|Home | About | Journals | Submit | Contact Us | Français|
Burkholderia cenocepacia mutants that fail to produce the siderophore ornibactin were obtained following mutagenesis with mini-Tn5Tp. These mutants were shown to be growth restricted under conditions of iron depletion. In eight of the mutants, the transposon had integrated into one of two genes, orbI and orbJ, encoding nonribosomal peptide synthetases. In the other mutant, the transposon had inserted into an open reading frame, orbS, located upstream from orbI. The polypeptide product of orbS exhibits a high degree of similarity to the Pseudomonas aeruginosa extracytoplasmic function (ECF) σ factor PvdS but possesses an N-terminal extension of approximately 29 amino acids that is not present in PvdS. Three predicted OrbS-dependent promoters were identified within the ornibactin gene cluster, based on their similarity to PvdS-dependent promoters. The iron-regulated activity of these promoters was shown to require OrbS. Transcription of the orbS gene was found to be under the control of an iron-regulated σ70-dependent promoter. This promoter, but not the OrbS-dependent promoters, was shown to be a target for repression by the global regulator Fur. Our results demonstrate that production of ornibactin by B. cenocepacia in response to iron starvation requires transcription of an operon that is dependent on the Fur-regulated ECF σ factor gene orbS. A mechanism is also proposed for the biosynthesis of ornibactin.
The genus Burkholderia belongs to the β-subgroup of the proteobacteria and includes a number of bacterial species that are pathogenic to animals and/or plants (13, 44, 78). One group, the Burkholderia cepacia complex (BCC), is noted for their ability to cause opportunistic infections in humans, particularly in patients with cystic fibrosis (CF) (12, 41, 73). BCC infections of the CF lung can result in a rapidly fatal necrotizing pneumonia, and treatment of infections is problematic due to the high intrinsic resistance of BCC isolates to many antimicrobials (1, 68, 73). The BCC contains at least nine closely related species, of which B. cenocepacia is the most prevalent in both CF and non-CF infections (12, 43, 60, 79, 80). The complete genome sequence has recently been determined for a strain of B. cenocepacia that caused an epidemic among CF patients in Canada and the United Kingdom (http://www.sanger.ac.uk/Projects/B_cenocepacia/).
Colonization of the CF lung by bacterial pathogens requires the expression of high-affinity iron uptake systems due to the presence of the iron-binding protein lactoferrin and the iron-sequestering property of CF respiratory mucus (24, 87). A commonly employed mechanism for iron acquisition by bacteria is the secretion of low-molecular-weight Fe(III)-binding compounds known as siderophores, which are then transported into the bacterium via specific outer membrane receptors (7). Most clinical isolates of B. cenocepacia produce the siderophores ornibactin and pyochelin under iron-limiting conditions (15, 51, 69, 83), and production of both siderophores has been correlated with morbidity and mortality in CF patients and/or shown to contribute to pathology in animal models of respiratory infection (69-72, 86).
Ornibactin is a tetrapeptide siderophore with an l-ornithine-d-hydroxyaspartate-l-serine-l-ornithine backbone (74) (Fig. (Fig.1A).1A). The N-terminal ornithine residue is modified at the δ-amino group (N5) by hydroxylation and the addition of a β-hydroxy acid (of variable chain length) via an amide linkage. The C-terminal ornithine is hydroxylated and formylated on the δ-amino group and amidated on the α-carboxyl with putrescine (1,4-diaminobutane). The modifications to the δ-amino groups of the two ornithine residues result in the formation of hydroxamate groups which, together with the side chain hydroxyl and carboxyl groups of the aspartate residue, form a hexadentate complex with Fe(III) possessing a stoichiometry of 1:1 (74). The modification of the side chain amino groups of two ornithine residues to produce iron-chelating hydroxamate groups is analogous to the situation in some of the pyoverdine siderophores produced by the fluorescent pseudomonads (50).
The production of ornibactin by members of the BCC has been shown to be regulated by iron availability and is completely inhibited by the presence of ≥15 μM ferric iron in the medium (51). Transposon mutagenesis studies have facilitated the identification of several genes that are required for ornibactin biosynthesis and transport in B. cenocepacia (70, 71). In one ornibactin-negative mutant, the transposon had inserted into a gene which is orthologous with the pvdA gene of Pseudomonas aeruginosa. The pvdA gene encodes l-ornithine N5-oxygenase, which serves to hydroxylate the δ-amino group of ornithine as required for the biosynthesis of type I pyoverdine (84). Downstream of the B. cenocepacia pvdA gene, two other genes were identified: orbA, encoding the 78-kDa ferric-ornibactin receptor, followed by a gene which is homologous to the P. aeruginosa pvdF gene, encoding N5-hydroxyornithine transformylase (45, 70). Mutation of orbA was shown to inhibit the uptake of ferric ornibactin.
It has been shown that the B. cenocepacia pvdA gene is induced during growth under low-iron conditions (71). However, the mechanism of iron-dependent regulation of this gene and the possible role of the iron-dependent repressor, Fur, were not investigated. Ornibactin synthesis was also shown to be negatively regulated by the B. cenocepacia CepR-CepI quorum-sensing system. Inactivation of cepR or cepI resulted in less than a twofold increase in production of ornibactin under iron-limiting and iron-replete conditions. Consistent with these observations, transcription of the B. cenocepacia pvdA gene was shown to be higher in a cepR mutant during stationary phase regardless of whether cells were grown in low- or high-iron medium (39, 40). In contrast to the situation with pyochelin, it was observed that the ability to utilize ornibactin is not a prerequisite for production of ornibactin and that synthesis of the OrbA receptor is not dependent on the presence of exogenous ornibactin (70).
As part of our investigation into iron-regulated gene expression in B. cenocepacia, we have used a mutagenesis approach to isolate mutants that are unable to produce ornibactin. Characterization of the mutants, together with the availability of the complete genome sequence for a B. cenocepacia strain, has allowed us to identify a genetic locus containing the genes specifically required for assembly and utilization of this siderophore, and a model is presented for the metabolic steps that occur during ornibactin synthesis. In addition, we have identified the promoters required for transcription of the ornibactin genes, and we demonstrate that the activity of these promoters is dependent upon a linked ECF σ factor gene which is, in turn, subject to regulation by the global iron regulator Fur.
