Isolation and characterization of Pseudomonas aeruginosa mutants blocked in the synthesis of pyoverdin

We have isolated and characterized by chemical and enzymatic analyses three distinct types of pyoverdin-defective (pvd) mutants of Pseudomonas aeruginosa PAO1. The pvd-1 mutant is an L-N5-hydroxyornithine (L-N5-OH-Orn) auxotroph unable to hydroxylate L-ornithine (L-Orn) in a cell-free system and requiring L-N5-OH-Orn for pyoverdin production. The other two types of mutants appear to be blocked in further steps of the biosynthetic pathway leading to pyoverdin, namely, the acylation of L-N5-OH-Orn (pvd-2) and chromophore synthesis (pvd-3). The different pvd mutations were all found to be located in the catA1 region at 47 min of the genetic map of P. aeruginosa PAO1.

L-N5-OH-Orn for pyoverdin production. The other two types of mutants appear to be blocked in further steps of the biosynthetic pathway leading to pyoverdin, namely, the acylation of L-N5-OH-Orn (pvd-2) and chromophore synthesis (pvd-3). The different pvd mutations were all found to be located in the catAl region at 47 min of the genetic map of P. aeruginosa PAO1.
Under iron limitation, Pseudomonas aeruginosa synthesizes two siderophores, pyochelin and pyoverdin. Pyochelin is a unique phenolate siderophore which has been assigned the chemical structure 2-[2-(o-hydroxyphenyl)-2-thiazoline-4-yl]-3-methyl-4-thiazolidine carboxylic acid (12). Ferripyochelin uptake occurs via two outer membrane proteins of 14 kDa (38) and 75 kDa (16). Pyoverdin is a hydroxamate siderophore composed of a 6,7-dihydroxyquinoline-containing fluorescent chromophore, bound to the N terminus of an octapeptide (D-Ser-L-Arg-D-Ser-L-N5-OH-Orn-L-Thr-L-Thr-L-Lys-L-N5-OH-Orn, in P. aeruginosa PA01) (7). Recent studies of different isolates of P. aenrginosa have demonstrated a strict strain specificity of the pyoverdin-mediated Fe(III) uptake system associated with a remarkable heterogeneity in the amino acid composition of the peptide moiety and in the outer membrane proteins involved in ferripyoverdin uptake (10). In P. aeruginosa PAO1, two high-molecularmass outer membrane proteins of 80 kDa (28) and 90 kDa (34) have been shown to function in ferripyoverdin transport. The different pyoverdins from fluorescent pseudomonads (or pseudobactins from plant-related isolates) and the yellow-green siderophore azotobactin from Azotobacter vinelandii have similar spectral characteristics and extensive structural homologies (reference 1 and references therein). As a general rule, the same dihydroxyquinoline-derived chromophore is present in all these compounds, with minor structural differences. Moreover, they all contain hydroxamate groups provided by one or two residues of L-N5-OH-Orn which participate in Fe(III) coordination with the quinoline hydroxyls of the chromophore (1). These similarities also extend to the DNA level, since an extensive homology has recently been shown for pyoverdin genes from different fluorescent Pseudomonas spp. (35).
Pyoverdins are powerful iron chelators, with Fe(III)binding constants about 25 orders of magnitude higher than that of pyochelin (11). The different affinities of the two P. aeruginosa siderophores for Fe(III) may also account for the * Corresponding author. higher growth-promoting activity of pyoverdin in an ironrestricted environment. In fact, the importance of pyoverdin compared with pyochelin in the removal of transferrinbound Fe(III) has been deduced from the fact that the ability of pyoverdin-deficient mutants to multiply in the presence of this serum protein is greatly reduced while the pyochelindeficient mutant grows as well as the wild type (5). Therefore, it has been proposed that pyoverdin plays a relevant role in the infectious process of P. aeruginosa (5).
