Parallel synthesis and biological evaluation of 5,6,7,8- tetrahydrobenzothieno[2,3-d]pyrimidin-4(3H)-one cytotoxic agents selective for p21-deficient cells

Article abstract of DOI:10.1016/j.bmcl.2005.07.072

A novel series of inhibitors of cancer cell proliferation, selective against p21 cell cycle checkpoint-disrupted cells vs. cells with intact p21 checkpoint, were identified by high-throughput screening. Optimization of both ends of the lead molecule to improve potency, using parallel synthesis and iterative design, is described. The 2-(1,4-dibenzodioxane)-substituted derivative 14 was identified as a highly selective and potent agent displaying an IC₅₀ of 91 nM in the p21-deficient cell line.

Full text of DOI:10.1016/j.bmcl.2005.07.072

Bioorganic & Medicinal Chemistry Letters 15 (2005) 4731–4735
Parallel synthesis and biological evaluation of
5,6,7,8-tetrahydrobenzothieno[2,3-d]pyrimidin-4(3H)-one
cytotoxic agents selective for p21-deficient cells
Lee D. Jennings,a,* Scott L. Kincaid,a Yanong D. Wang,b
Girija Krishnamurthy,c Carl F. Beyer,d John P. McGinnis,d Miriam Miranda,d
Carolyn M. Discafanid and Sridhar K. Rabindrand
aDepartment of Exploratory Chemistry, Wyeth Research, Pearl River, NY 10965, USA
bDepartment of Medicinal Chemistry, Wyeth Research, Pearl River, NY 10965, USA
cDepartment of Biophysics and Enzymology, Wyeth Research, Pearl River, NY 10965, USA
dDepartment of Oncology Research, Wyeth Research, Pearl River, NY 10965, USA
Received 21 June 2005; revised 18 July 2005; accepted 25 July 2005
Available online 6 September 2005
Abstract—A novel series of inhibitors of cancer cell proliferation, selective against p21 cell cycle checkpoint-disrupted cells vs. cells
with intact p21 checkpoint, were identified by high-throughput screening. Optimization of both ends of the lead molecule to improve
potency, using parallel synthesis and iterative design, is described. The 2-(1,4-dibenzodioxane)-substituted derivative 14 was iden-
tified as a highly selective and potent agent displaying an IC50 of 91 nM in the p21-deficient cell line.
ꢀ 2005 Elsevier Ltd. All rights reserved.
Cell cycle checkpoints are signal transduction pathways,
which ensure that each step in the cell division cycle, has
been successfully completed before the onset of the next
phase.1 Loss of checkpoint control is a hallmark of tu-
mor cells, as it is thought to increase the mutation rate,
allowing a more rapid progression of cells to the tumor-
igenic state.1,2 However, inactivation of these check-
points results in aberrant responses to cellular damage.
This failure of checkpoint responses in malignant cells
can be exploited in cancer drug discovery. Identification
of compounds that selectively kill checkpoint-deficient
cells can be expected to preferentially target tumor cells,
while sparing normal cells.3,4
environmental insults causing DNA damage7 or micro-
tubule perturbation.8 Disruption of this checkpoint by
deletion of the p21 protein leads to failure of the cell
to arrest, leading to endoreduplication, and finally, to
apoptosis.9,10 p21-deficient cells show increased sensitiv-
ity, compared to isogenic p21-proficient parental cells, to
a variety of clinically used antineoplastic drugs, validat-
ing the role of checkpoints in chemosensitivity.
We have prepared p21-deficient cells from human colon
carcinoma cell line HCT116 by targeted deletion of the
p21 gene.9 HCT116, the parental cell lines from which
the p21ꢀ/ꢀ cells were derived, is one of the few colon
cancer cell lines with an apparently intact p21 check-
point. We anticipate that a search for compounds that
preferentially kill the p21-deficient cells, compared to
the p21-proficient cells, will enable us to identify com-
pounds that are not simply toxic, but have selectivity to-
ward cells defective in checkpoint control. Targeting
these cells should improve the likelihood of obtaining
lead molecules that preferentially inhibit the growth of
tumor cells rather than normal cells, a requirement for
an anticancer agent. Furthermore, since loss of p21
checkpoint is a hallmark of cancer (primarily through
The p53 tumor suppressor gene is a major regulator of
the G1/S-phase checkpoint and one of the most com-
monly mutated genes in human cancer.5,6 p21, a down-
stream effector of p53, inhibits the cyclin-dependent
kinases and arrests cell cycle progression in response to
Keywords: Inhibitors of cell cycle progression; Parallel synthesis;
Tubulin binding.
