In vitro characterization of salmochelin and enterobactin trilactone hydrolases IroD, IroE, and Fes

Article abstract of DOI:10.1021/ja0522027

The iroA locus encodes five genes (iroB, iroC, iroD, iroE, iroN) that are found in pathogenic Salmonella and Escherichia coli strains. We recently reported that IroB is an enterobactin (Ent) C-glucosyltransferase, converting the siderophore into mono-, di-, and triglucosyl enterobactins (MGE, DGE, and TGE, respectively). Here, we report the characterization of IroD and IroE as esterases for the apo and Fe³⁺-bound forms of Ent, MGE, DGE, and TGE, and we compare their activities with those of Fes, the previously characterized enterobactin esterase. IroD hydrolyzes both apo and Fe³⁺-bound siderophores distributively to generate DHB-Ser and/or Glc-DHB-Ser, with higher catalytic efficiencies (k_cat/K_m) on Fe³⁺-bound forms, suggesting that IroD is the ferric MGE/DGE esterase responsible for cytoplasmic iron release. Similarly, Fes hydrolyzes ferric Ent more efficiently than apo Ent, confirming Fes is the ferric Ent esterase responsible for Fe ³⁺ release from ferric Ent. Although each enzyme exhibits lower k_cat's processing ferric siderophores, dramatic decreases in K _m's for ferric siderophores result in increased catalytic efficiencies. The inability of Fes to efficiently hydrolyze ferric MGE, ferric DGE, or ferric TGE explains the requirement for IroD in the iroA cluster. IroE, in contrast, prefers apo siderophores as substrates and tends to hydrolyze the trilactone just once to produce linearized trimers. These data and the periplasmic location of IroE suggest that it hydrolyzes apo enterobactins while they are being exported. IroD hydrolyzes apo MGE (and DGE) regioselectively to give a single linear trimer product and a single linear dimer product as determined by NMR.

Full text of DOI:10.1021/ja0522027

Published on Web 07/14/2005
In Vitro Characterization of Salmochelin and Enterobactin
Trilactone Hydrolases IroD, IroE, and Fes
Hening Lin,† Michael A. Fischbach,†,‡ David R. Liu,‡ and Christopher T. Walsh*,†
Contribution from the Department of Biological Chemistry and Molecular Pharmacology,
HarVard Medical School, Boston, Massachusetts 02115, and Department of Chemistry and
Chemical Biology, HarVard UniVersity, Cambridge, Massachusetts 02138
Received April 6, 2005; E-mail: christopher_walsh@hms.harvard.edu
Abstract: The iroA locus encodes five genes (iroB, iroC, iroD, iroE, iroN) that are found in pathogenic
Salmonella and Escherichia coli strains. We recently reported that IroB is an enterobactin (Ent)
C-glucosyltransferase, converting the siderophore into mono-, di-, and triglucosyl enterobactins (MGE, DGE,
and TGE, respectively). Here, we report the characterization of IroD and IroE as esterases for the apo and
Fe3+-bound forms of Ent, MGE, DGE, and TGE, and we compare their activities with those of Fes, the
previously characterized enterobactin esterase. IroD hydrolyzes both apo and Fe3+-bound siderophores
distributively to generate DHB-Ser and/or Glc-DHB-Ser, with higher catalytic efficiencies (kcat/Km) on
Fe3+-bound forms, suggesting that IroD is the ferric MGE/DGE esterase responsible for cytoplasmic iron
release. Similarly, Fes hydrolyzes ferric Ent more efficiently than apo Ent, confirming Fes is the ferric Ent
esterase responsible for Fe3+ release from ferric Ent. Although each enzyme exhibits lower kcat’s processing
ferric siderophores, dramatic decreases in Km’s for ferric siderophores result in increased catalytic efficiencies.
The inability of Fes to efficiently hydrolyze ferric MGE, ferric DGE, or ferric TGE explains the requirement
for IroD in the iroA cluster. IroE, in contrast, prefers apo siderophores as substrates and tends to hydrolyze
the trilactone just once to produce linearized trimers. These data and the periplasmic location of IroE suggest
that it hydrolyzes apo enterobactins while they are being exported. IroD hydrolyzes apo MGE (and DGE)
regioselectively to give a single linear trimer product and a single linear dimer product as determined by
NMR.
of EntB, EntD, EntE, and EntF5-8 and follows nonribosomal
peptide synthetase logic.2,9 Following synthesis in the bacterial
Introduction
Iron is an essential cofactor for most bacterial species.1 The
iron supply is normally limited in the environment because of
the low solubility of Fe3+ at neutral or basic pH. In mammalian
cytoplasm, Ent is exported from the cell in a process that
involves YbdA (EntS).10 Ent molecules that have successfully
bound to extracellular Fe3+ are recognized by the outer
membrane receptor FepA and are imported through an active
transport process.11,12 Iron release from ferric Ent (Fe-Ent)
requires the enzyme Fes, which has been shown to catalyze the
hydrolysis of Fe-Ent despite conflicting reports on whether Fes
prefers apo Ent or Fe-Ent as substrate.13-16 Since the DHB-
Ser monomer has a much lower affinity for Fe3+, the hydrolysis
hosts, the free Fe3+ concentration is further lowered by Fe3+
-
binding proteins to an estimated 10-24 M.1 To survive in an
iron-deficient environment, bacteria have evolved the ability
to biosynthesize dedicated Fe3+-binding small molecules, or
siderophores, to scavenge iron from the environment.2 One of
the best studied siderophores is enterobactin (Ent) (Figure 1), a
2,3-dihydroxybenzoylserine macrotrilactone produced by enteric
bacteria, such as E. coli and Salmonella enterica.1 Ent is the
(5) Gehring, A. M.; Bradley, K. A.; Walsh, C. T. Biochemistry 1997, 36, 8495-
8503.
strongest Fe3+ ligand known with an estimated KD of 10-49
M
(10-35 M at physiological pH).3,4 The genes involved in the
biosynthesis, transport, and processing of Ent are clustered into
a 20 kB region of the bacterial chromosome controlled by the
iron-dependent repressor Fur.2 The biosynthesis of Ent from
2,3-dihydroxybenzoic acid (DHB) and Ser requires the action
(6) Gehring, A. M.; Mori, I.; Walsh, C. T. Biochemistry 1998, 37, 2648-
2659.
(7) Shaw-Reid, C. A.; Kelleher, N. L.; Losey, H. C.; Gehring, A. M.; Berg,
C.; Walsh, C. T. Chem. Biol. 1999, 6, 385-400.
(8) Ehmann, D. E.; Shaw-Reid, C. A.; Losey, H. C.; Walsh, C. T. Proc. Natl.
Acad. Sci. U.S.A. 2000, 97, 2509-2514.
(9) Finking, R.; Marahiel, M. A. Annu. ReV. Microbiol. 2004, 58, 453-488.
(10) Furrer, J. L.; Sanders, D. N.; Hook-Barnard, I. G.; McIntosh, M. A. Mol.
Microbiol. 2002, 44, 1225-1234.
