Background

Antarctica is largely covered by glacial ice sheets, with ice-free areas making up ~0.32% of the entire continental land mass [1]. Of these ice-free areas, 90% are located along the continental coastline and occur on the Antarctic Peninsula and the Ross Sea Region (RSR). Soils of the RSR are exposed to a wide range of environmental extremes including physical extremes of temperature and elevated ultraviolet (UV) radiation, as well as geochemical extremes of high salinity, low water and low nutrient availability [2]. Together these environmental conditions make Antarctic soils some of the harshest environments on Earth.

Prior to the advent of molecular ecology techniques, cultivation- and microscopy-based studies had reported that Antarctic soils are dominated by a few, cosmopolitan groups of bacteria [3,4]. However, modern molecular methods have allowed for a more accurate assessment of bacterial community composition. Pyrosequencing and clone libraries of the 16S rRNA gene from RSR soils have identified representatives of 15 phyla including Acidobacteria, Actinobacteria, Armatimonadetes, Bacteroidetes, Chloroflexi, Cyanobacteria, Deinococcus-Thermus, Firmicutes, Gemmatimonadetes, Nitrospira, Planctomycetes, Proteobacteria, Spirochaetes, Verrucomicrobia and Candidate ‘TM7’ [5-13].

While modern techniques have aided our understanding of bacterial diversity in Antarctic terrestrial environments, they have also revealed adaptive mechanisms of psychrophilic organisms through genomic data. At the time of writing, genomes of 46 psychrophilic/psychrotolerant bacteria and archaea were complete and published (as reviewed by De Maayer et al. [14]). Of these, just four studies have investigated psychrophilic/psychrotolerant organisms isolated from Antarctic environments including Methanococcoides burtonii from Ace Lake, Vestfold Hills [15,16], Exiguobacterium antarcticum from microbial mats, Lake Fryxell [17], Octadecabacter antarcticus from Antarctic sea ice [18], and Cellulophaga algicola from the surface of a sea-ice diatom Melosira, East Antarctica [19]. These studies have reported the presence of cold adaptation relating to membrane modification, compatible solute accumulation, reactive oxygen species (ROS) detoxification, and significant changes in bacterial protein sequences including reduction in charged residues, hydrophobic clusters and proline content. These genomes have provided an insight into the lifestyle of psychrophilic microorganisms in permanently cold environments (<5°C) such as the Antarctic marine and lacustrine environments. However, there remains a gap in our knowledge of the adaptive mechanisms of psychrotolerant bacteria isolated from Antarctic terrestrial environments wherein large temperature fluctuations are common. In soils of the RSR, members of the Acidobacteria, Actinobacteria, Bacteroidetes, Deinococcus-Thermus and Proteobacteria phyla represent dominant taxa [5-7,9-12]. Of these, Actinobacteria are of special note. The phylum Actinobacteria is composed of phylogenetically diverse organisms that have been primarily investigated for their ability to cause disease in plants and animals, to produce anti-microbial compounds and anti-tumour agents, and to degrade recalcitrant molecules in soil environments [20]. Within Actinobacteria, members of the genus Arthrobacter are of note as they are among the most frequently isolated bacteria, occurring most commonly in soils and environments contaminated with industrial chemicals and radioactive materials. Their ubiquity can be attributed to their nutritional versatility and their resistance to environmental stressors [21]. At the time of writing, Arthrobacter included 82 species with validly published names (http://www.bacterio.net/a/). Of these, complete and published genomes are available for just six species, namely A. aurescens TC1, A. chlorophenolicus A6, Arthrobacter sp. FB24, A. nitroguajacolicus Rue61a, A. phenanthrenivorans Sphe3, and A. arilaitensis re117 [22].

In soils of the RSR, Arthrobacter species are both readily cultured and commonly observed in 16S rRNA gene clone libraries. Furthermore, they can be dominant in the soil environment as observed in soils on the Hatherton Drift, Transantarctic Mountains [5]. Despite their prevalence in soils of the RSR, very little is known about the physiological traits that allow these organisms to survive, flourish and establish dominance in the harsh Antarctic soil environment. A key question is if Antarctic strains of Arthrobacter owe their resilience to substantial genomic changes compared to Arthrobacter spp. isolated from temperate soil environments. Therefore, the three objectives of this study were (1) to investigate the abundance and diversity of Arthrobacter species found in soil microbial communities at four locations in the RSR, (2) to compare genomes of seven Antarctic Arthrobacter isolates with seven temperate Arthrobacter spp., focusing on traits that may contribute to survival and growth in the Antarctic soil environment, and (3) to investigate the metabolic versatility and salinity tolerance range of Antarctic Arthrobacter isolates compared to three temperate, soil-dwelling Arthrobacter species. For this, a combination of genotypic and phenotypic techniques including quantitative PCR (qPCR), whole genome sequencing and BIOLOG’s Phenotype Microarray (PM) technologies were utilised. To our knowledge this is the first study to provide genomic and phenotypic insights into the metabolic potential and ecological role of Arthrobacter strains isolated from RSR soils.

