Towards SPR biosensing combined with bioaffinity assisted nano-HILIC TOFMS identification of PSP toxins

Trends Trends in Analytical Chemistry, Vol. 28, No. 6, 2009

Towards Surface PlasmonResonance biosensing combinedwith bioaffinity-assisted nano HILICLiquid Chromatography / Time-of-flight Mass Spectrometryidentification of Paralytic ShellfishPoisonsG.R. Marchesini, H. Hooijerink, W. Haasnoot, J. Buijs,

K. Campbell, C.T. Elliott, M.W.F. Nielen

The potential for coupling technologies to deliver new, improved forms of bioanalysis is still in its infancy. We review a number of

examples in which coupling has been successful, with special emphasis on combining surface-plasmon-resonance biosensors with

mass spectrometry. We give an overview of current progress towards combining biosensor-based bioanalysis with chemical

analysis for confirmation of paralytic shellfish poisons that are marine toxins. This comprehensive approach could be an alter-

native to the official methods currently used (e.g., animal testing and high-performance liquid chromatography with fluorescence

detection) and could serve as a model for many more such applications.

ª 2009 Elsevier Ltd. All rights reserved.

Keywords: Bioaffinity; Bioanalysis; Biosensor; Identification; Mass spectrometry; Nanotechnology; Paralytic shellfish poison; Surface-plasmon

resonance; Time of flight; Toxin

G.R. Marchesini*,

H. Hooijerink,

W. Haasnoot,

M.W.F. Nielen

RIKILT, Institute of Food Safety, P.O. Box 230, 6700 AE Wageningen, The Netherlands

J. Buijs

GE Healthcare Europe AB, Bjorkgatan 30, 75125 Uppsala, Sweden

K. Campbell,

C.T. Elliott

Institute of Agri-Food and Land Use (IAFLU), Queen�s University, David Keir Building, Stranmillis

Road, Belfast, Northern Ireland, United Kingdom

M.W.F. Nielen

Wageningen University, Laboratory of Organic Chemistry, Dreijenplein 8, 6703 HB Wageningen,

The Netherlands

*Corresponding author.E-mail: [email protected]

792 0165-9936/$ - see front matter ª 2009 Elsev

1. Introduction

Paralytic shellfish poisons (PSPs) are pro-duced by microalgae and recurrentlycontaminate shellfish [1]. Besides being apotentially lethal health hazards to sea-food consumers, PSPs can cause a greatdeal of economic damage due to closure ofshellfish-production areas [2]. Since the1970s, algal blooms have been expandingin frequency, intensity and geographicdistribution worldwide. Such phenomenahas various explanations, ranging fromeutrophication to increased transportationof dinoflagellate-resting cysts in ballastwater of commercial ships, to unusual

ier Ltd. All rights reserved. doi:10.1016/j.trac.2009.04.008

mailto:[email protected]

Trends in Analytical Chemistry, Vol. 28, No. 6, 2009 Trends

climatic conditions [3]. Furthermore, due to theimprovement in monitoring programs and analyticaltechniques, new PSPs have been discovered [4,5].

PSPs elicit their toxicity by binding to the transmem-brane voltage-gated sodium-channel receptor (NaCR)present in excitable cells (e.g., nerves, muscle fibers andcardiac fibers). Upon binding, sodium transport isblocked, thus slowing or stopping the propagation of theaction potential. After ingestion of PSP-contaminatedshellfish, and depending on the dose, the respiratorymuscle paralysis symptoms appear within minutes inhumans [1].

PSPs are a group of closely related tetrahydropurinecompounds, currently divided into three sub-groups: theN-sulfocarbamoyl, the decarbamoyl and the carbamatesub-groups, in increasing order of toxicity as tested in themouse bioassay [6]. Currently, the mouse bioassay is themain tool used for PSP-monitoring programs through-out the world [1,7]. Besides the in vivo assay, otherin vitro bioassays for the detection of PSPs include theuse of hippocampal neural tissue [8] and a neuroblas-toma cell-culture [9,10], both considered functional as-says. Several biochemical assays based on biospecificrecognition have also been described [i.e. radioligandbinding assay based on the NaCR [11,12], biosensorassays based on the NaCR or antibodies and ELISAs (forreview see [7,13])]. The affinity profile of NaCR andantibodies for PSPs are usually dissimilar, so data gen-erated by immunoassays cannot be expressed in a tox-icity equivalence factor as is the case with functionalassays.

During the past decade, a considerable number ofchromatography-based techniques were developed forPSP analysis. Liquid chromatography (LC), gas chro-matography (GC) and capillary electrophoresis (CE) areable to separate the analytes based on their physico-chemical properties. Unequivocal structural informationcan be obtained if these chromatographic techniques arehyphenated with a spectrometric detection method [e.g.,mass spectrometry (MS)]. The advantage of such anapproach over bioassays or biochemical assays is thepossibility to separate and to identify individual PSPs.The downside of this approach is the lack of informationabout the overall toxicity of the PSP mixture. Interfacinga fast functional biochemical method with a confirma-tory method could provide an alternative to animaltesting.

1.1. Biochemical testing combined with chemicalanalysisInterfacing a biochemical functional assay and a chro-matographic technique was pioneered by Moorhouseet al. in 1969 [14]. Moorhouse added a new dimensionto a GC-flame ionization detector by hyphenating theantenna of an insect as a parallel detector for the dis-covery of insect pheromones. This marked the beginning

of several successful studies in recent years couplingbiosensors or other bioassay technologies with analyticaltechniques (for a comprehensive review see [15,16]).