The biosynthetic pathway leading to pyoverdin has not yet been elucidated. Taking into account that hydroxylation of w-amino acids is an enzymatic reaction involved in the formation of a number of hydroxamate siderophores (2,3,14,17,20,39), we assumed that conversion of L-Orn to L-N5-OH-Orn might represent an early biosynthetic step in the generation of the peptidic moiety of pyoverdin. In this report we substantiate this hypothesis by showing that a P. aeruginosa PA01 mutant defective in pyoverdin production lacked the L-Orn-N5-hydroxylating activity and required L-N5-OH-Orn for pyoverdin synthesis.
The strains of P. aeruginosa used in this study are listed in Table 1. The strains were grown in nutrient-yeast extract medium (32) or in the minimal medium M9 (21) supplemented with 20 mM sodium succinate (SM9), sodium benzoate, or mannitol as the carbon source. Amino acids required for auxotrophs were added at a concentration of 1 mM. Streptomycin sulfate was used at 500 ,ug/ml. To reduce iron availability, the chelator 2,2'-dipyridyl (9) was added to SM9 at 500 ,uM, corresponding to the maximal nontoxic concentration of the compound. L-N5-OH-Orn was obtained from acid hydrolysis of rhodotorulic acid (6). The molecule was checked by paper electrophoresis on Whatman no. 1 filter paper in a water-cooled apparatus at 30 V/cm with pyridine-acetic acid-water (7:5:465) (pH 5.5) as a solvent. The cathodic migration rate of L-N5-OH-Orn was 0.6 to 0.7 with respect to L-Orn, as determined after ninhydrin and alkaline tetrazolium spraying (6,39). L-N5-OH-Orn was stored at -20°C in acid solution (39) and employed in biochemical feeding assays at 400 ,uM. P. aeruginosa PA01 was mutagenized with ethyl methanesulfonate as previously described (24). For enzymatic analysis, cell-free lysates were prepared essentially as reported elsewhere (33). P. aeruginosa was grown to mid-logarithmic phase (A620, 0.30 to 0.40) in SM9 supplemented with either 150 ,uM nitrilotriacetic acid (to minimize protease activity and to reduce iron availability [8]) or 100 ,uM FeCl3. Cells (approximately 40 mg [dry weight]) were collected by centrifugation (2,000 x g, 15 min, 4°C), washed once with 0.85% NaCl, and resuspended in 1 ml of 100 mM sodium phosphate buffer (pH 7.0) containing 2 mg of lysozyme per ml and 20 mM EDTA. After 15 min of incubation at 20°C, cells were put in ice and subjected to ultrasonic disruption for a total of 3 min. The cell-free lysate was supplemented with 1 ml of 100 mM sodium phosphate buffer containing 40 ,ug of DNase per ml, 100 ,ug of RNase per ml, 4 mM dithiothreitol, and 4 mM glutamine. After 30 min of incubation at 4°C, cell debris was removed by centrifugation (2,000 x g, 15 min, 4°C) and the resulting supernatant (containing less than 102 viable cells per ml) was immediately assayed for hydroxylase activity. The enzyme assay was performed by mixing 2 ml of crude lysate and 2 ml of 100 mM sodium phosphate buffer containing 2 mM sodium pyruvate and 4 mM L-Orn. Incubation was carried out at 30°C for up to 6 h in 10-ml tubes flushed with air (4 h was used as the standard incubation time for most of the assays performed). In some experiments, the tubes were stoppered and the air was replaced with nitrogen. The reaction was terminated by the addition of 10% trichloroacetic acid. The supernatant obtained after centrifugation (6,000 x g, 10 min, 20°C) was assayed for the presence of hydroxamate and hydroxylamine nitrogen by iodine oxidation (13) with and without acid hydrolysis (for 30 min at 121°C in 6 N H2SO4 under N2), respectively. L-N5-OH-Orn was identified after ion-exchange chromatography and gel filtration. The hydroxylamine-positive material was loaded on a column (2 by 7 cm) of Dowex 50W-X8 (H+ form) resin and washed first with 100 ml of 0.1 N HCI and then with 50 ml of 6 N HCl. The latter fraction, containing all the hydroxylamine-posi-tive material, was freeze-dried, resuspended in 2 ml of water, and loaded onto a Sephadex G-10 column (2.5 by 52 cm, containing about 250 ml of resin) equilibrated and eluted with water. The hydroxylamine-positive material which was eluted at approximately 150 ml was freeze-dried and characterized as detailed above. Further analysis of L-N5-OH-Orn was carried out by performic acid oxidation (29). Glutamic acid formed upon performic acid oxidation of L-N5-OH-Orn was identified in an LKB Pharmacia 4400 amino acid analyzer. Pyoverdin from early-stationary-phase cultures in SM9 was purified by gel filtration and highpressure liquid chromatography as described elsewhere (1). Amino acid analysis was carried out after acid hydrolysis (6 N HCl, 12 h, 100°C, in sealed, evacuated Carius tubes) with an LKB Pharmacia 4400 analyzer. Pyochelin was extracted with ethyl acetate from Fe(III)-poor culture supernatants at pH 1.5 to 2.0 and analyzed by thin-layer chromatography on Silica Gel G (11).
