Emergence and spread of Mycobacterium ulcerans at different geographic scales

Abstract

The classical lineage of Mycobacterium ulcerans is the most prevalent clonal group associated with Buruli ulcer in humans. Its reservoir is strongly associated with the environment. We analyzed together 1,045 isolates collected from 13 countries on two continents to define the evolutionary history and population dynamics of this lineage. We confirm that this lineage spread over 7,000 years from Australia to Africa with the emergence of outbreaks in distinct waves in the 18th and 19th centuries. In sharp contrast with its global spread over the last century, transmission chains are now mostly local, with little or no dissemination between endemic areas. This study provides new insights into the phylogeography and population dynamics of M. ulcerans, highlighting the importance of comparative genomic analyses to improve our understanding of pathogen transmission.

Importance: Mycobacterium ulcerans is an environmental mycobacterial pathogen that can cause Buruli ulcer, a severe cutaneous infection, mostly spread in Africa and Australia. We conducted a large genomic study of M. ulcerans, combining genomic and evolutionary approaches to decipher its evolutionary history and pattern of spread at different geographic scales. At the scale of villages in an endemic area of Benin, the circulating genotypes have been introduced in recent decades and are not randomly distributed along the river. On a global scale, M. ulcerans has been spreading for much longer, resulting in distinct and compartmentalized endemic foci across Africa and Australia.

Keywords: Mycobacterium ulcerans; WGS; evolution; phylogeography.

Fig 1

Core-genome phylogeny of 307 M. ulcerans strains (Mu-A1 lineage) isolated from patients living in South-East Benin and South-West Nigeria. raxML phylogenetic tree of M. ulcerans isolates inferred from 1,520 core-genome SNP analysis by Parsnp and visualized with iTOL. Analyses were run using RAxML with a rapid 1,000 bootstrap analysis and a general time-reversible model of evolution according to a distribution with four rate categories (GTRGAMMA). The reference genome (Agy99) is shown. Based on the segregation indicated by this tree, the genomes were split into 10 monophyletic or paraphyletic genotypes. A specific color is assigned to each taxon.

Fig 2

Temporal and spatial distribution of the eight genotypes. (A) Distribution of Mycobacterium ulcerans genotypes according to diagnosis dates for Buruli ulcer patients in Benin and Nigeria. The distribution of genotypes was tested with contingency tables (Fisher’s exact test) comparing years and there is no significant difference. (B) Spatial cluster detection results for Mycobacterium ulcerans genotypes for Buruli ulcer patients in Benin. Two significant areas were detected along the Ouémé River (northern Ouémé and southern Ouémé). Composition of these two clusters relative to that expected for a random distribution.

Fig 3

Focus on genotype 8, its subgenotypes, and spatial distribution. (A) The phylogenetic tree for genotype 8 reveals the presence of potentially emerging genotypes. (B) Location of the clusters of subgenotypes. (B) Two significant spatial clusters were identified in which specific subgenotypes were overrepresented relative to other areas. Composition of these two clusters relative to that expected for a random distribution.

Fig 4

Time tree of the eight genotypes (G1–G8) of data set 1. BEAST-dated phylogenetic tree of strains built from non-recombining SNPs. Mean node ages and their associated 95% highest posterior density intervals are indicated in calendar years. The branches of the different genotypes are collapsed and colored. The red star indicates the clade at which a significant temporal signal was found and for which tip dating was performed.

Fig 5

The recent Mu lineage is composed of three sub-lineages, including five clades in the African sub-lineage. Core-genome phylogeny of 1,045 M. ulcerans strains isolated from patients living in Australia, Africa, and Papouasie New Guinea based on 16,500 SNPs. Analyses were run using RAxML with a rapid 1,000 bootstrap analysis and a general time-reversible model of evolution according to a distribution with four rate categories (GTRGAMMA). The M. marinum outgroup was not represented. The presence of a specific deletion for each genome is indicated by the colored dot corresponding to this specific sequence and is also reported in Table 4. The geographic origin of the patients is indicated by a line of color. A clade color identifies the SL3.1 and SL3.2 sub-lineages and five clades in the SL3.3 sub-lineage. Strains whose names are written in red are data set 1 strain.

Fig 6

Localization of the sub-lineages and clades of Mu’s recent lineage in West Africa and Australia. The size of the circles is not proportional to the sample size.

Fig 7

Distribution of the genome sizes according to sub-lineages and clades. The distribution of the size of each aligned genome is observed using a box-and-whisker plot according to their group (Sub-lineages SL3.1, SL3.2, and SL3.3 and clades of SL3.3).

Fig 8

Bayesian temporal analyses on a representative 300 isolate data set of Mu’s recent lineage. BEAST-dated phylogenetic tree of a subset of 300 Mu strains built from non-recombining SNPs. Mean node ages are indicated in calendar years. The branches of the different groups are collapsed and colored. An outgroup was included but not represented. The youngest and oldest isolates of this data set date to 1945 and 2019, respectively.