Introduction

Rhizobia is the common name for a group of soil α- and β-proteobacteria which can establish a nitrogen-fixing symbiosis with legumes (Poole et al. 2018). In this interaction, bacteria and plants establish a molecular dialogue that allows compatible rhizobia to infect legume roots and invade new root plant organs called nodules (Roy et al. 2020; Yang et al. 2022). Inside these organs, rhizobia are hosted by symbiotic plant cells in which they differentiate into bacteroids, the bacterial form able to fix N2 into ammonia. Bacteroids are surrounded by a plant-origin membrane and constitutes a kind of new organelles, the symbiosomes, that are fed by the plant with C and energy sources. Thus, rhizobia may exist as soil saprophytes, which competes for resources with other microorganisms, or as plant endosymbionts (Poole et al. 2018).

The molecular dialogue that is established between rhizobia and legumes begins with the exudation by the plant of phenolic compounds called flavonoids (López-Baena et al. 2016; Peck et al. 2013). When appropriate, these flavonoids interact with the bacterial protein NodD, a LysR transcriptional regulator that binds to conserved DNA sequences called nod boxes and activates the expression of rhizobial genes involved in the production and secretion of molecular signals called Nod factors (NFs). NFs, in turn, and when appropriate, are perceived by plant LysM receptors and trigger bacterial infection and nodule organogenesis (Oldroyd 2013; Roy et al. 2020). In some rhizobia, such as Sinorhizobium fredii HH103, NodD also activates the expression of another transcriptional regulator, TtsI, which is the positive regulator of a symbiotic type 3 secretion system (T3SS). This symbiotic T3SS delivers effector proteins (called T3Es) inside plant cells (López-Baena et al. 2008; Staehelin and Krishnan 2015). This secretion system, which is not present in all rhizobia, has a crucial role for some symbiotic associations, showing two opposite faces: it can be absolutely required for the formation of nodules, or it can block nodulation in different legume species or cultivars (Jiménez-Guerrero et al. 2022; Teulet et al. 2022). TtsI activates the expression of genes coding for the T3SS apparatus and T3Es through its binding to conserved DNA sequences named tts boxes. In addition to NFs and T3Es, rhizobial surface polysaccharides, such as exopolysaccharides (EPS), lipopolysaccharides (LPS), K-antigen capsular polysaccharides (KPS), and cyclic glucans (CG), may play crucial roles in symbiosis with legumes (Acosta-Jurado et al. 2016a, 2021; López-Baena et al. 2016).

S. fredii HH103 is a rhizobial strain able to establish effective symbiosis with dozens of different legumes (Acosta-Jurado et al. 2016b; Margaret et al. 2011; Vinardell et al. 2015). In HH103, the presence of genistein, an appropriate inducer flavonoid for this strain (Vinardell et al. 2004a), not only triggers the production and secretion of Nod factors and T3Es, but also represses EPS production and biofilm formation (Acosta-Jurado et al. 2016a) and activates bacterial surface motility (Alías-Villegas et al. 2022). These two latter processes depend on the presence of a functional NodD1 protein. Genistein-induced surface motility also requires the participation of TtsI. RNAseq studies have shown that the presence of genistein affected the expression of 100 genes in HH103 (Pérez-Montaño et al. 2016; Navarro-Gómez et al. 2023): 35 through nod boxes (and NodD1), 35 through tts boxes (and NodD1 and TtsI), and 30 through different mechanisms.

The bacterial flagellum is a complex macromolecular structure that allows bacterial cells to move through liquids (swimming) and, in certain cases, to propel on surfaces (Kearns 2010; Nedeljković et al. 2021). This machine and various chemotaxis systems are coded by a numerous group of genes that are normally clustered in specific regions of the genome and are well conserved in bacteria. Although flagella functioning and many of the genes involved have been studied in different rhizobial species, such as S. meliloti, Rhizobium leguminosarum and Bradyrhizobium diazoefficiens (Aroney et al. 2021), they remain to be studied in other rhizobial species such as S. fredii. In addition, rhizobial motility systems present different peculiarities that optimize their movements in soil and root surface environments. For example, as mentioned above, S. fredii HH103 surface motility is induced by the presence of nod gene inducing flavonoids such as genistein, in a process that is controlled by different symbiotic regulators (NodD1, TtsI, NolR, MucR1) and that depends on the presence of a functional T3SS (Alías-Villegas et al. 2022). Recently it has been shown that the presence of a non-ionic osmotic stress (400 mM mannitol) also induces flagellar-dependent surface motility in S. fredii HH103 (Fuentes-Romero et al. 2023).

In this work we have initiated the characterization of the S. fredii SFHH103_00346 gene, which codes for the flagellar protein FlgJ. The expression of this gene is enhanced by genistein in a NodD1 and TtsI dependent manner and it is driven from an imperfect tts box previously not described in the HH103 genome. We have obtained two independent strains with mutations in this gene. Inactivation of flgJ completely abolished swimming, partially impaired genistein-induced surface motility and increased biofilm formation in the absence of flavonoids. Moreover, although the mutation of flgJ did not affect the ability of HH103 to colonize soybean roots, it provoked a reduction of the bacterial symbiotic performance with this host plant when bacteria are not directly inoculated onto the root, a defect that is also exhibited by a HH103 ΔflaCBAD mutant.

