Plant material
Medicago truncatula cv. Jemalong A17 and A17::YC3.6, T. turgidum cv. Kronos and T. aestivum cv. Cadenza were used as WT. M. truncatula TILLING alleles were identified from A17 mutant collections (https://www.jic.ac.uk/research-impact/technology-research-platforms/molecular-genetics/medicago-truncatula/)16. The population was screened by direct sequencing of target amplicons on a capillary ABI3730 sequencer (Applied Biosystems), as described16. The primers used to amplify the target are listed in Supplementary Table 1. T. aestivum cv. Cadenza and T. turgidum cv. Kronos TILLING alleles were identified from the sequenced collections28 via Ensembl Plants (www.plants.ensembl.org). The third generation of backcrossed T. aestivum cv. Cadenza mutant and T. turgidum cv. Kronos mutant was generated by manual pollination with their respective WTs. After the third backcross, BC3F1 plants were self-pollinated, and near-isogenic lines (NILs) homozygous for the GOF mutation and without the mutation (that is, WT) were selected at the BC3F2 stage. Populations were genotyped as described below in ‘Genotyping of segregating F2 population’. F2 homozygous and WT M. truncatula cngc15aP98S was backcrossed to WT, and the segregation of the mutations with the phenotype was analysed in F2 plants. The M. truncatula cngc15GOF mutant lines were backcrossed to A17::YC3.6, dmi1-1 (C71)::YC3.6 (ref. 19), A17::ENOD11:GUS45 and dmi3-1 (ref. 46). Transformed roots expressing DMI1D470A-NLS:YC3.6, DMI1D470A-E521Q-NLS:YC3.6 DMI1-NLS:YC3.6 DMI1TVGYG-NLS:YC3.6, RNAiCaM2-NLS:YC3.6 or RNAiCHS were generated via A. rhizogenes-mediated gene transfer, as described previously47 using the A. rhizogenes strain AR1193.
Seed sterilization and germination
M. truncatula seeds were scarified using sandpaper, treated with 10% sodium hypochlorite for 4 min, washed in sterile dH2O and imbibed in sterile dH2O for 5 h. The seeds were stratified onto 2% agar for 6 days in darkness at 4 °C and germinated overnight in darkness at 23 °C. For phenotyping experiments, the germinated seedlings were grown on modified Fahraeus medium48 in controlled environment for 7 days (23 °C, 16-h photoperiod and 300 µmol m−2 s−1) before transfer into soil.
Triticum sp. seeds were sterilized with 2% sodium hypochlorite for 4 min then washed five times in sterile dH2O. The sterilized seeds were stratified for 7 days at 4 °C and germinated for 2 days at 23 °C in darkness. The germinated seedlings were grown on modified Fahraeus medium in a controlled environment for 5 days (23 °C, 16-h photoperiod and 300 µmol m−2 s−1) before transfer into soil.
Genotyping of segregating F2 population
Genomic DNA samples were extracted using DNeasy 96 Plant Kit (Qiagen) according to the manufacturer’s instructions. Single-nucleotide polymorphism of M. truncatula CNGC15GOF, dmi1-1, dmi3-1 and Triticum sp. CNGC15GOF was detected via Kompetitive allele-specific PCR49 using allele-specific forward primers conjugated with hexachloro-fluorescein fluorescent dye or 6-carboxyfluorescein fluorescent dye, and common reverse primer. The primers used were synthetized by Sigma-Aldrich and are listed in Supplementary Table 1. Reactions were set up in 384-well optically clear plates (4titude Ltd), including 25 ng of genomic DNA, 12 µM of allele-specific primers, 30 µM common reverse primer and 2× PACE master mix (3CR Bioscience). Amplifications were performed in Mastercycler pro 384 (Eppendorf) with the following thermal cycle: 95 °C for 15 min, 10× (95 °C for 20 s; touchdown, 65 °C, −1 °C per cycle and 25 s), 50× (95 °C for 10 s; 57 °C for 60 s). Fluorescent signal data were detected using the PHERAstar plate reader (BMG Labtech). The genotype of homozygous mutants was further confirmed via Sanger sequencing (Genewiz). Genotyping of the T. turgidum cv. Kronos population BC3F2 was performed using primers P69 and P70, listed in Supplementary Table 1.
Genotyping of ENOD11:GUS was performed via PCR using GUS specific primers listed in Supplementary Table 1.
