Choosing source of microorganisms and processing technology for next generation beet bioinoculant

The increase of human population and associated increasing demand for agricultural products lead to soil over-exploitation. Biofertilizers based on lyophilized plant material containing living plant growth-promoting microorganisms (PGPM) could be an alternative to conventional fertilizers that fits into sustainable agricultural technologies ideas. We aimed to: (1) assess the diversity of endophytic bacteria in sugar and sea beet roots and (2) determine the influence of osmoprotectants (trehalose and ectoine) addition during lyophilization on bacterial density, viability and salt tolerance. Microbiome diversity was assessed based on 16S rRNA amplicons sequencing, bacterial density and salt tolerance was evaluated in cultures, while bacterial viability was calculated by using fluorescence microscopy and flow cytometry. Here we show that plant genotype shapes its endophytic microbiome diversity and determines rhizosphere soil properties. Sea beet endophytic microbiome, consisting of genera characteristic for extreme environments, is more diverse and salt resistant than its crop relative. Supplementing osmoprotectants during root tissue lyophilization exerts a positive effect on bacterial community salt stress tolerance, viability and density. Trehalose improves the above-mentioned parameters more effectively than ectoine, moreover its use is economically advantageous, thus it may be used to formulate improved biofertilizers.


Results
Rhizosphere as a source of microbes in endosphere. Rhizosphere soil physicochemical parameters are different for sugar and sea beet. Majority of tested parameters was higher in sugar beet soil, but only in cases of CaCO 3 and Na + the difference was statistically significant. On the other hand, OC, P, Ca 2+ , Mg 2+ and N t were higher in sea beet soil and for the latter the difference was significant (Table 1). To check if these changed parameters influenced bacterial communities in rhizosphere soils, we sequenced 16S rRNA amplicon libraries.
Bacterial diversity in sugar beet roots is lower than in its wild ancestor. Bacterial diversity, evenness and species richness were the highest in 16S rRNA libraries coming from rhizosphere soil, regardless of plant genotype. Lyo- Table 1. Physico-chemical rhizosphere soil parameters (mean and standard deviation) obtained after three months of cultivation of sugar-and sea beet. [↑] significantly higher level based on Newman-Keuls test of rhizosphere soil parameter observed between the plant species. www.nature.com/scientificreports/ philized sea beet roots harbored more diverse community than sugar beet (Fig. 1). The number of OTUs was ca. three times higher in the wild beet than in the crop (Fig. 1A,B), while the diversity was around 1.5 times higher (Fig. 1C), and evenness was ca. 1.3 times greater (Fig. 1D).
Both endophytic and rhizosphere soil bacterial community is dominated by Proteobacteria. There were no significant differences in taxonomic composition of rhizosphere soil bacterial communities of sugar-and sea beet at the level of phylum ( Fig. 2A). At the level of genus, three taxa were differentially represented, all of them belonging to Alphaproteobacteria: two Rhizobiales-belonging genera, Pedomicrobium and an unknown genus of JG34.KF.361 family as well as Woodsholea (Caulobacteraceae) were more abundant in the crop (Fig. 2C). Differences in lyophilized roots communities were more pronounced, although still there were no taxa significantly differentially represented between osmolyte treatments. At the level of phyla Proteobacteria-derived reads were more abundant in libraries from sugar beet lyophilized roots, while Actinobacteria, Bacteroidetes, Acidobacteria, Verrucomicrobia and rare phyla were more abundant in its wild ancestor (Fig. 2B). Among genera significant differences were observed for Stenotrophomonas and Bacillus that were more abundant in the crop and for proteobacterial genera Novosphingobium, Devosia (Alphaproteobacteria), Hydrogenophaga, Polaromonas (Betaproteobacteria), Rhizobacter and Tahibacter (Gammaproteobacteria) as well as for rare and unclassified genera being more abundant in sea beet (Fig. 2D).  www.nature.com/scientificreports/ At the level of phyla , regardless of plant genotype, there were significant differences between soil and roots in all taxa but Firmicutes. Proteobacteria and Firmicutes were more abundant in roots than in soil, while abundance of the remaining phyla was lower in planta, and Gemmatimonadetes as well as Verrucomicrobia were absent from roots. At the level of genus, regardless of genotype, Pseudomonas and Rhizobium were significantly more abundant in roots than in soil, while Sphingomonas, Pedomicrobium, rare and unclassified bacteria were less frequent in roots than in rhizosphere. Novosphingobium, Pantoea, Hydrogenophaga, Polaromonas, Paenibacillus, Hyphomicrobium and Rhizobacter were significantly more abundant in wild beet roots than in soil, while Stenotrophomonas was the only genus that was more frequent in sugar beet roots than in soil. Variibacter, Chryseolinea, and Woodsholea were less abundant in wild beet roots than in soil, while Devosia and Hirschia were less frequent in sugar beet than in soil.
