Discovery, mechanism and metabolism studies of 2,3-difluorophenyl-linker-containing PARP1 inhibitors with enhanced in?vivo efficacy for cancer therapy

Article abstract of DOI:10.1016/j.ejmech.2017.06.053

Poly (ADP-ribose) polymerase 1 (PARP1) is overexpressed in a variety of cancers, especially breast and ovarian cancers, and tumor cell lines deficient in breast cancer gene 1/2 (BRCA1/2) are highly sensitive to PARP1 inhibition. In this study, with the help of molecular docking, we identified a novel series of 2,3-difluorophenyl-linker analogues (15–54) derived from olaparib (1) as PARP1 inhibitors. Lead optimization led to the identification of 47, which showed high selectivity and high potency against PARP1 enzyme (IC₅₀ = 1.3 nM), V-C8 cells (IC₅₀ = 0.003 nM), Capan-1 cells (IC₅₀ = 7.1 nM) and MDA-MB-436 cells (IC₅₀ = 0.2 nM). Compound 47 had more potent PARP1-DNA trapping and double-strand breaks (DSBs)-induction activities than 1 and induced G2/M arrest and caspase-dependent apoptosis. Compound 47 (50 mg/kg, 94.2%) had a more beneficial effect on tumor growth inhibition than 1 (100 mg/kg, 65.0%) in a BRCA1-mutated xenograft model and significantly inhibited tumor growth (40 mg/kg, 48.1%) in a BRCA2-mutated xenograft model, with no negative influence on the body weight of the mice. Collectively, these data demonstrated that 47 might be an excellent drug candidate for the treatment of cancer, especially for BRCA-deficient tumors.

Full text of DOI:10.1016/j.ejmech.2017.06.053

Accepted Manuscript
Discovery, mechanism and metabolism studies of 2,3-difluorophenyl-Linker-
Containing PARP1 inhibitors with enhanced in vivo efficacy for cancer therapy
Wenhua Chen, Ne Guo, Minghui Qi, Haiying Dai, Minghuang Hong, Longfei Guan,
Xiajuan Huan, Shanshan Song, Jinxue He, Yingqing Wang, Yong Xi, Xinying Yang,
Yanyan Shen, Yi Su, Yiming Sun, Yinglei Gao, Yi Chen, Jian Ding, Yun Tang, Guobin
Ren, Zehong Miao, Jian Li
PII:
S0223-5234(17)30507-X
10.1016/j.ejmech.2017.06.053
EJMECH 9548
DOI:
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 23 April 2017
Revised Date: 30 May 2017
Accepted Date: 25 June 2017
Please cite this article as: W. Chen, N. Guo, M. Qi, H. Dai, M. Hong, L. Guan, X. Huan, S. Song, J. He,
Y. Wang, Y. Xi, X. Yang, Y. Shen, Y. Su, Y. Sun, Y. Gao, Y. Chen, J. Ding, Y. Tang, G. Ren, Z. Miao,
J. Li, Discovery, mechanism and metabolism studies of 2,3-difluorophenyl-Linker-Containing PARP1
inhibitors with enhanced in vivo efficacy for cancer therapy, European Journal of Medicinal Chemistry
(2017), doi: 10.1016/j.ejmech.2017.06.053.
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. 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.

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT
Discovery, Mechanism and Metabolism Studies of
,3-Difluorophenyl-Linker-Containing PARP1 Inhibitors with
Enhanced In Vivo Efficacy for Cancer Therapy
2
a,†
b,d,†
c,†
a
c
Wenhua Chen , Ne Guo , Minghui Qi , Haiying Dai , Minghuang Hong ,
a
b
b
b
b
Longfei Guan , Xiajuan Huan , Shanshan Song , Jinxue He , Yingqing Wang , Yong
b
b
b
b
b
b
b
Xi , Xinying Yang , Yanyan Shen , Yi Su , Yiming Sun , Yinglei Gao , Yi Chen , Jian
b
a
c,
b,
a,
Ding , Yun Tang , Guobin Ren *, Zehong Miao *, Jian Li *
a
Shanghai Key Laboratory of New Drug Design, School of Pharmacy, East China
University of Science and Technology, Shanghai 200237, China
b
Division of Antitumor Pharmacology, State Key Laboratory of Drug Research,
Shanghai Institute of Materia Medica, Chinese Academy of Sciences, Shanghai
2
01203, China
c
Laboratory of Pharmaceutical Crystal Engineering & Technology, School of
Pharmacy, East China University of Science and Technology, Shanghai 200237,
China
d
University of Chinese Academy of Sciences, Beijing 100049, China
†
These authors contributed equally to this work.
*
To whom correspondence should be addressed: jianli@ecust.edu.cn,
zhmiao@simm.ac.cn or rgb@ecust.edu.cn.

