High-Throughput Carbon Substrate Profiling of Mycobacterium ulcerans Suggests Potential Environmental Reservoirs

Background Mycobacterium ulcerans is a close derivative of Mycobacterium marinum and the agent of Buruli ulcer in some tropical countries. Epidemiological and environmental studies pointed towards stagnant water ecosystems as potential sources of M. ulcerans, yet the ultimate reservoirs remain elusive. We hypothesized that carbon substrate determination may help elucidating the spectrum of potential reservoirs. Methodology/Principal findings In a first step, high-throughput phenotype microarray Biolog was used to profile carbon substrates in one M. marinum and five M. ulcerans strains. A total of 131/190 (69%) carbon substrates were metabolized by at least one M. ulcerans strain, including 28/190 (15%) carbon substrates metabolized by all five M. ulcerans strains of which 21 substrates were also metabolized by M. marinum. In a second step, 131 carbon substrates were investigated, through a bibliographical search, for their known environmental sources including plants, fruits and vegetables, bacteria, algae, fungi, nematodes, mollusks, mammals, insects and the inanimate environment. This analysis yielded significant association of M. ulcerans with bacteria (p = 0.000), fungi (p = 0.001), algae (p = 0.003) and mollusks (p = 0.007). In a third step, the Medline database was cross-searched for bacteria, fungi, mollusks and algae as potential sources of carbon substrates metabolized by all tested M. ulcerans; it indicated that 57% of M. ulcerans substrates were associated with bacteria, 18% with alga, 11% with mollusks and 7% with fungi. Conclusions This first report of high-throughput carbon substrate utilization by M. ulcerans would help designing media to isolate and grow this pathogen. Furthermore, the presented data suggest that potential M. ulcerans environmental reservoirs might be related to micro-habitats where bacteria, fungi, algae and mollusks are abundant. This should be followed by targeted investigations in Buruli ulcer endemic regions.


Introduction
Mycobacterium ulcerans is the etiologic agent of Buruli ulcer, a disabling infection of the cutaneous and subcutaneous tissues [1][2][3]. M. ulcerans has been discovered in Bairnsdale, Australia, where Buruli ulcer was initially described [4,5]. Buruli ulcer is a World Health Organization notifiable infection and has been reported at least once by 33 countries located in the rural tropical regions of Africa and South America, in addition to Australia and Japan [6,7]. Over the past ten years, 83.6% (80.89-86.30) of cases were declared by eight West African countries [8]. In these highly endemic regions, the exact reservoirs of M. ulcerans remain elusive [6,[9][10][11]. However, epidemiological studies conducted in West African countries all indicated a significant association between the prevalence of Buruli ulcer and the contact of populations with stagnant water sources [12][13][14][15][16][17] through routine activities such as washing, swimming, fishing and farming [18,19]. A significant progress was recently made by narrowing the possible sources down to contacts with rice fields in Côte d'Ivoire which are sources of stagnant water [16,18,20,21]. Parallel environmental investigations of stagnant water [20,22], water insects [23][24][25], fishes [26,27] and aquatic mammals [12] showed the presence of PCRamplified M. ulcerans insertion sequences (IS) IS2404, IS2606 and KR-B gene. Furthermore, M. ulcerans partial DNA coding sequences were also recovered from the soil in the vicinity of stagnant water [20,22,26,28,29]. This finding was strengthened by an experimental study confirming a four-month survival of M. ulcerans in soil [30]. M. ulcerans DNA has been also detected in water plants [28,31] and in Thryonhuomys swinderianus (agouti), a small mammal causing damages to rice fields and in close contacts with rural populations in West Africa [20].
Moreover, this compelling amount of information concerning the presence of M. ulcerans DNA-related sequences found in the environment has been strengthened by the isolation of five wild strains from those sources [3,32,33].
Here, we propose that a characterization of the metabolic profile of M. ulcerans may give clues to better define its natural environment including its environmental reservoirs. In this perspective, we used the Biolog Phenotype MicroArray (Biolog Inc., Hayward, CA) for highthroughput carbon substrate profiling of M. ulcerans. Indeed, Biolog Phenotype MicroArray was previously used to classify and characterize heterotrophic microbial communities from different natural habitats according to their sole-carbon-source utilization profiles [34]. Accordingly, this approach previously unraveled the phenotypic patterns of some Mycobacterium tuberculosis complex mycobacteria [35] and Mycobacterium avium subsp. paratuberculosis [36]. It is used here in the context of unique carbon metabolisms such as chitinase exhibited by M. ulcerans [37].

