Investigating the Role of Free-living Amoebae as a Reservoir for Mycobacterium ulcerans

Background The reservoir and mode of transmission of Mycobacterium ulcerans, the causative agent of Buruli ulcer, still remain a mystery. It has been suggested that M. ulcerans persists with difficulty as a free-living organism due to its natural fragility and inability to withstand exposure to direct sunlight, and thus probably persists within a protective host environment. Methodology/Principal Findings We investigated the role of free-living amoebae as a reservoir of M. ulcerans by screening the bacterium in free-living amoebae (FLA) cultures isolated from environmental specimens using real-time PCR. We also followed the survival of M. ulcerans expressing green fluorescence protein (GFP) in Acanthameoba castellanii by flow cytometry and observed the infected cells using confocal and transmission electron microscopy for four weeks in vitro. IS2404 was detected by quantitative PCR in 4.64% of FLA cultures isolated from water, biofilms, detritus and aerosols. While we could not isolate M. ulcerans, 23 other species of mycobacteria were cultivated from inside FLA and/or other phagocytic microorganisms. Laboratory experiments with GFP-expressing M. ulcerans in A. castellani trophozoites for 28 days indicated the bacteria did not replicate inside amoebae, but they could remain viable at low levels in cysts. Transmission electron microscopy of infected A. castellani confirmed the presence of bacteria within both trophozoite vacuoles and cysts. There was no correlation of BU notification rate with detection of the IS2404 in FLA (r = 0.07, n = 539, p = 0.127). Conclusion/Significance This study shows that FLA in the environment are positive for the M. ulcerans insertion sequence IS2404. However, the detection frequency and signal strength of IS2404 positive amoabae was low and no link with the occurrence of BU was observed. We conclude that FLA may host M. ulcerans at low levels in the environment without being directly involved in the transmission to humans.


Introduction
Mycobacterium ulcerans is a slow growing environmental pathogen responsible for a necrotizing cutaneous infection called Buruli ulcer (BU). The disease has been reported in over 30 countries worldwide mainly in tropical and subtropical climates and emerged as an increasing cause of morbidity in endemic rural communities in some West and Central African countries with Benin, Côte d'Ivoire and Ghana bearing the highest burden of disease [1].
Most BU endemic areas are found close to slow flowing or stagnant water bodies and it is therefore assumed that the aquatic ecosystem may be a source of M. ulcerans from which the bacterium is transmitted to humans. This is supported by several studies that have detected M. ulcerans DNA sequences in a variety of environmental specimens including fish, snails, detritus, biofilms, soil, water filtrands, insects and protozoa [2][3][4][5]. Recently in Australia, M. ulcerans DNA has been detected in mosquitoes, faecal matter and skin lesions of small terrestrial mammals (ringtail and brushtail possums) that are thought to harbor and vector the bacterium [6,7]. However, the main reservoir and modes of transmission of BU outside Australia still remain unknown.
Since the discovery that Legionella pneumophila is able to infect and replicate in free-living amoebae (FLA) [8], there has been an increasing number of studies on the role of FLA in the survival of pathogenic organisms [9]. Also, several species of mycobacteria (M. shottsii, M. pseudoshottsii, M. tuberculosis, M. leprae, M. ulcerans, M. marinum, M. bovis, M. avium subsp paratuberculosis and M. avium) have been shown to survive within protozoa [4,[10][11][12][13][14][15][16]. M. ulcerans bears characteristic genomic signatures that are typical of host restricted pathogens suggesting that M. ulcerans is unlikely to be free-living in the environment but is instead undergoing or has undergone adaptation to a specific ecological niche [17]. Internalization of infectious agents inside other parasites is a recurring theme in biology and represents an evolutionary strategy for survival that may sometimes enhance pathogenesis or transmissibility [18]: Bacteria ''hidden'' in their protozoan hosts may more easily infect vertebrate end hosts, multiplying within protozoans to escape immune reactions [15,18].
Water bodies in areas of high BU endemicity have been reported to contain significantly more FLA than in low endemic areas [19]. Recently, we demonstrated that M. ulcerans can be phagocytosed in vitro by Acanthamoeba polyphaga and persist for at least 2 weeks [4]. This study also showed a higher detection frequency of the IS2404 target in FLA cultures as compared to crude samples from the environment. The aim of the present study was to further explore FLA as a reservoir for M. ulcerans by screening M. ulcerans in FLA from aquatic environment sampled for 10 months and relating this to the BU notification rate in the same endemic area. Furthermore, we experimentally investigated the ability of M. ulcerans to survive and replicate within A. castellanii by infecting these amoebae with M. ulcerans expressing green fluorescence protein (GFP).

