Root-Associated Bacteria Community Characteristics of Antarctic Plants: Deschampsia antarctica and Colobanthus quitensis—a Comparison

Colobanthus quitensis (Kunth) Bartl. and Deschampsia antarctica Desv. are the only Magnoliophyta to naturally colonize the Antarctic region. The reason for their sole presence in Antarctica is still debated as there is no definitive consensus on how only two unrelated flowering plants managed to establish breeding populations in this part of the world. In this study, we have explored and compared the rhizosphere and root-endosphere dwelling microbial community of C. quitensis and D. antarctica specimens sampled in maritime Antarctica from sites displaying contrasting edaphic characteristics. Bacterial phylogenetic diversity (high-throughput 16S rRNA gene fragment targeted sequencing) and microbial metabolic activity (Biolog EcoPlates) with a geochemical soil background were assessed. Gathered data showed that the microbiome of C. quitensis root system was mostly site-dependent, displaying different characteristics in each of the examined locations. This plant tolerated an active bacterial community only in severe conditions (salt stress and nutrient deprivation), while in other more favorable circumstances, it restricted microbial activity, with a possibility of microbivory-based nutrient acquisition. The microbial communities of D. antarctica showed a high degree of similarity between samples within a particular rhizocompartment. The grass’ endosphere was significantly enriched in plant beneficial taxa of the family Rhizobiaceae, which displayed obligatory endophyte characteristics, suggesting that at least part of this community is transmitted vertically. Ultimately, the ecological success of C. quitensis and D. antarctica in Antarctica might be largely attributed to their associations and management of root-associated microbiota.


Introduction
Antarctica is a place of extremes [1]. Only two species of Magnoliophyta (flowering plants) managed to establish their presence in this remote and life-challenging region: a member of the family Poaceae-Deschampsia antarctica Desv. (Antarctic hairgrass) and Colobanthus quitensis (Kunth) Bartl. (Antarctic pearlwort), belonging to the family Caryophyllaceae [2]. Their vast distribution in maritime and coastal Antarctica has baffled scientists for decades and was a subject of many debates [2][3][4]. Originally from South America, those two Magnoliophyta were suspected of being migratory relics from the Oligocene-Pliocene colonization event [2], while other data hints towards their more recent arrival during the late Pliocene [5]. Nonetheless, their ecological success in harsh Antarctic conditions is undisputed and attributable mostly to the extensive adaptations to many stress factors like cold, freeze-thaw cycles, UV radiation, drought, flooding, high concentration of sodium, and varying levels of nutrient concentrations from severe deficit to extreme over manuring [4]. To cope with environmental stresses, Antarctic plants have evolved adaptations in terms of molecular, cellular, and physiological functions to boost their cold tolerance by increasing their metabolic rates upon cold acclimation [6], such as stress protein production, upregulation of antioxidant synthesis, and accumulation of compatible solutes like amino acid proline and non-structural carbohydrates [7,8] and secretion of antifreeze proteins into their apoplasts [9]. They also developed low temperature-efficient photosynthetic and respiratory systems whose maximal activity occurs at 13 °C, and 30% of its maximal photosynthetic capacity is retained at 0 °C [10,11]. Moreover, such ontogenesis-related features of Antarctic flowering plants like long life cycles, an extended primordium development