Azospirillum is one of the best studied plant growth promoting bacteria (PGPB) and is considered a general model for plant-bacteria interactions (Cassán et al. 2020). These bacteria can positively influence the growth of many plant species and have therefore become the basis of biological products designed to promote growth and increase plant productivity in an environmentally friendly and cost-effective manner (Nievas et al. 2023). The beneficial effects of these products, termed inoculants, are attributed to several processes of Azospirillum metabolism: biological fixation of atmospheric nitrogen, synthesis of phytohormones such as auxins, gibberellins, and cytokinins, production of siderophores, phosphate solubilization, and induction of plant tolerance mechanisms to abiotic and biotic stresses (Bashan and de-Bashan 2010; Cassán et al. 2020; Fukami et al. 2018). Among these mechanisms, the availability of nitrogen to the associated plant through biological fixation of atmospheric nitrogen and the synthesis of phytohormones, especially indoleacetic acid (IAA), are considered to be the most important players in growth promotion by Azospirillum (Cassán et al. 2020). However, for Azospirillum to exert these beneficial effects and promote plant growth, efficient colonization of the plant in question is required (Steenhoudt and Vanderleyden 2000). Therefore, not only the mechanisms to promote plant growth but also the factors involved in plant colonization by PGPB should be investigated to improve the inoculation technology.

Chemotaxis and motility have been described as important factors for root colonization by Azospirillum (Aroney et al. 2021; Liu et al. 2024; Wang et al. 2025). Azospirillum bacteria are highly mobile and show positive chemotaxis to various attractants such as organic acids, sugars, amino acids, and aromatic compounds, as well as root exudates (Steenhoudt and Vanderleyden 2000). In addition to motility, polar flagella are required for weak adhesion to root surfaces, as loss of polar flagella abolishes the ability of A. brasilense cells to adhere to wheat roots (Croes et al. 1993). The attachment of A. brasilense to plant roots is thought to be a process involving two phases: the primary adsorption phase, characterized by weak and reversible adhesion mediated by the polar flagellum, and the secondary anchorage phase, involving the formation of cell aggregates and firm and irreversible attachment associated with the production of extracellular polysaccharides by A. brasilense (Croes et al. 1993; Michiels et al. 1991; Steenhoudt and Vanderleyden 2000).

In addition to chemotaxis and motility, bacterial cell surface polysaccharides such as exopolysaccharides (EPS), lipopolysaccharides (LPS) and extracellular proteins have been associated with root colonization by Azospirillum. Disruption of noeL (encode GDP-mannose 4,6-dehydratase), associated with EPS biosynthesis, impaired the ability of strain Sp7 to aggregate and form biofilm on glass (Lerner et al. 2009). In a mutant of A. brasilense lacking dTDP-rhamnose biosynthesis, the LPS core structure was altered, which also affected the ability to colonize maize roots (Jofré et al. 2004). The major outer membrane protein (OmaA) has exposed domains on the cell surface that may be involved in the adsorption of Azospirillum to the root, and these exposed domains interact with surface domains of neighboring bacteria, allowing cell aggregation (Burdman et al. 2000). Using confocal microscopy, A. brasilense cells were observed to form aggregates and layers of micro-colonies connected by extracellular polymeric substances, indicating a mature biofilm structure (Viruega-Góngora et al. 2020). Analysis of the components of the biofilm matrix of A. brasilense revealed the presence of polar flagellin, exopolysaccharides, the major outer membrane protein (OmaA), and extracellular DNA (Viruega-Góngora et al. 2020). The matrix confers mechanical stability to biofilms, protects them from environmental stresses, and mediates bacterial adhesion to a surface (Flemming and Wingender 2010; Yaron and Römling 2014). Therefore, the ability of A. brasilense to auto aggregate and form biofilms is important for both plant root colonization and bacterial survival.

