HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via CrossRef Citing Articles via Google Scholar Google Scholar Articles by López, C. S. Articles by Crosa, J. H. Search for Related Content PubMed PubMed Citation Articles by López, C. S. Articles by Crosa, J. H. Agricola Articles by López, C. S. Articles by Crosa, J. H.

Identification of amino acid residues required for ferric-anguibactin transport in the outer-membrane receptor FatA of Vibrio anguillarum

¹ Department of Molecular Microbiology and Immunology, Oregon Health and Science University, Portland, OR 97239, USA
² Department of Health Sciences, College of Public and Allied Health, East Tennessee State University, Johnson City, TN, USA

Correspondence
Jorge H. Crosa
crosajor{at}ohsu.edu

	ABSTRACT

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

 
Vibrio anguillarum 775 is a fish pathogen that causes a disease characterized by a fatal haemorrhagic septicaemia. It harbours the 65 kbp pJM1 plasmid, which encodes an iron sequestering system specific for the siderophore anguibactin and is essential for virulence. The genes involved in the biosynthesis of anguibactin are located on both the pJM1 plasmid and the chromosome. However, the genes for the outer-membrane receptor FatA and the other transport proteins are only carried on the plasmid. With the aim of elucidating the mechanism of ferric-anguibactin transport mediated by FatA, this work focuses on the identification of FatA amino acid residues that play a role in the transport of ferric-anguibactin, by analysing the transport kinetics of site-directed mutants. The mutations studied were located in conserved residues of the lock region, which contains a cluster of ten residues belonging to the N-terminal and barrel domains, and of the channel region of FatA, which contains conserved glycines located in the 5-6 loop and a conserved arginine located in strand 11 of the -barrel. In the case of the FatA lock region, it is clear that although the residues analysed in this work (R95, K130, E505 and E550) are conserved among various outer-membrane receptors, their involvement in the transport process might differ among receptors. Furthermore, it was determined that in the FatA channel region double substitutions of the conserved glycines 131 and 143 with alanine resulted in a variant receptor unable to transport ferric-anguibactin. It was also shown that the conserved arginine 428 located in strand 11 is essential for transport. The results suggest that a conformational change or partial unfolding of the plug domain occurs during ferric-anguibactin transport.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

 
In the host and in natural environments, ferric iron is unavailable since under aerobiosis at biological pH it is quite insoluble (free iron 10^–18 M). In the vertebrate host the majority of this iron is found in red blood cells or bound by host proteins, such as transferrin and lactoferrin (Bullen & Griffiths, 1999). Therefore, during infection bacterial pathogens depend on their ability to scavenge this essential metal from the host-complexed state. These types of invasive organisms have the ability to produce low-molecular-mass iron sequestering molecules with high affinity for ferric iron, known as siderophores. These compounds are synthesized from carboxylic acids and amino acids via a nonribosomal peptide synthetase mechanism and contain side chains and functional groups able to coordinate ferric ions with high affinity and specificity (Crosa & Walsh, 2002).

Vibrio anguillarum is a Gram-negative, polarly flagellated, comma-shaped rod bacterium that is responsible for both marine and freshwater fish epizootics all over the world. This bacterium causes a highly fatal haemorrhagic septicaemic disease not only in salmonids, but also in other types of fish, including eels (Actis et al., 1999). Many virulent serotype O1 strains harbour the 65 kbp pJM1 virulence plasmid, which encodes an iron-sequestering system essential for their pathogenicity (Crosa, 1980). This system includes most of the biosynthetic genes for the siderophore anguibactin, as well as the transport proteins FatA, B, C and D, involved in the transport of ferric-anguibactin into the cytoplasm (Actis et al., 1988; Crosa, 1980; Koster et al., 1991; Tolmasky et al., 1988). This peptide siderophore is composed of one molecule each of 2,3-dihydroxybenzoic acid (DHBA), L-cysteine and N-hydroxyhistamine (Actis et al., 1986; Jalal et al., 1989).

The ferric-siderophore's receptors are specific in the recognition of these iron complexes, with the energy-dependent transport being driven by the proton-motive force across the cytoplasmic membrane (Bradbeer, 1993; Kadner, 1990). The energy transduction from the cytoplasmic membrane to the outer membrane requires the TonB, ExbB and ExbD proteins (Bassford et al., 1976; Fischer et al., 1989; Frost & Rosenberg, 1975; Wiener, 2005). The TonB protein, which is anchored in the cytoplasmic membrane, extends into the periplasm and interacts with the receptor during transport of the ferric-siderophore complex, transducing the energy from the cytoplasmic membrane; however, the exact mechanism of this energy transduction event is still unknown (Postle & Kadner, 2003). In the case of V. anguillarum two TonB systems are present but only one, the TonB2 system, operates in ferric-anguibactin transport (Stork et al., 2004).

