- Email: [email protected]
The validity of two HPLC methods and a colorimetric PP2A assay related to the mouse bioassay in quantification of diarrhetic toxins in blue mussels (Mytilus edulis)
Toxicon 39 (2001) 1387±1391
www.elsevier.com/locate/toxicon
The validity of two HPLC methods and a colorimetric PP2A assay related to the mouse bioassay in quanti®cation of diarrhetic toxins in blue mussels (Mytilus edulis) H. Ramstad a,*, J.L. Shen a, S. Larsen b, T. Aune a a
Department of Pharmacology, Microbiology and Food Hygiene, Norwegian School of Veterinary Science, PO Box 8146, Department 0033, Oslo, Norway b Department of Large Animal Clinical Sciences, Norwegian School of Veterinary Science, Oslo, Norway Received 13 October 2000; accepted 27 January 2001
Abstract Validity of two HPLC methods and a PP2A assay in relation to the mouse bioassay for diarrhetic shell®sh poisoning (DSP) toxins was evaluated. The mouse bioassay for DSP toxins was performed on a total of 177 mussel samples from the Sognefjord, Norway, using diethyl ether in the ®nal step of extraction. For ¯uorimetric HPLC analyses, either 4-bromomethyl-7-methoxycoumarin (BrMMC) or 9-anthryl diazomethane (ADAM) were used for analysis of 48 and 118 of the samples, respectively. The colorimetric PP2A inhibition assay was performed on all 177 samples that were analysed with the mouse bioassay. When comparing the HPLC-BrMMC, the HPLC-ADAM and the PP2A assays with the mouse bioassay, cut off values of #4, 5 and 6 mg okadaic acid (OA) equivalents (eq.)/5 g digestive gland (DG) was used. With reference to the results from the mouse bioassay, the total number of failure and correct classi®cation by HPLC-ADAM and the PP2A method was compared for the three cut off values. No signi®cant differences between the methods were detected. However, all differences were found in favour of HPLC-ADAM. All three methods could replace the mouse bioassay in detecting levels of diarrhetic toxins approved internationally for safe consumption of mussels. However, HPLC-ADAM seems to be the method of choice. q 2001 Elsevier Science Ltd. All rights reserved. Keywords: DSP toxins; HPLC; Mouse bioassay; PP2A
1. Introduction A well-known and widespread human toxic syndrome is called diarrhetic shell®sh poisoning (DSP) on account of its predominant gastrointestinal symptoms (Yasumoto et al., 1980). The most widely used screening method for these toxins is the mouse bioassay (van Apeldoorn et al., 1998), but the choice of extraction procedure highly in¯uences the outcome of the DSP mouse bioassay (Aune, 1997). Using diethyl ether in the ®nal step of extraction (Yasumoto et al., 1984) normally facilitate extraction of the truly diarrhetic toxins, while chloroform extraction includes a wider spec* Corresponding author: c/o Dr Kevin James, Cork Institute of Technology, Cork, Ireland. Fax: 1353-21-345-191. E-mail address: [email protected] (H. Ramstad).
trum of toxins, including pectenotoxins (PTXs), yessotoxins (YTXs) and among others, unknown toxin(s) with neurotoxic effects (Stabell et al., 1991; Ramstad et al., 2001a). YTX is still considered to belong to the group of DSP toxins although it does not induce diarrhea (Terao et al., 1990). Okadaic acid (OA) and the dinophysis toxins (DTXs) are among the most important toxins for the mussel industry due to their worldwide occurrence. (Yasumoto and Murata, 1993; Kumagai et al., 1986). The minimum doses of OA and DTX-1 necessary to induce diarrhea in adults have been estimated to be 40 and 36 mg, respectively (Hamano et al., 1986). By means of titration of toxicity with OA, one mouse unit (MU) corresponds with 4 mg OA, and is de®ned as the minimum quantity of toxin capable of killing a 20-g mouse within 24 h after intraperitoneal injection (i.p.) (Yasumoto et al., 1984, 1989). Most countries with established regulations for DSP toxins apply an acceptance level of 4±5 mouse
0041-0101/01/$ - see front matter q 2001 Elsevier Science Ltd. All rights reserved. PII: S 0041-010 1(01)00097-6
1388
H. Ramstad et al. / Toxicon 39 (2001) 1387±1391
units or 16±20 mg OA equivalents (eq.)/100 g mussel meat (Yasumoto et al., 1984; Aune, 1997). There are ethical and economic concerns associated with the mouse bioassay, and furthermore it lacks precision regarding death time (Toti et al., 1991). In addition, the mouse bioassay is time consuming and may give false positives with respect to risk of diarrhea if the ®nal extraction is acetone or chloroform. Due to increased availability of commercial standards, a number of biochemical and chemical methods have been developed (Apeldoorn et al., 1998). The diarrhetic toxins, measured as OA equivalents, were found to give repeatable and reliable results when using both ¯uorimetric HPLC methods and a colorimetric PP2A assay (Ramstad et al., 2001b). In this study, the validity of two HPLC methods and a colorimetric PP2A assay were evaluated against the mouse bioassay with ether extracts for quanti®cation of the diarrhetic toxins in blue mussels (Mytilus edulis). 2. Materials and methods 2.1. Reagents Solvents used for the HPLC analyses were acetonitrile and methanol of LC quality supplied by Romil (Cambridge, UK), and the water was Millipore-Q cleaned. The derivatizing agent 9-anthryl diazomethane (ADAM) was purchased from Molecular Probes (Oregon, USA) and the derivatizing agent 4-bromomethyl-7-methoxycoumarin (BrMMC) was from Sigma (USA). OA was supplied by Marine Analytical Chemistry Standard Program (Halifax, Canada) and DTX-1 was obtained from Wako Chemicals GmbH (Germany). pNitrophenylphosphate was supplied by SIGMA Bio Sciences (USA), and protein phosphatase 2A (25 U) isolated from rabbit skeletal muscle was bought from Promega (WI, USA). Other solvents and reagents used were of analytical grade. 2.2. Sampling stations Nine locations in the Sognefjord, Norway were selected for sampling of blue mussels (Mytilus edulis), covering a distance of about 200 km from the coast to the innermost part of the fjord (Ramstad et al., 2001a). Mussels (1±2 kg) were sampled every other week during the monitoring periods, from July until the end of October 1996, and from March to November, 1997. Additionally, eight samples from the period October to December 1998 were included. The mussels were evenly sampled from the surface to 6 m depth, except on the two innermost stations, where mussels do not survive the low salinity in the surface water. From the latter stations the samples were taken from 3 to 8 m depth. HPLC analyses for OA/DTX-1 were performed on 48 and 118 samples using BrMMC and ADAM, respectively, while the total samples of 177 were analysed using a colorimetric PP2A inhibition assay and the mouse bioassay.