† Harvard Medical School.
(11) Faraldo-Go´mez, J. D.; Sansom, M. S. P. Nat. ReV. Mol. Cell. Biol. 2003,
4, 105-116.
‡ Harvard University.
(1) Raymond, K. N.; Dertz, E. A.; Kim, S. S. Proc. Natl. Acad. Sci. U.S.A.
2003, 100, 3584-3588.
(12) Braun, V.; Braun, M. Curr. Opin. Microbiol. 2002, 5, 194-201.
(13) Bryce, G. F.; Brot, N. Biochemistry 1972, 11, 1708-1715.
(14) Langman, L.; Young, I. G.; Frost, G. E.; Rosenberg, H.; Gibson, F. J.
Bacteriol. 1972, 112, 1142-1149.
(2) Crosa, J. H.; Walsh, C. T. Microbiol. Mol. Biol. ReV. 2002, 66, 223-249.
(3) Harris, W. R.; Carrano, C. J.; Cooper, S. R.; Sofen, S. R.; Avdeef, A. E.;
McArdle, J. V.; Raymond, K. N. J. Am. Chem. Soc. 1979, 101, 6097-
6104.
(15) Greenwood, K. T.; Luke, R. K. J. Biochim. Biophys. Acta 1978, 525, 209-
218.
(4) Loomis, L. D.; Raymond, K. N. Inorg. Chem. 1991, 30, 906.
(16) Brickman, T. J.; McIntosh, M. A. J. Biol. Chem. 1992, 267, 12350-12355.
9
10.1021/ja0522027 CCC: $30.25 © 2005 American Chemical Society
J. AM. CHEM. SOC. 2005, 127, 11075-11084
11075

A R T I C L E S
Lin et al.
Figure 1. Structures of enterobactin, glucosylated enterobactins, and some of their degradation products. Next to each structure is a schematic representation
of the compound, where a blue triangle represents DHB-Ser, and a pink circle represents glucose. Similar schematic representations are used in subsequent
figures.
event facilitates the transfer of Fe3+ to iron-dependent enzymes,
such as cytochromes.
iroA cluster, how they differ in catalytic efficiency and
specificity, or if Fes is able to release Fe3+ from salmochelins.
These questions are the subject of the work reported herein.
We recently purified and characterized the proposed gluco-
syltransferase (Gtf) IroB24 and established that it can catalyze
the successive mono-, di-, and tri-C-glucosylation of Ent to
generate monoglucosyl enterobactin (MGE), diglucosyl entero-
bactin (DGE), and triglucosyl enterobactin (TGE) (Figure 1).
DGE is identical to salmochelin S4, while S1, S2, and SX
correspond to MGE and DGE hydrolysis products that may arise
from the action of IroD or IroE. The availability of pure, active
IroB24 enables the preparation of sufficient MGE, DGE, and
TGE to assay the enzymatic properties of IroD and IroE for
both catalytic efficiency and regioselectivity of cleavage. We
also compare the kinetic parameters of these two enzymes with
those of Fes.
Recently, C-glucosylated Ent analogues, termed salmochelins,
have been isolated from S. enterica and structurally character-
ized.17,18 Salmochelin production could be a mechanism by
which pathogenic bacteria subvert a mammalian host’s efforts
to obstruct bacteria iron acquisition. One observation that
supports this is that salmochelins are better siderophores than
Ent in the presence of serum albumin.17 Another observation is
that ferric Ent is sequestered by the protein siderocalin in a
mammalian host,19,20 and glucosylation of Ent presumably
decreases the affinity for siderocalin. The conversion of Ent to
salmochelins, such as S1, S2, and S4 (Figure 1), requires the
iroA gene cluster, which includes five genes, iroB, iroC, iroD,
iroE, and iroN.21 The IroN protein has homology to FepA and
was shown to be an outer membrane receptor for uptake of Fe3+
-
bound salmochelins.18,21 Similar to FepA and Cir, IroN is also
able to recognize several other siderophores.22 IroC is thought
to be an inner membrane transporter, functioning in the export
of apo siderophores.23 IroB has been shown to catalyze the
C-glucosylation of Ent.17,23 The two remaining proteins, IroD
and IroE, are homologous to Fes. IroD has been predicted to
be cytoplasmic, while IroE is thought to be periplasmic.18
Bacterial strains that harbor an iroA cluster lacking IroD or IroE
produce less of the hydrolyzed salmochelins (S1, S2, and SX),
suggesting that both enzymes are probably salmochelin es-
terases.17,23 It is not clear why two esterases are present in the
Materials and Methods
Preparation of Ent and Glucosylated Ents. Ent was chemically
synthesized as reported previously.24,25 The enzymatic synthesis of TGE
was described previously.24 For the synthesis of MGE and DGE, a
slightly modified method was used. To 150 mL of Tris buffer (75 mM,
pH 8.0) containing MgCl2 (5 mM), tris(2-carboxyethyl)phosphine (2.5
mM), UDP-Glc (1.5 mM), and Ent (1 mM) was added IroB to a final
concentration of 56 mg/L. The reaction was gently agitated at 23 °C
for 2 h, then quenched with 50 mL of 2.5 N HCl in methanol. After
filtering through a 0.45 µm membrane, the MGE and DGE products
were purified by reverse-phase HPLC using a gradient of 0-40% CH3-
CN in 0.1% TFA/water over 40 min. Fractions containing MGE and
DGE were lyophilized to give the products as white solids (∼50 mg
each). All compounds in apo forms were dissolve in DMSO to be used
in enzyme assays.
(17) Bister, B.; Bischoff, D.; Nicholson, G. J.; Valdebenito, M.; Schneider, K.;
Winkelmann, G.; Hantke, K.; Sussmuth, R. D. Biometals 2004, 17, 471-
481.
(18) Hantke, K.; Nicholson, G.; Rabsch, W.; Winkelmann, G. Proc. Natl. Acad.
Sci. U.S.A. 2003, 100, 3677-3682.
Preparation of Ferric Complexes of Ent, MGE, DGE, and TGE.
Since previous inconsistencies on Fes substrate specificities were
attributed to possible differences in the Fe-Ent samples, we tried two
methods to prepare the ferric complex. (A) Ent, MGE, DGE, or TGE
(19) Goetz, D. H.; Holmes, M. A.; Borregaard, N.; Bluhm, M. E.; Raymond,
K. N.; Strong, R. K. Mol. Cell. 2002, 10, 1033-1043.
(20) Flo, T. H.; Smith, K. D.; Sato, S.; Rodriguez, D. J.; Holmes, M. A.; Strong,
R. K.; Akira, S.; Aderem, A. Nature 2004, 432, 917-921.
(21) Baumler, A. J.; Norris, T. L.; Lasco, T.; Voight, W.; Reissbrodt, R.; Rabsch,
W.; Heffron, F. J. Bacteriol. 1998, 180, 1446-1453.
