Analytica Chimica Acta 850 (2014) 57–64

Multi-detection method for five common microalgal toxins based onthe use of microspheres coupled to a flow-cytometry system

María Fraga a, Natalia Vilariño a,*, M. Carmen Louzao a, Laura P. Rodríguez a,Amparo Alfonso a, Katrina Campbell b, Christopher T. Elliott b, Palmer Taylor c,Vítor Ramos d, Vítor Vasconcelos d, Luis M. Botana a,*aDepartamento de Farmacología, Facultad de Veterinaria, Universidad de Santiago de Compostela, 27002 Lugo, Spainb Institute for Global Food Security (IGFS), School of Biological Sciences, Queen's University Belfast, David Keir Building, Stranmillis Road, Belfast BT9 5AG,Northern Ireland, UKcDepartment of Pharmacology, Skaggs School of Pharmacy and Pharmaceutical Sciences, University of California at San Diego, La Jolla, CA 92093-0657,United Statesd Interdisciplinary Centre of Marine and Environmental Research, CIIMAR, and Faculty of Sciences, University of Porto, Rua dos Bragas 289, Porto 4050-123,Portugal

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

� Multiplexed method for the detec-tion of five microalgal toxin classes.

� Sensitive, easy-to-perform, rapid,semi-quantitative screening tech-nique.

� Useful for the detection of freshwatertoxins in cyanobacterial samples.

A R T I C L E I N F O

Article history:Received 17 June 2014Received in revised form 12 August 2014Accepted 17 August 2014Available online 21 August 2014

Keywords:Multi-detectionAquatic toxinsMicroalgal toxinsMicrosphere-based arrayFlow-cytometry systemScreening method

A B S T R A C T

Freshwater and brackish microalgal toxins, such as microcystins, cylindrospermopsins, paralytic toxins,anatoxins or other neurotoxins are produced during the overgrowth of certain phytoplankton and benthiccyanobacteria, which includes either prokaryotic or eukaryotic microalgae. Although, further studies arenecessary to define the biological role of these toxins, at least some of them are known to be poisonous tohumans and wildlife due totheiroccurrence in these aquatic systems.The World Health Organization (WHO)has established as provisional recommended limit 1 mg of microcystin-LR per liter of drinking water. In thiswork we present a microsphere-based multi-detection method for five classes of freshwater and brackishtoxins: microcystin-LR (MC-LR), cylindrospermopsin (CYN), anatoxin-a (ANA-a), saxitoxin (STX) and domoicacid (DA). Five inhibition assays were developed using different binding proteins and microsphere classescoupled to a flow-cytometry Luminex system. Then, assays were combined in one method for thesimultaneousdetectionofthetoxins.TheIC50'susingthismethodwere1.9 � 0.1 mg L�1MC-LR,1.3 � 0.1 mg L�1

CYN, 61 �4 mg L�1 ANA-a, 5.4 � 0.4 mg L�1 STX and 4.9 � 0.9 mg L�1 DA. Lyophilized cyanobacterial culturesamples were extracted usinga simple procedure and analyzed by the Luminex methodandby UPLC–IT-TOF-MS. Similar quantification was obtained by both methods for all toxins except for ANA-a, whereby theestimated content was lower when using UPLC–IT-TOF-MS. Therefore, this newly developed multiplexeddetectionmethodprovidesarapid,simple,semi-quantitativescreeningtoolforthesimultaneousdetectionoffive environmentally important freshwater and brackish toxins, in buffer and cyanobacterial extracts.

ã 2014 Elsevier B.V. All rights reserved.

Contents lists available at ScienceDirect

Analytica Chimica Acta

journa l home page : www.e l sev ier .com/ loca te /aca

* Corresponding authors. Tel.: +34 982822233; fax: +34 982822233.E-mail addresses: natalia.vilarino@usc.es (N. Vilariño), luis.botana@usc.es (L.M. Botana).

http://dx.doi.org/10.1016/j.aca.2014.08.0300003-2670/ã 2014 Elsevier B.V. All rights reserved.

58 M. Fraga et al. / Analytica Chimica Acta 850 (2014) 57–64

1. Introduction

Many phytoplankton species belonging to different groups ofmicroalgae, such as diatoms, dinoflagellates or cyanobacteria(formerly known as blue–green algae) have been described astoxin producers in freshwater, brackish and marine ecosystems[1,2]. Cyanobacteria have a wide geographical distribution andinclude numerous genera involved in the production of toxins(i.e., cyanotoxins), mainly in freshwater environments [3].Cyanobacterial harmful algal blooms have been associated withhuman and animal poisonings, which highlights the hazard tohuman health, the environment and the economy associated tofreshwater resources. Cyanotoxins include microcystins (MCs),cylindrospermopsin (CYN), anatoxin-a (ANA-a) and saxitoxin(STX), among few others [4]. Additionally, the marinetoxin domoic acid (DA) has been also described in brackishwaters. In this case, the microalgal producers are diatoms thatbelong to the genera Pseudo-nitzschia and Nitzschia [2]. Thesetoxins are usually classified by their chemical structures (Fig. 1)and toxin activity. MCs are cyclic heptapeptides that share ageneral D-Ala-X-D-MeAsp-Z-Adda-D-Glu-Mdha structure [5]. Var-iations of the peptides yield about 90 structural variants of MCs[6]. The amino acids Adda, D-Glu and Mdha play an important rolein the interaction with the catalytic center of protein phosphatasesand therefore, these toxins block the access of the enzymesubstrate [7]. CYN is a tricyclic alkaloid with a natural epimer,7-epicylindrospermopsin, and one analog occurring at theC7 hydroxyl group, 7-deoxycylindrospermopsin (deoxyCYN)[8,9]. CYN has been described as a protein synthesis inhibitorthat affects mainly the liver [10]. ANA-a is a potent agonist of thenicotinic acetylcholine receptors (nAChR) [11,12]. ANA-a and itsanalog, called homoanatoxin-a, are unstable so they will beoxidized and converted into epoxy and dihydro degradationproducts, depending on the environmental conditions [13,14].Paralytic shellfish poisoning toxins (PSTs) are a group of neuro-toxins with a common tetrahydropurine backbone that block iontransport by voltage-dependent sodium-channels. The represen-tative toxin of the PST class is STX, although more than 50 analogshave been described, some of them produced specifically byfreshwater cyanobacteria [15]. DA and its, at least, 9 isomers(epi- and iso-domoic A–H) constitute the class of amnesic shellfishpoisoning toxins (ASTs) [16]. DA binds to and activates kainatereceptors, a subclass of glutamate receptors [17].

To protect human health, a few countries have establishedlimits for the concentration of some freshwater toxins in drinkingand recreational waters [18], mostly based on World HealthOrganization (WHO) guidelines that provisionally recommend asupper limit 1 mg of microcystin-LR (MC-LR) per liter of drinking

Fig. 1. Structure of five common microalgal toxins: (A

water [19]. The development of detection methods for freshwatertoxins is necessary to comply with legal limits. Actually, manydifferent methods have been developed for the detection of thesetoxins, including bioassays, molecular and analytical techniques,some of them allowing multi-detection [20–25]. The current trendin the aquatic toxins field is the development of inexpensive andefficient multiplexed screening methods in which the same sampleis analyzed for the presence of multiple analytes in one assay [26].Multi-detection screening methods save sample volume andhasten the acquisition of results reducing the number of analysisto be performed with confirmatory and/or quantitative methods.