Protein sequence alignment
Amino acid sequences were obtained from the Mt4.0v1 and the International Wheat Genome Sequencing Consortium databases and are listed in Supplementary Table 2. Amino acid sequences were aligned using MUSCLE v.3.8.425 (ref. 50). Alignment of the S1 transmembrane domain of CNGC15 is listed in Supplementary Fig. 1.
Rhizobia infection and nodulation assays
Analysis of infection was performed using the S. meliloti strain Sm2011 (optical density at 600 nm (OD600) = 0.01)51 transformed with a constitutive hemA-β-galactosidase (LacZ) reporter gene fusion (pXLGD4) (ref. 52) for visualization. One-day-old seedlings grown on sterile Whatman filter paper and buffered nodulation agar medium (BNM)53 containing 0.1 µM l-α-(2-aminoethoxyvinyl) glycine were inoculated and grown, as described previously4. For LacZ staining, roots were fixed in 100 mM sodium phosphate (pH 7.0), 10 mM KCl, 1 mM MgCl2 and 2.5% glutaraldehyde and stained for β-galactosidase activity overnight in 0.1 M sodium phosphate (pH 7.2), 5 mM K3Fe(CN)6, 5 mM K4Fe(CN)6 and 0.02 M 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside yielding a blue precipitate (Fig. 1c) or 5-bromo-6-chloro-3-indolyl β-d-galactopyranoside yielding a magenta precipitate (Fig. 2f) at 30 °C. Infection threads and infection pockets were scored using a light microscope (Zeiss Axiophot).
For nodulation assays, 1-week-old plants were grown in terragreen/sand (Oil-Dri Ltd) to a ratio (1:1) and inoculated with Sm2011 (OD600 = 0.01). Plants were grown in a controlled-environment room at 22 °C (80% humidity, 16-h photoperiod and 300 µmol m−2 s−1). Nodules were scored as indicated. For nitrate experiments, plants were watered with BNM with or without 3 mM KNO3.
Mycorrhization assays
For M. truncatula mycorrhization experiments, 7-day-old seedlings were transferred in 90% 1:1 sand:terragreen (Oil-Dri Ltd) and 10% mycorrhizal inoculum (homogenized soil substrate containing Allium schoenoprasum roots colonized by R. irregularis DAOM197198). Plants were grown in a controlled-environment room at 22 °C (80% humidity, 16-h photoperiod and 300 µmol m−2 s−1).
For Triticum sp. mycorrhization experiments, 5-day-old seedlings were transferred in 80% 1:1 sand/terragreen mix and 20% mycorrhizal inoculum (as above). Wheat plants were grown in a controlled-environment room at 22 °C (35% daytime humidity and 50% night-time humidity, 16-h photoperiod and 500 μmol m−2 s−1).
Mycorrhizal fungal structures were visualized after acidic ink staining performed as follows: M. truncatula roots were cleared in 10% KOH for 5 min at 96 °C, rinsed three times in dH2O and stained in acidic ink (5% black ink; Waterman) and 5% acetic acid) for 3 min at 96 °C. Triticum sp. roots were cleared in 10% KOH for 25 min at 96 °C, rinsed three times in dH2O and stained in acidic ink for 10 min at 96 °C. Roots were destained in 70% chloral hydrate for 10 min and stored in dH2O. Mycorrhizal colonization were visualized using an M80 microscope (Leica) and quantified using the grid intersect method54.
Spontaneous nodulation assay
One-week-old M. truncatula seedlings were planted into 1:1 terragreen:sand (Oil-Dri Ltd) in phytaboxes under sterile conditions. The plants were grown in a controlled-environment room at 22 °C (80% humidity, 16-h photoperiod and 300 µmol m−2 s−1). Spontaneous nodules were scored after 98 days.
Shoot and root dry weight
The roots and shoots were separated and dried for 7 days at 28 °C before measuring dry weights with an analytical scale.
CHN analysis
Leaf tissues were dried for 7 days at 28 °C. Dried leaf material was ground in liquid nitrogen, and CHN elemental microanalysis was performed at Butterworth Laboratories. For M. truncatula, total dried leaves from five to seven plants were ground per biological replicate. For Triticum sp., the dried flag leaf from five to seven plants were ground per plot replicate.
