Biomimetic ferrichrome: Structural motifs for switching between narrow- and broad-spectrum activities in P. putida and E. coli

Article abstract of DOI:10.1039/c5dt02685g

A series of novel ferrichrome (FC) analogs was designed based on the X-ray structure of FC in the FhuA transporter of Escherichia coli. Two strategies were employed: the first strategy optimized the overall size and relative orientation of H-bonding inter

Full text of DOI:10.1039/c5dt02685g

Dalton
Transactions
View Article Online
View Journal
PAPER
Biomimetic ferrichrome: structural motifs for
switching between narrow- and broad-spectrum
Cite this: DOI: 10.1039/c5dt02685g
activities in P. putida and E. coli†
Evgenia Olshvang,‡a Agnieszka Szebesczyk,‡b Henryk Kozłowski,b Yitzhak Hadar,c
Elzbieta Gumienna-Kontecka*b and Abraham Shanzer*a
A series of novel ferrichrome (FC) analogs was designed based on the X-ray structure of FC in the FhuA
transporter of Escherichia coli. Two strategies were employed: the first strategy optimized the overall size
and relative orientation of H-bonding interactions. The second strategy increased H-bonding interactions
by introducing external H-donors onto analogs’ backbone. Tris-amino templates were coupled to succi-
nic or aspartic acid, and the second carboxyl was used for hydroxamate construction. Succinic acid pro-
vided analogs without substituents, whereas aspartic acid generated analogs with external amines (i.e.
H-donors). All analogs had similar physicochemical properties, yet the biological activity in Pseudomonas
putida and E. coli showed great variation. While some analogs targeted specifically P. putida, others were
active in both strains thus exhibiting broad-spectrum activity (as in native FC). Narrow-spectrum or
species-specificity might find application in microbial diagnostic kits, while broad-spectrum recognition
may have advantages in therapeutics as siderophore–drug conjugates. The differences in the structure
and range of microbial recognition helped us in formulating guidelines for minimal essential parameters
required for inducing broad-spectrum activity.
Received 15th July 2015,
Accepted 3rd September 2015
DOI: 10.1039/c5dt02685g
www.rsc.org/dalton
Ferrichrome14–16 (FC; Chart 1a) is a prototype of the natural
hydroxamate siderophores. It is composed of a chiral hexa-
Introduction
Iron is an essential trace metal for virtually all life forms,1–3 peptide ring template, non-symmetrically extended by three
and crucial for a wide variety of cell functions, ranging from identical arms that are terminated by hydroxamate moieties.
oxygen metabolism and electron transfer processes to DNA FC, similarly to most siderophores, is not species-specific; it
and RNA synthesis. However, under the oxidizing earth atmos- exhibits broad-spectrum activity and thus can be recognized by
phere iron forms insoluble hydroxyl polymers inaccessible to various types of microorganisms.17,18 Among the various pro-
microorganisms.
teins that control FC transport inside the cell, the outer mem-
To obtain iron, microorganisms have developed a unique brane transporter is responsible for the selective recognition of
iron acquisition system that strictly controls iron uptake.4–6 It FC.19,20 Ferric hydroxamate uptake protein component A
consists of (i) bio-synthesis and secretion of iron chelators, (FhuA) is a FC transporter from the outer membrane of Escher-
called siderophores,7 capable of solubilizing and sequestering ichia coli and is among the few membrane transporters with a
iron from ferric hydroxyl polymers, and (ii) a specific transport resolved crystal structure.21,22 The crystallographic data indi-
system that recognizes the siderophore-iron complex and cate the presence of two sets of H-bonding networks that domi-
transports it into the cytoplasm.8–13
nate the recognition process. The first H-bonding network
links between the oxygen of the hydroxamate groups and
proton-donors within the transporter, and the second network
connects the amide-carbonyls in the FC peptidic skeleton with
the transporter.
Biomimetic chemistry is a discipline that aims to mimic or
reproduce biological activity, rather than reproduce the
detailed structure of a substrate.23–26 Extensive studies on bio-
mimetic tris-hydroxamate siderophores27–30 established that:
(i) the cyclic hexapeptide ring of FC can be replaced by a C3
symmetric template;31–33 and (ii) the presence of the hydrox-
aDepartment of Organic Chemistry, The Weizmann Institute of Science,
Rehovot 76100, Israel. E-mail: abraham.shanzer@weizmann.ac.il
bFaculty of Chemistry, University of Wrocław, F. Joliot-Curie 14, 50-383 Wrocław,
Poland. E-mail: elzbieta.gumienna-kontecka@chem.uni.wroc.pl
cDepartment of Plant Pathology and Microbiology, The R.H. Smith Faculty of
Agriculture Food and Environment, The Hebrew University of Jerusalem,
Rehovot 76100, Israel
†Electronic supplementary information (ESI) available. See DOI: 10.1039/
c5dt02685g
‡These authors contributed equally to this work.
This journal is © The Royal Society of Chemistry 2015
Dalton Trans.