Flow-cytometry technology has been widely employed inclinical and research fields for the development of multiplexedassays in which different analytes present in the same sample aresimultaneously detected [27]. Luminex technology uses micro-sphere classes with different spectral properties and surfacecarboxyl groups for covalent ligand attachment. Each microsphereclass becomes specific for an analyte through the immobilizationof a specific ligand. A fluidic system separates individual micro-spheres where respective red and green lasers distinguishmicrosphere class and quantify the attached compound. Multi-plexing is provided by the incubation of a sample with multipleclasses of analyte-specific microspheres simultaneously.

In this work, a semi-quantitative microsphere-based multi-detection method for five freshwater and brackish microalgaltoxins has been developed using the Luminex technology. Thismethod integrates previously developed assays for STX and DA[28,29] with newly developed assays for MC, CYN and ANA-a toxinsinto a multi-detection assay. The performance of this microsphere-based method has been evaluated by comparison withultra-performance liquid chromatography coupled to an iontrap-time of flight mass spectrometer (UPLC–IT-TOF-MS).

2. Material and methods

2.1. Materials

Certified reference standard material of DA was obtained fromCIFGA (Lugo, Spain). Certified reference standard material of STXdihydrochloride was obtained from the Institute for MarineBiosciences, National Research Council (Halifax, Canada). DA forimmobilization was purchased from Merck Millipore (Darmstadt,Germany) and CIFGA (Lugo, Spain). ANA-a, CYN, MC-LR, MC-YRand MC-RR were obtained from ENZO (Farmingdale, NY).Biotin-a-bungarotoxin (a-BTX) was from Molecular Probes(Eugene, OR). Analytical standard of MC-LR, MC-YR and MC-RR,N-hydroxysuccinimide (NHS), sodium tetraborate decahydrate,jeffamine (2,20-(ethylenedioxy)bis(ethylamine)), ethylenediamine,

) MC-LR, (B) CYN, (C) ANA-a, (D) STX and (E) DA.

M. Fraga et al. / Analytica Chimica Acta 850 (2014) 57–64 59

boric acid, sodium phosphate monobasic, ethanolamine andTween-20 were purchased from Sigma–Aldrich (Madrid, Spain).1-ethyl-3-(3-dimethylaminopropyl)carbodiimide hydrochloride(EDC) was purchased from Pierce (Rockford, IL). The anti-MC-LR(MC-LR-Ab), anti-CYN (CYN-Ab), anti-DA (DA-Ab) and anti-gonyautoxin 2/3 (GT-13A-Ab) antibodies were obtained aspreviously described [21,22,30,31]. The acetylcholine bindingprotein from Lymnaea stagnalis (Ls-AChBP) was purified andcharacterized as previously described [32,33]. The followingselection of nine cyanobacterial strains from the LEGE (Laboratoryof Ecotoxicology, Genomics and Evolution) Culture Collection(CIIMAR, Porto, Portugal) were cultured and lyophilized asdescribed by Sanchez et al. [34]: Microcystis aeruginosa strainsLEGE 00063, 05195 and 91341; Aphanizomenon ovalisporum strainLEGE X-001; Cylindrospermopsis raciborskii strains LEGE 97047 and99043; Anabaena sp. strains LEGE X-002, 00233 and 00234.Phycoerythrin goat anti-mouse Ig antibody (PE-Ab) was purchasedfrom Invitrogen (Eugene, OR), phycoLink streptavidin-R phycoer-ythrin (PE-SA) from Prozyme (Hayward, CA) and sodium azidefrom Fluka (Steinheim, Germany). Carboxylated microspheres(LC10019–01, LC10027–01, LC10038–01, LC10050–01 andLC10054–01) were from Luminex Corporation (Austin, TX).Luminex sheath fluid, multiscreen 96 well filter plates, 33 mmMillex filter with 0.22 mm pore size, 0.45 mm pore size ultrafree-MC (Durapore1 membrane) or ultrafree-CL centrifugal filters (lowbinding Durapore1 PVDF membrane) were purchased fromMillipore (Billerica, MA). Acetonitrile and methanol were fromhigh performance liquid chromatography (HPLC) grade and formicacid from American Chemical Society (ACS) grade. Reagent gradesolvents and buffer constituents were employed. High puritywater was from a Millipore Direct 8/16 water system (Billerica,MA). Phosphate-buffered saline solution (PBS) was 130 mM NaCl,10 mM NaPO4, pH 7.4. PBS-BT solution was PBS supplemented with0.1% w/v BSA and 0.1% v/v Tween-20.

2.2. Toxin or binding protein immobilization on the microspheresurface

MC-LR, CYN, STX, DA and Ls-AChBP were covalently attached tothe carboxylated surface of five different microsphereclasses: LC10050-01, LC10019-01, LC10038-01, LC10054-01 andLC10027-01, respectively. The protocols for the immobilization ofSTX (National Research Council, Canada) and DA (Merck Milliporeand CIFGA) were described in Fraga et al. [29]. The same protocolswere followed with slight modifications for the attachment of CYN(ENZO) and MC-LR (ENZO). Briefly, for covalent binding of CYN,2 � 106microspheres were activated by adding 150 mL of 150 mgmL�1 EDC and 23 mg mL�1 NHS dissolved in water. After 30 min ofincubation, this mixture was removed and 20% jeffamine in boratebuffer (pH 8.5) was added for 1 h. After jeffamine removal, 41.5 mgfree-CYN in 35 mL of H2O and 16 mL of 37% formaldehyde wereadded to pre-activated microspheres and allowed to react for 24 h.Subsequently, unbound toxin was removed and unreacted carboxylgroups were inactivated by 1 M ethanolamine-HCl during 30 min.For MC-LR immobilization carboxylated microspheres wereactivated using a solution of 154 mg mL�1 EDC and 46 mg mL�1

NHS. Later 1 M ethylenediamine in borate buffer (pH 8.5) wasadded for 1 h. The free NHS-ester groups were inactivated by 1 Methanolamine-HCl for 20 min. Pre-activation of toxin involved20 mg of free-MC-LR in 10 mL of DMSO diluted with 40 mL of27 mg mL�1 EDC and 12 mg mL�1 NHS in 10 mM sodium acetatebuffer (pH 4.5). Ethanolamine was removed and the pre-activatedtoxin was added to the microspheres and allowed to react for 4 h.The immobilization of Ls-AChBP was described in Rodríguez et al.[35]. At the end of each toxin immobilization, each microsphereclass was washed with PBS and stored in PBS with 0.01% sodium

azide at 4 �C in the dark. To achieve maximum specific binding, theimmobilized toxin-microspheres should be allowed to equilibratefor one week with at least one storage solution renewal.Ls-AChBP-microspheres were stored in PBS-BT without azideand used right away. Incubations were in the dark with constantshaking (700 rpm).

