Stereospecificity of the Siderophore Pyochelin Outer Membrane Transporters in Fluorescent Pseudomonads*

Pyochelin (Pch) and enantio-pyochelin (EPch) are enantiomer siderophores that are produced by Pseudomonas aeruginosa and Pseudomonas fluorescens, respectively, under iron limitation. Pch promotes growth of P. aeruginosa when iron is scarce, and EPch carries out the same biological function in P. fluorescens. However, the two siderophores are unable to promote growth in the heterologous species, indicating that siderophore-mediated iron uptake is highly stereospecific. In the present work, using binding and iron uptake assays, we found that FptA, the Fe-Pch outer membrane transporter of P. aeruginosa, recognized (Kd = 2.5 ± 1.1 nm) and transported Fe-Pch but did not interact with Fe-EPch. Likewise, FetA, the Fe-EPch receptor of P. fluorescens, was specific for Fe-EPch (Kd = 3.7 ± 2.1 nm) but did not bind and transport Fe-Pch. Growth promotion experiments performed under iron-limiting conditions confirmed that FptA and FetA are highly specific for Pch and EPch, respectively. When fptA and fetA along with adjacent transport genes involved in siderophore uptake were swapped between the two bacterial species, P. aeruginosa became able to utilize Fe-EPch as an iron source, and P. fluorescens was able to grow with Fe-Pch. Docking experiments using the FptA structure and binding assays showed that the stereospecificity of Pch recognition by FptA was mostly due to the configuration of the siderophore chiral centers C4″ and C2″ and was only weakly dependent on the configuration of the C4′ carbon atom. Together, these findings increase our understanding of the stereospecific interaction between Pch and its outer membrane receptor FptA.

To access iron, aerobic bacteria produce siderophores which form complexes with Fe 3ϩ in the environment and deliver it via specific membrane transporters to the bacteria (1). The energy required for this process is provided by the proton motive force of the inner membrane by means of an inner membrane complex comprising TonB, ExbB, and ExbD (2,3). Pyochelin (Pch), 2 which is the focus of this study, is produced as a secondary siderophore under iron limitation by almost all strains of Pseudomonas aeruginosa and some closely related bacteria (4 -7). In pseudomonads, secondary siderophores (pyochelin, thioquinolobactin, etc.) are usually produced in smaller amounts, demonstrate lower iron affinity, and are rather simple molecules compared with pyoverdin (the main siderophore). It is hypothesized that under certain environmental or physiological conditions, secondary siderophores provide sufficient iron to the cell or fulfill functions other than iron sequestration (8). Pch was isolated for the first time from P. aeruginosa ATCC 15692 by Liu and Shokrani (9), and its structure was established later by Cox et al. (10) as (4ЈR,2ЉR,4ЉR)-2Ј-(2-hydroxyphenyl)-3Љ-methyl-4Ј,5Ј,2Љ,3Љ,4Љ,5Љ-hexahydro-[4Ј,2Љ]bithiazolyl-4Љ-carboxylic acid with three chiral centers located at C4Ј, C2Љ, and C4Љ (see Fig. 1). Pch is synthesized by P. aeruginosa from salicylate and two molecules of cysteine via a thiotemplate mechanism (11,12) involving proteins encoded by the two biosynthetic operons pchDCBA and pchEFGHI (13,14). Pch biosynthesis is autoregulated by a positive-feedback loop (14) requiring the transcriptional regulator PchR together with Pch as an effector molecule (15,16).
In the extracellular medium Pch chelates Fe 3ϩ with a 2:1 (Pvd:Fe 3ϩ ) stoichiometry (10,17,18), with one molecule of Pch tetradentately coordinated to Fe 3ϩ and the second molecule bound bidentately to complete the octahedral geometry (19). Pch has poor water solubility, and its Fe 3ϩ affinity was determined in ethanol as 2 ϫ 10 5 M (10). This is low compared with other siderophores, and it is possible that in aqueous solution at physiological pH the affinity for iron is higher. Once loaded with Fe 3ϩ , Pch is recognized at the cell surface of P. aeruginosa and transported into the periplasm by a specific outer membrane transporter, FptA. The structure of this protein (20) is typical of this class of transporters; a transmembrane 22-␤stranded barrel occluded by an N-terminal domain containing a mixed four-stranded ␤-sheet. The Pch binding pocket is principally composed of hydrophobic and aromatic residues, con-sistent with the hydrophobicity of this siderophore. In the x-ray FptA structure, a Fe-Pch complex was found in the binding site of FptA, where Pch provided a tetradentate coordination of iron. Ethylene glycol, originating from the purification procedure, provided the remaining bidentate coordination (20). The presence of such a complex in the x-ray FptA structure showed that one molecule of Pch is sufficient for a ferric Pch complex to be recognized by the transporter. The second chelator providing the remaining bidentate coordination does not interact with the transporter and can, therefore, be a different molecule, e.g. the siderophore cepabactin, which was found in such a ternary complex (21,22).
