Development of bioinformatic tools and exploration of antibiotic resistance genes and mobile genetic elements in the environment

Abstract

To investigate the environmental prevalence of antibiotic resistance genes (ARGs) and interacted mobile genetic elements (MGEs) through high-throughput sequencing technology, bioinformatic analysis and data mining are the primary research object of the thesis. Specifically, a user-friendly tool was developed in this thesis for accelerated ARGs classification and quantification. First, the preliminary integrated analysis pipeline was updated into ARGs-OAP 2.0 with an expanded database SARG 2.0 and a novel module based on Hidden Markov Models to clearly enhance the detection coverage of ARGs and supplement the ability to discover novel ARGs. In the further update into ARGs-OAP 3.0, visualization packages were added for efficient interpretation of ARG profiles. And the database SARG 3.0 was kept updating with public data and was indexed to support online searching. In addition, the temporal variation of resistome profile was characterized using archived nine-year monthly sampling activated sludge (AS) in a wastewater treatment plant (WWTP). It was demonstrated that both the abundance and structure of the resistome changed significantly every two to three years, implying a successive selection of resistome in the AS system over the study period. Annual variation of resistome was explained via structural equation modeling (SEM), which deciphered the structural linkages of determining factors such as the operational parameters, microbial community composition and horizontal gene transfer (HGT). Then, further analysis on risk assessment of ARGs in WWTPs was conducted in a comprehensive way based on both temporal and spatial sampling influent, effluent and AS from three WWTPs in HK. Prevalence information of ARGs and MGEs was compared with North America, South America, Europe, and Asia. Results demonstrated HK WWTPs were featured high ARG removal efficiency of 2.34 to 2.43 log reduction rate, and effluents were ranked at moderate levels of Level 2 and Level 3 according to the risk prioritizing scheme based on total ARG abundance. Furthermore, to compare among a broad scale of eco-systems in ARG pollution level, 767 publicly available metagenomes from 13 eco-systems were collected, covering industrial, urban, agricultural and natural environments and spanning five continents. The resulting ARG profiles revealed that the widely distributed ARGs were unevenly proliferated in various eco-systems. Further analyses to decipher the distinction and connection across eco-systems were facilitated by machine learning tools. Finally, this thesis identified the HGT mechanism of ARGs on integrative and conjugative elements (ICE) across taxonomy lineages by analyzing 199,679 genomes and then summarized 27 important ICE families and 21 widely distributed ARG types carried by ICEs. A well-developed tool to classify ICEs and ARGs simultaneously from bacterial genomes, termed ICEcream, was developed. With this tool, nine exclusive co-occurrence patterns between ICE families and ARG subtypes was found. More importantly, over 50% of important ICE families, including Tn916, Tn5801 and ICESa2603, carried at least one ARG and a higher possibility of carrying multiple ARGs simultaneously. Active transfer across taxonomy barriers over class levels was observed in diverse ICE families, and Tn916 and Tn5801 were found even transfer across phyla, highlighting the underestimated roles of ICEs in ARG dissemination and evolution in the environment.