Identification of novel purine and pyrimidine cyclin-dependent kinase inhibitors with distinct molecular interactions and tumor cell growth inhibition profiles

Article abstract of DOI:10.1021/jm990628o

Substituted guanines and pyrimidines were tested as inhibitors of cyclin B1/CDK1 and cyclin A3/CDK2 and soaked into crystals of monomeric CDK2. O⁶- Cyclohexylmethylguanine (NU2058) was a competitive inhibitor of CDK1 and CDK2 with respect to ATP (K(i) values: CDK1, 5 ± 1 μM; CDK2, 12 ± 3 μM) and formed a triplet of hydrogen bonds (i.e., NH-9 to Glu 81, N-3 to Leu 83, and 2-NH₂ to Leu 83). The triplet of hydrogen bonding and CDK inhibition was reproduced by 2,6-diamino-4-cyclohexylmethyloxy-5-nitrosopyrimidine (NU6027, K(i) values: CDK1, 2.5 ± 0.4 μM; CDK2, 1.3 ± 0.2 μM). Against human tumor cells, NU2058 and NU6027 were growth inhibitory in vitro (mean GI₅₀ values of 13 ± 7 μM and 10 ± 6 μM, respectively), with a pattern of sensitivity distinct from flavopiridol and olomoucine. These CDK inhibition and chemosensitivity data indicate that the distinct mode of binding of NU2058 and NU6027 has direct consequences for enzyme and cell growth inhibition.

Full text of DOI:10.1021/jm990628o

J . Med. Chem. 2000, 43, 2797-2804
2797
Id en tifica tion of Novel P u r in e a n d P yr im id in e Cyclin -Dep en d en t Kin a se
In h ibitor s w ith Distin ct Molecu la r In ter a ction s a n d Tu m or Cell Gr ow th
In h ibition P r ofiles
Christine E. Arris,# F. Thomas Boyle,§ A. Hilary Calvert,# Nicola J . Curtin,# J ane A. Endicott,¶
Elspeth F. Garman,¶ Ashleigh E. Gibson,# Bernard T. Golding,# Sharon Grant,# Roger J . Griffin,#
Philip J ewsbury,§ Louise N. J ohnson,¶ Alison M. Lawrie,¶ David R. Newell,*,# Martin E. M. Noble,¶
Edward A. Sausville, Robert Schultz, and Wyatt Yu¶
Cancer Research Unit and Department of Chemistry, University of Newcastle, Newcastle upon Tyne, NE2 4HH U.K.,
AstraZeneca Pharmaceuticals, Alderley Park, Cheshire, U.K., Laboratory of Molecular Biophysics and Department of
Biochemistry, University of Oxford, Oxford, OX1 3QU U.K., and Developmental Therapeutics Program, National Cancer
Institute, Bethesda, Maryland 20892-7458
Received December 23, 1999
Substituted guanines and pyrimidines were tested as inhibitors of cyclin B1/CDK1 and cyclin
A3/CDK2 and soaked into crystals of monomeric CDK2. O6-Cyclohexylmethylguanine (NU2058)
was a competitive inhibitor of CDK1 and CDK2 with respect to ATP (Ki values: CDK1, 5 ( 1
µM; CDK2, 12 ( 3 µM) and formed a triplet of hydrogen bonds (i.e., NH-9 to Glu 81, N-3 to
Leu 83, and 2-NH2 to Leu 83). The triplet of hydrogen bonding and CDK inhibition was
reproduced by 2,6-diamino-4-cyclohexylmethyloxy-5-nitrosopyrimidine (NU6027, Ki values:
CDK1, 2.5 ( 0.4 µM; CDK2, 1.3 ( 0.2 µM). Against human tumor cells, NU2058 and NU6027
were growth inhibitory in vitro (mean GI50 values of 13 ( 7 µM and 10 ( 6 µM, respectively),
with a pattern of sensitivity distinct from flavopiridol and olomoucine. These CDK inhibition
and chemosensitivity data indicate that the distinct mode of binding of NU2058 and NU6027
has direct consequences for enzyme and cell growth inhibition.
Ch a r t 1. Chemical Structures of CDK Inhibitors
In tr od u ction
The cyclin-dependent kinases (CDKs) are a family of
serine-threonine protein kinases which control cell cycle
progression in proliferating eukaryotic cells.1-5 The
activity of CDKs is dependent upon the presence of
cyclin partners whose levels are sequentially regulated
to ensure that the phases of the cell cycle proceed in
the correct order. For example, cyclins of the D family
complex with CDKs 4 and 6 during G1 phase, cyclin E
with CDK2 in late G1, cyclin A with CDK2 in S phase,
and cyclin B with CDK1 (also known as cdc2) in late
G2/M. Aberrant CDK control and consequent loss of cell
cycle checkpoint function have been directly linked to
the molecular pathology of cancer.6 For example, cyclin
overexpression (e.g., cyclin D), loss of function of en-
dogenous CDK inhibitors (e.g., p16 deletion or p21 loss
secondary to p53 defects), and CDK substrate alter-
ations (e.g., retinoblastoma gene mutations) have been
widely documented in human tumors.6 These CDK-
related events are among the most common genetic
changes found in human tumors and, clinically, they
often confer a poor prognosis.
CDK inhibitors should function as non-DNA-interactive
antiproliferative agents, thereby avoiding treatment-
induced DNA damage in normal tissues and the con-
sequent carcinogenic risk. Furthermore, when used in
combination with cytotoxic drugs, reestablishing an
otherwise deficient cell cycle checkpoint may enhance
the activity of conventional treatments. Studies with
first-generation CDK inhibitors support a role for the
use of these agents as both single agents and in
combination with cytotoxic drugs (see below).
In view of the strong link between CDKs and the
molecular pathology of cancer, CDKs are being inves-
tigated as possible targets for therapeutic intervention.7-9
* Address correspondence to this author at the Cancer Research
Unit, Medical School, University of Newcastle, Framlington Place,
Newcastle NE2 4HH, England, U.K. E-mail: herbie.newell@ncl.ac.uk.
Telephone: -44-191-222-8057. Fax: -44-191-222-7556.
#
Cancer Research Unit and Department of Chemistry, University
of Newcastle.
§ AstraZeneca Pharmaceuticals.
A lead CDK inhibitor is flavopiridol (Chart 1a) which
has recently completed Phase I clinical evaluation.10
Evidence of activity was seen in the Phase I trial of
¶
Laboratory of Molecular Biophysics and Department of Biochem-
istry, University of Oxford.
Developmental Therapeutics Program, National Cancer Institute.
