Role of the Regulatory Gene rirA in the Transcriptional Response of Sinorhizobium meliloti to Iron Limitation

Abstract

A regulatory network of Sinorhizobium meliloti genes involved in adaptation to iron-limiting conditions and the involvement of the rhizobial iron regulator gene (rirA) were analyzed by mutation and microarray analyses. A constructed S. meliloti rirA mutant exhibited growth defects and enhanced H₂O₂ sensitivity in the presence of iron, but symbiotic nitrogen fixation was not affected. To identify iron-responsive and RirA-regulated S. meliloti genes, a transcriptome approach using whole-genome microarrays was used. Altogether, 45 genes were found to be jointly derepressed by mutation of rirA and under different iron-limited conditions. As expected, a number of genes involved in iron transport (e.g., hmuPSTU, shmR, rhbABCDEF, rhtX, and rhtA) and also genes with predicted functions in energy metabolism (e.g., fixN3, fixP3, and qxtAB) and exopolysaccharide production (e.g., exoY and exoN) were found in this group of genes. In addition, the iron deficiency response of S. meliloti also involved rirA-independent expression changes, including repression of the S. meliloti flagellar regulon. Finally, the RirA modulon also includes genes that are not iron responsive, including a gene cluster putatively involved in Fe-S cluster formation (sufA, sufS, sufD, sufC, and sufB).

Iron is an essential micronutrient for almost all known organisms. Due to the ready interconversion between the reduced Fe²⁺ ferrous form and the oxidized Fe³⁺ ferric form, iron is a versatile component for incorporation as a biocatalyst or electron carrier into proteins that are involved in a number of essential metabolic and enzymatic functions. In particular, rhizobia, a diverse group of symbiotic soil bacteria belonging to the genera Rhizobium, Bradyrhizobium, Mesorhizobium, Azorhizobium, and Sinorhizobium, have a special demand for iron. These so-called root nodule bacteria are known for their ability to establish nitrogen-fixing symbioses with their legume hosts. During this symbiotic interaction root nodules are formed, in which the bacteria fix atmospheric nitrogen after differentiation into bacteroids (61). Many enzymes necessary for the reduction of atmospheric nitrogen (e.g., nitrogenase and nitrogenase reductase) contain iron compounds as cofactors. In addition, the bacteroids have a very high respiratory demand, requiring an abundance of cytochromes and other ferro-proteins (14). Thus, rhizobia not only need to compete successfully with other soil organisms for iron in order to propagate in the free-living state but also need to satisfy their high iron demand during symbiosis.

Although iron is abundant in the Earth's soil, under physiological conditions it is mainly present as insoluble ferric iron. Accordingly, bacteria have evolved a number of transport systems in order to ensure a sufficient supply of iron. A common mechanism employed by most bacteria is the synthesis and secretion of siderophores, which are low-molecular-weight ligands that specifically bind ferric iron with high affinity. The ferri-siderophore complexes are then transported into the cell via cognate transporters. A number of siderophores have been characterized in different rhizobia; these include the trihydroxamate vicibactin of Rhizobium leguminosarum (10, 11), the amino polycarboxylic acid siderophore of Sinorhizobium meliloti DM4 (59), and the citrate-based dihydroxamate rhizobactin 1021 produced by S. meliloti 1021 (46). In addition to their own or exogenous ferri-siderophores, rhizobia are also known to use a wide range of other iron sources, including ferric citrate, heme, and hemoglobin (3, 11, 39, 71). This flexibility with regard to iron utilization probably reflects adaptation to complex environmental conditions and underlines the importance of iron acquisition for rhizobia.

Despite the importance of iron, high concentrations of this metal can lead to generation of hydroxyl radicals as a result of the Fenton chemistry. Therefore, iron uptake is usually strictly regulated. In many gram-negative bacteria and some gram-positive bacteria the ferric uptake regulator (Fur) has been established as a central regulator of iron-responsive genes (1, 17, 22, 23). In the sequenced genome of S. meliloti 1021, the microsymbiont of alfalfa, a Fur homologue is encoded (9, 19), but recently we demonstrated that this protein is primarily involved in the regulation of the Mn²⁺ transporter operon sitABCD (12), a finding that was verified independently by Platero et al. (47). A similar role of Fur was also found in R. leguminosarum (15, 70). It is therefore obvious that in S. meliloti and R. leguminosarum iron-dependent regulation is implemented very differently than it is in other gram-negative bacteria. In R. leguminosarum, Todd and colleagues have identified a regulator termed RirA (rhizobial iron regulator) which has been shown to be involved in the regulation of several iron-responsive genes (64). Additionally, proteome analyses indicated that over 100 proteins were differentially expressed in an R. leguminosarum rirA mutant, and 10 of these proteins were identified by mass spectrometry (65).

The R. leguminosarum RirA protein does not exhibit significant sequence homology to known iron-responsive regulators, such as Fur, DtxR (7), or Irr (21), and belongs to the Rrf2 family of putative transcription regulators. Other members of this protein family include NsrR, which is involved in regulating a nitrite reductase in Nitrosomonas europaea (4), Rrf2, a regulator of cytochrome synthesis in Desulfovibrio vulgaris (30), and IscR, an iron-sulfur protein involved in the regulation of Fe-S cluster formation (54). Hence, the members of this family of regulators appear to have very diverse functions in different eubacterial species.

There is an rirA homologue in S. meliloti (64), but its role in this symbiont is still unknown. In this study we analyzed the relevance of the S. meliloti rirA gene for controlling iron homeostasis during free-living growth and symbiosis. As the S. meliloti rirA gene is a likely candidate for a global iron-dependent regulator, whole-genome microarrays were utilized to identify the extent of RirA-regulated genes. These experiments were complemented with transcriptome profiling of changes in global iron-dependent expression in order to identify the mechanism with which S. meliloti adapts to iron-limiting conditions and to define the relevance of rirA for coordinating the iron deficiency response.

MATERIALS AND METHODS

Bacterial strains, plasmids, and media.

The bacterial strains and plasmids used in this study are shown in Table 1. Strains of Escherichia coli were routinely cultured at 37°C in antibiotic medium no. 3 (Oxoid, Wesel, Germany). S. meliloti strains were cultivated at 30°C either in tryptone yeast (TY) complex medium (6) or in Vincent minimal medium (VMM) (67). VMM was prepared without iron. A final FeCl₃ concentration of 37 μM was used for iron-sufficient growth in VMM, and an FeCl₃ concentration of 0.37 μM was used for iron-deficient growth in VMM. Other iron sources and the concentrations used are indicated below. Iron-limited TY medium was prepared by adding the iron chelator 2,2′-dipyridyl to a final concentration 200 μM. When appropriate, antibiotics were added at the following concentrations: neomycin, 100 μg ml⁻¹; kanamycin, 50 μg ml⁻¹; and streptomycin, 600 μg ml⁻¹. Glassware was washed with 50 mM EDTA and 6 M HCl before it was thoroughly rinsed with water.

TABLE 1.