ACCEPTED MANUSCRIPT
Abstract
Poly (ADP-ribose) polymerase 1 (PARP1) is overexpressed in a variety of
cancers, especially breast and ovarian cancers, and tumor cell lines deficient in breast
cancer gene 1/2 (BRCA1/2) are highly sensitive to PARP1 inhibition. In this study,
with the help of molecular docking, we identified a novel series of
2
,3-difluorophenyl-linker analogues (15-54) derived from olaparib (1) as PARP1
inhibitors. Lead optimization led to the identification of 47, which showed high
selectivity and high potency against PARP1 enzyme (IC = 1.3 nM), V-C8 cells (IC
50
5
0
=
0.003 nM), Capan-1 cells (IC = 7.1 nM) and MDA-MB-436 cells (IC = 0.2 nM).
50 50
Compound 47 had more potent PARP1-DNA trapping and double-strand breaks
DSBs)-induction activities than 1 and induced G2/M arrest and caspase-dependent
(
apoptosis. Compound 47 (50 mg/kg, 94.2%) had a more beneficial effect on tumor
growth inhibition than 1 (100 mg/kg, 65.0%) in a BRCA1-mutated xenograft model
and significantly inhibited tumor growth (40 mg/kg, 48.1%) in a BRCA2-mutated
xenograft model, with no negative influence on the body weight of the mice.
Collectively, these data demonstrated that 47 might be an excellent drug candidate for
the treatment of cancer, especially for BRCA-deficient tumors.
Key words: Antitumor drug; BRCA1/2; PARP inhibitors; Molecular docking

ACCEPTED MANUSCRIPT
1
. Introduction
Poly (ADP-ribose) polymerases (PARPs) are ADP-ribosyl transferases that
+
cleave nicotinamide adenine dinucleotide (NAD ) to give nicotinamide and an
intermediate oxonium ion, which react with nucleophiles on the target proteins (such
as histones, topoisomerases, DNA polymerases, and DNA ligases) or themselves. [1-3]
PARPs are involved in a number of important biological processes, such as the
surveillance of genome integrity, progression of the cell cycle, initiation of the DNA
damage response, apoptosis, and regulation of transcription [4]. Among the PARP
family members, PARP1 is the most abundant and plays an essential role in the repair
of DNA single-strand breaks (SSBs) [5-6] It accounts for more than 90% of
ADP-ribosylation occurring in cells and is evolutionarily conserved in all advanced
eukaryotes. [7-8] PARP1 is an enzymatic 113 kDa protein containing three major
structural domains, including a 42 kDa DNA binding domain with two zinc fingers in
the N-terminal region, a 55 kDa catalytic domain in the C-terminal region and an
automodification domain in the central region.[9-11] The DNA binding domain can
recognize and locate the SSBs in damaged DNA and then bind to them to stimulate
the polymerization of ADP-ribose, which leads to the unwinding of DNA from
histones and exposes the damaged DNA for repair. The function of the catalytic
+
domain is to utilize NAD as a substrate to construct linear and branched polymers of
ADP-ribose onto its targets, which enables it to perform a pivotal role in DNA
damage repair.
PARP1 is overexpressed in a variety of cancers, and its expression has been
associated with overall prognosis in cancers, especially breast and ovarian cancers,
[
10-11] so PARP1 has emerged as an attractive anticancer drug target for effecting
DNA damage. Cell lines deficient in breast cancer genes 1/2 (BRCA1/2, two known

