Expression and function of the cdgD gene, encoding a CHASE–PAS-DGC-EAL domain protein, in Azospirillum brasilense

The plant growth-promoting bacterium Azospirillum brasilense contains several genes encoding proteins involved in the biosynthesis and degradation of the second messenger cyclic-di-GMP, which may control key bacterial functions, such as biofilm formation and motility. Here, we analysed the function and expression of the cdgD gene, encoding a multidomain protein that includes GGDEF-EAL domains and CHASE and PAS domains. An insertional cdgD gene mutant was constructed, and analysis of biofilm and extracellular polymeric substance production, as well as the motility phenotype indicated that cdgD encoded a functional diguanylate protein. These results were correlated with a reduced overall cellular concentration of cyclic-di-GMP in the mutant over 48 h compared with that observed in the wild-type strain, which was recovered in the complemented strain. In addition, cdgD gene expression was measured in cells growing under planktonic or biofilm conditions, and differential expression was observed when KNO3 or NH4Cl was added to the minimal medium as a nitrogen source. The transcriptional fusion of the cdgD promoter with the gene encoding the autofluorescent mCherry protein indicated that the cdgD gene was expressed both under abiotic conditions and in association with wheat roots. Reduced colonization of wheat roots was observed for the mutant compared with the wild-type strain grown in the same soil conditions. The Azospirillum-plant association begins with the motility of the bacterium towards the plant rhizosphere followed by the adsorption and adherence of these bacteria to plant roots. Therefore, it is important to study the genes that contribute to this initial interaction of the bacterium with its host plant.

Azospirillum brasilense is a species that has been employed as an inoculant worldwide because of its beneficial effects on host plants via multiple mechanisms, which are proposed to be responsible for the improved nutritional status of plants. These mechanisms include both direct effects on plant metabolism by providing nutrients (iron, phosphorus and nitrogen) and synthesizing molecules with phytohormone activity (auxins, cytokinins, gibberellins, and nitric oxide) and indirect effects such as antagonism against phytopathogens and increased resistance to pathogens, which define the mode of action of Azospirillum in plant development according to the additive hypothesis 1 . Therefore, this bacterium must contain an array of genes that participate in the control of the successful colonization of plants through such processes as chemotaxis and biofilm formation, which are the initial steps leading to the beneficial effects observed in plants of agronomic interest 2,3 .
Investigations into the role of the cyclic dimeric guanosine monophosphate (c-di-GMP) second messenger signaling system have shown that the predominant bacterial trait controlled by this system is the transition from a motile to a sessile lifestyle 4,5 and these studies have indicated that the accumulation of c-di-GMP promotes biofilm formation via the biosynthesis of biofilm matrix components, adhesins, and exopolysaccharides [6][7][8] , while a decrease in c-di-GMP promotes motility, inhibiting this process 9,10 . The enzymes catalyzing c-di-GMP formation are diguanylate cyclases (DGCs), while those degrading c-di-GMP are phosphodiesterases (PDEs), which are identified by their characteristic GGDEF and EAL domains, respectively 11 . In bacteria, genes with translational products that include both domains (GGDEF and EAL) are referred to as hybrid proteins 11 .

Construction and selection of the A. brasilense 12-A null mutant,
A. brasilense C-56A complemented mutant, and A. brasilense C-40A control strains. The A. brasilense cdgD::Km R null mutant, referred to as the 12-A strain, was constructed as described previously 16 . Initially, the pABcdgD suicide plasmid was produced by molecular cloning following the molecular biology protocols described by Green and Sambrook 17 . Briefly, a fragment containing the cdgD gene was obtained by PCR. The primers used for this purpose were 4571RC-F and 4571DC-R (Supplementary Table S2), and amplification was performed with Highfidelity Platinum Taq DNA polymerase (Invitrogen Thermo Fisher Scientific Life Technologies, Carlsbad, CA, USA). The fragment was cloned into the pCR 2.1 TOPO vector (Thermo Fisher Scientific Life Technologies) to yield the pCRcdgD plasmid. The pCRcdgD plasmid was digested with the restriction enzyme EcoRI, and the fragment was cloned into pSUP202, digested with the same restriction enzyme 18 , to yield pABcdgD. The km R gene was released from the pBSL98 plasmid with the restriction enzyme SmaI 19 and inserted into pABcdgD, which had been digested with the same restriction enzyme; this construct was named pABcdgD::km R . The obtained plasmids were transformed into E. coli S17.1 cells 18 . For conjugation assays, A. brasilense cells were grown on LB* solid medium 16 . Transconjugants were selected in K-malate medium supplemented with Km at 50 μg/mL. The obtained mutant was verified by PCR and sequencing, which showed that the ORF was disrupted. The A. brasilense C-56A complemented mutant was generated from a construct of the cdgD gene with its native region promoter obtained with the primers F-ORF210 (HindIII) and R-ORF-210 (XhoI), containing the restriction enzyme sequences indicated for direct cloning in the pVK100 broad host plasmid 20 , which was digested with the same restriction enzymes to yield the construct pVKpcdgD. Next, this construct was transformed into E. coli DH5α cells and subsequently to the E. coli S17.1 (pVKpcdgD) donor. This plasmid was then transferred by conjugation to A. brasilense 12-A. A control strain carrying the empty plasmid was also constructed to generate the A. brasilense C-40A strain. All strains were verified by PCR and sequencing using the described corresponding primers (Supplementary Table S2).
