2.2.1 In general
Because of the potential hazard to humans and animals, a quick, sensitive and specific method is needed to determine the presence of the PSP toxins in shellfish. Traditionally, the presence of PSP toxins has been determined using the mouse bioassay. However, the controversial issue of using mammals for testing in addition to the inherent problems and limitations of mammalian bioassays encourages the development of alternative assays such as pharmacological assays, immunoassays, chemical or separation assays and alternative bioassays to detect marine toxins in seafood (Mons et al., 1998).
in vivo assays
Presently the mouse bioassay still forms the basis of most shellfish toxicity monitoring programmes. The procedure was developed more than half a century ago and has been refined and standardized by the Association of Official Analytical Chemists (AOAC) to produce a rapid and reasonable accurate measurement of total PSP toxins (Hollingworth and Wekell, 1990). Twenty gram mice are injected with 1 ml of an acid extract of the shellfish and the time taken for the animal to die is recorded. Highly toxic extracts are diluted to ensure that mortality occurs within 5 to 15 minutes. The toxicity of the sample is then calculated with reference to dose response curves established with STX standards and expressed in mouse units (MU). In most countries the action level for closure of the fishery is 400 MU/100 g shellfish (1 MU is the amount injected toxin which would kill a 20 g mouse in 15 minutes and is equivalent to 0.18 mg of STX). The limit of detection of the assay is approximately 40 µg STX/100g of shellfish tissue with a precision of ± 15-20 percent. A known interference is a high salt content of samples which suppresses toxic effects (Schantz et al., 1958), whereas zinc accumulation in oysters has been reported to lead to lethal effects in mice at levels that present no health threat to humans (Aune et al., 1998). Highly toxic extracts may give extremely variable results (Park et al., 1986). The practical drawbacks to using the method are:
a colony of mice between 19 g and 22 g in weight must be maintained, however, at times of increased monitoring the supply of mice may fail;
the detection limit of the assay is strain dependent;
the death time versus toxin level is non-linear;
it is very labour-intensive to determine accurately the death time;
the sacrifice of a large number of animals is involved.
In spite of these difficulties, the assay has been employed on a wide range of molluscs and crustaceans and is still the official method in most countries that regulate PSP toxins in seafood.
In France, a proficiency study was conducted in which eight laboratories applied the mouse assay for the analysis of oyster samples, contaminated with PSP toxins at levels non-detectable and at levels of 153 and 335 mg STX/100 g meat. The authors concluded that on the basis of overall performance all eight participating laboratories were proficient in their use of the AOAC mouse assay. Within-laboratory variations and between-laboratory variations ranged from 5 to 10 and from 8 to 40 percent respectively. Low recoveries were reported for the spiked samples, which pointed at underestimation because of salt effects. This inaccuracy would require an adequate safety margin to protect consumers (LeDoux and Hall, 2000).
The mouse assay was also used in a pilot study on PSP toxins in freeze-dried mussels, organized by the Food Analysis Performance Assessment Scheme (FAPAS®) in 2003 (Earnshaw, 2003). Fifteen laboratories took part in this exercise, nine of which applied the mouse bioassay. The materials used in the study were prepared from certified reference materials (Van Egmond et al., 1998), in such a way that stability and homogeneity were satisfactory. The results of this pilot exercise were not impressive. Analysis results ranged from 1 to 383 mg/100 g (expressed as total PSP toxins on fresh weight basis) with a median value of 137 mg/100 g. Statistical evaluation of the results was not undertaken due to the variable nature of the results received.
in vitro assays
in vitro hippocampal slice assay
Kerr et al. (1999) investigated in vitro rat hippocampal slice preparations as a means of rapidly and specifically detecting the marine algal toxins STX, brevetoxin and domoic acid (DA) in shellfish tissue or finfish and identified toxin-specific electro-physiological signatures for each. It was concluded that hippocampal slice preparations are useful in detection and analysis of marine biotoxins in contaminated shellfish tissue.
