Evolutionary diversification of an ancient gene family (rhs) through C-terminal displacement

Abstract

Background: Rhs genes are prominent features of bacterial genomes that have previously been implicated in genomic rearrangements in E. coli. By comparing rhs repertoires across the Enterobacteriaceae, this study provides a robust explanation of rhs diversification and evolution, and a mechanistic model of how rhs diversity is gained and lost.

Results: Rhs genes are ubiquitous and comprise six structurally distinct lineages within the Enterobacteriaceae. There is considerable intergenomic variation in rhs repertoire; for instance, in Salmonella enterica, rhs are restricted to mobile elements, while in Escherichia coli one rhs lineage has diversified through transposition as older lineages have been deleted. Overall, comparative genomics reveals frequent, independent gene gains and losses, as well as occasional lateral gene transfer, in different genera. Furthermore, we demonstrate that Rhs 'core' domains and variable C-termini are evolutionarily decoupled, and propose that rhs diversity is driven by homologous recombination with circular intermediates. Existing C-termini are displaced by laterally acquired alternatives, creating long arrays of dissociated 'tips' that characterize the appearance of rhs loci.

Conclusion: Rhs repertoires are highly dynamic among Enterobacterial genomes, due to repeated gene gains and losses. In contrast, the primary structures of Rhs genes are evolutionarily conserved, indicating that rhs sequence diversity is driven, not by rapid mutation, but by the relatively slow evolution of novel core/tip combinations. Hence, we predict that a large pool of dissociated rhs C-terminal tips exists episomally and these are potentially transmitted across taxonomic boundaries.

Figure 1

A multiple alignment of rhs protein sequences from across the Enterobactericaeae. Scale in amino acid residues. The alignment is divided into four: 'clade-specific' N-terminal domain, core domain, (including a hyperconserved domain) and variable C-terminal domain. The region used in phylogenetic analyses is bordered by a dotted line and conserved amino acid motifs. Clade structure (refer to Figure 2) is shown at left.

Figure 2

Maximum likelihood phylogeny showing global rhs genetic diversity. The ML phylogram was estimated from a multiple alignment of 81 rhs core domain nucleotide sequences from 11 genera, using a GTR+Γ model. Scale is in substitutions/site. The topology is concordant with alternative trees estimated from protein sequence alignments, and with Bayesian phylogenies. Node support is provided by non-parametric bootstrap values/Bayesian posterior probabilities; an asterisk * denotes values of 100/1.00 respectively. Each terminal node is labelled with the species (in bold), strain and locus tag, where available. Existing rhs aliases are shown in red next to E. coli K12 rhs sequences. The tree is subdivided into six clades and the phylogenetic distribution of each bacterial genus is shown on the right.

Figure 3

Comparative genomics and phylogenetics of rhs genes in E. coli/Shigella spp. a. Rhs loci are found at 11 unique positions in E. coli. These are marked along the K12_MG1655 genome (scale in base-pairs, beginning at the thr operon). For each locus the following are noted: the existing gene name ('alias') where available, the clade to which it belongs (see Figure 2), the presence or absence of a contiguous vgrs gene, and its phylogenetic distribution across all strains (present: solid circle, pseudogene: crossed through, relic: half circle, or otherwise absent). Note that two sequences labelled 'III' belong to Clade III, rather than Clade I. The ML phylogeny shown at left was estimated from MLST concatenated sequence (see Methods), and is labelled with bootstrap values. Coloured boxes denote the inferred origins of rhs loci. b. A phylogenetic network estimated from HKY distances using a Neighbour-Net algorithm. Sequence labels are shaded by locus, as in a. A key is provided that relates strain names to sequence codes. Clades are linked to their corresponding positions by arrows.

Figure 4

Arrangement of associated and dissociated C-terminal tips in Serratia marcescens. The chromosome sequence is shown in grey, with gene models and G+C content shown plotted above. The regions between dotted lines correspond to rhs genes and fragments (core domain: red; hyperconserved domain: yellow; variable C-terminal: various). Scale in base-pairs.

Figure 5

Comparison of C-terminal tip types in E. coli, at position 4 (clade IV rhs). The phylogenetic relationships of seven E. coli strains are shown at left. Complete and fragmentary rhs genes at position 4 in each strain are shaded as in Figure 4. Two strains possess a downstream core domain fragment without any obvious associated tip; the fragment and its corresponding location in the core domain are shaded grey.

Figure 6

A hypothetical model of C-terminal tip displacement. Homologous recombination between the conserved core sequence within the downstream unattached tip and either the core region of the rhs gene or that of another unattached tip. This event would result in the production of a recombination proficient episomal circle carrying a conserved core region, of varying lengths, which is attached to a variable C-terminal tip and any intervening genes. After transfer to a second bacterial isolate homologous recombination would be required between the highly conserved core regions on the chromosomal rhs gene and identical sequences located on the episomal circle carrying the unattached tip. This second single cross-over event of a circular intermediate is required to explain how an attached tip can be displaced by a new tip without deleting or entirely replacing the old tip, but simply shunting it to a silent position downstream of the intact rhs gene.