Construction of the A. brasilense FPDm1 chromosomal transcriptional fusion for abiotic and plant assays. The pAZBR-T7mCh plasmid was constructed from the pJMS-Km R suicide vector 21 as follows.
The CMP-PstIF/CMP-KpnIR primers (Supplementary Table S2) were used to amplify a 5′ fragment of 700 bp that included the ORF (GenBank accession number CCC96879.1, encoding a membrane protein of unknown function), which was then cloned into pGEM-TEasy (Promega, Madison, WI, USA) to generate the pAB79 plasmid. Subsequently, the MTF-SpeIF/MTF-NcoIR primers (Supplementary Table S2) were used to amplify a 3′ fragment of 900 bp that include the ORF (GenBank accession number CCC96880.1, encoding a putative SAM-dependent methyltransferase), which was further cloned into pGEM-TEasy to yield the pAB80 plasmid. Both fragments were excised from the plasmids with the corresponding restriction enzymes and cloned into pJMS-Km R to obtain a suicide plasmid with the mCherry gene without a promoter. Subsequently, the amplicon of the terminator of the T7 sequence (T7 iorA, Supplementary Table S2) was introduced between the mCherry gene and the CCC96879.1 gene to guarantee the transcription termination of the ORF encoded by this gene. This plasmid was designated pAZBR-T7mCh ( Supplementary Fig. S1). In this plasmid, a fragment 456 bp upstream of the cdgD gene, including its promoter, which had been previously amplified with the primers P-cdgD-F/ (SnaB1) and P-cdgD-R/(XhoI) (Supplementary Table S2), was cloned into the 5´ region of the mCherry gene, which was previously cloned into pGEM-TEasy to yield the pGEM-pcdgD plasmid. Both the pAZBR-T7mCh www.nature.com/scientificreports/ and pGEM-pcdgD plasmids were digested with XhoI and SnaB1, followed by cloning to obtain the pAZBRpcdg-DT7mCh plasmid, which was verified by PCR and sequencing. The pAZBR-T7mCh and pAZBRpcdgDT7mCh constructs were subsequently transformed into the mobilizing E. coli S17.1 strain, and the resultant strains were then transferred by conjugation into the A. brasilense Sp245 strain (cloned into a neutral locus of the A. brasilense chromosomal region) to obtain the control and tagged mCherry strains, respectively. These last two strains were used in further experiments to determine the expression of the cdgD gene under culture conditions and in wheat plants (Supplementary Table S1 and Fig. S1).

Application of bioinformatics tools for the analysis and modeling of the CdgD protein. The
CdgD protein sequence from A. brasilense Sp245 was retrieved from the NCBI database (https ://www.ncbi. nlm.nih.gov/prote in; accession number; AZOBR_100210). The Rapid Annotation using Subsystem Technology (RAST, http://RAST.nmpdr .org) server was employed for the initial analysis. Protein domain predictions were made using SMART 22 (http://smart .embl-heide lberg .de), HMMER 23 and PROSITE 24 . To predict and identify transmembrane regions, the TMHMM 25 and TMpred tools were used 26 . The visualization of the transmembrane helices was performed with the HeliQuest helical wheel-drawing program 27 . Secondary and three-dimensional structure predictions were performed using the Phyre2 and I-Tasser programs to detect the domain organization and to identify a suitable template fold for the CHASE and PAS-GGDEF-EAL domains 28 . Three-dimensional models of the CHASE (residues 64-327) and PAS-GGDEF-EAL (residues 378-946) domains were obtained using the Phyre2 and I-Tasser packages and analysed with Chimera software 29  Motility assays. The motility assays were performed essentially as described previously 21 in media containing 0.25% agar and K-malate, K-fumarate and K-proline as a source of carbon at a 10 mM concentration. The wild-type, mutant, and complemented mutant were evaluated by measuring their motility in cm in three independent experiments, each of which included three technical replicates.
Determination of extracellular polymeric substance (EPS) production. EPS quantification was performed as previously described 21 . Briefly, the A. brasilense WT strain, its isogenic mutant, and the complemented strain were grown in NFB* medium supplemented with KNO 3 or NH 4 Cl as a nitrogen source to an OD at 590 nm of 0.01 and subsequently grown with shaking at 30 °C until they reached an OD at 590 nm of 1.2-1.4. These cultures were incubated at 30 °C under static conditions for 3 days. The cells (2 mL) were harvested by centrifugation at 10,000 rpm (Biofuge-Heraeus, Thermo Fisher Scientific) for 3 min and resuspended in 1 mL of NFB* medium, to which a 0.005% (w/v) CR colorant solution (Sigma-Aldrich, Chemical) was added to achieve a 40 µg/mL concentration. The cells were incubated with shaking (200 rpm) for 2 h. Next, the samples were pelleted by centrifugation at 10,000 rpm. The amount of CR remaining in the supernatant was determined by measuring the OD 490 nm of the solution and comparing the result with the calibration curve of CR (concentration from 10 to 300 µg/mL) to obtain the µg CR. The protein concentration was determined via the Bradford protein assay. CR binding was expressed as µg CR/mg protein. Each culture was evaluated by measuring the EPS content in three independent experiments, each of which included three technical replicates.
