Cyclin-dependent kinase inhibition by new C-2 alkynylated purine derivatives and molecular structure of a CDK2-inhibitor complex

Article abstract of DOI:10.1021/jm9911130

A new series of 2,6,9-trisubstituted purines, characterized by the presence of a common alkynyl substituent at C-2 and a range of different anilino/benzylamino groups at C-6, were synthesized. These compounds were evaluated for their capacity to inhibit c

Full text of DOI:10.1021/jm9911130

1282
J . Med. Chem. 2000, 43, 1282-1292
Cyclin -Dep en d en t Kin a se In h ibition by New C-2 Alk yn yla ted P u r in e
Der iva tives a n d Molecu la r Str u ctu r e of a CDK2-In h ibitor Com p lex
Michel Legraverend,*,§ Paul Tunnah,‡ Martin Noble,‡ Pierre Ducrot,§ Odile Ludwig,§ David S. Grierson,§
Maryse Leost,† Laurent Meijer,† and J ane Endicott*,‡
Section de Recherche, Institut Curie, UMR 176 CNRS, Baˆt. 110-112, Centre Universitaire, 91405 Orsay Cedex, France, Station
Biologique, CNRS, B.P. 74, 29682 Roscoff Cedex, France, and Department of Biochemistry, Laboratory of Molecular Biophysics
and Oxford Centre for Molecular Sciences, Rex Richards Building, South Parks Road, Oxford OX1 3QU, U.K.
Received J une 30, 1999
A new series of 2,6,9-trisubstituted purines, characterized by the presence of a common alkynyl
substituent at C-2 and a range of different anilino/benzylamino groups at C-6, were synthesized.
These compounds were evaluated for their capacity to inhibit cyclin-dependent kinase activity
(CDK1-cyclin B) in vitro. Compounds 4e (N-6-p-Cl-benzylamino derivative) and 5e (N-6-m-
Cl-anilino derivative) exhibited the strongest inhibitory activity with an IC50 of 60 nM. The
structure of compound 4b (N-6-p-methoxybenzylamino derivative) in complex with human
CDK2 was determined by X-ray crystallography, revealing the molecular basis of inhibition
by this molecule. Subsequent molecular modeling studies allowed us to rationalize the SAR
observed for these compounds.
In tr od u ction
include the potent purine derivatives bearing a diami-
nocyclohexane substituent at C-2, described by the
Novartis group,10 and purvalanols A and B, identified
by Schultz et al. by combinatorial library evaluation.11,12
Flavopiridol,13 indirubin 3′-monoxime14 (the active in-
gredient in a traditional Chinese herbal medicine
historically used to treat chronic myelocytic leukemia),
and kenpaullone15 represent different families of mol-
ecules (Figure 1) which also interact with the ATP
binding site in CDK1/CDK2.
Cyclin-dependent kinases (CDKs), particularly CDK1,
CDK2, CDK4, and CDK6, play important roles in cell
cycle regulation. Their sequential activation ensures the
correct timing and ordering of events required for cell
cycle progression.1 Over the past decade, intense efforts,
using biochemical and molecular biology techniques,
have helped to define the mechanism of action of the
CDKs, their regulation, and the identity of certain of
their protein substrates.2 To advance such studies, small
molecule cyclin-dependent kinase inhibitors (CKIs)
would be valuable biochemical tools.
Similarly, taking into consideration that CDK mis-
regulation is implied in a number of disease states, and
in particular cancer, enormous potential exists for small
molecule CKIs as chemotherapeutic agents.3 Indeed,
certain proteins that regulate CDKs (e.g. the CKI p16
and cyclin D) or are CDK substrates (e.g. the retino-
blastoma protein, pRb) are incorrectly or inappropriately
expressed in a number of proliferative disorders. Both
pRb and certain CKIs have been characterized as tumor
suppressors.4,5
The development of selective inhibitors for the ATP
binding site is, to a large extent, rendered difficult by
the high level of sequence homology that exists for this
region among protein kinases.6,7 However, the screening
of compound collections has revealed that small mol-
ecules are indeed able to selectively inhibit the CDK
family. For example, the 2,6,9-trisubstituted purines
olomoucine and roscovitine selectively inhibit CDK1-
cyclin B, CDK2-cyclin E/A, and CDK5-p35.8,9
In our laboratory, a new series of potent and selective
trisubstituted purine-based CKIs is being developed.16
These systems are characterized by the presence of a
hydroxyalkyl-substituted acetylene substituent at C-2
of the purine ring. For example, compound 4a is a more
potent inhibitor of CDK1-cyclin B (IC50 ) 200 nM) than
the lead molecule (R,S)-roscovitine (IC50 ) 650 nM).
Although no structures have been obtained to date
for CDK1, X-ray crystal structures of CDK2 complexes
with ATP,17 olomoucine,18 roscovitine,19 staurosporine,20
flavopiridol,13 purvalanol B,11 and indirubin14 have been
reported which demonstrate how the ATP binding site
can accommodate structurally diverse inhibitor types.
X-ray structures for CDK2-cyclin A-ATP,21 CDK2
alone,17 and CDK2-cyclin A plus p27KIP122 (natural
CKI) have also been determined. These structures have
shown how movements of the T-loop, PSTAIRE region,
and N- and C-terminal lobes induced by cyclin binding
and CDK2 Thr160 phosphorylation result in a catalyti-
cally active complex.21 They also reveal limited differ-
ences in the vicinity of the ATP adenine binding site
between different activation states of CDK2. This sug-
gests that information derived from monomeric CDK2-
inhibitor structures can guide the design of inhibitors
against the active binary complex.
Building upon this discovery, more optimized purine
CKIs have been described by several groups. These
* To whom correspondence should be addressed. M. Legraverend:
In this context, this report presents further advances
in the development of the novel C-2 alkynyl-substituted
purine family of CKIs. A series of molecules bearing a
range of different anilino/benzylamino groups at C-6 has
tel, 1-6986-3086; e-mail, Michel.Legraverend@curie.u-psud.fr.
§ UMR 176 CNRS.
† CNRS.
‡ Laboratory of Molecular Biophysics and Oxford Centre for Molec-
ular Sciences.
10.1021/jm9911130 CCC: $19.00 © 2000 American Chemical Society
Published on Web 03/11/2000

CDK Inhibition
J ournal of Medicinal Chemistry, 2000, Vol. 43, No. 7 1283
Ta ble 1. IC50 Values against CDK1-Cyclin B for Members of
Series 4 and 5
compda
R
IC50 (nM)
4a
H
200
230
180
430
60
200
300
150
4b (OL567)
p-OCH3
4c
4d
4e
4f
4g
4h
p-OCH2CH3
m,p-di-Cl
p-Cl
m,p-OCH2O
m,p-di-OCH3
p-N(CH3)2
5a
5b
5c
5d
5e
H
400
200
200
200
60
p-OCH3
p-OCH2CH3
m,p-di-Cl
m-Cl
olomoucine
(R,S)-roscovitine
7000
650
a
m ) meta, p ) para substituents of the phenyl group.
anilines or benzylamines (1.5 equiv) in ethanol contain-
ing N(Et)3 (4 equiv), under inert atmosphere at 50 °C
for 3-4 h (Scheme 1). In this way the C-6 substituted
purine derivatives 2 and 3 were obtained in high yields
(60-95%). Conversion of these 2-iodopurine intermedi-
ates to the target 2-ethynylpurines 4 and 5 was achieved
by reaction with 3-methylpent-1-yn-3-ol under pal-
ladium(0)-CuI catalyzed cross-coupling conditions de-
veloped by Sonogashira.26
F igu r e 1. Structures of CKIs.
thus been synthesized and evaluated for their capacity
to inhibit CDK1-cyclin B in vitro. All compounds
prepared proved to be more active than roscovitine, with
maximum activity observed for 4e and 5e (IC50 ) 60
nM). The structure of one of these compounds, OL567
(4b) in complex with CDK2, was determined by X-ray
crystallography, revealing the molecular basis of inhibi-
tion by this molecule. Subsequent molecular modeling,
employing the crystal structure data, has further per-
mitted us to rationalize the structure-activity relation-
ship (SAR) observed for these compounds.
Resu lts a n d Discu ssion
Compounds 4 and 5 were tested for their ability to
inhibit in vitro the CDK1-cyclin B complex from
starfish oocytes as described.9,16,25 In this assay the
transfer of γ-32P-labeled phosphate from ATP to histone
H1 in the absence and presence of inhibitor is measured
and expressed as a percent of kinase activity without
inhibitor. The IC50 is determined from dose-response
curves. The IC50 value for each of the compounds
determined under these assay conditions is shown in
Table 1.
