Introduction

In natural environments, plants interact simultaneously with a broad spectrum of both pathogenic and beneficial microorganisms that might influence plant performance and survival. Among soil microbes, arbuscular mycorrhizal (AM) fungi (subphylum Glomeromycotina1) establish a symbiosis with most plants living in wild and agroecosystems2. AM fungi colonise the root cortex, supplying mineral nutrients to plants in exchange for carbon compounds, thanks to the development of highly branched intracellular structures called arbuscules3. Besides improved mineral nutrition, plants colonised by AM fungi often display increased biomass and yield grain and a higher tolerance to biotic and abiotic stresses leading to a general improvement in plant fitness4,5,6,7. Considering the range of benefits provided by the fungal partner, the management of AM fungi in crop production is a cornerstone for future low-input and sustainable agriculture.

Many studies have focused on local and systemic transcriptomic and proteomic changes in rice8,9,10, maize11, Medicago truncatula12,13) and tomato plants14,15,16. By contrast, even though wheat (Triticum aestivum L.) is a major global crop, cultivated on more than 200 million hectares with more than 700 million tons of annual production17, its response to AM symbiosis has been poorly investigated. The main reason for this backwardness is that wheat has a hexaploid genome of 17 Gb in size, more than 80% of which is composed of repetitive transposable elements18. It is considered one of the most challenging genomes, since it has the genetic structure of three independent genomes in one species (AABBDD genome)19. A meta-analysis highlights the beneficial effects of mycorrhizal inoculation on wheat dry weight and phosphorus (P), nitrogen (N) and zinc (Zn) uptake20. However, the molecular determinants underlying the AM-related growth promotion and enhanced nutrient status in wheat are still poorly understood.

AM symbiosis is acknowledged to reduce damage caused by soil-borne pathogens including fungi, oomycetes21 and parasitic nematodes22,23. Similarly, in wheat, where pathogen attacks cause about 10–16% yield losses24, some of these stress events are alleviated by AM fungi25. The mechanisms involved in the bio-protective effect of AM fungi are not fully explained: they are not exclusively dependent on the improved mineral nutrition, but seem to be related to activation of plant defence mechanisms26. In addition, plant hormones and small RNA molecules (sRNAs) are attractive candidates for long-distance defence signals6. Plant hormones and small RNA molecules (sRNAs) are attractive candidates for long-distance defence signals6. Indeed, establishment of an AM symbiosis and production of AM signals activate defence-responsive genes in both shoot and root12,27,28,29. This boost of basal defences is known as priming, and it could be successfully triggered by various natural and artificial compounds, including AM fungi30. As a consequence, mycorrhizal plants are expected to be better protected against pathogen challenge than non-mycorrhizal plants: this phenomenon has been named mycorrhiza-induced resistance (MIR)4,6. MIR is dependent on the particular pathogen-mycorrhizal plant interaction and the plant organ under examination, i.e., root or shoot12,21,31,32,33,34.

The contrasting results obtained in such studies suggest that induction of resistance against pathogens depends on multiple mechanisms that may operate simultaneously4,31,33,34. The potential protective effect of mycorrhizal symbiosis in wheat has been poorly investigated35,36. The main goals of this work were, first, to define the responsiveness of T. aestivum cv. Chinese Spring to AM symbiosis, and second, to elucidate the molecular mechanisms underlying the mycorrhizal phenotype. We looked for the main pathways involved in enhancing plant biomass and mineral nutrition, and in promoting the bio-protective effect against a leaf pathogen. To address these issues, we combined phenotypic, and molecular metabolomic approaches. We explored the plant growth effect exerted by AM fungi in both greenhouse and controlled environment conditions, and further evaluated the impact of AM symbiosis against Xanthomonas translucens, which is a specific pathogen of wheat leaves. By integrating whole-transcriptome sequencing (RNA-seq) with shotgun nanoflow scale liquid chromatography-tandem mass spectrometry (LC-MS/MS), we provide a comprehensive functional overview of both local and systemic transcriptomic and proteomic changes in roots and leaves during the mycorrhizal combination (plant and AM fungus), as well as during the tripartite interaction (plant, AM fungus and pathogen). Mineral and amino acid contents in roots, leaves and seeds, and protein oxidation profiles in leaves, supported the omics data, providing new insight into the mechanisms exerted by AM symbiosis to confer positive effects on wheat development, and resistance to a wheat pathogen.