1.2. SPR biosensor screening combined with chemicalanalysisSurface-plasmon resonance (SPR) biosensors are andautomated and do not require specialized human re-sources for operation. Hence, a synergistic complementcan be obtained by combining SPR-biosensor technologybased on molecular biorecognition events with analyti-cal techniques based on the physico-chemical propertiesof the analytes. SPR biosensors were combined off-linewith micro-preparative high-performance LC (HPLC)[17], on-line as a detector for IgG when using CE forseparation [18], in parallel with LC-diode array detector(DAD)-MS for drug screening [19] and off-line and on-line with MS [20].

The choice of ionization technique amenable forhyphenating SPR biosensor technology with MS greatlydepends on the characteristics of the molecule to beanalyzed and on the assay format chosen to detect it.When a direct binding assay format is chosen, the ana-lyte is captured on the chip surface and can be eluted forfurther off-line or on-line LC-MS analysis. In contrast,with the inhibition biosensor immunoassay (iBIA) for-mat, the analyte interacting with the biorecognitionelement is usually discarded to waste.

The most commonly used technology in combinationwith SPR-biosensor assays is matrix-assisted laserdesorption ionization-time-of-flight-mass spectrometry(MALDI-TOF-MS). MALDI has the advantage of gener-ating intact vapor-phase ions from a solid of large mol-ecules co-crystalized with small volatile molecules(matrix). This gentle ionization technique in combina-tion with TOF-MS has achieved fmol sensitivity and al-lowed the MS analysis of large macromolecules (e.g.,proteins and protein complexes of megadalton (MDa)masses) [21]. Regardless of the sensitivity, the massaccuracy of MALDI-TOF-MS is in the range 0.1–0.01%for proteins above 30 kDa, hence it cannot yet be con-sidered a general method for protein identification [22].However, the identification of enzymatically-digestedproteins by peptide-mass fingerprinting based on data-base search is possible given the 10–50 ppm massaccuracy of MALDI-TOF-MS for small polypeptides [23].

Using the SPR/MALDI-TOF-MS combination for mac-romolecules has been extensively researched. SPRdetection and MALDI-MS confirmation of the peptidemycotoxin A from snake venom was achieved by directon-chip analysis [24], at the expense of yielding the chipunusable after one binding cycle. To overcome thislimitation, the analyte was eluted in a small volumecontained between two air bubbles and later respottedonto a MALDI plate [25–28]. Such an approach waslater improved with completely automated commands,

http://www.elsevier.com/locate/trac 793


including those for sample preparation needed prior toMALDI [29].

Although very limited, SPR/MALDI-TOF-MS has alsobeen applied to food analysis, for the detection of thelethal staphylococcal enterotoxin B in milk and mush-rooms [30,31]. Furthermore, imaging SPR has enabledSPR-MS array platforms for the high-throughputscreening of large numbers of interactants in the field ofproteomics [32].

The main shortcoming of MALDI was its mass reso-lution and accuracy. Nevertheless, with the advent ofnewer instrumentation, the main concerns shifted tosignal reproducibility and sample preparation. The latteraspect concerns the matrix interfering with ionizationand detection of small molecules [33]. Thus, the majordownside of SPR/MALDI-TOF-MS is its unreliability foranalyzing small molecular weight analytes (e.g., PSPtoxins).

As an alternative to MALDI, a different approach forSPR-MS analysis using a direct binding assay format, isthe elution of the captured analytes and their off-line oron-line analysis using electrospray ionization (ESI) cou-pled to MS. ESI is a gentle ionization technique thatproduces singly-charged or multiply-charged ions in agaseous phase directly from a liquid phase (either organicor aqueous). ESI overcomes the use of an interferingmatrix and solves in one step the transition of ions to gasphase needed for MS analysis, simplifying the interfacingwhen compared to MALDI. Previous research attemptedoff-line ESI-MS analysis of SPR eluates containing lowmolecular-weight protease inhibitors by a direct infusioninto a nano-ESI-ion trap-MS. However, it was not satis-factory, due to the chemical noise arising from the lack ofan analytical chromatographic step [34].

Recently, a similar off-line approach addressed thelack of chromatographic step using a larger chip surfaceand nano-ESI-LC-tandem MS (MS2) analysis of trypticdigests of the SPR eluate and successfully achieved theidentification of cyclic-nucleotide-binding proteins fromcell lysates [35].

An alternative to the laborious handling of the eluatesis to hyphenate the fluidic system of the biosensor di-rectly with a reversed-phase trapping column that issubsequently coupled to LC-ESI-MS analysis. This strat-egy has been successfully tested for MS-based amino-acidsequencing after on-chip digestion of proteins using asimple Biacore X instrument [36].

Bouffartigues et al. [37], attempted the analysis ofnucleoproteins using a Biacore 2000 instrument with asimilar system set-up. Unfortunately, the ion suppressioncaused by a buffer detergent prevented MS identification.

Finally, it was recently reported that proteins purifiedfrom a cell lysate were successfully identified using afully automated online SPR biosensor-nano-LC-MS2

system [38].

794 http://www.elsevier.com/locate/trac

1.2.1. SPR biosensor screening combined with chemicalanalysis of small molecular weight analytes in food. Manyof the analytes of interest in food analysis are drug res-idues or toxins, mainly of low molecular weight and arenot easily detectable with an SPR biosensor in a directbinding assay. These analytes require the use of the iBIAformat where the analytes of interest flow through to thewaste complexed with the biorecognition molecule. Al-though a large number of studies can be found in theliterature on SPR biosensor iBIA for food-relevant ana-lytes, its combination with MS is in its infancy and barelyreported. A complementary strategy is required forcoupling SPR iBIA with MS. One option is to use a split-flow strategy for the parallel coupling of the screeningand MS. Such an approach was proved using a yeastbioassay for the screening of androgens and estrogens incalf-urine combination with LC-ESI-quadrupole (Q)TOF-MS [39,40]. A similar parallel approach was used fordetection and confirmation of fluoroquinolone (FQ)antibiotics in chicken muscle tissue [41]. Two iBIAs,detecting six FQs, were used instead of the yeast bioassayand were able to screen a large number of samples. Onlywhen a sample was considered non-compliant duringscreening for FQs was it fractionated by LC and theeffluent was split toward two 96-well fraction collectors.The immunoactive LC fractions in one of the 96-wellplates were immuno-discriminated using the iBIAs, cre-ating an immunogram, and the positive well positionswere used in the second 96-well plate for immunoactive-oriented identification with ESI-TOF-MS.

