A LecA ligand identified from a galactoside-conjugate array inhibits host cell invasion by pseudomonas aeruginosa

Article abstract of DOI:10.1002/anie.201402831

Lectin LecA is a virulence factor of Pseudomonas aeruginosa involved in lung injury, mortality, and cellular invasion. Ligands competing with human glycoconjugates for LecA binding are thus promising candidates to counteract P. aeruginosa infections. We have identified a novel divalent ligand from a focused galactoside(Gal)-conjugate array which binds to LecA with very high affinity (K_d=82 nM). Crystal structures of LecA complexed with the ligand together with modeling studies confirmed its ability to chelate two binding sites of LecA. The ligand lowers cellular invasiveness of P. aeruginosa up to 90% when applied in the range of 0.05-5 μM. Hence, this ligand might lead to the development of drugs against P. aeruginosa infection.

Full text of DOI:10.1002/anie.201402831

Angewandte
Chemie
DOI: 10.1002/anie.201402831
Drug Discovery
A LecA Ligand Identified from a Galactoside-Conjugate Array
Inhibits Host Cell Invasion by Pseudomonas aeruginosa**
Alexandre Novoa, Thorsten Eierhoff, Jꢀrꢀmie Topin, Annabelle Varrot, Sofia Barluenga,
Anne Imberty,* Winfried Rçmer,* and Nicolas Winssinger*
Abstract: Lectin LecA is a virulence factor of Pseudomonas
aeruginosa involved in lung injury, mortality, and cellular
invasion. Ligands competing with human glycoconjugates for
LecA binding are thus promising candidates to counteract
P. aeruginosa infections. We have identified a novel divalent
ligand from a focused galactoside(Gal)-conjugate array which
binds to LecA with very high affinity (Kd = 82 nm). Crystal
structures of LecA complexed with the ligand together with
modeling studies confirmed its ability to chelate two binding
sites of LecA. The ligand lowers cellular invasiveness of
P. aeruginosa up to 90% when applied in the range of 0.05–
5 mm. Hence, this ligand might lead to the development of drugs
against P. aeruginosa infection.
model of infection that the lectin is involved in lung injury and
mortality[4–7] and recent data suggested that LecA promotes
bacterial invasion into host cells.[7] The lectin is a tetramer and
structural data demonstrated an overall rectangular shape
with pairing of neighboring binding sites separated by ca.
30 ꢀ. This structural knowledge has stimulated intense work
to design templates presenting galactose residues with a spac-
ing that matches the dimeric geometry of LecA neighboring
binding sites[8–16] with the potential of acting as anti-patho-
genic agents[17,18] or disrupting biofilm formation.[9,15] These
studies clearly highlighted the potential benefit of oligovalent
interactions since dissociation constants lower than 90 nm and
20 nm were obtained for tetravalent[11,16] and divalent[13]
compounds, respectively. Nevertheless, a more systematic
investigation in the linker, including more chemical diversity,
had not been carried out. Furthermore, no ligand has thus far
been shown to compromise host invasion.
In an independent work, we reported a technology to
rapidly access heteroglycoconjugate libraries emulating the
architectures of complex glycan with additional points of
diversity that included aromatic substituents probing benefi-
cial secondary interactions.[19] A profile of the lectin LecA
from P. aeruginosa revealed specific interactions that served
as the lead for a focused library (Figure 1). This library was
designed based on the selectivity of LecA for galactose, on the
ca. 30 ꢀ distance between the two adjacent binding sites, and
on the observed benefit of an aryl substituent in the linker.