Alphafold2 and structural homology modelling
The structure of M. truncatula CNGC15a homotetramer was predicted with AlphaFold2 multimer, as implemented through ColabFold (v.1.5.2)55 (https://colab.research.google.com/github/sokrypton/ColabFold/blob/main/AlphaFold2.ipynb). Four copies of the 710-residue CNGC15a sequence (UniProt accession G7IBJ4) were uploaded to the server, and each of the five independent models was subjected to 20 recycles to improve the accuracy of the predictions. With the exception of regions at the termini of each chain and a few short surface loops, the remainder of the sequence was well predicted according to the quality metrics (Supplementary Fig. 2). After removal of the termini, specifically residues 1–56 and 607–710 in each chain, the remaining core structures gave high overall predicted local distance difference test (pLDDT) scores in the range of 79.7–86.9. Superpositions of the CNGC15a AlphaFold2 models and related structures, as well as prediction of the structural consequence of P98S or P98L, were made using the Secondary Structure Matching algorithm within COOT v.1 (ref. 56). In pairwise comparisons, the five MtCNGC15a AlphaFold2 models (truncated as above) gave root-mean-square deviations in the range of 1.099–2.200 Å for the tetramers. The larger values being mostly the result of small changes in the relative tilts of the domains and subunits with respect to one another rather than localized differences. By contrast, superposing the regions encompassing only the transmembrane portion (specifically residues 57–400) of a single subunit from each model gave values in the range of 0.558–0.808 Å. All structural figures were prepared using ChimeraX v.1.5 (ref. 57).
Molecular cloning
The NLS:YC3.6, NLS:YC3.6-DMI1, NLS:YC3.6-DMI1TVGYG, NLS:YC3.6-DMI1D470A, NLS:YC3.6-DMI1D470A-E521Q and CNGC15cGOF–mCherry constructs were generated via Golden Gate cloning according to a previous study58. Assembled level 1 modules were cloned in a level 2 binary vector backbone, as presented in Supplementary Fig. 3. Level 0 modules were synthetized by Life Technologies (Thermo Fisher Scientific) except for DMI1TVGYG, DMI1D470A, DMI1D470A-E521Q and CNGC15cGOF, which were generated via site directed mutagenesis using the primers P32/P33 for DMI1TVGYG, P56/P57 for DMI1D470A, P56–P59 for DMI1D470A-E521Q and P34–P37 for CNGC15cGOF followed by Golden Gate cloning to generate the level 0 module. The primer sequences are indicated in Supplementary Table 1. The non-selective selectivity filter of DMI1 was mutated to the potassium selectivity filter identical to KcsA22, MthK24 and BK channels23. The amino acid positions 292–296 of MtDMI1 was mutated from ‘ADAGN’ to ‘TVGYG’.
The previously generated silencing construct RNAiCaM2 in pK7GWIWG2D(II)8 was modified as follows to assess the nuclear calcium oscillation. The pAtUBI:DsRed expression cassette was substituted by pAtUBI:NLS:YC3.6 via restriction digestion using AatII and ApaI (New England Biolabs) and ligated with T4 ligase (New England Biolabs) to obtain the pK7GWIWG2D(II)YC3.6 empty vector and pK7GWIWG2D(II)YC3.6::RNAiCAM2.
The RNAiCHS construct was generated using BP/LR Gateway cloning (Invitrogen). The sequence published in a previous study41 was amplified from M. truncatula cv. Jemalong A17 roots cDNA using the primers P60/P61 and cloned into the donor vector pDONR207 and subsequently into the destination vector pK7GWIWG2D(II)R following the manufacturer’s instructions (Invitrogen).
Analyses of Ca2+ oscillation
Ca2+ oscillation measurements were performed using a Nikon ECLIPSE FN1 equipped with an emission image splitter (OptoSplit II; Cairn Research) and an electron multiplying cooled charge coupled (Rolera Thunder EMCDD) camera (QImaging). enhanced cyan fluorescent protein (ECFP) was excited using light-emitting diode (OptoLED; Cairn Research) at 436 ± 20 nm and emitted fluorescence detected at 535 ± 30 nm (cpVenus) and 480 ± 40 nm (ECFP). Images were collected in 3-s intervals for 1 h and 30 min using MetaFluor software. Two-day-old seedlings or 2-week-old transformed roots were placed in a chamber made on a 48 × 64-mm coverglass (Solmedia) using high-vacuum grease (Dow Corning GmbH). The chamber was filled with 1 ml BNM. Only the root was covered with a coverslip to leave space for S. meliloti Nod factor application at a final concentration of 10−8 M. Nod factor was produced as described previously59. Ca2+ imaging was performed on 2-cm-long roots and on the root hair cells of the induction zone. Supplementary Videos 1 and 2 show root hair cells displaying the NF-induced Ca2+ spiking in WT and spontaneous Ca2+ spiking in cncg15aGOF, respectively. Each video represents 9 min of pseudo-coloured cpVenus to display calcium concentration variation from low to high (green to red). Ca2+ oscillation traces were analysed as described previously8.
