Pulmonary Isolation of Multidrug resistant "Mycobacterium simulans" and Mycobacterium tuberculosis from a patient in the Horn of Africa.

In low-income countries of the Horn of Africa, pulmonary infections are usually considered as tuberculosis, which diagnosis relies on clinical data and positive microscopic observation. This strategy allows non-tuberculous mycobacteria to escape detection, facilitating their emergence in populations. A non-tuberculous mycobacterium strain FB-527 was unexpectedly cultured from the sputum of a Djiboutian patient otherwise diagnosed with multi-drug resistant (MDR) tuberculosis. The sequencing of the rpoB and 16S rRNA genes showed that the isolate was identical to strain FI-09026 previously named "Mycobacterium simulans" and reported only once from a Somali patient. Strain FB-527 mimicked Mycobacterium tuberculosis colonies and enzymatic profile using API ZYM strip and was in vitro resistant to rifampicin and isoniazid. Isolation of two MDR mycobacteria complicated the diagnosis and therapeutic management of the patient. We here report on the complete description of strain FB-527 and strain FI-09026 including genome sequencing, finalizing the description of the proposed new species "Mycobacterium simulans".

Introduction

Tuberculosis remains a major public health problem in several African countries with an average incidence of 275/100.000 inhabitants in the continent, with 84% of cases presenting as pulmonary tuberculosis[1]. The Republic of Djibouti located in the Horn of Africa exhibits one of the highest incidence rates of tuberculosis worldwide with an estimated incidence of 378/100.000 inhabitants with 57% of cases presenting as pulmonary tuberculosis[1]. Such a high incidence rate parallels a high rate of antibiotic-resistant Mycobacterium tuberculosis strains[2,3]. There, the diagnosis of pulmonary tuberculosis does not rely on the isolation of the causative pathogen agent but rather on the microscopic examination of sputum smear after Ziehl-Neelsen staining[3]. This diagnostic approach may be uncertain regarding the exact identification of the mycobacteria, including M. tuberculosis and Mycobacterium canettii, the latter species being reported to cause 3–6% of pulmonary tuberculosis cases in the Republic of Djibouti[3-5]. Moreover, this culture-free diagnostic approach closes the door to the discovery of any additional mycobacterium which could divert the current laboratory diagnostic strategy from pulmonary tuberculosis identification in developing countries.

By investigating mycobacteria responsible for pulmonary tuberculosis in the Republic of Djibouti by culturing sputum and accurately identifying colonies using a polyphasic approach, we unexpectedly isolated a non-tuberculous mycobacterium strain FB-527 in one patient presenting with antibiotic-resistant pulmonary tuberculosis[4]. We report this patient case along with a complete polyphasic description of strain FB-527.

Results

Case report

On February 2016, a 40-year-old Djiboutian man presented at Chakib Saad Hospital, Djibouti, a hospital in charge of pulmonary pathologies. The patient was living in Balbala (Bouldhouqo) in Djibouti and working as a seller in a clothing store. He reported no travel outside Djibouti, no medical, surgical or tuberculosis histories, but a three-month cough. The patient was found to be HIV-negative. A chest radiograph revealed a retractile opacity of the right upper lobe of the lung and a para-aortic opacity with micronodules of the culmen (Fig. 1). Direct microscopic examination of the sputum smear after Ziehl-Neelsen staining exhibited acid-fast bacilli but the sputum was not cultured. The patient was diagnosed with pulmonary tuberculosis and received first-line antituberculosis drugs. After three months of treatment, the patient returned to the hospital with persistent symptoms. Direct examination of the sputum was positive and a rifampicin-resistant M. tuberculosis complex isolate was detected by GeneXpert® MTB/RIF lab test (Cepheid, Sunnyvale, CA). The patient was hospitalized and treated with daily kanamycin (1 g), moxifloxacin (400 mg), prothionamide (250 mg), clofazimine (100 mg), isoniazid (300 mg), ethionamide (250 mg) and pyrazinamide (400 mg), the doses being adjusted to the patient’s weight. On July 2016, a first positive MGIT (Becton Dickinson, Le Pont-de-Claix, France) culture obtained from sputum yielded strain 5175 identified as a M. tuberculosis complex isolate by SD BIOLINE TB Ag MPT64 rapid test® (Standard Diagnostics, Inc., Seoul, South Korea). In September 2016, improved clinical course contrasted with the positivity of sputum cultures in MGIT but the patient was readmitted to the hospital in November 2016 for persisting cough. In January 2017, the patient was diagnosed with treatment failure and was treated with kanamycin, levofloxacin, cycloserine, linezolid, para-aminosalicylic acid (PAS) and bedaquiline and Directly Observed Treatment (DOT) follow-up. Complementary microbiological investigations confirmed antibiotic-resistant M. tuberculosis and occasional isolation of strain FB-527 from respiratory material during follow-up.

