Rhizobium leguminosarum strains used in this work were Rhizobium leguminosarum bv. viciae (Rlv) 3841 (Johnston and Beringer, 1975) and derivative mutants prsD::Tn5 (Krehenbrink and Downie, 2008), pssA::Tn5 (Williams et al., 2008) and ΔrapA2 (Vozza et al., 2016). Strains were grown in TY (Beringer, 1974) or Y-minimal medium (Sherwood, 1970) containing mannitol (0.2 % wt/vol) as the carbon source at 28°C with the appropriate antibiotics. Bacterial growth was monitored at 600 nm using a visible light spectrophotometer. Plasmids were mobilized into Rhizobium spp. by triparental mating using Escherichia coli RK600 as a helper strain (Finan et al., 1986) or by biparental mating using the E. coli S17 strain (Quandt and Hynes, 1993) (for plasmid pK18mobsacB). Antibiotics in the culture media were used in the following concentrations: streptomycin 400 μg/ml, kanamycin 50 μg/ml, gentamycin 20 μg/ml, and chloramphenicol 50 μg/ml.

Biofilm-assocciated phenotypes were analyzed in Rlv 3841 derivated strains as previously described for polystyrene adhesion (Russo et al., 2006), macrocolony morphology (Vozza et al., 2016) and swimming and swarming motilities (Sorroche et al., 2018).

A rapD deletion mutant was generated in R leguminosarum bv. viciae 3841 strain by homologous recombination of the ΔrapD carried in the suicide plasmid pK18mobsacB/ΔrapD. Briefly, 2 sets of primers (ΔrapD-A-fwdBamHI: GAGTAGGGATCCT TCTTCTTCTCATAGGCGT, ΔrapD-A-revlinker: TCGATACT CTCCTTCTTTCAGTGGCTTTGCATTTGTCATTTG, ΔrapD-B-fwdlinker: CTGAAAGAAGGAGAGTATCGATTTTCGCTGT TTGCCTTGCTA and ΔrapD-B-revBamHI: ATTGGATCCC AATCCGACCCAGTGCGAG) were used to amplify proximal and distal ends of rapD gene (Uniprot: Q1MFT2, locus tag: RL2702) with a random linker sequence. The 1244 bp fragment obtained by recombinant PCR amplification was BamHI-digested and cloned into pK18mobsacB in the same restriction site (Schafer et al., 1994). The pK18mobsacB/ΔrapD plasmid was transferred from E. coli S17 to Rlv 3841 strain by biparental mating. Single cross-over events were selected on a Y-minimal medium (to avoid E. coli growth) supplemented with streptomycin and kanamycin. Then, single colonies were picked and grown overnight in liquid culture without antibiotics. Double recombinants were selected from TY agar plates containing 10% sucrose and streptomycin and kanamycin sensitivity was confirmed. The deletion of rapD was verified by PCR using the oligonucleotides ΔrapD-A-fwdBamHI and rapDfull-revXbaI: GCTCCCTCTAGACTATAGCAAGGCAAACA.

The double mutant ΔrapA2ΔrapD was also generated by homologous recombination but in a ΔrapA2 mutant strain background.

To overexpress RapD, rapD coding sequence (locus tag: RL2702, 834 nt) was amplified from 3841 strain using the oligonucleotides rapDfull-fwdKpnI: GAAGGTACCGACA AATGCAAAGCCAACT and rapDfull-revXbaI, which generates a product of 853 pb that was double digested with KpnI and XbaI and cloned in pBBR1-MCS2 (Kovach et al., 1995) under the lacZ promoter. Plasmid pBBR1rapD was then transferred to Rlv 3841 derivative strains via triparental mating using the helper strain E. coli RK600. The selection of rhizobial transconjugants was done on TY rich media agar plates using streptomycin and kanamycin (for single mutant or wild type strains) and gentamycin (for double mutant ΔrapA2ΔrapD). The presence of plasmid pBBR1rapD was checked by PCR using T7 and T3 universal primers.