View Article Online
Paper
Dalton Transactions
Chart 1 (a) Chemical structure of FC, which consists of a chiral hexapeptide ring (marked in blue) that is non-symmetrically extended by three
identical arms terminated by hydroxamate groups. (b) Chemical structure of FC analogs consisting of tris-amino derived templates (marked in blue)
extended with succinic acid (T1 and T4) or L/D-Asp (T2(α) and T3(β)), all terminated by retrohydroxamate groups.
amate moiety (which coordinates the ferric ion in native FC) is of a small library of FC biomimetics (Chart 1b) whose struc-
an essential feature in the siderophore design. Reversing the tures result from the implementation of both proposed
directionality of the hydroxamate groups to produce retro- methodologies. Coupling succinic acid to the template
hydroxamate (an artificial FC) has little biological effects; the through one of its carboxyls, while the second is used for the
activity of the FC retro-isomer is practically indistinguishable construction of the retrohydroxamate moiety (via hydroxyl-
from that of native FC.34,35
amination) provides a backbone void of substituents on the
Initial attempts by our group to mimic FC led to the con- basic skeleton, as represented by analogs T1 and T4. The
struction of analogs utilizing simple modular building blocks: incorporation of Asp through its carboxylic acid provides
a tris-carboxylate derived template extended with chelating analogs with polar amine side groups. Moreover, the way Asp
arms (based on chiral aliphatic amino acids) and linked to binds to the template (through the α- or β-carboxyl) provides
retrohydroxamate.36–38 This approach had several advantages: analogs with the amine at different positions; either α- or β
(i) synthetic simplicity; (ii) accessibility to enantio-pure iron with respect to the hydroxamate moiety and the iron center, as
complexes, which enabled us to explore the transporter’s in analogs T2 and T3. Using L/D-Asp in the synthesis of T2 and
stereo-specific preferences; and (iii) identification of structural T3 introduces a chiral center into the backbones of the
motifs on the basic skeleton, such as sites in close proximity analogs. Since iron is an additional chiral center, diastereo-
to the hydroxamate that can enable narrow-spectrum reco- meric pairs for each ligand can be obtained, forming a total of
gnition. However, the low overall number of species-specific four Asp structural isomers.
analogs capable of promoting bacterial growth, the lack of
This paper describes the synthesis of analogs T1–T4, their
analogs specific to E. coli, and the fact that no analog pro- physicochemical properties, and their growth promotion
moted activity across multiple species, unlike native FC, activity on two reference organisms, Pseudomonas putida and
remained major challenges.
Escherichia coli. While some derivatives were found to be
A new approach in designing FC analogs can possibly meet highly specific for P. putida, the succinic acid derivative T4
these challenges. In this design, two complementary methodo- showed a broader activity, and was capable of inducing growth
logies can be envisioned. The first is based on developing in both species. The new tripodal analogs are also compared
analogs possessing unsubstituted arms, as in the native FC with other structurally similar analogs in terms of bacterial
structure. The second methodology is based on enhancing the growth promotion. We also show how systematic modification
H-bonding networks (seen in the high resolution X-ray diffrac- of their backbones can lead to mimics with broader biological
tion analysis of the FhuA transporter with FC)21,22 by introdu- activity.
cing polar side groups that may serve as potential H-donors.
Structural optimization of each analog’s backbone and
improved fitting of the analogs to the FhuA binding site might
direct siderophore mimics toward recognition by new bacterial
species (particularly E. coli) or even provide broad-spectrum
Results and discussion
Design and synthesis
recognition. While narrow-spectrum activity offers the poten- Two types of tri-amino templates derived from commercially
tial to develop fast selective diagnostic tools,39–41 broad- available triols were prepared, both based on a quaternary
spectrum activity may have therapeutic advantages, as side- carbon center connected to three hydroxymethyl groups,
rophores have no mammalian targets and are therefore non- which were elongated with either an ethylene (compound 15)
toxic. Siderophore drug-conjugates might hold the promise of or a propylene (compounds 7 and 11) spacer (Scheme 1). In
novel antimicrobial agents.42–46
compounds 11 and 15, the fourth substituent on the quatern-
The incorporation of succinic or aspartic (Asp) acids into a ary carbon (the apical site) was a methyl group. In the case of
tris-amino template provides the means for the construction compound 7, the inert apical methyl group was replaced by an
Dalton Trans.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Dalton Transactions
Paper
Scheme 2 Synthesis of the succinic acid hydroxamate derivative (the
“chelating arm” for analogs T1 and T4). Reaction conditions: (a) Boc,
triethylamine (NEt3), THF/H2O, r.t.; (b) NaH, methyl iodide, DMF, r.t.;
(c) 4 N HCl in dioxane, 0 °C → r.t.; (d) succinic anhydride, K2CO3,
dioxane, r.t. For full experimental details, see the ESI.†
Scheme 1 Synthesis of the tripodal templates 7, 11 and 15. Reaction
conditions: (a) 40% KOH (aq.), acrylonitrile, dioxane, 0 °C → room temp-
erature (r.t.); (b) 1 M NaHCO3 (aq.), benzyl chloroformate, dioxane 0 °C
→ r.t.; (c) CoCl2·6H2O, di-tert-butyl dicarbonate (Boc), NaBH4, methanol
(MeOH), 0 °C → r.t.; (d) 10% Pd/C, H2, ethanol (EtOH), r.t.; (e) allyl chloro-
formate, 4 M NaOH (aq.), dioxane, 0 °C → r.t.; (f) 50% trifluoroacetic
acid (TFA) in CH2Cl2, triisopropylsilane, 0 °C → r.t.; (g) 7.5 M NaOH,
acrylonitrile, 0 °C → r.t.; (h) 4 N HCl in dioxane, 0 °C → r.t.; (i) allyl
bromide, NaH, DMF, 0 °C → r.t.; ( j) O3, CH2Cl2: MeOH, −78 °C. Thiourea,
EtOH, NaBH4, 0 °C → r.t.; (k) diphenyl phosphorazide, 1,8-diazabicyclo-
[5.4.0]undec-7-ene (DBU), DMF, 0 °C → 50 °C (5 hours) → r.t. For full
experimental details, see the ESI.†
Scheme 3 Synthesis of analog T1 from the succinic acid hydroxamate
derivative 21 and the tripodal template 7. Reaction conditions: (a)
1-hydroxy-7-azabenzotriazole (HOAt), N,N’-diisopropylcarbodiimide
(DIC), N,N-diisopropylethylamine (DIPEA), CH2Cl2/DMF, r.t.; (b) phenyl-
silane, Pd(PPh3)4, CH2Cl2, r.t.; (c) 10% Pd/C, H2, EtOH, r.t. For full experi-
mental details, see the ESI.†
amine group, generating an anchoring point for future intro-
duction of functional conjugates.