*
Corresponding author. Tel.: +1 845 602 3618; fax: +1 845 602
3045; e-mail: jenninl@wyeth.com
0960-894X/$ - see front matter ꢀ 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2005.07.072

4732
L. D. Jennings et al. / Bioorg. Med. Chem. Lett. 15 (2005) 4731–4735
mutations in the upstream activator, p53), these com-
pounds are widely used in treating a variety of them.
O
NC
sulfur
morpholine
EtOH, 60 oC, 1 hr
CONH2
R1
R1
NH2
NH2
O
High-throughput screening using the p21-knockout iso-
genic cell screen described above led to the identification
of two leads, pyrazolo[1,5,a]pyrimidin-7-yl phenyl amide
111 and 2-(3,4,5-trimethoxyphenyl)-tetrahydrobenzothi-
eno[2,3-d]pyrimidin-4(3H)-one 2.12 Preliminary SAR
studies by others in our group13 indicated that either the
4-methoxy or 3,4,5-trimethoxy substitution pattern on
the phenyl ring of 2 was necessary for its activity. Replace-
ment of the methoxy groups with chloro, methyl, or nitro,
or deletion or rearrangement of the methoxy groups gave
much less active compounds. The exocyclic oxygen was
important to activity and conservative modifications to
the tetrahydrobenzene ring were tolerated, giving slightly
more active compounds. We presumed that a more
thorough investigation of both the phenyl group and
the tetrahydrobenzene ring SAR could be carried out
using a parallel synthesis. Here, we report both the
strategies adopted by us and the results of our effort.
S
3a-j
R2
CHO
O
R1
NH
R2
HCl (cat)
N
S
n-butanol
80 oC, 10 hr
4-23
Scheme 1. General route for the preparation of thienopyrimidinones.
The collection of aldehydes available to us for use in syn-
thesis was so large that a physical property-based design
approach to select building blocks from this large set
(i.e., limiting the number of building blocks) had to be
adopted.15 First, physical properties [MW, hydrogen
bond donor and acceptor counts, flexible bond counts,
clogP, CMR, PSA, and MW of substituents in the ortho,
meta, and para positions (where applicable)] of the 133
available benzaldehydes and heteroaryl aldehydes were
calculated. Second, three principal components (vectors
of property space) of this resulting data set were identified
via multivariate analysis using SIMPCA-P.16,17 Two
selections of aldehyde building block sets were made using
this information: (1) a ÔdiversityÕ set of 12 aldehydes was
selected by D-optimal selection using the principal
components and (2) a ÔsimilarityÕ set of 13 aldehydes
was selected on the basis of closest similarity to 4-meth-
oxybenzaldehyde. Additionally, a third set was selected
by hand from a set of commercially available heterocyclic
aldehydes. In total, 47 aldehydes were selected. Products
of the reaction of 2-amino-4,5,6,7-tetrahydrobenzo[b]-
thiophene-3-carboxamide (3a) with the selected alde-
hydes were prepared18 and tested for both activity and
selectivity for the p21ꢀ/ꢀ cells over the p21+/+ cells
(Table 1). Alone out of the entire compound set, the
new thienopyrimidinone 4, having the 2-(1,4-dibenzodi-
oxane) substituent, evoked cell death in the p21ꢀ/ꢀ cell
line at a submicromolar dose (IC50 p21ꢀ/ꢀ cells,
0.42 lM, n = 2), with excellent selectivity for p21ꢀ/ꢀ cells
over the p21 proficient cells. This discovery represented a
substantial improvement over the lead.
H
N
O
O
NH
OMe
N
N
N
S
OMe
OMe
N
O
S
2
1
The compounds in this study were all made following the
published procedure for the preparation of thienopyri-
midinones.12 2-Amino-thiophene-3-carboxamides were
prepared from cyclic ketones via the Gewald thiophene
synthesis14 and the fused ring thiophene intermediates
were, in turn, condensed with aldehydes in parallel to
give the desired thienopyrimidinones (Scheme 1). As
the capacity of the team to assay new compounds was
limited to a certain extent, we rejected the idea of making
a full combinatorial library using large sets of ketone and
aldehyde building blocks. Instead, we first investigated
the effect of variation of the aromatic substituent
(right-hand side) on activity, while keeping the left-hand
side of the molecule constant.
