Pyridyl-pyrimidine benzimidazole derivatives as potent, selective, and orally bioavailable inhibitors of Tie-2 kinase

Article abstract of DOI:10.1016/j.bmcl.2008.11.056

Selective small molecule inhibitors of Tie-2 kinase are important tools for the validation of Tie-2 signaling in pathological angiogenesis. Reported herein is the optimization of a nonselective scaffold into a potent and highly selective inhibitor of Tie-2 kinase.

Full text of DOI:10.1016/j.bmcl.2008.11.056

Bioorganic & Medicinal Chemistry Letters 19 (2009) 424–427
Contents lists available at ScienceDirect
Bioorganic & Medicinal Chemistry Letters
journal homepage: www.elsevier.com/locate/bmcl
Pyridyl-pyrimidine benzimidazole derivatives as potent, selective,
and orally bioavailable inhibitors of Tie-2 kinase
Victor J. Cee a, , Alan C. Cheng b,f, Karina Romero a, Steve Bellon b,f, Christopher Mohr b,f
,
*
Douglas A. Whittington b,f, Annette Bak c, James Bready e, Sean Caenepeel e, Angela Coxon e, Holly L. Deak a,
Jenne Fretland d,g, Yan Gu b,f, Brian L. Hodous a, Xin Huang b,f, Joseph L. Kim b,f, Jasmine Lin d,g
,
Alexander M. Long b,f, Hanh Nguyen a, Philip R. Olivieri a, Vinod F. Patel a, Ling Wang e, Yihong Zhou d,g
,
Paul Hughes e, Stephanie Geuns-Meyer a
a Department of Medicinal Chemistry, Amgen Inc., One Kendall Square, Bldg. 1000, Cambridge, MA 02139, USA
b Department of Molecular Structure, Amgen Inc., One Kendall Square, Bldg. 1000, Cambridge, MA 02139, USA
c Department of Pharmaceutics, Amgen Inc., One Kendall Square, Bldg. 1000, Cambridge, MA 02139, USA
d Department of Pharmacokinetics and Drug Metabolism, Amgen Inc., One Kendall Square, Bldg. 1000, Cambridge, MA 02139, USA
e Department of Oncology Research, Amgen Inc., One Amgen Center Drive, Thousand Oaks, CA 91320, USA
f
Department of Molecular Structure, Amgen Inc., One Amgen Center Drive, Thousand Oaks, CA 91320, USA
g Department of Pharmacokinetics and Drug Metabolism, Amgen Inc., One Amgen Center Drive, Thousand Oaks, CA 91320, USA
a r t i c l e i n f o
a b s t r a c t
Article history:
Selective small molecule inhibitors of Tie-2 kinase are important tools for the validation of Tie-2 signaling
in pathological angiogenesis. Reported herein is the optimization of a nonselective scaffold into a potent
and highly selective inhibitor of Tie-2 kinase.
Received 30 August 2008
Revised 12 November 2008
Accepted 17 November 2008
Available online 20 November 2008
Ó 2008 Elsevier Ltd. All rights reserved.
Keywords:
Tie-2 kinase
Angiogenesis
Kinase inhibitor
Benzimidazole
Receptor tyrosine kinases expressed by vascular endothelial
cells are emerging as important mediators of pathological angio-
genesis.1 It is well established that activation of KDR (VEGFR2)
by vascular endothelial growth factor (VEGF) strongly promotes
angiogenesis by enhancing endothelial cell proliferation and sur-
vival.2 In contrast, the precise role of the Tie-2 receptor in patho-
logical angiogenesis has been difficult to ascertain, due to the
existence of numerous endogenous ligands, including Angiopoe-
itin-1 (Ang-1) and Angiopoeitin-2 (Ang-2), which exhibit context-
dependent activities.3 We hypothesized that a small molecule
Tie-2 kinase inhibitor with selectivity over KDR could be used to
clarify the role of Tie-2 signaling in pathological angiogenesis.4 Re-
ported herein is the optimization of a nonselective scaffold into a
potent, highly selective (>2500-fold over KDR), and orally available
inhibitor of Tie-2 kinase.
