Glycine max (L.) Merr. cv. Bragg wild type (wt), nts1007 (hyper nodulating nark-mutant) and nod139 (non-nodulating nfr5-mutant) (Carroll et al., 1985, 1986) were used in this work. Seeds were surface-disinfected for 10 min in sodium hypochlorite 5% (v/v) and washed successively with distilled water. The seeds were germinated on filter paper moistened with distilled water in a chamber at 28°C in the dark for 3 days. Then, seedlings were placed in aerated plastic trays with B&D solution (Broughton and Dilworth, 1971) supplemented with 0.625 mM KNO₃ and 0.313 mM Ca(NO₃)₂ (1,25 mM of total nitrogen concentration). The B&D solution was replaced every 7 days. Soybean plants were grown in a growth chamber under 16 h photoperiod (250 μmol m^-2 s^-1) at 26 ± 2°C.

Bradyrhizobium japonicum USDA 138 strain was cultured in yeast extract mannitol (YEM) medium (Vincent, 1971) at 28°C with constant agitation (180 rpm) for 5 days. For the treatments, the bacteria were washed and resuspended in sterile water (OD₆₀₀ = 0.8).

The root of 12-days-old soybean plants were inoculated with 2.5 mL of washed B. japonicum USDA 138 strain (OD₆₀₀ = 0.8) per liter of hydroponic medium. The first trifoliate leaf was collected at different times post-inoculation and immediately frozen or used for measurement.

For the DPI treatments, an NADPH-oxidase inhibitor (50 μM), was added 30 min before inoculating B. japonicum to either the root or leaf according to the experiment. The inhibitor was added to the hydroponic medium for the root application or sprinkled locally in the first trifoliate leaf. DPI application does not affect the normal growth of B. japonicum (Muñoz et al., 2012).

For the photooxidative stress treatment, the first trifoliate leaf of 12-days-old inoculated and non-inoculated (control) plants were treated with 1,1′-dimethyl-4, 4′-bipyridinium dichloride (PQ). PQ is a viologen that acts at the photosystem I level by intercepting the electrons that are going from ferredoxin to NADP+, thus enhancing chloroplastic ROS generation and inhibiting ascorbate regeneration (Asada, 1999). The leaves were splashed uniformly with 20 μM of PQ and 0.01% Tween 20 at 24 and 48 hpi and physiological parameters were measured after 24 h of PQ treatments. PQ was replaced by water in control treatments.

Purified Nod factors and chitosan treatments were performed in the same way than B. japonicum inoculation, in the hydroponic medium. Chitosan is a fungal elicitor.

Superoxide levels were determined histochemically with nitroblue tetrazolium (NBT) staining, which reacts with superoxide radicals to produce a blue formazan precipitate. Leaves were incubated in 0.01% (w/v) NBT solution in 25 mM K-Hepes buffer (pH 7.6) in the darkness at 28°C for 2 h. Color images were transformed in invert 8-bit images and gray value intensity of the blue stain was quantified using the ImageJ software (Schneider et al., 2012).

H₂O₂ generation was histochemically determined with 3,3′-diaminobenzidine (DAB) staining. The leaves were incubated in 0.02% (w/v) DAB solution in 50 mM Tris acetate buffer (pH 5) and incubated in the darkness at 28°C for 2 h. Color images were transformed into invert 8-bit images and gray value intensity of the DAB stain was quantified using Optimas 6 (Optimas Corporation, Bothell, WA, United States). Total optical density (OD) was calculated as log inverse gray value of pixels within an area boundary relative to the analyzed area.

Hydrogen peroxide was also estimated in leaf extracts as previously described by Guilbault et al. (1968). A blank with CAT was included for each sample. Frozen leaf samples were ground to a fine powder with liquid nitrogen and homogenized 1/10 (w/v) in 50 mM potassium phosphate buffer (pH 7.5), containing 1 mM EDTA and 1% polyvinylpolypyrrolidone (PVPP). Homogenates were centrifuged at 16,000 × g at 4°C for 25 min and the supernatant was used to determine protein and hydrogen peroxide concentrations.

