Dynamic multivalency for carbohydrate-protein recognition through dynamic combinatorial libraries based on FeII-bipyridine complexes

Article abstract of DOI:10.1002/chem.201300464

Molecular recognition: Dynamic combinatorial libraries (DCLs) exploiting multivalency effects and metal coordination have been employed for carbohydrate-protein recognition. The interaction of a three-component DCL based on 2,2′-bipyridine (bipy)-Fe^II complexes with concanavalin A (ConA; see scheme) results in enhanced binding by multivalent presentation with a bias towards mannose-containing library components. Copyright

Full text of DOI:10.1002/chem.201300464

COMMUNICATION
DOI: 10.1002/chem.201300464
Dynamic Multivalency for Carbohydrate–Protein Recognition through
Dynamic Combinatorial Libraries Based on FeII–Bipyridine Complexes
Philipp Reeh and Javier de Mendoza*[a]
Dynamic combinatorial Libraries (DCLs) have recently
been employed to enhance and optimize binding affinities
of carbohydrates to the surface of sugar-binding proteins
(lectins),[1] as their thermodynamic equilibrium is biased
towards the best-matching arrangement by simply adding
the respective target.[1a,2] It is widely accepted that multi-
valent presentation of binding sugar units is the method
of choice for enhancing binding affinities towards protein
surfaces,[3] the development of anti-pathogenic agents em-
ploying this concept is of high contemporary interest.[4] In
biological systems, polymeric sugar-based structures owe
Scheme 1. Formation of DCL-1 by ligand-scrambling around a FeII centre
(enantiomers of the complexes are omitted for clarity).
their high affinities largely to the so-called “cluster glyco-
side effect”.[5] Nevertheless, structural aspects such as the
nature of the sugar moieties involved and their orientation
at the protein surface have a large impact as well.[6]
homoleptic FeII complexes coordinated with 2,2’-bipyridine
(bipy) and 4,4’-dimethyl-bipy, respectively (Scheme 1). Com-
plexes 1 and 2 readily precipitate as PF6-salts, partially coor-
dinated side products were not detectable. This confirms
that the third binding constant of bipy ligands with FeII is
large enough to prevent partial coordination.[10] As expect-
ed, scrambling by ligand exchange to form the dynamic li-
brary containing the four possible compositions 1–4 (regard-
less of the different enantiomers) resulted in a statistical dis-
tribution after five days in acetonitrile solution (LCMS anal-
ysis, Figure 1).
Reported dynamic combinatorial approaches to lectin rec-
ognition have mostly relied on the reversible exchange of
two simple building blocks.[1b,c,7] One of the earliest exam-
ples of a prototype DCL was provided by Sasaki who syn-
thesized a homoleptic FeII complex from a galactosamine-
substituted 2,2’-bipyridine (bipy) ligand that provides the
system with four different stereochemical configurations,
merging into each other intramolecularly.[8] The reversible
system was shifted towards a specific stereoisomer in the
presence of Vicia villosa B4 lectin. Similarly, intermolecular
exchange of two terpyridyl CoII ligands was demonstrated
by Lehn and co-workers, however, not employing the strat-
egy for protein recognition.[9]
Despite their initial success, the above ideas have never
been combined to set up a multi-component exchanging
DCL for protein recognition. We report herein the interac-
tion of a three-component DCL based on bipy–FeII com-
plexes with concanavalin A (ConA), resulting in an expect-
ed bias towards mannoside components, as a proof-of-con-
cept to set up larger adaptive libraries containing a variety
of carbohydrates linked to the coordinative bipy backbone
through multiple orientations.
Figure 1. HPLC chromatogram of the FeII-based DCL-1 (reverse phase,
acetonitrile/water).
Then, bipy bis-d-mannoside (MMM), d-galactoside
(GGG) and l-fucoside (FFF) complexes were prepared to
set up a 10-member adaptive library, after mixing (DCL-2,
Scheme 2). The sugar moieties were connected to the coor-
dinative bipy backbone by a spacer long enough to provide
the necessary flexibility and distance to leave each of the
saccharide clusters to an individually induced-fit onto the
carbohydrate recognition domain (CRD) of the target
lectin.[11] The choice of sugars was guided by their known
binding affinities towards ConA (Man>Fuc,Gal),[12] and by
their predictable easy HPLC separation and identification
To test the reversibility of the resulting complexes, a
simple DCL (DCL-1) was first constructed by mixing two
[a] P. Reeh, Prof. J. de Mendoza
The Institute of Chemical Research of Catalonia (ICIQ)
Avda. Paꢀsos Catalans 16, 43007-Tarragona (Spain)
E-mail: jmendoza@iciq.es
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.201300464.
Chem. Eur. J. 2013, 19, 5259 – 5262
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
5259

