Phenotypic and Genotypic Characterization of Mycobacteria Isolates from Buruli Ulcer Suspected Patients Reveals the Involvement of Several Mycobacteria in Chronic Skin Lesions

Buruli ulcer is a cutaneous mycobacterial disease that occurs in tropical countries in sub-Saharan Africa, South-East Asia, Australia and America. The responsible pathogen is Mycobacterium ulcerans. Côte d'Ivoire is the most affected country, with more than 30 endemic health districts reporting a large number of chronic skin lesions. The clinical forms and the severity of ulcers vary from one patient to another. Samples from suspected patients were analyzed by PCR at Pasteur Institute of Côte d'Ivoire, as recommended by WHO. IS2404 sequence was detected in 61% of cases, incriminating M. ulcerans in chronic cutaneous lesions. For the other cases the etiology was not identified, thus raising several questions. Are all reported "Buruli ulcer" cases really caused by M. ulcerans? Would other mycobacteria be involved in the occurrence of chronic skin lesions considered as "Buruli ulcer"? BU suspected patients were enrolled in endemic areas of Côte d’Ivoire. Samples were collected from cutaneous lesions and transported to the lab at +4°C in 2 ml of Middlebrook 7H9 medium supplemented by Cetylpiridium chloride. The centrifugation pellet was taken with saline buffer to perform microscopic examination and mycobacteria isolation on Lowenstein-Jensen medium. Biochemical characteristics were described by the nicotinic acid detection according to Konno protocol, by the Nitrates reduction test, by the catalase activity detection at 22° and 68°C and the Wayne's Tween 80 hydrolysis test. Genotypic characteristics were determined by PCR with 1 ml of bacterial suspension targeting the insertion sequences (IS6110, IS2404, and IS2606), the plasmid virulence genes and Miru-VNTR loci (Miru-1, VNTR 6, VNTR 19, ST-1). A total of 47 mycobacterial strains were isolated with 3 different types of colonies whose microscopic examination showed Acid-Alcohol-Resistant Bacilli. 65.9% of isolates expressed biochemical characters in favor of M. ulcerans strains and 6.4% in favor of M. marinum strains. For 29.8% of isolates, the characteristics were related to atypical mycobacterial species. The genotyping targeting the IS6110, IS2606 and IS2404 insertion sequences allowed simultaneous amplification in 53.2% of isolates. IS2404 was amplified in 93.6% of isolates; IS6110 was amplified in 74.5% of isolates and IS2606 was amplified in 70.2% of isolates. Five genotypes were identified corresponding to various species of mycobacteria: genotypes 1 and 2 accounted for 63.8% with all the 3 insertion sequences and biochemical characteristics in favor of M. ulcerans strains; genotype 4 accounted for 6.4% of isolates with insertion sequences and biochemical characteristics in favor of M. marinum strains; The strains of genotypes 3 and 5 expressed molecular and biochemical characters relating to various non-M. ulcerans mycobacteria. Virulence genes were found in 72.3% of isolates corresponding to 90% of M. ulcerans strains and 60.7% of non-M. ulcerans mycobacteria. This study confirmed the involvement of several genotypes of M. ulcerans and other mycobacteria in chronic cutaneous lesions suspected as cases of Buruli ulcer in


Introduction
Buruli ulcer (BU) is a cutaneous mycobacterial disease that occurs in thirty tropical countries, especially in sub-Saharan Africa [1,2,3,4,5], South-East Asia [6,7,8,9], Australia [10,11,12] and Latin America [13,14,15]. The responsible pathogen of the disease is Mycobacterium ulcerans, a slow-growing mycobacterium. Côte d'Ivoire is one of the most affected countries, with more than 30 endemic health districts whose have been reporting for many years a large number of chronic skin lesions attributed rightly or wrongly to M. ulcerans [4]. The evolution of M. ulcerans infection occurs in two successive stages: a pre-ulcerative phase marked by the appearance of a nodule, a plaque or an oedema. The pre-ulcerative forms and the severity of ulcers vary from one patient to another. Some patients have small lesions that are easy to cure while others have large ulcers for a long time. Biological confirmation is routinely performed by Polymerase Chain Reaction (PCR) methods targeting the IS2404 insertion sequence and ketoreductase (KR) in clinical samples, as recommended by WHO [16]. Several samples of BU suspected patients were analyzed by the National Buruli Ulcer Reference Center of Pasteur Institute of Côte d'Ivoire. For several years, the molecular diagnosis results incriminated at most M. ulcerans in 61% of those samples [4,17]. For the other cases (39%), the etiology was not identified, thus raising several questions. Are all reported BU cases really caused by M. ulcerans? Would other mycobacteria be involved in the occurrence of chronic skin lesions liked observed in BU? Would there be a genetic diversity of M. ulcerans strains in relation to the clinical polymorphism observed in endemic areas? These questions remained for a long time unanswered, prompted the study of several isolated strains of BU suspected patients in several endemic areas of Côte d'Ivoire. The aim of this work was to study the bacteriological, biochemical and molecular characteristics of isolated mycobacterial strains from cutaneous lesions considered as cases of M. ulcerans infection.

