Global phylogenomic analysis of nonencapsulated Streptococcus pneumoniae reveals a deep-branching classic lineage that is distinct from multiple sporadic lineages

Abstract

The surrounding capsule of Streptococcus pneumoniae has been identified as a major virulence factor and is targeted by pneumococcal conjugate vaccines (PCV). However, nonencapsulated S. pneumoniae (non-Ec-Sp) have also been isolated globally, mainly in carriage studies. It is unknown if non-Ec-Sp evolve sporadically, if they have high antibiotic nonsusceptiblity rates and a unique, specific gene content. Here, whole-genome sequencing of 131 non-Ec-Sp isolates sourced from 17 different locations around the world was performed. Results revealed a deep-branching classic lineage that is distinct from multiple sporadic lineages. The sporadic lineages clustered with a previously sequenced, global collection of encapsulated S. pneumoniae (Ec-Sp) isolates while the classic lineage is comprised mainly of the frequently identified multilocus sequences types (STs) ST344 (n = 39) and ST448 (n = 40). All ST344 and nine ST448 isolates had high nonsusceptiblity rates to β-lactams and other antimicrobials. Analysis of the accessory genome reveals that the classic non-Ec-Sp contained an increased number of mobile elements, than Ec-Sp and sporadic non-Ec-Sp. Performing adherence assays to human epithelial cells for selected classic and sporadic non-Ec-Sp revealed that the presence of a integrative conjugative element (ICE) results in increased adherence to human epithelial cells (P = 0.005). In contrast, sporadic non-Ec-Sp lacking the ICE had greater growth in vitro possibly resulting in improved fitness. In conclusion, non-Ec-Sp isolates from the classic lineage have evolved separately. They have spread globally, are well adapted to nasopharyngeal carriage and are able to coexist with Ec-Sp. Due to continued use of PCV, non-Ec-Sp may become more prevalent.

Keywords: antibiotic nonsusceptibility; comparative genomics; integrative conjugative elements; pneumococcal isolates; whole-genome sequencing.

Fig. 1.—

Core genome tree of 131 nonencapsulated and 44 encapsulated S. pneumoniae. One hundred thirty-one sequences of non-Ec-Sp isolates from 17 different countries or states of America were included (shown in red, green, and blue). The classic lineage of exclusively non-Ec-Sp (light gray) contains the isolates of ST448 (green) and ST344 (red). Additionally published encapsulated pneumococci (indicated in pink) (Donati et al. 2010) clustered within other non-Ec-Sp (blue). COGs that are present in every isolate as a single copy and, in addition, having the same sequence length were selected for building the tree (363 COGs). The maximum-likelihood phylogeny was generated using 24,342 polymorphic sites within a 278,778 bp codon alignment. Three isolates of the recently published BC3-NT clade are also indicated (yellow) (Chewapreecha et al. 2014).

Fig. 2.—

Accessory genome diversity of 131 nonencapsulated and 44 encapsulated S. pneumoniae. The classic lineage of exclusively non-Ec-Sp is indicated (light gray) and contains the isolates of ST448 (green) and ST344 (red). The remaining isolates are sporadic non-Ec-Sp and Ec-Sp. To construct the input array data all isolates were compared pairwise. Of each pair the COGs (70% coverage, 70% similarity) which are not part of the core-genome and are present in both isolates were counted. The isolates in the x and y axes are ordered according to the Euclidean distance constructed dendrogram.

Fig. 3.—

Size and composition of accessory genome. The total number of accessory genes (A) of each isolate was determined by number of homologous groups (70% coverage, 70% similarity) which are present in the specific strain and not part of the core genome. The total genome sizes of all isolates are indicated (B). To detect genes of possible ICEs (C) in the different isolates, all CDS of each strain were aligned to the ICEberg database [1] (March 2014) using BLASTP (version 2.2.28+) [2]. Every CDS that showed homology (70% coverage, 70% similarity) to at least one ICE-protein was assumed to be ICE related. To determine the number of proteins that originate by phages (D), the genomes were scanned for possible prophage region using PhiSpy (version 2.2) [3]. All the CDS that lie in the region of a predicted prophage region were counted for each isolate separately. Means and SD (standard deviation) are indicated (A-D).

Fig. 4.—

Presence and absence of CDS of nonencapsulated and encapsulated S. pneumoniae compared with an ST344 reference genome. Each row illustrates a specific isolate genome with the presence or absence of CDS as defined in the reference genome (110.58). The rows are ordered by the phylogeny of the isolates (indicated on the left). The x axis represents the genomic position of the CDS in the genome of 110.58. The figure was constructed using R (version 3.1.0 alpha) with the “ggplot2” package. Two large surface proteins (Sp) were not correctly assembled in the illumina sequenced isolates and are therefore lacking. The reference genome (110.58) contains the genes aliB-like ORF1 and ORF2 at the cps (capsule) operon (Hathaway et al. 2004) which were absent and replaced by cps genes in Ec-Sp. RD, region of diversity; cps, capsule operon; ϕ, prophage.

Fig. 5.—

Genetic organization and nucleic acid alignment of ICEs. Artemis comparison tool (ACT) was used for the nucleic acid alignment of the ICESsu_BM44071 (S. suis), ICE₁SpST344 and ICESde₃₃₉₆ (Streptococcus dysagalagticae). Individual CDS are indicated in light blue. Conserved CDS between the three ICEs are indicated by red shading. CDS which were unique for ICE₁SpST344 are indicated in dark blue. Tn, transposon; pezAT, toxin–antitoxin genes: lrtA, reverse transcriptase A; prtP, PII-type proteinase precursor.

Fig. 6.—

Adherence of S. pneumoniae to human epithelial cells (A) and in vitro growth (B). Detroit nasopharyngeal epithelial cell lines were exposed to non-Ec-Sp containing or lacking ICE₁SpST344 and RD₁SpST344, respectively (A). Adherence was determined at 30 min and calculated as the proportion of recovered bacteria to the inoculum. Experiments were repeated three times (three times on three different days: Indicated are mean and SD (standard deviation)). Swiss and Thai strains were classic and sporadic non-Ec-Sp, respectively. For the nine adherence experiments, ordinary one-way ANOVA resulted in P = 0.0052. See text for details. MLST, multilocus sequence type. Isolates of the BC3-NT lineage have been recently defined (Chewapreecha et al. 2014). Measurement of planktonic growth was done in 96-well microtiter polystyrene plates and OD450nm was measured on an ELISA plate reader every 30 min for 22 h (B). Each experiment included three technical replicates and each experiment was performed three times.