2.3. Animals White, female mice, BOM: NMRI weighing between 15 and 20 g were used for mouse bioassays. 2.4. Toxin extraction for the mouse bioassay The mouse bioassay for the DSP toxins was performed essentially according to the method of Yasumoto et al. (1984) with slight modi®cations. Brie¯y, 20 g of homogenised digestive glands (DG) from each mussel sample was extracted with 200 ml acetone, ®ltered and taken to dryness on a rotary evaporator. The residue was partitioned between 16 ml petroleum ether and 8 ml 80% MeOH, and the petroleum ether phase was discarded. Defatting with 16 ml petroleum ether was repeated. The 80% MeOH-fraction was divided in two portions, and diethyl ether or chloroform was used in the ®nal step of extraction. In this study, only the ether fraction was studied. The 80% MeOH-fraction used for ether extraction was evaporated almost to dryness under nitrogen, and 3 ml water was then added. The aqueous suspension was extracted twice with 4 ml diethyl ether, and ®nally the combined ether fraction was backwashed with 1 ml water and then evaporated to dryness under nitrogen. All samples were stored at 2208C until use. 2.5. Mouse bioassay The residue from the mussel extract was suspended in 2 ml 1% Tween 60 solution and two mice received each i.p. 5 g DG per 20 g body weight (b.w.). The mice were observed for 24 h, and the de®nition of `dead' means that minimum one out of two mice analysed is dead. 2.6. Toxin extraction for biochemical and chemical analysis Toxin extraction was performed according to the method of Lee et al. (1987) when preparing extracts for HPLCADAM and the PP2A assay. The method of Shen et al. (1997) was used when preparing extracts for HPLCBrMMC. OA and DTXs are almost exclusively accumulated in digestive glands, which on average accounts for 20% of the mussel meat (Stabell et al., 1992), and the concentration of diarrhetic toxins are measured as OA eq./g DG. One gram DG corresponds to 5 g mussel meat and a factor of 20 was used to convert results from mg OA eq./g DG to mg OA eq./ 100 g mussel meat. 2.7. HPLC OA and DTX-1 were analysed by HPLC in two different ways; one method was according to Lee et al. (1987) with modi®cations in the silica SPE column clean-up step (Aase and Rogstad, 1997) using ADAM for derivatization, while derivatization was performed with BrMMC in the alternative method (Shen et al., 1997).
2 35 95 45 0.71 2 41 95 39 0.76 4 48 93 32 0.79 4 52 93 28 0.81 7 55 90 25 0.81 18 65 34 71 63 9 0.74
79 15 0.80
0 36 52 30 0.73 1 37 51 29 0.73 2 44 50 22 0.78 4 46 48 20 0.78 6 48 46 18 0.78 6 53 10 61 41 6 0.86
46 13 0.83
0 9 24 15 0.68 0 10 24 14 0.70 0 11 24 13 0.72 0 12 24 12 0.74 0 15 24 9 0.80 1 16 23 8 0.80 1 19 22 6 0.84
. 10 mg # 10 mg . 8 mg # 8 mg . 6 mg # 6 mg . 5 mg # 5 mg . 4 mg
1389
BrMMC n 48 Alive Dead Kappa Adam n 118 Alive Dead Kappa PP2A n 177 Alive Dead Kappa
The HPLC-BrMMC results in connection with the mouse bioassay indicate a preferable cut off value to be #4 mg OA eq./5 g DG (Table 1). Above this level, the agreement between the two methods was found substantially reduced, demonstrated by a large reduction in the kappa value. The connection between the HPLC-ADAM method and the mouse bioassay indicates a preferable cut off value of #6 mg OA eq./5 g DG. This choice seems to give a similar suf®cient agreement as the HPLC-BrMMC method in relation to the mouse bioassay (Table 1). Similarly, a cut off value of either #5 or #6 mg OA eq./5 g DG seems to be appropriate when comparing the PP2A assay with the mouse bioassay. Due to the limited number of HPLC-BrMMC analyses performed, only the HPLC-ADAM and the PP2A assays were internally compared. These two latter methods were compared with the mouse bioassay at concentrations from 4 to 6 mg OA eq./5 g DG [Table 2(a)±(c)]. The two methods displayed suf®cient agreement with kappa-values ranging from 0.88 to 0.94. With a cut off value of #4 mg, the two methods disagreed regarding classi®cation in 13 cases [Table 2(a)]. By using the mouse bioassay as the `golden standard', HPLC-ADAM correctly classi®ed ®ve of these 13 samples above and three below the limit. In contrast, PP2A
# 4 mg
3. Results
. 3 mg
All results are expressed in cross-tables as observed numbers. The agreement between the mouse bioassay and the chemical methods, and between the different chemical methods is presented by kappa-values (Agresti, 1990). For comparison of the chemical methods, McNehmar test was used (Altman, 1991).