ACCEPTED MANUSCRIPT
tumor suppressor genes whose mutations are associated with breast, ovary, prostate,
and pancreas cancers) were 1000-fold more sensitive to PARP inhibition than
wild-type or heterozygous mutant cells. Therefore, the inhibition of PARP1 can
induce highly selective killing of BRCA1/2-deficient tumor cells. In this regard, many
small molecules targeting PARP1 have been developed, either as chemosensitizers in
combination with ionizing radiation or DNA-damaging chemotherapeutic agents or as
stand-alone (synthetic lethal) therapies to kill cancers defective in DNA repair
mechanisms (esp. BRCA1/2 mutant breast or ovarian tumors) [12-23].
Many PARP inhibitors have been reported, and most can be regarded as
analogues of nicotinamide.[6, 24-25] It has been observed that several PARP
inhibitors (Figure 1) have been in late-phase (phases 2 and 3) clinical trials and were
even approved for the treatment of some ovarian cancers, including olaparib (1,
AZD-2281; AstraZeneca/Kudos) [26], rucaparib (2, AG-014699; Clovis/Pfizer)
[
27-28], niraparib (3, MK-4827; Tesaro/Merck)[29-30], veliparib (4, ABT-888;
AbbVie)[31-33], and talazoparib (5, BMN 673; BioMarin/LEAD/Medivation).[34]
Figure 1. PARP inhibitors currently in late-phase (phases 2 and 3) clinical trials or

ACCEPTED MANUSCRIPT
already approved for clinical use.
From the perspective of steric effects, the influence of fluorine is anticipated to
be marginal as fluorine is a small atom with a van der Waals radius of 1.47 Å, close to
the 1.20 Å value for hydrogen [35-37]. Moreover, the high electronegativity of
fluorine results in the high polarization of C−F bonds and potential new
hydrogen-bond interaction between fluorine atoms and drug targets. In addition,
fluorine is a powerful tool for modulating the pK of proximal functional groups and
a
the electron densities of (hetero)aromatic rings.[38-40] The pK of an ionizable center
a
in a drug molecule changes the lipophilic profile and influences solubility,
permeability, and protein binding, which can manifest as changes in potency,
selectivity, toxicity, and pharmacokinetic (PK) properties, including absorption,
distribution, metabolism, and excretion (ADME).[41-46] As a result, the
incorporation of fluorine into drug molecules is a commonly used tactic to increase
potency and optimize the ADME profile.[47-48]
Studies on 1, the first approved drug for some ovarian cancers, showed that the
phthalazine fragment was essential and the N-substituted piperazine was flexible for
PARP1 inhibition potency.[26] As other PARP inhibitors have come to the fore, 1
was found to demonstrate inferior antitumor efficacy, especially compared to 5.[35]
In an attempt to improve the antitumor efficacy of 1, we introduced a fluorine at the
3
-position of the phenyl linker in 1 and designed a new 2,3-difluorophenyl-linker
analogue (15) (Figure 2). Molecular docking analysis showed that not only the
-position fluorine atom of analogue 15 but also the 2-position fluorine atom formed a
3
hydrogen bond with Tyr889, in contrast to 1 (Figure 2). To explore the
structure-activity relationship (SAR) of the 2,3-difluorophenyl-linker analogues of 1,

