Sinorhizobium fredii HH103 RirA Is Required for Oxidative Stress Resistance and Efficient Symbiosis with Soybean

Members of Rhizobiaceae contain a homologue of the iron-responsive regulatory protein RirA. In different bacteria, RirA acts as a repressor of iron uptake systems under iron-replete conditions and contributes to ameliorate cell damage during oxidative stress. In Rhizobium leguminosarum and Sinorhizobium meliloti, mutations in rirA do not impair symbiotic nitrogen fixation. In this study, a rirA mutant of broad host range S. fredii HH103 has been constructed (SVQ780) and its free-living and symbiotic phenotypes evaluated. No production of siderophores could be detected in either the wild-type or SVQ780. The rirA mutant exhibited a growth advantage under iron-deficient conditions and hypersensitivity to hydrogen peroxide in iron-rich medium. Transcription of rirA in HH103 is subject to autoregulation and inactivation of the gene upregulates fbpA, a gene putatively involved in iron transport. The S. fredii rirA mutant was able to nodulate soybean plants, but symbiotic nitrogen fixation was impaired. Nodules induced by the mutant were poorly infected compared to those induced by the wild-type. Genetic complementation reversed the mutant’s hypersensitivity to H2O2, expression of fbpA, and symbiotic deficiency in soybean plants. This is the first report that demonstrates a role for RirA in the Rhizobium-legume symbiosis.


