Food Chemistry 135 (2012) 1323 1329

Contents lists available at SciVerse ScienceDirect

Food Chemistry
journal homepage: www.elsevier.com/locate/foodchem

Analytical Methods

Sensitive detection of ochratoxin A in wine and cereals using uorescence-based immunosensing

Beatriz Prieto-Simn , Isao Karube, Hiroshi Saiki
School of Bioscience and Biotechnology, Tokyo University of Technology, Katayanagi Institute 1404-1 Katakura, Hachioji, Tokyo 192-0982, Japan

a r t i c l e
Article history:

i n f o

a bs t r a c t
Ochratoxin A (OTA) is a mycotoxin found in a wide range of food and feedstuffs. Intake of OTA-contaminated food causes health concern due to the harmful effects reported on humans and animals. Much effort is currently devoted to set up and optimise highly sensitive and accurate methods of OTA analysis. This w ork describes the comparison of uorescence-based immunosensing strategies for the analysis of OTA. First, an indirect competitive uoroimmunoassay was designed and optimised. The assay enabled the quantication of the toxin at the levels set by the European legislation. Then, a ow-immunoassay based on kinetic exclusion measurements w as developed. It showed the theoretical lowest limit of detec1 tion enabled by the afnity of the anti-OTA antibody (IC 80 = 12 ng L in the assay solution). Wine and cereal samples were analysed using the optimised o w system. No signicant matrix effects were observed after simple pre-treatment of win e and OTA extraction from corn- akes samples. This simple and highly sensitive automated biosensing-system allows OTA quantication in food and beverages. It is envisaged as a powerful tool for rapid and reliable toxin screening. 2012 Elsevier Ltd. All rights reserved.

Received 18 January 2012 Received in revised form 10 April 2012 Accepted 11 May 2012 Available online 19 May 2012 Keywords: Ochratoxin A Immunosensor Fluoroimmunoassay KinExA Wine Cereals

1. Introduction Ochratoxin A (OTA) is a mycotoxin produced by several species of Aspergillus and Penicillium, fungi that grow on several types of food and feedstuff (Krogh, 1987). Its presence depends on several environmental factors, such as temperature, humidity and storage conditions. OTA is a moderately stable metabolite that can survive food processing. Once it enters the food chain, it becomes a serious health risk for humans and animals. Toxicological studies conrm that OTA can cause strong toxic effects in liver and kidneys (Fink-Gremmels, 1999; Richard & Payne, 2003). It also shows teratogenic, mutagenic and immunosuppressive effects in certain animals. It has been classied by the International Agency for Research on Cancer as a possible human carcinogen (group 2B) (Pfohl-Leszkowicz & Manderville, 2007). Several countries have adopted regulatory limits due to the increased awareness of the potential risk of OTA. The European Commission has set permissible limits of OTA in cereals (5 lg kg1 in raw cereal grains and 3 lg kg1 in processed cereal products for direct human consumption) and in wine and grape-containing beverages (2 lg L 1 ) (Commission Regulation, 2005). The development of validated ofcial analytical methods has become necessary to protect human and animal from OTA intake
Corresponding author. Address: Nanobioengineering Group, Institut de Bioenginyeria de Catalunya (IBEC), C/Baldiri Reixac, 10-12 08028 Barcelona, Spain, Tel.: +34 93 403 71 79; fax: +34 93 403 71 81. E-mail addresses: beatriups@gmail.com, bprieto@ibec.pcb.ub.es (B. Prieto-Simn).
0308-8146/$ - see front matter 2012 Elsevier Ltd. All rights reserved. ttp://dx.doi.org/10.1016/j.foodchem.2012.05.060

(European Mycotoxin Awareness Network (EMAN), 2012). These methods are mainly based on purication with an antibody-based afnity column followed by high-performance liquid chromatography (HPLC) with uorescence detection. HPLC is an accurate and reproducible technique, but requires highly skilled personnel, sophisticated technical equipment and time-consuming sample clean-up. OTA has been also analysed by thin-layer chromatography (Santos & Vargas, 2002). Nevertheless, this method shows low sensitivity. Additionally, immunoassay techniques such as ELISA have been developed as qualitative or semi-quantitative methods, useful for rapid screening (Giray, Atasayar, & Sahin, 2009; Radoi, Dumitru, Barthelmebs, & Marty, 2009). However, they still involve long incubation times and tedious washing steps. Researchers have focused their efforts on developing new accurate and fast screening tools to detect OTA. These tools should protect consumers from the intake of OTA-contaminated food (Pittet, 2005; Prieto-Simn, Noguer, & Camps, 2007). Dipsticks, based on immunoassays and a direct visual reading, have been developed as simple and powerful tools for the preliminary screening of OTAcontaminated samples (Bazin, Nabais, & Lopez-Ferber, 2010; Goryacheva et al., 2008; Kolosova, Saeger, Eremin, & Van Peteghem, 2007; Schneider et al., 2004; Shim, Dzantiev, Eremin, & Chung, 2009a; Shim, Kim, Ryu, Nam, & Chung, 2009b). They are cost-effective and allow fast analysis. A commercial o wthrough rapid test kit is available for OTA detection in wine, cereals and green coffee (Euro-Diagnostica) (Euro-Diagnostica, 2012). The main shortcoming of immunostrips is their limited sensitivity.

