Oyster samples were collected from the border of the Bohai and the Yellow seas (Weihai, Shandong Province, China), and all the oysters were transported to the laboratory alive. After that, the oysters were acclimated in artificial seawater (32 g/L) for 24 hrs. The seawater was produced using seawater crystals (Tianjin Binhai New Area Tanggu Hai Sheng Seawater Crystal Factory, Tianjin, China) according to Bielmyer et al. (35). Then, the oysters were exposed to seawater with DEL (2 μg /L) for 72 hrs. DEL for contamination was obtained from Bayer CropScience (Hangzhou, Zhejiang Province, China). Salinity (27.0 ± 1.0‰) and water temperature (18 ± 1°C) were maintained throughout. Photoperiod at 12 h light/12 h dark was maintained. To prevent the residual chlorine in the tap water from affecting the oyster culture process, the seawater was oxygenated for 48 hrs before use.

All the artificially contaminated oysters were divided into four groups, of which three groups were subjected to different processing methods: raw, steamed (i.e., steaming in an steam cooker for 5 min, until the oyster shell opens), and roasted (i.e., in an oven at 200°C for 20 min, only use downfire). Oyster tissue was taken out and divided into four parts: gill, viscera, mantle, and adductor muscle. Twenty-five specimens were removed from the treated oysters. For all samples, the tissue was separated from the oyster-shell, cleaned with running water. The collected tissues were then drained and homogenized with a blender (JY-200B, Zhongshan Jiuyuan Electric Appliance Co. Ltd, Zhongshan, Guangdong Province, China) until visually homogenized. These samples were stored at −20°C for further analyses (maximum 4 weeks of storage).

The bioaccessibility of DEL in oysters was assessed using an in vitro digestion INFOGEST procedure previously described by Brodkorb et al. (36) with proper modifications. In short, the simulated human digestion methodology was carried out in three different phases (oral, gastric, and intestinal) using four digestive enzymes: pepsin (Aladdin, P110927), lipase (Aladdin, L299012), trypsin from swine (Sigma, P7545) and cow bile (Solarbio, B8210). And each oyster sample was digested in triplicate. For each sample, 5 g of oyster tissue was digested at 37°C using a constant temperature water bath shaker (JieRuiEr THZ-82; Changzhou, Jiangsu Province, China). The simulated digestion was done using the following protocol: oral phase (the tissue is diluted 1:1 with simulated salivary fluid at pH 7 ± 0.2; 2 min), gastric phase (the oral bolus is diluted 1:1 with simulated gastric fluid; 150 min), and intestinal phase (the gastric chyme is diluted 1:1 with simulated intestinal fluid at pH 7 ± 0.2; 120 min). Studies have shown that the pH in the gastric phase is constantly changing during the digestion process (37). To better simulate the real physiological environment during the gastric digestion process, the pH is continuously reduced by adding hydrochloric acid (the total amount of hydrochloric acid added is 2 mL, 0.4 mL is added at the beginning of the gastric phase, and then 0.2 mL is added every 15 min) during the experiment. The change of pH in the gastric phase is shown in Figure 1. Each simulated digestion fluid was prepared just before simulated digestion to avoid the loss of enzymatic activity. At the end of the simulated digestion, the pH of gastric samples was adjusted to 8.0 ± 0.2 and the intestinal samples were heated in the water bath at 95°C for 10 min to stop the digestion process. After that, the digested samples were centrifuged at 2,930 × g at 4°C for 5 min to separate the bioaccessible and non-bioaccessible fractions.

Considering that the content of DEL in the digestive phase is changing, the steamed oyster group (one of the most common cooking method of oysters) was selected to study the concentration changes of DEL in cooked oysters during digestion. Samples were taken every 30 min during the simulated digestion process for DEL analysis.

Cell culture and transport assay were done using the methodologies previously described by Xu et al. (38) with modification. Briefly, NCM460 cells were spread onto a Transwell polyester permeable membrane support at a density of 1.0 × 105 cells/cm². The cells were grown at 5% CO₂ and 37°C in a humidified atmosphere in DMEM medium (DMEM: serum:double antibody = 9:1:0.1). The medium was changed every other day to allow the cells to differentiate for at least 21 days. For toxicity assay, cells were incubated with PBS containing DEL (0, 5, 10, 17.5, 30 ng/mL) at 37°C for 24 h. Then, the fluorescence was measured at 450 nm using a Microplate Photometer (Multiskan FC, Thermo Fisher Scientific Inc., Shanghai, China).

The transport assays of DEL were done in a Transwell membrane. NCM460 cell monolayers were incubated with 0.5 mL phosphate-buffered saline buffer (PBS, containing 1 mmol/L Ca²⁺ and 0.5 mmol/L Mg²⁺ ions) containing DEL (0, 2, 10, 30 ng/mL) on the apical side and 1.5 mL PBS on the basolateral side of the monolayers for 2 h and culture medium collected from both sides for GC-ECD. The transepithelial electrical resistance (TEER) of the monolayer was measured to ensure its value was >400 Ω/cm².

