RirA, an iron-responsive regulator in the symbiotic bacterium Rhizobium leguminosarum

Jonathan D. Todd1, Margaret Wexler1, Gary Sawers2, Kay H. Yeoman1, Philip S. Poole3 and Andrew W. B. Johnston1

School of Biological Sciences, University of East Anglia, Norwich NR4 7TJ, UK1
Department of Molecular Microbiology, John Innes Centre, Norwich Research Park, Colney Lane, Norwich NR4 7UH, UK2
Department of Animal and Microbial Sciences, University of Reading, Reading RG6 6AJ, UK3

Author for correspondence: Andrew W. B. Johnston. Tel: +44 1603 592264. Fax: +44 1603 592250. e-mail: a.johnston{at}uea.ac.uk


   ABSTRACT
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Mutations in a Rhizobium leguminosarum gene, rirA (rhizobial iron regulator), caused high-level, constitutive expression of at least eight operons whose transcription is normally Fe-responsive and whose products are involved in the synthesis or uptake of siderophores, or in the uptake of haem or of other iron sources. Close homologues of RirA exist in other rhizobia and in the pathogen Brucella; many other bacteria have deduced proteins with more limited sequence similarity. None of these homologues had been implicated in Fe-mediated gene regulation. Transcription of rirA itself is about twofold higher in cells grown in Fe-replete than in Fe-deficient growth media. Mutations in rirA reduced growth rates in Fe-replete and -depleted medium, but did not appear to affect symbiotic N2 fixation.

Keywords: iron-mediated regulation, rhizobia, RirA, siderophores

Abbreviations: CAS, chrome azural sulphonate; VB, vicibactin

The GenBank accession number for the RirA sequence is CAC35510.


   INTRODUCTION
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
With very few exceptions (see Archibald, 1983 ; Posey & Gherardini, 2000 ), all organisms require iron, a major constituent of many metallo-enzymes and an important component of many essential redox reactions. But, due to its insolubility in most biologically relevant conditions, organisms go to great lengths to acquire this metal, employing diverse mechanisms, such as the synthesis of siderophores, low-Mr compounds that bind avidly to ferric (Fe3+) iron, which are then imported by cognate transporters. Other bacterial Fe sources include haem (as a free molecule or bound to proteins, e.g. haemoglobin), Fe-containing proteins (e.g. transferrin), ferric citrate, or inorganic Fe either in its ferric or its reduced, ferrous forms. Although Fe is (usually) an essential metal, it can also be harmful, catalysing the production of potentially damaging reactive oxygen species via the Fenton reaction. In view of the balance between sufficiency and toxicity, it is not surprising that many genes involved in Fe uptake (and other aspects of Fe metabolism) are regulated, being expressed (or functional) only when there is a shortage of the metal (see Braun et al., 1998 ; Hantke, 2001 ).

Most work on Fe-responsive gene regulation has been on the ‘global’ ferric uptake regulator, Fur, first identified in Escherichia coli (Hantke, 1981 ). Although Fur homologues exist in many different bacteria (see Hantke, 2001 ), detailed molecular studies on this regulator are mostly confined to a few members of the {gamma}-proteobacteria (e.g. Escherichia, Vibrio, Salmonella and Pseudomonas). In such bacteria, in the presence of Fe2+, Fur protein binds to ‘Fur-boxes’, which are conserved sequences near promoters of genes whose transcription it regulates. When bound, Fur inhibits transcription of these promoters; thus the greater the concentration of available Fe, the lower the expression of the Fur-regulated operons. In E. coli and Pseudomonas aeruginosa, many operons are Fe-repressible, via the action of Fur (Panina et al., 2001 ; Ochsner & Vasil, 1996 ). Many of these genes are directly involved in Fe uptake (e.g. those for siderophore synthesis or uptake); others are more indirectly linked to Fe metabolism (e.g. those involved in aspects of pathogenicity). In addition to repressing a variety of genes, Fur is required for Fe-dependent induction of several operons. Recently, it was shown that E. coli Fur represses transcription of ryhB, which specifies a small, regulatory RNA that in turn inhibits expression of several genes, including sodB, whose transcription initially appeared to be activated by Fur (Masse & Gottesman, 2002 ). Thus, this ‘activation’ is due to repression of a repressor by Fur.

A second class of Fe-dependent regulator is exemplified by DtxR, first identified in Corynebacterium diphtheriae as a repressor of toxin production (Boyd et al., 1990 ). It seems that in Gram-positive bacteria, DtxR-like proteins assume at least some of the roles that Fur undertakes in other bacterial classes (Dussurget et al., 1996 ; Hantke, 2001 ). Although DtxR and Fur proteins have no sequence similarities, they do have similar overall structures (de Peredo et al., 2001 ).

Here we describe rirA (rhizobial iron regulator), another type of regulatory gene, mutations in which affect transcription of many genes in response to Fe availability. It was found in the bacterium Rhizobium leguminosarum, a nitrogen-fixing symbiont that nodulates roots of the legumes – peas, vetches clovers and beans. Bacteria known collectively as ‘rhizobia’ have great flexibility in their ability to use different Fe sources (Johnston et al., 2001 ). Thus, R. leguminosarum not only makes and imports vicibactin (VB), its native siderophore (Dilworth et al., 1998 ; Carson et al., 2000 ), but can use ferric citrate and, unusually for non-pathogenic bacteria, haem and haemoglobin (Carson et al., 2000 ; Noya et al., 1997 ; Wexler et al. 2001 ). Moreover, the genomic sequences of both Sinorhizobium meliloti (Galibert et al., 2001 ; http://sequence.toulouse.inra.fr/meliloti.html) and Mesorhizobium loti (Kaneko et al., 2000 ; http://www.kazusa.or.jp/rhizobase/), which respectively nodulate alfalfa and Lotus, have at least three operons, all of whose gene products resemble ABC transporters for inorganic Fe3+ uptake. In R. leguminosarum, transcription of one such operon, termed sfuABC, is enhanced by growth in low levels of Fe (J. D. Todd & M. Wexler, unpublished observations). As with haem uptake, such ‘Fbp-like’ transporters had hitherto been thought to be confined to bacterial pathogens. This wide array of Fe uptake systems in rhizobia may reflect their need to use many different Fe sources in the oligotrophic conditions of soils. It may also mean that, compared to E. coli, the rhizobia may have more complex, and/or different systems that regulate gene expression in response to Fe availability.

Homologues of Fur exist in different rhizobia, but it seems that their roles differ from that of Fur in ‘model’ bacteria, such as E. coli (Wexler et al., 2001 ; M. Wexler, unpublished observations). Thus, mutations in an R. leguminosarum gene with significant sequence similarity to fur of other bacteria (deLuca et al., 1998 ) had no detectable effect on the Fe-responsive repression of two operons, hmuPSTUV and orf1–tonB, both which are normally expressed only under low-Fe conditions and are required for the uptake of haem as sole Fe source (Wexler et al., 2001 ). The expression of several other R. leguminosarum operons whose products are involved in Fe uptake and whose transcription is depressed by high levels of Fe in the medium is also unaffected by mutations in this fur gene (M. Wexler, unpublished observations). In Bradyrhizobium japonicum, the symbiont of soybeans, fur is involved in regulating hemA, which specifies {delta}-aminolaevulinic acid synthase, the first dedicated step in haem biosynthesis. However, the fur mutants of this species did not have other features similar to those predicted for a fur mutant of (e.g.) E. coli (Hamza et al., 1999 ). Further, Nienaber et al. (2001) found that B. japonicum Fur is not involved in haem uptake.

