Scaffold hopping with molecular field points: Identification of a cholecystokinin-2 (CCK2) receptor pharmacophore and its use in the design of a prototypical series of pyrrole- and imidazole-based CCK2 antagonists

Article abstract of DOI:10.1021/jm049069y

A new molecular modeling approach has been used to derive a pharmacophore of the potent and selective cholecystokinin-2 (CCK₂) receptor antagonist 5 (JB93182), based on features shared with two related series. The technique uses "field points"

Full text of DOI:10.1021/jm049069y

6790
J. Med. Chem. 2005, 48, 6790-6802
Articles
Scaffold Hopping with Molecular Field Points: Identification of a
Cholecystokinin-2 (CCK2) Receptor Pharmacophore and Its Use in the Design of
a Prototypical Series of Pyrrole- and Imidazole-Based CCK2 Antagonists
Caroline M. R. Low,*,† Ildiko M. Buck,† Tracey Cooke,† Julia R. Cushnir,† S. Barret Kalindjian,† Atul Kotecha,†
Michael J. Pether,† Nigel P. Shankley,†,‡ J. G. Vinter,§ and Laurence Wright†
James Black Foundation, 68 Half Moon Lane, London, SE24 9JE, U.K., and Cresset BioMolecular Discovery Ltd.,
Spirella Building, Bridge Road, Letchworth, Herts, SG6 4ET, U.K.
Received November 18, 2004
A new molecular modeling approach has been used to derive a pharmacophore of the potent
and selective cholecystokinin-2 (CCK2) receptor antagonist 5 (JB93182), based on features
shared with two related series. The technique uses “field points” as simple and effective
descriptions of the electrostatic and van der Waals maxima and minima surrounding a molecule
equipped with XED (extended electron distribution) charges. Problems associated with the high
levels of biliary elimination of 5 in vivo required us to design a compound with significantly
lower molecular weight without sacrificing its nanomolar levels of in vitro activity. Two new
series of compounds were designed to mimic the arrangement of field points present in the
pharmacophore rather than its structural elements. In a formal sense, two of the three amides
in 5 were replaced with either a simple pyrrole or imidazole, while some features thought to
be essential for the high levels of in vitro activity of the parent compounds were retained and
others deleted. These compounds maintained activity and selectivity for this receptor over CCK1.
In addition, the reduction in molecular weight coupled with lower polarities greatly reduced
levels of biliary elimination associated with 5. This makes them good lead compounds for
development of drug candidates whose structures are not obviously related to those of the
parents and represents the first example of scaffold hopping using molecular field points.
Introduction
hybridization studies.15 Both human CCK1 and CCK2
receptors have been cloned and shown to belong to the
family of G-protein-coupled receptors.15-18 The problems
associated with designing drug candidates for the major
class of G-protein-coupled receptors (GPCRs) that are
activated by peptides are well described. In such cases
it is imperative that compounds should not be peptidic
because of the recognized problems of potential immu-
nogenicity, low metabolic stability, and bioavailability
associated with this class of molecule. However, the
search for suitable small-molecule candidates is also
hampered by the lack of detailed knowledge about the
structure of the receptors themselves. Hence, we are
limited to indirect methods of determining the require-
ments of high-affinity ligands for our target receptor
protein, in this case the CCK2 GPCR.
Conventional molecular modeling techniques place a
high emphasis on using structural comparisons as their
basis for explaining patterns of biological activity. These
methods have had some success in rationalizing the
behavior of closely related compounds but place signifi-
cant constraints on the development of distinct new
molecular entities. For example, the hormone CCK is a
large peptide and compounds described as competitive
CCK2 antagonists represent a diverse group of small-
molecule non-peptides ranging from benzodiazepines,
such as L-365,26019 and YF476,20 to peptoids, such as
We have reported several novel classes of cholecys-
tokinin-2 (CCK2) receptor antagonists that all show good
selectivity for this receptor over the closely related CCK1
receptor.1 These have included compounds based on
substituted dibenzobicyclo[2.2.2]octanes (BCO),2 two
related series based on bicyclic aromatic scaffolds,3,4 and
a series of novel benzo[h][1,4]diazonines.5 The two
subclasses of cholecystokinin receptors mediate a wide
range of peripheral and central biological processes, and
identification of selective ligands for either subtype has
attracted a large amount of interest.6-10 CCK1 receptors
occur centrally in the nucleus tractus solitarius11 but
are mainly located in the pancreas,12 gall bladder,13 and
colon. CCK2 receptors are the predominant subtype in
the brain and are widely distributed throughout the
cortex.14 They are also located on the ECL cell of the
stomach where they are involved in the regulation of
gastric acid secretion. There is strong evidence to
suggest that the central and peripheral populations of
receptors are homogeneous, on the basis of their selec-
tivity for a range of ligands and evidence from molecular
* To whom correspondence should be addressed. Phone: +44 20
7737 8282. Fax: +44 20 7274 9687. E-mail: caroline.low@kcl.ac.uk.
† James Black Foundation.
‡ Current Address: Johnson & Johnson Pharmaceutical Research
& Development, LLC, 3210 Merryfield Row, San Diego, CA 9212.
§ Cresset BioMolecular Discovery Ltd.
10.1021/jm049069y CCC: $30.25 © 2005 American Chemical Society
Published on Web 10/11/2005

CCK2 Receptor Pharmacophore
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 22 6791
PD134308,21 to our own bicylic aromatics, typified by
5.4 All of these compounds are selectively recognized by
the CCK2 receptor, and if we assume, in the absence of
any experimental evidence to the contrary, that they
bind to a similar site on the receptor, then we should
be able to identify common features relating one to
another. Analyzing their structures alone does not
provide us with a unifying rationale that would allow
us to move directly from one series to another. However,
the receptor binds all of these structures with nano-
molar affinity, suggesting that it uses additional fea-
tures to recognize and interact with ligands. This
observation is not confined to CCK2 receptor ligands and
we are not alone in believing that the processes of
molecular recognition and binding are governed by
electrostatic and steric features that are not explicitly
represented in the stick representations of molecular
models. It therefore follows that if two molecules with
diverse structures can generate similar electronic prop-
erties in a particular conformation, they will interact
with an enzyme or receptor in a similar way. This point
is recognized in other modeling techniques to some
extent, e.g., GRID analysis22 and other methods of
pharmacophore modeling.23,24 However, it is difficult to
apply these to automated analysis of diverse structures
because many require significant operator input to
generate alignments.
Chart 1. CCK2 Antagonist Based on the
Dibenzobicyclo[2.2.2]octane (BCO) Core
derived for sets of low-energy conformations of the
compound of interest. Sets of field points can be easily
compared with one another and the resulting overlay
displayed together with its structural origin, facilitating
interpretation of the results in chemical terms. This
approach avoids the problems of choosing structural
alignments that limit conventional QSAR methods, such
as CoMFA.24,31 In addition, the comparison of individual
sets of field points is a fast process, making it feasible
to compare many conformations of the same molecule
against those of another. Conversely, methods for
comparing molecular similarity using MEPs are gener-
ally more involved and many require significant manual
manipulation in generating the final overlay.32 The first
examples using this field point approach were reported
some time ago,25 and this paper will describe its ap-
plication to the design of a new series of CCK2 receptor
antagonists.
Our first examples of CCK2 antagonists, typified by
structure 1 (Chart 1), all contained a BCO core and had
micromolar affinity for this receptor.2 They were the
products of functionalizing the two bridgehead positions
with an alicyclic amide and a carboxylate, respectively.
A series of structural modifications rapidly showed that
both the activity and selectivity of these compounds
could be improved by extending the carboxylate side-
arm. This gave compounds, such as 2, with the same
levels of CCK1 affinity but whose affinity for CCK2
receptors had increased by a factor of 100 (Table 1). The
optimal substituents were found to be 1-adamantyl-
methyl on one arm, with the 3,5-dicarboxyanilide de-
rivative of L-Phe on the second.3 The bulky BCO was
not a unique requirement and could be replaced by
simple aromatic groups to give compounds with 10- to
30-fold lower activity (3 and 4). However, field point
analysis of the original BCO series had suggested that
a key feature of these compounds was not present in
the naphthalene analogue 3. Both molecules had a
series of electropositive field points of varying magni-
tude around the plane of the aromatic groups. In
addition, the BCO 2 also had a sizable negative field
point arising from the focus of the π-electrons of the two
isolated aromatic rings. This led us to devise the indole
derivative 5, which we compared to BCO 2 using our
field point overlay methodology, as previously de-
scribed.4 The desired relationship was maintained in
most of the pairwise comparisons generated. Compound
5 was synthesized and our expectations were confirmed
when it was shown to possess a profile of activity that
was indistinguishable from that of the BCO 2 in a
number of in vitro bioassays. This was clearly a signifi-
cant finding and demonstrated that this methodology
could be used in a meaningful fashion.
This conundrum has led us to develop a new molec-
ular modeling approach that uses “field points” as
simple and effective descriptions of the electrostatic and
van der Waals maxima and minima surrounding a
molecule.25 Sets of field points are calculated for indi-
vidual conformations equipped with XED (extended
electron distribution) charges26 rather than the atom-
centered charge descriptions used in conventional mo-
lecular mechanics packages. The XED concept was
originally based on results obtained by Hunter and
Sanders who showed that a three-center charge model
could be used to reproduce the stacking geometry of
porphyrins better than an equivalent atom-centered
charge model.27 We have now extended this approach
and derived a new molecular mechanics force field in
which appropriately parametrized XED charges have
been added to the Coulombic term. The package has
been evaluated against experimental data compiled by
Guntertofte et al.28,29 and found to be superior to most
of the commercially available packages.30 This was
particularly true in the case of aromatic-aromatic
π-stacking interactions and examples where the ano-
meric effect controls the choice of conformational space.
The field points that we have developed are based on
the widely used molecular electrostatic potential (MEP)
descriptors but resemble the results of distributed
multipole analysis when derived using XED charges.
However, accurate calculation of MEPs requires the use
of quantum mechanical methods, making their deriva-
tion time-consuming. We have found that the quality
of molecular electrostatic maps obtained using simpler
molecular mechanics methods can be improved using
XED charge descriptions. Further simplification has
been achieved by distilling these maps down to their
energy extrema, which we have termed field points. A
new method of defining surface interaction has been
added, and the resulting composite maps can be rapidly
At this point we had access to a family of three
compounds (2, 3, and 5) that differed in the nature of
the group to which the two amide sidearms were

