Synthesis, structure, and molecular dynamics of gallium complexes of schizokinen and the amphiphilic siderophore acinetoferrin

Article abstract of DOI:10.1021/ja048145j

A new general synthesis of the citrate-based siderophores acinetoferrin (Af) and schizokinen (Sz) and their analogues is described. The molecular structure of gallium schizokinen, GaSz, was determined by combined ¹H NMR, Hartree-Fock ab initio

Full text of DOI:10.1021/ja048145j

Published on Web 09/01/2004
Synthesis, Structure, and Molecular Dynamics of Gallium
Complexes of Schizokinen and the Amphiphilic
Siderophore Acinetoferrin
Evgeny A. Fadeev, Minkui Luo, and John T. Groves*
Contribution from the Department of Chemistry, Princeton UniVersity,
Princeton, New Jersey 08544
Received March 31, 2004; E-mail: jtgroves@princeton.edu
Abstract: A new general synthesis of the citrate-based siderophores acinetoferrin (Af) and schizokinen
(Sz) and their analogues is described. The molecular structure of gallium schizokinen, GaSz, was determined
1
by combined H NMR, Hartree-Fock ab initio calculations, DFT, and empirical modeling of vicinal proton
NMR spin-spin couplings. The metal-coordination geometry of GaSz was determined from NOE contacts
to be cis-cis with respect to the two chelating hydroxamates. One diaminopropane adopts a single chairlike
conformation while another is a mixture of two ring pucker arrangements. Both amide hydrogens are
internally hydrogen bonded to metal-ligating oxygen atoms. The acyl methyl groups are directed away
from each other with an average planar angle of ca. 130°. The kinetics of GaSz racemization were followed
1
by selective, double spin-echo inversion-recovery H NMR spectroscopy over the temperature range of
10-45 °C. The racemization proceeds by a multistep mechanism that is proton independent between pD
5 and 12 (k0 ) 1.47 (0.15 s-1)) and acid catalyzed below pD 4 (k1 ) 2.25 (0.15) × 104 M-1 s-1). The
activation parameters found for the two sequential steps of the proton independent pathway were ∆Hq )
25 ( 3 kcal M-1, ∆Sq ) 25 ( 7 cal M-1 K-1 and ∆Hq ) 17.1 ( 0.2 kcal M-1, ∆Sq ) 0.3 ( 2.7 cal M-1 K-1
.
The first step of the proton-independent mechanism was assigned to the dissociation of the carboxyl group.
The second step was assigned to complex racemization. The proton-assisted step was assigned to a
complete dissociation of the R-hydroxy carboxyl group at pD < 4. The ab initio modeling of gallium
acinetoferrin, GaAf, and analogues derived from the structure of GaSz has shown that the pendant trans-
octenoyl fragments are oriented in opposite directions with the average planar angle of ca. 130°. This
arrangement prevents GaAf from adopting a phospholipid-like structural motif. Significantly, iron siderophore
complex FeAf was found to be disruptive to phospholipid vesicles and is considerably more hydrophilic
than Af, with an eight-fold smaller partition coefficient.
microorganisms.13-15 Mycobactins, lipophilic siderophores of
Introduction
Mycobacter smegmatis, have shown beneficial activity for the
Siderophores are a class of low molecular weight organic
compounds that are produced by microorganisms in response
to iron deficiency. These compounds are capable of highly
selective chelation and delivery of inorganic iron1,2 to bacterial
cells via specific, receptor-mediated membrane transport mech-
anisms.3 Siderophore-mediated iron uptake is an important
determinant of bacterial growth3 and there is suggestive evidence
that iron proteins and siderophores are involved in cell signaling
cascades and quorum sensing.4-6 Amphiphiles are a significant
and widely distributed subset of siderophores,7 having been
found in pathogens,8-11 terrestrial symbionts,12 and marine
treatment of arteriosclerosis,16 breast cancer,17 and postischemic
reperfusion injury.18 A synthetic, amphiphilic analogue of
desferoxamine has been reported to inhibit malaria19,20 and
another lipophilic tripeptide ornithine-based hydroxamate
(9) Okujo, N.; Sakakibara, Y.; Yoshida, T.; Yamamoto, S. Biometals 1994, 7,
170-176.
(10) Snow, G. A. Biochem. J. 1965, 97, 166-175.
(11) Snow, G. A. Biochem. J. 1965, 94, 160-165.
(12) Persmark, M.; Pittman, P.; Buyer, J. S.; Schwyn, B.; Gill, P. R.; Neilands,
J. B. J. Am. Chem. Soc. 1993, 115, 3950-3956.
(13) Macrellis, H. M.; Trick, C. G.; Rue, E. L.; Smith, G.; Bruland, K. W. Mar.
Chem. 2001, 76, 175-187.
(14) Butler, A.; Martinez, J. S.; Barbeau, K. J. Inorg. Biochem. 2001, 86, 30-
30.
(1) Albrecht-Gary, A. M.; Crumbliss, A. L. Met. Ions Biol. Syst. 1998, 35,
239-327.
(2) Boukhalfa, H.; Crumbliss, A. L. Biometals 2002, 15, 325-339.
(3) Neilands, J. B. J. Biol. Chem. 1995, 270, 26723-26726.
(4) Fong, K. P.; Gao, L.; Demuth, D. R. Infect. Immun. 2003, 71, 298-308.
(5) Zhu, J.; Dizin, E.; Hu, X.; Wavreille, A.; Park, J.; Pei, D. Biochemistry
2003, 42, 4717-4726.
(6) Bollinger, N.; Hasset, D. J.; Iglewski, B. H.; Costerton, J. W.; McDermott,
T. R. J. Bacteriol. 2001, 183, 1990-1996.
(7) Groves, J. T. Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 3569-3574.
(8) Lane, S. J.; Marshall, P. S.; Upton, R. J.; Ratledge, C. Biometals 1998, 11,
13-20.
(15) Martinez, J. S.; Zhang, G. P.; Holt, P. D.; Jung, H. T.; Carrano, C. J.;
Haygood, M. G.; Butler, A. Science 2000, 287, 1245-1247.
(16) Rosenthal, E. A.; Bohlmeyer, T. J.; Monnet, E.; MacPhail, C.; Robertson,
A. D.; Horwitz, M. A.; Burchenal, J. E. B.; Horwitz, L. D. Circulation
2001, 104, 2222-2227.
(17) Pahl, P. M. B.; Horwitz, M. A.; Horwitz, K. B.; Horwitz, L. D. Breast
Cancer Res. Treat. 2001, 69, 69-79.
(18) Horwitz, L. D.; Sherman, N. A.; Kong, Y. N.; Pike, A. W.; Gobin, J.;
Fennessey, P. V.; Horwitz, M. A. Proc. Natl. Acad. Sci. U.S.A. 1998, 95,
5263-5268.
(19) Lytton, S. D.; Cabantchik, Z. I.; Libman, J.; Shanzer, A. Mol. Pharmacol.
1991, 40, 584-590.
9
10.1021/ja048145j CCC: $27.50 © 2004 American Chemical Society
J. AM. CHEM. SOC. 2004, 126, 12065-12075
12065

A R T I C L E S
Fadeev et al.
wherein its interactions with cell membranes play an essential
role. While the coordination chemistry of ferric schizokinen
FeSz40 and ferric aerobactin41 have been investigated spectro-
scopically, no molecular structures of this class of citrate-based
siderophores have been reported.
Here we present an efficient and general synthesis of Af, Sz,
and their analogues. We also describe the three-dimensional
structure of the Ga3+ complexes of Sz and Af derived from 1H
NMR data, Hartree-Fock ab initio, and DFT molecular
modeling. The coordination geometries of the GaAf and GaSz
hydroxamate groups are unambiguously determined to be cis-
1
cis. The two enantiomers of GaAf and GaSz are shown by H
NMR to equilibrate readily. A sequential, two-step mechanism
of racemization has been discerned from the kinetics of this
process. The solution structure of FeAf and the conformational
analysis of the FeAf aliphatic tails have been inferred from the
GaSz geometry. The resulting conformation of FeAf suggests
a special mode of interaction of the metal-Af complex with the
cell phospholipid membranes that is dictated by a change in
the arrangement of the lipophilic side chains upon metal binding.
Figure 1. Structure of acinetoferrin, Af (R ) trans-1-heptenyl) and Sz (R
) methyl).
chelator was found to effectively promote growth of Mycobacter
smegmatis.21 Strategies for the design of antibiotic “Trojan
Horse” siderophore conjugates have been described.22 The
ubiquity of amphiphilic siderophores, the mechanisms of their
exploitation by pathogens, and their promising pharmacological
activities focus attention on the structures and functions of these
little-studied natural products.
Our interest in amphiphiles23 and membrane assemblies24-27
has led us to examine the citrate-based siderophore acinetoferrin
(Af) of Acinetobacter haemoliticus, an antibiotic-resistant
bacterium that causes an increasing number of difficult-to-treat
infections.28 This bacterium produces two siderophores: acineto-
ferrin9 (principally) and smaller amounts of acinetobactin.29,30
Acinetoferrin belongs to a class of citrate-based siderophores
(Figure 1) in which the terminal carboxyl groups of citric acid
are peptide-coupled to two hydroxamate-bearing appendages.
