Gas chromatography/electron-capture negative ion mass spectrometry for the quantitative determination of 2- and 3-hydroxy fatty acids in bovine milk fat

Article abstract of DOI:10.1021/jf800647w

2- and 3-hydroxy fatty acids (2- and 3-OH-FAs) are bioactive substances reported in sphingolipids and bacteria. Little is known of their occurrence in food. For this reason, a method suitable for the determination of OH-FAs at trace levels in bovine milk fat was developed. OH-FAs (and conventional fatty acids in samples) were converted into methyl esters and the hydroxyl group was derivatized with pentafluorobenzoyl (PFBO) chloride to give PFBO-O-FA methyl esters. These derivatives with strong electron affinity were determined by gas chromatography interfaced to mass spectrometry using electron-capture negative ion in the selected ion monitoring mode (GC/ECNI-MS-SIM). This method proved to be highly sensitive and selective for PFBO-O-FA methyl esters. For the analysis of samples, two internal standards were used. For this purpose, 9,10-dideutero-2-OH-18:0 methyl ester (ISTD-1) from 2-OH-18:1(9c) methyl ester as well as the ethyl ester of 3-PFBO-O-12:0 (ISTD-2) was synthesized. ISTD-1 served as a recovery standard whereas ISTD-2 was used for GC/MS measurements. The whole-sample cleanup consisted of accelerated solvent extraction of dry bovine milk, addition of ISTD 1, saponification, conversion of fatty acids into methyl esters by use of boron trifluoride, separation of the methyl esters of OH-FAs from nonsubstituted FAs on activated silica, conversion of OH-FAs methyl esters into PFBO-O-FA methyl esters, addition of ISTD-2, and measurement by GC/ECNI-MS-SIM. By this method, ten OH-FAs were quantified in bovine milk fat with high precision in the range from 0.02 ± 0.00 to 4.49 ± 0.29 mg/100 g of milk fat.

Full text of DOI:10.1021/jf800647w

5500 J. Agric. Food Chem. 2008, 56, 5500–5505
Gas Chromatography/Electron-Capture Negative Ion
Mass Spectrometry for the Quantitative
Determination of 2- and 3-Hydroxy Fatty Acids in
Bovine Milk Fat
RAMONA JENSKE AND WALTER VETTER*
Institute of Food Chemistry, University of Hohenheim, Garbenstrasse 28, D-70599 Stuttgart, Germany
2- and 3-hydroxy fatty acids (2- and 3-OH-FAs) are bioactive substances reported in sphingolipids
and bacteria. Little is known of their occurrence in food. For this reason, a method suitable for the
determination of OH-FAs at trace levels in bovine milk fat was developed. OH-FAs (and conventional
fatty acids in samples) were converted into methyl esters and the hydroxyl group was derivatized
with pentafluorobenzoyl (PFBO) chloride to give PFBO-O-FA methyl esters. These derivatives with
strong electron affinity were determined by gas chromatography interfaced to mass spectrometry
using electron-capture negative ion in the selected ion monitoring mode (GC/ECNI-MS-SIM). This
method proved to be highly sensitive and selective for PFBO-O-FA methyl esters. For the analysis
of samples, two internal standards were used. For this purpose, 9,10-dideutero-2-OH-18:0 methyl
ester (ISTD-1) from 2-OH-18:1(9c) methyl ester as well as the ethyl ester of 3-PFBO-O-12:0 (ISTD-
2) was synthesized. ISTD-1 served as a recovery standard whereas ISTD-2 was used for GC/MS
measurements. The whole-sample cleanup consisted of accelerated solvent extraction of dry bovine
milk, addition of ISTD 1, saponification, conversion of fatty acids into methyl esters by use of boron
trifluoride, separation of the methyl esters of OH-FAs from nonsubstituted FAs on activated silica,
conversion of OH-FAs methyl esters into PFBO-O-FA methyl esters, addition of ISTD-2, and
measurement by GC/ECNI-MS-SIM. By this method, ten OH-FAs were quantified in bovine milk fat
with high precision in the range from 0.02 ( 0.00 to 4.49 ( 0.29 mg/100 g of milk fat.
KEYWORDS: r-Hydroxy fatty acids; ꢀ-hydroxy fatty acids; milk fat; quantification; GC/ECNI-MS-SIM
INTRODUCTION
accessible to gas chromatography (GC) with mass spectrometry
(MS). In many cases, the OH-FAs are converted into the
corresponding methyl esters either without derivatization of the
hydroxyl group or by the additional conversion of the hydroxyl
group into trimethylsilyl (TMS) ethers, which are measured by
electron ionization mass spectrometry (EI-MS) in the selected
ion monitoring (SIM) mode or by MS-MS (6, 10–12, 14–16).
In addition, methyl/pentafluorobenzoyl (ME/PFBO) derivatives
have been analyzed by GC/MS in the electron-capture negative
ion (ECNI) mode (8, 9). In order to increase both the selectivity
and sensitivity of analyses, PFBO derivatives have been
measured in the SIM mode by using the molecular ion [M]-
that specifies the corresponding OH-FA (7, 9).