The structures of several outer-membrane receptors have been solved, including those of FecA, FepA, FhuA and BtuB from Escherichia coli and FpvA and FptA from Pseudomonas aeruginosa (Buchanan et al., 1999; Chimento et al., 2003; Cobessi et al., 2005a, b; Ferguson et al., 1998, 2002). More recently, the crystal structure of the E. coli TonB protein in complex with the cobalamine receptor BtuB (Shultis et al., 2006) and the ferrichrome receptor FhuA (Pawelek et al., 2006) have also been solved. Taking into account these structures, several laboratories have identified specific amino acids intervening in the binding and/or transport of ferric-siderophore complexes (Cao et al., 2000; Chakraborty et al., 2003; Endriss et al., 2003; Mislin et al., 2006; Sauter & Braun, 2004; Shen et al., 2005). From these structures it was possible to determine that on the extracellular side, the strands of the barrel are connected to form solvent-accessible extracellular loops of variable lengths, while on the periplasmic side relatively shorter loops are present (van der Helm, 2004). The extracellular loops form a pocket to receive the ferric-siderophore complex from the external environment and there are no homologies observed in these loops or the apices formed by the N-terminal domain, which is consistent with their recognizing and binding to chemically and structurally diverse siderophores. Moreover, by sequence alignment of several receptors, it can be deduced that most of the conserved residues are in the N-terminal region, forming a plug domain, with the remainder located in the portion of the barrel embedded in the membrane (Chimento et al., 2005; Ferguson & Deisenhofer, 2002; van der Helm, 2004). The plug, formed by approximately 150 N-terminal residues, consists of a four-stranded mixed sheet inclined by about 45° with respect to the membrane plane (Chakraborty et al., 2003; Ferguson & Deisenhofer, 2002; Ferguson et al., 2002). The ferric-siderophore binding site involves amino acid residues located in the external loops as well as some residues from the three apices of the N-terminal plug domain. In FecA, the diferric-dicitrate receptor, the interaction of the ligand complex with the receptor significantly affects the conformation of the external loops as well as the plug domain (Ferguson et al., 2002; Yue et al., 2003). The major structural changes observed upon binding of the ferric-siderophore are in loops 7 and 8 and minor changes were also observed in loops 4, 5 and 9 (Ferguson et al., 2002). However, this conformational change has not been observed in the other ligand-bound structures. Binding of the ligand to FecA also induces conformational changes in the plug domain: apices A and C shift toward the diferric-dicitrate molecule, apex B shifts away from the siderophore, and there is a concerted downward movement of 5 and 6 toward the periplasm (Ferguson et al., 2002; Yue et al., 2003). These changes extend to the periplasmic side and might be responsible for the receptor's interaction with the TonB complex. Therefore, these allosteric transitions tightly regulate the opening of a putative transient channel or permeation path for the transport of the substrate into the periplasm (Shultis et al., 2006; van der Helm, 2004; van der Helm et al., 2002). As suggested by Shultis et al. (2006) the degree of this allosteric transition will be a function of the size and chemical characteristics of each ferric-siderophore complex.

The experiments delineated in this work explored the effects of mutations in specifically conserved residues on the transport function of FatA. Our molecular modelling and experimental results underscore the fact that although there is a low sequence identity between the known siderophore receptors, the mechanisms of transport will probably share some similarities among them. Moreover, our results also suggest that a conformational change or partial unfolding of the plug domain occurs during ferric-anguibactin transport.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

 
Bacterial cultures.
The strains and plasmids used in this work are listed in Tables 1 and 2 respectively. E. coli strains were grown at 37 °C in Luria–Bertani (LB) broth supplemented with antibiotics as appropriate: kanamycin 50 µg ml^–1, chloramphenicol 30 µg ml^–1 and ampicillin 100 µg ml^–1. V. anguillarum strains were grown at 25 °C in trypticase soy broth with 1 % (w/v) sodium chloride (TSBS) supplemented with antibiotics as appropriate: kanamycin 50 µg ml^–1, chloramphenicol 10 µg ml^–1 and rifampicin 100 µg ml^–1. Solid media were obtained by adding 1.5 % (w/v) agar. For iron-restricted growth V. anguillarum was grown in chemically defined CM9 minimal medium [60 g Na₂HPO₄ l^–1, 30 g KH₂PO₄ l^–1, 50 g NaCl l^–1, 10 g NH₄Cl l^–1, pH 7.2, 0.5 %, w/v, glucose and 0.2 %, w/v, Casamino acids (Crosa, 1980)] supplemented with 10 µM of the iron chelator ethylenediamine-di-(o-hydroxyphenylacetic) acid (EDDHA). When appropriate IPTG was added to the liquid broth to a final concentration of 1 mM. Bacterial growth was monitored by the optical density at 600 nm. Chrome azurol S (CAS) agar plates were used to detect siderophore production (Schwyn & Neilands, 1987).

RT-PCR assays.
Total RNA was extracted as described above and treated with Turbo DNA-free (Ambion). RT-PCR analysis was performed using the Superscript II reverse transcriptase (Invitrogen) according to the manufacturer's instructions. The PCR reactions were performed as follows: 95 °C for 5 min, 30 cycles of 95 °C for 30 s, 60 °C for 60 s and 72 °C for 90 s, with a final extension step at 72 °C for 10 min. The primer used for cDNA synthesis was RR (Table 3) and the primers used in the PCR reactions were RR and RU for the plasmid-encoded angR gene; BR and BU for the fatB gene; DR and DU for the fatD gene; and A4 and A3 for the fatA gene (Table 3).

