NopC Is a Rhizobium-Specific Type 3 Secretion System Effector Secreted by Sinorhizobium (Ensifer) fredii HH103

Sinorhizobium (Ensifer) fredii HH103 is a broad host-range nitrogen-fixing bacterium able to nodulate many legumes, including soybean. In several rhizobia, root nodulation is influenced by proteins secreted through the type 3 secretion system (T3SS). This specialized secretion apparatus is a common virulence mechanism of many plant and animal pathogenic bacteria that delivers proteins, called effectors, directly into the eukaryotic host cells where they interfere with signal transduction pathways and promote infection by suppressing host defenses. In rhizobia, secreted proteins, called nodulation outer proteins (Nops), are involved in host-range determination and symbiotic efficiency. S. fredii HH103 secretes at least eight Nops through the T3SS. Interestingly, there are Rhizobium-specific Nops, such as NopC, which do not have homologues in pathogenic bacteria. In this work we studied the S. fredii HH103 nopC gene and confirmed that its expression was regulated in a flavonoid-, NodD1- and TtsI-dependent manner. Besides, in vivo bioluminescent studies indicated that the S. fredii HH103 T3SS was expressed in young soybean nodules and adenylate cyclase assays confirmed that NopC was delivered directly into soybean root cells by means of the T3SS machinery. Finally, nodulation assays showed that NopC exerted a positive effect on symbiosis with Glycine max cv. Williams 82 and Vigna unguiculata. All these results indicate that NopC can be considered a Rhizobium-specific effector secreted by S. fredii HH103.


Introduction
Rhizobia are soil bacteria able to establish a symbiotic interaction with legumes that culminates in the formation of specialized plant organs, called nodules, on the roots of the host plant. Within these symbiotic structures atmospheric nitrogen is reduced to ammonia, which is assimilated by the host plant in exchange of a carbon source and an appropriate environment that promotes bacterial growth [1]. This process requires a complex interchange of molecular signals between the microorganism and the plant. Thus, certain flavonoids exuded by legume roots are recognized by the rhizobial protein NodD, which in turns binds to specific promoter sequences (nod boxes), activating the transcription of the nod genes. Proteins encoded by these genes are responsible for the biosynthesis and secretion of the Nod factors, which are recognized by specific plant receptors to initiate nodule organogenesis [2].
Plant flavonoids, besides inducing Nod factors production, attract the bacteria to the legume root [3], activate the rhizobial quorum sensing systems [4,5], and induce via NodD the secretion of proteins through the type 3 secretion system (T3SS) [6]. This specialized secretion apparatus is a common virulence mechanism shared by many plant and animal pathogenic Gram negative bacteria that delivers proteins directly into the host cells [7,8,9]. These secreted proteins are called effectors and function within the eukaryotic cell, where they interfere with signal transduction cascades and promote infection by suppressing host defenses [10,11].
In rhizobia, secreted proteins are collectively known as nodulation outer proteins (Nops) [12] and are involved in host-range determination and symbiotic efficiency [13]. Recent works have shown that the S. fredii HH103 T3SS is responsible of the suppression of early soybean defense responses to effectively nodulate this legume [14]. In addition, the T3SS of Bradyrhizobium elkanii USDA61 induces the formation of nodules in soybean in the absence of Nod factors when infecting by crack-entry or intercellular infection [15]. Synthesis and secretion of Nops are controlled by the transcriptional regulator TtsI, which binds to specific promoter sequences called tts boxes. TtsI is an intermediary in the regulatory cascade between NodD, previously activated by flavonoids, and the T3SS-related genes [6,16,17,18].
S. fredii HH103, hereafter HH103, is a broad host-range bacterium that nodulates many legumes including soybean, which is considered its natural host plant [19]. HH103 secretes at least eight proteins through the T3SS in response to the inducer flavonoid genistein: NopA, NopB, NopC, NopD, NopL, NopM, NopP, and NopX [20]. NopA, NopB, and NopX are extracellular components of the T3SS machinery [21,22,23] and the rest can be considered putative effectors (NopC, NopD, NopL, and NopM) with the exception of NopP, whose secretion to the interior of Vigna unguiculata nodule cells has been confirmed [24]. Interestingly, two of these proteins, NopL and NopP, are specific to rhizobia and have no homologues in plant or animal pathogens [13]. NopL is phosphorylated by plant kinases and probably interferes with plant signal transduction cascades that are responsible of the activation of plant defense genes [25]. In addition, NopL seems to be involved in the suppression of the nodule premature senescence observed in the symbiosis between S. fredii NGR234 and Phaseolus vulgaris [26]. NopP has also been described as phosphorylated by plant kinases but its function in symbiosis is still unknown [27]. In HH103, the inactivation of the nopP gene causes an increase in the number of nodules formed in American and Asiatic soybeans [28]. Finally, no reports about the possible function or the role in symbiosis of NopC have been published so far.
