Structure-based generation of a new class of potent Cdk4 inhibitors: New de novo design strategy and library design

Article abstract of DOI:10.1021/jm0103256

As a first step in structure-based design of highly selective and potent Cdk4 inhibitors, we performed structure-based generation of a novel series of Cdk4 inhibitors. A Cdk4 homology model was constructed according to X-ray analysis of an activated form of Cdk2. Using this model, we applied a new de novo design strategy which combined the de novo design program LEGEND with our in-house structure selection supporting system SEEDS to generate new scaffold candidates. In this way, four classes of scaffold candidates including diarylurea were identified. By constructing diarylurea informer libraries based on the structural requirements of Cdk inhibitors in the ATP binding pocket of the Cdk4 model, we were able to identify a potent Cdk4 inhibitor N-(9-oxo-9H-fluoren-4-yl)-N′-pyridin-2-ylurea 15 (IC₅₀ = 0.10 μM), together with preliminary SAR. We performed a docking study between 15 and the Cdk4 model and selected a reasonable binding mode which is consistent with the SAR. Further modification based on the proposed binding mode provided a more potent compound, N-[(9bR)-5-oxo2,3,5,9b-tetrahydro-1H-pyrrolo[2,1-a]isoindol-9-yl]-N′ -pyridin-2-ylurea 26a (IC₅₀ = 0.042 μM), X-ray analysis of which was accomplished by the soaking method. The predicted binding mode of 15 in Cdk4 was validated by X-ray analysis of the Cdk2-26a complex.

Full text of DOI:10.1021/jm0103256

J . Med. Chem. 2001, 44, 4615-4627
4615
Str u ctu r e-Ba sed Gen er a tion of a New Cla ss of P oten t Cd k 4 In h ibitor s: New d e
Novo Design Str a tegy a n d Libr a r y Design
Teruki Honma,* Kyoko Hayashi, Tetsuya Aoyama, Noriaki Hashimoto, Takumitsu Machida,
Kazuhiro Fukasawa, Toshiharu Iwama, Chinatsu Ikeura, Mari Ikuta, Ikuko Suzuki-Takahashi,
Yoshikazu Iwasawa, Takashi Hayama, Susumu Nishimura, and Hajime Morishima
Banyu Tsukuba Research Institute in collaboration with Merck Research Laboratories,
Okubo-3, Tsukuba 300-2611, Ibaraki, J apan
Received J uly 17, 2001
As a first step in structure-based design of highly selective and potent Cdk4 inhibitors, we
performed structure-based generation of a novel series of Cdk4 inhibitors. A Cdk4 homology
model was constructed according to X-ray analysis of an activated form of Cdk2. Using this
model, we applied a new de novo design strategy which combined the de novo design program
LEGEND with our in-house structure selection supporting system SEEDS to generate new
scaffold candidates. In this way, four classes of scaffold candidates including diarylurea were
identified. By constructing diarylurea informer libraries based on the structural requirements
of Cdk inhibitors in the ATP binding pocket of the Cdk4 model, we were able to identify a
potent Cdk4 inhibitor N-(9-oxo-9H-fluoren-4-yl)-N′-pyridin-2-ylurea 15 (IC50 ) 0.10 µM),
together with preliminary SAR. We performed a docking study between 15 and the Cdk4 model
and selected a reasonable binding mode which is consistent with the SAR. Further modification
based on the proposed binding mode provided a more potent compound, N-[(9bR)-5-oxo-
2,3,5,9b-tetrahydro-1H-pyrrolo[2,1-a]isoindol-9-yl]-N′-pyridin-2-ylurea 26a (IC50 ) 0.042 µM),
X-ray analysis of which was accomplished by the soaking method. The predicted binding mode
of 15 in Cdk4 was validated by X-ray analysis of the Cdk2-26a complex.
In tr od u ction
lead generation as a first step. The results of lead
optimization to improve Cdk4 selectivity will be reported
in an accompanying article.
Cyclins and cyclin-dependent kinases (Cdks) play
important roles in regulation of the cell cycle.1 In
particular, D-type cyclins, which are activated by rear-
rangement or amplification in several tumors,2 associate
with Cdk4/6. Cyclin D-Cdk4/6 complexes phosphorylate
the retinoblastoma protein (pRB) and regulate the cell
cycle during G1/S transition.3 Loss of function or deletion
of p16ink4a (an endogenous Cdk4/6 specific inhibitor
protein) frequently occurs in clinical cancer cells.4
Several Cdk inhibitors such as UCN-01,5 flavopiridol,6
purvalanol B,7 pyrido[2,3-d]pyrimidin-7-one,8 and in-
denopyrazole9 have been reported. Among these, non-
specific Cdk inhibitors, UCN-01 and flavopiridol, have
entered clinical trials. As a next generation of Cdk
inhibitors, selective inhibitors of only one target Cdk
are expected to cause cell cycle arrest specifically.10
Suppression of tumor growth by G1 arrest is thought to
reduce the stress for normal cells more than in other
phases, because normal cells are usually resting in the
G0-G1 phase. Therefore, we tried to design Cdk4 selec-
tive inhibitors that cause cell cycle arrest in the G1
phase.
Recently, several successful examples of structure-
based lead generation such as the HIV-1 integrase
inhibitor, FKBP-12 ligand, Factor Xa inhibitor, COX-2
inhibitor, and DNA gyrase inhibitor have been re-
ported.11 In most cases, methods for structure-based
lead generation were based on computational descrip-
tions of a binding site, as in the coordinates of atoms or
pharmacophores, and also based on techniques to search
for the configurational and conformational space of a
candidate molecule in the binding site, evaluating
potential energy and/or scoring of binding affinity. These
methods are classified into two categories: 3D database
search12 and de novo design.13 In the case of 3D database
search, programs such as DOCK,14 UNITY,15 and Cata-
lyst16 take each molecule from known compound data-
bases and attempt to position it in the active site of a
receptor or pharmacophore model. Three-dimensional
database search is more frequently used because hit
compounds are just listed in the commercial catalogues
or synthetically accessible. However, most 3D database
search programs have one or more of the following
critical problems: (1) the number of considered confor-
mations for each ligand is often not sufficient, (2) some
functional groups work as the ionic form in a physi-
ological condition (pH 7.4), whereas 3D database search
programs often use their neutral forms, and (3) novel
structures cannot be obtained. On the other hand, de
novo design programs such as LEGEND,17 LeapFrog,18
CONCERTS,19 PRO-LIGAND,20 and LUDI21 sequen-
tially build up structures which are predicted to fit the
To obtain highly selective and potent Cdk4 inhibitors,
we performed a structure-based design consisting of the
following two steps: (1) lead generation of a new class
of Cdk4 inhibitors based on a Cdk4 homology model and
(2) enhancement of Cdk4 selectivity of lead compounds
over Cdk1/2 and the other kinases based on the binding
modes and structural differences between Cdk4 and
other kinases. In this paper, we describe structure-based
* To whom correspondence should be addressed. Tel: 81-298-77-
2000. Fax: 81-298-77-2029. E-mail: honmatr@banyu.co.jp.
10.1021/jm0103256 CCC: $20.00 © 2001 American Chemical Society
Published on Web 12/13/2001

4616 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 26
Honma et al.
Ta ble 1. Reported X-ray Analyses of the Cdk Family and Hydrogen-Bonding Sites Used by Small Ligands23-25,29-33
hydrogen-bonding sites used by small ligands
PDB ID
molecule
E81 CO L83 NH L83 CO K33 NúH D86 Oδ Q131 Oꢀ D145 Oδ N132 Oδ Wat74a
1HCL
1B39
1HCK
1FIN
1J ST
Cdk2
Cdk2c-ATP
Cdk2-ATP
NH
NH
NH
NH
NH
NH
N
PO
PO
OH
OH
N
OH
OH
Cdk2-ATP-cyclin A2
N
Cdk2c-ATPb-cyclin A2
N
PO
PO
PO
1QMZ Cdk2c-ATPf-cyclin A2-substrated
N
OH
1DM2
1CKP
Cdk2-hymenialdisine
Cdk2-Purvalanol B
Cdk2-L828276e
Cdk2-isopentenyladenyne
Cdk2-olomoucine
Cdk2-roscovitine
Cdk2-staurosporin
Cdk2-Ckshs1
Cdk2-cyclinA2-p27
Cdk6-p16
Cdk6-p19
Cdk6-p19
CO
N
NH
NH
NH
CO
OH
OH
NH
CO
N
N
N
CO
OH
N
NH+,OH
NH
NH
OH
1AQ1
1BUH
1J SU
1BI7
1BI8
1BLX
NH
NH+
a
d
Wat74 is a water molecule in 1J ST. b ATPγS. c Thr160 phosphorylated Cdk2. Substrate peptide (HHASPRK). e des-Chloroflavopiridol.
f AMPPNP.
active site of the receptor. In the case of de novo design,
analysis of monomeric Cdk2 was reported by H. L. De
Bondt et al.23 In 1996, the structure of the Thr160
phosphorylated Cdk2-cyclin A-ATPγS complex, a final
activated form of Cdk2, was solved by A. A. Russo et
al.24 They revealed the activation mechanism of Cdk2
by analyzing conformational changes with cyclin A and
phosphorylation of Thr160. Although the structures of
the Cdk6-p16 complex and Cdk6-p19 complex were
already known, their cyclin binding region and the ATP
binding pocket in Cdk6 were both largely altered by the
endogenous inhibitors, p16 or p19.25 Therefore, despite
a very high sequence identity between Cdk4 and Cdk6
(70%), we could not use these structural data for
designing Cdk4 inhibitors. On the other hand, the level
of sequence homology between Cdk4 and Cdk2 is quite
high (45%) so that it seemed appropriate to construct a
Cdk4 model based on the final activated form of Cdk2
(Protein Data Bank (PDB) ID: 1J ST).
According to the sequence alignments by S. K. Hanks
et al.,26 we constructed an initial Cdk4 model using the
modeling software BIOCES[E].27 The model was mini-
mized with the CHARMm (Chemistry at Harvard
Molecular mechanics) force field28 with the exception
of the conserved region in Cdk4 and Cdk2. This mini-
mized structure was used for scaffold design in the de
novo design strategy.
considering conformations and forms of functional groups
would not be critical problems, because de novo design
programs generate functional groups in suitable forms
and suitable conformations to interact with the protein
surface. Furthermore, de novo design programs can also
construct structures including novel skeletons. We
applied this de novo design strategy to discover novel
lead compounds. However, in contrast to structures
obtained from 3D database search programs, output
structures obtained from de novo design programs
usually have greater difficulty in commercial avail-
ability and synthetic feasibility.
To overcome these problems in the conventional de
novo design strategies, we constructed the chemical
structure selection supporting system, SEEDS (system
for evaluation of availability of essential structures
generated by de novo ligand design programs),22 taking
advantage of database searches. The program SEEDS
automatically extracts an essential part of each struc-
ture by the de novo design program, and searches for
commercially available and/or synthetically accessible
derivatives of the essential parts. In this study, we used
the de novo design program LEGEND with this SEEDS
to generate novel scaffold candidates with molecular
weights of less than 350.
After the discovery of some promising scaffold can-
didates, it is important to validate whether the scaffold
candidates bind to Cdk4 in desired ways. Recently,
efficient validation methods based on NMR analysis
have been reported (SAR by NMR11c and needle
screening11f). However, the molecular weight of Cdk4
(44 kDa) is so large that measurement of its NMR is
not feasible. Thus, we synthesized several informer
libraries of a scaffold and, using the preliminary struc-
ture-activity relationship (SAR), validated the binding
mode by molecular modeling methods. Then modifica-
tions based on the predicted binding mode were per-
formed to discover more potent lead compounds.
St r u ct u r e-Ba sed Gen er a t ion of New Sca ffold
Ca n d id a tes. According to several binding modes of
ATPγS and ATP competitive Cdk2 inhibitors deter-
mined by X-ray analysis, we examined the structural
requirements for binding of the inhibitors in the ATP
binding pocket (Table 1). In Cdk2, the main chain NH
group in Leu83 would be the most important in binding
of the inhibitors because it serves as a hydrogen-bonding
donor in every structure reported so far. The main chain
carbonyl groups of Glu81 and Leu83 also seemed to be
important because most inhibitors form a hydrogen
bond(s) with them. In addition, Lys33 NúH, Asp86 Oδ,
Glu131 Oꢀ, and a highly conserved water molecule
(Wat74 in 1J ST) are utilized by some ligands. In terms
of hydrophobicity, some areas deep in the ATP binding
pocket around Glu81 and Leu83 are occupied by flat
aromatic rings of Cdk2 inhibitors, such as flavopiridol,29
staurosporin,30 and roscovitine.31 Indeed, these areas
Resu lts a n d Discu ssion
Hom ology Mod elin g of Cd k 4 Ba sed on th e X-r a y
Str u ctu r e of Cd k 2. X-ray analyses of the Cdk family
are summarized in Table 1. In 1993, the first X-ray

