Integrating Multiple Bacterial Phenotypes and Bayesian Network for Analyzing Health Risks of Pathogens in Plastisphere

Hong-Zhe Li,Wen-Jing Li,Zi-Jian Wang,Qing-Lin Chen,Mia Kristine Staal Jensen,Min Qiao,Li Cui

DOI: https://doi.org/10.1021/acs.analchem.4c01433

2024-07-16

Abstract:Plastic pollution represents a critical threat to soil ecosystems and even humans, as plastics can serve as a habitat for breeding and refuging pathogenic microorganisms against stresses. However, evaluating the health risk of plastispheres is difficult due to the lack of risk factors and quantification model. Here, DNA sequencing, single-cell Raman-D2O labeling, and transformation assay were used to quantify key risk factors of plastisphere, including pathogen abundance, phenotypic resistance to various stresses (antibiotic and pesticide), and ability to acquire antibiotic resistance genes. A Bayesian network model was newly introduced to integrate these three factors and infer their causal relationships. Using this model, the risk of pathogen in the plastisphere is found to be nearly 3 magnitudes higher than that in free-living state. Furthermore, this model exhibits robustness for risk prediction, even in the absence of one factor. Our framework offers a novel and practical approach to assessing the health risk of plastispheres, contributing to the management of plastic-related threats to human health.