Thieno[3,2-d]pyrimidin-4(3H)-one derivatives as PDK1 inhibitors discovered by fragment-based screening

Article abstract of DOI:10.1016/j.bmcl.2012.04.080

Ligand efficient fragments binding to PDK1 were identified by an NMR fragment-based screening approach. Computational modeling of the fragments bound to the active site led to the design and synthesis of a series of novel 6,7-disubstituted thienopyrimidin

Full text of DOI:10.1016/j.bmcl.2012.04.080

Bioorganic & Medicinal Chemistry Letters 22 (2012) 4023–4027
Contents lists available at SciVerse ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Thieno[3,2-d]pyrimidin-4(3H)-one derivatives as PDK1 inhibitors discovered
by fragment-based screening
Angeline C.-H. Lee , Pondy Murugappan Ramanujulu , Anders Poulsen, Meredith Williams,
Stéphanie Blanchard, Diana M. Ma, Zahid Bonday, Kay Lin Goh, Kee Chuan Goh, Miah Kiat Goh,
⇑
Jeanette Wood, Brian W. Dymock
S*BIO PTE LTD, 1 Science Park Road, #05-09 The Capricorn, Singapore Science Park II, Singapore 117528, Singapore
a r t i c l e i n f o
a b s t r a c t
Article history:
Ligand efficient fragments binding to PDK1 were identified by an NMR fragment-based screening
approach. Computational modeling of the fragments bound to the active site led to the design and syn-
thesis of a series of novel 6,7-disubstituted thienopyrimidin-4-one compounds, with low micromolar
inhibitory activity against PDK1 in a biochemical enzyme assay.
Received 10 March 2012
Revised 13 April 2012
Accepted 17 April 2012
Available online 25 April 2012
Ó 2012 Elsevier Ltd. All rights reserved.
Keywords:
3-Phosphoinositide dependent protein
kinase-1 (PDK1)
Fragment screening
Kinase inhibitor design
Ligand efficiency
3-Phosphoinositide dependent protein kinase-1 (PDK1) is
crucial for the activation of a series of AGC family kinases and as
a result plays an important role in a number of cellular signaling
pathways. Significantly, it is involved in the PI3K/Akt pathway dis-
regulation which, either through upregulation or mutation, has
been implicated in the development and progression of a range
of cancers. Acting downstream of PI3K, PDK1 activates Akt result-
ing in a range of signaling events linked to cell growth, survival and
proliferation. As a result, the search for inhibitors of PDK1 has at-
tracted much attention as potential therapeutic agents for the
treatment of cancer as well as other diseases.1–4
Fragment-based drug design has become an important and
powerful tool for the discovery of new drug leads.5–9 Fragment
screening involves the use of technologies which can detect weak,
but specific, binding of low molecular weight (100–300 Da) ligands
to a protein. Once the core binding motifs have been identified
they may be elaborated in a variety of ways, most commonly by
addition of new substituents to improve potency. Due to the weak
binding constants of the fragments, conventional screening meth-
ods are not always suitable to detect them. Among other methods,
NMR spectroscopy has been used broadly to study fragment bind-
ing due to its high sensitivity of detection within the affinity range
of fragments (100
M–10 mM).10,11
l
In this case, NMR screening was carried out on a carefully
selected focused library of commercially available compounds
with a view to identifying suitable fragments that bind to the
PDK1 enzyme. This work was carried out as part of a wider
program searching for orally available PDK1 inhibitors, for example
compound 1 which was moderately potent against PDK1, active in
prostate cell line PC3 and 65% orally bioavailable in mice (Fig. 1).12
Each screened fragment was selected from commercial and
in-house sources with potential to bind to the kinase hinge region.
A 500 MHz cryoprobe equipped NMR instrument was used for the
primary screen, employing a Water-LOGSY (Water-Ligand ob-
served via Gradient Spectroscopy) pulse sequence.7 The fragments
were screened as mixtures of 6 or 7 compounds in a single NMR
experiment. Compound mixtures were selected so that the diag-
nostic signals did not overlap. In this way, the library could be
screened more rapidly, in some cases, even identifying multiple
hits within one NMR experiment.