nonselective inhibitor of Tie-2 and KDR (Fig. 1). Further modifica-
tion involving replacement of the N,N0-diarylurea with N-arylbenz-
imidazole gave analog 2. This molecule exhibited significantly
weaker inhibition of Tie-2 as well as KDR. To our surprise, com-
pound 4, which lacks the lipophilic trifluoromethyl group, proved
to be a reasonably potent and highly selective Tie-2 inhibitor. This
finding was not expected based on the diminished potency and
H
N
H
N
H
N
H
N
Me
O
R
N
O
R
N
1 R = CF3
2 R = CF3
HN
Me
N
N
Tie-2 = 4 nM
Tie-2 = 470 nM
Structural modification of an existing kinase inhibitor series4c
provided a pyridyl-pyrimidine urea derivative 1 that is a potent,
KDR = 34 nM
KDR = 6,000 nM
3 R = H
4 R = H
Tie-2 = 460 nM
Tie-2 = 23 nM
KDR = 7,600 nM
KDR = >25,000 nM
* Corresponding author. Tel.: +1 805 313 5500; fax: +1 805 480 1337.
E-mail address: vcee@amgen.com (V.J. Cee).
Figure 1. Discovery of a potent and selective Tie-2 inhibitor.
0960-894X/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmcl.2008.11.056

V. J. Cee et al. / Bioorg. Med. Chem. Lett. 19 (2009) 424–427
425
Table 2
selectivity of analogous urea 3. Given the promising selectivity of 4,
an effort was initiated to further optimize this series.
SAR of the aminobenzimidazole region
X-ray co-crystal structures of related phenoxy-substituted pyr-
idyl pyrimidines bound to the Tie-2 kinase domain indicate that
the aminopyrimidine engages in a series of hydrogen bonds with
the hinge region of the kinase, with the central para-aminophen-
oxy ring occupying a hydrophobic pocket.4c Modifications to the
central ring were undertaken with the goal of improving Tie-2 po-
tency (Table 1). The initial monomethyl-substituted analog 4 has
an IC50 of 23 nM, with a 10-fold higher IC50 of 250 nM for inhibition
of Ang-1-stimulated Tie-2 phosphorylation in the Ea.hy926 endo-
thelial cell line. The desmethyl analog (5), regioisomer (6), and di-
methyl analog (7) all exhibit 10- to 20-fold reduced potency for
Tie-2 inhibition. Ultimately, naphthalene was found to be the ideal
central ring, with 8 exhibiting IC50 values of 1 nM for Tie-2 enzyme
and 20 nM for Tie-2 autophosphorylation in cells. Weak but mea-
surable inhibition of KDR is evident for compound 8 (1400 nM),
but the selectivity is still 1400-fold in favor of Tie-2. The impor-
tance of the benzimidazole appendage in achieving selectivity
against KDR is evident from the corresponding urea 9, which
exhibits only 8-fold selectivity.
H
N
R1
A
N
R2
N
O
HN
Me
N
N
a
Compound
A
R1
R2
Enzyme IC50 (nM)
KDR
Tie-2
1
8
NH
NH
NH
NMe
O
H
Me
H
H
H
H
H
0.6
1400 90
1570 170
69 10
10
11
12
13
14
110 14
27
240 170
16
44
CF3
H
6
250
30 0.2
88
3
H
H
2
4
S
H
5
a
Values are means of two to four experiments. Kinase assays run at the Km for
M for Tie-2 and 1 M for KDR.
ATP with final [ATP] of 5
l
l
genesis is not well understood from the literature, this off-target
activity was not addressed.7
With the optimal central ring identified, our attention turned to
modification of the aminobenzimidazole portion of molecule 8
(Table 2). Lipophilic substituents (Me, CF3) as the R1 and R2 groups,
respectively, reduce Tie-2 potency, with the trifluoromethyl-
The strategy for reducing inhibition of the Aurora kinases was
guided by an X-ray co-crystal structure of urea 1 bound to Aurora
A (Fig. 2).8 Unlike previously obtained co-crystal structures of 4-(2-
phenoxypyridin-3-yl)pyrimidines and Tie-2, the lipophilic methyl
substituent of the central aryl ring does not occupy the hydropho-
bic pocket. This finding suggests that the hydrophobic pocket of
the Aurora kinases is limited in size relative to Tie-2.
