Enterobactin protonation and iron release: Structural characterization of the salicylate coordination shift in ferric enterobactin

Article abstract of DOI:10.1021/ja062046j

The siderophore enterobactin (Ent) is produced by many species of enteric bacteria to mediate iron uptake. This iron scavenger can be reincorporated by the bacteria as the ferric complex [Fe^III(Ent)]^3- and is subsequently hydrolyzed by an esterase to facilitate intracellular iron release. Recent literature reports on altered protein recognition and binding of modified enterobactin increase the significance of understanding the structural features and solution chemistry of ferric enterobactin. The structure of the neutral protonated ferric enterobactin complex [Fe^III(H ₃Ent)]⁰ has been the source of some controversy and confusion in the literature. To demonstrate the proposed change of coordination from the tris-catecholate [Fe^III(Ent)]^3- to the tris-salicylate [Fe^III(H₃Ent)]⁰ upon protonation, the coordination chemistry of two new model compounds N,N′,N″-tris[2-(hydroxybenzoyl)carbonyl]cyclotriseryl trilactone (SERSAM) and N,N′,N″-tris[2-hydroxy,3-methoxy(benzoyl)carbonyl] cyclotriseryl trilactone (SER(3M)SAM) was examined in solution and solid state. Both SERSAM and SER(3M)SAM form tris-salicylate ferric complexes with spectroscopic and solution thermodynamic properties (with log β₁₁₀ values of 39 and 38 respectively) similar to those of [Fe^III(H₃Ent)]⁰. The fits of EXAFS spectra of the model ferric complexes and the two forms of ferric enterobactin provided bond distances and disorder factors in the metal coordination sphere for both coordination modes. The protonated [Fe^III(H₃Ent)] ⁰ complex (d_Fe-O = 1.98 A, σ² _stat(O) = 0.00351(10) A²) exhibits a shorter average Fe-O bond length but a much higher static Debye-Waller factor for the first oxygen shell than the catecholate [Fe^III(Ent)]^3- complex (d_Fe-O = 2.00 A, σ^2stat(O) = 0.00067(14) A²). ¹H NMR spectroscopy was used to monitor the amide bond rotation between the catecholate and salicylate geometries using the gallic complexes of enterobactin: [Ga^III(Ent)] ^3- and [Ga^III(H₃Ent)]⁰. The ferric salicylate complexes display quasi-reversible reduction potentials from -89 to -551 mV (relative to the normal hydrogen electrode NHE) which supports the feasibility of a low pH iron release mechanism facilitated by biological reductants.

Full text of DOI:10.1021/ja062046j

Published on Web 06/17/2006
Enterobactin Protonation and Iron Release: Structural
Characterization of the Salicylate Coordination Shift in Ferric
Enterobactin1
Rebecca J. Abergel,† Jeffrey A. Warner,‡ David K. Shuh,‡ and
Kenneth N. Raymond*,†,‡
Contribution from the Department of Chemistry, UniVersity of California, Berkeley, California
94720-1460, and the Chemical Sciences DiVision, Lawrence Berkeley National Laboratory,
Berkeley, California 94720
Received April 3, 2006; E-mail: raymond@socrates.berkeley.edu
Abstract: The siderophore enterobactin (Ent) is produced by many species of enteric bacteria to mediate
iron uptake. This iron scavenger can be reincorporated by the bacteria as the ferric complex [FeIII(Ent)]3-
and is subsequently hydrolyzed by an esterase to facilitate intracellular iron release. Recent literature reports
on altered protein recognition and binding of modified enterobactin increase the significance of understanding
the structural features and solution chemistry of ferric enterobactin. The structure of the neutral protonated
ferric enterobactin complex [FeIII(H3Ent)]0 has been the source of some controversy and confusion in the
literature. To demonstrate the proposed change of coordination from the tris-catecholate [FeIII(Ent)]3- to
the tris-salicylate [FeIII(H3Ent)]0 upon protonation, the coordination chemistry of two new model compounds
N,N′,N′′-tris[2-(hydroxybenzoyl)carbonyl]cyclotriseryl trilactone (SERSAM) and N,N′,N′′-tris[2-hydroxy,3-
methoxy(benzoyl)carbonyl]cyclotriseryl trilactone (SER(3M)SAM) was examined in solution and solid state.
Both SERSAM and SER(3M)SAM form tris-salicylate ferric complexes with spectroscopic and solution
thermodynamic properties (with log â110 values of 39 and 38 respectively) similar to those of [FeIII(H3Ent)]0.
The fits of EXAFS spectra of the model ferric complexes and the two forms of ferric enterobactin provided
bond distances and disorder factors in the metal coordination sphere for both coordination modes. The
protonated [FeIII(H3Ent)]0 complex (dFe-O ) 1.98 Å, σ2stat(O) ) 0.00351(10) Å2) exhibits a shorter average
Fe-O bond length but a much higher static Debye-Waller factor for the first oxygen shell than the
catecholate [FeIII(Ent)]3- complex (dFe-O ) 2.00 Å, σ2stat(O) ) 0.00067(14) Å2). 1H NMR spectroscopy was
used to monitor the amide bond rotation between the catecholate and salicylate geometries using the
gallic complexes of enterobactin: [GaIII(Ent)]3- and [GaIII(H3Ent)]0. The ferric salicylate complexes display
quasi-reversible reduction potentials from -89 to -551 mV (relative to the normal hydrogen electrode
NHE) which supports the feasibility of a low pH iron release mechanism facilitated by biological reductants.
Introduction
human immune system which sequesters and inactivates sid-
erophores, has heightened this significance.10 Functionalization
While enterobactin has been described as an archetype for
siderophore-mediated iron transport2 and much is known about
its selective affinity for Fe(III), microbial synthesis, and
recognition,3,4 recent research progress has opened new and
surprising aspects of this and related siderophores. Because the
growth of pathogenic bacteria depends on adequate iron supply,
iron metabolism is a key determinant in bacterial disease.5-9
The recent discovery of siderocalin, a protein produced by the
to increase siderophore water solubility alters the iron transport
capability in a host organism and also the pathogenicity of the
producing organism.11,12 Many species of Gram-negative enteric
bacteria such as Escherichia coli and Salmonella enterica
produce and utilize the siderophore enterobactin (Ent) (Figures
1 and 2).2 Enterobactin is predisposed for iron binding, with
(6) Dertz, E. A.; Raymond, K. N. In ComprehensiVe Coordination Chemistry-
II; McCleverty, J., Meyer, T., Eds.; Pergamon: Oxford, 2003; Vol. 8, pp
141-168.
† University of California, Berkeley.
(7) Hantke, K. Curr. Opin. Microbiol. 2001, 4, 172-177.
(8) Raymond, K. N.; Dertz, E. A. In Iron Transport in Bacteria: Molecular
Genetics, Biochemistry, Microbial Pathogenesis and Ecology; Crosa, J. H.,
Payne, S. M., Eds.; ASM Press: Washington, DC, 2004; pp 3-17.
(9) Raymond, K. N.; Telford, J. R. In Bioinorganic Chemistry An Inorganic
PerspectiVe of Life; Kessissoglou, D. P., Ed.; Kluwer Academic Publish-
ers: The Netherlands, 1995; Vol. 459, pp 25-37.
‡ Lawrence Berkeley National Laboratory.
(1) Coordination Chemistry of Microbial Iron Transport. 75. Part 74: Dertz,
E. A.; Raymond, K. N. Inorg. Chem. 2006, 128. In press.
(2) Raymond, K. N.; Dertz, E. A.; Kim, S. S. Proc. Natl. Acad. Sci. U.S.A.
2003, 100, 3584-3588.
(3) Earhart, C. F. In Iron Transport in Bacteria: Molecular Genetics,
Biochemistry, Microbial Pathogenesis and Ecology; Crosa, J. H., Payne,
S. M., Eds.; ASM Press: Washington, DC, 2004; pp 133-146.
(4) Walsh, C. T.; Marshall, C. G. In Iron Transport in Bacteria: Molecular
Genetics, Biochemistry, Microbial Pathogenesis and Ecology; Crosa, J. H.,
Payne, S. M., Eds.; ASM Press: Washington, DC, 2004; pp 18-37.
(5) Boukhalfa, H.; Crumbliss, A. L. BioMetals 2002, 15, 325-339.
(10) Goetz, D. H.; Holmes, M. A.; Borregaard, N.; Bluhm, M. E.; Raymond,
K. N.; Strong, R. K. Mol. Cell 2002, 10, 1033-1043.
(11) Fischbach, M. A.; Lin, H.; Liu, D. R.; Walsh, C. T. Nat. Chem. Biol. 2006,
2, 132-138.
(12) Luo, M.; Lin, H.; Fischbach, M. A.; Liu, D. R.; Walsh, C. T.; Groves, J.
T. ACS Chem. Biol. 2006, 1, 29-32.
9
8920
J. AM. CHEM. SOC. 2006, 128, 8920-8931
10.1021/ja062046j CCC: $33.50 © 2006 American Chemical Society