Although successful, the iBIA screening-MS parallelstrategy had some limitations. First, it required a rela-tively large sample volume. Second, due to the amountof sample handling required, the chances of samplecontamination were high. Third, the fractionation reso-lution was limited, thus hampering identification, giventhat complex samples may contain hundreds of peaksseparated during the chromatographic step, but pooledback together in the same fraction. Fourth, the frac-tionation process is not feasible when screening for bio-active compounds whose half-life is short or unknown.

Considering these limitations, a system requiringminimum sample volume and handling was developedusing the antibiotic enrofloxacin as a model analyte[42]. This system was subsequently adapted to analysisof PSP toxins (Fig. 1). The SPR-MS system interface wasbased on a nanoscale affinity chip that linked iBIAscreening with a nano-LC-TOF-MS system by recoveringonly the relevant analytes. The same biospecific recog-nition molecule was used in the iBIA and as a biosorbentin the recovery chip, which specifically captured theanalyte while the sample matrix flowed through.

The operating procedure of the SPR/nano-LC-TOF-MSsystem comprised four stages. After the sample extractswere prepared (Fig. 1, frame 1), they were screened with

Figure 1. Overview of the surface-plasmon resonance (SPR)/nano-hydrophilic interaction liquid chromatography (HILIC) time-of-flight (TOF) massspectrometry (MS) (nano-HILIC-TOF-MS) system and procedure. The operating procedure of this system comprises four stages: the samples areprepared (1); the samples are screened using a screening chip in the paralytic shellfish poison (PSP) inhibition biosensor immunoassay (iBIA)(2); samples suspected of being non-compliant are reinjected over the recovery chip and the analyte is captured at sub-ng level, released and depos-ited into a vial (3); and, this eluate is subsequently diluted and injected into a loop-type interface to be analyzed with nano-HILIC-TOF-MS (4).


the iBIA (Fig. 1, frame 2) and only those suspected non-compliant were reinjected onto the recovery chip anal-ogous to an affinity chromatographic step on a chip.Once captured, the analyte was eluted into a vial andautomatically diluted (Fig. 1, frame 3) to be later injectedon to a loop-type interface. From the loop type interface,the analytes enter the nano-LC system and are retainedin a trapping column followed by an analytical columnconnected to the TOF MS for mass analysis (Fig. 1, frame4). The main features of this strategy were; minimumsample volume consumption and the possibility ofacquiring in depth information about the performance ofthe bio-affinity sorbent prior to use and for periodicalquality control by placing the recovery chip into the SPRbiosensor.

1.3. Towards an SPR/nano-LC-TOF-MS system for PSPtoxinsAdvances in the implementation of an off-line SPR/nano-LC-MS application for PSP toxins based on a recoverychip were explored. The first step was implementation ofthe PSP iBIA previously developed using a monoclonal

antibody (MAb) GT13A as the biorecognition element[13,43]. MAb GT13A is an antibody with a group-spe-cific PSP affinity, thus providing the possibility to detectvarious PSP toxins in the same assay with different sen-sitivity for each PSP [44]. The second step was to obtain arobust recovery chip with MAb GT13A immobilized onits surface for the recovery of the relevant PSPs. The finalstep was to set up a nano-hydrophilic interaction liquidchromatography (HILIC)-TOF-MS system suitable foranalyzing recovery chip eluates.

2. Experimental

2.1. Chemicals and materialsSensor chips (CM5), HBS-EP buffer [pH 7.4, consisting of10 mM 4-(2-hydroxyethyl) piperazine-1-ethanesulfonicacid, 150 mM sodium chloride, 3 mM EDTA, 0.005%(v/v) surfactant polysorbate (P20)], HBS-N buffer [pH7.4, consisting of 10 mM 4-(2-hydroxyethyl) piperazine-1-ethanesulfonic acid, 150 mM sodium chloride], anamine coupling kit [containing 0.1 M N-hydroxy-


Table 1. Cross-reactivity and sensitivity of MAb GT13A for PSPs in HBS-EP and HBSN buffer

PSP Toxin Abbreviation HBS-EP Buffera HBS-N Bufferb

IC50 (ng/ml) CR % IC50 (ng/ml) CR %

1 Saxitoxin STX 4.7 100 3.2 1002 Decarbamoylsaxitoxin DCSTX 3.6 131 1.8 177.83 Neosaxitoxin NEO P 94.7 6 5 699.5c 0.54 Decarbamoylneosaxitoxin DCNEO P 86.3 6 5 - -5 Gonyautoxin 1 and 4 GTX 1/4 P 80.9 6 6 228.3 1.46 Gonyautoxin 5 GTX 5 4.3 109 2.2 145.57 Gonyautoxin 2 and Gonyautoxin 3 GTX 2/3 3.6 131 1.8 177.88 Decarbamoylgonyautoxin 2 and 3 DCGTX 2/3 2.7 174 1.4 228.69 C1 and C2 Toxins C1/C2 3.4 138 - -

aBuffer pH 7.4 consisting of 10 mM 4-(2-hydroxyethyl) piperazine-1-ethanesulfonic acid, 150 mM sodium chloride, 3 mM EDTA, 0.005% (v/v)surfactant polysorbate (P20).

bBuffer pH 7.4, consisting of 10 mM 4-(2-hydroxyethyl) piperazine-1-ethanesulfonic acid and 150 mM sodium chloride.cExtrapolated value – not measured.


