Determination of nitrofuran residues in tilapia tissue by enzyme-linked immunosorbent assay and confirmation by liquid chromatography tandem mass spectrometric detection

Article abstract of DOI:10.1002/jccs.200900086

Furazolidone is a broad-spectrumantibiotic that is frequently used in aquaculture on account of its excellent antibacterial properties. In this study, both the enzyme-linked immunosorbent assay (ELISA) and high-performance liquid chromatography-tandemmass

Full text of DOI:10.1002/jccs.200900086

Journal of the Chinese Chemical Society, 2009, 56, 581-588
581
Determination of Nitrofuran Residues in Tilapia Tissue by Enzyme-Linked
Immunosorbent Assay and Confirmation by Liquid Chromatography
Tandem Mass Spectrometric Detection
Chung-Wei Tsaia (
), Chi-Hsin Hsua (
) and Wei-Hsien Wanga,b,* (
)
aDepartment of Marine Biotechnology and Resources, National Sun Yat-Sen University,
Kaohsiung 804, Taiwan, R.O.C.
bNational Museum of Marine Biology and Aquarium, Pingtung 944, Taiwan, R.O.C.
Furazolidone is a broad-spectrum antibiotic that is frequently used in aquaculture on account of its ex-
cellent antibacterial properties. In this study, both the enzyme-linked immunosorbent assay (ELISA) and
high-performance liquid chromatography-tandem mass spectrometry (LC-MS/MS) methods were used to
analyze the content of residual 3-amino-2-oxazolidinone (AOZ), a metabolite of furazolidone in Tilapia
tissue. Homogenized fish samples were spiked with various amounts of AOZ, and following combined
acid-hydrolysis and derivatization of the homogenized tissue with 2-NBA (2-nitrobenzaldehyde), sample
clean-up was performed and the derived 2-nitrophenylmethylene-3-amino-2-oxazolidinone (NPAOZ)
was analyzed. Using the LC-MS/MS method, a linear correlation between measured concentration Y and
spiked concentration X was observed: Y = 0.4518X - 0.0166, R2 = 0.9972. The linear equation for the
ELISA method was Y = 0.9322X + 0.5168, R2 = 0.9066. These results demonstrated that the ELISA
method might overestimate the residual AOZ content at low concentrations. The detection limit and re-
covery of the known addition were 0.05 mg kg-1 and 108% for the LC-MS/MS method and 0.31 mg kg-1 and
305% for the ELISA method, respectively.
Keywords: Nitrofuran; Metabolite; ELISA; Furazolidone.
INTRODUCTION
the EU.4
Furazolidone [3-(5-nitrofurfurylideneamino)-2-ox-
azolidinone] is a nitrofuran antibacterial agent (antibiotic),
and is frequently used as an antibacterial veterinary drug in
the form of a food additive for the treatment of gastrointes-
tinal infections, including bacterial enteritis caused by
Escherichia coli and Salmonella in livestock, aquaculture
and poultry. If nitrofuran remains in food, it causes muta-
genesis, carcinogenicity and teratogenesis. Because of the
toxicity of nitrofuran, the European Union (EU) prohibited
the use of nitrofuran antibiotics in food-producing animals
by listing them in Annex IV of Council Regulation 2377/
90,1,2 and furazolidone was added to Annex IV in June
1995. A provisional maximum residue limit (MRL) of 5 mg
kg-1 was established by the EU for furazolidone,3 and the
total amount of nitrofuran residues in animals from food-
stuffs should not exceed this legal limit. Drugs in Annex IV
can no longer be registered for use in any farm animals in
In previous studies, it has been reported that furazo-
lidone overdose in mammals may not only cause muscle
convulsions but also can lead to nerve breakage throughout
the body.5-7 Furazolidone can also inhibit sperm production
in humans, mice and poultry.8 The majority of a dose of the
parent drug furazolidone taken by livestock is metabolized
and destroyed by bacteria in the intestine and stomach; a
smaller part is absorbed in the gut and transferred to the
liver,9 where the drug is broken down by enzymes, and an
even smaller proportion is eliminated in its original form
without being metabolized.10
Research into the half-life of furazolidone in pigs re-
ported that half of the parent drug was depleted in 2.6 h;11
the excretive pathway was 60-70% in the urine, 15% in the
bile and 19% in the excrement.12 Broilers were treated with
medicated feed containing 50 mg kg-1 of nifursol over a
7-day period, and the results showed that nifursol could be
* Corresponding author. Tel: +886-7-525-200 ext.5029; Fax: +886-7-525-1595; E-mail: whw@mail.nsysu.edu.tw

