Anda di halaman 1dari 6

BOLYGO ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 83, NO.

3, 2000 543

AGRICULTURAL MATERIALS

Determination of Histamine in Tomatoes by Liquid


Chromatography/Mass Spectrometry

ELEK BOLYGO, PAUL A. COOPER, K. MICHAEL JESSOP, and FRANK MOFFATT1


Zeneca Agrochemicals, Jealotts Hill Research Station, Bracknell, Berkshire RG42 6ET, UK

The histamine level in tomato fruits and pastes were of analytical purity grade (Sigma-Aldrich Co.). Buffers
was determined by 2 orthogonal techniques as a were prepared from analytical grade reagents (Merck Ltd.
means of comparing accuracy. Close statistical Poole, Dorset, UK). Solvents (acetonitrile, acetone, toluene)
agreement was found between assays for free were LC grade (Romil, Cambridge, UK). Deionized water
histamine by capillary electrophoresis (with UV was obtained from a Milli Q water purification system (Mil-
absorbance detection), and for the dansyl deriva- lipore Ltd., Watford, Hertfordshire, UK).
tive by reversed-phase liquid chromatography
(LC). Both techniques have adequate sensitivity Biological Materials
for the analysis of endogenous histamine in
Samples of tomatoes were obtained in-house from Zeneca
tomatoes, but LC/electrospray tandem mass
Plant Sciences or from a retailer (market sample). The green
spectrometry was more sensitive by at least
tomatoes were mature but unripe. All samples and extracts
an order of magnitude, down to a level of
were stored at < −18°C when not in use.
0.05 mg/kg.
Instruments

LC/tandem mass spectrometry (LC/MS–MS) was per-

B
iogenic amines such as histamine are natural compo-
nents of food and feed that can affect quality and formed on a Quattro II mass spectrometer (Micromass, Al-
physiology (1). This report provides evidence of the tringham, Cheshire, UK), fitted with a Waters Alliance 2690
accuracy of a previously reported assay which used capillary liquid chromatograph (Waters Ltd., Watford, Hertfordshire,
electrophoresis (CE)(2) for endogenous histamine in tomatoes UK). Off-line LC was performed on a standard HP1100 sys-
by comparing it with an orthogonal liquid chromatographic tem fitted with a quaternary pump and UV detector and a
(LC)-based method. Previous LC methods for histamine in Model 1046A fluorescence detector (Hewlett-Packard,
various plant matrixes, used either postcolumn (3) or precol- Bracknell, Berkshire, UK).
umn derivatization (4–6) with UV absorbance or fluores-
Sample Preparation
cence detection. Reversed- phase LC requires the use of an
ion-pairing agent (3) which would be incompatible with con- Samples of frozen tomatoes were homogenized with dry
ventional electrospray tandem mass spectrometry (MS). The ice. A subsample (20 g) was extracted by maceration with
use of such agents can be avoided through precolumn deriva- aqueous perchloric acid (0.2M), with cooling over an ice
tization, such as dansylation which is broadly applicable to a bath, at an extraction ratio of 0.33 g tomato fruit/mL extract
wide range of analytes, opening up options of detection by as described in an earlier publication (2). Samples were for-
fluorescence (7) and MS on a triple quadrupole instrument. tified with an internal standard, 1,3 diaminopropane (200
µg), prior to extraction in order to provide an independent
Experimental means to assess the efficiency of the procedure. An aliquot of
the clear extract was dansylated according to the method of
Reagents and Solvents Valleé et al. (6), except for an increased reaction time from 5
to 10 min. An aliquot of the extract (100 µL) was treated with
Histamine dihydrochloride (99% purity) and 1,3- saturated sodium carbonate solution (300 µL) in a screw-cap
diaminopropane dihydrochloride were used as analytical ref- vial (2 mL). Dansyl-chloride solution (400 µL, 7.5 mg/mL in
erence and internal standards, respectively (Sigma-Aldrich acetone) was added. The mixture was shaken and then left to
Co., Poole, Dorset, UK). L-Proline, 5-[dimethylamino]naph- derivatize (60°C, 10 min) in the dark. After cooling to room
thalene-1-sulfonyl (dansyl) chloride, and sodium carbonate temperature the excess reagent was quenched by the addition
of L-proline (100 µL, 100 mg/mL in water). After 15 min at
Received July 27, 1999. Accepted by RN November 30, 1999. room temperature, toluene (500 µL) was added; the mixture
1
Author to whom correspondence should be addressed. was vigorously shaken for 20–30 s, centrifuged at 4000 rpm
544 BOLYGO ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 83, NO. 3, 2000