Complementation of yeast lacking potassium transporters
DMI1 and DMI1TVGYG were cloned via directional topoisomerase-based cloning (TOPO) cloning using primers P54/55 (Supplementary Table 1) into the pDONR207 vector according to Thermo Fisher Scientific manual’s instructions. They were subsequently subcloned into the yeast expression vector pAG414GAL-ccdB-HA (Addgene plasmid #14239) using LR Clonase II Enzyme mix (Invitrogen) according to the manufacturer’s instructions. The vectors were transformed into the MAB2d strain, which lacks the two potassium transporters Trk1 and Trk2 (MATa ade2–1 can1–100 his3–11,15 leu2–3,112 trp1–1 ura3–1 mall0 ena1Δ::HIS3::ena4Δ nha1Δ::LEU2 trk1Δ::LEU2 trk2Δ::HIS3) (ref. 25), via lithium acetate method60. The transformants were grown on a synthetic dropout (SD) medium lacking tryptophan (Trp) and supplied with 100 mM KCl, 2% glucose and 2% agar. The transformants were selected after 3 days of growth at 28 °C and subsequently inoculated in 5 ml liquid SD medium lacking Trp and supplied with 100 mM KCl, 1% sucrose and 1% galactose at 28 °C overnight. To perform the growth assay, the yeast cultures were centrifuged, washed in water and resuspended to an OD600 = 0.01 in 2.5 ml of SD 1% galactose and 1% raffinose medium lacking Trp and supplemented with the indicated concentration of KCl. The cultures were grown in a 12-well CytoOne plate (Starlab Ltd) at 28 °C in a shaker incubator (200 rpm) for 1 and 7 days, as indicated. The OD600 was measured using a BioPhotometer Model #6131 (Eppendorf).
Wheat field trial
Field experiments were sown at the John Innes Centre experimental trial sites in Bawburgh, UK, at two different locations: 52° 37′ 50.7″ N, 1° 10′ 39.7″ E in 2023 and 52° 37′ 37.9″ N, 1° 10′ 54.0″ E in 2024. The soil of both fields is classified by the Land Information System at the Cranfield Environment Centre as slightly acidic loamy and clayey soils with impeded drainage. Each trial site was arranged in a split-plot design with two sections of 18 × 1.1 m in 2023, and 30 × 1.1 m in 2024. Commercial AM inoculum (Empathy rootgrow mycorrhizal fungi; Royal Horticultural Society) was applied to one section at a concentration of 1.11 kg m−2. A wide strip of 4 m separated the inoculated and non-inoculated sections to avoid contamination. In 2023, five independent replicates for each Triticum sp. WT/mutant plants were grown per section, and in 2024, three independent replicates for each Triticum sp. BC3 mutant NIL and WT NIL were grown per section. During the trial growing season, 2023 was slightly cooler than 2024, with lower average temperatures (2023, 11.6 °C; 2024, 12.1 °C). Both years experienced equivalent precipitation levels (2023, 156.4 mm; 2024, 165.9 mm), albeit 2023 had a 2-day event midway through the trial that accounted for 32% of the total rainfall. In terms of quantum radiation, 2023 had a higher average daily quantum radiation (435.8 μmol m−2 s−1) compared to 2024 (335.2 μmol m−2 s−1). Excel v.16.0 was used to calculate the average (Supplementary Table 3).
Field soil analysis
To assess the soil nutrient quality of the field, 700 g of soil per sample was collected around the section at the same time as the section was dug out for AM analysis. The soil samples were analysed by Dove Associates Ltd. The amounts of Ca2+, sodium, aluminium, sulfur, potassium, magnesium, manganese, boron, copper, iron, zinc and molybdenum were assessed via inductively coupled plasma optical emission spectroscopy. Extractable phosphorus was measured via the Olsen extraction method. Available ammonium and nitrate were extracted with 2 M potassium chloride solution and measured by colorimetry (Supplementary Table 4). Excel v.16.0 was used to calculate the average.
