Soil collection and characteristic analyses
Three types of soil were collected from different parts of China in August, 2016. Black soil, paddy soil and red soil samples were collected from Qiqihaer (123°62'E, 47°64'N) in Heilongjiang Province of Northeast China, Taizhou (120°20′ E, 32°44′ N) in Jiangsu Province of East China and Yingtan (116°94'E, 28°21'N) in Jiangxi Province of South China, respectively. The black, paddy and red soils are classified as Calcaric Chernozems, Orthic Acrisol and Ferralic Cambisol respectively, according to FAO/Unesco System of Soil Classification. Samples were obtained from the upper 20 cm of soil perennially covered by weeds. All samples were sieved with a 2 mm sieve removing visible plant tissues and stones, after which they were temporarily preserved in a portable storage box and transported to the lab immediately. Subsamples for measuring physiochemical properties were air-dried. Subsamples for pot experiment were stored at room temperature (25 oC) until use.
The following physio-chemical properties were measured to assess soil characteristics: (i) the soil pH was determined using a PHS − 3C mv/pH detector (Shanghai, China) at a soil-to-water ratio of 1:5; (ii) the available K (AK) in the soil was extracted with ammonium acetate and determined using flame photometry; (iii) the available P (AP) in the soil was extracted with sodium bicarbonate and then determined using the molybdenum blue method; (iv) the total N (TN) was determined via Kjeldahl digestion; (v) the total P (TP) and total K (TK) were extracted with HF–HClO4 and determined via molybdenum-blue colorimetry and flame photometry, respectively; (vi) the soil organic matter (OM) was determined using the potassium dichromate volumetric method; (vii) the cation exchange capacity (CEC) was measured using the ammonium acetate exchange method; (viii) the specific conductance (SC) was determined using 5:1 (liquid/soil) leach liquor and conductometer for calculation.
Experimental Setup
The maize used in this study is hybridized from male parent plant H9822 and female parent plant H9801, which is widely cultivated in coastal areas of China. Firstly, the seeds were soaked in sterilized water (25 oC) for 6 hours and then in 70% ethanol for 1 min. After that, the seeds were washed by sterile distilled water for three times. Then, the seeds were soaked in 2% Sodium hypochlorite solution for 10 min and washed by sterile distilled water for another three times. After surface sterilization, all seeds were spread on a sterilized plate for germinating in an artificial climate chamber under the condition of 16 h light at 22 oC and 8 h dark at 18 oC for 4 days.
The 4 days old seedlings were transplanted from plate to sterile (autoclaved) cylindrical pots (12 cm in diameter and 15 cm in height) with 1,000 g one type of fresh soil inside, with 2 seedlings per pot. Some pots were designated ‘bulk soil’ without plant. Including bulk soil controls, all pots were spatially randomized and placed in growth chambers providing 16 h light at 22 oC and 8 h dark at 18 oC and were watered during cultivation with sterile distilled water as an accessible imitation for rain water but without other chemical additives.
For time-series sampling, we established 20 replicated pots and 4 pots were randomly chosen for sampling at each sampling time (Fig. 1a and Additional file: Table S2). The rhizosphere and root samples were collected at successive intervals (0, 6, 9, 12, 16, and 20 days) with day 0 as the equivalent to soil and seed.
For transplanting experiment, after 8 days cultivation, a proportion of pots with the same soil were randomly chosen for sampling and the rest were used for transplanting (Fig. 1d and Additional file: Table S3). For transplantation, loose soil was manually removed from the roots without any damage on roots, and the roots were gently flushed with running water to wash soil away as possible. Then, roots were washed in sterile Phosphate Buffer (PBS-S, per litre: 6.33 g of NaH2PO4·H2O, 16.5 g of Na2HPO4·7H2O, 200 µL Silwet L-77) on shaking tables (80 Hz) for 30 min, followed by sterile distilled water washing for 3 times. After that, the plants were sonicated for 3 min with 30 s pulses at 60 Hz and 30 s breaks (power 220 V, ultrasonic cleaner KH5200DB, Kunshan ultrasonic instrument Co., Ltd., Kunshan, China) in order to wipe out tiny soil aggregates and root surface microorganisms [4, 8]. The washed plants were then transplanted to distinct soils and cultivated for another 8 days (Fig. 1d). Sampling was performed after cultivation (Additional file: Table S3).
