Synthesis, SAR and biological evaluation of 1,6-disubstituted-1H- pyrazolo[3,4-d]pyrimidines as dual inhibitors of Aurora kinases and CDK1

Article abstract of DOI:10.1016/j.bmcl.2012.01.019

Since the early 2000s, the Aurora kinases have become major targets of oncology drug discovery particularly Aurora-A and Aurora-B kinases (AKA/AKB) for which the selective inhibition in cells lead to different phenotypes. In addition to targeting these Aurora kinases involved in mitosis, CDK1 has been added as a primary inhibition target in hopes of enhancing the cytotoxicity of our chemotypes harboring the pyrazolopyrimidine core. SAR optimization of this series using the AKA, AKB and CDK1 biochemical assays led to the discovery of the compound 7h which combines strong potency against the 3 kinases with an acceptable microsomal stability. Finally, switching from a primary amide to a two-substituted pyrrolidine amide gave rise to compound 15a which exhibited the desired AKA/CDK1 inhibition phenotype in cells but showed moderate activity in animal models using HCT116 tumor cell lines.

Full text of DOI:10.1016/j.bmcl.2012.01.019

Bioorganic & Medicinal Chemistry Letters 22 (2012) 2070–2074
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Synthesis, SAR and biological evaluation of 1,6-disubstituted-1H-pyrazolo
[3,4-d]pyrimidines as dual inhibitors of Aurora kinases and CDK1
Jean-Yves Le Brazidec a, , Angela Pasis a, Betty Tam a, Christina Boykin a, Cheryl Black a, Deping Wang b,
⇑
Gisela Claassen a, Jer-Hong Chong a, Jianhua Chao a, Junhua Fan a, Khanh Nguyen a, Laura Silvian a,b
Leona Ling b, Lin Zhang a, Michael Choi a,b, Min Teng a, Nuzhat Pathan a, Shuo Zhao a, Tony Li a,
Art Taveras b
,
a Biogen Idec, 5200 Research Place, San Diego, CA 92122, USA
b Biogen Idec, 12 Cambridge Center, Cambridge, MA 02142, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
Since the early 2000s, the Aurora kinases have become major targets of oncology drug discovery partic-
ularly Aurora-A and Aurora-B kinases (AKA/AKB) for which the selective inhibition in cells lead to differ-
ent phenotypes. In addition to targeting these Aurora kinases involved in mitosis, CDK1 has been added
as a primary inhibition target in hopes of enhancing the cytotoxicity of our chemotypes harboring the
pyrazolopyrimidine core. SAR optimization of this series using the AKA, AKB and CDK1 biochemical
assays led to the discovery of the compound 7h which combines strong potency against the 3 kinases
with an acceptable microsomal stability. Finally, switching from a primary amide to a two-substituted
pyrrolidine amide gave rise to compound 15a which exhibited the desired AKA/CDK1 inhibition pheno-
type in cells but showed moderate activity in animal models using HCT116 tumor cell lines.
Ó 2012 Elsevier Ltd. All rights reserved.
Received 26 October 2011
Revised 5 January 2012
Accepted 9 January 2012
Available online 18 January 2012
Keywords:
Aurora kinase
Cyclin dependent kinase
Cell cycle arrest
Pyrazolopyrimidine
Cytotoxicity
The modulation of cell cycle by targeting kinases responsible for
regulating cancer cells has been the object of intense research
which resulted in the identification of putative targets such as cy-
clin dependant kinase 1 (CDK1)1 and Aurora-A (AKA) and Aurora-B
(AKB) kinases.2 These kinases are often over-expressed3 in numer-
ous cancer types and their specific inhibition by gene silencing or
small molecule treatment leads to mitotic arrest and eventual cell
death. In cells, CDK1 is functional when associated with a regula-
tory subunit such as cyclin A or cyclin B depending on the progress
of the cell cycle toward mitosis.1 During mitosis, AKA complexed
with TPX24 regulates centrosome separation and mitotic spindle
formation while AKB complexed with INCENP5 is important in
chromatid alignment and cytokinesis.
In an effort to enhance the cytotoxic effect associated with
selective inhibition of only one of these three kinases, we focused
our program on the development of a dual inhibitor of CDK1 and
AKA. In addition to the work previously published on the pan-
CDK/AKA inhibitor, JNJ7706621,6 this hypothesis was supported
by results of an in vitro experiment combining a selective AKA
inhibitor, MLN8054, with a selective CDK1 inhibitor, RO3306, in a
5-day MTS assay measuring the viability of HCT116 cells. The
combination treatment showed higher cytotoxicity and less neu-
tropenia, as measured by the CFU-GM ratio (a measure of neutro-
penic effect), compared to each of the inhibitors alone (Table 1).