Histochemical GUS staining
M. truncatula roots were fixed in ice cold 90% methanol for 2 h at −20 °C. The roots were washed three times for 10 min in a reaction buffer containing 0.1 M Na2HPO4–NaH2PO4 (pH 7) and incubated in a reaction buffer supplemented with 1 mM K3Fe(CN)6, 1 mM K4Fe(CN)6, 5 mM NaEDTA, 0.1% Triton X-100 and 2 mM 5‐bromo‐4‐chloro‐3‐indolyl‐β‐d‐glucuronide. Samples were gently vacuum-infiltrated for 30 min and then incubated at 37 °C, in the dark, for 24 h. The samples were imaged with a DM6000 microscope (Leica) equipped with a DFC420 colour camera (Leica).
Localization
M. truncatula A17 was transformed with A. rhizogenes AR1193 carrying the Golden Gate constructs p35S:mCherry:T35S-pNOS:CNGC15cGOF:GFP:TNOS or p35S:mCherry:T35S. Two-week-old transformed roots were analysed for green fluorescent protein (GFP) and mCherry fluorescence using the confocal laser scanning microscope Zeiss LSM 980 (GFP: excitation 488 nm and emission imaged between 500 and 530 nm; mCherry: excitation 587 nm and emission imaged between 600 and 620 nm).
RNA-seq analysis
M. truncatula seedlings were grown on BNM agar medium supplemented with 0.1 µM l-α-(2-aminoethoxyvinyl) glycine for 24 h before treatment. The seedlings were incubated in six-well plates supplemented with liquid BNM with and without 10−8 M Nod factor, and with or without 3 mM KNO3− for 3 h. Total RNA was extracted from root infection zones corresponding to 1 cm from the first elongated root hair upward (with RNeasy Plant Mini Kit; Qiagen) according to the manufacturer’s instructions. Genomic DNA was removed with TURBO DNA-free (Invitrogen) according to the manufacturer’s instructions. RNA library preparation and sequencing were performed by Novogene. RNA libraries were prepared with the Illumina TruSeq Stranded mRNA HT technology. RNA-seq was performed using the Illumina HiSeq platform with paired-end 150-bp (PE 150) strategy.
The resulting reads were quality controlled using FastQC v.0.11.8 and Trim Galore v.0.6.10, and mapped to M. truncatula v5 genome (MtrunA17r5.0-ANR) using STAR v.2.5.a. Gene read counts were retrieved using HTSeq v.0.9.1. Differentially expressed genes (DEGs) were identified by pairwise comparisons using DESeq2 v.3.18 package in R. DEGs with a false discovery rate-corrected P value < 0.05 were used for further analyses. Heat maps were generated in R using heat map package. Gene ontology over-representation analysis was performed with PANTHER v.18.0 online tool using the corresponding M. truncatula v4 gene IDs. Previously published RNA-seq reads used in this study were retrieved from the National Center for Biotechnology Information Sequence Read Archive (accession number SRP099836)35 and the Gene Expression Omnibus (accession number GSE154845)34.
Gene expression analyses
RNA was extracted from M. truncatula root induction zone of 24-h-old seedlings after 3-h treatment, as indicated or root inoculated with Sm2011 for 21 days with RNeasy Plant Mini Kit (Qiagen) and subsequently treated with TURBO DNA-free (Invitrogen) before performing the reverse transcription with 1 µg RNA using SuperScript IV reverse transcriptase (Invitrogen). The quantitative gene expression was monitored with SYBR Green (Sigma-Aldrich) based quantitative PCR on a Bio-Rad thermocycler using gene specific primers. UBC9 (TC106312)61 was used for normalization. The primers used for quantitative reverse transcription polymerase chain reaction (qRT-PCR) are listed in Supplementary Table 1.
Phylogeny
CNGC15 sequences were identified through BLASTp and BLASTn v.2.13 searches against genomes on Phytozome v.13 (ref. 62) (Supplementary Table 5). A phylogenetic tree was constructed, as described in a previous study8. IQ-TREE (v.2.2.3) was used to construct the phylogenetic tree with the maximum likelihood approach. ModelFinder was used to find the best-fit evolutionary model (TPM2u+R5), and ultrafast bootstrap (UFBoot2) with 1,000 replications was used to estimate branch support63. Three figures were generated with the interactive Tree Of Life v.6 (ref. 64).