# 3 mg
2.9. Statistical methods
. 2 mg
This assay is based on the ability of PP2A to dephosphorylate a colourless substrate to a yellow product in alkaline medium. The procedure is described in Promega's Technical Bulletin, No 537 (Anonymous, 1995). DTX-1 was used as a standard since it is the dominating DSP toxin in the Sognefjord (Lee et al., 1988), while OA only accounts for about 10% of the DSP toxicity (Stabell et al., 1992). The relative activity of PP2A in the presence of the various DTX-1 concentrations was calculated and used to produce a calibration curve (Anonymous, 1995). The linear area for determination of DTX-1 was in the range 0.5±4 ng DTX-1/ ml when using 0.125 U of enzyme per well with a total reaction volume of 110 ml. The reactions were carried out for 15 min at 308 C and the absorbency values were read on a Multiscan RC (Labsystem, Finland) at 405 nm. Each determination was performed at least in duplicate, and the concentration of toxin was calculated using the linear portion of the standard curve.
# 2 mg
2.8. PP2A inhibition assay
Table 1 Comparison of two HPLC methods and the PP2A assay with the mouse bioassay, using concentrations varying from 2 to 10 mg OA equivalents/5 g digestive gland (alive: two of two mice alive; dead: at least one mouse dead)
H. Ramstad et al. / Toxicon 39 (2001) 1387±1391
1390
H. Ramstad et al. / Toxicon 39 (2001) 1387±1391
Table 2 Internal comparison of HPLC-ADAM and the PP2A assay with the mouse bioassay at concentrations from 4 to 6 mg OA equivalents/ 5 g digestive gland PP2A
(a)
HPLC-ADAM
#4 .4
(b)
#5 .5
(c)
#6 .6
#4
.4
Alive Dead Alive Dead
43 14 3 3
2 5 4 44
Alive Dead Alive Dead
#5 47 18 1 2
.5 1 3 3 43
Alive Dead Alive Dead
#6 48 20 2 2
.6 0 4 2 40
total number of failure and correct classi®cation by the HPLC-ADAM and the PP2A method was compared for the three cut off values (Table 3). No signi®cant differences between the two methods were detected, but all differences found were in favour of HPLC-ADAM. The largest difference between the two methods was found when choosing the cut off value to #6 mg (P 0.15). 4. Discussion
correctly classi®ed two of the 13 samples below and three above the limit [Table 2(a)]. Consequently, HPLC-ADAM correctly classi®ed eight, and PP2A ®ve of the 13 samples of disagreement. Using cut off values of #5 and #6 mg, disagreements between the two methods occurred in seven and eight cases, respectively [Table 2(b) and (c)]. The most pronounced difference between the two methods in correct classi®cation by using the mouse bioassay as the golden standard was detected by choosing the cut off value of #6 mg [Table 2(c)]. In this situation, the HPLC- ADAM method correctly classi®ed four of the eight samples above and two below the limit, whereas PP2A only correctly classi®ed two of the eight samples above and none below the limit. Consequently, HPLC-ADAM correctly classi®ed six, and PP2A only two of the eight samples of disagreement. The least difference in correct classi®cation of the disagreed samples between the two methods was found for the chosen cut off value of #5 mg [Table 2(b)]. However, also in this situation HPLC-ADAM correctly classi®ed more samples compared to PP2A. With reference to the results from the mouse bioassay, the
The results indicate that the three methods were slightly different regarding sensitivity when compared with the mouse bioassay. The HPLC-ADAM method seemed to be the most sensitive, while the HPLC-BrMMC method seemed to be the least sensitive in comparison with the mouse bioassay, showing suf®cient agreement within a smaller range of cut off values. All three methods gave suf®cient agreement with the mouse bioassay at 4 mg OA equivalents (eq.) per 5 g DG. This level corresponds to 16 mg OA eq. per 100 g mussel meat, and is recommended by the European Union as the regulatory limit. With reference to the results of the mouse bioassay in the present study, the percentage of misclassi®cation according to the HPLC-ADAM or the PP2A assay at 4 mg OA eq. was 14 and 15%, respectively, while when choosing a cut off value of 6 mg OA eq. the percentage of misclassi®cation was 19 and 18%, respectively. The mouse bioassay occasionally showed relatively high toxicity in spite of relatively low concentrations of OA/ DTX-1, resulting in `false' positives related to risk of diarrhea in humans. In these cases the mice demonstrated symptoms deviating from the diarrhetic toxins with survival times often around 60 min, and indicated therefore a possible coextraction of non-diarrhetic toxins. The major toxin in the Sognefjord besides DTX-1 is YTX (Lee et al., 1988; Ramstad et al., 2001a). YTX is highly toxic towards mice when administered i.p., while its oral toxicity is at least 10 times lower (Ogino et al., 1997). Because YTX lacks any diarrhetic effect (Terao et al., 1990), the questioned has been raised in several publications whether YTX should be removed from the DSP toxin complex (Daiguji et al., 1998; Aune, 1997). By choosing the ether method for the
Table 3 Connections between HPLC-ADAM and the PP2A assay in failure and correct classi®cation related to the results of the mouse bioassay. The table is divided in three parts. One for each of the three cut-off values ,4, ,5 and ,6 mg OA equivalents/5 g digestive gland HPLC Right PP2A Total
Right
87
Failure
8 95
Failure ,4
Right
5
92
90
18 23
26 118
4 94
Failure ,5
Right
3
93
88
21 24
25 118
6 94
Failure ,6
2
90
22 24
28 118
H. Ramstad et al. / Toxicon 39 (2001) 1387±1391