6792 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 22
Low et al.
Table 1. Biological Data for Compounds Incorporated in the
CCK2 Receptor Pharmacophore
receptor ligands, in the absence of information about
the receptor, has been described;35 however, no biologi-
cal data were presented for any of the structures
proposed. An alternative approach used “fuzzy” molec-
ular representations called FEPOPS (3D feature point
pharmacophores) to suggest new leads based on com-
pounds already present in an in-house high-throughput-
screening database.36 All of these approaches require
the definition of a pharmacophore, and hence, our first
step was to define such a model for compounds 2, 3, and
5, based on our existing knowledge, which might be used
to indicate ways in which the rapid billiary excretion
observed in vivo could be reduced without compromising
the high levels of in vitro activity that had already been
achieved.
Chemistry
The pyrrole derivatives in Table 3, 10a-f, were
prepared according to the general route in Scheme 1.
Pyrroles 7a-d were prepared from the diketoesters (6)
following literature procedures.37,38 Hydrolysis of the
ethyl ester of intermediates 7a-d afforded the acids
8a-d, which were coupled with aminoisophthalic acid
dibenzyl ester (9a) or 3-aminobenzoic acid benzyl ester
(9b) via their acid chlorides to afford 10a-e, after
hydrogenolysis of the benzyl ester protecting group.
Reaction of the acid chloride formed from 8a with
â-alanine benzyl ester afforded 10f after hydrogenolysis.
The imidazole derivatives 17a-g were prepared by
the general route outlined in Scheme 2. Tricarbonyls
13 were prepared by oxidation39 of ylides 1240 and
converted to the trisubstituted imidazole carboxylic
acids 15a-e by reaction with benzaldehyde or 2-naph-
thaldehyde in the presence of acetic acid/ammonium
acetate,41 followed by hydrolysis of the ethyl ester.
Coupling of the aniline derivatives 9a-c to the acids
15a-e under standard active ester coupling conditions,
followed by the appropriate deprotection, afforded 17a-
g. All the compounds were tested as N-methyl-D-
glucamine salts. The salts were prepared by stirring a
mixture of the compound with 1 or 2 equiv of N-methyl-
D-glucamine, as appropriate, in dioxan-water (1:1) until
a solution was obtained. The solution was then freeze-
dried, and all compounds were obtained as fluffy white
solids.
a Compounds were tested as the bis(N-methyl-D-glucamine)
salts. b Reference 3. c Reference 4.
anchored. However, we were also aware that represen-
tative examples of these compounds were rapidly ex-
creted, unchanged, in the bile of both rats and dogs. This
presented us with the challenge of designing new
molecules with improved bioavailability while retaining
the high levels of potency and selectivity for CCK2
receptors that had been demonstrated in vitro. Biliary
excretion is a complex process whose mechanism is still
not fully understood. However, a possible cause for this
rapid clearance was the high molecular weights of these
compounds that, at around 700, were significantly above
the qualitative threshold of 325 that is commonly quoted
for biliary excretion in these two species. This threshold
is known to be species-dependent and rises to a value
of between 500 and 600 in man. However, it is well
recognized that removing molecular weight from ligands
without concomitant loss of activity is not a simple
task.33
A number of “scaffold-hopping” approaches have been
described in the literature. This recent concept involves
the computational comparison of chemical scaffolds to
generate new chemical entities with different molecular
frameworks. The overall aim is to change the physical
characteristics of a drug candidate to avoid adverse
properties or create compounds whose structural rela-
tionship to the starting point is not obvious. A number
of examples of cannabinoid receptor antagonists, whose
structures cannot be categorized as bioisosteres or
conformationally constrained analogues of rimonabant,
have been recently reviewed, although the processes
underlying their discovery are not discussed in any
detail.34 Software for the de novo design of new estrogen
Molecular Modeling
Molecular modeling analyses were carried out using
structures equipped with XED charges rather than
conventional atom-centered charges. We have found
that this modification allows us to reproduce experi-
mental thermodynamic data with good accuracy30 and
have incorporated these charge descriptions into our
standard modeling package, XED.42
In this study, structures were selected on the basis
of their biological activity with the emphasis on antago-
nists with nanomolar affinity and good selectivity for
CCK2 over the closely related CCK1 receptor (Table 1).
Each compound was submitted to a conformational
hunting routine in order to generate a set of conforma-
tions equipped with XED charges and field points whose
energies lay within 3 kcal/mol of the calculated global
minimum. This limit was chosen, in line with accepted
practice,43 because the Boltzmann equation predicts that

CCK2 Receptor Pharmacophore
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 22 6793
Scheme 1. Synthesis of Trisubstituted Pyrrolesa
a (a) K2CO3, NaI, Me2CO; (b) NH4OAc, AcOH; (c) NaOH, EtOH-H2O; (d) SOCl2, pyr; (e) Pd-C, H2, THF-MeOH.
Scheme 2. Synthesis of Trisubstituted Imidazolesa
a (a) Ph3PCHdCO2Et, EDC, DCM; (b) OXONE, THF-H2O; (c) R1CHO, NH4OAc, AcOH; (d) NaOH, EtOH-H2O; (e) EDC, HOBt, DMF;
(f) Pd-C, H2, THF-MeOH when X′ ) Y′ ) CO2Bn or X′ ) H and Y′ ) CO2Bn; (g) 1% K2CO3, THF-MeOH when X′ ) Y′ ) CH2OCO2Me.
<1% of available conformations will have energies
greater than 3 kcal/mol above the lowest energy con-
formation found at body temperature, i.e., 37 °C (310
K). This process typically gave a set of 10-40 conforma-
tions for each molecule.
tions within 15° on all torsions of any other were
removed. The energies of all remaining structures were
then minimized using a combined parabolic/Fletcher-
Reeves conjugate gradient technique, whose accuracy
was set at <0.01 cal/mol. XED charges were then added
to each conformation, and the energies of each structure
were minimized once again using the same procedure.
Conformations were collected over a 15.0 kcal/mol
energy range to a maximum of 1000 structures.
Conformational Hunting
The protocol used was as follows. Calculations were
carried out at dielectric constant of 2.0, allowing all
rotatable bonds to move. An initial set of conformations
was generated using a hard spin torsional minimizer.
Each input conformation was randomized 250 times
(Monte Carlo), twisting each bond by 10° torsional
increments, iterating through all bonds to 0.001 cal/mol
and storing all conformations within a 15.0 kcal/mol
range of the lowest energy structure found. Conforma-
Calculation of Field Points
Three classes of field point were generated for each
conformer: positive and negative electrostatic extrema
and a third class of “sticky” points defining the van der
Waals surface around the molecule. The electrostatic
field points are calculated using a constant point unit

6794 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 22
Low et al.
Scheme 3. Molecular Field Points Calculated for
points on one molecule and driving them over those of
the second under a Coulombic potential. The process
was then reversed such that A was compared with B,
and B with A. The interaction between the two mol-
ecules can then be quantified in terms of a scoring
function, which we have termed the “overlay energy”,
that is calculated as q1q2/r2 (where r is the distance
between two charges q1 and q2). This number reflects
the degree of similarity between the two sets of field
points. Thus, comparisons between pairs of molecules
are made entirely on the basis of the field points,
avoiding the problems of operator bias in choosing
structural alignments, a recognized limit of conventional
QSAR methods such as CoMFA.24,31 Goodness-of-fit was
ranked on the basis of the “overlay energy” parameter,
and the 50 best overlays, i.e., those with the lowest
overlay energies, were stored. A complete comparison
of 20 conformations of 2 with six conformations of 5 took
1 h to reach completion on a Silicon Graphics Octane
with dual 270 MHz IP30 processors.
Benzenea
a The aromatic carbon atoms are equipped with XED charges
(purple). Positive and negative electrostatic points are shown in
red and green, respectively, and van der Waals points are shown
in yellow. The size of the spheres is directly related the magnitude
of the field point.
charge, the size of an oxygen atom, of +1 for a positive
probe and -1 as a negative probe. The probes are placed
on a grid of points 3 Å above the molecular van der
Waals surface at a dielectric constant of 4.0. Full details
of the methodology used have been published else-
where.25 The interval between grid points is adjusted
so that the whole molecular surface is covered using a
maximum of 250 points. This number varies according
to the compromise between the interval and the area
of the molecule.
After allocation of an interaction energy to every grid
point, each point is allowed to find its appropriate
maximum or minimum, using a 3D simplex algorithm
that is now no longer confined to the grid itself. The
result is that many minimized positions eventually
coincide. After coincident extrema are filtered out, the
final positions of the field points are stored, together
with their type (positive or negative) and energy (well-
depth) information. These are plotted around the mol-
ecule as red (positive) and green (negative) spheres
whose radii reflect the depth of the energy well. The
final set of unique extrema is stored with Cartesian
coordinates, the field point class, and the interaction
well depth in energy units.
The results can be readily understood by considering
the pattern of field points generated for a molecule, such
as benzene (Scheme 3). In this case negative points are
found in the expected location of the π-electrons with a
ring of smaller positive points in the plane of the ring.
A third class of “sticky” points, shown in yellow, is
calculated using a neutral probe and a Morse potential
to represent the van der Waals properties of the
molecule. These broadly map the surface of the molecule
and suggest where the “stickiest” points occur.
An earlier description of this procedure has been
published25 and differs only in that all probe classes use
a Morse equation for the van der Waals interactions. A
simple Coulombic term is used for the electrostatic
probes.
Biology
CCK2 receptor activity was assessed by testing all
compounds in a radioligand binding assay in mouse
cortical membrane homogenates, in competition with 20
pM [125I]BH-CCK-8S, as previously described.44 The pKi
( SEM values were determined using results taken
from at least three separate experiments. Activity at
CCK1 receptors was established using an equivalent
assay in guinea pig pancreas cell homgenates.45 In the
same way pKi ( SEM values were determined in
competition with 20 pM [125I]BH-CCK-8S, using results
taken from at least three separate experiments.
The agonist/antagonist properties of two compounds,
10a and 17c, were examined in the isolated, lumen-
perfused immature rat stomach.46 pKB ( SEM values
were estimated from single shifts of pentagastrin tBoc-
CCK(29-33) concentration-effect curves, calculated
assuming an underlying Schild slope of unity and fitted
using the Gaddum-Schild equation.
Biliary elimination was measured in anesthetized rats
as the percentage of unchanged compound detected in
the bile following iv bolus administration of the com-
pound at a dose of either 1 or 10 µmol/kg. The bile duct
was cannulated and bile collected over 15 min intervals
for a total 150 min in each case. The unchanged
compound was detected by reverse-phase HPLC (Waters
C8 column), eluting with H2O/CH3CN (30:70) + 0.01%
CH3COOH (final pH 3.74) and monitoring at 286 nm.
Results and Discussion
We have already demonstrated the use of our field
point methodology in moving from the original BCO
series to the heterocyclic derivatives typified by indole
5.4 However, there is a clear structural relationship
between these compounds, and significant synthetic
effort had already shown that certain groups could not
be removed from these optimized structures without
concomitant loss of biological activity. This made it
difficult to envisage ways in which we could include
those features that we knew to be essential for the high
levels of in vitro activity in a compound with greatly
reduced molecular weight.
Comparison of Ligands Using Field Points
The set of field points generated for a particular
conformation of a ligand (A) was compared to all the
conformers of another molecule (B) on a pairwise basis.
This process involved inverting the charges of the field
One of the advantages of using field points to sum-
marize molecular properties is that it allows us to