Effect of osmolytes on diversity, viability, and tolerance to salinity of bacterial communities in lyophilized beet roots. Bacterial cell density in lyophilized roots depends on host genotype but not on presented, and significant differences between genotypes are marked either with m and h letters (panels A and C, significant differences between rhizosphere and endosphere in sea (m) or sugar (h) beet) or with asterisks (panels B and D, significant differences between genotypes, no differences due to osmolytes were found). www.nature.com/scientificreports/ osmolyte. In total, 72 bacterial strains were isolated and identified, 35 coming from sugar beet and 37 from sea beet. Proteobacteria were the most frequent phylum in fresh roots of both sugar and sea beet, followed by Actinobacteria in the crop and Firmicutes in the wild plant. Pseudomonas and Sphingomonas were characteristic for fresh roots of sugar beet, while Bosea and Sphingopyxis were found exclusively in sea beet roots before lyophilization ( Table 2). Density of culturable root endophytic bacteria was higher in sugar beet lyophilizates than in sea beet (ANOVA, p < 0.05, Fig. 3), regardless of the osmolytes addition. We observed no influence of osmolytes on sea beet endophytes density, while trehalose increased slightly, but significantly (ANOVA, p < 0.05) the density in sugar beet samples (Fig. 3).
Sea beet endophytes are more salt tolerant than sugar beet ones. Increasing salinity negatively affected growth of culturable fraction of microbiome regardless of origin (sea-vs. sugar beet), however stronger effect was observed for sugar beet. In control treatment the growth was inhibited (final cell density below the critical level of 0.2 OD 600 ) at 200 mM and 300 mM NaCl concentration for sugar and sea beet, respectively. Addition of osmolytes enhanced the growth in general and increased the inhibitory concentration to 400 and 700 mM, respectively (Supplementary Table 1). Influence of both osmolytes was similar, with trehalose performing slightly better at high NaCl concentrations., The effect was greater for sea beet, than for sugar beet (Fig. 4).
Bacterial viable cell density in lyophilized roots is associated with plant genotype and osmolyte. Cell viability in lyophilized beet roots was assessed by means of three, complementary methods: via plate counts, fluorescence Table 2. Identification of cultivable endophytic bacteria associated with roots of sugar-and sea beet before and after lyophilization without addition of any osmolyte (C) or supplemented either with ectoine (E) or trehalose (T). www.nature.com/scientificreports/ microscopy and flow cytometry (Fig. 5). Bacterial viability in sugar beet was consistently higher than in roots of its wild relative, regardless of osmolyte treatment, storage time and measurement methodology. Both trehalose and ectoine increased the viability compared to control, regardless of genotype, but the effect of the former was more pronounced (Fig. 5).

