GENES AND GENOMES

Identification and characterization of novel pyoverdine synthesis genes in Pseudomonas aeruginosa

  • Department of Biochemistry, University of Otago, PO Box 56, Dunedin, New Zealand
  • Correspondence
    Iain L Lamont
    iain.lamont{at}stonebow.otago.ac.nz
  • Microbiology 2003; 149(4):833–842 · https://doi.org/10.1099/mic.0.26085-0

    View at publisher PubMed

    Abstract

    Fluorescent pseudomonads secrete yellow-green siderophores named pyoverdines or pseudobactins. These comprise a dihydroxyquinoline derivative joined to a type-specific peptide and, usually, a carboxylic acid or amide. In Pseudomonas aeruginosa strain PAO1, six genes that encode proteins required for pyoverdine synthesis (pvd genes) have been identified previously. Expression of all of these genes requires an alternative sigma factor PvdS. The purpose of this research was to identify other genes that are required for pyoverdine synthesis in P. aeruginosa PAO1. Fourteen candidate genes were identified from the PAO1 genome sequence on the basis of their location in the genome, the functions of homologues in other bacteria, and whether their expression was likely to be PvdS-dependent. The candidate genes were mutated and the effects of the mutations on pyoverdine production were determined. Eight new pvd genes were identified. The presence of homologues of pvd genes in other strains of P. aeruginosa was determined by Southern blotting and in other fluorescent pseudomonads by interrogation of genome sequences. Five pvd genes were restricted to strains of P. aeruginosa that make the same pyoverdine as strain PAO1, suggesting that they direct synthesis of the type-specific peptide. The remaining genes were present in all strains of P. aeruginosa that were examined and homologues were present in other Pseudomonas species. These genes are likely to direct synthesis of the dihydroxyquinoline moiety and the attached carboxylic acid/amide group. It is likely that most if not all of the genes required for pyoverdine synthesis in P. aeruginosa PAO1 have now been identified and this will form the basis for a biochemical description of the pathway of pyoverdine synthesis.

    INTRODUCTION

    Fluorescent pseudomonads secrete yellow-green fluorescent siderophores termed pyoverdines or pseudobactins (Fig. 1). These enable acquisition of Fe(III) ions from the environment (reviewed by Meyer & Stintzi, 1998) and also serve as signalling molecules controlling gene expression inside the bacterial cells (Lamont et al., 2002; Visca et al., 2002). A large number of pyoverdines and pseudobactins have been characterized and all comprise a shared dihydroxyquinoline chromophore joined to an acyl (carboxylic acid or amide) group and a short (6–12 amino acid) type-specific peptide (Fig. 1) (reviewed by Budzikiewicz, 1993; Meyer, 2000). Pyoverdines/pseudobactins produced by a single strain all have the same peptide but they may differ in the nature of the acyl group. Strains of Pseudomonas can utilize heterologous pyoverdines and pseudobactins for iron acquisition and the spectrum of ferrisiderophores that can be used forms the basis of a strain identification method termed siderotyping (Meyer et al., 2002).

    Figure image not available in archive
    Fig. 1.

    Structures of pyoverdines and pseudobactins. (a) PyoverdinePAO from P. aeruginosa PAO1 (Abdallah, 1991). The acyl group can be a carboxylic acid or an amide (as shown). (b) The peptide components of pyoverdines and pseudobactins from different strains of Pseudomonas (data from Meyer, 2000). Amino acid residues that are in the d configuration are underlined. fOHOrn, N5-Formyl-N5-hydroxyornithine; aThr, allo-threonine; OHOrn, hydroxyornithine; c, cyclic.

    The pyoverdines that are produced by strains of P. aeruginosa are classified into three types (I–III) that are distinguished by their peptides (Fig. 1) (Meyer et al., 1997). The genes and enzymes that are required for synthesis of pyoverdine are best characterized in the type I strain PAO1. Most of the pyoverdine synthesis genes that have been identified in this strain are at about 47 min on the genetic map (Ankenbauer et al., 1986; Hohnadel et al., 1986; Rombel & Lamont, 1992; Stintzi et al., 1996; Tsuda et al., 1995) and these genes are listed in Table 1. The pvdA gene encodes an enzyme that catalyses synthesis of N5-hydroxyornithine (Visca et al., 1994) and the pvdF gene product catalyses the formylation of N5-hydroxyornithine to give N5-formyl-N5-hydroxyornithine, which is present in the type I pyoverdine (pyoverdinePAO) made by P. aeruginosa PAO1 (McMorran et al., 2001). The product of the pvdD gene is a peptide synthetase that directs incorporation of two l-threonine residues into the peptide of pyoverdinePAO (Merriman et al., 1995; Ackerley et al., 2003). The pvdIJK gene products also have the characteristics of peptide synthetases (Lehoux et al., 2000); resequencing of P. aeruginosa PAO1 DNA shows that pvdJ and pvdK are part of a single gene (see below), which will be referred to here as pvdJ. The product of the pvdE gene is likely to be an ABC transporter protein and it is essential for pyoverdine synthesis (McMorran et al., 1996) although the transported substrate has not been identified. A separate cluster of four genes (pvcABCD) at 66–70 min on the genetic map has been reported to be required for synthesis of the pyoverdine chromophore (Stintzi et al., 1996, 1999). However, pvc mutants are able to make pyoverdine in some growth media so that these genes are not essential for pyoverdine synthesis (P. Cornelis & U. Ochsner, personal communication).