The cells of A. brasilense can change their morphology and physiology in response to changes in the environment. Under stressful conditions such as high oxygen rates, desiccation, and nutrient limitation, A. brasilense cells transition from free swimming to agglomeration and then to flocculation if metabolic stress persists (Sadasivan and Neyra 1985; Bible et al., 2015). The formation of large macroscopic clusters (flocculation) of A. brasilense cells in liquid cultures is induced in a medium with a high C:N ratio (Sadasivan and Neyra 1985). Agglomeration is a prerequisite for flocculation, a process in which motile A. brasilense cells transform into immobile cyst-like forms embedded in a matrix of extracellular polysaccharides and accumulate granules of intracellular poly-β-hydroxybutyrate (PHB) (Sadasivan and Neyra 1985, 1987; Bible et al. 2015). These PHB granules are a carbon source that favors the survival and competitiveness of Azospirillum spp. cells in situations of nutrient and energy imbalance (Sadasivan and Neyra 1987).

The flocculation of A. brasilense cells is controlled by the response regulator FlcA both in culture and in association with plants (Pereg-Gerk et al. 1998). Azospirillum brasilense flcA mutants do not flocculate, do not differentiate from motile vegetative cells to “cyst-like” forms, and lack exopolysaccharide material on the cell surface (Pereg-Gerk et al. 1998; Hou et al. 2014). Hou et al. (2014) investigated proteins that are differentially expressed under flocculation conditions in A. brasilense Sp7 and its flcA knockout mutant. The functional properties of these proteins included carbohydrate metabolism, morphological transformation, nitrogen fixation, and stress tolerance, indicating the involvement of flcA in these processes. Among the differentially expressed proteins, the protein homologous to NarL (NarL-like) showed increased expression in the flcA knockout mutant compared to the wild-type strain Sp7 under flocculation conditions (Hou et al. 2014). Interestingly, RNA-Seq analysis performed three days after inoculation showed that the narL-like gene was strongly expressed in A. brasilense FP2 cells when associated with wheat (Triticum aestivum) (Camilios-Neto et al. 2014). In this analysis, other genes besides the narL-like gene were differentially expressed, suggesting that interaction with wheat causes high expression of A. brasilense genes that encode proteins related to the initial stages of plant-bacteria interaction (chemotaxis, adhesion, and biofilm formation), adaptation processes, and nitrogen fixation (Camilios-Neto et al. 2014), revealing candidate genes of A. brasilense involved in the interaction of this bacterium with wheat.

We have shown that the ΔnarL-like mutant and the nonflocculating ΔflcA mutant have a lower ability to bind Congo red than the wild-type strain FP2, suggesting a lack of or altered surface polysaccharide composition in these mutants (REFERENCE). We also show that the wheat- and Setaria viridis-associated epiphytic population and the maize-associated adherent population of the ΔnarL-like mutant were larger than those of FP2 (Camilios-Neto et al. 2014). These results support the idea that the altered surface polysaccharide composition in the ΔnarL-like mutant may have contributed to better colonization of the tested grasses. In this context, the aim of this work was to mutagenize and characterize the narL-like gene in the FP2 strain of A. brasilense and to compare the mutant strain with the flcA mutant.

In silico study of the flcA and narL-like genes of Azospirillum brasilense strain FP2

The target genes (flcA and narL-like) were located in the annotated assembly of the FP2 strain genome (Guizelini et al. 2016) using the Artemis software (Carver et al. 2012). The genes nucleotide sequences were compared with those deposited in the NCBI database, using the Blastp tool to confirm similarity to homologous genes from other organisms. From the genomic sequences, primers were designed for the genes amplification (Table S1) and subsequent mutagenesis. The known and predicted protein‒protein interaction networks of the protein products of the narL-like and flcA genes were analyzed using String software version 11.5 (Szklarczyk et al. 2019). The conserved domains of the NarL-like and FlcA proteins were analyzed using Pfam 31.0 (http://pfam.xfam.org/).

Growth of bacterial strains, DNA manipulations, and mutagenesis.