1324

B. Prieto-Simn et al. / Food Chemistry 135 (2012) 1323 1329

Sensitive and accurate biosensors for OTA detection have also been developed based on different transduction methods. The simplicity of gravimetric techniques, such as quartz crystal microbalance, has been used to develop immunosensors for the detection of OTA (Tsai & Hsieh, 2007; Vidal, Duato, Bonel, & Castillo, 2009). The high sensitivity of electrochemical techniques has been also used to develop immunosensors for OTA detection. Immunosensors based on voltammetric detection methods require electroactive labels (Alarcn et al., 2006; Heurich, Kadir, & Tothill, 2011; Liu et al., 2009; Prieto-Simn, Camps, Marty, & Noguer, 2008; Radi, Muoz-Berbel, Cortina-Puig, & Marty, 2009). Alternatively, impedimetricbased sensors show the advantage of being label-free (Khan & Dhayal, 2008; Khan & Dhayal, 2009; Radi, Muoz-Berbel, Lates, & Marty, 2009; Zamr et al., 2011). Different optical transduction methods have been used for OTA detection. Surface Plasmon Resonance (SPR)-based biosensors rely upon mass changes at the sensor surface. Due to the small size of OTA (403.8 Da), its direct label-free detection is not possible (Urusov, Kostenko, Sveshnikov, Zherdev, & Dzantiev, 2011; Yuan, Deng, Lauren, Aguilar, & Wu, 2009; Zamr et al., 2011). Optical waveguide light spectroscopy is another optical detection method used for OTA biosensors development. The technique is based on the variations in the refractive index due to changes in the layer thickness. Like SPR devices, competitive approaches to detect OTA are required due to the low sensitivity achieved by a direct non-competitive approach (Adnyi et al., 2007). The Naval Research Laboratory has developed fast, accurate and reliable biosensor arrays for the detection of OTA (Ngundi, Shriver-Lake, Moore, Ligler, & Taitt, 2006; Ngundi et al., 2005; Sapsford et al., 2006). They combine the high sensitivity of uorescence detection techniques and the advantages provided by biosensor arrays. Those advantages include the possibility to measure multiple samples and provide multimycotoxin proles in one assay. The developed arrays assess matrix effects and elucidate synergistic effects among mycotoxins. Kinetic Exclusion Assay (KinExA) is a kinetic method that has been successfully applied to detect low levels of different mycotoxins, such as zearalenone in standard solutions (Carter et al., 2000) and aatoxin B 1 in nut puree, peanut butter, and pistachio meal (Strachan, John, & Miller, 1997). It is a uorometric method based on a sequential injection immunoassay format performed with a commercial instrument. Our aim was to prove the ability to sensitively and reliably detect OTA in wine and cereal samples by combining highly sensitive uorescence detection with kinetic exclusion measurements. As a rst step, a uoroimmunoassay was developed, greatly lowering the limit of detection (LOD) previously reported for an equivalent ELISA assay (Prieto-Simn et al., 2008). Then, a ow-based immunoassay based on kinetic exclusion measurements was optimised. To the best of our knowledge, we report for the rst time the potential of KinExA to be used as a rapid, sensitive and precise technique for the analysis of OTA in food and beverages. 2. Experimental 2.1. Reagents and materials Monoclonal antibody (MAb, clone 5G9, developed in mouse) against OTA was obtained from Soft Flow Biotechnology (Gdll, Hungary). Cross-reactivity with ochratoxin B (OTB) is 9.3%, while no cross-reaction with ochratoxin a has been reported. Cyanine 5 uorescent dye (Cy5)-conjugated F(ab0) 2 fragment of goat antimouse IgG was purchased from Jackson ImmunoResearch Laboratories (West Grove, PA, USA). Bovine serum albumin (BSA), anti-mouse IgG-FITC, poly(vinylpyrrolidone) (PVP), and OTA (from Aspergillus ochraceous) were purchased from Sigma Chemical (St.

Louis, MO, USA). OTA standard solutions were rstly prepared in ethanol (25 g L 1 ) and subsequently diluted in 0.1 M phosphate buffered saline, pH 7.2 (PBS). EZ-Link Amine-PEO 3 -Biotin, avidin, N-hydroxysuccinimide (NHS), N-(3-dimethylaminopropyl)-N0ethyl-carbodiimide hydrochloride (EDC), N,N0-carbonyldiimidazole (CDI) and components of buffers were supplied by Wako Pure Chemical Industries, Ltd. (Osaka, Japan). All solutions were prepared using Milli-Q water. Red wine and corn- akes were purchased from local grocery stores. 96-Well Maxisorp Black microtiter plates were obtained from Nalge Nunc International KK (Tokyo, Japan). Poly(methyl methacrylate) (PMMA) beads (100- lm diameter) were supplied by Sapidyne Instruments (Boise, ID, USA). 2.2. Apparatus Fluorescence measurements were performed with an Ultra Evolution microplate reader (Tecan, Japan). A KinExA 3000 instrument (Sapidyne Instruments) was used as immunoassay platform for kinetic exclusion measurements. 2.3. Synthesis of OTABSA conjugate and biotinylated OTA OTA was conjugated to BSA and biotin according to previously described procedures (Prieto-Simn et al., 2008). Briey, the synthesis of OTA BSA conjugate started activating the carboxyl group of OTA. The activated groups reacted with the lysine groups of BSA to form amide bonds (Xiao, Clarke, Marquardt, & Frohlich, 1995). Biotinylated OTA was synthesised using a commercial carboxyl reactive biotin labelling reagent, slightly modifying the procedure detailed by Pierce (Prieto-Simn et al., 2008). Aliquots of conjugates were stored at 4 C for daily use. 2.4. Preparation of wine and cereal samples Wine (red wine, 2008, 12.5% alcohol) and cereal samples (cornakes) were spiked before or after treatment depending on the study. Treatment of wine samples was based on the addition of 0.1 g of PVP to a 10-mL wine aliquot. The wine/PVP mixture was shaken for 2 min at room temperature and, after ltration, the pH was adjusted to 7.2 (Prieto-Simn et al., 2008). To study OTA recovery three aliquots of OTA-free wine sample were spiked before treatment with the stock solution of OTA. Final concentrations were 3, 2 and 1 lg L 1 . To study matrix effects, an OTA-free wine sample was spiked after ltration to obtain an OTA concentration of 50 lg L 1 . From this spiked sample several dilutions from 0.3 ng L 1 to 6 lg L 1 were prepared. Corn-akes samples were milled to powder using a hand-held blender. To study the extraction efciency several lots of 2 g of milled cereals were spiked with the stock solution of OTA to obtain a nal concentration of 4, 3 or 2 lg kg1 before extraction. Extraction was performed with 20 mL 2% KHCO3 methanolic solution (70:30, v/v) on a sonication bath for 45 min. Extracts were centrifuged for 5 min at 5000 rpm to remove the solids. Collected supernatants were evaporated to dryness under nitrogen at room temperature. The residues were re-suspended in 6 mL PBS. Reconstituted solutions were ltered through 0.45 lm cut-off Whatman nylon membranes. Wine and cereal samples were mixed with an equal volume of a PBS solution containing a pre-determined amount of anti-OTA. 2.5. Fluoroimmunoassay protocol Two indirect competitive uoroimmunoassay protocols were performed depending on the OTA immobilisation procedure. Fluorometric checkerboards enabled the optimisation of all concentrations of the involved reagents. For the OTA BSA based protocol, the