Results and discussion

Abundance and diversity of Arthrobacter spp. in soils of RSR

A total of eight samples from two soil depths at four locations within the RSR was investigated by qPCR (Figure 1) to determine the relative abundance of members of the phylum Actinobacteria and genus Arthrobacter. Specificity of the qPCR assays were tested by clone library preparations with DNA from three Antarctic soil samples and a marine sponge sample. All sequenced clones belonged to the correct target group. The relative abundance of each bacterial taxon was calculated as a ratio of measured copy numbers for each taxon-specific qPCR assay to measured copy numbers for the ‘all Bacteria’ assay. Members of Actinobacteria and Arthrobacter were present at all sample sites. Actinobacteria represented approximately 10-40% of the bacterial community at all soil locations, consistent with published data based on 16S rRNA gene clone libraries (as observed in Figure 1) [5-9,23]. In this study, up to 4% of the bacterial community was comprised of Arthrobacter species, with the lowest relative abundance observed in soils of Minna Bluff and the highest in soils of Granite Harbour. These results were also broadly consistent with published data (as observed in Figure 1) [5-9,23].

Figure 1
figure 1

Relative abundance of bacteria belonging to the phylum Actinobacteria and genus Arthrobacter in soils of RSR determined by 16S rRNA gene clone libraries [ 5 - 7 , 9 , 23 ] and by qPCR data (this study). Black bar represents Actinobacteria, grey bar represents Arthrobacter. Bar represents standard deviation for qPCR data. Sample sites: SB, Scott Base; MP, Marble Point; MB, Minna Bluff; GH, Granite Harbour; LV, Luther Vale.

Full size image

Phylogenetic analysis was performed on 16S rRNA gene sequences of clones and isolates associated with Arthrobacter from soils of the RSR. This analysis revealed that clones and isolates clustered together, clearly illustrating that the cultured isolates are representative of Arthrobacter spp. observed in soil bacterial communities (Additional file 1). Of these, seven isolates that represent the phylogenetic diversity of Arthrobacter spp. occurring in RSR soils were chosen for subsequent comparative analyses by whole genome sequencing and phenotypic characterisation.

Genome analyses

Genome overview

Genomes of the seven Antarctic Arthrobacter isolates are composed of the chromosome, each constructed from varying numbers of DNA scaffolds ranging from 53 to 158. The completeness of the seven genomes ranged from 78-98%, assessed by the occurrence of essential, single-copy genes. Due to this incompleteness, one must regard with caution the apparent absence or low copy number of a given gene. General genome features of the seven Antarctic Arthrobacter strains, as compared with the seven temperate Arthrobacter spp., are listed in Table 1. The genome G + C content for the seven Antarctic strains ranges from approx. 61-65%, similar to that for the seven temperate Arthrobacter species. The genomes contain 3,429-4,772 open reading frames (ORFs) with an average coding density of 88%. While genomes of strains I3, H5 and H14 contain 4,703, 4,552 and 4,566 protein-coding sequences (CDSs) respectively, significantly fewer CDSs were observed in genomes of strains H20 (3,466), 35/47 (3,470), Br18 (3,575) and H41 (3,373) (P < 0.01). For each genome, of the total CDSs, approx. 74% of the CDSs were classified into clusters of orthologous groups (COG) categories. Notably, the highest percentage of genes was assigned to COG categories such as amino acid transport and metabolism [E], carbohydrate transport and metabolism [G], and transcription [K] (Figure 2). Putative horizontally transferred genes, identified by the Joint Genome Institute (JGI) annotation pipeline, constituted 1.53-3.79% of the total genes observed. All Antarctic strains harboured a high number of genes associated with mobile genetic elements (up to 2.4% of total genes) encoding for phage integrases, transposases, and other phage elements (Additional file 2). Two putative phage sequences were identified in the genome of strain Br18, and genomes of strains H20 and H14 contained one putative phage sequence each (Additional file 3) [24]. No phage sequences were identified in the remaining four Antarctic strains. Significantly fewer copies of the 16S rRNA gene were identified in Antarctic Arthrobacter genomes as compared to the temperate Arthrobacter spp. (P <0.05).

Table 1 General genome features of seven Antarctic Arthrobacter strains (this study) vs. seven temperate Arthrobacter spp. [22]
Full size table
Figure 2
figure 2

Comparison of gene content in seven temperate Arthrobacter spp. and seven Antarctic Arthrobacter isolates by COG categories. Asterisks represent abundant COG categories. Letters K, G, and E represent COG categories transcription, carbohydrate and amino acid transport and metabolism, respectively.

