Molecular basis of pyoverdine siderophore recycling in Pseudomonas aeruginosa

Abstract

The siderophore pyoverdine (PVD) is a primary virulence factor of the human pathogenic bacterium Pseudomonas aeruginosa, acting as both an iron carrier and a virulence-related signal molecule. By exploring a number of P. aeruginosa candidate systems for PVD secretion, we identified a tripartite ATP-binding cassette efflux transporter, here named PvdRT-OpmQ, which translocates PVD from the periplasmic space to the extracellular milieu. We show this system to be responsible for recycling of PVD upon internalization by the cognate outer-membrane receptor FpvA, thus making PVD virtually available for new cycles of iron uptake. Our data exclude the involvement of PvdRT-OpmQ in secretion of de novo synthesized PVD, indicating alternative pathways for PVD export and recycling. The PvdRT-OpmQ transporter is one of the few secretion systems for which substrate recognition and extrusion occur in the periplasm. Homologs of the PvdRT-OpmQ system are present in genomes of all fluorescent pseudomonads sequenced so far, suggesting that PVD recycling represents a general energy-saving strategy adopted by natural Pseudomonas populations.

Pseudomonas aeruginosa is among the most dreaded Gram-negative pathogens in the hospital setting and the main cause of lung decline and death in patients suffering from cystic fibrosis (CF). In response to iron depletion, P. aeruginosa utilizes the endogenous siderophore pyoverdine (PVD) as the primary iron source. In addition to its role in nutrition, PVD is implicated in biofilm control, cell-to-cell communication, and virulence regulation (1). By a mechanism known as surface signaling, PVD tunes the expression of virulence factors, including exotoxin A, exoprotease PrpL, and PVD itself (2). Accordingly, PVD is essential for P. aeruginosa virulence in several animal models of infection, and it has been found at biologically significant concentrations in the sputum from CF patients (reviewed in refs. 1 and 3).

PVD-related siderophores are produced by all fluorescent pseudomonads, and consist of a conserved fluorescent chromophore linked to a short peptide differing among species and strains. PVD synthesis and transport have been studied most intensively in the type strain P. aeruginosa PAO1. Almost all genes required for PVD synthesis and uptake cluster at a single (pvd) locus in the PAO1 genome (4). PVDs are typically produced by nonribosomal peptide synthesis, through a pathway involving several synthetases and accessory enzymes that generate nonproteinogenic amino acid precursors (1). Other gene products are also essential for PVD production, but their actual role has not yet been established. Some of these have a periplasmic localization (5, 6), in accordance with preliminary evidence of PVD maturation in the periplasmic compartment (7).

The mechanism of PVD secretion is so far unknown. Involvement of the multidrug efflux pump MexAB-OprM in PVD secretion has been proposed (8), but not confirmed yet. One candidate for PVD transport from the cytoplasm to the periplasm is the ATP-binding-cassette (ABC) inner membrane (IM) transporter PvdE, which is essential for PVD production (9). In addition, the pvd locus contains an operon for a tripartite efflux system of the ABC superfamily (PA2389–90-OpmQ), whose expression mirrors that of PVD synthesis enzymes, being controlled by the PVD-specific sigma factor PvdS. Deletion of this system reduces, but does not abrogate, PVD production (4).

PVD-mediated iron uptake occurs by binding of the ferri-PVD complex to the outer membrane (OM) receptor FpvA (10, 11), while a secondary PVD receptor, called FpvB, provides a minor contribution to PVD uptake (12). Iron release from PVD has been shown to occur in the periplasm, and that apo-PVD is then recycled to the extracellular milieu (13, 14). However, no protein candidates have been proposed for these processes and further studies are required to elucidate the fate of iron and PVD following dissociation in the periplasm (15).

Here, we demonstrate that the tripartite efflux system PA2389–90-OpmQ is implicated in PVD recycling by P. aeruginosa, and that PVD recycling and export probably involve alternative pathways.

Results

PA2389, PA2390, and opmQ Mutants Accumulate PVD in the Periplasm.

