Nanomolar inhibition of the enterobactin biosynthesis enzyme, EntE: Synthesis, substituent effects, and additivity

Article abstract of DOI:10.1016/j.bmcl.2006.04.024

2,3-Dihydroxybenzohydroxamoyl adenylate (I) was prepared as a potential product analog inhibitor of EntE (EC# 2.7.7.58), a 2,3-dihydroxybenzoate AMP ligase from Escherichia coli that is required for the biosynthesis of enterobactin. This compound, obtained by the aqueous reaction of imidazole-activated adenosine 5′-phosphate and 2,3-dihydroxybenzohydroxamic acid, is a competitive inhibitor with a K_i value of 4.5 × 10^-9 M. Deletion of the catecholic 3-OH group of (I), in compound (II), reduced inhibitory activity by a factor of 3.5, whereas, removal of both the 3-OH and 2-OH groups, in (III), reduced inhibitory activity by a factor of ～2000. Acetohydroxamoyl adenylate (IV), in which the entire catechol moiety of (I) is replaced by a hydrogen atom, gave {less-than or slanted equal to}10% inhibition at 6 × 10^-4 M, indicating a reduction in affinity by more than 10⁵. The binding free energy of (I) is nearly equivalent to the sum of the corresponding values for adenosine 5′-phosphate and 2,3-dihydroxybenzoate.

Full text of DOI:10.1016/j.bmcl.2006.04.024

Bioorganic & Medicinal Chemistry Letters 16 (2006) 3802–3805
Nanomolar inhibition of the enterobactin biosynthesis enzyme,
EntE: Synthesis, substituent effects, and additivity
Brian P. Callahan,* Joseph V. Lomino and Richard Wolfenden
Department of Biochemistry and Biophysics, University of North Carolina, Chapel Hill, NC 27599-7260, USA
Received 2 March 2006; revised 12 April 2006; accepted 12 April 2006
Available online 5 May 2006
Abstract—2,3-Dihydroxybenzohydroxamoyl adenylate (I) was prepared as a potential product analog inhibitor of EntE (EC#
2.7.7.58), a 2,3-dihydroxybenzoate AMP ligase from Escherichia coli that is required for the biosynthesis of enterobactin. This com-
pound, obtained by the aqueous reaction of imidazole-activated adenosine 50-phosphate and 2,3-dihydroxybenzohydroxamic acid,
is a competitive inhibitor with a Ki value of 4.5 · 10ꢀ9 M. Deletion of the catecholic 3-OH group of (I), in compound (II), reduced
inhibitory activity by a factor of 3.5, whereas, removal of both the 3-OH and 2-OH groups, in (III), reduced inhibitory activity by a
factor of ꢁ2000. Acetohydroxamoyl adenylate (IV), in which the entire catechol moiety of (I) is replaced by a hydrogen atom, gave
610% inhibition at 6 · 10ꢀ4 M, indicating a reduction in affinity by more than 105. The binding free energy of (I) is nearly equivalent
to the sum of the corresponding values for adenosine 50-phosphate and 2,3-dihydroxybenzoate.
ꢀ 2006 Elsevier Ltd. All rights reserved.
Siderophores are low molecular weight chelating agents
secreted by certain strains of infectious bacteria, includ-
ing Mycobacterium tuberculosis1 and Yersinia pestis2
(cause of plague), that require iron for virulence. Their
equilibrium constants (Ka) for association with free FeIII
are sufficiently high as to render chelation essentially
irreversible,3 and liberation of bound FeIII for cellular
use requires pathogen hydrolases that dismantle the fer-
ric siderophore itself.4 Because no equivalent systems
seem to be present in mammals, the enzymes involved
in siderophore biosynthesis5 represent logical targets
for the development of bactericidal agents.6–10
sis.11 Using the same number of serine residues, down-
stream enzymes tether the DHB moieties into a ring,
allowing each of the six catecholic hydroxyl groups to
coordinate FeIII (Scheme 1). Recent evidence suggests
that production of this siderophore by E. coli may be
correlated with virulence in urinary tract infections,12
and similar associations have been found for catecho-
late-type siderophores produced by the pathogens, Bru-
cella abortus and Vibrio anguillarum, which require the
action of EntE homologues (Table 1).
