The Identification of Potent, Selective, and Orally Available Inhibitors of Ataxia Telangiectasia Mutated (ATM) Kinase: The Discovery of AZD0156 (8-{6-[3-(Dimethylamino)propoxy]pyridin-3-yl}-3-methyl-1-(tetrahydro-2 H-pyran-4-yl)-1,3-dihydro-2 H-imidazo[4,5- c]quinolin-2-one)

Article abstract of DOI:10.1021/acs.jmedchem.7b01896

ATM inhibitors, such as 7, have demonstrated the antitumor potential of ATM inhibition when combined with DNA double-strand break-inducing agents in mouse xenograft models. However, the properties of 7 result in a relatively high predicted clinically efficacious dose. In an attempt to minimize attrition during clinical development, we sought to identify ATM inhibitors with a low predicted clinical dose (<50 mg) and focused on strategies to increase both ATM potency and predicted human pharmacokinetic half-life (predominantly through the increase of volume of distribution). These efforts resulted in the discovery of 64 (AZD0156), an exceptionally potent and selective inhibitor of ATM based on an imidazo[4,5-c]quinolin-2-one core. 64 has good preclinical phamacokinetics, a low predicted clinical dose, and a high maximum absorbable dose. 64 has been shown to potentiate the efficacy of the approved drugs irinotecan and olaparib in disease relevant mouse models and is currently undergoing clinical evaluation with these agents.

Full text of DOI:10.1021/acs.jmedchem.7b01896

Subscriber access provided by AUSTRALIAN NATIONAL UNIV
The Identification of Potent, Selective and Orally Available Inhibitors
of Ataxia Telangiectasia Mutated (ATM) Kinase: The Discovery of
AZD0156 (8-{6-[3-(dimethylamino)propoxy]pyridin-3-yl}-3-methyl-1-
tetrahydro-2H-pyran-4-yl)-1,3-dihydro-2H-imidazo[4,5-c]quinolin-2-one)
(
Kurt G. Pike, Bernard Barlaam, Elaine Cadogan, Andrew Campbell, Yingxue Chen, Nicola Colclough,
Nichola L Davies, Camila DeAlmeida, Sébastien L. Degorce, Myriam Didelot, Allan Dishington,
Richard Ducray, Stephen T. Durant, Lorraine A. Hassall, Jane L. Holmes, Gareth D Hughes, Philip
A MacFaul, Keith R. Mulholland, Thomas M. McGuire, Gilles Ouvry, Martin Pass, Graeme R. Robb,
Natalie Stratton, Zhenhua Wang, Joanne Wilson, Baochang Zhai, Kang Zhao, and Nidal Al-Huniti
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.7b01896 • Publication Date (Web): 23 Apr 2018
Downloaded from http://pubs.acs.org on April 23, 2018
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a service to the research community to expedite the dissemination
of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in
full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully
peer reviewed, but should not be considered the official version of record. They are citable by the
Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore,
the “Just Accepted” Web site may not include all articles that will be published in the journal. After
a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web
site and published as an ASAP article. Note that technical editing may introduce minor changes
to the manuscript text and/or graphics which could affect content, and all legal disclaimers and
ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or
consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W.,
Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
The Identification of Potent, Selective and Orally
Available Inhibitors of Ataxia Telangiectasia Mutated
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(ATM) Kinase: The Discovery of AZD0156 (8ꢀ{6ꢀ[3ꢀ
(
dimethylamino)propoxy]pyridinꢀ3ꢀyl}ꢀ3ꢀmethylꢀ1ꢀ
(tetrahydroꢀ2Hꢀpyranꢀ4ꢀyl)ꢀ1,3ꢀdihydroꢀ2Hꢀ
imidazo[4,5ꢀc]quinolinꢀ2ꢀone)
,
1
1
1
2
3
Kurt G. Pike,* Bernard Barlaam, Elaine Cadogan, Andrew Campbell, Yingxue Chen, Nicola
1
1
1
1,6
6
Colclough, Nichola L. Davies, Camila de-Almeida, Sebastien L. Degorce, Myriam Didelot, Allan
4
6
1
4
4
Dishington, Richard Ducray, Stephen T. Durant, Lorraine A. Hassall, Jane Holmes, Gareth D.
1
4
5
1
6
1
Hughes, Philip A. MacFaul, Keith R. Mulholland, Thomas M. McGuire, Gilles Ouvry, Martin Pass,
1
7
8
1
8
8
Graeme Robb, Natalie Stratton, Zhenhua Wang, Joanne Wilson, Baochang Zhai, Kang Zhao, Nidal
3
Al-Huniti.
1
Oncology, IMED Biotech Unit, AstraZeneca, Building 310, Cambridge Science Park, 319 Milton
2
Road, Cambridge CB4 0WG, UK. Pharmaceutical Sciences, AstraZeneca, Silk Road Business Park,
3
Macclesfield, SK10 2NA, UK. Oncology, IMED Biotech Unit, AstraZeneca, 35 Gatehouse Drive,
4
Waltham, MA 02451, USA. Oncology, IMED Biotech Unit, AstraZeneca, Alderley Park,
5
Macclesfield, Cheshire SK10 4TG, UK Chemical Development, AstraZeneca, Silk Road Business
6
Park, Macclesfield, SK10 2NA, UK. Oncology, IMED Biotech Unit, AstraZeneca, Centre de
ACS Paragon Plus Environment
1

Page 3 of 54
Journal of Medicinal Chemistry
7
Recherches, Z.I. la Pompelle, BP 1050, 51689 Reims Cedex 2, France. Discovery Sciences,
1
2
3
4
5
6
7
8
9
8
AstraZeneca, Alderley Park, Macclesfield, Cheshire SK10 4TG, UK. Pharmaron Beijing Co., Ltd., 6
Taihe Road BDA, Beijing, 100176, P.R. China.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ABSTRACT
ATM inhibitors, such as 7, have demonstrated the antiꢀtumor potential of ATM inhibition when
combined with DNA double strand breakꢀinducing agents in mouse xenograft models. However, the
properties of 7 result in a relatively high predicted clinically efficacious dose. In an attempt to minimize
attrition during clinical development we sought to identify ATM inhibitors with a low predicted clinical
dose (<50 mg) and focussed on strategies to increase both ATM potency and predicted human
pharmacokinetic halfꢀlife (predominantly through the increase of volume of distribution). These efforts
resulted in the discovery of 64 (AZD0156), an exceptionally potent and selective inhibitor of ATM
based on an imidazo[4,5ꢀc]quinolinꢀ2ꢀone core. 64 has good preclinical phamacokinetics, a low
predicted clincal dose and high maximum absorbable dose. 64 has been shown to potentiate the efficacy
of the approved drugs irinotecan and olaparib in disease relevant mouse models and is currently
undergoing clinical evaluation with these agents.
INTRODUCTION
The DNA within our cells is constantly under attack from endogenous sources, environmental
1
,2
mutagens and carcinogens with tens of thousands of DNA damage events occurring every day. To
effectively detect and repair the different types of DNA damage incurred the body has developed a suite
of different repair mechanisms and signaling pathways, collectively termed the DNA damage response
3
(DDR). Although not the most frequent form of DNA damage, DNA doubleꢀstrand breaks (DSBs) are
the most cytotoxic and are repaired by either the homologous recombination repair (HRR) pathway or
the nonꢀhomologous endꢀjoining (NHEJ) pathway. The HRR pathway is a relatively efficient and
ACS Paragon Plus Environment
2