10.1021/jm990628o CCC: $19.00 © 2000 American Chemical Society
Published on Web 07/11/2000

2798 J ournal of Medicinal Chemistry, 2000, Vol. 43, No. 15
Arris et al.
6-chloro-9-methylpurine, which was prepared by methylation
of 2-amino-6-chloropurine. 2,6-Diamino-4-cyclohexylmethoxy-
5-nitrosopyrimidine (NU6027, Chart 2c) and 2,6-diamino-4-
cyclohexylmethoxypyrimidine (NU6034, Chart 2d) were syn-
thesized from 2,6-diamino-4-chloropyrimidine as described
previously.29
Ch a r t 2. Structures of Compounds Designed to Probe
the Interaction of NU2058 with CDKs
Exp er im en ta l: NU2058. Cyclohexylmethanol (9.71 mmol)
and sodium hydride (3.54 mmol) were stirred in anhydrous
DMSO (8.0 mL) for 1 h, before addition of 2-amino-6-(1,4-
diazabicyclo[2.2.2]oct-1-yl)purinium chloride (1.78 mmol). After
the reaction was stirred for 48 h, glacial acetic acid was added
to adjust the pH to 7. The solvent was removed, and the crude
product was loaded onto a silica column, which was eluted with
10% MeOH/DCM. The desired product was obtained as a white
solid (51%), mp 202-204 °C; λmax (CH3OH)/nm 281; IR 3350,
3200, 2900, 1640 cm-1; 1H NMR δ 1.5 (m, 11H, C6H11), 4.5 (d,
2H, OCH2), 6.31 (s, 2H, NH2), 7.92 (s, 1H, C8H); 13C NMR δ
25.5 (C4′), 26.3 (C3′ and C5′), 29.5 (C2′ and C6′), 37.2 (C1′),
70.8 (OCH2), 137.9 (C8), 157.0 (C4), 160.1 (C2), 162.0 (C6);
HRMS (EI) m/z 247 [M+], 170, 151, 134, 109, 81. Anal.
(C12H17N5O) C, H, N.
flavopiridol, consistent with preclinical in vivo data,11,12
and Phase II trials are underway. However, enzyme
inhibition studies have shown that the CDK specificity
of flavopiridol is limited, CDKs 1, 2, and 4 being equally
sensitive to inhibition by the compound.8,13 Consistent
with the lack of CDK specificity, whole cell studies have
shown that following treatment with flavopiridol cell
cycle arrest can occur at either the G1/S and/or G2/M
checkpoints.13-16
In the search for more specific CDK inhibitors a
number of pharmacophores have been investigated,
prominent among which are the 6-aminopurine-based
inhibitors.17-24 Where studied, the 6-aminopurines have
been shown to be competitive inhibitors with respect to
ATP, and structural investigations with monomeric
CDK2 have shown that certain of the compounds (e.g.,
olomoucine, roscovitine, and purvalanol B (Chart 1b-
d, respectively) bind as expected in the ATP binding
site.21,25,26 However, the 6-aminopurines do not repro-
duce the hydrogen-bonding characteristics of ATP.
Specifically, ATP forms a single hydrogen bond with Leu
83 of CDK2 through the N-1 position of the purine ring,
and a second bond to Glu 81 through the N-6 position.
In contrast, the 6-aminopurines interact with Leu 83
through the N-7 and N-6 positions, and Glu 81 via
C-8.21,25,26
As part of a screening exercise, a series of guanine
analogues were evaluated for their ability to inhibit
CDK1. Surprisingly, given the lack of an exocyclic C-6
amino group and hence the potential for interaction with
Leu 83 via this position, certain O6-alkylguanines were
found to be CDK1 inhibitors (L. Meijer, personal com-
munication). This paper describes the inhibition of
CDKs 1 and 2 by selected O6-cyclohexylmethylguanines.
The tumor cell growth inhibitory properties of O6-cyclo-
hexylmethylguanines and O4-cyclohexylmethylpyrimi-
dines designed to probe hydrogen bond interactions
within the CDK2 active site, identified by crystal
structure analysis, have also been determined.
NU2017. Cyclohexylmethanol (0.8 mL, 6.5 mmol) and
sodium hydride (0.08 g, 3.33 mmol) were added to anhydrous
DMSO (8.0 mL) under N2, and the mixture was stirred for 1 h
at room temperature. 6-(1,4-Diazabicyclo[2.2.2]octyl)purinium
chloride (0.3 g, 1.13 mmol) was added, and stirring was
continued for 3 days. The reaction was neutralized by addition
of glacial acetic acid, and solvents were removed. Purification
by column chromatography (eluent 10% MeOH/DCM) yielded
1
a white solid (57%), mp 210-211 °C; H NMR δ 1.25 (m, 5H,
C6H11), 1.9 (m, 6H, C6H11), 4.4 (d, 2H, J ) 6 Hz, OCH2), 8.5
(s, 1H, C2H or C8H), 8.6 (s, 1H, C2H or C8H); HRMS (EI) m/z
232 (M+), 149, 135, 119, 81, 41. Anal. (C12H16N4O) C, H, N.
NU6052. 2-Amino-6-chloropurine (2.95 mmol) and potas-
sium carbonate (2.95 mmol) in DMF (2 mL) were stirred
vigorously for 30 min. Methyl iodide (3.25 mmol) was added,
and stirring continued at room temperature for 18 h. The
solvent was removed under reduced pressure. Purification by
column chromatography (eluent 10% MeOH/DCM) yielded the
N9 isomer (62%, white solid, mp > 300 °C) as the predominant
product. UV λmax 247, 311 nm (MeOH); IR 3412, 3332, 3210,
2990, 1647, 1607, 1562 cm-1 1H NMR δ 8.2 (s, 1H, C8H),
;
7.0 (s, 2H, NH2), 3.8 (s, 3H, CH3); HRMS (EI) m/z 183.030693
[M+ calc for C6H6N5Cl], 148. Sodium (2.46 mmol) was added
to cyclohexylmethanol (2 mL) at 90 °C. After 1 h, 2-amino-6-
chloro-9-methylpurine (0.82 mmol) was added, and the reac-
tion was stirred for a further 2 h. The reaction mixture was
cooled to room temperature, after which the pH was adjusted
to 7 by addition. The solvent was removed and the residue
extracted with water (100 mL) and EtOAc (2 × 100 mL). The
organic layers were combined, dried, and evaporated. The
residue was purified by column chromatography (eluent 3%
MeOH/DCM). Recrystallization from ethyl acetate/petrol yielded
a white solid product (60%), mp 174-175 °C. UV λmax 250, 283
nm (MeOH); IR 3382, 3338, 2922, 2850, 1651, 1607, 1582 cm-1
;
1H NMR δ 1.0-1.5 and 1.7-2.0 (m, 11H, C6H11), 3.7 (s, 3H,
CH3), 4.3 (s, 2H, OCH2), 6.5 (s, 2H, NH2), 7.9 (s, 1H, H8); HRMS
(EI) m/z 261.157974 [M+ calcd for C13H19N5O 261.158960], 178,
165, 148. Anal. (C13H19N5O) C, H, N.