mouse bioassay, most of the YTX present will remain in the water phase, but from time to time it is extracted to some extent by ether (Ramstad et al., 2001a). A few of the samples analysed by the mouse bioassay failed to detect relatively high concentrations of DTX-1. This phenomenon has not been discussed before and the reason is not known. The ethical aspects of using animals is highly questionable, and if the aim is to determine the presence of diarrhetic toxins in mussels, the mouse bioassay for DSP toxins could be replaced by either of these analytical methods used in the present study. However, if PTXs and YTXs are removed from the DSP toxin complex, tolerance level for these toxins must be established internationally before analytical methods can replace the mouse bioassays for commercial mussel control. In the meantime, both HPLC methods and the PP2A assay can at least be used as supplementary tests for detecting toxic levels of diarrhetic toxins in mussels, and thereby reduce the number of animals used for toxicity testing. Due to the validity of the three methods related to the mouse bioassay, the HPLC-ADAM method seems to be the method of choice. Acknowledgements The authors are grateful to Peter Hovgaard (Sogn and Fjordane College) for providing mussel samples. Technical assistance from Brit Heidenreich, Marianne Tomtum, Astrid Gravelle and Nanna Bruun Bremnes is very much appreciated. Financial support from the Norwegian Research Council and the Sogn and Fjordane County is gratefully acknowledged. Also, the Ministries of Fisheries, Agriculture and Labour are acknowledged for ®nancial support. References Anonymous, 1995. Protein phosphatase-2A catalytic subunit/ PNPP activity assay. Promega (WI, USA): Technical Bulletin No. 537. Aase, B., Rogstad, A., 1997. Optimization of sample cleanup procedure for determination of diarrhetic shell®sh poisoning toxins by use of experimental design. Journal of Chromatography A 764, 223±231. Agresti, A., 1990. Categorical Data Analysis. John Wiley and Sons, New York. Altman, D.G., 1991. Practical Statistics for Medical Research. 1st ed. Chapman and Hall, London. Apeldoorn, M.E.V., Egmond, H.P.V., Speijers, G.J.A., 1998. Diarrhoeic shell®sh poisoning: A review; 5722A00, p. 1. Aune, T., 1997. Health effects associated with algal toxins from seafood. Arch. Toxicol. 19(suppl.), 389±397. Daiguji, M., Satake, M., Ramstad, H., Aune, T., Naoki, H., Yasumoto, T., 1998. Structure and ¯uorometric HPLC determination of 1-desulfoyessotoxin, a new yessotoxin analog isolated from mussels from Norway. Nat. Toxins 6, 235±239.
1391
Hamano, Y., Kinoshita, Y., Yasumoto, T., 1986. Enteropathogenicity of diarrhetic shell®sh toxins in intestinal models. J. Food Hyg. Soc. Jpn. 27, 375±379. Kumagai, M., Yanagi, T., Murata, M., Yasumoto, T., Kat, M., Lassus, P., Rodriguez-Vazquez, J.A., 1986. Okadaic acid as the causative toxins of diarrhetic shell®sh poisoning in Europe. Agric. Biol. Chem. 50, 2853±2857. Lee, J.-S., Yanagi, T., Kenma, R., Yasumoto, T., 1987. Fluorometric determination of diarrhetic shell®sh toxins by highperformence liquid chromatography. Agric. Biol. Chem. 51, 877±881. Lee, J.S., Tangen, K., Dahl, E., Hovgaard, P., Yasumoto, T., 1988. Diarrhetic shell®sh toxins in Norwegian mussels. Bull. Japan Soc. Sci. Fish. 54, 1953±1957. Ogino, H., Kumagai, M., Yasumoto, T., 1997. Toxicologic evaluation of yessotoxin. Nat. Toxins 5, 255±259. Ramstad, H., Hovgaard, P., Yasumoto, T., Larsen, S., Aune, T., 2001a. Monthly variations in diarrhetic toxins and yessotoxin in shell®sh from coast to the inner part of the Sognefjord, Norway. Toxicon 39 (7), 1035±1043. Ramstad, H., Larsen, S., Aune, T., 2001b. The repeatability of two HPLC methods and the PP2A assay in quanti®cation of diarrhetic toxins in blue mussels (Mytilus edulis). Toxicon 39, 515± 522. Shen, J.L., Hummert, C., Luckas, B., 1997. Sensitive HPLC-¯uorometric and HPLC-MS determination of diarrhetic shell®sh poisoning (DSP)-toxins as 4-bromomethyl-7-methoxycoumarin esters. Fresenius J. Anal. Chem. 357, 101±104. Stabell, O.B., Yndestad, M., Heidenreich, B., 1991. Paralytic shell®sh toxins seem absent in extracts of diarrhetic shell®sh toxins. Environ. Toxicol. Chem. 10, 331±334. Stabell, O.B., Steffenak, I., Aune, T., 1992. An evaluation of the mouse bioassay applied to extracts of `diarrhoetic' shell®sh toxins. Food & Chemical Toxicology 30, 139±144. Terao, K., Ito, E., Oarada, M., Murata, M., Yasumoto, T., 1990. Histopathological studies on experimental marine toxin poisoning-5: the effects in mice of yessotoxin isolated from Patinopecten yessoensis and of a desulfated derivative. Toxicon 28, 1095±1104. Toti, L., Croci, L., De Medici, D., Gizzarelli, S., Di Pasquale, M., Ore®ce, L., Stazi, A., 1991. Evaluation of Yasumotos test for the determination of DSP toxin in shell®sh. Proceedings of Symposium on Marine Biotoxins, Paris, pp. 107±110. Yasumoto, T., Oshima, Y., Sugawara, W.F.Y., Oguri, H., Igarashi, T., Fujita, N., 1980. Identi®cation of Dinophysis fortii as the causative organism of diarrhetic shell®sh poisoning. Bull. Jap. Soc. Sci. Fish. 46, 1405±1411. Yasumoto, T., Murata, M., Oshima, Y., Matsumoto, G.K., Clardy, J. (Eds.), 1984. Diarrhetic shell®sh poisoning. In: Seafood Toxins. American Chemical Society, Washington DC, pp. 207±214. Yasumoto, T., Murata, M., Lee, J. S., Torigoe, K., 1989. Polyether toxins produced by dino¯agellates. In: Natori, S., Hashimoto, K., Ueno, Y. (Eds.), Mycotoxins and Phycotoxins '88. Elsevier, Amsterdam, pp. 375±382. Yasumoto, T., Murata, M., 1993. Marine toxins. Chem. Rev. 93, 1897±1909.