Following these considerations, the library included the
following elements: four different branching geometries and
lengths (R1.2–5) as well as the unbranched monovalent
control (R1.1); five different spacer groups with different
lengths and geometries (R2.1–5); the four permutations of
1,2- 1,3- 1,4- and 1,6-a-galactose disaccharide (R3.2–5) in
addition to the b-thioarylgalactose (R3.1); five different aryl
groups in the linker moiety. This afforded a library of 625
unique monovalent and divalent glycans with a connectivity
ranging from 28 to 47 atoms between the anomeric carbons of
the terminal galactoses and diverse conformational con-
strains. The library was synthesized using peptide nucleic acid
(PNA)-encoded synthesis (see Figure 1 for the structure of
fragments).[19] The quality of each synthetic step within the
library was verified by analysis of the cleavage product from
an aliquot of resin (see the Supporting Information (SI) for
full details of synthesis and analysis).
P
athogens generally initiate their infection by using host-
specific epitopes for adhesion and penetration. Frequently,
the host-specific marker is a glycoconjugate that is recognized
by a lectin present on the pathogen surface.[1] In the case of P.
aeruginosa, the structural basis for the preference of the LecA
lectin towards galactose and globotriaosylceramide (Gb3)
was elucidated.[2,3] It was demonstrated in a P. aeruginosa
[*] Dr. A. Novoa, Dr. S. Barluenga, Prof. N. Winssinger
Department of Organic Chemistry
University of Geneva (Switzerland)
E-mail: nicolas.winssinger@unige.ch
Dr. T. Eierhoff, Prof. W. Rçmer
Institute of Biology II, Albert Ludwigs University Freiburg
Schꢀnzlestrasse 1, 79104 Freiburg (Germany)
and
BIOSS Centre for Biological Signalling Studies
Albert Ludwigs University Freiburg
Schꢀnzlestrasse 18, 79104 Freiburg (Germany)
E-mail: winfried.roemer@bioss.uni-freiburg.de
Dr. J. Topin, Dr. A. Varrot, Dr. A. Imberty
CERMAV, Universitꢁ Grenoble Alpes and CNRS
38000 Grenoble (France)
E-mail: imberty@cermav.cnrs.fr
[**] A.I., A.V. and J.T. acknowledge support from Labex ARCANE (ANR-
11-LABX-003), Association Vaincre la Mucoviscidose, and COST
actions BM1003 and CM1102. Crystal data were collected at the
European Synchrotron Radiation Facility and the authors are
grateful for the assistance in using beamlines ID14-4 et ID23-2. The
technical help of Emilie Gillon in protein production and ITC
measurement is appreciated. W.R. and T.E. were supported by the
Excellence Initiative of the German Research Foundation (EXC 294)
and by an ERC grant (ERC-2011-StG 282105). N.W. was supported
by an ERC grant (ERC 201749) and the Swiss National Science
Foundation (200021_143247).
The library was hybridized at 2.5 mm to a microarray
containing 24 copies of each DNA sequence for statistical
analysis of binding data. At this concentration, each DNA
spot on the array is saturated with the ligand. Screening for
the preferred ligand for LecA at concentrations from 50 to
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201402831.
Angew. Chem. Int. Ed. 2014, 53, 8885 –8889
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
8885

.
Angewandte
Communications
Figure 1. Design of a focused library of ligands bridging the neighboring binding sites of LecA. Left: Image generated from pdb ID: 1OKO. Right:
Structure of the library component.
1.8 nm using 3-fold dilution gave a saturated signal for the best
ligand down to 10 nm. At 5.6 and 1.8 nm of LecA, a single
compound clearly stood out as the fittest ligand (1.3–2.5–3.1–
4.3; see Figure 2 for array image and Figure 3 for titration).
The signal corresponding to this compound had a standard
deviation of less than 3% across the 24 spots providing
reliable binding accuracy. This profile suggested a clear
binding preference for a unique linker combination with the
phenoxyacetate group and b-thiophenyl galactose. For com-
parison, the B-subunit of Shiga toxin which also binds to
Gb3,[20] was profiled with this focused library. While a much
higher concentration was required to obtain binding (200 nm),
a clear preference was obtained for a different linker
combination with 4-phenoxy benzoate and a1–4-linked
digalactoside.