Structure-Based Generation of Cdk4 Inhibitors
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 26 4617
F igu r e 1. Structural requirements of ATP-competitive Cdk inhibitors. A: L828276 (des-chloroflavopiridol)-Cdk2 complex.29 B:
staurosporin-Cdk2 complex (1AQ1 in PDB).30 C: roscovitine-Cdk2 complex.31 D: predicted structural requirements in the Cdk4
homology model.
are surrounded by the hydrophobic amino acid residues
Ile10, Val18, Ala31, Val64, Phe80, Phe82, Leu134, and
Ala144 and form a narrow cavity. The binding modes
of representative Cdk2 inhibitors and the locations of
the pharmacophores are shown schematically in Figure
1A-C. In Cdk4, most of the residues that are important
for hydrogen bonds or hydrophobic interactions are
conserved according to Hanks' multiple alignment.
Among the altered amino acid residues, those between
Leu83 in Cdk2 and Val96 in Cdk4 would not be critical
since only the main chain is used for hydrogen bond(s).
Therefore, we considered that the information concern-
ing these structural requirements is also useful for
discovering a new class of Cdk4 inhibitors.
To identify new scaffolds that satisfy these structural
requirements, we applied a de novo design program,
LEGEND. This program is based on the atom-by-atom
approach34 and is suitable for generation of drug
molecules in a deep, narrow cavity just like the ATP
binding pocket. With this program, we can specify the
directions of growth, the size of the molecules that will
be generated, and the hydrogen bonds that will be
formed during calculations. Specification of the calcula-
tion performed by LEGEND and the process of scaffold
generation are shown in Scheme 1.
available nor synthetically feasible. This problem ap-
pears to apply to most de novo design programs that
build up structures sequentially. To overcome this
disadvantage, we developed an in-house program,
SEEDS.22 The program SEEDS automatically picks out
an essential substructure of each output from LEGEND
and uses it in search of commercially available and/or
synthetically feasible derivatives. This process is out-
lined in Figure 2 using an example. Structure I is one
of the output structures generated by LEGEND. If we
searched compound databases using structure I itself
as a query, no commercially available derivatives could
be obtained because of simple alkyl side chains which
would not form hydrogen bonds. Therefore, SEEDS
extracted rings and functional groups that were pre-
dicted to form hydrogen bonds as an essential structure
(II). Subsequently, based on structure II, several queries
for searching compound databases and reaction data-
bases were automatically made by SEEDS. Query III
is one of the examples for searching compound data-
bases. Using the queries made by SEEDS, we first
searched for commercially available derivatives. When
attractive scaffold candidates from commercially avail-
able derivatives were not identified, we searched for
synthetically feasible derivatives using reaction data-
bases. On searching the Available Chemicals Directory
(ACD),35 4884 compounds with molecular weights of less
At a primary step, we obtained 1000 output struc-
tures; however, most of them were neither commercially

4618 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 26
Honma et al.
F igu r e 2. Schematic figure of the SEEDS. I: one of the output structures generated by the LEGEND. II: an essential part of
structure I. III: one of the queries for searching compound databases.
Sch em e 1. Process of Structure-Based Scaffold
Generation
than 350 were selected. Compounds with skeletons or
functional groups that were unacceptable for the devel-
opment of lead compounds were omitted, and finally 382
commercially available compounds remained. The com-
pounds were purchased for screening in cyclin D-Cdk4
F igu r e 3. Typical classes of hit compounds and their IC50
assays at concentrations up to 1 mM. We obtained 18
values in cyclin D-Cdk4 assay. Bold lines represent essential
parts of output structures by LEGEND.
compounds with IC50 values of under 500 µM as hits.
Most of the hits had large, flat aromatic rings and a
couple of neighboring hydrogen-bonding donors and
acceptors. Twelve of the 18 hits were clustered into one
of the following four classes according to their structural
profile: diarylurea, cyclic urea and thiourea, 4-aryl-2-
aminopyridine and triazine, and pteridine (Figure 3).
In these four classes, our interest was directed to the
diarylurea scaffold, because there were five potent
compounds, 1, 2, 3, 4, and 5, with a diarylurea structure
that is appropriate for rapid construction of structurally
diverse libraries. Therefore, we constructed diarylurea

Structure-Based Generation of Cdk4 Inhibitors
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 26 4619
Ta ble 2. Diarylurea Informer Library (2): Transformation of
7-Hydroxynaphthyl Group
F igu r e 4. Diarylurea informer library (1): transformation of
5-chloro-2-methylphenyl group.
informer libraries by parallel synthesis to validate the
potential of the scaffold and to obtain preliminary SAR.
Con str u ction of Dia r ylu r ea In for m er Libr a r ies.
The data obtained through LEGEND suggested that
neighboring NH and CO in the diarylurea scaffolds form
hydrogen bond(s) with the main chain(s) of Glu94 and/
or Val96, and aromatic rings are supposed to be located
in the hydrophobic regions as shown in Figure 1.
According to these insights, we designed informer
libraries as described below.
binding pocket. Therefore, as in the case of the first
informer library, we mainly selected aromatic amines
as building blocks. Using these building blocks, 410 urea
compounds were synthesized and tested by Cdk4 inhibi-
tory assay. These compounds were provided by the
coupling reaction between alkylamines and pyridine-2-
carbonyl azide. The structures and assay results for
representative compounds are summarized in Table 2.
This area required an aromatic ring, because none of
the aliphatic substituted compounds showed inhibitory
activity (14e and 14f are representatives). The existence
of a hydrogen-bonding acceptor in the left part (14a -c
and 15) led to a significant improvement of potency, in
particular, with compound 15 (IC50 ) 0.10 µM). Com-
pound 15 was selected as the lead compound for further
modification.
P r ed iction of th e Bin d in g Mod e of Dia r ylu r ea
15 in Cd k 4. To investigate the binding mode of 15 more
comprehensively, we performed a docking study be-
tween the Cdk4 model and 15. First we made 400
conformations of the Cdk4-15 complex model by manual
docking, random sampling, and minimization with the
modeling software QUANTA.36 These conformations
were clustered by root-mean-square deviation (RMSD)
of atomic coordinates (40 clusters), and representative
conformations in each cluster with the lowest energy
in the CHARMm force field were extracted. Then, the
representative conformations were filtered by the fol-
lowing two conditions based on structural requirements
for being ATP competitive Cdk inhibitors as shown in
Figure 1: (1) the NH group of Val96 forms a hydrogen
bond with 15 and (2) the aromatic ring(s) of 15 is located
in the hydrophobic regions. Four patterns of binding
modes were suggested (Figure 5A-D).
First, we searched for substituents for the 5-chloro-
2-methylphenyl group in the diarylurea 1, leaving the
7-hydroxynaphthyl group intact. Considering the deep,
narrow shape of the ATP binding pocket, we mainly
selected flat aromatic amines as building blocks to
synthesize diarylurea derivatives. Fifty-five urea com-
pounds were synthesized from 8-isocyanato-2-naphthol
and alkylamines in moderate yields (30-70%). In ad-
dition, we prepared an amide compound 13 to confirm
the importance of the urea moiety. The resultant SAR
is summarized in Figure 4. Substitutions with 2-pyridi-
nyl (14) and 2-thiazolyl (1d ) groups gave IC50 values of
7.6 µM and 23 µM, respectively, while substitutions with
3-pyridinyl (1b) and 4-pyridinyl (1c) groups showed poor
inhibitory activities. The relative position of the nitrogen
atom in the aromatic ring seemed to be important for
binding of the inhibitor. None of the compounds sub-
stituted with aliphatic amines showed enhanced po-
tency. Transformation of NH-3 (numbering of atoms is
shown in Figure 4) into CH2 (13) caused a remarkable
decrease in Cdk4 inhibition. These SAR suggested (1)
the nitrogen atom on 2-substituted pyridine and 2-sub-
stituted thiazole form a hydrogen bond with protein or
an intramolecular hydrogen-bonding site (i.e., NH-1 of
the urea moiety) and (2) the NH-3 of the urea moiety is
important as a hydrogen-bonding donor.
In all candidates, formation of an intramolecular
hydrogen bond between urea NH-1 (numbering of atoms
is shown in Figure 6) and the nitrogen atom (N-5) on
the pyridine ring was predicted (Table 3), and the
fluorenone moiety was located in the hydrophobic region
in Figure 1. According to SAR of the informer libraries,
Next we kept the 2-pyridinyl group intact and modi-
fied the 7-hydroxynaphthyl group in compound 14. The
7-hydroxynaphthyl group was predicted to be located
in the hydrophobic, narrow, deep area of the ATP

4620 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 26
Honma et al.
F igu r e 5. Four predicted binding mode candidates of 15 in the Cdk4 model. Right green ribbon: ribbon representation of the
Cdk4 model. Green residues: conserved amino acid residues in Cdk4 and Cdk2. Red residues: different amino acid residues in
Cdk4 and Cdk2. Broken lines: predicted hydrogen bonds.
Ta ble 3. Predicted Hydrogen Bonds in the Binding Mode
Candidatesa
binding
mode
hydrogen-bonding sites
candidates
CO-4
NH-1
CO-2
NH-3
N-5
mode A
mode B
mode C
mode D
Lys35 NúH N-5
Val96 NH
Val96 NH
Val96 CO NH-1
NH-1
N-5
Val96 NH
Val96 NH
N-5
N-5
Asp99 Oδ NH-1
Lys35 NúH
NH-1
a
Numbering of atoms at hydrogen-bonding sites is shown in
Figure 5. J udgment of formation of hydrogen bonds was performed
by QUANTA.
hydrophobic requirements (Figure 6). Compared to
mode A, CO-2 in mode C could not make a hydrogen
bond with any residues, and the pyridine ring
deviated from the locations for hydrophobic require-
ments. Therefore, we concluded that mode A was
preferable to mode C.
Design a n d Syn th esis Ba sed on th e P r ed icted
Bin d in g Mod e. In most conformations containing the
cluster of mode A, the angle between the fluorenone and
the pyridine was rather large (>30°) even in the narrow
cavity. This suggests some degree of steric repulsion
between the pyridine ring and the terminal benzene ring
in the fluorenone. The large angle caused by the
repulsion was supposed to be unfavorable for binding
affinity. On the other hand, the terminal benzene ring
F igu r e 6. Schematic figure of mode A.
we were able to propose additional hypotheses for the
validation of the binding modes as follows: (1) NH-3 of
the urea moiety makes a hydrogen bond with the
protein and (2) fluorenone CO-4 makes a hydrogen bond
with the protein. Modes B and D could not satisfy these
hypotheses at the same time, whereas modes A and C
satisfied them simultaneously.
Next, we made a comparison of modes A and C. In
mode A, the neighboring CO-2 and NH-3 formed two
hydrogen bonds with the main chain of Val96, and both
fluorenone and pyridine were located in accord with