of leaves and flowers [12], or diverse propagation strategies based on seed production derived from self-pollination and less often cross-pollination (C. quitensis and D. antarctica) or vegetative reproduction (only D. antarctica) enabled continuous habitat range expansion of those plants since 1960, presumably aided by the ongoing global climate warming [13,14]. However, the geographical and physical isolation of Antarctica still enforces low intra-species genetic diversity by restricting gene flow between Antarctica and South America into those populations [15,16].
The relations between plants and microbes are now fairly well understood for the flora of temperate and tropical regions [17] but are somewhat under researched for polar regions, especially for Antarctica [18]. It is the consensus that microbial communities associated with the host plants' roots exert the greatest influence on the plant health and development, most notably those that reside within two distinct rhizocompartments: the rhizosphere and the root endosphere [19]. Rhizospheric microbes inhabit the rootadjacent soil and feed on the root-derived organic exudates, whereas the endosphere community consists largely of plantspecific endosymbionts selectively recruited from the rhizosphere [20]. Some studies indicate that rhizospheric and endospheric microbial communities are not only spatially separated but also display different niche characteristics [21]. Rhizospheric bacteria are mostly engaged in mineral substrate dissolution by secreting organic acids (often gluconic acid), siderophores, or even cyanides consequently providing the plant with bioavailable phosphorus and biogenic metals. Their main role also includes biocontrol of soil-borne pathogens by competing with them for resources with the use of antimicrobial agents like bacteriocin-like peptides, different classes of antibiotics, and lytic enzymes: chitinase or β-1,3-glucanase. Rhizobacteria also have a direct input into root-growth stimulation by secretion of phytohormones such as indole-3-acetic acid. Endospheric bacteria on the other hand are engaged in plant stress responses mainly by reducing the inner-plant stress hormone ethylene levels through ACC deaminase-mediated hydrolysis. Endophytes also have a major contribution in plant-growth enhancement by providing fixed nitrogen, which is the main limiting element in soil environments [22].
As mentioned, studies on the Antarctic plant-associated microbiome (especially the bacterial part) are largely underrepresented in literature [23]. Available but scarce data allow us to draw only limited conclusions on microbiological phenomena connected to Antarctic Magnoliophyta [24][25][26]. This is largely due to the low resolution of the techniques used but also due to low sample diversity (mostly from only one particular site) or even exclusion of a vital part of the community, namely the endosphere, from analysis. Therefore, in this study, we employed cultivation-independent methods to compare root-associated microbial communities of C. quitensis and D. antarctica sampled from sites displaying contrasting edaphic, ecologic, and microclimatic characteristics. Our hypothesis states that the phylogenetic and metabolic diversity of microbial communities associated with native Antarctic plants is greater in sites of high stress-inducing factor intensity than in those providing milder growth conditions. To gain the necessary knowledge on the differences and similarities within the rhizospheric and endosphere microbial communities of Antarctic flowering plants, we made use of the high-throughput 16S rRNA gene fragment targeted sequencing (assessing the bacterial communities phylogenetic diversity) and community-level physiological profiling by Biolog EcoPlates.