In this work, we studied the transcriptional regulation of the nopC gene by NodD1, TtsI and flavonoids and the effect of a non-polar mutation in nopC on Nops secretion. We determined that the tts region of HH103 was expressed in young soybean nodules and showed that NopC was translocated into soybean root cells, confirming that this protein can be considered an effector secreted by the HH103 T3SS. Finally, we studied the role of NopC in the symbiosis with Glycine max (soybean) cv. Williams 82 and V. unguiculata.

Microbial and molecular techniques
Bacterial strains and plasmids used in this work are listed in Table 1. Sinorhizobium strains were grown at 28°C on tryptone yeast (TY) medium [29] or yeast extract mannitol (YM) medium [30]. Escherichia coli strains were cultured on Luria-Bertani (LB) medium [31] at 37°C. When required, the media were supplemented with the antibiotics ampicillin (Ap, 100 μg ml -1 ), rifampicine (Rif, 50 μg ml -1 ), spectinomycin (Spc, 50 μg ml -1 ), kanamycine (Km, 30 μg ml -1 ), tetracycline (Tc, 5 μg ml -1 ), and gentamycine (Gm, 5 μg ml -1 ). Genistein was dissolved in ethanol and used at 1 μg ml -1 to give a final concentration of 3.7 μM. Plasmids were transferred from E. coli to Sinorhizobium strains by conjugation as described by [32] using plasmid pRK2013 as helper. Recombinant DNA techniques were performed according to the general protocols of [31]. For hybridization, DNA was blotted to Hybond-N nylon membranes (Amersham, United Kingdom), and the DigDNA method of Roche (Switzerland) was employed following the manufacturer's instructions. PCR amplifications were performed as previously described [5]. Primer pairs used for the amplification of the S. fredii HH103 nopC and nopA genes were fy1secF and fy1secR ( Table 2). Plasmid pMUS986 was obtained by cloning into the broad hostrange vector pMP92 [33] a 1.3-kb PCR fragment containing the nopC and nopA genes and their upstream tts box.
The in frame nopC gene deletion was constructed by overlap extension polymerase chain reaction [34] using the pairs of primers nopC-1/nopC-2 and nopC-3/nopC-4 ( Table 2). Plasmid pK18mobsac was used for the homogenotization of the mutated version of nopC in S. fredii HH103 Rif R [35]. The deletion event was confirmed by PCR and hybridization. The nopC deletion was complemented in trans by the transference of plasmid pMUS986 and in cis by single recombination using plasmid pMUS1192.
Construction of the translational fusion nopC-cya was performed by cloning a DNA fragment containing the nopC gene without end codon, which was amplified using primers listed in Table 2, into plasmid pDONR207 (Invitrogen, USA). The resulting plasmid (pMUS1239) was used to clone the nopC gene into plasmid pLMS150 [24], which possesses recombination sites for the clonase II upstream the cya gene, resulting in plasmid pMUS1290. This gene fusion was confirmed by sequencing. Two primers ( Table 2) were designed to amplify the nopC-cya fusion from pMUS1290 and add EcoRI and BamHI restriction sites. Finally, the 1.5-kb PCR product was digested with EcoRI and BamHI and the resulting DNA fragment was cloned into pK18 mob to obtain plasmid pMUS1291. This plasmid was used for the integration of the nopC-cya fusion by single recombination in the symbiotic plasmids of S. fredii HH103 Rif R and in its ttsIΩ mutant derivative. These integrations were confirmed by PCR amplification using Table 2. DNA oligonucleotide primers used in this study.

Name
Sequence Usage primers described previously ( Table 2). Secretion of the NopC-Cya fusion protein by the T3SS was confirmed using specific antibodies against the cya epitope (data not shown).