Based on the resolved co-crystal structure of FC within the
FhuA transporter,21,22 the iron complex or the FC “head” is
embedded in the binding site, while the FC “tail” remains
solvent accessible. In similar tripodal templates, the substi-
tuted apical site did not interfere with iron-binding or trans-
porter recognition, therefore it was used for coupling
fluorescent chromophores for ‘signaling’,40,47 or to enable the
incorporation of a surface-adhesive moiety.48
with NaBH4 provided triol 13, which was converted to azide 14
using diphenyl phosphorazide, as suggested by Thompson
and co-workers.54 The desired triamine 15 was obtained by
hydrogenation of the azide group.
For the synthesis of analogs T1 and T4, the convergent syn-
thetic approach was followed, that is, the template and the
chelating arms were prepared separately and only then
assembled to generate the desired analog.
The chelating arms were prepared as shown in Scheme 2.
The synthesis of O-benzyl-N-methyl-hydroxylamine 19 was
accomplished by bocylation of commercially available O-benzyl-
hydroxylamine 16 to ensure monomethylation in the next
step.55 Following methylation of 17 by iodomethane,56 the Boc
group was removed by acidolysis.57 Acetylation of compound
19 with succinic anhydride 20 provided the succinic acid
hydroxamate derivative 21. Compound 21 was coupled
through its free carboxyl to the tripodal-triamine template 7 or
15, providing the protected analogs 22 (Scheme 3) and 25
(Scheme 4), respectively. Removal of the orthogonal protecting
groups, alloc and benzyl (Bzl), provided the target compound T1,
while deprotection of benzyl from 25 provided compound T4.
Compounds T2 and T3 were prepared in accordance with
the principles of linear synthesis, that is, by integrating Asp
Templates 7 and 11 (Scheme 1a and b respectively) were
prepared by Michael addition of tris(hydroxyme-thyl)amino-
methane
1
or 1,1,1-tris(hydroxymethyl)ethane
8
to
acrylonitrile,49,50 followed by the reduction of the trinitrile
intermediates using NaBH4 and CoCl2 in the presence of Boc
anhydride (in situ protection).51 Acidolysis of Boc provided the
desired triamine templates 7 and 11. The presence of the free
functional amine in compound 1 meant that additional syn-
thetic steps were required. The apical amine of compound 2
was protected with carboxybenzyl (Cbz), which is stable under
nitrile reductive conditions. After nitrile reduction, Cbz was
exchanged with allyloxycarbonyl (alloc),52 which enabled
orthogonal deprotection in later steps of the synthesis.
Template 15, which has an ethylene spacer (Scheme 1c),
was prepared by allylation of triol 8 using allyl bromide.53
Ozonolysis53 of 12 and further reduction of the intermediate
This journal is © The Royal Society of Chemistry 2015
Dalton Trans.

View Article Online
Paper
Dalton Transactions
system does not depend on the complex stability constant.
However, to function as an efficient iron transporter, a FC
analog must be able to form a stable ferric trihydroxamate
complex under physiologically relevant pH conditions. There-
fore, the pH-dependent distribution of the forms of the FC
complexes was obtained, and then the pFe value of each
analog, which reflects the efficacy with which it binds ferric
iron under the designated conditions, was determined.
In order to determine the stability constants of ferric com-
plexes and to obtain all their coordination characteristics, the
acid–base properties of T1–T4 free ligands were first deter-
mined (see the ESI†). Next, electrospray ionization mass spec-
trometry (ESI-MS) was used to characterize the stoichiometry
of the metal complexes in solution. Spectra of the Fe(ClO4)3/
ligand in a 1 : 1 reaction mixture were characterized by the
presence of a few major peaks attributable to the mononuclear
species (Table S2†).
To determine the stability constants of the T1–T4 ferric
complexes, classical spectrophotometric competition experi-
ments with ethylenediaminetetraacetic acid (EDTA) were per-
formed at pH 8 (Fig. S1†). The overall stability constants,
log KFeL, were calculated using eqn (S5) and (S6), the protona-
tion constants of the respective ligands (Table S1†) and of
EDTA,58 and the stability constant of Fe-EDTA.58 The values
obtained for T1 and T4 are given in Table 1. However, the T2
and T3 ferric complexes exhibited some instability before the
equilibrium with EDTA could be achieved and the calculated
constants were unreliable (data not shown). A different
method, which involved a combination of pH-dependent spec-
trophotometric and potentiometric titrations, was therefore
applied to all the studied compounds. As complex formation
for all the studied ligands started at pH of around 0 and the
complexation process was complete at pH < 2, the stability
constants were determined via two series of experiments
carried out in a pH range of (i) 0–2 and (ii) 2–8 (Fig. S2†). The
log KFeL values determined from both the EDTA competition
and the pH-dependent spectrophotometric/potentiometric
titrations are given in Table 1.
Scheme 4 Synthesis of analog T4 from the succinic acid hydroxamate
derivative 21 and the tripodal template 15. Reaction conditions: (a)
HOAt, DIC, DIPEA, CH2Cl2/DMF, r.t.; (b) 10% Pd/C, H2, EtOH, r.t. For full
experimental details, see the ESI.†
into the triamine template and then incorporating O-benzyl-N-
methylhydroxylamine (Scheme 5). This synthetic pathway
diminished the tendency of Asp to form the undesired asparti-
mide by-product.
