Drug-like property optimization: Discovery of orally bioavailable quinazoline-based multi-targeted kinase inhibitors

Article abstract of DOI:10.1016/j.bioorg.2020.103689

In an effort to develop new cancer therapeutics, we have reported clinical candidate BPR1K871 (1) as a potent anticancer compound in MOLM-13 and MV4-11 leukemia models, as well as in colorectal and pancreatic animal models. As BPR1K871 lacks oral bioavailability, we continued searching for orally bioavailable analogs through drug-like property optimization. We optimized both the physicochemical properties (PCP) as well as in vitro rat liver microsomal stability of 1, with concomitant monitoring of aurora kinase enzyme inhibition as well as cellular anti-proliferative activity in HCT-116 cell line. Structural modification at the 6- and 7-position of quinazoline core of 1 led to the identification of 34 as an orally bioavailable (F% = 54) multi-kinase inhibitor, which exhibits potent anti-proliferative activity against various cancer cell lines. Quinazoline 34 is selected as a promising oral lead candidate for further preclinical evaluation.

Full text of DOI:10.1016/j.bioorg.2020.103689

Journal Pre-proofs
Drug-Like Property Optimization: Discovery of Orally Bioavailable Quinazo-
line-Based Multi-Targeted Kinase Inhibitors
Shu-Yu Lin, Chun-Feng Chang, Mohane Selvaraj Coumar, Pei-Yi Chen, Fu-
Ming Kuo, Chun-Hwa Chen, Mu-Chun Li, Wen-Hsing Lin, Po-Chu Kuo,
Sing-Yi Wang, An-Siou Li, Chin-Yu Lin, Chen-Ming Yang, Teng-Kuang
Yeh, Jen-Shin Song, John T.A. Hsu, Hsing-Pang Hsieh
PII:
S0045-2068(20)30031-6
DOI:
https://doi.org/10.1016/j.bioorg.2020.103689
YBIOO 103689
Reference:
To appear in:
Bioorganic Chemistry
Received Date:
Revised Date:
Accepted Date:
6 January 2020
19 February 2020
20 February 2020
Please cite this article as: S-Y. Lin, C-F. Chang, M. Selvaraj Coumar, P-Y. Chen, F-M. Kuo, C-H. Chen, M-C.
Li, W-H. Lin, P-C. Kuo, S-Y. Wang, A-S. Li, C-Y. Lin, C-M. Yang, T-K. Yeh, J-S. Song, J.T.A. Hsu, H-P.
Hsieh, Drug-Like Property Optimization: Discovery of Orally Bioavailable Quinazoline-Based Multi-Targeted
Kinase Inhibitors, Bioorganic Chemistry (2020), doi: https://doi.org/10.1016/j.bioorg.2020.103689
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover
page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version
will undergo additional copyediting, typesetting and review before it is published in its final form, but we are
providing this version to give early visibility of the article. Please note that, during the production process, errors
may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
©
2020 Published by Elsevier Inc.

Drug-Like Property Optimization: Discovery of Orally Bioavailable
Quinazoline-Based Multi-Targeted Kinase Inhibitors
a,#
a,#
d,#
a
Shu-Yu Lin , Chun-Feng Chang , Mohane Selvaraj Coumar , Pei-Yi Chen , Fu-
a
a
a
a
a
Ming Kuo , Chun-Hwa Chen , Mu-Chun Li , Wen-Hsing Lin , Po-Chu Kuo , Sing-Yi
a
a
a
a
a
Wang , An-Siou Li , Chin-Yu Lin , Chen-Ming Yang , Teng-Kuang Yeh , Jen-Shin
a
a
a,b,c*
Song , John T.-A. Hsu , Hsing-Pang Hsieh
aInstitute of Biotechnology and Pharmaceutical Research, National Health Research
Institutes, 35 Keyan Road, Zhunan, Miaoli County 35053, Taiwan, R.O.C.
bDepartment of Chemistry, National Tsing Hua University, Hsinchu, 30013, Taiwan
cFrontier Research Center on Fundamental and Applied Sciences of Matters, National
Tsing Hua University, Hsinchu, 30013, Taiwan
dCentre for Bioinformatics, School of Life Sciences, Pondicherry University, Kalapet,
Puducherry 605014, India.
#These authors contributed equally to this work.
*
To whom correspondence should be addressed: Prof. Dr. Hsing-Pang Hsieh, Phone,
+
886-37-246-166 ext. 35708; fax, +886-37-586-456; e-mail, hphsieh@nhri.org.tw
1

Abstract
In an effort to develop new cancer therapeutics, we have reported clinical candidate
BPR1K871 (1) as a potent anticancer compound in MOLM-13 and MV4-11 leukemia
models, as well as in colorectal and pancreatic animal models. As BPR1K871 lacks oral
bioavailability, we continued searching for orally bioavailable analogs through drug-
like property optimization. We optimized both the physicochemical properties (PCP)
as well as in vitro rat liver microsomal stability of 1, with concomitant monitoring of
aurora kinase enzyme inhibition as well as cellular anti-proliferative activity in HCT-
1
16 cell line. Structural modification at the 6- and 7-position of quinazoline core of 1
led to the identification of 34 as an orally bioavailable (F% = 54) multi-kinase inhibitor,
which exhibits potent anti-proliferative activity against various cancer cell lines.
Quinazoline 34 is selected as a promising oral lead candidate for further preclinical
evaluation.
Keywords: aurora kinase inhibitor, quinazoline, physicochemical property, metabolic
stability, oral bioavailability, anti-cancer
2