CCK2 Receptor Pharmacophore
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 22 6795
Scheme 4. Proposed Bioactive Conformation of Indole
separate elements responsible for recognition and recep-
tor binding from their structural origins. This led us to
hope that we could introduce much greater levels of
diversity into our design process once we removed the
need to reproduce structural features directly. The first
step in this process was to define a field point pharma-
cophore that encompassed the properties of the current
series. We chose to construct this from compounds 2, 3,
and 5, each equipped with the optimal choice of side-
arms but with different nuclei. The affinities of these
compounds are not identical; however, they all behave
as competitive antagonists in the presence of penta-
gastrin,2-4 and so we assume, since we have no evidence
to the contrary, that they bind to the same site on the
CCK2 receptor. Our aim was to use molecular field point
analysis to identify conformations of 2, 3, and 5 with
the most similar electronic properties as a way of
identifying a ligand-based pharmacophore.
5a
To this end, 22 conformations of the naphthalene 3
were compared with 6 generated for the indole 5. The
results showed that the arrangement with the lowest
“overlay energy” was an essentially unique conformation
of the amidic side chains in which the adamantyl and
3,5-dicarboxyanilide groups were held in proximity by
the presence of a bifurcated H-bond between the car-
bonyl oxygens of the geminal amides and the anilide
NH. This provided the first clue to the bioactive
conformation of 3 and 5. Furthermore, we could test this
proposal by comparing indole 5 to BCO 2 on the same
basis. In this case the best field point overlay occurred
with the same conformation of indole 5 and a conforma-
tion of BCO 2 that also fitted the previous observations.
At this point we should stress that the comparison was
not biased by seeding the results with the common
structure identified in the first experiment but resulted
from the exhaustive comparison of the complete set of
conformations generated for a 3 kcal/mol range in each
case.
Because this study was one of the first applications
of the field point methodology, we restricted our original
choice of compounds to those that were generated in
house and fully characterized as competitive antago-
nists. The levels of structural diversity within this set
are not high but are analogous to those of most data
sets used in CoMFA analyses, where a significant
portion of the structure must be common to allow the
operator to manually superpose the molecules in the
training set. We have subsequently extended this ap-
proach to include other examples of the diverse set of
known CCK2 receptor ligands, such as the peptoid
PD13430847 and the benzodiazepine YF476.20 The con-
formation of indole 5 identified in these later studies
does not differ significantly from that described here
(unpublished results).
a Identified by field point comparison of multiple conformations
of BCO 2, naphthalene 3, and indole 5. Electrostatic field points
are shown in red (positive charge) and green (negative charge),
but van der Waals points have been omitted for clarity. The field
point pattern due to the Phe residue and the amidic carbonyl of
the adjacent adamantylmethyl sidearm are highlighted.
broadly dipolar, in common with our model of the
tetrapeptide fragment tBoc-CCK(30-33) on which the
design of the original series of BCOs was based.48 It is
not surprising that the groups that contribute most to
the population of negative field points are the carbonyls
of the amides and carboxylic acids. However, there is
also a significant density of positive field points gener-
ated by the edge of this electron-deficient indole system
and the NH adjacent to the adamantyl group.
The next step was to seek physical evidence for the
1
existence of this structure. A series of H NMR experi-
ments (in DMSO-d6) was carried out on indole 5 and
several closely related analogues. These showed a
number of consistent features that supported our pro-
posed structure. For example, both the nuclear Over-
hauser enhancement (NOE) difference and rotating-
frame Overhauser enhancement spectroscopy (ROESY)
experiments showed evidence of the proximity of the
ortho protons of the aniline dicarboxylate to the ada-
mantane methylenes. There were also a number of
ROEs between the phenyl protons of the Phe residue
and the indole group but none to the aniline, suggesting
that the Phe side chain bends back toward the hetero-
cycle. This is also supported by the existence of NOEs
between the â-methylene protons of the Phe and the
ortho protons of the aniline. In addition, there were
strong NOEs between the amide NH adjacent to the
adamantyl and the 7-CH of the indole, defining the
configuration of this sidearm. These findings are all in
line with our proposed structure and rule out several
other possible conformations. For example, the set of
conformations generated for indole 5 contains a number
of structures in which the 3,5-dicarboxyanilide is folded
back over the indole group. The existence of this
conformation would be reflected by the presence of
NOEs between the protons attached to the aniline and
The conformation of indole 5 identified by this process
shows a number of interesting features. For example,
the polar 3,5-dicarboxyanilide is located close to the
hydrophobic adamantyl, although we might have ex-
pected to find this group associated with one of the other
hydrophobic components of the molecule (Scheme 4).
This arrangement is maintained by hydrogen-bonding
between the amides of the two arms and is mirrored in
all the low-energy structures. The pattern of field points
around the molecule illustrates that the system is

6796 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 22
Low et al.
Table 2. Effect of N-Methylation of the Individual Amides on
the Activities of Indole 5 and Benzimidazole 18
Scheme 5. Electrostatic Field Point Patterns
Calculated for (a) Pyrrole and (b) Imidazole Equipped
with XED Chargesa
a Negative points are shown in green, and positive points are
shown in red.
loss of CCK2
activity
relative to
two H-bonding possibilities but supports our belief that
this group is essential to maintaining the overall
structure of the series. Smaller losses were observed on
methylation of the Phe NH in 19 and adamantyl amide
NHs in 21.
a
a
CCK2
parent
CCK1
compd
X
R1 R2 R3 pKi ( SEM (log units) pKi ( SEM
5
18
19
20
21
CH H
H
H
H
H
H
9.0 ( 0.2
8.3 ( 0.1
7.6 ( 0.1
6.3 ( 0.1
5.4 ( 0.2
5.8 ( 0.3
5.0 ( 0.2
5.4 ( 0.2
5.7 ( 0.2
N
N
N
H
Me H
0.7
2.0
1.2
Structure-activity studies had convinced us that the
most important features of indole 5 were the hydropho-
bic adamantyl group, the heteroaromatic group, and the
3,5-dicarboxyanilide.4 Furthermore, we believed that we
had identified two possible bioactive conformations of
this compound and had access to both structural and
field point representations of this pharmacophore. We
were now in a position to exploit this information in the
design of novel compounds. Our strategy was to tackle
the problem of biliary elimination by reducing the
molecular weight of the structure without sacrificing the
high levels of in vitro activity that had already been
attained.
Molecular mechanics calculations had repeatedly
shown that there was a consistent spatial relationship
between the three constituents of indole 5 to which we
alluded earlier. However, we had been unable to pin
down a role for the Phe side chain, which showed a high
degree of conformational mobility. In addition, structure-
activity studies had shown that the side chain of the
amino acid incorporated at this point could be altered
substantially without significant effect on the affinity
of the compounds.4 Hence, we decided to omit this group
from our initial survey of potential candidates in the
interests of reducing their overall molecular weight.
We also considered that it might be possible to reflect
the properties of the three amides responsible for
maintaining the proximity of the adamantyl and 3,5-
dicarboxyanilide with a small rigid heterocycle. Careful
analysis of the pattern of electrostatic field points
generated by the two amide groups attached to the
indole showed that there were three significant field
points, one positive and two negatives (Scheme 4), to
be considered. Good candidates for this purpose ap-
peared to be either pyrrole or imidazole because the
electrostatic field point patterns around each of these
is consistent with the pattern attributed to the amides
linking the Phe residue and its adjacent sidearm to the
indole group (Scheme 5). In addition, these five-
membered ring heterocycles allowed us to maintain the
spatial relationship between the adamantyl, indole, and
the 3,5-dicarboxyanilide moieties present in the phar-
macophore of compound 5.
H
Me H
H
CH H
Me 7.8 ( 0.1
a Compounds were tested as the bis(N-methyl-D-glucamine)
salts.
the indole group. None of the 1H NMR experiments
carried out to date have provided any evidence of the
existence of such a population in solution.
Hence, we have gathered some evidence that indole
5 can adopt the topology in solution that we propose
for our pharmacophore. Additional evidence of the
nature of the bioactive conformation was supplied by
the results of chemical modification of the indole 5 and
the closely related benzimidazole analogue 18. The
overall conformation of the proposed pharmacophore is
stabilized by hydrogen bonding between the carbonyl
oxygens and NHs of the various amides. In fact, two
conformations of similar energy have been identified
from our modeling experiments. One involves the in-
teraction of the Phe and anilide NHs with the carbonyl
group of the adamantyl amide, as shown in Scheme 4,
and the second relies on hydrogen bonding between the
oxygens of the two amide carbonyl groups attached to
the indole and the anilide NH. The energy difference
between these two alternatives is 1 kcal/mol, and so the
Boltzmann equation predicts that the ratio of the two
conformations will be 5:1 in favor of the first, lower
energy conformation at 37 °C. The role that the various
amides play in maintaining the overall structure was
further probed by methylation of the individual NHs
(19-21). Molecular modeling showed that methylation
of either the anilide or adamantylmethylamide NHs
reinforced the H-bond between the NH of the Phe and
the carbonyl oxygen of the adamantylmethylamide.
However, interpreting these results was not straight-
forward because methylation of either of the amide NHs
on the side chain attached to the 5-position of the indole
group changed the configuration of the relevant amide
bond from E to Z. In practice the compounds synthesized
showed a loss of CCK2 activity in all cases (Table 2). It
is also interesting to note that levels of CCK1 affinity
remained constant throughout this exercise. The most
pronounced effects were noted when the anilide NH was
methylated (20), resulting in a 100-fold loss of activity.
This result does not allow us to distinguish between the
A single amide was included so that the negative field
point generated by its carbonyl oxygen could be used to

