Target metabarcoding on bacterial 16S rRNA gene, fungal ITS region and Glomeromycete 28S rRNA gene

The bacterial 16S rRNA gene was amplified using the primer pair 341 F/785R, while the fungal ITS1 region was amplified using the ITS1F/ITS2 primer pair. PCR reactions were monitored in a final volume of 25 µl composed of 5 µl of 5X GoTaq® colorless reaction buffer (Promega, France), 0.5 µl of each primer (10 µM), 0.5 µl of dNTPs (10 mM), 0.125 µl of GoTaq® G2 DNA Polymerase (Promega, France), 1 µl (at 1 ng.µl^− 1) or 2.5 µl (at 5 ng.µl^− 1) of DNA extracted from roots or soil respectively, and nuclease-free water (q.s.p 25 µl). The fungal 28S rRNA gene was amplified using a nested PCR approach to target the Glomeromycota division. A first PCR was performed using the primer pair LR1/NDL22 specific to the eukaryote 28S rRNA gene. The obtained product was diluted to 1/100th and 5 µl were used as template for the second PCR using the primer pair FLR3/FLR4 specific to Glomeromycota. All analyses included PCR negative and positive controls. The sequences of the Illumina adapters and the primers, as well as the cycling conditions are listed in Additional file 2: Table S3. Further steps were carried out at the Plateforme Génome Transcriptome de Bordeaux (Cestas, France). The PCR products were purified with the platform-specific SPRI magnetic beads (1X ratio) and quantified using Quant-iT™ dsDNA assay kit (ThermoFisher, France). MID and Illumina sequencing adapters were added. Libraries were pooled in equimolar amounts using a Hamilton Microlab STAR robot and sequenced on an Illumina MiSeq platform using the MiSeq Reagent Kit v2 (2 × 250 bp). Obtained sequences were demultiplexed with index search at the PGTB facility. The quality of the obtained sequences were first checked with FastQC v.0.11.8 [33]. Sequences were quality filtered, trimmed, denoised, and clustered into Operational Taxonomy Units (OTUs) using FROGS pipeline from Galaxy instance [34, 35]. This involved assembling raw forward and reverse reads for each sample into paired-ended reads with a minimum overlapping of 50 nucleotides and 0.1 mismatch using the VSEARCH tool [36]. Primers were removed using Cutadapt [37], chimeras were detected and removed with UCHIME [38], and clustering was performed using SWARM [39] in the FROGS pipeline. The minimum sequence abundance proportion was set at 5e^− 5 to keep OTUs with a minimum prevalence fixed at 4. Taxonomic assignments of 16S rRNA, ITS, and 28S rRNA-based OTUs were performed using Silva138.1 [40], Unite8.2 [41], and MaarjAM (28S) [42], respectively. For the 16S rRNA gene of bacteria, the sequences “multi-affiliated” at the phylum level were removed, as well as those affiliated to grapevine mitochondrial and chloroplastic16S rRNA sequences. As the proportions of host 16S rRNA sequences were important in the root endosphere, the samples from root and rhizosphere were divided before rarefaction at the number of sequences in the sample containing the fewest ones. Only the samples containing more than 4,000 or 250 sequences were kept for the rhizosphere and the root endosphere, respectively. For the fungal ITS region, the sequences “multi-affiliated” or “unidentified” at the phylum level were removed and only the samples containing more than 4,000 sequences were kept. For the 28S rRNA gene of AMF, the sequences affiliated to “New_clade” at the class level were removed and only the samples containing more than 5,000 sequences were kept.