We embarked on optimizing the 1,6-disubstituted-1H-pyrazol-
o[3,4-d]pyrimidine chemotype to enable it to bind both CDK1 and
Aurora kinases. This was derived from a homologous chemotype,
2,8-disubstituted-pyrido[2,3-d]pyrimidin-7(8H)-one, optimized
for inhibition of CDK2 and CDK4 by Vanderwel et al.7
As previously reported,8 the synthesis of the pyrazolo-pyrimi-
dine series started with the coupling between the thiomethyl pyr-
azolopyrimidine 1 and alicyclic alcohols (ROH) leading to the
intermediates 2 (a–i) (Scheme 1). After oxidation of these latter
with m-CPBA, the sulfones 3 (a–i) were condensed with the methyl
3-amino-pyrrole-carboxylate 4 under basic conditions to give the
methyl esters 5 (a–i) which were hydrolyzed into the key carbox-
ylic acids 6 (a–i) in good overall yield. Treated with either ammo-
nium carbonate in the presence of (BOC)2O or with a secondary
amine under standard amide coupling conditions, the acids 6
(a–i) were converted, respectively, into the primary amides 7
(a–i) and the pyrrolidine amides 8 to 16 in acceptable yields.
In addition to increasing the biochemical potency against AKA
and CDK1, a second goal of this study was to design a compound
with better pharmaceutical properties by limiting its lipophilicity
(CLogP), which is an important molecular property that has been
⇑
Corresponding author.
E-mail address: jean@chempartner.com (J.-Y. Le Brazidec).
0960-894X/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2012.01.019

J.-Y. Le Brazidec et al. / Bioorg. Med. Chem. Lett. 22 (2012) 2070–2074
2071
Table 1
Combination of MLN8054 and RO3306 in MTS and neutropenia assays
HOOC
O
N
N
S
N
HN
NH
S
N
N
F
F
Cl
MLN8054
M)
RO3306
Compd
AKAa IC50
(l
CDK1a IC50
(l
M)
HCT116a IC50
(lM)
CFU-GM ratio13
RO3306
MLN8054
RO3306 + MLN8054
>10
0.0015
—
0.019
>100
—
3.2
0.69
0.14
16.7
4.5
1.9
a
For assay descriptions, see Ref. 9.
NH3
Cl
O
N
N
N
N
O
a
N
b
4
N
N
R1
N
O
N
S
N
N
S
N
c
S
N
H
R1
O
1
2a-i
3a-i
N
N
N
N
N
N
N
R1
N
R1
HN
N
N
R1
HN
N
HN
N
d
e
R2
N
N
N
O
N
HO
R3
O
O
O
6a-i
7a-i, 8-16
5a-i
Scheme 1. Reagents and conditions: (a) ROH, DIAD, PPh3, THF, À70 °C to rt, 30–80%; (b) m-CPBA, DCM, rt, 95%; (c) 4, K2CO3, DMSO, 80 °C; (d) KOH 5 N, MeOH/dioxane, 90%;
(e) NH4HCO3, (BOC)2O, pyridine, rt, 30–70% or R2R3NH, HATU, DIEA, THF, 50 °C, 50–80%.
correlated with drug-likeness.9 During the optimization of this
pyrazolo[3,4-d]pyrimidine series, the nature of R1 was first modi-
fied in order to maximize its interactions with the lipophilic side
chains from the amino acid residues such as Leu210, Leu263 and
Phe275 lining the ribose binding pocket.10 In Table 2, the results
showed that the isopropyl group conferred weak activity against
AKA, AKB and CDK1 in compound 7a but also conferred an
improvement in metabolic stability. Increasing the size and the
lipophilicity of R1 by replacement of the isopropyl with a monocy-
clic ring (7g and 7e), spirobicyclic rings (7b and 7f), fused bicyclic
rings (7d, 7h and 7i) and tricyclic rings (7c) led to a dramatic
potency enhancement but accompanied with a significant loss in
metabolic stability. The exception was compound 7h; it main-
tained a % Qh under 70% and reasonable potency.
amide-anilinyl group from the previous series.9 In order to improve
the overall kinase selectivity profile, the two-substituted pyrroli-
dine amides were investigated (Table 3). Dimethylaminocarbonyl
or methylaminomethyl substituents (8, 9) led to weak activity
against CDK1 and AKA which translated into reduced activity in
cells. For the other substituents, the IC50s in the biochemical assay
against AKA, AKB and CDK1 were all in a tight range from 0.002 to
0.022 lM except for the compounds 11 and 14 which were a little
weaker against CDK1. This strong biochemical activity gave rise to
a marked phenotypic effect yielding G2/M arrest at concentrations
between 0.01 and 0.03 lM. The HCT116 cell viability assay con-
firmed the potencies of these compounds in cells with IC50s within
the same range.
