Sample Preparation for Determination of Macrocyclic Lactone

LAGANÀ ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 86, NO. 4, 2003 729
FOOD CHEMICAL CONTAMINANTS
Sample Preparation for Determination of Macrocyclic Lactone
Mycotoxins in Fish Tissue, Based on On-Line Matrix Solid-Phase
Dispersion and Solid-Phase Extraction Cleanup Followed by
Liquid Chromatography/Tandem Mass Spectrometry
ALDO LAGANÀ, ALESSANDRO BACALONI, MARYANNA CASTELLANO, ROBERTA CURINI, ILARIA DE LEVA, ANGELO FABERI, and
STEFANO MATERAZZI
La Sapienza University, Department of Chemistry, Piazzale Aldo Moro 5, 00185 Rome, Italy
A new method based on matrix solid-phase dispersion (MSPD) on-line with a solid-phase extraction
(SPE) cleanup process followed by liquid chromatography with tandem mass spectrometry
(LC/MS/MS) is presented for the determination of
3 macrocyclic lactone mycotoxins in fish tissues:
zearalenone, a-zearalenol, and b-zearalenol. The
sample was prepared in a device that used a reversed-phase material (C18) or a normal-phase material (neutral alumina) as a matrix dispersing
agent, and a graphitized carbon black cartridge
was used for sequential cleanup by SPE.
LC/MS/MS was used for selective determination.
Isocratic elution with acetonitrile–methanol–water
was used for LC separation; for MS/MS, 2 types of
interfaces (a pneumatically assisted electrospray
ionization interface or an atmospheric pressure
chemical ionization interface) were evaluated and
compared in terms of the intensity of the total ion
current produced by each analyte. The use of
highly selective MSPD on-line with SPE for sample
preparation before analysis allowed the removal of
interfering matrix compounds `present in tissue
extracts that would otherwise cause severe ionization suppression of zearalenone and its metabolites during the ionization process. Average recoveries at 100 ng/g were between 83 and 103% with
C18 and ³67% with neutral alumina; the relative
standard deviations were <11% with C18 and <18%
with alumina. The limits of detection ranged from
0.1 to 1.0 ng/g. Sample preparation is simple to
perform, no special technical equipment is required, and solvent volumes are minimal.
he infection of food with several species of fungi has
been recognized as a potential threat to animal and human health. Mycotoxins are secondary metabolites of
T
Received October 21, 2002. Accepted by AP February 20, 2003.
Corresponding author’s e-mail: [email protected].
filamentous fungi or, more specifically, molds. Their presence
in food is a real problem. Over 300 mycotoxins have been isolated and chemically characterized (1). It can be assumed that
about 5–10% of the total cereal, fruit, and vegetable harvest is
destroyed by mold. Therefore, a significant degree of contamination is due to fungal metabolites, among which are many
mycotoxins that can be detected in food products and animal
feed. Zearalenone (ZON) is a nonsteroidal estrogenic
macrocyclic lactone mycotoxin produced by different strains
of fungi (Fusarium spp.), which often affect cereal crops in
temperate climate zones. This substance is known because of
its estrogenic activity: it binds to estrogen receptors influencing estrogen-dependent transcriptions in the nucleus and
thereby causes dysfunction to human and animal reproductive
systems (2, 3). Recent studies have demonstrated the potential
for ZON to stimulate the growth of human breast cancer cells
containing estrogen-response receptors (4, 5). It has been observed that ZON at approximately 500 mg/kg is a level at
which toxic effects can be found in livestock, and that
50 mg/kg can cause visible effects in the reproductive organs
of mammals (6).
Metabolism investigation has shown that ZON is transformed into 2 biologically active metabolites, a-zearalenol
(a-ZOL) and b-zearalenol (b-ZOL; Figure 1), and that the ratio of the metabolites varies among different species (7). Even
the latter compounds are estrogenic, behaving like endocrine
disrupters (8), and therefore must be included in an appropriate monitoring program.
Animal exposure to ZON occurs mainly by ingestion of
contaminated feed, whereas humans can also be damaged by
eating contaminated meat (9).
Rainbow trout (Oncorhynchus mykiss) is one of the most
widely diffused aquaculture species worldwide (in Italy there
are >1000 fish farms). Trout feed is composed of animal protein, fats, oils, semioleo byproducts, and cereals in grains (10);
because of the last component, trout can be particularly subject to mycotoxin exposure through ingestion of contaminated
feed (11). Therefore, it is important to have an analytical
method for monitoring mycotoxins in animals.
