Perturbation metatranscriptomics for studying complex microbial communities

Studying the functional state of natural or engineered microbial communities presents substantial challenges due to both the complexities of field sampling, and, in the laboratory context, the inability of culture or reactor systems to maintain community composition ex situ over long periods. Here, we use whole community shotgun transcriptomics—metatranscriptomics—performed on freshly sourced samples, in combination with short-term laboratory perturbations, to demonstrate recapitulation of the functional state of a natural community in a controlled setting. We illustrate this approach by studying nitrogen transformation in wastewater treatment using freshly sourced anoxic sludge, in combination with systematic oxygen perturbation that switches physiological state of the community from denitrification activity to nitrification activity. Sampling every 10 minutes we collected and analysed 20 samples in total (13 in the denitrification phase, and 6 in nitrification phase) using shotgun whole community transcriptomics (RNA-Seq) and physicochemical analysis. Analysis of N compounds demonstrated that this system captures the characteristics of a functioning community performing nitrogen removal, and expression data showing key N compound related genes are detectable (described using KEGG Orthology ids at a whole community level). We observed 182 genes where up-regulated, and 156 down-regulated, following the onset of aerobic conditions (adjusted P<0.05). The mean factor change in expression in these genes was 1.34 (range: 1.12-2.65) for genes up-regulated upon oxygenation, and 1.43 (range: 1.10-4.43) for genes down-regulated upon oxygenation. Comparable changes in gene expression were also observed when data was mapped predicted genes (ORFs) from a metagenome assembly from the same community. Using KEGG pathways, up-regulated genes were statistically (adjP<0.05) enriched for genes associated with ribosome and down-regulated genes enriched for nitrogen metabolism (mostly denitrification related, including narG, napA, nirK, norB and nosZ), two-component systems and phenylalanine metabolism (PaaABCDEK genes in the aerobic phenylacetate catabolic pathway). We also sampled in situ from anoxic and aerobic source tanks in the field, and compared expression levels between anoxic and aerobic samples in each study: strongly down-regulated genes were preserved between both settings, and an overall good correlation between orthologous genes (Pearson r=0.4). We also used these data to annotate uncharacterized genes in the Ca. nitrospira defluvii genome: finding clear evidence for several previously unrecognized denitrification related genes, using a combination of expression profiles and protein domain data. We conclude that: 1) using freshly sourced sludge samples in combination with controlled laboratory manipulation over short study periods (1-2 hours) captures the key functional states of critical bioprocesses, as defined at both physicochemical and transcript levels; 2) specific pathways are associated with the transition from anoxic to aerobic conditions, and are observable at a whole community level and 3) these data provide a means of identifying unannotated genes in reference genomes that are likely to be associated with specific functional processes. More broadly, our approach permits the systematic functional investigation of complex communities and is readily extended to more specific targeted perturbations, in a variety of settings.