ilable at ScienceDirect

Toxicon 54 (2009) 62–66

Contents lists ava

Toxicon

journal homepage: www.elsevier .com/locate/ toxicon

Rapid extraction combined with LC-tandem mass spectrometry(CREM-LC/MS/MS) for the determination of ciguatoxinsin ciguateric fish flesh

Richard J. Lewis*, Aijun Yang, Alun JonesInstitute for Molecular Bioscience, The University of Queensland, Qld 4072, Australia

a r t i c l e i n f o

Article history:Received 17 October 2008Received in revised form 11 March 2009Accepted 12 March 2009Available online 19 March 2009

Keywords:CiguateraHPLCMultiple reactant monitoringAnalytical detectionSolid phase extractionMonitoring

* Corresponding author: Tel.: þ617 3346 2984; faE-mail address: r.lewis@imb.uq.edu.au (R.J. Lewi

0041-0101/$ – see front matter Crown Copyright �doi:10.1016/j.toxicon.2009.03.013

a b s t r a c t

Ciguatera is a significant food borne disease caused by potent polyether toxins known asciguatoxins, which accumulate in the flesh of ciguateric fish at risk levels above 0.1 ppb.The management of ciguatera has been hindered by the lack of analytical methods todetect and quantify clinically relevant levels of ciguatoxin in easily prepared crude extractsof fish. Here we report a ciguatoxin rapid extraction method (CREM) that allows the rapidpreparation of fish flesh extracts for the detection and quantification of ciguatoxin bygradient reversed-phase liquid chromatography-tandem mass spectrometry (LC/MS/MS).CREM-LC/MS/MS delivers a linear response to P-CTX-1 spiked into fish prior to extraction.A similar response was obtained for P-CTX-1 spiked after extraction, indicating >95%extraction efficiency was achieved overall and 85% at the limit of quantification (0.1 ppb).Using this approach, levels �0.1 ppb P-CTX-1 could be detected and quantified from anextract of 2 g fish flesh, making it suitable as a confirmatory assay for suspect ciguatericcarnivorous fish in the Pacific Ocean. The approach is designed to simplify the extractionand analysis of multiple samples per day.

Crown Copyright � 2009 Published by Elsevier Ltd. All rights reserved.

1. Introduction

Ciguatera (fish poisoning) is a major economic andsocial problem throughout tropical and sub-tropicalwaters, with w 25,000 persons poisoned annually (Lewis,2001). The disease is characterised by neurological andgastrointestinal disorders, which typically appear from 1 to24 h following the consumption of contaminated fish andcan last for months or longer (Gillespie et al., 1986). Thetoxins involved are potent sodium channel activator toxinsknown as ciguatoxins (Lewis et al., 2000) that are producedby the benthic dinoflagellate Gambierdiscus spp. (Lewis andHolmes, 1993; Holmes and Lewis, 2002). The ciguatoxinsand structurally related brevetoxins compete at site 5 on

x: þ617 3346 2090.s).

2009 Published by Elsevier

voltage sensitive sodium channel (Lombet et al., 1987). Tworelated families of Pacific ciguatoxins (P-CTX) have beenidentified in ciguateric Pacific Ocean fish (Murata et al.,1990; Lewis et al., 1991, 1993; Satake et al., 1993). A thirdfamily of Caribbean ciguatoxins (C-CTX) contaminateciguateric fish of the Caribbean Sea (Lewis et al., 1998), witha fourth family of Indian Ocean ciguatoxins contaminatingfish of the Indian Ocean (Hamilton et al., 2002a,b). Allciguatoxins identified to date are heat stable polyethertoxins of 1023–1157 Da. P-CTX-1 remains the most potentciguatoxin characterised (Lewis et al., 1991), oftencontributing w90% of the total lethality of carnivorousciguateric fish capture in the western Pacific Ocean (Lewisand Sellin, 1992), and posing a health risk at levels�0.1 ppb(Lewis, 2001).

The traditional method of detecting the presence ofciguatoxins in fish involves testing lipid extracts by themouse bioassay (Lewis and Sellin, 1993). More recently,

Ltd. All rights reserved.