Figure 1

Initial chest radiograph done during the first diagnosis. It revealed a retractile opacity of the right upper lobe of the lung and a para-aortic opacity with micronodules of the culmen.

Microbial investigations

Further microbiological investigations were conducted in collaboration with the Hôpital d’Instruction des Armées Alphonse Laveran, Marseille, France and the Institut Hospitalier Universitaire Méditerranée Infection, Marseille, France. M. tuberculosis complex isolate strain 5175 cultured in July 2016 in Djibouti was sent to France in May 2017 where it was manipulated within a biosafety level 3 laboratory. Matrix-assisted laser desorption ionization-time of flight-mass spectrometry (MALDI-TOF-MS) using a Microflex LT MALDI-TOF mass spectrometer (Bruker Daltonics, Germany) identified a M. tuberculosis complex isolate[6]. Further analysis of regions of deletion by multiplex PCR precisely identified a M. tuberculosis species[7] and GeneXpert® MTB/RIF test found rifampicin resistance. Drug susceptibility tests using ETEST® confirmed rifampicin resistance [minimal inhibitory concentration (MIC) > 32 mg/L] and resistance to isoniazid (MIC > 4 mg/L), streptomycin (MIC > 1,024 mg/L) and ethambutol (MIC > 256 mg/L). On drug-supplemented solid media, M. tuberculosis strain 5175 was susceptible to chloramphenicol (20 mg/L) and clofazimine (1.5 mg/L) but resistant to pyrazinamide (100 mg/L) and minocycline (4 mg/L).

By the end of September 2016, one sputum specimen was cultured in Marseille on Coletsos medium (Bio-Rad, Marnes-la-Coquette, France). After a two-month incubation period at 37 °C, one rough colony was observed by the naked eye and microscopic examination after auramine-O staining was positive, the strain was then referred to as strain FB-527. Subcultures in BACTEC MGIT medium (Becton-Dickinson) and Middlebrook 7H10 (Becton-Dickinson) were positive after 7 and 10 days, respectively. Colonies tested negative for the GeneXpert® MTB/RIF test and yielded no identification by MALDI-TOF-MS analysis (Bruker database, version December 2015). The rpoB and 16S rRNA gene partial gene sequencing yielded a similarity of 99% with “M. simulans” FI-09026[4,8].

Phenotypic characterization

Strain FB-527 displayed rough and non-pigmented colonies growing at a temperature range of 25 °C to 37 °C after a 10-day incubation period on egg-based Coletsos medium. The colonies’ morphology resembled that of M. tuberculosis H37Rv (Fig. 2A). Optimal growth was obtained at 37 °C under microaerophilic atmosphere as well as in 5% CO2 atmosphere. Growth occurred up to only 1% NaCl. Ziehl-Neelsen staining showed dispersed pink bacilli and clumps (Fig. 2B). Further observation of colonies by electron microscopy showed rod-shaped bacilli with a length of 1.33 ± 0.17 µm and a width of 0.62 ± 0.06 µm (Fig. 2C). A reproducible MALDI-TOF-MS profile was generated which was easily distinguishable from that of M. tuberculosis H37Rv (Fig. 3). Further mass spectrometry analysis of mycolic acids of strains FB-527, FI-09026[8] and M. tuberculosis H37Rv as a positive control yielded good mass accuracy (below 5 ppm error). M. tuberculosis H37Rv yielded a previously well described mycolic acid pattern[9,10], including α- (C72-84), methoxy- (C81-90) and keto- (C80-89) forms (Fig. 4). Strains FI-09026 and FB-527 yielded a similar mycolic acid pattern including α- (C74-86), methoxy- (C80-90) and keto/epoxy/ω-1- (C82-89) mycolic acids. A low abundance of α‘- mycolic acids (C75-79) was observed for both strains (2 and 4%). In addition, FI-09026 and FB-527 strains presented profiles with twice more keto/epoxy/ω-1 forms (29 and 26%) and less methoxy- forms (22 and 29%) than M. tuberculosis H37Rv (13% keto-and 43% methoxy-). The overall relative percent of the α-subclass was equivalent in the three strains (48%, 41% and 45%, respectively) (Table 1).

Figure 2

Morphological characteristics of “M. simulans” strain FB-527. (A) Colony aspect of “M. simulans” strain FB-527 cultured on Coletsos medium compared to M. tuberculosis H37Rv. Both mycobacteria display rough colonies. (B) Ziehl-Neelsen staining showed pink bacilli with dispersed or clumped mycobacteria. (C) Transmission electron microscopy of “M. simulans” strain FB-527. The scale bar represents 500 nm.