Discussion
Bacterial diversity in beet rhizosphere. Differences in rhizosphere soil physicochemical properties observed in our study, may be due to greater nutritional demands of the two beet genotypes (TN, Na) or varying exudates composition (OC), as it was found that rhizodeposition is the primary organic carbon source in the rhizosphere 30 . Alternatively, they might be caused by changes in microbial activity resulting from microbial metabolic activity or interaction between microorganisms 31,32 . Greater microbiome diversity in rhizosphere compared to endosphere was commonly observed, and resulted from natural plant selection mechanisms [33][34][35] . Accordingly, in our study, the higher bacterial diversity, evenness and species richness were noted in rhizosphere soil of both investigated genotypes, than in roots. At the same time, in spite of slightly different TN, OC and Na levels, microbiome composition and diversity were similar in rhizosphere soils of both studied plant genotypes. This observation could be explained by the use of the same starting substrate (garden soil) and short culture period (three months), not allowing the rhizosphere differences to fully manifest. Culture-independent analysis revealed that dominating bacterial phyla were the same as those observed in rhizosphere of many plant species e.g. barley, alfalfa or wheat [36][37][38] . Only a few differences between the genotypes were noted at the genus level, mainly concerning Alphaproteobacteria. Pedomicrobium as well as JG34.KF.361_ge, more frequent in sugar beet, represent Rhizobiales, an order known for organisms that establish beneficial interactions with plants and comprises numerous bacteria with nitrogen-fixing capability 39 . The observed lower TN level in the sugar beet rhizosphere may indicate higher demand for nitrogen. Tsurumaru and colleagues 40 indicated that Mesorhizobium and Bradyrhizobium, also belonging to Rhizobiales, play an important ecological role in the taproot of sugar beet. Moreover, it was showed that higher levels of nitrogen (N) and potassium (K) significantly affect the growth parameters of sugar beet. Both elements were generally recognized as crucial for obtaining higher yields of this crop, favorably affecting organic metabolites biosynthesis and improving nutritional status 41 . Bacterial diversity in beet roots. The higher diversity both in rhizo-and endosphere of the wild plant compared to its crop counterpart was observed 42,43 . It was hypothesized that beneficial endophytes associated with wild plants were absent or fewer in domesticated crops 43 . Sugar beet as a cultivated plant grows under more controlled conditions regulated by farmers, while sea beet grows mainly in highly saline and nutrients poor coastal soil 28 . Growth under adverse environmental conditions requires support of microorganisms with a wide range of beneficial metabolic properties tailored for specific plant needs 23 . The loss of high tolerance to salt stress during the process of sea beet domestication was demonstrated 29 and might be associated with the loss of   www.nature.com/scientificreports/ microbes that increased tolerance of this plant to salinity. Concordantly, despite the lack of differences in rhizosphere soil microbial composition, lower diversity of endophytes in sugar beet compared to its wild ancestor was noted in our study. This difference might be explained by varying root system architecture, with fibrous root system of sea beet providing more opportunities for bacteria to enter the endosphere 33 , which affects stochastic community assembly. On the other hand microbe selection can be driven by the genetic makeup of two studied subspecies. We observed that sea beet caused decrease in the soil Na level, suggesting accumulation of Na ions in wild plant tissues. Accordingly, there was an increase in community salinity resistance in this plant, which pointed at higher level of halotolerant and halophytic microorganisms. In general, endophytic microbiome diversity and composition is related to soil properties as well as plant ecology and physiology 44 . Members of only three phyla (Proteobacteria, Actinobacteria and Firmicutes) were cultured in our experiment, this may be related to their high ability to grow on commercially available media 5,6,44,45 . It was emphasized that Proteobacteria distinctly predominate among culturable plant endophytes, then the presence of Firmicutes and Actinobacteria is common, and Bacteroidetes occur slightly less frequently 44 .