B. Prieto-Simn et al. / Food Chemistry 135 (2012) 1323 1329

1325

rst step was the adsorption of the conjugate onto microtiter wells using a concentration of 20 mg L 1 in 0.05 M carbonate buffer, pH 9.6. In the case of the biotinylated OTA-based protocol, the wells were rst modied with a 1 mg L 1 avidin solution in HEPES buffer. Then, a 20 mg L 1 biotin-OTA solution in PBS was added. After OTA immobilisation, both uoroimmunoassay protocols followed the same steps. After blocking the free binding sites on the wells with 5% skimmed milk solution in PBS, 1 h competition step was performed by incubation with 10 lL of 1 mg L 1 MAb against OTA and 90 lL of OTA standard solutions at different concentrations (from 76 ng L 1 to 5 mg L 1 ), both in 1% BSA solution in PBS (PBS BSA). Afterwards, 1 h incubation of FITC-conjugated antimouse secondary antibody was carried out. Fluorescence values were measured in PBS buffer at 518 nm after excitation at 494 nm. Each step was performed at 37 C with shaking and protected from light. The wells were thoroughly rinsed ve times with 0.1 M PBS containing 0.05% Tween 20 (PBS-T) between each step. Wells were rinsed two additional times with 0.1 M PBS before the uorescence measurement. Assays were performed in triplicate. 2.6. Kinetic Exclusion Assay protocol KinExA was set up as a competitive immunoassay. Again, two different protocols were followed depending on the OTA conjugate immobilised on the PMMA beads: OTA BSA conjugate or biotinylated OTA. KinExA instrument involves three main functional components: a o w cell where a bead column is packed, a uidics system that controls the o w of buffer, samples and secondary uorescent probe through the o w cell, and an epi-illumination uorescence detection system to measure uorescence from the o w cell (Darling & Brault, 2004). Details of the KinExA instrument performance have been described elsewhere (Blake, Pavlov, & Blake, 1999; Blake et al., 2001; Khosraviani et al., 2000; Kim, Shelver, Hwang, Xu, & Li, 2006; Ohmura, Tsukidate, Shinozaki, Lackie, & Saiki, 2003; PrietoSimn, Miyachi, Karube, & Saiki, 2010). Briey, the capillary o w cell of the automated o w immunoassay system is tted with a microporous screen where beads are packed acting as the solid phase. Solutions are drawn through the o w cell due to the negative pressure created by a syringe pump. Equilibrated mixtures of a xed anti-OTA concentration and different OTA concentrations (from 0.75 ng L 1 to 0.1 mg L 1 ) were drawn through the OTAmodied beads to accumulate unbound antibody. A nal labelling step with a 1 mg L 1 Cy5-conjugated IgG secondary antibody (against the source of the primary antibody) solution allowed the uorescence detection (Fig. 1). Control of the pumps and valves was achieved by a computer interface and the KinExA Pro software package. 2.7. Modication of PMMA beads PMMA beads were coated with OTA BSA conjugate or biotinylated OTA depending on the approach. Approximately 220 mg 100 lm diameter PMMA beads were suspended in 1 mL of 20 mg L 1 OTA BSA conjugate in PBS and rotated overnight. For biotinylated OTA- modied beads, rstly beads were suspended in 1 mL PBS containing 0.5 mg of avidin and rotated overnight. Avidin- modied beads were then re-suspended in 1 mL of a 20 mg L 1 biotinylated OTA solution in PBS and rotated for 4 h. After settling and removal of supernatant, blocking with a 10 g L 1 BSA solution was performed to minimise non- specic adsorptions. Coated beads were stored at 4 C and diluted to 27 mL with PBS prior to use.