Study sites and specimen collection
The study was carried out in five endemic communities (with recorded human BU cases): Ananekrom, Nshyieso, Serebouso, Dukusen and Bebuso, and two non-endemic communities (no recorded human BU cases): Mageda and Pataban in the Asante Akim North Municipal of Ghana (Table 1, Fig. 1). These communities are on average 18 km apart and were selected based on number of BU cases reported at the Agogo Presbyterian Hospital (APH), the Municipal health facility serving all communities ( Fig. 1). Week-long monthly field visits were made for 10 months between October 2008 and July 2009 to randomly collect environmental specimens: water, biofilm from plants and tree trunks, detritus and aerosols. The specimens were taken between 6:00 am and 8:00 am, the peak period of human contact activities in the water bodies. Biofilms (n = 428) were taken by scraping surfaces of tree trunks, floating logs and tree stumps with sterile scalpels and cotton swabs into 50 mL sterile Falcon tubes. Detritus (n = 45) were scooped by hand into 50 mL sterile Falcon tubes. Water specimens (n = 53) were taken from mid column with buckets of which 3-10 liters were concentrated via 0.45 mm membrane filters (Sartorius Stedim Biotech GmbH, Germany) depending on the turbidity of the water. During the last two months of sampling, non-nutrient agar (NNA) plates (n = 13) seeded with Escherichia coli were exposed for 30 minutes for isolation of FLA from aerosols generated next to the water bodies.

Specimen processing
Seven milliliters phosphate buffered saline (PBS) were added to the membrane filters (cut into smaller pieces), swabs and scalpels contained in 50 mL Falcon tubes and shaken vigorously to dislodge the substrate and biofilms from the surface. For detritus, specimens were processed as described by Gryseels et al. [4].

Isolation of FLA
A piece of the membrane filter and 2 to 3 drops of the suspensions (biofilms and detritus) were inoculated at the centre of 1.5% NNA plates seeded with E. coli for cultivation of FLA [20] at 28.5uC. The inoculated NNA plates were examined daily for the presence of trophozoites and cysts using the 106 objective of a bright field microscope. When FLA were observed, they were subcultured on new NNA plates seeded with E. coli [21]. After 3 or 4 subcultures, the FLA were harvested by scraping the surface with an inoculating loop and suspending them in 1.5 mL sterile distilled water.

DNA preparation
The modified Boom method was used for the extraction of DNA from FLA cultures as described previously [22,23].

Detection of M. ulcerans DNA
Two multiplex real-time PCR assays were performed on the DNA extracts to test the presence of three distinct sequences: IS2404, IS2606 and the ketoreductase B (KRB) domain in the M. ulcerans genome as described by Fyfe et al. [24]. All DNA extracts were first screened for the IS2404 target multiplexed with an internal positive control to check for PCR inhibitors such as humic and fulvic acids (commonly found in environmental specimens) [24]. The second PCR assay for detecting the presence of IS2606 and KRB was done on FLA cultures that turned out positive for the IS2404 target. Amplification and detection was carried out using the 7500 real-time PCR system (Applied Biosystems).

Identification of FLA
The identification of FLA of the genera Acanthamoeba, Naegleria and Vahlkampfiidae was confirmed using the primer sets JDP1/JDP2, ITSfw/ITSrv and JITSfw/JITSrv respectively as described by Gryseels et al. [4] and Eddyani et al. [19].