(22) Rabsch, W.; Methner, U.; Voigt, W.; Tschape, H.; Reissbrodt, R.; Williams,
P. H. Infect. Immun. 2003, 71, 6953-6961.
(24) Fischbach, M. A.; Lin, H.; Liu, D. R.; Walsh, C. T. Proc. Natl. Acad. Sci.
U.S.A. 2005, 102, 571-576.
(25) Ramirez, R. J. A.; Karamanukyan, L.; Ortiz, S.; Gutierrez, C. G.
Tetrahedron Lett. 1997, 38, 749-752.
(23) Baumler, A. J.; Tsolis, R. M.; Velden, A. W. M. v. d.; Stojiljkovic, I.;
Anic, S.; Heffron, F. Gene 1996, 183, 207-213.
9
11076 J. AM. CHEM. SOC. VOL. 127, NO. 31, 2005

Bacterial Iron Release from Enterobactin and Salmochelins
A R T I C L E S
Table 1. MS m/z Data for apo Ent, MGE, DGE, and TGE
(6.4 mM in DMSO) was mixed with 1.2 equiv of FeCl3 (6.4 mM in
water) for 2 min at 23 °C. To the colored solution was then added 75
mM HEPES buffer (pH 7.5) to make the final ferric complex
concentration 128 µM. This solution was then left at 23 °C for at least
1 h before being used in enzyme assays. (B) Ent, MGE, DGE, or TGE
(6.4 mM in DMSO, 200 µL) was mixed with 1.1 equiv of FeCl3 (6.4
mM in water, 220 µL) and 3.3 equiv of 1 M aqueous K2CO3 (4.2 µL)
for 1 min at 23 °C. The colored solution was then loaded onto an anion
exchange column packed with 10 mL of DEAE Sephacel resin
(Amersham). After washing with 3 column volumes of water, 3 column
volumes of 0.1 M NaCl, the ferric complex was eluted with 2-4 column
volumes of 1 M NaCl or a gradient of NaCl. To the pink solution
containing the ferric complex was added 4 volumes of ethanol. The
salt precipitate was removed by centrifugation, and the supernatant was
concentrated by rotary evaporation. The residue was then resuspended
in 80% ethanol (10 mL), and insoluble salt again was removed by
centrifugation, and the supernatant was concentrated by rotary evapora-
tion. The final residue was resuspended in 80% ethanol (0.5 mL), and
any salt precipitation was removed by centrifugation. The ferric
complex, by dissociation with HCl and HPLC analysis monitored at
316 and 220 nm, contained more than 95% intact macrolactone scaffold
and only very little side product. The concentration of the ferric complex
solution was determined by comparing the area of absorption in the
HPLC trace with that of a known concentration apo Ent solution. The
solution was then diluted with 75 mM HEPES pH 7.5 buffer to make
the final concentration 128 µM. The ferric siderophore solution thus
obtained was checked with UV-vis spectroscopy and compared with
the spectrum of the apo compound. The apo form has an absorption at
320 nm, while the ferric form has absorptions at 340 and 496 nm,
consistent with literature for Ent and Fe-Ent.3,26 We carried out enzyme
assays with ferric complexes prepared by both methods and found they
behaved essentially the same. The kinetic data presented in the result
section were obtained with ferric complexes prepared by method B.
Cloning, Expression, and Purification of IroD. The iroD gene
was amplified from E. coli CFT073 genomic DNA using the forward
primer 5′-ggaattccatatgctgaacatgcaacaacatc-3′ and the reverse primers
5′-gatcgaattctcaaccctgtagtaaaccaatcccg-3′ (pET-28b) and 5′-gatcgaat-
tcggaccctgtagtaaaccaatcccgtc-3′ (pET-22b). The forward primer intro-
duced an NdeI restriction site, and the reverse primers introduced EcoRI
restriction sites (underlined above). PCR reactions were performed with
Herculase DNA polymerase (Stratagene). The amplified gene sequences
were digested with NdeI and EcoRI (New England Biolabs), then ligated
into the expression vectors pET-28b and pET-22b, and transformed
into E. coli TOP10 cells (Stratagene). The identities of the resulting
pET-28b-IroD (N-His) and pET-22b-IroD (C-His) constructs were
confirmed by DNA sequencing. Expression constructs were transformed
into E. coli BL21(DE3) cells, grown to saturation in LB medium
supplemented with kanamycin (50 µg/mL) or ampicillin (100 µg/mL)
at 37 °C, and diluted 1:100 into LB medium supplemented with
kanamycin (30 µg/mL) or ampicillin (50 µg/mL). The expression of
N- and C-terminal His6 fusion proteins was induced at OD600 0.5-0.6
with 400 µM IPTG, and overexpression was allowed to proceed at 15
°C for 20 h. Cells from 2 L cultures were pelleted by centrifugation
(10 min at 6100g), resuspended in 15 mL of buffer A (20 mM Tris pH
8.0, 500 mM NaCl, 10 mM MgCl2, 5 mM imidazole) and lysed by
two passages through a cell disruptor (Avestin EmulsiFlex-C5) at
5000-15000 psi. Cell debris was removed by ultracentrifugation (30
min at 126000g), and the supernatant was incubated with 1 mL of Ni-
NTA resin (Qiagen) at 4 °C for 2 h. After discarding the unbound
fraction, the resin was resuspended in 3 mL of buffer A, loaded onto
a column, and washed with 10 mL of buffer A. IroD was eluted from
the column with a stepwise imidazole gradient (from 10 to 200 mM).
After SDS-PAGE analysis, fractions containing pure IroD were pooled
and dialyzed twice against 2 L of buffer B (25 mM Tris pH 8.0, 50
mM NaCl, 1 mM DTT, 10% (v/v) glycerol). The protein was flash-
Hydrolysis Products by IroD, IroE, and Fesa
Ent
MGE
DGE
TGE
macrolactone
linear trimer
linear dimer
monomer
m/z calc. (M+) 669.1 831.2
m/z obs. 669.6 831.6
993.3 1155.3
993.9 1155.4
m/z calc. (M+) 687.1 849.2 1011.3 1173.3
m/z obs.
m/z calc. (M+) 464.1 626.2
m/z obs. 464.9 626.7
m/z calc. (M+) 241.1 241.1
m/z obs.b
240.1 240.1
monomer (with Glc) m/z calc. (M+) NA
m/z obs. NA
687.4 849.9 1011.5 1173.2
626.2
626.8
241.1
240.1
403.1
404.1
788.2
788.8
NA
NA
403.1
403.8
403.1
404.1
a Product distribution for different enzymes/substrates is shown in Figure
3. b Detected with negative ion mode. All others were detected with positive
ion mode.
frozen in liquid N2 and stored at -80 °C. The concentrations of purified
IroD N-His6 and C-His6 were determined spectrophotometrically at 280
nm using calculated extinction coefficients of 72 030 M-1 cm-1 for
both proteins.
