Six Oryza sativa cultivars and three AMF genotypes belonging to three different species were selected and their characteristics are listed in Table 1 . Two indica (IR64 and Phka Rumduol) and four japonica (Azucena, Kitaake, Nipponbare and Zhonghua 11) subspecies were selected. Seeds were obtained from IRRI and propagated at IRD except for Phka Rumduol which was provided by CIRAD.

Funneliformis mosseae FR140 (FM), Rhizophagus intraradices FR121 (RIN) and Rhizophagus irregularis DAOM 197198 (RIR) were purchased from MycAgro Lab (Technopôle Agro-Environnement, Bretenière, France) in the form of individual granular inoculums (100 spores/g).

Rice seeds were dehusked and surface-sterilised by immersion in 70% ethanol for 3 min, then in 3.8% sodium hypochlorite supplemented with 1% Tween 20 under agitation (180 rpm) for 30 min. Seeds were rinsed three times with sterile water, three times with 2% filtered sodium thiosulfate and three more times with sterile water. They were incubated overnight at 28°C in sterile water in the dark and then germinated on sterile-soaked sand for three days at 28°C. To ensure the absence of contaminants, 100 µL of the last rinse water and imbibition water were plated on tryptic soy agar (Sigma-Aldrich) Petri dishes. Four homogeneous rice seedlings were then transferred to anti-coiling pots (Comptoir Vert, France) filled with 150 mL of clay beads (6-18 mm, Terres & Traditions, France) and 450 mL of sterile inert substrate composed of 70% of sand, 20% sieved perlite and 10% of vermiculite (Campo & San Segundo, 2020). This substrate was inoculated with either the AMF granular inoculum, or the granular inoculum without fungal spores (control) at a volume of 5% per pot.

Rice plants were grown for 2.5 months in a growth chamber (12 h day/night, 28°C day, 26°C night, 75% humidity). The substrate was moistened for one week and then watered three times a week with a Hoagland solution (Hoagland and Arnon, 1938) reduced in phosphate (2.5 mM Ca(NO₃)₂, 2.5 mM KNO₃, 1 mM MgSO₄, 0.25 mM (NH₄)₂SO₄, 25 µM KH₂PO₄ and trace elements, complete recipe in Supplementary Table 1 ), with the watering volume gradually increased according to the cultivar growth.

Maximum height, shoot and root fresh weights were measured for each plant (n = 20 per condition). Shoot dry weight was also measured after drying at 40°C for one week at 48 h (n = 20 par condition). Roots were stored in 70% ethanol at 4°C until mycorrhizal quantification (n= 5 pools of 4 root systems).

The root systems of four plants from the same pot were washed in tap water, placed in 70% ethanol and stored at 4°C until analysis. Fungal structures were stained using a blue ink-based protocol modified from Cao et al., 2013. Roots were heated to 80°C for 45 min in 10% KOH. They were rinsed three times with ultrapure water (MilliQ) and stained with a staining solution consisting of 5% blue ink (Waterman “Bleu Sérénité”) in 5% acetic acid at room temperature for 10 min. They were then rinsed three times with ultrapure water and fixed in 5% acetic acid overnight.

Five replicates of 20 to 25 fragments from coronary roots were mounted between slide and glass and the mycorrhizal index (global mycorrhization and arbuscular intensity) were assessed as in Trouvelot et al., 1986, on a Axiozoom Zeiss microscope.

Infection of Xanthomonas oryzae pv. oryzae PXO99 (Xoo) was carried out on 50-days old rice plants by leaf clipping as described in Niño-Liu et al., 2005. The extent of chlorosis and necrosis was assessed 14 days post inoculation (dpi) on each leaf (n = 18). Mock inoculations were made with sterile water to assess that the sole clipping of the leaf does not induce any disease symptoms.

