Direct identification of a siderophore import protein using synthetic petrobactin ligands

Article abstract of DOI:10.1002/anie.201005527

Nice catch! The immobilization of a synthetic petrobactin derivative on agarose was crucial for the isolation and identification of a bacterial siderophore import protein for the first time from crude cell extracts. The biochemical and genetic characteriz

Full text of DOI:10.1002/anie.201005527

Communications
DOI: 10.1002/anie.201005527
Siderophore Import
Direct Identification of a Siderophore Import Protein Using Synthetic
Petrobactin Ligands**
Nikolas Bugdahn, Florian Peuckert, Alexander G. Albrecht, Marcus Miethke,
Mohamed A. Marahiel,* and Markus Oberthꢀr*
The increase of bacterial resistance against almost all
clinically used antibiotics is one of the most pressing public
health problems. While existing drugs are becoming less and
less effective, only a few truly new antibiotics have found their
way to clinical application in the last decades. The inhibition
of novel biochemical pathways that are not addressed by
currently used antibiotics offers a potentially successful
strategy for the development of new drugs able to combat
infections caused by resistant bacteria. In this context, the
interference with the bacterial uptake of iron promoted by
siderophores has become the focus of attention.[1] Side-
rophores are polar low-molecular-weight molecules with
exceptionally high iron-binding affinities that are secreted
and reimported by microorganisms through dedicated trans-
port systems. In addition to blocking their biosynthesis, the
inhibition of siderophore export and import proteins offers a
promising approach for the development of new antibiotics,
because siderophore-promoted iron uptake is essential for
both the survival and virulence of pathogens.[2] For many
bacteria, however, there is only limited information available
regarding the export and import systems involved. In
addition, siderophore-binding proteins have been discovered
so far only indirectly based on homology searches or growth
phenotype analysis of mutants.
Herein, we report the first capture of a siderophore-
binding protein from cell extracts through a direct interaction
with its natural ligand.[3] To this end, we employed affinity
chromatography using an immobilized siderophore and sub-
sequent identification of the retained protein by mass
spectrometric analysis (Figure 1A). After overexpression in
Escherichia coli and purification, biochemical characteriza-
tion of the recombinant protein was possible. This novel
Figure 1. A) Identification of siderophore-binding proteins by affinity
chromatography using immobilized siderophores. The retained pro-
teins are identified by MS analysis of fragments obtained by tryptic
digest, and are then characterized biochemically. B) Chemical structure
of petrobactin (1) and biotinyl petrobactin (2). The groups that are
involved in iron binding are shown in red.
[*] N. Bugdahn,[+] Dr. M. Oberthꢀr
Fachbereich Chemie, Philipps-Universitꢁt Marburg
Hans-Meerwein-Strasse, 35032 Marburg (Deutschland)
Fax: (+49)6421-282-2021
E-mail: oberthuer@chemie.uni-marburg.de
F. Peuckert,[+] A. G. Albrecht, Dr. M. Miethke, Prof. Dr. M. A. Marahiel
Fachbereich Chemie, Philipps-Universitꢁt Marburg
Hans-Meerwein-Strasse, 35032 Marburg (Deutschland)
Fax: (+49)6421-282-2191
approach allows the selection of specific binding proteins out
of a pool of siderophore importers present in cell extracts.
For the capture of siderophore-binding proteins, we
selected the siderophore petrobactin (1), which most notably
is produced by the two pathogens Bacillus cereus and Bacillus
anthracis.[4] For B. cereus, two petrobactin transporters have
been reported recently based on sequence homologies,[5]
whereas genetic studies identified similar petrobactin
import systems in B. anthracis.[6] For the latter, petrobactin
(1; Figure 1B) is a virulence factor, because the second
siderophore produced by B. anthracis, bacillibactin, is inter-
cepted during an infection by a protein of the innate immune
system, siderocalin.[7] Petrobactin (1) is therefore essential for
E-mail: marahiel@staff.uni-marburg.de
[+] These authors contributed equally to this work.
[**] Financial support from the German Science Foundation (DFG) is
gratefully acknowledged. We thank Dr. Uwe Linne and Natalia
Fritzler for mass spectrometric analysis and Tanja Ellenberger and
Michael Kock (research students) for excellent technical assistance.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201005527.
10210
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2010, 49, 10210 –10213

