CCR2: Characterization of the antagonist binding site from a combined receptor modeling/mutagenesis approach

Article abstract of DOI:10.1021/jm030862l

We describe here a classical molecular modeling exercise that was carried out to provide a basis for the design of novel antagonist ligands of the CCR2 receptor. Using a theoretical model of the CCR2 receptor, docking studies were carried out to define pl

Full text of DOI:10.1021/jm030862l

4070
J . Med. Chem. 2003, 46, 4070-4086
CCR2: Ch a r a cter iza tion of th e An ta gon ist Bin d in g Site fr om a Com bin ed
Recep tor Mod elin g/Mu ta gen esis Ap p r oa ch
Theo A. Berkhout,† Frank E. Blaney,*,‡ Angela M. Bridges,§ David G. Cooper,‡ Ian T. Forbes,‡
Andrew D. Gribble,‡ Pieter H. E. Groot,† Adam Hardy,† Robert J . Ife,‡ Rejbinder Kaur,† Kitty E. Moores,†
Helen Shillito,† J ennifer Willetts,§ and J ason Witherington‡
Departments of Discovery Chemistry, Vascular Biology, and Gene Expression Sciences, GlaxoSmithKline, New Frontiers
Science Park, Third Avenue, Harlow, Essex, UK CM19 5AD
Received April 4, 2003
We describe here a classical molecular modeling exercise that was carried out to provide a
basis for the design of novel antagonist ligands of the CCR2 receptor. Using a theoretical model
of the CCR2 receptor, docking studies were carried out to define plausible binding modes for
the various known antagonist ligands, including our own series of indole piperidine compounds.
On the basis of these results, a number of site-directed mutations (SDM) were designed that
were intended to verify the proposed docking models. From these it was clear that further
refinements would be necessary in the model. This was aided by the publication of a crystal
structure of bovine rhodopsin, and a new receptor model was built by homology to this structure.
This latest model enabled us to define ligand-docking hypotheses that were in complete
agreement with the results of the SDM experiments.
In tr od u ction
found to exacerbate the disease.7 It has also been
reported that a truncated antagonist analogue of
MCP-1 (MCP-1 9-76) prevented the onset of arthritis
in the MRL-lpr mouse model.8 These data suggest
that lowering MCP-1 levels, or blocking CCR2, in man
may have useful therapeutic effects in treating chronic
inflammatory conditions such as atherosclerosis and/
or arthritis. The discovery and development of small
molecular weight CCR2 antagonists has therefore
been regarded as an important pharmaceutical op-
portunity.
Chemokines are small, inducible, secreted chemotac-
tic cytokines involved in the trafficking of white blood
cells, critical in immunosurveillance and inflammation.1
On the basis of the configuration of conserved cysteine
residues, four classes can be distinguished: CXC, CC,
CX3C, and C chemokines. Chemokines exert their
effects via activation of specific receptors that belong
to the superfamily of G protein-coupled transmembrane
spanning receptors. Many chemokine receptors are
promiscuous in ligand interaction, but they are selective
insofar that all accepted ligands are of the same
chemokine class. On the basis of their ligand preference,
receptors have been classified as CC, CXC, CX3C, and
C chemokine receptors. Recently, the IUPHAR nomen-
clature system for chemokines and their receptors has
been published.2
The CC-chemokine receptor 2 (CCR2) is expressed on
monocytes and recognizes the ligands MCP-1, MCP-2,
MCP-3, and MCP-4.3 In the past decade, much evidence
has emerged suggesting a pathological role of MCP-1
and CCR2 in diseases involving chronic inflammation,
in particular atherosclerosis.4 Thus, MCP-1 has been
shown to be highly expressed in human and mouse
atherosclerotic lesions by in situ hybridization or im-
munostaining.5 Furthermore, mice in which the genes
for MCP-1 or CCR2 have been silenced and bred on an
apoE -/- or LDL receptor -/- background6 show a sub-
stantial reduction in atherosclerotic lesions, while over-
expression of mouse MCP-1 (J E) in apoE -/- mice was
Over the past few years, there have been major
advances in the discovery of small molecule antagonists
of chemokine receptors.9 With regard to CCR2, four
chemically distinct series of antagonists have been
reported (Scheme 1). The cinnamide 1 (SB-282241)10 is
a potent functional antagonist of CCR2, but suffers from
lack of selectivity over 5-HT receptors. The spiropiperi-
dine 2 (RS-504393)11 has equal affinity at the R1B
adrenoceptor and also has lower functional potency than
would be predicted from its binding affinity. Several
publications from Takeda have highlighted the CCR5
receptor antagonist 4 (TAK-779),12 which also possesses
significant activity at the CCR2 receptor. Acceptable
bioavailability of this quaternary salt may be an issue,
so clearly the profile of the corresponding free base 3
is also of interest (vide infra). A series of amide der-
ivatives have been reported from Teijin as possessing
CCR2 antagonist activity, with 5 being identified as
a lead compound.13 Both compounds 2 and 5 are
reported to have poor affinities at rodent CCR2 recep-
tors, making biological evaluation of these compounds
difficult.
* To whom correspondence should be addressed. Phone: (+)44 1279
622143. Fax:(+)44 1279 622790. Email:frank_e_blaney@gsk.com.
† Department of Vascular Biology.
Thus, despite intense interest in the development of
CCR2 antagonists as potential drug candidates, the
‡ Department of Discovery Chemistry.
§ Department of Gene Expression Sciences.
10.1021/jm030862l CCC: $25.00 © 2003 American Chemical Society
Published on Web 08/13/2003

Study of the CCR2 Antagonist Binding Site
J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 19 4071
Sch em e 1. Structures of CCR2 Antagonists and Their Published Affinities
Sch em e 2. Synthesis of 1 (SB-282241)a
a
Reagents: (a) Br(CH2)5NHCOCHdCHPh(3,4-Cl2) (7), NaHCO3, DMF, 80 °C, 7 h (89%); (b) pyridine hydrochloride, sealed tube, 130
°C, 48 h (40%).
identification of potent and selective compounds pos-
sessing suitable pharmacokinetic properties still re-
mains an unmet challenge for the pharmaceutical
industry. With the goals of gaining a fuller understand-
ing of the nature of the interactions of diverse ligands
with the CCR2 receptor and thus identifying a phar-
macophore for antagonists at CCR2, we embarked upon
a combined modeling/SDM approach to aid the design
of novel structures possessing a profile superior to those
of the existing leads.
mediate 8, which was demethylated using pyridine
hydrochloride to afford the final product 1.
Compound 2 (RS-504393) was synthesized as shown
in Scheme 3. Ortho lithiation of BOC-protected 2-bromo-
4-methylaniline 9 followed by reaction with N-BOC-4-
piperidone yielded the spiro intermediate 10, which was
deprotected using ethanolic hydrogen chloride to afford
11. Alkylation of 11 with the bromide 12 provided the
target compound 2.
Compounds 3 and 4 were prepared as described by
Takeda.12
Ch em istr y
Compound 5 (Teijin lead) was synthesized as indi-
cated in Scheme 4. Reductive alkylation of 2,4-dimeth-
ylbenzaldehyde 12 with R-3-tert-butyloxycarbonylami-
Compound 1 (SB-282241) was synthesized as shown
in Scheme 2. Alkylation of 4-(5-methoxyindol-3-yl)-
piperidine 614 with the bromide 7 afforded the inter-

4072 J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 19
Sch em e 3. Synthesis of 2 (RS--504393).a
Berkhout et al.
Ta ble 1. Inhibition of Binding and Chemotaxis by CCR2
Antagonistsa
antagonist
binding pKi ( SEM chemotaxis pKb ( SEM
1
2
3
4
5
SB-282241
RS-504393
TAK-779 base 7.54 ( 0.07 (n ) 4)
TAK-779
7.30 ( 0.025 (n ) 28)
6.68 ( 0.13 (n ) 3)
7.71 ( 0.1 (n ) 3)
7.57 ( 0.09 (n ) 3)
8.63 ( 0.11 (n ) 3)
9.53 ( 0.15 (n ) 3)
6.40 ( 0.25 (n ) 3)
8.38 ( 0.14 (n ) 8)
6.54 ( 0.13 (n ) 4)
Teijin lead
a
Binding displacement of 125I-MCP-1 (0.14 nM) was measured
on CHO cell membranes expressing human CCR2. Inhibition of
chemotaxis in human monocytes was measured in the presence
of 1 nM MCP-1. TAK-799 base represents the free base of TAK-
799.
placement data and the chemotaxis data are expressed
as pKi and pKb values, respectively (Table 1),15 and are
in close agreement with literature data, where available.
As can be seen from Table 1, compounds 1-5 display a
range of different binding affinities at CCR2, with 4
(TAK-779) representing the most potent analogue and
5 (Teijin lead) representing the least potent. Unexpect-
edly, the free base 3 is ca. 10-fold less potent than the
quaternized form 4 (TAK-779), perhaps indicating a
difference in the way that these two compounds interact
with the receptor. In the functional assay, the antago-
nist profile of 1-5 was confirmed. However, compounds
1-4 are up to 10-fold more potent than in the binding
assay, but the same rank order of affinities is main-
tained. Interestingly, the Teijin lead 5 does not display
this increased functional potency, possessing a similar
pKi and pKb.
a
Reagents: (a) MeLi, tBuLi, THF, -78 °C, N-tBOC-piperidone,
room temperature, 18 h, (52%); (b) EtOH/HCl, room temperature,
1 h (95%); (c) 4-bromoethyl-5-methyl-2-phenyl-1,3-oxazole 12,
NaHCO3, DMF, 60 °C, 4 h (78%).
nopyrrolidine 13 provided the intermediate 14. Deprotec-
tion of 14 followed by standard carbodiimide coupling
with BOC-glycine afforded the homologated amide 15,
which was similarly deprotected to 16 and coupled with
3-trifluoromethylbenzoic acid, to furnish the target
structure 5.
CCR2 Recep tor Mod elin g a n d Liga n d Dock in g.
The initial model of the CCR2 receptor was based on a
template of frog rhodopsin that, in turn, was derived
from published cryoelectron diffraction data at 6 Å
resolution.16 In these papers, electron density was
depicted as a series of slices at 4 Å intervals going
through the plane of the membrane. Using a recently
published algorithm,17 these density slices were con-
verted into a set of three-dimensional images, which
were used to define the helical axes of the transmem-
brane domain. The rhodopsin sequence was then folded
ab initio around these axes, using experimental infor-
mation on the relative positioning of key residues in the
Biology
Initially, the four chemical series indicated above
were fully profiled on binding and functional studies in
our own laboratories, as comparative data for all
compounds were required prior to our mutation studies
discussed below. Compounds 1-5 were evaluated for
their ability to inhibit MCP-1 binding to membranes of
CHO cells stably expressing CCR2 (12000 receptors/
cell). For functional studies, inhibition of MCP-1-
stimulated chemotaxis of freshly isolated peripheral
human monocytes was investigated. The binding dis-
Sch em e 4. Synthesis of 5 (Teijin lead).a
a
Reagents: (a) R-3-tbutyloxycarbonylamino-pyrrolidine 13, sodium triacetoxyborohydride, 1,2-dichloroethane, room temperature (89%);
(b) (i) EtOH/HCl, room temperature, 1 h (∼100%), (ii) N-tBOC-glycine, diisopropylcarbodiimide, HOBT, DMF, room temperature, 18 h
(56%); (c) EtOH/HCl, room temperature, 1 h (∼100%); (d) 3-trifluoromethylbenzoic acid, diisopropylcarbodiimide, HOBT, DMF, room
temperature, 18 h (53%).

