The bacterial strains used in this work are described in Table 1. Escherichia coli strains were grown at 37°C in lysogeny broth (LB) supplemented with 100 μg/mL ampicillin (Ap), 15 μg/mL tetracycline (Tc), 50 μg/mL kanamycin (Km), chloramphenicol (Cm), 10 or 30 μg/mL gentamycin (Gm) when needed. E. coli DH5α and E. coli S17.1 were used for transformation and conjugation assays, respectively. The Azospirillum baldaniorum Sp245 used in this study was kindly provided by Dr. C. Elmerich (Institute Pasteur Paris France). The Azospirillum strains were grown at 30°C in K-malate minimal medium [(containing KH₂PO₄, 0.87 g; K₂HPO₄, 1.67 g; NaCl, 0.2 g; MgSO₄⋅7H₂O, 0.01 g; MnSO₄⋅H₂O, 0.01 g; FeCl₃, 0.01 g; Na₂MoO₄⋅H₂O, 0.02 g) L^–1]; pH 6.8, malic acid, 37.2 mM, and NH₄Cl 18.6 mM as carbon and nitrogen source, respectively (Ramírez-Mata et al., 2016), Congo red (CR) (Rodriguez Caceres, 1982), LB^∗ (Lysogeny broth supplemented with 10 mM, CaCl₂, and 10 mM, MgCl₂), supplemented with Tc, Km or Gm (15, 50, and 30 μg/mL, respectively), depending on the required, Nitrogen-Free Medium [(NFB, malic acid as carbon source, 4 mL vitamin solution (biotin, 10 mg; pyridoxal-HCl, 20 mg), Fe-EDTA (0.065 g L^–1), and micronutrient solution (CuSO₄.5H₂O, 0.04 g; ZnSO₄.7H₂O, 0.12 g; H₃BO₃, 1.40 g; Na₂MoO₄.2H₂O, 1.0 g; MnSO₄. H₂O, 1.175 g) L^–1)] (Baldani et al., 1987) or NFB^∗ modified to achieve a relation C:N = 2 using malic acid at 27.6 mM and supplemented with13.8 mM KNO₃ as N source (Arruebarrena Di Palma et al., 2013) media.

The domain architecture of CdgC was analyzed using InterPro (Mitchell et al., 2019) and SMART (Letunic and Bork, 2018) databases. Protein sequence alignment was performed using Clustal Omega (Sievers et al., 2011) and Multiple Align Show (Stothard, 2000) software. WspR (PA3702) (De et al., 2008) from P. aeruginosa (PDB accession number 3BRE) and PleD (Wassmann et al., 2007) from C. vibroides CB15 (PDB accession number 2V0N) were used as reference sequences in pairwise or multiple sequence alignments. The secondary structure of CdgC was predicted using the Phyre2 (Kelley et al., 2016) software. Three-dimensional structures were modeled employing the I-Tasser server (Yang et al., 2014), and figures were made using the UCSF Chimera program (Pettersen et al., 2004). Logos of active and inhibitory sites were created with WebLogo 3 Create (Crooks et al., 2004).

The oligonucleotides and plasmids used are listed in Table 2. The deletion construct pCDR-ΔcdgC was generated through the assembly of upstream and downstream flanking regions of 509 nucleotides including the GGDEF domain of cdgC in the suicide plasmid pCDR. These sequences were amplified through PCR using primer pairs CdgC13/14 and CdgC17/18. Amplicons were cloned into the vector pGEM T-Easy (Promega) and subcloned into pCDR after digesting with the restriction enzymes (REs) EcoRI/XbaI and NotI/XbaI, respectively, and ligating using a T4 DNA ligase (Promega). Gene deletion was achieved through double homologous recombination as previously reported (Ramirez-Mata et al., 2018).