Other citrate-based siderophores are aerobactin,31 arthrobactin,32
nannochelin,33 petrobactin,34-36 rhizobactin 1021,12 and schizok-
inen (Sz).37-39
Results
Synthesis of Siderophores. Synthesis of schizokinen42 and
acinetoferrin43 were reported previously. We developed a general
method (Figure 2) allowing synthetic access to both compounds
and their 2-E-butenoyl and 2-E-dodecenoyl homologues without
major changes in the procedure. The key feature of this approach
is the coupling of 2-tert-butyl-1,3-di-N-(hydroxy)succinimidyl
citrate,42 1, with 1-N-benzoyloxy-1,3-diaminopropane dihydro-
chloride 2. Compound 2 was conveniently obtained from
N-(3-(tert-butoxycarboylamino)-N-(1-benzoyloxy)-diamino-
propane43 upon treatment with dry HCl.
Attachment of the acyl fragments to 3 was achieved in high
yield via the appropriate acyl chloride. The resulting fully
protected siderophores, 4a-d, were amenable to purification
by column chromatography under conditions dictated by the
terminal acyl chain length. This strategy has the advantage of
retaining the protected hydroxamate groups until the molecular
scaffold was fully assembled, thus lessening the iron contamina-
tion acquired from the glassware and silica that is an inherent
difficulty with these compounds.43
The benzoyl protecting groups of 4a-d were removed under
basic conditions and the resulting crude products were separated
on a short column packed with a specially prepared, iron-free
silica.43 Final cleavage of the tert-butyl esters 5a-c was effected
with 95:5 TFA:H2O at 25 °C. We found that dry TFA led to
some dehydration of the citrate hydroxyl. The free siderophores
6a (Sz) and 6c were purified by gel filtration with sephadex
G-10. Compounds 6b (Af) and 6d were sufficiently lipophilic
to be separated from the residual TFA by the chloroform/water
extraction.
Stoichiometry and Overall Structure of the Iron Af and
Gallium Sz Complexes. The iron complexes Fe-6(a-d) were
prepared by metalation of 6(a-d) with ferric ammonium citrate.
Polyacrylamide gel electrophoresis showed that Fe-6(a-d) were
Acinetoferrin is a unique member of this family, having two
lipophilic side chains resembling a phospholipid structural motif.
The amphiphilic nature of this siderophore and its unusual
structure are suggestive of a special mechanism of action,
(20) Harpstrite, S. E.; Beatty, A. A.; Collins, S. D.; Oksman, A.; Goldberg, D.
E.; Sharma, V. Inorg. Chem. 2003, 42, 2294-2300.
(21) Lin, Y. M.; Miller, M. J.; Mollmann, U. Biometals 2001, 14, 153-157.
(22) Murray, A. P.; Miller, M. J. J. Org. Chem. 2003, 68, 191-194.
(23) Xu, G. F.; Martinez, J. S.; Groves, J. T.; Butler, A. J. Am. Chem. Soc.
2002, 124, 13408-13415.
(24) Lahiri, J.; Fate, G. D.; Ungashe, S. B.; Groves, J. T. J. Am. Chem. Soc.
1996, 118, 2347-2358.
(25) Groves, J. T.; Fate, G. D.; Lahiri, J. J. Am. Chem. Soc. 1994, 116, 5477-
5478.
(26) Groves, J. T.; Neumann, R. J. Am. Chem. Soc. 1989, 111, 2900-2909.
(27) Groves, J. T.; Neumann, R. J. Am. Chem. Soc. 1987, 109, 5045-5047.
(28) Rudant, E.; Courvalin, P.; Lambert, T. Antimicrob. Agents Chemother. 1997,
41, 2646-2651.
(29) Yamamoto, S.; Okujo, N.; Sakakibara, Y. Arch. Microbiol. 1994, 162, 249-
254.
(30) Dorsey, C. W.; Tolmasky, M. E.; Crosa, J. H.; Actis, L. A. Microbiology
2003, 149, 1227-1238.
(31) Gibson, F.; Magrath, D. I. Biochim. Biophys. Acta 1969, 192, 175-184.
(32) Schafft, M.; Diekmann, H. Arch. Microbiol. 1978, 117, 203-207.
(33) Kunze, B.; Trowitzschkienast, W.; Hofle, G.; Reichenbach, H. J. Antibiot.
1992, 45, 147-150.
(34) Barbeau, K.; Zhang, G.; Live, D. H.; Butler, A. J. Am. Chem. Soc. 2002,
124, 378-379.
(35) Bergeron, R. J.; Huang, G. F.; Smith, R. E.; Bharti, N.; McManis, J. S.;
Butler, A. Tetrahedron 2003, 59, 2007-2014.
(40) Plowman, J. E.; Loehr, T. M.; Goldman, S. J.; Sanders-Loehr, J. J. Inorg.
Biochem. 1984, 20, 183-197.
(36) Gardner, R. A.; Kinkade, R.; Wang, C. J.; Phanstiel, O. J. Org. Chem.
2004, 69, 3530-3537.
(41) Harris, W. R.; Carrano, C. J.; Raymond, K. N. J. Am. Chem. Soc. 1979,
101, 2722-2727.
(37) Akers, H. A. Appl. EnViron. Microbiol. 1983, 45, 1704-1706.
(38) Simpson, F. B.; Neilands, J. B. J. Phycol. 1976, 12, 44-48.
(39) Trick, C. G.; Kerry, A. Curr. Microbiol. 1992, 24, 241-245.
(42) Milewska, M. J.; Chimiak, A.; Glowacky, Z. J. Prakt. Chem./Chem.-Ztg.
1987, 329, 447-456.
(43) Wang, Q. X.; Phanstiel, O. I. J. Org. Chem. 1998, 63, 1491-1495.
9
12066 J. AM. CHEM. SOC. VOL. 126, NO. 38, 2004

+
Structures of Ga3 Siderophore Complexes
A R T I C L E S
protons of the terminal alkyl side chains. The GaSz spectrum
was more amenable to detailed analysis as it was better resolved
and not complicated by the presence of signals from the trans-
octenoyl fragments of GaAf.
1
The H NMR spectrum of the Na[GaSz] in D2O (Figure 3)
displayed remarkably well-defined multiplets in the range of δ
1.4-4.3 for the methylene and methyl protons and at δ 7.5 and
9.3 for the amide protons. There were additional broad
resonances more pronounced at higher pH (>6) that were not
assigned to a particular structure and probably account for a
partially hydrolyzed complex.
1
The H peaks of the terminal methyl groups and all the
methylene hydrogen atoms, excluding those on the citrate
moiety, were assigned to one of the N-n-diaminopropane
branches on the ligand on the basis of the bond connectivity
revealed by COSY and distance via NOESY spectra. Thus,
resonances 1-7 in Figure 3 belong to one diaminopropane chain
(designated chain I), and resonances 1′-7′ to the other chain
(chain II). Resonances from the methylene groups neighboring
the hydroxamate residues (2, 2′, 4, and 4′) were distinguished
by the presence of the NOE correlation peaks between the
resonances 2 and 2′ with the corresponding terminal methyl
protons. Hydrogen atoms on the methylene groups connected
to the amide nitrogens were identified by observing crosspeaks
in the COSY spectrum with WET water resonance suppression.
1
The H resonance chemical shift assignments for GaSz and
the spin-spin coupling values are shown in Table 1. The
observed spin-coupling pattern for chain I is consistent with a
single conformation with well-defined, staggered relationship
of the central methylene (resonances 6 and 7) with the two
neighboring methylene groups (resonances 2-5). Resonance 7
clearly originates from a proton bearing two trans-diaxial
relationships with its neighbors. Also, there is an additional NOE
signal (not observed for the chain II) between the two hydrogen
atoms (resonances 4 and 5) separated by three carbon atoms.
The protons in chain II, by contrast, display more average-sized
vicinal splittings. Also, the chemical shift dispersion within
geminal proton pairs is smaller in the chain II.
That every proton in the molecule has a distinctive chemical
shift indicated that the GaSz complex is chiral. The 1D 1H NMR
spectra of GaSz and GaAf did not show signs of exchange
broadening at room temperature but NOE spectra (Figures S15
and S16) showed slow exchange of chain I and chain II
resonances, indicating that the chains interchanged their identi-
ties by the racemization.44 The racemization kinetics were
followed by a Double Pulsed Field Gradient Spin-Echo
(DPFGSE) NOESY-1D experiment45 that has a great advantage
over 2D NOESY measurements for small molecules due to the
higher accuracy and speed. The kinetic data yielded accurate
racemization rate constants and activation parameters (see the
section below) allowing an insight into the racemization
mechanism.
Figure 2. General synthesis of the acinetoferrin siderophore family: (a) 3
equiv of TEA, CH3CN; (b) 2 equiv of acyl chloride, CH2Cl2, reflux, 1 h;
(c) 2 equiv of 5 N aqueous NaOH, CH3OH, 0 °C, 10 min; (d) TFA/H2O
19:1, 1 h. R: -CH3- 4-6a; trans-(CH)2(CH2)4CH3- 4-6b; trans-
(CH)2CH3- 4-6c; trans-(CH)2(CH2)8CH3- 4-6d.
negatively charged at neutral pH. Orange colored bands (λmax
) 400-426 nm) of the complex molecules were observed to
move toward the anode. Negative ion electrospray mass
spectrometry of FeAf and GaSz (Figures S1 and S2) complexes
showed one major peak at m/z 636 and 485, respectively,
corresponding to a 1:1 metal-siderophore complex bearing one
negative charge. A NMR titration of Sz with GaBr3 (Figure
S3) also indicated a 1:1 metal-to-ligand stoichiometry in
agreement with the conclusions of Plowman et al.40 for the FeSz
complex. Therefore, the ligands must saturate the Ga3+ and Fe3+
octahedral coordination and the R-hydroxy acid as well as the
two hydroxamates are deprotonated.
