An Antarctic bacteria strain (ID So64.6b) previously isolated from soil on Agar M1 (peptone 2.0 g/l, yeast extract 4.0 g/L, starch 10.0 g/L and agar 18 g/L; Lamilla et al., 2017) was used in this study. For storage, the bacterium was cultured in 3 ml of International Streptomyces Project-2 medium (ISP-2; yeast extract 4.0 g/L, malt extract 10.0 g/L, glucose 4.0 g/L, and agar 18 g/L) and cryopreserved at -80°C with 20% glycerol. Culture for DNA extraction was obtained by inoculating the isolated bacterial strain on the ISP-2 agar at 15°C for 7 days.

Genomic DNA extraction was performed using the UltraClean Microbial DNA Extraction Kit (MoBio Laboratories, Carlsbad, USA). DNA quality was assessed by fluorescence quantification using the QuantiFluor® ONE dsDNA System kit (Promega) on a Quantus device (Promega). Also, integrity was verified by 1% agarose gel and purity according to 260/280 and 260/230 absorption ratios. The sample was divided into two and sequencing using Illumina and Oxford Nanopore Technologies (ONT) technologies. The Illumina library was prepared with 2x150bp fragments using the Nextera XT DNA Sample Prep kit (Illumina, San Diego, CA) and sequenced on an Illumina MiSeq platform. The ONT library was prepared using the Rapid Sequencing kit SQK-RBK004 and sequenced on the MinION platform of the Extreme Environments Biotechnology Lab (Universidad de La Frontera, Chile), according to the manufacturer’s recommendations and using the MinKNOW software v.4.0.20. The basecalling of the ONT reads was performed with Guppy v3.1.5 software.

Assembly was performed using the two data libraries (i.e., Illumina and ONT). The quality of Illumina reads was verified by FastQC (Andrews, 2010), and fastp (Chen et al., 2018) was used to filter poor-quality reads using default parameters. The quality of ONT reads was verified with nanoPLOT (De Coster et al., 2018). Adapter and low quality reads (phred >10, size >5,000 bp) were removed with Porechop and NanoFilt (De Coster et al., 2018), respectively. De novo assembly was performed using the Unicycler software v.0.4.8 (Wick et al., 2017) with default parameters for hybrid assembly with short + long reads. The whole genome results of this study were deposited in the DDBJ/ENA/GenBank database under the accession CP048817.

Whole genome taxonomic identity was confirmed using Average Nucleotide Identity (ANI) values. The ANI value was calculated among the available Sphingomonas representative complete genomes on NCBI database (×49) using Pyani v0.2.10 software in ANIb mode (Pritchard et al., 2016). The reference genome with highest ANI value was used for DNA–DNA hybridization (DDH) estimation values, which was computed using the genome-to-genome distance calculator (GGDC v3; Meier-Kolthoff et al., 2022). In addition, phylogenetic distances between the closest taxonomic species were also determined by constructing the core proteome of the available representative complete genomes on the genus Sphingomonas (×49). First, core proteins were obtained with the web server M1CR0B1AL1Z3R¹ (Avram et al., 2019) using an 80.0% cutoff value for the minimum identity of proteins found in all compared genomes. Next, an unrooted phylogenetic construction of the core proteome was performed with a maximum likelihood algorithm using RAxML (Stamatakis, 2014) based on the inferred proteome alignment. Phylogenetic tree visualization was performed with the iTOL v3 tool (Letunic and Bork, 2016).

Finally, for same-species strains, the single enriched functions for each strain were studied based on Gene Ontology using the OrthoVenn2 server with default parameters (accessed on 12/15/2020; Xu et al., 2019). Protein sequences obtained from Prokka annotation (Seemann, 2014) were used as input. Briefly, similar proteins are obtained by all-against-all alignment with DIAMOND v0.9.24 (Buchfink et al., 2014) with a threshold of e-value = 0.05. Then, clusters are generated with MCL (inflation value = 1.5) and gene ontology (GO), and enrichment is assigned by similarity with the UniProt protein database (Apweiler et al., 2004). GO enrichment analysis computed the p-values for GO terms in a clusters overlapping using a hypergeometric distribution with the variables N = total count of all GO terms, M = count of GO terms in the overlap, n = total count of a GO term and x = count of the GO term that overlap on the following formula:

Genetic elements associated with the production of secondary metabolites (BGCs) on the So64.6b genome were identified using the antiSMASH v.5.1.0 tool (Blin et al., 2019) with default parameters and selecting all prediction features. Additionally, structure prediction of BGCs was performed on PRISM 3 (Skinnider et al., 2017). The closest known BGCs were assigned when matches were found compared to the MiBiG Cluster Repository of Known Biosynthetic Gene Clusters.

