Plastic production increased dramatically in the last couple of decades making microplastic (MP) a growing problem today. MPs are tiny ubiquitous plastic particles smaller than five millimeters (5 mm) in size and are composed of different polymer types. The presence of this anthropogenic pollutant in aquatic ecosystems concerns researchers worldwide since adverse effects on animal models are not fully understood. The aim of this study was to explore the long-term genotoxic and cytotoxic effects of microplastics on the rainbow trout (Oncorhynchus mykiss) larvae. Fish were exposed to polystyrene (PS), polyethylene terephthalate (PET) and polyethylene (PE) MPs. It is important to notice that environmentally relevant concentrations of MP were used.
Cytogenetic damage was assessed using erythrocyte nuclear abnormalities assay. The formation of micronuclei (MN), nuclear buds (NB), nuclear buds on filament (NBf), blebbed nuclei (BL) cells were assessed as genotoxicity endpoints, and 8-shaped nuclei, fragmented apoptotic (FA) and bi-nucleated (BN) cells as cytotoxicity endpoints. In addition, bean shaped nuclei, non-nucleated (EN) and vacuolated nuclei (VacNuc) cells were evaluated. Data analysis was performed using R [1] and RStudio [2] software. Correlations between variables were calculated separately in each treatment and control groups. In control group the highest correlation was between MN and BL (Pearson correlation coefficient $r = -0.43$), in control (with particles) group - between 8-shaped nuclei and EN (Pearson correlation coefficient $r = 0.72$), in PS group - between MN and 8-shaped nuclei (Pearson correlation coefficient $r = 0.67$), in PET group - between MN and NB (Pearson correlation coefficient $r = 0.57$) and in PE group - between MN and 8-shaped nuclei (Pearson correlation coefficient $r = 0.82$). The investigation of potential relationships between cytogenetic endpoints in rainbow trout using principal component analysis (PCA) yielded two principal components (PCs) which explained 43% of the total variance. Afterwards, PCA was done in each treatment group separately. Contributions of variables were different in different MP exposure groups which could suggest a possibility that MPs of different polymer types could induce nuclear and cellular damage by different mechanisms. Unfortunately, this area of study is not adequately researched. In this analysis, the Hopkins statistic was more than 0.5 ($H = 0.712$), indicating that the data tends to cluster. For this reason, k-means analysis was performed. Optimal number of clusters was calculated using elbow method (yielded 2 clusters), average silhouette method (yielded 2 clusters) and gap statistic method (with 500 bootstrap samples yielded 1 cluster). k-means clusters were compared with real treatment groups by performing cross-tabulation. Results of this analysis help to generate new hypotheses for future experiments, as further studies of the ecotoxicological impacts of MPs on fish (especially in early stages of development) are required.