Evolution of transcription regulatory genes is linked to niche specialization in the bacterial pathogen Streptococcus pyogenes

Abstract

Streptococcus pyogenes is a highly prevalent bacterial pathogen, most often giving rise to superficial infections at the throat or skin of its human host. Three genotype-defined subpopulations of strains exhibiting strong tropisms for either the throat or skin (specialists) or having no obvious tissue site preference (generalists) are recognized. Since the microenvironments at the throat and skin are distinct, the signal transduction pathways leading to the control of gene expression may also differ for throat versus skin strains of S. pyogenes. Two loci (mga and rofA/nra) encoding global regulators of virulence gene expression are positioned 300 kb apart on the genome; each contains alleles forming two major sequence clusters of approximately 25 to 30% divergence that are under balancing selection. Strong linkage disequilibrium is observed between sequence clusters of the transcription regulatory loci and the subpopulations of throat and skin specialists, against a background of high recombination rates among housekeeping genes. A taxonomically distinct commensal species (Streptococcus dysgalactiae subspecies equisimilus) shares highly homologous rof alleles. The findings provide strong support for a mechanism underlying niche specialization that involves orthologous replacement of regulatory genes following interspecies horizontal transfer, although the directionality of gene exchange remains unknown.

FIG. 1.

Phylogenetic trees of complete ORFs of transcriptional regulatory genes of S. pyogenes. Neighbor-joining tree (mid-point rooted) with maximum likelihood distance for mga derived from 19 isolates (A) and rofA/nra derived from 23 isolates (B). The rate matrices were optimized to the best-fit models, using the hierarchical likelihood ratio test: TrN+G (for mga) and TVM+G (for rofA/nra). Bootstrap values showing confidence intervals of ≥90% are indicated (1,000 replicates). Excluding alignment gaps, the total number of nucleotide sites is 1,579 for panel A and 1,482 for panel B. Three alleles have a premature termination signal or readthrough, relative to the stop codon of the majority of alleles for that locus (A735 and Manfredo for mga; MGAS8232 for rofA/nra); these sequences were trimmed to the same position as the majority of ORFs for that allele. For rofA/nra, the genome map position adjacent to the highly conserved hsp33 locus was confirmed by PCR. Taxon labels indicate S. pyogenes strains and are listed and described further in Table S1 in the supplemental material, except for strains D471, MGAS8232, and SF370, for which the GenBank accession numbers are M58461, AE010111, and AE006624 for mga and U01312, AE009963, and AE006482, for rofA/nra, respectively. GenBank accession numbers for strain B737 (CS101) are X68501 (for mga) and SPU49397 (for nra); for strain A735, the GenBank accession number is AF447492 (for rofA/nra); for strain Manfredo, sequences were obtained from www.sanger.ac.uk. Note that for rofA, the ORFs used for analysis begin with a codon for leucine, not methionine (25). The GenBank accession numbers for new mga and rofA/nra sequences are given in Materials and Methods.

FIG. 2.

Phylogenetic tree of rofA and rofCG. The phylogenetic tree is as described in the legend of Fig. 1B, with an additional six rofCG alleles. The rate matrix was optimized to the best-fit model (GTR+G), using the hierarchical likelihood ratio test. Taxon labels indicate streptococcal strains, as listed and described further in Table S1 (for S. pyogenes) or Table S2 in the supplemental material (32); included are three strains each of group C and G streptococci. For rofCG, the genome map position adjacent to the highly conserved hsp33 locus was confirmed by PCR. The GenBank accession numbers for six new rofCG sequences are given in Materials and Methods.