Quinic acid derivatives as sialyl Lewisx-mimicking selectin inhibitors: Design, synthesis, and crystal structure in complex with E-selectin

Article abstract of DOI:10.1021/jm050049l

A search for noncarbohydrate sLe^x mimics led to the development of quinic acid derivatives as selectin inhibitors. At Wyeth we solved the first cocrystal structure of a small molecule, quinic acid, with E-selectin. In the cocomplex two hydroxyls of quinic acid mimic the calcium-bound fucose of the tetrasaccharide sLe^x. The X-ray structure, together with structure based computational methods, was used to design quinic acid based libraries that were synthesized and evaluated for their ability to block the interaction of sLex with P-selectin. A large number of analogues were prepared using solution-phase parallel synthesis. Selected compounds showed decrease in leukocyte rolling in the IVM mouse model. Compound 2 inhibited neutrophil influx in the murine TIP model and demonstrated good plasma exposure.

Full text of DOI:10.1021/jm050049l

4346
J. Med. Chem. 2005, 48, 4346-4357
Quinic Acid Derivatives as Sialyl Lewisx-Mimicking Selectin Inhibitors: Design,
Synthesis, and Crystal Structure in Complex with E-Selectin
Neelu Kaila,*,† William S. Somers,† Bert E. Thomas,# Paresh Thakker,† Kristin Janz,† Silvano DeBernardo,†
Steve Tam,† William J. Moore,† Ruiyang Yang,§ Wojciech Wrona,§ Patricia W. Bedard,† Deidre Crommie,†
James C. Keith, Jr.,† Desiree H. H. Tsao,† Juan C. Alvarez,† Heyu Ni,| Erik Marchese,† John T. Patton,‡
John L. Magnani,‡ and Raymond T. Camphausen
Chemical & Screening Sciences and Cardiovascular and Metabolic Disease Research, Wyeth, 200 Cambridge Park Drive,
Cambridge, Massachusetts 02140, Arqule, 19 Presidential Way, Woburn, Massachusetts 01801, GlycoTech Corporation,
14915 Broschart Road Ste. 200, Rockville, Maryland 20850, and CBR Institute for Biomedical Research, Harvard Medical
School, Boston, Massachusetts 02115
Received January 17, 2005
A search for noncarbohydrate sLex mimics led to the development of quinic acid derivatives as
selectin inhibitors. At Wyeth we solved the first cocrystal structure of a small molecule, quinic
acid, with E-selectin. In the cocomplex two hydroxyls of quinic acid mimic the calcium-bound
fucose of the tetrasaccharide sLex. The X-ray structure, together with structure based
computational methods, was used to design quinic acid based libraries that were synthesized
and evaluated for their ability to block the interaction of sLex with P-selectin. A large number
of analogues were prepared using solution-phase parallel synthesis. Selected compounds showed
decrease in leukocyte rolling in the IVM mouse model. Compound 2 inhibited neutrophil influx
in the murine TIP model and demonstrated good plasma exposure.
Background
and 4-hydroxyls of fucose make interactions with pro-
tein side chains that bind the calcium. Some of the
hydroxyls of galactose and sialic acid also interact with
the selectins, as does the required acid group of sialic
acid, which binds to an arginine (E-selectin) or a serine
(P-selectin) side chain.
There are numerous reports of selectin inhibitors in
the literature. One popular strategy has been the design
of sLex mimetics with better pharmacokinetic properties
than the tetrasaccharide sLex. This has been ac-
complished by replacing some or all of the sugars with
other moieties such that the key interactions are
conserved.5 Our search for a noncarbohydrate mimic of
fucose led to the identification of quinic acid as a
replacement for fucose. In this paper, we describe the
design, synthesis and biological evaluation of quinic acid
derivatives as sLex mimetics through the use of X-ray
crystallography and molecular modeling. This is the first
report of the crystal structure of E-selectin in complex
with a small molecule.
Excessive recruitment of leukocytes from the circula-
tion can result in pathological conditions such as stroke,
reperfusion injury, and cardiovascular and allergic
diseases. Leukocyte migration into underlying tissues
is an orderly process mediated initially by low-affinity,
transient interactions of P- and E-selectin cell adhesion
molecules expressed on the vascular endothelium with
tetrasaccharide sialyl Lewis x (sLex)-containing ligands
on the leukocyte cell surface. Leukocytes thus slowed
to a roll by interactions with the selectins subsequently
arrest as a consequence of high-affinity integrin-based
interactions and then extravasate into the interstitium
where inflammatory reactions manifest. Constituting
the earliest adhesion event in the leukocyte recruitment
cascade, central to virtually all inflammatory processes,
and of low-affinity relative to other cell adhesion events,
selection-sLex interactions are appealing drug discovery
targets.1-10
Recently, we published the crystal structures of both
E- and P-selectin (lectin and EGF domains) in complex
with sLex, and P-selectin LE cocomplexed with the
PSGL-1 N-terminus.11 These structures have clearly
defined the interactions between the selectins and sLex
and are consistent with information gleaned from NMR
experiments and binding studies of selectively modified
sLex.5 Sialic acid, L-fucose and galactose are vital for
the calcium-mediated binding of sLex to selectins. In
addition to binding directly to the calcium ion, the 3-
Structural Studies
A number of small molecules that potentially mimic
the fucose residue of sLex were evaluated. Quinic acid
was chosen as the preferred template in anticipation
that this noncarbohydrate would prove more stable than
a carbohydrate-based inhibitor; at the time, there were
no literature references to quinic acid derivatives having
been evaluated as selectin inhibitors. Derivatization was
synthetically straightforward. It was expected that the
4- and 5-hydroxyls of quinic acid (one in an equatorial
position and one in an axial position) would coordinate
the calcium in a fashion similar to the 3- and 4-hy-
droxyls of fucose in sLex. To confirm this hypothesis,
we soaked quinic acid into preformed crystals of a
human E-selectin construct consisting of the lectin and
* To whom correspondence should be addressed. E-mail:
nkaila@wyeth.com.
† Wyeth.
§ Arqule.
| The CBR Institute for Biomedical Research, Inc.
‡ GlycoTech Corporation, MD.
# Present address: Guilford Pharmaceuticals.
Present address: Compound Therapeutics, Inc.
10.1021/jm050049l CCC: $30.25 © 2005 American Chemical Society
Published on Web 05/25/2005

Sialyl Lewisx-Mimicking Selectin Inhibitors
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 13 4347
EGF domains (E-LE). The protein structure in the
resulting complex is essentially identical to those previ-
ously described11,12 (Figure 1). However, close examina-
tion of the complex structure revealed the unexpected
result that the 3- and 4-hydroxyls of quinic acid (both
in an equatorial position) bind the calcium ion, while
the 5-hydroxyl forms a hydrogen bond with the side
chain of Glu107. The consequences of the 3- and
4-hydroxyls binding to the calcium ion are significant.
The geometry of the sLex molecule requires that the
fucose ring be perpendicular to the surface of the protein
when it is bound to the calcium ion. Quinic acid, which
is not constrained as part of a tetrasaccharide, binds to
the calcium ion with its ring parallel to the surface of
the protein. This significantly changes the directional
vector of substituents that are attached to the 1-position
of quinic acid.
Our focus was to find a P-selectin inhibitor, since it
is expressed on the endothelium within minutes after
stimulation and plays a key role in the early adhesion
cascade.13 Attempts to obtain a cocrystal structure of
quinic acid with P-selectin failed; the high concentra-
tions of ligand used in the soaking experiments caused
loss of diffraction with the P-selectin crystals.14 Since
E- and P-selectin bind to sLex in a similar fashion, we
used the cocrystal of E-selectin-quinic acid to generate
a model of the P-selectin-quinic acid complex. This
model was also used to generate virtual libraries
analoging the tertiary alcohol and carboxylic acid func-
tionalities to improve potencies.
Figure 1. The structure of quinic acid shown with purple
carbon atoms bound to E-selectin. The bound calcium ion is
shown in orange and an ordered water in red. The figure was
made using PYMOL.25
of multimeric complexes of an Fc construct of the
P-selectin ligand (PSGL-1) and a conjugated antibody.
The formation of these complexes is quite variable,
making it difficult to evaluate small gains in activity.
This problem is avoided in the Biacore assay, which
looks at monomeric binding of P-selectin to PSGL-1.
Compound 1 showed marginal activity at 10 mM in the
Biacore assay. All subsequent compounds were tested
in the Biacore assay.15
Next, we modeled simple modifications of 1 that could
be made through amide formation or a Suzuki reaction,
adding an additional aryl group. We hoped to obtain the
same kind of increase in potency and change in physical
properties that had resulted from adding the first aryl
group to quinic acid. Library A and library B were the
first two libraries made (Figure 2). Library A utilized
an amide group as a spacer between the two aryl groups,
while library B investigated the direct addition of an
additional aryl group to 1. Each library consisted of
three subgroups, with the new linker ortho, meta, or
para to quinic acid. The parent structures were each
modeled onto the P-selectin surface using a Monte Carlo
algorithm in QXP.16 Using the lowest energy structure
of each parent as a starting point, a virtual library
corresponding to the appropriate adducts was generated
in the presence of P-selectin using QXP. This method
of reagent selection was utilized for every library except
library H.
Initial modeling efforts were directed toward explor-
ing simple modifications of the 1-hydroxy group of quinic
acid. One particularly interesting modification was the
addition of a benzyl group (1). Modeling suggested that
Modeling studies for Library B showed that in the
ortho position, the new aryl ring does not make any
significant interactions with the protein surface, but
does orient possible substitution positions on the ring
toward the protein surface where new interactions could
be made. In the meta position, the new aryl ring lies
along the surface of the protein, interacting with the
side chain of Tyr-48. Possible substitution points on this
new ring are directed along the protein surface.
Several potent compounds were identified out of
library B. On the basis of this result, we decided to
replace the phenyl ring attached to quinic acid with a
heterocycle, giving library C. The attachment point for
substitutents presents a similar though slightly differ-
ent vector than the meta position on an aryl group.
Thus, molecular modeling suggested that substituents
a benzyl group would cause very little change in the
orientation of the quinic acid group with respect to
P-selectin but would result in a stacking interaction
between the benzyl group and the side chain of Tyr-94.
Compound 1 was synthesized and showed 61% inhibi-
tion at 10 mM in the P-selectin ELISA assay. Quinic
acid shows marginal activity in this assay at 10 mM.
At this time the Biacore assay was developed which
proved to be more stringent, robust, and reproducible
(see details for the Biacore assay in the results section).
The ELISA relied on amplification of signal by formation

