Comparative study of colloidal gold and quantum dots as labels for multiplex screening tests for multi-mycotoxin detection

Analytica Chimica Acta 955 (2017) 48e57

Contents lists available at ScienceDirect

Analytica Chimica Acta journal homepage: www.elsevier.com/locate/aca

Comparative study of colloidal gold and quantum dots as labels for multiplex screening tests for multi-mycotoxin detection Astrid Foubert*, Natalia V. Beloglazova, Sarah De Saeger Faculty of Pharmaceutical Sciences, Laboratory of Food Analysis, Ghent University, Ghent, Belgium

h i g h l i g h t s

g r a p h i c a l a b s t r a c t


Development of colloidal gold- and quantum dot-based multiplex lateral flow immunoassay.
Lateral flow immunoassays allow simultaneous detection of four mycotoxins.
In-use comparison in use of immunoreagents and performance characteristics.
Label-depended use of different materials and blocking solutions.
Validation according to European legislation allows qualitative detection.

a r t i c l e i n f o

a b s t r a c t

Article history: Received 26 August 2016 Received in revised form 13 November 2016 Accepted 18 November 2016 Available online 28 November 2016

Quantum dots (QDs) and colloidal gold nanoparticles (CG) were evaluated as labels for multiplex lateral flow immunoassay (LFIA) for determination of mycotoxins deoxynivalenol (DON), zearalenone (ZEN) and T2/HT2-toxin (T2/HT2) in cereal matrices. Both developed assays were based on the same immunoreagents (except for the labels), therefore their analytical characteristics could be objectively compared. For both LFIAs antigens (DON-ovalbumin (OVA), ZEN-OVA and T2-OVA) and rabbit anti-mouse immunoglobulin were immobilized on a nitrocellulose membrane as three test lines and one control line, respectively. Depending on the LFIA, monoclonal antibodies (mAb) against DON, ZEN and T2 were conjugated with CdSeS/ZnS QDs or CG. T2 and HT2 were detected by one test line (T2-OVA) with an antiT2 mAb which showed 110% cross-reactivity with HT2. Both tests were developed in accordance with the legal limits and were developed in such a way that they had the same cut-off limits of 1000 mg kg1, 80 mg kg1 and 80 mg kg1 for DON, ZEN and T2/HT2, respectively in order to allow a correct comparison. Applicability of these assays was demonstrated by analysis of naturally contaminated wheat samples. The results demonstrate that both the LFIAs can be used as rapid, cost-effective and convenient qualitative tool for on-site screening for simultaneous detection of DON, ZEN and HT2/T2 in wheat without special instrumentation. However, the QD-based LFIA consumed less immunoreagents and was more sensitive

Keywords: Quantum dots Colloid gold nanoparticle Bioconjugation Multiplex lateral flow immunoassay Mycotoxin

Abbreviations: BSA, bovine serum albumin; CDI, 1,10 -carbonyldiimidazole; CG, colloidal gold; CG-LFIA, CG-based LFIAs; CG-mAbs, CG-labelled mAbs; CMO, O-(carboxymethyl)hydroxylamine hemichloride; DCC, N,N0 -dicyclohexyl carbodiimide; DLS, dynamic light scattering; DMF, dimethylformamide; DMSO, dimethyl sulfoxide; DON, deoxynivalenol; EDC, 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride; ELISA, enzyme-linked immunosorbent assays; ESI, electrospray ionization; HF, HiFlow; HRP, horseradish peroxidase; HT2, HT2-toxin; LFIA, lateral flow immunoassays; LOD, limit of detection; mAb, monoclonal antibody; MRM, multiple reaction monitoring; NHS, N-hydroxysuccinimide; OTA, ochratoxin; OVA, ovalbumin; PBS, phosphate buffered saline; PBST, PBS þ Tween 20 0.05% (v/v); POD, probability of detection; QD, quantum dot; QD-mAbs, QD-labelled mAbs; RT, room temperature; SRM, selected reaction monitoring; sulfo-NHS, N-hydroxysulfosuccinimide; TBSU, Tris buffer containing 1% BSA and 1% sucrose; T2, T2-toxin; TMB, 3, 30 , 5, 50 -tetramethylbenzidine; ZEN, zearalenone. * Corresponding author. E-mail addresses: [email protected], [email protected] (A. Foubert). http://dx.doi.org/10.1016/j.aca.2016.11.042 0003-2670/© 2016 Elsevier B.V. All rights reserved.

A. Foubert et al. / Analytica Chimica Acta 955 (2017) 48e57

49

and economically beneficial. In addition, the results were easier to interpret, resulting in a lower false negative rate (<5%) which was in good agreement with Commission Decision 2002/657/EC regarding the performance of analytical methods intended for screening purposes. © 2016 Elsevier B.V. All rights reserved.

1. Introduction Initially, rapid screening tests found a wide application in clinical diagnostics with the pregnancy test as one of the most known and first commercialized example. The ease-of-use and rapid completion of these assays, together with the high sensitivity, selectivity, and accuracy, increased their appeal for field applications in food analysis and food control outside the controlled laboratory environment. Therefore, a huge variety of rapid immunochemical tests for detection of pathogens [1], allergens [2], drug residues [3] and mycotoxins [4] were developed and commercialized. Mycotoxins are common contaminants of many grains like wheat, barley, maize, and rice, and they can evoke a broad range of toxic properties including carcinogenicity, neurotoxicity, as well as reproductive and developmental toxicity [5,6]. On a global level, more than 30% of food and feed samples are co-contaminated [7]. For example, Fusarium mycotoxins, like deoxynivalenol (DON), zearalenone (ZEN), T2 and HT2-toxin (T2 and HT2), are widely distributed in the food chain in the EU and worldwide [8]. Important economic and trade implications arise from mycotoxin contamination. The economic costs are difficult to assess but quantitative estimates of economic losses associated with mycotoxin contamination may range from millions to billions of US $ annually [7]. Aside from economic implications, mycotoxins also have a significant impact on human and animal health. As a consequence, the presence of mycotoxins in foodstuffs is a major concern for food safety. Despite efforts to control fungal contamination, toxigenic fungi are omnipresent in nature and occur regularly in worldwide food supplies due to mold infestation of susceptible agricultural products [9]. Therefore the control of mycotoxin levels in food and feed has become an important objective for producers, regulatory authorities, and researchers worldwide. This, in combination with the co-occurrence of mycotoxins demands for rapid multiplex assays. Lateral flow immunoassays (LFIAs) are suitable for this purpose because they provide rapid on-site simultaneous screening of multiple mycotoxins in different matrices. In addition, these LFIAs can be performed by non-specialists and enable to take quick corrective actions when needed. However, LFIAs are considered to be less sensitive in comparison with assays that need instruments, like e.g. enzymelinked immunosorbent assays (ELISAs) [4] or biosensors [10]. An important part that contributes to the LFIA's sensitivity is the signal reporter/label that is used to enable a visible detection. Despite the fact that different nanoparticles have been used as label in LFIA, such as carbon nanoparticles [11], fluorescent dyes [12], liposomes [13], dye-doped nanoparticles [3] and magnetic nanoparticles [14], colloidal gold (CG) nanoparticles remain the most widely used label due to its simple and rapid way of synthesis, low cost and easy interpretation of the obtained analytical result [15]. In terms of sensitivity, the use of CG nanoparticles in a LFIA resulted in a slightly better result compared with carbon nanoparticles [16]. Several attempts have been made to compare CG-based LFIAs (CGLFIA) with fluorescent-based tests and this showed a higher sensitivity of the later mentioned device [17,18]. However, in the studies the CG-LFIAs were evaluated visually while the fluorescent

