INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

The multifunctional FhuA receptor protein in the outer membrane of Escherichia coli K-12 serves as a binding site for the phages T1, T5, 80, and UC-1, for the entry of colicin M and microcin 25, and for the uptake of ferrichrome and the structurally related albomycin. With the exception of the infection by phage T5, killing by the phages and the toxins and transport of the iron complexes require the Ton system, which is composed of the proteins TonB, ExbB, and ExbD (6). It is thought that by consumption of energy, provided by the electrochemical potential of the cytoplasmic membrane, FhuA is converted from the ground state to the energized conformation, which is recognized by the Ton-dependent phages and triggers DNA release from the phage head. Inhibition of phage T5 binding by ferrichrome provided evidence for a conformational change of FhuA mediated by TonB. Much less ferrichrome was required to inhibit T5 binding by ferrichrome in unenergized cells and tonB mutants than in energized tonB⁺ cells (12). It is further assumed that energized FhuA forms an open channel through which the toxins and ferrichrome, which bind to FhuA independent of the Ton system, enter the cytoplasm. Evidence for a FhuA channel was obtained by excision of most of a surface-exposed loop (21), which converted FhuA into a permanently open channel through which ferrichrome, sodium dodecyl sulfate (SDS), and certain antibiotics diffused in vivo (15) and which rendered artificial bilayer membranes permeable to anions and cations (15). The single-channel conductance of the FhuA deletion mutant was at least three times larger than that of the E. coli OmpC and OmpF porins (15). From these results the region comprising residues 322 to 355, or the entire loop (residues 316 to 356), was proposed to form a gating loop that closed the FhuA channel unless it was moved by input of conformational energy provided by the TonB protein. Accessibility of the gating loop from the cell surface was derived from proteolytic cleavage of FhuA in viable cells within inserted 4- and 16-residue peptides after amino acid 321 (21), reaction with a monoclonal antibody (27) and an antibody directed against a C3 viral reporter epitope inserted after residue 321 (25), and labeling of cysteine 318 and cysteine 329 after reduction of the disulfide bridge and of an inserted cysteine (Asp336Cys) with biotin-maleimide and fluorescein-maleimide (4, 5). Interaction of FhuA with TonB was evidenced by suppressor mutations in TonB that partially restored activity of FhuA carrying certain point mutations in the TonB box, a conserved pentapeptide located close to the N terminus of all Ton-dependent outer membrane proteins and colicins (32). Furthermore, degradation of overexpressed TonB by cellular proteases was prevented by overexpressed FhuA (11), by chemical cross-linking of FhuA to TonB, and by retention of TonB on a nickel column loaded with histidine-tagged FhuA (26). Interaction of FhuA with TonB was increased by the ferrichrome homolog ferricrocin (26), suggesting induction of an FhuA conformation by ferricrocin that favors binding to TonB. Indeed, ferrichrome induced a conformational change of FhuA, as was deduced from inhibition of FhuA degradation by added proteases in the presence of ferrichrome (13, 28). In isolated FhuA phage T5 opens a channel without cellular energy input and involvement of the Ton system (3). The physical characteristics of the channel are similar to the channel formed by the FhuA deletion derivative (15). FhuA inserted into liposomes induced the release of phage T5 DNA and its transfer inside the vesicles (30). Phage T5 triggered release of ferrichrome trapped in proteoliposomes containing FhuA (23). These data prove beyond a doubt that FhuA contains a closed channel which is opened upon binding of phage T5. Binding of the other FhuA ligands is not sufficient for opening of the channel but requires in addition an input of energy. However, the differences between the way T5 opens the channel and the ways the other FhuA ligands do so may not be as serious as they seem. Certain tonB point mutations increase T5 infection of certain FhuA mutants 100-fold (18), and host range mutants of phage T1 infect tonB deletion mutants with high efficiency (12).