The glutathione and ascorbate content in soybean leaves were determined as previously described by our group (Rodríguez et al., 2010). Leaf samples were homogenized (100 mg of fresh weight material) in 1 mL of cold 3% trichloroacetic acid (TCA) and 100 mg PVPP. The homogenate was centrifuged at 10,000 × g at 4°C for 15 min and the supernatant was collected to for analyze of glutathione and ascorbate content.

Superoxide dismutase activity was determined spectrophotometrically at 560 nm, as previously described by Casano et al. (1997). GR activity was assayed following the decrease at A340 nm because of NADPH oxidation, according to Schaedle and Bassham (1977). CAT activity was determined by measuring the decrease at 240 nm because of H₂O₂ degradation (Gallego et al., 1996). APX activity was measured according to Nakano and Asada (1981) by measuring the H₂O₂-dependent oxidation of ascorbate at 290 nm (the extract medium for this enzyme contained 5 mM ascorbate).

Soluble proteins were estimated according to Bradford (1976). Bovine serum albumin was used as a standard for calibration.

The Chlorophyll content of the first trifoliate leaf was estimated using a handheld SPAD CL01 meter (Hansatech Instruments, Pentney King’s Lynn, United Kingdom). The results are expressed as unitless parameter (SPAD units, 0 to 100), which is proportional to leaf chlorophyll.

Quantum efficiency of PSII photochemistry under ambient light conditions (250 μmol photon m^-2 s^-1, 25 ± 2°C) (ΦPSII) was measured using a pulse amplitude modulated fluorometer (FMS2, Hansatech Instruments, Pentney King’s Lynn, United Kingdom). F_v/F_m ratio was calculated using (F_m -F₀)/F_m, where F_m is maximal fluorescence yield of the dark-adapted state and F₀ is minimum fluorescence yield (Maxwell and Johnson, 2000).

Photosynthesis was determined using a portable infrared gas analyzer (Li-COR-6400, United States). Leaf chamber setting parameters were: LED light source of 800 μmol. m^-2 s^-1, leaf temperature 25–26°C, CO₂ concentration of 400 μmol. mol^-1 and 500 flow.

The performance index, is an indicator of leaf vitality, the measurements were performed with the Pocket PEA chlorophyll fluorimeter (Hansatech Instruments, Pentney King’s Lynn, United Kingdom).

The samples were homogenized using a mortar and pestle under liquid nitrogen and adding 80% EtOH. Then, centrifugation was carried out at 12,000 × g, 4°C during 10 min. This extracts were used to measure FRAP, lipid peroxidation (MDA content), chlorophylls and sugars.

For FRAP determination, 5 μL of supernatant was diluted 20 times with 80% EtOH and mixed with 100 μL reaction buffer (5 mL of acetate buffer 0.3 M pH 3.6; 0.5 mL of TPTZ 10 mM (2,4, 6 Tris (2 pyridyl) s-triazine) diluted in 40 mM HCl and 0.5 mL of FeCl₃ 200 mM). This mixture was left to stand for 20 min at room temperature to react and the reaction was subsequently measured at 593 nm. TROLOX was used as a standard to calculate FRAP capacity of the samples (Benzie and Strain, 1996).

Malondialdehyde levels were quantified according to Heath and Packer (1968). Briefly, 200 μL of each sample was mixed with 200 μL of 20% TCA + 0.5% TBA, incubated at 80°C for 20 min and immediately cooled in ice. The mix was centrifuged at 13,000 × g for 10 min and the absorbance of the supernatant was read at 532 nm and 600 nm.

Chlorophylls were calculated by spectrofluorometry at 654 nm in a reaction with 50 μL of extract and 450 μL of EtOH.

Sugars were determine in a reaction of 5 μL of each extract with 45 μL of EtOH and 150 μL of antrone that was incubated 20 min at 4°C, 30 min at 80°C and 20 min at 25°C. The reactions were subsequently measured at 620 nm. Glucose was used to calibrate a standard curve.