Because the Fur homologue of R. leguminosarum does not appear to play a central role in Fe-responsive transcription in Rhizobium (M. Wexler, unpublished observations), we sought to identify other genes that controlled its response to Fe. Here, we describe one such regulatory gene, rirA.


   METHODS
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Media and bacterial growth conditions.
Strains and plasmids used in this study are shown in Table 1. Strains of E. coli and R. leguminosarum were grown routinely as described by Beringer (1974) . In high-Fe medium, FeCl3 (20 µM in liquid, 200 µM in agar) was added; low-Fe medium had no added Fe, but 2,2'-dipyridyl (20 µM in liquid, 150 µM in agar) was present. These were concentrations of this chelator that reduced Fe concentrations sufficiently to allow expression of Fe-responsive promoters but which did not excessively inhibit bacterial growth. Where appropriate, haem (10 µM) and ferric citrate (20 µM) were sole Fe sources. Peas were inoculated, grown and assayed for N2 fixation as described by Beynon et al. (1980) . Vicia sativa seedlings were grown on Perlite with N-free mineral salts liquid medium. Dry weights of the plant tops were measured and the roots inspected for nodules after 42 days in a greenhouse.


View this table:
[in this window]
[in a new window]
 
Table 1. Strains and plasmids used in this work

 
Assays.
ß-Galactosidase assays were done as described by Rossen et al. (1985) . Naringenin (1 µM) was the inducer for the nodC–lacZ fusion (Firmin et al., 1986 ). Qualitative chrome azural sulphonate (CAS) tests for siderophores (Schwyn & Neilands, 1987 ) and hydroxamate assays were done as described by Yeoman et al. (1997) . Green fluorescent protein (Gfp) was detected by inspecting colonies after 4 days’ growth on agar, lit with a hand-held 365 nm UV light source (UVP model UVL-21).

In vivo genetic manipulations.
Plasmids were conjugated into R. leguminosarum using the helper plasmid pRK2013 (Figurski & Helinski, 1979 ). Wild-type strain J251 was mutagenized with Tn5 by using it as a recipient in a conjugation with E. coli harbouring the ‘suicide’ plasmid pSUP2021, which contains Tn5. R. leguminosarum transconjugants, which arose at a frequency of about 10-6, each contain a genomic copy of Tn5 (Simon et al., 1989 ). Random insertions of Tnlac into pBIO1382 were obtained essentially as described by Wexler et al. (2001) . The population of mutated plasmids was conjugated into a rirA mutant that also contained a gfp reporter; insertions in rirA in pBIO1382 were identified by their failure to correct siderophore overproduction and deregulated expression of the gfp fusion on Fe-replete medium. Mutations were transferred from their plasmid-borne locations into the genome by marker exchange, using pPH1JI as the incompatible plasmid (Downie et al., 1983 ).

In vitro DNA manipulations.
Routine transformations, restriction digestions, ligations, Southern blotting and hybridization were done essentially as described by Downie et al. (1983) . R. leguminosarum genomic DNA was isolated using a Promega genomic preparation kit. Sequencing was done by the dideoxy chain-termination method, in some cases by M. W. G. Ltd, Germany. Data were analysed with the DNA-Star package. Searches of databases used BLAST in the EGCG package. Tnlac insertions into cosmids were precisely located by cycle-sequencing reactions at M. W. G. Ltd, using a primer that corresponded to the 5' end of the lacZ gene.

To construct a genomic library of wild-type R. leguminosarum J251, genomic DNA was partially digested with EcoRI, ligated into the cosmid pLAFR1 (Friedman et al., 1982 ), then packaged using Gigapack III XL Packaging extract (Stratagene). After packaging, approximately 105 primary transfectants were obtained. Ten of these were checked to confirm they had different inserts, each of the expected total size of approximately 25–30 kb.

RNA isolation and primer extensions were done as described by Sawers & Böck (1989) . RNA was obtained from R. leguminosarum containing pBIO1382 (to increase levels of rirA transcripts) with an RNAwiz (Ambion) kit. The primer (5'-CGTCATTGGCAGCACAATACATCAAC-3') corresponded to a region in the 5' end of the coding region of rirA. Plasmid pBIO1438 DNA was used as the template for the corresponding DNA sequencing reactions To identify the rpoI promoter, the primer 5'-CGTTGACCACGGCATTTCCGTCC-3' was used.


   RESULTS
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Isolation of mutants defective in regulating an Fe-responsive promoter
We had previously identified several operons that, in different ways, are involved in Fe uptake, and whose expression is enhanced by growth in Fe-deficient conditions (see below; Carter et al., 2002a ; Stevens et al., 1999 ; Wexler et al., 2001 ; Yeoman et al., 1999 , 2000 ). To isolate R. leguminosarum mutants that were defective in Fe-dependent regulation of one or more of these operons, we exploited a transcriptional fusion plasmid, pBIO1371 (M. Wexler, unpublished observations). This plasmid contains one such Fe-responsive operon, orf1–tonB, fused to a gfp reporter gene in the wide-host-range promoter-probe plasmid pOT2 (Allaway et al., 2001 ). As in other bacteria, TonB is a transmembrane energy-transducing protein that is required to import both haem and the siderophore VB as Fe sources (Wexler et al., 2001 ). The function of orf1 is unknown, but it is in the same transcript as tonB, the orf1–tonB operon being expressed at enhanced levels in Fe-depleted conditions (Wexler et al., 2001 ). Thus, when pBIO1371 is present in the wild-type R. leguminosarum strain J251, the colonies fluoresce under UV light, but only in conditions (i.e. low-Fe) in which the orf1–tonB promoter is active (M. Wexler, unpublished; Fig. 1).



View larger version (53K):
[in this window]
[in a new window]
 
Fig. 1. Effect of RirA on siderophore production and on a gfp fusion to orf1–tonB. Colonies of wild-type, two rirA mutants and an rirA mutant corrected by cloned DNA, each containing the gfp reporter pBIO1371, were picked to Fe-replete medium. The plates contained the siderophore indicator CAS (left) or were viewed in UV light for green fluorescent protein (right).

 
To isolate R. leguminosarum mutants that were defective in Fe-dependent repression of the orf1–tonB promoter, wild-type strain J251 was mutagenized with Tn5 (see Methods). Approximately 18000 random mutants were pooled, cultured in liquid medium, and were used, en masse, as a recipient into which was conjugated the Gfp reporter plasmid, pBIO1371. Of approximately 9000 such transconjugants, two, termed J394 and J396, fluoresced brightly under UV light even on high-Fe medium (Fig. 1). This suggests that they had mutations that deregulated the expression of the orf1–tonB promoter under conditions that would be repressive in the wild-type.

Since J394 and J396 were apparently deregulated for at least one operon involved in siderophore uptake, they were examined on CAS siderophore indicator agar. As seen in Fig. 1, the haloes made by these two mutants were significantly larger than that of wild-type J251. This is further evidence for the synthesis of more, and/or the import of less, siderophore than the wild-type. Quantitative hydroxamate assays that detect VB showed that under Fe-replete conditions mutant J394 made about threefold more of this siderophore than did the wild-type (data not shown).