RNA was extracted from leaf samples collected during leaf-clipping (before infection, at 50 days post germination). They were ground to a fine powder using a TissueLyser II (Retsch) at 30 Hz, for 15 s twice. Each biological replicate consisted of two (Nipponbare) to three (IR64) leaf samples from the same pot, with four biological replicates for each condition. Total RNA was extracted using TriReagent (Sigma) and a DNAse (QIAgen) treatment was added in the protocol before purification with the RNA Clean & Concentrator kit (Zymo), according to the manufacturer’s instructions. The quantity and quality of total RNA was assessed using a NanoDrop 1000 spectrophotometer (ThermoFisher). Approximately 360 ng of total RNA from each biological replicate was used for retrotranscription into cDNA using SuperScript III Reverse Transcriptase (Thermo Fisher Scientific). cDNAs were diluted 5-fold and RT-qPCRs were performed using the Takyon™ Low ROX SYBR 2X MasterMix blue dTTP (Eurogentec) on a LightCycler 480 qPCR system (Roche). Plate preparation was automated using the epMotion 5070 pipetting robot (Eppendorf). Four independent biological replicates were analysed for each condition, each one analysed in triplicate. Relative gene expression is calculated by comparing each sample to the standard’s (number of cycles of EF1a, Δ C t ), and then to the control group (FM, RIN & RIR vs. CT, Δ Δ C t ). The fold change is calculated with 2 − m e a n Δ Δ C t and the logFC represents the relative expression of each marker gene as indicated in (Pfaffl, 2001). The list of marker genes, their function and the primers used in this study are listed in Table 2 .

Statistical analyses were performed with Rstudio (version 2022.2.0.443) and R (version 4.1.3) softwares using the packages “readxl”, “tidyverse”, “ggplot2”, “rstatix”, “ggpubr”, “multicompView”, “car”, “pcr”, “pheatmap”. As phenotypic data of mycorrhization and arbuscular intensity indexes fulfilled normality (Shapiro test, p>0.05) and homoscedasticity (Levene test, p>0.05) hypotheses, a two-way ANOVA was used to test the significance of rice cultivar and AMF genotype effects on these parameters, and one-way ANOVA with Tukey test were used for pairwise group comparisons. For rice growth and biocontrol traits, these data did not fulfil the normality and homoscedasticity hypotheses, thus a non-parametric ANOVA with Kruskal-Wallis test (α=0.05) followed by pairwise comparisons with Wilcoxon tests (α=0.05) were used for mean group comparisons of measured phenotypic traits.

For the analysis of the expression of rice marker genes, the “pcr” package was used. Linear regression was used to assess statistical differences between AMF inoculation and marker gene expression. To visualise the relative expression of each marker gene as a function of rice cultivars and AMF inoculation, a heatmap of LogFC was made with the “pheatmap” package.

To assess the symbiotic compatibility between the three AMF genotypes (RIN, RIR, FM) and the six rice cultivars (listed in Table 1 ), we analysed the mycorrhization rates in the 18 combinations. We focused on the global fungal colonisation rate and the percentage of visible arbuscules in the mycorrhizal roots (see Material & Methods). The use of mycorrhizal index of global mycorrhization and arbuscular intensity allowed us to quantify the interaction between AMF and rice. The mean of each mycorrhizal index for each combination are presented in Figure 1 and in Supplementary Table 2 .

We tested whether rice cultivar and AMF genotype have an impact on rice’s global mycorrhization and arbuscular intensity index, with a two-way ANOVA test. Both indexes are statistically significantly affected by either rice cultivar or AMF genotype (Two-Way ANOVA, F = 66.91593, p< 2.10^-6; F = 4.9315, p< 0.01).

All the rice cultivars tested were root colonised by each AMF genotype ( Figure 1 ; Supplementary Table 2 ). Each fungal organ (hyphal structures, spores, vesicles and arbuscules) was clearly visible on each combination as shown in Figure 2 and Supplementary Figure 1 . The global intensity of mycorrhization ranged from 27.20% (Phka Rumduol with RIN) to 83.90% (Nipponbare with RIR). Independently of the fungal inoculation, indica rice varieties have the less intense symbiotic percentage, ranging from 27.20% (Phka Rumduol - RIN) to 46.8% (IR64 - RIN) ( Supplementary Table 2 ). The percentage of mycorrhization of the japonica cultivars ranges from 43.20% (Kitaake with FM) to 83.90% (Nipponbare with RIR). Nipponbare is the most intensely mycorrhized cultivar with 79%, 77.90% and 83.90% for FM, RIN and RIR inoculation, respectively ( Supplementary Table 2 ).

The arbuscular percentage of the mycorrhizal roots ranged from 3.22% (Azucena with FM) to 49.40% (Nipponbare with RIR) ( Supplementary Table 2 ). Japonica rice cultivars showed the highest percentage of visible arbuscules in the mycorrhized system compared to indica rice, with RIN and ( Figure 1 ). On the other hand, each rice genotype in interaction with FM formed almost no arbuscules, with a maximum of 5.24% in Zhonghua 11 ( Supplementary Table 2 ).