The PAL activity was measure according to Beaudoin-Eagan and Thorpe (1985). The enzyme activity was expressed in nmol of trans-cinnamic acid.mg protein^-1.min^-1, where 1 U is defined as 1 nmol trans-cinnamic acid.mg protein^-1.min^-1. 300 mg of leaf samples were homogenized in 3 mL of Tris-HCl buffer 50 mM pH 8.5 with 2-mercaptoethanol 14.4 mM and 100 mg PVPP, and centrifuged at 6,000 × g for 10 min at 4°C. The supernatant were passed through a column of celite and centrifuged at 10,000 × g for 15 min at 4°C. Total protein concentration was determined using the Bradford (1976) assay.

Phytohormones were extracted from 200 g of dry weight leaf as previously described (Durgbanshi et al., 2005) with some modifications. Plant material was homogenized in an ultraturrax T25 basic homogenizer (IKA, Staufen; Germany) with 5 mL deionized water. D2-SA and D6-JA (Leibniz- Institute of Plant Biochemistry; Halle, Germany) were used as internal standards. A total of 50 ng of each standard was added to the samples. The samples were centrifuged at 1,540 × g for 15 min and the supernatant was adjusted to pH 2.8 with 15% (v/v) acetic acid and extracted twice with diethyl ether. The organic fraction was evaporated under vacuum. Dried extracts were dissolved in 1 mL methanol and filtered on a vacuum manifold at a flow rate below 1 mL.min^-1. The eluate was evaporated at 35°C under vacuum in SpeedVac SC110 (Savant Instruments; New York, NY, United States). Four biological replicates were used for the assays.

Hormones were separated from samples using reversed-phase high-performance liquid chromatography (HPLC). An Alliance 2695 separation module (Waters; Milford, MA, United States) equipped with a Resteck Ultra C18 column (100 × 2.1 mm, 3 μm) was used to maintain performance of the analytical column. Fractions were separated using a gradient of increasing methanol concentration, constant glacial acetic acid concentration (0.2% in water), and an initial flow rate of 0.2 mL min^-1. The gradient was increased linearly from 40% methanol/60% water-acetic acid at 25 min to 80% methanol/20% water-acetic acid. After 1 min, the initial conditions were restored, and the system was allowed to equilibrate for 7 min. Hormones were identified and quantified using a quadruple tandem mass spectrometer (Quattro Ultima, Micromass; Manchester, United Kingdom) fitted with an electrospray ion (ESI) source, in multiple reactions monitoring mode (MRM). This assay was performed using precursor ions and their transitions (m/z) to SA (m/z 137/93), D2-SA (m/z 141/97) and JA (m/z 209/59), D6-JA (m/z 215/59) with retention times of 4.35 and 14.30 min, respectively. Collision energies used were 20 eV for both and the cone voltage was 35 V. The analyses were accomplished using MassLynk version 4.1 (Micromass).

Prior to B. japonicum inoculation, 12-days-old plants were transferred to 580 cm³ glass jars containing 100 mL of B&D solution. Three plants per bottle were placed and inoculated with B. japonicum. The containers were sealed to prevent the loss of gaseous hormone. Subsequently, samples were taken through syringes of 10 mL 30 min post inoculation, and subsequently conserved in sealed 10 cm³ vial tubes at 4°C until quantification.

Ethylene content was determined by gas chromatography according to adaptations of the methodology described by Hardy et al. (1968). Briefly, 1 mL of air sample was injected and quantification was carried out for 2 min in a Hewlett Packard Series II 5890 gas chromatograph by determining the ET peak between the retention times of 1.4 and 1.55 min.

Three independent experiments with their corresponding number of replicates for each case were performed in different dates. All data are presented as the mean ± standard error (SE). Comparisons between different treatments were analyzed by ANOVA and Tukey tests using InfoStatTM Software (Di Rienzo et al., 2016). The results were considered significant when p < 0.05.

A multivariate statistical analysis was performed using a principal component analysis (PCA) (Johnson and Wichern, 2006) to explore associations between genotypes/treatments and the set of physiological variables.

