Identification of the rhizospheric microbe and metabolites that led by the continuous cropping of ramie (Boehmeria nivea L. Gaud)

Continuous cropping lowers the production and quality of ramie (Boehmeria nivea L. Gaud). This study aimed to reveal the metagenomic and metabolomic changes between the healthy- and obstacle-plant after a long period of continuous cropping. After 10 years of continuous cropping, ramie planted in some portions of the land exhibited weak growth and low yield (Obstacle-group), whereas, ramie planted in the other portion of the land grew healthy (Health-group). We collected rhizosphere soil and root samples from which measurements of soil chemical and plant physiochemical properties were taken. All samples were subjected to non-targeted gas chromatograph-mass spectrometer (GS/MS) metabolome analysis. Further, metagenomics was performed to analyze the functional genes in rhizospheric soil organisms. Based on the findings, ramie in Obstacle-group were characterized by shorter plant height, smaller stem diameter, and lower fiber production than that in Health-group. Besides, the Obstacle-group showed a lower relative abundance of Rhizobiaceae, Lysobacter antibioticus, and Bradyrhizobium japonicum, but a higher relative abundance of Azospirillum lipoferum and A. brasilense compared to the Health-group. Metabolomic analysis results implicated cysteinylglycine (Cys-Gly), uracil, malonate, and glycerol as the key differential metabolites between the Health- and Obstacle-group. Notably, this work revealed that bacteria such as Rhizobia potentially synthesize IAA and are likely to reduce the biotic stress of ramie. L. antibioticus also exerts a positive effect on plants in the fight against biotic stress and is mediated by metabolites including orthophosphate, uracil, and Cys-Gly, which may serve as markers for disease risk. These bacterial effects can play a key role in plant resistance to biotic stress via metabolic and methionine metabolism pathways.

bacteria, nitrifying bacteria, and aerobic nitrogen-fixing bacteria. Changes in the diversity of soil bacteria are correlated with soil physicochemical properties, for example, pH value, moisture, salinity, porosity, available nitrogen, and organic matter [13][14][15] . Accordingly, continuous cropping practices lower crop yields and quality by affecting the biosynthesis and metabolism of active plant components 16 .
Ramie (Boehmeria nivea L. Gaud), also known as "China grass, " is a perennial, diploid (2n = 28) and herbaceous plant of the Urticaceae family. Ramie is widely cultivated owing to its bast fiber, which offers excellent properties such as excellent thermal conductivity and high tensile strength 17 . Ramie fiber is extracted from stem bast and poses numerous excellent characteristics, for example, long strands, smooth texture, and high tensile strength 18 . Previous studies reported that continuous cropping of ramie reduces plant height, stem diameter, and fiber yield 5,12,19 . In our previous studies, we showed that the stem growth of ramie is impeded by continuous cropping 12 . Moreover, we found that the abundance of Firmicutes was correlated positively with the plant height, stem diameter, and ramie fiber yield. However, as the duration of continuous planting increases, fiber production decreases 12 . The above studies revealed the obstacles in the continuous cropping of ramie. However, none of the reports indicated whether soil microbial communities and metabolites were associated with continuous cropping obstacles.
In production practice, we found an interesting phenomenon: After 10 years of continuous cropping (single variety), ramie planted in some portions of the land were characterized by weak growth and low yield (Obstaclegroup), on the other hand, ramie planted in the other portions exhibited healthy growth (Health-group). Thus, we hypothesized that this difference may be attributed to the effects of soil microorganisms and could be revealed by soil metagenomics and metabolome analysis. This present study attempted to provide a better understanding of the relationship between soil bacteria and metabolites and continuous cropping obstacles and to elucidate on the growth of ramie under continuous cropping practice.

Results
Continuous cropping of ramie. After 10 years of continuous cropping (single variety), ramie planted in some portions of the grew healthy (defined as Health-group) whereas, ramie planted in the other portions grew weak in and with low yields (defined as Obstacle-group). Soil samples in Obstacle-group had lower contents of available P and K, and higher content of total K than Health-group (p < 0.05, Table 1). The values of plant height, stem diameter, bark thickness, weight, and fresh and dry fiber weight in the Obstacle-group were lower compared to those in Health-group (p < 0.0001, Fig. 1). These findings confirmed that growing the same ramie variety in different continuous cropping lands greatly varies.
Illumina gene sequencing. Illumina sequencing generated a total of 326.58 M clean reads from 6 rhizosphere soil samples with an average Q30 value of 94.59% and GC content of 64.37%. A total of 10,450,4244 scaffolds and 80,931 open reading frames (ORFs, longer than 200 bp) were retrieved from the data (Table S1). Additionally, 435,887 non-redundant gene catalogs or clusters were identified, including 49,444 annotated genes Table 1. Differences in soil chemical parameters between Health-group and Obstacle-group. Differences are analyzed using a t-test.