CCK2 Receptor Pharmacophore
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 22 6797
Scheme 6. Design of Prototype Pyrrole 22 Based on the
Indole Pharmacophore 5a
Scheme 7. Overlaying the Field Point Pattern of the
CCK2 Receptor Pharmacophorea
a Based on indole 5 (shown in yellow), with a 3 kcal/mol energy
range of conformations calculated for the prototype pyrrole 22
(shown in white). The indole, 3,5-dicarboxyanilide, and adamantyl
groups are mapped directly over one another in 10% of the overlays
stored.
of indole 5 (pKi ) 6.0 ( 0.1), and the reasons for this
are not clear. Comparing the two molecules (17g and
5) on the basis of their field point patterns gave the
same results as those noted above. Hence, the paucity
of “good” overlays, analogous to those in Scheme 7, must
suggest that the alternative orientations identified are
not tolerated by the receptor binding site. In fact,
replacing the pendent indole of 22 with a phenyl group
(10a) gave a higher level of structural congruence with
the desired elements of the pharmacophore (30% of
overlays), and these were the first examples to be
synthesized. Therefore, this initial example 10a (pKi )
6.9 ( 0.1, Table 3) and the close relatives (10b,c) are
analogues of the phthalate 4 (pKi ) 7.5 ( 0.3, Table 1)
rather than indole 5. These compounds show similar
affinities for the CCK2 receptor, and so the difference
is not as great as might appear at first sight. The
molecular weights of these compounds were now close
to the threshold value of 500 for biliary elimination in
man. We were pleased to see that the levels of CCK1
activity were also broadly similar to those of phthalate
4 and hence that the selectivity for CCK2 over CCK1
receptors was maintained. This demonstrated that the
field point methodology could be used to design drug
candidates with reasonable accuracy, and synthesis of
further examples of both pyrrole and imidazole proto-
types was begun.
A number of compounds were made in parallel in
order to assess differences between the pyrrole (10a-
e) and imidazole (17a-e) series. Modifying the sizes of
the 2-aryl and 5-cycloalkylmethyl groups did not in-
crease levels of biological activity, which were consis-
tently robust in both receptor bioassays (Table 3). A
single example of a pyrrole with an aliphatic rather than
aromatic acid at the 4-position (10f) showed a 25-fold
decrease in CCK2 activity. We would attribute this to
its failure to maintain the relationship between the acid
and adamantyl groups observed in the indole pharma-
cophore due to the increased flexibility of the linking
group between the acid and amide. Most noticeably, the
pKi results obtained for any particular pair of pyrroles
and imidazoles were not significantly different in most
cases. The largest difference was observed for 10c and
a The design process involved removal of the benzyl side chain
of the Phe residue in indole 5 (gray) and replacement of the two
amide groups highlighted (black box) with a pyrrole to give
compound 22. The indole, adamantane, and 3,5-dicarboxyanilide
groups were all thought to be essential for activity and were
retained in the initial structure.
reinforce that from one of the acid carbonyls. This
recreates the pattern of negative points observed around
the 3,5-dicarboxyanilide of the pharmacophore, which
we considered to be essential for the high levels of in
vitro activity because we had already shown that
removal of one of the carboxylic acids in the indole series
had led to a reduction in CCK2 activity.4 In field point
terms, removing either of the acids changed the pattern
significantly because of the propensity for the amide and
carboxylic acid substituents to sit anti to each other,
presumably as a result of dipole-dipole repulsion
between the two groups.
A model of a prototypical pyrrole (22) (Scheme 6) was
constructed and subjected to a full conformational hunt.
All 42 conformations whose energies lay within 3 kcal/
mol of the calculated global minimum were then
equipped with field points and compared to those of the
pharmacophore shown in Scheme 4. The best 50 over-
lays were then assessed on the basis of both their
electrostatic and steric elements. These could be clas-
sified into three broad groups. The results of this
experiment showed that there were a small number of
overlays (10%) in which the common adamantyl, indole,
and 3,5-dicarboxyanilide groups of the two molecules
were structurally superimposed (Scheme 7). In fact, the
3,5-dicarboxyanilides were superimposed in all overlays.
However, either the indole or adamantane groups of
pyrrole 22 were located over the phenylalanine residue
of indole 5 in the remainder (data not shown). Never-
theless, the desired superposition (Scheme 7) had been
achieved in an albeit small proportion of the overlays
examined. Unfortunately, synthesis of the 5-indolylpyr-
role 22 first proposed was not possible using the general
route described, although the cycloheptyl analogue of
the equivalent imidazole (17g) was accessible. The
affinity of this compound was greatly reduced from that

6798 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 22
Low et al.
Table 3. Comparison of Parallel Series of Pyrrole and Imidazole CCK2 Receptor Antagonists
a
a
CCK2
CCK1
compd
Z
Ar
R1
1-Ad
R2
pKi ( SEM
pKi ( SEM
10a
10b
10c
10d
10e
10f
CH
CH
CH
CH
CH
CH
N
C6H5
3,5-(CO2H)2C6H3
3,5-(CO2H)2C6H3
3,5-(CO2H)2C6H3
3,5-(CO2H)2C6H3
3-(CO2H)C6H4
6.9 ( 0.1
6.6 ( 0.1
6.7 ( 0.1
6.7 ( 0.1
5.9 ( 0.1
5.5 ( 0.1
6.8 ( 0.1
6.0 ( 0.1
6.2 ( 0.1
6.6 ( 0.1
6.1 ( 0.1
5.4 ( 0.1
6.0 ( 0.1
5.1 ( 0.1
5.1 ( 0.1
5.3 ( 0.1
5.6 ( 0.1
5.4 ( 0.1
5.0 ( 0.1
5.5 ( 0.1
5.1 ( 0.1
5.2 ( 0.1
5.5 ( 0.1
5.6 ( 0.1
5.5 ( 0.1
5.3 ( 0.1
C6H5
c-C7H13
1-AdCH2
c-C7H13
c-C7H13
1-Ad
C6H5
2-C10H7
2-C10H7
C6H5
(CH2)2CO2H
17a
17b
17c
17d
17e
17f
C6H5
1-Ad
3,5-(CO2H)2C6H3
3,5-(CO2H)2C6H3
3,5-(CO2H)2C6H3
3,5-(CO2H)2C6H3
3-(CO2H)C6H4
N
C6H5
c-C7H13
1-AdCH2
c-C7H13
c-C7H13
c-C7H13
c-C7H13
N
C6H5
N
2-C10H7
2-C10H7
2-C10H7
5-indolyl
N
N
3,5-(CH2OH)2C6H3
3,5-(CO2H)2C6H3
17g
N
a Compounds were tested as the bis(N-methyl-D-glucamine) salts.
Table 4. Effect of Modifying the Aniline Substituent of Benzimidazole Bis-amide CCK2 Antagonists on Biliary Elimination
a
a
CCK2
CCK1
% material
compd
X
Y
pKi ( SEM
pKi ( SEM
MW
log P b
excreted unchangedc
18
23
24
CO2H
CO2H
H
CH2OH
8.3 ( 0.1
6.4 ( 0.1
6.2 ( 0.1
5.8 ( 0.3
5.3 ( 0.1
5.4 ( 0.1
664
620
638
-2.3
1.1
3.7
45
39
22
CO2H
CH2OH
a Compounds were tested as the bis(N-methyl-D-glucamine) salts. b log P measured in octanol/buffer (pH 7.5). c % unchanged compound
detected in the bile by HPLC following iv bolus administration of the compound at either 1 or 10 µmol/kg to anesthetized rats. The bile
duct was cannulated and bile collected over 15 min intervals for a total of 150 min in each case.
17c where this was 0.5 log units. At this stage, these
results supported our original proposal that pyrrole and
imidazole could be used interchangeably.
particularly high for the original series of indoles and
benzimidazoles, to the extent that 45% of the benzimi-
dazole 18 was excreted unchanged in the bile of anes-
thetized rat (Table 4). This value fell slightly as the
polarity of the aniline substituents was reduced by
removing one of the carboxylic acids (23), a change that
was also reflected as an increase in log P. However,
following this trend further by replacing both of the
carboxylic acids by hydroxymethyl groups in the equiva-
lent positions (24) failed to reduce excretion below 22%
and also resulted in a 100-fold loss of activity for the
CCK2 receptor.
Early examples of the pyrrole series (10d and 10e)
were examined in the same model to see whether the
reduction in molecular weight had reduced biliary
excretion. The compounds were selected on the basis
that their in vitro activity was typical of the series and
that the molecules contained a naphthalene chro-
mophore, improving their detection by UV-visible
spectroscopy relative to the 2-phenyl series. We were
disappointed to find that the reduction in molecular
weight alone for the diacids did not reduce biliary
elimination because 59% of the compound was excreted
unchanged by this route (Table 5) with respect to the
In view of the radical change in structure in moving
from the original series, typified by indole 5, to these
new pyrroles and imidazoles, we have assessed a
number of compounds for their ability to behave as
agonists or antagonists at the CCK2 receptor. Repre-
sentative examples, 10a and 17c, were tested in the
isolated, lumen-perfused immature rat stomach assay.46
This is an in vitro assay in which the ability of
compounds to cause acid secretion, or block the actions
of the full agonist tBoc-CCK(29-33) (a small fragment
of the hormone CCK), can be examined. In both cases
the compounds behaved as simple competitive antago-
nists and no agonist response was observed. The pKB
values for the two compounds were 6.3 ( 0.4 (10a) and
5.6 ( 0.4 (17c), in broad agreement with their pKi
values obtained in the CCK2 receptor radioligand bind-
ing assay (Table 3). Therefore, our scaffold-hopping
approach has not altered this particular property of the
original series.
Biliary Elimination of Pyrroles and Imidazoles.
We were aware that levels of biliary elimination were

