Seeds of rice (O. sativa L. japonica cv. Nipponbare) were sterilized and germinated on 1/2 Murashige and Skoog (MS) medium with 0.8% agar. After one week, rice seedlings were transferred and grown in plastic tubes containing sterilized sands without (mock) or with sand inoculum containing spores of Rhizophagus irregularis (Ri, purchased from Mycorise^®ASP, Premier Tech, Rivière-du-Loup, Québec, Canada). Plants were grown in a phytochamber with a 12-h day/night cycle at 30/28°C and 70% air humidity. The plants were regularly watered for the first week after inoculation and fertilized every second day with one-half Hoagland solution containing 25 μM phosphate (Pi). At five weeks postinoculation (5 wpi), mock and mycorrhizal plants were divided into two batches, in which one batch was treated with fertilizer solution supplemented with 150 mM NaCl (saline condition) and the other batch was grown under nonsaline conditions. At 8 wpi, shoots and roots were collected and separated to detect the physiological, biochemical and molecular responses. One biological replicate was the combination of two plants, and three biological replicates were collected from each treatment.

Root samples were stained with Trypan blue, and mycorrhizal colonization was quantified with a modified gridline intersection procedure as described (Paszkowski et al., 2006).

For biomass measurement, the dry weight was measured after two days at 70°C in a hot-air oven. To monitor the tissue ion content, 0.05 g of chopped dry samples was digested with 2 ml 65% HNO₃ and 0.5 ml H₂O₂ (both Suprapur; Merck) in a MarsXpress microwave digestion system (CEM, Matthews, NC, USA). After that, the samples were diluted with Milli-Q water (Millipore Co., MA, USA) to 20 ml, filtered using a 0.45‐μm membrane filter and injected into an inductively coupled plasma (ICP) analyzer. The elemental profile of plant samples was determined using inductively coupled plasma‐optical emission spectrometry (ICP‐OES; PerkinElmer OPTIMA 5300) according to a previously described method with some modification (Shanmugam et al., 2011). The relative ion concentration (mg/L) of each sample was obtained based on the wavelength intensity and calibration standard curve. For 3,3’-diaminobenzidine (DAB) staining, the youngest fully expanded leaves detached from 8-week-old plants were soaked in DAB solution (0.1%) overnight and then decolorized in bleaching solution (96% ethanol:acetic acid:glycerol = 3:1:1) at 70°C until the brown spots appeared clearly. To statistically present the H₂O₂ content visualization by DAB staining, all pictures were first transformed into 8-bit grayscale images. The percentage of brown area (the area of brown spots divided by the area of whole leaf) was quantified by using ImageJ.

RNA extraction, cDNA synthesis, RT−PCR, and RT−qPCR were performed as previously reported (Gutjahr et al., 2008). Total RNA was extracted from 100 mg shoot or root tissue by using TRIzol Reagent (Invitrogen™) following the manufacturer’s instructions. For the following gene expression analysis, genomic DNA was removed from total RNA using DNase I (RNase-free, Invitrogen™). After DNase I treatment, purified RNA samples were reverse transcribed using the Moloney murine leukemia virus reverse transcriptase kit (Invitrogen™) with oligo (dT) primers for cDNA synthesis. The RT-qPCRs were performed with SYBR^® Green Supermix (2X) (Bio-Rad) on a CFX Connect Real-Time PCR Detection System (Applied Biosystems) as described in the manufacturer’s protocol. Transcript levels were normalized to constitutively expressed Cyclophilin2 (Gutjahr et al., 2008). The RT−qPCR primers were designed using Primer3 web version 4.1.0 (http://primer3.ut.ee/). All RT−qPCR primers are listed in Supplementary Table S9 .

