Introduction

Agricultural activities need a large amount of commercial nitrogen fertilizers to effectively promote crop growth, but it can also result in soil compaction and water eutrophication (Carpenter et al. 1998). Development of more environmentally benign nitrogen fertilizers such as biological N2 fixation would therefore be useful for the decremental application rate of commercial nitrogen fertilizers (Galloway et al. 2004). A biogenous azotobacter and combined nitrogen-fixing bacteria can colonize in rhizosphere of non-legume plants during the agricultural activities (Carvalho et al. 2016); thus, it is a more effective method to protect environmental pollution caused by nitrogen fertilizers when compared with that of symbiotic nitrogen-fixing bacteria.

It is widely accepted that mass number of non-symbiotic nitrogen-fixing bacteria could be utilized to promote plants’ growth after they were effectively colonized in rhizosphere (Fibach-Paldi et al. 2012; Ke et al. 2019; Maheshwari 2011; Puri et al. 2016). The colonization processes of these bacteria were primarily relied on the attraction effects of root exudates (el Zahar Haichar et al. 2014), such as sugars, organic acids, or some secondary metabolites (Huang et al. 2014). The migration of these bacteria to root surface (could be achieved by the bacterial effects of chemotaxis and swarming motility) and the formation of biofilm are two necessary processes for their early and persistent colonization, respectively (Allard-Massicotte et al. 2016; Beauregard et al. 2013; Gao et al. 2016). The chemotaxis responses of bacteria are referred to the flagella rotation to swim in a favorable direction attracted by the chemical compound compositions and concentrations in environment (Eisenbach 1996). Swarming motility is a rapid multicellular movement of bacteria with surviving the semi-solid surface (agar > 0.3%) driven by flagella rotation (Kearns 2010). It has been reported that a larger number of components in root exudates could act as the signaling molecules for the colonization of strains in their rhizosphere. For instance, carbohydrates, dicarboxylates, and l-amino acids in root exudates of corn, Arabidopsis thaliana, and tomato can attract the colonization of Bacillus velezensis S3-1, Azospirillum brasilense, and Pseudomonas fluorescens Pf0-1 in their rhizosphere, respectively (Jin et al. 2019; Oku et al. 2012; Singh et al. 2019).

At present, ~ 92 strains, distributed in bacteria and archaea, are known with a nitrogen fixation capacity (Dos Santos et al. 2012; McRose et al. 2017). In total, ~ 20 high-efficiency non-symbiotic nitrogen-fixing strains are preserved in China. We hypothesized that (i) the nitrogen fixation capacity of 9 representative non-symbiotic nitrogen-fixing bacteria was different even with a similar medium culture; (ii) the distinct attraction ability of crops on strains resulted from different components and concentrations of root exudates; (iii) signal molecules may be found in root exudates of the crop with a high attraction ability for nitrogen-fixing bacteria; and (iv) the efficient colonization of preferred stain will help to crop yield essential for environmental protect by reducing the application rate of commercial nitrogen fertilizers.

Materials and methods

Bacterial strains

Nine non-symbiotic nitrogen-fixing bacterial strains were purchased from China Center of Industrial Culture Collection (CICC), China General Microbiological Culture Collection Center (CGMCC), and Agricultural Culture Collection of China (ACCC), and the detailed information and enrichment medium of these strains are presented in Table S1.

Nitrogen fixation capacity evaluation

Each strain suspension (~ 107 colony-forming unit (CFU) mL−1 fluid prepared with an enrichment medium, Table S1) was respectively inoculated (1% v/v) in a conical flask with sterile Ashby nitrogen-free liquid medium (Table S1) and 10% soil solution (pH, 5.52; total nitrogen, 25.38 mg kg−1; ammonia nitrogen, 1.98 mg kg−1; and nitrate nitrogen, 1.26 mg kg−1). Ammonia nitrogen concentrations of the supernatant medium in days 5 and 7 were determined as indicated in the previous study (Bergersen 1980). Then, strain with a higher ammonium excretion (> 1.0 mg L−1) ability was selected to determine its nitrogenase activity in Ashby’s nitrogen-free medium (semi-solid medium, 0.5% agar) (Hardy et al. 1973).