582 J. Chin. Chem. Soc., Vol. 56, No. 3, 2009
Tsai et al.
detected only in the bile and the plasma 3 days after the ces-
sation of treatment; it could not be detected in the liver, kid-
ney or muscle samples of the broilers (limit of determina-
tion, LOD = 1-2 mg kg-1 for tissue).13
The ion source used for mass spectrometers is electrospray
ionization (ESI), which is a soft ion source that produces
signals for parent and daughter ions of the analyzed com-
pound.
The absorption and elimination half-lives of furazo-
lidone in the serum are 0.2 and 3.3 h, respectively,14 and
therefore the detection of the parent drug in the animal tis-
sue is generally not very applicable or useful. Some resi-
dues are metabolites of furazolidone, one of which is a 3-
amino-2-oxazolidinone (AOZ) side chain, which can be re-
leased from bound residues under acidic conditions and de-
tected by reaction with 2-nitrobenzaldhyde to produce 2-ni-
trophenylmethylene-3-amino-2-oxazolidinone (NPAOZ)15
(Fig. 1).
Another method commonly used to analyze AOZ res-
idues is ELISA,18 which involves the production of poly-
clonal antibodies capable of detecting AOZ following deri-
vation with 2-nitrobenzaldehyde. However, a commonly-
occurring problem of the ELISAmethod is one of false pos-
itive results.
In this study, we attempted to determine the presence
of nitrofuran residues in Tilapia tissue by ELISA and con-
firm the results using high-performance liquid chromatog-
raphy-tandem mass spectrometry (LC-MS/MS).
Many residue analysis methods have been described,
including high-performance liquid chromatography-ultra-
violet (HPLC-UV), liquid chromatography-mass spec-
trometry (LC-MS)16 and liquid chromatography-tandem
mass spectrometry (LC-MS/MS);17 however LC-MS and
LC-MS/MS are prohibitively expensive. Nowadays, a rapid
and high-capacity screening method is required in order to
process the large number of samples to be analyzed, and
hence enzyme-linked immunosorbent assay (ELISA) is a
common method more recently used in normal laboratories
for this purpose. ELISA is cheaper to carry out than LC-
MS/MS, and the ELISA method has a detection limit simi-
lar to that of LC-MS/MS.
EXPERIMENTAL
Chemicals and solvents
Several milligrams of AOZ were provided by Riedel-
de Haën (Sigma-Aldrich Laborchemikalien Gmbh, D-
30918 Seelze, Germany). An AOZ stock standard solution
(10 mg L-1) was prepared in methanol and stored at 4 °C in
the dark. Standard solutions were then prepared from the
stock solution at concentrations of 0.05, 0.1, 0.2, 0.5, 1.0,
2.0, 4.0 mg L-1 AOZ in methanol. Methanol (HPLC gradi-
ent grade), ethyl acetate (HPLC grade) and dimethyl sulf-
oxide (DMSO, spectroscopy grade) were supplied by
Merck (Darmstadt, Germany), and hydrochloric acid (pro
analysis grade) by Riedel-de Haën (Seelze, Germany). Wa-
ter was double-distilled prior to use. 2-nitrobenzaldehyde
(2-NBA) and ammonium acetate were obtained from Fluka
(Buchs, Switzerland), di-potassium hydrogen phosphate
Methods for the detection of NPAOZ in feeds and bi-
ological matrices have been detailed in the literature, and
analysis has been performed by reversed-phase (RP) high-
performance liquid chromatography-mass spectrometry.
Fig. 1. The structures of tissue-bonded AOZ (3-amino-2-oxazolidinone), released type of AOZ, derivative reagent 2-NBA
(2-nitrobenzaldehyde) and target derivative NPAOZ (2-nitrophenylmethylene-3-amino-2-oxazolidinone).