Determination of cyclic-di-GMP levels using a riboswitch biosensor. Azospirillum brasilense strains harboring the pFY4535 plasmid containing the c-di-GMP biosensor 34 were constructed via the conjugation of E. coli S17.1 (pFY4535) to A. brasilense Sp245, A. brasilense 12-A, A. brasilense C-56A and A. brasilense C-40A as described previously 16 . Next, the reporter strains were streaked onto a Congo red agar plate containing 30 μg/ml gentamicin and incubated for 48 h at 30 °C. Two colonies from each plate were inoculated into NFB* + KNO 3 broth containing 30 μg/mL Gm and incubated for 24, 48, and 72 h at 30 °C with shaking at 120 rpm. The Azospirillum cells were mounted on a coverslip and sealed with a 1% agar plug. Images were acquired using a Nikon TE2000U microscope equipped with a 100× objective (oil immersion objective). For the AmCyan fluorophore, excitation was performed at 457 nm, and the emission spectrum was obtained at 520 nm (green). For the TurboRFP fluorophore, the excitation spectrum was obtained at 553 nm, and emission was measured at 574 nm (red). The images were edited and analysed with Nikon NIS Elements and ImageJ RNA extraction and quantitative RT-qPCR. Azospirillum brasilense Sp245 was grown in NFB* medium supplemented with KNO 3 or NH 4 Cl as a nitrogen source 33 under planktonic or biofilm conditions for 10, 18, 24 or 48 h as described above. After the cells from 6 mL of the culture were pelleted and washed with TE (10 mM Tris HCl, 1 mM EDTA) buffer, pH 8.0, the obtained pellets were stored at − 70 °C until use. Total RNA was extracted from the pellets, resuspended in 550 µL of TE buffer plus 200 µg/mL (w/v) lysozyme and lysed by sonication at 30% for 5 s (Hielscher UP50H-Compact Lab Homogenizer, 12623 Berlin Germany). The cell suspension was maintained at 37 °C with shaking for 35 min, and 50 µL of 10% SDS (w/v) plus 400 µL of cetyl trimethyl ammonium bromide (CTAB 2% w/v in NaCl 2 M; CTAB, Sigma-Aldrich) was added. After mixing, the suspension was centrifuged at 10,000 rpm for 10 min. The supernatant was treated by mixing with 150 µL of phenol-Tris-HCl, pH 8.0, and RNA was extracted four times with 400 µL of chloroform/isoamyl alcohol (v/v 24:1). After that step, the RNA was precipitated with ethanol and centrifuged at 10,000 rpm for 10 min. The RNA was resuspended in 50 µL of water treated with diethylpyrocarbonate (DEPC, Sigma-Aldrich) at a concentration of 0.1% (w/v) and stored at − 70 °C until use. The concentration and quality of the RNA were determined using an EON-BioTeK spectrophotometer (BioTek Instruments Inc.). RNA integrity was confirmed by electrophoresis in 0.8% (w/v) agarose gels. Complementary DNA (cDNA) synthesis by reverse transcription (RT-PCR) was performed using the Maxima First Strand cDNA Synthesis system (Thermo Fisher Scientific) from 1 µg of extracted RNA. For RT-qPCR analysis, three liquid culture biological replicates were performed for each gene (cdgD and glyA). Specific RT-qPCR primers (qRT-cdgDF, qRT-cdgDR, and glyAF, glyAR 36 ); were used to amplify the reference and target genes. The reactions were carried out in triplicate using SYBR Green High ROX master mix (Thermo Fisher Scientific, USA). The thermocycling conditions were as follows: 95 °C for 3 min, 40 repetitions of 95 °C for 30 s and 60.5 °C for 30 s, followed immediately by a melting curve, in a StepOnePlus thermocycler (Applied Biosystems, Foster City, California, USA). Gene expression was normalized to the level of the endogenous housekeeping gene glyA and to the wild-type RNA reference sample following the 2 −ΔΔC T method. The standard curve was constructed (in duplicate) using chromosomal DNA. Melting curve analysis was used to confirm the production of a specific single product by each primer pair.

Plant growth conditions of the A. brasilense strains, inoculation, and visualization of colonization by confocal laser scanning microscopy (CLSM).