Study of the CCR2 Antagonist Binding Site
J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 19 4073
bundle. The positioning of conserved prolines in trans-
membrane (TM) helices 5 and 6 was facilitated by the
presence of well-defined kinks in the two helical axes.
This TM bundle was then used as a template to
construct the CCR2 receptor model, using the auto-
mated GPCR_Builder program18 developed some years
ago in our laboratory. This whole process, as used in
the construction of neurokinin receptor models, has been
described in greater detail in a recent publication in this
journal.17 Once the TM domain of the CCR2 model was
complete, the extracellular loops (ECLs) were then
added using the following procedure. Essentially, a
distance geometry approach (DGEOM9519) using con-
straints between consecutive helices, the known disul-
fide bond between TM3 and the second extracellular
loop, and already constructed adjacent loops was used
to generate 1000 random conformations for each loop
(in order of increasing size, i.e., ECL1, ECL3, and
ECL2). Each of these was minimized with CHARMm20
using 300 steps of steepest descent (SD) followed by
3000 steps of adopted basis Newton Raphson (ABNR),
keeping the helical ends of the loop region fixed. The
resulting conformations were initially ranked by the
number of violations in backbone dihedral (Ramachan-
dran) space. In the case of the CCR2 loops, a sufficient
number of conformations were found with no violations.
These were ranked in order of increasing energy and
the lowest 10 were subjected to molecular dynamics
simulations using the locally enhanced sampling pro-
cedure21 implemented in CHARMm (five copies, 1 ns
sampling time, and 400 K). The loop family showing the
greatest convergence of conformation and with no
backbone violations was chosen as the final structure.
The diverse ligands under study all contain a basic
positively charged center and this was therefore as-
sumed to interact with the glutamate residue (E291)
on TM7 (see Discussion). The ligands used in this study
were initially built using the Spartan program22 and
were further optimized at the semiempirical AM1 level
using the Vamp program.23 Atom centered point charges
used in the molecular mechanics docking and optimiza-
tion were ab initio natural atomic orbital charges,
calculated with Spartan using a 3-21G* basis set.
Electrostatic potential calculations (see Discussion) were
also calculated with Spartan using this same basis set.
The initial pharmacophore model was generated using
manual rms fitting of key features such as the basic
centers, hydrogen-bond acceptors, and hydrophobic
regions. The result was very similar to the pharma-
cophore generated by Catalyst,24 which is the program
we routinely use for database searching. Automated
docking programs are extensively used within our group
for the placement of ligands, and a number of these were
tried out in the case of the CCR2 model. For reasons
that will be described in the Discussion, they have not
been found to be successful with many 7TM receptors,
and manual docking has therefore been used for the
work described here. Each ligand was docked manually
into the receptor in a variety of starting orientations,
allowing only low-energy starting rotameric conforma-
tions for the ligand. Each placement of the ligand
was accompanied by adjustment of the side chain
dihedrals of the residues in the binding pocket, again
only using the allowed rotamers found in the Karplus
rotamer library.25 The complexes were then fully
minimized (500 steps of SD followed by 5000 steps of
ABNR) with the CHARMm program.20 Helical NOE
distance constraints within the range of 1.8-2.5 Å for
the i to i + 4 hydrogen bonds were used to maintain
the TM bundle. These were generated using a CHARMm
stream file that skipped the ith residue when it was a
proline. Using this distance range allows the helix to
adopt other conformational forms such as a 3/10 helix
conformation or a proline kink. With the exception of
these distance constraints, the receptor-ligand complex
was allowed to fully relax. In all cases, several binding
orientations were found to be possible. The residues in
the model that were found to have significant interac-
tions were used as the basis for deciding upon which
SDM experiments to carry out (see Discussion for
details).
After the initial modeling work, the X-ray crystal
structure of bovine rhodopsin was published.26 The
crystal structure turned out to be remarkably similar
to our model of rhodopsin. TM helices 3, 6, and most of
7 aligned almost perfectly with the crystal based models.
Helices 1 and 2 were correctly oriented toward the inner
face of the bundle but were approximately displaced by
one helical turn in the transmembrane axis. Helices 4
and 5 were correct in their placement in the membrane
axis but the inward faces were out by approximately
60°. In our previous work, this has proved to be
fortuitous because ligand binding very often occurs
primarily between TM3, TM6, and TM7. We have used
these earlier rhodopsin-based models successfully in the
design of ligands for serotonin27 and neurokinin17 recep-
tors. In the case of CCR2, however, significant binding
influences were found in TM1 and TM5. The main
differences affecting the model structure were the
relative placements of the TM1 tyrosine (Y49) and the
TM5 arginine (R206). The crystal structure was used
to rebuild a refined model of the CCR2 receptor, using
the homology building tools within the Quanta pro-
gram.28 Once again the extracellular loop regions were
added using the procedure described earlier. Using the
information from the results of these SDM studies, the
various ligands were redocked into this new model in
orientations that would best explain the experimental
data.
P la sm id Con str u ction a n d Site-Dir ected Mu -
ta gen esis. Full-length cloning of the wild-type MCP1/
CCR2 gene has been published.3 The mutants chosen
on the basis of the early receptor modeling were mainly
in TM7, TM3, and TM1. Their relative positions in these
helices are identified in Table 3 using the numbering
scheme of Ballesteros and Weinstein,29 together with
their actual numbering in the sequence. The latter
numbering will be used to simplify the discussion below,
although the alternative positioning will also be given
where appropriate. Mutants Y49F, Y120A, H121A,
H121F, D284A, Q288A, E291Q, T292A, Y120A/T292A,
and H121A/T292A were introduced using the Quick-
Change PCR-based mutagenesis kit (Stratagene) ac-
cording to the manufacturer’s protocol, using the wild-
type expression vector, pCDNA3.1-MCP1, as the template

