Rhizosphere sample collection from Italy, Ethiopia and Burundi locations
The rhizosphere samples were collected from sorghum grown in different geographical locations: Italy (coordinate: 45.981558 N, 13.198167 E), Ethiopia (coordinate: 9°03'49.8"N 40°52'29.2"E, 9°05'30.9"N 37°02'45.1"E, 8°40'13.3"N 39°29'14.8"E and 7°50'11.6"N 38°41'41.2"E) and Burundi (coordinate:4°00'03.3"S 30°04'26.2"E). In Italy 46 sorghum genotypes (Table S1) were selected as a subset of the broad genetic diversity sorghum association panel (SAP) United States [27, 34, 35]. Seeds were surface sterilized with 10% bleach for 10 min followed by five washes with sterilized water and then pre-germinated in the plant growth room to ensure uniformity in growth. The pre-germinated seeds were then placed in a homogenized soil mixture and grown in controlled conditions (25oC, 16 h photoperiod) for two weeks in a growth chamber. Watering was performed every 24h for the first three days and every three days starting from the fourth day onward. After 14 days, uniformly grown seedlings were transplanted to an experimental field in three replicates at ERSA (Agenzia Regionale per lo Sviluppo Rurale) center at Pozzuolo del Friuli, Italy. Three plants of each of the 46-sorghum genotype were transplanted in each field block, with a separation of 20 cm between each plant and 40 cm between each block. After nine weeks, root samples were collected from the 414 plants (three plants from each block of the same genotype). Additionally, nine soil samples (three from each block) were collected using a shovel, with each sample individually dipped 20 cm into a 50 ml Falcon tube containing removal buffer (6.75g KH2PO4, 8.75g of K2HPO4, and 1 ml of Triton X-100, in 1L). The rhizospheres were separated from the root as previously described [36]. The recovered rhizospheres of the three plants with the same genotype and collected from the same field block were pooled together as a technical replicate and stored at -80oC in 20% glycerol for further DNA isolation and bacterial isolation. The same procedure was followed for collecting rhizosphere samples from sorghum plants in Burundi and Ethiopia grown in agricultural fields as part of regular farming practices. A total of 16 sorghum rhizosphere DNA samples were collected from Burundi and 30 samples were collected from Ethiopia at the age of nine weeks post transplantation. In Burundi, rhizosphere samples were collected from four different sorghum varieties: IESH22005, IESH29068, ICSR93034, and Gambela. In Ethiopia, rhizosphere samples were collected from five different sorghum varieties: Gambella 1107, Chiro, Meko-1, Dinkinash and Merera. In summary, rhizosphere samples for DNA isolation were uniformly collected from all geographical locations, at the same plant age (pre-maturation stage), and following identical procedures.
DNA extraction, library preparation and sequencing
DNA was extracted from the sorghum rhizosphere compartment using DNeasy® PowerSoil® Kit from Qiagen (Cat. No. 12888-100). The DNA quality and quantity was measured by using the Nanodrop device (Thermo Scientific, Wilmington, DE, USA). The 16S rRNA gene amplicon libraries were prepared for each DNA sample following the manufacturer’s protocol (Illumina Inc., San Diego, CA, USA). Amplification of V3- V4 hypervariabe region of the 16S rRNA gene was performed by using long PCR primer [37] incorporating the Illumina adaptor sequences (Forward Primer 5’- TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGCCTACGGGNGGCWGCAG and Reverse Primer 5’- GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGGACTACHVGGGTAT CTAATCC). Following the first amplification, a cleaning step was performed using the AMPure XP bead clean-up (A63880l; Beckman Coulter Inc., Brea, CA, USA). Next steps for the library construction have been performed with a second PCR to attach dual index and Illumina sequencing adapters using the Nextera XT Index Kit; followed by a final AMPure XP bead clean-up. Amplicons size, integrity, and purity were checked using the Bioanalyzer equipment (Agilent Inc., Santa Clara, CA, USA) and the library concentration was measured by fluorimetric quantification using Qubit 2 (Invitrogen Inc., Carlsbad, CA, USA). Libraries were adjusted to a final concentration of 4nM in 5ul of 10mM Tris pH 8.5 and submitted for sequencing using 2x250 bp paired end reads on Illumina Miseq.
Amplicon bioinformatics data analysis
Comprehensive analysis of a large dataset was conducted using the fastq files generated from samples collected in Italy (141), Ethiopia (30), and Burundi (16), as mentioned above. Additionally, fastq files from 600 16S rRNA gene libraries were obtained from publicly available sources, specifically focusing on rhizosphere samples cultivated in the United States. These sequences were sourced from [27], and were downloaded from the NCBI BioProject ID PRJNA612320.
