Bradyrhizobium diazoefficiens USDA 110 (WT) (United States Department of Agriculture, Beltsville, MD, United States), and a bjgb deletion mutant (named 4001 strain), which was previously constructed by Cabrera et al. (2016), were used in this study. B. diazoefficiens strains were grown under aerobic conditions at 30°C in peptone-salts-yeast extract (PSY) medium added with 0.1% (w/v) L-arabinose (Regensburger and Hennecke, 1983). For inocula preparation, cells were collected by centrifugation at 8000 g for 10 min, washed twice and cultured aerobically for 48 h at 30°C in Bergersen minimal medium (Bergersen, 1977) where glycerol was substituted by 10 mM succinate as carbon source and L-glutamate was replaced by 10 mM KNO₃ as sole N-source. Chloramphenicol was added to the cultures at 20 μg ml^–1.

Soybean (G. max L. Merr., cv. Williams) seeds were surface-sterilized with ethanol for 5 min, immersed in H₂O₂ (30%, v/v) for 15 min, and finally washed with sterile distilled water. Then, seeds were germinated in 1% agar petri-dishes (8–9 seeds each) and incubated in darkness at 30°C for 72 h. Seedlings were sowed in autoclaved Leonard jars which contained vermiculite (Trung and Yoshida, 1983). Two soybean plants per jar were inoculated at sowing with 1 ml of a single bacterial strain (approx. 10⁸ cells ml^–1) and overlaid with autoclaved perlite. Plants were transferred to a plant growth chamber for 35 days (16–8 h day/night cycle, day/night temperatures of 26–22°C and photosynthesis photon flux density of 128–148 μmol photons m^–2 s^–1). Plants were cultivated using a mineral solution (Rigaud and Puppo, 1975) with or without 4 mM KNO₃. Treatment of plants with 4 mM KNO₃ does not inhibit nodule formation or nitrogenase activity as previously reported by Mesa et al. (2004). After growth for 28 days, a set of plants were kept under flooding for 7 days by immersing them to 1 cm above substrate level applying mineral solution as described previously (Sanchez et al., 2010). For determination of the plant N content acquired from biological N₂ fixation, plants were watered with 4 mM ¹⁵N-labeled KNO₃ (Potassium nitrate-¹⁵N; 5 atom% ¹⁵N; Cat. #CS01-185_272; Campro Scientific GmbH). Nodules were collected from 35-day-old plants. Plant physiological parameters were determined per plant: nodule number per plant (NNP), nodule dry weight per plant (NDWP), and plant dry weight (PDW).

These analyses were carried out as previously described (Sanchez et al., 2011b). Plants were oven-dried and then weighed and grounded in an IKA A 11 basic analytical mill (Rose Scientific Ltd., Alberta, Canada). For total nitrogen (TN) and ¹⁵N enrichment (δ¹⁵N), subsamples of 3 mg were analyzed with an elemental analyzer (EA1500 NC, Carlo Erba, Milan, Italy) combined to isotope-ratio mass spectrometer (Delta Plus XL, ThermoQuest, Bremen, Germany). The general precision of analyses for δ¹⁵N was ± 0.1‰. The stable composition was shown as δ¹⁵N values per mil: δ¹⁵N (‰) = (R_sample/R_standard − 1) × 1000, where R = ¹⁵N/¹⁴N. Commercial N₂ was the internal standard for the nitrogen isotopic analyses. δ¹⁵N data for all samples were standardized against internationally accepted reference materials (IAEA N1, δ¹⁵N = +0.4‰, IAEA N2, δ¹⁵N = +20.3‰ and USGS32 δ¹⁵N = +174.5 vs. AIR). The proportion of N derived from the atmosphere (%Ndfa) was calculated by following the formula: %Nfda = 100 × [1 − (A/B)], where A = Atom% ¹⁵N excess in inoculated plants, B = Atom% ¹⁵N excess in uninoculated plants. Atom% ¹⁵N excess = atom% ¹⁵N in labeled treatment – atom% ¹⁵N in non-labeled treatment. Atom% ¹⁵N was calculated as: δ¹⁵N (‰) × 100. To calculate the atom% ¹⁵N excess, a set of plants was grown only under N₂-fixing conditions to obtain the atom% ¹⁵N of the non-labeled treatment. The fixed-nitrogen content (FN) was calculated as: FN = (%Ndfa × TN)/100.