sodium channel blocking assay
The mechanism by which the PSP neurotoxins disrupt cell function has been suggested as an alternative method of assay. The toxins bind to sodium channels in nerve cell membranes disrupting normal depolarization. The amount of binding is proportional to toxicity. Davio and Fontelo (1984) described an assay in which the amount of radiolabeled STX displaced from a rat brain preparation is measured. An alternative approach has been developed in mouse neuroblastoma cells by Kogure et al. (1988) and Gallacher and Birkbeck (personal communication; Van Egmond et al., 1993). Mouse neuroblastoma cells swell and eventually lyse upon exposure to veratridine, which, when added together with ouabain, enhances sodium ion influx. In the presence of STX, which blocks the sodium channels, the action of the other two compounds is inhibited and the cells remain morphologically normal. In this bioassay the fraction of the cells protected from the actions of ouabain and veratridine is in direct proportion to the concentration of STX and its analogues.
Jellett et al. (1992) have modified this bioassay to improve its speed and convenience by eliminating the need to count individual cells to determine the STX equivalents. Instead, they have employed a microplate reader for automated determinations of absorption of crystal violet from stained neuroblastoma cells. When these changes and other minor technical modifications were tested in this tissue culture bioassay systematically, the lower detection limit was found to be around 10 ng STX equivalents per ml of extract (= 2.0 mg STX eq/100 g shellfish tissue). This version of the tissue culture bioassay was compared with the standard mouse bioassay using 10 acid extracts of dinoflagellates (Alexandrium excavata and Alexandrium fundyense) and 47 extracts of shellfish tissues, prepared according to the AOAC procedure. The tissue culture bioassay provided results virtually identical to those obtained with the mouse bioassay (r >0.96), and moreover, was considerably more sensitive. The results obtained from liquid chromatography (LC) analysis of a subset of 12 extracts were less consistent when compared with the results from both bioassay methods (Jellett et al., 1992). Truman and Lake (1996) also compared results of the neuroblastoma cell culture assay with results of the mouse bioassay. Twenty-nine extracts of shellfish gave negative results in both assays. Fifty-seven extracts gave positive results in at least one assay. In spiking studies with shellfish extracts the neuroblastoma assay showed a good response to added STX. The correlation between the assays for STX eq in shellfish was 0.876. The authors concluded that, although the results supported the use of the neuroblastoma assay as a screening procedure, results close to the regulatory limits should be confirmed by the mouse bioassay.
In principle the neuroblastoma cell assay could be a good alternative to the mouse bioassay for testing shellfish for PSP toxins. However, the procedure developed by Jellett et al. (1992) did not yield satisfactory results when it was tested in an AOAC International collaborative study in 1999. Disappointing performance in practice, also due to problems in the shipment of study materials, led to discontinuation of the studied method in the evaluation procedure of the Methods Committee on Natural Toxins of AOAC International (Personal information).
Another recent development to detect sodium channel-specific marine toxins like saxitoxin is the hemolysis assay developed by Shimojo and Iwaoka (2000). It is based on the principles of the mouse neuroblastoma tissue culture assay for sodium channel specific biotoxins using red blood cells from the red tilapia (Sarotherodon mossambicus). Veratridine and ouabain both react with red blood cells from tilapia by affecting the permeability of the cells membrane. Saxitoxin can inhibit this action (leaving the cell morphologically normal). By sequencing the addition of veratridine and ouabain, with either the extracted samples or saxitoxin to the red blood cells, PSP toxins can be detected. The authors reported that the test was able to detect saxitoxin in concentrations at 0.3 mg/ml, which is slightly above the limit of detection of the mouse bioassay. No information was provided about its value in screening shellfish in practice.
Both the mouse bioassay and the tissue culture bioassay measure total toxicity but not the content of the individual toxins.
Cheun et al. (1998) developed a tissue biosensor system consisting of a Na+ electrode covered with a frog bladder membrane integrated within a flow cell. The direction of Na+ transfer, investigated in the absence of Na+ channel blockers, established that active Na+ transport occurs across the frogs bladder membrane from the internal to the external side of the membrane. The tissue sensor response to each of a number of PSP toxins was recorded (GNTX1, 2, 3 and 4). Sensor output was inhibited in the rank order GNTX4 > GNTX3 > GNTX1 > GNTX2. Comparing these results with those obtained from the standard mouse bioassay showed good agreement except for GNTX2.