www.elsevier.com/locate/toxicon
The validity of two HPLC methods and a colorimetric PP2A assay related to the mouse bioassay in quanti®cation of diarrhetic toxins in blue mussels (Mytilus edulis) H. Ramstad a,*, J.L. Shen a, S. Larsen b, T. Aune a a
Department of Pharmacology, Microbiology and Food Hygiene, Norwegian School of Veterinary Science, PO Box 8146, Department 0033, Oslo, Norway b Department of Large Animal Clinical Sciences, Norwegian School of Veterinary Science, Oslo, Norway Received 13 October 2000; accepted 27 January 2001
Abstract Validity of two HPLC methods and a PP2A assay in relation to the mouse bioassay for diarrhetic shell®sh poisoning (DSP) toxins was evaluated. The mouse bioassay for DSP toxins was performed on a total of 177 mussel samples from the Sognefjord, Norway, using diethyl ether in the ®nal step of extraction. For ¯uorimetric HPLC analyses, either 4-bromomethyl-7-methoxycoumarin (BrMMC) or 9-anthryl diazomethane (ADAM) were used for analysis of 48 and 118 of the samples, respectively. The colorimetric PP2A inhibition assay was performed on all 177 samples that were analysed with the mouse bioassay. When comparing the HPLC-BrMMC, the HPLC-ADAM and the PP2A assays with the mouse bioassay, cut off values of #4, 5 and 6 mg okadaic acid (OA) equivalents (eq.)/5 g digestive gland (DG) was used. With reference to the results from the mouse bioassay, the total number of failure and correct classi®cation by HPLC-ADAM and the PP2A method was compared for the three cut off values. No signi®cant differences between the methods were detected. However, all differences were found in favour of HPLC-ADAM. All three methods could replace the mouse bioassay in detecting levels of diarrhetic toxins approved internationally for safe consumption of mussels. However, HPLC-ADAM seems to be the method of choice. q 2001 Elsevier Science Ltd. All rights reserved. Keywords: DSP toxins; HPLC; Mouse bioassay; PP2A
1. Introduction A well-known and widespread human toxic syndrome is called diarrhetic shell®sh poisoning (DSP) on account of its predominant gastrointestinal symptoms (Yasumoto et al., 1980). The most widely used screening method for these toxins is the mouse bioassay (van Apeldoorn et al., 1998), but the choice of extraction procedure highly in¯uences the outcome of the DSP mouse bioassay (Aune, 1997). Using diethyl ether in the ®nal step of extraction (Yasumoto et al., 1984) normally facilitate extraction of the truly diarrhetic toxins, while chloroform extraction includes a wider spec* Corresponding author: c/o Dr Kevin James, Cork Institute of Technology, Cork, Ireland. Fax: 1353-21-345-191. E-mail address: [email protected] (H. Ramstad).
trum of toxins, including pectenotoxins (PTXs), yessotoxins (YTXs) and among others, unknown toxin(s) with neurotoxic effects (Stabell et al., 1991; Ramstad et al., 2001a). YTX is still considered to belong to the group of DSP toxins although it does not induce diarrhea (Terao et al., 1990). Okadaic acid (OA) and the dinophysis toxins (DTXs) are among the most important toxins for the mussel industry due to their worldwide occurrence. (Yasumoto and Murata, 1993; Kumagai et al., 1986). The minimum doses of OA and DTX-1 necessary to induce diarrhea in adults have been estimated to be 40 and 36 mg, respectively (Hamano et al., 1986). By means of titration of toxicity with OA, one mouse unit (MU) corresponds with 4 mg OA, and is de®ned as the minimum quantity of toxin capable of killing a 20-g mouse within 24 h after intraperitoneal injection (i.p.) (Yasumoto et al., 1984, 1989). Most countries with established regulations for DSP toxins apply an acceptance level of 4±5 mouse
0041-0101/01/$ - see front matter q 2001 Elsevier Science Ltd. All rights reserved. PII: S 0041-010 1(01)00097-6
1388
H. Ramstad et al. / Toxicon 39 (2001) 1387±1391
units or 16±20 mg OA equivalents (eq.)/100 g mussel meat (Yasumoto et al., 1984; Aune, 1997). There are ethical and economic concerns associated with the mouse bioassay, and furthermore it lacks precision regarding death time (Toti et al., 1991). In addition, the mouse bioassay is time consuming and may give false positives with respect to risk of diarrhea if the ®nal extraction is acetone or chloroform. Due to increased availability of commercial standards, a number of biochemical and chemical methods have been developed (Apeldoorn et al., 1998). The diarrhetic toxins, measured as OA equivalents, were found to give repeatable and reliable results when using both ¯uorimetric HPLC methods and a colorimetric PP2A assay (Ramstad et al., 2001b). In this study, the validity of two HPLC methods and a colorimetric PP2A assay were evaluated against the mouse bioassay with ether extracts for quanti®cation of the diarrhetic toxins in blue mussels (Mytilus edulis). 2. Materials and methods 2.1. Reagents Solvents used for the HPLC analyses were acetonitrile and methanol of LC quality supplied by Romil (Cambridge, UK), and the water was Millipore-Q cleaned. The derivatizing agent 9-anthryl diazomethane (ADAM) was purchased from Molecular Probes (Oregon, USA) and the derivatizing agent 4-bromomethyl-7-methoxycoumarin (BrMMC) was from Sigma (USA). OA was supplied by Marine Analytical Chemistry Standard Program (Halifax, Canada) and DTX-1 was obtained from Wako Chemicals GmbH (Germany). pNitrophenylphosphate was supplied by SIGMA Bio Sciences (USA), and protein phosphatase 2A (25 U) isolated from rabbit skeletal muscle was bought from Promega (WI, USA). Other solvents and reagents used were of analytical grade. 2.2. Sampling stations Nine locations in the Sognefjord, Norway were selected for sampling of blue mussels (Mytilus edulis), covering a distance of about 200 km from the coast to the innermost part of the fjord (Ramstad et al., 2001a). Mussels (1±2 kg) were sampled every other week during the monitoring periods, from July until the end of October 1996, and from March to November, 1997. Additionally, eight samples from the period October to December 1998 were included. The mussels were evenly sampled from the surface to 6 m depth, except on the two innermost stations, where mussels do not survive the low salinity in the surface water. From the latter stations the samples were taken from 3 to 8 m depth. HPLC analyses for OA/DTX-1 were performed on 48 and 118 samples using BrMMC and ADAM, respectively, while the total samples of 177 were analysed using a colorimetric PP2A inhibition assay and the mouse bioassay.
