All bacterial strains, plasmids, and primers used in this study are shown in Tables 1, 2. A. caulinodans ORS571 was grown in TY or L3 media with ampicillin and nalidixic acid at 37°C. Escherichia coli DH5α strain was grown in LB medium at 37°C.

Antibiotics were used as follows: nalidixic acid (Nal), 25 μg/ml; kanamycin (Kan), 30 μg/ml; gentamicin (Gen), 30 μg/ml; tetracycline (Tc), 10 μg/ml; and ampicillin (Amp), 100 μg/ml. TY medium contains tryptone (5 g/L), yeast extract (3 g/L), and CaCl₂ (0.6 g/L). The L3 medium contains KH₂PO₄ (1.36 mg/L), NH₄Cl (0.53 mg/L), carbon source (adjusting the carbon according to the need, 10 mM), MgSO₄ (100 mg/L), NaCl (50 mg/L), CaCl₂ (40 mg/L), FeCl₃ (5.4 mg/L), Na₂MoO₄ (5 mg/L), biotin (2 mg/L), nicotinic acid (4 mg/L), and pantothenic (4 mg/L). LB medium contains tryptone (10 g/L), yeast extract (5 g/L), and NaCl (10 g/L).

The domains of ActR (AZC_0619) were predicted by InterPro annotation (Mitchell et al., 2019). The Kyoto Encyclopedia of Genes and Genomes (KEGG) database was searched for the protein sequences of FtcR-like proteins (accession number: K21603) (Kanehisa et al., 2021). Alignments of the protein sequences of FtcR-like protein were done using COBALT with the default parameters (Papadopoulos and Agarwala, 2007). The maximum likelihood (ML) tree of FtcR-like protein sequences was reconstructed using the MEGAX with the LG amino acid substitution model and gamma distributed with invariant sites (G+I) (Kumar et al., 2018). Furthermore, complete deletion of gaps and missing data was carried out to exclude highly variable regions. The AZC_0619 and selected proteins were shown in a logo that was generated by Weblogo with the default parameters (Crooks et al., 2004) and further analyzed with ESPript 3.0 (Robert and Gouet, 2014).

Genomic DNA was extracted from A. caulinodans ORS571, and the actR deletion mutant was constructed using a methodology of allelic exchange mutagenesis. First, fragments of upstream and downstream genes of actR were amplified from the genomic DNA of A. caulinodans ORS571 with two pairs of primers (actR-up-F and actR-up-R, and actR-down-F and actR-down-R), carrying KpnI, NdeI, AgeI, and SacI restriction enzyme sites, respectively (Table 1). These fragments were cloned into vector pEASY-Blunt Simple to generate plasmid pEASY::actR up and pEASY::actR down and then, respectively, linked with pCM351 to generate pCM351::actR up–down by restriction enzyme digestion. Then, pCM351::actR up–down was transformed into E. coli DH5α competent cells. Finally, pCM351::actR up–down was introduced into A. caulinodans ORS571 via allelic exchange, and actR gene was replaced. Mutations were selected in TY medium with Nal, Amp, and Gen and were verified by PCR using a pair of primers: actR-up-F and actR-down-R. The mutant with a deleted actR gene was named ΔactR.

To construct the complemented strain of ΔactR (ΔactR-C), the fragment of the entire open reading frame and predicted promoter of actR was amplified, then was digested with Xho I and BamH I, and cloned into plasmid pBBR1MCS-2 (Kovach et al., 1995). The recombinant plasmid pBBR-actR was introduced into ΔactR mutant by triparental conjugation. The complemented strain was named as ΔactR-C. The ActR overexpressed strain, ΔactR with empty plasmid pBBR1MCS-2, and wild type (WT) with empty plasmid pBBR1MCS-2 were constructed in the same way and was designated as WT-O, ΔactR-P, and WT-P, respectively.

WT and ΔactR mutant strains were cultured overnight with TY medium containing antibiotics. The overnight cultures were collected and adjusted to an optical density (OD) at OD₆₀₀ of 0.05. Normalized cultures measuring 500 μl of the WT and mutant were inoculated into 50 ml of TY medium and then were cultured in a rotary shaker at 37°C. The value of OD₆₀₀ was measured every 2 h. The data were recorded as means and SDs from three repetitions.

The swimming assay was analyzed on L3 soft agar (0.3% agar) plates containing succinate as the sole carbon source and 10 mM of NH₄Cl. Overnight cultures were adjusted to a density at OD₆₀₀ of 1.0. Aliquots measuring 5 μl of cell suspensions of the WT, ΔactR, ΔactR-C, ΔactR-P, WT-P, and WT-O were inoculated on L3 soft agar plates for 48 h at 37°C.