based LFIAs were instrumentally evaluated. This visual evaluation probably resulted in an underestimation of the sensitivity and did not allow a correct comparison of both tests. Quantum dots (QDs), semiconductor nanoparticles, are characterized by unique size-tunable optical properties which favor their use for biomedical diagnostics, chemical and therapeutic labeling, imaging and chemical analysis. They possess a broad absorbance band and a narrow size-dependent symmetrical shaped emission peak, a high fluorescence quantum yield, stability against photobleaching, an increased brightness and a higher signal-to-noise ratio compared with traditional organic fluorophores [19]. Beside this, QDs emitting in a different part of the spectrum (e.g. QDs with different core size) can be simultaneously excited with a single wavelength and can, because of their narrow emission peak, be combined in one single test without spectral overlap. This dramatically simplifies the realization of multiplex analysis [20]. In addition, by using QDs that emit different colors on a single strip it is easier to distinguish the results between the different analytes. QD have been exploited to prepare QD-labelled aptamers for the development of a LFIA for the detection of ochratoxin (OTA). The visual limit of detection (LOD) was set at 5 ng mL1, which was comparable with the LOD of a reported antibody-based CG method [21]. Recently, a QD submicrobead-based LFIA for quantitative detection of ZEN was developed. Here, numerous QDs were embedded in a polymer matrix in order to improve sensitivity of the LFIA due to their stronger fluorescence. They reported a LOD of 0.0625 ng mL1 for ZEN, which was 5.6 times lower than a previously reported quantitative CG-LFIA [22]. However, both tests were not developed with the same immunoreagents which makes a correct comparison impossible. Here, we developed a tricolor-QD and CG-based LFIA system for the detection of the mycotoxins DON, ZEN, T2 and HT2. By using the same immunoreagents, but labelled differently, a direct and correct comparison of development and performance characteristics was possible. To the best of our knowledge this is the first manuscript developing a multiplex tricolor QD-based LFIA for detection of mycotoxins in combination with a thorough comparison with a CGbased LFIA. 2. Materials and methods 2.1. Materials and reagents The mycotoxins ZEN and DON were purchased from Fermentek (Jerusalem, Israel). T2 and HT2 were obtained from Sigma Aldrich (Diegem, Belgium). 1-ethyl-3-(3-dimethylaminopropyl) carbodiimide hydrochloride (EDC), N-hydroxysulfosuccinimide (sulfoNHS), N,N0 -dicyclohexyl carbodiimide (DCC), N-hydroxysuccinimide (NHS), O-(carboxymethyl)hydroxylamine hemichloride (CMO), 1,10 -carbonyldiimidazole (CDI), phosphate buffered saline (PBS) pH 7.4, Tween 20, bovine serum albumin (BSA), ovalbumin (OVA), 2-(N-morpholino)ethanesulfonic acid sodium salt, TRIS-borate-EDTA buffer, sucrose, chloroauric acid, sodium citrate, sodium borohydride, sodium chloride, rabbit anti-mouse IgG secondary antibody labelled with horseradish peroxidase (HRP), casein, carbonate buffered saline (CBS, pH 0.05 M, pH 9.6) capsule

50

A. Foubert et al. / Analytica Chimica Acta 955 (2017) 48e57

and carboxyl functionalized CdSeS/ZnS QDs (525 nm, 575 nm and 665 nm) were purchased from Sigma Aldrich (Diegem, Belgium). Chloroform, sodium sulfate, dimethyl sulfoxide (DMSO), different nitrocellulose membranes (Hi-Flow (HF) plus 07502XSS, HF13502XSS, HF24002XSS), conjugate pad, sample pad and absorbent pad were supplied by Merck Millipore (Darmstadt, Germany). Membrane ‘Fusion 5’ and agarose were purchased from GE Healthcare (Diegem, Belgium). Ethyl acetate and dimethylformamide (DMF) were purchased from Acros organics (Geel, Belgium). Methanol was bought from Biosolve (Valkenswaard, The Netherlands). The protein concentrator tubes were supplied by Fisher Thermo Scientific (Erembodegem, Belgium). Water was purified using a Milli-Q Gradient System (Millipore, Brussels, Belgium). Colorburst™ Blue 3, 30 , 5, 50 -tetramethylbenzidine (TMB) substrate solution containing hydrogen peroxide was supplied by Alercheck (Springvale, ME, USA). Rabbit anti-mouse immunoglobulin (IgG, protein concentration 2.1 g L1) was purchased from DakoCytomation (Heverlee, Belgium). Monoclonal anti-ZEN [23] and anti-DON [24] antibodies were prepared at the Laboratory of Food analysis at Ghent University, Ghent, Belgium. Cross-reactivity of the ZEN mAb was 69% with a-zearalenol, 42% with a-zearalanol, 22% with zearalanone and none at all (<1%) with b-zearalenol and b-zearalanol. Crossreactivity of the DON mAb was 147% with 3-ADON, 65% with 15ADON, and minor cross-reactions with DON-3-G (31%) and DOM (28%), but no cross-reactivity towards the other trichothecenes. The anti-T2-antibody was described by Li et al. (2012) [25]. According to our data the specific anti-T2 antibody showed 110% cross-reactivity for HT2. Because of stability reasons the mycotoxins (and conjugates) and mAbs were stored at 20  C. Stock solutions were prepared in PBS-1% glycerol at 20  C. QD-labelled mAbs (QD-mAbs) and CG-labelled mAbs (CG-mAbs) were stored at 4  C. All working solutions were used for 3 months. The strips were cut by a guillotine cutter Dahle 508 (Germany). Maxisorp polystyrene 96-well plates were purchased from Nunc (Roskilde, Denmark). Immunoassay absorbance and UVevisible spectra were obtained by using the Biorad 550 microplate reader (Temse, Belgium) or the Safire 2 spectrophotometer from Tecan (Mechelen, Belgium). Dynamic light scattering (DLS) measurements were done by the Zetasizer Nano series from Malvern Instruments (Worcestershire, United Kingdom). The QD-based LFIA (QD-LFIA) strips were evaluated under UV light by using a TLC viewing cabinet with 2  15 W tubes of 365 and 254 nm (VWR, Leuven, Belgium). The gel electrophoresis equipment (midi horizontal unit and power supply EV265) was provided by Consort bvba (Turnhout, Belgium). 2.2. Preparation of mycotoxin-protein conjugates The ZEN-OVA conjugates were prepared based on a modified procedure of Thouvenot et al. (1983) [26]. ZEN does not contain any functional groups suitable for conjugation with proteins, as a consequence, a carboxymethyloxime derivate of ZEN (ZEN-CMO) was used for synthesis. 5 mg ZEN was dissolved in 0.5 mL pyridine and stirred for 24 h at room temperature (RT). The reaction mixture was dried under nitrogen at 50  C. The residue was mixed with 2.5 mL distilled water and the pH was adjusted to pH 8 with NaOH. The white residue was sonicated for 1.5 min. The unreacted ZEN was extracted 5 times with 1 mL chloroform. The conjugate was precipitated in the aqueous phase by addition of HCl (pH 3) and was extracted 4 times with 5 mL ethyl acetate. The extract was dried over anhydrous sodium sulfate and filtered. The ethyl acetate was evaporated under vacuum at 50  C. The reaction was tested by direct MS (Xevo TQ-S mass spectrometer) infusion (results not shown) and ZEN-CMO was stored at 20  C. Two mg NHS and