The phenotypic response of rice to AMF inoculation and symbiosis establishment was assessed by growth measurements. Maximum height, fresh and dry shoot weight and fresh root weight were measured for each combination (n = 20) and are shown as boxplots in Figure 3 . All the corresponding measured values are listed in Supplementary Table 2 . Developmental stage of each cultivar (at 10 weeks) is shown in Supplementary Figure 2 . Plants were still at developmental vegetative stage, except for Kitaake plants which flowered one week before harvest.

The dataset shows that AMF inoculation can result in an increase, decrease or no significant effect on plant growth parameters. We observed a significant decrease in the height of IR64 during the interaction with FM or RIN (-18.81% and -14.12% respectively, Figure 3A and Supplementary Table 2 ). The RIR genotype resulted in a significant increase in both root and shoot weights for Phka Rumduol, Kitaake, Zhonghua 11 and Nipponbare. However, this effect was not statistically significant for Azucena. The effect on rice’s dry weight wasn’t as significant as on height or fresh weight, but was still a good proxy for the beneficial effect of AMF on rice growth ( Figure 3 ).

Under our growth conditions, we observed different growth rates depending on the rice cultivar. Uninoculated Kitaake was the smallest rice cultivar both in size (24.65 cm) and weight (0.19 g, 0.11 g, 0.09 g on average for fresh shoot, root and dry shoot weight, respectively; Supplementary Table 2 ). Still, mycorrhization of Kitaake induced a clear improvement in growth: the highest dry weight among all combinations being obtained with Kitaake in association with RIN (0.30 g, Figure 3D ; Supplementary Table 2 ). This combination showed the greatest positive effect on rice’s growth on all variables: 36% taller, 259%, 270% and 221% heavier on biomass of fresh shoots and roots, and dry shoots, respectively ( Supplementary Table 2 ).

Globally, FM inoculation affected rice growth non-significantly (Nipponbare, Phka Rumduol, Zhonghua 11) or negatively (Azucena, IR64), in both height and weight ( Figure 3 ). The only significant positive interaction was with Kitaake: 20% taller, 163%, 177%, 125% heavier on its fresh shoot & root and dry shoot weights, respectively ( Figure 3 ; Supplementary Table 2 ). The effect of RIN inoculation was contrasting: beneficial on Kitaake and Zhonghua 11, negative on IR64 and Phka Rumduol or non-significant on Azucena and Nipponbare ( Figure 3 ). RIR was the AMF genotype that induce the most positive effects among all rice cultivars: +35% and +20% leaf height with Kitaake and Phka Rumduol, respectively ( Figure 3A ); +182% and 103% fresh shoot weight with Kitaake and Zhonghua 11, respectively ( Figure 3B ) and +196% and +149% fresh root weight with Kitaake and Phka Rumduol, respectively ( Figure 3C ).

Our results show that the effect of AMF inoculation on rice growth depends on both rice cultivars and AMF genotypes: ranging from negative, neutral to beneficial outcomes across the 18 combinations under study.

The potential of each fungal inoculum to induce systemic resistance in the leaves of each rice cultivar during a shoot phytopathogen infection was investigated. Rice plants were infected by leaf-clipping with Xanthomonas oryzae pv oryzae PXO99 (Xoo) and the extent of chlorosis and necrosis was recorded 14 days later. The results are shown as boxplots in Figure 4 and all the corresponding measured values are listed in Supplementary Table 2 .

The effect of AMF inoculation on the chlorosis and necrosis symptoms of rice induced by Xoo differed greatly between combinations. Only two combinations, both with RIN, showed a significant increase in leaf symptoms: Azucena on chlorosis (+69%) and Phka Rumduol on necrosis (+106%). Regarding the bio-protective effects of AMF, chlorosis symptoms were significantly reduced on IR64 in combination with FM (-24%) and RIR (-26%), on Zhonghua 11 in combination with FM (-44%), RIN (-28%) and RIR (-29%), and on Nipponbare with FM (-40%) and RIN (-34%) ( Figure 4 and Supplementary Table 2 ). For necrosis, only Zhonghua 11 and Nipponbare showed significant reductions of symptoms. These reductions in the size of necrosis were observed with the three AMF genotypes: -65%, -66% and -64% for Zhonghua 11 with FM, RIN and RIR respectively; and -78%, -87% and -64% for Nipponbare with FM, RIN and RIR, respectively ( Figure 4 and Supplementary Table 2 ).