Determination of Nitrofuran Residues in Fish
J. Chin. Chem. Soc., Vol. 56, No. 3, 2009 583
trihydrate from Merck, and sodium hydroxide from Riedel-
de Haën.
Table 1. Optimized conditions for the analysis of AOZ in fish
tissue by HPLC
HPLC analysis
HPLC gradient
Chromatographic analyses were performed using an
high performance liquid chromatography system (HPLC,
Agilent 1100 Series) and separation were achieved using an
reversed-phase C18 column (4.6 ´ 150 mm, 5 mm, Agilent,
ZORBAX SB-C18). The auto-sampler was equilibrated at
20 °C. The analytes were separated with a mobile phase
consisting of 0.5 mM ammonium acetate and 0.05% formic
acid in water (eluent A) and methanol (eluent B) at a flow
rate of 700 mL min-1. The linear gradient program was: 0-
0.5 min from 65 to 30% A; 0.5-6 min 30% A; 6-8 min from
30 to 65% A; 8-9 min 65% A (Table 1). The injection vol-
ume was 25 mL.
Eluent A (%)
0.5 mM ammonium
acetate
Eluent B (%)
100% Methanol, HPLC
grade
Time
(min)
0
65
30
30
30
65
65
35
70
70
70
35
35
0.5
3.5
6.0
8.0
9.0
HPLC system
Sample temperature
Injection volume
Column type
25 °C
25 mL
4.6 ´ 150 mm, 5 mm, Agilent,
ZORBAX SB-C18
25 °C ± 2 °C
700 mL min-1
Electrospray ionization mass spectrometer (ESI-MS-
MS)
Column temperature
Flow rate
Mass spectrometry analysis was carried out using a
API 4000 tandem quadrupole mass spectrometer (Applied
Biosystem, USA). The instrument was operated using an
electrospray (ESI) source in positive mode. ESI parameters
were: capiliary voltage 5500 V, source temperature 650 °C.
Collision-induced dissociation was performed using argon
as the collision gas at the pressure of 4 ´ 10-3 mbar in the
collision cell. In all cases, [M+H]+ ions were found to be
the most abundant. These ions were selected as the precur-
sor ions, the most abundant product ions were selected the
most sensitive transition for quantification purposes and a
second one for confirmation. It shows MS/MS transitions
for quantification and confirmation for each of the selected
compounds in Fig. 2. The parent/fragment ion combina-
tions and the optimizing parameters of ion path entrance
(DP), collisional focusing quadrupole (EP), offset on colli-
sion cell quads (CE) and Q3 entrance lens (CXP) are listed
in Table 2. The dwell time for each fragmentation pathway
was 150 ms. The electrospray voltage was set to 5500 V
and the collision energy to 15 eV. Nitrogen was used as
collision gas. Data acquisition was performed using Ana-
lyst 1.4.1 software for Applied Biosystem.
50, 150, 450, 1350 and 4050 ng L-1 AOZ in aqueous solu-
tion) standards of AOZ. Each level were prepared, which
included peroxidase-conjugated AOZ, anti-AOZ antibody,
and substrate/chromogen (containing tetramethylbnezi-
dine). A stop solution containing 1N sulfuric acid and a
sample-washing buffer made with 10 mM phosphate buffer
(pH = 7.4) were also prepared. The microtiter plate spectro-
photometer was set at 450 nm.
Sample extraction (ELISA standard method)
4 mL H2O, 0.5 mL 1 M HCl and 100 mL 10 mM 2-ni-
trobenzoic aldehyde (in DMSO) was added to a 1 ± 0.02 g
sample of homogenized fish tissue. After thorough shak-
ing, the tubes containing the samples were incubated at 37
°C overnight (for more than 16 h); 5 mL 0.1 M K2PO4, 0.4
mL 1 M NaOH and 10 mL ethyl acetate were then added
and the tubes shaken vigorously for 30 sec then centrifuged
3000 g for 10 min at room temperature (20-25 °C). A 2.5
mL portion of the organic layer (ethyl acetate) was then
transferred into a new vessel and reduced to dryness with
N2, after which the residue was dissolved in 1 mL n-hexane
and mixed thoroughly with 1 mL sample buffer. The ves-
sels were centrifuged again, then 50 mL of the aqueous
layer was placed in each well.
Enzyme immunoassay for the quantitative analysis of
AOZ (reagents and equipment)
Each immunoassay kit contained sufficient material
for 96 measurements. Each microtiter consisted of 96 wells,
which were coated with capture antibodies against anti-
AOZ-antibodies. There were six concentration levels (0,
Determination of antibody titer
All standards and samples (50 mL of each standard so-
lution or prepared sample) were added to a sufficient num-