1.7 mg DCC were dissolved in 60 mL DMF and added to 1.6 mg ZENCMO. The ratio ZEN-CMO/NHS/DCC was 1/2/3. The reaction mixture was stirred on an orbital shaker for 4 h at RT and was incubated overnight at 4  C (without shaking). The next day, the formed precipitate was removed by centrifugation. OVA was weighed in a ratio of 1/30 (to ZEN-CMO) and dissolved in 7 mL PBS with 380 mL DMF and the solution was cooled down at 4  C for 30 min. The OVA solution was added dropwise to the ZEN-CMO solution and stirred for 2 h at RT. After this the reaction mixture was incubated overnight at 4  C. The formed precipitate was removed by centrifugation for 30 min at 3500 g and the conjugate was purified and concentrated by use of Pierce concentrator tubes 20 K. DON was coupled to OVA through a carbamate linkage by N,N0 carbonyldiimidazole (CDI) reaction using a modified procedure from Xiao et al. (1995) [27] and Maragos et al. (2010) [28]. First, 1 mg (0.0034 mmol) DON was dissolved in 0.10 mL dry DMSO. This was mixed with 2.2 mg CDI (0.013 mmol), an amount in 4-fold molar excess of DON, stirring for 2 h at RT. The reaction mixture was then added slowly and dropwise to a solution of 3.9 mg OVA in 0.4 mL 0.1 M sodium bicarbonate buffer (NaHCO3), pH 8.5. Molar ratios of DON/OVA were 38/1. The resulted mixture was left for shaking overnight at 4  C. The solution was centrifuged for 10 min at 3500 g to remove the precipitation. The next day, the DON-OVA solution was purified and concentrated by using Pierce concentrator tubes (washed 4 times 45 min at 4500 g). For the synthesis of T2-OVA the protocol described by Chu et al. (1979) was followed [29]. T2 has no reactive group for coupling reactions, therefore, it was first converted to the T2-hemisuccinate, followed by conjugation to OVA. The mycotoxin conjugates were tested by indirect ELISA and their concentration was spectrophotometrically determined.

2.3. Indirect ELISA of mycotoxin-protein conjugates The synthesized mycotoxin-OVA conjugates were diluted in CBS (pH 9.6) and 100 ml per well was added to a 96-well microtiter plate and incubated for 2 h at 37  C. The plate was washed 3 times with PBST (PBS þ Tween 20 0.05% (v/v)). This was followed by blocking with 1% casein in PBS for 1 h at 37  C and 2 times washing with PBST. Mycotoxin standard solutions (50 ml/well, in PBS) and the specific anti-mycotoxin mAb (50 ml/well, in PBS) were added in an appropriate concentration in each well and incubated 1 h at 37  C. After washing 3 times with PBST, the HRP-secondary rabbit antimouse antibody (100 ml/well, dilution of 1/20 000, in PBS) was added in each well and incubated for 1 h at 37  C. Next, the plate was washed 4 times with PBST and 100 ml TMB substrate was added to each well and incubated 15 min at 37  C. The reaction was stopped by adding 50 ml of 2 M H2SO4, and absorbance was measured at 450 nm.

2.4. Labeling of the specific antibodies with quantum dots Green, yellow and red emitting carboxyl functionalized CdSeS/ ZnS QDs were coupled with anti-ZEN, anti-DON and anti-T2 mAb, respectively. First, the QDs were diluted with ultrapure water and activated by adding a combination of EDC and sulfo-NHS. After 3 h stirring at RT the antibody was added and left stirring for 5 h followed with incubation overnight at 4  C. The ratios QD/EDC/sulfoNHS/antibody were optimized and depended on the antibody. Characterization of the conjugation was done by gel electrophoresis and dynamic light scattering (DLS).

A. Foubert et al. / Analytica Chimica Acta 955 (2017) 48e57

2.5. Preparation of colloidal gold nanoparticles and conjugation to antibodies Colloidal gold (CG) was prepared by using the sodium citrate method as previously described [30]. The size of the gold nanoparticles was determined by DLS. Before the mAb were coupled to the gold nanoparticles, a flocculation curve was constructed according to a modified method by Urusov et al. (2011) [31]. First, different dilutions of mAb in PBS were made and 20 mL of these dilutions were dropped to 200 mL CG. Next, the mixtures were incubated for 10 min at RT and 20 mL of a 10% NaCl solution was added. The absorbance was measured at 580 nm after 10 min and a plot absorbance versus concentration of the antibody was constructed. The pH of the CG solution was adjusted with potassium carbonate, followed by the addition of antibody (anti-DON, antiZEN and anti-T2 mAb) diluted in 10 mM Tris buffer, pH 8.5. After 15 min incubation, a solution of BSA (CG/BSA: 40/1 (v/v)) was added and the mixture was vigorously stirred for 10 min. The CG were pelleted by centrifugation at 10 000 g for 20 min at 4  C. The pellet was resuspended in 10 mM Tris buffer, pH 8.5, containing 1% BSA and 1% sucrose (TBSU). The centrifugation was repeated twice and the obtained solution was stored at 4  C. Conjugation of the mAbs with CG was characterized by UVevis absorbance spectra and DLS. 2.6. Preparation and optimization of the lateral flow immunoassays The mycotoxin conjugates DON-OVA (1.0 mg mL1), ZEN-OVA (2.0 mg mL1), T2-OVA (2.0 mg mL1) and the rabbit anti-mouse immunoglobulin (2.1 mg mL1) were all diluted in PBS buffer and applied onto the nitrocellulose membranes as test eand control lines, respectively and with a distance of 5 mm. The mycotoxin conjugates were manually dot-blotted onto the membrane and left to dry for 30 min at 37  C. Different types of nitrocellulose membranes and blocking solutions were tested and compared. The prepared sheets were then cut into 0.5  5 cm strips using a paper cutter. CG or QD-labelled mAbs (CG-mAbs or QD-mAbs) (30 mL in total) were dispensed on the conjugate pad, which was treated with different blocking solutions to find the ideal solution for release of the labelled mAbs from the conjugate pad. All the membranes were left to dry for 30 min at 37  C. Strips from 5 cm length were composed as follows starting from the bottom: the sample pad (0.5  1.5 cm), the conjugate pad (0.5  1 cm), the nitrocellulose membrane (0.5  5 cm) and the absorbent pad (0.5  1.5 cm), with a 2 mm overlap (Fig. 1). The assays were performed at RT. The test was carried out by submerging the strip in 200 mL of sample for 15 min and the result was evaluated by the naked eye. In case of the QD-LFIA a UV light was necessary. 2.7. Sample preparation and recovery experiments The same sample preparation procedure was applied for both the LFIAs. Five grams of blank wheat samples were weighed and fortified with an appropriate amount of spike solution of mycotoxins and left to dry for 20 min in the dark. Next, extraction with 15 mL of 80% (v/v) methanol/water mixture happened by vigorous shaking for 4 min. The samples were left to rest for 5 min to allow sedimentation of particulate matter. One hundred mL of the supernatants were used to dilute with PBS in order to reduce the amount of methanol to 10% and subsequently used to perform the LFIA. After 15 min of incubation, the result could be visually assessed. Recovery experiments of the extraction procedure were carried out in triplicate at spiking levels equal to the desired cut-off limits