ACCEPTED MANUSCRIPT
we designed a novel series of 2,3-difluorophenyl-linker analogues (15-54) derived
from 1 (Figure 2) and exhaustively evaluated whether the type of substituent (R,
Figure 2) at the terminal nitrogen of piperazine would improve pharmacological
activity. Finally, a highly potent PARP1 enzymatic inhibitor (47) with excellent in
vitro and in vivo antitumor efficacy against both BRCA1- and BRCA2-deficient tumor
cells, good specificity for BRCA-deficient cells and a good safety profile was
identified as a unique candidate compound for the development of antitumor drug.
Figure 2. Chemical modification strategies (15–54) for lead compound 1.
2
. Results and Discussion
2
.1. Chemistry
As outlined in Scheme 1, 1,2-difluoro-3-iodo-5-methylbenzene (7) was prepared
from commercially available 1,2-difluoro-4-methylbenzene (6) in the presence of
n-BuLi and iodine in THF at -78 °C for 3 h in a 70% yield. Further treatment of 7
with cuprous cyanide in N,N-dimethylformamide under reflux for 20 h gave rise to
2
,3-difluoro-5-methylbenzonitrile (8) in a 62% yield. The bromination of 8 with
N-bromosuccinimide in CCl under reflux for 48 h yielded 9 in a 66% yield. The
4
hydrolysis of 9 provided the key intermediate 2,3-difluoro-5-formylbenzonitrile (10).
Dimethyl (3-oxo-1,3-dihydroisobenzofuran-1-yl)phosphonate (12) was prepared in a
5
2% yield by the reaction of 2-formylbenzoic acid with dimethyl phosphonate and
methanesulfonic acid in sodium methoxide solution at 0-20 °C for 3 h by following a

ACCEPTED MANUSCRIPT
literature procedure [24]. The coupling of 10 and 12 afforded compound 13 in an 86%
yield by reacting the two in the presence of triethylamine and tetrahydrofuran at
1
5-20 °C for 2 h. 2,3-Difluoro-5-((4-oxo-3,4-dihydrophthalazin-1-yl)methyl)benzoic
acid (14) was prepared in a 75% yield from 13 with hydrazine hydrate and sodium
hydroxide at 70-90 °C for 19 h. The subsequent condensation of acid 14 with an
appropriate piperazine substrate in the presence of HOBt and EDCI in
2
6
dichloromethane produced target analogues 15-54 in 30-70% yields. All analogues
were confirmed to be ≥95% purity (Table S1, Supporting Information). Additionally,
to confirm the 2,3-disubstituted positions of the two fluorine atoms, a fine single
crystal of 39 was obtained, and the structure was determined by X-ray scattering
analysis. The structure (Figure 3) clearly showed the two fluorine atoms at the 2- and
3
-positions. (Tables S2 and S3, Supporting Information).
Scheme 1. Synthesis of analogues 15-54.
Reagents: (a) I , n-BuLi, THF, -78 °C, 3 h; (b) DMF, CuCN, reflux, 20 h; (c) NBS,
2

ACCEPTED MANUSCRIPT
CCl , reflux, 48 h; (d) AgNO , H O, CH CN, reflux, 2 h; (e) MeONa, (CH O) PHO,
4
3
2
3
3
2
CH SO H, 0-20 °C, 3 h; (f) THF, Et N, 15-20 °C, 2 h; (g) (i) NaOH, 90 °C, 1 h; (ii)
3
3
3
NH -NH ·H O, 70 °C, 18 h; (h) HOBt, EDCI, DIPEA, piperazine, DMA, overnight.
2
2
2
Figure 3. The X-ray crystal structure of analogue 39.
2
.2. Lead Optimization and SAR
All synthesized analogues were first evaluated for PARP1 enzymatic inhibitory
activity using 1 as the reference standard in the assay. Considering the importance of
the fluorine atom, analogue 15 was designed by replacing the middle 2-fluorophenyl
linker with a 2,3-difluorophenyl linker. As expected, analogue 15 displayed slightly
better enzymatic potency than 1 (IC = 11.2 nM vs 19.1 nM), which was further
5
0
supported by molecular docking analysis (see Figure 8 below), the docking pose of 15
see Figure 8B below) displayed two additional hydrogen bonds in contrast to the
binding pose of 1, which were formed between the two fluorine atoms and Tyr889
see Figure 8A below). The hydrogen-bonding network changes might be a plausible
(
(