Recently it was shown that Pseudomonas fluorescens CHA0 produces the siderophore enantio-pyochelin (EPch, see Fig. 1), the optical antipode of Pch (23). EPch was found to promote growth under iron-limiting conditions in P. fluorescens and to act as an inducer of its own biosynthetic genes (23), thus resembling Pch, which is both a siderophore and a signaling molecule in P. aeruginosa (14,18). However, EPch did not promote growth in P. aeruginosa and neither did Pch in P. fluorescens (23), suggesting that both the EPch-and Pch-mediated iron uptake machineries are highly stereospecific in the two species. Fe-EPch is transported in P. fluorescens CHA0 by a homolog of the P. fluorescens Pf-5 transporter PFL_3498, which we have named FetA (23). Sequence comparison between FptA of P. aeruginosa and FetA of P. fluorescens showed that the two receptors are not closely related and that the amino acids interacting with Pch in the crystallized FptA receptor (20) are not conserved in FetA. We, thus, wondered whether FptA and FetA would be able to interact with the siderophore of the other species.
Using 55 Fe, we show here that both receptors, FptA and FetA, are highly stereoselective and do not bind and transport the ferrisiderophore of the other species. We also show by growth promotion experiments that the preference of P. aeruginosa and P. fluorescens for Pch and EPch, respectively, can be inverted by swapping the siderophore uptake systems between the two species. From our docking experiments on the FptA structure and from binding assays, we conclude that the stereospecificity of the FptA-Pch interaction relies to a large extent on the stereochemical configuration at the chiral centers C2Љ and C4Љ, whereas the conformation at C4Ј is less important.

EXPERIMENTAL PROCEDURES
Chemicals-Pch, neopyochelin (NeoPch), and EPch were synthesized and purified according to previously published protocols (23,24). Carbenicillin disodium salt was provided from Euromedex. The protonophore carbonyl cyanide m-chlorophenylhydrazone (CCCP) was purchased from Sigma. 55 FeCl 3 was obtained from PerkinElmer Life Sciences with a specific activity of 93.76 Ci/g. This radioactive 55 Fe solution was diluted with nonradioactive iron to 9.4 Ci/g. Siderophore- 55 Fe complexes ( 55 Fe-Pch, 55 Fe-NeoPch, or 55 Fe-EPch) were prepared at concentrations of 10 M 55 Fe with a siderophore:iron (mol:mol) ratio of 20:1. The solutions were prepared using a 10 mM solution of siderophores (in methanol). To 20 l of these solutions were added 40 l of a solution of 55 FeCl 3 (250 M, 9.4 Ci/g in HCl, 0.5 N), obtained by dilution of the stock solution.
The mixtures were incubated 15 min at room temperature before the addition of 940 l of 50 mM Tris-HCl (pH 8.0).
Media and Growth Conditions-Bacteria were routinely grown on nutrient agar and in nutrient yeast broth or LB (25,26) at 37°C (P. aeruginosa and Escherichia coli) or 30°C (P. fluorescens). When P. fluorescens and P. aeruginosa cells were used for binding and iron uptake assays, growth occurred in succinate medium (composition was 6.0 g/liter K 2 HPO 4 , 3.0 g/liter KH 2 PO 4 , 1.0 g/liter (NH 4 ) 2 SO 4 , 0.2 g/liter MgSO 4 ⅐7H 2 O, 4.0 g/liter sodium succinate, and the pH was adjusted to 7.0 by the addition of NaOH (27)) at 30°C. Siderophore-mediated growth promotion experiments were performed in minimal medium M9 (25) with 0.5% glycerol as a carbon source. Iron limitation was achieved in this medium by adding the iron chelator 2,2Ј-dipyridyl at 500 M. Where necessary, antibiotics were added to growth media at the following concentrations: 100 g/ml ampicillin, 100 g/ml streptomycin, and 25 g/ml tetracycline for E. coli and 100 g/ml tetracycline for Pseudomonas. To counterselect E. coli donor cells in gene replacement experiments, chloramphenicol was used at a concentration of 10 g/ml. Mutant enrichment was performed with tetracycline at 20 g/ml and carbenicillin (for P. aeruginosa) at 2 mg/ml or cycloserine (for P. fluorescens) at 50 mg/ml, respectively.