Sampling
Bulk soil samples were collected from 1 cm below the surface in the control pots. Soil samples were placed into 2 mL clean and sterile centrifuge tubes. The samples were flash-frozen using liquid nitrogen and stored in -80 oC until DNA extraction.
For rhizosphere sampling, the aboveground plant organs were aseptically removed, loose soil (> 1 mm, not the rhizosphere soil) was manually removed with sterile rubber gloves (sprinkled with 70% EtOH) leaving approximately 1 mm soil on roots. Roots were placed in a clean and sterile 50 mL tube containing 30 mL PBS-S solution and vortex at maximum speed for 15 s (Vortex-Gene 2, Scientific Industries, USA). Next, a new clean and sterile 50 mL tube with 100 µm nylon mesh cell strainer was used to filter the plant organism and large sediment from rhizosphere soil [4], and repeat this step with 10 mL PBS-S solution to obtain a more complete rhizosphere sample. The turbid filtrate was then centrifuged for 15 min at 3,500 g to form a pellet with nice sediment and microbes. We discarded the supernatant and added 1 mL PBS-S solution to resuspend by votex. Then the suspension was transported to a sterile 2 mL microfuge tubes and centrifuged for 5 min at 10,000 g to form firmly pellets. A flash-frozen was conduct using liquid nitrogen after discarding the supernatant and stored at -80 oC until DNA extraction.
For endosphere sampling, the roots after cell strainer filtering were transferred to a new sterile 50 mL tube with 30 mL PBS-S solution and sonicated at 60 Hz for 3 min (3 cycle: 30 s sonication and 30 s break) to remove the surface microbes. After sonication, the roots were transferred to a 2 mL centrifuge tube with sterile tweezers and flash-frozen using liquid nitrogen. The samples were stored in -80 oC until DNA extraction.
Dna Extraction And 16s Rrna Gene Amplicon Sequencing
The liquid nitrogen frozen root samples were preprocessed by bead beating using a plant grinder (DHS TL2020, 0401261, DHS Technology Co., Ltd., Beijing, China) for 5 min at 1,800 rpm (5 cycle: 30 s vibrate and 30 s break). Then, total DNA of all bulk soil, rhizosphere soil and root samples were extracted from 0.25 g of sample using the QIAGEN DNeasy PowerSoil Kit (Ref: 12888-100, Germany). To minimize DNA extraction bias, three successive DNA extractions of each sample were pooled before performing polymerase chain reaction (PCR). A NanoDrop ND-2000 spectrophotometer (NanoDrop, ND2000, Thermo Scientific, 111 Wilmington, DE, USA) was used to assess DNA quality according to the 260/280-nm and 260/230-nm absorbance ratios.
Amplification of the V4 hypervariable region of the bacterial 16S rRNA gene was performed to assess the bacterial community using the primers 515F: 5’-GTGYCAGCMGCCGCGGTAA-3’ and 806R: 5’-GGACTACNVGGGTWTCTAA-3’. PCR amplifications were combined in equimolar ratios and sequenced on an Illumina MiSeq instrument (300-bp pairedend reads). The sequencing data were processed using the UPARSE pipeline (http://drive5.com/usearch/manual/uparse_pipeline.html) [26]. Raw sequences were first trimmed at length of 220 bp using command “fastx_truncate” to discard shorter sequences, the paired-end sequences were assembled using the command “fastq_mergepairs”. Then, high-quality sequences were reaped by “fastq_filter” command and dereplicated by “fastx_uniques” command. Singleton and chimeric sequences were removed after dereplication. The remaining sequences were clustered into operational taxonomic units (OTUs) at 97% similarity and taxonomic assignment was performed using the Greengenes 16S rRNA database (released on 2013/5). For each treatment, we made zero if an OTU has less than 3/4 (time-series samples) or 4/6 (transplanting samples) detected values among duplicates. At last, a rarefied OTU table at 5000 reads per sample was created using the USEARCH command “otutab_norm”.