4348 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 13
Kaila et al.
Figure 3. A proposed binding mode of compound 3 to
P-selectin is shown. The proximal residues of the protein are
shown in orange with a Connelly surface. The bound calcium
is shown in pink. Tyr94 and Tyr48 are labeled to highlight
the interaction between the side chains of these residues and
the aryl groups of 3.
Scheme 1
Figure 2. Quinic acid libraries.
would have slightly altered interactions with the protein
relative to similar compounds from library B.
substituent (as in library G), but incorporated a car-
boxylic acid group. Diverse reagents were chosen for this
library.
Library D and E were based on further modification
of compound 2, a hit from library B. Modeling predicted
that the trifluoromethyl group from biphenyl 2 was
oriented in the direction of a small hydrophobic tunnel
(or pseudo pocket) on the surface of P-selectin. It was
proposed that the alkyl group from the amides might
interact with this hydrophobic tunnel.
Library F is based on 3, a potent hit from library B.
When 3 was modeled onto the P-selectin surface we
found that the first aryl ring stacks onto Tyr94, while
the second aryl group makes a T-interaction with the
side chain of Tyr48 (Figure 3). All three substitution
sites on the second aryl ring are directed along the
surface of the protein, allowing for the opportunity to
find the appropriate substituent that makes additional
productive interactions.
Library G explored amide substitution at the car-
boxylic acid of quinic acid. While the directional vector
with respect to the protein surface is not as favorable
for a substituent attached to the carboxylic acid as for
one attached to the hydroxyl we still decided to explore
this option. We selected amines with substituents that
could reach back and make interactions with the surface
of the protein and then used previously described
molecular modeling experiments to choose reagents out
of this set. Results from library G showed that a
carboxyl functionality was essential for activity. Thus,
we designed library H, which kept an amide linked
Chemistry
Libraries A-F, the Tertiary Alcohol-Substituted
Libraries. To rapidly prepare libraries of quinic acid
in a parallel format, we were looking for protecting
groups, which could be removed in such a manner as to
eliminate the need for one-by-one workup of each
reaction. For the tertiary alcohol libraries, we used the
carbonate-lactone functionality to protect the secondary
alcohols and carboxylic acid group (Scheme 1). Com-
mercially available quinic acid was converted to the
corresponding lactone 4 in refluxing DMF/benzene with
catalytic p-toluenesulfonic acid.17 Lactone 4 was then
treated with triphosgene in pyridine to form lactone
carbonate 5, which is the common intermediate for the
synthesis of libraries A-F. Treatment with lithium
hydroxide and quenching with acidic resin can remove
these groups in one step.
Library A (Scheme 2): Intermediate 5 was alkylated
with ortho, meta, or para substituted bromomethylben-
zoic acid tert-butyl ester to form ethers. Removal of the
tert-butyl group with TFA provided the free acids, 6.
Due to the wide range of amines selected for this library,
there was no universal set of conditions effective for
every amide coupling. However, a large percentage of
the compounds could be prepared through the interme-

Sialyl Lewisx-Mimicking Selectin Inhibitors
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 13 4349
Scheme 2
Scheme 3
Scheme 4
diacy of acid chloride 7. The amide products were
hydrolyzed with LiOH and quenched with acidic resin
to give library A, which contained 256 compounds.
Library B (Scheme 3A): Intermediate 5 was alkylated
with ortho, meta, or para substituted bromobenzyl
bromide or 1-bromo-2-bromomethylnaphthalene to give
intermediates 8a-d respectively. Suzuki coupling of
intermediates 8a-d was a challenging reaction for
parallel synthesis. It was difficult to find coupling
conditions that worked for all four cores 8a-d. The
conditions shown in Scheme 3 worked reasonably well;
other conditions tried were Pd(OAc)2/CsCO3 and Pd2-
(dba)3/PtBu3/CsCO3. The four cores, 8a-d, were crossed
with several boronic acids; hydrolyzed with LiOH, and
quenched with acidic resin to give library B, which
contained 140 compounds. During the purification of
some of these compounds, approximately 5-20% of
epimerized product 9 was isolated. Further studies
showed that the CsF in the Suzuki reaction was causing
this epimerization. Simply refluxing the core 8c in CH3-
CN/water in the presence of CsF, followed by LiOH
hydrolysis, gave triol 10a that contained about 20% of
epimerized product 10b. For the resynthesis of biologi-
cally active compounds from library B, modified Scheme
3B was used in order to avoid 9-like byproducts.
Library C (Scheme 4): The common intermediate 11
was easily synthesized by alkylation of 5 with propargyl
bromide. Reaction of 11 with nitrile oxides 12 gave

4350 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 13
Kaila et al.
Scheme 5
Scheme 8
Scheme 6
The aldehyde group was then reduced to the corre-
sponding alcohol, which was converted to the desired
bromide 15 with triphenylphophine dibromide. Core 16
was coupled with vinylzinc (17) and alkynylzinc (18)
reagents to give vinyl- and alkynyl-substituted biphenyl
quinic acids, respectively.
Scheme 7
Commercially available Grignard reagents were used
to obtain the vinyl zinc reagents. Grignard reagents
were first reacted with ZnCl2 to form the vinylzinc,
followed by Pd(PPh3)4 catalyzed coupling with core 16.
Hydrolysis with LiOH gave the final vinyl-substituted
products (Library F1, nine compounds).
Commercially available terminal alkynes were used
to obtain the alkynylzinc reagents. Terminal alkynes
were first treated with n-butyllithium. The alkynyl-
lithium reagents were then reacted with zinc chloride
to generate alkynylzinc reagents in situ, which were
coupled to core 16 using the previously established
conditions for vinyl substituted quinic acids. Deprotec-
tion with LiOH gave the final alkynyl substituted
products (Library F2, 4 compounds).
Libraries G and H, the C-1 Carboxylic Acid
Substituted Libraries. In these libraries the alcohol
groups were protected as acetates. Library G (Scheme
8): Quinic acid was treated with acetic anhydride
/pyridine to give the tetraacetyl intermediate 19. The
acid group was then activated by conversion to the
HOBt ester 20. Compound 20 was stable for long periods
of time and could be coupled with a variety of amines.
Hydrolysis under basic conditions gave library G (70
compounds).
Library H (Scheme 8): Coupling of intermediate 20
with methyl 3,5-diaminobenzoate gave the core for
library H, the reaction of core 21 with acid chlorides
worked well using PVP (polyvinylpyridine) as the
catalyst. Final deprotection with LiOH also hydrolyzed
the methyl ester to give a free acid in library H (42
compounds).
isoxazoles which, upon deprotection with LiOH, resulted
in library C. The required nitrile oxides 12 were formed
in situ by chlorination of oximes with N-chlorosuccin-
imide followed by dehydrohalogenation with triethyl-
amine. Library C contained 21 compounds.
Library D (Scheme 5): Alkylation of intermediate 5
with 4-(2-nitrophenyl)benzyl bromide, followed by re-
duction, gave the aniline core 13. Coupling of a variety
of acid chlorides with core 13, followed by hydrolysis
with LiOH, resulted in library D (37 compounds). It was
found that LiOH hydrolysis gave the methyl ester as a
byproduct when carried out in MeOH/ water solution;
therefore, the hydrolysis for this library was performed
in THF/water.
Library E (Scheme 6): Acid chloride core 14 was
prepared from intermediate 5 in three steps. Alkylation
with 4-(2-tert-butoxycarbonylphenyl)benzylbromide was
followed by TFA removal of the tert-butyl group, then
conversion to the acid chloride using oxalyl chloride.
Coupling of core 14 with several amines followed by
LiOH deprotection gave 41 new compounds.
Library F (Scheme 7): The common core for this
library, 16, was prepared by alkylation of intermediate
5 with biphenyl benzylbromides 15. Compounds 15 were
prepared in three steps starting with the Suzuki cou-
pling of a bromoiodobenzene with a formylboronic acid.
Coupling of 21 with sulfonyl chlorides was sluggish.
We could not find optimal conditions for the synthesis
of sulfonamides, and therefore this approach was aban-
doned.
Results and Discussion
Compounds were evaluated for inhibition of P-selectin
in a Biacore assay. This is an equilibrium assay, which
measures interactions between monomeric, soluble
P-selectin, and immobilized PSGL-1 under flow condi-