Cloning and identification of the rirA gene
To clone and identify the gene mutated in these deregulated mutants, one of them, J394, was used as a recipient in a conjugational cross with an E. coli culture containing plasmids of a gene bank of DNA fragments (about 25 kb in size), derived from wild-type R. leguminosarum strain J251 and cloned in the vector pLAFR1 (see Methods). The pOT2-based reporter plasmid, pBIO1371, which was present in J394, is compatible with pLAFR1 so both plasmids could stably coexist. The ‘gene library’ plasmids were transferred, en masse, selecting for Tetr, specified by pLAFR1, and transconjugants were screened on high-Fe agar for any in which the Gfp reporter was repressed. Plasmids, termed pBIO1382 and pBIO1383, were isolated from two such colonies, then transformed into E. coli. Their restriction digests showed that they contained overlapping cloned DNA, including a 2·5 kb BamHI fragment, which was cloned into the wide-host-range plasmid pRK415, also cut with BamHI, to form pBIO1412. When introduced into mutants J394 and J396, pBIO1412 corrected them both for deregulated expression of the orf1–tonB:gfp fusion and for their large haloes on CAS plates (Fig. 1).

It was confirmed that this BamHI fragment corresponded to the one in which Tn5 was inserted in J394 and J396 by using it to probe Southern blots of BamHI-digested genomic DNA from these two mutants. The genomic DNA from strains J394 and J396 each generated two hybridizing bands, of appropriate sizes, consistent with insertion of Tn5 (which has one BamHI site) into this fragment; in contrast, a single 2·5 kb hybridizing fragment was seen with the wild-type DNA.

The BamHI fragment, subcloned from pBIO1382 into pUC18, was sequenced. It contained one intact gene, which we term rirA (rhizobial iron regulator) and an incomplete ORF, 3' of rirA, which was identical to dppA, the first gene of dppABCDF, an operon that specifies a dipeptide transporter and which, by chance, we had analysed earlier (Carter et al., 2002b ; see below). Further sequencing of this region showed that the deduced product of the gene 5' of rirA was similar to that of iolA, which specifies a semialdehyde decarboxylase.

We precisely located the location of Tn5 in one of the mutants, J394. A PCR was done, with one primer corresponding to the end of Tn5 and the other located upstream of the rirA coding region, and J394 genomic DNA as template. The sequence of the resultant fragment positioned Tn5 within rirA, 338 bp downstream of its proposed ATG start site.

To demonstrate further that rirA was important in controlling expression of orf1–tonB, cosmid pBIO1382 (see above), which contained the cloned rirA region, was mutagenized with the reporter transposon Tnlac (see Methods). One insertion abolished the ability of pBIO1382 to correct both the constitutive expression of orf1–tonB and the large-CAS-halo phenotype of the rirA mutants J394 and J396. The insert in this mutant plasmid, termed pBIO1439, was mapped by restriction digestions, then by sequencing the junction of Tnlac and the Rhizobium DNA. This mutation, rirA1:Tnlac, was located in the rirA coding region, with lacZ and rirA in the same orientation (see below).

The rirA1:Tnlac mutation was marker-exchanged into the genome of wild-type strain J251 to form mutant J388, the chromosomal mutation being confirmed by hybridization with a rirA probe to digests of the genomic DNA. Not unexpectedly, the homogenote produced a larger halo on CAS plates than wild-type. A gentamicin-sensitive derivative of J388 (termed J399) was obtained which had lost the gentamicin-resistance plasmid pPH1JI (used to ‘force’ the marker exchange). The original orf1–tonB:gfp fusion plasmid, pBIO1371, was then introduced into J399; as expected, Gfp fluorescence was actively expressed in both high- and low-Fe conditions.

Analysis of the rirA gene and its deduced polypeptide product
The deduced protein product (Mr 17441) of rirA has very close homologues in two other rhizobial species, Sinorhizobium meliloti and Mesorhizobium loti (respectively 77% and 84% identical; see Fig. 2). In the S. meliloti annotated genome (Galibert et al., 2001 ), the corresponding ORF (SMc00785) was originally termed aau3, since it was thought to specify a protein involved in acetoacetate utilization (Charles et al., 1997 ). However, aau3 is, in fact, elsewhere in the genome and SMc00785 has no known function (T. C. Charles, personal communication). Intriguingly, the product of the gene immediately 5' of the rirA homologue resembles a component of an ABC transporter involved in Fe3+ uptake. In M. loti, the RirA homologue is the product of the gene Mlr1147; in this species, the product of the adjacent gene is also involved in Fe uptake, in this case a predicted ABC transporter for haem uptake. A close homologue of RirA also occurs in Agrobacterium tumefaciens (88% identical) where the gene is also erroneously termed aau3, and, as in S. meliloti, is adjacent to a gene that is likely to specify an inorganic Fe3+ transporter. Rhizobium is closely related to Brucella, which also has a RirA-like protein (79% identity), of no known function; we noted that the corresponding gene in this pathogen is separated by two ORFs from a homologue of a bacterioferritin gene.



View larger version (64K):
[in this window]
[in a new window]
 
Fig. 2. Comparison of RirA and RirA-like sequences. In the upper group, R. leguminosarum RirA protein (R. leg, CAC35510) is aligned with very close homologues in Agrobacterium tumefaciens (A. tum, AAL41224), Sinorhizobium meliloti (S. mel, AAF06014), Brucella melitensis (B. mel, AAL53949) and Mesorhizobium loti (M. lot, BAB48587). The lower group shows comparisons with more distantly related gene products from Ralstonia solanacearum (R. sol, CAD16894), Salmonella typhimurium (S. typ, AAL23187), Rickettsia conorii (R. con, AAL03359), Streptomyces coelicolor (S. coe, CAB7646), Listeria monocytogenes (L. mon, CAC98714), Thermoanaerobacter tengcongensis (T. ten, AAM24467) and Desulfovibrio vulgaris Rrf2 (D. vul, AAA72001). Sequences were from GenBank. Amino acids that are totally conserved among the upper group of comparisons have asterisks above them. Residues that are also highly conserved in other, more distantly related bacteria, are indicated with filled triangles. The three conserved cysteines alluded to in the text are labelled ‘c’ below the sequence.

 
Apart from these near-relatives of Rhizobium, there are no other close homologues of RirA in other bacterial proteome sequences, even in other {alpha}-proteobacteria such as Rhodobacter and Caulobacter. However, a very wide range of other bacteria, from Streptomyces to Salmonella, encode proteins with lesser (about 40% identity) similarity to RirA. Of these, the only one with a proposed function is Rrf2 of Desulfovibrio vulgaris (31% identical). Rrf2 mutants constitutively express a high-Mr cytochrome–redox protein complex, suggesting that Rrf2 may be a transcriptional regulator (Keon et al., 1997 ; Rossi et al., 1993 ). Sporadic residues are very highly conserved throughout all these homologues, one notable feature being a triad of conserved cysteines towards their C-termini (see Fig. 2), perhaps suggestive of metal-binding ability. However, given the overall lack of sequence similarity among these more distantly related RirA homologues, it is not necessarily the case that these are orthologues of each other and/or RirA.