A total amount of RNA (≥ 4 μg) per sample was used for transcriptome sequencing. The quantity and purity of the purified RNA was assessed using a NanoDrop ND-2000 (Thermo Scientific, Wilmington, MA, USA) and an Agilent 2100 Bioanalyzer (Agilent Technologies, Inc., Santa Clara, CA, USA) before Illumina NGS sequencing. To construct the sequencing libraries, only high-quality RNA samples were used (OD260/280 = 1.8~2.0, OD260/230 ≥ 2.0, RIN ≥ 7.0). After purification, end repair, and ligation to sequencing adapters, strand-specific sequencing libraries were constructed by using the Illumina HiSeq 4000 platform (2 × 150 bp paired end) and following the handbook for the NEBNext^® Ultra™ RNA Library Prep Kit for Illumina^® (Genomics Biotechnology Co., Ltd, Taipei, Taiwan). The read depth is 6 Gb. The quality control of raw reads was evaluated using FastQC software v0.11.8 (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/). The low-quality reads (Q-value < 20) and adaptors were removed using the sequence preprocessing tool Trimmomatic v0.39 (Bolger et al., 2014). After low-quality reads and adaptors were removed, all reads were mapped to Oryza sativa v7.0 from Phytozome 13 (Goodstein et al., 2012) with STAR (Dobin et al., 2013) and quantified with Salmon (Patro et al., 2017) using the nf-core/rnaseq pipeline (Ewels et al., 2020).

Genes with low expression (log2CPM < -3) were filtered out based on the CPM value. DEGs were identified using the edgeR (empirical analysis of DGE in R) package v3.10.5 by comparing the counts per million (CPM) values between treatment groups (Robinson et al., 2010). Three treatment groups (salt, AM and tissue) were categorized to identify DEGs responding to salt, AM symbiosis or tissue from pairwise comparison between samples with one factor difference. For example, salt-responsive genes were identified between the same tissues with the same mycorrhizal treatment but grown under control and salinity conditions. AM-responsive genes were identified between the same tissues with the same salinity treatment but grown under mock and mycorrhizal conditions. Tissue-responsive genes were identified between roots and shoots with the same salinity and mycorrhizal treatment. In each group, four pairwise comparisons were performed. An adjusted p value and false discovery rate (FDR) < 0.05 and an absolute value of log2-fold-change > 1 were considered significant thresholds. To further analyze the overlap of DEGs among different treatment groups, a custom Python script was used to plot the Venn diagram of the DEGs responsive to salt, AM symbiosis and tissue.

After obtaining the DEGs in each region from Venn diagram analysis, a gene ontology (GO) enrichment analysis of the DEGs was performed with CARMO, a web-based tool for GO analysis (Wang et al., 2015) with “MSU RGAP ID” selected. The output GO list from CARMO was submitted to REVIGO (Supek et al., 2011) for visualization for interpretation. The bubble plot in the manuscript was plotted with a custom Python script using the scatterplot function from the package seaborn (Waskom et al., 2017). The GO terms from each region were sorted by FDR and filtered based on FDR < 0.05. Only the top 15 terms were visualized on the bubble plot.

Fold change values were calculated by edgeR (Robinson et al., 2010). TPM values were calculated by Salmon (Patro et al., 2017). Custom Python scripts were used to visualize the fold change value of the selected genes with the clustermap function from the seaborn package (Waskom et al., 2017).

The experiment to observe phenotypic data was arranged with three to seven biological replicates in each treatment. The mean values were compared using two-way ANOVA followed by a least significant differences (LSD) post hoc test and analyzed by R software.

Sequence data from this article can be found in the Michigan State University Rice Genome Annotation Project database (http://rice.plantbiology.msu.edu) using the following accession numbers: granule-bound starch synthase II (OsGBSSII) (LOC_Os07g22930), xyloglucan endotransglucosylase/hydrolases 19 (OsXTH19) (LOC_Os03g01800), low phosphate root 5 (OsLPR5) (LOC_Os01g03640), purple acid phosphatase 7 (OsPAP7) (LOC_Os11g34720) and OsCyclophilin2 (LOC_Os02g02890). The sequencing files discussed in this publication have been deposited in NCBI’s Gene Expression Omnibus (Edgar et al., 2002) and are accessible through Gene Expression Omnibus (GEO) Series accession number GSE200863 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE200863).