2.3. Animals White, female mice, BOM: NMRI weighing between 15 and 20 g were used for mouse bioassays. 2.4. Toxin extraction for the mouse bioassay The mouse bioassay for the DSP toxins was performed essentially according to the method of Yasumoto et al. (1984) with slight modi®cations. Brie¯y, 20 g of homogenised digestive glands (DG) from each mussel sample was extracted with 200 ml acetone, ®ltered and taken to dryness on a rotary evaporator. The residue was partitioned between 16 ml petroleum ether and 8 ml 80% MeOH, and the petroleum ether phase was discarded. Defatting with 16 ml petroleum ether was repeated. The 80% MeOH-fraction was divided in two portions, and diethyl ether or chloroform was used in the ®nal step of extraction. In this study, only the ether fraction was studied. The 80% MeOH-fraction used for ether extraction was evaporated almost to dryness under nitrogen, and 3 ml water was then added. The aqueous suspension was extracted twice with 4 ml diethyl ether, and ®nally the combined ether fraction was backwashed with 1 ml water and then evaporated to dryness under nitrogen. All samples were stored at 2208C until use. 2.5. Mouse bioassay The residue from the mussel extract was suspended in 2 ml 1% Tween 60 solution and two mice received each i.p. 5 g DG per 20 g body weight (b.w.). The mice were observed for 24 h, and the de®nition of `dead' means that minimum one out of two mice analysed is dead. 2.6. Toxin extraction for biochemical and chemical analysis Toxin extraction was performed according to the method of Lee et al. (1987) when preparing extracts for HPLCADAM and the PP2A assay. The method of Shen et al. (1997) was used when preparing extracts for HPLCBrMMC. OA and DTXs are almost exclusively accumulated in digestive glands, which on average accounts for 20% of the mussel meat (Stabell et al., 1992), and the concentration of diarrhetic toxins are measured as OA eq./g DG. One gram DG corresponds to 5 g mussel meat and a factor of 20 was used to convert results from mg OA eq./g DG to mg OA eq./ 100 g mussel meat. 2.7. HPLC OA and DTX-1 were analysed by HPLC in two different ways; one method was according to Lee et al. (1987) with modi®cations in the silica SPE column clean-up step (Aase and Rogstad, 1997) using ADAM for derivatization, while derivatization was performed with BrMMC in the alternative method (Shen et al., 1997).
2 35 95 45 0.71 2 41 95 39 0.76 4 48 93 32 0.79 4 52 93 28 0.81 7 55 90 25 0.81 18 65 34 71 63 9 0.74
79 15 0.80
0 36 52 30 0.73 1 37 51 29 0.73 2 44 50 22 0.78 4 46 48 20 0.78 6 48 46 18 0.78 6 53 10 61 41 6 0.86
46 13 0.83
0 9 24 15 0.68 0 10 24 14 0.70 0 11 24 13 0.72 0 12 24 12 0.74 0 15 24 9 0.80 1 16 23 8 0.80 1 19 22 6 0.84
. 10 mg # 10 mg . 8 mg # 8 mg . 6 mg # 6 mg . 5 mg # 5 mg . 4 mg
1389
BrMMC n 48 Alive Dead Kappa Adam n 118 Alive Dead Kappa PP2A n 177 Alive Dead Kappa
The HPLC-BrMMC results in connection with the mouse bioassay indicate a preferable cut off value to be #4 mg OA eq./5 g DG (Table 1). Above this level, the agreement between the two methods was found substantially reduced, demonstrated by a large reduction in the kappa value. The connection between the HPLC-ADAM method and the mouse bioassay indicates a preferable cut off value of #6 mg OA eq./5 g DG. This choice seems to give a similar suf®cient agreement as the HPLC-BrMMC method in relation to the mouse bioassay (Table 1). Similarly, a cut off value of either #5 or #6 mg OA eq./5 g DG seems to be appropriate when comparing the PP2A assay with the mouse bioassay. Due to the limited number of HPLC-BrMMC analyses performed, only the HPLC-ADAM and the PP2A assays were internally compared. These two latter methods were compared with the mouse bioassay at concentrations from 4 to 6 mg OA eq./5 g DG [Table 2(a)±(c)]. The two methods displayed suf®cient agreement with kappa-values ranging from 0.88 to 0.94. With a cut off value of #4 mg, the two methods disagreed regarding classi®cation in 13 cases [Table 2(a)]. By using the mouse bioassay as the `golden standard', HPLC-ADAM correctly classi®ed ®ve of these 13 samples above and three below the limit. In contrast, PP2A
# 4 mg
3. Results
. 3 mg
All results are expressed in cross-tables as observed numbers. The agreement between the mouse bioassay and the chemical methods, and between the different chemical methods is presented by kappa-values (Agresti, 1990). For comparison of the chemical methods, McNehmar test was used (Altman, 1991).