The plant experiments were performed essentially as previously described 37 using seeds of the wheat (Triticum aestivum) variety 'Nana' and soil obtained from the Campo Agrícola Experimental del Valle de México (INIFAP). Briefly, the seeds were surface-sterilized with 1% (v/v) sodium hypochlorite with agitation (60 rpm) for 30 min after rinsing with sterile water and were then immersed in 40 mL of a solution containing 150 μg/mL cycloheximide, 250 μg/mL streptomycin, 20 μg/mL Tc, and 150 μg/mL fluconazole, with agitation at 60 rpm for 10 min. At the end of the disinfection process, the seeds were rinsed five times with demineralized sterile water for 10 min. Regarding seed inoculation, the A. brasilense Sp245, A. brasilense 12-A, A. brasilense C-40A and A. brasilense C-56A strains harboring the pMP2449-5 plasmid were grown on K-malate medium overnight at 30 °C with shaking (150 rpm). The bacterial cells were resuspended in K-malate medium at a final density of 10 9 colony-forming units (CFU)/mL (OD 600 = 1.2). The bacterial suspensions were diluted 100-fold in sterilized NFB* medium to a final density of approximately 5 × 10 6 -5 × 10 7 CFU/mL. The inoculated seedlings were maintained in tubes with 15 mL of Hoagland hydroponic solution supplemented with NH 4 NO 3 (4 mM) in a growth chamber for 7 days under a day-night cycle of 14 h light at 24 °C and 10 h darkness at 16 °C with humidity of 80% as previously described 37 . At 7 days postinoculation, the colonizing bacteria were evaluated by microscopy analysis, and root samples were cut, soaked in sterilized H 2 O, placed on glass slides with PBS and covered with glass coverslips. Confocal images were obtained using a Nikon C2 + CLSM (Nikon, Tokyo, Japan) equipped with a CFI Plan Apo Lambda 20× objective and two helium-neon lasers for the excitation of the mCherry fluorophore at wavelengths of 540 nm and 650 nm and an argon laser for the excitation of autofluorescence at 488 nm. The cell morphology and localization of the A. brasilense strains (sample and control) were evaluated in the roots to assess the presence of the bacteria and biofilms. cdgD gene promoter expression in wheat plants determined by confocal microscopy. Nana wheat seeds were sterilized as described in the previous section and germinated on sterile 1.0% H 2 O-agar for 72 h in the dark. The seedlings were then transferred to FluoroDish glass-bottom Petri dishes containing Hoagland hydroponic solution and placed in an environmental chamber (24 °C, 14 h light and 16 °C, 10 h dark photoperiod). The A. brasilense T7mCh and A. brasilense FPDm1 strains were grown overnight in LB* liquid medium, and the cultures were subsequently diluted in phosphate buffer and used to inoculate 3-day-old seedlings at 1 × 10 7 CFU/mL. The samples were visualized by confocal microscopy for 24 or 48 h to evaluate cdgD gene promoter expression in association with wheat plants. Confocal images were acquired using a Nikon C2 + CLSM (Nikon, Tokyo, Japan) equipped with a CFI Plan Apo Lambda 20× objective and two helium-neon lasers for the excitation of the mCherry fluorophore at wavelengths of 540 nm and 650 nm and an argon laser for the excitation of autofluorescence at 488 nm. The Nikon NIS Elements program provided by Nikon was used to acquire, analyse, and edit the images.
Plant colonization experiments under soil conditions. Wheat seed disinfection and the inoculation of seedlings 3 days after germination were performed as described above. The seedlings were placed in an environmental chamber maintained under a day-night cycle as described above for 3 days. Next, the seedlings were inoculated with 1 × 10 6 -1 × 10 7 CFU/mL of the A. brasilense Sp245, A. brasilense 12-A, A. brasilense C-56A www.nature.com/scientificreports/ or A. brasilense C-40A strain, grown in a 50 mL flask with sterile soil for 7 days postinoculation and handled as described above. Subsequently, the plants were carefully washed to remove all adhered soil with sterile water for 40 s. The colonizing bacteria were quantified by colony counting in CR medium supplemented with the appropriate antibiotic (WT, Ap, 150 µg/mL; A. brasilense 12-A, Km, 50 µg/mL; C-40A or C-56A, Tc, 20 µg/mL).

Statistical analyses.
For all experiments, the overall data were analysed by Student's t-test, to determine whether the mutant significantly affected all phenotypes tested and the expression of cdgD gene under the analysed growth conditions as well as the colonization tests. The difference were considered to be significant when the p value were < 0.05. Statistical analyses were performed using Sigma plot 11.0 software (Systat software, Sn Jose, CA).

Results and discussion
Homology modeling and in silico analysis. While biofilm formation is often described as a natural mode of microbial growth and the second messenger c-di-GMP is known to be a central molecule regulating the microbial transition from a planktonic (motile) to sessile (biofilm) lifestyle 4,15,38 , it is clear that we still lack knowledge regarding the genes encoding the proteins responsible for cyclic-di-GMP biosynthesis and degradation and the roles that they play in host plant colonization by A. brasilense. An initial analysis was conducted to confirm the domains constituting the translational product encoded by the cdgD gene (GenBank accession number AZOBR_100210) and further used for structural modeling. Alignments were generated and analysed using the Clustal Omega server (www.ebi.ac.uk/Tools /msa/clust alo/) against equivalent domains known to be catalytically active for which a structure has been deposited in the PDB; this approach employed to compare the motifs of the translational product of the cdgD gene to the WspR of Pseudomonas aeruginosa or the PleD of Caulobacter crescentus diguanylate cyclase as a reference for the amino acid motifs of the GGDEF protein domain 39,40 and to RocR from P. aeruginosa as a reference for the EAL-containing domain 41 to identify the GGDEF and EAL domains in the CdgD hybrid protein. The identified GGDEF domain consists of highly conserved sequence motifs that include the D/E catalytic site and other residues involved in the substrate binding and coordination of one of the two divalent cations 41 . Moreover, we identified a motif acting as a regulator of DGC function, referred to as the I site, which was determined to contain an RXXD motif immediately preceding the GGDEF catalytic motif 42 (Supplementary Fig. S2a,b). The EAL domain also exhibits all characteristics of the amino acid motif required for the binding of divalent cations and the substrate. In addition, a flexible loop (loop 6) that mediates dimerization and controls substrate and cation binding, which are required for catalytic activity, was also found 42,43 , ( Fig. 1 and Supplementary Fig. S2a,b). Notable characteristics were also observed in the cytoplasmic region of CdgD, including an α-helical lever (H-helix) that joins the GGDEF and EAL regions as well as a signaling α-helix (S-helix) located immediately adjacent to the GGDEF domain ( Fig. 1), which has been shown to be important in the folding and regulation of PDE function of RbdA from P. aeruginosa 32 . Therefore, we might categorize both domains as functional based on the presence of conserved amino acid motifs in the active sites of the two domains 11,12 . Moreover, the CdgD protein contains three other domains at the N-terminus, including two transmembrane (TMD) domains, a CHASE domain, and a PAS domain, which is a sensor domain that modulates activities in response to external stimuli such as light or small ligands, including O 2 , NO, and CO 44 . This PAS domain was identified using the structure of the cytoplasmic region (cRbdA) of RbdA from P. aeruginosa as a model that exhibits a highly conserved structural architecture in the cytoplasmic region of the CdgD protein (residues 241-557; PDB Code: 5XGB, resolution 2.28 Å, TM-score 0.918, RMSD 0.79 Å) 32 , (Supplementary Table S3, Figs. S3, S4, S5 and S6).