Structure-Based Generation of Cdk4 Inhibitors
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 26 4621
Ta ble 4. Designed Compounds and Their IC50 Values in Cyclin
D-Cdk4 Assay
F igu r e 7. Suitable directions to introduce substituents in
mode A. Cdk4 model: solvent accessible surface (probe: 1.4
Å) representation (colored by partial charge) by WebLab
viewer. 15: stick representation.
in the fluorenone was supposed to be important for
binding affinity in mode A since the terminal benzene
ring was appropriately located on the hydrophobic
surface of Cdk4 (Figure 7).
From these assumptions, we attempted to replace the
terminal benzene ring with various kinds of five-mem-
bered rings in order to eliminate unnecessary intramo-
lecular steric repulsion while maintaining favorable
hydrophobic interactions. Representative compounds 25,
26a , and 26b and their IC50 values are shown in Table
4, and their synthetic routes are summarized in Scheme
2. Fused pyrrole 19 was synthesized by the intramo-
lecular Heck reaction.37 Reduction of 19 by Fe/HCl
provided 23. Saturated chiral amines 21a and 21b were
obtained by hydrogenation with Pd/C and separation
using chiral HPLC. The absolute configuration of 21b
was determined by X-ray analysis of the (-)-camphanic
amide of 21b. Compounds 25, 26a , and 26b were
synthesized by coupling reactions between these amines
and pyridine-2-carbonyl azide. These modifications
provided potent compounds with IC50 values of up to
0.034 µM.
According to mode A, we also found some space for
the introduction of substituents on the aromatic rings.
We introduced substituents in the two positions shown
by arrows in Figure 7. As a result, 27 with a substituent
at the 5-position of the terminal pyrrole ring in 25
showed an IC50 value of 0.017 µM, and 28 with
2-thiophene at the 4-position of the pyridine showed an
IC50 value of 0.011 µM. We performed docking studies
between the Cdk4 model and 27/28 and recognized that
the molecules were not able to bind to Cdk4 in the way
of mode B, C, or D. This suggested that mode A would
be the best candidate.
26a (solubility at pH 7.4 in Tris-Cl buffer, 7.18 µg/mL)
with potent binding affinity. By using compound 26a ,
we succeeded in X-ray analysis of the Cdk2-26a
complex (PDB ID: 1GIH).38 A comparison between mode
A and the X-ray structure is shown in Figure 8. The
X-ray analysis showed that 26a formed hydrogen bonds
with the main chain NH and CO groups of Leu83
(corresponding to Val96 in Cdk4) and adopted a U-
shaped conformation in the ATP binding pocket of Cdk2.
This binding mode of 26a in Cdk2 (Figure 8B) was found
to be consistent with the predicted binding mode of 15
in Cdk4 (mode A, Figure 8A).
Supported by the results of the docking study with
the Cdk4 model, X-ray analysis of the Cdk2 complex,
and the SAR obtained by the Cdk4 inhibitory assay, we
concluded that mode A was the most preferable binding
mode of the diarylurea series of lead compounds in the
activated form of the cyclin D-Cdk4 complex.
Con clu sion
We employed a structure-based lead generation ap-
proach consisting of the following steps: (1) homology
modeling of the target protein, (2) scaffold generation
by using the de novo design program LEGEND and our
in-house structure selection supporting system SEEDS,
(3) construction of informer libraries, and (4) structural
modification based on the predicted binding mode. From
the result, we developed a novel diarylurea class of
potent inhibitor, 26a (IC50 ) 0.042 µM). Combination
Con fir m a tion of th e P r ed icted Bin d in g Mod e by
X-r a y An a lysis. For further validation of the binding
mode, we tried X-ray analysis of the Cdk2-15 complex
by the soaking method. However, compound 15 was so
insoluble (solubility at pH 7.4 in Tris-Cl buffer, <0.01
µg/mL) that soaking was not appropriate. During modi-
fication of 15, we obtained a relatively soluble compound

4622 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 26
Honma et al.
Sch em e 2. Synthetic Routes of Diarylurea Derivatives 25, 26a , 26b, 27, and 28a
a
Reagents: (a) (1) thionyl chloride; (2) pyrrole (for 17) or 2-acetylpyrrole (for 18), triethylamine, CH2Cl2, 36% (for 17), 67% (for 18); (b)
Pd(PPh3)4, KOAc, dimethylacetamide, 81% (for 19), 62% (for 20); (c) Fe, 6 N HCl, MeOH, 83% (for 23) or Pd/C, H2, MeOH-tetrahydrofuran,
100% (for 24); (d) pyridine-2-carbonyl azide, tetrahydrofuran, 30-70%; (e) Pd/C, H2, MeOH, 99%; (f) (1) 4-tributylstannylpyridine-2-
carbonyl azide, tetrahydrofuran, 79%; (2) Pd(PPh3)4, 2-bromothiophene, dimethylformamide, 76%; (g) p-nitrobenzoyl chloride, triethylamine,
CHCl3, 83%; (h) chiral HPLC; (i) LiBH4, MeOH, tetrahydrofuran, 22% (for 21a ), 34% (for 21b).
F igu r e 8. Comparison of modeling prediction and X-ray analysis. A: mode A of 15 in the Cdk4 model. B: X-ray analysis of the
Cdk2-26a complex (PDB ID: 1GIH).38 Green residues: conserved amino acid residues in Cdk4 and Cdk2. Red residues: different
amino acid residues in Cdk4 and Cdk2. Broken lines: predicted hydrogen bonds.
of the LEGEND and the SEEDS was so effective that
we were able to overcome the difficulty in utilizing
output structures from the de novo design program and
identified some promising scaffolds including diarylurea.
Construction of informer libraries, prediction of the
binding mode by molecular modeling, and modification
based on the predicted binding mode led efficiently to
potent lead compounds (0.01 µM order) from initial
compounds (10-100 µM order). Recently, high-through-
put screening (HTS) has become an essential technique
in lead identification. It is expected that it should
accelerate assay throughput and reduce assay costs.
However, the success rate of lead discovery for various
kinds of drug discovery targets is not supposed to be
changed dramatically, because the success rate depends
on whether the screening source has sufficient size and
structural diversity. When a reliable 3D structure or
model of a target protein is available, our methodology
of structure-based lead generation would be an alterna-
tive approach to development of lead compounds.
The binding mode of 15 in Cdk4 as predicted by
molecular modeling was confirmed by SAR and X-ray
analysis of the Cdk2-26a complex. Compound 26a
showed moderately higher selectivity (>50-fold) over the