All bacterial strains used are shown in Table Table1.1. B. cenocepacia strains were cultured at 37°C on M9 minimal salts agar (11) containing glucose (0.5%) as the carbon source and subsequently maintained on this medium at room temperature (~20°C). Escherichia coli strains were cultured at 37°C on MacConkey agar or on Luria-Bertani (LB) agar containing appropriate antibiotics (Iso-Sensitest agar [Oxoid] was used when the antibiotic was trimethoprim) and were maintained at 4°C for up to 3 months. Where indicated, Casamino Acids (CAA; Difco) were added to M9-glucose medium to a final concentration of 0.1% (i.e., M9-CAA medium). Antibiotics were used at the following concentrations: ampicillin, 100 μg/ml (E. coli); trimethoprim, 25 to 50 μg/ml (E. coli and B. cenocepacia as indicated); chloramphenicol, 25 μg/ml (E. coli) and 50 μg/ml on LB medium and in M9-CAA medium, 100 μg/ml on M9-glucose agar (B. cenocepacia). For strains carrying lacZ fusions, 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside was included in solid medium at 40 μg/ml. For measurement of the effect of iron availability on growth rates and/or promoter activity, B. cenocepacia strains were grown to stationary phase for ~20 h at 37°C in M9-CAA medium containing appropriate antibiotics (50 μg/ml chloramphenicol was used for maintaining pKAGd4 derivatives in these experiments). Cells from these cultures were inoculated at 100-fold dilution into fresh medium (5 ml for β-galactosidase assays and 40 ml for growth rate determinations) containing either 2,2′-dipyridyl (100 μM for B. cenocepacia) to generate iron starvation conditions or 50 μM ferric chloride for iron-replete conditions and were grown in flasks at 37°C with vigorous shaking. For E. coli, such experiments were performed using LB broth containing 175 μM dipyridyl or 50 μM ferric chloride. Chrome azurol S (CAS) agar was made by adding 10 ml of CAS reagent (65) to 90 ml Y minimal agar (0.169% l-glutamic acid, 0.3% Tris, 0.01% MgSO4 · 7H2O, 0.022% CaCl2 · 6H2O, 0.022% K2HPO4 · 3H2O, 1.5% agar [pH 6.8]). EDDHA plates were made by adding the ferric iron chelator ethylenediamine di(o-hydroxyphenyl acetic acid) to M9-CAA agar at a final concentration of 35 μg/ml.
All plasmids and transposons used are shown in Table Table1.1. Primers used for plasmid constructions are shown in the Table S1 in the supplemental material. Mini-Tn5Tp (0.74 kb) is carried on the suicide plasmid pUT (18, 28) and was constructed by replacing the 3.7-kb Ω-Cm interposon of pUTmini-Tn5Cm with the 0.64-kb SmaI fragment from p34E-Tp, containing the dfrB2 (Tpr) gene originating from plasmid R388. The absence of the transcription and translation terminators that border the Ω-Cm interposon (23) means that transposon insertion will not necessarily lead to polar effects on downstream gene expression. The broad-host-range, mobilizable transcription fusion vector pKAGd4 was constructed by replacing the HindIII-NotI fragment of pPR9TT, containing the lacZ gene, with the 3.34-kb HindIII-DraI fragment from pUC18NotlacZYA, containing the entire lacZ gene preceded by a Shine-Dalgarno sequence. This manipulation results in loss of the unique SmaI site together with the EcoRI and EcoRV sites from the pPR9TT multiple cloning site (MCS) and creation of unique BamHI, SphI, and XbaI sites upstream of the lacZ ribosome-binding site. pKAGd4ΔAp was constructed by inserting the 1.22-kb SmaI kanamycin resistance cassette from p34E-Km (E. H. Bull and M. S. Thomas, unpublished data) into the ScaI site within the bla (Apr) gene of pKAGd4. B. cenocepacia genomic DNA for amplification of ornibactin genes and promoters was obtained by using the GenomeStar system (Hybaid GmbH) or by boiling a colony for 4 min in 200 μl Tris-EDTA buffer. DNA was routinely amplified by hot start PCR in the presence of 5% dimethyl sulfoxide (DMSO), Pfu Turbo (Stratagene), or Accuzyme (Bioline). When a cell lysate was used as the source of template DNA, the proofreading DNA polymerase was used in combination with Taq DNA polymerase. The B. cenocepacia orbS gene, together with its promoter, was amplified from 715j genomic DNA as a 969-bp HindIII-BamHI fragment with primers orbSfor and orbSrev2 and cloned between the corresponding sites of pBBR1MCS, resulting in pBBR1MCS-orbS. pSHAFT-orbS::Tp was constructed by first inactivating the orbS gene on pBBR1MCS-orbS by insertion of the dfrB2 (Tpr) cassette from p34E-Tp into the EcoRI site within orbS. The inactivated orbS gene was then transferred as a SalI-NotI fragment from pBBR1MCS-orbS::Tp to the corresponding sites of the mobilizable suicide vector pSHAFT. To construct a pBBR1MCS-orbS derivative in which a translation stop codon was substituted for the first potential translation initiation codon, the 5′ end of orbS and upstream DNA were amplified with primers OrbSfor and OrbSstart-stop, and the resultant fragment was used to replace the HindII-NcoI fragment of pBBR1MCS-orbS. To construct derivatives in which the second potential start codon was replaced by a stop codon or an alanine codon and the third potential start codon was replaced by a stop codon or analternative valine codon, SOE mutagenesis was employed. The 5′ end of orbSand upstream DNA were amplified with OrbSfor in combination with OrbSstart2-stop, OrbSstart2-ala, OrbSstart3-stop, and OrbSstart3-val, and the products were used as megaprimers in a second PCR with OrbSrev2. The products of the second amplification were cloned between the HindIII and BamHI sites of pBBR1MCS.