51

for each mycotoxin and for every recovery experiment, two parallel sets of samples were used. One set was spiked with a known concentration prior to the start of the extraction procedure (begin spike). Another set was spiked after the extraction procedure was performed (end spike). In all cases, the extract was filtered and analyzed by liquid chromatography tandem mass spectrometry (LC-MS/MS). The ratio of the peak area of the begin spike to the peak area of the end spike were used to calculate the analyte recovery. For the recovery experiments a Waters Acquity UPLC system (Waters, Zellik, Belgium) and a Xevo TQ-S mass spectrometer were used for separation and detection of the analytes of interest. The analytical column used was a HSS T3, 100 mm  2.1 mm i.d., 1.7 mm (Agilent, Diegem, Belgium). Gradient elution with mobile phase consisted of solvent A (water) and solvent B (methanol) with addition of 0.01% acetic acid at a flow rate of 0.400 mL min1. The sample injection volume was 20 mL. The MS analyses were carried out in multiple reaction monitoring (MRM) mode and ionization was performed in the positive electrospray ionization (ESI) mode. The capillary voltage was 0.50 kV and nitrogen was used as the spray gas. Source and desolvation temperatures were set at 150 and 450  C. 2.8. Lateral flow immunoassay procedure After extraction, the diluted sample (200 mL) was applied onto the sample pad. The solution migrated up the membrane towards the absorbent pad by capillary forces. The labelled mAbs (CG-mAbs or QD-mAbs) were solubilized by the sample extract from the conjugate pad. In absence of mycotoxins the labelled mAbs were captured by the immobilized mycotoxin conjugates on the test lines via antibody-antigen interactions. The excess labelled mAb migrated further and interacted on the control line with the secondary antibody (control line). However, in presence of mycotoxins the labelled mAbs captured the mycotoxins and migrated over the membrane without (or in less extent) interaction on the test lines. The developed LFIA was based on a competitive immunoassay format (Fig. 1). Therefore, the color intensity of the test lines was inversely correlated with the mycotoxin concentration. After 15 min, the test results could be visually evaluated. A negative test resulted in 4 lines (1 control line and 3 test lines for DON, ZEN and T2/HT2). A positive test for all the mycotoxins gave only 1 line, the control line. If no control line was present, the test was considered to be invalid. 2.9. Validation of both lateral flow immunoassays The cut-off limits were defined as the lowest analyte concentration which did not give colored test lines and were set as 80% of the EU maximum permitted levels. The spiked concentration levels were different for each analyte which was based on the maximum limits of each analyte in the different foodstuffs as laid down by the European Commission Regulation [32,33]. The following seven performance characteristics are relevant for the validation of qualitative screening assays for mycotoxin determination; cut-off limits, specificity (the ability to detect negative samples as negative), sensitivity (the ability to detect truly positive samples as positive), false positive rate, false negative rate, positive predicted value and negative predicted value. A probability of detection (POD) study was done by constructing probability of positive response versus concentration (POD curves). Such POD curves were constructed by using series of blank wheat samples which were spiked at 7 concentration levels, in 10 replicates. Four concentrations were below, one level at and two levels above the pre-determined cut-off limits, which were

52

A. Foubert et al. / Analytica Chimica Acta 955 (2017) 48e57

Fig. 1. Schematic representation of a competitive quantum dot based lateral flow immunoassay.

different for each analyte and was based on the maximum limits. The pre-determined cut-off limits were optimized during prevalidation studies. The validation was performed according Commission Decision 2002/657/EC [32] and Commission Regulation (EC) No 401/2006 [34]. The latter refers to the guidelines for validation of binary qualitative chemistry methods drafted by the AOAC International [35]. To determine the performance characteristics blank wheat samples (n ¼ 40) were used. A set of 20 samples were used as control blanks (n ¼ 20 for each analyte) while the other set of 20 samples were fortified with the target analytes at the respective cut-off limits. Control and test experiments were always realized in parallel and each result was shown blindly to three other colleagues who visually evaluated the results. The following equations were used to calculate the other six performance parameters: sensitivity (Ntrue positive/(Ntrue positive þ Nfalse negative)  100 (%)), specificity rate (Ntrue negative/(Ntrue negative þ Nfalse positive)  100 (%)), false positive rate (Nfalse positive/(Ntrue positive þ Nfalse positive)  100 (%)), false negative rate (Nfalse negative/(Ntrue negative þ Nfalse negative)  100 (%)), positive predictive rate (Ntrue positive/(Ntrue positive þ Nfalse positive)  100 (%)), negative predictive rate (Ntrue negative/(Ntrue negative þ N false negative)  100 (%)) [36].

2.10. LC-MS/MS analysis of naturally contaminated samples For the analysis of naturally contaminated samples a Waters Acquity UPLC system (Waters, Zellik, Belgium) and a Micromass Quattro Premier XE triple quadrupole mass spectrometer were used, equipped with Masslynx software for data processing. The column used was a 150 mm  2.1 mm i.d. 5 mm Symmetry C18, with a 10 mm  2.1 mm i.d. guard column of the same material (Waters, Zellik, Belgium). A detailed description of the sample preparation can be found in Monbaliu et al. (2010) [37]. The column was kept at RT. The sample injection volume was 20 mL. Separation was done with a gradient elution with mobile phase consisting of solvent A (water/methanol/acetic acid, 94/5/1 (v/v/v) and 5 mM ammonium acetate) and solvent B (methanol/ water/acetic acid, 97/2/1 (v/v/v) and 5 mM ammonium acetate) at a flow rate of 0.3 mL min1 with a gradient elution program. The MS analyses were carried out in selected reaction monitoring (SRM) and ionization was performed in the positive ESI mode. The capillary voltage was 3.2 kV, and nitrogen was used as spray gas. Source and desolvation temperatures were set at 150 and 350  C, respectively.

3. Results and discussion 3.1. Preparation of mycotoxin-protein conjugates All three mycotoxins were conjugated to the protein OVA. Prior to their use in the LFIA the prepared mycotoxin conjugates (DONOVA, ZEN-OVA and T2-OVA) were tested by an indirect ELISA to check the reaction. As shown in Fig. 2 the mycotoxin conjugates DON-OVA, ZEN-OVA and T2-OVA could be used in dilutions of 1/ 15 000, 1/5000 and 1/1000, respectively and still gave a significant clear signal. So, it can be concluded that all the coupling reactions were successful. The concentrations of the mycotoxin-protein conjugates were spectrophotometrically determined as 1.0 mg mL1, 2.0 mg mL1, 2.0 mg mL1 for DON-OVA, ZEN-OVA and T2-OVA, respectively. 3.2. Synthesis and characterization of labelled antibodies 3.2.1. Quantum dot labelled monoclonal antibodies All mAbs against the mycotoxins were conjugated with the QDs by the activated ester technique. The activation of the QDs in a ratio of 50/50 EDC/sulfo NHS resulted in severe aggregation. Consequently, this aggregation resulted in no flow over the membrane. For each mAb against a specific mycotoxin the ratios of EDC/sulfoNHS and mAb were optimized and the conjugation was characterized by gel electrophoresis. The coupled and uncoupled QDs moved through the agarose gel at different rates, based on each particle's size and charge. As illustrated in Fig. 3 the ratios of the linkers (EDC/sulfo-NHS) and mAbs were optimized until the QDs stayed in front of the gel and did not show any precipitation. Next, the reaction was checked by DLS (Fig. 3). Uncoupled green, yellow and red emitting QDs showed an average size of ±10.1 nm. After conjugation with the mAbs an average size increase of 7.3 nm was observed. However, the yellow emitting QDs conjugates showed a higher increase (10.3 nm) in size, while the red emitting QDs conjugates showed only an increase of 7.1 nm. From all these data it was concluded that the conjugation with the mAbs was successful and that the size of the conjugates was small enough to be used in a LFIA. 3.2.2. Colloidal gold nanoparticle labelled monoclonal antibodies CG was produced according to a standard procedure by the reduction of HAuCl4 with sodium citrate. A size characterization by DLS of the produced CG showed a narrow size distribution and a size of ±10 nm. CG was conjugated to the mAbs by constructing the flocculation curves to determine the optimal loading of the gold

A. Foubert et al. / Analytica Chimica Acta 955 (2017) 48e57

53

Fig. 2. Determination of the mycotoxins by indirect ELISA (the dilutions of DON-OVA, ZEN-OVA and T2-OVA are 1/15 000, 1/5000 and 1/1000 dilution, respectively), n ¼ 3.