Patient's Enrollment and Mycobacterial Isolation
BU suspected cases (patients with nodule, plaque, oedema or ulcer) were enrolled in nine endemic areas of Côte d'Ivoire. Epidemiological and clinical informations were collected using UB 01, as recommended by WHO. Samples consisted of exudates or fine needle aspiration fluid [18] were discharged into a tube containing 2 ml of Middlebrook 7H9 medium supplemented by Cetylpiridium chloride (0.5%) and transported to the lab at +4°C. 10 ml of sterile water were added to the samples previously placed in a 15 ml Falcon tube. The mixture was centrifuged at 3000 rpm for 20 minutes The pellet was resuspended in 1 ml of saline buffer (NaCl, 9‰) to inoculate four tubes of Löwenstein-Jensen medium. The tubes were incubated at 32°C and a daily observation was instituted till the appearance of colonies [19].

Phenotypic and Molecular Characterization of Strains
Several tests have been performed to describe the phenotypic and genotypic characteristics of isolates. The cultural characteristics were described based on the timing of colony appearance, the pigment secretion, and the morphology of colonies. Microscopic examination to detect Acid-fast bacilli was performed by using dried smears made with 5 μl of bacterial suspension and staining by the Ziehl-Neelsen coloration [20,21]. Biochemical characteristics were described by the Nicotinic Acid detection according to Konno protocol (Niacin-test), by the Nitrates reduction test, by the catalase activity detection at 22° and 68°C, and the Wayne's Tween 80 hydrolysis test [1,21]. The genotypic characteristics were determined by molecular tests using 1 ml of bacterial suspension. The strains were first characterized using insertion sequences (IS6110, IS2404, IS2606) secondly the plasmid virulence factor coding gene (Ketoreductase, KR) and then a VNTR typing. The insertion sequence IS6110 is common to the genus Mycobacterium wilton et cousin 1992 [22] while the insertion sequences IS2404 and IS2606 [23] are carried by M. ulcerans's genome respectively in about 250 and 100 copies. The third tests targeted minisatellite markers: MIRU-VNTR (Mycobateria Interpersed Repeated Units-Variable Number of Tandem Repeated). The using method is as described by Stragier and several authors [24,25,26,27]. PCR were performed in a 9700 PCR system Thermocycler (Applied Biosystems, USA) [28,29,30,31,32]. The IS6110, IS2404, IS2606 and ketoreductase (KR) were amplified according to the following program: a pre-denaturation phase at 94°C for 2 minutes, followed by 35cycles consisted of a denaturation phase at 94°C for 1 minute, an annealing at 66°C for 1 minute and extension phase at 72°C for 1 minute. A final extension at 72°C for 5 minutes stopped the reaction. PCR products were revealed on a 1.5% gel electrophoresis containing Sybr Green. A 100 bp molecular weight marker was used to evaluate PCR products size. For Genotyping, the thermocycler (GeneAmp 9700) was programmed as following: an initial denaturation for 5 min at 95°C, followed by 40 cycles of 94°C for 1 min, 58°C for 1 min (MIRU-1, locus 6, VNTR 19) or 65°C (for ST-1) and 72°C for 1 min. DNA inhibition was tested by serial dilution of the DNA extract. Positive DNA samples of M. ulcerans and M. marinum from the Institute of Tropical Medicine were tested for each amplification.