Sialyl Lewisx-Mimicking Selectin Inhibitors
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 13 4351
tions. Compounds from our libraries with IC50s from 1
to 10 mM in the Biacore assay were considered reason-
able hits, as the natural ligand sLex has an IC50 of 15
mM in this assay.18
in activity. Nonetheless, a 500 µM hit in the rolling
assay is exciting and worthy of evaluation in vivo.
As mentioned above, the cesium fluoride used during
the synthesis of this library caused hydrolysis of the
carbonate protecting group and partial epimerization
of the C-3 hydroxyl group. The diastereomers for two
representative examples were separated and tested and
were found to have similar activities (2 versus 32 and
22 versus 33).
Compounds with IC50 e10 mM in the Biacore assay
were further evaluated in our secondary assays, an
NMR assay, and a cell-based flow assay. In the NMR
assay, transfer NOE (trNOE) experiments were per-
formed using P-selectin. The trNOE experiment is a
well-established technique for studying protein-ligand
interactions, including the determination of bound
ligand structures.19 In the trNOE experiment, the
protein-ligand system must be in rapid exchange for
the bound small molecule to be easily detectable. In this
case, strong negative trNOEs are observed. Compounds
that do not bind to the protein display a weak positive
NOE.
The cell-based in vitro flow assay measures selectin-
mediated rolling of human neutrophils on stimulated
human umbilical vein endothelial cells (HUVEC) in a
parallel plate flow chamber. Digital image analysis of
rolling and arrested cells is performed to quantitate
inhibition.
Addition of an extra aryl group to compound 1 using
an amide linker (Library A) did not result in any hits.
Many of these compounds had additional atoms between
the amide and the added aryl ring, in the form of small
alkyl chains. While the flexibility inherent in these
molecules allows for them to have a conformation that
fits the surface of the protein well and to produce a good
interaction energy with the protein from a molecular
force field, it is likely that the additional negative
entropy of binding counteracts any positive enthalpic
considerations.
In contrast to library A, library B produced several
hits. Adding the aryl ring directly to 1 was a success. It
should be noted that palladium contamination of the
products resulted in false positives. When the library
was resynthesized, ensuring that the compounds were
free of palladium contamination, eight compounds (2
and 3, 22-27) were identified with IC50 < 10mM in the
Biacore assay. Unfortunately, 3 (IC50 ) 1 mM) could
not be further evaluated due to the difficulty in syn-
thesizing this compound on a larger scale. All efforts
toward its scaled up synthesis resulted in polymeriza-
tion of the vinyl boronic acid. Seven compounds, 2 and
22-27, were evaluated in the NMR assay. Five com-
pounds (2, 22-25) showed positive binding by transfer
NOE with P-selectin. Addition of small amounts of
protein to 26 and 27 results in a significant broadening
of the tNOE peaks, much more than is observed for
compounds of similar Biacore potency. This suggests
that these compounds bind to P-selectin with high
stoichiometry, which is an unfavorable property for a
drug. From our data it is difficult to speculate how many
additional potential binding sites might exist on the
P-selectin surface.
The isoxazole compounds from library C did not show
significant activity. No further effort was invested in
this class of compounds.
Library D produced one hit, 28. This compound was
very potent in the cell based flow assay. Unfortunately,
28 showed nonstoichiometric binding to P-selectin in the
NMR assay. Library E, the reverse amide library of D,
produced no hits.
One hit, 29, was obtained from library F1. 29 showed
strong positive tNOE with P-selectin and had an IC50
of 1 mM in the cell rolling assay.
Library G produced no hits. The most potent com-
pound in this library showed only 20% inhibition at 10
mM. Two hits were obtained from library H, 30 and 31.
They were well behaved in both the NMR and cell
rolling assay. Replacement of the benzothiophene ring
in 31 with a benzofuran or thiophene ring resulted in
decreased activity.
The hits from all the libraries are summarized in
Table 1. There were seven compounds that were well
behaved in all three assays. These seven compounds
were evaluated for iv PK in the rat (Table 2). Several
of the quinic acid derivatives showed good pharmaco-
kinetic properties. Our decision to choose a noncarbo-
hydrate fucose mimic seems to have been a good one. A
comparison of AUC/IC50 in the cell-rolling assay (Figure
4) showed that 2 and 22 scored the highest, and these
two compounds were advanced to in vivo studies based
on these results.
The compounds were first evaluated in intravital
microscopy (IVM). In this study, the mesentery vessels
of mice are surgically exteriorized. This surgical ma-
nipulation results in up-regulation of P-selectin on the
vessel endothelium and causes leukocytes to roll on the
endothelium via the P-selectin PSGL-1 interaction.
Fifteen minutes after surgery, leukocyte rolling is video
recorded. At 50 mg/kg iv compounds 2 and 22 showed
a 49% and 48% decrease in leukocyte rolling respec-
tively, compared to vehicle (Figure 5). In comparison, a
P-selectin antibody completely abated rolling at a dose
of 2.5 mg/kg iv. These results clearly show that quinic
acid derivatives inhibit P-selectin dependent leukocyte
rolling in vivo.
2 and 22-27 were also evaluated in the cell-rolling
assay. Six compounds had IC50s between 100 and 1000
µM. One compound caused cell lysis. It is difficult to
deduce a meaningful SAR from our data. The only
conclusion that can be drawn is that a hydrophobic
substituent on the biphenyl results in a modest increase
Compound 2, which has somewhat better iv PK than
22, was evaluated in the murine thioglycolate induced

4352 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 13
Table 1. Hits from the Quinic Acid Libraries
Kaila et al.
Table 2. Abbreviated Pharmacokinetic Parameters after
6). In conclusion we have identified small molecule
P-selectin inhibitors with in vivo results comparable to
PSGL-1.
Single Dose 20 mg/kg iv
compound
CL, mL/min/kg
AUC, min‚µM
Cmax, µM
2
22
23
24
29
30
31
4
5
13159
9376
5233
1899
8480
5028
1332
211.47
245.71
60.69
48.7
Experimental Section
10
24
6
Crystallography. Crystals of E-LE were grown using the
hanging drop vapor diffusion method at 18 °C. The solution
contained protein at 30 mg/mL, 15% (w/v) PEG 4000, 200 mM
CaCl2, 10 mM Tris-HCl, and 100 mM HEPES at pH 7.5. The
crystals were transferred to an intermediate stabilizing solu-
tion consisting of 30% (w/v) PEG 4000, 200 mM CaCl2, 100
mM HEPES, and 200 mM quinic acid which was adjusted to
114
12
35
73.95
57.69
peritonitis (TIP) model. 2 at 50 mg/kg iv inhibited
neutrophil influx by 58% compared to vehicle (Figure