succinimide (NHS), 0.4 M N-ethyl-N-(3-dimethylamino-propyl) carbodiimide hydrochloride (EDC), and 1 Methanolamine hydrochloride-NaOH (pH 8.5) (EA)] andthe SPR-based biosensor system Biacore 3000, the sur-face-preparation unit (SPU) accessory as well as the fastprotein liquid chromatography system (AKTA purifier)and HI Trap protein G columns (1 mL) were supplied byGE Healthcare (Uppsala, Sweden). Acetonitrile (ACN)HPLC far-UV grade was from Lab-Scan Ltd. (Dublin,Ireland). Water was purified using a Milli-Q gradientA10 system (Millipore, Bedford, MA, USA). The 30 kDaand 50 kDa Amicon cut-off filters were from Millipore.All the PSP standards used throughout the study (seePSP toxin abbreviations in Table 1) were purchased fromthe National Research Council, Marine AnalyticalChemistry standards program (NRC-CRM), Institute forMarine Biosciences, Nova Scotia, Canada.

2.2. EquipmentThe system (Fig. 1) consisted of the Biacore 3000 SPRbiosensor, containing both a CM5 chip coated withsaxitoxin (STX) for the iBIA screening and the SPUaccessory containing the recovery chip, a CM5 chipcoated with MAb GT13A, for analyte capture. The FlowCell Carrier Type 2 (FCT2) on top of the recovery chipfeatured a serpentine-like flow cell (frame 3) (area 16mm2, height 50 lm and volume 800 nL). The plugcontaining the eluate from the recovery chip wasrecovered into a vial using the ‘‘MS recover’’ commandpresent in Biacore software. The eluate was diluted with12 lL of acetonitrile containing 1% formic acid and in-jected with a 10 lL Hamilton syringe into a 10 lL loop-type interface. The inlet of the loop interface was con-nected to the Agilent 1100 HPLC pump. The outlet ofthe loop-type interface was connected to the tee of anano-LC switching system (frame 4) adapted from Mei-ring et al. [45] using 1/16-inch connections having


short (0.02-inch ID) Peek sleeves and fused silica capil-lary tubing (360 lm OD, 150 lm ID, 60 cm). The flowwas split, with this switching system, between therestrictor (360 lm OD, 50 lm ID · 120 cm) and thetrapping column (TSK-GEL Amide-80 HILIC 5 lm50 lm ID, 360 lm OD, 2 cm length) between the first teeand the second tee. Following the second tee, the ana-lytical column (TSK-GEL Amide-80 HILIC 5 lm 50 lmID, 360 lm OD, 20 cm length) was directly butt-con-nected with the 30 lm emitter through the universalNanoFlow Sprayer mounted on the Waters model QTOF-micro MS system. The nano-LC gradient separation wasperformed using a mobile phase consisting of (A) ACN/water (95:5; v/v) and (B) water/ACN (95:5; v/v), bothmobile phases containing 2 mM ammonium formate and50 mM formic acid. Sample loading through the trap-ping column was performed at 10 lL/min for 5 min with100% solvent A. Upon valve switch, the precolumn splitwas activated and the restrictor back pressure allowed aflow of the order of 300 nL/min through the analyticalcolumn.

2.3. iBIA ProcedureFor the iBIA procedure, the sample extracts or standardsolutions in HBS-N buffer were mixed (4:1; v/v) withMAb GT13A in the biosensor and 50 lL were injected ata flow rate of 25 lL/min. The STX surface was regen-erated with 15 lL of a mixture of 0.2 M NaOH and ACN(4:1; v/v).

2.4. Recovery-chip preparationMAb GT13A was immobilized onto the entirecarboxymethylated surface of Biacore CM5 sensor chips.The chip surface was thoroughly rinsed with distilledwater and placed on to the surface-preparation unit. Thesurface was activated by injecting a mixture of 0.4 MEDC and 0.1 M NHS (1:1; v/v) for 20 min at 5 lL/min.


Finally, MAb GT13A antibody was diluted to a concen-tration of 50 lg/mL in acetate buffer (pH 5) and 100 lLwere injected onto the activated chip surfaces and theremaining active groups where deactivated by injecting35 lL of 1 M EA. The immobilization levels were eval-uated by measuring the relative responses of the chips inthe biosensor before and after immobilization.

2.5. SamplesSamples were obtained from mussel or cockle flesh (1 g)and extracted by homogenizing and adding 1.5 mLacetate buffer pH 5. After roller mixing for 30 min,the homogenate was centrifuged for 10 min at 3000 gand the supernatant was diluted (1:10; v/v) in HBS-Nbuffer. Two mL were passed through a 30 kDa cut-offfilter and the filtrate was stored at 4�C for furtherexperiments.

Figure 2. (A) Paralytic shellfish poison (PSP) inhibition biosensor immunoasscalibration curve; (B) PSP-toxin calibration curves in HBS-N buffer (0.01 M Htion curves for saxitoxin (STX) and a PSP mixture; and, (D) inhibition shown

Matrix-matched calibration curves were prepared withspiked mussel extracts and diluted in HBS-N buffer. TheEU action limit (AL) for PSP toxins is 800 lg STXequivalents/kg shellfish meat. The mussel homogenateswere spiked with either STX or a PSP mixture of 1 ALSTX equivalents (eqs.) [6] containing STX (50% eqs.,0.4 lg), DC STX (5% eqs., 0.78 lg), GTX 1/4 (25% eqs.,2.16 lg), GTX 2/3 (10% eqs., 1.86 lg) and dcGTX 2/3(10% eqs., 3.99 lg).