The two carboxyl groups of Asp could be attached either
α or β to the template, generating two analogs differing in the
location of the amine side group and chiral center relative to
the position of the iron center. Moreover, by using enantiopure
L/D-Asp precursors, chiral derivatives were prepared in order to
examine their effect on the enantioselective bacterial uptake
system.
In this synthetic pathway, the tripodal template 11 was
coupled to orthogonally protected Boc-Asp-OBzl 27a or Boc-
Asp-(OBzl)-OH 27b. Following benzyl deprotection, the inter-
mediates 29a and 29b were coupled to O-benzyl-N-methyl-
hydroxylamine 19. Acidolysis of Boc and hydrogenolysis of
benzyl provided the target compounds T2 and T3 as pure L or
D isomers.
All compounds were purified by chromatography and
characterized using spectroscopic techniques, as fully
described in the ESI.†
Physicochemical properties
Analysis of the pH-dependent distribution of the FC analog
complexes provided an understanding of the effect that modi-
fying the structure of the analogs had on their coordination
Binding properties and overall complex stability. Reco-
gnition of the ferrisiderophore complex by the bacterial uptake
Scheme 5 Synthesis of analogs T2 and T3. Reaction conditions: (a) 1,1’-carbonyldiimidazole (CDI), DIPEA, THF, 0 °C → r.t.; (b) 10% Pd/C, H2, MeOH,
r.t (c) HOAt, DIC, DIPEA, CH2Cl2/DMF, r.t.; (d) 50% TFA in CH2Cl2, triisopropylsilane, 0 °C → r.t. For full experimental details, see the ESI.†
Dalton Trans.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Dalton Transactions
Paper
the corresponding ferric complexes, which is in agreement
with the behavior of hydroxamate siderophores.
Table 1 Overall stability constants (log KFeL), E1/2 and pFe values for
iron(III) complexesa
pFe of ferric complexes. To compare the efficacy of iron che-
lation by the studied ligands with that of other trihydroxa-
mate-based natural or synthetic siderophores, the free
hexaaquairon(III) concentrations were calculated (pFe = −log
[Fe(III)(aq)] at a pH of 7.4 with a total ligand concentration of
10−5 M and a total Fe(III) concentration of 10−6 M).61 The
results (Table 1) clearly demonstrated that shortening the tem-
plate backbone by one methylene group did not affect the
binding properties of the molecule (pFe of 23.5 vs. 23.6 for T1
and T4, respectively), whereas the introduction of amino
groups as external substituents on the basic backbone slightly
decreased the Fe(III) binding efficacy (pFe of 22.4 vs. 22.5 for
T2 and T3, respectively). Therefore, even if the amino groups
were deprotonated and uncharged at pH 7.4, they still induced
steric hindrance, lowering the pFe by around one order of mag-
nitude. Generally, the pFe values for the non-substituted
analogs were less than two orders of magnitude lower than the
pFe values of natural FC, while those for the amino-substituted
analogs were almost three orders of magnitude lower. Never-
theless, the pFe values for all the studied tripodal FC analogs
remained within the range for effective iron chelators
(Table S5†).
CD (circular dichroism) of ferric complexes. The chiral pre-
ferences of the T2 and T3 ferric complexes in solution were
determined by CD. L- and D-T2 complexes displayed spectra
(Fig. 1) corresponding to tris-hydroxamate ferric complexes,
with a typical ligand-to-metal charge transfer (LMCT) band in
the region of 420 nm.62,63 As expected, L- and D-diastereomers
displayed opposite spectral configurations: a right handed-Δ
configuration in the case of L-Asp, and a left handed-Λ con-
figuration in the case of D-Asp acid derivatives. Both the
D-analog and the native FC63 preferably formed left handed-Λ
iron complexes.
log KFeL
Compound
Fe(III)
Fe(II)
E1/2 [mV per NHE]
pFe(III)
FC
T1
29.07b
8.5
8.9
−446d
−360(3)
25.2b
23.5
27.96(8)c
27.45(9)
27.92(6)
26.91(6)
27.80(9)c
27.34(1)
T2
T3
T4
10.0
7.9
8.3
−289(4)
−356(4)
−390(3)
22.4
22.5
23.6
a Conditions: T = 25 °C, I = 0.10 M NaClO4. b Anderegg et al.59
c Constants determined from EDTA competition titrations. Conditions:
clig = cFe(III) = 1.5 × 10−4 M, cEDTA = 0–1.5 × 10−3 M, I = 0.10 M NaClO4.
d Cooper et al.60
properties (Fig. S3–S5†). In general, for compounds T1 and T4,
two forms, namely dihydroxamate and trihydroxamate, pre-
dominated in solution, with the equilibrium between them
represented as pKNHOH (Table S3†).
For the T4 ferric system, pKNHOH was 2.48 and was only
slightly higher than the corresponding constant for the T1
ferric system, 2.32. These two compounds differ in the length
of the template spacer and the nature of the apical site (–CH3
vs. –NH2). The apical amino group of analog T1 is too far to
influence the di-/trihydroxamate equilibrium, whereas short-
ening the spacer in the template only slightly affected the
binding properties of FC analogs.
In the case of analogs T2 and T3, the presence of external
amino substituents in the vicinity of the hydroxamate moieties
hampered the formation of the trihydroxamate ferric com-
plexes; the pKNHOH values of 3.40 for T2 and 2.79 for T3 were
higher than the corresponding value for analog T1 (Table S3†).
Moreover, the presence of charged –NH3+ groups disturbed the
geometry of the di- and trihydroxamate complexes, as was
reflected by much lower ε values (Table S4†), and a trihydrox-
amate complex with the proper octahedral geometry was
formed only after their deprotonation (Table S3†). This
phenomenon was much more pronounced in analog T2, and
resulted from the charge and electronic effects of the proto-
nated amino groups located in close proximity to the hydrox-
amate functional groups.
