Genetic exchange across a species boundary in the archaeal genus ferroplasma

Abstract

Speciation as the result of barriers to genetic exchange is the foundation for the general biological species concept. However, the relevance of genetic exchange for defining microbial species is uncertain. In fact, the extent to which microbial populations comprise discrete clusters of evolutionarily related organisms is generally unclear. Metagenomic data from an acidophilic microbial community enabled a genomewide, comprehensive investigation of variation in individuals from two coexisting natural archaeal populations. Individuals are clustered into species-like groups in which cohesion appears to be maintained by homologous recombination. We quantified the dependence of recombination frequency on sequence similarity genomewide and found a decline in recombination with increasing evolutionary distance. Both inter- and intralineage recombination frequencies have a log-linear dependence on sequence divergence. In the declining phase of interspecies genetic exchange, recombination events cluster near the origin of replication and are localized by tRNAs and short regions of unusually high sequence similarity. The breakdown of genetic exchange with increasing sequence divergence could contribute to, or explain, the establishment and preservation of the observed population clusters in a manner consistent with the biological species concept.

Figure 1.—

Diagram illustrating the representation of single-nucleotide polymorphisms (SNPs) that distinguish individual sequence variants within the community (white boxes signify sequencing reads, lines link mate pairs). Reads are aligned to a reference sequence that is either an isolate genome sequence or a composite sequence derived from a community genomic data set (cream-colored box). (A) Over the region shown one read has 7 SNPs relative to the reference sequence and the other is identical to it. Colored bars highlight the substituted base at each location; cyan, pink, magenta, and yellow represent substitutions of the bases A, C, T, and G, respectively. (B and C) The assembly is viewed at lower magnification without individual nucleotides labeled. The colors indicating SNPs are now small tick marks. In the region shown, sequencing reads can be easily clustered on the basis of SNP frequency. Backgrounds colored in shades of brown (similar to F. acidarmanus) and blue (similar to Ferroplasma type II scaffolds) indicate these groupings. However, within each cluster there is some fine-scale variation. In B the reads in the brown cluster are grouped into two sequence types (one with SNPs and one without SNPs). One read, outlined in red, represents a recombination of the two types. The recombinant read can be compared to the reconstructed parent sequences, as shown in C. Thus, the divergence between the parent sequences can be calculated for that recombination event.

Figure 2.—

Images showing the clear separation of the Ferroplasma type I (brown) and Ferroplasma type II (blue) genomes. (A) Reads from the community genomic data set are aligned (using BLASTN) to 2.7 kb of the isolate genome of F. acidarmanus and displayed as described in Figure 1. Reads are grouped (indicated by background shades of blue and brown) on the basis of conserved SNP patterns. Genes encoded on this fragment (bars on top from left to right) are a hypothetical protein, a putative dihydroxy-acid dehydratase, and a putative acetolactate synthase (large subunit). (B) The genomewide distribution of read sequence alignments against the F. acidarmanus genome. Every 3000 bases, all overlapping reads were assigned to one or the other Ferroplasma type and the divergence of each read from F. acidarmanus was calculated. At each point, for each type, the 5, 25, 33, 66, 75, and 95% divergence quantiles were calculated. These points were then connected to map the distributions of sequence identity genomewide. Lightly shaded regions span the 5–95% quantiles; darker regions, the 25–75% quantiles; and the darkest region, the 33–66% quantiles.

Figure 3.—

Interspecies recombination illustrated (as described in Figure 1) using the Ferroplasma type II composite genome as the reference. (A) In the region displayed, most of the reads from Ferroplasma type I (bottom cluster) share their sequence type with one of the Ferroplasma type II strains (top cluster). In A, all SNPs are displayed as gray ticks instead of using coded colors, as in Figure 1. (B) A more detailed view of the right side of A. A shaded vertical bar in B indicates a small region (∼80 bp) showing unusually high sequence similarity between the Ferroplasma types.

Figure 4.—

Intraspecies recombination in (A) Ferroplasma type I and (B) Ferroplasma type II. Figures are rendered as described in Figure 1. F. acidarmanus is used as a reference sequence in A and Ferroplasma type II composite sequence in B. Reads are grouped (indicated by background shades of blue and brown) into strains on the basis of conserved SNP patterns. Red outlines indicate reads containing inferred recombination points and black rectangles indicate apparent recombination end points within reads. Mate pairs assigned to different strain types are connected by a diagonal red line to highlight recombination events between strain types.

Figure 5.—

Observed recombination rates plotted as a function of sequence divergence. Log-linear lines were fitted to data for Ferroplasma type I (A), Ferroplasma type II (B), and interspecies (C) using maximum-likelihood estimation. Red lines were plotted from 95% confidence values of parameters. Errors, vertical blue bars, are estimated from maximum-likelihood optimization assuming the same normal error distributions for all recombinant clone counts (before normalization). Values for divergences <3% (green circles in A and B) were excluded from estimation due to difficulty in identifying recombinations. (D) All three data sets superimposed with the average value for interspecies recombination.

Figure 6.—

Circular diagram (origin of replication at the top) showing a genomewide comparison of Ferroplasma type II to F. acidarmanus. The five rings of data are, progressing inward, (1) Ferroplasma type I genome showing predicted gene sequence locations color coded according to functional category (see supplemental Table S2 at http://www.genetics.org/supplemental/), (2) orthologs between Ferroplasma types I and II colored by percentage of identity (from red for 100% identity through the spectrum of colors to pale blue), (3) reconstructed Ferroplasma type II genome (colored from blue to purple as distance around the genome) highlighting rearrangements and syntenous regions, (4) tRNAs in F. acidarmanus (gray), and (5) interpopulation recombination points (red).