CCK2 Receptor Pharmacophore
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 22 6799
Table 5. Effect of Modifying the Aniline Substituent of Pyrrole
and Imidazole CCK2 Antagonists on Biliary Elimination
does not rely on relating common structural motifs, it
should also be applicable to the wide range of com-
pounds that have also been described as CCK2 receptor
antagonists in recent years,10 and future investigations
will be directed toward this end.
The new series of pyrroles and imidazoles are not as
potent as the original indoles in vitro but have greater
potential as drug candidates in terms of their reduced
tendency to biliary excretion. This makes them good
lead compounds for development of a new series whose
base structures are not obviously related to those of the
parent compounds. We have also demonstrated that the
reduction of molecular weight from 664 (18) to 511 (10d)
is not sufficient in itself to prevent biliary elimination;
indeed, this rose by 10% in the first examples examined.
However, a concomitant reduction in the polarity of both
the pyrrole and imidazole series, obtained by modifica-
tion of the aniline substituents, produced a significant
reduction in biliary elimination when compared to an
equivalent series of benzimidazoles and gave us two new
series of lead compounds. We have gone on to develop
the imidazole series into compounds with nanomolar
affinity and high bioavailability in vivo, and this work
is described elsewhere.49
% material
excreted
compd
Z
X
Y
MW log P a unchangedb
10d
10e
17d
17e
17f
CH CO2H
CH CO2H
N
N
N
CO2H
H
CO2H
H
511
467
512
468
1.4
2.9
0.7
2.8
59
6
53
5
1
CO2H
CO2H
CH2OH CH2OH 484
>4
a Measured in octanol/buffer (pH 7.5). b % unchanged compound
detected in the bile by HPLC following iv bolus administration of
the compound at 10 µmol/kg to anesthetized rats. The bile duct
was cannulated and bile collected over 15 min intervals for a total
of 150 min in each case.
benzimidazoles (Table 4). However, we were delighted
to find that significant improvements were observed in
the case of the pyrrole monoacid (10e) such that levels
of biliary excretion were now reduced to 6%. Further-
more, the loss of in vitro activity engendered by this
change was less than 10-fold compared to a loss of
almost 100-fold for the same change in the benzimida-
zole series (18-23). The same pattern of behavior was
found for the equivalent imidazoles where removal of
one of the carboxylic acids reduced elimination from 53%
for the diacid 17d to 5% for the monoacid 17e. Further
modification of the two aniline substituents to hy-
droxymethyl groups (17f) reduced this further to 1%.
This suggests that the changes in polarity of the aniline
substituent are important in determining the degree to
which a compound is susceptible to biliary elimination
but that their significance is greater for these lower
molecular weight compounds. While the small ad-
ditional loss of in vitro activity detracted from this
favorable result, the 3-substituted monoacids, 10e and
17e, were nonetheless our preferred choice as lead
compounds for the development of this class of CCK2
antagonists. The results of these studies will be de-
scribed in the accompanying paper.49
Experimental Section
Abbreviations. 1-Ad, adamantan-1-yl; EDC, 1-(3-dimethyl-
aminopropyl)-3-ethylcarbodiimide hydrochloride; DMAP,
4-(dimethylamino)pyridine; HOBt, 1-hydroxybenzotriazole;
OXONE, potassium hydrogen peroxymonosulfate sulfate
(2KHSO5‚KHSO4‚K2SO4, 5:3:2:2, CAS registry no. 70693-62-
8).
General. NMR spectra were recorded on a Bruker DRX 300
spectrometer with chemical shifts reported in ppm relative to
tetramethylsilane used as internal standard. Elemental analy-
ses were determined at the London School of Pharmacy and
are within 0.4% of the theoretical values. Merck silica gel 60
(40-63 µm) was used for column chromatography.
3-Adamantan-1-ylpropionic acid was prepared from 1-bro-
moadamantane,50 and cycloheptylacetic acid was obtained from
cycloheptanone.51 The aniline derivatives 9a and 9b were
prepared from the relevant nitrobenzoic acids, while 9c was
synthesized from 5-nitro-m-xylene-R,R′-diol.
The 1H-pyrrole-3-carboxylic acid intermediates 8a-d were
prepared by the method described below.
2-Adamantan-1-ylmethyl-5-phenyl-1H-pyrrole-3-car-
boxylic Acid (8a). Step a. To a solution of 4-adamantan-1-
yl-3-oxobutyric acid ethyl ester37 (3.00 g, 11.0 mmol) in acetone
(30 mL) was added sodium iodide (0.55 g, 3.67 mmol) and
anhydrous potassium carbonate (3.04 g, 22.0 mmol), followed
by a solution of 2-bromo-1-phenylethanone (2.38 g, 11.5 mmol)
in acetone (10 mL). The mixture was heated at reflux for 36
h, cooled to room temperature, and filtered. The filtrate was
evaporated, and the residue was dissolved in diethyl ether (50
mL) and washed with H2O (2 × 20 mL). The organic phase
was dried (MgSO4), and the solvent was evaporated. The
residue was purified by chromatography on silica gel using
hexanes-EtOAc (4:1) as eluant to afford 4-adamantan-1-yl-
3-oxo-2-(2-oxo-2-phenylethyl)butyric acid ethyl ester as a pale-
yellow oil (1.92 g, 46%). 1H NMR (CDCl3) δ 7.98 (2H, m), 7.65
(1H, m), 7.46 (2H, m), 5.59 (1H, m), 4.19 (3H, m), 3.60 (2H,
dd, J ) 7.2, 17.1 Hz), 2.50 (2H, m), 1.96 (3H, br s), 1.68 (12H,
m), 1.29 (3H, t, J ) 7.2 Hz).
Conclusion
We have demonstrated that field points can be used
as simple descriptors of molecular properties. This
approach has been successfully applied to the design of
a novel series of CCK2 antagonists in a process of
“scaffold hopping” from a known series of indoles with
high in vitro affinity and selectivity but poor bioavail-
ability. We intend to develop this molecular modeling
technique further because we believe that it represents
a simple way of comparing molecules on the basis of
their electrostatic and van der Waals properties rather
than their structural frameworks. The structural di-
versity of the compounds used as input for creation of
the pharmacophore was deliberately conservative be-
cause this allowed us to evaluate the results from the
field point overlays easily. However, since this method
Step b. 4-Adamantan-1-yl-3-oxo-2-(2-oxo-2-phenylethyl)-
butyric acid ethyl ester (1.10 g, 2.88 mmol) and ammonium
acetate (780 mg, 10.1 mmol) were stirred in acetic acid (1.4
mL) at 80 °C for 24 h. The reaction mixture was cooled, then
partitioned between CH2Cl2 (50 mL) and saturated aqueous
NaHCO3 (50 mL). The organic layer was dried over Na2SO4,
and the solvent was evaporated. The residue was crystallized

