Vaccination Drives Changes in Metabolic and Virulence Profiles of Streptococcus pneumoniae

Abstract

The bacterial pathogen, Streptococcus pneumoniae (the pneumococcus), is a leading cause of life-threatening illness and death worldwide. Available conjugate vaccines target only a small subset (up to 13) of >90 known capsular serotypes of S. pneumoniae and, since their introduction, increases in non-vaccine serotypes have been recorded in several countries: a phenomenon termed Vaccine Induced Serotype Replacement (VISR). Here, using a combination of mathematical modelling and whole genome analysis, we show that targeting particular serotypes through vaccination can also cause their metabolic and virulence-associated components to transfer through recombination to non-vaccine serotypes: a phenomenon we term Vaccine-Induced Metabolic Shift (VIMS). Our results provide a novel explanation for changes observed in the population structure of the pneumococcus following vaccination, and have important implications for strain-targeted vaccination in a range of infectious disease systems.

Fig 1. Shifts in associations between capsular serotype and metabolic types type following vaccination.

(A) Associations between the most frequent pneumococcal serotypes and MLST-defined STs, England (adapted from [4]) (B) Disease prevalence of three STs of serotype 19A among children <5 years in the United States, following vaccination in 2000 (adapted from [7]) (C) Partial screenshot of the metabolic profiles of 616 pneumococcal isolates (obtained from [22]) grouped by serotype (see also S1 Dataset). Each column represents the allelic profile of 876 metabolic/transport genes of a single isolate. Black bars indicate where VIMS (Vaccine-Induced Metabolic Shift) has occurred. (D) Linkage Disequilbrium (measured as D’) between pairs of metabolic loci & pairs of non-metabolic loci, plotted against their genetic distance on the chromosome. (E) Box-plots showing the distribution of D’ among metabolic and non-metabolic pairs of loci. Values of D’ were significantly higher among metabolic/transport loci than non-metabolic loci (Wilcoxon Test: W = 615, p<0.0001).

Fig 2. The effects of vaccination on pathogen population structure.

(A) Schematic describing how non-overlapping associations emerge between antigenic type (AT: a,b) and metabolic type (MT: 1,2). Vaccination against serotype a causes a shift in population structure, inducing a metabolic shift (VIMS) in serotype b (B) Strain dynamics following vaccination against serotype a, showing two forms of VIMS in serotype b: either with exclusion of MT2 (γ = 0.9) or in coexistence with it (γ = 0.75). Other parameters: R₀ (MT1) = 4.001, R₀ (MT2) = 4, σ = 4, μ = 0.02. (C) Regions of parameter space associated with (i) non-overlapping associations between AT & MT (red), (ii) competitive exclusion by MT1 (yellow), and (ii) an intermediate state (green). Regions of VIMS are shown in light blue (with exclusion) and dark blue (with coexistence). Orange areas indicate where vaccination has led to the loss of the less transmissible MT (here MT2). Prevalence of a strain had to exceed 5% of the total infections present to be recorded as contributing to the population structure (σ = 4, μ = 0.02).