R.J. Lewis et al. / Toxicon 54 (2009) 62–66 63

cytotoxicity (Manger et al., 1995), radioligand binding (Poliet al., 1997) and antibody-based sandwich (Oguri et al.,2003) assays have been shown to have potential as cost-effective screens for the detection of ciguateric fish.Analytical liquid chromatography-tandem mass spectrom-etry (LC/MS/MS) procedures have also been developed fordetermining ciguatoxins in fish extracts (Lewis et al., 1999).However, lack of a rapid extraction procedure to simplifyciguatoxin analysis limits the usefulness of this approach. Inthis report, we describe a ciguatoxin rapid extractionmethod (CREM) combined with gradient reversed-phaseliquid chromatography-tandem mass spectrometry (CREM-LC/MS/MS) approach for the detection and quantification ofclinically relevant levels of P-CTX-1 in fish flesh.

2. Materials and methods

2.1. Extraction of fish

Coral trout (Plectropomus maculatus), a species impli-cated in ciguatera outbreaks in Queensland, was obtainedfrom commercial outlets in Queensland, diced and keptfrozen at �20 �C until use. Two gram portions of the fishwere either spiked with P-CTX-1 (Lewis et al., 1991) or leftunspiked before being cooked at 70 �C for 20 min in capped50 ml Falcon tubes. Samples were then cooled beforehomogenisation (IKA Ultra Turrax T25, setting 5) with 8 mlmethanol/hexane (3:1) until no lumps of fish remained(2� 30 s). The homogenate was centrifuged at 4000 rpmfor 20 min at RT, the upper hexane phase carefully removedusing a pasture pipette and discarded, and the loweraqueous methanol phase (w5.5 ml) transferred intoa single use syringe and filtered through a 0.45-mm Milli-pore aqueous membrane filter (Millex@-HA) into a glassvial. The extraction procedure is designed for use on fishflesh with normal water content and variations in watercontent may influence extraction efficiency.

2.2. Solid phase extraction (SPE) of the crude extract

The crude extract was adjusted to 50–55% aqueousmethanol by the addition of 2 ml H2O, before cleanupthrough a C18 SPE cartridge. A number of reverse-phaseC18 cartridge types and elution conditions were trialled.The optimised procedure used a 900 mg C18 SPE cartridge(Alltech Prevail Maxi-Clean) pre-conditioned with 4 ml ofwater before loading the w7.5 ml sample and a 0.5 ml 65%methanol rinse of the vial. Initial studies showed thatspiked P-CTX-1 started eluting during the wash step whenmethanol was �70%, while no P-CTX-1 was detected in�65% aqueous methanol washes. To maximise cleanupwithout the loss of analyte, the cartridge was washed with6 ml 65% methanol (discarded) before P-CTX-1 was elutedwith 8 ml of 80% aqueous methanol.

To reduce matrix interference and the lengthy dryingstep associated with removal of water from the C18 cleanupstep, an additional orthogonal normal phase SPE cleanupstep was developed. To prepare the sample for this step, the80% methanol fraction was collected into a 50 ml Falcontube and made more polar by the addition of 4.2 ml of 1 MNaCl, before extraction with 6.7 ml chloroform with

vigorous shaking. The resulting two phases were separatedby centrifugation (w2000 rpm for 4 min on a bench-topcentrifuge), the upper aqueous methanol phase discarded,and the lower chloroform phase transferred to a scintilla-tion vial and dried under N2 and low heat to remove anyremaining methanol and water, which affect normal phaseSPE cleanup. After comparing a number of commerciallyavailable normal phase SPE cartridges, we selected a silicaSPE cartridge (Silica Plus, Waters), pre-conditioned with4 ml chloroform and loaded the sample in 4 ml chloroformand a 0.5 ml chloroform rinse of the vial. The cartridge wasthen washed with 4 ml chloroform before P-CTX-1 waseluted with 8 ml chloroform/methanol (9:1). For both SPEcleanup steps, syringe assisted positive displacement wasused to shorten the conditioning, sample loading, washingand final elution steps.