DNA fragments containing the orbH (415 bp), orbE (467 bp), orbI (467 bp), and full-length orbS (359 bp) promoters were amplified by PCR using primers pmbtHfor and pmbtHrev, ppvdEfor and ppvdErev, pnrpSfor and pnrpSrev, and OrbSfor4 and porbSrev, respectively, and ligated into pBluescript II KS as EcoRI-HindIII fragments. The cloned promoter fragments were then transferred from pBluescript II KS as BamHI-HindIII fragments into the BglII and HindIII sites of pKAGd4. orbS promoter derivatives with upstream endpoints at −69, −40, and +5 relative to the transcription start point were amplified with primers altpromfor, candpromfor, and OrbSfor3 in combination with the downstream porbS primer pOrbSrev and ligated between the HindIII and XbaI sites of pBluescript II KS and pKAGd4. The integrity of all cloned DNA fragments arising from PCR amplification was confirmed by DNA sequencing.
pUTmini-Tn5Tp was introduced into B. cenocepacia KLF1 by conjugal transfer using the Escherichia coli donor strain BW19851 as described previously (18, 28), and transposon mutants (arising at a frequency of 5.12 × 10−4 per recipient) were selected on CAS agar containing trimethoprim (40 μg/ml) and kanamycin (25 μg/ml). Following incubation at 37°C for 2 days, candidate ornibactin-deficient mutants were identified by the absence of an orange zone around the colony.
pSHAFT-orbS::Tp was mobilized from E. coli S17-1(λpir) into B. cenocepacia strain 715j, and recombinants were selected on M9-glucose agar containing trimethoprim and kanamycin (50 μg/ml each). Candidate double-crossover recombinants, in which vector sequences and the wild-type copy of orbS were lost, were identified by virtue of their sensitivity to 50 μg/ml chloramphenicol. The integrity of candidate orbS::Tp mutants was confirmed by two separate PCRs using primer pairs orbSfor with orbSrev and orbSfor5 with orbSrev5.
The analysis of pyochelin production by bacteria growing in liquid culture was carried out as described by Farmer and Thomas (22). To analyze ornibactin production, overnight cultures grown in M9-glucose medium were used to inoculate 3 ml of the same medium at a 100-fold dilution. Following 36 h of growth with aeration at 37°C, 1 ml of the resultant culture was centrifuged at 16,000 × g for 10 min, whereupon the supernatant was recovered and sterilized by passing through a 0.22-μm filter. Twenty microliters of the supernatant was subjected to isoelectric focusing in an ampholine PAG plate polyacrylamide gel (pH 3.5 to 9.5; Amersham Pharmacia Biotech), and the presence of ornibactin was analyzed using the CAS overlay method (33). Production of ornibactin is indicated by the presence of a single orange band migrating near the cathode.
B. cenocepacia genomic DNA was purified using the PureGene genomic DNA isolation kit (Gentra Systems, Inc.) or the GenomeStar system (Hybaid GmbH) as recommended by the manufacturers. Following overnight digestion with PstI (which does not cut within mini-Tn5Tp) and SalI (which cuts adjacent to the O end of the transposon), the DNA was fractionated in an 0.8% agarose gel and transferred to Hybond-N+ nylon membrane (Amersham Pharmacia Biotech) by Southern blotting (68), following which hybridization was carried out under normal stringency conditions using probe DNA labeled by using the ECL direct nucleic acid labeling and detection system (Amersham Pharmacia Biotech). The probe DNA used contained the dfrB2 (Tpr) gene from p34E-Tp. Genomic DNA fragments corresponding in size to the fragment hybridizing to the probe were eluted from gel slices, ligated to plasmid pUC18 or pBBR1MCS, and used to transform E. coli NM522 or MC1061. Selection of transformants was made on Iso-Sensitest agar (Oxoid, Basingstoke, United Kingdom) containing ampicillin (100 μg/ml) and trimethoprim (20 μg/ml). The sites of transposon insertion were then identified by DNA sequencing.
B. cenocepacia 715j was diluted from an overnight culture into 10 ml of M9-glucose medium containing Casamino Acids (0.1%) and grown to an optical density at 600 nm (OD600) of ~0.5 in the presence of FeCl3 (50 μM) or 2,2′-dipyridyl (100 μM), whereupon RNA was extracted using the Ambion RiboPure kit according to the manufacturer's instructions. Reverse transcription-PCR (RT-PCR) was carried out using the Access RT-PCR kit (Promega) according to the manufacturer's instructions, but with the addition of 2.5 μl of DMSO. Primer pairs are listed in Table S1 of the supplemental material. For all the pairs of primers used, products were not obtained when reverse transcriptase was omitted from the reaction mixture. However, all pairs of primers were shown to amplify fragments of the expected size from B. cenocepacia genomic DNA (results not shown).
B. cenocepacia cultures were grown in triplicate to an OD600 of 0.5 to 0.8 as described above, whereupon the cells were chilled on ice for at least 15 min (E. coli cultures were grown to an OD600 of 0.4). Duplicate assays were performed at 30°C on 25 to 200 μl of cells from each culture in a total volume of 1 ml following permeabilization of the cells with chloroform-sodium dodecyl sulfate (52). Values presented (in Miller units) are background corrected by subtraction of the activity measured in cells harboring the parental plasmid, pKAGd4, grown under identical conditions.
The Fur titration assay (FURTA) (75) was carried out by streaking overnight cultures of the E. coli fhuF-lacZ strain, H1717, transformed with p3ZFBS, pBluescript II KS (pBS), or pBS containing different orbS promoter fragments onto MacConkey-lactose agar supplemented with ampicillin and 40 μM Fe(NH4)2(SO4)2. Plates were incubated overnight at 37°C.
DNA fragments containing different orbS promoter derivatives, the orbH promoter, and both the orbE and orbI promoters were PCR amplified with Taq DNA polymerase (York Bio) in the presence of DMSO (5%) using the same primer pairs used to clone promoter fragments into pBluescript II KS for the Fur titration assay. The DNA products were then digested with HindIII and purified by elution following electrophoresis in a 6% polyacrylamide gel. The DNA was then end labeled with 20 μCi [α-32P]dATP (3,000 Ci/mmol) using Klenow fragment of DNA polymerase I (61). DNA (0.4 nM) was incubated for 30 min at 37°C with purified E. coli Fur in 1× binding buffer (10 mM bis-Tris-borate [pH 7.5], 40 mM KCl, 1 mM MgCl2, 0.1 mM MnCl2), 20 μg/ml sonicated calf thymus DNA, and glycerol (10%). Gels were loaded under tension, and electrophoresis was carried out at 4°C in a 6% polyacrylamide gel in the presence of 40 mM bis-Tris-borate (pH 7.5) and 0.1 mM MnCl2 in both the gel and the running buffer. DNA fragment mobility was visualized using a phosphorimager.