Fatty acids are principal constituents of food and biological
samples. They consist of a carboxyl group and usually a long
hydrophobic alkyl chain with 4-26 carbons. In addition to these
standard fatty acids, several substituted fatty acids, such as 2-
and 3- (or R- and ꢀ-) hydroxy fatty acids (2- and 3-OH-FAs),
are naturally occurring in lipids. For instance, 2-OH-FAs are
constituents of sphingolipids in the lipid membranes of plants
and animals (1–5). In addition, 2- and 3-OH-FAs are charac-
teristic compounds of Lipid A, the lipid component of the
lipopolysaccharides (LPS), which are located in the outer
membrane of Gram-negative bacteria, whereas 3-OH-FAs were
used as chemical markers for the determination of endotoxin
in clinical and environmental samples (6–11). While many
studies have dealt with this subject matter, little data existed
on the occurrence and quantities of 2- and 3-OH-FAs in
food (12, 13).
The aim of the present study was to develop a highly selective
and sensitive method for the quantitative determination of 2-
and 3-OH-FAs in bovine milk fat in the submilligram/100 g
range. Given the low amounts of OH-FAs to be detected, the
method included (i) separation of the methyl esters of hydrox-
ylated fatty acids from those of nonhydroxylated fatty acids,
followed by (ii) conversion of OH-FAMEs into the cor-
responding PFBO derivatives (PFBO-O-FAMEs). Due to their
electron-capturing properties, the PFBO derivatives were ana-
Quantitative determinations of 2- and 3-OH-FAs usually
require the conversion of these fatty acids into derivatives
* Corresponding author: tel +49 711 459 24016; fax +49 711 459
24377; e-mail w-vetter@uni-hohenheim.de.
10.1021/jf800647w CCC: $40.75  2008 American Chemical Society
Published on Web 06/21/2008

Hydroxy Fatty Acids in Bovine Milk Fat
J. Agric. Food Chem., Vol. 56, No. 14, 2008 5501
Standard Solutions. Esterification of free OH-FAs was performed
according to the official standard procedure (Standard Method of the
German Society for Fat Science) (18). A mixture containing ap-
proximately 5 µg each of 2- and 3-OH-FA and 25 µL of ISTD 1 solution
was treated (5 min/80 °C) with 0.5 mL of 0.5 M methanolic KOH.
After cooling, 1 mL of methanolic BF3 was added and the mixture
was heated for an additional 5 min at 80 °C. Furthermore, approximately
100 µg of 3-OH-12:0 was treated with 0.5 M ethanolic KOH and
ethanolic BF3 in the same way.
lyzed by GC/ECNI-MS in the SIM mode. In order to meet
quality control standards, two internal standards were synthe-
sized: a dideuterated hydroxy fatty acid for use as recovery
standard for the entire sample cleanup and a second internal
standard to even out variations in the performance of the mass
spectrometer.
MATERIALS AND METHODS
Materials and Chemicals. Bovine milk (lipid content 3.5%,
Weihenstephan, Freising, Germany) was used as food sample. 2-OH-
and 3-OH-FA standards were obtained from the following commercial
sources: 2-hydroxyoctanoic acid (2-OH-8:0), 2-hydroxydecanoic acid
(2-OH-10:0), 2-hydroxytetradecanoic acid (2-OH-14:0), γ-decalactone,
and γ-dodecalactone were from Sigma-Aldrich (Steinheim, Germany);
2-hydroxydodecanoic acid (2-OH-12:0), 2-hydroxyhexadecanoic acid
(2-OH-16:0), 2-hydroxyoctadecanoic acid (2-OH-18:0), 2-hydroxy-
eicosanoic acid (2-OH-20:0), 3-hydroxyoctanoic acid (3-OH-8:0),
3-hydroxydecanoic acid (3-OH-10:0), 3-hydroxydodecanoic acid (3-
OH-12:0), 3-hydroxytetradecanoic acid (3-OH-14:0), 3-hydroxyhexa-
decanoic acid (3-OH-16:0), 3-hydroxyoctadecanoic acid (3-OH-18:0),
and 2-hydroxy-(9c)-octadecenoic acid methyl ester [3-OH-18:1(9c)-
ME] were from Larodan (Malmo¨, Sweden).
Cyclohexane (purest; VWR, Darmstadt, Germany) and ethyl acetate
(purest; Acros Organics, Geel, Belgium) were combined (1:1 v/v) and
distilled to obtain an azeotropic mixture (54:46 v/v). n-Hexane (HPLC
gradient grade) and methanol (HPLC gradient grade) were purchased
from Fisher Scientific (Ulm, Germany); diethyl ether, ethanol (absolute),
and acetonitrile (HPLC gradient grade) were from Roth (Karlsruhe,
Germany); and benzene and triethylamine (TEA) were from Merck
(Darmstadt, Germany). All other solvents used in this study were
distilled before use.
Deuterium (99.9%) was obtained from Isotec (Miamisburg, OH);
Wilkinson’s catalyst [(Ph3P)3RhCl(I)] and pentafluorobenzoyl chloride
(99%, PFBO-Cl) were from Sigma-Aldrich (Steinheim, Germany);
diatomaceous earth (isolute-HM-N) was from Separtis (Grenzlach-
Wyhlen, Germany); discovery silver-ion SPE cartridges (750 mg/6 mL)
were from Supelco (Bellefonte, PA); and potassium hydroxide (KOH)
and sodium chloride (NaCl) were from Roth (Karlsruhe, Germany).