Figure 2. a) Chromatogram of the equilibrated DCL-2 from left to right.
b) Fraction of DCL-2 bound to the protein surface complexes.
Scheme 2. Setup of a 10-member DCL-2 from a three-complex scram-
bling around a FeII centre.
parison of their retention times with those of reference sam-
ples. The sequence of retention times GGG<MMM<FFF
was subsequently employed to correctly locate peaks of
identical masses by interpolation, such as GGM and GMM
in the first set of signals, GGF, GMF and MMF in the
second set, and finally GFF and MFF at higher retention
times. Minor peaks corresponding to uncomplexed glyco-
side–bipy ligands (10b, 11b and 12b) were also detected in
the chromatogram.
by mass spectrometry. It is likely that mannoside-containing
complexes should prevail upon contact with the protein.
The synthesis of the necessary ligands was straightforward
(Scheme 3). Glycosylation of d-mannose or -galactose with
N-carboxybenzyl-protected diethyleneglycolamine in the
presence of BF3·Et2O exclusively provided the correspond-
ing a-isomers.[13] For l-fucose, requiring a more active glyco-
syl donor, a trichloroacetimidate strategy was selected,[14]
The subsequent steps were similar for all three pyranosides:
amine deprotection (palladium hydroxide, H2, trifluoroacetic
acid (TFA), quantitative yield), followed by coupling to the
succinimidoyl derivative (9) of 2,2’-bipyridine-4,4’-dicarbox-
ylic acid (36–61% yields), and deacetylation (94–100%
yields). Pure homoleptic tris-coordinated FeII complexes
(MMM, GGG and FFF) were finally obtained by mixing
stoichiometric amounts of the corresponding ligand and
{FeCl2·4H2O} in water (Scheme 2).
DCL-2 containing all ten expected arrangements evolved
after mixing aqueous solutions of complexes MMM, GGG
and FFF within a few days. Complexes of DCL-2 tend to
dissociate at RT at the given concentration; therefore, the
setup required a lower temperature (58C). Equilibrium was
established after 14 days, when the expected statistical distri-
bution (Gaussian normal distribution) was reached. Stability
of the library components was sufficient to be analysed by
HPLC analysis (rp column, eluent: acetonitrile/water+
0.1% TFA); their identity was again confirmed by LCMS
analysis (Figure 2a). Since the masses of galactose and man-
nose are identical, a combination of MS and retention-time
interpolation was employed for peak assignments. The three
initial homoleptic complexes were easily identified by com-
Then the same library was constructed by scrambling a
mixture of the three initial homoleptic ligands in the pres-
ence of the target (sepharose-bound ConA) and the evolv-
ing library was allowed to equilibrate for 14 days, equilibri-
um was assumed when no further changes in the proportions
of library members in the reference sample were observed.
The use of a solid-phase supported ConA allowed for the
separation of binding and non-binding entities by using a fil-
tration/elution process.[1b] The solid residue containing the
protein and its bound counterparts was then re-suspended in
HCl (0.5m) to release the bound complexes from the lectin
surface. LCMS analysis (Figure 2b) of this fraction showed
the full mannoside complex MMM as the most prevalent
species; its overall abundance compared to the reference
sample (with no protein interaction) is increased by more
than 700% (Figure 3). Other compositions (metal-bound or
free) that contained a mannoside ligand as a high affinity
recognition unit for the protein surface were found to be en-
hanced according to the number of mannosides present in
the arrangement. At least two mannoside–bipyridyl ligands
are necessary to obtain a positive change in abundance in
the protein bound fraction of the dynamic library as ob-
served for GMM and MMF (Figure 3). Complexes contain-
5260
www.chemeurj.org
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 5259 – 5262