Comparison of results for neoSTX and dcSTX in the tissue biosensor system with the results in the standard mouse bioassay again showed good agreement (within 5 mg toxin/g wet tissue).
Lee et al. (2000) used the method above to examine the toxicity in cultured Alexandrium tamarensis strains under various environmental conditions. It appeared that the tissue biosensor system was able to measure very small quantities of PSP toxin within an individual plankton cell (5 fg). However, measurement of at least 100 cells is more desirable for increasing the sensitivity of the system. For comparison: at least 6 000 individual cells must be harvested to measure toxin production using the LC method.
2.2.3 Alternative bioassays
There is growing concern about the continued use of mammals for bioassay and one alternative may be to develop similar assays based on the use of invertebrates such as oyster embryos or fish larvae. One method employed to reduce the number of mouse tests in several European countries is to use enumeration of presumptive toxic algal cells in seawater for monitoring purposes (Hald et al., 1991). This technique could also be described as a qualitative assay but cannot be used for quality control of shellfish for commercial sale.
2.2.4 Biochemical assays
Indirect enzyme-linked immunosorbent assays (ELISA) that exploit antibody-antigen binding are increasingly used as dip-stick assays for a variety of compounds. One method for production of PSP assay systems has been described by Chu and Fan (1985). A STX antigen is prepared using bovine serum albumin and injected into rabbits. Antibodies raised by the rabbits are then collected and lyophilized. In the test system, antigens are coated to microtitre plates, STX standards or mussel extracts and appropriate dilutions of antibodies are added, and the amount of bound antibody is determined using goat antirabbit IgG peroxidase conjugate, with measurement by a colorimetric substrate assay. STX present in the mussel extract competes for binding with the STX antigen coated to the microtitre plates. Until recently, commercial ELISA test kits have only been developed for STX. However, these are not totally specific for STX and some reaction is induced to decarbamoyl-STX (dcSTX) and neoSTX. Cembella and Lamoreux (1991) described a polyclonal test kit which measures STX, neoSTX, GNTX1 and GNTX3. Although the kit has not yet been fully evaluated, it appears to be more sensitive than LC and more specific than the mouse bioassay.
Chu et al. (1996) compared three different direct competitive ELISAs for the analysis of a large number of contaminated shellfish and concluded that there was excellent agreement between the ELISA data and mouse assay results. Usleber et al. (1997) also concluded that ELISA results correlated well with mouse bioassay results when analysing scallops. Kasuga et al. (1996) concluded however that the mouse assay cannot be replaced with ELISA for the purpose of screening inshore shellfish samples, as unpredictable cross-reactions occurred, as well as underestimations of toxicity of some naturally contaminated shellfish samples, harvested in the sea near Japan.
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.
Kawatsu et al. (2002) developed a direct competitive enzyme immunoassay based on a gonyautoxin 2/3 (GNTX2/3)-specific monoclonal antibody and a saxitoxin-horseradish peroxidase conjugate. GNTX2/3, dc-GNTX2/3, C1/2, GNTX1/4, STX and neoSTX were detectable at concentrations lower than the regulatory limit of 80 µg/100 of shellfish tissue.
Several more publications have appeared recently about the application of ELISA to the analysis of shellfish for PSP toxins. In view of the presence of cross-reactions with lower binding specificity and the potential lack of response to other toxins than STX within the PSP group, the practical application of these assays probably will remain limited, unless acceptable performance characteristics can be demonstrated in formal collaborative studies according to AOAC International or ISO accepted procedures. Such studies have not yet been published.
2.2.5 Chemical assays
fluorometric and colorimetric techniques
The alkaline oxidation of PSP toxins yields fluorescent products, allowing simple determination using fluorometric techniques (Bates and Rapoport, 1975; Bates et al., 1978). However, such techniques are subject to several sources of variability. The adjustment of pH during extraction and before oxidation is critical, ion exchange column clean-up is necessary to remove interfering co-extractants, and the presence of a variety of metals can affect oxidation and subsequent fluorescent yield. Moreover, the toxins do not fluoresce equally, and for several of the carbamate toxins fluorescence is very weak. One way of circumventing the latter problem is to apply multiple fluorescence and colorimetric assays on the same samples. The fluorescence assay was reported to be an order of magnitude more sensitive, and the colorimetric assay slightly more sensitive, than the mouse assay (Mosley et al., 1985).