Further analysis of the FhuA gating loop by competitive peptide mapping revealed three subdomains involved in phage binding which are distributed over the entire loop (20). Deletion of residues 322 to 336 of the gating loop strongly reduced phage binding but not ferrichrome binding. FhuA with residues 322 to 336 deleted (FhuA 322-336) supported Ton-dependent ferrichrome transport at a somewhat reduced rate (16). In contrast, FhuA 335-355 did not bind ferrichrome and did not actively transport ferrichrome across the outer membrane. It formed a permanently open channel through which ferrichrome, SDS, and maltotetraose diffused into the periplasm (16). Apparently, the segment containing residues 335 to 355 mainly controls the permeability of FhuA.

If only half of the gating loop is essential for ferrichrome transport and the entire loop is required for phage sensitivity, analysis of the FhuA proteins of E. coli-related bacteria may disclose regions that determine specificity for ferrichrome transport and phage infection. Salmonella paratyphi, Salmonella typhimurium (10), and Pantoea agglomerans (formerly Erwinia herbicola) (2) transport ferrichrome and structurally related ferric hydoxamates (22), but only S. paratyphi is sensitive to the E. coli phages. Therefore, we sequenced the fhuA genes of these strains and found that the largest difference between the FhuA proteins is located in the predicted gating loops. The activities of the FhuA proteins are consistent with the proposal that ferrichrome uptake is mainly determined by the region equivalent to residues 335 to 355 of E. coli FhuA. In addition, deletion mutations in a predicted loop near the gating loop inactivated the ferrichrome transport activity of FhuA but left its role in phage infection and colicin M killing largely intact, which indicates its specific involvement in ferrichrome transport.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Bacterial strains, cosmids, plasmids, and growth conditions. The E. coli strains and plasmids used in this study are listed in Table 1. Cells were grown in TY medium (Bacto Tryptone [10 g/liter; Difco Laboratories], yeast extract [5 g/liter], NaCl [5 g/liter]) or NB medium (nutrient broth [8 g/liter], NaCl [5 g/liter] [pH 7]) at 37°C. To reduce the available iron of the NB medium, 2,2'-dipyridyl (0.2 mM) was added (NBD medium). The antibiotics ampicillin (40 µg/ml) and neomycin (50 µg/ml) were added when required.

Recombinant DNA techniques. Isolation of plasmids, use of restriction enzymes, ligation, agarose gel electrophoresis, and transformation were done by standard techniques (31). DNA was sequenced by the dideoxy chain termination method with fluorescence-labeled or unlabeled nucleotides (Auto Read Sequencing Kit; Pharmacia Biotech, Freiburg, Germany) and the ALF sequencer (Pharmacia).

Protein analytical methods. E. coli WM1576 transformed with various plasmids encoding the receptors were collected by centrifugation at an optical density at 578 nm of 0.4 and resuspended in 1 ml of M9 salts (24) supplemented with 0.4% glucose, 0.01% methionine assay medium, and 0.01% thiamine. After shaking the cells for 1 h at 27°C, T7 RNA polymerase synthesis was induced by shifting the temperature to 42°C for 15 min. Rifampin (10 µl; 5 mg/ml in methanol) was added, and incubation continued at 27°C for 20 min. [³⁵S]methionine was added, and the suspension was incubated for an additional 10 min before cells were collected by centrifugation and suspended in sample buffer. The radioactively labeled proteins were separated by SDS-polyacrylamide gel electrophoresis (SDS-PAGE) as described previously (17).

Phenotype assays. All phenotype assays were performed with freshly transformed E. coli K-12 strain HK97 aroB fhuA fhuE. The sensitivity of cells to the FhuA ligands (phages T1, T5, and 80 and colicin M and albomycin) was tested by spotting 10-fold diluted solutions (4 µl) on TY plates overlaid with 3 ml of TY soft agar that contained 10⁸ cells of the strain to be tested. The colicin M solution was a crude extract of a strain that carried the plasmid pTO4 cma cmi (29). Competition experiments between ferrichrome and phage infection and colicin M killing, respectively, were done with final ferrichrome concentrations of 10 and 100 µM in the 3 ml of TY soft agar that contained 10⁸ cells of the strain to be tested. The TY nutrient agar plates were incubated for 15 min at 37°C before the spot tests were performed as described above.