3. Results and discussion

One common problem when coupling the SPR biosensorwith MS is mobile-phase compatibility. To preserve bio-molecular interactions, the biorecognition element in thebiosensor usually needs a suitable buffer containing saltsthat might interfere with the MS analysis. Non-ionic

ay (iBIA) sensorgrams corresponding to the measurement of a saxitoxinEPES, pH 7.4,. 150 mM NaCl); (C) PSP iBIA matrix-matched calibra-by mussel and cockle sample extracts analyzed with the PSP iBIA.



surfactants (e.g., P20) are incompatible with MS analysisbecause they interfere with chromatographic analysis[42]. Nevertheless, P20 is used in the mobile phase ofbiosensors to avoid non-specific adsorption of matrix oranalytes to the fluidic system. Hence, the mobile phaseused in the biosensor analysis has to be carefully opti-mized and considered in the context of the type ofhyphenation chosen [46].

To prevent further interference, the PSP iBIA perfor-mance was evaluated using HSB-N [a simple MS com-patible buffer (10 mM HEPES (pH 7.4), 150 mM NaCl)].Calibration curves for seven of the toxins were measuredusing HBS-N buffer and compared with those previouslymeasured in HBS-EP that contained the P20 surfactant[13]. The sensorgrams showed no sign of non-specificbinding (Fig. 2A) and the chip could be re-used. All thecalibration curves could be fitted to the four-parameterfitting function used in BIA evaluation software(Fig. 2B). As shown in Table 1, the lack of surfactant inthe HBS-N buffer induced an overall increase in the PSPiBIA sensitivity for the highly cross-reacting toxins, STX,DCSTX, GTX 2/3, GTX 5 and DCGTX 2/3 (see Table 1 forPSP-toxin abbreviations). PSP toxins with lower affinityshow a similar trend; however, higher concentrations inHBS-EP would have to be tested to compare fairly.

PSP iBIA was considered suitable and was used for themeasurement of highly contaminated mussel and cocklesamples, which tested positive for PSP presence in themouse bioassay and HPLC fluorescence analysis com-plying with the AOAC method [47].

The measurement of the matrix-matched standardcalibration curves shows that, at the same AL level, thePSP mixture produced a much stronger inhibition thanthe STX (Fig. 2C). The strong inhibition shown by thePSP mixture is due to the strong cross-reactivity of MAbGT13A for GTX 2/3 and dcGTX 2/3, present in highconcentration in the PSP mixture. It is important topoint out that, if the NaCR had been used instead of MAbGT13A as a biorecognition element for this experiment,equal STX eqs. would in theory show similar inhibitionvalues.

Spiked mussel samples and mussel and cockle samplesconsidered non-compliant by reference methods (themouse bioassay and HPLC-fluorescence) were assayedwith the SPR biosensor in the PSP iBIA showing a highdegree of inhibition (Fig. 2D). Although the extractionprotocol used showed a 55–65% recovery efficiency,which is far from optimal, the mussel samples spiked atthe AL level before extraction showed around 50%inhibition in the PSP iBIA for STX. This indicated that,even under these extraction conditions, PSPs in thesesamples can be detected below regulatory levels.

3.1. Antibody GT13A–PSP toxin interactionGiven the good performance of MAb GT13A as biorec-ognition element in the iBIA, the logical step was to


evaluate its performance when covalently immobilizedas a biosorbent in a recovery-chip set-up. The interactionof MAb GT13A and the PSPs was studied in this directassay format using the SPR biosensor. In this directassay format, from the total 16 mm2 surface of therecovery chip, only 1 mm2 per flow channel are sensed.Given that the SPR signal depends on the mass of theanalyte within the sensing region, the signals expectedin this assay format are about an order of magnitudelower than those from the iBIA. This assay set-up al-lowed characterization of the biosorbent and tuning ofthe different parameters that favored the PSP-MAbGT13A interaction to achieve multi-analyte capture,retention and elution. Of particular interest was todetermine the conditions that would reduce the dissoci-ation constant (kd) to improve the half life of the analyteon the chip surface.

Fig. 3(A and B) shows the optimization of buffercomposition based on the maximum amount of STXbound to MAb GT13A, the inverse of the off rate (1/kd),the half-life of the interaction and the amount of STX onthe surface after a washing step with water to removenon-specifically bound compounds. The chosen ionicstrength and pH conditions indicated with an arrowshow that there was a strong ionic component in theinteraction between STX and MAb GT13A. An increasein the ionic strength of the buffer generated a strongdecrease in binding affinity, evidenced as a drop on themaximum amount of STX bound and an increase of thekd. A concentration of 150 mM was selected instead ofthe optimum below 100 mM, considering that thisinteraction occurs in a mussel-extract environment,likely to contain sea salts that will add to the total ionicstrength.

The pH also had a dramatic influence on the MAbGT13A–STX interaction (Fig. 3B). Any drop below pH7.4 causes the disruption of the interaction and a 1 pHpoint increase to pH 8.4 negatively influenced theinteraction. These results indicated that special careshould be taken to the uniformity of the sample extractssubmitted to the recovery chip.

Since HILIC was used in the nano-LC system, themaximum organic solvent concentration tolerable byMAb GT13A was of interest to achieve a low amount ofaqueous phase in the eluate. The resistance of MAbGT13A to a short pulse of organic solvent during elutionwas therefore evaluated (data not shown). The maxi-mum amount of acetonitrile tolerated by MAb GT13A inthe elution solution was 10%. Hence, a dilution of theeluate in organic solvent is required to achieve retentionin the HILIC trapping column.

3.2. Multi-analyte recovery chipTo assess the possibility of having a recovery chipcapable of capturing multiple analytes, the kineticparameters of seven PSPs were measured. The sensor-

Figure 3. Influence of mobile-phase conditions on the saxitoxin monoclonal antibody (STX-MAb) GT13A-interaction parameters in a direct assayset-up: (A) ionic strength; (B) pH; (C) binding and dissociation kinetics of different paralytic shellfish poisons (PSPs) to immobilized MAb GT13A.


grams obtained binding different PSP toxins to the MAbGT13A recovery chip are shown in Fig. 3C. All the PSPstested were able to bind, showing more than an order ofmagnitude difference in the Kd values (NEO and STX).Interestingly, when the chip surface was saturated withanalyte, the response normalized by the molecularweight should have yielded equal maximum responses.However, it was observed that the maximum responsesvaried almost four-fold (except for GTX 1/4 that did notreach equilibrium). It can be speculated that this was aneffect of the structural differences observed and not onlyof the molecular weight difference.

