Synthesis and evaluation of indenopyrazoles as cyclin-dependent kinase inhibitors. Part 4: Heterocycles at C3

Article abstract of DOI:10.1016/j.bmcl.2003.11.008

New indeno[1,2-c]pyrazol-4-one cyclin dependent kinase inhibitors have been disclosed. The most promising compounds are nanomolar enzyme inhibitors with excellent activity against tumor cells. The most advanced compound retains cell culture activity even in the presence of human serum proteins. The most advanced compound did not kill the normal fibroblast line AG1523.

Full text of DOI:10.1016/j.bmcl.2003.11.008

Bioorganic & Medicinal Chemistry Letters 14 (2004) 343–346
Synthesis and evaluation of indenopyrazoles as cyclin-dependent
kinase inhibitors. Part 4: Heterocycles at C3
Eddy W. Yue,* Susan V. DiMeo, C. Anne Higley, Jay A. Markwalder,
Catherine R. Burton, Pamela A. Benfield, Robert H. Grafstrom, Sarah Cox,
Jodi K. Muckelbauer, Angela M. Smallwood, Haiying Chen, Chong-Hwan Chang,
George L. Trainor and Steven P. Seitz
Bristol-Myers Squibb Company, Experimental Station, PO Box 80500, Wilmington, DE 19880-0500, USAy
Received 18 July 2003; accepted 5 November 2003
Abstract—New indeno[1,2-c]pyrazol-4-one cyclin dependent kinase inhibitors have been disclosed. The most promising compounds
are nanomolar enzyme inhibitors with excellent activity against tumor cells. The most advanced compound retains cell culture
activity even in the presence of human serum proteins. The most advanced compound did not kill the normal fibroblast line
AG1523.
# 2003 Elsevier Ltd. All rights reserved.
With recent advances in biology, the focus of research
has turned towards finding new targets for the treat-
ment of cancer. One such area involves the cell cycle
which functions as the vehicle for the crucial task of cell
division and proliferation.1 At the core of this machin-
ery are the cyclin-dependent kinases (CDKs) which
control the progression through the cycle.2 CDK2/E
and CDK4/D1 control the activity in G1 and entry into
S. CDK2/A along with CDK2/E regulates passage
through S and CDK1/Bcontrols the G2 checkpoint and
entry into mitosis. One of the better studied cell cycle
processes is the phosphorylation of the retinoblastoma
protein (pRb) by CDK4/D1. Once phosphorylated pRb
releases the transcription factor E2F leading to the
downstream activation of genes required for DNA
synthesis. Dysregulation of the Rb pathway is a key
event in transformation. Cyclins D, E and CDK4 have
also shown aberrant expression in tumors. Inhibitors of
CDKs have thus been suggested as a novel way to stop
the proliferation of tumor cells.
azole class of CDK inhibitors (1).3 Our previous studies
have revealed that ureas and semicarbazides are gen-
erally preferred at C5 while a variety of groups are tol-
erated at C3 including phenyls, heterocycles, and
alkyls.4,5 In this paper, we report on a series of C3
heterocycles substituted at C5 with semicarbazides.
The synthesis of indenopyrazoles proceeded according
to previously described chemistry.5 Treatment of 3-
nitrophthalic anhydride (2) with an appropriately sub-
stituted diketone (3) gave the triketone 4 (Scheme 1).
This reaction allowed the installation of a variety of
heterocyclic groups simply by varying the diketone (3,
R=heterocycle). These diketones in turn were made
from ethyl trifluoroacetate and the corresponding acetyl
precursor of the heterocycle using Claisen condensation
conditions.5 Triketone 4 was reduced to the aniline 5
We have been exploring the potential of the indenopyr-
* Corresponding author at present address: Incyte Corporation,
Experimental Station, PO Box 80336, Wilmington, DC 19880-0336,
USA. Tel.: +1-302-498-6902; fax: +1-302-425-2750; e-mail:
eddyyue@incyte.com
y
Present address: Bristol-Myers Squibb Company, Rt. 206 and Pro-
vince Line Road, Lawrenceville, NJ 08543, USA.
0960-894X/$ - see front matter # 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2003.11.008