Ch em istr y
To access the C-2 alkynylated purine compounds
4b-h and 5a -e, 6-chloro-2-iodo-9-isopropyl-(9H)-purine
(1), prepared as previously described,16,23-25 was reacted
with the appropriate meta- and/or para-substituted
Trisubstituted purine-based inhibitors of CDK1-
cyclin B, and in particular those bearing an amino
Sch em e 1a
a
(i) Aniline or benzylamine (1.5 equiv), N(Et)3 (4 equiv), ethanol, 50 °C, 3-4 h; (ii) (R,S)-3-methyl-1-pentyn-3-ol (1.5 equiv), CuI (0.05
equiv), dichlorobis(triphenylphosphine)palladium(II) (0.01 equiv), n-butylamine (1.5 equiv), degassed DMF (25 mL), 80 °C.

1284 J ournal of Medicinal Chemistry, 2000, Vol. 43, No. 7
Legraverend et al.
alcohol substituent at C-2, generally, also inhibit CDK2
and CDK5 to various extents.8,9,11 For this reason
compounds 4 and 5 were evaluated for their capacity
to inhibit CDK5-p25 (data not shown). With the excep-
tion of compound 4f, which is more active in the CDK5
assay (IC50 ) 130 nM), all compounds in both series
displayed reduced capacity to inhibit CDK5-p25. This
is especially true for the monochloro analogues 4e and
5e, which are respectively 4 and 8.3 times less active
against CDK5-p25. These results are not included in
our SAR study as X-ray crystal data for CDK5 is
currently unavailable. Note, however, that two recent
publications describe computer-generated models of
CDK5 either alone27 or complexed to a neuronal CDK5
activator (Nck5a).28
Taking roscovitine as reference (IC50 ) 650 nM) it was
found that the purine derivative 4a , which is unsubsti-
tuted in the phenyl ring of the 6-benzylamine group, is
3 times more active.16 Indeed, all the new C-2 alkyny-
lated purine compounds 4 containing a meta and/or para
substituent on the phenyl ring are more active than
roscovitine. These results underscore our preliminary
finding that there is a gain in activity upon exchange
of the 2-amino alcohol chain in roscovitine by a prop-
argyl alcohol motif.16
In the present study, the first modification of 4a to
be examined was the introduction of a p-OMe substitu-
ent.29,30 Interestingly, a slight loss in activity was
observed for the derived compound 4b (OL567). Intro-
duction of a second OMe group to give the m,p-
dimethoxyphenyl analogue 4g resulted in a further loss
in activity, whereas the methylenedioxy analogue 4f
remained equipotent with 4a . These results contrast
with the small gain in activity for the corresponding
p-OEt-substituted compound 4c.
The largest fluctuation in activity was observed for
the chloro-substituted analogues 4d and 4e. Compound
4d with chlorine atoms at both the meta and para
positions of the phenyl ring proved to be the least active
molecule in the series, whereas compound 4e with a
single p-chloro substituent displayed powerful capacity
to inhibit substrate phosphorylation by CDK1-cyclin
B (IC50 ) 60 nM). This latter finding suggested that the
presence of a polar substituent at the para position of
the phenyl ring of the 6-benzylamino should be favor-
able. The good level of activity observed for the p-NMe2-
substituted analogue 4h is consistent with this idea.
importance as the specificity and potency of olomoucine,
roscovitine, and OL567 are in part the consequence of
interaction of their 6-aminobenzyl substituent with a
region of the kinase that is highly conserved in the CDK
family and which does not interact with its natural
substrate, ATP.19,32 To obtain more information on this
point the structure of the CDK2-4b (OL567) complex
was determined by X-ray diffraction.
Cr yst a llogr a p h ic An a lysis: In t er a ct ion s b e-
tw een CDK2 a n d OL567. The structure of CDK2 is a
minimal protein kinase catalytic core consisting of an
N-terminal domain formed principally from â-sheet and
a C-terminal domain that is primarily R-helical.17,33
OL567 occupies the binding site in the cleft between the
CDK2 N- and C-terminal domains. Figure 2 shows the
hydrogen-bonding pattern and apolar contacts of OL567
bound to CDK2 as calculated by the program Ligplot.34
The apolar contacts correspond to contacts shorter than
3.9 Å between non-hydrogen-bonding atom pairs.
The binding of OL567 to CDK2 closely resembles that
of olomoucine,18 roscovitine,19 and purvalanol11 and
differs from that reported for ATP bound to CDK2.17
OL567 N7 accepts a hydrogen bond from the peptide
nitrogen of Leu83, and N6 of OL567 donates a hydrogen
bond to the backbone oxygen of Leu83. The ATP adenine
ring hydrogen bonds to the CDK2 backbone in this
region through interactions between N6 and the back-
bone oxygen of Glu81 and between N1 and the backbone
nitrogen of Leu83. The OL567 adenine ring is sand-
wiched between the side chains of Ile10 and Leu134 and
is not coplanar with that of the ATP adenine ring in
the CDK2-ATP structure. Minor variations in ring
position between OL567, olomoucine, roscovitine, and
purvalanol are observed.11,13,35,36
Although residues 80-84 constitute part of the hinge
between the CDK2 N- and C-terminal domains, the
interactions do not result in a significant change in
relative domain orientation.
OL567, like roscovitine, has an isopropyl group on N9
that forms hydrophobic contacts with the side chain of
Phe80. As commented upon in more detail further on,
the 3-methylpent-1-yn-3-ol group substituted at C-2
binds in the pocket occupied by the ATP ribose group
in the phosphorylated CDK2-cyclin A binary complex.37
Binding appears to favor the S isomer which places the
methyl group into a small hydrophobic patch formed
mainly by Val18 and backbone atoms of the glycine loop
at residues 11 and 12. However, interpretation of the
electron density map for this group is not unambiguous.
Val18 shows slight movement toward the inhibitor to
maximize the hydrophobic contact. Bound as the S
isomer, the hydroxyl group makes no direct interactions
with CDK2 but can hydrogen bond to a bound water
molecule (Figure 2).
The 4-methoxybenzylamino group on N6 is surface
exposed and binds outside the active side cleft contact-
ing residues in the hinge region and N-terminal lobe
(Figure 2). The phenyl ring is surrounded by the side
chains of Phe82 and Ile10, the side chain amino group
of Lys89, and the CDK2 backbone at His84. Olomoucine,
roscovitine, purvalanol B, and OL567 all have markedly
different ring orientations, that of OL567 being most
displaced from that of purvalanol B. Lys89 adopts a
different conformation to that seen in the olomoucine
The preparation of compounds 5a -e was undertaken
on the basis of independent reports by the Novartis
group, Schultz et al., and ourselves10,11,25,31 which
showed that trisubstituted purines possessing a sub-
stituted arylamine at C-6 were active. It was interesting
to find therefore that, relative to the parent anilino
compound 5a , good levels of activity were observed for
compounds 5b-e bearing meta-para substituents and
that analogue 5e with a m-chloro group is equipotent
with 4e.
To summarize the data, the best results were obtained
with m- or p-chloro substituents on the aminophenyl
(anilino) and benzylamino groups, respectively. These
results confirm that the capacity of the C-2 alkynyl-
substituted purines to inhibit CDK1-cyclin B can be
modulated through variation of the C-6 position aro-
matic ring substituents. This point is of considerable

CDK Inhibition
J ournal of Medicinal Chemistry, 2000, Vol. 43, No. 7 1285
F igu r e 2. Schematic representation of the contacts between OL567 and CDK2 produced using Ligplot. Hydrogen-bonding residues
are drawn in full, with the length of the hydrogen bond annotated on a dotted line. Residues making non-hydrogen-bonding
contacts at a distance of less than 3.9 Å are also drawn, with a ‘fan’ representation indicating the atom of CDK2 contacted.
structure,18 being oriented toward Asp86. To avoid a
steric clash, the 4-methoxy group displaces the side
chain of Glu8. The combination of this movement and
also that of Val 18 results in subtle differences in the
positions of the residues throughout the glycine-rich
loop.
view of the very high degree of sequence homology
between CDK1 and CDK2 for the amino acid residues
that comprise the ATP binding pocket (Table 2). Indeed,
two important residues are modified in the catalytic site
of the two kinases (CDK2: His84, Gln85 f Ser84,
Met85: CDK1), which both have their respective side
chains outside the binding pocket and are not directly
involved in ATP binding.
Compounds 4b-h and 5a -e were constructed in
Sybyl 6.5.1 (Tripos software) using the structure for
OL567 (4b) extracted from the crystallographic data.