2.3. Microsphere-based multi-detection assay for MC-LR, CYN, ANA-a,STX and DA

Five competition assays were performed simultaneously in thesame well. The procedure started with a first incubation of 60 mL ofthe sample or calibration solution with 60 mL of the bindingproteins (0.1 mg mL�1 MC-LR-Ab, 1.3 pg mL�1 CYN-Ab, 1 mg mL�1

GT-13A-Ab, 0.6 mg mL�1 DA-Ab and 3 � 103 pre-washed Ls-AChBP-coated microspheres in a filter plate well). After 1 h of incubation,100 mL of this mixture was transferred to a second microtiter filterplate containing previously washed toxin-coated microspheres(2 � 103MC-LR-microspheres + 2 � 103 CYN-microspheres + 2 � 103

STX-microspheres + 2 � 103 DA-microspheres). One hour later100 mL of a-BTX (50 nM) was added for 30 min. Finally, followinga thorough wash, 100 mL of PE-Ab (0.5 mg mL�1) and PE-SA(4 mg mL�1) were added for 1 h. After a washing step, themicrospheres were suspended in 100 mL of PBS-BT. All theincubations were carried out in the dark with constant shaking(700 rpm).

2.4. Quantification of binding signals

PE-fluorescence attached to the surface of the microsphereswas quantified with a Luminex 200TM analyzer (LuminexCorp,Austin, TX). Microspheres were classified with a 635 nm laser andPE-fluorescence was quantified with a 532 nm laser. The acquisi-tion volume was 75 ml and the minimum number for bead countwas 100.

2.5. Extraction method for cyanobacterial samples

Freeze-dried cyanobacterial strains were extracted with asimple procedure. 20 mg were reconstituted in 2 mL of 75%aqueous methanol, vortexed, sonicated (6 � 30 s) and centrifugedfor 10 min at 3000 � g at 20 �C. The pellet was twice re-extractedwith 2 mL 75% aqueous methanol and centrifugation. Supernatantswere pooled together. This extract was further diluted in PBS-BT toappropriate toxin content and filtered (0.45 mm) for the micro-sphere-based assay. The same extracts were analyzed by UPLC–IT-TOF-MS after their dilution in 10% aqueous methanol and filtrationthrough a 0.22 mm filter.

2.6. Detection of MCs, CYN and ANA-a by UPLC–IT-TOF-MS

An UPLC–IT-TOF-MS with an electrospray ionization (ESI)interface (Shimadzu, Kyoto, Japan) was used for the identificationand quantification of MC-LR, MC-RR, MC-YR, CYN and ANA-a. UPLCseparation was performed with an ACQUITY UPLC HSS T3 column(2.1 i.d. �100 mm, 1.8 mm particle size, 100 Å pore size, Waters,Milford, MA) coupled to an in-line filter kit. Mobile phases A and Bwere water and acetonitrile, respectively, both acidified with 0.1%formic acid. The UPLC flow rate was 0.45 mL min�1 for a binarygradient of A and B, and temperature was maintained at 35 �C. Thegradient of mobile phase B was as follows: 5% for 1 min, 5–60% over9 min, 80% for 1 min, 80–5% over 3 min and a final equilibration ofthe column at 5% for 5 min. All MSnwere operated in positive modewith the following ESI source conditions: nebulizing gas flow,1.5 L min�1; curved desolvation-line and heat block temperature,230 �C; drying gas pressure, 100 kPa; and detector voltage, 1.65 kV.

60 M. Fraga et al. / Analytica Chimica Acta 850 (2014) 57–64

Each toxin was analyzed using MS1 and MS2with ion accumulationtimes of 10 ms and 30 ms, respectively. For the quantification andidentification of the toxins, the following transitions wereemployed (�1.5 Da ion precursor width, quantification transitionsin italics): 995.54 > 599.35 and 995.54 > 553.31 for MC-LR (collisionenergy (E) 26%, collision gas (Cg) 21%); 519.78 > 440.23and 519.78 > 298.17 for MC-RR (E 22%, Cg 17%); 1045.52 > 599.35and 1045.52 > 603.29 for MC-YR (E 25%, Cg 15%); 416.12 > 336.17and 416.12 > 194.13 for CYN (E 20%, Cg 24%); 166.12 > 149.10 and166.12 > 131.08 for ANA-a (E 14%, Cg 5%). Previously reportedmisidentification of ANA-a with phenylalanine [36] was avoided bychromatographic separation and fragment characterization ofphenylalanine in these UPLC–IT-TOF-MS conditions. Calibrationsolutions for MC-LR, MC-RR, MC-YR, CYN and ANA-a were preparedfrom stock solutions of 1 mg L�1 by further dilution in 10% aqueousmethanol. Calibration levels ranged between 0.01 and 0.75 mg L�1.The injection volume was 10 mL. Sodium trifluoroacetate was usedfor mass calibration. The limits of detection (LOD) and quantifica-tion (LOQ) were: 0.05 and 0.15 mg L�1 for MC-LR, 0.03 and0.08 mg L�1 for MC-RR, 0.09 and 0.28 mg L�1 for MC-YR, 0.04 and0.12 mg L�1 for CYN and 0.03 and 0.10 mg L�1 for ANA-a,respectively.

2.7. Safety

Toxins should be handled with gloves and eye protection at alltimes. Appropriate disposal methods should also be utilized.

2.8. Data analysis

All experiments were performed in duplicate. The Student'st-test for unpaired data was used for statistical analysis. Calibrationcurves for the microsphere-based method were fitted usingGraphPad Prism 5.0 by a four-parameter logistic equation obtainedwith a nonlinear regression fitting procedure: Y = Rhi + (Rlo� Rhi)/(1 + 10^ ((Log IC50� X) � HillSlope)), where Rhi is the bottom or theresponse at infinite concentration, Rlo is the top or the response at0 concentration, IC50 is the half maximal inhibitory concentrationand X is the logarithm of concentration. Calibration curves andquantification of samples performed by UPLC–IT-TOF-MS werefitted using the LabSolutions LCMS software (Shimadzu, Kyoto,Japan).

3. Results and discussion

3.1. Optimization of five microsphere-based assays for the detection offreshwater and brackish toxins

Five microsphere-based assays were individually optimized todetect MC-LR, CYN, ANA-a, STX and DA. Four inhibition immuno-assays consisted of the competition of toxins attached to the

Table 1Calibration curve parameters for MC-LR, CYN, ANA-a, STX and DA when using microspmaximum (Max) and non-specific (Min) binding signals. Concentrations are expresse(mean � SEM, n = 3, *statistically significant difference from individual detection).

