• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of jbacterPermissionsJournals.ASM.orgJournalJB ArticleJournal InfoAuthorsReviewers
J Bacteriol. Mar 2005; 187(5): 1668–1676.
PMCID: PMC1064021

Common Evolutionary Origin for the Unstable Virulence Plasmid pMUM Found in Geographically Diverse Strains of Mycobacterium ulcerans


The 174-kb virulence plasmid pMUM001 in Mycobacterium ulcerans epidemic strain Agy99 harbors three very large and homologous genes that encode giant polyketide synthases (PKS) responsible for the synthesis of the lipid toxin mycolactone. Deeper investigation of M. ulcerans Agy99 resulted in identification of two types of spontaneous deletion variants of pMUM001 within a population of cells that also contained the intact plasmid. These variants arose from recombination between two 8-kb sections of the same plasmid sequence, resulting in the loss of a 65-kb region bearing two of the three mycolactone PKS genes. Investigation of nine diverse M. ulcerans strains by using PCR and Southern hybridization for eight pMUM001 gene sequences confirmed the presence of pMUM001-like elements (collectively called pMUM) in all M. ulcerans strains. Physical mapping of these plasmids revealed that like M. ulcerans Agy99, three strains had undergone major deletions in their mycolactone PKS loci. Online liquid chromatography-sequential mass spectrometry analysis of lipid extracts confirmed that strains with PKS deletions were unable to produce mycolactone or any related cometabolites. Interstrain comparisons of the plasmid gene sequences revealed greater than 98% nucleotide identity, and the phylogeny inferred from these sequences closely mimicked the phylogeny from a previous multilocus sequence typing study in which chromosomally encoded loci were used, a result that is consistent with the hypothesis that M. ulcerans diverged from the closely related organism Mycobacterium marinum by acquiring pMUM. Our results suggest that pMUM is a defining characteristic of M. ulcerans but that in the absence of purifying selection, deletion of plasmid sequences and a corresponding loss of mycolactone production readily arise.

Mycobacterium ulcerans causes Buruli ulcer (BU), a devastating human disease that is characterized by chronic, necrotic ulceration of subcutaneous fat. Left untreated, BU leads to high morbidity due to loss of limbs or limb function. Surgery is the only recommended treatment. Over the last 15 years the numbers of BU cases has been increasing dramatically, particularly in the rural regions of West and Central Africa, where in some instances the prevalence of BU now exceeds the prevalence of leprosy and is equal to or even higher than the prevalence of tuberculosis (1, 8, 32). The mode of transmission is unknown, but M. ulcerans is known to have aquatic environmental reservoirs. The bacteria can be cultured from the salivary glands of carnivorous aquatic insects and from the surfaces of aquatic plants in regions where BU is endemic (20, 21). In an experimental model of infection M. ulcerans but not other mycobacteria were able to colonize insect salivary glands (20). This observation suggests the presence of M. ulcerans-specific colonization factors. The mechanisms by which M. ulcerans can persist and grow within insect salivary glands remain to be determined, but one important factor may be the M. ulcerans-specific polyketide mycolactone (12). Apart from its potential role in the environment, purified mycolactone has cytotoxic, apoptotic, and immunosuppressive properties (11, 13, 25). It is also thought to be the primary virulence determinant for M. ulcerans because when injected subcutaneously into an animal model, purified mycolactone can reproduce the pathology seen in the natural M. ulcerans infection (12).

In the original description of mycolactone, George et al. (12) extracted acetone-soluble lipids from the Malaysian M. ulcerans strain 1615 and identified mycolactone A/B ([M + Na]+ at m/z 765) with Z- and E- isomers of a 12-member macrolactone linked via an ester bond to a fatty acyl side chain. The structure of mycolactone A/B has been confirmed by complete chemical synthesis (10) (Fig. (Fig.11).

FIG. 1.
Structures of mycolactone A (Z4′,5′) and B (E4′,5′) ([M + Na]+ at m/z 765).

We recently determined the genetic basis for mycolactone biosynthesis and showed that M. ulcerans has a 174-kb megaplasmid harboring three genes (mlsA1 [51 kb], mlsA2[7 kb], and mlsB [42 kb]). These unusually large genes encode the three type I polyketide synthases (PKS) that are responsible for the synthesis of mycolactone (30). For type I PKS, one PKS module catalyzes one round of polyketide chain extension. Modules are made up of at least three specific functional domains (see references 18 and 27 for reviews). MlsA1 is a putative 1.8-MDa protein composed of eight extension modules that links with MlsA2 (0.26 MDa, comprising one extension module) via specific docking domains (3). Together, these two PKS produce the mycolactone core structure. MlsB is a 1.2-MDa protein composed of seven extension modules, and it synthesizes the fatty acyl side chain. An unusual FabH-like ketosynthase that has predicted acyltransferase activity, encoded by another plasmid gene, MUP045, may be involved in linking the side chain to the mycolactone core. While the Mls proteins conform to the modular and domain structure of other type I PKS, they are also highly unusual because of their extreme size and their extraordinarily high level of interdomain sequence identity. Whereas other PKS from other bacterial species display levels of amino acid identity between domains having the same function of up to 75%, for the Mls PKS this value is greater than 97% (30).

