Pre-organization of the core structure of E-selectin antagonists

Article abstract of DOI:10.1002/chem.201102884

A new class of N-acetyl-dglucosamine (GlcNAc) mimics for Eselectin antagonists was designed and synthesized. The mimic consists of a cyclohexane ring substituted with alkyl substituents adjacent to the linking position of the fucose moiety. Incorporation into E-selectin antagonists led to the test compounds 8 and the 2'-benzoylated analogues 21, which exhibit affinities in the low micromolar range. By using saturation transfer difference (STD)-NMR it could be shown that the increase in affinity does not result from an additional hydrophobic contact of the alkyl substituent with the target protein E-selectin, but rather from a steric effect stabilizing the antagonist in its bioactive conformation. The loss of affinity found for antagonists 10 and 35 containing a methyl substituent in a remote position (and therefore unable to support to the stabilization of the core) further supports this hypothesis. Finally, when a GlcNAc mimetic containing two methyl substituents (52 and 53) was used, in which one methyl was positioned adjacent to the fucose linking position and the other was in a remote position, the affinity was regained.

Full text of DOI:10.1002/chem.201102884

DOI: 10.1002/chem.201102884
Pre-organization of the Core Structure of E-Selectin Antagonists
Daniel Schwizer,[a] John T. Patton,[b] Brian Cutting,[a] Martin Smiesˇko,[a]
Beatrice Wagner,[a] Ako Kato,[a] Cꢀline Weckerle,[a] Florian P. C. Binder,[a]
Said Rabbani,[a] Oliver Schwardt,[a] John L. Magnani,[b] and Beat Ernst*[a]
Abstract: A new class of N-acetyl-d-
glucosamine (GlcNAc) mimics for E-
selectin antagonists was designed and
synthesized. The mimic consists of a cy-
clohexane ring substituted with alkyl
substituents adjacent to the linking po-
sition of the fucose moiety. Incorpora-
tion into E-selectin antagonists led to
the test compounds 8 and the 2’-ben-
zoylated analogues 21, which exhibit
affinities in the low micromolar range.
By using saturation transfer difference
(STD)-NMR it could be shown that
the increase in affinity does not result
from an additional hydrophobic con-
tact of the alkyl substituent with the
target protein E-selectin, but rather
from a steric effect stabilizing the an-
tagonist in its bioactive conformation.
The loss of affinity found for antago-
nists 10 and 35 containing a methyl
substituent in a remote position (and
therefore unable to support to the sta-
bilization of the core) further supports
this hypothesis. Finally, when
a
GlcNAc mimetic containing two
methyl substituents (52 and 53) was
used, in which one methyl was posi-
tioned adjacent to the fucose linking
position and the other was in a remote
position, the affinity was regained.
Keywords: carbohydrates · molecu-
lar dynamics · pre-organization ·
STD-NMR · stereoselectivity · sur-
face plasmon resonance
Introduction
solution[10] and the bioactive[11,12] conformation of sLex have
been elucidated and it could be shown that the pre-organi-
zation of the pharmacophores in the bioactive conformation
contributes substantially to the affinity of E-selectin antago-
nists.[9a,13–15]
The selectins play a key role in the bodyꢀs defense mecha-
nism against inflammation.[1] They form a class of three cell
adhesion molecules (E-, P-, and L-selectin), which, in the
case of an inflammatory stimulus, are responsible for the ini-
tial steps of the inflammatory response, that is, the tethering
and rolling of leukocytes on the endothelial surface. It has
been shown with anti-selectin antibodies[2–4] as well as E-, P-,
and L-selectin knock-out (k.o.) mice[5,6] that these early
steps are a prerequisite for triggering the inflammatory cas-
cade, which finally leads to the extravasation of leukocytes
into the inflamed tissue. However, excessive infiltration of
leukocytes into the adjacent tissue leads to its destruction,
as observed in reperfusion injuries, stroke or rheumatoid ar-
thritis.[7] The antagonism of selectins is therefore regarded
as a valuable therapeutic approach in these indications.[8]
Since all physiological ligands of the selectins contain the
sialyl Lewisx motif (sLex, 1, Figure 1),[9] this provided the
lead structure in the search for E-selectin antagonists. The
In sLex, the hydroxy groups of the fucose moiety, the 4-
and the 6-OH of the galactose moiety and the COOH group
of the sialic acid residue act as pharmacophores, whereas
the N-acetylglucosamine moiety serves as a three-dimen-
sional spacer to position l-fucose underneath the b-face of
d-galactose (see 1, Figure 1).[16] Due to the shallow binding
site of E-selectin[17] and the considerable distances between
the functional groups involved in binding, the three-dimen-
sional orientation and pre-organization of the pharmaco-
phores becomes exceedingly important.
Ligand binding to a biological target is generally associat-
ed with an entropic penalty arising from the loss of internal
conformational degrees of freedom. To increase binding af-
finity, a well-known molecular design strategy is to intro-
duce conformational restrictions, which limits the degrees of
freedom a molecule can potentially lose upon binding.
These conformational locks can be realized by adding sub-
stituents to create steric constraints that favor a particular
conformation or by introducing ring closures. Various suc-
cessful applications of these strategies have been reported.
Examples are high affinity protein tyrosine phosphatase-1B
inhibitors with constraints based on additional substitu-
ents[18] or hepatitis C virus polymerase inhibitors, in which
the torsional angles were fixed by a ring closure.[19]
ˇ
[a] Dr. D. Schwizer, Dr. B. Cutting, Dr. M. Smiesko, B. Wagner,
Dr. A. Kato, Dr. C. Weckerle, F. P. C. Binder, Dr. S. Rabbani,
Dr. O. Schwardt, Prof. B. Ernst
Institute of Molecular Pharmacy, Pharmacenter, University of Basel,
Klingelbergstrasse 50, 4056 Basel (Switzerland)
Fax : (+41)61-267-1552
E-mail: beat.ernst@unibas.ch
[b] Dr. J. T. Patton, Dr. J. L. Magnani
GlycoMimetics Inc., Gaithersburg, MD20878 (USA)
With sLex (1) as a lead structure, numerous mimetics have
been synthesized.[16] (S)-cyclohexyl lactic acid turned out to
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/chem.201102884.
1342
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 1342 – 1351

FULL PAPER
Results and Discussion
Influence of the GlcNAc re-
placement on the core confor-
mation studied by molecular
dynamics: To elucidate the
impact of GlcNAc replacement
by other structural motifs, a
series of computer simulations
was performed on cyclohexane
and tetrahydropyran derivatives
with and without additional
methyl substituents (Figure 2
and Figure 3). As the structural
changes are rather conservative
regarding ring size (6-mem-
bered aliphatic rings are always
present) and conformation
(chair), a traditional approach
Figure 1. sLex (1) and sLex mimics in which Neu5Ac is replaced by (S)-cyclohexyl lactic acid (!2) and
GlcNAc by a flexible alkyl linker (!3), an aromatic linker (!4), (R,R)-cyclohexane-1,2-diol (!5) or tetrahy-
dropyran derivatives (!6 and 7). The methylcyclohexane replacement (!8a) is expected to combine the im-
portant structural elements leading to an improved pre-organization.
be a potent replacement of neuraminic acid (!2[20]).[13]
Originally, the mimetic approach for the N-acetylglucosa-
mine moiety was less successful. A substantial loss of affinity
resulted from a highly flexible (!3[14]) or geometrically in-
accurate linker (!4[21]). With the (R,R)-cyclohexane deriva-
tive (5[13]) or the tetrahydropyran derivative (6[14]) an expedi-
ent replacement of the GlcNAc moiety was identified.