3. Results and discussion Most of the analytical devices developed to detect OTA are based on immunoassay strategies. Due to the low-molecular weight of OTA, its direct detection by simple binding to the antiOTA antibody is not easily performed. Thus competitive formats are required. Two indirect competitive uoroimmunoassays were compared. Both assays differ on the solid supports used: polystyrene microtiter plates and PMMA beads. Both supports were modied with the same OTA conjugates: OTA BSA conjugate and biotinylated OTA. Fluoroimmunoassays were set up to show their ability to reduce the LODs below those previously found for ELISA (Alarcn et al., 2006; Prieto-Simn et al., 2008). The optimised uoroimmunoassay approach was used to develop an automated o w-imm u nosensing system as a powerful tool for OTA assessment in beverages and food samples. 3.1. Indirect competitive uoroimmunoassays A checkerboard assay conrmed the optimal working conditions previously found for ELISA assays (Prieto-Simn et al., 2008). Briey, saturated coating of microtiter wells was achieved using OTA-BSA at 20 mg L 1 or a 20 mg L 1 biotinylated OTA solution, the latter after modication with a 1 mg L 1 avidin solution. The optimum concentration of MAb in the competition step was 0.1 mg L 1 . The optimum dilution of the FITC-conjugated antimouse secondary antibody was 1:500. Negligible non- specic adsorption of the FITC-conjugates on bare wells was achieved thanks to the blocking and rinsing steps. Table 1 summarises the results for the non-linear four-parameter logistic regression tting of the plots obtained for both indirect competitive uoroimmunoassays, as well as the results previously published for equivalent indirect competitive ELISA. IC50 and LOD values derived from the regression equation are also included. The OTA biotin conjugate approach showed a 50% inhibition of binding of 5 lg L 1 , a value half of that obtained by the OTA BSA 1 conjugate-based assay (IC 50 = 10 lg L ). The OTA BSA approach showed a shorter linear working range and a higher LOD than that obtained by the OTA biotin approach. These results agree with those reported for ELISA. OTA was covalently linked to biotin using a long spacer that minimises the steric effects. Thus OTA was better exposed than for immobilised OTA BSA. Furthermore, as expected, uorescent assays showed higher sensitivity and lower detection limits than previously reported colorimetric assays (Prieto-Simn et al., 2008). 3.2. Indirect competitive KinExA measurements KinExA measurements for OTA detection were based on indirect competitive immunoassays, equivalent to those previously described for uoroimmunoassays. Two approaches were studied depending on the OTA conjugate immobilised on the PMMA beads packed as solid phase. Fig. 1 shows the sequence of steps for both approaches. Working conditions (ow rates and injected volumes and times) were set to provide a contact time lower than 0.5 s. The contact time must be short enough to capture a portion of the free antibody in solution and ensure negligible dissociation. Dissociation of the soluble antibody-toxin complex is kinetically excluded. This rapid separation prevents shift in the equilibrium of the aqueous phase binding. Furthermore, optimisation of the working conditions also involved obtaining approximately a signal of 1.0 V for the relative signal difference (RSD). The RSD is the difference between the signal 100% (signal produced when 100% of the binding

1326

B. Prieto-Simn et al. / Food Chemistry 135 (2012) 1323 1329

(A) OTA-BSA

OTA anti-OTA MAb

Cy5-IgG

(B) OTA-biotin Avidin

(a) Packed beads

(b) Flow equilibrated mixtures of anti-OTA

(c) Labelling with anti-mouse Cy5-IgG

Fig. 1. Schematic diagram depicting (A) OTA BSA- and (B) biotinylated OTA-based ow-immunosensing systems. (a) Fresh column of OTA- modied PMMA beads; (b) Equilibrated mixtures of anti-OTA and different OTA concentrations owing through the column; (c) Anti-mouse Cy5 IgG solution owing through the column to label the previously captured antibody against OTA. After removing excess of label by owing buffer, uorescent measurement is performed.

Table 1 Curve parameters derived from the non-linear four-parameter logistic regression tting of the plots obtained for indirect competitive uoroimmunoassay and ELISA measurements (n = 3). Strategy Biotinylated OTA/IgG-ALP/ELISA (Shim et al., 2009a) Biotinlyated OTA/IgG-HRP/ELISA (Shim et al., 2009a) Biotinylated OTA/immunouoroassay OTA-BSA/immunouoroassay
a

IC50 ( lg L 14 41 5 10

LOD ( lg L 4.8 9.9 0.2 3.7

SD (%) 5.0 3.8 4.4 4.8

Sigmoidal logistic equation % binding 0:5 % binding 0:1 % binding 20 % binding 11

148 10 :006 10 :05
OTA 0 :9

R 0.9925 0.9996 0.9994 0.9995

OTA 1 :1

102 OTA 0:4 10 :000 94 4

56 OTA 1:1 10 :02

SD values refer to the maximum SD values found in the calibration curve.

sites are free at a given MAb concentration) and the nonspecic binding (NSB, background binding for samples without antibody). KinExA uses the solid phase as a tool to capture the antibody, differing from other heterogeneous methods, such as uoroimmunoassays, ELISA or SPR. An excess of antigen was used to modify the micro beads as a way to avoid mass transport limitations and mobility effects. However, results suggested that the steric effects associated to OTA BSA conjugate could not ensure an excess of OTA on PMMA beads. Results in Table 2 show that IC50 values decreased as the antiOTA dilution increased. This trend led to minimum values for IC50 and LOD using an antibody concentration equal to or lower 10 than the equilibrium dissociation constant (K d ), 1.28 10 M (1:50,000 anti-OTA dilution). Results agreed with the KinExA theory, stating that the lowest detectable concentrations of target analyte correspond to antibody concentrations equal to or lower