Journal of Medicinal Chemistry
Page 4 of 54
accurate repair pathway for DNA DSBs but requires the presence of undamaged sister chromatid DNA,
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
4
,5
whereas the NHEJ is less accurate but is not dependent on the presence of replicated DNA. Cancer
cells typically have high levels of replication stress, defects in one or more DDR pathway and a higher
level of endogenous DNA damage resulting from a variety of factors associated with increased
proliferation and loss of cell cycle checkpoints. The result of these differences mean that cancer cells are
more sensitive to exogenous DNA damage than normal cells and supports the hypothesis that targeting
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
6
DDR pathways may provide an effective avenue for cancer therapies. The formation of DNA DSBs is
the primary driver for the cytotoxicity for a number of approved cancer chemotherapies, such as
topoisomerase I and II inhibitors (e.g. irinotecan), as well as for ionizing radiation (IR). In addition,
DNA DSBs can be induced by targeting specific DDR pathways, such as the inhibition of PARP with
the approved drug olaparib. Opportunities to interfere with the repair of DNA DSBs may therefore offer
the potential to enhance the cytotoxicity of DSB inducing agents.
Ataxia telangiectasia Mutated (ATM) derives its name from a rare human autosomal recessive
disorder ataxiaꢀtelangiectasia (AꢀT) that results from mutations in the ATM gene which causes patients
to suffer a variety of symptoms including immunodeficiency, extreme radiosensitivity and a
7
predisposition to cancer. ATM kinase is a member of the phosphatidylinositol 3ꢀkinase (PI3K)ꢀrelated
kinase (PIKK) family of atypical serine/threonine protein kinases (also comprising ATR, DNAꢀPKcs,
mTOR, SMG1 and the nonꢀenzymatic TRRAP) and plays a central role in both the early signaling of,
and the protection of cells against, DNA DSBs and reactive oxygen species (ROS) that radiotherapy and
8ꢀ10
a wide range of chemotherapies induce.
ATM is a large protein (350 kDa) comprising 3056 residues
and containing a kinase domain with a relatively high degree of similarity to all isoforms of the lipid
1
1
kinase PI3K. ATM, in its inactive dimeric form, is recruited to the site of a DNA DSB by the DNAꢀ
end tethering MRE11ꢀRAD50ꢀNBS1 (MRN) complex and results in the autophosphorylation of ATM
1
2,13
on Serine 1981 and subsequent dimer dissociation.
The activated ATM monomer phosphorylates a
broad range of downstream targets, including γH2AX, p53 Mdm2 and Chk2, thereby initiating a
ACS Paragon Plus Environment
3

Page 5 of 54
Journal of Medicinal Chemistry
1
4ꢀ18
signaling network capable of activating both cell survival and cell death pathways.
Loss of ATM
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
activity results in the sensitization of virtually all cells to IR whereas upregulation of ATM signaling has
1
9ꢀ21
been observed to result in both chemoresistance and radioresistance in cancer cells.
The central role
of ATM in coordinating both the HR and NHEJ repair pathways in response to induced DNA DSBs,
and its nonꢀcanonical responses to cellular stress, highlights this protein as a highly attractive
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
2ꢀ24
pharmacological target to sensitize tumor cells to both chemotherapy and radiotherapy.
A number of selective inhibitors of ATM have been disclosed the first of which, 2 (KUꢀ55933),
was developed by KuDOS Pharmaceuticals (now part of AstraZeneca) following the extensive
structureꢀactivity relationship (SAR) studies performed on chemical scaffolds related to the panꢀPI3K
25,26
and panꢀPIKK inhibitor 1 (LY294002), figure 1.
Compound 2 is a potent, ATPꢀcompetitive inhibitor
of ATM enzyme (IC = 0.013 ꢁM) with high levels of selectivity over the closely related enzymes
50
DNAꢀPK (IC = 2.5 ꢁM), ATR (IC >100 ꢁM), mTOR (IC = 9.3 ꢁM) and PI3K (IC = 17 ꢁM) and
5
0
50
50
50
has been shown to sensitize HeLa cells to the cytotoxic effects of topoisomerase I and II inhibitors and
2
7
to IR. The utility of 2 has been limited to in vitro assessment as a result of poor physicochemical
properties, in particular low aqueous solubility and low oral bioavailability. Subsequent optimization of
the scaffold resulted in the discovery of second generation ATM inhibitors, such as 3 (KUꢀ59403), with
2
8
improved potency and properties. Compound 3 is a potent inhibitor of ATM enzyme (IC = 0.003
5
0
ꢁM) with >300 fold selectivity over ATR, DNAꢀPK, mTOR and PI3K and showed effective
chemosensitization in cells at 1 ꢁM. Despite the improved solubility of 3 the compound remains
unsuited to oral administration (PO); however, intraperitoneal administration (IP) has allowed the study
of 3 in tumorꢀbearing mice. Tumor growth delay was observed following administration of 3 (25 mg/kg
IP BID) for 5 days when combined with either etoposide or irinotecan. More recently the quinazoline 4
has been reported as a modest inhibitor of ATM in cellꢀbased assays (IC = 1.2 ꢁM) but with good in
5
0
29
vivo pharmacokinetics in mouse (t = 19 h). Quinazoline 4 showed efficacy when evaluated at 10 ꢁM
1
/2
in a clonogenic assay in which MCF7 cells were treated with compound and irradiated with increasing
ACS Paragon Plus Environment
4

Journal of Medicinal Chemistry
Page 6 of 54
doses of IR (0, 2 and 4 Gy) but concerns about its selectivity remain (significant activity was observed
against 41 out of 451 kinases when tested at 3 ꢁM in a panel of diverse kinases).
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Figure 1: Structures of LY294002 (1), KUꢀ55933 (2), KUꢀ59403 (3) and quinazoline 4.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
(LY294002)
2 (KU-55933)
3
(KU-59403)
4
We have recently disclosed the optimization of a quinoline carboxamide scaffold from an initial
screening hit 5, a modestly potent inhibitor in cells with encouraging selectivity over closely related
targets, to deliver potent and selective compounds with oral bioavailability, such as 6 (which has been
3
0
given the external identifier AZ31) and 7, table 1. Such compounds represent a significant advance
allowing the more detailed investigation of ATM inhibition in vivo. Compound 6, administered orally at
1
00 mg/kg QD, has been shown to potentiate the efficacy of IR in an HT29 mouse xenograft model and
the efficacy of irinotecan, administered at 50 mg/kg IP Q7D, in an SW620 mouse xenograft model,
31,32
resulting in tumor regressions.
Compound 7, administered orally at 50 mg/kg BID (on days 2ꢀ4 of a
weekly schedule), has also been shown to potentiate the efficacy of irinotecan, administered at 50 mg/kg
3
0
IP Q7D, and cause tumor regression in an SW620 mouse xenograft model. As anticipated, no
significant activity in these models was observed for either 6 or 7 in the absence of a DNA DSBꢀ
inducing agent. The efficacious doses of 6 and 7 identified provide free plasma concentrations at, or
ACS Paragon Plus Environment
5

Page 7 of 54
Journal of Medicinal Chemistry
above, the cellular IC for ATM for approximately 24 hours in mice on the days that the inhibitors were
5
0
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
dosed. This level and duration of target engagement has subsequently been used to define the likely
efficacious clinical exposure. The increased separation between ATM potency and potency against the
hERG (human etherꢀaꢀgoꢀgo gene related gene) ion channel suggested a reduced risk of adverse
cardiovascular events for compound 7 compared to 6 and its potential as a clinical candidate was
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
3
assessed.
Table 1: Structure and profiles of screening hit 5 and optimized quinoline carboxamide examples AZ31
a
(6) and 7.
5
6 (AZ31)
LogD7.4
7
ATM
Cell IC50 Cell IC50
ATR
b
human /c
rat heps
hERG
IC50
Entry
Aq Sol
5
0.82
0.046
0.033
4.4
>30
>19
3.5
2.5
2.7
19
590
69
12 / 74
<2.6 / <5.3
<1.5 / 5.3
2.3
4.5
22
6
(AZ31)
7
a
All IC50 data are expressed in micromolar (ꢁM) and ATM / ATR potency values are the geometric mean of at least 3
independent measurements. Further details for the biological assays can be found in the supplemental section.
b
Thermodynamic aqueous solubility was measured at pH7.4 and the data are expressed in micromolar (ꢁM). No assessment
of solid state was made.
c
6
The data are expressed in ꢁL/min/10 cells
Physiologicallyꢀbased pharmacokinetic (PBPK) models attempt to portray the body as a series of
compartments that represent tissues and organs connected by the arterial and venous blood flow
ACS Paragon Plus Environment
6