A total of eight independent, nonfluorescent, pyoverdindefective (pvd) mutants were selected on SM9 agar plates after ethyl methanesulfonate mutagenesis. Thepvd mutants were of three distinct classes, designated pvd-1, pvd-2, and pvd-3. The pvd-1 mutant strain PALS124 produced neither hydroxylamine nor hydroxamate compounds when grown in SM9. Under these conditions, the pvd-2 mutant strain PALS128 synthesized low levels of hydroxylamine nitrogen (approximately 30 jiM), which also reacted quantitatively in the hydroxamate assay, while the pvd-3 mutants produced hydroxamate nitrogen at wild-type levels (from 95 to 272 ,uM) but no hydroxylamine compounds. The synthesis of hydroxylamine and hydroxamate compounds by the pvd-2 and pvd-3 mutants, respectively, was totally repressed by the addition of 100 ,uM FeCl3 to SM9, as was true also for pyoverdin synthesized by P. aeruginosa PAO1. In comparison with the wild type, the pvd-1 and pvd-2 mutants grew poorly in SM9 supplemented with 500 p,M 2,2'-dipyridyl, while the pvd-3 mutants grew as well as the parental strain, PAO1, did. However, all the pvd mutants produced pyochelin at wild-type levels (data not shown). Therefore, we hypothesized that the pvd-1 mutant might be blocked in the reactions generating the hydroxylamine groups of pyoverdin (i.e., the L-N5-OH-Orn residues) whereas the pvd-2 mutant might be impaired in the formation of the hydroxamate functions (by acylation of L-N5-OH-Orn). Indeed, the presence of L-N5-OH-Orn in culture supematants of strain PALS128 was confirmed by paper electrophoresis and amino acid analysis of the hydroxylamine-positive material purified by Dowex 50W-X8 ion-exchange chromatography and Sephadex G-10 gel filtration. Additional evidence for this hydroxylamine-positive material being the N5-hydroxy derivative of L-OM was provided by the formation of glutamic acid upon performic acid oxidation. The pvd-3 mutants are probably blocked in later steps of the biosynthetic pathway leading to pyoverdin, most likely in chromophore synthesis. Preliminary results allowed us to identify the hydroxamate-positive compound produced by the pvd-3A mutant PALS106 as an acidic, hydrophilic oligopeptide containing Asx-Ser-Glx-Gly-N5-OH-Orn, the apparent ratio being 1:2:3:1:2 (38a).
This molecule is endowed with a siderophore-like activity in the Schwyn and Neilands assay (37) and might be responsible for the growth of thepvd-3 mutants in SM9 supplemented with 500 ,uM 2,2'-dipyridyl.
To further characterize the nature of the pvd-1 mutation, biochemical complementation experiments were conducted.