Furthermore, a matched analysis of three pairs of enantiomers
in this series revealed that the (S) enantiomers were more potent
than their corresponding (R) isomers, especially in the case of the
primary alcohol 13a which displayed enantiomeric-selectivity ra-
tios of 25, 12 and 15 for AKA, AKB and CDK1, respectively (Table 4).
We then focused our attention on the amide moiety connected
to the N-methyl 3-amino-pyrrole group to further improve potency
while retaining reasonable metabolic stability. The N-methyl
3-amino-pyrrole is a key pharmacophore replacing the 4-sulfon-

2072
J.-Y. Le Brazidec et al. / Bioorg. Med. Chem. Lett. 22 (2012) 2070–2074
Table 2
SAR of 1-substituted-pyrazolopyrimidines
N
N
N
R1
HN
N
N
H2N
O
Compd
R1
AKA IC50
0.23
(lM)
AKB IC50
0.038
(l
M)
CDK1 IC50
(
lM)
G2/Ma arrest (
lM)
HCT116 IC50
1
(
lM)
cLogP
MLM Stability (% Qh)b
27
7a
0.16
1
1.2
7b
7c
0.15
0.019
0.018
0.05
0.65
0.3
0.29
0.27
2.1
4.0
89
89
0.025
0.038
H
7d
7e
0.021
0.087
0.012
0.023
0.016
0.015
0.3
0.23
0.14
2.7
2.9
76
87
H
0.23
7f
7g
7h
0.04
0.03
0.017
0.005
0.005
0.004
0.019
0.024
0.024
0.3
0.1
0.1
0.12
0.1
2.7
3.4
1.6
89
92
69
0.079
H
7i
0.013
0.022
0.013
0.1
0.066
2.7
85
H
a
The G2/M arrest corresponds to the minimum concentration for G2/M versus G1 peak ratio over 2.
% Qh = in vitro clearance/hepatic blood flow of the species.
b
Table 3
SAR of 2-substituted-pyrrolidine amides 8 to 16a
N
N
N
N
HN
N
N
O
R
Compd
R
AKA IC50
(lM)
AKB IC50
(
lM)
CDK1 IC50
(lM)
G2/M 100% (
l
M)
HCT116 IC50 (lM)
8
9
10
11
12
13a
CONMe2
CH2NHMe
Me
0.019
0.08
0.002
0.017
0.006
0.004
0.004
0.05
0.004
0.04
0.004
0.003
1.7
1.2
0.04
0.041
0.01
0.010
0.1
0.3
0.03
0.1
0.03
0.01
0.16
0.15
0.13
0.095
0.047
0.018
H
CONH2
CH2OH
14
0.01
0.005
0.005
0.047
0.022
0.03
0.03
0.011
0.01
N
15a
CH2OMe
0.004
O
16a
0.003
0.004
0.021
0.017
0.008
N

J.-Y. Le Brazidec et al. / Bioorg. Med. Chem. Lett. 22 (2012) 2070–2074
2073
Table 4
SAR of enantiomeric pairs of 2-substituted-pyrrolidine amides
N
N
N
HN
N
N
N
O
R
Compd
R
AKA IC50
(
lM)
AKB IC50
(lM)
CDK1 IC50
(l
M)
G2/M 100% (lM)
HCT116 IC50 (lM)
13a
13b
15a
15b
(S)CH2OH
(R)CH2OH
(S)CH2OMe
(R)CH2OMe
0.004
0.1
0.004
0.016
0.003
0.036
0.005
0.006
0.010
0.15
0.022
0.15
0.01
0.1
0.03
0.065
0.018
0.17
0.01
0.06
O
16a
16b
0.003
0.004
0.004
0.021
0.39
0.017
0.03
0.008
0.024
(S)
(R)
N
N
O
0.0015
Table 5
Selected pharmacokinetic parameters of 15a in mouse and rat.
L210(Gatekeeper)
Dose (mg/kg)
T1/2 (h)
AUC (h ng/mL)
CL (mL/min/kg)
Vz (L/kg)
25a
4b
1
5
6416
1228
65
55
3
5
R137
a
Mouse PK, intraperitoneal administration.