Angewandte
Chemie
iron acquisition by B. anthracis inside a human host, which
makes the inhibition of its uptake a potential strategy for
treatment of anthrax infections.
Petrobactin (1) contains two 3,4-dihydroxybenzoate units
attached to a citric acid backbone through spermidine linkers;
the two catechol units and the central carboxy and hydroxy
groups of the citric acid moiety (red in Figure 1B) act as
ligands in the formation of an octahedral iron complex.
Because the amino groups of the spermidine units are not
directly involved in iron binding, we chose them for the
attachment of a biotin tag, such that the siderophore can be
immobilized on agarose beads derivatized with streptavidin
for affinity chromatography. Accordingly, we synthesized the
biotinylated petrobactin 2 (Figure 1B), in which the tag is
attached to the central amino group of one of the spermidine
side chains by means of a g-aminobutyric acid spacer. In
addition, we also generated sufficient amounts of petrobactin
(1) for the biochemical characterization of the captured
transport protein.
The synthetic route to the required petrobactin spermi-
dine side chain 8 and the modified side chain 10, which
contains the Dde-protected g-aminobutyric acid linker,
started with the commercially available diethyl acetal of 3-
aminopropanal, 3, and the known benzoic acid 4[8]
(Scheme 1). Compared to previously reported syntheses,[8,9]
our optimized route to petrobactin side chains is short,
operationally simple, and high yielding. In addition, the
protection of the central amino group led to improved yields
in subsequent coupling reactions (Scheme 2). For the syn-
Scheme 2. Synthesis of 1 and 2. Reagents and conditions: a) 1. NHS,
DCC, THF; 2. 8, Et3N, 1,4-dioxane; 87%; b) AcOH, conc. HCl; c) H2
(45 bar), Pd–C, EtOH/H2O, then reversed-phase HPLC purification,
30% (1, 2 steps); 26% (2, 2 steps); d) 10, Et3N, DMF, 84%;
e) 1. NHS, DIC, THF; 2. 8, Et3N, 1,4-dioxane; 82%; f) N2H4, EtOH;
g) biotin NHS ester, iPr2EtN, DMF, 84% (2 steps). DCC=dicyclohex-
ylcarbodiimide, DIC=diisopropylcarbodiimide.
thesis of petrobactin (1), tert-butyl citrate 11 was activated as
the N-hydroxysuccinimide (NHS) ester as described by
Phanstiel et al.[8] and subsequently coupled with side-chain
amine 8 (87%). Removal of the tert-butyl group was
accompanied by the formation of small amounts of the
known imide side product[8,10] (see the Supporting Informa-
tion). Hydrogenation and purification of the residue by
reversed-phase HPLC then afforded 1 in 30% yield.
The synthesis of the biotinylated petrobactin 2 com-
menced with the reaction of tert-butyl-protected citric acid
anhydride 13 with the linker containing side chain 10, thus
affording monoamide 14 in excellent yield. NHS activation
and reaction with petrobactin side chain 8 then led to the fully
protected petrobactin derivative 15 (82%). Next, the biotin
group was introduced by selective removal of the Dde
protecting group using hydrazine and subsequent biotinyla-
tion of the liberated amine (84%). Finally, removal of the
Scheme 1. Synthesis of spermidine side chains 8 and 10. Reagents and
conditions: a) 4, EDC, HOBt, Et3N, DMF, 92%; b) PPTS, acetone/H2O,
408C, 96%; c) 6, NaBH(OAc)3, Et3N, MeOH, 71%; d) CbzCl, Et3N,
MeOH, 94%; e) TFA, CH2Cl2, 08C, 95% (8); 97% (10); f) 9, EDC,
HOBt, Et3N, DMF, 96%. EDC=1-(3-dimethylaminopropyl)-3-ethylcar-
bodiimide hydrochloride, HOBt=1-hydroxybenzotriazole, DMF=N,N-
dimethylformamide, PPTS=pyridinium para-toluenesulfonate, Boc=
tert-butoxycarbonyl, Cbz=benzyloxycarbonyl, Dde=1-(4,4-dimethyl-
2,6-dioxocyclohexylidene)ethyl, TFA=trifluoroacetic acid.
Angew. Chem. Int. Ed. 2010, 49, 10210 –10213
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 10211

