Dynamic combinatorial chemistry: Lysozyme selects an aromatic motif that mimics a carbohydrate residue

Article abstract of DOI:10.1002/anie.200462150

Lysozyme makes its choice: Hen egg-white lysozyme selects the optimal binder from a collection of aromatic motifs that mimic a carbohydrate unit upon exposure to a dynamic library of potential ligands that are directed towards its active site by a low-aff

Full text of DOI:10.1002/anie.200462150

Angewandte
Chemie
Combinatorial Chemistry
Dynamic Combinatorial Chemistry: Lysozyme
Selects an Aromatic Motif That Mimics a
Carbohydrate Residue**
Sandrine Zameo, Boris Vauzeilles,* and
Jean-Marie Beau*
Dedicated to Professor Julius Rebek, Jr.
on the occasion of his 60th birthday
In a series of recent developments, dynamic combinatorial
chemistry (DCC)[1] has proved its capacity to select potent
enzyme inhibitors.[2] The success of these preliminary vali-
dation studies was mainly based on the use of a previously
known high-affinity scaffold[2a,c,d,f] that could reversibly self-
assemble with several complementary building blocks. This
process results in the formation of a dynamic library of
potential ligands in which the target of interest, which acts as a
thermodynamic trap, can select its best binder. When the
experimental conditions are properly chosen,[3] binding leads
to an amplification of this particular compound which can be
detected by an appropriate analytical method such as HPLC.
In these early studies, the products that were selected bore
very close similarities, both in structure and affinity, to
previously known inhibitors of the target enzymes. We were
curious to challenge this process with an enzyme for which a
scaffold with relatively poor binding properties would hope-
fully direct members of the dynamic combinatorial library
(DCL) towards the active site. Such situations are frequently
found in glycobiology, an area of research that is rapidly
[*] S. Zameo, Dr. B. Vauzeilles, Prof. J.-M. Beau
Universitꢀ Paris-Sud
Laboratoire de Synthꢁse de Biomolꢀcules associꢀ au CNRS
Institut de Chimie Molꢀculaire et des Matꢀriaux
91405 Orsay Cedex (France)
Fax: (+33)1-6985-3715
E-mail: jmbeau@icmo.u-psud.fr
[**] We gratefully acknowledge the Conseil Gꢀnꢀral de l’Essonne for
financial support of this study and GlaxoSmithKline for their
interest.
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
Angew. Chem. Int. Ed. 2005, 44, 965 –969
DOI: 10.1002/anie.200462150
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
965

Communications
becoming fundamental in the post-genomic era.[4]
Indeed, many carbohydrate-binding proteins
including enzymes (termed Group II proteins)[5]
comprise shallow carbohydrate-binding sites that
result in relatively weak carbohydrate–protein
interactions with dissociation constants often in
the millimolar range. We therefore decided to
assess the capacity of DCC to probe such low
energy interactions and selected for an initial
proof-of-principle study a well-known and readily
available Group II glycosidase, namely hen egg-
white lysozyme (HEWL). Involved in peptidogly-
can degradation, HEWL is known to cleave N-
acetyl-glucosamine oligomers (chito-oligosac-
charides) into their smaller units, which, up to
the tetramer, behave as competitive inhibitors.[6]
The affinity of N-acetyl-d-glucosamine (d-
GlcNAc) for the HEWL-binding site lies in the
Scheme 2. Synthesis of carbohydrate scaffold A. a) PivCl, pyridine, CH2Cl2, 08C,
2 h (78%); b) 1) Tf2O, pyridine, CH2Cl2, À158C, 1 h; 2) NaNO2, DMF, 208C,
16 h; 3) NaHCO3, H2O, 208C, 5 min (64%); c) 1) Tf2O, pyridine, CH2Cl2, À158C,
1 h; 2) NaN3, Bu4NHSO4, DMF, 208C, 16 h (61%); d) MeONa, MeOH, 208C,
5 d (84%); e) H2, Pd/C (5%), MeOH, 208C, 4 h (90%). Piv=pivaloyl, Tf=tri-
fluoromethylsulfonate, DMF=N,N-dimethylformamide.