16S rRNA gene libraries generated in our study were dominated by the four phyla (Proteobacteria, Actinobacteria, Firmicutes, Bacteroidetes) commonly found in endosphere of glycophytes including maize (Zea mays L. 46 ), Dactylis glomerata L., Festuca rubra L. and Lolium perenne L. 47 as well as in halophytes such as Salicornia europaea 23 or para grass (Urochloa mutica 48 ). Sea beet was characterized by significantly higher frequency of Actinobacteria, Bacteroidetes, Acidobacteria, Verrucomicrobia and rare phyla compared to sugar beet, where Proteobacteria were observed more often. Zachow et al. observed greater frequency of Actinobacteria, Bacteroidetes and Verrucomicrobia in rhizosphere of wild beet cultivated in coastal soil than in sugar beet rhizosphere 42 . This fact, together with our results, may point at these bacterial taxa being preferred by sea beet regardless of soil.
Our 16S rRNA gene sequencing results also revealed significantly higher abundance of certain genera in sea beet endosphere, including: Novosphingobium, Devosia (Alphaproteobacteria), Hydrogenophaga, Polaromonas (Betaproteobacteria), Rhizobacter and Tahibacter (Gammaproteobacteria) as well as certain rare and unclassified bacteria. This set of microorganisms comprises extremophiles, e.g. Polaromonas 49 or Hydrogenophaga 50 and organisms modulating plant stress response, such as Novosphingobium 25 . In our study, only Stenotrophomonas and Bacillus genera were more frequent in roots of sugar beet than of sea beet. Stenotrophomonas and Pseudomonas sp. were identified in rhizospheric soil of sugar and sea beet, while the former together with Staphylococcus sp. were mainly observed in crop rhizosphere. Sea beet microbiome was found to be more diverse than that of sugar beet, which is explained by greater number of rare taxa. It was found that sugar beet rhizosphere was more frequently colonized by strains with antagonistic activity against plant pathogens and/or stress protection activity, while abiotic stress-releasing ones were more often found in sea beet's rhizosphere 42 . These facts together with our results suggest that pre-adaptation to stress observed in sea beet transcriptome 51 may also take place at the level of microbiome serving as a helper.

Osmoprotectants enhance bacterial viability and diversity in lyophilized beet roots.
Significantly higher cell density of culturable bacteria observed in sugar beet lyophilized roots can be attributed to high content of sucrose. This sugar acts as a natural osmoprotectant, allowing better viability of microorganisms during lyophilization 52 . Another explanation of obtained results can be associated with higher ability of sugar beet endophytes to grow on solid medium.
Sea beet endophytic microbiome was found to be more resistant to salinity. Microorganisms present in a more saline sea beet tissue most likely developed mechanisms of adaptation to high salt level, which provided them ability to grow in higher NaCl concentrations compared to the sugar beet microbiome. This fact may be related to higher sodium accumulation in this plant tissues 51 , which caused decrease of soil sodium concentration observed in our study.
Salinity-induced changes in community structure and adverse effects on microbial density, activity, biomass were reported by many scientists 53,54 . The decrease in number of culturable microorganisms related to increasing NaCl concentration was noted even in the case of endophytes associated with halophytes (Aster tripolium, Salicornia europaea) 5,6,55 . Obtained results were in line with the above trend, but apart from negative effect of salinity on sugar and sea beet bacterial density, a beneficial impact of trehalose and ectoine on salt stress mitigation was demonstrated. Although ectoine is a major osmolyte in aerobic chemoheterotrophic bacteria and is considered as a marker for halophytic bacteria 15 , a slightly better effect of trehalose, was confirmed by the results of microscopic analyzes, flow cytometry and culture tests. Protective effect of trehalose is explained by "water replacement hypothesis" that states that the compound lowers the phase transition temperature of membrane phospholipids, by replacement of water molecules occurring around the lipid head groups 56 , thus protecting membrane structure 57 . This suggests that the use of trehalose is a better and more economic solution providing high viability of bacterial cells after lyophilization. In the case of sugar beet the above mentioned positive sucrose impact was enhanced by trehalose addition. Similar effect was observed for rhizobial strains, where trehalose worked better than sucrose/peptone mixture 58 . In general, 16S rRNA gene sequencing results considering diversity of endophytes associated with sea and sugar beet root did not show any effect of applied osmoprotectants neither on alpha nor beta diversity of bacteria. This observation can be explained by the presence of 'relic DNA' , i.e. DNA coming from non-viable cells 59 in lyophilized samples.