6800 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 22
Low et al.
from hexanes-EtOAc (4:1) to afford 2-adamantan-1-ylmethyl-
5-phenyl-1H-pyrrole-3-carboxylic acid ethyl ester (7a) as a
white solid (750 mg, 71%). 1H NMR (CDCl3) δ 8.21 (1H, br s),
7.48-7.21 (5H, m), 6.88 (1H, d, J ) 2.7 Hz), 4.30 (2H, q, J )
7.0 Hz), 2.85 (2H, s), 1.96 (3H, br s), 1.60 (12H, m), 1.38 (3H,
t, J ) 7.0 Hz).
Step c. To a solution of 7a (750 mg, 2.06 mmol) in EtOH
(45 mL) was added 6 M sodium hydroxide (5 mL). The mixture
was heated at reflux for 48 h, allowed to cool to room
temperature, and concentrated to a small volume under
reduced pressure. The concentrated solution was diluted with
2 M HCl (40 mL), the precipitated solid was filtered, washed
with H2O, and dried to afford 8a (660 mg, 94%). 1H NMR
(CDCl3) δ 11.5 (1H, br s), 8.26 (1H, br s), 7.48 (2H, m), 7.40
(2H, m), 7.26 (1H, m), 6.94 (1H, d, J ) 3.0 Hz), 2.87 (2H, s),
1.97 (3H, br s), 1.63 (12H, m).
5-[(2-Adamantan-1-ylmethyl-5-phenyl-1H-pyrrole-3-
carbonyl)amino]isophthalic Acid (10a). Step d. To a
suspension of 8a (290 mg, 0.91 mmol) in CH2Cl2 (5 mL) was
added thionyl chloride (200 µL, 2.74 mmol) and 1 drop of DMF.
The mixture was stirred at room temperature for 30 min, and
the solvent was evaporated. Any remaining DMF was removed
by coevaporation with a further two portions of CH2Cl2 (5 mL).
5-Aminoisophthalic acid dibenzyl ester (9a) (361 mg, 1.00
mmol) and anhydrous pyridine (2 mL) were then added to the
residue. The solution was kept at room temperature for 16 h
and diluted with CH2Cl2 (30 mL). The organic phase was
washed with 2 M HCl (2 × 20 mL), brine (20 mL), and dried
(MgSO4), and the solvent was evaporated. The residue was
purified by chromatography on silica gel using CH2Cl2-
hexanes-EtOAc (9:9:2) as eluant to afford 5-[(2-adamantan-
1-ylmethyl-5-phenyl-1H-pyrrole-3-carbonyl)amino]isophthal-
ic acid dibenzyl ester as a pale-yellow solid (300 mg, 45%). 1H
NMR (CDCl3) δ 8.48 (3H, s), 8.30 (1H, br s), 7.69 (1H, s), 7.49-
7.27 (15H, m), 6.66 (1H, d, J ) 2.7 Hz), 5.40 (4H, s), 2.91 (2H,
s), 1.95 (3H, br s), 1.63 (12H, m).
d6) further characterized the product as the N-methyl-D-
glucamine salt (fluffy white solid). Anal. (C30H30N2O3‚C7H17NO5‚
2.5H2O) C, H, N.
3-[(2-Adamantan-1-ylmethyl-5-phenyl-1H-pyrrole-3-
carbonyl)amino]propionic Acid (10f). 1H NMR (DMSO-
d6) further characterized the product as the N-methyl-D-
glucamine salt (fluffy white solid). Anal. (C25H30N2O3‚C7H17NO5‚
H2O) C, H, N.
The 1H-imidazole-4-carboxylic acid intermediates 15a-d
were prepared by the representative method described below
(Scheme 2).
5-Cycloheptylmethyl-2-naphthalen-2-yl-1H-imidazole-
4-carboxylic Acid (15d). Step a. To a solution of cyclohep-
tylacetic acid (10.0 g, 64.0 mmol) and (carbethoxymethylene)-
triphenylphosphorane (22.4 g, 64.0 mmol) in CH2Cl2 (250 mL)
were added EDC (12.25 g, 64.0 mmol) and DMAP (cat.) at 0
°C. The solution was stirred at this temperature for 1 h, then
at room temperature for 16 h. The reaction mixture was
washed with saturated aqueous NaHCO3 (2 × 100 mL) and
dried (MgSO4). Filtration and evaporation of the solvent gave
the crude product, which crystallized from EtOAc to afford
4-cycloheptyl-3-oxo-2-(triphenyl-λ5-phosphanylidenebutyric acid
ethyl ester (23.6 g, 76%). 1H NMR (CDCl3) δ 7.68-7.27 (15H,
m), 3.73 (2H, m), 2.77 (2H, d, J ) 6.9 Hz), 2.07 (1H, m), 1.73-
1.41 (10H, m), 1.22 (2H, m), 0.67 (3H, m).
Step b. To a solution of 4-cycloheptyl-3-oxo-2-(triphenyl-λ5-
phosphanylidenebutyric acid ethyl ester (12.1 g, 25.0 mmol)
in THF (200 mL) and H2O (100 mL) was added OXONE (23.0
g, 37.5 mmol) at 0 °C. The solution was stirred at room
temperature for 16 h and diluted with H2O (300 mL), and the
product was extracted with EtOAc (2 × 150 mL). The organic
phase was dried and filtered, and the solvent was evaporated.
The crude product was purified by chromatography on silica
gel using CH2Cl2-EtOAc (9:1) as eluant, affording 4-cyclohep-
tyl-2,3-dioxobutyric acid ethyl ester hydrate as a pale-yellow
oil (5.70 g, 88%). 1H NMR (CDCl3) δ 4.99 (2H, br s), 4.30 (2H,
q, J ) 6.9 Hz), 2.51 (2H, d, J ) 6.9 Hz), 2.14 (1H, m), 1.67-
1.16 (15H, m).
Step c. To a slurry of ammonium acetate (9.0 g, 116.0 mmol)
in acetic acid (35 mL) was added 4-cycloheptyl-2,3-dioxobutyric
acid ethyl ester hydrate (3.0 g, 11.6 mmol) followed by
2-naphthaldehyde (3.6 g, 23.2 mmol). The mixture was stirred
in an oil bath heated to 70 °C for 2 h. The solution was cooled
to room temperature, and the acetic acid was evaporated. The
residue was dissolved in EtOAc (50 mL) and washed with
saturated aqueous NaHCO3 (2 × 50 mL), H2O (20 mL), and
brine (20 mL). The organic phase was dried (MgSO4), and the
solvent was evaporated. The crude product was purified by
crystallization from EtOAc to afford 5-cycloheptylmethyl-2-
naphthalen-2-yl-1H-imidazole-4-carboxylic acid ethyl ester
(14d) as a white solid (1.74 g, 40%). 1H NMR (CDCl3) δ 10.18
and 9.97 (1H, br s), 8.39 (1H, s), 7.90 (4H, m), 7.52 (2H, br s),
4.40 (2H, q, J ) 9.0 Hz), 2.89 (2H, m), 2.05-1.26 (16H, m).
Step d. To a suspension of 14d (1.73 g, 4.62 mmol) in EtOH
(25 mL) was added a solution of sodium hydroxide (1.29 g, 32.3
mmol) in H2O (5 mL). The reaction mixture was heated under
reflux for 48 h, allowed to cool to room temperature, and
concentrated under reduced pressure. The aqueous solution
was diluted with H2O (30 mL) and acidified to pH 2 with 1 M
HCl. The precipitate was collected by filtration, washed with
H2O, and dried to afford 15d as an off-white solid (1.53 g, 96%).
1H NMR (DMSO-d6) δ 8.77 (1H, s), 8.24-7.98 (4H, m), 7.64
(2H, m), 2.91 (1H, d, J ) 6.9 Hz), 2.64 (1H, d, J ) 6.9 Hz),
2.03 (1H, m), 1.69-1.24 (12H, m).
Step e. A round-bottom flask containing 5-[(2-adamantan-
1-ylmethyl-5-phenyl-1H-pyrrole-3-carbonyl)amino]isophthal-
ic acid dibenzyl ester (180 mg, 0.27 mmol), 10% palladium on
charcoal (50 mg), and THF-MeOH (1:1 mixture, 20 mL) was
evacuated and flushed with hydrogen three times. The mixture
was vigorously stirred overnight under an atmosphere of
hydrogen. The catalyst was removed by filtration and the
filtrate was evaporated to afford 10a as a white solid (130 mg,
1
98%). H NMR (DMSO-d6) δ 13.17 (2H, br s), 11.22 (1H, s),
9.80 (1H, s), 8.64 (2H, s), 8.13 (1H, s), 7.66 (2H, m), 7.39 (2H,
m), 7.19 (2H, m), 2.85 (2H, s), 1.88 (3H, br s), 1.53 (12H, m).
It was further characterized as the di(N-methyl-D-glucamine)
salt (fluffy white solid). Anal. (C30H30N2O5‚2C7H17NO5‚2.0H2O)
C, H, N.
Compounds 10b-d were prepared by a similar sequence for
8b-d. 10e was prepared from 8d by a similar sequence used
to obtain 10a except that 3-aminobenzoic acid benzyl ester (9b)
was used in place of 5-aminoisophthalic acid dibenzyl ester
(9a) in step d. 10f was prepared from 8a by a similar sequence
used to obtain 10a except that â-alanine benzyl ester was used
in place of 5-aminoisophthalic acid dibenzyl ester in step d.
5-[(2-Cycloheptylmethyl-5-phenyl-1H-pyrrole-3-carbo-
nyl)amino]isophthalic Acid (10b). 1H NMR (DMSO-d6)
further characterized the product as the di(N-methyl-D-glu-
camine) salt (fluffy white solid). Anal. (C27H28N2O5‚2C7H17NO5‚
2.0H2O) C, H, N.
5-{[2-(2-Adamantan-1-ylethyl)-5-phenyl-1H-pyrrole-3-
carbonyl]amino}isophthalic Acid (10c). 1H NMR (DMSO-
d6) further characterized the product as the di(N-methyl-D-
glucamine) salt (fluffy white solid). Anal. (C31H32N2O5‚
2C7H17NO5‚3.0H2O) C, H, N.
5-[(5-Cycloheptylmethyl-2-naphthalen-2-yl-1H-imida-
zole-4-carbonyl)amino]isophthalic Acid (17d). Step e. To
a solution of 15d (500 mg, 1.44 mmol) and 5-aminoisophthalic
acid dibenzyl ester (9a) (520 mg, 1.44 mmol) in DMF (3 mL)
were added HOBt (195 mg, 1.44 mmol), DMAP (cat.), and EDC
(280 mg, 1.44 mmol). The solution was kept at room temper-
ature for 72 h and poured over 1 M HCl (20 mL), and the
product was extracted with EtOAc (2 × 20 mL). The product
crystallized from the EtOAc extracts. The crystals were
collected by filtration, dried, and washed with MeOH to afford
5-[(2-Cycloheptylmethyl-5-naphthalene-2-yl-1H-pyrrole-
3-carbonyl)amino]isophthalic Acid (10d). 1H NMR (DMSO-
d6) further characterized the product as the di(N-methyl-D-
glucamine) salt (fluffy white solid). Anal. (C31H30N2O5‚
2C7H17NO5‚3.0H2O) C, H, N.
3-[(2-Cycloheptylmethyl-5-naphthalen-2-yl-1H-pyrrole-
3-carbonyl)amino]Benzoic Acid (10e). 1H NMR (DMSO-