In-frame deletions of the entire coding sequence of the tonB1, tonB2 and angA chromosomal genes were generated using splicing by overlap extension (SOE)-PCR (Senanayake & Brian, 1995), which resulted in a final product containing less than 10 % of the sequence of the wild-type gene. For the tonB1 gene the primers used for SOE were 1B1AU, 2B1AUR, 3B1ARU and 4B1AR and for tonB2 primers 1B2AU, 2B2AUR, 3B2ARU and 4B2AR (Table 3). The fragments containing each deletion were cloned in the plasmid pCR2.1-TOPO, then subcloned in the suicide vector pDM4 (Table 2) and transformed into E. coli S17-1 pir (Table 1). For the construction of the double fatA angA mutant strain we used the suicide plasmid pALE3 (Alice et al., 2005). The resulting suicide plasmids were mobilized by conjugation into V. anguillarum 775 and CSL-13 using the chloramphenicol resistance for selection. After the conjugation, rifampicin (100 µg ml^–1) was used for counter-selection against E. coli. The exconjugants were screened using a sucrose-resistance-based counter-selection strategy and all the isolated mutants were confirmed by PCR analysis. To serve as a control the pMMB208 vector was conjugated into each strain (Table 2).

Site-directed mutagenesis.
The fatA gene was amplified with the primers fatAUP and fatAD (Table 3) using the Pfu-Taq polymerase (Stratagene) and subcloned into the pCR-BluntII vector (Invitrogen). The resulting plasmid, pCL2, was used as a template for the site-directed mutagenesis, which was performed using the QuickChange site-directed mutagenesis kit (Stratagene) according to the manufacturer's instructions. The sequencing reactions were performed at the Vollum Institute Sequencing Core facility, using a 3130xL Genetic Analyser (Applied Biosystems) with the BigDye Terminator v3.1 Cycle Sequencing Ready Reaction Kit (Applied Biosystems) and the samples were run on a 50 cm capillary array with the POP-7 polymer.

Anguibactin purification.
The siderophore purification procedures were carried out using glassware that had been washed with 0.1 M HCl. The temperature during the purification procedure never exceeded 25 °C. Anguibactin was purified as described previously (Actis et al., 1986). The purity of the siderophore was confirmed by HPLC (Beckman Gold HPLC System), monitoring at 264 nm using a C18 reverse-phase HPLC column (YMC, AQ-312, 150x6 mm ID), and as mobile phase: solvent A [H₂O 0.01 %, v/v, trifluoroacetic acid (TFA)] and solvent B (acetonitrile 0.01 %, v/v, TFA). The gradient used was 100 % solvent A for 2 min, followed by an increasing concentration of solvent B from 0 to 100 % over 30 min. The biological activity of the purified anguibactin was assessed by bioassays using strain AC7 (Alice et al., 2005), which is siderophore-production-deficient but FatA-receptor-proficient.

⁵⁵Fe-anguibactin uptake.
The ⁵⁵Fe-anguibactin complex was prepared using a stock solution of ⁵⁵FeCl₃ in 0.5 M HCl (specific activity 103.46 mCi mg^–1, 3828 MBq mg^–1; Perkin Elmer). The ⁵⁵FeCl₃ was diluted in MOPS buffer (0.1 M pH 7.0) to a final concentration of 0.8 mM in the presence of 1.2 mM of the iron chelator EDDHA in order to maintain the ferric iron in solution and this complex was then incubated for 10 min at room temperature. After this incubation period, the deferrated anguibactin (1 mM), also prepared in 0.1 M MOPS pH 7.0, was added. Due to its instability, the ⁵⁵Fe-anguibactin complex was used fresh or stored for only 3 days at –80 °C. The cultures used for this assay were grown in CM9 minimal medium supplemented with 1 mM IPTG until exponential phase (OD₆₀₀ 0.5–0.8). The cultures were centrifuged and resuspended in CM9 supplemented with 10 µM EDDHA in order to chelate the available iron in the medium and induce the ferric-anguibactin transport system as well as the energy transduction system. These cultures were then incubated for 2 h at 25 °C in order to starve the cells of internal iron. It is worth mentioning that this EDDHA concentration, which is lower than the minimal inhibitory concentration detected for the V. anguillarum 775 strain (75 µM), is sufficient to prevent cellular multiplication during the incubation period. After this incubation period the cells were centrifuged and resuspended in half of the original volume of complete minimal medium (without EDDHA). Bacterial dilutions performed in the same medium were used to calculate the colony-forming units (c.f.u.). For the time- and concentration-dependent curves 50 ml tubes containing 20 ml complete minimal medium and the desired ⁵⁵Fe-anguibactin concentration were prepared. For each tube, 100 µl of the cellular suspension was added. During the experiment the cells were incubated under agitation at 200 r.p.m. at room temperature. After each incubation period the cellular suspension was quenched by the addition of a 100-fold dose of non-labelled ferric-anguibactin, rapidly filtered through a 0.22 µm filter (Millipore Corporation) and washed three times with 10 ml 2 M LiCl. The filters were air-dried and the radioactivity measured in a liquid scintillation counter. The K_m and V_max were calculated using ferric-anguibactin concentrations ranging from 0.01 to 75 nM. In the case of the G138A G143A double mutant, the concentrations used were from 0.01 to 200 nM. Kinetic parameters were determined from the initial rate of ⁵⁵Fe-anguibactin uptake, calculated at different substrate concentration at 20 sec and 1 min for all the strains. The kinetic parameters K_m and V_max were evaluated and treated as those for an enzyme reaction using the Graphic Pad Prism4 program. Non-specific binding has been subtracted for the determination of the transport rates and kinetic parameters by using the values obtained for the double fatA tonB2 V. anguillarum mutant complemented with the wild-type fatA gene. The Scatchard plot shown in Fig. 2 was obtained by linearizing the saturation binding data shown in the same figure using the Graphic Pad Prism4 program.

