Phaseolus vulgaris (common bean) seeds cultivar Lariat were surface sterilized by soaking in 4% sodium hypochlorite for 4 min followed by three washes with sterile water. Then, the seeds were swirled in 70% ethanol for 3 min and washed three times with sterile water.

Rice seeds cultivar M104 were surface sterilized in 4% sodium hypochlorite for 10 min followed by three washes with sterile water. Then, seeds were swirled in 70% ethanol for 10 min and extensively washed with sterile water.

Since we were interested in studying a possible malic acid present in the seed exudate and since malic acid is a polar molecule we collected seed exudates by soaking either common bean or rice seeds in MgCl₂ solution (0.2%) for 2, 24, and 48 h. Six seeds per 15 ml tubes (2 g seeds l^-1) with six replicates were used. Seed exudates were passed through 0.22 μm pore filter membranes and checked for eventual contamination by plating an aliquot of 50 μl in LB medium (10 g l^-1 tryptone, 5 g l^-1 yeast extract, 10 g l^-1 NaCl). Plates were incubated at 28°C for 24 h and then checked for bacterial growth.

ALB629 was grown overnight in LB medium and washed in MgCl₂ as described above and re-suspended to a final density of OD₆₀₀ = 0.8 in minimal medium (MSgg) [19.52 g l^-1 morpholinepropane sulfonic acid (MOPS) 100 mM (adjust pH to 7.0), 5 ml l^-1 glycerol, 5 g l^-1 glutamate, 0.41 g l^-1 MgCl₂, 0.87 g l^-1 potassium phosphate (pH 7), 0.05 g l^-1 tryptophan, 0.10 g l^-1 phenylalanine, 0.10 g l^-1 CaCl_2, 0.013 g l^-1 FeCl₃, 100 μl MnCl₂ (from stock of 50 mM), 100 μl thiamine (from stock of 2 mM), and 100 μl ZnCl₂ (from stock of 1 mM)] (Branda et al., 2001).

The obtained seed exudates were tested for its implication on the rhizobacteria biofilm formation, three concentrations of seed exudates [0.4, 1, and 5% v/v, or MgCl₂ solution (control)] were mixed with the bacterial suspension using the method of O’Toole and Kolter (1998). Samples of 100 μl of the diluted cells were aliquoted into sterile 96-well microtiter plates with eight wells per treatment as replicates. Plates were covered and incubated at 30°C without agitation for 48 h. Cells that had adhered to the well walls were treated with 0.1% crystal violet for 10–15 min at 25°C without agitation; the plates were drained of liquid via pipet, gently rinsed several times with water, and allowed to dry at room temperature. The dye that had stained the cells was solubilized in 200 μl of 95% (v/v) ethanol. Biofilm formation was quantified by measuring the optical density at 630 nm for each well using Wallac 1420 Manager plate reader.

To check the ALB629 growth with bean seed exudate presence, the bacteria were grown as described previously and re-suspended in MSgg to get a final density of OD₆₀₀ = 0.2. Then, bacterial suspensions were grown for 4 and 10 h at 28°C, 220 rpm with seed exudates at 1%. As a control, the bacteria were grown without seed exudates. At the referred time points, ALB629 suspensions were diluted and 10 μl was poured on Petri dishes with LB media. Plates were incubated at 28°C for 48 h and cells were counted for the number of colony forming units (CFU). Six replicates (n = 6) were used, three of which were biological replicates and two of which were technical replicates for treatment.

ALB629 was grown overnight in LB medium, washed in MgCl₂ as described above, and re-suspended in MgCl₂ solution to a final density of OD₆₀₀ = 0.8. Common bean seeds were treated with the bacterial solution for 2 h. In one of the treatments 1% of bean seed exudate, which was taken from 24 h of incubation, was added to the final bacterial solution. As a control, seeds were treated with water. Seeds were then fixed in 2% glutaraldehyde for 2 h, washed in filtered PBS buffer, and finally stained in SYTO^®13 Green Fluorescent Nucleic Acid Stain for 7 min. The seeds were analyzed in LSM 710 confocal microscopy.