Full size image

General genome comparisons

General comparisons between genomes of seven temperate Arthrobacter spp. and seven Antarctic Arthrobacter strains were carried out using CMG-Biotools [32]. Firstly, amino acid usage was calculated for all 14 Arthrobacter isolates using their protein sequences. The amino acid usage tree (Figure 3) shows three main clusters. This clustering pattern is similar to the clustering observed in the maximum likelihood tree based on 16S rRNA gene sequences (Figure 4). This analysis revealed that Ala, Gly, Leu, and Val are the most frequently used amino acids across all Arthrobacter genomes. Predicted proteome comparisons and a pan- and core-genome plot analysis were also performed on all 14 Arthrobacter genomes using CMG-Biotools [32]. Proteomes were predicted for each isolate using Prodigal [33] and then BLAST algorithm (Basic Local Alignment Search Tool)-based proteome comparisons were performed to identify whether proteins are shared between predicted proteomes [34]. In Figure 5, the main part of the matrix (shaded green) consists of pairwise proteome comparisons and the bottom row (shaded red) represents a self-comparison where a hit within the proteome to a protein other than the query is identified as an internal homolog or a paralog. The BLAST matrix illustrates that conservation between Antarctic Arthrobacter genomes (24.4-53.6%) is low as compared to genomes of Arthrobacter spp. isolated from temperate soil environments (42.2-76.6%) (excluding A. castelli that was isolated from the biofilm of a mural). This observation was also supported by the pan- and core-genome plot analysis (Additional file 4), as the Antarctic core- and pan-genome comprised 1,285 and 10,873 gene families respectively and the temperate core- and pan-genome comprised 1,559 and 9,798 gene families respectively. Approximately 4.8% of the total CDSs were in paralogous clusters across all Arthrobacter genomes. The final Arthrobacter pan-genome comprised 14,902 sequences, indicative of a large diversity of accessory genes. The Arthrobacter core-genome is comprised of 1,153 gene families, representing approx. 27% of the total genes in Arthrobacter genomes. A large proportion of genes in the Antarctic and temperate core-genomes were assigned to COG categories including amino acid transport and metabolism [E], carbohydrate transport and metabolism [G] and translation, ribosomal structure and biogenesis [J].

Figure 3
figure 3

Amino acid usage heatmap of seven temperate Arthrobacter spp. and seven Antarctic Arthrobacter isolates based on their protein content. The percentage of amino acid usage was plotted in gplots using R. Amino acids highlighted in bold face represent abundant amino acids.

Full size image
Figure 4
figure 4

Maximum likelihood phylogenetic tree based on 16S rRNA gene sequences from seven Antarctic Arthrobacter isolates and seven temperate Arthrobacter species. Filled circles indicate bootstrap support of >90%, and open circles represent >75% support (maximum parsimony, 1000 resamplings). Bar, 0.11 substitutions per compared nucleotide site. Outgroup comprises Microbacterium maritypicum, AM 181506 and M. profundi, EF623999.

Full size image
Figure 5
figure 5

BLAST matrix of an all against all protein comparison of 14 Arthrobacter genomes. The blue box contains genomes of seven temperate Arthrobacter spp. and the red box contains genomes of seven Antarctic Arthrobacter isolates. #, Note that proteins represent total CDSs and families represent CDSs not in paralogous clusters (unique CDSs). Red diamonds represent protein families in paralogous clusters.

Full size image

Genomic traits linked to environmental stress-related adaptation

Several known adaptive mechanisms for growth and survival in cold terrestrial environments were identified in the genomes of all 14 Arthrobacter isolates and are summarised in Table 2 (Additional file 5), and in the following categorical descriptions.

Table 2 List of genes linked to environmental stress response
Full size table

Sigma factors

Sigma factors are dissociable units of bacterial RNA polymerase that control the conditional expression of a specific set of genes in response to a particular stress or stimulus. Copies of genes of the σ70 factor, RpoD, more commonly referred to as the house-keeping or general stress response sigma factor, were identified in all Antarctic and temperate Arthrobacter genomes. The presence of multiple copies of genes for RpoD is a common feature in psychrophilic bacteria such as Planococcus halocryophilus [35] and Psychromonas ingrahamii [36]. Further analysis also revealed the presence of several copies of genes for the σ24 factor, RpoE, associated with regulating cellular responses to heat shock and other stresses on cell membrane and periplasmic proteins in all Antarctic and temperate Arthrobacter genomes. In Escherichia coli, RpoE also regulates cell lysis in a prolonged stationary phase, thus providing nutrients for the next generation of cells [37].

Oxidative stress response

Antarctic Arthrobacter genomes contain several copies of genes encoding (putative) oxidases that contribute to the abundance of endogenous H2O2 and other ROS (Additional file 6). Furthermore, ROS are formed at a higher abundance as a result of increased oxygen solubility at low temperatures [38]. Consequently, combating free radical damage alongside surviving exposure to UV radiation is crucial for survival in the Antarctic soil environment. Genomes of the Antarctic Arthrobacter isolates contain several genes that confer protection from free radical damage and allow for detoxification of ROS. This includes up to two copies of the superoxide dismutase gene, sodA and up to three copies of the catalase gene that were identified in all Antarctic Arthrobacter genomes. Additionally, several copies of the peroxiredoxin gene, bcp and thioredoxin genes, trx, trxB were identified in all Antarctic Arthrobacter genomes. Genes for antioxidant activity have also been identified in other cold-adapted bacteria including Colwellia psychrerythraea [39], Desulfotalea psychrophila [40], and P. halocryophilus [35]. These genes were identified in genomes of all temperate Arthrobacter species. Further analysis revealed the presence of up to two copies of the redox-sensitive transcriptional activator gene, soxR in isolates H14, H20 and H5. However, the regulatory gene, soxS was absent in all Antarctic and temperate Arthrobacter genomes. In E. coli, SoxR regulates the expression of transcription factor, SoxS in response to H2O2 and other superoxide compounds. SoxS in turn activates the expression of several superoxide stress response genes. SoxR is observed without SoxS in many organisms including Pseudomonas aeruginosa, P. putida and Streptomyces coelicolor. For these bacteria, it is hypothesized that the SoxR homolog regulates the redox-active secondary metabolite, pyocyanin, which is involved in redox homeostasis [41]. However, as genomes of the Antarctic Arthrobacter strains lack homologs of genes involved in phenazine biosynthesis, we hypothesize that the SoxR regulon may directly interact with proteins involved in the superoxide stress response.