20–50 mm range, whereas chitobiose and chitotriose display
higher affinities.[7] On the basis of this knowledge, we
designed a DCL of potential HEWL binders starting with a
d-GlcNAc (or d-Glc) motif as a scaffold and by using the
generation of imines to introduce diversity (Scheme 1). This
reversible connecting system, which functions under mild
conditions, has previously been reported to be suitable for the
formation of DCLs.[2a,c,d,8]
Two amino-derived carbohydrate compounds (amines A
and B in the d-GlcNAc and d-Glc series, respectively) and six
differently substituted aromatic aldehydes (1–6) were
selected.[9] A 4-methylumbelliferyl chromophore was intro-
duced at the anomeric position of the carbohydrate building
blocks A and B to allow equal HPLC detection of the
different library products at a specific near-UV wavelength
(l = 322 nm). The aromatic motifs may “mimic”, after imine
formation, a second carbohydrate ring and potentially
interact with complementary aromatic residues of the binding
site. Stacking interactions between aromatic residues of the
amino acid side chains and the hydrophobic faces of
carbohydrates are a common feature of almost every
carbohydrate–protein complex studied until present.[5a,10]
Amine A was synthesized from N-acetyl-d-glucosamine
following the sequence presented in Scheme 2. The 4-
methylumbelliferyl glycoside 7 was obtained by Royꢀs proce-
dure[11] and by using Hortonꢀs chloride.[12] Regioselective
pivaloylation of hydroxyl groups at the 3 and 6 positions,[13]
followed by epimerization at C-4[14] furnished the d-galacto
derivative 8. Triflate formation and nucleophilic substitution
by sodium azide provided the azido intermediate 9, which
after deprotection of the hydroxyl groups and catalytic
hydrogenation afforded amine A. A similar sequence was
applied to the preparation of amine B from d-galactose (see
Supporting Information).
When a mixture of amines A and B (0.4 mm) were
equilibrated with aldehydes 1–6 (0.4 mm each) at room
temperature in an aqueous phosphate buffer solution
(pH 6.2) in the presence of sodium cyanoborohydride
(3.6 mm), slow formation of the amines through reduction
of the iminium species was observed. After 24 h, the twelve
expected products could be detected by reverse-phase HPLC
(Figure 1a). Equilibration of the same library in the presence
of HEWL (0.4 mm) induced a detectable change in the
distribution of the amines in the mixture with enrichment of
Scheme 1. Structures of the building blocks and components of the dynamic combinatorial library (DCL).
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
966
www.angewandte.org
Angew. Chem. Int. Ed. 2005, 44, 965 –969

Angewandte
Chemie
Figure 1. Selected portion of HPLC chromatograms of the DCL made
from amines A and B, and aldehydes 1–6 over 24 h: a) in the absence
of HEWL and b) in the presence of HEWL (1 equiv). (* indicates re-
sidual 4-methylumbelliferone).
amines A2 and, to a lesser extent, A6 (resulting from
reductive alkylation of amine A with aldehydes 2 and 6,
respectively; see Figure 1b). Although HPLC–MS analysis
easily confirmed the structures, amplification of the active
components becomes difficult to evaluate when a strong
overlap occurs (see, for instance, amines A1 and B1 in
Figure 1), and this difficulty increases with the size of the
library.[9,15] We therefore constructed and analyzed sub-
libraries with amines A and B. As expected from the previous
experiments, amplification of amines A2 and A6 was
observed when comparing libraries synthesized from A in
the presence (Figure 2b) and in the absence (Figure 2a) of
HEWL. No amplification was detected in the presence of
HEWL in the sublibrary that was based on amine B (see
Supporting Information). This clearly indicates that the
enzyme selects amine A as a scaffold rather than amine B.
When the reaction was performed at a higher concentration
of HEWL (1.2 mm), the amplification was significantly
enhanced (Figure 2c). Also, the addition of chitotriose, a
good HEWL inhibitor, led to a reduction (0.4 mm chitotriose,
not shown) or disappearance (1.2 mm chitotriose, Figure 2d)
of the amplification effect.