Parameters
Health-group Obstacle-group p www.nature.com/scientificreports/ to bacterial taxonomy. Expression analysis showed that 5517 clusters were differentially expressed between the two groups. Notably, the numbers of annotated genes in the two groups were different (Fig. 2). About 98.26% of the clean reads originated from bacteria, followed by 1.28% and 0.3% from archaea and virus, respectively.

GC-MS/MS metabolomics.
TICs (total ion currents) revealed 554 peaks detected from all samples (including quality control samples). Metabolite annotation results showed 190 identified metabolites. PCA (Fig. S2A) and OPLS-DA (Fig. S2B) results showed that there were different metabolites in soil and root between the two groups. Moreover, we identified 20 and 31 differential metabolites in the soil and root samples between the two groups, respectively (Table S3), including 3 overlapping metabolites (threitol, glycerol, and 2-Deoxyerythritol).
Identification of the differential metabolites in soil samples. From the soil samples, there were higher levels of threitol, 2-Deoxyerythritol, uracil, malonic acid, and Cys-Gly (Cysteinylglycine and lower glycerol level in Obstacle-group than in Health-group (Table S3, Fig. 5A). KEGG (Kyoto Encyclopedia of Genes and   www.nature.com/scientificreports/ Genomes) pathway enrichment analysis showed that these 20 different metabolites in soil samples were enriched in 17 pathways. Notably, Cys-Gly, orthophosphate, uracil, malonate, and glycerol had a higher frequency in these pathways. Moreover, Uracil was involved in multiple pathways, such as metabolic pathway, pantothenate and CoA biosynthesis, pyrimidine metabolism, among others, whereas malonate was associated with pyrimidine metabolism and beta-Alanine metabolism.
Identification of the differential metabolites in root samples. From the root samples, the level of glycerol and threitol was higher in Obstacle-group compared to Health-group (Fig. 5B). On the contrary, the level of 2-Deoxyerythritol, maleic acid, and raffinose in the Obstacle-group was higher than in Health-group (p < 0.05). The KEGG enrichment analysis showed that 90 pathways were enriched, including Carbon metabolism (e.g. fumarate, 3-Hydroxypropanoate, citrate, d-Gluconic acid, d-Glucono-1,5-lactone, and pyruvate), Metabolic pathways (e.g. fumarate, 3-Hydroxypropanoate, citrate, glycerol, myo-Inositol, alpha,alpha-Trehalose, d-Gluconic acid, cysteamine, pyruvate, and Phytosphingosine), Biosynthesis of phenylpropanoids, and Biosynthesis of plant secondary metabolites (e.g. fumarate, citrate, gallate, l-Phenylalanine and pyruvate; Table S4). Of all the enriched differential metabolites in the pathways, orthophosphate, uracil and Cys-Gly were found to exhibit a high frequency.

Integrated analysis of the metagenome and metabolomics.
To identify the correlation between the differential microbial genes and metabolites in the two groups, we identified the same pathways associated with the differential genes and metabolites in soil and root samples. Of note, the oxidative phosphorylation (ko00190), pyrimidine metabolism (ko00240), beta-alanine metabolism (ko00410), glutathione metabolism (ko00480), metabolic pathways (ko01100), ABC transporters (ko02010), and two-component system (ko02020) were found to be correlated with the content of metabolites (e.g., orthophosphate, uracil, malonate and Cys-Gly) through the genes expressed in 5 Archaea, one virus and 206 bacteria (Table S3). For instance, Archaea Nitrososphaera viennensis, Candidatus Nitrososphaera gargensis and Candidatus Nitrososphaera evergladensis regulated two-component system (ko02020) and metabolic pathways (ko01100) by changing the metabolomic profiles of orthophosphate, uracil, and Cys-Gly. Genes expressed in Archaea, for example, cmpC glutamate dehydrogenase (GDH2) and glrR are listed in Table S3. Of which, WspR, expressed in Azoarcus sp. PA01 and Planococcus sp. PAMC_21323, regulated orthophosphate metabolism and was involved in the two-component system pathway; trxB was expressed by Nitrospira sp. SCGC_AG-212-E16 and Bradyrhizobium elkanii, which regulated pyrimidine metabolism by regulating the metabolism of uracil and malonate. The f accD (Nitrospira moscoviensis) and accC (Janthinobacterium lividum) genes could regulate orthophosphate, uracil, and Cys-Gly and were enriched in metabolic pathway (ko01100) (Table S3). Besides, we identified several genes involved in methionine metabolism, which regulates orthophosphate, uracil and Cys-Gly metabolism, they include gshB (encodes glutathione synthase, GSH) in Azospirillum lipoferum; pepA (encodes leucyl aminopeptidase, LAP) in Elusimicrobium minutum; phoB1 (encodes alkaline phosphatase synthesis response regulator PhoP) in A. brasilense; and mmuM (encodes homocysteine S-methyltransferase, HMT) in Desulfitobacterium metallireducens (Fig. 6A). In addition, an interaction network was constructed to reveal the relationship between metabolites and candidate genes (Fig. 6B). Results showed that Threitol, Cys-Gly, oxamide and d-(glycerol 1-phosphate) were highly associated with the genes. These findings uncovered the crucial roles of soil bacterial diversity in the rhizospheric and root metabolites, in particular, Cys-Gly, which participated in multiple pathways and showed rich interactions with other compositions. www.nature.com/scientificreports/