In order to assess the contribution of the lectin interacting
with a ligand hybridized to different DNA strands rather than
a unique divalent ligand (Figure 4), the concentration of
ligand hybridized on the array was reduced and the relative
intensity of the fittest divalent ligand (1.3–2.5–3.1–4.3) was
compared to the monomeric version (1.1–2.5–3.1–4.3). As
shown in in Figure 4, as the concentration of the ligand on the
surface is reduced (hybridizing from 2.5 to 0.03 mm of library,
3-fold dilution), the signal ratio between the fittest divalent
ligand and its monomeric version increases. This observation
is consistent with the speculation that at high concentration,
the ligand density is sufficiently high for the lectin to interact
with multiple ligands across different hybridization sites.[21] At
lower ligand concentration and hence surface density, the
binding to the divalent ligand is preferred compared to the
Figure 2. Binding profile of LecA (left) and the B-subunit of Shiga toxin (StxB) (right).
8886 www.angewandte.org
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2014, 53, 8885 –8889

Angewandte
Chemie
phenyl-b-galactoside (bPNPGal) had a dissociation constant
of 26 mm (see Table S1). The monovalent compounds SI-3 and
SI-4 containing the aryl-b-thiogalactoside and truncated
portion of the linker had higher affinity than bPNPGal (5.4
and 7 mm) but were still much weaker ligands than the
divalent SI-2, demonstrating the requirement of the two
galactose units to obtain nanomolar affinity. These results
clearly indicate a strong synergy of interaction between the
two galactose units of the divalent ligand (> 100 fold increase
in affinity per galactose unit).
The crystal structure of the fittest divalent ligand without
the nucleic acid tag (SI-2) complexed with LecAwas obtained
by cocrystallization under two different conditions (Li2SO4
and MgSO4 with a resolution of 1.65 and 1.57 ꢀ, respectively,
Table S2 and Figure S2–S4; PDB ID: 4CP9 and 4CPB). In
both structures, the asymmetric unit containing one LecA
tetramer provided clear density, however, better ligand
density was observed for the cocrystalisation using Li2SO4.
The structure obtained for the aryl-b-thiogalactoside is
consistent with previous structure of LecA having the thioaryl
Figure 3. Plot of fluorescence intensities (y-axis) measured at different
LecA concentrations (x-axis, nm, log scale) for four related com-
pounds.
ring establishing
a
T-shape interaction with His50
(Figure 5).[9,14] The triazole ring was also clearly seen and
was stabilized by a water-bridged interaction involving the
side chains of His50, Tyr36 and Asp47. Finally, the proline
ring on one side was also stabilized through water bridge
interaction with a sulfate ion. While the cocrystal structure
confirmed the divalent nature of the identified ligand and
provided information regarding interactions of the triazole
ring, a clear diffraction for the whole linker could not be
obtained (Figure S5). This may be the result of the close
packing effect of neighboring tetramer in addition to the
presence of sulfate ions from the crystallization buffer
affecting the solvent exposed portion of the linker. In order
to release the ligand from the packing forces, a molecular
dynamics trajectory was calculated with Amber force-field
using a dimer of LecA in water solution with manual building
of the ligand starting from the crystal structure observations.
Constraints on the position of the galactose residues were
maintained during 50 ns and then fully released for a further
110 ns trajectory (Figure S6–S8). During the unrestrained
trajectory, the two galactose residues were stable in their
binding sites, maintaining the coordination contacts with the
calcium ions. The adjacent thioaryl rings displayed some
mobility, but stayed in close contact with His50. The rest of
the linker had a flexible behavior, although a clustering
analysis demonstrated the emergence of preferred conforma-
tions (Figure 5). In several clusters representing a sizable part
of the trajectory, the phenoxy rings of the linker stabilized the
complex through hydrophobic interactions at the protein
surface. Stacking with neighboring thioaryl groups of the
same ligand was also observed, which may further contribute
to the stabilization.
Figure 4. Top: Schematic representation of oligomeric protein interact-
ing with monovalent versus divalent ligands on the array surface (top).