Clusters associated with the production of potentially novel antibiotic secondary metabolites were predicted with ARTS v2.0 (Mungan et al., 2020) with default parameters. A search for known resistance elements is performed by comparing with the curated databases: ResFams, The Comprehensive Antibiotic Resistance Database (CARD), The LACtamase Engineering Database (LACED), and The Jacobi and Bush Collection. In addition, essential genes are identified with TIGRfam Equivologs as the core genes from a reference set of genomes and classified according to: (a) rare duplication (possible resistant target genes), defined as all essential genes that exceed the median and standard deviation thresholds of the sum of essential genes identified in the bacterial family; (b) acquisition by horizontal transfer when comparing the topology of core gene phylogeny against species tree; and (c) their proximity to the BGCs identified with ANTISMASH 3 software (Mungan et al., 2020). This allowed the prediction of the mode of action of encoded compounds of an uncharacterized BGC based on the identification of resistant target genes (Mungan et al., 2020), improving the selection of the strains with the highest potential for antibiotic metabolites discovery.

The genome of the Antarctic strain So64.6b consists of 5,570,175 pb. Phylogeny based on core proteome showed that strain So64.6b is genetically close to S. alpina, followed by Sphinogmonas sp. strains AAP5 (isolated from alpine lake), HMP6 (isolated from ultra-oligotrophic lake in East Antarctica), and the species S. aliaeris DH-S5 (obtained from pork steak) and S. panacis (obtained from rusty ginseng root in South Korea based on Genbank repository records; (Singh et al., 2015; Sanangelantoni et al., 2018; Kim et al., 2020). The phylogeny shows a closer genetic relation of the Antarctic strain with Sphingomonas alpina DSM 22537 (S8-3T strain), a species isolated from the Alpes (Margesin et al., 2012; Figure 1). Both strains also share a 99.6% identity of 16S rRNA, assuming they belong to the same species. However, ANI values have no coincidence over the threshold that defines species (>95–96%) since they share 86% identity. Most similarities are between 76% and 78% with other species of Sphingomonas. (Supplementary Figure S1). The result confirms that S. alpina is the closest species, but in silico DDH calculates a 38% hybridization between S. alpina and our Antarctic isolate genome. Hence, our work suggests that S. alpina is a close relative of Sphingomonas sp. So64.6b, a potentially novel species of the genus Sphingomonas. Given the phylogenic results and to further explore the relation and differences that define each strain, we conducted an ortholog clusters analysis.

Both Sphingomonas strains shared 3,385 ortholog clusters, while 98 clusters were unique for the So46.6b strain and only 52 for the Alpes strain (Supplementary Figure S2). The Antarctic strain presented an enrichment of the transposition functions (value of p = 0.00076) and catabolism of aromatic compounds (value of p = 0.00019), while the Alpes strains have the greatest function in the oxidoreductase activity (value of p = 0.00061). Clusters on the So46.6b strain showed 10 transposases–grouped in two ortholog clusters– and four insertion elements related to genetic mobility. In addition, it contains multiple proteins for the catabolism of polluting agents such as benzoate, gallate, lignin, xylene, and tetrahydrofolate (Supplementary Table S1). That could be related to the presence of these phenolic pollutants in the Antarctic environments, where a higher rate of redox reactions on pollutants/heavy metals at a cold temperature might increase their harmful potential (Ju et al., 2017; Zangrando et al., 2019). Alpine strain does not have any of these genes and showed only proteins related to the catabolism of xylene and lignin, commonly found in the Sphingomonas genus (Menon et al., 2019). Therefore, we suggest genomic differences among the strains might be in part caused by a large accessory genome related with adaptation to the harsh Antarctic environment.

