Manipulation of electrostatic and saccharide linker interactions in the design of efficient glycopolypeptide-based cholera toxin inhibitors

Article abstract of DOI:10.1002/mabi.200900182

Multivalent, glycopolymer inhibitors designed for the treatment of disease and pathogen infection have shown improvements in binding correlated with general changes in glycopolymer architecture and composition. We have previously demonstrated that control

Full text of DOI:10.1002/mabi.200900182

Full Paper
Manipulation of Electrostatic and Saccharide
Linker Interactions in the Design of Efficient
Glycopolypeptide-Based Cholera Toxin Inhibitorsa
Ronak Maheshwari, Eric A. Levenson, Kristi L. Kiick*
Multivalent, glycopolymer inhibitors designed for the treatment of disease and pathogen infection
have shown improvements in binding correlated with general changes in glycopolymer archi-
tecture and composition. We have previously demonstrated that control of glycopolypeptide
backbone extension and ligand spacing significantly impacts the inhibition of the cholera toxin B
subunit pentamer (CT B5) by these polymers. In the studies reported here, we elucidate the role of
backbone charge and linker length in modulating the inhibition event. Peptides of the
sequence AXPXG (where X is a positive, neutral or negative amino acid), equipped with the
alkyne functionality of propargyl glycine, were designed and synthesized via solid-phase peptide
synthetic methods and glycosylated via Cu(I)-catalyzed alkyne-azide cycloaddition reactions. The
capacity of the glycopeptides to inhibit the binding of the B5 subunit of cholera toxin was
evaluated. These studies indicated that glycopeptides with a negatively charged backbone show
improved inhibition of the binding event relative to the other glycopeptides. In addition, variations
in the length of the linker between the peptide and the
saccharide ligand also affected the inhibition of CT by
the glycopeptides. Our findings suggest that, apart from
appropriate saccharide spacing and polypeptide chain
extension, saccharide linker conformation and the sys-
tematic placement of charges on the polypeptide back-
bone are also significant variables that can be tuned to
improve the inhibitory potencies of glycopolypeptide-
based multivalent inhibitors.
R. Maheshwari, K. L. Kiick
Department of Materials Science and Engineering, University of
Delaware, 201 DuPont Hall, Newark, Delaware 19716 USA
Introduction
Fax: þ1 (302) 831-4545; E-mail: kiick@udel.edu
E. A. Levenson
Multivalent interactions play a pivotal role in biology and
Department of Chemistry and Biochemistry, University of
Delaware, Newark, Delaware 19716 USA
are present in numerous biomolecular interactions, such as
those of glycoproteins, integrins, antibodies, and viruses.[1]
In particular, protein-carbohydrate interactions have been
extensively investigated, since they play an important role
in several processes, including pathogen recognition,
inflammation, cell signaling, differentiation, and adhesion
of various bacterial toxins.[2,3] Given the relatively low
affinity of the protein-carbohydrate interaction, nature
K. L. Kiick
Delaware Biotechnology Institute, 15 Innovation Way, Newark,
Delaware 19711 USA
a : Supporting information for this article is available at the bottom
of the article’s abstract page, which can be accessed from the
journal’s homepage at http://www.mbs-journal.de, or from the
author.
Macromol. Biosci. 2010, 10, 68–81
68
ß 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
DOI: 10.1002/mabi.200900182

Manipulation of Electrostatic and Saccharide Linker Interactions in the Design . . .
employs multivalent interactions between proteins and
carbohydrates to increase the avidity of the interaction and
provide specificity. This improvement in binding is often
referred to as the ‘‘cluster glycoside effect’’.[4]
glucose at controlled ratios and densities on dendrimers
and showed that the avidity of the multivalent interaction
canbeeasilymanipulatedbytuningthedensityandratioof
monovalent ligands.[7] The significance of ligand density in
the inhibition of anthrax toxin by peptide-modified
polymers was also illustrated by the Kane group; in these
studies, poly(N-acryloyloxysuccinimide) of controlled
molecular weights was modified at various densities with
the peptide ligand (HTSTYWWLDGAP), and the polymeric
multivalent ligands exhibited an optimal ligand density for
toxin inhibition.[33]
The dynamic nature of biological interactions suggests
that a certain degree of flexibility may be valuable in
designing multivalent constructs,[5] to optimize both
enthalpic and entropic effects during binding. In addition
to the flexibility that appropriately designed multivalent
polymer-ligand conjugates may offer, they also provide
locally high concentrations of ligands, to yield greater
binding avidity and specificity compared to their mono-
valent counterparts. The multivalent presentation of
ligands also reduces the apparent disassociation rate of
receptor-ligand complexes,[6] which, in the long-term
application of multivalent therapeutics, may afford
advantages in efficacy and residence time.[7] An additional
advantage of multivalent approaches to ligand design is
that polymer-based materials can be tailored to many
varied architectures or conformations to optimize interac-
tions with targets, via the utilization of appropriate
polymeric scaffolds, monomeric ligands and conjugation
strategies.[8,9]
We have investigated both the role of architectural
control of the polymer backbone, as well as ligand
presentation, in the binding of the B5 subunit of cholera
toxin (CT) by linear glycopolypeptide inhibitors.[34–36]
Cholera toxin is produced by the Vibrio cholerae bacterium
and has an AB5 architecture shared by the Escherichia coli
heat labile enterotoxins. The pentameric B5 structure binds
specifically to the ganglioside GM1 (Gal-b1-3GalNAc
b1-4(Neu5Aca2-3)Gal-b1-Glc-ceramide) present on the sur-
face ofintestinalepithelial cells andgains entryintothe cell
via endocytosis, upon which the A subunit is cleaved and
initiates the catalytic events that result in host dehydration
and diarrhea.[37] From the crystal structure of GM1-bound
CT it is known that two terminal sugars of the GM1
pentasaccharide, galactose and sialic acid, are mainly
Significant research has been undertaken for the
development of high affinity, polymer-derived multivalent
ligands to inhibit multivalent interactions and selectively
modify cell-signaling.[1,10–24] In particular, these studies
have focused on factors such as clustering of ligands, ligand
density and valency, as well as length and interactions of
intermolecular linkers. Many pioneering groups have con-
tributed to this continually developing field.[7,9,14,24–27] For
instance, the Kiessling group has created multivalent
neoglycopolymers to promote the proteolytic cleavage of
L-selectin and thereby potentially modulate inflammatory
responses. In addition, multivalent 2,4-dinitrophenyl (DNP)
antigens produced via ring-opening metathesis polymer-
ization (ROMP) permitted studies of the relationship
between B cell receptor (BCR) clustering, localization,
internalization and signal amplification. The extent of
signaling,degreeofBCRclusteringandantibodyproduction
were dependent on antigen valency, while the BCR
internalization was not.[28] Other works by this group
examined the influence of multivalent ligand epitope
density on the clustering of Con A.[10,29,30] Sampson and
coworkers also used ROMP to produce peptide-modified
polymers that have elucidated details of the interaction of
sperm protein fertilin b with its egg receptor.[31] In another
study directed at design of dendrimeric multivalent
ligands, Cloninger and coworkers investigated the binding
of a series of mannose-functionalized PAMAM dendrimers
(generations two through six) with monovalent and
divalent derivatives of Con A and reported a statistical
increment in binding with increase in the generation of
dendrimers.[18,32] They also incorporated mannose and
responsible for the interaction. The binding sites between
[37]
˚
adjacent B subunits are approximately 35 A apart.
The
known structure and biological functions of CT render it a
useful system for unraveling the design rules in the
production of multivalent inhibitors, as well as to provide
additional and detailed insights into the general design of
multivalent ligands.
In our previous studies,[34–36,38] we demonstrated
the production of well-defined, high molecular weight
and monodisperse glycopolypeptides via recombinant
methods, to study the role of multivalent interactions in
the inhibition of CT. With these recombinantly produced
glycopolypeptides, we investigated the role of architectural
control over the polymer backbone as well as ligand
presentation in the binding of the CT B5. Results from these
studies indicated that optimal spacing of the saccharide
˚
ligands at a distance of 35 A was an important parameter in
designing improved inhibitors of CT binding, and also
demonstrated that the composition of the polypeptide
chain influences binding separate from considerations of
saccharide spacing. When a high percentage of glycine
residues (approximately 45%) was included in sequences
that adopt a random-coil secondary structure, the glyco-
polypeptides adopted compact structures that hindered
binding, whereas charged polyglutamic acid-based glyco-
polymers exhibited improved inhibition owing to a larger
hydrodynamic volume and improved accessibility of
saccharide groups.[35] We postulated that the ionic charge
Macromol. Biosci. 2010, 10, 68–81
ß 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.mbs-journal.de
69

