Ligand amplification in a dynamic combinatorial glycopeptide library

Article abstract of DOI:10.1039/b507674a

N-acetyl glucosamine binding protein amplifies the concentration of one member in a dynamic combinatorial glycopeptide library based on exchanging disulfides. The Royal Society of Chemistry 2005.

Full text of DOI:10.1039/b507674a

View Article Online / Journal Homepage / Table of Contents for this issue
COMMUNICATION
www.rsc.org/chemcomm | ChemComm
Ligand amplification in a dynamic combinatorial glycopeptide library{
Tom Hotchkiss, Holger B. Kramer, Katie J. Doores, David P. Gamblin, Neil J. Oldham and
Benjamin G. Davis*
Received (in Cambridge, UK) 2nd June 2005, Accepted 28th June 2005
First published as an Advance Article on the web 9th August 2005
DOI: 10.1039/b507674a
the ability to employ entirely unprotected sugars in a convergent
strategy allowed syntheses in overall yields of 12–13% from parent
amino acids and carbohydrates.
N-acetyl glucosamine binding protein amplifies the concentra-
tion of one member in a dynamic combinatorial glycopeptide
library based on exchanging disulfides.
The conditions for the formation of a dynamic combinatorial
library based on these well-defined starting glycopeptides were
monitored using electrospray ionization mass spectrometry
(ESI-MS) in positive ion mode and established as follows.
Glycopeptides containing D-glucose (Glc) and N-acetyl-D-
glucosamine (GlcNAc), GlyCys(Glc) C and Cys(GlcNAc)Thr G,
respectively, were each stirred in NH4OAc buffer (pH 7.5),
containing catalytic dithiothreitol (DTT, y0.2 equiv.) for efficient
disulfide exchange, for 48 h. Under equivalent conditions at pH 4.5
(4% formic acid) no disulfide exchange was observed after 48 h or
even after 14 days. Therefore, reaction mixtures were quenched
and disulfide exchange halted by addition of formic acid prior to
analysis. Simultaneous observation of the peaks of all members of
the resulting DCL in a single spectrum required calibration of cone
voltage and capillary voltage of the mass spectrometer.
Gratifyingly, starting from just C and G, nine out of ten possible
disulfide exchange products A–M were identified in the library
under the chosen conditions (Fig. 2).
Dynamic combinatorial chemistry represents an evolutionary
approach to molecular design and therefore has the potential to
offer a very efficient route to specific inhibitors, catalysts or
receptor ligands.1 In order to achieve full potential, a library of
building blocks should exchange efficiently to form its constituents
by reversible reactions. Introduction of a template to the
thermodynamic equilibrium mixture results in binding of suitable
ligands. Stabilization of the bound states leads to a shift in
equilibrium favouring the respective library members and results in
their amplification. Other members, which do not bind well but
contain common building blocks, are at the same time reduced in
concentration. Determination of the changes in composition of the
library requires a means of concurrently monitoring the concen-
trations of all library members.
In this communication we wish to report our efforts towards a
dynamic combinatorial glycopeptide library based on 1-thiosugars
and cysteine containing dipeptides. These studies are the first
results in a programme aimed at using DCL approaches to
influence and detect dynamic protein glycosylation (Fig. 1).{
Glycopeptides C and G (Fig. 2) were synthesized (Scheme 1)
starting from suitably protected amino acids using solution phase
peptide coupling methods, followed by reduction of the resulting
disulfides with Bu3P and formation of the disulfide-linked
glycopeptides using Glyco-SeS methodology.2 It should be noted
that the minimal use of protecting groups in peptide syntheses and
Having established suitable conditions for the development and
analysis of libraries, we turned our attention towards the
concentration dependence of the ion count obtained. While a
clear dependence of ion count on the concentration of glycopep-
tides C and G could be observed, these measurements do not
provide a reliable measure of absolute concentration of the
respective compound. Therefore the ratio of peak height of the
observed compound to the sum of peak heights of the remaining
members of the library was taken as a measure of relative
concentration. This internal calibration allows observation of
changes in relative proportions of compounds in the library to be
made. Time-dependant measurements of DCLs derived from
A–M at room temperature demonstrated that disulfide exchange,
in fact, led to formation of mixtures containing the expected
disulfide exchange products within 10 h at pH 7.5.
After successfully establishing formation and monitoring of the
glycopeptide libraries, their use as probes of binding was
investigated. The dynamic addition (so-called ‘GlcNAc-ylation’)
and removal of GlcNAc to and from Ser/Thr is a process of
emerging importance in regulating the activity of proteins ranging
from nucleoporins to transcription factors and implicated in
gibberellin signalling and diabetes.3 While elegant methods for
tracking the existence of GlcNAc-modified amino acids at a given
point in time in proteins have been described4 to the best of our
knowledge no models of the dynamic addition and removal have
been proposed. With the longer term aim of creating and
understanding this process ultimately in proteins, we first wished
Fig. 1 Dynamic Protein GlcNAc-ylation may be considered to be an
example of a natural DCL.
Department of Chemistry, University of Oxford, Mansfield Road,
Oxford, UK OX1 3TA. E-mail: Ben.Davis@chem.ox.ac.uk;
Fax: +44 (0)1865 285002; Tel: +44 (0)1865 275652
{ Electronic supplementary information (ESI) available: Experimental
details. See http://dx.doi.org/10.1039/b507674a
4264 | Chem. Commun., 2005, 4264–4266
This journal is ß The Royal Society of Chemistry 2005