The GoTaq G2 Flexi DNA polymerase kit (Promega Corporation, USA) was used for the PCR mixes containing 0.2μM of each primer, 1.5μM, MgCl, 0.1μM dNTPs, 1 unit Taq polymerase, 1X of buffer and 3μl of DNA template for a final volume of 50μl.The revelation was made on a GelDoc Bioanalyzer (BioRad) after electrophoresis.
The size of the amplicons was used to determine the copy number of the targeted sequence and the profile result was given in code form characterizing one type of polymorphism [25,26].
The data of these methods applied to the reference strains are recorded in Table 2.

Microbiological and Biochemical
Characteristics of Isolates 47 isolates were obtained on Löwenstein-Jensen medium from 9 endemic areas. Microscopic examination showed Acid-Alcohol-Resistant Bacilli. Three types of colonies were described among the isolates. The first type accounted for 65% of the isolates and consisted of firm, rounded, bulging, yellow-gold colonies. The colonies appeared after 3 to 6 weeks of incubation at 32°C ( Figure 1A). The second type accounted for 25% of the isolates and consisted of rough, irregular-shaped, cauliflower-like, whitish, brittlelike colonies. The colonies appeared between 1 and 3 weeks of incubation at 32°C ( Figure 1B). The third type of culture represented 10% of the isolates and associated rough, whitish, cauliflower-like colonies and rough yellow-gold colonies with curved surfaces. The colonies appeared after 3 weeks of incubation at 32°C ( Figure 1C). ulcerans Agnéby-Tiassa The rate of mycobacteria isolation on Löwenstein-Jensen medium was 64.4%. The majority of isolates (45.2%) were obtained with samples collected on pre-ulcerative lesions and ulcer of category 1. Mycobacterial isolation was lower in ulcers of categories 2 and 3, with rates of 15.1% and 4.1%, respectively. In 35.6% of samples from BU suspected patients, no mycobacteria were isolated after 12 weeks incubation at 32°C (Figure 2).
Three groups of isolates were identified based on biochemical characters detected. Most of isolates (31/47) expressed biochemical characters in favor of a M. ulcerans strain. In 6.4% of the isolates (3/47), the biochemical characteristics were in favor of a M. marinum strain. For 27.6% of the isolates (13/47), the biochemical characteristics were related to atypical mycobacterial species (Table 3).

Molecular Characteristics of the Isolates
Molecular characterization of mycobacteria with the inserted elements IS6110, IS2606 and IS2404 allowed simultaneous amplification in 53.2% of the isolates (Table 3). Specifically, IS2404 was amplified in 93.6% of the isolates (44/47); IS6110 was amplified in 74.5% of isolates (35/47) and IS2606 was amplified in 70.2% of isolates (33/47). Depending on the number of amplified insertion sequences and biochemical characters expressed, five different genotypes were identified corresponding to various species of mycobacteria ( Table 3). The strains of genotype 1 accounted for 53% of the isolates that possessed the three insertion sequences and biochemical characteristics in favor of M. ulcerans strains. Genotype 2 is composed of 10.6% of the isolates (5/47) with the IS2606 and IS2404 insertion sequences and expressed biochemical characteristics of M. ulcerans strains. Both genotypes 1 and 2 accounted for 63.8% of isolates (30/47) from suspicious lesions of Buruli ulcer. Genotype 4 composed of strains with sequences IS6110 and IS2606 accounted for 6.4% of the isolates. Their microbiological and biochemical characteristics were identical to those of M. marinum strains. Genotype 3 accounted for 14.9% of isolates (7/47) and had sequences IS6110 and IS2404. Genotype 5 also accounted for 14.9% of the isolates (7/47) and had only IS2404 sequence. However, both genotypes 3 and 5 expressed biochemical characters relating to various strains of mycobacteria. The association of microbiological, biochemical and genotypic characters allowed the identification of M. chelonae and M. smegmatis species isolated (Figure 3).   72.3% of the isolates were carried the KR gene coding for the plasmid of virulence. 27 out of 30 M. ulcerans strains (90%) were amplified the KR gene. 71.4% of isolates from genotype 3 and 50% isolates from genotype 5 possessed the virulence genes. But, the virulence factor genes were not found in the strains from genotype 4 ( Figure 4).