Sites and Sampling
Samples were collected during the austral summer season of 2017-2018 from three sites on King George Island, South Shetland Islands, maritime Antarctica (Table 1). Several specimens (4-6 per site) were collected with the root adjacent soil with the use of sterile tools into sterile plastic containers and transported frozen (− 20 °C) to the laboratory in the Institute of Biochemistry and Biophysics, Polish Academy of Sciences (IBB, PAS). Additionally, bulk soil samples from those sites were gathered in triplicates (approximately 1.5 kg per site) for component analysis and transported in the same conditions.

Measurement of Soil Components
Soil pH (in 1 M KCl) and salinity (in double-distilled water) were measured with a CPC-411 Elmetron™ multiparameter probe [27]. Phosphates and nitrates were determined spectrophotometrically in a Shimadzu UV 1601 spectrophotometer and in an Epoll-Eco 20 spectrophotometer, respectively. Other elements were determined by atomic absorption spectroscopy [28].

Bacterial Extraction
Bacterial cells were extracted from rhizospheric soil and roots samples from nine C. quitensis individuals (three per site) and respectively the same number of D. antarctica individuals. The following method was devised based on the findings of Lunau et al. [29] regarding the separation of prokaryotic cells from mineral and organic debris and the guidelines provided by Szymańska et al. [30] regarding root-associated microbe isolation. To analyze the microbiome of the root-adjacent soil, a sample of the soil was carefully removed from between the roots with a sterile spatula onto a pre-sterilized aluminums foil piece. Approximately 1 g of the soil was weighed and placed in a 50-mL conical tube containing 20 mL of sterile and cool (4 °C) dilution liquid composed of 0.9% (w/v) NaCl and 10 mM tetrasodium pyrophosphate (Na 4 P 2 O 7 ). The suspension was then shaken for 30 min in a Tornado™ Vortexer at 2000 rpm at 4 °C. The tubes were then placed in a VWR Ultrasonic Cleaner USCTH filled with chilled water and sonicated for 60 s. The tubes were vortexed afterwards for 30 s to suspend detached cells. After brief centrifugation (1 min; 1000 rpm; 4 °C), the suspension was submitted to metabolic fingerprinting by the Biolog EcoPlate technique. To detach the rest of the adhering soil, the root system was washed in 60 mL of sterile NaCl/ Na 4 P 2 O 7 solution by shaking for 30 min in the aforementioned shaker (1000 rpm; 4 °C) and then rinsed 3 times in 5 mL of the same sterile and cooled solution by vortexing. Washed roots were sterilized by incubation in a cooled 10% hydrogen peroxide (H 2 O 2 ) solution for 5 min and then rinsed 3 times with sterile NaCl/Na 4 P 2 O 7 solution. The so surface-sterilized roots were placed in a pre-cooled sterile mortar. Two and a half milliliter of sterile NaCl/Na 4 P 2 O 7 was added with 0.6 g of sterile, sharp garnet sand (lysing matrix A, MP Biomedicals) and gently ground with a pestle, allowing the sharp angular garnet pieces to comminute the roots to an amorphous pulp. The pulp was transferred to a 50-mL conical tube containing 20 mL of sterile and cool (4 °C) NaCl/Na 4 P 2 O 7 solution and submitted to the above-mentioned procedure (shaking, ultrasonication, and vortexing). The resulting supernatant suspension was submitted to metabolic fingerprinting by the Biolog EcoPlate technique and DNA extraction. All reagents used in the bacterial extraction steps were molecular biology grade and were purchased from Sigma Aldrich.

DNA Extraction
Rhizosphere soil DNA was extracted using the PowerSoil® DNA isolation kit (QIAGEN GmbH, Hilden, Germany) according to manufacturer protocol. An approximately 0.2 g of soil was used. DNA solutions were kept at 4 °C for further analysis. The dilution liquid containing endosphere bacteria was passed through a sterile 47-mm Whatman polycarbonate filter (0.22 µm pore size). The DNA from the filter-trapped bacteria was extracted using the PowerWater® DNA isolation kit (QIAGEN, GmbH, Hilden, Germany) according to manufacturer protocol and kept at 4 °C.

16S rRNA Gene Amplification
The phylogenetic study was performed by targeted sequencing and analysis of the prokaryotic 16S ribosomal RNA gene. A fragment of the 16S rRNA gene containing the V3 and V4 variable regions was amplified using gene-specific primers: 16S_V3-F and 16S_V4-R positions 341-357F and 785-805R, respectively, according to Escherichia coli 16S rRNA gene reference sequence [31]. Illumina Nextera XT overhang adapter nucleotide sequences were included in addition to the 16S rRNA gene-specific sequences, which allowed sample indexing and pooling. Each PCR was conducted in triplicates using KAPA HiFi PCR kit (Roche, Basel, Switzerland) in a final volume of 20 µL per reaction according to the manufacturer's instructions.

Amplicon Sequencing
Obtained PCR products were pooled into 12 samples (2 rhizocompartments × 3 sampling sites × 2 plant species) in equimolar ratio and indexed using Nextera XT barcodes (Illumina, San Diego, CA, USA). Amplicon libraries were sequenced on Illumina MiSeq instrument (Illumina, San Diego, CA, USA) in the DNA Sequencing and Oligonucleotide Synthesis Laboratory (IBB, PAS). Sequencing was conducted in paired-end mode (2 × 300 bp) with the use of a v.3 (600 cycles) chemistry cartridge, which allowed the generation of long paired reads fully covering 16S V3-V4 amplicons.