The bacterial strains were cultured in TY medium for 24 h at 37°C. The overnight cultures were collected and washed twice with sterile phosphate-buffered saline (PBS), spotted on copper grids, and negatively stained with 2% phosphotungstic acid. The transmission electron microscopy (TEM) images of the WT and ΔactR mutant were taken at random with grid.

The total RNA was isolated from A. caulinodans ORS571 using TransZol Up Plus RNA kit. cDNA was generated through TransScript One-Step gDNA removal kit according to the manufacturer’s instructions. The synthesized cDNA was diluted 500-fold and used as template to analyze relative gene expression. The quantitative qPCR was performed with GoTaq^® qPCR Master Mix kit using gene specific primer pairs (Table 2). The quantitative PCR program consisted of an initial denaturation at 95°C for 2 min, followed by 40 cycles of 95°C for 15 s, 60°C for 30 s, and 72°C for 1 min. To evaluate the gene expression, the copy number of each gene was normalized to that of the 16S rRNA. The analysis of its results was performed using the comparative cycle threshold method (Schmittgen and Livak, 2008).

WT and ΔactR mutant were incubated overnight in the L3 medium at 37°C, and then the culture was adjusted to OD₆₀₀ of 2. Bacterial suspensions measuring 15 μl were added into glass tubes including 1.5 ml of L3 medium and were incubated for 3 days at 37°C. The L3 medium contained succinic acid (10 mM) as a sole carbon source, with nitrogen. After 3 days’ incubation, the glass tubes were gently washed three times with PBS (pH = 7.2) to remove free-floating bacteria and then stained with 2 ml of crystal violet (CV; 0.1%, w/v) for 20 min. CV was gently removed by washing three times with deionized water. Finally, biofilm formation was quantified by ethanol-solubilized CV from glass tube biofilms. The OD value of each tube was determined after CV staining with a wavelength of 590 nm.

EPS production was estimated using the method described by Jiang et al. (2016) with the following modifications. The WT and ΔactR mutant were incubated in the L3 medium at 37°C. Then, bacterial cultures were adjusted to an OD₆₀₀ of 0.8. Ten-microliter aliquots of bacterial suspensions were spotted on L3 solid agar (0.8% agar) plates with three different carbon sources (sodium lactate, glycerol, and malic acid) and incubated at 37°C. The L3 plates included 10 mM of carbon source, 10 mM of NH₄Cl, and 40 μg/ml of Congo red. Photographs were taken after 4 days of incubation. The quantitative analysis of EPS content was measured by a colorimetric method using anthrone and sulfuric acid and was evaluated by normalizing to OD₆₀₀ of the bacterial suspension.

Flocculation was measured using the method described by Jiang et al. (2016). Overnight cultures were normalized to an OD₆₀₀ of 1.0, and then 200 μl of the normalized cultures was inoculated into 10 ml of L3 medium containing 10 mM of sodium lactate as the carbon source and 0.5 mM of NH₄Cl as the nitrogen source. Photographs of the non-flocculated cells and flocculated cells were taken after 48 h of incubation. For the quantitative analysis, the bacterial suspensions were left to stand for 30 min. The non-flocculated cells were removed from the tube, and the OD₆₀₀ of the suspension (ODs) was measured. The flocculated cells that settled to the bottom of the tube were dispersed using a tissue homogenizer and thoroughly mixed with free cells; the density of the mixed culture was also measured at 600 nm and named as OD_t. The percent flocculation was calculated as follows: percent flocculation = (1 – OD_s/OD_t) × 100.

Seeds of S. rostrata were treated with concentrated sulfuric acid for 30 min, followed by rinsing five times with sterilized water. Sterile seedlings were germinated in the dark at 37°C for 72 h. WT and mutant strains were grown overnight in sterile TY medium; then overnight cultures were collected and adjusted to a density of 0.8. For competitive colonization, suspensions of the WT and mutant were mixed at 1:1, 1:5, and 1:10. Germinated S. rostrata seeds were inoculated with the mixed cultures for 1 h. The surfaces of seedlings were washed seven times using sterilized water. And then the root tips were vortexed, and bacteria were reisolated on TY agar plates with ampicillin. A total of 100 of the colony-forming units (CFUs) were selected, and the number of the WT and ΔactR was counted by plate streaking in TY agar plates with ampicillin and gentamicin. In addition, the WT and mutant colonies were also identified by PCR.