4074 J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 19
Berkhout et al.
Ta ble 2. PKd Values for 125I-MCP-1 Binding to Mutant CCR2
tors we had suitable receptor densities and MCP-1
affinities for the purposes of our comparative studies
below.
Receptorsa
receptors/cell
mean pKd
( SEM
pKd comparison
with WT t-testb
receptor variant
( SEM
Having established that MCP-1 binds in a manner
similar to wild type and all mutant receptors, the
binding profile of compounds 1-5 across all mutant
receptors was then investigated, using inhibition of 125I-
MCP-1 binding. The data obtained are shown in Tables
3 and 4, and a chart of key differences in affinities
between wild type and mutant receptors for compounds
1-5 are displayed in Figures 1 (single mutants) and 2
(double mutants).
WT
Y49F
Y120A
H121A
H121F
D284A
Q288A
T290A
E291Q
9027 ( 1815 10.18 ( 0.13
6949 ( 1082 10.01 ( 0.05
ns
ns
ns
ns
ns
4758 ( 1167
9.99 ( 0.01
9984 ( 2950 10.22 ( 0.13
5093 ( 2206 10.47 ( 0.13
7212 ( 1711 10.03 ( 0.12
12440 ( 1696 9.78 ( 0.24
8552 ( 2161 10.10 ( 0.13
P e 0.05
ns
P 0.001
ns
P 0.05
ns
9390 ( 2668
10369 ( 4827 10.09 ( 0.17
5202 ( 1402 9.56 ( 0.25
9152 ( 1043 10.26 ( 0.29
9.02 ( 0.15
T292A
It is readily observed from Table 3 and Figure 1 that
the most striking effects are observed with all five
compounds on E291Q, in which the affinities of 1, 2,
and 5 are reduced by over 100-fold, and the affinities of
3 and 4 are reduced by ca. 30-fold. A small but
significant decrease in affinity was observed with
compounds 1, 2, 3, and 5 at the D284A mutant. As will
be described later (see Discussion), this is probably not
due to direct ligand receptor interactions with this
residue but rather to a decrease in the electronegative
environment of the ligand binding pocket. These data
support the hypothesis that the glutamic acid residue
at position 291, but not the aspartic acid residue at
position 284, is involved in a key interaction with all
ligands examined to date and extends the results
reported from Roche,13 in which E291Q and E291A
mutations were shown to cause a pronounced decrease
in affinity with their lead compounds. The adjacent
threonine residue at position 292 is also an important
residue in the receptor for binding of 1-5, since affini-
ties of 2, 3, and 5 are reduced by 10-30-fold and
affinities of 1 and 4 are reduced by a lesser but
significant amount, in the T292A mutant receptor.
Compounds 2-4 exhibit a major reduction in affinity
at the Y120A mutant receptor (10-100-fold), with a
smaller reduction (<10-fold) observed with compounds
1 and 5. No significant effects were seen at the Q288A
mutant receptor. Major differences between the diverse
series represented by 1-5 are apparent with His 121
and Tyr 49 mutations. Thus, compounds 1, 2, and 5
show typically 10-fold reduction in affinity at the H121A
mutant receptor but possess relatively unchanged af-
Y120A/T292A
H121A/T292A
a
Kd values were determined from binding saturation experi-
ments using single site, saturation curve fitting Grafit 4 software
and converted to pKd values. Data presented are results from at
b
least three independent transfections. Significance compared to
wild type using a t-test for each comparison, assuming equal
variances; ns ) no significant difference.
for mutagenesis. The accuracy of all PCR-derived se-
quenceswas confirmed by dideoxy sequencing of the
mutant plasmids. Constructs were transfected into HEK
293 cells using lipofectAMINE Plus (Life Technologies,
Inc.) according to the manufacturer’s protocol, and after
48 h, cell membranes were isolated for binding studies.
Evalu ation of Receptor Mu tan ts. The eleven CCR2
mutant receptors shown in Table 2 were characterized
by establishing the receptor expression levels and
the affinity for MCP-1 (pKd) and by comparing these
data with those obtained from the wild type CCR2
receptor expressed in the same transient cell line (HEK
293). It can be seen from Table 2 that the receptor
expression levels were broadly similar, ranging from
∼5000 to 12000 receptors per cell, compared to
ca. 9000 receptors per cell for wild type. Generally, the
affinity for MCP-1 did not vary significantly between
wild type and mutant receptors, with the exception of
E291Q, where there was a 10-fold reduction. This
implies that MCP-1 has a significant but not crucial
interaction with the glutamic acid residue at position
291, but not with any of the other amino acid residues
under investigation. Thus, with these 10 mutant recep-
Ta ble 3. Profile of Antagonists Across Mutant CCR2 Receptorsa
pKi
receptor
variant
residue
1
2
3
4
5
positionb
(SB-282241)
(RS-504393)
(TAK-779 base)
(TAK-779)
(Teijin)
WT
Y49F
7.00 ( 0.09
7.78 ( 0.16
6.25 ( 0.19
5.60 ( 0.15
6.78 ( 0.04
6.64 ( 0.06
6.77 ( 0.10
6.89 ( 0.23*
ia
6.66 ( 0.016
7.23 ( 0.15
ia
5.60 ( 0.09
6.35 ( 0.26
6.07 ( 0.03
6.76 ( 0.20
6.28 ( 0.13*
ia
7.16 ( 0.10
6.82 ( 0.06
5.59 ( 0.09
6.52 ( 0.15
7.62 ( 0.03
6.23 ( 0.25
6.64 ( 0.22
6.98 ( 0.47
5.69 ( 0.24
6.12 ( 0.12
8.16 ( 0.12
7.56 ( 0.08
6.98 ( 0.13
7.75 ( 0.10
8.58 ( 0.13
7.90 ( 0.20
8.01 ( 0.13
7.67 ( 0.53
6.83 ( 0.16
7.54 ( 0.13
6.47 ( 0.04
6.50 ( 0.11
5.74 ( 0.17
5.59 ( 0.05
6.18 ( 0.09
6.14 ( 0.20
6.50 ( 0.16
NT
Y1:39
Y3:32
H3:33
H3:33
D7:32
Q7:36
T7:38
E7:39
T7:40
Y120A
H121A
H121F
D284A
Q288A
T290A
E291Q
T292A
ia
6.14 ( 0.10
5.20 ( 0.09
5.11 ( 0.04
a
HEK 293 cells were transiently transfected with point-mutated CCR2. Membranes of transfected cells were incubated with 0.13 nM
125I-MCP-1 in the presence of up to 10 µM antagonist and the pKi was determined. pKi values are expressed as ( standard error of mean
(SEM) from at least three independent transfections (except where indicated: * represents n ) 2). ia ) inactive (no significant displacement
b
of MCP-1 radioligand observed). NT ) not tested. Residues are identified according to the numbering scheme of Ballesteros and
Weinstein.29 The first number (1-7) represents the TM helix in which the residue is situated. The second number is the position of the
residue in that helix, relative to the most highly conserved residue of the rhodopsin family, which is numbered 50. These are the asparagine
of TM1, the arginine of TM3, and the proline of TM7.

Study of the CCR2 Antagonist Binding Site
J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 19 4075
Ta ble 4. Profile of Antagonists across Double Mutant CCR2 Receptorsa
pKi
1
2
3
4
5
receptor variant
(SB-282241)
(RS-504393)
(TAK-779 base)
(TAK-779)
(Teijin)
WT
7.00 ( 0.09
5.61 ( 0.05
5.68 ( 0.63
6.66 ( 0.016
7.16 ( 0.10
ia
5.32 ( 0.08
8.16 ( 0.12
5.87 ( 0.14
6.74 ( 0.10
6.47 ( 0.04
Y120A-T292A
H121A/T292A
ia
ia
ia
ia
a
HEK 293 cells were transiently transfected with point-mutated CCR2. Membranes of transfected cells were incubated with 0.13 nM
125I-MCP-1 in the presence of up to 10 µM antagonist and the pKi was determined. pKi values are expressed as ( standard error of mean
(SEM) from at least three independent transfections. ia ) inactive (no significant displacement of MCP-1 radioligand observed).
F igu r e 1. Summary of pKi differences for compounds 1-5 on mutant CCR2 receptors compared to wild type CCR2a.
also show a pattern consistent with different binding
modes for the different chemical classes. Thus, both 1
and 2 possess enhanced affinity at this Y49F mutant
receptor, while 3 and 4 show slightly reduced affinity,
and 5 displays no change in affinity. Similar effects
were observed with a Y49L mutant receptor (data not
shown).
Since major changes in affinities were observed with
the Y120A, H121A, and T292 mutations, it was of
interest to see if any of these effects were additive by
looking at double mutations (see Table 4 and Figure 2).
In Figure 2 it can be seen that additive effects on
lowering affinity of the antagonists are observed at all
the double mutant receptors Y120A-T292A and H121A-
T292A, with the exception of compound 1 at the
H121A-T292A variant. This provides further evidence
that amino acid residues in positions 120, 121, and 292
are closely involved with the binding of antagonists 1-5
to the CCR2 receptor.
Discu ssion
F igu r e 2. Summary of pKi differences for compounds 1-5
In itia l P h a r m a cop h or e Gen er a tion a n d Liga n d
on double mutant CCR2 receptors compared to wild type
Dock in g. One of the goals of this work was to define a
CCR2a.
pharmacophore for small molecule ligands of the CCR2
finities at H121F. Conversely, binding of compounds 3
and 4 are only affected to a minor extent at the H121A
receptor mutant, but they intriguingly show a signifi-
cant, albeit small, increase in affinity at H121F. This
implies that the Takeda analogues 3 and 4 have a
different mode of interaction with His 121 compared to
the other compounds being studied. Finally, the effects
of the CCR2 antagonists at the Y49F mutant receptor
receptor that could be used in the search of our “in-
house” databases for novel lead series. Despite the
structural diversity of the ligands studied here, it was
recognized that they do have several common features.
In addition to the basic center, all molecules contain at
least two hydrophobic/aromatic sites. They all also
contain an amide type carbonyl that could act as a
hydrogen-bond acceptor. Given that only five compounds

4076 J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 19
Berkhout et al.
F igu r e 3. (a) Pharmacophore model based on rms fit of key presumed binding features. (b) Revised pharmacophore model based
on overlap of the docked structures in the homology receptor model.
were being studied, a manual rms fit was made of these
features. Overlap of the basic nitrogens and two aro-
matic centers was readily identified for the compounds,
but it was impossible to find a fit for the carbonyl
oxygens (see Figure 3a). The lack of overlap of the amido
carbonyls could suggest that they played a structural
rather than hydrogen-bonding role in the receptor, or
simply that they bind to different H-bond donors. The
best hypothesis obtained with the pharmacophore search-
ing program Catalyst24 showed an overlap of the aro-
matic rings of compounds 1, 2, 4, and 5 and predicted
two hydrogen-bond acceptors. Unfortunately, it took no
account of the essential basic nitrogens. Search of our
databases with these pharmacophores failed to produce
any useful hits other than compounds closely related
to the four series used in their generation (results not
shown). We therefore decided to study their binding in
more detail by docking them into the first CCR2
receptor model described in the Experimental Section.
The intention was to confirm the binding hypotheses
for these ligands with SDM experiments.
Ligand docking into receptor or enzyme structures
has been a major topic of research for many years. Due
to its importance in the drug design field, many pro-
grams have been developed for the automated docking
of ligands into these structures. Such programs have
the advantage of being much faster than manual
docking. Because of their speed, however, they often
suffer from the problem of assuming a fixed protein
structure, without allowing for the influence of ligand-
induced fit. A few programs do allow some flexibility of
side chains, but none allow any major movement of the
backbone, which is frequently observed from crystal
structures. They also use some simplified scoring func-

Study of the CCR2 Antagonist Binding Site
J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 19 4077
F igu r e 4. Initial docking of SB-282241 into the theoretical model of CCR2. This shows the binding pocket and the various
residues chosen for the mutation studies.
tion to rank dockings rather than a computationally
more expensive full energy minimization with potential
inclusion of entropy terms. Their great advantage is that
they are fast and can therefore handle hundreds or even
thousands of compounds in a reasonable time. In our
laboratories, we have studied the use of FlexX,30 DOCK,31
GOLD,32 DockIt,33 ICM-Dock,34 and FLO35 with varying
degrees of success. The latter in particular allows some
degree of side chain flexibility in the protein. They
usually work best for systems with large accessible
binding sites, a situation that is not commonly found
in 7TM receptor models. Given the fact that only five
ligands were being studied here and that the intention
was to design suitable SDM experiments and use the
results to “fine-tune” the dockings, it was felt that
manual docking experiments were preferable in this
case. These allowed complete freedom of placement of
each ligand in the light of knowledge from the experi-
mental results as well as the use of optimized ligand
geometries, proper ab initio electrostatic descriptors,
and full controlled flexibility of both the ligand and the
protein.
Ea r ly Dock in g Mod els a n d th e Design of Mu -
ta n ts. At the outset of this study, little evidence was
available to confirm that the various ligands were bound
in the TM domain. However, there is a wealth of SDM
data from other GPCR proteins that supports the
binding of small molecule (ant)agonists within the helix
bundle.36 This, together with the general lipophilic
nature of the compounds, led us to construct early
binding models, where the ligands were placed in the
extracellular part of the TM bundle. As all the com-
pounds studied contain a positively charged basic
center, an acidic residue was sought as the primary
binding site for these. CCR2 has a glutamate at the top
of TM2 close to the first extracellular loop, and an
aspartate and glutamate on the extracellular side of
TM7. Initial extensive docking studies with SB-282241
and RS-504393 suggested that of these three residues,
only the TM7 glutamate could form salt bridge com-
plexes, where the remaining part of the ligands formed
favorable interactions with the TM bundle (see Figure
4 ). Both compounds showed hydrogen-bonding interac-
tions with polar residues in TM7 such as aspartate
D284, glutamine Q288, and threonine T290. T290, for
example, was predicted to hydrogen bond to the es-
sential amide carbonyl of SB-282241. Another interest-
ing observation with SB-282241 was that the indole NH
of SB-282241 formed a hydrogen bond with the hydroxyl
of tyrosine Y49 on TM1. The first round of mutation
studies carried out was based on these two docking
models. The mutants that were constructed were E291Q,
D284A, Q288A, T290A and Y49F. Both compounds
showed a reduction in binding of over 100-fold with the
E291Q mutant, confirming that this is indeed the
primary binding site of these compounds. On the other
hand, with the exception of a small decrease in affinity
with the D284A mutant, no significant effect on binding
was seen with the other TM7 mutants (Q288A and
T290A). These results with the glutamine and aspar-
tate, which are one and two helical turns, respectively,
above the key glutamate, would suggest that the ligands
are buried more deeply in the receptor pocket than the
models predicted. The lack of effect on binding with the
mutant T290A on any of the ligands suggested that
either the amide was playing a structural role in
maintaining some geometrical requirement for fit or