The bacterial strains and their relevant characteristics are listed in Table 1. Azospirillum brasilense strain FP2 was grown in NFbHPN-lactate liquid medium (Machado et al. 1991) at 30°C, under 120 rpm agitation for 24 h. Escherichia coli strains were grown at 37°C in LB medium or LA medium (LB medium containing 15 g/L agar) (Sambrook et al. 1989). Solid TY medium (Sambrook et al. 1989) was used, mixed with NFbHPN-lactate, for triparental conjugation. When necessary, the culture media were supplemented with the following concentrations of antibiotics: for E. coli, 250 µg.mL^-1 ampicillin (Ap), 5 µg.mL^-1 nalidixic acid (Nal), 50 µg.mL^-1 kanamycin (Km), 30 µg. mL^-1 chloramphenicol (Cm), 40 µg.mL^-1 streptomycin (Sm); for FP2 of A. brasilense, 10 µg.mL^-1 of Nal, 50 µg.mL^-1 of Km, 80 µg.mL^-1 of Sm and 10 µg.mL^-1 of tetracycline (Tc). The plasmids used in this study are listed in Table 1. Plasmid and total DNA preparations, agarose gel electrophoresis, and digestion with restriction endonucleases were performed according to standard protocols (Sambrook et al. 1989).

The ΔflcA and ΔnarL-like mutant strains of A. brasilense strain FP2 were obtained by the deletion mutagenesis. For this, primers (Table S1) were designed to amplify fragments upstream and downstream of the gene to be deleted (Fig. S1). The obtained amplicons were ligated into the pTZ57R/T vector and transformed in E. coli TOP10, resulting in plasmids pTZUflcA, pTZDflcA, pTZUnarL-like, and pTZDnarL-like. Plasmids pTZUflcA and pTZUnarL-like were then cleaved at the XbaI and XhoI sites, and plasmids pTZDflcA and pTZDnarL-like were cleaved at the XhoI and HindIII restriction sites; the relevant fragments were purified, and for subsequent ligation and cloning into the pTZ57R/T vector. The plasmids containing the upstream and downstream spliced DNA fragments of each gene (pTZUDflcA and pTZUDnarL-like) were cleaved at the XbaI and HindIII sites and subcloned into the pK18mobsacb vector containing a negative selection marker, sacB gene (Gay et al. 1983). The obtained constructs (pK18mobsacB vector + insert, plasmids pK18UDflcA and pK18UDnarL-like) were transferred by triparental conjugation to A. brasilense strain FP2 as described by Pedrosa and Yates (1984) using the pRK600 helper plasmid in E. coli strain HB101 (Fig. S2), with some adaptations.

The E. coli donor strains TOP10 (containing pK18mobsacB with the insert of interest) and HB110 (containing the pRK600 plasmid) were grown in LB liquid medium with antibiotics under constant shaking for 16 h at 37°C. After this period, the E. coli strains were reinoculated in a 1:100 ratio in LB liquid medium without antibiotics and incubated under agitation and appropriate temperatures for a period of approximately 3 h. The A. brasilense FP2 strain (recipient) was grown in NFbHPN-lactate liquid medium without antibiotics for about 6 h until reaching an OD_600nm = 1. Then, mixtures of the A. brasilense culture with the E. coli cultures were made in two different proportions. In one falcon tube, 5 mL of FP2 + 2.5 mL of E. coli HB101 + 2.5 mL of E. coli TOP10 were mixed, and in another tube, 9 mL of FP2 + 0.5 mL of E. coli HB101 + 0.5 mL of E. coli top10 were mixed. Then the mixtures were centrifuged for 10 min at 5000 rpm. The supernatant was discarded and the cells were resuspended in the remaining medium with a pipette. The suspension was plated in drops on TY-NFbHPN-lactate (1:1) solid medium and incubated at 30°C overnight. The solid TY–NFbHPN-lactate medium contained a reduced concentration of the phosphate mixture (45 mM) and NH₄Cl (18 mM).