344
E. W. Yue et al. / Bioorg. Med. Chem. Lett. 14 (2004) 343–346
thylthiazol-5-yl analogue (8i) was also quite potent in
CDK2/E and it exhibited good selectivity for CDK4/
D1. Unfortunately, the utility of these compounds was
limited by solubility.
In an attempt to improve solubility, semicarbazides
were synthesized. Data for semicarbazides of 4-amino-
morpholine are shown in Table 1. These compounds
exhibited comparable or improved binding affinity for
CDK2/E compared with the corresponding N,N-dime-
thyl semicarbazides. With the exception of 8n, all of the
compounds also had improved affinity for CDK4/D1.
Because of this increase, compounds such as 8k and 8l
had balanced profiles against CDK2/E and CDK4/D1.
However, the analogue which again stood out was the
2,4-dimethylthiazol-5-yl (8n) which had good binding
affinity for CDK2/E and maintained good selectivity
against CDK4/D1. Unfortunately, the solubility of
these semicarbazides was poor (8n: 4.4 mg/mL, 5%
mannitol).
Scheme 1. Reagents: (a) RC(¼O)CH2C(¼O)CF3 (3), Ac2O, Et3N,
25 ꢀC; (b) Zn, CaCl2, EtOH, H2O, reflux; (c) PhOC(¼O)Cl, Na2CO3,
acetone, 50 ꢀC; (d) R0=hydrazines, DMSO, 90 ꢀC; (e) H2NNH2,
EtOH, reflux; (f) (9)=R0CO2Ph, Et3N, DMSO, 90 ꢀC.
which allowed for functionalization at C5. Carbamate 6
was formed under standard conditions and upon treat-
ment with a variety of hydrazines (R0) gave semi-
carbazide triketones (7). Cyclization of triketone 7 gave
the desired indenopyrazole 8. In certain cases, the
synthesis could be reduced by one step when aniline 5
was treated directly with the preformed carbamate of
the hydrazine to be introduced (9).
The next series of compounds were derived from 1-
amino-4-methylpiperazine. Again, the most notable
analogue contained 2,4-dimethylthiazol-5-yl (8q) which
was still a potent inhibitor of CDK2/E but had lost
some of its selectivity against CDK4/D1 compared with
the other semicarbazides 8i and 8n. The switch of the
semicarbazide from morpholino to piperazino had a
dramatic effect on the solubility of 8q (2.4 mg/mL, 5%
mannitol).
The first semicarbazides examined were those derived
from 1,1-dimethylhydrazine (Table 1). As before, these
compounds showed greater affinity for CDK2/E than
CDK4/D1.5 Substitution at the 5-position of 2-thienyl
with small groups was preferred (8b and 8c). For the 3-
thienyl substitution the 5-chloro analogue proved to be
one of the most potent inhibitors (8f). The 2,4-dime-
With this data at hand, the focus shifted towards
examining the C3 heterocycle binding site in more
detail. Compound 8k was chosen because the ethyl ester
Table 1. C3 heterocyle, C5 semicarbazide SAR
Compd
R
R0
IC50 (nM)
CDK4/D1a
CDK2/Ea
HCT-116a
HCT-116 (protein)a,b
AG1523a
8a
8b
8c
8d
8e
8f
8g
8h
8i
8j
8k
8l
8m
8n
8o
8p
8q
Dimethylamino
Dimethylamino
Dimethylamino
Dimethylamino
Dimethylamino
Dimethylamino
Dimethylamino
Dimethylamino
Dimethylamino
Morpholin-4-yl
Morpholin-4-yl
Morpholin-4-yl
Morpholin-4-yl
Morpholin-4-yl
Thien-2-yl
5-(OMe)thien-2-yl
5-(Me)thien-2-yl
5-(CO2Et)thien-2-yl
Thien-3-yl
5-(Cl)thien-3-yl
2,5-(di-Me)thien-3-yl
Furan-2-yl
2,4-(di-Me)thiazol-5-yl
5-(Me)thien-2-yl
340
84
81
185
107
21
515
635
150
23
30
7
125
>370
30
36
32
25
28
24
7
26
26
4
10
31
10
<8
9
36
12
8
75
130
48
380
68
NT
NT
NT
NT
NT
NT
NT
NT
NT
NT
3200
NT
NT
500
NT
NT
260
905
1550
640
2800
605
615
1500
NT
64
110
160
NT
25
110
35
25
8
46
NT
285
>17,000
385
345
>19,000
NT
205
5-(CO2Et)thien-2-yl
5-(Cl)thien-3-yl
2,5-(di-Me)thien-3-yl
2,4-(di-Me)thiazol-5-yl
5-(CO2Et)thien-2-yl
2,5-(di-Me)thien-3-yl
2,4-(di-Me)thiazol-5-yl
4-(Methyl)piperazin-1-yl
4-(Methyl)piperazin-1-yl
4-(Methyl)piperazin-1-yl
90
46
24
14
>21,000
NT, not tested.
a Values correspond to n=2.
b Values determined in the presence of human serum albumin and a-1-acidglycoprotein.

