4592
D. A. Scott et al. / Bioorg. Med. Chem. Lett. 23 (2013) 4591–4596
F
F
F
F
N
NH
O
N
NH
O
N
NH
O
N
O
N
O
N
O
NH2
NH2
NH2
N
N
N
N
N
13
Figure 3. Cinnolines 13 and 14.
14
Figure 1. 1, AZ683.
active against the target, and binding to CSF-1R was shown to cor-
relate with damage to rat cardiomyocytes for a small set of
compounds.15a
L-type Ca-channel. However, the SAR of the NaV1.5 channel did
not follow the same trends, with the most basic compounds show-
ing reduced activity relative to AZ683. We wanted to establish
whether the reduced calcium channel activity and off-target phar-
macology from less basic, less lipophilic compounds would trans-
late into meaningful improvements in the cardiac action
potential and rat toxicity profiles.
A similar approach was reported during the optimization of a
series of arylamide CSF-1R inhibitors, as part of the work that led
to the identification of JNJ-28312141.7 Early lead compounds were
active in sodium and ion channel binding assays, and also a guinea
pig right atrium assay which measured the rate and force of con-
traction. Less basic compounds within that series delivered re-
duced activity in the ion channel assays.20
While AZ683 was progressing through the pivotaltoxicity stud-
ies, the 3-amido-4-anilinocinnolines were identified as another
series of highly selective CSF-1R inhibitors with an excellent phar-
macokinetic profile and improved cellular potency relative to the
quinoline series.21 In a cell proliferation assay (3T3 cells engi-
neered to express CSF-1R and stimulated with CSF-1),22 cinnolines
In a panel of 85 kinases AZ683 showed >20% inhibition for only
two kinases (CSF-1R 96%, ARK5 51%, 1 l
M concentration).10 The
compound was also profiled in a panel of 69 enzymes, receptors,
transporters and ion channels16 and of those targets associated
with the cardiovascular system,17 it was found to have activity
against those included in Table 1. Binding data (% inhibition at
10 lM) for the alpha 1A, 1D and 2A adrenoreceptors and the L-type
Ca channel, along with hERG and NaV 1.5 ion channel IC50 data is
shown. Due to the Ca (increases in AV conduction and PR interval)
and Na (increases in QRS duration) dependent cardiac electrophys-
iology changes in the guinea-pig and dog, we suspected ion chan-
nel activities were contributing to the cardiovascular toxicity
observed with AZ683.
Therefore, as part of our efforts to deliver a safe and selective
CSF-1R inhibitor, we sought to improve the overall toxicity profile
by identifying compounds with reduced off-target activities. Addi-
tional amidoquinolines were screened in a smaller secondary phar-
macology panel, focusing on the targets in which AZ683 had shown
activity, and others associated with cardiotoxicity (Table 1). Where
possible, matched pairs were included in the selection of com-
pounds, to assess the impact of specific structural changes.
The synthesis of compounds 1–3, 6–7, 9 and 11, along with their
biological activity, physical properties, and pharmacokinetic pro-
files, has been described previously.10 Compounds 8, 10 and 12
were prepared by similar chemistry.18 Compounds 4 and 5 were
also prepared through a similar approach, after introducing the
piperidine substituent via pyridine installation and reduction, as
shown in Scheme 1.19
Compounds incorporating a basic amine (2–7, 11) had a similar
pattern of off-target activity to AZ683, with a trend for increased
potency of the more lipophilic dichloro examples (3 vs 1 and 5
vs 4) against the various targets. The most basic compounds
(piperidines 4 and 5, and homopiperazine 11) also tended to have
higher potency against the targets where activity was observed,
and in some cases picked up additional off-target activity (data
not included). Morpholine 8, 4-hydroxypiperidine 9 and piperazi-
none 12 showed significantly cleaner profiles. Also of note was
hydroxyethyl piperazine 10, with reduced activity against the
13 and 14 (Fig. 3) had an IC50 of 0.013 and 0.025
compared with 0.23 M for AZ683. However, when profiled in the
GP-MAPD assay up to the maximum feasible dose (free plasma
concentrations of 1.5–2.2 M), cinnolines 13 and 14 showed simi-
lM respectively,
l
l
lar dose-dependent increases in AV conduction time and QRS dura-
tion to those seen with AZ683, and the compounds were not
progressed.
It was reported previously for both the quinoline and the cinn-
oline series that potency in the cellular proliferation assay was lar-
gely determined by a combination of the aniline substitution
pattern and the 7-alkoxy group. Cyclic amines at the 6-position,
such as N-methylpiperazine, bestowed suitable physical properties
to deliver good oral PK, and modifications here typically had little
impact on in vitro potency. Therefore, this position offered a suit-
able handle to tune the overall properties of the series. The combi-
nation of 2-F, 4-Me aniline and 7-MeO had delivered good potency
and physical properties in the cinnoline series when combined
with 6-N-methylpiperazine (compound 14), and this scaffold was
selected for further work.
A variety of substituents were
introduced at the 6-position to deliver compounds with a range
Dose 1
(5mg/kg)
Dose 2
(15mg/kg)
Washout
Dose 1
(5mg/kg)
Dose 2
(15mg/kg)
Washout
vehicle
AZ683
75
50
25
0
vehicle
AZ683
120
100
80
60
40
20
0
-25
-20
0
10
20
30
40
50
60
0
10
20
30
Time (min)
40
50
60
Time (min)
Figure 2. Effects of AZ683 on AV conduction time and QRS duration in the GP-MAPD assay.