4078 J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 19
Berkhout et al.
possibly that another residue was responsible for hy-
drogen bonding to the carbonyl.
From these results, we had to conclude that while
some aspects of the binding could be explained for each
compound studied, no binding mode was in complete
agreement with all the mutation data. Further refine-
ments in the model were therefore necessary.
The flexible nature of the alkyl linking chain in SB-
282241 allowed it to adopt a second conformation where
it could hydrogen bond to histidine H121 on TM3. The
initial dockings of RS-504393 and TAK-779, however,
suggested that these ligands were not hydrogen bonding
to the histidine but rather formed aromatic-aromatic
π interactions with these ligands. Interestingly, the
Teijin analogue 5 was not predicted to interact with this
histidine. Mutation of the histidine (H121A) did indeed
result in a reduction in binding of all five compounds.
To further test the feasibility of all the ligands forming
aromatic π interactions with H121, the mutation H121F
was carried out. The effect on binding with this mutant
was much less than H121A in the case of compounds 1,
2, and 5. On the other hand, the H121F mutation
resulted in a significant increase of about 0.5 pKi log
units for the Takeda compounds 3 and 4, suggesting
that an important increase in the aromatic π interaction
was occurring in this region. Of particular interest is
the fact that the corresponding residue in the CCR5
receptor is phenylalanine. This is one of the few differ-
ences between CCR2 and CCR5 in the TM domain, and
these compounds are both more potent at CCR5.
Refin em en t of th e Recep tor Liga n d Bin d in g
Mod els. As mentioned previously, it was at this time
that the first crystal structure of a Family A GPCR, viz.
bovine rhodopsin, was published.26 An homology model
of the CCR2 receptor, based on this structure, was
therefore used in an attempt to refine the binding modes
of the five ligands, in a manner that would explain the
observed SDM data. While the theoretically derived
model and the crystal structure based homology model
were quite close in many regions, some notable differ-
ences could be observed.
One such region of difference was the environment
around the histidine H121. The altered orientation of
TM5 in the homology model now meant that this
histidine was involved in a strong hydrogen-bond in-
teraction with the arginine R206 (R5:42) in TM5.
Indeed, this arginine also formed an additional hydro-
gen bond with a second histidine, H202 (H5:38), one
turn above it in TM5 (Figure 5a). Thus, mutation of
H121 to either alanine or phenylalanine would not only
alter potential interactions with the side chain itself,
but might also have an indirect effect by releasing the
arginine to form alternative interactions with ligands
in its own right.
The initial dockings of the Takeda and Teijin com-
pounds postulated a hydrogen-bonding role for threo-
nine T292 to the tetrahydropyran oxygen in 3 and 4 and
to one of the peptide backbone carbonyls of 5. No role
for this residue was predicted, however, for compounds
1 and 2. The mutation T292A resulted in a significant
reduction in binding for all five compounds as can be
seen in Figure 1. The dockings also showed that a large
face to face aromatic interaction was predicted for
compounds 3 and 4 with tyrosine Y120 in TM3. This
was weaker in compounds 1 and 2 and was predicted
not to have an effect with the Teijin compound 5. This
was the last of the mutations to be carried out in the
work described here. As with the T292A mutation, the
change of Y120A resulted in a large decrease in affinity
for all five compounds.
Another crucial difference between the homology
model and its earlier theoretical counterpart was ob-
served in the vicinity of tyrosine Y49. As with most
amino acids, the tyrosine side chain prefers to lie in one
of three rotameric states around the CR-Câ bond. In
the homology model, the tyrosine sits in the most
preferred state and in doing so the hydroxyl forms a
strong hydrogen bond with glutamate E291, thus lock-
ing it in this rotameric form (Figure 5b). The mutation
Y49F removes this hydrogen bond and allows the side
chain to adopt one of the other two rotameric states if
necessary. As will be seen shortly, this fact allowed us
to develop a rationale for the observed increase in
binding of the SB and Roche compounds, 1 and 2. Also
in the homology model the TM7 aspartate D284 forms
a salt bridge with lysine K38 (K1:28) (see Figure 5a).
Mutation of this aspartate as in D284A would increase
the positive charge in that local region and thus
decrease the strength of the salt bridge between the
ligands and glutamate E291, thus explaining the ob-
served slight decrease in affinity. The effect would be
expected to be least in the quaternary center of TAK-
779, where the electropositive charge is distributed over
the molecule. Ab initio electrostatic potential calcula-
tions show that the positive charge on the quaternary
center is, not suprisingly, centered mainly on those
hydrogens â to the nitrogen (see Figure 6)
Perhaps the most suprising result in this series of
mutations was that of the TM1 tyrosine Y49F. The
tyrosine hydroxyl was predicted to bind to the indole
NH; hence, mutation to a phenylalanine should result
in a reduction in observed binding. Instead, a significant
increase was observed with both compounds. Indeed,
with other analogues of SB-282241, this increase was
found to be greater than 100-fold.37 As this effect is
evidently not due to a hydrogen bond between the
receptor and the ligand, it must be the result of a broken
hydrogen bond in the receptor itself. This could result
in either the release of an adjacent side chain that is
involved in a H-bond with the tyrosine or in the release
of the aromatic ring itself, to form an interaction with
the ligand. Alternatively, it could be speculated that
the release of the tyrosine results in the rearrange-
ment of several residues that then form a new binding
pocket for the ligands. In the model, tyrosine Y49 does
form a hydrogen bond with the TM7 glutamine Q288,
but movement of residues upon mutation did not sug-
gest any explanation for the observed increase in
binding.
As with the earlier docking experiments, multiple
initial orientations were considered in the starting
conformations. The “best” orientations were chosen,
however, not only on the basis of calculated interaction
energy but also on the observed interactions with those
residues that had been shown to be important from the
SDM studies.

Study of the CCR2 Antagonist Binding Site
J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 19 4079
F igu r e 5. (a) CCR2 homology model showing the hydrogen-bonding network between histidines H121 and H202 and arginine
R206. At the top the salt bridge between aspartate D284 and lysine K38 can be seen. (b) Binding pocket residues in the CCR2
homology model. Tyrosine Y49 can be seen in its preferred rotameric state, hydrogen bonded to glutamate E291.
Revised Dock in g of RS-504393. As before and with
all compounds, the primary interaction was between the
charged amino group and the carboxyl of glutamate
E291. The bicyclic dihydrobenzoxazinone ring is sand-
wiched between the two aromatic residues on TM3,
forming a coplanar π-stacked interaction with tyrosine
Y120 and an orthogonal aromatic-aromatic interaction
with histidine H121 (Figure 7). Mutation of both these
residues to alanine would be expected to significantly
decrease the binding, and this is indeed observed. The
small loss of binding seen with the H121F mutation
might seem suprising at first, but as previously noted,
the histidine is rigidly held in a hydrogen bond with
arginine R206 (R5:42). Replacement of the histidine
by a phenylalanine causes a small steric clash with
this arginine, which results in the phenylalanine mov-
ing away from an ideal interaction with the ligand. As
can be seen from Figure 7, the oxazole ring forms a
hydrogen bond with threonine T292, and again this
is supported by the decrease in binding observed with
the T292A mutant. The other main observation is
that the 2-phenyl substituent on the oxazole is locked
into the pocket normally occupied by the tyrosine
Y49. It is necessary for this side chain to adopt the
gauche + rotameric state. This is also seen with SB-
282241 (see below) and is supported by the observed
increase in binding of these two compounds with the
Y49F mutant.

4080 J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 19
Berkhout et al.
F igu r e 6. Electrostatic potential of TAK-779 calculated on the electron density surface and displayed in both (a) transparent
and (b) opaque forms. In panel b in particular, the blue electropositive region of the quaternary center can be seen widely diffused
over the surface.
Revised Bin d in g of SB-282241. As with the Roche
compound, it was found that tyrosine Y49 had to
reorient to accommodate the indole ring of SB-282241.
By binding in this pocket it was found that the 5-hy-
droxyl group formed a hydrogen bond with threonine
T292. The protonated nitrogen of the piperidine forms
a particularly strong interaction with glutamate E291,
as evidenced also by the large decrease on binding with
the E291Q mutant. This orientation of the indole
piperidine allows the alkyl chain to adopt a low-energy
conformation that places the dichlorocinnamide group
in the aromatic pocket of TM3 (Figure 8). The tyrosine
ring of Y120 does not form an aromatic-aromatic
interaction with the dichlorophenyl group. Instead, it
forms a hydrophobic interaction with the olefinic bond
of the cinnamide. Thus, the effect of the Y120A mutation
is smaller than is observed with the Roche or Takeda
compounds. The main interaction in this pocket is an
orthogonal π-aromatic binding with histidine H121. This
is evidenced by the loss of binding with H121A and the
much smaller change with the H121F mutants.
Bin d in g of th e Ta k ed a Com p ou n d TAK-779 a n d
Its Ter tia r y Am in o An a logu e. Compared to the
dockings based on the earlier theoretical model of CCR2,
both the Takeda free base (3) and quaternary TAK-779
(4) interacted with glutamate E291 on TM7 (the earlier
model had suggested that E291 bound preferentially to
the amide NH in the quaternary analogue). The tet-
rahydropyran ring of both compounds forms a favorable
hydrophobic interaction with tyrosine Y49 in TM1. This
is further enhanced in the quaternary compound 4,
because the increased positive charge of the hydrogens,
â to the nitrogen, interacts favorably with the oxygen
of the tyrosine hydroxyl (Figure 9). Thus the greater
potency of the quaternary compound is readily ex-
plained. Mutation of this tyrosine to phenylalanine
results in a loss of both effects, which is reflected in the
lowering of observed binding, with the greater loss, as
expected, in the quaternary analogue. Both Takeda
compounds have the tetrahydropyran oxygen hydrogen-
bonded to the TM7 threonine T292. In the tertiary
analogue, the closer proximity of the nitrogen to