The antiSMASH results are summarized in Table 1. Sphingomonas sp. So46.6b has a total of 6 BGCs related to polyketides synthases (PKS), terpene, two RiPP Recognition Elements-containing (RRE), lassopeptide, and a hybrid cluster lassopeptide-homoserine lactone. None of the BGCs have similarities to previously reported clusters from neither Sphingomonas nor other bacteria genres, excluding the terpene, which presented a low similarity (30%) to the known carotenoid BGC. This result indicates the potential of new untapped natural products from the Antarctic strain. Furthermore, BGCs prediction with the ARTS tool demonstrates that three of those clusters were also associated with the production of potentially novel antibiotic compounds due to their proximity with duplicated essential genes (×13), known resistant genes (×36), and recent horizontal gene transfer events (×189).

Particularly we highlight the lassopeptide BGC-4 of Sphingomonas spSo64.6b as a potential novel antibiotic since it does not match with any other known BGC and has a low similarity with any other genetic cluster (Table 1, 23% shared genes with S. leidyi). Moreover, this cluster contains an essential gene related to resistance prediction due to duplication and Horizontal Gene Transfer (HGT) events, identified as a hypothetical protein (TIGR02118). A gene with this characteristic could be a resistant copy of the antibiotic target acquired by horizontal transfer along with the identified BGC (Alanjary et al., 2017). Most BGCs showed resistance genes related to typical functions of antibiotic activity, mainly in amino acid biosynthesis, protein production and transport, DNA synthesis, and central metabolism (Table 1). Although those BGCs might also produce unknown antibiotic compounds with usual targets, the BGC-4 cluster could be involved in more specific unknown activities; then, its lassopeptidic products could represent a new class of antibiotics with an unknown target molecule. The study of unknown targets for discovering new kinds of broad-spectrum antibiotics is considered one of the main strategies to fight against the antibiotic crisis (Matos De Opitz and Sass, 2020).

In addition, two molecular structures were predicted on PRIMS from the BGCs of So64.6b (Figure 2), one belonging to BGC-4. Both structures corresponded with lassopeptides types and showed high molecular weights (over 2,000 g/mol), which escapes from the detection spectrum of the antibiotic metabolites previously reported for this strain (Núñez-Montero et al., 2020). Such high molecular weight and predicted structure stand out the high molecular complexity of the compounds eventually produced by the BGC-4. This complexity in the lassopeptides structure appeals to their engineering regarding thermal stability, proteolytic protection, and chemical properties, promoting bioactive activities (Maksimov and Link, 2014).

Since we hypothesize that the Antarctic environment might be partially responsible for the unknown diversity of BGCs found on the strain So64.6b, a phylogenetic analysis using CORASON was performed to identify genetic divergences of a common BGC between the Antarctic bacteria and other closely related Sphingomonas strains. This analysis is an alternative for prioritizing potentially novel BGCs based on comparative phylogenomics. Instead of comparing between full and partial merged BGCs as done with other tools, CORASON avoids artifact errors by a glocal-type comparison, which first finds the longest common subcluster between the PFAM of a pair of BGCs (including our Antarctic query strain) and continues with an extension of this alignment applying match/mismatch penalties (Navarro-Muñoz et al., 2020). The analysis was performed using the BGC-6 of strain So64.6b since this was the only cluster shared with other Sphingomonas species, including Antarctic strains, and matching a known BGC on MiBIG database. The synteny of genes was based on the carotenoid cluster, specifically by comparing the gene ctrl, coding for the phytoene desaturase enzyme. This enzyme converts phytoene or squalene to lycopene by desaturation in four places.