Another interesting feature of the GaSz NMR data is the low-
field values of the amide hydrogen chemical shifts (resonances
1 and 1′) at δ 9.43 and 7.5, suggesting the presence of
intramolecular hydrogen bonds.46 To investigate properties of
NMR Spectroscopy of GaSz. The 1H NMR spectra of GaSz
(Figure 3) and GaAf (Figure S16) were very similar, indicating
very similar structures. Among the spectral similarities are the
1
ordering and shape of the H signals from the citrate and 1,3-
(44) Perrin, C. L.; Dwyer, T. J. Chem. ReV. 1989, 90, 935-967.
(45) Stott, K.; Keeler, J.; Van, Q. N.; Shaka, A. J. J. Magn. Reson. 1997, 125,
302-324.
(46) Becker, E. Hydrogen bonding. In Encyclopedia of Nuclear Magnetic
Resonance; John Wiley: New York, 1996; Vol. 4, p 4212.
diaminopropane methylene groups, and the presence of the two
NOE cross-peaks between one methylene proton adjacent to
the each hydroxamate and the vinyl (GaAf) or methyl (GaSz)
9
J. AM. CHEM. SOC. VOL. 126, NO. 38, 2004 12067

A R T I C L E S
Fadeev et al.
Figure 3. 1H NMR spectrum of aqueous Na[GaSz], pH ) 7, T ) 25 °C: (a) WET-1D 1H spectrum of the amide region in 10% D2O; (b) 1D 1H spectrum
in D2O. Spin-spin connectivities, the corresponding coupling magnitudes, and the NOE contacts are indicated. Spin-spin coupling assignments for chain
I are grouped above the spectral line and for chain II, below.
Table 1. 1H NMR Peak Assignments of Ga Schizokinen Complex, GaSza
Chain I
spin−spin couplings, Hz
resonance
δ
assignment
2
3
4
5
6
7
1
2*
3
4**
5**
6
9.43
4.27
3.81
3.42
2.74
2.11
1.45
-NH-
8.9-8.7
∼2
-CH2-NOH
-CH2-NH
-CH2-NOH
-CH2-NH
C-CH2-C
C-CH2-C
15.8
6.59
6.09
1.58
2.06
14.35
15.18
10.47
10.47
15.7
14.35
6.09
2.06
6.59
1.58
7
10.47
10.47
15.7
Chain II
spin−spin couplings, Hz
resonance
δ
assignment
2'
3'
4'
5'
6'
7'
1′
2′*
3′
4′
5′
7.5
-NH-
6.5
2.3-4
4.08
3.67
3.52
3.02
2.28
1.69
-CH2-NOH
-CH2-NH
-CH2-NOH
-CH2-NH
C-CH2-C
C-CH2-C
14.77
7.43
1.88
3.08
8.84
3.03
8.77
8.57
1.91
14.3
14.77
14.3
1.88
8.77
6′
7′
7.43
3.03
3.08
8.57
8.84
1.91
∼15
∼15
a Spectra taken in D2O at pD ) 7.5. *NOE contact with terminal methyl hydrogens. **Weak NOE signal.
these hydrogen bonds, we studied dependencies of amide
resonance chemical shifts on pD. The chemical shift of
resonance 1′ was slightly sensitive to solution pD in the alkaline
region (pD ∼ 11); in contrast resonance 1 did not change until
the complete hydrolysis at pD ∼ 12.
Kinetics of GaSz Racemization. The racemization rate was
determined from DPFGSE NOESY-1D spectra as a function
of pD at 15 °C and 25 °C (Figure 8). There were two regimes
of racemization, depending on acidity. At pD > 5 its rate was
nearly constant, krac ≈ 1.4 s-1, and the ∆Gq profile vs T (Figure
9
12068 J. AM. CHEM. SOC. VOL. 126, NO. 38, 2004

+
Structures of Ga3 Siderophore Complexes
A R T I C L E S
Figure 4. GaSz configurations, (a) cis-cis and (b) cis-trans.
Figure 6. GaAf structure in RRââ conformation of the core fragment and
in one of the minimum-energy octenoyl tail conformations. Dashed lines
denote the NH‚‚‚O hydrogen bonds.
Models of GaSz with different 11-membered chelate ring-
pucker conformations within both cis-cis and cis-trans
geometries were built by changing orientations of the amide
groups and the central methylene units on the two N-n-
propylacetamido chains. The orientations of those groups were
designated as R (“up”) or â (“down”).47 As a reference, the
molecular orientation was as shown in Figures 4 and 5.
Figure 5. Structures of the two dominant GaSz conformations and 1H NMR
resonance assignments. Arrows show the groups with different ring-pucker
arrangements in structures RRRR and RRââ.
7) obtained from the measured rated constants clearly showed
biphasic behavior. At T < 28 °C, ∆Hq was 25 ( 3 kcal M-1
and ∆Sq was 25 ( 7 cal M-1 K-1. Above 28 °C the activation
parameters sharply changed: ∆Hq became 17.1 ( 0.2 kcal M-1
and ∆Sq dropped to a value of 0.3 ( 2.7 cal M-1 K-1. At pD
< 4 the racemization sharply accelerated (Figure 8) and was
first-order in [D]+, where the rate constant krac could be
represented as krac ) k1 [D]+ + k0, with k1 ) (2.25 ( 0.15) 104
M-1 s-1, k0 ) 1.47 ( 0.15 s-1, R2 ) 0.9924.
Molecular Modeling of the Structures of GaSz and GaAf.
The geometry of the GaAf structure was approached by
molecular modeling in several steps. Ab initio Hartree-Fock
and DFT-optimized GaSz structures were examined for compat-
ibility with the experimental vicinal spin-spin coupling values,
the NOE contacts, and the dynamic features observed in the
NMR spectra. Candidate structures for the core fragment of
GaAf were chosen in cis-cis and cis-trans coordination
geometries and with various 11-membered chelation ring pucker
arrangements. The best geometries yielded a complete chemical
shift assignment of the GaSz 1H NMR spectrum shown in Figure
5. Finally, the favorable GaAf terminal alkyl chain orientations
(Figure 6) were determined by analyzing the torsional energy
profiles for the GaSz models extended with one terminal trans-
1-propenyl and the trans-1-heptenyl tail.
Ab Initio Hartree-Fock and DFT Modeling. Calculation
Strategy. The Hartree-Fock ab initio and DFT methods were
explored for the modeling of GaSz structures. Geometry
optimizations with B3LYP/6-31G(d) essentially reproduced the
literature X-ray structure geometrical parameters of gallium
hydroxamate complexes.48-52 By contrast, some of those
parameters in the RHF/6-31G(d) gas-phase geometry-optimized
structures deviated slightly from those known from literature.48-52
In particular, the hydroxamate C-O bond length in the RHF/
6-31G(d) structures (1.24 Å) was ∼0.03 Å shorter than is
typical. However, the relative conformational energies were
practically the same when obtained for both ab initio RHF/6-
31G(d) and DFT B3LYP/6-31G(d) methods.53 We conclude that
a slight inacuracy in the geometry predictions for the gallium
(47) The four-symbol combinations (RRRR, RRRâ, RγâR, etc.) were used to
define the various ring-pucker conformations of GaSz. The label characters
reflect orientations of the four groups labeled as i-iV in Figure S25b: left-
hand side methylene (i), left-hand side carbonyl (ii), right-hand side carbonyl
(iii), and right-hand side methylene (iV) groups, in that order. An additional
amide group conformation, "γ", was introduced after we obtained the
optimized structures. In that conformation the amide proton was hydrogen-
bonded to the uncoordinated citrate central carboxyl oxygen, instead of
one of the metal-bound oxygen atoms, as it was in all "R“ and ”â"
conformations.
(48) Keys, A.; Barbarich, T. J.; Bott, S. G.; Barron, A. R. J. Chem. Soc., Dalton
Trans. 2000w, 577-588.
(49) Beissel, T.; Powers, R. E.; Raymond, K. N. Angew. Chem., Int. Ed. Engl.
1996, 35, 1084-1086.
GaSz Model Construction. We considered two GaSz con-
figurations. The cis-cis configuration (Figure 4a),wherein both
hydroxamate NO fragments are cis with respect to the ligand
carboxyl group, appearing to be more relaxed in comparison to
the cis-trans configuration (Figure 4b), in which the hydrox-
amate NO fragment on the right part of molecule is trans- with
respect to the carboxyl group. A third, trans-cis configuration,
was unstable in ab initio computations.
(50) Dietrich, A.; Fidelis, K. A.; Powell, D. R.; Vanderhelm, D.; Engwilmot,
D. L. J. Chem. Soc., Dalton Trans. 1991, 231-239.
(51) Borgias, B. A.; Barclay, S. J.; Raymond, K. N. J. Coord. Chem. 1986, 15,
109-123.
(52) Dhungana, S.; White, P. S.; Crumbliss, A. L. J. Am. Chem. Soc. 2003,
125, 14760-14767.
(53) The relative GaSz conformation energies were almost the same when
calculated with RHF/6-31G(d)//RHF/6-31G(d), RHF/6-31G(d)//B3LYP/6-
31G(d), and B3LYP/6-31G(d)//B3LYP/6-31G(d) method combinations for
the geometry optimization and the single point energy calculation
9
J. AM. CHEM. SOC. VOL. 126, NO. 38, 2004 12069

A R T I C L E S
Fadeev et al.
