Chrysobactin-dependent Iron Acquisition in Erwinia chrysanthemi FUNCTIONAL STUDY OF A HOMOLOG OF THE ESCHERICHIA COLI FERRIC ENTEROBACTIN ESTERASE*

Under iron limitation, the plant pathogen Erwinia chrysanthemi produces the catechol-type siderophore chrysobactin, which acts as a virulence factor. It can also use enterobactin as a xenosiderophore. We began this work by sequencing the 5 (cid:1) -upstream region of the fct-cbsCEBA operon, which encodes the ferric chrysobactin receptor and proteins involved in synthesis of the catechol moiety. We identified a new iron-regulated gene ( cbsH ) transcribed divergently relative to the fct gene, the translated sequence of which is 45.6% identical to that of Escherichia coli ferric enterobactin esterase. Insertions within this gene interrupt the chrysobactin biosynthetic pathway by exerting a polar effect on a downstream gene with some sequence identity to the E. coli enterobactin synthase gene. These mutations had no effect on the ability of the bacterium to obtain iron from enterobactin, showing that a functional cbsH gene is not required for iron removal from ferric enterobactin in E. chrysanthemi . The cbsH

Iron is an essential but nevertheless potentially toxic element for most living organisms. The bioavailability of the ferric ion is extremely limited because of its poor solubility (at pH 7, K sp (solubility product) of Fe(OH) 3 ϭ 10 Ϫ17 M). A wide variety of microorganisms accommodate this situation by excreting siderophores. Siderophores are high-affinity Fe(III)-scavenging/solubilizing molecules that, once loaded with iron, are specifically imported into the cell. In Escherichia coli K12, it has been demonstrated that delivery of the ferric siderophore complex to the cell implicates active transport (1,2). The passage through the outer membrane requires a receptor that is a pore energized by cytoplasmic membrane-generated proton-motive force transduced by the TonB protein. The ferric complex then binds to the periplasmic component of a permease belonging to the ABC transporter family, which completes the passage to the cytosol. The fate of the ferric siderophore complex in the cytosol is not clearly understood. As the stability constants of ferric siderophore complexes are very high and the ferrous complexes dissociate near neutral pH, enzymatic reduction to the ferrous state has been proposed to be a plausible mechanism for iron removal (3,4). Ferric siderophore reductase activity has been found in cell extracts from several microorganisms (5)(6)(7)(8). However, the redox potentials for hexadentate catechol siderophores are out of the range of physiological reductants, and it is assumed that ligand degradation is required for transformation of the irreducible form of the complex into a reducible one. In E. coli, the ester bonds of the siderophore enterobactin (enterochelin), the cyclic trimer of 2,3dihydroxybenzoyl-L-serine (9) (see Fig. 1), are hydrolyzed by the ferric enterobactin esterase encoded by the fes gene, yielding 2,3-dihydroxybenzoyl-L-serine (10 -12). The redox potential of this compound is 2 orders of magnitude below that of ferric enterobactin. fes-negative mutants fail to grow if ferric enterobactin is the only iron source (13,14). The plant pathogenic enterobacterium Erwinia chrysanthemi strain 3937 provides another illustration of this problem.
Under iron limitation, E. chrysanthemi produces the catechol siderophore chrysobactin. This siderophore is essential for this pathogen to disseminate throughout its host plant and to cause systemic soft-rot symptoms (15). Chrysobactin is a bidentate ligand consisting of a monomer of 2,3-dihydroxybenzoyl-D-lysyl-L-serine (16) (see Fig. 1). Ferric chrysobactin is transported back into the cell via its specific TonB-dependent outer membrane receptor Fct (17)(18)(19) and a cytoplasmic membrane permease that is missing in a class of mutants deficient in ferric chrysobactin uptake (20). These mutants do not acquire iron from ferric enterobactin. Enterobactin is not synthesized by E. chrysanthemi cells, but promotes growth of a chrysobactindeficient mutant if supplied exogenously. An iron-regulated outer membrane protein with an apparent molecular mass of 88,000 Da and immunologically related to the E. coli ferric enterobactin receptor FepA is thought to play a similar function in E. chrysanthemi (21). In addition, E. chrysanthemi strain 3937 produces another high-affinity iron uptake system mediated by a citrate siderophore called achromobactin (22).