Osmotic stress response

Metabolites including compatible solutes, cryoprotectants, exopolysaccharides (EPS) and other protective polysaccharides can confer resistance to environmental stressors including UV radiation, osmotic stress, oxidation, and desiccation, thus playing a crucial role in the Antarctic soil environment. Glycogen and trehalose protect the cell from desiccation, osmotic stress and cold shock, and under nutrient limiting conditions can also serve as a source of carbon [42]. The glycogen biosynthesis pathway (from ADP-D-glucose) comprises three steps catalyzed by three enzymes, glucose-1-phosphate adenylytransferase (glgC), ADP-glucose type (glycosyl-transferring) starch/glycogen synthase (glgA) and glycogen branching enzyme (glgB). Genes (glgA, glgB and glgC) for the entire glycogen biosynthesis pathway were identified in all Antarctic and temperate Arthrobacter genomes. Genes encoding glycogen degradation enzymes, glucan phosphorylase (glgP) and glycogen debranching enzyme (glgX), were also identified in all Antarctic and temperate Arthrobacter genomes [43]. Genes for trehalose biosynthesis, trehalose-6-phosphate synthase (otsA) and trehalose-6-phosphate phosphatase (otsB) were present in all Antarctic and temperate Arthrobacter genomes. In E. coli, trehalose synthesis by enzymes OtsA and OtsB is induced by cold shock and is essential for cell viability [44]. In the psychrotolerant bacterium Arthrobacter strain A3, trehalose serves as a source of carbon, allowing cells to maintain normal metabolism at prolonged low temperatures [45]. Additional pathways for trehalose biosynthesis from maltose by the enzyme trehalose synthase (treS), and from maltodextrin by malto-oligosyl trehalose synthase (treY) and malto-oligosyl trehalose trehalohydrolase (treZ), were also identified in all Antarctic and temperate Arthrobacter genomes. Genome analysis has also revealed the presence of genes encoding a number of ABC-type transporter systems for mediating cytoplasmic accumulation of organic compatible solutes including choline, glycine betaine and proline (proV, proW, proX) [46] in all Antarctic and temperate Arthrobacter genomes. Genome surveys of psychrophilic bacteria and archaea also revealed the presence of multiple genes for the uptake and/or synthesis of compatible solutes, illustrating the importance of these compounds for osmoprotection and cryoprotection in cold environments [47,48].

A large number of genes involved in several sugar biosynthesis pathways were identified in all Antarctic and temperate Arthrobacter genomes (Additional file 7). EPS produced by members of the genera Pseudoalteromonas, Shewanella, Polaribacter, and Flavobacterium isolated from the Antarctic marine environment was largely composed of neutral sugars including glucose, fucose, and mannose, amino sugars including N-acetyl galactosamine and N-acetyl glucosamine and uronic acids including galacturonic and glucuronic acids. In these bacteria, EPS may protect cells from the extremes of low temperature and high salinity in the marine environment [49]. EPS production has also been reported in psychrophilic bacteria such as C. psychrerythraea [39] and Psychromonas ingrahamii [36]. The presence of a wide range of osmoprotection systems suggests that the Antarctic Arthrobacter isolates are well equipped to cope with desiccation and osmotic stress.

Cold shock response

Cold shock can result in the inhibition of bacterial cell growth and proliferation as a result of stabilization of DNA and RNA secondary structures, reduction in membrane fluidity and solute uptake. Therefore, upon exposure to a sudden temperature downshift, bacteria respond with a specific pattern of transient gene expression of members of a family of small, nucleic acid-binding cold-shock proteins (CSPs). CSPs are of note as they regulate messenger RNA (mRNA) translation, rate of mRNA degradation and transcription termination, all of which are dependent on the ribosome that is targeted by cold shock [50-52]. Several copies of the DNA-binding cold-shock proteins were identified in all Antarctic and temperate Arthrobacter genomes similar to the multiple copies observed in C. psychrerythraea [39], Psychrobacter arcticus [53] and Shewanella oneidensis [54]. Additionally, genes encoding cold-shock-inducible proteins: ribosome-binding factor A (rbfA), translation initiation factors, IF-1 and IF-2 (infA, infB), polynucleotide phosphorylase (pnp), RNA-binding cold-shock domain A (csdA) and NusA, N-using substance protein A (nusA) were identified in the Antarctic and temperate Arthrobacter genomes [55]. Homologs of the CspA-like cold acclimation protein, CapA, were identified in genomes of four Antarctic Arthrobacter isolates, 35/47, H14, H5 and I3. Unlike CSPs that are transiently expressed upon cold shock, in A. globiformis S155, CapA was over-expressed following cold shock and during prolonged growth at low temperatures [51]. The presence of the CapA homolog in Antarctic Arthrobacter genomes may explain the survival of these organisms at prolonged low temperatures. Genes for CapA were also identified in genomes of all temperate Arthrobacter species.