Introduction
As for almost all living organisms, iron is an essential nutrient for bacteria but in excess it becomes toxic. This active redox metal forms part of the prosthetic group (heme or iron-sulfur clusters) of many proteins that participate in important metabolic processes such as respiration, central metabolism, can react with iron present in the prosthetic group of metalloproteins, resulting in enzyme inhibition and the release of free iron, which could then promote the Fenton reaction [3]. Therefore, bacteria require mechanisms for iron homeostasis that allow them to maintain sufficient levels of the metal to support vital metabolic processes and at the same time prevent toxicity.
Iron homeostasis is achieved by controlling the expression of proteins involved in iron uptake, storage, consumption, and export. When intracellular levels of iron are insufficient, many bacteria produce siderophores, i.e., water soluble Fe 3+ chelating compounds, and express several high-affinity uptake systems to acquire iron from the environment [4]. Under iron sufficient conditions, expression of iron uptake systems is repressed and iron is incorporated into metalloproteins. Excessive iron is accumulated in a non-reactive form in iron storage proteins and/or expelled out of the cell by iron efflux transporters to mitigate cytotoxic effects [5][6][7]. All of these processes are coordinately controlled through the activity of iron response regulatory proteins.
In many bacteria, including Escherichia coli, the major iron response regulator is Fur (Ferric Uptake Regulator) [1,[8][9][10]. Fur is a transcriptional regulator that senses and responds to iron by directly binding this metal [10]. In the presence of certain levels of iron, the Fe 2+ -Fur complex binds to DNA sequences called Fur boxes and represses the transcription of iron uptake systems, preventing iron overload toxicity [1]. Remarkably, iron-responsive regulation differs from the Fur paradigm in rhizobia, i.e., soil-dwelling bacteria that are known for their ability to establish a symbiotic relationship into nitrogen-fixing nodules on the roots of leguminous plants [11]. Symbiotic nitrogen fixation is highly dependent on iron because different proteins involved in the process require the metal to be functional. This is the case of the nitrogenase complex responsible for the nitrogen fixation reaction, or the abundant nodule protein leghemoglobin that buffers oxygen levels to avoid inactivation of nitrogenase activity. Although a Fur-like protein is present in most rhizobia, functional studies performed on the fur genes of Sinorhizobium (Ensifer) meliloti and Rhizobium leguminosarum revealed that the encoded proteins are responsive to manganese and not to iron, which is the reason why they were renamed as Mur [12][13][14]. Instead, iron-responsive regulation in rhizobia is mediated by the Irr and RirA regulatory proteins that sense iron availability indirectly [15][16][17].
The Irr (Iron Responsive Regulator) protein belongs to the Fur superfamily. Described initially in the soybean symbiont Bradyrhizobium japonicum [18], Irr occurs in all rhizobia and is the only iron responsive regulator in Bradyrhizobiaceae. Under iron-deficient conditions, Irr usually represses genes encoding proteins that function under iron-sufficient conditions (i.e., proteins containing iron, heme and iron-sulfur cluster biosynthetic proteins). Upon sensing iron indirectly through the status of heme biosynthesis, Irr is either degraded (as occurs in B. japonicum) or becomes inactive (as is the case of R. leguminosarum), which allows transcription of Irr-repressed genes [17].
In rhizobia distinct from Bradyrhizobiaceae, iron homeostasis is achieved with the participation of a second regulatory protein, RirA (Rhizobial Iron Regulator A). This regulator, which is found exclusively in alphaproteobacteria, is present in Rhizobium, Sinorhizobium (Ensifer), and Mesorhizobium spp., but also in the phytopathogen Agrobacterium tumefaciens and in the animal pathogens Brucella and Bartonella [14,17]. RirA is an iron-sulfur protein that belongs to the Rrf2 family of putative transcription regulators. Under iron-replete conditions, this regulator acts as a repressor of many iron-responsive genes by binding to Iron-Responsive Operator sequences (5'-TGA-(N9)-TCA-3') known as IRO boxes [14,19]. A recent study has demonstrated that R. leguminosarum RirA contains a [4Fe-4S] cluster and that the holoprotein ([4Fe-4S]RirA) binds to an IRO box sequence [20]. Interestingly, the [4Fe-4S] cluster is sensitive to iron levels but also to O 2 , with low iron concentrations and aerobic conditions promoting the conversion of [4Fe-4S] cluster to [2Fe-2S] and subsequent cluster loss, which relieves gene repression [20]. Therefore, RirA seems to act as an iron sensor via iron-sulfur cluster availability.
In rhizobia, RirA has been functionally characterized only in R. leguminosarum and S. meliloti. In R. leguminosarum, RirA regulates several iron-responsive genes such as genes for the synthesis and uptake of the siderophore vicibactin (vbs), genes involved in heme uptake (hmu and tonB), genes for the synthesis of iron-sulfur clusters (suf ), the irrA gene as well as its own expression [21,22]. In S. meliloti, the RirA regulon comprises several iron-responsive genes, which include genes involved in iron uptake, energy metabolism, exopolysaccharide synthesis or iron storage, but also genes that are not iron-responsive (e.g., suf ) [23][24][25]. Several free-living phenotypes have been associated with rirA mutants. Thus, inactivation of rirA leads to: (i) constitutive expression of siderophores in R. leguminosarum, S. meliloti and in A. tumefaciens, (ii) species-specific growth phenotypes which can be iron-dependent or -independent, (iii) higher sensitivity to oxidants in S. meliloti and in A. tumefaciens, and (iv) decreased ability to cope with cobalt and nickel toxicity in A. tumefaciens [21,23,24,[26][27][28]. Hypersensitivity to peroxides and metals in rirA mutants was explained by the derepression of iron uptake systems and the consequent iron overload in the cell, which promotes the generation of ROS via Fenton chemistry.
The role of RirA in the establishment of plant-bacteria interactions has also been investigated. Loss of function of RirA in R. leguminosarum or S. meliloti does not affect symbiotic nitrogen fixation [21,23,24]. In A. tumefaciens, RirA was found to be important for tumor formation on tobacco leaf pieces [26] but not on potato disks [27]. The defect in tumorigenesis on tobacco leaves was associated with increased sensitivity to oxidants and reduced virulence gene expression exhibited by the rirA mutant.
S. fredii HH103 is a fast-growing rhizobial strain, which was isolated from a Chinese soil from the Hubei Province. HH103 is able to nodulate American and Asiatic varieties of soybean, as well as many other different legumes that form determinate or indeterminate nodules [29,30]. Iron homeostasis in broad host range rhizobia has not been investigated yet. The genome sequence of S. fredii HH103 reveals the presence of putative iron-responsive regulatory genes, including a rirA homologue [31,32]. With the aim of gaining insights into iron-mediated responses in HH103, in this study, we have investigated the role of RirA during free-living growth and in symbiosis with soybean. Our data demonstrate that, in S. fredii HH103, RirA is important for oxidative stress resistance and for effective symbiosis with soybean plants.