Structure-Based Generation of Cdk4 Inhibitors
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 26 4623
other kinases39 with the exception of the Cdk family.
However, 26a inhibited Cdk1/2 with comparable IC50
values (0.120 µM/0.078 µM). As a next step, we designed
highly selective Cdk4 inhibitors based on this binding
mode and structural information about differences
between Cdk4 and other kinases. The results are
reported in the following article.
(1H, t, J ) 7.8 Hz), 7.30 (1H, s), 7.46 (2H, d, J ) 5.6 Hz), 7.57
(1H, d, J ) 9.6 Hz), 7.76 (2H, t, J ) 9.6 Hz), 8.36 (2H, d, J )
5.6 Hz), 8.64 (1H, s), 9.39 (1H, s), 9.80 (1H, s). MS (FAB+):
m/z 280 (M+H)+. HR-MS calcd for C16H14N3O2 (M+H)+:
280.1086, found 280.1092.
N-(7-Hydr oxy-1-n aph th yl)-N′-(1,3-th iazol-2-yl)u r ea (1d).
1H NMR (300 MHz, DMSO-d6): δ 7.13 (2H, m), 7.24 (1H, t, J
) 7.8 Hz), 7.28 (1H, s), 7.39 (1H, d, J ) 5.6 Hz), 7.57 (1H, d,
J ) 9.6 Hz), 7.78 (2H, t, J ) 9.6 Hz), 7.83 (2H, d, J ) 5.6 Hz),
8.89 (1H, s), 9.84 (1H, s), 10.77 (1H, s). MS (FAB+): m/z 286
(M+H)+. HR-MS calcd for C14H12N3O2S (M+H)+: 286.0650,
found 286.0650.
N-Ben zyl-N′-(7-h yd r oxy-1-n a p h th yl)u r ea (1e). 1H NMR
(300 MHz, DMSO-d6): δ 4.34 (2H, d, J ) 5.6 Hz), 6.94 (1H, t,
J ) 5.6 Hz), 7.08 (1H, d, J ) 6.7 Hz), 7.16 (1H, t, J ) 7.8 Hz),
7.21-7.38 (6H, m), 7.44 (1H, d, J ) 7.8 Hz), 7.72 (1H, d, J )
9.6 Hz), 7.80 (1H, t, J ) 7.8 Hz), 8.32 (1H, s), 9.73 (1H, s). MS
(FAB+): m/z 293 (M+H)+. HR-MS calcd for C18H17N2O2
(M+H)+: 293.1290, found 293.1293.
Exp er im en ta l Section
Ma ter ia ls a n d Meth od s. All reagents and solvents were
of commercial quality and were used without further purifica-
tion unless otherwise noted. Melting points were determined
with a Yanaco MP micromelting point apparatus (Yanagimoto
Seisakusho Co., J apan) and were not corrected. 1H NMR
spectra were obtained on a Varian Gemini-300 (300 MHz) or
Gemini-200 (200 MHz) with tetramethylsilane as an internal
standard. Mass spectrometry was performed with a J EOL
J MS-SX 102A (FAB positive) or micromass QUATTRO II (ESI
positive). High-resolution mass spectra (HR-MS) were deter-
mined in a micromass Q-TOF2 (ESI positive). Optical rotations
were measured with a J asco DIP-370 polarimeter. Thin-layer
chromatography (TLC) analysis was performed on Merck
Kieselgel F254 precoated plates. Silica gel column chromatog-
raphy was carried out on Wako gel C-300 (Wako Pure
Chemicals Industries, J apan).
Gen er a l P r oced u r e for P r ep a r a tion of N-Alk yl-N′-(7-
h yd r oxy-1-n a p h th yl)u r ea . To a stirred solution of 1-amino-
7-hydroxynaphthalene (1.59 g, 10 mmol) in pyridine (30 mL)
was added methyl chloroformate (850 mL, 11 mmol) at 0 °C.
After the mixture was stirred for 30 min, methyl chloroformate
(430 mL, 5.6 mmol) was added. After being stirred for 15 min,
the reaction mixture was poured into water. The aqueous layer
was extracted with EtOAc. The combined organic layer was
washed successively with 2 N HCl, water, and brine, dried over
MgSO4, and concentrated in vacuo. The residue was purified
by silica gel column chromatography (EtOAc) to give methyl
7-hydroxy-1-naphthylcarbamate (2.20 g, 100%) as a purple oil.
1H NMR (300 MHz, CDCl3): δ 6.72 (1H, brs), 7.09 (1H, d, J )
9.6 Hz), 7.30 (1H, t, J ) 8.0 Hz), 7.63 (1H, d, J ) 8.0 Hz), 7.68
(1H, m), 7.73 (1H, d, J ) 9.6 Hz). MS (ESI+): m/z 218 (M+H)+.
HR-MS calcd for C12H12NO3 (M+H)+: 218.0817, found 218.0828.
A solution of methyl 7-hydroxy-1-naphthylcarbamate (58 mg,
0.27 mmol) in toluene (3 mL) was mixed with triethylamine
(82 mL, 0.59 mmol) and refluxed. After 10 min, B-chlorocat-
echolborane (82 mg, 0.53 mmol) was added. After 15 min, 1.2
equiv of amine was added and refluxed for 10 min. The
reaction mixture was poured into water. The aqueous layer
was extracted with EtOAc. The combined organic layer was
washed successively with saturated aqueous NaHCO3 and
brine, dried over MgSO4, and concentrated in vacuo. The
residue was purified by silica gel column chromatography to
give N-alkyl-N′-(7-hydroxy-1-naphthyl)urea. Yields ranged
from 30 to 70%.
N-(7-H yd r oxy-1-n a p h t h yl)-N′-isop r op ylu r ea (1f). 1H
NMR (200 MHz, DMSO-d6): δ 1.12 (6H, d, J ) 7.2 Hz), 3.77
(1H, heptet, J ) 7.2 Hz) 6.38 (1H, d, J ) 7.5 Hz), 7.07 (1H, d,
J ) 8.6 Hz), 7.16 (1H, d, J ) 7.5 Hz), 7.28 (1H, s), 7.42 (1H, d,
J ) 7.5 Hz), 7.72 (1H, d, J ) 8.6 Hz), 7.80 (1H, d, J ) 7.5 Hz),
8.07 (1H, s), 9.69 (1H, s). MS (FAB+): m/z 245 (M+H)+. HR-
MS calcd for C14H17N2O2 (M+H)+: 245.1290, found 245.1298.
N-Cyclop en tyl-N′-(7-h yd r oxy-1-n a p h th yl)u r ea (1g). 1H
NMR (300 MHz, DMSO-d6): δ 1.33-1.92 (8H, m), 3.98(1H,
m) 6.55 (1H, d, J ) 7.7 Hz), 7.08 (1H, d, J ) 9.0 Hz), 7.14 (1H,
t, J ) 7.7 Hz), 7.28 (1H, s), 7.40 (1H, d, J ) 7.7 Hz), 7.72 (1H,
d, J ) 9.0 Hz), 7.82 (1H, d, J ) 7.7 Hz), 8.06 (1H, s), 9.69 (1H,
s). MS (FAB+): m/z 271 (M+H)+. HR-MS calcd for C16H19N2O2
(M+H)+: 271.1447, found 271.1439.
N-(7-Hyd r oxy-1-n a p h th yl)-N′-p yr id in -2-ylu r ea (14). mp
1
224-225 °C. H NMR (300 MHz, DMSO-d6): δ 7.06 (1H, t, J
) 5.6 Hz), 7.12 (1H, d, J ) 6.7 Hz), 7.24 (1H, t, J ) 7.8 Hz),
7.37 (2H, m), 7.55 (1H, d, J ) 7.8 Hz), 7.78 (2H, m), 8.00 (1H,
d, J ) 6.7 Hz), 8.40 (1H, d, J ) 5.6 Hz), 9.82 (1H, s), 9.88 (1H,
s), 11.18 (1H, brs). MS (FAB+): m/z 280 (M+H)+. HR-MS calcd
for C16H14N3O2 (M+H)+: 280.1086, found 280.1088. Anal.
(C16H13N3O2‚0.33H2O) C, H, N.
N-(7-Hyd r oxy-1-n a p h th yl)-2-p h en yla ceta m id e (13). To
a stirred solution of 1-amino-7-hydroxynaphthalene (159 mg,
1.0 mmol) in tetrahydrofuran (10 mL) were added phenylacetic
acid (164 mg, 1.0 mmol) and dicyclohexyl carbodiimide (250
mg, 1.2 mmol), and the resultant mixture was stirred at room
temperature for 10 h. The reaction mixture was concentrated
in vacuo. The residue was purified by silica gel column
chromatography (hexane-EtOAc (2:1)) and precipitated with
1
EtOAc to give 13 (214 mg, 77%). H NMR (200 MHz, DMSO-
d6): δ 3.77 (2H, s), 7.10 (1H, d, J ) 8.2 Hz), 7.17-7.44
(7H, m), 7.48 (1H, d, J ) 8.2 Hz), 7.61 (1H, d, J ) 8.2 Hz),
7.76 (1H, d, J ) 9.4 Hz), 9.79 (1H, s), 9.95 (1H, s). MS
(FAB+): m/z 278 (M+H)+. HR-MS calcd for C18H16NO2
(M+H)+: 278.1181, found 278.1183.
The following compounds (1a , 1b, 1c, 1d , 1e, 1f, 1g, and
14) were synthesized from the appropriate amines and methyl
7-hydroxy-1-naphthylcarbamate by this method.
Gen er a l P r oced u r e for P r ep a r a tion of N-Alk yl-N′-
p yr id in -2-ylu r ea . To a stirred solution of thionyl chloride (10
mL) was added picolinic acid (2.47 g, 20 mmol) with heating
at 85 °C for 1 h. The reaction mixture was evaporated to afford
crude picolinic chloride as a dark purple solid. The crude
picolinic chloride was dissolved in water (20 mL) at 0 °C.
Sodium azide (1.84 g, 28 mmol) was added, and the resultant
mixture was stirred at room temperature for 15 min and
poured into saturated aqueous NaHCO3. The aqueous layer
was extracted with EtOAc. The combined organic layer was
washed with brine, dried over MgSO4, and concentrated in
vacuo. The residue was purified by rapid silica gel column
chromatography (hexane-EtOAc (1:1)) to yield pyridine-2-
N-(7-Hyd r oxy-1-n a p h th yl)-N′-p h en ylu r ea (1a ). 1H NMR
(200 MHz, DMSO-d6): δ 6.96 (1H, t, J ) 6.8 Hz), 7.12 (1H,
dd, J 1 ) 8.9, J 2 ) 1.6 Hz), 7.24 (1H, d, J ) 6.8 Hz), 7.30 (2H,
d, J ) 7.9 Hz), 7.33 (1H, m), 7.47 (2H, d, J ) 7.9 Hz), 7.51
(1H, m), 7.78 (2H, t, J ) 8.9 Hz), 8.46 (1H, s), 9.00 (1H, s),
9.98 (1H, s). MS (FAB+): m/z 279 (M+H)+. HR-MS calcd for
C
17H15N2O2 (M+H)+: 279.1134, found 279.1132.
N-(7-Hyd r oxy-1-n a p h th yl)-N′-p yr id in -3-ylu r ea (1b). 1H
NMR (300 MHz, DMSO-d6): δ 7.10 (1H, d, J ) 6.7 Hz), 7.22
(1H, t, J ) 7.8 Hz), 7.33 (2H, m), 7.55 (1H, d, J ) 7.8 Hz),
7.98 (1H, d, J ) 6.7 Hz), 8.18 (1H, d, J ) 5.6 Hz), 8.60 (2H, d,
J ) 9.6 Hz), 9.14 (1H, s), 9.80 (1H, s). MS (FAB+): m/z 280
(M+H)+. HR-MS calcd for C16H14N3O2 (M+H)+: 280.1086,
found 280.1091.
N-(7-Hyd r oxy-1-n a p h th yl)-N′-p yr id in -4-ylu r ea (1c). 1H
NMR (300 MHz, DMSO-d6): δ 7.11 (1H, d, J ) 6.7 Hz), 7.23
1
carbonyl azide (1.06 g, 34%). H NMR (300 MHz, DMSO-d6):
δ 7.55 (1H, dd, J 1 ) 9.0 Hz, J 2 ) 6.7 Hz), 7.88 (1H, t, J ) 9.0
Hz), 8.17 (1H, d, J ) 9.0 Hz), 8.77 (1H, d, J ) 6.7 Hz). MS
(ESI+): m/z 149 (M+H)+. A solution of pyridine-2-carbonyl
azide (25 mg, 0.17 mmol) in tetrahydrofuran (2 mL) was heated
at 85 °C. After 30 min, 1.0 equiv of amine was added, and the

