From the ganglioside GQ1bα to glycomimetic antagonists of the myelin- associated glycoprotein (MAG)

Article abstract of DOI:10.2533/chimia.2010.17

The tetrasaccharide 4, a substructure of ganglioside GQ1bα, shows a remarkable affinity for the myelinassociated glycoprotein (MAG) and was therefore selected as starting point for a lead optimization program. In our search for structurally simplified and pharmacokinetically improved mimics of 4, antagonists with modifications of the core disaccharide Galβ (1-3)GalNAc, as well as the terminal α (2-3)- and the internal α (2-6)-linked neuraminic acid were synthesized and tested in target-based binding assays. Compared to the reference tetrasaccharide 4, the most potent antagonist 17 exhibits a 360-fold improved affinity. Furthermore, pharmacokinetic parameters such as stability in the cerebrospinal fluid, logD and permeation through the BBB indicate the drug-like properties of antagonist 17. Schweizerische Chemische Gesellschaft.

Full text of DOI:10.2533/chimia.2010.17

From ChemiCal researCh to industrial appliCationsꢀ
CHIMIAꢀ2010,ꢀ64,ꢀNo.ꢀ1/2ꢀ 17
doi:10.2533/chimia.2010.17ꢀꢀ
Chimiaꢀ64ꢀ(2010)ꢀ17–22ꢀ ©ꢀSchweizerischeꢀChemischeꢀGesellschaft
From the Ganglioside GQ1bα to
Glycomimetic Antagonists of the
Myelin-Associated Glycoprotein (MAG)
BeatꢀErnst*,ꢀOliverꢀSchwardt,ꢀStefanieꢀMesch,ꢀMatthiasꢀWittwer,ꢀGianlucaꢀRossato,ꢀandꢀꢀ
AngeloꢀVedani
DedicatedꢀtoꢀProfessorꢀDanielꢀBellušꢀonꢀtheꢀoccasionꢀofꢀhisꢀ70thꢀbirthday
Abstract:ꢀTheꢀtetrasaccharideꢀ4,ꢀaꢀsubstructureꢀofꢀgangliosideꢀGQ1bα,ꢀshowsꢀaꢀremarkableꢀaffinityꢀforꢀtheꢀmyelin-
associatedꢀglycoproteinꢀ(MAG)ꢀandꢀwasꢀthereforeꢀselectedꢀasꢀstartingꢀpointꢀforꢀaꢀleadꢀoptimizationꢀprogram.ꢀInꢀourꢀ
searchꢀforꢀstructurallyꢀsimplifiedꢀandꢀpharmacokineticallyꢀimprovedꢀmimicsꢀofꢀ4,ꢀantagonistsꢀwithꢀmodificationsꢀofꢀ
theꢀcoreꢀdisaccharideꢀGalb(1-3)GalNAc,ꢀasꢀwellꢀasꢀtheꢀterminalꢀα(2-3)-ꢀandꢀtheꢀinternalꢀα(2-6)-linkedꢀneuraminicꢀ
acidꢀwereꢀsynthesizedꢀandꢀtestedꢀinꢀtarget-basedꢀbindingꢀassays.ꢀComparedꢀtoꢀtheꢀreferenceꢀtetrasaccharideꢀ4,ꢀ
theꢀmostꢀpotentꢀantagonistꢀ17ꢀexhibitsꢀaꢀ360-foldꢀimprovedꢀaffinity.ꢀFurthermore,ꢀpharmacokineticꢀparametersꢀ
suchꢀasꢀstabilityꢀinꢀtheꢀcerebrospinalꢀfluid,ꢀlogDꢀandꢀpermeationꢀthroughꢀtheꢀBBBꢀindicateꢀtheꢀdrug-likeꢀproper-
tiesꢀofꢀantagonistꢀ17.
Keywords:ꢀDockingꢀ·ꢀHaptenꢀinhibitionꢀassayꢀ·ꢀHomologyꢀmodelꢀ·ꢀGangliosidesꢀ·
Myelin-associatedꢀglycoproteinꢀ(MAG)ꢀ·ꢀMAGꢀantagonistsꢀ·ꢀSurfaceꢀplasmonꢀresonanceꢀ(Biacore)
Introduction
as its major axonal ligands.[18] Therefore, (Fig. 3).[24] In addition to the therefore
blocking MAG with potent glycomimetic required improvement of affinity, phar-
The injured adult mammalian central ner- antagonists may be a valuable therapeutic macokinetic issues as metabolic stability,
vous system (CNS) lacks the ability for approach to enhance axon regeneration.
e.g. sialidase stability[26] or permeation of
axon regeneration,[1,2] predominantly due Schnaar and coworkers[19] reported that the blood brain barrier have also to be ad-
to specific inhibitors expressed on residual a limited set of structurally related ganglio- dressed. For the in vivo application, it is
myelin and on astrocytes recruited to the sides like GT1b or GQ1bα (Fig. 2), known planned to add the antagonist by infusion
site of injury.[3–7] Several inhibitor proteins to be expressed on myelinated neurons in to the site of injury. Therefore, a prolonged
have been identified, one of them being the vivo, are functional ligands for MAG. Re- stability in the cerebrospinal fluid is also
myelin-associated glycoprotein (MAG).[8] cently, the MAG-affinity of a partial struc- required. Furthermore, to maintain the
MAG is a transmembrane glycoprotein[9] ture of GQ1bα, the tetrasaccharide 2, could necessary minimal therapeutic concentra-
belonging to the so-called Siglecs, a family clearly be correlated with its ability to re- tion in the CNS, a loss of the antagonist by
of the sialic acid-binding immunoglobulin verse MAG-mediated inhibition of axonal an active or passive transport mechanism
like lectins.[10,11] On the surface of neurons, outgrowth.[22] Since SAR studies indicate would be detrimental.