View Article Online
n
Scheme 1 (i) EDDQ, Et3N, DCM, 3: 84%, 7: 60%, (ii) Bu3P, THF–MeOH, 4: 95%, 8: 99%, (iii) Et3N, MeOH, C: 43%, G: 91%.
Fig. 2 Glycopeptide DCL members and their observation by ESI-MS.
to create a chemical analogue relying on the reversibility of
disulfide bond formation and that could be interrogated by one of
the same sugar-binding proteins used to detect natural GlcNAc-
ylation, wheat germ agglutinin (WGA).5 WGA (14.7 nmol) was
added to glycopeptide DCLs (y206 and 168 nmol of starting
components C and G respectively) containing both dynamically
glucosylated and GlcNAc-ylated peptides that had been allowed to
exchange for 78 h; significant changes of the relative concentra-
tions of select members of the library were observed. Specifically,
an increase in the concentration of GlcNAc-ylated glycopeptide B,
and a less significant increase in the concentration of GlcNAc-
ylated glycopeptide G were observed, while at the same time the
glucosylated glycopeptide F showed a significant decrease in
concentration (Fig. 3).
Fig. 3 Change in the relative concentration of B, F and G on addition of
carbohydrate specificity of WGA.6 The observed amplification of
WGA.
This is an important result which is coherent with the known
B can be understood in terms of relative stabilisation of the bound
species and consequent alteration in the reequilibration of the
reaction mixture. Upon quenching of the samples prior to
electrospray mass spectrometric determination, the lectin–
glycopeptide complex is expected to break down and the
determined concentration of the selected glycopeptide is larger
than the corresponding system prior to WGA addition.
Interestingly, the much stronger amplification of B, rather than
the other GlcNAc-containing species in the DCL, suggests that
this glycopeptide with a Gly–Cys backbone is bound more
strongly than GlcNAc-ylated glycopeptide G with a Cys–Thr
This journal is ß The Royal Society of Chemistry 2005
Chem. Commun., 2005, 4264–4266 | 4265