Through a series of Water-LOGSY experiments, we identified
several suitable fragments which then underwent a competition
screen with the addition of 1, an ATP competitive ligand. This com-
petition experiment identified fragments which bound to the ATP
site of the kinase. These confirmed competitive fragment hits were
then tested in a PDK1 ATPase biochemical assay.13 The 10 preferred
hits are shown in Table 1 with ligand efficiency (LE)14 and IC50
⇑
Corresponding author at present address: National University of Singapore,
Singapore. Tel./fax: +65 66012495.
E-mail address: phadbw@nus.edu.sg (B.W. Dymock).
These authors contributed equally to this work.
0960-894X/$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.bmcl.2012.04.080

4024
A. C.-H. Lee et al. / Bioorg. Med. Chem. Lett. 22 (2012) 4023–4027
on the basis of the extensive prior art and anticipated selectivity
OMe
challenges. Indole carboxylic acid 2 was not favoured for similar
reasons as well as a concern that the acid moiety would be a
required function. Assessment of fragment binding modes via frag-
ment docking studies15 contributed to our prioritization decisions.
Most compounds had an identifiable donor-acceptor motif which
inspired confidence in the postulated binding mode (see Fig. 2).
Fused pyrazole lactone 8 has a very good LE but we were unsure
of the stability of the lactone and wanted to avoid modification
to the core scaffold in the first instance. Benzthiazole 9 has the po-
tential to undergo deacetylation to the potentially toxic amino-thi-
azole so it was not selected for initial work. To a great extent these
selections are based on the quality of the hits, if faced with a num-
ber of attractive scaffolds the criteria for selection can be tightened.
The remaining hits, thienopyrimidinones, 3–5, pyrazolopyrazinone
O
HN
N
PDK1 IC50 0.15 µM
PC3 IC50 0.58 µM
F% = 65 (mouse)
N
N
N
N
H
N
H
1
Figure 1. Orally bioavailable purine series.
data. A range of structural types are represented all with LE >0.30
and IC50s in the range 50 M–1 mM.
Selection of the preferred hits to take forward into synthetic
analog studies is a critical decision. Purines 1 and 7 were rejected
l
Table 1
Competitive fragment hits
Compound
Chemical structure
MW
205
LEa
PDK1 IC50 (l
M)13
O
N
1
0.38
92
N
N
N
H
N
F
O
2
3
179
180
0.38
0.31
297
828
N
H
OH
N
S
N
N
N
H
H
H
O
N
4
5
152
208
0.42
0.39
492
55
S
S
O
N
O
O
H
N
N
N
6
7
150
241
0.43
0.34
472
N
H
O
HN
61
N
N
N
H
N
8
9
166
210
0.41
0.31
302
402
N
N
H
N
O
F
O
NH
S
O
O
NH
10
189
0.36
273
N
N
a
LE =
D
G/N, where N = number of heavy (non-hydrogen) atoms, units kcal molÀ1 per heavy atom.

A. C.-H. Lee et al. / Bioorg. Med. Chem. Lett. 22 (2012) 4023–4027
4025
Figure 2. Glide docking of 5.15 Top: Fragment 5 (thick tube with green carbon) docked into the PDK1 ATP-binding site (thin tube with gray carbon). Hydrogen bonds are
shown as purple/black dashed lines. The aromatic core of the scaffold has hydrophobic contacts to Leu212, Thr222 and Val143 in the bottom of the binding site, Leu159 the
gatekeeper residue as well as Leu88, Val96 and Ala109 in the top of the binding site. Bottom: Fragment 5 docked into PDK1 which is shown as a grey surface. There is space for
large or bulky substituents on the thiophene carbons (R1 and R2, Table 2).