NU6034. Cyclohexylmethanol (30 mL) and sodium (32
mmol) were heated under N2 at 150 °C for 90 min. 4-Chloro-
2,6-diaminopyrimidine (30 mmol) was added, and the reaction
mixture was refluxed at 180 °C for 2 h. The excess of
cyclohexylmethanol was removed, and the residue was purified
by column chromatography (eluent 10% MeOH/DCM). The
product was recrystallized from MeOH yielding a white solid
Ma ter ia ls a n d Meth od s
1
Ch em istr y. O6-Cyclohexylmethylguanine (NU2058, Chart
1e) was synthesized as previously reported,27,28 using either
2-amino-6-chloropurine or 2-amino-6-(1,4-diazabicyclo[2.2.2]-
oct-1-yl)purinium chloride as starting materials, respectively.
The latter method has the advantage of easier purification of
products and was also used to prepare O6-cyclohexylmethyl-
purine (NU2017, Chart 2a). O6-Cyclohexylmethyl-N9-methyl-
guanine (NU6052, Chart 2b) was synthesized from 2-amino-
(70%), mp 142 °C; H NMR δ 1.05 (m, 5H,), 1.85 (m, 6H), 4.0
(d, 2H, OCH2), 5.15 (s, 1H, C5H), 5.95 (s, 2H, NH2), 6.1 (s, 2H,
NH2); HRMS (EI) m/z 222 [M+]. Anal. (C11H18N4O) C, H, N.
NU6027. 2,6-Diamino-4-cyclohexylmethoxypyrimidine (1.26
mmol) was dissolved in warm 30% glacial acetic acid solution
(10 mL). The solution was heated to 80 °C, and aqueous sodium
nitrite solution (1.72 mmol in 5 mL of H2O) was added
dropwise over 1 h. The reaction mixture was allowed to cool

Purine and Pyrimidine CDK Inhibitors
J ournal of Medicinal Chemistry, 2000, Vol. 43, No. 15 2799
Ta ble 1. Statistics of the Data Sets Used and of the Refined
Structures
to room temperature, and the violet crystals were filtered off
and washed well with water. The crude product was purified
by recrystallization from EtOH (83%), mp 254 °C; H NMR δ
1.25 (m, 5H,), 2.0 (m, 6H), 4.4 (d, 2H, OCH2), 7.9 (s, 2H, NH2),
8.1 (s, 1H, NH), 10.2 (s, 1H, NH); HRMS (EI) m/z 251 [M+].
Anal. (C11H17N5O2) C, H, N.
1
CDK2/NU2058
CDK2/NU6027
cell dimensions (Å)
maximal resolution (Å)
observations
52.62, 71.05, 71.50 52.65, 69.90, 71.61
1.95
1.85
42 243
62 274
En zym e In h ibition Stu d ies. Inhibition of cyclin B1/CDK1
was assayed as previously described17 using enzyme prepared
from starfish oocytes (Marthasterias glacialis). Inhibition of
cyclinA3/CDK2 was determined using a similar assay and an
assay buffer comprised of 50 mM Tris-HCl pH 7.5 containing
5 mM MgCl2. Cyclin A3/CDK2 was prepared as previously
described30 (cyclin A3 is a C-terminal cyclin A fragment
encoding residues 171-432). The final ATP concentration in
both CDK assays was 12.5 µM ,and the IC50 concentration for
each compound is the concentration required to inhibit enzyme
activity by 50% under the assay conditions used. To determine
the Km for ATP for cyclin B1/CDK1 and cyclin A3/CDK2, and
Ki values for NU2058 and NU6027, assays were performed in
the absence of inhibitor and at two fixed inhibitor concentra-
tions (NU2058, 25 and 50 µM; NU6027, 5 and 10 µM), with
ATP concentrations ranging from 6.25 to 800 µM. Data were
fitted to the Michaelis-Menten equation using unweighted
nonlinear least squares regression.
unique reflections,
completeness (%)
Rmerge
17 940 (88.8)
22 399 (96.4)
a
0.065
12.3
2.04-1.95
84.3
2.6
0.313
2338
290
174 water
18 NU2058
20.00-1.8
19.5
26.9
25.2
0.057
16.3
1.93-1.85
97.2
2.27
0.372
2338
290
144 water
18 NU6027
20.00-1.85
21.3
28.1
35.9
mean I/(I)
highest resolution bin (Å)
completeness (%)
mean I/mean (I)
Rmerge
protein atoms
residues
other atoms
resolution range (Å)
b
Rconv
Rfree
c
mean main chain protein
temperature factors (Å)2
mean ligand temperature
factors (Å)2
16.9
42.6
Expr ession , P u r ification , an d Cr ystallization of Hu m an
CDK2. Human CDK2 was expressed in Sf9 insect cells using
a
Rmerge ) ∑h∑j|Ih,j - hIh|/∑h∑j|Ih,j| where Ih,j is the intensity of
b
the jth observation of unique reflection h. Rconv ) ∑h||Fo | - |Fc ||/
h
h
a
recombinant baculovirus encoding CDK2 and purified
∑h|Fo | where Fo and Fc are the observed and calculated structure
following slight modifications to a published method.30 Mono-
meric unphosphorylated CDK2 crystals were grown as previ-
ously described.31
h
h
h
c
factor amplitudes for reflection h. Rfree is equivalent to Rconv but
is calculated using a 5% disjoint set of reflections excluded from
the least squares refinement stages.
X-r a y Cr ysta llogr a p h y Da ta Collection a n d P r ocess-
in g. The CDK2/NU2058 data set was collected from a crystal
soaked for 48 h in 5 mM NU2058 in 1× mother liquor solution
(50 mM ammonium acetate, 10% PEG 3350, 15 mM NaCl, 100
mM HEPES, pH 7.4) plus 5% DMSO. The CDK2/NU6027 soak
conditions were 26 h in 5mM NU6027 in 1x mother liquor
solution plus 5% DMSO. CDK2/NU2017 soak conditions were
similar. Data for the CDK2/NU2058 complex was collected on
beamline 9.5, SRS, Daresbury, U.K. The CDK2/NU6027
complex data set was collected on X-RAY DIFFRACTION at
the Elettra Light source, and the CDK2/NU2017 data set was
collected on BM14 at the ESRF. In each case data were
collected at 100 K after crystals had been transferred briefly
to cryoprotectant (mother liquor containing 20% glycerol).