Transport assays. E. coli K-12 HK97 aroB fhuA fhuE freshly transformed with the plasmids to be tested was grown overnight on TY plates. Cells were washed and suspended in transport medium (M9 salts [24], 0.4% glucose) before the cell density was adjusted to an optical density at 578 nm of 0.5. Free iron ions were removed by adding 25 µl of 10 mM nitrilotriacetate, pH 7.0, to 1 ml of cells. After incubation for 5 min at 37°C, transport was started by adding 10 µl of 100 µM [⁵⁵Fe³⁺]ferrichrome. Samples (100 µl) were withdrawn; cells were harvested on cellulose nitrate filters (pore size, 0.45 µm; Sartorius AG, Göttingen, Germany) and washed twice with 5 ml of 0.1 M LiCl; the filters were dried; and the radioactivity was determined by liquid scintillation counting.

Computer-assisted sequence analysis. Computer-assisted sequence analysis was performed with the program package PC.GENE and the BLAST homology search (1).

Nucleotide sequence accession number. The nucleotide sequences reported in this study were deposited in the EMBL data bank under accession numbers Y14026 [fhuA(Pa) and partial fhuC(Pa)], Y14067 [fhuA(Sp) and partial fhuC(Sp)], and Y14025 [fhuA(St) and partial fhuC(St)].

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Nucleotide sequences of fhuA genes and derived amino acid sequences. fhuA(Pa), fhuA(Sp), and fhuA(St) have been cloned previously on the cosmids pUH62, pUH65, and pUH66, respectively (22). Restriction fragments were cloned on plasmids pT7-5 and pT7-6 and transformed into E. coli HK97 fhuA fhuE aroB to test sensitivity to albomycin and growth on ferrichrome as the sole iron source. The fhuA mutation of HK97 displays no polar effect on the transcription of the downstream-located fhuCDB genes, which are required for transport of albomycin and ferrichrome across the cytoplasmic membrane. Due to the aroB mutation, strain HK97 did not produce its own siderophore enterobactin, and therefore growth on NBD plates or transport of iron in M9 medium depended on the ability to take up ferrichrome.

View larger version (77K): [in this window] [in a new window]   View larger version (74K): [in this window] [in a new window]   FIG. 1.   Sequence alignment of various FhuA proteins. Asterisks denote identical residues, and dots indicate similar amino acids. Amino acids representing the gating loop are shown in shaded boxes. The amino acids deleted in FhuA 236-243 (8aa) and in FhuA 236-248 (13aa) are indicated. The TonB box sequences are underlined.

The largest difference between the FhuA proteins is located in the predicted gating loops. Alignment of the FhuA amino acid sequences, related to the FhuA(Ec) gating loop, disclosed a gap of 17 residues in FhuA(St) and a gap of 14 plus 3 residues in FhuA(Pa) (Fig. 1). There are a few other sequence gaps that are introduced to obtain the highest degree of sequence identity and homology between the FhuA proteins of which the largest consists of six residues. At this site FhuA(Pa) is six residues larger than FhuA(Sp), FhuA(St), and Fhu(Ec), the three of which are identical (Fig. 1). The sequence gaps are confined to the first half of the E. coli gating loop, for which a previous deletion analysis has revealed its importance for phage infection but not for the uptake of ferrichrome and albomycin (16).

Functional analysis of the predicted gating loop. The activities of the FhuA proteins of S. parathyphi, S. typhimurium, and P. agglomerans were determined in E. coli HK97 fhuA fhuE aroB transformed with the cloned heterologous fhuA genes. E. coli HK97 fhuA(Sp) was as sensitive to the phages T1, T5, and 80 and to colicin M and albomycin and grew as well on ferrichrome as the sole iron source as E. coli HK97 transformed with the E. coli fhuA(Ec) gene (Fig. 2). This result was expected because of the high sequence identity of FhuA(Sp) and FhuA(Ec). FhuA(St) took up albomycin and ferrichrome but conferred no sensitivity to the E. coli phages and to colicin M but did confer sensitivity to the S. typhimurium phage ES18 (Fig. 2). Similar results were obtained with E. coli fhuA(Pa), which was resistant to the E. coli and S. typhimurium phages and to colicin M but took up albomycin and ferrichrome (Fig. 2). The ferrichrome transport rates of E. coli HK97 transformed with fhuA(Ec), fhuA(Sp), fhuA(St), and fhuA(Pa) were very similar (Fig. 2 and 3). The initial transport rate of E. coli HK97 fhuA(Pa) was consistently higher than those of the other transformants, which could reflect a stronger binding of ferrichrome to FhuA(Pa) or the presence of somewhat larger amounts of FhuA(Pa) in the outer membrane. After transcription by the T7 RNA polymerase, the amounts of the [³⁵S]methionine-labeled proteins were very similar and no degradation products were observed, as revealed by SDS-PAGE (data not shown), which indicated a similar expression of the genes and a high stability of the proteins.