Mutations in rirA affect expression of several Fe-responsive operons
Having established that a mutation in rirA caused deregulated expression of the orf1–tonB operon, we wished to see if this potential regulatory gene affected the transcription of other Fe-responsive transcriptional units. To do this, we employed some previously constructed plasmids, each with a transcriptional fusion between a promoterless lacZ and the promoter of individual operons that are involved in one way or another with Fe uptake. These are described individually in more detail below (and see Tables 1 and 2). By assaying these lac fusions for ß-galactosidase, we could quantitate any effects of rirA on their expression. Each plasmid was transferred into wild-type J251 and into J397, which is derived from the rirA mutant strain J394, but which was cured of the original pOT2-basedfusion plasmid pBIO1371 (a precaution taken to avoid possible effects of the cloned orf1–tonB promoter). Transconjugants were grown in high and low levels of Fe, then assayed for ß-galactosidase. As shown in Table 2, the rirA mutation markedly affected expression of all of these lacZ fusions, as follows.


View this table:
[in this window]
[in a new window]
 
Table 2. Effect of rirA mutation on expression of Fe-responsive promoters in R. leguminosarum

 
rpoI. The rpoI gene specifies an ECF {sigma} factor that is very likely used to transcribe two operons, vbsGSO and vbsADL, each of which specify components of the VB biosynthetic pathway (Yeoman et al., 1999 ; Carter et al., 2002a ). Transcription of rpoI is elevated by growth in low levels of Fe, although there is significant expression in Fe-replete conditions (Yeoman et al., 1999 ; Table 2). It was found that expression of the rpoI::lacZ fusion (in pBIO1326) is enhanced (about fivefold in Fe-replete medium) in the rirA mutant, the response of rpoI transcription to Fe availability having been lost. Indeed, in the rirA mutant, rpoI expression is higher (about fourfold) than in the wild-type background, even in severely Fe-depleted conditions (Table 2).

vbsGSO and vbsADL. These two operons, which are transcribed divergently from each other, specify six of the proteins required for synthesis of the siderophore VB (Carter et al., 2002a ). In wild-type cells, in low-Fe conditions, their transcription is enhanced, at least fivefold, compared to when Fe is abundant. Their promoter regions differ in sequence from typical {sigma}70-type promoters, consistent with their being transcribed with RNA polymerase that likely contains the RpoI ECF {sigma} factor (Carter et al., 2002a ). Since transcription of rpoI is enhanced in the rirA mutant, it was not surprising that expression of the vbsGSO and vbsADL fusions (in plasmids pBIO1298 and pBIO1309) is also elevated in the rirA mutant (Table 2). In high-Fe medium, the expression of both these fusions is about sixfold higher in the rirA mutant than in the wild-type. However, in contrast to rpoI, there is still a response to Fe depletion, expression of vbsGSO and vbsADL being further elevated (about twofold) in low-Fe medium in the rirA mutant.

We know from earlier work (Carter et al., 2002a ; Yeoman et al., 2000 ; Stevens et al., 1999 ) that none of the other operons tested here requires rpoI for their transcription, so any effect of RirA on their expression cannot be explained in terms of a primary effect on rpoI transcription.

vbsC. This gene is thought to specify an acetylase involved in VB synthesis (Carter et al., 2002a ). Like vbsGSO and vbsADL, its expression is enhanced under low-Fe conditions. However, unlike these other two vbs operons, vbsC transcription does not require RpoI, although vbsC and rpoI are adjacent, with divergent transcript starts, separated by only 15 bp (K. H. Yeoman, unpublished). In the rirA mutant background, transcription of the vbsC–lacZ fusion (pBIO1306) is greatly elevated, and shows no evidence of Fe-responsive transcription. As with vbsGSO and vbsADL the activity of the vbsC–lacZ fusion in the rirA mutant is significantly higher than when it is present in the wild-type, even when grown in low-Fe medium (Table 2).

fhuA and fhuCDB. These two operons specify VB uptake proteins, their transcription being enhanced, more than fourfold, in Fe-depleted cells (Stevens et al., 1999 ; Yeoman et al., 2000 ). fhuA and fhuCDB also show marked differences in their expression patterns in the wild-type and in the rirA mutant, Fe-mediated regulation being lost in the rirA mutant. The expression from fhuA–lacZ and fhuCDB–lacZ (pBIO1125 and pBIO406 respectively) was much higher in J397 than in the wild-type, even when the latter was Fe-depleted (Table 2).

orf1–tonB and hmuPSTUV. Given that pBIO1371, which contains the orf1–tonBgfp fusion, was used to isolate the rirA mutants, it was not surprising that the corresponding lacZ fusion plasmid (pBIO1247) was also deregulated in the rirA mutant (Table 2). It was striking that not only was transcription of this fusion much higher in Fe-replete conditions in the rirA mutant than in the wild-type, but the levels were actually higher when Fe was abundant than when it was scarce. Such a switchover in relative levels of expression in the high- and low-Fe conditions is also seen with the fusion between hmuPSTUV and lacZ in plasmid pBIO1248. The hmuPSTUV operon encodes an ABC transporter for haem uptake, shows Fe-responsive transcription, and is transcribed divergently from orf1–tonB, the two transcript starts being only 11 bp apart (Wexler et al., 2001 ). This may be significant, given the ‘reversal’ of both these operons in their response to Fe availability when in the rirA mutant.

sfuABC. In addition to the above lacZ fusions, a transcriptional fusion plasmid, pBIO1378, contains gfp fused to the promoter of an R. leguminosarum operon (sfuABC) whose products resembled an ABC transporter of the Fbp-like family for Fe3+ uptake (M. Wexler, unpublished observations). Expression of this sfuABC operon is also enhanced in cells grown in Fe-depleted conditions, but this regulation is unaffected by mutations in the fur-like gene of R. leguminosarum (M. Wexler, unpublished observations). The sfuABC–gfp fusion plasmid was conjugated into the rirA mutant J397, and the transconjugants examined for fluorescence on high-and low-Fe agar. Compared to the wild-type, where fluorescence was seen only in Fe-deficient medium, in the rirA mutant there were very similar levels of Gfp fluorescence under both Fe regimes (not shown). Thus, expression of sfuABC also appears to be controlled by rirA.

fur and irr. R. leguminosarum has a homologue of the ‘global’, Fe-responsive regulator Fur (deLuca et al., 1998 ), but mutations in this gene do not affect transcription of any of the Fe-responsive genes described above (Wexler et al., 2001 ; M. Wexler, unpublished observations). Nevertheless, we examined the effects of the rirA mutation on fur transcription by conjugating pBIO939, a fur–lacZ fusion plasmid, into J397, measuring ß-galactosidase after growth in high- and low-Fe medium. As seen in Table 2, the overall levels of transcription were unaffected by the rirA mutation, expression of fur–lacZ being essentially the same in both media, just as in the wild-type (M. Wexler, unpublished observations).