Although both T2 and T3 were expected to exhibit a chiral
preference, L- and D-T3 presented a negligible CD signal (data
not shown). By contrast, the presence of chiral centers near
the iron binding site (α position) in T2 induced a larger energy
difference between the Δ and Λ configurations, with a prefer-
ence for the more stable form.64 Compared with T2, the chiral
center in compound T3 was located farther from the iron
CV (cyclic voltammetry) measurements. In order to study
the redox chemistry of FC analogs, electrochemical measure-
ments were performed. For ferric complexes of T1–T4, quasi-
reversible voltammograms were obtained (Fig. S6†), with ΔE in
the range of 95–135 mV. The values of E1/2 were recalculated
vs. normal hydrogen electrode (NHE) and are presented in
Table 1. Overall, in comparison with FC, the ferric complexes
of T1–T4 ligands presented a less negative redox potential,
reflecting their lower stability (vide supra). The approximate
stability constants of ferrous Fe(II)L complexes, calculated
using eqn (S10) and the log KFe(III)L from Table 1, were around
20 orders of magnitude lower than the stability constants for
Fig. 1 CD spectra of L- and D-T2 ferric complexes. Solvent: tris/HClO4
buffer (pH = 7.8, I = 0.1 M NaClO4). Complex concentration of 0.25 mM.
This journal is © The Royal Society of Chemistry 2015
Dalton Trans.

View Article Online
Paper
Dalton Transactions
binding site (β position). Moving of the chiral center away
from the chirogenic octahedral iron center (even by a single
carbon) practically abolishes chiral discrimination, thus
leading practically to a racemic mixture of Δ and Λ configur-
ations of T3.
Examination of the biological activity of the FC analogs.
The synthetic siderophore analogs were tested on two micro-
organisms: Pseudomonas putida mutant JM218, provided by
L. C. van Loon, Utrecht, the Netherlands;40 and Escherichia coli
mutant UT5600,65 obtained from the Coli Genetic Stock
center. These mutants possess transporters for FC but do not
produce their own siderophores; therefore, no exchange can
occur between heterologous and native siderophores. The
growth of these strains is proportional to the bio-availability of
ferric complexes.
The growth-promoting effects of the synthetic analogs were
compared with each other (Table 2) and with natural FC (posi-
tive control) and double-distilled water (DDW; negative
control). The siderophore iron-complexes were added to the
growth medium as the sole iron source. To remove iron traces,
modified King’s B (MKB) medium was treated with the amido-
xime-chelating resin Purolite® S910. Bacteria were inoculated
in iron-depleted MKB liquid medium and incubated at 30 °C
until the bacterial stationary phase. Their growth was followed
by measuring their optical density at 600 nm at fixed time
intervals. Each enantiomer was tested separately (for full
experimental procedures, see the ESI†).
A comparison of the uptake by P. putida and E. coli (Fig. S7
and S8†) indicated that both bacterial strains were able to
consume artificial siderophores in addition to FC (Table 2).
Compound T1 was only recognized by P. putida, however, a
slightly modified template, substituted with an ethylene
spacer instead of the propylene spacer and bearing methyl
instead of amine on the apical site, produced a new analog T4,
with excellent growth enhancement properties in both
P. putida and E. coli. Further improvement was obtained by
Chart 2 Representation of the systematic modification performed on
the backbones of the FC analogs. Modifications to the analog backbone
are marked in red. Compounds able to promote the growth of both
E. coli and P. putida are shown in blue.
reversing the direction of the amide linkage to produce an
analog (β-Ala) that promoted growth to the same extent as FC
in both P. putida and E. coli (Table 2, and Chart 2). β-Ala, like
analog T1, possesses an apical amino group, however the
different recognition of T1 and β-Ala indicates that this group
is not a key structural feature, and that the amine “tail” most
likely does not interact with the recognition domain of the
FhuA transporter. Analogs T4 and β-Ala were able to match the
FhuA transporter in that they could effectively promote the
growth of dissimilar microorganisms. Such broad-spectrum
activity, similar to that of FC, is rare among synthetic analogs.
Extending the arm with a propylene spacer, as in the GABA
analog, or shortening the arm to a single methylene spacer, as
in B9, had no effect on P. putida propagation, while completely
abolishing E. coli growth (Table 2, and Chart 2).
Optimal recognition by E. coli was achieved for analogs
with backbones having ethylene spacers in both the template
and the arms, indicating the need to retain well-defined dis-
tances in both the arms and the template. The fact that some
analogs failed to transport iron to E. coli, while being fully
active in P. putida, suggests that the FC uptake systems of
these two bacterial strains are not identical.
Our attempt to enhance the H-bonding network in the reco-
gnition-domain of FhuA by developing analogs carrying amine
side groups (T2 and T3) produced analogs that facilitated
growth in P. putida but not in E. coli. It seems that the reco-
gnition domain in P. putida is more tolerant (possibly less
crowded) than that of E. coli, as it was shown to tolerate ali-
phatic side groups as well as polar amine substituents.38 The
location of the amine group one position farther from the
metal binding site (i.e., in the β position with respect to the
iron center) improved P. putida recognition from high to full
activity (Table 2, T2 and T3, respectively).
Table 2 Biological activity of the artificial siderophores in P. putida
(JM218) and E. coli (UT5600)
Growth promotiona
Compound
E. colib
P. putidab
Reference
FC
T1
+++++
+++++
+++++
++++
+++++
+++++
+++++
+++++
+++++
−
Current work
Current work
Current work
Current work
−
L/D-T2
L/D-T3
T4
−
−
++++
+++++
−
Current work
β-Alac
GABAc
B9c
Besserglick et al.66
Besserglick et al.66
Jurkevitch et al.38
Current work
−
−
DDW
a Scale from full activity (+++++) to no activity (−) based on the optical
density at 600 nm in comparison with that of positive (FC) and
negative (DDW) controls. b The results represent the stationary phase
of the appropriate bacterial strain. c See Chart 2 for the structures of
GABA, β-Ala, and B9.