One might expect such high sequence homology to promote frequent genetic rearrangements and therefore a diversity of metabolites, but M. ulcerans strains from around the world have thus far been shown to produce a very restricted repertoire of mycolactones. A study of 34 M. ulcerans isolates collected worldwide showed that they all make the same lactone core with minor variations in the acyl side chain (22). These variations have been largely attributed to various degrees of oxidation at C-12′ of the side chain (15, 22), and it has been proposed that this is due to the activity (or lack of activity) of a specific P450 monooxygenase (encoded by the plasmid gene MUP053) (15, 30).

In this study we used a large-insert M. ulcerans DNA clone library to examine the stability of plasmid pMUM001. We then explored the distribution and structure of this plasmid in other M. ulcerans strains using PCR, DNA sequencing, pulsed-field gel electrophoresis (PFGE), and Southern hybridization.


Bacterial strains and culture conditions.

Escherichia coli strains DH10B [F mcrA Δ(mrr-hsdRMS-mcrBC) [var phi]80dlacZΔM15 ΔlacX74 deoR recA1 araD139 Δ(ara, leu)7697 galU galK rpsL endA1 nupG] and XL2-Blue (recA1 endA1 gyrA96 thi-1 hsdR17 supE44 relA1 lac [F′ proAB lacIqZΔM15 Tn10(Tetr) Amy Camr]) were cultivated in Luria-Bertani broth at 37°C. Mycobacterium marinum strain M was cultivated at 32°C in 7H9 Middlebrook medium (Becton Dickinson) supplemented with oleic acid-albumin-dextrose-catalase (Difco). The following 10 M. ulcerans clinical isolates were used: Agy99 (origin, Ghana, 1999; this strain was used for the M. ulcerans genome sequencing project), Kob (origin, Ivory Coast, 2001; kindly provided by Laurent Marsollier, Institut Pasteur), 1615 (origin, Malaysia, 1963; kindly provided by Pamela Small, University of Tennessee), Chant (origin, southeastern Australia, 1993), IP105425 (from the reference collection of the Institut Pasteur and derived from reference strain ATCC 19428; origin, southeastern Australia, 1948), 01G897 (origin, French Guiana, 1991), ITM-5114 (origin, Mexico, 1958), ITM-941331 (origin, Papua New Guinea, 1994), ITM-98912 (origin, People's Republic of China, 1997), and ITM-941328 (origin, Malaysia, 1994). M. ulcerans isolates were grown as described above for M. marinum. M. ulcerans isolates whose designations begin with ITM were kindly provided by Françoise Portaels (Institute for Tropical Medicine, Antwerp, Belgium).

Liquid chromatography-sequential mass spectrometry analysis of mycolactones.

Lipid fractions from M. ulcerans were extracted and analyzed for mycolactones as previously described (11, 15)

Oligonucleotides and DNA methods.

The oligonucleotides used in this study are shown in Table Table1.1. Standard methods were used for subcloning, PCR, and automated DNA sequencing. DNA sequences were assembled and annotated by using Gap4 and Artemis, respectively (2, 26).

Oligonucleotides used in this study

PFGE and Southern hybridization.

Mycobacterial DNA was prepared in agarose plugs as follows. Bacterial cells were grown to the mid-log phase in 7H9 Middlebrook medium and harvested by centrifugation. The cells were inactivated by addition of 800 μl of 70% ethanol for 30 min at 22°C. The ethanol was then removed, and the cell pellet was washed once in 1% Triton X-100 and resuspended in Tris-EDTA (TE) buffer (10 mM Tris, 1 mM EDTA [pH 8.0]) by using as a guide 150 μl of TE buffer for every 10 mg (wet weight) of cells. The cells were mixed with an equal volume of 2% (wt/vol) low-melting-temperature agarose (Bio-Rad) at 45°C and dispensed immediately into plug molds (Bio-Rad). Up to 10 plug slices (4 by 7 mm) were then incubated for 18 h at 37°C in a 30-ml solution containing 0.5 M EDTA (pH 8.0), 0.5% Sarkosyl, 60 mg of deoxycholic acid, and 100 mg of lysozyme. The plugs were washed once in 1× TE buffer and incubated for a further 48 h at 50°C in a 30-ml solution containing 0.5 M EDTA (pH 8.0), 0.5% Sarkosyl, and 30 mg of proteinase K. The plugs were then washed extensively in 1× TE buffer at 4°C. Prior to restriction enzyme (RE) digestion, each plug slice was equilibrated for 30 min at room temperature in 400 μl of the RE buffer. Each plug slice was then incubated for 18 h at 37°C in 300 μl of RE buffer with 1% (wt/vol) bovine serum albumin and 40 U of XbaI. PFGE was performed by using the Bio-Rad CHEF DRII system (Bio-Rad) with 1.0% agarose in 0.5× Tris-borate-EDTA at 200 V with 3- to 15-s switch times for 15 h. DNA was visualized by staining with 0.5 μg of ethidium bromide per ml. Southern hybridization analysis was performed as follows. M. ulcerans genomic DNA, separated by PFGE as described above, was transferred to Hybond N+ nylon membranes by overnight alkaline transfer in 0.4 M NaOH. The gels were subjected to a 1,200-mJ UV treatment prior to transfer. DNA was fixed to the nylon membranes by cross-linking (1,200 mJ of UV irradiation) and then incubated in prehybridization buffer (5× SSC, 0.1% sodium dodecyl sulfate [SDS], 1% skim milk [1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate]) for at least 2 h at 68°C. DNA probes were prepared by random prime labeling of PCR products by using a HighPrime random labeling kit (Stratagene) and incorporation of [γ-32P]dCTP. Probes were denatured by heating at 100°C and were then added to hybridization buffer (5× SSC, 0.1% SDS, 1% skim milk) to a final concentration of approximately 10 ng/ml. Hybridization was performed at 68°C for 18 h. The hybridization solution was then removed, and three stringency washes were performed (one wash for 5 min in 2× SSC-0.1% SDS at room temperature and then two washes for 10 min in 0.1× SSC-0.1% SDS at 68°C). Each membrane was then washed in 2× SSC and sealed in clear plastic film before detection with a Storm PhosphorImager (Molecular Dynamics). Probe stripping was performed by washing the membrane twice for 20 min at 68°C with 0.1% SDS-0.2 M NaOH. The sizes of DNA restriction fragments were estimated with the Sigmagel software (Jandel Scientific) by using a lambda low-range DNA size ladder (New England Biolabs) to calibrate the gel and blot images.