Boron trifluoride-methanol complex solution (13-15% BF3 in metha-
nol) was from Riedel-de-Hae¨n (Seelze, Germany), and ethanolic BF3
(∼10%, ∼1.3 M, purris) and silica gel G60 were from Fluka (Steinheim,
Germany). Prior to use, silica gel G60 was dried (activated) overnight
at 120 °C.
After the reactions, vials were cooled in an ice bath, successively 2
mL of n-hexane and 2 mL of saturated sodium chloride solution were
added, and the organic phase including the methylated (or ethylated)
FAs (FAMEs or FAEEs) was separated, evaporated to dryness under
a gentle stream of nitrogen at room temperature, and subjected to the
PFBO derivatization procedure as described below. After conversion,
the corresponding 3-PFBO-O-12:0-EE was used as internal standard
for GC/MS measurements (ISTD 2).
Sample Preparation. The bovine milk sample was lyophilized prior
to extraction. Lipids were extracted by accelerated solvent extraction
(ASE, Dionex, Idstein, Germany) by using 11 mL extraction cells filled
with approximately 2.0 g of diatomaceous earth (isolute-HM-N). The
azeotropic mixture of cyclohexane and ethyl acetate was used as
solvent (19, 20). After removal of the solvent, the lipid phase was
transesterified according to the official standard procedure (18). After
addition of 25 µL of ISTD 1 solution, a portion (15-20 mg) of the
isolated lipids was treated as described above.
The residue was dissolved in 1 mL of an n-hexane-ethyl acetate
mixture (98:2 v/v). This solution was transferred onto a silica gel
column (0.8 g of dried silica gel in a Pasteur pipet) equilibrated with
3 mL of n-hexane-ethyl acetate (98:2, v/v). The column was washed
with 10 mL of the same solvent mixture to eliminate the non-OH-
FAMEs. OH-FAMEs were eluted into a separate fraction by rinsing
the column with 6 mL of ethyl acetate. The residue obtained after the
removal of the solvent was subjected to the PFBO derivatization
procedure (see next section).
PFBO Derivatization. The PFBO derivatives were formed by adding
150 µL of 35% PFBO-Cl solution and 100 µL of 2% TEA solution
(both in acetonitrile) to the dried samples and then heating for 1 h at
100 °C (8). After the sample was cooled to room temperature, distilled
water (2 mL) and n-hexane (3.5 mL) were added, and the products
were separated with the organic phase. Prior to GC/ECNI-MS analyses,
50 µL of ISTD 2 (3- PFBO-O-12:0-EE) was added to 950 µL of the
organic phase including the OH-FAs as their ME/PFBO derivatives.
Gas Chromatography with Electron-Capture Negative Ion Mass
Spectrometry. GC/MS measurements were performed with a CP-3800
GC coupled to a 1200 triple-quadrupole mass spectrometer (Varian,
Darmstadt, Germany). Helium (purity 5.0) was used as carrier gas. The
injector and transfer-line temperatures were set at 250 and 280 °C,
respectively. A scan rate of 2 cycles/s was applied, and the filament
emission current was set at 50 µA. GC analyses were performed with
a Factor Four VF-5 ms column (30 m, 0.25 mm i.d., 0.25 µm df;
Varian). The oven temperature program started at 60 °C (hold time
1.5 min), which then was raised at 40 °C/min to 180 °C (hold time 2
min), at 2 °C/min to 230 °C (hold time 9 min), and finally, at 10 °C/
min to 300 °C (hold time 7.5 min). The total run time was 55 min.
Injections were performed in splitless mode and the injection volume
was 1 µL. A constant flow rate of 1 mL/min was used throughout
the measurements. A solvent delay of 5 min was applied. The ion source
temperature was kept at 150 °C. Methane (purity 5.0) was used as the
reagent gas at approximately 8.8 Torr. In the full-scan mode, m/z
50-600 was recorded throughout the run. In the SIM mode, PFBO-
O-FAMEs were determined as follows: m/z 368/369 for 2- and 3-PFBO-
O-8:0-ME was measured from 5 to 13 min; m/z 396/397 for 2- and
3-PFBO-O-10:0-ME was measured from 13 to 18 min; m/z 424/425
for 2- and 3-PFBO-O-12:0-ME was measured from 18 to 21.5 min;
m/z 438/439 for ISTD 2 (3-PFBO-O-12:0-EE) was measured from 21.5
to 25 min; m/z 452/453 for 2- and 3-PFBO-O-14:0-ME was measured
from 25 to 32 min; m/z 480/481 for 2- and 3-PFBO-O-16:0-ME was
measured from 32 to 41 min; m/z 508/509 for 2- and 3-PFBO-O-18:
0-ME and m/z 510/511 for ISTD 1 (2-PFBO-O-18:0-ME-9,10-d2) was
Synthesis of 9,10-Dideutero-2-hydroxyoctadecanoic Acid Methyl
Ester. 2-OH-18:1(9c)-ME (5 mg) was dissolved in sodium-dried and
distilled benzene with a 50-mL pear-shaped flask as reaction vessel.
