Identifying candidate genetic determinants associated with invasion in pneumococcal lineages

Abstract

Serotype invasiveness estimates help identify high-risk serotypes and interpret disease replacement following pneumococcal conjugate vaccination. Because serotype and lineage are strongly associated, their independent contributions must be separated to identify unexpectedly invasive lineages and investigate underlying mechanisms. Using temporally matched carriage and invasive-disease isolates from children aged <7 years in South Africa and the USA, we fitted a logistic mixed-effects model with Global Pneumococcal Sequence Cluster (GPSC) as a random effect and serotype and country as fixed effects. We conducted a species-wide genome-wide association study (GWAS) of invasion and compared the prevalence of significant unitigs in GPSC1 and GPSC21 to identify invasion-associated variants differentiating them. We additionally examined capsule-locus variation that could contribute to their contrasting invasiveness. After serotype and country adjustment, GPSC1 and GPSC21, which both express 19F and share an immediate common ancestor, ranked fifth and 149th of 150 lineages; odds multipliers were 1.51 (95% conditional interval 1.06–2.15) and 0.61 (0.41–0.93), respectively. Of 1,569 invasion-associated unitigs from the species-wide GWAS, 272 differentiated GPSC1 from GPSC21. Among these, the strongest signal mapped to hk11 (desK) and was associated with GPSC1; bgaA was the highest-ranked known antibody-binding target. Within GPSC1 CC271 had additional variants in the ATP-dependent Clp protease ATP-binding subunit ClpX compared to the CC420 subclade. Capsule-protein folds were conserved with local effects indistinguishable from prediction uncertainty. Separating serotype and lineage effects can identify unexpectedly invasive lineages and prioritize candidate serotype-independent determinants, vaccine antigens and capsule variation for experimental investigation. Data Summary Code implementing the mixed-effects invasiveness model is available at https://github.com/rgladstone/pneumo-lineage-invasion. Deidentified isolate metadata, sequence accessions, analysis inclusion and supporting data underlying the figures and tables are available in the Supplementary Tables. Impact Statement Among pneumococcal isolates collected from children younger than 7 years, pneumococci with the same capsule type differed in their capacity to cause invasive disease. Separating the effects of capsule type and genetic lineage makes it possible to identify unexpectedly invasive lineages and investigate the reasons for their increased disease potential. Applying this approach highlighted candidate proteins that may help explain invasiveness and could be explored as vaccine antigens with relevance beyond individual lineages. Several differences were found in and around the capsule-producing region; predicted protein structures were conserved, although subtle effects on capsule composition or expression cannot be ruled out. This work therefore provides a way to uncover both non-capsular factors and overlooked capsule variation associated with disease, helping to prioritize mechanisms, vaccine candidates and structures for experimental investigation.