These results are in agreement with the role of GlcNAc as a
geometrically restricted linker of the adjacent d-Gal and l-
Fuc moieties as depicted by Magnani et al.[9a] and later con-
firmed by the X-ray structure of sLex in complex with E-se-
lectin.[17] Furthermore, when an additional methyl group ad-
jacent to the fucose linking position was introduced into the
tetrahydropyran moiety (!7), the improved pre-organiza-
tion of the pseudo-trisaccharide core as deduced from the
NOEs between H-5-Fuc and H-2-Gal led to reduced entro-
py costs upon binding and therefore a six-fold improvement
of affinity compared with the unsubstituted tetrahydropyran
derivative 6.[14] An evaluation of the sLex mimetics 2–7 re-
veals that a further improvement of affinity can be expected
for 8a, since it combines both structural elements already
identified to lead to improved affinities: 1) the cyclohexane
linker leading to a three-fold improvement of affinity com-
pared with the tetrahydropyran linker and 2) the conforma-
tional restriction caused by the methyl substituent leading to
a six-fold improvement of affinity by forcing the adjacent l-
Fuc moiety into the bioactive conformation.
Figure 2. Torsion angle O(4)-C(4)-C(3)-O(3) [8] for antagonist 2 com-
pared with 5–10; the torsion angles determine the mutual position of d-
Gal and l-Fuc from MD simulations for tetrahydropyran and cyclohex-
ane derivatives.
of modeling the lowest energy states would yield only small
differences, which might not be helpful to rationalize the
relatively large variations in binding activities (39 to 280 mm,
see Figure 1). Since the GlcNAc unit and its mimics are not
forming direct interactions to the target protein but are
rather responsible for pre-organizing the pharmacophores of
the antagonist in the bioactive conformation,[13] a molecular
dynamics (MD) approach providing a detailed conforma-
tional description over a given period of time (up to 1.0
nanosecond) was performed.[22]
In this communication, we report the stereoselective syn-
thesis of the E-selectin antagonist 8a as well as derivatives
thereof and correlate (by NMR and molecular dynamics cal-
culations) the affinities with their flexibility and the degree
of their pre-organization in the bioactive conformation.
In general, the cyclohexane derivatives 5, 8a, and 10 ex-
hibit a population of conformations having the torsion angle
Chem. Eur. J. 2012, 18, 1342 – 1351
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
1343

B. Ernst et al.
Scheme 1. a) (2R)-(ꢀ)-MTPA-Cl, DMAP, pyr, RT, 60 h (quant.);
b) Ph3CCl, CH2Cl2, DBU, RT, 45 h (71%); c) mCPBA, NaHCO3,
CH2Cl2, 08C to RT, 5 h (72%); d) RLi, CuCN, BF3·Et2O, THF, ꢀ788C to
ꢀ308C, 22–51 h (16a: 85%, 16b: 70%, 16c: 82%, 16d: 81%); e) i) 17,
Br2, CH2Cl2, 08C, 30–50 min, ii) TEAB, MS 3 ꢂ, DMF, CH2Cl2, RT, 1–
4 d, iii) ZnBr2, TES, CH2Cl2, RT, 45 min–8 h, (18a: 49%, 18b: 55%, 18c:
67%, 18d: 60%, yields are over two steps); f) Pd/C, H2, THF, RT, 30 min
(18e, 77%); g) i) methyl acrylate, Grubbs cat. 2nd gen., CH2Cl2, reflux,
9 d, ii) Pd/C, H2, THF, RT, 30 min (18 f, 25% over two steps). mCPBA=
Figure 3. Distance C(2)Gal-C(5)Fuc [ꢂ] for antagonist 2 compared with
5–10; calculated distribution of the distances between d-Gal and l-Fuc
from MD simulations for tetrahydropyran and cyclohexane derivatives.
meta-chloroperoxybenzoic acid, DBU=1,8-diazabicyclo
ene, TEAB=triethylammonium bicarbonate.
ACHTUNGTREN[NGNU 5.4.0]undec-7-
O(4)-C(4)-C(3)-O(3) comparable to the one for the parent
compound 2. In contrast, the tetrahydropyran derivatives 6,
7, and 9 show a shift to larger torsion angles leading to a dif-
ferent spatial orientation of the d-Gal and the
l-Fuc moiety (Figure 2). For both ring systems, a methyl
group adjacent to the linking position of the l-Fuc as pres-
ent in 7, 8a, and 9 stabilizes the core conformation with a
maximal population at a distance of 4.2 ꢂ between C-2 of
d-Gal and C-5 of l-Fuc. Two of them, compounds 8a and 9,
exhibit profiles almost identical to the one for parent com-
pound 2, which indicates comparable flexibility of the
GlcNAc mimetics. When the methyl group is missing or in a
remote position (see compounds 5, 6, and 10), the core
gains flexibility leading to a broader distribution of the dis-
tances between the C-2 of d-Gal and the C-5 of l-Fuc
(Figure 3). In summary, the core conformation present in
parent compound 2 is best mimicked regarding torsion
angle O(4)-C(4)-C(3)-O(3) and the C(2)Gal-C(5)Fuc dis-
tance by (1R,2R,3S)-3-methyl-cyclohexane-1,2-diol as pres-
ent in 8a. Whether a sterically more demanding substituent
in the 3-position would further improve the stabilization of
the core and thus reduce entropy costs upon binding re-
mains to be tested.
lites of the methyl resonance of the (S)-Mosher ester 12.
After halogen-metal exchange in 11 with tBuLi and subse-
quent hydrolysis (!13) and tritylation, ether 14 was ob-
tained in 71%. Epoxidation with mCPBA yielded a separa-
ble 3:1 mixture of the epoxides anti-15 (72%) and syn-15
(23%). The observed diastereoselectivity is the result of the
directing effect of the sterically demanding trityl group. A
similar selectivity was reported earlier for racemic TBDMS-
protected cyclohex-2-enol.[25] The regioselective opening of
the epoxide anti-15 with a variety of higher-order cyanocup-
rates in the presence of BF3·Et2O yielded the GlcNAc
mimics 16a–d in good to excellent yields.[26]
For the subsequent fucosylation with donor 17[27]
(Scheme 1), the in situ anomerization procedure developed
by Lemieux and co-workers[28] was applied, yielding almost
exclusively the a-anomers. After detritylation with ZnBr2
the disaccharide mimics 18a–d were obtained in 49–67%
overall yield for the glycosylation and deprotection step.
The ethyl-substituted pseudo-disaccharide 18e was obtained
by Pd-catalyzed hydrogenation of the vinyl derivative 18d.
Finally, the 2-methoxycarbonylethyl-substituted disaccharide
mimic 18 f was synthesized by cross-metathesis of 18d with
methyl acrylate in the presence of Grubbsꢀ 2nd generation
catalyst[29] followed by hydrogenation of the cis/trans mix-
ture.