To deoxygenate the reaction mixture, nitrogen was introduced and the
solution was stirred vigorously for 2 min. This process was repeated
three times with nitrogen and twice with deuterium. Then Wilkinson’s
catalyst (4 mg) was added, and the flask was flushed three times with
deuterium. The reaction mixture was stirred at room temperature for
8 h. Afterward, the solvent was evaporated under reduced pressure,
and diethyl ether (20 mL) was added to the residue. The colored ethereal
solution was subjected to adsorption chromatography (1 cm i.d. glass
column filled with 5 g of dried silica gel) for purification (17). Due to
the fact that the deuteration did not proceed completely, the saturated
2-OH-18:0-ME-9,10-d2 was separated from the monounsaturated 2-OH-
18:1(9c)-ME by means of discovery silver-ion SPE cartridges. For this
purpose, an aliquot [4 mL, containing both 2-OH-18:0-ME-9,10-d2 and
2-OH-18:1(9c)-ME] of the ethereal solution was evaporated under a
stream of nitrogen, redissolved in 1 mL of n-hexane, and applied onto
a silver-ion SPE cartridge previously conditioned with 4 mL of distilled
acetone and equilibrated with 4 mL of n-hexane. The saturated 2-OH-
18:0-ME-9,10-d2 was eluted with 10 mL of a n-hexane-acetone mixture
(9:1 v/v). The monoenoic 2-OH-18:1(9c)-ME was collected in a separate
fraction eluted with 10 mL of n-hexane-acetone (1:1 v/v). This
separation step was repeated five times. The combined fractions
containing about 1.15 mg of 2-OH-18:0-ME-9,10-d2 (yield ∼ 23%)
were evaporated to dryness, redissolved in 4 mL of n-hexane, and used
as recovery standard (internal standard 1, ISTD 1).

5502 J. Agric. Food Chem., Vol. 56, No. 14, 2008
Jenske and Vetter
of the 12C/13C signal did not differ more than 10% from the
theoretical isotope ratio.
Under the above-mentioned GC/ECNI-MS-SIM conditions,
the detection limit was 0.010-0.105 mg/100 g for 2-PFBO-O-
FAMEs and 0.003-0.026 mg/100 g for 3-PFBO-O-FAMEs
(depending on the chain length; for individual detection limits,
see Table 1). This method is not only very selective for the
determination of PFBO-O-FAMEs but also very sensitive and,
thus, was suitable for the determination of trace amounts of 2-
and 3-OH-FAs in foodstuffs.
Sample Preparation for Bovine Milk Fat Samples. Despite
the high sensitivity and selectivity of the determination method,
it should be noted that OH-FAMEs are only minor constituents
of the lipids of foodstuffs. After transfer of OH-FAs into the
corresponding PFBO-O-FAMEs, conventional FAMEs are also
found in the extract. Thus, the direct injection of small amounts
of PFBO-O-FAMEs would be accompanied with the coinjection
of high amounts of FAMEs onto the GC column. In order to
prevent an overload of the GC column, the bulk of non-OH-
FAMEs was separated from the OH-FAMEs prior to PFBO
derivatization. For this purpose, adsorption chromatography on
dried silica was applied. In order to avoid hazardous halogenated
solvents, a solvent mixture made of n-hexane and ethyl acetate
(98:2) for the elution of non-OH-FAMEs into a first fraction
(which was discarded) was chosen. After this step, OH-FAMEs
were eluted with 100% ethyl acetate. This eluent selection
occurred according to Youhnovski et al. (21), although they
used cyclohexane instead of n-hexane. It turned out that this
sample preparation step could be accomplished faster with
n-hexane.
Figure 1. Derivatization of methyl esters of 2-OH-FAs (x ) 0, y ) 5-17)
or 3-OH-FAs (x ) 1, y ) 4-14) with PFBO-Cl.
measured from 41 to 45 min; and finally, m/z 536/537 for 2- and
3-PFBO-O-20:0-ME was measured from 45 to 55 min.
RESULTS AND DISCUSSION
GC/ECNI-MS Capability of PFBO Derivatives. Because
2- and 3-OH-FAs are only minor constituents in foodstuffs, it
is imperative to utilize a very sensitive method for quantification.
Consequently, OH-FAs were converted into ME/PFBO deriva-
tives (Figure 1) and analyzed by GC/ECNI-MS-SIM (7). The
results of our own experiments concerning the influence of the
reaction temperature on the PFBO derivatization aligned with
previous reports (7, 8).
Under the present GC conditions, separation of the 2- and
3-PFBO-O-FAME isomers was achieved. In addition, it was
evident from the GC/ECNI-MS-SIM chromatograms that the
2- and 3-PFBO-O-FAMEs gave a different response compared
to ISTD 2 (Table 1). In general, 3-PFBO-O-FAMEs showed a
better response than the corresponding 2-PFBO-O-FAMEs.
Furthermore, the detector response of ME/PFBO derivatives
decreased with increasing chain length. This is explainable by
the decreasing ratio of the part with electron-capturing properties
to the rest of the molecule (i.e., the alkyl chain). However, due
to the varying response factors of individual OH-FAs, the
application of external standards was essential for quantification.