View larger version (33K): [in this window] [in a new window]   FIG. 2.   Schematic representation of wild-type fhuA genes. The enzymes used to construct the chimeric receptor derivatives are indicated. The sensitivities to the phages T1, T5, 80, and ES18, to colicin M (ColM), and to albomycin (Albo) are given as the last of a 10-fold dilution series which resulted in a clear zone of growth inhibition. Numbers in parentheses indicate zones of turbid growth inhibition. Dashes indicate no growth inhibition. Growth promotion by ferrichrome (Fc) was tested by placing filter paper disks supplemented with 10 µl of ferrichrome (1 mM) onto NBD plates overlaid with NB top agar containing 108 cells of the strain to be tested. The results are given as the diameter of the growth zone (in millimeters) around the filter paper disk (6 mm) (the diameter of the discs was not subtracted from the values given). Transport indicates ferrichrome transport rates in units of 1,000 Fe3+ ions transported per cell per minute (average of three determinations without corrections). The positions of the gating loop and of the loop containing residues 236 to 257 as well as the number of the exchanged amino acids of FhuA(St) and FhuA(Sp) are indicated. The gating loop of FhuA(Ec) and of FhuA(Sp) consists of 41 amino acids (aa), while the gating loop of FhuA(St) and of FhuA(Pa) consists of only 24 aa.

View larger version (17K): [in this window] [in a new window]   FIG. 3.   Time-dependent transport of [55Fe3+]ferrichrome (1 µM) into E. coli HK97 fhuA fhuE aroB expressing the plasmid-encoded FhuA wild-type proteins of E. coli(pHK763), P. agglomerans(p75Pa), S. paratyphi(p76Sp), and S. typhimurium(p76St).

View larger version (56K): [in this window] [in a new window]   FIG. 4.   Comparison of outer membrane preparations of [35S]methionine-labeled FhuA hybrid proteins [hybrids of FhuA(St) and FhuA(Sp)] (lanes 2 to 7), FhuA(Sp) (lane 1), and FhuA(St) (lane 8) after SDS-PAGE (see Table 1 and Fig. 2 for details).

Identification of an additional site in FhuA for ferrichrome uptake. The sequences of all four mature FhuA proteins are identical from residues 234 to 248 (Fig. 1). The amino acid sequence from residues 236 to 257 of FhuA(Ec) was predicted to form a surface-exposed loop adjacent to the gating loop (21). To examine whether this highly conserved region is involved in ligand binding, small (8- and 13-amino-acid) deletions were made in FhuA(Ec). The fhuA(Ec) deletion derivatives were constructed by ligating shortened PCR fragments using introduced BamHI restriction sites. The deletions extended from residues 236 to 243 (FhuA 236-243) and from residue 236 to 248 (FhuA 236-248) (Fig. 1) and comprised 8 and 13 FhuA amino acids, respectively, but the genetic technique used to construct these deletions introduced two residues (glycine and serine). In the outer membrane the proteins of both deletion mutants were found in amounts similar to those of wild-type FhuA(Ec) (data not shown). FhuA 236-243 and FhuA 236-248 were resistant to albomycin and showed no growth on ferrichrome (Table 2). This was confirmed by transport assays with [⁵⁵Fe³⁺]ferrichrome, in which both deletion derivatives showed no transport and no binding of ferrichrome (Fig. 5). However, the deletion derivatives conferred full sensitivity to phages T1 and T5 and a reduced sensitivity to phage 80 and to colicin M (Table 2). Phage and colicin sensitivity provided an additional means to measure ferrichrome binding, because ferrichrome competes with these FhuA ligands, except phage T5. Only TonB mutants are protected from T5 infection by ferrichrome unless high ferrichrome concentrations are used (12). Sensitivity of E. coli HK97 that synthesized wild-type FhuA(Ec) was reduced 10-fold to phage T1, 80, and colicin M by 10 µM ferrichrome added to the soft agar and 1,000-fold to 80 by 100 µM ferrichrome, whereas the sensitivity of cells that synthesized the FhuA deletion derivatives was not altered by ferrichrome (Table 2).