Results
2.1. The Lack of RirA in S. fredii HH103 Confers a Siderophore-Independent Growth Advantage under Low Iron Conditions S. fredii HH103 encodes homologues of Fur (SFHH103_03131), Irr (SFHH103_03814), and RirA (SFHH103_00388). The chromosomal SFHH103_00388 gene encodes a 154 amino acid-long protein annotated as an iron-responsive regulator of the Rrf2 family of transcriptional regulators. The predicted protein, which is identical to one encoded by S. fredii NGR234, exhibits high sequence similarity to the RirA proteins of S. meliloti, A. tumefaciens and R. leguminosarum bv. viciae (93.5%, 83.4%, and 79.4% identity, respectively), and conserves the three Cys residues predicted to ligate iron-sulfur clusters in Rrf2 family regulators [20] ( Figure S1). The genetic context of the rirA gene is similar in HH103 and in S. meliloti. In the two rhizobia, the gene is flanked downstream by a cluster of dpp genes involved in the uptake of dipeptides and a heme precursor [33], and upstream by the fbpA gene that potentially codes for a periplasmic ferric binding protein, which is regulated by RirA in S. meliloti [23].
In order to investigate the role of the rirA homologue in HH103, mutant SVQ780 was constructed by inserting a lacZ∆p-Gm R cassette into the coding sequence of this gene (see Material and Methods). A battery of free-living phenotypes that have been associated with rirA mutants were evaluated such as siderophore production, growth in media with different iron content, resistance to H 2 O 2 , and expression of iron-responsive genes.
Under our experimental conditions, Chrome Azurol S (CAS) assays performed using cultures of the wild-type strain SVQ269 (HH103-Rif R ) grown in iron-deficient media failed to show any siderophore activity. In fact, CAS assays performed on forty S. fredii strains isolated from different locations in China only showed activity in two of them (strains S2 and S5; Figure S2), suggesting that siderophore production is a rare attribute in Chinese S. fredii strains. RirA loss-of-function leads to overexpression of siderophore biosynthesis and transport genes in R. leguminosarum, S. meliloti and A. tumefaciens [22][23][24]26]. However, no CAS activity was detected upon inactivation of the rirA homologue in HH103 ( Figure S2c), indicating that this strain does not acquire iron via CAS-detected siderophore production.
Depending on the bacterial species, inactivation of rirA results in different growth phenotypes ranging from iron-dependent or -independent reduced growth to no effect at all [21,23,27]. In our study, we evaluated the growth rates of S. fredii wild-type and rirA mutant strains in minimal medium (MM) agar plates supplemented with different concentrations of FeCl 3 as the added iron source, as well as in un-supplemented MM with ambient iron levels. As shown in Figure 1, the rirA mutant exhibited similar growth capacities in un-supplemented MM and in MM supplemented with 0.22, 2.2 and 22 µM FeCl 3 . The wild-type and the rirA mutant exhibited similar growth rates on tryptone-yeast extract (TY) medium. Likewise, no differences were detected between the two strains on MM plates supplemented with 22 µM FeCl 3 , which is the optimum concentration of iron for the wild-type strain under the conditions tested. Interestingly, on MM containing lower concentrations of iron (i.e., un-supplemented and supplemented with 0.22 and 2.2 µM FeCl 3 ) the rirA mutant showed a slight advantage in growth with respect to SVQ269 and the complemented strain SVQ780C. The low iron availability-dependent growth advantage associated with the rirA mutant might be caused by derepression of as yet uncharacterized high-affinity iron uptake systems. In MM with the highest iron content (220 µM FeCl 3 ), all three strains exhibited similar growth reductions compared with MM containing 22 µM FeCl 3 , suggesting probable iron toxicity.
of iron (i.e., un-supplemented and supplemented with 0.22 and 2.2 µM FeCl3) the rirA mutant showed a slight advantage in growth with respect to SVQ269 and the complemented strain SVQ780C. The low iron availability-dependent growth advantage associated with the rirA mutant might be caused by derepression of as yet uncharacterized high-affinity iron uptake systems. In MM with the highest iron content (220 µM FeCl3), all three strains exhibited similar growth reductions compared with MM containing 22 µM FeCl3, suggesting probable iron toxicity.

RirA Plays A Role in Oxidative Stress Resistance in HH103
It has been shown that rirA mutants of S. meliloti and A. tumefaciens are more sensitive to oxidative stress than their corresponding wild-type strains [23,26,27]. Therefore, we decided to test whether inactivation of rirA in HH103 alters the bacterial resistance to hydrogen peroxide.
During these experiments, we noticed that HH103 was much more sensitive to H 2 O 2 than S. meliloti or A. tumefaciens. Thus, approximately only 1 out of 10 7 cells survived after exposing SVQ269 cultures to 10 mM H 2 O 2 for only 1 h. When SVQ269 cultures were exposed to a 100-fold lower concentration of H 2 O 2 (100 µM), the survival ratio of the wild-type strain was slightly reduced to 71%. Under the same conditions, the viability of rirA mutant cells was extremely low (only 20 out of every 10 7 cells could survive after treatment with 100 µM H 2 O 2 ). The higher sensitivity of the rirA mutant to H 2 O 2 compared to the wild-type strain was confirmed by growing serial dilutions of cells on TY plates containing different concentrations of the oxidant (Figure 2).

