in vivo assays
The most commonly used assay method is the mouse bioassay developed by the Japanese Ministry of Health and Welfare (Yasumoto et al., 1978; Japanese Ministry of Health and Welfare, 1981). Toxins are extracted from shellfish tissue using acetone and after evaporation the residue is dissolved in a small volume of 1% Tween 60. The extract is injected intraperitoneally into mice with a body weight of about 20 g and the survival is monitored from 24 to 48 hours. One mouse unit (MU) is defined as the minimum quantity of toxin needed to kill a mouse within 24 hours. The toxicity of the sample (MU/g whole tissue) is determined from the smallest dose at which two mice or more in a group of three die within 24 hours. In many countries, the regulatory level is set at 0.05 MU/g whole tissue. In this mouse assay, all DSP components are likely to be detected including those DSP toxins which do not cause diarrhoea (PTXs and YTXs) and have an unknown toxicity for humans. Other unknown toxin groups exhibiting ichthyotoxic and hemolytic properties may also cause mortality of mice in this bioassay.
Therefore major disadvantages of this assay are the lack of specificity (no differentiation between the various components of DSP toxins), subjectivity of death time of the animals, and the maintaining and killing of laboratory animals. In addition, this assay is time consuming and expensive, may give false positives because of interferences by other lipids (notably free fatty acids have shown to be very toxic to mice (Suzuki et al., 1996) and shows variable results between whole body and hepatopancreas extracts (Botana et al., 1996 and Van Egmond et al., 1993).
The problems observed with the original mouse bioassay of Yasumoto et al. (1978) have led to several modifications (Yasumoto et al., 1984a; Lee et al., 1987; Marcaillou-Le Baut et al., 1990). This, in turn, has led to a situation where different countries use different variants of the mouse assay, which calls for harmonization. In an attempt to standardize the methodology of the mouse bioassay, the EU has included directions on how to perform this assay, in its new directive on toxins of the DSP complex (EU, 2002a). The EUs Community Reference Laboratory on Marine Biotoxins has conducted an intercalibration exercise that showed reasonable agreement between EU National Reference Laboratories, when they applied the mouse bioassay on unknown shellfish extracts (CRL, 2001).
Fernández et al. (1996) warned that some bioassay procedures involve hexane washing steps to avoid false positive results when free fatty acids are present. The hexane washing step should be reconsidered taking into account the possible losses of low-polar DSP toxins, which may be solubilized in the hexane layer. This step must be avoided when analysing samples of unknown origin and with unknown DSP profiles.
suckling mouse assay
In this procedure, an extract of shellfish tissue is administered intragastrically to four to five day old mice. The degree of fluid accumulation in the gastrointestinal tract is determined after a four hour period by measuring the ratio of intestine mass to that of the remaining body. Ratio values above 0.8 to 0.9 indicate a positive reaction. The assay time is shorter than with the mouse bioassay but quantification of the results is much more difficult. Only diarrhoea causing substances (OA, DTXs) produce positive reactions. Detection limits for OA and DTX1 are 0.05 and 1 MU, respectively (Hallegraeff et al., 1995 and Van Egmond et al., 1993)
This assay is based on diarrhoea induction in rats. The (starved) animals are fed with suspect shellfish tissue (mixed into the diet) and observed during 16 hours for signs of diarrhoea, consistency of the faeces and food refusal. The method is at best semi-quantitative and does not detect PTXs and YTXs (Hallegraeff et al., 1995 and Van Egmond et al., 1993). The test is still used routinely in the Netherlands and it is an officially allowed procedure in EU legislation.
Daphnia magna assay
An assay in Daphnia magna was developed and used to analyse OA in mussel extracts. This method was reported to be inexpensive and sensitive. The method can be used in replacement of the mouse bioassay for the screening of okadaic acid and some co-extracting toxins in mussels. The extraction method used allows okadaic acid and DTX1 to be determined. The Daphnia bioassay can measure OA levels 10 times below the threshold of the mouse bioassay method (Vernoux et al., 1994).