View larger version (21K): [in this window] [in a new window]   Fig. 2. Binding of 55Fe-anguibactin. Determination of binding to FatA in the presence of 1 mM potassium cyanide () or 1 mM dinitrophenol () as respiratory inhibitors. The tonB2 mutant strain () was used as a control without respiratory inhibitors. The Scatchard plot of the results is shown in the inset.

Sequence alignments.
The sequences of the outer-membrane receptors from different organisms were simultaneously aligned using CLUSTALW (Thompson et al., 1994).

FatA modelling.
To obtain a structural model for FatA the structures of the E. coli receptors FhuA (Ferguson et al., 1998; Locher et al., 1998), FepA (Buchanan et al., 1999), BtuB (Chimento et al., 2003) and FecA (Ferguson et al., 2002) and the P. aeruginosa FpvA receptor (Cobessi et al., 2005b) together with the multiple sequence alignment of 29 different ferric-siderophore and haem transporters (van der Helm, 2004) were used. Since the E. coli FhuA receptor displays the highest identity to the V. anguillarum FatA receptor (23 % identity) it was chosen as the platform for the FatA model. In the multiple sequence alignment of 29 proteins in this family, about 25 residues are conserved or identical; half of these residues belong to the plug region of the proteins and the rest are spread over a number of transmembrane strands from strand 11 to strand 22 (van der Helm, 2004). Interestingly, none of the conserved residues occur on extracellular loops with the exception of those at the beginning of loop 11. The conserved residues in the plug region occur at identical locations and orientation in five different crystal structures (see Results and Fig. 3b). These structural features of the plug region and the presence of conserved residues allow the sequence of FatA to be fitted in the structure of this region of FhuA, although it required several small deletions, insertions and a number of mutations. The insertions and deletions required were made in the middle of the extracellular loops using the SPDBV program (Guex & Peitsch, 1997). Construction of the remainder of the model was adversely affected by the fact that, as previously mentioned, the identity between FatA and FhuA is low and by the observation that extracellular loops in the proteins with known structure vary greatly in length. In addition, for many transmembrane strands the secondary structure extends far above the surface of the outer membrane. However, the multiple sequence alignment together with the structural information from the strands and the location of the conserved residues in the known structures allowed the assignment of the transmembrane strands of the 22-stranded barrel of FatA, obtaining a better certainty for the latter half of the barrel. The resulting FatA model was refined by molecular dynamics (MD) according to Brunger et al. (1998) and subsequently with energy minimizations using the SPDBV program (Guex & Peitsch, 1997). After convergence was obtained, the complete water structure of FhuA was added to the FatA model and once more MD and energy minimization were applied. The final model did not show unusual intermolecular contacts and none of the bond lengths, bond angles or conformational angles showed unusual energy values.

View larger version (29K): [in this window] [in a new window]   Fig. 3. Molecular modelling of FatA. (a) Representation of the barrel model of FatA viewed from one side. For clarity some of the strands have been removed from the left figure to show the plug domain of FatA, which is shown in blue. The yellow spheres in the structure represent two residues of the ‘lock region’ R95 and K130, the purple spheres two conserved glycines of the ‘channel region’ G131 and G143, and the red one the conserved arginine R428 in strand 11. (b) Representation of the conserved amino acid cluster of the quadrupole in the V. anguillarum outer-membrane receptor FatA and in the E. coli outer-membrane receptors FepA, FecA and FhuA.

Western blots.
For the Western blots 5 µg of the outer-membrane isolated fractions were separated on 12 % SDS-PAGE (Criterion–XT Precast gel, Bio-Rad Laboratories) using XT-MOPS buffer (Bio-Rad Laboratories) and transferred for 4 h at 400 mA into 0.45 µm Immobilon-P membranes (Millipore) using a Trans-Blot Electrophoretic Transfer Cell (Bio-Rad Laboratories). The transfer buffer used was 25 mM Tris, 192 mM glycine and 20 % (v/v), methanol, pH 8.3. After transfer the membranes were blocked overnight at 4 °C using 10 % (w/v), BSA in TBST buffer (20 mM Tris pH 7.6, 0.3 M NaCl and 0.05 %, v/v, Tween-20). The membranes were washed four times with TBST buffer and then incubated for 1 h with anti-FatA polyclonal antibody diluted 1/15 000 in TBST 5 % (w/v), BSA (TBST-BSA). After this incubation the membranes were washed four times with TBST buffer and then incubated for 1 h using as secondary antibody goat anti-rabbit IgG, [H+L, horseradish peroxidase (HRP) conjugated (Pierce Biotechnology) diluted 1/20 000 in TBST-BSA. The membranes were washed four times with TBST buffer and then incubated for 5 min with the Immobilon Western Chemiluminiscent HRP substrate (Millipore). The Westerns were developed using Kodak BioMax MR film and the images captured and analysed as described above.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