BAC library construction.

A whole-genome M. ulcerans bacterial artificial chromosome (BAC) library was constructed as described previously for Mycobacterium tuberculosis (4). Briefly, genomic DNA from M. ulcerans Agy99 was prepared in agarose plugs as described above and subjected to partial HindIII digestion. The DNA was separated under PFGE conditions. Partially digested DNA in the size range from 40 to 120 kb was cloned into the unique HindIII site of the vector pBeloBAC11 and then used to transform E. coli DH10B by electroporation. The resulting clones were stored in Luria-Bertani broth containing 15% glycerol in a 96-well format at −80°C.

BAC plasmid DNA preparation.

BAC DNA for automated sequencing was extracted by the method of Brosch et al. (4). For subcloning of BACs, DNA was prepared from 40-ml overnight E. coli cultures, and the plasmid DNA was extracted as previously described (4).

Phylogenetic analysis.

The sequences from the four plasmid loci (repA, parA, mls, MUP045) that were present in all 10 M. ulcerans strains were concatenated in frame to produce a 1,266-bp semantide for each strain. These sequences were then aligned with CLUSTAL W (31). In the same way the plasmid sequences obtained from the seven M. ulcerans strains that contained the repA, parA, MUP011, mls load, mlsAT(II), MUP038, and MUP045 loci were concatenated in frame to produce a 2,208-bp semantide composed of these seven loci. Phylogenetic analysis was performed with the MEGA software, version 2.1 (19). P distances were used throughout this study as the overall level of sequence divergence was small. Values for synonymous (dS) and nonsynonymous (dN) mutation frequencies were calculated by the method of Nei and Gojobori (23), and standard errors of the means of these values were estimated by the method of Nei and Jin (24). The dS and dN calculations were performed by using the dSdNqw program (7).


M. ulcerans plasmid pMUM001 is unstable in M. ulcerans Agy99.

The 11 different functional domains of the mycolactone polyketide synthase genes (mlsA1, mlsA2, and mlsB) have an unprecedented level of interdomain nucleotide identity (>97%). The high level of sequence repetition within the locus is displayed in the Dotter plot shown in Fig. Fig.2.2. Our hypothesis was that this DNA homology could act as a substrate for recombination and could manifest itself as inherent instability and variability of the mls locus within and between M. ulcerans strains.

FIG. 2.
Dotter analysis of the pMUM001 DNA sequence, highlighting regions where there are repetitive DNA sequences. Direct repeat sequences are indicated by lines that are parallel to the main diagonal, while inverted repeats are indicated by lines that are perpendicular ...

The first evidence that this was indeed the case was obtained in the course of determining the complete sequence of pMUM001 when several M. ulcerans BAC clones, derived from a single DNA preparation of M. ulcerans Agy99, were found to represent two different deletion variants of the 174-kb plasmid. These variants were represented by clones 22A01 and 22D03, and they were discovered by DNA end sequencing of an M. ulcerans genomic BAC library of 176 clones. Sequence analysis revealed 22 clones containing pMUM001-related sequences (hereafter collectively referred to as pMUM). These 22 clones were then further grouped into two subfamilies based on two distinct types of PstI RE profiles. Some of the clones in each subfamily had end sequences that indicated that they had been cloned into pBeloBAC11 at a single (but varying) M. ulcerans HindIII site, raising the possibility that we had cloned the entire M. ulcerans plasmid. However, this hypothesis was discounted as the insert size of these clones was either 65 or 110 kb, which was much less than the expected size, 174 kb. Curiously, the sum of the sizes of these two BAC clones was 175 kb, leading us to suspect that these clones represented deletion variants of pMUM001. A representative clone from each family was fully sequenced and annotated. Comparisons of the complete sequence of each clone with the complete sequence of pMUM001 indicated that these clones were indeed deletion derivatives that had arisen as a result of recombination between two identical 8,237-bp sequences overlapping the beginning of mlsA1 and mlsB (Fig. (Fig.22 and 3A and B). This arrangement was confirmed by PstI RE digestion and Southern hybridization of all BAC clones containing M. ulcerans plasmid sequences (Fig. 3C and D). These alternate plasmid forms were not detectable by PFGE and Southern hybridization of M. ulcerans genomic DNA (Fig. (Fig.4A)4A) and probably represented subpopulations of the predominant 174-kb plasmid form. It is possible that they represented deletion variants that arose by recombination in E. coli, but the presence of several examples of the same variations, cloned at different HindIII sites, (Fig. (Fig.3C),3C), and the existence of similar variants in spontaneous M. ulcerans mycolactone mutants (Fig. (Fig.4)4) argue against this proposition and support the idea that this was a real phenomenon, reflecting the inherent instability of the locus.

