Discovery of a series of benzopyrimidodiazepinone TNK2 inhibitors via scaffold morphing

Article abstract of DOI:10.1016/j.bmcl.2020.127456

The protein kinase TNK2 (ACK1) is an emerging drug target for a variety of indications, in particular for cancer where it plays a key role transmitting cell survival, growth and proliferative signals via modification of multiple downstream effectors by unique tyrosine phosphorylation events. Scaffold morphing based on our previous TNK2 inhibitor XMD8-87 identified urea 17 from which we developed the potent and selective compound 32. A co-crystal structure was obtained showing 32 interacting primarily with the main chain atoms of an alanine residue of the hinge region. Additional H-bonds exist between the urea NHs and the Thr205 and Asp270 residues.

Full text of DOI:10.1016/j.bmcl.2020.127456

Journal Pre-proofs
Discovery of a Series of Benzopyrimidodiazepinone TNK2 Inhibitors via
Scaffold Morphing
Zhengnian Li, Chelsea E. Powell, Brian J. Groendyke, Thomas W. Gero,
Frederic Feru, John Feutrill, Bailing Chen, Bin Li, Hilary Szabo, Nathanael S.
Gray, David A. Scott
PII:
S0960-894X(20)30567-9
DOI:
Reference:
https://doi.org/10.1016/j.bmcl.2020.127456
BMCL 127456
To appear in:
Bioorganic & Medicinal Chemistry Letters
Received Date:
Revised Date:
Accepted Date:
18 May 2020
23 July 2020
24 July 2020
Please cite this article as: Li, Z., Powell, C.E., Groendyke, B.J., Gero, T.W., Feru, F., Feutrill, J., Chen, B., Li, B.,
Szabo, H., Gray, N.S., Scott, D.A., Discovery of a Series of Benzopyrimidodiazepinone TNK2 Inhibitors via
Scaffold Morphing, Bioorganic & Medicinal Chemistry Letters (2020), doi: https://doi.org/10.1016/j.bmcl.
2
020.127456
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover
page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version
will undergo additional copyediting, typesetting and review before it is published in its final form, but we are
providing this version to give early visibility of the article. Please note that, during the production process, errors
may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
©
2020 Published by Elsevier Ltd.

Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com
Discovery of a Series of Benzopyrimidodiazepinone TNK2 Inhibitors
via Scaffold Morphing
Zhengnian Li,a,b Chelsea E. Powell, Brian J. Groendyke, Thomas W. Gero, Frederic Feru,a,b John
a,b
a,b
a,b
Feutrill, Bailing Chen, Bin Li, Hilary Szabo, Nathanael S. Gray,a,b ,* and David A. Scott ,a,b ,
d
d
d
c
*
a
Department of Cancer Biology, Dana-Farber Cancer Institute, Boston, MA 02115, USA.
Department of Biological Chemistry & Molecular Pharmacology, Harvard Medical School, 360 Longwood Ave, Boston, MA 02115, USA.
b
c Vivid BioSciences, 50 Northern Ave, Boston, MA 02210, USA
d SYNthesis med chem, 425 Changyang Street, Suzhou Industry Park, Suzhou, Jiangsu, China.
ARTICLE INFO
ABSTRACT
Article history:
Received
Revised
Accepted
Available online
The protein kinase TNK2 (ACK1) is an emerging drug target for a variety of indications, in particular
for cancer where it plays a key role transmitting cell survival, growth and proliferative signals via
modification of multiple downstream effectors by unique tyrosine phosphorylation events. Scaffold
morphing based on our previous TNK2 inhibitor XMD8-87 identified urea 17 from which we
developed the potent and selective compound 32. A co-crystal structure was obtained showing 32
interacting primarily with the main chain atoms of an alanine residue of the hinge region. Additional
H-bonds exist between the urea NHs and the Thr205 and Asp270 residues.