Strains and Plasmids-All strains and plasmids used in this study are listed in Table 1. Gene replacement mutants were generated with suicide plasmids as described previously (28 -30). Briefly, the suicide plasmids were mobilized from E. coli to P. aeruginosa or P. fluorescens using the helper plasmid pME497 and chromosomally integrated, with selection for tetracycline resistance. Excision of the vector via a second crossing-over was obtained by enrichment for tetracycline-sensitive cells using carbenicillin (for P. aeruginosa) or cycloserine (for P. fluorescens). The suicide plasmid pME7531 used to generate a 1025-bp deletion in the fetA gene (ϭPFL_3498 in P. fluorescens Pf-5) was constructed as follows. Two PCR fragments of 0.6 kilobase carrying the 3Ј and 5Ј ends of fetA, respectively, were generated from chromosomal DNA of P. fluorescens CHA0 using the primers rez-1 (ACGTGAATTCATGGG-TAGCCGCGTTGCGC) together with rez-2 (ACGTGGATC-CGATCTGGTTGACCACGCCC) and using rez-3 (ACGTG-GATCCGCAATCGGAAGTGGGGGTC) together with rez-4 (ACGTAAGCTTACCAGCGATAGCGCAGGCT). Fragment 1, cleaved with EcoRI and BamHI, was ligated to BamHI-and HindIII-trimmed fragment 2 and cloned into the suicide vector pME3087 between the EcoRI and HindIII sites. Plasmid pME7531 was then introduced into P. fluorescens CHA400 and CHA1085 to generate the corresponding mutants CHA1080 and CHA1169, respectively.
The pyoverdine-negative P. aeruginosa mutant PAO6382 was obtained by deleting the pvdF gene in strain PAO1 using the previously described suicide plasmid pME7152 (31). All gene replacement mutants were checked by PCR.
To swap the Pch and EPch transport genes between the two Pseudomonas species, the genes were cloned under the control of the lac promoter in pME6000 as follows. The P. aeruginosa Pch transport genes fptABCX were excised from pME7036 on a 5.9kilobase HindIII-XbaI fragment and ligated to HindIII-XbaIcleaved pME6000 to give pME9629. To generate plasmid pME9630, the CHA0 homologs of PFL_3498 to PFL_3503, which encode FetA and a putative ABC transporter, were PCR-amplified from chromosomal DNA of CHA0 using primers XL1-ftp (ACGTGGATCCTGCTCGAGCCTGCTCCATCC) and XL1rtp (ACGTAAGCTTTCGTTCGCCCGAAGGTTACC). The resulting 7.5-kilobase fragment was trimmed with XhoI and Hin-dIII and cloned into pME6000, cleaved with the same enzymes.
DNA Manipulation and Sequencing-Small-and large-scale preparations of plasmid DNA were made with the QIAprep Spin Miniprep kit (Qiagen, Inc.) and Jetstar kit (Genomed GmbH), respectively. DNA fragments were purified from agarose gels with the Geneclean II kit (Bio 101, La Jolla, CA) or the MinElute and QIAquick Gel Extraction kits from Qiagen. DNA manipulations were performed according to standard procedures (25). Transformation of E. coli, P. aeruginosa, and P. fluorescens was carried out by electroporation (32). Sequencing was performed with the BigDye Terminator Cycle Sequencing kit and an ABI-PRISM 373 automatic sequencer (Applied Biosystems) or was carried out commercially. The DNA sequence of the P. fluorescens CHA0 fetA gene was deposited at GenBank TM under accession number FJ641192. Data base searches were conducted at NCBI using BLAST algorithms.
Ligand Binding Assays Using 55 Fe-The in vivo dissociation constants (K d ) of 55 Fe-Pch, 55 Fe-NeoPch, and 55 Fe-EPch complexes to FptA and FetA were determined according to the following procedure (33). PAD07 and CHA1085 cells were washed twice with an equal volume of fresh medium and resuspended in 50 mM Tris-HCl (pH 8.0) buffer at an A 600 of 0.3 and 1.5, respectively. Cells were then first incubated for 15 min in the presence of 200 M CCCP to avoid iron uptake and afterward for 1 h at 0°C in a final volume of 500 l with 0.1-200 nM 55 Fe-Pch, 55 Fe-NeoPch, or 55 Fe-EPch, prepared as described above under "Chemicals." The mixtures were then centrifuged at 12,000 ϫ g for 2 min, and the supernatants containing the unbound siderophore-55 Fe were removed. The tubes containing the cell pellets were counted for radioactivity in scintillation mixture. To evaluate a nonspecific binding of 55 Fe-Pch, 55 Fe-NeoPch, or 55 Fe-EPch, binding experiments were repeated in parallel for each concentration of siderophore with the same strains grown in LB medium. In this iron-rich medium, neither FptA nor FetA receptors were expressed. K d values were determined using the Scatchard representation.