We observed contrasted patterns of symbiotic compatibility among our AMF-cultivar combinations. In order to link these observed differences with the expression level of leaf marker genes, we selected two rice cultivars with contrasting AMF responses: Nipponbare and IR64. The first is a japonica model cultivar and was the most intensely mycorrhized, regardless of the AMF genotype, with the interaction having non-significant to beneficial effects on its growth and tolerance to Xoo infection ( Figures 1 , 3 , 4 ). The latter is an indica model cultivar, that was significantly less mycorrhized, with non-significant to negative effects of the AMF interaction on its growth, but with beneficial effects on its tolerance to Xoo infection ( Figures 1 , 3 , 4 ). We selected 19 markers genes of development, nutrient homeostasis, hormonal balances and defence and their expression was normalised to that of EF1a reference gene. The list of marker genes, their function and the primers used in this study are listed in Table 2 . A summary of the statistical comparison of gene expression for each combination, for both Nipponbare and IR64, is provided in Supplementary Table 3 and Supplementary Figures 3 , 4 . Their expression was visualised as a heatmap in Figure 5 .

The expression of two cellular growth marker genes, OsYABBY6, responsible for abaxial-adaxial polarity and whose expression is needed for leaf development (Jiang et al., 2015), and OsXTH17, a xyloglucan endotransglucosylase/hydrolases involved in primary cell wall formation (Lin et al., 2019) was recorded. A non-significant induction of OsXTH17 expression was observed for Nipponbare in interaction with RIN (p= 0.055), while that of OsYABBY6 is repressed with FM (p=0.009) ( Figure 5 ; Supplementary Table 3 ; Supplementary Figure 3 ). In IR64, their expression was not significantly affected independently of the AMF genotype ( Supplementary Figure 4 ).

To assess the effect of mycorrhization on mineral homeostasis in leaves, one iron transporter (OsIRO2), one nitrate-reductase (OsNIA1) and three Pi transporters, marker genes of Pi-starvation response (OsMGD2, OsPAP23 and OsSPX3) were selected. The expression of almost all mineral marker genes was significantly reduced in Nipponbare leaves (OsNIA1, OsMGD2, OsPAP23 with RIN, OsSPX3 with either AMF genotype), except for a non-significant strong induction of OsIRO2 expression when associated with RIN (p= 0,09) ( Figure 5 ; Supplementary Table 3 ). In leaves of IR64, the expression of OsMGD2 and OsPAP23 was not significantly affected. The expression of the other mineral marker genes was reduced only significantly for OsSPX3 in RIN-mycorrhized leaves, and for OsNIA1 and OsIRO2 in FM-mycorrhized leaves ( Figure 5 ; Supplementary Table 3 ; Supplementary Figures 3 , 4 ).

Mycorrhization is known to affect the hormonal balance in mycorrhized plants. Its effect on the expression of jasmonate (JA: OsLOX4, a lipoxygenase responsible for the biosynthesis of JA (Nguyen et al., 2022) and OsJAMyb & OsJAZ6, both responsible for JA signalling (Yokotani et al., 2013; Lu et al., 2016), ethylene (ET: OsACS1, 1-aminocyclopropane-1-carboxylate synthase responsible for ethylene biosynthesis) and salicylic acid (SA: OsNPR1, mediating SA biosynthesis and responsives genes (Kumari et al., 2016) & OsWRKY45, a transcription factor mediating SA signalling) pathways was investigated. Overall, jasmonate- and ethylene-related genes expression was not significantly repressed in Nipponbare and IR64 leaves ( Figure 5 ; Supplementary Table 3 ). SA-related genes were not significantly repressed, except for OsNPR1 in RIN mycorrhized-IR64 leaves ( Figure 5 ; Supplementary Table 3 ; Supplementary Figures 3 , 4 ).