# 3 mg
2.9. Statistical methods
. 2 mg
This assay is based on the ability of PP2A to dephosphorylate a colourless substrate to a yellow product in alkaline medium. The procedure is described in Promega's Technical Bulletin, No 537 (Anonymous, 1995). DTX-1 was used as a standard since it is the dominating DSP toxin in the Sognefjord (Lee et al., 1988), while OA only accounts for about 10% of the DSP toxicity (Stabell et al., 1992). The relative activity of PP2A in the presence of the various DTX-1 concentrations was calculated and used to produce a calibration curve (Anonymous, 1995). The linear area for determination of DTX-1 was in the range 0.5±4 ng DTX-1/ ml when using 0.125 U of enzyme per well with a total reaction volume of 110 ml. The reactions were carried out for 15 min at 308 C and the absorbency values were read on a Multiscan RC (Labsystem, Finland) at 405 nm. Each determination was performed at least in duplicate, and the concentration of toxin was calculated using the linear portion of the standard curve.
# 2 mg
2.8. PP2A inhibition assay
Table 1 Comparison of two HPLC methods and the PP2A assay with the mouse bioassay, using concentrations varying from 2 to 10 mg OA equivalents/5 g digestive gland (alive: two of two mice alive; dead: at least one mouse dead)
H. Ramstad et al. / Toxicon 39 (2001) 1387±1391
1390
H. Ramstad et al. / Toxicon 39 (2001) 1387±1391
Table 2 Internal comparison of HPLC-ADAM and the PP2A assay with the mouse bioassay at concentrations from 4 to 6 mg OA equivalents/ 5 g digestive gland PP2A
(a)
HPLC-ADAM
#4 .4
(b)
#5 .5
(c)
#6 .6
#4
.4
Alive Dead Alive Dead
43 14 3 3
2 5 4 44
Alive Dead Alive Dead
#5 47 18 1 2
.5 1 3 3 43
Alive Dead Alive Dead
#6 48 20 2 2
.6 0 4 2 40
total number of failure and correct classi®cation by the HPLC-ADAM and the PP2A method was compared for the three cut off values (Table 3). No signi®cant differences between the two methods were detected, but all differences found were in favour of HPLC-ADAM. The largest difference between the two methods was found when choosing the cut off value to #6 mg (P 0.15). 4. Discussion
correctly classi®ed two of the 13 samples below and three above the limit [Table 2(a)]. Consequently, HPLC-ADAM correctly classi®ed eight, and PP2A ®ve of the 13 samples of disagreement. Using cut off values of #5 and #6 mg, disagreements between the two methods occurred in seven and eight cases, respectively [Table 2(b) and (c)]. The most pronounced difference between the two methods in correct classi®cation by using the mouse bioassay as the golden standard was detected by choosing the cut off value of #6 mg [Table 2(c)]. In this situation, the HPLC- ADAM method correctly classi®ed four of the eight samples above and two below the limit, whereas PP2A only correctly classi®ed two of the eight samples above and none below the limit. Consequently, HPLC-ADAM correctly classi®ed six, and PP2A only two of the eight samples of disagreement. The least difference in correct classi®cation of the disagreed samples between the two methods was found for the chosen cut off value of #5 mg [Table 2(b)]. However, also in this situation HPLC-ADAM correctly classi®ed more samples compared to PP2A. With reference to the results from the mouse bioassay, the
The results indicate that the three methods were slightly different regarding sensitivity when compared with the mouse bioassay. The HPLC-ADAM method seemed to be the most sensitive, while the HPLC-BrMMC method seemed to be the least sensitive in comparison with the mouse bioassay, showing suf®cient agreement within a smaller range of cut off values. All three methods gave suf®cient agreement with the mouse bioassay at 4 mg OA equivalents (eq.) per 5 g DG. This level corresponds to 16 mg OA eq. per 100 g mussel meat, and is recommended by the European Union as the regulatory limit. With reference to the results of the mouse bioassay in the present study, the percentage of misclassi®cation according to the HPLC-ADAM or the PP2A assay at 4 mg OA eq. was 14 and 15%, respectively, while when choosing a cut off value of 6 mg OA eq. the percentage of misclassi®cation was 19 and 18%, respectively. The mouse bioassay occasionally showed relatively high toxicity in spite of relatively low concentrations of OA/ DTX-1, resulting in `false' positives related to risk of diarrhea in humans. In these cases the mice demonstrated symptoms deviating from the diarrhetic toxins with survival times often around 60 min, and indicated therefore a possible coextraction of non-diarrhetic toxins. The major toxin in the Sognefjord besides DTX-1 is YTX (Lee et al., 1988; Ramstad et al., 2001a). YTX is highly toxic towards mice when administered i.p., while its oral toxicity is at least 10 times lower (Ogino et al., 1997). Because YTX lacks any diarrhetic effect (Terao et al., 1990), the questioned has been raised in several publications whether YTX should be removed from the DSP toxin complex (Daiguji et al., 1998; Aune, 1997). By choosing the ether method for the
Table 3 Connections between HPLC-ADAM and the PP2A assay in failure and correct classi®cation related to the results of the mouse bioassay. The table is divided in three parts. One for each of the three cut-off values ,4, ,5 and ,6 mg OA equivalents/5 g digestive gland HPLC Right PP2A Total
Right
87
Failure
8 95
Failure ,4
Right
5
92
90
18 23
26 118
4 94
Failure ,5
Right
3
93
88
21 24
25 118
6 94
Failure ,6
2
90
22 24
28 118
H. Ramstad et al. / Toxicon 39 (2001) 1387±1391