R. Maheshwari, E. A. Levenson, K. L. Kiick
of the glutamic acid residues might also afford additional
advantages in binding, particularly if these charges
could be placed judiciously and selectively along the
polymer backbone, as is possible only via recombinant
methods.
hydrodynamic volume and secondary structures of the
glycopeptides. A direct enzyme linked assay (DELA) was
used to determine the glycopeptide-based inhibition of the
binding of CT B5, and illustrated the effect of charge and
linker length on inhibition.
The crystallographic details of CT B5 reveal that the
regions around and between adjacent GM1 binding sites of
CT B5 contain multiple positively charged amino acid
residues (as shown in Figure S1).[37] The incorporation of
glutamic acid residues in the chain and particularly around
the saccharide ligand sitemight thereforeplay adual role in
both extending the polymer chain and complementing the
positive charges around the saccharide-binding site of
the CT B5, resulting in additional improvements in the
interaction of glycopeptides and glycopolypeptides with
the CT B5 surface.
Experimental Part
Materials and Methods
Fmoc-protected natural amino acids, Fmoc-protected propargyl-
glycine and 2-(1H-benzotriazole-1-yl)-1,1,3,3-tetramethyluronium
hexafluorophosphate (HBTU) were purchased from EMD Bios-
ciences Inc. (San Diego, CA). The cholera toxin B5 subunit
horseradish peroxidase conjugate (CT B5-HRP) was obtained from
List Biological Laboratories (Campbell, CA). Ganglioside GD1b and
C96 Maxisorp microtiter plates were obtained from Matreya
(Pleasant Gap, PA) and Fisher Scientific (Pittsburgh, PA) respec-
tively. Azido b-D-galactopyranoside, 1,2,3,4,6-penta-O-acetyl-b-D-
galactopyranoside, dimethylsulfoxide (DMSO), OHꢀ[CH2]2ꢀBr,
OHꢀ[CH2]3ꢀBr, BF3.Et2O and all other reagents were obtained
from Sigma-Aldrich Co. (St. Louis, MO) and were used as received
without any further purification.
In this work, we report three random-coil-based bivalent
glycopeptides designed to carry different charges in testing
of this hypothesis. Peptides that are either negatively
charged [35-RCE-2 (XG AEAEPEG AEAEPEG AEAEPEG
AEAEPEG AEAEPXG)], positively charged [35-RCK-2
(XG AKAKPKG AKAKPKG AKAKPKG AKAKPKG AKAKPXG)],
or neutral [35-RCG-2 (XGAGAGPSG AGAGPSG AGAGPSG
AGAGPSG AGAGPXG)] were designed and synthesized via
solid-phase synthetic methods. The single letter abbrevia-
tions denote the appropriate amino acids, while the
position X is representative of the terminal alkyne-
functionalized amino acid propargylglycine. Root mean
Peptide Synthesis
Peptides, with the sequences shown in Table 1, were synthesized
via standard solid-phase peptide synthesis methods on a PS3TM
automated peptide synthesizer (Protein Technologies, Inc., Tucson,
AZ), usingRinkamideMBHAresin asa polymersupport. Theamino
acid residues were activated with HBTU in the presence of 0.4 M
methyl morpholine in DMF. Deprotection reactions were carried
out with 20% piperidine in DMF. Cleavage of the glycopeptides or
peptide from the resin was performed in 95:2.5:2.5 trifluoroacetic
acid (TFA)/triisopropylsilane (TIS)/water for 6–7 h. TFA was
evaporated and cleavage products were precipitated in ether.
The water-soluble glyco/peptides were dissolved in water and
lyophilized. Peptides and glycopeptides were purified via reverse
phase HPLC (RP-HPLC) (Waters, MA, USA), using a Symmetry C-18
column. Identities of peptides and glycopeptides were confirmed
via ESI or MALDI mass spectroscopy and 1H NMR spectroscopy.
˚
square distances of ca. 32–40 A between propargylglycines
were calculated by assuming a freely jointed chain model
and were corroborated by molecular dynamic simulations
and random flight models. The alkyne functionality of
propargylglycine allows the coupling of azido-functiona-
lized galactose with various length linkers via Cu-catalyzed
azide alkyne [3 þ 2] Huisgen cycloaddition.[39] Three azido-
functionalized galactopyranosides – azido-b-D-galactopyr-
anoside, 2-azidoethyl-O-b-D-galactopyranoside, and 3-azi-
dopropyl-O-b-D-galactopyranoside – were employed to
modify the above peptides. Reverse-phase high perfor-
mance liquid chromatography (RP-HPLC), electrospray
ionization mass spectrometry (ESI-MS) and 1H nuclear
magnetic resonance (1H NMR) spectroscopy confirmed the
purity and quantitative modification of the peptides with
the galactopyranosides. Gel permeation chromatography
(GPC) and circular dichroic (CD) spectroscopy indicated the
Synthesis of Azido-Functionalized
Galactopyranosides
Syntheses of azido-functionalized galactopyranosides as shown in
Scheme1wereconductedaspreviouslyreported.[40] Threegramsof
Table 1. Composition of synthesized peptides. X is propargylglycine.
Peptide
Sequence
35-RCE-1
35-RCE-2
35-RCK-2
35-RCG-2
GAEAEPEGAEAEPEGAEAEPEGAEAEPEGAEAEPXGGGG
XGAEAEPEGAEAEPEGAEAEPEGAEAEPEGAEAEPXGGGG
XGAKAKPKGAKAKPKGAKAKPKGAKAKPKGAKAKPXGGGG
XGAGAGPSGAGAGPSGAGAGPSGAGAGPSGAGAGPXGGGG
Macromol. Biosci. 2010, 10, 68–81
70
ß 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
DOI: 10.1002/mabi.200900182

Manipulation of Electrostatic and Saccharide Linker Interactions in the Design . . .
Synthesis of Monovalent
Saccharides and Monovalent/
Bivalent Glycopeptides
For the synthesis of monovalent saccharide
controls, Fmoc-propargylglycine was attached
to Rink amide MBHA resin and the N-terminus
of the amino acid was acetylated using acetic
anhydride. The alkyne of the propargylglycine
was then modified with azido-b-D-galactopyr-
anoside, 2-azidoethyl-O-b-D-galactopyranoside
or 3-azidopropyl-O-b-D-galactopyranoside via
Huisgen 3 þ 2 cycloaddition protocols, as
shown in Scheme 2.[41] Reactions were con-
ducted on resin (solid phase) and the general
protocol for the cycloaddition reaction was as
follows: 1 eq of alkyne, 3 eq of azide, 0.5 eq of
CuAc and 0.5 eq of tris(triazolyl amine) were
dissolved in 2 mL DMSO, and the reaction was
Scheme 1. Synthesis of azido-functionalized galactopyranosides with ethyl or propyl
linkers; n ¼ 2 or 3, a: BF3.OEt2, CH2Cl2, 0 8C, b: NaN3, DMF, 100 8C, and c: NaOMe in MeOH,
Amberlite 15 resin.
carried out at 80 8C for 24 h. Then the resin was thoroughly washed
with thiourea, DMF and DCM to remove copper and other
unreacted reactants. The glycosylated product was then cleaved
from the resin as described above. For the synthesis of the
monovalent and bivalent glycopeptides, the peptide sequences
shown in Table 1 were first produced on the Rink amide MBHA
resin, as described above. After synthesis of the peptide, the
propargylglycine residues were glycosylated via the same proce-
dure just described. All monovalent saccharides and all glycopep-
tides were purified using RP-HPLC and identities were confirmed
via 1H NMR spectroscopy and ESI-MS (see the Supporting
Information).
1,2,3,4,6-penta-O-acetyl-b-D-galactopyranoside (7.7 mmol) and
1.5 eq of OHꢀ[CH2]nꢀBr were dissolved in dry DCM (25 mL)
˚
containing molecular sieves 4 A (1.5 g). The solution was cooled
to 0 8C and BF3 ꢁ Et2O (5 mL) was added slowly over a time period of
2 min. The reaction mixture was stirred overnight at 0 8C and then
diluted with excess DCM, followed by washing with a solution of
cold water (200 mL), saturated NaHCO3 (200 mL) and NaCl (200 mL)
and drying using MgSO4. The solution was filtered and concen-
trated via rotary evaporation at 55 8C. The identity of the product
(1a in Scheme 1, light yellow syrup) was confirmed via 1H NMR
spectroscopy, 13C NMR spectroscopy (CDCl3) and ESI-MS (see the
SupportingInformation). 1aandsodiumazide(4 eq)weredissolved
in DMF (15 mL) and stirred overnight at 100 8C. Again the reaction
mixture was diluted in DCM and unreacted sodium azide was
removed by washing with excess cold water, saturated NaCl and
drying using MgSO4. The solution was filtered and concentrated.
De-acetylation of the azido-sugars (1b) was conducted by reaction
with 5 eq of 25 wt.-% NaOMe/MeOH in dry methanol for 2 h.
Sodium ions were exchanged with protons via treatment with
Amberlite 15 resin. The final product (1c) was filtered and dried via
rotary evaporation; evaluation of 1H NMR and ESI-MS spectra (see
the Supporting Information), coupled with comparisons to those
previously reported,[40] confirmed the identity and the essentially
complete deacetylation of the saccharide products.
Instrumental Methods
NMR Spectroscopy
1H NMR spectra were acquired on a Bruker DRX-400 NMR
spectrometer under standard quantitative conditions at ambient
temperature. Electrospray ionization (ESI) mass spectrometry on
saccharides, peptides and glycopeptides were performed at the
Mass Spectrometry Facility in the Department of Chemistry and
Biochemistry at the University of Delaware.
Synthesis of Tris(triazolyl amine)
Tripropargyl amine (7.5 mmol) and benzyl
azide (26.3 mmol) were dissolved in 10 mL of
DMSO. Copper sulfate (0.378 mmol) and
sodium ascorbate (1.89 mmol) were dissolved
in 1 mL of water. The two solutions were mixed
and stirred at 60 8C for 12 h. The reaction
product, tris(triazolyl amine), was purified
via re-crystallization in CHCl3/ethyl ether.[41]
Theoretical and experimental (M þ Na)þ
¼
553.3 Da. The product was also characterized
via 1H NMR spectroscopy (see the Supporting
Information), confirming its identity.[41]
Scheme 2. General protocol for cycloaddition reactions performed on the solid phase.
Macromol. Biosci. 2010, 10, 68–81
ß 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.mbs-journal.de
71