Salicylate Coordination Shift in Ferric Enterobactin
A R T I C L E S
Despite thorough literature precedent,21-24 confusion about
coordination geometry at the metal center has persisted.25
Previous studies suggested a structural change for the ferric
enterobactin complex from a catecholate to salicylate geometry
around the metal ion, where protonation occurs at the meta
hydroxyl oxygen of the catechol, and coordination of Fe(III)
shifts from the two catecholate oxygens to the ortho hydroxyl
oxygen and the amide oxygen (Figure 1).23,24 The work
presented here confirms the solution and solid-state structural
alteration of ferric enterobactin upon protonation. Two analogues
of enterobactin based on the same serine trilactone scaffold,
SERSAM and SER(3M)SAM, reproduce the salicylate coordi-
nation environment of the protonated ferric enterobactin complex
(Figure 2). A variety of thermodynamic and spectroscopic
measurements (UV-vis, FTIR, NMR, EXAFS) of the catecho-
late [FeIII(Ent)]3-, [GaIII(Ent)]3-, [FeIII(TRENCAM)]3- com-
plexes26 and the salicylate analogues [FeIII(SERSAM)]0, [FeIII-
(SER(3M)SAM)]0, [FeIII(TRENSAM)]0, and [FeIII(TREN(3M)-
SAM)]0,23 structurally characterize both coordination modes.
Corresponding data as well as DFT calculated structures were
obtained for the two protonated [FeIII(H3Ent)]0 and [GaIII(H3-
Ent)]0 complexes, establishing their common salicylate geom-
etry.
Figure 1. Ferric enterobactin (top) and the coordination shift from
catecholate (bottom left) to salicylate (bottom right) upon protonation.
three catecholate units attached to a serine trilactone scaffold
through amide linkages, and exhibits a very high affinity for
ferric ion (pFe ) 34.3).13 The metabolism of this powerful iron
chelator is tightly controlled by the Fur (Fe uptake regulation)
protein.7 Much is known about the biosynthesis and TolC-
dependent exportation of the apo-siderophore as well as the
TonB-dependent FepA recognition of the enterobactin ferric
complex.3,4 The iron release mechanism from the ferric entero-
bactin complex to intracellular iron carriers occurs primarily
through enzymatic hydrolysis of the siderophore trilactone
backbone by the esterase Fes.14,15 However, synthetic analogues
of enterobactin, which are not susceptible to this specific
hydrolysis, can be 5% as effective as enterobactin in delivering
iron to the cell, and thus sufficient for growth promotion.16,17
This result implies the existence of a secondary pathway for
intracellular iron release. While ferric enterobactin at neutral
pH cannot release iron by reduction,18,19 protonation of ferric
enterobactin makes reduction much easier.20 Furthermore, low
pH binding of ferric enterobactin raises the question of whether
a second type of structure might be recognized by proteins
involved in ferric enterobactin metabolism. Studies in progress
are investigating this issue.
Results
Ligand and Metal Complex Synthesis. The two new
enterobactin analogues SERSAM and SER(3M)SAM were
synthesized by using identical methods, as shown in Scheme
1. The serine trilactone (1) was reacted with an excess of either
2-(benzyloxy)benzoyl chloride (2) or 2-(benzyloxy),3-(meth-
oxy)benzoyl chloride (3), and triethylamine in THF. The crude
tribenzyl-SERSAM (4) and tribenzyl-SER(3M)SAM (5) were
purified by separation on chromatography columns. Subsequent
hydrogenation over Pd/C under hydrogen pressure (1 atm)
provided pure SERSAM (6) and SER(3M)SAM (7) in good
yields. The ligands were both combined with ferric chloride in
methanol, with pyridine as a base, to provide the corresponding
neutral ferric complexes, [FeIII(SERSAM)]0 and [FeIII(SER(3M)-
SAM)]0. Enterobactin (Ent) was also combined with Fe(acac)3
or Ga(acac)3 in methanol to afford the neutral protonated ferric
and gallic complexes, [FeIII(H3Ent)]0 and [GaIII(H3Ent)]0, re-
spectively. Addition of KOH in the same reaction conditions
allowed the isolation of the deprotonated catecholate ferric
([FeIII(Ent)]3-) and gallic ([GaIII(Ent)]3-) enterobactin com-
plexes. Electrospray mass spectrometry (ESMS) showed the
molecular peaks of the mononuclear 1:1 ligand to metal
complexes for all complexes. No peaks corresponding to
complexes having a 2:1 ligand to metal stoichiometries were
observed.
It has been shown that protonation of ferric enterobactin
occurs in three discrete one-proton steps upon acidification,
resulting eventually in a triprotonated, neutral ferric complex.21,22
(13) Loomis, L. D.; Raymond, K. N. Inorg. Chem. 1991, 30, 906-911.
(14) Brickman, T. J.; McIntosh, M. A. J. Biol. Chem. 1992, 267, 12350-12355.
(15) Lin, H.; Fischbach, M. A.; Liu, D. R.; Walsh, C. T. J. Am. Chem. Soc.
2005, 127, 11075-11084.
FTIR and UV-Vis Spectroscopic Properties. The solid-
state infrared spectral data for SERSAM and SER(3M)SAM
exhibit a shift of the carbonyl amide stretching frequencies to
lower wavenumbers upon complexation with Fe3+: the carbonyl
stretch of the free ligands SERSAM (1641 cm-1) and SER-
(16) Ecker, D. J.; Matzanke, B. F.; Raymond, K. N. J. Bacteriol. 1986, 167,
666-673.
(17) Heidinger, S.; Braun, V.; Pecoraro, V. L.; Raymond, K. N. J. Bacteriol.
1983, 153, 109-115.
(18) Cooper, S. R.; McArdle, J. V.; Raymond, K. N. Proc. Natl. Acad. Sci.
U.S.A. 1978, 75, 3551-3554.
(22) Pecoraro, V. L.; Harris, W. R.; Wong, G. B.; Carrano, C. J.; Raymond, K.
N. J. Am. Chem. Soc. 1983, 105, 4623-4633.
(19) 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.
(23) Cohen, S. M.; Meyer, M.; Raymond, K. N. J. Am. Chem. Soc. 1998, 120,
6277-6286.
(20) Lee, C. W.; Ecker, D. J.; Raymond, K. N. J. Am. Chem. Soc. 1985, 107,
6920-6923.
(21) Cass, M. E.; Garrett, T. M.; Raymond, K. N. J. Am. Chem. Soc. 1989,
111, 1677-1682.
(24) Cohen, S. M.; Raymond, K. N. Inorg. Chem. 2000, 39, 3624-3631.
(25) Ratledge, C.; Dover, L. G. Annu. ReV. Microbiol. 2000, 54, 881-941.
(26) Stack, T. D. P.; Karpishin, T. B.; Raymond, K. N. J. Am. Chem. Soc. 1992,
114, 1512-1514.
9
J. AM. CHEM. SOC. VOL. 128, NO. 27, 2006 8921

A R T I C L E S
Abergel et al.
Figure 2. Enterobactin and its synthetic analogues: the catecholate TRENCAM and salicylate SERSAM, SER(3M)SAM, TRENSAM, and TREN(3M)-
SAM ligands. The coordinating oxygen atoms are indicated in red.
Scheme 1. Synthesis of the Two Salicylate Analogues of Enterobactin SERSAM and SER(3M)SAM
(3M)SAM (1642 cm-1) shift to lower energy by 35 and 38 cm-1
for their corresponding ferric complexes. These observations
are in agreement with those reported for similar salicylate
complexes and support the assignment of the salicylate mode
of coordination.23 The metal enterobactin complexes also display
characteristic IR frequencies, depending on their protonation
states, as detailed in a previous study.22 Similar infrared spectra
are observed for the corresponding ferric or gallic complexes.
In aqueous media, [FeIII(SERSAM)]0 and [FeIII(SER(3M)-
SAM)]0 both show a strong absorption in the UV due to π-π*
transitions at 304 nm (ꢀ ) 9520 M-1 cm-1) and 312 nm (ꢀ )
9300 M-1 cm-1), respectively. However, [FeIII(SERSAM)]0
exhibits a deep red color resulting from ligand-to-metal-charge-
transfer (LMCT) transitions at λ ) 446 nm (ꢀ ) 3270 M-1
cm-1), which does not correlate well with that of [FeIII(H3Ent)]0.
The presence of the methoxy groups in the deep purple [FeIII-
(SER(3M)SAM)]0 (λ ) 502 nm, ꢀ ) 3270 M-1 cm-1) provides
a better spectroscopic model for [FeIII(H3Ent)]0.13
Ligand Protonation Constants and Stability of Ferric
Salicylate Complexes. The stepwise proton association con-
stants as defined by eqs 1, 2, and 3 and the ferric complex
formation constant (eq 4) were determined by spectrophoto-
metric titrations for the two new salicylate ligands. Due to
limited solubility in water, SERSAM and SER(3M)SAM were
9
8922 J. AM. CHEM. SOC. VOL. 128, NO. 27, 2006