To examine whether mycorrhizal rice plants could maintain better growth under salt stress, 5-week-old rice seedlings inoculated with the AM fungus R. irregularis (Ri) or without (mock) were grown under nonsaline (0 mM NaCl) or saline (150 mM NaCl) conditions for another three weeks. To investigate the effect of salinity on fungal growth, the fungal colonization level was quantified. Mock plants did not show the presence of AM fungi under either saline or nonsaline conditions (data not shown). The average fungal colonization levels reached 91% and 93% under nonsaline and saline conditions, respectively, indicating that AM fungi successfully colonized rice roots. The abundance of vesicles was slightly reduced by salinity, and the level of extraradical hyphae was higher under salt stress. The levels of the remaining fungal structures were not significantly different between nonsaline and saline conditions ( Figure 1A ). These results indicated that salt stress had a mild impact on AM symbiosis.

Mycorrhizal plants showed fewer wilted blade tips than mock plants under saline conditions ( Figure 1B ). Under nonsaline conditions, both the shoot and root biomass of mock plants and mycorrhizal plants were not different ( Figure 1C ), reflecting the fact that rice is an AM-nonresponsive plant (Smith et al., 2004; Li et al., 2006; Hata et al., 2010; Smith and Read, 2010). On the other hand, under saline conditions, the shoot biomass of mock plants but not mycorrhizal plants was severely decreased by salt stress ( Figure 1C ). The dry weight of mycorrhizal plant shoots was significantly higher than that of the mock group (1.5-fold) under salt stress. In contrast, AM symbiosis did not influence root biomass significantly under either nonsaline or saline conditions ( Figure 1C ). These results suggested that AM symbiosis may help rice plants maintain better shoot growth under salt stress.

We then aimed to further examine whether AM symbiosis also had positive effects on reproductive growth (panicle development) under salt stress. Compared with the mock plants, the mycorrhizal plants showed a 2.17- and 1.93-fold increase in differentiated spikelet numbers under control and salt stress conditions, respectively ( Figure 1E ). In addition, AM-colonized plants contained fewer salt-injured spikelets ( Figure 1D ), and the spikelet degeneration rate (number of salt-injured spikelets divided by total number of spikelets) in mycorrhizal plants was significantly lower than that in mock plants under salt stress ( Figure 1F ). These results further indicated that AM symbiosis could enhance spikelet tolerance to salt stress.

To evaluate the effect of salt stress on nutrient content, the concentrations of Na⁺, K⁺ and P were measured. Under salt stress, the Na⁺ concentration significantly increased compared to nonsaline conditions in both mock and mycorrhizal plants. Strikingly, mycorrhizal shoots presented a lower concentration of Na⁺ than mock shoots under salt stress ( Supplementary Figure S1A ). The ability to maintain a high cytosolic K⁺/Na⁺ ratio is an indicator of plant salt tolerance (Gregorio and Senadhira, 1993; Wu et al., 2013), so the K⁺ concentration was also measured to calculate the K⁺/Na⁺ ratio. The only difference in K⁺ concentration between mock and mycorrhizal plants was found in shoot tissue under salt stress, in which mock shoots accumulated more K⁺ than mycorrhizal shoots ( Supplementary Figure S1B ). Salt stress significantly reduced the K⁺/Na⁺ ratio in both mock and mycorrhizal plants. In roots, the K⁺/Na⁺ ratio of mycorrhizal plants was significantly higher than that of mock plants under nonsaline conditions ( Supplementary Figure S1C ). However, in shoots, the K⁺/Na⁺ ratio of mycorrhizal plants was significantly higher than that of mock plants under saline conditions ( Supplementary Figure S1D ). Even though the shoot/root Na⁺ ratio was significantly increased by salt stress, the shoot/root Na⁺ ratio was significantly lower in mycorrhizal plants than in mock plants, suggesting that mycorrhizal plants restrict sodium movement from roots to shoots ( Supplementary Figure S1E ). Previous studies showed that phosphate enhanced plant growth under saline conditions (Okusanya and Fawole, 1985). Therefore, we also measured the phosphorus content of plants. The results showed that the phosphorus content in mycorrhizal plants significantly increased in both shoots and roots compared with mock plants under both nonsaline and saline conditions ( Supplementary Figure S1F ). The positive effect of AM symbiosis not only enhanced salt tolerance, as shown by the K⁺/Na⁺ ratio, but also increased the efficiency of phosphate uptake.

