Mycolicibacterium aurum L1^T was isolated from a vinyl chloride polluted soil collected at Arnhem, Netherland (Hartmans and de Bont, 1992). The strain was deposited by Dr. Sybe Hartmans (Wageningen Agricultural University, Netherlands) at the German Collection of Microorganisms and Cell Cultures (DSMZ) and the Spanish Type Culture Collection (CECT) under accession numbers DSM 6695^T and CECT 8761^T, respectively. The strain included in this study was obtained from CECT and the strain designation L1^T was used in the whole manuscript to avoid confusion. Mycolicibacterium sphagni DSM 44076^T (obtained from the DSMZ) was found to be the close phylogenetic neighbour of strain L1^T. These strains were maintained on proteose peptone-meat extract-glycerol agar (PMG; DSMZ 250 medium) at 28°C and conserved as bacterial suspensions in 30%, v/v glycerol at −80°C.

The cultural properties of strain L1^T were evaluated on different agar media: PMG, glucose-yeast extract-malt extract agar (GYM; DSMZ medium 65), International Streptomyces Project (ISP2; Shirling and Gottlieb, 1966), Löwenstein-Jensen medium (LJ; Jensen, 1932), Middlebrook 7H10 agar (MB7H10; Lorian, 1968) and tryptic soy agar (TSA; MacFaddin, 1985) and in presence of at a wide range of temperature 4°C, 10°C, 15°C, 25°C, 28°C, 37°C and 45°C. Anaerobic growth test of strain L1^T was examined using an anaerobic bag system (Sigma-Aldrich 68,061).

Strain L1^T and its phylogenetic neighbour, M. sphagni DSM 44076^T, were examined for a broad range of biochemical tests known to be of value in mycobacterial systematics: arylsulfatase after 3 and 14 days (Tomioka et al., 1990), catalase (de Waard and Robledo, 2007), heat stable catalase (Sequeira de Latini and Barrera, 2008), nitrate reduction (Vincent et al., 2003) and potassium tellurite tolerance (Kilburn et al., 1969; Kent and Kubica, 1985). Moreover, the growth of these strains was also evaluated in the presence of a wide range of carbon, nitrogen substrates and inhibitory compounds using GENIII microplates. The latter were inoculated with a bacterial suspension as described by Nouioui et al. (2017) and then incubated at 28°C for 5 days in an Omnilog device (Biolog Inc., Hayward, United States). The resultants data were analysed using opm package version 1.3 (Vaas et al., 2012, 2013). Furthermore, the enzymatic activities of strains L1^T and M. sphagni DSM 44076 ^T were determined using API coryne kit and following the manufacturer’s instruction (Biomérieux, France). Each test was performed in duplicate.

The chemotaxonomic markers relevant to the genus Mycolicibacterium were examined for strain L1^T and its neighbour M. sphagni DSM 44076^T. Biomass was harvested from cultures, prepared on medium DSMZ 250 and shaked at 250 rpm for 5 days at 28°C. The pellets were washed twice with sterile distilled water and then freeze-dried. Diaminopimelic acid (Schleifer and Schleifer and Kandler, 1972), whole-organism sugars (Lechevalier and Lechevalier, 1970; Staneck and Roberts, 1974) and polar lipids (Minnikin et al., 1984; Kroppenstedt and Goodfellow, 2006) were performed. Cellular fatty acids were extracted following the protocol of Miller (1982) modified by Kuykendall et al. (1988). Gas chromatography (Agilent 6,890 N instrument) was used to analyse the fatty acid methyl esters which were identified using microbial identification (MIDI) system version 4.5 and the MYCO 6 database (Sasser, 1990). Thin-layer chromatographic analyses of mycolic acid extracts of strains L1^T and M. sphagni DSM 44076^T were performed following the protocol of Goodfellow et al. (1976).