Journal of Medicinal Chemistry
Page 8 of 54
3
4,35
pathways.
These models utilize measured physiological data to describe the tissues and organs by
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
their volume and the blood flow to them and can be used to predict the disposition of novel compounds
in human. Predicted human pharmacokinetic parameters can be used to simulate the dose required to
achieve a target concentration for a defined duration which, when combined with an understanding of
the target engagement required for efficacy in preclinical models, can be used to predict the efficacious
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
6
clinical dose. PBPK modelling suggested that 7 would possess a relatively short halfꢀlife in humans
(~4 hours) and would, therefore, require an oral dose of 700 mg QD to deliver the desired clinical
exposure. The simulation also predicts that a maximum unbound concentration (free Cmax) of 1.3 ꢁM
would be reached following this dose. This relatively high predicted free Cmax increases the likelihood
that nonꢀATM mediated pharmacology may become manifest thus reducing compound tolerability.
Whilst it is appreciated that alternative dosing regimens, for example BID dosing, could be employed in
an effort to maintain the desired level and extend of target engagment whilst reducing both total dose
and free Cmax, such regimens are generally considered less attractive as they place increased burden on
patients to remain compliant.
Many approaches to minimize compound attrition during clinical development have been
suggested and recent analyses have recognized the link between an increased risk of nonꢀtarget
mediated toxicity (including idiosyncratic toxicity) and increasing clinical dose (and by inference
3
7,38
increasing free Cmax).
As a result of these analyses, where possible, AstraZeneca has adopted the
approach of selecting compounds for clinical development which have a low predicted clinically
efficacious dose (<50 mg). However, attrition in clinical development is not solely due to toxicity and
consideration should also be given to the developability of a clinical candidate. For orally administered
compounds the parameter of maximum absorbable dose (D ) is often used, in conjunction with a
abs
predicted clinical dose, to understand whether a significant investment in formulation will be required to
achieve the desired exposure. Poor developability can add significant cost and time to the clinical
development of a clinical candidate and could limit the achievable clinical exposure preventing the
ACS Paragon Plus Environment
7

Page 9 of 54
Journal of Medicinal Chemistry
3
9ꢀ41
biological hypothesis from being properly tested.
Computational simulation software (for example
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
TM
GastroPlus or the GIꢀSim model) can be used to predict Dabs based on easily measured in vitro
parameters, such as aqueous solubility and permeability, combined with knowledge of intestinal surface
4
2
area and transit times. In addition to selecting clinical candidates with a low predicted clinically
efficacious dose ,where possible, AstraZeneca looks to select candidates with a predicted Dabs at least
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
0ꢀfold above the predicted clinically efficacious dose. As a consequence of good permeability in
ꢀ6
MDCK cells overexpressing MDR1 (Papp AꢀB = 12 x 10 cm/s, efflux ratio = 8.7) and moderate
aqueous solubility (69 ꢁM), the Dabs of 7 was predicted to be approximately 500 mg. Whilst the
predicted clinically efficacious dose and Dabs would not necessarily preclude the development of 7 as a
clinical candidate we sought to continue compound optimization with an emphasis on reducing
predicted clinical dose and increasing D . Such compounds, we hypothesized, would carry a reduced
abs
risk of attrition during clinical development.
CHEMISTRY
Compounds reported herein were synthesized as shown in Schemes 1–6. The versatile 6ꢀbromoꢀ4ꢀ
chloroꢀ7ꢀfluoroquinolineꢀ3ꢀcarboxamide (10) was prepared in 4 steps from commercially available 4ꢀ
30
bromoꢀ3ꢀfluoroaniline 8. Reaction of 8 with 1,3ꢀdiethyl 2ꢀ(ethoxymethylidene)propanedioate,
followed by a high temperature cyclisation gave intermediate 9 which could be easily converted to
intermediate 10 by hydrolysis, chlorination and amide formation. Suzuki crossꢀcoupling methodology
was used to install the required aryl substituent into the 6ꢀposition of the quinoline ring to give 11. S Ar
N
4
3
reaction with the appropriate amine, followed by Hoffman rearrangement and subsequent methylation,
gave compounds 12 – 15. Compounds 18 – 20 were prepared from 10 by employing the same
transformations, albeit with the sequence altered. Compound 16 was prepared from 10 by S Ar reaction
N
with 3ꢀ(aminomethyl)pentane followed by Suzuki crossꢀcoupling and a subsequent S Ar reaction on the
N
resultant fluoropyrimidine, scheme 1.
ACS Paragon Plus Environment
8

Journal of Medicinal Chemistry
Page 10 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
a
Scheme 1: Synthesis of 12 – 15, 16 and 18 ꢀ 20.
O
N
R
O
N
O
Cl
N
O
Br
F
N
f,g,h
N
NH2
NH2
F
N
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
F
1
1
1
2 R = 1,4-trans-4-methoxycyclohexyl
3 R = 1,3-cis-3-methoxycyclobutyl
4 R = 1,3-trans-3-methoxycyclobutyl
8
11
a,b
15 R = 1,3-cis-3-hydroxycyclobutyl
e
O
O
Cl
O
H
N
Br
F
Br
N
c,d
i,j,k
O
NH2
N
NH O
N
N
N
H
F
N
NH2
9
10
F
l,m,n,o
16
p
O
N
O
O
O
N
R
O
N
N
NH O
q,r,s
O
N
Br
F
N
NH2
N
F
N
F
N
1
8 R = MeOCH2-
9 R = Me NCH CH CH O-
17
20
1
2
2
2
2
a
Reagents and conditions: (a) 1,3ꢀdiethyl 2ꢀ(ethoxymethylidene)propanedioate, EtOH, 80 °C, 4 h (84%); (b) Ph O, 240 °C,
2
2
.5 h (64%); (c) NaOH, EtOH, water, 75 °C, 2 h (89%); (d) SOCl , DMF, 75 °C, 2 h, NH OH (100%); (e) aryl boronic acid,
2
4
Cs CO , Pd(PPh ) , dioxane:water, 90 °C, 4 h (72%); (f) amine, DIPEA, DMA, 30ꢀ100 °C, 5ꢀ6h (44ꢀ72%); (g)
2
3
3 4
trichloroisocyanuric acid, DBU, MeOH, r.t., 0.25ꢀ2 h (56ꢀ70%); (h) MeI, TBAB, NaOH, DCM, water, r.t., 2.5ꢀ24 h, (21ꢀ
9%); (i) 3ꢀ(aminomethyl)pentane, DIPEA, DMA, 100 °C, 5 h (100%); (j) 2ꢀfluoroꢀ5ꢀ(tetramethylꢀ1,3,2ꢀdioxaborolanꢀ2ꢀ
yl)pyrimidine, Cs CO , Pd(PPh ) , dioxane:water, 95 °C, 3 h (82%); (k) 2ꢀ(pyrrolidinꢀ1ꢀyl)ethanꢀ1ꢀamine, DIPEA, DMA, 34
4
2
3
3 4
°
C, 1 h (80%); (l) (S)ꢀ1ꢀ(tetrahydroꢀ2Hꢀpyranꢀ4ꢀyl)ethanamine, DIPEA, DMA, 100 °C, 24 h (70%); (m) trichloroisocyanuric
acid, DBU, MeOH, r.t., 20 min. (100%); (n) MeI, TBAB, NaOH, DCM, water, r.t., 2.5 h, (98%); (o) aryl boronic acid,
Cs CO , Pd(PPh ) , DMF:water, 90 °C, 2 h (37%); (p) tetrahydroꢀ2Hꢀpyranꢀ4ꢀamine, DIPEA, DMA, 100 °C, 24 h (78%); (q)
2
3
3 4
trichloroisocyanuric acid, DBU, MeOH, r.t., 15 min. (98%); (r) MeI, TBAB, NaOH, DCM, water, r.t., 24 h, (26%); (s) aryl
boronic acid, Cs CO , Pd(PPh ) , dioxane:water, 80ꢀ90 °C, 1.5ꢀ2 h (40ꢀ74%).
2
3
3 4
ACS Paragon Plus Environment
9