Several analytical techniques, such as thin-layer chromatography, enzyme-linked immunosorbent assay, gas chroma-
730
LAGANÀ ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 86, NO. 4, 2003
Figure 1. Chemical structures of selected analytes.
is more efficient, because it combines disruption of the gross
architecture of the sample with the dissolution and dispersion
of sample components into the bound organic phase on the
surface of particles. Thus, MSPD extraction is a promising approach for extraction of mycotoxins from solid or semisolid
samples.
In this paper, we report the development of an on-line
MSPD–SPE cleanup followed by LC/MS/MS for the determination of ZON and its metabolites in fish tissues. A nonpolar
(C18) or a polar (aluminum oxide activity I) dispersing agent
was used for sample dispersion. The elution profile and further cleanup using graphitized carbon black (GCB) were optimized. The analytes in the extracts were selectively determined by multiple reaction monitoring (MRM) LC/MS/MS to
allow detection at the ng/g level.
Experimental
tography, liquid chromatography (LC), and LC/mass spectrometry (LC/MS), have been used for the determination of
ZON, a-ZOL, and $-ZOL in different matrixes such as various foods and feeds, animal tissues, blood, and urine (12–17).
A common belief is that, because LC with tandem mass
spectrometry (LC/MS/MS) is very selective and specific,
sample preparation may be reduced to a minimum or even
omitted. On the contrary, some researchers have recently
demonstrated without any doubt that chromatographic
coelution of matrix compounds can severely affect the ion formation process in both electrospray and atmospheric pressure
chemical ionization (APCI) interfaces, resulting in a decrease
in the accuracy and reproducibility of LC/MS/MS analyses (18, 19). The use of an internal standard can reduce this
phenomenon, but cannot prevent it completely. Therefore, an
efficient sample preparation and cleanup process is needed to
minimize it. With complex matrixes, such as meat and vegetable tissues, this task is often long, tedious, and difficult: it consists not simply in extraction of the analytes from the solid matrix, but also in their separation from many interfering
substances of biological origin; the cellular structure of the
sample needs to be disrupted, and there is a high abundance of
proteins and lipids.
Current methodologies for preparation of samples containing macrocyclic lactone mycotoxins are preferably based on
solid-phase extraction (SPE). Zöllner et al. (20–22) have recently developed LC/MS/MS methods to determine ZON and
its metabolites in various matrixes: beer, urine, grain, and
meat. These methods involve an extraction step using C18 SPE
columns for liquid samples; homogenization with various organic solvents is used for solid samples, followed by an SPE
cleanup with a C18 or immunoaffinity column. The use of SPE
in analytical protocols has some drawbacks such as the presence of particulates that impede and block the flow because
they occupy the spaces in the solid-phase support materials.
Centrifugation or filtration is used to remove the particulates,
but it can potentially alter the results and lead to variability.
Recently, matrix solid-phase dispersion (MSPD) extraction has been applied to a variety of matrixes for the extraction
of analytes with a broad range of polarity. The MSPD process
Chemicals and Reagents
(a) 2,4-Dihydroxy-6-(10-hydroxy-6-oxo-trans-1-undecenyl)
benzoic acid m-lactone (ZON), 2,4-dihydroxy-6-(6a,10dihydroxy-trans-1-undecenyl)benzoic acid m-lactone (a-ZOL),
2,4-dihydroxy-6-(6b,10-dihydroxy-trans-1-undecenyl)benzoic
acid m-lactone (b-ZOL), and 2,4-dihydroxy-6-(10-hydroxy6-oxo-undecenyl)benzoic acid m-lactone (ZAN).—Sigma
(Milwaukee, WI).
(b) C18 (Lichroprep® RP 18, 25–40 mm) and neutral alumina (aluminum oxide activity I).—Merck (Darmstadt, Germany); both were used as supplied, without any preliminary
treatment.
(c) Carbograph-4.—A particular type of GCB; supplied
by L.A.R.A. srl (Rome, Italy). The particle size range was
120–400 mm. No particular precautions were taken in storing
the GCB. The carbograph cartridge was prepared by filling a
polypropylene tube, 6 ´ 1.3 cm (Supelco, Bellefonte, PA),
with 0.250 g adsorbent material. A polyethylene fritted disk
and ca 1 cm pressed quartz wool were placed below the
sorbent bed, and another polyethylene fritted disk was placed
above the sorbent bed.
(d) Polytetrafluoroethylene (PTFE) filter.—0.2 mm,
25 mm diameter (Alltech, Milan, Italy).
(e) Acetonitrile and methanol, both LC grade.—Carlo-Erba
(Milan, Italy).
(f) Hexane and dichloromethane.—Carlo-Erba.
(g) Deionized water.—Purified in a Milli-Q RG system
(Millipore, Bedford, MA).