The genomic DNA extraction was performed according to Amaro et al. (2008) and the 16S rRNA gene sequence was generated using the Sanger method (Sanger and Coulson, 1975; Sanger et al., 1977) as a quality control step for the identity of the strain. The genome sequencing was performed on a MiSeq instrument (Illumina) as previously described by Sangal et al. (2015). 300 bp paired-end reads were assembled into contigs using SPAdes 3.9.0 with a k-mer length of 127 (Bankevich et al., 2012). The draft genome sequence was annotated through RAST server (Aziz et al., 2012) and deposited in GenBank database under accession number JACBJQ000000000. The pairwise comparison of average nucleotide identity (ANI) values was performed using the OrthoANIu algorithm and ANI Calculator web tool (Yoon et al., 2017a), at the EzBioCloud portal. Digital DNA–DNA hybridization (dDDH) between the draft genome sequence of strain L1^T and their close phylogenetic neighbours were estimated according to the methods described by Meier-Kolthoff et al. (2013).

An almost complete 16S rRNA gene sequence (1531 bp, accession number MT478173) was extracted from the draft genome of strain L1^T and found to be identical to the sequence obtained by Sanger method. However, the 16S rRNA gene sequence of the nearest neighbours was retrieved from the EzTaxon database (Yoon et al., 2017b). BLAST of the full 16S rRNA gene sequence of isolate L1^T was performed against those of validly named species available in EzBioCloud portal (Yoon et al., 2017b). A multiple sequence alignment of all 16S rRNA gene sequences was performed using MUSCLE (Multiple Sequence Comparison by Log- Expectation) algorithm (Edgar, 2004). Phylogenetic trees were constructed using MEGA X software (Kumar et al., 2018) and including Neighbour-Joining (NJ; Saitou and Nei, 1987) and Maximum-Likelihood (ML; Nei and Kumar, 2000) methods with 1,000 bootstrap iterations. The evolutionary distances were calculated using Kimura’s two parameter (Kimura, 1980) and General time reversible (Nei and Kumar, 2000) models.

Phylogenomic tree was inferred from the genome distances calculated with the BLAST distance phylogeny approach (GBDP) using the Type Strain Genome Server pipeline (Meier-Kolthoff and Göker, 2019). The type-based species and subspecies affiliation of strain L1^T and the 17 type strains included in the analysis were performed based on the pairwise comparisons mentioned above and according to the thresholds previously reported (Meier-Kolthoff et al., 2013; Yoon et al., 2017a). Tree annotations and visualisations were carried out using the Interactive Tree Of Life (iTOL) webtool (Letunic and Bork, 2021).

The coding sequences comprising the putative gene cluster of ethene (ETH) and vinyl chloride (VC) assimilation pathway were manually mapped and annotated on the draft genome sequence of strain L1^T considering criteria of GC reading-frame content (Bibb et al., 1984) and protein domain similarity using ARTEMIS (Berriman and Rutherford, 2003). The ORFs were screened based on their similarity with protein domains of the previously described etnEABCD gene cluster of ‘Mycobacterium smegmatis JS623’ (accession number: FJ602754.1) and ‘Mycobacterium chubuense NBB4’ genome (accession number: NC_018027), which were evidenced after comparison with the Conserved Domains Database (CDD) of NCBI (Marchler-Bauer et al., 2012). The genomic sub-regions of the clusters were compared using BLASTN (Johnson et al., 2008) and visualised with EasyFig 2.2 software (Sullivan et al., 2011).

A sequence similarity network (SSN) of the epoxyalkane coenzyme M transferase (EaCoMT) encoded by the etnE gene (Coleman and Spain, 2003b) was constructed to evaluate the taxonomic distribution of the Mycolicibacterium etnEABCD gene cluster based on the functional-sequence space of EaCoMT in homologous protein families and its genome context (Copp et al., 2018, 2019). The SSN was generated using EFI-EST (Zallot et al., 2019),² with 1000 homologous proteins from UniProtKB database³ and an e-value clustering threshold of 1E-3. The final network was processed and visualised using the organic layout within Cytoscape v. 3.2.0 (Shannon et al., 2003). The genome context of the closest proteins of validly named species was clustered along with the EaCoMT of strain L1^T. The EaCoMT clusters were preliminary visualised through EFI-GNT⁴ (Zallot et al., 2019) and fully mapped as described above. Only strains with publicly available genome sequences were analysed in this present report. The sequence of EaCoMT of the five mutants, Mycolicibacterium smegmatis JS623 (M1-M5) and the partial one of Mycolicibacterium rhodesiae JS60 (Coleman and Spain, 2003a,b), was included for the SSN analysis but not considered for further studies.