than the K d . Usually the sensitivity of the detection method limits the ability to work under these conditions. However, KinExA can achieve the theoretical limits of detectability mainly due to two reasons: (1) KinExA favours the continuous accumulation of free antibody onto the beads. (2) During the measurement, there is only equilibrium between antigen and antibody in solution. Other reactions are kinetically excluded. Moreover, binding events take place between unmodied molecules in solution, avoiding dependence on how immobilisation or labelling alter native binding reactions (Glass, Ohmura, & Saiki, 2007; Ohmura, Lackie, & Saiki, 2001; Su, Endo, Saiki, Xing, & Ohmura, 2007). Additionally, KinExA relies on a highly sensitive detection method, such as uorescence. As a result of the advantages of KinExA measurements, very low LODs were achieved. Biotinylated OTA-based assay could detect 12 ng L 1 (considered as 80% binding), using a 1:50,000 dilution of anti-OTA. Moreover, the assay showed very good reproducibility (maximum standard deviation of 3.8%).

Table 2 Curve parameters derived from the non-linear four-parameter logistic regression tting of the plots obtained for OTABSA- and biotinylated OTA-based indirect competitive KinExA measurements using different antibody dilutions (n = 3). Strategy/anti-OTA dilution OTA BSA/1:1000 OTA BSA/1:5000 Biotinylated OTA/1:5000 Biotinylated OTA/1:50,000
a

IC50 ( lg L 0.83 0.30 0.19 0.10

LOD ( lg L 0.17 0.15 0.06 0.01

SD (%) 7.0 3.4 3.2 3.8

Sigmoidal logistic equation % binding 6 % binding 0:1 % binding 6 % binding 6

10OTA :000 7 95
0:9

R 0.9996 0.9973

10 :000 3 10 :00099 2
OTA 1 :4

OTA

2 :1

0.9987 0.9990

86 0:9 10OTA :000 1 86

SD values refer to the maximum SD values found in the calibration curve.

B. Prieto-Simn et al. / Food Chemistry 135 (2012) 1323 1329

1327

3.3. Indirect competitive KinExA measurements for OTA detection in cereals and wine samples Once the ability of KinExA to detect OTA in buffered solutions was proven, our purpose was to use the optimised assay to determine OTA in more complex samples. OTA is ubiquously found in a wide range of food commodities. It was rst reported in corn. Since then, most OTA surveys have been carried out in cereals. We selected two samples: cereals, as an important food commodity, and wine, as a beverage widely exported. In fact, OTA contamination is not only a problem for the warmer areas of the world, but also for importing countries. Moreover, cereals and wine were chosen as examples of solid and liquid samples that follow different pre-treatments prior to OTA analysis. Wine samples were simply pre-treated with PVP to remove polyphenols and subsequently pH neutralised. Polyphenols content must be minimised to avoid antibody inactivation and interfering effects to uorescent measurements. Corn-akes samples followed an extraction procedure with a methanolic solution. Evaporation to dryness and nal reconstitution in PBS buffer were necessary to avoid the inhibition of the antibody-antigen binding caused by methanol. Preliminary assays showed high NSB signals for wine samples. Unlike other immunoassays, the protocol for KinExA measurements also involves the presence of wine at the detection step. Thus the presence of certain uorescent compounds in wine could cause an interfering effect. Consequently, wine samples were diluted. Percentages of slope deviation between calibration curves in PBS and in several diluted wine samples (1:4, 1:10 and 1:100) were lower than 7% and 5% for OTA BSA and biotinylated OTA approaches respectively. Therefore, no signicant interfering effects could be observed from diluted wine samples. Fig. 2 shows the non-linear tting of the plots generated by KinExA assays with OTA dilutions prepared in buffer or in diluted wine samples. Results show good performance of OTA BSA and biotinylated OTA-based approaches for 1:10 and 1:100 diluted wine samples respectively. IC50 values for calibration curves in PBS and in diluted wine samples were very similar using the same approach and anti-OTA dilution (299 and 296 ng L 1 when working with PBS or 1:10 diluted wine samples respectively for the

OTA BSA approach; 96 and 112 ng L 1 when working with PBS or 1:100 diluted wine samples respectively for the biotinylated OTA approach). Fig. 2 also includes the calibration curve for a 1:4 diluted wine sample using the OTA BSA approach, corresponding to a 1:2000 anti-OTA dilution. The percentage of slope deviation between this calibration curve and that of the samples prepared in PBS is lower than 4%, showing no signicant matrix effects. 1 The IC50 value (550 ng L ) was lower than that found for a 1:1000 anti-OTA dilution in PBS, but not as low as that found for a 1:5000 anti-OTA dilution (Table 2). It is interesting to notice that using different antibody and/or sample dilutions the working range, and thus the LOD, can be shifted. The system can be initially used for a preliminary screening. Then, working conditions can be changed to modify working range and LOD for a more accurate detection of OTA. After evaluation of matrix effects, recovery studies were performed using aliquots of wine spiked before pre-treatment with OTA at 3, 2 and 1 lg L 1 . Table 3 summarises the percentages of OTA recovered working with 1:10 and 1:100 wine dilutions for the OTA BSA and biotinylated OTA approaches respectively. Results are in good agreement with the amounts of OTA spiked before pre-treatment for OTA concentrations equal to or higher than the regulatory limit set by the European Commission (2 lg L 1 ) (European Mycotoxin Awareness Network (EMAN), 2012). Thus, the system is suitable for simply and rapidly assess OTA in real samples. Nevertheless, wine samples spiked with lower OTA amounts show overestimated recoveries (115% and 123% for OTA BSA- and biotinylated OTA-based approaches), with high relative standard deviation values (20% and 36%, respectively). The suggested hypothesis to explain these results refers to the high percentages of binding associated to low OTA concentrations, close to the end of the linear working range. As previously mentioned, results in Table 3 show that choosing the proper anti-OTA and sample dilutions, the working range and LOD can be shifted (e.g. a 1:50,000 anti-OTA dilution using biotinylated OTA approach enables a reliable determination of an OTA concentration one order of magnitude lower than using a 1:5000 anti-OTA dilution with the OTA BSA approach). We have shown that anti-OTA dilution should be carefully optimised, enabling the detection of the theoretical lowest LOD for a certain antibody.