Most of the proteins involved in chrysobactin-mediated iron transport are encoded by a 50-kb contiguous region of the E. chrysanthemi chromosome (20). The fct-cbsCEBA operon codes for the receptor Fct and the enzymes leading to the catechol moiety in chrysobactin biosynthesis (23) (see Fig. 1). Analysis of the fct gene sequence (18) revealed a strong resemblance of the promoter region to the bidirectional promoter controlling the expression of the fepA-entD and fes-entF operons in E. coli (24,25) (Fig. 1). The fepA gene codes for the receptor FepA (26,27). The entD and entF genes ( Fig. 1) encode the EntD and EntF proteins, respectively, which are two components of the enterobactin synthase multienzyme complex (28 -31). In E. chrysanthemi, as in E. coli and many other bacterial species, control by iron is achieved via a fur gene that encodes a protein highly similar to the E. coli Fur regulator (32). In the presence of ferrous iron as a cofactor, the E. coli Fur protein acts as a transcriptional repressor by binding to operator-specific sequences (Fur or iron boxes) (33,34).
Sequence analysis of the 5Ј-upstream region of the fct gene revealed the existence of a gene (cbsH) transcribed in the opposite direction to the fct gene that shares identity with the E. coli fes gene. The functional analysis of the cbsH gene product is presented.

EXPERIMENTAL PROCEDURES
Bacterial Strains, Plasmids, and Microbiological Techniques-The bacterial strains used are listed in Table I. Plasmids are described in Table I and Fig. 2A. The cbsE-1 and acs-37 mutations were transduced into strain L2 cbsH-19 with phage EC2 as described previously (35). Insertional mutagenesis with the MudI1734 prophage in plasmid pCS2 and marker exchange recombination into the chromosome were performed as described previously (36). Expression of the cbsH-17::lacZ fusion was monitored as reported previously (37). The rich media used were L broth and L agar (38). L agar was iron-depleted by adding EDDHA 1 (Sigma) to give a final concentration of 100 g/ml. Tris medium was used as the low-iron minimal medium (35). For iron-rich conditions, it was supplemented with 20 M FeCl 3 . Glucose (2 g/liter) was used as the carbon source. The antibacterial agents and chemicals used were as reported previously (35).
General DNA Methods-All DNA manipulations were carried out as described previously (35,36). The aphA-3 cassette, obtained by SmaI digestion of pUC18K (39), was inserted into AfeI-digested pCS2 and then recombined into the chromosome. The AfeI site lies within the cbsH gene at a position corresponding to amino acid 274.
Deletion Analysis, Nucleotide Sequence Determination, and Primer Extension-Serial truncations (200 -250 bp) of pDE34 (see Fig. 2A) were carried out from the SalI site using the double-stranded nested deletion kit from Amersham Biosciences AB (Uppsala, Sweden). The deleted subclones were polyethylene glycol-purified (40) and sequenced using the Sequenase kit (U. S. Biochemical Corp.) and [␣-35 S]dATP according to the manufacturer's instructions. The second-strand sequence was determined by extension from specific oligonucleotides (17mers). Data were analyzed using the UWGCG software package provided by BISANCE (41). Two programs (BLAST and Kanehisa) were used for amino acid sequences comparisons.
RNA templates were isolated from cultures of E. chrysanthemi strain 3937 and E. coli strain JM101 harboring pCS1 and grown in Tris medium, reaching an absorbance at 600 nm of 0.8 and 0.6 for high-and low-iron conditions, respectively. RNA isolation and primer extension analyses were performed as described previously (18). A 32 P-labeled oligonucleotide complementary to nucleotides 158 -142 was used for primer extension.
Identification of the CbsH Product-The procedure described by Tabor and Richardson (42) was used to produce proteins encoded by the pT7 derivative plasmid. Samples were radiolabeled as described previously (37). Proteins were separated by electrophoresis on SDS-8% polyacrylamide gel.
Siderophore Detection-Catechol was determined by the chemical assay of Arnow (43) using dihydroxybenzoic acid as the standard. Siderophore activity was detected as chrome azurol S-reacting material in the culture supernatant (44) using desferrioxamine B (Desferal, Novartis Pharma S. A., Rueil Malmaison, France) as the standard. The biological activities of chrysobactin and enterobactin were determined in bioassays as described previously (16) using strains RW193 and RW818-60 for enterobactin and strains L2 cbsE-1 and L2 fct-34 for chrysobactin as indicators.