Phenotype Fingerprinting with Biolog EcoPlate™
The Biolog EcoPlate assay determines the ability of a mixed microbial community to use any of 31 carbon compounds as the sole carbon source (plus a single control well with nocarbon). Microbial communities were characterized for their ability to catabolize 10 different carbohydrates, 9 carboxylic and acetic acids, 4 polymers, 6 amino acids, and 2 amines [32]. Root-associated bacterial suspensions were adjusted with sterile 0.9% NaCl to optical transmittance of 0.9. One hundred microliter aliquots of each suspension were added to each well of EcoPlate microplates (Biolog Inc., Hayward, CA, USA). The plates were incubated in darkness at 10 °C. The temperatures were chosen to accommodate the activity range of the resident microbial communities: the psychrophiles and the psychrotrophs [33]. The color development was read at 590 nm (A 590 ) in a Varioskan plate reader (Thermo Fisher Scientific, Waltham, MA, USA), and cellular respiration was measured kinetically by determining the colorimetric reduction of tetrazolium dye. Data were collected approximately twice a week over a 65-day period. The prolonged incubation of EcoPlates was based on our previous observations [34][35][36]. Data from the 36th day of incubation were used as there was no further color development after this date. Final absorbance data were first blanked against the time zero reading and then blanked against the respective control well containing no carbon source. Readings that had the A 590 value of 0.25 or higher were scored as a positive EcoPlate response (PER).

Data Analysis
Raw sequencing data were cleaned, aligned, and classified automatically by the EzBioCloud platform using the PKSSU4.0 database [37]. Chimeric, low quality, and nontarget (chloroplast, mitochondrial, and archaeal) amplicons were automatically discarded. The operational taxonomic unit was defined as a group of sequences that exhibit greater than 97% similarity to each other. Illumina reads were deposited in the NCBI Sequence Read Archive (SRA) as BioProject PRJNA726953. All results were compiled using Excel 2016 (MS Office) for Windows. A two-sample t-test was applied to compare different data sets. Variance within the sets was assessed using the f-test beforehand. Correlations between biological and geochemical parameters were calculated using Pearson's correlation coefficient. Principal component analysis (PCA) was performed using the singular value decomposition method. Data visualization and statistical analysis have been performed using the R software (R v.4.0.2) and the following packages: ggplot2, fmsb, Hmisc, ggpubr, corrplot, and autoplot [38].

Results
Site 1 soil was characterized by high magnesium, very high calcium, and relatively high sodium contents and salinity, and also pH was high (8.3). Soil from site 2 had the highest potassium content among the examined soil samples with high copper and iron concentrations and a low pH (4.3). Site 3 soils had the highest concentration of nitrates, phosphates, manganese, and zinc with the lowest reported pH (4.0) (Fig. 1C).  (Fig. 3). photographs. C Sampling sites 1-3 geochemical composition of the soil; N, mg NO 3 /100 g soil; P, mg P 2 O 5 /100 g soil; K, mg K 2 O/100 g soil; Mg, mg Mg/100 g soil; Ca, mg Ca/100 g soil; Na, mg Na/100 g soil; salinity, g NaCl/L; Mn, mg Mn/kg soil; Zn, mg Zn/kg soil; Cu, mg Cu/kg soil; Fe, mg Fe/kg soil Several bacterial families were present in considerable amounts in the rhizosphere, although the occurrence of some of them was highly site specific (Fig. 4A). Relative abundance of the family Micrococcaceae showed severe differences for C. quitensis (av. 6.3%, sd. 10) as well as for D. antarctica   PCA showcased several phenomena within both plants' root-associated microbial communities (Fig. 6). The PCA clustering based on the relative abundance of family-rank groups indicated that the rhizosphere community differs in structure from the endosphere community, both for C. quitensis and D. antarctica. The bacterial communities of C. quitensis showed great differences between sampling sites, while those of D. antarctica clustered according to the rhizocompartment of origin. The situation was similar for the EcoPlate-based clustering, mainly for C. quitensis, where no apparent clustering was observed. Rhizospheric and endospheric D. antarctica communities formed a loose cluster in this analysis. PCA based on a combination of phylogenetic and physiological data revealed a clear distinction between C. quitensis and D. antarctica root-associated communities. Two tight clusters emerged, separately harboring the rhizospheric community and the endospheric community of D. antarctica, while for C. quitensis, there were no apparent similarities between the samples.
Significant differences in phylogenetic and physiological levels emerged between microbial communities of C. quitensis and D. antarctica (Fig. 7 A and B). In the rhizosphere of D. antarctica, relative abundance of the family Chitinophagaceae was significantly higher (p = 0.049) than in the C. quitensis rhizosphere. The endosphere communities of D. antarctica were significantly richer in sequences of the family Rhizobiaceae (p = 0.00091) and Sphingobacteriaceae (p = 0.0014). Phenylethylamine catabolism was significantly more pronounced in the D. antarctica root endosphere (p = 0.0073).
The endospheric core community was based on the relative abundances of family-rank groups (Fig. 7C). Those that displayed an abundance of 1% or higher were scored as core community members. Only five families were at sufficient abundance in the endosphere of both examined plant species: Sphingobacteriaceae (5.9-19.1%), Sphingomonadaceae (1.1-6.7%), Microbacteriaceae