The observation that there was no difference between the
growth promoting effects of the enantiomer pairs of T2 and T3
(Λ and Δ isomers) vis-à-vis P. putida is quite puzzling, given
Dalton Trans.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Dalton Transactions
Paper
that this result is in direct contrast to the enantioselectivity of sitions, with analogs T2 and T3 substituted by amine side-
FC67,68 and of similar artificial chiral siderophores.64 The lack groups. In addition, we present a synthetic methodology for
of stereospecificity observed in analogs T2 and T3 may reflect introducing functional-conjugates at the template’s apical site
the lability of iron-complexes with respect to isomerization in to allow future therapeutic applications, as represented by
aqueous solutions.24 Ligand exchange can shift complexes analog T1.
possessing a Δ-configuration to the FhuA-preferred Λ-configur-
The iron-complex formation characteristics and stability of
ation, thus allowing utilization of both isomers. An alternative the analogs were investigated in detail. All the investigated FC
possibility is that several distinct siderophore transporters par- analogs bind ferric ions in a 1 : 1 stoichiometry, as demon-
ticipate in the transport of different isomers (e.g. FhuA in the strated by ESI-MS and further confirmed by potentiometric
transport of the Λ-isomer and FhuE69 in the transport of the and spectrophotometric monitoring. Chiral preferences were
Δ-isomer). It is, however, unlikely that different transporters investigated by CD. Chiral centers located near the metal
will generate growth promotion curves with practically identi- binding site display LMCT bands with Δ and Λ configurations
cal kinetic parameters (Fig. S8†).
typical for trishydroxamate-ferric complexes. However, distan-
The extent to which transport is governed exclusively by a cing the chiral center from the metal binding site practically
single transporter has been addressed in previous studies and abolishes chiral discrimination. The affinities of the iron-side-
in this work. We have previously shown, by radioactive com- rophores, as reflected by pFe, show that the ferric complexes of
petition uptake assays, that the uptake of tripodal analogs in all the presented analogs are about 1–2 orders of magnitude
P. putida is primarily carried out through the FC trans- less stable than those of FC. Although the amine substituents
porter.38,64 In a similar experiment, Schalk suggested that are located in close proximity to the metal binding site, the
P. aeruginosa takes-up the tripodal artificial analog and native amines are not involved in iron coordination. The external
FC with equal efficiency.70
amine substituents slightly destabilize the ferric complex,
The uptake and release pathway in E. coli UT5600 was most likely due to electrostatic repulsion between the neigh-
studied using fluorescent tripodal β-Ala labeled with naphtha- boring arms, thus preventing the formation of the stabilizing
limide (see the ESI†). The bound iron(III) quenches the fluoro- H-bonds observed in native FC.63 Nevertheless, the stability of
phore emission, once the iron is taken-up in the cytosol the all the complexes lies within the range of natural sidero-
green fluorescence is restored and the distribution of the iron- phores,7,24,72 and therefore these FC analogs can be utilized as
free siderophore recorded. In the case of E. coli K-12, which potent siderophores.
lacks an external transporter (the ΔfhuA mutant obtained
The biological potency of the new analogs was monitored
from the Coli Genetic Stock center),71 no fluorescence was by growth promotion experiments. The analogs exhibit biologi-
observed, indicating that uptake is exclusively carried out by cal activities that can be divided into two sub-groups: one
the FhuA receptor (Fig. S9†).
(T1–T3) that shows high specificity toward P. putida, but are
not recognized by E. coli; and another (T4) that is recognized
by both P. putida and E. coli. The existence of the first group
indicates that the FC recognition domains of these two bac-
teria differ considerably, whereas the second group shows that
Summary and conclusions
FC is an attractive model for studies since a detailed X-ray the simple platform of ethylene-amide-ethylene-retrohydrox-
structure of FhuA21,22 is available, so allowing researchers to amate is capable of matching different transporters in dissimi-
focus attention on the parameters important for transporter lar microorganisms, as does the native FC.
recognition. The X-ray structure shows two sets of H-bonding
The less-tolerant structure of the E. coli transporter com-
interactions between FC and the FhuA transporter. One set pared with the accommodating structure of the P. putida trans-
involves the hydroxamate oxygens and the second set involves porter allows us to establish the minimal essential structural
carbonyls on the peptide backbone. We assumed that the parameters for broad-spectrum recognition; an elusive target
H-bonding interactions with hydroxamate or retrohydroxamate that we and others have been pursuing for years. The govern-
are practically identical. Thus, the only accessible means to ing rules for designing analogs with multiple-uptake capabili-
match the remaining H-bonding and control the analog’s ties are: (i) to mimic the transporter with the most stringent
overall size was via the spacers. Spacers inserted into the tem- requirements, such as the FhuA transporter in E. coli; (ii) to
plate backbone and/or the arms can optimize the directionality develop a library of analogs in which each building group can
and the relative orientation of the functional groups involved be systematically and gradually modified until optimization is
in H-bonding. Alternatively, additional H-donor groups, such obtained, as was demonstrated through step-by-step structural
as external amines, can be introduced to try to enhance the adjustments of the modular system described in this work.
H-bonding network.