Author Summary
Mycobacterium ulcerans, the causative agent of Buruli ulcer (BU) is an environmental pathogen known to reside in aquatic habitat. However, the reservoir and modes of transmission to humans still remain unknown. M. ulcerans can probably not live freely due to its natural fragility and inability to withstand exposure to direct sunlight. This study investigated the hypothesis that free-living amoebae (FLA) can serve as a reservoir of M. ulcerans by testing for its presence in amoebae isolated from water bodies in BU endemic and non-endemic communities and whether the pathogen can remain viable when experimentally infected in amoebae in the laboratory. We detected only one (IS2404) of the three (IS2606 and KRB) targets for the presence of M. ulcerans in amoebae cultures and found no correlation between its presence in the environment and BU notification rate. M. ulcerans remained viable at low levels in amoebae for 28 days in vitro. We therefore conclude that FLA may host M. ulcerans at low levels in the environment without being directly involved in the transmission to humans.

Isolation of M. ulcerans and other intracellular mycobacteria from specimens
Specimens were processed to kill extracellular mycobacteria as described by Gryseels et al. [4], decontaminated using the oxalic acid method [25], inoculated on LJ medium, and incubated at 30uC with weekly examination for a year. Supernatants containing extracellular mycobacteria were cultured for week at the same temperature to monitor the effectiveness of the isolation method.

Identification of intracellular mycobacteria
Direct smear examination (DSE) by ZN staining for acid fast bacilli (AFB) was performed on single colonies grown on LJ media. Cultures positive for AFB were subjected to a nested PCR specific for the mycobacterial 16S rRNA gene [26,27] after heatinactivation of mycobacterial suspensions in TE for 10 minutes. Amplicons were sequenced by the VIB Genetic Service Facility (Antwerp, Belgium). The sequenced data were compared with known sequences in the GenBank database and interpreted using the BlastN algorithm (available on http://www.ncbi.nlm.nih.gov/ BLAST/). The 16S rRNA sequences were also matched against entries in the RIDOM (Ribosomal Differentiation of Medical Microorganisms) database (http://www.ridom-rdna.de/). Direct detection of mycobacterial DNA in FLA cultures was done using the same 16S rRNA PCR assay.
Infection of A. castellanii with M. ulcerans M. ulcerans strains JKD8083 (which expresses GFP) and 04126204 (which does not express GFP) were grown in 7H9 broth or 7H10 agar supplemented with OADC (Difco). Real-time PCR was performed on M. ulcerans strains to estimate cell numbers as previously described [28]. Primers targeting the 16S rRNA gene were used for detection of mycobacteria. Known amounts of M. ulcerans Agy99 genomic DNA were used to construct a standard curve and cell numbers were estimated based on the predicted mass of an M. ulcerans chromosome [24].
A. castellanii was cultured in peptone-yeast extract-glucose (PYG) medium at 22uC in the dark as described by Moffat and Tompkins [29]. Trophozoites were harvested at 4006 g for 10 minutes (Eppendorf, 5810R) and adjusted to a final concentration of 10 6 cells ml 21 in Acanthamoebae (AC) buffer as described [29].
Bacterial strains were concentrated by centrifugation at 60006 g for 15 minutes at room temperature and then resuspended in AC buffer. A preparation of 1610 6 cells of A. castellanii was mixed with 1610 6 cells of either M. ulcerans (JKD8083) or (04126204) in 20 mL PYG broth and incubated at 22uC for 30 minutes. Co-cultures were washed three times in AC buffer and treated with amikacin (150 mg/ml) as described [30] to kill extracellular bacteria. Trophozoites were then washed and resuspended in 50 mL AC buffer. Three milliliters samples were taken at 1, 2, 7, 14, 21 and 28 days for analysis.

Flow cytometry
At each time point, samples were washed with FACS buffer six times before a final resuspension in 500 ml of FACS buffer. FACS was carried out using uninfected amoebae to identify trophozoite populations. The subsequent infected samples were gated only on these relevant populations. All samples, including the controls (uninfected amoebae and bacteria only) were analyzed with a FACS (Becton Dickinson) equipped with a 488 nm argon laser. At each time point 50,000 events were counted. Background fluorescence was determined by using infections of A. castellanii with non-fluorescent M. ulcerans (04126204). Percentages of fluorescing amoebae were then calculated using Flowjo (v8.7).