CCK2 Receptor Pharmacophore
Journal of Medicinal Chemistry, 2005, Vol. 48, No. 22 6801
5-[(5-cycloheptylmethyl-2-naphthalen-2-yl-1H-imidazole-4-car-
bonyl)amino]isophthalic acid dibenzyl ester (16d) as a white
Supporting Information Available: Analytical data for
compounds used in biological tests. This material is available
free of charge via the Internet at http://pubs.acs.org.
1
solid (453 mg, 46%). H NMR (DMSO-d6) δ 13.00 (1H, br s),
10.50 (1H, s), 8.84 (2H, s), 8.61 (1H, s), 8.28 (2H, m), 8.00 (3H,
m), 7.50 (12H, m), 5.41 (4H, s), 2.98 (2H, d, J ) 6.0 Hz), 2.00
(1H, m), 1.69-1.16 (12H, m).
References
(1) Black, J. W.; Kalindjian, S. B. Gastrin agonists and antagonists.
Pharmacol. Toxicol. 2002, 91, 275-281.
(2) Kalindjian, S. B.; Bodkin, M. J.; Buck, I. M.; Dunstone, D. J.;
Low, C. M. R.; McDonald, I. M.; Pether, M. J.; Steel, K. I. M. A
new class of nonpeptidic cholecystokinin-B gastrin receptor
antagonists based on dibenzobicyclo[2.2.2]octane. J. Med. Chem.
1994, 37, 3671-3673.
(3) Kalindjian, S. B.; Buck, I. M.; Cushnir, J. R.; Dunstone, D. J.;
Hudson, M. L.; Low, C. M.; McDonald, I. M.; Pether, M. J.; Steel,
K. I.; Tozer, M. J. Improving the affinity and selectivity of a
nonpeptide series of cholecystokinin-B/gastrin receptor antago-
nists based on the dibenzobicyclo[2.2.2]octane skeleton. J. Med.
Chem. 1995, 38, 4294-4302.
Step f. 16d (450 mg, 0.65 mmol) was deprotected using the
same procedure as in step e in the preparation of 10a to afford
17d as white solid (310 mg, 94%). 1H NMR (DMSO-d6) δ 10.75
(1H, s), 8.77 (3H, m), 8.40 (1H, d, J ) 6.0 Hz), 8.25 (1H, s),
8.13 (1H, d, J ) 6.0 Hz), 8.02 (2H, m), 7.65 (2H, m), 3.05 (2H,
d, J ) 9.0 Hz), 2.11 (1H, m), 1.75-1.31 (12H, m). The product
was further characterized as the di(N-methyl-D-glucamine) salt
(fluffy white solid). Anal. (C30H29N3O5‚2C7H17NO5‚4.8H2O) C,
H, N.
17a-c were prepared by a similar sequence from the
carboxylic acid intermediates 15a-c, respectively.
(4) Kalindjian, S. B.; Buck, I. M.; Davies, J. M.; Dunstone, D. J.;
Hudson, M. L.; Low, C. M.; McDonald, I. M.; Pether, M. J.; Steel,
K. I.; Tozer, M. J.; Vinter, J. G. Non-peptide cholecystokinin-B/
gastrin receptor antagonists based on bicyclic, heteroaromatic
skeletons. J. Med. Chem. 1996, 39, 1806-1815.
(5) McDonald, I. M.; Dunstone, D. J.; Kalindjian, S. B.; Linney, I.
D.; Low, C. M.; Pether, M. J.; Steel, K. I.; Tozer, M. J.; Vinter,
J. G. 2,7-Dioxo-2,3,4,5,6,7-hexahydro-1H-benzo[h][1,4]diazonine
as a new template for the design of CCK(2) receptor antagonists.
J. Med. Chem. 2000, 43, 3518-3529.
(6) Jensen, R. T. Involvement of cholecystokinin/gastrin-related
peptides and their receptors in clinical gastrointestinal disorders.
Pharmacol. Toxicol. 2002, 91, 333-350.
(7) Noble, F.; Roques, B. P. CCK-B receptor: chemistry, molecular
biology, biochemistry and pharmacology. Prog. Neurobiol. 1999,
58, 349-379.
(8) Wank, S. A. G protein-coupled receptors in gastrointestinal
physiology. I. CCK receptors: an exemplary family. Am. J.
Physiol. 1998, 274, G607-G613.
(9) Steel, K. Gastrin and gastrin receptor ligands. A review of recent
patent literature. IDrugs 2002, 5, 689-695.
(10) McDonald, I. M. CCK2 receptor antagonists. Expert Opin. Ther.
Pat. 2001, 11, 445-462.
5-[(5-Adamantan-1-ylmethyl-2-phenyl-1H-imidazole-4-
carbonyl)amino]isophthalic Acid (17a). 1H NMR (DMSO-
d6) further characterized the product as the di(N-methyl-D-
glucamine) salt (fluffy white solid). Anal. (C29H29N3O5‚
2C7H17NO5‚0.4H2O) C, H, N.
5-[(5-Cycloheptylmethyl-2-phenyl-1H-imidazole-4-car-
1
bonyl)amino]isophthalic Acid (17b). H NMR (DMSO-d6)
further characterized the product as the di(N-methyl-D-glu-
camine) salt (fluffy white solid). Anal. (C26H27N3O5‚2C7H17NO5‚
2.4H2O) C, H, N.
5-{[5-(2-Adamantan-1-ylethyl)-2-phenyl-1H-imidazole-
4-carbonyl]amino}isophthalic Acid (17c). 1H NMR (DMSO-
d6) further characterized the product as the di(N-methyl-D-
glucamine) salt (fluffy white solid). Anal. (C30H31N3O5‚
2C7H17NO5‚6.2H2O) C, H, N.
3-[(5-Cycloheptylmethyl-2-naphthalen-2-yl-1H-imida-
zole-4-carbonyl)amino]benzoic Acid (17e). 17e was pre-
pared from 15d by a similar sequence used to obtain 17d
except that 3-aminobenzoic acid benzyl ester (9b) was used
in place of 5-aminoisophthalic acid dibenzyl ester (9a) in step
(11) Hill, D. R.; Shaw, T. M.; Graham, W.; Woodruff, G. N. Autora-
diographical detection of cholecystokinin-A receptors in primate
brain using 125I-Bolton Hunter CCK8 and 3H-MK-329. J. Neu-
rosci. 1990, 10.4, 1070-1081.
1
d. H NMR (DMSO-d6) further the product characterized as
the N-methyl-D-glucamine salt (fluffy white solid). Anal.
(C29H29N3O5‚C7H17NO5‚4.8H2O) C, H, N.
(12) Jebbink, M. C. W.; Jansen, J. M. B. J.; Mooy, D. M.; Rovati, L.
C.; Lamers, C. B. H. W. Effect of the specific cholecystokinin
receptor antagonist Loxiglumide on bombesin stimulated pan-
creatic enzyme secretion in man. Regul. Pept. 1991, 32, 361-
368.
(13) Douglas, B. R.; Jebbink, M. C.; Tjon a Tham, R. T.; Jansen, J.
B.; Lamers, C. B. The effect of loxiglumide (CR-1505) on basal
and bombesin-stimulated gallbladder volume in man. Eur. J.
Pharmacol. 1989, 166, 307-309.
(14) Moran, T. H.; Robinson, P. H.; Goldrich, M. S.; McHugh, P. R. 2
Brain cholecystokinin receptors. Implications for behavioral
actions. Brain Res. 1986, 362, 175-179.
(15) Lee, Y. M.; Beinborn, M.; McBride, E. W.; Lu, M.; Kolakowski,
L. F.; Kopin, A. S. The human brain cholecystokinin-B gastrin
receptor. Cloning and characterization. J. Biol. Chem. 1993, 268,
8164-8169.
(16) Ulrich, C. D.; Ferber, I.; Holicky, E.; Hadac, E.; Buell, G.; Miller,
L. J. Molecular-cloning and functional expression of the human
gallbladder cholecystokinin-a receptor. Biochem. Biophys. Res.
Commun. 1993, 193, 204-211.
(17) Pisegna, J. R.; Deweerth, A.; Huppi, K.; Wank, S. A. Molecular-
cloning, functional expression, and chromosomal localization of
the human cholecystokinin type-a receptor. Ann. N. Y. Acad. Sci.
1994, 713, 338-342.
(18) Deweerth, A.; Pisegna, J. R.; Huppi, K.; Wank, S. A. Molecular-
cloning, functional expression and chromosomal localization of
the human cholecystokinin type-a receptor. Biochem. Biophys.
Res. Commun. 1993, 194, 811-818.
(19) Bock, M. G.; DiPardo, R. M.; Evans, B. E.; Rittle, K. E.; Whitter,
W. L.; Veber, D. E.; Anderson, P. S.; Freidinger, R. M. Benzo-
diazepine gastrin and brain cholecystokinin receptor ligands:
L-365,260. J. Med. Chem. 1989, 32, 13-16.
5-Cycloheptylmethyl-2-naphthalen-2-yl-1H-imidazole-
4-carboxylic Acid (3,5-Bis-hydroxymethylphenyl)amide
(17f). 17f was prepared from 15d and 9c according to step d,
and the intermediate 16f was deprotected according to the
procedure of step g.
5-Cycloheptylmethyl-2-naphthalen-2-yl-1H-imidazole-4-car-
boxylic acid (3,5-(dimethoxycarbonyloxymethyl)phenyl)amide
(16f) was obtained according to step e in the sequence used
for the preparation of 17d except that 3,5-dimethoxycarbonyl-
oxymethylaniline (9c) was used in place of 5-aminoisophthalic
acid dibenzyl ester (9a).
Step g. To a solution of 16f (438 mg, 0.72 mmol) in MeOH
(50 mL) was slowly added a 1% aqueous solution of potassium
carbonate. The mixture was stirred at room temperature for
2 h, then heated at reflux for 1 h. The reaction mixture was
allowed to cool to room temperature, and the MeOH was
evaporated. The precipitate was collected by filtration, washed
with H2O, and dried to afford 17f as a white solid (290 mg,
83%). 1H NMR (DMSO-d6) experiments were conducted. Anal.
(C30H33N3O3) C, H, N.
5-[(5-Cycloheptylmethyl-2-(1H-indol-5-yl)-1H-imidazole-
4-carbonyl)amino]isophthalic Acid (17g). 17g was pre-
pared by a sequence similar to the preparation of 17d from
15e. 1H NMR (DMSO-d6) further characterized the product
as the di(N-methyl-D-glucamine) salt (fluffy white solid). Anal.
(C28H28N4O5‚2C7H17NO5‚3.0H2O) C, H, N.
Acknowledgment. We thank Dr. Elaine Harper
and Mr. Eric Griffin for providing the radioligand
binding data and Dr. David Neuhaus, of the MRC
Laboratory of Molecular Biology, Cambridge, U.K., for
his assistance with the NMR experiments described.
(20) Semple, G.; Ryder, H.; Rooker, D. P.; Batt, A. R.; Kendrick, D.
A.; Szelke, M.; Ohta, M.; Satoh, M.; Nishida, A.; Akuzawa, S.;
Miyata, K. (3R)-N-(1-(tert-butylcarbonylmethyl)-2,3-dihydro-2-
oxo-5-(2-pyridyl)-1H-1,4-benzodiazepin-3-yl)-N′-(3-(methylami-
no)phenyl)urea (YF476): a potent and orally active gastrin/
CCK-B antagonist. J. Med. Chem. 1997, 40, 331-341.