The result shows a unique and shared syntenic organization of the BGC across nine strains of the genus Sphingomonas, where 54–68% of the genes were found in all the strains (Figure 3A). Furthermore, the phylogenetic result showed a clear grouping of the cluster in three clades, with the Antarctic strain cluster being the most distant one from the rest of the sequences (Figure 3B). In this context, it is interesting to notice that the Antarctic So64.6b strain shares a clade with strains from similar extreme environments, including the PAMC strain obtained from the Arctic (Shin et al., 2012) –and the AAP strain, isolated from an Alpine lake (Kopejtka et al., 2020). Such environments shared extreme conditions such as high altitude, high UV radiations and exposure to freeze/thaw cycles. Besides, strains related to different environments, such as those isolated from plant microbiome or soil, are grouped apart from those of extreme environments (Figure 3B). The production of pigments has been reported in other Antarctic organisms to cope with high UV radiation (Rodrigues et al., 2021; Singh et al., 2022). Then, this carotenoid variations might play a role on the resistance to UV radiation. The result supports our hypothesis showing a genetic divergence on an Antarctic strain’s BGC; while such genetic differences might drive the variations observed in the biosynthetic modules, highlighting its potential for finding new molecules. In addition, that genetic divergence could be associated with the adaptation of this bacteria to an extreme environment. Specialized metabolites have evolved to mediate important ecological functions, explaining their diversity of chemical products (van Bergeijk et al., 2020). Our hypothesis established that some of those metabolites might be driven by the Antarctic environmental harsh conditions. Other authors have reports findings that are consistent with this theory. Among then, chromomycin, an antibiotic produced by Streptomyces flaviscleroticus showed antioxidant effect that confers resistance oxidative stress (Prajapati et al., 2019); the lipid-based esters, and Cys-GSH isomers were related to antioxidant response, significantly shifted by Cadmium concentrations in Scenedesmus obliquus (Mangal et al., 2020); secondary metabolites production and additional genes for specific peptides were found to potentially enhance the response to environmental changes on members of the order Candidatus Thermoprofundales, enabling them inhabit marine hydrothermal vents, deep subsurface oil reservoirs and hot springs (Liu et al., 2022); and the products of shikimate pathway including mycosporines and mycosporines-like-amino-acids which are well-known radiation resistant substances (Sarkar et al., 2018; Gacem et al., 2021).

To further support this idea we checked whether the environmental conditions influences the overall composition and diversity of BGCs on Sphingomonas strains by comparing the predicted BGCs within this genus and between the representative genomes of Sphingomonas species isolated from different environments available on the Genbank database. An average of four clusters were found on 49 genomes analyzed, ranging from one in Sphingomonas sp. LM7, to eight in Sphingomonas sp. AP4-R1 (Figure 4). Previous studies regarding genome-mining of proteobacteria species also showed that the most common types of BGC were nonribosomal peptide synthetases (NRPS), PKS, and hybrids; however, lanthipeptides and terpenes are also present (Brotherton et al., 2018). NRPS and PKS are peptide biosynthesis machinery of natural products with multiple bioactivities (Giordano et al., 2015). Likewise, lanthipeptides and terpenes have been related to antimicrobial activity compounds (Giordano et al., 2015; Oladipo et al., 2022).

The total amount of clusters in the strains was variable; then, to understand whether it can be affected by the origin of each sample, they were grouped by source of isolation from animals (×6), plants (×5), soil (×19), water (×12) or from an unknown source (×7) when it was not reported on Biosample Genbank information. The samples were also classified into extreme or mesophilic environments regarding the expected conditions of the region of the isolation source. Strains from water showed the greatest average of the total amount of predicted clusters (×4), while animal-isolated bacteria were the lowest. Nonetheless, soil isolated bacteria have the maximum number of clusters (×8), while plant isolated bacteria showed the lowest maximum (×4). In addition, animal-isolated bacteria had the greatest genetic content devoted to BGC in their genome (6.02%) compared to strains isolated from plants (2.66%), soil (2.60%), and water (4.20%; Figure 5C). This result suggests that the total amount of BGCs in the Sphingomonas might be related to the conditions they face in their specific samples. Also, strains from animals seem to have fewer BGCs, but with more genetic content on each cluster, which could be associated with a stronger regulation due to the interspecies communication or additional biosynthetic genes for modifications on the scaffold molecule to produce a larger number of derived products. Besides, soil and water samples were related to a larger amount of BGCs, mainly NRPS, PKS, and ribosomally synthesized and post-translationally modified peptides (RiPPs). These results are similar to other studies on Streptomyces mining for specialized metabolism and marine Pseudoalteromonas pangenome analysis (Cordero and Polz, 2014; Maansson et al., 2016; Nicault et al., 2020)