To generate a trans genetic complementation construct, the ORF of the cdgC gene and its upstream intergenic region were amplified with the oligonucleotide pair Pr100206F/Pr100206R and cloned into the low-copy-number vector pJB3Tc20 (Blatny et al., 1997). Both the amplicon and the destination plasmid were digested with the REs EcoRI and KpnI and ensembled using a T4 DNA ligase (Promega). The vector pAZBR-cdgC was generated to allow for cis genetic complementation of the ΔcdgC mutant strain by inserting cdgC and its promoter region in a neutral locus in the chromosome (Cruz-Pérez et al., 2021) of A. baldaniorum 59C. First, we constructed the pGEM-Gm^R plasmid by cloning into the pGEM-T Easy vector, the gentamycin resistance gene from the pBBR1MCS-5 vector (Kovach et al., 1994) which was amplified using the GmF and GmR primers. Next, we generated the suicide plasmid pAZBR-T7-Gm by subcloning the gentamycin resistance gene from pGEM-Gm^R into pAZBR-T7mCh (Cruz-Pérez et al., 2021) by digesting with the REs XhoI and SpeI. This plasmid enables the insertion of exogenous sequences through homologous recombination. The cdgC gene, and its native promoter were extracted from the plasmid pJB-cdgC using the REs SnaBI and EcoRI, this fragment was next cloned into pAZBR-T7-Gm. The pAZBR-cdgC construct was mobilized into A. baldaniorum 59C by biparental mating using E. coli S17.1 as a donor strain.

The pGEX-cdgC construct was generated to analyze the effect of the expression of cdgC on the c-di-GMP levels of the heterologous host E. coli S17.1. The cdgC gene was amplified with the cdgC-F and cdgC-R primers and the product was cloned into vector pGEM-T Easy generating pGEM-cdgC. The latter was digested with the REs EcoRI and SalI and cloned into the expression plasmid pGEX-4T-1 generating pGEX-cdgC. The pGEX-cdgC plasmid was then used to transform E. coli S17.1 competent cells harboring the c-di-GMP biosensor pDZ-119 (Martínez-Méndez et al., 2021).

The pMP2450 vector containing the gene encoding the enhanced green fluorescent protein (eGFP) was constructed for competence and wheat root colonization assays. This plasmid was generated by cloning the kanamycin resistance cassette from the pJMS-Km (Ramírez-Mata et al., 2016) vector into the vector pMP2444 (Clontech). The kanamycin resistance cassette was extracted using the RE XbaI and cloned into a unique XbaI site in pMP2444 to generate pMP2450. All amplicons obtained for cloning were sequenced at the Sequencing Unit at the Universidad Nacional Autonoma de Mexico (UNAM). All strains generated in this study are listed in Table 1.

Biofilm formation was performed employing NFB^∗ medium (C/N = 2) under static conditions, as previously described (Arruebarrena Di Palma et al., 2013), or under nitrogen fixation conditions using NFB medium. Briefly, a 3-day-old colony of Azospirillum strains (A. baldaniorum, A. baldaniorum 59C, A. baldaniorum C21, A. baldaniorum C87, A. baldaniorum-GFP, A. baldaniorum 59C-GFP, A. baldaniorum 2A65, and A. baldaniorum 2D8) grown in CR medium were inoculated into LB^∗ medium and incubated at 30°C with shaking at 150 rpm until the culture reached an optical density at 600 nm (OD600) of 1.1–1.4. Cells were harvested by centrifugation (5,000 rpm), washed with phosphate buffer (66 mM) pH 7.0, and concentrated to an OD600 of 2. Biofilm formation was determined by the crystal violet assay as previously reported (Arruebarrena Di Palma et al., 2013). Alternatively, for CLSM, the cells were diluted at 1:100 in NFB^∗ medium, and 3.9 mL of this suspension was transferred to a glass-bottom FluoroDish device (Thermo Fisher Scientific). Calcofluor-White colorant (CWC) (Sigma-Aldrich, United States) was added to a final concentration of 85 μM. Cultures were incubated under static conditions at 30°C for 5 days. Then, biofilm formation and exopolysaccharide production were evaluated with an inverted CLSM Nikon Eclipse Ti-E C2+ (Nikon Instruments, Tokyo, Japan) using a 60x Plan lambda objective. eGFP fluorescence excitation and emission were set at wavelengths of 488 and 509 nm, respectively. CWC was excited at 380 nm and its emission was acquired at a wavelength of 475 nm.