Sample preparation for the determination of OH-FAMEs in
bovine milk consisted of several steps. Thus, the loss of OH-
FAs during the sample preparation (accompanied with the
falsification of the results) cannot be excluded. Therefore, there
was a need for a recovery standard (ISTD 1) to be added to the
foodstuff samples before the transesterification step (see next
section). Another problem in GC/ECNI-MS is the instability
of the instrument from day to day, sometimes even from
injection to injection. Repeated analyses of samples by GC/
ECNI-MS measurements are subject to unavoidable variations
in the total area (uniform for all compounds). To even out
variations between standards and samples in subsequent GC
runs, an internal standard (ISTD 2) was required to calibrate
the GC/ECNI-MS performance.
Representative GC/ECNI-MS mass spectra of 2- and 3-PFBO-
O-FAMEs are shown in Figure 2. Both GC/ECNI-MS spectra
of the PFBO-O-18:0-ME isomers showed the base peak for the
molecular ion at m/z 508 and little fragmentation. In addition
to [M]-, only three relevant fragment ionssm/z 167 (C6F5),
m/z 196 (C6F5HCO), and m/z 211 (C6F5CO2)swere observed
in the GC/ECNI-MS spectrum of 2-PFBO-O-18:0-ME (Figure
2A). These fragment ions were much less abundant in the mass
spectrum of 3-PFBO-O-18:0-ME (Figure 2B). The lower
fragmentation of 3-PFBO-O-FAMEs is rooted in the higher
temperature stability of these compounds. Mielniczuk et al. (8)
have shown that the relative intensities of the ions in the GC/
ECNI-MS mass spectra of 2-PFBO-O-FAMEs were strongly
dependent on the temperature of the ion source. While they did
not observe any molecular ion at an ion source temperature of
150 °C (8), on our GC/MS system, [M]- was the base peak in
the ECNI mass spectra of all investigated 2- and 3-PFBO-O-
FAMEs. The stronger fragmentation of the 2-PFBO-O-FAMEs
compared to 3-PFBO-O-FAMEs was accompanied by worse
detection limits when molecular ions were used in the SIM mode
(Table 1). In addition, the 13C isotopic peak was used as
confirmatory ion for verification (Table 1). 2- and 3-PFBO-O-
FAMEs were positively identified when the experimental ratio
Selection of Internal Standards. The ideal standards ISTD
1 and ISTD 2 should not be present in foodstuff and they should
not coelute with any relevant fatty acid with the same SIM mass.
In addition, the recovery standard should run through all sample
preparation steps in an analogous manner as the OH-FAs
contained in foodstuffs.
In previous studies, odd-numbered OH-FAs (7–9) or a
perdeuterated nonhydroxy fatty acid (10) were used as internal
standards. However, odd-numbered OH-FAs did not come into
consideration, given that it could not be excluded that these
OH-FAs do occur in the food samples (12, 22, 23). Deuterated
non-hydroxy fatty acids could not be used either, since their
physicochemical properties are different from those of OH-FAs.
Hence, they do not run through all sample preparation steps in
the same way as OH-FAs. However, deuterium-labeled OH-
FAs appeared to be a suitable recovery standard (ISTD 1). Since
deuterated 2- and 3-OH-FAs are not commercially available,
2-OH-18:0-ME-9,10-d2 was synthesized from the unsaturated
2-OH-18:1(9c)-ME following the protocol for the deuteration
of unsaturated (nonhydroxy) methyl fatty esters (17). While the

Hydroxy Fatty Acids in Bovine Milk Fat
J. Agric. Food Chem., Vol. 56, No. 14, 2008 5503
Table 1. Gas Chromatographic Retention Times, Relative Response Factors, Quantification Ions, and Detection Limits and Amounts of 2- and
3-PFBO-O-FAMEs in Bovine Milk Fat
PFBO
retention
time, min
quantification
detection limits,b
mg/100 g
bovine milk fat (n ) 5),
mg/100 g
no.
derivative of
RRFa
(qualification) ion, m/z
1
2
3
4
5
6
7
8
2-OH-8:0-ME
3-OH-8:0-ME
10.03
10.21
14.29
14.47
20.05
20.23
22.11
26.67
26.85
33.76
33.96
42.76
42.79
42.92
46.31
1.109
3.780
0.886
2.944
0.477
2.242
1.000
0.356
1.630
0.224
0.828
0.061
0.093
0.202
0.092
368 (369)
368 (369)
396 (397)
396 (397)
424 (425)
424 (425)
438 (441)
452 (455)
452 (455)
480 (483)
480 (483)
510 (511)
508 (509)
508 (509)
536 (537)
0.010
0.003
0.014
0.004
0.019
0.005
4.49 ( 0.29
0.79 ( 0.09
0.05 ( 0.00
3.19 ( 0.25
0.02 ( 0.00
3.41 ( 0.12
(ISTD 2)
0.49 ( 0.02
2.47 ( 0.16
1.05 ( 0.01
4.20 ( 0.10
(ISTD 1)
ndc
nd
nd
2-OH-10:0-ME
3-OH-10:0-ME
2-OH-12:0-ME
3-OH-12:0-ME
3-OH-12:0-EE
2-OH-14:0-ME
3-OH-14:0-ME
2-OH-16:0-ME
3-OH-16:0-ME
2-OH-18:0-ME-9,10-d2
2-OH-18:0-ME
3-OH-18:0-ME
2-OH-20:0-ME
0.047
0.007
0.066
0.013
9
10
11
12
13
14
15
0.105
0.026
0.105
a Relative response factors of 2- and 3-PFBO-O-FAMEs related to 3-PFBO-O-12:0-EE (based on the peak areas of the GC/ECNI-MS-SIM chromatogram). b Detection
limit based on 20 mg sample purified and made up in 1 mL of solvent. The detection limit of standards ranged from 300 fg to 2 pg of 2-OH-PFBO-O-FAMEs and from 50
to 500 fg of 3-PFBO-O-FAMEs. c Not detectable.