Figure 3

MALDI-TOF-MS spectra of “M. simulans” strain FB-527 in comparison with that of M. tuberculosis H37Rv.

Figure 4

ESI-MS spectra of the [M-H]- mycolic acid ions. (A) Mycobacterium tuberculosis H37Rv (control), (B) strain FI-09026 and (C) Strain FB-527.

Table 1

Identified mycolic acids for “Mycobacterium simulans” strains FI-09026 and FB-527 and Mycobacterium tuberculosis H37Rv (control).

Mycolic acid subclass	Formula	Calculated [M-H]	Mycobacterium simulans FI-09026			Mycobacterium simulans FB-527			Mycobacterium tuberculosis H37RV
			Measured [M-H]-	Error (ppm)	%a	Measured [M-H]	Error (ppm)	%a	Measured [M-H]	Error (ppm)	%a
α	C72H140O3	1052.07297							1052.06986	−3.0	0.3
	C74H144O3	1080.10427	1080.10151	−2.6	0.4	1080.10569	1.3	0.1	1080.10151	−2.6	1.5
	C75H146O3	1094.11992	1094.12099	1.0	0.1	1094.11679	−2.9	0.1	1094.11679	−2.9	0.4
	C76H148O3	1108.13557	1108.13354	−1.8	4.1	1108.13354	−1.8	1.1	1108.13354	−1.8	8.6
	C77H150O3	1122.15122	1122.15435	2.8	0.5	1122.15009	−1.0	0.8	1122.15009	−1.0	1.3
	C78H152O3	1136.16687	1136.16903	1.9	10.4	1136.16474	−1.9	5.4	1136.16903	1.9	17.3
	C79H154O3	1150.18252	1150.18015	−2.1	0.6	1150.18015	−2.1	0.2	1150.18015	−2.1	1.5
	C80H156O3	1164.19817	1164.19906	0.8	17.8	1164.19472	−3.0	14.4	1164.19906	0.8	9.5
	C81H158O3	1178.21382	1178.21126	−2.2	0.4	1178.21126	−2.2	0.2	1178.21126	−2.2	0.7
	C82H160O3	1192.22947	1192.22827	−1.0	12.3	1192.22827	−1.0	14.9	1192.22827	−1.0	2.8
	C83H162O3	1206.24512				1206.24429	−0.7	0.8
	C84H164O3	1220.26077	1220.26228	1.2	1.3	1220.25784	−2.4	2.6	1220.25784	−2.4	0.7
	C86H168O3	1248.29207				1248.2897	−1.9	0.3
α′	C75H148O3	1096.13557	1096.13596	0.4	0.5	1096.13596	0.4	0.4
	C77H152O3	1124.16687	1124.16512	−1.6	1.3	1124.16512	−1.6	1.5
	C79H156O3	1152.19817				1152.19429	−3.4	2.3
keto/epoxy/ω − 1b	C80H156O4	1180.19309							1180.19739	3.6	1.3
	C82H160O4	1208.22439	1208.22738	2.5	1.7				1208.22738	2.5	1.2
	C83H162O4	1222.24004	1222.24353	2.9	2.8				1222.24353	2.9	1.1
	C84H164O4	1236.25569	1236.25558	−0.1	1.7	1236.25558	−0.1	2.2	1236.26005	3.5	1.6
	C85H166O4	1250.27134	1250.27109	−0.2	9.8	1250.27109	−0.2	7.7	1250.27558	3.4	2.6
	C86H168O4	1264.28699	1264.28423	−2.2	1.0	1264.28423	−2.2	2.8	1264.28423	−2.2	1.2
	C87H170O4	1278.30264	1278.30276	0.1	9.6	1278.30276	0.1	10.2	1278.30276	0.1	3.5
	C88H172O4	1292.31829				1292.31631	−1.5	0.8
	C89H174O4	1306.33394	1306.33275	−0.9	1.8	1306.33275	−0.9	2.5	1306.33735	2.6	0.4
methoxy	C80H158O4	1182.20874				1182.20704	−1.4	0.3
	C81H160O4	1196.22439	1196.22575	1.1	0.6	1196.23014	4.8	1.1	1196.22575	1.1	3.0
	C82H162O4	1210.24004	1210.23864	−1.2	0.6	1210.23422	−4.8	0.4	1210.24306	2.5	1.2
	C83H164O4	1224.25569	1224.25307	−2.1	2.4	1224.25752	1.5	1.5	1224.25752	1.5	8.0
	C84H166O4	1238.27134				1238.27213	0.6	1.2	1238.26766	−3.0	3.0
	C85H168O4	1252.28699	1252.28104	−4.8	6.0	1252.28104	−4.8	5.2	1252.28554	−1.2	11.5
	C86H170O4	1266.30264	1266.29636	−5.0	1.4	1266.29636	−5.0	2.4	1266.30089	−1.4	4.5
	C87H172O4	1280.31829	1280.31236	−4.6	6.0	1280.31326	−3.9	8.4	1280.31691	−1.1	5.3
	C88H174O4	1294.33394	1294.33233	−1.2	1.9	1294.33233	−1.2	3.2	1294.33233	−1.2	4.8
	C89H176O4	1308.34959	1308.34587	−2.8	1.5	1308.34587	−2.8	2.8	1308.34587	−2.8	0.8
	C90H178O4	1322.36524	1322.36553	0.2	1.2	1322.36091	−3.3	2.1	1322.37015	3.7	0.6