Rat PK, intravenous administration.
Ribose pocket
b
2hr
6hr
9hr
24hr
R220
350.0
300.0
250.0
200.0
150.0
100.0
50.0
Figure 1. Docking of compound 15a against AKA.
µM
0
-
0
0.003 0.01 0.03 0.1 0.3
1
0.0
Nocodazole
+
+
+
+
+
+
+
-50.0
-100.0
-150.0
PH1
PH3
pAKA
pAKB
pAKC
PH3
Figure 3. Pharmacodynamic experiment data for 15a in mouse at low serum
concentrations.
PH1
Actin
The kinase selectivity screening of 15a against a panel of 262 ki-
nases resulted in the inhibition of 100 kinases with IC50 P 1 lM,
Figure 2. Effect of 15a on the phosphorylation level of AKA, AKB and AKC in K562
cells.
suggesting that the compound is very selective. To further deter-
mine whether the biochemical assay, which was performed without
regulatory factors, was reflective of the in vivo pharmacodynamics,
the levels of phosphorylation of AKA and AKB in cells were deter-
mined upon treatment of K562 leukemia cells with different con-
centrations of 15a. Quantitation of the levels of phosphorylated
Aurora kinases by Western blot shows that 15a inhibits AKA phos-
phorylation at much lower concentrations than the concentrations
required to fully inhibit AKB phosphorylation (Fig. 2). Possibly the
discrepancies between the IC50s observed in the biochemical assay
and PD assay can be explained not including the co-factors TPX2
and INCENP in the AKA and AKB biochemical assays.
The docking study for this series is exemplified in Figure 1
showing the key interactions between 15a and the ATP binding site
of AKA. Two hydrogen bonds are predicted between the pyrazolo-
pyrimidine core and the hinge region. More interestingly, the 2-
methoxymethyl-pyrrolidine amide moiety binds in an optimal
manner with the side-chains of Arg137 and Arg220 of AKA.11 The
R-methoxymethyl (15b) is expected to weaken this interaction.
This moiety is also predicted to interact with Lys9 and Lys89 in
CDK1.

2074
J.-Y. Le Brazidec et al. / Bioorg. Med. Chem. Lett. 22 (2012) 2070–2074
H3 was observed, corresponding to an AKA inhibition phenotype
HCT116 nu/nu
as observed for AKA selective inhibitors such as MLN805412
(Fig. 3).
Control QDx21
1000
800
600
400
200
0
15a 12.5 mg/kg QDx21 n=8
15a 25 mg/kg QDx21 n=8
15a 40 mg/kg QDx21 n=10
Irinotecan 10 mg/kg QDx5
As 15a displayed acceptable PK properties, was effective in
modulating the AKA and CDK1 biomarkers and exhibited the
highest neutropenic index of 32 (other data not shown), we con-
ducted an efficacy study in the HCT116 nu/nu mouse models by
administering it daily using IP route (Fig. 4). It resulted in modest
tumor growth inhibition (30–40%) and moderate weight loss
(2–9%) along with a range of side effects such as reluctance to
move and hunched postures (Fig. 5). In addition, deaths of a few
animals in this study indicated that 15a was not tolerated at this
dosing regimen.
In conclusion, the optimization of the SAR for this series against
AKA, AKB and CDK1 has been achieved. The PK and PD studies on
15a and other analogs (data not shown) revealed that this class of
compounds reached serum levels which are sufficient to efficiently
modulate the histone biomarkers pH1 and pH3 corresponding,
respectively, to CDK1 and AKA inhibition phenotypes. However,
the modest efficacy and the lack of tolerability in mouse models,
probably due to both ubiquitous nature and medium clearance of
this series, prevented us from evaluating this class of compounds
any further.
0
5
10
15
20
25
30
35
40
Days Post Tumor Inoculation
Figure 4. Efficacy study of 15a in HCT116 nu/nu mouse models.
Acknowledgments
672/CB/HCT116 nu/nu
Control QDx21
Special thanks to Dr. Adeela Kamal, Dr. Srinivas R. Kasibhatla,
Dr. Francis J. Burrows, and Dr. Marcus F. Boehm for their contribu-
tions to this program.