Strain Ac63 chemotaxis with root exudates

Sterilized seedlings of five common vegetables, including Amaranthus tricolor (AT), Willow amaranth (WA), Chrysanthemum coronarium (CC), cabbage, and lettuce, were cultured in sterile Hoagland medium (discharged every 2 days) and were illuminated with the cool white fluorescent light with a daylight cycle of 16 h (28 ± 2 °C) for 4 weeks. Crops were transferred in sterile water for 2 days to collect root exudates after their roots were washed with deionized water for three times; then, the solution of root exudates was filtered with a 0.22-μm membrane and then stored at − 20 °C (Haier BCD-625WDGEU1, China). Total organic carbon (TOC) concentrations of root exudates were measured using a TOC analyzer (Shimadzu TOC-LCPH, Japan). The chemotaxis effects of Ac63 in response to root exudates were carried out according to the previously described method (Ling et al. 2011). Briefly, a 200-μL pipette tip containing 100 μL of Ac63 suspension (~ 107 CFU mL−1 fluid prepared with an Azotobacter medium, Table S1) was attached to 1-mL tuberculin syringe with a 4-cm 25-gauge needle for 2 h. The syringes contained 200 μL root exudates (TOC concentrations, 0.1, 0.5, 1.0, 3.0, 5.0, and 10.0 mg L−1) or sterile water (as a control groups). Plate counting method was used to evaluate the influences of different concentrations of root exudates of the five crops on their chemotaxis performances of strain Ac63. Components in root exudates of the most differential chemotaxis effects of two crops (AT and lettuce) were identified with a gas chromatography–mass spectrometry (GC-MS) (Shimadzu QP2010 Ultra, Japan) as indicated with the previous derivatization processes (Lisec et al. 2006) and operating conditions (Guo et al. 2018). One-microliter sample was injected into the GC-MS with a splitless mode. Chromatographic separation was set with a DB-5 integrated guard column (30 m × 0.25 mm ID, 0.25-mm film thickness, Agilent, USA). Carrier gas flow rate was 1 mL min−1. Column temperature was initially maintained at 70 °C for 1 min, followed by a 1 °C min−1 ramp to 76 °C, then increased to 330 °C with a rate of 5 °C min−1, and a final hold for 6 min at 330 °C. Mass spectra were recorded at 20 scans per second with a scanning range of 50–1000 m/z. The concentration of glutamic acid in root exudates of five crops was measured with the glutamic acid analysis kit (A074 Nanjing Jiancheng, China).

Root exudates effected on chemotaxis, swarming, in vitro, and biofilm formation of Ac63

The different components in root exudates of AT and lettuce were respectively selected for a chemotaxis assay with Ac63 as the description in the previous study (Ling et al. 2011). Each component with an excellent chemotactic effect were selected for swarming experiments, and then in vitro assay was performed with the component with an obvious swarming effect on Ac63 as indicated in the previous research (Ling et al. 2011). Component, which shows a significant recruiting ability for Ac63 in chemotaxis, swarming, and in vitro experiments, was considered as the colonization signal molecule of this strain. Sterile water effected on this strain was used as the control treatments.