RNA isolation, cDNA synthesis and quantitative RT-PCR
qRT-PCR assays were performed as described by [4]. Briefly, S. fredii strains HH103 Rif R , HH103 Rif R nodD1::lacZ-Gm R , HH103 Rif R ttsI::Ω, and HH103 Rif R ΔnopC were grown with shaking at 28°C in YM medium supplemented with genistein when necessary. When cultures reached an OD 600 of 0.8, cells were harvested and RNA was extracted using the High Pure RNA Isolation kit following the manufacturer's instructions (Roche, Switzerland). Two independent RNA extractions were performed. RNA concentration was quantified using a Nanodrop 2000 spectrophotometer (Thermo Scientific, USA) and the integrity of the RNA was assayed on 2% agarose gels. cDNA was synthesized using the Quantitect Reverse Transcription kit according to the manufacturer's instructions (Qiagen, Switzerland).
To quantify the HH103 nopC, nopA, and nodA gene expression using quantitative RT-PCR, primers nopCq_F, nopCq_R, nopAq_F, nopAq_R, nodAq_F, and nodAq_R were designed ( Table 2). The reactions were performed in a 10 μl final volume containing 25 ng of cDNA, 0.6 pmol of each primer and 5 μl of FastStart SYBR Green Master Mix (Roche, Switzerland). PCR was conducted on a Light Cycler 480 II (Roche) with the following conditions: 95°C, 10 min; 95°C, 30 sec; 50°C, 30 sec; 72°C, 15 sec; 45 cycles, followed by the melting curve profile from 65 to 95°C to verify the specificity of the reaction. The threshold cycles (Ct) were determined with the Light Cycler 480 II software and the individual values for each sample were generated by averaging three technical replicates that varied less than 0.5 cycles. Expression was calculated relative to the parental strain grown without flavonoids. The HH103 RNA 16S gene was used as an internal control to normalize gene expression ( Table 2). The fold change in the target gene, normalized to RNA 16S and relative to the gene expression in the control sample was calculated.

Purification and analysis of nodulation outer proteins
Extracellular proteins from several HH103 strains were recovered from 50 ml of YM bacterial cultures grown on an orbital shaker (180 r.p.m.) for 40 h (approximately 10 9 c.f.u. ml -1 ). Cultures were centrifuged for 20 min at 10000 g at 4°C. The supernatants were mixed with 3 volumes of cold acetone and maintained at -20°C for 24 h. The mixtures were centrifuged for 45 min at 22000 g at 4°C. Dried pellets were resuspended in 300 μl of sample buffer (62.5 mM Tris-HCl [pH 6.8], 2% SDS [m/v], 10% glycerol [v/v], 5% β-mercaptoethanol [m/v], and 0.001% bromophenol blue [m/v]). Extracellular proteins were separated by SDS-PAGE using the discontinuous buffer system of Laemmli [36]. Electrophoresis was performed on SDS 15% (m/v) polyacrylamide gels and proteins were visualized by silver staining.
For immunostaining, extracellular proteins were separated on SDS 15% (m/v) polyacrylamide gels and electroblotted to Immun-Blot polyvinylidene difluoride membranes (Bio-Rad, USA) using a Mini Trans-Blot electrophoretic transfer cell (Bio-Rad). Membranes were blocked with TBS containing 2% (m/v) bovine serum albumin (BSA) and then incubated with antibodies raised against NopA, NopB, NopC, NopP, and NopX [21] or the cya epitope (Cell Signaling Technologies, USA) diluted 1:1000 in the same solution. Anti-rabbit immunoglobulin AP-conjugated secondary antibody was used and reaction results were visualized using NBT-BCIP.