Study of the CCR2 Antagonist Binding Site
J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 19 4081
F igu r e 7. Docking of RS-504393 into the CCR2 homology model. Note in particular the less favored rotameric state of tyrosine
Y49 necesssary to accommodate the phenyloxazole moiety in its pocket.
F igu r e 8. Revised docking of SB-282241 into CCR2. Once again, tyrosine Y49 must adopt the less favored gauche+ rotameric
form to accommodate the indole. The 5-hydroxy group forms a hydrogen bond with threonine T292.
glutamate E291 means that the ring oxygen is also
closer to threonine T292 than in the quaternary salt.
This is reflected in the observed smaller loss of binding
of TAK-779 with the T292A mutation. The biaryl rings
are sandwiched between the tyrosines Y120 and Y259
(Y6:51) (on TM6) and to a lesser extent histidine H121.
The interaction of this histidine with the TM5 arginine
R206 means that there is only a weak aromatic-
aromatic binding between the ligands and the histidine.
Mutation of this residue to alanine would leave a cavity,

4082 J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 19
Berkhout et al.
F igu r e 9. Binding of TAK-779 to CCR2. The π π aromatic interaction of the ligand with histidine H121 is weakened by the
latter’s H-bond to arginine R206. The corresponding phenylalanine in CCR5 does not have this interaction and can therefore
adopt the alternative rotameric conformation and increase the π-π interaction.
which would result in a lowering of binding, although
as observed, this effect is less than with the other
compounds that have a stronger direct binding to
H121. On the other hand, the mutation H121F would
allow the phenylalanine to stack with the biaryl
rings, increasing the binding. This is reflected in the
small but significant increase observed on binding the
Takeda compounds. In the CCR5 receptor histidine
H121 is a phenylalanine. The H121F switch is one of
the main differences between the CCR2 and CCR5
receptors in the transmembrane domain, and is re-
flected in the increased potency of the Takeda com-
pounds at the latter target, for which they were origi-
nally synthesized.
observed between this compound and the Y120A and
H121A mutants, compared to the other compounds
studied.
It can be concluded from the above results that the
revised docking hypotheses for the five compounds
studied are in good agreement with all the SDM
experimental data. In particular, the binding modes for
SB-282241 and RS-504393 explain the unusual increase
with the Y49F mutant. The decrease in binding of the
Takeda compounds with this mutant and the slight but
significant increase with H121F can also clearly be
explained. The binding pocket available for ligands in
the CCR2 is quite extensive, and while all the com-
pounds studied share common features, such as interac-
tions with glutamate E291 and threonine T292, they
differ in the way they interact with the aromatic pocket
around TM3 and with the region at the top of TM1. The
clear definition of this pocket with the combination of
docking and SDM experiments has enabled us to
develop a reliable model of the CCR2 receptor. In
particular, it has allowed us to refine the original
pharmacophore model used in the database searches.
Thus, by extracting the five structures as they are
docked into the receptor, the new overlay (Figure 3b)
shows significant differences from that produced previ-
ously (Figure 3a). In particular, the nitrogen atoms in
the five ligands are not superimposed but can be as
much as 5.4 Å apart. Although they all bind to the same
glutamate, this is not suprising in retrospect, because
the glutamate is seen to adopt two low-energy rotameric
states depending on the ligands bound. The models and
revised pharmacophore are being used currently in the
design of novel antagonist ligands for this important
pharmaceutical target.
Revised Bin d in g of th e Teijin Com p ou n d 5. The
Teijin lead compound was found to form a series of
hydrogen-bonding interactions along its peptide back-
bone. Thus, in addition to the salt bridge with the
pyrrolidine nitrogen, glutamate E291 also formed an
interaction with the amide NH attached to the ring. The
carbonyl of this amide also formed a hydrogen bond with
the hydroxyl of tyrosine Y49, but this was at the expense
of breaking the interaction between Y49 and E291, with
no net gain in free energy. Thus, this is in keeping with
the observed lack of effect with the Y49F mutation. The
electronegative trifluoromethyl group has a favorable
electrostatic interaction with lysine K38 (K1:28), which
itself forms the salt bridge to aspartate D284. Threonine
T292 also formed a strong hydrogen bond with the
second amide carbonyl (Figure 10). The 2,4-dimeth-
ylphenyl ring does not extend fully into the TM3
aromatic pocket, and the main interactions are more
hydrophobic in nature, between one of the methyl
groups and the rings of tyrosine Y120 and histidine
H121. This is reflected in the smaller drop in binding

Study of the CCR2 Antagonist Binding Site
J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 19 4083
F igu r e 10. Docking of compound 5 (Teijin) in the CCR2 homology model. Note the hydrogen bonds from threonine T292 and
tyrosine Y49 to the ligand backbone carbonyls. There is also a favorable electrostatic interaction between the trifluoromethyl
group and lysine K38.
were determined using a Fisons VG 302 single quadrupole
mass spectrometer. Elemental analysis were within 0.4% of
the theoretical values. Tetrahydrofuran (THF) and N,N-
dimethylformamide (DMF) were purchased from Aldrich as
anhydrous solvents. Other solvents and reagents were of
standard commercial grade and used without purification.
Organic extracts were dried over anhydrous sodium sulfate
before evaporation at reduced pressure. Chromatography was
performed using commercially available Biotage silica car-
tridges or using Merck Art. 7734 silica gel.
Su m m a r y a n d Con clu sion s
Using a theoretically derived model of the CCR2
receptor, the binding of the various key antagonist series
described to date was studied, and plausible binding
modes were defined for each. These were used as the
basis for the design of a number of SDM experiments,
which would verify (or not) the proposed hypotheses.
The results of the SDM showed that while various
aspects of the ligand interactions could be rationalized,
the binding modes overall were clearly not in agreement
with the experimental data and further refinements in
the model would be necessary. The publication of a
crystal structure of the related G-protein-coupled recep-
tor bovine rhodopsin allowed us to construct a new
model of CCR2. While the two models were close in
many aspects, some key differences were apparent and
these were shown to have important influences on the
binding modes. The various ligands were reevaluated
in the light of this model and the experimental SDM
results. In this way plausible binding modes were
developed for all compounds, which were in complete
agreement with the SDM data. The new models high-
light a potential problem in the design of pharmacoph-
ores in that the latter often overemphasize the impor-
tance of direct overlap of key binding features such as
basic centers, because they assume a common binding
mode with no major motions of the protein molecule.
Currently these models are being used in the design of
further novel antagonists. This work and the effects of
these compounds on the various mutant receptors will
be described in a future paper in this series.
(E)-N-(5-Br om op en t yl)-3-(3,4-d ich lor op h en yl)a cr yla -
m id e (7). A solution of 5-aminopentanol (42.3 g, 0.41 mol),
N,N-diisopropylethylamine (72 mL), and dichloromethane (750
mL) was cooled (ice bath) and treated with 3,4-dichlorophe-
nylacrylolyl chloride (47.7 g, 0.37 mol) in dichloromethane (150
mL) over 10 min and then the solution was stirred for a further
20 min. The solvent volume was reduced to (ca. 300 mL), water
(1 L) was added, and the solid was collected (99.65 g). A sample
of this solid (15.1 g, 0.05 mol) was suspended in dichlo-
romethane (250 mL) containing carbon tetrabromide (18.3 g,
0.055 mol). Triphenylphosphine (14.1 g, 0.055 mol) was added
in portions over 1 h, then the solution was stirred for 3 h. The
solution was poured onto a flash silica column that was eluted
with dichloromethane to give the title as a white solid that
crystallized from hexane (15.96 g): 1H NMR (CDCl3) δ 1.6 (4H,
m), 1.9 (2H, m), 3.4 (4H, m), 5.82 (1H, br t), 6.36 (1H, d, J )
17 Hz), 7.31 (1H, m), 7.46 (1H, d, J ) 17 Hz), 7.56 (2H, m);
MS m/e 364, 366 (MH+).
(E)-3-(3,4-Dich lor oph en yl)-N-{5-[4-(5-m eth oxy-1H-in dol-
3-yl)-p ip er id in -1-yl]p en tyl}a cr yla m id e (8). A stirred mix-
ture of 4-(5-methoxyindol-3-yl)piperidine (1.15 g, 5.0 mmol),
bromide 7 (2.0 g, 5.5 mmol and sodium hydrogen carbonate
(1.0 g) in dimethylformamide (30 mL) was heated at 80 °C for
7 h. The reaction mixture was allowed to cool, poured onto
brine, and extracted twice with ethyl acetate. The combined
organic layers were washed with brine, dried, and evaporated
to afford the crude product. Chromatography on a Biotage
cartridge eluting with 3-6% methanol in dichloromethane
containing 0.5% aqueous ammonia afforded 2 (2.3 g, 89%) as
an off-white solid: 1H NMR (CDCl3) δ 1.39-1.55 (2H, m),
Exp er im en ta l Section
Ch em istr y. NMR spectra were determined with Bruker
AC-200, AC-250, or AM-400 spectrometers using tetrameth-
ylsilane as internal standard. Electron impact mass spectra