The qualitative assessment of Ac63 biofilm formation on the roots’ surface was visualized with the laser confocal microscopy (Zeiss LSM880, Germany) (Rudrappa et al. 2008). To quantify the biofilm formation as affected by root exudates (AT and lettuce) and single signal molecule in a MSNc liquid medium (potassium phosphate buffer 5.0 mM, Mops (pH 7) 0.1 M, MgCl2 2.0 mM, MnCl2 0.05 mM, ZnCl2 1 μM, thiamine 2.0 μM, CaCl2 700 μM, NH4Cl 0.2% (w/v), cellobiose 0.5% (w/v)) (Beauregard et al. 2013), the modified crystal violet (CV) assay method was used for evaluation of biofilm formation based on previous study (O’Toole 2011; Wang et al. 2017). MSNc liquid medium (600 μL) and root exudates (20 μL) or signal molecules (20 μL) were added into a 48-well plate, respectively. The final concentrations of root exudates, and signal molecule, were 10 mg L−1 and 10, 25, and 50 μM, respectively. Sterile water effected on this strain was used as the control treatments. A 40 μL Ac63 bacterial suspension (~ 107 CFU mL−1 fluid prepared with an Azotobacter medium, Table S1) was inoculated. The 48-well plate was placed at 28 °C for 12, 24, and 48 h, and biomass of biofilm formation was quantitatively determined.

Soil pot trial with Ac63

Pot experiments were performed in a greenhouse of Jinan University, Guangzhou, in May 2018 (daily temperature, 27 to 33 °C). Vegetables (i.e., AT/lettuce) with/without the secretion of signal molecule (l-glutamic acid) were selected for the comparative tests, and each vegetable was set up with four treatments: without nitrogen fertilizer and uninoculated (C), without nitrogen fertilizer and inoculated with Ac63 (B), nitrogen fertilizer and uninoculated (N), and nitrogen fertilizer and inoculated with Ac63 (NB). Each pot contained with 1.5 kg unsterilized soil mixed with fertilizers: potassium 50 mg kg−1 (KCl), phosphate 100 mg kg−1 (Ca(H2PO4)2·H2O), and nitrogen (N and NB treatments) 20 mg kg−1 (NH4NO3). We tried to sterilize the seeds of AT and lettuce on 75% ethanol for 30 s; then, we transferred them into 2% sodium hypochlorite for 3 min, and then washed 4 times with sterile water. Then, seeds were transferred into seedling pots consisted of unsterilized soil mixed with potassium 50 mg kg−1 (KCl) and phosphate 100 mg kg−1 (Ca(H2PO4)2·H2O). The culture conditions were illuminated under cool white fluorescent light with a 16-h daylight cycle (28 ± 2 °C) for 14 days. After seeding (14 days), seedlings were transferred into soil and inoculated with 5 mL Ac63 suspension (CFU mL−1 > 109) at day 5, 10, 15, and 30. Meanwhile, 5 mL sterile water was added into soil as the control treatment. All crops were harvested after a 45-day cultivation. The chlorophyll concentrations in leaf were determined with a spectrophotometry method at 652 nm (Arnon 1949). Approximately 3.0 g rhizosphere soils of each pot were obtained from samples adhering to plant roots. Total nitrogen in roots and leaves and available nitrogen/ammonia nitrogen in rhizosphere soils were determined with the methods of Horneck and Miller (Horneck and Miller 1998) and ISO/TS 14256–1:2003, respectively. Samples for gene quantification was quickly frozen in liquid nitrogen and then stored at − 80 °C (Haier DW-86L728, China).

For root colonization assay, Ac63-gfp was successfully labeled with the plasmid pMP2444 (named as Ac63-gfp) (Stuurman et al. 2000) by previously described electroporation method (Xue et al. 1999). Then, it was tested for the colonization ability in roots after 45-day cultivation with a similar pot condition as mentioned above. The roots were washed with sterile water (three times) to remove root-attached soil and then observed by the laser confocal microscope (Zeiss LSM710, Germany) (Chen et al. 2013).