Bacterial strains and plasmids

Strain or plasmid	Characteristics	Reference
Sinorhizobium meliloti strains
Rm1021	Spontaneous mutant of wild-type strain RU47, Smr	35
Rm1021-TR2	Rm1021 derivative, ΔrirA	This study
Rm1021-TR2-3	Rm1021 derivative, ΔrirA ΔrhrA	This study
Escherichia coli strains
DH5αMCR	F−endA1 supE44 thi-1 λ−recA1 gyrA96 relA1 deoR Δ(lacZYA-argF)U169 φ80dlacZΔM15 mcrA Δ(mrr hsdRMS mcrBC)	20
S17-1	E. coli 294::[RP4-2(Tc::Mu)(Km::Tn7)] pro res ΔrecA Tpr	57
Plasmid
pK18mobsacB	pUC18 derivative, sacB lacZα Kmr, mobilizable	53

DNA manipulations.

The protocols of Sambrook et al. (52) were used for routine manipulations of plasmid and chromosomal DNA. Mutated DNA fragments containing either a 342-bp deletion in the rirA gene or a 810-bp deletion in the rhrA gene were constructed by gene SOEing (27). In a first PCR regions up- and downstream of the desired deletion were amplified, and then they were fused in a second PCR. The deletion constructs obtained were subsequently cloned into the suicide vector pK18mobsacB, which allows sucrose selection for vector loss (53). The resulting plasmids were conjugated into S. meliloti via E. coli S17-1 to introduce deletions by allelic exchange. Mutants were verified by PCR and Southern hybridization.

CAS siderophore assay.

Chrome azurol S (CAS) assay mixtures for siderophore detection were prepared as described by Schwyn and Neilands (55). Supernatants of S. meliloti cultures grown in VMM containing various concentrations of FeCl₃ were mixed 1:1 with a CAS assay solution. After equilibrium was reached, the absorbance at 630 nm was measured. The relative siderophore activity was determined by measuring optical density ratios of different cultures.

H₂O₂ challenge.

To test for H₂O₂ sensitivity, overnight cultures of S. meliloti wild-type strain Rm1021 and rirA mutant Rm1021-TR-2 were washed, and equal amounts of the two strains were incubated for 4 h in iron-free VMM and VMM containing 60 μM FeCl₃, respectively. After this, the samples were exposed to 50 mM H₂O₂ for 2 h. The survival rate was determined by comparing the number of CFU for H₂O₂-treated samples to the number of CFU for untreated control samples after incubation for 2 days on TY agar plates at 30°C.

Assays to determine nitrogen fixation efficiency in the S. meliloti-Medicago sativa symbiosis.

Nodulation tests were performed as described by Rolfe et al. (50). Solutions a to c for the plant agar plates (solution a contained 294 g/liter CaCl₂ · 2H₂O; solution b contained 50 g/liter KH₂PO₄; solution c contained 123 g/liter MgSO₄, 87 g/liter K₂SO₄, 0.247 g/liter H₃BO₃, 0.288 g/liter ZnSO₄, 0.1 g/liter CuSO₄ · 5H₂O, 0.056 g/liter CoSO₄ · 7H₂O, and 0.048 g/liter NaMo₄ · 2H₂O) were autoclaved separately. Then 0.5 ml of each solution was added to 1 liter autoclaved water containing 15 g agar. FeSO₄ was added as an iron source at final concentrations of 5, 10, and 15 μM to the medium. Alfalfa (M. sativa L. cv. europe) seeds were sterilized with 32% HCl for 30 min and then washed with sterile water. After germination, seedlings were placed on nodulation plates and inoculated with equal amounts (6 × 10⁷ CFU) of washed wild-type and mutant cells. Plants were weighed after 30 days of growth, and the nitrogen fixation activity was tested by the acetylene reduction assay (50).

S. meliloti transcript profiling using the genome-wide SM6kOligo microarray.

For identification of the RirA modulon, S. meliloti wild-type strain Rm1021 and rirA mutant Rm1021-TR2 were cultivated in VMM containing 37 μM FeCl₃. For transcriptome profiling of iron-regulated genes, the S. meliloti wild-type Rm1021 strain was grown in either TY medium (iron-sufficient complex medium), TY medium containing 200 μM 2,2′-dipyridyl (iron-limited complex medium), VMM containing 37 μM FeCl₃ (iron-sufficient minimal medium), or VMM containing 0.37 μM FeCl₃ (iron-limited minimal medium). All cultures were incubated in 250-ml Erlenmeyer flasks with shaking at 150 rpm at 30°C until the optical density at 580 nm was 0.9 before they were harvested. Cells were centrifuged (10,000 × g, 1 min, 4°C) and then immediately frozen in liquid nitrogen. For total-RNA isolation an RNeasy mini kit (QIAGEN, Hildesheim, Germany) was used. Cells were disrupted in the RLT buffer provided in the kit in Fast Protein tubes (Qbiogene, Carlsbad, CA) using a Ribolyser (30 s; speed, 6.5; Hybaid, Heidelberg, Germany) before RNA isolation using the RNeasy mini kit RNA purification protocol. Fluorescent labeling of cDNA by amino-allyl dye coupling was performed as described by de Risi (http://www.microarrays.org/protocols.html). For this study the Sm6kOligo microarrays described by Krol and Becker (32) were used. Hybridization and image acquisition of the microarrays were performed as described previously (5, 51). For acquisition of the mean signal and mean local background intensity for each spot of the microarray, the ImaGene 5.0 software (Biodiscovery Inc., Los Angeles, CA) for spot detection, image segmentation, and signal quantification was used. The log₂ value of the intensity ratios (M_i) was calculated for each spot as follows: M_i = log₂(R_i/G_i), where R_i = I_ch1i − Bg_ch1i and G_i = I_ch2i − Bg_ch2i (I_ch1i and I_ch2i are the intensities of spots in channels l and 2, respectively, and Bg_ch1i and Bg_ch2i are the background intensities of spots in channels 1 and 2, respectively). The mean intensity (A_i) was calculated for each spot as follows: A_i = log₂(R_iG_i)^0.5. A normalization method based on local regression that accounts for intensity spatial dependence in dye biases was used (73). Normalization and t-statistics were carried out using the Emma 1.1 microarray data analysis software (16). The expression of a gene was considered significantly different if the P value was ≤0.05, the log₂ ratio of the intensities (M value) was ≥1 or ≤−1, and the mean intensity (A value) was ≥7. The microarray results were verified for specific genes (rhbA, shmR, and hmuS) by quantitative reverse transcription-PCR using a QuantiTect SYBR Green reverse transcription-PCR kit (QIAGEN, Hildesheim, Germany) according to the manufacturer's instructions. The measurements were obtained with a LightCycler instrument (Roche, Mannheim, Germany). The rhbA and shmR genes were chosen since previous reports showed that there was iron-responsive regulation of these genes (3, 34). Iron-dependent regulation of the hmuS genes was confirmed by glucuronidase assays (data not shown).

RESULTS AND DISCUSSION

Mutational analysis of the S. meliloti rirA gene suggested that it is involved in the regulation of iron metabolism.

One of the objectives of this work was to elucidate iron-dependent regulation in S. meliloti. As we demonstrated previously that the S. meliloti fur gene is not involved in iron-dependent regulation, we focused on the rhizobial iron regulator gene rirA. The rirA gene was initially identified in R. leguminosarum as a possible regulator of iron-responsive genes (64). In S. meliloti a gene (SMc00785) homologous to the R. leguminosarum rirA gene was identified, which has been annotated to encode a hypothetical protein. The deduced protein had a predicted molecular mass of 17 kDa and exhibited very high levels of homology to the RirA repressor of R. leguminosarum (84% identity, 93% positive). Consequently, we renamed the corresponding S. meliloti open reading frame (SMc00785) rirA. Other close homologues of the S. meliloti RirA protein occur in the closely related plant pathogen Agrobacterium tumefaciens, Mesorhizobium loti, and different Brucella and Bartonella species. In none of these proteins could a significant helix-turn-helix motif, which is typical of many bacterial transcriptional regulators, be identified by bioinformatic approaches. The S. meliloti rirA gene is flanked downstream by genes encoding a putative dipeptide transporter (dppA1-F1) and upstream by a gene coding a putative periplasmic iron binding protein (SMc00784).