Conformations were generated for the phenyl moiety by
systematic rotations (60° step) around all bonds at-
tached to the aromatic ring; then each conformer was
docked separately. For the protein, all hydrogens and
lone pairs were added, and Kollman charges were
included such that the amine and carboxylic acid
functions were all ionized (NH3+, CO2-; pH 7). In
addition, hydrogens were added to the ligand, and
Gasteiger-Hu¨ ckel charges were calculated within
SYBYL. Docking calculations, were then carried out
after residues within 15 Å from the ligand (OL567) were
extracted. The molecules were docked using the DOCK
module within Sybyl, maintaining the binding site rigid
Molecu la r Mod elin g. OL567 is intermediate in its
capacity to inhibit CDK1-cyclin B relative to the
monochloro compound 4e and compound 4d which
possesses a m,p-dichlorobenzylamino substituent at C-6.
Several factors could be acting individually or col-
lectively to give rise to this range of activities, including
steric interactions due to the presence of functionality
on the phenyl ring and electronic effects. Steric interac-
tions can lead to a local repositioning of the benzyl side
chain in its binding zone (rotation about C-C/C-N
bond) or to a repositioning of the entire molecule in the
ATP binding pocket. To better understand the influence
of the C-6 substituent on inhibition within this series
of purine derivatives, and to study the binding of the
corresponding anilino compounds 5 in the ATP pocket
of CDK1, molecular models were developed using the
crystal data for the OL567-CDK2 complex as the
starting point. This approach was considered valid in

1286 J ournal of Medicinal Chemistry, 2000, Vol. 43, No. 7
Legraverend et al.
F igu r e 3. Superimposition of N6-benzylaminopurine series 4b-h (yellow) and N6-phenylaminopurine series 5b-e (magenta)
as modeled in the ATP binding site of human CDK2. The enzyme is colored using a lipophilicity range from brown (highest
lipophilic area of the molecule) to blue (highest hydrophilic area); intermediate lipophilicity is in green.
during minimizations. After minimization was com-
pleted, the best conformer hits for each compound were
reminimized within the binding site without any rigid
body. Figure 3 shows the superimposition of the mini-
mum energy structures for the two series docked in the
ATP binding site.
In Figure 3, one sees that, relative to OL567 (4b ),
there is essentially no change in the position of the
purine ring and the C-2 acetylene side chain in com-
pounds 4c-h . A complete superposition of these ele-
ments in compounds 5 was also observed, indicating
that modification of the C-6 substituent does not pro-
voke any significant displacement of the molecule in the
ATP site. This is consistent with the relatively small
variations in activity that are observed.
As might be expected, the aromatic rings of the C-6
benzylamino and anilino groups in analogues 4 and 5
occupy different regions of space in the enzyme. Con-
sequently, there is a distinct set of interactions for the
functionalized aryl rings in each series with the amino
acid residues in their vicinity.
In Figure 4 a more focused view of the positioning
and environment of the aromatic ring of the N-benzyl
compounds 4b-h in the ATP pocket is given. In
particular, one sees that the pocket is composed of two
hydrophobic residues (Ile10, Phe82), which interact and
limit the movement of the aromatic ring, as well as three
important polar residues (His84, Lys20, Glu8).
of the aromatic ring is globally more crowded by the
presence of Phe82, a larger volume is available in the
immediate vicinity of the meta carbon.
As this advantage is lost for compounds 4f and 4g
with more bulky and extended alkoxy substituents, they
prefer to adopt the meta′/para conformation (∆E ) 6.42
kcal). Oriented in this way, the oxygen substituents
experience the more hydrophilic environment created
by residues Lys89 and Glu8 and the backbone amide
carbonyl oxygens of Ile10 and Lys9 located toward the
exterior of the kinase. Steric crowding remains, how-
ever, a problem in this orientation. This is particularly
true for compound 4g, in which the m′-OMe group
points toward the para substituent, rather than away
from it, in an effort to avoid Lys89 and the acetylenic
side chain. In the methylenedioxy analogue 4f this steric
interaction is diminished.
Although no ortho-substituted compounds were evalu-
ated in this study, the model predicts that there is no
place to accommodate functionality in the ortho confor-
mation due to the presence of the Phe82 residue.
Equally severe problems would be encountered for the
ortho′ conformer due to unfavorable interactions with
both Lys89 (at 2 Å) and the purine ring of the inhibitor
itself (3.4 Å from the ethyl chain of the R isomer, 4.8 Å
from the OH of the S isomer, and 4.5 Å from the triple
bond).
In contrast to the dichloro analogue 4d , compound 4e
with one p-chloro substituent proved to be the most
active inhibitor of CDK1-cyclin B. From comparison of
this analogue with the parent unsubstituted compound
4a and the olomoucine-CDK2 structure, it may be
expected that the presence of the electron-withdrawing
chloro substituent in 4e might be detrimental to a
CH-π type interaction38 with the side chain of Ile10.
However, it is difficult to appreciate the influence of the
chloro substituent on this already inherently weak
Considering the different analogues individually, it
is striking that the m,p-dichloro analogue 4d , which
displays the weakest activity, is oriented in a manner
opposite to the two other more bulky meta,para-disub-
stituted analogues 4f and 4g. Indeed, comparison of the
relative energies of the meta/para orientation of 4d with
its alternative meta′/para conformer indicates that the
latter is 3 kcal/mol less stable. This energy difference
results from the fact that, even though the “meta” side

CDK Inhibition
J ournal of Medicinal Chemistry, 2000, Vol. 43, No. 7 1287
F igu r e 4. Binding model of 4b-h in the CDK2 active site. See legend of Figure 3 for colors.
bulky groups are present on the phenyl ring, it is
interaction, as there is very little difference in the
disposition of these entities in the two crystal structures.
In the olomoucine-CDK2 structure, interaction of the
phenyl ring of the 6-benzylamine substituent with the
δ-amino group of Lys89 was also considered important.18
In our structure of OL567 complexed to CDK2, the
Lys89 side chain adopts a different orientation, and this
interaction was consequently not present in our model.
Overall, it is probable that any weakening of hydropho-
bic interactions of the aryl ring in 4e resulting from
p-chloro substitution can be more than compensated for
by formation of a hydrogen bond between the chlorine
atom and the protonated side chain carboxylic acid func-
tion of Glu8. However, as this acidic residue is present
in its carboxylate form in the model enzyme, this inter-
action could not be observed in the docking experiment.
Returning to the m,p-dichloro analogue 4d , relative
to 4a and 4e, a slight rotation of the phenyl ring is
induced to relieve steric interactions involving the
m-chloro atom and Phe82. This phenomenon, combined
with the accentuated decrease in electron density in the
aromatic ring, may well lead to a weakening of the
CH-π interaction with Ile10 located directly below it
and thereby to the observed sharp drop in activity.
The increased activity of the p-methoxy-substituted
compound 4b (OL567) and the corresponding p-OEt
analogue 4c compared to the disubstituted compound
4g is most probably a consequence of their reduced
steric volume. However, as they remain essentially
equipotent with the parent compound 4a , it is clear that
there is no net advantage in effecting p-O-alkoxy
substitution of the aromatic ring. One can argue that
for these molecules there would be an enhanced positive
interaction with Ile10, but it is apparent, even from the
X-ray data for OL567, that a nonnegligible steric
interaction exists between these groups and Glu8.
Taking into consideration this subtle balance between
the stabilizing hydrophobic π type interaction with Ile10
and the destabilizing steric interaction evoked when
pertinent to note the good level of activity observed for
the N-dimethyl compound 4h . Due to the planar char-
acter of the nitrogen atom in this analogue, transfer of
electron density to the ring may strengthen the Phe-
Ile10 contact. In addition, the orientation of the di-
methyl substituent relative to Glu8 was found to be
different to that of 4f and 4g and may also be at the
origin of the increased inhibition of CDK1-cyclin B by
this molecule.
For the series of 6-anilino analogues 5 the aromatic
ring has a different orientation such that it is deeper
in the cleft and rotated by about 45° with respect to the
planar nitrogen (Figure 5). As a result, the Ile10 side
chain moved slightly so as to preserve CH-π interac-
tions.
The presence of a substituent on the aromatic ring
appears to be important for these 6-anilino analogues,
as compounds 5b-e are all more active than 5a . In
particular, a hydrogen bond between the alkoxide
oxygen and the side chain amino group of Lys89 (NH-O
) 2.70 Å) is observed for the p-OMe-substituted com-
pound 5b and the p-OEt-substituted compound 5c. Not
neglecting that the para position in compounds 5 is
surface exposed, advantage can be taken of the contact
with Lys89 in the development of new analogues of our
2-alkynylated purines.