The expression of defence-related genes was recorded to assess how mycorrhization affects the defence response in healthy leaves. OsPR5 is a pathogenesis-related protein, OsPAL4 is a broad-spectrum disease resistance-related gene and OsTGAP1 & OsDXS3 are responsible for phytoalexines production in rice (Okada et al., 2009; Delteil et al., 2012; Petitot et al., 2017; Valette et al., 2020). OsMPK10 and OsWRKY30 are responsible for early disease-mediated signalling, the latter also responsive to SA and JA treatments (Ryu et al., 2006; Nguyễn et al., 2014). Globally, defence genes appeared to be more induced in mycorrhized IR64 leaves than in Nipponbare leaves ( Figure 5 ; Supplementary Figures 3 , 4 ). In Nipponbare leaves, we observed i) a significant repression of OsPR5 expression, irrespective of the AMF species, ii) a significant down-regulation of OsPAL4 expression with FM, iii) a non-significant down-regulation with RIN ( Supplementary Table 3 ). In leaves of IR64, phytoalexin biosynthesis-related genes (OsDXS3 and OsTGAP1) were not significantly induced in plant associated with AMF. When mycorrhized with RIR, the expression of OsWRKY30 is significantly induced but not the one of OsMPK10 (p-value = 0.03 and 0.07, respectively) ( Figure 5 ; Supplementary Table 3 ).

In this study we have characterised the symbiotic compatibility between multiple rice cultivars from japonica and indica subspecies in association with three AMF genotypes (F. mosseae FR140, R. intraradices FR121 and R. irregularis DAOM197198). First, we observed that all rice cultivars selected for this study were colonised by each fungal inoculum, their colonisation intensity differing between rice cultivars (Phka Rumduol being the lowest and Nipponbare the highest in terms of global mycorrhization). A similar pattern was also observed at the arbuscular level, with FM developing fewer arbuscules when interacting with rice, compared to the two Rhizophagus genotypes. We also observed an effect of AMF colonisation on rice growth that could be either beneficial, neutral, or negative. However, there was no direct relationship between the level of colonisation and the growth phenotype: both indica rice cultivars had similar AMF colonisation and arbuscular levels, but the tendency of IR64 to be negatively affected on growth wasn’t observed for Phka Rumduol ( Figures 1 , 3 ). The japonica cultivars used in this study were generally beneficially affected in their growth (either in height or weight; Figure 3 ). A special case was observed with the cultivar Kitaake. As stated earlier, early mortality and a general lack of growth were observed for this cultivar in our inert substrate. We hypothesised that Kitaake’s growth wasn’t optimal under these drastic conditions. However, plants that have managed to grow were well colonised by the three AMF (43%, 69% and 71% of roots colonised by FM, RIN, and RIR respectively). The highest levels of growth improvement were recorded for this cultivar: 37% taller, 259%, 270% and 221% heavier on biomass of fresh shoots and roots and dry shoots ( Supplementary Table 2 ). AMF symbiosis is known to enhance root anchoring and nutrient uptake in poor soils, thereby improving plant health (Paszkowski and Boller, 2002; Gutjahr et al., 2009; Chiu et al., 2018). However, early growth depression following mycorrhization has also been documented in wheat, barley, and soybean (Jacott et al., 2017). Such depression was eventually overcome (or not) in the later life cycle of the crop (Jacott et al., 2017). This growth depression can be explained by the genetic variability of AMF genotypes and their symbiotic effectiveness, but the recurring hypothesis is related to the trade-off between plant photosynthates and soil nutrients recovered by AMF (Jin et al., 2017). Our differences in the phenotypic growth effects of mycorrhization among rice cultivars could be explained by an imbalance between early life stage development and the photosynthetic carbon cost for AMF establishment and function (Jacott et al., 2017; Jin et al., 2017). Our growth conditions were adapted from recent studies on wheat and rice mycorrhization, composed of a mixture of sand, sieved perlite, and vermiculite. It mimics sandy soil and is easy to sterilise. Though, this inert substrate is still poor in essential nutrients, and a modified Hoagland’s solution depleted in phosphate was used to water the plants. Several studies have shown that similar conditions allow both crop species to grow and the establishment of an efficient AMF symbiosis (Gutjahr et al., 2008; Fiorilli et al., 2018; Campo and San Segundo, 2020; Campo et al., 2020; Guo et al., 2022).

