Discovery of N6-phenyl-1H-pyrazolo[3,4-d]pyrimidine-3,6-diamine derivatives as novel CK1 inhibitors using common-feature pharmacophore model based virtual screening and hit-to-lead optimization

Article abstract of DOI:10.1016/j.ejmech.2012.08.007

Aberrant activation of casein kinase 1 (CK1) has been demonstrated to be implicated in the pathogenesis of cancer and various central nervous system disorders. Discovery of CK1 inhibitors has thus attracted much attention in recent years. In this account, we describe the discovery of N6-phenyl-1H- pyrazolo[3,4-d]pyrimidine-3,6-diamine derivatives as novel CK1 inhibitors. An optimal common-feature pharmacophore hypothesis, termed Hypo2, was firstly generated, followed by virtual screening using Hypo2 against several chemical databases. One of the best hit compounds, N6-(4-chlorophenyl)-1H-pyrazolo[3,4-d] pyrimidine-3,6-diamine, was chosen for the subsequent hit-to-lead optimization under the guide of Hypo2, which led to the discovery of a new lead compound (1-(3-(3-amino-1H-pyrazolo[3,4-d]pyrimidin-6-ylamino)phenyl)-3-(3-chloro-4- fluorophenyl)urea) that potently inhibits CK1 with an IC₅₀ value of 78 nM.

Full text of DOI:10.1016/j.ejmech.2012.08.007

European Journal of Medicinal Chemistry 56 (2012) 30e38
Contents lists available at SciVerse ScienceDirect
European Journal of Medicinal Chemistry
journal homepage: http://www.elsevier.com/locate/ejmech
Original article
Discovery of N6-phenyl-1H-pyrazolo[3,4-d]pyrimidine-3,6-diamine derivatives as
novel CK1 inhibitors using common-feature pharmacophore model based virtual
screening and hit-to-lead optimization
,
b
,
,
,
Ling-Ling Yang a 1, Guo-Bo Li a 1, Heng-Xiu Yan a 1, Qi-Zheng Sun a, Shuang Ma a, Pan Ji a, Ze-Rong Wang a,
Shan Feng a, Jun Zou a, Sheng-Yong Yang a
,
*
a State Key Laboratory of Biotherapy and Cancer Center, West China Hospital, West China Medical School, Sichuan University, #1 Keyuan Road 4, Sichuan 610041, China
b College of Chemical Engineering, Sichuan University, Sichuan 610041, China
a r t i c l e i n f o
a b s t r a c t
Article history:
Aberrant activation of casein kinase 1 (CK1) has been demonstrated to be implicated in the pathogenesis
of cancer and various central nervous system disorders. Discovery of CK1 inhibitors has thus attracted
much attention in recent years. In this account, we describe the discovery of N6-phenyl-1H-pyrazolo[3,4-
d]pyrimidine-3,6-diamine derivatives as novel CK1 inhibitors. An optimal common-feature pharmaco-
phore hypothesis, termed Hypo2, was firstly generated, followed by virtual screening using Hypo2
against several chemical databases. One of the best hit compounds, N6-(4-chlorophenyl)-1H-pyrazolo
[3,4-d]pyrimidine-3,6-diamine, was chosen for the subsequent hit-to-lead optimization under the guide
of Hypo2, which led to the discovery of a new lead compound (1-(3-(3-amino-1H-pyrazolo[3,4-d]pyr-
imidin-6-ylamino)phenyl)-3-(3-chloro-4-fluorophenyl)urea) that potently inhibits CK1 with an IC50
value of 78 nM.
Received 3 February 2012
Received in revised form
2 August 2012
Accepted 3 August 2012
Available online 14 August 2012
Keywords:
Casein kinase 1
Pharmacophore
Virtual screening
Lead optimization
Cancer
Ó 2012 Elsevier Masson SAS. All rights reserved.
Central nervous system disorders
1. Introduction
inhibitors have been thought as promising interfering agents in the
treatment and prevention of these diseases.
Protein kinase casein kinase 1 (CK1) belongs to the serine/
threonine kinase superfamily that functions as a regulator of signal
transduction pathways in most eukaryotic cell types [1]. CK1 iso-
forms, which are highly conserved in their kinase domains (53%e
98% identical) but significantly differ in their regulatory C-
terminal region, are involved in Wnt signaling, circadian rhythms,
nucleo-cytoplasmic shuttling of transcription factors, DNA repair,
and DNA transcription [1e3]. Aberrant functional regulation of CK1,
such as its over-expression or excessive activation, has been
demonstrated to be implicated in the pathogenesis of many
diseases including Alzheimer’s disease [4], Parkinson’s disease [5],
sleep disorders [6], inflammation [7] and cancers [8]. Thus CK1
Due to the potential application of CK1 inhibitors in various
diseases, the development of CK1 inhibitors have attracted much
attention in recent years. Though some CK1 inhibitors have been
reported, there is no CK1 inhibitor in clinical studies so far. Thus,
discovering more potent CK1 inhibitors is still needed, which could
provide more lead candidates for drug development. Formerly, the
discovery of CK1 inhibitors was achieved through high-throughput
screening (HTS) [9] and subsequent experience-based structural
optimization, which usually suffered a high cost and a low success
rate. Currently, the fast developing computer-aided drug discovery
methods, including virtual screening [10e13] and quantitative
structureeactivity relationship (QSAR)-guided structural optimi-
zation [14e16], provide economic and rapid approaches to the lead
discovery. In this investigation, we shall describe the discovery of
N6-phenyl-1H-pyrazolo[3,4-d]pyrimidine-3,6-diamine derivatives
as novel CK1 inhibitors, which were obtained through virtual
screening based on a common-feature pharmacophore model of
CK1 inhibitors, and subsequent pharmacophore model guided hit-
to-lead optimization.
Abbreviations: CK1, casein kinase 1; VS, virtual screening; HTS, high-throughput
screening; QSAR, quantitative structureeactivity relationship; PDB, protein data
bank; HA, hydrogen-bond acceptor; HD, hydrogen-bond donor; H, hydrophobic
feature; DS, discovery studio.
* Corresponding author. Tel.: þ86 28 85164063; fax: þ86 28 85164060.
E-mail address: yangsy@scu.edu.cn (S.-Y. Yang).
These authors contributed to this work equally.