Smooth colonies of strain L1^T acquired yellow-orange colour, after 5 days of incubation on DSMZ 65 and 250, LJ and MB7H10 media at 28°C and 37°C. Optimal growth was observed on DSMZ 250 medium, pH 7 after 5 days of incubation at 28°C. No colonies were developed under anaerobic condition and neither at 4°C, 15°C, 25°C nor 45°C.

Strain L1^T and M. sphagni DSM 44076^T were unable to reduce nitrate but were able to produce arylsulfatase (after 3 and 14 days) and catalase and reduce potassium tellurite. However, only strain L1^T produced a heat stable catalase at 68°C and could be distinguished from its close neighbour by a wide range of metabolic features as shown in Table 1. Strain L1^T metabolised D-trehalose and methyl pyruvate (carbon source); butyric acid and citric acid (organic acids); and D-serine and L-arginine (amino acids). It was found to be resistant to nalidixic acid, vancomycin and was able to grow in the presence of guanidine hydrochloride, lithium chloride, up to 4% NaCl and 1% sodium lactate (Table 1).

The chemotaxonomic properties of strain L1^T were consistent with its affiliation to the genus Mycolicibacterium. Strain L1^T showed quantitative and qualitative variations in polar lipid pattern comparing to its close phylogenetic neighbour M. sphagni DSM 44076^T. The major polar lipids for strains L1^T and DSM 44076^T were diphosphatidylglycerol, phosphatidylethanolamine and phosphatidylinositol (Supplementary Figure S1). The whole-cell hydrolysates of both strains were rich in meso-diaminopimelic acid (Supplementary Figure S2), galactose, glucose, mannose and ribose as whole-cell sugars (Supplementary Figure S3). The mycolic acid profiles of strain L1^T and M. sphagni DSM 44076^T contained α-mycolate, methoxymycolate and ketomycolate (Supplementary Figure S4).

The fatty acids patterns of the strain L1^T and M. sphagni DSM 44076^T consisted of C_16:0, C_17:1 ω7c/18 alcohol, C_18:1ω9c, 10Me-C_18:0 (tuberculostearic) and 20:0 ALC 18.838/ 20:0 ALC as shown in Supplementary Table S1.

Strain L1^T showed 16S rRNA gene sequence (1531pb) similarity values of 98.5% with M. sphagni DSM 44076^T and 98.8% with Mycolicibacterium houstonense ATCC 49403^T, Mycolicibacterium senagalense CIP 104941^T and Mycolicibacterium setense DSM 45070^T. These results were not in line with their phylogenetic positions based on the ML and NJ phylogenetic trees (Figures 1A,B). Strain L1^T formed a poorly supported distinct branch that is loosely associated to a clade housed the type strains of Mycolicibacterium alvei, Mycolicibacterium fortuitum subsp. fortuitum, M. houstonense, Mycolicibacterium lutetiense, Mycolicibacterium peregrinum, Mycolicibacterium setense and M. senagalense (Figures 1A,B). Hartmans et al. (1985) proposed strain L1^T as Mycobacterium aurum which was emended as Mycolicibacterium aurum (Gupta et al., 2018) as stated above. However, the 16S rRNA gene sequence similarity between strain L1^T and the type strain Mycolicibacterium aurum DSM 43999^T was 97.3% which well below the cut-off point of 98.65% for prokaryotic species demarcation (Kim et al., 2014). Based on this data, strain L1^T potentially needs to be defined as a different species.