Under our growing conditions, japonica rice appeared to be more intensely colonised than indica rice. ( Figure 1 ; Supplementary Table 2 ). The result of upland japonica rice cultivars being more colonised and responding more to AMF colonisation than flooded indica rice cultivars is shared by several studies, considering their different root architecture and the co-evolution between AMF and upland rice in aerobic selective condition, optimizing their interaction (Diedhiou et al., 2016; Davidson et al., 2019). It is important to note that we selected two indica rice cultivars compared to four japonica, depending on the valuable factors listed in Table 1 . This study should be extended to more rice genotypes from both subspecies to understand if this pattern can be generalised. The global understanding of how rice mycorrhization is affected by host genotype is currently limited. In a published report, the mycorrhizal growth response (MGR, corresponding to the ratio between the dry weight of mycorrhized and control plants) was measured for 64 rice genotypes from different subspecies 4 weeks after inoculation with F. mosseae. Differences between genotypes were observed but not related to indica or japonica origin (Suzuki et al., 2015). A possible explanation is that variations in AMF recognition receptors between host genotypes affect their symbiotic compatibility. OsCERK1, a LysM receptor-like kinase that is essential for AMF recognition and activation of the symbiosis pathway, is also responsible for root branching in rice (Chiu et al., 2018; Choi et al., 2018). It has been proposed that natural variation between allelic variants of the OsCERK1 gene in different rice cultivars may affect their symbiotic compatibility with R. irregularis DAOM 197198 (Huang et al., 2020). A higher level of AMF colonisation, 14 days post inoculation, was reported for eleven indica rice cultivars, compared to eight japonica. The difference was proposed to be related to an underrepresented OsCERK1 haplotype, absent in japonica rice and present in their selected indica cultivars (Lefebvre, 2020). In our study, we obtained opposite results, but we chose to harvest the rice after two months of growth, which, in addition to the different substrate used, may explain the difference. Several reports have shown that root colonisation of Nipponbare becomes clearly visible after 14 days post inoculation, and arbuscules after 3 to 4 weeks (Gutjahr et al., 2015; Guo et al., 2022; Liu et al., 2022). It would be interesting to assess whether rice cultivar also affects the kinetics of AMF establishment and functioning in the long term.

As plant colonisation by AMF has often been shown to be associated with a better protection against pathogens, we conducted biocontrol trials against Xoo. We observed a general tendency for symptoms to decrease, which was statistically significant for some AMF-rice cultivar combinations ( Figure 4 ; Supplementary Table 2 ). Only two among 18 showed an increase in the leaf symptoms when associated with an AMF (Phka Rumduol and Azucena associated with RIN). Almost all well-colonised japonica rice species showed a significant reduction of at least one Xoo-induced symptoms thanks to MIR, particularly with Zhonghua 11 and Nipponbare cultivars. The most-colonised rice cultivar, Nipponbare, see chlorosis and necrosis symptoms reduced in both FM and RIN conditions. With the most compatible one, this reduction of symptoms can be noticed but is not statistically significant. This reduction of symptoms becomes statistically significant in a repeated study on 20 non-mycorrhizal Nipponbare and 20 associated with RIR ( Additionnal file 6 : Supplementary Figure 3 ). An induction of 39%, 107% and 187% of Nipponbare’s maximum height, shoot and root dry weight, respectively was shown, linked with a reduction of chlorosis symptoms by 32% at 14 days post-clipping. It is noteworthy that IR64, a rice variety three times less intensively mycorrhized, also shows reduced chlorosis symptoms when mycorrhized with FM or RIR. A possible explanation could reside in the differences between rice genotypes themselves, being more or less sensitive to AMF colonisation and their established interaction, having then important or little to no effects on rice growth phenotype, but notable impacts on rice nutrient physiology and defence responses. Our results highlight here that symbiotic compatibility between AMF and rice species states for a sufficient amount of colonisation allowing significant nutrient starvation response inhibition and phytopathogen-induced symptoms reduction.

The potential of biological control by AMF symbiosis is well documented in the literature, although pathogen-dependent. Mycorrhization of japonica rice cultivars by AMF genotypes such as R. intraradices, R. irregularis or F. mosseae showed both a local and a systemic defence response against Pyricularia oryzae, sometimes associated with an increase in shoot biomass (Campos-Soriano et al., 2010 ; Campos-Soriano et al., 2012; Campo et al., 2020). In Triticum aestivum cv. Chinese Spring associated with F. mosseae, the AMF symbiosis confers both a positive effect on growth and resistance to X. translucens (Fiorilli et al., 2018). Two American rice cultivars inoculated with a mixture of AMF genotypes showed increased susceptibility to two insects and Rhizoctonia solani infections, but without growth defects or nutrient losses, suggesting an effect of symbiosis on defence vs growth trade-off (Bernaola et al., 2018b).

