A highly conserved core bacterial microbiota with nitrogen-fixation capacity inhabits the xylem sap in maize plants

Site description

The six long-term fertilisation field experiments were located across a latitudinal gradient in China from north to south, spanning three climate zones from the middle temperate zone to the subtropical zone. These sites were chosen to represent the three main agricultural production areas in China. The soils at the sites were black soils (Udolls or Typic Hapludoll according to USDA Soil Taxonomy, BSA) in Hailun (Heilongjiang Province) and Changchun (Jilin Province); fluvo-aquic soils (Aquic Inceptisol according to USDA Soil Taxonomy, FSA) in Yucheng (Shandong Province) and Yuanyang (Henan Province); and red soils (Ultisols according to USDA Soil Taxonomy, RSA) in Jinxian (Jiangxi Province) and Qiyang (Hunan Province). The two most distant sites (from Hailun to Qiyang) were more than 2,500 km apart, and the two closest test sites (from Yucheng to Yuanyang) were at least 300 km apart. Three treatments, i.e., no fertiliser (Control), chemical fertiliser N, P, and K (NPK), and organic manure plus chemical fertiliser (NPKM), have been applied in triplicate plots in each field for 29 years or more. Climate data corresponding to the sampling site coordinates were obtained from the China Meteorological Data Network (http://data.cma.cn/). Further details of experimental sites are provided in Supplementary Data 1.

Sample collection

Sampling was performed during the silking-maturity period of maize in 2019 and 2020, with the exact date of sampling depending on the developmental stage of plants at each location (Supplementary Data 2). In the FSA and RSA soils, three individual maize plants were selected from each subplot (a total of 27 maize plants per site). From each maize plant, we collected the following compartments in the field: mixed leaves, xylem sap, stem, roots, bulk soil. The mixed leaves sample consisted of the 2nd, 4th, and 6th leaves, which were removed from the plant stem using ethanol-sterilised scissors. To collect xylem sap, a proxy for xylem, we cut off the stem mid-way between the 2nd and 3rd node from the base of the plant, and sterilised absorbent cotton in sterilised bags was placed on the cut end of the shoot (see Supplementary Movie 1 for details of this procedure). Meanwhile, we inserted a steel stick (sterilised, 2 cm diameter, 30 cm length) into the soil at each subplot to simulate the collection of xylem sap and check for contaminants during the field operations (Supplementary Fig. 11). The stem sample consisted of the upper region between the 2^nd and 3^rd nodes, collected into sterilised bags. To collect root samples, we shook whole roots vigorously to remove all loose soil. The roots and root-adhered soil particles were collected for further separation of the roots and rhizosphere soil in the laboratory. The bulk soil sample was collected from between the rows of maize plants. At the sites with BSA soils, we only collected one individual maize plant from each subplot (a total of nine maize plants per site), because these two long-term experiments have strict requirements for sampling to avoid large-scale damage to the entire test field. Thus, for each plant, the 1^st and 2^nd, the 3^rd and 4^th, and the 5^th and 6^th leaves were collected as three replicates. Similarly, fine roots (< 2 mm) and thick roots (> 2 mm) were collected separately. Except for this difference, the other operations were the same as those at the other four experimental sites. All samples were placed on ice for transport and further processing within 48 h. The soil parameters of pH, total C (TC) and N (TN), ammonium (NH₄⁺) and nitrate (NO₃⁻), and soil available P (AP) and K (AK) are listed in Supplementary Table 1.

Sample processing

To recover the xylem sap absorbed in the cotton, each cotton ball was placed into a 50-mL sterile centrifuge tube with a filter and centrifuged at 6000 × g for 5 min. The collected sap was divided into two parts; one part was used for bacterial isolation, and the other part (stored at −80 °C) was used for culture-independent bacterial 16 S rRNA gene profiling.

