Propagation of viral genomes by replicating ammonia-oxidising archaea during soil nitrification

Abstract

Ammonia-oxidising archaea (AOA) are a ubiquitous component of microbial communities and dominate the first stage of nitrification in some soils. While we are beginning to understand soil virus dynamics, we have no knowledge of the composition or activity of those infecting nitrifiers or their potential to influence processes. This study aimed to characterise viruses having infected autotrophic AOA in two nitrifying soils of contrasting pH by following transfer of assimilated CO₂-derived ¹³C from host to virus via DNA stable-isotope probing and metagenomic analysis. Incorporation of ¹³C into low GC mol% AOA and virus genomes increased DNA buoyant density in CsCl gradients but resulted in co-migration with dominant non-enriched high GC mol% genomes, reducing sequencing depth and contig assembly. We therefore developed a hybrid approach where AOA and virus genomes were assembled from low buoyant density DNA with subsequent mapping of ¹³C isotopically enriched high buoyant density DNA reads to identify activity of AOA. Metagenome-assembled genomes were different between the two soils and represented a broad diversity of active populations. Sixty-four AOA-infecting viral operational taxonomic units (vOTUs) were identified with no clear relatedness to previously characterised prokaryote viruses. These vOTUs were also distinct between soils, with 42% enriched in ¹³C derived from hosts. The majority were predicted as capable of lysogeny and auxiliary metabolic genes included an AOA-specific multicopper oxidase suggesting infection may augment copper uptake essential for central metabolic functioning. These findings indicate virus infection of AOA may be a frequent process during nitrification with potential to influence host physiology and activity.

Fig. 1. Contigs and MAGs from low GC mol% DNA and in situ growth of AOA in pH 4.5 and 7.5 soils determined from hybrid analysis of ¹²C- and ¹³C-enriched DNA.

A Distribution of total prokaryotic and AOA genomes in CsCl gradients determined from the relative abundance of 16S rRNA genes and amoA genes, respectively. Example profiles are shown for pH 4.5 soil only (see Supplementary Fig. 2 for pH 7.5 profiles). Vertical error bars are the standard error of the mean relative abundance and horizontal bars (mostly smaller than the symbol size) the standard error of the mean buoyant density of individual fractions from three independent CsCl gradients, each representing an individual microcosm. Genomic DNA in fractions highlighted in blue or pink areas were pooled for each replicate microcosm for metagenomic sequencing. B Distribution of assembled contigs (≥5 kb) from metagenomic libraries prepared from LBD or HBD DNA in sequence coverage vs. GC mol% plots. Contigs binned into ten medium- and high-quality nitrifier MAGs and predicted viruses of nitrifiers (≥10 kb) are highlighted, with numbers in parenthesis beside each MAG identifier giving estimated completeness and contamination (%). C Mean and standard error of the relative proportion of metagenomic reads from ¹²C- and ¹³C-enriched HBD DNA mapped onto contigs of nine AOA and one Nitrobacter MAGs. A significantly greater relative proportion of reads in ¹³C-derived libraries are indicated with * (p < 0.05, two-sample Student’s t-test or Welsch’s t-test when variances were not homogenous). GTDB genus [16] and amoA gene lineage-affiliation [17] are given.

Fig. 2. Identification and determination of activity of viruses infecting AOA in pH 4.5 and 7.5 soils using a hybrid analysis of ¹²C- and ¹³C-enriched DNA.

A Heatmap showing the relative abundance (reads per kb, ln transformed) of 64 vOTUs identified in LBD DNA libraries from pH 4.5 and 7.5 soil (≥1× coverage, ≥75% contig breadth). To identify activity, reads from ¹²C- and ¹³C-enriched HBD DNA were mapped onto LBD-derived contigs with a significantly greater relative proportion of reads in ¹³C-derived libraries indicated with * (p < 0.05, two-sample Student’s t-test or Welsch’s t-test when variances were not homogenous). B Proteomic tree showing genome-wide sequence similarities between AOA viral contigs and 1,981 curated virus genomes. Values at dotted circles represent a distance metric based on normalized tBLASTx scores plotted on a log scale. C Maximum likelihood phylogenetic analysis of derived large sub-unit terminase (TerL) protein sequences (336 unambiguously aligned positions, LG substitution model) identified in this and other metagenomic studies or in proviruses of cultivated AOA. NCBI accession numbers are given in parenthesis. Circles at nodes represent percentage bootstrap support from 1000 replicates, scale bar denotes an estimated 0.05 changes per position and colour-coding describes the comparative relative abundance in each soil. D Genetic map of AOA-infecting virus-derived contig 053596 containing a type 1 multicopper oxidase (MCO1) gene. Maximum likelihood phylogenetic tree describes placement of MCO1 within the AOA MCO/NirK family of cultivated AOA following designations of Kerou et al. [22] (206 unambiguously aligned positions, LG substitution model, invariant and gamma distributed sites). NCBI accession numbers are given in parenthesis. Circles at nodes represent percentage bootstrap support from 1000 replicates and the scale bar denotes an estimated 0.05 changes per position.