Materials and methods

Basic molecular and microbiological techniques

Bacterial strains and plasmids used in this work are listed in Table S1. Sinorhizobium fredii HH103 RifR (Madinabeitia et al. 2002) and its mutant derivatives were grown at 28ºC on TY medium (Beringer 1974), yeast extract/mannitol (YM) medium (Vincent 1970), Bromfield medium (BM) (0.04% tryptone, 0.01% yeast extract, and 0.01% CaCl2 2H2O) (Sourjik and Schmitt 1996) or minimal medium (MM) containing glutamate (6.5 mM), mannitol (55 mM), mineral salts (1.3 mM K2HPO4, 2.2 mM KH2PO4·3H2O, 0.6 mM MgSO4·7H2O, 0.34 mM CaCl2·2H2O, 0.022 mM FeCl3·6H2O, 0.86 mM NaCl), and vitamins (0.2 mg/l biotin, 0.1 mg/l calcium pantothenate) (Beringer 1974). Escherichia coli was cultured on LB medium (Sambrook and Russell 2001) at 37ºC. When required, the media were supplemented with the appropriate antibiotics as described by Vinardell et al. (2004a). Genistein was dissolved in ethanol at a concentration of 1 mg/ml and used at 1 µg/ml.

Plasmids were transferred from E. coli to rhizobia by triparental mating using plasmid pRK2013 as a helper as described by Simon (1984).

To improve reproducibility, all liquid cultures of S. fredii HH103 and its derivatives were routinely initiated from glycerol stocks. The ability of the different strains to grow in liquid BM and MM was monitored every 2 h in a Genesys 20 visible spectrophotometer (ThermoFisher Scientific).

Details about construction of mutants are provided in Supplementary Methods. Recombinant DNA techniques were performed according to the general protocols of Sambrook and Russell (2001). For hybridisation, DNA was blotted to Amersham HybondTM-N nylon membranes, and the DigDNA method of Roche was employed according to manufacturer’s instructions. PCR amplifications were performed as previously described (Vinardell et al. 2004b). qPCR assays were performed as described by Pérez-Montaño et al. (2016). Briefly, total RNA was isolated using a High Pure RNA Isolation Kit (Roche) and RNAase Free DNAse (Qiagen) according to the manufacturer’s instructions. This (DNA free) RNA was reverse transcribed to cDNA using a QuantiTec Reverse Transcription Kit (Qiagen). Quantitative PCR was performed using a LightCycler 480 (Roche, Switzerland) with the following conditions: 95ºC, 10 min; 95ºC, 30 s; 50ºC, 30 s; 72ºC, 20 s; forty cycles, followed by the melting curve profile from 60 to 95ºC to verify the specificity of the reaction. The S. fredii HH103 RNA 16 S gene was used as an internal control to normalize gene expression. The fold changes of two biological samples with three technical replicates in each condition were obtained using the ∆∆Ct method (Pfaffl et al. 2002). All primer pairs used in this work are listed in Table S2.

Bioinformatic analyses

Search of protein sequence homologies were performed with Blastp (https://blast.ncbi.nlm.nih.gov/Blast.cgi). Protein domains were identified by using NCBI Conserved Domain Search, which is available at https://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi (Lu et al. 2020). Structural homologies were analysed by using the Phyre2 (Protein Homology/analogY Recognition Engine V 2.0) software that it is available at http://www.sbg.bio.ic.ac.uk/phyre2/html/page.cgi?id=index (Kelley et al. 2015).

Motility assays

The methodology employed for this kind of assays is thoroughly described in Fuentes-Romero et al. (2024). Briefly, surface motility was analyzed on semisolid MM plates, containing 0.4% agarose D1 Low EEO (Condalab, Spain). Plates were supplemented with ethanol (which is the solvent used for flavonoids) or genistein and were inoculated with 2 µL of washed 10-fold concentrated cultures grown in TY broth to the late exponential phase (OD600nm = 1). Plates were prepared following a rigorous protocol according to Alías-Villegas et al. (2022) and Fuentes-Romero et al. (2024). Swimming was examined on plates prepared with BM containing 0.3% Bacto-agar with ethanol or genistein and inoculated with 3 µL aliquots of rhizobial cultures grown in TY (OD600nm = 1). The migration zone was determined as the colony diameter (mm) after 72 h (for swimming and surface motility) of incubation. In the case of surface motility experiments performed on semisolid MM, in which fractal patterns with characteristic tendrils were formed, migration zones were calculated as the average length of the two sides of a rectangle able to exactly frame each colony. Swimming and surface motility experiments were performed at least with two and three biological replicates, respectively, and three technical replicates each one. In all cases, statistical comparisons were performed by using the non-parametric test of Mann–Whitney.

Transmission electron microscopy (TEM)

Cells for TEM observations were obtained from the edge of surface colonies after 24 h of incubation in the presence of genistein or ethanol at 28ºC. Carbon-coated Formvar grids were placed for 5 min on top of a drop of water previously applied to the colony border. The grids were then washed twice in water for 1 min and stained with 2% (w/v) uranyl acetate for 3 min. Grids were allowed to air dry for at least 1 h and visualized using a Zeiss Libra 120 TEM a 100 kV beam at the Microscopy Service in Centro de Investigación, Tecnología e Innovación (CITIUS, University of Seville).

Biofilm formation and EPS quantification

Assays for biofilm formation on plastic surfaces were carried out as described by Rodríguez-Navarro et al. (2014). Data presented are the average of at least three independent experiments performed in duplicate; in each experiment, at least 12 wells for each treatment were measured. For EPS quantification, bacterial cultures were grown in YM for 96 h at 28 ºC with shaking (OD600 = 1.2–1.3). Cells were removed by centrifugation (20,000 g, 15 min) and total carbohydrate amounts of the EPS-containing supernatants were measured by using the anthrone-H2SO4 method, which determines the amount of glucose equivalents (i.e., the total reducing sugar content) in a given sample, as described by Abarca-Grau et al. (2012). Data presented are the average of three independent experiments performed in duplicate. Statistical comparisons for analyses of biofilm formation and EPS quantification were performed by using the non-parametric test of Mann–Whitney.