Hungerford et al. (1991) have automated a fluorescence method by using flow injection analysis. The method allows automatic correction for background fluorescence and rapid screening of shellfish samples for the presence of PSP toxins.
Techniques based on liquid chromatography (LC) are the most widely used non-bioassay methods for determination of PSP compounds. During the last decade considerable effort has been applied to the development of an automated LC method for routine analysis of PSP toxins. The assays are generally based on separation of the toxins by ion-interaction chromatography and use of a post-column reactor that oxidizes the column effluent to produce readily detectable derivatives. The methodology developed by the United States Food and Drug Administration was reported to be capable of resolving 12 carbamate and sulfocarbamoyl PSP toxins (Sullivan, 1988). The methodology has been validated against the mouse bioassay and the correlations between the techniques is generally good (r > 0.9) (Sullivan, 1988). Detection limits are generally an order of magnitude lower than with the mouse assay. In practice, the method of Sullivan (1988) has shown difficulties in separating STX from dcSTX (Van Egmond et al., 1994) and has therefore gone out of use in most European laboratories involved in PSP analysis.
Although the LC approach is an interesting development, the system requires a considerable amount of skill and dedicated time to make it operate routinely. Furthermore, the LC technique is not free of problems. Thielert et al. (1991) has shown that the 6-decarbamoyl toxins are not resolved by the method of Sullivan (1988). Improved resolution was achieved by sequential analysis of the samples using different buffer and ion-pair reagent systems. Ledoux et al. (1991) described problems with discrimination of the C-group toxins from fluorescent material present in non-toxic mussels. Waldock et al. (1991) also reported that the LC technique was not sufficiently rapid or robust to cope with the large number of samples generated during bloom events.
Peak spreading is also a problem due to the large volume of post-column reaction tubing. One method of circumventing this problem is to prepare fluorescent derivatives before LC separation (Lawrence and Menard, 1991; Lawrence et al., 1991a), but as yet not all of the known PSP toxins have been separated by this method because some of the known toxins (e.g. GNTX2 and GNTX3) lead to the same oxidation products.
Furthermore, for accurate quantitation, it is essential to calibrate the system continually using PSP toxins standards. This is because of differences in the chemistry of each PSP toxin that result in different oxidation rates for each compound in the post-column reactor. Until recently only a STX standard was commercially available and accurate estimation of the amounts of the other PSP toxins in the mixture was impossible. In 2003, certified standards of STX, neoSTX, GNTX 1-4, GNTX 2/3 and GNTX 5 were commercially available (Laycock et al., 1994; NRC, 2003), and their availability significantly improves the quality of the data that is obtained by the LC method (Wright, 1995). Researchers should be careful when changing from one standard to another as a discontinuity of data may occur. Concentration differences up to 20 percent have been noticed between STX concentrations of three different suppliers (Quilliam et al., 1999).
LC methods were used (in addition to the mouse bioassay, see Section 2.2.2.) in a pilot study on PSP toxins in freeze-dried mussels organized by the Food Analysis Performance Assessment Scheme (FAPAS®) in 2003 (Earnshaw, 2003). Fifteen laboratories took part in this exercise and seven of them applied LC. Practically all laboratories analysed the test materials for STX and dcSTX, some also determined the amounts of neoSTX; GNTX1/4; GNTX2/3; GNTX5, GNTX6, C1/2 and C3/4. The results obtained for STX ranged from non-detectable to 83 mg/100 g (on fresh weight basis), those for dcSTX ranged from 25 to 130 mg/100 g. The test material actually contained < 3.5 mg/100 g for STX and ~ 80 mg/100 g for dcSTX. An analysis of the analytical procedures used showed that those laboratories that found positive values for STX all used HCl, with boiling in the extraction step (as in the mouse assay according to the AOAC-procedure (Hollingworth and Wekell, 1990). In contrast, laboratories that applied acetic acid without boiling in the extraction step found hardly or no saxitoxin. The reason for this is that HCl extraction with boiling leads to partial hydrolysis of certain PSP toxins, leading to conversion of some PSP toxins into more toxic analogues (e.g. GNTX5 is converted into STX). Acetic acid without boiling is a milder extraction procedure, which leaves the toxin profile of the sample practically intact. The sample used in the FAPAS study did not contain STX but it did contain GNTX5. Awareness of this phenomenon and standardization of methodology may largely solve this problem, and may lead to better agreement in analytical results, as demonstrated in a Dutch proficiency study (Van Egmond et al., 2004)
A project was carried out from 1993 to 1997 within the framework of the European Commissions Standards, Measurements and Testing Programme (SMT) (previously called and also known as the Bureau Communautaire de Référence or BCR) to develop shellfish reference materials with certified mass fractions of some PSP toxins. The work was carried out by a consortium of 13 public laboratories and six universities, representing the five main shellfish producing countries in the European Union (EU) and some other EU member countries that had an interest in the area of PSP-determinations. A preliminary inter-laboratory study in the EU had already shown that there was a basis for the development of reference materials (Van Egmond et al., 1994).