Table 2. Variations in GC/ECNI-MS-SIM Areas of ISTD 1 and ISTD 2 and
Ratios of Both Compounds in Three Injections
area of ISTD 1
area of ISTD 2
ratio of
injection
[kcts] (m/z 510)
[kcts] (m/z 438)
ISTD1:ISTD 2
1
2
3
8830 (1.000)a
7746 (0.877)
7545 (0.854)
14.6%
495 100 (1.000)
433 800 (0.876)
423 600 (0.856)
14.4%
0.01783
0.01786
0.01782
0.2%
maximum deviationb
a Values in parenthesess are normalized to the value of injection 1, which was
set at 1.0. b Lowest area divided by the highest area (%).
Thus, the inclusion of ISTD 2 was essential for precise
determination of hydroxylated fatty acids as PFBO-O-FA methyl
esters by GC/ECNI-MS-SIM.
Application of the Method to Bovine Milk Fat. Ten OH-
FAs (even-numbered 2- and 3-OH-FAs with chain lengths from
8 to 16 carbons) were identified by means of the correct retention
time and the correct ratio of the 12C/13C signal by GC/ECNI-
MS-SIM (see Figure 3 and Table 1). These results expand the
range of compounds described in bovine milk by Parks (12).
This author identified C6-C16 3-OH-FAs after conversion into
the corresponding TMS/ME derivatives using GC/MS (12).
The amounts of 2- and 3-OH-FAs in the isolated bovine milk
fat, listed in Table 1, are in each case a mean value of five
determinations (n ) 5, five times the complete sample cleanup
of bovine milk fat). The recovery of ISTD 1 was 93.4% ( 6.5%.
The amounts of 2-OH-FAs in 100 g of bovine milk fat were
0.02-4.5 mg, and the amounts of 3-OH-FAs ranged from 0.79
to 4.2 mg. Except for the abundant 2-OH-8:0, the corresponding
3-OH-FAs were 4-fold to >100-fold more abundant than the
corresponding 2-OH-FAs. Even at very low concentrations of
<0.1 mg/100 g, as observed for 2-OH-10:0 and 2-OH-12:0, the
standard deviation proved to be excellent (Table 1). Thus, it is
demonstrated that our method is suitable for quantification of
traces of 2- and 3-OH-FAs in foodstuffs.
Figure 2. GC/ECNI-MS full-scan mass spectra of (A) 2-PFBO-O-18:0-
ME (13) and (B) 3-PFBO-O-18:0-ME (14).
purified 2-PFBO-O-18:0-ME-9,10-d2 (see Materials and Meth-
ods) was not fully GC-separated from the native 2-PFBO-O-
18:0-ME (compare Table 1), the molecular ions ([M]-) of both
compounds differ by 2 atom mass units. Consequently, 2-OH-
18:0-ME-9,10-d2 was suitable as recovery standard (ISTD 1)
for GC/ECNI-MS-SIM.
For the correction of GC/MS instabilities (see above), the
PFBO derivative of 3-OH-12:0 ethyl ester (3-PFBO-O-12:0-
EE) was selected. This compound was added as ISTD 2 to
sample solutions directly before the GC/MS measurements.
Compared to the methyl ester, the retention time of the ethyl
ether was shifted by ∼1.9 min to higher retention times (Table
1). The efficiency of ISTD 2 was tested in a mixture with ISTD
1 (Table 2). Three injections provided variations in the peak
areas of ISTD 2 by up to 15%. However, ISTD 1 showed exactly
the same behavior. This can be seen from the fact that the ratio
of the peak areas differed only in the fourth digit (Table 2).
In addition, other constituents of bovine milk fat ascertainable
with this method were detected (see Figure 3). Since these
additional compounds (marked A-E in Figure 3) showed the
correct ratio of quantification ion (12C signal) to verification
ion (13C signal) in GC/ECNI-MS-SIM, they must be isomers
of 2- and 3-OH-FAs. A multitude of γ- and δ-lactones and their
precursors (4- and 5-hydroxy fatty acids bound to glycerides
or phospholipids) from C8 to C16 exist in bovine milk

5504 J. Agric. Food Chem., Vol. 56, No. 14, 2008
Jenske and Vetter
Supporting Information Available: GC-ECNI-MS-SIM chro-
matogram of ME/PFBO derivatives of lactones. This material
is available free of charge via the Internet at http://pubs.acs.org.
LITERATURE CITED
(1) Bohn, M.; Heinz, E.; Lu¨thje, S. Lipid composition and fluidity of
plasma membranes isolated from corn (Zea mays L.) roots. Arch.