To generate specific antibodies against RapD, a truncated version of RapD was constructed using the coding sequence of RapD without the Ra/CHDL domain (aa 122 – 277) (Uniprot: Q1MFT2, locus tag: RL2702). This region was amplified from Rlv 3841 using the primers rapD-NdeI-fwd: GGAGGCCATATGTACCATGAAAACCTGGAT and rapD-NotI-rev: CTCGCGGCCGCATATAGCAAGGCAAA. The NdeI-NotI double digested product was cloned into the same restriction sites of plasmid pET-22b to generate the vector pET-22brapDtrunc. After induction of the E. coli BL21 (DE3) cultures carrying pET-22brapDtrunc with 0.5 mM isopropylthiogalactoside overnight at 18°C (OD of 0.6–0.8), cells were resuspended in a buffer containing 20 mM Tris-HCl, 0.5 M NaCl, pH 8 with the addition of 1 mM phenylmethanesulfonyl fluoride and 10 mM imidazole. Cells were disrupted in a French pressure cell at 18,000 p.s.i. Then, the cell extract was centrifuged at 100,000 × g for 1 h and the supernatant was applied to a Ni-NTA HiTrap Chelating column (GE Healthcare, Amersham, United Kingdom) (#RPN2232) and eluted with a buffer containing 500 mM imidazole during a lineal gradient. Fractions containing truncated RapD were pooled and concentrated by ultrafiltration. Mouse polyclonal antisera against the truncated form of RapD was generated by injecting the SDS-PAGE 15% gel-excised band of 18 kDa containing the fragmented RapD into BALB/c mice. The initial dose started with 100 μg of protein/mouse using Freund’s complete adjuvant. Each mouse received the same concentration of booster every 2 weeks with Freund’s incomplete adjuvant up to a total of four immunizations.

The polyclonal antisera against RapD were performed using the animal procedures and management protocols approved by the local Institutional Committee (“Comité de Bioética Fundación Instituto Leloir”) protocol number 2009 01 BFIL.

Complete RapD coding sequence (aa 1 – 277) (Uniprot: Q1MFT2, locus tag: RL2702) was amplified using primers rapDfull-NdeI-fwd: GCCCTCCATATGACAAATGCAAAGCCA ACT and rapD-NotI-rev: CTCGCGGCCGCATATAGCAAGGCA AA and the NdeI-NotI-digested product of 854 pb was cloned in pET-22b digested with the same restriction endonucleases to obtain pET22brapD. E. coli BL21 (DE3) carrying pET22brapD (OD of 0.6–0.8) was incubated with 0.1 mM isopropylthiogalactoside overnight at 18°C and cells were resuspended in a buffer containing 20 mM Tris-HCl, 0.5 M NaCl, pH 8 with the addition of 10 mM imidazole and 1 mM CaCl₂. Then, cells were disrupted in a French pressure cell at 18,000 p.s.i. The cell extract was centrifuged at 100,000 × g for 1 h, the supernatant was applied to a Ni-NTA HiTrap Chelating column (GE Healthcare, Amersham, United Kingdom) (#RPN2232) and eluted with a buffer containing 500 mM imidazole during a lineal gradient. Fractions containing full-length RapD were dialyzed against 20 mM Tris-HCl, 0.15 M NaCl, 10 mM EGTA, pH 8 overnight and applied to Superdex 200 column (Pharmacia Corp.) pre-equilibrated in 20 mM Tris-HCl and 150 mM NaCl, pH 8. Purified full-length RapD was pooled, concentrated by ultrafiltration, and stored at –80°C.

To analyze secreted proteins, rhizobia were grown at 28°C and monitored at 600 nm with a visible light spectrophotometer. Culture supernatant proteins were concentrated by precipitation with 10% trichloroacetic acid (TCA) as described previously (Economou et al., 1990). After precipitation, TCA was removed by washing the precipitate with cold acetone twice. For SDS-PAGE analysis and Western blot, TCA precipitates were resuspended with a loading buffer containing urea and β-mercaptoethanol.

For LC-MS/MS analysis, starter cultures were diluted to an OD of 0.01 and grown for approximately 24 h in TY rich media to an OD of 0.7–0.8 (exponential growth phase). TCA precipitates were resuspended with 50 mM ammonium bicarbonate, quantified using Bio-Rad Bradford reagent according to manufacturer instructions, and analyzed by the CEQUIBIEM service (Departamento de Química Biológica, Facultad de Ciencias Exactas y Naturales, UBA). Cytoplasmic protein RhiA (Krehenbrink and Downie, 2008) was used as a lysis marker for the extracellular samples.

Analyses of surface proteins were performed with equivalent amounts of cells as follows. Starter cultures were diluted to an OD of 0.01 and grown for 24 h in rich (TY) or minimal (Y) medium to an OD of 0.6–0.8 (exponential growth phase). Standardized bacterial pellets were washed with Tris 30 mM pH 8.4 and resuspended in a high ionic strength buffer containing 30 mM Tris pH 8.4 and 1.5 M NaCl. Cells were submitted to mechanical pressure using a 27-gauge needle 20 times. Bacterial debris was removed, the supernatant was centrifuged for 1 h at 21,000 × g and the proteins were precipitated with TCA 10% as described above.