It has been reported that these GGDEF and EAL domain-containing proteins often harbor a diverse array of additional domains that are important for the biochemical function of the specific protein, such as a CHASE domain, which is a periplasmic domain that is defined as a sensory domain and predicted to play a role in small molecule recognition in a wide range of proteins and various bacterial chemotaxis receptors 45,46 and may be a ligand-binding domain (LBD) 47 , (Fig. 1). This CHASE domain was located in the N-terminus of the protein between two TMDs (including 264 amino acid residues); according to the presence of these predicted transmembrane segments, the protein may be associated with the cytoplasmic membrane. Structural analysis of the CHASE domain was performed using the same domain found in Xanthomonas campestris pv campestris histidine kinase PcrK (residues 39-302; PDB Code: 6K62, resolution 2.55 Å, TM-Score 0.955, RMSD 0.82 Å) 30 as a reference model ( Fig. 1 and Supplementary Table S3, Figs. S3a,b and S4a,b). The structure showed the highest-confidence resolution scores (TM 0.928 and RMSD 0.74 Å) and fit well with the CHASE domain, and it was recognized as a structural homologue of the CdgD protein, a notably similar structure showing the predicted formation of a homodimer between the two TMDs, with the predicted function of a binding domain involved in periplasmic signaling (Fig. 1). Such a situation has also been described for the Arabidopsis thaliana histidine kinase (AHK4) 4-10 protein (PDB Q9C5U0) 31 , which has been experimentally shown to be able to bind with cytokinin and both models to provide structural patterns that match the CHASE domain of the CdgD protein well. It was also predicted that the CHASE domain is composed of two PAS domains (membrane-distal and membrane-proximal PAS domains) displaying the same numbers of ß-strands and α helices, even though CdgD sharing only 25% and 19% sequence identity with PcrK or AHK4, respectively (Fig. 1). These results suggest that the CdgD sensor domain adopts a conserved CHASE topology overall, which is consistent with previous findings 30,31 . The occurrence of CHASE and PAS domains has been observed in other hybrid proteins, in which the module mediating the signaling pathway has an effect on the regulation of the dual function of DGC-PDE proteins 48,49 .
Therefore, the high conservation of all these motifs in GGDEF and EAL domains, as determined through the sequence alignment and analysis of the three-dimensional (3D) structure predicted for the translational product of the cdgD gene suggest that it is likely bifunctional and may be classified as a class I CdgD hybrid protein 11 Table S1).

Phenotypic studies of the cdgD gene. Biofilm formation and extracellular polymeric substance (EPSs)
production. To elucidate the molecular basis of the role of the cdgD gene of the A. brasilense Sp245 strain in biofilm formation, motility and the colonization of wheat roots, we constructed the A. brasilense 12-A mutant. Sequence analysis of this mutant showed that the insertion of the Km resistance cassette interrupted the ORF of the translational product. Additionally, we constructed a corresponding complemented mutant strain harboring the wild-type cdgD gene to compare the effect of the mutation on biofilm formation and motility against the WT strain. The different strains and plasmids that were constructed and used in this study are shown in Supplementary Table S1. To characterize the differences between A. brasilense Sp245 and the derived strains, the strains were grown in NFB* medium supplemented with two different nitrogen sources and cultured under static conditions. Under such conditions, we observed that biofilm formation by the WT was greater in NFB* + KNO 3 than NFB* + NH 4 Cl medium (Fig. 2a,b). However, when the bacteria were grown in each type of nitrogen NFB* medium, a marked decrease in the production of biofilm was observed in the mutant compared with the WT, which was recovered in the complemented strain. Extracellular polymeric substances have been reported 4,6,50,51 to play essential role in the formation of the biofilm matrix. To quantify extracellular polymeric substances such as exopolysaccharides, extracellular DNA (eDNA) and amyloid compounds, we tested their capacity to bind the colorant Congo Red (CR) 8,51-54 , and the production of these components was assessed in the strains growing in NFB* medium supplemented with either KNO 3 or NH 4 Cl. As anticipated, the mutant growing in NFB* + KNO 3 medium showed a decrease in EPSs production (3.2-fold reduction) compared with the WT, whereas EPSs production similar to that in the WT strain was observed in the complemented strain (Fig. 2c). This data showed that the decreases in biofilm production and extracellular polymeric substance production observed in the mutant strain were corelated in NFB* + KNO 3 medium (Fig. 2a,c). The mutation of the predicted CdgD, which contained a highly conserved www.nature.com/scientificreports/ GGDEF motif, decreased biofilm formation and affected the production of EPSs matrix components, as determined by quantification based on staining with CV 53 and CR colorant binding to components of the matrix structure of biofilms 51,54 , this results are consistent with this protein being a