4624 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 26
Honma et al.
resultant mixture was stirred for 2 h at the same temperature.
The reaction mixture was evaporated, and the residue was
precipitated from hexane-EtOAc (1:1) to give N-alkyl-N′-
pyridin-2-ylurea. Yields ranged from 10 to 70%.
) 7.2 Hz, J 2 ) 5.1 Hz), 7.28-7.33 (2H, m), 7.46 (1H, t, J ) 7.7
Hz), 7.76-7.82 (1H, m), 8.29-8.32 (2H, m), 9.95 (1H, s), 11.19
(1H, brs). MS (FAB+): m/z 309 (M+H)+. HR-MS calcd for
C17H17N4O2 (M+H)+: 309.1352, found 309.1353. [R]D 162.0°
20
(c 1.0, CHCl3). Anal. (C17H16N4O2‚0.33H2O) C, H, N.
The following compounds (14a , 14b, 14c, 14d , 14e, 14f, 15,
25, 26a , 26b, 27, and 28) were prepared from the appropriate
amines and pyridine-2-carbonyl azide by this procedure.
N-(1H-In d a zol-6-yl)-N′-p yr id in -2-ylu r ea (14a ). mp 220-
N-[(9b S)-5-Oxo-2,3,5,9b -t et r a h yd r o-1H -p yr r olo[2,1-a ]
isoin d ol-9-yl]-N′-p yr id in -2-ylu r ea (26b). mp 213-215 °C.
1H NMR (300 MHz, DMSO-d6): δ 1.06-1.20 (1H, m), 2.30-
2.43 (2H, m), 2.52-2.57 (1H, m), 3.28-3.35 (1H, m), 3.50-
3.60 (1H, m), 4.83 (1H, dd, J 1 ) 10.0 Hz, J 2 ) 5.7 Hz), 7.06
(1H, dd, J 1 ) 7.2 Hz, J 2 ) 5.1 Hz), 7.28-7.33 (2H, m), 7.46
(1H, t, J ) 7.7 Hz), 7.76-7.82 (1H, m), 8.29-8.32 (2H, m),
9.95 (1H, s), 11.19 (1H, brs). MS (FAB+): m/z 309 (M+H)+.
1
221 °C. H NMR (300 MHz, DMSO-d6): δ 6.95-6.99 (2H, m),
7.48 (1H, d, J ) 8.2 Hz), 7.66 (1H, d, J ) 8.2 Hz), 7.75 (1H, t,
J ) 8.2 Hz), 7.95 (1H, s), 8.06 (1H, s), 8.30 (1H, d, J ) 4.7
Hz), 9.49 (1H, s), 10.73 (1h, brs). MS (ESI+): m/z 254 (M+H)+.
HR-MS calcd for C13H12N5O (M+H)+: 254.1042, found 254.1065.
Anal. (C13H11N5O‚0.25H2O) C, H, N.
HR-MS calcd for
309.1351. [R]D -156.0° (c 1.0, CHCl3). Anal. (C17H16N4O2‚
0.2H2O) C, H, N.
C
17H17N4O2 (M+H)+: 309.1352, found
20
N-P yr id in -2-yl-N′-qu in olin -5-ylu r ea (14b). mp 213-215
1
°C. H NMR (300 MHz, DMSO-d6): δ 7.06 (1H, dd, J 1 ) 7.3
Hz, J 1 ) 5.1 Hz), 7.42 (1H, d, J ) 8.2 Hz), 7.64-7.71 (1H, m),
7.73-7.83 (3H, m), 8.23 (1H, dd, J 1 ) 6.7 Hz, J 1 ) 2.1 Hz),
8.41 (1H, dd, J 1 ) 5.1 Hz, J 1 ) 2.1 Hz), 8.54 (1H, d, J ) 8.8
Hz), 8.95 (1H, dd, J 1 ) 4.1 Hz, J 1 ) 1.4 Hz), 9.88 (1H, s), 11.41
(1H, brs). MS (ESI+): m/z 265 (M+H)+. HR-MS calcd
for C15H13N4O (M+H)+: 265.1089, found 265.1096. Anal.
(C15H12N4O) C, H, N.
N-(3-(1-Hydr oxyvin yl)-5-oxo-5H-pyr r olo[2,1-a ]isoin dol-
9-yl)-N′-p yr id in -2-ylu r ea (27). mp >300 °C. 1H NMR (300
MHz, DMSO-d6): δ 6.32-6.35 (1H, m), 6.74 (1H, s), 7.07 (1H,
dd, J 1 ) 7.2 Hz, J 2 ) 5.2 Hz), 7.19 (1H, s), 7.26 (1H, s), 7.40
(1H, t, J ) 8.0 Hz), 7.47 (1H, d, J ) 8.6 Hz), 7.66 (1H, dd, J 1
) 7.9 Hz, J 2 ) 1.5 Hz), 7.78-7.83 (1H, m), 8.25 (1H, dd, J 1
)
5.2 Hz, J 2 ) 1.6 Hz), 8.47 (1H, dd, J 1 ) 8.0 Hz, J 2 ) 1.6 Hz),
10.06 (1H, s), 10.85 (1H, brs), 11.96 (1H, s). MS (FAB+): m/z
347 (M+H)+. HR-MS calcd for C19H15N4O3 (M+H)+: 347.1144,
found 347.1149.
N-(1,3-Dioxo-2,3-d ih yd r o-1H-isoin d ol-4-yl)-N′-p yr id in -
1
2-ylu r ea (14c). mp >300 °C. H NMR (300 MHz, DMSO-d6):
δ 7.04-7.10 (1H, m), 7.19 (1H, d, J ) 8.3 Hz), 7.41 (1H, d, J
) 6.9 Hz), 7.71-7.82 (2H, m), 8.35 (1H, dd, J 1 ) 4.9 Hz, J 1
)
N-(5-Oxo-2,3,5,9b-tetr ah ydr o-1H-pyr r olo[2,1-a ]isoin dol-
9-yl)-N′-(4-th ien -2-ylp yr id in -2-yl)u r ea (28). A solution of
4-(tributylstannyl)pyridine-2-carbonyl azide (300 mg, 0.69
mmol) in tetrahydrofuran (10 mL) was heated at 85 °C. After
30 min, 21 (150 mg, 0.79 mmol) was added, and the resultant
mixture was stirred for 2 h at the same temperature. The
reaction mixture was evaporated, and the residue was
precipitated from hexane-EtOAc (1:1) to yield N-(5-oxo-
2,3,5,9b-tetrahydo-1H-pyrrolo[2,1-a]isoindol-9-yl)-N′-[4-(tribu-
tylstannyl)pyridin-2-yl]urea (326 mg, 79%). mp 149-150 °C.
1H NMR (300 MHz, CDCl3): δ 0.89 (9H, t, J ) 8.0 Hz), 1.04-
1.20 (1H, m), 1.10-1.15 (6H, m), 1.27-1.36 (6H, m), 1.51-
1.58 (6H, m), 2.40-2.48 (2H, m), 2.64-2.68 (1H, m), 3.46-
3.50 (1H, m), 3.75-3.79 (1H, m), 4.80 (1H, dd, J 1 ) 10.5 Hz,
J 2 ) 5.5 Hz), 6.92 (1H, s), 7.07 (1H, d, J ) 4.9 Hz), 7.47 (1H,
t, J ) 7.9 Hz), 7.53 (1H, d, J ) 7.9 Hz), 8.01 (1H, brs), 8.09
(1H, d, J ) 4.9 Hz), 8.36 (1H, d, J ) 7.9 Hz), 12.07 (1H, brs).
HR-MS calcd for C29H43N4O2120Sn (M+H)+: 599.2408, found
599.2418. To a solution of N-(5-oxo-2,3,5,9b-tetrahydo-1H-
pyrrolo[2,1,-a]isoindol-9-yl)-N′-[4-(tributylstannyl)pyridin-2-yl]-
urea (200 mg, 0.34 mmol) in dimethylformamide (15 mL) were
added 2-bromothiophene (164 mg, 1.0 mmol) and tetrakis-
(triphenylphosphin)palladium (100 mg, 0.086 mmol). The
resultant mixture was stirred for 10 h at 110 °C, and poured
into water. The aqueous layer was extracted with EtOAc. The
combined organic layer was washed successively with satu-
rated aqueous NH4Cl, water, and brine, dried over MgSO4, and
concentrated in vacuo. The residue was purified by silica gel
column chromatography (3% MeOH in CHCl3) and precipitated
from EtOAc-Et2O to provide 28 (102 mg, 76%) as a white
solid. mp 245-247 °C. 1H NMR (300 MHz, DMSO-d6): δ 1.08-
1.20 (1H, m), 2.26-2.45 (2H, m), 2.48-2.61 (1H, m), 3.29-
3.39 (1H, m), 3.49-3.59 (1H, m), 4.80-4.88 (1H, m), 7.23 (1H,
t, J ) 4.0 Hz), 7.32 (1H, d, J ) 6.6 Hz), 7.38 (1H, d, J ) 4.0
Hz), 7.47 (1H, t, J ) 6.6 Hz), 7.59 (1H, brs), 7.70-7.76 (2H,
m), 8.30 (2H, d, J ) 6.6 Hz), 9.97 (1H, s), 11.03 (1H, br).
1.6 Hz), 8.74 (1H, d, J ) 8.3 Hz), 10.19 (1H, s), 11.29 (1H, s),
12.57 (1H, brs). MS (ESI+): m/z 283 (M+H)+. HR-MS calcd
for C14H11N4O3 (M+H)+: 283.0831, found 283.0822. Anal.
(C14H10N4O3) C, H, N.
N-P h en yl-N′-p yr id in -2-ylu r ea (14d ). mp 186-188 °C. 1H
NMR (300 MHz, DMSO-d6): δ 6.97-7.04 (2H, m), 7.30 (1H,
d, J ) 7.4 Hz), 7.46-7.53 (3H, m), 7.71-7.76 (1H, m), 8.27
(1H, dd, J 1 ) 5.0 Hz, J 1 ) 1.9 Hz), 9.41 (1H, s), 10.48 (1H, s).
MS (ESI+): m/z 214. (M+H)+. HR-MS calcd for C12H12N3O
(M+H)+: 214.0980, found 214.0977. Anal. (C12H11N3O) C, H,
N.
N-Ben xyl-N′-p yr id in -2-ylu r ea (14e). mp 143-145 °C. 1H
NMR (300 MHz, CDCl3): δ 4.63 (1H, d, J ) 5.9 Hz), 6.79 (1H,
d, J ) 8.2 Hz), 6.83-6.87 (1H, m), 7.23-7.41 (5H, m), 7.56
(1H, t, J ) 8.2 Hz), 8.12 (1H, d, J ) 4.1 Hz), 8.60 (1H, s), 9.83
(1H, br). MS (ESI+): m/z 228. (M+H)+. HR-MS calcd
for C13H14N3O (M+H)+: 228.1137, found 228.1135. Anal.
(C13H13N3O‚0.1H2O) C, H, N.
N-Cycloh exyl-N′-p yr id in -2-ylu r ea (14f). mp 122-123 °C.
1H NMR (300 MHz, CDCl3): δ 1.26-1.51 (5H, m), 1.59-1.79
(3H, m), 2.00-2.06 (2H, m), 3.82-3.86 (1H, m), 6.82 (1H, d, J
) 8.4 Hz), 6.84-6.88 (1H, m), 7.58 (1H, t, J ) 8.4 Hz), 8.18
(1H, dd, J 1 ) 5.1 Hz, J 1 ) 1.1 Hz), 8.27 (1H, brs), 9.32 (1H,
m). MS (ESI+): m/z 220. (M+H)+. HR-MS calcd for C12H18N3O
(M+H)+: 220.150, found 220.1450. Anal. (C12H17N3O) C, H,
N.
N-(9-Oxo-9H-flu or en -4-yl)-N′-p yr id in -2-ylu r ea (15). mp
>300 °C. 1H NMR (300 MHz, DMSO-d6): δ 7.07 (1H, dd, J 1
)
8.3 Hz, J 2 ) 5.1 Hz), 7.34-7.45 (4H, m), 7.64-7.69 (2H, m),
7.78-7.84 (1H, m), 8.04 (1H, d, J ) 7.9 Hz), 8.08 (1H, d, J )
7.7 Hz), 8.29 (1H, dd, J 1 ) 5.0 Hz, J 2 ) 1.2 Hz), 10.03 (1H, s),
11.12 (1H, brs). MS (FAB+): m/z 316 (M+H)+. HR-MS calcd
for C19H14N3O2 (M+H)+: 316.1086, found 316.1084. Anal.
(C19H13N3O2) C, H, N.
N-(5-Oxo-5H-p yr r olo[2,1-a ]isoin d ol-9-yl)-N′-p yr id in -2-
HR-MS calcd for
391.1228.
C
21H19N4O2S (M+H)+: 391.1229, found
1
ylu r ea (25). mp 268-269 °C. H NMR (300 MHz, DMSO-d6):
δ 6.34 (1H, t, J ) 3.1 Hz), 6.65 (1H, d, J ) 3.1 Hz), 7.08 (1H,
dd, J 1 ) 6.7 Hz, J 2 ) 5.6 Hz), 7.24-7.29 (3H, m), 7.38 (1H, d,
J ) 7.3 Hz), 7.77-7.83 (1H, m), 8.27 (1H, d, J ) 8.2 Hz),
8.31 (1H, dd, J 1 ) 5.1 Hz, J 2 ) 1.1 Hz), 10.06 (1H, brs), 10.97
(1H, brs). MS (FAB+): m/z 305 (M+H)+. HR-MS calcd for
1-(2-Br om o-3-n itr oben zoyl)-1H-p yr r ole (17). To a stirred
solution of thionyl chloride (20 mL) was added 2-bromo-3-
nitrobenzoic acid (2.0 g, 8.1 mmol), and the resultant mixture
was heated at 80 °C for 3 h. After cooling to room temperature,
the reaction mixture was concentrated in vacuo. The residue
was dissolved in dichloromethane (20 mL). Then pyrrole (2.3
mL, 33 mmol) and triethylamine (5 mL, 36 mmol) were added.
The reaction mixture was stirred at room temperature for 10
h and poured into water. The aqueous layer was extracted with
EtOAc. The combined organic layer was washed successively
C
17H13N4O2 (M+H)+: 305.1039, found 305.1035.
N-[(9bR)-5-Oxo-2,3,5,9b-tetr a h yd r o-1H-p yr r olo[2,1-a ]i-
soin d ol-9-yl]-N′-p yr id in -2-ylu r ea (26a ). mp 213-215 °C. 1H
NMR (300 MHz, DMSO-d6): δ 1.06-1.20 (1H, m), 2.30-2.43
(2H, m), 2.52-2.57 (1H, m), 3.28-3.35 (1H, m), 3.50-3.60 (1H,
m), 4.83 (1H, dd, J 1 ) 10.0 Hz, J 2 ) 5.7 Hz), 7.06 (1H, dd, J 1