Synthesis of GlcNAc mimetics for E-selectin antagonists:
The enantiomerically pure GlcNAc mimetics were obtained
by different approaches, that is, enantioselective reduction
(Scheme 1), separation of diastereomers (Scheme 3) or en-
zymatic hydrolysis (Scheme 4).
The synthesis of the GlcNAc mimics 16a–d (Scheme 1)
started from 2-cyclohexenone, which was transformed[23]
into the prochiral bromoketenone. (1R)-2-bromocyclohex-2-
en-1-ol (11) was then obtained in 93% yield by asymmetric
CBS reduction.[24] An enantiomeric excess (ee) of 96% was
determined by optical rotation and based on the 13C-satel-
Galactosylation with donor 19[30] promoted by dimethyl-
(methylthio)sulfonium triflate (DMTST) afforded b-selec-
tively the tetrasaccharide mimics 20a–c, 20e, and 20 f
(Scheme 2). The final deprotection of 20a and 20e by de-
benzylation, followed by saponification with lithium hydrox-
ide and ion exchange chromatography yielded the fully de-
1344
www.chemeurj.org
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 1342 – 1351

Pre-organization of the Core Structure of E-Selectin Antagonists
FULL PAPER
Scheme 2. a) DMTST, MS 3 ꢂ, CH2Cl2, RT, 43 h to 65.5 h (20a: 59%,
20b: 80%, 20c: 87%, 20e: 78%, 20 f: 79%); b) i) Pd/C, H2, EtOH, cat.
AcOH, RT, ii) LiOH, MeOH/H2O, RT, 2 d, iii) Dowex (Na+), Sephadex-
G15 (8a: 74%, 8e: 30%); c) i) Pd(OH)2/C, H2, dioxane/H2O, RT, ii)
NaOMe, MeOH, RT, iii) Dowex (Na+), Sephadex-G15 (21a: 79%, 21b:
56%, 21c: 72%, 21e: 77%, 21 f: 73%). DMTST=dimethyl(methylthio)
sulfonium triflate.
Scheme 3. a) Rh/Al2O3, cyclohexane, THF, H2, RT, 50 h (78%, rac-23 :
rac-24=4:1); b) Bu2SnO, BzCl, NEt3, CH2Cl2, 08C, 1.5 h (53%, rac-25 :
rac-26=4:1); c) 3,5-dinitrobenzoic acid, PPh3, DEAD, toluene, RT, 22 h
(rac-27: 48%, rac-28: 12%); d) MeOH, NEt3, RT, 30 min (89%);
e) Bu4NBr, CuBr2, CH2Cl2, DMF, MS 4 ꢂ, RT, 17 h; f) LiOH, MeOH/
H2O, 50 8C, 3.5 h (32: 35% and 33: 32%, over two steps); g) DMTST,
CH2Cl2, MS 4 ꢂ, RT, 65 h (66%); h) i) H2, Pd(OH)2, dioxane/H2O, RT,
4 h; ii) NaOMe, MeOH, RT, 16 h (83%); i) LiOH, MeOH/H2O, RT, 18 h
(73%). DEAD=diethyl azodicarboxylate.
protected antagonists 8a and 8e. However, when the saponi-
fication was carried out with catalytic amounts of sodium
methoxide in methanol, the 2’-monobenzoylated derivatives
21a–c, 21e, and 21 f were obtained. Selectin antagonists acy-
lated in the 2’-position of the galactose moiety are of inter-
est, since Thoma et al.[15] have shown that a benzoyl sub-
stituent in the 2’-position improves the affinity for E-selectin
by approximately a factor of 3, presumably by contributing
by steric means to the stabilization of the Lewisx mimic in
its bioactive conformation.
Finally, antagonists 52 and 53 containing a GlcNAc mimic
with two methyl substituents, one adjacent to the linking po-
sition of the l-Fuc moiety and one in a remote position,
were synthesized (Scheme 4). The synthesis started from cis-
1,2,3,6-tetrahydrophthalic anhydride, which was transformed
into the dimethyl ester 37 by acid catalysis in methanol.[33]
Desymmetrization was accomplished by enantioselective hy-
drolysis of the symmetrical diester by pig liver esterase
(PLE).[34] The optical purity of the resulting half-ester 38
was analyzed by optical rotation (96.4% ee) and by chiral
GC (96.0% ee). Subsequent Barton decarboxylation[35]
As negative control, antagonist 10 was synthesized
(Scheme 3). Antagonist 10 contains a methyl group in a
remote position in which it cannot exert a repulsive effect
on the l-Fuc moiety. The synthesis of 10 starts from 4-meth-
ylcatechol (22), which was hydrogenated to an inseparable
4:1-mixture of the 4-methyl-1,2-cyclohexanediols rac-23 and
rac-24. After benzoylation of the equatorial (!rac-25 and
rac-26) and inversion of the axial hydroxy group under Mit-
sunobu conditions[31] a diastereomeric mixture of the dini-
trobenzoates rac-27 and rac-28 was obtained, which could
be separated by chromatography. For the protection of rac-
27, 3,5-dinitrobenzoate was preferred over the correspond-
ing para-nitrobenzoate, because the subsequent hydrolysis
to rac-29 could be performed under milder conditions
(NEt3, MeOH), avoiding the partial cleavage of the adjacent
benzoate. Fucosylation of rac-29 with 17[27] and Bu4NBr/
CuBr2 as promoters[32] yielded a diastereomeric mixture of
30 and 31, which was separated by chromatography after de-
benzoylation (!32 & 33). The structural assignment of 32 is
A
(R)-cyclohex-3-ene carboxylic acid (40) with an enantiomer-
ic excess of 96.3%, determined by chiral GC. Iodolactoniza-
tion (!41) and DBU induced dehydroiodination afforded
lactone 42.[36] Transesterification with methanol in the pres-
ence of sodium bicarbonate and protection of the secondary
alcohol with tert-butyl-dimethylsilyl chloride yielded 43. Due
to the directing effect of the sterically demanding TBDMS
group, epoxidation with mCPBA yielded preferentially anti-
epoxide 44.[25a] Regioselective opening of the epoxide with
Me2Cu(CN)Li2 in the presence of BF3·Et2O finally provided
45.[26]
1
based on a comparison with the H NMR spectra of the cor-
responding partial structure of the non-methylated antago-
nist 5.[13] By galactosylation of 32 with 19,[30] the pseudo-tet-
rasaccharide 34 was obtained in 66% yield. Finally, the step-
wise final deprotection gave the 2’-benzoylated compound
35 and the fully deprotected antagonist 10.
To prevent desilylation, a-fucosylation of 45 was per-
formed under in situ anomerization conditions[32] in the
presence of 2,6-di-tert-butyl-4-methylpyridine to yield the
Chem. Eur. J. 2012, 18, 1342 – 1351
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
1345

B. Ernst et al.
or BSA neoglycoproteins to quantify binding to recombi-
nant selectins.[40] The target-based assay used in this commu-
nication utilized a polyacrylamide-type glycoconjugate as
synthetic ligand for immobilized E-selectin.[38,40,41] Briefly,
microtiter plates were coated with E-selectin/IgG,[12] blocked
with BSA, and incubated with a fixed concentration of sLea-
polyacrylamide (sLea-PAA) and various concentrations of
the antagonists. The binding reaction was revealed by the
addition of 3,3’,5,5’-tetramethylbenzidine (TMB) substrate
reagent and quantified spectrophotometrically at 450 nm.