Soybean root colonization assays

For these assays, we followed the methodology described by Pérez-Montaño et al. (2014) with some modifications. Briefly, pre-germinated Glycine max (L.) Merrill cv. Williams seeds were aseptically transferred to a stainless-steel lattice placed in a glass cylinder containing 150 ml of Fåhraeus solution (Vincent 1970), pH 6.8, and grown in a controlled environment chamber with a 16-h day, 8-h night cycle (light intensity of 500 µE x m−2 x s−1) and a relative humidity of 50%. Growth temperatures were set to 26 °C during the day and 18 °C during the night. The hydroponics system was inoculated at the time of transferring the pre-germinated seed with approximately 107-108 bacteria per glass cylinder. Three days after inoculation, 2-cm of the upper part of the soybean roots were excised, washed two times in 35-mL of sterile deionized water to eliminate those bacteria that were not firmly attached, and individually introduced in a microcentrifuge vial containing 1 mL of sterile deionized water. Each vial was subjected to two 1-minute ultrasonic cycles in a Bandelin (Berlin, Germany) DL 102-H digital ultrasonic bath at 40% of maximum power. The liberated bacteria were quantified by plate count of serial dilutions in TY medium supplemented with the appropriate antibiotics. For each bacterial strain, at least three independent experiments with 6 plants per treatment were carried out. Results, obtained as c.f.u. (colony forming units) per cm of root, were expressed as the % observed for each treatment compared to the value obtained for the wild-type strain.

In addition, roots of soybean seedlings inoculated with HH103 or its flgJ mutant derivatives carrying the DsRed fluorescent marker (Kelly et al. 2013) were visualised eleven days upon inoculation on a Zeiss Apotome.2, following the method described by Pérez-Montaño et al. (2014) and using λ of 558 and 583 for excitation and detection, respectively. Soybean root auto-fluorescence (green) is shown in white for more clarity. Pictures were processed and merged by using Zen 3.4 blue edition and ImageJ software. The numbers of bacteria attached per mm2 of root were calculated by using ImageJ software and compared by using the non-parametric test of Mann-Whitney. Two independent assays were carried out, and three images per treatment of each assay were analysed.

Nodulation assays

Nodulation assays on Glycine max (L.) Merr. cv. Williams (soybean) were carried out in Leonard jars containing two plants, as previously described by Hidalgo et al. (2010). Germinated seeds were transferred to the upper compartment of sterilized Leonard jars. This compartment contains a vermiculite and perlite mix (2:1) supplemented with 250 mL of Fåhraeus nutrient solution. The upper part of each jar was connected by a cotton wick to a lower compartment containing 1 L of the same nutrient solution. Each Leonard jar contained two plants that, upon inoculation, were grown for seven weeks with a 16 h-photoperiod at 25ºC in the light and 18ºC in the dark. Two different kind of nodulation assays were performed: in one of them, each Leonard jar was inoculated in the surface centre with approximately 108 bacteria, whereas in the second type the root of each seedling was directly inoculated with approximately 5 × 107 bacteria. For each type of experiment, two independent assays were carried out. Figure 1 and S2 show a representative experiment of each type. Shoots were dried at 70 °C for 48 h and weighed. Bacterial isolation from surface-sterilized nodules was carried out as previously described by Buendía-Clavería et al. (2003). Nitrogenase activity of nodules was assessed by ARA as described previously (Acosta-Jurado et al. 2019). Data are given by plant, except for ARA (in this case, data are given by Leonard jar). For the different parameters analysed, the values of each treatment were compared to those of S. fredii HH103 by using the non-parametric test of Mann-Whitney.

Fig. 1
figure 1

Symbiotic performance of S. fredii HH103 and different mutants with Glycine max cv. Williams 82 following two different kinds of inoculation A, Directly onto seedling roots. B, In the centre on the top of the Leonard jar. Every treatment was compared to the wild-type strain following the Mann–Whitney non-parametric test. Significant differences are denoted by asterisks: *, p < 0.05, **, p < 0.01, ***, p < 0.001; ****, p < 0.0001. ARA, Acetylene Reduction Assay

Full size image

Results

The S. fredii HH103 00346, 00347 and 00348 genes code for proteins involved in flagellar assembly

In a previous work we analysed the effect of the presence of the flavonoid genistein on global gene expression of S. fredii HH103 (Pérez-Montaño et al. 2016). Among the HH103 genes showing genistein-induced expression, three contiguous chromosomal ORFs, showing the same transcriptional orientation and called SFHH103_00346, SFHH103_00347, and SFHH103_00348, were found (Genbank accession NC_016812) (Fig. 2A).

Fig. 2
figure 2

A, Genetic organization of the SFHH103_00346 (flgJ), SFHH103_00347 (flgN), and SFHH103_00348 (motF) genes. TSS, transcription start site as detected by cappable-seq. B, Agarose gel electrophoresis of samples resulting from PCR and RT-PCR experiments. The templates used were cDNA of HH103 grown in the absence (-) or presence (+) of genistein, HH103 genomic DNA (gDNA), or absence of DNA (- DNA), and are indicated on the top of each line. MW stands for 100-bp ladder of New Englands Biolabs (Massachusetts, US). Primers used, as represented in panel A, were: (1) q346_F/q346_R; (2) 346–347_F/346–347_R; (3) 347–348_F/347–348_R; (4) rt-16S_F2/rt-16S_R2.

