Potent quinoxaline-based inhibitors of PDGF receptor tyrosine kinase activity. Part 1: SAR exploration and effective bioisosteric replacement of a phenyl substituent

Article abstract of DOI:10.1016/S0960-894X(03)00654-1

Novel substituted 2-anilino- and 2-cycloalkylaminoquinoxalines have been found to be useful and selective inhibitors of PDGF-R autophosphorylation. Replacement of an anilino-substituent with substituted cyclohexylamino- or norbornylamino substituents led

Full text of DOI:10.1016/S0960-894X(03)00654-1

Bioorganic & Medicinal Chemistry Letters 13 (2003) 3091–3095
Potent Quinoxaline-Based Inhibitors of PDGF Receptor
Tyrosine Kinase Activity.Part 1: SAR Exploration and
Effective Bioisosteric Replacement of a Phenyl Substituent§
Michael R. Myers,* Wei He, Barbara Hanney, Natalie Setzer, Martin P. Maguire
Allison Zulli, Glenda Bilder, Helen Galzcinski, Dilip Amin, Saul Needle
and Alfred P. Spada
Aventis Pharmaceuticals, Route 202/206, Bridgewater, NJ 08807, USA
Received 14 April 2003; accepted 20 May 2003
Abstract—Novel substituted 2-anilino- and 2-cycloalkylaminoquinoxalines have been found to be useful and selective inhibitors of
PDGF-R autophosphorylation. Replacement of an anilino-substituent with substituted cyclohexylamino- or norbornylamino sub-
stituents led to significant improvements in the pharmacokinetic profile of these analogues.
# 2003 Elsevier Ltd. All rights reserved.
Percutaneous transluminal coronary angioplasty
(PTCA) is a highly successful surgical procedure used to
clear blocked coronary arteries. The process is relatively
non-invasive and easily performed; it is estimated that
more than one million procedures are performed
annually worldwide. The incidence of acute side effects
is very low and patients recover rapidly. However,
PTCA leads to a process termed restenosis that involves
the migration, proliferation, matrix formation and
remodeling associated with the initial injury. This pro-
cess can lead to significant coronary artery blockage
over a period of up to 6 months in 25–40% of all
patients.
The key role of platelet derived growth factor (PDGF)
and its receptor (PDGF-R) in the process leading to
long-term restenosis after PTCA has been investigated
in depth.3 The growth factor and its receptor have also
been implicated in wound healing, tumorigenesis,
angiogenesis and atherosclerosis. Our early discovery
efforts to find a selective inhibitor of PDGF-R auto-
phosphorylation4 resulted in the identification of a now
well-studied prototype, RPR101511A (Table 1), which
was ultimately evaluated in the Yucatan mini-pig
model.5 The positive activity seen in this rigorous model
of restenosis gave us significant confidence in proceed-
ing with our optimization program. Our objectives were
thus: (1) to improve in vitro activity, (2) to improve
pharmacokinetic profile, (3) to maintain selectivity and
(4) to expand the SAR of the series.
New surgical technologies including the placement of
metallic stents have been employed in attempts to limit
restenosis.2 Despite the successes in stent technology for
acute restensosis, the rate of long-term restenosis
remains relatively constant. A host of clinical trials over
the last 15 years with a large variety of pharmacological
agents have also failed to impact the rate of restenosis
significantly. However, recent success in lowering the
rate of restenosis has been reported using drug-coated
metallic stents.
Herein, we report our first written communication on
these optimization efforts and we present an interesting
SAR study regarding the activity of this class of ATP-
competitive PDGF-R inhibitors.
Hetereoatom linked molecules, generically described as
1 in Figure 1, were initially targeted as PDGF receptor
tyrosine kinase inhibitors. It was thought that the ease
of synthesis and the potential for a rapid exploration of
SAR would be profitable. The general methods of pre-
paration for these compounds have already been repor-
§This workwas presented in part at the conferences shown in ref 1.
*Corresponding author at present address: Lilly Corporate Center,
Indianapolis, IN 46285, USA. Fax: +1-317-276-1177.. E-mail:
myers_michael_r@lilly.com
ted in
a
series of published PCT applications.6
0960-894X/$ - see front matter # 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/S0960-894X(03)00654-1

3092
M. R. Myers et al. / Bioorg. Med. Chem. Lett. 13 (2003) 3091–3095
Figure 1. RPR101511 and hetereoatom linked analogues.