Cloning, Expression, and Purification of IroE N-30. IroE N-30
truncation was PCR amplified from E. coli CFT073 genomic DNA
using the forward primer 5′-ggaattccatatgtatgcgaagccggatatgcg-3′ and
reverse primer 5′-gatcgaattcgggtggcttaactcatgacaacctgc-3′, digested, and
ligated into pET-22b. The expression and purification followed the same
procedure as described for IroD. The concentrations of purified IroE
N-30 C-His6 were determined spectrophotometrically at 280 nm using
calculated extinction coefficient of 45 041 M-1 cm-1
.
Cloning, Expression, and Purification of Fes N-His. Fes was PCR
amplified from E. coli CFT073 genomic DNA using the forward primer
5′-ggaattccatatgtttgaggtcactttctggtg-3′ and reverse primer 5′-gatcgaat-
tctcaactcctgtcatggaaaagtg-3′, digested, and ligated into pET-28b. The
expression in BL21(DE3) cells was carried out at 15 °C for 3 days
with 4 L of culture and no IPTG induction. Unless otherwise noted,
the same procedure as described for IroD was followed. The fractions
containing impure Fes from the Ni-NTA purification were pooled and
further purified by FPLC with a Superdex 200 (Amersham) gel filtration
column and a MonoQ 10/100 GL (Amersham) anion exchange column
at 4 °C. After SDS-PAGE analysis, the fraction containing the
relatively pure Fes protein was concentrated using Amicon ultracen-
trifugation filter devices (Millipore), dialyzed, flash-frozen with liquid
N2, and stored at -80 °C. The concentration of purified Fes N-His6
was determined spectrophotometrically at 280 nm using calculated
extinction coefficient of 124 704 M-1 cm-1
.
Initial Enzyme Activity Assay with apo Ent, MGE, DGE, and
TGE. For the initial activity assay with apo siderophores, all reactions
were carried out in 75 mM HEPES buffer pH 7.5 with 32 µM substrates.
IroD, IroE, and Fes frozen stocks were thawed on ice, diluted to 1 µM
with cold buffer B, and then added to the reaction mixture to a final
concentration of 20 nM. The reactions were quenched with 0.5 volumes
of 2.5 N HCl in methanol and analyzed by LCMS with a gradient of
0-35% CH3CN (with 0.1% formic acid) in water (with 0.1% formic
acid) over 8 min. The MS m/z data for different substrates and
hydrolysis product are listed in Table 1.
Initial Enzyme Activity Assay with Ferric Ent, MGE, DGE, and
TGE. For the initial activity assay with ferric siderophores, reaction
conditions were the same as those for apo siderophores, except that
the siderophores were mixed with 1.2 equiv of FeCl3 in the reaction
buffer for 10 min before the addition of enzymes.
Enzyme Kinetics with apo Ent, MGE, DGE, TGE. All reactions
were carried out in duplicate in 200 µL of 75 mM HEPES buffer pH
7.5 at 23 °C, with substrate concentrations ranging from 2 to 128 or
256 µM. Reactions were quenched with 100 µL of 2.5 N HCl in
methanol and then frozen on dry ice. For HPLC analysis, the samples
were thawed just before injection. Product quantification was based
(26) Corey, E. J.; Bhattacharyya, S. Tetrahedron Lett. 1977, 45, 3919-3922.
9
J. AM. CHEM. SOC. VOL. 127, NO. 31, 2005 11077

A R T I C L E S
Lin et al.
on the area of absorption monitored at 316 nm, assuming hydrolysis
of the ester bond does not affect the absorption. Negative control
reactions without enzyme were also carried out so that the background
hydrolysis could be subtracted from the data. For IroE, the enzyme
concentration used was 20 nM, and reactions were quenched after 40
s. For IroD, the enzyme concentration used was 5 nM, and reactions
were also quenched after 40 s. For Fes, 27 nM enzyme was used for
Ent and MGE hydrolysis, and reactions were quenched after 40 s; 36
nM enzyme was used for DGE hydrolysis, and reactions were quenched
after 5 min.
Enzyme Kinetics with Ferric Ent, MGE, DGE, TGE. Unless
otherwise noted, reaction conditions were identical to those involving
apo Ent, MGE, DGE, and TGE. The ferric siderophores were prepared
using method B. The substrate concentrations used in the assays ranged
from 0.125 to 16 µM. For IroE, the enzyme concentration used was
320 nM, and reactions were quenched after 2 min for Fe-Ent, Fe-
MGE, and Fe-DGE, and 1 min for Fe-TGE. For IroD, the enzyme
concentration used was 5 nM, and reactions were quenched after 40 s.
For Fes, consistent with a previous report,16 we found DTT and MgCl2
increased the hydrolysis rate severalfold. Therefore, all reactions with
Fes were carried out with 2 mM DTT and 2 mM MgCl2. The Fes
concentration used was 14 nM for Fe-Ent hydrolysis, and reactions
were quenched after 40 s; 36 nM was used for Fe-MGE hydrolysis,
and reactions were quenched after 3 min.
Preparation of MGE and DGE Hydrolysis Products for NMR
Characterization. To 150 mL of HEPES buffer (75 mM, pH 7.5)
containing MGE or DGE (250 µM) was added IroD C-His6 (2 µM, 6
mL) to a final concentration of 80 nM. The reaction was gently agitated
at 23 °C for 2 min (MGE) or 1.5 min (DGE) and then quickly quenched
with 50 mL of 2.5 N HCl in methanol. The hydrolysis products were
purified by HPLC using a gradient of 0-40% CH3CN in 0.1% TFA/
water over 40 min. The major product obtained this way was the linear
trimer. By extending the reaction time to 4 min, the major product
obtained was the linear dimer. The linear trimers and dimers from MGE
and DGE hydrolysis were dissolved in methanol-d4, and the NMR
experiments (1H, 13C, COSY, HSQC, HMBC) were done using a Varian
500 MHz NMR spectrometer in the Chemistry and Chemical Biology
Department of Harvard University. The spectra can be found in the
Supporting Information.
Figure 2. SDS-PAGE analysis of purified IroD (with either N- or
C-terminal His6 tag), IroE (N-terminal 30 or 40 aa deletion, with C-terminal
His6 tag), and Fes (with N-terminal His6 tag).
The expression and purification of Fes (374 aa, 43 kDa),
previously characterized by Brickman and MacIntosh,16 proved
to be challenging. Fes was not soluble under conditions used
for IroD overexpression and purification. Unlike IroE, N-
terminal truncation did not remedy this insolubility. Sufficient
soluble Fes was obtained in roughly 50% purity by limiting
expression of Fes to the background level resulting from the
absence of IPTG in the growth media. During anion exchange
and gel filtration purification steps, the Fes protein copurified
with two other proteins, suggesting the possibility of a tight
physical association with these proteins. One chromatographic
fraction contained Fes in ∼90% purity; assays with Fes were
conducted using this fraction (Figure 2).
Initial Enzyme Activity Assay of IroD, IroE, and Fes with
apo Substrates. Initial studies were performed with the apo
siderophores. Ent was chemically synthesized as previously
reported,25 while MGE, DGE, or TGE was synthesized enzy-
matically with IroB from Ent and UDP-Glc.24 Although TGE
was not isolated from iroA-harboring bacterial strains and may
not be physiologically relevant, we included TGE in our assays.