IC20(mg L�1) I

MC-LR Individual 0.6 � 0.2

Multiplex 0.6 � 0.1

CYN Individual 0.7 � 0.1

Multiplex 0.7 � 0.1

ANA-a Individual 24 � 5

Multiplex 30 � 4

STX Individual 1.4 � 0.2 4Multiplex 1.6 � 0.3 5

DA Individual 1.4 � 0.3

Multiplex 1.9 � 0.2 4

microspheres (MC-LR, CYN, STX or DA) with toxins in solution(sample or calibration solution) for binding to specific antibodies(MC-LR-Ab, CYN-Ab, GT-13A-Ab or DA-Ab). The ANA-a assay wasbased on the competition of ANA-a with biotin-labeled a-BTX forbinding to the Ls-AChBP, previously immobilized on the micro-sphere surface. Specific antibody and a-BTX bound to themicrospheres were finally quantified using PE-Ab and PE-SA.Optimization was based on adjusting the immobilization proce-dure, antibody/detector-protein dilutions and/or incubation times.

Immobilization of a toxin or binding protein on the microspheresurface was the first optimization step. Immobilization resultswere evaluated with the Luminex system through the maximumbinding signal (Max) displayed after contact with a specificantibody or a-BTX. STX and DA immobilization had beenpreviously optimized for high Max [29] (Table 1). For MC-LRtwo different immobilization couplings, through the amine orcarboxyl moieties of the toxin, were tested. Immobilizationthrough carboxyl moieties provided a higher Max (Table 1) thanthrough amine moieties and were used for further assaydevelopment. The protocol for the immobilization of CYN, beingdevoid of a carboxyl moiety, was similar to that one used for STX(Table 1). Finally, for the optimization of the Ls-AChBP attachment,concentrations from 0.32 to 6.4 mg mL�1 were tested. Higher Maxwas achieved with 3.2 mg mL�1 Ls-AChBP (Table 1).

Optimization of single inhibition assays was done for MC-LR, CYNand ANA-a following the protocols for STX and DA previouslypublished,andalreadyincludedinamulti-detectionassayformarinetoxins [29]. With the final aim of multiplexing, some experimentalconditions were maintained constant for the four immunoassays(MC-LR, CYN, STX and DA): 1 h incubation times and 0.5mg mL�1

PE-Ab concentration. In these conditions several primary antibodyconcentrations were tested for each toxin: 0.09–0.18 mg mL�1

MC-LR-Ab, 0.95–1.9 pg mL�1 CYN-Ab, 0.2–2 mg mL�1 GT-13A-Aband 0.02–1.9 mg mL�1 DA-Ab. Binding signals and calibration curveswere evaluated for adequate performance. Calibration solutionsranged from 0.01 to 50 mg L�1 for MC-LR, from 0.004 to 20 mg L�1 forCYN, from 0.3 to 3000 mg L�1 for ANA-a, from 0.04 to 370 mg L�1 forSTX and from 0.003 to 310 mg L�1 for DA. The final primary antibodyconcentrations, 0.1 mg mL�1 MC-LR-Ab, 1.3 pg mL�1 CYN-Ab, 1 mgmL�1 GT-13A-Ab and 0.6 mg mL�1 DA-Ab, were selected consideringsufficient Max/non-specific binding signal (Min) ratio and the IC50 ofthe calibration curves (Fig. 2A, B, D and E, and Table 1). Finally, theLs-AChBP-based assay optimized for spirolide detection [35] wastested for ANA-a detection (Fig. 2C and Table 1).

To explore the assay capabilities in terms of sensitivity, longerincubation times and higher antibody dilutions were tested forMC-LR detection. Overnight incubations for the competition stepwere combined with lower antibody concentrations, ranging from4 to 18 ng mL�1. The highest sensitivity, with a dynamic range from0.09 to 1.66 mg L�1 and an IC50 of 0.3 mg L�1, was provided byovernight incubation of free toxin (sample or calibration solution)

here-based single- and multi-detection methods. Dynamic ranges (IC20–IC80), IC50,d in mg L�1, maximum and minimum binding signals in raw response units (RU)

C50(mg L�1) IC80(mg L�1) Max, Min (RU)

1.8 � 0.2 5.5 � 0.1 1639 � 161, 44 � 21.9 � 0.1 6.2 � 0.4 1394 � 100, 38 � 21.4 � 0.2 2.7 � 0.2 350 � 20, 5 � 0.41.3 � 0.1 2.6 � 0.3 260 � 7*, 5 � 0.593 � 11 389 � 20 950 � 58, 86 � 261 � 4 * 130 � 9 * 2030 � 289*, 78 � 6.2 � 0.4 15 � 0.3 443 � 64, 4 � 0.4.4 � 0.4 25 � 7 363 � 30, 29 � 5*

5.1 � 1.1 18 � 3 2912 � 141, 25 � 6.9 � 0.9 13 � 4 2544 � 148, 26 � 6

Fig. 2. Effects of multiplexing on the calibration curves of MC-LR, CYN, ANA-a, STXand DA when using the microsphere-based Luminex method. Single- and multi-detection calibration curves in buffer solution were obtained simultaneously for (A)MC-LR, (B) CYN, (C) ANA-a, (D) STX and (E) DA (mean � SEM, n = 3, *statisticallysignificant difference from individual detection).

M. Fraga et al. / Analytica Chimica Acta 850 (2014) 57–64 61

with 18 ng mL�1 MC-LR-Ab, followed by 2 h incubation withMC-LR-coated microspheres (data not shown).

3.2. Optimization of the multi-detection microsphere-based assay

Once the experimental protocol for individual detection wasoptimized, the five assays were combined in a multi-detectionmethod. The final assay consisted of the incubation of five bindingproteins with the calibration solution or sample, this mixture waslater added to toxin-coated microspheres, and finally detected bythe addition of a-BTX and the PE-labeled proteins. Individual andmulti-detection assays were compared using calibration curvesobtained simultaneously with calibration solutions containing amixture of MC-LR, CYN, ANA-a, STX and DA (Fig. 2). Multi-detection and single assays performed similarly (Fig. 2). The IC50,dynamic ranges (IC20–IC80), curve slopes, Max and Min forindividual and multiplexed assays did not show statisticallysignificant differences for MC-LR, CYN, STX and DA, except forCYN Max that was slightly lower in multi-detection (Fig. 2A, B, Dand E, Table 1). The calibration curve obtained for ANA-a in multi-detection was quite different from the curve obtained in the singledetection assay showing a statistically different steepness of theslope, a reduction of the dynamic range and a slight increase ofsensitivity (Fig. 2C, Table 1). The modification of ANA-a calibrationcurve in multi- versus single-detection might be due tonon-specific antibody binding to the a-BTX which would beconsistent with the significant increase of Max raw binding signals(Table 1). Although the characteristics of the assay change withmultiplexing, as shown for other microsphere-based detectionmethods [37], the assay is still useful to detect ANA-a.

LODs (estimated using the IC20, Table 1 [38,39]) of this multi-detection microsphere-based method are close to previouslypublished immunoassays [40]. Nowadays, provisional WHOrecommendations establish that the upper limit of MC-LR indrinking water should be 1 mg L�1 [19]. Considering that the LOD ofthe microsphere-based method for MC-LR is 0.6 mg L�1, the assaywould have enough sensitivity to detect MC-LR at the recom-mended limit. Additionally its sensitivity could be increased withlonger incubations and lower antibody concentrations, loweringthe LOD of the MC-LR assay to 0.09 mg L�1. Although the sensitivityfor MC-LR is adequate for current recommendations, futureadjustments for all toxins may be necessary with the developmentof new regulations.