Iron Uptake-Iron uptake assays were carried out as previously reported for the FptA/Pch system (22). After an overnight culture (20 h), bacteria were prepared in 50 mM Tris-HCl (pH 8.0) at A 600 of 1 for P. aeruginosa or at A 600 of 5 for P. fluorescens and incubated at 37°C. Transport assays were started by adding 100 nM 55 Fe-Pch, 55 Fe-NeoPch, or 55 Fe-EPch prepared as described above under "Chemicals." To separate siderophore- 55 Fe transported into P. aeruginosa cells from unbound siderophore- 55 Fe, aliquots (100 l) of the suspensions were removed at different times and filtered, and the retained radioactivity was counted in scintillation mixture. For P. fluorescens, aliquots (300 l) of the cell suspensions were removed, mixed with 600 l of cold Tris buffer, and centrifuged (12,000 ϫ g for 3 min). The supernatants containing the unbound siderophore were removed, and the tubes containing the cell pellets were counted for radioactivity in scintillation mixture. This difference in the protocols used for P. aeruginosa cells and P. fluorescens cells is due to different levels of FptA and FetA expression, which led us to use much higher cell concentrations for experiments with P. fluorescens. As controls, the experiments were repeated with cells incubated at 0°C in the presence of 200 M CCCP and in the absence of cells.
Docking-Standard settings of the docking program GOLD Version4.0 (34) were used to dock Fe-PchI and Fe-EPchI. Ligand three-dimensional structures (mol2 format) were obtained from two-dimensional sketches using Marvin Sketch (ChemAxon Ltd) and further converted into three-dimensional coordinates using Corina (Molecular Networks GmbH). The iron-bound EPchI molecule assumed a tetravalent coordination of the iron atom, as seen in the FptA-bound structure of FptA (20). The FptA binding cavity was defined as any atom within a 6.5 Å radius sphere centered on the center of mass of PchI in the ptA-PchI x-ray structure (20). Docking poses were further analyzed by docking score (Goldscore), and the percentage of ligand surface was buried upon binding.

RESULTS
Binding of Pch Diastereoisomers to FptA and FetA-The specificity of iron uptake in Gram-negative bacteria is regulated at the cell surface by siderophore outer membrane transporters. These proteins are highly specific for one or a few siderophores. To investigate the binding specificities of FptA and FetA, we carried out binding assays using 55 Fe-loaded Pch diastereoiso-mers. To avoid competition with endogenous siderophores, we used the P. aeruginosa Pch-and Pvd-deficient strain PAD07 and the P. fluorescens EPch-and Pvd-deficient strain CHA1085. To measure only the binding of the siderophore- 55 Fe to the outer membrane transporter without any uptake into the bacteria, the binding assays were carried out in the presence of the protonophore CCCP. In the presence of this compound, the siderophore outer membrane transporter is still able to bind its ferrisiderophore, but the TonB machinery and, therefore iron uptake, is inhibited. The Pch stereoisomers used for these experiments were synthesized and purified as reported previously (23,24). The purification procedure used did not separate the diastereoisomer couples. However, in the presence of a metal ion (Fe 3ϩ and Zn 2ϩ ), PchII isomerizes into PchI (35,36) by epimerization of the C2Љ carbon atom (Fig. 1). The thioamine function included in the thiazolidine ring can undergo ring opening and closure (37). The configuration of the C2Љ carbon atom, located between the thiazolidine nitrogen and sulfur heteroatoms, is therefore highly labile, especially in the presence of a Lewis acid like a metal ion. Thus, the metal ion may drive its ligand isomerization to optimize the chelation properties through a template effect. For Pch diastereoisomers, where the hydrogen atoms H2Љ and H4Љ are in cis, the carboxylate, the thiazoline nitrogen, the thiazolidine nitrogen, and the phenol oxygen have a suitable geometry to form a tetradentate ligand of the metal. On the contrary, for the diastereoisomers where the hydrogen atoms H2Љ and H4Љ are in trans, the carboxylate is too distant from the metal to coordinate; these isomers are bare chelators. Similar metal-induced epimerization of C2Љ may also occur in the other diastereoisomer pairs (Fig. 1). Therefore, the mixtures used were: PchI/II, which is PchI and PchII in the absence of iron and mainly PchI in the presence of iron; NeoPchI/II, which is NeoPchI and NeoPchII in the absence of iron and mainly NeoPchI when loaded with iron; EPchI/II, which is EPchI and EPchII in the absence of iron and mainly EPchI in the presence of iron. ENeoPch diastereoisomers here such that in all four isomers of type I there is now a cis relationship between H2Љ and H4Љ, whereas in all isomers of type II there is now a trans relationship between these two hydrogen atoms.
We measured binding of Pch, NeoPch, and EPch to the outer membrane receptors of P. aeruginosa (FptA) and P. fluorescens (FetA). For 55 Fe-Pch, a K d of 2.5 nM Ϯ 1.1 was determined for FptA, whereas no binding was observed to the P. fluorescens receptor (Table 2). Likewise, 55 Fe-NeoPch bound FptA, although with a 10-fold lower affinity (29 nM Ϯ 11) than 55 Fe-Pch, and no binding occurred to FetA ( Table 2). Binding of 55 Fe-NeoPch to FptA was reported previously by our group (22). However, in those experiments a siderophore preparation containing some Pch (ratio Pch/NeoPch of 3:7) was used. Our present data, obtained with a pure preparation of NeoPch, thus demonstrate clearly that NeoPch binds FptA and that the binding reported previously (22) was not only because of the presence of low amounts of Pch. 55 Fe-EPch bound the P. fluorescens receptor FetA with a K d of 3.7 Ϯ 2.1 nM but was unable to interact with the P. aeruginosa receptor ( Table 2). The affinities of ENeoPch to FptA or FetA were not determined, as we do not have a sufficiently pure sample of this molecule. The concentrations of the 55 Fe-loaded siderophores used here were below 200 nM. At higher concentrations, an increase of radioactivity was monitored even in the absence of cells, probably because of a minor precipitation of free 55 Fe. Therefore, we cannot totally exclude a binding of 55 Fe-EPch to FptA and of 55 Fe-Pch to FetA with an affinity of at least 200-fold lower.