Keywords:
TNK2
Kinase inhibitor
Benzopyrimidodiazepinone
Scaffold morphing
2
009 Elsevier Ltd. All rights reserved.
Tyrosine Kinase Nonreceptor 2 (TNK2) is a cytoplasmic
kinase, and was originally identified based on its binding to the
87 as a potent and highly selective TNK2 inhibitor (Kinome Scan
S(10) = 0.03), identified through kinase selectivity profiling.
4
,13
1
,2
cell-cycle regulator, CDC42, also known as ACK1 (activated
CDC42-associated kinase). TNK2 aberrant activation or mutation
occurs in multiple types of cancers such as prostate, breast,
HN
3
O
schwannoma, pancreatic, ovarian, non-small cell lung cancer,
N
N
4
and leukemia. TNK2 can interact with several oncogenic
HN
N
N
N
N
O
NH2
N
signaling pathways, including modulation of the AKT signaling
O
N
N
5
N
pathway and phosphorylation of androgen receptor, resulting in
MeO
6
NH
androgen-independent prostate cancer growth. Additionally,
7
TNK2 negatively regulates the tumor suppressor Wwox. These
O
N
H
HO
findings suggest that inhibition of TNK2 might have therapeutic
1
(AIM-100)
2
3
8
potential across various cancers.
H
O
Cl
N
N
N
Several groups have reported small-molecule TNK2 inhibitors
HN
N
NH
HN
N
N
9
-13
(
Figure 1).
AIM-100 has been used as a tool compound to
O
study TNK2 biology, though it was initially disclosed as a
O
1
4
screening hit in a Lck inhibitor program. Amgen reported a
series of potent pyrazolopyrimidines and obtained an X-ray co-
crystal structure with 2, which showed the typical donor-acceptor
H-bond interactions between the hinge and the anilinopyrimidine
and an additional H-bond to the gatekeeper Thr205 residue from
N
N
N
N
4 (R)-9b
5 (XMD8-87)
Figure 1. TNK2 inhibitors.
9
the dimethylaniline substituent. Compound 3 was described by
OSI as a relatively selective (KinomeScan S(10) = 0.08) TNK2
We also recently described our efforts to optimize the PK
profile of the benzopyrimidodiazepinone series, but observed
1
1
inhibitor with good oral PK. Compound 4 was developed
through the preparation of a library of compounds incorporating
fragments of other TNK2 inhibitors, and has been used to explore
15
consistently high in vivo clearance for compounds lacking a
substituent on the amide nitrogen. This was in contrast to the
1
2
the role of TNK2 in prostate and triple negative breast cancer.
We previously reported the benzopyrimidodiazepinone XMD8-
16
amide N-methyl analogs reported as inhibitors of ERK5 and
17,18
LRRK2,
which demonstrated good IV and oral PK profiles.
We pursued a scaffold-morphing approach to try and develop

TNK2 inhibitors with better PK properties, in which the tricyclic
amide core of XMD8-87 was replaced with a urea scaffold.
Next, we introduced substituents on the right-hand side of the
molecule. A methyl group was introduced at each position on the
phenyl ring. 9-Methyl (21) and 10-methyl (22) demonstrated
about 4-fold and 8-fold increase of potency in the biochemical
assay (IC50 = 224 nM) and (IC50 = 91 nM) compared with 17.
We also introduced fluorine at different positions of the phenyl
ring, and further substituents at the C9 position, including
methoxy (28) and cyclopropyl (27). Unlike the amide scaffold, in
which even small substituents on the right-hand phenyl ring were
An efficient three-step synthetic route was developed to prepare
the urea compounds (17, 18, and 20-39). The synthesis is
outlined in Scheme 1. First, a Suzuki coupling of 5-amino-2,4-
dichloropyrimidine and a 2-aminophenylboronic acid pinacol
ester gave diamine intermediate 8 in moderate yield. Treatment
of the diamine with triphosgene at room temperature resulted in
cyclization to the tricyclic urea intermediate 9. Anilines were
installed on the pyrimidine of intermediate 9 via Pd coupling
chemistry as the final step. Compound 19 was synthesized from
methylamino phenylboronic acid pinacol ester 13, using similar
procedures to those in Scheme 1. 13 was prepared in two steps
from 2-aminophenylboronic acid pinacol ester 10 by alkylation
with benzotriazole alcohol 11, and reduction of benzotriazole
boronate ester 12 with sodium borohydride (Scheme 2).19
1
5
not well-tolerated, we found that several substituents on the
phenyl ring of the urea scaffold led to increased TNK2 potency.