(A)
120 100 100 80 80 60 PBS buffer / 1:5000 anti-OTA dilution -1 IC50= 299 ng L Wine samples, dilution 1:4 / 1:2000 anti-OTA dilution -1 IC50= 550 ng L Wine samples, dilution 1:10 / 1:5000 anti-OTA dilution -1 IC50= 296 ng L 0,001 0,01 0,1 1
-1

(B)

%binding

40

%binding

60

40 PBS buffer / 1:50000 anti-OTA dilution -1 IC50= 96 ng L Wine samples, dilution 1:100 / 1:50000 anti-OTA dilution -1 IC50= 112 ng L

20

20

-20 10

0 0,0001

0,001

0,01

0,1
-1

10

log([OTA] (g L )

log([OTA] (g L )

Fig. 2. Calibration curves obtained for OTA in PBS and in diluted wine samples using competitive KinExA measurements: (A) OTA BSA- and (B) biotinylated OTA-based approaches. Error bars are standard deviations of the mean with n = 3.

1328

B. Prieto-Simn et al. / Food Chemistry 135 (2012) 1323 1329

Table 3 OTA recovery percentages from spiked wine and cereal samples using competitive KinExA measurements. Spiked OTA ( lg L1 ) OTA BSA strategy/1:5000 anti-OTA dilution [OTA] in the assay solution ( lg L1 ) a Wine samples 3 2 1 0.3 0.2 0.1 % binding 51 3 68 3 88 4 72 3 79 3 68 2 [OTA] detected ( lg L1 ) 0.29 0.02 0.21 0.01 0.12 0.02 0.19 0.01 0.16 0.01 0.21 0.01 % recovered 98 6 105 7 115 20 95 7 106 9 105 4 Biotinylated OTA strategy/1:50,000 anti-OTA dilution [OTA] in the assay solution ( lg L1 ) a 0.03 0.02 0.01 0.04 0.03 0.02 % binding 70 1 75 1 80 3 66 2 70 1 75 1 [OTA] detected ( lg L1 ) 0.029 0.002 0.020 0.002 0.012 0.004 0.038 0.005 0.029 0.002 0.020 0.002 % recovered 97 7 99 9 123 36 96 13 97 7 99 9

Corn-akes samples 4 0.2 3 0.15 2 0.2

a

A dilution step was performed to obtain percentages of binding within the working range.

Moreover, the high sensitivity of KinExA-based immunosensors often requires the dilution of samples to t them in the working range of the calibration curves. Thus dilution can be the key step to allow accurate and reliable results. As a result, anti-OTA and sample dilutions can be selected depending on the sensitivity required for each sample to analyse, offering high versatility to the method. Corn-akes samples spiked before extraction were used to perform recovery studies. These studies showed excellent extraction efciency. Results included in Table 3 show that KinExA can accurately measure the concentration of OTA in spiked corn-akes samples, yielding average recoveries from 95% to 106% for both OTA BSA and biotinylated OTA approaches and different spiked OTA amounts (from 2 to 4 lg kg1 ). Results show this technique as a powerful analysis tool able to detect OTA in cereal samples under the limit set by the European Commission (3 lg kg1 for processed cereal products for direct human consumption) (European Mycotoxin Awareness Network, 2012). 4. Conclusions Two uorescence-based immunosensing strategies have been set up and optimised to detect OTA at the limits set by the European legislation. The optimised uoroimmunoassay showed LOD values lower than those previously reported for equivalent colorimetric ELISA. However, the assay involved several steps, being a time-consuming method. Better performance characteristics were achieved using a ow-immunosensing system based on KinExA measurements. KinExA-based assays were able to detect the theoretical 1 lowest LOD set by the afnity of the antibody (IC 80 = 12 ng L in the assay solution). Moreover, each sample measurement could be performed in less than 10 min in an almost automated form. To the best of our knowledge, this is the rst time that KinExA has been used to determine OTA. We have used the optimised KinExA protocol to analyse OTA- fortied wine and cereal samples. The system is shown as a versatile tool: after an initial screening, optimisation of sample and anti-OTA dilutions tailors the LOD and working range. Therefore, OTA quantication can be performed with high accuracy. The developed ow-immunosensing strategies showed KinExA as a suitable bioanalytical tool to simply and rapidly assess OTA in food and beverages. Its simplicity and high sensitivity prove its capacity to perform as an accurate complementary tool to less sensitive techniques for toxin screening. Acknowledgements Dr. B. Prieto-Simn is grateful to the Japan Society for the Promotion of Science for a fellowship supporting her research at the