Discussion
Continuous cropping obstacles are primarily related to disorders or deterioration of rhizosphere microorganisms 20,21 . Our present study confirmed that continuous cropping of ramie is correlated with the changing rhizospheric soil microbes and soil metabolomics as well as the root metabolism. The present study identified that the abundance of some bacteria, including Rhizobia, L. antibioticus, and Bradyrhizobium japonicum, were lower in Obstacle-group compared to Health-group, a finding that concurs with previous reports [22][23][24][25] . A higher abundance of Rhizobia can increase soil biomass, as was confirmed in previous experiments [26][27][28][29] . Among many soil bacteria, Rhizobia can produce higher levels of indolacetic acid (IAA) 30 , which is one of the most physiologically active auxins. IAA is a common product of l-tryptophan metabolism produced by several microorganisms including Plant Growth Promoting Rhizobacteria (PGPR) 31 . The endogenous or exogenous IAA can stimulate cell elongation by modifying certain conditions, for example, increasing in osmotic contents of the cell, increasing the permeability of water into cells, reducing wall pressure, increasing cell wall synthesis and inducing specific RNA and protein synthesis 32 . In addition to Rhizobia, we identified two bacteria (including A. lipoferum and A. brasilense) that were highly abundant in the Obstacle-group compared to Health-group. Notably, A. lipoferum and A. brasilense are growth-promoting bacteria and are capable of impoving plant resistance to abiotic stresses partially via enhancing plant abscisic acid level 33,34 . Both A. lipoferum and A. brasilense can also produce IAA 35 .  37 . Moreover, the overexpression of GSH in plants has been shown to contribute to improved plant stress tolerance and productivity 38 . It is worth noting that A. lipoferum expresses the GSH encoding gene gshB, while D. metallireducens expresses a gene mmuM encoding HMT involving in plant tolerance and defense response 39,40 . Although no l-amino acids were detected among the differential metabolites in soil and ramie root, our study uncovered that these bacterial genes regulated the metabolism of orthophosphate, uracil, and Cys-Gly, thus participates in metabolic pathway and methionine metabolism. Therefore, we speculated that the increased abundance of A. lipoferum and A. brasilense in the soil may be associated with the resistance to the continuous cropping induced abiotic stress in ramie root. In the present study, L. antibioticus was revealed to play important roles in various metabolic pathways via the regulation of metabolites such as orthophosphate, uracil, and Cys-Gly. Moreover, L. antibioticus could produce phenazine antibiotics 41 . In the recent past, Laborda et al. showed that L. antibioticus OH13 produced antifungal p-aminobenzoic acid. Another study reported that L. antibioticus exhibited biocontrol potential as it produced 4-hydroxyphenylacetic acid and several lytic enzymes against phytophthora blight 42 . In summary, these studies showed that L. antibioticus exerts a positive effect on plants in fighting biological stress. Low levels of L. antibioticus in the Obstacle-group may imply a decline in plant defense mechanism. Moreover, metabolites such as orthophosphate, uracil, and Cys-Gly may serve as mediators for L. antibioticus to play an active role. In addition, Bradyrhizobium japonicum, a microsymbiotic nitrogen-fixing bacteria, is known to enhance legume-root nodulation, which when introduced when planting, improve crop yields, especially legumes 43 . As far as we know, no report has explored whether Bradyrhizobium japonicum is related to the growth of ramie. However, based onour findings, we speculate that Bradyrhizobium japonicum may also play an active role in maintaining the health of ramie.
Our metagenome analysis showed a lower relative abundance of several soil probiotics in the Obstacle-group, which may have been induced by long-term continuous cropping. Additionally, the NMDS and PCoA analyses using different methods revealed the distinct separations among samples from different groups. This showed that differences exist in the microenvironment of rhizosphere soil between the Obstacle-group and Healthgroup. The PCoA based on the weighted UniFrac algorithm clearly demonstrated the variations among different groups. And in the dimension of PC1, the two groups can be distinguished to the greatest extent (79.15%). These differences in soil microorganisms might be the most important factor that ultimately led to a greater difference between the two groups. Han et al. showed that 5-year continuous cropping of cotton lowered not only the richness of nutrition bacteria type in soil but also the abundance of ammonifying bacteria, nitrobacteria, and aerobic nitrogen-fixing bacteria. Also, a different study reported that two plant growth-promoting bacteria Azospirillum brasilense Cd and B. pumilus ES4 could enhance the growth of microalga by inducing the production of total carbohydrates, chlorophyll a and total lipids in microalga 34 . Besides, numerous previous studies found that Bradyrhizobium and Rhizobium (Proteobacteria) are symbiotic bacterial partners that promote the formation of nitrogen-fixing nodules on legumes, thereby enhancing the plant growth, resistance, and tolerance to abiotic stresses and pathogens [44][45][46] . Similarly, our present study reported higher relative abundances of R. leguminosarum, A. chroococcum and R. mesoamericanum (Proteobacteria) in the Health-group than that in the Obstacle-group, which was inversely associated with the inhibition of ramie growth. The above-mentioned bacteria potentially play a key role in keeping plants healthy. Previous findings revealed the Lysobacter species as a potential source of novel antimicrobial plant metabolites 41,47,48 . Notably, L. antibioticus has been revealed to be capable of producing phenazine antibiotics 41 . For example, a recent study by Laborda et al. showed that L. antibioticus OH13 produced antifungal p-aminobenzoic acid. In our study, we suggested that the lower level of L. antibioticus in the Obstacle-group may signify a decline in plant defense.

