Assembly of viral metagenomes from yellowstone hot springs

Abstract

Thermophilic viruses were reported decades ago; however, knowledge of their diversity, biology, and ecological impact is limited. Previous research on thermophilic viruses focused on cultivated strains. This study examined metagenomic profiles of viruses directly isolated from two mildly alkaline hot springs, Bear Paw (74 degrees C) and Octopus (93 degrees C). Using a new method for constructing libraries from picograms of DNA, nearly 30 Mb of viral DNA sequence was determined. In contrast to previous studies, sequences were assembled at 50% and 95% identity, creating composite contigs up to 35 kb and facilitating analysis of the inherent heterogeneity in the populations. Lowering the assembly identity reduced the estimated number of viral types from 1,440 and 1,310 to 548 and 283, respectively. Surprisingly, the diversity of viral species in these springs approaches that in moderate-temperature environments. While most known thermophilic viruses have a chronic, nonlytic infection lifestyle, analysis of coding sequences suggests lytic viruses are more common in geothermal environments than previously thought. The 50% assembly included one contig with high similarity and perfect synteny to nine genes from Pyrobaculum spherical virus (PSV). In fact, nearly all the genes of the 28-kb genome of PSV have apparent homologs in the metagenomes. Similarities to thermoacidophilic viruses isolated on other continents were limited to specific open reading frames but were equally strong. Nearly 25% of the reads showed significant similarity between the hot springs, suggesting a common subterranean source. To our knowledge, this is the first application of metagenomics to viruses of geothermal origin.

FIG. 1.

Transmission electron microscopy images of virus-like particles directly isolated from YNP hot springs. Images from Bear Paw (A and B) and Octopus (C and D) hot springs are shown. The bar in each panel is 200 nm.

FIG. 2.

Sensitivity of linker-facilitated anonymous-DNA amplification. Decreasing amounts of lambda DNA (50 ng, 5 ng, 500 pg, 50 pg, and 5 pg, as indicated) were sheared, end repaired, linker ligated, and size selected on an agarose gel. The resulting material was amplified using Vent DNA polymerase and a single oligonucleotide complementary to the linker sequences. One-tenth of the reaction mixture was resolved by agarose gel electrophoresis as shown. As negative controls, no input DNA (lane 0) or only DNA-free gel slices from otherwise-identical reactions were similarly processed. DNA molecular weight markers (mkr) are indicated.

FIG. 3.

Genes and gene order are highly conserved between a cultured crenarchaeal virus and a consensus contig from the Bear Paw library. Contig 372 (5,492 bp; 71 reads) was assembled at ≥50% identity from the Bear Paw library. ORFs identified by the GeneMark algorithm were compared by BLASTp to proteins in GenBank. Similarities to PSV proteins are shown with percent coding identity. The gene names are based on the annotation in GenBank and are named in order of their locations on the viral chromosome. The directions of transcription are indicated by the arrows.

FIG. 4.

Alignment of nucleotide polymorphisms with coding sequences in a 16.5-kb consensus contig from the Octopus hot spring. Contig 722 was assembled at ≥50% identity from the Octopus library. Sequence coverage is shown at the top, with each line representing a separate read. SNPs per 10 base pairs were normalized to the number of reads covering the respective nucleotide (middle) and are aligned with predicted ORFs from the consensus sequence in the contig and the gene name with the strongest BLASTx similarity (bottom). The directions of transcription are shown by the arrows. Similarities to known genes were identified by BLASTp.

FIG. 5.

Broad classification of viral metagenomic contigs based on tBLASTx similarities. Contigs assembled at 95% identity from Bear Paw and Octopus reads (A and B, respectively) were compared to sequences in GenBank to infer phylogeny. Shown are frequencies of contigs with no significant sequence similarity in GenBank (E < 0.001) and those with sequence similarity to Bacteria, Archaea, Eukarya, and their respective viruses.

FIG. 6.

Alignment of Octopus and Bear Paw viral metagenomic library contigs with six cultured virus genomes. Contigs assembled at >95% identity from the viral metagenomic libraries were compared by tBLASTx to the genomes of PSV, SIRV1, ARV, ATV, STSV, and YS40. Each bar represents the alignment of a unique metagenomic sequence to the indicated location on the cultivated viral genome, shown on the horizontal axis. Percent coding sequence identities are shown on the vertical axis. The threshold for inclusion of a contig is an E value of <10⁻³. The red bars indicate Bear Paw alignments; the blue bars indicate Octopus alignments. Also shown are the known or predicted functions of the conserved coding sequences (rep, replication related; vir, virion component; gt, glycosyltransferase; tnp, transposase; cp, coat protein; dam, adenine DNA methylase; ts, thymidylate synthase; dut, dUTPase; dcm, cytosine DNA methylase; hel, helicase; rec, recombinase; rnr, ribonucleotide reductase (, , , , , , , , , ; D. Prangishvili, personal communication).