We reasoned that methyl-substituted analog 15 (Table 3) would
not be accommodated by the Aurora kinases due to the presence of
two substituents ortho to the phenoxy linkage. Relative to the par-
ent compound 8, compound 15 retains excellent potency for Tie-2
enzyme inhibition (IC50 = 2 nM), high selectivity against KDR
(>2500ꢀ), and exhibits better potency for inhibition of Tie-2 auto-
phosphorylation in Ea.hy926 cells (IC50 = 2 nM). Confirming our
hypothesis, 15 exhibits approximately 2- and 8-fold less inhibitory
activity for Aurora A and B enzymes, respectively, and most impor-
tantly, 15 does not induce phenotypic changes at concentrations
substituted 11 exhibiting
a striking increase in potency for
inhibition of KDR, resulting in a relatively nonselective dual inhib-
itor. Methylation or replacement of the benzimidazole ring N–H
results in reduced potency for inhibition of Tie-2, but significantly
enhanced inhibition of KDR (12–14).
Compound 8 emerged as a very potent inhibitor of Tie-2 with
excellent selectivity against KDR (1400ꢀ), and as such appeared
to be an ideal molecular tool for further studies of Tie-2 in patho-
logical angiogenesis. Profiling against a panel of 20 tyrosine and
serine/threonine kinases indicated generally good selectivity, with
significant activity (IC50 < 100 nM) for Aurora A/B and JNK kinases.
The inhibitory activity of 8 toward the purified Aurora kinase en-
zymes was confirmed in a cell-based assay: 8 induced aborted
cytokinesis and P4 N DNA content in HeLa cells with an EC50 of
120 nM.5 Lack of meaningful Aurora kinase selectivity raised a po-
tential concern as the Aurora kinases are important in the function
of normal cell division.6 As the impact of JNK inhibition on angio-
up to 1.2 lM in a cell-based assay. Therefore, compound 15 pos-
sesses excellent cellular selectivity (>600ꢀ) against the Aurora ki-
nases. This structural change did not, however, reduce the ability of
15 to inhibit the JNK family of kinases.
The synthesis of 15 (Scheme 1) is representative of the 4-(2-
phenoxypyridin-3-yl)pyrimidines 4–15. Coupling of 4-amino-2-
methylnaphthalen-1-ol and the previously reported 4-(2-chloro-
pyridin-3-yl)-N-methylpyrimidin-2-amine 164c occurred readily
in the presence of cesium carbonate at high temperature. Conver-
sion of the aniline into the corresponding isothiocyanate was
accomplished with di-2-pyridyl thionocarbonate.9 Polymer-sup-
ported carbodiimide, a convenient desulfurizing agent,10 facilitated
Table 1
SAR of the central aryl ring (4–9)
R2
H
N
H
N
H
N
H
N
R1
O
N
O
N
N
O
HN
Me
N
N
HN
Me
N
N
4-8
R2
9
H
N
H
N
Me
O
Hydrophobic Pocket
R1
Enzyme IC50 (nM)
Cellular IC50 (nM)
Tie-2 (Ea.hy926)
a
b
Compound
O
N
CF3
Tie-2
23
KDR
HN
Me
N
N
4
5
6
7
8
9
Me
H
H
H
H
Me
Me
4
>25,000
>25,000
>25,000
>25,000
1400 90
40 14
250 70
ND
ND
250 30
590 70
310 30
Me
ND
–(CH)4–
1
5
0.6
0.1
20c
ND
a
Values are means of two to four experiments. Kinase assays run at the Km for
M for Tie-2 and 1 M for KDR.
Inhibition of Ang-1 stimulated Tie-2 phosphorylation in Ea.hy926 cells. ND, not
determined.
ATP with final [ATP] of 5
l
l
Ala 213
b
c
One experiment.
Figure 2. X-ray co-crystal structure of Aurora A and urea 1.

426
V. J. Cee et al. / Bioorg. Med. Chem. Lett. 19 (2009) 424–427
Table 3
Elimination of cellular Aurora activity
H
N
H
N
N
N
O
R
HN
Me
N
N
a
Compound
R
Enzyme IC50 (nM)
Cellular IC50 (nM)
Tie-2
KDR
Aur A/B
19 11/10
28 1/80 50
Tie-2b
20d
Aurorac
8
H
1
2
0.6
0.7
1400 90
>5000
2
120d
15
Me
2
0.4
>1200d
a
Values are means of two to four experiments. Kinase assays run at the Km for
M for Tie-2, 1 M for KDR, 8.2 M for Aurora A and
ATP with final [ATP] of 5
l
l
l
23
lM for Aurora B.
b
Inhibition of Ang-1 stimulated Tie-2 phosphorylation in Ea.hy926 cells.