1
0223-5234/$ e see front matter Ó 2012 Elsevier Masson SAS. All rights reserved.
http://dx.doi.org/10.1016/j.ejmech.2012.08.007

L.-L. Yang et al. / European Journal of Medicinal Chemistry 56 (2012) 30e38
31
2. Results and discussion
(DS) 3.1 program package. After the running of program, 10 phar-
macophore hypotheses were created for each complex with the top
ranking one being assigned as the final pharmacophore hypothesis
for the corresponding complex, which are shown in Fig. 1. Fourthly,
all of the eight pharmacophore hypotheses were superimposed
together. Chemical features with the same property and same
location in the three dimensional (3D) space were identified as
common features, which were used for constructing the overall
pharmacophore model with the aid of Hypoedit module within
Catalyst (AccelrysR) (Fig. 2A). We defined the average position of
the feature centers in each chemical feature cluster as the center of
common pharmacophore feature. The tolerances were set to the
default values for corresponding features in Catalyst. The finally
obtained common features are shown in Fig. 2A, which includes
one hydrogen-bond acceptor (HA1), two hydrogen-bond donors
(HD1 and HD2) and three hydrophobic features (H1, H2 and H3).
Since the two features, HD1 and H1, approach closely (the distance
2.1. Generation of common-feature pharmacophore model of CK1
inhibitors
In this section, we shall describe the development of a common-
feature pharmacophore model of CK1 inhibitors. As one of the
major computer-aided drug discovery methods, the pharmaco-
phore method has been widely used in the virtual screening for
retrieving hit compounds. A pharmacophore model could be
developed by methods either ligand-based or complex-based [17].
Pharmacophore models developed by ligand-based methods could
involve information from a broad range of active compounds con-
tained in a so-called training set [18e20]. Nevertheless, the
generated pharmacophore models might not correctly reflect the
interaction mode between the ligands and target, which stems
from the possible defects of algorithms for the ligand conforma-
tional generation. On the contrary, pharmacophore hypotheses
built by complex-based methods are often directly derived from the
interaction mode between the ligand and target in one complex,
which inevitably leads to that the generated pharmacophore model
contains very limited ligand information [17,21]. Apparently an
ideal pharmacophore model should not only properly reflect the
interaction mode between the ligand and receptor, but also contain
information from more active compounds. Thus, in order to
develop such an ideal pharmacophore model, we proposed a new
pharmacophore modeling protocol for CK1 inhibitors, which is
described as follows.
Firstly, all the publicly reported CK1 inhibitors were collected
from literature [22e28]. Eight representative CK1 inhibitors were
chosen to form a training set (see Chart 1); these compounds were
selected since they are the most active compounds and each one of
them possesses a different scaffold. Secondly, each compound in
the training set was docked to the active site of CK1 (PDB: 2CSN) to
form a ligandereceptor complex (see Fig. 1). Here the GOLD
program was used since it is one of the best programs for flexible
docking calculations [29]. Thirdly, each complex was taken as the
input to create pharmacophore hypotheses by using the ‘Receptor-
Ligand Pharmacophore Generation’ module in Discovery Studio
ꢀ
between their centers is less than 1 A), we thus constructed two
five-feature pharmacophore hypotheses, Hypo1 and Hypo2 (Fig. 2B
and C), with the other four features combined with either HD1 or
H1, which is for the convenience of virtual screening.
2.2. Evaluations of the performances of Hypo1 and Hypo2 in virtual
screening
In order to evaluate the performances of Hypo1 and Hypo2 in
virtual screening, we built a test set, called M-ZINC. M-ZINC
contains 110 known CK1 inhibitors (see Supplementary Table S1)
and 3300 decoys [30] (the selection method for the decoys is given
in the Supplementary data) from the ZINC chemical database [31].
The performance of virtual screening, including the yield
(percentage of predicted compounds in known inhibitors), hit rate
(percentage of known inhibitors in predicted compounds), and
enrichment factor (ratio of hit rate to the percentage of known
inhibitors in M-ZINC database), were evaluated.
The screening results are given in Table 1. From Table 1, we can
see that the number of the predicted positives (Fitvalue ꢀ 3.0) is
661, and that of the hits is 42 with a yield of 38.18% (42 out of the
110 inhibitors) for the use of Hypo1. The hit rate and enrichment
Fig. 1. The docking poses and pharmacophore models of eight known CK1 inhibitors in the crystal structure of CK1 (PDB code: 2CSN). (A)e(H) the green sphere represents hydrogen
bond acceptor; puniceous represents the hydrogen bond donor; and light blue represents the hydrophobic feature. (For interpretation of the references to colour in this figure
legend, the reader is referred to the web version of this article.)

32
L.-L. Yang et al. / European Journal of Medicinal Chemistry 56 (2012) 30e38
Fig. 2. Generation of common-feature pharmacophore models of CK1. (A) The superposition plot of the eight pharmacophore models. Hydrophobic features in dashed cycles
represent the clustered hydrophobic features; (B) The pharmacophore model Hypo1; (C) The pharmacophore model Hypo2.
factor are 6.35% and 1.97, respectively. For Hypo2, the number of the
predicted positives is 101, including 27 hits and 74 decoys; the yield
is 24.55% (27 out of 110 inhibitors). The hit rate and enrichment
factor for Hypo2 are 26.73% and 8.29, respectively. Here one may
think that combined use of Hypo1 and Hypo2 may have a better
performance compared with individual use of Hypo1 and Hypo2.
We thus examined the effect of their combination. The yield, hit
rate and enrichment factor are 41.82, 6.50 and 2.01, respectively. It
is obvious that, though the yield increased a little, the hit rate and
enrichment factor for the combined use of Hypo1 and Hypo2
drastically decreased compared with that for Hypo2. Considering
that the hit rate and enrichment factor are the most concerned of in
VS, Hypo2 was then chosen as the reference pharmacophore model
in the subsequent VS for retrieving novel CK1 inhibitors.
selected from the top-ranking compounds sorted by Fitvalue (see
Supplementary Table S2). Among them, the compound N6-(4-
chlorophenyl)- 1H-pyrazolo[3,4-d]pyrimidine-3,6-diamine (1,
Fig. 3A) attracted our attention due to that it possesses a novel
scaffold and has the smallest molecular weight.