Activity assays of IroE were carried out at pH 7.5 with 20
nM IroE and 32 µM Ent, MGE, DGE, or TGE. Reactions were
monitored by LCMS. The time courses of the IroE-catalyzed
hydrolysis reactions shown in Figure 3 indicate that (i) Ent,
MGE, DGE, and TGE are all substrates of IroE, but Ent and
MGE seem to be hydrolyzed faster than DGE and TGE. This
trend can also be seen from the catalytic efficiency values shown
in Table 2. (ii) The initial linear trimer products of trilactone
hydrolysis are not efficient substrates for further hydrolysis,
enabling the reaction to stop at the linear trimer stage. (iii) IroE
acts with little regioselectivity relative to the placement of the
C-glucosyl substituent on the trilatone since at least two closely
eluting product peaks (separable upon analytical HPLC analysis)
were observed with the same mass for MGE and DGE
hydrolysis. Due to the difficulty in separating these products,
we did not further pursue their structural characterization, unlike
the IroD case described below.
Results
Overexpression and Purification of IroD, IroE, and Fes.
The iroD, iroE, and fes genes were cloned from E. coli CFT073
into E. coli protein expression vectors as N-terminal or
C-terminal His6 fusions. IroD (410 aa, 45 kDa) was over-
expressed from E. coli BL21/DE3 cells and purified by Ni-
NTA affinity purification to >95% purity (Figure 2).
In contrast, IroE (318 aa, 35 kDa), the predicted periplasmic
protein, was insoluble under similar overexpression conditions.
Hydropathy analysis (http://www.predictprotein.org) predicted
that residues 20-30 form a transmembrane region. We hypoth-
esized that deletion of the transmembrane region might increase
the solubility of the resulting truncated IroE protein. Therefore,
genes encoding IroE variants lacking the first 10, 20, 30, or 40
residues were constructed, and the corresponding proteins were
overexpressed. Deletion of the first 10 or 20 amino acids still
yielded insoluble proteins, but deletion of the first 30 or 40
amino acids (N-30, N-40) generated soluble proteins that could
be purified to >95% purity (Figure 2). Initial enzyme activity
assays with Ent and MGE showed that both the N-30 and N-40
truncations of IroE were active. Subsequent kinetic studies were
carried out with the N-30 truncation, containing residues 31-
318 of IroE.
9
11078 J. AM. CHEM. SOC. VOL. 127, NO. 31, 2005

Bacterial Iron Release from Enterobactin and Salmochelins
A R T I C L E S
Figure 3. Reaction time courses of IroE-, IroD-, and Fes-catalyzed hydrolysis of apo Ent, MGE, DGE, and TGE. Reaction aliquots were quenched at
different time points and analyzed by LCMS. The assignment of the hydrolysis products is based on the MS data, and the schematic representations of the
hydrolysis products are shown. For the linear trimer and dimer products from MGE and DGE hydrolysis, the schematic representation of only one possible
regioisomer is shown.
The activity assays of IroD were carried out in a similar
manner. The assay results shown in Figure 3 are consistent with
three conclusions: (i) apo Ent, MGE, DGE, and TGE are all
substrates for hydrolysis by IroD; (ii) the initial linear trimer
products are, in turn, very good substrates for further hydrolytic
cleavage by IroD, leading to complete degradation of the
trilactone to DHB-Ser and/or Glc-DHB-Ser monomers; (iii)
only one isomer of the linear trimer product and one isomer of
9
J. AM. CHEM. SOC. VOL. 127, NO. 31, 2005 11079

A R T I C L E S
Lin et al.
Table 2. Kinetics Data of Fes, IroD, and IroE
Ent
Fe
−Ent
MGE
Fe
−MGE
DGE
Fe
−DGE
TGE
Fe−TGE
Fes
kcat (min-1
m (µM)
kcat/Km (min-1 µM-1
)
>520a
>130a
4.0
1060 ( 60
40 ( 5
26
375 ( 40
16 ( 4
23
9 ( 1
<0.1b
>90
74 ( 4
0.12 ( 0.04
617
3.0 ( 0.1
3.4 ( 0.2
0.9
850 ( 65
100 (15
8.5
3720 ( 360
60 ( 15
62
430 ( 10
29 ( 2
14
1.4 ( 0.1
0.29 ( 0.05
4.8
46 ( 2
0.08 ( 0.02
575
3.2 ( 0.3
4.8 ( 0.7
0.7
>78a
NA
NA
NA
32 ( 2
0.12 ( 0.02
267
2.5 ( 0.3
4.6 ( 1.0
0.5
NA
NA
NA
4460 ( 470
160 ( 30
28
450 ( 30
155 ( 30
3
NA
NA
NA
K
>130a
0.6
)
)
IroD
IroEc
k
cat (min-1
m (µM)
)
6500 ( 540
120 ( 20
54
320 ( 20
39 ( 6
8
84 ( 2
0.43 ( 0.04
195
K
kcat/Km (min-1 µM-1
kcat (min-1
Km (µM)
)
>8a
>16a
kcat/Km (min-1 µM-1
)
0.5
a Cannot be determined because no saturation was observed up to the highest concentrations tested. b Cannot be determined because saturation was
reached even at the lowest concentration (0.125 µM) used. c At higher concentrations of Fe-Ent, Fe-MGE, and Fe-TGE, IroE was substrate-inhibited.
Figure 4. IroE, IroD, and Fes catalyze the hydrolysis of Fe3+-bound forms of Ent, MGE, DGE, and TGE more slowly than the apo forms under the same
conditions used. IroE- and IroD-catalyzed hydrolysis of MGE versus Fe-MGE and Fes-catalyzed hydrolysis of Ent versus Fe-Ent are shown for comparison.
the linear dimer product from MGE or DGE hydrolysis were
observed, suggesting that IroD hydrolyzes MGE and DGE
regioselectively. Interestingly, the linear dimer products from
MGE and DGE hydrolysis both have a molecular weight of
627 Da, indicating the presence of a single glucose group in
both dimers.
Under similar assay conditions, we found that E. coli Fes
catalyzes the hydrolysis of apo Ent and apo MGE, but catalyzes
the hydrolysis of apo DGE very poorly, and does not process
apo TGE at all (Figure 3). These results indicate that Fes exhibits
a bias against processing the more extensively glucosylated
trilactone substrates.