In order to assess the role of the S. meliloti rirA gene in regulation of global iron metabolism, a marker-free S. meliloti rirA deletion mutant designated Rm1021-TR2 (Table 1) was constructed and used for further studies. As we hypothesized that the rirA gene might be the central regulator of iron metabolism, we tested the influence of iron status on the growth of the rirA mutant. To do this, S. meliloti wild-type strain Rm1021 and rirA mutant Rm1021-TR2 were grown in VMM containing different iron sources. In medium containing 37 μM FeCl₃ as the sole iron source, the rirA mutant exhibited a reduced growth rate (Fig. 1a). This phenotype was partially recovered in minimal medium containing only 0.37 μM FeCl₃ (Fig. 1b). This reduction in growth of the rirA mutant was apparently connected to iron availability and may have been caused by oxidative stress due to derepressed iron uptake. With hemin as the iron source, virtually no growth of the rirA mutant was observed (Fig. 1c). Supplementation with other iron sources did not restore growth if hemin was present (data not shown), indicating that deregulated accumulation of this potentially toxic compound (63), rather than the inability of the rirA mutant strain to utilize hemin as an iron source, was responsible for the growth defects.

FIG. 1.

Effects of the rirA mutation on growth rates of S. meliloti strains. Overnight cultures of S. meliloti wild-type strain Rm1021 (WT) and rirA mutant Rm1021-TR2 (ΔrirA) were washed and diluted in fresh minimal medium containing 37 μM FeCl₃ (a), 0.37 μM FeCl₃ (b), or 10 μM hemin (c) as the sole iron source. Samples were taken, and their optical densities at 580 nm (OD₅₈₀) were determined. The error bars indicate the standard deviations calculated from six independent cultures.

Further evidence for the involvement of the rirA gene in regulating iron acquisition was obtained by semiquantitative liquid CAS assays in which the siderophore concentrations in supernatants of the rirA mutant were compared to those in wild-type cultures. When cultivated under low-iron conditions (no FeCl₃), both strains exhibited siderophore production, but the amount in the rirA mutant was ∼125-fold higher. Addition of 37 μM FeCl₃ to the culture medium abolished siderophore production in the wild type but did not affect siderophore accumulation in the rirA mutant (data not shown). Thus, the sensitivity of the rirA mutant to high iron concentrations might be at least partially caused by derepression of siderophore-dependent iron uptake mechanisms.

Since oxidative stress caused by the Fenton reaction in the presence of iron can be enhanced by low doses of H₂O₂, we tested the effect of the rirA mutation on H₂O₂ sensitivity in the presence of FeCl₃ to ascertain whether the growth deficits of the rirA mutant were a result of enhanced oxidative stress. When preincubated in medium containing no added iron sources before challenge with H₂O₂, the rirA mutant showed only a slightly lower survival ratio (46% ± 3%) than the wild type (58% ± 4%). However, the survival ratio of rirA mutant cells which were preincubated in iron-containing medium was reduced to only 4% ± 0.5%, whereas the viability of the wild type remained largely unchanged (40% ± 2%).

In conclusion, the bioassays indicated that a mutation in rirA leads to harmful accumulation of FeCl₃ or hemin, resulting in reduced viability due to oxidative stress. In other gram-negative bacteria, like E. coli (66), Yersinia pestis (62), and Pseudomonas aeruginosa (24), a fur mutation led to iron overload, resulting in iron-dependent growth phenotypes similar to those exhibited by the S. meliloti rirA mutant. Our data therefore suggested that the importance of the S. meliloti rirA gene for maintaining the intracellular iron balance below toxic levels is similar to the importance of fur in other bacteria. It is interesting that in R. leguminosarum an rirA mutation resulted in an iron-independent growth phenotype (64), indicating that there may be a different mode of rirA-mediated regulation and oxidative stress avoidance in R. leguminosarum.

Since deregulation of iron uptake caused by the rirA mutation had strong effects on the viability of S. meliloti, we speculated that the symbiotic properties might also be affected. To test this, M. sativa seedlings were inoculated with either the wild-type strain or the rirA mutant on plant medium containing FeSO₄ at concentrations between 5 and 15 μM. However, no significant differences in nodulation efficiency, foliage fresh weight, or nitrogen fixation rate between plants inoculated with the wild type and plants inoculated with the mutant were found (data not shown). In conclusion, our results indicated that rirA is essential for maintaining a balanced iron content in the cell during iron-sufficient free-living growth, but not during symbiosis. It is therefore possible that in planta iron uptake is regulated independent of rirA; alternatively, in nodules the iron availability might be limited, or iron could be provided in a form that does not require strict regulation by rirA.

Transcriptional profiling of the S. meliloti rirA mutant and the wild-type strain supported the hypothesis that the rirA gene has a regulatory function in iron metabolism.

In order to define the RirA modulon, transcriptome studies utilizing Sm6kOligo microarrays (32) were performed. The complete set of genes affected by the rirA mutation was identified by comparing the transcriptomes of the rirA mutant Rm1021-TR2 and the wild-type strain Rm1021 cultivated in iron-sufficient VMM. Duplicate samples were taken from three independent cultures of each strain, which yielded a total of six replicates for each experimental microarray set. RNA isolation, labeling, hybridization, and data analysis were performed as described in Materials and Methods. Only genes with statistically significant (P ≤ 0.05) changes in expression of twofold or more were considered. Using this approach, the expression of 195 genes was found to be significantly changed by the rirA mutation. Of these 195 genes, 132 were induced, whereas only 63 genes were repressed in the rirA mutant compared to the wild type. Figure 2a shows the results of microarray experiments expressed by plotting the log₂ expression ratio (M value) versus the mean signal intensity (A value) for each gene. As expected, many of the most highly induced genes (induction levels, 8- to 181-fold) belonged to the previously characterized rhizobactin 1021 synthesis and uptake cluster (including rhbABCDEF and rhtX) (34), further confirming the role of rirA as a repressor of iron uptake genes. In contrast, the only highly repressed gene that was found was the rirA gene itself, whose expression was reduced fourfold or more, an obvious result of deletion of this gene.

FIG. 2.

Scatter plots of the microarray-based analysis of S. meliloti gene expression affected by iron limitation and the rirA mutation. The plots show the logarithmic mean signal ratio (M-value) versus the logarithmic mean signal intensity (A-value) obtained by comparison of the transcriptomes of S. meliloti rirA mutant Rm1021-TR2 (ΔrirA) and S. meliloti wild-type strain Rm1021 (WT) (a), S. meliloti wild-type cells grown in iron-limited VMM and cells grown in iron-sufficient VMM (b), and S. meliloti wild-type cells incubated in iron-limited TY medium and cells grown in iron-sufficient TY medium (c). A number of genes with the greatest changes in mRNA abundance are indicated. The functions of the genes indicated are shown in Tables 2 and 3.