Table 2. GaSz cis-trans Conformation RHF/6-31G(d)//
For the cis-cis geometry, the total energy was more sensitive
to the pucker in the left N-n-propylacetamido chain. The changes
in energy due to the pucker variations in the left and the right
N-n-propylacetamido branches were 0.3-4.4 kcal/mol and
0-1.9 kcal/mol, respectively, and these contributions to the total
energy were independent of each other to the precision of 1
kcal/mol. Also, the greater rigidity of the left N-n-propylaceta-
mido branch was supported by the facts that its “Râ” conforma-
tion was unstable and the corresponding gallium hydroxamate
five-membered ring was more distorted vs the other two five-
membered chelation rings.
B3LYP-6-31G(d) Energies
conformation
Hartrees
kcal/mol
RRRR
RRRâ
RRâR
RRââ
RγRR
RγRâ
RγâR
Rγââ
âRRR
âRRâ
âRâR
âRââ
ââRR
ââRâ
âââR
ââââ
-3443.329 729
-3443.338 531a
unstableb
3.16
0.0
unstableb
-3443.315 913
-3443.321 360
unstableb
11.82
8.41
unstableb
-3443.323 965
-3443.328 808
unstableb
6.77
3.74
In the cis-trans geometry, the right N-n-propylacetamido
chain (having the hydroxylamino oxygen in the trans- position
with respect to the citrate carboxyl) was more conformationally
constrained. Consequently, all structures with the right amide
carbonyl in the “â” conformation were unstable and automati-
cally converged to the “R” conformation. The “ââ” conforma-
tion of the left chain was also unstable. Energetic separations
between the cis-trans GaSz conformations (3.1-8.6 kcal/mol)
due to change of conformation on a single node (“R”, “â”, or
“γ”) were typically larger than those for the cis-cis configu-
ration (0-3.3 kcal/mol). Therefore, the computation provides
compelling evidence of a diminished conformational freedom
of the cis-trans GaSz configuration compared to the cis-cis
geometry.
unstableb
unstableb
unstableb
unstableb
unstableb
a B3LYP/6-31G(d)//B3LYP/6-31G(d) energy of this conformer is
-3443.348535 Hartrees. b Geometry converged to that of another in the
table or the Ga3+ center had lost its ligation to some of the chelating groups.
Table 3. GaSz Conformation Energies in cis-cis Configuration
B3LYP/6-31G(d)a
B3LYP/6-31G(d) dipolea,b
conformation
Hartrees
kcal/mol
Hartrees
kcal/mol
RRRR
RRRâ
RRγR
RRââ
RγRR
RâRâ
RγâR
Rγââ
âRRR
âRRâ
âRγR
âRââ
âγRR
âââR
âγââ**
ââââ
-3443.343320
unstable*
-3443.341621
-3443.343364
-3443.338458
unstable*
-3443.338458
-3443.339718
-3443.338223
unstable*
-3443.337988
-3443.339641
-3443.336341
-3443.338592
-3443.337257
-3443.339332
0.03
-3443.355297
0.54
1.09
0
3.08
-3443.346283
-3443.356151
-3443.343919
6.19
0
7.68
Onsager Model Solution Phase Calculations of GaSz. To
investigate the effect of the aqueous medium on the GaSz
structures and their energies, we used the Onsager dipole self-
consistent reaction field (SCRF) model54,55 at the B3LYP/6-
31G(d) level of theory. These calculations were confined to the
cis-cis geometry, since that geometry proved to be preferable
(see the following section). The cavity radius a0 was chosen at
5.80 Å as determined by the molecular volume calculation in
Gaussian9856 for the gas-phase geometry optimized B3LYP/6-
31G(d) cis-cis RRRR conformation. That cavity radius and
dielectric constant of 78.39 were then used in the SCRF
calculations for all conformations. Calculated energies of the
SCRF geometry-optimized structures are listed in Table 3. As
expected for the Onsager theory, the overall energy trend is such
that the change in the absolute energy upon placing the
molecules into the dielectric continuum is proportional to the
square of their dipole moments (Figure S21 and Table S1).
Relative energies of most conformers have increased upon
placement in the dielectric continuum with respect to the ground-
state structures, only those energies of the âRRR and âRââ
conformers have slightly decreased. Importantly, the lowest-
energy conformations in the gas phase remained the lowest-
energy conformations in the SCRF solution phase calculations,
3.08
2.29
3.23
-3443.343919
-3443.345997
-3443.352253
7.68
6.37
2.45
3.37
2.34
4.41
2.99
3.83
2.53
-3443.342807
-3443.353064
-3443.348225
-3443.348213
-3443.345997
-3443.349505
8.37
1.94
4.97
4.98
4.17
4.17
a Both structure optimization and single point energy calculations were
done with the same method. b Onsager dipole solvent reaction field (SCRF)
model, a0 ) 5.80 Å. *Geometry was found to converge to some other
hexacoordinated structure listed in the table or (for cis-trans configuration)
loose one of the hydroxamate Ga-O bonds during the geometry optimiza-
tion. **cis-cis structure labeled âγââ strongly resembles structure ââââ
with a slight difference of the amide group orientation; amide in âγââ is
hydrogen bonded to the Ga-bound carboxyl oxygen instead of the carbonyl
oxygen as in ââââ.
hydroxamates at the RHF level of theory was not accompanied
by a deterioration of accuracy in the relative conformation
energies.
Gas-Phase Calculations of GaSz. Thirteen unique confor-
mations were found for the cis-cis geometry; three more were
unstable. The cis-trans configuration yielded only six stable
conformations, as it was considerably more demanding on the
pucker of the right side N-n-propylacetamido fragment. The
B3LYP/6-31G(d) energies for the corresponding structures of
both configurations are presented in Tables 2 and 3. Several
conformations of the cis-cis geometry had indistinguishable
energies. The B3LYP/6-31G(d) conformation energies are nearly
additive with respect to the orientations of the chosen conforma-
tion-defining groups (i-iV in Figure S25b), further confirming
that such calculations produce chemically accurate energy
values.
(54) Onsager, L. J. Am. Chem. Soc. 1936, 58, 1486-1493.
(55) Wong, M. W.; Frisch, M. J.; Wiberg, K. B. J. Am. Chem. Soc. 1991, 113,
4776-4782.
(56) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M.
A.; Cheeseman, J. R.; Zakrzewski, V. G.; Montgomery, J. J. A.; Stratmann,
R. E.; Burant, J. C.; Dapprich, S.; Millam, J. M.; Daniels, A. D.; Kudin,
K. N.; Strain, M. C.; Farkas, O.; Tomasi, J.; Barone, V.; Cossi, M.; Cammi,
R.; Mennucci, B.; Pomelli, C.; Adamo, C.; Clifford, S.; Ochterski, J.;
Petersson, G. A.; Ayala, P. Y.; Cui, Q.; Morokuma, K.; Malick, D. K.;
Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Cioslowski, J.; Ortiz,
J. V.; Baboul, A. G.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.;
Komaromi, I.; Gomperts, R.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham,
M. A.; Peng, C. Y.; Nanayakkara, A.; Gonzalez, C.; Challacombe, M.;
Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M. W.; Andres, J. L.;
Gonzalez, C.; Head-Gordon, M.; Replogle, E. S.; Pople, J. A. Gaussian
98, version A.7; Gaussian, Inc.: Pittsburgh, PA, 1998.
9
12070 J. AM. CHEM. SOC. VOL. 126, NO. 38, 2004

+
Structures of Ga3 Siderophore Complexes
A R T I C L E S
consistent with their dipole moments being one of the largest
within the whole set of conformers.
the pro-S methylene group protons on the right side. The two
protons, one on chain I (resonance 2) and one on chain II
(resonance 2′), displaying NOE contact with the corresponding
terminal methyl group in Sz (Figures 3, S15, and Table 1), and
the R-methyne group in Af (Figure S16), do not exchange upon
racemization, and therefore must be both pro-S or both pro-R.
All conformations of the cis-cis configuration (Figures 5 and
S23) except the âRRR and âRââ strictly matched the above
observations. The latter two structures have calculated gas-phase
and SCRF energies of 2.6-3.2 kcal/mol and 1.94-2.45 kcal/
mol above the ground state conformation, respectively, and
therefore should not represent the GaSz structure at room
temperature. By contrast, all conformations of the cis-trans
configuration (Figures S24 and S25c) did not agree with the
observed NOE and racemization exchange patterns. Therefore,
only the cis-cis configuration is consistent with the experi-
mental data.
Determination of Favorable Octenoyl Side-Chain Orien-
tations in FeAf. Favorable orientations of Af octenoyl pendant
side chains were investigated by analysis of the calculated
torsional energy profiles for the two rotatable C-C bonds: one
(CCO-CR) was modeled on the GaSz structure with one of the
terminal methyl groups replaced by a 2-trans-methylvinyl group
(inset in Figure S18), and the second (Câ-Cγ) was modeled on
an analogous structure, modified with the 2-trans-propylvinyl
group (inset in Figure S17).