Table 1. Inhibition of p21+/+ and p21ꢀ/ꢀ cell proliferation by 5,6,7,8-tetrahydrobenzothieno[2,3-d]pyrimidin-4(3H)-ones19,21
O
NH
R1
N
S
Compound
R1
p21+/+ IC50 (lM)
p21ꢀ/ꢀ IC50 (lM)
Ratio
2
4
3,4,5-Trimethoxyphenyl
>20
>20
8.4
2.8
0.42
1.7
>9
>47
5
2,3-Dihydro-benzo[1,4]dioxin-6-yl
4-Hydroxy-naphthalen-1-yl
[2,20]Bithiophenyl-5-yl
5
6
>20
>20
>20
>20
>20
9.3
9.3
>2
>2
>2
>2
>2
7
2-Bromo-3,4-dimethoxyphenyl
3-Acetyloxyphenyl
8
12
9
10
3-Chloro-4-hydroxy-5-methoxyphenyl
4-Isopropenyl-cyclohex-1-enyl
12.5
12.8

L. D. Jennings et al. / Bioorg. Med. Chem. Lett. 15 (2005) 4731–4735
4733
In the second phase of our optimization of 2, we pre-
pared a 9 · 6 combinatorial array comprised of the reac-
tion of nine differently substituted 2-amino thiophene-3
carboxamides 3b–j, prepared from nine commercially
available cyclohexanones and piperidin-4-ones, with
six aldehydes. Thirty-two out of 54 possible reaction
products were successfully prepared and purified in par-
allel. Result from testing these products in our assay
(Table 2) confirmed the desirability of the 2-(2,3-
benzo[1,4]dioxinyl)-thienopyrimidin-4-one substituent
group and revealed further novel compounds 11–23 use-
ful for inducing apoptosis in p21ꢀ/ꢀ cells in preference
to p21+/+ cells. Compound 11 (IC50 p21ꢀ/ꢀ cells,
0.091 lM, n = 3) represented a second significant ad-
vance in potency over the lead. Furthermore, enrich-
ment of the second library in active compounds
(IC50 < 10 lM), relative to the first library, was pro-
nounced, a testimony to our success at identifying
right-hand side substituents in the first iteration that
improved activity.
The most potent compounds from the two libraries were
selected for further characterization in a panel of human
colon carcinoma cell lines (Table 3). Compounds dis-
played moderate to good potency across the panel.
Potency seemed to track with potency of the lethal effect
on p21ꢀ/ꢀ cells. To assess the therapeutic potential of
the best compound from this series, 11 was tested in a
human LoVo cell xenograft model in nude mice. Disap-
pointingly however, 11 was not active at either 30 or
Table 2. Inhibition of p21+/+ and p21ꢀ/ꢀ cell proliferation by fused thieno[2,3-d]pyrimidin-4-ones20,21
O
R1
NH
R2
N
S
Compound
R1
R2
p21+/+ IC50 (lM)
p21ꢀ/ꢀ IC50 (lM)
Ratio
42
O
N
N
N
N
N
N
11
2,3-Dihydrobenzo[1,4]dioxin-6-yl
3.8
0.091
O
O
12
13
14
15
16
17
18
19
20
21
22
23
2-Benzo[1,3]dioxol-5-yl
4-Methoxyphenyl
6.8
>20
>20
>20
>20
11
0.68
1.9
1.9
2
10
>10
>10
>10
>6
O
O
O
O
4-Isopropenylcyclohex-1-enyl
3,4,5-Trimethoxyphenyl
4-(Methylthio)phenyl
O
O
O
O
3.6
0.8
8.9
1.8
2.8
2.2
2.8
5.6
O
Ph
Ph
O
4-Methoxyphenyl
14
N
O
O
3,4,5-Trimethoxyphenyl
2,3-Dihydrobenzo[1,4]dioxin-6-yl
4-Methoxyphenyl
>20
20
>2
N
O
N
11
O
O
N
>20
>20
>20
>20
>7
3,4,5-Trimethoxyphenyl
3,4,5-Trimethoxyphenyl
2,3-Dihydrobenzo[1,4]dioxin-6-yl
>9
>7
>4

4734
L. D. Jennings et al. / Bioorg. Med. Chem. Lett. 15 (2005) 4731–4735
Table 3. Activity comparison in the colon cell line panel
Compound LoVo SW620 DLD1
IC50 (lM) IC50 (lM) IC50 (lM) IC50 (lM)