Page 11 of 54
Journal of Medicinal Chemistry
The versatile intermediate ethyl 6ꢀbromoꢀ4ꢀchloroꢀ7ꢀfluoroquinolineꢀ3ꢀcarboxylate (21) could be easily
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
prepared from 9 by treatment with thionyl chloride. An S Ar reaction was used to install the
N
methoxycyclobutyl motif, followed by ester hydrolysis, Curtius rearrangement and subsequent
methylation to afford 22. Compounds 24, 25 and 28 were prepared using Suzuki cross coupling
methodology to install a simply substituted pyridyl ring followed by standard manipulations to complete
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
4
4
the elaboration of the pyridyl substituent. Miyaura borylation of 22 followed by Suzuki coupling
afforded compound 27. Compounds 29 ꢀ 33 were synthesized from intermediate 22 by Suzuki coupling
of fully elaborated boronates, scheme 2.
a
Scheme 2: Synthesis of 24, 25 and 27 – 33.
g
9
23
24 R = H
25 R = Me
a
f
b,c,d,e
h
21
22
26
n
i
j,k,l,m
28
2
9 R = HOCH - , X = H
27
2
30 R = MeSO CH - , X = H
2
2
3
3
3
1 R = Me NCH CH O- , X = H
2 2 2
2 R = Me NCH CH CH O- , X = H
2
2
2
2
3 R = Me NCH CH CH O- , X = F
2
2
2
2
ACS Paragon Plus Environment
10

Journal of Medicinal Chemistry
Page 12 of 54
a
Reagents and conditions: (a) SOCl , DMF, 80 °C, 4 h (70%); (b) 3ꢀmethoxycyclobutanꢀ1ꢀamine.HCl, DIPEA, DMA, 85
2
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
°C, 3 h (77%); (c) NaOH, EtOH, water, 40 °C, 2 h (81%); (d) DPPA, TEA, DMF, 60 °C, 2 h, (77%); (e) MeI, TBAB, NaOH,
DCM, water, r.t., 24 h, (75%); (f) (6ꢀfluoropyridinꢀ3ꢀyl)boronic acid, Cs CO , Pd(PPh ) , dioxane:water, 90 °C, 3 h; (g)
2
3
3 4
amine, DMSO, 110 °C, 1ꢀ2 h (47ꢀ54%); (h) B Pin , (dppf)PdCl , KOAc, Dioxane , 80 °C, 2 h (76%); (i) aryl boronic acid,
2
2
2
Cs CO , Pd(PPh ) , dioxane:water, 80 °C, 2 h (17%); (j) (6ꢀ[[(tertꢀbutoxy)carbonyl]amino]pyridinꢀ3ꢀyl)boronic acid,
2
3
3 4
Cs CO , Pd(PPh ) , dioxane:water, 100 °C, 3 h (70%); (k) 4ꢀchlorobutanoyl chloride, DIPEA, DCM, r.t., 16 h (66%). (l)
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
3
3 4
TFA, DCM, r.t., 30 min (99%); (m) dimethylamine, NaI, THF, 60 °C, 16 h, (19%); (n) arylboronic ester, Cs CO , Pd(PPh ) ,
2
3
3 4
dioxane:water, 80 °C, 2ꢀ6 h (22ꢀ62%);.
Analogous transformations employing a Hoffman rearrangement were used to prepare compounds 37 –
0, 42, 43, 45 – 48, 50 and 51 from 6ꢀbromoꢀ4ꢀchloroquinolineꢀ3ꢀcarboxamide (35), scheme 3, or using
4
a Curtius rearrangement to prepare compounds 57 – 61, 63 and 64 from ethyl 6ꢀbromoꢀ4ꢀ
chloroquinolineꢀ3ꢀcarboxylate (52), scheme 4. NꢀOxide 65 was made by the biocatalytic oxidation of
6
4, scheme 4.
a
Scheme 3: Synthesis of 1Hꢀimidazo[4,5ꢀc]quinolinꢀ2(3H)ꢀones by Hoffmann rearrangement.
ACS Paragon Plus Environment
11

Page 13 of 54
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
g
3
4
36
37
e,f
a,b,c,d
h
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
i,j,k,l
3
5
m,n,o
40
38 R = Me
39 R = isopropyl
q
p
44
4
2 R = H
43 R = F
s
41
r
t,u
4
4
4
4
5 R = 3-(methylamino)propoxyl
6 R = 3-(azetidin-1-yl)propoxyl
7 R = 3-(pyrrolidin-1-yl)propoxyl
8 R = 3-(piperidin-1-yl)propoxyl
49
5
5
0 R = morpholino
1 R = 4-hydroxypiperidinyl
a
Reagents and conditions: (a) 1,3ꢀdiethyl 2ꢀ(ethoxymethylidene)propanedioate, EtOH, 78 °C, 24 h (90%); (b) Ph O, 240 °C,
2
1
1
h; (c) NaOH, EtOH, water, 75 °C, 1.5 h (96%); (d) SOCl , DMF, 70 °C, 2 h; NH OH (95%); (e) 3ꢀ(4,4,5,5ꢀtetramethylꢀ
2
4
,3,2ꢀdioxaborolanꢀ2ꢀyl)pyridine, Cs CO , Pd(PPh ) , dioxane:water, 100 °C, 5 h (72%); (f) tetrahydroꢀ2Hꢀpyranꢀ4ꢀ
2
3
3 4
amine.HCl, DIPEA, DMF, 80 °C, 16 h (19%); (g) trichloroisocyanuric acid, DBU, MeOH, r.t., 30 min. (87%); (h)
DMFDMA, DMF, 80 °C. 1.5 h (43%) or iodopropane, NaH, DMF, 55°C, 30 min, (9%); (i) oxetanꢀ3ꢀamine, DIPEA, DMA,
100 °C, 18 h (100%); (j) trichloroisocyanuric acid, DBU, MeOH, r.t., 24 h. (19%); (k) MeI, TBAB, NaOH, DCM, water, r.t.,
3
6 h, (92%); (l) boronic ester, dichloro [1,1’ꢀ bis(diꢀtertbutylphosphino)ferrocene] palladium(II), K CO , Dioxane, 80 °C, 1 h
2 3
(
58%); (m) 3ꢀmethoxycyclobutanꢀ1ꢀamine.HCl, DIPEA, DMA, 100°C, 16 h (68%); (n) trichloroisocyanuric acid, DBU,
MeOH, r.t., 18 h. (99%); (o) MeI, TBAB, NaOH, DCM, water, r.t., 12 h, (75%); (p) arylboronate, Cs CO , Pd(PPh ) ,
2
3
3 4
dioxane:water, 80ꢀ90 °C, 2ꢀ3 h (36ꢀ75%); (q) (6ꢀfluoropyridinꢀ3ꢀyl)boronic acid , K CO , Pd(PPh ) , dioxane:water, 90 °C, 4
2
3
3 4
ACS Paragon Plus Environment
12