MAG interacts with two classes of targets: that not only the terminal, α(2-3)-linked,
In a first approach, we focused on a
Proteins of the family of Nogo receptors but also the internal, α(2-6)-linked sialic reduction of the structural complexity
(NgR)[12,13] and brain gangliosides (GD1a acid is essential for MAG binding, various of GQ1bα (1) and, at the same time, an
and GT1b)[11,14–16] (Fig. 1). Although the partial structures of 1[20] as well as sulfated improvement of pharmacodynamic and
relative roles of gangliosides and NgRs as analogs, e.g. 3[21] were synthesized.
MAG ligands have yet to be resolved,[8,17]
in some systems, MAG inhibition is com-
pletelyreversedbysialidasetreatment,sug- Design of Glycomimetics
gesting that MAG uses sialylated glycans
pharmacokinetic properties. From vari-
ous structure-affinity relationship studies
(SAR),[27,28] the tetrasaccharide 4 was
identified as the minimal carbohydrate
epitope. Detailed binding information
High-affinity MAG antagonists with of epitope 4 was obtained by STD NMR
concurrent drug-like pharmacokinetic experiments[24] (Fig. 3). They indicated
properties would provide a valuable tool important lipophilic interactions of the
for the investigation of the exact physio- glycerol side chain of the α(2-3)-linked
logical role of MAG in the inhibitory N-acetyl neuraminic acid (Neu5Ac), the
cascade leading to the collapse of growth b-face of the galactose moiety and the N-
cones, the reason for the failure of regener- acetates of both Neu5Ac residues. In addi-
ation of injuries in the CNS. Because of the tion, the carboxylates of the two Neu5Ac
shallow binding site typically present in moieties are involved in salt bridges and
lectins, carbohydrate ligands often exhibit the C(9)-OH of the α(2-3)-linked Neu5Ac
*Correspondence: Prof. Dr. B. Ernst
Institut für Molekulare Pharmazie
Pharmazentrum der Universität Basel
Klingelbergstrasse 50
CH-4056 Basel
Tel.: + 41 61 267 15 51
only modest, i.e. milli- to micromolar af- is forming a relevant hydrogen bond.[25]
A
finities.[23] This also proved true for MAG verification of these findings by docking
with a 180 micromolar affinity for tetrasac- studies to a homology model of MAG[29,30]
charide 4, the binding epitope of GQ1bα revealed the corresponding amino acids
Fax: + 41 61 267 15 52
E-mail: beat.ernst@unibas.ch

18ꢀ CHIMIAꢀ2010,ꢀ64,ꢀNo.ꢀ1/2ꢀ
From ChemiCal researCh to industrial appliCations
Fig.ꢀ1.ꢀMyelin-associatedꢀglycoproteinꢀ(MAG),ꢀNogoꢀ66ꢀandꢀ
oligodendrocyteꢀmyelinꢀglycoproteinꢀ(OMgp)ꢀbindꢀtoꢀtheꢀNogoꢀreceptorꢀ
(NgR).ꢀViaꢀtheꢀneurotrophinꢀreceptorꢀp75NTR,ꢀtheꢀinhibitoryꢀsignalꢀisꢀ
transducedꢀintoꢀtheꢀcytosolꢀofꢀtheꢀneuron.ꢀMAGꢀalsoꢀbindsꢀtoꢀtheꢀ
gangliosidesꢀGD1aꢀandꢀGT1b.ꢀAgain,ꢀwithꢀco-receptorꢀp75NTR,ꢀtheꢀ
inhibitoryꢀsignalꢀisꢀtransducedꢀintoꢀtheꢀcytosol.ꢀIntracellularly,ꢀtheꢀsmallꢀ
GTPaseꢀRhoAꢀisꢀactivated,ꢀwhichꢀleadsꢀtoꢀaꢀcollapseꢀofꢀtheꢀgrowthꢀ
conesꢀofꢀtheꢀinjuredꢀaxonꢀ(adaptedꢀfromꢀFilbinꢀet al.[5]).
Fig.ꢀ3.ꢀBindingꢀepitopeꢀofꢀtetrasaccharideꢀ4ꢀasꢀdeterminedꢀbyꢀSTDꢀ
NMR.[24]ꢀBesidesꢀtwoꢀsaltꢀbridgesꢀbyꢀtheꢀcarboxylatesꢀofꢀtheꢀNeu5Acꢀ
moietiesꢀandꢀanꢀimportantꢀhydrogenꢀbondꢀdonatedꢀbyꢀtheꢀC(9)-OHꢀ
ofꢀtheꢀα(2-3)-linkedꢀNeu5Ac,[25]ꢀtheꢀbindingꢀepitopeꢀinvolvesꢀlipophilicꢀ
interactionsꢀofꢀtheꢀglycerolꢀsideꢀchainꢀofꢀtheꢀα(2-3)-linkedꢀNeu5Ac,ꢀ
theꢀb-faceꢀofꢀtheꢀgalactoseꢀmoietyꢀandꢀtheꢀN-acetatesꢀofꢀtheꢀNeu5Acꢀ
residues.