3.3. Assay cross-reactivity

The cross-reactivity of the assay for MC-LR, MC-RR and MC-YR,three common MCs, was evaluated using the microsphere-basedmethod. A calibration curve was obtained for each analog withcalibration solutions prepared in PBS-BT. The calibration solutionswere simultaneously assayed in separate wells using the MCdetection protocol. The immunoassay provided similar calibrationcurves for the three toxins, with IC50 values of 1.20, 2.72 and1.79 mg L�1 for MC-LR, MC-RR and MC-YR, respectively (Fig. 3). Thepercentage of cross-reactivity of the MC-LR-Ab for the threeanalogs was calculated considering MC-LR the reference toxin:% cross reactivity (% CR) = (IC50 of MC-LR/IC50 of MC analog) � 100.Cross-reactivity values for MC-RR and MC-YR were 44% and 67%,respectively (Fig. 3). Therefore, the MC assay can detect at leastthree analogs of this toxin class. Very little information is availableabout MCs relative toxicity. The estimated intraperitoneal LD50

values in mice for MC-LR, -RR and -YR (around 50, 300 and110.6 mg kg�1, respectively) indicate that MC-LR is the most potentof the three analogs [23,41]. Therefore, considering these data, thecross-reactivity of the MC-Ab used in this study is an acceptablematch to the relative toxicity of these analogs. Actually,

Fig. 3. Cross-reactivity of MC-LR-immunoassay for MC-LR, MC-RR and MC-YR. (A)Calibration curves for MC-LR, MC-RR and MC-YR obtained simultaneously using themicrosphere-based Luminex assay. (B) IC50, dynamic range (IC20–IC80) andpercentage of cross reactivity for the three toxins. The percentage of cross-reactivity was calculated considering MC-LR as reference toxin. The cross-reactivityis expressed in percentage and the concentrations in mg L�1 (mean � SEM, n = 3).

62 M. Fraga et al. / Analytica Chimica Acta 850 (2014) 57–64

the detection of several toxic analogs is an important feature of theassay, since 1998, WHO recommendations are provisional coveringonly MC-LR based on the lack of toxicological data onother microcystins [42]. This guideline will probably evolve inthe near future to recommend limits of other microcystins, andother freshwater toxins, as it is already happening with legalregulations of several countries [4,18,42].

In regards to the other toxin groups, the multi-detectionmicrosphere-based method is capable of detecting more than oneanalog per group. The GT-13A-Ab has been demonstrated to detectseveral PST analogs in marine samples, but poor cross-reactivitywith N-1 hydroxy PST analogs was repeatedly reported[28,31,38,43]. STX, decarbamoyl (dc) STX, gonyautoxin (GTX)2/3, GTX1/4, dcGTX2/3, neosaxitoxin (NEO), and C1/2, amongothers, have been found in freshwater ecosystems and theirtoxicities are assumed equal to the same compounds from marineorigin [4,44]. Ideally, the cross-reactivity of the antibody shouldmatch the toxicity equivalency factor for all the toxins of the group.Unfortunately some of the N-1 hydroxy analogs, such as NEO, areamong the more toxic PSTs. Toxicity evaluation by this assay wouldimprove with an antibody specific for N-1 hydroxy PSTs. Samplescreening with the current PST assay will require re-testing ofnegative samples by analytical methods, until a N-1 hydorxy PSTassay is available. The previous characterization of the CYN-Abindicated 2–11% cross-reactivity for deoxyCYN versus CYN [22].DeoxyCYN did not show lethal toxicity by intraperitonealadministration in mice, contrary to CYN, which had an estimatedLD50 of 200 mg kg�1 [9,45]. Ls-AChBP binds to ANA-a and othertoxins from different origins that could be detected with this assay,however none of them has been described in brackish orfreshwaters to the extent of our knowledge [32,35,46]. Finally,there are no data that could provide information about the cross-reactivity of DA and ANA-a assays for other members of these toxinclasses. The individual assays compiled in the multiplexedtechnique were specific since none of them detected toxins ofthe other classes included in the multi-detection method.

3.4. Comparative analysis of cyanobacterial strains by themicrosphere-based assay and UPLC–IT-TOF-MS

Freeze-dried cyanobacterial biomass was extracted and analyzedby the microsphere-based assay and UPLC–IT-TOF-MS. A simpleextraction method using aqueous methanol was developed torecover the five toxin classes and to warrant compatibility withprotein-binding assays [28,29,35]. Aqueous methanol was used asextraction solution due to the hydrophilic nature of these toxins, andto avoid interferences generated by ethanol [29]. Preliminary testsusing two different extraction solutions containing two ratios ofmethanol:water (1:3 and 3:1) showed that a higher content ofmethanol provideda slightly higher toxin recovery (data not shown).A cyanobacterial sample from a strain negative for the presence ofthese toxins (LEGE 233) was extracted for matrix interference tests.Several dilutions of LEGE 233 extract, 1:10, 1:100 and 1:1000 (v/v),were assayed with the multi-detection microsphere assay. Max andMin signals obtained with these sample dilutions were compared tosignalsobtainedsimultaneouslyinbuffer.Dilution1:10(v/v)showedinterferences with Max signal associated to CYN-, STX- and LsAChBP-coated microspheres (data not shown). Therefore, calibration curvesof the five toxins were obtained simultaneously in buffer and in a1:100 dilution of the LEGE 233 extract. The calibration curvesdemonstrated the lack of interference of cyanobacterial extract forthe five assays (Fig. A.1). Cyanobacterial samples from in vitrocultures were assayed for the presence of the five toxin groups usingthe microsphere-based assay. All samples were assayed using a1:100 dilution of the initial extract, however, positive samples wereout of the linear range of the assay. Therefore, positive samples werere-assayed using dilutions of the extract ranging from 1:1000 to1:10000. The same extracts were analyzed for MCs, CYN and ANA-aby UPLC–IT-TOF-MS (Fig. A.2–A.5). Both techniques were used toestimate the toxin contents of the nine strains (Table 2). The finalresults fromtheLuminexassay werereportedasrepresentativetoxinequivalents. The estimated contents of MCs and CYN were similar forboth methods (Table 2). Toxic and non-toxic strains from a samecyanobacterial species were discriminated by both Luminex andUPLC–IT-TOF techniques, although this LC–MS/MS (liquid chroma-tography–tandem mass spectrometry) method was not optimizedfor high sensitivity. The analysis of the strains positive for MCproduction by UPLC–IT-TOF-MS demonstrated the presence ofMC-LR, MC-RR and MC-YR (Fig. A.3). These MCs were quantifiedusing analytical standards and the amount of the three moleculeswas added to report the total content of MCs for comparisonwith theLuminex assay (Table 2). Quantification of ANA-a in LEGE X-002 wasalmost four times higher when using the microsphere-based assaythan when using UPLC–IT-TOF-MS. The quantification byUPLC–IT-TOF-MS was done considering only ANA-a. However, theLEGE X-002 sample contained also a mass of m/z 182.12 (M + H+)consistent with the presence of epoxy-ANA-a. Its fragmentation(MS2) yielded transitions described for epoxy-ANA-a (Fig. A.5) [47].The epoxy-ANA-a amount cannot be calculated due to the lack ofanalytical/certified standards. Additionally, the analysis of extractsfrom LEGE X-002 using a QTRAP LC/MS/MS instrument points to theproduction of three additional ANA-a analogs by this cyanobacterialstrain [34]. Therefore, in these experimental conditions it is clearthat the UPLC–IT-TOF-MS method underestimated the amount ofANA-a toxins. In addition to this demonstrated underestimation ofANA-a toxins by LC–MS, it is also possible that the Luminex assayoverestimates the amount of ANA-a toxins. Cross-reactivity couldnot be determined for the members of this toxin class, and a higheraffinity of some analogs for the Ls-AChBP would certainly yield anoverestimation of toxin content when compared to ANA-a standard.