2. Hanahan, D.; Weinberg, R. A. Cell 2000, 100, 57.
3. Torrance, C. J.; Agrawal, V.; Vogelstein, B.; Kinzler, K.
W. Nature Biotechol. 2001, 19, 940.
4. Prendergast, G. C. Nature Biotech. 2001, 19, 919.
5. Levine, A. J. Cell 1997, 88, 323.
6. Stewart, Z. A.; Pietenpol, J. A. Chem. Res. Toxicol. 2001,
14, 243.
7. Vogelstein, B.; Lane, D.; Levine, A. J. Nature 2000, 408,
307.
8. Stewart, Z. A.; Leach, S. D.; Pietenpol, J. A. Mol. Cell.
Biol. 1999, 19, 205.
9. Waldman, T.; Kinzler, K. W.; Vogelstein, B. Cancer Res.
1995, 55, 5187.
HT29
2
4
0.14
0.08
0.02
0.05
0.11
0.18
0.17
0.19
0.19
0.12
0.03
0.07
0.14
0.24
0.17
0.42
0.29
0.15
0.04
0.09
0.14
0.29
0.18
0.41
0.24
0.18
0.04
0.12
0.15
0.36
0.15
0.42
11
12
13
15
17
19
10. Waldman, T.; Lengaur, C.; Kinzler, K. W.; Vogelstein, B.
Nature 1996, 381, 713.
11. Gopalsamy, A.; Ellingboe, J. W.; Tsou, H.-R.; Zhang, N.;
Wang, D.; Honores, E.; Johnson, S.; Powell, D.; Miranda,
M.; McGinnis, J. P.; Rabindran, S. K. Bioorg. Med.
Chem. Lett. 2005, 5(6), 591.
12. Cruceyra, A.; Gomez Parra, V.; Madronero, R. Anales de
Quimica (1968–1979) 1975, 71, 103, The authors actually
reported the preparation of 2-(3,4,5-trimethoxyphenyl)-
2,3,5,6,7,8-hexahydro-1H-benzo[4,5]thieno[2,3-d]pyrimidin-
4-one (CAS # 37556-33-5) but probably, in fact, synthesized
the tetrahydrothieno pyrimidinone 2.
13. Wang, Y. D.; Johnson, S.; Powell, D.; McGinnis, J.;
Miranda, M.; Rabindran, S. K. Bioorg. Med. Chem. Lett.
2005, 15, 3763.
Table 4. Binding of thienopyrimidinone cell proliferation inhibitors to
tubulin in a fluorescence binding assay22
Compound
Kd(apparent)
11
12
16
1.8 lM 0.14
4.0 lM 0.4
1.6 lM 0.26
50 mg/kg (oral dosing, once-daily). Solubility of 11 was
low (4 lg/mL at pH 7.4). The lack of in vivo efficacy
may be on account of the low bioavailability caused
by poor physical properties. However, a pharmacokinet-
ic study on 11 was not done.
14. Gerwald, K.; Schinke, E.; Bottcher, H. Chem. Ber. 1966,
99, 94.
The anti-tumor compound colchicine is known to work
by inhibiting microtubule formation. Colchicine is also
an inhibitor of p21ꢀ/ꢀ cell proliferation (IC50 p21ꢀ/ꢀ
0.02 lM) with approximately 10-fold selectivity for
p21 deficient cells over p21 proficient cells. We were
interested knowing whether or not the thienopyrimidi-
none series killed p21ꢀ/ꢀ cells by the same mechanism.
Using a fluorescence binding assay,22 11 was found to
bind saturably to the tubulin dimer with a one-to-one
stoichiometry and showed a Kd(apparent) of 1.8 lM. How-
ever, tubulin binding within the series (Table 4) was
essentially invariant, while potency of induction of
p21ꢀ/ꢀ cell apoptosis varied by orders of magnitude,
indicating that there is no correlation between tubulin
binding and induction of cell death in this series.