Acetylene reduction activity (ARA) was analyzed with fresh detached nodules from plants provided with mineral solution containing 4 mM KNO₃. About 20 nodules per replica were placed in 20-ml headspace vials (SUPELCO^®) containing 100 μl mineral solution added with 4 mM KNO₃. Tubes were sealed, filled with 1 ml of pure acetylene and incubated at 30°C. After incubation for 2 and 4 h, gas samples (0.5 ml) were extracted from the tubes for ethylene analysis. A Hewlett-Packard model 5890 gas chromatograph (Agilent Technologies, S.L., Madrid) with a flame ionization detector and a molecular sieve 5A (60–80 mesh) column (180 cm × 0.32 cm) (Agilent Technologies, S.L.) was used. N₂ at 60 ml min^–1 was used as carrier gas. Temperatures of oven, injector, and detector were 60, 90, and 110°C, respectively. Ethylene concentration in each sample was calculated from standards of 2% (v/v) ethylene. Acetylene reduction rate was calculated by measuring the increase of ethylene (C₂H₄) production inside the vials headspace determined in the lineal range (0, 2, and 4 h of incubation) (Supplementary Figure S1) and calculated as: Δ C₂H₄ nmol (4–2 h)/Δ time (4–2 h). Results presented in Figure 1A were expressed as: [nmol C₂H₄ min^–1 (g NFW) ^–1] where NFW = nodule fresh weight.

Leghemoglobin (Lb) was quantified in nodules from plants provided with mineral solution containing 4 mM KNO₃ following the protocol described by Sanchez et al. (2010). Fluorimetric detection of Lb was analyzed in a Shimadzu (Shimadzu Scientific Instruments, Kyoto, Japan) spectrophotofluorometer with a mercury-xenon lamp and a RF-549 red-sensitive photomultiplier. The excitation wavelength was 405 nm and the emission monochromator adjustment was 650 nm. Heme protein concentration was proportional to the difference in fluorescence between heated and unheated samples and therefore it was indicative of the leghemoglobin concentration in the samples. Lb content in each sample was calculated using bovine hemoglobin as standard. Lb content was expressed as: [mg Lb (g NFW) ^–1].

Nodules were detached from roots, frozen in liquid nitrogen and stored at −80°C until further use. For RNA isolation, nodules were distributed in 2-ml Eppendorf tubes and were disrupted and homogenized by using a TissueLyser (Mixer Mill MM 301; Retsch GmbH, Haan, Germany) (three times for 50 s at 30 Hertz) with a tungsten carbide bead (3 mm; Qiagen). Then, RNA was isolated by following the hot phenol-extraction protocol described elsewhere (Babst et al., 1996). RNA integrity was confirmed by agarose gel electrophoresis. Precipitated RNA was treated with DNaseI amplification grade (Invitrogen). Next, RNA samples were cleaned up with RNeasy Mini spin columns (Qiagen). Genomic DNA contamination was checked by PCR using fixN4_For and fixN4_Rev primers (Supplementary Table S1). Then, 2 μg of total RNA was reverse transcribed to cDNA employing random hexamers and SuperScript II reverse transcriptase (Invitrogen) following the supplier’s instructions. For quantitative reverse transcriptase-polymerase chain reaction (qRT-PCR) analyses, an iQTM5 Optical System (Bio-Rad, Foster City, CA, United States) was used. Each PCR reaction (with a total volume of 19 μl) was performed using 9.5 μl iQ^TM SYBR Green Supermix (Bio-Rad), 2 μM (final concentration) of individual primers (Supplementary Table S1) and proper dilutions of different cDNA samples. The PCR program included an initial denaturation and Taq activation step of 5 min at 95°C, followed by 40 cycles of 15 s at 95°C, 45 s at 60°C, and 45 s at 72°C. PCR reactions were performed in triplicate. The formation of specific PCR products was checked by melting curve analyses. Expression of the gene encoding the 16S rRNA was used as reference for normalization (primers 16S_qRT_For and 16S_qRT _Rev; Supplementary Table S1). Relative changes were calculated by using the Pfaffl method (Pfaffl, 2001). Expression of narK in WT flooded nodules was determined relative to that observed in WT non-flooded nodules. For nifH and norC expression, values were calculated relative to the levels of nifH or norC expression in WT non-flooded nodules. The expression levels given as fold-change for both the WT and bjgb mutant were compared with those from WT under non-flooding conditions.