In the genome-based phylogeny, strain L1^T occupied a well-supported distinct branch closely related to M. sphagni ATCC 33027^T (GenBank accession number: GCA_002250655) which was next to a subclade housing Mycolicibacterium sarraceniae JCM 30395^T and Mycolicibacterium helvum JCM 30396^T (Figure 2). However, M. senegalense DSM 43656^T, M. houstonenese ATCC 49403^T and M. setense DSM 45070^T were in distant clade (Figure 2). More confidence can be attributed to the topology of the phylogenomic tree since it is generated from millions of unit characters (Nouioui et al., 2018) and therefore, M. sphagni ATCC 33027^T is considered as the close neighbour to strain L1^T.

Strain L1^T and M. sphagni DSM 44076 ^T have genome sizes of 7.1 Mb and 6.0 Mb with 66.6 and 65.9% G + C content, 6,914 and 5,690 coding sequences and 52 and 56 RNAs, respectively. The ANI and dDDH values between the draft genome sequences of strain L1^T and its close relative, M. sphagni ATCC 33027^T, were 84.7% and 28.5%, values well below the threshold of 95–96% and 70% used for prokaryotic species delineation, respectively (Wayne et al., 1987; Goris et al., 2007; Richter and Rosselló-Móra, 2009; Lee et al., 2016; Jain et al., 2018).

Since VC degradation trait has been associated with strains belong to M. aurum, M. chubuense, M. rhodesiae and M. smegmatis species, the taxonomic affiliation of the reported mycobacterial strains as assimilators of VC was evaluated based on genomic approaches. The dDDH and ANI values between the genome sequence of M. rhodesiae NBB3, M. rhodesiae JS60, M. smegmatis JS623 and M. chubuense NBB4 and those of the type strains of their corresponding species were below the defined threshold cited above and confirm that these strains were misclassified and form novel species within the genus Mycobacterium sensu lato (Supplementary Table S2). The misclassification of strain JS623 to M. smegmatis species was already reported by Garcia and Gola (2016). Therefore, these strains should be referred as Mycolicibacterium sp. to avoid confusion and misleading conclusion. In addition, genome mining for VC gene cluster of the reference strains showed that the type strains of M. aurum, M. chubuense, M. rhodesiae and M. smegmatis species devoid from VC gene cluster.

The resulting SSN for the EaCoMT (etnE) of strain L1^T was filtered to include edges with a minimum edge alignment score of 170 (identity >73.5%) and proteins within a minimum and maximum length of 330 and 500 residues (Figure 3). The full SSN₁₇₀ was integrated by 995 protein sequences, segregated into 45 isofunctional clusters and all of them correspond to the catalytic domain of the cobalamin-independent synthase II family (PF01717; Figure 3A). The cluster 5 contained the EaCoMT aminoacidic sequence of strain L1^T along with those of Mycobacteriaceae and Nocardioidaceae (Figure 3B) species, with exception of Mycolicibacterium moriokaense GAS496 whose EaCoMT protein sequence showed low similarity with all the other sequences included in the analysis and was subsequently not grouped into any cluster. The sequences incorporated into the cluster 5 belong to strains from the genus Mycolicibacterium (6 sequences), Amycolatopsis (2 sequences) and Carbonactinospora, Nocardioides, Rhodococcus, Kribella and Marmoricola (1 sequence for each); all these taxa belong to the class of Actinobacteria.

The EaCoMT sequence of strain L1^T showed 98.9, 98.6, 92.6 and 92.3% similarities to those of M. rhodesiae JS60, M. chubuense NBB4, M. smegmatis JS623 and M. rhodesiae NBB3, respectively (Figure 3B). However, the sequence identity value decreased to 76.7% between EaCoMT sequence of strain L1^T and Nocardioides sp. JS614. The latter is known as ETH and VC assimilators (Coleman et al., 2002, 2006; Coleman and Spain, 2003b; Figure 3B).