Full size image

SFHH103_00346 is 546-bp long, has been annotated as flgJ, and codes for a protein of 181 residues (CCE94846.1) that is 100% identical to the orthologous proteins of S. fredii strains CCBAU45436 and USDA205, 98% to that of S. fredii NGR234, and more than 74% to the corresponding proteins of other Sinorhizobium species (Supplementary Dataset S1). During flagellar assembly, the different structural subunits crystallise beneath cap structures that are located at the distal tip of the nascent flagellum. FlgJ functions as the rod cap (Evans et al. 2014). The FlgJ proteins of β- and γ-proteobacteria are bifunctional, harbouring a N-terminal domain that is required for flagella assembly and a C-terminal domain possessing peptidoglycan lytic activity (Herlihey et al. 2014). In contrast, the FlgJ protein of the α-proteobacterium Rhodobacter sphaeroides only contains the domain involved in flagellar assembly, whereas the peptidoglycan lytic domain is harboured by a different protein called SltF (de la Mora et al. 2007; González-Pedrajo et al. 2002). This appears to be also the situation in other α-proteobacteria as the members of the Sinorhizobium genus, since HH103 FlgJ only contains a rod-binding (pfam10135) domain (Supplementary Dataset S1), which is characteristic of proteins that are involved in the assembly of the prokaryotic flagellar rod. In addition, the SFHH103_00334 gene from HH103 codes for a 188-amino acid protein which contains an Slt (soluble lytic transglycosilase) domain which is characteristic of muramidases like SltF.

SFHH103_00347 and SFHH103_00348 are 369-bp and 567_bp long, respectively, and code for two hypothetical proteins (CCE94847.1 and CCE94848.1) lacking known domains. The predicted product of SFHH103_00347 is a 122-residue polypeptide that is 100%, 99%, 88% and 65% identical to the proteins N181_01515, NGR_c03050, SMc03072, and CN09_15695 of S. fredii USDA 205, S. fredii NGR234, S. meliloti 1021, and Rhizobium (Agrobacterium) rhizogenes, respectively (Supplementary Dataset S2). No significant sequence similarities to proteins having an assigned function could be found when the predicted product of SFHH103_00347 was used as a query in Blastp searches. However, the use of the Phyre2 (Protein Homology/analogY Recognition Engine V 2.0) web utility predicted structural identities of this polypeptide (93.0% of confidence and 90.0% of coverage) to Salmonella enterica subsp. enterica serovar Typhimurium FlgN (Supplementary Dataset S3), a FlgK type chaperone involved in flagellar assembly (Halte and Erhardt 2021; Rossi et al. 2023). SFHH103_348 codes for a putative polypeptide of 188 residues that show identities ranging from 73 to 100% to the orthologous proteins of different strains belonging to the Sinorhizobium genus, including the FliL protein of Sinorhizobium terangae (WP_153438383.1, 189 residues, 86% Id) and Smc03057 of S. meliloti (CAC45263, 188 residues, 80% Id) (Supplementary Dataset S4). The latter protein has been recently described in S. meliloti RU11/001 as MotF, a paralog of the FliL protein (Sobe et al. 2022). FliL is a transmembrane protein that is associated to the flagellar membrane-embedded basal body and that is important for flagella functioning in viscous environments (Takekawa et al. 2019). Interestingly, the SFHH103_00348 encoded putative polypeptide also presented structural homology (> 95.9% of confidence and 50.0% of coverage) to FliL of Vibrio alginolyticus and Helicobacter pylori (Supplementary Dataset S3), although no significant homologies were found between the encoded product of SFHH103_00348 and the FliL protein of HH103 (coded by SFHH103_00324). Since the orthologs genes of SFHH103_00347 and SFHH103_00348 in S. meliloti has been renamed as flgN and motF (Sobe et al. 2022), hereafter we used the same nomenclature.

The S. fredii flgJ, flgN, and motF genes do not form a single transcriptional unit

The pattern of induction with genistein of flgJ, flgN, and motF suggested that these three genes might belong to a single transcriptional unit. To analyse this hypothesis, we performed RT-PCR experiments for searching putative mRNAs covering the 3’-end of flgJ and the 5’-end of flgN (primers 346–347_F/346–347_R), and the 3’-end of flgN and the 5’-end of motF (primers 347–348_F/347–348_R) by using cDNA obtained from HH103 cultures grown either in the absence or in the presence of genistein (Fig. 2B). As a positive control, primers q346_F/q346_R and rt-16S_F2/rt-16S_R2, which allow the amplification of internal fragments of the flgJ and rRNA 16 S genes respectively, were employed. In addition, S. fredii HH103 genomic DNA (gDNA) and RNA samples were used as positive and negative controls for PCR amplification, respectively.

As expected, all the primer pair tested allowed the amplification of fragments of the expected size when gDNA was used as template, whereas PCR reactions with RNA samples gave negative results. When cDNA samples (regardless of they were obtained from cultures grown in the absence or presence of genistein) were used as template, the internal fragments of flgJ and the rRNA 16 S gene, as well as the fragment connecting the flgN and motF genes could be amplified. Instead, the amplification of a fragment connecting the flgJ and flgN failed with the two cDNA samples tested. We tried a second primer pair that would lead to the amplification of a fragment connecting the flgJ and flgN genes (primers 346–347_F2/346–347_R2) with the same negative result. Thus, our results suggest that flgN and motF, but not flgJ, belong to the same transcriptional unit.