To validate the method, unspiked fish flesh sampleswere either spiked with P-CTX-1 just prior to the final N2

drying step, or left unspiked as negative controls. Allsamples were evaporated under N2 and low heat beforebeing redissolved in 200 ml of 50% aqueous methanol priorto LC/MS/MS analysis. To confirm its suitability for naturallycontaminated fish, a ciguateric giant queenfish (Scomber-oides commersonnianus) implicated in human ciguaterapoisoning in Queensland was also extracted and analysedby CREM-LC/MS/MS.

2.3. LC/MS/MS

The LC system comprised a C18 column (5 mm Phe-nomenex Luna, 2.1�250 mm) fitted with a pre-column(Phenomenex C18, 4� 2.1 mm, 5 mm). Solvent A wasaqueous 2 mM ammonium formate and 0.1% formic acid,and Solvent B was 95% acetonitrile with 2 mM ammoniumformate and 0.1% formic acid. The column was eluted at400 ml/min with a linear gradient from 35% B to 100% B over5 min. 100% B was held for 2 min before returning to 35% Bat 7.1 min. The column was then equilibrated for 5 min with35% B prior to the next run, allowing a 12 min turn-aroundbetween analyses.

Two triple-quadrupole mass spectrometers with Turbo-Ion-Spray ionization (AB Sciex Instruments) were used forthe detection of P-CTX-1 in this study. Both mass spec-trometers detected positive ions using multiple reactantmonitoring (MRM) with resolution for both Q1 and Q3 setat low. The API 2000 was employed for the initial devel-opment work optimising the reversed-phase C18 SPEcleanup step. However, the API 4000 (QTRAP), with itsgreater sensitivity due in part to its orthogonal spray sourceand design of the collision cell, was used for subsequentanalyses. MS/MS conditions were established using pureP-CTX-1 ([MþNH4]þ m/z 1128.7) dissolved in 50% B andinjected directly into the mass spectrometer at 10 mL/min.MS/MS signals were optimised for the dominant productions originating from the [MþNH4]þ ion. For the API 4000,a declustering potential of 110 V and a collision energy of32 eV were used (CUR 25, TEM 300, EP 10 and CXP 15). Asfound previously using an API-III triple-quadrupole MS(Lewis et al., 1999), three dominant fragment ions couldreadily be generated from the [MþNH4]þ ion of P-CTX-1(m/z 1128.7 / 1093.7; 1128.7 / 1075.7; 1128.7 / 1057.7)

R.J. Lewis et al. / Toxicon 54 (2009) 62–6664

allowing the sensitive detection of P-CTX-1 in fish extractsby LC/MS/MS. The peak height above baseline wasmeasured after summing the individual response data togenerate a TIC response for each analysis, with averageddata presented as the mean� standard deviation.

To minimise the extent of matrix effects on MS responseto P-CTX-1, various injection volumes from 5 to 100 ml of the200 ml sample after silica cleanup were analysed by LC/MS/MS. The optimal volume was found to be 20 mL, which wasused for the analysis of all the samples and pure P-CTX-1.

2.4. Chemicals used

Chloroform (99.8% pure, Ajex Finechem) containedethanol, methanol or alkyl phenols stabilizers at <0.2%. Allother solvents used were HPLC grade or equivalent.

3. Results and discussion

3.1. Rapid extraction of ciguatoxins

Previous approaches to the extraction of the lipidsoluble ciguatoxins are cumbersome. Most procedurestypically extract 50–100 g of fish with acetone, dry thefiltrate, remove low polarity lipids from an aqueous

Fig. 1. Flow diagram of the ciguatoxin rapid extraction method (CREM). CREM wastwo orthogonal solid phase extraction (SPE) cleanup steps prior to LC/MS/MS analyssteps are performed under a stream of N2 with mild heating, avoiding the need to uthe sample vial. The conditioning and wash steps (1 and 3) are discarded, and ciguatdifferent Pacific Ocean, Caribbean Sea and Indian Ocean ciguatoxins (Lewis, 2001analysis.