FIG. 3.
Mapping of the deletion variants of pMUM001. (A) Scaled, circular maps of pMUM001 and the two types of deletion derivatives, with a proposed model for recombination-mediated deletion. The positions of all HindIII sites are indicated. On the outer circles, ...
FIG. 4.
Mapping of pMUM in seven M. ulcerans strains. (A) PFGE and Southern hybridization analyses with five selected PCR-derived probes from pMUM001 and undigested and XbaI-digested genomic DNA extracted from M. ulcerans and M. marinum. Lane 1, M. ulcerans Agy99; ...

All M. ulcerans strains contain a related plasmid.

To explore interstrain plasmid variation, a panel of nine M. ulcerans clinical isolates with geographically diverse origins was screened by PCR for the presence of eight M. ulcerans plasmid markers. The results of this analysis are summarized in Table Table2.2. The presence of key plasmid replication and maintenance genes (repA and parA) and sections of the mycolactone biosynthesis genes (mls loading domain and MUP045) in all isolates indicated that they all contain an element closely related to pMUM001.

PCR analysis of 10 different M. ulcerans strains for the presence of eight plasmid-associated genes

Plasmid variation between strains.

The absence of several of the other plasmid markers in some of the isolates pointed to plasmid variation. Most notable was the absence in three isolates of key mycolactone accessory genes, such as MUP038 (encoding a type II thioesterase), and one of the mls acyltransferase domains; the absence of the latter sequence indicated that these isolates should be unable to produce mycolactone.

PFGE and Southern hybridization were used to study in more detail the structure of the plasmids in 7 of the 10 M. ulcerans strains. M. ulcerans DNA was separated by PFGE. This DNA was then hybridized with a pool of probes derived from five of the plasmid markers described in Table Table2.2. The results are shown in Fig. Fig.4A4A and demonstrate that there was a considerable difference in plasmid size among the isolates, with the sizes ranging from 59 to 174 kb. M. ulcerans strains harboring plasmids less than 110 kb long would not be expected to produce mycolactone as the Mls biosynthetic cluster is encoded by genes encompassing approximately 110 kb of DNA. Screening of lipid extracts from the seven isolates by liquid chromatography-mass spectrometry confirmed this prediction and that of the PCR analysis, as neither mycolactone nor its cometabolites were detected in extracts from M. ulcerans Kob (a recent West African isolate with a 101-kb plasmid), M. ulcerans ITM-5114 (a Mexican isolate with a 59-kb plasmid), and M. ulcerans IP105425 (an isolate from the culture collection of the Institut Pasteur, derived from reference strain ATCC 19428, with a 76-kb plasmid). Digestion with XbaI and hybridization with the five pooled plasmid markers resulted in a profile consisting of two, three, or four bands. For each strain, the sum of its XbaI fragments was equal to the size of the linear plasmid form in the absence of XbaI digestion (Fig. (Fig.4).4). This demonstrated that none of the plasmids had new, additional XbaI fragments. Hybridization experiments with individual probes then permitted linking of plasmid markers to particular XbaI fragments and construction of low-resolution maps (Fig. (Fig.4B).4B). The three mycolactone-minus strains had large deletions of 75, 98, and 115 kb. The hybridization data showing the absence of MUP038 (encoding the type II thioesterase), together with the PCR data showing an absence of the acyltransferase domain of module 5 in mlsA1 and the acyltransferase domain of modules 1 and 2 in mlsB, confirmed that these deletions had occurred, at least in part, within their respective mls loci.

Only the strains with four XbaI fragments (M. ulcerans Agy99, 1616, Chant, and ITM-941331) produced mycolactone, and thus by definition they must all contain an intact mls locus. This fact was supported by the presence of conserved 54- and 13-kb fragments, corresponding to the locus harboring the mlsA genes and MUP038. Therefore, the size variations detected among these four strains occurred in the regions flanking the mls genes.

Plasmid variation correlates with the presence of different mycolactone cometabolites.

For M. ulcerans strains Chant and ITM-941331, some of the plasmid size variation could be attributed to the absence of a region that includes the gene MUP053 (encoding a P450 hydroxylase). The product of MUP053 is predicted to hydroxylate the mycolactone side chain at C-12′ to produce mycolactone A/B with a mass of [M + Na]+ at m/z 765 (30). Strains lacking the hydroxyl group at C-12′ produce a mycolactone with a mass of [M + Na]+ at m/z 749. This metabolite has been called mycolactone C (22), and it is a characteristic of Australian strains. The absence of MUP053 in the Australian M. ulcerans strain Chant correlates well with the presence of mycolactone C and the absence of mycolactone A/B (Fig. (Fig.5).5). However, M. ulcerans ITM-941331 also lacks MUP053, yet this strain produces the same mycolactone profile as M. ulcerans Agy99 (15) (data not shown).