The S. fredii HH103 flgJ gene is preceded by a functional tts box and is positively regulated by TtsI and NodD1

Previous results of our group indicated that flgJ, flgN, and motF belongs to the nod regulon (Pérez-Montaño et al. 2016). RNAseq analyses revealed that all these three genes exhibited enhanced expression in the presence of genistein (+ 7.6, + 7.0, and + 4.3 respectively), and this positive effect of the presence of the flavonoid was absent in nodD1 or ttsI HH103 mutants (Pérez-Montaño et al. 2016). For the flgJ gene, these transcriptomic data were validated by qPCR analysis (Pérez-Montaño et al. 2016). Due to the expression pattern and the putative function of the product encoded by flgJ, we decided to investigate this gene and analyse its possible role on bacterial motility and symbiosis with soybean, a host legume of HH103. For this purpose, we constructed two independent mutants in this gene, as described in Supplementary Methods. One of them is a non-polar mutant obtained by in frame deletion and called HH103 ΔflgJ. The second HH103 flgJ derivative was generated by insertion of the lacZΔp-GmR cassette into the flgJ gene (hereafter HH103 flgJ::lacZ).

To further validate the pattern of expression of flgJ, we used HH103 flgJ::lacZ (in which flgJ is fused to a lacZ lacking its own promoter) for β-galactosidase assays performed both in the presence or in the absence of genistein, an effective inducer flavonoid for strain HH103 (Vinardell et al. 2004a) (Fig. 3A). As a positive control we used a HH103 derivative carrying a Tn5-lacZ insertion into nodA, a gene that is positively regulated by NodD1 in the presence of appropriate flavonoids (Vinardell et al. 2004b). As expected, the presence of genistein enhanced the expression of nodA nearly 8-fold. Regarding flgJ, its expression was 2.1 times higher in the presence of the flavonoid.

Fig. 3
figure 3

The S. fredii HH103 flgJ expression is driven by a tts box and depends on the NodD1 and TtsI transcriptional regulators. A, β-galactosidase assays of strain HH103 nodA::Tn5-lacZ and HH103 flgJ::lacZΔp-GmR carried out in the presence or absence of genistein 3.7 µM. Asterisks denote significant differences at p = 0.001 as determined by using the non-parametric test of Mann-Whitnney. B, Nucleic acid sequences of the HH103 tts box consensus sequence and the tts box that is located upstream of the flgJ gene. C, β-galactosidase assays of S. fredii HH103 and its nodD1::Ω and ttsI::Ω mutant derivatives harbouring plasmid pMUS1393 and carried out in the presence or absence of genistein 3.7 µM. Asterisks denote significant differences at p = 0.05 as determined by using the non-parametric test of Mann-Whitnney.

Full size image

In a previous work, our group had located 18 putative tts boxes in the S. fredii HH103 genome (Vinardell et al. 2015), although only 11 out of those 18 tts boxes appeared to be functional (Pérez-Montaño et al. 2016) since they were proven to drive gene expression upon induction with genistein and depending on the presence of functional NodD1 and TtsI proteins. None of those 18 putative tts boxes was located upstream of flgJ. However, the analysis of the upstream sequence of flgJ revealed the presence of a sequence (positions − 152 a -125 with respect to the translational start site) that partially matched the consensus sequence used for identification of tts boxes in HH103, mainly in the 5’ part (Fig. 3B). In order to check whether this sequence actually functions as a tts box, a 191-bp fragment containing this sequence was PCR-amplified (Table S2) and subcloned into vector pMP220, a TcR broad-host range vector with contains a lacZ gene lacking a promoter sequence. The resulting plasmid, called pMUS1393, was transferred to S. fredii HH103 and its nodD1 and ttsI derivatives, and β-galactosidase assays were performed both in the absence and presence of genistein. As it is shown in Fig. 3C, lacZ expression was enhanced by genistein in the wild-type strain, but not in its nodD1 or ttsI derivatives, which strongly suggests that the putative tts box harboured by that fragment is functional and able to drive gene expression in the presence of inducing flavonoids and functional copies of NodD1 and TtsI.

S. fredii HH103 mutants in flgJ are negatively affected in bacterial motility

Since FlgJ is predicted to be a flagellar protein, we decided to investigate its putative role on bacterial motility. On the one side, bacterial swimming motility was analysed in Bromfield medium supplemented with 0.3% w/v of agar. On the other side, surface motility was studied in minimal medium supplemented with 0.4% w/v of agarose (Alías-Villegas et al. 2022; Fuentes-Romero et al. 2024). Since flgJ shows enhanced expression in the presence of flavonoids, and S. fredii HH103, in our experimental conditions, only exhibit surface motility in the presence of nod gene inducing flavonoids such as genistein, our studies were carried out in the absence or presence of this flavonoid. In all these experiments, an HH103 ΔflaCBAD, which lacks flagella and its fully impaired in swimming and partially in surface motility (Acosta-Jurado et al. 2016b; Alías-Villegas et al. 2022), was also included.

The growth abilities in either minimal or Bromfield medium of S. fredii HH103 and its two independent mutants in flgJ were tested. As it is shown in Fig. S1, none of those mutations has a significant effect on the growth of strain HH103 in those media.

Swimming assays revealed that both flgJ mutants were fully impaired in this kind of motility when compared to the parental strain, regardless of the presence or absence of genistein (Fig. 4). In fact, these mutants exhibited a similar behaviour to that of HH103 ΔflaCBAD.