Structure-Based Generation of Cdk4 Inhibitors
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 26 4625
with saturated aqueous NH4Cl, water, and brine, dried over
MgSO4, and concentrated in vacuo. The residue was purified
by silica gel column chromatography (hexane-EtOAc (2:1)) to
(44 mL, 0.32 mmol) and p-nitrobenzoyl chloride (59 mg, 0.32
mmol). After being stirred for 3 h, the reaction mixture was
poured into a saturated NaHCO3 solution. The aqueous layer
was extracted with CHCl3. The combined organic layer was
washed with brine, dried over MgSO4, and concentrated in
vacuo. The residue was purified by silica gel column chroma-
tography (2% MeOH in CHCl3) to give racemate 22 (45 mg,
1
afford 17 (865 mg, 36%) as a colorless oil. H NMR (300 MHz,
CDCl3): δ 6.26 (1H, d, J ) 2.0 Hz), 6.38 (2H, t, J ) 2.3 Hz),
6.83 (1H, d, J ) 2.1 Hz), 7.63 (2H, d, J ) 5.0 Hz), 7.93 (1H, d,
J ) 4.9 Hz). MS (ESI+): m/z 296 (M+H)+.
1
83%) as a slight yellow foam. H NMR (300 MHz, CDCl3): δ
2-Acetyl-1-(2-br om o-3-n itr oben zoyl)-1H-p yr r ole (18).
To a stirred solution of thionyl chloride (30 mL) was added
2-bromo-3-nitrobenzoic acid (10 g, 41 mmol), and the resultant
mixture was heated at 80 °C for 3 h. After cooling to room
temperature, the reaction mixture was evaporated. The resi-
due was dissolved in dichloromethane (100 mL). Then
2-acetylpyrrole (8.9 g, 81 mmol) and triethylamine (20 mL, 163
mmol) were added. The resultant mixture was stirred at room
temperature for 10 h, and poured into water. The aqueous
layer was extracted with EtOAc. The combined organic layer
was washed successively with saturated aqueous NH4Cl,
water, and brine, dried over MgSO4, and concentrated in vacuo.
The residue was purified by precipitation from hexane-CHCl3
(2:1) to give 18 (9.2 g, 67%) as a white solid. mp 170-172 °C.
1H NMR (300 MHz, CDCl3): δ 2.0.27 (3H, s), 6.47 (1H, t, J )
3.3 Hz), 7.38 (1H, d, J ) 3.3 Hz), 7.49 (1H, d, J ) 7.9 Hz),
7.61-7.68 (2H, m), 8.11 (1H, d, J ) 7.9 Hz). MS (ESI+):
m/z 336 (M+H)+. HR-MS calcd for C13H10N2O479Br (M+H)+:
336.9824, found 336.9835.
1.25-1.31 (1H, m), 2.20-2.27 (1H, m), 2.33-2.42 (2H, m),
3.42-3.47 (1H, m), 3.68-3.72 (1H, m), 4.92 (1H, dd, J 1 ) 11.2
Hz, J 2 ) 5.9 Hz), 7.53 (1H, d, J ) 7.8 Hz), 7.68 (1H, t, J ) 7.8
Hz), 7.82 (1H, d, J ) 7.8 Hz), 8.12 (2H, d, J ) 8.8 Hz), 8.22
(1H, brs), 8.38 (2H, d, J ) 8.8 Hz). HR-MS calcd for C18H16N3O4
(M+H)+: 338.1141, found 338.1134.
N-[(9bR)-5-Oxo-2,3,5,9b-tetr a h yd r o-1H-p yr r olo[2,1-a ]i-
soin d ol-9-yl]-4-n itr oben za m id e (22a ) a n d N-[(9bS)-5-Oxo-
2,3,5,9b-tetr a h yd r o-1H-p yr r olo[2,1-a ]isoin d ol-9-yl]-4-n i-
tr oben za m id e (22b). Compound 22 (22 g, 66 mmol) was
separated by chiral HPLC (CHIRALPAK AD, 50φ × 500 mm,
hexane-EtOH (1:4), flow rate 100 mL/min) to give 22a
(retention time: 30 min, 10.1 g) and 22b (retention time: 22
1
min, 11.2 g) as pale yellow foams. 22a : H NMR (300 MHz,
CDCl3): δ 1.25-1.31 (1H, m), 2.20-2.27 (1H, m), 2.33-2.42
(2H, m), 3.42-3.47 (1H, m), 3.68-3.72 (1H, m), 4.92 (1H, dd,
J 1 ) 11.2 Hz, J 2 ) 5.9 Hz), 7.53 (1H, d, J ) 7.8 Hz), 7.68 (1H,
t, J ) 7.8 Hz), 7.82 (1H, d, J ) 7.8 Hz), 8.12 (2H, d, J ) 8.8
Hz), 8.22 (1H, brs), 8.38 (2H, d, J ) 8.8 Hz). HR-MS calcd for
9-Nitr o-5H-pyr r olo[2,1-a ]isoin dol-5-on e (19). To a stirred
solution of 17 (865 mg, 2.9 mmol) in dimethylacetamide (20
mL) were added potassium acetate (575 mg, 6.0 mmol) and
tetrakis(triphenylphosphin)palladium (100 mg, 0.086 mmol),
and the resultant mixture was heated at 130 °C for 10 h. After
cooling to room temperature, the reaction mixture was poured
into water. The aqueous layer was extracted with EtOAc. The
combined organic layer was washed successively with water
and brine, dried over MgSO4, and concentrated in vacuo. The
residue was purified by silica gel column chromatography
(hexane-EtOAc (5:1)) to yield 19 (509 mg, 81%) as a red solid.
C
18H16N3O4 (M+H)+: 338.1141, found 338.1135. [R]D 134.4°
20
(c 1.0, CHCl3). 22b: 1H NMR (300 MHz, CDCl3): δ 1.25-1.31
(1H, m), 2.20-2.27 (1H, m), 2.33-2.42 (2H, m), 3.42-3.47 (1H,
m), 3.68-3.72 (1H, m), 4.92 (1H, dd, J 1 ) 11.2 Hz, J 2 ) 5.9
Hz), 7.53 (1H, d, J ) 7.8 Hz), 7.68 (1H, t, J ) 7.8 Hz), 7.82
(1H, d, J ) 7.8 Hz), 8.12 (2H, d, J ) 8.8 Hz), 8.22 (1H, brs),
8.38 (2H, d, J ) 8.8 Hz). HR-MS calcd for C18H16N3O4
(M+H)+: 338.1141, found 338.1128. [R]D -132.8° (c 1.0,
20
CHCl3).
(9b R)-9-Am in o-1,2,3,9b -t et r a h yd r o-5H -p yr r olo[2,1-a ]
isoin d ol-5-on e (21a ). To a stirred solution of 22a (10.0 g, 29.6
mmol) in tetrahydrofuran (300 mL) was added lithium boro-
hydride (3.86 g, 177 mmol) at room temperature. Then MeOH
(9.7 mL, 237 mmol) in tetrahydrofuran (10 mL) was added in
a dropwise manner. The resultant mixture was stirred for 5 h
at room temperature and poured into 3 N HCl. The aqueous
layer was neutralized with saturated aqueous NaHCO3 and
extracted with CHCl3. The combined organic layer was washed
with brine, dried over MgSO4, and concentrated in vacuo. The
residue was purified by silica gel column chromatography (2%
MeOH in CHCl3) and recrystallized from EtOAc to give 21a
(1.2 g, 22%, 100% ee). mp 151-152 °C. 1H NMR (300 MHz,
DMSO-d6): δ 0.80-0.93 (1H, m), 2.10-2.30 (2H, m), 2.43-
2.51 (1H, m), 3.18-3.24 (1H, m), 3.38-3.47 (1H, m), 4.50 (1H,
dd, J 1 ) 10.0 Hz, J 1 ) 5.5 Hz), 5.34 (2H, s), 6.72 (1H, d, J )
7.9 Hz), 6.76 (1H, d, J ) 7.9 Hz), 7.11 (1H, t, J ) 7.9 Hz).
HR-MS calcd for C11H13N2O (M+H)+: 189.1028, found 189.1027.
1
mp 152-153 °C. H NMR (300 MHz, CDCl3): δ 6.33 (1H, t, J
) 3.2 Hz), 7.08 (1H, d, J ) 3.2 Hz), 7.19 (1H, d, J ) 3.2 Hz),
7.35 (1H, t, J ) 8.3 Hz), 7.92 (1H, d, J ) 8.3 Hz), 8.27 (1H, d,
J ) 8.3 Hz). MS (ESI+): m/z 215 (M+H)+.
3-(1-Hyd r oxyvin yl)-9-n itr o-5H-p yr r olo[2,1-a ]isoin d ol-
5-on e (20). To a stirred solution of 18 (2.0 g, 5.9 mmol) in
dimethylacetamide (40 mL) were added potassium acetate
(1.16 g, 12 mmol) and tetrakis(triphenylphosphin)palladium
(700 mg, 0.61 mmol), and the resultant mixture was heated
at 110 °C for 2 h. After cooling to room temperature, the
reaction mixture was poured into water. The aqueous layer
was extracted with EtOAc. The combined organic layer was
washed successively with water and brine, dried over MgSO4,
and concentrated in vacuo. The residue was purified by silica
gel column chromatography (hexane-EtOAc (2:1)) to yield 20
(941 mg, 62%) as an yellow solid. mp 208-210 °C. 1H NMR
(300 MHz, DMSO-d6): δ 6.33-6.36 (1H, m), 6.81 (1H, s), 6.94-
6.97 (1H, m), 7.25-7.27 (1H, m), 7.61 (1H, t, J ) 7.9 Hz), 8.32
(1H, d, J ) 7.9 Hz), 8.45 (1H, d, J ) 7.9 Hz), 12.00 (1H, brs).
MS (ESI+): m/z 257 (M+H)+. HR-MS calcd for C13H9N2O4
(M+H)+: 257.0562, found 257.0560.
9-Am in o-1,2,3,9b-tetr ah ydr o-5H-pyr r olo[2,1-a ]isoin dol-
5-on e (21). To a stirred solution of 19 (50 mg, 0.23 mmol) in
MeOH (10 mL) was added 10% Pd/C (25 mg), and the mixture
was stirred at room temperature for 2 h under an atmospheric
pressure of hydrogen. The catalyst was removed by filtration,
and the filtrate was evaporated to yield racemate 21 (48 mg,
99%) as a white solid. mp 149-152 °C. 1H NMR (300 MHz,
DMSO-d6): δ 0.80-0.93 (1H, m), 2.10-2.30 (2H, m), 2.43-
2.51 (1H, m), 3.18-3.24 (1H, m), 3.38-3.47 (1H, m), 4.50 (1H,
dd, J 1 ) 10.0 Hz, J 1 ) 5.5 Hz), 5.34 (2H, s), 6.72 (1H, d, J )
7.9 Hz), 6.76 (1H, d, J ) 7.9 Hz), 7.11 (1H, t, J ) 7.9 Hz). MS
(ESI+): m/z 189 (M+H)+. HR-MS calcd for C11H13N2O
(M+H)+: 189.1028, found 189.1025.
20
[R]D 87.6° (c 1.0, CHCl3).
(9b S)-9-Am in o-1,2,3,9b -t et r a h yd r o-5H -p yr r olo[2,1-a ]
isoin d ol-5-on e (21b). To a stirred solution of 22b (11.0 g, 32.6
mmol) in tetrahydrofuran (300 mL) was added lithium boro-
hydride (4.25 g, 195 mmol) at room temperature. Then MeOH
(16.0 mL, 391 mmol) in tetrahydrofuran (10 mL) was added
in a dropwise manner. The resultant mixture was stirred for
5 h at room temperature and poured into 3 N HCl. The
aqueous layer was neutralized with saturated aqueous NaH-
CO3 and extracted with CHCl3. The combined organic layer
was washed with brine, dried over MgSO4, and concentrated
in vacuo. The residue was purified by silica gel column
chromatography (2% MeOH in CHCl3) and recrystallized from
1
EtOAc to give 21a (2.1 g, 34%, 100% ee). mp 151-152 °C. H
NMR (300 MHz, DMSO-d6): δ 0.80-0.93 (1H, m), 2.10-2.30
(2H, m), 2.43-2.51 (1H, m), 3.18-3.24 (1H, m), 3.38-3.47 (1H,
m), 4.50 (1H, dd, J 1 ) 10.0 Hz, J 1 ) 5.5 Hz), 5.34 (2H, s), 6.72
(1H, d, J ) 7.9 Hz), 6.76 (1H, d, J ) 7.9 Hz), 7.11 (1H, t, J )
7.9 Hz). HR-MS calcd for C11H13N2O (M+H)+: 189.1028, found
189.1034. [R]D20-87.0° (c 1.0, CHCl3).
4-Nitr o-N-(5-oxo-2,3,5,9b-tetr a h yd r o-1H-p yr r olo[2,1-a ]-
isoin d ol-9-yl)ben za m id e (22). To a stirred solution of 21 (30
mg, 0.16 mmol) in CHCl3 (10 mL) were added triethylamine