Quantitative Determination of Ferric Chrysobactin in Cell Lysates-A culture (grown exponentially in L broth) of the strain to be studied was diluted 1:40 in 20 ml of Tris medium supplemented with glucose and 5 M FeCl 3 . Cultures were grown aerobically for 12-14 h. Cells were washed, suspended in 1 ml of Tris medium, and disrupted in a Vibra Cell apparatus (Sonics and Materials Inc.). The lysis mixture was centrifuged at 7000 rpm for 30 min at 4°C in a microcentrifuge (Medical Scientific Equipment, Leicester, United Kingdom). Pelleted cell debris was discarded, and the supernatant was checked for the presence of ferric chrysobactin in a bioassay. The concentrations of ferric chrysobactin in cell lysates and in culture supernatants equivalent to 5 ϫ 10 8 colony-forming units was determined spectrophotometrically. The ferric complex (at pH 7.4) had an absorption maximum at 525 nm (⑀ 525 ϭ 3.2 mM cm Ϫ1 ) (45).
Assay for Enzymatic Hydrolysis of Ferric Chrysobactin-Hydrolysis of ferric chrysobactin was assayed in cell extracts from the bacterial strains tested and prepared as following. Cells were grown aerobically in 500 ml of Tris medium supplemented with glucose until an absorbance at 600 nm of 0.6 -0.9 was reached. Cells were washed, suspended in 2 ml of 0.1 M Tris-HCl (pH 7.5) and disrupted as described above. Lysates were supplemented with dithiothreitol at a final concentration of 0.005 mM and centrifuged for 20 min at 20,000 ϫ g. The enzymatic activity in the supernatants was immediately tested. The reaction mixture (incubated at 37°C for 1 h) contained 0.300 ml of lysate, 0.430 ml of 0.1 M Tris-HCl (pH 7.5), and 0.020 ml of the bis complex of ferric chrysobactin to give a final concentration of 0.028 mM. Ferric chrysobactin was prepared by adding FeCl 3 to chrysobactin in 0.1 M Tris (pH 7.5) at a ligand/iron ratio of 4:1. Chrysobactin (a gift from Dr. J. S. Buyer) was synthesized according to the procedure described previously (46). Protein concentration in cell lysates was determined with the Bradford reagent. Hydrolytic activity was determined spectrophotometrically as described above. Enzymatic activity is expressed in nanomoles of ferric chrysobactin hydrolyzed per mg of protein/h. Diisopropyl fluorophosphate was added to the reaction mixture at a final concentration of 0.036 mM. For each strain, three independent experiments were performed.
Assay for Ferric Enterobactin Esterase Activity-For ferric enterobactin esterase activity, cell extracts from the bacterial strains tested were prepared as described above. Lysates were assayed as reported by Langman et al. (13). Enzymatic activity is expressed in nanomoles of enterobactin hydrolyzed per mg of protein/h. For each strain, three independent experiments were performed. Ferric enterobactin was prepared according to the procedure reported by Greenwood  BZB1013 was lyophilized and extracted with ethyl acetate. As a final purification step, ferric enterobactin dissolved in methanol was passed through a column of Sephadex LH-20 (30 g) with methanol as the eluent. Fractions were collected, evaporated, and dissolved in 0.1 M phosphate buffer (pH 7). Catechol-positive fractions were bioassayed using strains RW193 and RW818-60 as indicators and checked by ultraviolet spectrometry. The purest fraction yielded ϳ50 mol of ferric enterobactin.
Transport Experiments-An overnight culture in L broth of the strain to be studied was diluted 1:40 in Tris medium supplemented with glucose and incubated with shaking until the required absorbance at 600 nm was reached. Bacterial cells were harvested by centrifugation, washed with Tris medium without phosphate, suspended in phosphatefree Tris medium supplemented with glucose, and kept on ice until use. The transport medium was Tris medium containing 40 -50 M dihydroxybenzoic acid equivalents of chrysobactin and supplemented with 1 M 59 FeCl 3 (0.1 mCi/ml iron(III) chloride in 0.1 M HCl; Amersham Biosciences AB). For transport experiments, the bacterial suspension was diluted in transport medium to give A 600 nm ϭ 0.4 in a total volume of 5 ml and placed in a 50-ml Erlenmeyer flask. At intervals of 5-30 min, 200 l was withdrawn and immediately filtered through a filter with 0.45-pores that had been soaked for at least 12 h in Tris medium supplemented with 20 M unlabeled FeCl 3 . Filters were immediately washed with 20 ml of Tris medium without phosphate. The filters were placed in scintillation vials and air-dried, and radioactivity was measured by liquid scintillation counting. Two 20-l samples of each bacterial culture were counted to check the total amount of radioactivity. For each strain, experiments were performed in triplicate.