584 J. Chin. Chem. Soc., Vol. 56, No. 3, 2009
Tsai et al.
Table 2. Parent/fragment ion combinations used for multiple reaction monitoring
Transition Reactions
Conditions of MRM
(m/z)
Analyte
NPAOZ
DP
EP
CE
CXP
Quantitation
236.2 ® 134.0
236.2 ® 104.0
58
58
10
10
18
17
11
11
Analyte confirmation
RESULTS AND DISCUSSION
Validation of the method
ber of wells in the microwell holder to ensure that each was
run in duplicate, and the positions of the standards and sam-
ples were recorded. A new pipette tip was used for each
standard or sample. 50 mL of the enzyme conjugate was
placed in the bottom of each well, and 50 mL of the anti-
body was added to each well, followed by gentle mixing by
rocking the plate manually, then incubation for 1 h at room
temperature (20-25 °C) in the dark. 100 mL of the stop solu-
tion was then added to each well and the solutions gently
mixed by manual rocking of the plate. The absorbance at
450 nm was then measured against an air blank and the re-
sults were read within 60 min of the addition of the stop so-
lution.
The purpose of this study was to use liquid chroma-
tography-tandem mass spectrometric detection to confirm
the results obtained using the enzyme-linked immunosor-
bent assay method to extract AOZ residue from the tissue
of Tilapia fish. The method used was as described in the il-
lustration accompanying the ELISA kit. There were two
groups of gradation standards: the first was the AOZ stan-
dard in each ELISA kit (0, 50, 150, 450, 1350, 4050 ng
kg-1) (Fig. 3), which was determined by two methods with-
out direct extraction; the second was determined using
spiked specific concentrations in the homogenized samples
Fig. 2. Mass spectrum of NPAOZ and its typical fragmentation pattern.

Determination of Nitrofuran Residues in Fish
J. Chin. Chem. Soc., Vol. 56, No. 3, 2009 585
of Tilapia tissue (0, 0.05, 0.1, 0.5, 1.0, 2.0, 4.0 mg kg-1). Af-
ter extraction using the ELISA standard method, the results
were determined by antibody titer and confirmed by LC-
ESI-MS/MS (Fig. 4).
curves obtained for the nitrofuran metabolites (n = 8) in the
0.05~4 mg kg-1 range were linear, with correlation coeffi-
cients R2 > 0.998 for AOZ (Fig. 5).
Linearity (ELISA)
Analysis of control fish tissue
In the same way, the spiked specific concentration of
Analysis of the control fish tissue by LC-MS/MS con-
firmed that no AOZ was present. The analysis result for the
control fish tissue obtained by antibody titer was 0.14~0.48
mg kg-1, with a mean of 0.3 mg kg-1 and a standard deviation
(S.D.) of 0.11 mg kg-1 (Table 3). There must therefore have
been some disturbance of the matrix in the ELISA method.
Compared with the control group about the LC-MS/MS
were all none determination.
Linearity (LC-ESI-MS-MS)
The AOZ standard in the ELISA kit was determined
by LC-ESI-MS-MS, the result of which was perfect linear-
ity (y = 1.0003x + 0.0007, R2 = 1). The linearity of the de-
termination of the gradient standard without fish tissue was
y = 0.8052x - 0.0241, R2 = 0.9982. External calibration
Fig. 3. ELISA standard curve for RIDASCREEN Ni-
trofuran (AOZ) (Art. No.: R3701) (Y = –0.6298X
+ 2.4468, R2 = 0.9686).
Fig. 4. LC-MS/MS chromatograms (MRM) of spiked 5.0 ng/g AOZ in Tilapia fish tissue. (A) sum of the chromatogram, (B)
confirmation of analyte AOZ (236.2 ® 104.0 amu), (C) quantification of analyte AOZ (236.2 ® 134.0 amu).

586 J. Chin. Chem. Soc., Vol. 56, No. 3, 2009
Tsai et al.
tion of the true content (Fig. 7). Diblikova et al. published
about the ELISA method that the detection capability was
0.4 mg kg-1 for shrimp, poultry, beef and pork muscle.18
Furthermore, they showed the blank tissue which were
overestimated 0.1-1.7 mg kg-1 and 0.1-1.4 mg kg-1 in shrimp
and chicken, respectively.
Table 3. Precision of the ELISA and LC-MS/MS methods in
control fish tissue
Control fish tissue
ELISA (mg L-1)
LC-MS/MS (mg L-1)
group (n = 10)
1
2
3
4
5
6
7
8
0.25
0.38
0.34
0.39
0.48
0.26
0.20
0.14
ND
ND
ND
ND
ND
ND
ND
ND
CONCLUSION
This study involved the quarantine of aquaculture
products for medicament determination, in which two com-
monly-applied methods were used to quantify AOZ in fish
muscle tissue.
Mean
S.D.
0.30
0.11
0
0
Electrospray ionization-tandem mass spectrometry
was used for detection, and two fragment ions per analysis
were used as additional identifiers, taking the EU regula-
tions into account. The results of this study showed that the
LC-MS/MS method resulted in a smaller distribution, but
the equipment required for LC-MS/MS analysis is more
expensive than that required for the ELISA method. In
addition, sample extraction and pretreatment were more
complicated in the ELISA method than in the LC-MS/MS
method. For these reasons, LC-MS/MS equipment is com-
monly found in the laboratories of Taiwan, and the ELISA
method is familiarly used for analysis of aquaculture prod-
ucts for medicament determination. If samples are found to
be positive by the ELISAmethod, the results should then be
reconfirmed by the LC-MS/MS method, as the AOZ con-
tent is overestimated when using the ELISA method.
Our future research will focus on reducing the error in
AOZ in the control fish tissue was determined as 0.05~4 mg
kg-1. Then, analysis of AOZ in the samples with and with-
out tissue was performed by ELISA. The linearity of the
proportion of AOZ in the tissue was y = 0.9322x + 0.5168
and R2 = 0.9066, for which the percentage of extraction
was 93% as compared with perfect extraction. The linearity
was therefore worse and the intercept was too high (0.5168
mg kg-1) (Fig. 6). In the spiked specific concentration analy-
sis of samples without fish tissue, the percentage of extrac-
tion was found to be 170% and the intercept 0.3 mg kg-1.
Fish tissue samples must therefore cause a disturbance of
the matrix in the ELISA method of about 0.2 mg kg-1, and
there must be some error in the ELISA method accounting
for 0.3 mg kg-1. Use of the ELISA method to determine the
AOZ content in fish tissue therefore leads to overestima-
Fig. 5. Calibration curves for: (a) spiked specific concentrations in samples with fish tissue, (b) spiked specific concentra-
tions in samples without fish tissue, (c) standard solutions in the ELISA kits.