Fig. 3A depicts the mapping of 1 with Hypo2. From Fig. 3A, we
can see that except one hydrophobic feature, the other four features
are mapped very well. This mapping provides some important
information, which could be used to guide the structural optimi-
zation of compound 1 (Fig. 3B). Firstly, a hydrophobic group could
be added through a ‘linker’ to chlorobenzene to match the hydro-
phobic feature. Secondly, according to the orientations of H3 and
H2, the “linker” added should be at the meta position (3-position)
other than the para position (4-position) of phenyl. Thirdly, since
the distance between the center of H2 and the center of H3 is about
ꢀ
8.5 A, the ‘linker’ chain should be in the range of 4e5 bonds’ length.
2.3. Virtual screening and hit-to-lead optimization based on Hypo2
Fourthly, there may exist a hydrogen bonding interaction between
the ‘linker’ and the ASP94 (Fig. 1E). Fifthly, it would be better that
the position of H3 hydrophobic feature is occupied by some
aromatic rings, which is common in the known inhibitors. Based on
these analyses, we designed 15 compounds, which are given in the
Supplementary Figure S1. In the present work, we just chose three
The optimal common feature pharmacophore model Hypo2 was
used to screen several chemical databases including the commer-
cial SPECS as well as an in-house chemical database to retrieve
potential CK1 inhibitors with novel scaffolds. The number of
maximum omitted features was set to 1. 50 compounds were finally
Fig. 3. Hit-to-lead optimization of compound 1 guided by Hypo2. (A) The mapping of the selected compound 1 from screening results to Hypo2; (B) The schematic of hit-to-lead
optimization.

L.-L. Yang et al. / European Journal of Medicinal Chemistry 56 (2012) 30e38
33
Fig. 4. The docking poses together with Hypo2 in the crystal structure of CK1 of (A) Compound 2, (B) Compound 3 and (C) Compound 4. The green dashed line represents hydrogen
bond interaction. This interaction was predicted by using Discovery Studio 3.1 (Accelrys, Inc., USA). (For interpretation of the references to colour in this figure legend, the reader is
referred to the web version of this article.)
representative compounds to synthesize since the main purpose
here is to test the correctness of the pharmacophore model and our
postulation. Synthesis of other compounds as well as further
structural optimization will be carried out in succeeding studies,
which results will be reported in the near future.
By the way, currently there are many rational design strategies
for structural optimization to improve the potency of kinase
inhibitors. For example, Liu and Gray [32] and Okram [33] have
suggested a design strategy for the generation of type-II kinase
inhibitors based on strategic placement of specific aryl amide or
urea substituents onto a type-I inhibitor; type-II kinase inhibitors
are generally thought to outperform corresponding type-I inhibi-
tors in terms of the potency as well as the selectivity. The structural
optimization strategy adopted here, which was based on and
guided by the established pharmacophore model, could be another
good option for the structural optimization of, not only kinase
inhibitors, but also other hit compounds.
2.4. Chemistry
We then synthesized the selected three compounds, namely 1-(3-
(3-amino-1H-pyrazolo[3,4-d]pyrimidin-6-ylamino)phenyl)-3-(3-chl-
oro-4-fluorophenyl)urea (2), (R)-1-(3-(3-amino-1H-pyrazolo[3,4-d]
pyrimidin-6-ylamino)phenyl)-3- (1-phenylethyl)urea (3), and 1-
(3-(3-amino-1H-pyrazolo[3,4-d]pyrimidin-6-ylamino) phenyl)-3-(3-
tert-butyl-1H-pyrazol-5-yl)urea (4). For comparison, we also synthe-
sized two additional compounds, 1-(4-(3-amino-1H-pyrazolo[3,4-d]
pyrimidin -6-ylamino)phenyl)-3-(3-chloro-4-fluorophenyl)urea (5),
and (R)-1-(4-(3-amino-1H- pyrazolo[3,4-d]pyrimidin-6-ylamino)
phenyl)-3-(1-phenylethyl)urea (6), in which the hydrophobic group is
attached to the 4-position of phenyl.
A general scheme for the synthesis of compounds 2e6 is shown
in Scheme 1. Reactions of commercially available 2-methyl-2-
thiopseudourea sulfate (2:1) (7) with (E)-ethyl-2-cyano-3-
ethoxyacrylate produced compound 8, which was then reacted
Chart 1. Structures and activities of CK1 inhibitors in the training set.

34
L.-L. Yang et al. / European Journal of Medicinal Chemistry 56 (2012) 30e38
Table 1
respectively. Compound 5 and 6, in which the hydrophobic groups
are attached on the 4-position of phenyl, have no activity
(IC50 > 10 mM). Fig. 4 depicts the interaction modes of 2, 3 and 4 in
Comparison of the performances of Hypo1, Hypo2 as well as the combination of
Hypo1 and Hypo2 in virtual screening against an independent test set, M-ZINC (110
CK1 inhibitors þ 3300 decoys).
the binding site of CK1, which shows that they all match very well
in shape with the active pocket of CK1. Compound 2 has the best
mapping with the hydrophobic feature (H3) compared with
compound 3 and 4. Compound 5 and 6 were also docked into the
active site of CK1, which showed that there were some collisions
occurred between 5/6 and residues of CK1, in addition to that H3
can not be mapped. These results are consistent with our analyses
given above.
We further examined the selectivity of the most active
compound. 12 common kinases, including ALK, Axl, mTOR, Plk1,
Met, EGFR, CDK1, FGFR2, GSK3a, Flt3, PDGFRa, and KDR, were
chosen. The inhibition activity was expressed as the percentage of
the residual kinase activity with the concentration of the
compound fixed at 10 mM. Table 3 presents the inhibition activity of
Model
Predicted
positive
Number
of hits
Yield
(%)
Hit rate
(%)
Enrichment
factor
Hypo1
Hypo2
Hypo1þHypo2
661
101
708
42
27
46
38.18
24.55
41.82
6.35
26.73
6.50
1.97
8.29
2.01
with phosphorus oxychloride to give compound 9. Subsequently
compound 9 was under cyclization reaction by using hydrazine
hydrate to produce compound 10, followed by addition of diluted
hydrochloric acid to give compound 11. Then, compound 11 was
reacted with phosphorus oxychloride in the solvents of DMF and
DMA to give the compound 12. On the other side, the 3-nitroaniline
or 4-nitroaniline (13) was reacted with isocyanates bearing
different R1 substituents to give the compound 14, which was then
reduced by using iron powder to provide compound 15. Finally, the
target compounds 2e6 were obtained from the reactions of
compound 12 and 15 with different R1 substituents. More detailed
description regarding reaction conditions and materials are given
in Scheme 1 and Experimental section.
compound 2 against CK1 and 12 selected kinases. From Table 3, we
can see that compound 2 can completely inhibit the CK1 kinase
activity (the inhibition rate ¼ 100%) at the concentration of 10
mM.