The sequence of mini-Tn5Tp has been deposited in GenBank and has been assigned accession number DQ266302. The complete sequences of orbS and orbE and partial sequences of orbB, orbI, and ubiC of strain 715j have been deposited in GenBank and have been assigned accession numbers DQ279459, DQ279460, and DQ269988.
To generate mutants of B. cenocepacia that do not produce ornibactin (Orb− phenotype), we employed a spontaneous pyochelin-negative (Pch−) derivative, KLF1, of the Orb+ Pch+ strain 715j (15, 21). Following mutagenesis of this strain with a newly constructed transposon, mini-Tn5Tp (see Materials and Methods), ornibactin mutants (OM) were selected by plating the cells directly onto CAS agar plates containing trimethoprim and kanamycin. Of a total of 8,600 trimethoprim-resistant mutants, arising from two independent experiments, 9 were identified in which ornibactin production was abolished, as judged by the absence of an orange halo surrounding the colony. Another mutant, OM14, exhibited a reduction in ornibactin production and was retained for further investigation (Fig. (Fig.2A).2A). Confirmation that ornibactin production is abolished in all of these mutants was obtained by analysis of supernatants from iron-starved cultures using the isoelectric focusing (IEF)-CAS overlay assay. As an example, analysis of ornibactin production by OM3 is shown in Fig. Fig.2B2B.
Southern analysis of PstI-digested genomic DNA revealed that single mini-Tn5Tp insertions had occurred at unique locations in each of the 10 mutants (results not shown). Genomic PstI or SalI fragments of 1.2 to 4.7 kb containing the integrated minitransposon from all the mutants were cloned into pBBR1MCS or pUC18, and the sequence of the cloned genomic DNA was determined.
Using the completed genome sequence of B. cenocepacia J2315, it was found that for eight of the mutants (OM1, OM4, OM6, OM11, OM12, OM13, OM15, and OM17), the transposon had inserted into either of two tandemly organized genes which are similar to nonribosomal peptide synthetase (NRPS) genes required for the biosynthesis of pyoverdines by pseudomonads. In particular, in OM6 and OM12, the transposon had inserted into an open reading frame (ORF) of 1,670 codons in length, orbJ, encoding a polypeptide homologous to the products of the P. aeruginosa pvdD (locus tag PA2399) and pvdJ (PA2400/1) genes. In the other six mutants, the transposon had integrated into a 3,223-codon ORF, orbI, located immediately upstream of orbJ (Fig. (Fig.3A).3A). The product of this ORF most closely resembles pyoverdine side chain peptide synthetase IV (PSPTO2150) of Pseudomonas syringae strain DC3000, which is responsible for assembly of the d-Asp-l-Ser component of the pyoverdine peptide side chain, and is also highly homologous to part of the pvdI (PA2402) gene product of P. aeruginosa. The pvdI, pvdJ, and pvdD genes of P. aeruginosa strain PAO1 encode NRPSs required for assembly of the type I pyoverdine (pyoverdinePAO) peptide chain (36, 38, 48).
Located downstream of orbI and orbJ are five genes which are homologous to genes involved in siderophore biosynthesis and transport in other bacteria. Three of these genes, pvdA, orbA, and pvdF, have been previously identified (70, 71). The orbA gene encodes a TonB-dependent outer membrane receptor specific for ferric-ornibactin, while the additive role of the pvdA and pvdF gene products is to convert the δ-amino group of ornithine to a hydroxamate group via a hydroxy intermediate (45, 84). Flanking the pvdA-orbA-pvdF genes are two similar ORFs, orbK and orbL. These genes encode proteins which are similar to N-acetyltransferases, such as IucB, which acetylates the -amino group of hydroxylysine during the biosynthesis of the siderophore aerobactin (16). It therefore is likely that the role of one, or both, of these genes is to promote the derivatization of the N5 nitrogen atom of N5-hydroxyornithine with a β-hydroxy acid prior to incorporation of this amino acid at the N terminus of ornibactin. Downstream of orbL is a gene which encodes a predicted cytoplasmic peptidase or amidohydrolase with no obvious connection to siderophore biosynthesis. This gene is followed by a sequence that resembles a ρ-independent transcriptional terminator.
Located upstream of orbI are a further seven genes with predicted roles in siderophore biosynthesis and utilization (Fig. (Fig.3A).3A). orbG encodes a polypeptide exhibiting homology to members of the family of α-ketoglutarate-dependent hydroxylases (27). In particular, there are strong matches to polypeptides involved in the biosynthesis of pyoverdine by P. syringae strain DC3000 (PSPTO2146) and Pseudomonas putida strain KT2440 (PP4222), but a similar protein is not encoded by the pyoverdine loci of P. aeruginosa strain PAO1 (locus tags PA2385-2413 and PA2424-2426). The pyoverdines produced by P. aeruginosa are distinguished from those of the other two pseudomonads by the absence of d-hydroxyaspartate in the peptide side chain (59). Therefore, the orbG gene product is most likely required for hydroxylation of the β-carbon atom of aspartic acid prior to its incorporation into ornibactin. The orbH gene, located immediately upstream from orbG, encodes a small protein of 80 amino acids which is very similar to MbtH-like proteins that participate in the biosynthesis of some other siderophores, such as mycobactin and pyoverdinePAO (36, 55, 57). However, the role of the MbtH-like proteins remains a mystery.
The orbB gene product is homologous to the E. coli ferric-dicitrate transporter periplasmic component, FecB, and the iron-ferrichrome transporter periplasmic component, FhuD. The orbC gene product is similar to the E. coli FecE and FhuC ATP-binding components of the ferric-dicitrate and iron-ferrichrome transporters, respectively. orbD encodes a polypeptide that possesses similar N-terminal and C-terminal domains, each of which is homologous to the integral membrane components (FecC and FecD) of the E. coli ferric-dicitrate transporter. OrbD is also homologous over its entire length to the pseudodimeric integral membrane component (FhuB) of the E. coli iron-ferrichrome transporter (34). Thus, the ferric-ornibactin transporter is most likely an ATP-binding cassette transporter (34, 64) comprised of a periplasmic ferric-ornibactin binding protein (OrbB) plus an integral membrane protein, OrbD, containing two domains, each of which interacts with a monomer of the ATPase, OrbC, on the cytoplasmic face of the membrane.