1
. Introduction
Cancer is a global disease with high mortality rate. Lung, breast, prostate and colorectal
cancer are the major cancer types reported world over [1]. Many pharmaceutical
companies and academic institutes are in pursuit of novel anti-cancer therapeutics that
could target aberrant proteins/pathways responsible for the development or progression
of cancer [2]. One of the important classes of drug targets for anti-cancer drug
development are protein kinases [3-5]. Protein kinases are enzymes that transfer
phosphate group from ATP to serine, threonine or tyrosine residues in proteins;
phosphorylation of proteins acts as a switch to turn-on or turn-off the proteins function,
thereby playing an important role in normal physiological signaling process. More than
5
00 protein kinases have been identified in the humans that are involved in a variety of
function and maintain homeostasis [5]. Aberration in many protein kinase signaling
process has been reported, either due to the overexpression or due to mutations in the
protein. These signaling aberrations results in alteration in normal cellular functions,
leading to a variety of diseases, including cancer [6, 7].
During the last two decades, researches have shown that the altered kinase responsible
for the cancer could be targeted by ATP competitive small molecule inhibitors. These
small molecules bind to the ATP binding site of the protein kinase and block the
aberrant downstream signaling, resulting in the inhibition of cancer progression [3, 6,
3

®
8
]. For example, in 2001 the first kinase inhibitor imatinib (Gleevec ) was approved by
U.S. Food and Drug Administration (FDA) for the treatment of chronic myeloid
leukemia (CML). Imatinib targets the ATP site of Abl kinase and prevents the
progression of the disease caused by the Bcr-Abl fusion protein [9, 10]. Currently, 52
small molecule inhibitors that target various kinase (Abl, EGFR, HER2, etc) have been
approved world over for the treatment of various cancers (CML, lung cancer, breast
cancer, etc) [5, 7, 8, 11]. The successful launch of several kinase inhibitors endorses
the strategy of blocking the aberrant kinase as a new avenue for targeted cancer therapy.
Aurora kinase belongs to serine/threonine kinase family and are normally expressed
during mitosis process. There are three distinct subtypes – Aurora A kinase , Aurora B
kinase and Aurora C kinase – with different function and location in the nucleus,
regulating the mitosis process [12]. Aurora A is located to the central of the
chromosome and to the pole of spindle which is involved in the process of centrosome
maturation and separation. Aurora B is a component of the chromosomal passenger
complex (CPC) that plays a critical role in the mitosis process; Aurora B is essential for
chromosomal condensation and cytokinesis. The third subtype Aurora C kinase has
high similarity and overlapped functions with Aurora B. Aurora A/B have been found
to be overexpressed or amplified in many malignant tumors, like glioma, breast, lung,
ovarian, colon and thyroid cancers. To date, several Aurora kinase inhibitors have been
4

reported to show anti-cancer efficacy in pre-clinical animal models and few are in
different stage of clinical trial as anti-cancer therapeutics [13–17] (Figure 1). These
observations led us to develop different series of inhibitors against Aurora kinase for
the treatment of cancers [18–20].
H
N
N
NH
N
N
H
CH3
N
N
N
NH
HN
N
N
N
HN
N
O
H
N
N
O
O
MeO
NH HN
O
S
O
N
N
S
N
HN
O
N
N
H2N
N
N
H3C
MK-0457 (VX-680)
Withdrawn from Phase II
PHA-739358 (Danusertib)
Phase II (Advanced solid
tumors and CML)
AMG-900
Phase I (Advanced solid
tumors and CML)
AT-9283
Phase II (Advanced solid
tumors, AML and CML)
(CML & AML)
NH
N
HN
N
N
CH3
OMe
N
O
COOH
OMe
N
N
NH
N
N
N
N
N
F
H
N
N
N
H
H3C
ENMD-2076
Phase II (Advanced solid
tumors, Hematological Malignancies
and multiple myeloma)
Cl
N
MLN-8237 (Alisertib)
Phase III (Advanced solid
tumors and Leukemias)
GSK1070916
Phase I (Advanced solid tumors)
F
NH
N
Cl
O
N
OH
N
O
OH
COOH
P
HN
NH
NH
HN
N
S
O
O
NH
N
O
N
S
N
F
N
O
N
Cl
N
O
N
H
BPR1K871 (1)
Approved for Phase I
Gastrointestinal cancers)
MK-5108 (VX-689)
Phase I (Advanced and/or
refractory solid tumors)
AZD-1152 (Barasertib)
Phase III (AML and
Advanced solid tumors)
(
Figure 1. Representative Aurora kinase inhibitors in clinical trials. BPR1K871 (1) is a
quinazoline based multi-kinase inhibitor with Aurora and FLT3 kinase inhibition
reported by us.
In our previous effort to develop Aurora kinase inhibitors, compound 2 was
identified as initial hit from a kinase-focused furanopyrimidine library as Aurora kinase
A inhibitor (Figure 2). The following introduction of a phenylurea group on the aniline
5