R. Maheshwari, E. A. Levenson, K. L. Kiick
Circular Dichroic Spectroscopy
and linker arm on the CT B5-inhibitory potency of the low
valency and low molecular weight glycopeptides; these
studies were intended to elucidate the salient features to be
incorporated into recombinantly derived glycopolypep-
tides. The synthesized peptides (Table 1) with alkyne
functionality from propargylglycine residues were mod-
ifiedwithazido-functionalizedgalactopyranosidesviaCu(I)
alkyne-azide cycloaddition reactions.[39] The glycosylation
of the peptides (with protected amino acid side-chains) was
carried out on the solid phase by using one of the
three azido-functionalized galactopyranosides: azido-b-D-
galactopyranoside, 2-azidoethyl-O-b-D-galactopyranoside
or 3-azidopropyl-O-b-D-galactopyranoside. Glycopeptides
were then deprotected (i.e., deprotection of the protected
amino acid side chains) and cleaved from the resin, and
purified via HPLC and characterized using ESI-MS and
1H NMR spectroscopy (see the Supporting Information).
The 1H NMR spectra confirmed the maintenance of the
b-linkage of the saccharides. The degree of glycosylation
was measured via 1H NMR spectroscopy and MS analyses,
and was indicated to be greater than 90% for all
glycopeptides, as shown in Table 2.
Circular dichroic spectra were recorded on a JASCO 810 spectro-
photometer (Jasco, Inc., Easton, MD) using a 1 mm path length
quartz cuvette. Samples with concentrations of approximate
50 ꢂ 10ꢀ6 M were prepared using PBS buffer (10 ꢂ 10ꢀ3 M phos-
phate, 0.150 M NaCl, pH 7.3)andscans wererecordedby subtracting
the PBS background. 400 mL of samples were loaded in the quartz
cuvette and the data points for the wavelength-dependent CD
spectra were recorded with a 1 nm bandwidth.
Gel Permeation Chromatography
Gel permeation chromatography (GPC) was carried out to
determine the relative hydrodynamic volume of peptides and
glycopeptides for comparison with a control PGA30 peptide
[poly(glutamic acid) with a degree of polymerization of approxi-
mately 30]. Samples were dissolved at approximately 1 mg ꢁ mLꢀ1
in phosphate-buffered saline (PBS, pH 7.3), and filtered through a
0.22 mm filter. The samples were separated using a Waters
Ultrahydrogel Linear column (7.8 ꢂ 300 mm) followed by
a
Ultrahydrogel 250 (7.8 ꢂ 300 mm) column. Detection was achieved
via the use of a Waters 2996 Photodiode Array Detector and a
Waters 2414 Refractive Index Detector.
CT GD1b Direct Enzyme-Linked Assay
The GD1b direct enzyme-linked assays (DELA) were carried out in a
96-well format as previously reported.[35,42] Microtiter plates were
incubated at 37 8C for 16 h with 100 mL (2 mg/mL) ganglioside GD1b
dissolved per well in PBS (pH 7.4). Unattached ganglioside was
removed by washing the wells three times with PBS. Additional
binding sites on the plate surface were blocked by incubating the
wells with 200 mL of a 1% (w/v) bovine serum albumin (BSA)-PBS
solution for 30 min at 37 8C and then washing them three times
with 0.05% Tween 20-PBS. All the glycopeptide samples were
prepared by serial dilution from their stock solutions. Samples
consisted of 6 ng ꢁ mLꢀ1 cholera toxin B subunit conjugated to
horseradish peroxidase (CTB5-HRP) incubated for 2 h in the
presence of ligand at varying concentrations. 100 mL of these
samples were incubated in a well for 30 min at room temperature
and unbound toxin was removed by washing three times with
0.05% Tween 20-PBS. Toxin bound to GD1b was then revealed by
addition of 100 mL of Ultra TMB solution (Pierce) for 10–15 min
followed by 100 mL of 2 M H2SO4 and recording of the absorbance at
450 nm on a microtiter plate reader. IC50 values were calculated
from at least 10 different concentrations of saccharide ligands via
non-linear regression analysis, as described previously,[5] using the
Microcal Origin software package.
Secondary Structures of Glycopeptides
Circular dichroic (CD) spectroscopy was conducted in order
to evaluate the secondary structures of the peptides and
glycopeptides. The mean residue ellipticity as a function of
wavelength was recorded for peptide and glycopeptide
solutions (50 ꢂ 10ꢀ6 M in PBS buffer at pH 7.3) at 25 8C. As
indicated in Figure 1, the spectra for all samples shows a
minima near 198 nm, confirming the random-coil second-
ary structure of all peptides and glycopeptides,[43] and
confirming, as expected, that the modification of peptides
with saccharides, with the concomitant formation of the
triazole ring, does not affect the secondary structure of the
peptide backbone. As also illustrated in the spectrum,
the region near 220 nm shows identical positive bands
(maxima) for the glutamic acid-rich and lysine-rich
peptides and glycopeptides, consistent with previous
spectra for extended poly-L-(glutamic acid) or poly-L-lysine
at pH 7,[44] suggesting a similar extended, random-coil
secondary structure of the charged glycopeptides. In
contrast, the spectrum of glycine-rich peptides and
glycopeptides showed the absence of any positive band
in this region, suggesting that they form a more compact
random-coil secondary structure as would be expected
based on their high glycine content (approximately 50% for
the peptides).
Results
Synthesis of Glycopeptides
A series of well-defined, low molecular weight glycopep-
tides of random-coil secondary structure and different
electrostatic charge were synthesized. These peptides were
glycosylated with galactopyranosides with various linkers
in order to evaluate the effects of electrostatic interactions
Gel Permeation Chromatography
In our previous studies, we demonstrated that the
hydrodynamic volumes of glycopolypeptides play a vital
Macromol. Biosci. 2010, 10, 68–81
72
ß 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
DOI: 10.1002/mabi.200900182

Manipulation of Electrostatic and Saccharide Linker Interactions in the Design . . .
Table 2. DELA results of monovalent saccharides and glycopeptides.
Inhibitor
Degree of
IC50 (saccharide
Improvement
over galactose
Molecular
weight
glycosylation
concentration)
10ꢀ3 T M
Da
Galactose
Gal-PGa)
NA
NA
30–45
2.8 ꢃ 0.1d)
0.4 ꢃ 0.03e)
0.25 ꢃ 0.02e)
0.66 ꢃ 0.05e)
2.6 ꢃ 0.2e)
4.0 ꢃ 0.1d)
0.88 ꢃ 0.03
>3
1
180
359
15
Gal-35-RCE-1
Gal-35-RCE-2
Gal-35-RCG-2
Gal-35-RCK-2
Gal-Et-PGb)
100 ꢃ 1
100 ꢃ 2
95 ꢃ 2
95 ꢃ 3
NA
100
160
60
3 876
4 176
3 287
4 163
403
15
10
Gal-Et-35-RCE-2
Gal-Et-35-RCK-2
Gal-Prop-PGc)
Gal-Prop-35-RCE-2
Gal-Prop-35-RCK-2
100 ꢃ 1
94 ꢃ 2
NA
45
4 264
4 251
417
NA
8
5.0 ꢃ 0.2d)
>1
100 ꢃ 2
92 ꢃ 2
NA
NA
4 292
4 279
>3
a)Gal-PG: propargylglycine modified with azido-b-D-galactopyranoside; b)Gal-Et-PG: propargylglycine modified with 2-azidoethyl-O-b-D-
galactopyranoside; c)Gal-Prop-PG: propargylglycine modified with 3-azidopropyl-O-b-D-galactopyranoside; d)IC50 values of monovalent
saccharide ligands are significantly different from one another (Student’s t-test showed p < 0.05); e)IC50 values of monovalent and divalent
glycopeptides are significantly different from one another (p < 0.05).
role in their inhibitory potencies toward CT B5.[35] Thus, the
relative retention times (and hydrodynamic volumes) of
the peptides and glycopeptides (35-RCG-2, Gal-35-RCG-2,
35-RCE-2, Gal-35-RCE-1, Gal-35-RCE-2, Gal-Et-35-RCE-2 and
Gal-Prop-35-RCE-2) were evaluated via GPC to evaluate the
effects of charge on the hydrodynamic volumes relative to
the PGA30 [poly(glutamic acid) with a degree of polymer-
ization of approximately 30] employed in our original
studies; the resulting data are shown in Figure 2. The
absorbance at 214 nm was monitored as a function of time
for solutions of peptides and glycopeptides (1 mg ꢁ mLꢀ1) in
PBS at pH 7.3; the observed peak widths for the peptides/
glycopeptides were consistent with those obtained for
monodisperse poly(ethylene oxide) (PEO) standards. As
shown in the figure, PGA30, with the greatest charge
density, eluted at the earliest time (14.4 min) and is
therefore indicated to have the largest hydrodynamic
volume. The somewhat pronounced shoulder in this peak
results from the presence of a lower molecular mass species
in this polydisperse macromolecule as it is received from
Figure 1. Circular dichroic spectroscopy results for peptides and
glycosylated peptides (PBS, 25 8C).
Figure 2. Gel permeation chromatography of peptides and gly-
copeptides compared to the standard peptide PGA30.
Macromol. Biosci. 2010, 10, 68–81
ß 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.mbs-journal.de
73