The research programme involved:
studies on the improvement and evaluation of the chemical methodology;
identification and determination of purity of PSP standards, and their stability in solution;
two inter-comparison studies of analytical methods;
preparation of reference materials, including homogeneity and stability studies;
a certification exercise.
Initially the laboratories were asked to analyse solutions of STX and PSP-containing shellfish extracts with a method of their choice but in the final certification study design only LC-methods involving precolumn or postcolumn derivatization were included. The project was finalised with a report describing the certification of the mass fractions of STX and dcSTX in two mussel reference materials (BCR-CRMs 542 & 543) including the identification of several other PSP-toxins, and a spiking procedure based on an enrichment solution (CRM 663) with a certified mass concentration of STX (Van Egmond et al., 1998; Van den Top et al., 2000, 2001).
Two of the methods used in the SMT project that showed good performance characteristics in the SMT project (Lawrence and Menard, 1991; Franco and Fernandez, 1993) were selected for standardization by the European Committee for Standardization (CEN). At the time of writing, the Franco method had appeared as European Prestandard (CEN, 2002a), and the method of Lawrence as Draft European Standard (CEN, 2002b). The latter method was successfully applied in a proficiency study on PSP in shellfish, carried out in the Netherlands in 2001 at national level (Van Egmond et al., 2004). The method was also further modified by Lawrence and the modification was evaluated in 2002 in an international collaborative study (Lawrence et al., 2003).
Various methods for separation of PSP toxins have been developed using gel and paper electrophoresis (Boyer et al., 1979; Onoue et al., 1983; Ikawa et al., 1985; Thibault et al., 1991). Used in batch mode and in a single dimension, the technique could allow rapid screening of a number of samples. However, quantitation appears to be a major stumbling block, and most methods employ a peroxide spray and a UV lamp to visualise the toxins on the electrophoretic plate. Perhaps one way forward in this area would be the use of scanning fluorescence detectors (Van Egmond et al., 1993).
Capillary electrophoresis (CE) is a relatively new technique and to date there have been few applications in the field of toxin analysis, however the flexibility of CE systems suggests that it is a promising area for research. In essence, the technique employs a narrow (~100 mm id) fused silica capillary in place of the electrophoretic gel, and nanolitre amounts of the sample are introduced to the end of the column before it is used to bridge two buffer reservoirs. The toxins migrate through the column when high voltage is applied and may be detected as they pass through a UV or fluorescence cell. The technique is applicable to broad classes of compounds with electrophoretic mobility and even where no net charge occurs it is possible to trap compounds in micelles which will then migrate.
Wright et al. (1989) applied a CE system coupled to a laser fluorescence detector for the determination of STX standards. The technique allowed detection of STX at the 1 mg/kg level. Even though the injection volume is necessarily small (1 to 10 nl), the theoretical detection limits for samples are in the mg/kg range. The present drawbacks to the technique are that the same separation has not been demonstrated for biota with mixed toxins, the equipment is not commercially available and is expensive, and the methodology suffers from the same problems as LC in that a fluorescent derivative must be prepared before separation or detection.
