A core genome multilocus sequence typing (cgMLST) analysis of Mycoplasma bovis isolates

Mycoplasma bovis (M. bovis) is an emerging major bovine pathogen, causing economic losses worldwide in the dairy and beef industry. Whole-genome sequencing (WGS) now allows high resolution for tracing clonal populations. Based on WGS, we developed the core genome multilocus sequence typing (cgMLST) scheme and applied it onto 151 genomes of clonal and non-clonal strains of M. bovis isolated from China, Australia, Israel, Denmark, Canada, and the USA. We used the complete genome of M. bovis PG45 as the reference genome. The pairwise genome comparison of these 151 genome sequences resulted in 478 cgMLST gene targets present in > 99.0 % clonal and non-clonal isolates with 100 % overlap and > 90 % sequence similarity. A total of 478 core genes were retained as cgMLST target genes of which an average of 90.4-99 % were present in 151 M. bovis genomes, while M. agalactiae (PG2) had 17.0 % and M. mycoides subsp. capri (PG3), M. ovipneumoniae (Y98), and M. arginine resulted in 0.0 % of good targets. When tested against the clonal and non-clonal strains, we found cgMLST clusters were congruent with the MLST-defined clonal groups, which had various degrees of diversity at the whole-genome level. Notably, cgMLST could distinguish between clonal and epidemiologically unrelated strains of the same clonal group, which could not be achieved using traditional MLST schemes. Our results showed that ninety-two M. bovis genomes from clonal group isolates had > 10 allele differences and unambiguously differentiated from unrelated outgroup strains. Additionally, cgMLST revealed that there might be several sub-clones of the emerging ST-52 clone. The cgMLST phylogenetic analysis results showed substantial agreement with geographical and temporal information. cgMLST enables the use of next-generation sequencing technology to bovine mycoplasma epidemiology at both the local and global levels. In conclusion, the novel cgMLST scheme not only showed discrimination resolution highly as compared with MLST and SNP cgMLST in sub-typing but also indicated the capability to reveal more population structure characteristics than MLST.