FIG. 5.
Liquid chromatography-mass spectrometry analysis of the lipid extract from Australian isolate M. ulcerans Chant, showing the absence of a mycolactone ([M+Na]+ at m/z 765.5) and the presence of the nonhydroxylated mycolactone ([M+Na] ...

Sequence analysis indicates a common origin for pMUM.

Comparisons of the DNA sequences obtained from the four plasmid markers common to all M. ulcerans strains revealed nucleotide identity scores of >98%. For each strain, the four sequences obtained were concatenated in frame in the order repA-parA-MUP045-mls in the loading domain to produce a 422-codon semantide. The sequences were aligned, and a summary of the 16 variable sites detected by this analysis is shown in Fig. Fig.6A.6A. A phylogenetic relationship was then inferred from these sequences, and this produced a dendrogram with a topology that closely mimicked the topology produced by a similar analysis of seven chromosomally encoded genes in a previous multilocus sequence typing study (Fig. 6C and E) (29). The congruence of these trees strongly suggests that pMUM was acquired in a single event and has coevolved with its host. Comparisons of the frequencies of synonymous substitution in coding sequences are a measure of the time that a given sequence has been extant relative to another sequence (16). Thus, similar synonymous substitution frequencies for the plasmid-borne gene sequences and the chromosomally encoded gene sequences are consistent with the idea that plasmid acquisition coincided with the divergence of M. ulcerans from a common progenitor. The values of dS (where dS is number of synonymous substitutions per 100 synonymous sites) for the plasmid and chromosomal sequences were not significantly different (for plasmid-borne gene sequences the mean dS was 0.59 [standard error, 0.24]; for chromosomal gene sequences the mean dS was 0.54 [standard error, 0.17]).

FIG. 6.
Phylogenetic analysis of 10 M. ulcerans strains with selected plasmid markers. (A) Alignment of 1,266-bp sequences derived from the four concatenated pMUM protein-coding loci present in all 10 M. ulcerans strains. Only variable nucleotides are shown. ...

Seven of the ten strains had seven of the eight plasmid markers. Therefore, to try to obtain further discrimination, the sequences of these strains were treated as described above. Thus, for a given strain the seven sequences were concatenated in frame in the order repA-parA-MUP011-mls load-mlsAT(II)-MUP038-MUP045 to produce a 736-codon semantide. These sequences were aligned and exhibited more than 99% nucleotide identity (Fig. (Fig.6B).6B). The inferred phylogeny was entirely consistent with that produced by using the four plasmid markers and multilocus sequence typing (Fig. (Fig.6D6D).

MUP053, encoding a putative P450 monooxygenase with a possible role in modifying mycolactone, had an uneven distribution in the strains. However, MUP053 was present in strains from Africa, Malaysia, the People's Republic of China, and Mexico, and these strains spanned the known genetic diversity of the species. The levels of DNA and amino acid identity for MUP053 in these strains were 98 and 96%, respectively, which were equal to the values for other plasmid sequences (Fig. (Fig.6F).6F). This suggests that MUP053 was present in a progenitor M. ulcerans and was subsequently lost from some strains as the species evolved.


A rapidly accumulating body of evidence highlights the important role that horizontal or lateral gene transfer (LGT) plays as a driver of adaptive evolution in microbes (9). However, until now, LGT among the pathogenic mycobacteria was thought to be a rare event. There is clear evidence from whole-genome sequence data that obligate intracellular mycobacterial pathogens, such as M. tuberculosis and Mycobacterium leprae, have evolved by gene loss rather than gene acquisition (5). M. ulcerans provides the first direct evidence of the importance not only of gene loss but also of LGT in the evolution of pathogenesis in the mycobacteria. M. ulcerans is an example of an emerging mycobacterial pathogen that has evolved by acquiring a plasmid (pMUM) that confers a virulence phenotype and, probably more critically for the organism, a fitness advantage for a particular niche environment. Previous multilocus sequence typing studies have shown that at a nucleotide level, M. ulcerans is highly related to M. marinum, which is a natural pathogen of fish and phenotypically is quite distinct from M. ulcerans. However, the two species were shown to exhibit more than 98% DNA identity for seven nonlinked genes when 40 diverse strains were examined (28). Phylogenetic analysis strongly suggested that M. ulcerans evolved from a common M. marinum progenitor, and from this result it was hypothesized that the divergence of M. ulcerans as a discrete clonal group was assisted by acquisition of foreign DNA. Subsequent work revealed the presence of the virulence plasmid pMUM in M. ulcerans, and in the present study we found that pMUM is a key attribute of M. ulcerans and that it is present in a range of strains obtained from around the world. Comparisons of pMUM gene sequences of these strains with chromosomal gene sequences revealed congruent tree topologies and identical frequencies of synonymous substitution, strongly suggesting that acquisition of pMUM marked the divergence of the species from a single M. marinum progenitor. Plasmid acquisition was then followed by other independent genome changes within strains from different areas to produce the regiospecific phenotypes and genotypes that we now see (6, 28, 29).