And it has also an inhibition rate greater than 90% against several
other kinases including Met, EGFR, Flt3, and KDR, although it has
a very low activity or almost no activity against the remaining seven
kinases. All of these results indicate that compound 2 is not a good
selective CK1 inhibitor. The optimization of the selectivity for this
compound is still underway, for which results will be reported in
the near future.
2.5. In vitro CK1 kinase inhibitory assay
Five synthesized compounds (listed in Table 2) were submitted
to test their inhibitory potency against CK1. The IC50 values of these
compounds against CK1 are given in Table 2. From Table 2, we can
see that compound 2, 3, and 4, in which the hydrophobic groups are
linked to the 3-position of phenyl, have inhibition activity against
CK1 with IC50 values being 78 nM, 2470 nM, and 142 nM,
3. Conclusion
In this investigation, pharmacophore models of CK1 inhibitors
were developed using
a pharmacophore modeling protocol
Scheme 1. Reagents and conditions: (a) (E)-ethyl-2-cyano-3-ethoxyacrylate, K2CO3, EtOH, H2O, reflux, 12 h; (b) PhMe, Phosphorus oxychloride, reflux, 6 h; (c) (i) Hydrazine hydrate,
n-Butanol, rt, 30 min; (ii) n-Butanol, reflux, 5 h; (d) EtOH, HCl, 80 ꢁC, overnight; (e) Phosphorus oxychloride, DMF, DMA, 100 ꢁC, 2 h; (f) (i) Triphosgene, Et3N, THF, rt; (ii) R1-NH2,THF,
45 ꢁC, 2 h; (g) Fe, NH4Cl, EtOH, H2O, 1 h; (h) HCl, n-Butanol, 110 ꢁC, 5 h.

L.-L. Yang et al. / European Journal of Medicinal Chemistry 56 (2012) 30e38
35
Table 2
pyrimidine-3,6-diamine was chosen from the top-ranking hits for
further structural optimization since it is the smallest one and
possesses a novel scaffold. A simple structural optimization, which
was carried out under the guide of Hypo2, led to three active
compounds with the highest one having an IC50 value of 78 nM
against CK1. Collectively, this investigation reports the discovery of
a lead compound targeting CK1 through a pharmacophore-based
virtual screening and subsequent pharmacophore model guided
structural optimization. This strategy used here could potentially
be applied to the hit and lead discovery for other targets.
The kinase inhibition activity against CK1 of the designed compounds 2e6.
4. Experimental section
4.1. Docking calculations
Compound
Position
3
R1
IC50 (nM)
78
All the docking calculations were performed using GOLD
(version 5.0), which is one of the best docking programs [34]. The
X-ray crystal structure of CK1 (PDB code 2CSN) [22] complexed
with ligand CKI-7 was used as the reference receptor structure in
the docking calculations. The preprocess of the protein structure
was carried out by using Discovery Studio 3.1 (Accelrys, Inc., USA)
software package, which includes removing all the water mole-
cules, adding hydrogen and assigning the CHARMm forcefield. Then
the binding site was defined as a sphere containing the residues
2
3
4
5
6
3
3
4
4
2470
ꢀ
that stay within 10 A around the ligand in this crystal structure.
Here we adopted the flexible docking strategy with the side chains
of Val153, Leu138, Ser91, Asn136, Asp154 and Lys41 being set to
flexible. The parameters used in the docking processes are given as
follows: the ‘Number of dockings’ was set to 20 without constraints
and early termination; the ‘Detect Cavity’ was turn on; ‘Flip Ring
Corners’ was used to explore the rings conformations; the ‘Flip
Amide Bonds’ was turn on; the planar of R-NRIR2 was flipped all
and the protonated carboxylic acids were fixed; the genetic algo-
rithm (GA) parameter was set as ‘GOLD Default’.
N
NH
142
>10,000
>10,000
4.2. Receptoreligand pharmacophore generation
The ‘Receptoreligand pharmacophore generation’ module of DS
3.1 program package was used to generate pharmacophore models.
The parameters used are given as follows: the number features
were delimited in the range of 3e6, and the ‘Maximum pharma-
cophore’ was set to 10; the option of ‘Keep Water Molecules’ and
‘Add Shape Constraint’ were both set to false; the maximum
distances of charge, hydrogen bond and hypophobic group were set
to 5.6, 3.2 and 5.5, respectively, and the minimum inter-feature
distance was set to 2.0; the ‘Check Pharmacophore’ was turn on;
and the others keep as default.
proposed by us. The optimal pharmacophore hypothesis, Hypo2,
was adopted to retrieve potential CK1 inhibitors from the
commercial chemical database SPECS and an in-house chemical
database. Compound N6-(4-chlorophenyl)-1H-pyrazolo[3,4-d]
4.3. Chemistry
All the reagents were purchased from market and were used
without further purification unless otherwise indicated. Analytical
thin layer chromatography (TLC) was performed using aluminum
sheet silica gel Merck 60F254, and the spots being developed with
UV light. All the final compounds were purified to >95% purity, as
determined by high-performance liquid chromatography (HPLC). 1
HNMR and 13C NMR spectra were recorded on a Bruker AV-400
(400 MHz) spectrometer at 400 MHz and 100 MHz, respectively.
Table 3
The kinase inhibition activity of compound 2.
Kinase
Inhibition (%) @10 mM
CK1
ALK
Axl
mTOR
Plk1
Met
EGFR
CDK1
FGFR2
100
26
46
0
0
99
92
42
35
34
90
82
96
Chemical shifts (d) are quoted in parts per million (ppm) relative to
tetramethylsilane (TMS) as an internal standard. Multiplicities are
given as s (singlet), d (doublet), dd (doubleedoublet), q (quadru-
plet), t (triplet), and m (multiplet). Low-resolution and High-
resolution mass spectral (MS) data were determined on an Agi-
lent 1100 Series LCMS with UV detection at 254 nm and a low
resonance electrospray mode (ESI).