Surface-sterilized seedlings of S. rostrata were grown in soil supplemented with sterilized vermiculite and low nitrogen plant nutrient solution for 4 weeks. The bacterial suspension (OD₆₀₀ = 0.6) of the WT and ΔactR mutant strains was mixed at ratios of 1:1 and 1:5; then the mixed cultures were used to paint the stem surface of S. rostrata. After 30 days’ inoculation at 27°C in the greenhouse, nodules were harvested, and bacteria were reisolated and plated on TY agar medium from stem nodules. The WT and mutant bacteria were identified and distinguished by PCR using the primer pair actR in-F and actR in-R, and the ratios between the WT and mutant were further counted.

Mean and standard errors were measured based on each experiment repeated three times. The Statistical Package for the Social Sciences (SPSS version 20) was used to analyze the least significant difference test (p < 0.05).

From the genome sequence of A. caulinodans ORS571, we searched for the presence of FtcR homolog, AZC_0619. Only one open reading frame encoding for the FtcR-like protein was identified, which was named as ActR. Sequence analysis showed that ActR contains 222 amino acids, encoding one N-terminal receiver domain (IPR001789) and one C-terminal OmpR/PhoB-type DNA binding domain (IPR001867) (Supplementary Figure 1). A BLASTp analysis exhibited that AZC_0619 shared about 53.2% of identity with a flagellar master regulator FtcR of B. melitensis 16M. actR gene was located downstream of two chemotactic genes (cheY and cheZ), mltE gene encoded for transmembrane protein, and three flagellar genes (motB, motC, and fliK) in A. caulinodans (Figure 1A). And flagellar genes were found in the proximity of ftcR-like gene in most species (Figure 1A), indicating that FtcR-like protein may play a role in the regulating flagellar motility among Rhizobiales.

To clarify the evolutionary relationships of the FtcR-like proteins, a phylogenetic tree was generated. As shown in Figure 1B, the FtcR-like tree exhibited extensive genetic diversity. FtcR-like proteins among Rhizobiales members were monophyletic and exhibited topology that was generally congruent with the species taxonomy [KEGG Organisms in the National Center for Biotechnology Information (NCBI) Taxonomy], suggesting that FtcR-like proteins originated from a last common Rhizobiales ancestor and underwent divergent evolution during species differentiation. Within this phylogenetic tree, we found that FtcR-like proteins in Azorhizobium formed a monophyletic subclade with long branch lengths (Figure 1B). Notably, FtcR-like proteins from A. caulinodans ORS571 and B. melitensis 16M are clustered into separated clades.

We then systematically examined the prevalence and evolution of FtcR. Based on the KEGG orthology database (KO: K21603) (Kanehisa et al., 2021), FtcR-like proteins were present in orders Rhizobiales, Rhodobacterales, and Polymorphum of Alphaproteobacteria. FtcR-like proteins were widespread in most genus/species among Rhizobiales and had a sporadic distribution among Rhodobacterales (genera Pseudovibrio, Pannonibacter, and Labrenzia). A total of 117 protein sequences of FtcR-like were recovered from the KEGG database (KO: K21603) and exhibited a significant divergence (more than 35.4% identity) at the amino acid level (Figure 2). The FtcR-type sequences are conserved in genus Azorhizobium (A. caulinodans, Azorhizobium doebereinerae, and Azorhizobium oxalatiphilum) with the pairwise sequence identities in the range of 93.24–95.59% (Figures 1A, 2); this result was consistent with that of Azorhizobium FtcR proteins that formed a monophyletic subclade in Figure 1B. Our analysis indicated that FtcR-like protein is conserved and that it may play an important role among Azorhizobium. Although Léonard et al. (2007) provided insights on the role of flagellar master regulator FtcR in Brucella, the function of its homology in rhizobia has remained unclear. Therefore, we will investigate the biological function of ActR in A. caulinodans ORS571.

ActR was predicted to be a flagellar two-component response regulator. To characterize the role of ActR in regulating motility, ΔactR mutant strain was constructed through allelic exchange mutagenesis (see section “Materials and Methods,” Supplementary Figure 2), and the swimming motility behavior of the ΔactR mutant and complementary strain were compared with those of the WT (Figures 3A,B). We found that the mutant (ΔactR-P) with an empty plasmid pBBR1MCS-2 was devoid of motility ability, which could be rescued by introducing the pBBR1MCS-2 carrying WT actR gene and its native promoter into the mutant (ΔactR-C) (Figure 3A). We also found that an overexpressing strain (WT-O) exhibited significant increase of swimming motility when compared with that of the WT-P. The WT-O and WT-P were constructed by the introduction of an empty plasmid pBBR1MCS-2 and a pBBR1MCS-2 carrying WT actR gene and its native promoter, respectively. The quantitative data clearly confirmed the result, too (Figure 3B).