Figure 2. Selected portion of HPLC chromatograms showing the prod-
ucts of the DCL prepared from amine A and aldehydes 1–6 over 24 h:
a) in the absence of HEWL, b) in the presence of 1 equivalent of
HEWL, c) in the presence of 3 equivalents of HEWL, and d) in the
presence of 1 equivalent of HEWL and 3 equivalents of chitotriose.
(* indicates residual 4-methylumbelliferone).
suggests that the d-GlcNAc scaffold may occupy subsite À2
with the selected aromatic motif occupying subsite À3.
If this reasoning is correct, amines A2 or A6 may display
inhibitory activities which are comparable to that of chito-
biose.[19] To test this hypothesis, the most amplified compound
(amine A2, Scheme 3) was synthesized separately,[20] and its
All these simple experiments clearly indicate an amplifi-
cation of the library members selected through interaction
with the HEWL active site. Interaction of scaffold A with the
active site rather than scaffold B further indicates that the 2-
acetamido group is an essential feature for recognition of the
library members by HEWL.[7b,16] The active site of the enzyme
is a cleft composed of six carbohydrate-binding subsites (A to
F[17] or À4 to +2[18]). Subsites À4, À2, and +1 are d-GlcNAc-
binding sites which are sensitive to the acetamido substituent
by strict hydrogen-bonding and hydrophobic interactions,
whereas the three other sites are not.[17] As chitotriose is
known to tightly bind to the À4 to À2 subsites,[17] the extra
stabilization of the precursors of amines A2 and A6 above
Scheme 3. Amine A2, which results from reductive alkylation of amine
A with aldehyde 2.
inhibitory activity was measured and compared with the
chito-oligomers. We used the lysis rate of Micrococcus
lysodeikticus, a Gram-positive bacteria that is very sensitive
to HEWL,[21] as a measure of the enzyme activity. Line-
weaver–Burk analysis is presented in Figure 3. The inhibition
constants (Ki) measured for N-acetyl-d-glucosamine and
chitotriose were the same as those reported (not shown).[7]
The value found for chitobiose was in the range reported
(0.6 mm instead of 0.2 mm reported, see Supporting Informa-
Angew. Chem. Int. Ed. 2005, 44, 965 –969
www.angewandte.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
967

Communications
removed from these solutions and guanidine thiocyanate (4.7 mg), a
good denaturant of HEWL, was added prior to HPLC analysis to
ensure the release of any bound ligand from the active site. This
treatment was also applied in the control experiment without HEWL.
After 1 h, acetic acid (20 ml) was added, and the solution was diluted
with methanol (120 ml). Analytical chromatography was performed
on a JASCO LC-1500 system equipped with a Phenomenex LUNA
C18 (2) 5m reversed-phase HPLC column (150 ꢁ 4.60 mm) with UV/
Vis detection at l = 322 nm. A binary solvent gradient (solvent A:
0.1% trifluoroacetic acid in 95:5 H2O/CH3CN, solvent B: 0.1%
trifluoroacetic acid in 95:5 CH3CN/H2O) was optimized to separate
most of the DCL compounds: from 90:10 A/B to 70:30 A/B over
20 min, with a flow rate of 0.8 mLminÀ1. HPLC peaks were assigned
by means of LC–MS analysis. Some products were synthesized and
characterized separately to validate their assignments on the chro-
matograms.
Figure 3. Lineweaver–Burk analysis (double reciprocal plot) of the
observed initial lysis rate (Vi) as a function of the concentration of
Received: September 29, 2004
Published online: January 11, 2005
*
Micrococcus lysodeikticus ([S]) in the absence ( ) and in the presence
(ꢂ) of amine A2 (214 mM).
Keywords: carbohydrates · combinatorial chemistry ·
.
enzyme inhibitors · molecular recognition · reduction
tion). The inhibition assays carried out with amine A2
(Figure 3) yielded a competitive type of inhibition with a Ki
value that was similar to the value observed for chitobiose
(Ki = 0.6 mm).