Transcriptome profiling of the S. meliloti wild-type strain grown with high and low iron concentrations defined the iron deficiency stimulon.

Assuming that RirA is a global regulator of iron-responsive genes, a large overlap between genes affected in expression by the rirA mutation and genes affected in expression by iron limitation was expected. Accordingly, two sets of microarray experiments were performed under iron-limited conditions to assess the role of rirA in coordinating gene expression in response to iron availability and also to identify other rirA-independent iron-regulated genes. Two different types of media were used for these experiments, defined VMM and complex TY medium, both either iron deficient or iron sufficient, in order to distinguish between iron-specific and medium-specific expression changes. The optimal culture conditions under which an iron deficiency response was elicited but which still allowed consistent growth were ascertained by monitoring the viability of S. meliloti cells incubated either in TY medium with increasing concentrations of the iron chelator 2,2′-dipyridyl or in VMM with various concentrations of FeCl₃ as the sole iron source. At the same time the siderophore production was measured by using a semiquantitative CAS assay as a marker for an iron deficiency response (data not shown). Consequently, we decided to compare the transcriptome of S. meliloti wild-type cells grown in VMM containing 0.37 μM FeCl₃ (iron-limited VMM) to the transcriptome of cells cultivated in VMM containing 37 μM FeCl₃ (iron-sufficient VMM); likewise, the gene expression of cultures grown in TY medium containing the iron chelator 2,2′-dipyridyl at a concentration of 200 μM (iron-limited TY medium) was compared to the gene expression of cells grown in TY medium without additives.

In iron-limited VMM the expression patterns of 378 genes were altered compared to the patterns observed for growth in iron-sufficient VMM. A total of 199 of these genes were induced by iron limitation, and 179 were repressed. An overview of genes with the greatest increases in mRNA abundance in iron-limited VMM, as shown in a scatter plot (Fig. 2b), revealed a significant consensus for the most highly induced genes in the S. meliloti rirA mutant (Fig. 2a). In contrast, the most highly repressed genes during growth in iron-limited VMM were involved in chemotaxis and cell motility, which was not apparent in the rirA mutant.

Finally, with 2,2′-dipyridyl-induced iron limitation in TY medium 318 genes were found to be differentially expressed (Fig. 2c). Of these 318 genes, 184 were upregulated, and 134 were downregulated. The scatter plots derived from these microarray results demonstrated that there was a high correlation between the most highly induced and repressed genes under both sets of iron-limited conditions, indicating that a comparison of these conditions is suitable for elucidating the transcriptome response of S. meliloti for dependence on iron availability.

Venn mapping identified the S. meliloti genes whose expression was affected by iron availability and/or the rirA mutation.

A complete overview of all genes with significant changes in expression is provided in Table S1 in the supplemental material. Tables 2 and 3 list genes which are differentially expressed under at least two different conditions. For a closer analysis of these data sets, genes that were differentially expressed during growth in iron-limited VMM, in iron-limited TY medium, or in the rirA mutant were plotted in a Venn diagram (Fig. 3).

TABLE 2.