Root total DNA and rhizosphere soil DNA were extracted using a DNeasy Plant Mini Kit (Qiagen, Germany) and DNeasy PowerSoil Kit (Qiagen, Germany), respectively. The unsterilized soils were intended for pot experiments as a negative control for the DNA extraction. In accordance with the whole genome of A. chroococcum (txid1328314), the specific primer (nifH-F 5′ CGAATACGATCCGACCGCCAAG 3′, nifH-R 5′ TTCCATAACGCCGAACTCCATCAG 3′) of nitrogen-fixing gene nifH in Ac63 was designed, and real-time quantitative PCR (qPCR) was carried out with a real-time PCR detection system (CFX96 BIORAD, USA). The reaction system includes 1 μL (50 ng) DNA template, 1 μL forward primers (10 μM), 1 μL reverse primers (10 μM), 10 μL 2× QuantiNova SYBR Green PCR Master Mix (Qiagen, Germany) and DNase-free water 7 μL. The reaction conditions follow with the lists: denaturized at 95 °C for 2 min, and 40 cycles include 95 °C for 5 s, then 57 °C for 10 s and 72 °C for 5 s, and the dissolution curve was set to 95 °C for 10 s, and 65 to 95 °C for 0.5 °C plate read once. T plasmid was serially diluted into 10-folds and then subjected to qPCR measurement simultaneously, and the qPCR reactions were carried out separately, in triplicate. The copy number of each sample was calculated in accordance with the previous method (Thonar et al. 2012). Then, CFU was generated with the gene copy number according to referred method (Mosimann et al. 2017).

Data analysis

Relative chemotactic effects of Ac63 in response to root exudates or each component:

$$ {\mathrm{CFU}}_{\mathrm{rcr}}=\frac{{\mathrm{CFU}}_{\mathrm{exu}}}{{\mathrm{CFU}}_{\mathrm{ck}}} $$
(1)

where CFUrcr, CFUexu, and CFUck indicate a value for the evaluation of relative chemotactic response, the bacterial colony number in a syringe needle with root exudates or each component, and the bacterial colony number in a syringe needle with sterile water, respectively.

Relative swarming motility of Ac63 to each component

$$ {L}_{\mathrm{rsr}}=\frac{L_{\mathrm{exu}}}{L_{\mathrm{ck}}} $$
(2)

where Lrsr, Lexu, and Lck represent the relative swarming mobility of bacteria, the diffusion diameter of Ac63 in semi-solid plates with component of root exudates, and the diffusion diameter of Ac63 in semi-solid plates with sterile water, respectively.

For qPCR assay, CFX manager 3.1 (Bio-Rad, USA) was used to calculate the qPCR quantification data. Ensuring the amplification efficiency (E) of qPCR is between 90 and 110%, R2 > 0.98, the dissolution curve is single peak and coincides with the positive control, and no template control (NTC) fluorescence intensity is lower than threshold, and Ct values were between 10 and 35 cycles; the qPCR quantification data (nifH abundance) was converted to the equivalent number into copies per gram.

Three parallels for each treatment were set up separately for all treatments. One-way ANOVA was performed in SPSS 22.0 software (IBM, USA), and Duncan’s multiple range tests with P < 0.05 or P < 0.01 were considered the significant levels.

Results

Nitrogen fixation capacity of strains

Strain number 6 (Azotobacter chroococcum CICC22663, Ac63) showed a highest ammonium excretion ability (Fig. S1), and its nitrogenase activity reached to 77.1 nmol mg−1 h−1. Compared with the previously described non-symbiotic nitrogen-fixing bacteria (Chen et al. 2018; Goswami et al. 2016), Ac63 has a stronger ability in N2 fixation. However, there is no report about the stable colonization performance of this strain on its potential host plant.

Strain Ac63 chemotaxis with root exudates

Chemotactic response of strain Ac63 to root exudates of crops, except for lettuce, increased with the TOC concentrations of root secretions (Fig. 1). When these concentrations reached to 10 mg L−1, the relative chemotactic response (CFUrcr) value of AT, WA, CC, cabbage, and lettuce was 3.0, 1.7, 1.4, 1.3, and 0.9, respectively. Compared with the control treatment (water), a strongest chemotactic response of Ac63 was observed in AT treatment (P < 0.05), but no chemotactic effect was observed in root exudates of lettuce treatment (P > 0.05) (Fig. 1).