Since AM symbiosis maintained shoot growth under salt stress ( Figure 1C ), whether the ROS level was different between mock and mycorrhizal shoots was further analyzed by visualizing H₂O₂ content histochemical stained with 3,3′-diaminobenzidine (DAB). It was obvious that the area of brown coloration was larger in mock shoots than in mycorrhizal shoots under salt stress ( Figure 2 ). The results suggested that AM symbiosis may enhance the reduction in H₂O₂ levels in the shoots to protect the rice plant from oxidative damage under salt stress.

To better understand the salt tolerance mechanism in mycorrhizal plants at the molecular level, RNA sequencing was used to investigate differential gene expression in response to AM symbiosis under salt stress. RNA-seq data were generated from the roots and shoots of mock and mycorrhizal plants grown under nonsaline and saline conditions with two to three biological replicates. Twenty‐two paired‐end libraries ( Supplementary Table S1 ) were generated, and 683,233,862 and 656,077,984 paired-end 150 bp raw reads were obtained from the control and salt treatment samples, respectively. After removing the low-quality raw reads (Q-value < 20) and trimming adaptor sequences, a total of 671,602,666 cleaned reads with > 90% Q30 bases from nonsaline samples and 635,393,116 from saline samples were selected as high-quality reads for further analysis ( Supplementary Table S1 ). All the high-quality reads served as input to the nfcore/rnaseq pipeline. The genome reference of Oryza sativa v7.0 was downloaded from Phytozome 13 (https://phytozome-next.jgi.doe.gov/). Sample correlation matrix analysis revealed a strong similarity between the biological replicates ( Supplementary Figure S2 ).

Differentially expressed genes (DEGs) affected by one of the three factors, salinity, AM symbiosis and tissue, were identified with cutoffs of | log₂ (fold-change) | > 1 and FDR < 0.05. For tissue-preferred expression, approximately nine thousand genes showed differential expression patterns between shoots and roots in each condition, and among these genes, approximately 45% showed higher expression in the roots than in the shoots. Moreover, this ratio was similar and not affected by salinity or AM symbiosis ( Figure 3A ). For AM-regulated expression, the number of DEGs regulated by AM symbiosis varied. For example, 2025 and 135 genes were regulated by AM symbiosis in the shoots grown under control and salinity conditions, respectively, and 1480 and 818 genes were regulated by AM symbiosis in the roots grown under control and salinity conditions, respectively. The number of AM-regulated DEGs was dramatically reduced in the roots and shoots grown under salt stress compared to those under control condition. Interestingly, more genes were downregulated in shoots but upregulated in roots by AM symbiosis under both control and salinity conditions ( Figure 3B ). For salinity-regulated expression, the number of DEGs regulated by salinity also varied. For example, 4058 and 676 genes were regulated by salinity in the shoots under mock and mycorrhizal conditions, respectively, and 1702 and 1223 genes were regulated by salinity in the roots under mock and mycorrhizal conditions, respectively. The number of salinity-regulated DEGs was significantly reduced in the tissues colonized by AM fungi, especially in shoots. In addition, salinity stress upregulated more genes in roots than no stress under both mock and mycorrhizal conditions. However, more genes were downregulated by salinity stress in mock shoots but upregulated in mycorrhizal shoots ( Figure 3C ). These results showed that the effect of AM symbiosis and salinity on gene expression was less pronounced under salinity and mycorrhizal conditions, respectively. In addition, compared to mock shoots, mycorrhizal shoots showed differential responses to salt stress.