Thibault et al. (1991) have applied a CE method to samples of marine biota. Separations of neoSTX and STX were achieved and using UV spectrometry a detection limit of 5 mM (approximately 1.5 mg/ml) was demonstrated. The authors suggest that the CE-UV technique holds considerable promise for the routine screening of these toxins in natural extracts, but presently detection limits appear to be too high to be of use in monitoring programmes.
The application of mass spectrometry in the field of marine biotoxins was promoted during the last two decades not only by the development of LC-MS interfacing but also by integration of separation, detection and computer technologies. The key to the growth and success of LC-MS (including LC-tandem MS) is (and will be) in the informing power, reliability, affordability and availability of commercial systems (Willoughby et al., 1998).
Already in the late 1980s, Quilliam et al. (1989) reported the determination of STX by LC-MS applying ion-sprayä as ionization technique, being a trademarked name to describe pneumatically assisted electrospray (Sparkman, 2000). In single ion recording (SIM mode) and focussing on positive ions, a concentration detection limit of 0.1 µM (1 µL injection) was estimated from flow injection analysis, which is about five times more sensitive than the AOAC mouse bioassay. Full scan spectra were recorded of 100 ng of STX, as well as product ion (daughter ion) spectra of the single protonated molecule ([M+H]+), providing information useful for confirmation of identity and for development of an SRM method.
Pleasance et al. (1992a) reported on analysis of PSP toxins applying LC-MS and CE-MS. LC-MS (SIM and full scan-MS1 mode) was used to monitor purification of saxitoxin isolated from dinoflagellate cell extracts. Additionally, tandem mass spectrometry (MS2) has been used to provide structural information. It appeared possible to detect 10 pg injected, that is equivalent to a concentration of 0.03 µM. The improvement was obtained by change of the mobile phase in combination with a reduced flow rate. A calibration curve was shown for standard solutions (external calibration) having a concentration range with ratio 55 (highest/lowest conc.) Although the picture looks fine, values are missing for the linearity and reproducibility indicators (r2 and s.d. respectively). The applicability of Flow Injection Analysis (FIA) to the determination of PSP toxins in more complex marine extracts was also clearly discussed. It was judged to have serious limitations.
Quilliam et al. (1993) reported on an LC-MS study with qualitative aspects. The study focused on the characterization of periodate oxidation products of PSP toxins. Mass spectra (mostly MS1-spectra) were acquired of the various oxidation products, however sensitivity (relative response) was greatly reduced over that for the parent toxins, and the authors concluded that The overall sensitivity is such that pre-column oxidation combined with LC/MS will not be a competitive method for the trace level analysis of PSP toxins.
Jaime et al (2001) mentioned a PSP quantification method using a linkage of ion exchange chromatography with electrospray ionization (ESI)-mass spectrometry. The chromatographic separation was achieved by gradient elution. Measurements were carried out in SIM mode. Descriptions for automated systems were depicted. The focus was on limits of detection (LOD) and linearity; the LODs obtained for the individual PSP toxins were comparable to those obtained by other methods based on ion-pair chromatography with chemical oxidation and fluorescence detection, and well suited for determination of PSP toxins in biological materials (regulatory limit mentioned for mussels and shellfish: 800 µg PSP/kg). Linearity was demonstrated by good correlation coefficients (> 0.99). These were obtained notwithstanding the limited calibration concentration range (approximately one decade on average).
Quilliam et al. (2001; 2002) presented various LC-MS methods for the determination of PSP toxins, especially the method where they used hydrophylic interaction liquid chromatography coupled with electrospray ionization tandem mass spectrometry detection (HILIC-ESI-MS/MS). The authors claim to have a method that detects all PSP toxins in a single analysis run. So far, the methods have been presented but not published.
Oikawa et al. (2002) have used LC-MS to confirm the accumulation of PSP toxins (GNTXs and C-toxins) in edible crab. A description of quantification with LC-MS was not reported. Partial purification was conducted for ESI-MS analyses, that is successive treatment with activated charcoal and a Bio-Gel P2 column.
To summarise, in the field of PSP toxin analysis, LC-MS articles mainly concern qualitative aspects and reflect conventional use of MS1 mode, although tandem instruments are used. The application of LC-tandem MS for PSP toxin analysis has recently been presented but has not yet been published.