Confocal and electron microscopy
Aliquots of infected amoebae in AC buffer were again pelleted and washed in 16 PBS before DAPI staining according to the manufacturer's instructions (Invitrogen). Samples were imaged using a LAS700 confocal microscope (Zeiss) with a 1006 oil immersion lens.
At different time points 1 mL aliquots were used to examine for M. ulcerans within A. castelanii as described previously [31] using the electron microscope.

Statistical analysis
Statistical analyses were performed in SPSS 18.0 (SPSS Inc., Chicago, IL) software. Standard multiple regression was used to investigate whether the isolation frequency of FLA in communities (water bodies) was related to the waterbody-specific prevalence of Table 1. Information on sampling sites. the IS2404 target and mycobacterial DNA (in those FLA). Hierarchical multiple regression was also used to assess the ability of some parameters (detection of IS2404 positive FLA, detection of mycobacterial DNA in FLA, and isolation of intracellular mycobacteria) to predict BU notifications (number of reported cases/number of inhabitants/month), after controlling for the influence of time. Logistic regression was performed to assess the influence of time (months) on the detection of IS2404 in FLA, isolation of intracellular mycobacteria and frequency of isolated FLA. The relationship between the isolation of FLA and the type of specimen and the communities sampled was investigated using the Pearson product-moment and Spearman Rank Order correlation (rho) coefficient. Kruskal-Wallis and Mann-Whitney U Tests were used to compare the isolation frequency of FLA and detection frequency of IS2404 between the different types of specimen and communities. P values ,0.05 were considered significant.  (Table S2). FLA were isolated from all specimen types (Table 2), and showed a statistically significant difference across the specimen types, (x 2 (4, n = 539) = 14.532, p = 0.006) with aerosols recording the highest mean rank (354.50) compared to the other specimen types. A Mann-Whitney U Test also showed that FLA were more frequently isolated from aerosols than plant biofilm (p = 0.006) (Bonferroni adjustment alpha level = 0.008). FLA were isolated from all communities, the isolation frequency showed a significant difference across communities ((x 2 (6, n = 539) = 14.955, p = 0.021) with Nshyieso recording the highest mean rank (308.22). A Mann-Whitney U Test showed no significant difference in FLA isolation between Mageda and Nshyieso (p = 0.878) but showed a difference between Nshyieso and the communities Dukusen, Ananekrom and Serebouso (p = 0.000, p = 0.002, p = 0.006).

Identification of intracellular mycobacteria
We could not cultivate M. ulcerans from the FLA, however, other mycobacteria were cultured from intracellular origins in 162 (30.06%) specimens; 109 (67.28%) among these originated from specimens from which we also isolated FLA. All isolates were confirmed by DSE for AFB and partial sequencing of the 16S rRNA gene. There was no growth of bacteria after the supernatants containing the killed extracellular mycobacteria were cultured for a week. One hundred and thirty one isolates showed .99% sequence similarity match with the available database in GenBank (NCBI and RIDOM 16S rDNA) comprising a total of