intestinal loop assays
Fluid accumulation in the intestine of intact rabbits and mice has been used to detect DSP. Suspensions of DSP toxins in 1% Tween 60 saline are injected into intestinal loops. A positive result is obtained when the ratio of the volume of accumulated fluid (ml) to the length of the loop (cm) was greater than 1.0 (Hungerford and Wekell, 1992)
The diarrhoeic activity of algal toxins in blue mussels is determined quantitatively in ligated intestinal loops of the rat by Edebo et al. (1988a). Hepatopancreas from toxic mussels is disintegrated by freeze-pressing, and the homogenized tissue suspended in an equal amount (w/v) of buffer or in the liquid recovered after steaming. Rapid fluid secretion is seen after injection of the suspension into ligated loops of rat small intestine; maximum is reached within two hours (about 300 mg of weight increase per cm of intestine). Within a range of 50-200 mg/cm dose-response relationship is close to linear. Average deviation from the mean is ± 9 mg/cm (SD= ± 4.9). Mussels yielding less than 100 mg/cm of weight increase per g hepatopancreas were allowed for human consumption, a quantity agreeing with the allowed level of okadaic acid. The minimum quantity of OA which produces significant secretion in the rat intestinal ligated loop test is approximately 0.5 mg.
Although mammalian bioassays for DSP toxicity are applied worldwide, there are large differences in the performance of, for instance, the mouse bioassay (toxicity criterion: animal death; no consensus on appropriate observation time) among different countries, resulting in differences in specificity and detectability. A major problem is the fact that the mouse bioassay detects all DSP components and probably also other toxins. On the other hand, the rat bioassay detects only OA and DTXs because the criteria in this assay are soft stool, diarrhoea and feed refusal which effects are known to be caused by OA and DTXs only. In addition, there is an increasing pressure to replace mammalian bioassays, not only because they are considered less suitable for quantitative purposes, but also because of ethical reasons. In the EU, a recommendation with supportive and convincing documentation has recently been issued by representatives of government institutions in Germany, the United Kingdom and the Netherlands to the members of the Scientific Advisory Committee (ESAC) of the European Centre for the Validation of Alternative Methods (ECVAM) to stimulate the development of methodology that can replace the existing bioassays, not only for DSP, but also for PSP (Grune et al., 2003).
The European Commission has recognised the needs of the analytical community to develop methods alternative to animal testing. A relevant call for proposals in the Commissions Sixth Framework Programme in Area 5: Food Quality and Safety appeared (EC, 2003) in which one of the objectives is to develop cost-effective tools for analysis and detection of hazards associated with seafood from coastal waters such as Diarrhoeic Shellfish Poisons, Yessotoxins, Pectenotoxins and Azaspiracid Shellfish Poisons. If granted, this will mean that progress can be expected in the coming years.
in vitro assays
An assay based on morphological changes in fresh rat hepatocytes when exposed to DSP toxins has been developed by Aune et al. (1991). Using this method, it is possible to differentiate between the diarrhoeic DSP toxins OA and DTX1, and the non diarrhoeic toxins PTX1 and YTX. OA and DTX1 induce irregular-shaped cells with surface blebs, PTX1 induces dose-dependent vacuolization and YTX does not cause changes in the shape of the cells but induces blebs on the surface. For OA and DTX1 the first signs appear at 0.5 mg/ml, for PTX at 5 mg/ml and for YTX at 10 mg/ml. This method is a valuable research tool in the separation between diarrhoeic and non-diarrhoeic DSP toxins. However, there are some disadvantages too as it is time consuming and confusing results may be obtained in the presence of mixtures of different algal toxins.
OA has high toxicity for KB cells (a human cell line derived from epidermoid carcinoma) apparent already after three hours of contact. Amzil et al. (1992) developed a method to determine the minimal active concentration (MAC) based on direct microscopic study of toxin-induced changes in cell morphology. A high correlation is found between the MAC of tested extracts and corresponding OA concentrations in mussel hepatopancreas as measured by LC (see Chapter 3.2.3).
Daiguji et al. (1998) reported that PTX2 showed cytotoxicity against KB cells at 0.05 µg/ml whereas pectenotoxin-2 seco acid and 7-epi-pectenotoxin-2 seco acid did not show cytotoxicity at a dose of 1.8 µg/ml. This implied that the cyclic structure of the PTXs is important to express cytotoxicity.
Tubaro et al. (1996b) developed a quantitative assay for OA using also KB cells. The method has shown itself to be effective in detecting OA in mussel samples at a detection limit of 50 ng/g digestive gland tissue in a 24 hours endpoint assay. The dose-dependent cytotoxicity assay is based upon the metabolic conversion of a tetrazolium dye (MTT) to yield a blue-coloured formazan product which can be read for absorbance with a microplate scanning spectrophotometer.