Many research studies focus on the association between R. irregularis and Nipponbare (Gutjahr et al., 2009; Gutjahr et al., 2015; Campo et al., 2020; Guo et al., 2022; Liu et al., 2022). Since they are both models (sequenced genomes, easy to grow, transform and store, controlled lifecycle), deepening our knowledge of their interaction at different life stages of both organisms may shed light on how rice interacts with and benefits from AMF and vice-versa. Under our conditions, Nipponbare was the rice cultivar with the best level of mycorrhization, regardless of the AMF inoculated ( Figure 1 ; Supplementary Table 2 ). There is no significant negative effect of mycorrhization on its growth, with a tendency of AMF symbiosis to increase both height and weight ( Figure 3 ). After Xoo infection, a significant reduction of symptoms was observed with FM and RIN ( Figure 4 ). To further confirm this, the experiment was repeated (n = 20) for Nipponbare in combination with R. irregularis DAOM197198 and showed then a clear positive effect on growth promotion (height, root and leaf weight) and on biocontrol against Xoo ( Supplementary Figure 5 ). At the molecular level, RT-qPCR analysis showed that the expression of cellular development marker genes was induced by mycorrhization, even after two months of growth ( Figure 5 ; Supplementary Table 3 ). Under these conditions, the overall expression of nutrient transporter and defence genes was down-regulated. As these genes are known to be induced upon starvation, Nipponbare in symbiosis with AMF could then be considered to be in a healthy state.

Our results are in agreement with previous work. On Nipponbare, Glomus intraradices, which could be classified as R. irregularis according to the current AMF phylogeny, colonises more than 60% of the large lateral roots and increases both dry weight and coronary root length (Gutjahr et al., 2009). Colonisation kinetics of R. irregularis on this cultivar showed high colonisation rates (up to 87% of root length, of which 59% contained arbuscules) and the presence of viable arbuscules and vesicles up to six weeks after inoculation (Gutjahr et al., 2015; Liu et al., 2022). Here we show that even after ten weeks, arbuscules and vesicles were still clearly visible and in larger numbers. A study on the effect of pre-inoculation of F. mosseae on Nipponbare before transplanting in the field showed a root colonisation rate of 65%, an arbuscule content of 1.3% and a vesicle content of 6% (Sisaphaithong et al., 2017), consistent with our results. Growing conditions and practices may have an impact on this low arbuscule content. After 10 weeks of association with F. mosseae, eight temperate japonica cultivars used in Spanish or Italian fields had between 20 and 60% of arbuscules in their mycorrhized roots, with non-significant to beneficial effects on both rice growth and P. oryzae tolerance (Campo et al., 2020). The geographical origin and genotype of the host may determine its symbiotic compatibility with different AMF genotypes.

The description of the phenotypic responses (rice growth and tolerance to Xoo) of different rice cultivars to inoculation with each AMF genotype allowed us to assess the symbiotic compatibility between them. To understand how AMF symbiosis affects the molecular functioning in two rice cultivars with contrasting phenotypic responses to AMF, the leaf molecular responses of Nipponbare and IR64 after six weeks of interaction were studied. Local molecular responses to the establishment of the mycorrhization have been well studied in rice roots (Güimil et al., 2005; Gutjahr et al., 2009; Campos-Soriano et al., 2010, reviewed in Choi et al., 2018), but the systemic effects on the responses of rice leaves are still poorly described. A panel of rice marker genes involved in development, nutrient status, phytohormone signalling and defence responses (listed in Table 2 , detailed in the results) was selected, and their expression analysed in healthy 50-day old leaves of Nipponbare and IR64.

First of all, the response of Nipponbare leaves to root mycorrhization was consistent with our phenotypic results. We observed a modulation of developmental gene expression, coupled with a reduced starvation response (repression of Pi starvation’s marker genes and a nitrate reductase marker gene, Figure 5 ; Supplementary Table 3 ). Defence related genes expression was not significantly to negatively affected by AMF symbiosis in Nipponbare, suggesting that mycorrhization does not induce a systemic defence response under healthy conditions. Overall, mycorrhization of the japonica rice Nipponbare highlights an improvement of rice’s development and nutritional status, to be linked with the reported increase in growth in this most intensely mycorrhized rice. This imprint on rice’s expression pattern hints for a better overall state and tolerance against abiotic as well as biotic stressors. The study of the effect of either one of them on mycorrhized Nipponbare is interesting to assess which specific systemic mechanisms will act on the trade-off between growth and tolerance.