Processing of root-associated samples

We used a modified protocol¹⁵ to separate the microbiome living on the plant surface (epiphytes) from the microbiome living within the plant (endophytes). Briefly, 5 g root tissue was weighed into a 100 mL conical flask containing 80 mL sterile PBS and 5 μL Tween 80. The mixture was vortexed, and the liquid was collected as the rhizosphere (root epiphyte) sample. To extract rhizosphere DNA, the sample was centrifuged at 10,000 × g for 5 min, and then 500 mg of the resulting tight pellet containing fine sediment and microorganisms was placed in a Lysing Matrix E tube (supplied in the FastDNA™ Spin Kit for Soil). To obtain endophytes from the root samples, the roots were washed with fresh PBS until the buffer was clear after vortexing. The roots were then sonicated using an ultrasonic cell disruptor (Scientz JY 88- IIN, Ningbo Scientz Biotechnology Co., Ltd., Zhejiang, China) at a low frequency for 10 min (30-s bursts followed by 30-s rests).

Processing of leaf-associated samples

The method for washing leaf samples was similar to that used to wash the roots, except for an additional step before collecting the phyllosphere. Each leaf sample (5 g) was added to a conical flask containing sterile PBS, which was subjected to two 5-min treatments in an ultrasound bath (25 °C, 40 KH_Z), with vortexing for 30 s between the two ultrasonication treatments. This procedure released most of the phyllosphere microbes from the leaves. Then, the filtrate was collected and the leaves were washed again using the above procedure. Phyllosphere samples were collected by centrifugation (at 10,000 × g for 20 min) of the accumulated filtrate and were resuspended in 1 mL sodium phosphate buffer (FastDNA™ Spin Kit for Soil) before being transferred to Lysing Matrix E tubes (FastDNA™ Spin Kit for Soil). Finally, the leaf and stem samples were washed and sonicated in the same way as the roots. Sonicated root, leaf, and stem samples were snap-frozen in liquid N₂ and stored at −80 °C until analysis.

DNA extraction, PCR amplification and sequencing

Total DNAs were extracted from the aforementioned samples with a FastDNA™ Spin Kit for Soil (MP Biomedicals, Solon, OH, USA) following the manufacturer’s instructions. The DNA concentration and purity were measured using a NanoDrop2000 spectrophotometer (NanoDrop2000, Thermo Fisher Scientific, Waltham, MA, USA). The DNAs extracted from soils (bulk and rhizosphere soil) were diluted 10-fold. For 16 S rRNA gene libraries, the V5–V7 region was amplified using the primers 799 F and 1193 R (Supplementary Table 5). Each DNA template was amplified in triplicate (together with a water control) in a 25-μL reaction volume. The PCR conditions were as follows: 12.5 µL 2× EasyTaq PCR SuperMix (TransGen Biotech, Beijing, China), 1.25 µL forward primers (10 µM), 1.25 µL barcoded reverse primers (10 µM), 1.25 µL template DNA, and 8.75 µL ddH₂O. The PCR amplification program was as follows: 94 °C for 3 min; 28 cycles of 94 °C for 30 s, 55 °C for 30 s, 72 °C for 90 s; and 72 °C for 90 s. The products were stored at 4 °C until use. After mixing the triplicate PCR products of each sample, the bacterial 16 S rRNA gene amplicons were extracted from a 1% agarose gel using a Gel Extraction Kit (Omega Bio-tek Inc., Norcross, GA, USA). The DNAs were measured using a Quant-iT™ PicoGreen™ dsDNA Assay kit (Thermo Fisher Scientific) and pooled in equimolar concentrations. Sequencing libraries were generated using an Illumina TruSeq DNA PCR-Free Library Preparation Kit (Illumina, San Diego, CA, USA) and were sequenced on the NovaSeq-PE 250 platform (Illumina).