The IC50 defines the molar concentration of the test com-
pound that reduces the maximal specific binding of sLea-
PAA polymer to E-selectin by 50%. To ensure comparabili-
ty of different antagonists, the reference compound 5
(CGP69669[4,13]) was tested in parallel on each individual mi-
crotiter plate. The affinities are reported relative to 5 as
rIC50 in Table 1. The relative IC50 (rIC50) is the ratio of the
IC50 of the test compound and the IC50 of 5 (Table 1,
entry 2).
Scheme 4. a) Amberlyste 15, MeOH, CHACHTNUTRGNE(UNG OCH3)3, RT, 9 d (88%);
b) PLE, buffer pH 7, aq. NaOH, 208C, 56.5 h (90%, 96% ee);
c) i) (COCl)2, CH2Cl2, cat. DMF, RT, 3 h, ii) 2-mercaptopyridine-1-oxide
sodium salt, tBuSH, cat. DMAP, THF, reflux, 3 h (83%); d) PLE, buffer
pH 7, aq. NaOH, RT, 11 h (84%, 96% ee); e) KI, I2, NaHCO3, H2O, RT,
24 h (95%); f) DBU, THF, reflux, 20 h (94%); g) i) NaHCO3, MeOH,
RT, 12 h, ii) TBSCl, DBU, CH2Cl2, RT, 12 h (quant.); h) mCPBA,
CH2Cl2, 108C to RT, 17 h (78%); i) MeLi, CuCN, BF3·Et2O, THF,
ꢀ788C, 5 h (78%); j) TBAB, 2,6-di-tert-butyl-4-methylpyridine, MS 4 ꢂ,
CuBr2, DMF, CH2Cl2, RT, 20 h (76%); k) LiAlH4, THF, 08C, 1 h (84%);
l) 1-chloro-N,N,2-trimethylpropenylamine,[37] DCE, RT, 45 min (85%);
m) Bu3SnH, AIBN, THF, 908C, 90 min (71%); n) AcOH, THF, 808C, 4 h
(!50, 68%); o) DMTST, CH2Cl2, RT, 4 h (62%); p) i) H2, Pd(OH)2, di-
oxane/H2O, RT, 4 h, ii) NaOMe, MeOH, RT, 4 h (80%); q) LiOH,
MeOH/H2O, RT, 18 h (80%). AIBN=azobisisobutyronitrile.
pseudo-disaccharide mimic 46, which was then reduced to
the primary alcohol 47. In the next step, chloride 48 ob-
tained by treatment with 1-chloro-N,N,2-trimethylpropenyl-
ACHTUNGTRENNUNG
amine[37] was reduced with nBu3SnH to 49. After removal of
the silyl protection (!50), glycosylation with donor 19[30]
using DMTST as promoter afforded b-selectively the tetra-
saccharide mimic 51. Debenzylation and saponification with
catalytic amounts of sodium methoxide in methanol yielded
the 2’-benzoylated antagonist 52. Finally, treatment with
LiOH in MeOH gave the fully deprotected antagonist 53.
Figure 4. a) Representative sensorgram showing the reference subtracted
signals for compound 21a binding to immobilized E-selectin/IgG;
b) Binding affinity curve extracted from the steady state of the sensor-
gram.
Surface plasmon resonance (SPR): The interactions between
E-selectin and the antagonists 5, 8a, 21a, 21c, 21e, and 21 f
were analyzed by surface plasmon resonance (Biacore) at
258C.[39] 6000–7000 RU of E-selectin/IgG were immobilized
on the chip surface by using anti-human Fc antibody as de-
scribed in the Experimental Section. An example of a repre-
sentative sensorgram is presented in Figure 4. All the sen-
sorgrams were referenced to a reference flow cell containing
only the anti-human Fc antibody. The equilibrium response
units were extracted from sensorgrams (Figure 4a) and plot-
ted as a function of the concentration (Figure 4b). Control
Determination of the affinity of the E-selectin antagonists:
Affinities were determined with two different assay formats,
that is, a classical competitive binding assay[38] and surface
plasmon resonance (SPR) experiments.[39]
Competitive binding assay: For the characterization of selec-
tin/ligand interactions, several in vitro assays are reported.
Most of these are cell-based, and measure leukocyte binding
to activated endothelial cells or recombinant selectins. Cell-
free assays utilize the carcinoembryonic antigen, glycolipids,
1346
www.chemeurj.org
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 1342 – 1351

Pre-organization of the Core Structure of E-Selectin Antagonists
FULL PAPER
Table 1. Affinity of selectin antagonists 2, 5, 8a, 8e, 10, 35, 53, 52, 21a–c, 21e, and 21 f binding to E-selectin/IgG. IC50 values were determined with a
competitive binding assay.[38] Relative IC50 values (rIC50) were calculated by dividing the IC50 of the substance of interest by the IC50 of the reference
compound 5 (entry 2). Dissociation constants KD were determined by surface plasmon resonance (SPR) experiments.[39]
Entry
Antagonist
SPR
Entry
Antagonist
SPR
assay
assay
IC50 rIC50
KD
IC50 rIC50
KD
[mM]
[mM]
[mM]
[mM]
1
2
280 3.49
n.d.[a]
8
52
5.3 0.06
5.2 0.06
n.d.
2
3
4
5
6
7
5
8a
8e
10
35
53
80.2
1
44.9
7.6
9
21a
21b
21c
21e
21 f
2.1
n.d.
5.4
1.5
1.6
13 0.16
8.3 0.10
10
11
12
13
9
11
4
0.11
0.14
0.05
0.10
n.d.
n.d.
n.d.
n.d.
94 1.17
22.3 0.28
13.6 0.17
8
[a] n.d=not determined.
experiments were performed in the presence of EDTA to
check the specificity and the calcium ion dependency of the
binding.
within the series of cyclohexane derivatives (Table 1, en-
tries 2 to 13), these factors are no longer present and should
allow a comparison of the affinities with the focus on pre-or-
ganization.
When the GlcNAc moiety in antagonist 2 is replaced by
(R,R)-cyclohexane-1,2-diol (!5), the binding affinity is im-
proved by more than a factor of 3 (DDG8ꢁ2 kJmolꢀ1), al-
though the carbocyclic exchange is only partially mimicking
the core geometry of Lewisx (see Figures 2 and 3). The im-
proved affinity of 5 is therefore rather associated with physi-
cochemical parameters related to the structural alteration.
Firstly, hydration effects of polar groups within, but also
about the periphery of the binding epitope can strongly in-
fluence association.[42,43] Secondly, the interaction of E-selec-
tin with its ligand is entropically driven[44] and hence differs
from typical protein–carbohydrate interactions.[45] The
switch from d-GlcNAc to (R,R)-cyclohexane-1,2-diol may
further influence this entropic balance, for example the en-
tropy costs related to flexibility of ring substituent of d-
GlcNAc, and therefore influence affinity as well. However,
For our studies, we selected antagonist 5 as reference
compound. When an alkyl substituent adjacent to the link-
ing position of the l-Fuc moiety was introduced, a sixfold
(8a, Table 1, entry 3) and tenfold (8e, Table 1, entry 4) im-
provement of affinity was achieved. This is in perfect agree-
ment with the predicted enhanced degree of pre-organiza-
tion of the core of the antagonists (Figures 2 and 3). Fur-
thermore, with a methyl substituent in a remote position,
that is, when the steric effect leading to pre-organization is
lost, an affinity in the range of the starting point was ob-
tained (10, Table 1, entry 5). With antagonists with two
methyl groups, one adjacent to the linking position of l-Fuc
moiety and one in a remote position (53, Table 1, entry 7)
affinity was regained, indicating that the steric effect of one
methyl group is responsible for the gain in affinity and the
Chem. Eur. J. 2012, 18, 1342 – 1351
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
1347

B. Ernst et al.
one in the remote position does not contribute to affinity.