Mössbauer Measurements-For each Mössbauer measurement, a 2-liter bacterial culture in 5-liter Erlenmeyer flasks was required to obtain ϳ1 ml of packed cells. Cultures of strains L2 cbsE-1 and L2 cbsH-19 were grown in Tris medium supplemented with glucose for 12 h. The absorbance at 600 nm was 0.65. 57 Fe-Labeled chrysobactin was added to the cell suspensions at a final concentration of 1.3 M (ligand/iron ratio of 4:1.3). Cells were grown for an additional 30, 60, and 120 min. At 0 min and each additional time, cells were cooled to 4°C within 2 min, harvested, washed in Tris medium, and transferred to Delrin Mössbauer sample holders. All sample volumes were ϳ1 ml. Sample thickness did not exceed 9 mm. The containers were quickly frozen in liquid nitrogen and kept in a liquid nitrogen storage vessel until measurements were carried out. The Mössbauer spectra were recorded in the horizontal transmission geometry using a constant acceleration spectrometer operated in conjunction with a 512-channel analyzer in the time-scale mode. The source was at room temperature and consisted of 1.15 GBq of 57 Co diffused in rhodium foil (AEA, Braunschweig, Germany). The spectrometer was calibrated against a metallic ␣-iron foil at room temperature, yielding a standard line width of 0.24 mm/s. The Mössbauer cryostat was a helium bath cryostat (MD306, Oxford Instruments). A small field of 20 milliteslas perpendicular to the ␥-beam was applied to the tail of the bath cryostat using a permanent magnet. Isomer shift (␦), quadrupole splitting (⌬E Q ), and percentage of the total absorption area were obtained by least-squares fits of lorentzian lines to the experimental spectra.

Characterization of the cbsH Gene and Its Translation Product-
The sequence of the 2.1-kb SspI-SalI fragment ( Fig. 2A) begins at nucleotide ϩ3 of the fct-cbsCEBA transcript and ends 100 bp upstream from the SalI site. Two contiguous ORFs (ORF1 and ORF2) and the beginning of a third (ORF3) were identified ( Fig. 2A). ORF1 has two putative ATG initiation codons (ATG1 position 142 and ATG2 position 277) (Fig. 2B) and terminates with a TAG stop codon (position 1444). Potential ribosome-binding sites are located 6 and 8 bp upstream from the initiation codons, respectively. ORF2 starts with an ATG codon (position 1557), terminates with a TGA codon (position 1796), and has a ribosome-binding site 9 bp upstream from the start codon. A GTG initiation codon (position 1815) preceded by a Shine-Dalgarno sequence (5Ј-AGGA-3Ј, position 1804) suggests the beginning of ORF3. ORF1 is separated from ORF2 by 100 bp, and ORF2 is separated from ORF3 by 18  Sequence analysis predicted a potential P promoter, overlapping the PЈ promoter of the fct-cbsCEBA operon characterized previously (Fig. 2B). To determine the transcriptional start site of the cbsH gene, total RNA from iron-replete and -depleted cultures of E. chrysanthemi strain 3937 and E. coli strain JM101 harboring pCS1 was used as a template in extension reactions primed with a 32 P-labeled 17-mer oligonucleotide complementary to the sequence between positions 158 and 142. For both strains, the reactions yielded iron-regulated cDNAs that comigrated with an A residue (Fig. 2C). The occurrence of a transcriptional start at a T nucleotide (position 49) is consistent with the predicted P promoter. Two putative Fur-binding sites overlapping the Ϫ10/Ϫ35 sequences of the P promoter ( Fig. 2B) account for the observed iron regulation. Regulation by iron was confirmed by monitoring expression of the chromosomal cbsH-17::lacZ fusion constructed in L2 cells (Table I and below) grown in Tris medium with and without iron supplementation (Fig. 2C).