The rate of MIRU-VNTR loci amplification varied from one locus to another. ST-1marker was the most amplified locus by 80.8% of the isolates (38/47) with an amplification size of 340 base pairs (bp) corresponding to two replicates ( Figure 5D). VNTR 6 locus was amplified by 78.7% of the isolates (37/47) and by two reference controls at a length of 500 bp, corresponding to three replicates ( Figure 5B). In the third position, Miru-1 locus was amplified by 61.7% of isolates (29/47) and by all 3 controls, with 3 tandem repeats at a length of 500 bp ( Figure 5A). VNTR 19 was the least amplified locus by 51% of the isolates (24/47) according to the reference controls at a length of 340 bp, corresponding to one tandem repeat ( Figure 5C).
Seven VNTR profiles were identified among the mycobacteria isolates (

Discussion
The rate of bacterial isolation was different according to the stage of the lesions. It was better when lesions were early (45.2%) and the skin lesions were not subject to any antibiotic treatment. So the bacilli were more viable to grow easily on Löwenstein-Jensen medium. Conversely, the isolation rate was low when lesions were old, which had mostly undergone antibiotic treatments that destroyed or weakened the bacilli. In 2009, Ahoua and al. obtained an isolation rate of 22.1%. However, she didn't refer to either the stage of the lesions or the antibiotic treatments received by the patients [17].
8.5% of the strains were isolated from lesions after the end of antibiotic treatment according to WHO protocol. That treatment combining Rifampicin and Streptomycin was directly and daily administered by health workers. Mycobacteria isolation after 8 weeks treatment appeared abnormal and caused suspicions of strains that may have particular characteristics. Different phenotypes were identified among the isolates based on microbiological characters described. For the dominant phenotype (65.9%), bacteriological and biochemical characteristics were identical to those of a M. ulcerans strain. 6.4% of the isolates expressed bacteriological and biochemical characteristics in favor of a M. marinum strain. For 27.6% of the isolates, these characteristics were associated with atypical mycobacteria, such as M. chelonae, M. smegmatis ( Table 3).
The results of molecular typing targeting the insertion sequences IS6110, IS2606 and IS2404 showed differences between the isolates. Indeed, those targets were amplified in different proportions. IS2404 was amplified in the majority of the isolates (93.6%), followed by IS6110 (74.5%) and IS2606 (70.2%), respectively. Five genotypes were identified according to the number of amplified sequences. A typical M. ulcerans strain possesses both IS2606 and IS2404 insertion sequences, as described by Stinear in 1999 [23]. In this study, only the strains of genotype 1 (53.2%) and 2 (10.6%) had microbiological, biochemical and molecular characteristics identical to those of a M. ulcerans strain. Thus in our context, M. ulcerans was responsible for 63.8% of skin lesions treated as "Buruli ulcer". The strains of the genotype 4 had IS6110 and IS2606 sequences and expressed microbiological and biochemical characteristics identical to those of a M. marinum strain (6.4%). The strains of genotypes 3 and 5 (29.8%) expressed microbiological, biochemical and molecular characteristics relating to various atypical mycobacteria, such as M. smegmatis and M. chelonae ( Table 3). Because of the possession of IS2404 sequence, those pathogens were reported and treated as M. ulcerans strains. But they were not.
Geographical distribution of the isolates showed a different circulation of genotypes in the endemic areas of Côte d'Ivoire. M. ulcerans strains of genotype 1 were found everywhere patients were enrolled. In some localities of the Sud-Comoé region (Bonoua and Adiahou), the genotype 1 was the only one detected. Both localities are in agricultural areas drained by the Comoé River and numerous wetlands. M. ulcerans strains of genotype 2 were found in central (Gbêkê, Marahoué) and southern (Mé, Agnéby-Tiassa) regions of Côte d'Ivoire. The geographical distribution of genotypes 1 and 2 provided information on the circulation of M. ulcerans in endemic areas of Cote d'Ivoire. Other mycobacteria of genotypes 3 and 5 with IS2404 sequence were found in almost areas (7/9). Their presence could explain the high number of cases reported as "Buruli ulcer" in many localities, which may be mistaken for BU hyper-endemic areas. Some of those mycobacteria were isolated from patients after 8 weeks of well-conducted specific antibiotic therapy. This may explain the slow-healing lesions where Mycolactone and viable bacilli persisted after several weeks of antibiotic therapy [19,33,34].
