Journal Pre-proof Paralytic shellfish toxin profiles in mussel, cockle and razor shell under post-bloom natural conditions: Evidence of higher biotransformation in razor shells and cockles Evidence of

15 Concentrations of the paralytic shellfish toxins GTX6, C1+2, GTX5, C3+4, dcSTX, dcNEO and dcGTX2+3 were 16 determined by LC-FLD in composite samples of whole soft tissues of mussels ( Mytilus galloprovincialis ) , cockles 17 ( Cerastoderma edule ) and razor shells ( Solen marginatus ) after exposure to a Gymnodinium catenatum bloom . 18 Specimens were harvested weekly during three months under natural depuration conditions in the Mira 19 branch of Aveiro lagoon, Portugal. Under the decline of G. catenatum cell densities, elimination or 20 transformation of the uptake toxins associated with the ingestion of toxic cells differed among the surveyed 21 species. Ratio between the toxins dcSTX plus dcGTX2+3 plus dcNEO and toxins GTX6 plus GTX5 plus C1+2 plus 22 C3+4 was used to illustrate the biotransformation occurring in the bivalves. Enhancement of the ratios was 23 observed for razor shells and cockles seven weeks after the peak of the algal bloom. Most likely it reflects more intense biotransformation in razor shells and cockles than in mussels. Conversion into toxins of higher 2 toxicity may prolong the bivalve toxicity. These results show the complexity of toxin elimination in bivalves 3 under post-bloom conditions and emphasize the pertinence of monitoring programs of bivalve toxicity in order 4 to protect human health.


9
Twelve individuals of each species were randomly selected for molecular taxonomic analyses to assure the 10 identification of the collected specimens. Taxonomic identification of the three species followed the procedure

12
Ten specimens of mussels and razor shells, and 40 specimens of cockles were used for toxin determinations.

13
These specimens were sacrificed and dissected to obtain composite samples of whole soft tissues.  sodium hydroxide solution and vortex mixed, then the   2   C18-cleaned extract or the PST standard solution was added, and the mixture was thoroughly mixed and   3 allowed to react for 2 min at room temperature; then glacial acetic acid was added and vortex mixed before 4 LC-FLD analysis. In parallel, in order to quantify N-hydroxylated toxins a C18-cleaned extract (for dcNEO), a 5 SPE-COOH fraction (for NEO, GTX1+4, C3+4 and GTX6) or a PST standard solution (for calibration) was added to 6 a matrix modifier solution prepared with PST-free oysters, and to that it was added the periodate oxidant (0.3 7 mol L -1 ammonium formate, 0.3 mol L -1 disodium phosphate, 0.03 mol L -1 periodic acid, with adjusted pH to 8 8.2); the solution was thoroughly mixed and allowed to react for 1 min, then glacial acetic acid was added and 9 the mixture was allowed to stand for further 10 min before LC-FLD analysis. were performed with car and PMCMR packages, respectively. Data were tested for normality and 10 homogeneity of variance with Shapiro-Wilk and Levene tests, respectively. Since these assumptions could not 11 be met, the non-parametric Kruskal-Wallis test was used to assess differences in temporal variation of toxin 12 concentrations and decarbamoil to N-sulfocarbamoil ratios over the study period. Whenever Kruskal-Wallis 13 test was significant, the post-hoc Conover test was performed for multiple comparisons between groups. The 14 probability lower than 0.05 was considered as statistically significant.   ). Consequently, 17 harvesting in that production area was interdicted by the National Authority (IPMA).