Journal of Medicinal Chemistry
Page 14 of 54
h (86%); (r) alcohol, NaH, DMF or DMA, r.t., 10 min ꢀ 24 h (25ꢀ75%) (for 45 an additional step: TFA, DCM, r.t. 1 h, was
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
required); (s) (6ꢀ(3ꢀbromopropoxy)pyridinꢀ3ꢀyl)boronic ester, Cs CO , Pd(PPh ) , dioxane:water, 90 °C, 1 h (27%); (t) MsCl,
2
3
3 4
DIPEA, DMF, r.t., 30 min. (u) morpholine, DIPEA r.t, (9% over 2 steps).
a
Scheme 4: Synthesis of 1Hꢀimidazo[4,5ꢀc]quinolinꢀ2(3H)ꢀones by Curtius rearrangement.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
O
O
O
O
N
O
O
F
N
O
N
N
Br
g
O
N
N
N
N
H
N
N
57
56
a
f
O
R
O
R
N
O
Cl
N
O
N
N
h
N
Br
b,c,d,e
N
N
Br
O
N
N
52
53 R = (3S)-tetrahydropyran-3-yl
54 R = (3R)-tetrahydropyran-3-yl
55 R = cis-3-hydroxycyclobutyl
58 R = (3S)-tetrahydropyran-3-yl
59 R = (3R)-tetrahydropyran-3-yl
60 R = cis-3-hydroxycyclobutyl
i,j,k,l,m
n,o,p,q,r
O
O
O
O
N
O
O
F
N
s
R
O
O
N
N
N
N
N
N
N
N
N
N
6
6
3 R = piperidyl
4 R = Dimethylamino
62
61
t
6
5 R = Dimethylamine-oxide
a
Reagents and conditions: (a) SOCl , DMF, 75 °C, 16 h (83%); (b) amine, DIPEA, DMA, 80ꢀ90 °C, 1.5ꢀ16 h (83ꢀ97%); (c)
2
NaOH, THF, MeOH, Water, 60 °C, 1.5ꢀ3 h (95ꢀ100%); (d) DPPA, TEA, DMF, 60 °C, 1ꢀ16 h, (80ꢀ98%); (e) MeI, TBAB,
NaOH, DCM, water, r.t., 1ꢀ12h or DMFDMA, DMF, 80°C, 3h (75ꢀ96%); (f) (6ꢀfluoropyridinꢀ3ꢀyl)boronic acid, Cs CO ,
2
3
Pd(PPh ) , dioxane:water, 100 °C, 3 h (87%); (g) 3ꢀ(piperidinꢀ1ꢀyl)propanꢀ1ꢀol, NaH, THF, r.t., 24 h (28%); (h) aryl boronic
3
4
ester, Cs CO , Xꢀphosꢀ2Gen or Pd(PPh ) , dioxane:water, 90ꢀ100 °C, 2ꢀ3 h (41ꢀ62%); (i) tetrahydroꢀ2Hꢀpyranꢀ4ꢀamine,
2
3
3 4
DIPEA, DMF, 60 °C, 16h (100%);(j) 2ꢀ(methoxymethyl)ꢀ5 ꢀboronic esterꢀpyridine Cs CO , Pd(PPh ) , dioxane:water, 100
2
3
3 4
°C, 2 h (95%) ; (k) NaOH, THF, Water, 60 °C, 4 h (92%); (l) DPPA, TEA, DMF, 60 °C, 7 h, (79%); (m) MeI, TBAB,
ACS Paragon Plus Environment
13

Page 15 of 54
Journal of Medicinal Chemistry
NaOH, DCM, water, r.t., 72 h (62%); (n) tetrahydroꢀ2Hꢀpyranꢀ4ꢀamine, DIPEA, DMF, 60 °C, 16h (100%); (o) NaOH,
MeOH, Water, 70 °C, 2 h (86%); (p) DPPA, TEA, DMF, 60 °C, 2 h, (99%); (q) MeI, TBAB, NaOH, DCM, water, r.t., 16 h
(93%); (r) (6ꢀfluoropyridinꢀ3ꢀyl)boronic acid, Cl Na Pd, DTBPPS, K CO , dioxane:water, 100 °C, 16 h (96%); (s) alcohol,
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
4
2
2
3
NaH, DMA or DMF, r.t. to 50 °C, 2ꢀ16 h, (82ꢀ90%); (t) NADP, K PO , Codexis KRED, Codexis BVMO, 2ꢀpropanol, water,
2
4
r.t., 41 h (39%).
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Imidazo[4,5ꢀc][1,5]naphthyridinꢀ2ꢀone 70 was synthesized in 8 steps from 6ꢀchloroꢀ3ꢀaminopyridine
7. Enamine formation followed by cyclisation and hydrolysis gave bicycle 68. Chlorination and amide
6
formation followed by S Ar and Hoffman rearrangement led to intermediate 69 which was converted to
N
70 by methylation and finally Suzuki crossꢀcoupling with the appropriate boronate, scheme 5.
a
Scheme 5: Synthesis of 70.
a,b,c
d,e,f
g,h
68
69
70
67
a
Reagents and conditions: (a) 1,3ꢀdiethyl 2ꢀ(ethoxymethylidene)propanedioate, EtOH, 80 °C, 4 h (90%); (b) Ph O, 220 °C, 2
2
h (37%); (c) NaOH, EtOH, water, 100 °C, 0.5 h (92%); (d) SOCl , DMF, 80 °C, 2 h; NH OH (100%); (e) tetrahydroꢀ2Hꢀ
2
4
.
pyranꢀ4ꢀamine HCl, DIPEA, DMA, r.t., 18 h (37%); (f) trichloroisocyanuric acid, DBU, MeOH, r.t., 30min. (75%); (g) MeI,
TBAB, NaOH, DCM, water, r.t., 12 h (75%); (h) boronic ester, Cs CO , Pd(PPh ) , dioxane:water, 90 °C, 2 h (28%).
2
3
3 4
Imidazo[4,5ꢀc]cinnolinꢀ2ꢀone 74 was made in 10 steps from 4ꢀbromophenylhydrazine 71. Initial
reaction with diethyloxomalonate was followed by hydrolysis of both esters and acid chloride
formation. Titanium chloride induced cyclisation afforded cinnoline 72. Chlorination, S Ar, ester
N
hydrolysis and Curtius rearrangement led to the formation of tricylcic compound 73 and methylation
followed by a Suzuki crossꢀcoupling produced 74, scheme 6.
ACS Paragon Plus Environment
14

Journal of Medicinal Chemistry
Page 16 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
a
Scheme 6: Synthesis of 74
a,b,c,d
e,f,g,h
i,j
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
72
74
71
73
a
Reagents and conditions: (a) Diethyloxomalonate, aq EtOH, r.t., 18 h (92%); (b) NaOH, EtOH, 78 °C, 50 min. (81%); (c)
SOCl , 44°C, 2 h, (97%); (d) TiCl , nitrobenzene, 95 °C, 18 h then EtOH, H SO , 95 °C, 5 h (34%); (e) SOCl , 75°C, 3 h,
2
4
2
4
2
(
100%); (f) tetrahydroꢀ2Hꢀpyranꢀ4ꢀamine.HCl, TEA, THF, r.t., 30 min. (98%); (g) NaOH, MeOH, water, r.t., 1 h; (h) DPPA,
TEA, DMF, 60 °C, 3 h (60% over 2 steps); (i) DMFDMA, DMF, 80 °C, 4 h (91%); (j) Boronic ester, Cs CO , Pd(PPh ) ,
2
3
3 4
dioxane:water, 80 °C, 2 h (8.5%).
RESULTS AND DISCUSSION
A number of strategies for reducing the predicted clinically efficacious dose can be conceived;
however, to identify the optimal strategy it is important to consider the extent to which different
parameters contribute to the dose prediction. For example, increasing potency (whilst keeping all other
parameters constant) will reduce the predicted dose in a linear fashion; however, increases in the
pharmacokinetic halfꢀlife can have a much more profound influence (especially where the predicted
halfꢀlife is significantly shorter than the desired duration of target engagement). Given the relatively
short predicted human halfꢀlife of 7, especially in the context of achieving target engagement for a 24
hour period, we embarked on a strategy to increase halfꢀlife whilst maintaining or improving potency,
selectivity and physicochemical properties. The use of PBPK modelling approaches to accurately
predict human pharmacokinetics requires detailed in vitro and in vivo compound profiling and as such
places a limit on the number of compounds for which such models can be built. However, a reasonable
first estimate of human pharmacokinetic parameters can be produced based on a more limited set of in
vitro parameters (i.e. plasma protein binding and the intrinsic clearance in human hepatocytes)
ACS Paragon Plus Environment
15