One of the unusual features of pMUM001 is the unprecedented DNA homology among the functional domains of the mls genes. While the mls genes occupy 105 kb of pMUM001, this region contains less than 10 kb of unique sequence (30). This extraordinary economy of sequence is reflected in Fig. Fig.22 and suggests that the mls genes were created de novo by successive recombination events, such as in-frame duplications and deletions from a core set of PKS sequences. The precise origin of such a core gene set remains obscure as DNA database searches have revealed no orthologous genes, but the significant levels of amino acid identity to PKS sequences from other species of mycobacteria and streptomyces point to a likely origin among the actinomycetes. In addition to suggesting a recent evolutionary origin for mycolactone biosynthesis, the extended DNA sequence homology also implies that such an arrangement is inherently unstable and acts as a substrate for general recombination. In this study we showed that in M. ulcerans Agy99, pMUM001 is unstable and that recombination between two homologous sequences gave rise to two deletion variants. The larger 109-kb variant, represented by BAC clone 22D03, contains an intact origin of replication and is thus likely to be maintained within a cell population. Cells harboring the 22D03 variant should be unable to produce mycolactone but could theoretically still produce the acyl side chain. However, the smaller 65-kb deletion variant, represented by BAC clone 22A01, should be lost from the population upon cell division as it is not capable of autonomous replication, despite having the genes required for synthesis of the mycolactone core. Spontaneous mycolactone-minus and avirulent M. ulcerans mutants were first reported by George et al. (12) and were used to demonstrate the key role of mycolactone in virulence. Mycolactone confers a pale yellow color to colonies, and mycolactone-minus mutants are readily observed as white colony variants when they are grown on Löwenstein-Jensen medium. We attempted to isolate white colony variants of M. ulcerans Agy99 to try to identify the 109-kb deleted form of pMUM001. While white colonies were readily detected on Löwenstein-Jensen medium, their growth after subculturing was highly impaired, and we were unable to generate the biomass required for additional studies, such as PFGE. Nevertheless, investigation of other M. ulcerans strains revealed deleted forms of pMUM similar to those identified in M. ulcerans Agy99 (in particular, in M. ulcerans Kob), and these deleted forms had corresponding toxin-minus phenotypes. Each strain tested had a different plasmid size, and the mapping data showed that deletions had occurred to various extents and in different regions of pMUM. Recombination between homologous sequences is one explanation for this variety, but given the large number of insertion sequences in pMUM (30), another possibility is that insertion sequences also mediated some of these plasmid rearrangements.

It is probably significant that no pMUM-minus M. ulcerans strains were found. While such mutants may exist, the recent finding that pMUM contains an active partition (par) locus (T. P. Stinear et al., Microbiology, in press) means that spontaneous curing is likely to be an infrequent event. par loci are cis-acting elements that ensure that daughter cells faithfully receive a copy of an episome during cell division.

Based on the assumption that the clinical isolates used in this study were originally mycolactone proficient and thus contained intact pMUM, it appears that spontaneous toxin-minus mutants, caused by deletion of M. ulcerans plasmid DNA, are a common occurrence. We have not been able to calculate the frequency with which deletion mutants arise, but for some strains it appears to be very high. M. ulcerans Agy99 and Kob were recent clinical isolates from West Africa with minimal laboratory passaging. The DNA used for the M. ulcerans Agy99 BAC library was prepared from a liquid culture that was at its fourth passage since primary isolation, and M. ulcerans Kob was at its third passage. One outcome of this work is to highlight the care that researchers must take to continually test the plasmid and mycolactone status of the M. ulcerans strains used in their work. Periodic passaging of lab strains through an animal virulence model (e.g., mice) may be required to ensure maintenance of a homogeneous population of mycolactone-producing bacteria.

Plasmid instability contrasts most strikingly with the fact that M. ulcerans isolates recovered from diverse geographic locations around the world produce a relatively homogeneous range of mycolactones (22). This apparent paradox leads compellingly to the notion that there is strong purifying selection for maintenance of a mycolactone-proficient form of pMUM, presumably because mycolactone has a key function in M. ulcerans in the environment. It is probably unlikely that the cytotoxic properties of mycolactone for human cells are part of the primary survival strategy of the bacterium. However, given the highly episodic and geographically compact epidemiology of Buruli ulcer, where waves of M. ulcerans infection can rapidly appear and then disappear from a given region, one possibility is that deleterious recombination and loss of the plasmid function interrupt the chain of transmission at some point. Perhaps mycolactone is a factor required for colonization or persistence in insect salivary glands (19) or establishment of a biofilm on plant surfaces (20). In other clonal bacterial pathogens, such as Yersinia pestis, a modest number of genetic changes have led to dramatically different routes of transmission and modes of pathogenesis compared with their progenitors. Indeed, despite their radically different disease pathologies, there are many parallels between Y. pestis and M. ulcerans; in the case of the agent of plague, acquisition of the plasmid-encoded genes ymt and hms has conferred the ability to resist digestion in the midgut of fleas and the ability to persist on the surface of spines that line the interior of the proventriculus, respectively, thus facilitating an arthropod-linked mode of transmission (14, 17). Determining the role(s) of mycolactone and its mode(s) of action is now the subject of investigations in our laboratory.