16 S rRNA gene amplicon sequence processing

Paired-end reads were checked by FastQC v.0.10.1³⁷ and merged using the USEARCH 11.0.667³⁸ fastq_mergepairs script. Reads were assigned and demultiplexed to each sample according to the unique barcodes by QIIME 1.9.1³⁹. After removing barcodes and primers, low-quality reads were filtered and non-redundant reads were identified using VSEARCH 2.12.0⁴⁰. Unique reads with ≥ 97% similarity were assigned to the same OTU. Representative sequences were selected using UPARSE⁴¹ and the classify.seqs command in mothur⁴² was used to taxonomically classify each OTU with reference to the SILVA 138 database⁴³. The OTUs classified as host plastids, cyanobacteria, and others not present in our samples were removed from the dataset.

Statistical analysis

Alpha and beta-diversity analyses were conducted with R script as described in previous studies⁴⁴ and on the QIIME2³⁹ platform. PerMANOVA analyses were performed using the ‘Adonis’ function implemented in the vegan package⁴⁵ of R. We used the microbial source-tracking method FEAST⁴⁶ to determine the potential origin of the microbiota inhabiting the various compartments of maize plants. Bacterial OTUs were assigned into multiple functional groups using FAPROTAX v.1.2.1⁴⁷. All analyses were conducted in the R Environment⁴⁸ except for beta-diversity analyses. All plots were generated with ggplot2⁴⁹ and GraphPad Prism 8.0.0 (GraphPad Software, San Diego, CA, USA, www.graphpad.com).

Distance-decay relationship analyses

We conducted distance-decay relationship analyses to assess the relationship between the similarity of communities in individual plant compartments and spatial distance, edaphic distance, and climatic distance. We used the Geosphere package to calculate the geographic distances in km from the latitude and longitude coordinates, and calculated edaphic and climatic distances separately as the Euclidian distance. We used the ‘vegdist’ function in the vegan package to calculate the Bray–Curtis similarity of microbial communities. Here, the variation in the slope of distance decay reflects the degree to which the similarity of microbial communities in plant compartments varies with environmental distances. The relationships between the Bray–Curtis similarity of each compartment and specific soil or climatic factors were determined by calculating Pearson’s correlation (r) values. The significance of r values was assessed with the Mantel test implemented in the vegan package.

Differential abundance testing

A negative binomial generalised linear model was implemented with the edgeR package⁵⁰ to detect differences in OTU abundance among samples. We compared individual plant compartments (RS, RE, VE, SE, LE, and P) against bulk soil (BS) and conducted pairwise comparisons among fertilisation treatments. For each comparison, after constructing the DGEList object and filtering out low counts, the calcNormFactors function was used to obtain normalisation factors and the estimateDisp function was used to estimate tagwise, common, and trended dispersions. We then used the glmFit function to test the differential OTU abundance. The corresponding P values were corrected for multiple tests using FDR with α = 0.05.

Core taxa selection

We used the UpSetR package⁵¹ to visualise the OTUs that overlapped among all plant compartments and soils. The overlapping OTUs were defined as those detected in at least one sample from each compartment. We further identified the union of overlapped OTUs and enriched OTUs in the xylem using the EVenn online tool⁵². Importantly, abundance–occupancy analyses were conducted to identify the core OTUs across environmental gradients. We calculated occupancy with the most conservative approach, which restricted the core to only those OTUs that were detected in all xylem sap samples (i.e., occupancy=1).

Triple-qPCR to verify FAPROTAX results

To detect contamination with plant organelles, sample DNAs were amplified using the universal primer pair 799 F/1193R⁵³, which amplifies both bacterial 16 S and mitochondrial 18 S rRNA but not chloroplast sequences; the mitochondrial-specific primer pair mito1345F/mito4130R^54,55, which only amplifies mitochondrial 18 S rRNA, and the nifH gene primers PolF/PolR⁵⁶ (Supplementary Table 5). To reflect the potential N-fixation of bacterial communities in each plant compartment, the following ratio was calculated:

All qPCR assays were run on a QuantStudio 6 Flex using SYBR Green Pro Taq HS Premix (Accurate Biotechnology, Changsha, China) in a 20 µL volume containing 200 nM of each primer and approximately 50 ng DNA per reaction. All three primer sets were amplified in a three-step qPCR⁵⁷ run at 95 °C for 15 s, 55 °C for 30 s and 72 °C for 40 s for 40 cycles followed by a melting curve analysis. The amplification efficiency varied from 83 to 117%.