ACCEPTED MANUSCRIPT
factor that induces the difference in enzymatic inhibitory activity between 1 and 15.
Encouraged by this result, we turned our attention to explore the SAR by
incorporating various substituents (R) on the N-atom of the piperazine ring in
analogue 15.
First, a small set of analogues with various acyl substituents (16-20, Table 1) on
the terminal nitrogen of the piperazine ring were examined. The enzymatic inhibitory
activity of acetyl analogue 16 was slightly better than that of analogue 15 (IC = 7.1
5
0
nM vs 11.2 nM). Replacing cyclopropane with tetrahydrofuran or morpholine gave
rise in analogues 18-19, which led to a dramatic improvement in the enzymatic
inhibitory activity (IC50 = 2.4 nM and 2.3 nM, respectively). Interestingly, the
benzoyl (17) and 4-methylbenzoyl (20) analogues showed significantly different
enzymatic activities (IC = 6.1 nM vs 29.1 nM).
5
0
To further develop this 2,3-difluorophenyl linker, we next explored analogues
2
1-26, in which the acyl fragment was replaced by alkyl groups (Table 1). Alkyl
analogues 21-22 and 24-26 clearly exhibited improved enzymatic inhibitory activities
IC = 1.4 nM, 2.1 nM, 2.6 nM, 3.5 nM and 0.8 nM, respectively); by contrast, 23
(
5
0
showed similar enzymatic inhibition as 15 (IC = 9.7 nM vs 11.2 nM). These data
5
0
indicated that bulky substituents (e.g., 26) at the terminal nitrogen of the piperazine
ring could play a favorable role in enzyme inhibition. Thus, we next assessed
compounds with arylmethyl (27-28) and aryl (29) groups at the terminal of the
piperazine ring (Table 1). Analogues 27-28 (IC50 = 96.8 nM and 156.7 nM) with
(hetero)arylmethyl substitutions showed approximately 8-fold and 14-fold less
enzymatic inhibitory activity than 15, respectively. Significantly, analogue 29 with a

ACCEPTED MANUSCRIPT
phenyl substituent demonstrated a high potency against PARP1 enzyme compared to
1
5 (IC = 0.5 nM vs 11.2 nM).
50
Considering the advantage of the phenyl substituent at the terminal nitrogen of
the piperazine ring, we next tested the efficacy of differently substituted phenyl
analogues (30-44, Table 1). To our surprise, electron-withdrawing or
electron-donating substituents on the phenyl ring indistinguishably affected the
inhibitory activity (e.g., 30 vs 40 vs 43). In addition, which positions on the phenyl
ring were substituted also had no consistent impact on enzymatic activity, for example,
3
0 vs 37 (IC = 0.4 nM vs 55.8 nM), 33 vs 39 (IC = 5.4 nM vs 0.02 nM), or 32 vs
50 50
3
6 (IC = 103.0 nM vs 4.2 nM).
5
0
With the intention of extending the substitution scope at the terminal nitrogen of
the piperazine moiety, we replaced the phenyl ring with other heteroaryl rings (45-54,
Table 1). Adding a pyrid-2-yl group (45) (IC = 0.1 nM) and a pyrid-4-yl group (46)
5
0
(
IC = 0.9 nM) significantly enhanced the enzymatic inhibitory activity (45-46 vs 1
50
and 15). Thus, we explored the inhibitory effects of incorporating several
electron-withdrawing substituents on the pyridine ring (47-48) and using double
heteroatom-containing heteroaryl rings (49-54). Analogues 47, 49 and 53 displayed
potent enzymatic inhibitory activity (IC = 1.3 nM, 2.9 nM and 0.2 nM, respectively).
5
0
We observed that the size of the heteroaryl ring was very important for the enzymatic
inhibitory activity and bulky substitutions (50-52) as well as fused aromatic rings (54)
were not tolerable.