4084 J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 19
Berkhout et al.
1.55-1.70 (4H, m), 1.81-2.3 (6H, m), 2.4-2.51 (2H, m), 2.75-
2.91 (1H, m), 3.12 (2H, d, J ) 12 Hz), 3.42 (2H, q, J ) 7 Hz),
3.88 (3H, s), 5.89 (1H, b s), 6.39 (1H, d, J ) 15 Hz), 6.86 (1H,
dd, J ) 9 and 2 Hz), 6.96 (1H, d, J ) 2 Hz), 7.08 (1H, d, J )
2 Hz), 7.2-7.33 (2H, m), 7.40 (1H, d, J ) 8 Hz), 7.51 (1H, d, J
) 15 Hz), 7.57 (1H, m), 7.89 (1H, b s); MS m/e 514, 516 (MH+).
washing the organic extracts with water followed by evapora-
tion afforded the crude product. Purification was effected by
chromatography on
a Biotage cartridge eluting with 5%
methanol in dichloromethane containing 0.5% aqueous am-
monia afforded 14 (2.9 g, 89%): 1H NMR (CDCl3) δ 1.43 (9H,
s), 1.56-1.58 (1H, m), 2.1-2.3 (2H, m), 2.33 (3H, s), 2.31 (3H,
s), 2.51-2.58 (2H, m), 2.76 (1H, m), 3.53 (2H, s), 4.14 3.53 (2H,
s), 4.14 (1H, b s), 4.84 (1H, b s), 6.9-7.03 (2H, m), 7.23 (1H, d,
J ) 8 Hz); MS m/e 304 (MH+).
(E)-3-(3,4-Dich lor oph en yl)-N-{5-[4-(5-h ydr oxy-1H-in dol-
3-yl)-p ip er id in -1-yl]p en tyl}a cr yla m id e (1). A stirred mix-
ture of 2 (1.0 g, 1.9 mmol) and pyridine hydrochloride (3.0 g)
was heated in a sealed tube to 130 °C for 48 h. The reaction
mixture was digested in methanol/chloroform, washed with
aqueous ammonia and brine, dried, and evaporated to afford
the crude product. Chromatography on a Biotage cartridge
eluting with 6% methanol in dichloromethane containing 0.5%
aqueous ammonia afforded 2 (375 mg, 40%) as an off-white
solid: 1H NMR (DMSO) δ 1.2-1.8 (7H, m), 1.8-2.1 (4H, m),
2.2 (2H, m), 2.6-2.7 (1H, m), 2.97 (2H, d, J ) 10 Hz), 3.1-3.2
(3H, m), 6.58 (1H, dd,, J ) 9 and 2 Hz), 6.71 (1H, d, J ) 15
Hz), 6.83 (1H, d, J ) 2 Hz), 6.95 (1H, d, J ) 2 Hz), 7.10 (1H,
d, J ) 10 Hz), 7.40 (1H, d, J ) 15 Hz), 7.58 (1H, dd, J ) 9 and
2 Hz), 7.69 (1H, d, J ) 10 Hz), 7.85 (1H, d, J ) 2 Hz), 8.10
(1H, t, J ) 7 Hz), 8.50 (1H, s), 10.41 (1H, s); MS m/e 500, 502
(MH+). Anal. (C27H31N3O2Cl2) C, H, N.
1′-[Spir o{4H-3,1-6-m eth ylben zoxazin e-4,4′-piper idin e}-
2(1H)-on e]ca r ba m ic Acid ter t-Bu tyl Ester (10). A solution
of N-BOC-2-bromo-4-methylaniline (2.86 g, 10 mmol) in dry
tetrahydrofuran was cooled to 0 °C under argon. 1.5 M
Methyllithium in tetrahydrofuran (8 mL, 12 mmol) was added
dropwise over 5 min, and the mixture stirred for 30 min. After
cooling the solution to -78 °C, 1.7 M tert-butyllithium (13 mL,
20 mmol) was added dropwise, and stirring continued for 1 h.
A solution of N-BOC-4-piperidone (1.99 g, 10 mL) in tetrahy-
drofuran (5 mL) was added, and the solution was then allowed
to warm to room temperature. Potassium tert-butoxide (50 mg)
was added, and the mixture stirred for 17 h at room temper-
ature. The reaction mixture was poured onto brine and ex-
tracted twice with ethyl acetate. The combined organic extracts
were washed with brine, dried and evaporated to afford the
crude product. Chromatography on silica using 30-60% ethyl
acetate in hexane afforded 10 (1.7 g, 52%) as an off-white
solid: 1H NMR (CDCl3) δ 1.49 (9H, s), 1.8-2.1 (4H, m), 2.31
(3H, s), 3.33 (2H, m), 4.14 (2H, m), 6.72 (1H, d, J ) 9 Hz),
6.91 (1H, s), 7.05 (1H, m), 8.06 (1H, b s); MS m/e 331 (M-H+).
6-Meth ylspir o{4H-3,1-ben zoxazin e-4,4′-piper idin}-2(1H)-
on e Hyd r och lor id e (11). The N-BOC intermediate 10 (1.7
g, 5.1 mmol) was dissolved in ethanol saturated with hydrogen
chloride. After stirring for 1 h, the solution was evaporated to
dryness to afford 11 (1.3 g, 95%) as a light brown foam: 1H
NMR (DMSO) δ 2.1-2.4 (4H, m), 2.27 (3H, s), 3.0-3.2 (2H,
m), 3.25-3.4 (2H, m), 8.30 (1H, d, J ) 8 Hz), 6.98 (1H, s), 7.12
(1H, m), 9.0 (2H, b s), 10.25 (1H, b s); MS m/e 233 (MH+).
1′-[2-(5-Met h yl-2-p h en yloxa zol-4-yl)et h yl]-6-m et h yl-
sp ir o{4H-3,1-ben zoxa zin e-4,4′-p ip er id in }-2(1H)-on e (2).
The amine 11 (1.0 g, 3.7 mmol) was dissolved in dimethylfor-
mamide (20 mL) containing 4-bromoethyl-5-methyl-2-phenyl-
1,3-oxazole (1.0 g, 3.7 mmol) and sodium hydrogen carbonate
(1.0 g). The reaction mixture was heated to 60 °C for 4 h,
allowed to cool, poured onto brine, and extracted twice with
ethyl acetate. The combined organic layers were washed with
brine, dried, and evaporated to afford the crude product.
Chromatography on a Biotage cartridge eluting with 5%
methanol in dichloromethane containing 0.5% aqueous am-
monia afforded 2 (1.2 g, 78%) as an off-white solid: 1H NMR
(CDCl3) δ 2.1-2.2 (4H, m), 2.30 (3H, s), 2.35 (3H, s), 2.6-3.95
(8H, m), 7.15 (1H, d, J ) 8 Hz), 6.9-7.05 (2H, m), 7.4-7.44
(3H, m), 7.98-8.0 (2H, m), 8.29 (1H, b s); MS m/e 418 (MH+).
Anal. (C25H27N3O3) C, H, N.
{[(R)-1-(2,4-Dim eth ylben zyl)p yr r olid in -3-ylca r ba m oyl]-
m eth yl}ca r ba m ic Acid ter t-Bu tyl Ester (15). The N-BOC
compound 14 (2.9 g, 9.5 mmol) was dissolved in ethanol
saturated with hydrogen chloride. After stirring for 1 h, the
solution was evaporated to dryness to afford the corresponding
amine dihydrochloride (2.5 g, ∼100%) as a light brown foam.
This material was dissolved in dimethylformamide (30 mL)
and added to a solution of diisopropylethylamine (2.6 mL),
N-BOC-glycine (1.84 g, 10.5 mmol), N-hydroxybenzotriazole
(1.6 g, 10.5 mmol), and diisopropylcarbodiimide (1.3 g, 10.5
mmol) in dimethylformamide (20 mL). The solution was stirred
for 18 h and then poured onto brine. The mixture was
extracted with ethyl acetate, and the combined organic layers
were washed with sodium carbonate solution and brine, dried,
and evaporated to afford the crude product. Purification was
effected by chromatography on a Biotage cartridge eluting with
2-5% methanol in dichloromethane containing 0.5% aqueous
ammonia to afford 15 (1.9 g, 56%) as an off-white foam: 1H
NMR (CDCl3) δ 1.44 (9H, s), 1.6-1.63 (1H, m), 2.2-2.26 (2H,
m), 2.29 (3H, s), 2.32 (3H, s), 2.5-2.54 (2H, m), 2.81-2.87 (1H,
m), 3.54 (2H, s), 3.72 (2H, d, J ) 6 Hz), 4.44 (1H, m), 5.04
(1H, m), 6.30 (1H, m), 6.92-6.97 (2H, m), 7.10 (1H, d, J ) 8
Hz); MS m/e 362 (MH+).
2-Am in o-N-[(R)-1-(2,4-d im eth ylben zyl)p yr r olid in -3-yl]-
a ceta m id e Dih yd r och lor id e (16). The N-BOC compound 15
(1.9 g, 5.26 mmol) was dissolved in ethanol (40 mL) saturated
with hydrogen chloride. After stirring for 2 h, the solution was
evaporated to dryness to afford the corresponding amine
dihydrochloride 16 (1.8 g, ∼100%) as a light brown foam: MS
m/e 262 (MH+); This material was used directly in the next
step.
N-{[(R)-1-(2,4-Dim eth ylben zyl)p yr r olid in -3-ylca r ba m -
oyl]m eth yl}-3-m eth ylben za m id e (5). A solution of the
amine 16 (334 mg, 1.0 mmol) in dimethylformamide (3 mL)
was added to a solution of diisopropylethylamine (0.6 mL),
3-trifluoromethylbenzoic acid (210 mg, 1.1 mmol), N-hydroxy-
benzotriazole (170 mg, 1.1 mmol), and diisopropylcarbodiimide
(140 mg, 1.1 mmol) in dimethylformamide (5 mL). The solution
was stirred for 18 h and then poured onto brine. The mixture
was extracted with ethyl acetate, and the combined organic
layers were washed with sodium carbonate solution and brine,
dried, and evaporated to afford the crude product. Purification
was effected by chromatography on a Biotage cartridge eluting
with 2-5% methanol in dichloromethane containing 0.5%
aqueous ammonia to afford 5 (230 mg, 53%) as an off-white
foam: 1H NMR (CDCl3) δ 2.2-2.4 (3H, m), 2.29 (3H, s), 3.30
(3H, s), 2.5-2.62 (2H, m), 2.85-2.91 (1H, m), 3.55 (2H, q, J )
13 Hz), 4.08 (2H, d, J ) 8 Hz), 4.79 (1H, m), 6.45 (1H, d, J )
13 Hz), 6.9-7.15 (4H, m), 7.54 (1H, dd, J ) 12 Hz), 7.75 (1H,
d J ) 12 Hz), 7.98 (1H, d, J ) 12 Hz), 8.09 (1H, s); MS m/e
434 (MH+). Anal. (C23H26N3O2F3) C, H, N.
Site-Dir ected Mu ta gen esis. All oligonucleotides were
synthesized by Sigma-Genosys Ltd (Pampisford, U.K.). Full-
length cloning of the wild-type MCP-1/CCR2 gene has been
published.3 Mutants Y49F, Y120A, H121A, H121F, D284A,
Q288A, E291Q, T292A, Y120A/T292A, and H121A/T292A were
introduced using the QuickChange PCR-based mutagenesis
kit (Stratagene) according to the manufacturer’s protocol, using
the wild-type expression vector, pCDNA3.1-MCP1, as the
template for mutagenesis. The accuracy of all PCR-derived
sequences was confirmed by dideoxy sequencing of the mutant
plasmids.
[(R)-1-(2,4-Dim et h ylb en zyl)p yr r olid in -3-yl]ca r b a m ic
Acid ter t-Bu tyl Ester (14). A solution of 2,4-dimethylben-
zaldehyde (1.5 g, 11 mmol), R-3-tert-butyloxycarbonylaminopy-
rrolidine (2.0 g, 10.7 mmol) in dichloroethane (40 mL) was
treated with sodium triacetoxyborohydride (4.5 g). The mixture
was stirred for 3 h and then poured onto aqueous sodium
carbonate solution. Extraction with dichloromethane and
Primer sequences (base changes introducing the mutations
are italic):