The binding of exopolysaccharides to calcofluor-white colorant (CWC) was performed as described previously (Spiers et al., 2003). Briefly, 3-day-grown colonies of Azospirillum strains were inoculated into LB^∗ medium and incubated at 30°C with agitation at 150 rpm until the cultures reached an OD600 of 1.1–1.4. Cells were harvested by centrifugation (5,000 rpm), washed with phosphate buffer (66 mM) pH 7.0 and resuspended to a 2 OD600 biomass. Cells were diluted 1:100 in NFB^∗ or NFB media with an initial OD600 of 0.1 (∼ 2 × 10⁷ CFU/mL), and 2 mL of this suspension were transferred to a flat bottom polystyrene well (Corning Incorporated) and incubated under static conditions at 30°C for 5 days. The supernatant of each well was removed and 1 mL NFB^∗ or NFB supplemented with CWC at a concentration of 50 μg/mL was added. Cells were incubated at 30°C for 2 h under static conditions. Afterward, the media with CWC was replaced by 1 mL of 96% ethanol. Next, the content of each well was recovered and harvested by centrifugation at 13,000 rpm for 5 min. Then, μg/mL of CWC binding colorant to cells was determined in a standard curve and subsequently normalized with OD600 biomass (Spiers et al., 2003; O’Toole, 2010).

Motility assays were performed as previously described (Alexandre et al., 2000; Cruz-Pérez et al., 2021) with some modifications. Briefly, strains were grown in LB^∗) and at 30°C and 150 rpm for 18 h. Next, a 5 μL drop containing approximately 5 × 10⁶ CFU/mL of each evaluated strain was spotted onto soft agar plates with K-minimal medium, agar 0.25% (w/v), and succinate, malate, pyruvate, proline, and fructose as the carbon source to be tested as an attractant to a final concentration of 10 mM. Plates were incubated under static conditions at 30°C for 24 and 48 h. Swimming rings were measured and statistically analyzed.

The analysis of c-di-GMP levels was done in E. coli S17.1 strains carrying the pGEX-cdgC/pDZ-119, pGEX-4T-1/pDZ-119, or pDZ-119 vectors, that were grown in LB medium supplemented with ampicillin at 100 μg/mL and/or chloramphenicol at 10 μg/mL. CdgA, a functional and characterized DGC from A. brasilense Sp7 (Ramírez-Mata et al., 2016), expressed from the pLIL-2 vector was used as a positive control to validate the assay. c-di-GMP accumulation was evaluated as reported by Zhou et al. (2016). Briefly, E. coli cultures were grown to an OD600 of 0.6, afterward IPTG was added to a final concentration of 0.01 or 0.1 mM to induce the expression of the DGCs. Induction was carried out at 30°C for 24 h with shaking at 150 rpm. Cultures were concentrated 10-fold and resuspended in water. c-di-GMP production was analyzed macroscopically relating color intensity with the production of the second messenger. Microscopic assessment of c-di-GMP was performed using a Nikon Eclipse TE2000U fluorescence microscope. A drop of the induced culture was deposited on a coverslip and covered with a 1% agarose plug. The excitation and emission of the calibrator AmCyan fluorophore were recorded at 457 and 520 nm, respectively. The reporter TurboRFP fluorophore was excited at 553 nm, while its emission was measured at 574 nm. Images obtained were edited with the Nikon NIS Elements software.

The competition assay and root colonization experiments were performed as described previously (Ramirez-Mata et al., 2018). Briefly, wheat seeds (Triticum aestivum) of the “Nana” variety were surface sterilized with 70% ethanol for 1 min, 1% sodium hypochlorite for 30 min and a mix of cycloheximide, 150 μg/mL; streptomycin, 250 μg/mL; tetracycline, 20 μg/mL, and fluconazole, 150 μg/mL. Subsequently, the seeds were washed 4 times with sterile deionized water for 1 min, placed on Petri plates with agar 0.6% (w/v), and germinated for 2 days at 30°C. Seedlings were transferred into sterile glass tubes containing 15 mL of Hoagland hydroponic solution supplemented with 25 mM KNO₃, maintained for 7 days, and placed into an environmental chamber at 80% relative humidity with a photoperiod of 14 h light at 25°C and 10 h dark at 16°C. Six plants for each strain were inoculated by dip 10⁷ CFU/mL A. baldaniorum 2449 or A. baldaniorum 2450; for the competition assay Azospirillum cells were inoculated to a 1:1 (5 × 10⁶ CFU/mL of each strain) (Ramírez-Mata et al., 2018). All inoculated plantlets were maintained under the above conditions for 7 days. Seven days postinoculation (dpi), the plant roots were macerated with PBS buffer (pH 6.8), and bacterial colonization was determined by colony counting using CR medium supplemented with Gm for 2449 strain or Km for 2450 strain. For competition assay bacteria were plated in RC medium without antibiotic, and were further growing in CR media supplemented with Km or Gm. Additionally, seedling inoculated root sections were examined using an inverted CLSM Nikon Eclipse Ti-E C2+ (Nikon Instruments, Tokyo, Japan) to evaluate the colonization patterns of these strains. Images obtained were edited employing NIS Elements software.