After SDS-PAGE, proteins were transferred to a PVDF membrane (Millipore, Billerica MA, United States), blocked with 5% skimmed milk in TBS-0.05% Tween 20 and incubated with the proper antibodies: Primary antibodies against RapA2, diluted 1:5000 (Abdian et al., 2013) or RapD, diluted 1:8000 were prepared in TBS-0.05% Tween with 1% BSA. RapA2 was revealed by an anti-rabbit secondary antibody conjugated to HRP and RapD antiserum was revealed by an anti-mouse secondary antibody conjugated with HRP (Jackson Immunoresearch Laboratories Inc., United States) (#115-036-062). Secondary antibodies conjugated with HRP were revealed with ECL (GE Healthcare, Amersham, United Kingdom) (#RPN2232) according to manufacturer instructions.

To compare RapD secretion under different culture conditions, the coefficient between the specific RapD signal and total proteins signal was estimated. Coomassie blue staining of total proteins was analyzed by densitometry with the ImageJ software (NIH, United States) (https://imagej.nih.gov/ij/download.html). For RapD-specific signals, Western blots were also quantified by densitometry using the same software.

For colony blots, 10 μl of standardized bacterial suspensions from 48 h starter cultures (OD 0.05) were grown at 28°C on a sterilized nitrocellulose filter placed on TY or Y agar plates. After 72 h, the macrocolonies were washed away from the nitrocellulose filter with MilliQ, and immunodetection was performed as described above for the Western blot analysis.

To analyze RapD localization, immunofluorescence analyses of whole bacterial cells were performed. Starter cultures were diluted to an OD of 0.025 and grown for 24 h in rich TY medium with proper antibiotics. Cells were harvested during exponential growth (OD 0.8) and fixed with paraformaldehyde 4% in PBS at 28°C for 20 min. After blocking with gelatin 0.2% for 1 h at room temperature, incubation with an anti-RapD antiserum (1:200) in PBS-BSA 1% for 1.5 h at room temperature was performed. Then, anti-mouse secondary antibodies conjugated with Cy3 (1:200) in PBS-BSA 1% were incubated for 1 h at room temperature. Cells were spotted onto glass slides with agarose 1% pads and visualized in a Zeiss Axio Observer 3 inverted fluorescence microscope (Carl Zeiss Microscopy, LLC, United States). For images, the fluorophore signal was pseudo-colored cyan.

The extracellular medium was obtained from culture supernatants as follows. Cells were grown in Y- mannitol medium for 5 days at 28°C. NaCl was added to a final concentration of 0.25 mM in cell-free supernatants and EPS was precipitated with 2½ volumes of 100% ethanol. EPS precipitate was washed with increasing ethanol concentrations from 70 to 90% and let to air dry before being resuspended in MilliQ water. Polysaccharides were quantified by the determination of hexuronic acids by the meta-hydroxidiphenyl method (Filisetti-Cozzi and Carpita, 1991). Proteinase K treated polysaccharide was obtained by incubating 1 mg of uronic acids equivalents with 2 mg/ml proteinase K at 60°C for 1 h. Proteinase K was heat-inactivated at 28°C for 40 min and centrifuged at 16,000 × g for 30 min. Soluble EPS was then precipitated as described above.

Extracellular medium preparations (1 mg/ml uronic equivalents) were heated at 100°C for 10 min and centrifuged at 10,000 × g to remove denatured proteins as described previously (Vozza et al., 2016). Treated EPS was fractioned by gel filtration chromatography in a Superose 6 HR 10/30 column (Amersham Biosciences) previously equilibrated in 0.1 M NaCl and 0.1 M sodium phosphate buffer pH 7. Hexoses in every 1 ml fraction were quantified according to Loewus (1952).

The interaction between RapD and the EPS from the Rhizobium leguminosarum 3841 strain was analyzed as described previously (Abdian et al., 2013) with some modifications. Briefly, the microplates were coated with 100 μl of a 0.1 mg/ml EPS/Xanthan solution in binding buffer (50 mM sodium carbonate pH 9.7) overnight at 4°C. For conventional Enzyme-Linked ImmunoSorbent Assay (ELISA), increasing concentrations of RapD ranging from 0.05 μg/ml to 10 μg/ml were incubated in TBS, 1 mM CaCl₂ for 2 h at room temperature. For binding inhibition assays (BIA), RapD (1 μg/ml) was preincubated before addition to the wells with serial 10-fold dilutions of the putative inhibitory compound for 30 min at room temperature. The initial concentration of the inhibiting polysaccharides was 0.1 mg/ml and 5 mM for each monosaccharide. Bound RapD was detected with a polyclonal antiserum against the C-terminal domain of RapD (truncated RapD) and revealed with an HRP conjugated secondary antibody. OPD was used as a substrate in the peroxidase reaction which was stopped after 4 min and absorbance at 492 nm was measured with a multiwell scanner. The data represent the mean values of a representative experiment of three independent assays done in triplicate.