Results

Greenhouse experiment

Table 1 presents the results of a 2 year greenhouse trial to evaluate the impact of Funneliformis mosseae inoculation in Chinese Spring wheat. ANOVA analysis showed that all measured agronomic traits, such as tillering capacity, vegetative biomass and yield, as well as qualitative traits, such as kernel weight and size, are significantly different between mycorrhizal (M) and non-mycorrhizal control (C) plants. The presence of F. mosseae correlated with greater tillering capacity and plant biomass and also with increased yield, kernel weight and size (Fig. S1a). AM fungal inoculation also led to significant increases in concentrations of P, Mg and Zn in M seeds. In addition, total amino acid (AA) content increased (methionine, ornithine, tyrosine and tryptophan were more abundant), while lysine content decreased (Fig. S1b,c).

Table 1 The mean values of agronomic and qualitative traits are reported.
Full size table

Phenotypic assessment under controlled conditions

To confirm the morphometric data recorded in the greenhouse, C and M plants were grown in controlled conditions in a growth chamber, and the biomass of their epigeous and hypogeous parts was measured at 50 and 63 dpi. Growth of both tissues was increased significantly in M plants compared with C plants (Fig. 1A, B). To better investigate the effect of AM symbiosis on plant yield, spike weight was evaluated in M and C plants at the end of their natural life cycle. M plants displayed higher spike weight than C plants (Fig. 1C).

Figure 1
figure 1

Effect of AM symbiosis on wheat biomass in different plant organs. Fresh weight of roots (A) and leaves (B) of the control (C) and mycorrhizal (M) wheat plants harvested at 50 and 63 days post AM fungus inoculation. (C) Spike fresh weight of control and mycorrhizal plants evaluated at the end of wheat natural life cycle. Data (means ± SD, n ≥ 6) were subjected to one-way analysis of variance (ANOVA). The asterisks indicated significant differences at the 5% level using Tukey’s test.

Full size image

In the same experiment, 12 pots were devoted to investigating the impact of the AM fungus on leaf infection with X. translucens. These plants were identified as MX, while control plants infected with the pathogen were identified as CX. For all samples, mycorrhizal success was evaluated 50 days post inoculation (dpi) by calculating the total length of root colonisation (F%) and the total number of arbuscules (A%) in M plants (F%: 53.5 ± 16.9; A%: 26.3 ± 9) and in MX plants (F%: 59.7 ± 12.6; A%: 34.5 ± 14.2). Similar colonisation values were detected 63 dpi in M (F%: 61.5 ± 16.3; A%: 38.5 ± 10.2) and MX plants (F%: 60.5 ± 19.2; A%: 40.5 ± 12.2), revealing that pathogen inoculation of the leaves did not inhibit root colonisation by the AM fungus, in the short term. At the same time (63 dpi, i.e., 14 days after inoculation with X. translucens), disease symptoms were evident. Lesion length was significantly reduced in MX plants compared with CX plants (Fig. 2A,B).

Figure 2
figure 2

Phenotypic evaluation of disease symptoms caused by the bacterial pathogen Xanthomonas translucens in control (C) and mycorrhizal (M) plants. (A) Disease area (cm) was assessed on leaves from control (LC) and mycorrhizal (LM) plants 24 h post inoculation (hpi) and 14 days post inoculation (dpi). (B) The pictures show lesions provoked by X. translucens on LC and LM 14 dpi. Data (means ± SD, n ≥ 6) were subjected to one-way analysis of variance (ANOVA). The asterisks indicated significant differences at the 5% level using Tukey’s test.

Full size image

These experiments demonstrate that AM symbiosis exerts a positive effect on wheat growth and provides protection against X. translucens.

A quantitative overview of transcript and proteomic data sets

RNAs and proteins were isolated from leaves (L) and roots (R) of wheat plants, grown in the absence (LC and RC) or in the presence of the mycorrhizal fungus F. mosseae (LM and RM), and following infection with the bacterial pathogen X. translucens (LMX and RMX).