Strains PALS124 (pvd-1) and PALS128 (pvd-2) and the prototypicpvd-3 mutant PALS106 were grown to stationary a Assays were carried out with cell extracts of bacteria grown for 10 h at 37°C in SM9. The assay system consisted of 4 ml of 100 mM sodium phosphate buffer, pH 7.0, containing 0.5 mg of lysozyme per ml, 10 ,ug of DNase per ml, 25 ,ug of RNase per ml, 5 mM EDTA, 1 mM dithiothreitol, 1 mM sodium pyruvate, 1 mM L-glutamine, 2 mM L-OM, and the equivalent of approximately 40 mg (dry weight) of bacterial lysate. The assay system without any added L-OrM served as a control. The values were corrected by subtracting the corresponding values of the control. An endogenous activity was noted in the controls of strains PAO1, PALS128, and PALS106 (<8% of that in the L-Orn-supplemented sample), which could arise from a small amount of L-Orn in the bacterial extract.
b Hydroxylamine nitrogen was determined by iodine oxidation (13) with hydroxylamine hydrochloride as the standard. The lower sensitivity limit of the method was 6.25 ,uM hydroxylamine nitrogen.
c Hydroxamate nitrogen was determined by the Csalky assay (13) after acid hydrolysis. The hydroxamate nitrogen concentration was corrected by subtracting the hydroxylamine nitrogen concentration for each sample. phase in SM9 supplemented with 400 ,uM L-N5-OH-Orn. Under this condition, pyoverdin production was fully restored in the pvd-I mutant but not in the other pyoverdindeficient strains. Conversion of L-N5-OH-Orn to pyoverdin by PALS124 never exceeded 60%, assuming a theoretical L-N5-OH-Orn/pyoverdin ratio of 2:1. This is probably because at pH values near or above the pK of the hydroxylamine group, which is about 5, L-N5-OH-Orn is oxidatively decomposed (14) and might be partially degraded in the culture medium during bacterial growth. L-N5-OH-Orn restored not only pyoverdin synthesis by the pvd-1 mutant but also its growth in SM9 supplemented with 2,2'-dipyridyl. This effect was not observed for the pvd-2 mutant. Furthermore, pyoverdin synthesis by the pvd-1 mutant fed with L-N5-OH-Orn was totally repressed upon addition of 100 ,uM FeCl3 to SM9 (data not shown).
Pyoverdin produced by biochemical complementation of the putative L-N5-OH-Orn auxotroph PALS124 was purified and analyzed in detail. The molecule showed fluorescence spectra identical to those of authentic pyoverdin from P. aeruginosa PAO1, with excitation and emission maxima at 405 and 455 nm, respectively. Spectrophotometric titrations of the siderophore with Fe(III) in 200 mM Tris-HCl at pH 8.8 showed a shift in the absorption maxima from 425 nm for the desferrisiderophore form to 410 nm for the ferrisiderophore, as also shown for pyoverdin from wild-type PAO1 (18). Amino acid analysis of the peptide moiety of the molecule following acid hydrolysis with HCl and reductive hydrolysis with HI gave the same composition as pyoverdin from P. aeruginosa PAO1, with Thr-Ser-N5-OH-Orn-Lys-Arg in the ratio 2:2:2:1:1.
The results of enzymatic analysis confirmed expectations based on phenotypic characterization and biochemical complementation tests. Cell-free lysates of wild-type PAO1 and pvd-2 and pvd-3 mutants formed hydroxylamine and/or hydroxamate nitrogen when incubated at 30°C under aeration with L-Orn but not with other putative substrates or precursors, e.g., L-Lys or L-Glu (23), at the same concentration. In contrast, the L-N5-OH-Orn auxotroph PALS124 was unable to convert L-Orn to the hydroxylamine and/or the hydroxamate derivative (Table 2). It appears also that the wild-type PAO1 and, to a greater extent, the pvd-3 mutant formed much higher amounts of hydroxamate than hydroxylamine nitrogen. This is probably because hydroxylation of L-Orn might be readily followed by formylation or cyclization in cell-free lysates of these strains. The requirement of molecular oxygen was demonstrated by the complete inhibition of the hydroxylation process when air was replaced with N2 in the reaction mixture. Hydroxylation was also affected by the growth stage of the cells and by the Fe(III) concentration of the culture medium; it was maximal in early-logarithmic-phase cells from Fe(III)-depleted medium and was absent when bacteria were grown to late stationary phase or in the presence of 100 puM FeCl3 (data not shown).