The use of a Scatchard representation to determine the affinity of a ligand for its receptor allowed us to estimate the receptor concentration. According to our experimental data, the concentrations in an overnight culture (A 600 of 1) are 13.4 nM for FptA in P. aeruginosa PAD07 and 0.15 nM for FetA in P. fluorescens CHA1085. There is, thus, an unexplained 100-fold difference in receptor expression between the two species.
Ability of Pch Stereoisomers to Transport 55 Fe in P. aeruginosa and P. fluorescens-The ability of the Pch stereoisomers to transport iron in P. aeruginosa and P. fluorescens was measured with 55 Fe (Fig. 2). Cells were incubated in the presence of the siderophore- 55 Fe complexes, and the amount of transported 55 Fe was monitored. As controls, all experiments were repeated under conditions when iron uptake is inhibited, such as in the presence of the protonophore CCCP at 0°C or in the absence of cells (data not shown). To avoid 55 Fe uptake by endogenous siderophores, we used, in the case of P. aeruginosa, the Pch-and Pvd-deficient PAD07 strain (Fig. 2A). The data obtained clearly showed that PAD07 was able to transport 55 Fe-Pch and, with a lower uptake rate, also 55 Fe-NeoPch. However, no iron uptake occurred with 55 Fe-EPch. When the fptA mutant strain K2388 was used, no uptake of 55 Fe-Pch and 55 Fe-NeoPch occurred, demonstrating that the incorporation of both iron-loaded sid- PAD07 (expressing FptA) and CHA1085 (expressing FetA) cells were prepared as described under "Experimental Procedures" and were incubated with a series of concentrations of 55 Fe-Pch or 55 Fe-EPch for 1 h at 0°C. The mixtures were centrifuged, and the radioactivity in the pellet was counted. The constants were determined from a Scatchard representation, and the errors were determined from multiple Scatchard plots.

Siderophores
Outer   55 Fe-Pch (E and ‚ for, respectively, PAD07 and K2388), 55 Fe-NeoPch (Ⅺ for PAD07, data not shown for K2388), or 55 Fe-EPch (छç for PAD07, data not shown for K2388). The experiments were repeated at 0°C in the presence of 200 M CCCP (F for 55 Fe-Pch, ࡗ for 55 Fe-EPch, and data not shown for 55 Fe-NeoPch) and in the absence of cells (data not shown). B, after 20 h of culture, cells of P. fluorescens CHA400 (Pvd Ϫ ) and CHA1080 (Pvd Ϫ and FetA Ϫ ) at an A 600 of 5 were incubated at 37°C for 15 min in 50 mM Tris-HCl (pH 8.0) before transport assays were started by the addition of 100 nM 55 Fe-Pch (E for CHA400, data not shown for CHA1080), 55 Fe-NeoPch (Ⅺ for CHA400, data not shown for CHA1080), or 55 Fe-EPch (छ and ‚ for, respectively, CHA400 and CHA1080). The experiments were repeated at 0°C in the presence of 200 M CCCP (data are shown only for CHA400 with 55 Fe-EPch (ࡗ)) and in the absence of cells (data not shown). erophores occurs via FptA. All these uptake assays were carried out with 20-h cultures of P. aeruginosa. When a Pvd-and EPch-deficient P. fluorescens strain was used (CHA1085), no 55 Fe uptake in the presence of 55 Fe-EPch was detected with cells grown for 20 h, and only a very modest uptake occurred with cells from a 48-h culture (data not shown). As mentioned above, the concentration of FetA in P. fluorescens was found to be considerably lower at a given optical density than the concentration of FptA in P. aeruginosa. Moreover, we suspect that fetA expression may require EPch, thus resembling the regulation of fptA by PchR and Pch in P. aeruginosa (16). It is, thus, possible that the FetA concentration was too low in the EPch-negative strain CHA1085 to detect iron uptake with 55 Fe-EPch under the experimental conditions used. We, therefore, performed the experiments with the Pvdnegative but EPch-producing strain CHA400 and its fetA-negative derivative CHA1080 (Fig. 2B). With CHA400 cells from a 20-h culture, 55 Fe-EPch but not 55 Fe-Pch was able to transport 55 Fe. 55 Fe-EPch uptake was abolished in the Pvd-negative ⌬fetA mutant CHA1080, demonstrating that EPch-mediated iron uptake depends on the FetA receptor.
