Aurora kinase A inhibitors: Identification, SAR exploration and molecular modeling of 6,7-dihydro-4H-pyrazolo-[1,5-a]pyrrolo[3,4-d]pyrimidine-5,8-dione scaffold

Article abstract of DOI:10.1016/j.bmcl.2008.01.068

Tricyclic 6,7-dihydro-4H-pyrazolo[1,5-a]pyrrolo[3,4-d]pyrimidine-5,8-dione was identified as a novel scaffold for Aurora kinase A inhibition through virtual screening. SAR exploration coupled with molecular modeling of 8a reveals the minimum pharmacophore requirements for Aurora kinase A inhibition.

Full text of DOI:10.1016/j.bmcl.2008.01.068

Available online at www.sciencedirect.com
Bioorganic & Medicinal Chemistry Letters 18 (2008) 1623–1627
Aurora kinase A inhibitors: Identification, SAR exploration
and molecular modeling of 6,7-dihydro-4H-pyrazolo-
[1,5-a]pyrrolo[3,4-d]pyrimidine-5,8-dione scaffold
Mohane Selvaraj Coumar,a Jian-Sung Wu,a, Jiun-Shyang Leou,a, Uan-Kang Tan,b
Chung-Yu Chang,a Teng-Yuan Chang,a Wen-Hsing Lin,a John T.-A. Hsu,a,c
Yu-Sheng Chao,a Su-Ying Wua,* and Hsing-Pang Hsieha,*
aDivision of Biotechnology and Pharmaceutical Research, National Health Research Institutes,
No. 35, Keyan Road, Zhunan Town, Miaoli County 350, Taiwan, ROC
bDepartment of Chemical and Materials Engineering, Technology and Science Institute of Northern Taiwan, Taipei 112, Taiwan, ROC
cDepartment of Chemical Engineering, National Tsing Hua University, Hsinchu 300, Taiwan, ROC
Received 23 October 2007; revised 12 January 2008; accepted 16 January 2008
Available online 19 January 2008
Abstract—Tricyclic 6,7-dihydro-4H-pyrazolo[1,5-a]pyrrolo[3,4-d]pyrimidine-5,8-dione was identified as a novel scaffold for Aurora
kinase A inhibition through virtual screening. SAR exploration coupled with molecular modeling of 8a reveals the minimum phar-
macophore requirements for Aurora kinase A inhibition.
Ó 2008 Elsevier Ltd. All rights reserved.
Inhibition of Aurora kinase, a member of serine/threo-
nine kinase involved in the regulation of cell division is
emerging as a new molecular targeted cancer treatment
option.1–4 Three isoforms of Aurora kinase, A, B, and
C are known, all of which have a conserved ATP bind-
ing site, but differ in amino acid length and sequence at
the N-terminal domain.3,5 Aurora A localizes on centro-
somes from early S phase and is involved in centrosome
maturation and separation, bi-polar spindle assembly,
mitotic entry, and exit. Aurora B is a chromosome pas-
senger protein, which localizes to the kinetochores from
prophase to metaphase and to the central spindle and
midbody during cytokinesis. It is essential for accurate
chromosomal segregation, protein localization to the
centromere and kinetochore, and for proper cytokinesis
to take place. Aurora C co-localizes and complements
the functions of Aurora B.3,5
mosome instability, oncogenic transformation, tumor
progression, and development of chemoresistance.6,7
Inhibitors of Aurora kinase, regardless of their Aurora
isoform specificity have shown to promote cancer cell
death by induction of apoptosis and mitotic catastro-
phe. A handful of Aurora inhibitors are in various
stages of development,4 among them MK-0457/VX-
6808–10 (1), AZD115211,12 (2), and PHA-73935813 (3)
are in clinical development (Fig. 1). VX-680 is a potent
inhibitor of all three Aurora kinases with Ki values of
0.6, 1.8, and 4.6 nM for A, B, and C isoforms, respec-
tively.9 It is undergoing clinical trials for solid tumors
and hematological malignancies. AZD1152 is a quinaz-
oline prodrug with selective Aurora kinase B/C inhibi-
tion profile. PHA-739358 is a non-selective agent like
VX-680, with IC50 values of 13, 79, and 61 nM for Aur-
ora A, B, and C, respectively.13
More importantly, both Aurora A and B are overex-
pressed in many human cancers and are linked to chro-
Based on the current success of Aurora kinase inhibitors
in the development of kinase-based cancer therapy, we
have initiated a virtual screening program for the identi-
fication of Aurora kinase inhibitors. Through virtual
screening14 of around 60,000 compound library,15 we
have identified a few potential hits with high priority
scores. These compounds were purchased and assayed
for Aurora kinase A inhibition. The most promising
hit in terms of activity and novelty in structure,
Keywords: Aurora kinase inhibitors; Cancer; Molecular modeling.