ACCEPTED MANUSCRIPT
Analysis of the data shown in Table 1 revealed some noteworthy observations
from the SAR study of analogues 1 and 15-54: (1) replacement of the middle
2
-fluorophenyl linker with 2,3-difluorophenyl linker could be tolerated (1 vs 15); (2)
the presence of an R substituent at the terminal nitrogen of the piperazine ring was
critical for high potency, and arylalkyl (27-28), bromo- or methoxy(hetero)aryl (31,
4
1-42, 50-51), and fused aryl (54) substituents were not beneficial, whereas alkyl
(especially the bulky cycloalkyl group in 26), aryl (especially the 2-CN-phenyl group
in 39) and heteroaryl (especially the unsubstituted heteroaryl groups in 45 and 53)
substituents could improve the inhibitory activity; and (3) in the studied sets of
monosubstituted (hetero)aryl group (Table 1), the contribution to potency
improvement was in the order of CN > CH > F = NO = OH = CF > Br = Cl =
3
2
3
OCH3.
2
.3. Cell Proliferation Inhibition of Potent Analogues
From the results described above, 21 analogues, i.e., 18-19, 21-22, 24-26, 29-30,
4-36, 39-40, 43-47, 49 and 53, were identified as potent PARP1 inhibitors with IC50
3
values lower than 5 nM. Because analogues 25 and 49 slightly precipitated out of
solution in the PARP1 enzyme activity test, the remaining 19 analogues were further
evaluated for their inhibition of cell proliferation. As shown in Table 2, the selected
analogues together with reference compound 1 were tested in a matched pair of
previously well-characterized Chinese hamster lung fibroblast V79 (wild-type) and
V-C8 (BRCA2-deficient) cells and breast cancer MDA-MB-436 (BRCA1-deficient)
cells.

ACCEPTED MANUSCRIPT
Interestingly, all the analogues, except for 29, 36 and 40, showed stronger
inhibition of the proliferation of BRCA1-deficient cells than 1. Although analogues 19,
2
1, 29, 35-36, 40, 43 and 46 demonstrated better enzymatic inhibitory activities than 1
(
IC = < 5 nM vs 19.1 nM), they were much less potent in terms of their cellular
5
0
activity in BRCA2-deficient cells (IC > 620 nM). Moreover, a similar trend was
5
0
observed for analogues 26, 30 and 34-35, which displayed potent enzymatic
inhibitory activity (IC50 < 1 nM), but their cellular activity in the V-C8 and
MDA-MB-436 assays was slightly better than or similar to that of 1. It was noted that
all the analogues showed less potent anti-proliferation activities against
BRCA2-deficient cells than against BRCA1-deficient cells. There were 10 analogues
(
18, 24, 26, 30, 34, 39, 44-45, 47 and 53) that displayed better cell inhibitory activity
IC = 3.2-284.2 nM vs 565.6 nM against V-C8, and IC = 0.03-10.8 nM vs 39.3 nM
(
5
0
50
against MDA-MB-436) and better selectivity against BRCA2-deficient cells than 1.
Although analogue 45 exhibited both good enzymatic (IC = 0.1 nM) and cellular
5
0
inhibitory activities (IC50 = < 0.03 nM and < 3.2 nM), it also showed modest
proliferation inhibition in wild-type V79 cells (IC = 924.3 nM). Two of the most
50
potent inhibitors (39 and 47) displayed not only excellent enzymatic (IC = 0.02 nM
5
0
and 1.3 nM) and cellular inhibitory activities (IC = 8.8 nM and 0.003 nM against
50
V-C8, IC = < 0.05 nM and 0.2 nM against MDA-MB-436, respectively) but also
5
0
good selectivity, which confirmed that the tested PARP1 inhibitors could selectively
kill BRCA2-deficient cells. Therefore, we decided to focus on these two analogues for
further development.

ACCEPTED MANUSCRIPT
Table 1. In vitro PARP1 enzyme activities of analogues 15-54.
PARP1 enzyme
IC (nM)
PARP1 enzyme
PARP1 enzyme
PARP1 enzyme
No.
R
No.
R
NO.
R
No.
R
IC (nM)
IC (nM)
IC (nM)
50
50
50
50
a
1
19.1±11.2
1
5
6
11.2
25
26
3.5
0.8
35
36
0.8
45
46
0.1
0.9
a
1
7.1
6.1
4.2±2.8
a
1
7
8
27
28
96.8
37
38
55.8
47
48
1.3±0.3
a
a
1
2.4±0.6
156.7
323.7
5.1±2.3

ACCEPTED MANUSCRIPT
a
1
9
2.3
29
0.5
39
0.02±0.006
49
2.9
2
0
1
29.1
1.4
30
31
0.4
40
41
<0.1
50
51
174.9
110.0
2
83.4
112.2
22
23
24
2.1
9.7
2.6
32
33
34
103.0
5.4
42
43
44
296.6
0.6
52
53
54
70.5
0.2
0.4
0.9
52.2
a
Average of more than two experiments.