The meta′ position is in close contact with the inner
surface of the binding pocket (Asp86 and Lys89).
However, the presence of two chlorine atoms (5d )
decreased the activity, due to unfavorable interactions
between the p-chlorine atom and the backbone carbonyl
group of His84. Furthermore, due to the orientation of
the phenyl ring toward the more hydrophilic outside of
the cleft, unfavorable interactions are created by the
presence of two chlorine atoms which increase its
hydrophobic character.
In line with the activities observed for the Novartis
compound10 as well as purvalanols A and B,12 m-chloro

1288 J ournal of Medicinal Chemistry, 2000, Vol. 43, No. 7
Legraverend et al.
F igu r e 5. Binding model of 5b-e in the CDK2 active site. See legend of Figure 3 for colors.
substitution as in 5e resulted in a very pronounced
increase in activity. From the X-ray structure of purv-
alanol B-CDK211 it was deduced that an H-bond
between the 3-Cl and Asp86 is important. However, the
meta′ orientation was privileged only to the extent of
2:1.11 In our model, this tendency is inversed, the
conformer with the chloro substituent in the meta
orientation being more stable by 1.5 kcal. This meta′/
meta inversion is a consequence of the protonation state
of the amine/carboxylic acid functionalities in the pro-
tein model (CO2-, NH3+; pH 7) and also the result of a
different orientation in the Asp86 side chain in our
crystal structure compared to the purvalanol B-CDK2
X-ray crystal structure. Conversely, in its protonated
state, the carboxylate of Asp86 should be engaged in a
weak hydrogen bond with the m′-chlorine. In any event,
irrespective of the preferred conformation for the 3-chlo-
ro substituent, it is clear that there is sufficient room
available in the enzyme for simultaneous substitution
of meta and meta′ positions. This suggestion from our
model is currently being explored, but already Schultz
et al. have recently shown that a meta,meta′-disubsti-
tuted analogue of purvalanol (m-Cl,m′-NH2, 97) is
the entire series of analogues 4 and 5 in the ATP pocket
of CDK2. Given the high level of sequence homology
between CDK1 and CDK2 (Table 2), the derived models
were used to interpret the results obtained in our
CDK1-cyclin B activity assay.
For the further design of new purine-based CDK1/2
inhibitors, the results suggest that small polar substit-
uents can be introduced at the para position of series 4
and the meta or para position of 5, but that substitution
at the ortho position of the aromatic ring in both series
should be avoided. This study also indicates that only
small substituents will be tolerated at the phenyl meta
and even more at the meta′ positions in the anilino
series. However, any meta′ substituent able to make
strong favorable electrostatic or hydrogen-bonding in-
teractions with Lys89 would be expected to result in
increased activity.
From the modeling study, it would also appear that
key interactions may be conserved if the aromatic ring
in compounds 4 and 5 is replaced by certain heterocyclic
systems.
Confirmation that the acetylene moiety in 4 and 5
effectively occupies the ribose binding pocket of ATP was
another important aspect of our structural study. In-
deed, it is remarkable that the 2-nitrogen which is
characteristic of all other purine-based CKIs can be
replaced by a sp-hybridized carbon atom. Pertinent in
this repect is the observation that partial or total
reduction of the acetylenic moiety leads to compounds
with greatly attenuated activities (IC50 > 1000 nM).16
Finally, comparison of the activities of purvalanol A and
the Novartis compounds with 5e, as well as OL567 with
the corresponding purvalanol derivative with a p-OMe
group (compound 75 in ref 12; IC50 ) 230 nM), shows
that our acetylene-based molecules are of comparable
activity. This observation offers a number of interesting
possibilities for the construction of new molecules modi-
fied both at the C-6 position and in the propargyl side
chain, which bind strongly to CDK1/2. In particular, this
equipotent (IC50 ) 33 nM) with purvalanol A (IC50
)
35 nM) itself.12
No ortho/ortho′-substituted products were evaluated,
but the presence of functionality at these positions
would again generate very unfavorable steric interac-
tions with both neighboring residues (ortho with Phe82,
ortho′ with Asp86 and Lys89) and the adenine ring.
Con clu sion
The X-ray crystal structure of the OL567-CDK2
complex gives a direct view of the positioning of OL567
in the ATP binding site of the kinase and thereby
provided the opportunity to analyze the interactions
that exist between this inhibitor and the protein. This
information further provided a valid starting point for
a broader molecular modeling study of the binding of

CDK Inhibition
J ournal of Medicinal Chemistry, 2000, Vol. 43, No. 7 1289
Ta ble 2. Selected Amino Acids in CDK2 That Interact with
Inhibitors* and Corresponding Amino Acids§ in CDK1, CDK4,
CDK5, and CDK6
by column chromatography which provided pure material after
crystallization from heptane.
(2-Iod o-9-isop r op yl-9H-p u r in -6-yl)(ben zyl)a m in e (2a )
has already been described.25
CDK2a
CDK1b
CDK5c
CDK4a
CDK6a
(2-Iod o-9-isop r op yl-9H-p u r in -6-yl)(4-m eth oxyben zyl)-
a m in e (2b). This compound was obtained in 75% yield with
Glu8
Glu
Ileu
Glu
Gly
Thr
Val
Ala
Lys
Phe
Glu
Phe
Leu
Ser
Met
Asp
Lys
Gln
Asn
Leu
Ala
Glu
Ileu
Glu
Gly
Thr
Val
Ala
Lys
Phe
Glu
Phe
Cys
Asp
Gln
Asp
Lys
Gln
Asn
Leu
Ala
Ala d
Ileu
Va l
Gly
Ala
Val
Ala
Lys
Phe
Glu
His
Va l
Asp
Gln
Asp
Th r
Glu
Asn
Leu
Ala
Ala
Ileu
Glu
Gly
Ala
Val
Ala
Lys
Phe
Glu
His
Va l
Asp
Gln
Asp
Th r
Gln
Asn
Leu
Ala
Asp
Ileu10
Glu12
Gly13
Thr14
Val18
Ala31
Lys33
Phe80
Glu81
Phe82
Leu83
His84
Gln85
Asp86
Lys89
Gln131
Asn132
Leu134
Ala144
Asp145
1
mp 188-190 °C. H NMR δ: 8.76 (s, 1H), 4.83 (m, 1H), 1.55
(d, 6H, J ) 6.7). Anal. (C16H18N5O1I1) C, H, N.
(2-Iod o-9-isop r op yl-9H -p u r in -6-yl)(4-et h oxyb en zyl)-
a m in e (2c). This compound was obtained in 85% yield with
mp 174-176 °C. 1H NMR δ: 7.69 (s, 1H, H-8), 7.28 (d, 2H,
Phe, J ) 8.6), 6.85 (d, 2H, Phe, J ) 8.6), 6.31 (br s, 1H, NH),
4.80 (m, 1H, CH(CH3)2), 4.70 (m, 2H, CH2Phe), 4.00 (q, 2H,
OCH2CH3, J ) 7), 1.55 (d, 6H, CH(CH3)2, J ) 6.8), 1.4 (t, 3H,
CH3CH2O, J ) 6.9). Anal. (C17H20N5O1I1) C, H, N.
(2-Iod o-9-isop r op yl-9H-p u r in -6-yl)(3,4-d ich lor oben zyl)-
a m in e (2d ). A crude solid was obtained in 97% yield with mp
1
184-186 °C. H NMR δ: 7.73 (s, 1H, H-8), 7.47-7.20 (m, 3H,
Phe), 6.42 (br s, 1H, NH), 4.89-4.79 (m, 3H, CH(CH3)2, CH2-
Phe), 1.56 (d, 6H, CH(CH3)2, J ) 6.8). Anal. (C15H14N5Cl2I1)
C, H, N.
(2-Iod o-9-isop r op yl-9H -p u r in -6-yl)(4-ch lor ob en zyl)-
a m in e (2e). This compound was obtained in 87% yield with
mp 156-158 °C. 1H NMR δ: 7.68 (s, 1H, H-8), 7,30 (s, 4H,
Phe), 6.28 (br s, 1H, NH), 4.81-4.78 (m, 3H, CH(CH3)2, CH2-
Phe), 1.55 (d, 6H, CH(CH3)2, J ) 6.8). Anal. (C15H15N5Cl1I1)
C, H, N.
(2-Iod o-9-isop r op yl-9H -p u r in -6-yl)(3,4-m et h ylen ed i-
oxyben zyl)a m in e (2f). This compound was obtained in 93%
yield with mp 116-118 °C. 1H NMR δ: 7.85 (s, 1H, H8), 6.89-
6.77 (m, 3H, Phe), 5.95 (s, 2H, OCH2O), 5.05 (m, 1H, CH(CH3)2),
4.77 (m, 2H, CH2Phe), 1.96 (br s, NH,), 1.57 (d, 6H, CH(CH3)2).