aRelative intensity was calculated from sum of the identified mycolic acids; bketo -, epoxy - and ω-1 - forms could not be distinguished because of identical chemical formula.

As for the biochemical tests, the catalase test was positive at room temperature but niacin production and Tween 80 hydrolysis were negative. Alkaline phosphatase, esterase (C4), lipase esterase (C8), lipase (C14), leucine arylamidase, acid phosphatase and phosphoamidase were positive for FB-527 and FI-09026, as detected by using the API ZYM strip (bioMérieux, Craponne, France). This pattern should be undistinguishable from that of M. tuberculosis H37Rv used as a positive control (Supplementary Table 1). However, inoculation of the API CORYNE strip (bioMérieux) indicated that strain FB-527 was positive for nitrate reductase, phosphatase alcaline, ß-glucosidase and fermentation of D-glucose, D-maltose and D-saccharose while strain FI-09026 was positive for pyrazinamidase, phosphatase alcaline and ribose fermentation and M. tuberculosis H37Rv was positive for alkaline phosphatase only (Supplementary Table 1). Antibiotic susceptibility pattern tested using ETEST® showed that FB-527 was in vitro susceptible to streptomycin (MIC < 0.064 mg/L), amikacin (MIC, 3 mg/L), clarithromycin (MIC, 0.032 mg/L), ethambutol (MIC, 0.5 mg/L), linezolid (MIC, 0.125 mg/L) and trimethoprim-sulfamethoxazole (MIC < 0.002 mg/L); intermediate to azithromycin (MIC, 8 mg/L), levofloxacin (MIC, 2 mg/L) and rifampicin (MIC, 2 mg/L); and resistant to doxycycline (MIC > 12 mg/L), imipenem (MIC > 32 mg/L), meropenem (MIC > 32 mg/L) and isoniazid (MIC > 256 mg/L) (Table 2). Strain FB-572 was also susceptible to chloramphenicol (20 mg/L), clofazimine (1.5 mg/L) and minocycline (MIC 4 mg/L)[4]. Furthermore, strain FI-09026 was shown to be in vitro susceptible to chloramphenicol (20 mg/L), azithromycin (MIC < 0.19 mg/L) and doxycycline (MIC, 0.25 mg/L); and resistant to levofloxacin (MIC, 12 mg/L), clofazimine (1.5 mg/L) and minocycline (4 mg/L) (Table 2).

Table 2

Minimum inhibitory concentration (MIC) of selected antibiotics against “M. simulans” strains. MIC values are given in g/L.

Drug	MIC (Interpretationa)
	Strain FB-527	Strain FI-09026
Azithromycin	8 (I)	0.125–0.19 (S)
Clarithromycin	0.032 (S)	0.5 (S)b
Levofloxacin	2 (I)	12 (R)
Amikacin	3 (S)	8 (S)b
Streptomycin	<0.064 (S)	>64 (R)b
Ethambutol	0.50 (S)	4 (I)b
Rifampicin	1–2 (I)	4 (I)b
Isoniazid	>256 (R)	(R)b
Minocycline (4 mg/L)	Sc	R
Doxycycline	12–16 (R)	0.25 (S)
Clofazimine (1.5 mg/L)	Sc	R
Chloramphenicol (4 mg/L)	Sc	S
Linezolid	0.125 (S)	0.5 (S)b
TRIM/SULFA	<0.002 (S)	4–76 (R)b
Imipenem	>32 (R)	>32 (R)
Meropenem	>32 (R)	>32 (R)

aI, intermediate; S, susceptible; R, resistant.

bResults according to Tortoli et al.[8].

cResults according to Bouzid et al.[4].