RirA Plays A Role in Oxidative Stress Resistance in HH103
It has been shown that rirA mutants of S. meliloti and A. tumefaciens are more sensitive to oxidative stress than their corresponding wild-type strains [23,26,27]. Therefore, we decided to test whether inactivation of rirA in HH103 alters the bacterial resistance to hydrogen peroxide.
During these experiments, we noticed that HH103 was much more sensitive to H2O2 than S. meliloti or A. tumefaciens. Thus, approximately only 1 out of 10 7 cells survived after exposing SVQ269 cultures to 10 mM H2O2 for only 1 h. When SVQ269 cultures were exposed to a 100-fold lower concentration of H2O2 (100 µM), the survival ratio of the wild-type strain was slightly reduced to 71%. Under the same conditions, the viability of rirA mutant cells was extremely low (only 20 out of every 10 7 cells could survive after treatment with 100 µM H2O2). The higher sensitivity of the rirA mutant to H2O2 compared to the wild-type strain was confirmed by growing serial dilutions of cells on TY plates containing different concentrations of the oxidant ( Figure 2). No differences in growth were observed for the wild-type and mutant strains on TY plates in the absence of H2O2. While viability of SVQ269 remained unchanged in the presence of 5 µM H2O2, survival of the rirA mutant was already severely affected. The presence of higher concentrations of H2O2 (7.5 and 10 µM) decreased the viability of the wild-type and especially that exhibited by the rirA mutant. The hypersensitive phenotype of SVQ780 could be reversed by introducing a functional No differences in growth were observed for the wild-type and mutant strains on TY plates in the absence of H 2 O 2 . While viability of SVQ269 remained unchanged in the presence of 5 µM H 2 O 2 , survival of the rirA mutant was already severely affected. The presence of higher concentrations of H 2 O 2 (7.5 and 10 µM) decreased the viability of the wild-type and especially that exhibited by the rirA mutant. The hypersensitive phenotype of SVQ780 could be reversed by introducing a functional rirA gene in cis (strain SVQ780C). These data indicate that inactivation of rirA in S. fredii HH103 results in reduced resistance to oxidative damage.

Role of RirA in Iron-Responsive Gene Expression
Inactivation of rirA in R. leguminosarum, S. meliloti, and A. tumefaciens causes upregulation of several genes whose transcription is normally reduced/repressed under iron-sufficient conditions. These include genes involved in the synthesis and uptake of siderophores, heme uptake, inorganic ferric ion transport, biosynthesis of Fe-S clusters, and the regulatory irr and rirA genes [21][22][23][24]26,27].
The expression levels of potentially iron-responsive genes were determined by Reverse Transcription-quantitative Polymerase Chain Reaction (RT-qPCR) in the wild-type strain SVQ269 after growth in iron-deficient or iron-replete media. Based on the bacterial growth shown in Figure   To examine the effects of RirA on iron homeostasis, the transcript levels of fbpA, hmuS, sufS and irr were also determined by RT-qPCR in rirA mutant cells after growth in iron-sufficient (22 µM FeCl3) media (Figure 3b). Under these conditions, no differential expression was detected for hmuS between the mutant and the wild-type (1.5 ± 0.3 fold increase in rirA mutant versus wild-type strain). In the case of sufB and irr, only slight decreases in expression were observed in the mutant compared To examine the effects of RirA on iron homeostasis, the transcript levels of fbpA, hmuS, sufS and irr were also determined by RT-qPCR in rirA mutant cells after growth in iron-sufficient (22 µM FeCl 3 ) media (Figure 3b). Under these conditions, no differential expression was detected for hmuS between the mutant and the wild-type (1.5 ± 0.3 fold increase in rirA mutant versus wild-type strain). In the case of sufB and irr, only slight decreases in expression were observed in the mutant compared to the wild-type (2 ± 1.3 and 4.4 ± 3.3 fold decrease in rirA mutant versus wild-type strain for sufB and irr, respectively). In contrast, inactivation of rirA caused a clear upregulation of fbpA (95.4 ± 39.6 fold increase in rirA mutant versus wild-type strain). No differential expression for fbpA was detected between the wild-type and the complemented strain. This suggests that RirA represses transcription of fbpA, either directly or indirectly.
Since the promoterless lacZ-Gm R cassette used to obtain the rirA mutant was inserted in the same transcriptional orientation as that of the mutated gene, rirA expression levels could be determined by measuring β-galactosidase activity in SVQ780 and SVQ780C strains grown in un-supplemented MM or in MM supplemented with different iron concentrations. As shown in Figure 3c, expression of rirA in the SVQ780 mutant increased slightly with increased iron content in the medium, exhibiting the lowest and highest expression levels in un-supplemented MM and in MM containing 220 µM FeCl 3 , respectively. These results are in agreement with data obtained in R. leguminosarum and in A. tumefaciens rirA mutants, in which rirA expression was downregulated under iron-deplete conditions [21,27]. In the complemented SVQ780C strain, expression of rirA was low irrespective of the concentration of iron in the medium, indicating an auto-regulatory role similar to that of RirA in R. leguminosarum and in A. tumefaciens [21,27].