    Table 1.

    Genes involved in synthesis or transport of pyoverdine in P. aeruginosa PAO1

    Expression of all of the pyoverdine-synthesis genes that have been characterized to date requires an alternative sigma factor protein, PvdS (reviewed by Visca et al., 2002). Promoters that are recognized by RNA polymerase containing PvdS contain a sequence motif, the IS box, at about 33 bp from the transcription start sites and this forms part of the promoter sequence (Rombel et al., 1995; Wilson et al., 2001). A second sequence CGT at about −10 bp is also required for promoter recognition by PvdS (S. Tsao, M. J. Wilson & I. L. Lamont, unpublished data). The activity of PvdS is regulated post-translationally by an anti-sigma factor FpvR (Lamont et al., 2002) and in addition expression of the pvdS gene is repressed in iron-rich cells (Cunliffe et al., 1995; Leoni et al., 1996), providing two levels of control of pyoverdine production.

    The structural complexity of pyoverdinePAO suggests that the biosynthetic pathway will involve a number of enzymes in addition to those that have been identified to date. The sequence of the genome of P. aeruginosa strain PAO1 (Stover et al., 2000) provides a new approach for identifying pyoverdine synthesis genes. In this study we identified possible pyoverdine synthesis genes in the genome sequence, mutated them, and determined the effects of the mutations on pyoverdine production. In addition, we determined which genes are present in other pseudomonads and which are restricted to strains that, like P. aeruginosa PAO1, make type I pyoverdine.

    METHODS

    Bacterial strains and growth conditions.

    Bacterial strains and plasmids used in this study are listed in Table 2. Escherichia coli was grown in Luria (L-) broth (Sambrook et al., 2000) and P. aeruginosa in L-broth or King's B broth (King et al., 1954) at 37 °C with aeration for liquid cultures. Media were solidified by the addition of agar (1·5 %) and supplemented with antibiotics or with the iron-chelating compound ethylenediamine(o-hydroxy)phenylacetic acid (EDDA) as described previously (McMorran et al., 2001). Gentamicin was added to a final concentration of 4 μg ml−1 (E. coli) and 20 μg ml−1 (P. aeruginosa) where required.

    Table 2.

    Bacterial strains and plasmids

    Molecular biology methods.

    Plasmid DNA was prepared using the High Pure Plasmid Isolation kit (Roche) and genomic DNA was prepared from P. aeruginosa as described by Chen & Kuo (1993). DNA was amplified from P. aeruginosa DNA by PCR using primers designed from the P. aeruginosa PAO1 genome sequence (); details of primers are available on request. Restriction digestion, gel electrophoresis and DNA cloning were done by standard methods (Sambrook et al., 2000) with cloning into pGEM-T Easy carried out using the protocol recommended by the manufacturer (Promega). All plasmid constructs were verified by DNA sequencing (Centre for Gene Research, University of Otago, Dunedin). Sequencing a cloned PCR fragment spanning the junction of PA2400 and PA2401 in the P. aeruginosa genome showed that a GC base-pair was missing from the genome sequence at position 2669175 and when this was included PA2400 and PA2401 form a single reading frame PA2400/1 (pvdJ). Southern blotting was carried out by standard methods (Sambrook et al., 2000), using as probes radiolabelled PCR fragments or cloned restriction fragments corresponding to individual genes, except that pvdN and pvdO were part of the same PCR fragment. Membranes were washed at 65 °C in 0·1 % SDS/0·1× SSC prior to autoradiography.

    Gene disruptions in P. aeruginosa.