When the 2’-position of d-Gal was benzoylated, the report-
ed[15] beneficial effect leading to a fourfold improvement of
affinity could be observed (35, Table 1, entry 6). A similar
effect was observed for the couple of 53 (Table 1, entry 7)
and 52 (entry 8).
In the last series, the effect of different alkyl substituents
adjacent to the linking position of the l-Fuc moiety was in-
vestigated. The highest affinity was achieved with an ethyl
substituent (21e, Table 1, entry 12), whereas the methyl sub-
stituent gave a slightly less active antagonist (21a, Table 1,
entry 9). All extended substituents, n-butyl (21b, Table 1,
entry 10), cyclopropyl (21c, Table 1, entry 11), and 2-me-
thoxycarbonylethyl (21 f, entry 13) resulted in reduced affin-
ities.
NMR analysis of the selectin antagonists 5 and 8a: To sub-
stantiate our assumption that the higher binding affinity of
antagonist 8a compared with 5 is a consequence of the
closer stacking of l-Fuc to d-Gal, both the unbound solution
conformations, as well as the STD-derived binding epitopes,
were studied by STD-NMR experiments. Similarities in the
STD-derived binding epitopes correlate with similarities in
the mode of binding.[46,47]
The investigation of the unbound conformation was per-
formed through selective one-dimensional ROESY meas-
urements.[48] The interglycosidic ROE between H-5-Fuc and
H-2-Gal (Figure 5) was used to determine the proximity of
l-Fuc to d-Gal in 5 and 8a. Assuming a distance of 2.60 ꢂ
for H-5-Fuc to H-3-Fuc, the average distance between H-5-
Fuc and H-2-Gal was quantified to be 2.98 ꢂ for 5 and
2.73 ꢂ for 8a (for details see the Experimental Section),
which clearly indicates a closer stacking in 8a (Figure 5c).
These distances are in good agreement with the molecular
modeling results reported in Figure 3. The reduced distance
between H-5-Fuc and H-2-Gal in inhibitor 8a could addi-
tionally be confirmed by the chemical shift of the H-5-Fuc
proton.[14] In compound 8a the H-5-Fuc is shifted downfield
to d=4.83 ppm compared with the shifts of d=4.60 ppm for
5.[15]
The measured ROE values and the chemical shift of the
H-5-Fuc proton provide evidence for the closer stacking of
the a-face of the l-Fuc moiety to the b-face of the d-Gal
moiety and suggest that the improved affinity of 8a is a con-
sequence of an increased degree of pre-organization of the
core structure in the biologically active conformation.
To determine the similarity in the binding modes of 5 and
8a, the STD-NMR spectra were measured to derive the
binding epitopes (Figure 6).[50] The value of the STD trans-
fer of magnetization from E-selectin to the different hydro-
gen atoms of the two antagonists is indicated in a color code
and tabulated in Figure 7. STD values are referenced with
respect to the value for H-2 of lactic acid, defined as 100. In-
spection of the STD epitopes for 5 and 8a reveals that these
two antagonists have a nearly identical mode of binding.
The largest difference, albeit with only 8%, between the an-
alyzed resonances common to 5 and 8a was found for H-1-
Figure 5. Selective 1D ROESY experiments measuring the transfer of
magnetization from H-5-Fuc to H-2-Gal in compound 8a. Analysis of the
time-dependence of the transfer of magnetization to the target peak (a)
from the source peak (b) provides the cross-relaxation rate constant that
is used to determine the internuclear distance. c) Internuclear distances
for compounds 5 and 8a were determined through linear regression of
ratio of the signal of the target signal divided by the source signal verses
the mixing time.[49] In all four cases, the source signal was the H-5-Fuc,
whereas the target signal was either the H-2-Gal or the H-3-Fuc.
Fuc. However, this resonance is close to the unsuppressed
solvent with poorer baseline properties and therefore exhib-
its some inaccuracy. The only chemical difference between 5
and 8a is the methyl group attached to the cyclohexanediol
moiety in 8a. Since this methyl group shows by far the
smallest STD value and hence interacts only minimally or
not at all with E-selectin, the greater affinity of 8a com-
pared with 5 is necessarily a conformational issue.
Conclusion
A new family of GlcNAc mimetics was prepared and used
for the synthesis of new, high-affinity E-selectin antagonists.
For the determination of the affinities, two independent
assay formats were used; a competitive binding assay as well
as surface plasmon resonance experiments. As predicted by
MD simulations, equatorial alkyl groups in the 2-position in
1348
www.chemeurj.org
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 1342 – 1351

Pre-organization of the Core Structure of E-Selectin Antagonists
FULL PAPER
the carbocyclic GlcNAc mimetic restrict the conformational
flexibility of the core of the antagonist and entropically im-
prove binding affinities. The similarity of epitopes as deter-
mined by STD-NMR experiments indicate similar enthalpic
contributions of antagonists 5 and 8a to binding and there-
fore further support the hypothesis that the increased affini-
ty results from a gain in entropy. In agreement with earlier
observations by Thoma et al.[15], a further 3- to 5-fold en-
hancement of affinity could be accomplished when a benzo-
yl substituent was introduced in the 2’-position of the galac-
tose unit.
Overall, the affinity of the parent compound sLex (1) was
improved more than 660-fold (21e; KD =1.5 mm), predomi-
nantly by pre-organizing the Lex-core with a novel GlcNAc
mimetic, whereas the pharmacophores as present in sLex (1)
remained untouched.
Figure 6. STD spectra were measured for compounds 5 (a) and 8a (b) at
18.7 T (800 MHz). Reference STD spectra were as well obtained in
order to determine the binding epitopes displayed in Figure 7. The reso-
nances that were sufficiently free of overlap in both compounds 5 and 8a
are indicated. In addition, the methyl group in (b) from the cyclohexane
spacer of compound 8a is labeled.
Experimental Section
Synthesis: The entire synthetic part, including compound characterization
data, can be found in the Supporting Information.
Molecular dynamics: Molecular dynamics simulations were carried out in
MacroModel v9.5207 (Schrodinger Inc.) using the OPLS/2005 force field
and an implicit water solvation model.[22] High-quality parameters were
used for all force field terms. As a template structure, ligand 9 from the
crystal structure of E-selectin Lectin/EGF domains complexed with sLex
(1) (PDB ID: 1 g1t), was chosen. All MD simulations were carried out
following the standard protocol: 1) complete minimization of the mole-
cule to a gradient of 0.01 kcalmolꢀ1; 2) pre-equilibration phase: duration
100 ps, time step 0.1 fs, thermodynamic temperature 300 K; 3) production
phase: duration 1 ns, time step 0.1 fs, thermodynamic temperature 300 K.