Anal. (C16H16N5O2I1) C, H, N.
Asp
Asp
Asp
*Olomoucine, roscovitine, and OL567. § According to sequence
a
comparisons of human CDKs, as described; see Brotherton39 and
b
d
Lawrie,20 De Bondt,17 c Meyerson.40 Amino acid residues in
bold type indicate sequence differences with respect to CDK2.
information will be valuable to current efforts to estab-
lish the molecular basis for the observed selectivity of
our 2-acetylenylpurine analogues for inhibition of CDK1-
cyclin B. Indeed, these molecules, as well as roscovitine
and other trisubstituted purines, show a pronounced
specificity for inhibition of CDK1, CDK2, and CDK5,8,9,11
displaying essentially no capacity to inhibit CDK4- and
CDK6-cyclin complexes. Indeed, although the crystal
structure of CDK6 bound to p19INK4d has been published
recently,39 crystal structures of CDK4 or CDK6, either
alone or bound to a cyclin, will be very helpful to
understand the absence of activity of most synthetic
inhibitors of CDK1 and CDK2 toward CDK4 and CDK6.
However, this inactivity is probably the consequence of
critical sequence differences between CDK4 and CDK6
(see Table 2) and CDK1, CDK2, and CDK5. These latter
three enzymes show a high degree of homology particu-
larly in the catalytic site (Table 2). Several CDK2
residues which in this study have been shown to interact
with OL567 and other members of series 4 and 5 by
sequence alignment are different in CDK4 and CDK6.
In conclusion, the present study has contributed to a
better understanding of the SAR of the new class of
acetylenic purine CKIs. This insight will serve as a
guide in the design of more potent and selective CKIs.
(2-Iod o-9-isop r op yl-9H-p u r in -6-yl)(3,4-d im eth oxyben -
zyl)a m in e (2g). This compound was obtained in 97% yield
1
with mp 177-179 °C. H NMR δ: 7.91 (br s, 1H, H-8), 7.03-
6.81 (m, 3H, Phe), 4.87 (sept, 1H, CH(CH3)2, J ) 6.7), 4.72 (br
d, 2H, CH2Phe), 3.91 (s, 3H, OCH3), 3.87 (s, 3H, OCH3), 2.26
(br s, NH), 1.58 (d, 6H, CH(CH3)2, J ) 6.7). Anal. (C17H20N5O2I1)
C, H, N.
(2-Iodo-9-isopr opyl-9H-pu r in -6-yl)(4-dim eth ylam in oben -
zyl)a m in e (2h ). This compound was obtained in 72% yield
1
with mp 202-204 °C. H NMR δ: 7.61 (s, 1H, H-8), 7.26 (d,
2H, Phe, J ) 8.5), 6.76 (d, 2H, Phe, J ) 7.9), 6.31 (br s, 1H,
NH), 4.80 (sept, 1H, CH(CH3)2, J ) 6.9), 4.65 (m, 2H, CH2-
Phe), 2.95 (s, 6H, N(CH3)2), 1.54 (d, 6H, CH(CH3)2, J ) 6.8).
Anal. (C17H21N6I1) C, H, N.
(2-Iod o-9-isop r op yl-9H-p u r in -6-yl)(p h en yl)a m in e (3a ).
This compound was obtained in 90% yield with mp 124-127
°C. 1H NMR δ: 7.80-7.73 (m, 2H, Phe), 7.61 (s, 1H, H-8),
7.43-7.35 (m, 2H, Phe), 7.15-7.10 (m, 1H, Phe), 4.86 (sept,
1H, CH(CH3)2)), 1.60 (d, 6H, CH(CH3)2, J ) 6.7). Anal.
(C14H14N5I1‚0.3 heptane) C, H, N.
(2-Iod o-9-isop r op yl-9H-p u r in -6-yl)(4-m eth oxyp h en yl)-
a m in e (3b). This compound was obtained in 95% yield with
mp 147-149 °C. 1H NMR δ: 7.86 (s, 1H, H-8), 7.65 (d, 2H,
Phe, J ) 8.6), 6.92 (d, 2H, Phe, J ) 8.7), 4.87 (sept, 1H,
CH(CH3)2, J ) 6.7), 3.82 (s, 3H, OCH3), 1.60 (d, 6H, CH(CH3)2,
J ) 6.7). Anal. (C15H16N5O1I1) C,H, N.
(2-Iod o-9-isop r op yl-9H -p u r in -6-yl)(4-et h oxyp h en yl)-
a m in e (3c). This compound was obtained in 98% yield with
mp 136-138 °C. H NMR δ: 7.89 (br s, 1H, NH), 7.84 (s, 1H,
H-8), 7.64 (d, 2H, Phe, J ) 9), 6.92 (d, 2H, Phe, J ) 9), 4.87
(sept, 1H, CH(CH3)2, J ) 6.7), 4.05 (q, 2H, OCH2CH3, J ) 7),
1.60 (d, 6H, CH(CH3)2, J ) 6.7), 1.42 (t, 3H, CH3CH2O, J )
6.9). Anal. (C16H18N5O1I1) C, H, N.
Exp er im en ta l Section
1
Ch em istr y. The melting points were taken on a Kofler hot
stage apparatus and are uncorrected. Elemental analyses were
performed by the “Service de Microanalyse”, CNRS, ICSN,
91198 Gif sur Yvette, France. Proton nuclear magnetic reso-
nance (1H NMR) spectra were recorded (CDCl3) at 200 MHz
on a Bruker AC 200 spectrometer (J values in Hz).
Gen er a l P r oced u r e for th e P r ep a r a tion of Com p ou n d s
2 a n d 3. A solution of 6-chloro-2-iodo-9H-9-isopropylpurine
(1)25 in ethanol (40 mL) containing triethylamine (4 equiv) and
1.5 equiv of the appropriate amine was stirred under inert
atmosphere (N2) at 50 °C for 3-4 h. After evaporation of the
volatile material in vacuo, workup (CH2Cl2 extraction, water
washing and MgSO4 drying of the organic phase) was followed
(2-Iodo-9-isopr opyl-9H-pu r in -6-yl)(3,4-dich lor oph en yl)-
a m in e (3d ). This compound was obtained in 36% yield with
1
mp 184-186 °C. H NMR δ: 8.2 (br s, 1H, NH), 8.02 (m, 1H,
Phe), 7.94 (s, 1H, H-8), 7.67-7.62 (m, 1H, Phe), 7.45-7.40 (m,
1H, Phe), 4.89 (m, 1H, CH(CH3)2), 1.62 (d, 6H, CH(CH3)2, J
) 6.7). Anal. (C14H12N5Cl2I1) C, H, N.
(2-Iod o-9-isop r op yl-9H -p u r in -6-yl)(3-ch lor op h en yl)-
a m in e (3e). This compound was obtained in 73% yield with

1290 J ournal of Medicinal Chemistry, 2000, Vol. 43, No. 7
Legraverend et al.
mp 108-112 °C. 1H NMR δ: 8.01 (s, 1H, H-8), 7.87 (m, 2H,
Phe, NH), 7.66-7.63 (m, 1H, Phe), 7.34-7.30 (m, 1H, Phe),
7.11-7.07 (m, 1H, Phe), 4.92-4.88 (m, 1H, CH(CH3)2)), 1.60
(d, 6H, CH(CH3)2, J ) 6.8). Anal. (C14H13N5Cl1I1) C, H, N.
Gen er a l P r oced u r e for th e P r ep a r a tion of Com p ou n d s
4 a n d 5. A solution of 6-aryl-2-iodo-9H-9-isopropylpurine 2 or
3 (1 mmol), CuI (0.05 mmol), bis(triphenylphosphine)palla-
dium dichloride (0.01 mmol), n-butylamine (1.5 mmol), and
(R,S)-3-methyl-1-pentyn-3-ol (1.5 equiv) in degassed DMF (25
mL) was stirred under argon atmosphere at 80 °C until
disappearance of the iodopurine, as judged by TLC in CH2-
Cl2-EtOH: 95-5. After evaporation of the solvent the residue
was dissolved in CH2Cl2 and H2S gas was bubbled into the
solution for 2 min.The suspension was evaporated to dryness
and subjected to silica gel column chromatography. When
necessary, analytical samples were obtained after recrystal-
lization from heptane. Melting points refer to analytical
samples.