The effect of RapA2 on RapD-EPS interaction was carried out using a Binding-Inhibition Assay (BIA). Fixed EPS was either preincubated with increasing RapA2 concentrations (0.05 μg/ml to 10 μg/ml) for 2 h at room temperature or incubated simultaneously with RapA2 (0.05 μg/ml to 10 μg/ml) and RapD (5 μg/ml) for 2 h at room temperature. Then, proteins were washed and the remaining bound RapD was quantified following the procedure mentioned above.

Analyses were done in 15 mM Tris pH 8, 15 mM NaCl with the addition of the corresponding CaCl₂ (0–1.25 mM), NaCl (0–500 mM) or EPS concentration (0–12 μg/ml uronic equivalents). Far-Ultraviolet Circular Dichroism (Far-UV-CD) measurements were carried out at 25°C on a Jasco J-815 spectropolarimeter (JASCO Corp, Japan). The recombinant RapD protein was diluted to 3.75 μM (0.12 mg/ml) and placed in a cell with a 1-mm path length. Spectra were acquired over the wavelength range of 195–260 nm. Measurements were performed in a Peltier-thermostatted cell holder using a 1-mm path length cell. The spectra acquired were the average of six scans to reduce background noise.

Fluorescence emission spectra of recombinant RapD were recorded at 25°C between 305 and 400 nm with an excitation wavelength of 295 nm using excitation and emission bandwidths of 5 nm in a Jasco FP-6500 spectrofluorometer (JASCO Corp., Japan). Fluorescence emission data were analyzed by first subtracting the buffer background. The RapD protein concentration was 3.75 μM.

A total of 100 μl of RapD (1 mg/ml) were loaded on a Superdex 200 column (Pharmacia Biotech, Cytiva, United States) (#17-5175-01) and size exclusion chromatography was carried out at 1 ml/min in Tris 20 mM pH 8, 150 mM NaCl and CaCl₂ (ranging from 0.1 to 1.25 mM). Fractions (1 ml) were concentrated by precipitation with 10% TCA, resuspended in loading buffer, and analyzed by SDS-PAGE. Molecular weight markers used were Ferritin (440 kDa), BSA (67 kDa), OVA (45 kDa), and Lysozyme (15 kDa). Blue dextran and acetone were used for V0 and V0+Vi calibration.

The average MW of RapD in solution was determined on a Precision Detectors PD2010 90° light scattering instrument (Bellingham, MA, United States) tandemly connected to high-performance liquid chromatography and an LKB 2142 differential refractometer. A Superdex 200 column was used. Hundred μl of the purified recombinant RapD at 1 mg/ml pre-incubated with 1.25 mM CaCl₂ was injected into the column, and the chromatographic runs were performed with a buffer containing 20 mM Tris-HCl, 0.15 M sodium chloride, pH 8, under isocratic conditions at a flow rate of 0.4 ml min^–1 at 20°C. The molecular weight of each sample was calculated relating its 90° and RI signals and comparison of this value with the one obtained for BSA (molecular mass: 66.5 kDa) as a standard using the software Discovery32. The average and standard deviation values correspond to the peak center ± 0.08 ml.

The coding sequence for RapD (RL2702) (Supplementary Table 1) was used as a query for a protein blast analysis excluding R. leguminosarum species and limited to Rhizobium genus. The orthologous proteins found were used in local pairwise alignments to detect similarities as well as for global alignment analyses with Clustal Omega (EMBL-EBI, United Kingdom).

In addition to the Ply glycanases, the RapA lectins and the RapB and RapC proteins, another member of the Rap family, named RapD (RL2702), was predicted by an in silico search of Ra/CHDL domain-harboring proteins in the Rlv 3841 genome (Abdian et al., 2013). The Ra/CHDL domains consist of about 110 residues found in several bacterial proteins with a high degree of structural similarity with the cadherin-family of eukaryotic calcium-binding proteins involved in cell-to-cell interactions. CHDL domains are present in numerous bacterial and archaeal proteins and were previously proposed to confer carbohydrate-binding ability (Cao et al., 2005). RapA1 and RapA2 lectins are the smallest members of the Rap family containing only two Ra domains (Ra1 and Ra2) (Ausmees et al., 2001; Abdian et al., 2013). Both the size and the domain structure of RapD are similar to those of RapB since it contains an N-terminal Ra2 domain (aa. 1–122) that includes a VCBS repeat (aa. 47–124) and a C-terminal region (aa. 123–277) of unknown function, sharing an overall 49% similarity with RapB (aa. 1–248) (Figure 1; Supplementary Table 1). The VCBS repeats comprise about 100 residues and are domains found in multiple copies in large proteins of Vibrio, Colwellia, Bradyrhizobium, and Shewanella whose functions are related to cell-to-cell interactions. Local (PBlast) and global (Clustal) alignments showed a degree of similarity greater than 85% between the RapD orthologues of species belonging to the Rhizobium genus such as R. indigoferae, R. laguerrae, R. ecuadorense, R. valis, R. phaseoli, R. etli, and R. esperanzae, among others (Supplementary Figure 1), suggesting that RapD is highly conserved in this genus. Iterative BLAST using as query sequences either the complete or the C-terminal region resulted in no orthologous outside the Rhizobium genus.