4626 J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 26
Honma et al.
(1S,4R)-N-[(9bS)-5-Oxo-2,3,5,9b-tetr a h yd r o-1H-p yr r olo-
[2,1-a ]isoin d ol-9-yl]-4,7,7-t r im et h yl-3-oxo-2-oxa b icyclo-
[2,2,1]h ep ta n e-1-ca r boxa m id e. To a stirred solution of 21b
(54 mg, 0.29 mmol) in CH2Cl2 were added (1S)-(-)-camphanic
chloride (126 mg, 0.58 mmol), triethylamine (80 µL, 0.57
mmol), and 4-(dimethylamino)pyridine (3.5 mg, 0.029 mmol).
The resultant mixture was stirred for 5 h at room temperature,
poured into saturated aqueous NaHCO3, and extracted with
CHCl3. The combined organic layer was washed with brine,
dried over MgSO4, and concentrated in vacuo. The residue was
purified by silica gel column chromatography (hexane-EtOAc
(5:1)) to give (1S,4R)-N-[(9bS)-5-oxo-2,3,5,9b-tetrahydro-1H-
pyrrolo[2,1-a]isoindol-9-yl]-4,7,7-trimethyl-3-oxo-2-oxabicyclo-
[2,2,1]heptane-1-carboxamide (88 mg, 82%). mp 189 °C. 1H
NMR (300 MHz, CDCl3): δ 1.01 (3H, s), 1.17 (3H, s), 1.19 (3H,
s), 1.25-1.47 (1H, m), 1.73-1.82 (1H, m), 1.98-2.09 (2H, m),
2.22-2.33 (1H, m), 2.34-2.45 (2H, m), 2.58-2.69 (1H, m),
3.38-3.50 (1H, m), 3.66-3.78 (1H, m), 4.76 (1H, dd, J 1 ) 10.2
Hz, J 1 ) 5.9 Hz), 7.47 (1H, t, J ) 8.0 Hz), 7.62 (1H, t, J ) 8.0
Hz), 8.03 (1H, d, J ) 8.0 Hz), 8.08 (1H, brs). HR-MS calcd for
Refer en ces
(1) Pines, J . Cyclins and cyclin dependent kinases: take your
partners. Trends Biochem. Sci. 1993, 18, 195-197.
(2) Sherr, C. J . D-type cyclins. Trends Biochem. Sci. 1995, 20, 187-
190
(3) (a) Taya, Y. Cell cycle-dependent phosphorylation of the tumor
suppressor RB protein. Mol. Cells 1995, 5, 191-195. (b) Wein-
berg, R. A. The retinoblastoma protein and a cell cycle control.
Cell 1995, 81, 323-330. (c) Kato, J .; Matsushime, H.; Hiebert,
S. W.; Ewen, M. E.; Sherr, C. J . Direct binding of cyclin D to
the retinoblastoma gene product (pRB) and pRB phosphorylation
by the cyclin D-dependent kinase CDK4. Genes Dev. 1993, 7,
331-342. (d) Baldin, V.; Lukas, J .; Marcote, M. J .; Pagano, M.;
Draetta, G. Cyclin D1 is nuclear protein required for cell cycle
progression in G1. Genes Dev. 1993, 7, 812-821.
(4) Kamb, A. Cell-cycle regulators and cancer. Trends Genet. 1995,
11, 136-140.
(5) Akinaga, S.; Sugiyama, K.; Akiyama, T. UCN-01 (7-hydroxys-
taurosporine) and other indolocarbazole compounds:
generation of anticancer agents for the new century? Anti-Cancer
Drug Des. 2000, 15, 43-52.
a new
(6) (a) Sedlacek, H. H.; Czech, J .; Naik, R.; Kaur, G.; Worland, P.;
Losiewicz, M.; Parker, B.; Carlson, B.; Smith, A.; Senderowicz,
A.; Sausville, E. Flavopiridol (L86-8275, NSC-649890), a new
kinase inhibitor for tumor therapy. Int. J . Oncol. 1996, 9, 1143-
1168. (b) Senderowicz, A. M.; Headlee, D.; Stinson, S. F.; Lush,
R. M.; Kalil, N.; Villalba, L.; Hill, K.; Steinberg, S. M.; Figg, W.
D.; Tompkins, A.; Arbuck, S. G.; Sausville, E. A. Phase I trial of
continuous infusion flavopiridol, a novel cyclin-dependent kinase
inhibitor, in patients with refractory neoplasms. J . Clin. Oncol.
1998, 16, 2986-2999.
C
21H25N2O4 (M+H)+: 369.1814, found 369.1804. [R]D20 -92.4°
(c 1.0, CHCl3).
9-Am in o-5H -p yr r olo[2,1-a ]isoin d ol-5-on e (23). To a
stirred solution of 19 (100 mg, 0.47 mmol) in methyl alcohol
(15 mL) were added Fe (powder, 200 mg) and then 6 N HCl
(500 mL). The resultant mixture was stirred at room temper-
ature for 30 min, poured into saturated aqueous NaHCO3, and
the aqueous layer was extracted with EtOAc. The combined
organic layer was washed successively with water and brine,
dried over MgSO4, and concentrated in vacuo. The residue was
purified by silica gel column chromatography (hexane-EtOAc
(5:1)) to afford 23 (71 mg, 83%) as a red solid. mp 119-120
(7) Glay, N. S.; Wodicka, L.; Thunnissen, A.-M.; Norman, T. C.;
Kwon, S.; Espinoza, F. H.; Morgan, D. O.; Barnes, G.; Leclerc,
S.; Meijer, L.; Kim, S.-H.; Lockhart, D. J .; Schultz, P. G.
Exploiting chemical libraries, structure, and genomics in the
search of kinase inhibitors. Science 1998, 281, 533-538.
(8) Barvian, M.; Boscheli, D.; Cossrow, J .; Dobrusin, E.; Fattaey,
A.; Fritsch, A.; Fry, D.; Harvey, P.; Keller, P.; Garrett, M.; La,
F.; Leopold, W.; McNamara, D.; Quin, M.; Trumpp-Kallmeyer,
S.; Toogood, P.; Wu, Z.; Zhang, E. Prido[2,3-d]pyrimidin-7-one
inhibitors of cyclin-dependent kinases. J . Med. Chem. 2000, 43,
4606-4616.
(9) Nugiel, D. A.; Etzkorn, A.-M.; Vidwans, A.; Benfield, P. A.;
Boisclair, M.; Burton, C. R.; Cox, S.; Czerniak, P. M.; Doleniak,
D.; Seitz, S. P. Indenopyrazoles as novel cyclin dependent kinase
(CDK) inhibitors. J . Med. Chem. 2001, 44, 1334-1336.
(10) Fry, D. W.; Garret, M. D. Inhibitors of cyclin dependent kinases
as therapeutic agents for the treatment of cancer. Curr. Opin.
Oncol. Endocr. Metab. Invest. Drugs 2000, 2, 40-59.
(11) (a) Nicklaus, M. C.; Neamati, N.; Hong, H.; Mazumder, A.;
Sunder, S.; Chen, J .; Milne, G. W. A.; Pommier, Y. HIV-1
integrase pharmacophore: discovery of inhibitors through three-
dimensional database searching. J . Med. Chem. 1997, 40, 920-
929. (b) Babine, R. E.; Bleckman, T. M.; Kissinger, C. R.;
Showalter, R.; Pelletier, L. A.; Lewis, C.; Tucker, K.; Moomaw,
E.; Parge, H. E.; Villafranca, J . E. Design, synthesis and X-ray
crystallographic studies of novel FKBP-12 ligands. Bioorg. Med.
Chem. Lett. 1995, 5, 1719-1724. (c) Shuker, S. B.; Hajduk, P.
J .; Meadows, R. P.; Fesik, S. W. Discovering high-affinity ligands
for proteins: SAR by NMR. Science 1996, 274, 1531-1534. (d)
Manduskuie, T. P., J r.; McNamara, K. J .; Ru, Y.; Knabb, R. M.;
Stouten, F. W. Rational design and synthesis of novel, potent
bis-phenylamidine carboxylate factor Xa inhibitors. J . Med.
Chem. 1998, 41, 53-62. (e) Stewart, K. D.; Loren, S.; Frey, L.;
Otis, E.; Klinghofer, V.; Hulkower, K. I. Discovery of a new
cyclooxygenase-2 lead compound through 3-D database search-
ing and combinatorial chemistry. Bioorg. Med. Chem. Lett. 1998,
8, 529-534. (f) Boehm, H.-J .; Boehringer, M.; Bur, D.; Gmuend-
er, H.; Huber, W.; Klaus, W.; Kostrewa, D.; Kuehne, H.;
Luebbers, T.; Meunier-Keller, N.; Mueller, F. Novel inhibitors
of DNA gyrase: 3D structure based biased needle screening, hit
validation by biophysical methods, and 3D guided optimization.
A promising alternative to random screening. J . Med. Chem.
2000, 43, 2664-2674.
(12) Reviewed in (a) Desjarlais, R. L. Generation and use of three-
dimensional databases for drug discovery. In Practical Applica-
tion of Computer-aided Drug Design; Charifson, P. S., Ed.;
Marcel Dekker: New York, 1997; pp 73-104. (b) Good, A. C.;
Mason J . S. Three-dimensional structure database searches. In
Reviews in Computational Chemistry; Lipkowitz, K. B., Boyd,
D. B., Eds.; VCH Publishers: New York, 1996; pp 67-117.
(13) Reviewed in Murcko M. A. An introduction to de novo ligand
design. In Practical Application of Computer-aided Drug Design;
Charifson, P. S., Ed.; Marcel Dekker: New York, 1997; pp 304-
354.
1
°C. H NMR (300 MHz, CDCl3): δ 3.85 (2H, brs), 6.02 (1H, d,
J ) 3.1 Hz), 6.17 (1H, t, J ) 3.1 Hz), 6.78 (1H, d, J ) 8.0 Hz),
6.99-7.04 (2H, m), 7.16 (1H, d, J ) 8.0 Hz). MS (ESI+): m/z
185 (M+H)+. HR-MS calcd for C11H9N2O (M+H)+: 185.0715,
found 185.0714.
9-Am in o-3-(1-h yd r oxyvin yl)-5H-p yr r olo[2,1-a ]isoin d ol-
5-on e (24). To a stirred solution of 20 (200 mg, 0.78 mmol) in
MeOH-tetrahydrofuran (1:1, 20 mL) was added 10% Pd/C (25
mg), and the mixture was stirred at room temperature for 10
h under an atmospheric pressure of hydrogen. The catalyst
was removed by filtration, and the filtrate was evaporated.
The residue was purified by precipitation from EtOAc to yield
24 (208 mg, 100%) as a slightly green solid. mp 198-200 °C.
1H NMR (300 MHz, DMSO-d6): δ 5.92 (2H, brs), 6.29 (1H, d,
J ) 2.9 Hz), 6.62 (1H, s), 6.95 (1H, t, J ) 5.2 Hz), 7.10 (2H, d,
J ) 5.2 Hz), 7.14 (1H, d, J ) 2.9 Hz), 7.21 (1H, brs), 12.05
(1H, brs). MS (ESI+): m/z 227 (M+H)+. HR-MS calcd for
C
13H11N2O2 (M+H)+: 227.0821, found 227.0823.
Cyclin D-Cd k 4 Assa y. In vitro Cdk assays were per-
formed as previously described40 with some modifications. The
recombinant human cyclin D2-Cdk4 complex was expressed
in Sf9 cells with a baculovirus expression system and then
purified by HPLC. The purified cyclin D2-Cdk4 was incubated
with 50 µM ATP (0.5-1.5 µCi of [33P]-ATP (3000 Ci/mmol)),
100 µM or 400 µM G1 peptide, R buffer (20 mM Tris-HCl pH
7.4, 10 mM MgCl2, 4.5 mM 2-mercaptoethanol, 1 mM EGTA),
and diluted compounds at 30 °C for 45 min. The reactions were
terminated by addition of 350 mM H3PO4. Peptides were
trapped and washed using P81 paper filter 96-well plates
(Millipore) and 75 mM H3PO4. 33P incorporation was measured
with a Top Count (Packard).
Ack n ow led gm en t. It is our pleasure to acknowl-
edge the contributions of Miss Hiroko Oki (for Cdk4
inhibitory assay), Mr. Hirokazu Ohsawa (for mass
spectral analysis), Mrs. Chihiro Sato (for elemental
analysis), and Dr. Kenji Kamata (for X-ray analysis).
We thank Dr. Hiroshi Funabashi and Dr. Takehiro
Fukami for their useful suggestions. We are also grate-
ful to Ms. Kimberley A. Marcopul for her critical reading
of this manuscript.
(14) Kuntz, I. D.; Blaney, J . M.; Oatley, S. J .; Langridge, R.; Ferrin,
T. E. A geometric approach to macromolecule-ligand interactions.
J . Mol. Biol. 1982, 161, 269.