of 2 gave compound 3 as a potent hit, with 14.8-fold enhanced Aurora kinase inhibition
21]. In continuation works, we reported utilizing 3D-QSAR-assisted drug design and
[
a FLT3 homology model to develop two potential lead compounds, 4 with quinazoline
core and 5 with a thiazole containing urea chain [22, 23].
As the quinazoline core is considered as a privileged structure for the inhibition of
ATP-dependent kinases with lower lipophilicity, a series of lead-to-candidate
optimization led to the identification of BPR1K871 (1) as a multi-kinase inhibitor with
potent enzymatic (Aurora A IC = 22 nM; Aurora B IC = 13 nM; FLT3 IC = 19 nM)
5
0
50
50
and cellular activities in AML and multiple solid tumor cell lines. In addition, the
inhibitor 1 demonstrated in vivo efficacy in leukemia (MOLM-13 and MV4-11),
colorectal (COLO205) and pancreatic (Mia-PaCa2) xenograft mouse models upon
intravenous (IV) administration [18]. Currently, 1 has been approved as investigational
new drug (IND) by US FDA and TDFA for testing in humans for gastrointestinal
cancers or gastroenterology and hepatology related cancer treatment.
Notably, the majority of the medical indication of kinase inhibitors are approved
for the treatment of malignancies as oral oncologics, since oral prescription drugs offer
a number of advantages including easy storage, lower costs, patient preference and
convenience, etc. However the IND candidate 1 could not be developed as oral drug
due to poor oral bioavailability and need to be administered though IV route in clinics.
6

Due to this drawback, we undertook the optimization of 1 for the identification of orally
bioavailable compounds with anticancer potential in this work For this purpose, the
substitutions at three different possible metabolic sites in 1, the C-6 and C-7 position of
the quinazoline core and CH CH linker of thiazole urea side chain, were modified to
2
2
alter the drug-like properties while maintaining the Aurora kinase enzymatic as well as
cellular anticancer activities.
3D-QSAR-assisted drug design
ChemMedChem, 2013, 8, 136.
Introduction of
urea functional group
H
H
H
H
N
N
crucial
NH2
N
N
O
O
HN
N
HN
HN
O
O
essential for Auora A
increase 14.8-fold activity
N
N
4
N
Lead compound
O
N
2
O
N
Auora A IC50 = 25 nM
FLT3 IC50 = >10,000 nM
3
Second hit
initial hit
Auora A IC50 = 43 nM
computer-aid drug design
Auora A IC50 = 640 nM
Journal of Medicinal Chemistry, 2010, 53, 4980.
ChemMedChem, 2014, 9, 953.
N
NH
HN
S
NH
O
N
essential
for FLT3
Cl
O
N
This work
5
Lead compound
N
Auora A IC50 = 43 nM
FLT3 IC50 = 322 nM
NH
HN
S
NH
R1
R2
6
O
N
lead to drug
optimization
R3
R3 = H, Cl
7
N
6
Possible sites for
modification
N
Essential fuctional
group for dual activity
crucial
HN
R1 = H, F, OMe
R2 = H, F, Cl, OMe, OEt, CF3,
NH
clinical drug candidate
modification
S
NMe2
NH
H
O
N
O
O
Cl
O
NMe2
S
reduced
rotable bond nubmer
,
O
OMe
O
N
Me
N
O
N
,
H
O
N
BPR1K871 (1)
Investigational New Drug (FLT3/Auora A inhibitor)
O
O
NHAc ,
O
,
O
,
O
Auora A IC50 = 22 nM
FLT3 IC50 = 19 nM
I.V. Drug with no oral bioavailability
Oncotarget, 2016, 7, 86239.
Figure 2. Flow chart illustrating the discovery of drug candidate BPR1K871 (1) and
the design strategy in this work for improving oral bioavailability of 1.
2
. Chemistry
One of the key intermediate 7 was prepared according to our previous reports [18].
7

The chloro group of 7 was replaced by nucleophilic substitution with sodium methoxide
to give 8. On the other hand, substitution with sodium azide, followed by hydrogenation
over Pd/C provided the amine 10. Reacting 10 with methanesulfonyl chloride or acetic
anhydride produced the corresponding sulfonamide 11 and the acetamide 12.
Cl
Cl
O
O
N
N
NH
NH
NH
NH
HN
N
S
HN
N
S
N
N
b
R1
O
N3
O
9
7
R1 = Cl
a
8
R1 = OMe
Cl
Cl
O
O
N
S
N
NH
NH
NH
NH
HN
HN
N
S
c
N
d or e
N
R2
H2N
O
N
N
O
H
10
1 R2 = -SO Me
2
2 R2 = -COMe
1
1
Scheme 1. Reaction conditions and reagents: (a) NaOMe, KI, DMF, 100 °C, 16 h, 34%;
b) NaN , DMSO, 60 °C, 16 h, 95%; (c) H , 10% Pd/C, rt, 16 h, 64%; (d) MeSO Cl,
(
3
2
2
Et N, DMF, CH Cl , rt, 16 h, 15%; (e) Ac O, pyridine, CH Cl , rt, 48 h, 35%.
3
2
2
2
2
2
For compounds 16–37, quinazolinone derivatives 13a–v were either
commercially available or prepared according to reported methods. The starting
quinazolinone 13a–v were first chlorinated by thionyl chloride or phosphoryl chloride,
followed by nucleophilic aromatic substitution with tert-butyl 5-(2-aminoethyl)thiazol-
2
-yl-carbamate to give 14a–v. De-protection of the Boc group of 14a–v under acidic
8