Tokyo University of Technology. She also acknowledges the Ministerio de Ciencia e Innovacin for a Juan de la Cierva fellowship. References
Adnyi, N., Levkovets, I. A., Rodriguez-Gil, S., Ronald, A., Vradi, M., & Szendro, I. (2007). Development of immunosensor based on OWLS technique for determining aatoxin B1 and ochratoxin A. Biosensors & Bioelectronics, 22, 797 802. Alarcn, S. H., Palleschi, G., Compagnone, D., Pascale, M., Visconti, A., & Barna-Vetr, I. (2006). Monoclonal antibody based electrochemical immunosensor for the determination of ochratoxin A in wheat. Talanta, 69, 1031 1037. Bazin, I., Nabais, E., & Lopez-Ferber, M. (2010). Rapid visual tests: Fast and reliable detection of ochratoxin A. Toxins, 2, 2230 2241. Blake, D. A., Jones, R. M., Blake, R. C., II, Pavlov, A. R., Darwish, I. A., & Yu, H. (2001). Antibody-based sensors for heavy metal ions. Biosensors & Bioelectronics, 16, 799 809. Blake, R. C., II, Pavlov, A. R., & Blake, D. A. (1999). Automated kinetic exclusion assays to quantify protein binding interactions in homogeneous solution. Analytical Biochemistry, 272, 123 134. Carter, R. M., Blake, R. C., II, Mayer, H. P., Echevarria, A. A., Nguyen, T. D., & Bostanian, L. A. (2000). A uorescent biosensor for the detection of zearalenone. Analytical Letters, 33, 405 412. Commission Regulation (EC) No. 123/2005 of 26 January 2005 amending Regulation (EC) No. 466/2001 as regards ochratoxin A. Ofcial Journal of the European Communities L 25/3 (current as of January 2012 at http://eur-lex.europa.eu/ LexUriServ/LexUriServ.do?uri=OJ:L:2005:025:0003:0005:EN:PDF). Darling, R. J., & Brault, P.-A. (2004). Kinetic exclusion assay technology: characterization of molecular interactions. Assay and Drug Development Technologies, 2, 647 657. European Mycotoxin Awareness Network (EMAN). Accessed 9 January 2012 http:// www.mycotoxins.com/. Fink-Gremmels, J. (1999). Mycotoxins: their implications for human and animal health. Veterinary Quarterly, 21, 115 120. Flow-through rapid test kit for ochratoxin A in wine (Euro-Diagnostica). http:// www.elisa-tek.com/wp-content/themes/elisatek/documents/Ochratoxin%20AWine%20FTRT.pdf Accessed 9 January 2012. Giray, B., Atasayar, S., & Sahin, G. (2009). Determination of ochratoxin A and total aatoxin levels in corn samples from Turkey by enzyme-linked immunosorbent assay. Mycotoxin Research, 25, 113 116. Glass, T. R., Ohmura, N., & Saiki, H. (2007). Least detectable concentration and dynamic range of three immunoassay systems using the same antibody. Analytical Chemistry, 79, 1954 1960. Goryacheva, I. Y., Basova, E. Y., Van Peteghem, C., Eremin, S. A., Pussemier, L., Motte, J.-C., et al. (2008). Novel gel-based rapid test for non-instrumental detection of ochratoxin A in beer. Analytical and Bioanalytical Chemistry, 390, 723 727. Heurich, M., Kadir, M. K. A., & Tothill, I. E. (2011). An electrochemical sensor based on carboxymethylated dextran modied gold surface for ochratoxin A analysis. Sensors and Actuators B: Chemical, 156, 162 168. Khan, R., & Dhayal, M. (2008). Nanocrystalline bioactive TiO2 chitosan impedimetric immunosensor for ochratoxin-A. Electrochem. Commun., 10, 492 495. Khan, R., & Dhayal, M. (2009). Chitosan/polyaniline hybrid conducting biopolymer base impedimetric immunosensor to detect ochratoxin-A. Biosensors & Bioelectronics, 24, 1700 1705. Khosraviani, M., Blake, R. C., II, Pavlov, A. R., Lorbach, S. C., Yu, H., et al. (2000). Binding properties of a monoclonal antibody directed toward lead-chelate complexes. Bioconjugate Chemistry, 11, 267 277. Kim, H.-J., Shelver, W. L., Hwang, E.-C., Xu, T., & Li, Q. X. (2006). Automated ow uorescent immunoassay for part per trillion detection of the neonicotinoid insecticide thiamethoxam. Analytica Chimica Acta, 571, 66 73. Kolosova, A. Y., Saeger, S. D., Eremin, S. A., & Van Peteghem, C. (2007). Investigation of several parameters inuencing signal generation in ow-through membranebased enzyme immunoassay. Analytical and Bioanalytical Chemistry, 387, 1095 1104.