Images were integrated with the DENZO package and sub-
sequently scaled and merged using SCALEPACK.32 Statistics
for the CDK2/NU2058 and CDK2/NU6027 data sets are given
in Table 1.
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/inhibitor
complexes was the CDK2/ATP structure, PDB code 1HCK.25
This model included protein residues 1-36 and 41-298.
Refinement of the structure was begun by carrying out rigid
body refinement using REFMAC,36 using sequentially higher
resolution data. As the resolution of the data included was
increased from 3.0 Å to approximately 2.0 Å, an increasing
number of rigid bodies were used, so that initially the whole
molecule was treated as a single rigid body, and finally five-
residue segments were allowed to refine independently. Fol-
lowing rigid body refinement, the (Fo - Fc)Rcalc maps included
readily interpretable electron density for the bound inhibitors.
Clear electron density also defined those residues involved in
interactions with the inhibitors. The structure was then
compared with 2Fo - Fc electron density maps using the
program “O”,33 and minor structural changes were introduced.
Following the first round of model building, a model of the
appropriate ligand generated within SYBYL34 was built into
the electron density map and included in subsequent refine-
ment steps. Appropriate parameters for the geometric re-
straints in subsequent refinement by the program REFMAC
were taken from the SYBYL minimized structure. Refinement
was then pursued with alternating cycles of interactive model
building and maximum likelihood refinement using the pro-
gram REFMAC.35,36 Toward the end of the refinement, water
Ta ble 2. Inhibition of CDK1 and CDK2 and Cell Growth by
Cyclohexylmethyl Guanine and Pyrimidine Derivatives
CDK1 inhibition
CDK2 inhibition
mean GI50
compounda
Ki or IC50 (µM)b
Ki or IC50 (µM)b
(µM)c
NU2058
NU2017
NU6052
NU6027
NU6034
Ki ) 5 (1
Ki ) 12 (3
13 ( 7
55
>100
10 ( 6
>100
IC50 ) 18 ( 7
IC50 > 100
Ki ) 2.5 ( 0.4
IC50 > 10d
IC50 ) 13(10-16)
IC50 > 100
Ki ) 1.3 ( 0.2
IC50 > 10d
a
b
For structures, see Charts 1 and 2. IC50 is the concentration
of inhibitor required to inhibit enzyme activity by 50% under the
assay conditions used, and Ki is the associated inhibition constant.
Values are from independent studies or the mean ( SD where
n ) >3 experiments or the mean and range for n ) 2. c The mean
GI50 is the mean of the concentrations of the compound required
to inhibit cell growth by 50% in each of the cell lines (total ) 57),
d
mean ( SD where n ) >2 experiments. Less than 10% inhibition
at 10 µM, the maximum achievable concentration.
molecules were added with ARP.36,37 Statistics for the final
models are given in Table 1 and the atomic coordinates are
available in PDB files 1E1V (CDK2/NU2058) and 1E1X
(CDK2/NU6027).
Cell Gr ow th In h ibition Stu d ies. The growth inhibitory
activity of the compounds was evaluated in the NCI in
vitro cell line panel using the standard 48 h exposure assay
and inhibitor concentrations ranging from 10-9 to 10-4 M.38
Relationships between the profile of cell growth inhibition
produced by the novel CDK inhibitors and that of stand-
ard anticancer agents, and the established CDK inhibitors
olomoucine and flavopiridol, was investigated using the
COMPARE algorithm.39
Resu lts a n d Discu ssion
The lead guanine-based CDK inhibitor O6-cyclo-
hexylmethylguanine (NU2058, Chart 1e) was found to
be an inhibitor of both CDK1 and CDK2 (Table 2).
Enzyme kinetic studies demonstrated that the Km for
ATP for both CDK1 and CDK2 under the assay condi-
tions used was 50 µM (51 ( 5 µM and 53 ( 6 µM,

2800 J ournal of Medicinal Chemistry, 2000, Vol. 43, No. 15
Arris et al.
respectively, mean ( SD, n ) 4), and the Ki values for
NU2058 were as follows: CDK1, 5 ( 1 µM; CDK2,
12 ( 3 µM (mean ( SD, n ) 4). Against both enzymes,
inhibition by NU2058 was competitive with respect
to ATP.
Inhibitors were soaked into crystals of monomeric,
unphosphorylated, CDK2 and the structures refined to
the resolutions given in Table 1. The structure of CDK2
represents the minimal catalytic kinase core, consti-
tuted by two domains: a smaller N-terminal domain of
approximately 80 residues formed principally from
â-sheet, and a larger (∼210 residue) predominantly
R-helical C-terminal domain. The first complex charac-
terized was CDK2/NU2058. Consistent with the enzyme
inhibition data, NU2058 was found to bind within the
ATP site of CDK2. A comparison of the refined structure
to that of apo-CDK2,40 and to other CDK2-inhibitor
complexes,21,25,26,31,41,42 revealed that NU2058 employs
a number of novel interactions to bind to CDK2. Figure
1 shows the site of binding of NU2058 to CDK2 and the
hydrogen-bonding pattern between NU2058 and the
CDK2 backbone.
As shown in Figure 1b, N-9 of NU2058 donates a
hydrogen bond to the peptide oxygen of Glu 81, the NH
group of Leu 83 donates a hydrogen bond to NU2058
N-3, and 2-NH2 acts as a donor to the carbonyl group
of Leu 83. Although residues 80-84 constitute part of
the hinge between the N- and C-terminal domains of
CDK2, the interactions do not result in a significant
change in relative domain orientation. Importantly,
NU2058 binds into the ATP binding site cleft in a
different orientation from that of the purines olo-
moucine, roscovitine, or purvalanol B (Figure 2).21,25,26
The hydrogen bonds to the carbonyl group of Glu 81,
and the amide group of Leu 83, are also made by
isopentenyladenine and ATP, while the hydrogen bond
with the carbonyl group of Leu 83 is also seen in the
complexes of CDK2 with olomoucine, roscovitine, and
purvalanol B. Hydrogen bonds to these amino acids are
also observed with members of the indirubin class of
CDK2 inhibitors,42 although these are structurally
dissimilar from NU2058. The binding of the guanine
ring of NU2058 is not coplanar with that of the ATP
adenine ring in the CDK2/ATP structure,25 and minor
variations in ring position are observed among the
inhibitors described here. The cyclohexylmethyl group
at the O6-position of NU2058 occupies space close to
where the ribose ring of ATP binds, although it cannot
mimic the hydrogen-bonding interactions of the ribose
2′- and 3′-hydroxyls.