In a preliminary search for candidate PVD secretion systems, we compared the PVD levels between culture supernatants of wild-type P. aeruginosa PAO1 and transposon mutants in genes encoding the MexAB-OprM, PvdE, and PA2389–90-OpmQ transport systems [supporting information (SI) Table S1]. No significant growth differences between wild-type PAO1 and putative PVD secretion mutants were observed under both iron-deplete (iron-deplete Casamino acids or DCAA medium) and -replete (DCAA medium plus 50 μM FeCl₃) conditions (Fig. S1). In the iron-deplete medium, PVD production was: (i) marginally affected by mutation in any of the mexA, mexB, and oprM genes; (ii) significantly reduced (approximately 50–60% of the wild-type levels) by mutation in any of the PA2389, PA2390, and opmQ genes; and (iii) completely abolished in the pvdE and pvdA mutants (Fig. 1A).

Fig. 1.

Extracellular and cell-associated PVD in candidate PVD export mutants of P. aeruginosa PAO1. Wild-type and mutant strains were grown for 14 h at 37 °C in DCAA, or in DCAA plus 50 μM FeCl₃ (+Fe). (A) Relative levels of extracellular PVD (black bars; Left y axis) were determined spectrophotometrically in culture supernatants. Intracellular PVD was determined fluorimetrically in 1-ml cell lysates from 5 × 10⁹ cells (white bars; Right y axis). Values are mean (± SD) of three independent assays. (B) Confocal microscopy of P. aeruginosa mutant cells. For each strain/condition, the left and right panels show the visible and the corresponding 405-nm laser line confocal microscopy pictures, respectively.

We took advantage of the fluorescent properties of apo-PVD to determine its amount in cell lysates of wild-type PAO1 and isogenic mutants by fluorimetric measurements (see Fig. 1A). The mexA, mexB, and oprM mutants showed intracellular levels of PVD comparable to or even lower than those of the wild type, consistent with their profile of PVD production. PVD was undetectable in cell lysates of the pvdE mutant, suggesting that PvdE is not involved in PVD secretion or, alternatively, that it directs the export of a nonfluorescent PVD precursor named ferribactin (1, 7). Conversely, PA2389, PA2390, and opmQ mutants accumulated five- to sixfold higher amounts of PVD than wild-type PAO1, suggesting an involvement of these genes in PVD export. Therefore, the PA2389 and PA2390 genes were renamed pvdR and pvdT, respectively, to emphasize their role in PVD transport or recycling.

To visualize PVD accumulation by pvdR, pvdT, and opmQ mutants, intracellular PVD was detected in living cells by confocal laser-scanning microscopy (Fig. 1B). The fluorescence emission levels of pvdR-, pvdT-, and opmQ-mutant cells largely exceeded that of wild-type cells. No fluorescence was detected in the PVD-defective pvdA mutant or in iron-replete wild-type cells (see Fig. 1B and Fig. S2), validating the identification of the cell-associated fluorescent molecule as apo-PVD. Once more, mexA-, mexB-, and oprM-mutant cells appeared similar to wild-type PAO1, thus excluding the involvement of MexAB-OprM in PVD export. No fluorescence was detected in pvdE-mutant cells.

To determine the cellular compartment where PVD is accumulated by pvdR, pvdT, and opmQ mutants, cell-fractionation experiments were performed. For this purpose, the periplasm was extracted from P. aeruginosa cells and the amount of PVD was compared between the periplasmic fraction and the resulting spheroplasts. Notably, nearly all intracellular PVD was associated with the periplasmic fraction of pvdR, pvdT, and opmQ mutants (Fig. 2), irrespective of the mutation. Complementation of the pvdR, pvdT, and opmQ mutations with the plasmid-borne pvdRT-opmQ operon restored nearly wild-type levels of both extracellular and periplasmic PVD, as demonstrated by fluorimetric and microscopic analyses (see Figs. 1 and 2). Therefore, the entire PvdRT-OpmQ system (i.e., the IM ABC transporter PvdT, the periplasmic fusion protein PvdR, and the OM porin OpmQ) appears to be involved in the transport of PVD from the periplasm to the extracellular milieu.

Fig. 2.

Subcellular localization of cell-associated PVD in pvdR, pvdT, and opmQ mutants. Strains were grown for 14 h at 37 °C in DCAA. PVD was determined fluorimetrically in 1-ml periplasmic (gray bars) and spheroplasts (white bars) fractions from 5 × 10⁹ cells. Values are mean (± SD) of three independent assays.