 
Construction and characterization of a V. anguillarum fatA mutant strain
To understand the mechanism of transport mediated by FatA we needed to construct a fatA null mutant to use as a recipient in complementation experiments. However, because the fatA gene is part of the iron-transport biosynthesis (ITB) operon it was required that this mutant was non-polar (Fig. 1a). The absence of polar effects due to the replacement of the fatA gene with an antibiotic resistance cassette was assessed by RT-PCR (Fig. 1a). The results clearly showed that the replacement of the fatA gene by a kanamycin resistance cassette did not affect the expression of downstream genes, such as the positive regulator and biosynthesis gene angR (Wertheimer et al., 1999). The expression of the upstream genes fatB and fatD was also confirmed by RT-PCR (Fig. 1a). As expected from a receptor with no N-terminal regulatory domain, such as those present in FecA in E. coli and FpvA in P. aeruginosa, the absence of the outer-membrane protein receptor FatA in V. anguillarum caused a significant overproduction of the siderophore anguibactin, as assessed by using chrome azurol S (CAS) medium (data not shown). The null mutation was also confirmed by Western blot analysis. Fig. 1(b) shows the amount of receptor produced in the fatA mutant complemented with the wild-type gene (lane 2) as compared to that expressed from the pJM1 indigenous plasmid (lane 1). Also shown is the result obtained with the fatA mutant harbouring the empty vector pMMB208 (lane 3). Densitometric analysis indicated that in the mutant complemented with the wild-type gene, receptor expression is at least twofold higher than in the wild-type 775 (Fig. 1b). To quantify this difference we used the nonspecific band denoted as X that cross-reacts with the FatA antiserum (Fig. 1b) as a loading control. Further comparisons of the amounts of FatA in the site-directed mutants were carried out using the complemented strain as a control.

View larger version (27K): [in this window] [in a new window]   Fig. 1. Characterization of a V. anguillarum fatA mutant. (a) Scheme of the ITB operon in V. anguillarum and RT-PCR analysis of RNA extracted from the V. anguillarum 775 wild-type (right panel) and the fatA mutant strain (left panel). The reverse transcription reactions were carried out using a primer complementary to the angR gene (RR) and the PCR reactions with primers complementary to the fatD (lanes 1), fatB (lanes 2), fatA (lanes 3) and angR (lanes 4) genes as indicated in Methods. M, low molecular mass marker. (b) Western blot using FatA antiserum of the wild-type V. anguillarum 775 (lane 1), the single fatA mutant complemented with the wild-type fatA gene (lane 2), and the empty vector pMMB208 (lane 3). X denotes a non-specific cross-reacting band.

The results indicated that FatA has a K_D of 26±3 nM for the complex with ferric-anguibactin. To assess whether the TonB2 protein plays a role in the binding of the ferric-anguibactin complex to FatA, we also determined the K_D value for the FatA receptor in the absence of this energy-transducing protein. The K_D values obtained for FatA in a tonB2 background determined in the presence and absence of KCN, 29±3 nM and 23±1 nM respectively, indicated that TonB2 is not essential for the binding process.

FatA modelling
The FatA model shown in Fig. 3a is based on the FhuA structure as described in the Methods (Ferguson et al., 1998; Locher et al., 1998). The model shows the well-known plug region (Fig. 3a, b) with the central 4-stranded sheet and the 22-stranded barrel (Fig. 3a). This model was developed from a set of 25 strictly conserved and identical residues identified using multiple sequence alignment, together with the observed identical locations of these residues within the crystal structures of FepA, FhuA, FecA, BtuB and FpvA. It is notable that all the conserved residues are located below the A, B and C apices, the binding sites of the ferric siderophores, and are therefore not involved in the primary binding of these substrates (van der Helm, 2004). The presence of several conserved residues within the plug region, named the ‘lock region’, and its strand structure observed in the crystals, allowed us to design a reliable structure for this region in the FatA model (Fig. 3a, b). However, there are no strictly conserved residues in the multiple sequence alignment or structurally conserved residues in the first 10 transmembrane strands. Furthermore, no conserved residues are found in the extracellular loops; the assignment of the strands 1–9 in FatA was derived from their position in the known crystal structures delineated in the multiple sequence alignment. Thus, their locations are to some extent uncertain, especially taking into account the different lengths of the extracellular loops in the various structures. Therefore, the model will not be used to analyse the region of the substrate primary binding site on the extracellular loops. Fortunately, due to the presence of strictly conserved residues, this uncertainty does not occur for strands 10–22, which are also structurally conserved in the known crystal structures. The point mutations described in this paper involved only the conserved and identical residues of the plug region and strands 11, 14 and 16 of the barrel of FatA. The conserved residues were also the basis for the development of the FatA model and one can therefore expect them to be in the correct position and show the proper interactions with other conserved residues. As ferric-siderophore transport is a process that should be common to all the proteins in this family, our hypothesis is that these conserved residues are involved in the TonB2-dependent transport function of the FatA protein.

⁵⁵Fe-anguibactin transport in V. anguillarum
The investigations of ⁵⁵Fe-anguibactin uptake transport performed with the fatA mutant indicated that FatA is the only outer-membrane receptor able to transport anguibactin in V. anguillarum 775 (Fig. 4a). As controls in this study, we used the wild-type 775 strain together with the single and double tonB1 and tonB2 mutant strains and the results obtained confirmed that anguibactin transport is mediated by the TonB2 system (data not shown), as it was previously described in our laboratory by using bioassays (Stork et al., 2004). Additionally, we have characterized the transport properties of a V. anguillarum fatA mutant unable to produce anguibactin. Since, as we have previously described, the chromosomally encoded angA gene is essential for anguibactin biosynthesis, we constructed the double fatA angA mutant strain (Alice et al., 2005). The results obtained with this mutant (Fig. 4a) indicated that the de novo synthesis of anguibactin does not affect the interpretation of the experiments described in the following sections. These results were confirmed even at the lowest ⁵⁵Fe-anguibactin concentration used in the transport and binding assays (data not shown).