For transcriptomic analysis, each treatment was sequenced in triplicates, with 37 million reads on average per replicate, and a minimum of 27 million and maximum of 41 million reads per replicate. Pearson correlation coefficients for biological replicate samples sharing the same treatment and tissue were always above 0.9 (Table S3). Also, for proteomic investigation, each treatment was analysed in triplicate leading to 2,750 proteins identified on average per replicate, with a minimum of 2,659 and a maximum of 2,800 proteins per replicate. Pearson correlation coefficients for biological replicate samples sharing the same treatment and tissue ranged from 0.95 to 0.99 (Table S4). All genes with a false discovery rate (FDR) below 0.05 and log2FC over 0.5 and proteins with FDR below 0.01 and log2FC over 0.3 were considered differentially expressed (differentially expressed genes, DEGs; and differentially expressed proteins, DEPs).

The overall changes in gene expression detected in the different comparisons are represented in a Venn diagram (Fig. 3). As expected, the AM fungus had a greater impact on the root system than bacterial infection (RM vs RC: 5,155 DEGs and RCX vs RC: 150 DEGs, respectively). The AM fungus had a deep impact on the leaf profile (LM vs LC: 9,097 DEGs) as well as the bacterial inoculation (LCX vs LC: 8,408 DEGs). The presence of Xanthomonas on leaves of mycorrhizal plants (LMX vs LM) led to a higher number of DEGs (13,302) than other comparisons, and 43% (5,777) of these DEGs were exclusively regulated in this contrast (Fig. 3B), suggesting that a synergistic effect occurs. As a consequence of the huge number of DEGs in LMX and LCX, their direct comparison led to a low DEGs number (97).

Figure 3
figure 3

Venn diagrams of DEGs modulated in the different comparisons in roots and leaves. Venn diagrams illustrating the relationships between DEGs in the different contrasts among the same organ (A) roots and (B) leaves in the absence (C) or presence (M) of the AM fungus and following (CX; MX) or not pathogen infection.

Full size image

Overall, these data reveal that the AM fungus has a strong local and systemic impact, while the pathogen exerts a local effect during the binary interaction, and both local and systemic ones when inoculated on mycorrhizal plants.

For a deeper analysis, we compared the expression of all identified proteins in the four samples (M, C, MX, X) by cluster analysis, on the basis of the modulation of their expression. Figure 4 shows some of the protein profiles obtained.

Figure 4
figure 4

Protein profiles of roots and leaves. For each sample three biological replicas were considered (indicated as 1, 2, 3). (A and B) Indicate profiles containing respectively up or down regulated proteins in mycorrhizal samples (RM, RMX, LM, LMX).

Full size image

Differentially regulated genes and proteins in wheat leaves and roots following colonisation by F. mosseae

RNA-seq analysis revealed 3,607 up-regulated and 1,549 down-regulated genes in M roots (Fig. 5A and Table S5). An even higher gene modulation was observed in leaves: 6,632 were up-regulated and 2,464 were down-regulated genes (Fig. 5C and Table S6). RNA-seq analysis revealed 532 and 2,220 transcripts that were exclusively expressed in roots (RM vs RC) and leaves (LM vs LC) of M plants, respectively.

Figure 5
figure 5

Global overview of the transcriptional and proteomic changes in the two organs (leaves -L and roots- R) in the absence (C) or presence (M) of the AM fungus. Mean expression versus log2 fold change plots (MA-plots, left side; Volcano plot, right side) were computed for these comparisons: (A,B) RM vs RC, and (C,D) LM vs LC. Called DEGs (A,C) and DEPs (B,D) (FDR 0.05) are plotted in color.

Full size image

Unlike wheat, rice has been much studied as a model plant for AM symbiosis9,10,37. To determine whether a common core of genes responds to AM symbiosis, we compared the available AM-rice root RNA-seq data set10 with the wheat data in the present work. Using a Reciprocal Best Hits (RBH), we identified 114 of the 1,088 up-regulated rice genes, which contained a corresponding sequence in mycorrhizal wheat roots (Table S7). Among them were several AM marker genes identified in other AM host plants: Glycerol-3-phosphate acyltransferase (OsRAM2 homolog)38, Gibberellin response modulator protein (LjRAD1 homolog)39, LysM domain-containing protein (OsLysM homolog)10, Ammonium transporter and Inorganic phosphate transporters (see nutrient uptake paragraph) and ABC-2 type transporter (OsSTR1 homolog)40 (Table S7). These genomic commonalities further support a key role for these genes in establishment of the AM symbiosis.