Fig. 4
figure 4

Swimming abilities of S. fredii HH103 and different mutants in the absence and the presence of genistein at 72 h. All data points are shown in box and whisker plots from at least two biological replicates performed with three technical replicates. Asterisks (****) indicate significant differences with the corresponding control sample using the non-parametric test of Mann–Whitney, p < 0.0001. Representative pictures of the motilities exhibited in the presence of genistein by the different S. fredii HH103 mutants analysed are also shown. Values under images represent the average and standard deviation of migration (given in millimetres and determined as described in the text)

Full size image

As shown in Fig. 5, both flgJ mutants showed decreased surface motility in the presence of this flavonoid when compared to the wild-type strain. Actually, the surface motility ability of the two HH103 flgJ mutants was similar to that of HH103 ΔflaCBAD.

Fig. 5
figure 5

Surface motility of S. fredii HH103 and different mutants in the absence and the presence of genistein at 72 h. All data points are shown in box and whisker plots from at least three biological replicates performed with three technical replicates. Asterisks (****) indicate significant differences with the corresponding control sample using the non-parametric test of Mann–Whitney, p < 0.0001. Representative pictures of the motilities exhibited in the presence of genistein by the different S. fredii HH103 mutants analysed are also shown. Values under images represent the average and standard deviation of surface migration (given in millimetres and determined as described in the text)

Full size image

The fact that inactivation of flgJ negatively affected bacterial motility and the predicted function of its encoded product prompted us to investigate whether these mutants might be affected in flagellum biosynthesis. Transmission Electronic Microscopy (TEM) analyses (Fig. 6) revealed that both flgJ mutants lack flagella.

Fig. 6
figure 6

Transmission electron microscope (TEM) images of S. fredii HH103 and different mutants. Cells were isolated from the edge of colonies grown on semisolid MM 0.4% agarose in the presence of genistein or ethanol after 24 h of incubation at 28 ºC and stained with 2% uranyl acetate. Scale bars correspond to 1 μm

Full size image

Inactivation of flgJ affects the ability to form biofilms on plastic surfaces but not the colonisation of soybean roots

Due to the pattern of expression and the involvement of flgJ in bacterial motility, we decided to investigate whether the inactivation of this gene influences the bacterial abilities to form biofilms on plastic surfaces and to colonize soybean (an HH103 host-legume) roots.

The ability of S. fredii HH103 to form biofilms on plastic surfaces is negatively affected by the presence of flavonoids (Acosta-Jurado et al. 2016a). As shown in Fig. 7, the amount of biofilm formed by both flgJ mutants (ΔflgJ and HH103 flgJ::lacZ) was indistinguishable from that of the wild-type strain when grown in the presence of genistein, but significantly higher when flavonoids were not present.

Fig. 7
figure 7

Biofilm formation ability on plastic microtiter plates of S. fredii HH103 (wt) and its ΔflgJ and flgJ::lacZ mutants grown in the presence (+) or absence (-) of genistein. The presence of three asterisks denotes a significant difference (p < 0.001, non-parametric test of Mann–Whitney) with respect to the wild-type strain grown in the same condition

Full size image

EPS plays a key role in the formation of biofilms in rhizobia, including S. fredii HH103 (Downie 2010; Rodríguez-Navarro et al. 2014). The increased ability to form biofilms exhibited by the flgJ mutants of HH103 prompted us to investigate whether they were affected in the production of EPS. Since HH103 EPS production is repressed in the presence of nod-gene inducing flavonoids (Acosta-Jurado et al. 2016a, 2020), we investigated this trait both in the presence or absence of genistein. The production of EPS was studied by quantification of the amount of glucose equivalents found in the extracellular milieu of cultures grown in YM broth either in the presence or absence of genistein. No differences could be scored among the amount of glucose equivalents (µg mL−1) produced by both flgJ mutants and the wild-type strain, either in the absence (55.9 ± 9.8 and 57.6 ± 9.5 for HH103 flgJ::lacZ and HH103 ΔflgJ vs. 60.1 ± 8.7 for HH103) or the presence or genistein (36.1 ± 1.3 and 37.3 ± 4.4 vs. 33.6 ± 4.5).

We have also studied whether the HH103 flgJ mutants might be affected in the bacterial ability to colonize soybean roots. For this purpose, we followed two methodological approaches. On the one hand, we quantified the number of bacteria attached to soybean roots 3 days after inoculation by sonicating these roots and calculating the number of CFUs (colony forming units) that were present in the resulting bacteria suspensions. Both flgJ mutants showed similar levels of colonization of soybean roots when compared to the parental strain (115.5 ± 14.3% for ΔflgJ and 104.7 ± 13.7% for HH103 flgJ::lacZ). On the other hand, we introduced plasmid pFAJDsRed (Kelly et al. 2013) that contains the DsRed flourescent marker in both flgJ mutants and in the wild-type strain, thus allowing the visualization of bacterial biofilms on soybean roots by fluorescence microscopy (Fig. 8) as well as the quantification of the number of bacteria attached per mm2 of root. These numbers, expressed as average ± SEM, were similar (and not significantly different by using the non-parametric test of Mann-Whitney) for the wild-type strain (2,120 ± 133) and the flgJ::lacZ and ΔflgJ mutants (2,128 ± 331 and 2,220 ± 330, respectively). In addition, as shown in Fig. 8, the patterns of bacterial cell attachment to the root were also similar for the three strains tested. In general, bacteria were homogeneously distributed over the root surface, although occasional accumulation of bacteria could also be seen for the three strains analysed.