All the 787 fastq files were imported into qiime2 [38] and quality filtering, denoising and amplicon sequence variants detection have been done using DADA2 [39] and taxonomic assignment to each Amplicon Sequence Variants ASVs was carried out based on Silva database (release 138) [40]. Diversity parameters were calculated at a rarefaction depth of 1385 reads. ASVs that have been classified as a chloroplast and mitochondria were filtered out. The obtained dataset was imported into R using qiime2R package [41], and the subsequent analysis was done using phyloseq [42], vegan [43] and microbiome R package [44]. Alpha and Beta diversity were measured using Shannon and Unifrac unweighted distance respectively. The prevalence of each taxon was determined at ASVs and Genus level.
We considered as MPT (most prevalent taxa) or “core microbiome” the genera occurring (i) in more than 80% of the total samples with a relative abundance > 0.5% per sample and also (ii) present in > 70% of samples from each location, regardless to their relative abundance. [13, 45]. To infer co-occurrence using the fastspar [46], we exported the table at the genus level, filtered to retain the genera occurring in at least 5 samples, and computed the SparCC value [47] to estimate the co-occurrence of bacterial genera among the samples. Correlations with a p-value > 0.05 and those in the range − 0.3–0.3 were eliminated. A bottleneck analysis was performed to identify pivotal nodes within the microbial community structure. This analysis entails constructing a network in which nodes represent various taxa or microbial entities, and edges denote interactions or relationships between them, such as co-occurrence patterns [48]. Cytoscape was used to compute the statistics and depict the generated network [49]. According to [14, 50, 51], keystone bacterial taxa were considered the nodes presenting the highest betweenness centrality and node degree. The network analysis was performed either using the whole dataset or restricted to the Italian samples.
For monitoring the presence and co-existence of the 8 strains used in the designed consortia, gene community sequencing was employed both in the co-existence/compatibility experiment (in vitro) and in planta tests (plant growth room and open field). All fastq files were imported into qiime2 and processed following the same pipeline as describe above. For the samples derived from the compatibility experiment in plant growth room the diversity parameters were calculated at a rarefaction depth of 4100 reads, and the number of reads per sample ranged from 14387 to 72399 with an average of 44035.37 reads. For the samples derived from the sorghum rhizosphere persistence experiment in field, diversity parameters were calculated at a rarefaction depth of 900 reads, and the number of reads per sample ranged from 16 to 29136 with an average of 12649,08 reads.
For the evaluation of the consortia compatibility and ability of the 8 selected isolates to colonize the sorghum rhizosphere a customized dataset of 16S rRNA gene sequences from the 8 isolates used in this study was created and used for the taxonomic assignment. For clustering and comparing the ASVs based on sequence similarity, the cd-hit program was used, setting the parameter of sequence similarity equal to 98% for two sequences to be considered part of the same cluster. All the ASVs not matching with any of the 16S rRNA gene sequences of the customised dataset were removed from the analysis. The obtained dataset was imported in R using the package qiime2R [38] and the analysis and graphs were generated using phyloseq and vegan packages in R [42, 43].
Isolation of culturable collection of sorghum rhizosphere-associated bacteria
A large set (321 strains) of culturable bacteria associated with the sorghum root microbiome have been isolated using aliquots of the rhizospheric fractions from Italy samples which were stored at -80oC with 20% glycerol. Different dilutions of the samples were plated in various solid media; Tryptic Soy Agar TSA, (TSA-supplemented with NaCl, CuSo4 and and pH 4.5), NBRIP, KBA, YEM Congo red, Jonson medium (JM). All the media compositions used are presented in Table S2. The plates were incubated at 30oC for 2–5 days and pure single colonies showing distinct colony morphology were picked up independently and streaked on 1/5 TSA plates to ensure the purity of the colonies. The pure colony culture was then stored at -80oC in 1/5 Tryptic Soy Broth (TSB) and 20% glycerol. Colony PCR was performed after boiling (10’ at 98oC) a colony suspension in 50 ul of sterile H2O to amplify the complete 16S rRNA gene by using primer fD1Funi 16S (5’- AGAGTTTGATCCTGGCTCAG-3’) and rP2Runi 16S (5’-ACGGCTACCTTGTTAGGACTT-3’). PCR products were purified by using PCR Clean-Up purification kit (Euroclone S.p.A). The sequencing was performed with the primer 518F (5’- CCAGCAGCCGCGGTAATACG-3’) and 907R (5’- CCGTCAATTCMTTTRAGTTT-3’) and the stains were identified by BLAST analysis at NCBI (https://www.ncbi.nlm.nih.gov).
Co-growth tests of sorghum rhizospheric most prevalent and keystone strains and strategy for designing multi-strain consortia
From the bacterial strains identified as sorghum-MPT and keystone using the 16S rRNA gene profiling approach, we carefully chose those that could be cultured/grown under laboratory conditions and exhibited the highest 16S rRNA gene sequence homology (99%) with the ASVs identified as sorghum-MPT and keystone. A set of eight bacterial strains was carefully selected (Pantoea sp., Enterobacter sp., Pseudomonas sp., Bacillus sp., Rhizobium sp., Streptomyces sp., Paenibacillus sp. and Paraburkholderia sp.) and based on the genomic characterization and in vitro PGP features, three consortia were designed as follows: Consortium 1(Pantoea dispersa SRG11, Enterobacter asburiae SRG25 and Pseudomonas chlororaphis subsp. aurantiaca SRG 32), Consortium 2 (Priestia megaterium SRG70, Rhizobium wenxiniae SRG248 and Streptomyces sp. SRG181) and Consortium 3 (Paenibacillus illinoisensis SRG287and Paraburkholderia sp. SRG18).