Conclusions
Soil bacteria may primarily affect health and obstacles to the growth ramie after long-term continuous cropping. Bacteria such as Rhizobia can synthesize IAA and are likely to reduce the biotic stress of ramie. For instance, L. antibioticus exerts a positive effect on plants in fighting biotic stress. In this study, we showed that lower levels of Scientific Reports | (2020) 10:20408 | https://doi.org/10.1038/s41598-020-77475-3 www.nature.com/scientificreports/ L. antibioticus in the Obstacle-group may imply a decline in plant defense. Metabolites including orthophosphate, uracil, and Cys-Gly might serve as mediators for L. antibioticus to play an active role. Notably, subjecting ramie to biological stress may increase in abundance of A. lipoferum and A. brasilense. Consequently, the bacterial changes are associated with various metabolic pathways through the regulation of metabolites such as orthophosphate, uracil and Cys-Gly. Particularly, Cys-Gly participated in multiple pathways and showed rich interaction in the interaction analysis. A low abundance of beneficial bacteria in Obstacle-group might be associated with the resistance to the continuous cropping induced abiotic stress in ramie root. These bacterial changes may play a key role in plant resistance to biotic stress via the metabolic and methionine metabolism pathways. Thus, this present study provides novel insights into the association of soil microbe and metabolome. These findings might show some indications on the agricultural practice for continuous ramie cropping. year, 10 consecutive years). The test field was 15 acres (areas were close to each other with a similar initial soil structure). All the acres were farmed under the same cropping system and conditions, including annual planting time, harvest time, fertilization and irrigation measures, etc. Routine agronomic management and fertilization were performed annually during the ramie growing period. In the later period of the experiment, we found that the ramie planted in some portions of the land grew well (after 10 years of continuous cropping, some ramie were healthy, and were defined as Health-group), while the ramie planted the other portions had obvious growth obstacles due to continuous cropping (mainly characterized by short plants and low fiber production, and were defined as Obstacle-group). Rhizosphere soil samples were collected from the Health-group and Obstacle-group whereby a total of 30 soil samples were obtained in each group on June 23, 2017. The fields in the two groups were about 10 m apart, and each field was separated by a 1.5-m-wide ditch. Every 5 samples were pooled, and a total of 6 rhizosphere soil samples were prepared. The pooled soil samples in each group were divided into 2 portions, for metagenomics sequencing analysis (randomly selecting three samples, n = 3) and metabolome analysis, respectively. We also collected the ramie root samples. A total of 30 root samples were collected, every 5 samples were pooled, and a total of 6 ramie root samples were prepared. All the 6 root samples were used for metabolome analysis. All collected samples were stored at − 80 ℃ immediately after preparation.