Nitric oxide was detected in fresh detached nodules (about five nodules) from plants provided with mineral solution added with 4 mM KNO₃. For NO scavenger treatment, nodules were incubated for 1 h in the dark at 30°C in a solution containing 2-[4-carboxyphenyl]-4,4,5,5-tetramethylimidazoline-1-oxyl-3-oxide (c-PTIO) (Sigma-Aldrich) at a final concentration of 3 mM. Then, NO was detected by dipping the nodules in 7 μM of 4,5-diaminofluorescein diacetate (DAF-2 DA) solution (Genaxxon bioscience) for 2 h in the dark at 30°C, while control nodules were incubated in distilled water under the same conditions. The relative fluorescence units of the DAF-2 DA solution were determined using a fluorometer (MD-5020, Photon Technology International, Birmingham, NJ, United States) with 495 nm excitation and 515 nm emission wavelength (2 nm band width). Nodular NO accumulation was expressed as relative fluorescence units per nodule dry weight (NDW): [RFU (mg NDW)^–1].

N₂O was measured in fresh detached nodules from plants provided with mineral solution supplemented with 4 mM KNO₃. Essentially, detached nodules from roots were placed in 20-ml headspace vials (SUPELCO^®) (about 20 nodules) provided with 100 μl (for nodules incubated under non-flooding conditions) or 5 ml (to maintain nodules under flooding treatment) of mineral solution with 4 mM KNO₃. Acetylene 10% (v/v) was injected into each tube in order to inhibit nitrous oxide reductase (Yoshinari and Knowles, 1976). Samples were incubated for 6 h at 30°C before measuring N₂O production. Then, gaseous aliquots (1 ml) were taken from the headspace using a gas-tight syringe and they were manually injected into an HP 4890D gas chromatograph with an electron capture detector and a Porapak Q 80/100 MESH (8 ft) packed column. N₂ at 28 ml min^–1 was the carrier gas. The injector, column and detector temperatures were 125, 60, and 375°C, respectively. Concentrations of N₂O in the samples were determined employing 2% (v/v) N₂O as standard. Total N₂O concentration was calculated considering N₂O in headspace and dissolved N₂O using Bunsen water solubility coefficient (54.4% at 25°C). For each replicate, N₂O flux was recorded after 4 and 6 h of incubation, which was the lineal range of N₂O emission. The N₂O emission flux was calculated as: Δ N₂O molar concentration (6–4 h)/Δ time increase (6–4 h). Results were expressed as: [nmol N₂O h^–1 (g NDW) ^–1].

The total number of replicates for each assay is shown in each figure and table. For each parameter, descriptive statistical analyses (media and standard deviation) were calculated and data were checked for normal distribution and homoscedasticity. The analyses of variance (ANOVA) within treatments were performed using the post hoc Tukey–Kramer test (P ≤ 0.05). For inferential statistical analyses, GNU-PSPP open-source software v0.9.0¹ was used.