E. W. Yue et al. / Bioorg. Med. Chem. Lett. 14 (2004) 343–346
Table 2. C3 thienylamide, C5 morpholino semicarbazide SAR
345
Compd
R
IC50 (nM)
HCT-116a
CDK4/D1a
CDK2/Ea
HCT-116 (protein)a,b
AG1523a
10a
10b
10c
10d
10e
10f
10g
10h
10i
10j
10k
10l
10m
10n
10o
10p
10q
2-(Dimethylamino)ethylamino
2-(Pyrrolidin-1-yl)ethylamino
2-(Piperidin-1-yl)ethylamino
2-(Morpholin-4-yl)ethylamino
Piperidin-1-yl
3-(Dimethylamino)piperidin-1-yl
4-(Dimethylamino)piperidin-1-yl
Piperazin-1-yl
4-(Methyl)piperazin-1-yl
4-(Ethyl)piperazin-1-yl
3-(Amino)pyrrolidin-1-yl
3-(Methylamino)pyrrolidin-1-yl
3-(Dimethylamino)pyrrolidin-1-yl
Azepan-1-yl
15
23
22
86
22
13
8
4
10
5
35
5
12
17
3
21
25
31
51
25
24
23
38
32
19
39
11
44
23
6
>800
475
20
860
52
188
210
700
110
63
>800
230
64
NT
NT
NT
NT
NT
680
255
NT
NT
NT
1000
NT
485
NT
NT
NT
330
NT
NT
63
16,750
11,825
>15,100
15,400
9940
14,275
NT
8500
2540
8325
>19,700
870
7100
53
195
19
[1,4]Diazepan-1-yl
4-(Methyl)-[1,4]diazepan-1-yl
4-(Ethyl)-[1,4]diazepan-1-yl
4
7
15
14
33
NT
NT, not tested.
a Values correspond to n=2.
b Values determined in the presence of human serum albumin and a-1-acidglycoprotein.
at the 5-position of the thiophene ring provided a useful
handle for derivatization. A large number of amines
were coupled with 8k, and some examples are shown in
Table 2. It was apparent that both primary and sec-
ondary amides inhibited CDK4/D1 and CDK2/E simi-
larly or better than the parent ester 8k. Compound 10a
was an important analogue as it led to amides derived
from cyclic amines (10e–q). By constraining the methyl-
ene units of the side chain of 10a, it was hoped that an
increase in binding affinity could be achieved. Indeed
the results for these amides, ranging from five- to seven-
membered rings showed an increase in binding affinity
for CDK4/D1. Even more significant was that for the
first time, CDK4/D1 selectivity was achieved and in the
case of 10h it was about 10-fold more selective than
CDK2/E. This selectivity appears to be influenced by
the presence of a distal basic nitrogen. For example,
compound 10e showed a relatively balanced inhibition
profile for CDK4/D1 versus CDK2/E whereas both
compounds 10h and 10i were more selective for CDK4/
Figure 1. The structure of 8q in the CDK2 ATP binding pocket. Pro-
tein carbons are colored light green and inhibitor carbons are colored
white.
D1.
An X-ray crystal structure of 8q complexed with CDK2
was obtained in order to determine the binding con-
formation. The conformation of 8q is very similar to the
C3-alkyl indenopyrazole previously determined5 and the
binding also occurs in the ATP pocket of the enzyme.
The major interaction involves the pyrazole nitrogens of
8q forming hydrogen bonds with Leu83 (Fig. 1). The
amine of Leu83 is 2.9 A from N1 and the carbonyl
oxygen of Leu 83 is 2.4 A from N2. The folding of the
C5 semicarbazide toward the C3 region is not as pro-
nounced as in the previous X-ray structure,5 however,
both the C5 and C3 groups are still partially solvent
exposed. The X-ray crystal structure of 8q is consistent
with the observed activity of CDK inhibitors. The
structure suggests there is ample room to accommodate a
variety of groups at either C5 or C3 and is consistent with
the fact that many such analogues are good inhibitors of
CDK2/E.
Further profiling of these compounds was done in cell-
based assays. A colon carcinoma-derived cell line,