Initial Enzyme Activity Assay of IroD, IroE, and Fes with
Fe-Ent, Ferric MGE (Fe-MGE), Ferric DGE (Fe-DGE),
and Ferric TGE (Fe-TGE). Release of iron from Fe-Ent
molecules imported to the cytoplasm is thought to occur by Fes-
catalyzed hydrolysis of the trilactone scaffold of Fe-Ent.1 By
analogy, iron release from glucosylated Ent following import
to the bacterial cell could be achieved by IroD- and/or IroE-
catalyzed hydrolysis. However, previous studies on Fes reached
disparate conclusions about its preference for apo or Fe3+-bound
Ent as substrate,13-16 with several reports suggesting that Fes
preferentially hydrolyzes apo Ent. We therefore tested whether
IroD and IroE can catalyze the hydrolysis of Fe3+-bound
siderophores (Figure 4). To our surprise, we found that under
conditions similar to those for the apo siderophore hydrolysis
assays, IroE hydrolyzes the Fe3+-bound siderophores very
inefficiently. IroD catalyzes the hydrolysis of Fe3+-bound
siderophores, with rates faster than those of IroE for ferric
substrates, but about 100-fold lower than those of IroD for apo
substrates. We also examined the hydrolysis of Fe-Ent by Fes
with the purified Fes protein and found that the rate is also about
100-fold slower for Fe-Ent than for apo Ent. The reluctance
of Fes, IroD, and IroE to hydrolyze Fe3+-bound siderophores
seemed at odds with their presumed physiological function of
releasing bound Fe3+. The detailed kinetics studies described
below resolved this apparent contradiction.
Kinetic Parameters of Fes, IroD, and IroE. The kinetic
parameters of Fes-, IroD-, and IroE-catalyzed hydrolysis of both
apo and Fe3+-bound forms of Ent, MGE, DGE, and TGE were
measured and are summarized in Table 2.
For Fes-catalyzed apo Ent hydrolysis, no saturation was
observed at Ent concentrations up to 128 µM, arguing against
the apo form as natural substrate. We observed that kcat/Km
)
4.0 µM-1 min-1, Km is greater than 128 µM and, therefore,
inferred that kcat is greater than 512 min-1. In contrast, for Fes-
catalyzed Fe-Ent hydrolysis, the reaction reaches saturation
even at 0.125 µM Fe-Ent concentration, which was the lowest
concentration tested given the sensitivity limit of the HPLC
assay used. We determined the kcat of Fes on Fe-Ent to be 9
min-1, Km to be less than 0.1 µM, and, therefore, inferred that
kcat/Km is greater than 90 µM-1 min-1. The decrease in kcat for
Fes-catalyzed hydrolysis of Fe-Ent versus apo Ent explains
why we observed a much slower reaction for Fe-Ent hydrolysis
in the initial enzyme assays. However, the catalytic efficiency
(kcat/Km) of Fes for processing Fe-Ent is much higher than that
for apo Ent, suggesting that Fe-Ent is the physiological
substrate of Fes.
Fes catalyzes the hydrolysis of Fe-MGE, albeit 20-fold less
efficiently than Fe-Ent, but does not catalyze the hydrolysis
9
11080 J. AM. CHEM. SOC. VOL. 127, NO. 31, 2005

Bacterial Iron Release from Enterobactin and Salmochelins
A R T I C L E S
Figure 5. Partial HMBC spectra of MGE linear trimer (A) and dimer (B) obtained from IroD-catalyzed hydrolysis reaction. Partial assignments are shown,
which are based on HSQC, COSY, and HMBC. The pink arrows indicate possible glucose attachment sites. The blue arrows indicate carbon-proton 2J and
3J couplings present in the molecules that are used to determine the structures.
of Fe-DGE or Fe-TGE. The catalytic efficiency of Fes for
Fe-MGE is very small compared to the catalytic efficiency of
Fes for Fe-Ent or the catalytic efficiency of IroD for Fe-MGE
and Fe-DGE. Therefore, Fes is unlikely to be the enzyme that
hydrolyzes Fe-MGE or Fe-DGE to release Fe3+, explaining
the need for a separate salmochelin hydrolase.
IroD catalyzes the hydrolysis of both the apo and Fe3+-bound
forms of Ent, MGE, DGE, and TGE. Furthermore, we observed
a kcat/Km preference for the Fe3+-bound form of Ent, MGE,
DGE, and TGE over the apo forms. Both kcat’s and Km’s are
smaller for the Fe3+-bound forms than the apo forms, but the
decreases in Km’s are more dramatic, resulting in higher kcat/
Km values for the Fe3+-bound form.
IroE also hydrolyzes both apo and Fe3+-bound forms of Ent,
MGE, DGE, and TGE. However, unlike IroD, IroE hydrolyzes
apo forms more efficiently. The catalytic efficiencies for the
apo forms are generally about 20-fold higher than those for the
Fe3+-bound forms. Furthermore, at higher concentrations, Fe-
Ent, Fe-MGE, and Fe-DGE inhibit IroE (Supporting Informa-
tion). These data suggest that IroE is not the enzyme responsible
for iron release from Fe-MGE, Fe-DGE, or Fe-Ent in vivo.
Therefore, IroD is probably the only enzyme that hydrolyzes
Fe-MGE and Fe-DGE to release iron in vivo.
Regioselectivity of IroD on apo MGE and DGE. As
mentioned above, IroD catalyzes the hydrolysis of apo MGE
and DGE regioselectively, to give a single linear trimer product
and a single linear dimer product for each. This regioselectivity
is especially interesting given the recently reported structure of
microcin E492,27 in which a linearized MGE is covalently
attached the microcin peptide’s C-terminus via the C6 hydroxy
of the glucose. IroB and IroD homologues are found in the
biosynthesis gene cluster of microcin E492.28,29 The linearized
MGE found in microcin E492 bears the glucosyl exclusively
on the DHB-Ser with the free carboxylic acid. Therefore, it
would be interesting to determine IroD’s regio-
selectivity and see whether it is similar to that of the homologues
found in the microcin biosynthesis pathway. We performed
larger-scale MGE and DGE hydrolysis reactions and isolated
the trimer and dimer products in multimilligram quantities by
1
reverse-phase HPLC purification. Using H NMR, 13C NMR,
COSY, HSQC, and HMBC experiments (Figure 5 and Sup-
(27) Thomas, X.; Destoumieux-Garzon, D.; Peduzzi, J.; Afonso, C.; Blond, A.;
Birlirakis, N.; Goulard, C.; Dubost, L.; Thai, R.; Tabet, J. C.; Rebuffat, S.
J. Biol. Chem. 2004, 279, 28233-28242.
(28) Azpiroz, M. F.; Lavina, M. Antimicrob. Agents Chemother. 2004, 48, 1235-
1241.
(29) Corsini, G.; Baeza, M.; Monasterio, O.; Lagos, R. Biochimie 2002, 84,
539-544.
9
J. AM. CHEM. SOC. VOL. 127, NO. 31, 2005 11081

A R T I C L E S
Lin et al.