The molecular typing targeting Miru-VNTR loci confirmed the variety of strains by showing the circulation of three VNTR profiles among M. ulcerans strains. Our results were similar to those found by Quinet and al. in 2017 [35]. The same loci were amplified in clinical samples and isolates from regions of Côte d'Ivoire in the same order of magnitude: ST-1 locus (57.4%), MIRU-1 locus (42.3%), VNTR 6 locus (30.8%) and VNTR 19 locus (23.1%). By contrast, Kakou [36] had amplified in 2015 those four loci but in proportions different from ours. In fact, she had found ST-1 in 85.7%, MIRU-1 and VNTR 19 in 61.9% and VNTR 6 in 52% of the samples tested, respectively. The MIRU-VNTR are variable repeat sequences with a new deletion or insertion in repetition loci. These different profiles would be evidence of a mutation in MIRU-VNTR loci in human strains. RFLP typing with better resolution showed differences between environmental mycobacterial strains and human M. ulcerans strains. Variations of MIRU-VNTR profiles could be a marker of adaptation of mycobacteria genotypes during the transition from the environment to humans. On the one hand, these studies confirmed a diversity of VNTR profiles between human M. ulcerans strains, and on the other hand between strains of atypical mycobacteria isolated from human skin lesions. 90% (27/30) of M. ulcerans strains expressed virulence genes. The lack of amplification of the virulence genes could result from the loss of the plasmid by strains under the isolation or conservation conditions in the lab. In contrast, virulence genes were amplified in 63.6% of non-M. ulcerans strains possessing IS2404 sequence. This sequence is also carried by the plasmid pMUM001 as described by Stinear [23,37,38]. Its presence as well as that of the virulence genes could translate an acquisition of the plasmid by other mycobacteria living in the same ecosystem as M. ulcerans. Our results corroborated those reported by Vakou et al. [39] which isolated environmental fast-growing mycobacteria with both IS2404 and ketoreductase sequences. Transfer of the plasmid from M. ulcerans to environmental mycobacteria could confer virulence or enhance pathogenicity leading to necrotic skin lesions observed in patients. Conversely, virulence genes were not detected in the M. marinum strains identified in patients.
Molecular diagnosis of M. ulcerans infection based on IS2404 sequence and Ketoreductase gene detection may present limitations. The first limitation is that this method doesn't allow a differentiation between M. ulcerans strains and fast-growing or slow-growing mycobacteria, possessing IS2404 sequence or having acquired the plasmid pMUM001. Many cutaneous lesions probably caused by other mycobacteria could be attributed to M. ulcerans. The second limitation of this method is that it does not target IS2606 sequence found in M. marinum also involved in chronic skin ulcerations. The third limitation is that it could target environmental mycobacteria that contaminate skin ulcers. The best molecular diagnostic approach would be to target both insertion sequences IS2404-IS2606 and ketoreductase gene. On the one hand, this algorithm would discriminate between M. ulcerans and other environmental mycobacteria that also possess IS2404 sequence. On the other hand, it would determine the involvement of M. marinum in chronic skin lesions.

Conclusion
Mycobacterium ulcerans is an environmental mycobacterium that causes an endemic skin disease called "Buruli ulcer". Every year several health districts in Cote d'Ivoire report a large number of chronic ulcers, rightly or wrongly, as cases of Buruli ulcer.
Phenotypic and genomic characterization of strains isolated from Buruli ulcer suspected cases showed the presence of different mycobacteria species: M. ulcerans (63.8%), M. marinum (6.4%) and other atypical mycobacteria, such as M. chelonae and M. smegmatis (29.8%). Most of M. ulcerans strains (90%) and atypical mycobacterial strains (63.6%) expressed genes encoding mycolactone bio synthesis, with the exception of M. marinum strains. Geographical distribution of strains showed that M. ulcerans co-circulated with other mycobacteria strains in the same ecosystems. Some of those non-M. ulcerans mycobacteria possessed insertion sequences (IS2404, IS2606) and virulence genes.