Sialyl Lewisx-Mimicking Selectin Inhibitors
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 13 4353
Table 3. Statistics for the Complex of Quinic Acid with E-LE
space group
P212121
unit cell parameters
a (Å)
34.5
72.0
77.9
2.0
b (Å)
c (Å)
maximum resolution (Å)
reflections
65483
12915
5.5
unique reflections
Rmerge (%)
completeness (%)
shell 2.00-2.07 Å
Rmerge (%)
completeness (%)
model refinement
R-value
94.2
21.1
73.7
Figure 4. Quinic acid monomer lead compounds, AUC/IC50.
0.201
0.237
0.010
1.44
Rfree-value
rmsd bonds (Å)
rmsd angles (deg)
lowest energy structure of each parent structure was used as
a starting point for designing the library. A virtual library for
each parent structure was generated in the presence of
P-selectin using QXP employing the same method as described
above. The low energy structure of each construct was retained
and ranked using the QXP scoring function.16 The top 200
structures were visually screened and the best structures
retained for library synthesis. For library G, a diverse set of
amines was chosen using molecular fingerprints and atom-
pair fingerprints as descriptors, and hierarchal clustering
through Selector.21
Figure 5. Effect of quinic acid derivatives 2 and 22 on
leukocyte rolling in mouse mesenteric.
Chemistry. Reactions were run using commercially avail-
able starting materials and anhydrous solvents, without
further purification. Proton NMR spectra were recorded at 300
MHz on a Varian Gemini 2000 or on a 400 MHz Bruker AV-
400 spectrometer using TMS (δ 0.0) as a reference. Combustion
analyses were obtained using a Perkin-Elmer Series II 2400
CHNS/O analyzer. CHN analyses were carried out by
Robertson-Microlit. Where analyses are indicated by symbols
of the elements, analytical results obtained for those elements
were (0.4 of the theoretical values. Low resolution mass
spectra were obtained using a Micromass Platform Electro-
spray Ionization Quadrupole mass spectrometer. High resolu-
tion mass spectra were obtained using a Bruker APEXIII
Fourier transform ion cyclotron resonance (FT-ICR) mass
spectrometer equipped with an actively shielded 7 T super-
conducting magnet (Magnex Scientific Ltd., UK) and an
external Bruker APOLLO electrospray ionization (ESI) source.
Preparative HPLC was run using a Shimadzu reverse phase
prep HPLC with Bischoff C18 SH (10 µm) column. The flow
rate was 40 mL/min and mobil phase A was water with TFA
as modifier; mobile phase B was CH3CN. Purity in two solvent
systems [H2O-CH3CN (method 1) and H2O-MeOH (method
2)] was determined using an Agilent 1100 HPLC instrument,
and all compounds analyzed were >98% pure.
General Procedure for Library Synthesis. To a 2-dram
vial were added all the reagents. The vials were then capped,
shaken, and heated (if needed) overnight. The volatiles were
removed under vacuum, and the residue was redissolved in
EtOAc (2 mL) and water (1 mL) added. The vials were vortexed
and centrifuged. The organic layer was collected and the
aqueous layer extracted with EtOAc (1 mL × 2). At this stage,
for some of the libraries, resins were added to remove the
excess acidic or basic reagents. For example, for library B,
Amberlite basic resin was added, and the vials were shaken
for 1 h at room temperature to remove the excess boronic acid.
The solvent was then transferred to another vial, and solvent
was removed on Savant. The residues were redissolved in 2
mL of THF, 250 µL of water, and 800 µL of 0.5 M LiOH in
water and the resulting mixture was shaken at room-temper-
ature overnight (for some of the libraries, methanol was used
instead of THF). Amberlite acidic resin was added to the vial,
and the vial was shaken for 1 h at room temperature. The
solution was transferred to a tared vial and an aliquot removed
for QC analysis. Products with less than 90% purity were
Figure 6. Compound 2 in the TIP model.
pH 7.5 with the addition of small amounts of HCl and NaOH.
After 15 h the crystals were transferred into an identical
solution but with the CaCl2 concentration reduced to 2 mM.
After 1 h in the second solution the crystals were rapidly cooled
by plunging into liquid propane held at liquid nitrogen
temperatures. Diffration data were collected on a Raxis II
detector mounted on a Rigaku RU200 generator with Yale/
Molecular Structure Corp. focusing mirrors. The data were
reduced using DENZO and Scalepack (HKL research INC)
giving the statistics in Table 3.
Rigid body refinement in XPLOR20 resulted in phases that
gave clear electron density for a bound Quinic acid. Refinement
of the complex was performed using XPLOR giving the
statistics in Table 3.
Molecular Modeling. Parent structures for each library
were modeled onto the P-selectin site by overlaying the quinic
acid moiety of the ligand onto quinic acid in the P-selectin/
quinic acid model structure. Each structure was optimized
using a Monte Carlo simulation of 1000 steps in which the
ligand was allowed to rotate, translate, and flex while the
protein was held rigid using QXP.16 The 3- and 4-hydroxyls of
the quinic acid moiety were constrained to be proximal to the
selectin-bound calcium through he use of zero-order bonds. The