for 3 min, and then transferred to a freezer at < −18°C. The and fluorescence at excitation and emission wavelengths of
toluene layer was separated from the frozen aqueous layer, 259 and 512 nm, respectively. The gain setting was used to
and evaporated to dryness under nitrogen at 40°C. The resi- increase the sensitivity of the detector by a factor of 16 at 10.5
due was dissolved in acetonitrile (200 µL), ultrasonicated for min into the gradient, between the 2 analyte peaks due to
15 s, transferred into a vial, and centrifuged (4000 rpm, differences in the response efficiencies between the dansyl
15 min) before analysis by LC/MS–MS. The crop-to-solvent derivatives of 1,3-diaminopropane and histamine. The gradi-
ratio was 0.17 g/mL for tomato fruit. To demonstrate the per- ent used was: A = water, B = acetonitrile; 0.0–6.0 min,
formance of the method, recovery experiments were con- 60–75% B; 6.0–8.0 min, 75% B; 8.0–13.0 min, 75–95% B;
ducted at various levels of histamine (Table 1) and at 10 mg/ 20– 25 min, 95% B. The flow rate was 1 mL/min.
kg for the internal standard. CE was performed as described previously (2). A voltage
gradient of 5–30 kV was used for 13 min. The sodium citrate
Preparation of Derivatized Standards buffer (20 mM, pH 2.5) was replenished after every injection.
After each sample the capillary was rinsed: water 1 min,
Aqueous solutions of histamine and 1,3-diaminopropane aqueous sodium hydroxide (0.1M for paste extracts, 0.5M for
dihydrochloride containing perchloric acid (0.2M) were de- fruit extracts) 1 min, water 1 min, running buffer 1 min. De-
rivatized as described above. These solutions were used to tection was at 212 nm and the operating temperature was set
determine the procedural recoveries in samples fortified to 35°C. Samples and standards were injected hydrodynami-
with known amounts of histamine and 1,3-diaminopropane cally (50 mbar for 10 s).
(Table 1).
Results and Discussion
Analysis Conditions
The dansylation method of Valleé et al. (6) was applied to
A Prodigy LC column was used (25 cm × 3.2 mm id, ODS tomato extracts. The efficiency of the procedure was found to
2, 5 µm particle size, Phenomenex, Macclesfield, Cheshire, be reliable in all instances as evidenced by the good agree-
UK). The following solvent gradient was used: A = 0.2% ment between the mean recoveries for histamine and the in-
acetic acid in water, B = acetonitrile; 0.0–6.0 min, 60–75% B; ternal standard (1,3-diaminopropane) in samples that had
6.0–8.0 min, 75–98% B; 8.1–13.0 min, 98% B. LC condi- been fortified (Table 1). Although the dansyl derivatives of
tions were: flow rate 1 mL/min, column temperature 40°C, the internal standard and histamine could be separated by re-
sample temperature 10°C, injection volume 10 µL, flow split versed-phase LC, the choice of detector became a critical is-
with ca 20% to MS, 80% to UV/waste. MS data were ac- sue in the examination of samples. Both UV absorbance and
quired by multiple reaction monitoring (MRM) in positive fluorescence were subject to lack of sensitivity, and interfer-
ion electrospray mode. The primary transitions monitored ence by matrix impurities (Figures 1 and 2). The response of
were m/z 578 ➔ m/z 170 for histamine and m/z 541 ➔ m/z the dansylated internal standard in fluorescence was approxi-
170 for the internal standard with additional confirmatory mately 16 times greater than that of the histamine derivative,
transitions between m/z 578 ➔ m/z 315 (histamine) and m/z suggesting that an additional quenching of the fluorescence
541 ➔ m/z 307 (internal standard). The source temperature occurs in the latter case. The sensitivity of the quantum yield
was maintained at 110°C and a cone voltage of 50 V was used for fluorescence of dansyl derivatives to the intra- and inter-
for all experiments. The laboratory collision energy was 30 molecular environment is well known (8,9).
eV for histamine and 35 eV for the internal standard. Nitro- The problems of sensitivity and selectivity were solved by
gen was used as both drying gas (300 L/h) and nebulizing gas MRM MS. The full scan mass spectrum of the derivatized
(50 L/h). The collision gas was argon containing 10% nitro- histamine (Figure 3) contains a prominent signal at m/z 578,
gen at a gas cell pressure of 1.5 × 10−3 mbar. the (M + H)+ ion of histamine with 2 dansyl additions (9).
Offline LC was performed with UV absorbance at 338 nm MS–MS of the m/z 578 ion produces a major fragment ion at