mouse bioassay, most of the YTX present will remain in the water phase, but from time to time it is extracted to some extent by ether (Ramstad et al., 2001a). A few of the samples analysed by the mouse bioassay failed to detect relatively high concentrations of DTX-1. This phenomenon has not been discussed before and the reason is not known. The ethical aspects of using animals is highly questionable, and if the aim is to determine the presence of diarrhetic toxins in mussels, the mouse bioassay for DSP toxins could be replaced by either of these analytical methods used in the present study. However, if PTXs and YTXs are removed from the DSP toxin complex, tolerance level for these toxins must be established internationally before analytical methods can replace the mouse bioassays for commercial mussel control. In the meantime, both HPLC methods and the PP2A assay can at least be used as supplementary tests for detecting toxic levels of diarrhetic toxins in mussels, and thereby reduce the number of animals used for toxicity testing. Due to the validity of the three methods related to the mouse bioassay, the HPLC-ADAM method seems to be the method of choice. Acknowledgements The authors are grateful to Peter Hovgaard (Sogn and Fjordane College) for providing mussel samples. Technical assistance from Brit Heidenreich, Marianne Tomtum, Astrid Gravelle and Nanna Bruun Bremnes is very much appreciated. Financial support from the Norwegian Research Council and the Sogn and Fjordane County is gratefully acknowledged. Also, the Ministries of Fisheries, Agriculture and Labour are acknowledged for ®nancial support. References Anonymous, 1995. Protein phosphatase-2A catalytic subunit/ PNPP activity assay. Promega (WI, USA): Technical Bulletin No. 537. Aase, B., Rogstad, A., 1997. Optimization of sample cleanup procedure for determination of diarrhetic shell®sh poisoning toxins by use of experimental design. Journal of Chromatography A 764, 223±231. Agresti, A., 1990. Categorical Data Analysis. John Wiley and Sons, New York. Altman, D.G., 1991. Practical Statistics for Medical Research. 1st ed. Chapman and Hall, London. Apeldoorn, M.E.V., Egmond, H.P.V., Speijers, G.J.A., 1998. Diarrhoeic shell®sh poisoning: A review; 5722A00, p. 1. Aune, T., 1997. Health effects associated with algal toxins from seafood. Arch. Toxicol. 19(suppl.), 389±397. Daiguji, M., Satake, M., Ramstad, H., Aune, T., Naoki, H., Yasumoto, T., 1998. Structure and ¯uorometric HPLC determination of 1-desulfoyessotoxin, a new yessotoxin analog isolated from mussels from Norway. Nat. Toxins 6, 235±239.
1391
Hamano, Y., Kinoshita, Y., Yasumoto, T., 1986. Enteropathogenicity of diarrhetic shell®sh toxins in intestinal models. J. Food Hyg. Soc. Jpn. 27, 375±379. Kumagai, M., Yanagi, T., Murata, M., Yasumoto, T., Kat, M., Lassus, P., Rodriguez-Vazquez, J.A., 1986. Okadaic acid as the causative toxins of diarrhetic shell®sh poisoning in Europe. Agric. Biol. Chem. 50, 2853±2857. Lee, J.-S., Yanagi, T., Kenma, R., Yasumoto, T., 1987. Fluorometric determination of diarrhetic shell®sh toxins by highperformence liquid chromatography. Agric. Biol. Chem. 51, 877±881. Lee, J.S., Tangen, K., Dahl, E., Hovgaard, P., Yasumoto, T., 1988. Diarrhetic shell®sh toxins in Norwegian mussels. Bull. Japan Soc. Sci. Fish. 54, 1953±1957. Ogino, H., Kumagai, M., Yasumoto, T., 1997. Toxicologic evaluation of yessotoxin. Nat. Toxins 5, 255±259. Ramstad, H., Hovgaard, P., Yasumoto, T., Larsen, S., Aune, T., 2001a. Monthly variations in diarrhetic toxins and yessotoxin in shell®sh from coast to the inner part of the Sognefjord, Norway. Toxicon 39 (7), 1035±1043. Ramstad, H., Larsen, S., Aune, T., 2001b. The repeatability of two HPLC methods and the PP2A assay in quanti®cation of diarrhetic toxins in blue mussels (Mytilus edulis). Toxicon 39, 515± 522. Shen, J.L., Hummert, C., Luckas, B., 1997. Sensitive HPLC-¯uorometric and HPLC-MS determination of diarrhetic shell®sh poisoning (DSP)-toxins as 4-bromomethyl-7-methoxycoumarin esters. Fresenius J. Anal. Chem. 357, 101±104. Stabell, O.B., Yndestad, M., Heidenreich, B., 1991. Paralytic shell®sh toxins seem absent in extracts of diarrhetic shell®sh toxins. Environ. Toxicol. Chem. 10, 331±334. Stabell, O.B., Steffenak, I., Aune, T., 1992. An evaluation of the mouse bioassay applied to extracts of `diarrhoetic' shell®sh toxins. Food & Chemical Toxicology 30, 139±144. Terao, K., Ito, E., Oarada, M., Murata, M., Yasumoto, T., 1990. Histopathological studies on experimental marine toxin poisoning-5: the effects in mice of yessotoxin isolated from Patinopecten yessoensis and of a desulfated derivative. Toxicon 28, 1095±1104. Toti, L., Croci, L., De Medici, D., Gizzarelli, S., Di Pasquale, M., Ore®ce, L., Stazi, A., 1991. Evaluation of Yasumotos test for the determination of DSP toxin in shell®sh. Proceedings of Symposium on Marine Biotoxins, Paris, pp. 107±110. Yasumoto, T., Oshima, Y., Sugawara, W.F.Y., Oguri, H., Igarashi, T., Fujita, N., 1980. Identi®cation of Dinophysis fortii as the causative organism of diarrhetic shell®sh poisoning. Bull. Jap. Soc. Sci. Fish. 46, 1405±1411. Yasumoto, T., Murata, M., Oshima, Y., Matsumoto, G.K., Clardy, J. (Eds.), 1984. Diarrhetic shell®sh poisoning. In: Seafood Toxins. American Chemical Society, Washington DC, pp. 207±214. Yasumoto, T., Murata, M., Lee, J. S., Torigoe, K., 1989. Polyether toxins produced by dino¯agellates. In: Natori, S., Hashimoto, K., Ueno, Y. (Eds.), Mycotoxins and Phycotoxins '88. Elsevier, Amsterdam, pp. 375±382. Yasumoto, T., Murata, M., 1993. Marine toxins. Chem. Rev. 93, 1897±1909.