Salicylate Coordination Shift in Ferric Enterobactin
A R T I C L E S
[HnL]
[Hn-1L][H+]
Hn-1L + H+ T HnL K0ln
L + hH+ T HhL âolh
)
)
(1)
[HhL]
[L][H+]h
(2)
(3)
[HnL]
â0ln
K0ln
)
)
[Hn-1L][H+]
â0l(n-1)
[MmLlHh]
[M]m[L]l[H+]h
mM + lL + hH+ T MmLlHh âmlh
)
(4)
Spectrophotometric titrations, monitoring wavelengths be-
tween 250 and 700 nm, were also performed for ferric SERSAM
and ferric SER(3M)SAM from pH 5.5 to 2 and are shown in
Figures 3 and S1. Increasing the pH above 7 resulted in the
complete disappearance of the charge-transfer band due to
formation of iron hydroxide. The complexes are stable only in
neutral to acidic conditions. As the pH is lowered, the intensity
of the LMCT band decreases and finally vanishes around pH
) 2. At low pH, the intense π-π* transitions resolve as a
narrower band (λ ) 300 nm, ꢀ ) 5025 M-1 cm-1 and λ ) 314
nm, ꢀ ) 7018 M-1 cm-1 for SERSAM and SER(3M)SAM,
respectively), the final spectrum being identical to the spectrum
calculated for LH3 species. The nonlinear least-squares refine-
ments27 of the overall formation constants â110 included in each
case the three protonation constants derived from spectropho-
tometric titrations (Table 1) and the metal hydrolysis products,
whose equilibrium constants were fixed to the literature values28
(log â10-1 ) -2.61, log â10-2 ) -5.66, log â20-2 ) 2.86) and
which do not absorb significantly.
Figure 3. Spectrophotometric titrations of SER(3M)SAM by KOH (top)
and [FeIII(SER(3M)SAM)]0 by HCl (bottom) in water. I ) 0.1 (KCl), T )
25.0 °C, l ) 1 cm.
Table 1. Refined Protonation and Ferric Complex Formation
DFT Calculations on Catecholate and Salicylate Entero-
bactin Metal Complexes. The relative conformational stability
of the two most probable binding modes of protonated ferric
enterobactin [FeIII(H3Ent)]0 was investigated by molecular
modeling. Density functional theory calculations29 were per-
Constants for SERSAM and SER(3M)SAMa
log âmlh
species
m l h
SERSAM
SER(3M)SAM
LH
0 1 1
0 1 2
0 1 3
1 1 0
9.1 (1)
17.5 (2)
23.3 (2)
39 (3)
8.8 (1)
16.7 (1)
23.3 (2)
38 (2)
formed on the ferric and gallic complexes [FeIII(Ent)]3-
,
LH2
LH3
FeL
[GaIII(Ent)]3-, cat-[FeIII(H3Ent)]0, cat-[GaIII(H3Ent)]0, sal-[FeIII(H3-
Ent)]0, and sal-[GaIII(H3Ent)]0 where the prefixes “cat” and “sal”
designate the catecholate and salicylate possible binding modes.
The geometry of each complex was optimized with Jaguar at
the b3lyp/lacvp** level. The complete d-shell of GaIII provides
complexes with a lower spin-state than that of the equivalent
ferric complexes, which facilitates the convergence of the
calculation. In the FeIII case, iguess ) 10 and iacscf ) 2 values
were added to the default parameters to satisfy the SCF
convergence criteria. Frequency calculations, which were
performed on all structures, verified that the optimized geom-
etries were minima (no negative frequencies) on the potential
energy surface. For both metals, the same trends were observed.
The results indicate that the protonated salicylate binding mode
is energetically favored over the protonated catecholate mode
with a heat of formation difference of 10 kcal‚mol-1 (24
kcal‚mol-1) in the ferric (gallic respectively) complexes;
geometric parameters and relative energies are summarized in
Tables 2, S2, and S3. As expected,30 the calculated M-O bond
a I ) 0.1 (KCl), T ) 25 °C. Figures in parentheses give the uncertainty
determined from the standard deviation between three independent titrations.
titrated at low concentration by monitoring the spectral changes
as a function of pH throughout the pH range of 4.5-10.0. The
compounds were titrated only from low to high pH to avoid
base-catalyzed hydrolysis of the backbone.19 In both cases, the
intense UV absorption band of the salicylate functionality
increases in intensity and shifts to longer wavelengths as the
pH is increased to 10 (Figures 3 and S1). Factor analysis27 of
the titration curves (each data set comprising about 60 spectra
and 151 wavelengths for each replicate) indicated for each ligand
three sequential protonation equilibria corresponding to log Ka
values reported in Table 1. The best nonlinear least-squares
refinements were obtained with the models including four
species: L3-, LH2-, LH2-, and LH3. The refined constants are
comparable to the values previously reported for the stepwise
protonation of the tripodal TREN-based analogues TRENSAM
and TREN(3M)SAM.23
(28) Perrin, D. D.; Dempsey, B.; Hall, C. a., Ed. London, 1974.
(29) Jaguar 5.5, 5.0.1 ed.; Schrodinger, L. L. C.: Portland, OR, 2003.
(30) Meyer, M.; Trowitzsch-Kienast, W. Res. DeVel. Phys. Chem. 2004, 7, 127-
149.
(27) Gans, P.; Sabatini, A.; Vacca, A. Ann. Chim. (Rome) 1999, 89, 45-49.
9
J. AM. CHEM. SOC. VOL. 128, NO. 27, 2006 8923

A R T I C L E S
Abergel et al.
Table 2. Calculated Heats of Formation, Atomic Distances, and Bond Lengths for Different Binding Modes of Ferric Enterobactin
∆H
∆H
Fe
−Oortho
Fe
−Ometa
Fe
−Ocarbonyl
average Fe−O
complex
(hartrees)
(kcal/mol)
distance (Å)
distance (Å)
distance (Å)
length (Å)
[FeIII(Ent)]3-
-2571.497
-1613640
2.100
2.097
2.094
1.891
1.892
1.892
1.936
1.933
1.935
2.005
2.007
2.010
2.275
2.271
2.275
4.477
4.475
4.478
6.283
6.280
6.271
6.022
6.024
6.027
2.103
2.100
2.100
2.052
cat-[FeIII(H3Ent)]0
sal-[FeIII(H3Ent)]0
-2573.279
-2573.295
-1614759
-1614769
2.083
2.018
Figure 4. Calculated frontier molecular orbital energy level diagrams for [FeIII(Ent)]3- (left) and sal-[FeIII(H3Ent)]0 (right).
lengths for [GaIII(Ent)]3- are substantially shorter than for
[FeIII(Ent)]3-. In both cases, the distances between the catechol
oxygen atoms and the metal ion are inequivalent. The distance
between metal and the ortho catechol oxygen atom Oortho is
longer than the corresponding distance of the meta catechol
oxygen atom Ometa, which is consistent with the parameters
obtained from the [VIV(Ent)]2- X-ray structure.31 The M-O
bond lengths of the protonated models also vary significantly
from the deprotonated catecholate model; the average M-O
bond distance is shorter for the salicylate (and longer for the
catecholate) protonated complex [MIII(H3Ent)]0 than for
[MIII(Ent)]3-. The assignment of the frontier orbitals, for both
[FeIII(Ent)]3- and sal-[FeIII(H3Ent)]0, are consistent with D3
symmetric high-spin d5 iron complexes (Figure 4), with d-orbital
energy levels that are significantly lower for sal-[FeIII(H3Ent)]0
than for [FeIII(Ent)]3-. It has been shown previously that the
broad LMCT band in the absorption spectra of ferric tris-
catecholate complexes is composed of two overlapping, x,y-
polarized, one-electron transitions: a2 f ea and eπ1 f ea.32 The
calculated energy difference between the ea orbitals of sal-
[FeIII(H3Ent)]0 and [FeIII(Ent)]3- is ∆Ee ) 9.3 eV. This
a
a
corresponds to ∆λe ) 141 nm, causing the color change from
deep red to violet upon protonation of ferric enterobactin.13 Last,
the LUMO orbital of the sal-[FeIII(H3Ent)]0 model is primarily
(31) Karpishin, T. B.; Raymond, K. N. Angew. Chem., Int. Ed. Engl. 1992, 31,
466-468.
(32) Karpishin, T. B.; Gebhard, M. S.; Solomon, E. I.; Raymond, K. N. J. Am.
Chem. Soc. 1991, 113, 2977-2984.
9
8924 J. AM. CHEM. SOC. VOL. 128, NO. 27, 2006