Metabolite profiling analysis
Roots from M. truncatula inoculated with S. meliloti strain Sm2011 (OD600 = 0.01) for 21 days, R. irregularis for 5 weeks or non-inoculated were freeze dried and ground in liquid nitrogen. The root systems from two to three M. truncatula plants were used per biological replicate with a minimum of 36 mg ground root tissue used for analysis. Metabolite profiling was carried out by MS-Omics by Biogenity as follows. The analysis was carried out using a Vanquish LC (Thermo Fisher Scientific) coupled to a Orbitrap Exploris 240 mass spectrometer (Thermo Fisher Scientific). An electrospray ionization interface was used as ionization source. Analysis was performed in positive and negative ionization modes under polarity switching. The ultra-performance liquid chromatography–mass spectrometry (LC–MS) was performed, as described65 with the following modification. Peak areas were extracted using Compound Discoverer 3.3 (Thermo Fisher Scientific). Identification of compounds was performed at four levels. Level 1 is identification by retention times (compared against in-house authentic standards), accurate mass (with an accepted deviation of 3 ppm) and MS/MS spectra. Level 2a is identification by retention times (compared against in-house authentic standards) and accurate mass (with an accepted deviation of 3 ppm). Level 2b is identification by accurate mass (with an accepted deviation of 3 ppm) and MS/MS spectra. Level 3 is identification by accurate mass alone (with an accepted deviation of 3 ppm). All the raw data are presented in Supplementary Table 6.
Targeted metabolite analysis of wheat roots
The liquiritigenin analysis in 7-week-old wheat roots was carried out by Biogenity as follows. Methanol, formic acid, acetonitrile and the standard substances of liquiritigenin were obtained from Sigma-Aldrich. For the sample preparation, 0.1 g of the root samples was weighed, and 4 ml 80% MeOH was added. The samples were sonicated for 30 min at room temperature. The sonicated samples were centrifuged at 10,000 rpm for 10 min, and 1 ml of the clear supernatant was transferred to a new tube and dried using a vacuum concentrator. The dried extracts were reconstituted in 200 µl 80% MeOH and centrifuged at 10,000 rpm. The supernatant was transferred to a high-performance liquid chromatography (HPLC) vial for LC–MS analysis. LC–MS analysis was performed using a Vanquish Core HPLC system (Thermo Fisher Scientific) coupled with an Orbitrap Exploris 120 mass spectrometer (Thermo Fisher Scientific). A Waters ACQUITY UHPLC HSS T3 column (2.1 × 100 mm; 1.7 µm) was used for the separation. The mobile phases consisted of (A) water with 0.1% formic acid and (B) acetonitrile with 0.1% formic acid. The elution gradient was as follows: 0 min at 10% B, linearly increasing to 100% B over 10 min, maintaining 100% B for 2 min, returning to 10% B at 12.1 min and re-equilibrating at 10% B until 15 min. The flow rate was set at 0.3 ml min−1, and the column temperature was 40 °C. The injection volume was 2 µl. Orbitrap Exploris 120 was operated in the electrospray ionization (ESI) positive mode, spray voltage was 3.5 kV, sheath gas was 50 a.u., Aux gas 10 a.u., ion transfer tube temperature was set to 325 °C and vaporization temperature was set to 350 °C. For detection, a SIM scan was used with an isolation window of 2 m/z, resolution of 30,000 and centre mass of 257.0808 for liquiritigenin. For the data analysis and quantification, a stock solution of 1 mM was prepared for liquiritigenin in 80% MeOH. This solution was further diluted to obtain final concentrations of 1.58, 0.78, 0.39, 0.19, 0.097 and 0.048 µM to construct the calibration curve. The quantification was performed by linear regression of the peak area against concentration. Data processing and analysis were conducted using TraceFinder 5.1 (Thermo Fisher Scientific).