Fig. 8
figure 8

Fluorescence microscopy of soybean roots inoculated with S. fredii strains harbouring plasmid pFAJDsRed. Bars correspond to 100 μm

Full size image

Inactivation of flgJ has a negative effect on symbiosis with soybean when bacterial cultures are not directly applied onto the roots

We have also investigated whether the inactivation of flgJ has an effect on the symbiotic performance with a host legume of S. fredii HH103: Glycine max cv. Williams, an American variety of soybean (Margaret et al. 2011). Due to the involvement of this gene in bacterial motility, we carried out two different kinds of nodulation assays. In the first type, inoculation of the plants was carried out onto each seedling (which is the usual procedure in our nodulation assays). In the second type, bacterial cultures were applied in the middle of the surface of the vermiculite containing vessel, in an equidistant position (approximately 2–3 cm) with regard to the different seedlings located in that compartment. Regardless of the type of nodulation assay carried out, four different parameters were analysed per plant: number of pink (nitrogen-fixing) nodules, fresh mass of nodules, dry mass of the aerial part (shoot), and nitrogen-fixation ability (estimated by ARA). Figure 1 shows a representative experiment of each type. Although in both kind of assays the HH103 flgJ mutants were able to induce the formation of nitrogen-fixing nodules, the results obtained were different depending on the type of inoculation employed.

When the bacterial inoculum was applied onto the soybean seedlings (Fig. 1A), no differences could be scores between the flgJ mutants and the wild-type strain in the different parameters analysed. When the inoculum was applied at a certain distance of the seedlings (Fig. 1B), both flgJ mutants induced the formation of fewer nodules than the wild-type strain, although the differences were not significant. In contrast, the values of the fresh mass of nodules, the shoot dry mass, and nitrogen-fixation were significantly lower for these two mutants than for the parental strain.

Similar results were obtained when we analysed the symbiotic behaviour of a HH103 ΔflaCBAD mutant with Williams soybean (Figure S2). This strain behaved similarly to HH103 when inoculants were applied onto root seedlings. However, both the fresh mass of nodules and the shoot dry mass of plants inoculated with this mutant were significantly lower than the respective values obtained in plants inoculated with the wild-type strain when inoculants were applied at 2–3 cm from soybean seedlings.

Discussion

The expression of rhizobial symbiotic genes is under the control of a complex regulatory circuit driven by flavonoids exuded by the plant and sensed by the bacterial LysR regulator NodD. Additional regulators can participate in this regulatory circuit, including TtsI, which is the transcriptional activator of the symbiotic T3SS (López-Baena et al. 2016). Previous results of our group have shown that, in Sinorhizobium fredii HH103, the presence of the flavonoid genistein not only triggers the production of two symbiotic signals (NFs and T3Es) but also represses the production of EPS and induces bacterial surface motility (Acosta-Jurado et al. 2016a; Alías-Villegas et al. 2022). The analysis of transcriptomic data of the effect of flavonoids on rhizobial gene expression might shed light on the regulation of certain bacterial traits related to symbiosis (Jiménez-Guerrero et al. 2018). In this work we show that the flgJ gene, whose expression is activated by genistein and NodD1 through the action of TtsI, is involved in genistein-induced surface motility. We also demonstrated that flgJ expression is driven by a tts box, previously unidentified, that is located upstream of this gene. A recently cappable-seq analysis of HH103 total RNA performed by our group confirmed the presence of a transcription start site located between that tts box and the flgJ gene.

Inactivation of flgJ completely abolished swimming motility, accordingly with the fact that this gene codes for the flagellar protein FlgJ. In α-Proteobacteria, this protein has a single domain and acts as scaffold or cap essential for flagellar rod assembly (de la Mora et al. 2007; Nambu et al. 2006). In addition, we show that mutation of HH103 flgJ results in the loss of flagella and, consequently, in that of swimming motility. Downstream of flgJ, there are two other genes SFHH103_00347 and SFHH103_00348, that code for hypothetical proteins that maybe involved in flagellar motility. Recently, an article about the S. meliloti SFHH103_00346-SFHH103_00348 orthologues (called SMc03071, SMc03072 and SMc03057 respectively) has been published (Sobe et al. 2022). In that work the authors demonstrate that these three genes are involved in motility since mutation of any of them led to swimming impairment. They also demonstrate that SMc03057 is a distantly related paralog of FliL that has been named as MotF and works as a motor protein required for proper flagella functioning. According to the nomenclature established in S. meliloti (Sobe et al. 2022), we renamed SFHH103_00347 and SFHH103_00348 as flgN and motF, respectively. Although previous RNAseq experiments carried out in S. fredii HH103 indicated that flgN and motF are also induced in the presence of genistein in a NodD1 and TtsI dependent manner, in this work we show that flgJ is not co-transcribed with them. Because of this, in the present work we have focused in flgJ, but we plan to investigate the putative roles of HH103 flgN and motF in motility and/or symbiosis in the near future.

As previously described, S. fredii HH103 exhibits surface motility upon induction with flavonoids in a NodD1- and TtsI-dependent manner (Alías-Villegas et al. 2022). Moreover, this surface motility is the result of two different mechanisms, one dependent (swarming) and one independent (unidentified) of flagella. The regulation of surface motility in HH103 made flgJ a good candidate to be involved in genistein-induced motility, as it has been demonstrated in this work. Interestingly, the two flgJ mutants tested retain part of the wild-type ability for this kind of motility, at a similar level than a non-flagellated HH103 derivative (HH103 ΔflaCBAD). Most probably, according to its function, flgJ is involved in the flagella-dependent genistein-induced HH103 surface motility. Additional research is required to elucidate the genistein-induced flagella-independent surface motility exhibited by HH103.