methanol phase with hexane (2�), remove solvent, extractether-solubles and P-CTX-1 from an aqueous ethanol phasewith ether (3�) and solvent removal. At this point theextract can be assayed using mice (Lewis, 2003) or run ona silica column before LC/MS/MS analysis (Lewis et al.,1999). To improve the efficiency of ciguatoxin analysis, wehave developed an optimised ciguatoxin rapid extractionmethod (CREM) that simplifies the detection and quanti-fication of P-CTX-1 by LC/MS/MS. CREM streamlines theextraction of ciguatoxin from fish flesh by (i) replacing theinitial acetone extraction and filtration step with a one-stepextraction and hexane cleanup, (ii) reducing the numbertransfers and drying steps, (iii) using centrifugation toseparate phases, (iv) keeping the liquid–liquid extractionsto single steps, and (v) reducing the quantity of fish fleshextracted (�2 g) to allow the incorporation of twoorthogonal SPE cleanup steps. A flow diagram of CREM isshown in Fig. 1. The initial cooking step, which is includedto denature fish flesh proteins and improve extractionefficiency, can be skipped when cooked fish are analysed.

3.2. LC/MS/MS

In parallel with the development of CREM, we devel-oped a gradient HPLC procedure that provided sufficient

designed to streamline the initial extraction of ciguatoxin and incorporatesis. The filtration step uses an aqueous Millipore filter and the two evaporationse a rotavapor. The SPE wash volumes indicated include 0.5 ml used to rinse

oxin elutes at step 4. Given the similar physical and chemical properties of the), CREM may have applicability to the extraction of multiple CTXs prior to

Fig. 3. Gradient reversed-phase LC/MS/MS response to P-CTX-1. (A) Coe-luting responses detected at m/z 1128.7 / 1093.7; 1128.7 / 1075.7600;1128.7 / 1057.7 obtained for 600 pg P-CTX-1 spiked into non-toxic fishflesh prior to extraction. Matching responses (1075/1093/1057) obtained forunspiked fish flesh samples are also shown for comparison. (B) TIC obtainedfor 20, 60, 200 and 600 pg P-CTX-1 spiked into non-toxic fish flesh prior toextraction.

R.J. Lewis et al. / Toxicon 54 (2009) 62–66 65

separation between the remaining lipids and P-CTX-1 witha relatively quick (12 min) turn-around time betweensample analyses. Finally, conditions for P-CTX-1 detectionusing an API 4000 QTRAP MS were optimised. Using thisapproach, pure P-CTX-1 was readily quantified by LC/MS/MS, giving a linear response from 6 to 600 pg (Fig. 2). Incontrast, responses to P-CTX-1 were reduced by w75% afterspiking into non-toxic fish extract, presumably due to thepresence of lipids coeluting with P-CTX-1 that reduce theefficiency of the ion evaporation/ionisation process andincrease baseline noise. Despite this reduction in sensi-tivity, a linear response was obtained after spiking 0.2–6 ngP-CTX-1 directly into 2 g of non-toxic fish flesh. An equiv-alent response was also obtained when P-CTX-1 was spikedinto an extract from 2 g fish just prior to solvent evapora-tion (Fig. 2). Comparing the slopes of responses obtainedfor spiking before and after extraction (see Fig. 2) gives anestimated recovery of P-CTX-1 from fish flesh of >90%,a significant improvement over more traditional extractionprocedures (Lewis and Sellin, 1993). The lower limit ofquantification of P-CTX-1 in 2 g fish was w20 pg usinga 20 ml injection. Comparing responses obtained for spiking0.2 ng (0.1 ppb) P-CTX-1 before (270� 62) and after(316�112) extraction estimates recovery of 85% for P-CTX-1 spiked at 0.1 ppb. Thus the approach allows the quanti-fication of P-CTX-1 in fish flesh at levels above 0.1, making itsufficiently sensitive to detect the clinically relevant levelsof P-CTX-1 (0.1–5 ppb) contaminating the flesh of cigua-teric carnivorous fish in the Pacific Ocean.