Each individual compound was dissolved in an appropriate
volume of methanol to obtain a 1 mg/mL stock solution. An
aliquot of each stock solution was mixed and diluted with
methanol to obtain a primary working solution containing
each analyte at 100 ng/mL. Working solutions at the appropriate concentrations were prepared daily by dilution of the primary solution with methanol. All stock solutions and the primary working solution were stored at –18°C and brought to
room temperature before use.
LAGANÀ ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 86, NO. 4, 2003 731
Instrumentation
A Series 200 Perkin-Elmer pump equipped with a
Rheodyne Model 7125 injector with a 50 mL loop was used for
LC. Chromatographic separation was achieved under
isocratic conditions on an Alltima Prevail C18 column, 250 ´
4.6 mm id, average particle size 5 mm (Alltech, Deerfield, IL)
with a Supelguard precolumn, 20 ´ 4.6 mm id (Supelco). The
mobile phase was acetonitrile–methanol–water (37 + 16 +
47, v/v/v). The flow rate was set at 1 mL/min, but only 1/5
(200 mL/min) of the total column effluent was diverted to the
mass spectrometer.
A PE Sciex (Concord, ON, Canada) API 365 triple-quadrupole mass spectrometer, equipped with a
TurboIonSpray™ (TISP) source and operated in the negative-ion mode, was used for MS/MS. MassChrom 1.1.1 software on a Power Macintosh G3 was used for data processing.
All ion source and MS/MS instrumental parameters were
optimized for high sensitivity by infusing a standard solution
of each analyte (1 ng/mL) at a flow rate of 10 mL/min by using
an infusion pump (Harvard Apparatus, South Natick, MA).
The Q1 full scans of the analytes were conducted from 80 to
350 m/z, in 0.5 m/z increments with a dwell time of 10 ms; the
MS/MS product ion scans of the precursor ion [M–H]– were
conducted from 50 to 400 m/z, with the same increments and
dwell time. The temperature of the turbo gas (nominal heating-gun temperature) was set at 350°C. High-purity nitrogen
was used as the nebulizer, curtain, and collision gas with respective (arbitrary) settings of 10, 10, and 2 as characteristic
values for the Sciex API 365 instrument.
The MS/MS analysis was performed in the MRM mode.
The deprotonated molecular species of ZON (m/z 317) and of
a-ZOL, b-ZOL, and ZAN (m/z 319) were selected as the precursor ions, and the following fragment ions were selected:
m/z 175, 160, and 131 for ZON; m/z 174, 160, and 130 for
a-ZOL; m/z 160, 144, and 130 for b-ZOL; and m/z 205 and
161 for ZAN.
The dwell time for each monitored transition was 600 ms
except that for the internal standard, which was set at 200 ms.
For each analyte, the best declustering potential (OR) was
chosen as follows: –30 V for ZON, –28 V for a-ZOL, –29 V
for b-ZOL, and –51 V for ZAN. The collision energy was adjusted by varying the voltage between the high-pressure entrance quadrupole, Q0, and the collision cell quadrupole, R02.
For each individual monitored transition, the value giving the
most intense signal was selected (for a-ZOL, 32.0 eV for
m/z 174, 38.0 eV for m/z 160, and 46.0 eV for m/z 130; for
b-ZOL, 40.0 eV for m/z 160, 38.0 eV for m/z 144, and 44.5 eV
for m/z 130; for ZON, 31.5 eV for m/z 175, 37.0 eV for
m/z 160, and 41.0 eV for m/z 131; for ZAN, 38.0 eV for both
m/z 205 and 161).
Data acquisition was divided into 3 time periods in order to
use longer dwell times for each MRM transition. Unit mass
resolution was used for both Q1 and Q3 mass-resolving
quadrupoles (full peak width at half height was ca 0.7 Da).
Sample Preparation
Portions of the edible part or liver of a trout, purchased in a
local supermarket, were chopped and kept refrigerated at
–18°C. A 0.5 g portion of tissue was placed in a glass mortar,
homogenized with a pestle, and then spiked with 50 or 5 ng of
each target analyte (a-ZOL, b-ZOL, and ZON). These compounds were contained in 1 mL acetone. The sample was then
allowed to air-dry at room temperature to eliminate all the organic solvent. A 2 g portion of C18 was then added to the sample, and the contents of the mortar were vigorously mixed with
the pestle to obtain homogeneity. The resulting powder was
dried overnight in a ventilated oven set at 40°C. Finally, the material was packed into a 6 mL polypropylene cartridge tube;
polyethylene fritted disks were placed above and below the
packing material. Meanwhile, the carbograph-4 cartridge was
prepared. The analytes were extracted by placing the MSPD
packed column, stacked on-line with the cartridge containing
the GCB, onto a vacuum extractor as shown in Figure 2.