The two potential translational start codons ATG1 and ATG2 (Fig. 2B) have good matches with the Shine-Dalgarno sequence (5Ј-GGAGG-3Ј and 5Ј-GGACG-3Ј, respectively). To determine which of these codons is functional, we analyzed the translation products of two constructs (pCS3 and pLR1) placed under the control of the T7 10 promoter by SDS-PAGE. pCS3, which contains the 2.1-kb DraI-SalI fragment ( Fig. 2A), includes both ATG codons and the 5Ј-untranslated region. pLR1, in which the 158-bp SspI-EcoNI fragment present in pCS3 has been deleted (Fig. 2, A and B), lacks the ATG1 codon. For both constructs, the same polypeptide migrating in the 43,000-Da range and induced at 42°C only was identified (data not shown). Thus, under the conditions described, ATG2 (position 277) (Fig. 2B) is the functional translational start codon for the cbsH gene.
A cbsH Mutation Has No Effect on Iron Acquisition from Enterobactin in E. chrysanthemi-We investigated the protein encoded by the cbsH gene by isolating mutants (L2 cbsH-17 and L2 cbsH-19) ( Table I) using insertional mutagenesis. Insertion cbsH-19 was mapped to position 943 by sequencing and was analyzed further. As the MudI1734 prophage generates polar mutations, we first investigated whether the mutant was able to produce chrysobactin. It did not grow on EDDHA/L agar medium. It produced catechol compounds, but did not release a functional siderophore, as shown by the chrome azurol S assay and growth stimulation experiments (data not shown). The introduction of pCS2, which carries the cbsH gene and ORF2, did not enable the mutant cells to grow on EDDHA/L agar medium. We therefore concluded that the mutant did not synthesize chrysobactin because of the polar effect on the downstream gene that shares identity with the E. coli entF gene.
As ferric enterobactin esterase is an essential component of the enterobactin-mediated iron transport pathway in E. coli, we investigated whether the L2 cbsH-19 mutant could use ferric enterobactin as an iron source (Table II). Ferric enterobactin promoted the growth of this mutant as efficiently as for a cbsH-positive strain. This mutant may be able to utilize ferric enterobactin because it produced achromobactin or catechol. We therefore transduced the mutant with mutations acs-37 and cbsE-1. The transductant L2 cbsH-19 acs-37 cbsE-1 utilized ferric enterobactin as efficiently as did the simple mutant (Table II). This shows that a functional cbsH gene is not required in E. chrysanthemi for iron acquisition from enterobactin.
Lack of Functional Complementation of the E. coli fes Mutation by the cbsH Gene-These data led us to verify whether the cbsH gene could functionally complement an E. coli fes mutant. MM272-60 cells were transformed with plasmid pCS2 or pLR2. The transformants did not grow on EDDHA/L agar medium, and their growth was not stimulated by ferric enterobactin (Table II). A similar result was obtained with MM272-60 cells harboring pTF12, which contains the cbsH gene on a low-copynumber vector (Table II). We checked that the lack of complementation did not result from the production of nonphysiological levels of the CbsH protein by testing low-iron cultures of cells harboring pLR2 for the presence of enterobactin. All culture supernatants tested were strongly positive in the bioassay (data not shown).
A cbsH Mutation Affects Iron Acquisition from Chrysobactin in E. chrysanthemi-To determine whether the CbsH protein is a component of the chrysobactin-dependent iron transport pathway, we assessed the stimulation of growth of the L2 cbsH-19 acs-37 mutant by chrysobactin. After 24 h of incubation, the mutant had not grown; but after 72 h, a halo of growth became visible (Table II). The introduction of pCS2 into the mutant restored its growth in 24 h. Rescue was also observed after the introduction of pLR2, which carries the only cbsH gene (Table II). Thus, the mutant phenotype did not result from a polar effect of the mutation on downstream genes of the same operon. In contrast, the E. coli fes mutant (MM272-60), in which the ferric chrysobactin receptor fct gene is present on plasmid pLR3 (unlike pLR2), grew normally if supplied with ferric chrysobactin as an iron source (Table II). The halo of growth was similar with strain MM272-60 carrying pTF12, which contains the cbsH gene (Table II). Thus, the protein encoded by the cbsH gene is not required in E. coli cells if ferric chrysobactin is the iron source. One possible interpretation of these data is that a cbsH-negative mutant of E. chrysanthemi was affected in the transport of ferric chrysobactin.