View larger version (32K): [in this window] [in a new window]   Fig. 4. (a,b) 55Fe-anguibactin uptake kinetics of the single fatA mutant, double fatA angA mutant and the FatA site-directed mutants of the quadrupole region. The values represent the mean±SD of at least three independent experiments. The 55Fe-anguibactin concentration used was 10 nM. (c) SDS-PAGE of isolated outer-membrane fractions of the FatA mutants. Locations of FatA and OmpU are denoted with arrows. M, molecular size markers.

The FatA model described above, together with the sequence alignments suggest that some of the residues belonging to the ‘lock region’ in FatA consist of the charged amino acid residues R95, K130, E505 and E550 (Fig. 3b). Therefore, these residues were mutated to alanine and to glutamine in order to determine both the transport kinetics and the binding parameters of ferric-anguibactin. We determined the latter by means of dissociation constants of the wild-type and each site-directed mutant. Characterization of transport kinetics of the wild-type FatA indicated that this receptor transports ⁵⁵Fe-anguibactin with a K_m of 8 nM and a V_max of 56 fmol min^–1 per 10⁸ cells (Table 4).

The expression of the FatA variant proteins shown in Fig. 4(c) indicated that in all of the site-directed mutants the receptor is located in the outer membrane. Densitometry analysis of the gels indicated that the mutants synthesized the receptor in similar amounts as compared to the wild-type (Fig. 4c). We only detected lower amounts of FatA in the R95Q and F522A receptor mutants, with the levels of these variant receptors twofold lower as compared to the wild-type as determined by analysing outer-membrane fractions of three independent cultures. These results confirmed that the absence of transport in several of the mutants is not due to a problem in the expression of the proteins but to their structural characteristics. Curiously, for the R95A, R95Q, G96A, K130A, K130Q and F522A mutants we observed two bands at the FatA location (Fig. 4c). The densitometric analyses mentioned above were performed with the higher molecular mass band; however, these results suggest that there might be increased FatA degradation in these mutants because the lower molecular mass band also showed cross-reactivity with the FatA antisera (data not shown). The K_D values shown in Table 4 indicate that this degradation product does not affect the proper binding of the ferric-siderophore complex.

The FatA variant receptor in which the conserved glycine G96 was replaced with alanine showed similar K_m values as compared to the wild-type receptor; however, there is a sixfold decrease in the V_max of this variant receptor (Table 4, Fig. 4a). The F522 residue, which might support the side chain of the R95 residue, when replaced with alanine resulted in variant protein with lower transport rates as compared to that of the wild-type (Fig. 4a) but a similar K_m value (Table 4). These results suggest that although these two residues are supporting the quadrupole cluster they are not indispensable for ferric-anguibactin transport.

The R95 residue in FatA will form salt bridges with the glutamic acid residues at position E505 and E550, while the K130 will form a salt bridge with the E550 (Fig. 3b). The substitutions of these glutamic acids by glutamines should allow the formation of hydrogen bonds instead of the salt bridges previously mentioned. However, an alanine substitution will not even allow the formation of a hydrogen bond. Our results indicate that a change in the basic group at position 95 by a polar but uncharged glutamine or by a non-polar aliphatic residue such as alanine affects the points of interaction between the globular domain and the barrel, without affecting the binding properties of the transporter (Fig. 4a, Table 4).

The sequence alignment of various siderophore receptors showed another conserved arginine in the quadrupole (data not shown). This residue can also be a lysine in some receptors, such as K130 in FatA. One significant feature about the FatA quadrupole is that mutations in this basic group, either to alanine or to glutamine, resulted in a complete loss of transport, suggesting that this residue, as is the case for the R95, is also essential for the proper interaction of the plug domain with the barrel. Table 4 shows that the binding properties of the K130A and K130Q variant proteins were similar to that of the wild-type strain.

In the same vein, substitution of the glutamic acid at position 550 by alanine also affects this region, not allowing the formation of hydrogen bonds and preventing ferric-anguibactin transport (Fig. 4b). However, its replacement by glutamine resulted in a FatA variant receptor that had wild-type K_m and V_max values (Table 4). The K_m and V_max values of the mutated E505A receptor were similar to those of the wild-type (Table 4). In the case of the E505Q receptor mutant, the K_m showed similar values as compared to the wild-type receptor; however, we observed a twofold decrease in the V_max value (Table 4). The results obtained with these mutants are in agreement with our hypothesis in which the residue E550 most probably forms bonds with both the R95 and K130 residues.

Site-directed mutations in the ‘channel region’ of FatA
In addition to the mutations of residues located in the ‘lock region’ of FatA, we also performed site-directed mutagenesis in several conserved glycines in a specific region we called the ‘channel region’ (Fig. 5a). We hypothesize that the 5-6 loop of the plug needs to change its conformation to allow the ferric-siderophore to move from the apices towards the periplasm (Fig. 5b). We propose that these conserved glycines allow this conformational change to occur. Therefore, a glycine to alanine substitution would make this structural rearrangement more difficult or even impossible, leading to a decrease or inhibition of anguibactin transport.