When RM samples were compared with RC samples, proteomic analysis revealed 586 up-regulated and 395 down-regulated proteins (Fig. 5B and Table S8), and a comparison between LM and LC samples revealed 175 up-regulated and 226 down-regulated proteins (Fig. 5D and Table S9).

To gain an overview of the overlap between transcriptomic and proteomic data sets, differentially regulated genes and proteins detected using the two high-throughput techniques were compared. In roots and leaves of mycorrhizal plants, transcriptomic and proteomic data sets shared 192 (3.7% of DEGs and 19% of DEPs) and 82 (0.9% of DEGs and 20% of DEPs) elements, respectively (Fig. S2; Tables S10 and S11).

Taking advantage of bioinformatics tools such as agriGO v2.041 and over-represented Gene Ontology (GO) categories, we identified the pathways elicited by the AM fungus locally (root) and systemically (leaf). In the following paragraphs, we illustrate those pathways that might better explain the effects on growth and bio-protection detected in mycorrhizal plants.

Nutrient uptake

Nutrient uptake is a crucial trait of the AM symbiosis; however, in wheat, AM-induced nutrient transporter genes are still poorly characterised. To help plug this knowledge gap, the transcription profiles of some phosphate transporters (PTs), ammonium transporter 3 member 1 (TaAMT3.1), high-affinity sulfate transporter 2 (TaSulfTr2), potassium channel (TaAKT1) and oligopeptide transport (TaOPT) were investigated by both RNA-seq and qRT-PCR analyses. We monitored transcript levels of the PT genes previously described in wheat as highly induced in M roots (TaPT10, TaPT11 and TaPT12)42,43, and clustered with the AM-induced PT genes (OsPT11; MtPT4; LjPT444,45,46; (Fig. S3) as well as the putative inorganic phosphate transporter 1–13 (TaPT13), which shows high homology with the AM-induced OsPT1347 (Fig. S3). All of these PT genes were strongly induced in RM versus RC (Fig. S4). The transcripts of TaAMT3.1, which shows high similarity to the AM-induced OsAMT3.1 from rice10,48, were detected exclusively in M roots (Fig. S4). The same gene expression profile was observed for TaAKT1 and TaOPT, whose transcripts were detected only in M roots. By contrast, although TaSulfTr2 was strongly induced in M roots, it was also expressed in C roots (Fig. S4). A comparable expression profile was detected for LjSultr1;2 which was induced in L. japonicus by both sulphur starvation and mycorrhizal formation49.

AM colonisation led to the differential regulation of several proteins involved in nutrient uptake. In agreement with the transcriptomic data, TaPT10 and TaAMT3.1 proteins were accumulated in RM. In addition, two proteins involved in N uptake were up-regulated in RM vs RC: an AMT which shows high similarity to OsAMT3.2 and a nitrate transporter with similarity to the tomato AM-inducible LeNRT2;350. Accordingly, two glutamine synthases, involved in N assimilation, were more strongly expressed in RM than in RC.

Several proteins involved in iron (Fe) uptake also accumulated in RM: one Fe-phytosiderophore transporter, some nicotianamine synthases (NAS) and two deoxymugineic acid synthases. Finally, a H+-ATPase and a copper (Cu)-transporting ATPase were also induced in RM. H+-ATPase shows high homology to OsHa1 and MtHA1, which energise nutrient uptake during mycorrhizal symbioses in rice and Medicago truncatula51. The Cu-transporting ATPase is 90% similar to HMA4 of Oryza sativa, which is involved in Cu accumulation in root vacuoles106. A false discovery rate threshold of 0.05 was set for DEG calling. Sample clustering and principal component analyses were performed upon variance stabilizating transformation of expression data. Transcripts were considered differentially expressed when the adjusted FDR values are below 0.05 and the logarithmic fold change over 0.5 or when the logarithmic fold change was over 2. The transcriptomics data related to roots and leaves of mycorrhizal plant and roots and leaves of mycorrhizal plants infected by X. translucens have been deposited in the European Bioinformatics Institute (EMBL-EBI) ArrayExpress database (https://www.ebi.ac.uk/arrayexpress) with the dataset identifier E-MTAB-5898, while the transcriptomic data from control plants and control plants infected by X. translucens with the dataset indentifier E-MTAB-589173. Unless otherwise stated, further graphical outputs were generated with custom R and Python scripts.