Statistical analysis
Relative abundance of one phylogenetic group was defined as the number of sequences affiliated with that group divided by the total number of sequences per sample. The Shannon diversity index (α-diversity) calculation, Bray-Curtis dissimilarity-based PCA analysis and permutational multivariate analysis of variance (PERMANOVA) were performed based on the rarefied OTU table using the vegan R package (v.2.5-2) (https://cran.r-project.org/package=vegan). Tukey’s HSD test was used to calculate the significance between two samples. All statistical analyses were performed using R software (v.3.5.1).
The phylogenetic tree of the top 100 OTUs in relative abundance was constructed by FastTree (v.2.1.3) [27] and visualized by iTol (https://itol.embl.de/). The ternary plots were conducted using edgeR R package (v.3.22.3) (http://www.bioconductor.org/packages/release/bioc/html/edgeR.html) and visualized by ggplot2 R package (v.2.2.1) (https://cran.r-project.org/package=ggplot2). The Venn diagrams were calculated and visualized using the online tool Venny (https://bioinfogp.cnb.csic.es/tools/venny/). Sankey plots reveal the OTU flows as well as the composition among different samples [28] were constructed using the custom scripts based on D3.js (v.5.14.2) (d3js.org).
Network analysis was used to represent the co-occurrence pattern in a complex community. We conduct co-occurrence network analysis in different rhizosphere samples based on rarefied OTU table. A valid co-occurrence was considered as a statistically significant correlation between OTUs if the spearman’s correlation coefficient r > 0.75 and P-value < 0.01. The P-values were adjusted by multiple testing correction using the Benjamini-Hochberg’s FDR (false discovery rate) method [29] to reduce the chance of obtaining false-positive results.
The network analyses were performed using the psych R package (v.1.8.12) (https://cran.r-project.org/package=psych). Co-occurrence networks were visualized using Gephi software (v.0.9.2) [30]. We constructed bipartite network to feature the OTU sharing among samples [31] through the “Edge-weighted spring-embedded algorithm” method after trimming the OTU table at a 0.05% relative abundance threshold. The bipartite network was visualized using Cytoscape (v.3.6.1) [32].
The PICRUSt software [16] was applied to predict KEGG Ortholog (KO) functional profiles [33] of microbial communities using 16S rRNA gene sequences (Additional file: Figure S1). The overall functional characteristics (“Total Function”) displayed in ternary plot and PCA plots were analyzed using the KOs count table supplied by PICRUSt software. For segmented function analysis, firstly, the importance of OTU in each network was defined using the degree of each OTU in co-occurrence network. Secondly, the OTUs were segmented into a series of functional OTU clusters from the most important OTU to the least important OTU. Thirdly, “Segmented Predicted Function” of every OTU cluster was predicted using PICRUSt software, respectively. Meanwhile, a “Segmented Theoretical Function” of each OTU cluster was calculated using relative abundance (“OTU RA”, the relative abundance of each segmented OTU cluster) adjusted overall community function (Eq. 1).
Importantly, we summed the “Segmented Predicted Function” of all OTU clusters and found the results are exactly the same as the “Total Function”. To define the enriched or depleted functions of each OTU cluster, we calculated the “Segmented enriched or depleted function” using the “Segmented Predicted Function” and “Segmented Theoretical Function” (Eq. 2). The enriched or depleted functions were visualized in MATLAB (v.7.14.0.739).