In vivo monitoring of the activation of the tts box upstream nopC during the initial stages of the symbiosis with soybean The Flux vector (Table 1) was digested with BamHI and EcoRI and the resulting 5.8 kb DNA fragment carrying the luxCDABE genes was cloned into the broad host-range plasmid pMP92 to obtain plasmid pMUS1199. Two primers ( Table 2) were designed to amplify the tts box upstream nopC from pMUS986 and add EcoRI and HindIII restriction sites. The resulting 0.2 kb PCR fragment was digested with the restriction enzymes indicated previously and cloned into plasmid pMUS1199 previously digested with EcoRI and HindIII, obtaining plasmid pMUS1207. The luxCDABE operon encodes the luciferase enzyme and therefore instant in vivo promoter activity can be monitored by measuring bioluminescence. Then, two different experiments were carried out. In the first one, nine pre-germinated G. max cv. Williams 82 seeds were aseptically transferred to a recipient containing vermiculite and 150 ml of a Fåhraeus solution 1M pH 6.8 and were grown in a controlled environment chamber with a 16 h day/8 h night cycle and a relative humidity of 70%. Growth temperatures were set to 26°C during the day period and to 18°C during the night. The system was inoculated at the time of transferring the pre-germinated seeds with a bacterial culture of about 10 8 c.f.u. ml -1 of S. fredii HH103 Rif R carrying plasmid pMUS1207 or plasmid pMUS1199. In the second one, pre-germinated seeds were placed on sterilized pouches containing the same nitrogen-free Fåhraeus solution, inoculated with 1 ml of the previously described bacterial strains and the formation of nodules and bioluminescence were monitored at 14,16,19,21,23,26, and 29 days post-inoculation (d.p.i.). In both cases experiments were performed twice with five replicates and bioluminescence was quantified using a photon-counting camera IVIS Lumina II (Caliper Life Science, USA). Images were analyzed with the Living Image 4.0 software (Caliper Life Science). A rootflavonoid diffusion assay was used to validate the T3SS-dependent bioluminescence. Thus, pre-germinated soybean seeds were placed in squared Petri dishes and roots were covered with a volume of 20 ml of TY (1.2% agar) mixed with 10 ml bacterial cultures of strains HH103 Rif R (pMUS1207), HH103 Rif R nodD1::lacZ-Gm R (pMUS1207), HH103 Rif R ttsI::Ω (pMUS1207), and HH103 Rif R (pMUS1199). Bioluminescence was quantified 72 h after inoculation.

Adenylate cyclase (cya) assay
To examine whether S. fredii HH103 T3SS translocates the NopC-Cya fusion protein into soybean root cells, the protocol described by [24] was used with some modifications. Eighteen pregerminated soybean seeds were aseptically transferred to a recipient containing vermiculite and 150 ml of a Fåhraeus solution 1M pH 6.8 and grown in a controlled environment chamber with a 16 h day/8 h night cycle and a relative humidity of 70%. Growth temperatures were set to 26°C during the day period and to 18°C during the night. The system was inoculated at the time of transferring the pre-germinated seeds with bacterial cultures of about 10 8 c.f.u.ml -1 of the HH103 Rif R or the ttsIΩ mutant strains, both containing a chromosomal integration of the nopC-cya fusion. Cyclic AMP (cAMP) accumulation was measured in nodules harvested 18 d. p.i. Nodules were frozen in liquid nitrogen, ground to a fine powder and resuspended in a 0.1 M hydrochloric acid solution. The suspension was centrifuged and the supernatant was used for cAMP measurement using the cyclic AMP (direct) EIA kit (Cayman Chemical Company, USA) according to the manufacturer's instructions. Each sample was diluted for quantification to measure cAMP concentration in the detection range of the assay. The HH103 Rif R strain without the nopC-cya fusion was used as a control for quantification.

Plant assays
Nodulation assays on G. max cv. Williams 82 or V. unguiculata were performed as described by [37]. Each Leonard jar contained two soybean or two V. unguiculata plants. Each

Results
NopC is a Rhizobium-specific T3SS secreted protein The nopC and nopA genes are located in the symbiotic plasmid of HH103 (pSfHH103d). The analysis of the genome sequence of HH103 revealed the presence of a conserved tts box situated 43 pb upstream nopC (297 pb). The nopA gene (216 pb) was located 80 pb downstream nopC (Fig 1a). Therefore, the organization and position of nopC within the tts region was similar to those described in S. fredii NG234 [21].
The genome of HH103 contains two genes, nopL and nopP, which code for two type 3-secreted proteins exclusive of rhizobia. NopC also seems to be Rhizobium-specific since no homologues have been detected in animal or plant pathogens. Neighbor joining phylogenies of the NopC tree of several rhizobial strains showed two clearly separated branches. At the bottom of the branch we found the three Bradyrhizobium species in which the NopC protein has been identified and annotated. On the other hand, the top branch is split between Mesorhizobium and Sinorhizobium (Ensifer) species. Interestingly, the neighbor joining of the NopC sequence of the S. fredii strains showed identical gene sequence (Fig 1b). This result was confirmed by aligning the HH103 (AAY33494), USDA257 (not annotated in the databases) and NGR234 (YP_052972) NopC sequences and verifying 100% sequence identity (data not shown).