(3R,S)-1-[9-Isop r op yl-6-(4-m et h oxyp h en yla m in o)-9H -
p u r in -2-yl]-3-m eth ylp en t-1-yn -3-ol (5b). This compound
was obtained in 95% yield with mp 169-171 °C. H NMR δ:
8.01 (br s, 1H, H-8), 7.69 (d, 2H, Phe, J ) 8.7), 6.92 (d, 2H,
Phe, J ) 8.8), 3.82 (s, 3H, OCH3), other protons are identical
to 5a . Anal. (C21H25N5O2) C, H, N.
1
(3R,S)-1-[6-(4-Eth oxyp h en yla m in o)-9-isop r op yl-9H-p u -
r in -2-yl]-3-m eth ylp en t-1-yn -3-ol (5c). This compound was
1
obtained in 48% yield with mp 146-148 °C. H NMR δ: 7.98
(br s, 1H, H-8), 7.68 (d, 2H, Phe, J ) 8.8), 6.90 (d, 2H, Phe, J
) 8.8), 4.03 (q, 2H, OCH2CH3, J ) 7.0), 1.42 (t, 3H, OCH2CH3,
J ) 7.0), other protons are identical to 5a . Anal. (C22H27N5O2)
C, H, N.
(3R,S)-1-[6-(3,4-Dich lor op h en yla m in o)-9-isop r op yl-9H-
p u r in -2-yl]-3-m eth ylp en t-1-yn -3-ol (5d ). This compound
1
was obtained in 66% yield with mp 175-178 °C. H NMR δ:
8.23 (s, 1H, H-8), 7.94 (br s, 1H, NH), 7.61-7.57 (m, 1H, Phe),
7.41-7.37 (m, 1H, Phe), 4.95 (sept, 1H, CH(CH3)2, J ) 6.7),
1.87 (q, 2H, CH2CH3, J ) 7.3), 1.64 (s, 3H, CH3), 1.61 (d, 6H,
CH(CH3)2, J ) 6.8). 1.19 (t, 3H, CH2CH3, J ) 7.3). Anal.
(C20H21N5O1Cl2) C, H, N.
(3R,S)-1-[6-Ben zyla m in o-9-isop r op yl-9H-p u r in -2-yl]-3-
m eth ylp en t-1-yn -3-ol (4a ) has already been described.16
(3R,S)-1-[9-Isop r op yl-6-(4-m et h oxyb en zyla m in o)-9H -
p u r in -2-yl]-3-m eth ylp en t-1-yn -3-ol (4b). This compound
(3R,S)-1-[6-(3-Ch lor op h en yla m in o)-9-isop r op yl-9H-p u -
r in -2-yl]-3-m eth ylp en t-1-yn -3-ol (5e). This compound was
obtained in 68% yield with mp 152-155 °C. 1H NMR δ: 8.25-
8.05 (br m, 3H, H-8, Phe, NH), 7.63-7.58 (m, 1H, Phe), 7.31-
7.26 (m, 1H, Phe), 7.08-7.05 (m, 1H, Phe), 4.95 (sept, 1H,
CH(CH3)2, J ) 6.7), 1.87 (q, 2H, CH2CH3, J ) 7.4), 1.64 (s,
3H, CH3), 1.61 (d, 6H, CH(CH3)2, J ) 6.8), 1.19 (t, 3H,
CH2CH3, J ) 6.8). Anal. (C20H22N5O1Cl1) C, H, N.
Exp r ession , P u r ifica tion , a n d Cr ysta lliza tion of Hu -
m a n CDK2. Human CDK2 was expressed in Sf9 insect cells
using a recombinant baculovirus encoding CDK2 and purified
following the published method.33 Monomeric unphosphory-
lated CDK2 crystals were grown as described.20
1
was obtained in 97% yield with mp 133-136 °C. H NMR δ:
7.32 (d, 2H, Phe, J ) 8.4), 6.86 (d, 2H, Phe), 3.80 (s, 3H, OCH3),
other protons are identical to 5d . Anal. (C22H27N5O2) C, H, N.
(3R,S)-1-[6-(4-Eth oxyben zyla m in o)-9-isop r op yl-9H-p u -
r in -2-yl]-3-m eth ylp en t-1-yn -3-ol (4c). This compound was
1
obtained in 64% yield with mp 110-114 °C. H NMR δ: 7.30
(d, 2H, Phe, J ) 8.4), 6.85 (d, 2H, Phe, J ) 8.3), 4.01 (q, 2H,
OCH2CH3, J ) 6.9), 1.40 (t, 3H, OCH2CH3), other protons are
identical to 4d . Anal. (C23H29N5O2) C, H, N.
(3R,S)-1-[6-(3,4-Dich lor oben zyla m in o)-9-isop r op yl-9H-
p u r in -2-yl]-3-m eth ylp en t-1-yn -3-ol (4d ). This compound
1
was obtained in 72% yield with mp 114-117 °C. H NMR δ:
7.89 (br s, 1H, H-8), 7.50 (m, 1H, Phe), 7.38 (m, 1H, Phe), 7.22
(m, 1H, Phe), 6.6 (br s, 1H, NH), 4.91 (m, 3H, CH(CH3)2, CH2-
Ph), 1.83 (q, 2H, CH2CH3, J ) 7.3), 1.61 (s, 3H, CH3), 1.58 (d,
6H, CH(CH3)2, J ) 6.8), 1.14 (t, 3H, CH2CH3, J ) 7.4). Anal.
(C21H23N5O1Cl2) C, H, N.
Da ta Collection a n d P r ocessin g. The CDK2-OL567
dataset was collected from a crystal soaked for 60 h in 1 mM
OL567 in 1× mother liquor solution (50 mM NH4Ac, 10% PEG
4K, 0.1 M HEPES, pH 7.4) plus 20% DMSO. Data were
collected on beamline X-RAY DIFFRACTION at Elettra,
Trieste, operating at a wavelength of 0.90 Å using a Mar 345
detector and oscillations of 1.5°/frame. The crystal was trans-
ferred briefly to cryoprotectant (1× mother liquor plus 25%
glycerol) before mounting and freezing to 100 K in a nylon loop.
Images were integrated with the DENZO41 package and
subsequently scaled and merged using SCALEPACK.
Str u ctu r e Solu tion a n d Refin em en t. The starting model
for the structure solution and refinement of the CDK2-OL567
structure was that of CDK2 in complex with a small molecule
inhibitor refined at 1.4 Å resolution.42 This model includes
protein residues 1-35 and 44-296. Residues 36-43 prior to
the C-helix are disordered and have not been built in any
reported monomeric CDK2 structure. Residues 297 and 298
in this structure are also disordered. Refinement was begun
by carrying out rigid body refinement of the CDK2 structure.
As the resolution of the data included was increased from 2.5
to 1.9 Å an increasing number of rigid bodies was used, so
that initially the whole molecule was treated as two rigid
bodies, and finally 10 amino acid segments were allowed to
refine independently. The OL567 structure was built in Sybyl43
and at this point the (Fo- Fc)calc maps included readily
interpretable electron density for the bound inhibitor. Refine-
ment of the model was then pursued with alternating cycles
of interactive model building44 and maximum likelihood
refinement using the program REFMAC.45 Toward the end of
refinement, water molecules were added using ARP.46 Statis-
tics relating to the quality of the dataset and of the refined
model are presented in Table 3.
(3R,S)-1-[6-(4-Ch lor oben zyla m in o)-9-isop r op yl-9H-p u -
r in -2-yl]-3-m eth ylp en t-1-yn -3-ol (4e). This compound was
1
obtained in 54% yield with mp 138-142 °C. H NMR δ: 7.86
(br s, 1H, H-8), 7.50 (m, 1H, Phe), 7.29-7.24 (m, 4H, Phe), 6.4
(br s, 1H, NH), 4.89 (m, 3H, CH(CH3)2, CH2Ph), 1.82 (q, 2H,
CH2CH3, J ) 7.3), 1.58 (d, 6H, CH(CH3)2, J ) 6.8), 1.53 (s,
3H, CH3), 1.12 (t, 3H, CH2CH3, J ) 7.3). Anal. (C21H24N5O1-
Cl1) C, H, N.
(3R,S)-1-[9-Isop r op yl-6-(3,4-m eth ylen ed ioxyben zyla m i-
n o)-9H-p u r in -2-yl]-3-m eth ylp en t-1-yn -3-ol (4f). This com-
pound was obtained in 68% yield with mp 133-135 °C. 1H
NMR δ: 7.85 (br s, 1H, H-8), 6.93-6.70 (m, 3H, Phe), 5.93 (s,
2H, OCH2O), 5.00-4.66 (m, 3H, CH(CH3)2, CH2Ph); 1.84 (q,
2H, CH2CH3, J ) 7.4), 1.60 (s, 3H, CH3), 1.56 (d, 6H, CH-
(CH3)2, J ) 6.7), 1.13 (t, 3H, CH2CH3, J ) 7.4). Anal.