Genetic characterization

Strain FB-527 16S rRNA gene sequence (GenBank Accession Number: LT935784) was 99.6% similar to that of strain FI-09026 (FJ786255). A 16S rRNA gene sequence-based phylogenetic tree showed that strain FB-527 was a member of the Mycobacterium szulgai complex closely related to Mycobacterium riyadhense (Fig. 5). We also sequenced a 733-bp rpoB gene fragment in FB-527 strain (LT935785)[11]. This sequence’s highest similarity rate was of 99.3% with strain FI-09026 (FJ786254) which enforced the identity of strain FB-527 at the species level. Genome sequencing of strains FB-527 and FI-09026 yielded 49 and 105 contigs indicative of one 6,251,405 bp-long chromosome (64.5% GC content) and one 6,192,024 bp-long chromosome (64.6% GC content), respectively, without any evidence of extra-chromosomal replicon. FB-527 genome encodes for 5,684 proteins and 51 RNAs including 48 tRNA, 2 rRNA and 1 tmRNA. A total of 3,563 (62.1%) genes were assigned with a putative function. The remaining genes were annotated as hypothetical proteins 2,172 (37.9%). FI-09026 genome encodes for 5,570 proteins and 54 RNAs including 50 tRNA, 3 rRNA and 1 tmRNA. A total of 3,490 (62%) genes were assigned a putative function while 2,134 (38%) genes were annotated as hypothetical proteins. Annotated genome sequences of strains FB-527 and FI-09026 have been deposited (GenBank accession number: OCTY01000001-OCTY01000049 and OCVX01000001-OCVX01000105, respectively).

Figure 5

Phylogenetic tree based on the 16S rRNA gene sequence bootstrapped 1,000 times indicating the phylogenetic position of “M. simulans” strain FB-527 and strain FI-09026 here referred to as “M. simulans”. The tree was constructed using the MEGA7 software. Bootstrap values ≥60% are indicated at nodes. Bar: 0.002 substitutions per nucleotide position.

Comparing whole genome sequences, strain FB-527 exhibits an average nucleotide identity (ANI)[12] of 97.88% with strain FI-09026. In silico DNA-DNA hybridization analysis (DDH)[13] comparing strains FB-527 and FI-09026 yielded a value of 79.90% [76.9–82.5%] (Table 3). ANI and DDH similarity values being greater than 95–96%[12] and 70%[13], respectively, indicated that both strains belonged to the same species. Further DDH analysis with available M. szulgai complex species genomes based on 16S rRNA similarity including M. szulgai, M. angelicum and M. riyadhense yielded values <70% (Table 3).

Table 3

Comparison of “M. simulans” strains with related mycobacteria species using GGDC, formula 2 (DDH estimates based on identities/HSP length).

Strains	FB-527	FI-09026
M. szulgai ACS1160	22.50% [20.2–24.9%]	22.50% [20.3–25%]
M. szulgai 1146974.2	22.70% [20.4–25.2%]	22.70% [20.4–25.2%]
M. szulgai DSM 44166	25.10% [22.7–27.5%]	25.10% [22.8–27.6%]
Mycobacterium angelicum DSM 45057	25.30% [23–27.8%]	25.10% [22.8–27.6%]
M. riyadhense DSM 45176	44.50% [42–47.1%]	44.40% [41.8–46.9%]
FI-09026	79.90% [76.9–82.5%]	100%
FB-527	100%	79.90% [76.9–82.5%]

Discussion

We unexpectedly isolated “M. simulans” from a sputum specimen of a Djiboutian patient in treatment for confirmed antibiotic-resistant pulmonary tuberculosis[8].

The identification of “M. simulans” strain FB-527 isolate was firmly established by rpoB and 16S rRNA gene sequencing[4]. This isolate is the second one to be reported worldwide. Indeed, the index case led to the partial description of this non-tuberculous Mycobacterium species[8]. The first isolation of “M. simulans” strain FI-09026 originated from the sputum of a Somali patient with severe cavitary pulmonary disease who had been first diagnosed with MDR tuberculosis[8]. Medical history of this patient included pulmonary tuberculosis diagnosed four years before with interrupted therapy[8]. The only two patients reported to harbor “M. simulans” were not immunocompromised and originated from the same geographical area, the Horn of Africa (Somalia and the Republic of Djibouti) which otherwise remains an endemic area of tuberculosis[1].

The case herein reported gave us the opportunity to completely describe this emerging species now comprising two isolates, strain FB-527 and strain FI-09026, with phenotypic, genetic and genomic data. Strain FB-527T has been deposited in the CSUR collection under number P4791. Our investigations confirmed “M. simulans” as a member of the M. szulgai complex closely related to M. riyadhense[8], another non-tuberculous Mycobacterium species phenotypically mimicking M. tuberculosis[14]. “M. simulans” is therefore the 4th species of the M. szulgai complex.