DGC. However, an unexpected result was obtained when the strains were grown in NFB* + NH 4 Cl medium. In effect, the mutant showed an increase in EPS production of 1.6-fold compared with the WT strain (Fig. 2d). In contrast to the data obtained on biofilm formation, in the case of extracellular polymeric substance production, a significant increase was notably observed in the mutant in medium supplemented with NH 4 Cl. Furthermore, when the EPSs quantification data for the WT strain were compared to analyse growth in NFB* + KNO 3 and NFB* + NH 4 Cl (13.13 µg/mg protein compared with 23.4 µg/mg protein, respectively), 1.7-fold higher production was observed in the medium supplemented with ammonium (Fig. 2c,d). Similar results were obtained for cells of the 12-A mutant growing under the same conditions (4.1 µg/mg protein versus 37.3 µg/mg protein), corresponding to 9.1-fold higher production. These data indicated that the nitrogen source causes changes in cell surface properties that contribute to CR binding (Fig. 2c,d). Therefore, the data obtained in each type of nitrogen-supplemented NFB* medium suggest a differential contribution of the cdgD gene to cell surface properties, as demonstrated by changes in binding to CR. These results are notable because amongst the EPS, the production of extracellular polysaccharides in flocs plays an important role in the infection bacterial ability to colonize the surface of plant roots 2,15,54-57 . Furthermore, the nitrogen source had an effect on biofilm formation, which was more significant in presence of KNO 3 compared to NH 4 Cl (Fig. 2). These data are in keeping with findings obtained in both A. brasilense 58,59 and other bacterial species such as Pseudomonas stutzeri A150 52 showing that the nitrogen source affects the formation of biofilms and the production of EPSs, particularly the exopolysaccharide components that allow nitrogen fixation by the bacterium 52,59 . Moreover, in Cellulomonas spp., nitrogen source limitation induces biofilm formation, which is coupled to the biosynthesis of a curdlan-type biofilm matrix. However, when a nitrogen source is available, the biofilm curdlan matrix is solubilized and used as a carbon and energy source 60 . In A. brasilense, different classes of exopolysaccharides are produced under high carbon source and low nitrogen source availability, which were the conditions applied in this study 56,58 . Furthermore, in terms of functional redundancy, several genes presumably encoding DGCs, PDEs or DGC-PDEs have been described in A. brasilense 12,21 , this gene products could provide compensatory effects that could be responsible for the production of different exopolymeric substance contributing to the changes in cell surface properties observed in cultures grown in medium supplemented with NH 4 Cl Motility assay. Next, we tested the ability of the 12-A mutant and the complemented C-56A mutant to swim in semisolid agar minimal medium supplemented with malate, fumarate or proline as a carbon source. All strains had been previously tested in NFB* + KNO 3 medium and showed similar growth (Supplementary material Fig. S7). All rings sizes formed by the 12-A mutant or the C56-A are larger than those formed by the wild type strain in all carbon sources tested, however when added to the K-minimal medium, the effect of proline was more pronounced (3.4 times greater than in the WT strain) (Fig. 3a,b).The constructed complemented mutant was generated in the intermediate-copy-number vector pVK100, under which condition the cdgD gene was overexpressed and cells grown on soft agar plates exhibited increased swimming motility. Furthermore, cells grown in broth medium exhibited distinct swimming patterns, and the cells of WT strain reduced their motility and began to form small flocs, whereas cells from 12-A mutant or complemented mutant C56-A (Supplemental Movies S1, Movie S2, Movie S3 and Movie S4) continued to move. It has been reported that aerotaxis is a major behavioral response in A. brasilense, and the bacterium depends on sensing oxygen gradients. It has also been shown that flocs are formed when oxygen is increased 3,59,61 . Therefore, we suggest that WT cells sensed oxygen more efficiently than the 12-A and C56-A cells did and that they continuously moved mostly without changing the direction of swimming; however, that behavior warrants further research. In addition, it has also been reported that motile cells of A. brasilense undergo chemotaxis towards several nutrients found in root exudates, which has been shown to be an essential prerequisite for efficient root colonization 61 . CHASE domains have been observed to sense environmental signals, such as chemoattractants, including organic acids, amino acids and other compounds that are found in root exudates 62,63 . The ability of proline to increase A. brasilense motility might be due to the presence of the CHASE domain, which is also able to bind amino acids and other small molecules 47 . Alternatively, other DGCs/PDEs may produce the phenotype in the absence of cdgD and perhaps even amplify the effect of CdgD on C-56A. Nevertheless, future studies will clearly establish the nature of the signal recognized by the predicted ligand-binding domain (LBD) of the CHASE domain and elucidate the ligand and signal transduction mechanism.