This microsphere-based method can be considered a useful toolto detect the presence of cyanotoxins in cyanobacterial samples.The sensitivity, calculated considering our extraction protocol and

Table 2Quantification of MC, CYN and ANA-a in cyanobacterial samples by the multiplexed microsphere-based Luminex assay or UPLC–IT-TOF-MS. Nine cyanobacterial strains wereassayed by the microsphere-based method for MC, CYN, ANA-a, DA and STX and analyzed by UPLC–IT-TOF-MS for MC-LR, MC-RR, MC-YR, CYN and ANA-a. Concentrations areexpressed in mg of toxin per mg of biomass. For the Luminex assay toxin content is expressed as representative toxin equivalents (toxin eq.). For all samples the content of STXand DA by the Luminex assay was below the limit of detection (<LOD). For the UPLC–IT-TOF-MS method the total amount of MCs is the result of adding the amounts of MC-LR,MC-RR and MC-YR quantified versus analytical standards of the three compounds (amount of individual analogs in brackets below total amount). The total amount of ANA-a isthe result of ANA-a quantification without considering other members of the class that cannot be quantified (mean � SEM, n = 3).

Cyanobacterial strains mg tox per mg of biomass

MCs CYN ANA-a

LuminexMC-LR eq.

UPLC–IT-TOF-MS(MC-LR + MC-RR + MC-YR)

LuminexCYN eq.

UPLC–IT-TOF-MS LuminexANA-a eq.

UPLC–IT-TOF-MS

Cylindrospermopsis raciborskii LEGE 97047 <LOD <LOD 1.9 � 0.1 1.9 � 0.3 <LOD <LODAphanizomenon ovalisporum LEGE X-001 <LOD <LOD 3.4 � 0.2 2.9 � 0.7 <LOD <LODCylindrospermopsis raciborskii LEGE 99043 <LOD <LOD <LOD <LOD <LOD <LODMicrocystis aeruginosa LEGE 00063 1.9 � 0.01 2.0 � 0.7 (0.6 + 0.5 + 0.9) <LOD <LOD <LOD <LODMicrocystis aeruginosa LEGE 05195 3.3 � 0.2 2.5 � 0.1 (0.8 + 0.5 + 1.2) <LOD <LOD <LOD <LODMicrocystis aeruginosa LEGE 91341 <LOD <LOD <LOD <LOD <LOD <LODAnabaena sp. LEGE X-002 <LOD <LOD <LOD <LOD 7.2 � 0.7 1.9 � 0.2Anabaena sp. LEGE 00233 <LOD <LOD <LOD <LOD <LOD <LODAnabaena sp. LEGE 00234 <LOD <LOD <LOD <LOD <LOD <LOD

M. Fraga et al. / Analytica Chimica Acta 850 (2014) 57–64 63

the minimum dilution compatible with the assay, was0.014 mg mg�1 of lyophilized cyanobacteria for MC-LR,0.016 mg mg�1 for CYN, 0.67 mg mg�1 for ANA-a, 0.036 mg mg�1

for STX and 0.043 mg mg�1 for DA. Moreover, the multiplexed assaycould also be used for the analysis of water samples from the fieldafter optimization of a sample preparation method, which wouldrequire organic solvents and sonication for cell disruption and,possibly, minor concentration of the extract due to the highsensitivity of the assay.

These results demonstrate that the multiplexed method wouldprovide a cost-effective, rapid evaluation of the presence of fivetoxin classes. Each batch of 2 � 106 analyte specific microspheresallows the analysis of up to 400 samples in duplicates. Thesimultaneous estimation of the toxins content in 40 pre-extractedsamples can be performed in 5 h. Sample number can be increasedthrough the use of additional microtiter plates.

Other options for multi-detection are analytical methods, suchas LC–MS, which offer accurate quantification and identification ofmultiple toxin classes and analogs as far as certified referencestandards are available [20,24,48]. For this work, UPLC–IT-TOF wasselected in order to identify different analogs of these toxin classeseven if analytical standards were not available. However, forfreshwater toxin detection triple quadrupoles are more often usedfor LC–MS/MS detection, which provide higher sensitivities thanTOF mass spectrometers [20,24,48,49], but the analysis of40 samples in 5 h is not possible. Additionally, analytical methodsrequire the use of expensive instrumentation and highly trainedpersonnel, which increases the cost of sample analysis. Recently, amultiplexed method has been developed for the simultaneousimmuno-detection of MC-LR, CYN and STX using a microfluidicchip [25], however, the detector device is not commerciallyavailable yet.

4. Conclusion

These results demonstrate the performance of a microsphere-based multi-detection assay as a semi-quantitative screening toolfor the detection of freshwater/brackish toxins belonging to MC,CYN, ANA-a, PST and AST classes. This method would allow savingprecious resources such as time and sample volume due to thesimultaneous detection of five toxin classes. Additionally, thisassay is easy-to-perform, rapid and suitable for microalgal sampleswith a previous simple extraction procedure. In fact, the

co-occurrence and apparent bioaccumulation of CYN and PSTsupport the need for the development of multi-detection methodsof freshwater microalgal toxins [50,51].

Conflict of interest

The authors declare no competing financial interest.

Acknowledgments

The research leading to these results has received fundingfrom the following FEDER cofunded-grants. From CDTI andTechnological Funds, supported by Ministerio de Economía yCompetitividad, AGL2012-40185-CO2-01 and Consellería deCultura, Educación e Ordenación Universitaria, GRC2013-016,and through Axencia Galega de Innovación, Spain,ITC-20133020 SINTOX, IN852A 2013/16-3 MYTIGAL. From CDTIunder ISIP Programme, Spain, IDI-20130304 APTAFOOD.

From the European Union's Seventh Framework Programmemanaged by REA–Research Executive Agency (FP7/2007-2013)under grant agreement Nos. 265,409 mAQUA, 315,285 CIGUA-TOOLS and 312,184 PHARMASEA. The project MARBIOTECH-NORTE-07–0124-FEDER-000047 and the PortugueseGovernmental Foundation for Science and Technology (FCT)throught the projectsPesT-C/MAR/LA0015/2013. This researchwas partly funded from a European Sustainable Programme2007–2013 under the European Regional Development Fund(“ERDF”) as the Advanced Asset Project.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.aca.2014.08.030.