Cell membrane adaptations

In the Antarctic soil environment, maintaining membrane permeability and fluidity at sub-zero temperatures is crucial for continued cell viability. Genome analyses of Antarctic Arthrobacter isolates have revealed the presence of several genes for fatty acid desaturases (des), which are important in this context. Upon cold shock, the expression of des genes is regulated by a two-component signal transduction pathway, comprising a membrane-integrated histidine kinase that senses change in membrane fluidity and a response regulator that binds the promoter region of des genes. Together they increase the production of desaturases that add double bonds into pre-existing fatty acid tails within the membrane, thus restoring membrane fluidity. Genes for desaturases were also observed in genomes of all temperate Arthrobacter spp., similar to the copies observed in the psychrotolerant bacterium, Exiguobacterium sibiricum [56]. In Bacillus subtilis, the DesKR system regulates the expression of the des gene coding for ∆5 desaturase [57,58]. All Antarctic Arthrobacter isolates contained several copies of genes associated with signal transduction mechanisms (approx. 3% of the total CDSs) (Additional file 8). A similar proportion of genes (approx. 4% of the total CDSs) were attributed to signal transduction mechanisms in genomes of the temperate Arthrobacter species.

Carotenoid biosynthesis also contributes to cold adaptation by stabilizing the cell membrane, maintaining proton permeability and by promoting oxidative stress resistance [59,60]. Pigmentation is a common feature of Antarctic Arthrobacter isolates. Genes involved in the synthesis of lycopene (red-coloured) have been identified in Antarctic Arthrobacter strains H41 and Br18. Decaprenoxanthin (yellow-coloured) biosynthesis genes have been identified in Antarctic Arthrobacter strains H14, H20, and 35/47 [61]. These findings are in agreement with the coloured pigments produced by these isolates (data not shown). Genome analysis of P. halocryophilus also revealed the presence of genes for lycopene biosynthesis. It is hypothesized that these genes are responsible for the bright orange colouration observed in P. halocryophilus cultures [35]. The decaprenoxanthin biosynthesis pathway includes genes, crtE (geranylgeranyl pyrophosphate synthase), crtI (phytoene desaturase) and crtB (phytoene synthase) for the production of lycopene and genes, crtEB (lycopene elongase) and crtYef (lycopene epsilon cyclise) for the production of decaprenoxanthin. Antarctic Arthrobacter isolates I3 and H5 only contained copies of genes crtE and crtI. In contrast, just two of the seven temperate Arthrobacter spp., A. aurescens TC1 and A. castelli DSM 16402, contained genes involved in decaprenoxanthin biosynthesis.

Phenotypic characterisation of Antarctic Arthrobacter strains

Seven Antarctic Arthrobacter isolates and three temperate, soil-dwelling Arthrobacter species were selected for phenotypic characterisation and comparative analysis by BIOLOG PM1 and PM2A plates assessing carbon utilisation and plate PM9 assessing salinity tolerance.

Carbon utilisation profiles

Antarctic Arthrobacter isolates demonstrated significantly lower metabolic versatility as compared to temperate Arthrobacter isolates (P <0.05), as the temperate species were able to utilise 123-140 different C sources and the Antarctic strains were only able to utilise 98-121 different C sources (Additional file 9). The diverse patterns of C utilisation were further reflected in the MDS plot as four separate clusters were observed (Figure 6a).

Figure 6
figure 6

MDS plots of BIOLOG substrate usage data of seven Antarctic Arthrobacter isolates and three closely related temperate Arthrobacter species. a. Carbon utilisation profile determined by BIOLOG PM1 and PM2A assay plates. b. Salinity tolerance profile determined by BIOLOG PM9 assay plate. Antarctic strains: I3, H14, H5, H20, 35/47, Br18, H41. Temperate strains: A. chlorophenolicus A6, A. phenathrenivorans Sphe3, Arthrobacter sp. FB24.