Before the extraction, the GCB cartridge was sequentially
washed with 10 mL dichloromethane–methanol (80 +
20, v/v), 5 mL methanol, 20 mL water acidified with hydrochloric acid (10mM), 10 mL water and, finally, 10 mL methanol–water (70 + 30, v/v). When the matrix dispersion was performed with aluminum oxide instead of C18, the last solution
was replaced with 10 mL methanol–water (50 + 50, v/v).
The analytes were extracted and purified sequentially in a
few operations. The analytes were first eluted from the upper
MSPD cartridge with 15 mL methanol–water (70 + 30, v/v).
The vacuum was regulated to obtain an average flow of ca
1 mL/min. After the solvent passed through the upper cartridge, that cartridge was removed; in this step, the eluate from
the GCB cartridge was discarded. The cartridge containing
GCB was sequentially washed with 10 mL water, 10 mL
methanol acidified with 10mM formic acid, and 3 mL methanol. Finally, a 0.22 mm PTFE filter was inserted at the bottom
of the cartridge, and the analytes were eluted with 15 mL dichloromethane–methanol (80 + 20, v/v). The eluate was collected in a round-bottom glass vial. The eluate was evaporated
to dryness under a gentle stream of nitrogen. The residue was
reconstituted with 250 mL water–acetonitrile–methanol (40 +
42 + 18, v/v/v) containing the internal standard (ZAN) at
0.5 ng/mL. ZAN was chosen as the internal standard because it
does not occur in nature. For the LC/TISP–MS/MS analysis,
50 mL final extract was injected into the LC/MS/MS system.
With aluminum oxide instead of C18, the procedure was
similar to that described above: 0.5 g sample, spiked as described above, was placed in a glass mortar and homogenized
with 2 g aluminum oxide activity I to obtain a yellowish (when
muscular tissues are analyzed) or brownish powder (when
liver tissues are analyzed). The contents of the mortar were
dried overnight in a ventilated oven set at 40°C. Then the material was packed into a 6 mL polypropylene cartridge tube in
the same manner as described above. Subsequently, the
packed column was sequentially washed with 5 mL hexane
and 5 mL methanol.
732
LAGANÀ ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 86, NO. 4, 2003
The extraction was performed on-line with GBC by using
30 mL methanol–water (50 + 50, v/v). The flow was adjusted,
by regulating the vacuum, to ca 0.5–1 mL/min. After all of the
solvent passed through the upper cartridge, that cartridge and
the eluting solution were discarded. The subsequent operation
was performed in the same manner as described for the extraction using the C18 procedure.
To assess accuracy and precision, with either the C18 or the
alumina procedure, the following experimental design was
adopted: 12 different samples of muscular tissue and
12 different samples of liver were spiked at 100 ng/g as described above; the first 6 samples of each type were processed
with the C18 protocol, and the remaining 6 were processed
with the alumina protocol. The final extracts were then analyzed by LC/MS/MS. Recoveries were assessed by comparing
ratios of the analyte peak area to the internal standard peak
area with those obtained by injecting a standard solution at the
same concentration level. This protocol was then repeated for
samples spiked at 10 ng/g.
Results and Discussion
LC/MS/MS
We used the chromatographic separation developed previously (23) involving a ternary phase (acetonitrile–methanol–water) that produced a good chromatographic separation
in a short time.
The relatively clean extracts coming from the MSPD and
GCB cartridges allowed the use of a shorter chromatographic
run time, because interferences arising from biological
sources were drastically reduced and thus ion suppression
phenomena were minimized.
In a confirmation of the findings of other researchers (24),
the negative ionization mode was found to be more sensitive
for the selected analytes, compared with the positive mode.
The deprotonated molecule [M–H]– was chosen as the precursor ion, and product ion (MS/MS) spectra were generated with
precursor ions at m/z 319 for a-ZOL and b-ZOL and m/z 317
for ZON. The product ions selected for MRM transition were
the most abundant in the fragmentation spectra obtained under
the optimized experimental conditions.
Because of the structural similarity of the analytes, they exhibit similar fragmentation patterns and, therefore, mutual interference in the total ion chromatograms. Thus, chromatographic separation before mass spectrometric detection
cannot be omitted.
To optimize MS/MS conditions and the sensitivity for individual analytes, MRM analysis was performed by using a different setting of the ion optics and MS/MS tuning conditions
to achieve the lowest level of quantification for the
macrocyclic lactones studied. Moreover, to achieve better
sensitivity, the chromatographic run was divided into
3 acquisition time periods, permitting the use of a longer dwell
time (600 ms) for each monitored transition. Because the concentration of the internal standard was constant, its dwell time
was not increased.