M. ulcerans strains
This experimental study investigated M. ulcerans strain CU001 (a gift from Pr V. Jarlier, Paris, France), a clinical isolate representative of the West African epidemic, M. ulcerans ATCC 19423 isolated in Australia, M. ulcerans ATCC 33728 isolated in Japan, M. ulcerans ATCC 25900 isolated in the USA and Mycobacterium buruli ATCC 25894 isolated in Uganda [38]. These strains were manipulated into a BLS3 laboratory and a clinical isolate of Mycobacterium marinum was isolated in our laboratory [39]. All strains were cultured at 30˚C in Middlebrook 7H10 agar medium supplemented with 10% (v/v) oleic acid/albumin/dextrose/catalase (OADC) (Becton Dickinson, Sparks, MD, USA) and 0.5% (v/v) glycerol in a microaerophilic atmosphere for one week for M. marinum and four weeks for M. ulcerans.

Biolog Phenotype microarray
The Biolog Phenotype MicroArray (Biolog Inc.), which consists of 96-well microtiter plates containing each a defined medium that incorporates a unique carbon source (plates PM1 and PM2A for 190 different carbon sources) plus a dye indicator of cell respiration was used, according to the previously reported standard Biolog Inc. protocol [40,41]. M. ulcerans and M. marinum colonies were removed from Middlebrook 7H10 medium using a cotton swab previously dipped in 0.1% Tween 80 (WGK Germany, Sigma Aldrich). Mycobacteria were taken with the wet swab off the agar plate culture by gently sweeping on the surface of the culture and then rubbed against the wall of a dry glass tube containing glass beads. The cells were then suspended in GN/GP-IF-0a (Biolog inoculating fluid n˚133), the suspension was vigorously vortexed, passed three times through a 29-gauge needle in order to separate aggregates and adjusted to 81% transmittance using a turbidimeter (Biolog Inc). The PM-additive solutions for each plate were prepared according to Table 1. The inoculating fluid (Table 2) consisted of 20 mL of IF-0a GN/GP (1.2 x), 0.24 mL of dye mix G (100x) and 2.0 mL of PM additive (12x) added to the M. ulcerans or M. marinum suspension in IF-0a GN/GP (1.76 mL). Each PM plate was then inoculated in duplicate with 100 μL of inoculating fluid. The PM plates were incubated in the OmniLog PM System (Biolog Inc.) which measures the growth of mycobacteria every fifteen minutes for eight days at 30˚C. In each well the substrate was reduced to a purple color which was directly proportional to the growth of the mycobacteria. The intensity of the purple color was recorded as dye reduction value, which was then plotted as area under the curve (AUC) by Biolog's parametric software. Negative control wells containing non-inoculated additive solutions in each PM1 and PM2 plates were run at the same time as a quality control element. The threshold separating the wells which exhibited a positive reaction from those with a negative reaction was set for each plate according to the value of the area under the curve (AUC) of the negative control Well (NCW). We defined moderately positive growing wells (MPW) and highly positive growing wells (HPW) as follows: MPW is when the AUC value of the well is equal to or lower than 1.25 times the AUC value of the negative control well, and HPW is when the AUC value of the well is equal to or higher than 1.50 times the AUC value of the negative control. PM plates were further examined visually at the end of each incubation period to ensure an independent verification of the results.

Environmental sources of substrates metabolized by all tested M. ulcerans strains
In order to find the potential environmental origin of the carbon substrates metabolized by M. ulcerans, we used the PubMed database to obtain information on the environmental sources for each of the 190 carbon substrates present in the PM1and PM2 plates. The environmental sources were organized in 10 categories (plants, fruits and vegetables, bacteria, algae, fungi, nematodes, mollusks, mammals, insects and the inanimate environment). The Chi-square test was used to compare the proportion of each category for substrates not metabolized by M. ulcerans versus substrates metabolized by all tested M. ulcerans strains; a P value < 0.05 was used as the criterion for statistical significance. We then used the PubMed database to match each substrate, used as a key-word, with all environmental sources significantly associated with substrates metabolized by all tested M. ulcerans strains, used as the second key-word (e.g., D-glucosamine and fungi). We calculated the number of hits obtained in this research and compared it to the number of hits obtained by searching only for the key word corresponding to the environmental sources (e.g., fungi).