F igu r e 1. Interaction of NU2058 with monomeric CDK2: (a)
NU2058 binding to CDK2 in the ATP binding site cleft. The
structure is shown in a cartoon representation. The N-terminal
domain (residues 1-82) is colored white, with the exceptions
of the glycine-rich loop (residues 11-24) colored magenta, and
the C-helix (PSTAIRE helix, residues 46-55) colored gold. The
chain is broken at residues 36-43 for which there is no
electron density in the maps. NU2058 is shown in ball-and-
stick representation at the interface between the N- and
C-terminal domains. The C-terminal domain is colored pink,
with the activation segment (residues 145-172) highlighted
in cyan. (b) Schematic drawing of NU2058 bound to CDK2.
Conserved hydrogen bonds between the CDK2 backbone at
residues Glu 81 and Leu 83 and NU2058 are illustrated by
thin lines. CDK2 residues within 4 Å of the ligand are drawn
with carbon atoms colored as for Figure 1a. Selected CDK2
residues are labeled.
In the apo-CDK2 structure, and in other monomeric
CDK2-inhibitor complex structures, two regions of
CDK2 have been reported to be flexible.25,31,40 The
missing residues constitute the loop connecting strand
â3 to helix RC, and the “activation segment” containing
Thr 160 which is phosphorylated in the fully active
CDK2-cyclin A complex.43 NU2058 binding to CDK2 is
accompanied by a marked decrease in the main chain
temperature factors for all residues, but it is particularly
noticeable for residues in strand â1 and the activation
loop and results in a readily interpretable electron den-
sity map for both these regions (Figure 3a). Additional
interactions between the NU2058 O6-substituent and
the glycine-rich loop accompany stabilization of the
activation loop (Figure 3a). These interactions are pre-
dominantly hydrophobic between the cyclohexyl group
and a hydrophobic patch on the glycine loop contributed
by the side chains of residues Val 18 and the peptide
backbone around residue Gly 11 (Figure 3c). Residues
in the activation loop do not interact directly with
NU2058. The activation segment conformation is simi-
lar to that first reported by DeBondt et al.,40 for the
structure of the monomeric CDK2/ATP complex. Order-
ing of both the glycine-rich loop and the activation

Purine and Pyrimidine CDK Inhibitors
J ournal of Medicinal Chemistry, 2000, Vol. 43, No. 15 2801
6-NH2 group. As shown in Table 2, NU6027 was indeed
a potent inhibitor of both CDK1 and CDK2, with Ki
values of 2.5 ( 0.4 µM and 1.3 ( 0.2 µM, respectively
(mean ( SD, n ) 4). As in the case of NU2058, the
pyrimidine NU6027 was a competitive inhibitor of both
CDK1 and CDK2 with respect to ATP.
NU6027 was soaked into crystals of monomeric CDK2
and the structure refined to a resolution of 1.85 Å. As
shown in Figure 4, as anticipated, the 6-NH2 group of
NU6027 donates a hydrogen bond to the peptide oxygen
of Glu 81, the NH group of Leu 83 donates a hydrogen
bond to N-1, and the NH2 group attached to C-2 acts as
a donor to the carbonyl group of Leu 83. Importantly,
the predicted intramolecular hydrogen bond between
the 5-nitroso and 6-NH2 groups of NU6027 was ob-
served, and this interaction was apparently the only
contribution to binding made by the 5-nitroso group. To
confirm the importance of the intramolecular hydrogen
bond between the 5- and 6-substituents of NU6027, the
derivative lacking the 5-nitroso group was synthesized
(2,6-diamino-O4-cyclohexylmethyloxypyrimidine, NU6034,
Chart 2d). As shown in Table 2, NU6034 did not inhibit
CDK1 or -2 at the concentrations tested, suggesting that
the intramolecular hydrogen bond between the 5- and
6-substituents of NU6027, and hence the correct orien-
tation of the 6-NH2 group, is required for activity.
F igu r e 2. Overlay of the CDK2/olomoucine structure with
that of the CDK2/NU2058 complex in the vicinities of the
inhibitor binding sites. Conserved hydrogen bonds between the
CDK2 backbone, at residues Glu 81 and Leu 83, and NU2058
are drawn as thin lines. Only CDK2 residues Phe 80 to Leu
83 in the CDK2/NU2058 structure are included for clarity.
CDK2 carbon atoms are colored as for Figure 1a; NU2058
carbon atoms are green, and olomoucine carbon atoms are
yellow. The O6-cyclohexyl substituent of NU2058 and the
2-hydroxyethyl group of olomoucine occupy the ATP ribose
binding site. The olomoucine N6-group binds outside the ATP
cleft. Roscovitine and purvalanol B bind to CDK2 in the same
orientation as olomoucine.
Having defined the unique interactions of the novel
guanine- and pyrimidine-based CDK inhibitors with
CDK2, the effects of the compounds on the growth of
human tumor cells was investigated. Fifty-seven of the
cell lines in the National Cancer Institute in vitro cell
line panel were used for these studies in order to allow
comparisons with other compounds, both standard
anticancer agents and known CDK inhibitors. Table 2
shows the mean GI50 data (concentrations required to
inhibit cell growth by 50% over 48 h) for the compounds
investigated as CDK inhibitors. In the purine series,
NU2058 was the most potent compound with a mean
GI50 value of 13 ( 7 µM. The compound lacking the
2-NH2 substituent (NU2017, Chart 2a) was less potent
than NU2058, a result consistent with the reduced
activity of the compound as a CDK1 inhibitor (Table 2).
The N-9 methyl derivative of NU2058 (NU6052, Chart
2b), which cannot by definition form a purine N-9 to Glu
81 hydrogen bond and did not inhibit CDK1 or -2 in the
isolated enzyme assays, was not an inhibitor of cell
growth in most of the cell lines (GI50 > 100 µM in 50/57
cell lines studied). Similar in vitro growth inhibition
studies were performed with the pyrimidines NU6027
(Chart 2c) and NU6034 (Chart 2d). NU6027 was growth
inhibitory with mean GI50 values of 10 ( 6 µM whereas
NU6034, which cannot form an intramolecular hydro-
gen bond in order to orientate correctly the 6-NH2
substituent, was not growth inhibitory (GI50 > 100 µM
in 49/57 cell lines studied).
segment is not observed in all guanine- and pyrimidine-
based CDK2/inhibitor complex structures (results not
shown). For example, binding of NU6027 to CDK2 does
not result in ordering of these regions (Figure 3b).