It should be noted that minor modification may abolish
Here, we present a new series of FC analogs. The series uti- broad-spectrum activity, so inducing the formation of narrow-
lizes two C3 symmetric tris-amino derived templates extended spectrum analogs utilizing practically the same platform.
by succinic (T1 and T4) or Asp (T2 and T3) acids, with all com-
Although we have not identified an artificial analog target-
pounds terminated by retrohydroxamates. The resulting FC ing exclusively E. coli, understanding the unique requirements
analogs differ with respect to their skeletal lengths and compo- of FhuA may facilitate its preparation. The interactions
This journal is © The Royal Society of Chemistry 2015
Dalton Trans.

View Article Online
Paper
Dalton Transactions
between FC and its transporter may serve as a prototype for 19 R. Chakraborty, E. Storey and D. van der Helm, Biometals,
other hydroxamate siderophores for which the detailed X-ray 2007, 20, 263.
structure is lacking. Assuming similar H-bonding interactions 20 D. van der Helm and R. Chakraborty, in Microbial transport
to those described for FC in its transporter, we suggest follow-
ing similar optimization processes.
systems, ed. G. Winkelmann, Wiley-VCH, Weinheim, 2002,
ch. 11, p. 261.
In future work, we will try to extend the guidelines for 21 A. D. Ferguson, E. Hofmann, J. W. Coulton, K. Diederichs
optimization and multiple-recognition to other bacteria and and W. Welte, Science, 1998, 282, 2215.
other siderophores with an emphasis on mammalian patho- 22 K. P. Locher, B. Rees, R. Koebnik, A. Mitschler,
genic targets. We also plan to initiate a computer ‘docking’
analysis aimed at understanding the origin of recognition in
L. Moulinier, J. P. Rosenbusch and D. Moras, Cell, 1998,
95, 771.
E. coli and at resolving the lack of stereo-specificity observed in 23 R. Breslow, Chem. Soc. Rev., 1972, 1, 553.
both enantiomer pairs derived from Asp in P. putida.
24 K. N. Raymond, G. Muller and B. F. Matzanke, Top. Curr.
Chem., 1984, 123, 49.
25 A. Shanzer, C. E. Felder and Y. Barda, in The Chemistry of
Hydroxylamines, Oximes and Hydroxamic Acids, ed.
Z. Rappoport and J. F. Liebman, John Wiley & Sons, Ltd,
2009, ch. 16, p. 751.
Acknowledgements
E. G.-K. and A. Sz. are grateful to the Polish National Science
Center (NCN, UMO-2011/03/B/ST5/01057) and Wroclaw Centre 26 M. J. Miller, Chem. Rev., 1989, 89, 1563.
of Biotechnology (The Leading National Research Centre 27 S. Heggemann, U. Mollmann, P. Gebhardt and L. Heinisch,
Program, KNOW, 2014–2018) for financial support.
Biometals, 2003, 16, 539.
28 M. Gaspar, M. A. Santos, K. Krauter and G. Winkelmann,
Biometals, 1999, 12, 209.
29 S. K. Sharma, M. J. Miller and S. M. Payne, J. Med. Chem.,
1989, 32, 357.
References
1 J. B. Neilands, Annu. Rev. Nutr., 1981, 1, 27.
2 C. Wandersman and P. Delepelaire, Annu. Rev. Microbiol.,
2004, 58, 611.
3 M. L. Guerinot, Annu. Rev. Microbiol., 1994, 48, 743.
4 J. B. Neilands, J. Biol. Chem., 1995, 270, 26723.
5 G. Winkelmann, Biochem. Soc. Trans., 2002, 30, 691.
30 S. Dhungana, J. M. Harrington, P. Geblhardt, U. Mollmann
and A. L. Crumbliss, Inorg. Chem., 2007, 46, 8362.
31 B. H. Lee, M. J. Miller, C. A. Prody and J. B. Neilands,
J. Med. Chem., 1985, 28, 317.
32 B. H. Lee, M. J. Miller, C. A. Prody and J. B. Neilands,
J. Med. Chem., 1985, 28, 323.
6 C. Ratledge and L. G. Dover, Annu. Rev. Microbiol., 2000, 54, 33 K. Matsumoto, T. Ozawa, K. Jitsukawa and H. Masuda,
881. Inorg. Chem., 2004, 43, 8538.
7 R. C. Hider and X. Kong, Nat. Prod. Rep., 2010, 27, 34 T. Emery, L. Emery and R. K. Olsen, Biochem. Biophys. Res.
637. Commun., 1984, 119, 1191.
8 I. J. Schalk, G. L. A. Mislin and K. Brillet, in Current Topics 35 R. K. Olsen and K. Ramasamy, J. Org. Chem., 1985, 50,
in Membranes, ed. J. M. Arguello and S. Lutsenko, Elsevier,
2012, vol. 69, ch. 2, p. 37.
9 I. J. Schalk and L. Guillon, Amino Acids, 2013, 44,
1267.
10 K. D. Krewulak and H. J. Vogel, Biochim. Biophys. Acta,
2008, 1778, 1781.
11 M. Miethke and M. A. Marahiel, Microbiol. Mol. Biol. Rev.,
2007, 71, 413.
2264.
36 A. Shanzer, J. Libman, R. Lazar and Y. Tor, Pure Appl.
Chem., 1989, 61, 1529.
37 A. Shanzer, J. Libman, R. Lazar, Y. Tor and T. Emery,
Biochem. Biophys. Res. Commun., 1988, 157, 389.
38 E. Jurkevitch, Y. Hadar, Y. Chen, J. Libman and A. Shanzer,
J. Bacteriol., 1992, 174, 78.
39 O. Ardon, R. Nudelman, C. Caris, J. Libman, A. Shanzer,
Y. Chen and Y. Hadar, J. Bacteriol., 1998, 180, 2021.
12 V. Braun and K. Hantke, in Microbial transport systems, ed.
G. Winkelmann, Wiley-VCH, Weinheim, 2002, ch. 12, 40 H. Weizman, O. Ardon, B. Mester, J. Libman, O. Dwir,
p. 289.