In total, 15,679 DEGs were identified from the three treatment groups ( Figure 3A ). To further reveal the difference in DEGs among the three treatments, a Venn diagram was generated, and these DEGs were categorized into the seven regions ( Figure 3D ). In region 1, 8,123 genes (51.81%) were differentially expressed between shoots and roots and were not regulated by AM symbiosis or salinity treatment. In region 2, 172 genes (1.10%) were regulated by AM symbiosis but did not show tissue-preferred or salinity-responsive expression patterns. In region 3, 605 genes (3.86%) were regulated by salt stress but did not show tissue-preferred or AM-responsive expression patterns. In region 4, 1,289 genes (8.22%) showed AM-responsive and tissue-preferred expression patterns but did not respond to salinity treatment. In region 5, 3,380 genes (21.56%) showed salt-responsive and tissue-preferred expression patterns but did not respond to AM symbiosis. In region 6, 134 genes (0.85%) were regulated by both AM symbiosis and salinity but showed similar expression levels in both shoots and roots. In region 7, 1,976 genes (12.60%) were not only responsive to AM symbiosis and salt stress but also showed either shoot- or root-preferred expression patterns ( Figure 3D ).

To identify the functional categories of each region, DEGs of each region in Figure 3D were submitted to CARMO for GO enrichment analysis. Enriched GO terms with FDR < 0.05 were retained for inspection and visualization. Only region 1, region 4, region 5, and region 7 obtained enriched GO terms that were significant ( Supplementary Table S2 ). In region 1, genes showing tissue-preferred expression patterns were enriched in terms such as “thylakoid membrane organization”, “photosynthesis”, and “ATP binding” ( Supplementary Table S2 ). In region 4, genes showing AM-responsive and tissue-preferred expression patterns were enriched in terms such as “biosynthetic process”, “lipid metabolic process” and “sequence-specific DNA binding transcription factor activity” ( Supplementary Table S2 ). In region 5, genes showing salinity-responsive and tissue-preferred expression patterns were enriched in terms such as “secondary metabolic process”, “response to abiotic stimulus” and “electron carrier activity” ( Supplementary Table S2 ). In region 7, genes showing tissue-preferred expression in response to both AM symbiosis and salinity might be important for AM-enhanced salt stress tolerance. Among the 10-15 GO terms with the smallest FDR value belonging to either the cellular component (CC), biological process (BP), or molecular function (MF) category, several GO terms contained similar descriptions. For example, four and two GO terms contained “membrane” and “cell wall”, respectively. Three and two GO terms contained “transport” and “hydrolase activity”, respectively. Six GO terms were related to signal transduction, including “protein tyrosine kinase activity” and “protein serine/threonine kinase activity”. In addition, the most enriched GO term is “cytoplasmic membrane-bounded vesicle” ( Supplementary Table S2 and Figure 4 ).