To quantify bacterial internalization to wheat roots, an external disinfection protocol was implemented, as described above. Seedlings were transferred into sterile glass tubes containing 15 mL of Hoagland hydroponic solution supplemented with 25 mM KNO₃ and without antibiotics. Seven-day-old plants were inoculated by dip with 10⁷ CFU/mL of A. baldaniorum-GFP, A. baldaniorum 59C-GFP, or A. baldaniorum 2A65 (Ramírez-Mata et al., 2018) and maintained as above for 7 days. Subsequently the post-inoculated roots were immersed in sodium hypochlorite (0.6%) for 15 s with subsequent washes using phosphate buffer pH 6.8. Roots were weighed out and macerated. Endophytic colonization of A. baldaniorum-GFP, A. baldaniorum 59C-GFP were quantified by CFU plating in CR plates supplemented with Gm and for A. baldaniorum 2A65 by CFU plating in CR plates supplemented with both Gm and Km.

Data obtained were analyzed by Student’s t-test to determine if a statistically significant difference existed between the examined strains. Differences were cataloged as statistically significant when p-values were < 0.05.

A. baldaniorum contains 20 genes that encode putative proteins with homology to DGC proteins (Ramirez-Mata et al., 2018). In this work, we focus on a DGC encoded by the AZOBR_100206 gene, located on the bacterial chromosome (Wisniewski-Dyé et al., 2011). The AZOBR_100205 gene encoding a putative Nudix hydrolase, is located immediately downstream of AZOBR_100206 (Figures 1A,B). The analysis of the domain architecture of this protein of 294 amino acids in length revealed the presence of two domains: a REC or receptor domain located at the N-terminus of the protein and a GGDEF domain at the C-terminus (Figures 1C,D and Supplementary Figure 1A). We named this DGC CdgC (WP_014240083.1). Our analysis revealed that CdgC has a domain architecture that resembles that of WspR, a key regulator of exopolysaccharide production in P. aeruginosa (Güvener and Harwood, 2007; Ganesh and Rai, 2018). We speculated that CdgC could play similar roles in processes associated with surface attachment and biofilm formation in A. baldaniorum. We further analyzed the conservation pattern of amino acid motifs important for domain function, using WspR and PleD as reference sequences (Güvener and Harwood, 2007; Malone et al., 2007). The REC domain of CdgC spans amino acids from position 1 to 114, with the putative phosphorylation site being an aspartate residue at position 51 (Supplementary Figures 1A,B). This aspartate residue aligns with aspartate residues 70 and 52 from WspR and PleD, respectively, which have been proposed as the phosphorylation targets in these response regulators (Güvener and Harwood, 2007; Wassmann et al., 2007; Figure 1D). The GGDEF domain spans amino acid residues from position 120 to 293. The GGEEF amino acid residues involved in catalysis, were located in positions 210–214, while and inhibitory site (I-site) that matches the consensus RXXD was found in positions 201–204 (Chan et al., 2004; Vorobiev et al., 2012; Figures 1E,F). The catalytic site of CdgC (Gly²¹⁰, Gly²¹¹, Glu²¹², Glu²¹³, Phe²¹⁴), is located between the β₂ and β₃ strands. The I-site (Arg²⁰¹, Pro²⁰², Gly²⁰³, Asp²⁰⁴) is located five amino acids upstream of the GGEEF motif (Supplementary Figure 1B). It is noteworthy that both WspR and PleD have a GGEEF motif and an I-site. The GTP-interacting and additional catalytic residues observed in previous DGC domain structures (Malone et al., 2007; Römling et al., 2013, 2017) are conserved in CdgC, strongly supporting that this DGC might be active.

CdgC showed 33.33, 25, and 25.56% identity with the first REC and second REC domains of PleD and the REC domain of WspR, respectively (Supplementary Figure 1A). On the other hand, the DGC domain of CdgC showed 41 and 39.66% identity with the DGC domains of the PleD and WspR proteins, respectively (Supplementary Figure 1B). The secondary structure of the C-terminal domain is made from five α-helices and seven short β-strands with a sequence α1–α2–β1–α3–β2–β3–α4–β4–β5–β6–α5–β7 with the GGDEF motif located within a β-hairpin between β2 and β3, these features are conserved in the response regulators WspR and PleD (Figure 1E and Supplementary Figure 1B).