Throughout the production phase the key torsion angle defined by exit
vectors from the GlcNAc [O(4)-C(4) and C(3)-O(3)] and the distance be-
tween carbon atom C(2) of d-Gal and C(5) of l-Fuc were monitored
with samples being taken every 0.02 ps (in total 50000 conformations
were saved). The sampled data were analyzed using statistical methods
and distribution plots were constructed (Figure 2 and Figure 3).
Cell-free E-selectin ligand binding assay: Wells of a microtiter plate
(plate 1, Falcon Probind) were coated with E-selectin/hIg chimera[12] by
incubation with 100 mL of the purified chimeric protein at a concentra-
tion of 200 ng/well in Tris pH 7.4 (50 mm), NaCl (0.15m), CaCl2 (Tris-
Ca2+; 2 mm). After 2 h, a 1:1 mixture (100 mL) of 1% BSA in Tris-Ca2+
and Stabilcoat (SurModics, Inc.) were added to each well and incubated
at 228C to block nonspecific binding. During this incubation, inhibitory
test compounds, diluted in Tris-Ca2+, 1% BSA were titrated by a twofold
serial dilution in a second U-shaped bottom low-bind microtiter plate
(plate 2, Costar, Inc.). An equal volume of a preformed complex of a bio-
tinylated sialyl Lewisa polyacrylamide polymer (sLea-PAA, Glycotech,
Inc.) and horseradish peroxidase-labeled streptavidin (KPL, Gaithers-
burg, MD) at 1 mgmLꢀ1 in Tris-Ca2+, 1% BSA was added to each well.
After 2 h at 228C, plate 1 was washed with Tris-Ca2+ and 100 mL/well
were transferred from plate 2 to plate 1. The binding reaction was al-
lowed to proceed for 2 h at 228C while rocking. Plate 1 was then washed
with Tris-Ca2+, and 100 mL of TMB substrate reagent (KPL, Gaithers-
burg, MD) was added to each well. After 3 min, the colorimetric reaction
was stopped by adding 100 mL/well of 1m aq. H3PO4 and the optical den-
sity was determined at 450 nm. The concentration of antagonist required
to inhibit binding by 50% was determined and is reported as the IC50
value.
Figure 7. STD determined binding epitopes of compounds 5 and 8a to E-
selectin/IgG. The contribution to the STD epitope of each hydrogen is
quantified by dividing the intensity of the corresponding resonance in the
STD spectrum with that of the STD reference spectrum. The result of
the division to obtain the STD value for each hydrogen is normalized to
H-2-Lac. STD values greater than 100 represent interactions to E-selec-
tin/IgG greater than that of the H-2-Lac to E-selectin/IgG. Further de-
tails concerning the sample preparation and measurement are described
in the experimental section. Excluding the methyl substituent in com-
pound 8a, the hydrogen signals in 5 and 8a, that for both compounds
were well-resolved and unambiguously assigned, were analyzed. These
eight hydrogens showed less than a 2% root mean square difference in
the STD values, indicating a high degree of similarity in the mode of
binding for 5 and 8a. The STD for the methyl substituent in compound 5
showed the smallest value and is in agreement with the finding that this
group is not significantly involved in binding to E-selectin/IgG.
Biacore assay: Surface plasmon resonance (Biacore) assays were per-
formed on a Biacore 3000 instrument (GE Healthcare) at 258C. The run-
ning buffer I was HBS-P (10 mm HEPES, pH 7.4, 150 mm NaCl, 0.002%
v/v surfactant P20). For all experiments, a polyclonal goat anti-human Fc
Chem. Eur. J. 2012, 18, 1342 – 1351
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
1349

B. Ernst et al.
antibody (Sigma) was first immobilized onto a research grade CM5
sensor chip through amine coupling using the manufacturerꢀs protocol
(Biacore). In a standard coupling procedure, a solution (20 mgmLꢀ1) of
the polyclonal antibody diluted in acetate buffer (10 mm NaOAc, pH 5)
was injected at 10 mLminꢀ1 for 10 min over all flow cells. Following poly-
clonal antibody coupling, the flow path was changed to exclude the refer-
ence flow cell containing only the immobilized polyclonal antibody. A so-
lution (50 mgmLꢀ1) of E-selectin/IgG diluted in acetate buffer (10 mm
NaOAc, pH 5.5) was injected at 5 mLminꢀ1 for 20 min over a single flow
cell, designated the active flow cell. The reference and active flow cells
were equilibrated at 5 mLminꢀ1 for 12 h in running buffer II (HBS-P with
20 mm CaCl2). Before injecting the antagonist, the system was equilibrat-
ed for 2 h in running buffer II supplemented with DMSO (2.5%, v/v).
Different concentrations of antagonists were prepared in running buffer
II supplemented DMSO (2.5% v/v). Antagonist dilutions were injected
using the KINJECT command with 60 s association and dissociation
times into the reference and the active flow cell. To compare the binding
affinity of different selectin antagonists, the sensorgrams from SPR ex-
periments were processed with Scrubber 2.0a (BioLogic Software). The
response observed in the reference flow cell was subtracted from the
active flow cell. The response at equilibrium was used to determine the
dissociation constant KD applying a simple steady state affinity 1:1 bind-
ing model. Because of the influence of DMSO on the binding signal, a
calibration was necessary. For that purpose different solutions of DMSO
were injected before each cycle of measurement. Signal corrections
based on the calibration solutions were directly performed during binding
evaluation. The surface was equilibrated overnight with HBS-EP buffer
(10 mm HEPES, pH 7.4, 150 mm NaCl, 3 mm EDTA, 0.002% v/v surfac-
tant P20).
buffer, but without E-selectin. The pulse sequence used for the STD ex-
periments was slightly modified from a published procedure.[54] A train of
40 cosine-modulated E-Burp-1 pulses (50 ms each) was used to saturate
E-selectin,[55] allowing the simultaneous irradiation of E-selectin at 0 and
8 ppm. This was achieved by placing the carrier frequency at 4 ppm
during the pulse train and by modulating the cosine waveform by
3200 Hz. The root-mean-square power of each sideband was 23.7 Hz. Ap-
plying the modulated pulse train provided a sensitivity of nearly twice
that of the mono-chromatic irradiation at 0 ppm. Following the non-se-
lective excitation pulse, a T1rho filter of 30 ms was applied with the
strength of 2.5 kHz to remove residual E-selectin magnetization. Follow-
ing the T1rho filter, a DPFGSE water suppression sequence was append-
ed,[56] and the remaining signal removed through digital filtering. For the
STD experiment, 1792 scans were measured with a 2 s delay and 0.7 s
measurement of the fid. For the STD reference experiment, 128 scans
were measured with a 2 s prescan delay and 0.7 s measurement of the fid.
The NMR data were analyzed using XWINNMR 3.0. The spectra were
apodized with an exponential decay function with 2 Hz line broadening.
Acknowledgements
The authors gratefully acknowledge the financial support by the Swiss
National Science Foundation (grant no. 200020–103875/1). They also
thank Mr. Werner Kirsch, Microanalytical Services of the Organic Insti-
tute of the University of Basel, for performing the microanalysis and
Prof. Dr. Michael Przybylski, Department of Chemistry, University of
Konstanz, Germany for high resolution MS. Finally, we are thankful to
Prof. Dr. Stephan Grzesiek, Biocenter of the University of Basel, Swit-
zerland for the measurement time at 800 MHz for the STD-NMR.