Scheme 3. Synthesis and resolution of norbonyl analogues.
Scheme 1. Synthesis of quinoxaline analogues: (i) RNH2/180 ꢀC;
(ii) RONa/THF.
Scheme 4. Asymmetric synthesis of the norbonyl analogue.
inversion using pthalimide followed by hydrazine-medi-
ated liberation of the amine provided 10. We then
applied the palladium-mediated coupling conditions of
Buchwald10 in order to conserve the precious chiral
amine 10. Optical rotation studies confirmed that the
levorototory isomer is 39.
Scheme 2. Synthesis of quinoline analogues: (i) cyclohexanone,
NaBH3(CN); (ii) NaNO2/HCl; (iii) cyclohexanol/DEAD/Ph3P.
Our early prototype PDGF-R inhibitor, RPR101511,
has an IC50 of 461 nM in an ELISA-based assay for
PDGF-R autophosphorylation. Simple insertion of a
nitrogen between the two aromatic rings leads to a
nearly 2-log improvement in activity (17 vs
RPR101511). For comparison, the des-dimethoxy deri-
vative, 11, is 1 log less potent than RPR101511. How-
ever, a single methoxyl substitution can improve
potency by up to 100-fold depending on the position of
substitution (compare 13, 14, 15, 16 vs 11). It is clear
that substitutions at the 5 and 8 positions of quinoxaline
are less tolerated than those at the 6 and 7 positions (15,
16 vs 13, 14). The fact that the dimethyl substituted
anilinoquinoxaline 12 is 60-fold less potent than the
dimethoxyl analogue 17 seems to suggest that the two
oxygen atoms might be involved in hydrogen bonding
and that this region of the quinoxaline could be used as
a handle for chemical manipulations in order to change
the chemical and physical properties of target mole-
cules. Unfortunately, these anilino analogues suffered a
similar pharmacokinetic profile to RPR101511 in that
they induced p450 enzymes (in rat) that led, over the
course of several days dosing, to a greatly diminished
exposure via oral administration as discussed further in
the following paper.
Generally, the amino-substituted quinoxaline deriva-
tives 3 were prepared by heating a >5-fold excess of the
amine or aniline with the appropriate 2-chloroquinox-
aline 2 for 3–12 h at 180 ꢀC (Scheme 1). These reaction
conditions limited our ability to prepare a diverse array
of analogues from sensitive or precious amines directly.
Significant amounts of des-methyl impurities were often
isolated in the coupling of less reactive amines with 2-
chloro-6,7-dimethoxyquinoxaline. The preparation of
the oxy and thio derivatives 4 (X=O, S) from the cor-
responding alkali salts was straightforward. Such reac-
tions proceeded at much lower temperature and were
complementary to the fusion reaction for the synthesis
of aminoquinoxalines 3.
The preparation of the quinoline derivative, 24, was
carried out by reductive amination with 3-amino-6,7-
dimethoxyquinoline 5b and the appropriate ketone. The
oxy-substituted quinoline 25 was prepared via Mitsu-
nobu reaction with 6,7-dimethoxy-3-hydroxyquinoline 6
and the corresponding alcohol (Schemes 2 and 3).
The racemic exo-norbonyl derivative 34, prepared by
the fusion reaction of compounds 7 and 8, was resolved
by chiral HPLC to afford two single enantiomers 38 and
39. An asymmetric synthesis was conducted to confirm
the absolute stereochemical assignments as outlined in
Scheme 4. The preparation begins with the commer-
cially available (+)-(R)-endo-norboneol; Mitsunobu
The first significant breakthrough towards an improve-
ment in PK profile was realized when compound 22
was evaluated side-by-side with RPR101511 in an in
vitro assay of p450 upregulation (data not shown). It
was determined that 22 did not induce upregulation.

M. R. Myers et al. / Bioorg. Med. Chem. Lett. 13 (2003) 3091–3095
3093
Moreover, single-dose po administration over 4 days
demonstrated that exposure levels were maintained
between days 1 and 4, albeit with low exposure. These
positive results and the fact that 22 was only 4-fold
more potent than RPR101511 led us to explore fur-
ther the SAR of substituted cycloalkyl derivatives. The
highlights of some of our results are presented in
Table 1.