Page 17 of 54
Journal of Medicinal Chemistry
combined with a prediction of the volume of distribution (V ). When combined with a knowledge of
ss
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
target potency and the required extent and duration of target engagement an estimate of clinical dose can
be made. This simplified prediction, termed early dose to man prediction (eD2M) within AstraZeneca,
assumes complete absorption of a compound across the gut (i.e. Fabs = 100%) which would only be
45
expected to hold true for highly permeable compounds. However, if available, V and F data
ss
abs
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
determined experimentally in rodents can be incorporated to refine the prediction. The use of eD2M and
Dabs predictions enables a comparison of compound quality across a large number of compounds and
was used to guide the compound optimization described herein.
Pharmacokinetic halfꢀlife is intrinsically linked to both the clearance (CL) and the V of a
ss
compound and both parameters were considered for optimization. Compound 7 displays low metabolic
turnover in human hepatocyte incubations suggesting that a further reduction in CL would be a
challenge; therefore, a strategy to increase V was considered more profitable. Many studies have been
ss
reported attempting to identify those molecular properties that influence V with ionization state often
ss
4
6
found to dominate. In particular, basic compounds generally show increased V when compared with
ss
neutral or acidic compounds thus suggesting that the incorporation of a basic substituent into our novel
series of ATM inhibitors, whilst keeping other parameters constant, would deliver an increased
predicted human halfꢀlife and reduced predicted clinical dose.
Figure 2: Structure of basic quinoline carboxamide 16
16
ACS Paragon Plus Environment
16

Journal of Medicinal Chemistry
Page 18 of 54
Wide ranging SAR exploration in the 6ꢀposition of the quinoline carboxamide scaffold revealed
the potential to incorporate a basic substituent, as exemplified by 16, figure 2. Compound 16 showed an
unexpected increase in ATM cell potency (IC = 0.0086 ꢁM) whilst selectivity over ATR was retained
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
5
0
(
IC >30 ꢁM) and the basic nature of the compound resulted in an increased predicted human V when
50 ss
compared to neutral compound 7 (3.5 L/kg c.f. 1.2 L/kg). The improved primary potency and increased
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
V of 16 culminated in a low predicted clinical efficacious dose (eD2M = 11 mg QD) thereby
ss
highlighting how optimization of primary potency and pharmacokinetic properties can influence the
predicted dose. Unfortunately, despite this low dose prediction the permeability of the 16 was poor and
ꢀ6
a significant efflux liability was observed in MDCK cells (Papp AꢀB = 0.8 x 10 cm/s, efflux ratio = 28)
suggesting that this compound is unlikely to have good oral bioavailability or Fabs and would, therefore,
be unsuitable for further development. It should be noted that the anticipated poor Fabs of 16 will
adversely affect the accuracy of the eD2M prediction. The poor permeability observed with a lipophilic
compound such as 16 (log D7.4 = 2.8) emphasizes the impact that ionization and hydrogen bonding
groups (in particular donating groups) can have on permeability. Previously reported SAR studies have
established the importance of the 4ꢀamino and 3ꢀcarboxamide motifs within this scaffold culminating in
the hypothesis that an internal hydrogen bond is formed between these motifs to organize the molecule
3
0
into a bioactive conformation. As a consequence, all attempts to reduce hydrogen bond donor count
within this scaffold have been unsuccessful and it was, therefore, considered improbable that a
permeable, basic compound could be delivered without increasing lipophilicity to an unacceptable level.
Accordingly, it was appreciated that an inherently more permeable series of selective ATM inhibitors
would be required to allow the continuation of our strategy to increase both primary potency and
pharmacokinetic halfꢀlife through the introduction of basic substituents.
Numerous examples of chemical scaffold hopping exist in which a cyclic motif within a
molecule is replaced with a “pseudo” ring system (constrained by an intramolecular hydrogen bond)
47
leading to compounds with similar conformations and biological activities. The presence of a
ACS Paragon Plus Environment
17

Page 19 of 54
Journal of Medicinal Chemistry
“
pseudo” cyclic system within the quinoline carboxamide scaffold prompted the consideration of
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
suitable cyclic systems which could replace this “pseudo” ring and retain the ability to correctly position
any appended substituents. Such a strategy, if successful, was anticipated to maintain ATM potency and
selectivity whilst reducing the number of hydrogen bond donors and increasing permeability. A number
of possible tricyclic systems with the potential to fulfil the desired criteria were conceived but the
imidazo[5,4ꢀc]quinolinꢀ2ꢀone scaffold was prioritized based on inꢀhouse data on novel, proprietry
AstraZeneca compounds that was generated during routine cross project selectivity screening. This data
highlighted the potential for compounds based on this scaffold to display potent activity against a range
of PIKK family members. It should be noted that a similar scaffold is also present in a dual PI3K /
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
mTOR inhibitor, NVPꢀBEZ235, which has been reported as undergoing clinical evaluation; NVPꢀ
48ꢀ50
BEZ235 is reported to have activity against ATM.
Although this activity for NVPꢀBEZ235 was not
appreciated at the time of the work described it further supports the proprietry AstrsaZeneca data
supporting the potential of imidazo[5,4ꢀc]quinolinꢀ2ꢀone containing compounds to deliver potent ATM
inhibitors whilst also further highlighting the challenge to achieve PIKK family selectivity with such a
scaffold. A comparison of 38 with 36 reveals that ATM potency is broadly maintained following the
scaffold hop confirming that both cores can correctly position substituents in a bioactive conformation,
table 2. However, selectivity against closely related kinases appeared reduced with 38 being more potent
against ATR than against ATM. Permeability is increased, and efflux decreased, for 38 when compared
to 36 despite the compounds possessing similar levels of lipophilicity supporting the importance of
restricting hydrogen bond donors when optimizing permeability. Compound 38 retained good stability
in hepatocyte incubations although aqueous solubility was reduced. Further investigation of the
imidazo[5,4ꢀc]quinolinꢀ2ꢀone demonstrated that the removal of a methyl group to deliver 37 was
broadly tolerated, although the addition of a hydrogen bond donor may be expected to reduce
permeability. Replacing the methyl with a larger substituent, 39, resulted in the loss of activity against
both ATM and ATR suggesting that there is little opportunity to increase primary potency or selectivity
in this region of the molecule.
ACS Paragon Plus Environment
18

Journal of Medicinal Chemistry
Page 20 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
a
Table 2: Structures and profiles of 36 – 39.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
36
38 R = Me
3
7 R = H
i
3
9 R = Pr
36
38
0.36
37
39
ATM Cell IC50
ATR Cell IC50
0.95
3.1
1.6
>30
0.087
2.2
0.051
>27.6
LogD7.4
1.9
2.4
12
ꢀ
2.9
41
ꢀ
Aqueous Solubility
Human / Rat Heps
hERG IC50
>610
<4 / 8.5
2.1
17
<4.9 / 9.4
11
ꢀ
ꢀ
b
MDCK Papp AꢀB / efflux ratio
4.8 / 10
15 / 4.5
30 / 1.5
60 / 0.6
ꢀ
ꢀ
b
Caco2 Papp AꢀB / efflux ratio
ꢀ
ꢀ
a
b
Data reported as in table 1.
ꢀ6
Papp AꢀB data are expressed in 10 cm/s and the efflux ratio is expressed as P BꢀA / Papp AꢀB
app
In order to better understand how these compounds bind to ATM an in silico model of the
protein was built using the MOE/2014 software (Chemical Computing Group) and the structure of
related protein kinase mTOR as a template (PDB code = 4JSX). Alignment was performed based on
sequence similarity and conserved features of protein kinases. Ten candidate homology models of ATM
were built, retaining the original mTOR ligand to prevent collapse of the binding site and the best model
in terms of protein structure quality criteria was selected and further refined. Standard procedures as
outlined in the MOE documentation were employed throughout. Example ligands were docked into the
binding site using the default docking algorithm in MOE and twenty poses were kept for each with the
best pose selected manually, based on SAR knowledge. When modelled into the ATPꢀbinding site of
ACS Paragon Plus Environment
19