In the environment-type context, seven bacteria were classified as isolated from extreme environments, 35 strains were isolated from mesophilic environments, and seven could not be classified. Bacteria from extreme environments had the greatest average of BGC (five), while mesophilic environments had the lowest (three). However, the same maximum number of clusters (eight) were found on strains from all environments. Interestingly, bacteria isolated from extreme environments also presented the greatest average of genetic content devoted to BGC (4.84% vs. 4.17% in mesophilic environment, Figure 5D). The results also indicate a major capacity for production of secondary metabolites in Sphingomonas isolated extreme environments, compared to the average percentage of 3.7% (± 3.1%) reported for proteobacteria phyla (Cimermancic et al., 2014).

To explore whether those differences were also found for particular types of BGCs –because of their different functionalities– the amount of each type of BGCs across the genomes was compared and related to the isolation origin. More than 16 types of BGCs were found across the Sphingomonas genomes. Interestingly, a maximum number of five clusters associated with the type of RRE-containing metabolites were found on species isolated from water. Conversely, other BGC types were less commonly found in the sum of similar samples, and, particularly, the RRE-containing type was present in less than two isolated species in other types of samples. In contrast, the plant isolated species did not have this type of BGC (Figure 5A). Besides, a higher number of lassopeptides-related clusters were found in soil and animal isolated species (four and three, respectively); one was identified in water and an unknown source, while none was present for plant isolated species (Figure 5A). In addition, the greatest amount of lassopeptides and RRE-containing metabolites were found in mesophilic environments (five and four, respectively). On the other hand, strains from extreme environments exclusively present clusters related to lanthipeptide class II and redox cofactor (Figure 5B). These can be associated with the adaption to the oxidative stress and low-temperature conditions of the Antarctic environment (Simpson and Codd, 2011; Grimalt-Alemany et al., 2021). The overall results showed that the number and some types of BGCs found on Sphingomonas genus seems to be associated with the environment they inhabit. Then BGC content in thus genus might be drive as evolutionary response to the requirements of the specific environment. Particularly, the identified BGCs in our soil-extreme environment bacteria might be related to an ecological advantage instead of taxonomic characteristics. This is expected since in those environments, bacteria deal with multiple physico-chemical complex changes, requiring a plethora of physiological and metabolic response (Cordero and Polz, 2014). Here we highlight the relevance of elucidating new natural products from strain So64.6b –isolated from polyextreme soil–, as others of biosynthetic interest in the Sphinogmonas species.

Since the Antarctic strain So64.6b had characteristics related to the metabolite production potential, we continue to study if this strain has other genetic elements particularly related to its adaptability to the Antarctic environment explaining its divergent genome content. From the pangenome analysis of the Sphingomonas genus, 34,242 genetic families were obtained, 1,546 were unique for the Antarctic strain, including metabolic pathways for the degradation of aromatic compounds previously mentioned. The result evidenced a significant genetic diversity of the genus with a vast number of unique genes belonging to one species. This is consistent with similar studies on comparative genomics of Sphingomonas, where it has been suggested that the presence of this group in a wide diversity of environments has led to a large genetic difference and a small core genome for the genus (Kim et al., 2020).

We identify multiple ribonucleases, duplicated genes for the synthesis of asparagine, kinases, and aminotransferases of serine exclusively present on the genome of the Antarctic So46.6b strain. Considering that a great genetic variability might be associated with the adaptation mechanisms of Sphingomonas species to specific conditions, this can be part of adaptation to extreme temperatures. For example, it has been described that highly catalytic enzymes at low temperatures have a greater amount of asparagine and serine than their mesophilic homologs, with more arginine and lysine (Qin et al., 2014).