4354 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 13
Kaila et al.
purified by HPLC. For compounds with greater than 90%
purity, the solvent was removed and the vial weighed, and a
determined volume of DMSO was added to the product to
make a 60 mM solution, which was formatted into 96-well
polypropylene microtiter plates. These solutions were further
diluted as described in the Experimental Section for the
Biacore assay. If these compounds showed e10 mM IC50 in
the Biacore assay they were resynthesized as powders and
evaluated in the NMR and cell-based flow assay. Procedures
for the resynthesis of hits (i.e. compounds from Table 1) are
given below.
mmol). NaHCO3 (3.3 mmol), Pd(PPh3)4 (50 mg), CH3CN (30
mL), and water (10 mL). The reaction system was evacuated
and then filled with nitrogen. This was repeated at least three
times. Then the resulting mixture was refluxed overnight
under nitrogen. After cooling to room temperature, the reaction
was quenched with aqueous NH4Cl solution and extracted with
EtOAc. After the organic layer was concentrated, the residue
was dissolved in methanol (40 mL), followed by addition of
LiOH (1 M, 4 mL). The resulting mixture was stirred at room-
temperature overnite. Amberlite acidic resin was added to
neutralize the reaction mixture to a pH around 5. After
filtration, the solution was concentrated and the residue was
purified by reverse phase HPLC to give compound 2 (300 mg,
1,3,4-Trihydroxy-6-oxa-bicyclo[3.2.1]octan-7-one (4). In
a 500 mL round-bottom flask fitted with a stirring bar, reflux
condensor, Dean-Stark trap, and argon inlet, 20 g of quinic
acid (104 mmol) was placed and 20 mL of dry DMF was added
via syringe and the slurry stirred at room temperature. Next,
benzene (200 mL, added via cannula) and p-toluenesulfonic
acid (2.0 g, 10.4 mmol) were added, and the slurry was heated
to reflux for 25 h. TLC after 25 h showed completion of
reaction. A 1:1 mixture of EtOAc and heptane (100 mL) was
added to the cooled reaction mixture. The mixture stirred for
1 h at room temperature and filtered. The collected solid was
again stirred with a 1:1 mixture of EtOAc and heptane (100
mL) for 1 h at room temperature and filtered. Tituration was
repeated one more time with a 1:1 mixture of EtOAc and
heptane (100 mL) and the precipitate collected to give 16.9 g
1
70%). H NMR (400 MHz, DMSO-d6+D2O) δ ppm 1.93 (m, 3
H), 2.09 (d, J ) 13.1 Hz, 1 H), 3.34-3.42 (m, 1 H), 3.75-3.84
(m. 1 H), 3.86-3.94 (m, 1 H), 4.27 (d, J ) 10.9 Hz, 1 H), 4.54
(d, J ) 11.1 Hz, 1 H), 7.23 (d, J ) 7.6 Hz, 2 H), 7.35 (t, J ) 8.7
Hz, 3 H), 7.57 (t, J ) 7.6 Hz, 1 H), 7.67 (t, J ) 7.58 Hz, 1 H),
7.79 (d, J ) 8.1 Hz, 1 H). Anal. (C21H21F3NaO6.CH3OH) C, H.
HRMS calcd. for C21H21F3O6 (M + H)+ 427.1363, found:
427.1351.
3,4,5-Trihydroxy-1-(4′-trifluoromethoxy-biphenyl-3-yl-
methoxy)-cyclohexanecarboxylic Acid (22). This com-
pound was prepared using the same procedure as for com-
pound 2. 1H NMR (400 MHz, DMSO-d6+D2O) δ ppm 1.97 (m,
2 H), 2.01 (m, 1 H), 2.03 (m, 1 H), 3.43 (dd, J ) 6.4, 3.16 Hz,
1 H), 3.78-3.88 (m, 1 H), 3.91-4.02 (m, 1 H) 7.38 (d, J ) 7.6
Hz, 1 H), 7.48 (t, J ) 7.8 Hz, 3 H), 7.59 (d, J ) 7.6 Hz, 1 H),
7.64 (s, 1 H), 7.78 (d, J ) 8.6 Hz, 2 H). Anal. (C21H21F3O7‚
H2O) C, H. HRMS calcd. for C21H21F3O7 (M + H)+ 443.1312,
found: 443.1325.
1-(3-Benzofuran-2-yl-benzyloxy)-(3R,5R)-3,4,5-trihy-
droxy-cyclohexanecarboxylic Acid (23). This compound
was prepared using the same procedure as for compound 2.
1H NMR (400 MHz, DMSO-d6+D2O) δ ppm 1.85-2.03 (m, 3
H), 2.05-2.15 (m, 1 H), 3.38 (dd, J ) 6.6, 3.0 Hz, 1 H), 3.75-
3.84 (m, 1 H), 3.82-3.90 (m, 1 H), 4.31 (d, J ) 11.1 Hz, 1 H),
4.57 (d, J ) 11.12 Hz, 1 H), 7.24 (t, J ) 7.45 Hz, 1 H), 7.27-
7.36 (m, 3 H) 7.45 (t, J ) 7.71 Hz, 1 H), 7.59 (d, J ) 8.1 Hz, 1
H), 7.64 (d, J ) 7.6 Hz, 1 H), 7.79 (d, J ) 7.8 Hz, 1 H), 7.83 (s,
1 H). Anal. (C22H22O7.CH3OH) C, H. HRMS calcd. for C22H22O7
(M + H)+ 399.1439, found: 399.1447.
1-(3′,5′-Bis-trifluoromethyl-biphenyl-4-ylmethoxy)-
(3R,5R)-3,4,5-trihydroxy-cyclohexanecarboxylic Acid (24).
This compound was prepared using the same procedure as for
compound 2. 1H NMR (400 MHz, DMSO-d6d6+D2O) δ ppm
1.95 (dd, J ) 13.0, 7.45 Hz, 2 H), 1.99-2.08 (m, 1 H), 2.10-
2.20 (m, 1 H), 3.44 (dd, J ) 6.3, 3.0 Hz, 1 H), 3.81-3.88 (m, 1
H), 3.94-4.02 (m, 1 H), 4.35 (d, J ) 11.1 Hz, 1 H), 4.61 (d, J
) 11.4 Hz, 1 H), 7.52 (d, J ) 8.3 Hz, 2 H), 7.81 (d, J ) 8.3 Hz,
2 H), 8.08 (s, 1 H), 8.30 (s, 2 H). Anal. (C22H20F6O6‚H2O) C, H.
HRMS calcd. for C22H20F6O6 (M + H)+ 495.1237, found:
495.125.
(3R,5R)-3,4,5-Trihydroxy-1-(3-naphthalen-2-yl-benzyl-
oxy)-cyclohexanecarboxylic Acid (25). This compound was
prepared using the same procedure as for compound 2. 1H
NMR (400 MHz, DMSO-d6+D2O) δ ppm 1.91-2.08 (m, 3 H),
2.16 (dd, J ) 13.1, 3.54 Hz, 1 H), 3.43 (dd, J ) 6.3, 3.0 Hz, 1
H), 3.81-3.88 (m, 1 H), 3.94-4.02 (m, 1 H), 4.39 (d, J ) 11.1
Hz, 1 H), 4.64 (d, J ) 10.9 Hz, 1 H), 7.38 (d, J ) 7.6 Hz, 1 H),
7.48-7.60 (m, 3 H), 7.74 (d, J ) 7.8 Hz, 1 H), 7.76-7.88 (m, 2
H), 7.96 (d, J ) 7.3 Hz, 1 H), 7.99-8.08 (m, 2 H), 8.17-8.26
(m, 1 H). HPLC (Method 1: 100%, Method 2: 100%); HRMS
calcd for C24H24O6 (M + H+) 409.1646, found: 409.1649.
1-(4′-Benzyloxy-biphenyl-3-ylmethoxy)-(3R,5R)-3,4,5-
trihydroxy-cyclohexanecarboxylic Acid (26). This com-
pound was prepared using the same procedure as for com-
pound 2. 1H NMR (400 MHz, DMSO-d6+D2O) δ ppm 1.83-
2.03 (m, 3 H), 2.04-2.16 (m, 1 H), 3.37 (dd, J ) 6.1, 2.8 Hz, 1
H), 3.72-3.79 (m, 1 H), 3.87-3.96 (m, 1 H), 4.28 (d, J ) 11.1
Hz, 1 H), 4.53 (d, J ) 11.1 Hz, 1 H), 5.11 (s, 2 H), 7.07 (d, J )
8.6 Hz, 2 H), 7.23 (d, J ) 7.3 Hz, 1 H), 7.39 (m, 6 H), 7.50-
7.56 (m, 4 H). Anal. (C27H28O7) C, H. HRMS calcd. for C27H28O7
(M + H465.1908, found: 465.1901.
1
of intermediate 4 (93% yield). H NMR (300 MHz, CD3OD) δ
1.89 (t, J ) 11.7 Hz, 1H), 2.01-2.08 (m, 1H), 2.21-2.27 (m,
1H), 2.49 (d, J ) 11.3 Hz, 1H), 3.72 (ddd, J ) 11.3, 6.8, 4.5
Hz, 1H), 4.00 (t, J ) 4.6 Hz, 1H), 4.72 (t, J ) 5.4 Hz, 1H).
8-Hydroxy-3,5,10-trioxa-tricyclo[6.2.1.02,6]undecane-
4,9-dione (5). To intermediate 4 (15.0 g, 86.2 mmol) in 42.5
mL of dry pyridine (kept over KOH) at 0 °C was added
triphosgene (10.2 g, 34.5 mmol) portionwise over about 5 min.
The ice bath was removed and the reaction stirred at room
temperature for 1 h. TLC showed some unreacted starting
material. Another 1 g of triphosgene was added and the
reaction stirred for another 1 h. TLC showed completion of
reaction. Pyridine was removed on the high vaccum pump at
room temperature, and then the thick syrup was poured into
water. The aqueous layer was repeatedly extracted with EtOAc
until TLC showed no leftover product in the aqueous layer.
The organic layer was dried over magnesium sulfate, filtered,
and evaporated under reduced pressure. The resulting solid
was extracted with hot methanol and filtered while hot. The
filtrate was evaporated under reduced pressure to give the
pure product (13.1 g, 76% yield). 1H NMR (300 MHz, CD3OD)
δ 2.16 (dd, J ) 14.9, 3.8 Hz, 1H), 2.34 (d, J ) 12.6 Hz, 1H),
2.50-2.64 (m, 1H), 4.90-5.00 (m, 2H), 5.16 (dt, J ) 11.6, 3.8
Hz, 1H).
8-(3-Bromobenzyloxy)-3,5,10-trioxa-tricyclo[6.2.1.02,6]-
undecane-4,9-dione (8c). To intermediate 5 (3.0 g, 15.0
mmol) in 16 mL of dry DMF at 0 °C (ice/NaCl) was added NaH
(900 mg, 22.5 mmol, 60% dispersion in oil). The reaction was
stirred at 0 °C for 30 min. Next the 4-bromobenzyl bromide
(7.5 g, 30 mmol) was added and the reaction left in the ice
bath for 30 min. The ice bath was then removed and the
reaction stirred at room temperature for 1 h. TLC showed no
starting material, and the reaction was worked up by quench-
ing with saturated aqueous ammonium chloride and extracting
into EtOAc. Column chromatography using silica gel and
eluting with hexanes/EtOAc gave pure product in 76% yield.
1H NMR (300 MHz, CDCl3), δ 2.06 (m, 1H), 2.22 (dd, J ) 15
Hz, 3.6 Hz, 1H), 2.57-2.64 (m, 2H), 4.51(dd, J ) 3.4, 9.9 Hz,
2H), 4.84 (m, 2H), 5.03 (dt, J ) 7.8, 3.6 Hz, 1H), 7.12 (d, J )
7.54 Hz, 2H), 7.40 (d, J ) 7.8 Hz, 2H). MS(electrospray), 369
(M + H)+.
(3R,5R)-3,4,5-Trihydroxy-1-(2′-trifluoromethyl-biphen-
yl-3-ylmethoxy)-cyclohexanecarboxylic Acid (2). Inter-
mediate 8c (369 mg, 1 mmol) was dissolved in methanol (10
mL), followed by addition of LiOH solution (1 M, 2 mL, 2 mmol)
at 0 °C. The resulting mixture was stirred at room temperature
for 2 h. The solution was then concentrated. To the residue
was added o-trifluoromethylphenylboronic acid (209 mg, 1.1