S. meliloti genes significantly induced under iron-limiting conditions and in the rirA mutant^a

Open reading frame	Gene and/or description	Expression ratiob
		VMM	TY medium	ΔrirA
SMa0002	fdoG, probable FdoG formate dehydrogenase-O alpha subunit	−0.42	2.13	1.05
SMa0142	Possible protease	1.60	1.29	1.08
SMa0172	Conserved hypothetical protein	1.27	1.17	0.96
SMa0312	Hypothetical protein	2.43	1.74	2.09
SMa0314	Hypothetical protein	1.24	0.92	1.21
SMa0316	Conserved hypothetical protein	2.31	0.90	1.56
SMa0320	Putative dehydrogenase	1.81	0.49	1.68
SMa0612	fixN3, cytochrome c oxidase subunit 1	1.97	1.61	1.26
SMa0617	fixP3, cytochrome c oxidase membrane-anchored subunit	1.27	1.05	1.62
SMa0621	fixI2, E1-E2-type cation ATPase	1.00	0.52	1.11
SMa0994	Hypothetical protein	1.37	1.92	1.08
SMa1461	Putative muconate cycloisomerase	1.13	0.66	1.47
SMa1683	Putative arylsulfatase	2.57	1.02	1.74
SMa1686	Putative two-component response regulator	1.05	1.09	0.54
SMa1745	Putative iron ABC transporter permease	1.09	0.51	1.39
SMa1746	Putative iron ABC transporter periplasmic binding protein	1.98	1.96	2.41
SMa1749	Putative transcriptional regulator	0.47	1.16	1.23
SMa1860	Putative ABC transporter periplasmic binding protein	2.13	1.05	1.02
SMa1985	Hypothetical protein	1.09	0.83	1.56
SMa2259	Hypothetical protein	1.25	0.77	1.36
SMa2294	mrcA2, probable penicillin binding protein	1.35	1.47
SMa2337	rhtX, rhizobactin 1021 transporter	4.95	3.39	4.96
SMa2339	Putative siderophore biosynthesis protein	5.09	3.17	6.28
SMa2400	rhbA, rhizobactin 1021 biosynthesis protein	7.86	5.20	7.50
SMa2402	rhbB, rhizobactin 1021 biosynthesis protein	5.81	4.85	5.78
SMa2404	rhbC, rhizobactin 1021 biosynthesis protein	4.34	2.02	4.74
SMa2406	rhbD, rhizobactin 1021 biosynthesis protein	5.21	3.11	5.40
SMa2408	rhbE, rhizobactin 1021 biosynthesis protein	7.19	4.66	6.70
SMa2410	rhbF, rhizobactin 1021 biosynthesis protein	6.38	4.47	6.38
SMa2412	rhrA, transcriptional activator	1.40	1.75	2.42
SMa2414	rhtA, rhizobactin 1021 receptor	4.72	3.96	4.27
SMb20022	Conserved hypothetical protein	1.22	1.23	1.36
SMb20072	Putative rhizopine binding protein	1.06	3.46	0.37
SMb20099	Putative trehalose synthase protein	1.81	1.17	1.18
SMb20127	Hypothetical protein	1.36	0.56	2.88
SMb20203	cbbR, probable transcriptional regulator protein	1.07	0.63	1.38
SMb20545	Hypothetical protein	1.19	0.15	1.31
SMb20600	Hypothetical protein	1.73	1.43	0.75
SMb20755	pccB, putative propionyl-coenzyme A carboxylase beta chain protein	1.18	2.07	0.24
SMb20757	bhbA, methylmalonyl-coenzyme A mutase protein	1.40	2.07	0.52
SMb20924	abfA, putative alpha-L-arabinofuranosidase protein	1.76	0.70	1.44
SMb20946	exoY, galactosyltransferase protein	2.06	1.43	1.18
SMb20949	exoV, putative pyruvyltransferase protein	1.46	0.90	1.04
SMb20959	exoO, glucosyltransferase protein	0.70	1.12	1.37
SMb20960	exoN, UDP-glucose pyrophosphorylase protein	1.72	1.41	1.41
SMb20993	Putative monooxygenase protein	1.05	0.45	1.09
SMb21285	Conserved hypothetical protein	1.57	1.65	2.05
SMb21337	Putative oxidoreductase	1.43	1.48	0.65
SMb21432	Putative iron ABC transporter periplasmic solute binding protein	1.35	2.18	2.10
SMb21456	Hypothetical protein	1.88	1.33	0.60
SMb21464	Putative transcriptional regulator GntR family protein	1.07	0.10	1.42
SMb21487	cyoA, putative cytochrome o ubiquinol oxidase chain II protein	−0.51	2.88	2.94
SMb21488	cyoB, putative cytochrome o ubiquinol oxidase chain I protein	−0.29	2.76	2.55
SMb21489	cyoC, putative cytochrome o ubiquinol oxidase chain III protein	−0.09	2.64	2.08
SMb21690	exoW, glucosyltransferase protein	2.07	1.22	1.13
SMc00062	Hypothetical protein	1.05	0.88	1.63
SMc00086	cycG, putative diheme cytochrome c type signal peptide protein	1.90	−0.01	1.52
SMc00136	Putative oxidoreductase protein	1.39	1.37	0.60
SMc00235	trpD, probable anthranilate phosphoribosyltransferase protein	1.44	−0.06	1.74
SMc00301	Conserved hypothetical protein	0.40	1.02	1.31
SMc00338	Conserved hypothetical protein	1.78	1.23	0.82
SMc00371	Conserved hypothetical protein	1.09	1.11	0.27
SMc00401	Conserved hypothetical protein	1.72	1.70	2.99
SMc00402	Putative iron-regulated protein A	1.44	1.49	2.72
SMc00512	Conserved hypothetical protein	1.07	1.07	0.29
SMc00537	Putative transport protein	1.15	0.23	1.43
SMc00591	Hypothetical/unknown signal peptide protein	0.21	2.24	1.77
SMc00592	Hypothetical transmembrane protein	0.76	1.81	1.34
SMc00764	Hypothetical or unknown transmembrane protein	1.23	1.43	0.63
SMc00784	Putative iron ABC transporter periplasmic binding protein	1.42	2.86	2.94
SMc00819	katA, catalase	3.17	−0.02	1.16
SMc00830	Hypothetical protein	−0.32	1.20	1.08
SMc00832	glcD, probable glycolate oxidase subunit protein	1.14	0.56	1.32
SMc00922	Putative transporter transmembrane protein	1.20	0.55	1.06
SMc01016	Hypothetical protein	1.09	1.75	0.71
SMc01022	Putative cytochrome transmembrane protein	2.16	1.00	1.28
SMc01095	mexF1, probable multidrug efflux system transmembrane protein	0.21	1.04	2.81
SMc01169	ald, probable alanine dehydrogenase oxidoreductase protein	1.23	1.48	0.56
SMc01266	Conserved hypothetical protein	1.48	1.48	0.98
SMc01267	Conserved hypothetical protein	0.91	1.31	1.13
SMc01471	senC, putative cytochrome c oxidase assembly factor transmembrane protein	−0.13	1.47	1.29
SMc01489	Hypothetical protein	2.09	1.30	1.04
SMc01512	hmuT, putative hemin ABC transporter periplasmic binding protein	1.33	1.32	3.00
SMc01513	hmuS, Putative hemin ABC transporter ATPase	2.39	3.04	4.62
SMc01514	Conserved hypothetical protein	2.20	3.56	5.01
SMc01516	Conserved hypothetical protein	1.34	1.77	2.54
SMc01517	Conserved hypothetical protein	0.53	1.18	2.58
SMc01658	Putative transport protein	1.36	0.98	3.11
SMc01659	Putative iron ABC transporter periplasmic binding protein	1.21	1.87	5.51
SMc01718	Hypothetical transmembrane protein	1.64	1.29	0.66
SMc01747	hmuP, putative hemin transport protein	2.30	1.52	3.68
SMc01765	Hypothetical transmembrane protein	0.90	1.43	1.04
SMc01788	Hypothetical protein	2.45	1.92	1.21
SMc01814	Probable glutamate synthase small-chain protein	0.15	2.15	1.17
SMc02047	gcvT, probable aminomethyltransferase	1.58	2.68	0.24
SMc02075	Conserved hypothetical protein	1.06	1.12	−0.01
SMc02084	exbD, putative polymer transport protein	1.21	1.45	1.99
SMc02085	exbB, putative polymer transport protein	1.96	2.60	3.57
SMc02156	Conserved hypothetical protein	1.86	1.34	0.96
SMc02179	Conserved hypothetical protein	1.34	1.05	0.44
SMc02254	qxtB, putative quinol oxidase subunit II	1.09	2.66	1.08
SMc02255	qxtA, putative quinol oxidase subunit I	1.77	4.62	2.37
SMc02266	Conserved hypothetical protein	1.37	1.48	−0.14
SMc02708	Conserved hypothetical protein	1.04	1.05	0.38
SMc02726	shmR, hemin binding outer membrane receptor	3.98	2.90	5.98
SMc02727	Hypothetical protein	1.98	0.46	2.23
SMc02887	Hypothetical protein	2.19	0.78	2.17
SMc03149	Hypothetical protein	0.34	1.38	1.58
SMc03787	Hypothetical protein	1.64	3.19	2.70
SMc03971	mexF2, putative multidrug efflux system protein	1.19	1.01	1.10
SMc04049	Putative sulfite oxidase protein	1.11	1.03	0.21
SMc04162	Putative transcription regulator protein	1.39	−0.03	1.69
SMc04164	Hypothetical protein	1.40	1.56	0.92
SMc04206	Putative hemolysin type calcium binding protein	4.90	1.93	2.43

^aGenes were considered differentially expressed when according to t-statistics the P value was <0.05 with an induction level of >twofold. Only genes differentially expressed during at least two sets of microarray experiments were considered.

^bThe expression ratios are expressed as log₂ values.

TABLE 3.