References

[1] H.W. Paerl, R.S. Fulton, P.H. Moisander, J. Dyble, Harmful freshwater algalblooms, with an emphasis on cyanobacteria, Sci. World J. 1 (2001) 76–113.

[2] Y. Kotaki, N. Lundholm, H. Onodera, K. Kobayashi, F.F.A. Bajarias, E.F. Furio,M. Iwataki, Y. Fukuyo, M. Kodama, Wide distribution of Nitzschia navis-varingica, a new domoic acid-producing benthic diatom found in Vietnam,Fisheries Sci. 70 (2004) 28–32.

[3] H.W. Paerl, V.J. Paul, Climate change: links to global expansion of harmfulcyanobacteria, Water Res. 46 (2012) 1349–1363.

64 M. Fraga et al. / Analytica Chimica Acta 850 (2014) 57–64

[4] S. Merel, D. Walker, R. Chicana, S. Snyder, E. Baures, O. Thomas, State ofknowledge and concerns on cyanobacterial blooms and cyanotoxins, Environ.Int. 59 (2013) 303–327.

[5] W.W. Carmichael, V. Beasley, D.L. Bunner, J.N. Eloff, I. Falconer, P. Gorham, K.Harada, T. Krishnamurthy, M.J. Yu, R.E. Moore, et al., Naming of cyclicheptapeptide toxins of cyanobacteria (blue–green algae), Toxicon 26 (1988)971–973.

[6] L. Pearson, T. Mihali, M. Moffitt, R. Kellmann, B. Neilan, On the chemistrytoxicology and genetics of the cyanobacterial toxins, microcystin, nodularin,saxitoxin and cylindrospermopsin, Mar. Drugs 8 (2010) 1650–1680.

[7] S.R. Pereira, V.M. Vasconcelos, A. Antunes, Computational study of the covalentbonding of microcystins to cysteine residues-a reaction involved in theinhibition of the PPP family of protein phosphatases, FEBS J. 280 (2013)674–680.

[8] R. Banker, S. Carmeli, M. Werman, B. Teltsch, R. Porat, A. Sukenik, Uracil moietyis required for toxicity of the cyanobacterial hepatotoxin cylindrospermopsin,J. Toxicol. Environ. Health 62 (2001) 281–288.

[9] R.L. Norris, G.K. Eaglesham, G. Pierens, G.R. Shaw, M.J. Smith, R.K. Chiswell, A.A.Seawright, M.R. Moore, Deoxycylindrospermopsin, an analog of cylindrosper-mopsin from Cylindrospermopsis raciborskii, Environ. Toxicol. 14 (1999)163–165.

[10] K. Terao, S. Ohmori, K. Igarashi, I. Ohtani, M.F. Watanabe, K.I. Harada, E. Ito, M.Watanabe, Electron microscopic studies on experimental poisoning in miceinduced by cylindrospermopsin isolated from blue–green alga Umezakianatans, Toxicon 32 (1994) 833–843.

[11] C.E. Spivak, B. Witkop, E.X. Albuquerque, Anatoxin-a: a novel potent agonist atthe nicotinic receptor, Mol. Pharmacol. 18 (1980) 384–394.

[12] P. Thomas, M. Stephens, G. Wilkie, M. Amar, G.G. Lunt, P. Whiting, T. Gallagher,E. Pereira, M. Alkondon, E.X. Albuquerque, et al., (+)-Anatoxin-a is a potentagonist at neuronal nicotinic acetylcholine receptors, J. Neurochem. 60 (1993)2308–2311.

[13] K.J. James, A. Furey, I.R. Sherlock, M.A. Stack, M. Twohig, F.B. Caudwell, O.M.Skulberg, Sensitive determination of anatoxin-a, homoanatoxin-a and theirdegradation products by liquid chromatography with fluorimetric detection,J. Chromatogr. A 798 (1998) 147–157.

[14] D.K. Stevens, R.I. Krieger, Stability studies on the cyanobacterial nicotinicalkaloid anatoxin-a, Toxicon 29 (1991) 167–179.

[15] M. Wiese, P.M. D'Agostino, T.K. Mihali, M.C. Moffitt, B.A. Neilan, Neurotoxicalkaloids: saxitoxin and its analogs, Mar. Drugs 8 (2010) 2185–2211.

[16] J. Clayden, B. Read, K.R. Hebditch, Chemistry of domoic acid, isodomoic acids,and their analogues, Tetrahedron 61 (2005) 5713–5724.

[17] J.T. Slevin, J.F. Collins, J.T. Coyle, Analogue interactions with the brain receptorlabeled by [3H]kainic acid, Brain Res. 265 (1983) 169–172.

[18] I. Chorus, in: Chorus I (Ed.), Current Approaches to Cyanotoxin RiskAssessment, Risk Management and Regulations in Different Countries, FederalEnvironment Agency (Umweltbundesamt), Dessau-Rosslau, 2012.

[19] WHO, Guidelines for Drinking-water Quality. Addendum to Volume 2, HealthCriteria and Other Supporting Information, second ed., WHO Publishing,Geneva, 1998.

[20] J. Chen, T. Yan, J. Xu, S. He, P. Zhao, X. Yan, Simultaneous determination of toxinsin algae and water samples by high-performance liquid chromatography withtriple quadrupole mass spectrometry, J. Sep. Sci. 35 (2012) 1094–1101.

[21] S. Devlin, J.P. Meneely, B. Greer, C. Greef, M.J. Lochhead, C.T. Elliott, Nextgeneration planar waveguide detection of microcystins in freshwater andcyanobacterial extracts, utilising a novel lysis method for portable samplepreparation and analysis, Anal. Chim. Acta 769 (2013) 108–113.

[22] C.T. Elliott, C.H. Redshaw, K. George, First development and characterisation ofpolyclonal and monoclonal antibodies to the emerging fresh water toxincylindrospermopsin, Harmful Algae 24 (2013) 10–19.

[23] N. Gupta, S.C. Pant, R. Vijayaraghavan, P.V.L. Rao, Comparative toxicityevaluation of cyanobacterial cyclic peptide toxin microcystin variants (LR, RRYR) in mice, Toxicology 188 (2003) 285–296.

[24] M. Maizels, W.L. Budde, A LC/MS method for the determination ofcyanobacteria toxins in water, Anal. Chem. 76 (2004) 1342–1351.

[25] J. Zhang, S. Liu, P. Yang, G. Sui, Rapid detection of algal toxins by microfluidicimmunoassay, Lab Chip 11 (2011) 3516–3522.

[26] N. Vilarino, M.C. Louzao, M. Fraga, L.P. Rodriguez, L.M. Botana, Innovativedetection methods for aquatic algal toxins and their presence in the foodchain, Anal. Bioanal. Chem. 405 (2013) 7719–7732.