Fig. 3. Gel electrophoresis of coupled/uncoupled quantum dots (QDs). a) uncoupled yellow emitting QDs and QD-anti-DON mAb (ratio QD/EDC/sulfo-NHS/Ab: 1/5/5/3), b) uncoupled green emitting QDs and QD-anti-ZEN mAb (ratio QD/EDC/sulfo-NHS/Ab: 1/4/4/2) and c) coupled QD-anti-T2 mAb (ratio QD/EDC/sulfo-NHS/Ab: 1/4/4/1) and uncoupled red emitting QDs (A.). Size distribution curves from green (green line), yellow (orange line) and red emitting QDs (red line) before and after conjugation with anti-ZEN mAb (green dashed line), anti-DON mAb (orange dashed line) and anti-T2 mAb (red dashed line), respectively (B.). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

nanoparticles. When not enough antibody was present an increase in absorbance at 580 nm upon addition of NaCl was observed. For all the used mAbs this point was more or less the same and concentrations of 25 mg mL1 anti-DON mAb, 16.7 mg mL1 anti-ZEN mAb and 19 mg mL1 anti-T2 mAb were necessary to form a stable CG suspension. These conjugations were based on electrostatic interaction and were characterized by using UVevis absorption spectra and DLS. The surface plasmon absorption band was located at 520 nm (Fig. 4). Due to the strong electrostatic interaction between the CG and the mAbs, the surface plasmon absorption band shifted to 530 nm for all the CG-mAbs, which indicated the conjugation took place. In addition, DLS was performed to determine the sizes of the used CG nanoparticles and the obtained conjugates (Fig. 4). While CG was characterized with a size of ±10 nm, the conjugates had sizes between 15 and 16 nm. So the conjugation resulted in an increase of ±7 nm (6.3 nm, 7.3 nm and 7.3 nm for CG-anti-DON mAb, CG-anti-ZEN mAb and CG-anti-T2 mAb, respectively).

is used for the extraction of DON from cereals. For this reason a combination of these two solvents was chosen for the extraction of DON, ZEN and T2/HT2. It was decided to perform extraction of 5 g of sample by using 15 mL 80% (v/v) methanol/water. Although it was proven that a higher extraction volume (e.g. 25 mL or 50 mL) resulted in a higher analyte recovery 15 mL was chosen because it resulted in more concentrated supernatant. This allowed to dilute the sample more to reduce the methanol concentration. The recovery data for the different mycotoxins in wheat are shown in Table 1. The performance of a LFIA can be affected by the presence of a high percentage of organic solvents. Sample solutions containing 20e30% methanol resulted in a disturbed antigen-antibody interaction and flow properties. When the sample was too diluted (containing less than 5% methanol) the development of the test strip took too long. For this reason it was decided to take 100 mL of the supernatant and dilute it with 720 mL PBS, which corresponds with a sample that contains 10% methanol.

3.3. Optimization of the sample preparation

3.4. Development and optimization of the lateral flow immunoassay

In order to develop a rapid screening test, a simple and rapid sample preparation must be developed. Methanol is often used for the extraction of ZEN and T2/HT2 from cereals, while water or PBS

3.4.1. Selection of the membrane type and blocking solution Different nitrocellulose membranes with different pore size

54

A. Foubert et al. / Analytica Chimica Acta 955 (2017) 48e57

Fig. 4. UVevis absorption spectra of colloidal gold (CG) (dark blue line) and its conjugates, CG-anti-DON mAb (red line), CG-anti-T2 mAb (green line), CG-anti-ZEN mAb (light blue) (A.). Size distribution curves for CG before (blue line) and after conjugation with anti-ZEN mAb (grey line), anti-DON mAb (yellow line) and anti-T2 mAb (orange line) (B.). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Table 1 Recovery data for DON, ZEN, T2 and HT2 using 15 mL extraction solvent in wheat. SD: standard deviation. Mycotoxin

Average recovery (%) ± SD

DON ZEN T2 HT2

103 ± 10.0 97 ± 5.9 100 ± 8.3 93 ± 11.4

were compared. The pore size of the membrane directly affects the flow rate and sensitivity of the LFIA, i.e. a smaller pore size results in a slower flow rate. Four different nitrocellulose membranes, HF 240, 135, 090 and 075 were tested. A final choice of membrane was done by using the intensity of the control line as criterion. When the membranes were not blocked, a bright line intensity for the CGLFIA with the HF 135 and HF 240 membrane was visually observed. However, for the QD-LFIA a bright line intensity was only received using the HF 240 membrane. As the aim was to compare the CG -and QD-LFIA the HF 240 membrane was selected. PBS (pH 7.4) was chosen as diluting buffer for the mycotoxin conjugates. To investigate the effect of a blocking buffer (less non-specific binding, decrease of variability) on the control line intensity the membrane was blocked with different blocking solutions. In case of the CGLFIA, the blocking with different percentages of BSA, casein or sucrose did not show any improvement in reduction of the variability and line intensity. For the QD-LFIA the blocking resulted in absence of test -and control lines, even after washing of the membrane with PBS. So for both the LFIA strips it was decided not to include a blocking step.

3.4.2. Choice of conjugate pad and blocking solutions Selection of the optimal conjugate pad and its blocking was a challenge. The conjugate pad can cause non-reproducible results because the solution can diffuse in an irregular manner along and across the fiber. It is important that the used conjugate pad has low non-specific binding, consistent flow characteristics and bed volume. Two kinds of conjugate pads were tested namely (i) glass fiber membrane and (ii) Fusion 5, a single layer matrix membrane. First, the conjugate pads were tested without preliminary blocking. For the QD-LFIAs this resulted in almost no flow from both the conjugate pads. Therefore both pads were selected for further optimization. In case of the CG-LFIAs, the Fusion 5 gave slightly better colored control -and test lines in comparison with the glass fiber pad (Fig. 5). Hereby, Fusion 5 was selected for further development of the CG-LFIA. A possible explanation for the bad flow of QD-mAbs