EC50 for induction of P4 N phenotype by cellular imaging in HeLa cells.
One experiment.
c
d
Figure 3. (A) X-ray co-crystal structure of 15 bound to JNK3. (B) Homology model
of 15 bound to Tie-2. (C) Overlay of Tie-2 apo (purple) and bound (wheat) with
surface rendering of benzimidazole binding pocket. (D) X-ray co-crystal structure of
11 bound to KDR.
NH2
N
Cl
nobenzimidazole occupies a strikingly different position, with
a
b
N
O
hydrogen-bonding interactions with
a-C-helix residue Glu885
HN
N
N
N
92%
63%
Me
and the backbone N–H of activation loop residue Asp1046. The
activation loop is positioned in the inactive DFG-out conforma-
tion,13 with the lipophilic trifluoromethyl-arene portion of the
benzimidazole occupying the extended hydrophobic pocket. Pre-
sumably, the absence of the lipophilic trifluoromethyl group in
compounds 8 and 15 results in suboptimal interactions in the ex-
tended hydrophobic pocket, leading to poor binding to KDR.
Comparison of the bound structure of the KDR-selective Tie-2
inhibitor 15 derived from the JNK3/15 co-crystal with the bound
structure of the KDR-selective Tie-2 inhibitor 19 derived from a
previously reported Tie-2/19 co-crystal4c reveals that the benz-
imidazole ring of 15 and the morpholine ring of 19 occupy similar
space (Fig. 4). As structure–activity relationships have demon-
strated that the benzimidazole of 15 and morpholine of 19 are
key to achieving selectivity for Tie-2 over KDR, both molecules ap-
pear to benefit by occupying the hydrophobic pocket of Tie-2 iden-
tified in Figure 3C.
Compound 15 possessed acceptable pharmacokinetic properties
in rat (Table 4) characterized by a moderate clearance (Cl) and
large volume of distribution (Vss) which resulted in a relatively
long half life of 3.7 h. When orally dosed as a suspension to rats,
the bioavailability was 26%.
The ability of 15 to modulate Tie-2 phosphorylation levels
in vivo was assessed in a mouse pharmacodynamic model, in
which Tie-2 phosphorylation in mouse lung tissue is stimulated
by human Angiopoeitin-1 (h-Ang-1) (Fig. 5).4c A dose-dependent
inhibition of stimulated Tie-2 phosphorylation was observed, with
Tie-2 phosphorylation reduced to sub-basal levels with a single
oral dose of P30 mg/kg. The 10 mg/kg dose provided an approxi-
mately 50% reduction in stimulated Tie-2 phosphorylation, with
a plasma concentration of 260 nM 15. This concentration is 130-
Me
HN
Me
N
N
16
17
H
N
H
N
NCS
N
c
N
O
N
O
Me
57%
Me
HN
Me
N
N
HN
N
Me
18
15
Scheme 1. Synthesis of 15. Reagents and conditions: (a) 4-amino-2-methylnaphth-
alen-1-ol, Cs2CO3, DMSO, 130 °C; (b) di-2-pyridyl thionocarbonate, CH2Cl2; (c) 1,2-
phenylene diamine, PS-carbodiimide, THF, 70 °C.
the condensation of 18 and 1,2-phenylenediamine to give amino-
benzimidazole 15.
While it was not possible to obtain a structure of 15 bound to
Tie-2, an X-ray co-crystal structure was obtained for 15 bound to
JNK3 (Fig. 3A). The molecule occupies the ATP binding pocket, with
the aminopyrimidine within hydrogen-bonding distance of the
hinge residue Met149. The central naphthalene ring projects into
the hydrophobic pocket. The aminobenzimidazole is positioned
to hydrogen bond with the backbone C@O of the activation loop
residue Leu206 and the side chain of Lys93. The terminal phenyl
portion of the benzimidazole occupies a relatively hydrophobic
groove bounded on one side by the
a-C-helix and by leucine side
chains from the N-terminal kinase domain on the other.