15a 12.5 mg/kg QDx21
16
15a 25 mg/kg QDx21
11
15a 40 mg/kg QDx21
References and notes
Irinotecan 10 mg/kg QDx5
6
1
1. Lapenna, S.; Giordano, A. Nat. Drug Disc. 2009, 8, 547.
2. Kitzen, J. J.; De Jonge, M. J.; Verweij, J. Crit. Rev. Oncol. Hematol. 2010, 73, 99.
3. Andrews, P. Oncogene 2005, 25, 5005.
4. Bayliss, R.; Sardon, T.; Vernos, I.; Conti, E. Mol. Cell 2003, 12, 851.
5. Sessa, F.; Mapelli, M.; Ciferri, C.; Tarricone, C.; Areces, L. B.; Schneider, T. R.;
Stukenberg, P. T.; Musacchio, A. Mol. Cell 2005, 18, 379.
-4
-9
6. Emmanuel, S.; Rugg, C. A.; Gruninger, R. H.; Lin, R.; Fuentes-Pesquera, A.;
Connolly, P. J.; Wetter, S. K.; Hollister, B.; Kruger, W. W.; Napier, C.; Jolliffe, L.;
Middleton, S. Cancer Res. 2005, 19, 9038.
7. Vanderwel, S. N.; Harvey, P. J.; McNamara, D. J.; Repine, J. T.; Keller, P. R.; Quin,
J., III; Booth, R. J.; Elliott, W. L.; Dobrusin, E. M.; Fry, D. W.; Toogood, P. L. J. Med.
Chem. 2005, 48, 2371.
-14
-19
-24
8. Zhang, L.; Fan, J.; Chong, J.-H.; Cesena, A.; Tam, B.; Gilson, C.; Boykin, C.; Wang,
D.; Aivazian, D.; Marcotte, D.; Xiao, G.; le Brazidec, J.-Y.; Piao, J.; Lundgren, K.;
Hong, K.; Vu, K.; Nguyen, K.; Gan, L.-S.; Silvian, L.; Ling, L.; Teng, M.; Reff, M.;
Takeda, N.; Timple, N.; Wang, Q.; Morena, R.; Khan, S.; Zhao, S.; Li, T.; Lee, W.-
C.; Taveras, A. G.; Chao, J. Bioorg. Med. Chem. Lett. 2011, 21, 5633.
9. Leeson, P. D.; Springthorpe, B. Nat. Rev. Drug Disc. 2007, 6, 881.
10. Cheetham, G. M. T.; Knegtel, R. M. A.; Coll, J. T.; Renwick, S. B.; Swenson, L.;
Weber, P.; Lippke, J. A.; Austen, D. A. J. Biol. Chem. 2002, 277, 42419.
11. The model of 15a in AKA was generated using Glide 5.5 (Schrodinger, LLC)
based on the published crystal structure of Aurora A(pdb: 3R21). Graphics
generated in PyMol (Delano Scientific, South San Francisco, version 0.99).
12. Manfredi, M. G.; Ecsedy, J. A.; Meetze, K. A.; Balani, S. K.; Burenkova, O.; Chen,
W.; Galvin, K. M.; Hoar, K. M.; Huck, J. J.; LeRoy, P. J.; Ray, E. T.; Sells, T. B.;
Stringer, B.; Stroud, S. G.; Vos, T. J.; Weatherhead, G. S.; Wysong, D. R.; Zhang,
M.; Bolen, J. B.; Claiborne, C. F. Proc. Natl. Acad. Sci. U.S.A. 2007, 104, 4106.
13. CFU-GM assay: Approximately 3.75 Â 104 bone marrow mononuclear cells
(Stem Cell Technologies) were plated in duplicate in Methocult media designed
for CFU-GM assays of human cells containing rhSCF, rhG-CSF, rhGM-CSF, rhIL-3
containing vehicle or various concentrations of indicated drug. Colonies were
allowed to form under humidified conditions at 37 °C, 5% CO2 for 14 days. At
which time, CFU-GM colonies were scored manually under light microscopy.
EC50s were determined by GraphPad analysis.
15
20
25
30
35
40
Days Post Tumor Inoculation
Figure 5. Body weight loss after administration of 15a in HCT116 nu/nu mouse
models.
Pharmacokinetic studies of 15a in mice and rat showed that this
compound exhibited medium to high clearance along with low dis-
tribution volumes (Table 5). Subsequently, a pharmacodynamic
experiment confirmed the CDK1 inhibition phenotype at high
serum concentrations (after 2 h) with 60% and 90% decreases in
biomarkers phosphorylated histone H1 (pH1) and phosphorylated
histone H3 (pH3), respectively. At lower serum concentrations
(after 6 and 9 h), a dramatic increase in phosphorylated histone