Fig. 1
figure 1

Chemotactic responses of Ac63 to root exudates of five crops. Relative chemotactic response is the value of bacterial number in root exudates/control (water) treatments. Lowercase letters represent the significant differences in 6 treatments according to Duncan’s test (P < 0.05)

Full size image

Components of root exudates and their effects on strain Ac63

As shown in Fig. 2, in total, root exudates of AT (highest CFUrcr) consisted of 9 amino acids, 14 organic acids, 3 sugars, 3 alcohols, and 1 amine, and for lettuce (lowest CFUrcr), root exudates consisted of 7 amino acids, 14 organic acids, 1 alcohol, and 1 amine (Table S2). Chemotactic effects of strain Ac63 to each component in root exudates (AT and lettuce) are presented with Fig. 3. The colony number of Ac63 attracted by l-glutamic acid, acetic acid, and malic acid was significantly higher (i.e., by 289%, 13%, and 7%, respectively) than that of the control treatment (water) (P < 0.05). However, the average number of Ac63 colonies attracted by the remaining substances was lower than that of the control treatment (water), and 15 compounds showed an inhibitory effect (P < 0.05) (Fig. 3). The application of l-glutamic, acetic, and malic acids for bacteria swarming experiments showed that the diffusion diameter of acetic and malic acids on the surface of Ac63 semi-solid medium did not result in a significant diffusion promotion (P > 0.05), whereas the diameter of Ac63 diffusion of l-glutamic acid increased by 20% (P < 0.05) when compared with that of control treatment (water). Furthermore, the results of in vitro experiments suggested that after adding l-glutamic acid into crop’s root, the average number of Ac63 colonies attracted by root was 1.51 × 107 CFU groot−1, which was 738.0% higher than that of the control treatment (water) (P < 0.01) (Fig. 4a). Laser confocal microscopy indicated that the crop’s roots with l-glutamic acid secretion function formed a dense biofilm-like structure with a strong green fluorescence (Fig. 4b), whereas control treatment (water) presented with an individual and relatively weak fluorescence (Fig. 4c). Chemotaxis, swarming, and in vitro experiments demonstrated that l-glutamic acid was the signal molecule for Ac63 for its colonization in crops’ roots (Figs. 3a, d and 4a).

Fig. 2
figure 2

The components chromatogram of root exudates collected from Amaranthus tricolor and lettuce

Full size image
Fig. 3
figure 3

Relative chemotactic and swarming responses of Ac63 towards carbon source components (30 μM) in root exudates of Amaranthus tricolor and lettuce. The asterisk indicates a significant difference from the control condition (water, 1.00 ± 0.15) according to Duncan’s test (P < 0.05), and red asterisk and green asterisk indicate a relative chemotaxis response index > 1 or < 1, respectively

Full size image
Fig. 4
figure 4

Root colonization and biofilm formation of Ac63 with l-glutamic acid and water treatments: the population levels of Ac63 recruited by roots of Amaranthus tricolor after the addition of l-glutamic acid (a) (Duncan’s test, **P < 0.01); biofilm formed by the addition of l-glutamic acid (30 μM, 10 μL) (b), or water (c). Green indicates individual Ac63 visualized by staining with SYTO9; white arrows indicate biofilm-like structure formed by Ac63

Full size image

l-Glutamic acid concentration in root exudates of these five crops followed the lists of AT > WA > cabbage > CC > lettuce. When TOC concentrations of root exudates were 10 mg L−1, the corresponding glutamic acid concentrations were 38.0 ± 3.2, 22.4 ± 2.1, 14.0 ± 3.2, 11.9 ± 2.6, and 0.4 ± 0.3 μM (Fig. 5), respectively. In summary, l-glutamic acid concentration in root exudates of these five crops was significantly positively correlated with the chemotactic effects of root exudates on Ac63 (P < 0.01) (Fig. 5).