Isolation of bacteria from xylem sap

To isolate strains, four gradient dilutions (10⁻³, 10⁻⁴, 10⁻⁵, and 10⁻⁶) of xylem sap were incubated on TSB, R2A, and Ashby’s Nitrogen-Free Agar media for 5–7 days at 30 °C (Supplementary Data 9). After incubation, colonies were selected based on their character and colony morphology and were purified by triple serial colony isolation. The isolates were subjected to Sanger sequencing and identified on the basis of PCR analyses with 27 F and 1492 R primers, and alignment against reference. 16S rRNA gene sequences using the BLAST algorithm. Isolates belonging to the core taxa were identified by comparing the 16 S rRNA V5–V7 regions against the highly abundant OTUs (> 0.01%) using UCLUST with 98.65% similarity;⁵⁸ this threshold has been reported to accurately distinguish two species. Cladograms were visualised by iTOL.v6.4⁵⁹. The isolated cultures were stored in 30% (v/v) glycerol.

Nitrogen-fixing capacity of bacterial isolates

We used three different methods to evaluate the N-fixing capacity of bacterial isolates. First, we observed growth on Ashby’s N-Free medium, and documented which strains grew well after streaking of diluted cultures. Then, these strains were analysed by PCR to detect nifH with the PolF/PolR primer set⁵⁶. The positive control was the N-fixing strain, Azotobacter chroococcum ACCC10006 (Agricultural Culture Collection of China). Strains that did not yield a PCR product with this primer set were analysed using other nitrogenase gene primers including nifH-F/nifH-R⁶⁰ primers and the nested PCR primers FGPH19/PolR (outer primers) and PolF/AQER (inner primers)⁵⁶ (Supplementary Table 5). We also conducted acetylene reduction assays (ARA)^61,62,63 to quantify the nitrogenase activity of putative N-fixing strains. Each tested strain was initially incubated overnight in TSB medium and then washed twice with sterile 0.9% NaCl solution. After centrifugation and re-suspension, the bacterial pellet was added to a 20 mL serum vial containing 5 mL Dobereiner’s N-free liquid medium (Supplementary Data 9), reaching a final OD₆₀₀ of ~0.1. The vials were first flushed with argon to evacuate air, and then 1% and 10% of the headspace was replaced with pure and fresh O₂ and C₂H₂, respectively. After incubation at 30 °C for 12 h, the gas phase was analysed with a gas chromatograph (Agilent Technologies 6890 N). Data are presented as mean values from five replicate cultures. To test the hypothesis that the core non-N-fixers might assist N-fixation by modifying the oxygen concentration, the nitrogenase activity was measured as described above except that the headspace atmosphere in the sealed vial was not adjusted to 1% O₂ by flushing with argon gas so that the initial oxygen concentration was that found in ambient air.

Draft whole-genome sequencing of cross-referenced core strains

Isolated genomic DNA was extracted with a TIANamp Bacteria DNA Kit (Tiangen Biotech, Beijing, China). The purified genomic DNA was used to construct a sequencing library, which was generated using the NEB Next® Ultra™ DNA Library Prep Kit for Illumina (NEB, Beverly, MA, USA) following the manufacturer’s recommendations. Pooled libraries were sequenced on the NovaSeq-PE 150 platform. After trimming low-quality reads by fastq⁶⁴, the clean reads were assembled into draft genomes (excluding contigs of < 300 bp) by SPAdes 3.13.1⁶⁵. Gene prediction and annotation were performed by NCBI PGAP⁶⁶, and the putative genes were further annotated by searching against the eggNOG database⁶⁷ by emaper. The functional mapping and analysis pipeline (FMAP 0.15)⁶⁸ was used to align the filtered reads using BLAST against a KEGG Filtered UniProt⁶⁹ reference cluster (e < 1e-5, identity > 70%) and to calculate the number of reads mapping to each KEGG Orthologous group (KO). Other data manipulation was performed using perl scripts developed in-house.