Protein extraction and Liquid Chromatography-Mass Spectrometry (LC-MS/MS) analysis

Total proteins were extracted from T. aestivum shoots and roots of M, C, CX and MX plants as described by Garcia-Seco et al.73.

MS analysis was performed on a QExactive mass spectrometer coupled to a nano EasyLC 1000 (Thermo Fisher Scientific Inc., Waltham, MA). Solvent composition at the two channels was 0.1% formic acid for channel A and 0.1% formic acid, 99.9% acetonitrile for channel B. For each sample, 4 μL of peptides were loaded on a self-made column (75 μm × 150 mm) packed with reverse-phase C18 material (ReproSil-Pur 120 C18-AQ, 1.9 μm; Dr. Maisch GmbH, Ammerbuch, Germany) and eluted at a flow rate of 300 nL/min by a gradient from 2 to 35% B in 80 min, 47% B in 4 min and 98% B in 4 min. Samples were acquired in a randomized order. The mass spectrometer was operated in data-dependent mode (DDA), acquiring a full-scan MS spectra (300–1700 m/z) at a resolution of 70,000 at 200 m/z after accumulation to a target value of 3,000,000, followed by HCD (higher-energy collision dissociation) fragmentation on the twelve most intense signals per cycle. HCD spectra were acquired at a resolution of 35,000 using a normalized collision energy of 25 and a maximum injection time of 120 ms. The automatic gain control (AGC) was set to 50,000 ions. Charge state screening was enabled and singly and unassigned charge states were rejected. Only precursors with an intensity above 8300 were selected for MS/MS (2% underfill ratio). Precursor masses previously selected for MS/MS measurement were excluded from further selection for 30 s, and the exclusion window was set at 10 ppm. The samples were acquired using internal lock mass calibration on m/z 371.1010 and 445.1200.

Proteomic data processing and bioinformatic analysis

Mass spectrometer raw files were analyzed by MaxQuant (version 1.5.3.28 and 1.5.3.30) with the match between runs (matching time window of 2 min) and label free quantification (LFQ) options selected. Tandem MS spectra were searched against UniProt T. aestivum (Version 2015–10, 100,800 entries), Uniprot Rhizophagus irregularis (Version 2015–10, 29,847) and Uniprot Xanthomonas translucens (Version 2015–10, 14,378 entries). Trypsin/P was chosen as the protease, cysteine carbamidomethylation was set as fixed modification, and oxidation of methionine and acetylation of the N-terminal as variable modifications. Peptide tolerance was set to 4.5 ppm, while MS/MS tolerance was set to 0.5 Da. Peptide-spectrum matches (PSMs) and proteins were validated with 1% FDR. Only PSMs with a minimum length of 7 amino acids were kept.

The raw data was first processed by Perseus (MaxQuant package) and the incorrected protein identifications (contaminants, decoy and “only identified by site” entries) were removed from the main data frame. LFQ intensities were Log2 transformed. Only protein groups detected in almost two of three biological replicate samples sharing the same treatment and tissue were considered unambiguously identified and were used for assessment of significant change. Missing value were imputed using the R package ‘imputeLCMD’ (https://cran.rproject.org/web/packages/imputeLCMD/). The experimental quality was checked using the multi-scatter plots tool for the analysis of Pearson correlation between the samples. To determine the differentially expressed proteins (DEPs) among considered conditions, we performed a multiple sample test using Anova with a permutation-based false discovery rate (FDR) cutoff of 0.01.