The PvdRT-OpmQ System Is Responsible for PVD Recycling.

PVD-mediated iron uptake occurs through the OM receptor FpvA, which translocates ferri-PVD into the periplasmic space. While iron is internalized into the cytoplasm, PVD is recycled to the extracellular milieu by a yet unknown mechanism (13, 14). Because the PvdRT-OpmQ efflux system is implicated in PVD export from the periplasmic space, its function in PVD recycling was investigated. To this aim, the pvdA mutant and the pvdApvdR and fpvApvdA double mutants, all defective in PVD synthesis, were grown in iron-deplete medium supplemented or not with 50-μM exogenous PVD.

Confocal microscopy and fluorimetric measurements of PVD in subcellular fractions showed that the pvdApvdR double mutant accumulated in the periplasm approximately sixfold higher amounts of PVD than the pvdA mutant (Fig. 3 A and B). Periplasmic PVD levels were strongly reduced in the fpvApvdA double mutant lacking the FpvA receptor for the uptake of extracellular PVD, and completely absent in the pvdApvdR mutant incubated without PVD (see Fig. 3 A and B). The kinetics of exogenous PVD (50 μM) accumulation by iron-deplete cells of pvdA, pvdApvdR, and fpvApvdA mutants were consistent with the above results. In pvdA cells, the amount of internalized PVD reached a plateau between 30 and 60 min of incubation with exogenous PVD, while pvdApvdR cells accumulated PVD for an additional hour, attaining more than threefold higher concentration of intracellular PVD than pvdA cells. In contrast, fpvApvdA cells accumulated very low levels of PVD (Fig. 3C), as expected for a PVD uptake mutant.

Fig. 3.

PVD recycling from the periplasmic space. (A) Confocal microscopy of P. aeruginosa cells grown for 14 h at 37 °C in DCAA with or without 50-μM exogenous PVD. For each strain/condition, the left and right panels show the visible and the corresponding 405-nm laser line confocal microscopy pictures, respectively. (B) Subcellular localization of PVD recycling. PVD was determined in periplasmic (gray bars) and spheroplasts (white bars) fractions as described in Fig. 2. (C) Kinetics of intracellular accumulation of PVD by iron-deplete cells incubated with 50-μM exogenous PVD. (D) PVD recycling (white symbols) and PVD accumulation (black symbols) by iron-deplete intact cells incubated with 1 μM ferri-PVD. Symbols for (C) and (D) are: pvdA, diamonds; pvdApvdR, squares; fpvApvdA, circles. (E) Utilization of PVD as iron source by pvdA (black bars) and pvdApvdR (white bars) measured as growth yields after 14 h in DCAA plus 600 μM dipyridyl and exogenously-added PVD (0–80 μM). All values are mean (± SD) of three independent assays.

Next, we sought to investigate the involvement of the PvdRT-OpmQ efflux system in recycling PVD after FpvA-mediated ferri-PVD uptake. To this aim, iron-deplete cells of pvdA, pvdApvdR, and fpvApvdA mutants were incubated with the nonfluorescent ferri-PVD complex (1 μM). Recycling of apo-PVD was monitored fluorimetrically as the appearance of PVD-specific fluorescence in culture supernatants (14). The pvdA strain efficiently recycled apo-PVD in the culture medium, while both the pvdApvdR and fpvApvdA double mutants did not (Fig. 3D). Nevertheless, pvdApvdR cells accumulated apo-PVD (see Fig. 3D), indicating that they are able to remove iron from ferri-PVD in the periplasm but are unable to recycle apo-PVD. Moreover, pvdA and pvdApvdR cells showed similar ability to use PVD as an iron source, as determined by a feeding assay in which exogenously added PVD was used to sustain bacterial growth under conditions of extreme iron restriction (DCAA plus 600 μM dipyridyl) (Fig. 3E). Taken together, our results substantiate the hypothesis of Greenwald et al. (14), according to which PVD recycling and PVD-iron dissociation in the periplasm are independent processes.

The PvdRT-OpmQ System Is Not Involved in Export of Endogenous PVD.

