Genome sequencing of disease and carriage isolates of nontypeable Haemophilus influenzae identifies discrete population structure

Abstract

One of the main hurdles for the development of an effective and broadly protective vaccine against nonencapsulated isolates of Haemophilus influenzae (NTHi) lies in the genetic diversity of the species, which renders extremely difficult the identification of cross-protective candidate antigens. To assess whether a population structure of NTHi could be defined, we performed genome sequencing of a collection of diverse clinical isolates representative of both carriage and disease and of the diversity of the natural population. Analysis of the distribution of polymorphic sites in the core genome and of the composition of the accessory genome defined distinct evolutionary clades and supported a predominantly clonal evolution of NTHi, with the majority of genetic information transmitted vertically within lineages. A correlation between the population structure and the presence of selected surface-associated proteins and lipooligosaccharide structure, known to contribute to virulence, was found. This high-resolution, genome-based population structure of NTHi provides the foundation to obtain a better understanding, of NTHi adaptation to the host as well as its commensal and virulence behavior, that could facilitate intervention strategies against disease caused by this important human pathogen.

Keywords: NTHi; genomics; population genetics.

Fig. 1.

(A) Minimum spanning tree (MST) based on the allelic profiles of all isolates present in the MLST database. Each node is an ST, and it is colored according to the serotypes of the isolates. The size of the nodes is proportional to the number of isolates. By applying the traditional definition of clonal complex (CC) (i.e., six out of seven identical MLST genes), we found 464 different CCs, of which 151 included more than one ST. Most capsulated isolates were concentrated in a small number of CCs whereas the nontypeable STs were dispersed in a high number of CCs, many of which included a single ST. (B) STs of the sequenced isolates colored according to disease. Although there is some association between serotype B and invasive disease, carriage, COPD, and otitis media are associated with variable capsular genotype.

Fig. 2.

A 3D scatterplot of the first three principal components of the DAPC of the polymorphic sites. Each dot is one isolate, colored according to the classification into one of the six clusters. Clusters II and III are closely related whereas all other clusters are clearly separated by DAPC.

Fig. 3.

Neighbor-joining phylogenetic tree of the sequenced strains. The tree was rooted using Haemophilus haemoliticus as outgroup. The tips are colored according to the assigned cluster in the DAPC analysis. The DAPC classification identifies six monophyletic clades. Note the position of the strain RM603375, the only discordance between the DAPC and the phylogenetic analysis. Shades indicate isolates that have most of the capsule biosynthetic locus intact although they do not express capsule. The serotype inferred by homology of the capsule operon is indicated (B*, D*, F*).

Fig. 4.

A 3D scatterplot of the first three principal components of the DAPC of the presence/absence profiles. Points are colored according to the clades defined using the core SNPs (Figs. 2 and 3). Although clades I and IV–VI are clearly separated by the analysis of the presence/absence profiles, isolates of clades II and III are not separable.

Fig. 5.

Distribution of the hif, hia/hsf, hmw1/hmw2, igaB, iga1, and hap genes. Green marks isolates where the gene is present, white where it is absent. There is a good correlation between presence/absence of the hif, hia/hsf, hmw1/hmw2, and igaB genes and the six clades. Clades I and VI are relatively more heterogeneous whereas clades II–V are composed of a very homogeneous set of isolates.