19
3.4. Toxin profiles of mussels, cockles and razor shells exposed to an algal bloom 20 Figure 4 shows the toxin profiles, expressed as molar fractions, of the quantified toxins in the specimens 21 collected on 9 th January. GTX6, C1+2, GTX5, C3+4, dcSTX and dcGTX2+3 were the major contributors to the 22 toxin profiles in composite samples of the three species. Toxin dcNEO was a minor contributor in mussels and 23 cockles, and was undetected in razor shells. The compound GTX6 accounted for 35, 33 and 30% of total 24 quantified toxins in mussels, cockles and razor shells, respectively. Furthermore, the results showed higher 25 proportion of C1+2 in cockles (34%) than in mussels (24%) and in razor shells (16%), while the compound GTX5 26 reached 24% in razor shells, clearly above the 13% in mussels and 11% in cockles. Differences among species 27 were also observed in the toxins dcSTX, dcGTX2+3 and dcNEO that were present in lower proportions. 9 1 3.5. Toxin concentrations in mussels, cockles and razor shells under post-bloom conditions 2 Concentration of all detected PSTs in bivalve whole soft tissues collected weekly in the the study area 3 between 9 th January and 3 rd April 2017 are given as supplementary data (supplementary Table S1). GTX6, 4 C1+2, GTX5, C3+4, dcSTX, dcNEO and dcGTX2+3 were detected in most of the samples of the three species.

5
The toxins GTX2+3, STX, GTX1+4 and NEO were below the detection limit throughout the observation period. , respectively) and in mussels (7.1 and 8.7 µg g ). The same pattern was observed for the toxins GTX5, C3+4, dcSTX and 12 dcNEO, although toxin concentrations in cockles and mussels were only 2 to 4 times higher than in razor shells.

13
Comparison of dcNEO could not be done because values in razor shells were below the detection limit. On 9 th 14 January, mean concentrations of dcGTX2+3 were higher in cockles (1.3 µg g -1 ) than in mussels (0.44 µg g 2 3 3.6. Time-course variation of bivalve toxicity 4 Figure 6 shows the variation of total toxicities in the bivalve specimens collected from 9 th January to 3 rd April, 5 2017. During this period, mean toxicities decreased from 4013 to 165 (mussels), 4721 to 125 (cockles) and 6 1176 to 133 (razor shells) µg STX di-HCl equivalents kg -1. . In all species higher toxicities were registered one 7 week after the enhancement of toxic phytoplankton cells (2 nd January, 2320 cells L -1 ). Toxicity values of 8 mussels and cockles were higher than of razor shells. Enhancement of toxic cells on 30 th January seems to have 9 contributed to the increase of mussel toxicities in the following dates, although relative constant values were 10 observed in cockles and razor shells. During four weeks, toxicity in the three species displayed values above 11 the RL for PSTs. Approximately three months after the initial time of observations, toxicity levels observed in 12 razor shells, mussels and cockles decreased to 11, 5 and 3% of the respective initial values. 16 Figure 7 shows the mean R calculated for mussels, cockles and razor shells under post-bloom conditions, 17 between 9 th January and 3 rd April. During the first three weeks, the ratios were higher for razor shells than for 18 cockles and mussels.

25
Elevated ratios for razor shells decreased abruptly from 0.38 to 0.06 between 6 th and 13 th March. In the same 26 period, cockles ratios decreased from 0.36 to 0.21, and mussels ratios fluctuated between 0.14 and 0.08. below the RL in the following week, contrasting to toxicities of mussels that remained above RL for another 21 four weeks. However, during the surveyed period the compounds that contributed to the total toxicity differed among the bivalve species. Contribution of decarbamoil derivatives (dcSTX, dcGTX2+3 and dcNEO) for the 23 toxicity of razor shells was higher than in mussels and cockles that, on average, remained within the narrow 24 interval of 54-58% during both periods, above and below the RL. One possible explanation for these results is 1 on paralytic shellfish toxins quantification and toxicity estimation in mussels exposed to Gymnodinium          (Oshima, 1995b). Solid arrow: hydrolysis; dashed arrow: epimerisation. NEO -neosaxitoxin; STX -saxitoxin, GTX1 to GTX4gonyautoxins; GTX5, GTX6 and C1 to C4 -N-sulfocarbamoil toxins; dcGTX1 to dcGTX4 -decarbamoil toxins.