In order to determine if RapD is a substrate of the PrsDE system and is co-secreted with the other PrsDE-dependent proteins, the secretome of the Rlv 3841 wild type strain and the isogenic prsD::Tn5 secretion mutant was analyzed by LC/MS-MS as described in Methods. Several proteins were absent in the extracellular medium of the prsD mutant. As was previously shown (Russo et al., 2006; Krehenbrink and Downie, 2008), the secretion of RapA2 (Q1M755), RapC (Q1MER4), PlyA, and PlyB (Q1MD30 and Q1MEW2) was PrsDE-dependent (Table 1). In addition, the RapD protein (RL2702) was found in the secretome of the wild type strain but was absent in the extracellular medium of the prsD secretion mutant. The secretion of other PrsDE-dependent proteins such as three calcium-binding RTX proteins (Q1MF23, Q1M7J7, and Q1MGL7), a beta-helix domain-containing protein (Q1M8U8), and an NDK-nucleoside diphosphate kinase (Q1MIY5) was corroborated (Krehenbrink and Downie, 2008). Besides, three novel PrsDE dependent extracellular proteins were detected in Rlv 3841: a calcium-binding RTX Zn-metalloprotease (Q1ML69), the RapB protein (Q1MCD0), and the PlyC glycanase (Q1MEW3). These proteins were previously proposed in silico as putative TISS substrates (Krehenbrink and Downie, 2008; Table 1 and Supplementary Table 1).

The presence of RapD in the extracellular medium of Rlv 3841 grown in TY rich medium was confirmed by Western blot using specific polyclonal antibodies generated against the C-terminal region of RapD (Figure 2). We also included in this analysis a mutant that was obtained by deletion of the rapD gene (ΔrapD) and a wild type derivative-strain that overexpresses rapD from a multicopy plasmid (pBBRrapD). In line with the proteomic analysis, RapD was absent in the extracellular medium of the prsD secretion mutant (Figure 2A). High levels of extracellular RapD were observed in the rapD-overexpressing cells from pBBRrapD. In relation to the total extracellular protein amount produced by the wild type strain both in rich (TY) and Y-mannitol minimal media, RapD showed a steady accumulation. However, an estimation by densitometry using ImageJ suggests that the amount of RapD in the extracellular medium at an OD of 0.8 in minimal Y-mannitol medium in relation to the total extracellular proteins was considerably higher (up to fourfold) than that observed in the rich medium (Figure 2B). This observation is interesting since copious amounts of EPS are produced in Y-mannitol and Rlv forms structured biofilms in this medium both in vitro and on pea roots (Russo et al., 2006; Williams et al., 2008; Vozza et al., 2016). Taken together, these results show that RapD is co-secreted with the other Rap(s) in a PrsDE-dependent manner. Noteworthy, RapD was detected in the wt strain either as a faint or a strong band that slightly differs in the molecular weight between TY or Y media (Figure 2B). The reasons for the difference in RapD behavior under different culture conditions are so far unknown.

To determine whether RapD is also secreted in biofilms, macrocolonies that are developed on semisolid media were used as a biofilm model (Branda et al., 2005; Tseng et al., 2018). A moderate signal corresponding to extracellular RapD was detected in the wild type strain and, as expected, absent in the prsD and ΔrapD mutants, both in TY and Y media (Figure 3A). Besides, a very strong signal was observed in the membrane where the macrocolony of the rapD-overexpressing strain (pBBRrapD) was grown. These observations support the idea that RapD is secreted under sessile conditions. In order to analyze RapD secretion in the macrocolony of a strain deficient in EPS production, the pssA mutant (Williams et al., 2008) was also analyzed. Surprisingly, for this strain both in TY and Y media, a strong extracellular RapD signal was observed (Figure 3A). This seems not to be due to an increased expression/secretion of RapD in the EPS-deficient strain since analysis by Western blot of the extracellular proteins showed RapD levels that were even lower than those of the wild type strain (Figure 3B). As expected, colonies of the pssA mutant were visibly dry on both TY medium and Y medium (Figure 3A). Therefore, the strong signal of extracellular RapD may be related somehow to the EPS-deficient phenotype.