*
Corresponding authors. Tel.: +886 37 246166x35713; fax: +886
37586456 (S.-Y.W.); tel.: +886 37 246166x35708; fax: +886 37586456
(H.-P.H.); e-mail addresses: suying@nhri.org.tw; hphsieh@nhri.org.
tw
These authors contributed equally to this work.
0960-894X/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2008.01.068

1624
M. S. Coumar et al. / Bioorg. Med. Chem. Lett. 18 (2008) 1623–1627
H
N
N
HN
HN
N
O
H
O
N
N
N
O
H
N
O
N
S
N
N
N
1 (MK-0457/VX-680)
3 (PHA-739358)
F
N
O
NH
O
N
NH
O
HO P
HO
N
N
O
HN
N
O
N
N
H
O
O
N
N
2
(AZD1152)
8a
(CSV1A1957S0)
Figure 1. Aurora kinase inhibitors.
CSV1A1957S0 (8a) had a low micromolar Aurora A
inhibition profiles with 90% inhibition at 50 lM concen-
tration. Literature search showed that the 6,7-dihydro-
4H-pyrazolo[1,5-a]pyrrolo[3,4-d]pyrimidine-5,8-dione
scaffold was reported as inhibitor of acyl CoA: choles-
terol acetyl transferase,16 but as such was unreported
as inhibitor for any protein kinases. Considering this
fact, the development of this new Aurora inhibitor could
reveal interesting insight into the binding modes of 8a to
Aurora kinase A. This paper describes the SAR studies
and molecular modeling of 6-benzyl-2-(furan-2-yl)-6,7-
dihydro-4H-pyrazolo[1,5-a]pyrrolo[3,4-d]pyrimidine-5,8-
dione 8a for Aurora kinase A inhibition.
alkylating with methyl iodide or 4-(2-chloroethyl)-mor-
pholine hydrochloride using potassium carbonate. The
structures of all the synthesized compounds were con-
1
firmed by H NMR and LC–MS analysis.
Aurora kinase A inhibition assay17 was performed at
two compound concentrations of 50 and 10 lM. For
compounds with a good activity profile, IC50 was also
determined. The original virtual screening hit 8a,
showed an IC50 of about 15 lM for Aurora A inhibition
(Table 1). Initial attempts were focused on the replace-
ment of furan ring (R2 group), but either a thiophene
(8b) or phenyl ring (8c) led to a decrease in activity.
Complete removal of furanyl group (8e) or replacement
with methyl group (8d) led to a dramatic loss of enzyme
inhibition activity. This SAR trend clearly shows that
presence of heterocyclic group (R2 group) is essential
for maintaining contact with the Aurora kinase. In the
central ring NH group, substitution with either methyl
(8f) or 4-ethyl morpoline group (8g) led to complete loss
of kinase inhibition, suggesting the importance of free
NH group for binding to Aurora kinase.
The general synthetic method16 for the preparation of
the fused tricyclic compounds is shown in Scheme 1.
Commercially available amines 4 were alkylated with
ethyl acrylate in absolute ethanol by stirring at RT to
give the intermediate amines 5. In the case of aniline
(for 8h compound), refluxing for 24 h is required for
the completion of reaction. Intermediate amines 5
formed were cyclized in-situ to the pyrrolidinones 6
using diethyl oxalate and sodium ethoxide in refluxing
ethanol. Condensation of pyrrolidinones 6 with various
substituted pyrazoles 7 in refluxing glacial acetic acid
gave the final tricyclic compounds 8a–e, h–u in 20–25%
overall yields after crystallizing from methanol–DMSO
mixture. Compounds 8f–g were synthesized from 8a by
Next, we turned our attention to the pyrrolidinone ring
substitution (R1 group; Table 2). Decreasing the linker
length that separates the phenyl ring from the tricyclic
core by one-carbon (8h) or increasing the length to
two- (8i) or three-carbon (8j) led to decreased activity.