While the repetitive nature of the mls locus has not yet led to heterogeneity among mycolactones, one DNA deletion identified in this study can be linked with the production of variant toxin. The plasmid gene MUP053 encodes a putative P450 monooxygenase, an enzyme thought to be required for hydroxylation of mycolactone at position C-12′ of its fatty acid side chain to produce mycolactone A/B (m/z 765). As predicted, the Australian strain M. ulcerans Chant lacks MUP053 and produces a lower-mass metabolite at m/z 749 (mycolactone C) that corresponds with the absence of a hydroxyl group. The fact that M. ulcerans ITM-941331 from Papua New Guinea also lacks MUP053 but still produces oxidized mycolactones suggests that in some strains there may be chromosomal P450 genes encoding hydroxylases that are active against the molecule.

This study showed that there is considerable mutational dynamism in pMUM. It may be that there is constant genetic flux within the Mls genes such that new mycolactones are continuously being created within a given M. ulcerans population. However, if new metabolites do not confer a fitness advantage, then cells with such changes do not persist. Nevertheless, it seems that there is potential to discover further structural heterogeneity among mycolactones by screening a wider selection of isolates, as demonstrated by the recent discovery of a new mycolactone produced by an M. ulcerans strain from the People's Republic of China (H. Hong et al., Chem. Biochem., in press).


We gratefully acknowledge the financial support of the Génopole program, the World Health Organization, the Association Française Raoul Follereau, and the National Health and Medical Research Council, Australia, Follereau and the award of a Chaire Internationale de Recherche Blaise Pascal (to P.F.L.).