N
N
N
N
O
O
a
b
O
N
N
B
B
H
N
O
NH2
OH
10
11
12
H
N
O
H
O
NH2
N
N
NH2
Cl
N
6a
N
O
N
c
d
NH
B
Cl
N
HN
N
1N
HN
N
O
HN
Cl
N
1
N
O
O
H
R
13
6
14
H
O
N
N
N
N
N
NH2
N
O
N
H
O
HN
N
5
(XMD8-87)
Urea Series
e
N
O
N
N
N
N
Cl
N
Figure 2. Scaffold morphing of amide to urea series.
N
N
To understand the structure-activity relationships (SAR) of the
urea analogs, we employed a biochemical TNK2 assay (HTRF
LanthaScreen Assay)20 and a cell viability assay based on Ba/F3
cells expressing TNK2 D163E, (CellTiter-Glo Luminescent Cell
Viability Assay). We also conducted a cellular counter-screen
assay in parental BaF3 cells, and observed at least 10 fold
selectivity for the majority of compounds. Compound 17 (Table
1
5
16
19
Scheme 2. Synthesis of 19. Reagents and conditions: (a) EtOH, rt, overnight,
70% yield; (b) NaBH , THF, 75°C, 4 h, 69% yield; (c) Pd(dppf)Cl , XPhos,
4
2
2
1
i
2.0 M Na
2
CO
,4-dioxane, rt, 1h, 58% yield; (e) Pd
00°C, 2h, 60% yield.
3
/1,4-dioxane, 100°C, 4h, 35% yield; (d) Triphosgene, Pr
2
EtN,
1
1
2
(dba) , XPhos, K CO , 1,4-dioxane,
3
2
3
1
) had a TNK2 IC50 of 738 nM, and an EC50 of 0.48 µM in our
Ba/F3 cellular assay, less potent than the amide XMD8-87 (IC50
of 44 nM, cell EC50 of 0.19 µM). We introduced a methyl group
at each urea nitrogen (18, 19) but both compounds were
significantly less active. This was consistent with our
observations on the amide scaffold, in which a methyl group on
the amide nitrogen of XMD8-87 led to reduced potency in both
the enzyme and cell assays, with the NH adjacent to the
pyrimidine ring assumed to make an H-bond interaction with the
Thr205 gatekeeper of TNK2.
Table 1. Phenyl ring SAR; compounds 17-28
R1
O
N
5
N
R2
7
N
HN
N
8
O
11
9
10
R3
N
N
R1
R2
R1
NH
Cell
TNK2
D163E Ba/F3
Cell
Parental
N
TNK2
IC50
(nM)
NH
N
O
a
R1
R2
R3
Compound
B
Cl
N
a
R2
Cl
N
Cl
O
NH2
(M)
(M)
H2N
1
1
19
20
21
22
2
24
25
2
2
2
a
7
8
-
-
-
-
-
-
738
0.48
>10
nd
8.42
2.77
nd
9.58
>10
2.68
9.53
8.47
4.84
6.20
6
7
8
Me
>10000* 5.87
1
R = H, CH3
R1
N
-
-
-
-
-
-
-
-
-
-
Me
4770
942
224
91
1070
248*
683
204
135
394
4.98
0.21
0.15
0.05
0.59
0.21
0.48
0.25
0.07
0.21
O
R1
-
-
-
-
-
-
-
-
-
8-Me
9-Me
10-Me
11-Me
8-F
9-F
10-F
9-cyclopropyl
9-OMe
O
N
N
NH
N
HN
N
NH
c
b
3
Cl
N
R2
R3
R2
6
7
8
9
17, 18, 20-39
Scheme 1. Synthesis of benzopyrimidodiazepinones. Reagents and
conditions: (a) Pd(dppf)Cl , XPhos, 2.0 M Na CO /1,4-dioxane, 100°C, 4h,
9-74% yield; (b) Triphosgene, iPr EtN, 1,4-dioxane, rt, 1h, 33-35% yield;
c) Aniline, XPhos, Pd (dba) , K CO , 1,4-dioxane, 100°C, 2h, 60-65% yield.