Environmental sources for substrates metabolized by all tested M. ulcerans strains
Comparing the potential environmental sources in search of substrates metabolized by all tested M. ulcerans strains versus non-metabolized substrates, we found a significant association between M. ulcerans metabolized substrates and bacteria (p = 0.000), fungi (p = 0.001), algae (p = 0.003) and mollusks (p = 0.007). The differences were not significant for plants (p = 0.535), fruits and vegetables (p = 0.870), mammals (p = 0.064), insects (p = 0.234) and the inanimate environment (p = 0.477). No carbon source was found to be associated with nematodes. Further MedLine research incorporating bacteria, fungi, algae and mollusks as keywords disclosed that 16/28 (57%) metabolized substrates were associated with bacteria, 5/28 (18%) were associated with alga, 3/28 (11%) were associated with mollusks and 2/28 with fungi. Discarding bacteria because of a potential bias since Biolog was designed for the study of bacterial metabolism, 15/28 (54%) metabolized substrates were associated with fungi whereas 6/28 (21%) were associated with the algae and 6/28 (21%) with mollusks (Table 4).

Discussion
We determined that five different strains of M. ulcerans could use 28 different substrates as sources of carbon. These results were authenticated by the negativity of the negative controls introduced in every plate and the reproduction of data over two replicates. Moreover, stringent criteria were used to ensure the predictive value of the positive results. However, only seven of these 28 substrates were found to be specifically used by M. ulcerans and not by the phylogenetically closest species M. marinum. Three of these seven carbon sources indeed contain indispensable amino-acids. The carbon sources here determined for M. ulcerans may by incorporated in culture media in the perspective of enhancing the isolation and culture of this pathogen. Indeed, M. ulcerans is a slow-growing mycobacterium and the availability of an improved method for its culture would improve the diagnosis of Buruli ulcer patients and the quest for environmental reservoirs [32]. As an example, it has been shown that the incorporation of chitin into the Middlebrook 7H9 broth enhances the growth of M. ulcerans [37]. Accordingly, our study points towards a possible association of M. ulcerans with fungi as a potential source of chitin, a polysaccharide possibly degraded by M. ulcerans' genome-encoded chitinase [42]. Likewise, the other carbon sources here disclosed should be tested for their potential to increase the cultivation of M. ulcerans.
Moreover, our analyses suggested that M. ulcerans may have found some sources of carbon in microbial communities including alive and dead bacteria, fungi and algae. As for bacteria, it has been previously reported that M. ulcerans was isolated in environments where 17 other mycobacteria species were also isolated, including M. fortuitum as a constant co-inhabitant [3,32,33]. These results suggest cross-feeding between various bacterial complexes including mycobacteria, for the acquisition of carbon. Likewise, green algae extracts have been shown to halve the in vitro doubling time of M. ulcerans and promote the formation of biofilm [31]. We observed that M. ulcerans metabolizes D-galacturonic acid, the main component of pectin contained in the primary cell walls of terrestrial plants, and putrescine, a foul-smelling chemical derived from the decomposition of dead plants, which indicates that M. ulcerans may live in assemblages of dead aquatic plants. This finding is reinforced by the observation that M. ulcerans's genome encodes five putative cutinases. Cutinases are mainly produced by phytopathogenic fungi to hydrolyze cutin (a main component of the cuticle which covers the aerial surfaces of plants) during plant colonization process [43].
Green algae are among the main food of freshwater mollusks pointed out in our study; mollusks are herbivores like other species of the freshwater snail family [44]. The principal genera of mollusks met in freshwater in West Africa are Bulinus, Planorbis, Pila, Lanistes, Melania, Bithynia, Lymnaea, Biomphalaria, Mutela, Aspatharia and Sphaerium [23,45]. Previous molecular investigations reported the detection of specific M. ulcerans DNA sequences in Bulinus spp. [23,46], in Planorbis spp. [23] and in mollusks of different Gastropoda order, Bivalvia order and Basommatophora order [26]. Furthermore, the experimental infection of Pomacea canaliculata (Ampullariidae) and Planorbis planorbis (Planorbidae) by plants contaminated by M. ulcerans-showed through optic microscopy digestive tract observation that snails remained infected by viable mycobacteria up to 25 days [23]. Small mollusks are also known to be a prey for water bugs which are involved in the transmission of M. ulcerans in Buruli ulcer endemic regions [3]. In West Africa, approximately 76% of the population lives next to rivers, lakes, and other water bodies contaminated with intermediate hosts such as snails [47].
In conclusion, our study is suggesting paths to improve culture media for the enhanced isolation of M. ulcerans by mimicking the natural ecosystem of M. ulcerans which is probably living in microbial communities with other bacteria, fungi and algae. These data support the recent hypothesis that mollusks could be part of a larger food chain including several hosts giving appropriate shelters to M. ulcerans, as recently reported [48]. Small mollusks should be further investigated using culture-based appropriate methods in the search for M. ulcerans.