NMR experiments: All ROESY experiments were carried out at 300 K
on a Bruker DMX-500 (500 MHz) spectrometer, equipped with Z-gradi-
ent SEI probe. The samples consisted of approximately 5 mg of 5 or 8a
solvated in of 99.8% D2O (200 mL) (Armar Chemicals). Chemical shifts
were referenced to the residual water resonance (d=4.70 ppm). The
doubly-selective homonuclear Hartmann-Hahn scheme[51] was used to se-
lectively transfer magnetization from H-6-Fuc to H-5-Fuc. The selective
excitation of H-5-Fuc allowed an accurate quantification of this reso-
nance by avoiding the excitation of residual H2O. To remove any remain-
ing magnetization from H-6-Fuc, a selective gradient echo at the frequen-
cy of H-5-Fuc was applied. A 200 ms REBURP 1800 refocusing pulse[52]
was applied to the H-5-Fuc resonance. The REBURP pulse was sand-
wiched by a pair of Gaussian shaped gradients of 1 ms each with an am-
plitude of 20 G cmꢀ1. The jump-symmetrized CW-ROESY variation of
the ROESY sequence was used in all experiments to minimize TOCSY
artifacts.[53] During the ROESY period, the transmitter frequency was
shifted up or downfield during the first or second half of the mixing-time,
respectively. The high-field spin lock was applied at d=4.9 ppm and the
low-field at d=0.9 ppm. The spin lock was a rectangular pulse with a
2 kHz-amplitude. To record a build-up curve of the ROE transfer, for
each compound a sequence of 10 experiments with increasing durations
of the spin lock from 50 ms to 500 ms in steps of 50 ms was recorded. Fol-
lowing the application of the spin lock, the transmitter was returned to
2.9 ppm and the fid measured using 4096 complex points to sample a
bandwidth of d=7 ppm. To achieve a high signal-to-noise required for
the accurate distance evaluations of Figure 5c, 1024 scans were measured
for each mixing time. The NMR data were analyzed using XWINNMR
3.0. The spectra were apodized with an exponential decay function with
2 Hz line broadening. To determine the internuclear distances, the NOEs
of the target proton were normalized to the intensity of the diagonal
peak of H-5-Fuc.[48] Plotting these normalized intensities against the
mixing time results in a linear function for each pair of protons (Fig-
ure 5c). The distances rij were then calculated from the slopes s of the
linear regression according to rij =rref (sref/sij)1/6, in which rref =2.60 ꢂ is
the assumed distance for H-5-Fuc—H-3-Fuc.
[1] Physiology and Pathophysiology of Leukocyte Adhesion (Eds.: D. N.
Granger, G. W. Schmid-Schçnbein), Oxford University Press, 1995.
[2] M. S. Mulligan, J. Varani, M. K. Dame, C. L. Lane, C. W. Smith,
D. C. Anderson, P. A. Ward, J. Clin. Invest. 1991, 88, 1396–1406.
[3] K. Ley, D. C. Bullard, M. L. Arbones, R. Bosse, D. Vestweber, T. F.
Tedder, A. L. Beaudet, J. Exp. Med. 1995, 181, 669–675.
[4] K. E. Norman, G. P. Anderson, H. C. Kolb, K. Ley, B. Ernst, Blood
1998, 91, 475–483.
[5] R. P. McEver, K. L. Moore, R. D. Cummings, J. Biol. Chem. 1995,
270, 11025–11028.
[6] T. F. Tedder, D. A. Steeber, A. Chen, P. Engel, FASEB J. 1995, 9,
866–873.
[7] S. A. Mousa, Drugs Future 1996, 21, 283–289.
[8] B. Ernst, J. L. Magnani, Nat. Rev. Drug Discovery 2009, 8, 661–677.
[9] a) E. L. Berg, M. K. Robinson, O. Mansson, E. C. Butcher, J. L.
Magnani, J. Biol. Chem. 1991, 266, 14869–14872; b) M. J. Polley,
M. L. Phillips, E. Wayner, E. Nudelman, A. K. Singhal, S. Hako-
mori, J. C. Paulson, Proc. Natl. Acad. Sci. USA 1991, 88, 6224–6228;
c) M. L. Phillips, E. Nudelman, F. C. Gaeta, M. Perez, A. K. Singhal,
S. Hakomori, J. C. Paulson, Science 1990, 250, 1130–1132.
[10] Y.-C. Lin, C. W. Hummel, D.-H. Huang, Y. Ichikawa, K. C. Nico-
laou, C.-H. Wong, J. Am. Chem. Soc. 1992, 114, 5452–5454.
[11] M. Rinnbauer, B. Ernst, B. Wagner, J. Magnani, A. J. Benie, T.
Peters, Glycobiology 2003, 13, 435–443.
[12] L. Scheffler, B. Ernst, A. Katopodis, J. L. Magnani, W. T. Wang, R.
Weisemann, T. Peters, Angew. Chem. 1995, 107, 2034–2037; Angew.
Chem. Int. Ed. Engl. 1995, 34, 1841–1844.
[13] a) H. C. Kolb, B. Ernst, Chem. Eur. J. 1997, 3, 1571–1578; b) H. C.
Kolb, B. Ernst, Pure Appl. Chem. 1997, 69, 1879–1884.
[14] G. Thoma, J. L. Magnani, J. T. Patton, B. Ernst, W. Jahnke, Angew.
Chem. 2001, 113, 1995–1999; Angew. Chem. Int. Ed. 2001, 40, 1941–
1945.
All STD experiments were carried out at 300 K on a Bruker DRX-800
(800 MHz) spectrometer, equipped with a Z-gradient cryoprobe. The
samples contained 5 or 8a (1.5 mm) and E-selectin/IgG (20 mm) in buffer
(99.8% D2O, 10 mm CaCl2, maintained at pH 7.5 with 25 mm d11-TRIS).
Control samples of antagonist 5 and 8a were prepared in the same
[15] G. Thoma, R. Bꢃnteli, W. Jahnke, J. L. Magnani, J. T. Patton,
Angew. Chem. 2001, 113, 3756–3759.
[16] a) O. Schwardt, H. C. Kolb, B. Ernst, in The Organic Chemistry of
Sugars (Eds.: P. Fꢄgedi, D. E. Levy), Marcel Dekker, New York,
1350
www.chemeurj.org
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Chem. Eur. J. 2012, 18, 1342 – 1351

Pre-organization of the Core Structure of E-Selectin Antagonists
FULL PAPER
2005; b) N. Kaila, B. E. Thomas, Med. Res. Rev. 2002, 22, 566–601;
c) E. E. Simanek, G. J. McGarvey, J. A. Jablonowski, C.-H. Wong,
Chem. Rev. 1998, 98, 833–862.
[37] a) B. Haveaux, A. Dekoker, M. Rens, A. R. Sidani, J. Toye, L.
Ghosez, Org. Synth. 1980, 59, 26–34; b) L. Ghosez, Angew. Chem.
1972, 84, 901–902; Angew. Chem. Int. Ed. Engl. 1972, 11, 852–853.