In general terms, MAb GT13A seems to be a robustbiorecognition element for a reusable recovery chip. Itremains to be thoroughly tested how these PSPs com-peting for MAb GT13A binding sites would influence

each others� binding kinetics and the performance of themulti-analyte recovery chip when simultaneously pres-ent in a sample extract.

3.3. nano-HILIC TOF MS analysis of mussel and cocklesamplesPSPs are highly hydrophilic, generally dicationic (butalso neutral and monocationic) compounds with a tri-cyclic structure and a dense arrangement of heteroatoms(some structures are shown in Fig. 3C). PSPs are alsonon-volatile and thermally labile, hence posing aninteresting chromatographic challenge, since theyremove the possibility of separation with GC or traditionalLC with UV detection. Previous studies on PSP-toxinseparation were usually based on either ion-pairing LCor HILIC [48,49]. The reagents needed for ion-pairing LC


Figure 4. (A) Background subtracted extracted ion-mass chromatograms of a recovery chip eluate obtained from the injection of an extract from amussel sample spiked with a mixture of paralytic shellfish poisons (PSPs); and, (B) MS spectrum where in-source collision-induced dissociation(CID) is observed for saxitoxin (STX).


hinder the possibility of MS detection due to ion sup-pression. However, PSP analysis with ion-pairing LCcoupled with post-column oxidation and fluorescentdetection has been successful [6]. HILIC seems to be analternative if confirmation with MS is desired. In thiscase, the retention of PSPs is based on the interactionbetween PSPs and an aqueous stagnant layer covering ahydrophilic stationary phase embedded in a mobilephase with a high percentage of the organic component.In this case, the sample has to be loaded in a highcontent of organic phase to achieve column retention.Therefore both the trapping column and the analyticalcolumn in the nano-HILIC-TOF-MS (Fig. 1 frame 4) wereof HILIC material and the recovery-chip eluate plug had


to be pulled backward from the recovery-chip flow cell into the needle and deposited into a vial where it wasautomatically diluted with acetonitrile. Further injectioninto the sample loop was manual, which proved to beeffective in this double HILIC column system.

Nano-HILIC TOF MS was used to analyze recovery-chip eluates containing PSPs recovered from mussel andcockle extracts, which were considered as suspected non-compliants (Fig. 2D). Fig. 4A exemplifies the extractedion current (XIC) chromatogram obtained from sucheluates with nano-HILIC-TOF-MS after subtraction of arelevant blank injection. Four of the five peaks observedin the XIC chromatograms correspond to either theparent [M+H]+ ion or a fragment ion of the PSP toxins

Table 2. Critical comparison of the PSP toxins detected with the recovery chip / nano-HILIC-TOF-MS and those used for spiking or measuredwith an independent method

PSP Toxin Observed m/z Samples

Spiked Mussel (1 AL MIX) Mussel A Cockle 37262

Spikedb Rt in RC/ nanoLC-MS Reference c Rta in RC/ nanoLC-MSd Referencec Rt in RC/ nanoLC MS

STX 300 0.4 15.52 0.17 14.33 0.87 15.52DCSTX 257 0.78 15.78 14.53 0.94 15.58NEO 316 0.13 14.41 15.52DCNEO - 0.93 n.d.

12.2GTX 1/4 [412] fi332 2.16 13.2 1.49 13.01 0.72 13.2GTX 5 - 9.54 n.d.GTX 2/3 [396] fi378 1.86 12.33 0.48 n.d 0.15 n.d.DCGTX 2/3 [353] fi273 3.99 13.72 0.35 n.d.C1/C2 - 0.86 n.d. 4.37 n.d.

aShifted Rts are due to a new column.bAdded in lg/g.cAnalyzed by HPLC-fluorescence in lg/g.dRetention time in XIC chromatogram of Fig. 4.


used for spiking the sample. The only peak that remainsunrecognized is one with a [M+H]+ ion mass of 255,which might correspond to a transition due to the loss of–SO3 and –H2O from DCGTX 3. However, the retentiontime appeared to be too low for DCGTX3, since it wasexpected before STX [48].

Comparison of the PSPs used for spiking or detectedwith the HPLC-reference method and those found in theeluate with nano-HILIC-TOF-MS are presented in Table2. A mussel sample (Mussel A) extract, which showedthe highest percentage of inhibition in the PSP iBIA(Fig. 3B), was injected into the MAb GT13A recoverychip for further nano-HILIC-TOF-MS analysis of theeluate. The recovery chip was capable of recovering atleast three out of the five PSPs found with the referenceLC-fluorescence method as shown by the XICchromatograms (data not shown). The DCSTX [M+H]+

ion (m/z = 257) was observed in this sample but was notreported by the reference method, indicating that DCSTXwas either not detected by the reference method, or thatthis [M+H]+ ion belongs to the in-source CID degrada-tion of GTX 5. The latter is unlikely, given the transitionsthat would need to occur {e.g., [380] fi 300 (-SO3) fi257 (-NHCO)}. However, it is interesting to note in thiscase that NEO was successfully recovered, even though itshowed a cross-reactivity of only 0.5% in the PSP iBIA.By contrast, when MAb GT13A was immobilized, NEOappeared to have an affinity with MAb GT13A highenough to saturate its binding sites rapidly (Fig. 3C). Thenano-HILIC-TOF-MS analysis of a recovery-chip eluatefrom the non-compliant cockle sample (37262) showedthat three out of the eight PSPs expected in the samplewere observed. The lack of these PSPs in the eluatetriggers the challenge of investigating whether these

PSPs were extracted at all, given that cockle is a differentmatrix, or captured by the MAb GT13A recovery chipbelow the limit of detection of the nano-HILIC-TOF-MSsystem.