This case report provides a new addition to the list of Mycobacterium species responsible for pulmonary infection or colonization in exposed patients in the Horn of Africa (Table 4). According to the criteria proposed by the American Thoracic Society, the case here reported was described as colonization rather man infection by “M. simulans”. Indeed, in addition to M. tuberculosis complex members, sporadic reports showed pulmonary infections due to non-tuberculous mycobacteria with M. tuberculosis co-infection for three cases[3,4,8,15]. Therefore, there is a necessity for an accurate documentation of cases as the antibiotic susceptibility pattern of these eight different species is not the same, thus impeding the antibiotic treatment of patients. Even within the same species, we showed that the profile of drug susceptibility differs between strain FB-527 and strain FI-09026 (Table 2). Indeed, there is an urgent need to expand culture and drug susceptibility-testing capacity in tuberculosis diagnostic services according to WHO recommendations[16].

Table 4

Mycobacterium species responsible for pulmonary infection or colonization in exposed patients in the Horn of Africa.

Mycobacterium	Geographical exposure
M. tuberculosis	Djibouti[4], Ethiopia[14], Somalia[27], Eritrea[1]
M. canettii	Djibouti[28], Somalia[28]
M. bovis	Djibouti[29], Ethiopia[14], Eritrea[30]
M. africanum	Djibouti[29]
M. chelonae	Djibouti[3]
M. fortuitum	Djibouti[3]
M. peregrinum	Djibouti[3]
M. intracellulare	Ethiopia[14]
M. flavescens	Ethiopia[14]
M. simiae	Ethiopia[14]
M. kansasii	Djibouti[4]
“M. simulans”	Somalia[8], Djibouti (Our study)

Moreover, this case indicates that an accurate identification of Mycobacterium species responsible for pulmonary infection should rely on culturing sputum specimens, which allows to isolate the colonies to be identified, rather than bypassing culture by direct examination after Ziehl-Neelsen staining or some diagnostic assay based on nonspecific DNA tests. The index “M. simulans” FI-09026 isolate was misidentified as a member of the M. tuberculosis complex by using the GenoType© Mycobacterium CM test. In addition, confusing results were obtained after reverse hybridization test (GenoType MTBC) and GenoType MTBDRplus test[8]. By comparing “M. simulans” FB-527 and M. tuberculosis H37Rv phenotypes, we have also shown that characterization based on colony morphology or the profile of enzymatic reactions can lead to misidentification. However, MALDI-TOF-MS[6] and rpoB gene sequencing[11] gave an accurate identification. We urge for further research in culturing Mycobacterium species and the implementation of effective culture facilities in emerging countries instead of the sole development of the PCR amplification-based approach for the diagnosis of pulmonary tuberculosis and related syndromes[17].

Methods

Strain isolation and identification

This study has been performed in accordance with relevant guidance and regulations and was approved by the IHU Méditerranée Infection, Ethics Committee Approval n°2016-025, Marseille, France. Collection of sputum was part of the patients’ routine care activity. After being informed, patients who agreed to participate signed an informed anonymised consent. Isolation and first-line identification procedure of “M. simulans” strain FB-527 was described previously[4]. Culture was maintained on egg-based Coletsos (Bio-Rad, Marnes-la-Coquette, France) and Middlebrook 7H10 (Becton Dickinson, Le Pont de Claix, France) supplemented with 10% oleic acid-albumin-dextrose-catalase (OADC) (Becton Dickinson). This strain was deposited in the CSUR collection under number P4791. Strain FI-09026 was purchased from the DSMZ collection under number DSM 45395. Freeze-dried colonies were cultured on Middlebrook 7H10 supplemented with 10% OADC (Becton Dickinson) and in the mycobacteria Growth Indicator Tube (MGIT) liquid medium (BACTEC™ MGIT™ 960, Becton Dickinson). Identification was confirmed by partial sequencing of rpoB gene and 16S rRNA[11,18]. This stain was then deposited in the CSUR collection under number P4792. The M. tuberculosis strain 5175 conserved in Djibouti was cultured in France as indicated above in a biosafety level 3 laboratory. Identification was first performed by MALDI-TOF- MS[7]. The species level has been specified by multiplex PCR amplification of regions of difference as previously described[8]. Rifampicin resistance was assessed by GeneXpert® MTB/RIF lab test (Cepheid, Sunnyvale, CA). Drug susceptibility testing to rifampicin, isoniazid, streptomycin and ethambutol were performed using ETEST® (bioMérieux, La Balme les Grottes, France). In addition, susceptibility to chloramphenicol (20 mg/L), clofazimine (1.5 mg/L), minocycline (4 mg/L) and pyrazinamide (100 mg/L) was tested in 6-well plates including two drug-free wells and four drug-supplemented wells on solid media by inoculating 106 colony-forming units (CFUs) onto MOD9 solid medium[19] incorporating the tested antibiotic.

