Genomic lineages of Rhizobium etli revealed by the extent of nucleotide polymorphisms and low recombination

Abstract

Background: Most of the DNA variations found in bacterial species are in the form of single nucleotide polymorphisms (SNPs), but there is some debate regarding how much of this variation comes from mutation versus recombination. The nitrogen-fixing symbiotic bacteria Rhizobium etli is highly variable in both genomic structure and gene content. However, no previous report has provided a detailed genomic analysis of this variation at nucleotide level or the role of recombination in generating diversity in this bacterium. Here, we compared draft genomic sequences versus complete genomic sequences to obtain reliable measures of genetic diversity and then estimated the role of recombination in the generation of genomic diversity among Rhizobium etli.

Results: We identified high levels of DNA polymorphism in R. etli, and found that there was an average divergence of 4% to 6% among the tested strain pairs. DNA recombination events were estimated to affect 3% to 10% of the genomic sample analyzed. In most instances, the nucleotide diversity (π) was greater in DNA segments with recombinant events than in non-recombinant segments. However, this degree of recombination was not sufficiently large to disrupt the congruence of the phylogenetic trees, and further evaluation of recombination in strains quartets indicated that the recombination levels in this species are proportionally low.

Conclusion: Our data suggest that R. etli is a species composed of separated lineages with low homologous recombination among the strains. Horizontal gene transfer, particularly via the symbiotic plasmid characteristic of this species, seems to play an important role in diversity but the lineages maintain their evolutionary cohesiveness.

Figure 1

SNP assessment on lower coverage. Paired comparisons between E. coli strains and K12 strain as reference genome. We used two set of E. coli strains: Complete genomes obtained from Sanger reads (green box) and Sanger reads simulated from complete genomes with coverage approximate of 1X (yellow box). For each alignment, we determined the percentage of SNPs by gene fragment (SNP number/length of contig; Y axis) by our methodology. Boxes inside the graphic include the median values (middle line) and the first and third quartiles (lower and upper lines) of the distribution. Abscissa: E. coli strains.

Figure 2

Paired comparisons between R. etli strains. We performed four paired comparisons used our methodology: draft genomes of R. etli against CFN42 (gray boxes), draft genomes of R. etli against CIAT652 (blue boxes), CIAT652 against CFN42 (red boxes) and finally R. leguminosarum bv viciae 3841 against CFN42 (green boxes). For all comparisons the Y axis is the percentage of SNPs by gene fragment (SNP number/length of contig). Boxes inside the graphic include the median values (middle line) and the first and third quartiles (lower and upper lines) of the distribution. Abscissa: R. etli and R. leguminosarum bv viciae 3841 strains.

Figure 3

Average nucleotide variation. We calculated the Average nucleotide variation (middle lines of each graphic) from the median SNP percentages (dots indicates by dashes lines) for each aligned comparison (Y axis) of test strain versus the reference strains, CFN42 (blue) or CIAT652 (red). Average Confidence interval was adjusted (arrows with dashes lines) to the medians of the length distributions of the aligned fragments (genes). The medians SNPs that exceed the average confidence interval were outliers. Abscissa: BRASIL5, CIAT894, GR56, IE4771, KIM5 and 8C-3.

Figure 4

SNP distribution profiles. Alignments were performed on a total of 240 sequence segments available for all tested strains of Rhizobium etli. Each nucleotide position in the alignment is represented by a consensus. In instances where half of the strains had the same nucleotide and the other half a different nucleotide, the consensus was defined as the nucleotide present in R. etli CIAT652. Common segments were concatenated according to the gene order found in the CFN42 genome (chromosome and after plasmids), yielding 71,630 aligned base pairs. The numbers of nucleotides differing from the consensus are plotted as bars, across independent windows of 250 nucleotides. The black bars (running downwards) show SNPs present in a single strain; the gray areas indicate when the same SNP pattern was present in at least two strains at the same position within the alignment (patterns A and B); and the white bars indicate polymorphic sites where at least three alleles were present in at least two strains, again within the alignment. Segments showing significant recombination events are indicated by bars at the bottom of the plot, and with bars indicating the genomic location of segments with respect to CFN42 (chromosome, white; plasmids, black).

Figure 5

Genetic relatedness. Network joining network phylogeny inferred from 186 concatenated regions shared among strains of Rhizobium etli and Rhizobium leguminosarum bv viciae 3841 (see Methods). The tree is unrooted and has six internal branches, indicated by split numbers on each internal branch. The scale bar denotes the expected number of nucleotide substitutions per site.

Figure 6

Detection of recombination in quartets. Six groups of quartets of orthologous segments were built with the Mauve program; they included shared sequences from CFN42, CIAT652, Rhizobium leguminosarum bv viciae 3841, and each one of the six test strains (incomplete genomes). Test of recombination were performed for all the quartets as described in Methods section. The number (as well as the percentage) of recombinant segments predicted for each group of quartets is indicated above the quartet diagram. The total number of quartets analyzed in each group is indicated at the left side of the diagrams. The percentages of recombinant quartets between pairs of strains are shown inside the diagram and by dashed and continuous lines defined below the diagram. To facilitate searching each test strains o incomplete genome has its own color.

Figure 7

Genetic diversity in recombinant segments. A) For each homologous segment (quartet; regardless of evidence of recombination), we calculated the nucleotide diversity phi (Y axis; see Methods). The dots indicate the distribution means and the bars represent the 95% confidence intervals. Blue and red dots indicate recombinant and non-recombinant segments, respectively. Moreover, we determined the nucleotide diversity of the sequence regions shared across all of the tested strains of R. etli (green dot). Abscissa: BRASIL5, CIAT894, GR56, IE4771, KIM5, 8C-3 and shared regions. B) Magnification of the results from the 240 common sequence segments shared by all tested strains. The average percentage of SNPs was 12.89% per segment. The sphere sizes indicate the proportions of ts/tv for both recombinant segments (red circles) and non-recombinant segments (blue circles). The Y axis denotes the SNP percentage by gene segment, while the X axis shows the nucleotide diversity.