Examples of the LC/MS/MS response to 600 pg P-CTX-1spiked into non-toxic fish flesh for m/z 1128.7 / 1093.7,1128.7 / 1075.7, and 1128.7 / 1057.7 are shown in Fig. 3A.Also shown is a control (unspiked sample) for comparison.These data illustrate the lack of background signal and theclear individual MS responses obtained. These data weresummed to give the TIC traces shown in Fig. 3B which wereused to quantify P-CTX-1. Care needs to be taken whenanalysing responses to pure P-CTX-1, which we found werereduced by 95% if pure P-CTX-1 was prepared in plasticinstead of glass prior to analysis (data not shown).However, P-CTX-1 with fish extract gave similar responsesin glass and plastics, indicating that the fish matrixminimise ciguatoxin absorption to plastic. CREM-LC/MS/

Fig. 2. Graph of LC/MS/MS total ion current (TIC) responses obtained fordifferent concentrations of either pure P-CTX-1, P-CTX-1 spiked into 2 g rawfish flesh prior to cooking at 70 �C, or P-CTX-1 spiked into crude extracts of2 g fish flesh. Linearity (r2) was >0.98 across the three sets of data(mean� SD, n¼ 6). Slopes� SE (an index of sensitivity) are indicated in theinset.

MS was also used to quantify P-CTX-1 in a suspect cigua-teric fish flesh sample. Using this approach, 0.6� 0.053 ppbP-CTX-1 (range 0.48–0.74 ppb, n¼ 4) was detected,consistent with the sample being implicated in ciguaterapoisoning. Unfortunately, lack of a validated analyticalmethod for ciguatoxin precluded a direct assessment ofinterfering substances that may be present in wild caughtciguateric fishes.

We also investigated removing the silica SPE cleanupstep from CREM. This truncated procedure resulted innoticeable quantities of lipid remaining after removal ofsolvent and solubilisation of extract in 200 ml of 50%aqueous methanol was more difficult. Despite the poorersolubility and additional matrix, both the C18 alone andC18þ silica procedures gave similar MS responses forP-CTX-1 spiked into fish flesh before cooking (data notshown). However, increased background noise reduced thelimit of detection to w0.3 ppb for the truncated method.Thus, while it is possible to analyse extracts after only C18cleanup, optimisation of the LC/MS/MS method is requiredto allow detection of all clinically relevant samplescontaminated with P-CTX-1 using a truncated method.

4. Conclusions

CREM-LC/MS/MS simplifies the detection and quantifi-cation of clinical relevant levels of ciguatoxin in crudeextracts of fish. The method provides significant improve-ments in sensitivity and is designed to allow multipleanalyses per day. Given their similar physical and chemicalproperties, CREM/LC/MS/MS should be investigated to

R.J. Lewis et al. / Toxicon 54 (2009) 62–6666

determine its applicability to detect and quantify clinicallyrelevant levels of C-CTX-1 (Lewis et al., 1999) and I-CTX-1(Hamilton et al., 2002b) in suspect fish from the CaribbeanSea and Indian Ocean, respectively. As MS instrumentationimproves, it is anticipated that the limits of detectionreported here will be further lowered to allow furtherminiaturisation and the detection of the full range ofciguatoxins contributing to ciguatera. Once appropriatestandards are available, CREM-LC/MS/MS could haveapplicability in environmental monitoring studies aimed atdeveloping a better understanding of the distribution andfactors influencing ciguatera risk.

Acknowledgements

The work was supported by an ARC Special ResearchCentre for Functional and Applied Genomics. We thank Dr.Geoff Eaglesham for advice on optimising the MS condi-tions for ciguatoxins.

Conflict of interest

There is no conflict of interest in the reported work.

References

Gillespie, N.C., Lewis, R.J., Pearn, J.H., Bourke, A.T., Holmes, M.J., Bourke, J.B.,Shields, W.J., 1986. Ciguatera in Australia: occurrence, clinical features,pathophysiology and management. Med. J. Aust. 145, 585–590.

Hamilton, B., Hurbungs, M., Jones, A., Lewis, R.J., 2002b. Multiple cigua-toxins present in Indian Ocean reef fish. Toxicon 40, 1347–1353.