Like many adenylate forming enzymes, EntE binds its
product with high affinity, perhaps to protect DHB–
AMP from being exposed to attack by adventitious
nucleophiles in the cell and ensure that the labile acyl
group is available to holo EntB for the next enzymatic
transformation. By measuring the catalytic turnover of
pyrophosphate released from EntE, a rate constant of
0.4 minꢀ1 has been determined for dissociation of the
EntE–DHB–AMP complex,11 much lower than dissoci-
ation rate constants of typical E–P complexes.13
Here, we describe a new class of aryl adenylate analogs
with potent inhibitory activity against EntE, an enzyme
from Escherichia coli that catalyzes the formation of 3
equivalents of 2,3-dihydroxybenzoyl adenylate (DHB–
AMP) during the early stages of enterobactin biosynthe-
Abbreviations: DHB, 2,3-dihydroxybenzoic acid; DHB–AMP, 2,3-di-
hydroxybenzoyl adenylate; AMP, adenosine 50-monophosphate;
Me-DHB, methyl 2,3-dihydroxybenzoic acid; DHB-Hx, 2,3-dihydro-
xybenzohydroxamic acid.
Keywords: Adenylate analog; Inhibitor; Additivity; EntE; Enterobac-
tin; Siderophore; Adenylate-forming; Bisubstrate; Iron; Inhibition;
Binding energy.
We attempted to exploit the high affinity of the product
binding site by designing a stable mimic of DHB–AMP
as a potential inhibitor. This analog contains an N-acyl
hydroxamoyl phosphate group in place of the labile car-
boxylic–phosphoric anhydride linkage, a modification
that increases the distance between the AMP and aryl
*
Corresponding author at present address: Division of Infectious
Disease, Wadsworth Center, New York State Department of Health,
Albany, NY 12201, USA. Tel.: +1 518 473 6067; fax: +1 518 402
2633; e-mail: callahan@wadsworth.org
˚
moieties by 2 A and renders the molecule unreactive.
The choice of a phosphoryl group, rather than an
0960-894X/$ - see front matter ꢀ 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2006.04.024

B. P. Callahan et al. / Bioorg. Med. Chem. Lett. 16 (2006) 3802–3805
3803
Scheme 1.
Table 1. DHB–AMP forming enzymes in microbial siderophore biosynthesis
EntE homologue
Catecholate siderophore
Microbe
DhbEa
AngEb
VibEc
DhbEd
AgbEe
Bacillibactin
Anguibactin
Vibriobactin
Brucebactin
Agrobactin
Bacillus subtilis
Vibrio anguillarum (fish pathogen)
Vibrio cholerae (cause of cholera)
Brucella abortus (ruminant pathogen)
Agrobacterium tumefaciens (plant pathogen)
a May, J. J.; Wendrich, T. M.; Marahiel, M. A. J. Biol. Chem. 2001, 276, 7209.
b Lorenzo, M. D.; Stork, M.; Tolmasky, M. E.; Actis, L. A.; Farrell, D.; Welch, T. J.; Crosa, L. M.; Wertheimer, A. M.; Chen, Q.; Salinas, P.;
Waldbeser, L.; Crosa, J. H. J. Bacteriol. 2003, 185, 5822.
c Wyckoff, E. E.; Stoebner, J. A.; Reed, K. E.; Payne, S. M. J. Bacteriol. 1997, 179, 7055.
d Bellaire, B. H.; Elzer, P. H.; Hagius, S.; Walker, J.; Baldwin, C. L.; Roop, R. M. 2nd. Infect. Immun. 2003, 71, 1794.
e Sonoda, H.; Suzuki, K.; Yoshida, K. Genes Genet. Syst. 2002, 77, 137.
uncharged surrogate, was based on the X-ray structure
of the homologous enzyme from Bacillus subtilis, DhbE
(47% identical based on its amino acid sequence), which
shows a potentially favorable interaction between the
side chain of a conserved active site histidine residue
(H234) and the 50-phosphoryl group of bound DHB–
AMP.14
imidazolium group of AMP-Im18 and the respective
hydroxamic acid.19 For III and IV, reaction times were
extended to 24 h and the temperature lowered to 4 ꢁC.