Nitrogen-Fixing Symbiosis with Soybean Plants Is Impaired in an S. fredii HH103 rirA Mutant
Previous studies have shown that mutations in rirA in R. leguminosarum or in S. meliloti do not affect nodulation or symbiotic nitrogen fixation with their respective hosts (peas and vetch for R. leguminosarum, and alfalfa for S. meliloti) [21,23,24]. The symbiotic phenotype of the S. fredii rirA mutant was investigated in soybean (Glycine max) cv. Williams, which forms determinate nodules. The mutant was able to induce nodule formation on soybean roots. However, plants inoculated with the rirA mutant exhibited symptoms of impaired symbiotic effectiveness since they were smaller than plants inoculated with the wild-type strain and their leaves showed a chlorotic appearance similar to that exhibited by non-inoculated soybeans ( Figure 4). These symptoms were suggestive of nitrogen deficiency.
No significant differences were detected in the number of nodules developed in the roots of soybean plants inoculated with the wild-type or the rirA mutant. However, the nodules induced by the mutant exhibited a significant reduction in size and fresh weight and were not as red as those induced by the wild-type strain ( Figure 4 and Table 1). Moreover, acetylene reduction assays (ARA) revealed that nitrogenase activity of the mutant was 5.6-fold reduced with respect to that of SVQ269 (Table 1). In order to better understand the symbiotic deficiency of the mutant, nodules induced by the wild-type and the rirA mutant were analyzed by light microscopy ( Figure 5). mutant was investigated in soybean (Glycine max) cv. Williams, which forms determinate nodules. The mutant was able to induce nodule formation on soybean roots. However, plants inoculated with the rirA mutant exhibited symptoms of impaired symbiotic effectiveness since they were smaller than plants inoculated with the wild-type strain and their leaves showed a chlorotic appearance similar to that exhibited by non-inoculated soybeans (Figure 4). These symptoms were suggestive of nitrogen deficiency.     The results revealed that infected cells in nodules induced by the mutant contained fewer bacteria than those in nodules formed by the wild-type strain. The complemented strain SVQ780C regained the wild-type phenotype in symbiosis with soybean, including efficient nitrogen fixation ( Figure 4 and Table 1) and degree of bacterial infection in nodule cells ( Figure 5). Altogether, these data indicate that RirA is important for efficient symbiosis with soybean plants.

Discussion
In this work, we investigated the role of RirA, the rhizobial iron regulator, in S. fredii HH103. This study represents the first approach to tackle iron homeostasis in a broad-host range Rhizobium. Despite the pivotal role of iron in the establishment of efficient Rhizobium-legume symbiosis, studies dealing with how the microsymbiont controls the intracellular iron pools have only been performed The results revealed that infected cells in nodules induced by the mutant contained fewer bacteria than those in nodules formed by the wild-type strain. The complemented strain SVQ780C regained the wild-type phenotype in symbiosis with soybean, including efficient nitrogen fixation (Figure 4 and Table 1) and degree of bacterial infection in nodule cells ( Figure 5). Altogether, these data indicate that RirA is important for efficient symbiosis with soybean plants.