Since we are interested in how AM symbiosis enhanced salt stress tolerance, a 3-way Venn diagram of the enriched GO terms from region 4 (tissue-preferred and AM-responsive), region 5 (tissue-preferred and salinity-responsive), and region 7 (tissue-preferred and AM/salinity-responsive) was generated to find specific GO terms derived from each region ( Supplementary Figure S3 and Table S3 ). The GO terms that overlapped between the three regions (g7 region in Supplementary Figure S3 ) included more general terms such as “carbohydrate metabolic process”, “catalytic activity”, “cation binding”, “hydrolase activity”, “metabolic process”, and “transport”. Stress-related terms such as “response to abiotic stimulus” and “response to stress” and signaling-related terms such as “protein kinase activity”, “protein phosphorylation”, “protein serine/threonine kinase activity” and “protein tyrosine kinase activity” were also found ( Supplementary Table S3 ). The GO terms that were specific to tissue-preferred and AM-responsive DEGs (g1 region in Supplementary Figure S3 ) included “ATPase activity”, “fatty acid biosynthetic process” and “transmembrane receptor protein serine/threonine kinase signaling pathway” ( Supplementary Table S3 ). The GO terms that were specific to tissue-preferred and salinity-responsive DEGs (g2 region in Supplementary Figure S3 ) included transport-related terms such as “amino acid transport”, “divalent metal ion transport” and “sugar transmembrane transporter activity”. Secondary metabolite-related terms such as “secondary metabolic process” and “ent-kaurene synthase activity” and photosynthesis-related terms such as “photosynthesis, light harvesting” and “photosynthesis, light reaction” were also found. Interestingly, reproduction-related terms such as “pollen−pistil interaction”, “recognition of pollen” and “stamen development” were also identified ( Supplementary Table S3 ). Finally, GO terms specific to DEGs representing tissue-preferred and AM/salinity-responsive expression patterns might be involved in AM-enhanced salt tolerance (g3 region in Supplementary Figure S3 ). In the g3 region, GO terms could be classified into several categories. One category was related to esters, such as “carboxylic ester hydrolase activity” and “hydrolase activity, acting on ester bonds”. Another category was related to reproduction, such as “pollen tube growth” and “pollination”. GO terms related to the cell wall, such as “cell wall modification”, “cellular glucan metabolic process”, “glucan biosynthetic process”, “pectinesterase activity” and “xyloglucan:xyloglucosyl transferase activity”, were also found. Intriguingly, many GO terms related to Pi homeostasis were found in the g3 region, such as “acid phosphatase activity”, “cellular phosphate ion homeostasis”, “cellular response to phosphate starvation”, “myo-inositol hexakisphosphate biosynthetic process”, “phosphate ion transport” and “positive regulation of cellular response to phosphate starvation” ( Supplementary Table S3 ). The expression pattern of genes belonging to GO terms specific to the g3 region will be closely investigated in future studies.

The expression of DEGs belonging to region 7-specific GO terms (g3 region) was visualized by heatmap. Most genes belonging to cell wall-related GO terms, such as “cell wall modification”, “cellular glucan metabolic process” and “glucan biosynthetic process”, showed higher expression in shoots than in roots ( Figure 5 and Supplementary Table S4 ), suggesting their possible critical role in the shoots. Interestingly, in the shoots, the expression patterns of these genes in response to salinity and AM symbiosis were similar. For example, pectinesterase (LOC_Os01g21034) was induced by salinity in the mock shoots (salinity (Mock_shoot) column) and upregulated by AM symbiosis in the shoots under control conditions (Ri (control_shoot) column) ( Figure 5 and Supplementary Table S4 ). These results implied that under control conditions, AM symbiosis already had a similar effect as salinity on the expression of these genes. Similar results were also observed in the expression pattern of genes belonging to “ester-related” or “oxidoreductase activity, acting on CH-OH group of donors” GO terms ( Figure 6 and Supplementary Table S5 ) and those belonging to reproduction-related GO terms ( Supplementary Figure S4 and Supplementary Table S6 ). To closely investigate the expression pattern and validate RNA-seq data, the expression profile from RNA-seq data and the corresponding RT−qPCR results for two genes were represented. For granule-bound starch synthase II (GBSSII) (Baysal et al., 2020) identified from the “glucan biosynthetic process” GO term, its expression was downregulated by AM symbiosis under control conditions and by salinity under mock conditions in both shoots and roots. Therefore, its expression was similar in mock and mycorrhizal tissue under salt stress. The RT−qPCR results were consistent with the RNA-seq findings ( Supplementary Figure S5 ). For xyloglucan endotransglucosylase/hydrolases 19 (XTH19) (Yokoyama et al., 2004) identified from the “cellular glucan metabolic process” GO term, salinity or AM symbiosis did not influence the expression in the roots. However, its expression was upregulated by AM symbiosis under control conditions and by salinity under mock conditions in the shoots. Therefore, its expression was similar in mock and mycorrhizal shoots under salt stress. The RT−qPCR results were consistent with the RNA-seq findings ( Supplementary Figure S5 ).