Evolution of study parameters during the sampling period
The detection of intracellular mycobacteria peaked in April 2009 followed by a peak in the detection of IS2404 positive FLA in June 2009 and the isolation of FLA in July 2009. The highest number of BU cases was however reported four months later in November 2009 after FLA isolation peaked (Figure 2).
Time accounted for a 13.80% variance in the BU notification rate (new BU cases per month) using a hierarchical multiple regression model (F (4, 534) = 26.00, p,0.001). The three other parameters, intracellular mycobacteria, detection of IS2404 target and detection of mycobacterial DNA in FLA, together explained an additional 2.5% of the variance in BU notification rate, after controlling for time, R squared change = 0.025, F change (3, 534) = 5.251, p,0.001. In the final model, only two parameters were statistically significant, with time recording a higher beta value (beta = 0.312, p,0.001) than detection of mycobacterial DNA in FLA (beta = 0.152, p,0.001). The isolation frequency of FLA varied significantly through time ((x 2 (1, N = 539) = 28.479, p,0.001) as well. There was a positive correlation of BU notification rate with detection of mycobacterial DNA in FLA (r = 0.27, n = 539, p,0.0005) but not with detection of the IS2404 target in FLA (r = 0.07, n = 539, p = 0.127).
M. ulcerans persists in amoebae for up to 28 days M. ulcerans infections of A. castellanii were performed over a four week period and quantified using flow cytometry. Methods involving the removal of extracellular bacteria using amikacin have been reported previously [4,30] and were independently tested here ( Figure 3A). M. ulcerans JKD8083 bacteria alone (10 6 -10 8 ) were treated with amikacin for 7 days to test the effect of the antibiotic on extracellular bacteria. Treatment with 150 mg/ml amikacin for 7 days left 18.7% of 10 8 extracellular bacteria (5.92% of 10 7 , 1.53% of 10 6 ) fluorescing above background ( Figure 3A).
Using flow cytometry, 50,000 events were counted at each time point. Amoebae were experimentally infected with M. ulcerans by placing them in co-culture at a multiplicity of infection of 1 for 30 min, and killing remaining extracellular mycobacteria with amikacin. At day 0, 9.51% of the amoebae were infected with M. ulcerans, but this percentage gradually decreased until at day 28 (end of experiment) only 0.7% of amoebae were infected with M. ulcerans ( Figure 3B). Figure 3 (D-F) shows gated trophozoites following 7 days co-incubation of amoebae alone (D), M. ulcerans JKD8083 (E) and M. ulcerans 04126204 (F). Amoebae infected with non-fluorescing bacteria however, fluoresced at levels less than 0.05% over the 28-day time course (Figure 3B), therefore we neglected this background fluorescence.

An intracellular location of M. ulcerans within A. castellanii
Localisation of the bacteria with respect to the amoebae was determined by both fluorescent confocal microscopy and electron microscopy ( Figure 3C; Figure 4). Confocal optical sections demonstrate a mycobacterium within A. castellanii three hours post infection ( Figure 3C). Video S1. shows fluorescing bacteria in