from the conjugate pads could be the presence of the hydrophilic coating around the QDs which causes high non-specific binding. To improve the solubilization of the labelled mAbs and gain a better line formation with an adequate migration rate different blocking solutions (casein, sucrose, BSA and Tween 20) on the Fusion 5 (for the CG-LFIA) and both pads (for the QD-LFIA) were tested. Blocking of the glass fiber conjugate pad resolved the release problem of the QD-mAbs but still gave a lot of non-specific binding on the Fusion 5 pad (Fig. 5). In addition, when the Fusion 5 pads were stored overnight in the fridge a shift in color of the QD-mAb conjugates was observed. Hereby, the glass fiber conjugate pad was selected for the development of the QD-LFIA. For both the CG-LFIA and the QD-LFIA Tween 20 was necessary to remove all the mAbs from the conjugate pad. Further optimization resulted in selection of a 4% sucrose, 1% BSA and 0.5% Tween 20 and a 5% sucrose, 1% BSA and 0.5% Tween 20 in water solution for the CG-LFIA and QD-LFIA, respectively. 3.5. Cut-off limits of the lateral flow immunoassays Sensitivity of each immunochemical method depends on the concentration and affinity of the used immunoreagents. Therefore the determination of the necessary amount of mycotoxin conjugates and labelled mAbs is a key factor in the development of a LFIA. The concentrations of immunoreagents were optimized to satisfy the following criteria: (i) the appearance of a clear color on the test lines for negative samples; (ii) the difference between positive and negative samples could be easily distinguished with the naked eye and (iii) minimum reagent consumption. These parameters were optimized by a checkerboard titration until the desired cut-off limit (DON: 1000 mg kg1; ZEN: 80 mg kg1 and T2/HT2: 80 mg kg1, both for the CG and QD-LFIA) was reached. The cut-off limits were set as 80% of the EU maximum established levels as laid down by the European Commission Regulation [32,33]. For both the LFIAs the same cut-off limits were chosen in order to make a correct comparison. For the optimization of the CG-LFIA the prepared solutions of mycotoxin CG-mAbs were already quite diluted and the checkerboard titrations happened by varying the volumes of the mycotoxin CG-mAbs until an optimum was reached. The synthesis of the QD-mAbs resulted in samples with a higher concentration of nanoparticles which needed to be diluted in PBS (pH 7.4) followed by application on the conjugate pad. LFIAs should be widely usable for a qualitative on-site testing of the target analyte at threshold levels. Therefore, the difference between negative and positive results must be easy detectable by the naked eye. For the CG-LFIA a series of conjugate dilutions (1/2; 1/5; 1/10) were

A. Foubert et al. / Analytica Chimica Acta 955 (2017) 48e57

55

Fig. 5. Choice of conjugate pad (glass fiber or Fusion 5) with or without blocking for colloidal gold based lateral flow immunoassay (A.) and quantum dot based lateral flow immunoassay (B.).

made for DON-OVA, ZEN-OVA and T2-OVA. At the same time different volumes of labelled mAb were applied on the conjugate pad. The optimal concentrations were chosen as 25 mL CG-antiDON mAbs, 20 mL CG-anti-ZEN mAb and 25 mL CG-anti-T2 mAb. The exact amount of all three mycotoxin conjugates and CG-mAbs that needed to be immobilized are depicted in Table 2. In case of the QD-LFIA, dilution of the QD-mAbs was necessary in order to receive a fluorescence intensity that was visible by the naked eye under UV irradiation. A too high concentration resulted in a high background fluorescence due to the excess of QD-mAbs. The exact amount of QD-mAbs that were immobilized on the conjugate pad could be calculated and was compared with the amount needed for the development of a CG-LFIA (Table 2). The same was done for the mycotoxin conjugates. As a conclusion, the development of a LFIA with QDs as label required ±20 times less labelled mAbs and ±4 times less mycotoxin conjugates than the CG-LFIA with the same cut-off limits. Furthermore, the signals of the QD-LFIA were much brighter and easier to interpret than the ones from the CG-LFIA. 3.6. Specificity of the developed lateral flow immunoassays Although the haptens DON, ZEN and T2/HT2 have different chemical structures it is important to check the cross-reactivity in a multiplex assay. Here, different mycotoxin conjugates were immobilized on the membrane. Fig. 6 represents the specificity of CG-anti-DON mAb, CG-anti-ZEN mAb and CG-anti-T2/HT2 mAb for corresponding test lines. The color development was observed at all the test lines and control line, when a mixture of CG-anti-DON mAb, CG-anti-ZEN mAb and CG-anti-T2/HT2 mAb was applied on the conjugate pad and run with blank sample extract. When antiDON mAb-CG, anti-ZEN mAb or anti-T2/HT2 mAb were added separately to the conjugate pad, only the color of the respective test line was visible. It could be concluded that there was no-cross reaction for multi-analysis of DON, ZEN and T2/HT2 using the LFIA.

Table 2 Amounts of labelled mAbs and mycotoxin conjugates to develop a quantum dot (QD) eand colloidal gold (CG) based lateral flow immunoassay (LFIA) with a cut-off limit of 1000 mg kg1 DON, 80 mg kg1 ZEN and 80 mg kg1 T2/HT2.

Labelled mAb (mg) Anti-DON mAb Anti-ZEN mAb Anti-T2 mAb Mycotoxin conjugate (ng) DON-OVA ZEN-OVA T2-OVA

CG-LFIA

QD-LFIA

1.5 0.9 0.9

0.066 0.049 0.041

1.5 4 4

0.38 1 1

Fig. 6. Specificity of colloidal gold (CG)-anti-DON mAb, CG-anti-ZEN mAb and CG-antiT2 mAb. Strip 1: conjugate pad was treated with all the CG-mAb conjugates, strip 2 was treated with CG-anti-DON mAb, strip 3 was treated with CG-anti-ZEN mAb and strip 4 was treated with CG-anti-T2 mAb.

3.7. Validation of the developed lateral flow immunoassays The performance characteristics of the CG eand QD-LFIAs were defined and validated. To assess the sensitivity of the CG and QDLFIA wheat matrices were spiked with different levels below, equal and above the cut-off limits (0/0/0/0, 250/25/15/15, 500/50/ 25/25, 800/70/30/30, 1000/80/40/40, 1500/150/80/80 mg kg1) of DON/ZEN/T2/HT2, respectively. This is for both CG and QD-LFIA depicted in Fig. 7. Both LFIAs were developed/optimized this way that there was no color development anymore at or above the cutoff limits. So the test lines disappeared at  1000 mg kg1 DON, 80 mg kg1 ZEN, 40 mg kg1 T2 and 40 mg kg1 HT2. From the POD curves the uncertainty region for the different mycotoxins could be determined for the CG eand QD-LFIA. This uncertainty corresponds to false response rates (Table 3). The uncertainty regions were broader when the CG-LFIA was used to construct the POD curves. All the curves had a sigmoidal shape (data not shown). Comparing the CG-LFIA with the QD-LFIA it can be concluded that the QD-LFIA gave more intense signals, whereas the CG-LFIA signals were faint. This was also reflected in the qualitative performance characteristics obtained for DON, ZEN, T2 and HT2 (Table 3). Table 3 shows the performance characteristics obtained in wheat samples for DON, ZEN and T2/HT2. Both LFIAs showed a good specificity (100%) for all four mycotoxins. However, the CG-

56

A. Foubert et al. / Analytica Chimica Acta 955 (2017) 48e57

Fig. 7. Visible cut-off limits (strip 5) of the colloidal gold -and quantum dot based lateral flow immunoassay for DON, ZEN, T2 and HT2. The numbers on the strip indicate the concentrations of the mycotoxins DON/ZEN/T2/HT2. Strip 1: 0/0/0/0, strip 2: 250/25/15/15, strip 3: 500/50/25/25, strip 4: 800/70/30/30 strip 5: 1000/80/40/40, strip 6: 1500/150/80/ 80 mg kg1.