LecA is tetrameric protein with two pairs of binding sites on opposite
sides. For clarity, only half of the tetrameric protein with the two
binding sites on the same face is depicted. Bottom: Intensity ratio
between divalent ligand (1.3–2.5–3.1–4.3) and monovalent ligand (1.1–
2.5–3.1–4.3) at decreasing ligand density (concentration of PNA used
in the hybridization is the concentration for a single PNA conjugate,
that is, 1/625 of library concentration). Experiments performed with
5.6 nm of LecA.
monomeric ligand and the ratio of observed intensity reaches
a factor of 12-fold.
We next investigated the potential of the identified ligand
to block bacterial entry in the human lung epithelial cell line
H1299 using an Amikacin protection assay. In this assay,
aPNPGal (3) used at a concentration of 10 mm resulted in ca.
40% inhibition of bacterial uptake. However, when using
compound SI-1 (1.3–2.5–3.4–4.3) hybridized to its comple-
mentary DNA at a 2000 times lower concentration (5 mm), the
Next, the fittest divalent ligand (Figure 2: 1.3–2.5–3.1–4.3:
SI-1, see Figure S1 for detailed structure) was resynthesized
substituting the PNA tag for an arginine residue (SI-2) and its
affinity to LecA was measured by isothermal calorimetry
(ITC). The divalent compound displayed very strong affinity
(Kd = 82 nm) while the monovalent reference para-nitro-
Angew. Chem. Int. Ed. 2014, 53, 8885 –8889
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
8887

.
Angewandte
Communications
Figure 5. Left: Crystal structure of LecA complexed with divalent galactose compound (SI-2) and close view of the binding site with larger
fragment located by X-ray analysis (PDB ID: 4CP9). Right: Two snapshots of molecular dynamics trajectory representative of two clusters of
conformation populated for 10% and 36% of the trajectory for the top and bottom panels, respectively.
Multivalent glycocompounds have been shown to inter-
fere with lectin-mediated pathogen adhesion.[17] Glycopoly-
mers, glycodendrimers, and glyconanoparticles bearing
a large number of epitopes are efficient at binding lectins
and pathogens but they induce aggregation resulting in poor
performance as therapeutics. Conversely, glycans designed to
chelate neighboring sites[22] on a small lectin oligomer are
more promising, granted sufficient affinity can be achieved.
Such an approach has been very successful against the
pentameric B-subunit of Shiga toxin with a “starfish”
shaped compound reaching nanomolar affinity.[23] Targeting
neighboring sites of LecA has been performed using a variety
of tetravalent and divalent compounds with the best divalent
Figure 6. Inhibition of P. aeruginosa invasion of human lung epithelial
ligand reported reaching affinity range of 20–100 nm, which is
a three order of magnitude improvement compared to
monomeric galactose.[8,12–13,16] Molecular modeling suggests
that such gains in affinity are the result of bridging neighbor-
ing sites 30 ꢀ apart. The divalent ligand (SI-2) reported
herein could be characterized by a crystal structure, confirm-
ing that the linker indeed spans the distance between the sites.
This linker maintains some flexibility, as was observed in the
only two other crystal structures of lectin/multivalent com-
plexes, that is, Shiga toxin complexed with starfish inhibitor[23]
and wheat germ agglutinin complexed with bivalent GlcNAc
ligand.[24] Molecular dynamics based on structural data points
towards a stabilizing role of phenoxy residues within the
linker that can interact with hydrophobic patches on the
protein surface. The combinatorial chemistry approach
favored the selection of flexible ligands with optimal contact
with the surface and with a gain in enthalpy, which appears
complementary to the structure-based design of more rigid
cells. Bacterial cells were left untreated or incubated with the indicated
concentrations of compound SI-2 hybridized to its complementary
DNA prior to inoculation with the human lung epithelial cell line
H1299. Bacterial invasion was measured by the Amikacine protection
assay. The number of intracellular, ligand-treated bacteria was normal-
ized to untreated bacteria (values ÆSD).
internalization was reduced by 90% (Figure 6). At 50 nm, the
same ligand was still able to lower the uptake of P. aeruginosa
by 83%. Higher order oligomer of the divalent ligand
obtained by hybridization of SI-1 to various DNA or PNA
templates, or the ligand without the nucleic acid tag (SI-2)
also exhibited inhibitory activity at low micromolar concen-
trations in a dose-dependent manner, decreasing cell entry by
30–86% (Figure S9). However, the use of higher order
oligomers of the divalent ligand did not result in a significant
inhibitory benefit.