Pouchus et al. (1997) compared the activity of contaminated mussel extracts on KB cells by direct interpretation of morphological changes and by a colorimetric method estimating the number of viable cells after staining. The latter technique reveals interferences, not detected by the former, with mussel cytotoxins. The results show that the technique, based on determination of the minimal active concentration of DSP toxic extracts inducing morphological changes, is specific for OA and preferable to the determination of a 50 percent inhibition concentration (IC50) by a cell culture method.
OA and related compounds in mussels possess a high toxicity to Buffalo green monkey (BGM) kidney cell cultures. A detection method for OA and related compounds based on the morphological changes in BGM cell cultures has been developed. A high correlation is found between the official mouse bioassay (Yasumotos bioassay, observation time five hours) and this cytotoxicity test conducted on naturally contaminated samples of Mytilus galloprovincialis (Croci et al., 1997, 2001).
Other cytotoxicity assays for DSP toxins make use of fibroblasts (Diogene et al., 1995) as well as human cell lines (Oteri et al., 1998; Fairy et al., 2001; Flanagan et al., 2001). Further endpoints used to assess the cytotoxicity of DSP toxins include neutral red uptake (Draisci et al., 1998), vital staining (Flanagan et al., 2001) and inhibition of cell aggregation and apoptosis (Fladmark et al., 1998).
Cytotoxicity (hepatocytes, KB cells) assays seem to work well for OA and DTX1. However, their value in practice is to be awaited from ongoing inter-laboratory validation studies being carried out in the EU. Marcaillou-Le Baut et al. (1994) reported that results of the cytotoxicity assay with KB cells correlated well with results in the LC or the mouse test (by linear regression analysis).
3.2.2 Biochemical assays
There are several immunodiagnostic methods available for the detection of DSP toxins, configured as either RIA or ELISA tests, all of which incorporate antibodies prepared against a single diarrhoeic agent OA (Hallegraeff et al., 1995). A radioimmunoassay (RIA) for OA has been developed by Levine et al. (1988) (Hallegraeff et al., 1995). Antibodies to OA are prepared by immunizing rabbits with okadaic acid conjugated at the carboxy function to form an amide bond with an amino group of the immunogenic carrier, bovine albumin (using carbodiimide). Competitive binding of OA with 3H-OA in the test system and measurement by scintillation counting allows detection of 0.2 pmoles of toxin (about 0.2 pg/ml). Structurally related marine toxins (a.o. maitotoxin, palytoxin and brevetoxin) do not inhibit binding of tritiated OA to the antibody.
Enzyme-linked immunosorbent assay (ELISA) test kits have been developed and are commercially available. The DSP-Check® ELISA test kit from UBE Industries, Tokyo, Japan has been used throughout the world for screening OA and DTX1 at a claimed detection limit of 20 ng/g. Reports about its performance in practice vary. Inconsistencies including false positive responses when applied to either phytoplankton or shellfish samples have been reported many times. However, in a comparative experiment with LC (method of Lee et al., 1987), the DSP-Check® test kit was capable of detecting quantitatively DSP toxins in all tested contaminated samples containing only okadaic acid, provided that the parent toxins were within the range of detection and were not in the ester form (Vale and De M. Sampayo, 1999). The test was found to be more sensitive, specific and faster than LC. The monoclonal antibody in the DSP-Check® test kit cross-reacts with DTX1 at a level comparable to OA but PTXs and YTXs are not reactive (Hallegraeff et al., 1995).
The Rougier Bio-Tech® ELISA test kit utilizes an anti-OA monoclonal antibody and an anti-idiotypic antibody which competes with OA for binding sites on the anti-OA antibody. The antibody in this test kit exhibits a much higher sensitivity (10-20 fold) for OA than either DTX1 or DTX2, and methyl-, diol- and alcohol derivatives of OA will also bind to the antibody, whereas DTX3 and brevetoxin-1 do not cross-react at all. This test kit has undergone extensive comparison with alternative analytical methods for DSP toxins such as HPLC and LC-MS and is found to be rather reliable for OA quantification in both mussel extracts and phytoplankton (Hallegraeff et al., 1995)
Morton and Tindall (1996) compared the DSP Check® test and the Rougier Bio-Tech® test with LC (modification of method of Lee et al., 1987) and found both ELISA kits to provide accurate estimations of okadaic acid in extracts which were free of methylokadaic acid. However, the DSP Check® test underestimated quantities of total okadaic acid in extracts containing both analogues. Since outbreaks of DSP have been associated with okadaic acid, methyl okadaic acid or a mixture of these and other related compounds, the ELISA kits may not accurately assess the total toxicity of shellfish samples.