Fig. 5
figure 5

Correlation analysis of glutamate concentrations in root exudates (TOC, 10 mg L−1) with chemotactic responses of Ac63. Two asterisks indicate the significant correlation at P = 0.01 level (both sides, Pearson correlation coefficient)

Full size image

Effects of root exudates or l-glutamic acid on biofilm formation of strain Ac63

Figure 6 shows the effects of root exudates collected from crops of AT and lettuce on biofilm formation of strain Ac63. Biomass of Ac63 biofilm formed in MSNc medium with root exudates of AT was ~ 8 times higher than that of control treatment (water) (P < 0.05), but its formation in lettuce treatment was similar with that of control treatment (water) (P > 0.05). In addition, the biofilm formation biomass of Ac63 in MSNc medium with different l-glutamic acid concentrations was 3–34 times higher than that of control treatment (water).

Fig. 6
figure 6

Ac63 biofilm formed in MSNc medium with water (CK), 10, 25, and 50 μM l-glutamic acid (Glu), Amaranthus tricolor and lettuce root exudates (TOC = 10 mg L−1), respectively

Full size image

Effects of Ac63 on Amaranthus tricolor and lettuce in soil pot

As shown in Table 1, without nitrogen fertilizer, the physiological indicators, including plant heights, chlorophyll concentrations, and total nitrogen concentrations in roots and shoots, and soil bioavailable nitrogen concentrations of AT and lettuce treatments inoculated with Ac63 were promoted significantly when compared with that of the corresponding non-inoculated treatments (P < 0.05). Except for total nitrogen concentrations in roots, the enhancements of other indicator rates of AT were higher than those of lettuce. With nitrogen fertilizer, the above physiological indicators and soil bioavailable nitrogen concentrations in AT treatment inoculated with Ac63 were significantly higher than that of non-inoculated treatments (P < 0.05). Except for total nitrogen in roots and ammonia nitrogen in rhizosphere soil, the enhancements of other parameter rates of AT were greater than those of lettuce (Table 1).

Table 1 Effects of inoculation with Ac63 on bio-characteristic parameters and soil nitrogen of Amaranthus tricolor (AT) and lettuce
Full size table

Ac63 showed a relatively excellent colonizing ability in roots of AT, rather than that of lettuce (Fig. 7(I)); the abundances of nitrogen-fixing gene nifH in roots of AT with and without nitrogen fertilizer treatments after strain Ac63 inoculation were 5.0 × 107 and 3.5 × 107, which represents the corresponding colony number of Ac63 in these roots were 8.5 × 107 and 6.0 × 107 CFU g−1 root, respectively. However, the abundance of nitrogen-fixing gene nifH and the colony number of Ac63 in the roots of lettuce were lower than detection limit after Ac63 inoculation (Fig. 7(II)). Meanwhile, the abundance of nitrogen-fixing gene nifH of the unsterilized soils intended for pot experiments (negative control for the DNA extraction) was also lower than that of the detection limit; the number of Ac63 colonies in the rhizosphere soil of AT was significantly higher than that of lettuce (P < 0.05) (Fig. 7(III)).

Fig. 7
figure 7

Colonization of Ac63 in roots and rhizosphere soil of Amaranthus tricolor (AT) and lettuce. Visualization of root colonization by strain Ac63-gfp under a laser confocal microscope (scale bar = 200 μm) (I); nifH abundance and colonization number of Ac63 in roots (II) and rhizosphere soil (III) of AT and lettuce. Four treatments: without nitrogen fertilizer and uninoculated (C), without nitrogen fertilizer and inoculated with Ac63 (B), nitrogen fertilizer and uninoculated (N), nitrogen fertilizer and inoculated with Ac63 (NB). Different lowercase letters represent significant differences among the treatments (including AT and lettuce) (Duncan’s test, P < 0.05), no letter above indicates the nifH abundance is below the detection limit. Red/white arrows indicate individual Ac63 and biofilm-like structure formed by Ac63, respectively

Full size image

Discussion

The results demonstrated l-glutamic acid is a signal molecule for attractions of high-efficiency nitrogen-fixing strain Ac63. Pseudomonas sojae zoospores (Suo et al. 2016), Pseudomonas chlororaphis PCL1606 (Polonio et al. 2017), nitrogen-fixing bacteria A. chroococcum X-50 (Gupta Sood 2003), and rhizosphere-promoting bacteria Bacillus amyloliquefaciens SQR9 (Feng et al. 2018) also could be attracted by l-glutamic acid in a chemotactic experiment as indicated in the previous studies.