View larger version (14K): [in this window] [in a new window]   FIG. 5.   Time-dependent transport of [55Fe3+]ferrichrome (1 µM) into E. coli HK97 fhuA fhuE aroB transformed with plasmid pHK763, which encodes wild-type FhuA(Ec), or with plasmids pB3/4 and pB4/5, which encode the FhuA(Ec) deletion derivatives FhuA 236-234 and FhuA 236-248, respectively.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

The kinetics of ferrichrome transport into E. coli are composed of two rates, high initial binding to FhuA and lower transport into cells. The site for binding and transport was localized to one-half of the gating loop by the deletion of amino acids 335 to 355, which reduced binding and transport to less than 10% of that of FhuA wild-type and opened unspecifically the FhuA channel for SDS and maltodextrins, whereas the deletion of residues 322 to 336 did not affect ferrichrome binding and reduced the transport rate to 60% (16). The conclusion that the region containing residues 335 to 355 mainly determines the permeability of the FhuA channel is supported by the structures and properties of the FhuA proteins described in this work. FhuA(Sp), FhuA(St), and FhuA(Pa) contain the region equivalent to the region containing residues 335 to 355 of FhuA(Ec) and transport ferrichrome. Lack of the region containing residues 322 to 336 in FhuA(St) and FhuA(Pa) indicates that this region is dispensable for binding and transport of ferrichrome and, in general terms, for closing and opening of the FhuA channel. Only FhuA(Sp) contains a surface loop of the size and sequence of FhuA(Ec). FhuA(Sp) expressed in E. coli HK97 fhuA confers the same degree of sensitivity to the E. coli phages and to colicin M as does FhuA(Ec). This result per se does not indicate that the region containing residues 322 to 355 is required for phage infection and colicin killing, since the overall identity of FhuA(Sp) and FhuA(Ec) is 92%, but it does agree with the previous data which demonstrate the importance of the entire E. coli gating loop for phage infection (17, 20). In the mature proteins only conservative amino acid replacements occur (replacement of E by Q, E by D, F by Y, S by A, T by V, K by R, V by I, and N by Q), except at five sites, where KGS of E. coli is replaced by DRA, KDGN is replaced by ENGK, RP is replaced by AA, PE is replaced by SA, and R is replaced by G in S. paratyphi B. In the proposed FhuA(Ec) transmembrane model (21), the nonconservative replacements are contained in surface loops, except one which is located in a cytoplasmic turn (RP replaced by AA). This finding is consistent with the rule that loops and turns are more variable and tolerate nonconservative amino acids replacements more easily than membrane-spanning regions. In addition, surface loops determine ligand binding specificity. The locations of the conservative and nonconservative amino acid replacements support the transmembrane topology model of FhuA(Ec) and suggest the same arrangement for FhuA(Sp), FhuA(St), and FhuA(Pa).