Sialyl Lewisx-Mimicking Selectin Inhibitors
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 13 4355
(3R,5R)-3,4,5-Trihydroxy-1-(2-thianthren-1-yl-benzyl-
oxy)-cyclohexanecaroxylic Acid (27). This compound was
prepared using the same procedure as for compound 2. 1H
NMR (400 MHz, DMSO-d6+D2O) δ ppm 1.72-1.94 (m, 4 H),
3.31-3.40 (m, 1 H), 3.67-3.76 (m, 1 H), 3.76-3.86 (m, 1 H),
3.93 (d, J ) 12.0 Hz, 1 H), 4.35 (d, J ) 11.4 Hz, 1 H), 7.13 (d,
J ) 7.6 Hz, 1 H), 7.20 (d, J ) 7.6 Hz, 1 H), 7.28-7.45 (m, 5
H), 7.49 (t, J ) 7.6 Hz, 1 H), 7.57-7.70 (m, 3 H). Anal. (C26H23-
NaO6S2) C, H. HRMS calcd. for C26H24O6S2 (M + H)+ 514.1352,
found: 514.1360.
chromatography using hexane/EtOAc to give intermediate 15.
To intermediate 5 (3.0 g, 15.0 mmol) in 16 mL of dry DMF at
0 °C (ice/NaCl) was added NaH (900 mg, 22.5 mmol, 60%
dispersion in oil). The reaction was stirred at 0 °C for 30 min.
Next, intermediate 15 (9.8 g, 30.0 mmol) was added and the
reaction left in the ice bath for 30 min. The ice bath was then
removed and the reaction stirred at room temperature for 1
h. The reaction was worked up by quenching with saturated
aqueous ammonium chloride and extracting into EtOAc.
Column chromatography using hexane/EtOAc gave pure prod-
1
uct 16 in 70% yield. H NMR (300 MHz, CD3OD) δ ppm 2.29
8-(2′-Amino-biphenyl-4-ylmethoxy)-3,5,10-trioxa-
tricyclo[6.2.1.02,6]undecane-4,9-dione (13). Intermediate
5 (3.00 g, 14.9 mmol) was dissolved in dry DMF under argon
at 0 °C. To the cold solution was added NaH (900 mg, 22.4
mmol), and the reaction was allowed to stir for 30 min. To the
mixture was added 3′-bromomethyl-2-nitro-biphenyl22 (4.37 g,
14.9 mmol) dissolved in dry DMF, via syringe under argon.
The reaction was allowed to reach room-temperature over-
night. The mixture was poured into a cold solution of saturated
aqueous ammonium chloride (200 mL) and extracted with
EtOAc (2 × 200 mL). The combined organic phases were dried
over sodium sulfate and filtered. The organics were concen-
trated to an oil and titurated in hexanes to yield 2.63 g of a
yellow powder. The yellow solid (2.63 g, 6.4 mmol) was
dissolved in EtOH/DMF (1:1). To the solution was added SnCl2‚
2H2O (8.65 g, 38.3 mmol), and the reaction was allowed to stir
at room-temperature overnight. Solid NaHCO3 was added until
basic and the mixture stirred for 60 min. The mixture was
poured into water and extracted with EtOAc (2 × 200 mL).
The combined extracts were dried over sodium sulfate, filtered,
and concentrated under vacuum to provide an oil. The oil was
purified by column chromatography (5:1 hexane/EtOAc) to give
intermediate 13 as a yellow powder (3.1 g, 55% yield). 1H NMR
(300 MHz, DMSO) δ ppm 2.11 (dd, J ) 14.8, 3.8 Hz, 1H), 2.42
(d, J ) 12.5 Hz, 1H), 2.35-2.52 (m, 2 H), 4.29 (d, J ) 10.9 Hz,
1 H), 4.58 (d, J ) 10.8 Hz, 1 H), 4.95-5.1 (m, 2 H), 5.15-5.26
(m, 1 H), 6.90-7.02 (m, 2 H), 7.12-7.25 (m, 2 H), 7.37-7.44
(m, 4 H). MS(electrospray), 382 (M + H)+.
(d, J ) 13.5 Hz, 1H), 2.38 (dd, J ) 17.0, 3.1 Hz, 1H), 2.69-
2.88 (m, 2H), 4.68 (dd, J ) 2.5, 11.0 Hz, 2H), 4.86(d, J ) 8.1
Hz, 1H), 4.96 (m, 1H), 5.04-5.14 (m, 1H), 7.18-7.62 (m, 8H).
MS(electrospray), 445 (M + H)+.
(3R,5R)-3,4,5-Trihydroxy-1-(4′-propenyl-biphenyl-4-yl-
methoxy)cyclohexanecarboxylic Acid (29). Under argon,
ZnCl2 solution (0.5 M in THF, 0.5 mL) was added to propen-
ylmagnesium bromide solution (0.5 M in THF, 0.5 mL) at room
temperature. After 5 min, this mixture was transferred to a
solution of intermediate 16 (50 mg, 0.1 mmol) and Pd(PPh3)4
(12 mg, 0.01 mmol) in THF (2 mL) under argon. The resulting
mixture was refluxed for 3 h. The reaction was quenched with
aqueous NH4Cl solution and extracted with EtOAc (2 mL ×
2). After concentration of the organic layer, the residue was
dissolved in methanol (2 mL), followed by addition of LiOH (1
M in water, 0.6 mL). The resulting solution was stirred at r.t.
overnight. Amberlite acidic resin was added to neutralize the
solution to a pH around 5. After filtration, the solution was
concentrated and the residue was purified by reverse phase
HPLC to give 29 (25.4 mg, 57%). 1H NMR (400 MHz, DMSO-
d6) δ ppm 1.88 (d, J ) 6.3 Hz, 1 H), 1.92 (dd, J ) 7.2, 1.64 Hz,
3 H), 1.94-2.08 (m, 2 H), 2.10-2.18 (m, 1H), 3.43 (dd, J )
6.4, 3.2 Hz, 1 H), 3.80-3.88 (m, 1 H), 3.93-4.00 (m, 1 H), 4.35
(d, J ) 10.9 Hz, 1 H), 4.60 (d, J ) 11.1 Hz, 1 H), 5.79-5.92
(m, 1 H), 6.43-6.52 (m, 1 H), 7.30-7.37 (m, 1 H), 7.42-7.47
(m, 2 H), 7.48 (d, J ) 7.3 Hz, 1 H), 7.55-7.71 (m, 4 H). Anal.
(C23H26O6‚H2O) C, H. HRMS calcd for C23H26O6 (M + H+)
399.1802, found: 399.1799.
(3R,5R)-3,4,5-Trihydroxy-1-{2′-[(5-methyl-2-phenyl-2H-
[1,2,3]triazole-4-carbonyl)-amino]- biphenyl-4-ylmethoxy}-
cyclohexanecarboxylic Acid (28). To intermediate 13 (760
mg, 2 mmol) in 25 mL of CH3CN was added 1.4 mL of Hunig’s
base followed by 5-methyl-2-phenyl-2H-[1,2,3]triazole-4-car-
bonyl chloride (556 mg, 2.5 mmol). The reaction was heated
at 60 °C for 48 h. The volatiles were removed, and the residue
was dissolved in methylene chloride and water. The water
layer was extracted twice with methylene chloride. The
combined organics were dried with MgSO4 and concentrated
under vacuum. The residues were redissolved in 20 mL of
THF, followed by addition of LiOH (8 mL of 1 M solution in
water). The resulting solution was stirred at room-temperature
overnight. Amberlite acidic resin was added to neutralize the
solution to pH around 5. After filtration, the solution was
concentrated and the residue was purified by reverse phase
HPLC to give compound 28 (650 mg, 60% yield). 1H NMR (400
MHz, DMSO-d6+D2O) δ ppm 1.69-1.83 (m, 2 H), 1.85-1.93
(m, 1 H), 2.11 (d, J ) 11.6 Hz, 1 H), 2.46 (s, 3H), 3.39-3.49
(m, 1 H), 3.58-3.76 (m, 2 H, 4.43 (s, 2 H), 7.29 (t, J ) 7.3 Hz,
1 H), 7.32-7.45 (m, 7 H), 7.59 (t, J ) 8.0 Hz, 2 H), 7.80 (d, J
) 7.8 Hz, 2 H), 8.03 (d, J ) 7.8 Hz, 1 H). HPLC (Method 1:
99%, Method 2: 99%); HRMS calcd for C30H30N4O7 (M + H+)
559.2187, found: 559.2193.
8-(4′-Bromo-biphenyl-4-ylmethoxy)-3,5,10-trioxa-
tricyclo[6.2.1.0.2,6]undecane-4,9-dione (16). To 4′-bromo-
biphenyl-3-carbaldehyde23 (2.6 g, 10.0 mmol) in 30 mL THF
was added 14 mL of a 1 molar solution of borane in THF (14.0
mmol). The reaction was stirred overnight and then quenched
with water and extracted with EtOAc (3×). After removal of
the solvent, the crude reaction mixture was dissolved in 15
mL of CH2Cl2 and to this was added a suspension of triphen-
ylphosphine dibromide (4.6 g, 11 mmol) in 20 mL of CH2Cl2.
The reaction was stirred for 4 h and then washed with water.
The organic layer was dried with magnesium sulfate and the
solvent evaporated. The crude mixture was purified by column
3-Amino-5-[(1,3,4,5-tetraacetoxy-cyclohexanecarbonyl)-
amino]-benzoic Acid Methyl Ester (21). To a stirred
solution of 5.00 g (26.0 mmol) quinic acid in 50 mL anhydrous
pyridine, cooled in an ice bath under argon, was added
dropwise 50 mL acetic anhydride. At the end of the addition
the reaction was allowed to warm slowly to room temperature.
After 24 h the solvents were removed under vacuum at <35
°C and the residue was azeotroped twice with toluene-CH3-
CN, then dried under high vacuum to a yellow foam, which
was used directly in the next step. MS(electrospray), 359 (M
- H)-.
The crude, peracetylated quinic acid 19 (26 mmol) and 3.5
g (26.0 mmol) 1-hydroxybenzotriazole were dissolved in a
mixture of 120 mL of dry CH2Cl2 and 120 mL of dry THF, and
the solution was cooled under argon in an ice bath. After
addition of 5.4 g (28.8 mmol) of dicyclohexylcarbodiimide,
the reaction was stirred at 0-4 °C for 1 h, then 2 h at room
temperature. The solid dicyclohexylurea was removed by
filtration and washed with CH2Cl2, and the combined fil-
trate and washings were evaporated under reduced pressure.
The residue was taken up in 100 mL of CH2Cl2, and some
residual dicyclohexylurea was removed by filtration. After
evaporation of the filtrate, the residue was dried under high
vacuum to a pale yellow foam, which was used directly in the
next step.
The crude activated ester 20 (26 mmol) was dissolved under
argon in 110 mL of dry CH2Cl2, and the solution transferred
via syringe into a flame dried 1 L, three-necked flask under
argon. After addition of 17.0 g (78.0 mmol) of 3,5-diaminoben-
zoic acid methyl ester, the stirred suspension was cooled under
argon in an ice-water bath for 20 min. A solution of 5.7 mL
(40.6 mmol) of triethylamine in 80 mL of anhydrous DMF was
then added from a dropping funnel over 0.5 h while stirring
at 0-4 °C under argon. Care should be taken that the pH does
not exceed ∼10 during the addition. At the end of the addition,