For this experiment, 1% of seed exudate concentration from the 24 h seed exudate was used to check the expression of the following TasA (biofilm), EpsD (EPS), and Spo0A (sporulation) at 0, 4, and 10 h of bacterial growth. However, prior to gene expression, normalization was performed with the housekeeping gene RecA, which has been used as a reference gene in various studies involving bacteria (Rivas et al., 2005; McMillan and Pereg, 2014; Da Silva et al., 2016). The bacterium was grown overnight in LB medium and washed in MgCl₂ as described above and re-suspended in MSgg medium to a final density of OD₆₀₀ = 0.2A. Bean seed exudate at 1% was added into the bacterial suspension and grown for 0, 4, or 10 h at 28°C 220 rpm. As a control, the bacteria were grown without seed exudates. Total RNA was isolated using the Macherey-Nagel^TM Kit from 1.5 ml bacterial suspension, following the manufacturer’s instructions. For 0 h, RNA extraction was performed from the bacterial suspension samples at the first time point of bacterial growth. For RT-PCR, cDNA was synthesized using M-MuLV reverse transcriptase (New England Biolabs) from 500 ng of RNA according to the Applied Biosystems protocol, followed by PCR amplification using DyNAzyme II DNA polymerase (DyNAzyme). The gene-specific primers for the genes RecA, TasA, EpsD, and Spo0A are listed in Table 1 (Qiu et al., 2014). Three experimental replicates for each treatment were used in two different experiments. The band intensity was taken from agarose gel images and quantified by Image J 1.47v¹ (National Institute of Health, United States).

Identification of malic acid in the exudates from non-inoculated seeds was carried out using high-performance liquid chromatography (HPLC)-MS Q Exactive, Thermo Scientific, H-ESI. The device was operating in a full negative mode scan at a resolution setting of 70,000, with a spray voltage of spray 4000 V, sheath gas flow rate set to 12 arb. units with three auxiliary arb. units, capillary temperature of 300°C, and S-lens of 60 arb. units. The sample at 600 ppm was diluted in methanol with 0.1% formic acid. The standard solution used was prepared by malic acid Synth PA and Milli-Q water. Data were obtained by Xcalibur software.

To study the ALB629 colonization of 22-day-old bean seedlings submitted to drought stress, a spontaneous rifampicin/nalidixic acid (ALB629^rif-nal) was selected from a B. amyloliquefaciens strain ALB629 based on Medeiros et al. (2009). To select the ALB629 with the selection marker, increasing amendments of both antibiotics were added to the LB medium up to 100 ppm at each bacterial culture. The spontaneous mutant strain was previously compared to the wild type for its growth and biological control efficacy and the mutation did not impact such variables (Martins et al., 2014).

In order to find out the capability of ALB629 in sustaining plant growth under abiotic stress, we selected the best time point regarding seed treatment in the previous assay to proceed with the seed treatment. A treatment using malic acid (0.5%) (Kleijn et al., 2010) was used in order to find out its effect on biological seed treatment. For both treatments, pH was brought to 5.6–5.8. Bacterial suspensions were submitted to the seed treatment for 2 h from a bacterial culture of 2-day-old as described before. As a control water and water + malic acid was used in the drought assays.

Pots filled with 0.3 l sterilized ultralite top soil and with one seedling per pot were kept under growth chamber conditions: 200 μE m^-2 s^-1 12 h of photoperiod of 23°C/15°C (day/night), and relative humidity of 55/70% (day/night) according to Shevyakova et al. (2013) for abiotic stress.

