The Interaction between Bradyrhizobium Japonicum E109 and Azospirillum Brasilense Az39 Improves Bradyrhizobium-Soybean Symbiosis: The Secrets Behind Co-Inoculation

Daniela Torres Universidad Nacional de Río Cuarto Florencia Donadio Universidad Nacional de Río Cuarto Gastón López Universidad Nacional de Río Cuarto Romina Molina Universidad Nacional de Río Cuarto So a Nievas Universidad Nacional de Río Cuarto Sanja Ćavar Zeljković Palacky University Martín Díaz-Zorita Universidad Nacional de La Pampa Nuria De Diego Palacky University Fabricio Dario Cassan (  fcassan@exa.unrc.edu.ar ) Universidad Nacional de Río Cuarto https://orcid.org/0000-0002-9776-0262


Introduction
Rhizobia are diazotrophic bacteria whose symbiotic interactions with plants from the Fabaceae family result in the formation of speci c structures where biological nitrogen xation (BNF) takes place (Stacey et al. 1992). Some rhizobia, mostly belonging to the genus Bradyrhizobium such as B. japonicum, have been extensively studied in terms of their association with soybean as sustainable alternatives to N-fertilizers (Hungria and Nogueira 2019). On the other hand, Azospirillum are plant-growth promoting bacteria that have attracted particular attention over the last four decades because they improve legume growth, development, and yield when co-inoculated with rhizobia (Bashan and de Bashan 2010). These positive effects are mostly related to Azospirillum's ability to produce phytohormones such as auxins (Crozier et al. 1988; Zimmer and Bothe 1988), cytokinins (Horemans et al. 1986; Cacciari et al. 1989), gibberellins (Bottini et al. 1989;Janzen et al. 1992), abscisic acid and ethylene (Perrig et al. 2007), as well as certain polyamines such as cadaverine (Cassán et al. 2009). The production of auxins, mainly indole-3-acetic acid (IAA), is considered the most important plant-growth promoting mechanism in A. brasilense (Cassán et al. 2014). However, in the case of Bradyrhizobium, the mode of action is still unclear, but many bene ts for the plant have been described yet (Jaiswal et al. 2021). The practice of co-inoculation could improve plant performance and the establishment of symbiosis under both controlled and eld conditions (Ferri et al. 2017; Rodelas et al. 1999;Remans et al. 2008;Cassán et al. 2020). Hungria et al. (2013) and  reported an increase in soybean [Glycine max (L.) Merrill] grain yield of over 15% and 200 kg.ha −1 , respectively, through a combination of Bradyrhizobium and Azospirillum as A. brasilense. Studies on soybean coinoculation in Argentina and Brazil also found respective increases in nodulation of over % (Hungria et  Therefore, Azospirillum's ability to produce phytohormones like auxins and cytokinins, and the morphofunctional changes thus induced in the soybean root system (Rondina et al. 2020) enhance the bene ts of Bradyrhizobiumsoybean symbiosis (Srinivasan et al. 1996; Molla et al. 2001;Vessey and Buss 2002). In the speci c case of B. japonicum E109, analysis of its genome revealed the existence of three putative pathways for IAA biosynthesis, but the bacterium is unable to produce it, and can instead degrade both natural and synthetic auxins, including IAA and NAA (Torres et al. 2018). Nevertheless, its biomass and exopolysaccharide production increased with the exogenous addition of IAA to the liquid culture medium where it was grown, which in turn modi ed its symbiotic behavior with soybean: more nodules were created, and a higher percentage of plants were nodulated (Torres et al. 2018). This supports the idea that it is the active compounds synthesized by Azospirillum which improve nodulation by Bradyrhizobium, yet the mechanisms underlying these effects remain poorly understood. With this in mind, we aimed to elucidate whether symbiosis between Bradyrhizobium and soybean is effective at least in part due the release of active molecules by co-inoculated A. brasilense, and to investigate some particularities behind this kind of co-inoculation.

Plant material
Seeds of soybean [Glycine max (L). Merrill] var. "Don Mario 3810 RR" were used for these studies. Quality control parameters were established by the International Seed Test Association (ISTA) (http://www.ista.org).