forming the binding site (Figs 3 and 6). Neu5Ac moieties in the appropriate spa- four-fold reduction of affinity compared
Based on this information, a rational ap- tial orientation (Fig. 4). In addition, the to tetrasaccharide 4[36] (Table 1). For an
proach for the design of MAG antagonists biphenyl linker enables a lipophilic con- additional structural simplification, the
was envisaged.
tact with the binding site and at the same α(2-6)-linked Neu5Ac moiety was re-
time reduces the high polarity of the lead placed by acetic acid leading to antagonist
structure 4. Starting from glycosyl donor 10, which showed a slightly lower affin-
6,[37] the building blocks 7 and 8 were ity than 5. To further fine-tune the spatial
Replacement of the
Galb(1-3)GalNAc Core
In a first approach, the Galb(1-3) synthesized, permitting the formation of orientation of the carboxylates, the biphe-
GalNAc core, establishing a lipophilic the protected test compound 9 by Suzuki nyl linker was replaced by a 1,2,3-triazol-
contact with MAG,[24] was replaced by coupling in an excellent yield (Scheme 1). 4-yl-phenyl moiety, a modification with
biphenyl (5), which acts as a linker Mimic 5 was obtained after deprotection practically no influence on the affinity for
to position the carboxylates of the two under Zemplén conditions and showed a MAG.[38]
Lipophilic Substituents on the
HO
OH
α(2-3)-linked Sialic Acid
C13H27
OH
OH
A pivotal simplification of the tetrasac-
charide lead structure 4 was reported by
Kelm and Brossmer who modified the α(2-
3)-linked sialic acid in the 2-, 5- or 9-posi-
tion to obtain up to a ten-fold enhancement
of affinity compared to lead 4.[28,39,40] Fur-
ther optimization of these three positions
led to antagonist 17 (Scheme 2) with a
360-fold improved affinity, i.e. 500 nano-
molar.[25,41]
Starting from the Boc-protected neur-
aminic acid derivative 12,[41] 14 was ob-
tained by deprotection with TMSCl and
PhOH (13) followed by acylation with
fluoroacetyl chloride. Glycosylation using
2,3-difluorobenzyl alcohol (15), ami-
dation using modified Staudinger condi-
tions[42] (16) and final deprotection gave
the test compound 17.
OH
OH
HO
AcHN
O
AcHN
HO
C16H34
HN
O
CO2
O
H
O
HO
OH
O
HO
O
CO2
H
HO
O
OH
O
OH
OH
HO
OH
O
O
O
AcNH
OH
HO
HO
O
O
O
O
HO
OH
O
O
OH
NHAc
OH
AcHN
HO
CO2
H
O
O
O
O
OH
OH
OH
OH
AcHN
HO
CO2
H
O
Me
HO
O
CO2
H
AcHN
2[20]
OH
AcHN
COOMe
1, GQ1bα[19]
OH
O
OSO3H
OH
HO
HO
HO
HO
AcHN
O
CO2
H
O
AcHN
O
O
O
H
O
OH
O
OH
H
O
OH
OH
OH
OH
OH
OH
HO
AcHN
HO3SO
CO2
O
O
O
C13H27
O
OH
3[21]
HN
C17H35
OH
OH
O
Fig.ꢀ2.ꢀGQ1bαꢀ(1)[19]ꢀandꢀpartialꢀstructuresꢀthereof.[20,21]ꢀSinceꢀtheꢀreportedꢀaffinityꢀdataꢀwereꢀ
obtainedꢀinꢀdifferentꢀassayꢀformats,ꢀtheyꢀshouldꢀbeꢀcomparedꢀwithꢀcaution.ꢀForꢀbindingꢀofꢀMAG-
transfectedꢀCOSꢀcellsꢀtoꢀimmobilizedꢀgangliosideꢀ1ꢀorꢀtoꢀtheꢀdisulfatedꢀGM1bꢀanalogꢀ3,ꢀapparentꢀ
KDsꢀofꢀ38ꢀandꢀ7.4ꢀpmol/wellꢀwereꢀreported.[21]ꢀInꢀaꢀcompetitiveꢀbindingꢀassayꢀwithꢀtetrasaccharideꢀ
2,ꢀanꢀIC50ꢀofꢀ300ꢀnMꢀwasꢀdetermined.[20]

From ChemiCal researCh to industrial appliCationsꢀ
CHIMIAꢀ2010,ꢀ64,ꢀNo.ꢀ1/2ꢀ 19
Fig.ꢀ4.ꢀReplacementꢀofꢀtheꢀGalb(1-3)GalNAcꢀ
coreꢀofꢀ4ꢀbyꢀaꢀbiphenylꢀ(5)ꢀleadsꢀtoꢀonlyꢀ
marginalꢀchangesꢀofꢀtheꢀrequiredꢀspatialꢀ
orientationꢀofꢀtheꢀcarboxylateꢀofꢀtheꢀα(2-
6)-linkedꢀNeu5Ac,ꢀbutꢀresultsꢀinꢀaꢀfour-foldꢀ
reductionꢀinꢀaffinity.[36]
5[36]
Schemeꢀ1.ꢀa)ꢀ4-Bromophenol,ꢀBnNEt3Br,ꢀ
aq.ꢀNaOH/DCM,ꢀ40ꢀ°C,ꢀ2.5ꢀh,ꢀ56%;ꢀb)ꢀ
bis(picanolato)diborane,ꢀKOAc,ꢀPdCl2(dppf),ꢀ
dppf,ꢀdioxane,ꢀMWꢀ120ꢀ°C,ꢀ45ꢀmin,ꢀ85%;ꢀc)ꢀ
3-iodophenol,ꢀBnNEt3Br,ꢀaq.ꢀNaOH/DCM,ꢀ40ꢀ
°C,ꢀ2.5ꢀh,ꢀ46%;ꢀd)ꢀAg2CO3,ꢀPd(PPh3)4,ꢀdioxane,ꢀ
MWꢀ120ꢀ°C,ꢀ7ꢀh,ꢀ77%;ꢀe)ꢀi.ꢀNaOMe,ꢀMeOH,ꢀr.t.,ꢀ
17ꢀh,ꢀii.ꢀaq.ꢀNaOH,ꢀr.t.,ꢀ6ꢀh,ꢀiii.ꢀDowexꢀ50ꢀ×ꢀ8ꢀ
(Na+),ꢀ56%.