Mycorrhized IR64 plants showed a different response to AMF symbiosis. A non-significant repression of developmental and nutrient starvation response marker genes was observed, coupled with a non-significant induction of the defence response ( Figure 5 ; Supplementary Table 3 ). Under our growth conditions, mycorrhization of IR64 does not seem to be as beneficial as of Nipponbare and this is still to be linked with phenotypic results (i.e. a non-significant to negative impact of mycorrhization on IR64’s growth with a less mycorrhized cultivar).

The nutritional status of both our rice cultivar were investigated thanks to iron, phosphate and nitrate-related genes expression. There is a strong trend of induction of OsIRO2 expression in leaves of Nipponbare interacting with RIN ( Figure 5 ; Supplementary Table 3 ; Supplementary Figure 3 ). This transcription factor is induced under Fe deficiency, modulating key genes involved in iron uptake in rice (OsNAS1, OsTOM1 or OsYSL15), with IRO2-overexpressing rice showing improved Fe-deficiency tolerance compared to the non-transgenic lines (Ogo et al., 2006; Ogo et al., 2011). Mycorrhization may affect iron homeostasis, via its effect on IRO2 expression, affecting Fe uptake and translocation to shoots and grains (Ogo et al., 2011) but little research has been done on this subject. Leaves of the japonica rice Senia show an increase of IRO2 expression during mycorrhization with RIR, while wheat symbiosis with FM triggers an accumulation of Fe-uptake related proteins in roots, linked with a translocation of iron from roots to shoots, suggesting both a beneficial effect of mycorrhization on iron homeostasis in mycorrhized hosts (Campos-Soriano et al., 2012; Fiorilli et al., 2018).

AMF have been demonstrated to increase the bioavailability of essential nutrients such as nitrogen and phosphate to their host roots. This capacity can be linked with the repression of OsNIA1 and Pi-starvation-related genes expression in the leaves of our two studied rice cultivars. ( Figure 5 ; Supplementary Table 3 ; Supplementary Figures 3 , 4 ). These results are in line with previous reports, and specifically results showing the repression of OsSPX3, OsPAP23 and OsMGD2 expression in Loto rice leaves in association with FM (Liu et al., 2007; Gutjahr et al., 2008; Fiorilli et al., 2018; Campo and San Segundo, 2020; Vannini et al., 2021).

The expression of marker genes for phytohormones biosynthesis or signalling in leaves was not significantly affected by AMF symbiosis, except for a repression of the SA perception gene (OsNPR1) in leaves of IR64 ( Figure 5 ; Supplementary Table 3 ; Supplementary Figure 4 ). Recent literature reports contradictory results on the effect of mycorrhization on SA pathways in wheat or rice: the first is not affected on any SA-related pathways during FM mycorrhization, while OsNPR1 is induced in some japonica rice cultivars (Campos-Soriano et al., 2012; Fiorilli et al., 2018; Tian et al., 2019). Mycorrhization effect on SA-related pathways may be plant species-dependent (Campo and San Segundo, 2020).

Under our conditions, ethylene and jasmonate related genes were not significantly repressed in both cultivars ( Figure 5 ; Supplementary Table 3 ). Ethylene biosynthesis is known to be induced in tomato and wheat and repressed in rice leaves under mycorrhization, not significantly in our case (Fiorilli et al., 2018; Campo and San Segundo, 2020). Jasmonate-related genes are previously reported to be modulated by mycorrhization in different rice cultivars, both on its biosynthesis or signalling (Campos-Soriano et al., 2012; Campo and San Segundo, 2020). MIR is reported to occur via JA-related pathways, leading to reduced symptoms and pathogen load in multiple host-pathogen interactions (Dowarah et al., 2021). The molecular responses to Xoo infection may allow us to understand if the reduction of symptoms occurring in both our rice cultivars are linked with MIR, a primed induction of defence responses, responding faster and more efficiently than non-mycorrhized controls, or by an overall better rice health state (related to both growth and nutritional status improvement). If the MIR hypothesis appears to be true, deeper transcriptional analyses are to be conducted to understand by which mechanisms MIR occurs.

Overall, the systemic response of rice to AMF symbiosis is dependent on rice cultivar and AMF genotype but can be linked to an overall improvement in rice health.