Table 3 Summarized performance characteristics for colloidal gold based -and quantum dot based LFIA obtained in wheat samples (n ¼ 40; 20 blank samples and 20 spiked samples at assay cut-off limits). Performance parameters (%)

CG-LFIA DON

ZEN

T2/HT2

DON

ZEN

T2/HT2

False positive rate False negative rate Sensitivity Specificity Positive predictive value Negative predictive value Uncertainty region (lower limit - upper limit) (mg kg1)

0 0 100 100 100 100 545e836

0 13 87 100 100 87 38e85

0 9 91 100 100 91 17e40

0 0 100 100 100 100 550e805

0 0 100 100 100 100 40e72

0 5 95 100 100 95 20e39

LFIA was less sensitive to ZEN (sensitivity 87%) and T2/HT2 (sensitivity 91%). Compared with the QD-LFIA the sensitivity to ZEN (100%) and T2/HT2 (100%) was higher. The lower sensitivity of the CG-LFIA to ZEN and T2/HT2 can be explained by their low cut-off limits. This is strengthened by the fact that the sensitivity of the QD-LFIA was higher. However, the false positive rate was 0% for all mycotoxins for both LFIAs. Similarly, 0% false negative rate was obtained for the analysis of DON for both the LFIAs, whereas a 13% and 9% rate was obtained for ZEN and T2/HT2, respectively, in case of the CG-LFIA. The QD-LFIA showed a lower false negative rate for T2/HT2 of 5%. According to Commission Decision 2002/657/EC, only those analytical techniques for which it can be demonstrated in a documented traceable manner that they are validated and have false compliant rate of less than 5% at the levels of interest shall be used for screening purposes. In case a non-compliant outcome occurs, the result should be verified by a complementary method (EC, 2002) [38]. This means that the QD-LFIA fulfills this criterion, which is not the case for the CG-LFIA. The positive predictive values were 100% for all mycotoxins of the CG-LFIA, the same as the QD-LFIA.

QD-LFIA

However, the negative predictive value was 100%, 87% and 91% for DON, ZEN and T2/HT2 respectively for the CG-LFIA. The QD-LFIA gave negative predictive values of 100% for DON and ZEN, whereas for T2 it was 95%. In general, it can be said that the QD-LFIA gave rise to a more sensitive assay of a better quality. It should be noticed that both the LFIAs were stable for at least 1 months at RT. Blank and spiked samples gave a similar result with the use of fresh and stored LFIAs in terms of color development and sensitivity. 3.8. Analysis of naturally contaminated wheat samples To assess method accuracy, the QD- and CG-LFIA performance was verified by LC-MS/MS analysis in naturally contaminated wheat samples. The samples were analyzed in triplicate. Table 4 shows the different results obtained with the naturally contaminated samples as well as the concentration of the toxins measured in these samples following LC-MS/MS analysis. Both LFIAs gave a positive result whenever a sample was contaminated with mycotoxins higher than the cut-off limit, which was confirmed by LC-

Table 4 Comparative analysis of CG- and QD-LFIA and LC-MS/MS for qualitative analysis of wheat samples. Visual results CG

Wheat Wheat Wheat Wheat Wheat

1 2 3 4 5

LC-MS/MS results (mg kg1) ± SD

Visual results QD

DON

ZEN

T2/HT2

DON

ZEN

T2/HT2

DON

ZEN

T2/HT2

þþþ þþþ þþþ e ±±±

þþþ þþþ þþþ e ±þþ

þþþ þ±± þþþ þþþ þþþ

þþþ þþþ þþþ e ±±±

þþþ þþþ þþþ e þþþ

þþþ þþ± þþþ þþþ þþþ

100 ± 50


SD: standard deviation, LOD: limit of detection. þ: an obvious line was observed. -: no line was observed. ±: weak color at test line.

A. Foubert et al. / Analytica Chimica Acta 955 (2017) 48e57

MS/MS analysis. However, when the values were around the cut-off limit (e.g. sample wheat 2 and 5) the interpretation of the LFIA results was more difficult in case of the CG-LFIA but still correct. In case of the QD-LFIA, results obtained with positive samples gave a very good visual contrast compared to those with negative samples. Consistently, the in-parallel results demonstrate that both the developed LFIAs are reliable and sensitive enough to be used for qualitative analyses.

[12]

[13] [14]

[15]

4. Conclusion [16]

In this study, we report a simple and sensitive QDs based rapid detection method for 4 mycotoxins. This was followed by a thorough comparison with a CG-LFIA, by using the same immunoreagents. The QD-LFIA showed to be robust, sensitive and allowed detection of 4 mycotoxins. The assay protocol was simple, easy to use and reproducible. From a practical point of view, the QD-LFIA offers few advantages over the CG-LFIA. First and foremost, the QD-LFIA is more sensitive which results in less antibody and antigen use which makes it a more economical assay. The ‘brightness’ of QDs as label made it easier to interpret and distinguish between positive and negative results. By using a covalent linking strategy the stability of the QDs-mAbs increases because it does not have the competition problem that exists when (bio)molecules are conjugated to CG by electrostatic interaction. Besides the fact that a QD-LFIA requires a portable UV lamp, it fulfills all the criteria of a sensitive, qualitative screening tool that can be used “on site”. Acknowledgements The authors would like to thank the BOF Special Research Fund from Ghent University, GOA project no. 01G02213 and the Russian Ministry of Science and Education (project 4.1708.2014/K). We gratefully acknowledge Prof. Dr. S. De Smedt for the opportunity to perform the DLS measurements in his laboratory and Prof. Dr. D. Deforce for allowing us to use the Infinite Tecan Plate Reader.

[17]

[18]

[19] [20] [21]

[22]

[23]

[24]

[25]

[26]

[27]

References [1] W. Jin, K. Yamada, M. Ikami, N. Kaji, M. Tokeshi, Y. Atsumi, M. Mizutani, A. Murai, A. Okamoto, T. Namikawa, Y. Baba, M. Ohta, Application of IgY to sandwich enzyme-linked immunosorbent assays, lateral flow devices, and immunopillar chips for detecting staphylococcal enterotoxins in milk and dairy products, J. Microbiol. Methods 92 (2013) 323e331. [2] A.N. Berlina, N.A. Taranova, A.V. Zherdev, M.N. Sankov, I.V. Andreev, A.I. Martynov, B.B. Dzantiev, Quantum-dot-based immunochromatographic assay for total IgE in human serum, PLoS One 8 (2013) e77485. [3] J. Zhou, K. Zhu, F. Xu, W.J. Wang, H.Y. Jiang, Z.H. Wang, S.Y. Ding, Development of a microsphere-based fluorescence immunochromatographic assay for monitoring lincomycin in milk, honey, beef, and swine urine, J. Agric. Food Chem. 62 (2014) 12061e12066. [4] J.-J. Wang, B.-H. Liu, Y.-T. Hsu, F.-Y. Yu, Sensitive competitive direct enzymelinked immunosorbent assay and gold nanoparticle immunochromatographic strip for detecting aflatoxin M1 in milk, Food Control 22 (2011) 964e969. [5] J.W. Bennett, M. Klich, Mycotoxins, Clin. Microbiol. Rev. 16 (2003) 497e516. [6] Q. Wang, L. Xu, Beauvericin, a bioactive compound produced by fungi: a short review, Molecules 17 (2012) 2367e2377. [7] L. Pinotti, M. Ottoboni, C. Giromini, V. Dell'Orto, F. Cheli, Mycotoxin contamination in the EU feed supply chain: a focus on cereal byproducts, Toxins 8 (2016) 24. [8] S.Q. Song, N. Liu, Z.Y. Zhao, E.N. Ediage, S.L. Wu, C.P. Sun, S. De Saeger, A.B. Wu, Multiplex lateral flow immunoassay for mycotoxin determination, Anal. Chem. 86 (2014) 4995e5001. [9] A.M. Santini, S. Uhlig, A. Ritieni, Fusaproliferin, beauvericin and enniatins: occurence in food - a review, World Mycotoxin J. 5 (2012) 71e81. [10] S. Joshi, A. Segarra-Fas, J. Peters, H. Zuilhof, T.A. van Beek, M.W.F. Nielen, Multiplex surface plasmon resonance biosensing and its transferability towards imaging nanoplasmonics for detection of mycotoxins in barley, Analyst 141 (2016) 1307e1318. [11] J. Bogdanovic, M. Koets, I. Sander, I. Wouters, T. Meijster, D. Heederik, A. van Amerongen, G. Doekes, Rapid detection of fungal alpha-amylase in the work