Bacterial motility and biofilm formation are often oppositely regulated (Amaya-Gómez et al. 2015). As we show in this work, the lack of flgJ clearly increase the ability of HH103 to form biofilms on plastic surfaces. As in most bacteria, HH103 EPS is crucial for biofilm formation (Rodríguez-Navarro et al. 2014). As mentioned before, and differently to other rhizobia, nod gene inducing-flavonoids repress EPS production in S. fredii HH103 (Acosta-Jurado et al. 2016a, 2020). Because of this, biofilm formation by HH103 and its flgJ mutants also diminishes in the presence of genistein. Also, as expected, mutation of flgJ did not affect the amounts of EPS produced either in the presence or absence of flavonoids, which suggest that the increase on biofilm formation caused by the lack of FlgJ in the absence of flavonoids might be related to the absence of flagella-dependent motility (either swimming or swarming). In contrast, the lack of flgJ and, consequently, that of functional flagella, does not influence the ability of S. fredii HH103 to colonize soybean roots as we have determined by two different methodologies.

Previous works have shown that rhizobial motility is not essential for nodulation and nitrogen fixation but may positively influence the efficiency of the symbiotic interaction with legumes by facilitating bacteria to reach infection sites and/or rhizobial progression inside infection threads (Aroney et al. 2021; Fournier et al. 2008; Wheatley et al. 2020).

Coherently with these previous observations, the symbiotic phenotype of HH103 mutants in flgJ varied depending on the mode of bacterial inoculation of the seedlings. When bacteria were applied onto the seedling roots, the symbiotic performance of both flgJ mutants was undistinguishable of that of the wild-type strain in all the parameters scored in this work. It is necessary to remark that flavonoids present in soybean root exudates promote surface spreading of HH103 and that the flagella-independent mechanism is actually working in the flgJ mutants. However, when the bacterial inoculum was deposited several centimetres far from the seedling roots, the two mutants showed a certain degree of impairment on symbiosis with soybean. In fact, although the number of nodules were similar to those formed by plants inoculated with the wild-type strain, in both flgJ mutants the fresh mass of nodules, the shoot dry mass and the nitrogenase activity were significantly lower than those of the plant inoculated with the wild-type strain (Fig. 1). These results indicate that the absence of flgJ, although did not affect soybean root colonization in our laboratory conditions, may provoke a diminution in the ability of the bacterium to reach the infection sites and/or to move inside infection threads and reach the developing nodules. Similar results, as commented above, have previously reported for other rhizobial strains defective in motility (Aroney et al. 2021; Bernabéu-Roda et al. 2015; Caetano-Anollés et al. 1988; Fournier et al. 2008; Wheatley et al. 2020). In fact, the symbiotic phenotype of both flgJ mutants matches that of an HH103 fla mutant, that also shows reduced fresh mass of nodules and shoot dry mass when the inoculum is applicated at a certain distance from the seedlings (Fig. S2), whereas not differences could be scored with the wild-type strain when bacteria are directly applicated on the seedlings. These results also suggest that the symbiotic phenotype exhibited by both flgJ mutants is most probably a consequence of the absence of flagella rather than a specific effect of the lack of the FlgJ protein.

One interesting point is why only three HH103 flagellar genes (flgJ, flgN, motF) are differentially expressed in the presence of nod gene inducing flavonoids, a condition that promotes surface spreading in this rhizobial strain (Pérez-Montaño et al. 2016; Alías-Villegas et al. 2022; Navarro-Gómez et al. 2023). It is possible that an enhanced expression of these genes might act as a checkpoint required for flagellar-dependent surface motility. Recently we have carried out transcriptomic studies that show that NolR, NodD2, and SyrM repress the expression of flgJ, flgN, and motF (Acosta-Jurado et al. 2020; Navarro-Gómez et al. 2023). However, the former protein is absolutely required for HH103 genistein-induced surface spreading, whereas NodD2 and SyrM appears to be dispensable for this kind of motility (Alías-Villegas et al. 2022). In addition, another HH103 symbiotic regulator, MucR1, has a clear positive impact on the expression of genes related to flagella biosynthesis and functioning (Acosta-Jurado et al. 2016c), although the set of genes affected varied according to the absence or presence of flavonoids. In fact, in a HH103 mucR1 mutant, the expression of flgJ, flgN and motF was significantly different (increases of 4.5–11.5 fold) to that found in the wild-type strain only when flavonoids were absent. However, HH103 mucR1 was affected neither in swimming (regardless the absence or presence of flavonoids) nor in surface motility in the absence of flavonoids, but it exhibited lower ability for surface spreading than the wild-type strain in the presence of genistein (Alías-Villegas et al. 2022). Thus, the regulation of flagellar-related genes and its impact on bacterial surface motility in HH103 appears to be extremely complex and clearly requires further investigation.

In conclusion, we show in this work that the flgJ gene is symbiotically regulated and participates in both swimming and surface motility. In coherence with its regulation, the inactivation of this gene negatively affects the efficiency of the symbiotic interaction in natural conditions, when bacteria must reach the infection zone of the root. To the best of our knowledge, this is the first time in which a rhizobial motility gene has been described as a member of the nod regulon and shown to have symbiotic relevance.