Biochem. Biophys. 2001, 387, 35–40.
(2) Smith, C. R.; Wolff, I. A. Characterization of naturally occurring
R-hydroxylinolenic acid. Lipids 1969, 4, 9–14.
(3) Hirono, H. Lipids of myelin, white matter and gray matter in a
case of generalized deficiency of cytochrome b5 reductase in
congenital methemoglobinemia with mental retardation. Lipids
1979, 15, 272–275.
(4) Niederer, M.; Bollhalder, R. Identification of species specific
central nervous tissue by gas chromatography-mass spectrometry
(GC-MS)-a possible method for supervision of meat products
and cosmetics. Mitt. Lebensmittelunters. Hyg. 2001, 92, 133–144.
(5) Biedermann, W.; Lucker, E.; Hensel, A. Detection of tissues of
the central nervous system (CNS) as specified risk material (SRM)
in meat products by means of gas chromatography-mass spec-
trometry (GC-MS). Berl. Muench. Tieraerztl. Wochenschr. 2002,
115, 131–133.
Figure 3. GC/ECNI-MS-SIM chromatogram (Factor Four VF-5 ms column)
of bovine milk fat after the sample preparation procedure described above.
Compound index: 3, 2-PFBO-O-10:0-ME; 4, 3-PFBO-O-10:0-ME; 5,
2-PFBO-O-12:0-ME; 6, 3-PFBO-O-12:0-ME; 8, 2-PFBO-O-14:0-ME; 9,
3-PFBO-O-14:0-ME; 10, 2-PFBO-O-16:0-ME; 11, 3-PFBO-O-16:0-ME. For
amounts of 2- and 3-OH-FAs, see Table 1. An enlarged excerpt of the
ME/PFBO derivatives of the 12C- (higher abundance peaks) and 13C-
(lower abundance peaks) ion traces of 2- (left peak) and 3-OH-14:0 is
shown as an inset. Isomers of 2- and 3-OH-FAs are marked as follows:
A, PFBO-O-10:0-ME isomers; B, PFBO-O-12:0-ME isomers; C, PFBO-
O-14:0-ME isomers; D, PFBO-O-16:0-ME isomers; E, PFBO-O-18:0-ME
isomers.
(6) Szponar, B.; Norin, E.; Midtvedt, T.; Larsson, L. Limitations in
the use of 3-hydroxy fatty acid analysis to determine endotoxin
in mammalian samples. J. Microbiol. Methods 2002, 50, 283–
289.
(7) Mielniczuk, Z.; Alugupalli, S.; Mielniczuk, E.; Larsson, L. Gas
chromatography-mass spectrometry of lipopolysaccharide 3-hy-
droxy fatty acids: comparison of pentafluorobenzoyl and trim-
ethylsilyl methyl ester derivatives. J. Chromatogr. 1992, 623, 115–
122.
(8) Mielniczuk, Z.; Mielniczuk, E.; Larsson, L. Gas chromatography-
mass spectrometry methods for analysis of 2- and 3-hydroxylated
fatty acids: application for endotoxin measurements. J. Microbiol.
Methods 1993, 17, 91–102.
fat (24, 26). Note that the lactones (inner esters of 4- and 5-OH-
FAs) are converted into 4- and 5-OH-FAMEs in the course of
the transesterification and subsequently derivatized into the
corresponding PFBO-O-FAMEs by the procedure used in our
study. This was tested with γ-decalactone (inner ester of 4-OH-
10:0) and γ-dodecalactone (inner ester of 4-OH-12:0) (see
Figure S1, Supporting Information). Consequently, our method
is not suited to distinguish lactones from their precursors since
both compound classes will be converted into the same
derivative. In either case, it was found that a peak at the retention
time of 4-PFBO-O-10:0-ME was detected in the bovine milk
sample (14.97 min), but the retention time of 4-PFBO-O-12:
0-ME (20.80 min) was shorter than the peak detected in the
sample (21.01 min). Both results are in agreement with data of
Wyatt et al. (25), who identified 4-OH-10:0 but not 4-OH-12:0
in bovine milk. Since δ-lactones and their precursors, the 5-OH-
fatty acids, are more common in bovine milk, some of the
isomeric peaks detected in this study may originate from
5-PFBO-O-FAMEs. Given the fact that additional standards of
γ- and δ-lactones or their precursors were not available to us,
additional unknown peaks in the chromatograms could not be
identified. Future investigations will have to be carried out to
give exact evidence concerning this matter. It appears that the
present GC/ECNI-MS-SIM method will be suited for exploring
this topic.
(9) Sonesson, A.; Larsson, L.; Schu¨tz, A.; Hagmar, L.; Hallberg, T.
Comparison of the Limulus amebocyte lysate test and gas
chromatography-mass spectrometry for measuring lipopolysac-
charides (endotoxins) in airborne dust from poultry-processing
industries. Appl. EnViron. Microbiol. 1990, 56, 1271–1278.
(10) Lee, A. K. Y.; Chan, C. K.; Fang, M.; Lau, A. P. S. The 3-hydroxy
fatty acids as biomarkers for quantification and characterization
of endotoxins and gram-negative bacteria in atmospheric aerosols
in Hong Kong. Atmos. EnViron. 2004, 38, 6307–6317.