View Article Online
approach to the investigation of binding preferences in lectins8
and in detection of peptide glycosylation.
In conclusion, we have established conditions for the successful
formation of a glycopeptide library based on disulfide exchange of
1-thiosugars with cysteine containing peptides and monitoring by
ESI-MS. Selective amplification of a glycopeptide component by
lectin was observed for the first time. The lectin–ligand complex
was identified directly by native mass spectrometry. The potential
of this method is currently being used to interrogate chemically
and naturally GlcNAc-ylated protein systems.
Notes and references
{ First details of this work were described at the GlycoTrain network
Seville meeting, 12 December 2003. For an example of mixed disulfide
library formation that shows component binding but not amplification or
selection see: S. Sando, A. Narita and Y. Aoyama, Bioorg. Med. Chem.
Lett., 2004, 14, 2835. Unfortunately, in this prior work none of the
individual members were individually synthesised; in our hands this
mixture-based approach carries ambiguity in results.
Fig. 4 ESI-MS analysis of WGA lectin following exposure to the DCL.
* 5 WGA dimer; % 5 WGA dimer + glycopeptide B; X 5 WGA dimer +
2 6 glycopeptide B. Peak multiplicity is due to the heterogeneous nature
of the two WGA chains.
backbone or even 1,1-linked diGlcNAc species I. This illustrates
that WGA is not only GlcNAc specific but in the context of
GlcNAc-ylation the peptide backbone may have a strong
additional influence.
1 G. R. L. Cousins, S.-A. Poulsen and J. K. M. Sander, Curr. Opin. Chem.
Biol., 2000, 4, 270.
2 D. P. Gamblin, P. Garnier, S. van Kasteren, N. J. Oldham,
A. J. Fairbanks and B. G. Davis, Angew. Chem., Int. Ed., 2004, 43, 828.
3 (a) K. Vosseller, K. Sakabe, L. Wells and G. W. Hart, Curr. Opin. Chem.
Biol., 2002, 6, 851–857; (b) G. Hart, K. Greis, L. Dong, M. Blomberg,
T. Chou, M. Jiang, E. Roquemore, D. Snow, L. Kreppel and R. Cole,
Adv. Exp. Med. Biol., 1995, 376, 115; (c) W. Lubas and J. Hanover,
J. Biol. Chem., 2000, 275, 10983.
4 (a) D. J. Vocadlo, H. C. Hang, E.-J. Kim, J. A. Hanover and
C. R. Bertozzi, Proc. Natl. Acad. Sci. USA, 2003, 100, 9116; (b)
N. Khidekel, S. Arndt, N. Lamarre-Vincent, A. Lippert, K. G. Poulin-
Kerstien, B. Ramakrishnan, P. K. Qasba and L. Hsieh-Wilson, J. Am.
Chem. Soc., 2003, 125, 16162–16163.
5 C. S. Arnold, G. V. W. Johnson, R. N. Cole, D. L.-Y. Dong, M. Lee and
G. W. Hart, J. Biol. Chem., 1996, 271, 28741.
6 M. M. Burger and A. R. Goldberg, Proc. Natl. Acad. Sci. USA, 1967, 57,
359; A. K. Allen, A. Neuberger and N. Sharon, Biochem. J., 1973, 131,
155; I. J. Goldstein, S. Hammarstrom and G. Sundblad, Biochim.
Biophys. Acta, 1975, 405, 63.
We sought to confirm directly the selection and binding of
GlcNAc-ylated glycopeptide B by WGA. The lectin-substrate
complexes (WGA)2 + B and (WGA)2 + B2 were directly observed
by native mass spectrometry7 at approximately 526 ¡ 2 Da higher
than the free lectin (Fig. 4). This striking result highlights the
strong potential of native MS in DCL and GlcNAc-ylation
investigation, confirms the observed amplification of glycopeptide
B (Mr 527) and highlights the multiplicity of binding of WGA that
further enhances its role as an interrogating protein.
While the detailed reasons for the clear selection of glycopeptide
B in the presence of other GlcNAc-ylated ligand possibilities
remains to be explored, the success of the developed methodology
opens clear possibilities to explore further not only GlcNAc-
ylation processes using such techniques but the influence of the
peptide backbone in glycopeptides on binding to other lectins. Our
results confirm the usefulness of a dynamic combinatorial
7 R. H. H. van den Heuvel and A. J. R. Heck, Curr. Opin. Chem. Biol.,
2004, 8, 519.
8 For a leading example of lectin binding in DCLs see O. Ramstrom and
J. M. Lehn, ChemBioChem, 2000, 1, 41.
4266 | Chem. Commun., 2005, 4264–4266
This journal is ß The Royal Society of Chemistry 2005