In silico interactions and deep neural network modeling for toxicity profile of methyl methanesulfonate.

Abstract

In this study, the toxicity induced by the alkylating agent methyl methanesulfonate (MMS) in Allium cepa L. was investigated. For this aim, bulbs were divided into 4 groups as control and application (100, 500 and 4000µM MMS) and germinated for 72h at 22-24°C. At the end of the germination period root tips were collected and made ready for analysis by applying traditional preparation methods. Germination, root elongation, weight, mitotic index (MI) values, micronucleus (MN) and chromosomal abnormality (CAs) numbers, malondialdehyde (MDA) levels, superoxide dismutase (SOD) and catalase (CAT) activities and anatomical structures of bulbs were used as indicators to determine toxicity. Moreover the extent of DNA fragmentation induced by MMS was determined by comet assay. To confirm the DNA fragmentation induced by MMS, the DNA-MMS interaction was examined with molecular docking. Correlation and principal component analyses (PCA) were performed to examine the relationship between all parameters and understand the underlying structure and relationships among these parameters. In the present study, a deep neural network (DNN) with two hidden layers implemented in Matlab has been developed for the comparison of the estimated data with the real data. The effect of MDA levels, SOD and CAT activities at 4 different endpoints resulting from administration of various concentrations of MMS, including MN, MI, CAs and DNA damage, was attempted to be estimated by DNN model. It is assumed that the predicted results are in close agreement with the actual data. The effectiveness of the model was evaluated using 4 different metrics, MAE, MAPE, RMSE and R2, which together show that the model performs commendably. As a result, the highest germination, root elongation, weight gain and MI were measured in the control group. MMS application caused a decrease in all physiological parameters and an increase in cytogenetic (except MI) and biochemical parameters. MMS application caused an increase in antioxidant enzyme levels (SOD and CAT) up to a concentration of 500µM and a decrease at 4000µM. MMS application induced different types of CAs and anatomical damages in root meristem cells. The results of the comet assay showed that the severity of DNA fragmentation increased with increasing MMS concentration. Molecular docking analysis showed a strong DNA-MMS interaction. The results of correlation and PCA revealed significant positive and negative interactions between the studied parameters and confirmed the interactions of these parameters with MMS. It has been shown that the DNN model developed in this study is a valuable resource for predicting genotoxicity due to oxidative stress and lipid peroxidation. In addition, this model has the potential to help evaluate the genotoxicity status of various chemical compounds. At the end of the study, it was concluded that MMS strongly supports a versatile toxicity in plant cells and the selected parameters are suitable indicators for determining this toxicity.