Previous works have shown that the HH103 nopA expression is regulated in a flavonoid-, NodD1-and TtsI-dependent manner [17]. Therefore, it was presumable that the expression of the nopC gene was regulated in the same manner because both genes are preceded by the same tts box. This was confirmed by quantitative real-time PCR assays in which the expression of nopC in the parental strain HH103 Rif R and in a nodD1 and a ttsI mutant backgrounds, in the absence or presence of the flavonoid genistein, was studied. As shown in Fig 2a, induction with genistein increased the transcription of nopC about 45-fold in the parental strain. By contrast, transcription of this gene was not detected either in the nodD1 or the ttsI mutants, confirming that the expression of nopC depends on flavonoids and on the transcriptional regulators NodD1 and TtsI.
Previous reports detected NopC in the supernatants of NGR234 and HH103 cultures grown in the presence of inducer flavonoids [20,21]. To confirm that secretion of NopC depends, together with flavonoids, on NodD and TtsI, proteins from strains HH103 Rif R , HH103 Rif R nodD1::lacZ-Gm R , and HH103 Rif R ttsI::Ω culture supernatants, in the presence or absence of inducing flavonoids, were extracted and separated by SDS-PAGE. The band of about 11 kDa corresponding to NopC was only observed in the supernatants of HH103 Rif R cultures induced with genistein and not in the nodD1 and ttsI mutants (data not shown). As expected, westernblots assays showed that the antibody raised against NopC specifically detected a protein of about 11 kDa in these supernatants. No signal was detected in the lanes corresponding to the nodD1 or ttsI mutant strains with or without genistein (Fig 2b).
In frame deletion of nopC did not alter secretion of other nodulation outer proteins or the expression of nopA As previously mentioned, nopA was located downstream nopC in the HH103 genome. Previous reports indicated that the inactivation of the NGR234 nopA gene completely abolished secretion of Nops and confirmed, together with the analysis of the interaction of NopA with NopB and NopX, that NopA was the major component of the T3SS pilus [21]. In HH103, the mutation of nopA also abolished secretion of Nops to the extracellular medium (data not shown). As the nopC and nopA genes are transcribed from the same tts box, the construction of an in frame mutation of the nopC gene was necessary to study the effect of the inactivation of nopC on protein secretion because a polar mutation would block secretion due to its effect on nopA transcription. Therefore, to confirm that the expression of nopA was not affected by the mutation, transcription of nopA in the ΔnopC mutant background was quantified by quantitative real-time PCR in strains HH103 Rif R , HH103 Rif R ΔnopC, and HH103 Rif R ttsI::Ω in the absence or presence of the flavonoid genistein. As shown in S1 Fig, induction with genistein increased the transcription of nopA about 34-fold and 32-fold in the parental and the nopC mutant strain, respectively. This increase in gene expression was not detected in the ttsI mutant induced with flavonoids. As a control, expression of nodA was quantified in all the treatments previously analyzed. Results showed increased expression in the presence of genistein in all the strains assayed (S1 Fig). The analysis of protein supernatants from strain HH103 Rif R ΔnopC induced with genistein showed the presence of all the Nops secreted by HH103 with the exception of the band corresponding to NopC, indicating that type 3 secretion was not blocked in this mutant (Fig 3). This result was confirmed with western-blot analyses using antibodies raised against NopC, which showed the presence of a band of the deduced size of NopC (~11 kDa) in the supernatants from the parental strain HH103 Rif R and the complemented strain HH103 Rif R ΔnopC (pMUS986) in the presence of genistein and not in the nopC mutant (Fig 4). To confirm that the absence of NopC did not affect secretion of the rest of Nops, western-blots assays with antibodies raised against NopA (~6 kDa), NopB (~20 kDa), NopP (~32 kDa), and NopX (~60 kDa) were performed. Thus, signals corresponding to the expected sizes of the Nops analyzed were detected in all the treatments when induced with genistein (Fig 4). All these results showed that the in frame deletion of nopC did not block secretion of the rest of Nops.