We therefore determined the ability of the mutant to trans- port 59 Fe-labeled chrysobactin and compared it with that of the parental strain (Fig. 3). Strains were grown in Tris medium, and uptake experiments were conducted with cells harvested at A 600 nm ϭ 0.6. After 5 h, the growth of mutant L2 cbsH-19 acs-37 had slower considerably, indicating that iron was poorly assimilated (Fig. 3A). The ferric chrysobactin transport rate was higher in the mutant than in the parental strain (Fig. 3B). Thus, the mutation had no effect on ferric chrysobactin transport per se. The transport rate seems to depend on the intracellular metabolic state and presumably reflects the level of derepression by iron of the entire protein machinery involved in transport.
Accumulation of Ferric Chrysobactin in a Nonpolar cbsHnegative Mutant-We investigated the protein encoded by the cbsH gene by constructing a nonpolar mutant with the aphA-3 cassette (39). Mutant L2 cbsH::aphA-3 acs-37 gave rise to colonies with a red color that was not observed in polar mutants. The E. coli fes mutant was also red on L agar medium. The mutant did not grow on EDDHA/L agar medium. If supplied with ferric chrysobactin, there was a time lag before it started growing, as observed with polar mutants (Table II). This mutant grew very slowly in Tris medium, but transported 59 Felabeled chrysobactin very quickly, indicating that the bacterial cells were severely iron-depleted (Fig. 3B). These observations suggest that an iron-binding compound accumulated inside the cells. To determine whether this compound was the ferric chrysobactin complex, cell extracts of L2 cbsH::aphA-3 acs-37 were compared with those of a fur mutant (L37 acsA-1 fur) that also overexpresses chrysobactin biosynthesis and transport proteins in Tris medium supplemented with FeCl 3 . Bioassay showed that extracts from cbsH-negative cells promoted the growth of a chrysobactin-deficient strain very efficiently (Fig.  4A). This effect was not observed with extracts from fur-deficient cells. The ferric chrysobactin complex was quantitatively determined in culture supernatants and cell lysates (Fig. 4B).
The cbsH mutant accumulated ferric chrysobactin intracellularly, unlike the fur strain, for which most of the ferric complex was present in the culture supernatant.
The CbsH Protein Is a Peptidase Hydrolyzing Chrysobactin-The accumulation of the ferric chrysobactin complex in the cytosol of cbsH-negative cells indicates that this molecule was not degraded following its transport. As chrysobactin possesses a peptide bond, one possibility was that the cbsH gene encodes a peptidase. Indeed, the catalytic mechanism of certain esterases involving the formation of an acetyl-enzyme intermediate during the reaction is analogous to that of serine proteases (47). The alignment of ferric enterobactin esterase-like protein sequences from various bacterial genera present in data banks (Fig. 5) reveals the presence of a common signature, GXSXG-GDH found in the family of prolyl oligopeptidases (48). These residues are within ϳ130 residues of the C terminus, and the N-terminal parts of the molecule are more or less variable. Therefore, we investigated whether cbsH-positive cells had an enzyme enabling them to catalyze the hydrolysis of ferric chrysobactin that was lacking in mutant cells. Enzymatic activity was determined in cell extracts from low-iron cultures of the parental strain L2 cbsE-1 and the mutant L2 cbsH-19 (Table  III). We observed the disappearance of the ferric chrysobactin complex only in cbsH-positive cells. This enzymatic activity was thiol-dependent, like a number of other cytosolic peptidases. The addition of diisopropyl fluorophosphate, an inhibitor of serine proteases. totally blocked the reaction. These results show that the CbsH protein has a ferric chrysobactin peptidase activity. To also determine whether this enzyme has a ferric enterobactin esterase activity, we used ferric enterobactin as a substrate. Although cbsH-positive cells had a significant level of ferric enterobactin esterase activity compared with mutant cells, the specific enzymatic activity was 10 times lower than that found for the hydrolysis of ferric chrysobactin under the same conditions.
Iron Removal from Chrysobactin in Vivo Does Not Require a Functional cbsH Gene-To obtain basic information on the metabolic utilization of chrysobactin-bound iron, in situ Mössbauer spectroscopy of whole cells was performed at various states of chrysobactin uptake. In principle, in situ Mössbauer spectroscopy enables the simultaneous identification of all main iron metabolites at a qualitative as well as quantitative level without destruction of the cellular assembly (49,50). Moreover, time-dependent changes can be followed, the resolution of which is merely limited by the time required for sample preparation (50).