Salicylate Coordination Shift in Ferric Enterobactin
A R T I C L E S
from the ligand’s electron density in proximity of the coordina-
tion sphere, and is 8.1 eV lower in energy as compared to that
of the LUMO of [FeIII(Ent)]3-, which displays electron density
only in the trilactone region.
EXAFS Spectroscopic Analysis. Samples of protonated
[FeIII(H3Ent)]0 and deprotonated [FeIII(Ent)]3- as well as the
salicylate ferric complexes [FeIII(SERSAM)]0 and [FeIII(SER-
(3M)SAM)]0 were subjected to temperature-dependent (30, 100,
200, and 300 K) Fe K-edge extended X-ray absorption fine
structure measurements. The ferric complexes of TRENSAM,
TREN(3M)SAM, and TRENCAM (Figure 2) were used as
model compounds for the salicylate and catecholate binding
modes, respectively.23,26 The XANES spectra of the compounds
were identical, as expected from the similar coordination
environments of the ferric ion in all complexes. Differences in
the EXAFS spectra were expected to be small since there is
only a slight change in bond distance between the iron-catechol
oxygen and the iron-amide oxygen (0.03-0.13 Å). In fact, in
all complexes this distance is below the resolution limit of
EXAFS (1/∆k), where ∆k is the usable data k-range. The
temperature-dependent EXAFS data, however, allow the deter-
mination of static Debye-Waller factors, which show differ-
ences in the positional disorder of the catechol and amide
oxygens, and are representative of the catecholate and salicylate
coordinations. EXAFS fits were extended to the second carbon
shell of the ferric complexes. Although good quality data
extended to at least k ) 12.0 Å-1 in all cases, a limited k-range
was Fourier transformed (2.5-9.0 Å-1) because of temperature-
dependent changes in the spectra occurring in the region of 10
Å-1. These changes did not seem to affect the fits for the first
two shells and cannot be attributed to spin-crossover effects33
since Mo¨ssbauer and EPR experiments have confirmed the high-
spin state of the ferric enterobactin and analogues complexes
from 4 K to room temperature.34,35 Due to the short k-range
employed, there were only a few degrees of freedom available
(11 in most cases) for varying the fitting parameters. The
analysis focused on fitting the first two shells and relied on
parameters obtained from crystal structures of the ferric
complexes of the model compounds, [FeIII(TRENSAM)]0, [FeIII-
Figure 5. k-Space spectra (left), Fourier transforms (right, solid lines) and
r-space fits (right, dashed lines) of [FeIII(Ent)]3- at 30, 100, 200, and 300
K.
modes.36,37 For atoms positioned further than 3 Å away from
the iron center, the competition between noise and back/multiple
scattering contributions becomes non-negligible;38 therefore,
only the first oxygen-coordination shell and the second carbon-
coordination shell were studied. Analysis of the temperature
dependence of the mean-square relative displacement (MSRD)
2
allowed the determination of the static Debye-Waller (σstat
)
component, by elimination of the temperature-dependent thermal
disorder Debye-Waller component (σvib2) (eq 5).39
p2
2ΘEkBµ
ΘE
pωE
σvib2(T) )
coth
,
ΘE )
(5)
(TREN(3M)SAM)]0, and [FeIII(TRENCAM)]3- 23,26
The final
.
( )
2T
kB
parameters were determined on the basis of low residuals and
robustness of the fits. Special attention was paid to parameter
correlation, specifically (E0, R, C3) and (A, σ2, CN) which are
highly correlated among each other. In general, these correlations
were avoided by fixing parameter values iteratively and using
values established from the model compounds. In the same
manner, the robustness of each parameter was established. The
best EXAFS fits for the three model compounds were obtained
at 200 K, and the experimental parameters were in very good
agreement with those determined by X-ray diffraction at 172
K.23,26 Multiple temperature data were fit using an Einstein
model, which considers the pair of absorber and backscatterer
atoms as an independent oscillator with frequency ωE, and is
particularly suited for monitoring intramolecular vibrational
The Einstein characteristic temperature ΘE was established
for each compound, indicative of the overall Debye-Waller
factor for the shell as a function of temperature, and of the
effective bond-stretching force constant. The k-space and r-space
spectra are shown in Figure 5 and S2-4. Results of fits to the
data are summarized in Table 3. The temperature-averaged
distances between the iron center and the first-shell oxygens
and the second-shell carbons display consistent trends for
catecholate vs salicylate coordination. The salicylate [FeIII-
(SERSAM)]0, [FeIII(SER(3M)SAM)]0, [FeIII(TRENSAM)]0, and
[FeIII(TREN(3M)SAM)]0 complexes have shorter Fe-O bond
distances and longer Fe-C distances than their respective
catecholate analogues, [FeIII(Ent)]3- and [FeIII(TRENCAM)]3-
.
(33) Paulsen, H.; Grunsteudel, W.; Meyer-Klaucke, W.; Gerdan, M.; Grunsteudel,
H. F.; Chumakov, A. I.; Ruffer, R.; Winkler, H.; Toftlund, H.; Trautwein,
A. X. Eur. Phys. J. B 2001, 23, 463-472.
(36) Sevillano, E.; Meuth, H.; Rehr, J. J. Phys. ReV. B 1979, 20, 4908-4911.
(37) Stern, E. A. X-ray Absorption; Wiley: New York, 1988.
(38) Scherk, C. G.; Ostermann, A.; Achterhold, K.; Iakovleva, O.; Nazikkol,
C.; Krebs, B.; Knapp, E. W.; Meyer-Klaucke, W.; Parak, F. G. Eur. Biophys.
J. 2001, 30, 393-403.
(39) Boyce, J. B.; Bridges, F.; Claeson, T.; Nygren, M. Phys. ReV. B 1989, 39,
6555-6566.
(34) Pecoraro, V. L.; Wong, G. B.; Kent, T. A.; Raymond, K. N. J. Am. Chem.
Soc. 1983, 105, 4617-4623.
(35) Spartalian, K.; Oosterhuis, W. T.; Neilands, J. B. J. Chem. Phys. 1975, 62,
3538-3543.
9
J. AM. CHEM. SOC. VOL. 128, NO. 27, 2006 8925

A R T I C L E S
Abergel et al.
Table 3. EXAFS Averaged Fit Results for the First-Shell Oxygens and Second-Shell Carbons at All Temperaturesa
complex
shell
CN
d (Å)
σ
2 (Å2)
σ
2 static (Å2)
θE (K)
[FeIII(Ent)]3-
Fe/O
Fe/C
Fe/O
Fe/C
Fe/O
Fe/C
Fe/O
Fe/C
Fe/O
Fe/C
Fe/O
Fe/C
Fe/O
Fe/C
6.0
6.0
6.0
6.0
6.0
6.0
6.0
6.0
6.0
6.0
6.0
6.0
6.0
6.0
2.00
2.82
1.98
2.83
2.00
2.91
1.96
2.94
1.99
2.79
1.97
2.95
1.96
2.94
0.0053 ( 0.0008
0.0052 ( 0.0009
0.0076 ( 0.0006
0.0049 ( 0.0009
0.0107 ( 0.0015
0.0042 ( 0.0007
0.0103 ( 0.0009
0.0097 ( 0.0001
0.0025 ( 0.0008
0.0049 ( 0.0006
0.0086 ( 0.0012
0.0061 ( 0.0002
0.0169 ( 0.0004
0.0059 ( 0.0009
0.00067 ( 0.00014
0.00044 ( 0.00047
0.00351 ( 0.00010
0.00064 ( 0.00024
0.00597 ( 0.00039
0.00225 ( 0.00065
0.00628 ( 0.00026
0.00397 ( 0.00072
0.00213 ( 0.00013
0.00051 ( 0.00027
0.00366 ( 0.00021
0.00067 ( 0.00019
0.00636 ( 0.00017
0.00021 ( 0.00032
490 ( 10
566 ( 42
548 ( 10
523 ( 17
505 ( 32
473 ( 27
559 ( 28
508 ( 47
497 ( 10
612 ( 29
476 ( 15
450 ( 9
[FeIII(H3ent)]0
[FeIII(SERSAM)]0
[FeIII(SER(3M)SAM)]0
[FeIII(TRENCAM)]3-
[FeIII(TRENSAM)]0
[FeIII(TREN(3M)SAM)]0
515 ( 15
472 ( 17
a CN is the coordination number, d is the interatomic Fe-O and Fe-C distance ((0.01 Å). Other parameters established from the analysis were for:
[FeIII(TRENCAM)]3-, S0 ) 0.77 and ∆E0 ) -3.47; [FeIII(TRENSAM)]0, S0 ) 0.86 and ∆E0 ) -3.60; [FeIII(TREN(3M)SAM)]0, S0 ) 0.87 and ∆E0 )
2
2
2
2
2
2
-1.46; [FeIII(Ent)]3-, S0 ) 0.87 and ∆E0 ) -3.63; [FeIII(H3Ent)]0, S0 ) 0.86 and ∆E0 ) -3.76, [FeIII(SERSAM)]0, S0 ) 0.87 and ∆E0 ) -4.19;
2
[FeIII(SER(3M)SAM)]0, S0 ) 0.87 and ∆E0 ) -2.25.
Figure 6. Einstein model fits of first-shell oxygens for: [FeIII(TREN(3M)-
SAM)]0, [FeIII(SERSAM)]0, [FeIII(SER(3M)SAM)]0, [FeIII(TRENSAM)]0,
[FeIII(H3Ent)]0, [FeIII(Ent)]3-, and [FeIII(TRENCAM)]3- from top to bottom.
The triprotonated species [FeIII(H3Ent)]0 also exhibits shorter
Fe-O and longer Fe-C bonds than [FeIII(Ent)]3-, consistent
with conversion to salicylate coordination. The static Debye-
Waller factor (σstat2) was determined from Einstein model fits
to the temperature-dependent EXAFS data (Figure 6). The
salicylate compounds [FeIII(SERSAM)]0, [FeIII(SER(3M)-
SAM)]0, [FeIII(TRENSAM)]0, and [FeIII(TREN(3M)SAM)]0
show a similar pattern in that the oxygen σstat2 values (and to a
lesser extent the carbon σstat2 values) are much larger than those
Figure 7. Variation of the 1H NMR chemical shift as a function of pD for
gallic enterobactin (symbols). Solid lines represent the fitted data. Dashed
lines designate the pKa values of the complex.
observed for the catecholate compounds [FeIII(Ent)]3- and
[FeIII(TRENCAM)]3-, indicating a larger positional disorder in
these shells. In the same manner, the σstat2 values are larger for
[FeIII(H3Ent)]0 than for [FeIII(Ent)]3-. Furthermore, the Einstein
characteristic temperatures ΘE are greater for the oxygen shell
than for the carbon shell for all salicylate compounds, whereas
the opposite trend is seen for the catecholate compounds.
Structural Characterization of Gallic Enterobactin by
NMR. Previous spectroscopic studies have shown that the
gallium complex of enterobactin exhibits the same behavior
upon acidification as ferric enterobactin.13 To provide more
information on the coordination shift, the protonation steps of
Acidification of the complex resulted in drastic changes in the
chemical shifts of the various protons (Figure 7). Least-squares
refinements of the chemical shifts variation produced pKa values
similar to those obtained spectrophotometrically (pKa1 ) 5.05-
(1), pKa2 ) 4.19(1), pKa3 ) 2.93(1)).13 More importantly, the
coordination shift happens during the third protonation of the
complex. Indeed, a major upfield shift followed by an inflection
point at pD ≈ 3.3 is observed for the protons Ha and Hc in the
proximity of the amide carbonyl oxygens, at low pD values
(below the second pKa). As determined with DFT calculations,
both of these sets of protons should point toward the periphery
1
1
this diamagnetic complex were followed by H NMR. The H
NMR titration was conducted from pD ) 9.6 to 2.4 in D2O.
9
8926 J. AM. CHEM. SOC. VOL. 128, NO. 27, 2006