B. Prieto-Simn et al. / Food Chemistry 135 (2012) 1323 1329 Krogh, P. (1987). Ochratoxins in food. In P. Krogh (Ed.), Mycotoxins in Food (p. 97). London: Academic Press. Liu, X.-P., Deng, Y.-J., Jin, X.-Y., Chen, L.-G., Jiang, J.-H., Shen, G.-L., et al. (2009). Ultrasensitive electrochemical immunosensor for ochratoxin A using gold colloid-mediated hapten immobilization. Analytical Biochemistry, 389, 63 68. Ngundi, M. M., Shriver-Lake, L. C., Moore, M. H., Lassman, M. E., Ligler, F. S., & Taitt, C. R. (2005). Array biosensor for detection of ochratoxin A in cereals and beverages. Analytical Chemistry, 77, 148 154. Ngundi, M. M., Shriver-Lake, L. C., Moore, M. H., Ligler, F. S., & Taitt, C. R. (2006). Multiplexed detection of mycotoxins in foods with a regenerable array. Journal of Food Protection, 69, 3047 3051. Ohmura, N., Lackie, S. J., & Saiki, H. (2001). An immunoassay for small analytes with theoretical detection limits. Analytical Chemistry, 73, 3392 3399. Ohmura, N., Tsukidate, Y., Shinozaki, H., Lackie, S. J., & Saiki, H. (2003). Combinational use of antibody afnities in an immunoassay for extension of dynamic range and detection of multiple analytes. Analytical Chemistry, 75, 104 110. Pfohl-Leszkowicz, A., & Manderville, R. A. (2007). Review on Ochratoxin A: An overview on toxicity and carcinogenicity in animals and humans. Molecular Nutrition & Food Research, 51, 61 99. Pittet, A. (2005). Modern methods and trends in mycotoxin analysis. Mitt. Lebensm. Hyg., 96, 424 444. Prieto-Simn, B., Camps, M., Marty, J.-L., & Noguer, T. (2008). Novel highlyperforming immunosensor-based strategy for ochratoxin A detection in wine samples. Biosensors & Bioelectronics, 23, 995 1002. Prieto-Simn, B., Miyachi, H., Karube, I., & Saiki, H. (2010). High-sensitive owbased kinetic exclusion assay for okadaic acid assessment in shellsh samples. Biosensors & Bioelectronics, 25, 1395 1401. Prieto-Simn, B., Noguer, T., & Camps, M. (2007). Emerging biotools for assessment of mycotoxins in the past decade. Trends in Analytical Chemistry, 26, 689 702. Radi, A.-E., Muoz-Berbel, X., Cortina-Puig, M., & Marty, J.-L. (2009). An electrochemical immunosensor for ochratoxin A based on immobilization of antibodies on diazonium-functionalized gold electrode. Electrochimica Acta, 54, 2180 2184. Radi, A.-E., Muoz-Berbel, X., Lates, V., & Marty, J.-L. (2009). Label free impedimetric immunosensor for sensitive detection of ochratoxin A. Biosensors & Bioelectronics, 24, 1888 1892. Radoi, A., Dumitru, L., Barthelmebs, L., & Marty, J.-L. (2009). Ochratoxin A in some French wines: Application of a direct competitive ELISA based on an OTA-HRP conjugate. Analytical Letters, 42, 1187 1202. Richard, J. L., & Payne, G. A. (2003). Mycotoxins: Risks in Plant, Animal and Human Systems. Raleigh: CAST Task Force Report.

1329

Santos, E. A., & Vargas, E. A. (2002). Immunoafnity column clean-up and thin layer chromatography for determination of ochratoxin A in green coffee.. Food Additives & Contaminants, 19, 447 458. Sapsford, K. E., Ngundi, M. M., Moore, M. H., Lassman, M. E., Shriver-Lake, L. C., Taitt, C. R., et al. (2006). Rapid detection of foodborne contaminants using an array biosensor. Sensors and Actuators B: Chemical, 113, 599 607. Schneider, E., Curtui, V., Seidler, C., Dietrich, R., Usleber, E., & Mrtlbauer, E. (2004). Rapid methods for deoxynivalenol and other trichothecenes. Toxicology Letters, 153, 113 121. Shim, W.-B., Dzantiev, B. B., Eremin, S. A., & Chung, D.-H. (2009a). One-step simultaneous immunochromatographic strip test for multianalysis of ochratoxin A and zearalenone. Journal of Microbiology and Biotechnology, 19, 83 92. Shim, W.-B., Kim, G., Ryu, H.-J., Nam, M., & Chung, D.-H. (2009b). Development of one-step simultaneous immunochromatographic assay for rapid analysis of aatoxin B1 and ochratoxin A. Food Science and Biotechnology, 18, 641 648. Strachan, N. J. C., John, P. G., & Miller, I. G. (1997). Application of an automated particlebased immunosensor for the detection of aatoxin B1 in foods. Food and Agricultural Immunology, 9, 177 184. Su, F.-Y., Endo, Y., Saiki, H., Xing, X.-H., & Ohmura, N. (2007). Simple and sensitive bacterial quantication by a ow-based kinetic exclusion uorescence immunoassay. Biosensors & Bioelectronics, 22, 2500 2507. Tsai, W. C., & Hsieh, C.-K. (2007). QCM-based immunosensor for the detection of ochratoxin A. Analytical Letters, 40, 1979 1991. Urusov, A. E., Kostenko, S. N., Sveshnikov, P. G., Zherdev, A. V., & Dzantiev, B. B. (2011). Ochratoxin A immunoassay with surface plasmon resonance registration: Lowering limit of detection by the use of colloidal gold immunoconjugates. Sensors and Actuators B: Chemical, 156, 343 349. Vidal, J. C., Duato, P., Bonel, L., & Castillo, J. R. (2009). Use of polyclonal antibodies to ochratoxin A with a quartz-crystal microbalance for developing real-time mycotoxin piezoelectric immunosensors. Analytical and Bioanalytical Chemistry, 394, 575 582. Xiao, H., Clarke, J. R., Marquardt, R. R., & Frohlich, A. A. (1995). Improved methods for conjugating selected mycotoxins to carrier proteins and dextran for immunoassays. Journal of Agriculture and Food Chemistry, 43, 2092 2097. Yuan, J., Deng, D., Lauren, D. R., Aguilar, M.-I., & Wu, Y. (2009). Surface plasmon resonance biosensor for the detection of ochratoxin A in cereals and beverages. Analytica Chimica Acta, 656, 63 71. Zamr, L.-G., Geana, I., Bourigua, S., Rotariu, L., Bala, C., Errachid, A., et al. (2011). Highly sensitive label-free immunosensor for ochratoxin A based on functionalized magnetic nanoparticles and EIS/SPR detection. Sensors and Actuators B: Chemical, 159, 178 184.