Molecular evolution and mosaicism of leptospiral outer membrane proteins involves horizontal DNA transfer

Abstract

Leptospires belong to a genus of parasitic bacterial spirochetes that have adapted to a broad range of mammalian hosts. Mechanisms of leptospiral molecular evolution were explored by sequence analysis of four genes shared by 38 strains belonging to the core group of pathogenic Leptospira species: L. interrogans, L. kirschneri, L. noguchii, L. borgpetersenii, L. santarosai, and L. weilii. The 16S rRNA and lipL32 genes were highly conserved, and the lipL41 and ompL1 genes were significantly more variable. Synonymous substitutions are distributed throughout the ompL1 gene, whereas nonsynonymous substitutions are clustered in four variable regions encoding surface loops. While phylogenetic trees for the 16S, lipL32, and lipL41 genes were relatively stable, 8 of 38 (20%) ompL1 sequences had mosaic compositions consistent with horizontal transfer of DNA between related bacterial species. A novel Bayesian multiple change point model was used to identify the most likely sites of recombination and to determine the phylogenetic relatedness of the segments of the mosaic ompL1 genes. Segments of the mosaic ompL1 genes encoding two of the surface-exposed loops were likely acquired by horizontal transfer from a peregrine allele of unknown ancestry. Identification of the most likely sites of recombination with the Bayesian multiple change point model, an approach which has not previously been applied to prokaryotic gene sequence analysis, serves as a model for future studies of recombination in molecular evolution of genes.

FIG. 1.

Comparison of DNA and amino acid (AA) sequence nonidentity among leptospiral genes. Sequences of the 16S, lipL32, lipL41, and ompL1 genes from 38 leptospiral strains were compared, revealing striking differences in the degree of DNA and amino acid sequence variability between genes. The 16S and lipL32 genes were less variable, and most lipL32 and lipL41 nucleotide mutations were synonymous. In contrast, significantly greater DNA and amino acid sequence variability was observed for the lipL41 and ompL1 genes. Differences in DNA and amino acid sequence variability were significant for all genes (two-tailed t test for paired samples, P < 0.0001). Error bars indicate standard deviations.

FIG. 2.

Localization of OmpL1 variable regions. Comparison of the OmpL1 amino acid sequences of 38 leptospiral strains reveals four variable regions corresponding to the four largest surface-exposed loops. (A) Histogram of OmpL1 amino acid sequence variability. Variable amino acids are clustered in four variable regions corresponding to the four largest surface-exposed loops of OmpL1. Variable region 1 (VR1) is located in the first surface-exposed loop (SEL1) and contains the highest number of variable amino acids. Variable region 2 (VR2) is located in the second surface-exposed loop (SEL2), variable region 3 (VR3) is located in the fourth surface-exposed loop (SEL4), and variable region 4 (VR4) is located in the fifth surface-exposed loop (SEL5). The height of each bar in the histogram indicates the degree of amino acid sequence variation from the consensus amino acid residue at that location. The OmpL1 amino acid sequence is numbered from the N-terminal amino acid of the mature protein. (B) Alignments of OmpL1 variable regions. Alignments of the four OmpL1 variable regions are shown by using the consensus amino acid sequences for the horizontally transferred peregrine allele (trans) and each of the six Leptospira species examined in this study: L. interrogans (inter), L. kirschneri (kirsch), L. noguchii (noguc), L. borgpetersenii (borgp), L. santarosai (santa), and L. weilii (weili). Locations of variable amino acids are indicated by dark boxes. The OmpL1 amino acid sequence is numbered from the N-terminal amino acid of the mature protein.

FIG. 3.

Phylogenetic trees for the 16S, lipL32, lipL41, and ompL1 genes. These unrooted phylogenetic trees summarize the posterior distribution of the evolutionary relationship among leptospiral strains inferred for the 16S, lipL32, lipL41, and ompL1 genes with MrBayes. Sequences from the same 38 strains were used in each of the four trees. Each strain is represented in each gene tree by a lowercase letter (i, L. interrogans; k, L. kirschneri; n, L. noguchii; s, L. santarosai; b, L. borgpetersenii; w, L. weilii) followed by the 4-digit strain code (detailed descriptions of strains used in this study are given in Table 1). Sequences suggestive of horizontal transfer of DNA (bHB10* in the lipL41 tree and the lowest branch of the ompL1 tree) are indicated by asterisks. Numbers provided above major branches are branch lengths and are reported in the expected number of nucleotide substitutions per site.

FIG. 4.

Most probable breakpoint locations along the ompL1 gene. The posterior distributions of breakpoint locations are plotted for 8 mosaic ompL1 genes. Peaks on the distribution graphs indicate the most probable position of recombination. Three mosaic patterns are shown. The most common mosaic pattern has four breakpoints (shown by shaded bands covering their 95% Bayesian credible intervals). A second pattern, shown for nAS10, has two breakpoints (shown by dark shaded bands covering their 95% Bayesian credible intervals). The third pattern, occurring in kCA02, with three breakpoints, and is a hybrid of the first two patterns. Each strain is represented by a lowercase letter (i, L. interrogans; k, L. kirschneri; n, L. noguchii; b, L. borgpetersenii) followed by the 4-digit strain code (detailed descriptions of strains used in this study are given in Table 1).

FIG. 5.

Mosaic ompL1 gene patterns. Three distinct mosaic ompL1 gene patterns were revealed by the BMCP model. The posterior probabilities of various lineages (red, L. interrogans; purple, L. kirschneri; blue, L. noguchii; green, the peregrine allele) giving rise to a region are plotted for the ompL1 genes from L. interrogans serovar Lai (A), L. kirschneri strain CA02 (B), and L. noguchii strain AS10 (C). The peregrine allele occurs in all three mosaic patterns between nucleotides 48 and 226 and between nucleotides 508 and 595.