As mentioned above, the establishment of the symbiotic relationship between legumes and rhizobia depends on both local and systemic complex signaling pathways (Ferguson et al., 2018). In this regard, although the participation of local ROS signaling in roots after symbiont perception has been widely studied (Cárdenas et al., 2008; Muñoz et al., 2012, 2014a; Damiani et al., 2016), much less is known about the ROS systemic production during the legume–rhizobium interaction. We evaluated histochemically the systemic ROS (O₂^- and H₂O₂) generation in leaves after 30, 60, 120 and 240 min of root inoculation with B. japonicum (Figure 1A,B). The pictures of the stained leaves were transformed in numeric values using image processing software as described in “Materials and Methods.” Besides, the content of H₂O₂ in leaves was also quantified by spectrofluorometer (Figure 1C). Interestingly, the H₂O₂ and O₂^- generation were differentially modulated in leaves after root inoculation (Figure 1). The inoculated plants displayed H₂O₂ generation from 30 to 120 min after root inoculation and the levels were higher than those of the control plants (Figure 1A,C). However, these levels decreased after 240 min post inoculation even below the levels of H₂O₂ observed in the control plants (Figure 1A).

In the case of O₂^-, a transient peak of O₂^- generation was induced at 30 min post inoculation, whereas no differences were observed in the later times with respect to the non-inoculated control plants (Figure 1B). Moreover, this systemic and transient response was specific for B. japonicum. Indeed, the addition of purified Nod factors to the roots induced a transient systemic ROS induction, whereas treatments with chitosan, a fungal elicitor, induced a sustained systemic ROS induction (Supplementary Figure S1).

In order to further characterize the cellular redox state changes in leaves after root inoculation, the content of ascorbic acid, glutathione, MDA and the activities of SOD, GR, CAT, and APX were measured (Figure 2, 3).

The total ascorbic acid content increased in leaves only at 30 min post inoculation compared to leaves of non-inoculated plants (Figure 2A). No significant differences were observed in the ascorbic acid redox state (reduced/total ratio) at any of the evaluated times (Figure 2B). On the other hand, total glutathione content in leaves remained unaltered with the inoculation. The glutathione redox state (reduced/total ratio), however, decreased at 30 min in leaves of inoculated plants compared with control plants (Figure 2D,E). Likewise, the content of MDA, which is an intermediary metabolite of lipid peroxidation used as an oxidative stress marker, significantly increased in leaves of inoculated plants after 30 min of root inoculation thereby correlating with ROS induction (Figure 2C).

NADPH oxidase complex plays important roles in local signaling during symbiotic interaction (Peleg-Grossman et al., 2007; Cárdenas et al., 2008; Montiel et al., 2012; Muñoz et al., 2012; Robert et al., 2018) as well as in systemic signaling triggered by wounding, heat, cold, high-intensity light, and salinity stress (Miller et al., 2009). To investigate the participation of this complex during the systemic ROS induction in leaves after root inoculation, we pretreated the roots or leaves with 50 μM DPI, a well-known inhibitor of flavoprotein enzymes that is extensively used as an NADPH oxidase inhibitor (Figure 3). Interestingly, the induced systemic ROS generation observed in leaves of inoculated plants was completely abolished when DPI was applied in leaf, and a partial inhibited when DPI was applied in root (Figure 3A). Moreover, this inhibitor completely abolished the induction of SOD, GR, and CAT activities observed in leaves 30 min post inoculation (Figure 3B–D). Nevertheless, no significant differences were observed in the APX activity (Figure 3E).

In addition, JA, SA, and ET levels remained apparently unaltered in leaves 30 min after root inoculation with B. japonicum (Figure 4).

To further study the systemic redox changes induced during the soybean–B. japonicum interaction, we analyzed ROS generation in leaves of non-nodulating nfr5-mutant and hyper-nodulating nark-mutant soybean plants after 30 min of root inoculation (Figure 5). The leaves of nark-mutants showed a systemic ROS induction in response to B. japonicum inoculation similarly to that of the wt soybean plants (Figure 5). However, neither the systemic generation of H₂O₂ nor O₂^- were induced in leaves of nfr5-mutant soybean plants 30 min post inoculation (Figure 5).