It can be seen that the cyclohexylamino quinoline ana-
logue 24 was slightly more potent than quinoxaline 22.
Interestingly, the oxo quinoline derivative 25 was
ꢁ5-fold less potent than 22 and 24 but the cyclohex-
yloxy-quinoxaline, 23, was 2-fold more potent. Going
further it can be seen that cyclopentyl is nearly equivalent
to cyclohexyl for the oxo and amino linkers (26 vs 22 and
27 vs 23). Addition of one carbon atom (33) significantly
Table 1. PDGF-R activity
Compd
R1
R2
R3
R4
X
Y
R
PDGF-R (IC50, mM)
RPR101511
0.461
5.1
0.34
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
H
H
MeO
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
H
Me
H
H
H
H
Me
H
H
MeO
H
H
H
H
MeO
H
H
H
H
H
H
H
H
H
H
H
H
H
H
NH
NH
NH
NH
NH
NH
NH
NMe
NH
NH
NH
NH
O
NH
O
NH
O
S
S
N
N
N
N
N
N
N
N
N
N
N
N
N
CH
CH
N
N
N
Ph
Ph
Ph
Ph
Ph
Ph
Ph
Ph
11% @ 1 mM
5% @ 1 mM
1.6
0.042
0.006
0% @ 1 mM
0.004
8% @ 1 mM
0.092
0.124
MeO
MeO
MeO
MeO
MeO
MeO
MeO
MeO
MeO
MeO
MeO
MeO
MeO
MeO
MeO
MeO
MeO
MeO
MeO
MeO
MeO
MeO
MeO
MeO
MeO
MeO
MeO
MeO
MeO
MeO
MeO
3Cl–Ph
4Cl–Ph
Bn
Cyclohexyl
Cyclohexyl
Cyclohexyl
Cyclohexyl
Cyclopentyl
Cyclopentyl
Cyclohexyl
Cyclopentyl
3-Methylcyclohexyl
4-Ethylcyclohexyl
0.065
0.096
0.601
0.128
0.159
0% @ 1 mM
0.226
0.041
H
H
H
H
H
H
N
N
N
NH
O
16% @ 1 mM
32
33
34
35
36
37
38
39
H
H
H
H
H
H
H
H
MeO
MeO
MeO
MeO
MeO
MeO
MeO
MeO
MeO
MeO
MeO
MeO
MeO
MeO
MeO
MeO
H
H
H
H
H
H
H
H
NH
NH
O
N
N
N
N
N
N
N
N
0.076
12% @ 1 mM
0.025
O
0.265
NH
NH
NH
NH
0.020
0.879
0.015
0.742

3094
M. R. Myers et al. / Bioorg. Med. Chem. Lett. 13 (2003) 3091–3095
Table 2. Comparison of inhibition of PDGF-R autophos phorylation and mitogenesis in human aortic smooth muscle cells
Compd
PDGF-R (IC50, mM)
HASMC (IC50, mM)
Ratio activity mito./enzyme
32
34
38
39
0.076
0.025
0.015
0.742
0.353
0.395
0.061
0.585
5
15
4
0.8
decreases activity. Exchange of the amine or oxo linker
for sulfur also leads to a dramatic loss in activity for the
cyclohexyl analogue 28, whereas a loss of only 2-fold or
less is seen for the thio cyclopentyl derivative (29). This
set of results underscores the significant steric con-
straints imposed by the lipophilic binding pocket adja-
cent to the ATP-binding domain and are consistent with
the steric restrictions seen in our earliest studies.7 This
binding domain is often cited as being crucial to kinase
selectivity and potency. Insensitivity to replacing the
NH-linker by a sulfur or oxygen also implies that there
is no absolute requirement for a hydrogen bond donor.
at the 4-position of cyclohexyl ring resulted in sig-
nificant loss of activities (31 vs 22).