Dynamic Multivalency for Carbohydrate–Protein Recognition
COMMUNICATION
FFF are comparably reduced in
the ConA bound fraction. Com-
pared to the original amount of
library members, approximately
80% remained in solution (due
to the excess of library used),
10% was bound to the protein
and another 10% were de-
stroyed during equilibration
(deducted from HPLC chroma-
togram integration areas), but
the only ligand that exhibits sig-
nificant binding to the carbohy-
drate
recognition
domain
(CRD) is mannoside (10b)
which is dominant at a reten-
tion time of 12.5 min in the
chromatogram (Figure 2b).
Our results clearly confirm
the expected higher affinity of
the arrangements containing a
large number of sugars with
strong interactions to the pro-
tein surface. If we assume that
complexes would bind to the
protein only by a single build-
ing block, the resulting chroma-
togram would resemble the ref-
erence sample regarding distri-
bution of species. However, the
strong bias towards DCL mem-
bers bearing higher amounts of
mannosides indicates that the
well-known multivalency effect
for protein recognition is also
present for the dynamic multi-
valent system described here.
In summary, we have shown
Scheme 3. Preparation of sugar-containing bipyridine-based ligands 10b, 11b, and 12b as building blocks for
DCL-2. CBZ=carbobenzyloxy.
that dynamic combinatorial chemistry using metal–ligand
coordination as a reversible interaction and the well-known
concept of enhanced binding to lectin surfaces by multiva-
lent presentation of sugar subunits can be combined to ana-
lyse the binding properties of a multivalent dynamic combi-
natorial library (DCL). LCMS analysis was found to be a
method of choice to detect all possible structures resulting
from the intermolecular exchange process. Analysis of the
species bound to the protein surface showed an obvious bias
towards the entities that are able to present larger numbers
of carbohydrate units of high affinity to the protein surface.
This example of a multivalent self-assembling system able to
freely exchange its binding entities on multiple (more than
two) connecting points to bind to a target proves that multi-
valency improves binding interactions in dynamic self-
assembling systems too.
Figure 3. Abundance change for the different library components. Equili-
brated DCL-2 in absence of protein and the fraction of the DCL-2 that
gets bound to the protein are being compared (determined from the re-
spective HPLC chromatograms).
ing only galactoside and/or fucoside ligands showed a lower
affinity in this protein-bound fraction. It becomes also appa-
rent that both galactoside and fucoside appear to have simi-
lar affinities for the protein, since GGG, GGF, GFF and
The approach shown here can be easily extended to wider
or more complex libraries, based on natural or unnatural
components, presented under different orientations by a
Chem. Eur. J. 2013, 19, 5259 – 5262
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
5261

