Two genotyping schemes, multi-locus sequence typing (MLST) and Streptococcus canis M-like protein (SCM) gene-based typing, are used to characterise Streptococcus canis, a multi-host pathogen commonly isolated from dogs and occasionally reported in cases of human infection. Using whole genome sequencing (WGS) we assessed the accuracy of these typing schemes against core genome single nucleotide polymorphism (SNP) analysis, and we investigated S. canis host tropism.
We analysed 59 S. canis genomes originating from multiple hosts. We created three core SNP phylogenies, two constrained to be monophyletic for MLST and SCM types (Figure 1), and compared the tree topologies by an approximately unbiased (AU) test. MLST and SCM clustering was compared to core genome SNP (CGS) phylogenetic clustering using the adjusted Wallace (AW) coefficient. Signs of host adaptation were investigated using phylogenetic analysis, accessory genome clustering and pan-genome wide association analysis.
Figure 1. A – core SNP phylogeny; B – MLST-constrained core SNP phylogeny; C – SCM-constrained core SNP phylogeny.
Based on the AU test, the MLST and SCM schemes failed to represent the core SNP-based genomic diversity. The accuracy of both schemes, expressed by the AW index with 95% confidence intervals (CI), was comparable (AWMLSTCGS: 0.575; CI 0.495-0.654. AW SCMCGS: 0.540; CI 0.467-0.613). Host adaptation of S. canis strains was not detected.
The MLST and SCM schemes lack sufficient resolution for fine typing such as outbreak scenarios. SCM genotyping performed as accurately as MLST, and due to its lower resource requirements would be preferable when WGS is not available. S. canis inter-species transmission, including zoonotic transfer, is strongly supported.