Discussion
A large body of literature dedicated to root-associated microbiomes indicates that bacterial and fungal communities dwelling in the plant rhizosphere and endosphere are host-specific species [39][40][41]. Nonetheless, the physiological status of this host plant influences the phylogenetic structure and metabolic capabilities of the associated microbiome [42]. This physiological status however is dependent on the edaphic and climatic conditions experienced by the plant [43].
Our results show that Antarctic flowering plants shape their root-associated microbiome differently, resulting in divergent microbial communities. The microbiome of C. quitensis root system bears different characteristics in each of the examined locations. In site 1 characterized by low essential nutrient concentrations (N-P-K) and high salinity and pH, the rhizosphere contained a highly diverse bacterial community, both phylogenetically and metabolically. The family Micrococcaceae (Actinobacteria) seemed to be a vital component of the rhizosphere in such conditions [44]. Members of this family along with other Actinobacteria have been observed in large quantities in salt marsh plants' rhizosphere, indicating their stress alleviating effect in low water activity substratum [45,46]. The corresponding C. quitensis endosphere was occupied by a fraction of metabolically active opportunitroph bacteria as indicated by the relatively low OTU numbers accompanied by high numbers of positive EcoPlate responses. Those were mainly the Pseudomonadaceae and Oxalobacteraceae family members. Inoculation with different Pseudomonadaceae strains has improved the salt-tolerance of Zea mays, which was connected to the water-binding exopolysaccharides produced by those bacteria [47], whereas the Oxalobacteraceae were enriched by the same plant species in nitrogen-poor soils, stimulating lateral root growth, consequently increasing nitrogen compound and other resources acquisition [48]. Furthermore, as these two families harbor mostly copiotrophic bacteria that display a multitude of metabolic features [49], their relative abundances in the endosphere were significantly correlated with the catabolism intensity of some of the carbon sources, most notably plant cell wall components: D-cellobiose and D-xylose. Cellulases and xylanases are essential in allowing bacterial entry into plant roots [50,51]. At site 2, the structure of the root-associated communities diverges considerably from those at site 1. High phylogenetic diversity was accompanied by low metabolic activity and numbers of utilized carbon sources. This indicates that the community consists of either low activity bacteria or that respiratory activity was restricted at this location in C. quitensis rhizosphere. In tundra soils such as this one, nitrate or mineral nitrogen compounds may be deficient as most are bound in organic matter, which decompose extremely slowly under polar conditions [52,53]. Some hypothesize that in this scenario, plants might exhibit microbivory by releasing proteases into the rhizosphere to liberate the microbe-bound nitrogen but also by destroying the cells of the endosphere microbes through oxidizing agent production on root cell plasma membranes [54]. Furthermore, the examined endosphere contained bacteria that are not usually found in this rhizocompartment, namely those belonging to the phylum Acidobacteria and Gemmatimonadetes [20]. In this regard, active endocytosis has been detected in Arabidopsis thaliana, which internalized and digested non-endophyte microbes [55]. At site 3, the root-associated microbiome of C. quitensis displays yet a different structure. This site is particularly rich in nitrogen, not only in the examined nitrates, but also in ammonia and organic forms [28]. Here the phylogenetic diversity in the rhizosphere community is relatively moderate, but the metabolic diversity is high, while the endosphere community displays moderate phylogenetic diversity and very low microbial activity. The predominant bacterial family in the rhizosphere were the Xanthomonadaceae, which displayed a positive correlation with soil nitrate contents. On several occasions, this bacterial group was observed to increase in numbers when organic and mineral nitrogen fertilization was applied [56,57]. Together with the Microbacteriaceae and some other low abundance families, they constituted the bulk of the C. quitensis endosphere. The mentioned low activity of these bacteria might be due to host-defense mechanisms, enhanced by the heightened nitrate levels, as they were proved to promote defense signal molecules production like spermine and spermidine [58].