In P. aeruginosa PAD07, 30 pmol of 55 Fe/ml (A 600 ) were transported after 30 min in the presence of Pch. When the same experimental conditions were used with P. fluorescens CHA400 cells, the uptake rate of 55 Fe/ml (A 600 ) was only 7 pmol with EPch. This difference cannot be because of a difference in the affinities of the transporters for their respective siderophore ( Table 2) but probably reflects the lower expression level of FetA in P. fluorescens compared with the expression of FptA in P. aeruginosa. Previous studies have shown that more transporters than TonB proteins are produced. A 10-fold excess was reported for the ferrichrome transporter FhuA and the enterobactin transporter FepA of E. coli (38 -41). As TonB is required for ferric-siderophore uptake across the outer membrane, only a small amount of the transporters can be active. Therefore, a 100-fold lower expression of FetA compared with FptA does not involve a proportional decrease in the 55 Fe uptake rates.
Role of the Outer Membrane Receptors in Enantiospecific Growth Promotion with Pch and EPch-We evaluated the role of FptA and FetA in Pch and EPch utilization in P. aeruginosa and P. fluorescens during cell growth. As shown in Fig. 3, B and G, unmodified M9-glycerol medium contained sufficient iron to support the growth of P. fluorescens CHA1085 (Pvd Ϫ , EPch Ϫ ) and P. aeruginosa PAO6383 (Pvd Ϫ , Pch Ϫ ), but when the iron chelator 2,2-dipyridyl was added to the medium, these mutants were no longer able to grow. The EPch-and Pch-positive precursor strains CHA400 (Fig. 3A) and PAO6382 (Fig.  3F), respectively, grew well in the presence of the iron chelator, although CHA400 exhibited a longer lag phase. Interestingly, Pch addition delayed growth of CHA400 even more, suggesting that the heterologous siderophore Pch competes with EPch for iron. As this effect was not observed in P. aeruginosa, we suspect that the kinetics of siderophore production and/or receptor expression are different in the two species. Growth of the siderophore-negative P. fluorescens mutant CHA1085 was restored by adding 20 M EPch (Fig. 3B), whereas the addition of Pch produced no such effect. Reciprocally, Pch promoted growth in PAO6383, but no growth promotion was achieved with EPch (Fig. 3G). These results, which have been reported previously (16), demonstrate that growth promotion is a stereospecific process. A mutation in fptA abolished Pch-mediated growth promotion in P. aeruginosa (Fig. 3H (31, 42)). Similarly, we have found here that the FetA receptor is essential for EPch utilization in P. fluorescens (Fig. 3C). Complementation of PAO6541 (Pvd Ϫ , Pch Ϫ , FptA Ϫ ) with the Fe-Pch uptake operon (fptABCX) carried by pME9629 fully restored Pch-mediated growth promotion (Fig. 3I), and heterologous expression of fpt-ABCX in the siderophore-negative fetA mutant CHA1169 enabled this strain to grow with Pch instead of EPch as an iron source (Fig. 3E). Reciprocally, the fetA mutation of CHA1169 was complemented by pME9630, a plasmid constitutively expressing fetA and adjacent genes specifying a putative ABC transporter (Fig. 3D). When pME9630 was introduced into the fptA mutant PAO6541, growth promotion occurred with EPch, whereas Pch was not utilized (Fig. 3J). These experiments, thus, confirm and extend the results from the above binding and 55 Fe uptake assays. They show that FptA and FetA are able to discriminate between the two siderophore enantiomers and that the preference of P. aeruginosa and P. fluorescens for Pch and EPch, respectively, can be inverted by swapping the respective ferrisiderophore transport genes between the two species.
Ligand Docking-To better understand the interaction mechanism of the FptA binding site with the Pch stereoisomers, we carried out docking experiments using the x-ray structure of FptA loaded with Fe-PchI (20) (Fig. 1). The program GOLD (Genetic Optimization for Ligand Docking) (34), a genetic algorithm for docking flexible ligands into protein binding sites, was used to dock all herein investigated ligands. Using automated docking, we obtained a Fe-Pch conformation very close to that of its x-ray structure (root mean square deviations of 0.3 Å from the heavy atoms) and with a high Goldscore of 63.8 ( Table 3). The docking of Fe-NeoPch has been described previously and gave similar results (22). Docking of Fe-EPchI afforded quite different results. First, the docking score obtained for that compound (48.72) is significantly lower than that computed for Fe-PchI (63.81, Table 3). Second, the predicted binding mode of Fe-EPchI to FptA involves only apolar contacts with the hydrophobic binding pocket. Contrary to Fe-PchI, no hydrogen bond to the receptor could be proposed for any possible docking solution of Fe-EPchI. As a consequence, there is no strong directionality for anchoring Fe-EPchI to the FptA binding pocket, which could explain the herein reported absence of detectable affinity for the latter siderophore. A structural explanation to the absence of hydrogen bonds lies in the inverted configuration of the C4Љ carbon atom, which expels the carboxylate moiety away from the binding pocket, as noticed by the decreased buried surface area of Fe-EPchI with respect to that of Fe-PchI (Table 3).