Statistical analysis was done using the IBM Statistical Package for the Social Sciences, SPSS Statistics 26 (IBM Corp., Armonk, NY, USA). One-way ANOVA followed by Duncan's multiple range tests were used to determine the statistical significance. Spearmans' correlation (non-parametric bivariate correction) was established between nutritious substance and DEL levels. Differences were considered significant at p < 0.05.

DEL levels in different oyster tissues are shown in Figure 2. DEL toxicity ranged from 677 to 2,750 μg /Kg in different tissues of oysters. The gill sample had the highest levels of DEL. The lowest DEL concentrations were in the adductor muscle samples. The concentration of DEL in the mantle was higher than in the viscera. Statistically significant differences (p < 0.05) were observed in the concentration of DEL in the different tissues. The content of DEL in the whole oyster had no significant difference with the content in viscera.

In addition to DEL, the proximate composition (crude protein, fat, total carbohydrate, and ash) in various tissues were also measured (Figure 2). The highest protein content and the lowest fat content were found in adductor muscle. A statistical correlation of DEL levels and tissue proximate composition gave a positive (r = 0.729) and highly significant correlation with ash content. No significant correlations were observed between DEL and crude protein (r = −0.200, p = 0.475), fat (r = 0.318, p = 0.248) and carbohydrate (r = 0.325, p = 0.237).

The DEL GC spectrum from oyster meat is shown in Figure 3. There were 2 peaks with retention times of 37.5 min (DEL-1) and 38.2 min (DEL-2). However, there were significant differences between the peak areas of DEL-1 and DEL-2 in oyster meat with different processing methods, and the proportions of DEL-1 was 6.2% (raw), 24.6% (steamed), and 20.4% (roasted).

Total DEL of oyster meats with the various digestion phases (gastric initial and final and intestinal initial and final) with the different cooking methods are shown in Figure 4. There was a similar trend among the three processing methods. The content of DEL at the end of the gastric phase was higher than that at the beginning of the gastric phase, and the content of DEL at the end of the intestinal phase was higher than that at the beginning. The total amount of DEL in the digested liquids was always lower than that in the undigested oyster meat (p < 0.05). And the amount of DEL in the raw group was higher than that in the other two groups at each digestive stage. At the end of the simulated digestion, the content of DEL in the raw, steamed, and roasted samples were 2,300 ± 300, 2,020 ± 80, and 1,360 ± 40 ng, respectively.

High percentages of bioaccessible DEL were found with all the cooking methods using the in vitro model: 82 ± 10, 65 ± 6, and 51 ± 3% of the total DEL was released into the digestive juice with the raw, steam, and roasting, respectively (Figure 5). There was a significant difference between these cooking methods (p < 0.05) and a reduction of bioaccessibility was observed after cooking. No significant difference was found between the steamed and roasted samples, but the bioaccessibility of the roasted group was slightly lower than the steamed group.

The digestion using the different gastric pH vs. a pH constant group. A total of 2 mL HCl was added in the gastric phase. Two hundred microliter HCl was added every 15 min in the pH varied group, and 2 mL HCl was added directly at the beginning of the gastric phase in the pH constant group. As shown in Figure 6, a significant increase in DEL levels was seen in the pH varied group, whereas a significant decrease was observed in the pH constant group during the gastric digestive stage. At the end of the gastric phase, the concentration of DEL in the pH varied group was significantly higher than that in the pH constant group. The switch to intestinal digestion led to an immediate decrease in DEL concentration. At this stage, the concentration of DEL in the pH varied group was higher than in the pH constant group. Furthermore, there were significant differences in the bioaccessibility of DEL between the two groups during gastrointestinal digestion (p < 0.05).

As shown in Figure 7A, after treatment with 5, 10, 17.5, and 30 ng/mL DEL for 24 h, there was no significant difference in the cell viability between control, 5, 10, and 17.5 ng/mL, whereas a significant reduction of cell viability was found at 30 ng/mL (p < 0.05). It was assumed that no harm came to the cells used in the transport study because the transport time used was 2 h, much shorter than 24 h for DEL toxicity. Figure 7B shows the linear curve of cell viability with DEL concentration at 24 h. As the concentration of DEL increases, cell viability continues to decrease. The linear equation obtained was y = 102.1–0.69 x, and the Pearson correlation coefficient was −0.97. Transport of DEL across NCM460 cell monolayers is shown in Figure 7C. The concentration of DEL in the 2 h group was slightly higher than that in the 0.5 h group.

The degradation of DEL was not observed within the cell, as evidenced by no significant differences (p > 0.05) between the total amount of DEL toxin added and the sum of this toxin found in the apical and the basolateral sections after 2 h.

After a short exposure experiment, the content of DEL was the highest in the gills of oysters, followed by the mantle and viscera, and the lowest in the muscle. The digestive pathway of DEL in oysters is from gills to internal organs and then to muscles. The mantle is a membrane that wraps the internal organs and tissues. The high content of DEL in mantle may be due to the direct contact between the mantle and water, and the lipophilicity of DEL allowed it to be absorbed in the mantle. In addition, the concentration of DEL and the proximate composition in whole meat was basically the same as the average value in each organ. Whole oyster meat is normally consumed. According to the data in Figure 4, there was no significant difference in the total content of DEL in raw, steamed, and roasted oysters. However, the concentration of DEL in steamed and roasted oysters was higher than that in raw oysters because the steaming and roasting process will cause water loss. This was consistent with the studies by Hess et al. (39), who suggested that the concentration of azaspiracids (AZA) in cooked shellfish was 2-fold higher than the uncooked shellfish. And Wiech et al. (40) also show the similar result.