Based on the results of the PCA clustering, the microbial communities of D. antarctica displayed a high degree of similarity between samples within a particular rhizocompartment as compared to the vastly divergent C. quitensis communities. However, the microbiome of D. antarctica was also prone to restructuring enforced by the prevailing abiotic conditions. D. antarctica in the rhizosphere and the endosphere harbored in the majority of cases a phylogenetically low diversity community but highly active in terms of variety of catabolized carbon compounds. An exception was the salt-stressed site 1 community, where no specific bacterial group was enriched in the rhizosphere. However, in the endosphere of this site, Bacteroidetes families were strongly featured: Flavobacteriaceae and Sphingobacteriaceae. The latter was a consistent inhabitant of the D. antarctica endosphere, and their mean relative abundance was significantly higher compared to C. quitensis endosphere.
Members of this family were noted to proliferate in saltstressed plants' rhizosphere and root tissues [59,60] and were proven to confer tolerance to osmotically challenging conditions [61]. Those families were also observed in the invasive grass Poa annua L. communities presumably aiding its establishment in Antarctica, especially the Flavobacteriaceae [62]. An interesting case is the relative abundance of the family Rhizobiaceae, known to hold key species of plant beneficial rhizobacteria [63]. The occurrence of Rhizobiaceae and Sphingobacteriaceae was positively correlated with phenylethylamine catabolism intensity. For the Rhizobiaceae, this connection was previously described by [64] and was thought to indicate the formation of nitrogen-fixing bacteroids within the plant host cells. Despite some sitespecific anomalies, the root-associated communities of D. antarctica displayed far greater stability across the sampling locations than C. quitensis communities. This could indicate that at least part of the D. antarctica root-associated microbial community is transmitted vertically, either by seeds or vegetatively due to the scattering of turf pieces. Monocotyledons of the Poaceae family were proven to be superior in carrying a beneficial bacteria load with their seeds as compared to other plant species [65]. This load of selected plant beneficial bacteria can be responsible for the ecological success of D. antarctica in the Antarctic region and the wider ecological niche than C. quitensis [66,67] but also for its relatively low genetic diversity [15]. The variability that is introduced by sexual reproduction might diminish the grass' compatibility with its associated microbial community which could have evolved since the Pliocene colonization event. Such compatibility loss was frequently observed in cultivars of genetically altered agricultural crops [68,69]. While the majority of the D. antarctica root microbiome might contain facultative endophytes, dispersing through the soil and colonizing C. quitensis as indicated by the common core microbiome consisting of five bacterial families, this might not apply to the Rhizobiaceae. They seem to be exclusive D. antarctica obligatory root endophytes, as they were not observed in considerable abundance in the rhizospheric soils nor the endospheres of here examined C. quitensis specimens nor the invasive in Antarctica grass P. annua (yet still present in European P. annua samples) [62].
In conclusion, the Antarctic-native flowering plants display different strategies in assembling their root-associated microbiomes. C. quitensis seems to adjust its resident microbial community to the prevailing conditions even making use of microbivory, presumably due to the lack of associated efficient nitrogen fixers. D. antarctica on the other hand is inclined to rely on a fixed subset of bacteria that are presumably vertically passed to the daughter plant. This grass species holds to some obligatory nitrogen-fixing endophytes as well as other taxa that do not colonize C. quitensis roots, yet a shared core microbiome is likely to exist. Consequently, the "enigma" behind the presence of only two flowering plants in Antarctica might be strongly connected to their unique relationships with rhizospheric and root-dwelling bacteria.