Salicylate Coordination Shift in Ferric Enterobactin
A R T I C L E S
Figure 8. 2D 1H NMR NOESY spectrum of [GaIII(Ent)]3- (0.1 M) in
DMSO-d6, τm ) 0.7 s. Selected NOEs are indicated in red.
of the molecule, in the same direction as the carbonyl oxygens
in the case of [GaIII(Ent)]3- and opposite that direction for
[GaIII(H3Ent)]0, and should be sensitive to a rotation of the amide
bond. Precipitation of the protonated gallic enterobactin occurred
when the pD value reached 2.4. The precipitate was filtered,
rinsed with water, dried, and dissolved in DMSO-d6 to be
1
identified as the complex [GaIII(H3Ent)]0, using H and 13C
NMR. The 13C NMR chemical shifts in DMSO-d6 are also
indicative of a salicylate shift. Indeed, as the catecholate
complex [GaIII(Ent)]3- is protonated to [GaIII(H3Ent)]0, the
amide carbonyl peak is shifted downfield from 168.7 to 180.1
ppm, whereas the benzoyl meta carbon peak displays an upfield
shift of 9.1 ppm.40 Furthermore, 2D-gradient NOESY experi-
ments were performed in DMSO-d6 on the two gallic complexes
[GaIII(Ent)]3- and [GaIII(H3Ent)]0 (Figure 8). Spectra were
collected at a variety of mixing times (τm from 0.15 to 1 s) to
observe the growth of the NOE signals. The rate of the NOE
cross-peak intensity growth is inversely proportional to the sixth
power of the distance between nuclei (Figure S5).41 As
suggested by the DFT calculated structures, the amide bond
rotates upon acidification, which causes the intensities of the
cross-peaks between the amide proton and the benzoyl ortho
proton or one of the seryl â-protons to increase significantly.
These correlations are highlighted in Figure 8.
Figure 9. Cyclic voltammogram of [FeIII(TREN(3M)SAM]0 at 50 mV/s
(top), reduction wave for [FeIII(SER(3M)SAM]0 at different scan rates: 2000
to 25 mV/s for spectra 1 to 8 (middle), sweep rate dependence of the
reduction current peak for all complexes (bottom).
ferrocene as a standard (Figure 9). Quasi-reversible signals
centered at -991, -919, -797, and -529 mV/0.01 M AgNO3
were found for the iron complexes of TRENSAM, TREN(3M)-
SAM, SERSAM, and SER(3M)SAM, respectively. In all cases,
the peak separation between the cathodic and the anodic waves
is independent of the sweep rate in the range 50-200 mV/s,
and the peak current Ired is a linear function of the square root
of the sweep rate, indicating diffusion-limited reduction pro-
Electrochemical Characterization of Ferric Salicylate
Complexes. Due to the low solubility of these neutral complexes
in water, the cyclic voltammograms of the salicylate analogues
of ferric enterobactin were recorded in acetonitrile, using
cesses (Figure 9). The half-wave potential vs NHE in water can
1/2
be estimated by eq 6;42 ENHE (H2O), E1/2 (S), ∆Gtr, and aAg
+
+
Ag /Ag
stand for the estimated reduction potential (vs NHE) in water,
the measured reduction potential vs the Pleskow electrode in
(40) Llinas, M.; Wilson, D. M.; Neilands, J. B. Biochemistry 1973, 12, 3836-
3843.
(41) Kumar, A.; Wagner, G.; Ernst, R. R.; Wutrich, K. J. Am. Chem. Soc. 1981,
103, 3654-3658.
(42) Gagne, R. R.; Koval, C. A.; Lisensky, G. C. Inorg. Chem. 1980, 19, 2854-
2855.
9
J. AM. CHEM. SOC. VOL. 128, NO. 27, 2006 8927

A R T I C L E S
Abergel et al.
Table 4. Reduction Potential Values for Relevant Natural Siderophores and Synthetic Analogues
/
/
/
complex
E1 2 (mV/NHE)
complex
E1 2 (mV/NHE)
complex
E1 2 (mV/NHE)
[Fe(TRENSAM)]
[Fe(SERSAM)]
[Fe(TRENHOPO)]
[Fe(hopobactin)]
-551
[Fe(TREN(3M)SAM)]
[Fe(SER(3M)SAM)]
[Fe(ferrichrome A)]
[Fe(ferrioxamine B)]
-450
-89
[Fe(TRENCAM)]
[Fe(Ent)]
-104044
-99044
-357
-43545
-34245
-44618
-45418
Table 5. Thermodynamic Parameters of Enterobactin
the organic solvent, the free energy change of transfer of Ag+
from water to the organic solvent, and Ag+ ion activity in the
reference compartment, respectively. The ∆Gtr values (-23.2
kJ mol-1 for acetonitrile) have been tabulated by IUPAC.43
log K
H3Ent f Ent + 3H+
-22.0513
4913
Fe3+ + Ent f [FeIII(Ent)]3-
[FeIII(Ent)]3- + 3H+ f [FeIII(H3Ent)]0
Fe3+ + H3Ent f [FeIII(H3Ent)]0
10.9713
37.92
1/2
1/2
Ag+/Ag
E
(H2O) ) E
(S) + E0NHE(Ag+, H2O) -
NHE
{∆Gtr + RT ln aAg+}/nF ) E1A/g2 /Ag(S) + 440 mV (6)
+
in similar conditions. In both cases, the salicylate binding mode
was found to be substantially lower in energy than the
catecholate coordination for triprotonated complexes. The
calculated d-orbital energy levels were consistent with the
LMCT energy shift observed in the absorption spectra of
[FeIII(Ent)]3- and its protonated form.13 Moreover, as established
by the EXAFS and NMR analyses of the respective ferric and
gallic complexes, the average metal-oxygen bond lengths
shorten upon protonation; this trend is consistent with the
calculated parameters of the protonated salicylate models as
opposed to the elongation of bonds found for the protonated
catecholate models. The EXAFS experimental results showed
a higher σstat2(O) parameter value for salicylate complexes than
for the corresponding catecholate ones, describing an environ-
ment with a higher static disorder in the first coordination shell
of oxygen around the metal center. Indeed, for the salicylate-
bound compounds, the coordinating oxygen atoms are no longer
all phenolate oxygens, three of them are carbonyl oxygens.
Another set of key parameters is noticeably different in both
coordination modes: in the catecholate coordination, the Ein-
stein characteristic temperature ΘE is greater for the carbon shell
than for the oxygen shell, as opposed to the salicylate binding
mode. The metal-oxygen bond strengths are comparable for
both coordination modes, whereas the deformation vectors
corresponding to the metal-carbon shell have a larger contribu-
tion in the salicylate geometry. This causes the metal-carbon
force constants to be weaker in the salicylate case and therefore
affects the relative value of the carbon shell Einstein temper-
ature. NMR characterization of the gallic complexes of entero-
bactin also confirmed the hypothesis of a switch to salicylate
coordination upon protonation. Acidification of the catecholate
[GaIII(Ent)]3- complex results in rotation of the amide bonds
to permit binding of the metal ion to the carbonyl oxygens. The
1H and 13C magnetic resonance peaks are therefore shifted down-
or upfield depending on the proximity of the nuclei to the
gallium center. The intensities of the NOE cross-peaks for each
complex provide relative measurements of the distances between
the nuclei and show that the amide protons approach the benzoyl
ortho proton as well as the seryl â-proton as the change from
catecholate to salicylate coordination occurs.
The reduction potentials for the ferric complexes [FeIII-
(TRENSAM)]0, [FeIII(TREN(3M)SAM)]0, [FeIII(SERSAM)]0,
and [FeIII(SER(3M)SAM)]0 were calculated using eq 6 and are
reported in Table 4. These experimental values are much higher
than those observed for the corresponding catecholate sidero-
phore complexes [FeIII(TRENCAM)]3- and [FeIII(Ent)]3- 44
. The
ligands built on a serine trilactone scaffold form ferric complexes
with higher reduction potentials than the equivalent TREN-based
compounds. This trend has been observed earlier with hydroxy-
pyridonate enterobactin analogues.45
Discussion
Salicylate Ferric Complexes as Model Compounds. What
is the structure of protonated ferric enterobactin and how might
the change in structure alter recognition and function? The
SERSAM and SER(3M)SAM ligand systems were designed to
explore the structural and spectroscopic features of a tris-
salicylate ligand built around the enterobactin triserine backbone.
A previous study had demonstrated the validity of the salicylate
coordination with the TREN-based analogues TRENSAM and
TREN(3M)SAM bound to hard metal cations such as Fe3+ and
Al3+ 23 As anticipated, the IR and UV-vis spectra of the [FeIII-
.
(SERSAM)]0 and [FeIII(SER(3M)SAM)]0 complexes exhibited
similar features to those of [FeIII(TRENSAM)]0 and [FeIII-
(TREN(3M)SAM)]0 respectively,23 which corroborates the
formation of tris-salicylato complexes. In addition, the 3-meth-
oxy-substituted [FeIII(SER(3M)SAM)]0 proves to be the best
spectroscopic model for the protonated [FeIII(H3Ent)]0 complex,
as emphasized by the nearly identical UV spectra of the two
species. The formation constants determined for both new
salicylate ferric complexes [FeIII(SERSAM)]0 and [FeIII(SER-
(3M)SAM)]0 correlate very well with the calculated hypothetical
formation constant of the triprotonated [FeIII(H3Ent)]0, as shown
by the equations depicted in Table 5, and therefore confirm that
the salicylate coordination mode is an excellent model for
[FeIII(H3Ent)]0.13,21
Structural Studies of the Catecholate and Salicylate
Coordinations. The density functional theory calculations
conducted on ferric and gallic enterobactin complexes in the
deprotonated and fully protonated states imply that these closely
related metal cations are subject to the same type of coordination
Reduction Potentials of Ferric-Salicylate Complexes. A
secondary mechanism for iron release by enterobactin relies on
the hypothesis that protonation of ferric enterobactin increases
its reduction significantly.18,19 The reduction potentials of the
salicylate analogues of ferric enterobactin (Table 4) are well
within the range of biological reductants and much higher than
those typically observed for the corresponding catecholate
(43) Marcus, Y. Pure Appl. Chem. 1983, 55, 977-1021.
(44) Rodgers, S. J.; Lee, C. W.; Ng, C. Y.; Raymond, K. N. Inorg. Chem. 1987,
26, 1622-1625.
(45) Meyer, M.; Telford, J. R.; Cohen, S. M.; White, D. J.; Xu, J.; Raymond,
K. N. J. Am. Chem. Soc. 1997, 119, 10093-10103.
9
8928 J. AM. CHEM. SOC. VOL. 128, NO. 27, 2006