S. meliloti genes significantly repressed under iron-limiting conditions and in the rirA mutant^a

Open reading frame	Gene and/or description	Expression ratiob
		VMM	TY medium	ΔrirA
SMa0249	Conserved hypothetical protein	−1.12	−1.06	−0.07
SMa0252	Conserved hypothetical protein	−2.32	−1.09	−0.40
SMa1289	Hypothetical protein	−1.68	−1.53	−0.10
SMa2032	Putative nonheme chloroperoxidase	−1.70	−1.92	−1.11
SMb20293	Hypothetical protein	−1.47	−1.21	−0.03
SMb20397	Putative oxidoreductase protein	−2.85	−1.69	−0.47
SMb20649	nadE1, putative NH3-dependent NAD+ synthetase protein	−1.09	−1.62	−0.02
SMb20650	Putative long-chain fatty acid coenzyme A ligase protein	−1.18	−1.71	−0.02
SMb20651	Hypothetical protein	−2.30	−1.96	0.10
SMb20764	phnL, putative phosphonate uptake ABC transporter ATP binding protein	−2.76	−2.31	−0.29
SMb20986	narB, putative nitrate reductase large-subunit protein	−1.61	−1.39	0.39
SMb21170	Putative guanylate kinase	−1.08	−1.41	−0.25
SMb21174	phoT, phosphate uptake ABC transporter permease protein	−1.07	−1.75	−0.11
SMb21175	phoE, phosphate uptake ABC transporter permease protein	−1.70	−1.90	−0.31
SMb21177	phoC, phosphate uptake ABC transporter ATP binding protein	−1.36	−1.31	−0.15
SMc00158	Hypothetical protein	−1.41	−2.25	−0.22
SMc00159	Hypothetical protein	−1.70	−2.12	0.12
SMc00283	Putative transcription regulator protein	−3.82	−3.05	−0.64
SMc00335	rpsA, 30S ribosomal protein s1	−0.44	−1.14	−1.05
SMc00565	rplI, probable 50S ribosomal protein 19	−0.63	−1.12	−1.05
SMc00578	Hypothetical protein	−0.47	−2.05	−1.20
SMc00765	mcpZ, probable methyl-accepting chemotaxis protein	−3.68	−3.29	−0.23
SMc00887	Hypothetical protein	−3.35	−3.43	−0.50
SMc00888	Hypothetical protein	−3.08	−3.01	−0.08
SMc00975	mcpU, probable chemoreceptor (methyl-accepting chemotaxis) transmembrane protein	−4.11	−4.14	−0.81
SMc00986	Hypothetical protein	−1.83	−1.79	−0.45
SMc00998	Conserved hypothetical signal peptide protein	−1.63	−1.19	0.08
SMc01104	mcpX, probable chemoreceptor (methyl-accepting chemotaxis) transmembrane protein	−2.98	−3.67	−0.36
SMc01160	Putative transcriptional regulator protein	−2.51	−0.47	−1.17
SMc01319	rplJ, probable 50S ribosomal protein	−1.33	−1.10	−1.51
SMc01326	tufB, probable elongation factor Tu protein	−1.04	−0.42	−1.25
SMc01469	mcpW, probable methyl-accepting chemotaxis transmembrane protein	−4.38	−3.41	−0.63
SMc01842	Putative methyltransferase transcription regulator protein	−0.49	−2.05	−1.00
SMc01847	Putative methyltransferase protein	−1.40	−1.25	−0.92
SMc02392	Hypothetical protein	−1.39	−1.81	−0.76
SMc02634	Hypothetical transmembrane protein	−2.60	−2.12	−0.91
SMc03004	mcpE, putative chemoreceptor (methyl-accepting chemotaxis protein)	−1.36	−1.44	−0.57
SMc03005	Conserved hypothetical protein	−1.21	−1.85	−0.30
SMc03007	cheA, chemotaxis protein (sensory transduction histidine kinase)	−2.80	−3.04	−0.57
SMc03008	cheW1, chemotaxis protein	−2.04	−2.29	−0.37
SMc03009	cheR, chemotaxis protein methyltransferase	−4.21	−3.68	−0.43
SMc03012	cheD, chemotaxis protein	−2.18	−2.00	−0.63
SMc03013	Conserved hypothetical protein	−2.07	−2.03	−0.52
SMc03017	Conserved hypothetical protein	−1.61	−1.80	−0.84
SMc03021	fliM, flagellar motor switch transmembrane protein	−2.41	−2.83	−0.71
SMc03022	motA, chemotaxis (motility protein A)	−2.66	−2.81	−0.80
SMc03023	Conserved hypothetical protein	−2.42	−2.69	−0.76
SMc03024	flgF, flagellar basal body rod protein	−3.14	−3.36	−1.00
SMc03027	flgB, flagellar basal body rod protein	−3.25	−3.50	−0.84
SMc03028	flgC, flagellar basal body rod protein	−1.35	−1.74	−0.59
SMc03029	fliE, flagellar hook basal body complex protein	−3.20	−2.72	−0.79
SMc03030	flgG, flagellar basal body rod protein	−3.35	−3.30	−0.89
SMc03031	flgA, flagellar precursor transmembrane protein	−1.16	−1.88	−0.69
SMc03034	flgH, flagellar 1 ring protein precursor	−1.62	−1.97	−0.56
SMc03035	fliL, flagellar transmembrane protein	−2.66	−3.11	−1.19
SMc03037	flaA, flagellin protein	−3.95	−4.37	−0.11
SMc03038	flaB, flagellin protein	−3.97	−3.76	−0.42
SMc03040	flaC, flagellin protein	−2.28	−2.36	−0.45
SMc03042	motB, chemotaxis protein	−1.84	−2.00	−0.83
SMc03043	motC, chemotaxis protein precursor	−1.92	−2.49	−0.96
SMc03044	motD, chemotaxis protein	−2.97	−3.04	−0.82
SMc03045	Hypothetical transmembrane protein		−4.50	−1.41
SMc03047	flgE, flagellar hook protein	−1.66	−1.61	−0.50
SMc03048	flgK, putative flagellar hook-associated protein	−2.50	−1.90	−0.81
SMc03049	flgL, putative flagellar hook-associated protein	−3.14	−2.25	−0.85
SMc03051	flbT, putative flagellin synthesis repressor protein	−2.21	−1.90	−0.68
SMc03052	flgD, putative basal body rod modification protein	−1.94	−1.71	−0.64
SMc03057	Conserved hypothetical transmembrane protein	−1.57	−2.56	−0.74
SMc03071	Hypothetical protein	−1.68	−2.06	−0.56
SMc03072	Conserved hypothetical protein	−3.35	−2.48	−0.62
SMc03242	Probable GTP binding protein	−0.21	−1.10	−1.61
SMc03857	ffh, probable signal recognition particle protein	−0.84	−1.16	−1.64
SMc04007	Conserved hypothetical protein	−1.18	−0.86	−1.19
SMc04009	Conserved hypothetical protein	−0.85	−1.03	−1.14
SMc04059	Hypothetical protein	−2.86	−3.23	−0.66
SMc04300	afuC, probable iron ABC transporter ATPase	−0.75	−1.40	−1.39
SMc04317	afuA, probable iron ABC transporter periplasmic binding protein	−1.52	−1.55	−0.90

^aGenes were considered differentially expressed when according to t-statistics the P value was <0.05 with a repression level of >two fold. Only genes differentially expressed during at least two sets of microarray experiments were considered.

^bThe expression ratios are expressed as log₂ values.

FIG. 3.

Venn mapping of S. meliloti genes with significantly altered expression obtained from three microarray experiments. The amounts of significantly induced genes (left) and repressed genes (right) were derived from the microarray-based global profiling of iron-responsive and RirA-regulated gene expression in S. meliloti. Of special interest were genes that were substantially up- or downregulated under both low- and high-iron conditions but not in the rirA mutant (subset A), during all three microarray experiments (subset B), and only in the rirA mutant compared to the wild type (WT) (subset C). The Venn diagrams are based on the microarray results shown in Table S1 in the supplemental material.

This led to identification of 136 genes (71 induced genes and 65 repressed genes) that were differentially expressed in both iron-limited media compared to growth under iron-sufficient conditions (Fig. 3, subsets A and B). These up- and downregulated genes are likely to be generally involved in the adaptation of S. meliloti to low-iron conditions and were considered to represent the S. meliloti iron deficiency stimulon. Of the 71 induced genes of the iron deficiency stimulon, 45 were also found to be induced in the rirA mutant (Fig. 3, subset B), thus supporting the hypothesis that rirA has a role as a major repressor of iron-responsive genes. In contrast, only 4 of the 65 repressed genes belonging to the iron deficiency stimulon were found to be repressed in the rirA mutant and under iron-limiting conditions, further indicating that rirA is involved primarily in repression rather than in induction of genes with dependence on iron.