The second component observed spectroscopically corresponds to a ferric high-spin species. The 57 Fe content of the cells grew with increasing incubation time (increasing total absorption area), although it was slightly slower for the cbsHpositive strain. Whereas after 30 min of incubation the ferric iron species contributed only little to the Mössbauer absorption, it represented the major component after 2 h. This species exhibited Mössbauer parameters very similar to those of bacterioferritin found in E. coli (54,55). Comparison of the Mössbauer parameters obtained from a spectrum measured at 86 K (data not shown) with those derived from a spectrum at 4.3 K (Fig. 6) revealed a significant increase in ⌫ (from 0.505 to 0.814 mm/s) and a concomitant decrease in relative transmission. This considerable line broadening is typically found at temperatures close to superparamagnetic transitions (53,54). E. coli- type bacterioferritins display magnetic broadening below 4.3 K and eventually show magnetically split spectra at temperatures below 1 K (56). Fig. 7B displays the Mössbauer spectrum of cbsH-negative mutant cells measured at 1.8 K. Indeed, the ferric iron species is missing in this spectrum; instead, a magnetically broadened absorption is visible, as expected for a bacterioferritin-type protein. Therefore, we attribute the ferric iron species to a bacterioferritin-like compound. DISCUSSION In this study, we report the functional analysis of a new gene (cbsH) that belongs to the chrysobactin-dependent iron transport gene cluster of E. chrysanthemi strain 3937. The cbsH gene is the first gene of an operon involved in chrysobactin biosynthesis and transcribed from the iron-regulated divergent promoter fct-cbsH, which controls, in the opposite orientation, the transcription of the fct-cbsCEBA operon, identified earlier (23). The cbsH gene is 64% identical to the E. coli fes gene for the 871 nucleotides from positions 272 to 1143. It encodes a polypeptide with an apparent molecular mass of 43,000 Da, a size similar to that reported for the E. coli Fes protein (25). The CbsH protein is 45.6% identical to the Fes protein of E. coli, 46% identical to the Fes protein of Yersinia enterocolitica (57), and 42% identical to Salmonella enterica IroD (iroD gene, GenBank TM /EBI Data Bank accession number U97227), with the level of identity uniform over the entire amino acid sequence (according to the program BLAST) The presence in E. chrysanthemi of a homolog of the ferric enterobactin esterase of E. coli was expected. E. chrysanthemi has a ferric enterobactin transport system that supplies the cell with iron. As the hydrolysis of ferric enterobactin is essential in E. coli cells, we thought it likely that this molecule would have the same fate in E. chrysanthemi. We have shown that the CbsH protein is not required for the removal of iron from ferric enterobactin in E. chrysanthemi. These data indicate that cleavage of the ester bonds of ferric enterobactin is not required in E. chrysanthemi for iron reduction and release. This was not because of the presence of an additional fes-like gene on the E. chrysanthemi chromosome, as shown by DNA/DNA hybridization analysis (data not shown). In contrast, the Fes homolog from Y. enterocolitica appears to be absolutely required for ferric enterobactin utilization in this bacterium (57). In addition, the viuB gene from Vibrio cholerae, which is involved in vibriobactin processing, can complement the E. coli fes mutation (58). No functional complementation of the E. coli fes mutation was observed with the E. chrysanthemi cbsH gene.
These results show that iron release from enterobactin is not CbsH-dependent. Instead, a Fes/CbsH-independent mechanism has to be considered.
We should point out that the role of the E. coli ferric enterobactin esterase has been much debated. In particular, several experimental aspects have remained unexplained (10,11,59). For instance, this enzyme is required for the removal of iron from enterobactin analogs devoid of ester bonds (60,61). The redox potential of a ferric siderophore depends on the binding constant for iron and thus on the capacity of the molecule to be protonated at neutral pH (62). If the internal pH of E. chrysanthemi were slightly lower than that of E. coli, then iron would be easier to extract in E. chrysanthemi than in E. coli. Cohen et al. (59) have reported that enterobactin, like synthetic analogs such as TRENSAM, may adopt a Tris salicylate mode of binding if sequentially protonated, with iron release facilitated by a biological reductant.
On the basis of amino acid sequence comparisons, we found that the CbsH protein displays characteristics of the S9 prolyl oligopeptidase family (48), viz. the conservation of amino acids around the catalytic triad Ser, Asp, and His (Fig. 5). The S9 family contains serine peptidases with a varied range of restricted specificities, including oligopeptidase B from eubacteria, which cleaves arginyl and lysyl bonds. In agreement with sequence predictions, we showed that the CbsH protein is an enzyme able to degrade ferric chrysobactin in the cytosol. This hydrolytic activity is thiol-dependent and inhibited by fluorophosphates such as diisopropyl fluorophosphate. Given the chrysobactin structure (Fig. 1), it is very likely that this enzyme cleaves the lysyl bond, thus forming 2,3-dihydroxybenzoyllysine and serine.