RapA lectins are partly retained by the capsular polysaccharide (CPS) on the cell surface and modulate cellular cohesion (Vozza et al., 2016). In order to evaluate the interplay between RapA and RapD, a RapA2-deficient strain was cultured under the same sessile conditions. Similar to that observed for the EPS-defective mutant, a prominent RapD signal was observed in the ΔrapA2 mutant grown in TY medium, in which this strain shows a dry colony phenotype (Vozza et al., 2016; Figure 3A). Again, the levels of extracellular RapD in this strain grown under planktonic conditions in the TY medium were similar to that of the wild type (Figure 3B), suggesting that the high RapD signal is not due to an increased RapD secretion. Since the pssA mutant (on both Y-mannitol and TY media) and the rapA2 mutant (only on TY) develop dry macrocolonies (Figure 3A; Vozza et al., 2016), the simplest explanation for these observations is that extracellular RapD is more efficiently retained by the nitrocellulose membrane in EPS altered/deficient (dry) colonies while in the mucoid ones a considerable proportion of it is washed out together with the bacterial biomass, probably due to a strong interaction of RapD with the EPS or the CPS. However, another non-exclusive possibility is that RapD secretion in these deficient/altered EPS strains is increased when the bacteria are grown in biofilms.

As mentioned, upon secretion, a proportion of the RapA protein is retained by the CPS on the bacterial cell surface (Vozza et al., 2016). To explore RapD destination, a surface-associated protein preparation from the wild type strain was analyzed by Western blot using anti-RapD antibodies. In contrast with RapA2, the RapD protein was not detected in the surface-associated protein fraction neither in the rich medium nor in the Y-minimal medium (Figure 2C). Furthermore, RapD was absent in the cell surface fraction of the rapD-overexpressing bacteria (Figure 2C). Accordingly, immunofluorescence analysis of the Rlv 3841 pBBRrapD cells indicated that RapD is almost not retained on the cell surface (Figure 4). Few bacterial cells showed a polar extracellular signal, probably due to a transient detection of the protein after secretion (Supplementary Figure 2). These observations suggest that similar to RapA, RapD is secreted by the PrsDE system in a polar manner.

In conclusion, it seems that the final fate of RapD is different from that of RapA proteins. While RapA proteins are partly retained on the cell surface interacting with the CPS, RapD is completely released into the extracellular environment, probably interacting with the polysaccharide in its released form (EPS).

RapA2 and the N-terminal region of RapD share domain homology with eukaryotic cadherins (Figure 1; Abdian et al., 2013). As calcium ions affect the conformation and dynamics of both eukaryotic cadherins and RapA2, we sought to determine if calcium ions also affect the conformation of RapD protein using circular dichroism (CD) and fluorescence spectroscopy (Figure 5). The CD spectrum of the Apo-RapD (decalcified) is indicative of a protein with a mixed content of secondary structure elements with a minimum centered at ∼210 nm. The addition of calcium in the millimolar range induces a decrease in the ∼210 negative component and the consolidation of a negative band at ∼217 nm, indicating that β sheets dominate the structure of the Ca⁺²-bound RapD (Figure 5A). The change of the CD signal at ∼210 nm saturates with the addition of 0.5 mM of calcium to 3.75 μM of RapD (Figure 5C). The addition of calcium to Apo-RapA2 (Abdian et al., 2013) and to eukaryotic cadherins (Pokutta et al., 1994) also induces the consolidation of protein conformation enriched in β sheets in agreement with the requirement of calcium for the proper folding of RapD cadherin-like domains. On the other hand, the addition of divalent magnesium at the concentrations used in the calcium experiment (Supplementary Figure 3) did not induce any change in the CD spectrum of the Apo-RapD confirming the exquisite preference for calcium ions.

To further investigate the effect of calcium ions on RapD conformation, the Apo-RapD protein was titrated with an increasing concentration of calcium ions following the fluorescent emission spectrum of the unique tryptophan residue (Figure 5B). As shown in Figure 5B, the addition of calcium led to a small red shift in the fluorescent spectrum suggesting that the unique tryptophan residue of RapD present in the Ra/CHDL domain exposes partially to the solvent. The titration curves of RapD with calcium analyzing the CD signal and fluorescence emission can be superimposed indicating that both signals report the same conformational transitions (Figure 5C). Together, these data suggest that calcium-binding induces the stabilization of the β-sheet prone regions and a local rearrangement of the environment of the tryptophan residue that partially exposes itself to the solvent. The addition of 3 mM EGTA to a calcium saturated Rap sample completely restores the CD and the fluorescence emission spectra to that observed for the Apo form indicating that the calcium-induced conformational transition is fully reversible.