An additional 87 genes of the iron deficiency stimulon (26 upregulated genes and 61 downregulated genes) (Fig. 3, subset A) were not affected by the rirA mutation, indicating that other regulatory mechanisms are involved during adaptation of S. meliloti to growth under low-iron conditions. Finally, 90 genes were found to be differentially expressed exclusively in the rirA mutant, independent of iron availability (Fig. 3, subset C). Several explanations might account for these transcriptome changes: (i) the rirA gene might also exert regulatory control on genes in an iron-independent way, (ii) the iron limitation used in this study might not have been sufficient to change the expression above the threshold value that we considered significant, and (iii) the deregulated iron uptake in the rirA mutant and the resulting changes in the intracellular iron content might have led to secondary transcriptome changes independent of actual regulation mediated by rirA.

S. meliloti genes involved in iron acquisition, energy metabolism, and exopolysaccharide production are induced in the rirA mutant, as well as under iron-limiting conditions.

The phenotypic analyses suggested that the rirA gene is involved in the repression of iron uptake and siderophore synthesis. In agreement with this, a large number of the 45 genes induced both under iron-limiting conditions and by the rirA mutation (Fig. 3, subset B) could be connected to iron acquisition (Table 2). Some of the most highly induced genes were the complete rhbABCDEF rhizobactin 1021 synthesis operon (34) and genes encoding the cognate outer membrane receptor (rhtA) and the recently characterized inner membrane transporter (rhtX) (13) (Fig. 4a). The transcriptional activator rhrA, which was demonstrated to be involved in the induction of siderophore synthesis and uptake genes (34), is located in the same gene cluster. Since rhrA was induced by iron limitation as well as the rirA mutation, indirect regulation of the rhizobactin 1021 synthesis and uptake cluster mediated by rirA via rhrA appeared to be likely. In fact, in a constructed rhrA/rirA double mutant designated Rm1021-TR2-3, siderophore overproduction was abolished, as indicated by CAS assays (data not shown), providing evidence that the rirA-mediated regulation of siderophore synthesis is dependent on the rhrA gene. A putative siderophore synthesis gene (SMa2339) which is possibly involved in the formation of rhizobactin 1021 in a hitherto unknown way is located in the immediate vicinity of rhtA. Moreover, a number of putative siderophore type ABC transporter genes (SMa1746, SMb21432, and SMc01659) were found among the genes induced under iron-limiting conditions and in the rirA mutant. It is of interesting that according to the original annotation SMb21431 codes for the C terminus of a periplasmic binding protein, but resequencing of the corresponding DNA region showed that in the published sequence an error resulting in a frameshift is present. In fact, SMb21431 and SMb21432 constitute a single open reading frame and are referred to as SMb21432 in this report. In addition to putative siderophore type ABC transporter genes, a chromosomal region encoding a putative hemin ABC transporter was also induced by iron limitation and in the rirA mutant (Fig. 4a). This putative hemin transporter is encoded by hmuP, hmuS, hmuT, hmuU, and hmuV, which are homologous to heme uptake genes of Bradyrhizobium japonicum (38), R. leguminosarum (71), and Y. pestis (26). The expression ratios of these genes are shown in Fig. 4a. In addition to these genes, shmR, encoding an outer membrane receptor for hemin (3), was also significantly induced in the rirA mutant and under both iron-limiting conditions. The deregulation of these genes in the rirA mutant is also likely to be a major factor contributing to the observed sensitivity to hemin (Fig. 1c). Interestingly, genes coding for hypothetical proteins in the immediate vicinity of the hmu transporter genes (Fig. 4a) were also induced both under iron-limiting conditions and in the rirA mutant (SMc01514, SMc01516, SMc01517, and SMc01518). Overall, these findings suggest that the secretion and uptake of siderophores and the induction of heme utilization systems are the main strategy that S. meliloti uses to counter iron deficiency and that the repression of these systems is mediated by rirA. The only nonsiderophore/heme-type iron transporter gene found to be expressed more highly during iron limitation and in the rirA mutant was SMc00784, which encodes the periplasmic binding protein of a ferric type ABC transporter. This protein exhibits 52% identity to FbpA of Mannheimia (formerly Pasteurella) hemolytica, which is involved in iron acquisition during pathogenic host invasion (31, 56). Interestingly, no other components of an ABC transporter were located in the vicinity of SMc00784. None of these putative transporters have been characterized yet, and our results indicate that they may play a major role in the adaptation of S. meliloti to low-iron conditions. For rhbA, shmR, and hmuS the results of the microarray experiments were verified by quantitative PCR. The same tendencies were found, although the induction ratios obtained in the quantitative PCR experiments were higher than the values obtained in the microarray experiments (data not shown). This is consistent with other experiments in which the Sm6kOligo microarray was used (32).

FIG. 4.

Clustered S. meliloti genes whose expression is affected by iron availability and by the rirA mutation. (a and b) Genetic maps of the rhizobactin 1021 synthesis/uptake and putative hemin uptake clusters (a) and a selection of putative energy metabolism gene clusters which are induced in the rirA mutant and under iron-limited conditions (b). (c) Gene cluster putatively involved in Fe-S cluster formation which is induced only in the S. meliloti rirA mutant. The numbers below the genes indicate the log₂ expression ratios of the genes derived from the microarray-based transcriptome analysis. Genes that were significantly induced in at least one experiment are indicated by solid arrows. The values in parentheses are expression ratios that were below the threshold used in this study.

Curiously, genes encoding the subunits of a putative ferric-type transporter (afuA and afuC) were downregulated instead of upregulated during iron-limited growth and in the rirA mutant. Recent work of Krol and Becker suggested that these genes might in fact be under the control of the phosphate-dependent regulator PhoB (32), and it is possible that despite high levels of homology to ferric-type transporters this transporter might play a different role in S. meliloti.

In addition to iron transporter genes, SMc00402, exbB, and exbD were also found to be derepressed in the rirA mutant and in either iron-limited medium. The deduced gene product of SMc00402 exhibits low levels of homology to IrpA (iron-regulated protein A) of Synechococcus sp. strain PCC7942 (48), which was proposed to be involved in iron acquisition, and a similar role for SMc00402 might be proposed for S. meliloti. The ExbB and ExbD proteins are generally involved in forming the energy-transducing TonB complex, which energizes the transport of iron substrates by specific receptors across the outer membrane (8). It is therefore likely that the S. meliloti exbB and exbD genes are involved in the transport of ferri-siderophores or heme compounds by the corresponding outer membrane receptors. We found that in the genome of S. meliloti only one tonB homologue could be detected (SMc01515), which was not differentially expressed in response to iron limitation or by the rirA mutation. An individual analysis of the microarray experiments with S. meliloti grown in TY medium or in VMM under iron-limiting conditions revealed no other genes with homologies to iron acquisition genes, demonstrating that all iron acquisition genes induced under iron-limiting conditions are also affected by the rirA mutation, which is strong evidence that the S. meliloti RirA protein is the main regulator of iron uptake.

In addition to iron acquisition genes, other genes involved in energy metabolism, noniron transport, and exopolysaccharide synthesis exhibited enhanced levels of expression during iron limitation and in the rirA mutant (Table 2). Altogether, 10 electron transport-associated genes were found to be expressed at an elevated level in the rirA mutant and under at least one type of iron-limiting growth conditions (SMc01022, fixN3, fixP3, qxtA, qxtB, cyoA, cyoB, cyoC, and cycG) (Table 2). The fix-3 gene cluster encodes a putative cytochrome oxidase and represents another iteration of the previously characterized fix-1 and fix-2 gene clusters (2). While at least either a functional fix-1 or fix-2 gene cluster is known to be essential for establishing a successful symbiosis (49), the fixN3 and fixP3 genes obviously have functions other than thsoe of their paralogues and, according to our results, are also regulated differently. Overall, the microarray analyses suggest that iron availability might influence the path of the respiratory chain in S. meliloti, which is consistent with the results of similar studies of other bacteria (40, 42, 43, 45, 58).