6 and pyrazole 10 had similar binding modes through the pyrimid-
inone and were all attractive starting points. A balance of structural
novelty, synthetic feasibility, potency and high ligand efficiency
prompted us to select the thieno[3,2-d]pyrimidin-4(3H)-ones 4
and 5 as our first-choice start points for medicinal chemistry and
we now report their initial evolution to single-digit micromolar
leads.
Sparse literature reports increased our confidence that frag-
ments 4 and 5 were quite novel hence more likely to be developed
into novel structural types.
Initial synthetic efforts were focused on adding functionalities
to the thiophene carbons. Synthetic targets were selected with
the aid of computational docking using the reported X-ray struc-
ture of PDK1.15 With reference to Figure 2, the docking clearly
shows that the pyrimidinone motif is forming key hydrogen bonds
to the hinge region of the kinase. Opportunities for growth of the
compound are greatest from the thiophene carbons. Table 2 shows
a selection of mono- and di-substituted derivatives of 4 and 5 pre-
pared on the basis of the binding mode analysis.
The preparation of the core scaffold 4 was carried out in two
steps according to the literature.16 Compounds 14 and 15 are com-
mercially available. Methyl derivative 16 was prepared according
to the literature.16 Nitration of 4 gave 11 (Scheme 1), which, fol-
lowing reduction of the nitro group, gave amine 12 in good yield.
Further modifications of this amine, such as reductive amination
and acylation, were performed to produce novel analogs 17, 18
and 19 (Table 2), which exhibited improved activity against
PDK1 (IC50 = 154, 109 and 95 lM, respectively).
Intermediate 12 can be further derivatized through halogena-
tion to yield 13,16 which was subsequently used as a core interme-
diate for Suzuki coupling with various boronic acids or esters
(Scheme 2). Compound 20, prepared via Suzuki coupling using

4026
A. C.-H. Lee et al. / Bioorg. Med. Chem. Lett. 22 (2012) 4023–4027
Table 2
Synthesised thienopyrimidin-4-one derivatives
R1
N
R2
N
S
H
O
Compound
R1
H
R2
H
MW
152
cLogPa
PDK1 IC50
492
(l
M)11
LEb
4
0.12
0.42
5
H
H
208
258
1.95
2.16
55
86
0.42
0.31
14
O
F
15
16
17
H
246
166
181
2.37
0.62
0.52
122
461
154
0.32
0.42
0.36
CH3
H
H
NH
18
19
N
H
195
271
0.80
1.40
109
95
0.42
0.29
NH
H
H
O
20
22
Ph
228
260
2.01
2.56
15
0.41
0.30
CH3
Ph
106
F
F
F
23
24
322
382
3.65
3.33
33
11
0.27
0.25
O
O
25
26
348
312
5.06
2.00
39
0.24
0.33
F
F
N
5.2
NH
a
Calculated using ChemDraw Ultra v8.0.
LE =
b
D
G/N, where N = number of heavy (non-hydrogen) atoms, units kcal molÀ1 per heavy atom.
Intermediate 21 was used to generate a series of di-substituted
analogs, 22–26 (Scheme 2).
NO2
NH2
N
N
O
N
O
Table 2 depicts PDK1 enzyme potency for the key fragments 4
and 5 and synthesized compounds along with their molecular
weight, calculated LogP and LE. Modification of R2 with the bulky
lipophilic tert-butyl in the fragment hit 5 realized nearly an order
of magnitude increase in potency with complete retention of LE.