[28] [29] [30]

[31]

[32]

[33]

[34]

[35] [36] [37]

[38]

57

environment with a lateral flow immunoassay, J. Allergy Clin. Immun. 118 (2006) 1157e1163. C. Chen, J. Wu, A fast and sensitive quantitative lateral flow immunoassay for Cry1Ab based on a novel signal amplification conjugate, Sens. Basel 12 (2012) 11684e11696. J.A. Ho, R.D. Wauchope, A strip liposome immunoassay for aflatoxin B1, Anal. Chem. 74 (2002) 1493e1496. D. Serrate, J.M. De Teresa, C. Marquina, J. Marzo, D. Saurel, F.A. Cardoso, S. Cardoso, P.P. Freitas, M.R. Ibarra, Quantitative biomolecular sensing station based on magnetoresistive patterned arrays, Biosens. Bioelectron. 35 (2012) 206e212. B. O'Farrell, Lateral flow technology for field-based applications-basics and advanced developments, Top. Companion Anim. M. 30 (2015) 139e147. B. Liu, L. Wang, B. Tong, Y. Zhang, W. Sheng, M. Pan, S. Wang, Development and comparison of immunochromatographic strips with three nanomaterial labels: colloidal gold, nanogold-polyaniline-nanogold microspheres (GPGs) and colloidal carbon for visual detection of salbutamol, Biosens. Bioelectron. 85 (2016) 337e342. D. Pyo, Comparison of fluorescence immunochromatographic assay strip and gold colloidal immunochromatographic assay strip for detection of microcystin, Anal. Lett. 40 (2007) 907e919. N. Khreich, P. Lamourette, B. Lagoutte, C. Ronco, X. Franck, C. Creminon, H. Volland, A fluorescent immunochromatographic test using immunoliposomes for detecting microcystins and nodularins, Anal. Bioanal. Chem. 397 (2010) 1733e1742. A.P. Alivisatos, W. Gu, C. Larabell, Quantum dots as cellular probes, Annu. Rev. Biomed. Eng. 7 (2005) 55e76. K.E. Sapsford, T. Pons, I.L. Medintz, H. Mattoussi, Biosensing with luminescent semiconductor quantum dots, Sensors 6 (2006) 925e953. L. Wang, W. Chen, W. Ma, L. Liu, W. Ma, Y. Zhao, Y. Zhu, L. Xu, H. Kuang, C. Xu, Fluorescent strip sensor for rapid determination of toxins, Chem. Commun. 47 (2011) 1574e1576. H. Duan, X.L. Chen, W. Xu, J.H. Fu, Y.H. Xiong, A. Wang, Quantum-DoT submicrobead-based immunochromatographic assay for quantitative and sensitive detection of zearalenone, Talanta 132 (2015) 126e131. N.A. Burmistrova, I.Y. Goryacheva, E.Y. Basova, A.S. Franki, D. Elewaut, K. Van Beneden, D. Deforce, C. Van Peteghem, S. De Saeger, Application of a new antizearalenone monoclonal antibody in different immunoassay formats, Anal. Bioanal. Chem. 395 (2009) 1301e1307. Y. Guo, M. Sanders, A. Galvita, A. Heyerick, D. Deforce, M. Bracke, S.A. Eremin, S. De Saeger, Heterologous screening of hybridomas for the development of broad-specific monoclonal antibodies against deoxynivalenol and its analogues, World Mycotoxin J. 7 (2014) 257e265. Y.S. Li, Y.R. Zhang, W.M. Shi, Z.H. Wang, J.Z. Shen, S.X. Zhang, Determination of T-2 toxin and HT-2 toxin in milk: a comparison of three formats of immunoassays, Anal. Lett. 45 (2012) 2425e2435. D. Thouvenot, R.F. Morfin, Radioimmunoassay for zearalenone and zearalanol in human serum: production, properties, and use of porcine antibodies, Appl. Environ. Microbiol. 45 (1983) 16e23. H. Xiao, J.R. Clarke, R.R. Marquardt, A.A. Frohlich, Improved methods for conjugating selected mycotoxins to carrier proteins and extran for immunoassays, J. Agric. Food Chem. 43 (1995) 2092e2097. C.M. Maragos, M. Busman, Rapid and advanced tools for mycotoxin analysis: a review, Food Addit. Contam. A 27 (2010) 688e700. F.S. Chu, S. Grossman, R.D. Wei, C.J. Mirocha, Production of antibody against T2 toxin, Appl. Environ. Microb. 37 (1979) 104e108. N.V. Beloglazova, I.Y. Goryacheva, R. Niessner, D. Knopp, A comparison of horseradish peroxidase, gold nanoparticles and quantum dots as labels in non-instrumental gel-based immunoassay, Microchim. Acta 175 (2011) 361e367. A.E. Urusov, S.N. Kostenko, P.G. Sveshnikov, A.V. Zherdev, B.B. Dzantiev, Immunochromatographic assay for the detection of ochratoxin A, J. Anal. Chemþ 66 (2011) 770e776. E. Commission, Commission regulation (EC) No. 1881/2006, Off. J. Eur. Union L255 (14) (2007). http://eur-lex.europa.eu/legal-content/en/ALL/?uri¼CELEX% 3A32006R1881. E. Commission, Commission recommendation 2013/165/EU, Off. J. Eur. Onion L91 (4) (2013). http://eur-lex.europa.eu/legal-content/NL/TXT/?uri¼CELEX% 3A32013H0165. E. Commission, Commission regulation No 401/2006, Off. J. Eur. Union L70 (44) (2006). http://eur-lex.europa.eu/legal-content/NL/TXT/?uri¼CELEX% 3A32006R0401. A. INTERNATIONAL, Guidelines for validation of qualitative binary chemistry methods, J. AOAC Int. 97 (2014) 1492e1495. E. Trullols, I. Ruisanchez, F.X. Rius, Validation of qualitative analytical methods, Trac-Trend Anal. Chem. 23 (2004) 137e145. S. Monbaliu, C. Van Poucke, C. Detavernier, F. Dumoulin, M. Van De Velde, E. Schoeters, S. Van Dyck, O. Averkieva, C. Van Peteghem, S. De Saeger, Occurrence of mycotoxins in feed as analyzed by a multi-mycotoxin LC-MS/ MS method, J. Agric. Food Chem. 58 (2010) 66e71. E. Commission, Commission decision 2002/657/EC of 12 August 2002, Off. J. Eur. Union L221 (2002) 8e36. http://eur-lex.europa.eu/legal-content/HR/ALL/ ?uri¼CELEX:32002D0657.