 amu difference in mass and a peak height ratio
of 1:1. Doublet cluster recognition was achieved both manually
and with an algorithm designed for this purpose in proteomics.51
Fatty acids were then identified by MS/MS analysis. Figure 6C
is the reconstructed ion chromatogram of fatty acids identified
by this method. Fifteen fatty acids were identified in the human
serum sample.
Absolute quantitation of serum fatty acids was achieved by
using isotopically derivatized internal standards. It has been
2
common to use H-coded fatty acids as internal standards for GC
2
realized by H-coding of derivatization reagents. The reported
or HPLC/MS quantitation of serum fatty acids. It is relatively
methods were validated by the analysis of fatty acids in human
serum. Finally, it is concluded that the new technique described
here should find wide application in future studies that require
the analysis of fatty acids in complex samples.
2
conveniently achieved through the addition of H-coded fatty acids
2
into the sample as internal standards. Nevertheless, H-coded fatty
acids are expensive, but they are commercially available by custom
2
synthesis. The strategy used here was to generate H-coded esters
of fatty acids as internal standards by derivatizing fatty acids with
ACKNOWLEDGMENT
(
(
50) Carlson, M. G.; Oeser, L.; Swift, L. Diabetes 1997, 46, 221A.
51) Zhang, X.; Hines, W.; Adamec, J.; Asara, J/ M.; Naylor, S.; Regnier, F. E. J.
Am. Soc. Mass Spectrom. 2005, 16, 1181-1191.
This work was supported by Grant AG13319 from the National
Institute of Health.
(
(
(
52) Shimomura, Y.; Sugiyama, S.; Takamura, T.; Kondo, T.; Ozawa, T. J.
Chromatogr. 1986, 383, 9-17.
53) Kanicky J. R.; Poniatowski, A. F.; Mehta, N. R.; Shah, D. O. Langmuir 2000,
Received for review February 13, 2007. Accepted April 10,
2007.
1
6, 172-177.
54) Kanicky, R.; Shah, D. O. J. Colloid Interface Sci. 2002, 256, 201-207.
AC070311T
Analytical Chemistry, Vol. 79, No. 14, July 15, 2007 5157