Overall, it was noted that MAb GT13A recovery chipwas capable of capturing multiple PSPs on its surface,allowing their recovery. The capture capacity of therecovery chip is limited by the number of active bindingsites on its surface. Hence, it is obvious that a highernumber of analytes would lower the absolute amountcaptured and their concentration in the eluate. Thisimplies that sub-ng to pg sensitivity is required for thenano-HILIC-TOF-MS system. The poor chromatographyobtained in this nano-HILIC split-flow system hinderedproper separation of the toxins. A peak intensity rapidlydecaying with column use indicated a reduction in theretention capacity of the trapping column. After 12chromatographic runs, the intensity remained stable tothe level of the XIC chromatograms seen in Fig. 4A. Aside-effect of the poor capacity of these HILIC columnswas the necessity to keep the MS cone voltage at 40 V.This maximized the sensitivity and allowed detection ofthe PSPs, but generated in-source CID. An example isshown in Fig. 4B, where the mass spectrum depicts theloss of H2O from STX.

Decreasing column capacity, low separation powerand in-source CID hindered the possibilities of perform-ing MS2 for a correct MS identification.

Although HILIC is the best alternative available forhydrophilic compounds, such as the PSPs, there is lim-ited knowledge on the basic chemistry of HILIC. Thesecolumns usually have a lower performance and robust-ness compared to reversed-phase LC columns, mainlydue to their sensitivity to flow regularity and properconditioning [50]. Both parameters are difficult to opti-



mize with the split-flow nano-LC system used through-out the study. The current availability of better nano-LChardware (e.g., pumps capable of delivering a steadypressure at a flow rate of nL/min) looks promising forHILIC. Stable column capacity and separation power ofthe columns over time will allow a higher signal-to-noiseratio of the relevant [M+H]+ ions for performing MS2

and confirmation of the current toxins.

4. Conclusions and future outlook

The iBIA/recovery chip/nano-HILIC-TOF-MS couplingwas achieved for the first time for PSP toxins by using arecovery-chip interface, which had the biorecognitionelement immobilized on its surface, was able to capturethe relevant analytes and could be re-used. The analyteswere further eluted and analyzed with nano-HILIC-TOF-MS. PSP toxins in non-compliant samples were detectedusing this analytical strategy at and below the regula-tory limits during the screening of mussel and cocklesamples.

Further research should focus on implementing morerobust nano-LC hardware to improve the performance ofnano-HILIC-TOF-MS and the possibility of on-line dilu-tion, which will allow an automated transition betweenthe recovery chip and the loop-type interface.

We can conclude that this analytical approach hasgreat potential for the identification of emerging bio-toxins, especially if combined with functional receptorsas biorecognition elements.

AcknowledgementsThis project is part of the European Union project ‘‘Newtechnologies to screen multiple chemical contaminantsin foods’’ (acronym BioCop) and financially supported bythe European Commission (Contract FOOD-CT-2004-06988) and the Dutch Ministry of Agriculture, Natureand Food Quality. We are also thankful to Cowan Hig-gins from the Marine Biotoxin Unit, Chemical Surveil-lance Branch, Agri-Food and Biosciences Institute, forhis HPLC measurements of the contaminated musselsamples with the AOAC method.

References[1] H.P. van Hegmond, G.J.A. van Apeldoorn, G.J.A. Speijers, Marine

Biotoxins, Food and Agriculture Organization (FAO) of the United

Nations, Rome, Italy, 2004.

[2] X.A. Alvarez-Salgado, U. Labart, M.J. Fernandez-Reiriz, F.G.

Figueiras, G. Roson, S. Piedraco, R. Filgueir, J.M. Cabanas,

Harmful Algae 7 (2008) 849.

[3] G.M. Hallegraeff, in: G.M. Hallegraeff, D.M. Anderson, A.D.

Cembella (Editors), Manual on Harmful Marine Algae, Inter-

governmental Oceanographic Commission, Paris, France, 1995,

p. 1.


[4] P. Vale, I. Rangel, B. Silva, P. Coelho, A. Vilar, Toxicon 53 (2009)

176.

[5] A. Negri, D. Stirling, M. Quilliam, S. Blackburn, C. Bolch, I.

Burton, G. Eaglesham, K. Thomas, J. Walter, R. Willis, Chem. Res.

Toxicol. 16 (2003) 1029.

[6] Y. Oshima, J. AOAC Int. 78 (1995) 528.

[7] L.M. Botana, A. Alfonso, A. Botana, M.R. Vieytes, C. Vale, N.

Vilarino, C. Louzao, Trends Anal. Chem. 28 (2009), in press,

doi:10.1016/j.trac.2009.02.014 .

[8] D.S. Kerr, D.M. Briggs, H.I. Saba, Toxicon 37 (1999) 1803.

[9] P. Truman, R.J. Lake, J. AOAC Int. 79 (1996) 1130.

[10] A.R. Humpage, A. Ledreux, S. Fanok, C. Bernard, J.F. Briand, G.

Eaglesham, J. Papageorgiou, B. Nicholson, D. Steffensen, Environ.

Toxicol. Chem. 26 (2007) 1512.

[11] S.R. Davio, P.A. Fontelo, Anal. Biochem. 141 (1984) 199.

[12] L.E. Llewellyn, Chem. Res. Toxicol. 19 (2006) 661.

[13] K. Campbell, L.D. Stewart, G.J. Doucette, T.L. Fodey, S.A.

Haughey, N. Vilarino, K. Kawatsu, C.T. Elliott, Anal. Chem. 79

(2007) 5906.