After about 24 h, the mass of cells was scraped, suspended in 1 mL of liquid NFbHP-lactate, and then an aliquot of 200 µL was plated on solid NFbHPN-lactate medium containing the appropriate antibiotics for selection of the transconjugants. Kanamycin, used as a positive selection marker, enabled the selection of bacteria that had integrated the plasmid into the chromosome (first homologous recombination event). The kanamycin-resistant transconjugants were subsequently grown in NFbHP-lactate medium with and without 10% sucrose. The second recombination event was identified by sucrose sensitivity, as the sacB gene present in the vector is lethal in the presence of this compound. Thus, colonies resistant to kanamycin and sensitive to sucrose were considered candidates for carrying the desired mutation in strain FP2. Mutation confirmation was performed by PCR using the Fwd Upt and Rev Dwt primers (Figs. S1 and S2), followed by sequencing.

Flocculation and Congo red binding assay

Flocculation was evaluated as described by Hou et al. (2014) with some modifications. Cultures of A. brasilense strain FP2 were grown in NFbHPN-lactate liquid medium (Machado et al. 1991) until reaching an OD _600nm of 0.8–0.9, after which the cells were collected by centrifugation at 3000xg for 5 min. The pellet was washed in 5 mL of NFbHP-lactate medium and inoculated in the flocculation medium (NFbHP-lactate medium supplemented with 8 mM fructose and 0.5 mM KNO₃) to an OD _600nm of 0.3–0.4. The experiments were performed in 50 mL conical flasks containing 10 mL of flocculation medium, which were incubated at 30°C on a shaker at 120 rpm for 6h. Flocculation was observed under an Olympus BX51 microscope.

For the Congo red binding assay five microliters of liquid cultures grown in NFbHPN-lactate medium (OD _600nm = 0.8–0.9) were plated on solid NFbHPN-lactate medium containing 45 µg/mL Congo red; the resulting colonies were analyzed for color and morphology by stereomicroscopy.

Structural organization of the flcA and narL-like gene regions in the genome of Azospirillum brasilense strain FP2 and in silico analysis of the proteins

The genome of A. brasilense strain FP2 consists of one chromosome and 6 plasmids (Guizelini et al. 2016). Using the Artemis software (Carver et al. 2012) and the annotated genome sequence (GCF_000404045.1), we identified that the flcA and narL-like genes were located on the chromosome. The position on the chromosome and general characteristics of these genes are listed in Table S2. The flcA gene is located upstream and in the same transcriptional direction as the narL-like gene, with both genes flanking the divergently-transcribed (Fig. S3).

Previous work showed that the narL-like gene was highly expressed in wheat roots (Triticum aestivum) colonized by A. brasilense strain FP2 (Camilios-Neto et al. 2014). flcA gene has been previously reported in the literature to be important for bacterial-plant interactions (see references in Table S2). Furthermore, because the flcA and narL-like genes are close to each other in the FP2 genome (Fig. S3), we wanted to investigate whether a functional interaction relationship exists between these genes using the String software version 11.5 (Szklarczyk et al. 2019). However, no predicted functional interaction of narL-like with flcA was identified.

Analysis of the conserved domains of FlcA and NarL-like proteins with the Pfam 31.0 program (http://pfam.xfam.org/) showed that both have similar domains typical of transcriptional regulators: the response regulatory protein receptor domain (REC) and the DNA-binding helix turn-helix domain (HTH) of the LuxR type (Fig. S4). We also identified that the ORF found between the two (Fig. S3), which contains a REC domain, suggesting it may encode a putative response regulator. The results confirmed the structural similarity of the protein encoded by the narL-like gene with the NarL response regulatory protein, which together with the NarX protein forms the NarX/NarL two-component signaling system in other bacteria such as E. coli and P. aeruginosa (Butcher and Tabor 2022; Moreno-Vivián et al. 1999; Musial and Gmiter 2025; Van Alst et al. 2007). However, no narX-like gene was found in the genomic neighborhood suggesting that the narL-like gene is not involved in nitrate metabolism.