Chemotactic response of bacteria to a specific component includes two steps: (i) the signaling molecule binds to its corresponding methyl-accepting chemotaxis proteins (MCPs) on the surface of bacteria, and it will promote the phosphorylation of CheA kinase to transfer phosphate group to CheY protein; (ii) CheY protein acts as motor proteins to induce flagellum move and turn, resulting in the bacteria tumbling and producing chemotaxis (Hazelbauer 2012). In general, ~ 14 MCP genes are found in bacterial genome (Lacal et al. 2010), and different MCPs are the receptor proteins for their specific signal molecules, respectively. For instance, CheA1 and CheA2, McpC, and Tip12 are the receptors of Pseudomonas putida KT2440, B. subtilis, and Campylobacter jejuni for the recognition of maize’s root exudates, glutamate and glutamate, respectively (Glekas et al. 2012; López-Farfán et al. 2019; Lübke et al. 2018). Unfortunately, there are few reports about the information of MCPs in A. chroococcum and their functional verification. We performed a sequence comparison analysis based on the protein annotation results in the NCBI database. The query cover of amino acid sequences between the MCP of A. chroococcum (WP_089168023.1) and McpC and Tip12 are 64% and 29%, respectively. Therefore, the homolog of McpC may mediate the chemotactic response of Ac63 to l-glutamic acid. In addition, bacterial chemotaxis is also related to the concentrations of signal molecules in their growth environment (Martín-Mora et al. 2016). Although the in situ concentration of root exudates in soil environment is hard to determine (Feng et al. 2018), quantitative analysis of root exudate components based on the GC-MS and then using of each component with different concentration rates could efficiently simulate the chemotactic effect of a specific component on bacteria than that of a certain concentration (30 μM).

The results of chemotactic responses to root exudates were based on the activities of all chemoreceptors (Feng et al. 2019). Six MCPs of Ralstonia pseudosolanacearum Ps29, including McpA, McpT, Mcp09, McpM, Mcp15, and Mcp19, presented with a negative chemotaxis response to ethanol (Oku et al. 2017). The negative chemotactic effects of strain Ac63 were found with the components of isoleucine, glycine, aspartic, tyrosine, threonine, lactic, hydroxypropionic, citric, oxalic, 4-aminobutyric, 3-hydroxybutyric, glyceric, glycolic, and hexadecenoic acids in root exudates of AT, whereas a positive chemotactic effect was observed with mixture compounds of root exudates. These results demonstrated that l-glutamic acid could act as a really strong attractant for the attraction of Ac63 in rhizosphere. Feng reported that malic acid could be used as an attractant for B. velezensis SQR9 (Feng et al. 2019), inducing a significantly positive chemotaxis for this strain even at a relatively low concentration (10 μM). Acetic and malic acids in root exudates of lettuce presented a weak positive chemotaxis effect, but the lack of strong attractant l-glutamic acid and the existence of many negative chemotactic components resulted in no significant Ac63 chemotaxis in its root.

Swarming is also an important factor for the colonization ability of bacteria in rhizosphere (Gao et al. 2016). Nutrient-rich and soft substrates in rhizosphere soils may also promote the swarming motility and bacterial colonization. However, swarming is determined by the collision between bacterial cells instead of tumbling and running; thus, swarming effects of bacteria are unlikely to be driven by chemotaxis (Darnton et al. 2010). During the colonization of B. subtilis SWR01 onto tomato roots, the swarming motility rather than chemotactic response plays a major role (Gao et al. 2016). Results indicated that l-glutamic acid could also significantly increase the swarming motility of Ac63, which is consistent with glutamate inducing the swarming motility of P. aeruginosa (Köhler et al. 2000). Glutamate affects the activity of protein DipA and then regulates c-di-GMP amount in intracellular (Mattingly et al. 2018). Different concentrations of c-di-GMP can induce or inhibit the swarming motility of P. aeruginosa PA14 by regulating on the flagellar motor stator proteins MotCD and MotAB, respectively (Kuchma et al. 2015).