To explore the reason for lost motility ability, it was first checked that growth properties of the mutant strain was not impaired (Supplementary Figure 3). What is really interesting was that TEM of the WT and the mutant strains revealed the absence of flagella in the mutant strain (Figure 3C).

Next, we tested whether ActR regulator regulated the expression of flagellar synthesis-related genes by quantitative real-time PCR. It is true that the deletion of any flagellar component is sufficient to abolish flagellar synthesis, such as FliM and FliN (Shen et al., 2018). The results showed that the expression levels of eight flagellar genes (motB, fliF, fliL, fliG, fliM, flgI, flgG, and flhB) in ΔactR mutant were significantly downregulated than those in the WT (Figure 3D). For example, the expression values of fliL and fliG declined 27 and 88%, respectively. These results suggested that ActR was closely associated with the synthesis of flagella and positively regulated the motility of A. caulinodans ORS571.

Biofilm formation and motility are modulated directly by flagellar power (Subramanian and Kearns, 2019), and biofilm forming on plant surfaces is important for bacterial symbiotic interaction with host plant (Xu et al., 2019). To investigate whether the deletion of actR has an influence on bacterial biofilm formation, we tested the biofilm of the WT and the mutant. As shown in Figure 4A, the mutant produced less biofilm than the WT in the L3 medium with nitrogen (L3+N medium), and quantitative data confirmed the results. In the L3+N medium, the biomass of biofilm in ΔactR mutant was 32% less than that of the WT strain (Figure 4B).

We next tested the production of EPS by Congo red staining. EPS is essential for the maintenance of biofilm formation in bacteria (Cugini et al., 2019). And it is involved in bacterial symbiotic nodulation with S. rostrata (Sun et al., 2020). To study the EPS production ability of ΔactR, the colony morphologies and quantitative analysis of the EPS were examined using minimal medium plates containing different carbon sources (sodium lactate, glycerol, malic acid, and succinic acid). A significant difference between the WT and the mutant was observed. The colonies of the WT cells produced more EPS than ΔactR cells regardless of the kind of carbon sources were assayed (Figures 4C,D). These data indicated that ActR positively affected EPS production and biofilm formation in A. caulinodans ORS571.

Flocculating substances are secreted by many microorganisms in the culture broth, which comprise polysaccharides, proteins, and lipids (Salehizadeh and Yan, 2014). To study whether ActR influences flocculation in A. caulinodans ORS571, we tested the flocculation morphologies between the WT and mutant strains in the L3 medium with 5 mM of NH₄Cl as a nitrogen source. Figure 5A shows that flocs formed by the WT were larger and more abundant than those of the mutant strain. Quantitative analysis suggested that the WT flocculated more than the mutant strain and that the flocculation formation of both strains increased over time (Figure 5B). These results indicated that ActR was involved in regulating flocculation formation.

The colonization on the surface of the host plant is a key step for successfully establishing symbiosis. To investigate the symbiotic role of ActR in A. caulinodans, competitive colonization experiments were performed. We counted the number of cells reisolated from the seedlings to verify the efficiency of competitive colonization. As shown in Figure 6A, the ΔactR mutant was less competitive than the WT. When the WT and the mutant cells were mixed in equal proportion, 100% of the bacteria reisolated from the root surface belonged to the WT, showing that the mutant could not compete. When the proportions between the WT and mutant were 1:5 and 1:10, a few of the cells of ΔactR mutant strains (14.3 and 23.1%, respectively) were reisolated from the root system. This result suggested that the deletion of actR did not disable bacterial colonization ability but affected A. caulinodans with an effective competitive ability for root colonization.

Induction of nodule morphogenesis is associated with the EPS production and colonization of the root system (Liu and Xie, 2019). To study whether ActR plays an important role in nodule formation, the competitive nodulation assay was performed by counting the number of cells (WT and mutant strains) reisolated from the stem nodules. As shown in Figure 6B, the ratios of ΔactR mutant were 8 and 31% when the WT and mutant strains were mixed at ratios of 1:1 and 1:5, respectively, indicating that the ΔactR mutant could still nodulate S. rostrata but not compete with the WT strain. Taken together, these results indicated that ActR could provide A. caulinodans an effectively competitive ability for symbiosis.