Seedling emergence was recorded daily and used to calculate the speed emergence index (SEI) as described by Teixeira and Machado (2003). When the first fully expanded leaves appeared, irrigation was suspended for 2 weeks, the time point in which plants started wilting. Then, plants were collected and checked growth promotion parameters: fresh and dry weights, fw/dw ratio, stem size, measurements of leaf width and length. Additionally, root tissue was washed twice with SDW, weighed, and transferred to 2 ml tubes containing 900 μl of SDW. The tissue was crushed, vortexed, and serially diluted to 10^-5 and then plated by spreading on LB medium with 100 ppm of rifampicin/nalidixic acid. Plates were incubated at 28°C for 48 h and data were transformed to log₁₀ CFU g of fresh tissue weight^-1.

All experiments were carried out randomized complete block with 8 and 10 replicates, respectively, for the in vitro and in vivo (drought stress) tests. Data were submitted to one-way analysis of variance (ANOVA) and for significant means Scott–Knott test, Tukey’s multiple range tests, or Student’s t-test (p < 0.05) were applied. For all analyses, the assumptions of normality of variance were checked by the Shapiro–Wilk test and no transformation was necessary.

The effects of seed exudates of common bean on ALB629 growth and biofilm formation were tested. Different concentrations (0.4, 1, and 5% v v^-1) of seed exudates collected were directly tested on biofilm formation in ALB629. Seed exudates were incubated and streaked on LB plates to negate contamination. Common bean seed exudates at the concentrations of 1 and 5% increased biofilm formation by ALB629 regardless of the time point (2, 24, and 48 h) that they were collected when compared to the negative control (ALB629 without seed exudate) (Figure 1). To evaluate the specificity of the seed exudation to trigger biofilm formation in ALB629, we checked the seed exudates from a non-legume plant (rice) on biofilm formation in ALB639. Our data show that rice seed exudate also triggers biofilm formation in ALB629 similar to bean seed exudates, suggesting the non-specificity of seed exudates to induce biofilm formation (Supplementary Figure 1).

To evaluate the biofilm, the bacterial colony formation units (CFU) in ALB629 exposed to seed exudates were also evaluated. There was a higher growth for ALB629 in the presence of common bean seed exudates for post 4 (p = 0.0001) and 10 h (p = 0.0002) of growth (Figure 2A). To validate the biofilm formation and bacterial association triggered by seed exudates, confocal microscopy was utilized. Confocal micrographs showed that addition of 1% seed exudates triggered bacterial association on seed surface (Figures 2B–D).

Seed exudates from common bean up-regulated biofilm operons in ALB629 in a TasA-and EpsD-dependent manner at 4 h by about ∼two and sixfold, respectively, higher than the untreated control (Figures 3A,B,D). On the other hand, in the absence of added seed exudate, ALB629 biofilm operons TasA and EpsD increased their expressions only at 10 h by about 1.5- and 4-fold, respectively. However, the difference between the treatments regarding Spo0A was not statistically significant regardless of the time point tested (Figure 3C).

When the seed exudates were analyzed by HPLC-MS a molecule was found at the relative abundance of 133.01308 (Figure 4C) and with a fragmentation pattern of 115.00246 and 71.01253 (Figure 4D), which showed the same relative abundance as malic acid PA 133.01314 (Figure 4A) and its fragmentation pattern 115.00247 and 71.01253 (Figure 4B), confirming the presence of malic acid as a major organic acid in the seed exudates of common bean.

When subjected to drought trials for 2 weeks, seedlings from seeds treated with ALB629^rif-nal and malic acid showed higher colonization by the rhizobacteria compared to the lone ALB29 treatments (Figure 5).

In addition, the seed treatment with ALB629 and malic acid increased the seedling emergence index (Figure 6C), leaf length (Figure 6D), plant fresh weight, and plant water holding capacity compared to the water control or water associated with malic acid (Figures 5B, 6B,E). Water holding capacity was represented as a ratio between fresh weight and dry weight. Moreover, ALB629 itself could increase the seedling stem size (Figure 6A). When the time point used for the biological seed treatment was tested, the optimal result regarding the number of bacterial cells recovered from the seeds was 2 h of seed immersion using a 2-day-old bacterial suspension (Supplementary Figures 2a,b).