[38] G. Thoma, J. L. Magnani, R. Oehrlein, B. Ernst, F. Schwarzenbach,
R. O. Duthaler, J. Am. Chem. Soc. 1997, 119, 7414–7415.
[39] a) R. L. Rich, D. G. Myszka, Curr. Opin. Biotechnol. 2000, 11, 54–
61; b) T. A. Morton, D. G. Myszka, Methods Enzymol. 1998, 295,
268–294; c) K. Nagata, H. Handa, Real-Time Analysis of Biomolec-
ular Interactions: Applications of Biacore, Springer-Verlag, Berlin,
2000.
[40] G. Weitz-Schmidt, D. Stokmaier, G. Scheel, N. E. Nifant’ev, A. B.
Tuzikov, N. V. Bovin, Anal. Biochem. 1996, 238, 184–190, and refer-
ences cited therein.
[41] J. Chang, J. T. Patton, A. Sarkar, B. Ernst, J. L. Magnani, P. S. Fren-
ette, Blood 2010, 116, 1779–1786.
[42] R. U. Lemieux, M.-H. Du, U. Spohr, J. Am. Chem. Soc. 1994, 116,
9803–9804.
[43] P. R. Connelly, R. A. Aldape, F. J. Bruzzese, S. P. Chambers, M. J.
Fitzgibbon, M. A. Fleming, S. Itoh, D. J. Livingston, M. A. Navia,
J. A. Thomson, K. P. Wilson, Proc. Natl. Acad. Sci. USA 1994, 91,
1964–1968.
[44] M. K. Wild, M.-C. Huang, U. Schulze-Horsel, P. A. van der Merwe,
D. Vestweber, J. Biol. Chem. 2001, 276, 31602–31612.
[45] E. J. Toone, Curr. Opin. Struct. Biol. 1994, 4, 719–728.
[46] M. Mayer, B. Meyer, J. Am. Chem. Soc. 2001, 123, 6108–6117.
[47] J. Angulo, C. Rademacher, T. Biet, A. J. Benie, A. Blume, H. Peters,
M. Palcic, F. Parra, T. Peters, Methods Enzymol. 2006, 416, 12–30.
[48] A. A. Bothner-By, R. L. Stephens, J. Lee, C. D. Warren, R. W. Jean-
loz, J. Am. Chem. Soc. 1984, 106, 811–813.
[17] W. S. Somers, J. Tang, G. D. Shaw, R. T. Camphausen, Cell 2000, 103,
467–479.
[18] E. Black, J. Breed, A. L. Breeze, K. Embrey, R. Garcia, T. W. Gero,
L. Godfrey, P. W. Kenny, A. D. Morley, C. A. Mishull, A. D. Panni-
fer, J. Read, A. Rees, D. J. Russell, D. Toader, J. Tucker, Bioorg.
Med. Chem. Lett. 2005, 15, 2503–2507.
[19] K. Ikegashira, T. Oka, S. Hirashima, S. Noji, H. Yamanaka, Y. Hara,
T. Adachi, J. I. Tsuruha, S. Doi, Y. Hase, T. Noguchi, I. Ando, N.
Ogura, S. Ikeda, H. Hashimoto, J. Med. Chem. 2006, 49, 6950–6953.
[20] For the synthesis of 2 see the Supporting Information.
[21] J. C. Prodger, M. J. Bamford, M. I. Bird, P. M. Gore, D. S. Holmes,
R. Priest, V. Saez, Bioorg. Med. Chem. 1996, 4, 793–801.
[22] MacroModel, Version 9.5, Schrçdinger, LLC, New York, NY, 2008.
[23] C. J. Kowalski, A. E. Weber, K. W. Fields, J. Org. Chem. 1982, 47,
5088–5093.
[24] N. Holub, J. Neidhçfer, S. Blechert, Org. Lett. 2005, 7, 1227–1229.
[25] a) B. A. McKittrick, B. Ganem, Tetrahedron Lett. 1985, 26, 4895–
4898; b) C. G. Chavdarian, C. H. Heathcock, Synth. Commun. 1976,
6, 277.
[26] a) A. Alexakis, D. Jachiet, J. F. Normant, Tetrahedron 1986, 42,
5607–5619; b) A. Ghribi, A. Alexakis, J. F. Normant, Tetrahedron
Lett. 1984, 25, 3075–3078.
[27] H. Lohn, Carbohydr. Res. 1985, 139, 105–113.
[28] a) R. U. Lemieux, J. Hayami, Can. J. Chem. 1965, 43, 2162–2173;
b) R. U. Lemieux, K. B. Hendriks, R. V. Stick, K. James, J. Am.
Chem. Soc. 1975, 97, 4056–4062.
[29] T. M. Trnka, R. H. Grubbs, Acc. Chem. Res. 2001, 34, 18–29.
[30] B. Ernst, B. Wagner, G. Baisch, A. Katopodis, T. Winkler, R. Oehr-
lein, Can. J. Chem. 2000, 78, 892–904.
[49] S. Macura, R. R. Ernst, Mol. Phys. 1980, 41, 95–117.
[50] M. Mayer, B. Meyer, Angew. Chem. 1999, 111, 1902–1906; Angew.
Chem. Int. Ed. 1999, 38, 1784–1788.
[31] O. Mitsunobu, Synthesis 1981, 1–28.
[32] S. Sato, M. Mori, Y. Ito, T. Ogawa, Carbohydr. Res. 1986, 155, C6–
C10.
[33] G. Cerichelli, C. Grande, L. Luchetti, G. Mancini, J. Org. Chem.
1991, 56, 3025–3030.
[51] B. Boulat, R. Konrat, I. Burghardt, G. Bodenhausen, J. Am. Chem.
Soc. 1992, 114, 5412–5414.
[52] H. Geen, R. Freeman, J. Magn. Reson. 1991, 93, 93–141.
[53] J. Schleucher, J. Quant, S. J. Glaser, C. Griesinger, J. Magn. Reson.
1995, 112, 144–151.
[34] a) H. J. Gais, K. L. Lukas, Angew. Chem. 1984, 96, 140–141; Angew.
Chem. Int. Ed. Engl. 1984, 23, 142–143; b) S. M. Roberts, Biocata-
lysts for fine chemicals synthesis, John Wiley and Sons, Chichester,
1999.
[54] B. Meyer, T. Peters, Angew. Chem. 2003, 115, 890–918; Angew.
Chem. Int. Ed. 2003, 42, 864–890.
[55] B. Cutting, S. V. Shelke, Z. Dragic, B. Wagner, H. Gathje, S. Kelm,
B. Ernst, Magn. Reson. Chem. 2007, 45, 720–724.
[35] a) D. H. R. Barton, D. Crich, W. B. Motherwell, J. Chem. Soc. Chem.
Commun. 1983, 939–941; b) D. H. R. Barton, D. Crich, W. B. Moth-
erwell, Tetrahedron Lett. 1983, 24, 4979–4982.
[56] T.-L. Hwang, A. J. Shaka, J. Magn. Reson. Ser. A 1995, 112, 275–
279.
Received: September 14, 2011
[36] J. A. Marshall, S. Xie, J. Org. Chem. 1995, 60, 7230–7237.
Published online: December 30, 2011
Chem. Eur. J. 2012, 18, 1342 – 1351
ꢁ 2012 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chemeurj.org
1351