[1] J.-M. Lehn, Chem. Eur. J. 1999, 5, 2455; I. Huc, R. Nguyen,
Comb. Chem. High Throughput Screening 2001, 4, 109; S. J.
Rowan, S. J. Cantrill, G. R. L. Cousins, J. K. M. Sanders, J. F.
Stoddart, Angew. Chem. 2002, 114, 938; Angew. Chem. Int. Ed.
2002, 41, 898.
These results indicate an additional binding effect from
the aromatic motifs, with amine A2 being around 100-fold
more active than the starting N-acetyl-d-glucosamine. With
the peptidoglycan substrate, HEWL subsite À3 hosts a
carbohydrate moiety (a N-acetylmuramic acid unit) which is
known to interact with complementary aromatic residues
through stacking interactions.[17] The Ki values, which are
similar for A2 and chitobiose, suggest that the enzyme selects
from the DCL an aromatic motif with a precise substitution
pattern that best fits in subsite À3.
[2] a) I. Huc, J.-M. Lehn, Proc. Natl. Acad. Sci. USA 1997, 94, 2106;
b) T. Bunyapaiboonsri, O. Ramstrꢂm, S. Lohman, J.-M. Lehn, L.
Peng, M. Goeldner, ChemBioChem 2001, 2, 438; c) M. Hoch-
gꢃrtel, H. Kroth, D. Piecha, M. W. Hofmann, C. Nicolau, S.
Krause, O. Schaaf, G. Sonnenmoser, A. V. Eliseev, Proc. Natl.
Acad. Sci. USA 2002, 99, 3382; d) M. Hochgꢃrtel, R. Biesinger,
H. Kroth, D. Piecha, M. W. Hofmann, S. Krause, O. Schaaf, C.
Nicolau, A. V. Eliseev, J. Med. Chem. 2003, 46, 356; e) T.
Bunyapaiboonsri, H. Ramstrꢂm, O. Ramstrꢂm, J. Haiech, J.-M.
Lehn, J. Med. Chem. 2003, 46, 5803; f) M. S. Congreve, D. J.
Davis, L. Devine, C. Granata, M. OꢀReilly, P. G. Wyatt, H. Jhoti,
Angew. Chem. 2003, 115, 4617; Angew. Chem. Int. Ed. 2003, 42,
4479.
[3] Z. Grote, R. Scopelliti, K. Severin, Angew. Chem. 2003, 115,
3951; Angew. Chem. Int. Ed. 2003, 42, 3821; P. T. Corbett, S. Otto,
J. K. M. Sanders, Chem. Eur. J. 2004, 10, 3139.
[4] R. U. Lemieux, Acc. Chem. Res. 1996, 29, 373; R. A. Dwek,
Chem. Rev. 1996, 96, 683; C. Bertozzi, L. L. Kiessling, Science
2001, 291, 2357; P. M. Rudd, R. A. Dwek, Crit. Rev. Biochem.
Mol. Biol. 1997, 32, 1; H. J. Gabius, Biochim. Biophys. Acta 2002,
1572, 165; H. Kogelberg, D. Solis, J. Jimꢄnez-Barbero, Curr.
Opin. Struct. Biol. 2003, 13, 646.
We have shown that hen egg-white lysozyme selects from
a dynamic library of potential active-site ligands an optimal
binder, which comprises an aryl group that mimics a
carbohydrate unit. In such a dynamic combinatorial assay,
the system has shown its capacity to discriminate subtle
affinity variations induced by the different substitutions on
the aromatic moiety of the aldehydes. Further structural
studies to confirm this hypothesis and experiments to develop
and extend this approach are currently in progress in our
group.
[5] a) F. A. Quiocho, N. K. Vyas in Bioorganic Chemistry: Carbo-
hydrates (Ed.: S. M. Hecht), Oxford University Press, Oxford,
1999, pp. 441 – 457; b) P. Sears, C.-H. Wong, Angew. Chem. 1999,
111, 2446; Angew. Chem. Int. Ed. 1999, 38, 2300.