Yields for each of these analogs were >50%.20
Competitive inhibition experiments indicate that com-
pound I was bound by EntE with a Ki(app) value of
9 nM. That affinity is 102 more favorable than that of
DHB (Km, 2.7 lM) and 105 more favorable than that
of ATP (Km, 1120 lM),11 suggesting that I may interact
with both substrate binding sites at once.
The preparation of hydroxamate-linked adenylate ana-
logs was conceptually similar to the syntheses of sulfa-
moyl-7 or tetrazolide-9linked analogs, but did not
require catalysts, protecting groups, or anhydrous
solvent conditions. Compound I was obtained by com-
bining imidazole-activated adenosine 50-phosphate
(AMP-Im)15 (0.2 mmol) with 2,3-dihydroxybenzohydr-
oxamic acid (DHB-Hx) (0.4 mmol) in 6 ml H2O at
Binding of compound I is competitive with respect to
substrate DHB. Figure S3 shows that increasing the
concentration of I results in an equivalent increase in
the apparent Km value of DHB. The Ki value of I, cor-
rected for substrate competition, is 4.5 nM. The strength
of this interaction appears to be substantially greater
than that reported for an adenylate analog inhibitor of
DhbE in which a sulfamoyl group replaces the native
phosphoryl group of DHB–AMP (Ki, 85 nM, 37 ꢁC).10
A second sulfamoyl-linked analog has been reported
to inhibit the salicyl activating enzyme, YbtE, from
M. tuberculosis, but binding was noncompetitive with
respect to salicylate.7 Although trivial explanations for
different inhibitor modalities are possible, a likely expla-
nation is that, despite their sequence similarity (42%
identical based on amino acid sequence), EntE and
YbtE may follow different kinetic mechanisms (random
vs ordered, respectively).
31
25 ꢁC, pH 7.0. After 1 h, P NMR16 showed 76% con-
version to I. The solution was passed over a Q-cellulose
column (1.25 · 35 ml) at a flow rate of 1.1 ml/min of
potassium phosphate buffer (0.025 M, pH 6.2) at 4 ꢁC,
with the results shown in Figure S1. Fractions containing
I were pooled, concentrated in vacuo, and desalted using
a Sephadex G-10 column (38 · 1 cm). Results of high
1
resolution ESI/MS and H NMR were consistent with
the assigned structure.17 Stock solutions were stored at
ꢀ80 ꢁC.
Compounds II–IV were prepared in a similar manner
except that the pH of the reaction was adjusted to a
value roughly midway between the pKa values of the

3804
B. P. Callahan et al. / Bioorg. Med. Chem. Lett. 16 (2006) 3802–3805
To investigate the basis for the strong interaction be-
tween EntE and I, compounds II–IV were prepared
and tested as potential inhibitors, with the results sum-
marized in Table 2. Deletion of the catecholic 3-OH
group of I, in compound II, resulted in only a minor
(3.5-fold) reduction in the observed inhibitory activity.
For compound III, in which both hydroxyl groups are
removed, enzyme affinity was reduced by a factor of
2000. This result is in reasonable agreement with the
substrate specificity of EntE, described by Rusnak
et al.11 Thus, 2-hydroxybenzoate is activated by EntE
10-fold less efficiently (kcat/Km) than DHB, whereas 3-
hydroxybenzoate and benzoate are not acted upon by
EntE to any measurable extent. Together, these results
indicate the importance of binding interactions between
EntE and the catecholic 2-OH group. The superior inhi-
bition displayed by III, compared with IV (610% inhibi-
tion at 5.9 · 10ꢀ4 M), suggests that noncovalent forces
of attraction are also at work between the unsubstituted
benzene ring and the EntE active site (Table 2).