The genome neighbourhood analysis of the etnE gene (EaCoMT) for strain L1^T revealed the presence of etnABCD genes, encoding for the protein components of the AkMO (Hartmans et al., 1991), consisting on the monooxygenase β-subunit (EtnA), the monooxygenase coupling-effector protein (EtnB), the monooxygenase α-subunit (EtnC) and the alkene reductase (EtnD; Figure 4). The AkMO complex is responsible for the oxidations of ETH and VC to epoxyethane and chlorooxirane, which are conjugated to CoM yielding 2-hydroxyethyl-CoM and the putative 2-chloro-hydroxyethyl-CoM, respectively. The latter involves in the first steps of the catabolic pathway (Coleman and Spain, 2003a; Mattes et al., 2005). All mycobacterial strains of cluster 5 (Figure 3B) presented the AkMO genes (etnABCD) in their genomes (Figure 4) while the type strains of M. aurum, M. chubuense, M. rhodesiae, M. smegmatis and M. sphagni lacked etnACBD cluster.

The Coenzyme M (2-mercaptoethanesulfonic acid), which is necessary for hydroxyalkyl-CoM derivatives generation, is produced by the xcbB1,C1,D1,E1 biosynthetic gene cluster (Mattes et al., 2010; Partovi et al., 2018) and was also present in all mycobacterial genomes of cluster 5, and Nocardioides sp. JS614, with a slightly different synteny (Figure 4). The gene xcbB1 encodes for the phosphosulfolactate synthase (ComA) that catalyses a nucleophilic addition of a sulphite to phosphoenolpyruvate yielding (R)-phosphosulfolactate while argininosuccinate lyase encoded by gene xcbC1 is responsible for releasing the phosphate group by mean of β-elimination and generating sulphoacrylic acid. However, the adenylosuccinate lyase encoded by xcbD1 was found to presumably catalyse an undetermined co-substrate addition across the sulphoacrylic acid double bond. The xcbE1 gene product, D-cysteine desulfhydrase, should be the responsible for the thiolation of the unknown intermediary substrate generated after the XcbD1 reaction, in order to produce the final CoM (Partovi et al., 2018). A new xcbF1 gene associated to hypothetical protein of 308 amino acid residues is placed between genes xcbB1 and xcbE1, showing a strong synteny in all the analysed genomes (Figure 4). This protein found to have a high degree of conservation between the studied strains and an amino acid sequence identity value over 62% (Supplementary Figure S5). Further studies are necessary to determine the function of this protein in the Coenzyme M biosynthesis. Its high conservation within the cluster could be a starting point to decipher the missing intermediaries in the CoM biosynthesis (Partovi et al., 2018).

Two genes encoding for acyl-CoA synthetase (ACoAS) and two component CoA-transferase (CoATa and CoATb; Figure 4) were found in the upstream of gene etnE for strain L1^T. These genes are presumably involved in the generation of Acyl-CoA and the subsequent transference of a CoA group to the malonate in the last steps of Ethene/Vinyl chloride pathway (Mattes et al., 2010).

Following upstream in the genome sequences, strain L1^T together with all mycobacterial strains of cluster 5, showed the putative CdaR family transcription factor and the methanogen/methylotroph DcmR Sensory domain (MEDS), which found to be involved in the negative regulation of dichloromethane degradation on methylotrophic bacteria (La Roche and Leisinger, 1991). Both genes are responsible for the transcription of all the catabolic genes downstream including the etnEABCD genes (Figure 4).

Furthermore, in the upstream of MEDS regulator, strain L1^T housed an alcohol dehydrogenase-associated gene (SDR; Figure 4) which is involved in the dehydrogenation of 2-hydroxyethyl-CoM to 2-ketoethyl-CoM and a CoA-disulphide reductase which could act as a reductive decarboxylase needed to complete the malonyl-CoA assimilation reactions (Mattes et al., 2010). All the analysed strains showed sequence identities above 65% for the conserved etnEABCD gene cluster, accessory epoxyalkane catabolic genes (ScoTa and ScoTb), regulators and the genes involved in the coenzyme M biosynthesis.