The orbF gene, located between orbD and orbB, encodes a protein that is homologous to the putative E. coli iron-ferrioxamine B reductase, FhuF (53). The proposed role of FhuF-like proteins is in the reductive release of ferrous iron from ferric-siderophore complexes, and therefore we suggest that the role of OrbF is to make available iron that is bound to internalized ferric-ornibactin. The presence of overlapping stop and start codons separating each gene within the orbG-orbC-orbD-orbF-orbB unit suggests that they are translationally coupled. This unit is followed by a predicted Rho-independent transcription terminator. Downstream of orbB, and transcribed in the opposite direction, is a gene encoding a polypeptide exhibiting a high similarity to the P. aeruginosa PvdE protein (PA2397) that is required for the production of pyoverdine (46). Like PvdE, OrbE possesses the characteristics of an ATP-binding cassette transport protein, and by analogy to PvdE, we suggest that orbE is probably required for the export of ornibactin across the cytoplasmic membrane to the periplasm.
In the other mutant that caused complete abolition of ornibactin production, OM3, the transposon had inserted into a gene located immediately upstream of orbH (Fig. (Fig.3A).3A). The product of this gene, OrbS, exhibits a high degree of similarity (36.5% identity over the matching region) to PvdS, an ECF σ factor of P. aeruginosa that is required for iron-regulated transcription of pyoverdine genes (55, 85). Finally, in mutant OM14, the transposon had inserted into a predicted chorismate lyase gene (ubiC) which is not linked to the ornibactin locus. Chorismate lyase catalyzes the removal of pyruvate from chorismate to produce 4-hydroxybenzoate and is required for the biosynthesis of ubiquinones (47, 77). As this gene does not appear to play a specific role in ornibactin synthesis or utilization, the OM14 mutant was not investigated further.
To examine the effects of disruption of the orbI, orbJ, and orbS genes on growth of B. cenocepacia under iron-limiting conditions, all the mutants (except OM14) were streaked on EDDHA plates. On this medium, growth of all the mutants was found to be completely inhibited, whereas the efficiency of plating (colony-forming ability) of the parent strain, KLF1, was the same as that when cultured in the absence of EDDHA (although the colony size was somewhat smaller) (results not shown). The growth of the ornibactin mutants was also monitored in liquid culture under iron-replete and iron-limiting conditions. Under iron-replete conditions, all mutants grew at a similar rate to KLF1. In the presence of the iron chelator 2,2′-dipyridyl, the growth of the mutants was strongly retarded and the rate of growth was similar for all the mutants. The growth rate of the parent strain, KLF1, was also significantly decreased under iron-limiting conditions, but to a lesser degree than for the ornibactin mutants, and this decrease was not apparent until the cell density was above an OD600 of 0.1. Figure Figure4A4A shows a comparison of the growth rates of two mutants, OM1 (orbI) and OM3 (orbS), with KLF1.
The insertion of mini-Tn5Tp into the orbS gene in OM3 occurred such that the dfrB2 gene was transcribed in the opposite direction to orbS (Fig. (Fig.3A).3A). To confirm that the ornibactin-negative phenotype of OM3 is not due to polar effects of the inserted minitransposon, we carried out a complementation experiment using pBBR1MCS-orbS. This plasmid was shown to restore the ability of OM3 to produce normal amounts of ornibactin in an iron-dependent manner. Thus, OM3 harboring pBBR1MCS-orbS produced an orange halo on CAS agar, whereas the mutant containing pBBR1MCS did not produce a halo (Fig. (Fig.2C).2C). Furthermore, analysis of culture supernatants by the IEF-CAS overlay assay confirmed that ornibactin was produced by OM3 containing pBBR1MCS-orbS when grown under iron-limited conditions (Fig. (Fig.2B).2B). This result demonstrates that inactivation of the orbS gene per se is responsible for the inability of OM3 to produce ornibactin.
To determine whether pyochelin production in B. cenocepacia requires orbS, we introduced an orbS::Tp allele into the Orb+ Pch+ strain, 715j, by allelic replacement. Although the resultant strain, 715j-orbS::Tp, produced a normal-sized orange zone on CAS agar, the IEF-CAS overlay assay revealed that it does not produce ornibactin. Thin-layer chromatography analysis of ethyl acetate-extracted culture supernatants revealed that 715j-orbS::Tp produced normal amounts of pyochelin (results not shown), indicating that the orbS gene is not required for production of pyochelin in B. cenocepacia. The normal-sized halo produced by the mutant growing on CAS plates is therefore likely to be due to compensatory production of pyochelin. As expected, the ability of the mutant to produce ornibactin was fully restored by pBBR1MCS-orbS.
The effect of orbS disruption on growth of a Pch+ strain in iron-limited medium was also investigated. We observed that, like OM3, 715j-orbS::Tp failed to grow on EDDHA plates (results not shown). In liquid medium, the mutant grew at a similar rate to 715j under iron-replete conditions. Under iron-limiting conditions, the growth rate of 715j, like KLF1, was retarded relative to iron-fed cells (Fig. (Fig.4B).4B). The growth rate of the orbS mutant was also retarded under these conditions. However, the profile of the growth curve was different from that of the parent strain: unlike the parent strain, in which a decreased growth rate was observed once the cell culture had reached an OD600 of 0.1, the decreased growth rate of the mutant was evident at lower cell densities and was maintained beyond the point at which the growth rate of 715j showed a marked decline.
The longest possible orbS coding sequence initiates with an ATG located 7 bp downstream from a potential Shine-Dalgarno sequence (Fig. (Fig.5B).5B). This would result in OrbS having an N-terminal extension of 31 amino acids relative to the N terminus of PvdS. However, the precise length of this extension cannot be predicted with any degree of certainty, as another methionine codon is located 3 bp downstream from the first methionine codon and a GTG codon is located a further 6 bp downstream (Fig. (Fig.5B).5B). To determine whether any of these codons serves as the true initiation codon for translation of orbS mRNA, we introduced mutations into all three codons and assayed their effect on the ability of pBBR1MCS-orbS to complement the orbS mutant, OM3, for production of ornibactin. The results show that substitution of the GTG codon by a stop codon completely abolishes complementation of the orbS mutant, whereas substitution by an alternative valine codon (GTC) does not (Fig. (Fig.2C).2C). This observation shows that the initiation codon for orbS translation is upstream of the valine codon. Substitution of the first ATG codon by a stop codon causes a decrease in the ability of orbS to complement OM3, whereas substitution of the second ATG codon by a stop codon completely abolishes complementation (Fig. (Fig.2C).2C). Substituting the second methionine codon by an alanine codon (GCC) almost completely abolishes complementation of the orbS mutant. This result suggests that the second methionine codon most likely serves as the initiation codon for orbS translation and indicates that OrbS has a 29-amino-acid N-terminal extension that is not present in PvdS. The inhibitory effect of replacing the first methionine codon with a stop codon probably results from subtle effects on the interaction of the 30S ribosome subunit with mRNA sequences flanking the true initiation codon, although it is possible that some translation may initiate from the first codon.