1. Amofah, G., F. Bonsu, C. Tetteh, J. Okrah, K. Asamoa, K. Asiedu, and J. Addy. 2002. Buruli ulcer in Ghana: results of a national case search. Emerg. Infect. Dis. 8:167-170. [PMC free article] [PubMed]
2. Bonfield, J. K., K. F. Smith, and R. Staden. 1995. A new DNA sequence assembly program. Nucleic Acids Res. 24:4992-4999. [PMC free article] [PubMed]
3. Broadhurst, R. W., D. Nietlispach, M. P. Wheatcroft, P. F. Leadlay, and K. J. Weissman. 2003. The structure of docking domains in modular polyketide synthases. Chem. Biol. 10:723-731. [PubMed]
4. Brosch, R., S. V. Gordon, A. Billault, T. Garnier, K. Eiglmeier, C. Soravito, B. G. Barrell, and S. Cole. 1998. Use of a Mycobacterium tuberculosis H37Rv bacterial artificial chromosome library for genome mapping, sequencing, and comparative genomics. Infect. Immun. 66:2221-2229. [PMC free article] [PubMed]
5. Brosch, R., A. S. Pym, S. V. Gordon, and S. T. Cole. 2001. The evolution of mycobacterial pathogenicity: clues from comparative genomics. Trends Microbiol. 9:452-458. [PubMed]
6. Chemlal, K., K. De Ridder, P. A. Fonteyne, W. M. Meyers, J. Swings, and F. Portaels. 2001. The use of IS2404 restriction fragment length polymorphisms suggests the diversity of Mycobacterium ulcerans from different geographical areas. Am. J. Trop. Med. Hyg. 64:270-273. [PubMed]
7. da Silva, J., and A. L. Hughes. 1998. dSdNqw, 1.0 ed. Pennsylvania State University, University Park.
8. Debacker, M., J. Aguiar, C. Steunou, C. Zinsou, W. M. Meyers, A. Guedenon, J. T. Scott, M. Dramaix, and F. Portaels. 2004. Mycobacterium ulcerans disease (Buruli ulcer) in rural hospital, Southern Benin, 1997-2001. Emerg. Infect. Dis. 10:1391-1398. [PMC free article] [PubMed]
9. Dobrindt, U., B. Hochhut, U. Hentschel, and J. Hacker. 2004. Genomic islands in pathogenic and environmental microorganisms. Nat. Rev. Microbiol. 2:414-424. [PubMed]
10. Fidanze, S., F. Song, M. Szlosek-Pinaud, P. L. Small, and Y. Kishi. 2001. Complete structure of the mycolactones. J. Am. Chem. Soc. 123:10117-10118. [PubMed]
11. George, K. M., L. P. Barker, D. M. Welty, and P. L. Small. 1998. Partial purification and characterization of biological effects of a lipid toxin produced by Mycobacterium ulcerans. Infect. Immun. 66:587-593. [PMC free article] [PubMed]
12. George, K. M., D. Chatterjee, G. Gunawardana, D. Welty, J. Hayman, R. Lee, and P. L. Small. 1999. Mycolactone: a polyketide toxin from Mycobacterium ulcerans required for virulence. Science 283:854-857. [PubMed]
13. George, K. M., L. Pascopella, D. M. Welty, and P. L. Small. 2000. A Mycobacterium ulcerans toxin, mycolactone, causes apoptosis in guinea pig ulcers and tissue culture cells. Infect. Immun. 68:877-883. [PMC free article] [PubMed]
14. Hinnebusch, B. J., A. E. Rudolph, P. Cherepanov, J. E. Dixon, T. G. Schwan, and A. Forsberg. 2002. Role of Yersinia murine toxin in survival of Yersinia pestis in the midgut of the flea vector. Science 296:733-735. [PubMed]
15. Hong, H., P. J. Gates, J. Staunton, T. Stinear, S. T. Cole, P. F. Leadlay, and J. B. Spencer. 2003. Identification using LC-MSn of co-metabolites in the biosynthesis of the polyketide toxin mycolactone by a clinical isolate of Mycobacterium ulcerans. Chem. Commun. 22:2822-2823. [PubMed]
16. Hughes, A. L., R. Friedman, and M. Murray. 2002. Genomewide pattern of synonymous nucleotide substitution in two complete genomes of Mycobacterium tuberculosis. Emerg. Infect. Dis. 8:1342-1346. [PMC free article] [PubMed]
17. Jarrett, C. O., E. Deak, K. E. Isherwood, P. C. Oyston, E. R. Fischer, A. R. Whitney, S. D. Kobayashi, F. R. DeLeo, and B. J. Hinnebusch. 2004. Transmission of Yersinia pestis from an infectious biofilm in the flea vector. J. Infect. Dis. 190:783-792. [PubMed]
18. Katz, L., and S. Donadio. 1993. Polyketide synthesis: prospects for hybrid antibiotics. Annu. Rev. Microbiol. 47:875-912. [PubMed]
19. Kumar, S., K. Tamura, I. B. Jakobsen, and M. Nei. 2001. MEGA2: molecular evolutionary genetics analysis software. Bioinformatics 17:1244-1245. [PubMed]
20. Marsollier, L., R. Robert, J. Aubry, J. P. Saint Andre, H. Kouakou, P. Legras, A. L. Manceau, C. Mahaza, and B. Carbonnelle. 2002. Aquatic insects as a vector for Mycobacterium ulcerans. Appl. Environ. Microbiol. 68:4623-4628. [PMC free article] [PubMed]
21. Marsollier, L., T. Stinear, J. Aubry, J. P. Saint Andre, R. Robert, P. Legras, A. L. Manceau, C. Audrain, S. Bourdon, H. Kouakou, and B. Carbonnelle. 2004. Aquatic plants stimulate the growth of and biofilm formation by Mycobacterium ulcerans in axenic culture and harbor these bacteria in the environment. Appl. Environ. Microbiol. 70:1097-1103. [PMC free article] [PubMed]
22. Mve-Obiang, A., R. E. Lee, F. Portaels, and P. L. Small. 2003. Heterogeneity of mycolactones produced by clinical isolates of Mycobacterium ulcerans: implications for virulence. Infect. Immun. 71:774-783. [PMC free article] [PubMed]
23. Nei, M., and T. Gojobori. 1986. Simple methods for estimating the numbers of synonymous and nonsynonymous nucleotide substitutions. Mol. Biol. Evol. 3:418-426. [PubMed]
24. Nei, M., and L. Jin. 1989. Variances of the average numbers of nucleotide substitutions within and between populations. Mol. Biol. Evol. 6:290-300. [PubMed]
25. Pahlevan, A. A., D. J. Wright, C. Andrews, K. M. George, P. L. Small, and B. M. Foxwell. 1999. The inhibitory action of Mycobacterium ulcerans soluble factor on monocyte/T cell cytokine production and NF-kappa B function. J. Immunol. 163:3928-3935. [PubMed]
26. Rutherford, K., J. Parkhill, J. Crook, T. Horsnell, P. Rice, M. A. Rajandream, and B. Barrell. 2000. Artemis: sequence visualization and annotation. Bioinformatics 16:944-945. [PubMed]
27. Staunton, J., and K. J. Weissman. 2001. Polyketide biosynthesis: a millennium review. Nat. Prod. Rep. 18:380-416. [PubMed]
28. Stinear, T., J. K. Davies, G. A. Jenkin, F. Portaels, B. C. Ross, F. Oppedisano, M. Purcell, J. A. Hayman, and P. D. R. Johnson. 2000. A simple PCR method for rapid genotype analysis of Mycobacterium ulcerans. J. Clin. Microbiol. 38:1482-1487. [PMC free article] [PubMed]
29. Stinear, T. P., G. A. Jenkin, P. D. R. Johnson, and J. K. Davies. 2000. Comparative genetic analysis of Mycobacterium ulcerans and Mycobacterium marinum reveals evidence of recent divergence. J. Bacteriol. 182:6322-6330. [PMC free article] [PubMed]
30. Stinear, T. P., A. Mve-Obiang, P. L. Small, W. Frigui, M. J. Pryor, R. Brosch, G. A. Jenkin, P. D. Johnson, J. K. Davies, R. E. Lee, S. Adusumilli, T. Garnier, S. F. Haydock, P. F. Leadlay, and S. T. Cole. 2004. Giant plasmid-encoded polyketide synthases produce the macrolide toxin of Mycobacterium ulcerans. Proc. Natl. Acad. Sci. USA 101:1345-1349. [PMC free article] [PubMed]
31. Thompson, J. D., D. G. Higgins, and T. J. Gibson. 1994. CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Res. 22:4673-4680. [PMC free article] [PubMed]
32. van der Werf, T. S., T. Stinear, Y. Stienstra, W. T. van der Graaf, and P. L. Small. 2003. Mycolactones and Mycobacterium ulcerans disease. Lancet 362:1062-1064. [PubMed]

Articles from Journal of Bacteriology are provided here courtesy of American Society for Microbiology (ASM)
PubReader format: click here to try