“M. simulans” strain FB-527 strain was cultured in egg-based Coletsos medium (Bio-Rad) or Middlebrook 7H10 (Becton Dickinson) supplemented with 10% oleic acid-albumin-dextrose-catalase (OADC) (Becton Dickinson) at 25–42 °C for four weeks. Inoculated plates were inspected daily by naked eye to determine growth time. Colonies were microscopically examined after Ziehl-Neelsen staining (RAL diagnostics, Martillac, France). The size of the microorganisms was determined by transmission electron microscopy after negative staining of bacteria. After an overnight fixation with 2.5% glutaraldehyde at 4 °C, bacterial suspension was applied to the top of a formvar carbon 400 mesh nickel grid (FCF400-Ni, EMS) and stained with 1% ammonium molybdate (1–800-ACROS, USA). Electron micrographs were acquired on a Tecnai G20 transmission electron microscope (FEI) operated at 200 Kev. MALDI-TOF- MS protein analysis was carried out after direct colony deposit as previously described[7]. Salt tolerance was tested by supplementing the Middlebrook 7H10 solid medium with 0–5% NaCl as previously described[20]. Niacin production was detected using BBL™ Taxo™ TB Niacin test strips (Becton Dickinson) as described by the manufacturer. Tween 80 hydrolysis test was performed as previously described[21].

Enzymatic activities were determined for strains FB-527 and FI-09026 by inoculating API® ZYM and API® Coryne strips (bioMérieux, Bruz, France) as indicated by the manufacturer using M. tuberculosis H37Rv as positive control. The MIC of the major antimycobacterial agents was determined using ETEST® (bioMérieux, Craponnes, France). Then, susceptibility to 1.5 mg/L clofazimine, 4 mg/L minocycline and 5 mg/L chloramphenicol was tested by inoculating 106 colony-forming units (CFUs) onto MOD9 solid medium[19] incorporating the tested antibiotic.

Extraction and analysis of mycolic acids

Strains FB-527, FI-09026 and M. tuberculosis H37Rv (control) were cultured on Middlebrook 7H10 agar medium supplemented with 10% OADC for two weeks. Mycolic acids were prepared as detailed previously with modifications[9,22]. At least 10 inoculation loops were collected from a culture plate and transferred into 2 mL of potassium hydroxyde 9 M. Mycolic acids were hydrolysed at 100 °C during 2 hours. Free mycolic acids were then extracted with 2 mL of low pH chloroform by adding 3 mL of 6 N hydrochloric acid to the aqueous phase. The organic layer was collected and dried at 40 °C under a stream of nitrogen. Free mycolic acids were then dissolved in 100 μL of a methanol-chloroform mixture (50:50, v/v) and subjected to electrospray-mass spectrometry analysis after a 1,000-fold dilution in methanol. Samples were analyzed in the Sensitivity Negative ionization mode using a Vion IMS QTof high resolution mass spectrometer (Waters, Guyancourt, France). Samples were infused at 10 μL/min, after fluidics wash with chloroform/methanol (50/50), and monitored from 500 to 2000 m/z during 2 minutes. Ionization parameters were set as follows: capillary voltage of 2.5 kV, cone voltage of 50 V, source and desolvation temperatures comprised between 120/650 °C. Mass calibration was adjusted automatically during analysis using a Leucine Enkephalin solution at 50 pg/μL (554.2620 m/z). Mass spectra were combined between 1000 and 1400 m/z for subsequent data interpretation. Mycolic acids were described according to previously detailed structures[23]. Here, keto, epoxy and ω-1 mycolic acid subclasses could not be distinguished.