For the annotation of the unknown proteins a blast search was made against the Uniprot database viridiplantae (Version 2015–10, 3398870 entries), taking the first hit with a valid annotation. For the GO analysis of DEPs we used agriGO v2.0 (Tian et al. 2017) and for the cluster analysis, we used the “Profile plot” tool of Perseus program.

Aminoacid analysis

HPLC-grade water, HPLC-grade methanol (MeOH), formic acid, aspartic acid (Asp), asparagine (Asn), glutamine (Gln), glutamic acid (Glu), serine (Ser), threonine (Thr), cystine (Cys), alanine (Ala), proline (Pro), valine (Val), methionine (Met), tyrosine (Tyr), leucine (Leu), phenylalanine (Phe), tryptophan (Trp), lysine (Lys), histidine (His) arginine (Arg) ornithine (Orn), pipecolic acid (Pip), citrulline (Cit), GABA and two deuterated internal standards (L-Phenyl-d5-alanine and L-alanine 15N) were purchased from Sigma–Aldrich Co. (Dorset, UK).

Stock solutions for all compounds were prepared in distilled water (10 mM). A working mixture (1 mM of each aminoacid, AA) was prepared and used for the calibration (range 2–15 µM).

For the AA extraction, all samples were lyophilized and 0.1 g of each sample was re-suspended in 10 mL of 0.1% (v/v) formic acid in water/methanol (50:50). 10 μl of 10 mM deuterated internal standards were added. The mixture was then vortexed for 4 h in the dark, sonicated for 15′, centrifuged and the supernatant was collected.

The HPLC analysis was performed in a Finningan Surveyor MS plus HPLC system (Thermo Electron Corporation, CA, USA). Separation was achieved using C18 column (Phenomenex, synergi 4 u fusion-RP 80a 150 × 2.00 mm). The mobile phase was composed of (A) water with 0.1% (v/v) formic acid and (B) methanol/water (50:50) 0.1% (v/v) formic acid with flow rate 150 µL/min; gradient 0–3.0 min/2% (v/v) B, 3–16 min/2–50% (v/v) B. For the mass spectrometry quantification, a Finningan LXQ linear ion trap mass spectrometer, equipped with an ESI ion source (Thermo Electron Corporation, CA, USA), was used. The analyses were done in positive (spray voltage 4,5 kV, capillary temperature 270 °C) and in the multiple reaction monitoring (MRM) mode. The optimization of collision energy for each substance, the tuning parameters and the choice of fragments to confirm the identity of target compounds were done in continuous flow mode, by using standard solution at concentration of 5 μM (Table S2). The linearity of the method was considered adequate when square correlation coefficient (R2) was higher than 0.98, based on peak area. The limits of detection (LOD) and quantification (LOQ) were fixed at 1 µM and at 2 µM, respectively.

Minerals content analysis

All frozen samples, were lyophilized, dried in an oven at 60 °C for 8 h and then mineralized with 1 mL of HCl and 1 mL of hydrogen peroxide (H2O2). Samples were reconstituted in 1% HNO3 in Milli-Q water. Blanks were made with the same solvents and chemicals employed in the treatment and digestion of the samples, or with just 1% HNO3 in Milli-Q water. Calibration standard solutions were prepared from 1000 mgLl−1 standard solutions of Mg, Fe, Cu, Zn and K (Baker Instra-Analyzed). The determination of minerals was performed on a Thermo Fisher Solaar M6 atomic absorption spectrometer. Ca, Mg, Fe, Zn, K were determined at ppm levels by flame atomic absorption spectrometry (FAAS) with deuterium lamp background correction; Cu was determined at ppb levels by graphite furnace atomic absorption spectrometry (GFAAS) and Zeeman background correction. All parameters such as the wavelength and the bandpass were set according to the recommendations of the instrument Cookbook.

The phosphorous content determination was performed as reported by Chen and Toribara107.

Detection of carbonylated proteins

Carbonylated proteins were detected in all samples collected at 1 dpi and 14 dpi, as described previously71. Protein carbonylation index (arbitrary units) was measured as ratio between the optical density (OD) obtained from the whole lane of the immunoblot and the OD of Coomassie stain. For each set (Leave and Roots) data (means ± SD, n = 4) were subjected to one-way analysis of variance (ANOVA) and post-hoc Tukey’s test.