Study of the CCR2 Antagonist Binding Site
J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 19 4085
Y49FFOR
CTCCTGCCTCCGCTCTTCTCGCTGGTGTTCATC
Y49FREV
GATGAACACCAGCGAGAAGAGCGGAGGCAGGAG
TTATTCACAGGGCTGGCTCACATCGGTTATTTTGG
CCAAAATAACCGATGTGAGCCAGCCCTGTGAATAA
TTCACAGGGCTGTATGCCATCGGTTATTTTGGC
GCCAAAATAACCGATGGCATACAGCCCTGTGAA
TTCACAGGGCTGTATTTCATCGGTTATTTTGGC
GCCAAAATAACCGATGAAATACAGCCCTGTGAA
CCAGTCAACTGGCCCAAGCCACGCAGG
CCTGCGTGGCTTGGGCCAGTTGACTGG
GACCAAGCCACGGCGGTGACAGAGACTC
GAGTCTCTGTCACCGCCGTGGCTTGGTC
GCCACGCAGGTGACACAGACTCTTGGGATGACT
AGTCATCCCAAGAGTCTGTGTCACCTGCGTGGC
CACGCAGGTGACAGAGGCTCTTGGGATGACTCAC
GTGAGTCATCCCAAGAGCCTCTGTCACCTGCGTG
CAAATTATTCACAGGGCTGGCCGCCATCGGTTATTTTGGCGG
CCGCCAAAATAACCGATGGCGGCCAGCCCTGTGAATAATTTG
Y120AFOR
Y120AREV
H121AFOR
H121AREV
H121FFOR
H121FREV
D284AFOR
D284AREV
Q288AFOR
Q288AREV
E291QFOR
E291QREV
T292AFOR
T292AREV
H121A/Y120AFOR
H121A/Y120AREV
Tr a n sfection a n d Mem br a n e P r ep a r a tion . Mutant and
wild type CCR2 receptor cDNA, subcloned into the mammalian
expression vector pcNA3.1, were transiently transfected into
50-80% confluent HEK-293T cells using lipofectAMINE Plus
(Life Technologies, Inc.) together with a cocktail of plasmids
expressing human G-proteins (Gqi5, Gqo5, Gγs5, GR15, and
GR16 all in pCDN expression vector) according to the manu-
facturer’s protocol. Forty-eight hours posttransfection, cells
were harvested and suspended into 10 mM HEPES pH 7 with
0.25 M sucrose, 1 mM EDTA, and 1 µg/mL each of pepstatin,
antipain, aprotonin, and leupeptide. Cells were disrupted by
sonication (Soniprep 150, GallenKamp) and cell debris was
removed by low-speed centrifugation, 500g for 15 min, 4 °C
(Beckman GS-6R centrifuge). The resulting supernatant was
spun at 105 000g for 22 min, 4 °C (Beckman L8-70M Ultra
centrifuge), the pellet resuspended in the same buffer, and the
centrifugation was repeated. The final membrane was resus-
pended in assay buffer (50 mM Hepes pH 7.4, 1 mM CaCl2, 5
mM MgCl2) and stored at -80 °C.
Refer en ces
(1) Saunders: J .; Tarby, C. M. Drug Discovery Today 1999, 4, 80-
92.
(2) Murphy, P. M.; Baggiolini, M.; Charo, I. F.; Hebert, C. A.; Horuk,
R.; Matsushima, K.; Miller, L. H.; Oppenheim, J . J .; Power, C.
A. International Union of Pharmacology. XXII. Nomenclature
for Chemokine Receptors. Pharmacol. Rev. 2000, 52, 145-176.
(3) Berkhout, T. A.; Sarau, H. M.; Moores, K.; White, J . R.;
Elshourbagy, E.; Apelbaum, E.; Makwana, J .; Foley, J . J .;
Schmidt, D. B.; Imburgia, C.; McNulty, D.; Mathews, J .;
O’Donnell, K.; O’Shannessy, D.; Scott, M.; Groot, P. H. E.;
Macphee, C. Cloning, in Vitro Expression, and Functional
Characterisation of a Novel Human CC Chemokine of the
Monocyte Chemotactic Protein (MCP) Family (MCP-4) that
Binds and Signals through the CC Chemokine Receptor 2B. J .
Biol. Chem. 1997, 272, 16404-16413.
(4) (a) Ross, R. Atherosclerosis- An Inflammatory Disease. New
England J Med. 1999, 340, 115-126. (b) Reape, T. J .; Groot, P.
H. E.; Chemokines and Atherosclerosis. Atherosclerosis 1999,
147, 213-215.
(5) (a) Yla-Herttuala, S.; Lipton, B. A.; Rosenfeld, M. E.; Sarkioja,
T.; Yoshimura, T.; Leonard, E. J .; Witztum, J . L.; Steinberg, D.
Expression of Monocyte Chemotactic Protein in Macrophage-
rich Areas of Human and Rabbit Atherosclerotic Lesions. Proc.
Natl. Acad. Sci. U.S.A. 1991, 88, 5252-5256. (b) Takeya, M.;
Yoshimura, T.; Leonard, E. J .; Takahashi, K. Detection of
Monocyte Chemoattractant protein-1 in Human Atherosclerotic
Lesions by an Anti-monocyte Chemoattractant Protein-1 Mono-
clonal Antibody. Hum. Pathol. 1993, 24, 534-539. (c) Nelken,
N. A.; Coughlin, S. R.; Gordon, D.; Wilcox, J . N. Monocyte
Chemotactic Protein-1 in Human Atheromatous Plaques. J . Clin.
Invest. 1991, 88, 1121-1127. (d) Rayner, K.; Van Eersel, S.;
Groot, P. H. E.; Reape, T. J . Localisation of mRNA for J E/MCP-1
and its Receptor CCR2 in Atherosclerotic Lesions of the ApoE
Knockout Mouse. J . Vasc. Res. 2000, 37, 93-102.
(6) (a) Gu, L.; Okada, Y.; Clinton, S. K.; Gerard, C.; Sukhova, G.
K.; Libby, P.; Rollins, B. J . Absence of Monocyte Chemotactic
Protein-1 Reduces Atherosclerosis in Low-Density Lipoprotein-
deficient Mice. Mol. Cell. 1998, 2, 275-281. (b) Boring, L.;
Gosling, J .; Cleary, M.; Charo, I. F. Decreased Lesion Formation
in CCR2-/- Mice Reveals a Role for Chemokines in the Initiation
of Atherosclerosis. Nature 1998, 394, 894-897.
Ra d ioliga n d Bin d in g Assa y. Saturation binding studies
were performed with increasing concentrations of 125I-MCP-1
(0.02-2 nM, Bolton and Hunter labeled, Amersham, specific
activity ) 2000 Ci/mmol) to membrane protein, (cell equivalent
of 200 000 cells/well) in binding buffer with 0.5% albumin
fraction V (fatty acid free) and incubated for 2 h at room
temperature. Membranes were harvested as described.10 The
Kd and Bmax were determined by single site saturation curve
fitting using GraFit 4 (Erithacus Software Ltd.). Nonspecific
binding was determined by measuring binding of 125I-MCP-1
to untransfected HEK 293 cell membranes or by incubation
in the presence of 300 nM MCP-1. The binding displacement
assay was performed as described previously.10
Ch em ota xis Assa y. The chemotaxis assay was essentially
carried out as previously described.3 Vehicle or compound were
added/diluted from a 10 mM stock solution in DMSO, giving
final concentrations in the range 10 nM to 10 µM, and 1 nM
MCP-1 was used as the agonist.
(7) Aiello, R. J .; Bourassa, P.-A. K.; Lindsey, S.; Weng, W.; Natoli,
E.; Rollins, B. J .; Milos, P. M. Monocyte Chemoattractant
Protein-1 Accelerates Atherosclerosis in Apolipoprotein E-defi-
cient Mice. Arteriosclerosis 1999, 19, 1518-1525.
(8) Gong, J .-H.; Ratkay, L. G.; Waterfield, J . D.; Clark-Lewis, I. An
Antagonist of Monocyte Chemoattractant Protein-1 (MCP-1)
Inhibits Arthritis in the MRL-lpr Mouse Model. J . Exp. Med.
1997, 186, 131-137.
Sta tistica l An a lysis. Differences between data for mutant
and WT receptors for individual compounds were calculated
along with their statistical significance based on t-tests using
estimates of the standard deviation of the pIC50/pKi calculated
by pooling the within-compound variation across all com-
pounds for a particular mutant.
(9) Trivedi, B. K.; Low, J . E.; Carson, K.; LaRosa, G. J . Chemok-
ines: Targets for Novel Therapeutics. In Annual Reports in
Medicinal Chemistry Doherty, A. M., Ed.; Academic Press: New
York, 2000; Vol. 35, pp 191-200.
(10) Forbes, I. T.; Cooper, D. G.; Dodds, E. K.; Hickey, D. M. B.; Ife,
R. J .; Meeson, M.; Berkhout, T.; Gohil, J .; Moores, K. CCR2B
Receptor Antagonists: Conversion of a Weak HTS Hit to a
Potent Lead Compound. Bioorg. Med. Chem. Lett. 2000, 10,
1803-1806.
Ack n ow led gm en t . We are grateful to Deirdre
Hickey for helpful discussions and Lynne Roxbee-Cox
for skilled technical assistance. Special thanks are
also due to Ian Wall for his help with the Catalyst
program.

