On the origin of Mycobacterium ulcerans, the causative agent of Buruli ulcer

Abstract

Background: Mycobacterium ulcerans is an unusual bacterial pathogen with elusive origins. While closely related to the aquatic dwelling M. marinum, M. ulcerans has evolved the ability to produce the immunosuppressive polyketide toxin mycolactone and cause the neglected tropical disease Buruli ulcer. Other mycolactone-producing mycobacteria (MPM) have been identified in fish and frogs and given distinct species designations (M. pseudoshottsii, M. shinshuense, M. liflandii and M. marinum), however the evolution of M. ulcerans and its relationship to other MPM has not been defined. Here we report the comparative analysis of whole genome sequences from 30 MPM and five M. marinum.

Results: A high-resolution phylogeny based on genome-wide single nucleotide polymorphisms (SNPs) showed that M. ulcerans and all other MPM represent a single clonal group that evolved from a common M. marinum progenitor. The emergence of the MPM was driven by the acquisition of the pMUM plasmid encoding genes for the biosynthesis of mycolactones. This change was accompanied by the loss of at least 185 genes, with a significant overrepresentation of genes associated with cell wall functions. Cell wall associated genes also showed evidence of substantial adaptive selection, suggesting cell wall remodeling has been critical for the survival of MPM. Fine-grain analysis of the MPM complex revealed at least three distinct lineages, one of which comprised a highly clonal group, responsible for Buruli ulcer in Africa and Australia. This indicates relatively recent transfer of M. ulcerans between these continents, which represent the vast majority of the global Buruli ulcer burden. Our data provide SNPs and gene sequences that can differentiate M. ulcerans lineages, suitable for use in the diagnosis and surveillance of Buruli ulcer.

Conclusions: M. ulcerans and all mycolactone-producing mycobacteria are specialized variants of a common Mycobacterium marinum progenitor that have adapted to live in restricted environments. Examination of genes lost or retained and now under selective pressure suggests these environments might be aerobic, and extracellular, where slow growth, production of an immune suppressor, cell wall remodeling, loss or modification of cell wall antigens, and biofilm-forming ability provide a survival advantage. These insights will guide our efforts to find the elusive reservoir(s) of M. ulcerans and to understand transmission of Buruli ulcer.

Figure 1

Comparative genome content of selected study isolates to references M. marinum and M. ulcerans.  Circos [36] plot of de novo assembled contigs, for a representative sample of the study isolates, mapped against the reference genomes of M. marinum M and M. ulcerans Agy99. The key below the figure describes the content. Moving inwards, the tracks are genes coloured by functional group (lipid metabolism – black; insertion seqs – aqua; others – green). The next track marks insertion sequences in pink. The remaining tracks show selected isolates and their coverage when de novo contigs are mapped against the reference. The isolates moving inwards are: Mu_06-3845 (Benin), Mu_05142109 (Australia), Mu_L15 (USA), Mu_CC240299 (Israel), Mu_1G897 (French Guiana), Mu_8765 (Japan), Mm_1726 (USA) and M. marinum “M”. The SNPs present in the M.ulcerans core genome are marked by type with coloured triangles in the innermost rings. The 12 loci overlapping the core M. ulcerans – pan M. marinum regions are shown with the outer most labels MUL_nnnn. The M. marinum region spanning the Esx-1 locus is scaled up x30 to highlight how it has been successively affected by deletion and is no longer part of the core M. ulcerans genome (blue circle).

Figure 2

Comparative genome content of plasmid pMUM001. Circos [36] plot of de novo assembled contigs, for a sample of the study isolates, mapped against reference chromosome plasmid pMUM001. ISs and known genes are labelled in the outermost ring. Moving inwards, the tracks are genes coloured by functional group (lipid metabolism – black, insertion seqs – aqua, others – green). The next track marks insertion sequences in pink. The remaining tracks show selected isolates and their coverage when de novo contigs are mapped against the reference. The isolates moving inwards are: Mu_06-3845 (Benin), Mu_05142109 (Australia), Mu_L15 (USA), Mu_CC240299 (Israel), Mu_1G897 (French Guiana). The cyp150 gene (mup053) has been highlighted (grey wedge) to show its presence in African isolates and absence in other isolates.

Figure 3

Open genome analysis. Median number of novel genes added per isolate (solid line) and the fitted power law model (equation inset). The median is for 1,000 random orderings of 32 of the study isolates. The dashed line shows the +/−1 standard deviation from the median.

Figure 4

MuMC phylogenomic analysis. Neighbour joining dendrogram based on 128,463 variable common nucleotide positions across 34 sequenced isolates produced by SplitsTree4 [41] using uncorrected ‘P’ distances. The tree was rooted using M. tuberculosis as an outgroup. The major clustering of isolates are M. marinum isolates (4) – blue; Fish and frog isolates (5) – green; Japanese isolate (1) – Mu_8765; French Guiana isolate – Mu_1G897; Australian isolates (10) – orange; African isolates (13) – red. The scale bars show the median pairwise divergence for the set of isolates they span. The isolates that produce mycolactone are highlighted in pale yellow. The inset shows the pMUM plasmid SNP tree of 315 SNPs with edge lengths and a topology matching the corresponding isolates in the main tree. The details of the isolates can be found in Table 1.

Figure 5

African isolates tree and map. Unrooted neighbour joining tree produced by SplitsTree4 [41] using uncorrected ‘P’ distances and based on 396 variable common nucleotide positions from a core of 5,190,553 bp among the 13 isolates from Ghana and Benin. The map shows their geographic distribution with pin colours corresponding to the tree. The isolate Mu_06-3846 is from Benin but clusters with the Ghanian isolates (see text). Map produced using Google Maps.

Figure 6

Venn diagrams, showing DNA shared and unique among isolates. Venn diagrams showing the unique genomic material in the isolates when partitioned by species, host and region. All regions were identified by finding the core genome of the isolates within the set and removing the pan-genome of all isolates excluded from the set. The unique material for each partition is shown as total size in base pairs, number of segments in partition with a minimum size of 200 bp, the length of the largest segment and the number of CDS with an overlap of >10%. The intersection of the sets was determined by finding the core genome of the union of the sets of isolates and shows the size of the core and the number of CDS with an overlap of >10%. (A) Shows the partition between all 30 M. ulcerans isolates and all four M. marinum isolates. (B) Shows the partition between the four M. ulcerans fish and frog host isolates and the M. ulcerans human host isolates. The human host isolates are a strict subset of the fish and frog isolates with no unique DNA. (C) Shows the partitioning between the 10 Australian and 13 African M. ulcerans isolates and their separation from the isolates from the rest of the world (6 isolates). The combined African and Australian isolates have no unique DNA when compared to the rest of the world isolates.

Figure 7

Inferred pseudogenes and deleted genes. Counts of inferred pseudogenes and deleted genes among the study isolates. The counts were inferred by analysis of SNPs that would render M. marinum reference CDS inactive and read coverage mapping showing partially or fully deleted CDS. The isolates are coloured by groups; African isolates – red, Australian isolates – gold, French Guiana isolate – purple, Japanese isolate – pink, fish and frog isolates – green.

Figure 8

IS 2404 and IS 2606 distribution among isolates. Counts of IS2404 and IS2606 among the study isolates. The counts were inferred from coverage of reads mapping to each IS. The isolates are coloured by groups; African isolates – red, Australian isolates – gold, French Guiana isolate – purple, Japanese isolate – pink, fish and frog isolates – green. IS2404 and IS2606 were not detected in the non-MPM M. marinum isolates