R. Maheshwari, E. A. Levenson, K. L. Kiick
themanufacturer. Thepeptides35-RCE-2and35-RCE-1(and
their corresponding glycopeptides), with approximately
35% glutamic acid content, exhibited mainly monomodal
peaks with longer retention times (15.1 min for both
peptides) and lower hydrodynamic volume compared to
PGA30. A lower molecular mass shoulder is observed only
for the Gal-RCE-1, and may result from the presence of a
small amount of degradation products and/or unglycosy-
lated species in this sample. The uncharged, glycine-rich
35-RCG-2 had the longest retention time of the peptides/
glycopeptides (16.3–16.4 min), and the correspondingly
lowest hydrodynamic volume, as expected. The high degree
of glycine residues (approximately 50% for the peptide) and
absence of any electrostatic repulsion causes 35-RCG-2 and
Gal-35-RCG-2 to form a more compact conformation
reflected in their higher retention times. The differences
in the hydrodynamic volume of the peptides and their
glycopeptides were negligible (the elution profiles for all
35-RCE-2 and 35-RCE-1-based peptides and glycopeptides
overlapped), consistent with our previous results in which
significant differences in the hydrodynamic volume of a
polypeptide and its corresponding glycopolypeptide were
only observed when there was a change in charge density
due to glycosylation or methylation.[35]
Figure 4. Dose response curves for the inhibition of CTB5 by
monovalent saccharides (Gal-PG, Gal-Et-PG and Gal-Prop-PG),
glycopeptides (Gal-35-RCE-2, Gal-Et-35-RCE-2 and Gal-Prop-35-
RCE-2) and galactose as determined via DELA.
for these various inhibitor species were calculated from the
dose response inhibition curves and the inhibition
improvements over monovalent galactose were compared,
as shown in Table 2. Unmodified 35-RCE-Z (Z ¼ 1 or 2) at
concentrations greater than 1 ꢂ 10ꢀ3 M and 35-RCK-2 and
Inhibition Assays
35-RCG-2 at concentrations greater than 1.5 ꢂ 10ꢀ3
M
The potential of these glycopeptides and monovalent
saccharides to act as efficient inhibitors of the binding of
CT B5 was tested using competitive DELA experiments
developed by Minke and coworkers.[35,42] Inhibition curves
from DELA experiments for different glycopeptides and
monovalent saccharides as a function of saccharide
concentration are shown in Figure 3 and 4. IC50 values
exhibited only a small degree of non-specific binding (less
than 20%, not shown). There was no visible aggregation
after addition of any inhibitors to the horseradish
peroxidase (HRP)-linked CTB5 (CTB5-HRP). The concentra-
tions of bifunctional glycopeptides were therefore varied
from 0 to 3 ꢂ 10ꢀ3
M saccharide concentration (0 to
1.5 ꢂ 10ꢀ3 M peptide concentration) to minimize nonspe-
cific binding, while the concentration of the monovalent
galactose was varied from 0 to 100 ꢂ 10ꢀ3 M.
Non-linear regression analysis of the dose response
inhibition curves of glycopeptides yielded IC50 values
(saccharide concentration) of 0.4 ꢂ 10ꢀ3
,
0.25 ꢂ 10ꢀ3
,
0.88 ꢂ 10ꢀ3, 0.66 ꢂ 10ꢀ3 and 2.63 ꢂ 10ꢀ3 M for Gal-35-RCE-
1, Gal-35-RCE-2, Gal-Et-35-RCE-2, Gal-35-RCG-2 and Gal-35-
RCK-2, respectively. Galactose exhibits an IC50 of
40 ꢂ 10ꢀ3 M in these assays, consistent with previous
reports.[42,45] The improvements in inhibition for these
glycopeptidescomparedtomonovalentgalactosewere100,
160, 45, 60 and 15, respectively. Gal-Prop-35-RCE-2 did not
show any inhibition up to 1 ꢂ 10ꢀ3 M concentration and
Gal-Et-35-RCK-2 and Gal-Prop-35-RCK-2 did not show any
inhibition up to 3 ꢂ 10ꢀ3 M concentration (see the Support-
ing Information). IC50 values for these glycopeptides were
therefore not evaluated, because nonspecific interactions
from the peptide backbone might begin to interfereat these
highconcentrationvalues. DELAexperimentsconducted on
the three monovalent saccharides Gal-PG, Gal-Et-PG and
Figure 3. Dose response curves for the inhibition of CT B5 binding
by the glycopeptides (Gal-35-RCE-2, Gal-35-RCK-2, Gal-35-RCG-2
and Gal-35-RCE-1) and monovalent galactose as determined via
DELA.
Macromol. Biosci. 2010, 10, 68–81
74
ß 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
DOI: 10.1002/mabi.200900182