O
O
(i)
O
(ii)
NH2
O
R1
N
R1
R1 NH
OH
4
5
O
6
O
O
N
N
(iii)
(iv)
N
N
R1
R2
R1
R2
N
N
8a
N
N
R3
8f-g
N
H
HN
H2N
R2
O
O
7
8a-e, h-u
Scheme 1. Reagents and conditions: (i) ethyl acrylate, ethanol, rt, 12 h; (ii) diethyl oxalate, sodium ethoxide, ethanol, reflux, 3 h; (iii) gl. acetic acid,
reflux, 2 h; (iv) methyl iodide or 4-(2-chloroethyl)-morpholine hydrochloride, K2CO3, dichloromethane.

M. S. Coumar et al. / Bioorg. Med. Chem. Lett. 18 (2008) 1623–1627
1625
Table 1. Inhibition of Aurora kinase A by compounds 8a–g
Table 2. Inhibition of Aurora kinase A by compounds 8h–u
O
O
N
O
N
N
N
R2
N
R1
N
N
H
N
R3
O
O
Compound R1
% inhibition of
Aurora kinase Aa
IC50
(lM)
Compound R2
R3
% inhibition of Aurora
kinase Aa
at 50 lM at 10 lM
at 50 lM at 10 lM
O
S
8a
8h
8i
90.2
73.7
83.9
84.4
79.2
35.6
7.6
15.1
—
8a
8b
H
H
90.2
72.3
35.6
(IC50 15.1 lM)
29.4
25.9
8c
H
47.2
19.6
16.4
21.6
—
8d
8e
CH3
H
H
20.8
14.5
1.8
1.6
H
8j
—
O
O
8f
CH3
3.4
1.1
2.3
0.0
8k
—
8g
N
O
N
a Values are expressed as the mean of three independent determina-
tions and are within 15%.
8l
79.9
30.4
—
Changing the phenyl ring to pyridyl (8k) or naphthyl
ring (8l) system also has no positive effect on activity.
As benzyl group was found optimal, the effects of sub-
stituents on the phenyl ring were investigated. Introduc-
tion of electron-donating substitution such as a methoxy
group at three different positions (8m–o) of the phenyl
ring led to the retention of activity. Of particular interest
was the ortho substituted compound 8o, which showed
an enhanced activity with almost threefold improve-
ments in IC50 of about 5 lM when compared to the
unsubstituted lead compound 8a. The meta substituted
compound 8n showed the lowest activity among the
three compounds. When an additional methoxy substi-
tuent was introduced the activity level declined as in
the case of 8p and 8q. Also, when a dimethoxy equiva-
lent methylene dioxy group (8r) was introduced the
activity level declined. On the other hand, introduction
of electron-withdrawing substituent such as a fluoro
substitution at para position (8s) led to a decreased
activity; while acetamido group (8t) in the meta position
retained the activity. In contrast to all of the above mod-
ifications, when the hydrophobic benzyl group of 8a was
replaced with a hydrophilic 4-ethyl morpholine group,
the resulting compound 8u completely lost the activity,
showing that the binding region for this part is hydro-
phobic in nature and hydrophilic substitution may not
be tolerated. Additionally, the two-carbon linker be-
tween the morpholine group and the tricyclic core could
have contributed to the complete loss of activity, as one-
carbon linker was found to be optimal (8a vs 8i).
8m
8n
88.2
90.0
42.4
29.3
8.0
OCH
3
OCH
3
15.2
H CO
3
8o
8p
95.1
75.3
51.6
10.1
5.6
—
OCH
3
OCH
3
OCH3
8q
8r
8s
70.8
75.2
89.8
7.8
12.3
24.0
—
—
—
OCH3
O
O
F
O
NH
8t
87.8
35.8
18.3
N
8u
10.7
0.9
—
O
a Values are expressed as the mean of three independent determina-
tions and are within 15%.
In order to understand the observed SAR trend and to
study the possible binding modes of this class of com-
pounds to Aurora kinase A, docking studies were car-

1626
M. S. Coumar et al. / Bioorg. Med. Chem. Lett. 18 (2008) 1623–1627
Figure 3. Superposition of the docking model of 8a (magenta) with 8o
(green) in Aurora A protein.
Figure 2. Docking model of 8a in Aurora A protein.
with those of 8a. The phenyl ring substituted with the
methoxy group extended deeper into the hydrophobic
pocket as compared with 8a and the methoxy group
had interactions with Lys162, Leu194, and Leu210.