For this study, ZAN was used as the internal standard, because the similarity of its structure to those of the target
analytes allowed efficient compensation for fluctuations in the
detector signal. The use of a deuterated internal standard
should be superior, but isotopically labeled compounds may
not be readily available because of difficult synthesis and/or
cost.
When real samples are analyzed by LC/MS, interferences
from the matrix can affect the intensity of the signal. For this
reason, the sensitivity of the LC/MS/MS system was investigated by comparing the peak areas obtained for a solution prepared as follows: a suitable known volume of working standard solution was dissolved in the eluant used for the final
elution from the carbograph-4 cartridge, and the solution was
evaporated to dryness as described in the Experimental section. This task was performed with the C18-based procedure
only. The effect of additives on the ionization efficiency was
evaluated by using both TISP and APCI–heated nebulizer
(APCI–HN) interfaces. The results of these studies are summarized in Table 1, which shows that the TISP interface without any addition to the mobile phases was the most effective,
producing the most intense signal, although the addition of
ammonium acetate to the mobile phases enhanced the signal
of the APCI interface and made it similar to that produced
with the TISP interface. However, the use of the latter was
preferred because minor amounts of impurities were sprayed
Figure 2. Schematic representation of the on-line
MSPD sample preparation procedure.
LAGANÀ ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 86, NO. 4, 2003 733
Table 1. Effect of carrier modifiersa on the relative intensities of the extracted ion current of MRM acquisition
Signal, %b (RSD, %)c
TISP
APCI
Analyte
I
II
III
I
II
III
IV
a-ZOL
100 (3.1)
29 (6.5)
76 (3.9)
49 (2.6)
60 (2.8)
39 (7.0)
85 (3.2)
b-ZOL
100 (4.0)
30 (7.4)
80 (4.7)
60 (4.0)
78 (3.9)
45 (5.9)
104 (3.3)
ZAN
100 (2.9)
44 (5.9)
69 (3.3)
62 (3.7)
67 (5.0)
54 (9.1)
96 (4.0)
ZON
100 (4.3)
29 (7.0)
75 (5.5)
59 (4.0)
57 (5.6)
48 (8.3)
80 (3.6)
a
b
c
I = Acetonitrile–methanol–water (37 + 16 + 47, v/v/v); II = acetonitrile–methanol–water (37 + 16 + 47, v/v/v) with 2mM ammonium acetate;
III = acetonitrile–methanol–water (37 + 16 + 47, v/v/v) with postcolumn methanol and 40mM ammonium hydroxide at 0.11 mL/min;
IV = acetonitrile–methanol–water (37 + 16 + 47, v/v/v) with 15mM ammonium acetate.
The signal intensities obtained with the TISP interface by using neutral mobile phases were arbitrarily set to 100. Injection: 10 mL of a 1 ng/mL
solution. Chromatographic conditions: isocratic elution at 1 mL/min.
n = 5.
onto the orifice plate. The addition of ammonium hydroxide
postcolumn had negative effects on the signal and therefore
should be avoided.
Figure 3 shows the chromatograms obtained, respectively,
from 0.5 g muscle tissue and liver tissue extracted by using
C18 or aluminum oxide; no interfering peaks are present in the
background, or in chromatograms obtained from analyses of
blank samples.
Linear Dynamic Range
This set of measurements was made by injecting different
known amounts of standard solution into the liquid
chromatograph. For each amount injected, measurements
were made in triplicate. The average peak areas were plotted
versus the concentrations injected. The resulting plots indicated that the linear response of the detector was in the range
of 1–100 ng injected. These data apparently suggest that quantitative analysis could be performed without the use of an internal standard; however, we recommend the use of an internal standard to achieve better precision, because signal
intensities may vary as a result of matrix interference in the
analysis of real samples.
Sample Preparation
We investigated the feasibility of using 2 different sample
dispersing agents for the determination of ZON and its
metabolites in fish liver and muscle tissues.
The application of MSPD methodology to the isolation of a
specific compound or class of compounds has grown tremendously in recent years.
As far as the analysis of animal tissues is concerned, this
technique is simpler than previous techniques based mainly
on SPE. In fact, the other techniques require sample homogenization, centrifugation, and removal of tissue debris before
column separation. The addition of homogenates directly to
the top of a column invariably leads to cessation of the flow
because of the clogging of the fritted disk or of the upper layers of the column packing.
Through the mixing of homogenate with a sample dispersing agent, the analyst eliminates the need to precipitate cellular components and to centrifuge the sample to separate the
debris. Moreover, the surface area of the sample exposed to
the solvent is increased so that it is entirely exposed to the solvent. Proceeding in this way, we optimized 2 extraction procedures by using both the reversed- and the normal-phase
sorbing materials.