A homology model of Tie-2/15 was constructed based on the
JNK3/15 structure (Fig. 3B).11 The major residue differences be-
tween JNK3 and Tie-2 are Met146 to Ile902 ‘gatekeeper residue’,
Ala108 to Phe869 and Leu95 to Met857. These differences do not
appear to significantly hinder binding of 15 to Tie-2. The position
of the benzimidazole ring in the Tie-2/15 homology model is
remarkably similar to the inhibitory position observed for
Phe983 in the reported Tie-2 apo crystal structure (Fig. 3C).12
Apparently, KDR cannot adopt this binding mode or is much less
accommodating of 15 in this binding mode.
O
Me
N
H
N
N
O
O
HN
Me
N
N
19
CF3
Tie-2IC50 = 10 nM
KDR IC50 = 310 nM
The inability of KDR to accommodate compound 15 led us to
speculate that the nonselective dual inhibitors 11–14 bind to
KDR in a significantly different conformation. The structure of 11
(KDR IC50 = 69 nM) bound to KDR was obtained (Fig. 3D). The ami-
Figure 4. Overlay of selective Tie-2 inhibitors 15 (green) and 19 (orange) from X-
ray co-crystal structures with JNK3 and Tie-2, respectively.

V. J. Cee et al. / Bioorg. Med. Chem. Lett. 19 (2009) 424–427
427
Table 4
15 may be a suitable tool for the in vivo evaluation of Tie-2 signal-
ing in angiogenesis.
Pharmacokinetic properties of 15 in male Sprague–Dawley rats
Cl (mL/min/kg)a
28
Vss (mL/kg)a
T1/2 (h)a
F (%)b
Acknowledgments
5900
3.7
26
a
iv, 0.5 mg/kg, DMSO.
po, 5 mpk, OraPlus pH = 2.
b
We thank Melanie Cooke for formulation support, Bethlyn
Jaramilla for bioanalytical support, Shane Johnson for plasma pro-
tein binding, Marc Payton and Patrick Eden for the HeLa cell assay,
and Jin Tang for purification of Aurora A/B proteins.
60
50
40
30
20
10
0
3500
3000
2500
2000
1500
1000
References and notes
1. (a) Folkman, J. Nat. Rev. Drug Disc. 2007, 6, 273; (b) Shawver, L. K.; Lipson, K. E.;
Fong, T. A. T.; McMahon, G.; Plowman, G. D.; Strawn, L. M. Drug Discovery Today
1997, 2, 50; (c) Folkman, J. Nat. Med. 1995, 1, 27.
2. Ferrara, N.; Gerber, H.-P.; LeCouter, J. Nat. Med. 2003, 9, 669.
3. (a) Kim, K. L.; Shin, I.-S.; Kim, J.-M.; Choi, J.-H.; Byun, J.; Jeon, E.-S.; Suh, W.; Kim,
D.-K. Cardiovasc. Res. 2006, 72, 394; (b) Yu, Q. Future Med. 2005, 1, 475; (c)
Peters, K. G.; Kontos, C. D.; Lin, P. C.; Wong, A. L.; Rao, P.; Huang, L.; Dewhirst,
M. W.; Sankar, S. Recent Prog. Horm. Res. 2004, 59, 51; (d) Jones, N.; Iljin, K.;
Dumont, D. J.; Alitalo, K. Nat. Rev. Mol. Cell Biol. 2001, 2, 257.
*
500
0
*
*
Vehicle
10 mpk 30 mpk
+ h-Ang-1 + h-Ang-1 + h-Ang-1 + h-Ang-1
100 mpk
Vehicle
4. (a) Johnson, N. W.; Semones, M.; Adams, J. L.; Hansbury, M.; Winkler, J. Bioorg.
Med. Chem. Lett. 2007, 17, 5514; (b) Cee, V. J.; Albrecht, B. K.; Geuns-Meyer, S.