Salicylate Coordination Shift in Ferric Enterobactin
A R T I C L E S
1
a white solid. Yield: 0.28 g (30%). H NMR (400 MHz, CDCl3): δ
4.20 (dd, J ) 6.8 Hz, J′ ) 4.4 Hz, 3H), 4.37 (dd, J ) 4.4 Hz, J′ ) 6.4
Hz, 3H), 5.00-5.11 (m, 9H), 6.90-7.58 (m, 24H), 8.15 (d, J ) 1.6
Hz, 3H), 8.54 (d, J ) 6.8 Hz, 3H). (+)-FABMS: m/z 892.6 (MH+).
Anal. Calcd (Found) for C51H45N3O12‚CH3OH: C, 67.60 (67.78); H,
5.35 (5.22); N, 4.55 (4.35).
siderophores. It is noteworthy that the reduction potentials
observed for the hydroxamate siderophores ferrichrome A and
ferrioxamine B are in the same range (Table 4).18 The triserine
trilactone stabilizes the ferrous over the ferric state compared
with the TREN scaffold;45 this trend is also seen in the salicylate
compounds, since the reduction potentials of [FeIII(SERSAM)]0
and [FeIII(SER(3M)SAM)]0 are higher than those of [FeIII-
(TRENSAM)]0 and [FeIII(TREN(3M)SAM)]0.
N,N′,N′′-Tris[2-benzyloxy,3-methoxy(benzoyl)carbonyl]cyclot-
riseryl Trilactone, Tribenzyl-SER(3M)SAM (5). Tris(N-hydrochloride-
L-serine) trilactone (0.371 g, 1.0 mmol) was suspended in 40 mL of
dry and degassed THF and cooled in an ice/water bath. Solutions of
2-(benzyloxy),3-(methoxy)benzoyl chloride (1.38 g, 5.0 mmol) in 10
mL of THF and triethylamine (1.01 g, 10.0 mmol) were added
simultaneously dropwise via syringes over 10 min into this suspension
while stirring under nitrogen. The mixture was allowed to warm to
room temperature and stirred overnight. It was then filtered, concen-
trated, applied to a silica gel column, and eluted with 96:4 CH2Cl2/
MeOH. Fractions were combined and evaporated to a white solid.
Conclusion
Enterobactin is inactivated in human serum by nonspecific
binding to human serum albumin and is bound by the human
immune protein siderocalin more strongly than by FepA, the
bacterial receptor for the siderophore.7,10 Recent studies have
shown that enzymatic functionalization of enterobactin makes
water-soluble derivatives that evade these traps.11,12 At low pH,
ferric enterobactin is protonated, remains intact, and undergoes
a marked structural change. The EXAFS and NMR structural
studies reported here show that ferric enterobactin goes through
a change from the catecholate coordination of [FeIII(Ent)]3- to
a salicylate mode of binding for [FeIII(H3Ent)]0 upon protonation.
The new synthetic hexadentate trilactone-based analogues
SERSAM and SER(3M)SAM form tris-salicylate ferric com-
plexes as do the previously made TRENSAM and TREN(3M)-
SAM chelators. These salicylate analogues of enterobactin are
excellent spectroscopic and thermodynamic models for the
triprotonated [FeIII(H3Ent)]0 complex. In addition, they exhibit
reduction potentials high enough to be in the range of those of
biological reductants. Protonation of ferric enterobactin generates
a large change in coordination mode and a corresponding overall
alteration in molecular shape and charge, as well as a weakened
iron binding affinity.
1
Yield: 0.62 g (63%). H NMR (400 MHz, CDCl3): δ 3.91 (s, 9H),
4.02 (dd, J ) 8.0 Hz, J′ ) 2.8 Hz, 3H), 4.15 (dd, J ) 4.4 Hz, J′ ) 6.4
Hz, 3H), 4.87-4.92 (m, 3H), 5.09 (m, 6H), 7.07-7.45 (m, 21H), 7.64
(d, J ) 6.4 Hz, 3H), 8.47 (d, J ) 7.2 Hz, 3H). (+)-FABMS: m/z 982.5
(MH+). Anal. Calcd (Found) for C54H51N3O15: C, 66.05 (65.92); H,
5.23 (5.48); N, 4.28 (4.63).
N,N′,N′′-Tris[2-(hydroxybenzoyl)carbonyl]cyclotriseryl Trilac-
tone, SERSAM (6). Absolute ethanol (14 mL) was added to a
suspension of tribenzyl-SERSAM (275 mg, 0.3 mmol) in 100 mL of
ethyl acetate. The solution was hydrogenated over 10% Pd-C (55.0
mg) at room temperature and atmospheric hydrogen pressure for 24 h.
The reaction mixture was filtered over Celite, washed with acetone,
and evaporated under vacuum. The product was collected as a white
1
powder. Yield: 177 mg (88%). H NMR (500 MHz, MeOD-d4): δ
4.63 (m, 6H), 5.03 (t, J ) 5.0 Hz, 3H), 6.88 (m, 6H), 7.38 (t, J ) 8.0
Hz, 3H), 7.78 (d, J ) 8.0 Hz, 3H). 13C NMR (500 MHz, MeOD-d4):
δ 52.2, 64.5, 115.6, 116.8, 119.0, 128.6, 133.7, 158.8, 168.6, 169.2.
(+)-FABMS: m/z 622 (MH+). Anal. Calcd (Found) for C30H27N3O12‚
0.5EtOH: C, 55.77 (55.79); H, 4.53 (4.31); N, 6.29 (6.05). IR: ν 1749
Experimental Section
s, 1641 s, 1604 s, 1531 s cm-1
.
N,N′,N′′-Tris[2-hydroxy, 3-methoxy(benzoyl)carbonyl]cyclot-
riseryl Trilactone, SER(3M)SAM (7). Absolute ethanol (14 mL) was
added to a suspension of tribenzyl-SER(3M)SAM (610 mg, 0.6 mmol)
in 100 mL of ethyl acetate. The solution was hydrogenated over 10%
Pd-C (122 mg) at room temperature and atmospheric hydrogen
pressure for 24 h. The reaction mixture was filtered over Celite, washed
with acetone, and evaporated under vacuum. The product was collected
as a white powder. Yield: 300 mg (70%). 1H NMR (500 MHz, MeOD-
d4): δ 3.87 (s, 9H), 4.64 (dd, J ) 4.0 Hz, J′ ) 11.5 Hz, 6H), 5.03 (t,
J ) 5.0 Hz, 3H), 6.84 (t, J ) 8.0 Hz, 3H), 7.10 (d, J ) 8.0 Hz, 3H),
7.40 (d, J ) 8.0 Hz, 3H). 13C NMR (500 MHz, MeOD-d4): δ 53.9,
56.8, 66.0, 116.2, 117.8, 120.2, 121.8, 149.3, 149.6, 169.4, 170.6. (+)-
FABMS: m/z 718 (MLi+). Anal. Calcd (Found) for C33H33N3O15: C,
55.70 (56.06); H, 4.67 (5.01); N, 5.90 (5.50). IR: ν 1748 s, 1642 s,
Synthesis. General. All chemicals were obtained from commercial
suppliers and were used as received. The starting materials tris(N-
hydrochloride-L-serine) trilactone45 (1), 2-(benzyloxy)benzoyl chlo-
ride23,24 (2), 2-(benzyloxy),3-(methoxy)benzoyl chloride23,24 (3), as
well as the ligand enterobactin31 and the ferric complexes [FeIII-
(TRENCAM)]3- 26 [FeIII(TRENSAM)]0,23 and [FeIII(TREN(3M)SAM)]0,
,
were prepared according to procedures described in the designated
references. Flash silica gel chromatography was performed using Merck
40-70 mesh silica gel. Melting points were taken on a Bu¨chi melting
apparatus and are uncorrected. All NMR spectra were recorded at
ambient temperature on Bruker FT-NMR spectrometers at the NMR
Laboratory, University of California, Berkeley. Microanalyses were
performed by the Microanalytical Services Laboratory, College of
Chemistry, University of California, Berkeley. Mass spectra were
recorded at the Mass Spectrometry Laboratory, College of Chemistry,
University of California, Berkeley. Infrared spectra were measured using
a Thermo Nicolet IR Avatar 370 Fourier transform spectrometer. UV-
visible absorption spectra were taken on a Varian Cary 300 UV-vis
spectrometer.
N,N′,N′′-Tris[2-(benzyloxybenzoyl)carbonyl]cyclotriseryl Trilac-
tone, Tribenzyl-SERSAM (4). Tris(N-hydrochloride-L-serine) trilac-
tone (0.371 g, 1.0 mmol) was suspended in 40 mL of dry and degassed
THF and cooled in an ice/water bath. Solutions of 2-(benzyloxy)benzoyl
chloride (1.23 g, 5.0 mmol) in 10 mL of THF and triethylamine (1.01
g, 10 mmol) were added simultaneously dropwise via syringes over
10 min into this suspension while stirring under nitrogen. The mixture
was allowed to warm to room temperature and stirred overnight. It
was then filtered, concentrated, applied to a silica gel column, and eluted
with 96:4 CH2Cl2/MeOH. Fractions were combined and evaporated to
1586 s, 1530 s cm-1
.
[FeIII(SERSAM)]0 (8). To a degassed MeOH solution of SERSAM
(74.4 mg, 0.11 mmol) was added anhydrous FeCl3 (19.4 mg, 0.12
mmol) and 0.12 mL of pyridine. The solution immediately turned dark
red, with rapid precipitation of the product as red powder. After the
reaction mixture was stirred for 1 h under nitrogen, the solid was
collected by filtration, washed with cold MeOH, and dried overnight.
Yield: 76 mg (91%). Mp > 300 °C. (+)-ESMS: m/z 697.1 ([M +
Na]+). Anal. Calcd (Found) for C30H24FeN3O12‚H2O‚CH3OH‚0.5HCl:
C, 48.93 (48.66); H, 4.04 (3.73); N, 5.52 (5.13). IR: ν 1751 s, 1606 s,
1574 s, 1534 s cm-1
.
[FeIII(SER(3M)SAM)]0 (9). To a degassed MeOH solution of SER-
(3M)SAM (100 mg, 0.14 mmol) was added anhydrous FeCl3 (22.7 mg,
0.14 mmol) and 0.55 mL of pyridine. The solution immediately turned
dark purple, with rapid precipitation of the product as purple powder.
9
J. AM. CHEM. SOC. VOL. 128, NO. 27, 2006 8929