Hamilton, B., Hurbungs, Mira, Vernoux, Jean-Paul, Jones, Alun,Lewis, Richard J., 2002a. Isolation and characterisation of IndianOcean ciguatoxin. Toxicon 40, 685–693.

Holmes, M.J., Lewis, R.J., 2002. Dinoflagellate toxins. In: Menez, A. (Ed.),Perspectives in Molecular Toxinology. John Wiley and Son, Ltd, WestSussex, pp. 39–65.

Lewis, R.J., 2001. The changing face of ciguatera. Toxicon 39, 97–106.

Lewis, R.J., 2003. Detection of ciguatoxins. In: Hallegraeff (Ed.), UNESCOMonographs on Oceanographic Methodology, Manual on HarmfulMarine Microalgae. UNESCO, pp. 253–263.

Lewis, R.J., Holmes, M.J., 1993. Origin and transfer of toxins involved inciguatera. Comp. Biochem. Physiol. 106C, 615–628.

Lewis, R.J., Sellin, M., 1992. Multiple ciguatoxins in the flesh of fish. Tox-icon 30, 915–919.

Lewis, R.J., Sellin, M., 1993. Recovery of ciguatoxin from fish flesh. Toxicon31, 1333–1336.

Lewis, R.J., Sellin, M., Poli, M.A., Norton, R.S., MacLeod, J.K.,Sheil, M.M., 1991. Purification and characterization of ciguatoxinsfrom moray eel (Lycodontis javanicus, Muraenidae). Toxicon 29,1115–1127.

Lewis, R.J., Norton, R.S., Brereton, I.M., Eccles, C.D., 1993. Ciguatoxin-2 isa diastereomer of ciguatoxin-3. Toxicon 31, 637–643.

Lewis, R.J., Vernoux, J.-P., Brereton, I.M., 1998. Structure of Caribbeanciguatoxin isolated from Caranx latus. J. Am. Chem. Soc. 120,5914–5920.

Lewis, R.J., Jones, A., Vernoux, J.P., 1999. HPLC/tandem electrospray massspectrometry for the determination of sub-ppb levels of Pacific andCaribbean ciguatoxins in crude extracts of fish. Anal Chem 71,247–250.

Lewis, R.J., Molgo, J., Adams, D.J., 2000. Pharmacology of toxins involvedin ciguatera and related fish poisonings. In: Botana, L. (Ed.), MarinePharmacology. Marcel Dekker, Inc., New York, pp. 419–447.

Lombet, A., Bidard, J.N., Lazdunski, M., 1987. Ciguatoxin and brevetoxinsshare a common receptor site on the neuronal voltage-dependentNaþ channel. FEBS Lett. 219, 355–359.

Manger, R.L., Leja, L.S., Lee, S.Y., Hungerford, J.M., Hokama, Y., Dickey, R.W.,Granade, H.R., Lewis, R., Yasumoto, T., Wekell, M.M.J., 1995. Assoc. Off.Anal. Chem. 78, 521–527.

Murata, M., Legrand, A.M., Ishibashi, Y., Yasumoto, T., 1990. Structures andconfigurations of ciguatoxin from the moray eel Gymnothorax jav-anicus and its likely precursor from the dinoflagellate Gambierdiscustoxicus. J. Am. Chem. Soc. 112, 4380–4386.

Oguri, H., Hirama, M., Tsumuraya, T., Fujii, I., Maruyama, M., Uehara, H.,Nagumo, Y., 2003. Synthesis-based approach toward direct sandwichimmunoassay for ciguatoxin CTX3C. J. Am. Chem. Soc. 125,7608–7612.

Poli, M.A., Lewis, R.J., Dickey, R.W., Musser, S.M., Buckner, C.A.,Carpenter, L.G., 1997. Identification of Caribbean ciguatoxins as thecause of an outbreak of fish poisoning among U.S. soldiers in Haiti.Toxicon 35, 733–741.

Satake, M., Murata, M., Yasumoto, T., 1993. The structure of CTX3C,a ciguatoxin congener isolated from cultured Gambierdiscus toxicus.Tetrahedron Lett. 34, 1975–1978.