It was proposed that cadherins mediate cell to cell contact through dimerization of their extracellular domains and that this process is promoted and maintained by calcium ions (Nagar et al., 1996; Pertz et al., 1999). To evaluate the propensity of RapD to oligomerize in the presence of increasing calcium concentrations we performed size exclusion chromatography (SEC) experiments (Figure 5D). The experiment was performed by loading 100 μl of a 37.5 μM protein sample in a Superdex 200 column. It must be considered that the elution volume observed in a SEC experiment depends on the hydrodynamic radius of the molecule that is, in turn, affected by both the oligomeric state and the conformational compaction. The RapD-Apo protein elutes as a single broad peak at 14.5 ml corresponding to a 118 kDa globular protein. A protein sample pre-incubated with 0.75 mM calcium elutes as two peaks, at 16.1 ml and 18.3 ml respectively. Both peaks correspond to RapD species with a smaller hydrodynamic radius compared to the Apo form. The peak eluting at 18.3 ml behaves as a globular protein of 34.7 kDa and is compatible with the globular monomer form of the RapD protein (MW of 32 kDa). As both protein and calcium initially present in the sample volume dilute upon injection on the column, we analyzed a sample containing 1.25 mM of calcium ion. In this condition, the Rap protein elutes in two peaks at 13.8 and 16.5 ml. These peaks that correspond to protein species with a hydrodynamic radius higher than that of the RapD globular monomer must correspond to oligomeric forms of the compacted monomer, indicating that calcium induces the oligomerization of RapD. Multimerization was verified by an SLS (static light scattering) assay in the presence of 1.25 mM CaCl₂. Two defined peaks of 90 and 32 kDa (A and C) were identified in the presence of 1.25 mM CaCl₂ (Supplementary Figure 4). Both calculated MWs fall in line with the predicted sizes of the monomer (31.8 kDa) and the trimer (95.4 kDa). The third peak (B) shows a shoulder with an estimated MW of 43 kDa which could be attributed to the equilibrium between the monomeric and dimeric states of 32 and 64 kDa respectively (Supplementary Figure 4). Taken together, SEC and CD titration experiments suggest that calcium induces RapD compaction and a subsequent oligomerization at the expense of the stabilization of the β-sheet secondary structure.

We sought to investigate whether the presence of the EPS exerts an effect on RapD secondary structure. Analysis by CD showed that in the presence of 1.25 mM CaCl₂, increasing concentrations of the EPS induce a slight change in the secondary structure of RapD, mainly an increase in the ellipticity, which might suggest that the EPS favors the formation of RapD aggregates or multimerization (Figure 6A). The same effect was also observed by fluorescence spectroscopy which showed a spectrum red shift upon the addition of EPS in the presence of calcium. To note, these changes were only seen in the presence of calcium as EPS alone was not able to trigger any change in the protein structure whatsoever (Figure 6B). To discern if non-specific carbohydrate-binding was responsible for the changes observed, RapD was incubated with xanthan, a negatively charged exopolysaccharide produced by Xanthomonas campestris. However, xanthan did not trigger the EPS-related conformational changes observed in the previous section (Supplementary Figure 5). The dependence of RapD-EPS interaction to promote these changes was analyzed by CD spectra. RapD was incubated with EPS in the presence or the absence of calcium and increasing concentrations of NaCl (Figure 6C). The results showed that NaCl had no effect on the CD spectra, suggesting that the interaction of RapD-EPS cannot be dissociated by ionic displacement or that once triggered, the conformational changes induced by the EPS cannot be reverted by separating RapD from the EPS.

Previous studies have shown that the Ra/CHDL domains of RapA2 confer the ability to bind the acidic polysaccharide either as CPS or in its released form (Abdian et al., 2013). The N-terminal domain of RapD consists of a Ra/CHDL domain which shares a 41% identity with the C-terminal Ra domain (Ra2) of RapA2 (Figure 1 and Supplementary Figure 6), suggesting that RapD would be able to bind the polysaccharide through this domain. In line with said hypothesis, we showed that protein-free-EPS also exerts a conformational change in RapD probably by means of direct interaction between the EPS and this domain. To assess whether RapD directly interacts with the EPS, we first analyzed by ELISA the presence of endogenous RapD associated with EPS preparations obtained from the extracellular medium of the Rlv wild type strain compared with those of rapD or secretion mutants (Figure 7A). The polysaccharides were immobilized on the wells and endogenous RapD was revealed with anti-RapD antibodies. The signal levels observed in the EPS preparation from the wild-type strain were significantly higher than those of the ΔrapD, ΔrapD ΔrapA2 or prsD mutants, and a proteinase K-treated EPS, indicating that RapD is retained by the EPS. Interestingly, an increment of (30 ± 8) % in the EPS-associated RapD was consistently observed in the rapA2 mutant (Figure 7A), suggesting that there is an interplay between both Rap(s) and their interaction with the EPS.