Figure 1. Estimated free energy of dissociation of EntE with substrates
and inhibitors [DGdiss = ꢀRTln(Kdiss)]. Asterisks indicate the algebraic
sum of the free energies of dissociation of the substrate, using Km
values reported by Rusnak et al., and AMP using the Ki(app) value of
9 · 10ꢀ4 M determined in this work. SAL: 2-hydroxybenzoate.
If compound I is represented as A–B, and its compo-
nents, AMP and DHB, are represented as A and B, it
would be of interest to know whether the free energy
of binding of A–B is equivalent to the sum of the free
energies of binding of A and B individually:
ibility of the linker between the aryl and AMP moi-
eties of compound I might disturb that relationship,
and permit its two fragments to be bound
synergistically.
AꢀB
¼ DGAbind þ DGBbind
Compound I is a powerful inhibitor of EntE, suggest-
ing that the N-acyl hydroxamoyl phosphate group
effectively mimics the acyl phosphate group present
in DHB–AMP. Inhibitor synthesis is carried out in
water, gives high yields, and uses reactants that are
storable and in some cases commercially available.
Comparisons with the inhibitory activity of derivatives
II and III indicate that the catecholic 2-OH group is
critical for tight binding. Tight binding of the EntE–
I complex also appears to be a consequence of the
near additivity of the free energies of binding of its
AMP and DHB moieties.
DG
bind
Such a comparison, evaluating the additivity of inhib-
itor binding interactions, might be expected to be
sensitive to differences in entropy, strain, or solvation,
associated with binding of A–B relative to A or B.21
As shown in Figure 1, however, the binding of com-
pounds I and II appears to be purely additive.22 In
future experiments, it would be of interest to explore
the possibility that adjustments in the length and flex-
Table 2. Ki(app) values with EntE (uncorrected for substrate compe-
tition)
NH2
Acknowledgments
N
N
We thank Dr. Christopher Walsh and Michael Fisch-
bach for the kind gift of purified EntE used in this study
and Dr. Michael Caplow for comments on the manu-
script. This work was supported by National Institutes
of Health Grant GM-18325 (R.W.).
O
N
N
R
O
P
O
O
O
OH OH
Inhibitor
R
Ki(app) (M)
9 · 10ꢀ9
Supplementary data
OH
O
HO
I
N
Experimental details for all the syntheses and the inhi-
bition results are available in Supplementary data. Sup-
plementary data associated with this article can be
found in the online version at doi:10.1016/
j.bmcl.2006.04.024.
H
OH O
II
3.7 · 10ꢀ8
1.3 · 10ꢀ5
P1 · 10ꢀ3
N
H
O
III
IV
N
H
References and notes
O
1. de Voss, J. J.; Rutter, K.; Shroeder, B. G.; Su, H.; Zhu, Y.;
Barry, C. E., III Proc. Natl. Acad. Sci. U.S.A. 2000, 97,
2352.
H3C
N
H

B. P. Callahan et al. / Bioorg. Med. Chem. Lett. 16 (2006) 3802–3805
3805
31
16. The P chemical shift of the 50phosphoryl group of I is
ꢀ1.4 ppm, slightly upfield of AMP at 0.7 ppm.
2. de Almeida, A. M.; Guiyoule, A.; Guilvout, I.; Iteman, I.;
Baranton, G.; Caraniel, E. Microb. Pathog. 1993, 14, 9.
3. Neilands, J. B. J. Biol. Chem. 1995, 270, 26723.
4. Lin, H.; Fischbach, M. A.; Liu, D. R.; Walsh, C. T. J. Am.
Chem. Soc. 2005, 127, 11075.