DISCUSSION
Under iron limitation, P. aeruginosa produces the two diastereoisomers PchI and PchII, whereas some P. fluorescens strains are known to produce the optical antipodes, EPchI and EPchII. The four other diastereoisomers, NeoPchI/II and ENeoPchI/II, have so far never been isolated from any siderophore-producing strain. In the presence of a metal ion, PchII isomerizes into PchI (24,36), and NeoPchII isomerizes into NeoPchI. This suggests a metal-driven template effect giving a cis H2Љ-H4Љ relationship that is necessary for optimal ferric ion tetracoordination by the siderophore. Such an isomerization may also exist for the two other couples of diastereoisomers. This isomerization suggests that in the 55 Fe-PchI/II, 55 Fe-NeoPchI/II, or 55 Fe-EPchI/II preparations, we have, respectively, mostly 55 Fe-PchI, 55 Fe-NeoPchI, or 55 Fe-EPchI. Therefore, in the present work Fe-Pch, Fe-NeoPch, and Fe-EPch are considered as, respectively, Fe-PchI, Fe-NeoPchI, and Fe-EPchI. 55 Fe uptake assays showed that Pch is the siderophore used by P. aeruginosa for iron uptake and EPch is the one used by P. fluorescens (Fig. 2). Absolutely no 55 Fe incorporation was observed when P. aeruginosa cells were incubated in the presence of 55 Fe-EPch or P. fluorescens in the presence of 55 Fe-Pch. Clearly two different iron uptake pathways exist in these two species, one using only Pch and the other, EPch. The outer membrane transporters involved are, respectively, FptA for Fe-Pch in P. aeruginosa and FetA for Fe-EPch in P. fluorescens. When their corresponding genes were mutated, 55 Fe-Pch and 55 Fe-EPch uptakes were abolished in P. aeruginosa and P. fluorescens, respectively. The use of the protonophore CCCP demonstrated that the two uptake pathways, Fe-Pch via FptA and Fe-EPch via FetA, are TonB-dependent; inhibition of the protonmotive force abolished iron uptake. The ability of FptA and FetA to discriminate between the two siderophore enantiomers was also observed in growth promotion assays (Fig. 3), and the siderophore preference of P. aeruginosa and P. fluorescens could be inverted by swapping the ferrisiderophore transport genes between the two species. The Scatchard representation allowed us to determine a K d of 2.5 nM Ϯ 1.1 for the binding of 55 Fe-Pch to FptA and a K d of 3.7 nM Ϯ 2.1 for the binding of 55 Fe-EPch to FetA (Table 2). Both K d values are, thus, very similar to each other and to the K d values determined for other ferrisiderophores and their specific outer membrane transporters (33,43,44). It seems that an affinity in the nanomolar range of ferrisiderophores for their corresponding transporters is a conserved parameter during evolution. A lower affinity may not be sufficient for bacteria in their competition for iron. It is important also to point out that, consistent with 55 Fe uptake assays, no binding was observed between FetA and 55 Fe-Pch and between FptA and 55 Fe-EPch under these experimental conditions. If binding does exist, it clearly would occur with a much lower affinity and would probably not allow an efficient Fe 3ϩ uptake. 55 Fe uptake and binding assays showed that FetA was unable to bind and transport 55 Fe-NeoPch. However, 55 Fe-NeoPch was bound by FptA (K d ϭ 29 nM Ϯ 11) and was also incorporated into the cells, although iron uptake was not as efficient as with Pch. By analogy with these findings, we predict that ENeoPch might be able to interact with FetA but not with FptA. Future experiments will address this issue.
The previously published x-ray structure of FptA-Fe-PchI and docking experiments were used to better understand the stereoselectivity of siderophore binding. The siderophore binding pocket is mainly composed of hydrophobic and aromatic residues, consistent with the hydrophobicity of Pch. In the FptA-Fe-PchI structure, the ferric ion is hexacoordinated by a Goldscore is the fitness function of the GOLD program (34), which quantifies protein-ligand interactions according to four terms: protein-ligand H-bond interaction energy, protein-ligand van der Waals interaction energy, ligand internal van der Waals energy, ligand torsional strain energy. The data concerning Fe-NeoPchI have been published previously (22). b Root mean square deviations (in Å) of the heavy atoms from the x-ray configuration of Fe-PchI. c Possible H-bonds are indicated by an X. d Buried surface area (in percentage of the total surface) of the FptA-bound ligand.