(C22H25N5O3) C, H, N.
(3R,S)-1-[6-(3,4-Dim et h oxyben zyla m in o)-9-isop r op yl-
9H-p u r in -2-yl]-3-m eth ylp en t-1-yn -3-ol (4g). This compound
was obtained in 96% yield with mp 75-77 °C. 1H NMR δ:
7.05-6.70 (m, 3H, Phe), 3.87 (s, 6H, 2x OCH3), other protons
are identical to 4d . Anal. (C23H29N5O3) C, H, N.
(3R,S)-1-[6-(4-Dim eth ylam in oben zylam in o)-9-isopr opyl-
9H-p u r in -2-yl]-3-m eth ylp en t-1-yn -3-ol (4h ). This compound
1
was obtained in 88% yield with mp 196-198 °C. H NMR δ:
7.27 (d, 2H, Phe, J ) 8.0), 6.75 (d, 2H, Phe, J ) 8.3), 2.94 (s,
6H, N(CH3)2), other protons are identical to 4d . Anal.
(C23H30N6O1) C, H, N.
(3R,S)-1-[9-Isop r op yl-6-(p h en yla m in o)-9H-p u r in -2-yl]-
3-m eth ylp en t-1-yn -3-ol (5a ). This compound was obtained
CDK1-Cyclin B Assa y. CDK1-cyclin B was extracted in
homogenization buffer (60 mM â-glycerophosphate, 15 mM
p-nitrophenylphosphate, 25 mM MOPS (pH 7.2), 15 mM
EGTA, 15 mM MgCl2, 1 mM dithiothreitol, 1 mM sodium
vanadate, 1 mM NaF, 1 mM phenylphosphate, 10 µg leupeptin/
mL, 10 µg aprotinin/mL, 10 µg soybean trypsin inhibitor/mL
and 100 µM benzamidine) from M phase starfish (Marthast-
1
in 95% yield with mp 129-131 °C. H NMR δ: 7.98 (br s, 1H,
H-8), 7.95-7.85 (m, 2H, Phe), 7.48-7.28 (m, 2H, Phe), 7.20-
7.02 (m, 1H, Phe), 4.93 (sept, 1H, CH(CH3)2, J ) 6.7), 1.87 (q,
2H, CH2CH3, J ) 7.4), 1.63 (s, 3H, CH3); 1.58 (d, 6H, CH-
(CH3)2, J ) 6.7). 1.17 (t, 3H, CH2CH3, J ) 7.4). Anal.
(C20H23N5O1‚1/4H2O) C, H, N.

CDK Inhibition
J ournal of Medicinal Chemistry, 2000, Vol. 43, No. 7 1291
(5) Hui, A. M.; Sun, L.; Kanai, Y.; Sakamoto, M.; Hirohashi, S.
Reduced p27(Kip1) expression in hepatocellular carcinomas.
Cancer Lett. 1998, 132, 67-73.
(6) 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.
(7) Taylor, S. S.; Radzio-Andzelm, E. Three protein kinase structures
define a common motif. Structure 1994, 2, 345-355.
(8) Veseley, J .; Havlicek, L.; Strnad, M.; Blow, J . J .; Donella-Deana,
A.; Pinna, L.; Letham, D. S.; Kato, J .-Y.; Detivaud, L.; Leclerc,
S.; Meijer, L. Inhibition of cyclin-dependent kinases by purine
analogues. Eur. J . Biochem. 1994, 224, 771-786.
(9) Meijer, L.; Borgne, A.; Mulner, O.; Chong, J . P. J .; Blow, J . J .;
Inagaki, N.; Inagaki, M.; Delcros, J . G.; Moulinoux, J . P.
Biochemical and cellular effects of roscovitine, a potent and
selective inhibitor of the cyclin-dependent kinases cdc2, cdk2 and
cdk5. Eur. J . Biochem. 1997, 243, 527-536.
Ta ble 3. Statistics of the Datasets Used and of the Refined
Structure
data collected (space group P212121)
CDK2-OL567
cell dimensions (Å)
maximal resolution (Å)
observations
53.14, 71.47, 71.91
1.80
67163
unique reflections, completeness (%)
Rmerge
25001, 95.8
0.072
a
mean I/σ(I)
protein atoms
residues
8.4
2338
1-35, 44-298
298 water
29 OL567
20.00-1.85
0.19
0.25
24.0
34.2
other atoms
resolution range (Å)
b
Rconv
Rfree
(10) Imbach, P.; Capraro, H. G.; Furet, P.; Mett, H.; Meyer, T.;
Zimmermann, J . 2,6,9-trisubstituted purines: Optimization
towards highly potent and selective CDK1 inhibitors. Bioorg.
Med. Chem. Lett. 1999, 9, 91-96.
c
mean protein temperature factors (Å)b
mean ligand temperature factors (Å)b
(11) Gray, N. S.; Wodicka, L.; Thunnissen, A. M. W. H.; Norman, T.
C.; Kwon, S. J .; 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 for kinase inhibitors. Science 1998, 281, 533-538.
(12) Chang, Y. T.; Gray, N. S.; Rosania, G. R.; Sutherlin, D. P.; Kwon,
S.; Norman, T. C.; Sarohia, R.; Leost, M.; Meijer, L.; Schultz, P.
G. Synthesis and application of functionally diverse 2,6,9-
trisubstituted purine libraries as CDK inhibitors. Chem. Biol.
1999, 6, 361-375.
(13) De Azevedo, W. F.; 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-
2740.
(14) Hoessel, R.; Leclerc, S.; Endicott, J .; Noble, M.; Lawrie, A.;
Tunnah, P.; Leost, M.; Damiens, E.; Marie, D.; Marko, D.;
Niederberger, E.; Tang, W.; Eisenbrand, G.; Meijer, L. Indirubin,
the active constituent of a chinese antileukeamia medicine,
inhibits cyclin-dependent kinases. Nature Cell Biol. 1999, 1, 60-
67.
(15) Schultz, C.; Link, A.; Leost, M.; Zaharevitz, D. W.; Gussio, R.;
Sausville, E. A.; Meijer, L.; Kunick, C. Paullones, a series of
cyclin-dependent kinase inhibitors: synthesis, evaluation of
CDK1/cyclin B inhibition, and in vitro antitumor activity. J . Med.
Chem. 1999, 42, 2909-2919.
(16) Legraverend, M.; Ludwig, O.; Bisagni, E.; Leclerc, S.; Meijer, L.
Synthesis of C2 alkynylated purines, a new family of potent
inhibitors of cyclin-dependent kinases. Bioorg. Med. Chem. Lett.
1998, 8, 793-798.
(17) 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.
(18) Schulze-Gahmen, U.; Brandsen, J .; J ones, H. D.; Morgan, D. O.;
Meijer, L.; Vesely, J .; Kim, S.-H. Multiple modes of ligand
recognition: Crystal structures of cyclin-dependent protein
kinase 2 in complex with ATP and two inhibitors, olomoucine
and isopentenyladenine. Proteins Struct. Funct. Genet. 1995, 22,
378-391.
(19) De Azevedo, W. F.; Leclerc, S.; Meijer, L.; Havlicek, L.; Strnad,
M.; Kim, S. H. Inhibition of cyclin-dependent kinases by purine
analogues - Crystal structure of human cdk2 complexed with
roscovitine. Eur. J . Biochem. 1997, 243, 518-526.
a
Rmerge ) ∑h∑j|Ih,j - I |/∑h∑j|Ih,j|, where Ih,j is the jth observa-
tion of reflection h. Rconvh ) ∑h||Foh| - |Fch||/∑h|Foh|, where Foh and
b
Fch are the observed and calculated structure factor amplitudes,
respectively, for reflection h. Rfree is equivalent to Rconv for a 5%
subset of reflections not used in the refinement.
c
erias glacialis) oocytes and purified by affinity chromatography
on p9CKShs1-Sepharose beads, from which it was eluted by
free p9CKShs1 as previously described. The kinase activity
was assayed in buffer C (homogenization buffer but 5 mM
EGTA, no NaF and no protease inhibitors) with 1 mg histone
H1 (Sigma type III-S)/mL, in the presence of 15 µmol [γ-32P]-
ATP (3000 Ci/mmol; 1 mCi/mL) in a final volume of 30 µL and
at 30 °C. After a 10-min incubation at 30 °C, 25-µL aliquots of
supernatant were spotted onto 2.5 × 3 cm pieces of Whatman
P81 phosphocellulose paper, and 20 s later, the filters were
washed five times (for at least 5 min each time) in a solution
of 10 mL of phosphoric acid/L of water. The wet filters were
transferred into 6-mL plastic scintillation vials, 5 mL of ACS
(Amersham) scintillation fluid was added and the radioactivity
was measured in a Packard counter. Blank values were
subtracted from the data and activities calculated as pmoles
of phosphate incorporated in protein acceptor for a 10-min
incubation. Controls were performed with appropriate dilu-
tions of Me2SO. IC50 values were calculated from dose-
response curves.