In the minimum energy CCO-CR torsional orientation, the
CdC bond is cisoid coplanar with respect to the hydroxamate
carbonyl. The transoid orientation is very unfavorable due to a
severe steric interaction of the vinyl group with the n-propyl
moiety. Rotation of the vinyl group out of the hydroxamate plane
is restricted by ∼5 kcal/mol due to the loss of conjugation with
the hydroxamate π-electron system. The most favorable orienta-
tion is in agreement with the NOE spectrum of GaAf (Figure
S16) showing strong NOE signals between the vinyl R-protons
(doublet at ∼δ 6.38) and low-field methylene protons of the
corresponding 1,3-diaminopropane fragment (multiplets at δ
4.47 and δ 4.35). The Câ-Cγ bond torsional energy profile
(Figure S17) has two energetically indistinguishable global
minima around the angles of (60°. One of the minimum-energy
conformations of the pendant GaAf octenoyl fragments is shown
in Figure 6. The two octenoyl tails are clearly oriented in the
opposite directions, separated by the average planar angle of
∼130° (for all minimum-energy conformations), unlike the
parallel side-chain arrangement in the phospholipid membrane
molecules.
Vesicle Fusion Induced by FeAf. TEM experiments were
performed to investigate the process of vesicle fusion in the
presence of Af and FeAf. Upon incubation of DMPC small
unilamellar vesicles (SUV) prepared as described in the
literature23 with 2.5 µM Af, some vesicle aggregation was
apparent, but a large percentage (50%) of vesicles retained a
normal structure (Figure S22). The fusion process was obviously
enhanced in the presence of 2.5 µM FeAf, with almost 100%
SUV (Figure S22) aggregated into large vesicle clusters and
with a significant amount of broken vesicles. Therefore, we
concluded that FeAf facilitates vesicle fusion and their disrup-
tion. That Af is considerably more hydrophobic than FeAf was
shown by the significant change in octanol-water partition
coefficient upon binding iron, 1.78 and 0.260, respectively.
1
Complete Structural H Chemical Shift Assignment and
Experimental Validation of the Favorable Conformations.
A first problem on the way to a full 1H NMR resonance
assignment was attributing the chain I and chain II resonances
to the two n-propyl fragments in the cis-cis GaSz structure.
Those two chains differ in the orientation of the corresponding
hydroxamate group with respect to the R-hydroxy carboxyl
moiety: one (the left side in Figure 4a) has the hydroxamate
carbonyl trans to the citrate hydroxyl and the another (the right
side in Figure 4a) is cis.
According to the NMR data, chain I is more rigid than chain
II and must adopt a single staggered conformation of the propyl
fragment. On the other hand, the gas phase and SCRF
calculations showed that the two lowest-energy conformations,
RRRR and RRââ, have identical pucker arrangements for the
left propyl fragment. Therefore chain I was assigned to the left
side of the structure.
1
The next task toward a full H NMR spectrum assignment
3
was verification of observed J spin-spin couplings with the
calculated GaSz geometries. The computed GaSz conformations
had the Haasnoot-model57 3J vicinal coupling values for the left
n-propyl chain in excellent agreement with the measured
couplings on chain I. By contrast, none of the conformations
had 3J vicinal coupling values that agreed well with the observed
splittings on chain II (Table 4). The obviously averaged vicinal
couplings observed in chain II (Figure 3 and Table 1) and the
absence of a clear diaxial spin-spin coupling pattern for either
of the chain II central methylene hydrogen atoms suggested a
conformational mixture. Indeed, there was an excellent agree-
ment between the experimental and aVerage calculated 3J
HCCH57 and HNCH58-61 coupling constants (Table 4, columns
2 and 3) for the two lowest energy conformations (RRRR and
RRââ, Table 3 and Figure 5).
Discussion
Synthesis. The general synthesis of acinetoferrin, schizokinen,
and their analogues described here features the use of a new
siderophore building block, the 1-N-benzoyloxy-1,3-diamino-
propane bis-hydrochloride. The method minimizes the risk of
iron contamination and could be adapted for solid phase and
combinatorial preparations.
The NMR data in the amide region are also consistent with
the RRRR and RRââ GaSz conformations. Remarkably, the ab
initio and DFT calculations predicted the intramolecular hy-
Coordination Geometry of GaSz. The preferred modes of
hydroxamate coordination of GaSz and, thereby, GaAf were
shown to be cis-cis in aqueous solution. The assignment
became clear after inspection of the racemization-induced
crosscorrelation and NOE-contact signals in the NOE spectra
(Figures S15 and S16, respectively). Racemization can only
exchange pro-R methylene group protons on the left side with
(57) Haasnoot, C. A. G.; Deleeuw, F.; Altona, C. Tetrahedron 1980, 36, 2783-
2792.
(58) Demarco, A.; Llinas, M.; Wuthrich, K. Biopolymers 1978, 17, 637-650.
(59) Gavrilov, Y. D.; Solkan, V. N.; Bystrov, V. F. Bull. Acad. Sci. USSR DiV.
Chem. Sci. 1975, 24, 2368-2373.
(60) Aubry, A.; Giessner. C.; Cung, M. T.; Marraud, M.; Neel, J. Biopolymers
1974, 13, 523-536.
(61) Ramachandran, G. N.; Chandrasekaran, R.; Kopple, K. D. Biopolymers
1971, 10, 2113-2131.
9
J. AM. CHEM. SOC. VOL. 126, NO. 38, 2004 12071

A R T I C L E S
Fadeev et al.
Table 4. Comparison of Measured 1H Vicinal Spin-Spin
Couplings with the Calculated Values for the Two Lowest Energy
cis-cis GaSz Conformers
calculated couplings for
energy-minimized structures
Jx
J
obs (Hz)
mean
RRRR
RRââ
-y
Chain I
0.8
12.1
6.4
1.1
1.1
6.3
12.2
1.2
5.57
4.15
6-4
7-4
6-2
7-2
5-6
3-6
5-7
3-7
1-3
1-5
-
0.8
12.2
6.5
1
0.9
11.9
6.3
1.1
1.1
6.1
12.1
1.2
6.15
2.4
10.47
6.59
1.58
--
6.09
10.47
2.06
8.8
1
6.4
12.3
1.1
5.0
6.29
2
Chain II
3.3
8.3
7.4
3.3
1.0
9.7
8.8
0.9
2'-7'
2'-6'
4'-7'
4'-6'
7'-5'
7'-3'
6'-5'
6'-3'
1'-3'
1'-5'
3.03
7.43
8.57
3.08
1.91
8.77
8.84
1.88
6.5
4
2.5
4.1
12.5
2.2
1.3
12.1
5.9
1.1
4.81
6.35
12.5
2.2
4.5
0.7
7.2
11.7
0.7
7.0
2.37
Figure 7. Plot of GaSz racemization ∆Gq vs temperature, T. Circles:
experimental data, solid line: plot of experimental data vs T at temperatures
below and above 28 °C, pH ) 7.
uration, and thus no ligand field stabilization, forms labile
complexes that racemize readily. There is only one known case
of chiral gallium hydroxamate resolution, which was only
achieved in aprotic solvent.63 The racemization reported here
(krac ≈ 1.4 s-1) is an order of magnitude slower than the known
racemizations of gallium tris-chelates.64-66 The unusually slow
racemization GaSz is clearly related to the rigidity of the
molecular framework and the two internal hydrogen bonds. Such
intramolecular hydrogen bonds have been previously suggested
to further stabilize the iron-siderophore complexes.67,68
Clearly, there are two regimes of racemization, depending
on the acidity (Figure 8). Above pH 5 this process is proton-
independent and has a biphasic temperature dependence of ∆Gq
(Figure 7). Only a mechanism with two or more consecutiVe
steps can explain the L-shaped racemization ∆Gq profile vs T.
Furthermore, the first step should be dissociative, manifested
by the positive activation entropy. In the alternative case with
two competing mechanisms, the low-entropy process would
dominate in the low-temperature regime, resulting in the
opposite ∆Gq profile vs T, where the smaller slope at lower T
would be followed by a steeper slope at higher T. The activation
parameters of the second step, rate limiting at slightly elevated
5.9
4.36
3.2
drogen bonds between the amide hydrogen atoms and metal-
bound hydroxamic and R-hydroxy carboxylic oxygen atoms
(Figure S25). These hydrogen bonds form despite inducing an
appreciable overall distortion of the tetrahedral carbon skeleton
angles. The insensitivity of N-H resonance 1 chemical shift to
pD, in contrast to that of resonance 1′, indicates that N-H 1
undergoes a slower exchange with water, in agreement with
the stronger corresponding intramolecular hydrogen bond, and
thus, a higher 1H chemical shift value (δ 9.43 for the resonance
1 vs 7.5 for the resonance 1′, Figure 3 and Table 1). The
calculated hydrogen bond lengths were also fully consistent with
the relative magnitudes of the corresponding amide proton
chemical shifts: the shorter hydrogen bond in chain I (1.77 Å
on average, δ 9.43) and the ∼0.1 Å longer one on chain II
(1.87 Å on average, δ 7.5).
Thus, the cis-cis configuration of GaSz, conformations
RRRR and RRââ, best agree with the experiments and represent
a structural model for the core fragment of the GaAf. Figure 5
shows a complete structural assignment of the methylene,62 the
amide, and the terminal methyl 1H NMR resonances for GaSz.
As can be seen, the two structures are very similar in the overall
arrangement. The left side has the same conformation in both
structures. The ring-pucker conformation of the right side
equilibrates between “RR” and “ââ”. Both amide protons are
hydrogen-bonded to one of the metal-coordinating oxygen
atoms. The planar angle between the two hydroxamate acetyl
C-C bonds in GaSz is 138° in the RRRR structure and 120° in
the RRââ structure, having a strong effect on the geometry of
a longer side-chain analogue, GaAf.
temperatures, likely involves a trigonal-prismatic rearrangement,
69-71
known as the Bailar twist,
since the measured activation
parameters, a modest ∆Hq and a low ∆Sq, are similar to the
(63) Brumaghim, J. L.; Raymond, K. N. J. Am. Chem. Soc. 2003, 125, 12066-
12067.