View larger version (37K): [in this window] [in a new window]   Fig. 5. The ‘channel region’ of FatA. (a) Sequence alignment of part of the sequences harbouring the ‘channel region’ of 23 different outer-membrane receptor proteins. The conserved glycines are indicated in bold and the conserved arginine located in strand 11 of the barrel is in underlined bold. (b) Schematic representation of the barrel of FatA viewed from the top. The barrel is indicated in white and for clarity the extracellular loops have been removed from the figure. The plug structure is shown with secondary structure. The green spheres represent the four residues of the ‘lock region’, the purple spheres the conserved glycines of the ‘channel region’ and the red sphere the conserved arginine located in strand 11.

Most of the variant receptors of the ‘channel region’ showed similar K_D values within the standard deviation range as compared to the wild-type, indicating that the mutations did not affect the binding of ferric-anguibactin (Table 5). We only detected an increased K_D value for the single G138A mutant and the double G138A G143A mutant strain (Table 5).

Although the single G131A variant receptor showed slower transport rates, after 15 min of incubation the accumulated ⁵⁵Fe-anguibactin complex reached similar values within the cells to that of the wild-type (Fig. 6a). Moreover, this mutant showed a fivefold increase in the K_m value and a 10-fold decrease in the V_max (Table 5). This defect in transport is expected because this residue is adjacent to residue K130, an essential component of the quadrupole as stated in the previous section. The single G138A variant receptor showed similar transport rates to the wild-type (Fig. 6a). However, the K_m value is threefold higher as compared to the wild-type receptor. The lower V_max detected for this receptor suggested that its transport efficiency was lower than that of the wild-type receptor (Table 5). In the case of the single G143A mutant, we observed a similar transport rate to the wild-type; however the K_m value determined is threefold higher (Table 5).

View larger version (27K): [in this window] [in a new window]   Fig. 6. 55Fe-anguibactin uptake kinetics of FatA site-directed mutants. (a) Single mutants of the ‘channel region’. (b) Double and triple site-directed mutants of the ‘channel region’. The values represent the mean±SD of at least three independent experiments. The 55Fe-anguibactin concentration used was 10 nM.

It could also be possible that a conserved arginine located on strand 11 of the barrel (R428 in FatA), which is also conserved in all the structures, might be part of this putative ‘channel region’ (Fig. 5a, b). In the case of this conserved residue, we replaced it with both alanine and glutamic acid. Our results indicated that when this basic residue is changed to a non-polar aliphatic residue such as alanine or to a residue with an opposite charge, such as glutamic acid, transport of the ferric-siderophore is abolished (Fig. 6a), indicating the importance of this arginine residue during the transport process.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

 
In this work we characterized the transport properties of site-directed mutants in the outer-membrane receptor FatA of V. anguillarum. The mutated residues were selected based on sequence alignment and structural comparisons of several outer-membrane receptors. The quantitative determination of the dissociation constants in the presence of a respiratory inhibitor showed normal binding of the ferric-anguibactin complex for all the mutants examined, which is concomitant with the selected residues being located below the binding site. The binding affinity of FatA to ferric-anguibactin is about 100-fold lower than those described for the E. coli FepA and ferric enterobactin (Chakraborty et al., 2003; Annamalai et al., 2004) as well as for FhuA and ferrichrome (Annamalai et al., 2004) and similar to the low transport rates for enterobactin and corynebactin described by Jin et al. (2006) in the Gram-positive L. monocytogenes. Whether the low transport rate observed in the FatA receptor of V. anguillarum is an intrinsic characteristic of this receptor or a general characteristic of this species' ferric-siderophore receptors remains to be determined.

Studies performed on the E. coli outer-membrane receptors FepA, FecA, FhuA and BtuB indicated that some regions of these proteins are conserved, as assessed not only by sequence alignments but also by structural analysis (Chimento et al., 2005). In fact, the important feature of these particular regions is that they are structurally conserved in location, orientation and bonding pattern in all these proteins. Therefore using molecular modelling, we identified amino acid residues that could form the ‘lock region’ quadrupole in FatA. Our modelling suggests that this region contains the charged residues R95, K130, E505 and E550, which form two separate points of contact between the plug or globular domain of the receptor and the barrel, which keep the globular domain in its place within the barrel. We were able to confirm this prediction by site-directed mutagenesis and uptake kinetics experiments. The transport kinetics of these mutants indicated that arginine 95 and lysine 130, located in the quadrupole, are essential for ferric-anguibactin transport because mutations of these residues to both alanine and glutamine resulted in a complete loss of transport. Moreover, the binding properties of these mutants were similar to that of the wild-type, indicating that the variant receptors are properly folded in the outer membrane. In the case of FepA, the conserved R75 when changed to glutamine showed a strong adverse effect on transport, suggesting that it must be essential for the ‘lock region’. The determinations performed previously on the E. coli ferrichrome receptor FhuA, in contrast, indicated that the substitution of R133 to alanine did not have any effect on transport (Endriss et al., 2003).