GSK3
Flt3
a
PDGFR
KDR
a

36
L.-L. Yang et al. / European Journal of Medicinal Chemistry 56 (2012) 30e38
4.3.1. 4-Hydroxy-2-(methylthio)pyrimidine-5-carbonitrile (8)
To a solution of 2-Methyl-2-thiopseudourea sulfate (19 g,
100 mmol) and (E)-ethyl 2-cyano-3-ethoxyacrylate (17 g,
100 mmol) in ethanol (380 ml), the K2CO3 solution (276 g,
200 mmol in 100 mL water) was added at room temperature. The
reaction mixture was stirred at 85 ꢁC overnight. The solvents were
distilled on a rota-evaporator. The crude mixture was dissolved in
water (100 ml) and adjusted the pH ¼ 7e8 by using 6 N HCl. During
the acidification, a buff solid was formed and collected by filtration.
The filter cake was washed with a small amount of water and dried
in a vacuum oven to give the product 8 (10.8 g). Yield: 65%; Purity:
92%; LC-MS m/z: 168.0 [M þ H]þ.
10 mmol) in THF (10 mL) by using the constant pressure dropping
funnel. NEt3 (3 mL, 21 mmol) was then added slowly to the reaction
mixture after the nitroaniline (13) was added completely. After
evaporation of the organic solvent, THF (80 ml) and another amine
(10 mmol) were added directly to the residue. The reaction mixture
was stirred at 45ꢁCfor 2 h. The solvents were distilled on a rota-
evaporator and the crude mixture was washed with water and
acetone to give the product 14 (yield: 48%e71%).
The compound 14 (5 mmol) was taken in EtOH (30 mL), and iron
powder (1.4 g, 25 mmol) was added at 50ꢁC-55 ꢁC followed by
NH4Cl solution (133 mg, 2.5 mmol in 15 mL water). The reaction
mixture was refluxed for 1 h. Most of the iron powder was filtered
while hot, the filtrate was concentrated under reduced pressure.
The residue was dissolved in water (10 ml) basified with NaHCO3
solution (pH 7e8) and extracted several times with ethyl acetate,
the combined organic extracts were dried (Na2SO4), and concen-
trated under reduced pressure to furnish 9 (yield about 90%). This
material was taken up for the next step without any purification.
4.3.2. 4-Chloro-2-(methylthio)pyrimidine-5-carbonitrile (9)
To a cooled solution of 4-hydroxy-2-(methylthio)pyrimidine-5-
carbonitrile (8) (10.8 g, 65 mmol) in toluene (150 ml), phosphorus
oxychloride (23.8 ml, 260 mmol) was slowly added. The reaction
mixture was stirred at 110 ꢁC for 6 h. After evaporation of the
organic solvent, the residue was cooled to room temperature and
added a lot of ice water. At this time, a brown solid was formed and
collected by filtration. The crude product was dried in a vacuum
oven to give 4-chloro-2-(methylthio)pyrimidine -5-carbonitrile
9(6.8 g). This product was taken up for the next step without any
purification. Yield: 58%; Purity: 90%; LC-MS m/z: 186.0 [M þ H]þ.
4.3.7. 1-(3-(3-Amino-1H-pyrazolo[3,4-d]pyrimidin-6-ylamino)
phenyl)-3-(3-chloro-4-fluorophenyl)urea (2)
6
N HCl (0.8 ml) was added to the mixture of 1-(3-
aminophenyl)-3-(3-chloro-4-fluorophenyl)urea (15) (1.2 g,
4.4 mmol) in n-Butanol (20 ml). The reaction mixture was stirred at
room temperature for 20 min. Then, a solution of (E)-N0-(6-chloro-
1H-pyrazolo[3,4-d]pyrimidin-3-yl)-N,N-dimethylformimidamide
(12) (1.0 g, 4.4 mmol) in n-Butanol (50 ml) was added into the
former mixture and heated to 100 ꢁC for 3.5 h. The solvent was
distilled on a rota-evaporator. The crude mixture was added 100 ml
water and the aqueous layer was extracted with ethyl acetate
(2 ꢂ 120 ml). Then the combined organic layers were dried
(Na2SO4) and concentrated. The residue was finally recrystallized
from ethanol, ethyl acetate and petroleum ether to give the product
2 (1.52 g). Yield: 83%; Purity: 97%; LC-MS m/z: 413.1 [M þ H ]þ.1H
4.3.3. 6-(methylthio)-1H-pyrazolo[3,4-d]pyrimidin-3-amine (10)
A mixture of 4-chloro-2-(methylthio)pyrimidine-5-carbonitrile
(9) (6.8 g, 36.5 mmol) and hydrazine hydrate (3.6 ml,73 mmol) in
n-Butanol (120 ml) was stirred at room temperature for 0.5 h. Then,
the solution was concentrated under reduced pressure. The residue
was dissolved in n-Butanol (150 ml) and heated to reflux for 5 h,
and the solution was concentrated again. The residue was washed
with a small amount of water to give the product 6-(methylthio)-
1H-pyrazolo[3,4-d]pyrimidin-3-amine (10) (6.05 g). Yield: 92%;
Purity: 92%; LC-MS m/z:182.0 [M þ H ]þ. 1H NMR (400 MHz, DMSO-
NMR (400 MHz, DMSO-d6): d11.77 (s, 1H), 9.56 (s, 1H), 8.88 (s, 1H),
d6):
d12.28 (s, 1H), 8.88 (s, 1H), 5.97 (s, 2H), 2.51 (s, 3H)ppm.
8.80 (s, 1H), 8.72 (s, 1H), 7.82 (d, J ¼ 3.2 Hz, 1H), 7.65 (s, 1H), 7.56 (s,
1H), 7.34e7.18 (m, 4H), 5.75 (s, 2H)ppm. 13C NMR (100 MHz, DMSO-
4.3.4. 3-Amino-1H-pyrazolo[3,4-d]pyrimidin-6-ol (11)
d6): d 158.3, 155.3, 152.5, 152.4, 148.6, 141.1, 139.4, 137.1, 128.6, 119.3,
A solution of 6-(methylthio)-1H-pyrazolo[3,4-d]pyrimidin-3-
amine (10) (6.05 g, 33.2 mmol) in ethanol (45 ml) was added 6 N
HCl (45 ml). The mixture was heated at 85 ꢁC overnight and then
cooled to room temperature, during which time a lot of yellow
precipitate was formed. The resulting solid was collected by
filtration to give the product 11 (4.04 g). Yield: 81%; Purity: 95%; LC-
118.4, 118.3, 116.9, 116.7, 113.3, 112.0, 109.4, 100.3 ppm.