four atoms from Pch (the nitrogen atoms of the thiazoline and thiazolidine rings and the oxygen atoms of the phenol and carboxylate groups) and two oxygen atoms from ethylene glycol, with a 1:1:1 stoichiometry (20). The observed diastereoisomer of Pch in the FptA structure is PchI. Previous docking experiments predict that Fe-PchI and Fe-NeoPchI adopt the same conformation when bound to FptA, whereas the receptorbound conformation of Fe-PchII and Fe-NeoPchII are incompatible with strong metal chelation because of a trans relationship between the H2Љ and H4Љ protons repelling the carboxylic acid moiety far from the iron coordination sphere (22). For Fe-PchI and Fe-NeoPchI, the phenol moiety is predicted to bind similarly to a hydrophobic subsite (Tyr-356, Phe-358, Ala-144, Leu-117), and the two five-member heterocycles face the central non-polar part of the binding site (Met-271, Tyr-334, Trp-702). However, the carboxylate moiety is proposed to H-bond to two different polar environments, the Leu-116 and Leu-117 backbone nitrogen atoms (22). Docking of Fe-EPchI gave a low docking score ( Table 3), suggesting that the Pch enantiomer cannot be recognized by FptA. The predicted binding mode involves only apolar contacts with the hydrophobic binding site (Fig. 4). These different docking predictions are fully consistent with our binding data on the ability of FptA to interact with the different Pch diastereoisomers (Table 2).
NeoPchI only differs from PchI by the stereochemistry of the C4Ј chiral center. Because both Fe-Pch and Fe-NeoPch are able to bind to FptA with only a 10-fold difference in affinity, the configuration of the C4Ј chiral center seems less important for the interaction with the transporter. Moreover, when the chiral center C4Ј is removed by replacing the thiazoline moiety by a thiazole ring in hydroxy-phenyl-thiazolyl-bishydro-thiazole (HPTBT) (Fig. 1), only a slight decrease in the affinity for FptA is observed, and the iron transport properties of Fe-hydroxyphenyl-thiazolyl-bishydro-thiazole resemble those of Fe-Pch (22). In contrast, the conformation of C4Љ appears to be more critical for the interaction of Pch with FptA. When Fe-EPchI was docked in the FptA binding site, the inverted configuration of the C4Љ carbon expelled the carboxylate moiety from the binding pocket, and the hydrogen-bond with Leu-116 and Leu-117 was no longer possible (Fig. 4). The absence of this hydrogen-bond may explain why there is no binding of Fe-EPch to FptA. Concerning the last chiral center, C2Љ, an epimerization occurs in the presence of metal for an optimal ferric ion tetracoordination by the siderophore (17,36). In the case of Pch, this chiral center may always have the configuration R when loaded with metal and bound to FptA, whatever the configuration of the two other chiral centers may be. To illustrate the importance of this chiral center C2Љ in the interaction of FptA with its siderophore, it is interesting to remember the biological properties of hydroxy-phenyl-thiazolyl-thiazole (HPTT-COOH) (Fig. 1). In this molecule the C4Ј and C2Љ chiral centers have been removed. As a consequence, the ferric-siderophore became more rigid, its affinity for FptA was slightly decreased, and it was no longer translocated across the outer membrane (22). Although Fe-hydroxyphenylthiazolylthiazole-COOH could be docked into the binding site with a score very similar to Fe-PchI, we predicted a complete inversion of the binding mode to FptA. Taken together, all these data clearly suggest that the configuration of the chiral centers C2Љ and C4Љ are important for the binding and uptake of Fe-Pch by FptA, whereas the configuration of the C4Ј carbon atom has nearly no influence.
The mechanism of interaction between proteins and chiral ligands is a fascinating topic in biology. Stereospecificity of iron uptake systems has been observed previously with several other siderophores such as parabactin, rhodotorulic acid, ferrichrome, and enterobactin and their corresponding enantiomers (45)(46)(47)(48)(49)(50)(51)(52). However, so far no siderophore outer membrane transporters have been described with opposite binding enantioselectivities. FptA and FetA are the first transporters with such properties; FptA is specific to Fe-Pch, and FetA is specific to the enantiomer siderophore, Fe-EPch. The investigation of the FptA binding properties allowed us to show that  (34), is displayed in the FptA binding site (residues lining the cavity are represented as sticks) and compared with the x-ray structure of FptA complexed with Fe-PchI (panel A). The accessible surface of the Pch cavity is represented by a transparent surface, and the main chain atoms of the FptA protein are displayed as a ribbon. Hydrogen bonds of the ligands to FptA are indicated by a yellow solid line with the corresponding distance in Å between interacting heavy atoms. The iron atom of the ferrisiderophores is displayed as a cyan ball. the configuration of the chiral centers C2Љ and C4Љ are important for the binding and uptake of Fe-Pch by this transporter, whereas the configuration of the C4Ј carbon atom has nearly no influence. Further studies are currently undertaken to determine the FetA-Fe-EPchI structure and to understand the mechanism driving enantioselectivity of this transporter.