Manipulation of Electrostatic and Saccharide Linker Interactions in the Design . . .
Gal-Prop-PG (at concentrations of 0 to 40 ꢂ 10ꢀ3 M) exhib-
due to charge or peptide chain conformations.[48] Since the
negatively charged 35-RCE-Z (Z ¼ 1 or 2) and positively
charged 35-RCK-2 peptide chains have similar character-
istic ratios, they should exhibit similar conformational
flexibility and hydrodynamic volumes (an extended ran-
dom-coil secondary structure was suggested by the CD
spectra),[44,48] allowing assessment of the impact of back-
bone charge on the binding event. 35-RCG-2, a glycine-rich
peptide in contrast, should exhibit a more flexible and
compact conformation compared to the charged peptides
35-RCE-2 and 35-RCK-2. The root mean square (RMS)
distances between the reactive propargylglycine residues
(PG) in all three peptides were calculated assuming a freely
jointed chain model with glycine as the point of joints. The
RMS value in case of 35-RCG-2 was further confirmed by the
random flight model using the formula hR2i ¼ (Cn np
l2p),[47,49] where Cn is the characteristic ratio, np the number
of segments, and lp the length of each segment (here the
distance between a-carbon atoms in the peptide bond, ca.
ited IC50 values of 2.8 ꢂ 10ꢀ3, 4.0 ꢂ 10ꢀ3 and 5.0 ꢂ 10ꢀ3
M
respectively, as shown in Figure 4 and Table 2, correspond-
ing to 15-, 10- and 8-fold improvements over galactose.
Although the differences in the slopes of these inhibition
curves suggest that there may be slight variation in the
modeofinhibitionorthataggregationmaybeoccurring, no
aggregation was apparent in these samples or in previous
samples analyzed via DLS.[46]
Discussion
Design of Glycopeptides
In our previous studies, we reported that glycopolypeptides
with large hydrodynamic volumes (Vh) arising from their
high negative charge density, and with spacing between
adjacent saccharide ligands similar to the CT binding sites,
exhibit higher inhibitory potencies against CT relative to
more flexible and more neutral glycopolypeptides.[35] Our
previous results showed that the larger Vh of glycopolypep-
tides, even for those glycopolypeptides of relatively low
molecular weight, was correlated with improved toxin
inhibition most likely due to the better accessibility of
pendantsaccharides. Whilethe previousglycopolypeptides
were negatively charged, similar chain extension might be
possible via the use of positively charged amino acids, and
variations in electrostatic interactions may render one type
of chain an improved inhibitor. The solved crystal structure
of the complex of CTB5 and GM1 reveals that the regions
between the GM1 binding sites of CTB5 are composed of
mainly positively charged (and neutral) amino acid
residues as illustrated in Figure S1.[37] The path between
˚
3.8 A). Although Cn is dependent on the chain length, it is
equal to C for poly(Gly) chains with n > 10,[50,51] so Cn was
1
assumed to be equal to C given that the peptides were
1
40 amino acids long; C was previously calculated to be
1
2–3.2 for glycine-rich peptides.[47,49] The calculated average
˚
distance between PGs in 35-RCG-2 was 32–40 A (by
employing Cn ¼ 2–3.2). Similarly, calculated distances
between PGs in 35-RCE-2 and 35-RCK-2 were within the
˚
range 35–42 A, calculated assuming the freely jointed chain
model and confirmed by molecular dynamic simulation
studies in the case of the 35-RCE-2 (see the Supporting
Information).Thus, thevariationinchargesinthesepeptide
sequences should permit assessment of the differences in
inhibition that arise due to differences in electrostatic
interactions or peptide chain conformation.
˚
two adjacent binding sites, approximately 35 A in length, is
The inclusion of alkyne-derivatized side-chains (via the
use of propargylglycine) permits facile modification of the
peptides with three different azido-functionalized galacto-
pyranosides. These strategies are also useful for future
studies of glycopolypeptide-based multivalent ligands,
owing to the facile incorporation of the alkyne-functiona-
lized homopropargylglycine in recombinantly derived
proteins.[52,53] The use of the three azido-functionalized
galactopyranosides indicated in Scheme 2 results in linker
arms of different lengths between the saccharide and
peptide backbone, with corresponding differences in
triazole positions with respect to the terminal galactose
moiety. A comparison of the relative binding activity of the
three monovalent saccharides therefore provides informa-
tion about the interactions due to specific molecular
contacts that could be made between the triazole ring
and the CT B5 binding site. Comparison of the inhibition by
glycopeptides with the various saccharide ligands affords
additional insight into the combined role of the linker and
peptide backbone and their joint impact on inhibition.
not severely hindered by any amino acid residue side-
chains. The use of negatively charged glycopeptides, rather
than positively charged chains, should likely enhance
binding through additional electrostatic interactions with
the basic, positively charged surface of CTB5 around the
saccharide binding pocket. Recombinant methods of
glycopolypeptide design would allow the synthesis of such
polypeptides in which the position of negatively charged
amino acids can be specified.
Given our ultimate interests in the design of such
glycopolypeptides, peptides of random-coil secondary
structure, with the sequences shown in Table 1, were
synthesized to compare the role of overall backbone
charge on the inhibitory potencies of these molecules.
The inclusion of high percentages of charged residues or
glycines, with their different characteristic ratios (C for
1
Gly-rich polypeptides of ꢄ2–3.2 and C for Glu- or Lys-rich
1
polypeptides of approximately 9),[47] in the target glyco-
peptidespermittedexplorationoftheinhibitiondifferences
Macromol. Biosci. 2010, 10, 68–81
ß 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.mbs-journal.de
75

R. Maheshwari, E. A. Levenson, K. L. Kiick
Determining the Effects of Charge on the Inhibition
of CT by Glycopeptides
no inhibition at these IC50 concentrations, and hence
nonspecific electrostatic interactions of the peptide back-
bone with the CT B5 were not the primary origin of binding.
Glycopeptides that were either negatively charged (Gal-35-
RCE-Z), positively charged (Gal-35-RCK-2), or neutral (Gal-
35-RCG-2) were tested as inhibitors of the cholera toxin B5
subunit via the use of the competitive direct enzyme-linked
assay (DELA) in which binding of CT B5 to a GD1b-modified
surface was monitored as a function of ligand concentra-
tion.[35,42] The improvements in inhibition for these
glycopeptides compared to monovalent galactose were
100, 160, 60 and 15 for Gal-35-RCE-1, Gal-35-RCE-2, Gal-35-
RCG-2 and Gal-35-RCK-2, respectively. Given that the inter-
residue spacing of the propargylglycines (and thus the
saccharides) were indicated to be approximately the same
for all peptides, the differences in inhibition improvement
for the glycopeptides must arise due to the differences in
either conformation and/or charge and suggest that
negatively charged glycopeptides offer measurable advan-
tages in inhibition. Although the differences in the actual
free energy of binding are small for the glycopeptides, their
statistically significant difference (see below) suggests the
potential for electrostatic charge in altering the avidity
of the binding event in glycopolypeptides of similar
composition.
Role of Charge in the Inhibition of CTB5
by Glycopeptides
Glycopeptides Gal-35-RCE-2 and Gal-35-RCK-2 possess
similarities in charge density (although opposite charge),
side-groups, characteristic ratio,[48] extended random-coil
secondary structure (from CD spectroscopy studies) and
hydrodynamic volumes. Thus, the saccharides in these two
glycopeptides should have similar spacing and levels of
accessibility for CT binding sites. However, Gal-35-RCK-2
exhibited only a 15-fold improvement in inhibition over
monovalent galactose, as shown in Figure 5, demonstrating
a significant difference in the inhibition potencies of these
glycopeptides (p ¼ 0.002). The regions proximal to the CT B5
saccharide-binding sites contain basic amino acid residues
(Figure S1), which may interact with the glycopeptide, and
thus the presence of negative or positive charges on the
glycopeptide might result in respective electrostatic
attraction or repulsion with the CT B5 surface upon
saccharide binding. Consistent with this hypothesis, the
data for Gal-RCE-2 and Gal-RCK-2, along with the data
aboveforGal-RCE-1, corroboratetheroleofthechargeofthe
peptide in enhancing or diminishing the glycopeptide
interactions with CT B5. Toone and coworkers have also
reported similar effects in the inhibition of Shiga-like toxin
(SLT) by hydrophobic and hydrophilic bivalent glycopep-
tides (5 amino acids long) equipped with Pk trisaccharide
ligands. Both the glycopeptides showed enhancements in
inhibition over the monovalent Pk trisaccharide ligand and
the extent of inhibition depended on the nature of the
peptide.Anapproximately20-foldenhancementinbinding
Role of Valency in the Inhibition of CTB5 by
Negatively Charged Glycopeptides
The bivalent chargedglycopeptides can interact with CT via
interactions with the saccharide ligands, the linkers, and
the charged backbone of the glycopeptide. Interactions of
negatively charged monovalent (Gal-35-RCE-1) and diva-
lent (Gal-35-RCE-2) glycopeptides with CT B5 were com-
pared to assess the differences in inhibition due to the
valency of the saccharide unit. As shown in Table 2,
there was a statistically significant difference (p < 0.03) in
theinhibitionofCTbyGal-35-RCE-1(100-foldimprovement
over galactose) and Gal-35-RCE-2 (160-fold) over mono-
valent galactose. Given that the conformations (from CD)
andhydrodynamicvolumes(fromGPC)ofGal-35-RCE-1and
Gal-35-RCE-2 were essentially identical, the differences in
binding can be ascribed to the number of saccharides on the
scaffold. The origins of the significant inhibition exhibited
by the Gal-35-RCE-1 (100-fold improvement over galactose)
likely arise from initial interaction of the saccharide with
the CT B5, strengthened by secondary interactions of the
charged peptide backbone with the surrounding basic
chargedaminoacidresiduesontheCTB5.Theimportanceof
both the saccharide and peptide backbone interactions
withCTB5 inmediatingthisbindingissuggestedbythefact
that the monovalent saccharide (Gal-PG) showed only a
15-fold improvement in binding over galactose, and by the
additional fact the unmodified 35-RCE-Z peptides showed
Figure 5. Comparison of the inhibition improvement values of
glycopeptides Gal-35-RCE-2, Gal-35-RCG-2 and Gal-35-RCK-2 over
ꢅ
galactose. Indicates statistically significant differences in inhi-
bition versus other scaffolds (p < 0.05).
Macromol. Biosci. 2010, 10, 68–81
76
ß 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
DOI: 10.1002/mabi.200900182

