Metagenomic de novo assembly of an aquatic representative of the verrucomicrobial class Spartobacteria

Abstract

The verrucomicrobial subdivision 2 class Spartobacteria is one of the most abundant bacterial lineages in soil and has recently also been found to be ubiquitous in aquatic environments. A 16S rRNA gene study from samples spanning the entire salinity range of the Baltic Sea indicated that, in the pelagic brackish water, a phylotype of the Spartobacteria is one of the dominating bacteria during summer. Phylogenetic analyses of related 16S rRNA genes indicate that a purely aquatic lineage within the Spartobacteria exists. Since no aquatic representative from the Spartobacteria has been cultured or sequenced, the metabolic capacity and ecological role of this lineage are yet unknown. In this study, we reconstructed the genome and metabolic potential of the abundant Baltic Sea Spartobacteria phylotype by metagenomics. Binning of genome fragments by nucleotide composition and a self-organizing map recovered the near-complete genome of the organism, the gene content of which suggests an aerobic heterotrophic metabolism. Notably, we found 23 glycoside hydrolases that likely allow the use of a variety of carbohydrates, like cellulose, mannan, xylan, chitin, and starch, as carbon sources. In addition, a complete pathway for sulfate utilization was found, indicating catabolic processing of sulfated polysaccharides, commonly found in aquatic phytoplankton. The high frequency of glycoside hydrolase genes implies an important role of this organism in the aquatic carbon cycle. Spatiotemporal data of the phylotype's distribution within the Baltic Sea indicate a connection to Cyanobacteria that may be the main source of the polysaccharide substrates.

FIG 1

Emerging self-organizing map of the metagenome contigs. Pixels are colored according the taxonomic annotation of the contig(s) that occupies the pixel. Background color represents the distance in data space between the pixels in the neighborhood; hence the white ridges represent borders between regions of highly dissimilar tetranucleotide frequency distributions.

FIG 2

Maximum likelihood tree based on a concatenated alignment of 31 conserved genes of “Spartobacteria baltica” and representative genomes of Verrucomicrobia (10 representatives), Chlamydiae (5 representatives), Planctomycetes (8 representatives), and Spirochaetes/Actinobacteria (21 representatives). The tree was rooted using the Spirochaetes/Actinobacteria as an outgroup. All groups except the Verrucomicrobia have been grouped into wedges for clarity. Dots indicate bootstrap values of >98%.

FIG 3

Phylogenetic tree of nonredundant sequences of >1,200 bp in the Spartobacteria class obtained from the Silva database 111 SSU Ref NR. “Candidatus Methylacidiphilum infernorum” was used to root the tree. The tree was calculated using the RAxML algorithm with rapid bootstrap analysis (1,000 bootstraps). Only nodes supported by high bootstrap values are marked (filled circles, >95%). The origins of the sequences are indicated by the accession number, the isolation source, and the length of the sequence in bp.

FIG 4

Analyses of “Spartobacteria baltica” GH5 sequences. (A) Modular structure of the GH5 protein sequence (gene id 2119805716 in Table S3). (B) A maximum likelihood tree of selected bacterial GH5 catalytic module sequences, including “Spartobacteria baltica” (gene id 2119806690 in Table S3). (C) A maximum likelihood tree of the subfamily GH5_46, including “Spartobacteria baltica” (gene id 2119805716 in Table S3), and selected bacterial sequences from related GH5 subfamilies. The phylogenetic analysis was restricted to the catalytic module.