Full size image

Antarctic Arthrobacter isolates were able to utilise C substrates that form components of bacterial and fungal cell walls [62,63] including carbohydrates, N-acetyl glucosamine, glucose, mannose, xylose, arabinose, rhamnose and amino acids, Ala, Glu, Gly, and Lys, glycogen and trehalose that serve as microbial carbon and energy storage molecules [43,64] and sucrose, mannitol and arabitol that serve as storage molecules in lichens [65]. With the exception of ornithogenic soils formed under penguin rookeries, soils of the RSR typically contain low concentrations of organic C, ranging from 0.01 to 0.96 mg C g−1 soil [66]. Recalcitrant compounds including pectin and lignin are limited in Antarctic soil environments due to the absence of higher plants [67]. Instead, Antarctic soil environments largely contain C compounds as a result of aeolian distribution of organic C derived predominantly from: (i) cyanobacterial mats and mosses from lacustrine and marine ecosystems, (ii) endolithic microbial communities, and (iii) soil-dwelling mosses, lichens and microbial communities [66]. In addition, the Antarctic isolates lacked the ability to utilise plant cell wall components including carbohydrate, allose; carboxylic acids, citric acid, 4-hydroxybenzoic acid and galacturonic acid; and polymer, pectin, all of which are typically absent in RSR soils [68]. In contrast, all three temperate Arthrobacter spp. were able to utilise these C sources.

Salinity tolerance patterns

Antarctic Arthrobacter isolates showed a significantly narrower salinity tolerance range as compared to temperate Arthrobacter species (P <0.01) (Additional file 10). Differences in salinity tolerance are clearly illustrated in the MDS plot where two separate clusters of temperate and Antarctic Arthrobacter spp. are observed (Figure 6b). A positive phenotype for all Antarctic Arthrobacter isolates was observed with up to 7% NaCl, 4% KCl and 5% sodium sulphate and up to 100 mM ammonium sulphate (pH 8) and 100 mM sodium nitrate. Salinity is a key feature of soils in the RSR with some soils, such as those at lower Taylor Valley, containing 62.04 kg of salts m-2 [69]. The salt composition of RSR soils is largely dependent on geographic location as coastal soils are largely comprised of chloride and nitrate salts while inland soils are dominated by sulphate and nitrate salts [67]. As a result, an ability to withstand high concentrations of these salts is essential for survival in the soil environment. As compared to temperate Arthrobacter spp., all Antarctic strains showed significantly reduced tolerance to organic salts, Na formate (<1%) and Na lactate (<1%), which are likely to be absent in Antarctic soil environments.

Conclusions

We have undertaken a comparative genomic study of seven Antarctic and seven temperate Arthrobacter spp. to identify genomic features that may be essential for growth and survival in the Antarctic terrestrial environment. Genomes of all Antarctic Arthrobacter isolates contained several features that are also observed in psychrophilic/psychrotolerant bacteria and archaea. These included genes for sigma factors, ROS detoxification, osmoprotection systems, cold shock response and a carotenoid biosynthesis pathway. However, a large proportion of these genes were also identified in temperate Arthrobacter spp., suggesting that these genes may be important for growth and survival in a range of soil habitats. Further investigations by transcriptomic- and proteomic-based techniques, as previously reported for A. chlorophenolicus [70,71] and A. phenanthrenivorans [72], are essential to reveal the expression profiles of these genes as well as identify novel traits or genes that are crucial for cold adaptation.

It should be noted that relative to temperate species, and notwithstanding the incompleteness of the genome assemblies, four Antarctic Arthrobacter isolates (Br18, H20, H41 and 35/47) contained significantly fewer CDSs. Phenotypic characterisation assessing carbon utilisation profiles of these isolates revealed lowered metabolic versatility. In addition, of the total CDSs identified in the genomes of these isolates significantly fewer CDSs were assigned to COG categories, transcription [K] and carbohydrate transport and metabolism [G] (P < 0.01). The fewer CDSs, the lowered metabolic versatility, and the significant reduction in CDSs associated with transcription and carbohydrate transport and metabolism, suggest the occurrence of genome content scaling in four of the seven Antarctic Arthrobacter isolates. This occurrence was also reported in three strains of Paenibacillus darwinianus that were isolated from gamma-irradiated soils of the RSR [73,74]. In bacteria, an increase in genome size is often linked with an increase in metabolic versatility, allowing bacteria to produce new enzymes that exploit a wide range of environmental conditions [75]. However, an increase in versatility is linked with a four-fold increase in regulatory proteins associated with transcription and two-component signal transduction systems [76,77]. In environments such as soil, efficient regulation of enzyme expression, enabling exploitation of scarce yet diverse, complex nutrients can offer a selective advantage, thus lowering the penalty of slow growth. This growth strategy is common amongst dominant bacteria in soil environments [78]. In the Antarctic soil environment, organic residues are scarce yet labile, with C and N being mineralisable within a relatively short period of time (90 days) under optimal conditions [79]. In the harsh Antarctic soil environment, maintenance of metabolic versatility comes at a higher cost and, more importantly, reproductive efficiency (promoted by smaller genomes containing fewer CDSs) is beneficial for survival and growth. Finally, these genome sequences allow further investigations into the expression of physiological traits that enable survival under extreme conditions and, more importantly, into the ability of these bacteria to respond to future perturbations including climate change and human impacts.