D.; Hughes, P. E.; Bellon, S.; Bready, J.; Caenepeel, S.; Chaffee, S. C.; Coxon, A.;
Emery, M.; Fretland, J.; Gallant, P.; Gu, Y.; Hodous, B. L.; Hoffman, D.; Johnson,
R. E.; Kendall, R.; Kim, J. L.; Long, A. M.; McGowan, D.; Morrison, M.; Olivieri, P.
R.; Patel, V. F.; Polverino, A.; Powers, D.; Rose, P.; Wang, L.; Zhao, H. J. Med.
Chem. 2007, 50, 627; (c) Hodous, B. L.; Geuns-Meyer, S. D.; Hughes, P. E.;
Albrecht, B. K.; Bellon, S.; Bready, J.; Caenepeel, S.; Cee, V. J.; Chaffee, S. C.;
Coxon, A.; Emery, M.; Fretland, J.; Gallant, P.; Gu, Y.; Hoffman, D.; Johnson, R. E.;
Kendall, R.; Kim, J. L.; Long, A. M.; Morrison, M.; Olivieri, P. R.; Patel, V. F.;
Polverino, A.; Rose, P.; Tempest, P.; Wang, L.; Whittington, D.; Zhao, H. J. Med.
Chem. 2007, 50, 611. and references therein.
5. Polyploidy is a hallmark of Aurora B or Aurora A/B inhibition: Yang, H.; Burke,
T.; Dempsey, J.; Diaz, B.; Collins, E.; Toth, J.; Beckmann, R.; Ye, X. FEBS Lett. 2005,
579, 3385.
6. Carmena, M.; Earnshaw, W. C. Nat. Rev. Mol. Cell Biol. 2003, 4, 842.
7. The JNK inhibitor SP600125 is reported to impact tumor growth and
angiogenesis, but this is likely an off-target activity: Schmidt, M.;
Budirahardja, Y.; Klompmaker, R.; Medema, R. EMBO Rep. 2005, 6, 866.
8. Atomic coordinates for the Aurora A/1, JNK3/15, and KDR/11 crystal structures
have PDB Codes 3EFW, 3DA6, and 3EWH, respectively.
Figure 5. Compound 15 inhibits h-Ang-1 stimulated Tie-2 phosphorylation in
mouse lung. Data points represent the mean SD, n = 3: (*) p 6 0.05 versus
vehicle + h-Ang-1 by one-way ANOVA with Dunnett’s post-hoc test. Bars represent
phosphorylated Tie-2, and circles represent plasma concentration of 15.
fold greater than the cellular IC50 of 15 (2 nM). However, taking
into account nonspecific binding to mouse plasma proteins
(15 = 99.6% bound in mouse plasma), the free unbound fraction
of 15 is 1 nM, approximately equivalent to the IC50 determined in
the cell-based assay (2 nM). The robust inhibition of Tie-2 phos-
phorylation upon oral administration of 15 indicates that this mol-
ecule may be a suitable tool for the in vivo evaluation of Tie-2
signaling in angiogenesis.
Conclusion: An effort to reduce KDR inhibitory activity of the
dual Tie-2/KDR inhibitor 1 resulted in 15, a potent Tie-2 inhibitor
(IC50 = 2 nM) with excellent selectivity over KDR (>2500ꢀ) and
selectivity in cell-based assays over Aurora (>600ꢀ). Structural
information from a JNK3/15 co-crystal suggests a DFG-in binding
mode in Tie-2 that is not tolerated by KDR. In a mouse pharmaco-
dynamic model, 15 exhibited a dose-dependent inhibition of Ang-
1-stimulated Tie-2 phosphorylation in lung tissue, indicating that
9. Kim, S.; Yi, K. Y. Tetrahedron Lett. 1985, 26, 1661.
10. Cee, V. J.; Downing, N. S. Tetrahedron Lett. 2006, 47, 3747.
11. SCWRL 3.0 was used for initial side-chain rotamer replacement using the Tie-2
sequence, and the resulting model underwent restrained energy minimization
using the AMBER89 force-field implemented in MOE 2006.08 (Chemical
Computing Group, Montreal, Canada).
12. Shewchuk, L. M.; Hassell, A. M.; Ellis, B.; Holmes, W. D.; Davis, R.; Horne, E. L.;
Kadwell, S. H.; McKee, D. D.; Moore, J. T. Structure 2000, 8, 1105.
13. Cherry, M.; Williams, D. H. Curr. Med. Chem. 2004, 11, 663.