Structure-Based Generation of Cdk4 Inhibitors
J ournal of Medicinal Chemistry, 2001, Vol. 44, No. 26 4627
(15) UNITY, developed and distributed by Tripos Associates, St.
Louis, MO.
(16) Catalyst, developed by BioCAD and now distributed by Molec-
ular Simulations Inc., Burlington, MA.
(17) (a) Nishibata, Y.; Itai, A. Automatic creation of drug candidate
structures based on receptor structure. Starting point for
artificial lead generation. Tetrahedron 1991, 47, 8985-90. (b)
Nishibata, Y.; Itai, A. Confirmation of usefulness of a structure
construction program based on three-dimensional receptor
structure for rational lead generation. J . Med. Chem. 1993, 36,
2921-8.
(18) LeapFrog manual, SYBYL version 6.1, Tripos Associates, St.
Louis, MO.
(19) Pearlman D. A.; Murcko M. A. CONCERTS-dynamic connection
of fragments as an approach to de novo ligand design. J . Med.
Chem. 1996, 39, 1651-1663.
(20) Clark, D. E.; Frenkel, D.; Levy, S. A.; Li, J .; Murray, C. W.;
Robson, B.; Waszkowycz, B.; Westhead, D. R. PRO-LIGAND: an
approach to de novo molecular design. 1. Application to the
design of organic molecules. J . Comput.-Aided Mol. Des. 1995,
9, 13-32.
(29) de Azevedo, W. F., J r.; Mueller-Dieckmann, H.-J .; Schulze-
Gahmen, U.; Worland, P. J .; Sausville, E.; Kim, S.-H. Structural
basis for specificity and potency of a flavonoid inhibitor of human
CDK2, a cell cycle kinase. Proc. Natl. Acad. Sci. U.S.A. 1996,
93, 2735-40.
(30) Lawrie, A. M.; Noble, M. E. M.; Tunnah, P.; Brown, N. R.;
J ohnson, L. N.; Endicott, J . A. Protein kinase inhibition by
staurosporine revealed in details of the molecular interaction
with CDK2. Nat. Struct. Biol. 1997, 4, 796-801.
(31) De Azevedo, W. F., J r.; Leclerc, S.; Meijer, L.; Havlicek, L.;
Strnad, M.; Kim, S.-H. Inhibition of cyclin Dependent kinases
by purine analogs. Crystal structure of human cdk2 complexed
with roscovitine. Eur. J . Biochem. 1997, 243, 518-526.
(32) Russo, A. A.; J effrey, P. D.; Patten, A. K.; Massague, J .;
Pavletich, N. P. Crystal structure of the p27Kip1 cyclin depend-
ent-kinase inhibitor bound to the cyclin A-Cdk2 complex. Nature
1996, 382, 325-331.
(33) Meijer, L.; Thunnissen, A. M. W. H.; White, A. W.; Garnier, M.;
Nikolic, M.; Tsai, L. H.; Walter, J .; Cleverley, K. E.; Salinas, P.
C.; Wu, Y. Z.; Biernat, J .; Mandelkow, E. M.; Kim, S. H.; Pettit,
G. R. Inhibition of cyclin dependent kinases, GSK-3b and CK1
by hymenialdisine, a marine sponge constituent. Chem. Biol.
2000, 7, 51-63.
(21) (a) Boehm H. J . The computer program LUDI: a new method
for the de novo design of enzyme inhibitors. J . Comput.-Aided
Mol. Des. 1992, 6, 61-78. (b) Boehm H. J . LUDI: rule-based
automatic design of new substituents for enzyme inhibitor leads.
J . Comput.-Aided Mol. Des. 1992, 6, 593-606.
(22) Iwama, T.; Honma, T.; Ikeura, C. System for evaluation of
availability of essential structures generated by de novo design
programs (SEEDS). Unpublished results.
(34) Sequential buildup methods fall into the following two catego-
ries: atom-by-atom approaches and fragment-by-fragment ap-
proaches. In atom-by-atom approaches, ligands are constructed
atom by atom. Murcko M. A. An introduction to de novo ligand
design. In Practical Application of Computer-aided Drug Design;
Charifson, P. S., Ed.; Marcel Dekker: New York, 1997; pp 322-
325.
(23) De Bondt, H. L.; Rosenblatt, J .; J ancarik, J .; J ones, H.. D.;
Morgan, D. O.; Kim, S.-H. Crystal structure of cyclin dependent
kinase 2. Nature 1993, 363, 595-602.
(24) (a) J effrey, P. D.; Russo, A. A.; Polyak, K.; Gibbs, E.; Hurwitz,
J .; Massague, J .; Pavletich, N. P. Mechanism of CDK activation
revealed by the structure of a cyclin A-CDK2 complex. Nature
1995, 376, 313-20. (b) Russo, A. A.; J effrey, P. D.; Pavletich, N.
P. Structural basis of cyclin dependent kinase activation by
phosphorylation. Nat. Struct. Biol. 1996, 3, 696-700. (c) Also
see a review: Pavletich, N. P. Mechanism of cyclin dependent
kinase regulation: structures of Cdks, their cyclin activators,
and cip and ink4 inhibitors. J . Mol. Biol. 1999, 287, 821-828.
(25) (a) Russo, A. A.; Tong, L.; Lee, J .-O.; J effrey, P. D.; Pavletich,
N. P. Structural basis for inhibition of the cyclin dependent
kinase Cdk6 by the tumor suppressor p16INK4a. Nature 1998,
395, 237-243. (b) J effrey, P. D.; Tong, L.; Pavletich, N. P.
Structural basis of inhibition of CDK-cyclin complexes by INK4
inhibitors. Genes Dev. 2000, 14, 3115-3125. (c) Brotherton, D.
H.; Dhanaraj, V.; Wick, S.; Leonardo, B.; Domaille, P. J .;
Volyanik, E.; Xu, X.; Parisini, E.; Smith, B. O.; Archher, S. J .;
Serrano, M.; Brenner, S. L.; Blundell, T. L.; Laue, E. D. Crystal
structure of the complex of the cyclin D-dependent kinase Cdk6
bound to the cell-cycle inhibitor p19INK4d. Nature 1998, 395,
244-250.
(35) The available chemicals directory distributed by MDL Informa-
tion Systems, San Leandro, CA.
(36) QUANTA modeling environment, December 1998; Molecular
Simulations Inc.: San Diego, CA, 1998.
(37) Grigg, R.; Sridharan, V.; Stevenson, P.; Sukirthanlingam, S.;
Worakun, T. The synthesis of fused ring nitrogen heterocycles
via regiospecific intramolecular Heck reactions. Tetrahedron
1990, 46, 4003-4018.
(38) Ikuta, M.; Kamata, K.; Fukasawa, K.; Honma, T.; Machida, T.;
Hirai, H.; Suzuki-Takahashi, I.; Hayama, T.; Nishimura, S.
Crystallographic approach to identification of cyclin dependent
kinase 4 (CDK4) specific inhibitors by using CDK4 mimic CDK2
protein. J . Biol. Chem. 2001, 276, 27548-27554.
(39) We investigated the following kinases: PKA, PKC, PKBR (Akt-
1), CaMK II, p38R, ERK1, MEK1, Src, Lck, Flt-1, ZAP, EGFR,
FGFR1, and PDGFRâ. Honma, T.; Yoshizumi, T.; Hashimoto,
Hayashi, K.; Kawanishi, N.; Fukasawa, K.; Takaki, T.; Ikeura,
C.; Ikuta, M.; Suzuki-Takahashi, I.; Hayama, T.; Nishimura, S.;
Morishima, H. A novel approach for the development of selective
Cdk4 inhibitors: library design based on locations of Cdk4
specific amino acid residues. J . Med. Chem. 2001, 44, 4628-
4640.
(40) (a) Kitagawa, M.; Okabe, T.; Ogino, H.; Matsumoto, H.; Suzuki-
Takahashi, I.; Kokubo, T.; Higashi, H.; Saitoh, S.; Taya, Y.;
Yasuda, H.; Ohba, Y.; Nishimura, S.; Tanaka, N.; Okuyama, A.
Butyrolactone I, a selective inhibitor of cdk2 and cdc2 kinase.
Oncogene 1993, 8, 2425-2432. (b) Kitagawa, M.; Higashi, H.;
J ung, H.-K.; Suzuki-Takahashi, I.; Ikeda, M.; Tamai, K.; Kato,
J .; Segawa, K.; Yoshida, E.; Nishimura, S.; Taya, Y. The
consensus motif for phosphorylation by cyclin D1-Cdk4 is
different from that for phosphorylation by cyclin A/E-Cdk2.
EMBO J . 1996, 15, 7060-7069.
(26) (a)
http://www.nih.go.jp/mirror/Kinases/pkr/pk•catalytic/
pk•hanks•seq•align•long.html. (b) Hanks, S. K.; Quinn, A.
M.; Hunter, T. The protein kinase family: conserved features
and deduced phylogeny of the catalytic domains. Science 1988,
241, 42-52. (c) Hanks, S. K.; Quinn, A. M. Protein kinase
catalytic domain sequence database: identification of conserved
features of primary structure and classification of family mem-
bers. Methods Enzymol. 1991, 200, 38-62. (d) Hanks, S. K.;
Hunter, T. Protein kinases. 6. the eukaryotic protein kinase
superfamily: kinase (catalytic) domain structure and classifica-
tion. FASEB J . 1995, 9, 576-96.
(27) BIOCES[E], developed by NEC Co., Tokyo, J apan.
(28) Brooks, B. R.; Bruccoleri, R. E.; Olafson, B. D.; States, D. J .;
Swaminathan, S.; Karplus, M. CHARMM: a program for mac-
romolecular energy, minimization, and dynamic calculation. J .
Comput. Chem. 1983, 4, 187-217.
J M0103256