Nitrogenase activity

Although the ΔnarL-like mutant appeared to show increased nitrogenase activity (52 nmol ethylene/min.mg) when compared with the wild type (35 nmol ethylene/min.mg protein), although the difference was not statistically significant at p ≤ 0.05 by t test (Fig. 1). The ΔflcA mutant showed nitrogenase activity levels close to those of the wild-type strain (~ 30 nmol ethylene/min.mg protein) (Fig. 1). This result indicates that the ability of the strains to fix nitrogen in vitro was not impaired by the gene deletion in the mutants under the conditions tested. Furthermore, the ΔnarL-like and ΔflcA mutants were cultivated on solid NHbHP-lactate medium supplemented with 5 mM potassium nitrate (KNO₃) as the sole nitrogen source, demonstrating their ability to utilize nitrate as the exclusive nitrogen source, similarly to the wild-type strain.

Flocculation and Congo red binding assays

Cells of the wild-type strain FP2 and mutant strain ΔnarL-like flocculated under in the flocculation medium (NFbHP lactate containing 8 mM fructose and 0.5 mM KNO₃), whereas ΔflcA did not (Fig. 2). Flocculation of the ΔnarL-like strain was increased compared to the parental FP2 strain. These results confirms the phenotype previously described in the literature for flcA mutants, while revealing an opposite phenotype in the ΔnarL-like mutant.

To study the formation of surface polysaccharides by A. brasilense, the ability of Congo red to bind to strains was investigated, as this dye interacts with various external polysaccharides of bacteria. The wild-type strain formed dark red, drier, and rougher surface colonies when cultured on agar medium with Congo red, while ΔflcA and ΔnarL-like formed bright red and smooth surface colonies (Fig. 3). These results indicate that ΔnarL-like is capable of flocculation but has altered cell surface polysaccharide composition compared with the wild-type strain, as does the ΔflcA mutant.

Colonization and adhesion assays

There were no statistically significant differences between strains in the number of cells adhering to the surface of wheat roots (Fig. S5). At the time periods studied (1, 3, 7, and 12 DAI), the planktonic and epiphytic populations of the strains ranged from 10⁵ to 10⁸ CFU.mL^-1 and 10⁷ to 10⁸ CFU.gram^-1 (fresh weight) of wheat root, respectively (Fig. S5 and 4). The planktonic population of ΔnarL-like was apparently higher over time than that of the wild-type strain. However, this increase was not statistically significant (p ≤ 0.05) by t test (Fig. S5).

The epiphytic population of the ΔflcA strain was similar to those of the wild strain (Fig. 4). Interestingly, the ΔnarL-like mutant strain had a higher number of CFUs than the wild-type FP2 at all time points examined, with the result being statistically significant from 3 DAI (Fig. 4).

To further evaluate the ability to attach to roots, maize plants were inoculated in a hydroponic system with 10⁵ CFU of each tested strain. At all time points tested, the ΔnarL-like strain seemed to show a higher number of root-attached bacterial cells, although this result was statistically significant only at DAI 0 (root adhesion assay) (Fig. 5, Table S3). The epiphytic population of the ΔnarL-like strain also appeared to be higher than that of the other strains at 5 and 7 days after inoculation. However, this difference was not statistically significant (p ≤ 0.05) by t test (Fig. 6, Table S3).

Fluorescence microscopy of A.brasilense strains colonization of roots

Because the ΔnarL-like strain appeared to have higher epiphytic populations, we introduced plasmids expressing fluorescent protein to directly observe root colonization by this strain compared with wild type. Fluorescently labeled strains were inoculated into Setaria viridis A10, and colonization was assessed after 3 days. When strains were inoculated individually, the ΔnarL-like mutant had a higher number of CFU per gram of root at 3 DAI regardless of the fluorescent plasmid used, although this was only significant for strains with plasmid pMP4655 (Fig. 7). The difference in colonization could be observed by confocal microscopy, as shown in Fig. 8A and confirmed by counting CFU (Fig. 8B). When inoculated with the red-labeled ΔnarL-like strain, the root exhibits a high density of attached bacteria compared to the red-labeled wild-type bacteria. These results indicate that the ΔnarL-like mutant is a better colonizer than the wild-type strain.