Materials and methods

qPCR analysis

Site descriptions and sampling strategy

Soil samples were collected from four sites: Scott Base (77°55’S, 166°45’E), Granite Harbour (77°24’S, 162°31’E), Minna Bluff (78°31’S, 166°46’E) and Marble Point (77°25’S, 163°41’E). To the best of our knowledge all sites sampled were free of recent anthropogenic disturbances, with a possible exception of foot traffic. Soil classification and descriptions are as previously described [80]. At each site, three pits (~ 50 m from each other) representing biological replicates were dug. Following removal of the desert pavement, four soil samples (~ 15 g each), two samples at depths 0-5 cm and two at 5-10 cm, representing technical replicates were collected. Soil samples were stored in LifeGuard™ Soil Preservation Solution (MoBio), 2 ml of solution per g soil, frozen at -20°C and transported back to New Zealand for processing. For long-term storage, samples were frozen at -80°C.

DNA extraction

DNA was extracted in duplicate (0.5 g soil/tube) from one technical replicate at each depth at all sample sites by mechanical cell disruption (bead-beating) as previously described [81]. DNA extracts were quantified by Qubit® dsDNA BR assay kit (Life Technologies) and their purity (A260/A280) was assessed on a NanoDrop™ ND-1000 Spectrophotometer (Biolab).

qPCR validation and protocol

qPCR assays were designed for the selective amplification of: (1) all Bacteria, (2) phylum Actinobacteria, (3) genus Arthrobacter. Primer sequences for each reaction are listed in Additional file 11. Specificity of all primer sequences was tested in silico using the Probe Match tool in the RDP-Release 10 (Ribosomal Database Project) classifier program [82]. Primer reactions were optimised with genomic DNA from the following pure cultures: Arthrobacter spp. H5, H14, H20 and 35/47, Nocardioides sp. D26, Pontibacter sp. D8, Modestobacter sp. Br44, Paenibacillus darwinianus BrT, and Escherichia coli NZRM 916. Finally, primer specificity was tested by clone library preparations as described by Aislabie et al. [8] with DNA from: three Antarctic soil samples (Scott Base, Minna Bluff and Marble Point) and a marine sponge sample (Rhopaloeides odorabile). Twenty clones per DNA sample were selected for identification by 16S rRNA gene sequencing.

All qPCR reactions were performed in 10 μl reaction volumes in 384-well clear optical reaction plates (Applied Biosystems) on a 7900HT Fast Real-Time PCR System (Applied Biosystems). Each reaction contained: 5 μl of Platinum® Green qPCR SuperMix-UDG with ROX (Life Technologies), 0.2 μM forward and reverse primers for all Bacteria and Actinobacteria-specific assays; 0.5 μM forward and reverse primers for Arthrobacter-specific assays, 1 μl template DNA (1 ng/μl) and ddH2O up to 10 μl. General qPCR cycling conditions were 2 min at 50°C, 2 min at 95°C, followed by 40 cycles of 95°C for 15 s and 1 min at the respective annealing temperature. Annealing temperatures were individually optimized for each assay. Each plate included triplicate reactions for all DNA samples, DNA standards and no-template controls. Melting curve analysis was performed at the end of each reaction to confirm the attribution of the fluorescence signal to a specific PCR product and not primer-dimers or other artifacts. DNA standards were prepared from a linear plasmid containing the entire 16S rRNA gene, amplified and cloned as described by Aislabie et al. [8] from genomic DNA of Arthrobacter sp. FB24 (NR_074590).

Standard curves and amplification efficiencies were calculated using 7900HT SDS software version 2.4 as described elsewhere [83]. Correlation coefficient (R2) and amplification efficiency (E) across all qPCR assays were >0.99 and 1.80-1.86, respectively.

Phylogenetic analyses of Arthrobacter from soils of the RSR

Phylogenetic analysis

Nearly full-length 16S rRNA gene sequences of Arthrobacter spp. from soils of the RSR were obtained from GenBank, aligned via the SINA web aligner and imported into the ARB phylogenetic package using the SILVA 108 database for analysis by the maximum likelihood, RAxML method. The topology of the tree was tested in MEGA v6.0 by bootstrap analysis based on 1000 resamplings [84,85].

Selection of isolates

Seven Arthrobacter isolates from RSR soils including: I3, H5, H14, H20, 35/47, Br18 and H41 were obtained from the Antarctic culture collection of Dr Jackie Aislabie (Landcare Research, Hamilton). These isolates were routinely cultured on R2A agar (Difco™, BD) at 15°C. Temperature range data for isolate 35/47 were determined by growth on R2A plates for two weeks at a range of temperatures, namely 5, 10, 15, 18, 20, 25, and 37°C.

