Genomic and functional determinants of host spectrum in Group B Streptococcus

Abstract

Group B Streptococcus (GBS) is a major human and animal pathogen that threatens public health and food security. Spill-over and spill-back between host species is possible due to adaptation and amplification of GBS in new niches but the evolutionary and functional mechanisms underpinning those phenomena are poorly known. Based on analysis of 1,254 curated genomes from all major GBS host species and six continents, we found that the global GBS population comprises host-generalist, host-adapted and host-restricted sublineages, which are found across host groups, preferentially within one host group, or exclusively within one host group, respectively, and show distinct levels of recombination. Strikingly, the association of GBS genomes with the three major host groups (humans, cattle, fish) is driven by a single accessory gene cluster per host, regardless of sublineage or the breadth of host spectrum. Moreover, those gene clusters are shared with other streptococcal species occupying the same niche and are functionally relevant for host tropism. Our findings demonstrate (1) the heterogeneity of genome plasticity within a bacterial species of public health importance, enabling the identification of high-risk clones; (2) the contribution of inter-species gene transmission to the evolution of GBS; and (3) the importance of considering the role of animal hosts, and the accessory gene pool associated with their microbiota, in the evolution of multi-host bacterial pathogens. Collectively, these phenomena may explain the adaptation and clonal expansion of GBS in animal reservoirs and the risk of spill-over and spill-back between animals and humans.

Fig 1. Diagram illustrating host specialism levels in Group B Streptococcus (GBS) clonal groups (CG).

A) Host generalist lineages show extensive between-lineage homologous recombination and the three host-associated accessory gene clusters (scpB, Lac.2, Locus 3) are primarily found in isolates from their associated host, while they are lacking from isolates from other hosts. Host restricted lineages can be associated with reductive evolution (e.g., gene loss of function and genome reduction as in CG260, CG261 and CG552), they carry host-associated genes (e.g., CG260, CG261 and CG552 all carry Locus 3) and they either show absence of recombination (CG260, CG261 and CG552) or recombination limited within CG (e.g., CG61 and CG91). Host-adapted CG are primarily associated with one host and they show all the characteristics of the host restricted lineages except for genome reduction and pseudogenisation. B) Thresholds used for the categorisation of CG in the three levels of host specialism; the prevalence of genomes within each CG associated with the dominant host species (x-axis) was plotted to identify cut-offs, where the y-axis represents the number of CG corresponding to a given host prevalence.

Fig 2. Population structure and pangenome of Group B Streptococcus (GBS).

A) Maximum-likelihood phylogenetic tree of 1,254 GBS genomes; leaf colours indicate host of origin, and external strips show sublineage (SL), clonal group (CG) and homologous recombination; tree was rooted on an out-group of five reference genomes from Streptococcus pyogenes (hidden); B) Prevalence of host species within each CG; C) Correlation of number of genes and assembly size; SL552 shows a marked pseudogenisation and reduced genome size; D) Correlation of average number of recombination bases and recombination blocks of each CG, as well as recombination to mutation (r/m) rate; E) Recombination observed in the three categories of host-specialism; the difference between groups is statistically significant (p-value<0.0001).

Fig 3. Accessory gene distance network of 1,254 Group B Streptococcus (GBS) genomes.

A) Major host groups (human, bovine, fish and camel) are overlaid to the nodes; B) Sublineages (SL) defined by fastbaps and renamed based on an inheritance principle from 7-gene MLST nomenclature are shown. The two panels show association of accessory gene clusters with SL, and a lack of clustering based on host species, unless when this is a direct result of host-specific lineages (e.g., SL552, SL91, SL61).

Fig 4. Host-associated accessory gene clusters in Group B Streptococcus (GBS).

A) Manhattan plot of human-associated unitigs mapped to reference genome NZ_CP019979 (only genes/elements detected as significant by both GWAS methods are here annotated); B) Manhattan plot of bovine-associated unitigs mapped to reference genome CP008813; C) Circular maximum-likelihood phylogenetic tree of 1,254 GBS genomes. Leaf colours indicate host of origin, whereas the three external strips show presence/absence of the three main host-associated accessory gene clusters (human-scpB, bovine-Lac.2, fish-Locus 3, respectively).

Fig 5. Locus 3 gene cluster and survival curves of Nile tilapia (Oreochromis niloticus).

A) Locus 3 gene cluster organisation and associated gene functions; B) Tilapia challenged intraperitoneally (IP) with Group B Streptococcus (GBS) Sequence Type (ST) 7 wild type (WT) or its isogenic mutant (ΔLocus3); C) Tilapia challenged with GBS ST7 (WT) or its isogenic mutant (ΔLocus3) through cohabitation with IP-challenged fish; D) Tilapia challenged IP with GBS ST283 (WT) or its isogenic mutant (ΔLocus3); E) Tilapia challenged with GBS ST283 (WT) or its isogenic mutant (ΔLocus3) through cohabitation with IP-challenged fish. All experiments (B-E) included a negative control (mock challenge with phosphate buffered saline). Below survival curves, number at risk has been indicated for each timepoint.

Fig 6. Time-scaled phylogenies of two clones of public health interest in Group B Streptococcus (GBS).

Branch colour shows host-jumps, leaf colour the host of isolation, and external strips show presence-absence of genes and features of interest. A) Clonal group (CG) 283 is predicted to have originated around 1986 from humans. Its population shows diversification in multiple sub-clades over the years. Genomes associated with human cases from the 2015 outbreak of foodborne infection in Singapore are part of a sub-clade in which most of the genomes carry the human-associated transposon bearing scpB. In recent years, CG283 has been described in Brazil (first isolates available are from 2015); this introduction likely happened from Vietnam, and it is associated with a scpB-negative genotype. The two main scpB loss events are indicated with arrows on the phylogeny; B) Sublineage (SL) 23 likely originated in humans at the beginning of the 18^th century. It is described as a host-generalist clonal complex (CC). However, when assessing host prevalence within its two CG (CG23 and CG24), as well as the distribution of human and bovine-associated accessory gene clusters (scpB and Lac.2, respectively) the two CG show distinct patterns of host-tropism. CG24 appears as human-adapted, whereas isolates of CG23, whose common ancestor is predicted to have acquired Lac.2 in the second half of the 1800s (red arrow), show a tendency towards bovine-specialism, but retain the ability to cause human infections (scpB is highly conserved).