    PCR fragments (1·0–1·5 kb) corresponding to genes to be mutated were cloned into pGEM T-Easy, excised using restriction enzymes (usually HindIII and EcoRI) corresponding to sites that were incorporated into the PCR primers, and subcloned into pEX18Tc or pEX18Gm. Kanamycin-resistance cassettes were then cloned into restriction sites within the target genes unless the cloned fragment was internal to the gene to be mutated. pEX constructs were transferred into P. aeruginosa PAO1 by triparental conjugation using the helper plasmid pRK2013 as described previously (McMorran et al., 1996) with selection for transconjugants in which plasmid DNA had integrated into the chromosome of P. aeruginosa by homologous recombination. For heterodiploid strains in which conjugation gave rise to bacteria containing both a wild-type and a mutant (kanamycin-disrupted) gene, plasmid DNA containing the wild-type gene was cured from the bacteria by subculture in L-broth containing kanamycin, followed by sucrose-selection for plasmid-lacking strains (Hoang et al., 1998). DNA from all recombinant P. aeruginosa strains was analysed by PCR and Southern blotting to ensure that the intended mutations had been generated.

    Phenotypic analysis of bacteria.

    P. aeruginosa strains were analysed for production of pyoverdine by growth on King's B agar and on agar supplemented with EDDA; EDDA prevents the growth of P. aeruginosa strains that are unable to make or take up (ferri)pyoverdine (Ankenbauer et al., 1986), and pyoverdine gives a yellow-green pigmentation around Pvd+ colonies (King et al., 1954). Pyoverdine production was quantified by growing cultures of bacteria in King's B broth as described previously (McMorran et al., 2001).

    Computational analysis.

    DNA sequences were obtained from the P. aeruginosa genome project () and the P. aeruginosa genome database (). Sequences were manipulated using Seqed (Devereux et al., 1984) and analysed using NLDNA and Codonuse as described previously (Merriman et al., 1995). Database searches and analysis of the genomes of other fluorescent pseudomonads were carried out at the National Center for Biotechnology Information () with blast algorithms.

    RESULTS AND DISCUSSION

    Identification of candidate pyoverdine synthesis and transport genes

    The pvd mutations identified previously spanned a region of the P. aeruginosa PAO1 genome extending from approximately pvdA (PA2386) to pvdS (PA2426) (Tsuda et al., 1995) (Fig. 2). ORFs in this part of the genome were identified as part of the P. aeruginosa genome sequencing project (Croft et al., 2000; Stover et al., 2000). For all of these ORFs, codon usage was found to be similar to that of other P. aeruginosa genes (Grocock & Sharp, 2002; West & Iglewski, 1988) (data not shown). Two approaches were taken to identify ORFs in this region of the genome that may contribute to pyoverdine synthesis or transport. Firstly, the ORFs were screened to identify those that are preceded by a probable PvdS-dependent promoter, or may be part of an operon that is preceded by a probable PvdS-promoter, and so are likely to be co-expressed with pyoverdine synthesis genes. Secondly, blast searches were carried out to determine whether homologous genes are involved in synthesis of siderophores or other secondary metabolites. This resulted in the identification of 14 previously uncharacterized genes that may be required for pyoverdine synthesis or transport (Table 3).

    Figure image not available in archive
    Fig. 2.

    The pyoverdine locus of P. aeruginosa PAO1. The orientations of ORFs are shown, with numbers corresponding to those in the P. aeruginosa genome (; ). Gene names are also shown, with genes that were identified in this study in bold. The positions of PvdS-dependent promoters, and likely promoters, are indicated by black arrowheads and other known promoters are indicated by hatched arrowheads.

    Table 3.

    Putative pyoverdine synthesis/secretion genes analysed in this study

    Mutational analysis of candidate genes

    Mutations were introduced into eleven of the genes listed in Table 3 as described in Methods. Three of the mutant strains (in ORFs PA2389, PA2411 and PA2417) retained the ability to make pyoverdine. These mutants were able to grow in the presence of EDDA, indicating that uptake of ferri-pyoverdine was also unaffected by the mutations. The remaining eight mutant strains did not make any detectable pyoverdine and were also unable to grow in the presence of the iron-chelating compound EDDA (Table 3). These phenotypes indicate that the corresponding genes are required for pyoverdine synthesis and they were assigned the names pvdGpvdQ (Table 3, Fig. 2). Three of these genes (pvdM, pvdN and pvdG) are predicted to be in operons upstream of other pvd genes (see Fig. 2) and it is possible that mutations in these genes cause a Pvd phenotype because of polar effects on expression of the downstream gene(s). However we have named these pvd genes, as different genes within a biosynthetic operon invariably encode products that contribute to the same biochemical pathway.