Bacillus sp. was the only species identified among the strains representing the Firmicutes phylum isolated from the lyophilized osmolytes-treated roots of both investigated genotypes. In the control variant the presence of Psychrobacillus sp. and Paenibacillus sp. inside sea and sugar beet root was additionally found, respectively. The viability of the above-mentioned bacteria after lyophilization was probably associated with their commonly known ability to form endospores and higher tolerance to environmental changes [60][61][62]  www.nature.com/scientificreports/ to be sensitive to lyophilization, while Proteobacteria remarkably well tolerated it, and additional osmolytes promoted the incidence of culturable bacteria belonging to the latter phylum.

Conclusions
Our research revealed that plant genotype played a pivotal role in the shaping of its endophytic microbiome diversity and physicochemical rhizosphere soil properties, affecting soil sodium content, but not soil bacterial community structure. Bacterial diversity was lower in sugar beet roots than in its wild ancestor tissues. At the same time sea beet endophytic microbiome was more salt resistant and consisted of genera characteristic for extreme environments. Supplementing osmoprotectants during root tissue lyophilization had a positive effect on bacterial salt stress tolerance, viability and density. Trehalose proved to improve these parameters more effectively than ectoine, moreover its use was economically advantageous. Plant and soil samples preparation. Plants were carefully uprooted, and 10 g of soil adhering to roots (rhizospheric soil) was collected, frozen at − 80 °C and lyophilized before DNA isolation for metagenomic analysis. Roots were washed with tap water to remove soil and were separated from shoots and leaves. Then, they were surface sterilized with 70% ethanol and 15% hydrogen peroxide mixture (1:1 v:v) for 5 min and subsequently rinsed six times with 0.9% NaCl. Efficiency of the sterilization process was evaluated by plating the last rinse on Luria-Bertani (Difco LB Agar, Miller) and potato dextrose extract (Lab A Neogen Company) media. Only properly sterilized plant material was used for subsequent analyzes. Approximately 100 g of fresh root material was homogenized in 100 ml of 0.9% NaCl by using surface sterilized (rinsed with 70% ethanol and UV-irradiated) blender. Homogenates were used to evaluate bacterial density and to prepare lyophilizates.

Roots lyophilization.
Homogenized sugar and sea beet roots were used to prepare three variants of lyophilizates including (1) no osmolytes addition (control-C) (2) trehalose (T) and (3) ectoine (E) supplemented. Three biological replicates were prepared for each tested plant species (9 samples per plant species, in total 18 samples were used for downstream analyzes). Either 1 ml of 0.9% NaCl (control) or 1.0 mg of trehalose (Tre) or 1.0 mg of ectoine (Ect) were mixed with 50 g of homogenized roots. The mixtures were lyophilized in Telstar LyuQues (DanLab) until completely dry (approximately 24 h).
Estimation of bacterial density. Serial dilutions were prepared directly from the homogenized fresh roots and lyophilizates re-suspended in 0.9% NaCl (1:9 m:v). The dilutions (10 −3 to 10 −8 ) were plated in triplicates on LB plates supplemented with nystatin (Sigma, 100 µg/ml) to prevent fungal growth, and the plates were incubated for 5 days at 26 °C. Colony counts (expressed as CFU per 1 g of fresh or dry weight for homogenates and lyophilizates, respectively) were based on plates with 30-300 colonies. At least six bacterial isolates were purified per experimental variant.