condition gave the key intermediates 15a–v. Reaction of phenyl isocyanate or 3-
chlorophenyl isocyanate with 15a–v gave the final urea compounds 16–37.
N
N
NH2
NHBoc
O
N
HN
S
HN
N
S
R1
R2
R1
R2
R1
R2
b
NH
a
N
N
N
13a-o
15a-o
14a-o
1
1
3-15g, 22 : R1 = H, R2 = H, R3 = Cl
2
3 : R1 = H, R2 = H, R3 = H
3-15h, 24 : R1 = H, R2 = OMe, R3 = H
R3
25 : R1 = H, R2 = OEt, R3 = H
O
1
1
3-15i,
N
26 : R1 = H, R2 = Cl, R3 = H
3-15j,
NH
NH
3-15k, 27 : R1 = H, R2 = F, R3 = H
1
HN
N
S
28 : R1 = H, R2 = CF , R = H
3
R1
13-15l,
3
c or d
N
13-15m,
9 : R1 = F, R2 = H, R3 = H
2
13-15n, 30 : R1 = F, R2 = OMe, R3 = H
3-15o, 31 : R1 = F, R2 = OEt, R3 = H
3-15p 32 : R1 = F, R2 = F, R3 = H
R2
16-37
1
1
1
1
1
1
1
1
3-15a, 16: R1 = H, R2 = 1-methylpiperidinol-yl, R3 = Cl
3-15b, 17: R1 = H, R2 = CH N(CH ) , R = Cl
3-15c, 18: R1 = H, R2 = (S)-tetrahydrofuranol-yl, R3 = Cl
3-15d, 19: R1 = H, R2 = (tetrahydrofuran-3-yl)methanol- yl, R3 = Cl
3-15e, 20: R1 = F, R2 = 3-(dimethylamino)propanol-yl, R3 = Cl
3-15f, 21: R1 = F, R2 = (tetrahydrofuran-3-yl)methanol- yl, R3 = Cl
3: R1 = H, R2 = OMe, R3 = Cl
4 : R1 = H, R2 = F, R3 = Cl
5 : R1 = F, R2 = H, R3 = Cl
6 : R1 = F, R2 = F, R3 = Cl
7 : R1 = F, R2 = OEt, R3 = Cl
3
3
3
3
3
3
2
3 2
Scheme 2. Reaction conditions and reagents: (a) (i) SOCl or POCl , DMF, reflux, 15
2
3
h, (ii) tert-butyl 5-(2-aminoethyl)-thiazol-2-yl carbamate, Et N, EtOH, reflux, 15 h, 28–
3
9
7% for two steps; (b) 6.0 N HCl , rt, 10 h, 58–99% or TFA, CH Cl , rt, 12 h, 62%;
(aq) 2 2
(
c) 3-Cl-PhNCO, MeOH, CH Cl , rt, 16 h, 21-86%; (d) PhNCO, MeOH, CH Cl , rt, 16
2 2 2 2
h, 19–94%.
In a manner similar to compounds 16–37, compound 40 was prepared by
nucleophilic aromatic substitution of the 4-chloro quinazoline 38 [24] with tert-butyl
5
-(2-aminoethyl)thiazol-2-yl-carbamate. De-protection of 39 under acidic condition,
followed by reaction with 3-chlorophenyl isocyanate installed the urea group in 40.
9

Nucleophilic substitution reaction of compound 40 with dimethylamine gave the
desired compound 41.
Cl
O
NH
N
N
NH
NHBoc
Cl
N
HN
S
HN
N
S
O
N
a
O
O
b
O
O
N
N
Cl
O
Cl
N
Cl
38
39
40
Cl
O
N
NH
NH
HN
N
S
c
O
O
N
N
41
Scheme 3. Reaction conditions and reagents: (a) tert-butyl 5-(2-aminoethyl)-thiazol-2-
ylcarbamate, Et N, EtOH, reflux, 16 h, 78%; (b) (i) TFA, CH Cl , rt, 2 h, (ii) 3-Cl-
3
2
2
PhNCO, MeOH, CH Cl , rt, 16 h, 66% for two steps; (c) 40% (CH ) NH , DMF,
2
2
3 2
(aq)
reflux, 3 h, 80%.
3
. Results and discussion
Biological evaluation of Aurora A /B kinase inhibitory activities and SAR analysis
All the synthesized key compounds were evaluated for their Aurora A/B inhibitory
activities using the previously established Aurora kinase assay and cellular activity in
HCT-116 colon cancer cell line, as well as acute cellular toxicity in Detroit 551
embryonic skin cell lines [18,19]. The compounds were also evaluated for their
metabolic stability in in vitro rat liver microsomal stability (RLMS) assay, to select the
compound with favorable PK properties and good stability for further in vivo PK study
1
0