[14] J.E. Moorhouse, R. Yeadon, P.S. Beevor, B.F. Nesbitt, Nature

(London) 223 (1969) 1174.

[15] H.A. Fishman, D.R. Greenwald, R.N. Zare, Annu. Rev. Biophys.

Biomol. Struct. 27 (1998) 165.

[16] M. Trojanowicz, M. SzewczyImageskac, Trends Anal. Chem. 24

(2005) 92.

[17] E. Nice, M. Lackmann, F. Smyth, L. Fabri, A.W. Burgess, J.

Chromatogr., A 660 (1994) 169.

[18] R.J. Whelan, R.N. Zare, Anal. Chem. 75 (2003) 1542.

[19] M. Minunni, S. Tombelli, M.A. Mascini, A. Bilia, C.M. Bergonzi,

F.F. Vincieri, Talanta 65 (2005) 578.

[20] D. Nedelkov, R.W. Nelson, Trends Biotechnol 21 (2003) 301.

[21] R.J. Wenzel, U. Matter, L. Schultheis, R. Zenobi, Anal. Chem. 77

(2005) 4329.

[22] C. Fenselau, Anal. Chem. 69 (1997) 661A.

[23] J.R. Yates 3rd, J. Mass Spectrom. 33 (1998) 1.

[24] J.R. Krone, R.W. Nelson, D. Dogruel, P. Williams, R. Granzow,

Anal. Biochem. 244 (1997) 124.

[25] C.P. Sonksen, E. Nordhoff, O. Jansson, M. Malmqvist, P. Roe-

pstorff, Anal. Chem. 70 (1998) 2731.

[26] D. Nedelkov, R.W. Nelson, Biosens. Bioelectron. 16 (2001) 1071.

[27] J. Borch, P. Roepstorff, Anal. Chem. 76 (2004) 5243.

[28] F. Lopez, C. Pichereaux, O. Burlet-Schiltz, L. Pradayrol, B.

Monsarrat, J.-P. Esteve, Proteomics 3 (2003) 402.

[29] A. Zhukov, M. Schurenberg, O. Jansson, D. Areskoug, J. Buijs, J.

Biomed. Technol. 15 (2004) 112.

[30] D. Nedelkov, A. Rasooly, R.W. Nelson, Int. J. Food Microbiol. 60

(2000) 1.

[31] D. Nedelkov, R.W. Nelson, Appl. Environ. Microbiol. 69 (2003)

5212.

[32] D. Nedelkov, Anal. Chem. 79 (2007) 5987.

[33] L.H. Cohen, A.I. Gusev, Anal. Bioanal. Chem. 373 (2002) 571.

[34] C. Sonksen, P. Roepstorff, P.U. Markgren, H. Danielson, M.D.

Hamalainen, O. Jansson, Eur. J. Mass Spectrom. 7 (2001) 385.

[35] N.F.C. Visser, A. Scholten, R.H.H. van den Heuvel, A.J.R. Heck,

ChemBioChem 8 (2007) 298.

[36] T. Natsume, H. Nakayama, O. Jansson, T. Isobe, K. Takio, K.

Mikoshiba, Anal. Chem. 72 (2000) 4193.

[37] E. Bouffartigues, H. Leh, M. Anger-Leroy, S. Rimsky, M. Buckle,

Nucleic Acids Res. 35 (2007) 1.

[38] T. Hayano, Y. Yamauchi, K. Asano, T. Tsujimura, S. Hashimoto,

T. Isobe, N. Takahashi, J. Proteome Res. 7 (2008) 4183.

[39] M.W.F. Nielen, T.F.H. Bovee, M.C. van Engelen, P. Rutgers,

A.R.M. Hamers, J.A. van Rhijn, L.A.P. Hoogenboom, Anal. Chem.

78 (2006) 424.

[40] M.W.F. Nielen, E.O. van Bennekom, H.H. Heskamp, J.A. van

Rhijn, T.H. Bovee, L.A.P. Hoogenboom, Anal. Chem. 76 (2004)

6600.

http://dx.doi.org/


[41] G.R. Marchesini, W. Haasnoot, P. Delahaut, H. Gercek, M.W.F.

Nielen, Anal. Chim. Acta 586 (2007) 259.

[42] G.R. Marchesini, J. Buijs, W. Haasnoot, D. Hooijerink, O. Jansson,

M.W.F. Nielen, Anal. Chem. 80 (2008) 1159.

[43] E.S. Fonfria, N. Vilarino, K. Campbell, C. Elliott, S.A. Haughey, B.

Ben-Gigirey, J.M. Vieites, K. Kawatsu, L.M. Botana, Anal. Chem.

79 (2007) 6303.

[44] K. Kawatsu, Y. Hamano, A. Sugiyama, K. Hashizume, T. Noguchi,

J. Food Prot. 65 (2002) 1304.

[45] H.D. Meiring, E. van der Heeft, G.J. ten Hove, P.J.M. de Jong, J.

Sep. Sci. 25 (2002) 557.

[46] H. Larsericsdotter, O. Jansson, A. Zhukov, D. Areskoug, S.

Oscarsson, J. Buijs, Proteomics 6 (2006) 2355.

[47] J.F. Lawrence, B. Niedzwiadek, C. Menard, J. AOAC Int. 88 (2005)

1714.

[48] C. Dell�Aversano, P. Hess, M.A. Quilliam, J. Chromatogr., A 1081

(2005) 190.

[49] J.F. Lawrence, C. Menard, J. Assoc. Off. Anal. Chem. 74 (1991)

1006.

[50] T. Ikegami, K. Tomomatsu, H. Takubo, K. Horie, N. Tanaka, J.

Chromatogr., A 1184 (2008) 474.


Towards SPR biosensing combined with bioaffinity assisted nano-HILIC TOFMS identification of PSP toxins

Documents