The protein–ligand intramolecular energy calculated
by InsightII 2000.1 (Accelrys Inc., San Diego, USA)
are À67.3 (kcal/mol) and À70.8 (kcal/mol) for com-
pounds 8a and 8o, respectively, suggesting that the bind-
ing of 8o with Aurora A is more energetically favorable
and consequently leads to the improved potency of 8o.
ried out. Compound 8a was first docked into the active
site/ATP binding site of the Aurora kinase A by Gold
3.0 (CCDC Software Limited, Cambridge, UK) and
then subjected to extensive docking by Glide 4.5 (Schro-
dinger, L.L.C., New York, USA) using the structure of
Aurora A (PDB code: 1MQ4) as the template. The
docking model (Fig. 2) showed the tricyclic ring fitted
into the hinge region where the free NH group in the
central ring formed the H-bond with the carbonyl group
of Ala213 and the pyrrolidine ring carbonyl group ‘O’
atom formed the other H-bond with the NH on the
main chain of Ala213. The important H-bond interac-
tions of tricyclic ring with the hinge region could give
the explanation for the loss of Aurora activity in 8f
and 8g, where the central ring free NH group is substi-
tuted with methyl and 4-ethyl morpoline groups, respec-
tively, which could interrupt in the hinge binding.
Moreover, the tricyclic ring made hydrophobic interac-
tions with the residues Leu139, Gly216, and Leu263.
The furanyl ring formed a H-bond with Arg137 and
had close contacts with Arg 220. In addition, there
would be cation-p interactions between the furanyl aro-
matic ring and these two Arg residues.18 The removal of
furanyl ring, as seen in 8c–8e, might loose these interac-
tions with the protein thus resulting in the loss of activ-
ity. Finally, the phenyl ring of compound 8a extended
into the hydrophobic pocket next to the hinge region
and formed hydrophobic interactions with Val147,
Leu194, Leu210, Leu263, Ala273, and Asp274. Also,
the docking pose of 8a was compared with that of a
known pyrimidine Aurora inhibitor (PDB code:
2NP8). It was found that both of them form two H-
bonds with the hinge residue Ala213, and one H-bond
with Arg137, while their phenyl groups orient differently
in the hydrophobic region of the Aurora kinase.
In conclusion, from virtual screening we have identified
a novel tricyclic ring system showing micromolar Aur-
ora kinase A inhibition. Since it is important and critical
to develop Aurora kinase inhibitors possessing novel
core structure, development of 6,7-dihydro-4H-pyrazol-
o[1,5-a]pyrrolo[3,4-d]pyrimidine-5,8-dione
scaffold,
which is unprecedented in the kinase domain is worth-
while and interesting. Molecular modeling of 8a and
8o reveals the tricyclic core made the important H-bond-
ing interactions with the hinge region while the furanyl
ring and the phenyl ring provided additional H-bond
and hydrophobic interactions with the protein. More-
over, the modeling study is consistent with the SAR ob-
served in this series of compounds and gives reasons for
the improved potency of 8o. Further skillful manipula-
tion of the substituents around the central tricyclic core
for lead optimization is intensively undergoing.
Acknowledgments
The study was financially supported by National Health
Research Institutes and National Science Council
(Grant Nos. NSC-95-2113-M-400-001-MY3 and NSC-
95-2752-B-007), Taiwan, ROC.
To understand the improved potency of compound 8o
with an ortho methoxy substituent in the phenyl ring,
docking was done following the same procedure as de-
scribed for compound 8a. As shown in Figure 3, the tri-
cyclic ring and the furanyl ring of 8o superimposed well
Supplementary data
Superposition of the docking pose of 8a and known
pyrimidine Aurora inhibitor (PDB code: 2NP8); Syn-

M. S. Coumar et al. / Bioorg. Med. Chem. Lett. 18 (2008) 1623–1627
1627
thetic procedure and spectral data (1H NMR and LC–
MS) of 6a, h–u and 8a–u; Bioassay protocol for Aurora
kinase inhibition available as supplementary data. Sup-
plementary data associated with this article can be
found, in the online version, at doi:10.1016/
j.bmcl.2008.01.068.
H. K.; Zarrinkar, P. P.; Lockhart, D. J.; Sawyers, C. L.;
Kuriyan, J. Cancer Res. 2006, 66, 1007.