Bacterial growth and seed inoculation
Seeds were inoculated with individual or combined cultures of BjE109 and AbAz39, according to the treatments described below. The BjE109 titer was adjusted to 5.0 E+9 cfu.ml −1 obtained at late exponential growth phase in yeast mannitol broth (YEM), as described by Vincent (1970). The AbAz39 titer was adjusted to 5.0 E+8 cfu.ml −1 obtained at late exponential growth phase in Luria broth (LB), as described by Molina et al. (2018). Inoculation doses were adjusted to obtain a nal volume of 12 ml.kg −1 of soybean seeds in all experiments. Five treatments were performed: i) uninoculated seeds treated with phosphate buffer solution (12 ml.kg −1 ) (control); ii) seeds inoculated with equal volume of BjE109 and phosphate buffer solution (1:1) (BjE109); iii) seeds inoculated with equal volume of AbAz39 and phosphate buffer solution (1:1) (AbAz39); iv) seeds co-inoculated with BjE109 and AbAz39 in a 1:1 ratio (BjE109 + AbAz39), and v) seeds inoculated with a mix, in a 1:1 ratio, of BjE109 and AbAz39 (BjE109-AbAz39). In the treatment iv) each microorganism was applied separately during the inoculation process, while for the treatment v) equal volumes of both microorganisms were mixed and maintained at room temperature (25°C) during 24 h before the inoculation of the seeds. The incubation time for treatment BjE109-AbAz39 was established following previous results by Torres et al. (2018) about the ability of BjE109 to degrade 40 µg.ml −1 IAA and increase its own exopolysaccharide (EPS) synthesis and biomass production in YEM culture medium. We also evaluated each microorganism without dilution as a control. After the application of the treatments, the seeds were maintained under sterile laminar air ow conditions at room temperature (25ºC) until analysis.

Bacterial count and physiological state
To determine the number of viable cells (cfu.ml −1 ) in each culture or in the mix of both, the microdrop quanti cation method was performed (Miles and Misra 1938). Plates containing YEM medium (Vincent 1970) or Congo Red medium (Rodríguez Cáceres 1982) were used for BjE109 and AbAz39, respectively. The plates were analyzed after incubation at 30°C for 7 days in the case of BjE109 and 37°C for 4 days in the case of AbAz39, as described by Cassán et al. (2014) with modi cations. In the case of the combined cultures, plates containing both YEM and Congo Red medium were used at temperatures and incubation time mentioned before.

IAA quanti cation
Quanti cation of IAA was performed by spectrophotometry (Glickmann and Dessaux 1995) and con rmed by HPLC following Torres et al. (2021). Aliquots of 1000 µl of bacterial culture were centrifuged at 11,300 g for 10 min. Next, the samples were ltered (0.2 µm) and 500 µl of the supernatant were mixed with 500 µl of Salkowski´s reagent and gently shaken in an inverted position at least 10 times. The samples were incubated in the dark for 30 min and absorbance was measured at 530 nm. An aliquot of ltered supernatants was injected into the HPLC equipment as

Exopolysaccharide (EPS) quanti cation
Quanti cation of EPS followed Torres et al. (2018). Brie y, the samples were centrifuged and then the supernatants were ltered and treated with DNase I and proteinase K. The EPSs were precipitated with ethanol and dried at room temperature before being resuspended in deionized water. Total carbohydrate content (EPS.mg -1 biomass) was measured through the phenol-sulfuric acid method, with glucose as a standard.

Seed recovery assays
Soybean seeds that had been either inoculated or co-inoculated (section 3.3) were maintained in aseptic environmental conditions at room temperature (≅ 25°C) for 4 hours before initial analysis and then 6 days before a second analysis, as described by Torres et al. (2018). Bacterial cell count and survival factor percentage (SFP) were determined for BjE109 and AbAz39 following the criteria by Penna et al. (2011). The plates were incubated at 30°C for 7 days in the case of BjE109 and at 37°C for 4 days in that of AbAz39. Additional counts were performed on uninoculated seeds (control). Results are expressed as the number of viable cells recovered from the seeds