[6] J. A. Rupley, Proc. R. Soc. London, Ser. B 1967, 167, 416.
[7] a) F. W. Dahlquist, L. Jao, M. Raftery, Proc. Natl. Acad. Sci. USA
1966, 63, 26; b) D. M. Chipman, N. Sharon, Science 1969, 165,
454.
[8] H. Li, P. Williams, J. Micklefield, J. M. Gardiner, G. Stephens,
Tetrahedron 2004, 60 , 753.
[9] For clarity, only six out of the thirty-six aromatic aldehydes that
were tested are shown in this work (see Supporting Informa-
tion).
Experimental Section
Preparation of a DCL from amines A and B and aldehydes 1–6: Stock
solutions of amines A (12 mm) and B (2.4 mm) were prepared in
distilled water. Solutions of aldehydes (0.6 mm) in phosphate buffer
(30 mm, pH 6.2) and sodium cyanoborohydride (36 mm) in distilled
water were prepared just prior to use. Solutions of amine A (16.7 ml),
amine B (83.5 ml), and the aldehydes (333 ml) were introduced into an
Eppendorf tube, which was equipped with a small magnetic stirrer
bar, and the volume was adjusted to 450 ml with distilled water. After
stirring, a 225-ml aliquot of the mixture was removed and introduced
into an Eppendorf tube that contained HEWL (1.5 mg, 1 equiv). Both
Eppendorf tubes were stirred for 30 min before solutions of sodium
cyanoborohydride (25 ml) were added. The resulting mixtures (250 ml)
were equilibrated at room temperature for 24 h. 20-ml aliquots were
[10] C.-H. Wong, Acc. Chem. Res. 1999, 32, 376.
[11] D. Carriꢅre, S. J. Meunier, F. D. Tropper, S. Cao, R. Roy, J. Mol.
Catal. A 2000, 154, 9.
968
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
Angew. Chem. Int. Ed. 2005, 44, 965 –969

Angewandte
Chemie
[12] D. Horton, Org. Synth. 1966, 46, 1.
[13] L. Lay, F. Nicotra, L. Panza, G. Russo, E. Adobati, Helv. Chim.
Acta 1994, 77, 509.
[14] F. Santoyo-Gonzalez, C. Uriel, J. A. Calvo-Asin, Synthesis 1998,
1787.
[15] Out of thirty-six aromatic aldehydes tested (see Supporting
Information), the product derived from aldehyde 2 was the most
amplified by the enzyme.
[16] A. Neuberger, B. M. Wilson, Nature 1967, 215, 524.
[17] C. C. F. Blake, D. F. Koenig, G. A. Mair, A. C. T. North, D. C.
Phillips, V. R. Sarma, Nature 1965, 206, 757; L. N. Johnson, D. C.
Phillips, Nature 1965, 206, 761; C. C. F. Blake, L. N. Johnson,
G. A. Mair, A. C. T. North, D. C. Phillips, V. R. Sarma, Proc. R.
Soc. London Ser. B 1967, 167, 378; K. Maenaka, M. Matsushima,
H. Song, F. Sunada, K. Watanabe, I. Kumagai, J. Mol. Biol. 1995,
247, 281.
[18] G. J. Davies, K. S. Wilson, B. Henrissat, Biochem. J. 1997, 321,
557; D. J. Vocadlo, G. J. Davies, R. Laine, S. G. Withers, Nature
2001, 412, 835.
[19] It should, however, be noted that the true binders are one of the
two hemiaminals, the aminal, the imine, or their protonated
species and that the corresponding reduced compound may have
much poorer binding properties.
[20] The imine was preformed in methanol (amine A, aldehyde 2,
MeOH, 3 ꢆ molecular sieves, 48 h) and then reduced (NaBH4,
16 h, 83%).
[21] D. Shugar, Biochim. Biophys. Acta 1952, 8, 302; A. L. N. Prasad,
G. Litwack, Anal. Biochem. 1963, 6, 328.
Angew. Chem. Int. Ed. 2005, 44, 965 –969
www.angewandte.org
ꢀ 2005 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
969