4.3.8. 1-(4-(3-Amino-1H-pyrazolo[3,4-d]pyrimidin-6-ylamino)
phenyl)-3-(3-chloro-4-fluorophenyl)urea (5)
The title compound was synthesized from 1-(4-aminophenyl)-
3-(3-chloro-4- fluorophenyl)urea (15) (1.2 g, 4.4 mmol) and (E)-N0-
(6-chloro-1H-pyrazolo[3,4-d] pyrimidin-3-yl)-N,N-dimethylformi-
midamide (12) (1.0 g, 4.4 mmol) using a procedure similar to that of
2. Yield: 81%; Purity: 98%; LC-MS m/z: 413.1 [M þ H ]þ. 1H NMR
MS m/z:152.1 [M þ H ]þ. 1H NMR (400 MHz, DMSO-d6):
J ¼ 6.8 Hz, 2H), 9.29 (s, 1H), 8.66 (s, 2H)ppm.
d12.62 (d,
4.3.5. N0-(6-chloro-1H-pyrazolo[3,4-d]pyrimidin-3-yl)-N,N-
(400 MHz, DMSO-d6): d11.76 (s,1H), 9.58 (s,1H), 8.80 (s,1H), 8.52 (s,
dimethylformimidamide (12)
1H), 8.30 (s, 1H),7.82 (s, 1H), 7.60 (d, J ¼ 3.2 Hz, 2H), 7.34 (d,
Phosphorus oxychloride (25 ml) was added to a mixture of 3-
amino-1H-pyrazolo[3,4-d]pyrimidin-6-ol (11) (4.04 g, 26.7 mmol),
DMF (3 ml) and DMA (2 ml). The resulting mixture was heated at
100 ꢁC for 2 h. After cooling to room temperature, the redundant
phosphorus oxychloride was removed. The residue was poured into
ice-water and adjusted the pH to 7e8. Then the mixture was
extracted several times with CH2Cl2. The final combined organic
extracts were dried (Na2SO4) and concentrated under reduced
pressure. The residue was then crystallized from ethyl acetate and
petroleum ether to give the product 12 (3.25 g). Yield: 54%; Purity:
90%; LC-MS m/z: 225.1 [M þ H ]þ. 1H NMR (400 MHz, DMSO-d6):
J ¼ 4.4 Hz, 2H), 7.06 (d, J ¼ 3.2 Hz, 2H), 5.72 (s, 2H)ppm. 13C NMR
(100 MHz, DMSO-d6): d 158.2, 155.2, 152.7, 151.0, 148.6, 141.1, 137.2,
135.5, 133.1, 119.9, 119.5, 118.9, 118.3, 116.9, 116.7, 100.2 ppm.
4.3.9. (R)-1-(3-(3-amino-1H-pyrazolo[3,4-d]pyrimidin-6-ylamino)
phenyl)-3-(1-phenyl ethyl) urea (3)
The title compound was synthesized from (R)-1-(3-
aminophenyl)-3-(1-phenylethyl) urea (15) (1.1 g, 4.4 mmol) and
(E)-N0-(6-chloro-1H-pyrazolo[3,4-d]pyrimidin-3-yl)
-N,N-dime-
thylformimidamide (12) (1.0 g, 4.4 mmol) using a procedure similar
to that of 2. Yield: 79%; Purity: 97%; LC-MS m/z: 389.2 [M þ H ]þ. 1H
d13.25 (s, 1H), 9.10 (s, 1H), 8.34 (s, 1H), 3.12 (s, 3H), 3.02 (s, 3H)ppm.
NMR (400 MHz, DMSO-d6): d11.74 (s, 1H), 9.47 (s, 1H), 8.78 (s, 1H),
8.33 (s, 1H), 7.48 (d, J ¼ 6.0 Hz, 2H), 7.34 (d, J ¼ 6.0 Hz, 4H),
7.26e7.23 (m, 1H), 7.16e7.08 (m, 2H), 6.61 (d, J ¼ 8.0 Hz 1H), 5.73 (s,
2H), 4.83e4.80 (m, 1H), 1.38 (d, J ¼ 7.2 Hz, 3H)ppm. 13C NMR
4.3.6. General procedure for synthesis of compounds 15
The nitroaniline (13) (1.38 g, 10 mmol) dissolved in THF (80 ml)
was dropped slowly to a stirred solution of triphosgene (2.98 g,
(100 MHz, DMSO-d6): d 158.4, 155.3, 154.3, 152.4, 148.6,145.2, 140.9,

L.-L. Yang et al. / European Journal of Medicinal Chemistry 56 (2012) 30e38
37
140.4, 128.5, 128.3, 126.6, 125.8, 112.5, 111.2, 108.6, 100.4, 48.5,
References
23.1 ppm.
[1] U. Knippschild, A. Gocht, S. Wolff, N. Huber, J. Lohler, M. Stoter, The casein
kinase 1 family: participation in multiple cellular processes in eukaryotes, Cell.
Signal. 17 (2005) 675e689.
[2] O. Marin, V.H. Bustos, L. Cesaro, F. Meggio, M.A. Pagano, M. Antonelli,
C.C. Allende, L.A. Pinna, J.E. Allende, A noncanonical sequence phosphorylated
by casein kinase 1 in beta-catenin may play a role in casein kinase 1 targeting
of important signaling proteins, Proc. Natl. Acad. Sci. U. S. A. 100 (2003)
10193e10200.
[3] J.K. Cheong, D.M. Virshup, Casein kinase 1: complexity in the family, Int. J.
Biochem. Cell. Biol. 43 (2011) 465e469.
[4] D.P. Hanger, H.L. Byers, S. Wray, K.Y. Leung, M.J. Saxton, A. Seereeram,
C.H. Reynolds, M.A. Ward, B.H. Anderton, Novel phosphorylation sites in tau
from Alzheimer brain support a role for casein kinase 1 in disease patho-
genesis, J. Biol. Chem. 282 (2007) 23645e23654.
[5] D.I. Perez, C. Gil, A. Martinez, Protein kinases CK1 and CK2 as new targets for
neurodegenerative diseases, Med. Res. Rev. 31 (2011) 924e954.