5. For an extensive review see Crosa, J. H.; Walsh, C. T.
Microbiol. Mol. Biol. Rev. 2002, 66, 223.
6. Finking, R.; Neumuller, A.; Solsbacher, J.; Konz, D.;
¨
Kretzchmar, G.; Scheitzer, M.; Krumm, T.; Marahiel, M.
A. ChemBioChem 2003, 4, 903.
7. Ferraras, J. A.; Ryu, J.-S.; Di Lello, F.; Tan, D. S.;
Quadri, L. E. N. Nat. Chem. Biol. 2005, 1, 29.
8. Kerbarh, O.; Bulloch, E. M.; Payne, R. J.; Sahr, T.;
Rebeille, F.; Abell, C. Biochem. Soc. Trans. 2005, 33, 763.
9. Somu, R. V.; Boshoff, H.; Qiao, C.; Bennett, E. M.; Barry,
C. E., III; Aldrich, C. C. J. Med. Chem. 2006, 49, 31.
10. Miethke, M.; Bisseret, P.; Beckering, C. L.; Vignard, D.;
Eustache, J.; Marahiel, M. A. FEBS J. 2006, 273, 409.
11. Rusnak, F.; Faraci, S. W.; Walsh, C. T. Biochemistry
1989, 28, 6827.
17. 1H NMR (D2O) d 8.23 (s, 1H), 8.06 (s, 1H), 6.79 (d, 1H,
J = 8 Hz), 6.72 (d, 1H, J = 8 Hz), 6.48 (t, 1H, J = 8 Hz),
5.96 (d, 1H, J = 5.5 Hz), 4.65 (t, 1H), 4.49 (m, 1H), 4.38
(ddd, 1H, J = 11, 5, 2 Hz), 4.33 (m, 1H), 4.14 (dt, 1H,
J = 11, 3.7 Hz).
18. pKa value is ꢁ6.5, for imidazolium group of AMP-Im.
Ruzicka, F. J.; Frey, P. A. Bioorg. Chem. 1993, 21, 238.
19. pKa values, respectively, are 9.4, 8.88, and 7.43 for
acetohydroxamic, benzohydroxamic, and salicylhydroxa-
mic acids Jencks, W. P.; Regenstein, J. Handbook of
Biochemistry; Chemical Rubber Company Press: Cleve-
land, OH, 1968, J159.
20. The robust yields of each analog, despite the presence of a
potentially nucleophilic solvent, appear to be the result of
a tight transition that forms when anions react at the
formally neutral phosphoryl group of AMP-Im. (Calla-
han, B.P. unpublished result.)
21. Jencks, W. P. Proc. Natl. Acad. Sci. U.S.A. 1981, 78,
4046.
12. Demir, M.; Kaleli, I. Clin. Microbiol. Infect. 2004, 10,
1011.
22. Additive binding has also been reported with ‘fragment
linked’ inhibitors of sweet almond beta-glucosidase
(Li, Y. K.; Byers, L. D. Biochim. Biophys. Acta 1989,
999, 227); FK506 binding protein (Shuker, S. B.; Hajduk,
P. J.; Meadows, R. P.; Fesik, S. W. Science 1996, 274,
1531); and thrombin (Howard, N.; Abell, C.; Blakemore,
W.; Chessari, G.; Congreve, M.; Howard, S.; Jhoti, H.;
Murray, C. W.; Seavers, L. C. A.; van Montfort, R. L. M.
J. Med. Chem. 2006, 49, 1346).
13. The enzyme’s equilibrium binding affinity for DHB–AMP
does not appear to be known. Attempts to determine this
value are likely frustrated by the instability of aryl
adenylates under alkaline conditions where EntE is most
active Kellerman, G. M. J. Biol. Chem. 1958, 231, 427.
14. May, J. J.; Kessler, N.; Marahiel, M. A.; Stubbs, M. T.
Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 12120.
15. Cramer, F.; Neunhoeffer, H. Chem. Ber. 1962, 95, 1664.