To assess the involvement of the PvdRT-OpmQ efflux system in secretion of de novo synthesized PVD, we investigated the effect of PvdRT-OpmQ inactivation in an FpvA-deficient mutant, which retains the ability to produce PVD (2) but is severely impaired in its uptake (see Fig. 3). Remarkably, the amount of PVD stored in the periplasm did not differ significantly between fpvA and fpvApvdR mutants, while it was much lower than that of PAO1 and pvdR mutant (Fig. 4A). Extracellular PVD levels were also comparable between fpvA and fpvApvdR (see Fig. 4A), denoting no defect in secretion of de novo produced PVD by the fpvApvdR mutant. Accordingly, confocal microscopy failed to reveal significant intracellular accumulation of PVD by both fpvA and fpvApvdR mutants (Fig. 4B). The barely detectable fluorescence shown by fpvApvdR cells was comparable to that of a fpvApvdRpvdA triple mutant fed with exogenous PVD (see Fig. 4B), suggesting that the weak fluorescence detectable in the fpvApvdR background originates from impaired recycling of some exogenous PVD acquired via an FpvA-independent route, likely through the secondary receptor FpvB. Indeed, no residual fluorescence was detected in a tonB1pvdR double mutant (see Fig. S2), in which receptor-dependent siderophore uptake is completely abrogated (16).

Fig. 4.

Effect of impaired PVD uptake on PVD accumulation in the periplasm. P. aeruginosa mutants were grown for 14 h at 37 °C in DCAA. (A) Quantification of extracellular (black bars, Left y axis), periplasmic and spheroplasts-associated PVD (gray and white bars, respectively; Right y axis). Values are mean (± SD) of three independent assays. (B) Confocal microscopy images of cells. For each strain, the left and right panels show the visible and the corresponding 405-nm laser line confocal microscopy pictures, respectively.

In conclusion, these results indicate that periplasmic PVD accumulation by PvdRT-OpmQ-deficient P. aeruginosa cells originates from impaired recycling of exogenous PVD, rather than from a defect in secretion of endogenous PVD.

Discussion

Prerequisites for a system to be implicated in any step of PVD secretion are the accumulation of intracellular PVD by the corresponding mutants, together with reduced or impaired PVD secretion. We therefore addressed the role of MexAB-OprM, PvdE, and PvdRT-OpmQ systems in PVD secretion by measuring the amount of secreted PVD and the level of PVD-related fluorescence retained by P. aeruginosa cells. Our results rule out any significant contribution of the MexAB-OprM pump to PVD secretion (see Fig. 1), while highlighting an essential role of PvdE in PVD biogenesis, as shown by the PVD-deficient phenotype of the pvdE mutant. In fact, PVD-related fluorescence was undetectable in pvdE mutant cells (see Fig. 1), and no hydroxamate-containing PVD precursors could be detected in pvdE culture supernatants (SI Text). We therefore suggest that PvdE could export the nonfluorescent PVD precursor ferribactin from the cytosol to the periplasm without being involved in the secretion of mature PVD into the extracellular milieu. Accordingly, maturation of the PVD chromophore has been proposed to occur in the periplasm (7).

Only mutation of the PvdRT-OpmQ exporter resulted in both reduction of extracellular PVD levels and remarkable intracellular accumulation of PVD (see Fig. 1). PvdRT-OpmQ displays the typical architecture of Gram-negative tripartite ABC-type exporters, consisting of an IM component with both ATPase and permease domains, a periplasmic membrane fusion protein, and a TolC-like OM protein (see SI Text for details). The pvdRT-opmQ gene cluster was found to be iron-regulated and under the tight control of the PVD-specific PvdS sigma factor (Table S2), in line with previous transcriptomics data (4).

Cell fractionation assays showed that PVD is accumulated in the periplasm of pvdRT-opmQ mutants, independent of the deleted component of the tripartite system (see Fig. 2), thus indicating that the PvdRT-OpmQ system is involved in secretion of PVD from the periplasmic space to the extracellular milieu. Previous studies localized the substrate binding site of some resistance-nodulation division-type (RND) efflux pumps in the periplasmic compartment, and suggested that substrate extrusion from the periplasm could represent a strategy to protect the cytoplasm from potential hazards (17–19). This work provides substantial evidence that ABC exporters can also extrude substrates from the periplasm, in line with a previous report showing that the macrolide efflux pump MacAB-TolC of Escherichia coli is also implicated in secretion of periplasmic enterotoxin II (20). Although PvdRT-OpmQ and MacAB-TolC share remarkable structural similarity (SI Text), the PvdRT-OpmQ efflux system does not provide any contribution to P. aeruginosa resistance to antibiotics belonging to the classes of tetracyclines, aminoglycosides, β-lactams and macrolides, and chloramphenicol (Table S3).