Discussion
In this work, we investigated the role of RirA, the rhizobial iron regulator, in S. fredii HH103. This study represents the first approach to tackle iron homeostasis in a broad-host range Rhizobium. Despite the pivotal role of iron in the establishment of efficient Rhizobium-legume symbiosis, studies dealing with how the microsymbiont controls the intracellular iron pools have only been performed in three rhizobia: B. japonicum, R. leguminosarum and S. meliloti. Whereas important insights have been obtained in B. japonicum, knowledge about iron homeostasis in Rhizobiaceae is limited. This is probably due to the fact that the different iron-related genes that have been characterized in R. leguminosarum and S. meliloti, including the iron regulators irr and rirA, are not essential for the formation of effective symbiosis with their respective host plants. Our study demonstrates that RirA is important for the symbiosis that S. fredii HH103 establishes with soybean.
S. fredii HH103 encodes homologues of three regulatory proteins that have been involved in iron homeostasis in different bacteria: Fur, Irr, and RirA. In R. leguminosarum and in S. meliloti, Irr and RirA participate in iron-dependent regulation in a rather complex pathway [17,25]. Inactivation of rirA in R. leguminosarum, S. meliloti and A. tumefaciens results in differential expression of a battery of genes with the remarkable upregulation of genes involved in iron uptake, whose expression is normally reduced under iron-sufficient conditions [21][22][23][24]26,27]. Our data indicate that S. fredii HH103 RirA functions as a regulatory protein. Firstly, like in R. leguminosarum and A. tumefaciens [21,27], rirA transcription in HH103 is subject to autoregulation since transcript levels decreased notably in the complemented strain compared with those observed in the mutant. It should be mentioned that, whereas rirA expression remained relatively constant in the complemented strain regardless of the iron levels in the medium, a slight but significant increase in transcriptional activity of rirA was detected in the rirA null mutant when grown in iron-rich medium compared to iron-deficient medium. This might indicate that rirA expression is controlled not only by its own product but also by another iron responsive regulator (e.g., Irr). In fact, a similar expression pattern was described in A. tumefaciens, where Irr was shown to repress rirA transcription [27]. Secondly, our data show that HH103 RirA not only controls its own expression but also that of the upstream iron-responsive gene fbpA, either directly or indirectly. Under iron-sufficient conditions, inactivation of rirA in HH103 leads to increased fbpA transcript levels compared with those detected in the wild-type. Similar results were obtained in transcriptomic studies performed in S. meliloti where the fbpA gene was induced under iron-limiting conditions and by rirA mutation [23]. Moreover, in silico predictions (RegPrecise (http://regprecise.lbl.gov/RegPrecise/regulog.jsp?regulog_id=1959) [34]) include FbpA as part of the RirA regulon in different rhizobia as well as in Brucella melitensis. In HH103, fbpA codes for a 347 aa-long protein, which shows significant homology to the ferric ABC transporter substrate-binding protein FbpA that is present in many bacteria, including pathogens [35]. This periplasmic protein participates in iron uptake by shuttling iron from the outer to the inner membrane. Although functional characterization of fpbA in HH103 is needed to determine its role in iron uptake, our expression data suggest that, similarly to other Rhizobiaceae members, HH103 RirA participates in iron homeostasis.
It is known that RirA regulates gene expression by binding to IRO boxes (5'-TGA-(N9)-TCA-3') located in the promoter region [14,19,20]. The intergenic fbpA-rirA region does not contain any IRO box. The closest IRO box sequence is located 426 nucleotides upstream of the rirA start codon and within the fbpA coding region. Additional studies are required in order to determine if this sequence or alternative imperfect IRO box sequences that could be present in the intergenic fbpA-rirA region might be binding sites for HH103 RirA.
The absence of RirA leads to species-specific growth phenotypes and increased sensitivity to oxidants [21,23,27]. Whereas inactivation of rirA in S. meliloti results in reduced growth only in iron-rich medium probably as a result of toxic intracellular iron levels, the same mutation in R. leguminosarum and A. tumefaciens does not cause an iron-dependent growth phenotype, although the R. leguminosarum rirA mutant grows slower than the wild-type on all tested media. In S. fredii HH103, inactivation of rirA leads to a growth advantage in media with low iron content. This phenotype is not caused by constitutive expression of a siderophore, since no CAS activity could be detected in either the wild-type or the rirA mutant. In fact, siderophore production seems to be a rare trait among soybean-nodulating rhizobia. B. japonicum USDA 110, another soybean symbiont, does not synthesize siderophores [36], and our data revealed that only two out of 40 different S. fredii strains tested exhibited CAS activity. Recently, iron availability has been shown to be the main factor influencing the distribution of soybean-nodulating rhizobia in China, with S. fredii strains associated with alkaline soils with low iron content, while Bradyrhizobium strains are found in soils with high iron content [37]. Therefore, it is tempting to suggest that the increased capacity of the S. fredii HH103 rirA mutant to grow under iron-limiting conditions might be caused by derepression of as yet unknown high-affinity iron uptake systems. This together with the observed fbpA upregulation in the rirA mutant, supports the notion that RirA controls iron homeostasis in HH103.