Several region 7-specific GO terms related to Pi homeostasis were selected for closer examination. For genes belonging to the GO terms “GO: 0016036 cellular response to phosphate starvation”, “GO:0006817 phosphate ion transport” and “GO:0003993 acid phosphatase”, most genes showed higher expression in the roots than in the shoots ( Figure 7 and Supplementary Table S7 ). Interestingly, similar to the genes mentioned above, the expression patterns of these genes in response to salinity and AM symbiosis were similar in the shoots. For example, SPX domain containing protein (LOC_Os10g25310) and Ser/Thr protein phosphatase family protein (LOC_Os11g34720) were repressed or induced by salinity in the mock shoots (salinity (Mock_shoot) column) and repressed or induced by AM symbiosis in the shoots under control conditions (Ri (control_shoot) column), respectively ( Figure 7 and Supplementary Table S7 ). These results implied that under nonsaline conditions, AM symbiosis influenced the expression of these genes in a similar way as salinity. Among these genes, the expression of low phosphate root 5 (LPR5) (Cao et al., 2016) identified from the “cellular response to phosphate starvation” GO term was downregulated by AM symbiosis under control conditions and by salinity under mock conditions in both shoots and roots. Therefore, its expression was similar in mock and mycorrhizal tissues under salt stress. The RT−qPCR results were consistent with the RNA-seq findings ( Supplementary Figure S5 ). For purple acid phosphatase 7 (PAP7) (Zhang et al., 2011) identified from the “acid phosphatase” GO term, salinity or AM symbiosis did not influence its expression in the roots. However, its expression was upregulated by AM symbiosis under control conditions and by salinity under mock conditions in the shoots. Therefore, its expression was similar in mock and mycorrhizal shoots under salt stress. The RT−qPCR results were consistent with the RNA-seq findings ( Supplementary Figure S5 ). Overall, these data implied that AM symbiosis influenced the expression of genes belonging to cell wall modification, ester-related, reproduction, oxidoreductase activity and phosphate-related pathways as salinity did, and this mediation might help plants prepare while encountering salinity stress.

To identify AM-regulated genes under salt stress, especially from the g3 region, the cutoff of log2(fold change) was set to one to obtain AM-up or downregulated genes under salt stress. AM-upregulated genes, especially in the roots under salinity, included AMP-binding enzyme, GRETCHEN HAGEN 3.2 (GH3.2), glucose-1-phosphate adenylyltransferase, inorganic phosphate transporter, multicopper oxidase and starch synthase. AM-downregulated genes, especially in the roots under salinity, included kinases such as calcium/calmodulin dependent protein kinase (CAMK) and receptor-like protein kinase; Ser/Thr protein phosphatase; and cell wall-related genes such as endoglucanase, expansin and pectinesterase ( Supplementary Figure S6 and Table S8 ). AM-upregulated genes, especially in the shoots under salinity, included cation/hydrogen exchanger 15 (CHX15); Pi-related genes, such as SPX domain-containing protein, inorganic phosphate transporter and multicopper oxidase; cell wall-related genes, such as endoglucanase, expansin and pectinesterase; inositol-3-phosphate synthase; and potassium channel Arabidopsis K⁺ transporters 2/3 (AKT2/3). Fewer genes were downregulated by AM symbiosis, especially in shoots under salt stress, and a high proportion of terpene synthase was identified from this category ( Supplementary Figure S7 and Table S8 ). Interestingly, AM symbiosis also had a similar effect as salinity on the expression of these genes under control conditions, especially in the shoots ( Supplementary Figures S6 and S7 and Table S8 ).