On the basis of the structure shown in Figure 1, two
additional O6-cyclohexylmethylguanine derivatives were
prepared in order to determine the relative importance
of the hydrogen bonds at the 2-NH2 and N-9 positions
of NU2058. These were O6-cyclohexylmethylpurine
(NU2017, Chart 2a) and O6-cyclohexylmethyl-N9-meth-
ylguanine (NU6052, Chart 2b). As shown in Table 2,
both compounds were less active than NU2058 as
inhibitors of CDK1 and/or CDK2, implying that the
hydrogen bonds formed by the 2-NH2 and N-9 functions
of NU2058 can both contribute to the potency of the
molecule. Of the two NU2058 derivatives NU2017
retained activity, particularly against CDK2, indicating
that the hydrogen bond formed by the 2-NH2 is not
indispensable. The CDK2/NU2017 structure confirmed
that NU2017 binds to CDK2 within the ATP binding
site in the same orientation as NU2058 (data not
shown).
In an attempt to reproduce the triplet of hydrogen
bonds observed for NU2058 (Figure 1b) in a molecule
lacking the purine nucleus, the pyrimidine NU6027 (2,6-
diamino-O4-cyclohexylmethyloxy-5-nitrosopyrimidine,
Chart 2c) was prepared. In designing NU6027, note was
taken of the need to maintain the 6-amino group
(equivalent to N-9 in NU2058) in the correct orientation
to enable hydrogen bonding with the peptide oxygen of
Glu-81. To do so, a nitroso group was introduced at the
5-position in the expectation that the 5-nitroso group
would form an intramolecular hydrogen bond with
the 6-amino group, thereby establishing the correct
geometry for the remaining hydrogen atom on the
A COMPARE analysis was performed against the
standard anticancer agent database for both NU2058
and NU6027. Neither compound generated Pearson
correlation coefficients of >0.5 for the GI50 COMPARE
analysis, and hence these CDK inhibitors have a pattern
of cellular activity distinct from that of known classes
of antitumor agents. Interestingly, the COMPARE
analysis for the relationship between the GI50 sensitivity
profiles of NU2058 and NU6027 versus flavopiridol and

2802 J ournal of Medicinal Chemistry, 2000, Vol. 43, No. 15
Arris et al.
F igu r e 3. (a, b) Ribbon diagrams of CDK2/inhibitor complexes colored by temperature factor such that deep blue corresponds to
a temperature factor at least 10 Å2 less than the average main chain B-factor and deep red corresponds to a temperature factor
greater than or equal to 40 Å2 greater than the average main chain B-factor. The inhibitors are shown in ball-and-stick
representation. (a) CDK2/NU2058 complex, (b) CDK2/NU6027 complex. (c) The shape and surface properties of the CDK2 ATP
binding site cleft as seen in complex with NU2058. The Connolly molecular surface of CDK2 is rendered looking onto the N-terminal
domain at the surface of the glycine-loop sequence to show the hydrophobic region that interacts with the NU2058 O6-cyclohexyl
group.45 The surface is colored such that regions showing a preference for hydrophobic interactions (as calculated by the program
GRID46) range from yellow (strong preference) to gray (no particular preference). A hydrophobic patch generated by the side
chain of Val 18 and the peptide backbone around Gly 11 is apparent.
olomoucine also yielded low Pearson correlation coef-
ficients, i.e., 0.22-0.38. These low values are in contrast
to the value for the relationship between flavopiridol
and olomoucine (0.57), and NU2058 and NU6027 (0.65).
Together with the enzyme inhibition and structural
studies, these in vitro chemosensitivity data suggest
that the distinct mode of binding of NU2058 and
NU6027 to CDKs has direct consequences in terms of
cell growth inhibition and differential activity with
respect to other CDK inhibitors.
CDKs or are CDK substrates (e.g., retinoblastoma
protein). Preclinical11,12 and more recently clinical10
studies with the prototype CDK inhibitor flavopiridol
have shown that CDK inhibitors can have single agent
antitumor activity, and initial combination studies with
cytotoxic drugs are encouraging.44
The first generation of CDK inhibitors, exemplified
by flavopiridol and olomoucine, have either limited
specificity for individual CDKs (flavopiridol13) or rela-
tively poor potency as inhibitors of the CDK enzymes
and tumor cell growth (olomoucine17,18). Hence a num-
ber of groups have initiated analogue development pro-
grams focused largely on the 6-aminopurine pharma-
cophore.17-24 These latter studies have shown that it
The CDKs are prominent among novel targets for
cancer therapy because of the frequency of abnormali-
ties in genes which encode proteins that either directly
(e.g., p 16, cyclin D) or indirectly (e.g., p 53) control

Purine and Pyrimidine CDK Inhibitors
J ournal of Medicinal Chemistry, 2000, Vol. 43, No. 15 2803
to Glu 81 and purine 2-NH2 to Leu 83. The inhibitors
so produced had either reduced activity against CDK1
(NU2017) or no activity (NU6052) at the concentrations
tested. Importantly, the same relative effects were seen
in tumor cell growth inhibition studies, i.e., reduced
activity with NU2017 and no activity with NU6052,
suggesting that growth inhibition in the whole cell
activity was indeed a reflection of kinase inhibition.
Detailed studies of the cellular pharmacology of NU2058
and NU6027, in comparison to olomoucine, are currently
ongoing. Preliminary studies in MCF7 cells have shown
that, consistent with inhibition of both CDK1 and
CDK2, treatment with NU2058 and NU6027 causes a
reduction in the number of cells in S-phase but not G1
or G2/M (J . Bentley and C. Arris, unpublished results).
An important feature of protein structure-based drug
design is the ability to predict novel pharmacophores
once important sites for inhibitor-protein interactions
have been identified. In the current study, identification
of the triplet of hydrogen bonds formed between NU2058
and the ATP binding site of CDK2 led to the design of
NU6027. Further detailed structure-activity studies
based on both the guanine and pyrimidine pharmaco-
phores are underway and will be described elsewhere.
F igu r e 4. NU6027 bound to CDK2. Conserved hydrogen
bonds between the CDK2 backbone at residues Glu 81 and
Leu 83 and NU6027 are illustrated by thin lines. The (2Fo
-
Fc)Rcalc electron density for the CDK/NU6027 complex in-
cluded in the figure was calculated at the end of the refinement
(Rconv ) 21.3%, Rfree ) 28.1%) using map coefficients output
from REFMAC with resolution between 20.0 and 1.85 Å. The
map is contoured at a level of 0.3 eÅ-3, corresponding to 1.1
times the root-mean-square deviation of the map from the
mean value. The intramolecular bond between the 5-nitroso
group and the 6-amino group establishes the correct geometry
for the 6-amino group to act as a hydrogen bond donor to the
backbone oxygen atom of Glu 81.