Conversely, when the conserved glutamic acid E505 on FatA was mutated to alanine or glutamine, no defect in transport was observed. Mutations in the E550 residue demonstrated its essential role in ferric-anguibactin transport. The substitutions of the glutamic acids by glutamines would be permissive for the formation of hydrogen bonds. Therefore, these hydrogen bonds might replace the salt bridges that are present in the wild-type receptor. However, substitution of a simple aliphatic residue, as is the case for alanine, would not result in the formation of hydrogen bonds. This prediction is corroborated by the results of the site-directed mutants. In another outer-membrane receptor, FepA, substitutions of residue E567 for glutamine results in a greater defect in transport when compared with substitutions in E511 (Chakraborty et al., 2003). Similar results were obtained in the case of the FatA E505A and E505Q residues. In the ferrichrome receptor FhuA, substitutions of the conserved glutamates E522 and E571 by alanines did not show any defect in transport (Endriss et al., 2003). Similar results were obtained when the E541 residue of FecA was substituted by alanine. Substitution of glutamate E587 by alanine in this receptor resulted in reduced transport rates as compared with the wild-type FecA (Sauter & Braun, 2004).

Comparison of our results with those described for FepA, FhuA and FecA led us to hypothesize that the outer-membrane receptor quadrupoles might have some similarities and particularities in their involvement in the transport process. Although some results suggest that the transport mechanism would be comparable for both V. anguillarum and E. coli cells, only some of the residues of the quadrupole appeared to be essential for the transport of a specific ferric-siderophore complex. Taking into account our results together with those described for FepA, FhuA and FecA, it is therefore possible to speculate that the essentiality of some of the quadrupole amino acid residues belonging to the ‘lock region’ might depend on the type of siderophore, because the size and chemical properties of each ferric-siderophore complex are fairly different.

Two different theories explain the passage of the ferric-siderophore complex through the outer-membrane receptors. One of them postulates that the plug domain remains within the -barrel and that during ferric-siderophore transport, allosteric transitions of the plug domain and the barrel would result in the opening of a channel through which the ligand permeates into the periplasm (Ferguson et al., 2002). Conversely, another theory proposes that the plug domain is removed from the barrel during the passage of the ferric-siderophore. Recently, the three-dimensional structure of the E. coli TonB protein in complex with the cobalamine receptor BtuB (Shultis et al., 2006) and the ferrichrome receptor FhuA (Pawelek et al., 2006) were solved. In both articles the authors suggest that a conformational change or localized unfolding of the plug domain must occur to open a permeation path that would allow siderophore translocation into the periplasm. Only two studies performed in FhuA (Eisenhauer et al., 2005; Endriss et al., 2003) suggest that a partial unfolding of the plug domain occurs during transport through this outer-membrane receptor. However, whether the plug is partially or totally removed is still a matter of controversy. Although, the removal of the plug domain during TonB-dependent transport is unlikely to occur because this domain forms a large amount of intramolecular polar contacts with the barrel (Buchanan et al., 1999; Chimento et al., 2003; Ferguson et al., 1998, 2000, 2002, 2001; Locher et al., 1998; Yue et al., 2003), in accordance with Faraldo-Gömez et al. (2003) and Chimento et al. (2005), the extensive solvation of the barrel-plug interface might offer a solution to this problem.

In this study we postulate that the glycines present in the 5-6 loop of the outer-membrane receptors are important for the conformational change or partial unfolding necessary to open a channel that allows transport through the barrel lumen. In contrast to other amino acids, glycines have a larger degree of freedom and a much larger range of allowed values for the phi/psi values in a Ramachandran plot. Therefore, rotations around glycines are favoured when a conformational change takes place within a protein or domains, since they serve as hinges between different domains. In the structure of FepA there are two glycines located at positions G127 (i) and G134 (i+7) that are part of the loop connecting the strands 5 and 6, located in the middle two strands of the central mixed -sheet of the plug domain (Chakraborty et al., 2003, and Fig. 5a). The results obtained with FepA indicated that single mutations of these two conserved glycines to alanine affected transport but not binding. Moreover, only the double mutant of G127A (i)/G134A (i+7) showed a 12-fold increase in the K_m value without affecting binding (Chakraborty et al., 2003). In the case of the V. anguillarum FatA receptor, single mutations at positions G131 (i), G138 (i+7) and G143 did not affect transport; however, it was affected by a triple mutation of these conserved residues. Furthermore, double mutations either did not affect transport, such as the G131A G138A and G138A G143A mutations, or abolished it, as was the case for the G131A G143A mutation. Altogether these results are more consistent with a model in which the conserved glycines located in the 5-6 loop could allow the rotation of these strands to form a transient channel (Fig. 5b). This model could be easily integrated with that in which a conformational change or partial unfolding of the plug might occur during ferric-siderophore transport. Since mutations of these three conserved glycines to alanine will result in small local changes in the FatA structure, it is unlikely that they will impede the complete removal of the whole plug domain. However, at this point we cannot rule out the possibility that a partial removal of the plug might occur during the transport of the ferric-anguibactin complex. Experiments to test our hypotheses are currently being performed.

	ACKNOWLEDGEMENTS

 
We are grateful to Drs Dick van der Helm and Lothar Esser for their contribution in designing the FatA model and to Dr van der Helm for stimulating discussions. We are also grateful to Dr M. Jalal for advice concerning anguibactin purification. This work was supported by grants AI19018 and GM64600 from the National Institutes of Health to J. H. C.

Edited by: J. Tommassen

	REFERENCES

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

 
Actis, L. A., Fish, W., Crosa, J. H., Kellerman, K., Ellenberger, S. R., Hauser, F. M. & Sanders-Loehr, J. (1986).