8888 www.angewandte.org
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2014, 53, 8885 –8889

Angewandte
Chemie
spacer proposed by other groups, gaining in entropy.[8,13,16,25]
The ability of this ligand to block cellular invasion might lead
to the development of promising intervention strategies
against P. aeruginosa.
[13] F. Pertici, N. J. de Mol, J. Kemmink, R. J. Pieters, Chem. Eur. J.
2013, 19, 16923 – 16927.
[14] J. Rodrigue, G. Ganne, B. Blanchard, C. Saucier, D. Giguere,
T. C. Shiao, A. Varrot, A. Imberty, R. Roy, Org. Biomol. Chem.
2013, 11, 6906 – 6918.
[15] J.-L. Reymond, M. Bergmann, T. Darbre, Chem. Soc. Rev. 2013,
42, 4814 – 4822.
[16] M. L. Gening, D. V. Titov, S. Cecioni, A. Audfray, A. G. Gerbst,
Y. E. Tsvetkov, V. B. Krylov, A. Imberty, N. E. Nifantiev, S.
Vidal, Chem. Eur. J. 2013, 19, 9272 – 9285.
Received: February 26, 2014
Revised: April 24, 2014
Published online: July 7, 2014
Keywords: bacterial invasion · glycan array · LecA · lectins ·
P. aeruginosa
[17] A. Bernardi, et al., Chem. Soc. Rev. 2013, 42, 4709 – 4727.
[18] Y. M. Chabre, R. Roy, Chem. Soc. Rev. 2013, 42, 4657 – 4708.
[19] A. Novoa, T. Machida, S. Barluenga, A. Imberty, N. Winssinger,
ChemBioChem 2014, DOI: 10.1002/cbic.201402280; for previous
examples of multivalent heteroglycoconjugates, see: V. Witt-
mann, S. Seeberger, Angew. Chem. 2004, 116, 918 – 921; Angew.
Chem. Int. Ed. 2004, 43, 900 – 903; E. Kolomiets, E. M. V.
Johansson, O. Renaudet, T. Darbre, J.-L. Reymond, Org. Lett.
2007, 9, 1465 – 1468; K. Gorska, K. T. Huang, O. Chaloin, N.
Winssinger, Angew. Chem. 2009, 121, 7831 – 7836; Angew. Chem.
Int. Ed. 2009, 48, 7695 – 7700; K. T. Huang, K. Gorska, S.
Alvarez, S. Barluenga, N. Winssinger, ChemBioChem 2011, 12,
56 – 60; M. Ciobanu, K. T. Huang, J. P. Daguer, S. Barluenga, O.
Chaloin, E. Schaeffer, C. G. Mueller, D. A. Mitchell, N. Wins-
singer, Chem. Commun. 2011, 47, 9321 – 9323; I. S. MacPherson,
J. S. Temme, S. Habeshian, K. Felczak, K. Pankiewicz, L.
Hedstrom, I. J. Krauss, Angew. Chem. 2011, 123, 11434 – 11438;
Angew. Chem. Int. Ed. 2011, 50, 11238 – 11242; S. Horiya, J. K.
Bailey, J. S. Temme, Y. V. G. Schippe, I. J. Krauss, J. Am. Chem.
Soc. 2014, 136, 5407 – 5415.
[20] L. Johannes, W. Romer, Nat. Rev. Microbiol. 2010, 8, 105 – 116.