Page 21 of 54
Journal of Medicinal Chemistry
ATM both 36 and 38 showed highly similar binding modes in which the molecules were anchored
through a hydrogenꢀbond from the quinoline nitrogen to Cys2770 in the hinge region of the kinase,
figure 3. A second hydrogenꢀbond was formed from the pyridine ring to the conserved catalytic lysine
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
(
Lys2717), while the same ring bypasses the gatekeeper residue (Leu2767) to take part in an edgeꢀface
πꢀ interaction with Tyr2755, within the back pocket. In order to make these interactions the pyridine
ring is twisted ~35° out of the plane of the quinoline ring. These predicted interactions are consistent
with the observed activities. The postulated internal hydrogen bond was predicted for 36 and is again
consistent with the observed SAR. The tetrahydropyran group was not predicted to make any specific
interaction and was predicted to partly occupy the ribose binding pocket
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 3a: In silico model of ATM with 36 bound into the ATP binding site.
Figure 3b: In silico model of ATM with 38 bound into the ATP binding site.
ACS Paragon Plus Environment
20

Journal of Medicinal Chemistry
Page 22 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
To further probe the hypothesized binding modes the methoxymethylpyridine motif, present in
both 6 and 7 and known to increase selectivity for ATM over ATR in the quinoline carboxamide
scaffold, was incorporated into the imidazo[5,4ꢀc]quinolinꢀ2ꢀone scaffold to produce 61, table 3.
Encouragingly, 61 retained activity against ATM whilst activity against ATR was dramatically reduced
thereby restoring the desired selectivity profile. Compound 61 was highly permeable and stable in
human hepatocyte incubations although aqueous solubility was low. The incorporation of a fluorine in
the 7ꢀposition of the imidazo[5,4ꢀc]quinolinꢀ2ꢀone core, 18, was found to improve solubility and reduce
metabolic turnover in rat hepatocyte incubations, albeit ATM potency was slightly diminished. Chloroꢀ
and methoxyꢀsubstituents were also tolerated in this position, albeit with a 3ꢀ10 fold reduction in ATM
potency (data not shown). This SAR is analogous to that observed with the quinoline carboxamide
scaffold further supporting the modelled binding modes. Based on the quinoline carboxamide SAR,
larger substituents in this position would not be expected to have good affinity for ATM and hence no
further investigation was performed. Interestingly, the impact of the fluorine on hERG affinity appears
ACS Paragon Plus Environment
21

Page 23 of 54
Journal of Medicinal Chemistry
less marked in the imidazo[5,4ꢀc]quinolinꢀ2ꢀone scaffold. Divergent SAR between the two series was
observed when replacing the tetrahydropyran motif where the introduction of a previously optimized
chiral amine to the imidazo[5,4ꢀc]quinolinꢀ2ꢀone scaffold, 20, failed to deliver the anticipated increase
in potency. This observation suggests that the positioning of this substituent differs slightly between the
two scaffolds which is also consistent with the proposed binding modes. More thorough evaluation of
the SAR in this region uncovered a number of small cyclic alkyl groups with improved ATM potency,
good selectivity and high levels of permeability, for example 12 – 15, table 3. The addition of a
hydrogen bond donor in this region did result in a reduction in permeability, 15, albeit the overall profile
remained encouraging. Pharmacokinetic evaluation of 13 in rat revealed the compound to have good
bioavailability (>100%) and moderate CL (22.5 mL/min/kg). However, as anticipated for a neutral
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
compound, V was relatively low (0.98 L/kg) leading to a modest halfꢀlife (2.7 h). The properties of 13
ss
resulted in a large predicted clinically efficacious dose (eD2M >10000 mg QD) and thus highlighted
that although the imidazo[5,4ꢀc]quinolinꢀ2ꢀone scaffold had the potential to deliver selective ATM
inhibitors, significant optimization of both primary potency and predicted human pharmacokinetics was
required.
a
Table 3: SAR exploration of neutral imidazo[5,4ꢀc]quinolinꢀ2ꢀones.
ATM
Cell IC50 Cell IC50
ATR
human /
rat heps
hERG
IC50
Papp AꢀB /
Efflux ratio
Entry
R1
R2
H
LogD7.4 Aq Sol
6
1
1
8
0.37
0.45
>13
>20
2.3
2.4
19
99
<3 / 63
<4 / 7.5
>33
27
32 / 1.4
43 / 0.8
F
2
0
F
0.49
>30
2.7
>1000
22 / 1.6
15
27 / 0.8
ACS Paragon Plus Environment
22

Journal of Medicinal Chemistry
Page 24 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
12
F
F
F
F
0.24
0.10
0.16
0.14
>26
>30
>28
>30
3.2
2.7
3.0
2.0
84
37
10 / 10
8.8 / 4.5
8.4 / 4.3
2.7 / 6.0
27
9.0
8.9
15
20 / 0.6
38 / 0.4
28 / 0.6
14 / 2.6
1
1
3
4
49
15
270
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
a
Data reported as in table 1 and table 2.
With an understanding of how the two series bind we were able to use our detailed knowledge of
the quinoline carboxamide SAR to inform the design of compounds 29, 30 and 24, table 4. Consistent
with the quinoline carboxamide SAR, the methoxy motif could be replaced by hydroxyl (29) or
methylsulfonyl (30) leading to reduced lipophilicity with only limited impact on ATM potency;
however, permeability was compromised to some extent and solubility remained low. The introduction
of a basic substituent (24) resulted in a significant increase in ATM potency and solubility and whilst
permeability was compromised the profile was considered superior to that for basic quinoline
carboxamide compound 16. The concept of Lipophilic Ligand Efficiency (LLE) has been widely
reported in the literature and is routinely used within the Medicinal Chemistry community as a measure
5
1,52
of compound quality.
When comparing the LLE (ATM Cell pIC – logD ) of 13 and 24 a clear
50 7.4
improvement was observed (LLE = 4.3 and 6.5 respectively) indicating the advantageous nature of the
basic side chain. This advantage is more apparent when comparing the predicted clinically efficacious
clinical dose (24: eD2M = 24 mg QD) and was further validation for our focus on primary potency and
predicted human pharmacokinetics. The low predicted clinically efficacious dose for 24 prompted the
more thorough exploration of basic substituents with a focus on improving permeability whilst
maintaining or increasing ATM potency. Removal of the hydrogen bond donor by methylation of the
nitrogen, 25, did increase permeability although a reduction in potency was observed, table 4. Replacing
the nitrogen with oxygen, 31, was also found to increase permeability and the impact on ATM potency
was much diminished resulting in a more attractive balance of properties. Amidic linker groups were
ACS Paragon Plus Environment
23

Page 25 of 54
Journal of Medicinal Chemistry
also investigated, 27 and 28, but no improvement in permeability was observed. The high potency of 28
implied that longer basic chains could be accommodated and it was quickly established that a three
carbon chain, linked to the pyridine through an oxygen atom (32), was optimal. Compound 32 was a
highly potent and selective ATM inhibitor with excellent solubility, reasonable hepatic stability and no
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
activity against hERG. The increased basicity of 32 compared to 31 (pK = 9.3 and 8.5, respectively) led
a
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
to a reduction in LogD and whilst a slight reduction in permeability was observed the profile remained
7.4
highly attractive. The dramatic improvement in ATM potency for 32 compared to neutral compound 13,
alongside the increased predicted V associated with a basic group (predicted human V = 2.7 L/kg),
ss
ss
culminated in a >1000ꢀfold reduction in predicted clinically efficacious dose (32: eD2M = 16 mg QD).
a
Table 4: Structure and profile of compounds 24, 25, 27, 28, 31, and 32.
ATM
Cell IC50 Cell IC50
ATR
human /
rat heps
hERG
IC50
Papp AꢀB /
Efflux ratio
Entry
R
LogD7.4 Aq Sol
2
3
9
0
0.23
0.16
0.006
0.23
0.13
15
2.0
1.3
1.7
2.2
1.5
23
18
ꢀ / 5.0
ꢀ / 2.4
23
14
29 / 1.1
14 / 2.4
3.4 / 11
6.3 / 3.5
2.5 / 11
S
O O
>16
>30
>30
>30
24
>1000
>1000
980
2.8 / 9.6
5.1 / 18
4.0 / 2.9
24
2
2
5
7
5.7
>33
ACS Paragon Plus Environment
24