Data was generated at BioDuro or ThermoFisher (*). IC50s against TNK2
2
2
3
4
(
2
was obtained with LanthaScreen Binding activity assays and reported as the
average of two replicates. nd = no data.
2
3
2
3

We next focused our attention on the aniline substituent using
the 9-methyl substituted core (21), rather than the more potent
OMe
N
40
131
0.37
>10
1
0-methyl substituted core (22) which showed a less selective
kinase activity profile (data not shown here). Aniline SAR was
generally consistent with what was previously observed for the
amide series (Table 2). Compound 29, lacking the 2-MeO group,
was 2-fold more potent than 21 in the enzyme assay. Moving the
methoxy group to the aniline 3-position (30) significantly
improved TNK2 potency in both enzyme and cell assays.
a
Data was generated at BioDuro. IC50s against TNK2 was obtained with
LanthaScreen Binding activity assays and reported as the average of two
replicates. nd = no data.
Piperazine derivatives 31 and 32 showed similar enzyme and
cell potency to the N-methylpiperazine 21. Fluorine was
introduced to the phenyl ring to reduce its electron density and
potential for oxidation. This resulted in reduced potency when
the F was ortho to the aniline NH (33), but substitution at the
meta position (34) was tolerated. We also installed N-
methylpiperidines (35, 36, and 37) to avoid the possibility of 1,4-
quinone diamine formation, but these compounds were
consistently less potent than the piperazines. We also prepared a
However,
our
previous
data
with
the
related
benzopyrimidodiazepinone amide scaffold suggested that the 3-
MeO aniline would result in reduced kinase selectivity.15 We
thought that 2-MeO substitution was most likely to give a good
balance of potency and selectivity and focused on a set of
anilines with this group with the potential to impact
2
2
metabolism.
15
set of compounds with a bicyclic aniline and of these,
tetrahydroisoquinoline 40 was the most potent. Overall, there was
a good correlation between the enzyme and TNK2 cell assays for
the urea series.
Table 2. Enzyme and cell data for aniline SAR of 9-Me;
compounds 21, 29-40
H
N
O
N
NH
HN
R
N
TNK2
Cell TNK2 Cell
Compoun
d
R
IC50
D163E
Parental
a
(nM)
(M)
Ba/F3 (M)
OMe
2
2
3
1
9
0
224
113
13
0.15
0.11
0.02
nd
N
N
N
N
N
>10
MeO
N
5.62
OMe
HO
3
3
1
2
164
102
0.34
0.29
6.56
9.93
N
N
OMe
HN
N
OMe
3
3
3
4
N
N
3935
120
5.07
0.32
3.50
nd
F
OMe
N
N
N
F
OMe
3
3
3
3
3
5
6
7
8
9
405
3410
364
313
279
0.58
2.84
0.70
1.11
0.47
8.53
9.33
4.95
7.55
9.62
Figure 3. (A) X-ray co-crystal structure of TNK2 in complex with compound
3
2, refined at 2.87 Å resolution (PDB: 6VPM). Compound 32 is shown in
stick representation with yellow carbon atoms. (B) Hydrogen bonds are
indicated by dotted lines. Main interacting residues are shown and labeled.