The results obtained through the Pfam program indicate that the NarL-like protein, like the FlcA protein, is a transcriptional regulator of the LuxR family, containing the REC receptor domain and the helix-turn-helix domain (HTH-helix-turn-helix) of binding to DNA, characteristic of two-component signal transduction systems (Fig. S4). Two-component systems (TCS) are versatile signaling circuits and represent one of the main mechanisms used by bacteria to regulate gene expression in response to environmental stimuli. These systems regulate essential processes in bacteria, such as quorum sensing, secondary metabolism, biofilm formation, motility, virulence, and antibiotic resistance (Alvarez and Georgellis 2023; Francis et al. 2017; Song et al. 2023; Tiwari et al. 2017).

The fact that NarL-like is a possible transcriptional regulator associated with the fact that the expression level of the narL-like gene was significant in the transcriptome of Azospirillum in association with wheat (Camilios-Neto et al. 2014), suggests that this gene acts in gene transcriptional regulation and that this regulation may be related to the interaction of this bacterium with the host plant. FlcA has a domain organization (RR and DNA binding HTH) very similar to that of NarL-like. However, no histidine kinase gene was found in their genomic region, remaining elusive the stimulus triggering the phosphorylation and activation of these proteins.

To better understand the role of the narL-like gene, we constructed deletions of strains of this gene and of flcA. The ΔflcA- and ΔnarL-like mutants showed less binding to the dye Congo red, bright red coloration, and smoother morphology than the wild-type strain. The ΔflcA mutant was unable to flocculate. In contrast, the ΔnarL-like mutant exhibited flocculation phenotype increased compared with the parental strains. These results are consistent with previous studies on flcA mutants of A. brasilense. For example, the ΔflcA mutant of strain Sp7 was also unable to flocculate and bound Congo red to a lesser extent than the wild-type strain (Hou et al. 2014). The dye Congo red binds to extracellular polysaccharides produced by bacteria, including A. brasilense (Pereg-Gerk et al. 1998; Hou et al. 2014). The difference in binding of colonies of the mutant strains compared to the wild-type strain to the dye Congo red and the differences in the morphology of these colonies suggest the absence or altered composition of these external polysaccharides in the mutants. Interestingly, the ΔflcA- and ΔnarL-like mutants show different flocculation and Congo red binding phenotypes, suggesting different changes in the composition of extracellular polysaccharides in the ΔflcA- and ΔnarL-like mutants. Consequently, these transcriptional regulators appear to interact to control the synthesis of cell wall components with complementary functions.

Although the differences in nitrogenase activity among the three strains were not significant, it is noticeable that the ΔnarL-like strain showed increased level. Despite the unexpectedness of this result, it can be explained by the fact that exopolysaccharides can alter cell wall composition and contribute to oxygen diffusion, possibly enhancing low oxygen conditions required for the optimum nitrogenase activity (Dong Z et al. 2002; Hartmann and Burris 1987; Shao et al. 2022; Whitfield GB et al 2025).

The results of this study show that deletion of the narL-like gene in A. brasilense strain FP2 results in changes in the ability to bind Congo red and colonize plants. These data suggest that the observed change in external polysaccharide composition in the Δnarl-like mutant contributes to enhanced colonization of roots of the tested grasses compared with the wild strain. However, it is still unclear how this gene affects the tested traits. To our knowledge, this is the first report describing the function of the NarL-like protein, and further studies are needed to better characterize these mutants.

In summary, the results presented here support the role of flcA in A. brasilense and provide evidence for the possible involvement of the narL-like gene in Azospirillum interactions with wheat, maize, and S. viridis, and for the importance of these genes in Azospirillum-plant interactions.