Journal of Medicinal Chemistry
Page 26 of 54
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
2
8
0.0082
0.023
>30
>30
>30
1.1
2.0
1.8
870
>1000
>980
6.5 / 7.0
8.3 / 12
7.9 / 6.6
18
0.4 / 77
16 / 1.6
6.8 / 6.2
31
>33
>33
3
2
0.0011
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
a
Data reported as in table 1 and table 2.
The positive impact of the basic substituent on both aqueous solubility and hERG affinity
prompted the reꢀevaluation of the 7ꢀfluoro substituent. Removal of the fluorine, 42, delivered a further
increase in ATM potency with little affect on other properties, table 5. Interestingly, incorporation of the
fluorine substituent on the pyridyl ring, 43, resulted ATM potency similar to 32, whilst the inclusion of
two fluorine atoms, 33, gave a minor reduction in potency. This pattern of activity was seen across a
range of fluorinated and nonꢀfluorinated matched sets (data not shown) and a possible explanation for
this could involve the preferred conformation of the biaryl bond linking the pyridyl ring to the
imidazo[5,4ꢀc]quinolinꢀ2ꢀone core. The proposed binding mode contains a dihedral angle of ~35°
allowing a πꢀinteraction between a ring hydrogen and Tyr2755 of the protein. Quantum mechanics
calculations (DFT, B3LYP, 6ꢀ31G**) of this biaryl system confirm the binding mode prediction that a
nonꢀplanar conformation is preferred and the variation in conformational strain as a function of dihedral
angle is shown in figure 4. When no fluorine atoms are present (e.g. 42) the energy minimum sits at 36°,
while a fluorine at the 7ꢀposition of the core (e.g. 32) increases the dihedral of the lowest energy
conformation to 41°. The energy penalty incurred for fluorinated analogues to adopt the preferred
binding conformation is consistent with the subtle reduction in potency observed. This hypothesis is
also consistent with the observed SAR for the introduction of a chloroꢀ or methoxyꢀsubstituent in this
region.
ACS Paragon Plus Environment
25

Page 27 of 54
Journal of Medicinal Chemistry
a
Table 5: Structure and profile of compounds 42, 43 and 33.
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
4
2
43
33
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ATM
ATR
human /
rat heps
hERG
IC50
Papp AꢀB /
Efflux ratio
Entry
LogD7.4 Aq Sol
Cell IC50 Cell IC50
4
4
3
2
3
3
0.00025
0.00076
0.0026
>30
>30
>30
1.6
2.0
2.1
540
8.7 / 5.1
>33
>33
22
7.8 / 3.0
4.2 / 3.2
10 / 2.5
>1100
>1000
15 / 9.3
14 / 9.0
a
Data reported as in table 1 and table 2.
Figure 4: Relative energy for the conformational strain of a biaryl system as a function of dihedral with
different substitution patterns, from quantum mechanics (DFT, B3LYP, 6ꢀ31G**).
X = H, Y = H
X = F, Y = H
X = F, Y = F
20
1
5
0
5
0
1
0
30
60
90
120
150
180
Dihedral Angle (degrees)
Pharmacokinetic evaluation of 42 in both rat and dog revealed moderate clearance (19.5 and 27.6
mL/min/kg, respectively) but the relatively high V (5.7 and 11 L/kg, respectively), driven by the basic
ss
ACS Paragon Plus Environment
26

Journal of Medicinal Chemistry
Page 28 of 54
nature of the compound (pK = 9.5), resulted in a halfꢀlife of approximately 5 hours in both species. The
a
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
permeable nature of 42 led to good bioavailability in both rat and dog (50% and 65%, respectively).
PBPK modelling suggested that 42 would have a moderate to high clearance in humans (~13
mL/min/kg) but that a high V (>10 L/kg) would produce a long halfꢀlife (~15 hours). The predicted
ss
clinically efficacious dose was estimated to be 3 mg QD with a free Cmax in the region of 0.001 ꢁM. GIꢀ
Sim modelling predicted the Dabs for 42 would to be 3500 mg, thereby suggesting that compound 42
could be well suited for clinical development. However, the moderate to high predicted human
clearance of 42 was considered to consitute a risk for lower than desired bioavailability in man, due to
high first pass metabolism, especially when you consider the errors associated with predicted values.
Therefore, despite the highly promising nature of 42 we believed there remained an opportunity for
further compound optimization and in particular we sought to investigate the potential impact of
cytochrome P450 mediated oxidative demethylation of the basic center. The methylamino analogue 45,
a potential product of such oxidative metabolism, was found to retain excellent potency and selectivity
for ATM, table 6. As a consequence of the removal of a lipophilic methyl group, combined with the
increased basicity of the compound (pK = 10.3), the LogD is dramatically reduced and this, in
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
a
7.4
conjunction with an increase in Hꢀbond donnor count, results in compromised permeability. The
incorporation of cyclic bases, 46 – 48, was well tolerated and moderate improvements in metabolic
stability were observed. However, detailed in vitro metabolite ID studies for 42 highlighted that
oxidative dealkylation of the base was not the major metabolic pathway. Similar studies for 45
suggested that the cyclic base has the potential to undergo oxidative ring opening giving rise to
potentially reactive species. Surprisingly, the morpholine containing and hydroxypiperidine containing
compounds, 50 and 51, have lost much of their affinity for ATM. It is currently unclear whether this
was driven by a reduction in basicity (pK = 6.7 and 8.4, respectively) or the inability of the protein to
a
accommodate hydrophilic atoms in that region.
ACS Paragon Plus Environment
27

Page 29 of 54
Journal of Medicinal Chemistry
a
Table 6: Structure and profile of compounds 45 – 48, 51 and 52.
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ATM
Cell IC50 Cell IC50
ATR
human /
rat heps
hERG
IC50
Papp AꢀB /
Efflux ratio
Entry
R
LogD7.4
Aq Sol
4
4
4
5
6
7
0.00025
0.00029
0.00017
0.00020
>28
4.4
>24
>24
>2.6
>30
>30
0.4
930
4.4 / 4.8
3.3 / 2.5
5.3 / <1.3
3.0 / 4.0
ꢀ / 28
>33
>33
>33
14
ꢀ
0.65 / 41
3.5 / 7.8
2.5 / 9.4
5.4 / 1.9
ꢀ
1.4
1.8
2.2
1.4
0.6
770
>1100
>1000
980
48
5
5
0
1
27
750
ꢀ / <1
ꢀ
ꢀ
a
Data reported as in table 1 and table 2.
To further enhance our understanding of the SAR we were keen to establish whether
observations in different regions were independent of each other (i.e. additive SAR). Such knowledge
was anticipated to help inform future compound design. A FreeꢀWilson type regression model, in which
specific values are attached to individual structural motifs, was built using available data for the ATM
cell pIC50 and found to predict well (RMSE <0.3) suggesting that the SAR is broadly additive in
5
3
nature. The use of such models has been employed by a number of groups to efficiency explore
chemical space by helping to identify the combination of substituents most likely to result in a desired
5
4ꢀ57
profile.
A similar strategy was adopted in which virtual compounds containing all possible
combinations of tolerated substituents were prioritized for synthesis based on the likelihood of
ACS Paragon Plus Environment
28