Parallel Evolution of Group B Streptococcus Hypervirulent Clonal Complex 17 Unveils New Pathoadaptive Mutations

Abstract

Group B Streptococcus (GBS) is a commensal of the gastrointestinal and genitourinary tracts, while a prevailing cause of neonatal disease worldwide. Of the various clonal complexes (CCs), CC17 is overrepresented in GBS-infected newborns for reasons that are still largely unknown. Here, we report a comprehensive genomic analysis of 626 CC17 isolates collected worldwide, identifying the genetic traits behind their successful adaptation to humans and the underlying differences between carriage and clinical strains. Comparative analysis with 923 GBS genomes belonging to CC1, CC19, and CC23 revealed that the evolution of CC17 is distinct from that of other human-adapted lineages and recurrently targets functions related to nucleotide and amino acid metabolism, cell adhesion, regulation, and immune evasion. We show that the most distinctive features of disease-specific CC17 isolates were frequent mutations in the virulence-associated CovS and Stk1 kinases, underscoring the crucial role of the entire CovRS regulatory pathway in modulating the pathogenicity of GBS. Importantly, parallel and convergent evolution of major components of the bacterial cell envelope, such as the capsule biosynthesis operon, the pilus, and Rib, reflects adaptation to host immune pressures and should be taken into account in the ongoing development of a GBS vaccine. The presence of recurrent targets of evolution not previously implicated in virulence also opens the way for uncovering new functions involved in host colonization and GBS pathogenesis. IMPORTANCE The incidence of group B Streptococcus (GBS) neonatal disease continues to be a significant cause of concern worldwide. Strains belonging to clonal complex 17 (CC17) are the most frequently responsible for GBS infections in neonates, especially among late-onset disease cases. Therefore, we undertook the largest genomic study of GBS CC17 strains to date to decipher the genetic bases of their remarkable colonization and infection ability. We show that crucial functions involved in different steps of the colonization or infection process of GBS are distinctly mutated during the adaptation of CC17 to the human host. In particular, our results implicate the CovRS two-component regulator of virulence in the differentiation between carriage- and disease-associated isolates. Not only does this work raise important implications for the ongoing development of a vaccine against GBS but might also drive the discovery of key functions for GBS adaptation and pathogenesis that have been overlooked until now. Author Video: An author video summary of this article is available.

Keywords: CovR; GBS vaccine; ST17; antibiotic resistance; eubacteria; evolution; genomics; group B Streptococcus; virulence.

FIG 1

Core genome phylogeny of CC17. Shown is a phylogenetic tree of 626 CC17 genomes built with RAxML (22) and based on the core and recombination-free alignment of 12,584 SNPs along a 1.47-Mb sequence. Outgroup strain B83 is depicted with a thicker branch. Isolates are color coded according to the geographical origin (branch tips) and clinical state (outside circle), as indicated in the central key. The four main distinct clades in the phylogeny are differently colored and labeled with roman numerals.

FIG 2

Parallel evolution of coding sequences above neutral expectation. (A) Mutation frequency per gene observed in the CC17 population (black) in relation to the average substitution rate across the COH1 reference genome (blue line). Red dots correspond to the 152 genes with a statistically significant (P < 0.05, exact Poisson test) mutational bias compared to a neutral model of evolution (Table S2). (B) Venn diagram depicting the proportion of genes with a significant mutational signature across CC17, CC1, CC19, and CC23. (C) Functional classification of the 152 significant genes in CC17, based on the eggNOG database (24). Fold change corresponds to the proportional difference in the number of genes among those with a mutational bias compared to the COH1 coding sequences (in parentheses). Statistical significance was assessed with a Fisher exact test. *, P < 0.05; **, P < 0.01.

FIG 3

Mutation frequencies per gene in carriage- and disease-associated isolates. Shown is a plot of a linear model assessing the correlation between the mutations acquired per gene for each clinical status. Red, gray, and blue dots depict genes that were more, equally, or less mutated in strains associated with disease or carriage according to the key at the top. The size of each data point is proportional to the number of genes found, as indicated in the key at the top. Outlier genes were detected with a Bonferroni-adjusted outlier test, and only those with a P value of <0.05 are represented by the symbol ×. Those highlighted with a red circle and whose names are shown were significantly associated with disease independently of population structuring.

FIG 4

Pigmentation of strains with covRS-related mutations. (A) Level of pigment production plotted against the phylogeny of 18 different CC17 strains. Strains WT1 to WT9 correspond to isolates with no mutations in genes assumed to affect the activity of CylE (covR, covS, abx1, stk1, and cylE). For the remaining strains, mutations potentially affecting covRS or cylE are indicated next to the heat map. Strains WT1, WT2, and WT4 to WT9 were obtained from carriers, while all others were collected from infections. (B) Plate image of the strains tested, with average values and standard deviations of the pigment levels obtained from four independent experiments depicted below the spots. Ratios ranging from 0 to 1 were the result of normalization against the sample with the highest intensity in each test, calculated with ImageJ (https://imagej.nih.gov/ij).

FIG 5

Phase variation of rib between carriage and disease. Shown is the normalized coverage of the gene coding for the alpha-like surface protein Rib in carriage- and disease-specific strains, as also depicted in Fig. S3. Statistical significance values were calculated with a two-tailed t test. ***, P < 0.001.

FIG 6

Genetic characterization of antibiotic resistance. (A) Antibiotic resistance determinants present in the ResFinder database (47) that were detected in both the sequencing reads and the assembled genomes of the CC17 strains, plotted according to their core genome phylogeny. The number of isolates with each antibiotic resistance gene is depicted below each column. Gene absence is represented in gray, while the remaining colors illustrate the different classes of antibiotics. (B) Distribution of the antibiotic resistance genes present within a genomic island that was first detected among CC17 strains from China (20). The phylogenetic tree corresponds to a particular clade containing all 14 CC17 strains that have been sequenced in China (20). For each gene, dark red boxes indicate presence and white indicates absence.

FIG 7

Persistence and short-term evolution of CC17 in neonates. Shown are mutations differentiating three pairs of isolates collected 1 month apart from different blood cultures of infected newborns. The number of days since birth is indicated below each strain. The name of the gene affected by the mutation and its effect on the protein sequence are presented as shown in the key at the lower right. Percentages alongside Rib represent changes in normalized sequencing coverage.