[6] N.P. Shanware, J.A. Hutchinson, S.H. Kim, L. Zhan, M.J. Bowler, R.S. Tibbetts,
Casein kinase 1-dependent phosphorylation of familial advanced sleep phase
syndrome-associated residues controls PERIOD 2 stability, J. Biol. Chem. 286
(2011) 12766e12774.
4.3.10. (R)-1-(4-(3-amino-1H-pyrazolo[3,4-d]pyrimidin-6-
ylamino)phenyl)-3-(1-phenylethyl) urea (6)
The title compound was synthesized from (R)-1-(4-
aminophenyl)-3-(1-phenylethyl) urea (15) (1.1 g, 4.4 mmol) and
(E)-N0-(6-chloro-1H-pyrazolo[3,4-d]pyrimidin-3-yl)
-N,N-dime-
thylformimidamide (12) (1.0 g, 4.4 mmol) using a procedure similar
to that of 2. Yield: 79%; Purity: 98%; LC-MS m/z: 389.2 [M þ H ]þ. 1H
NMR (400 MHz, DMSO-d6): d11.68 (s, 1H), 9.38 (s, 1H), 8.74 (s, 1H),
8.23 (s, 1H), 7.63 (d, J ¼ 8.8 Hz, 2H), 7.34 (d, J ¼ 4.0 Hz, 4H),
7.27e7.23 (m, 3H), 6.54 (d, J ¼ 8.0 Hz, 1H), 5.69 (s, 2H), 4.83e4.80
(m, 1H), 1.37 (d, J ¼ 6.8 Hz, 3H)ppm.
4.3.11. 1-(3-(3-Amino-1H-pyrazolo[3,4-d]pyrimidin-6-ylamino)
phenyl)-3-(5-tert-butyl-1H-pyrazol-3-yl) urea (4)
[7] S. Luz, P. Kongsuphol, A.I. Mendes, F. Romeiras, M. Sousa, R. Schreiber,
P. Matos, P. Jordan, A. Mehta, M.D. Amaral, K. Kunzelmann, C.M. Farinha,
Contribution of casein kinase 2 and spleen tyrosine kinase to CFTR trafficking
and protein kinase A-induced activity, Mol. Cell. Biol. 31 (2011) 4392e4404.
[8] U. Knippschild, S. Wolff, G. Giamas, C. Brockschmidt, M. Wittau, P.U. Wurl,
T. Eismann, M. Stoter, The role of the casein kinase 1 (CK1) family in different
signaling pathways linked to cancer development, Onkologie 28 (2005) 508e514.
[9] P. Gribbon, A. Sewing, Fluorescence readouts in HTS: no gain without pain?
Drug Discov. Today 8 (2003) 1035e1043.
[10] D.B. Kitchen, H. Decornez, J.R. Furr, J. Bajorath, Docking and scoring in virtual
screening for drug discovery: methods and applications, Nat. Rev. Drug Dis-
cov. 3 (2004) 935e949.
[11] P.D. Lyne, Structure-based virtual screening: an overview, Drug Discov. Today
7 (2002) 1047e1055.
The title compound was synthesized from 1-(3-aminophenyl)-
3-(5-tert-butyl -1H-pyrazol-3-yl)urea (15) (1.2 g, 4.4 mmol) and
(E)-N0-(6-chloro-1H-pyrazolo[3,4-d]
pyrimidin-3-yl)-N,N-dime-
thylformimidamide (12) (1.0 g, 4.4 mmol) using a procedure similar
to that of 2. Yield: 79%, Purity: 98%; LC-MS m/z: 407.2 [M þ H ]þ. 1H
NMR (400 MHz, DMSO-d6): d11.98 (s, 1H), 11.80 (s, 1H), 9.65 (s,
1H),9.15 (b, 1H), 9.00 (s, 1H), 8.89 (s, 1H), 7.65 (s, 1H), 7.52 (d,
J ¼ 7.2 Hz, 1H), 7.21e7.15 (m, 2H), 5.80 (s, 2H),1.41 (s, 9H)ppm. 13C
NMR (100 MHz, DMSO-d6): d 158.3, 155.3, 152.9, 152.4, 152.0, 148.6,
147.4, 140.9, 139.6, 128.6, 113.3, 111.9, 109.4, 100.4, 90.6, 30.6,
29.9 ppm.
[12] G. Klebe, Virtual ligand screening: strategies, perspectives and limitations,
Drug Discov. Today 11 (2006) 580e594.
[13] H. Eckert, J. Bajorath, Molecular similarity analysis in virtual screening:
foundations, limitations and novel approaches, Drug Discov. Today 12 (2007)
225e233.
4.4. In vitro kinase inhibitory assay
[14] P. Gedeck, R.A. Lewis, Exploiting QSAR models in lead optimization, Curr.
Opin. Drug Discov. Dev. 11 (2008) 569e575.
[15] N. Brown, R.A. Lewis, Exploiting QSAR methods in lead optimization, Curr.
Opin. Drug Discov. Dev. 9 (2006) 419e424.
[16] P.R. Murumkar, V.P. Zambre, M.R. Yadav, Development of predictive phar-
macophore model for in silico screening, and 3D QSAR CoMFA and CoMSIA
studies for lead optimization, for designing of potent tumor necrosis factor
alpha converting enzyme inhibitors, J. Comput. Aided. Mol. Des. 24 (2010)
143e156.
All the kinase inhibitory assays were carried out through the
KinaseProfiler service provided by Millipore [35]. CK1 was incu-
bated with 200
MgAcetate and [
pmol, concentration as required). The buffer composition includes
20 mM MOPS, 1 mM EDTA, 0.01% Brij-35, 5% Glycerol, 0.1%
m
M specific substrate KRRRALS(p)VASLPGL, 10 mM
g-
33P-ATP] (specific activity approx. 500 cpm/
b
-
[17] S.Y. Yang, Pharmacophore modeling and applications in drug discovery:
challenges and recent advances, Drug Discov. Today 15 (2010) 444e450.