346
E. W. Yue et al. / Bioorg. Med. Chem. Lett. 14 (2004) 343–346
Table 3. Enzymatic and cellular activity of 8q
uation. It displayed activity in CDK2/A and CDK1/B
and good selectivity against other kinase targets such as
the tyrosine kinase c-abl and protein kinases A and C
(Table 3). Compound 8q was also tested in a broad
panel of human and murine tumor cell lines. Very good
activity against these cell lines with several IC50 values
less than 50 nM was seen.
IC50 (nM)a
IC50 (nM)a
87
CDK2/A
CDK1/B6
c-abl
PKA
PKC
AG1523 (arrested)
B16-F1
H1299
13
HT29
HT1080
30
17,600
>42,000
<42,000
>21,000
160
L1210
MCF7
MiaPaCa2
NCI-H358
NCI-H460
PC3
105
23
26
18
104
125
25
The interesting profile of 8q resulted in extensive eval-
uation of its effects on the cell cycle and profiling in
vivo. The compound displayed good activity in HCT116
xenografts and in a MMTV neu transgenic model. These
results will be reported in detail in due course.
12
55
HMEC
Skut 1A
a Values correspond to n=2.
The current SAR studies have led to several significant
findings. Compound 8q was found to be quite active in
HCT116 even in the presence of plasma protein. Just as
important was the fact that it was inactive in AG1523
which may be an indicator of an enhanced therapeutic
window. In addition, 8q displayed good selectivity
against other kinase targets and very good activity in a
large panel of tumor cell lines. These results will allow
us to further refine our efforts in developing the optimal
CDK inhibitor in the indenopyrazole series.
HCT116, was used to determine the effectiveness of
these inhibitors in deterring cellular proliferation with
and without plasma proteins (human serum albumin
and a-1-acidglycoprotein). In addition, some of the
compounds were tested in AG1523, a normal human
fibroblast. AG1523 was chosen to give some indication
of the potential therapeutic window these inhibitors
possessed between proliferating cells versus arrested
normal cells. Most of the compounds in Table 1 showed
good translation from the enzymatic assays into the
cellular assay. Most notable were the thiazolyl analo-
gues 8n and 8q. Not only were they quite active in inhi-
biting HCT116, they retained their potency in the
presence of plasma proteins.
References and notes
1. (a) Lee, M.-H.; Yang, H.-Y. Cell. Mol. Life Sci. 2001, 58,
1907. (b) Owa, T.; Yoshino, H.; Yoshimatsu, K.; Nagasu,
T. Curr. Med. Chem. 2001, 8, 1487. (c) Sausville, E. A.;
Johnson, J.; Alley, M.; Zaharevitz, D.; Senderowicz, A. M.
Annals of the New York Academy of Sciences 2000, 910,
207. (d) McDonald, E. R., III; El-Deiry, W. S. Intl. J.
Oncol. 2000, 16, 871. (e) Fry, D.; Garrett, M. D. Cur.
Opin. Oncol. Endocrine Metab. Investig. Drugs 2000, 2, 40.
(f) Sielecki, T. M.; Boylan, J. F.; Benfield, P. A.; Trainor,
G. L. J. Med. Chem. 2000, 43, 1.
2. (a) Morgan, D. O. Annu. Rev. Cell Dev. Biol. 1997, 13,
261. (b) Pines, J. Semin. Cell Biol. 1994, 5, 399. (c) Hart-
well, L. H.; Kastan, M. B. Science 1994, 266, 1821. (d)
Sherr, C. J. Cell 1993, 73, 1059.
3. Nugiel, D. A.; Etzkorn, A.-M.; Vidwans, A.; Benfield,
P. A.; Boisclair, M.; Burton, C. R.; Cox, S.; Czerniak,
P. M.; Doleniak, D.; Seitz, S. P. J. Med. Chem. 2001, 44,
1334.
4. Nugiel, D. A.; Vidwans, A.; Etzkorn, A.-M.; Rossi, K. A.;
Benfield, P. A.; Burton, C. R.; Cox, S.; Doleniak, D.;
Seitz, S. P. J. Med. Chem. 2002, 45, 5224.
5. Yue, E. W.; Higley, C. A.; DiMeo, S. V.; Carini, D. J.;
Nugiel, D. A.; Benware, C.; Benfield, P. A.; Burton, C. R.;
Cox, S.; Grafstrom, R. H.; Sharp, D. M.; Sisk, L. M.;
Boylan, J. F.; Muckelbauer, J. K.; Smallwood, A. M.;
Chen, H.; Chang, C.-H.; Seitz, S. P.; Trainor, G. L. J.
Med. Chem. 2002, 45, 5233.
For the thienylamides of Table 2, the results in HCT116
showed moderate to good enzymatic to cellular transla-
tion. Unfortunately most of these thienylamides were
highly protein bound with IC50’s greater than 2500 nM
with the exception being 10p.
Analysis of the AG1523 data revealed that most of the
thienylamides tested had activity in AG1523 which was
deemed unacceptable (Table 2). For example compound
10p which had the best HCT116 protein-adjusted IC50
result was also quite potent against AG1523. Exami-
nation of the data in Table 1 revealed that semi-
carbazides derived from 1,1-dimethylhydrazine showed
poor selectivity for HCT116 over AG1523. The results,
however, improved with the cyclic semicarbazides.
Compound 8k had an IC50 in AG1523 greater than
17,000 nM. Although encouraging the HCT116 protein-
adjusted IC50 was still quite high. Further evaluation
revealed that both thiazolyl compounds, 8n and 8q were
inactive in AG1523 yet had activity in HCT116 even in
the presence of plasma proteins.
With its potent activity in HCT116 and large differential
in AG1523, compound 8q was selected for further eval-