Discussion
It has been suggested that M. ulcerans persists with difficulty in the environment as a free-living organism due to its natural fragility [32] and may be maintained in a commensal or parasitic relationship with hosts that protect the bacilli against potentially unfavorable environments. This hypothesis is supported by the observation that M. ulcerans, unlike its close environmental relatives, has degraded genome, with many pseudogenes, such as a mutation in the crt locus. This locus harbors genes responsible for the production of light-inducible carotenoids that affects its ability to withstand exposure to direct sunlight and hence diminishes its capacity to live freely [17].
A previous study by our group detected the IS2404 target twice as often in FLA cultures as in environmental specimens [4]. Extending this study, we investigated the role of FLA and other phagocytic microorganisms as a reservoir of M. ulcerans in the aquatic environment for 10 months and tested the survival of M. ulcerans within A. castellanii in vitro.
Acanthamoeba sp. and Vahlkampfiidae sp. were isolated more frequently than Naegleria from the sampled specimens. FLA were isolated more frequently from aerosols and detritus than from biofilms. Acanthamoeba has been implicated in a number of diseases including granulomatous amoebic encephalitis [33], a cerebral abscess [34] and chronic keratitis [35]. We also isolated potentially pathogenic V. avara, N. canariensis, N. philippinensis and A. lenticulata.
In this study, we detected the M. ulcerans IS2404 target in 4.64% of FLA cultures, significantly more often in Acanthamoeba sp. and Vahlkampfiidae sp. as has been reported by Gryseels et al. [4]. Our inability to detect the targets IS2606 and KR-B in IS2404 positive FLA was not surprising since most of the C T values of the IS2404 target recorded were indicative of low mycobacterial DNA concentrations as observed in other studies carried out in the same sampling sites [4,36].
The importance of protozoa harboring human-pathogenic bacteria has recently been given much attention, especially in the case of fragile bacteria whose environmental phase would be difficult without the protection of a protozoan host [37]. Moreover, phagocytic protozoans such as FLA strongly resemble vertebrate macrophages; and it has been shown that infection success and internal proliferation is enhanced when bacteria such as Legionella and M. avium had previously resided inside protozoans [15,38]. A number of intracellular Mycobacterium sp. were isolated from unknown hosts in the specimens, which previously have been shown to live/survive intracellularly in amoebae: M. simiae, M. fortuitum, M. septicum, M. peregrinum, M. terrae, M. gordonae, M. intracellulare and M. lentiflavum [16,39]. The most frequently isolated species were: M. arupense, M. fortuitum and M. lentiflavum are potentially pathogenic species. These species were isolated from the environment [40][41][42] and M. arupense was isolated from wild African rodents [43]. Intracellular mycobacteria were more frequently isolated from specimens from which we also isolated FLA that may indicate their role as a reservoir for these mycobacteria. Thomas et al. [37] also reported a significant association between the presence of amoebae and the presence of mycobacteria. FLA have the additional advantage that they can form cysts, which allow them to persist through harsh periods and be dispersed via the air. It has been suggested that some infections can be acquired by inhaling aerosols containing FLA cells filled with bacteria [44], for example in the case of Legionella [45]. The detection of the IS2404 target in two of the thirteen aerosolized FLA suggests that they may act as vehicles for these mycobacteria. The aerosol transmission hypothesis of M. ulcerans was first postulated by Hayman [46], but received little attention, due to the unlikelihood of M. ulcerans being airborne as a free-living organism. The possibility that IS2404 positive mycobacteria including M. ulcerans are carried by aerosolized protozoan cysts changes this perspective. More research is, however, needed to explore this transmission route further, and in a subsequent study we investigated the presence of M. ulcerans on the skin of healthy inhabitants in the same endemic communities (manuscript in preparation).
BU notification rates varied significantly through time with the highest number of cases recorded in November 2009. BU prevalence has increased during the last quarter of the year in this locality as well as in some endemic regions of Ghana (data not shown); in this case there was no community awareness during this period. BU notification rates correlated positively with detection frequency of mycobacterial DNA in FLA cultures but not with detection of IS2404 in FLA cultures, suggesting that detection of IS2404 in FLA cannot predict concurrent BU incidence. However, the time series of this data set was not long enough to test for a potential lagging phase between peaks of IS2404 detection in FLA and BU incidence.
Similar to our previous study [4], over the time course of experimental infection of A. castellanii with M. ulcerans, M. ulcerans is present within amoebae for up to 28 days albeit at low levels. In addition, both electron microscopy and standard fluorescence microscopy revealed the presence of intracellular bacteria within cysts at 22 days post infection ( Figure 4I-J). This is not unexpected due to the previously demonstrated presence and survival of a variety of environmental mycobacteria in cysts [39].
The persistence of strong GFP fluorescence of M. ulcerans within A. castellanii throughout the experiment indicated that the mycolactone polyketide synthases genes are abundantly expressed intracellularly, as GFP gene expression is under the control of the mlsA1 promoter [26]. These data also suggest that mycolactone may be produced by the bacteria within the vacuole. While this study was not designed to test the effect of M. ulcerans on A. castellanii, our observations of fluorescing M. ulcerans persisting through 28 days within intact A. castellanii suggest that A. castellanii is not adversely affected by mycolactone or the presence of the bacteria as was also shown for A. polyphaga [4].
The ability of M. marinum to persist within amoebae is widely documented [39,47,48]. Following the initial time points a decrease in M. ulcerans-infected amoebae as reported previously for M. ulcerans, M. shottsii and M. pseudoshottsii [4,10] was seen which suggests that M. ulcerans does not replicate within amoebae and is not as well adapted as M. marinum to resist initial amoebic digestion, but is perhaps able to persist once within the vacuolar compartment by preventing lysosomal maturation of the vacuole by as yet undetermined mechanisms.
This study showed the occurrence of the IS2404 marker in FLA, especially in the genera Acanthamoeba and Vahlkampfiidae. After co-culturing amoebae and M. ulcerans the pathogen persisted at low levels suggesting that it probably only transiently occupies FLA and that it is unlikely that protozoa are a long-term reservoir for this pathogen. While the data we present here confirm that FLA can host mycobacteria that harbor the IS2404 marker (including M. ulcerans), the lack of predictive power of detection of IS2404 positive FLA in predicting BU notifications suggests FLA are not directly involved in transmission of M. ulcerans to humans. We suggest future work should focus on reservoirs that act as M. ulcerans amplifiers that link protozoans with humans.