Determination of cellular cyclic-di-GMP levels using a riboswitch biosensor. Taken together, the results suggest that CdgD is capable of c-di-GMP synthesis, thereby validating its contribution, and indicate that this function is important for biofilm formation and might modulate extracellular polymeric substance production. We estimated the intracellular levels of c-di-GMP in the wild-type (Sp245), 12-A mutant, C-40A control carrying www.nature.com/scientificreports/ out the vector empty and complemented mutant (C-56A) strains using a fluorescent c-di-GMP reporter. This analysis was performed by quantifying the riboswitch fluorescence intensity, which was directly proportional to the c-di-GMP levels 34 . As expected, a low c-di-GMP level was determined in the 12-A mutant compared with the wild-type strain; an approximately 61% or 24% reduction in c-di-GMP levels was observed in the 12-A mutant strain compared with WT cultures at 24 h or 48 h, respectively, which was fully recovered in the complemented mutant. However, there was no significant difference in intracellular c-di-GMP levels compared with the WT strain in the cells after 72 h of growth, potentially due to compensatory effects of other DGCs occurring in the genome of the A. brasilense WT strain. In accordance with the data regarding biofilm and EPSs component production from the CR assay, the mutation reduced the levels of c-di-GMP in cells growing in NFB* + KNO 3 medium. These data indicate that overexpression of CdgD allows a temporary intracellular second messenger levels in A. brasilense (Fig. 4). Although bioinformatic analysis showed that all amino acid motifs were conserved in the EAL domain, the phenotyping studies were unable to show that CdgD functioned as a PDE protein. , and C-40A strains containing the c-di-GMP biosensor (pFY4535) were grown in NFB* + KNO 3 broth containing 30 μg/mL Gm with incubation for 24, 48, or 72 h at 30 °C. Then, Azospirillum cells were attached to the surface of a coverslip and sealed with a 1% agar plug. Cell images were collected at these time points after inoculation using a Nikon TE2000U microscope equipped with a 100× objective (oil immersion objective). Merge images represented the overlay of the fluorescence images AmCyan green, TurboRFP red and both yellow. (b) RFI represents the ratio between the TurboRFP and AmCyan fluorescence intensities and is directly proportional to c-di-GMP levels, as analysed using ImageJ software 35 . The RFI values represent the standard deviations of three biological replicates, and significant differences are indicated at *P < 0.05 according to Student's t-test by SigmaPlot (Systat Software, San Jose, CA). The bar corresponds to 10 µm. Both strains showed similar growth to the wild-type strain, as determined by the 10 9 CFU/mL value obtained in the exponential phase of growth (18 h), (Fig. 5). In both media, the expression of cdgD was initiated at 10 h and continued until 48 h, in cultures growing in planktonic and static conditions. These observations indicate that the expression of the gene is initiated at the beginning of bacterial growth and continues until the later stages of growth. The amount of fluorescence produced was higher in cells growing in NFB* + KNO 3 medium comparative to NFB* + NH 4 Cl, (Fig. 5b), indicating that the addition of KNO 3 to the medium increased transcript levels relative to those obtained in presence of NH 4 Cl, where expression was further increased when the structure of  www.nature.com/scientificreports/ the biofilm was structurally mature (48 h) (Fig. 5a). Such data indicate a role of the nitrogen source in determining gene expression, which may be direct or indirect. In addition, the data were correlated well with both biofilm production (Fig. 2a) and the production of extracellular polymeric substances (Fig. 2c).This suggests a function for CdgD as a DGC under static culture conditions, which is also supported by elevated cellular c-di-GMP levels (Fig. 4). Considering the data obtained in this study, we suggest that the regulation of the cdgD gene by nitrogen source might occur at the transcriptional level with CdgD protein activity as either a DGC or PDE or regulated at the post-translational level, as observed in other bacteria 10,13,14 . However, how signaling involving the CHASE domain modulates the DGC-PDE functions of the CdgD protein remains to be studied. Next, we attempted to monitor cdgD gene promoter expression during the colonization of wheat roots under hydroponic medium conditions by using the tagged chromosomal A. brasilense FPDm1 strain and A. brasilense T7mCh, carrying the mCherry gene devoid of the promoter as a negative control. Both strains showed similar growth to the wild-type strain, as determined by the 10 9 CFU/mL value obtained in the exponential phase of growth (18 h). To visualize the localization of CdgD in vivo, we expressed the mCherry gene fused upstream of the promoter of the cdgD gene. In effect, this hypothesis was verified because A. brasilense FPDm1 cells that were recovered following 48 h of exposure to the root environment of wheat seedlings showed significant fluorescence in cells maintained on root hairs, and rhizoplane-forming microcolonies were observed, which represent a preliminary step in biofilm formation; zones exhibiting the formation of a mature biofilm were also observed, indicating the expression of the cdgD gene, which was clearly lacking in A. brasilense T7mCh (Fig. 6). These findings suggest that the activity of the encoded protein is relevant to the early stages (48 h) of colonization in our model rhizosphere assay and that the bacterium is able to enter the roots and thereby colonize the apoplast tissue of wheat roots. In addition, our data suggest that biofilm formation is stimulated as a response to wheat rhizosphere colonization (Fig. 6). This results is in keeping with data obtained in the plant-growth-promoting strain Pseudomonas fluorescens, in which genes involved in the metabolism of c-di-GMP are expressed during the colonization of the wheat rhizosphere 14,38 , as well as data from Pseudomonas putida KT2440, in which the promoter of the rup4959 gene, encoding a dual GGDEF-EAL protein, was found to be expressed in the maize rhizosphere and upregulated by corn plant root exudates 13 .