The retusin analysis in wheat roots was carried out as follows. T. aestivum cv. Cadenza WT and Tacngc15aGOF were grown in 1 l 3:1 terragreen:sand in a controlled-environment greenhouse for 7 weeks. All harvested wheat roots were dried in VirTis freeze dryer, and then 200 mg dry weight root powder was mixed with 8 ml of 80% methanol (Sigma-Aldrich) containing 0.2 μM digitoxin (Sigma-Aldrich) as internal control and shaken at room temperature for 1 h. Samples were centrifuged at 13,000 rpm for 10 min. The supernatant was filtered through a Puradisc syringe filter, 0.45 µm, polytetrafluoroethylene (PTFE) membrane (Sigma-Aldrich). Samples were evaporated in Geneva HPLC mode overnight and redissolved in 200 μl methanol (Sigma-Aldrich). Samples were analysed on an ACQUITY Premier ultra-performance liquid chromatography equipped with a TQ Absolute tandem mass spectrometer (Waters). Chromatography was performed on a Kinetex EVO C18 column (100 × 2.1 mm, 2.6 μm, Phenomenex), remained 40 °C. The following gradient of 100% water with 0.2% formic acid (solvent A) versus acetonitrile with 0.2% formic acid (solvent B) was performed in the system: 0 min, 2% B; 10 min, 98% B; 11 min, 98% B, 11.1 min, 2% B; and 15 min, 2% B. The flow rate was 0.6 ml min−1, and the injection volume was 5 μl. MS detection was by positive mode electrospray monitoring the transition m/z 359 [M + H+] → 300.9 at a cone voltage of 52 and 32 V collision energy. The identity of retusin was confirmed by checking alternate fragments at m/z 106.8 (52 V collision energy), m/z 257.7 (44 V) and m/z 328.9 (28 V). Retention time was assessed for compound identification. The spray chamber conditions were 3 kV capillary voltage, 900 l h−1 drying gas at 500 °C and 150 l h−1 cone gas. To quantify retusin, 1 mM stock of retusin (Sigma-Aldrich) was dissolved in methanol. Serial dilutions (0–179.17 ng μl−1) of retusin were prepared in methanol containing 2 μM digitoxin. The quantification of the standard curve was analysed using MassLynx v.4.2 software.
External flavonoid and flavonol applications
For M. truncatula nodulation assays, 1-week-old plants were transferred into terragreen/sand (Oil-Dri Ltd) to a ratio of 3:1 and watered with either water or water containing 0.1 µM naringenin, 0.1 µM liquiritigenin or 0.1 µM retusin. After 2 days, all plants were inoculated with Sm2011 (OD600 = 0.01). The plants were grown in a controlled-environment room at 22 °C (80% humidity, 16-h photoperiod and 300 µmol m−2 s−1) and were watered twice a week with water with or without 0.1 µM naringenin (Sigma-Aldrich), 0.1 µM liquiritigenin (Sigma-Aldrich) or 0.1 µM retusin (Sigma-Aldrich). Nodules were scored for small and mature nodules 14 days post-inoculation with Sm2011.
For M. truncatula mycorrhization experiments, 7-day-old seedlings were planted into 84% 1:1 sand/terragreen (Oil-Dri Ltd) and 16% mycorrhizal inoculum (homogenized soil substrate containing A. schoenoprasum roots colonized by R. irregularis DAOM197198) and watered with either water or water containing 0.1 µM naringenin, 0.1 µM liquiritigenin or 0.1 µM retusin. Plants were grown in a controlled-environment room at 22 °C (80% humidity, 16-h photoperiod and 300 µmol m−2 s−1) and were watered twice a week with addition water or 0.1 µM naringenin, 0.1 µM liquiritigenin or 0.1 µM retusin solution.
For Triticum sp. mycorrhization experiments, 5-day-old seedlings were transferred in 80% 1:1 sand:terragreen mix and 20% mycorrhizal inoculum. Wheat plants were grown in a controlled-environment room at 22 °C (35% daytime humidity and 50% night-time humidity, 16-h photoperiod and 500 μmol m−2 s−1) and watered with water with or without 0.1 µM retusin twice a week.
Statistical analyses
Statistical significance was determined by two-tailed unpaired t-test with a previous F-test for homoscedasticity, one-way ANOVA followed by post hoc, as indicated using GraphPad Prism version 8 unless stated otherwise. Dot plots were used to show individual points. P values over 0.05 were considered nonsignificant. For metabolite analysis, the data were filtered, so only metabolites with at least 60% valid values across all the samples or 70% within an experimental group were used in the downstream data analysis. Filtration and statistical analysis were performed in the R programming by Biogenity using limma v.3.18 (ref. 66).
Accession numbers
Accession of the sequences from this study are listed in Supplementary Tables 2 and 5.
Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