Figure 6. Summary of IroD regioselectivity on intact macrolactones and linear trimers. The structures of the linear trimers and dimer from IroD-catalyzed
hydrolysis of MGE and DGE are shown, with the preferred hydrolysis site indicated at each step. The regioselectivity of IroD can be generalized from these
results. On the basis of the presence (+) or absence (-) of glucose on the carboxyl-side DHB ring and the hydroxyl-side DHB ring, each ester bond can be
characterized with one of the following symbols, (+/+), (+/-), (-/+), and (-/-). The first ‘+’ or ‘-’ sign indicates the presence or absence of glucose
on the carboxyl-side DHB ring, while the second sign indicates the presence or absence of glucose on the hydroxyl-side DHB ring. The order of IroD
preference is (+/-) > (+/+), (-/-) > (-/+) for the hydrolysis of intact macrolactones and linear trimers.
porting Information), we determined the glucose attachment
positions in the dimer and trimer products resulting from MGE
and DGE hydrolysis. Figure 5A shows the partial HMBC spectra
of the trimer from IroD-catalyzed MGE hydrolysis, with partial
the DHB with the glucose attached. Therefore, from this
information, we could only conclude that the glucose unit is
either on the DHB1 or DHB3, but not on DHB2. Using similar
methods, we conclude that in the dimer product arising from
MGE hydrolysis, the glucose is attached to DHB1 (Figure 5B).
To generate this dimer product requires that the glucose in the
trimer product must be attached to DHB1. Therefore, both the
trimer and dimer structures from MGE hydrolysis were eluci-
dated as the structures shown in Figure 6. It should be noted
that the structure of the linear MGE matches that from microcin
E492. The trimer and dimer structures generated by IroD-
catalyzed DGE hydrolysis were solved similarly, and the results
are summarized in Figure 6.
Interestingly, the dimers generated by IroD-catalyzed DGE
hydrolysis and MGE hydrolysis are the same compound, with
one glucose on the DHB-Ser with the free carboxylic acid.
Careful analysis of the regioselectivity indicates that IroD prefers
to hydrolyze the ester bond with glucose attached to the
carboxyl-side DHB-Ser unit and no glucose on the hydroxyl-
side DHB-Ser unit. If the most favorable ester bond is not
present, then IroD will hydrolyze the ester bond with glucose
1
assignments of the H and 13C spectra. The assignment of the
three sets of R and â protons from the three serine units was
based on both COSY and HSQC spectra. The two â protons
with lower chemical shifts were assigned to Ser3 with the free
hydroxyl group since the â-OH acylated Ser1 and Ser2 â protons
should have higher chemical shifts due to the electron-
withdrawing properties of the acyl groups. This analysis is
confirmed in HMBC spectra because Ser3 â protons are only
coupled to the Ser3 carbonyl carbon, while Ser1 and Ser2 â
protons are each coupled to two Ser carbonyl carbons. With
the assignment of the Ser3 â proton, the rest of the R and â
protons, and all the carbonyl carbons, including the DHB
carbonyl carbons and the Ser carbonyl carbons, can be assigned
2
by walking through the HMBC spectrum following the J and
3J couplings shown in Figure 5A. The DHB1 and DHB3
carbonyl carbons for the trimer appear to have the same chemical
shift, and this carbonyl 13C peak is coupled to the C6 proton of
9
11082 J. AM. CHEM. SOC. VOL. 127, NO. 31, 2005

Bacterial Iron Release from Enterobactin and Salmochelins
A R T I C L E S
Figure 7. A schematic representation showing the possible functions of Fes, IroD, and IroE.
on both DHB-Ser units or no glucose on either DHB-Ser unit,
while the ester bond with no glucose on the carboxyl-side
DHB-Ser units but with glucose on the hydroxyl-side DHB-
Ser unit is least favored (Figure 6).
that Fes is the Fe-Ent esterase responsible for iron release from
Fe-Ent (Figure 7A).
The kinetic data also explain the apparent inconsistency
regarding the substrate specificity of Fes (apo vs Fe3+-bound
Ent) in the literature.13-16 The disparate results are likely due
to different concentrations of substrates used in the assays. If
high concentrations of Fe-Ent and apo Ent were used, then
the specific activities reported would reflect kcat values and lead
to the conclusion that Fes hydrolyzes apo Ent more efficiently.
In contrast, if low concentrations were used, then the specific
activities would reflect kcat/Km values and lead to the conclusion
that Fes hydrolyzes Fe-Ent preferentially. Indeed, in the work
of Longman et al.,14 which reports that Fes prefers Fe-Ent as
substrate, substrate concentrations of 0.15 µM are used; in
contrast, Brickman and McIntosh,16 who arrive at the opposite
conclusion, use substrate concentrations of 750 µM.
IroD is the Fe-MGE/Fe-DGE Esterase and It Prefers
Ferric Forms over apo Forms as Substrates. The presence
of Fes homologues IroD and IroE in the iroA cluster suggests
that Fes might not be efficient at hydrolyzing Fe-MGE and
Fe-DGE for iron release. Our kinetic data indeed show that
Fes hydrolyzes Fe-MGE poorly and does not hydrolyze Fe-
DGE or Fe-TGE at all, confirming that IroD, IroE, or both
are required for Fe-MGE and Fe-DGE hydrolysis.
Discussion
The catecholic siderophore Ent binds Fe3+ with a estimated
Kd of 10-49 M (10-35 M at physiological pH). Cytoplasmic
release of Fe3+ from such a high affinity complex is generally
believed to result from Fes-catalyzed hydrolysis of the trilactone
scaffold of Fe-Ent, although reduction of Fe3+ to Fe2+ might
also play a role.1 The substrate specificity of Fes, however, has
been confusing due to conflicting reports.13-16 We address this
issue in the context of this study.
The iroA gene cluster from certain pathogenic E. coli and
Salmonella strains encodes two Fes homologues, IroD and IroE,
which have been proposed to be glucosylated enterobactin
hydrolases based on bioinformatics and genetic studies.17,18 We
have overexpressed and purified all three enzymes and carried
out detailed kinetics studies on both apo and Fe3+-bound forms
of Ent, MGE, DGE, and TGE. The results provide valuable
information about the in vivo functions of Fes, IroD, and IroE
(summarized in Figure 7).
Fes is the Fe-Ent Esterase and It Prefers Fe-Ent over
apo Ent. In the initial activity assay, we found that Fes
hydrolyzes apo Ent much faster than Fe-Ent, which is
consistent with some reports,16 but not others.14 Detailed kinetic
study indicates that interpreting catalytic turnover (kcat) or the
catalytic efficiency (kcat/Km) yields different conclusions. If Fes
is operating at low concentrations of Ent/Fe-Ent in the
cytoplasm (which is probably true in vivo), the apparent second-
order rate constant, kcat/Km, will be the relevant parameter for
measuring throughput. By this criterion, Fes prefers Fe-Ent
over apo Ent as substrate. The catalytic efficiency is much higher
for Fe-Ent (>90 min-1 µM-1) than for apo Ent (4 min-1
µM-1). If both Ent and Fe-Ent are present at similar concentra-
tions and the concentration of Fes is limiting, then nearly all
Fes molecules will be bound by Fe-Ent because of its much
smaller Km, and hydrolysis of Fe-Ent will dominate. In this
sense, Fe-Ent is a strong inhibitor of apo Ent hydrolysis by
Fes. Therefore, our results suggest that Fes prefers Fe-Ent over
apo Ent as the substrate and are consistent with the conclusion
Both IroD and IroE catalyze the hydrolysis of apo Ent, MGE,
DGE, and TGE. However, their activities differ substantially
on Fe3+-bound Ent, MGE, DGE, and TGE. IroD hydrolyzes
Fe3+-bound forms very efficiently, while IroE hydrolyzes Fe3+
-
bound forms inefficiently (Table 2). Furthermore, at higher
concentrations, Fe3+-bound substrates inhibit the hydrolysis
reactions catalyzed by IroE (Supporting Information), while no
substrate inhibition was observed for IroD. Therefore, our results
suggest that IroD is the Fe-MGE/Fe-DGE esterase responsible
for iron release in strains harboring the iroA gene cluster (Figure
7B).