4356 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 13
Kaila et al.
the solution was stirred in ice-water for 1 h, then at room-
Pharmacokinetics in Rats. Jugular vein cannulated
Sprague Dawley male rats weighing between 195 and 215 g
were used. Animals were fasted and housed at constant
temperature-humidity environment at 12-12 h day-night
cycles. Three rats were administered single 20 mg/kg iv bolus
(10 mg/mL, 50% DMSO:50% PEG400). Blood was collected
from each rat at 5 min, 0.5, 0.75, 1, 2, 4, 7, 24 h after dosing.
Equal volume of saline was given to replace the lost blood.
Plasma levels of test compounds were assayed via LC/MS
method. In this method an aliquot of plasma was treated (with
2× of CH3CN to precipitate plasma protein) after centrifuga-
tion, the supernatant was then further diluted with format
buffer (50 mM) and inject onto an LC/MS. A linear trapezoid
noncompartmental analysis was used in pharmacokinetic
analysis.
Biology. Biacore P-Selectin/PSGL-1 Inhibition Assay.
All surface plasmon resonance (Biacore) assays were per-
formed at 25 °C on a Biacore 3000 instrument (Biacore Inc.
Piscataway, NJ). 19ek, a purified, monomeric, truncated form
of human PSGL-1 that contains all the necessary P-selectin
binding determinants, was biotinylated via amine chemistry
(Sulfo-NHS-LC-Biotin, Pierce) at a unique C-terminal lysine
residue and immobilized on a Biacore streptavidin-coated SA
sensor chip.11 Small molecule P-selectin inhibitors were incu-
bated for 1 h in 100 mM HEPES, 150 mM NaCl, 1 mM CaCl2,
1 mM MgCl2, 0.05% P40, 10% DMSO. Solutions were filtered
to remove any possible precipitate. A soluble recombinant
truncated form of human P-selectin comprised of the lectin
and EGF domains (P-LE), expressed in CHO cells, was
delivered to the 19ek coated sensor chip at 30 µL/minute in
the presence and absence of small molecule inhibitors. The
final P-LE solution contained 500 nM P.LE, 100 mM HEPES,
150 mM NaCl, 1mM CaCl2, 1 mM MgCl2, 0.05% P40, 10%
DMSO and compound from 0 to 6 mM. Sample injections were
run in duplicate, and each compound was assayed at least
twice. Percent inhibition of binding was calculated by dividing
the inhibited signal, measured in resonance units (RU), by the
uninhibited signal, subtracting this value from one, then
multiplying by one hundred. Inhibitors, with greater than 50%
inhibition at 6mM, were assayed again using a series of 2-fold
dilutions. The data from this titration were plotted RU vs
concentration, and the IC50 concentration was determined by
extrapolation from the plot. All RU values were blank and
reference subtracted prior to percent inhibition and IC50.
In Vitro Leukocyte Rolling Assay. A parallel plate flow
chamber (GlycoTech, Rockville, MD) was used to measure
neutrophil/endothelial cell adhesion events under flow as
previously described,24 with changes in protocol noted as
follows. Confluent monolayers of HUVECs (human umbilical
cord endothelial cells) were activated with IL-4 to up-regulate
P-selectin expression on the cell surface just prior to the
experiment. Freshly isolated human neutrophils were incu-
bated for 20 min with or without compound in 4 mL of serum-
free cell media containing histamine (to further activate the
HUVEC once introduced into the flow chamber). This solution
was flowed over the HUVEC monolayer to generate a shear
stress of at approximately 1 dyn/cm2. For each test article, data
was collected by digital imaging and videotape on three
different monolayers. The number of rolling cells was an
average value from 10 fields in each of the three monolayers.
Active compounds were titrated to determine an IC50 concen-
tration. Glycyrrhizin trisodium salt (TCI) was used as a
positive control, inhibiting 50% of rolling flux at 1 mM.
Intravital Microscopy in Mouse Mesentery Venules.
C57/black mice were placed briefly under a heat lamp to dilate
blood vessels and then injected with small molecule in 8%
PEG400, 0.2% Tween80/water (w/w) pH 6-8 or vehicle by
intravenous tail vein injection. The mice were then anesthe-
tized using avertin, and the mesentery was exteriorized
through a midline abdominal incision. Venules were visualized
with a Zeiss Axiovert 135-inverted microscope (32x, 0.4NA;
Zeiss, Oberkochen, Germany). Messenteric tissues were bathed
periodically in PBS warmed to 37 °C, during the experiment.
Fifteen minutes after the mesentery was exteriorized, vessels
temperature overnight.
The reaction was diluted with 800 mL CH2Cl2 and washed
in sequence with 2 × 400 mL of H2O, 2 × 500 mL of 0.1 N
HCl, 500 mL of diluted (1:10) brine, and 500 mL of semisatu-
rated brine. The organic phase was dried (Na2SO4) and the
solvent evaporated under vacuum. The residue was dissolved
in 60 mL of CH2Cl2, and the solution was purified by flash
chromatography on silica gel eluting with cyclohexane/EtOAc.
Evaporation of the pertinent fractions afforded a residue which
crystallized from hexanes/EtOAc to afford 7.22 g of pure
intermediate 21 as colorless crystals. A second crop (0.30 g) of
pure material was obtained from the mother liquors. The total
yield was 60% over the three steps. 1H NMR (300 MHz, DMSO-
d6+D2O) δ ppm 1.88-1.99 (m, 10 H), 2.04 (s, 3 H), 2.11 (s, 3
H), 3.76 (s, 3 H), 5.04 (dd, J ) 10.0, 3.7 Hz, 1 H), 5.22-5.37
(m, 1 H, 5.37-5.47 (m, 1 H), 6.89 (s, 1 H), 7.11 (s, 1 H), 7.30
(s, 1 H), 9.70 (s, 1 H). MS(electrospray), 509 (M + H)+.
3-[(3-Chloro-1-benzothien-2-ylcarbonyl)amino]-5-
({[3R,5R)-1,3,4,5-tetrahydroxylcyclohexyl]carbonyl}-
amino)benzoic Acid (30). To a solution of 509 mg (1.0 mmol)
of intermediate 21 in 6 mL dry CH3CN was added 0.35 mL
(2.0 mmol) of diisopropylethylamine, dropwise via syringe,
while stirring at 0 °C under argon, followed by a solution of
246 mg (1.3 mmol) benzo[b]thiophene-2-carbonyl chloride in
4 mL of anhydrous CH3CN, over 5 min. The reaction was
stirred 15 min in the ice-water bath, then 7 h at room
temperature, after which point no starting material was
present. The solution was concentrated under vacuum to a
small volume, diluted with 150 mL of EtOAc, and washed in
sequence with water (50 mL), ice cold 0.5 N HCl (50 mL), ice-
cold semisaturated NaHCO3 (50 mL), and semisaturated brine
(2 × 50 mL). All the aqueous washings were extracted with
EtOAc (150 mL). The combined organic layers were dried (Na2-
SO4) and evaporated under vacuum. The residue was purified
by flash chromatography over silica gel, eluting with cyclo-
hexane/EtOAc. Evaporation of the appropriate fractions gave
582 mg of a pale tan solid. To a suspension of this solid in 15
mL of CH3CN was added a solution of LiOH‚H2O (177 mg,
4.20 mmol) in 10 mL water. After stirring 8 h at room
temperature, a second portion (60 mg, 1.4 mmol) of LiOH‚H2O
was added. One hour later, the conversion was complete.
Stirring with 7 g of Amberlite acidic resin for one minute
quenched the reaction. After removal by filtration, the resin
was washed with 1:1 CH3CN /H2O. The combined filtrate and
washings were evaporated under vacuum, and the residue was
re-evaporated several times from CH3CN to afford, after drying
1
under vacuum over P2O5, 390 mg (75%) of compound 30. H
NMR (400 MHz, DMSO-d6+D2O) δ ppm 1.80 (dd, J ) 12.8,
11.0 Hz, 1 H), 1.85-1.95 (m, 3 H), 3.26 (dd, J ) 8.8, 2.8 Hz, 1
H), 3.75-3.80 (m, 1 H), 3.97-4.03 (m, 1 H), 7.57-7.63 (m, 2
H), 7.91-7.98 (m, 3 H), 8.05-8.13 (m, 1 H), 8.26 (t, J ) 1.8
Hz, 1 H). Anal. (C23H21ClN2O8S‚ H2O) C, H. HRMS calcd for
C23H21ClN2O8S (M + H+) 521.078, found: 521.0781.
3-[(1-Benzothien-2-ylcarbonyl)amino]-5-({[3R,5R)-
1,3,4,5-tetrahydroxycyclohexyl]carbonyl}amino)-
benzoic Acid (31). This compound was prepared using the
same procedure as for compound 30. 1H NMR (400 MHz,
DMSO-d6) δ ppm 1.84 (dd, J ) 13.0, 10.5 Hz, 1 H), 1.89-1.98
(m, 3 H), 3.28 (dd, J ) 8.6, 2.8 Hz, 1 H), 3.75-3.93 (m, 1 H),
4.00-4.08 (m, 1 H), 4.78 (s, 2 H), 5.33 (s, 2 H),5.76 (s, 1 H),
7.45-7.55 (m, 2 H), 7.99-8.10 (m, 3 H), 8.13 (t, J ) 1.8 Hz, 1
H), 8.43 (s, 1 H), 8.49 (t, J ) 2.0 Hz, 1 H), 10.7 (s, 1 H). Anal.
(C23H22N2O8S‚2H2O) C, H.
Transferred NOE (trNOE) Experiment. The NMR
samples were prepared in D2O buffer, with 20 mM Imidazole,
pH ) 7.4, 150 mM NaCl, 5 mM CaCl2, 0.02% NaN3, with 1
mM small molecule and 40 µM P-selectin-hIgG chimera
protein. The 2D trNOE experiments were performed on a
Varian Unity Plus 600 MHz instrument at 25 °C, with 170
ms mixing time. The data were processed with nmrPipe
software.19 Negative control trNOE experiments were also
performed either without any added protein, or in the presence
of IgG instead of P-selectin.FC, run under the above conditions.