[25]. The microsomal stability assay is used to primarily investigate in vitro intrinsic
clearance of a compound under phase I metabolism and offer an approach for prediction
of the hepatic metabolism of an inhibitor in drug candidate development pipeline. The
5
0% inhibitory concentrations for Aurora kinase enzymatic assays (IC ) and cell
50
growth (CC ), the selectivity indices (SI; CC of Detroit 551/HCT-116) as well as rat
5
0
50
liver microsomal stability are presented in Tables 1 and 2. Compound 1 was used as the
starting point for the structure-activity and drug optimization studies.
In the field of medicinal chemistry, Lipinski’s rule of five and Veber's rule are the
guideline for medicinal chemist searching drug-like chemical properties with oral
effective potential. It is understood that fine tuning the physicochemical properties
(PCPs) of a lead molecule could alter the ADME property [26-30]. Hence, close
scrutiny was placed on the PCPs, by calculating the guiding parameters including
molecular weight (MW), number of hydrogen bond acceptors (HBA), number of
hydrogen bond donors (HBD), calculated LogP (cLogP), number of rotatable bonds
(NRB) and polar surface area (PSA) [30-32]. Compounds with optimal PCPs (MW <
2
5
00, HBA <10, HBD <5, LogP < 5, NRB < 10 and PSA < 140 Å ) have higher
probability for oral absorption. The clinical candidate 1 showed no oral bioavailability
and had high clearance in vivo. Only 14.5% of the drug remained after 30 minutes
incubation with rat liver microsomes, suggesting faster drug metabolism could be one
1
1

reason for poor oral bioavailability. In addition, it was found to violate two of the
empirical PCP guidelines (MW>500; NRB>10).
The initial focus of optimization was on the side chain bearing the solubilizing
group attached to the 7-position of quinazoline core of 1. In our previous studies with
different heteroaromatic core structures for kinase inhibition, we have shown that
introduction of N,N-dimethylamino solubilizing group into the heteroaromatic core
structures resulted in improved kinase inhibition as well as anti-proliferative activity
[18, 19]. However, it may be noted that oxidative deamination or N-demethylation of
substituted amine is one of the most common phase I metabolic reaction carried out by
cytochrome 450 (CYP) [33]. The metabolic reaction could inactivate a drug and also
lower oral bioavailability. Hence, we replaced the terminal N,N-dimethyl group in the
solubilizing side chain with alternate groups including methoxide, methylsulfonamide
and acetamide (i.e., 8 ,11, 12). The replacements enhanced the Aurora kinase inhibition
potency, microsomal stability and selectivity as well as cellular activity in HCT-116
cell assay, as compared to the lead 1.
The ability to alter the metabolic stability along with the retention of biological
activity, further enthused us to modify the solubilizing side chain. To add rigidity and
decrease the flexibility of the solubilizing side chain at 7-position of quinazoline core,
compounds 16–19 were synthesized. The rigid cyclic functional group with less NRB
1
2

at 7-position maintained their bioactivity against Aurora A and improved microsomal
stability, but increased the toxicity in Detroit 551 cell line (cf. 16, 18, and 19 with CC50
=
1.3–1.8 μM). The tetrahydrofuran-3-yl-methoxy group displayed 8.1-fold more
cellular activity in HCT-116 assay than 1 (cf. 1 versus 19). Modification of the
solubilizing dimethylaminopropoxy moiety of by replacement with
dimethylaminomethyl group resulted in loss of potency for both Aurora A/B inhibition
cf. 1 versus 17).
In addition, we have introduced an electron-donating methoxy group and electron-
1
(
withdrawing fluoro groups onto the C-6 position of the lead 1 to give compounds 41
and 20, respectively. Both compounds 20 and 41 showed improved stability (4.4–5.4-
fold) in rat liver microsome, as compared to 1. Even though the two compounds showed
good aurora kinase inhibition, the methoxy bearing 41 showed poor growth inhibition
against HCT-116 cells. Introduction of F-substituent at C-6 position of 1 enhanced
1
0.5-fold of the anti-proliferarive activity in HCT-116 assay with a selectivity indices
of 403. Encouraged by the activity of 6-fluoro substituted analog 20, a F-substituent
was also introduced at the C-6 position of 19 to obtain compound 21, which has similar
level of metabolic stability and anti-proliferative activity in HCT-116 with a selectivity
ratio of 115. Replacement of N,N-dimethylamino solubilizing functional group at 7-
position led to increased microsomal stability while retaining its bioactivities,
1
3

suggesting the substitution of the 7-position N,N-dimethylamino moiety is tolerated.
Moving forward to reduce the molecular weight and the number of rotatable bonds
in the inhibitors, for complying with PCPs guidlines, we investigated compound 22
with removal of substituent group from 7-postion of the quinazoline core and
compound 23 with additionally lacking the 3-Cl substitution in the terminal phenyl ring
of the urea side chain of 1 (Table 2). Both 22 and 23 displayed excellent Aurora
enzymatic inhibition and cellular growth inhibitory activity as well as higher
microsomal stability than that of 1 (cf. 1 versus 22 and 23). Encouraged by these results
that the removal of functional groups were well tolerated at quinazoline ring,
compounds 24–34 were synthesized by attaching electron-donating (ethoxy and
methoxy) and electron-withdrawing (chloro, fluoro and trifluoromethyl) groups at the
C-6 and C-7 position of the quinazoline ring. All these compounds showed improved
metabolic stability (RLMS = 41–80%) compared with the inhibitor 1. Moreover, most
of these compounds retained the Aurora A inhibition and had improved anti-
proliferative activity in HCT-116 cell line, except for compound 28 with
trifluoromethyl group. Particularly, compound 34 with a 7-F substitutent, had enhanced
Aurora kinase activity (IC , Aurora A IC = 9 nM, Aurora B IC = 30.5 nM) and anti-
5
0
50
50
proliferative activity in HCT-116 assay with high selectivity (i.e. 34, CC = 16.0 nM;
5
0
SI > 900-fold). Moreover, all of the target compounds reported in Table 2 possessed
1
4