Potted plant experiment

Two potted plant experiments were performed to (1) reproduce the endophytic behaviour of isolates from xylem sap; and (2) verify their N-fixation potential in maize. We constructed SynComs consisting of two diazotrophs (K. variicola MNAZ1050 and Citrobacter sp. MNAZ1397) and two non-N-fixers (Acinetobacter sp. ACZLY512 and R. epipactidis YCCK550) based on a N-fixing capacity test. Each individual strain was cultured overnight in TSB medium at 30 °C and 180 rpm, then cells were collected by centrifugation and the pellet was suspended in sterile 0.9% NaCl solution. Four bacterial suspensions were mixed in equal amounts to a final OD₆₀₀ of ~0.2.

The potted plants were grown in plastic pots filled with loose a soilless mixture consisting of perlite and vermiculite (sterilised by autoclaving). The nutrients needed for plant growth were added as base fertilisers (Supplementary Data 10). The seeds of maize “Zhengdan 958” were surface-disinfected for 15 min with sodium hypochlorite (approximately 2% active chlorine, with 200 μL Tween 80) and washed for 5 min, five times, with sterile water. The final rinse water (100 μL) was spread on TSA medium to check for other attached bacteria. Seeds were allowed to germinate, and then germinated seeds with similar primary root lengths were selected for inoculation with SynComs. Maize plants were grown in a greenhouse under a 16-h light/8-h photoperiod at 30 °C/25 °C (day/night).

Colonisation of maize tissues by GFP-tagged SynComs

To verify the endophytic behaviour, each of the four members of SynComs was tagged with green fluorescent protein (GFP) (vector pCPP6529-GFPuv). One GFP-tagged and three other wild bacterial suspensions were mixed as described above, giving a total of four different GFP-tagged combinations. The control was SynComs with no GFP tags. For inoculation, maize seedlings with the endosperm removed were soaked in four GFP-tagged combined solutions for 30 min. The same bacterial suspension was applied to the potted plants at days 10 and 30 after transplanting. After 63 days, stem samples from between the 2^nd and 3^rd nodes were surface-sterilised with 70% ethanol, collected, and then sectioned using a Leica VT 1000 S vibratome (Leica, Nussloch, Germany). Thin sections (60 μm) and xylem sap were observed under a confocal laser scanning microscope (CLSM, Zeiss LSM 880 confocal microscope, Jena, Germany).

¹⁵N isotope dilution method

To verify the N-fixation potential of endophytes in maize, the N fertiliser was replaced with ¹⁵N-labelled (NH₄)₂SO₄ (30 % ¹⁵N atom, Shanghai Research Institute of Chemical Industry, China). Maize seedlings with endosperm removed were soaked in a bacterial suspension of SynComs (Treatment group) or autoclaved SynComs (Control group) for 30 min. The same SynComs, active or autoclaved, were re-applied to the potted plants at 10 and 30 days after transplanting, as described above. The roots, stems and leaves were harvested separately from each treatment (four replicates) on day 65. The roots were washed with deionised water to remove adhering isotope residues. The N content and ¹⁵N enrichment of plant tissue were determined using Elementar vario PYRO cube elemental analyser (Vario PYRO Cube, Elementar, Hanau, Germany) and Isoprime 100 isotope mass spectrometer (Isoprime, Cheadle, United Kingdom). The plants inoculated with autoclaved SynComs were used as the reference to calculate BNF with the following equations³⁴:

where %Ndfa is the percentage of N derived from air, %¹⁵Na.e. (%¹⁵N atom excess) is the enrichment in plants inoculated with SynComs (I) and autoclaved SynComs (UI), N₂-fixed is N derived from air, and N yield._I is the total N content of the whole inoculated plant.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Source: Ecology - nature.com