The ELISA developed by Biosense® for yessotoxin is new (see Direct cELISA YTX assay at www.biosense.com). In-house data show that this ELISA probably detects a multitude of yessotoxin analogues and an international inter-laboratory study to test its performance was planned for 2003 (Kleivdal, H. Personal information, 2002)
Garthwaite et al. (2001) developed an integrated ELISA screening system for ASP, NSP, PSP and DSP toxins (including yessotoxin). The system detects suspected shellfish samples. Thereafter the suspected samples have to be analysed by methods approved by international regulatory authorities. Alcohol extraction gave good recovery of all toxin groups.
Immuno technology has also been applied in the development of biosensors for DSP toxins. Botrè and Mazzei (2000) defined a biosensor as a self-consistent bioanalytical device incorporating a biologically active material, either connected to, or integrated within, an appropriate physico-chemical transducer, for the purpose of detecting-reversibly and selectively-the concentration or activity of chemical species in any type of sample. Marquette et al. (1999) described a semi-automated membrane-based chemiluminescent immunosensor for okadaic acid in mussels. The sensor is integrated in a flow injection analysis system. Anti-OA monoclonal antibodies were labelled with horseradish peroxidase for their use in a competitive assay, in which the free antigen of the sample competes with OA, immobilized on commercially available polyethersulfone membranes. The authors investigated the operational stability of the sensor over 38 OA determination cycles and found a stable response for the first 34 measurements. In addition, the performance of five immunosensors (five different membranes) showed good repeatability for critically contaminated and blank mussel homogenates, with CVs of 12.6 and 7 percent respectively. It may be expected that the development and application of biosensors for the determination of toxins of the DSP complex will advance rapidly in the coming years.
Another antibody-based technique in DSP analysis is the application of immunoaffinity columns (IAC) to purify shellfish extracts prior to the determinative step in analysis procedures, usually LC. Puech et al. (1999) described the recent development and the characterization of IAC, which were elaborated using anti-okadaic acid monoclonal antibodies, for a specific retention of the OA group of toxins. The coupling yield and the stability of these columns were investigated as well as their capacity to remove interfering compounds. Cross-reactivity was observed between the antibodies and the DTX1 and the DTX2, allowing the detection of the different toxins in a single analysis. Different spiked or naturally-contaminated matrices (mussel digestive gland and algae) were tested, and recoveries varied from 55 to 95 percent according to the matrices. The IAC purification was then included as a step of a global IAC/LC/spectrofluorimetric detection method and the performance of the method was evaluated. Estimations of the linearity and the accuracy (percentages of the presumptive response for OA were in the range + 101 percent to 114 percent) were satisfactory in accordance with the method validation criteria. IACs have great potential as clean-up techniques in analytical methods, but their value in practice still has to be proven in inter-laboratory validation studies.
acid phosphatase assays
An assay for DSP based on acid phosphatase activity in the protozoan Tetrahymena pyriformis has been developed. Toxins are extracted from shellfish using acetone/ ether and cleaned up by silicic acid chromatography. Tetrahymena is cultured in the presence of the extract for 24 hours and the 50 percent acid phosphatase activity inhibitory concentration and the growth inhibitory concentration are determined and expressed as mouse unit equivalents (Van Egmond et al., 1993 and Hallegraeff et al., 1995)
The specific inhibition of protein phosphatase Type 1 (PP1) and Type 2A (PP2A) by certain DSP analogues (OA and DTX1) was used to develop a phosphatase radio assay using 32P-phosphorylase. The assay is used directly on shellfish extracts and on fractions collected after HPLC separation of the toxins from digestive gland extracts. Although the original technique, which is coupled with toxin fractionation by LC is not widely used as regulatory tool, it has been used frequently in screening the phosphatase inhibition activity of putatively phycotoxic compounds and partially purified extracts of phytoplankton and shellfish. In its current format, this assay is based on the inhibition of PP1 by OA with a limit of detection as low as 10 fg OA/100 g tissue. A relatively rapid radioactive protein phosphatase (PP)-based assay has been developed and used by Honkanen et al. (1996a, b) to detect OA in oyster (Crassostrea virginica) extracts. In more than 320 assessments with spiked oyster samples, all samples containing ³0.2 mg OA/g were positive. From the samples spiked with 0.1 mg OA/g, 16.7 percent were positive. Control samples and samples spiked with 0.02 mg OA/g were negative. A high correlation is seen between the results of this assay and LC.