good PCPs within the acceptable cut-off limits for oral bioavailability/drug-likeness.
Based on the in vitro potency, microsomal stability, toxicity and PCPs profile, we
selected compounds 20–23, 27, and 34 to perform in vivo PK studies.
In vivo pharmacokinetics investigation of selected compounds
Before we embarked to investigate the in vivo pharmacokinetics profile of promising
compounds, we carried out ADMET prediction using the Discovery Studio software.
The ADMET descriptor model predicts human intestinal absorption (HIA) and blood-
brain penetration (blood brain barrier, BBB) after oral administration. This model was
developed using the method reported by Egan et al. [34, 35], which ulilize multivariate
pattern recognition based on the predicted PSA and AlogP property (atom-type
partition coefficient) with 95% and 99% confidence as ellipses. Well-absorbed
compounds are expected to be found within the 95% confidence ellipse region. And in
general, oral absorption tends to drop off quite rapidly outside the 95% ellipse region.
Compound 1 with no oral bioavailability in rat PK study was located outside 95%
ellipse, and compounds 11 and 12 were found outside the 99% ellipse duo to high PSA
(
> 140 Å2).
According to the prediction model, 22, 23, 27, and 34 with no violated PCPs are
considered likely to be well-absorbed.
Based on the PCPs profile, RLMS, in vitro Aurora kinase inhibition and selective
1
5

anti-proliferative activity in HCT-116, 20–23, 27, and 34 were selected to carry out in
vivo pharmacokinetic studies in rats (Table 4). We found that both 20 and 21 showed
poor oral bioavailability (20, F = Not detected; 21, F = 6 %) similar to 1, despite 20 and
2
1 possessed higher in vitro metabolic stability. Notably, compound 22 showed 30%
and 23 showed 46% oral bioavailability by removal of the substituent from quinazoline
core. These encouraging results suggest that compounds with good in vitro metabolic
stability and appropriate PCPs would have acceptable oral bioavailability. Next, 27 and
3
4 with a 7-fluoro group in the quinazoline ring were chosen for PK evaluation, as they
possessed a high RLM stability and optimal in vitro activity profile. Accordingly, 34
was found to have better drug exposure on IV administration with 5-fold improved
th
AUC and only 1/4 the clearance as that of the initial clinical candidate BPR1K871 (1).
Moreover, oral administration of 34 resulted in an oral bioavailability of 54% and high
AUC, making it feasible for oral administration during pharamodynamics evaluation in
in vivo. The results indicated that compounds 22, 23, 27, and 34 with acceptable oral
absorption were located inside innermost oval in the ADMET prediction model.
1
6

Table 1. SAR investigations of compound 1, focusing on the quinazoline ring substitutions.a.b
N
NH
HN
N
S
NH
R1
R2
O
N
Cl
Selectivity
(Detroit
Detroit
551
compound
no.
Aurora A
IC (nM)
Aurora B@ 100 HCT-116
R1
R2
RLMS (%)
nM
CC (nM)
551/HCT-
5
0
50
CC (µM)
5
0
1
16 CC )
50
N
O
O
O
1
(BPR1K871)
8
H
H
H
H
14.5
39.1
61.5
59.9
22
IC = 13 nM
141
22.4
67.0
46.4
4.0
1.8
2.7
5.8
29
80
5
0
14.0
4.0
74%
80%
82%
O
O
S
1
1
1
2
N
H
O
40
O
N
H
O
8.0
125
1
7

N
1
1
1
1
4
2
2
6
7
8
9
1
0
1
H
H
81.6
41.3
74.5
52.3
75.7
63.8
66.3
38.0
262.0
26.0
22.0
6.0
ND
7%
320.4
2962
119.7
17.4
1.4
N.D.
1.3
4
O
N
N.D
11
O
H
55%
O
O
H
83%
1.8
103
44
O
N
O
OMe
F
IC = 12 nM
287.7
13.4
12.6
5.4
5
0
N
O
10.0
15.9
86%
78%
403
115
O
F
44.4
5.1
O
aThe rat liver microsomal stability (RLMS) assays offer a method to evaluate the rate of hepatic metabolism of a compound in liver microsomal
b
incubation. Data was represented as % remaining at 30 min by incubation of the compound in rat microsomes.; IC values represent the mean
5
0
of at least two independent experiments.; N.D.: Not determined.
1
8

Table 2. SAR investigations of compound 1, focusing on both the quinazoline ring and urea side chain substitutions.a,b
N
NH
HN
N
S
NH
R1
R2
O
N
R3
Selectivity
(Detroit
Detroit
compound
no.
Aurora A Aurora B
IC (nM) @ 100 nM
HCT-116
R1
R2
R3
RLMS (%)
551 CC50
CC (nM)
551/HCT-
5
0
50
(µM)
1
16 CC )
50
IC = 13
5
0
N
O
1
H
Cl
14.5
22
141
4.0
6.6
29
nM
71%
53%
2
2
2
3
H
H
H
H
Cl
H
80.4
90.0
15.0
11.1
10.0
29.0
660
586
17.0
1
9