OR1
AcO
OR1
OR1
O
AcHN
O
COOMe
7
B
O
AcHN
R1
O
OAc
OAc
AcO
a,b)
CO2R2
O
O
AcO
OAc
O
R1
O
d)
AcHN
O
+
COOMe
AcHN
O
O
OAc
AcO
Cl
c)
R1
O
OH
OAc
R2OOC
OAc
OAc
OR1
6
AcHN
R1
9, R1=Ac, R2=Me
[36], R1=H, R2=Na
e)
HO
O
O
CO2Me
5
O
AcHN
O
O
COONa
OH
I
HO
HO
NaOOC
OH
10[38]
8
HO
COONa
AcHN
OH
O
O
OH
N
N
NaOOC
11[38]
N
Lipophilic and Hydrophilic
assay formats were applied; a fluorescent
For the KD determination in the Bi-
Replacements of the α(2-6)-linked
hapten binding assay[43] and a surface acore assay, MAG -Fc could not be im-
Sialic Acid
plasmon resonance based biosensor (Bi- mobilized by amdi1n-3e coupling, because
Since the carboxylate of the α(2-6)- acore) experiment[30,41] (Fig. 5). For the three lysines are positioned in proximity
linked Neu5Ac in tetrasaccharide 4 forms hapten inhibition assay, a recombinant to the carbohydrate binding site (Fig. 5B).
a salt bridge with Lys67 and the N-acetate protein consisting of the three N-terminal Therefore, MAGd1-3-Fc was immobilized
a lipophilic contact with Tyr69 (Figs 3 and domains of MAG and the Fc part of hu- on a dextran chip containing a surface of
6A), hydrophilic as well as lipophilic sub- man IgG (MAGd1-3-Fc) was produced by covalently bound protein A. A reference
stitutes were explored. A replacement by expression in CHO cells and affinity puri- cell providing only protein A was used to
lactic acid (22) or biphenylmethyl (23) fication on proteinA-agarose[43] (Fig. 5A). compensate unspecific binding to the ma-
yielded affinities in the range of tetrasaccha- The relative inhibitory concentrations trix (Figs 5C and 5D).
ride 4 (Scheme 3).Combined with the most (rIC ) of the test compounds as competi-
successful modification of the 9-position tive 5l0igands were determined in microti- Stability in Cerebrospinal Fluid
of the α(2-3)-linked Neu5Ac (Scheme 2) ter plates coated with fetuin as binding
For nerve regeneration, MAG antago-
antagonist 29 with low micromolar affinity target for MAGd1-3-Fc. By complexing nists will most likely be applied to the CNS
could be identified (Scheme 4).[30]
the Fc-part with alkaline phosphatase- by a local infusion. We therefore tested the
labeled anti-Fc antibodies and measuring stability of the fluoroacetate 17 in artifi-
the initial velocity of fluorescein release cial cerebrospinal fluid (aCSF)[44] for 19 h
from fluorescein diphosphate, the amount at 37 °C and, as a control, in buffer solu-
of bound MAGd1-3-Fc can be determined. tion. According to LC-MS analysis, more
The affinities were measured relative to than 95% of the initial concentrations of
Biological Evaluation
Determination of Affinity for MAG
For the evaluation of the binding prop- the reference compound 4 (rIC50 of 1, 17 were recovered from both media, pre-
erties of these new MAG antagonists two Table 1).
dicting a high stability in the CNS, the

20ꢀ CHIMIAꢀ2010,ꢀ64,ꢀNo.ꢀ1/2ꢀ
From ChemiCal researCh to industrial appliCations
Tableꢀ1.ꢀRelativeꢀinhibitoryꢀconcentrationsꢀ(rIC50)ꢀofꢀMAGꢀantagonists,ꢀKDꢀ
valuesꢀofꢀcompoundsꢀ4,ꢀ17ꢀand 29.