Table 1. Analytical quality control recovery data for LC/MS–MS


Tomato sample Histamine fortification, mg/kg Histamine recovery, % Internal standarda recovery, %

Green 10 85, 88, 92 81, 80, 84, 87, 89


Red ripe 2 74 84
10 91, 89, 93, 80 81, 81, 81, 69, 84, 82, 86
20 74 61
50 83 76
Mean, RSD %, (n) 84.9, 8.3% (10) 80.4, 8.9% (15)

a
Internal standard used to fortify samples at 10 mg/kg.
BOLYGO ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 83, NO. 3, 2000 545

Figure 1. Liquid chromatograms of histamine-containing tomato samples using UV detection at 338 nm. Standard
2.5 µg/mL. R @ 10 mg/kg refers to fortification with histamine and the internal standard at this level. Samples 1
and 2 are examples of typical chromatograms for tomato samples.

m/z 170, corresponding to the ion formed by fragmentation level previously reported for the CE method (2). The detector
of the dansyl group (Figure 4). The LC/MS–MS method was response was linear over the range of 0.34–12.5 µg/mL.
therefore set up to monitor the transitions m/z 578 ➔ m/z 170 The extraction method was validated by calculating the
for histamine and m/z 541 ➔ m/z 170 for the internal stan- procedural recovery values for samples after fortification
dard. Figure 5 shows a typical LC/MS–MS chromatogram with histamine at several levels. The mean recoveries were
from a 10 µL injection of a 1 µg/mL standard mixture of de- 84.9%, (relative standard deviation [RSD] 8.3%, n = 10) and
rivatized histamine and internal standard. The limit of deter- 80.4%, (RSD 8.9%, n = 15) for histamine and internal stan-
mination using this method was 0.05 mg/kg (S/N 4:1), an dard, respectively (Table 1). Recovery levels for a subsample
order of magnitude increase in sensitivity over the 0.6 mg/kg of a tomato fortified with histamine at 10mg/kg and for the

Figure 2. Liquid chromatograms of histamine-containing tomato samples using fluorescence detection at 259 nm
excitation and 512 nm emission. Standard, 2.5 µg/mL. R @ 10 mg/kg refers to fortification with histamine and the
internal standard at this level. Samples 1 and 2 are examples of typical chromatograms for tomato samples.
Figure 3. Full scan mass spectrum of dansyl-derivatized histamine.

Figure 4. Daughter ion mass spectrum of dansyl-derivatized histamine.

Figure 5. LC/MS–MS chromatogram of a 1 µg/mL histamine and internal standard mixture.