R. leguminosarum, like other rhizobia, contains another gene, termed irr, which is another member of the fur superfamily. Found first in Bradyrhizobium japonicum (Hamza et al., 1998 ), irr specifies a regulator that has sequence similarity to Fur and represses hemB, whose product is involved in haem biosynthesis. In R. leguminosarum, too, Irr is involved in regulating haem biosynthesis and expression of irr itself was affected by Fe in the medium, but only to a small extent (M. Wexler, unpublished observations). An irr–lacZ fusion plasmid, pBIO1373, was conjugated into the wild-type J251 and into the rirA mutant J397. The levels of expression of this fusion were virtually doubled in the presence of Fe in the rirA mutant, compared to the wild-type (Table 2). In this sense, the irr–lacZ fusion behaves analogously to the orf1–tonB and hmuPSTUV fusions (see above), in that in the rirA mutant irr expression is greater in high-Fe than low-Fe conditions, whereas in the wild-type background, the reverse is the case.

hemA. As shown by M. Wexler (unpublished observations) transcription of the R. leguminosarum hemA gene (see above) is enhanced by growth of cells in high levels of Fe. A transcriptional hemA–lacZ fusion plasmid, pBIO1437, was mobilized into the wild-type and into the rirA mutant. As seen in Table 2, the Fe-enhanced expression of hemA was unaffected by the rirA mutation.

nodC. Expression of all the promoters above is affected by Fe availability, and all are related to iron nutrition. To see if RirA affected other types of regulated promoters in R. leguminosarum, we examined the effects of RirA on a nodulation (nodABCIJ) operon whose expression is enhanced by particular flavonoid molecules (Firmin et al., 1986 ). pIJ1477, a translational fusion plasmid containing R. leguminosarum nodC fused to lacZ, was conjugated into wild-type J251 and into the rirA mutant J397. Transconjugants were grown with and without the nod gene inducer naringenin. As expected, this flavanone enhanced expression of the fusion (about 20-fold), but this occurred to the same extent in the wild-type and the rirA mutant (Table 2).

Effect of rirA mutation on growth rates and on symbiotic N2 fixation
The effects of the rirA mutation on the regulation of genes involved in Fe uptake might be expected to affect the physiology of the mutant strains, particularly under Fe-stress conditions. Therefore, the growth rates of wild-type J251 and the rirA:Tn5 mutant J397 were compared in minimal media that varied with regard to Fe status. Cells were grown in normal, Fe-replete conditions, in media with haem or with ferric citrate as sole Fe sources, or in medium depleted of Fe by adding dipyridyl. The rirA mutant was slower-growing than wild-type on all these media (Fig. 3). Thus, although the rirA mutation increased doubling time, this was not exacerbated by Fe stress, since the relative reduction in growth rate of the rirA mutant was similar in Fe-replete and Fe-limited media.



View larger version (18K):
[in this window]
[in a new window]
 
Fig. 3. Effects of rirA mutation on growth rates of R. leguminosarum. Wild-type strain J251 ({blacktriangleup}) and rirA mutant J397 ({blacksquare}) were diluted into minimal media and were grown for 5 days, samples being taken and measured for their absorbance at 600 nm. Different media contained: (a) FeCl3 (20 µM), (b) ferric dicitrate (20 µM), (c) haem (10 µM) or (d) no added Fe but the chelator dipyridyl (20 µM) was present.

 
Strains J251 and the rirA mutant J397 were inoculated onto peas and the vetch Vicia sativa. On these two hosts, nodulation appeared to be unaffected under the conditions used. On both, nodules appeared at the same time with the two strains and the final numbers, gross morphology and N2-fixing ability were indistinguishable. On Vicia, the dry weights of plants inoculated by J394 were not significantly different from those inoculated by the wild-type.

Transcription of rirA is controlled by dual promoters and is subject to autoregulation
In the rirA1 mutant allele, lacZ and rirA are in the same orientation (see above), allowing expression of rirA to be monitored in different environments and genetic backgrounds. Given the apparent role of RirA in regulating Fe uptake, the effects of Fe on the expression of rirA itself were determined. To do this, plasmid pBIO1439, which contains the rirA1:Tnlac mutation, was mobilized into wild-type J251 and the transconjugants grown in Fe-replete or Fe-deficient minimal media. As seen in Table 2, ß-galactosidase activity was about twofold higher in Fe-replete medium, but, even in low-Fe conditions, there was significant expression from this rirA–lac fusion plasmid.

Two genes, rpoI and fur (see above), might potentially affect Fe-dependent regulation of rirA. However, neither rpoI nor fur regulated rirA expression, since ß-galactosidase activities from the rirA1:Tnlac fusion plasmid pBIO1439, in both high- and low-Fe conditions, were not altered in the rpoI mutant (J256) or fur mutant (J325) backgrounds, compared to wild-type (not shown).

To test if rirA affected its own transcription, pBIO1439 was mobilized into the rirA mutant J397, and the transconjugant was grown in high- and low-Fe media, then assayed for ß-galactosidase. Expression of rirA–lacZ was significantly enhanced in J397, compared to the wild-type background. This autoregulation was especially noticeable in cells grown in the high-Fe medium (Table 2).

The location of the rirA promoter(s) was determined by primer extension analysis. Total RNA was harvested from wild-type R. leguminosarum J251 containing cloned rirA in pBIO1382 (to enhance levels of rirA mRNA), the cells having been grown either in Fe-replete or Fe-deficient media. Two separate transcription start sites for rirA were discernible (Fig. 4). The more 5' one (TS1 in Fig. 4) is more pronounced with RNA isolated from cells grown in Fe-depleted medium. The 5' end of this transcript is located 76 bp upstream of the translational initiation codon of the rirA gene (Fig. 4). Upstream of the transcription start site there are the sequences TTGACG and CATAAG, putative -35 and -10 RNA polymerase recognition sequences, respectively (Fig. 4). The second, more 3', transcript start (TS2) is 51 bp downstream of TS1 and was increased about 1·5- to 2-fold in the presence of high Fe. The TS2 signal was much more intense than TS1 and therefore accounts for the overall increase in rirA–lacZ expression in the presence of Fe (see above). Unlike TS1, which has a single 5' end, three major 5' ends, spanning three contiguous GAA nucleotides, were observed for TS2 (Fig. 4). The TS2 site had TATATA and TTGTTC as putative -10 and -35 sequences (Fig. 4).



View larger version (44K):
[in this window]
[in a new window]
 
Fig. 4. Identification of transcript starts of the R. leguminosarum rirA gene. RNA isolated from R. leguminosarum grown with (+) and without (-) added Fe was used for primer extensions with an oligonucleotide within the rirA coding region. Two transcript start regions (TS1 and TS2) are indicated, together with the neighbouring sequences obtained from the sequencing reaction (see lanes G, A, T and C). The primer extension products at TS2 are shown in enlarged form. Below, the transcript starts are shown relative to the rirA ATG start site (at the 3' end of the sequence shown), the two sets of -10 and -35 sequences being underlined.

 
Cloned rirA suppresses another putative regulatory mutant of R. leguminosarum
We had noted that a ‘wild-type’ strain of R. leguminosarum, originally termed 8401pRL1JI (Downie et al., 1983 ), has several attributes of an rirA mutant (Wexler et al., 2001 ). It has a larger CAS halo than the genuine wild-type strain, J251, and transcription of several genes involved in VB synthesis (vbs genes), VB uptake (fhu genes) or haem uptake (hmu genes) occurs constitutively, even under Fe-replete growth conditions (Wexler et al. 2001 ; unpublished observations). We believe that this ‘rogue’ strain (hereafter called J4) acquired a spontaneous mutation early in its history that affects several characters associated with Fe uptake. It seemed, a priori, that J4 is mutated in rirA; the following observations showed that this is not the case.

As mentioned above, rirA is immediately upstream of the dppABCDF operon, which we had studied earlier (Carter et al., 2002b ). In that study, the source of the cloned dpp (and flanking) DNA was a cosmid, pBIO1068, which contained DNA cloned from strain J4. Sequencing of the cloned rirA gene in pBIO1068, plus DNA to at least 1 kb either side of it, revealed no differences compared to that of the rirA region of the bone fide strain J251. Further, the BamHI fragment that spanned rirA and which was equivalent to the one cloned in pBIO1412 from the bona fide wild-type J251 (see above), was subcloned from pBIO1068 into pLAFR1 to form pBIO1069. When pBIO1069 was conjugated into the rirA mutant J394, it corrected it for hyperexpression of the orf1–tonB:gfp fusion and for the large-CAS-halo phenotypes. To date, we have not succeeded in cloning or in transposon-tagging the mutated locus in strain J4.