(64) Karpishin, T. B.; Stack, T. D. P.; Raymond, K. N. J. Am. Chem. Soc. 1993,
115, 6115-6125.
(65) Kersting, B.; Telford, J. R.; Meyer, M.; Raymond, K. N. J. Am. Chem.
Soc. 1996, 118, 5712-5721.
(66) Kersting, B.; Meyer, M.; Powers, R. E.; Raymond, K. N. J. Am. Chem.
Soc. 1996, 118, 7221-7222.
(67) Matsumoto, K.; Ozawa, T.; Jitsukawa, K.; Einaga, H.; Masuda, H. Inorg.
Chem. 2001, 40, 190-191.
(68) Tor, Y.; Libman, J.; Shanzer, A.; Felder, C. E.; Lifson, S. J. Am. Chem.
Soc. 1992, 114, 6661-6671.
The Mechanism of GaSz Racemization. As discussed
above, the dynamic racemization of GaSz was used to assign
the preferred geometry as cis-cis. Ga3+, having a d10 config-
(69) Casanova, D.; Cirera, J.; Llunell, M.; Alemany, P.; Avnir, D.; Alvarez, S.
J. Am. Chem. Soc. 2004, 126, 1755-1763.
(70) Montgomery, C. D.; Shorrock, C. J. Inorg. Chim. Acta 2002, 328, 259-
262.
(62) The citrate methylene 1H NMR resonances were the only signals not reliably
assigned to a particular conformation, but it is likely that the ab-quartet
centered at δ 2.45 belongs to the conformationally restrained chain I, and
the singlet at d 2.55 to the more flexible and equilibrating chain II.
(71) It is clear that the analogue of the Bailar twist mechanism (Figure S27 a
and b), if it is not preceded by dissociation of at least one chelating group,
cannot be operative in the GaSz epimerization. Such a process requires
rotation of all the chelating groups in the same direction, thus minimizing
9
12072 J. AM. CHEM. SOC. VOL. 126, NO. 38, 2004

+
Structures of Ga3 Siderophore Complexes
A R T I C L E S
Figure 9. Change of Af conformation upon coordinating the metal ion.
Figure 10. Dissociative mechanism of the GaSz epimerization.
groups and the steric repulsion from the corresponding n-
diaminopropyl fragments. The Câ-Cγ bond has two energeti-
cally equivalent favorable torsional angles. The resulting GaAf
model (Figures 6 and 9) shows that the two trans-octenoyl
groups are oriented in the opposite directions. These results
indicate a significant structural reorganization of Af upon metal
binding, since apo-Af can adopt many relaxed structures with
phospholipidlike parallel arrangements of the lipophilic side
chains while GaAf clearly cannot (Figure 9).
Figure 8. GaSz racemization rate constant vs pD: circles: 25 °C, squares:
15 °C.
values observed for such a mechanism for gallium catecholates
(Figure 10),65,66,72 8-hydroxyquinolinates,73 and acetylaceto-
nates.74
The dissociation in the first step at pH > 5 should occur at
the R-hydroxy carboxylate position, rather than at the hydrox-
amate position, since the latter is a much stronger ligand.
Further, the dissociation is most likely to occur at the carboxylate
position rather than the alkoxide position,75 since there is no
pH dependence for this process at pH > 5 and there is a strong
evidence76 that the carboxylate is a much more labile ligand to
gallium.
Below pH 4, the marked first-order [H+] dependence was
assigned to complete dissociation of the R-hydroxy-carboxyl
moiety followed by a more facile rearrangement of the bis-
hydroxamate intermediate. It is known that at pH ∼ 2-3 the
ionic charge of FeSz switches from -1 to +1 upon simultaneous
acceptance of two protons, while the resulting complex is best
described as a bis-hydroxamate.40
Geometry of the Alkyl Tails of GaAf and Biological
Implications. The unusual feature of the Af molecule is the
presence of two unsaturated side chains conjugated to the
hydroxamate carbonyl groups. Significantly, for all reasonable
structures of GaAf and GaSz, the two substituent alkenyl or
alkyl groups are held rigidly in an antiparallel arrangement at
an angle of about 130°. The CCO-CR bond of the octenoyl
fragment is rotationally constrained by the π-electronic conjuga-
tion of the vinyl fragments with the hydroxamate carbonyl
Trivalent gallium and iron citrates have very similar structures
and share the same bidentate coordination mode with the
R-hydroxy carboxylic moiety.76,77 Such structural similarity
between Fe3+ and Ga3+ complexes is typical, for example, for
tripodal salicylate-based ligands78 and has been used in structural
studies of several siderophores.79-81 Thus, we anticipate that
FeAf and FeSz will have structures very similar to their gallium
analogues. The large 8-fold change in partition coefficient upon
binding iron indicates that Af is considerably more hydrophobic
than FeAf, despite the facts that metal binding buries six
hydrophilic oxygen atoms and there is no net change in
molecular charge. We suggest that the skewed orientation of
the octenoyl side chains that is enforced by iron binding is the
origin of this change. Significantly, a similar change in apparent
hydrophobicity upon iron binding was reported by us recently
for the amphiphilic marine siderophore, marinobactin E,23
suggesting that such differences are part of a larger iron
acquisition strategy. Thus, membrane binding and lipid traf-
ficking would provide directed avenues for iron extraction and
delivery. The facility with which FeAf was observed to disrupt
phospholipid vesicles, despite its greater hydrophilicity, and
ongoing studies of membrane binding and permeability of a
range of amphiphilic siderophores support this view.82
the unfavorable nonbonding interactions in the transition state. Such a
process, if applied to the GaSz, would lead to an unstable trans-trans isomer
(Figure S27 b). Thus, a nondissociative epimerization of GaSz would require
a more complex and energetically prohibitive pathway.
(72) Raymond, K. N.; Dertz, E. A.; Kim, S. S. Proc. Natl. Acad. Sci. U.S.A.
2003, 100, 3584-3588.
(77) Matzapetakis, M.; Raptopoulou, C. P.; Tsohos, A.; Papaefthymiou, V.;
Moon, N.; Salifoglou, A. J. Am. Chem. Soc. 1998, 120, 13266-13267.
(78) Cohen, S. M.; Petoud, S.; Raymond, K. N. Inorg. Chem. 1999, 38, 4522-
4529.
(73) Gromova, M.; Jarjayes, O.; Hamman, S.; Nardin, R.; Beguin, C.; Willem,
R. Eur. J. Inorg. Chem. 2000, 545-550.
(74) Hutchison, J. R.; Gordon, J. G.; Holm, R. H. Inorg. Chem. 1971, 10, 1004-
1017.
(79) Wasielewski, E.; Atkinson, R. A.; Abdallah, M. A.; Kieffer, B. Biochemistry
2002, 41, 12488-12497.
(75) Additional evidence that the dissociation of carboxylate is favorable was
obtained by ab initio modeling of the dissociative transition state geometry
at the RHF/3-21G level of theory (Figure S26). In the transition state Ga3+
essentially has tetragonal prismatic coordination with the two hydroxamate
five-membered chelation rings in a nearly parallel arrangement. This
transition state geometry permits a simultaneous binding of an additional
ligand, such as H2O, to the distal part of the molecule further stabilizing
the transition state compared to the nondissociative mechanism.
(76) Hawkes, G. E.; O’Brien, P.; Salacinski, H.; Motevalli, M.; Abrahams, I.
Eur. J. Inorg. Chem. 2001, 1005-1011.
(80) Atkinson, A. R.; Din, A. L. M. S.; Keiffer, B.; Lefevre, J.-F.; Abdallah,
M. A. Biochemistry 1998, 37, 15965-15973.
(81) Stephan, H.; Freund, S.; Meyer, J. M.; Winkelmann, G.; Jung, G. Liebigs
Ann. Chem. 1995, 43-48.
(82) Our preliminary results show rapid paramagnetic NMR line broadening of
DMPC choline methyl groups on the interior surface of the DMPC small
unilamellar vesicles upon mixing with FeAf solution. This indicates that
FeAf is capable of rapidly diffusing across the lipid layers. A further account
of these phenomena will be reported elsewhere.
9
J. AM. CHEM. SOC. VOL. 126, NO. 38, 2004 12073

A R T I C L E S
Fadeev et al.
13C NMR (CDCl3) δ: 13.92, 22.36, 27.03, 27.73, 27.80, 31.22, 32.45,
36.39, 44.07, 46.15, 73.98, 82.66, 117.79, 126.53, 129.01, 130.90,
134.65, 149.89, 164.63, 166.91, 169.99, 172.67.
Experimental Section
Synthesis. Solvents were freshly distilled prior to use. Sephadex
LH-20 and G-10 was obtained from the Sigma Chemical Company.
Silica gel 60 was purchased from Sorbent Technologies. Crotonyl
chloride, trans-2-octenoic acid, and trans-2-dodecenoic acid were
purchased from Lancaster Synthesis, anhydrous citric acid from Fisher
Scientific, 1,2-Dimyristoyl-sn-Glycero-3-Phosphocholine (DMPC) from
Avanti Lipids, and all other reagents from Aldrich Chemicals. All
reagents were used without further purification, unless otherwise noted.
ESI-MS: m/z 849 (MH+), 871 (MNa+).