Although aromatic ring bearing 14 and 15 were also more potent
they suffered from a lower LE due to their increase in MW for mod-
est potency gains. Initial R1 modifications with small groups or the
large benzamide (16–19) were only slightly more active. However
phenyl derivative 20 enjoyed a >30-fold potency increase over 4
while completely maintaining LE. Unfortunately this discovery at
R1 is not additive with R2: when the phenyl at R1 is combined with
4-fluorophenyl at R2 the potency dropped slightly with a concom-
itant significant drop in LE due to the higher MW. Reducing the size
of R1 back to methyl (22) did not provide any benefit and increas-
ing the size of R1 in combination with R2 only served to increase
MW, cLogP and decrease LE. However pyrazole 26 exhibited the
a
b
N
N
N
S
S
S
H
H
H
O
4
11
12
R
R'
N
N
O
17 R = H; R' = Me
18 R = R' = Me
c
N
19 R = H; R' = COPh
S
H
Reagents and reaction conditions: a) conc. HNO3, conc. H2SO4,
100 °C, 91%. b) Fe powder, NH4Cl, EtOH, H2O, 80 °C, 78%. c)
HCHO (aq), Na(OAc)3BH, CH2Cl2, 31% (17) and 55% (18);
PhCOCl, Et3N, CH2Cl2, 89% (19).
Scheme 1. Preparation of R1 nitrogen analogs.
phenyl boronic acid, exhibited a marked increase in inhibitory
best potency of this series with PDK1 IC50 of 5.2 lM, acceptable
activity against PDK1 with an IC50 of 15
l
M. On the other hand,
cLogP of 2.0, MW 312 and LE of 0.33.
We have shown that elaboration of a low molecular weight
fragment resulted in an improved potency from 492 M in 4 to
fragment hit 5 with a tert-butyl group at R2 also showed good
PDK1 activity. Hence, it was logical to next explore functionaliza-
tion of both positions of the thiophene ring (Table 2).
l
5.2 lM in lead compound 26. Further work will focus on under-
Compound 15 was used as a starting point for accessing di-
substituted ring systems. Bromination can be achieved in the pres-
ence of bromine and acetic acid at room temperature (Scheme 1).
standing the interplay between R1 and R2 while aiming to drive
down potency, maintaining LE above 0.3 and assessing enzyme
selectivity and cellular potency.

A. C.-H. Lee et al. / Bioorg. Med. Chem. Lett. 22 (2012) 4023–4027
4027
References and notes
Br
1. Vivanco, I.; Sawyers, C. L. Nat. Rev. Cancer 2002, 2, 489.
N
O
N
O
a
b
2. Maurer, M.; Su, T.; Saal, L. H.; Koujal, S.; Hopkins, B. D.; Barkley, C. R.; Wu, J.;
Nandula, S.; Dutta, B.; Xie, Y.; Chin, Y. R.; Kim, D.; Ferris, J. S.; Gruvberger-Saal,
S. K.; Laakso, M.; Wang, X.; Memeo, L.; Rojtman, A.; Matos, T.; Yu, J. S.; Cordon-
Cardo, C.; Isola, J.; Terry, M. B.; Toker, A.; Mills, G. N.; Zhao, J. J.; Murty, V. V. V.
S.; Hibshoosh, H.; Parsons, R. Cancer Res. 2009, 69, 6299.
3. Tsukamoto, Y.; Nakada, C.; Noguchi, T.; Tanihawa, M.; Nguyen, L. T.; Uchida, T.;
Hijiya, N.; Maysuura, K.; Fujioka, T.; Seta, M.; Moriyama, M. Cancer Res. 2010,
70, 2339.
4. Lim, M. A.; Yang, L.; Zheng, Y.; Wu, H.; Dong, L. Q.; Liu, F. Oncogene 2004, 23,
9348.
5. Fragment-Based Drug Discovery: A Practical Approach; Zartler, E. R., Shapiro, M. J.,
Eds.; John Wiley & Sons, Ltd: Chichester, 2008.
12
N
N
S
S
H
H
13
20
Reagents and reaction conditions: a) i) NaNO2, HBr, 0 °C. ii) CuBr,
HBr, 90 °C, 52%. b) Phenylboronic acid, Pd(dppf)Cl2:CH2Cl2, aq.
Na2CO3, PhMe/MeOH, 100 °C, 63%.
6. Bembenek, S. D.; Tounge, B. A.; Reynolds, C. H. Drug Discovery Today 2009, 14,
278.