Ack n ow led gm en t . The authors thank David
Morgan for his generous gift of the AcCDK2 baculovirus
and Tim Hunt for the cyclin A3 expression construct.
The staff at station 9.5, Daresbury, beamline BM14,
ESRF, and X-RAY DIFFRACTION, Elettra, provided
excellent facilities and advice during data collection.
At the LMB the authors thank I. Taylor, R. Bryan,
Y. Huang, K. Measures, and S. Lee. The authors are
also grateful to Dr. Hans Hendriks of the New Drug
Development Office (Amsterdam, The Netherlands) for
coordinating the initial screening of compounds by Dr.
Laurent Meijer and Ms. Sophie LeClerq (CNRS, Roscoff,
France) to whom they are also indebted. The assistance
of Ms. Susanne Radkte (NCI Liaison Office, Brussels)
and Ms. Ronnie Verdon and Dr. Sally Burtles (CRC
Drug Development Office, London, U.K) in facilitating
the evaluation of compounds in the NCI cell line screen
is also much appreciated. This research was supported
by grants from the Cancer Research Campaign, The
Royal Society, The Medical Research Council, and the
BBSRC.
is possible to develop CDK inhibitors with nanomolar
potency as CDK inhibitors and mean GI50 values in
the NCI cell line screen, or IC50 values for other
human tumor cell lines, in the low micromolar range.
The challenge in the area of small molecule CDK
inhibitor development is currently to identify com-
pounds with improved specificity for individual CDKs,
coupled with greater in vitro potency against tumor cell
lines. To meet this challenge, additional CDK inhibitor
pharmacophores would be valuable, and the current
report describes the identification of the guanine phar-
macophore and novel pyrimidine-based inhibitors de-
rived from a knowledge of the binding of the guanine-
based inhibitors to CDK2. The lead compounds in the
two series (NU2058 and NU6027) display CDK inhibi-
tory potency similar to that of olomoucine, which has
IC50 values against cyclinB/CDK1 and cyclinA or E/CDK2
of 7 µM,17 and activity against cell lines equivalent
or superior to that of roscovitine (the mean GI50 of
roscovitine in the NCI cell line screen is 16 µM18). These
data suggest that NU2058 and NU6027 may be able to
enter cells more readily than the 6-aminopurine-based
inhibitors.
Refer en ces
(1) Pines, J . Cyclins and cyclin-dependent kinases: A biochemical
view. Biochem. J . 1995, 308, 697-711.
(2) Chevalier, S.; Blow, J . J . Cell cycle control of replication initia-
tion in eukaryotes. Curr. Opin. Cell Biol. 1996, 8, 815-821.
(3) Nurse, P.; Masui, Y.; Hartwell, L. Understanding the cell cycle.
Nature Med. 1998, 4, 1103-1106.
High-throughput cell line screening and protein struc-
ture-based drug design are two of the most important
contemporary approaches to drug discovery. However,
it can be difficult to show that effects in whole cells of
chemical changes to inhibitors, introduced on the basis
of information from inhibitor-protein structures, are in
fact due to interactions with the intended target of the
compound. The approach taken here was to synthesize
compounds designed to delete what appeared, from the
NU2058-CDK2 structure, to be critical hydrogen bond
interactions, namely, the bonds between purine NH-9
(4) Sherr, C. J . Cancer cell cycles. Science 1996, 274, 1672-1677.
(5) Morgan, D. O. Cyclin-dependent kinases: Engines, Clocks and
Microprocessors. Annu. Rev. Cell Dev. Biol. 1997, 13, 261-291.
(6) Hall, M.; Peters, G. Genetics alterations of cyclins, cyclin-
dependent dependent kinases, and Cdk inhibitors in human
cancer. Adv. Cancer Res. 1996, 68, 67-108.
(7) Gray, N.; Detivaud, L.; Doerig, D.; Meijer, L. ATP-site inhibitors
of cyclin-dependent kinases. Curr. Med. Chem. 1999, 6, 859-
875.
(8) Walker, D. H. Small-molecule inhibitors of cyclin-dependent
kinases: Molecular tools and potential therapeutics. Curr. Top.
Microbiol. Immunol. 1998, 227, 149-165.
(9) Garrett, M. D.; Fattaey, A. CDK inhibition and cancer therapy.
Curr. Opin. Genet. Dev. 1999, 9, 104-111.

2804 J ournal of Medicinal Chemistry, 2000, Vol. 43, No. 15
Arris et al.
(10) 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.
(11) Drees, M.; Dengler, W. A.; Roth, T.; Labonte, H.; Mayo, J .;
Malspeis, L.; Grever, M.; Sausville, E. A.; Fiebig, H. H. Flavo-
piridol (L86-8275): Selective antitumour activity in vitro and
activity in vivo for prostate carcinoma cells. Clin. Cancer Res.
1997, 3, 2273-279.
(12) Arguello, F.; Alexander, M.; Sterry, J . A.; Tudor, G.; Smith, E.
M.; Kalavar, N. T.; Greene, J . F.; Koss, W.; Morgan, C. D.;
Stinson, S. F.; Siford, T. J .; Alvord, W. G.; Klabansky, R. L.;
Sausville, E. A. Flavopiridol induces apoptosis of normal lym-
phoid cells, causes immunosuppression, and has potent anti-
tumour activity in vivo against human leukaemia and lymphoma
xenografts. Blood 1998, 91, 2482-2490.
(13) Carlson, B. A.; Dunbay, M. M.; Sausville, E. A.; Brizuela, L.;
Worland, P. J . Flavopiridol induces G1 arrest with inhibition of
cyclin-dependent kinase (CDK) 2 and CDK4 in human breast
carcinoma cells. Cancer Res. 1996, 56, 2973-2978.
(14) Kaur, G.; Stetler-Stevenson, M.; Sebers, S.; Worland, P.;
Sedlacek, H.; Myers, C.; Czech, J .; Naik, R.; Sausville, E.
Growth inhibition with reversible cell cycle arrest of carcinoma
cells by flavone L86-8275. J NCI, J . Natl. Cancer Inst.1992, 84,
1736-1740.