4c: 1H NMR (CDCl3) δ: 1.48 (s, 9H, tert-butyl), 1.85(m, 10H, CH3
and CH2), 2.70 (ab-quartet, 4H, citrate CH2), 3.37 (m, 4H, CH2), 3.93
(m, 4H, CH2), 6.08 (d, 2H, CH), 7.05(m, 2H, CH), 7.52 (t, 4H,
aromatic), 7.68 (t, 2H, aromatic), 8.11(d, 4H, aromatic).
13C NMR (CDCl3) δ: 18.32, 27.01, 27.80, 36.41, 44.02, 46.15, 73.97,
82.66, 119.28, 126.49, 129.02, 130.11, 134.68, 144.99, 164.63, 166.63,
170.01, 172.67.
1-N-Benzoyloxy-1,3-diaminopropane Dihydrochloride (2). N-(Ben-
zoyloxy)-3-(tertbutoxycarbonylamino)propylamine43 (3.7 g, 12.5 mmol)
was dissolved in 250 mL of ether and 50 mL of methanol. Anhydrous
HCl was bubbled through the solution for 1 h without cooling. After
removal of the solvent, 3.1 g of 2 was collected as a white solid (92%
yield). The presence of methanol was essential for complete removal
of the Boc group.
1H NMR (CD3OD) δ: 2.16 (m, 2H,CH2), 3.18 (t, 2H, CH2), 3.57 (t,
2H, CH2), 7.57 (d, 2H, aromatic), 7.75 (m, 1H, aromatic), 8.1 (m, 2H,
aromatic).
ESI-MS: m/z 759 (MNa+).
4d: 1H NMR (CDCl3) δ: 0.82 (t, 6H, CH3), 1.23 (m, 24H, CH2),
1.39 (m, 4H, CH2), 1.48 (s, 9H, tert-butyl), 1.87(m, 4H, CH2), 2.17
(m, 4H, CH2), 2.71 (ab-quartet, 4H, citrate CH2), 3.40 (m, 4H, CH2),
3.96(m, 4H, CH2), 6.08(d, 2H, CH), 7.05m, 2H, CH), 7.12 (t, 2H,
amide), 7.57 (t, 4H, aromatic), 7.73 (t, 2H, aromatic), 8.14(d, 4H,
aromatic).
2-tert-Butyl-1,3-((3-N-hydroxy-3N-acyl)propyl)diamide Citrate
(5). General Procedure. A 0.65 mmol amount of 2 was dissolved in
20 mL of methanol in a polyethylene container loaded with a Teflon-
coated magnetic stirrer. The solution was cooled to 0 °C and 287 µL
of 5 N NaOH (2.2 equiv) in water was added. After 5 min 287 µL of
5 N HCl was added. Methanol and water were removed on a rotary
evaporator and the residue was subjected to column chromatography
(methanol/chloroform) on pretreated silica as described in ref 43. Yields
after purification were 80%, (94%) for 5b.
5a: 1H NMR (CD3OD) δ: 1.48 (s, 9H, tert-butyl), 1.80 (t, 4H, CH2),
2.10 (s, 6H, CH3), 2.62 (ab-quartet, 4H, citrate CH2), 3.19 (t, 4H, CH2),
3.62 (t, 4H, CH2).
13C NMR (CD3OD) δ: 20.44, 27.66, 28.33, 37.74, 46.26, 46.66,
75.32, 83.77, 172.11, 173.86, 174.61.
5b: 1H NMR (CD3OD) δ: 0.90 (t, 6H, CH3), 1.35 (m, 8H, CH2),
1.48 (m, 13H, CH2 and tert-butyl), 1.82(m, 4H, CH2), 2.23 (m, 4H,
CH2), 2.65 (ab-quartet, 4H, citrate CH2), 3.20 (m, 4H, CH2), 3.72 (t,
4H, CH2), 6.61 (d, 2H, CH), 6.85 (m, 2H, CH).
13C NMR (CD3OD) δ: 14.49, 23.61, 27.72, 28.32, 29.05, 29.29,
32.61, 33.52, 37.82, 45.19, 46.99, 75.28, 83.77, 120.51, 148.36, 168.61,
172.15, 174.49, 176.91.
ESI-MS: m/z 641 (MH+), 663 (MNa+).
13C NMR (CD3OD) δ: 23.41 (CH2), 39.15 (CH2), 49.20 (CH2),
130.28 (CH, aromatic), 130.95 (CH, aromatic), 134.37 (CH, aromatic),
135.93 (C, aromatic), 166.38 (CO, carbonyl).
Anal. Calcd for C10H16Cl2N2O2: C, 44.96; H, 6.04; Cl, 26.54; N,
10.49. Found: C, 44.70; H, 5.94; Cl, 26.41; N, 10.24.
2-tert-Butyl-1,3-((3-N-benzoyloxy)propyl)diamide Citrate (3). Pro-
tected di-N-hydroxysuccinimidyl citrate42 1, (3 g, 6.78 mmol) was
dissolved in 50 mL of dry acetonitrile. To this was added 3.27 g (12.2
mmol) of solid 2. After that, 5.47 mL (39.3 mmol) of triethylamine in
50 mL of dry acetonitrile was added dropwise to the stirred above
suspension. The reaction mixture was stirred at 25 °C for 1 h. After
removing the solvent the reaction mixture was partitioned between
chloroform and water, and the organic layer evaporated to dryness.
The crude product was purified by silica column chromatography with
4%CH3OH/CHCl3 used as eluent. Yield: 2.22 g (55%).
1H NMR (CDCl3) δ: 1.46 (s, 9H, tert-butyl), 1.83 (m, 4H, CH2),
2.65 (ab-quartet, 4H, citrate CH2), 3.2 (t, 4H, CH2), 3.41 (t, 4H, CH2),
6.9 (t, 2H, amide), 7.45 (t, 4H, aromatic), 7.59 (t, 2H, aromatic), 8.00
(d, 4H, aromatic).
13C NMR (CDCl3) δ: 26.73 (CH2, 1,3-diaminopropane), 27.49
(CH3), 36.8 (CH2), 43.63 (CH2), 49.53 (CH2, 1,3-diaminopropane),
74.00 (C, tertiary), 82.10 (C, tertiary), 128.10, 128.24, 128.97, 133.07,
166.33, 170.06, 172.58.
5c: 1H NMR (CD3OD) δ: 1.48 (s, 9H, tert-butyl), 1.82 (t, 4H, CH2),
1.90 (ab-quartet, 6H, CH3), 2.64 (ab-quartet, 4H, citrate CH2), 3.18
(t(broad), 4H, CH2), 3.7 (t(broad), 4H, CH2), 6.63(d(broad), 4H, CH),
6.86(m, 2H, CH).
HRMS (EI) m/z calcd for C30H41N4O9 (M+): 601.2874. Found:
13C NMR (CD3OD) δ: 18.44, 27.69, 28.29, 37.76, 45.29, 46.89,
75.27, 83.74, 121.96, 143.36, 168.62, 172.12, 174.52, 176.87.
ESI-MS: m/z 529 (MH+), 551 (MNa+).
601.2898.
2-tert-Butyl-1,3-((3-N-benzoyloxy-3N-acyl)propyl)diamide Citrate
(4). General Procedure. 2-tert-Butyl-1,3-((3-N-benzoyloxy)propyl)-
diamide citrate 3, 0.9 g (1.5 mmol), was dissolved in 20 mL of dry
dichloromethane. The solution was heated to reflux temperature and
2.1 equiv of acyl chloride in 10 mL of dry dichloromethane was added.
The reaction mixture was refluxed for 1 h. After evaporating the solvent
the oily residue was subjected to column chromatography (silica/ethyl
acetate-hexanes). Yields after purification were 0.82 g (80%) for 4a,
0.98 g (77%) for 4b, 0.82 g (75%) for 4c, and 1.12 g (78%) for 4d.
4a: 1H NMR (CD3OD) δ: 1.42 (s, 9H, tert-butyl), 1.83 (m, 4H,
CH2), 2.05 (broad s, 6H, CH3), 2.6 (ab-quartet, 4H, citrate CH2), 3.25
(t, 4H, CH2), 3.87 (t, 4H, CH2), 7.59 (t, 4H, aromatic), 7.78 (t, 2H,
aromatic), 8.15 (d, 4H, aromatic).
13C NMR (CDCl3) δ: 20.32 (CH3), 26.94, 27.80, 36.36, 44.11, 45.80,
73.94, 82.74, 126.45, 129.02, 130.05, 134.73, 164.57, 169.93, 172.77.
4b: 1H NMR (CDCl3) δ: 0.82 (t, 6H, CH3), 1.25 (m, 8H, CH2),
1.40 (m, 4H, CH2), 1.48 (s, 9H, tert-butyl), 1.87(m, 4H, CH2), 2.15
(m, 4H, CH2), 2.75 (ab-quartet, 4H, citrate CH2), 3.40 (m, 4H, CH2),
3.93 (m, 4H, CH2), 6.04 (d, 2H, CH), 7.02(m, 2H, CH), 7.52 (t, 4H,
aromatic), 7.69 (t, 2H, aromatic), 8.12(d, 4H, aromatic).
5d: 1H NMR (CD3OD) δ: 0.90 (t, 6H, CH3), 1.32 (m, 28H, CH2),
1.45 (s, 9H, tert-butyl), 1.82(m, 4H, CH2), 2.23 (m, 4H, CH2), 2.62
(ab-quartet, 4H, citrate CH2), 3.19 (m, 4H, CH2), 3.73 (t, 4H, CH2),
6.61 (d, 2H, CH), 6.85 (m, 2H, CH).