Manipulation of Electrostatic and Saccharide Linker Interactions in the Design . . .
was observed for glycopeptides with a hydrophobic spacer
while only a 10-fold enhancement in binding was observed
for glycopeptides with a hydrophilic spacer, as compared to
the monovalent saccharide ligand.[54]
The importance of the charge of the backbone was
further underscored by comparisons of the inhibition
improvements exhibited by Gal-RCE-2 (160-fold), Gal-
RCK-2 (15-fold) and Gal-35-RCG-2, which exhibited a
60-fold improvement in inhibition relative to the mono-
valent galactose (Figure 5). The combination of CD spectro-
scopy and GPC studies showed that the compact glycine-
rich glycopeptides exhibited a smaller hydrodynamic
volume than the charged glycopeptides. In light of our
previous studies that illustrated inhibition improvements
with increasing hydrodynamic volumes, the Gal-RCG-2
glycopeptides may have been expected to show poorer
inhibition than that of the charged glycopeptides of greater
hydrodynamic volume, due to a combination of entropic
and saccharide accessibility effects. The fact that the Gal-
RCG-2 instead exhibited significantly improved inhibition
(p ¼ 0.005) over that of the Gal-RCK-2 confirms the
important role of charge in the inhibitory event. Thus,
these studies suggest a key advantage in the use of
glycopeptide and glycopolypeptide approaches in the
design of inhibitors: charged residues can be placed
judiciously on the chain to affect inhibition via multiple
mechanisms. The chain extension afforded by the inclusion
of the glutamic acid residues may serve not only to improve
saccharide accessibility, but also to reduce entropic
penalties experienced by the chain upon binding, as the
flexibility of the chain is reduced. In addition, complemen-
tary matching of the charges on the chain with the charges
on the binding target also improves binding. The basic
amino acid residues on CT B5 in the region around the GM1
Figure 6. Comparison of the inhibitory potencies of glycopeptides
and glycopolypeptides[35] as a function of retention time.
the CT B5,[35] exhibits inhibition that is improved over that
expected on the basis of its size. The newly synthesized
glycine-rich Gal-35-RCG-2 (Point C), shows inhibition
improvements that are smaller than that of the larger
Cap-35-RC-6 (Point B), as expected owing to the lower
hydrodynamic volume of Gal-35-RCG-2, but that are still
greater than those expected for a molecule of its size. With
the additional increase in hydrodynamic volume of the
negatively charged Gal-35-RCE-2 (Point A, the largest of the
three species despite its low molecular mass), concomitant
improvements in inhibition of CT B5 were observed, again,
with improvements greater than those expected on the
basis of its size. This glycopeptides also shows inhibition
greater than that of the other well-defined molecules. Thus,
the inhibition exhibited by these glycopeptides and
glycopolypeptides supports that molecules with ligand
spacing matched to the target receptors provide improve-
mentsininhibitiongreaterthanthoseexpectedonthebasis
of hydrodynamic volume, and that increasing hydrody-
namic volume with appropriate electrostatic complemen-
tarity provides further improvements in binding. These
comparisons thus suggest that increasing the number of
ligands by producing high molecular weight glycopolypep-
tides with a similar sequence as that of 35-RCE-Z should
result in inhibitors with much greater inhibitory potency.
Thus, despite previous reports that correlate mainly
polymer molecular mass with inhibition potency,[55–57] our
study suggests that both variations in hydrodynamic
volume (rather than molecular mass) and differences in
charge can be exploited to enhance or diminish the
inhibitory potencies of glycopeptide- and glycopolypep-
tide-based inhibitors. Toone and coworkers had shown
significant differences in the enhancement in inhibition of
SLT by varying the hydrophobicity of glycopeptides
(20- versus 10-fold);[54] our findings demonstrate a much
larger difference in the inhibition (160- versus 15-fold) of
˚
binding sites are present at approximately 10.6 and 14 A
when measured from the Ca of the basic residue to the C4 of
the terminal galactose of the GM1-bound ligand. In the
sequence of Gal-35-RCE-2, glutamic acid residues adjacent
to the saccharide are in the proximity of these distances,
electrostatically complementing the basic charged amino
acid residues of CT B5, while the remaining glutamic acids
provide chain extension. Such desired placement can also
be expanded to polymeric polypeptide systems via
recombinant methods.
Figure 6 shows the inhibition improvements for a range
of glycopolypeptide and glycopeptide inhibitors as a
function of their retention time in the GPC experiment;
these data illustrate the potential impact on inhibition of
chain extension and matched architecture. The trend line in
Figure 6 highlights the relationship of inhibition with
retention time for heterogeneous glycopolypeptides; it is
clear from these data that the architecturally well-defined
glycopolypeptide Cap-35-RC-6 of our previous studies
(Point B), with a saccharide spacing matched to that of
Macromol. Biosci. 2010, 10, 68–81
ß 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.mbs-journal.de
77

R. Maheshwari, E. A. Levenson, K. L. Kiick
CT B5 by variations in the charge on the glycopeptide
backbone. In addition, the improvements in inhibition
exhibited by the glycopeptides here (up to 160-fold over
galactose) are of the range usually observed for large and
high valency linear glycopolymers and dendrimers.[32,58–61]
Although the inhibitory potencies by these glycopeptides
are much smaller than the small pentavalent molecule
inhibitors reported by Fan and coworkers,[62] these designs
allow assessment of the role of complementary secondary
interactions in multivalent binding events. Hence these
glycopeptide designs serve as the basis for designing new
high molecular weight glycopolypeptide based inhibitors
of higher binding avidity via appropriate choice of amino
acid residues and architectural design.
position of the triazole was varied to determine its impact
on the inhibitory event.
Various monovalent saccharide linkers were made by
modifying propargylglycine with azido-functionalized
galactopyranosides [azido-b-D-galactopyranoside (Gal-PG),
2-azidoethyl-O-b-D-galactopyranoside
(Gal-Et-PG),
or
3-azidopropyl-O-b-D-galactopyranoside (Gal-Prop-PG)] via
Cu(I)-catalyzed azide-alkyne cycloaddition (CuAAC),[39,71]
resulting in the formation of a triazole ring whose position,
with respect to the galactose moiety, was varied by
adjusting the length of the linker, as shown in Figure 7,
which illustrates the lowest energy conformations of the
three saccharide linkers determined at a temperature of
300 K by energy minimization calculations utilizing MM2
force fields.[72,73] DELA evaluation of these three mono-
valent saccharides showed quite small but statistically
significant differences (p < 0.002) in inhibition, with
inhibition improvements of 15-, 10- and 8-fold over
galactose for Gal-PG, Gal-Et-PG and Gal-Prop-PG respec-
tively. The binding improvements observed for the Gal-PG
are similar to N-(e-aminocaproyl)-b-D-galactosylamine
(data not shown),[42] illustrating that the inclusion of a
linker arm containing a triazole ring does not significantly
reduce binding relative to a saccharide with an alkyl linker
(or relative to the unmodified saccharide), although the
specific origins of the improvements afforded by the
inclusion of the triazole are not clear in the absence of
crystal structure data. Interestingly, although the increased
Linker Architecture and Effect of Triazole Placement
Apartfrom polymerarchitectural features, thedesignofthe
linker which presents the saccharide or other ligand
moieties is also important for designing inhibitors for
bacterial toxins.[63,64] GM1, a ganglioside found on the
surface of intestinal epithelial cells, is the natural ligand for
CT with a dissociation constant in the 10ꢀ9 M range.[65] The
majority of interactions in the CT-GM1 complex are from
terminal saccharide moieties (from galactose with addi-
tional contributions from the sialic acid group) and thus
maintenance of the terminal galactose as an anchor point
for linkers of various compositions has been an approach
previously adopted.[66,67] For example, m-nitrophenyl a-
galactoside (MNPG) was shown by Fan and coworkers to
exhibit favorable interactions with the binding sites of
CT B5. In this saccharide ligand, the m-nitrophenyl group
played a key role in the excellent inhibition observed,
with X-ray crystallography studies on the complex of
CT B5-MNPG revealing that this saccharide ligand makes
favorable contacts with the CT B5 binding pocket via
interactions of the nitro group with surrounding amino
acids.[68] In addition, N-(e-aminocaproyl)-b-D-galactosyl-
amine has been shown to elicit improvements in inhibition
of CT B5 through favorable hydrophobic interactions with
the CT B5 binding pocket.[17,34,35] In this work, we employed
alkyne-azide cycloaddition strategies, which result in the
formation of a triazole group; this coupling strategy has
been widely employed to modify polymers by attaching
saccharide or other ligands.[26,69] However, the impact on
the binding event of the triazole and its placement has not
yet been thoroughly probed in the design of multivalent
systems. The large dipole and steric size of the 1,2,3-triazole,
and the capacity of the N2 and N3 electron lone pairs to
serve as hydrogen bonding acceptors,[70] could affect the
interaction of the triazole with the CT B5 and alter binding.
We thus maintained the terminal galactose and introduced
the triazole group on a short hydrophobic chain; the
˚
length (ca. 16–17A between the Gal C4 and the a-carbon of
the amino acid) and hydrophobicity of the alkyl linker
arm of N-(e-aminocaproyl)-b-D-galactosylamine has been
shown to improve this molecule’s inhibition of CT B5 over
that of galactose,[42] in our studies here, the
Gal-PG, with the shortest and least hydrophobic linker
˚
(ca. 7–8 A), shows slightly improved binding over that
˚
˚
exhibited by Gal-Et-PG (ca. 11 A) and Gal-Prop-PG (ca. 8–9 A,
shorterowingtothenonlineartrajectory). Theplacementof
the triazole ring near the terminal galactose moiety in the
Gal-PG may slightly improve binding via direct interaction
of the triazolewith the CT B5 bindingpocket, as observed for
thenitrophenylgroupinMNPG(seeabove)inthestudiesby
Fan and coworkers. These effects may also arise from the
Figure 7. Lowest energy conformation of saccharide linkers:
a) Gal-PG, b) Gal-Et-PG, and, c) Gal-Prop-PG.
Macromol. Biosci. 2010, 10, 68–81
78
ß 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
DOI: 10.1002/mabi.200900182