    While this manuscript was being prepared, another study described mutations in many of the genes characterized in this study (Ochsner et al., 2002). The phenotypes obtained by these researchers are listed in Table 3 and are consistent with those described here. The role of pvdL in pyoverdine synthesis has also been described very recently (Mossialos et al., 2002).

    A total of 15 pvd genes have now been identified that are essential for pyoverdine synthesis in P. aeruginosa PAO1 and it is likely that most, if not all, of the genes that are essential for pyoverdine synthesis in this strain are now known. An early study mapped two mutations that affected pyoverdine synthesis to a locus at 23 min on the recalibrated genetic map of P. aeruginosa PAO1 (Hohnadel et al., 1986) but further study of these mutants has not been reported. The pvc genes lie at about 66–70 min on the genetic map (Stintzi et al., 1996), about 240 kb away from pvdS. All other pvd mutations, including all of the 24 mutations identified by Tsuda et al. (1995), have been mapped to the 47 min region of the P. aeruginosa chromosome that corresponds to the part of the genome represented in Fig. 2. We have not mutated the 17 genes in this interval that did not meet our criteria for candidate pyoverdine synthesis genes and so cannot exclude the possibility that they contribute to pyoverdine synthesis. However, two of these genes (PA2403 and PA2407) were mutated by Ochsner et al. (2002) and the mutant bacteria retained the ability to make pyoverdine. In addition the sequences of several of these genes suggest that they have functions other than pyoverdine synthesis. For example, PA2414 has 55 % sequence identity with l-sorbosone dehydrogenase from Acetobacter liquefaciens and PA2416 has 55 % identity with a periplasmic trehalase from E. coli (data not shown). A number of mutations in this part of the genome did not affect pyoverdine synthesis (Tsuda et al., 1995).

    Detection of pvd gene homologues in other Pseudomonas strains

    The three different kinds of pyoverdines (types I–III) that are made by strains of P. aeruginosa all have the same dihydroxyquinoline component and are distinguished by the compositions of their peptides (Meyer et al., 1997; Meyer, 2000) (Fig. 1). It is likely that synthesis of the shared dihydroxyquinoline group, with its attached carboxylic acid or amide, has the same biosynthetic pathway in all strains and that the enzymes for this are encoded by orthologous genes in different strains. In contrast, type-specific genes probably direct synthesis of the peptides that distinguish the different pyoverdines.

    Pyoverdine synthesis genes from P. aeruginosa PAO1 were used as hybridization probes with genomic DNA from other P. aeruginosa strains that produce type I, type II or type III pyoverdine (Fig. 3, Table 4). Hybridizations were carried out under conditions of high stringency to ensure that only very similar DNA sequences would hybridize. Some of the probes hybridized with DNA from all of the strains tested, indicating that orthologous genes are present in all of these strains. Other genes were only present in type I strains.

    Figure image not available in archive
    Fig. 3.

    Detection of pvd gene homologues by Southern blotting. Chromosomal DNA from P. aeruginosa strains producing type I pyoverdine (PAO1, 58.40 and 59.41) (lanes 1–3), type II pyoverdine (strains Pa4, 58.35 and 59.40) (lanes 4–6) and type III pyoverdine (strains Pa6, 58.36 and 59.20) (lanes 7–9) was digested with PstI. Following electrophoresis, the DNA was analysed by Southern blotting using the following pvd genes from P. aeruginosa PAO1 as probes: (a) pvdA; (b) pvdF; (c) pvdJ; (d) pvdN/pvdO.

    Table 4.

    Presence of homologous genes in other fluorescent pseudomonads

    Partial or complete genome sequences are available for strains of P. fluorescens, P. putida and P. syringae. The predicted sequences of the products of P. aeruginosa pvd genes were used in blast searches in order to identify homologues in other Pseudomonas species; the results are shown in Table 4. Many of the genes that were detected by hybridization in all strains of P. aeruginosa (PA2389, pvdP, the pvdM–pvdO operon, PA2411, pvdH, pvdL and pvdS) had homologues with over 60 % sequence identity in the other species. These are likely to be orthologues of the P. aeruginosa genes and to have common functions in all of the fluorescent pseudomonads. In addition, pvdQ and pvdG, which were detected in all P. aeruginosa strains, had homologues in the other species but with lower amounts of sequence similarity and these may also be orthologues. PvdS is an ECF sigma factor that is required for expression of other pyoverdine synthesis genes. The other pvd genes that are present in all fluorescent pseudomonads are most likely to be required for synthesis of the dihydroxyquinoline and amide/carboxylic acid moiety that is present in all pyoverdines. The pvdL gene-product has all of the characteristics of a peptide synthetase and corresponding genes are present in different strains of P. aeruginosa, and other pseudomonads (Table 4). Synthesis of the dihydroxyquinoline chromophore of pyoverdines is known to require amino acid precursors (Baysse et al., 2002; Budzikiewicz, 1993). It therefore seems likely that PvdL catalyses synthesis of a peptide that is modified by other enzymes (encoded by the shared pvd genes) to form the dihydroxyquinoline derivative coupled to an amide/carboxylic acid that is present in pyoverdines. Similar conclusions were reached by Mossialos et al. (2002) in an independent study of pvdL.