Cyclohexane has long been considered a bioisosteric
equivalent for a phenyl group mainly based on size and
lipophillic nature. In our case, replacement of the
phenyl with cyclohexane results in a 20-fold loss in
activity and a significant change in PK profile. A similar
improvement in PK profile was reported previously by
Tucker et al.9 in a study of thrombin inhibitors. How-
ever, in their case the improvement in exposure did not
provide a compound with improved efficacy due to
strong protein binding. For our purposes, it appears
that both norbornane and 3-methylcyclohexane are
equivalent bioisosteres of phenyl for PDGF-R. Poten-
tially significant advantages are readily apparent with
these bioisosteres of benzene: (1) greater number of sites
for substitution allowing for greater diversity; (2) crys-
tallinity parameters can be significantly changed (inter-
rupt p–p stacking) possibly leading to improved
dissolution; (3) a significant reduction in p450 upregu-
lation leading to reduced exposures, and (4) potential
for reduced p450 enzyme interactions due to the non-
aromatic pharmacophore.
To this last point, replacement of NH with a NMe-lin-
ker, (18 vs 17), leads to significant loss of activity in the
aryl series. We suggest that this is due to steric con-
straints of the lipophillic binding domain and/or con-
formational issues with the orientation of the aryl or
cycloalkyl lipophillic substituent rather than loss of a
significant H-bond between the NH-linker and the
protein backbone.8 We found that the NH-linked
analogues generally exhibited superior physicochemical
profiles relative to the oxo or thia derivatives although
we continued to exploit all of these for our SAR
studies.
In conclusion, highly optimized PDGF-R tyrosine
kinase inhibitors have been identified. The initial ani-
lino-leads suffered from significant issues with bioavail-
ability and induction of p450 enzymes. However,
optimized analogues like 32 (RPR127963) have been
shown to have superior profiles making them useful
compounds for further in vivo profiling as presented in
the accompanying paper. Simple bioisosteric replace-
ments for the aryl group were shown to have a sig-
nificant positive impact on PK properties; a principle
which should be generally applicable to other families of
inhibitors.
The next significant advance in our studies came with
the observation that the racemic oxy-linked norbornyl
analogue 34 exhibited 25 nM activity, thus recovering
much of the loss seen on going from anilino analogues
(17) to simple cycloalkyls (22, 26). Clearly the exo
orientation is preferred over endo (34 vs 35). Simple
replacement of the oxygen with an NH actually leads to
no change or a slight loss in activity for the exo and
endo racemic mixtures (34 vs 36 and 35 vs 37). A 50-fold
preference for the exo derivative 38 versus its enantiomer
39 is worthy of note. When evaluated for in-cell inhibi-
tion of PDGF-stimulated mitogenesis these two analo-
gues demonstrate a 10-fold separation in IC50 values
(Table 2). Generally, our most potent compounds in the
ELISA assay tend to be the most active in the mitogen-
esis assay. In situ autophosphorylation can be used to
confirm that mitogenesis activity is indeed associated
with inhibition of PDGF autophosphorylation.
Acknowledgements
The authors wish to thankthe analytical chemistry
department for obtaining MS and combustion data. We
would also like to acknowledge the work of our ADME
colleagues especially Zaid Jayyosi and his team for the
p450 induction workand Ken Page and his team for the
bioavailability studies performed over the course of this
program.
The next active hit from the series was identified from a
small library of compounds prepared via reductive
amination with a series of 3-aminoquinoline derivatives
and a number of aldehydes and ketones (data not
shown). Subsequent coupling of 3-methylcyclohexyl-
amine with 2-chloro-6,7-dimethoxyquinoxaline pro-
vided the target as a mixture of cis- and trans-isomers,
30, which was 3-fold more potent than the parent com-
pound 22. In contrast, substitution with an ethyl group
References and Notes
1. (a) He, W.; Myers, M. R.; Spada, A. P.; Hanney, B.; Setzer,
N. Abstracts of Papers, 217th ACS National Meeting, Anaheim,

M. R. Myers et al. / Bioorg. Med. Chem. Lett. 13 (2003) 3091–3095
3095
CA, March 21–25, 1999; ORGN-30. (b) He, W.; Myers, M.