OBz
O
AcO
AcO
OAc
Ph
AcHN
O
O
SEt
O
OAc
a
OAc
AcO
CO2Me
Entry Compound
rIC50
KDꢀ[µM]
180b
AcO
AcO
O
OAc
20
a,b)
Ref [30]
O
O
OH
OH
HO
HO
AcHN
18
19
CO2Na
O
HO
OH
O
1
1.00
HO
O
HO
OBz
O
OH
AcO
OH
O
HO
AcO
O
O
AcHN
O
O
AcHN
O
OMe
O
O
OH
NHAc
OH
HO
CO2Na
OAc
OAc
OAc
AcO
OH
CO2Me
4
21
OH
HO
OH
AcHN
c,e)
d,e)
HO
O
CO2Na
n.d.c
O
2
2.39
HO
CO2Na
H3
C
AcHN
O
O
OH
OH
O
OH
O
HO
HO
HO
O
NaO2
C
HO
HO
HO
OH
OH
HO
OH
O
O
5
O
O
AcHN
AcHN
O
O
O
O
O
OH
OH
HO
HO
HO
CO2Na
CO2Na
OH
OH
AcHN
O
O
22[30]
23[30]
n.d.c
n.d.c
CO2Na
OH
3
4
HO
4.00
3.64
NaO2
C
OH
10
HO
Schemeꢀ3.ꢀa)ꢀDMTST,ꢀMSꢀ3Å,ꢀDCM,ꢀ0ꢀ°C,ꢀ2ꢀd,ꢀ79%;ꢀb)ꢀ80%ꢀaq.ꢀAcOH,ꢀ
60ꢀ°C,ꢀ3ꢀh,ꢀ89%;ꢀc)ꢀi.ꢀBu2SnO,ꢀbenzene,ꢀ80ꢀ°C,ꢀ3ꢀh,ꢀii.ꢀethylꢀ(S)-2-O-
(trifluoromethylsulfonyl)-lactate,ꢀCsF,ꢀDME,ꢀr.t.,ꢀ1ꢀd,ꢀ85%;ꢀd)ꢀi.ꢀbiphenyl-
4-carbaldehyde,ꢀcat.ꢀp-TsOH,ꢀMeCN,ꢀ70ꢀ°C,ꢀ4ꢀh,ꢀii.ꢀMe3N⋅BH3,ꢀAlCl3,ꢀ
THF,ꢀr.t.,ꢀ3ꢀh,ꢀ51%;ꢀe)ꢀNaOMe/MeOH,ꢀr.t.,ꢀ1ꢀd,ꢀthenꢀaq.ꢀNaOH,ꢀrt,ꢀ7ꢀh,ꢀ
22:ꢀ68%,ꢀ23:ꢀ96%.
CO2Na
AcHN
O
O
OH
HO
N
NaO2
C
OH
N
N
11
Cl
H
N
OH
F
CO2Na
O
OH
F
0.5d
O
H
N
5
6
0.0002
1.02
O
O
HO
CH2
F
17
CO2Na
H3
C
OH
O
HO
O
HO
O
HO
AcO
O
O
AcHN
n.d.c
AcHN
O
O
SMe
OAc
O
O
OAc
O
O
OH
OAc
OH
HO
HO
AcO
O
RO
AcO
O
N3
HO
CO2Me
CO2Na
22
AcO
OAc
OAc
OH
25
a)
c)
O
O
AcHN
O
O
OAc
OAc
N3
CO2Me
OAc
24[30]
26 (R = H)
27 (R = Ac)
b)
OH
O
O
HO
HO
HO
OH
7
8
0.80
n.d.c
O
AcHN
O
O
O
OH
HO
OAc
O
O
OH
O
AcO
AcO
HO
HO
O
CO2Na
OH
AcO
HO
OH
d)
Cl
O
O
O
AcHN
AcHN
O
23
O
O
O
O
O
O
OAc
OAc
OAc
OH
N
H
N
H
CO2Me
CO2Na
OH
29[30]
28
Cl
OH
O
HO
HO
O
HO
2.83d
O
O
AcHN
0.0027
O
O
O
OH
OH
N
H
CO2Na
Schemeꢀ4.ꢀa)ꢀNIS/TfOH,ꢀMeCN,ꢀ–40ꢀ°C,ꢀ61%;ꢀb)ꢀAc2O,ꢀpyr,ꢀ80%;ꢀc)ꢀ
p-Cl-BzCl,ꢀPPh3,ꢀDCE,ꢀr.t.,ꢀ16ꢀh,ꢀ75%;ꢀd)ꢀNaOMe/MeOH,ꢀr.t.,ꢀ3ꢀh,ꢀthenꢀ
OH
29
Cl
aq.ꢀNaOH,ꢀr.t.,ꢀ16ꢀh,ꢀ75%.
arIC50ꢀisꢀtheꢀconcentrationꢀwhenꢀ50%ꢀofꢀtheꢀproteinꢀareꢀ
inhibited,ꢀmeasuredꢀrelativeꢀtoꢀreferenceꢀcompoundꢀ4;ꢀ
bdeterminedꢀbyꢀSTDꢀNMR;ꢀcn.d.ꢀnotꢀdetermined;ꢀddeterminedꢀ
byꢀsurfaceꢀplasmonꢀresonanceꢀ(Biacore.)
Schemeꢀ2.ꢀa)ꢀ4ꢀMꢀPhOH,ꢀ4ꢀMꢀTMSClꢀinꢀDCM,ꢀ
70%;ꢀb)ꢀFAcCl,ꢀNEt3,ꢀDMAP,ꢀDCM,ꢀrt,ꢀ16ꢀh,ꢀ
85%;ꢀc)ꢀ2,3-difluorobenzylꢀalcohol,ꢀNIS,ꢀTfOH,ꢀ
MeCN,ꢀα-isomer:ꢀ60%;ꢀb-isomer:ꢀ11%;ꢀd)ꢀ
PPh3,ꢀp-Cl-BzCl,ꢀDCE,ꢀr.t.,ꢀ48%;ꢀe)ꢀ10%ꢀaq.ꢀ
LiOH,ꢀTHF/H2O;ꢀDowexꢀ50ꢀ×ꢀ8ꢀ(Na+),ꢀ30%.