Although a high iron concentration in the medium (220 µM) negatively interferes with S. fredii HH103 growth indicating a toxic effect, no differences were detected between the wild-type and the rirA mutant. A possible explanation is that HH103 encodes effective systems that protect cells from an iron overload (e.g., iron storage proteins and/or iron efflux transporters), and that are not under RirA control. In this context, the regulation and role in iron homeostasis of SfHH103_03210 (bfr), a gene that putatively codes for bacterioferritin are worth further investigation. In S. meliloti, bfr expression is enhanced under iron-sufficient conditions and the gene affects the oxidative stress response and symbiotic phenotype of this Rhizobium [25]. Another non-excluding possibility that might explain why the rirA mutant shows similar resistance to high iron levels as the wild-type implies the existence of additional regulatory control on iron uptake systems in HH103 that function under high iron conditions. While the role of Irr awaits investigation, the participation of the regulatory protein MucR1 in regulating the expression of iron uptake systems (afuABC, hmuS and fbpA) has been reported in S. fredii strains [38,39].
Like the rirA mutants of S. meliloti and A. tumefaciens, S. fredii HH103 rirA is hypersensitive to H 2 O 2 . Hypersensitivity to peroxides in the former mutants was explained by the increased production of highly reactive oxygen radicals resulting from increased intracellular iron levels upon derepression of iron uptake systems. The same rationale could serve to explain why HH103 rirA is more sensitive to oxidative stress. The nature of the iron uptake systems that contribute to a metal overload in the absence of RirA have not been investigated in this study. Nevertheless, considering the upregulation of fbpA in the mutant, it is reasonable to venture the participation of ferric type ABC transporters. Iron and H 2 O 2 toxicities are intertwined. However, the behavior of HH103 suggests that the cell damaging effects of H 2 O 2 are more intense than those caused by the excess of iron. Compared to S. meliloti and A. tumefaciens, HH103 is much more sensitive to hydrogen peroxide. This fact suggests that HH103 does not possess an efficient antioxidant response.
In contrast to the symbiotic behavior of rirA mutants of R. leguminosarum and S. meliloti with their respective hosts, inactivation of rirA in S. fredii HH103 impairs the establishment of effective symbiosis with soybean plants, a legume that develops determinate-type nodules. The mutant is able to induce nodule formation on soybean roots but is less efficient in symbiotic nitrogen fixation than the wild-type strain as determined by the reduced nitrogenase activity and plant dry weight. The significant reduction in the number of bacteria present in infected cells in nodules induced by the rirA mutant could explain the lower effectiveness of the symbiotic interaction. The reasons for the reduced number of rirA cells within soybean nodules are unknown, although it is tempting to speculate that the mutant's hypersensitivity to oxidants might play a role. During the establishment of the symbiotic process, ROS are generated [40][41][42][43][44][45]. In order to establish a chronic infection within the host, rhizobia need to counteract the plant oxidative burst by using different strategies [44]. Due to the inability to repress iron uptake systems, ROS produced in the symbiotic interaction could have more detrimental effects on a rirA mutant than in the wild-type. Nevertheless, it remains to be determined if the role of HH103 RirA in contributing to the development of an efficient symbiosis is exclusive for the interaction with soybean plants, or can be extended to other legume hosts, including those that form indeterminate nodules. In A. tumefaciens, RirA is important for tumor formation on tobacco leaf pieces [26] but not on potato disks [27], indicating that the host plays an important role in the outcome of the interaction with mutants lacking this regulatory protein.
In summary, this work represents a basis for the study of iron homeostasis in S. fredii HH103. Although investigations on iron homeostasis regulation in members of Rhizobiaceae are limited, data obtained indicate that it is a very complex process. This complexity has been highlighted in a recent study performed in R. leguminosarum where iron-responsive genes were found to be regulated by O 2 in an RirA-independent fashion [20]. Despite the complexity of this biological process in which different regulators and signals might be involved, the identification and characterization of the different elements that participate in controlling intracellular iron pools deserves attention as a means to unveil genes that, like RirA, are crucial for the free-living and symbiotic lifestyles of broad host range rhizobia.