To further understand the role of this enzyme, we analyzed the cellular distribution and redox state of iron following the transport of ferric chrysobactin into the cytosol using Mössbauer spectroscopy. After 30 min of incubation with 57 Fe-labeled chrysobactin, the Mössbauer spectra of the cbsH-positive strain and of a cbsH-negative mutant showed mainly ferrous iron. The total lack of ferric chrysobactin and the initially observed high-ferrous iron contribution in the cell spectra of Erwinia clearly demonstrate that ferric chrysobactin transport is followed by a very rapid intracellular enzymatic iron reduction. Because ferric chrysobactin is transported across the cell membranes via a highly specific receptor-mediated pathway (18), the reduction obviously occurs at the level of the cytoplas-  mic membrane or in the cytosol. The affinity of catecholate siderophores for ferrous iron is very low. Even water is a better chelator of ferrous high-spin iron than these siderophores. Therefore, the presence of ferrous iron provides evidence for a rapid reductive release of the metal from its carrier, preventing an observable intracellular concentration of 57 Fe-labeled chrysobactin. Although a ferrous hexaquo complex is stable in a strict reductive (and anaerobic) environment, it is very likely that the reduced metal is complexed by a specific intracellular chelator to prevent Haber-Weiss-Fenton chemistry (50). Previously, we found that ferrous iron constitutes one of the major cellular iron species in many microorganisms under conditions of siderophore-controlled growth (52, 53). The corresponding compound has been isolated from E. coli and from Pantoea agglomerans 2 and partially characterized as an oligomeric sugar phosphate (52). It was termed ferrochelatin (53). Based on the previous studies, we attribute the detected ferrous iron to ferrochelatin. Whereas ferrochelatin-bound iron keeps its intracellular concentration at a certain level (ϳ0.3% effect), the second component of the Mössbauer spectra increases its contribution by time and represents the major component after 2 h of incubation. Based on the temperature-dependent Mössbauer spectra and their parameters, the conclusion must be drawn that this component represents a bacterioferritin-like iron storage compound. Thus, ferrous iron released from chrysobactin is immediately transferred into the iron storage form, where it is oxidized again at the ferroxidase site (71,72). In summary, the Mössbauer spectroscopic analysis shows significant differences neither in the chrysobactin-mediated iron uptake between the parental strain and its mutant nor in the metabolic distribution pattern. Based on the results of this investigation, it is safe to state that metabolic utilization of both enterobactin-and chrysobactin-bound iron is not cbsH-dependent (see scheme shown in Fig. 8).
As described above, there is good evidence for hydrolytic cleavage of chrysobactin by CbsH. In addition, long-term growth inhibition has been observed in cbsH-negative mutants. This ligand hydrolysis occurs obviously after iron removal, and lack of hydrolysis in the mutants results in growth inhibition. This unexpected finding might be linked either to utilization of the aromatic systems of chrysobactin for anabolic reactions or to a role of cbsH in intracellular iron homeostasis. Within the rationale of bacterial iron metabolism, we favor the latter line of thought. The free chrysobactin ligand is thermodynamically capable of extracting ferric iron from all intracellular ferric iron sources exhibiting a lower complex formation constant than that of ferric chrysobactin. In addition, recent studies on the uptake of iron(III) by chrysobactin have shown that the carboxyl group of the serine residue in chrysobactin strongly influences the kinetics of formation of the ferric complex. 3 To prevent iron removal from metabolically active enzymes or from any accessible intracellular iron pool, either the ligand has to be re-excreted (which is known for some bacterial siderophore uptake systems), or it must be degraded or modified. At this point, it is important to note that the nonpolar cbsHnegative mutant behaves as if it was severely iron-depleted, although it contains high levels of ferric chrysobactin. This finding fits well with our hypothesis because an intracellular post-transport recomplexation of iron by the non-degraded ligand seems to occur. In summary, taking all pieces of circumstantial evidence together, we suggest that hydrolytic degradation of chrysobactin by CbsH is aimed at keeping the intracellular iron distribution at a well regulated level (iron homeostasis) in E. chrysanthemi strain 3937.