OAc
N3
OAc
OAc
N3
F
COOMe
SMe
COOMe
SMe
COOMe
N3
OAc
OAc
AcO
OAc
b)
c)
F
H
N
O
O
H
O
O
RHN
N
O
O
AcO
CH2F
AcO
CH2F
12[41], R = Boc
13, R = H
14
15
a)
Cl
Cl
H
N
H
N
OAc
OAc
OH
F
F
d)
COOMe
O
e)
COONa
O
O
OH
F
F
O
O
O
H
H
N
O
N
O
AcO
HO
CH2F
17[41]
16

From ChemiCal researCh to industrial appliCationsꢀ
CHIMIAꢀ2010,ꢀ64,ꢀNo.ꢀ1/2ꢀ 21
[1] M. E. Schwab, C. E. Bandtlow, Nature 1994,
371, 658.
A
[2] S. Ramon
y
Cajal, ‘Degeneration and
Regeneration of the Nervous System’, Oxford
Univ. Press, London, 1928.
[3] M. E. Schwab, P. Caroni, J. Neurosci. 1988, 8,
2381.
[4] A. Sandvig, M. Berry, L. B. Barrett, A. Butt, A.
Logan, GLIA 2004, 46, 225.
[5] M. T. Filbin, Nature Rev. Neurosci. 2003, 4,
703.
[6] Z. G. He, V. Koprivica, Ann. Rev. Neurosci.
2004, 27, 341.
[7] P. Caroni, T. Savio, M. E. Schwab, Prog. Brain
Res. 1988, 78, 363.
[8] R. H. Quarles, Neurochem. Res. 2009, 34, 79.
[9] R. H. Quarles, J. Neurochem. 2007, 100, 1431.
[10] P. R. Crocker, E. A. Clark, M. T. Filbin, S.
Gordon, Y. Jones, J. H. Kehrl, S. Kelm, N. Le
Douarin, L. Powell, J. Roder, R. L. Schnaar,
D. C. Sgroi, K. Stamenkovic, R. Schauer, M.
Schachner, T. K. van den Berg, P. A. van der
Merwe, S. M. Watt, A. Varki, Glycobiology
1998, 8, Glycoforum 2, v.
[11] S. Kelm, A. Pelz, R. Schauer, M. T. Filbin, S.
Tang, M. E. de Bellard, R. L. Schnaar, J. A.
Mahoney, A. Hartnell, P. Bradfield, Curr. Biol.
1994, 4, 965.
[12] J. Lauren, F. Hu, J. Chin, J. Liao, M. S.
Airaksinen, S. M. Strittmatter, J. Biol. Chem.
2007, 282, 5715.
[13] L. A. Robak, K. Venkatesh, H. Lee, S. J. Raiker,
Y. Duan, J. Lee-Osbourne, T. Hofer, R. G.
Mage, C. Rader, R. J. Giger, J. Neurosci. 2009,
29, 5766.
Fig.ꢀ5.ꢀA)ꢀPhysiologicalꢀMAGꢀconsistingꢀofꢀaꢀVꢀdomain,ꢀfourꢀconsensusꢀrepeats,ꢀaꢀtransmembraneꢀ
domainꢀandꢀaꢀcytosolicꢀtail.ꢀItꢀwasꢀexpressedꢀasꢀanꢀFcꢀconjugateꢀwithꢀtwoꢀVꢀdomainsꢀ(MAGd1-3
-Fc[43]),ꢀallowingꢀaꢀnon-covalentꢀimmobilizationꢀonꢀproteinꢀA,ꢀwhichꢀonꢀitsꢀpartꢀwasꢀcovalentlyꢀ
immobilizedꢀbyꢀamineꢀcoupling;ꢀB)ꢀTheꢀV-domainꢀofꢀMAGꢀcontainsꢀthreeꢀlysines,ꢀoneꢀofꢀwhichꢀisꢀ
locatedꢀcloseꢀtoꢀtheꢀcarbohydrateꢀbindingꢀsite.ꢀTherefore,ꢀimmobilizationꢀonꢀtheꢀbioacoreꢀchipꢀbyꢀ
amineꢀcouplingꢀledꢀtoꢀaꢀsurfaceꢀwhichꢀwasꢀnotꢀsaturable;ꢀC)ꢀ&ꢀD)ꢀWithꢀthisꢀstableꢀsurface,ꢀBiacoreꢀ
experimentsꢀleadingꢀtoꢀreproducibleꢀblockꢀsignalsꢀandꢀKDꢀdeterminationꢀwereꢀperformed.[41]
[14] L. J. S. Yang, C. B. Zeller, N. L. Shaper, M.
Kiso, A. Hasegawa, R. E. Shapiro, R. L.
Schnaar, Proc. Nat. Acad. Sci. USA 1996, 93,
814.
[15] S. Tang, Y. J. Shen, M. E. DeBellard, G.
Mukhopadhyay, J. L. Salzer, P. R. Crocker, M.
T. Filbin, J. Cell Biol. 1997, 138, 1355.
[16] M. Vinson, P. J. L. M. Strijbos, A. Rowles, L.
Facci, S. E. Moore, D. L. Simmons, F. S. Walsh,
J. Biol. Chem. 2001, 276, 20280.
[17] V. Wörter, R. Schweigreiter, B. Kinzel, M.
Mueller, C. Barske, G. Böck, S. Frentzel,
C. E. Bandtlow, PLoS ONE 2009, 4, e5218;
doi:10.1371/journal.pone.0005218.