Biofilm formation is one of the marks of bacterial colonization in niche (Ramey et al. 2004). Bacteria form a biofilm matrix through extracellular proteins, EPS, and eDNA, thereby joining free cells into a membrane (Chen et al. 2015). The present study showed that l-glutamic acid can effectively stimulate Ac63 to form biofilm in roots and medium culture. Glutamate also promotes B. subtilis (Yu et al. 2016) and Pseudomonas stutzeri (Wang et al. 2017) to form a biofilm as indicated in the previous researches. Related studies showed that acetic acid affected on Ywbl protein of B. subtilis and promoted the biofilm formation of B. subtilis through holin-antiholin-like protein (YwbHG, YsbAB, and YxaKC) and eDNA efflux (Chen et al. 2015). Citric acid and malic acid affected on B. amyloliquefaciens SQR9 genes epsD and tapA, resulting in the secretion of EPS/extracellular protein and the formation of bacterial biofilm (Zhang et al. 2015). Recent study indicated that gene gltA encoded glutamate synthetase (GOGAT) with a function to convert glutamine into glutamate, and the increasing expression of this gene could significantly promote the biofilm formation of Enterococcus faecalis (Yu 2019). Thus, this gene possibly promoted biofilm formation by synthesizing glutamic acid. However, the specific mechanism on l-glutamic acid to promote the bacterial biofilm formation is unclear. Interestingly, we found that the biofilm formation of Ac63 was decreased with the concentration of l-glutamic acid in the medium from 10 to 25 μM or 50 μM. Similar studies have showed that the low concentrations (0.01 mM, 0.1 mM) of threonine could significantly increase the biofilm formation of Bacillus cereus AR156 than that of a high concentration (10 mM) (Wang et al. 2019).

Furthermore, results of soil pot experiment indicated that l-glutamic acid can strongly attract strain Ac63 to colonize in the unsterilized soil. The colonization number of Ac63 in roots (8.5 × 107 and 6.0 × 107 CFU g−1 root) is close to the previous reports (Ansari and Ahmad 2018; Ke et al. 2019; Puri et al. 2016). This study also found that the amount of Ac63 in rhizosphere soils of two crops after Ac63 inoculation could reach up to ~ 106 CFU g−1 soil, which might result from the short interval, mass inoculation, and well survivability of Ac63 in this study. In general, only a few of the rhizosphere inoculum infiltrates the micropores of soil matrix which could directly attach to the root surface of crops. Therefore, with the attraction of signal molecule (l-glutamic acid), mass inoculation of strains Ac63 attracted and colonized the roots of AT; however, lettuce was not. Rhizosphere soils have a high content of these bacteria, which slightly promotes the growth of lettuce. The amount of Ac63 in roots of AT decreased significantly after the application of nitrogen fertilizer, and the growth indicators of plant also decreased apparently. These results indicated that the application of nitrogen fertilizer affected the colonization number of Ac63 in AT root and its plant-promoting effects.

Conclusion

l-Glutamic acid is a signal molecule for the colonization of highly efficient nitrogen-fixing strain Ac63. The root system of Amaranthus tricolor can secrete a high concentration of l-glutamic acid, inducing Ac63 colonization in its roots and promoting its growth, whereas lettuce cannot secrete l-glutamic acid, and no corresponding effects were observed in this study. This means that crops which can secrete l-glutamic acid may promote the colonization of strain Ac63, exert a good growth-promoting effect, and decrease the using rate of commercial nitrogen fertilizers. These findings can enhance the understanding of mutually beneficial symbiotic relationships between non-symbiotic nitrogen-fixing bacteria and crops, which provides an important information for the application of high-efficiency nitrogen-fixing strains during the agricultural activities.