DNA extraction and sequencing

High molecular weight DNA was extracted from seven Arthrobacter isolates: I3, H5, H14, H20, 35/47, Br18 and H41 using a modified CTAB (hexadecyltrimethylammonium bromide) and protein lysis method [86]. Briefly, cells were scraped off R2A agar plates and re-suspended in 740 μl TE buffer containing 20 μl lysozyme (100 mg/ml), 40 μl 10% SDS and 8 μl proteinase K (20 mg/ml). These cells were incubated overnight at 37°C. Following overnight incubation, 100 μl of 5 M NaCl and CTAB/NaCl solutions were added to the reaction and incubated at 65°C for 10 min. Subsequently 0.5 ml chloroform:isoamyl alcohol (24:1) was added, and the entire reaction was centrifuged at 16,000 g for 15 min. The aqueous phase was transferred to a clean microcentrifuge tube containing 0.5 ml phenol:chloroform:isoamyl alcohol (25:24:1) and the reaction was centrifuged at 16,000 g for 15 min. The aqueous phase was transferred to a clean microcentrifuge tube containing 0.6 vol isopropanol. To allow for DNA precipitation, all reactions were incubated at room temperature for 60 min, then centrifuged at 16,000 g for 30 min. The resulting DNA pellet was washed with 70% ethanol and re-suspended in TE buffer containing RNAse (99 μl TE buffer + 1 μl RNAse (10 mg/ml)) and incubated at 37°C for 20 min. DNA extracts were quantified by Quant-iT™ PicoGreen® dsDNA assay kit (Life Technologies) and their purity (A260/A280) was assessed on a NanoDrop™ ND-1000 Spectrophotometer (Biolab). Additionally, quality of each DNA extract was tested by electrophoresis on a 1% agarose gel.

Following extraction, high molecular weight DNA was sent to Macrogen (Seoul, South Korea) for sequencing on the Illumina HiSeq 2000 platform using 100 bp paired end libraries. With a sequencing output of 35 Gb, estimated coverage was 700X per genome.

De novo assembly, annotation and comparative analyses

FASTQ files for each genome were trimmed and quality filtered using the FASTQ Trimmer tool of the FASTX-toolkit v0.0.13 [87] and Sickle (https://github.com/ucdavis-bioinformatics/sickle) respectively. High-quality reads (Q > 30) were assembled into contigs using Velvet v1.2.10 [88]. Following initial assembly, PAGIT (post assembly genome improvement toolkit) tools including IMAGE (iterative mapping and assembly for gap elimination) and iCORN (iterative correction of reference nucleotides) were utilized for gap elimination and sequencing error correction [89]. Finally, SSPACE basic version 1.0 (stand-alone scaffolder of pre-assembled contigs using paired-read data) was utilized to build scaffolds from assembled contigs [90]. Genome completeness was assessed as previously described [91]. Gene prediction and genome annotation was performed using the automated JGI pipeline (Joint Genome Institute) [92].

For comparative analyses, A. aurescens TC1, A. castelli DSM 16402, A. chlorophenolicus A6, A. globiformis NBRC 12137, A. nitroguajacolicus Rue61a, A. phenanthrenivorans Sphe3 and Arthrobacter sp. FB24 were selected based on their genome completeness and habitat. These analyses were performed using JGI-IMG (Integrated Microbial Genomes)-Expert Review [85]. CMG biotools was utilised for pan- and core-genome plot analysis and predicted proteome comparisons [32]. Putative phage sequences were identified by the Phage Search Tool (PHAST) [24].

Availability of supporting data

This whole genome shotgun project has been deposited at DDBJ/EMBL/GenBank under the following accession numbers: 35/47-AZHY00000000, H14-AZRX00000000, H20-AZRY00000000, H41-AZRZ00000000, Br18-AZSA00000000, I3-AZSB00000000 and H5-AZSC00000000.

Phenotypic characterisation by BIOLOG

Salinity tolerance and carbon substrate utilisation

Carbon utilisation (PM1, PM2A) and salinity tolerance (PM9) were tested by Phenotype Microarray (PM) technology (BIOLOG). Ten Arthrobacter isolates, including seven Antarctic isolates: I3, H5, H14, H20, 35/47, Br18 and H41 and three temperate isolates: A. phenanthrenivorans DSM 18606 T, A. chlorophenolicus DSM 12829 T and Arthrobacter sp. FB24 DSM 22572T, were included for characterisation. Type strains of the temperate bacteria were obtained from DSMZ, Germany. Antarctic isolates were routinely cultured on R2A agar plates at 15°C and temperate isolates were routinely cultured on TSB (tryptic soy broth) agar (Bacto™, BD) plates at 28°C. Prior to inoculation of PM plates, Antarctic isolates were grown at 15°C for three days on R2A agar plates and temperate isolates were grown at 28°C for 12 h on TSB agar plates. PM plates were inoculated with 150 μl of cell suspension, prepared to the appropriate cell density as per the manufacturer’s instructions. PM plates inoculated with Antarctic isolates were incubated at 15°C and PM plates with temperate isolates were incubated at 28°C. Colour development (OD600nm) on all plates was recorded at 24 h intervals on an EnSpire® multimode plate reader over 10 days for Antarctic isolates and over four days for temperate isolates.

Data analysis

Isolate response profiles were built based on the response of each isolate to each substrate, determined as either positive or negative (binary data – usage/non-usage). Response profiles were analysed with the multivariate statistics package, Primer-E v6 (UK) by non-metric multi-dimensional scaling (MDS) plots on Bray Curtis similarity matrices constructed from transformed data (presence/absence).