6802 Journal of Medicinal Chemistry, 2005, Vol. 48, No. 22
Low et al.
(21) Horwell, D. C.; Hughes, J.; Hunter, J. C.; Pritchard, M. C.;
Richardson, R. S.; Roberts, E.; Woodruff, G. N. Rationally
designed “dipeptoid” analogues of CCK. R-Methyltryptophan
derivatives as highly selective and orally active gastrin and
CCK-B antagonists with potent anxiolytic properties. J. Med.
Chem. 1991, 34, 404-414.
(22) Goodford, P. J. A computational procedure for determining
energetically favorable binding sites on biologically important
macromolecules. J. Med. Chem. 1985, 28, 849-857.
(23) Klebe, G.; Abraham, U.; Mietzner, T. Molecular similarity indices
in a comparative analysis (CoMSIA) of drug molecules to
correlate and predict their biological activity. J. Med. Chem.
1994, 37, 4130-4146.
(24) Cramer, R.; Patterson, D.; Bruce, J. Comparative molecular field
analysis (CoMFA). 1. Effect of shape on binding of steroids to
carrier proteins. J. Am. Chem. Soc. 1988, 110, 5959-5967.
(25) Vinter, J. G.; Trollope, K. I. Multiconformational composite
molecular-potential fields in the analysis of drug-action. 1.
Methodology and first evaluation using 5-HT and histamine
action as examples. J. Comput.-Aided Mol. Des. 1995, 9, 297-
307.
involving competitive ester migrations to C-2, C-4 and N.
Tetrahedron 1990, 46, 3439-3456.
(39) Wasserman, H. H.; Vu, C. B. Formation of vicinal tricarbonyl
compunds by selective oxidation of ylides using potassium
peroxymonosulfate. Tetrahedron Lett. 1990, 31, 5205-5308.
(40) Wasserman, H. H.; Ennis, D. S.; Blum, C. A.; Rotello, V. M. The
conversion of carboxylic acids to keto phosphorane precursors
of 1,2,3-vicinal tricarbonyl compounds. Tetrahedron Lett. 1992,
33, 6003-6006.
(41) Brackeen, M. F.; Stafford, J. A.; Feldman, P. L.; Karanewsky,
D. S. An efficient and mild synthesis of highly substituted
imidazoles. Tetrahedron Lett. 1994, 35, 1635-1638.
(42) XED Software from Cresset BioMolecular Discovery Ltd., Spire-
lla Building, Bridge Road, Letchworth, Herts, SG6 4ET, U.K.
(43) Goodman, J. M. Chemical Applications of Molecular Modelling;
Royal Society of Chemistry: Cambridge, U.K., 1998; pp 186-
188.
(44) Harper, E. A.; Roberts, S. P.; Shankley, N. P.; Black, J. W.
Analysis of variation in L-365,260 competition curves in radio-
ligand binding assays. Br. J. Pharmacol. 1996, 118, 1717-1726.
(45) Hull, R. A. D.; Shankley, N. P.; Harper, E. A.; Gerskowitch, V.
P.; Black, J. W. 2-Naphthalenesulphonyl L-aspartyl-(2-phen-
ethyl)amide (2-Nap), a selective cholecystokinin CCK(A)-receptor
antagonist. Br. J. Pharmacol. 1993, 108, 734-740.
(46) Welsh, N. J.; Shankley, N. P.; Black, J. W. Comparative analysis
of the vagal stimulation of gastric acid secretion in rodent
isolated stomach preparations. Br. J. Pharmacol. 1994, 112, 93-
96.
(47) Horwell, D. C.; Beeby, A.; Clark, C. R.; Hughes, J. Synthesis
and binding affinities of analogues of cholecystokinin-(30-33)
as probes for central nervous system cholecystokinin receptors.
J. Med. Chem. 1987, 30, 729-732.
(48) Low, C. M.; Black, J. W.; Broughton, H. B.; Buck, I. M.; Davies,
J. M.; Dunstone, D. J.; Hull, R. A.; Kalindjian, S. B.; McDonald,
I. M.; Pether, M. J.; Shankley, N. P.; Steel, K. I. Development
of peptide 3D structure mimetics: rational design of novel
peptoid cholecystokinin receptor antagonists. J. Med. Chem.
2000, 43, 3505-3517.
(49) Buck, I. M.; Black, J. W.; Cooke, T.; Dunstone, D. J.; Gaffen, J.
D.; Griffin, E. P.; Harper, E. A.; Hull, R. A. D.; Kalindjian, S.
B.; Lilley, E. J.; Linney, I. D.; Low, C. M. R.; McDonald, I. M.;
Pether, M. J.; Roberts, S. P.; Shankley, N. P.; Shaxted, M. E.;
Steel, K. I. M.; Sykes, D. A.; Tozer, M. J.; Watt, G. F.; Walker,
M. K.; Wright, L.; Wright, P. T. Optimization of the in vitro and
(26) Vinter, J. G. Extended electron distributions applied to the
molecular mechanics of some intermolecular interactions. J.
Comput.-Aided Mol. Des. 1994, 8, 653-668.
(27) Hunter, C.; Sanders, J. The nature of π-π interactions. J. Am.
Chem. Soc. 1990, 112, 5525.
(28) Gundertofte, K.; Liljefors, T.; Norrby, P.-O.; Pettersson, I. A
comparison of conformational energies calculated by several
molecular mechanics methods. J. Comput. Chem. 1996, 17, 429-
449.
(29) Halgren, T. http://www.ccl.net/cca/data/ff_evaluation_suite/.
(30) Chessari, G.; Hunter, C. A.; Low, C. M.; Packer, M. J.; Vinter,
J. G.; Zonta, C. An evaluation of force-field treatments of
aromatic interactions. Chemistry 2002, 8, 2860-2867.
(31) Marshall, G. R.; Cramer, R. D., 3rd. Three-dimensional structure-
activity relationships. Trends Pharmacol. Sci. 1988, 9, 285-289.
(32) Lemmen, C.; Lengauer, T. Computational methods for the
structural alignment of molecules. J. Comput.-Aided Mol. Des.
2000, 14, 215-232.
(33) Hann, M. M.; Leach, A. R.; Harper, G. Molecular complexity and
its impact on the probability of finding leads for drug discovery.
J. Chem. Inf. Comput. Sci. 2001, 41, 856-864.
(34) Lange, J. H.; Kruse, C. G. Keynote review: Medicinal chemistry
strategies to CB1 cannabinoid receptor antagonists. Drug Dis-
covery Today 2005, 10, 693-702.
(35) Lloyd, D. G.; Buenemann, C. L.; Todorov, N. P.; Manallack, D.
T.; Dean, P. M. Scaffold hopping in de novo design. Ligand
generation in the absence of receptor information. J. Med. Chem.
2004, 47, 493-496.
(36) Jenkins, J. L.; Glick, M.; Davies, J. W. A 3D similarity method
for scaffold hopping from known drugs or natural ligands to new
chemotypes. J. Med. Chem. 2004, 47, 6144-6159.
(37) Wierenga, W.; Skulnick, H. I. General, efficient, one-step
synthesis of â-keto-esters. J. Org. Chem. 1979, 44, 310-311.
(38) Chiu, P. K.; Sammes, M. P. The synthesis and chemistry of
azolenines. Part 18. Preparation of 3-ethoxycarbonyl-3H-pyrroles
via the Paal-Knorr reaction, and sigmatropic rearrangements
in vivo properties of
a novel series of 2,4,5-trisubstituted
imidazoles as potent cholecystokinin-2 (CCK2) antagonists. J.
Med. Chem. 2005, 48, 6803-6812.
(50) Ohno, M.; Ishizaki, K.; Eguchi, S. Synthesis of adamantane
derivatives by bridgehead radical addition to electron-deficient
unsaturated bonds. J. Org. Chem. 1988, 53, 1285-1288.
(51) Yokohama, S.; Miwa, T.; Aibara, S.; Fujiwara, H.; Matsumoto,
H.; Nakayama, K.; Iwamoto, T.; Mori, M.; Moroi, R.; Tsukada,
W.; et al. Synthesis and antiallergy activity of [1,3,4]thiadiazolo-
[3,2-a]-1,2,3-triazolo[4,5-d]pyrimidin-9(3H)-one derivatives. II.
6-Alkyl- and 6-cycloalkylalkyl derivatives. Chem. Pharm. Bull.
(Tokyo) 1992, 40, 2391-2398.
JM049069Y