   DISCUSSION
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
It is clear that Fe-responsive gene regulation in R. leguminosarum is significantly different from that in the widely studied enteric bacteria. Thus, the R. leguminosarum homologue of the ‘global’ regulator Fur, which in E. coli represses many genes involved in Fe uptake and metabolism, does not have an observable effect on the expression of many Fe-responsive operons that are intimately involved in Fe nutrition (Wexler et al., 2001 ; M. Wexler, unpublished observations). Similarly, Fur of the closely related Brucella abortus does not mediate Fe-dependent control of at least some genes involved in siderophore synthesis (M. Roop, personal communication). It thus seems likely that rhizobia have other transcriptional regulators that undertake at least some of the functions of Fur in E. coli and related bacteria.

Given the known phenotypes of rirA mutants, two, different, general models for its function may be considered. In one, RirA- mutants would, for unknown reasons, cause a reduction in the available level of Fe in the cells, which would respond by activating the array of genes responsible for Fe acquisition. A second general explanation is that RirA is directly involved in gene expression, possibly being a transcriptional regulator of genes involved in Fe uptake. Although we cannot discount the former explanation, we feel it is less likely. The main reason for this is that expression of some genes (e.g. rpoI, fhuCDB) in the rirA mutant background is very much higher than in the wild-type, even when it is subjected to severe Fe limitation. Such hyperexpression of target genes, seen in constitutive mutants, compared to when the low-Mr co-repressor molecule is absent is a characteristic of negatively regulated genes, recognized in the classic studies of Jacob and Monod on the lac operon.

Circumstantial evidence that RirA is a DNA-binding protein that may act as a transcriptional repressor comes from in silico studies in which molecular threading of the RirA N-terminal region, using the Fugue molecular modeller, suggests that this region has significant matches with known transcriptional regulators, despite there being no similarity in primary amino acid sequences. Interestingly, one of the closest matches is DtxR (a significant Z value >6·0 in the Fugue analysis), which, as mentioned above, is an Fe-dependent transcriptional repressor (A. Dore, personal communication).

If RirA is a transcriptional regulator, it might be expected that at least some operons whose expression is altered in rirA mutants would share some cis-acting regulatory sequences at or near their promoters. However, we found no extended conserved sequences in the promoter regions of all those operons whose expression was affected by rirA mutations. There was, though, a conserved 10mer, 5'-AAACTTGACT, which spanned the -35 sites, relative to the transcriptional starts of rpoI and hmuPSTUV (G. Sawers & K. H. Yeoman, unpublished observations). Biochemical studies with purified RirA protein will be required to establish if this represents a RirA-binding site. In any event, the lack of any other conserved sequence close to the promoters of other operons (e.g. fhuCDB) indicates that RirA-dependent gene control cannot be attributed to a simple interaction between the RirA protein and a conserved cis-acting sequence in the regulatory regions of all the Fe-responsive operons that are deregulated in rirA mutants.

Two sets of observations suggest the presence of other Fe-responsive transcriptional regulators in Rhizobium, in addition to RirA. Firstly, in the rirA mutant, transcription of the divergently transcribed hmuPSTUV and orf1–tonB operons is not only much enhanced in Fe-replete conditions, but their expression is actually enhanced when cells are grown in the presence of Fe. The Fe-dependent response of the irr gene also ‘switches’, being Fe-repressible in the wild-type but ‘Fe-inducible’ in the RirA mutant background. Thus, R. leguminosarum may have a positively acting Fe-responsive regulator whose presence is masked by the overriding, negative effects of RirA. Perhaps this hypothetical regulator only operates under particular (unknown) environmental conditions that overcome the influence of RirA. It will be of interest to know what these conditions are and to see if there is an interplay between positively and negatively acting Fe-responsive regulators in the nodulation process.

The second pointer that R. leguminosarum has other Fe-responsive regulators came from the serendipitous fact that strain J4, which had long been used in our laboratory (and in those of other researchers on R. leguminosarum) as ‘wild-type’ R. leguminosarum had, in fact, acquired a mutation early in its history which also affects Fe-responsive gene regulation. Although the phenotypes seen in J4 appear to be very similar to those of a rirA mutant, it was shown clearly that this strain is not defective in rirA. As yet the molecular basis of the defect in J4 is unknown.

Primer extensions and rirAlacZ fusions both showed that rirA transcription is enhanced about twofold in Fe-replete conditions. Interestingly, it appears that there are two promoters for rirA, with the weaker, upstream one being more active under low-Fe conditions, and the strong promoter being enhanced in response to Fe availability. It is currently not understood what the function of the weak low-Fe-inducible TS1 promoter is. Nevertheless, the overall effect is an increase in transcription in Fe-replete conditions, consistent with the lacZ expression data. This has adaptive sense, in that the greater the levels of Fe in the medium, the greater the expression of a gene whose product reduces expression of genes involved in Fe uptake. However, such a magnification of RirA-dependent repression is also counteracted, to some extent, since RirA is moderately autoregulatory. In this regard, RirA behaves like many other transcriptional repressors, including, for example, NodD of R. leguminosarum (Rossen et al., 1985 ). Comparison of the potentially cis-acting regulatory sequences around the two rirA promoters revealed no obvious consensus sequence. Note also that even in the rirA mutant background, rirA–lacZ expression was about twofold greater in cells grown in Fe-sufficient, compared to Fe-depleted media. Therefore, RirA itself does not mediate this Fe-dependent enhanced transcription of rirA and the mechanisms involved in this process are not known. Further, the Fe-responsive expression of the rirA–lacZ fusion pBIO1439 was unaffected when the reporter plasmid was present in R. leguminosarum strains with mutations in the potential Fe-dependent regulatory genes fur and irr. Therefore, regulation of rirA is unaffected by Irr or Fur proteins (data not shown).

To date, mutations in rirA affect expression of all the known R. leguminosarum promoters whose transcription is increased in low-Fe media, including that of irr, whose activity is only modestly enhanced in such conditions. This effect of rirA on irr may be indirect, perhaps through effects on intracellular Fe levels. It is clear that not all promoters are affected by rirA mutations. Thus, expression of fur is unaffected by rirA, as is the inducible nodC gene in response to nod inducer flavonoids. It will be of interest to make a more thorough inventory of those genes whose expression is altered by rirA mutations. The genome sequence of R. leguminosarum, currently in progress (http://www.sanger.ac.uk/Projects/R_leguminosarum/), will be very useful in addressing this question, since it will allow meaningful proteomic and transcriptomic comparisons to be made between wild-type and rirA mutant strains. Such post-genomic studies will also be of value in cataloguing the genes whose expression is affected by Fe availability and in establishing which of these transcriptional units RirA affects.

Given the numbers of genes whose expression is changed in the rirA mutants, it is perhaps surprising that they do not have more pronounced phenotypes. The rirA mutants were somewhat slower-growing in several media that varied in their Fe status, but this reduction could not be attributed to defects in Fe metabolism per se. In other bacteria, mutations in the analogous gene, fur, have different effects, depending on the particular genus. Thus, fur mutants of E. coli grow essentially normally in many growth media, but those of Vibrio anguillarum are believed to be non-viable (Tolmasky et al., 1994 ). Given that we do not know all the genes whose expression is affected by rirA mutations, it is not clear which particular targets contribute most to the slow growth of rirA mutants.