Periplasmic accumulation of fluorescent PVD in P. aeruginosa cells deficient in PvdRT-OpmQ could be explained by two nonmutually exclusive mechanisms. According to the “de novo pathway,” PVD production could require PvdE-dependent transport of the ferribactin precursor through the IM. Ferribactin would undergo maturation in the periplasm before extracellular release through a secretion device whose impairment would cause periplasmic accumulation of PVD. In the “recycling pathway,” mature PVD would enter the periplasm by FpvA-mediated uptake, where it would be retained in the absence of a recycling system.

To verify the involvement of PvdRT-OpmQ in either of these processes, accumulation of PVD by PvdRT-OpmQ-deficient cells was assessed in PVD synthesis- or uptake-defective backgrounds. With respect to the biosynthetic (pvdA) mutant, the pvdApvdR double mutant accumulated higher amounts of exogenously-provided PVD in the periplasm (see Fig. 3 A–C), and was impaired in recycling of periplasmic PVD (see Fig. 3D). On the other hand, comparable levels of extracellular PVD and no periplasmic PVD retention were observed in the PVD-producing uptake-mutant fpvA, irrespective of the pvdR mutation (see Fig. 4). These results lead us to propose that the PvdRT-OpmQ system is responsible for exogenous PVD recycling, while excluding its involvement in secretion of de novo synthesized PVD. Coherently, P. aeruginosa mutants with an impaired PvdRT-OpmQ system secrete up to 180 μM PVD in DCAA medium (see Fig. 1 and ref. 4), and our preliminary results suggest that the lower PVD levels in their culture supernatants are a result of reduced PVD synthesis rather than of impaired secretion of de novo synthesized PVD. A schematic of PVD trafficking in P. aeruginosa is shown in Fig. 5.

Fig. 5.

PVD trafficking across the P. aeruginosa cell envelope. Ferribactin is synthesized in the cytoplasm and secreted by PvdE into the periplasm, where maturation to PVD is suggested to take place (1, 7). Export of de novo synthesized PVD occurs by a still uncharacterized mechanism. Ferri-PVD is translocated by FpvA into the periplasm, where Fe(III) is dissociated, plausibly by reduction to Fe(II) (14). Internalized PVD is recycled by the PvdRT-OpmQ system. It is hypothesized that Fe(II) enters the cytoplasm by a classic periplasmic binding protein (PBP)/IM ABC importer (15). Experimentally confirmed and hypothetical steps are in yellow and gray, respectively.

What is the benefit for P. aeruginosa of having a system for PVD recycling? The energetic cost of PVD synthesis is prohibitive, and this is confirmed by the fact that PVD producers are outcompeted by nonproducers in growth competition assays (21). Although PVD recycling does not provide a great advantage to iron-deplete laboratory cultures, where the PVD concentration attains up to 350 μM, circumstances may differ in natural environments, such as the CF lung, where PVD is approximately 1 μM (22). Moreover, P. aeruginosa adopts a biofilm mode of growth in the CF lung, which could further reduce the local availability of PVD. Notably, PVD-deficient mutants have been isolated from sputa of CF patients (23, 24), indicating that P. aeruginosa occurs in the CF lung as a mixed community of PVD producers and PVD-deficient cheaters. In vitro and in vivo studies demonstrated that both growth and virulence of these mixed populations are negatively affected by increasing proportions of PVD cheaters over PVD producers (reviewed in ref. 25). The ability to recycle PVD and, thus, to make it available for virtually endless cycles of iron uptake, would allow the P. aeruginosa community to maintain PVD levels at biologically meaningful concentrations with minor costs for the whole population. The biological relevance of an energy-saving device for PVD recycling is testified by the presence of pvdRT-opmQ orthologs in the genome of all fluorescent pseudomonads sequenced so far (26) (Table S4).

Materials and Methods

Growth Conditions.