A R T I C L E S
Abergel et al.
After the reaction mixture was stirred for 1 h under nitrogen, the solid
was collected by filtration, washed with cold MeOH, and dried
overnight. Yield: 68 mg (63%). Mp > 300 °C. (+)-ESMS: m/z 765.0
([MH]+). Anal. Calcd (Found) for C33H30FeN3O15‚0.5H2O: C, 51.25
(51.00); H, 4.04 (4.42); N, 5.43 (5.36). IR: ν 1754 s, 1604 s, 1580 s,
Titration Solutions and Equipment. Corning high performance
combination glass electrodes (response to [H+] was calibrated before
each titration)49 were used together with either an Accumet pH meter
or a Metrohm Titrino to measure the pH of the experimental solutions.
Metrohm autoburets (Dosimat or Titrino) were used for incremental
addition of acid or base standard solutions to the titration cell. The
titration instruments were fully automated and controlled using LabView
software.50 Titrations were performed in 0.1 M KCl supporting
electrolyte under positive Ar gas pressure. The temperature of the
experimental solution was maintained at 25 °C by an externally
circulating water bath. UV-visible spectra for incremental titrations
were recorded on a Hewlett-Packard 8452a spectrophotometer (diode
array). Solid reagents were weighed on a Metrohm analytical balance
accurate to 0.01 mg. All titrant solutions were prepared using distilled
water that was further purified by passing through a Millipore Milli-Q
reverse osmosis cartridge system. Titrants were degassed by boiling
for 1 h while being purged under Ar. Carbonate-free 0.1 M KOH was
prepared from Baker Dilut-It concentrate and was standardized by
titrating against potassium hydrogen phthalate using phenolphthalein
as an indicator. Solutions of 0.1 M HCl were similarly prepared and
were standardized by titrating against sodium tetraborate to Methyl Red
endpoint.
1536 s cm-1
.
[FeIII(Ent)]3- (10). To a degassed MeOH solution of enterobactin
(50 mg, 0.075 mmol) was added a solution of Fe(acac)3 (26.4 mg,
0.0750 mmol) in degassed MeOH. The solution immediately turned
dark purple. A KOH solution in MeOH (0.225 mmol) was then added
via cannula, and the solution turned deep red. The solution was taken
to dryness and applied to a Sephadex LH-20 column in MeOH. The
ferric complex eluted as a single red band, which was dried in vacuo
to afford a deep red powder. Yield: 42 mg (58%). Mp > 300 °C.
(-)-ESMS: m/z 360.2 ([M + H]2-). Anal. Calcd (Found) for C30H21-
FeK3N3O15‚4H2O‚2MeOH: C, 39.51 (39.61); H, 3.83 (3.70); N, 4.32
(3.95). UV-vis (5% DMAA in H2O): λ ) 498 nm (ꢀ ) 5700 M-1
cm-1), 338 nm (ꢀ ) 15 100 M-1 cm-1). IR: ν 1752 s, 1606 s, 1582 s,
1539 s, 1462 s, 1440 s, 1256 (sh), 1218 s cm-1
.
[FeIII(H3Ent)]0 (11). To a degassed MeOH solution of enterobactin
(50 mg, 0.075 mmol) was added a solution of Fe(acac)3 (26.4 mg,
0.0750 mmol) in degassed MeOH. The resulting mixture immediately
turned dark purple. It was then evaporated to dryness and applied to a
Sephadex LH-20 column in MeOH. The ferric complex eluted as a
single purple band which was dried in vacuo to yield a deep purple
powder. Yield: 33 mg (61%). Mp: 171-173 °C. (-)-ESMS: m/z 721.1
([M - H]-). UV-vis (5% DMAA in H2O): λ ) 309 nm (ꢀ ) 9000
M-1 cm-1). IR: ν 1754 s, 1640 s, 1620 s, 1584 s, 1547 s, 1442 s, 1222
Incremental Spectrophotometric Titrations. The protonation
constants of SERSAM and SER(3M)SAM as well as the formation
constants of their respective ferric complexes were determined by
spectrophotometric titration due to solubility issues. Solutions were
assembled from a weighed portion of compound and the supporting
electrolyte solution (containing no more than 4% of N,N-dimethylac-
etamide), with resulting ligand (or complex) concentrations between
50 and 100 µM. Constant buffering of the solution was assured by the
addition of NH4Cl, Hepes, and Mes buffers (500 µM). The solutions
were incrementally perturbed by the addition of either base (KOH,
ligand titrations) or acid (HCl, complex titrations) titrant, followed by
a time delay for equilibration (90 s for protonation studies; 2 h for
ferric complex titrations). An average of 40-50 data points were
collected in each ligand titration, each data point consisting of a pH
measurement and an absorbance spectra over the pH range 4.5 to 10.
An average of 25 data points was collected over the pH range 2.0-5.5
for the ferric complexes. All absorbance measurements used for
calculation of formation constants were less than 1.05 AU.
s cm-1
.
[GaIII(Ent)]3- (12). To a degassed MeOH solution of enterobactin
(66.9 mg, 0.1 mmol) was added a solution of Ga(acac)3 (36.7 mg, 0.1
mmol) in degassed MeOH. The solution turned slightly yellow. A KOH
solution in MeOH (0.3 mmol) was then added via cannula. The solution
was stirred for 24 h under N2, taken to dryness, and applied to a
Sephadex LH-20 column in MeOH. The gallic complex eluted as a
single yellow band which was dried in vacuo to afford a yellow powder.
Yield: 51 mg (60%). Mp > 300 °C. 1H NMR (500 MHz, DMSO-d6):
δ 3.74 (d, J ) 11.5 Hz, 3H), 5.07 (m, 3H), 5.16 (d, J ) 11.5 Hz, 3H),
6.03 (t, J ) 7.5 Hz, 3H), 6.18 (d, J ) 1.5 Hz, 3H), 6.64 (d, J ) 1.5
Hz, 3H), 11.88 (m, 3H). (-)-ESMS: m/z 810.0 ([M - K]-). UV-vis
(5% DMAA in H2O): λ ) 346 nm (ꢀ ) 14500 M-1 cm-1). IR: ν
1754 s, 1612 s, 1587 s, 1540 s, 1468 s, 1445 s, 1257 (sh), 1217 s
Data Treatment. All results presented are the average of at least
three independent titrations. All equilibrium constants were defined as
cumulative formation constants, âmlh according to eq 2, where the
ligands are designated as L. Stepwise protonation constants, Kan, may
be derived from these cumulative constants according to eq 3 (describes
proton association constants). Nonlinear least-squares refinement of the
protonation and formation constants was accomplished using the
program pHab.27 The proton concentration was allowed to vary in the
spectrophotometric studies, and all other concentrations were held at
estimated values determined from the volume of standardized stock or
the weight of ligand (measured to 0.01 mg). Refined concentrations
were within 5% of the analytical values. For spectral titrations, all
species formed with the ligands L were considered to have significant
absorbance to be observed in the UV-vis spectra.
DFT Calculations. Computational studies were conducted at the
Molecular Graphics and Computation Facility, College of Chemistry,
University of California, Berkeley. Density functional theory calcula-
tions were performed using lacvp**-b3lyp parameters for geometry
optimizations, with the computational software Jaguar 5.5.29 For a better
localization of the minima, the original coordinates before optimization
were established from the crystal structures of [VIV(Ent)]2- (JOSLOS)
and [FeIII(TRENSAM)]0 (HAKQAL) found on the Cambridge data-
base.23,31
cm-1
.
[GaIII(H3Ent)]0 (13). To a degassed MeOH solution of H6-
(enterobactin) (66.9 mg, 0.100 mmol) was added a solution of Ga-
(acac)3 (36.7 mg, 0.100 mmol) in degassed MeOH. The resulting
mixture turned slightly yellow. It was then evaporated to dryness and
applied to a Sephadex LH-20 column in MeOH. The gallic complex
eluted as a single band which was dried in vacuo to yield a beige
powder. Yield: 45 mg (61%). Mp ) 186-187 °C. 1H NMR (500 MHz,
DMSO-d6): δ 3.96 (br d, J ) 10.5 Hz, 3H), 5.07 (br, 3H), 5.16 (d, J
) 9.0 Hz, 3H), 6.29 (t, J ) 7.5 Hz, 3H), 6.57 (br, 3H), 7.01 (br, 3H),
8.89 (br, 3H), 11.22 (br, 3H). 13C NMR (400 MHz, DMSO-d6): δ 57.6,
62.8, 114.3, 118.7, 119.5, 124.2, 145.8, 163.6, 180.1, 185.3. (-)-
ESMS: m/z 734.1 ([M - H]-). UV-vis (5% DMAA in H2O): λ )
304 nm (ꢀ ) 9000 M-1 cm-1). IR: ν 1754 s, 1641 s, 1620 s, 1586 s,
1548 s, 1451 s, 1220 s cm-1
.
Solution Thermodynamics. Ligand protonation and complex for-
mation constants were determined using procedures and equipment
following previous descriptions.46-48
(46) Cohen, S. M.; O’Sullivan, B.; Raymond, K. N. Inorg. Chem. 2000, 39,
4339-4346.
(47) Johnson, A. R.; O’Sullivan, B.; Raymond, K. N. Inorg. Chem. 2000, 39,
2652-2660.
(48) Xu, J.; O’Sullivan, B.; Raymond, K. N. Inorg. Chem. 2002, 41, 6731-
(49) Gans, P.; O’Sullivan, B. Talanta 2000, 51, 33-37.
(50) LABVIEW, 5.0.1 ed.; National Instruments Corp.: Austin, TX.
6742.
9
8930 J. AM. CHEM. SOC. VOL. 128, NO. 27, 2006