2
2
2
2
2
2
3
3
3
3
3
4
5
6
7
8
9
0
1
2
3
4
H
H
H
H
H
F
OMe
OEt
Cl
H
H
H
H
H
H
H
H
H
Cl
Cl
41.8
60.5
67.3
80.7
41.3
71.1
42.0
47.4
71.1
77.5
77.8
18.8
14.0
24.0
13.0
46.0
14.0
4.2
74%
75%
21.7
17.7
23.0
66.0
229.2
79.4
21.2
22.4
83.2
30.4
16.0
7.6
6.7
350
379
613
244
38
50%
14.1
16.1
8.6
F
50%
CF3
H
32%
52%
> 20
5.8
> 252
274
290
> 240
95
F
OMe
OEt
F
73%
F
12.0
39.0
9.0
69%
6.5
F
35%
> 20
2.9
H
H
OMe
F
89%
9.0
83% (IC50
14.5
906
2
0

=
30.5 nM)
67%
3
3
5
6
F
F
H
F
Cl
Cl
78.9
87.7
19.1
13.3
23.0
22.8
14.7
12.8
639
561
45%
8
0% (IC50
3
7
F
OEt
Cl
74.5
8.4
52.0
4.5
87
=
64 nM)
aThe rat liver microsomal stability (RLMS) assays offer a method to evaluate the rate of hepatic metabolism of a compound in liver microsomal
b
incubation. Data was represented as % remaining at 30 min by incubation of the compound in rate microsomes.; IC values represent the mean
5
0
of at least two independent experiments.
2
1

Table 3. The physicochemical properties (PCPs) of the synthesized inhibitorsa.
compound
no.
PCPs
violationsb
cLogP MW HBA HBD NRB
PSA
BPR1K871 (1) 4.2
526.1
513.0
576.1
540.0
538.1
482.0
511.0
525.0
544.0
543.0
424.9
390.5
420.5
434.5
424.9
408.5
458.5
408.5
438.5
452.5
426.4
454.9
442.9
442.9
460.9
487.0
556.1
9
9
3
3
4
4
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
11
11
12
11
8
132.54
138.53
183.85
158.39
132.54
123.3
2
2
4
3
1
0
1
1
2
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
3
8
3.9
3.0
3.2
4.1
4.1
3.8
4.0
4.4
4.2
4.0
3.3
3.3
3.7
4.0
3.5
4.3
3.5
3.5
3.9
3.7
4.0
4.2
4.2
4.4
4.5
4.2
1
1
2
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
1
11
10
9
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
4
8
8
9
8
138.53
138.52
132.54
138.53
120.07
120.07
129.3
9
9
9
11
9
9
7
6
7
6
8
7
8
8
129.3
7
6
120.07
120.07
120.07
120.07
129.29
129.29
120.07
129.29
120.07
120.07
120.07
129.29
141.77
7
6
7
7
7
6
8
7
8
8
7
6
8
7
7
6
7
6
7
6
8
8
10
12
aMW (molecular weight), cLogP (calculated LogP), HBA (number of hydrogen bond
acceptors), HBD (number of hydrogen bond donors), NRB (number of free rotatable
bond) and PSA (polar surface area) were calculated by using Discovery Studio.;
bViolations according to the Lipinski’s rule of five and Veber’s suggestion.
2
2

Table 4. Pharmacokinetics profile of selected compounds in rata.
IV (Dose: 5 mg/kg)
compound
PO (Dose: 20 mg/kg)
F (%)
t1/2
CL
(ml/min/kg)
60.4
Vss
(l/kg)
34.2
12.6
3.5
AUC(0-inf)
(ng/mL*h)
1464
t1/2
(h)
Cmax
(ng/mL)
N.D.
N.D.
125
Tmax
(h)
AUC(0-inf)
(ng/mL*h)
N.D.
no.
(h)
1
23.5
11.4
1.7
N.D.a
N.D.
2.8
N.D.
N.D.
5.3
N.D.
2
2
2
2
2
3
0
1
2
3
7
4
42.6
1958
N.D.
N.D.
6.3
27.1
3272
818
4.6
52.5
5.7
1599
5.1
488
0.8
1951
30.0
46.0
19.9
54.0
1.3
61.7
4.0
1396
2.7
679
1.0
3014
6.8
24.3
3.2
3478
4.9
1111
4146
0.7
2768
3.6
13.0
1.1
7451
3
3.3
16065
aN.D.: Not determined.
2
3