The S. fredii HH103 NopC is delivered to the cytoplasm of soybean root cells
To determine whether NopC was delivered to the interior of the soybean root cells we analyzed its T3SS-dependent translocation into plant cells using the adenylate cyclase assay. The Bordetella pertussis calmodulin-dependent adenylate cyclase (Cya) toxin is activated within eukaryotic cells and hence increase cAMP levels. However, no activation of this protein is detected in prokaryotic cells so it can be used to identify bacterial effector proteins [38]. The Cya reporter was fused to the carboxy terminus of NopC and integrated into the symbiotic plasmid of HH103 by simple homologous recombination. This method allowed the transcription of the nopC-cya fusion by means of the native tts box located upstream nopC.
Previous works [24,39] demonstrated translocation of effector-Cya fusions into nodules of Macroptilium atropurpureum and V. unguiculata. In the case of soybean, no reports about translocation of effectors into soybean nodule cells have been published so far. In addition, some authors [40] showed that Nops can be detected in soybean infection threads but not in mature nodules. The Cya assay to determine protein translocation to the interior of root cells is time-consuming and expensive. Therefore, it was necessary to optimize the assay using only those parts of the root in which the T3SS was expressed. Thus, a previous in vivo monitoring of the activation of the tts box upstream nopC during symbiosis with soybean was performed using a bioluminescent reporter system. The use of bioluminescence would indicate instant expression of the tts genes. First, a root-flavonoid diffusion assay was used to validate the tts box-dependent bioluminescence. As shown in S2 Fig, a high bioluminescent signal was mainly observed in plants inoculated with the parental strain carrying plasmid pMUS1207. Then, a time-course bioluminescence monitoring assay carried out in pouches to monitor gene expression in intact plants showed that the bioluminescence was first detected in the upper region of the main root and then concentrated in the nodule, reaching a peak of expression at 21-23 days after inoculation and then declined (S3 Fig). Finally, some but not all nodules from plants grown in vermiculite, harvested at 18 d.p.i. and inoculated with the parental strain carrying the plasmid with the tts box-luxCDABE fusion, showed high bioluminescence, confirming that HH103 T3SS-regulated genes are expressed in this symbiotic organ at that time-point (Fig 5A). As expected, no bioluminescence was detected in plants inoculated with the strain carrying the empty vector (data not shown). This result allowed us the selection of young soybean nodules (18 d.p.i. nodules) as the optimum tissue for cAMP quantification. Further Cya assays showed very low levels of cAMP in the nodules formed by the parental strain and the ttsI mutant strain expressing the nopC-cya fusion. By contrast, cAMP accumulation was significantly higher in  nodules from plants inoculated with the parental strain expressing nopC-cya (Fig 5B). These results indicated that NopC was translocated into the host cells via the T3SS of S. fredii HH103.

NopC is beneficial for the symbiosis with G. max and V. unguiculata
To elucidate the role of NopC in the symbiosis established between HH103 and soybean, the symbiotic properties of the HH103 Rif R , HH103 Rif R ΔnopC, the HH103 Rif R ΔnopC mutant complemented in cis and the HH103 Rif R nopA::lacZ-Gm R strains were determined in plant infection tests ( Table 3). The number and fresh mass of the nodules formed and the plant-top dry mass of the soybean plants were significantly lower (α = 5%) in plants inoculated with the nopA mutant strain (around 40% less in all symbiotic parameters) when compared with those inoculated with the parental strain. Interestingly, in the case of the nopC mutant no differences in plant-top dry masses were observed, but the fresh mass and the number of nodules formed were statistically lower (α = 5% and α = 10%, respectively) in plants inoculated with this mutant in comparison to plants inoculated with the parental strain. Complementation in cis of the nopC mutant restored all symbiotic phenotypes. These results indicated that NopC exerted a positive effect in the symbiosis between HH103 and soybean cv. Williams 82.