The ssuABC genes responsible for the uptake of organosulfur compounds, such as sulphate esters, sulfamates, sulfonates and alkanesulfonates during sulphur limited conditions (Beale et al., 2010), were detected approximately 2.7 Kb downstream of CoM biosynthesis genes in the genome of L1^T. The product of ssuA gene corresponds to a periplasmic aliphatic sulfonate-binding protein which binds to the extracellular organosulfur compounds in order to be incorporated by an ABC transporter (proteins SsuB and SsuC) to the cell (Figure 4). An additional gene encoding for S/N-Oxide reductase (T or A; Cheng and Weiner, 2007) was also found next to the ssuABC genes suggesting its participation in the reduction of the uptaked extracellular organosulfur compounds (Figure 4). The sulphate ABC transporter encoding genes (SsuABC) were present in all the analysed strains except Nocardioides sp. JS614.

The boundaries of the described gene cluster showed the presence of genes encoding for transposases, integrases, mobile elements and also flanking direct repeats in most of the studied genomes with exception of M. smegmatis JS623 (Figure 4). These genetic elements are known to be associated with genomic islands and lead to further mobilisation, deletion or/and insertion of a complete genomic region (Schmidt and Hensel, 2004; Juhas et al., 2009; da Silva Filho et al., 2018). These findings indicate that these mycobacterial studied strains have acquired the VC degrader feature via horizontal gene transfer in order to adapt and survive in the environment.

M. vinylchloridicum (vi.nyl.chlo.ri’di.cum. N.L. neut. n. vinylchloridicum, vinyl chloride; N.L. neut. Adj. vinylchloridicum, related to vinyl chloride).

Aerobic, fast growing actinobacterium that develops colonies with yellow-orange colour, after 5 days of incubation on DSMZ 65 and 250, LJ and MB7H10 media. Optimal growth is observed after 3 days of incubation at 28°C on DSMZ 250 medium, pH 7. It is able to metabolise D-fructose, D-glucose, D-mannitol, glycerol and myo- inositol, D-trehalose (carbon source); acetic acid, butyric acid, β-hydroxy-butyric acid, citric acid, bromo-succinic acid, l-malic acid, propionic acid, sodium formate and methyl pyruvate (organic acids); and D-serine and L-arginine (amino acids). It is resistant to nalidixic acid, rifamycin sv, vancomycin and able to grow in the presence of aztreonam, guanidine hydrochloride, lithium chloride, 1% sodium lactate and tetrazolium blue, tetrazolium violet, and up to 4% (w/v) NaCl. It produces alkaline phosphatase, arylsulfatase after 3 and 14 days, catalase and heat stable catalase and reduce potassium tellurite. Whole-cell hydrolysates are rich in meso-diaminopimelic acid and galactose, glucose, mannose and ribose as cell sugars. The polar lipid pattern of strain L1^T contains diphosphatidylglycerol, phosphatidylethanolamine, phosphatidylinositol, phosphoglycolipid, unidentified glycolipids and unknown phosphoaminolipid. The mycolic acid profile of strain L1^T contains α-mycolate, methoxymycolate and ketomycolate. The major fatty acids (>10%) consist of C_16:0, C_18:1 ω9c, C_17:1 ω 7c/18 alcohol and 10Me-C_18:0. The G + C content is 66.6 mol% and the genome size is 7.1 Mbp.

The type strain L1^T (DSM 6695^T = CECT 8761^T) was isolated from vinyl chloride polluted soil, collected at Arnhem, Netherland. The GenBank accession number of the 16S rRNA gene is MT478173. The Whole-Genome Shotgun project has been deposited at DDBJ/ENA/GenBank under the accession JACBJQ000000000. The version described in this paper is version JACBJQ010000000.