The legume–rhizobium symbiotic interaction can mediate an increased tolerance to abiotic stresses in host plants by an ISR modulated by JA and ET (Dimkpa et al., 2009; Pieterse et al., 2014). Furthermore, genetic and cell biological evidences revealed the role of ROS as a second messenger during the plant responses to the environment (Miller et al., 2007, 2008; Melchiorre et al., 2009). In this regard, the NADPH oxidase complex not only initiates ROS production but also amplifies them (Foreman et al., 2003; Torres and Dangl, 2005; Miller et al., 2009; Robert et al., 2009).

Stress conditions, in general, are associated with increases in the ROS production, mainly in the chloroplasts, which ultimately leads to photooxidative stress. In order to evaluate the systemic beneficial effects of the symbiotic interaction, the tolerance of non-inoculated and inoculated plants to PQ treatments, an herbicide that increase the chloroplasts ROS production (Lascano et al., 2003, 2012; Melchiorre et al., 2009; Robert et al., 2009), were evaluated. The first trifoliate leaf of non-inoculated and inoculated plants was treated with PQ at 24 and 48 h post inoculation. The selected physiological parameters related to photosynthesis to be assessed after 24 h of PQ application were SPAD, PSII, and F_v/F_m (Supplementary Figure S2). Under control conditions (without PQ), no significant differences were observed between non-inoculated and inoculated plants (Supplementary Figure S3). Likewise, no significant differences were observed between non-inoculated and inoculated plants treated with PQ at 24 h after root inoculation (Figure 6). However, the leaves of inoculated plants were more tolerant than leaves of non-inoculated plants when PQ was applied after 48 h of root inoculation (Figure 6).

To evaluate the involvement of Nod-factors and AON signaling in an ISR/PGPR-like response induced during soybean–B. japonicum interaction, nrf5- and nark-mutant soybean plants were used. Leaves of mutant and wt plants were treated with PQ 48 h after root inoculation, and then diverse physiological–biochemical parameters were assessed (Figure 7). The performance index is an integrative and sensitive parameter that gives quantitative information on the plant performance reflecting the functionality of both photosystems I and II (Strasser et al., 2000). After PQ treatments, leaves of wt inoculated plants showed an increase in the performance index in comparison to leaves of non-inoculated wt plants (Figure 7A). However, the leaves of inoculated and non-inoculated nfr5-mutant plants showed no significant differences after this treatment (Figure 7A). Curiously, the nark-mutant plants showed a lower tolerance upon inoculation (Figure 7A). Furthermore, the photosynthesis rates were significantly lower in wt non-inoculated with respect to wt inoculated plants, whereas no effect of inoculation was observed in leaves of nfr5- and nark-soybean mutants (Figure 7B). In a PCA performed to identify associations between physiological–biochemical parameters and the different plant genotypes under photooxidative stress treatments, the first two principal components (PC 1 and PC 2) of the analysis explained 85.1% of total variability in the data (Figure 7C). In a PCA, the cosine of the angle between two parameter vectors approximates the association among the parameters, with acute and obtuse angles indicating positive and negative correlations, respectively, and right angles denoting no correlation between parameters. The PC1 of the biplot indicated that SPAD, performance index, F_v/F_m, photosynthesis, stomatic conductance and FRAP were positively associated and these vectors were oriented toward wt-inoculated plants (Figure 7C). Interestingly, control and inoculated nfr5- and nark-mutant genotypes were located closed to non-inoculated wt plants (Figure 7C).

Phenylalanine ammonia lyase activity, a key enzyme of the phenylpropanoid pathway and another ISR marker, was evaluated. The analysis of PAL activity revealed that this activity was induced in leaves of wt inoculated plants as compared to wt non-inoculated, while no significant differences were observed between non-inoculated and inoculated nfr5- and nark-mutant plants (Figure 8).

Likewise, to analyze the participation of hormones related to ISR or AON, we assessed the levels of JA and SA in leaves after 72 h of root inoculation (Supplementary Figure S4). The content of these hormones was similar among treatments and genotypes (Supplementary Figure S4).