Greenhouse assays
The treated soybean seeds from each treatment listed in section 3.3 were sown in plastic pots (300 ml volume capacity) containing vermiculite as a solid substrate. Three seeds per pots were planted in six pots and irrigated with sterile N-free Hoagland's solution (25% v/v) (Hoagland and Boyer 1936) keeping them at eld capacity during the study. The seedlings were maintained inside a growth chamber for 21 days with a photoperiod of 16/8 h light regime, 30/20°C temperature and a relative humidity of 80%. At the end of the experiment, the following parameters

Statistical analysis
The treatments for greenhouse experiments were performed in with six replicates per treatment, while the eld experiment was placed in randomized blocks with six replicates per treatment. The values shown represent the mean ± standard error of mean (SEM). The data was analyzed for variance factor differences using ANOVA followed by

Bacterial count and physiological state
To de ne an experimental model based on the interaction between BjE109 and AbAz39, mixes of different proportions of each strain (1:9, 1:1 and 9:1) were evaluated for the EPS production, microbial biomass and IAA concentration in the culture medium after 24 h interaction. The results of this assessment are showed in Supplementary material (Table S1). An equal proportion of each microorganism (1:1) was selected, since it rendered the highest production of biomass and EPS, the two parameters of our main interest in this experiment according to Torres et al. (2018). Table 1 summarizes the cell number, EPS content, IAA concentration and biomass production of BjE109 and AbAz39 cultures obtained individually (control), in combination (1:1) with buffer phosphate (PB), or with both bacteria incubated for 15 minutes (T0) or 24 h (T24). Table 1 Cell number, EPS production, IAA concentration and biomass production of BjE109 and AbAz39 cultures obtained individually (control), in combination (1:1) (1:1), which could be ascribed to the presence of the growth medium for AbAz39. Contrarily, in the co-inoculation the IAA levels were 14.3% lower than in the AbAz39 culture diluted in phosphate buffer (1:1), and this may be attributed to BjE109's ability to degrade the hormone. When the T24 mix was examined, the values of EPS content and biomass were still increased with three and two times higher than in the BjE109 phosphate buffer culture, respectively, likely due to the interaction between both bacteria. IAA concentration, meanwhile, had fallen below detectable levels because of BjE109's degrading activity. Table 2 summarized the values for cell recovered and survival factor on seeds treated with the strains individually or in combination. No microorganisms compatible with the ones used in this study could be detected on untreated seeds. On seeds inoculated with AbAz39, viable bacteria and the survival factor percentage (4.87%) could be measured only 4 h after inoculation and were lower than those obtained for seeds inoculated with BjE109. The survival factor 4 h post inoculation was 14.54% and 0.8% after six days, i.e., survival was longer despite the signi cant fall since the 4-day measurement. The 4 h post-inoculation value for BjE109 was ramped up by 36.77% when it was inoculated at the same time as AbAz39 in a 1:1 proportion (T0), while the 6 d post-inoculation value rose almost two times under the same conditions. Co-inoculation at T0 measured 4 h after inoculation also increased AbAz39's survival by 25%. The pattern observed here suggests both strains have better chances of surviving when they can interact among each other on the seeds.

Seed recovery assays
In the 1:1 mix inoculated at T24, the survival rate for BjE109 rose approximately 75% 4 h post inoculation and 90% 6 d post-inoculation with respect to the values obtained at T0. The differences were bigger if these results were compared to the individual BjE109 inoculation, reaching 2-fold and 10-fold higher values, respectively. The T24 mix, moreover, increased AbAz39's count and survival by around 9% with respect to the T0 treatment, and 33% with respect to the strain inoculated on its own. These results con rmed the mutual bene t relationship between the two strains for longer survival. Figure 1 shows the recovery at different post-inoculation times of viable BjE109 cells on seeds inoculated with the strain on its own or in combination with AbAz39 (1:1), 15 m after the mix was prepared (T0) o 24 h later (T24). Grey bars represent recovery after treatment with the T24 mix, considered the most successful as regards the survival values up to 15 days after inoculation compared to individual treatment, and 9 days compared to T0. When BjE109 was inoculated on its own, recovery was indeed possible but at very low values and only up to 6 days after inoculation. The difference between the T24 and T0 treatments was at its peak immediately after inoculation, and still observable though ever decreasing between day 3 and day 15 after inoculation.