A direct ELISA assay was also performed using immobilized proteinase K-treated-EPS from Rlv 3841 wild type strain exposed to several concentrations of purified recombinant RapD ranging from 0.05 to 10 μg/ml in the presence of calcium. This analysis revealed a concentration-dependent binding of RapD with the EPS (Figure 7B). Surprisingly, under the assayed conditions, xanthan was also able to interact with recombinant RapD protein although to a lesser extent. This observation may be due to structural similarities between the EPS of Rlv and xanthan since both are anionic and are made up of repeating units of three sugars: two neutral and one anionic; glucose, glucuronic acid and galactose (in the case of EPS from Rhizobium) and glucose, galacturonic acid and mannose (in the case of xanthan) (Casas et al., 2000).

To further study the interaction of RapD with the EPS, a binding inhibition assay (BIA) using a modified ELISA technique was performed. Briefly, before incubation of recombinant RapD with immobilized-proteinase K treated EPS, binding inhibition was evaluated by adding several dilutions of proteinase K treated-EPS, xanthan, or vehicle (buffer), in the presence of calcium. Only the Rlv EPS (and not the xanthan) inhibited the binding of RapD to immobilized EPS in a concentration-dependent manner, suggesting that there is specific recognition of RapD for the Rlv EPS (Figure 7C).

As mentioned, the EPS of Rlv 3841 consists of octasaccharide repeating units containing glucose, glucuronic acid, and galactose (5:2:1). We have previously shown that the binding of RapA2 to the CPS/EPS was inhibited by glucuronic acid in the range of 20–100 mM (Abdian et al., 2013). To test whether RapD recognizes a particular monosaccharide on the EPS molecule, BIA analyses were carried out in the presence of free sugars and calcium. Results indicated that the RapD interaction with the EPS was partially inhibited in the presence of high concentrations of galactose in the mM range but not by glucose or glucuronic acid (Figure 7D), suggesting that the galactose might be implicated in the oligosaccharide structure recognized by RapD on the EPS molecule.

To investigate whether RapA2 can occupy RapD binding sites or recognize similar structures on the polysaccharide, a competition assay was performed preincubating the immobilized EPS with increasing concentrations of RapA2 or incubating simultaneously the immobilized EPS with RapD in the presence of increasing concentrations of RapA2. Under these conditions, no inhibition of the EPS-RapD binding was observed, suggesting that the sites recognized by these Rap(s) could be different (Figure 7E).

To evaluate whether RapD affects properties related to biofilm formation, diverse biofilm-associated phenotypes were analyzed in the rapD mutant and rapD-overexpressing strain. However, no obvious differences in macrocolony morphology, adhesion on polystyrene, swimming and swarming motilities, and EPS production were observed between the wild type strain and the ΔrapD,ΔrapD/pBBRrapD, and wild type/pBBRrapD strains grown in rich or Y-minimal media (Supplementary Tables 2, 3).

As mentioned earlier, we have previously shown that RapA levels influence the EPS size distribution. To evaluate if the absence of RapD or the absence of both RapA and RapD modify the EPS profile, the EPS obtained from the extracellular medium of the wild type and isogenic mutants were analyzed by size exclusion chromatography (SEC), using a Superose 6 HR 10/30 column as described in Methods. This analysis showed an increased proportion of lower molecular weight species in comparison with the wild type strain both in the rapD and rapA2 single mutants, which was even more evident in the rapA2 rapD double mutant (Figure 8A). We also studied the effect on the EPS profile of RapD overproduction in the strain that showed the most marked difference in size species (the rapA2 rapD double mutant). Overproduction of RapD restored the EPS profile to higher MW species (Figure 8A).

The differences in the molecular weights of the EPS species in the rap mutants may be due to a greater activity of EPS-glycanases from the Rap family, such as PlyB. It should be noted that the PlyA and PlyB glycanases are only active on the cell surface (Zorreguieta et al., 2000). Another possibility is that in the absence of RapA/RapD, the polysaccharide chains emerging from the cell surface are more susceptible to glycanase action. We evaluated glycanase activity using carboxymethyl cellulose (CMC) as a substrate model on a plate assay (Finnie et al., 1997). The extracellular β 1,4-glycanase activities were similar in the wild type and rapD mutant, showing only slightly increased activity in rapA2 and rapA2 rapD mutants with no apparent differences between these two strains (Figure 8B). These results suggest that differences observed in the EPS-profiles of the wild type and rapD mutant or the rapA2 and rapA2 rapD mutant cannot be accounted only for an increased glycanase activity in the extracellular medium.

Taken together, these observations suggest that the differences in the EPS profiles of the Rap mutant strains compared with those of the wild type strain may be due to the interaction of RapA/D proteins with the emerging polymer chains, which, in turn, would modulate the susceptibility of these molecules to Ply glycanases.