Deciphering activated sludge microbial communities with large-scale metagenomics sequencing

Abstract

As the increasing throughput of sequencing platforms and decreasing of sequencing prices, next generation sequencing (NGS) is becoming more important in detangling the microbial composition, interaction and activity. In this study, activated sludge (AS) samples were collected and sent to perform 16S rRNA gene amplicon, metagenomics and metatranscriptomics sequencing with NGS platforms. This thesis aimed to: (1) study microbial community dynamics using large-scale 16S rRNA gene with different timescales, i.e. monthly samples over five years for one wastewater treatment plant (WWTP) , biweekly samples of two WWTPs over 12 months and daily samples over 13 months at one WWTP; (2) develop an absolute quantification method at cell level for metagenomics and metatranscriptomics data, and then apply the method to study the activities of taxa and functional genes in foaming AS; (3) establish a reference gene catalog (RGC) for AS from globally collected 226 deeply sequenced metagenomics data sets; (4) develop bioinformatics tools to fast screen a specific gene and to estimate the average rrn operon copy number (ARCN) of a metagenomics data set. The five years 16S rRNA gene in Shatin WWTP revealed high diversity and abundance variations of bulking and foaming bacteria (BFB). The dynamics of BFB and their correlated bacteria and environmental parameters were investigated. Moreover, BFB could be well predicted by environmental interaction network (EIN) module. Two WWTPs with significantly different influent composition and operational conditions were compared to temporal microbial community dynamics. The communities were discovered to be major governed by influent composition and operational conditions and temperature drove the temporal dynamics in each plant. Daily resolution data showed huge diversity and dynamics of bacteria were neglected in sparse sampling by simulation analyses. The autocorrelation effect in time series was demonstrated to be an important source of spurious correlation and could influence on network analysis. Moreover, an absolute quantification method based on the cell number estimation to normalize gene abundance and expression to cell level was developed with combined metagenome and metatranscriptome experimental design. The biosurfactant synthetic genes analysis and host tracking showed the diverse profiles of both the biosurfactant types and host bacteria, which indicated that sludge foaming was possibly induced by many bacteria. A ~ 39 M non-redundant RGC for AS was created for the first time. The catalog could be a useful reference for further studies on metagenomics and metatranscriptomics of AS. Finally, two bioinformatics tools were developed to quantify a specific gene (like ARGs) and to calculate the ARCN in metagenomic datasets. The ARGs screening online platform (ARGs-OAP) was to accelerate the identification and quantification of ARGs in the environment. The same approach could be applied to other specific genes, like ppk1 gene for PAOs and amoA gene for nitrifiers, by using the corresponding databases. Applying “AroC” to large collections of amplicon and metagenomics samples, it was observed that different environmental samples hold significantly different ARCNs.