Results obtained in nodulation assays with V. unguiculata showed also a positive effect of the presence of NopC in the formation of nodules, since a lower number and fresh mass of nodules were obtained in plants inoculated with the nopC mutant strain with respect to plants inoculated with the parental strain (differences were statistically significant in the case of the number of nodules, α = 5%) (Table 4). Interestingly, the opposite phenotype was observed in S. fredii HH103 mutants were individually compared with the parental strain HH103 Rif R by using the Mann-Whitney non-parametrical test. Numbers on the same column followed by an asterisk (*) are significantly different at the level α = 5%. Numbers on the same column followed by two asterisks (**) are significantly different at the level α = 10%. S. fredii HH103 mutants were individually compared with the parental strain HH103 Rif R by using the Mann-Whitney non-parametrical test. Numbers on the same column followed by an asterisk (*) are significantly different at the level α = 5%.
plants inoculated with the nopA mutant, i.e. higher number and fresh mass of nodules (α = 5%) in comparison to plants inoculated with the parental strain ( Table 4). Complementation of the nopC mutant restored all symbiotic phenotypes.

Discussion
In silico analysis of the recently sequenced HH103 genome showed that the gene organization of the tts region in this strain was similar to that of S. fredii NGR234 [41]. Previous reports [21] have shown that the inactivation of the NGR234 nopA or nopB genes completely abolishes secretion of Nops and the effect of their mutation on nodulation was similar to that observed when mutating any structural component of the T3SS, suggesting that NopA and NopB were essential components of the T3SS machinery. Further experiments confirmed that both proteins were the main components of the T3SS extracellular appendages or T3SS-pilus [42]. In addition, bioinformatics analysis of the NopA amino acid sequence showed that this protein possesses the requisite secondary structural characteristics typical of T3SS pili (very high αhelical content, especially at the C-terminus region). In spite of the absence of any scientific report, NopC has also been considered in the literature as a component of the T3SS extracellular appendages. However, the analysis of the NopC amino acid sequence has revealed that this protein does not possess the secondary structures found in NopA [21]. The nopC gene has been found and annotated only in a very small group of rizobial genomes and similar proteins have not been detected in other pathogenic bacteria (Fig 1). Therefore, NopC can be considered a Rhizobium-specific T3SS secreted protein.
Results shown in this work indicated that the biosynthesis of the HH103 NopC is regulated by flavonoids and by the transcriptional regulators NodD1 and TtsI, because both gene expression and protein secretion were detected in HH103 Rif R bacterial cultures supplemented with genistein but not in nodD1 and ttsI mutant backgrounds (Fig 2). This regulation cascade for components or proteins secreted through the T3SS is shared by all strains possessing a T3SS, with the exception of B. elkanii USDA61 [43]. Interestingly, the inactivation of the nopC gene did not block secretion of the rest of Nops, indicating that NopC was not an essential component of the T3SS machinery (Figs 3 and 4). The in frame mutation of nopC did not affect transcription of nopA, discarding a possible polar effect of the mutation on the transcription of downstream genes (S1 Fig).
Transcription of the T3SS-genes are down-regulated in soybean bacteroids [44] and previous reports detected S. fredii USDA257 NopX (previously called NolX) in infection threads but not in mature nodules of soybean [40], suggesting that the T3SS was induced early in the symbiosis with soybean but repressed at later stages. However, results shown in this work of the in vivo monitoring of the activity of the tts box upstream nopC fused to lux genes showed that the HH103 T3SS was expressed in some but not all young soybean nodules (Fig 5A). The fact that bioluminescence was not detected in all nodules could have some explanations. First, plasmid pMUS1207 (derived from the high-copy plasmid pMP92) could be lost by part of the rhizobial population. Both bioluminescence and nitrogen fixation require high energy consumption and the promoter cloned shows very high activity with inducer flavonoids. Therefore, the presence of the plasmid and the production of bioluminescence since the early stages of the infection process could have a negative impact on bacterial fitness. In addition, luciferase needs oxygen and ATP for its activity and the atmosphere in the symbiosome, necessary for nitrogen fixation, is deficient in oxygen. Therefore, bioluminescence would not be detected in mature nodules, making this method only suitable for instant detection of gene expression in the early stages of the symbiosis, time period that coincides with the moment in which the T3SS shows its highest activity. Nevertheless, the objective of the use of this reporter was the selection of a particular root region/tissue with the best characteristics to further measure the production of cAMP in the Cya assay.