Manipulation of Electrostatic and Saccharide Linker Interactions in the Design . . .
impact of the triazole on the conformation of the linker arm
(see below).
by Gal-Prop-35-RCE-2 is even poorer than that of the
monovalent Gal-Prop-PG. Any potential differences in
conformational entropy loss of the peptide backbone for
the three triazole-linked glycopeptides would further
complicate these comparisons of inhibition; isothermal
titration calorimetry experiments, as well as crystal-
lography characterization, would shed more detail on these
speculations. Nevertheless, it is clear from these studies
that the position of the triazole ring, with its impact on the
presentation of saccharide ligands attached to the peptide
backbone, is an important variable in the design of related
glycopolypeptide-based inhibitors of CT B5. Taken with the
results above, our studies therefore suggest the high
inhibitory potential of negatively charged, recombinantly
produced polypeptide backbones, equipped with alkyne
groups and modified via CuAAC protocols with Gal-PG-
based saccharides. Production and characterization of such
scaffolds is underway.
These three saccharides were conjugated to the glutamic
acid-rich bivalent peptide 35-RCE-2. GPC results showed
that there was essentially no variation in the hydrody-
namic volumes, and CD spectra (data not shown) indicated
no measurable differences in conformation, thus any
differences in inhibition can be ascribed to differences in
saccharide linker architecture. DELA experiments were
performed on these three glycopeptides and the inhibition
improvements over galactose were compared. Gal-35-RCE-
2 and Gal-Et-35-RCE-2 exhibit IC50 values of 250 ꢂ 10ꢀ6 and
880 ꢂ 10ꢀ6 M (saccharide concentrations) which correspond
to an inhibition improvement of 160- and 45-fold
respectively over galactose (p < 0.001). Gal-Prop-35-RCE-2
did not show any inhibition up to 1 ꢂ 10ꢀ3 M peptide
concentration, after which nonspecific interactions of the
peptide backbone with the CT B5 become apparent. The
presentation of these three saccharides on the positively
charged peptide backbone (35-RCK-2) showed a similar
trend in the inhibition of CT binding. The Gal-35-RCK-2
exhibited the best inhibition of the three positively charge
peptides, with the lowest IC50 value (2.63 ꢂ 10ꢀ3 M), as
compared to Gal-Et-35-RCK-2 and Gal-Prop-35-RCK-2 with
IC50 values of greater than 3 ꢂ 10ꢀ3 M (p < 0.05, SI). Given
the marginal differences in inhibition by the monovalent
saccharides, these results suggest that the disparity in the
inhibition of CT B5 by the glycopeptides may be due to
differences in the accessibility of the saccharides on the
different linkers when those linkers are anchored to the
peptide chain. The charge of the chain further exacerbates
these differences.
Conclusion
In this work, we have reported three bivalent glycopeptides
withrandom-coilsecondarystructuresanddifferentcharge
identities. The presence of negative or positive charges on
the glycopeptide resulted in respective electrostatic attrac-
tion or repulsion with the CT B5 surface near the binding
pocket, which significantly affected the inhibition exhib-
ited by the glycopeptides. The negatively charged glyco-
peptides showed the greatest inhibition and the positively
charged glycopeptides showed the worst inhibition; a
neutral glycine-rich glycopeptide showed intermediate
inhibition despite its smaller size and lower saccharide
accessibility, underscoring the role that charge plays in
modulating the binding of polypeptide-based inhibitors to
the CT B5 target, and suggesting opportunities to specifi-
cally design those interactions in polymeric inhibitors
produced via recombinant methods. In addition, the
coupling of azido-functionalized galactose with various
length linkers via Cu(I)-catalyzed azide-alkyne cycloaddi-
tion has also been shown to be useful in the generation of
efficient inhibitors. The position of the triazole ring can
be easily varied by adjusting the length of the linker,
and our results demonstrate that a saccharide ligand
equipped with the triazole ring adjacent to the galactose
moiety shows better inhibition as compared to other
geometries, most likely due to both the placement of the
triazole ring and corresponding better accessibility of
the terminal galactose on the glycopeptide scaffold. Our
results point to the optimal design features for the
production of inhibitors of bacterial toxins, and suggest
the now-available possibility of employing such stringent
design control in the production of improved multivalent
ligands.
The lengths of linkers used in this study result in the
formation of a triazole ring at different locations, resulting
in a different conformation for each of the linkers (Figure 7).
The most drastic deviation from a linear trajectory for the
linkers occurred for the 3-azidopropyl-O-b-D-galactopyra-
noside derivative as shown in Figure 7c; there was also
some evidence of a preferred nonlinear chain trajectory
for the 2-azidoethyl-O-b-D-galactopyranoside derivative
(Figure 7b). This increased length and preference for a
nonlinear chain trajectory corresponds with slightly
decreased inhibitory potencies for the monovalent tria-
zole-linked saccharides that were substantially magnified
when these saccharides were presented on a bivalent
glycopeptide scaffold. In the simplest interpretation, the
introduction of a kink in the linker arm with increasing
linker length (when the saccharide is attached to the
peptide) may force the galactose to lie closer to the
peptide backbone resulting in a loss of accessibility and
specificity for galactose in its interactions with CT. The
accessibility of galactose in the case of Gal-Prop-PG
is the worst due to the maximum deviation from a
linear trajectory and consequently the inhibition of CT
Macromol. Biosci. 2010, 10, 68–81
ß 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.mbs-journal.de
79