After 35 days of growth, NNP, NDWP, and PDW were analyzed (Table 1). As previously reported (Mesa et al., 2004), the presence of nitrate in the nutrient solution did not inhibit NNP or NDWP in plants inoculated with either the WT or the bjgb mutant (Table 1). Flooding treatment did not significantly alter NNP in plants grown either in the presence or in the absence of nitrate, independently of the strain used as inoculum (Table 1). On the contrary, flooding provoked a significant inhibition of NDWP of plants cultivated with or without nitrate. Interestingly, those plants inoculated with the bjgb mutant and grown with nitrate showed higher NDWP in response to flooding (about 30%) than plants that were inoculated with the WT (Table 1). As expected, the presence of nitrate increased PDW compared to non-nitrate treated plants. As shown in Table 1, in the absence of nitrate, flooding did not significantly affect PDW in plants inoculated with any of the strains. However, in nitrate-treated plants, flooding provoked a negative effect on PDW that was only significant (about 28%) in those plants inoculated with the WT strain but not in those inoculated with the bjgb mutant (Table 1). As control of symbiotic nitrogen fixation, a set of uninoculated plants were included in the analyses. After 35 days growth, nodulation did not take place in uninoculated plants and very low levels of PDW were observed in those plants grown without nitrate (Table 1). In plants that were only nitrate dependent, flooding decreased PDW about 36% (Table 1).

As is indicated in Table 2, flooding stress provoked a significant decrease in the TN content of shoots from inoculated plants grown with 4 mM KNO₃ (Table 2). Nevertheless, plants inoculated with bjgb showed an increase of about 12% in TN compared to plants where the WT was used as inoculum (Table 2). In uninoculated plants cultured with nitrate, flooding did not affect TN (Table 2). In order to differentiate between N acquired from N₂-fixation or nitrate assimilation, we used the ¹⁵N isotope dilution technique that allowed us to determine the proportion of N derived from the atmosphere (Ndfa%) in plants that were cultivated with 4 mM ¹⁵N-labeled KNO₃. Here, flooding decreased %Ndfa of plants, being this reduction significantly higher in plants inoculated with the WT compared to those inoculated with the bjgb mutant (55% vs. 30%). Consequently, the content of fixed nitrogen (FN) of flooded plants inoculated with bjgb was about 47% higher than that of flooded plants inoculated with the WT (Table 2). These results clearly indicate that inoculation of the plants with the bjgb mutant confers tolerance of symbiotic nitrogen fixation to flooding. By using the ¹⁵N isotope dilution technique, we also calculated the % of ¹⁵N atom in excess (atom% ¹⁵N excess) of the shoots. As shown in Table 2, ¹⁵N excess in uninoculated plants was significantly higher compared to inoculated plants and no effect of flooding in ¹⁵N excess of uninoculated plants was perceived. However, in plants inoculated either with the WT or bjgb mutant, flooding increased the % of ¹⁵N excess. These results indicate that soybean nitrogen fixation is more sensitive to flooding than nitrate assimilation supporting previous findings (Bacanamwo and Purcell, 1999; Sanchez et al., 2011b).

In order to confirm the flooding effect on N₂ fixation, nitrogenase activity was measured by determining ARA in nodules incubated for 2 and 4 h (Supplementary Figure S1). Values for C₂H₄ produced in nmol per min and g NFW (Figure 1A) showed a decrease in ARA (about 55%) for WT nodules isolated from plants submitted to flooding compared to non-flooded nodules. However, nitrogenase activity only decreased about 30% in bjgb nodules from flooded plants compared to non-flooded bjgb nodules (Figure 1A). Functionality of the nodules was also estimated by measuring the Lb content (Figure 1B). Consistent with ARA determinations, flooding provoked a smaller decrease of Lb content in nodules produced by the bjgb mutant (about 22%) compared to that observed in WT nodules (about 55%).