Translocation of the effectors NopE1 and NopE2 of B. japonicum USDA110 and NopP of S. fredii USDA257 within eukaryotic host cells has been determined in the symbioses with M. atropurpureum and V.unguiculata, respectively [24,39]. In both cases, effectors were fused to a cya reporter gene to test translocation into the plant cell cytoplasm. This specific bacterial enzyme catalyzes the cAMP production in the presence of ATP and host calmodulin-like proteins. Therefore, an increase in the cAMP levels indicates effector translocation into eukaryotic cells. Translocation of NopC to the interior of soybean cells was confirmed using this reporter assay (Fig 5B) and the quantified levels of the cAMP induced by the NopC-Cya fusion protein in soybean root cells were similar to those obtained in previous reports [24,39].
Rhizobial proteins secreted through the T3SS are involved in host-range determination and symbiotic efficiency [13]. The positive, negative or neutral effects on symbiosis of each effector are determined by the host plant. For instance, in the symbioses between S. fredii HH103 and its host legumes, the inactivation of the T3SS is beneficial for the symbioses with several cultivars of G. max (including Williams 82) and Glycyrrhiza uralensis and detrimental on Erythrina variegata [17]. Results shown in this work indicated that NopC had a positive role in the symbiosis between HH103 and both soybean cv. Williams 82 and V. unguiculata, since the mutation of this gene caused a decrease in the number and the fresh mass of the nodules formed in both plants (Tables 3 and 4). This positive effect on soybean symbiosis was also observed in plants inoculated with the nopA mutant strain (Table 3). Interestingly, the symbiotic effect in V. unguiculata of the mutation of nopA was opposite to that observed with the nopC mutant ( Table 4), indicating that both Nops are playing different roles in the symbiotic process. Besides, results suggest that NopC is contributing, together with still unknown effectors, to an effective symbiosis with both plants.
As previously mentioned, NopC has no homologues in pathogenic bacteria and it does not possess any domain or conserved feature that could give clues about its role in symbiosis. It could be possible that NopC functions as a chaperone (T3SC) to facilitate the assembly of the secretion apparatus or secretion of effectors to the interior of the host cell. T3SCs are typically small and acidic cytoplasmic proteins that remain within the bacterial cell and some of them are encoded by a gene within an operon that carries genes encoding components of the secretion apparatus. In addition, T3SCs that bind effectors exhibit a contiguous conserved set of structural folds (α-β-β-β-α-β-β-α) and their absence results in lack of secretion of their corresponding effectors [45]. However, NopC was translocated to the interior of soybean cells, secretion of other Nops was not abolished when nopC was inactivated and the analysis of the NopC secondary structure showed low α-helical content at its carboxy terminus and the conserved structure of T3SCs was not detected. Further studies are necessary to determine the specific function of this Rhizobium-specific effector protein within the plant cell. qRT-PCR analysis of the expression of nopA (A) and nodA (B) in the parental strain HH103 Rif R and the mutant strains HH103 Rif R ttsI::Ω and HH103 Rif R ΔnopC in the absence (-) or presence (+) of the inducer flavonoid genistein (3.7 μM). Final expression was calculated relative to the expression in the HH103 Rif R strain in the absence of flavonoids. Expression data shown are the mean (± standard deviation of the mean) for three biological replicates performed at least in triplicates. Each expression value was individually compared with the HH103 Rif R strain in the absence of flavonoids using the Mann-Whitney non-parametrical test. Asterisks indicate that numbers are significantly different at the level α = 5% (p< 0.05). (TIF) S2 Fig. Validation of the tts box-luxCDABE fusions by in vivo activation with flavonoids exuded by soybean roots. S. fredii strains carrying plasmids pMUS1199 (= pMP92-luxCDABE) or pMUS1207 (= pMP92-tts box::luxCDABE) were assayed in squared Petri dishes with pre-germinated soybean seeds. A. HH103 Rif R (pMUS1199). B. HH103 Rif R (pMUS1207). C. HH103 Rif R nodD1::lacZ-Gm R (pMUS1207). D. HH103 Rif R ttsI::Ω (pMUS1207). Bioluminescence was measured 72 hours after inoculation. (TIF)

S3 Fig.
In vivo monitoring of the activation of the tts box upstream nopC using soybean plants grown in pouches. Bioluminescence was measured in soybean plants inoculated with the HH103 Rif R strain carrying plasmid pMUS1207 (plasmid pMP92 containing the tts box fused to luxCDABE). Bioluminescence in nodules is indicated with a white arrow. The grey arrow shows a nodule without bioluminescence. Bioluminescence was measured at 14,16,19,21,23,26, and 28 days after inoculation. (TIF)