(11) Keina¨nen, M. M.; Korhonen, L. K.; Martikainen, P. J.; Vartiainen,
T.; Miettinen, I. T.; Lehtola, M. J.; Nenonen, K.; Pajunen, H.;
Kontro, M. H. Gas chromatographic-mass spectrometric detection
of 2- and 3-hydroxy fatty acids as methyl esters from soil,
sediment and biofilm. J. Chromatogr. B 2003, 783, 443–451.
(12) Parks, O. W. Isolation and characterization of nonesterified
3-hydroxy acids in milk. J. Diary Sci. 1977, 60, 718–720.
(13) Chance, D. L.; Gerhardt, K. O.; Mawhinney, T. P. Gas-liquid
chromatography-mass spectrometry of hydroxyl fatty acids as their
methyl ester tert-butyldimethylsilyl ethers. J. Chromatogr. A 1998,
793, 91–98.
The present method already proved to be suitable for
quantitative determination of 2- and 3-OH-FAs in complex food
matrices like bovine milk. The preseparation of non-OH-FAMEs
followed by the conversion of the OH-FAMEs into the
corresponding PFBO derivatives allowed GC/ECNI-MS-SIM
determination of 2- and 3-OH-FAs with high sensitivity and
selectivity. Application of a recovery standard and a measuring
standard enabled a determination with high precision and good
quality assurance, as can be seen from the high recovery rate
of the internal standard. Thus, significant losses of 2- and 3-OH-
FAs during sample preparation could be excluded. Since bovine
milk is rather a complex foodstuff sample, it appears to be
warranted that this novel method can be applied as well to other
matrices for establishing the content of 2- and 3-OH-FAs in
foodstuffs.
¨
(14) Ozen, H. C.; Bashan, M.; Keskin, C.; Toker, Z. Fatty acid and
3-hydroxy fatty acid composition of two Hypericum species from
Turkey. Eur. J. Lipid Sci. Technol. 2004, 106, 68–70.
(15) Ferrando, R.; Szponar, B.; Sanchez, A.; Larsson, L.; Valero-
Guillen, P. L. 3-Hydroxy fatty acids in saliva as diagnostic markers
in chronic periodontitis. J. Microbiol. Methods 2005, 62, 285–
291.
(16) Wakeham, S. G.; Pease, T. K.; Benner, R. Hydroxy fatty acids in
marine dissolved organic matter as indicators of bacterial mem-
brane material. Org. Geochem. 2003, 34, 857–868.
(17) Lie Ken Jie, M. C. F.; Choi, Y. C. Mass spectral studies of
deuterium-labelled picolinyl fatty esters in the determination of
double-bond position. J. Am. Oil Chem. Soc. 1992, 69, 1245–
1247.

Hydroxy Fatty Acids in Bovine Milk Fat
J. Agric. Food Chem., Vol. 56, No. 14, 2008 5505
(18) Standard Method of the German Society for Fat Science (DGF)
1998;C-VI 11d.
(23) Jenske, R. Vetter, W. Highly selective and sensitive gas chroma-
tography-electron-capture negative-ion mass spectrometry method
for the indirect enantioselective identification of 2- and 3-hydroxy
fatty acids in food and biological samples. J. Chromatogr. A 2007,
1146, 225–231.
(24) Wyatt, C. J.; Pereira, R. L.; Day, E. A. The lactone precursor in
fresh milk fat: identification of the hydroxy fatty acids. Lipids
1966, 2, 208–211.
(25) Wyatt, C. J.; Pereira, R. L.; Day, E. A. Lactone precursor in fresh
milk fat: Isolation and characterization of the precursor. J. Dairy
Sci. 1967, 50, 1760–1763.
(26) Dimick, D. S.; Patton, S.; Kinsella, J. E.; Walker, N. J. The
prevalence of aliphatic delta-lactones or their precursors in animal
fats. Lipids 1966, 1, 387–390.
(19) Thurnhofer, S.; Vetter, W. A gas chromatography/electron ioniza-
tion-mass spectrometry-selected ion monitoring method for de-
termining the fatty acid pattern in food after formation of fatty
acid methyl ester. J. Agric. Food Chem. 2005, 53, 8896–8903.
(20) Weichbrodt, M.; Vetter, W.; Luckas, B. Microwave-assisted
extraction and accelerated solvent extraction with the solvent
mixture ethyl acetate/cyclohexane (1:1, v:v) in view of quantitative
determination of organochlorines in fish tissue. J. AOAC Int. 2000,
83, 1334–1343.
(21) Youhnovski, N.; Schulz, D.; Schwarz, C.; Spiteller, G.; Schubert,
K. Determination of hydroxyoctadecadienoic acids. Z. Natur-
forsch. 2002, 58c, 268–276.
(22) Moldovan, Z.; Jover, E.; Bayona, J. M. Gas chromatographic and
mass spectrometric methods for the characterisation of long-chain
fatty acids-application to wool wax extracts. Anal. Chim. Acta
2002, 465, 359–378.
Received for review March 2, 2008. Revised manuscript received May
8, 2008. Accepted May 10, 2008.
JF800647W