Bacterial strains and plasmids used in this study are listed in Table S1. P. aeruginosa transposon insertion mutants were purchased from a P. aeruginosa Transposon Mutant Library (www.genome.washington.edu/UWGC/pseudomonas). P. aeruginosa was grown at 37 °C in DCAA medium (27), supplemented or not with 50 μM FeCl₃ as iron-deplete and -replete conditions, respectively. Exogenous PVD was added as PVD-conditioned medium (28), corresponding to a culture filtrate of the pyochelin-defective P. aeruginosa PA6331 grown in DCAA medium. As a control, the medium conditioned by the PVD- and pyochelin-defective PA6331pvdA was used.

Construction of Deletion Mutants and Complementing Plasmids.

E. coli was routinely used for recombinant DNA manipulations. Site-specific excision of pvdR and pvdA genes was performed as previously described (29). For complementation of pvdR, pvdT, and opmQ mutations, a 4.7-kb fragment encompassing the pvdRT-opmQ operon with its putative promoter (region 2642032–2646727 of the P. aeruginosa PAO1 chromosome, www.pseudomonas.com) was amplified by PCR and cloned in pUCP19, yielding plasmid pUCPpvdRT-opmQ (see Table S1).

Subcellular Fractions.

Intracellular PVD determinations were performed on P. aeruginosa cells grown for 14 h in DCAA, and washed three times with 30 mM Tris-HCl (pH 7), 150 mM NaCl. Bacterial pellets (5 × 10⁹ cells) were resuspended in 1 ml of 10 mM Tris-HCl (pH 8), 100 mM NaCl, and lysed by sonication. Cell debris were removed by low-speed centrifugation (7,000 × g, 10 min, 4 °C), and PVD concentration was determined using appropriate dilutions of cell lysates in 100 mM Tris-HCl (pH 8).

Periplasmic fractions were obtained by spheroplasting P. aeruginosa cells with lysozyme and MgCl₂ as described (30). Briefly, bacterial pellets (5 × 10⁹ cells) were suspended in 1 ml of 10 mM Tris-HCl (pH 8.4), 200 mM MgCl₂, 0.5 mg/ml lysozyme, and incubated for 30 min at room temperature with gentle shaking. Suspensions were centrifuged (11,000 × g, 15 min, 4 °C) to collect the periplasm-containing supernatants. The resulting pellets were suspended in 1 ml of 10 mM Tris-HCl (pH 8), 100 mM NaCl, and lysed by sonication. Cell debris were removed to obtain the spheroplasts fractions. The specificity and purity of cell fractions were verified by using specific subcellular protein markers as described (30).

PVD Measurements.

PVD from culture supernatants was diluted in 100 mM Tris-HCl (pH 8), and measured as OD₄₀₅ normalized by the cell density of bacterial cultures (OD₆₀₀). Intracellular PVD was quantified fluorimetrically by recording the emission at 455 nm upon excitation at 398 nm in a LS50B Luminescence spectrometer (Perkin–Elmer), using a standard curve generated with known PVD concentrations (0–3 μM in 100 mM Tris-HCl, pH 8).

Kinetics of PVD Uptake and Recycling.

Kinetic studies of PVD uptake were performed by incubating 3 × 10⁹ cells/ml in the presence of 50 μM PVD in DCAA medium supplemented with 300 μg/ml kanamycin (Km) and 100 μg/ml chloramphenicol (Cm) to halt bacterial growth. At given time points, cells from 1-ml samples were washed, lysed by sonication, and cell debris removed by centrifugation for fluorimetric PVD measurements.

PVD recycling studies were performed by incubating 2 × 10⁹ cells/ml with 1 μM ferri-PVD in DCAA medium supplemented with 300 μg/ml Km and 100 μg/ml Cm. At given time points, 1-ml samples were centrifuged, and the appearance of apo-PVD was monitored fluorimetrically in both supernatants and intact cells.

Confocal Microscopy.

Cells (5 × 10⁸) from iron-deplete or -replete cultures were washed and mounted onto agarose-coated slides as described (31). Images were electronically acquired through a Leica SP5 confocal laser scanning microscope (Leica Microsystems) using a violet laser diode emitting at 405 nm. Laser intensity was kept constant at 11% for all images, without adjustments of contrast and brightness.

Note Added in Proof.

Since the work described in this paper was completed and submitted for publication, the involvement of PvdE in ferribactin secretion from the cytoplasm has been proposed by Yeterian E, et al. (32).