Determination of Nitrofuran Residues in Fish
J. Chin. Chem. Soc., Vol. 56, No. 3, 2009 587
Fig. 6. Calibration curves for: (a) spiked specific concentrations in samples with fish tissue, (b) spiked specific concentra-
tions in samples without fish tissue, (c) standard solutions in the ELISA kits.
Fig. 7. Comparison of calibration curves for ELISA method, standard sample and LC-MS/MS method.
2. Commission Regulation (EC) 1442/95, Off. J. Eur. Com.
1995, L143, 26.
the results obtained by the ELISA method and modifying
the method in order that the accuracy and speed matches
that of the LC-MS/MS method.
3. Commission Regulation n. 2701/94. Off. J. Eur. Com. 1994,
L287, 7.
4. Macri, A.; Mantovani, A. J. Exp. Clin. Cancer Res. 1995,
14, 119.
ACKNOWLEDGEMENT
This research was supported by National Museum of
Marine Biology and Aquarium of Taiwan, ROC.
5. Egyed, M. N.; Shlosberg, A.; Nobel, T. A.; Opfer, U.;
Davidson, M.; Yakobson, B.; Perl, S. Vet. Bull. 1983, 53,
3513.
6. Hofmann, W.; Thiel, W.; Reinacher, M.; Dirksen, G. Deut.
Tier. Woch. 1977, 84, 1-7.
Received December 3, 2008.
7. Litjens, C. G. Vet. Bull. 1976, 46, 296.
8. Nelson, W. O.; Bunge, R. G. J. Urol. 1957, 77, 275-281.
9. Cohen, J. Dissert. Abstr. Intern. 1980, 40, 3687B.
REFERENCES
1. Council Regulation (EEC) 2377/90, Off. J. Eur. Com. 1990,
L224, 1.

588 J. Chin. Chem. Soc., Vol. 56, No. 3, 2009
Tsai et al.
10. Tatsumi, Y.; Takahashi, Y. Chem. Pharm. Bull. 1982, 30,
3435-3438.
2003, 30, 2.
15. Cooper, K. M.; Caddell, A.; Elliott, C. T.; Kennedy, D. G.
Anal. Chim. Act. 2004, 520, 79-86.
11. Tennent, D. M.; Ray, W. Proc. Soc. Exp. Biol. Med. 1971,
138, 808.
16. Draisci, R.; Giannetti, L.; Palleschi, L.; Brambilla, G.; Serpe,
L.; Gallo, P. J. Chromatogr. A. 1997, 777, 201-211.
17. Szilagyi, S.; Calle, B. d. l. Anal. Chim. Act. 2006, 572, 133-
120.
12. Pietsch, W.; Belitz, B.; Trolidenier, H.; Meister, B. Vet. Bull.
1979, 49, 951.
13. Zuidema, T.; Mulder, P. P. J.; Rhijn, J. A. V.; Keestra, N. G.
M.; Hoogenboom, L. A. P.; Schat, B.; Schat, B.; Kennedy, D.
G. Anal. Chim. Act. 2005, 529, 339-346.
18. Diblikova, I.; Cooper, K. M.; Kennedy, D. G.; Franek, M.
Anal. Chim. Act. 2005, 540, 285-292.
14. Guo, J. J.; Tung, M. C.; Chou, H. N.; Liao, I. C. J. Fish. Soc.