We further tested the role of the cdgD gene in the colonization of wheat roots, and we estimated endorhizosphere colonization by the A. brasilense 12-A mutant and compared it with colonization by the WT and complemented A. brasilense C-56A mutant strains. Each strain was labeled with the autofluorescent mCherry protein, which was carried by a stable plasmid maintained in A. brasilense cells 37 . The plants were grown in glass tubes with Hoagland hydroponic solution and observed by confocal microscopy at 7 days postinoculation. Microscopy observations of these reporter strains indicated that all strains were able to colonize the rhizoplane and root hairs of wheat plants and form biofilms (Fig. 7a). However, differential colonization was visualized in the A. brasilense 12-A mutant and its C-40A control strain harboring the empty vector, which exhibited significant rhizosphere colonization. Next, the colonization of the A. brasilense 12-A mutant was quantified under two conditions: under hydroponic growth and in sterile soil and compared to the fitness of the WT and the complemented mutant.  www.nature.com/scientificreports/ The colonization of each strain was determined by counting the colonies obtained at 5 days postinoculation. Nonsignificant differences in fitness and the impairment of colonization were determined in the tested strains under most controlled hydroponic conditions (data not shown). However, the A. brasilense 12-A mutant and its control strain harboring the empty vector showed a statistically low ability to colonize wheat roots compared with the WT and complemented mutant C-56A strains tested in sterile soil (Fig. 7b), implying that the cdgD gene is involved in the colonization of wheat roots. However, in our study, the cells of the tagged mutant were observed more frequently than those of the WT and complemented mutant strains in the rhizoplane and rhizosphere (Fig. 7a). This result was surprising because A. brasilense Sp245 is an endophytic strain that has been found in the apoplast of plants 63,64 . A different colonization profile was observed in wheat roots associated with the 12-A mutant strain versus the Sp245 or C-56A strain. This effect could be explained by considering that cellulolytic activity might be crucial to the ability of bacteria to penetrate plant roots. Putative complex carbohydrate-degrading enzymes have been found in the analysed genome sequences of three Azospirillum species, conferring greater cellulolytic activity 64 . Therefore, these data strongly suggest that the CdgD protein plays a role in bacterial entry into the host plant; however, further research are needed to test this hypothesis.
In addition, under growth in sterile soil conditions, we tested the contribution of the cdgD gene to the colonization of the rhizosphere by comparing the ability of the mutant versus the WT and complemented mutant strains to colonize the roots of wheat seedlings. The results indicated less colonization by the mutant compared to the WT and complemented mutant strains, suggesting that unidentified factors, other than motility and biofilm formation, are affected in the 12-A mutant and contribute to the fitness of A. brasilense in the rhizosphere colonization (Fig. 7b).

Conclusions
Azospirillum brasilense is a PGPB that has been shown to promote plant growth in several host plants and must therefore successfully colonize the plants, but how colonization is established by the bacterium has not been fully elucidated. Single-gene mutations in predicted related genes encoding DGC and PDE proteins do not result in striking changes in phenotypes that are typically associated with these proteins, such as swimming motility, the production of extracellular polymeric substance components of biofilm matrixes, or host colonization in other bacteria [13][14][15]38,49,52,63 . Here we show that the inactivation of a gene encoding a hybrid protein that likely behaves as a DGC to decrease c-di-GMP cellular levels functions in motility, production of extracellular polymeric substance components of biofilm matrixes, or plant root surface colonization, indicating its participation in key functions implicated in the interaction of the bacterium with the host plant. Our results also highlights that nutritional status, such as observed here in response to nitrogen availability, has a critical and differential effect on biofilm formation and the production of surface-exposed polymeric substances by A. brasilense cells. The increase in the diameter of the swimming motility ring is observed in the presence of proline as a carbon source in the parent and mutant strains overexpressing cdgD from a plasmid. These data suggest proline may affect CdgD protein function, possibly through its CHASE signaling domain. Motility mediated by the polar flagellum is critical to the performance of the bacterium swimming in soil water, its chemotaxis towards root exudates and its subsequent attachment to the rhizoplane of root plants, thereby maintaining the sustainable growth of its community in the natural environment. After biofilm formation, the surrounding extracellular matrix provides a structure for the maintenance of biofilm cells, keeping them in long-term close proximity; thus, intense interactions occur with such effects as protecting the community from stresses, such as desiccation, contact with biocides and heavy metals, and ultraviolet radiation, which is essential in the competitive habitat of the soil. An mCherry transcriptional reporter fused to the promoter of the cdgD gene confirmed that expression of the cdgD gene was increased in the stationary phase under abiotic and biotic conditions in vitro and during the association of A. brasilense with wheat roots. This result is consistent with the colonization assay, suggesting that CdgD functions to promote an association with plant roots strong interaction that substantially contributes to better adaptation to changing environmental conditions and guarantees the beneficial effects of the bacterium on the plant host.

Data availability
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation, to any qualified researcher.