Communications
remaining protecting groups and HPLC purification afforded
biotinylated petrobactin 2.
When an aqueous solution of biotinyl petrobactin (2) was
treated with FeCl3 at pH 8, iron complexation was clearly
evident because of a strong color change to purple. The
formation of the siderophore–iron(III) complex was also
confirmed by mass spectrometry. Following incubation of
streptavidin-derivatized agarose beads with the iron-loaded
siderophore, the purple beads were transferred to a column
and equilibrated with PBS buffer. To test whether the
immobilized petrobactin derivative 2 is able to capture
relevant binding proteins from cell extracts, we chose Bacillus
subtilis as a model organism. B. subtilis does not produce
petrobactin (1) but is able to scavenge this siderophore from
the environment[11] using an import system that was not
known at the outset of our investigations. The cell lysate of a
B. subtilis culture was loaded onto the affinity column. After
washing and elution, the obtained fractions were subjected to
a tryptic digest, and the fragments were then analyzed by
mass spectrometry using the MASCOT software and the
MSDB database (see the Supporting Information).[12]
Figure 2. Final OD600 values (after 18 h) of DdhbC and DdhbCDfpiA
cultures with several additives. The gray arrow indicates the total
growth inhibition of the double mutant in the presence of apo-PB.
apo-BB=bacillibactin (iron-free), apo-PB=petrobactin (iron-free).
In our initial experiments, a number of proteins were
retained by the column. A negative control using a column
loaded with unmodified streptavidin–agarose beads, however,
showed that most of these proteins bound to the matrix itself.
After extensive experimentation, we were able to reduce
unspecific binding by pre-incubation of the siderophore-
derivatized column with bovine serum albumin. In subse-
quent experiments, YclQ was the only protein that was
retained repeatedly owing to a specific interaction with the
petrobactin-derivatized agarose beads (see the Supporting
Information). YclQ is an ABC transporter binding protein
that is part of the yclNOPQ gene cluster. It has only recently
been characterized to be the ferric petrobactin-binding
protein in B. subtilis by the group of Raymond based on
sequence homologies,[13] thus confirming the viability of our
approach. Because of its function, we propose to rename this
gene cluster fpiBCDA (fpi: ferric petrobactin import) and,
accordingly, rename YclQ FpiA.
To elucidate the role of the identified protein on the
genetic level, a B. subtilis DfpiA mutant was created, which is
also incapable of producing 2,3-dihydroxybenzoic acid and
bacillibactin as endogenous high-affinity chelators as a result
of its DdhbC background.[14] The resulting DdhbCDfpiA
double mutant and the DdhbC mutant (as control) were
grown under iron limitation, and supplementation of FeCl3,
bacillibactin, or petrobactin, respectively (Figure 2). Both
mutants showed an increased rate of growth compared to iron
limitation when iron(III) or iron-free bacillibactin was added
as a result of the uptake of iron by siderophore-independent
import systems and the bacillibactin-mediated import of iron
present in the culture, respectively.
The recombinantly produced protein FpiA was analyzed
for binding of the native ligand petrobactin (1), its biotiny-
lated analogue 2, and several other possible ligands by
fluorescence spectroscopy (see the Supporting Information).
The FpiA binding constant obtained for petrobactin (51 nm)
is in good accordance with previously published data.[13] In
comparison, the ferric complex of the biotinylated petrobac-
tin 2 is bound almost three orders of magnitude more weakly
by FpiA. Nevertheless, the capture of FpiA clearly shows that
this decreased binding affinity is still sufficient enough for the
successful retention of binding proteins.[15]
The binding stoichiometry of FpiA and the native iron-
loaded ligand 1 is rather unusual. In our fluorescence titration
experiments we determined a protein to ligand ratio of 1:4. To
corroborate the fluorescence measurements, we further
examined the protein–siderophore interaction using ligand-
dependent melting-point analysis.[16] The greatest increase in
the melting point because of the ligand-induced stabilization
of the complex was again achieved with a protein to ligand
ratio of 1:4 or higher (Figure 3). The same stoichiometry was
In contrast to the DdhbC strain, however, the DdhbC-
DfpiA double mutant did not grow in the presence of iron-
free petrobactin, which indicates that FpiA is the only
petrobactin-binding protein of B. subtilis. The growth inhib-
ition observed in the case of the DdhbCDfpiA mutant can be
explained by the action of petrobactin as an antibiotic through
the removal of remaining traces of iron from the medium.
Figure 3. Petrobactin-dependent thermal stabilization of FpiA. The
melting points were determined by CD spectroscopy, and maximal
stabilization (DT) was observed at a protein/ligand ratio of 1:4 or
higher.
10212 www.angewandte.org
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2010, 49, 10210 –10213