Next, we tested the effect of deleting bjgb on the expression of the nifH gene which is responsible for the synthesis of the Fe-protein from nitrogenase complex (Table 3). Transcript levels for nifH were examined in WT or bjgb nodules by qRT-PCR. As observed in Table 3, nifH expression was not significantly affected in the nodules induced by the bjgb mutant (−1.31 fold-change) compared to WT nodules, both collected from non-flooded plants. Flooding decreased nifH expression by ∼16-fold in WT nodules compared to that observed in WT non-flooded nodules. However, nifH mRNA levels decreased by ∼10-fold in bjgb flooded nodules compared to those observed in WT non-flooded nodules.

In order to ascribe the effect of bjgb inoculation to the presence of Bjgb in the nodules, we checked the expression of the bjgb gene by analyzing transcript levels of narK, the lead gene of the narK-bjgb-flp-nasC transcriptional unit (Cabrera et al., 2016) (Table 3). As shown in Table 3, narK expression increased ∼11-fold in flooded nodules compared to non-flooded nodules collected from plants inoculated with the WT strain.

The contribution of B. diazoefficiens Bjgb in NO homeostasis in nodules was investigated by analyzing the capacity to accumulate NO as well as to produce N₂O, the product of the nitric oxide reductase (Nor). To perform these experiments, we used nodules from plants grown under the conditions that induced NO and N₂O accumulation, as previously reported (Sanchez et al., 2010; Tortosa et al., 2015). Thus, nodules were collected from soybean plants inoculated with B. diazoefficiens USDA 110 (WT) or 4001 (bjgb mutant) strains, cultivated with 4 mM nitrate and submitted or not to flooding conditions for 7 days before harvesting. Free NO was detected by using the DAF-2DA specific fluorescent probe. As shown in Figure 2A, very low levels of NO were observed in WT and bjgb nodules of non-flooded plants. However, flooding significantly induced NO formation in WT nodules confirming previous results (Sanchez et al., 2010). This induction was also observed in nodules from plants inoculated with the bjgb mutant that accumulated approximately 2-fold less NO than WT nodules in response to flooding conditions (Figure 2A). In order to prove that the increase of fluorescence perceived in flooded nodules was caused by NO, WT, and bjgb flooded nodules were incubated with a NO scavenger (c-PTIO). After incubation of the nodules with c-PTIO, nodular NO accumulation was significantly reduced (Figure 2A), indicating that the fluorescence signal was mostly due to NO production.

We also analyzed Nor activity by measuring N₂O production by soybean nodules using gas chromatography. While non-flooded nodules produced by WT or bjgb strains showed basal levels of N₂O, flooding conditions induced N₂O production in B. diazoefficiens WT nodules, consistent with previous observations (Tortosa et al., 2015). Interestingly, nodules produced by the bjgb mutant accumulated about twofold more N₂O than WT nodules in response to flooding (Figure 2B). The lower NO levels as well as higher N₂O production capacity observed in the nodules from the bjgb mutant compared to WT levels (Figures 2A,B), suggest that Nor activity that reduces NO to N₂O is induced in bjgb bacteroids. In order to establish if the differences of Nor activity observed between WT and bjgb bacteroids could be explained by changes in nor gene expression, norC transcripts were measured by performing qRT-PCR analyses. RNA was isolated from nodules harvested from soybean plants inoculated with B. diazoefficiens WT or bjgb mutant strains. In nodules from plants that were not subjected to flooding, bjgb mutation did not significantly affect norC expression compared to wild-type (+1.12 fold-change). Flooding provoked a notable increase of norC expression in either WT or bjgb flooded-nodules compared to WT non-flooded nodules (+67.53 and +89.16 fold-change, respectively). Interestingly, the induction of norC expression by flooding was about 33% higher in bjgb nodules than in WT nodules (Table 3).