Whatever genes are affected by rirA, it is apparent that the deregulated expression of none of them leads to a major defect in symbiotic N2 fixation, at least under the conditions used in this study. At present, we do not know which are the major sources of Fe for rhizobia during nodule development and while fixing N2 as bacteroids in nodules (see Johnston et al., 2001 ). Mutations that abolish siderophore production or haem uptake have no major effects on the symbiosis (Carter et al., 2002a ; Lynch et al., 2001 ; Stevens et al., 1999 ; Wexler et al., 2001 ; Yeoman et al., 2000 ), and it is not clear if there are special, nodule-specific Fe uptake systems. If such transporters exist, perhaps they are regulated by mechanisms that do not involve RirA.

Lynch et al. (2001) identified a cluster of genes involved in the biosynthesis, uptake and regulation of a rhizobactin 1021, a siderophore made by some strains of Sinorhizobium meliloti. They showed that one of the structural genes, rhtA, which is required for siderophore synthesis is positively regulated by the product of rhrA, which has characteristics of an AraC-type transcriptional activator. It is not known if RhrA affects the Fe-responsive regulation of other genes in S. meliloti, or if its influence is confined to the nearby cluster of rht structural genes. However, the RhrA protein has no apparent similarity to RirA in terms of their sequences.

Close homologues of RirA occur in the deduced proteomes of near-relatives of Rhizobium, including the important pathogen Brucella. Interestingly, in Sinorhizobium, Agrobacterium and Mesorhizobium the rirA gene is adjacent to operons which, from their sequences, are likely to be involved in Fe uptake, and, in Brucella, a gene that appears to encode a ferritin is very closely linked to the rirA-like gene. It will be of interest to know if rirA mutations in these bacteria also have wide-ranging effects on Fe-dependent gene regulation. Excepting Rrf2 of Desulfovibrio, none of the more divergent homologues of RirA in, for example, E. coli has been predicted to be regulatory; indeed, none of these genes have been allocated any function. Although these more distant relatives may not be involved in Fe-responsive gene regulation, they might well have some regulatory roles. Certainly, the findings on RirA described here, and those of M. Wexler (unpublished observations) on the Fur protein of R. leguminosarum, add further evidence that E. coli and its closely related {gamma}-proteobacteria may not be ideal models for Fe-responsive gene control in other classes of Gram-negative bacteria.


   ACKNOWLEDGEMENTS
 
The work was funded by the UK BBSRC. We very grateful to Anja Hohl for experimental assistance, and to Marty Roop and Trevor Charles for providing unpublished data. We are especially grateful to Andrew Dore for very helpful advice.


   REFERENCES
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Allaway, D., Schofield, N. A., Leonard, M. E., Gilardoni, L., Finan, T. M. & Poole, P. S. (2001). Use of differential fluorescence induction and optical trapping to isolate environmentally induced genes. Environ Microbiol 3, 397-406.[Medline]

Archibald, F. (1983). Lactobacillus plantarum, an organism not requiring iron. FEMS Microbiol Lett 19, 29-32.

Beringer, J. E. (1974). R factor transfer in Rhizobium leguminosarum. J Gen Microbiol 84, 188-198.[Medline]

Beynon, J. L., Beringer, J. E. & Johnston, A. W. B. (1980). Plasmids and host-range in Rhizobium leguminosarum and Rhizobium phaseoli. J Gen Microbiol 120, 421-429.

Boyd, J., Oza, M. N. & Murphy, J. R. (1990). Molecular cloning and DNA-sequence analysis of a diphtheria toxin iron-dependent regulatory element (dtxR) from Corynebacterium diphtheriae. Proc Natl Acad Sci U S A 87, 5968-5972.[Abstract]

Braun, V., Hantke, K. & Koster, W. (1998). Bacterial iron transport: mechanisms, genetics, and regulation. Metal Ions Biol Syst 35, 67-145.

Carson, K. C., Meyer, J. M. & Dilworth, M. J. (2000). Hydroxamate siderophores of root nodule bacteria. Soil Biol Biochem 32, 11-21.

Carter, R. A., Worsley, P. S., Sawers, G. & 7 other authors (2002a). The vbs genes that direct synthesis of the siderophore vicibactin in Rhizobium leguminosarum: their expression in other genera requires ECF {sigma} factor RpoI. Mol Microbiol 44, 1153–1166.[Medline]

Carter, R. A., Yeoman, K. H., Klein, A., Hosie, A. H. F., Sawers, G., Poole, P. S. & Johnston, A. W. B. (2002b). dpp genes of Rhizobium leguminosarum specify uptake of {delta}-amino-levulinic acid. Mol Plant-Microbe Interact 15, 69-74.[Medline]

Charles, T. C., Cai, G. Q. & Aneja, P. (1997). Megaplasmid and chromosomal loci for the PHB degradation pathway in Rhizobium (Sinorhizobium) meliloti. Genetics 146, 1211-1220.[Abstract/Free Full Text]

deLuca, N. G., Wexler, M., Pereira, M. J., Yeoman, K. H. & Johnston, A. W. B. (1998). Is the fur gene of Rhizobium leguminosarum essential? FEMS Microbiol Lett 168, 288-295.

de Peredo, A. G., Saint-Pierre, C., Latour, J. M., Michaud-Soret, I. & Forest, E. (2001). Conformational changes of the ferric uptake regulation protein upon metal activation and DNA binding; first evidence of structural homologies with the diphtheria toxin repressor. J Mol Biol 310, 83-91.[Medline]

Dilworth, M. J., Carson, K. C., Giles, R. G. F., Byrne, L. T. & Glenn, A. R. (1998). Rhizobium leguminosarum bv. viciae produces a novel cyclic trihydroxamate siderophore, vicibactin. Microbiology 144, 781-791.

Downie, J. A., Hombrecher, G., Ma, Q. S., Knight, C. D., Wells, B. & Johnston, A. W. B. (1983). Cloned nodulation genes of Rhizobium leguminosarum determine host-range specificity. Mol Gen Genet 190, 359-365.

Dussurget, O., Rodriguez, M. & Smith, I. (1996). An ideR mutant of Mycobacterium smegmatis has derepressed siderophore production and an altered oxidative-stress response. Mol Microbiol 22, 535-544.[Medline]

Figurski, D. H. & Helinski, D. R. (1979). Replication of an origin-containing derivative of plasmid RK2 dependent on a plasmid function provided in trans. Proc Natl Acad Sci U S A 76, 1648-1652.[Abstract]

Firmin, J. L., Wilson, K. E., Rossen, L. & Johnston, A. W. B. (1986). Flavonoid activation of nodulation genes in Rhizobium reversed by other compounds present in plants. Nature 324, 90-92.

Friedman, A. M., Long, S. R., Buikema, W. J. & Ausubel, F. M. (1982). Construction of a broad host-range cosmid and its use in the genetic analysis of Rhizobium mutants. Gene 18, 289-296.[Medline]

Galibert, F., Finan, T. M., Long, S. R. & 52 other authors (2001). The composite genome of the legume symbiont Sinorhizobium meliloti. Science 293, 668–672.