OMe
N
To confirm the binding mode of the urea series, we solved the
co-crystal structure of compound 32 with TNK2 at a resolution of
OMe
N
2
.87 Å with the ligand occupying the ATP site and neighboring
regions (Figure 3A). As expected, the pyrimidine N and aniline
NH interacted with the hinge, forming H-bonds with the amide
NH and carbonyl of Ala208, respectively. The aniline substituent
extended into the solvent channel. The urea NH adjacent to the
pyrimidine interacted with the Thr205 gatekeeper residue, and
this likely contributed to the overall kinase selectivity of the urea
series, as we had suspected was the case for the previously
OMe
N
OMe
N

reported benzopyrimidodiazepinone amides.15 The NH adjacent
to the phenyl ring interacted with the Asp270 of the DFG motif at
the beginning of the activation loop. Methylation at either of the
urea nitrogens, to give compounds 18 and 19, resulted in a
significant loss of enzyme and cell potency. Compared to the
amide series, the phenyl ring of the tricyclic urea scaffold is
positioned in a more open region of the protein, allowing the
introduction of substituents at several positions to further
improve potency and selectivity.
Selectivity
Score:
S(10)=0.01
SRMS
ABL1
TNK2
GAK
We tested the kinase selectivity of 32 using KINOMEscan at a
concentration of 1 uM. The results show that 32 displays
excellent kinase selectivity with an S (10) score of 0.01,
calculated by dividing the number of kinases with ctrl% ≤10% (n
=
14) by the total number of kinases tested (n = 468). (See
Supporting Table 1 and Figure 1 for the full data).
Metabolic stability of compounds 31 and 32 was assessed in
mouse and human liver microsomes and mouse hepatocytes, and
both compounds demonstrated good in vitro metabolic stability
Figure 4. KINOMEscan results for compound 32. Selectivity profile was
determined for 468 kinases as percent DMSO control at 1 µM inhibitor
concentration (excluding mutant, lipid, atypical and pathogen). S (10) of
compound 32 against kinases with ctrl% ≤10% containing TNK2 highlighted
as blue.
(
Table 3). Compound 31 and 32 were further evaluated for their
pharmacokinetic properties in male CD1 mice after intravenous
IV) and oral (PO) administration (Table 4).
(
In summary, we have demonstrated a scaffold-morphing
approach to discover a novel series of urea TNK2 inhibitors. The
introduction of substituents on the phenyl ring improved the
potency of the series relative to 17. The expected binding mode
of the series was confirmed by a co-crystal structure with 32, and
this compound was found to have a good overall kinase
selectivity profile. However, as with our series of amide
benzopyrimidodiazepinone TNK2 inhibitors, we were not able to
identify an example with good PK, despite promising in vitro
metabolic stability data.
Hydroxyethylpiperazine 31 displayed a half-life (T1/2) of 0.6 h, a
very high clearance (Cl) of 150 mL/min/kg, and a low oral
bioavailability of 6%. Dimethylpiperazine 32 shows a limited
improvement over compound 31, with clearance (Cl) of 77
mL/min/kg and oral bioavailability of 9%. The disconnect
between the in vitro and in vivo clearance was disappointing, and
we speculated that despite the encouraging in vitro stability data,
oxidation to the quinone diamine might still be involved. We
therefore profiled an example with a bicyclic aniline (40) but its
overall PK profile was similar to that of 32.
a
Table 4. Mouse microsome and hepatocyte stability, and in vivo mouse PK data (3mpk IV, 10mpk PO)
Compound ID
Mouse microsomes
Mouse hepatocytes
IV Cl
(ml/min/kg)
IV T1/2
(hr)
F %
Auc inf (PO)
(ng*hr/ml)
(T
1/2
)
(T1/2
)
3
3
4
a
1
2
0
57
145
-
>500
>500
-
150
77
94
0.6
1.8
1.8
6
9
18
101
200
405
PK studies were conducted in male CD-1 mice, and all values are an average from three animals
Acknowledgments
Ba/F3 cells expressing TNK2 D163E were generously provided by
Jeffrey W. Tyner (Oregon Health and Science University, Portland, OR).