Ack n ow led gm en t. We are indebted to the Associa-
tion pour la Recherche sur le Cancer (ARC, Subvention
No. 9814 to M.L. and ARC 9314 to L.M.) as well as the
Conseil Re´gional de Bretagne (to L.M.) for financial
support. We thank the beamline scientists at SRS,
Daresbury, ESRF, Grenoble, and Elettra, Trieste, for
their assistance during data collection. At the LMB,
Oxford, we thank A. Lawrie for preparation of CDK2,
E. Garman for assistance during data collection, I.
Taylor, and S. Lee. This work was supported by the
MRC, BBSRC, and The Royal Society.
(20) 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. Nature Struct. Biol. 1997, 4, 796-801.
(21) 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 cyclinA-CDK2 complex. Nature
1995, 376, 313-320.
Su p p or tin g In for m a tion Ava ila ble: Analytical data.
This material is available free of charge via the Internet at
http://pubs.acs.org.
(22) Russo, A. A.; J effrey, P. D.; Patten, A. K.; Massague´, J .;
Pavletich, N. P. Crystal structure of the p27Kip1 cyclin-dependent-
kinase inhibitor bound to the cyclin A-Cdk2 complex. Nature
1996, 382, 325-331.
(23) Legraverend, M.; Boumchita, H.; Zerial, A.; Huel, C.; Lemaitre,
M.; Bisagni, E. Synthesis of new (()-3,5-dihydroxypentyl nucleo-
side analogues from 1-amino-5-(benzyloxy)pentan-3-ol and their
antiviral evaluation. J . Med. Chem. 1990, 33, 2476-2480.
(24) Legraverend, M.; Boumchita, H.; Bisagni, E. A new route to 2,5-
diamino-4,6-dichloropyrimidine, a key precursor of 9-substituted
guanines. Synthesis 1990, 587-590.
Refer en ces
(1) Keyomarsi, K.; Grana, X. In Progress in Cell Cycle Research;
Meijer, L. G., Philippe, M., Eds.; Plenum Press: New York, 1997;
pp 171-191.
(2) Morgan, D. O. Principles of CDK regulation. Nature 1995, 374,
131-134.
(3) Ball, K. L.; Lain, S.; Fahraeus, R.; Smythe, C.; Lane, D. P. Cell-
cycle arrest and inhibition of CDK4 activity by small peptides
based on the carboxy-terminal domain of p21WAF1. Curr. Biol.
1996, 7, 71-80.
(25) Legraverend, M.; Ludwig, O.; Bisagni, E.; Leclerc, S.; Meijer,
L.; Giocanti, N.; Sadri, R.; Favaudon, V. Synthesis and in vitro
evaluation of novel 2,6,9-trisubstituted purines as cyclin-de-
pendent kinases inhibitors. Bioorg. Med. Chem. 1999, 7, 1281-
1293.
(4) Kiyokawa, H.; Kineman, R. D.; Manova-Todorova, K. O.; Soares,
V. C.; Hoffman, E. S.; Ono, M.; Khanam, D.; Hayday, A. C.;
Frohman, L. A.; Koff, A. Enhanced growth of mice lacking the
cyclin-dependent kinase inhibitor function of p27Kip1. Cell 1996,
85, 721-732.

1292 J ournal of Medicinal Chemistry, 2000, Vol. 43, No. 7
Legraverend et al.
(26) Sonogashira, K.; Tohda, Y.; Hagihara, N. A convenient synthesis
of acetylenes: Catalytic substitutions of acetylenic hydrogen
with bromoalkenes, iodoarenes, and bromopyridines. Tetrahe-
dron Lett. 1975, 4467-4470.
(27) Sharma, P.; Steinbach, P. J .; Sharma, M.; Amin, N. D.; Barchi,
J . J .; Pant, H. C. Identification of substrate binding site of cyclin-
dependent kinase 5. J . Biol. Chem. 1999, 274, 9600-9606.
(28) Chou, K. C.; Watenpaugh, K. D.; Heinrikson, R. L. A model of
the complex between cyclin-dependent kinase 5 and the activa-
tion domain of neuronal Cdk5 activator. Biochem. Biophys. Res.
Commun. 1999, 259, 420-428.
(29) Norman, T. C.; Gray, N. S.; Hoh, J . T.; Schultz, P. G. A structure-
based library approach to kinase inhibitors. J . Am. Chem. Soc.
1996, 118, 7430-7431.
(30) Schow, S. R.; Mackman, R. L.; Blum, C. L.; Brooks, E.; Horsma,
A. G.; J oly, A.; Kerwar, S. S.; Lee, G.; Shiffman, D.; Nelson, M.
G.; Wang, X.; Wick, M. M.; Zhang, X.; Lum, R. T. Synthesis and
activity of 2,6,9-trisubstituted purines. Bioorg. Med. Chem. Lett.
1997, 7, 2697-2702.
(31) Zimmerman, J .; Capraro, H. G.; Peterli, P.; Furet, P. Purine
derivatives and processes for their preparation. Patent WO 97/
16452, 1997.
(36) Schulze-Gahmen, U.; De Bondt, H. L.; Kim, S.-H. High-resolution
crystal structures of human cyclin-dependent kinase 2 with and
without ATP: Bound waters and natural ligand for inhibitor
design. J . Med. Chem. 1996, 39, 4540-4546.
(37) Russo, A.; J effrey, P. D.; Pavletich, N. P. Structural basis of
cyclin-dependent kinase activation by phosphorylation. Nature
Struct. Biol. 1996, 3, 696-700.
(38) Nishio, M.; Hirota, M. CH/π interaction: Implications in organic
chemistry. Tetrahedron 1989, 45, 7201-7245.
(39) Brotherton, D. H.; Dhanaraj, V.; Wick, S.; Brizuela, L.; Domaille,
P. J .; Volyanik, E.; Xu, X.; Parisini, E.; Smith, B. O.; Archer, 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 p19(INK4d). Nature 1998,
395, 244-250.
(40) Meyerson, M.; Enders, G. H.; Wu, C.-L.; Su, L.-K.; Gorka, C.;
Nelson, C.; Harlow, E.; Tsai, L.-H. A family of human cdc2-
related protein kinases. EMBO J . 1992, 11, 2909-2917.
(41) Otwinowski, Z.; Minor, W. Processing of X-ray diffraction data
collected in oscillation mode. Methods Enzymol. 1997, 276, 307-
326.
(42) Calvert, A. H. Unpublished results.
(32) Meijer, L.; Kim, S. H. Chemical inhibitors of cyclin-dependent
kinases. Methods Enzymol. 1997, 283, 113-128.
(33) Brown, N. R.; Noble, M. E. M.; Lawrie, A. M.; Morris, M. C.;
Tunnah, P.; Divita, G.; J ohnson, L. N.; Endicott, J . A. Effects
of phosphorylation of threonine 160 on cyclin-dependent kinase
2 structure and activity. J . Biol. Chem. 1999, 274, 8746-
8756.
(34) Wallace, A. C.; Laskowski, R. A.; Thornton, J . M. LIGPLOT: a
program to generate schematic diagrams of protein-ligand
interactions. Protein Eng. 1995, 8, 127-134.
(35) Rosenblatt, J .; DeBondt, H.; J ancarik, J .; Morgan, D. O.; Kim,
S. H. Purification and crystallisation of human cyclin-dependent
kinase 2. J . Mol. Biol. 1993, 230, 1317-1319.
(43) Sybyl, 6.3 ed.; SYBYL Molecular Modeling Software, Tripos
Associates Ltd.: St. Louis, MO, 1992.
(44) J ones, T. A.; Zou, J . Y.; Cowan, S. W.; Kjeldgaard, M. Improved
method for building models in electron density maps and the
location of errors in these models. Acta Crystallogr. 1991, A47,
110-119.
(45) Murshudov, G. N.; Vagen, A. A.; Dodson, E. J . Refinement of
macromolecular structures by the maximum -likelihood method.
Acta Crystallogr. 1997, D53, 240-255.
(46) Lamzin, V. S.; Wilson, K. S. Automated refinement of protein
models. Acta Crystallogr. 1993, D49, 129-147.
J M9911130