Greenhouse assays
The data shown in the Table 3 contains the growth and nodulation values for soybean seedlings 4 hours or 6 days after inoculation with either BjE109 or AbAz39, individually, or the combination of both at T0 or T24.  hours before inoculation (T24) is bene cial for the nodulation and the plant growth compared to T0 and the treatment only with BjE109. Although all measurements describing the symbiosis and the plant growth signi cantly dropped 6 days after inoculation (6 dpi), both co-inoculation treatments continued outperforming the one with BjE109 alone, a further indication of the advantages of co-inoculation over the long term. In a similar manner to the earlier determinations, MRN, SRN and RN were 11.3%, 11.8% and 3.1% higher, respectively, 6 dpi with the T24 mix than at T0. SDW and BT were also higher (3.7% and 7.0%). The co-inoculation at T0 could only outperform the T24 treatment when it came to RDW, both 4 dpi and 6 dpi.
The Fig. 2 shows the nodulation in the roots of soybean seedlings developed from seed inoculated with either BjE109 or AbAz39, individually, or the combination of both at T0 or T24. Both MRN and RN signi cantly decreased over time regardless of the treatments. Nevertheless, the results were signi cantly better up to 15 days after inoculation (15 dpi) with the T24 mix than with the treatment at T0 or with BjE109. The co-inoculation at T0 outperformed the seed inoculation with the individual strain, except on day 3 (3 dpi) when there were no statistically signi cant differences between both treatments. In summary, both mixtures, but particularly the one inoculated at T24, were able to revert the decreasing nodulation trend over time.

Field assay
The data in the Table 4 show how, under eld conditions, the inoculation with either BjE109 or AbAz39, individually, or the combination of both at T0 or T24 modi ed the growth of soybeans crops and its symbiosis. greater than in those developed from uninoculated seeds. After the combination of BjE109 and AbAz39, the biomass of the nodules was greater than the observed when single BjE109 or AbAz39 treatments were applied to the seeds. The co-inoculation at T0 also increased 12.9% the soybean shoot biomass compared to the individual inoculation with BjE109 and 17.6% in the case T24 without differences between both combined treatments. The nodule biomass varied from 1.038 g without inoculation to 1.631 g with co-inoculation at T0 and 1.685 g at T24 representing an increase of 14.4% and 18.2% respectively. In the treatment only with BjE109 this value was 1.425 g and represented an increase of 7.1% in comparison to the control without inoculation. In terms of nodule size, the T24 induced the highest number of nodules greater than 5mm (range I) and smaller than 2 mm (range III), while the T0 treatment rendered more nodules from 2 to 5 mm (range II). In both cases, the plants developed from the co-inoculation treatments produced more nodules within all three ranges than the single inoculation with BjE109. At the same time, this last treatment improved the number of nodules in the ranges I and II than those obtained with the single inoculation with AbAz39 but reduced those in the range III. Nodule size after the treatment with AbAz39 was like the uninoculated control in ranges I and II, and this fact suggests the capability of Azospirillum to interact with the rhizobia of the soil promoting nodulation.
The nodule number and its location in the root system of the plants was also assessed. The highest MRN, SRN and total number of nodules (TN) was observed after co-inoculation with the 1:1 mix at T24. The SRN and TN were signi cantly higher after inoculation with AbAz39 on its own than in the control. On the other hand, the relative difference for MRN, was not signi cant, which might indicate a positive interaction between Azospirillum and the native bradyrhizobia in the soil where the assay was carried out. The shoot and the root biomass of the plants was also signi cantly increased after the co-inoculation (Table 4). However, whereas the SDW was signi cantly higher in those plants co-inoculated at T24 compared to T0, these differences were not observed for RDW. The intensity of the green color of the upper leaves measured in SPAD index units did not show differences among the treatments.
Insertion Table 5  As can be seen in Table 5, the greater nodulation and biomass production observed, under eld conditions, with the co-inoculation of AbAz39 and BjE109 at T24 (Table 4) correlated with a higher soybean grain yield. The treatment of co-inoculation at T24 favored a grain yield production 17.0% greater than the control without inoculation, 15.0% than with the single inoculation with AbAz39, 9.1% than only with BjE109, and 3.3% than the co-inoculation at T0. There were no mean differences in the single grain weight and in the concentration of N in the grains suggesting that most of the yield differences were related with better growth conditions among all the growing season and keeping an e cient biological N xation.
Based on the integrated analysis of the results, we interpret that the co-inoculation with AbAz39 and BjE109 has many advantages in soybean plant growth so improves the quality and the quantity of the nodulation and the crop grain yield. To validate this interpretation of the results we performed a correlation matrix representing the evaluated parameters from all treatments (Fig. 3). As results, we obtained that all the crop growth and yield related parameters, except for the green intensity of the upper leaves, signi cantly correlated with the nodulation parameters. Also, it is observed that the plant growth has more in uence on the number of nodules and its location than in their single size.