R. Maheshwari, E. A. Levenson, K. L. Kiick
Acknowledgements: This work was funded in part by grant
numbers 1-P20-RR017716 and 1-P20-RR015588 (instrument facil-
ities) from the National Center for Research Resources (NCRR), a
component of the National Institutes of Health (NIH). Its contents
are solely the responsibility of the authors and do not necessarily
represent the official views of the NCRR or the NIH. This work was
also funded in part by the Center for Neutron Science at the
University of Delaware under award (US Department of Commerce)
number 70NANB7H6178. Senthil Kandasamy and Ronald Larson
(University of Michigan) are thanked for conducting molecular
dynamics simulations.
[21] J. D. Reuter, A. Myc, M. M. Hayes, Z. Gan, R. Roy, R. Yin, L. T.
Piehler, R. Esfand, D. A. Tomalia, J. R. J. Baker, Bioconjugate
Chem. 1999, 10, 271.
[22] E. K. Woller, M. J. Cloninger, Org. Lett. 2002, 4, 7.
[23] E. K. Woller, E. D. Walter, J. R. Morgan, D. J. Singel, M. J.
Cloninger, J. Am. Chem. Soc. 2003, 125, 8820.
[24] T. K. Dam, C. F. Brewer, Biochemistry 2008, 47, 8470.
[25] Y. Lee, N. S. Sampson, Curr. Opin. Struct. Biol. 2006, 16, 544.
[26] V. Ladmiral, G. Mantovani, G. J. Clarkson, S. Cauet, J. L. Irwin,
D. M. Haddleton, J. Am. Chem. Soc. 2006, 128, 4823.
[27] J. C. Lee, K. A. Parker, N. S. Sampson, J. Am. Chem. Soc. 2006,
128, 4578.
[28] E. B. Puffer, J. K. Pontrello, J. J. Hollenbeck, J. A. Kink, L. L.
Kiessling, ACS Chem. Biol. 2007, 2, 252.
[29] C. W. Cairo, J. E. Gestwicki, M. Kanai, L. L. Kiessling, J. Am.
Chem. Soc. 2002, 124, 1615.
Received: May 18, 2009; Revised: July 15, 2009; Published online:
September 24, 2009; DOI: 10.1002/mabi.200900182
[30] P. Mowery, Z.-Q. Yang, E. J. Gordon, O. Dwir, A. G. Spencer, R.
Alon, L. L. Kiessling, Chem. Biol. 2004, 11, 725.
[31] K. A. Baessler, Y. Lee, K. S. Roberts, N. Facompre, N. S. Sampson,
Chem. Biol. 2006, 13, 251.
[32] S. L. Mangold, M. J. Cloninger, Org. Biomol. Chem. 2006, 4,
2458.
Keywords: glycopeptides; glycopolypeptides; multivalency;
toxin inhibition
[33] K. V. Gujraty, A. Joshi, A. Saraph, V. Poon, J. Mogridge, R. S.
Kane, Biomacromolecules 2006, 7, 2082.
[34] B. D. Polizzotti, K. L. Kiick, Biomacromolecules 2006, 7, 483.
[35] B. D. Polizzotti, R. Maheshwari, J. Vinkenborg, K. L. Kiick,
Macromolecules 2007, 40, 7103.
[36] S. Liu, K. L. Kiick, Macromolecules 2008, 41, 764.
[37] E. A. Merritt, S. Sarfaty, F. Vandenakker, C. Lhoir, J. A. Martial,
W. G. J. Hol, Protein Sci. 1994, 3, 166.
[1] J. Huskens, Curr. Opin. Chem. Biol. 2006, 10, 537.
[2] M. Mammen, S.-K. Chio, G. M. Whitesides, Angew. Chem. Int.
Ed. 1998, 37, 2755.
[38] Y. Wang, K. L. Kiick, J. Am. Chem. Soc. 2005, 127, 16392.
[39] H. C. Kolb, M. G. Finn, K. B. Sharpless, Angew. Chem. Int. Ed.
2001, 40, 2004.
[3] A. Varki, Glycobiology 1993, 3, 97.
[4] Y. C. Lee, R. T. Lee, Acc. Chem. Res. 1995, 28, 321.
[5] W. P. Bowen, J. C. Jerman, Trends Pharm. Sci. 1995, 16,
413.
[40] J. A. F. Joosten, V. Loimaranta, C. C. M. Appeldoorn, S. Haataja,
F. A. ElMaate, R. M. J. Liskamp, J. Finne, R. J. Pieters, J. Med.
Chem. 2004, 47, 6499.
[6] K. Nishikawa, S. Sawasdikosol, D. A. Fruman, J. Lai, Z.
Songyang, S. J. Burakoff, M. B. Yaffe, L. C. Cantley, Mol. Cell
2000, 6, 969.
[41] Q. Wang, T. R. Chan, R. Hilgraf, V. V. Fokin, K. B. Sharpless,
M. G. Finn, J. Am. Chem. Soc. 2003, 125, 3192.
[42] W. E. Minke, C. Roach, W. G. J. Hol, C. L. M. J. Verlinde,
Biochemistry 1999, 38, 5684.
[7] M. L. Wolfenden, M. J. Cloninger, Bioconjugate Chem. 2006, 17,
958.
[8] K. L. Kiick, Science 2007, 317, 1182.
[43] V. P. Saxena, D. B. Wetlaufer, Proc. Natl. Acad. Sci. USA 1971,
68, 969.
[9] L. L. Kiessling, J. E. Gestwicki, L. E. Strong, Angew. Chem. Int. Ed.
2006, 45, 2348.
[44] M. L. Tiffany, S. Krimm, Biopolymers 1973, 12, 575.
[45] I. Vrasidas, N. J. De Mol, R. M. J. Liskamp, R. J. Pieters, Eur. J.
Org. Chem. 2001, 4685.
[10] E. J. Gordon, W. J. Sanders, L. L. Kiessling, Nature 1998,
392, 30.
[11] N. Nagahori, S.-I. Nishimura, Biomacromolecules 2001,
2, 22.
[46] B. K. Shoichet, J. Med. Chem. 2006, 49, 7274.
[47] L. Mandelkern, W. L. Mattice, Biochemistry 1971, 10, 1934.
[48] W. G. Miller, C. V. Goebel, Biochemistry 1968, 7, 3925.
[49] W. Shen, J. A. Kornfield, D. A. Tirrell, Soft Matter 2007, 3, 99.
[50] T. H. Evers, E. M. W. M. van Dongen, A. C. Faesen, E. W. Meijer,
M. Merkx, Biochemistry 2006, 45, 13183.
[12] M. Mammen, G. Dahmann, G. M. Whitesides, J. Med. Chem.
1995, 38, 4179.
[13] H. Kamitakahara, T. Suzuki, N. Nishigori, Y. Suzuki, O. Kanie,
C.-H. Wong, Angew. Chem. Int. Ed. 1998, 37, 1524.
[14] E. J. Baird, D. Holowka, G. W. Coates, B. Baird, Biochemistry
2003, 42, 12739.
[51] P. R. Schimmel, P. J. Flory, Proc. Natl. Acad. Sci. USA 1967, 58,
52.
[15] H. Lis, N. Sharon, Chem. Rev. 1998, 98, 637.
[16] P. I. Kitov, H. Shimizu, S. W. Homans, D. R. Bundle, J. Am.
Chem. Soc. 2003, 125, 3284.
[52] J. C. van Hest, K. L. Kiick, D. A. Tirrell, J. Am. Chem. Soc. 2000,
122, 1282.
[53] K. L. Kiick, J. C. M. van Hest, D. A. Tirrell, Angew. Chem. Int. Ed.
2000, 39, 2148.
[17] E. K. Fan, Z. S. Zhang, W. E. Minke, Z. Hou, C. Verlinde, W. G. J.
Hol, J. Am. Chem. Soc. 2000, 122, 2663.
[54] J. J. Lundquist, S. D. Debenham, E. J. Toone, J. Org. Chem. 2000,
65, 8245.
[18] S. L. Mangold, J. R. Morgan, G. C. Strohmeyer, G.A.M., M. J.
Cloninger, Org. Biomol. Chem. 2005, 3, 2354.
[19] T. K. Dam, S. Oscarson, R. Roy, S. K. Das, D. Page, F. Macaluso,
C. F. Brewer, J. Biol. Chem. 2005, 280, 8640.
[20] E. J. Gordon, L. E. Strong, L. L. Kiessling, Bioorg. Med. Chem.
1998, 6, 1293.
[55] J. E. Gestwicki, C. W. Cairo, L. E. Strong, K. A. Oetjen, L. L.
Kiessling, J. Am. Chem. Soc. 2002, 124, 14922.
[56] K. Totani, T. Kubota, T. Kuroda, T. Murata, K. I. P. J. Hidari, T.
Suzuki, Y. Suzuki, K. Kobayashi, H. Ashida, K. Yamamoto, T.
Usui, Glycobiology 2003, 13, 315.
Macromol. Biosci. 2010, 10, 68–81
80
ß 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
DOI: 10.1002/mabi.200900182

Manipulation of Electrostatic and Saccharide Linker Interactions in the Design . . .
[57] A. Spaltenstein, G. M. Whitesides, J. Am. Chem. Soc. 1991, 113,
686.
[66] A. Bernardi, A. Checchia, P. Brocca, S. Sonnino, F. Zuccotto,
J. Am. Chem. Soc. 1999, 121, 2032.
[58] J. E. Gestwicki, L. E. Strong, L. L. Kiessling, Chem. Biol. 2000, 7, 583.
[59] K. H. Schlick, R. A. Udelhoven, G. C. Strohmeyer, M. J. Clo-
ninger, Mol. Pharmaceutics 2005, 2, 295.
[67] A. Bernardi, D. Arosio, L. Manzoni, D. Monti, H. Posteri, D.
Potenza, S. Mari, J. Jimenez-Barbero, Org. Biomol. Chem. 2003,
1, 785.
[60] C. C. M. Appeldoorn, J. A. F. Joosten, F. A. el Maate, U. Dobrindt,
J. Hacker, R. M. J. Liskamp, A. S. Khan, R. J. Pieters, Tetrahe-
dron-Asymmetry 2005, 16, 361.
[68] D. D. Mitchell, J. C. Pickens, K. Korotkov, E. Fan, W. G. J. Hol,
Bioorg. Med. Chem. 2004, 12, 907.
[69] J. Geng, G. Mantovani, L. Tao, J. Nicolas, G. Chen, R. Wallis,
D. A. Mitchell, B. R. G. Johnson, S. D. Evans, D. M. Haddleton,
J. Am. Chem. Soc. 2007, 129, 15156.
[61] S. Yamaguchi, Y. Nishida, K. Sasaki, M. Kambara, C. L. Kim, N.
Ishiguro, T. Nagatsuka, H. Uzawa, M. Horiuchi, Biochem.
Biophys. Res. Commun. 2006, 349, 485.
[70] A. Brik, J. Alexandratos, Y. C. Lin, J. H. Elder, A. J. Olson, A.
Wlodawer, D. S. Goodsell, C. H. Wong, ChemBioChem 2005, 6,
1167.
[62] Z. S. Zhang, E. A. Merritt, M. Ahn, C. Roach, Z. Hou, C. Verlinde,
W. G. J. Hol, E. Fan, J. Am. Chem. Soc. 2002, 124, 12991.
[63] P. Rai, C. Padala, V. Poon, A. Saraph, S. Basha, S. Kate, K. Tao, J.
Mogridge, R. S. Kane, Nat. Biotechnol. 2006, 24, 582.
[64] D. Arosio, M. Fontanella, L. Baldini, L. Mauri, A. Bernardi, A.
Casnati, F. Sansone, R. Ungaro, J. Am. Chem. Soc. 2005, 127, 3660.
[65] J. A. Mertz, J. A. McCann, W. D. Picking, Biochem. Biophys. Res.
Commun. 1996, 226, 140.
[71] B. H. M. Kuijpers, S. Groothuys, A. R. Keereweer, P. J. L. M.
Quaedflieg, R. H. Blaauw, F. L. vanDelft, F. P. J. T. Rutjes, Org.
Lett. 2004, 6, 3123.
[72] N. L. Allinger, J Am Chem Soc 1977, 99, 8127.
[73] ‘‘CS Chem 3D Ultra’’, in: CambridgeSoft Corp, Cambridge,
MA 2001.
Macromol. Biosci. 2010, 10, 68–81
ß 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
www.mbs-journal.de
81