In vitro and in vivo PK studies were conducted at BioDuro, Shanghai,
under the supervision of Cindy Liang and Ying Zhou. Hyuk-Soo Seo
helped upload the structural data to the PDB and Milka Kostic assisted
with the preparation of the manuscript. We thank Vivid BioSciences for
funding support.
8.
Fox, M.; Crafter, C.; Owen, D. Biochem Soc Trans 2019, 47 (6),
1715.
Kopecky, D. J.; Hao, X.; Chen, Y. et al. Bioorg Med Chem Lett
9
1
1
1
1
.
2
008, 18 (24), 6352.
0. Jiao, X.; Kopecky, D. J.; Liu, J. et al. Bioorg Med Chem Lett 2012,
2 (19), 6212.
1. Jin, M.; Wang, J.; Kleinberg, A. et al. Bioorg Med Chem Lett
013, 23 (4), 979.
2. Lawrence, H. R.; Mahajan, K.; Luo, Y. et al. J Med Chem 2015,
8 (6), 2746.
2
2
References and notes
5
3. Miduturu, C. V.; Deng, X.; Kwiatkowski, N. et al. Chem Biol
2011, 18 (7), 868.
1
.
Prieto-Echague, V.; Miller, W. T. J Signal Transduct 2011, 2011,
42372.
7
14. DiMauro, E. F.; Newcomb, J.; Nunes, J. J. et al. Bioorg Med
Chem Lett 2007, 17 (8), 2305.
15. Groendyke, B. J.; Powell, C. E.; Feru, F. et al. Bioorg Med Chem
Lett 2020, 30 (4), 126948.
2
3
4
.
.
.
Mahajan, K.; Mahajan, N. P. Cancer Letters 2013, 338 (2), 185.
Mahajan, K.; Mahajan, N. P. Oncogene 2015, 34 (32), 4162.
Maxson, J. E.; Abel, M. L.; Wang, J. et al. Cancer Res. 2016, 76,
1
27.
Mahajan, K.; Challa, S.; Coppola, D. et al. Prostate 2010, 70 (12),
274.
16. Deng, X.; Yang, Q.; Kwiatkowski, N. et al. ACS Med Chem Lett
2011, 2 (3), 195.
17. Deng, X.; Dzamko, N.; Prescott, A. et al. Nat Chem Biol 2011, 7
(4), 203.
18. Deng, X.; Elkins, J. M.; Zhang, J. et al. Eur J Med Chem 2013, 70,
758.
5
6
7
.
.
.
1
Mahajan, N. P.; Liu, Y.; Majumder, S. et al. Proc Natl Acad Sci U
S A 2007, 104 (20), 8438.
Mahajan, N. P.; Whang, Y. E.; Mohler, J. L. et al. Cancer Res
2
005, 65 (22), 10514.
19. Ogiyama, T.; Yamaguchi, M.; Kurikawa, N. et al. Bioorg Med
Chem 2017, 25 (7), 2234.

2
2
0. Enzyme testing was conducted at BioDuro or at ThermoFisher,
using an HTRF LanthaScreen assay. For compounds tested at both
sites, there was an excellent correlation between the two assays,
with the ThermoFisher assay typically showing slightly increased
sensitivity (compounds up to 2-fold more potent).
1. We used a cell viability assay based on Ba/F3 cells expressing
TNK2 D163E. Parental Ba/F3 cells were cultured with an
additional 1 ng/mL recombinant mouse IL-3. Cells were plated at
a volume of 50 µL with a density of 105 cells/mL in white 384-
well plates (Corning 3570). 100 nL of compound in DMSO from
stock plates was added by pin transfer using a Janus Workstation
(PerkinElmer). After 72 hr, cell viability was evaluated using the
CellTiter-Glo Luminescent Cell Viability Assay (Promega)
following the manufacturer’s standards. IC50s were modeled from
3 biological replicates using GraphPad Prism 8 software.
22. Mesaros, E. F.; Thieu, T. V.; Wells, G. J. et al. J Med Chem 2012,
55 (1), 115.
Supplementary Material