Methods
Detection of soil chemical properties. The soil chemical properties, including the contents of available nitrogen (N), phosphorus (P), potassium (K), and total N, P, K were measured according to the methods reported previously 49 . In each group, 15 samples were randomly collected from 0 to 60 cm soil, mixed into three pools, air dried and ground. Soil total N, P and K was detected using the GB 7173-1987 (NY/T  Sequencing data preprocessing. The paired-end raw reads in the format of FASTQ were quality controlled using Trimmomatic by trimming and filtering adaptors and low quality reads. SOAP denovo (v4.5.4; https ://soap.genom ics.org.cn/) was used for the metagenome assembly 51 . ORF prediction of assembled scaffolds was performed using prodigal (v2.6.3) and was translated into amino acid sequences. All effective tags were clustered by CDHIT (v4.5.4; https ://www.bioin forma tics.org/cd-hit/) 52 . With the thresholds of 95% identity and 90% coverage, a non-redundant gene set was assembled. The relative abundance of each gene in the clean reads was analyzed using fast gapped-read alignment with Bowtie2 (v2.1.0) 53  www.nature.com/scientificreports/ using R (v3.2.0). The annotation of the gene products were performed in databases including NR, SWISSPROT, COG (Clusters of Orthologous Groups of proteins), Gene Ontology (GO) and KEGG with the threshold of e ≤ 1e−5. PCoA and NMDS analysis was performed based on the Euclidean distance algorithm to identify the beta diversity among samples and groups. The differentially expressed genes in metagenome between the two groups were identified using ANOVA in QIIME (V1.8, https ://qiime .org/scrip ts/alpha _raref actio n.html). The KEGG pathways associated with differentially expressed genes were identified in KEGG database (https ://www.genom e.jp/KEGG/pathw ay.html) with the threshold of p < 0.05.
Preparation of quality control (QC) samples. QC samples were prepared for the non-targeted metabolomics analysis by pooling equal aliquots of individual sample (n = 6) within each group. Three QC samples were prepared.
Non-targeted chromatographic parameters. The analytical instrument used in this experiment is a 7890b-5977a gas chromatograph-mass spectrometer (GS/MS) from Agilent Technologies Inc. (USA). The chromatographic conditions are: DB-5MS column (30 m × 0.25 mm × 0.25 μm; Agilent J&W Scientific, Folsom, CA, USA); high purity helium (99.99%) with a flow rate of 1.0 mL/min; injection temperature of 260 °C. The temperature of column was increased from 60 to 125 °C at a speed of 8 °C/min, then increased to 210 °C at a speed of 4 °C/min, to 270 °C at a speed of 5 °C/min, to 305 °C at 10 °C/min and kept for 3 min. MS conditions are: electron impact ion (EI) source at 230 °C; quadrupole temperature at 150 °C; 70 eV high-energy. EI-MS spectra were recorded at a scan range of m/z 50-500. System stability and accuracy was validated using QC samples with an interval of 5 samples.
Non-targeted metabolomics data analysis. MS raw data (total ion current, TIC) was converted into file format using ChemStation (version E.02.02.1431, Agilent Technologies Inc). ChromaTOF (version 4.34, LECO, St Joseph, MI) was used to analyze the data, and NIST and Fiehn database were used for the annotation of the metabolites. After alignment with Statistic Compare component, the 'raw data array' (.cvs) was obtained from raw data including peak names, retention time-m/z and peak intensities. All internal standards and pseudo positive peaks were removed. Data was transformed by log2 and then imported into SIMCA software package (14.0, Umetrics, Umeå, Sweden). Unweighted principle component analysis (PCA) and (orthogonal) partial least-squares-discriminant analysis (OPLS-DA, with sevenfold cross validation and response permutation testing, 200 times randomly permutated) were performed to visualize the metabolism difference between groups. Metabolites with variable important in projection (VIP) > 1 and p value < 0.05 by two-tailed Student's t-test were used for identification of differential metabolites. Metabolites between groups with |fold change (FC)|≥ 1 were considered as differential metabolites. The KEGG pathways associated with the differential metabolites were identified from KEGG database (https ://www.genom e.jp/KEGG/pathw ay.html) with the threshold of p < 0.05. Statistical analysis. GraphPad Prism 6 software was used for data statistical analysis. Data were expressed as the mean ± SD and differences between groups were analyzed by t-test. A value of p < 0.05 was considered statistically significant.