The occurrence and profiles of DEL in oysters may not only be related to the tissue and cooking method, but also related to the structural characteristics of DEL itself. DEL contains 3 chiral centers, indicating that it has 8 stereoisomers (41, 42). Corcellas et al. (43) showed that the cis isomers of pyrethroids were easier to enrich than trans isomers in organisms. But the degradation rate of trans isomers was faster than cis isomers (44). Usually, there were two peaks in the deltamethrin gas chromatogram (45). As shown in Figure 3, there was a higher accumulation of DEL-1 in steamed or roasted oyster compared to raw oyster, whereas DEL-2 was the predominant toxin accumulated. However, the structure of DEL-1 is not understood. DEL-1 and DEL-2 may be two different stereoisomers of DEL and the conversion of DEL-2 to DEL-1 may have occurred due to heating. Therefore, more research regarding enantioselective accumulation and enantiomeric toxicology is needed to establish which enantiomers have a greater health risk.

Compared to other pyrethroids, DEL has a higher toxicity, which makes it important to study its bioaccessibility. The in vitro digestion model indicated that the DEL was gradually released from oysters during digestion. The simulated digestion, showed a lower concentration of DEL with intestinal digestion than in the prior gastric digestion. This may be due to the dilution of DEL by intestinal digestive juice such as bile.

The results observed in the present study show that DEL bioaccessibility in steamed oysters was around 65%, and in roasted oysters was around 51%, which is significantly lower than in raw oysters (82%). These results were consistent with the previous observations that the bioaccessibility values of AZA were lower in mussels after steaming (46). The hydrophobic peptides become exposed with surface denaturation. Since AZA are lipophilic, a weak binding between AZA and a 45 KDa protein was observed during cooking. DEL is also lipophilic, and this binding force might explain the lower bioaccessibility after heating. Considering the high bioaccessibility, some studies have shown that the content of harmful substances in food can be reduced by washing and soaking (47). Before eating oysters, we can try to reduce the harm of harmful substances to human body by washing and soaking.

The DEL content decreased significantly after intestinal digestion, which might be explained two ways. One is that DEL was present in the biologically unavailable parts (materials not digested), and the other is that the presence of pancreatin and bile salts in the digestive juice of the intestine hydrolyzed or degraded DEL. Previous studies showed that some pyrethroids including DEL may undergo significant transformations in the intestinal fluid (48). Carboxyl ester lipase has an important role in lipid metabolism and is synthesized primarily in the pancreas (49). Crow et al. (50) also observed a correlation between carboxylesterase and pyrethroids, indicating trans-permethrin were effectively cleaved, while DEL and bioresmethrin were not metabolized. The degree of hydrolysis varied for different pyrethroids. Carboxylesterase may therefore hydrolyze pyrethroids during simulated digestion in vitro. The degree of hydrolysis may also be related to the concentration of carboxylesterase. And the toxicity of the hydrolysate may be higher or lower than that of its parent compound (17). Thus, the hydrolytic metabolites and their potential biological activity should be investigated in future research.

The difference between the in vitro simulated digestion model and the existing static in vitro model is mainly reflected in the pH of gastric phase. The importance of pH values with in vitro simulated digestion in the bioaccessibility has been previously suggested (51, 52). Gastric pH is not a constant parameter and its value changes continuously during digestion (37). The concentration of DEL in the constant pH group was lower than the variable pH group for most stages of digestion as was its bioaccessibility.

In vitro simulated digestion experiments are simpler and more convenient than in vivo experiments partly because only specific enzymes are used in vitro. There are many in vitro digestion models, and the bioaccessibility data obtained by different digestion models may be different (53). Other factors when studying the bioavailability of pyrethroids such as the diversity of small intestinal microbiome and microbial diversity, as well as other enzymes and biosurfactants in digestive juices should be considered (48).

To reduce the influence of the complex components in the simulated digestive juice on the cell viability during cell transport, a standard solution of DEL with a similar concentration instead of intestinal digestion chyme was used. The transmembrane transport efficiency of DEL by NCM460 cells were 35, 45, and 43% at 2, 10, and 30 ng/mL, respectively. After simulated digestion, the concentration of DEL was about 10 ng/mL. Using a trans-transport efficiency of 45%, the bioaccessibility of DEL after passing through small intestinal epithelial cells were 37% (raw), 29% (steamed), and 23% (roasted). Shellfish like oysters, most often consumed raw, are the most hazardous, while foods consumed soon after heat treatment have fewer hazards (25).

The bioaccessibility of DEL with simulated digestion should help provide further information for the risk assessment of pyrethroids. However, there are some differences between bioavailability and bioaccessibility. The NCM460 cells were was used to study bioaccessibility, but the relationship between it and bioavailability still needs further study.