Discussion
In plants, co-inoculation is performed with a combination of different bacterial genera or species with the aim to modify plant growth and development synergistically. A great number of non-symbiotic or free-living bacteria are known to promote symbiosis in legumes when co-inoculated with rhizobia. A recent analysis of over 300 cases of positive plant responses to Azospirillum in 12 countries (Cassán and Díaz-Zorita 2016) found that the rise in its coinoculation with rhizobia led to a 6.6% increase in legume yield over rhizobia-only treatments. In the last decades, In this study, with the aim of nding further evidence that may elucidate the relative importance of the soybean-Azospirillum and Bradyrhizobium-Azospirillum interactions, we evaluated the effect of co-inoculating soybean seeds with BjE109 and AbAz39 at different ratios and interaction times. Both, in vivo and in planta assays were carried out to compare the performance between co-inoculation and single inoculation treatments with the bacterial strains on their own. The co-inoculation, combining Bradyrhizobium and Azospirillum strains, in a 1:1 proportion when treating the soybean seeds, represents the frequent co-inoculation treatment applied at farmers' elds. In this case, both strains are mixed and start their interaction with each other at the same time they are inoculated on the seeds. In this scenario, most of the effects of the presence of Azospirillum have direct impact on the plant. If the mix of microorganisms is prepared 24 hours before treating the seeds (T24), the strains interact with each other before reaching the seed coat and the derived bene ts on nodulation and soybean growth could be interpreted from the effects of Azospirillum on Bradyrhizobium performance.
Because of its backbone contribution in the nitrogen nutrition of soybean plants, the main microorganism in the coinoculation mixture is BjE109. Initially, to analyze the physiological behavior of BjE109, we assessed different proportions of the combination with Azospirillum (1:9, 1:1 and 9:1). The 1:1 treatment was chosen because it had the greatest values of EPS production, bacterial biomass, and IAA degradation for which AbAz39 is responsible (Table S1). These parameters were selected with reference to previous studies published by our laboratory (Torres et al. 2018;2021), in which we con rmed that exogenous IAA addition to pure BjE109 cultures triggered physiological changes in the bacterium, such as the ability to increase its biomass and produce EPS, both are advantages on the rhizobia cell survival on soybean seeds and on its symbiotic performance. We also obsserved that BjE109 can degrade IAA when is added to the culture medium through the action of a 3-phenylpropanoate dioxygenase-like enzyme (subunit alpha and beta) within a cluster named iac (Torres et al. 2018;2021). This must be relevant for a better performance of the co-inoculation so the mix at T24 doubled the EPS content and signi cantly increased the biomass production in comparison to the T0 treatment (Table 1). In turn, T0 outperformed inoculation with BjE109 alone. An increase in the production of EPS by rhizospheric bacteria like rhizobia is related to higher plant tolerance to stress caused by water, oxidation, low temperature, and When the BjE109 survival was evaluated, the co-inoculation allowed recovering of more viable cells compared to the rest of treatments, being better in the T24 mix than in T0 ( . These ndings suggest that the production of IAA by AbAz29 in the culture medium of the mix may be partly responsible for the increase in EPS, which may offer BjE109 a better chance at surviving on the seeds. Auxins have long been posited to play a major role in nodule ontogenesis within legumerhizobium symbiosis (Thiman 1936), and there are many reports on alterations in nodule organogenesis due to changes in auxin content because of inoculation with auxin-producing bacteria or exogenous hormone addition (Schmidt et al. 1988). The co-inoculation with rhizobia and Azospirillum is not an exception. For example, coinoculation with IAA-de cient Sinorhizobium meliloti and IAA-producing A. brasilense on alfalfa generated signi cantly more nodules on the primary root (Schmidt et al. 1988). Remans et al. (2008) also provided direct evidence on the enhancing effects of IAA by co-inoculating beans with R. etli and a mutant IAA-de cient A. brasilense. The literature, then, might lead us to believe that the improvement observed in legume growth after inoculation with Azospirillum is explained by a hypothetical interaction between the bacteria and the plant, in which phytohormone synthesis (primarily IAA) could be crucial.
The co-inoculation could also favor AbAz39 performance that was conditioned by the interaction time. In this regard, it has been observed that the interaction of Bradyrhizobium with Azospirillum in soybeans seeds highly improved drought tolerance (Rondina et al. 2020)  uncontrolled growth conditions ( eld) compared to T0, and much more when the bacteria are used individually (Table 4 and 5). However, we also demonstrate that the number but not the size of the nodule in uences the quality and quantity of the plant yield (Fig. 3). In the last decades, several authors have analyzed the contribution of coinoculation with Azospirillum sp. to legume productivity. In the pampas region (Argentina), 21 eld trials performed on alfalfa showed that combined inoculation of Ensifer meliloti and A. brasilense was almost twice as effective as inoculation with rhizobia alone (Díaz-Zorita et al. 2012). Hungria et al. (2013) also reported increased yield for soybean and common bean when complementing seed inoculation with rhizobia with the in-furrow application of A. brasilense at four locations in Brazil. They found that inoculation of soybean with Bradyrhizobium resulted in an 8.4% increase in yield, against the 16.1% increase achieved by the combination of strains. For common beans, individual inoculation with R. tropici boosted yield by 8.3%, but the addition of A. brasilense raised this gure to 19.6%. A metadata analysis by Barbosa et al. (2021) revealed that soybean co-inoculation was related to a 2.8% increase in grain yield and a 3.2% increase in N concentration in the grains with respect to the inoculation only with Bradyrhizobium. In other 37 eld trials, soybean co-inoculation with Bradyrhizobium and A. brasilense also increased mean yield by 227 kg ha −1 with respect to the inoculation with Bradyrhizobium alone and by 335 kg ha −1 with respect to the uninoculated control ). Finally, co-inoculation of soybean was found to raise the nodulation percentage by around 5% in Brazil