(15) Schrump, D. S.; Matthews, W.; Che, G. A.; Mixon, A.; Altorki,
N. K. Flavopiridol mediates cell cycle arrest and apoptosis in
esophageal cancer cells. Clin. Cancer. Res. 1998, 4, 2885-2890.
(16) Parker, B. W.; Kaur, G.; Nieves-Neira, W.; Taimi, M.; Kohlhagen,
G.; Taimi, M.; Kohlhagen, G.; Shimizu, T.; Losiewicz, M. D.;
Pommier, Y.; Sausville, E. A.; Senderowicz, A. M. Early induction
of apoptosis in haematopoietic cell lines after exposure to
flavopiridol. Blood 1998, 91, 458-465.
(17) Vesely, 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.
(18) Meijer, L.; Borgne, A.; Mulner, O.; Chong, J . P. J .; Blow, J . J .;
Inagaki, N.; Inagaki, M.; Delcros, J .-L.; Moulinoux, J .-P. Bio-
chemical and cellular effects of roscovitine, a potent and selective
inhibitor of cyclin-dependent kinases cdc2, cdk2 and cdk5. Eur.
J . Biochem. 1997, 243, 527-536.
(19) Brooks, E. E.; Gray, N. S.; J oly, A.; Kerwar, S. S.; Lum, R.;
Mackman, R. L.; Norma, T. C.; Rosete, J .; Rowe, M.; Schow, S.
R.; Schultz, P. G.; Wang, X.; Wick, M. M.; Schiffman, D. CVT-
313, a specific and potent inhibitor of CDK2 that prevents
neointimal proliferation. J . Biol. Chem. 1997, 272, 29207-29211.
(20) Schow, S. R.; Mackman, R. L.; Blum, C. L.; Brooks, E.; Horsma,
A. G.; J oly, A.; Kerwar, S. S.; Lee, G.; Schiffman, 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.
(21) Gray, N. S.; Wodicka, L.; Thunnissen, A.-M. W. H.; 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 for kinase inhibitors. Science 1998, 281, 533-538.
(22) 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.
(23) 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.
(24) Chang, Y.-T.; Gray, N. S.; Rosania, G. R.; Sutherlin, D. P.; Kwon,
S.; Norman, T. C.; Sarohia, R.; Leost, M.; Meijer, M.; Schultz,
P. G. Synthesis and application of functionally diverse 2,6,9-
trisubstituted purine libraries as CDK inhibitors. Chem. Biol.
1999, 6, 361-375.
(26) de Azevedo, W. F., J r.; 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.
(27) Arris, C. E.; Bleasdale, C.; Calvert, A. H.; Curtin, N. J .; Dalby,
C.; Golding, B. T.; Griffin, R. J .; Lunn, J . M.; Major, G. N.;
Newell, D. R. Probing the active site and mechanism of
action of O6-alkylguanine-DNA alkyltransferase with substrate
analogues. Anticancer Drug Design 1994, 9, 401-408.
(28) Lembicz, N. K.; Grant, S.; Clegg, W.; Griffin, R. J .; Heath, S. L.;
Golding, B. T. Facilitation of displacements at the 6-position of
purines by the use of 1,4-diazabicyclo[2,2,2]octane as leaving
group. J . Chem. Soc., Perkin Trans. 1 1997, 185-186.
(29) Pfleiderer, W.; Lohrmann, R. Synthesis of 2-amino-4-alkoxy-
pteridine. Chem. Ber. 1961, 94, 12-18.
(30) 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 CDK2 structure and
activity. J . Biol. Chem. 1999, 274, 8746-8756.
(31) Lawrie, A. M.; Noble, M. E. M.; Tunnah, P.; Brown, N. R.;
J ohnson, L. N.; Endicott, J . A. Protein kinase inhibition by
staurosporine: Details of the molecular interaction determined
by the X-ray crystallographic analysis of a CDK2-staurosporine
complex. Nature Struct. Biol. 1997, 4, 796-802.
(32) Otwinowski, Z. Oscillation data reduction program. In DL/SC1/
R34; Sawyer, L., Isaacs, N., Bailey, S., Eds.; SERC Laboratory:
Daresbury, Warrington, U.K., 1993; pp 56-62.
(33) 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.
(34) Sybyl Molecular Modelling Software; Tripos Associates Inc.:
St. Louis, MO, 1992.
(35) Murshudov, G. N.; Vagen, A. A.; Dodson, E. J . Refinement of
macromolecular structures by the maximum-likelihood method.
Acta Crystallogr. 1997, D53, 240-255.
(36) Collaborative Computational Project, Number 4. The CCP4
suite: programs for protein crystallography. Acta Crystallogr.
1994, D50, 760-763.
(37) Lamzin, V. S.; Wilson, K. S. Automated refinement of protein
models. Acta Crystallogr. 1993, D49, 129-147.
(38) Monks, A.; Scudiero, D.; Skehan, P.; Shoemaker, R.; Paull, K.;
Vistica, D.; Hose, C.; Langley, J .; Cronise, P.; Vaigro-Wolff, A.;
Gray-Goodrich, M.; Campbell, H.; Mayo, J .; Boyd, M. Feasibility
of a high-flux anticancer drug screen using a diverse panel of
cultured human tumor cell lines. J NCI, J . Natl. Cancer Inst.
1991, 83, 757-766.
(39) Paull, K. D.; Shoemaker, R. H.; Hodes, L.; Monks, A.; Scudiero,
D. A.; Rubinstein, L.; Plowman, J .; Boyd, M. R. Display and
analysis of differential activity of drugs against human
tumor cell lines: Development of mean graph and COMPARE
algorithm. J NCI, J . Natl. Cancer Inst. 1989, 81, 1088-1092.
(40) 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.
(41) de Azevedo, W., 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-
2740.
(42) Hoessel, R.; Leclerc, S.; Endicott, J . A.; Noble, M. E. 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 antileukaemia medicine,
inhibits cyclin-dependent kinases. Nature Cell Biol. 1999, 1,
60-67.
(43) 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.
(44) Ongkeko, W.; Ferguson, D. J . P.; Harris, A. L.; Norbury, C.
Inactivation of Cdc2 increases the level of apoptosis induced by
DNA damage. J . Cell Sci. 1995, 108, 2897-2904.
(45) Connolly, M. Molecular surface triangulation. J . Appl. Crystal-
logr. 1985, 18, 499-505.
(25) 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 structure of cyclin-dependent protein kinase
2 in complex with ATP and two inhibitors, olomoucine and
isopentenyladenine. Proteins 1995, 22, 378-391.
(46) Goodford, P. Multivariate characterization of molecules for
QSAR. J . Chemom. 1996, 10, 107-117.
J M990628O