BOLYGO ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 83, NO. 3, 2000 547

Figure 6. LC/MS–MS chromatogram of an extract of red ripe tomato which had been fortified (preextraction) with
histamine at 10 mg/kg.

the internal standard were 89 and 81%, respectively (Figure poses of this study. Analytical values of samples by both CE
6). The histamine values were corrected for the presence of and LC were in close agreement (Table 2), with no signifi-
endogenous material by analyzing appropriate control cant difference between the 2 methods, as judged by the Wil-
samples, i.e., similar subsamples from the same tomatoes, coxson non-parametric test (n = 6, p < 0.05).
which had not been fortified. The analysis of a typical control
indicates a natural background histamine level of 2.2 mg/kg Conclusions
(Figure 7).
Data on the level of histamine found in tomatoes (10–13) The accuracy of histamine determination in tomatoes has
are limited, but values range from 0.3 to 24.6 mg/kg. The been confirmed by 2 orthogonal methods involving CE and
levels found in this study by CE/UV and LC/MS–MS were in LC. The LC/MS method was more sensitive but involved ad-
the ranges of 2.0–8.4 mg/kg and 2.1–6.5 mg/kg, respectively ditional sample preparation. In the LC and CE methods, fluo-
(Table 2). For tomato paste samples, a matrix-related signal rescence and UV absorbance detection both lacked specific-
enhancement was noted by LC/MS–MS. In our experience, ity; therefore, these methods may give false-positive or
this can be compensated for by using a matrix-matched stan- inaccurate results because of coelution of endogenous matrix
dard, but further investigation was not necessary for the pur- components.

Figure 7. LC/MS–MS chromatogram of an extract of red ripe tomato, showing an endogenous histamine level of
2.2 mg/kg.
548 BOLYGO ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 83, NO. 3, 2000

Table 2. Comparison of CE with LC/MS–MS results (2) Bolygo, E., Cooper, P.A., Jessop K.M., & Moffatt, F. (1999)
for histamine in tomato J. AOAC Int. 83, 89–94
(3) Beljaars, P.R., Van Dijk, R., Jonker, K.M., & Schout, L.J.
Sample (1998) J. AOAC Int. 81, 991–998
description CEa, mg/kg LC/MS–MSb, mg/kg
(4) Busto, O., Valero, Y., Guasch, J., & Borrul, F. (1994)
Chromatographia 38, 571–578
Green 4.7 4.1 (5) Ibe, A., Saito, K., Nakazato, M., Kikuchi, Y., Fujinuma, K.,
Green 3.1 3.0
& Nishima, T. (1991) J. AOAC Int. 74, 695–698
(6) Valleé, M., Malle, P., & Bouquelet, S. (1997) J. AOAC Int.
Red ripe 2.1 2.2 80, 49–56
Red ripe 2.0 2.1 (7) Fürst, P., Pollack, L., Graser, T.A., Godel, H., & Stehle, P.
Red ripe market 5.2 4.9 (1989) J. Liq. Chromtogr. 12, 2733–2760
(8) Haughland, R.P. (1996) Handbook of Fluorescent Probes
Red ripe market 8.4 6.5
and Research Chemicals, 6th Ed., Molecular Probes, Inc.,
c
Red ripe from field trials n/a 5.9, 4.7, 6.8, 5.6 Eugene, OR
a
(9) Molecular Probes Inc., Internet address: www.probes.com
CE results are mean of 3 replicate injections. (10) Warshawsky, A., Altman, J., Kahana, N., Arad-Yellin, R.,
b
LC/MS–MS results are derived from a single analysis.
c Dehse, A., Hasson, H., Shoef, N., & Gottlieb, H. (1989)
n/a = not analyzed.
Synthesis 11, 825–829
(11) Mahendradatta, M., & Schwedt, G. (1996) Dtsch. Lebensm.
References Rundsch. 92, 218–221
(12) Jadhav, S.S., & Kulkarni, P.R. (1981) J. Food Sci. Technol.,
(1) Sorensen, H., Sorensen, S., Bjergegaard, C., & Michaelsen, India 18, 156–157
S. (1999) Chromatography and Capillary Electrophoresis in (13) Mahendradatta, M., & Schwedt, G. (1998) Z. Lenbensm.
Food Analysis, Royal Society of Chemistry, Cambridge, UK Unters. Forsch. A 206, 246–250