Three of the S. meliloti exopolysaccharide genes (exoW, exoY, and exoN) were induced under all three conditions, and two additional genes (exoO and exoV) were induced in the rirA mutant and during growth in iron-limited TY medium and VMM, respectively. These findings are in agreement with previous reports, which described enhanced production of exopolysaccharides in other bacteria as a result of nutrient stress (68, 72). In addition, katA encoding a monofunctional catalase was found to be induced in the rirA mutant and in iron-limited VMM. Previously, the S. meliloti katA gene was found to be inducible by H₂O₂ (25) and to be highly expressed in bacteroids (28). Our results demonstrate that the regulation of this gene is also affected by iron availability and the rirA gene. Finally, in addition to the activator gene rhrA mentioned above, a number of putative transcription regulator genes (SMa1749, SMc04162, SMc01160, and SMb21464) were differentially expressed depending on the rirA mutation and under at least one type of iron-limiting conditions. Hence, it is possible that in some cases rirA-mediated iron-dependent regulation might be exerted indirectly via these putative regulators, and it would be interesting to elucidate how these regulators contribute to the rirA-mediated regulatory network.

In R. leguminosarum iron-responsive regulators were detected, which are potential RirA binding sites (74). We conducted in silico searches in the upstream region of genes induced in the S. meliloti rirA mutant but were not able to identify any iron-responsive regulator-like sequences.

S. meliloti genes involved in cell motility are repressed under iron-limiting conditions but not in the rirA mutant.

Altogether, 87 genes were differentially expressed during growth in both iron-limited media but were not affected by the rirA mutation (Fig. 3, subset A). No function could be assigned to a large portion of these genes (57% of the induced genes and 21% of the repressed genes). A striking observation was that 30 of the downregulated genes are involved in motility. These genes represent 46% of all predicted motility genes of this bacterium. Twenty-four of these repressed motility genes are located in a cluster which represents the S. meliloti flagellar regulon (60).

These genes are known to be regulated in a hierarchical order, with VisN and VisR acting as global activators. In iron-limited VMM the visN and visR genes were also found to be significantly repressed two- to threefold (see Table S1 in the supplemental material), which probably accounts for the massive reduction of motility genes. This is also likely to be the case during growth in iron-limited TY medium, despite the fact that the visNR operon was repressed only ∼1.4-fold, which is below the threshold that we used in this study.

Moreover, seven additional genes annotated to encode hypothetical proteins (SMc03005, SMc03013, SMc03017, SMc03023, SMc03057, SMc3071, and SMc3072) located in the motility cluster and four other genes which are putatively involved in chemotaxis (SMc00765, SMc01469, SMc01104, and Smc00975) but are not located in this cluster were also found to be downregulated during iron limitation. It has been reported that C, N, or phosphate starvation in S. meliloti results in a reduction in motility (69). Here we provide evidence that iron limitation also negatively affects the expression of motility genes and that this may be caused by reduced transcription of the master activators of the flagellar regulon.

S. meliloti genes involved in the biosynthesis of Fe-S clusters are differentially expressed in the rirA mutant but not under iron-limiting conditions.

Altogether, 90 genes involved in a variety of functions were differentially expressed in the rirA mutant (Fig. 3, subset C) but were not affected by iron availability. Although we are not able to easily distinguish between direct rirA-mediated regulation and secondary effects, as proposed above, in at least some cases reasonable hypotheses can be advanced to explain the observed regulation. For instance, a cluster of genes (SMc00530, SMc00531, SMc00532, SMc00533, SMc00302, and SMc00301), including a putative nifS gene, was derepressed in the S. meliloti rirA mutant background (Fig. 4 c). The products encoded by genes in this cluster were annotated as either hypothetical proteins or putative ABC transporters, but our renewed annotation showed that the products of SMc00301, SMc00530, SMc00531, SMc00532, and SMc00533 exhibited homology to the E. coli SufA, SufB, SufC, SufD, and SufS proteins, respectively, which are involved in assembly of Fe-S clusters (18). In E. coli and Erwinia chrysanthemi these genes are important for iron acquisition and were found to be induced by oxidative stress and iron starvation (33, 36, 44). Thus, it has been assumed that the suf genes are specifically adapted to synthesize Fe-S clusters when iron or sulfur metabolism is disrupted by iron starvation or oxidative stress (41). In both enterobacteria the iron-dependent regulation of the suf genes was mediated by Fur. Since we demonstrated that in S. meliloti numerous usually fur-mediated functions are fulfilled by rirA, it is resonable to hypothesize that rirA might also be involved in the regulation of the suf genes. Another possibility is that the oxidative stress generated in the rirA mutant by deregulated iron uptake might indirectly lead to the induction of the suf homologues in S. meliloti. In a recent report an suf cluster in R. leguminosarum was found to be induced in an rirA mutant, which is in accordance with our results. Interestingly, the induction was found to be highest with high iron concentrations (65), possibly indicating that iron-generated oxidative stress also contributes to the regulation of the suf genes in R. leguminosarum.

Among the downregulated genes, the most striking groups of genes (17 genes) are involved in protein biosynthesis, including the synthesis of ribosomal proteins and elongation factors (see Table S1 in the supplemental material). This observation is probably a secondary effect due to the reduced growth of the S. meliloti rirA mutant.

Conclusions.

Iron acquisition is an important aspect of the rhizobial life cycle, and iron limitation severely inhibits the effectiveness of the rhizobium-legume symbiosis (29, 37). The large number of genes differentially regulated in response to iron availability found in this study stresses the importance of this micronutrient for S. meliloti. Bioassays with the S. meliloti rirA mutant clearly demonstrated the importance of the rirA gene for maintaining intracellular iron concentrations below toxic levels under free-living conditions. This finding was supported by the results of the microarray experiments, which implied that rirA has a central role in coordinating the transcriptional response with iron availability. In the case of siderophore synthesis and the uptake cluster, we obtained evidence that the rirA-dependent regulation is exerted indirectly via the transcriptional activator gene rhrA. Furthermore, the number and variety of regulated genes indicate that regulation mediated by rirA is not limited to iron acquisition. While it was not the goal of this work to define the RirA regulon on a molecular level, our analyses identified a number of possible directly RirA-regulated genes that may be used as targets for further studies. Overall, the extent of rirA-mediated regulation and the phenotypic analyses of the S. meliloti rirA mutant revealed striking parallels to fur-mediated regulation. It appears that in S. meliloti and likely also in R. leguminosarum RirA has the role of a central coordinator of iron-dependent regulation, while the Fur proteins in these rhizobia are involved mainly in the regulation of manganese uptake (12, 15, 47). In this study a large number of genes whose expression was previously not known to be affected by iron, as well as uncharacterized genes with previously unknown functions, were identified. This clearly illustrates the need for further analyses to truly understand the iron metabolism of S. meliloti. Our study thus was the first global analysis of iron metabolism and its novel form of regulation by rirA in a rhizobial species. The full data sets for genes whose expression is affected by iron availability and the rirA mutation constitute a sound basis for targeted mutagenesis experiments.