Performing the extraction and the cleanup step “on-line”
has 2 remarkable advantages: first, the entire procedure is simple (it combines extraction and purification in a few operations and does not use highly technical equipment); second,
possible bias in analyte concentrations that may involve risks
of loss and contamination due to handling is avoided (the extracts coming from the upper cartridge fall directly into the
second cartridge).
Dispersing Phase C18
Some researchers emphasize the need to prewash the phase
before use. We omitted this operation because we noticed no
substantial differences or interference when we used washed
material. Besides, preconditioning the phase is not needed because fish tissue dissolves in the C18 powder simply by pestle
homogenization. In the case of particularly dry samples, a few
drops of eluting solution can be added to help dissolution
without affecting the results. Methanol–water (70 + 30, v/v)
was found to be an effective extraction solution for the target
compounds. The quantity of the extracting solution was optimized by collecting fractions (5 mL each) of spiked blank
(without tissue) samples and by analyzing each one separately
until all of the analytes were recovered completely. The optimal quantity for the extraction was found to be 15 mL.
The use of a less-polar extracting medium is theoretically
more effective and allows the use of small quantities of solvents, but it consequently has the side effect of coeluting too
many lipid and protein substances from the matrix, as reported
by Barker et al. (25), who performed C18 MSPD analysis of
animal tissues with 100% methanol as the extracting solvent.
734
LAGANÀ ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 86, NO. 4, 2003
upper cartridge is subjected to a further cleanup in the
carbograph-4 cartridge.
Carbograph-4 (a particular GCB) is stronger than other
nonpolar sorbing materials. It is able to retain molecules having a broad range of polarities because of its behavior as both a
nonspecific (i.e., Van der Waals interaction) and an anion-exchange (electrostatic) sorbent (27). To achieve
desorption, it is necessary to use mixtures of high eluant
power (28). This allows the cartridge to be washed, before elution, with an acidified solvent to displace compounds retained
in anionic form (i.e., organic acids). Thus, the GCB cartridge
can be a valuable tool for removing matrix interferences that
are not retained by the MSPD C18 cartridge.
The average recoveries for this procedure (Table 2) obtained from analyses of liver and edible tissue samples, spiked
at 10 and 100 ng/g, were good, ranging from 80 to 103%.
These results show that target compounds are completely
eluted from the dispersed phase by the medium polar extraction solvent, methanol–water (70 + 30, v/v), even when the
matrix is present, and that they are retained almost quantitatively by the packing material in the lower column.
Figure 3. Chromatograms obtained from the extraction
of spiked samples (10 ng/g): (1) Al2O3 dispersion of
muscle tissue; (2) Al2O3 dispersion of liver tissue; (3) C18
dispersion of muscle tissue; and (4) C18 dispersion of
liver tissue.
Moreover, we found that if the methanol percentage in the
mixture is raised, the reproducibility of the analytes dissolved
in such a solution and retained by the GCB cartridge is poor.
The lipid and protein component is a problem when the target compounds are quantified by LC/MS/MS, as noted by
Shang et al. (26) in the analysis of fish tissue extracts, because
ion formation is suppressed and therefore signal intensity is altered. The proposed methodology overcomes this inconvenience because a consistent part of the interfering substances
is retained by the MSPD column, and then the extract from the
Dispersing Phase Al2O3
Several papers have reported the use of normal-phase aluminum oxide as a column packing material for cleanup of the
extracts of animal tissues (29, 30), whereas aluminum oxide is
used less frequently as a sample dispersing agent (31).
Dispersing the sample in this solid support appears more
complicated when compared with other reversed-phase
sorbing materials (chemically modified silica-based supports), because aluminum oxide has no ability to “dissolve”
the tissue. The result is obtained with mechanical effort; it is
necessary to finely mill the biological sample with the inert
support to completely disrupt the tissues.
The most effective eluting solution was found to be methanol–water (50 + 50, v/v). As described in the Experimental
section, the volume needed for extraction is twice that used
with the C18 packing. This additional volume increases the to-
Table 2. Accuracy, precision, and limits of detection (LODs) obtained for selected analytes by using 2 different
dispersing agents in the MSPD process for 2 fish tissues spiked at 2 levels
Recovery, % (RSD, %)a
Muscle tissue
Liver tissue
Dispersing agent Spiking level, ng/g
a-ZOL
b-ZOL
ZON
a-ZOL
C18
LOD, ng/g
a
n = 6.