The transcriptional units present within the ornibactin gene cluster were analyzed by RT-PCR. Pairs of primers were designed that would amplify a segment of DNA spanning each intergenic region within the ornibactin gene cluster. These were used to amplify cDNA generated by reverse transcriptase extension of one or both primers on B. cenocepacia total RNA. For RNA prepared from cells growing under iron-limited conditions, RT-PCR products spanning each intergenic region within the ornibactin gene cluster were generated, except for the region between orbE and orbI (Fig. (Fig.3B).3B). There was also no product obtained for the pair of primers annealing to the 5′ end of orbS and the 5′ end of the upstream, divergently orientated ORF (BCAL1687), nor for the pair of primers annealing to the 3′ end of orbL and the 5′ end of the downstream ORF (BCAL1703). When cells were growing under conditions of iron sufficiency, transcripts were not detectable in this assay (Fig. (Fig.3B).3B). Taking into consideration the fact that orbE is transcribed in the opposite direction to the other ornibactin genes, these results suggest that the ornibactin gene cluster is organized into at least three transcriptional units comprising orbS to orbB, orbE to orbB, and orbI to orbL (Fig. (Fig.3A)3A) (see Discussion, below).
The similarity of the orbS gene product to PvdS prompted us to examine the region preceding each ornibactin gene for sequences resembling PvdS-dependent promoters. RNA polymerase core enzyme complexed with PvdS recognizes the consensus sequence TAAAT(N)16CGT (55, 85, 88). We observed very strong matches to this sequence upstream of orbH, orbE, and orbI (Fig. (Fig.5A).5A). In the case of orbE, the sequence is an exact match to the PvdS-dependent promoter consensus sequence, while in the other two cases an A residue is present at position 5 in the −35 region rather than a T. The location of the three putative OrbS-dependent promoters is consistent with the RT-PCR results and suggests that a fourth transcriptional unit (orbH-orbB) is present and overlaps the orbS-orbB transcriptional unit (Fig. (Fig.3A3A).
To determine whether the sequences identified above serve as iron-regulated promoters, DNA fragments containing these sequences were cloned into a new broad-host-range transcription fusion vector, pKAGd4. Each cloned promoter fragment included at least the first eight codons of the first downstream ORF and at least 140 bp of DNA upstream of the predicted −35 region. The promoter-lacZ fusions were introduced into 715j, and promoter activities were measured in exponential-phase cells growing under iron-limited and iron-sufficient conditions. The results show that all three DNA fragments contain promoters that are strongly repressed in B. cenocepacia during growth in the presence of high concentrations of ferric iron (Fig. (Fig.6A).6A). To determine whether OrbS is required for transcription of the ornibactin biosynthesis and utilization genes, the activities of the orbH, orbE, and orbI promoters were analyzed in the orbS::Tp mutant. The absence of any significant β-galactosidase activity in orbS mutant cells growing under conditions of iron limitation demonstrates that the activity of all three promoters is absolutely dependent upon OrbS (Fig. (Fig.6A6A).
To determine whether orbS is under control of an iron-regulated promoter, we cloned a fragment containing 209 bp of DNA upstream from the start codon for orbS translation into pKAGd4. This DNA region was shown to contain a strongly iron-regulated promoter (Fig. (Fig.6A).6A). As genes encoding many ECF σ factors are autoregulated, we examined whether the activity of this promoter requires functional OrbS. The results showed that the activity of porbS in iron-starved bacteria is not affected by the presence or absence of a functional orbS gene (Fig. (Fig.6A).6A). This is consistent with the absence of any sequences resembling OrbS-dependent promoters in this region. Interestingly, the activity of porbS in mutant bacteria growing in iron-replete medium was >4-fold higher than in wild-type bacteria.
The orbS promoter region contains two candidate σ70-dependent promoter sequences located 71 bp and 38 bp upstream of the orbS translation initiation codon (Fig. (Fig.5B).5B). To determine which of these serves as the orbS promoter, we measured the activity of orbS promoter fragments containing upstream deletions. The orbS promoter fragment used above (“full-length” promoter fragment) has an upstream endpoint at −178 relative to the predicted transcription start site for the downstream putative promoter. Other derivatives were constructed with endpoints located at 69 and 40 bp upstream of the predicted transcription initiation site for this promoter (Fig. (Fig.5B).5B). Deletion of upstream DNA as far as position −40 (which removes the upstream promoter-like sequence) has no significant effect on the activity of the orbS promoter in cells growing under iron starvation conditions. Furthermore, iron regulation of orbS promoter activity is retained, although there appears to be a progressive decrease in the degree of iron-dependent repression as the extent of deletion increases. Deletion of the downstream promoter-like sequence (deletion endpoint at +5) abolishes promoter activity (Fig. (Fig.6B).6B). These results indicate that the downstream promoter-like sequence is likely to constitute the orbS promoter and that iron regulation of this promoter requires sequences downstream of −40.
We also observed that the full-length orbS promoter is very active in E. coli MC1061 cells growing under iron starvation conditions and is strongly downregulated (~20-fold) by iron availability (results not shown). The results were qualitatively the same when the −69 and −40 promoter derivatives were analyzed, although, unlike the situation in B. cenocepacia, downregulation of the −69 and −40 promoter derivatives was as efficient as for the full-length promoter (results not shown). These results suggest that a conserved iron-responsive factor, such as the Fur protein (5, 20, 26, 42, 82), regulates porbS through interacting with the DNA downstream of position −40. Consistent with this, we identified a sequence (GTAAACGCAAATCATTCTC) spanning positions −33 to −15 which exhibits a 13/19 match (underlined) (on the template strand) to the consensus Fur-binding sequence (Fig. (Fig.5B).5B). To test whether a functional Fur box is present at the orbS promoter, we measured the activity of this promoter in an E. coli fur mutant. To do this, QC1732 was cotransformed with pKAGd4ΔAp containing the full-length orbS promoter and a compatible plasmid expressing B. cenocepacia fur. The results show clearly that iron regulation of the orbS promoter in E. coli only occurs when B. cenocepacia fur is provided in trans, thereby strongly implicating Fur as the regulator of the orbS gene in B. cenocepacia (Fig. (Fig.6C6C).