ESI-MS: m/z 753 (MH+), 775 (MNa+).
Schizokinen, Acinetoferrin, and Analogues (6a-d). tert-Butyl
deprotection was effected as in Wang and Phanstiel,43 with TFA at
room temperature. After the deprotection was complete, TFA was
removed on a rotovap and finally under high vacuum. No heating was
used at any time during TFA removal. Compounds 6b and 6d could
be separated from the residual TFA by partition between the water
and the chloroform and evaporation of the chloroform phase, while
compounds 6a and 6c could be separated by repetitive lyophilization
from aqueous solutions. Compounds 6a and 6c were purified by
sephadex G-10 gel filtration, eluting with water, compounds 6b and
6d - by sephadex LH-20 eluted with CHCl3:ethanol (15:85).
6a (Sz): 1H NMR (D2O) δ: 1.69 (m, 4H, CH2), 2.1 (s, 6H, CH3),
2.61(ab-quartet, 4H, citrate CH2), 3.17 (m, 4H, CH2), 3.63 (t, 4H, CH2).
9
12074 J. AM. CHEM. SOC. VOL. 126, NO. 38, 2004

+
Structures of Ga3 Siderophore Complexes
A R T I C L E S
13C NMR (CD3OD) δ: 20.43, 27.72, 37.76, 46.28, 46.64, 75.36,
172.17, 173.85, 177.21.
The racemization rate constant was determined by fitting the temporal
buildup of the exchange peak and the parent peak integral intensity
ratio to a quadratic polynomial a t2 + b t. The fitting parameter b was
used for the rate constant value. The quadratic polynomial produced
better fitting results than linear because the initial linear buildup period
was too short. The ∆Gq for the dynamic racemization of GaSz was
calculated by the Eiring equation using the determined rate constants
in the temperature range of 10-45 °C at pD ) 8.5.
Computations. The Gaussian 9856 package was used for all
computations. All torsional angle energy profiles and structural
geometry optimizations were performed with RHF/6-31G(d) and
B3LYP/6-31G(d)85-87 methods. Solution-phase computations were done
with the Onsager dielectric continuum SCRF model54,55 at the continuum
dielectric constant value of 78.39. The cavity radius a0 ) 5.80 Å was
used as determined by the molecular volume calculation in Gaussian98.
The molecular structure figures were prepared with the MolMol
software.88
6b (Af): 1H NMR (CD3OD) δ: 0.94 (t, 6H, CH3), 1.35 (m, 8H,
CH2), 1.48 (m, 4H, CH2), 1.82 (m, 4H, CH2), 2.14 (m, 4H, CH2), 2.69
(ab-quartet, 4H, citrate CH2), 3.20 (t, 4H, CH2), 3.73 (t, 4H, CH2), 6.62
(broad d, 2H, CH), 6.83 (m, 2H, CH).
13C NMR (CD3OD) δ: 14.42, 23.55, 27.35, 29.07, 29.32, 32.65,
33.53, 37.77, 45.23, 47.04, 75.28, 120.54, 148.42, 168.58, 172.31,
177.25.
Anal. Calcd for C28H48N4O9: C, 57.52; H, 8.27; N, 9.58; O, 24.63.
Found: C, 57.60; H, 8.42; N, 9.34.
6c: 1H NMR (CD3OD) δ: 1.81 (m, 4H, CH2), 1.93 (d, 6H, CH3),
2.70 (ab-quartet, 4H, citrate CH2), 3.20 (t, 4H, CH2), 3.61 (t, 4H, CH2),
6.63 (d, 2H, CH), 6.85 (m, 2H, CH).
13C NMR (CD3OD) δ: 18.43, 27.65, 37.78, 45.24, 46.87, 75.25,
121.98, 143.32, 168.67, 172.24, 177.25.
ESI-MS: m/z 473 (MH+), 495 (MNa+).
6d: 1H NMR (CD3OD) δ: 0.90 (t, 6H, CH3), 1.29 (m, 24 H, CH2),
1.47 (m, 4H, CH2), 1.82 (m, 4H, CH2), 2.24 (m, 4H, CH2), 2.68 (ab-
quartet, 4H, citrate CH2), 3.20 (t, 4H, CH2), 3.70 (t, 4H, CH2), 6.62
(broad d, 2H, CH), 6.84 (m, 2H, CH).
Transmission Electron Microscopy. TEM images (Figure S22)
were obtained with a JEOL 100C electron microscope using uranyl
acetate (1.5 wt %) as the negative staining reagent. Freshly prepared
1,2-dimyristoyl-sn-glycero-3-phosphocholine (DMPC) SUV (0.05 mM,
final concentration) was incubated for 12 h at 25 °C with blank buffer,
Af (1:20, final molar ratio between apo-Af and DMPC) and ferric-
acinetoferrin, FeAf (1:20, final molar ratio between FeAf and DMPC).
A thin layer of mixed solution was then deposited onto a gold-coated
grid. After being stained by aqueous uranyl acetate (1.5 wt %), the
grid was dried in the air at room temperature for 15 min and then
subjected to TEM image analysis. DMPC SUV in the absence of either
Af or FeAf could be distributed homogeneously on the gold-coated
grid, while DMPC SUV incubated with either Af or FeAf exhibited a
heterogeneous distribution. Therefore, the images of the latter were
recorded by selecting vesicle-aggregated regions. Partition coefficients
were determined at 25 °C by the standard octanol-water procedure.
13C NMR (CD3OD) δ: 14.62, 23.85, 27.76, 30.44, 30.55, 30.65,
30.78, 33.16, 33.84, 45.26, 47.31, 75.25, 120.53, 148.34, 168.62, 172.28,
177.19.
Anal. Calcd for C36H64N4O9: C, 62.04; H, 9.26; N, 8.04; O, 20.66.
Found: C, 62.30; H, 9.55; N, 7.70.
Iron Acinetoferrin Complex (FeAf). Acinetoferrin 6c, 1 equiv from
2 mM solution in 100 mM pH 7.4 HEPES (aqueous + 20% vol. of
DMF or methanol to solubilize the acinetoferrin) buffer was mixed
with 1 equiv of 2 mM (in Fe concentration) solution of ferric ammonia
citrate in 100 mM pH 7.4 HEPES buffer and kept in a dark place at 25
°C for 24 h to allow for slow iron acquisition from ferric citrate clusters
to complete.
Iron and Gallium Schizokinen Complexes (FeSz and GaSz).
Schizokinen 6a, 1 equiv from 2 mM aqueous solution was mixed with
1 equiv of an aqueous solution of ferric nitrate (or GaBr3). The resulting
solution was slowly neutralized by aqueous NaOH.
NMR Spectroscopy. All NMR spectra were recorded on Varian,
Inc. 300, 400, and 500 MHz instruments. Deuterated solvents were
purchased from Cambridge Isotope Laboratories, Inc. The 1H 1D,
COSY and NOESY spectra of GaSz were recorded in D2O using
conventional pulse sequences, and in a 9:1 H2O:D2O mixture using
the corresponding pulse sequences modified with a WET water
suppression pulse train.83 GaSz racemization rates were measured with
a NOESY-1D experiment in H2O:D2O 1:1 without solvent suppression
using a double pulsed field gradient spin-echo (DPFGSE) selective
multiplet excitation technique.45 The methylene proton signal at δ 1.5
was selectively inverted and integral intensities of that multiplet and
of a signal at δ 2.2 were recorded as a function of mixing time. The
mixing time was within a range of 0.025-0.5 s. The sample pH was
adjusted by adding small volumes (∼10 µL) of 0.1M DCl or NaOD
solutions into the sample NMR tube and the pD was determined with
an Aldrich NMR pH electrode. A 0.2 pH unit correction was applied
to a pH meter reading to get the sample acidity instead of the suggested
0.4 since the latter value was determined for the pure D2O solutions.84
Acknowledgment. We thank Prof. Robert Pascal and Dr.
Adrzej Jarzecki for helpful discussions on quantum calculations,
Dr. Istvan Pelczer and Dr. Carlos Pacheco for assistance with
the NMR experiments, and Dr. Joseph Goodhouse for assistance
with TEM imaging. Financial support of this research by the
National Science Foundation (CHE-0221978), through the
Environmental Molecular Science Institute (CEBIC) at Princeton
University is gratefully acknowledged.
Supporting Information Available: Mass and NMR spectra,
charts with calculated torsion energy profiles, TEM images,
images and coordinate tables of calculated structures, table with
calculation parameters, and a scheme of the GaSz epimerization
mechanism. This material is available free of charge via the
Internet at http://pubs.acs.org.
JA048145J
(84) Glasoe, P. K.; Long, F. A. J. Phys. Chem. 1960, 64, 188-190.
(85) Becke, A. D. Phys. ReV. A 1988, 38, 3098-3100.
(86) Lee, C. T.; Yang, W. T.; Parr, R. G. Phys. ReV. B 1988, 37, 785-789.
(87) Vosko, S. H.; Wilk, L.; Nusair, M. Can. J. Phys. 1980, 58, 1200-1211.
(88) Koradi, R.; Billeter, M.; Wuthrich, K. J. Mol. Graphics 1996, 14, 51-55.
(83) Smallcombe, S. H.; Patt, S. L.; Keifer, P. A. J. Magn. Reson., Ser. A 1995,
117, 295-303.
9
J. AM. CHEM. SOC. VOL. 126, NO. 38, 2004 12075