Although the use of radiolabels in the PP assay leads to low limits of detection, colorimetric and fluorometric assays have been developed to allow a more widespread adoption of the PP assays (Quilliam, 1998a).
A colorimetric phosphatase-inhibition bioassay has been developed for the quantitative measurement of OA by Simon and Vernoux (1994). The assay uses an artificial substrate, p-nitrophenylphosphate, and a semi-purified protein phosphatase PP2A containing extract prepared from rabbit muscle. The lowest detectable concentration of OA is 4 ng/ml in aqueous solutions and 40 ng/ml (i.e. 100 ng of OA per g of mussel tissue) in crude methanol mussels extracts. The rapidity, accuracy, reproducibility (within the laboratory), specificity and simplicity of the procedure provide a simple way to assay OA in buffered or complex solutions.
Tubaro et al. (1996a) developed a colorimetric PP assay using p-nitrophenylphosphate and a commercially available PP2A preparation to assess the presence of OA in mussels. The assay, which is employed in the microplate format, is accurate and reproducible (within the laboratory). OA is detected in concentrations as low as 0.063 ng/ml in aqueous solutions and 2 ng/g in mussel digestive glands. Thirty naturally contaminated mussel samples were submitted to the PP2A inhibition assay as well as to an ELISA and a MTT cytotoxicity assay, with similar results. The assay is sensitive, rapid and does not require expensive equipment according to the authors.
Lower limits of detection are possible with fluorometric PP assays. Vieytes et al. (1997) developed a fluorescent enzyme inhibition assay for OA using 4-methylumbelliferyl phosphate and fluorescein diphosphate as substrates for enzyme PP2A. The detection limit of OA is 12.8 ng/g hepatopancreas in shellfish extracts. According to the authors this assay can also be used for very dilute samples, such as phytoplankton samples.
Fluorometric protein phosphatase inhibition assays have not only been shown to perform better than colorimetric assays, but also to agree well with the mouse bioassay and LC techniques (Quilliam, 1998a; Vieytes et al., 1997; Mountfort et al., 1999). However Mountfort et al. (2001) have modified the fluorometric assay to overcome the lack of sensitivity towards the ester derivatives of okadaic acid and analogues and to reduce significantly the incidence of false negatives observed previously. At the time of writing, a European collaborative study of the fluorometric protein phosphatase inhibition method was ongoing to establish the performance characteristics of this method for okadaic acid and DTX1. If the results are acceptable, the method will be standardized by CEN and it is likely that the method will subsequently be approved for regulatory purposes in the EU.
YTX inhibited the hydrolysis of p-nitrophenyl phosphate by PP2A. The IC50 was 0.36 mg/ml. The potency was lower than that of OA by four orders of magnitude. Hence, interference by YTX coexisting with OA in shellfish can be disregarded in the enzyme inhibition assay for OA or DTX1 (Ogino et al., 1997).
Detection methods based on immunology (ELISA, RIA) are not yet fully developed and certainly not formally validated for all toxins involved. Nunez and Scoging (1997) reported that the ELISA assay detecting OA and/or DTX1 did not accurately detect low concentrations compared with the LC assay, the colorimetric phosphatase inhibition assay and the mouse bioassay. Gucci et al. (1994) did not find either a clear quantitative agreement between four different test methods for DSP (mouse bioassay, rat bioassay, ELISA test and LC method). Also Draisci et al. (1994) reported that the ELISA method did not give always quantitatively reliable results compared to the mouse bioassay and the LC method. Morton and Tindall (1996) compared the LC-fluorescence method with two commercially available ELISA test kits for the detection of OA and DTX1 in dinoflagellate cells (Prorocentrum hoffmanium and P. lima). Although false positive and false negative samples were not detected by the ELISA test kits, both test kits may underestimate total toxins present. Acid phosphatase inhibition assays also seem to work well for OA and DTX1. However, their value in practice will be ascertained by ongoing inter-laboratory validation studies being carried out in the EU.