ZON
10
83 (8.1)
86 (7.5)
90 (10.1)
80 (11.0)
98 (9.8)
94 (10.9)
100
85 (7.8)
88 (7.0)
92 (9.3)
83 (9.1)
103 (8.2)
95 (10.2)
0.2
0.2
0.1
0.1
0.2
0.1
10
68 (12.9)
70 (13.6)
71 (17.8)
67 (12.8)
71 (14.5)
67 (23.7)
100
71 (9.3)
67 (9.7)
70 (14.3)
68 (10.2)
70 (9.9)
69 (17.3)
0.4
0.8
0.5
0.5
1.0
0.4
LOD, ng/g
Al2O3
b-ZOL
LAGANÀ ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 86, NO. 4, 2003 735
tal analysis time, because the flow during the extraction step
must be kept below 1 mL/min to avoid the formation of preferential paths in the column.
Preliminary experiments in which just the target analytes,
without the biological tissues, were dispersed showed that
they were strongly retained by the solid phase with the use of
organic solvents only. The results allowed us to develop a
cleanup and extraction strategy by using C18 to retain nonpolar
compounds such as lipids; aluminum oxide does not have this
capability. The presence of fatty compounds, as mentioned
above, interferes with the subsequent GCB cleanup by decreasing the reproducibility of the entire analysis and yielding
high relative standard deviations (RSDs; data not shown).
Moreover, the coeluted compounds can interfere with the determination step by influencing ion formation in the interface.
The preliminary cleanup with hexane and methanol removes part of this interference. However, the eluates from the
alumina packed column were more opaque than those from
the C18 column, and the evaporation of the final solution
yielded a white residue that tests with ninhydrin showed was
mainly protein, which is not retained by the matrix dispersing
agent (in contrast to C18).
The average recoveries (Table 2) obtained from analyses
of liver and edible tissue samples, spiked at 10 and 100 ng/g,
were acceptable, but lower than those obtained with the C18
procedure; moreover, the RSDs were higher. The only advantage of using aluminum oxide is its lower cost, which is
about 1/10 of the cost of the chemically modified silica-based support.
Recovery, Precision, and Method Detection Limit
Before the samples for the recovery study were spiked, a
blank sample for each method of sample preparation was analyzed. Peaks of the target compounds in the blank
chromatograms were absent. The recoveries of all the compounds from analyses using C18 or neutral alumina are shown
in Table 2.
The precision of the methods, expressed as the RSD of the
recoveries, was determined by repeating each analysis
5 times. As Table 2 shows, the data obtained with the C18
method are better both in terms of recovery and precision, indicating that this matrix dispersing agent is more effective and
reproducible, because for both liver and muscle tissues the
RSD values are lower.
The between-day precision of the methods was assessed by
analyzing 5 times during 1 work week a sample spiked at
100 ng/g. The results showed that the method is reproducible;
the RSDs were <13% for the C18-based method and <28% for
the Al2O3-based method. The results are comparable with
those for other published procedures for the analysis of biological tissues.
The limits of detection (LODs) of both techniques were
calculated by using a signal-to-noise (S/N) ratio of 3 (ratio of
the signal intensity and the intensity of the noise was used).
These data are shown in Table 2. The data reflect the superiority of the “clean” extracts obtained with the reversed-phase
material. The intraday precision of the LOD was £20%, which
confirmed that the entire procedure is very reproducible.
Conclusions
The results obtained with this methodology demonstrate
that the sample preparation protocol, in combination with determination by LC/MS/MS, is well suited for the analysis of
biological tissues for ZON and its metabolites at the ng/g
level. Two dispersing agents were tested: chemically modified silica (C18) and neutral alumina. The data obtained
showed that the C18-based protocol is well suited for the determination of the macrocyclic lactones studied (recoveries of
³83%), whereas normal-phase alumina can be used for the
same purpose, in the same way, as a low-cost alternative to
produce recoveries that are still acceptable.
The analysis by means of MSPD is easy to use, does not require any special equipment, and needs minimal quantities of
solvent. The target compounds can be selectively determined
by the LC/MS/MS technique, which allows determination of
as little as 0.1 ng/g.
The improvement in the LODs attained for the method presented here, which are lower than those obtained for other
methods used to determine ZON and its metabolites by
LC/TISP–MS/MS, is largely due to the absence of matrix
components in the sample extract provided by the extraction
and SPE cleanup.
Thus, it can be concluded that the method is suitable for use
in monitoring studies, e.g., for metabolism and
bioaccumulation studies of ZON in fishes species, because the
fate of this substance has not yet been studied in rainbow trout.
Moreover, analytical schemes similar to those proposed can
be applied to the determination of different substances in food.