[18] L. J. S.Yang, I. Lorenzini, K.Vajn,A. Mountney,
L. P. Schramm, R. L. Schnaar, Proc. Nat. Acad.
Sci. USA 2006, 103, 11057.
target compartment of an in vivo applica-
Trp22 and Tyr124. Whereas 4 develops a
tion. Furthermore, the logDoctanol/water value
second salt bridge between the carboxylate
of –0.26 is beneficial for an intrathecal ap-
plication, since this distribution coefficient
suggests a loss from the CNS compartment
by a passive transport mechanism to be un-
of the α(2-6)-linked Neu5Ac and Lys67,
the mimics 17 and 29 establish prominent
interactions with two hydrophobic pock-
ets. Glu131 and Tyr127 are homing the p-
chloro benzamide substituent and the side
likely. This hypothesis is further supported
by the results of the BBB-PAMPA assay
chains of Trp59, Tyr60, Tyr69 and Tyr116
showing a log P value of –10 for 17. For
values below –5e.7, no passive permeation
through the BBB is expected.[45]
are lining the main hydrophobic pocket
and accommodate the 2,3-difluorobenzene
(17) and biphenylmethyl moiety (29),
respectively.
[19] B. E. Collins, M. Kiso, A. Hasegawa, M.
B. Tropak, J. C. Roder, P. R. Crocker, R. L.
Schnaar, J. Biol. Chem. 1997, 272, 16889.
[20] O. Blixt, B. E. Collins, I. M. van den
Nieuwenhof, P. R. Crocker, J. C. Paulson, J.
Biol. Chem. 2003, 278, 31007.
Docking of Antagonists to a
Homology Model of MAG
Conclusion
For docking studies of the antagonists
4, 17 and 29, a homology model of MAG
based on the three-dimensional structure
of sialoadhesin was used.[29,30] The ligands
were first manually docked to the binding
pocket of the MAG model using the salt
bridge to Arg118 and the hydrogen bond
of the 9-OH to the backbone carbonyl of
Phe129 as anchor points. Next, the pro-
tein–ligand complex was minimized in
aqueous solution and then subjected to a
molecular-dynamics equilibration proto-
col. The lowest-energy binding modes of
tetrasaccharide 4 and mimics 17 and 29 are
compared in Fig. 6.
All three antagonists, tetrasaccharide 4
(Fig. 6A) and the mimics 17 (Fig. 6B) and
29 (Fig. 6C) form salt bridges withArg118.
In addition, the three antagonists establish
a crucial hydrophobic interaction between
the 5-amido groups and the side chains of
Overall, with antagonist 17 the affin-
ity of lead 4 could be improved more than [21] H. Ito, H. Ishida, B. Collins, S. Fromholt, R.
Schnaar, M. Kiso, Carbohydr. Res. 2003, 338,
350-fold. Its pharmacokinetic properties
1621.
certify the drug-like properties of the best
[22] A. A. Vyas, O. Blixt, J. C. Paulson, R. L.
so far identified MAG antagonist. A fur-
Schnaar, J. Biol. Chem. 2005, 280, 16305.
ther important issue to be addressed is
the metabolic stability of the presented
oligosaccharide mimics. In general, the
substrate specificity of mammalian siali-
dases is determined by the linkage type of
[23] L. Herfurth, B. Ernst, B. Wagner, D. Ricklin, D.
S. Strasser, J. L. Magnani, A. J. Benie, T. Peters,
J. Med. Chem. 2005, 48, 6879.
[24] S. Shin, H. Gäthje, O. Schwardt, G. Gao, B.
Ernst, S. Kelm, B. Meyer, ChemBioChem 2008,
9, 2946.
the terminal sialic acid residue (2-3, 2-6 [25] S. V. Shelke, G. P. Gao, S. Mesch, H. Gäthje,
S. Kelm, O. Schwardt, B. Ernst, Bioorg. Med.
or 2-8) and does not depend on the struc-
Chem. 2007, 15, 4951.
ture of the underlying oligosaccharide.[46]
[26] a) E. Bieberich, S. S. Liour, R. K. Yu, Methods
Therefore, it cannot be excluded that
Enzymol. 2000, 312B, 339; b) J. Kopitz, K.
Sinz, R. Brossmer, M. Cantz, Eur. J. Biochem.
1997, 248, 527.
the presented mimics are metabolically
cleaved by sialidases. Nevertheless, this
new class of MAG blockers constitute an
important step toward the development of
potent oligosaccharide mimics.
[27] a) K. Hotta, H. Ishida, A. Hasegawa, J.
Carbohydr. Chem. 1995, 14, 491; b) H. Ito,
H. Ishida, S. Ando, M. Kiso, Glycoconjugate
J. 1999, 16, 585; c) N. Sawada, H. Ishida, B.
E. Collins, R. L. Schnaar, M. Kiso, Carbohydr.
Res. 1999, 316, 1; d) B. E. Collins, H. Ito, N.
Received: January 20, 2010

22ꢀ CHIMIAꢀ2010,ꢀ64,ꢀNo.ꢀ1/2ꢀ
From ChemiCal researCh to industrial appliCations
Sawada, H. Ishida, M. Kiso, R. L. Schnaar, J.
Biol. Chem. 1999, 274, 37637; e) H. Ito, H.
Ishida, M. Kiso, J. Carbohydr. Chem. 2001, 20,
207.