The three different types of mutations affecting pyoverdin production were mapped by chromosome mobilization with plasmid R68.45 (15) and by phage E79-tv2-mediated transduction (32). Initial attempts to score the pyoverdin-proficient phenotype by selection on 2,2'-dipyridyl plates failed because at high cell densities the pvd mutants were not completely inhibited by the maximal nontoxic concentration of the chelator. Therefore, the eight pvd mutants were converted into donors by conjugative transfer of plasmid R68.45 from strain PA025(R68.45) and mated with suitable streptomycin-resistant recipients. Exconjugants were selected for auxotrophic or catabolic markers and tested for coinheritance with pvd by being screened for fluorescence      Table 3. The three different types of pvd mutations were closely linked to catAl and mtu-9002, located at 47 and 48 min on the PAO1 map (19), respectively. A lower degree of linkage with met-9020 and met-9011, both located at 40 min, was shown. On the basis of these results, three-factor crosses were performed with strain PA04032. Recombinants at both met-9020 and mtu-9002 or at both catAl and mtu-9002 gave 100% coinheritance with pvd-1,pvd-2, and pvd-3.
The positions of the pvd mutations were also determined by phage E79-tv2-mediated transduction. Thepvd-1,pvd-2, and pvd-3 mutations were all cotransducible with catAI (Table  4). No transductional linkage was found with mtu-9002. These results are in agreement with previous reports (4,18,35) and indicate that the different mutations affecting pyoverdin biosynthesis might be clustered within the chromosome region near the genes responsible for benzoate catabolism, at 47 min of the revised PAO1 map (19).
The results presented in this article define at least three genetic defects affecting pyoverdin biosynthesis. This is a minimum estimation, since recent studies of different Pseudomonas spp. indicate that several genes are required for the biosynthesis of pyoverdins, in agreement with the structural complexity of these fluorescent siderophores (25,26,31,35). It has also been reported that the synthesis of the fluorescing group is preceded by the synthesis of the peptide part of the siderophore (26). Our results strengthen this hypothesis and indicate that the formation of L-N5-OH-Orn is an earlier step in the peptide synthesis by P. aeruginosa PA01. Physiological and enzymatic criteria indicate that the pvd-J mutant is an L-N5-OH-Orn auxotroph, lacking the L-OM-N5-hydroxylating enzyme. The pvd-2 mutant, producing low levels of hydroxylamine nitrogen identified as N5-OH-Orn, might be blocked in the following step of pyoverdin synthesis, i.e., the acylation of the hydroxylamine group. Accordingly, cell-free lysates of the pvd-2 mutant converted L-OM to the corresponding hydroxylamine but not to the hydroxamate derivative. All the other pyoverdin-deficient mutants (pvd-3) produced hydroxamate compounds in iron-poor medium and converted L-Orn to the hydroxamate derivative in a cell-free enzyme assay. Of note, the formation of hydroxamate and/or hydroxylamine compounds by the wild type and by pvd-2 and pvd-3 mutants both in liquid cultures and in a cell-free system was strictly iron regulated and oxygen dependent, as expected for a hydroxylating activity involved in siderophore synthesis.
In conclusion, although hydroxamate siderophores include a variety of structurally distinct chelators, the enzymes for the hydroxylation of the w-amino groups seem to have been conserved over a wide evolutionary range, from bacteria to fungi (2,3,14,17,20,39). These enzymes catalyze a unique microbial reaction, which does not occur in animal and plant cells, and hence they might constitute potential targets for new antimicrobial drugs capable of interfering with in vivo iron uptake by pathogenic microorganisms.