Like Fes, IroD prefers the Fe3+-bound form of Ent, MGE,
DGE, and TGE as substrates by catalytic efficiency analysis.
Even though the absolute kcat values are smaller, the kcat/Km
values for Fe3+-bound forms are at least 4-fold higher than those
for the corresponding apo forms, and the Km values are at least
330-fold lower (Table 2). The low Km’s (∼100 nM) of Fes and
9
J. AM. CHEM. SOC. VOL. 127, NO. 31, 2005 11083

A R T I C L E S
Lin et al.
IroD on Fe3+-bound substrates suggest that the in vivo
concentrations of these ferric siderophores are also very low.
ring-opened salmochelins that are subsequently secreted into
the culture medium.
IroE Hydrolyzes apo Siderophores More Efficiently Than
Fe3+-Bound Forms, and In Vivo Probably Only Works on
apo Forms While They Are Being Exported. IroE was
predicted to be a periplasmic protein.18 The observation that
full-length IroE is insoluble while the deletion of the first 30 or
40 amino acids provided soluble IroE supports the prediction
that IroE is tethered to the outer leaflet of the cytoplasmic
membrane by an N-terminal membrane insertion motif. Our
kinetic data suggest that IroE prefers apo Ent, MGE, DGE, and
TGE as substrates because of its very low catalytic efficiencies
for processing the Fe3+-bound forms, and also because it is
substrate-inhibited at higher concentrations of Fe-Ent, Fe-
MGE, and Fe-DGE. Previous genetic studies indicate that IroE
functions in vivo because less of the hydrolyzed salmochelins
are isolated from bacterial strains harboring the iroA cluster
lacking iroE.17 These results suggest a model in which IroE
acts in the periplasm and hydrolyzes some or all of apo Ent,
MGE, and DGE while they are being exported out of the cell
(Figure 7C). If the inhibition by Fe-Ent, Fe-MGE, Fe-DGE
also happens in vivo (that is, if IroE has access to Fe-Ent, Fe-
MGE, and Fe-DGE while they are being imported), then the
hydrolysis of apo forms would only happen shortly after the
Even though Fes and IroD also catalyze the hydrolysis of
apo siderophores in vitro, we do not know whether the activity
on apo substrates is physiologically relevant or not. It has been
proposed that proteins involved in Ent biosynthesis are membrane-
associated and the synthesis of Ent is closely coupled to the
export machinery to prevent damage by intracellular apo
Ent.2,30,31 If this is true, it is possible that Fes and IroD do not
have access to apo siderophores in vivo under normal conditions.
IroD Hydrolyzes MGE and DGE Regioselectively. IroE
hydrolyzes MGE and DGE with little regioselectivity, generating
at least two linearized trimer products. In contrast, IroD
regioselectively hydrolyzes the trilactone in apo MGE and DGE.
We have determined the structures of the linearized trimer and
dimer products from IroD-catalyzed MGE and DGE hydrolysis
by NMR. The linear trimer product from MGE hydrolysis is
consistent with the linearized MGE structure present in microcin
E492,27 in which the glucose unit is linked to the DHB-Ser
that bears the free carboxylic acid group. Therefore, our result
agrees with the reported microcin E492 structure and suggests
a biosynthetic role for MceD, the IroD homologue found in the
microcin E492 gene cluster.27-29 We could not determine
whether the linearized MGE in microcin E492 arises from apo
MGE hydrolysis or Fe-MGE hydrolysis because HPLC analysis
showed that the products from Fe-MGE hydrolysis by IroD
are the same as those from apo MGE hydrolysis (Supporting
Information), suggesting that the regioselectivity is preserved
for the hydrolysis of both forms. It is also possible that IroD
hydrolyzes MGE after it is attached to the microcin peptide.
cells initiate the synthesis of these siderophores and no Fe3+
-
bound forms are being imported. However, it is possible that
during import, the siderophores are bound to transporter proteins
and, therefore, are not accessible to IroE. The fact that bacteria
utilize IroE to hydrolyze apo siderophores while being exported
suggests that the linearized siderophores may behave differently
from their macrocyclic counterparts.
The initial hydrolysis of the trilactone scaffold of DGE results
in a linear trimer product that has two glucose units which are
attached to the DHB-Ser with the free carboxylic acid and to
the central DHB-Ser. Interestingly, the linear dimer product
from DGE hydrolysis is the same as the dimer product from
MGE hydrolysis, with one glucose unit attached to the DHB-
Ser bearing the free carboxylic acid. This regioselectivity of
IroD can be explained by its preference for Glc-DHB-Ser on
the carboxyl-side and DHB-Ser on the hydroxyl-side of the
ester bond being hydrolyzed, as illustrated in Figure 6.
In summary, as shown in Figure 7, our in vitro study of IroD,
IroE, and Fes provides important information about their in vivo
functions and advances our understanding about how bacteria
use enterobactin and glucosylated enterobactins as virulence
factors to obtain iron from the environment.
Acknowledgment. This work is supported in part by NIH
AI 47238 (C.T.W.) and by NIH R01GM065400 (D.R.L.), a
postdoctoral fellowship from the Jane Coffin Childs Memorial
Fund (H.L.), and a graduate fellowship from the Hertz Founda-
tion (M.A.F.).
The linear trimers and dimers from MGE and DGE hydrolysis
by IroD are different from salmochelins S1 and S2,17 suggesting
that salmochelins S1 and S2 detected in S. enterica culture broth
are probably the products of IroE-catalyzed hydrolysis. Since
the complete structural characterization of all salmochelins has
not been performed, we cannot rule out that IroD also produces
Supporting Information Available: The reaction time courses
of IroD-, IroE-, and Fes-catalyzed hydrolysis of Fe-Ent, Fe-
MGE, Fe-DGE, and Fe-TGE; substrate inhibition of IroE; the
NMR (1H, 13C, COSY, HSQC, HMBC) spectra for linear trimer,
dimer products from IroD-catalyzed MGE, DGE hydrolysis.
This material is available free of charge via the Internet at
http://pubs.acs.org.
(30) Hantash, F. M.; Earhart, C. F. J. Bacteriol. 2000, 182, 1768-1773.
(31) Armstrong, S. K.; Pettis, G. S.; Forrester, L. J.; McIntosh, M. A. Mol.
Microbiol. 1989, 3, 757-766.
JA0522027
9
11084 J. AM. CHEM. SOC. VOL. 127, NO. 31, 2005