Three-dimensional structural analysis of the CdgC protein was performed by comparison with the crystal structure of WspR (PDB number access 3BRE). The superposition of the REC (C-score 1.38) domain of CdgC with the REC domain of WspR (Figure 1D) revealed a similar spatial distribution. Whilst the modeling of the DGC domain of CdgC (C-score 1.71) proved to be constituted by five α helices and seven β-strands (Figure 1E), it is worth mentioning that a C-score close to 2 gives the model greater reliability. In addition, the active site showed the presence of glycines, which occur in the corresponding position for the substrate GTP in the predicted structure of the GGEEF domain protein that is conserved and provides space for ribosyl sugars and phosphates, thus explaining the conservation of these residues, as shown in the Web-logo model (Figure 1F). Comparison with previously solved DGC structures shows a similarity of the active and inhibitory sites; the latter is characterized by an RXXD sequence, and both are conserved, similar to WspR and PleD, the most well-investigated diguanylate cyclases (Chan et al., 2004; Römling et al., 2017).

To evaluate if CdgC, like WspR and PleD, participates in physiological processes involved in surface attachment and biofilm formation we first generated a variant strain of A. baldaniorum that is isogenic except for the deletion of cdgC (A. baldaniorum 59C). Additionally, two complemented variants were generated, one where cdgC was inserted into the chromosome of A. baldaniorum 59C (A. baldaniorum C21) and another one where cdgC was expressed exogenously from a plasmid (A. baldaniorum 3A13).

We first analyzed the growth rate in liquid media of the 59C mutant strain and the complemented variants (Supplementary Figure 2). Our results showed that the strains exhibited similar growth characteristics. Next, to evaluate biofilm architecture, we mobilized into the parental strain, the mutant strain A. baldaniorum 59C, and the complemented strain A. baldaniorum 3A13, a plasmid expressing the fluorescent protein eGFP (A. baldaniorum-GFP, A. baldaniorum 59C-GFP, and A. baldaniorum 2A65 strains, respectively).

Biofilms were grown under denitrification or nitrogen fixation (NFB) conditions for 5 days, stained with CWC, and visualized employing an inverted CLSM. The CWC fluorophore binds to compounds with β-1,3 and β-1,4 bonds and it has been used to detect exopolysaccharides (Del Gallo et al., 1989). Data obtained from confocal microscopy showed no difference between wild-type, mutant, and complemented strains when strains were grown under denitrification conditions (minimal medium plus KNO₃). The three-dimensional architecture of the biofilms (Z-projection of x-y stacks with 20.1 μm, optical sections) showed similar thicknesses (Supplementary Figures 3A–C). In contrast, in cells growing under nitrogen-fixing conditions (NFB) a considerable alteration in exopolysaccharides stained with CWC was observed in the mutant strain, showing a decrease in the production of a calcofluor-binding polysaccharide compared to the wild-type strain, and this deficiency was complemented by exogenously expressing cdgC (Figure 2). Furthermore, we observed severe defects in biofilm structure in the ΔcdgC mutant strain compared to the wild-type strain (Figure 2). These observations suggest that CdgC is important for exopolysaccharides production in biofilms and that biofilms grown under nitrogen-fixing conditions are more sensitive to the absence of CdgC.

To quantify biofilm formation, we used the crystal violet staining method. Our results revealed that biofilm formation is more proficient under denitrification growth conditions compared to nitrogen-fixing conditions (Figure 3A). Similar to our observations regarding biofilm structure, we did not observe differences in biofilm formation under denitrification conditions in any of the analyzed strains. In contrast, we did observe a reduction in biofilm formation in the mutant strain A. baldaniorum 59C and A. baldaniorum C87 (control mutant strain), which was complemented by a single copy of cdgC inserted into the chromosome (A. baldaniorum C21) (Figure 3A).

Together, these results strongly suggest that CdgC plays an important role in biofilm formation especially under conditions where the nitrogen source is obtained through nitrogen fixation. The defects on biofilm formation observed in the mutant strain A. baldaniorum 59C, are likely linked to an impairment in exopolysaccharides production.