Y. Hadar, Y. Chen and A. Shanzer, J. Am. Chem. Soc., 1996,
118, 12368.
41 T. Zheng and E. M. Nolan, Metallomics, 2012, 4, 866.
42 M. J. Miller, J. A. Mckee, A. A. Minnick and E. K. Dolence,
Biol. Met., 1991, 4, 62.
13 V. Braun and H. Killmann, Trends Biochem. Sci., 1999, 24,
104.
14 J. B. Neilands, J. Am. Chem. Soc., 1952, 74, 4846.
15 S. J. Rogers, R. A. Warren and J. B. Neilands, Nature, 1963,
200, 167.
43 U. Mollmann, L. Heinisch, A. Bauernfeind, T. Kohler and
D. Ankel-Fuchs, Biometals, 2009, 22, 615.
16 J. B. Neilands, Science, 1967, 156, 1443.
17 D. E. Crowley, Y. C. Wang, C. P. P. Reid and P. J. Szaniszlo, 44 C. Ji, R. E. Juarez-Hernandez and M. J. Miller, Future Med.
Plant Soil, 1991, 130, 179. Chem., 2012, 4, 297.
18 E. Barness, Y. Hadar, Y. Chen, A. Shanzer and J. Libman, 45 J. M. Roosenberg, Y. M. Lin, Y. Lu and M. J. Miller, Curr.
Plant Physiol., 1992, 99, 1329. Med. Chem., 2000, 7, 159.
Dalton Trans.
This journal is © The Royal Society of Chemistry 2015

View Article Online
Dalton Transactions
Paper
46 S. J. Milner, A. Seve, A. M. Snelling, G. H. Thomas, 59 G. Anderegg, F. L’Eplattenier and G. Schwarzenbach, Helv.
K. G. Kerr, A. Routledge and A. K. Duhme-Klair, Org.
Biomol. Chem., 2013, 11, 3461.
47 R. Nudelman, O. Ardon, Y. Hadar, Y. N. Chen, J. Libman
and A. Shanzer, J. Med. Chem., 1998, 41, 1671.
48 Y. Gafni, H. Weizman, J. Libman, A. Shanzer and
I. Rubinstein, Chem. – Eur. J., 1996, 2, 759.
Chim. Acta, 1963, 46, 1409.
60 S. R. Cooper, J. V. Mcardle and K. N. Raymond, Proc. Natl.
Acad. Sci. U. S. A., 1978, 75, 3551.
61 W. R. Harris, C. J. Carrano, S. R. Cooper, S. R. Sofen,
A. E. Avdeef, J. V. Mcardle and K. N. Raymond, J. Am. Chem.
Soc., 1979, 101, 6097.
49 G. R. Newkome, X. F. Lin and J. K. Young, Synlett, 62 K. N. Raymond and C. J. Carrano, Acc. Chem. Res., 1979, 12,
1992, 53. 183.
50 R. Kikkeri, H. Traboulsi, N. Humbert, E. Gumienna-Kon- 63 D. Van der Helm, J. R. Baker, D. L. Engwilmot,
tecka, R. Arad-Yellin, G. Melman, M. Elhabiri,
A. M. Albrecht-Gary and A. Shanzer, Inorg. Chem., 2007, 46,
2485.
M. B. Hossain and R. A. Loghry, J. Am. Chem. Soc., 1980,
102, 4224.
64 A. Shanzer, J. Libman, P. Yakirevitch, Y. Hadar, Y. Chen
and E. Jurkevitch, Chirality, 1993, 5, 359.
65 M. E. Elish, J. R. Pierce and C. F. Earhart, J. Gen. Microbiol.,
1988, 134, 1355.
66 J. Besserglick, Thesis for the degree Doctor of Philosophy,
Weizmann Institute of Science, 2013.
67 G. Winkelmann, FEBS Lett., 1979, 97, 43.
51 S. Caddick, A. K. D. Haynes, D. B. Judd and
M. R. V. Williams, Tetrahedron Lett., 2000, 41, 3513.
52 C. M. Stevens and R. Watanabe, J. Am. Chem. Soc., 1950, 72,
725.
53 S. Hanessian, H. Prabhanjan, D. X. Qiu and S. Nambiar,
Can. J. Chem., 1996, 74, 2540.
54 A. S. Thompson, G. R. Humphrey, A. M. Demarco, 68 G. Winkelmann and V. Braun, FEMS Microbiol. Lett., 1981,
D. J. Mathre and E. J. J. Grabowski, J. Org. Chem., 1993, 58,
5886.
11, 237.
69 M. Sauer, K. Hantke and V. Braun, J. Bacteriol., 1987, 169,
55 B. H. Lee and M. J. Miller, J. Org. Chem., 1983, 48, 24.
2044.
56 R. Grigg, Z. Rankovic and M. Thoroughgood, Tetrahedron, 70 M. Hannauer, Y. Barda, G. L. A. Mislin, A. Shanzer and
2000, 56, 8025. I. J. Schalk, J. Bacteriol., 2010, 192, 1212.
57 G. Han, M. Tamaki and V. Hruby, J. Pept. Res., 2001, 58, 71 T. Baba, T. Ara, M. Hasegawa, Y. Takai, Y. Okumura,
338.
M. Baba, K. A. Datsenko, M. Tomita, B. L. Wanner and
H. Mori, Mol. Syst. Biol., 2006, 2.
72 H. Boukhalfa and A. L. Crumbliss, Biometals, 2002, 15, 325.
58 L. D. Pettit and H. K. Powell, Timble, York, UK, 2001,
(http://www.acadsoft.co.uk, version 5.6).
This journal is © The Royal Society of Chemistry 2015
Dalton Trans.