7. Dalvit, C. J. Biomolecular NMR 2001, 21, 349.
Br
N
O
N
O
a
8. Structure-Based Drug Discovery: An Overview; Hubbard, R. E., Ed.; RSC
Publishing, 2006; pp 142–172.
9. Fattori, D. Drug Discovery Today 2004, 9, 229.
10. Hubbard, R. E.; Davis, B.; Chen, I.; Drysdale, M. J. Curr. Top. Med. Chem. 2007, 7,
1568.
F
F
N
N
S
S
H
H
15
21
11. Zartler, E. R.; Mo, H. Curr. Top. Med. Chem. 2007, 7, 1592.
12. Blanchard, S.; Soh, C. K.; Lee, C. P.; Poulsen, A.; Bonday, Z.; Goh, K. L.; Goh, K. C.;
Goh, M. K.; Pasha, M. K.; Wang, H.; Williams, M.; Wood, J. M.; Ethirajulu, K.;
Dymock, B. W. Bioorg. Med. Chem. Lett. 2012. http://dx.doi.org/10.1016/
j.bmcl.2012.02.058.
R
22 R = Me
23 R = Ph
N
O
b
24 R = 3,4-dimethoxyphenyl
25 R = styryl
F
N
S
H
26 R = 4-pyrazole
13. PDK1 assay: The kinase reactions were initiated by adding 25
substrate solution (2 M biotin–casein, 4 M ATP, 10 Ci/ml [
-33P] ATP in 1Â
assay buffer: 30 mM Hepes pH 7.5, 5 mM MgCl2, 30 mM KCl, 1 mM DTT and
0.2 mg/ml BSA) to 25
L of 2Â enzyme solution (100 nM PDK1 in 1Â assay
buffer). After 1 h incubation at room temperature, the reaction was terminated
with 25 L of stop buffer (45 mM EDTA, 50 mM ATP, 0.1% Triton-X, 30 mM
Hepes pH 7.5) and 60 L of the stopped reaction was transferred to a 384-well
Streptavidin FlashPlate HTS PLUS (Perkin Elmer Cat SMP410). After
minimum 30 min incubation at room temperature, the FlashPlate was
washed 3Â with 90 L/well of H2O/0.02% Tween 20, air-dried and counted
lL
of 2Â
l
l
l
c
l
Reagents and reaction conditions: a) Br2, acetic acid, 89%. b) R-
boronic acid, Pd(dppf)Cl2:CH2Cl2, aq. Na2CO3, PhMe/MeOH, 100
°C, 56-91%.
l
l
#
Scheme 2. Preparation of R1 phenyl analog.
l
on a Wallac MicroBeta (Perkin Elmer) scintillation counter.
14. Hopkins, A. L.; Groom, C. R.; Alex, A. Drug Discovery Today 2004, 9, 430.
15. PDB entry 1Z5M. Feldman, R. I.; Wu, J. M.; Polokoff, M. A.; Kochanny, M. J.;
Dinter, H.; Zhu, D.; Biroc, S. L.; Alicke, B.; Bryant, J.; Yuan, S.; Buckman, B.;
Lentz, D.; Ferrer, M.; Whitlow, M.; Adler, M.; Finster, S.; Chang, Z.; Arnaiz, D. O.
J. Biol. Chem. 2005, 280, 19867. All computational chemistry software used is
from Schrödinger (www.schrodinger.com). The PDK1 X-ray structures were
downloaded from the Protein Data Bank (www.pdb.org) and prepared with the
Protein Preparation Wizard as recommended by Schrodinger. Ligands were
prepared with Ligprep and docked into the protein structures using Glide and
the XP scoring function with the settings recommended by Schrödinger for
fragment docking.
Acknowledgments
We thank Eric T. Sun and Kantharaj Ethirajulu for critical
reviews.
Supplementary data
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.bmcl. 2012.04.080.
16. Atherall, J. F.; Hough, T. L.; Lindell, S. D.; O’Mahony, M. J.; Saville-Stones, E. A.
PCT WO 98/49899, 1998.