15. Martin, E. J.; Blaney, J. M.; Siani, M. A.; Spellmeyer, D.
C.; Wong, A. K.; Moos, W. H. J. Med. Chem. 1995, 38,
1431.
16. (a) Eriksson, L.; Johanasson, E.; Kettaneh-Wold, N.;
Wold, S. Introduction to Multi- and Megavariate Data
Analysis Using Projection Methods (PCA
% PLS);
Umetrics AB: Umea, Sweden, 1999; (b) Calculations
performed using software product SIMPCA-P (Umetrics
AB, Umea, Sweden).
17. The most significant property contributors to the principal
components were as follows: [t]1, correlation to clogP and
reciprocal correlation to #HBD, #HBA, PSA, and
MWortho substituent; [t]2, correlation to MW; [t]3, correla-
tion to MWpara
MWmeta substituent
and reciprocal correlation to
substituent
.
18. 2-Amino-4,5,6,7-tetrahydro-benzo[b]thiophene-3-carbox-
amide (3a) (29 mg, 0.15 mmol) was suspended in 1 ml n-
butanol. 1,4-Benzodioxane-6-carboxaldehyde (24.6 mg, 1
equiv) and 10 lL conc. HCl were added and the reaction
mixture was heated to 80 ꢁC for 10 h. The reaction
product was collected by filtration, washed with water,
and dried to give 30 mg 4 [LC/MS (Hewlett Packard
1100 MSD with ChemStation Software, Keystone Aqua-
sil C18 50 mm · 2 mm column, 5 lm particle size, at
40 ꢁC, 10 mM NH4OAc–acetonitrile solvent system at
0.8 mL/min flow rate, 254 nm DAD detection, API-ES
scanning mode, fragmentor 70 mV): m/z 341 (M+H);
retention time 2.59 min]. Compounds that were not
judged to be sufficiently pure (>90%) by LC were
purified by RP-HPLC with automatic fraction collection
(Gilson Semi-Preparative HPLC system with Unipoint
Software v. 1.71, Phenomenex C18 Luna column, 21.2
mm · 100 mm, 5 lm particle size, water–acetonitrile sol-
vent system with added 0.05% NH4OH buffer, at
22.5 mL/min).
In summary, we have successfully used a parallel synthe-
sis to rapidly improve the potency of an initial lead. A
physical property-based approach was used to select
the building blocks for synthesis. Although the preferred
2-(1,4-dibenzodioxane) substituent was selected on ac-
count of its similarity to the reference, 4-methoxybenzal-
dehyde, it was not clear from this study if a selection
based on diversity or one based on similarity to a refer-
ence point was preferred. Probably, a combination
approach, as practiced here, is the best. This effort has
demonstrated the value of a parallel synthesis strategy
for the optimization of leads which led to the identifica-
tion of several analogs with submicromolar activity in a
panel of colon cell lines.
References and notes
19. Compounds not shown in the table had IC50 > 20 lM on
80S14, as well as HCT116 cell lines.
1. Paulovich, A. G.; Toczyski, D. P.; Hartwell, L. H. Cell
1997, 88, 315.
20. Compounds not shown on the table had IC50 > 10 lM on
80S14 as well as HCT116 cell lines.

L. D. Jennings et al. / Bioorg. Med. Chem. Lett. 15 (2005) 4731–4735
4735
21. p21+/+ and p21ꢀ/ꢀ isogenic cells and other colon
carcinoma cell lines (LoVo, SW620, DLD1, and HT29)
were plated in 96-well tissue culture plates. The following
day, various dilutions of compounds were added and cells
were cultured for an additional 4 days (colon cell panel) or
5 days (isogenic cell lines). Cell survival was determined
using sulforhodamine B, a protein binding dye. The
concentration of compound that inhibits cell growth by
50% (IC50) was estimated from cytotoxicity curves. When
multiple independent experiments were performed, IC50s
shown represent the mean value.
22. Krishnamurthy, G.; Cheng, W.; Lo, M.-C.; Aulabaugh,
A.; Raznikov, V.; Ding, W.; Loganzo, F.; Zask, A.;
Ellestad, G. Biochemistry 2003, 42, 13484.