J. de Mendoza and P. Reeh
bull, T. Velasco-Torrijos, S. Vidal, S. Vincent, T. Wennekes, H. Zuil-
hof, A. Imberty, Chem. Soc. Rev. 2012; DOI: 10.1039/c2cs35408j.
[5] a) J. J. Lundquist, E. J. Toone, Chem. Rev. 2002, 102, 555–578;
b) R. T. Lee, Y. C. Lee, Glycoconjugate J. 2000, 17, 543–551;
c) S. M. Dimick, S. C. Powell, S. A. McMahon, D. N. Moothoo, J. H.
Naismith, E. J. Toone, J. Am. Chem. Soc. 1999, 121, 10286–10296.
[6] a) W. I. Weis, K. Drickamer, Annu. Rev. Biochem. 1996, 65, 441–
473; b) R. D. Cummings, S. Kornfeld, J. Biol. Chem. 1982, 257,
11230–11234.
[7] a) M. Demetriades, I. K. H. Leung, R. Chowdhury, M. C. Chan,
M. A. McDonough, K. K. Yeoh, Y.-M. Tian, T. D. W. Claridge, P. J.
Ratcliffe, E. C. Y. Woon, C. J. Schofield, Angew. Chem. 2012, 124,
6776–6779; Angew. Chem. Int. Ed. 2012, 51, 6672–6675; b) S.
Andrꢂ, Z. Pei, H.-C. Siebert, O. Ramstrçm, H.-J. Gabius, Bioorg.
Med. Chem. 2006, 14, 6314–6326; c) B. Danieli, A. Giardini, G.
Lesma, D. Passarella, B. Peretto, A. Sacchetti, A. Silvani, G. Pratesi,
F. Zunino, J. Org. Chem. 2006, 71, 2848–2853.
suitable choice of the position, length and rigidity of the
linkers or the bipyridine anchoring. This opens new perspec-
tives in the understanding of the so-far elusive and delicate
rules governing the binding of carbohydrates to protein sur-
faces.
Acknowledgements
This work was supported by the Ministry of Economy and Competivity
(MINECO, Spain) (projects CTQ2008-00183 and CTQ2011-28677), the
ICIQ Foundation, and Consolider Ingenio 2010 (Grant CSD2006-0003).
Keywords: carbohydrates
·
combinatorial chemistry
·
molecular recognition · proteins · self-assembly
[8] S. Sakai, Y. Shigemasa, T. Sasaki, Tetrahedron Lett. 1997, 38, 8145–
8148.
[9] V. Goral, M. I. Nelen, A. V. Eliseev, J.-M. Lehn, Proc. Natl. Acad.
Sci. USA 2001, 98, 1347–1352.
[10] T. Saski, M. Lieberman, Tetrahedron 1993, 49, 3677–3689.
[11] T. Hasegawa, T. Yonemura, K. Matsuura, K. Kobayashi, Bioconju-
gate Chem. 2003, 14, 728–737.
[1] a) F. B. L. Cougnon, J. K. M. Sanders, Acc. Chem. Res. 2012, 45,
2211–2221; b) O. Ramstrçm, J. M. Lehn, ChemBioChem 2000, 1,
41–48; c) R. Caraballo, H. Dong, J. P. Ribeiro, J. Jimenez-Barbero,
O. Ramstrom, Angew. Chem. 2010, 122, 599–603; Angew. Chem.
Int. Ed. 2010, 49, 589–593.
[2] S. R. Beeren, J. K. M. Sanders, in History and Principles of Dynamic
Combinatorial Chemistry, Wiley-VCH, Weinheim, 2010, pp. 1–22.
[3] N. Jayaraman, Chem. Soc. Rev. 2009, 38, 3463–3483.
[4] A. Bernardi, J. Jimenez-Barbero, A. Casnati, C. C. De, T. Darbre, F.
Fieschi, J. Finne, H. Funken, K.-E. Jaeger, M. Lahmann, T. K. Lind-
horst, M. Marradi, P. Messner, A. Molinaro, P. V. Murphy, C. Nativi,
S. Oscarson, S. Penades, F. Peri, R. J. Pieters, O. Renaudet, J.-L.
Reymond, B. Richichi, J. Rojo, F. Sansone, C. Schaffer, W. B. Turn-
[12] See: http://www.functionalglycomics.org/.
[13] Z. Yang, W. Lin, B. Yu, Carbohydr. Res. 2000, 329, 879–884.
[14] a) R. R. Schmidt, J. Michel, Angew. Chem. 1980, 92, 763–764;
Angew. Chem. Int. Ed. Engl. 1980, 19, 731–732; b) R. R. Schmidt,
B. Wegmann, K.-H. Jung, Liebigs Ann. Chem. 1991, 121–124.
Received: February 5, 2013
Published online: March 11, 2013
5262
www.chemeurj.org
ꢁ 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2013, 19, 5259 – 5262