3.2.3 Chemical assays
thin layer chromatography (TLC)
DSP toxins can be detected by thin layer chromatography (TLC). After a clean-up (silica gel column chromatography or gel permeation) of the extracts, fractions are applied directly to a silica gel plate and eluted with a toluene-acetone-methanol mixture. The acidic DSP toxins themselves appear as a weak UV-quenching spot at Rf 0.4. Both the diol esters and the free acid toxins give a characteristic pinkish-red stain after spraying with a solution of vanillin in concentrated sulphuric acid-ethanol and standing at room temperature for several minutes. The free acids produce a bright pinkish-red colour whereas the colour is duller with the diol esters. When clean material is applied to a TLC plate, 1 mg of the toxin could be detected; with cruder fractions, 2 to 3 mg is required before detection was possible (Hallegraeff et al., 1995). These rather high detection limits are a limiting factor for the use of TLC determining DSP toxins.
gas chromatography (GC)
Gas chromatography (GC) methods have been developed to detect and separate OA toxins. The toxins from diethyl ether extracts of dinoflagellate cultures are first isolated and purified using silicic acid, gel permeation chromatography and reversed-phase partition chromatography. GC analysis of trimethylsilyl derivatives of intact toxin and methyl esters is carried out with hydrogen flame ionization detection (Hungerford and Wekell, 1992). In practice this technique is rarely used.
liquid chromatography (LC)
The method described below is one of the most commonly used analytical techniques for determination of OA and DTX1. The original method (Lee et al., 1987) involves sequential extraction of shellfish tissue with methanol, ether and chloroform; derivatization with 9-anthryldiazomethane (ADAM); silica Sep-pak clean-up; determination by HPLC with fluorescence detection. The ADAM method is very sensitive for DSP toxins being able to detect 10 pg of the OA derivative injected on the column. The minimum detectable concentration in shellfish tissue, however, is limited not by detector sensitivity but by chemical background, which can vary considerably between samples. The practical quantitation limit is about 100 ng/g tissue. If digestive glands only are used in the analysis this limit is equivalent to 10 to 20 ng/g for whole tissue of mussels.
Aase and Rogstad (1997) optimized the sample clean-up procedure for determination of OA and DTX1 with the ADAM derivatization method. The use of a solid-phase extraction silica column of 100 mg and washing solvents composed of dichloromethane instead of chloroform were proposed to minimize the effect of stabilizing alcohol.
The unstable nature of ADAM and its limited availability have led several researchers to look for alternative derivatization reagents including 1-pyrenyldiazomethane and 1-bromoacetyl-pyrene, N-(9-acridinyl)-bromoacetamide, 4-bromomethyl-7-methoxycoumarin, 2,3-(anthra-cenedicarboximido) ethyltrifluoro-methanesulphonate. The polyaromatic hydrocarbon reagents ADAM, 1-pyrenyldiazomethane (PDAM) and 1-bromoacetylpyrene (BAP) have proved to be the most successful as they are less prone to interferences from reagent and reaction artefact compounds (James et al., 1997). The ADAM-LC method has been collaboratively studied in an inter-laboratory validation study conducted by the German Federal Laboratory for fish and fish products (GFL, 2001) for the determination of OA and DTX1 in mussel. This method is now in the stage of standardization by CEN and expected to become a European Standard in 2004. Whereas ADAM-LC seems to work reasonably well for OA and DTX1, this is not the case for the other toxins of the DSP complex.
DTX3 cannot be analysed directly by this method but must first be converted back to OA, DTX1 or DTX2 via alkaline hydrolysis. The diol esters of the DSP toxins as well as YTXs and some of the PTXs cannot be analysed by the ADAM-LC method (Hallegraeff et al., 1995 and Van Egmond et al., 1993).