To demonstrate that Fur can bind to the orbS promoter in vivo, we employed the FURTA. In this assay, the presence of Fur-binding sites on DNA fragments cloned in multicopy plasmids titrates Fur and results in derepression of a Fur-regulated promoter-lacZ fusion on the E. coli chromosome (75). This is recognized by the formation of colonies with a Lac+ phenotype on MacConkey-lactose agar containing added ferric iron. The results showed that, as with the positive control plasmid, p3ZFBS, containing the consensus Fur-binding site, plasmids containing the full-length orbS promoter fragment, and derivatives with upstream endpoints at −69 and −40, give rise to a strong Lac+ phenotype (Fig. (Fig.7A).7A). However, the plasmid containing the +5 to +181 orbS promoter fragment gives a negative result in this assay. Thus, a site located downstream from position −40, with respect to the predicted orbS transcription start site, is able to titrate E. coli Fur in vivo. Part or all of this Fur-binding site is removed by deletion to +5. DNA fragments containing the orbH, orbE, and orbI promoters give a negative result in this assay, suggesting that the iron-dependent decrease in the activity of these promoters is not due to repression by Fur (Fig. (Fig.7B7B).
The ability of Fur to bind to the orbS promoter was confirmed in vitro by the electrophoretic mobility shift assay using the orbS promoter fragments described above. The mobility of porbS DNA fragments having upstream endpoints at −69 and −40 was specifically retarded by purified E. coli Fur at a concentration of 50 nM (Fig. 7C and D). At this concentration of Fur, two DNA fragments with retarded mobilities were observed, with the most slowly migrating fragment being present in the greatest abundance. It is possible that the most slowly migrating fragment is bound by two Fur dimers, while the more rapidly migrating fragment may be bound by a single Fur dimer. At 400 μM Fur, a further shift of the −69 fragment was observed (Fig. (Fig.7C).7C). The mobilities of the +5 orbS promoter fragment and DNA fragments containing the orbH, orbE, and orbI promoters were unaffected by Fur, even at a concentration of 600 nM (Fig. (Fig.7E).7E). These results confirm that the Fur regulatory protein can specifically interact with a DNA site overlapping the orbS promoter and strongly suggest that iron-dependent regulation of orbS transcription occurs through the direct action of Fur.
We isolated nine B. cenocepacia mutants which completely failed to produce ornibactin. Eight of the transposon insertions are located in the two NRPS genes, orbJ and orbI. The functional organization of the two NRPSs encoded by these genes can be predicted by database search and alignment tools. NRPSs have a modular arrangement, consisting mainly of repeated tri-domain peptide elongation units, each of which is comprised of an amino acid adenylation domain (A), which serves to activate the amino acid building block, a peptidyl carrier protein (PCP) domain, which is derivatized with 4′-phosphopantetheine and is used to covalently attach the amino acid to the NRPS via a thioester bond, and a condensation domain (C), which catalyzes peptide bond formation between the phosphopantetheine-anchored amino acid in one module and a similarly anchored amino acid in the preceding unit (for a review, see reference 10). The initiation unit occurs at the N terminus of one of the NRPSs participating in the biosynthesis of a peptide and is usually comprised of only an A and a PCP domain. Once the peptide chain is completed, in many systems it is hydrolytically removed from the final phosphopantetheine “arm” by a thioesterase domain, either located at the C terminus of the NRPS or present as a separate polypeptide (10).
As expected, a total of four PCP domains are present in the ornibactin synthesis machinery, three in OrbI and one in OrbJ (Fig. (Fig.1B).1B). As the N terminus of OrbI begins with an adenylation domain, synthesis of ornibactin is likely to be initiated here with the binding of N5-hydroxyornithine (which may also be derivatized with a β-hydroxy acid at this stage, as shown in Fig. Fig.1B).1B). Consistent with this idea, the adenylation domains of the first and second elongation units of OrbI are predicted to bind aspartate and serine based on the NRPSpredictor program (58). The second elongation unit also contains a predicted epimerase domain located between the subsequent PCP and C domains. It is likely that this domain is responsible for the conversion of the l-form of hydroxyaspartate to the d-form. OrbJ is effectively a single elongation unit which would be expected to bind N5-formyl-hydroxyornithine, but it contains an additional C domain near the C terminus instead of a thioesterase domain. As with the synthesis of the Vibrio cholerae siderophore vibriobactin, where VibH (a free-standing C domain protein) catalyzes nucleophilic displacement of 2,3-dihydroxybenzoate from the phosphopantetheine prosthetic group of VibB by the primary amine norspermidine (29, 30), the C-terminal condensation domain of OrbJ is likely to catalyze an analogous nucleophilic attack on the tetrapeptide thioester by free putrescine, thus liberating a completed ornibactin molecule (Fig. (Fig.1B).1B). Thus, the predicted enzymatic steps required for the biosynthesis of ornibactin can be accounted for by the products of the orbG, orbI-L, pvdA, and pvdF genes, together with a phosphopantetheinyl transferase required to activate the NRPSs. However, the role of OrbH in the biosynthesis of this siderophore is unclear.
The results of the RT-PCR analysis, in combination with the promoter analysis, suggest that the ornibactin cluster is organized into four transcriptional units and that some of these overlap. The orbS gene is transcribed in an iron-regulated manner from a σ70-dependent promoter (porbS) located a short distance upstream of the orbS translation initiation codon. Transcription from this promoter reads through orbS into the downstream genes, which are also transcribed from an OrbS-dependent promoter (porbH). Although readthrough transcription of orbH originating from porbS will not be OrbS dependent, it will be subject to regulation by Fur, i.e., the abundance of both types of orbH transcript is iron regulated. Our results suggest that both types of orbH transcript also include the orbG, orbC, orbD, orbF, and orbB cistrons and presumably terminate at the predicted Rho-independent terminator (ATCGCCGGCACACTGCCACTGCAGCGTGCCGGCGATTTTTT) (region of dyad symmetry underlined) located between orbB and orbE. The reason that RT-PCR analysis identified transcripts containing segments of both orbB and orbE can be explained if transcription originating from another OrbS-dependent promoter, porbE, reads through orbE into orbB, resulting in production of antisense orbB mRNA. Transcription originating from the third OrbS-dependent promoter, porbI, reads through the orbJ, orbK, pvdA, orbA, pvdF, and orbL genes.
