Detailed investigation of the microbial community in foaming activated sludge reveals novel foam formers

Abstract

Foaming of activated sludge (AS) causes adverse impacts on wastewater treatment operation and hygiene. In this study, we investigated the microbial communities of foam, foaming AS and non-foaming AS in a sewage treatment plant via deep-sequencing of the taxonomic marker genes 16S rRNA and mycobacterial rpoB and a metagenomic approach. In addition to Actinobacteria, many genera (e.g., Clostridium XI, Arcobacter, Flavobacterium) were more abundant in the foam than in the AS. On the other hand, deep-sequencing of rpoB did not detect any obligate pathogenic mycobacteria in the foam. We found that unknown factors other than the abundance of Gordonia sp. could determine the foaming process, because abundance of the same species was stable before and after a foaming event over six months. More interestingly, although the dominant Gordonia foam former was the closest with G. amarae, it was identified as an undescribed Gordonia species by referring to the 16S rRNA gene, gyrB and, most convincingly, the reconstructed draft genome from metagenomic reads. Our results, based on metagenomics and deep sequencing, reveal that foams are derived from diverse taxa, which expands previous understanding and provides new insight into the underlying complications of the foaming phenomenon in AS.

Figure 1. Abundance and divergence of the major genera in foam samples and corresponding AS samples.

The taxonomic classification was performed by RDP Classifier at confidence thresholds of 80% (for 454 pyrotags over 300 bps) and 50% (for 16S rDNA itags, 130–190 bp). The heatmap lists the top 32 genera over 0.5% abundance in at least one sample. The column above the heatmap shows percentages of tags assigned into a genus. The bar chart on the left exhibits the divergence of distribution for each genus between the foam samples and the corresponding foaming AS samples (using the logarithmic ratio between them). Only the ratios of those genera over 0.5% in foam or the corresponding AS are shown; for the others, the chart is marked by an asterisk. ND means “not detected”.

Figure 2. Variations of the 39 OTUs in AS before, during and after the foaming events in March 2010.

These OTUs had more than 0.5% abundance in the foaming AS. A black block in the heatmap means less than 0.1% or not detected. The list of the OTUs (from top to bottom) was determined by the ratio of abundance in the foaming AS and the highest abundance of the non-foaming AS samples. The taxonomic names are the lowest classified level in RDP Classified at a confidence threshold of 80%, by one representative sequence. The number after the taxonomic name indicates that more than one OTU could be assigned to the taxa and the number was sorted by abundance in the foaming AS.

Figure 3. Mycobacterial population in two foam samples by deep-sequencing of the rpoB gene amplicons.

Only the validated sequences that had more than 96% similarity to a reference were identified as the reference related mycobacteria.

Figure 4. Deposition of the uncultured Gordonia clone from the foam former in a phylogenetic tree of Gordonia, based on the near full-length 16S rRNA gene (approximately 1,200 bp).

The neighbor-joining method and Jukes-Cantor model were applied with 1,000 bootstrap replications. The sequences from Mycobacterium tuberculosis and Rhodococcus rhodochrous DSM 43241 were set as out-groups during the tree construction in MEGA 5. Only the bootstrap values of more than 50% were shown. The scale bar represents 2% divergence.

Figure 5. Genome bin of the uncultured Gordonia sp enriched in foam.

The raw reads were assembled into contigs and only contigs over 1 kb were involved in the genomic binning. Only contigs over 2 kb were shown in this figure. Two dimensional coverages were referred to separate the contigs from different genomes. The marked cluster was the genomic bin of the uncultured Gordonia sp. with average coverages of 9 and 58 in the activated sludge and the foam datasets, respectively. It contained 83 essential single copy genes as highlighted in colored circles.

Figure 6. Amino acid sequences similarities between the reconstructed draft genome and 22 Gordonia genomes and an out-group member, Mycobacterium smegmatis (A) and intra-species and inter-species Gordonia strains (B).

The collection of the ORFs of the reconstructed draft genome was used as the database for BLASTx by the genes of all the Gordonia genomes in Group A. In Group B, the ORFs genomes used as the databases were G. terrae NBRC100016, G. polyisoprenivorans VH2, G. namibiensis NBRC108229 and G. amarae NBRC15530, respectively. The ORFs from G. terrae C-6, G. polyisoprenivorans NBRC16320, G. rubripertincta NBRC 101908 and the reconstructed draft genome were queries to perform BLASTp against the corresponding databases, respectively. Therefore, the denominators here are the total ORFs in the queries. The percentages in the pie charts are the average amino acid identities for the shared ORFs.