Salicylate Coordination Shift in Ferric Enterobactin
A R T I C L E S
Extended X-ray Absorption Fine Structure Measurements.
EXAFS data were collected on the sector 20 bending magnet beamline
at the Advanced Photon Source (APS), Argonne National Laboratory,
Illinois, U.S.A. To prevent contamination by higher harmonics, a
harmonic rejection mirror with an energy cutoff of 11.2 keV was placed
in the X-ray beam. A slightly detuned Si(111) monochromator double-
crystal was used to collect iron K-edge (7112 eV) spectra and the X-ray
beam was defined using 1 mm × 8 mm slits in front of the first ion
chamber. Samples were run as powders mixed with boron nitride (1:
6) in transmission mode to k ) 15 Å-1. Nitrogen was used in all ion
chambers with 10% helium added in the first ion chamber. Transmission
spectra were analyzed using RSXAP,51,52 an r-space fitting and data
reduction suite of programs. The pre-edge absorption was removed by
fitting the transmission data to a Victoreen formula, and the resulting
spectra were normalized to unity at the post-edge step. The post-edge
background was removed by subtracting the isolated atom absorption
(µo), obtained from a series of cubic splines. After this subtraction,
choice of an energy origin (E0) and conversion from energy space to
photoelectron wavevector space, selected k3-weighted spectra are shown
in Figures 5 and S2. An increase in coordination number (N) would be
expected to increase the amplitude in a linear manner at all k values,
whereas an increase in disorder (σ) would attenuate the EXAFS signal
exponentially and the effect would increase exponentially as k increases.
The fitted range of k values was 2.5-13.1 Å-1 for the data collected
at 20 K with an r-space fit from 1.2 to 3 Å, giving 14 total degrees of
freedom (dof) using Stern’s rule (dof ) (2 × ∆r∆k/π) + 2).53 The
Protonation constants were determined by nonlinear least-squares
refinement using the program HypNMR55 with an estimated uncertainty
in chemical shift of 0.01 ppm and corrected for the deuterium effect
after the refinement according to the correlation log KD ) 0.32 + 1.044
log KH.28
NOE Growth Rates. Samples of [GaIII(Ent)]3- and [GaIII(H3Ent)]0
were prepared as 0.1 M solutions in DMSO-d6. Two-dimensional
NOESY spectra were recorded at 348.1 K on a Bruker Avance AV
500 spectrometer, with varying mixing times (150, 300, 400, 500, 600,
700, 850, and 1000 ms). NOE cross-peak intensity was plotted versus
mixing time (τm). Using the diagonal peaks for each proton as
references, the relative NOE growth rates σ for the distances of interest
1/6 41
could be evaluated, according to rAB ) rXY × (σXY/σAB
) .
Electrochemical Measurements. Cyclic voltammograms of 0.3 mM
ferric complex solutions in acetonitrile containing 0.1 M tetrabutylam-
monium hexafluorophosphate (Fluka, electrochemical grade) as sup-
porting electrolyte were recorded on a BAS100A electrochemical
analyzer. Working and auxiliary platinum electrodes were used with a
freshly prepared Pleskow reference electrode (0.01 M AgNO3 in 0.1
M N(t-Bu)4‚PF6 in CH3CN/Ag). The half-wave potentials were also
measured against the ferrocinium/ferrocene internal reference electrode.
Acknowledgment. This research was supported by the
National Institutes of Health (Grant AI11744-28). We thank Dr.
K. Durkin for assistance with the computational studies (NSF
CHE-0233882) and Dr. H. van Halbeek for help with the NMR
experiments. Use of the Advanced Photon Source was supported
by the U.S. Department of Energy (DOE), Office of Science,
Office of Basic Energy Sciences, under Contract No. W-31-
109-ENG-38. We thank Steve Heald and the staff of PNC Sector
20 bend magnet beamline for their assistance. J.A.W. and D.K.S.
are supported by the Director, Office of Science, Office of Basic
Energy Sciences, Division of Chemical Sciences, Geosciences,
and Biosciences of the U.S. Department of Energy at Lawrence
Berkeley National Laboratory under Contract No. DE-AC02-
05CH11231.
2
amplitude reduction factor, S0 , was estimated to be 0.75 from low-
temperature first-shell fits of both experimental samples. The energy
origin, E0, was obtained by averaging first-shell fit results over all scans
at all temperatures. These values were subsequently fixed and applied
to fits over the full fit range. The fitting model included single-scattering
shells from first-shell oxygens and second-shell carbons. Final fits
generally allowed only five unconstrained variables from a total of 14
parameters. Photoelectron scattering amplitudes and phase shifts were
calculated using the FEFF7 code54 and crystal structures of [FeIII-
(TRENSAM)]0 and [FeIII(TRENCAM)]3- 23,26
.
pD-1H NMR Spectroscopic Titration. A combination electrode
(Corning) filled with 3 M KCl (Corning filling solution) was calibrated
in water by a strong acid/strong base titration. A solution of 1 mM
[GaIII(Ent)]3-, 0.1 M KCl, and 1 mM 3-(trimethylsilyl)-1-propane-
sulfonic acid (DSS) shift agent in D2O was prepared and titrated by
addition of DCl with values of pD calculated according to the
relationship pD ) 0.4 + pH.28 Aliquots of 0.450 mL each were retrieved
after each DCl addition. 1H NMR spectra of each aliquot were
performed on a Brucker Avance AV-500 MHz equipped with a TBI
probe, using a pulse angle of 90° and 128 scans per sample. Chemical
shifts were referenced to the CH3-Si peak of DSS (0.000 ppm).
Supporting Information Available: Figures showing selected
spectral data obtained for the spectrophotometric titration of
SERSAM and [FeIII(SERSAM)]0, and the EXAFS characteriza-
tion of all ferric complexes; table of EXAFS fit parameters;
selected NOE build-up rates and tables of calculated heats of
formation and bond lengths for gallic enterobactin complexes.
This material is available free of charge via the Internet at
http://pubs.acs.org.
(51) Hayes, T. M.; Boyce, J. B. Solid State Phys. AdV. Res. App. 1982, 37,
173-351.
(52) Li, G. G.; Bridges, F.; Booth, C. H. Phys. ReV. B 1995, 521, 6332-6348.
(53) Stern, E. A. Phys. ReV. B 1993, 48, 9825-9827.
(54) Zabinsky, S. I.; Rehr, J. J.; Ankudinov, A.; Albers, R. C.; Eller, M. J. Phys.
ReV. B 1995, 52, 2995-3009.
JA062046J
(55) Gans, P. Protonic Software: Leeds, U.K., 1999.
9
J. AM. CHEM. SOC. VOL. 128, NO. 27, 2006 8931