Bacterial viability assessment: fluorescence microscopy and flow cytometry. Ten-miligram samples of ground lyophilized roots were mixed with 10 ml of PBS (pH = 7.4) and incubated for 2 days at 26 °C with mixing. The mixtures were filtered through a 40 µm cell strainer (Biologix) and 2 ml were centrifuged for 3 min at 1000 × g at RT to pellet the residual plant debris. Cells in the supernatant were stained with Cell Viability kit (BectonDickinson) as per the manufacturer's protocol, than bacterial viability was analyzed using fluorescence microscopy (after 6 and 12 months of storage) and flow cytometer (after 12 months storage). Preparations were photographed in red and green channel under 40 × magnification upon fluorescence excitation with 433 nm light on Axiostar plus fluorescence microscope (Zeiss) equipped with Delta Optical camera. Percentage of live cells was based on counts from at least 30 view fields per sample. Flow cytometric analysis was performed on samples stained as described above with FACS Aria III (BectonDickinson) using 488 nm laser for excitation. Fluorescence was collected at 530 ± 30 nm (for thiazole orange-TO) and 616 ± 26 nm (for propidium iodide-PI) bands and seventy-micrometer nozzle was used. Parameters were optimized basing on pure environmental strains and their mixtures analyses and autoclaved lyophilizate samples served as negative controls. Isolates identification by 16S rRNA gene sequencing. Genomic DNA was isolated from purified strains using GeneMatrix Bacterial and Yeast Genomic DNA Purification Kit (EurX) according to the manufacturer's protocol with modified homogenization step (FastPrep-24 bead-beater, one cycle of 20 s at 4.0 m/s). The DNA was analyzed spectrophotometrically (NanoDrop 2000). 16S rRNA gene fragment was amplified using 27F and 1492R primers 63 , following the procedure described in Szymańska et al. 6 . The products were purified with GeneMatrix PCR/DNA Clean-Up DNA Purification Kit (EurX) according to the manufacturer's protocol.
Sanger sequencing was performed with BrightDye Cycle Sequencing kit (Nimagen), using 40 ng of template DNA, 1.5 pmol of primer and 1 µl of kit and 1.5 µl of BD buffer in 10 µl volume. The reactions were EtOH/NaAc precipitated and read out at IBB PAS, Warsaw, Poland.
16S rRNA gene fragment library construction and sequencing. Metagenomic DNA was isolated and V3-V4 16S rRNA gene fragment libraries for Illumina sequencing were prepared as described earlier 64 . They were sequenced on Illumina MiSeq using 600 cycles v.3 kit at CMIT NCU. www.nature.com/scientificreports/ Statistical analysis and bioinformatics. Bioinformatics analyses of Illumina reads was performed as described earlier 64 . Briefly, the reads were denoised, merged and chimeras were removed with dada2 65 , then amplicon variant sequences were exported together with abundance information and processed in Mothur v.1.39 66 : aligned against SILVA v.132 database, screened for those covering the 6428-22400 positions of the alignment, filtered to remove gap-only and terminal gap-containing positions, pre-clustered to remove residual noise and clustered into 0.03 dissimilarity OTUs. Representative OTU sequences were classified using naïve Bayesian classifier 67 and SILVA database 68 . Sanger reads were manually inspected in Chromas to remove obvious errors, the corrected sequences were merged with CAP3 69 , and classified using naïve Bayesian classifier with SILVA v.132 reference files. Significance of differences between means was assessed with ANOVA test with Tukey's post-hoc analysis implemented in Statistica 10.0 (StatSoft). Normality of data was tested with Shapiro-Wilk's test and homogeneity of variance was assessed with Levene's test. When the assumptions were violated, non-parametric Kruskal-Wallis test with Dunn's test as a post-hoc analysis was used. Significance level of 0.05 was assumed.

Data availability
Sequences generated during this study were deposited in the SRA repository and are accessible via BioProject no. PRJNA606174. www.nature.com/scientificreports/