R.; Spada, A. P.; Hanney, B.; Setzer, N.; Maguire, M.; Mail-
liet, P.; Gontier, S.; Guegen, J.-C.; Cans, P.; Orton, E.; Cheve,
M.; Kubiak, G. G.; Shah, H. C.; O’Brien, M. K. Abstracts of
Papers, 219th ACS National Meeting, San Francisco, CA,
March 26–30, 2000; ORGN-225. (c) He, W.; Myers, M. R.;
Spada, A. P.; Hanney, B.; Setzer, N.; Mailliet, P.; Gontier, S.;
Guegen, J.-C.; Cans, P.; Orton, E.; Cheve, M.; Kubiak, G. G.;
Shah, H. C.; O’Brien, M. K.; Bilder, G. E.; Allen, E.; Bissery,
M.-C.; Page, K.; Natarajan, C.; Jayyosi, Z.; Kelley, M. F.;
Toutain, H.; Constan, A.; Amin, D.; Needle, S.; Galczenski,
H.; Wang, W.; Perrone, M. H. Abstracts of Papers, 219th ACS
National Meeting, San Francisco, CA, March 26–30, 2000;
MEDI-307. (d) Myers, M. R. The SCI-Fine Chemicals spon-
sored conference Protein Tyrosine Kinases, Good Targets for
New Drugs?, London, May 22, 2001.
5. (a) Bilder, G. E.; Wentz, T.; Leadley, R.; Amin, D.; Dugan,
L.; O’Connor, B.; Needle, S.; Myers, M. R.; Spada, A.; Page,
K.; Perrone, M.; Dunwiddie, C. Circulation 1996, 94 (Suppl.
I). (b) Bilder, G. E.; Wentz, T.; Leadley, R.; Amin, D.; Byan,
L.; O’Conner, B.; Needle, S.; Galczenski, H.; Bostwick, J.;
Kasiewski, C.; Myers, M. R.; Spada, A. P.; Merkel, L.; Ly, C.;
Persons, P.; Page, K.; Perrone, M.; Dunwiddie, C. Circulation
1999, 99, 3292.
6. (a) Myers, M.; Spada, A. P.; Persons, P. E.; Maguire, M. P.
PCT Int. Appl., WO9854156 A1 19981203. (b) Spada, A. P.;
He, W.; Myers, M. PCT Int. Appl., WO9854157 A1 19981203.
(c) Myers, M.; He, W.; Spada, A. P.; Maguire, M. P. PCT Int.
Appl., WO9854158 A1 19981203.
7. Maguire, M. P.; Sheets, K. R.; McVety, K.; Spada, A. P.;
Zilberstein, A. J. Med. Chem. 1994, 37, 2129.
8. Myers, M. R.; Setzer, N. N.; Spada, A. P.; Persons, P. E.;
Ly, C. Q.; Maguire, M. P.; Zulli, A. L.; Cheney, D. L.; Zil-
berstein, A.; Johnson, S. E.; Franks, C. F.; Mitchell, K. J.
Bioorg. Med. Chem. Lett. 1997, 7, 421.
2. Bassett, P. Drug Market Develop. 2000, 327.
3. Boschelli, D. H. Drugs Future 1999, 24, 515.
4. For a few recent reports of PDGF-R inhibitors see: (a) Mat-
suno, K.; Nakajima, T.; Ichimura, M.; Giese, N. A.; Yu, J.-C.;
Lokker, N. A.; Ushiki, J.; Ide, S.; Oda, S.; Nomoto, Y. J. Med.
Chem. 2002, 45, 4513. (b) Mahboobi, S.; Teller, S.; Pongratz,
H.; Hufsky, H.; Sellmer, A.; Botzki, A.; Uecker, A.; Beckers, T.;
Baasner.; S.; Schaechtele, C.; Ueberall, F.; Kassack, M.U.;
Dove, S.; Boehmer, F.-D. J. Med. Chem. 2002, 45, 1002.
9. Tucker, T. J.; Lumma, W. C.; Lewis, S. D.; Gardell, S. J.;
Lucas, B. J.; Baskin, E. P.; Woltmann, R.; Lynch, J. L.; Lyle,
E. A.; Appleby, S. D.; Chen, I.-W.; Dancheck, K. B.; Vacca,
J. P. J. Med. Chem. 1997, 40, 1565.
10. Wolfe, J. P.; Wagsaw, S.; Buchwald, S. L. J. Am. Chem.
Soc. 1996, 118, 7215.