In order to perform the first set of compatibility growth experiments, the eight bacterial sorghum MPT and keystone rhizospheric strains were independently grown in liquid media (TSB), washed twice and resuspended in phosphate-buffered-saline (PBS) solution. An equal amount of each strain (1.00E + 08 CFU/ml) was then pooled in a PBS solution in order to form the three different consortia. This mixed solution containing the strains in a comparable amount was used as inoculum for in vitro and in planta compatibility tests as described here below. In all these conditions, each bacterial-strain consortium was monitored and assayed using gene amplicon sequencing and analysis.
Co-growth on agar plate media
an aliquot of 100µl of each strain from the mixed inoculum in PBS, as described above, was spotted on TSA plates and was incubated at 30oC. Samples were collected after 96 hours, DNA was purified and used for 16S rRNA gene library preparation and sequencing. As a control, all eight strains were mixed in equal amount and DNA was purified immediately (0 hour).
In planta co-growth
sorghum seeds were surface sterilized, as described above and then submerged in each consortium-inoculum for 30 minutes. The seeds were then transferred to a 50ml Falcon tube containing semi-solid agar (0.4% agar) with ½ Hoagland solution. The plants were grown in a growth chamber under controlled conditions, including a photoperiod of 16 hours of light and 8 hours of darkness, as well as a humidity level of 80%. After nine days post-inoculation, the rhizosphere was collected and DNA was extracted for 16S amplicon microbiome determination.
In growth chamber and in field plant growth promotion experiments by applying multi-strain bacterial consortia
For the in planta experiments performed in the growth-chamber, surface-sterilized sorghum seeds were carefully submerged in each consortium-inoculum for 30 minutes. As control, sorghum seeds were submerged in 1xPBS without the bacterial suspensions. After inoculation, both the treated and no-treated seeds as control were transferred to the soil; seeds were inserted 1 cm deep and covered with soil. In brief, the soil was prepared using a mixture of garden soil, perlite/vermiculite, and sand in a specified ratio (1:1:5). After seven days post-germination and root development, a second furrow inoculation was performed using an aliquot of 500 µl of the three consortia and 1xPBS for control was applied near the roots. After the second inoculation, five plants were collected for rhizosphere DNA extractions and 16S rRNA gene amplicon library preparation and ten plants were collected for plant agronomic parameters detection. The experiments involved two time points, TP1: 25 days post-seed inoculation (includes 10 days after furrow inoculation), TP2: 60 days post-seed inoculation (includes 50 days after furrow inoculation).
For the in field experiments, the same seed-inoculum procedure was followed, as described above. The soil was prepared in three blocks, each receiving a different percentage of chemical fertilizer (Nitrogen, P2O5, K2O followed by N (Urea) at 4/5 leaf stage plant growth): Block 1: 100% chemical fertilizer, Block 2: 50% chemical fertilizer, Block 3: 0% chemical fertilizer (Figure S1). Both the treated and control seeds were transferred to the soil in each block. An aliquot of 2.5 ml of each consortium and 1xPBS for control was applied near the roots using furrow inoculation. Five plants were collected for rhizosphere at TP1: 40 days post-seed inoculation (includes 20 days after furrow inoculation), TP2: 80 days post-seed inoculation (includes 60 days after furrow inoculation) and ten mature sorghum plants were collected for plant agronomic parameters determination (shoot dry biomass, panicle dry weight, seeds dry weight and seeds count) analysis at days 115 post-inoculation (includes 95 days after furrow inoculation). At each time point the total DNA was extracted, 16S rRNA gene amplicon library prepared and microbiome monitored.
Whole Genome sequencing of the eight bacterial strains isolates forming the three multi-strain consortia
The eight bacterial strains used in the consortia studied here were whole genome sequenced. Each strain was cultured overnight in TS broth at 30°C. Following cultivation, the cultures were centrifuged at 6000 RPM for 5 minutes, the cells were collected and DNA was subsequently extracted using the Norgen Biotek Corp. Bacterial Genomic DNA Isolation Kit. The complete genomes were sequenced with the Illumina NovaSeq 6000 platform using 150bp paired-end reads and following the tagmentation Illumina Nextera XT protocol (Illumina Inc., San Diego, CA, USA). The assembly was performed with Unicycler v. 0.5.0 and the assembly statistics were recorded with QUAST v. 5.2.0 [52]. The assembled genomes were uploaded in the Integrated Microbial Genomes and Metagenomes (IMG/M) database and automatically annotated, using annotation pipeline IMG Annotation Pipeline v.4.16.6 [53]. Functional annotation and phylogenetic characterization were performed by DFAST [54] ran from https://dfast.ddbj.nig.ac.jp/.