Bacterial Strains, Plasmids, and Growth Conditions
Bacterial strains and plasmids used in this study are listed in Table S1. Escherichia coli strains were grown in Luria-Bertani (LB) medium [46] at 37 • C; S. fredii strains were grown at 28 • C in tryptone-yeast extract (TY) medium [47] or in minimal medium (MM) containing different concentrations of FeCl 3 [48]. The MM was prepared without iron (un-supplemented) and 100-fold concentrated stock solutions of FeCl 3 were added to obtain the desired concentration of FeCl 3 .

Construction of the S. fredii rirA Mutant and Complementation.
The rirA mutant strain SVQ780 was obtained by allelic exchange. To obtain the mutant allele, a 1391 bp fragment containing the rirA gene of S. fredii SVQ269, a Rif R derivative of strain HH103, was PCR amplified using primers rirA-F and rirA-R (Table S2), and cloned into pGEM-T-Easy generating plasmid pMUS1265. After DNA sequencing, the rirA gene of the construct was disrupted by inserting a cassette into a unique BglII site located in its coding sequence. For that, the pMUS1265 plasmid was digested with BglII, treated with Klenow DNA Polymerase to make blunt ends, and ligated to a 4.3 kb SmaI fragment containing the lacZ-Gm R cassette previously isolated from pAB2001 [49]. The resulting construct, pMUS1276, was digested with EcoRI, and the 5.7 kb fragment containing the disrupted rirA gene was subcloned into pK18mobsacB to yield plasmid pMUS1287. This plasmid was introduced into SVQ269 via conjugation with E. coli strain S17-1, and allele replacement events were selected as described previously [50]. Allelic replacement in SVQ780 was confirmed by PCR and DNA-DNA hybridization.
For complementation of the rirA mutation present in SVQ780, the wild-type version of rirA was PCR amplified as described above and subcloned as an EcoRI/KpnI fragment into pK18mob. The resulting plasmid, pMUS1406, was conjugated into SVQ780 and Km R transconjugants were selected in order to isolate clones in which plasmid pMUS1406 had been integrated by a single recombination event. One of these transconjugants, named SVQ780C, was selected for further studies. The genomic organization in SVQ780C shown in Figure S3 was determined by PCR.

Siderophore Detection
The determination of siderophores in liquid cultures was performed using the Chrome Azurol S (CAS) assay solution described by Schwyn and Neilands [51]. Supernatants of S. fredii cultures grown in un-supplemented MM or MM containing different concentrations of FeCl 3 were mixed 1:1 with the CAS assay solution. After reaching equilibrium, the absorbance was measured at 630 nm. Alternatively, siderophore production was assessed by the appearance of an orange halo on CAS-agar plates upon contact with bacterial cells previously grown on TY or MM plates.

Sensitivity to H 2 O 2
Rhizobial cultures were grown in TY medium up to an OD 600nm of 1 and subject to different treatments. To determine the rate of survival upon H 2 O 2 challenge, the cultures were exposed to the desired concentration of H 2 O 2 (10 mM or 100 µM) for 1 h. The survival rate was calculated by determining the colony forming units (CFU) obtained on TY after plating serial dilutions of non-treated cultures and cultures exposed to H 2 O 2 . To test growth on plates containing H 2 O 2 , rhizobial cells grown in TY broth were washed and resuspended in the same volume of liquid MM. The cell suspension was then tenfold serially diluted and an aliquot (10 µL) of each dilution was spotted onto TY plates containing different concentrations of H 2 O 2 . The experiment was repeated at least twice to ensure the reproducibility of the results.

Reverse Transcription Quantitative PCR (RT-qPCR)
Total RNA was isolated using the High Pure RNA Isolation Kit (Roche, Mannheim, Germany) and RNase Free DNA Set (Qiagen, Hilden, Germany) according to the manufacturer's instructions. This (DNA free) RNA was reverse transcribed to cDNA by using PrimeScript RT reagent Kit with gDNA Eraser (Takara, Mountain View, CA, USA). Quantitative PCR was performed using a LightCycler 480 (Roche, Basel, Switzerland) with the following conditions: 95 • C, 10 min; 95 • C, 30 s; 50 • C, 30 s; 72 • C, 20 s; forty cycles, followed by the melting curve profile from 60 to 95 • C to verify the specificity of the reaction. The S. fredii HH103 16S rRNA 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 [52]. Selected genes and primers are listed in Table S2.

Nodulation Assays
Nodulation assays on Glycine max (L.) Merr. cv. Williams (soybean) were carried out as previously described [53]. Each Leonard jar contained two plants, and each plant was inoculated with about 5 × 10 8 bacteria. Plants were grown for 7 weeks in a plant growth chamber with a 16 h photoperiod at 25 • C in the light and 18 • C in the dark. Plant tops 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. [54]. Nitrogenase activity of nodules was assessed by ARA as described by Buendía-Clavería and associates [55]. For the different parameters analyzed, the values of each treatment were compared to those of S. fredii SVQ269 by using the Mann-Whitney non-parametrical test. At least two independent plant tests were carried out for each strain and host plant. Table 1 shows a representative experiment.

Microscopy Studies
For optical microscopy studies, small fragments of nodules (15-20 days old) were immediately fixed in 4% (v/v) glutaraldehyde in 0.1 M cacodylate buffer, pH 7.2 for 2 h at 4 • C after collection. Samples were washed several times in 0.1 M cacodylate buffer, pH 7.2, dehydrated in acetone at progressively higher concentrations and embedded in Epon (epoxy embedding medium). The semi-thin sections (0.3-0.5 µm) were obtained on a Reichert-Jung Ultracut E ultramicrotome, stained with Toluidine blue and viewed in a Leitz (Aristoplan) light microscope (Leica, Munich, Germany).