[28] K. Strenge, R. Schauer, N. Bovin, A. Hasegawa,
H. Ishida, M. Kiso, S. Kelm, Eur. J. Biochem.
1998, 258, 677.
[29] A. Bhunia, O. Schwardt, H. Gäthje, G. Gao, S.
Kelm, A. J. Benie, M. Hricovini, T. Peters, B.
Ernst, ChemBioChem 2008, 9, 2941.
[30] O. Schwardt, H. Gaethje, A. Vedani, S. Mesch,
G. Gao, M. Spreafico, J. von Orelli, S. Kelm, B.
Ernst, J. Med. Chem. 2009, 52, 989.
[31] F. Mohamadi, N. G. Richards, W. Guida, R.
Liskamp, M. Lipton, C. Caufield, G. Chang,
T. Hendrickson, W. C. Still, J. Comput. Chem.
1990, 11, 440.
[32] S. J. Weiner, P. A. Kollmann, D. A. Case, U. C.
Singh, C. Ghio, G. Alagona, S. Profeta Jr., P.
Weiner, J. Am. Chem. Soc. 1984, 106, 765.
[33] J. J. P. Stewart, J. Comput. Aided Molec. Design
1990, 4, 1.
[34] D. E. Shaw Research, ‘Desmond Molecular
Dynamics System’, 2.0109; New York, NY,
2008.
[35] W. Humphrey, A. Dalke, K. Schulten, J. Molec.
Graphics 1996, 14, 33.
[36] D. Schwizer, H. Gäthje, S. Kelm, M. Porro, O.
Schwardt, B. Ernst, Bioorg. Med. Chem. 2006,
14, 4944.
[37] R. Kuhn, P. Lutz, D. L. MacDonald, Chem. Ber.
1966, 99, 611.
[38] G. Gao, M. Smiesko, O. Schwardt, H. Gäthje, S.
Kelm, A. Vedani, B. Ernst, Bioorg. Med. Chem.
2007, 15, 7459.
[39] S. Kelm, R. Brossmer, R. Isecke, H.-J. Gross,
K. Strenge, R. Schauer, Eur. J. Biochem. 1998,
255, 663.
[40] S. Kelm, R. Brossmer, PCT Patent WO
03/000709A2, 2003.
[41] S. Mesch, D. Moser, D. S. Strasser, A. Kelm,
B. Cutting, G. Rossato, A. Vedani, H. Koliwer-
Brandl, M. Wittwer, S. Rabbani, O. Schwardt,
S. Kelm, B. Ernst, J. Med. Chem. 2010, in press;
doi: 10.1021/jm901517k.
[42] P. Boullanger, V. Maunier, D. Lafont,
Carbohydr. Res. 2000, 324, 97.
[43] N. Bock, S. Kelm, in ‘Methods in Molecular
Biology’, Vol. 347, Ed. I. Brockhausen,
Humana Press Inc, Totowa NJ, 2006, p 359.
[44] R. Mitchell, D. Herbert, C. Carman, J. Appl.
Physiol. 1965, 20, 27.
[45] L. Di, E. Kerns, K. Fan, O. McConnell, Eur. J.
Med.Chem. 2003, 38, 223.
[46] J. Kopitz, K. Sinz, R. Brossmer, M. Cantz, Eur.
J. Biochem. 1997, 248, 527.
Fig.ꢀ6.ꢀTheꢀthree-dimensionalꢀstructuresꢀofꢀ4,ꢀ17ꢀandꢀ29ꢀwereꢀgeneratedꢀ
usingꢀtheꢀMacroModelꢀsoftware[31]ꢀandꢀoptimizedꢀinꢀaqueousꢀsolutionꢀbyꢀ
meansꢀofꢀtheꢀAMBER*ꢀforceꢀfield.[32]ꢀAtomicꢀpartialꢀchargesꢀ(MNDO/ESP)ꢀ
wereꢀthenꢀgeneratedꢀusingꢀMOPAC.[33]ꢀTheꢀligandsꢀwereꢀfirstꢀmanuallyꢀ
dockedꢀandꢀtheꢀprotein–ligandꢀcomplexꢀwasꢀminimizedꢀinꢀaqueousꢀ
solutionꢀandꢀthenꢀsubjectedꢀtoꢀaꢀmolecular-dynamicsꢀequilibrationꢀ
protocolꢀ(24ꢀpsꢀatꢀ10ꢀK,ꢀheatingꢀtoꢀ300ꢀKꢀduringꢀ48ꢀps,ꢀ1ꢀpsꢀ=ꢀ10–12
ꢀ
s),ꢀfollowedꢀbyꢀaꢀmolecularꢀdynamicꢀatꢀ300ꢀKꢀforꢀ4ꢀnsꢀperformedꢀwithꢀ
Desmond[34]ꢀ(atꢀ2.4ꢀpsꢀintervals).ꢀTheꢀstructuresꢀofꢀtheꢀtrajectoryꢀalongꢀ
theꢀmolecularꢀdynamicꢀsimulationꢀhaveꢀbeenꢀsampledꢀthroughꢀaꢀ
hierarchicalꢀclusterꢀlinkageꢀmethodꢀatꢀ2.4ꢀpsꢀintervals.ꢀTheꢀimagesꢀhaveꢀ
beenꢀgeneratedꢀusingꢀVMD.[35]ꢀA)ꢀTetrasaccharideꢀ4,ꢀaꢀsubstructureꢀofꢀ
gangliosideꢀGQ1bα,ꢀB)ꢀantagonistꢀ29;ꢀantagonistꢀ17.