Conclusions
The ndings presented here show that the pre-culture combining BjE109 and AbAz39 before the inoculation to the soybean seeds has bene ts in plant nodulation and hence production, more than the single inoculation with BjE109 or AbAz39, or the immediate co-inoculation of both strains. The higher values for EPS production and rhizobium biomass obtained for this mix in its culture media mean that the rhizobium may be better able to survive on the seeds and establish symbiosis more successfully. This must be due to the bene cial effects on BjE109 growth and survival from the IAA produced by AbAz39. Consequently, the better bacteria performance, and its growth, have positive effects in planta with better growth and yield under controlled conditions, and also when are grown at dryland eld conditions. The success in the co-inoculation with B. japonicum and A. brasilense, partially depends on the time that the microorganisms are allowed to interact prior inoculation. Further studies must be done in the future to understand not only the best conditions for co-inoculation of B. japonicum and A. brasilense, but also with other bene cial bacteria in which the pre-culture must be one of the most relevant points.

Acknowledgments
To the Instituto de Investigaciones Agrobiotecnológicas (INIAB); Universidad Nacional de Río Cuarto (UNRC); Consejo Nacional de Investigaciones Cientí cas y Tecnológicas (CONICET), Fondo Nacional de Ciencia y Tecnología (FONCyT). FC is Researcher of CONICET at the UNRC. DT and GL are former Postdoc at the UNRC granted by CONICET. RM is Postdoc at the UNRC granted by CONICET. FD is a former PhD student at the UNRC granted by CONICET.  Matrix correlation between nodulation and plant production related parameters of soybean plants from seed uninoculated, inoculated with either BjE109 or AbAz39, individually, or the combination of both at T0 or T24 according to Pearson correlation. Red asterisks indicate signi cant correlations. P > 0.05 *; P > 0.01 **; P > 0.001 ***

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. Torresetal.2021TableS1.docx