4086 J ournal of Medicinal Chemistry, 2003, Vol. 46, No. 19
Berkhout et al.
(11) Mirzadegan, T.; Diehl, F.; Ebi, B.; Bhakta, S.; Polsky, I.;
McCarley, D.; Mulkins, M.; Weatherhead, G. S.; Lapierre, J .-
M.; Dankwardt, J .; Morgans, D.; Wilhelm, R.; J arnagin, K.
Identification of the Binding Site for a Novel Class of CCR2B
Chemokine Receptor Antagonists. J . Biol. Chem. 2000, 275,
25562-25571.
(12) (a) Shiraishi, M.; Kitayoshi, T.; Aramaki, Y.; Honda, S. WO
Patent 99/ 32468. (b) Baba, M.; Nishimura, O.; Kanzaki, N.;
Okamoto, M.; Sawada, H.; Iizawa, Y.; Shiraishi, M.; Aramaki,
Y.; Okonogi, K.; Ogawa, Y.; Meguro, K.; Fujino, M. A Small
Molecule Non-peptide CCR5 Antagonist with Highly Potent and
Selective Anti-HIV Activity. Proc. Natl. Acad. Sci. U.S.A. 1999,
96, 5698-5703. (c) Shiraishi, M.; Aramaki, Y.; Seto, M.; Imoto,
H.; Nishikawa, Y.; Kanzaki, N.; Okamoto, M.; sawada, H.;
Nishimura, O.; Baba, M.; Fujino, M. Discovery of Novel, Potent,
and Selective Small-molecule CCR5 Antagonists as Anti-HIV
Agents: Synthesis and Biological Evaluation of Anilide Deriva-
tives with a Quaternary Ammonium Moiety. J . Med. Chem.
2000, 43, 2049-2063.
J . Am. Chem. Soc. 1990, 112, 9161-75. (b) Roitberg, A.; Elber,
R. Modeling side chains in peptides and proteins: Application
of the locally enhanced sampling and the simulated annealing
methods to find minimum energy conformations. J . Chem. Phys.
1991, 95, 9277-87.
(22) SPARTAN SGI, version 5.1.3; Wave function Inc., 18401 Von
Karman, Suite 370, Irvine, CA 92612, 1999.
(23) Clark, T.; Alex, A.; Beck, B.; Chandrasekhar, J .; Gedeck, P.;
Horn, A.; Hutter, M.; Rauhut, G.; Sauer, W.; Steinke, T. Vamp
7.0; Oxford Molecular Ltd., 1998.
(24) Catalyst, v4.7 (579); Accelrys Inc.: San Diego, CA, 2002.
(25) Dunbrack, R. L., J r.; Karplus, M. Backbone-dependent rotamer
library for proteins. Application to side-chain prediction. J . Mol.
Biol. 1993, 230, 543-74.
(26) Polczewski, K.; Kumasaka, T.; Hori, T.; Behnke, C. A.; Mo-
toshima, H.; Fox, B. A.; Le Trong, I.; Teller, D. C.; Okada, T.;
Stenkamp, R. E.; Yamamoto, M.; Miyano, M. Crystal structure
of rhodopsin: A G protein-coupled receptor. Science (Washington,
D. C.) 2000, 289, 739-745.
(13) (a) Shiota, T. WO Patent 99/25686. (b) Tarby, C. M.; Endo, N.;
Moree, W.; Kataioka, K.; Ramirez-Weinhouse, M. M.; Imai, M.;
Bradley, E.; Saunders, J .; Kato, Y.; Meyers, P. In 218th National
ACS Meeting in New Orleans, 1999, Poster Abstract No. 82.
(14) Shigenaga, S.; Manabe, T.; Matsuda, H.; Fujii, T.; Hiroi, J .
Synthesis and Pharmacological Properties of N-[4-[4-(1H-indol-
3-yl)piperidinoalkyl]-2-thiazolyl]alkanesulfonamides as Novel
Antiallergic Agents. Chem. Pharm. Bull. 1993, 41, 1589-1595.
(15) Studies on 1 (unpublished), 2 (Lapierre, J . M. 26th National
Medicinal Chemistry Symposium, 14-16 J une, 1998, Richmond,
VA), and 4 (ref 12c) have indicated that these small molecule
antagonists are competitive inhibitors of 125I-MCP-1 binding, so
the expression of results as pKi’s and pKb’s is valid
(16) (a) Unger, V. M.; Hargrave, P. A.; Baldwin, J . M.; Schertler, G.
F. X. Arrangement of rhodopsin transmembrane R-helixes.
Nature (London) 1997, 389, 203-206. (b) Baldwin, J . M.;
Schertler, G. F. X.; Unger, V. M. An alpha-carbon template for
the transmembrane helixes in the rhodopsin family of G-protein-
coupled receptors. J . Mol. Biol. 1997, 272, 144-164.
(17) Blaney, F. E.; Raveglia, Luca, F.; Artico, M.; Cavagnera, S.;
Dartois, C.; Farina, C.; Grugni, M.; Gagliardi, S.; Luttmann, M.
A.; Martinelli, M.; Nadler, G. M. M. G.; Parini, C.; Petrillo, P.;
Sarau, H. M.; Scheideler, M. A.; Hay, D. W. P.; Giardina, G. A.
M. Stepwise Modulation of Neurokinin-3 and Neurokinin-2
Receptor Affinity and Selectivity in Quinoline Tachykinin
Receptor Antagonists. J . Med. Chem. 2001, 44, 1675-1689.
(18) Blaney, F. E.; Tennant, M. Computational tools and results in
(27) (a) Bromidge, S. M.; Dabbs, S.; Davies, D. T.; Davies, S.;
Duckworth, D. M.; Forbes, I. T.; Gaster, L. M.; Ham, P.; J ones,
G. E.; King, F. D.; Mulholland, K. R.; Saunders, D. V.; Wyman,
P. A.; Blaney, F. E.; Clarke, S. E.; Blackburn, T. P.; Holland, V.;
Kennett, G. A.; Lightowler, S.; Middlemiss, D. N.; Trail, B.; Riley,
G. J .; Wood, M. D. Biarylcarbamoylindolines Are Novel and
Selective 5-HT2C Receptor Inverse Agonists: Identification of
5-Methyl-1-[[2-[(2-methyl-3-pyridyl)oxy]-5-pyridyl]carbamoyl]-6-
trifluoromethylindoline (SB-243213) as a Potential Antidepres-
sant/Anxiolytic Agent. J . Med. Chem. 2000, 43, 1123-1134. (b)
Gaster, L. M.; Blaney, F. E.; Duckworth, D. M.; Ham, P.;
J enkins, S.; J ennings, A. J .; J oiner, G. F.; King, F. D.; Mulhol-
land, K. R.; Starr, S.; Wyman, P. A.; Hagan, J . J .; Hatcher, J .;
J ones, B. J .; Middlemiss, D. N.; Price, G. W.; Riley, G.; Roberts,
C.; Routledge, C.; Selkirk, J .; Slade, P. D. The Selective 5-HT1B
Receptor Inverse Agonist 1′-Methyl-5-[[2′-methyl-4′-(5-methyl-
1,2,4-oxadiazol-3-yl)biphenyl-4-yl]carbonyl]-2,3,6,7-tetrahydrospiro-
[furo[2,3-f]indole-3,4′-piperidine] (SB-224289) Potently Blocks
Terminal 5-HT Autoreceptor Function Both in Vitro and in Vivo.
J . Med. Chem. 1998, 41, 1218-1235.
(28) Quanta98, version 98.1111; Accelrys Inc.: San Diego, CA, 1999.
(29) Ballesteros, J . A.; Weinstein, H. Integrated methods for the
construction of three-dimensional models and computational
probing of structure-function relations in G-protein coupled
receptors. Methods Neurosci. 1994, 25, 366-428.
(30) FlexX; Tripos, Inc., 1699 South Hanley Rd., St. Louis, MO.
(31) Ewing, T. J . A.; Kuntz, I. D.. Critical evaluation of search
algorithms for automated molecular docking and database
screening. J . Comput. Chem. 1997, 18, 1175-1189.
(32) GOLD; Cambridge Crystallographic Data Centre, 12 Union Rd.,
Cambridge CB2 1EZ, UK.
the construction of
G protein-coupled receptor models. In
Membrane Protein Models; Findlay, J . B. C., Ed.; Oxford, Bios
Scientific Publishers Ltd., 1996; pp 161-176.
(19) Blaney, J . M.; Crippen, G. M.; Dearing, A.; Dixon, J . S.;
Spellmeyer, D. C. DGEOM95. Available from the Quantum
Chemistry Program Exchange, Bloomington, IN, 1995.
(33) DockIt; Metaphorics, LLC., 27401 Los Altos #360, Mission Viejo,
CA, 92691.
(20) Brooks, B. R.; Bruccoleri, R. E.; Olafson, B. D.; States, D. J .;
Swaminathan, S.; Karplus, M. CHARMM: A program for
macromolecular energy, minimization, and dynamics calcula-
tions. J . Comput. Chem. 1983, 4, 187-217. CHARMm, Version
25.2, Revision 98.0731; Accelrys Inc.: San Diego, CA, 1999.
(21) (a) Elber, R.; Karplus, M. Enhanced sampling in molecular
dynamics: Use of the time-dependent Hartree approximation
for a simulation of carbon monoxide diffusion through myoglobin.
(34) ICM, v2.7; MolSoft, San Diego, CA, 1998.
(35) FLO; Thistlesoft Inc., Morris Township, NJ .
(36) Bikker, J . A.; Trumpp-Kallmeyer, S.; Humblet, C. G-Protein
Coupled Receptors: Models, Mutagenesis, and Drug Design. J .
Med. Chem. 1998, 41, 2911-2927.
(37) Blaney, F. E.; et al. Manuscript in preparation.
J M030862L