For the determination of the yessotoxins and pectenotoxin-2, alternative LC procedures have been developed making use of a dienophile reagent DMEQ-TAD (4-[2-(6,7-dimethoxy-4-methyl-3-oxo-3,4-dihydroquinoxalinyl)ethyl]-1,2,4-triazoline-3,5-dione) for fluorescent labelling (Yasumoto and Takizawa, 1997; Sasaki et al., 1999). While the authors claim that these methods are superior to the mouse bioassay in rapidity, sensitivity and specificity, no inter-laboratory validations have yet been performed to establish their performance characteristics.
Fernández et al. (1996) warned that any procedure (LC-FD, LC-MS) used to characterize all the DSPs present in shellfish should take into account that the hexane layer, usually discarded, can be very rich in low-polar DSP toxins.
micellar electrokinetic chromatography (MEKC)
MEKC with UV detection was applied to the determination of non-derivatized DSP toxins. OA was detected in mussels spiked with 10 ng/g whole tissue, and the presence of OA and DTX2 was observed in the crude extract of the dinoflagellate Prorocentrum lima (Bouaïcha et al., 1997a).
Hallegraeff et al. (1995) reported the analysis of diol esters of OA, DTX1 as well as DTX3 toxins. LC combined with electrospray ionization mass spectrometry (LC-ESI-MS) appears to be a sensitive and rapid method of analysis for DSP toxins. A detection limit can be achieved of 1 ng/g in whole edible shellfish tissue. Various analytical procedures continue to be developed for the determination of DSP toxins and recent reviews have described a comprehensive range of methods (Quilliam, 2001). DSP profiling of bivalves (scallops and mussels) with LC-MS has been reported by Suzuki et al. (2000). They focussed on OA, DTX1 and PTX6. Negative ESI-mode was found to be much more efficient than positive ESI-mode.
Matrix effects in the DSP analysis with LC-ESI-MS have been tackled in different ways. Suzuki et al. (2000) successfully used an alumina B column for sample clean-up. Hummert et al. (2000) applied size exclusion chromatography (SEC) for the clean-up of raw extracts from algae and mussel tissue containing either microcystins or DSP toxins. Although it is likely that improvements were obtained, the article fails in demonstrating that matrix effects could be removed completely (recovery data are missing, spiking was not applied). Goto et al. (2001) paid more attention to the chemical properties of the different DSP compounds by applying different extraction solvents and solvent partitioning.
Ito and Tsukada (2001) conducted an explicit study on matrix effects. They demonstrated a better performance by applying the standard addition method to each separate sample, which however requires two LC-MS runs per analysis. An alternative method, where the response factor was based on one model sample, was less satisfactory. The study nicely demonstrated and emphasizes the matrix effect from shellfish extracts and demonstrates how that effect can be tackled for quantification purposes.
In the second half of 2002, an inter-laboratory study took place of a new LC-MS method for determination of ASP and DSP toxins in shellfish (Holland and McNabb, 2003). The eight participating laboratories generally obtained consistent sets of data for the broad group of analyte toxins down to low levels (< 5ng/ml, equivalent to 0.05 mg/kg). In general, sensitivity was adequate to achieve the LODs required. Most of the participating laboratories could detect the analyte toxins; greater differences were observed for quantitation of some toxins, especially when no analytical standards were present. The participants used different MS detection modes: some used single MS detection (SIM/SIR), others used tandem MS detection (MRM), and some used both. Although the use of MRM mode is attractive in order to enhance specificity, it requires additional care for quantitation. To sum up, the study was very stimulating and encouraging for those who are interested in using an alternative method for the mouse bioassays which are not supported by any statistical validation data, are well known to have a relatively high rate of false positives, have inadequate detection capability for some toxins and are ethically unacceptable for routine food monitoring. Additionally, method 40.105 (the method tested) can reliably detect ASP toxins and a range of other toxins and metabolites such as azaspiracids and pectenotoxin seco acids which may not respond in mouse assays (Holland and McNabb, 2003).
Chemical methods (LC) are useful for identification and quantification of selected diarrhoeic toxins (usually OA or DTXs), and the first validated method for OA and DTX1 approaches the phase of standardization by CEN. For the other DSP toxins, some LC methods exist but have not yet been validated. The rapid developments in LC-MS methodology are promising; however, improvement and inter-laboratory studies will be necessary before these techniques can become generally accepted tools in regulatory analysis. A serious problem is that pure analytical standards and reference materials are hardly or not readily available, which hampers the further development and validation of analytical methodology for DSP toxins.