Genome sequencing, assembly and annotation

Genomic DNA (gDNA) of “M. simulans” strains FB-527 and FI-09026 was extracted in two steps. A mechanical treatment was first performed by acid-washed glass beads (G4649-500g Sigma) using a FastPrep BIO 101 instrument (Qbiogene, Strasbourg, France) at maximum speed (6.5) for 90 s. Then, after a 2-hour lysozyme incubation period at 37 °C, gDNA was extracted on the EZ1 biorobot (Qiagen) with EZ1 DNA tissue kit. The elution volume was of 50 µL. gDNA was quantified by a Qubit assay with the high sensitivity kit (Life technologies, Carlsbad, CA, USA) to 8.6 ng/µL and 4.5 ng/µL for strains FB-527 and FI-09026 respectively. gDNA of strains FB-527 and FI-09026 were sequenced on the MiSeq Technology (Illumina Inc, San Diego, CA, USA) with paired-end and barcode strategies in order to be mixed with 15 others projects constructed according the Nextera XT library kit (Illumina). To prepare the paired-end library, dilution was performed to require 1 ng of each genome as input. The « tagmentation » step fragmented and tagged the DNA. Then limited cycle PCR amplification (12 cycles) completed the tag adapters and introduced dual-index barcodes. The library profile was validated on an Agilent 2100 BioAnalyzer (Agilent Technologies Inc, Santa Clara, CA, USA) with a DNA High sensitivity labchip and the fragment size was estimated to be of 1.5 kb. After purification on AMPure XP beads (Beckman Coulter Inc, Fullerton, CA, USA), the libraries were then normalized on specific beads according to the Nextera XT protocol (Illumina). Normalized libraries were pooled for sequencing on the MiSeq. Automated cluster generation and paired-end sequencing with dual index reads were performed in a single 39-hour run in 2 × 250-bp. For strain FB-527, total information of 6.9 Gb was obtained from a 714 k/mm2 cluster density with a cluster passing quality control filters of 96.6% (13,376,000 passed filtered clusters). Within this run, the index representation for strain FB-527 was determined to be of 1.72%. The 230,075 paired-end reads were trimmed and filtered according to the read qualities. For strain FI-09026, total information of 2.8 Gb was obtained from a 277 K/mm2 cluster density with 98.2% (5,333,000 clusters) of the clusters passing quality control filters. Within this pooled run, the index representation of strain FI-09026 was determined to be of 6.57%. Then, 350,497 paired-end reads were filtered.

gDNA of strain FB-527 was also sequenced with mate pair application and was barcoded in order to be mixed with 11 others projects for the Nextera Mate Pair sample prep kit (Illumina). The mate pair library was prepared with 598 ng of gDNA using the Nextera mate pair Illumina guide. The gDNA sample was simultaneously fragmented and tagged with a mate pair junction adapter. The pattern of the fragmentation was validated on an Agilent 2100 BioAnalyzer (Agilent Technologies Inc, Santa Clara, CA, USA) with a DNA 7500 labchip. The DNA fragments ranged in size from 1.5 kb up to 11 kb with an optimal size at 3.33 kb. No size selection was performed and 148.8 ng of tagmented fragments were circularized. The circularized DNA was mechanically sheared to small fragments with an optimal size at 331 bp on the Covaris device S2 in T6 tubes (Covaris, Woburn, MA, USA). The library profile was visualized on a High Sensitivity Bioanalyzer LabChip (Agilent Technologies Inc, Santa Clara, CA, USA) and final concentration library was measured at 1.11 nmol/l. The libraries were normalized at 2 nM and pooled. After a denaturation step and dilution at 20 pM, the pool of libraries was loaded onto the reagent cartridge and then onto the instrument along with the flow cell. Automated cluster generation and sequencing run were performed in a single 39-hour run in a 2 × 151-bp. Total information of 4.7 Gb was obtained from a 461 K/mm2 cluster density with a cluster passing quality control filters of 97.8% (9,187,000 passing filter paired reads). Within this run, the index representation for strain FB-527 was determined to be of 7.48%. The 687,466 paired reads were trimmed then assembled with the paired-end reads.

For both strains, reads were assembled using the SPAdes software (http://bioinf.spbau.ru/spades)[24] and gaps were closed if possible by GapFiller[25]. Open reading frames (ORFs) and annotation were predicted using prokka[26] with default parameters. Average nucleotide identity (ANI) between strains FB-527 and FI-09026 was determined by OrthoANI Tool version 0.93.1[12] using the value of 95–96% for species demarcation[12]. Genome sequences of strains FB-527 and FI-09026 were further incorporated in silico DNA-DNA hybridization[13] and were compared with whole genome sequences of M. szulgai complex members selected based on 16S rRNA gene proximity. DDH values were estimated using the GGDC version 2.0 online tool using the value of 70% as a cut-off for species delineation[13].

Phylogenetic analysis

The phylogenetic tree based on 16S rRNA sequences was constructed using the MEGA7 software. The evolutionary history was inferred by using the Maximum Likelihood method based on the Hasegawa-Kishino-Yano model. Initial trees for the heuristic search were obtained automatically by applying the neighbor-joining and BIONJ algorithms to a matrix of pairwise distances estimated using the maximum composite likelihood (MCL) approach. Statistical support for internal branches of the trees was evaluated by bootstrapping with 1000 iterations.

Dataset 1