Motility and biofilm formation are typically inversely regulated by the second messenger c-di-GMP (Ryjenkov et al., 2006; Ramírez-Mata et al., 2016; Little et al., 2019). Hence, we analyzed if the absence of cdgC affects the motility of A. baldaniorum in soft agar plates. Our results did not reveal differences in motility for any of the strains analyzed. We speculate that this could be due to compensatory effects of several other genes encoding DGCs found in the genome of A. baldaniorum (Supplementary Figure 4).

As mentioned above, CdgC has all the structural and sequence elements of active diguanylate cyclases (Figure 1). To demonstrate that CdgC can promote c-di-GMP accumulation we coexpressed it with a c-di-GMP biosensor in the heterologous host E. coli S17.1. The usefulness of this c-di-GMP biosensor to measure relative abundance of c-di-GMP has been previously reported (Zamorano-Sánchez et al., 2019; Wu et al., 2020; Martínez-Méndez et al., 2021). This biosensor produces two fluorescent proteins, AmCyan and TurboRFP, both proteins are under the control of a constitutive promoter. However, the production of TurboRFP is regulated by two c-di-GMP riboswitches located in tandem upstream of the turborfp gene. These riboswitches prevent production of TurboRFP when the levels of c-di-GMP are low (Zamorano-Sánchez et al., 2019; Wu et al., 2020; Martínez-Méndez et al., 2021). Under the conditions tested, most of the E. coli S17.1 pDZ-119 cells expressing the c-di-GMP biosensor showed green fluorescence, indicating that mostly the AmCyan fluorophore is being produced and that c-di-GMP levels are very low in this strain (Figure 4A). The additional presence of the empty expression plasmid pGEX did not affect c-di-GMP levels (Figure 4A). In contrast, expression of either cdgC or cdgA from an IPTG inducible promoter in pGEX-CdgC and pQE-CdgA, respectively, resulted in an increase in red fluorescence likely due to elevated production of TurboRFP and c-di-GMP compared to the control strains (Figure 4B). The CdgA protein was used as a positive control for the production of c-di-GMP since its function as a DGC was previously characterized (Ramírez-Mata et al., 2016). These data strongly suggest that CdgC is a functional DGC capable of producing c-di-GMP.

Since we showed that CdgC is required for exopolysaccharides production we next asked if its absence affects bacterial colonization of wheat roots. We speculated that alterations in colonization patterns might be observed in single-strain infections or competition assays. To test these possibilities, we tagged the wild-type strain with the mCherry protein (A. baldaniorum 2449) and the ΔcdgC mutant strain with the eGFP protein (A. baldaniorum 2450). The number of bacteria colonized (CFU/mL) to wheat roots was determined 7 days post-inoculation for the individual strains and the competition assay. The number of cells colonizing the roots was not statistically different between strains 2449 and 2450, however, we observed reduced root-colonization of cells in the competition assay (Figure 5). The data indicated that the WT-2449 and (59C-2450)-tagged strains were able to grow as individual infections at the same level (1.6 × 10¹⁰ /g wet root and 8.4× 10⁹ /g root, respectively) colonizing the wheat plant roots well. However, when a competition assay was performed inoculating both strains with the same proportion of CFU/mL (1:1), a slight decrease in mutant 59C-2450 and WT-2449 bacteria colonizing the wheat roots was observed, with a statistically significant difference when two strains occur in a mixed infection (5.8 × 10⁹/g wet root). These data would reflect that the mutant strain is able to grow as efficiently as the wild type (Figure 5) and that the presence of the mutant strain upsets the growth of the wild-type strain, which might suggest an effect of the mutant (59C-2450) on the growth of WT-2449 into wheat roots.

To further evaluate the process of infection of the wild-type strain and the ΔcdgC mutant strain we analyzed the bacterial-plant association using confocal microscopy (Figure 6). The wild-type strain A. baldaniorum 2449 and the mutant strain A. baldaniorum 2450 showed distinct colonization patterns (Figure 6). A. baldaniorum 2449 showed the typical endophyte colonization pattern showing bacteria (in red fluorescence) internalized in plant (Figures 6A,B) cells. In contrast, the A. baldaniorum 2450 strain remained on the root surface, forming dispersed bacterial aggregates (green fluorescence) (Figures 6C,D). These results suggest that the absence of CdgC renders the cells unable to internalize efficiently into wheat roots.

Finally, to corroborate the data obtained by confocal microscopy, we quantified the endophytic bacteria of colonized wheat roots. As expected, the mutant strain showed an alteration in its capability to internalize to wheat roots compared to the wild-type and complemented strains (Figure 7).