Angewandte
Chemie
[3] Using
a related methodology, the siderophore pyoverdin,
recently obtained for the petrobactin-binding proteins FatB
and FpuA from B. cereus.[5] Nevertheless, since most side-
rophore-binding proteins interact with their ligand in a 1:1
ratio, further analysis of the FpiA–petrobactin interaction is
warranted.
immobilized on gold-plated glass chips, has recently been used
to capture and identify the bacterium Pseudomonas aeruginosa
(but no specific importer): D. D. Doorneweerd, W. A. Henne,
R. G. Reifenberger, P. S. Low, Langmuir 2010, 26, 15424 – 15429.
[4] a) M. K. Wilson, R. J. Abergel, K. N. Raymond, J. E. Arceneaux,
B. R. Byers, Biochem. Biophys. Res. Commun. 2006, 348, 320 –
325; b) K. Hotta, C. Y. Kim, D. T. Fox, A. T. Koppisch, Micro-
biology 2010, 156, 1918 – 1925; c) A. T. Koppisch, C. C. Browder,
A. L. Moe, J. T. Shelley, B. A. Kinkel, L. E. Hersman, S. Iyer,
C. E. Ruggiero, BioMetals 2005, 18, 577 – 585.
[5] A. M. Zawadzka, R. J. Abergel, R. Nichiporuk, U. N. Andersen,
K. N. Raymond, Biochemistry 2009, 48, 3645 – 3657.
[6] P. E. Carlson, Jr., S. D. Dixon, B. K. Janes, K. A. Carr, T. D.
Nusca, E. C. Anderson, S. E. Keene, D. H. Sherman, P. C.
Hanna, Mol. Microbiol. 2010, 75, 900 – 909.
In conclusion, we have demonstrated that the biotinylated
petrobactin derivative 2 described here is a valuable tool for
the direct identification of siderophore-binding proteins using
affinity chromatography. The robustness of our method is
shown by the successful retention and identification of the
membrane-bound protein FpiA, the principal petrobactin-
binding protein in B. subtilis. This novel approach currently
extends to the isolation of petrobactin-binding proteins from
pathogenic bacteria, for example, B. anthracis. Importantly, it
should be useful for the capture of binding proteins that have
no homologies to known siderophore transporters. The
identification and subsequent biochemical and structural
characterization of such transport proteins represent first
steps towards the development of antibiotics that interfere
with bacterial iron transport.
[7] R. J. Abergel, M. K. Wilson, J. E. Arceneaux, T. M. Hoette,
R. K. Strong, B. R. Byers, K. N. Raymond, Proc. Natl. Acad. Sci.
USA 2006, 103, 18499 – 18503.
[8] R. A. Gardner, R. Kinkade, C. Wang, O. Phanstiel IV, J. Org.
Chem. 2004, 69, 3530 – 3537.
[9] R. J. Bergeron, G. Huang, R. E. Smith, N. Bharti, J. S. McManis,
A. Butler, Tetrahedron 2003, 59, 2007 – 2014.
[10] The propensity of citric acid based siderophores to form imides
has long been known: B. H. Lee, M. J. Miller, J. Org. Chem. 1983,
48, 24 – 31, and references therein.
Received: September 3, 2010
Published online: November 29, 2010
[11] R. J. Abergel, A. M. Zawadzka, K. N. Raymond, J. Am. Chem.
Soc. 2008, 130, 2124 – 2125.
[12] D. N. Perkins, D. J. Pappin, D. M. Creasy, J. S. Cottrell, Electro-
phoresis 1999, 20, 3551 – 3567.
[13] A. M. Zawadzka, Y. Kim, N. Maltseva, R. Nichiporuk, Y. Fan, A.
Joachimiak, K. N. Raymond, Proc. Natl. Acad. Sci. USA 2009,
106, 21854 – 21859.
[14] M. Miethke, O. Klotz, U. Linne, J. J. May, C. L. Beckering, M. A.
Marahiel, Mol. Microbiol. 2006, 61, 1413 – 1427.
[15] FpiA could show higher affinity for biotinyl petrobactin 2 on a
solid support because of a multivalent binding mode.
[16] F. Peuckert, M. Miethke, A. G. Albrecht, L.-O. Essen, M. A.
Marahiel, Angew. Chem. 2009, 121, 8066 – 8069; Angew. Chem.
Int. Ed. 2009, 48, 7924 – 7927.
Keywords: affinity chromatography · immobilization ·
petrobactin · siderophore import proteins · siderophores
.
[1] a) M. Miethke, M. A. Marahiel, Microbiol. Mol. Biol. Rev. 2007,
71, 413 – 415; b) K. D. Krewulak, H. J. Vogel, Biochim. Biophys.
Acta Biomembr. 2008, 1778, 1781 – 1804; c) R. C. Hider, X.
Kong, Nat. Prod. Rep. 2010, 27, 637 – 657.
[2] The exploitation of siderophore uptake systems to deliver
siderophore conjugates that contain antibiotic moieties (side-
romycins or “Trojan-horse antibiotics”) into bacterial cells has
been studied more extensively than the inhibition of siderophore
uptake (see Ref. [1] for an overview).
Angew. Chem. Int. Ed. 2010, 49, 10210 –10213
ꢀ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 10213