11. Mortlock, A. A.; Foote, K. M.; Heron, N. M.; Jung, F. H.;
Pasquet, G.; Lohmann, J. J.; Warin, N.; Renaud, F.; De
Savi, C.; Roberts, N. J.; Johnson, T.; Dousson, C. B.; Hill,
G. B.; Perkins, D.; Hatter, G.; Wilkinson, R. W.; Wedge, S.
R.; Heaton, S. P.; Odedra, R.; Keen, N. J.; Crafter, C.;
Brown, E.; Thompson, K.; Brightwell, S.; Khatri, L.; Brady,
M. C.; Kearney, S.; McKillop, D.; Rhead, S.; Parry, T.;
Green, S. J. Med. Chem. 2007, 50, 2213.
12. Wilkinson, R. W.; Odedra, R.; Heaton, S. P.; Wedge, S.
R.; Keen, N. J.; Crafter, C.; Foster, J. R.; Brady, M. C.;
Bigley, A.; Brown, E.; Byth, K. F.; Barrass, N. C.; Mundt,
K. E.; Foote, K. M.; Heron, N. M.; Jung, F. H.;
Mortlock, A. A.; Boyle, F. T.; Green, S. Clin. Cancer
Res. 2007, 13, 3682.
13. Fancelli, D.; Moll, J.; Varasi, M.; Bravo, R.; Artico, R.;
Berta, D.; Bindi, S.; Cameron, A.; Candiani, I.; Cappella,
P.; Carpinelli, P.; Croci, W.; Forte, B.; Giorgini, M. L.;
Klapwijk, J.; Marsiglio, A.; Pesenti, E.; Rocchetti, M.;
Roletto, F.; Severino, D.; Soncini, C.; Storici, P.; Tonani,
R.; Zugnoni, P.; Vianello, P. J. Med. Chem. 2006, 49, 7247.
14. Details of virtual screening will be published in our next
communication.
References and notes
1. Glover, D. M.; Leibowitz, M. H.; McLean, D. A.; Parry,
H. Cell 1995, 81, 95.
2. Gautschi, O.; Mack, P. C.; Davies, A. M.; Lara, P. N., Jr.;
Gandara, D. R. Clin. Lung Cancer 2006, 8, 93.
3. Matthews, N.; Visintin, C.; Hartzoulakis, B.; Jarvis, A.;
Selwood, D. L. Expert Rev. Anticancer Ther. 2006, 6, 109.
4. Naruganahalli, K. S.; Lakshmanan, M.; Dastidar, S. G.;
Ray, A. Curr. Opin. Investig. Drugs 2006, 7, 1044.
5. Carmena, M.; Earnshaw, W. C. Nat. Rev. Mol. Cell Biol.
2003, 4, 842.
6. Katayama, H.; Brinkley, W. R.; Sen, S. Cancer Metastasis
Rev. 2003, 22, 451.
7. Fu, J.; Bian, M.; Jiang, Q.; Zhang, C. Mol. Cancer Res.
2007, 5, 1.
8. Cheetham, G. M.; Charlton, P. A.; Golec, J. M.; Pollard,
J. R. Cancer Lett. 2007, 251, 323.
9. Harrington, E. A.; Bebbington, D.; Moore, J.; Rasmussen,
R. K.; Ajose-Adeogun, A. O.; Nakayama, T.; Graham, J.
A.; Demur, C.; Hercend, T.; Diu-Hercend, A.; Su, M.;
Golec, J. M.; Miller, K. M. Nat. Med. 2004, 10, 262.
10. Young, M. A.; Shah, N. P.; Chao, L. H.; Seeliger, M.;
Milanov, Z. V.; Biggs, W. H., 3rd; Treiber, D. K.; Patel,
15. Commercially available compound database obtained
from the Maybridge Chemical Company (Tintagel, Corn-
wall, UK).
16. Larsen, S. D.; Spilman, C. H.; Bell, F. P.; Dinh, D. M.;
Martinborough, E.; Wilson, G. J. J. Med. Chem. 1991, 34,
1721.
17. Modified from Ref.: Koresawa, M.; Okabe, T. Assay Drug
Dev. Technol. 2004, 2, 153.
18. Mecozzi, S.; West, A. P., Jr.; Dougherty, D. A. Proc. Natl.
Acad. Sci. U.S.A. 1996, 93, 10566.