[27] J.S. Kim, F.S. Ligler, Utilization of microparticles in next-generation assays formicroflow cytometers, Anal. Bioanal. Chem. 398 (2010) 2373–2382.

[28] M. Fraga, N. Vilarino, M.C. Louzao, K. Campbell, C.T. Elliott, K. Kawatsu, M.R.Vieytes, L.M. Botana, Detection of paralytic shellfish toxins by a solid-phaseinhibition immunoassay using a microsphere-flow cytometry system, Anal.Chem. 84 (2012) 4350–4356.

[29] M. Fraga, N. Vilarino, M.C. Louzao, P. Rodriguez, K. Campbell, C.T. Elliott, L.M.Botana, Multidetection of paralytic, diarrheic, and amnesic shellfish toxins by

an inhibition immunoassay using a microsphere-flow cytometry system, Anal.Chem. 85 (2013) 7794–7802.

[30] T.F. McGrath, K. Andersson, K. Campbell, T.L. Fodey, C.T. Elliott, Development ofa rapid low cost fluorescent biosensor for the detection of food contaminants,Biosens. Bioelectron. 41 (2013) 96–102.

[31] K. Campbell, L.D. Stewart, G.J. Doucette, T.L. Fodey, S.A. Haughey, N. Vilarino, K.Kawatsu, C.T. Elliott, Assessment of specific binding proteins suitable for thedetection of paralytic shellfish poisons using optical biosensor technology,Anal. Chem. 79 (2007) 5906–5914.

[32] S.B. Hansen, Z. Radic, T.T. Talley, B.E. Molles, T. Deerinck, I. Tsigelny, P. Taylor,Tryptophan fluorescence reveals conformational changes in the acetylcholinebinding protein, J. Biol. Chem. 277 (2002) 41299–41302.

[33] S.B. Hansen, T.T. Talley, Z. Radic, P. Taylor, Structural and ligand recognitioncharacteristics of an acetylcholine-binding protein from Aplysia californica,J. Biol. Chem. 279 (2004) 24197–24202.

[34] J.A. Sanchez, P. Otero, A. Alfonso, V. Ramos, V. Vasconcelos, R. Araoz, J. Molgo,M.R. Vieytes, L.M. Botana, Detection of anatoxin-a and three analogs inAnabaena spp. cultures: new fluorescence polarization assay and toxin profileby LC–MS/MS, Toxins (Basel) 6 (2014) 402–415.

[35] L.P. Rodriguez, N. Vilarino, J. Molgo, R. Araoz, M.C. Louzao, P. Taylor, T. Talley, L.M. Botana, Development of a solid-phase receptor-based assay for thedetection of cyclic imines using a microsphere-flow cytometry system, Anal.Chem. 85 (2013) 2340–2347.

[36] M. Gugger, S. Lenoir, C. Berger, A. Ledreux, J.C. Druart, J.F. Humbert, C. Guette, C.Bernard, First report in a river in France of the benthic Cyanobacteriumphormidium favosum producing anatoxin-a associated with dog neurotox-icosis, Toxicon 45 (2005) 919–928.

[37] J. Peters, M. Bienenmann-Ploum, T. de Rijk, W. Haasnoot, Development of amultiplex flow cytometric microsphere immunoassay for mycotoxins andevaluation of its application in feed, Mycotoxin Res. 27 (2011) 63–72.

[38] K. Campbell, S.A. Haughey, H. van den Top, H. van Egmond, N. Vilariño, L.M.Botana, C.T. Elliott, Single laboratory validation of a surface plasmon resonancebiosensor screening method for paralytic shellfish poisoning toxins, Anal.Chem. 82 (2010) 2977–2988.

[39] E. Usleber, R. Dietrich, C. Burk, E. Schneider, E. Martlbauer, Immunoassaymethods for paralytic shellfish poisoning toxins, J. AOAC Int. 84 (2001)1649–1656.

[40] M. Weller, Immunoassays and biosensors for the detection of cyanobacterialtoxins in water, Sensors 13 (2013) 15085–15112.

[41] Y.-M. Chen, T.-H. Lee, S.-J. Lee, H.-B. Huang, R. Huang, H.-N. Chou, Comparisonof protein phosphatase inhibition activities and mouse toxicities of micro-cystins, Toxicon 47 (2006) 742–746.

[42] M.D. Burch, Effective doses, guidelines and regulations, Adv. Exp. Med. Biol.619 (2008) 831–853.

[43] E.S. Fonfría, N. Vilariño, K. Campbell, C. Elliott, S.A. Haughey, B. Ben-Gigirey, J.M. Vieites, K. Kawatsu, L.M. Botana, Paralytic shellfish poisoning detection bysurface plasmon resonance-based biosensors in shellfish matrixes, Anal.Chem. 79 (2007) 6303–6311.

[44] M.E. van Apeldoorn, H.P. van Egmond, G.J.A. Speijers, G.J.I. Bakker, Toxins ofcyanobacteria, Mol. Nutr. Food Res. 51 (2007) 7–60.

[45] R. Li, W.W. Carmichael, S. Brittain, G.K. Eaglesham, G.R. Shaw, Y. Liu, M.M.Watanabe, First report of the cyanotoxins cylindrospermopsin and deoxy-cylindrospermopsin from Raphidiopsis curvata (cyanobacteria), J. Phycol. 37(2001) 1121–1126.

[46] Y. Bourne, Z. Radic, R. Araoz, T.T. Talley, E. Benoit, D. Servent, P. Taylor, J. Molgo,P. Marchot, Structural determinants in phycotoxins and AChBP conferring highaffinity binding and nicotinic AChR antagonism, Proc. Natl. Acad. Sci. U. S. A.107 (2010) 6076–6081.

[47] K.J. James, J. Crowley, B. Hamilton, M. Lehane, O. Skulberg, A. Furey, Anatoxinsand degradation products, determined using hybrid quadrupole time-of-flightand quadrupole ion-trap mass spectrometry: forensic investigations ofcyanobacterial neurotoxin poisoning, Rapid Commun. Mass Spectrom. 19(2005) 1167–1175.

[48] S.A. Oehrle, B. Southwell, J. Westrick, Detection of various freshwatercyanobacterial toxins using ultra-performance liquid chromatography tandemmass spectrometry, Toxicon 55 (2010) 965–972.

[49] M. Holcapek, R. Jirasko, M. Lisa, Recent developments in liquid chromatogra-phy–mass spectrometry related techniques, J. Chromatogr. A 1259 (2012)3–15.

[50] J.P. Berry, A. Jaja-Chimedza, L. Davalos-Lind, O. Lind, Apparent bioaccumulationof cylindrospermopsin and paralytic shellfish toxins by finfish in lakeCatemaco (Veracruz, Mexico), Food Addit. Contam. Part A Chem. Anal. ControlExpo. Risk Assess. 29 (2012) 314–321.

[51] J.P. Berry, O. Lind, First evidence of paralytic shellfish toxins and cylindro-spermopsin in a Mexican freshwater system, Lago Catemaco, and apparentbioaccumulation of the toxins in Tegogolo snails (Pomacea patula catema-censis), Toxicon 55 (2010) 930–938.