Evolutionary dynamics of insertion sequences in Helicobacter pylori

Abstract

Prokaryotic insertion sequence (IS) elements behave like parasites in terms of their ability to invade and proliferate in microbial gene pools and like symbionts when they coevolve with their bacterial hosts. Here we investigated the evolutionary history of IS605 and IS607 of Helicobacter pylori, a genetically diverse gastric pathogen. These elements contain unrelated transposase genes (orfA) and also a homolog of the Salmonella virulence gene gipA (orfB). A total of 488 East Asian, Indian, Peruvian, and Spanish isolates were screened, and 18 and 14% of them harbored IS605 and IS607, respectively. IS605 nucleotide sequence analysis (n = 42) revealed geographic subdivisions similar to those of H. pylori; the geographic subdivision was blurred, however, due in part to homologous recombination, as indicated by split decomposition and homoplasy tests (homoplasy ratio, 0.56). In contrast, the IS607 populations (n = 44) showed strong geographic subdivisions with less homologous recombination (homoplasy ratio, 0.2). Diversifying selection (ratio of nonsynonymous change to synonymous change, >>1) was evident in approximately 15% of the IS605 orfA codons analyzed but not in the IS607 orfA codons. Diversifying selection was also evident in approximately 2% of the IS605 orfB and approximately 10% of the IS607 orfB codons analyzed. We suggest that the evolution of these elements reflects selection for optimal transposition activity in the case of IS605 orfA and for interactions between the OrfB proteins and other cellular constituents that potentially contribute to bacterial fitness. Taken together, similarities in IS elements and H. pylori population genetic structures and evidence of adaptive evolution in IS elements suggest that there is coevolution between these elements and their bacterial hosts.

FIG. 1.

Unrooted, radial gene tree generated by the ML method by using a 927-bp concatenated sequence from IS605 orfA (378 bp) and orfB (549 bp). All available sequences were used in the phylogenetic reconstruction. orfA sequences with stop codons are indicated by double daggers. The most appropriate model for IS605 sequence evolution (TIM + I + Γ) was used for phylogenetic reconstruction with a discrete gamma distribution (Γ) shape parameter (α = 0.803) and an assumed proportion of invariate sites (I = 0.69). The TIM model (a constrained submodel of the general time reversible model) specifies a rate substitution matrix in which transitions (A↔G and C↔T) have only one rate category and the four possible transversions (A↔C, G↔T, A↔T, and C↔G) have four distinct rate categories. Bootstrap values of ≥50 are indicated at the nodes. H. pylori strain designations indicate the geographic origins, as follows: HUP, Spain; SJM, Peru; I, India; OKI, Okinawa, Japan; F or CPY, Honshu, Japan; and HK, DUQT, or Pcm, Hong Kong. Major sequence similarity clusters are circled. nt, nucleotide.

FIG. 2.

(A) Unrooted, radial gene tree generated by the ML method by using an 1,100-bp concatenated sequence from IS607 orfA (537 bp) and orfB (563 bp). Sequences with stop codons in orfA and orfB are indicated by double daggers and solid diamonds, respectively. The TVM + I + Γ model was used in phylogenetic reconstruction with α = 0.304 and I = 0.46. The TVM rate substitution matrix specifies that transitions have two distinct rate categories and transversions have only two specified rate categories (A↔C and G↔T have one rate and A↔T and C↔G have one rate). Bootstrap values of ≥50 are indicated at the nodes; distinct IS607 sequence similarity clusters are circled. H. pylori strain GS1 is from Honshu, Japan; for an explanation of all other strain designations see the legend to Fig. 1. nt, nucleotide. (B) Summary of polymorphisms in IS607 subpopulations: Venn diagram summarizing the unique, shared, and fixed polymorphisms observed in IS607group Ia (n = 9), group Ib (n = 11), and group II (n = 19). Fixed differences between populations are indicated below the arrows outside the conjoined circles.

FIG. 3.

Split decomposition analysis of IS605. (A) Annotated splits graph of the concatenated 927-bp IS605 orfA and orfB sequence generated with a pairwise TIM + I + Γ distance matrix. A fit parameter of 7.9 indicated the virtual absence of a tree-like structure. (B) Recalculation of the splits graph after removal of IS605 sequences from clusters IndA and IndB improved the fit parameter and resolved networks linking the East Asian and Indo-European elements. (C) Splits graph of IS605 showing population subdivision within IS605, which was masked by a history of recombinational exchanges between the two lineages. All branch lengths are drawn to scale.

FIG. 4.

Split decomposition analysis of IS607. (A) Annotated splits graph of the 1,100-bp concatenated sequence from IS607 orfA and orfB generated with a pairwise TVM + I + Γ distance matrix. A fit parameter of 28.7 indicated that all conflicts in the data set could not be resolved. Removal of distant outgroups in the data set improved the fit parameter (indicating better resolution of conflicts), as shown in the annotated splits graph for group I IS607 (B) and the annotated splits graph for group II IS607 (C). All branch lengths are drawn to scale.