[21] D. Sicard, S. Cecioni, M. Iazykov, Y. Chevolot, S. E. Matthews,
J. P. Praly, E. Souteyrand, A. Imberty, S. Vidal, M. Phaner-
Goutorbe, Chem. Commun. 2011, 47, 9483 – 9485.
[22] V. Wittmann, R. J. Pieters, Chem. Soc. Rev. 2013, 42, 4492 – 4503.
[23] P. I. Kitov, J. M. Sadowska, G. Mulvey, G. D. Armstrong, H.
Ling, N. S. Pannu, R. J. Read, D. R. Bundle, Nature 2000, 403,
669 – 672.
[24] D. Schwefel, C. Maierhofer, J. G. Beck, S. Seeberger, K. Die-
derichs, H. M. Moller, W. Welte, V. Wittmann, J. Am. Chem. Soc.
2010, 132, 8704 – 8719.
[25] C. Fasting, C. A. Schalley, M. Weber, O. Seitz, S. Hecht, B.
Koksch, J. Dernedde, C. Graf, E. W. Knapp, R. Haag, Angew.
Chem. 2012, 124, 10622 – 10650; Angew. Chem. Int. Ed. 2012, 51,
10472 – 10498.
.
[1] A. Imberty, A. Varrot, Curr. Opin. Struct. Biol. 2008, 18, 567 –
576.
[2] G. Cioci, E. P. Mitchell, C. Gautier, M. Wimmerova, D.
Sudakevitz, S. Perez, N. Gilboa-Garber, A. Imberty, FEBS
Lett. 2003, 555, 297 – 301.
[3] B. Blanchard, A. Nurisso, E. Hollville, C. Tetaud, J. Wiels, M.
Pokorna, M. Wimmerova, A. Varrot, A. Imberty, J. Mol. Biol.
2008, 383, 837 – 853.
[4] C. Hardalo, S. C. Edberg, Crit. Rev. Microbiol. 1997, 23, 47 – 75.
[5] V. E. Wagner, B. H. Iglewski, Clin. Rev. Allergy Immunol. 2008,
35, 124 – 134.
[6] C. Chemani, A. Imberty, S. de Bentzmann, M. Pierre, M.
Wimmerova, B. P. Guery, K. Faure, Infect. Immun. 2009, 77,
2065 – 2075.
[7] T. Eierhoff, B. Bastian, R. Thuenauer, J. Madl, A. Audfray, S.
Aigal, S. Juillot, G. E. Rydell, S. Mꢁller, S. de Bentzmann, A.
Imberty, C. Fleck, W. Rçmer, Proc. Natl. Acad. Sci. USA,
submitted.
[8] S. Cecioni, R. Lalor, B. Blanchard, J.-P. Praly, A. Imberty, S. E.
Matthews, S. Vidal, Chem. Eur. J. 2009, 15, 13232 – 13240.
[9] R. U. Kadam, M. Bergmann, M. Hurley, D. Garg, M. Cacciarini,
M. A. Swiderska, C. Nativi, M. Sattler, A. R. Smyth, P. Williams,
M. Cꢂmara, A. Stocker, T. Darbre, J.-L. Reymond, Angew.
Chem. 2011, 123, 10819 – 10823; Angew. Chem. Int. Ed. 2011, 50,
10631 – 10635.
[10] S. Cecioni, V. Oerthel, J. Iehl, M. Holler, D. Goyard, J.-P. Praly,
A. Imberty, J.-F. Nierengarten, S. Vidal, Chem. Eur. J. 2011, 17,
3252 – 3261.
[11] S. Cecioni, J.-P. Praly, S. E. Matthews, M. Wimmerova, A.
Imberty, S. Vidal, Chem. Eur. J. 2012, 18, 6250 – 6263.
[12] F. Pertici, R. J. Pieters, Chem. Commun. 2012, 48, 4008 – 4010.
Angew. Chem. Int. Ed. 2014, 53, 8885 –8889
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 8889