An additional site important for ferrichrome uptake was localized in the surface loop containing residues 236 to 257, with residues 236 to 248 having an identical amino acid sequence in all four FhuA proteins. Deletion of residues 236 to 243 and residues 236 to 248 in FhuA(Ec) inactivated ferrichrome transport and rendered cells resistant to albomycin. In contrast, the FhuA deletion derivatives conferred high phage and colicin sensitivity, which indicates that their conformation was not grossly altered compared to that of wild-type FhuA and that they were properly inserted into the outer membrane. The deletion derivatives also did not bind ferrichrome, as shown by ferrichrome transport assays and the failure to prevent phage and colicin M binding by ferrichrome, which suggests that this loop is part of the ferrichrome binding site or contributes to the conformation of the ferrichrome binding site. In a previous study, insertion of a tetrapeptide after residue 241 reduced ferrichrome uptake, rendered cells albomycin resistant and reduced colicin M sensitivity 32-fold. Insertion of a 16-residue peptide reduced ferrichrome uptake and reduced sensitivity to colicin M 128-fold, to phage 80 10-fold, and to phage T5 100-fold (21). Insertion of a dipeptide after residue 239 reduced sensitivity to 80 10-fold, to phage UC-1 100-fold, and to colicin M 8-fold (8). The inserted heterologous peptides may have reduced access of the ligands to their binding sites by steric hindrance or may have distorted the conformation of the binding sites. This loop is in the two-dimensional transmembrane model (21) adjacent to the gating loop and probably contributes to the channel structure.

The specificity of the loss of ferrichrome uptake in the FhuA deletion derivatives FhuA 236-248 and FhuA 322-355 is supported by mutations in another proposed surface loop (residues 454 to 477) close to the gating loop which did not affect ferrichrome uptake. Insertion of a 12-amino-acid peptide after residue 456 only reduced sensitivity to colicin M 64-fold (21), and deletion of residues 457 to 479 resulted in mutant cells in which the FhuA-related phages formed turbid plaques (16). Monoclonal antibodies directed to residues 417 to 450 inhibited only colicin M killing and inactivation of phage T5 by E. coli cells (27).

Despite 74% sequence identity, hybrid proteins of FhuA(Sp) and FhuA(St) were only fully active when large fragments were exchanged. FhuA3/2 and FhuA9/9, composed of two-thirds of FhuA(Sp) and one-third of FhuA(St), displayed full FhuA(Sp)-specific activity and no FhuA(St)-specific activity. FhuA5/21, which consists of the central portion of FhuA(Sp) (residues 213 to 482) that forms the overlapping region in FhuA3/2 and FhuA9/9, flanked on both sides by FhuA(St) fragments, conferred full sensitivity only to phage T5 and conferred a reduced sensitivity to albomycin (caused by a reduced uptake), as evidenced by a reduced growth promotion by ferrichrome, and a very weak sensitivity to phage 80. Apparently, the gating loop of FhuA(Sp) and loops containing residues 236 to 257, 404 to 433, and 454 to 477 [according to the FhuA(Ec) model (21)] inserted in FhuA(St) are not sufficient to confer FhuA(Sp) activity. Full sensitivity to T5 and lack or reduction of activity to the other FhuA ligands point to a somewhat distorted interaction of the hybrid protein with TonB. Sensitivity to the S. typhimurium phage ES18 is observed when the hybrid protein (FhuA11/15) contains two-thirds of the N-terminal segment of FhuA(St) but not when it contains two-thirds of the C-terminal segment (FhuA1/2) of FhuA(St). Hybrid proteins of FhuA(Ec) and FhuA(Pa), which display 55.4% sequence identity, confer only a reduced sensitivity to phage T5 and colicin M when they contain residues 320 to 714 of mature FhuA(Ec). The inactive or partially active hybrid proteins may lack amino acid side chains required for binding of the ligands, or the active sites cannot adopt completely their native conformation.

Chimeric proteins consisting of the central part of FhuA(Ec), extending from amino acid 161 to 370, and the N- and C-terminal parts of FhuE (coprogen receptor) and FoxA (ferrioxamine B receptor), respectively, transported ferrichrome. The transport rates, relative to FhuA, were 38% for the FhuE-FhuA hybrid and 8% for the FoxA-FhuA hybrid (19). These results are consistent with the data presented in this paper, which localized the ferrichrome binding sites to the loops containing residues 236 to 257 and residues 316 to 356. Taken together these data and the results presented in this work demonstrate that several segments of FhuA are involved in ligand binding, in transport of ferrichrome, albomycin, and colicin M, and in phage infection and that these segments adopt certain conformations which are determined by the entire FhuA polypeptide.