Sialyl Lewisx-Mimicking Selectin Inhibitors
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 13 4357
(9) Thoma, G.; Schwarzenbach, F. Simplified sialyl Lewis x ana-
logue.s with improved E-selectin Inhibition. Helv. Chim. Acta
2003, 86, 855-864.
(10) Simanek, E. E.; McGarvey, G. J.; Jablonowski, J. A.; Wong, C.-
H. Selectin-Carbohydrate Interactions: From Natural Ligands
to Designed Mimics. Chem. Rev. 1998, 98, 833-862 and refer-
ences therein.
(11) Somers, W. S.; Tang, J.; Shaw, G. D.; Camphausen, R. T. Insights
into the Molecular Basis of Leukocyte Tethering and Rolling
Revealed by Structures of P- and E- Selectin Bound to sLex and
PSGL-1. Cell 2000, 103, 467-479.
(12) Graves, B. J.; Crowther R. L.; Chandram C.; Rumberger J. M.;
Li S.; Huang K.-S.; Presky D. H.; Familletti P. C.; Wolitzky B.
A. Burns D. K. Insight into e-selectin/ligand interaction from
the crystal structure and mutagenesis of the lec/EGF domains.
Nature 1994, 367, 532-538.
measuring 100-200 µm were identified and leukocyte rolling
was video-recorded in the selected vessels. The number of
rolling cells/minute was determined by counting the number
of cells passing a given plane perpendicular to the length of
the vessel. This number was measured in five vessels per
mouse and averaged. Blood samples were collected for analysis
of plasma levels of the small molecules just before the animals
were euthanized with 100% CO2 gas.
Thioglycollate-Induced Peritonitis. Female Balb/c E-
selectin -/- mice were injected intravenously, by tail vein
injection with small molecules in 8% PEG400/ water (w/w) pH
4-9, vehicle or antibody. Ten minutes after injection of test
article, 1 mL of 3% thioglycollate (Sigma) solution was injected
into the peritoneal cavity. Two hours later the animals were
deeply anesthetized with CO2 gas, and peritoneal lavage was
performed by injecting 5 mL of complete RPMI containing 10
units mL-1 heparin into the peritoneal cavity, massaging the
abdominal wall, and removing the injected liquid. Total cell
counts were calculated in the lavage fluid using a Baker
Hematology Analyzer. Cytospins were prepared and stained
with Diff-Quick, to perform differential counts. Just before the
animals were euthanized with 100% CO2 gas, blood samples
were collected for the analysis of plasma levels of the test
articles. A blocking anti-P-selectin antibody, RB40.34 (Pharm-
ingen), was used as a positive control (2.5 mg/kg).
(13) Geng J. G.; Bevilacqua M. P.; Moore K. L.; McIntyre T. M.;
Prescott S. M.; Kim J. M.; Bliss G. A.; Zimmerman G. A.; McEver
R. P. Rapid neutrophil adhesion to activated endothelium
mediated by GMP-140. Nature 1990, 343, 757-760.
(14) Somers, W. Unpublished results towards P-selectin-quinic acid
soaks.
(15) The IC50 for natural ligand sLex in the Biacore and ELISA assay
is ∼15 and 8 mM, respectively.
(16) McMartin, C.; Bohacek, R. S.QXP: powerful, rapid computer
algorithms for structure-based drug design J. Comput.-Aided
Mol. Des. 1997, 11, 333-334.
(17) Hanessian, S., Pan, J., Carnell, A., Bouchard, H., Lesage, L.,
Total synthesis of (-) Reserpine using the chiron approach J.
Org. Chem. 1997, 62, 465-473.
(18) It is well demonstrated in the literature that due to the nature
of the selection biology this high IC50 translates to in vivo
efficacy.1-10
(19) (a) Poppe, L.; Brown, G. S.; Philo, J. S.; Nikrad, P. V.; Shah, B.
H. Conformation of sLex Tetrasaccharide, Free in Solution and
Bound to E-, P- and L-Selectin. J. Am. Chem. Soc. 1997, 119,
1727-1736. (b) Ni, F. Recent developments in transferred NOE
methods. Prog. Nucl. Magn. Reson. Spectrosc. 1994, 26, 517-
606. (c) Meyer, B.; Thomas, W.; Peters, T. Screening mixtures
for biological activity by NMR. Eur J. Biochem. 1997, 246, 705-
709.
Acknowledgment. We would like to thank Dr.
Jasbir Seehra for his encouragement and support, Jin
Tang for protein purification and Richard Zollner for
cell culture. We would also like to thank Soo Peang Khor
and Michael Shanler for their expertise in pharmaco-
kinetics. Finally we are thankful to the Discovery
analytical group for spectral data.
Supporting Information Available: List of compounds
prepared for Libraries A-H. This material is available free of
charge via the Internet at http://pubs.acs.org.
(20) Brunger, A. T.; Adams, P. D.; Clore, G. M.; DeLano, W. L.; Gros,
P.; Grosse-Kunstleve, R. W.; Jiang, J.-S.; Kuszewski, J.; Nilges,
N.; Pannu, N. S.; Read, R. J.; Rice, L. M.; Simonson, T.; Warren,
G. L. Crystallography & NMR System: a new software suite
for macromolecular structure determination Acta Crystallogr.
1998, D54, 905-921.
References
(1) Ebnet, K.; Vestweber, D. Molecular mechanisms that control
leukocyte extravasation: the selectins and the chemokines.
Histochem. Cell Biol. 1999, 112, 1-23.
(2) Sears, P.; Wong, C.-H. Carbohydrate Mimetics: A New Strategy
for Tackling the Problem of Carbohydrate-Mediated Biological
Recognition. Angew. Chem., Int. Ed. 1999, 38, 2300-2324.
(3) Ley, K. Integration of inflammatory signals by rolling neutro-
phils Immunol. Rev. 2002, 186, 8-18.
(4) Thoma, G.; Banteli, R.; Jahmke, W.; Magnani, J. L.; Patton, J.
T. Preorganization of the Bioactive Conformation of Sialyl Lewis
x Analogues Correlates with Their Affinity to E-Selectin. Angew.
Chem., Int. Ed. 2001, 40, 3644-3647.
(5) For a summary on design and synthesis of selectin inhibitors
see refs 5-9 and references therein. Kaila, N.; Thomas IV, B.
E. Design and Synthesis of Sialyl Lewis x Mimics as E- and
P-Selectin Inhibitors. Med. Res. Rev. 2002, 22, 566-601.
(6) Dey, P. M.; Witczak, Z. J. Funtionalized S- Thio-di- and
S-Oligosaccharide Precursors as Templates for Novel Slex/a
Mimetic antimetastatic Agents. Mini Rev. Med. Chem. 2003, 3,
271-280.
(7) Kaila, N.; Thomas, B. E. Selectin Inhibitors Expert Opin. Ther.
Pat. 2003, 13, 305-317.
(8) Chhabra, S. R.; Rahim, A. S. A.; Kellam, B. Recent Progress in
the Design of Selectin Inhibitors Mini Rev. Med. Chem. 2003,
3, 679-687.
(21) Tripos, Inc., St. Louis, MO.
(22) DeVita, R. J.; Bochis, R.; Frontier, A. J.; Kotliar, A.; Fisher, M.
H.; Schoen, W. R.; Wyvratt, M. J.; Cheng, K.; Chan W.-S.; Butler,
B.; Jacks, M. T.; Hickey, G. J.; Schleim, K. D.; Leung, K.; Chen,
Z.; Chiu, S.-H.; Feeney, W. P.; Cunningham P. K.; Smith R. G.
A Potent, Orally Bioavailable Benzazepinone Growth Hormone
Secretagogue. J. Med. Chem. 1998, 41, 1716-1728.
(23) This compound was prepared by Suzuki coupling of 1-bromo-4-
iodobenzene with 3-formylphenylboronic acid. Simoni, D.; Gi-
annini, G.; Baraldi, P. G.; Romagnoli, R.; Roberti, M.; Rondanin,
R.; Baruchello, R.; Grisolia, G.; Rossi, M.; Mirizzi, D.; Invidiata,
F. P.; Grimaudo, S.; Tolomeo, M. A convenient synthesis of
unsymmetrically substituted terphenyls of biologically active
stilbenes via a double Suzuki cross-coupling protocol Tetrahedron
Lett. 2003, 44, 3005-3008.
(24) Thoma, G.; Patton, J. T.; Magnain, J. L.; Ernst, B.; Ohrlein, R.;
Duthaler, R. O. Versatile Functionalization of Polylysine: Syn-
thesis, Characterization, and Use of Neoglycoconjugates. J. Am.
Chem. Soc. 1999, 121, 5919-5929.
(25) DeLano, W. L. The PyMOL Molecular Graphics System 2002,
DeLano Scientific, San Carlos, CA. (http://www.pymol.org.).
JM050049L