References
(1) Betina, V. (1984) in Mycotoxins—Production, Isolation, Separation and Purification, V. Betina (Ed.), Elsevier,
Amsterdam, The Netherlands, pp 25–36
(2) Kolb, E. (1984) Z. Gesamte Inn. Med. Ihre Grenzgeb. 39,
353–358
(3) Pfohl-Leszkowicz, A., Chekir-Ghdirra, L., & Bacha, H.
(1995) Carcinogenesis 16, 2315–2320
(4) Ahamed, S., Foster, J.S., Bukovsky, A., & Wimalasena, J.
(2001) Mol. Carcinogen. 30, 88–98
(5) Withanage, G.S., Murata, H., Koyama, T., & Ishiwata, I.
(2001) Vet. Hum. Toxicol. 43, 6–10
(6) Scudamore, K.A., Nawaz, S., & Hetmanski, M.T. (1998)
Food Addit. Contam. 15, 30–55
(7) Hussein, S.H., & Brasel, J.M. (2001) Toxicology 167,
101–134
(8) Cheeke, P.R. (Ed.) (1998) in Natural Toxicants in Feeds,
Forages and Poisonous Plants, Interstate Publishers, Inc.,
Danville, IL, pp 87–136
(9) Harrison, P.T.C., Holmes, P., & Humfrey, C.D.N. (1997) Sci.
Total Environ. 205, 97–106
(10) Francis, G., Makkar, H.P.S., & Backer, K. (2001)
Aquaculture 19, 197–227
736
LAGANÀ ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 86, NO. 4, 2003
(11) Arukwe, A., Grotmol, T., Hauger, T.B., Knudsen, F.R., &
Goksøyr, A. (1999) Sci. Total Environ. 236, 153–161
(12) Liu, M.T., Ram, B.P., Hart, L.P., & Pestka, J.J. (1985) Appl.
Environ. Microbiol. 50, 332–336
(13) Warner, R., Ram, B.P., Hart, L.P., & Pestka, J.J. (1986) J.
Agric. Food Chem. 34, 714–728
(14) Hagler, W.M., Mirocha, C.J., Pathre, S.V., & Behrens, J.C.
(1979) Appl. Environ. Microbiol. 37, 849–853
(15) Seidel, W., Poglits, E., Schiller, K., & Lindner, W. (1993) J.
Chromatogr. 635, 227–234
(16) Lawrence, J.F., & Scott, P.M. (1993) in Techniques, Application and Quality Assurance, D. Barcelo (Ed.), Elsevier,
Amsterdam, The Netherlands, p. 273
(17) Schuhmacher, R., Krska, R., Grasserbauer, M., Edinger, W.,
& Liew, H. (1998) Fresenius Z. Anal. Chem. 360, 241–245
(18) Fu, I., Woolf, E.J., & Matuszewsky, B.K. (1997) in Pharmaceutical and Biomedical Analysis, 8th International
Symposium, Orlando, FL, May 4–8, Abstract M/P-A9
(19) Kebarle, P., & Tang, L. (1993) Anal. Chem. 65, 972–986
(20) Zöllner, P., Jodlbauer, J., & Lindner, W. (1999) J.
Chromatogr. A 858, 167–174
(21) Jodlbauer, J., Zöllner, P., & Lindner, W. (2000)
Chromatographia 51, 681–687
(22) Zöllner, P., Berner, D., Jodlbauer, J., & Lindner, W. (2000) J.
Chromatogr. B 738, 233–241
(23) Laganà, A., Fago, G., Marino, A., & Santarelli, D. (2001)
Rapid Commun. Mass Spectrom. 15, 304–310
(24) Rosenberg, E., Krska, R., Wissiak, R., Kmetov, V., Josephs,
R., Razzazi, E., & Grasserbauer, M. (1998) J. Chromatogr. A
819, 277–288
(25) Barker, S.A., Long, A.R., & Short, C.R. (1989) J.
Chromatogr. 475, 353–361
(26) Shang, D.Y., Ikonomou, M.G., Macdonald, R.W., Vonguyen,
L., & Raymond, K. (1998) Society of Environmental
Toxicology and Chemistry, 19th Annual Meeting, Charlotte,
NC, Abstract Book, November, p. 201
(27) Crescenzi, C., Di Corcia, A., Passariello, G., Samperi, R., &
Turnes-Carou, M.I. (1996) J. Chromatogr. A 733, 41–50
(28) Wells, M.J.M., & Yu, L.Z. (2000) J. Chromatogr. A 885,
237–250
(29) Tolls, J., Haller, M., & Sijm, D.T.H.M. (1999) J.
Chromatogr. A 839, 109–117
(30) Zhao, M., Van der Wielen, F., & De Voogt, P. (1999) J.
Chromatogr. A 837, 129–138
(31) Kishida, K., & Furusawa, N. (2001) J. Chromatogr. A 937,
49–55