[18] G.B. Li, L.L. Yang, S. Feng, J.P. Zhou, Q. Huang, H.Z. Xie, L.L. Li, S.Y. Yang,
mercaptoethanol, and 1 mg/ml BSA. The reaction was initiated by
the addition of the MgATP mix. After incubation for 40 min at room
temperature, the reaction was stopped by the addition of 3%
Discovery of novel mGluR1 antagonists:
a multistep virtual screening
phosphoric acid solution. 10 mL of the reaction was then spotted
approach based on an SVM model and a pharmacophore hypothesis signifi-
cantly increases the hit rate and enrichment factor, Bioorg. Med. Chem. Lett.
21 (2011) 1736e1740.
onto a P30 filtermat and washed three times for 5 min in 75 mM
phosphoric and once in methanol prior to drying and scintillation
counting. IC50 values of compounds were determined from
doseeresponse curves obtained from the assays at 10 different
concentrations of each compound. All the assays were repeated
twice and the mean values of activity were calculated.
[19] S. Thangapandian, S. John, S. Sakkiah, K.W. Lee, Potential virtual lead identi-
fication in the discovery of renin inhibitors: application of ligand and
structure-based pharmacophore modeling approaches, Eur. J. Med. Chem. 46
(2011) 2469e2476.
[20] P. Purushottamachar, J.B. Patel, L.K. Gediya, O.O. Clement, V.C. Njar, First
chemical feature-based pharmacophore modeling of potent retinoidal retinoic
acid metabolism blocking agents (RAMBAs): identification of novel RAMBA
scaffolds, Eur. J. Med. Chem. 47 (2012) 412e423.
[21] J. Zou, H.Z. Xie, S.Y. Yang, J.J. Chen, J.X. Ren, Y.Q. Wei, Towards more accurate
pharmacophore modeling: multicomplex-based comprehensive pharmaco-
phore map and most-frequent-feature pharmacophore model of CDK2, J. Mol.
Graph. Model. 27 (2008) 430e438.
[22] R.M. Xu, G. Carmel, J. Kuret, X. Cheng, Structural basis for selectivity of the
isoquinoline sulfonamide family of protein kinase inhibitors, Proc. Natl. Acad.
Sci. U. S. A. 93 (1996) 6308e6313.
[23] N. Mashhoon, A.J. DeMaggio, V. Tereshko, S.C. Bergmeier, M. Egli,
M.F. Hoekstra, J. Kuret, Crystal structure of a conformation-selective casein
kinase-1 inhibitor, J. Biol. Chem. 275 (2000) 20052e20060.
Acknowledgment
This work was supported by the National Natural Science
Foundation of China (81172987), SRFDP (20100181110025),
and partly by the 863 Hi-Tech Program (2012AA020301,
2012AA020308).
[24] F. Lovering, S. Kirincich, W. Wang, K. Combs, L. Resnick, J.E. Sabalski, J. Butera,
J. Liu, K. Parris, J.B. Telliez, Identification and SAR of squarate inhibitors of
mitogen activated protein kinase-activated protein kinase 2 (MK-2), Bioorg.
Med. Chem. 17 (2009) 3342e3351.
Appendix A. Supplementary material
[25] G. Rena, J. Bain, M. Elliott, P. Cohen, D4476, a cell-permeant inhibitor of CK1,
suppresses the site-specific phosphorylation and nuclear exclusion of
FOXO1a, EMBO Rep. 5 (2004) 60e65.
[26] N. Oumata, K. Bettayeb, Y. Ferandin, L. Demange, A. Lopez-Giral, M.L. Goddard,
V. Myrianthopoulos, E. Mikros, M. Flajolet, P. Greengard, L. Meijer, H. Galons,
Supplementary data associated with this article can be found in
the online version, at http://dx.doi.org/10.1016/j.ejmech.2012.08.
007. These data include MOL files and InChiKeys of the most
important compounds described in this article.

38
L.-L. Yang et al. / European Journal of Medicinal Chemistry 56 (2012) 30e38
Roscovitine-derived, dual-specificity inhibitors of cyclin-dependent kinases
and casein kinases 1, J. Med. Chem. 51 (2008) 5229e5242.
[30] N. Huang, B.K. Shoichet, J.J. Irwin, Benchmarking sets for molecular docking,
J. Med. Chem. 49 (2006) 6789e6801.
[27] G.P. Sapkota, L. Cummings, F.S. Newell, C. Armstrong, J. Bain, M. Frodin,
M. Grauert, M. Hoffmann, G. Schnapp, M. Steegmaier, P. Cohen, D.R. Alessi, BI-
D1870 is a specific inhibitor of the p90 RSK (ribosomal S6 kinase) isoforms
in vitro and in vivo, Biochem. J. 401 (2007) 29e38.
[31] J.J. Irwin, B.K. Shoichet, ZINCea free database of commercially available
compounds for virtual screening, J. Chem. Inf. Model. 45 (2005) 177e182.
[32] Y. Liu, N.S. Gray, Rational design of inhibitors that bind to inactive kinase
conformations, Nat. Chem. Biol. 2 (2006) 358e364.
[28] K.M. Walton, K. Fisher, D. Rubitski, M. Marconi, Q.J. Meng, M. Sladek, J. Adams,
M. Bass, R. Chandrasekaran, T. Butler, M. Griffor, F. Rajamohan, M. Serpa,
Y. Chen, M. Claffey, M. Hastings, A. Loudon, E. Maywood, J. Ohren, A. Doran,
T.T. Wager, Selective inhibition of casein kinase 1 epsilon minimally alters
circadian clock period, J. Pharmacol. Exp. Ther. 330 (2009) 430e439.
[29] Z. Zhou, A.K. Felts, R.A. Friesner, R.M. Levy, Comparative performance of
several flexible docking programs and scoring functions: enrichment studies
for a diverse set of pharmaceutically relevant targets, J. Chem. Inf. Model. 47
(2007) 1599e1608.
[33] B. Okram, A. Nagle, F.J. Adrian, C. Lee, P. Ren, X. Wang, T. Sim, Y. Xie,
G. Xia, G. Spraggon, M. Warmuth, Y. Liu, N.S. Gray, A general strategy for
creating “inactive-conformation” abl inhibitors, Chem. Biol. 13 (2006)
779e786.
[34] G. Jones, P. Willett, R.C. Glen, A.R. Leach, R. Taylor, Development and valida-
tion of a genetic algorithm for flexible docking, J. Mol. Biol. 267 (1997)
727e748.
[35] Millipore Inc. KinaseProfiler Servce http://www.millipore.com/life_sciences/
flx4/ld_kinases.