    Genes that are present in only some strains (pvdA, pvdD, pvdF, pvdI and pvdJ in P. aeruginosa PAO1) most likely direct synthesis of the peptide components of pyoverdines. This has been demonstrated biochemically for the products of pvdA and pvdF that catalyse hydroxylation of ornithine and formylation of hydroxyornithine, respectively, to generate N5-formyl- N5-hydroxyornithine, which is present in the peptide of type I pyoverdine (McMorran et al., 2001; Visca et al., 1994). Homologues of pvdA have been shown to be present in a strain of P. aeruginosa that makes type II pyoverdine and also in strains of P. fluorescens and Burkholderia cepacia, though not in a strain of P. putida (Visca et al., 1994). A homologue of pvdA from Pseudomonas sp. B10 complemented a pvdA mutation in P. aeruginosa PAO1 (Ambrosi et al., 2000), showing that it is an orthologue of pvdA. In this study, homologues of pvdA were present in all strains of P. aeruginosa that were tested, and also in P. fluorescens and P. putida, though not P. syringae (Table 4). However homologues of pvdF were not found outside P. aeruginosa type I strains and synthesis of formylhydroxyornithine must involve a different biosynthetic process in other strains/species that incorporate this compound into pyoverdine or pseudobactin.

    pvdD, pvdI and pvdJ were also detected by hybridization only in strains of P. aeruginosa making type I pyoverdine. These genes are thought to encode peptide synthetases (Lehoux et al., 2000; Merriman et al., 1995), a family of proteins with many conserved sequence features (Marahiel et al., 1997). PvdD directs incorporation of two l-threonine residues into pyoverdinePAO (Ackerley et al., 2003) and pvdI and pvdJ are very likely to encode peptide synthetases that direct incorporation of the remaining six amino acids into the peptide of type I pyoverdines. Homologues of these gene products are present in the other fluorescent Pseudomonas species but the levels of sequence identity (41–56 %) were no higher than those of paralogues of pvdD, pvdI and pvdJ in the P. aeruginosa genome (data not shown). The homologues in the other species are very likely to encode peptide synthetases that direct incorporation of different amino acids into pyoverdines/pseudobactins, or other secondary metabolites.

    The pvdE gene product has all the characteristics of an ABC-type transporter protein (McMorran et al., 1996) although its substrate(s) has not been identified. Homologues were not detected by Southern blotting in other strains of P. aeruginosa, suggesting that the substrate is strain-specific although homologues are present in other fluorescent pseudomonads. ABC transporter proteins that have many shared sequence features may have different substrates (Higgins, 1992, 2001). It remains to be determined whether the PvdE homologues present in other species transport the same substrate as in P. aeruginosa PAO1, or whether they transport a different substrate with the sequence similarities reflecting shared structural features.

    In conclusion, the research described here has identified eight previously undescribed genes that are required for synthesis of pyoverdine. Analysis of the distribution of pyoverdine synthesis genes amongst fluorescent pseudomonads, along with analysis of their sequences, indicates their possible roles in the biochemical pathway of pyoverdine synthesis. This will provide the basis for biochemical characterization of individual enzymes and a complete description of the pathway of pyoverdine synthesis.

    Acknowledgments

    This research was supported by a grant from the New Zealand Lotteries Board (Health). We thank Jean-Marie Meyer and Herbert Schweizer for providing strains, Pierre Cornelis, Mike Vasil and Urs Ochsner for communicating results prior to publication, and Clive Ronson and Paul Beare for comments on an earlier version of this manuscript. We acknowledge the Pseudomonas Sequencing Consortium and Interactive Pseudomonas Genome Project for providing access to the annotated P. aeruginosa PAO1 genome and the US DOE Joint Genome Institute and The Institute for Genomic Research for providing access to unfinished Pseudomonas genome sequences.

    References