Hamza, I., Chauhan, S., Hassett, R. & O’Brian, M. R. (1998). The bacterial Irr protein is required for coordination of heme biosynthesis with iron availability. J Biol Chem 273, 21669-21674.[Abstract/Free Full Text]

Hamza, I., Hassett, R. & O’Brian, M. R. (1999). Identification of a functional fur gene in Bradyrhizobium japonicum. J Bacteriol 181, 5843-5846.[Abstract/Free Full Text]

Hantke, K. (1981). Regulation of ferric uptake iron transport in Escherichia coli K12 – isolation of a constitutive mutant. Mol Gen Genet 182, 288-292.[Medline]

Hantke, K. (2001). Iron and metal regulation in bacteria. Curr Opin Microbiol 4, 172-177.[Medline]

Hirsch, P. R. & Beringer, J. E. (1984). A physical map of pPH1JI and pJB4JI. Plasmid, 12, 139-141.[Medline]

Johnston, A. W. B., Wexler, M. & Yeoman, K. H. (2001). Metals and the rhizobial-legume symbiosis – uptake, utilization and signalling. Adv Microb Physiol 45, 113-156.[Medline]

Kaneko, T., Nakamura, Y., Sato, S. & 19 other authors (2000). Complete genome structure of the nitrogen-fixing symbiotic bacterium Mesorhizobium loti. DNA Res 7, 331–338.

Keen, N. T., Tamaki, S., Kobayashi, D. & Trollinger, D. (1988). Improved broad-host range plasmids for DNA cloning in Gram-negative bacteria. Gene 70, 191-197.[Medline]

Keon, R. G., Fu, R. D. & Voordouw, G. (1997). Deletion of two downstream genes alters expression of the hmc operon of Desulfovibrio vulgaris subsp. vulgaris Hildenborough. Arch Microbiol 167, 376-383.[Medline]

Lynch, D., O’Brien, J., Welch, T., Clarke, P., Cuiv, P. O., Crosa, J. H. & O’Connell, M. (2001). Genetic organization of the region encoding regulation, biosynthesis, and transport of rhizobactin 1021, a siderophore produced by Sinorhizobium meliloti. J. Bacteriol 183, 2576-2585.[Abstract/Free Full Text]

Masse, E. & Gottesman, S. (2002). A small RNA regulates the expression of genes involved in iron metabolism in Escherichia coli. Proc Natl Acad Sci U S A 99, 4620-4625.[Abstract/Free Full Text]

Messing, J., Crea, R. & Seeburg, P. H. (1983). A system for shotgun DNA sequencing Nucleic Acids Res 9, 309-314.[Abstract]

Nienaber, A., Hennecke, H. & Fischer, H. M. (2001). Discovery of a haem acquisition system in the soil bacterium Bradyrhizobium japonicum. Mol Micriobiol 41, 787-800.

Noya, F., Arias, A. & Fabiano, E. (1997). Heme compounds as iron sources for nonpathogenic Rhizobium species. J Bacteriol 179, 3076-3078.[Abstract]

Ochsner, U. A. & Vasil, M. L. (1996). Gene repression by the ferric uptake regulator in Pseudomonas aeruginosa: cycle selection of iron-regulated genes. Proc Natl Acad Sci U S A 93, 4409-4414.[Abstract/Free Full Text]

Panina, E. M., Mironov, A. A. & Gelfand, M. S. (2001). Comparative analysis of FUR regulons in gamma-proteobacteria. Nucleic Acids Res 29, 5195-5206.[Abstract/Free Full Text]

Posey, J. E. & Gherardini, F. C. (2000). Lack of a role for iron in the Lyme disease pathogen. Science 288, 1651-1653.[Abstract/Free Full Text]

Rossen, L., Shearman, C. A., Johnston, A. W. B. & Downie, J. A. (1985). The nodD gene of Rhizobium leguminosarum is autoregulatory and in the presence of plant root exudate induces the nodABC genes. EMBO J 4, 3369-3374.

Rossi, M., Pollock, W. B. R., Reij, M. W., Fu, R. D. & Voordouw, G. (1993). The hmc operon of Desulfovibrio vulgaris subsp. vulgaris Hildenborough encodes a potential transmembrane redox protein complex. J Bacteriol 175, 4699-4711.[Abstract]

Sawers, G. & Böck, A. (1989). Novel transcriptional control of the pyruvate formate-lyase gene: upstream regulatory sequences and multiple promoters regulate anaerobic expression. J Bacteriol 171, 2485-2498.[Medline]

Schwyn, B. & Neilands, J. B. (1987). Universal chemical assay for the detection and determination of siderophores. Anal Biochem 160, 47-56.[Medline]

Simon, R., Quandt, J. & Klipp, W. (1989). New derivatives of transposon Tn5 suitable for mobilisation of replicons, generation of operon fusions and induction of genes in Gram negative bacteria. Gene 80, 161-169.[Medline]

Spaink, H. P., Okker, R. J. H., Wijffelman, C. A., Pees, E. & Lugtenberg, B. J. J. (1987). Promoters in the nodulation region of the Rhizobium leguminosarum symbiotic plasmid pRL1JI. Plant Mol Biol 9, 27-39.

Stevens, J. B., Carter, R. A., Hussain, H., Carson, K. C., Dilworth, M. J. & Johnston, A. W. B. (1999). The fhu genes of Rhizobium leguminosarum specifying siderophore uptake proteins: fhuCDB are adjacent to a pseudogene version of fhuA. Microbiology 145, 593-601.[Abstract]

Tolmasky, M. E., Wertheimer, A. M., Actis, L. A. & Crosa, J. H. (1994). Characterization of the Vibrio anguillarum fur gene – role in regulation of expression of the FatA outer-membrane protein and catechols. J Bacteriol 176, 213-220.[Abstract]

Vieira, J. & Messing, J. (1982). The pUC plasmids, an M13mp7-derived system for insertion mutagenesis and sequencing with synthetic universal primers. Gene 19, 259-268.[Medline]

Wexler, M., Yeoman, K. H., Stevens, J. B., de Luca, N. G., Sawers, G. & Johnston, A. W. B. (2001). The Rhizobium leguminosarum tonB gene is required for the uptake of siderophore and haem as sources of iron. Mol Microbiol 41, 801-816.[Medline]

Wood, W. B. (1966). Host specificity of DNA produced by Escherichia coli; bacterial mutations affecting the restriction and modification of DNA. Mol Biol 16, 118-133.

Yeoman, K. H., Delgado, M. J., Wexler, M., Downie, J. A. & Johnston, A. W. B. (1997). High affinity iron acquisition in Rhizobium leguminosarum requires the cycHJKL operon and the feuPQ gene products, which belong to the family of two-component regulators. Microbiology 143, 127-134.[Abstract]

Yeoman, K. H., May, A. G., DeLuca, N. G., Stuckey, D. B. & Johnston, A. W. B. (1999). A putative ECF {sigma} factor gene, rpoI, regulates siderophore production in Rhizobium leguminosarum. Mol Plant-Microbe Interact 12, 994-999.[Medline]

Yeoman, K. H, Wisniewski-Dye, F., Timony, C., Stevens, J. B., deLuca, N. G., Downie, J. A. & Johnston, A. W. B. (2000). Analysis of the Rhizobium leguminosarum siderophore-uptake gene fhuA: differential expression in free-living bacteria and nitrogen-fixing bacteroids and distribution of an fhuA pseudogene in different strains. Microbiology 146, 829-837.[Abstract/Free Full Text]

Received 10 July 2002; revised 3 September 2002; accepted 4 September 2002.