An African origin for Mycobacterium bovis

Abstract

Background and objectives: Mycobacterium bovis and Mycobacterium caprae are two of the most important agents of tuberculosis in livestock and the most important causes of zoonotic tuberculosis in humans. However, little is known about the global population structure, phylogeography and evolutionary history of these pathogens.

Methodology: We compiled a global collection of 3364 whole-genome sequences from M.bovis and M.caprae originating from 35 countries and inferred their phylogenetic relationships, geographic origins and age.

Results: Our results resolved the phylogenetic relationship among the four previously defined clonal complexes of M.bovis, and another eight newly described here. Our phylogeographic analysis showed that M.bovis likely originated in East Africa. While some groups remained restricted to East and West Africa, others have subsequently dispersed to different parts of the world.

Conclusions and implications: Our results allow a better understanding of the global population structure of M.bovis and its evolutionary history. This knowledge can be used to define better molecular markers for epidemiological investigations of M.bovis in settings where whole-genome sequencing cannot easily be implemented.

Lay summary: During the last few years, analyses of large globally representative collections of whole-genome sequences (WGS) from the human-adapted Mycobacterium tuberculosis complex (MTBC) lineages have enhanced our understanding of the global population structure, phylogeography and evolutionary history of these pathogens. In contrast, little corresponding data exists for M. bovis, the most important agent of tuberculosis in livestock. Using whole-genome sequences of globally distributed M. bovis isolates, we inferred the genetic relationships among different M. bovis genotypes distributed around the world. The most likely origin of M. bovis is East Africa according to our inferences. While some M. bovis groups remained restricted to East and West Africa, others have subsequently dispersed to different parts of the world driven by cattle movements.

Keywords: Mycobacterium bovis; bovine tuberculosis; molecular clock; phylogeography; whole-genome sequencing; zoonosis.

Figure 1.

Geographic distribution of the M.bovis samples used in this study according to isolation country. The circles correspond to pie charts and are coloured according to clonal complexes

Figure 2.

Maximum-likelihood phylogeny of 476 of the 3364 genomes included in this study (redundant genomes were removed), and inferred from 22 492 variable positions. The scale bar indicates the number of substitutions per polymorphic site. The phylogeny is rooted on a M.tuberculosis Lineage 6 genome from Ghana (not shown) and bootstrap values are shown for the most important splits. The coloured bars on the side of the phylogeny show the different clonal complexes. Other ‘unknown’ monophyletic clades are coloured in black and the branches of these are coloured to show their phylogenetic position more precisely. The pie charts mapped on the tree represent the summary posterior probabilities (from 100 runs) of the reconstructed ancestral geographic states and are coloured according to geographical UN region. Inferred spoligotype patterns from WGS described in M.bovis spoligotype database [30] are indicated for the unknown clades. The red circles at the tips correspond to the eight newly sequenced genomes. Regions of difference (RD) as in [31] are indicated; superscript + and − refer to presence of the region or its deletion, respectively

Figure 3.

The inferred age of main monophyletic clades according to LSD and BEAST dating analyses. For BEAST, we report the results of the two analyses that resulted in the lowest clock rate (subset1 Bayesian Skyline) and in the highest clock rate (subset3 exponential population growth). The confidence intervals reported correspond to the merged HPD interval of the two BEAST analyses mentioned above. The BEAST analysis was based on 300 genomes and the LSD analysis was based on 2058 genomes (see Methods section for sub-sampling strategy). Only one genome from the Af1 clonal complex was included in the dating analyses and therefore the dates reported correspond to the node where Af1 diverged