Three OrbS-dependent promoters were identified within the ornibactin operon. The location and orientation of these promoters can account for OrbS-dependent transcription of all the predicted ornibactin genes except orbS itself. The sequences of these promoters are similar to the consensus sequence for PvdS-dependent promoters (55, 85). However, position 5 of the −35 region of PvdS-dependent promoters is usually occupied by a T residue, whereas two of the OrbS-dependent promoters have an A residue at this position. In addition, conservation within the core elements of the OrbS-dependent promoters appears more extensive than for the PvdS-dependent promoters, as in all three OrbS-dependent promoters the first position downstream of the conserved −10 sequence is a C residue and the first position downstream of the −35 region is an A residue. Moreover, sequences outside the core elements are conserved at the OrbS-dependent promoters. For example, three G · C base pairs occur immediately upstream of the −35 region, and nine consecutive G · C base pairs are present at exactly the same position in the spacer region separating the −35 and −10 elements. Furthermore, the −35 region of OrbS-dependent promoters is part of an uninterrupted 8-bp A · T-rich sequence. None of the PvdS-dependent promoters has these particular features in precisely this arrangement (data not shown).
Our results strongly suggest that iron regulation of the OrbS-dependent promoters is due to modulation of OrbS activity. OrbS activity, in turn, appears to be modulated by control of orbS transcription in response to the intracellular iron concentration and is mediated by the Fur repressor. We have identified the orbS promoter and demonstrated that Fur binds to a sequence located between −40 and +5 with respect to the transcription start site of the orbS gene. Fur is therefore in a position to block access of RNA polymerase to this promoter. The 19-bp consensus Fur-binding sequence, GATAATGATAATCATTATC (the Fur box), is comprised of two overlapping inverted repeat sequences (having either a 6-1-6 or 7-1-7 organization), with each inverted repeat binding a separate Fur dimer (5, 37). This is supported by predictions based on the structure of Fur, which also suggest that two Fur dimers are required to bind to a Fur box (56). The predicted Fur-binding site at the orbS promoter exhibits only a 13/19 match to the consensus Fur box sequence, which suggests that Fur may have a low affinity for the orbS promoter. Nevertheless, the affinity of Fur for the orbS promoter is strong enough to allow Fur to effect efficient repression of porbS in vivo. The reason for the additional shift of the −69 orbS promoter fragment at very high Fur concentrations is due to low-affinity binding of Fur to sequences immediately upstream of the primary Fur-binding site. This is likely to be a nonspecific interaction, but it may be facilitated by interactions with Fur dimers bound at the primary binding site, as occurs at the pvdS promoter (54).
We observed that orbS promoter activity was higher in orbS mutant bacteria than in the wild-type strain growing under iron-replete conditions (Fig. (Fig.6B).6B). A possible reason for this is that the orbS mutant can only use the pyochelin system to acquire ferric iron. Pyochelin is likely to be a less efficient siderophore than ornibactin, on account of its relatively low binding coefficient for ferric iron (14, 86). Consistent with this, there is good evidence to suggest that, under iron starvation conditions, the pyochelin system cannot compensate for inactivation of the ornibactin system in B. cenocepacia (86). Our plate assays and growth rate measurements support this conclusion. The lower efficiency of iron acquisition by the pyochelin system would lead to a decreased iron pool in orbS mutant cells in a situation where bacteria are being supplied with ferric iron as an iron source. As a consequence, Fur-mediated repression of iron-regulated genes would be partially relieved to allow sufficient expression of iron acquisition genes to achieve both an intracellular iron pool and a growth rate comparable to the wild type (Fig. (Fig.4B4B).
The activities of several ECF σ factors that are required for transcription of siderophore genes are regulated by cytoplasmic membrane-anchored signal transducers. These proteins interact with the N terminus of the cognate ferric-siderophore receptor via their periplasmic domains. Such receptors are characterized by possession of a long periplasmic N-terminal extension, and induction of the system requires the binding of the ferric-siderophore complex to the receptor (for reviews, see references 8, 32, and 63). For example, the activity of PvdS is modulated by the membrane-anchored regulator FpvR, which in turn interacts with the ferric-pyoverdine receptor, FpvA (6). Thus, PvdS activity is controlled by two overlapping regulatory circuits, one of which involves Fur and the other requiring FpvR. However, the ferric-ornibactin receptor, OrbA, does not contain an N-terminal extension, and the presence of ornibactin is not required to induce the system (70). A survey of the B. cenocepacia genome sequence also failed to identify a candidate anti-σ factor for OrbS (results not shown). It would therefore appear that OrbS activity is controlled mainly as a result of the action of Fur at the orbS promoter and does not involve a signal transduction system.
This work was funded by an MRC postgraduate studentship (G78/7919 awarded to K.A.), two University of Sheffield PG studentships (awarded to C.A.L. and K.L.F.), and the Wellcome Trust (project grant 073917).
We are indebted to S. Andrews (University of Reading) who generously provided purified E. coli Fur protein. We also thank T. Brickman (University of Minnesota) for p3ZFBS, M. Kovach (Baldwin Wallace College) for pBBR1MCS, S. Valla (NUST, Trondheim, Norway) for pPR9TT (U.S. patent no. 6,258,565), K. Hantke (Tubingen, Germany) for E. coli H1717, and D. Touati (CNRS, Paris, France) for E. coli QC1732. We also acknowledge Anne Cook, Gil Shalom, and Katie Wynne for technical support. We are grateful to Beowulf Genomics and the Sanger Institute Pathogen Sequencing Unit for nucleotide sequence determination of the B. cenocepacia strain J2315 genome.
†Supplemental material for this article may be found at http://jb.asm.org/.
University at Buffalo, South Campus
Buffalo, NY 14215