In addition, the Antarctic strain shows multiple unique genes related to antibiotic and heavy metal resistance, such as CzcC protein for cobalt, zinc, and cadmium resistance; CorA for cobalt and magnesium transportations; copper resistance and transportation proteins; receptors and iron transporters (FepC and FieF); regulators for the resistance and reduction of mercury; metalo-beta-lactamases; multidrug excretion pumps (EmrA, EmrB, AcrB, MedtA-C-G, MexA, NorM, Stp, OqxB13-17); SugE for quaternary ammonium resistance, and penicillin-binding proteins. These genes might also have an important role in the Antarctic environment. Studies have reported the presence of pollutants in the Antarctic continent, including heavy metals (cadmium, zinc, vanadium, arsenic, and gold), antibiotics, and pesticides transported by natural processes through the air and water (Corsolini, 2009; Lo Giudice et al., 2013). Also, in this place, active volcanoes contribute a significant amount of heavy metals to the soil (Singh et al., 2007). Regarding antibiotic resistance, it has been described that the introduction of foreign microorganisms due to tourism and scientific activities increased the acquisition of these genes by HGT (Cowan et al., 2011).

Our results highlight the genetic dynamism of the Sphingomonas genus. Specifically for the Antarctic strain, we reported enrichment of ortholog genes related to transposons and resistance genes, usually transferable by plasmids, and pointed out multiple resistance genes. To confirm whether those unique genes of the Antarctic strain could have been acquired by mobile elements, driving the genetic differences among the Antarctic strain and the Alpine strain (closer relative), we studied the genomic islands in its genome. More genomic islands were identified on the genome of the Antarctic strain (Figure 6A) and most of them were unique for this genome (Figure 6B). Within those regions, we identified the unique genes of this strain related to degradation of benzoate, hydrolysis of ferulic acids, resistance to copper, nickel, and cobalt, as well as multiple transposases and insertion elements. Meanwhile, genomic island on Alpine strain were mostly composed of phage genes, transposases or hypothetical proteins. Some were already present in the Antarctic strain and only one gene of antibiotic resistance were found (erythromycin esterase) as evidence of adaptability genetic material. Hence these results suggest that the differential genetic elements of the Antarctic strain So64.6b could be recently acquired by HGT environments to cope with the Antarctic poly-extreme conditions. Previous studies regarding genetic diversity showed a correlation between taxonomic distance and the production of secondary metabolites featuring unique bioactive natural products (Hoffmann et al., 2018). Also, similar results were observed in Pseudoalteromonas sp., a model of adaptation to extreme environments. A notable genomic diversity on this group has been attributed to recent acquisition a large set of mobile elements, mainly by HGT (Parrilli et al., 2021). The results support our hypothesis that some of the BGCs detected in this strain might be linked to the adaptation to harsh conditions of the Antarctic; for example, those without similarity to any other clusters within its genus or phylogenetically closer to strains from similar environments. Therefore, the Sphingomonas sp. So64.6b strain should be considered a new source for mining and bioprospecting antimicrobials compounds.

Strains belonging to the family Sphingomonadaceae have been explored due to their biotechnological applications in bioremediation and degradation of refractory pollutants and in the production of valuable biopolymers called sphinganes (Huang et al., 2009), which have an important role as growth promoters in plants. Particularly the genus Sphingomonas has been characterized as a degrader of multiple environmental pollutants, so, in addition to its role as a soil conditioner, it has a relevant ecological role in the recycling of complex molecules (Bhatt et al., 2020; Zhang et al., 2020). Also, due to its capacity to store heavy metals, it has improved plant resistance in environments with high zinc, cadmium, and copper contents, in addition to collaborating in stress conditions such as drought (Asaf et al., 2020). This background agrees with the genomic results observed in the Antarctic strain, showing a high capacity for resistance to metals and degradation of pollutants, which appear to be unique to the strain and absent even its closest relative. This highlights the influence that the selective pressure of the Antarctic environment may have exerted on the evolution, shaping a likely new species; and the genetic conformation of the Antarctic strain under study related with an essential role it probably plays in the Antarctic soil by protecting other organisms from harmful compounds. Moreover, the high genetic variation observed in this work, which has been previously reported for the genus Sphingomonas, together with the unique BGCs, make this Antarctic strain an important target in the searching for new metabolic pathways and the discovery of natural products.