Figure 3. ADMET properties predicted for the quinazoline compounds.
Compounds located inside the 95% confidence ellipse are better for this parameter.
Profiling of compound 34 in kinase panel and cancer cell lines
Based on the in vitro biological activities and in vivo PK data, compound 34 was
selected as suitable candidate to further subject it to kinase profiling using the
KINOMEscan technology against a panel of 468 kinases (containing 403 non-mutant
kinases) at a concentration of 100 nM (Figure 4) [36]. The results for binding
2
4

interactions were reported as %control. Inhibitor 34 displayed broad and high inhibition
against several oncogenic kinases, such as Aurora kinase (2.5–14%), DDR1 (2.7%),
DDR2 (7.6%), FLT3 (10%), HIPK4 (8.9%), KIT (20%), MERTK (18%), MKNK2
(
6.7%), and TRK family (4.7–12%). Based on the profile, the compound 34 IC s for
50
these kinases were evaluated by KINOMEscan platform (Figure 5). Aurora kinase
family members were potently inhibited at single-digit nanomolar concentrations of 34.
In addition, 34 demonstrated high potency (IC ) against several kinases especially the
5
0
tyrosine kinases, including FLT3 (129 nM), DDR1 (16.7 nM), DDR2 (143 nM), TRKA
23.6 nM), TRKB (35.5 nM), and TRKC (12.5 nM). These results overall suggested
(
that 34 is a potent multi-kinase inhibitor.
Figure 4. Kinase profiling of 34 using the KINOMEscan technology.
2
5

Figure 5. KINOMEscan profile of compound 34 at 100 nM for kinases with % control
50.
<
Next, 34 was tested in a panel of in-house cancer cell lines, including human colon
cancer (Colo205, HCT-116), uterine sarcoma (MESSA), human gastric cancer (MKN-
5), human lung cancer (H1975, H446), and acute myeloid leukemia (MOLM-13,
4
MV4-11). For comparison, the two known multi-kinase inhibitors, linifanib and
sorafenib, were also examined side-by-side. The results revealed that compound 34
showed better potency against colon cancer HCT-116 and leukemia cell lines MOLM-
1
1
3 and MV4-11 (Table 5). 34 was more potent than the clinical candidate 1 in HCT-
16 colon cancer and MKN45 human gastric cancer cell line; but was less potent in the
case of leukemia cell lines MOLM-13, MV4-11, which harbor FLT3 mutations.
2
6

Table 5. Anti-proliferative activity of compound 34 in a panel of in-house cell lines.
compoun
d no.
Cell lines CC (nM)
5
0
Solid tumor type
Leukemia type Normal type
Colo
05
HCT-
116
MOLM MV4-
Detroit 551
MESSA MKN45
H1975
H446
2
-13
5
11
4
1
(BPR1K8
34
141
216
355
495
–
4000
7
1)
3
4
144
–
16
506
–
128
–
491
7010
8853
102
–
76
82
38
73
43
82
14500
Sorafenib
Linifanib
8800
8070
–
–
–
–
–
–
4
. Conclusion
Discovering a new drug, particularly drugs that can be orally administered to
patients is a challenging task. Nevertheless, orally administered drugs have the
advantage of easy administration as compared to injectable forms. Administering a drug
through oral route requires that the molecule be orally bioavailable, i.e., get absorbed
through the gut and enter the systemic circulation without being extensively
2
7

metabolized in the liver. Means to manipulate and improve the oral bioavailability of a
drug without compromising the activity is pursued during lead optimization stage.
In this respect, here we report our efforts to optimize our previous lead BPR1K871
(1), which is a clinical candidate for cancer treatment, to improve the pharmacokinetics
property and identify orally bioavailable lead. Careful consideration of
physicochemical properties (PCP) and in vitro rat liver microsomal stability (RLMS)
data provided a path to carry out a systematic medicinal chemistry optimization of lead
1
. During PCP driven optimization it was found that the solubilizing functional group
at the 7-position of quinazoline ring was not essential for maintaining the activity and
also the group is responsible for poor metabolic stability and oral absorption. A detailed
structural investigation at the 6- and 7-position of quinazoline ring identified 34 as an
orally bioavailable multi-kinase inhibitor with in vitro enzyme/cellular activity profile
very similar to lead 1. The results suggest that 34 is a promising lead candidate for
further evaluation of antitumor efficacy in animal models.
2
8

5
5
5
. Experimental
.1. Synthesis
.1.1. General methods
All reagents were purchased from commercial suppliers and were used without further
purification unless otherwise noted. All reactions were carried out under dry nitrogen
or argon atmosphere and were monitored by TLC using Merck 60 F254 silica gel glass
backed plates; zones were detected visually under UV irradiation (254 nm) or by
spraying with phosphomolybdic acid reagent (Sigma-Aldrich) followed by heating at
8
2
0 °C. Flash column chromatography using silica gel (Merck Kiesegel 60, no. 9385,
1
30–400 mesh ASTM). H NMR spectra were recorded on a Varian Mercury-300 or
Varian Mercury-400 spectrometers, and the chemical shifts were recorded in parts per
million (ppm, δ) and reported relative to TMS or the solvent peak. Low-resolution mass
spectra (LRMS) data were obtained from Agilent MSD-1100 ESI-MS/ MS system and
Agilent 1200 series APCI-ESI-MS/MS system. Purity of the final compounds were
determined with a Hitachi 2000 series HPLC system using C-18 column (Agilent
ZORBAX Eclipse XDB-C18 5 μm. 4.6 mm × 150 mm) operating at 25 C. For method
A, elution was carried out using water containing 0.1% formic acid + 10 mmol
NH OAc as mobile phase A, and acetonitrile as mobile phase B. Elution conditions: 0
4
to 45 min, 10 to 90% mobile B; 45 to 60 min, 90% to 10% mobile B. For method B,
2
9