ACCEPTED MANUSCRIPT
Table 2. Cell proliferation inhibition activities of some representative analogues.
Cell activity IC50 (nM)
Cell activity IC50 (nM)
MDA-MB-436 V-C8
No.
MDA-MB-436
V-C8
V79
No.
V79
-
/-
-/-
+/+
-/-
-/-
+/+
(
BRCA1 )
(BRCA2 )
565.6±105.4
68.1
(BRCA2 )
>10000
(BRCA1 )
(BRCA2 )
624.3
(BRCA2 )
>10000
a
a
1
39.3±8.0
35
36
23.4
18
19
21
4.9
>10000
395.8
1599.3
7881.6
a
a
25.7
12.9
937.8
987.1
>10000
>10000
39
40
<0.05
405.2
8.8±0.7
4336.1±61.5
>10000
2076.1

ACCEPTED MANUSCRIPT
2
2
4
<0.03
10.6
148.7
284.2
771.4
43
44
27.1
0.3
3966.3
31.7
>10000
7833.1
2
>10000
26
29
30
34
6.0
82.4
10.8
9.4
42.6
1944.6
90.1
>10000
6025.9
2853.5
>10000
45
46
47
53
<0.03
3.7
<3.2
3287.8
924.3
>10000
>10000
>10000
a
a
0.2±0.01
6.3
0.003±0.001
136.7
206.0
a
Average of more than two experiments.

ACCEPTED MANUSCRIPT
2
.4. The Water Solubility Evaluation
Since water solubility is an important property of drugs, we next tested the
solubility of the two potent PARP1 inhibitors selected (39 and 47). Analogues 39 and
4
7 showed poor solubility compared to 1 (S = < 0.2 µg/mL and 1.1 µg/mL vs S =
1
03.9 µg/mL), and this lower solubility may be due to the more lipophilic terminal
(hetero)aryl groups. Considering the presence of basic groups (tertiary amine) in the
structures of analogues 39 and 47, we explored the solubility of the corresponding
salts in an attempt to improve their water solubility. Unfortunately, the two analogues
failed to be salified in salt screening experiments. Considering the relatively good
water solubility of 47 and its excellent in vitro biological profiles (Tables 1-2),
compound 47 was further evaluated in mechanism-based in vitro assays and in vivo
xenograft tumor models. Moreover, to ensure the mice could be orally dosed with 47
in the in vivo test, a formulation of 47 was developed. To meet the target drug
concentration value of 2 mg/mL, 10 mg of 47 was dissolved in a mixed solution (5.0
mL) that consisted of 70% citrate buffer solution (3.5 mL), 20% PEG 200 (1.0 mL),
and 10% vitamin E TPGS (0.5 mL).
2
.5. Compound 47 Induces PARP1-DNA Trapping and DNA Double-Strand Breaks
DSBs)
PARP1-DNA trapping has been proposed as one of anticancer mechanisms of
(
PARP inhibitors [35, 49]. Compound 47 was shown to increase PARP1-DNA trapping
in both BRCA1-deficient MDA-MB-436 cells and BRCA2-deficient Capan-1 cells. As
shown in Figure 4A, 47 displayed a potency similar to that of 5 but stronger than that
of 1. Consequently, 47 increased the levels of γH2AX, a DSBs biomarker, in a

ACCEPTED MANUSCRIPT
concentration-dependent manner in both MDA-MB-436 and Capan-1 cells to a
greater degree than 1 (Figure 4B). These data indicate that compared to parent
compound 1, 47 has improved PARP1-DNA trapping and DSBs-induction activities.
Figure 4. Compound 47 induces PARP1-DNA trapping (A) and DSBs accumulation
(
B) in BRCA1-deficient MDA-MB-436 cells and BRCA2-deficient Capan-1 cells. A,
PARP1-DNA trapping were increased by compound 1, 5 and 47 both in MDA-MB-
36 and Capan-1 cells determined by Western blotting. B, the γH2AX accumulation
4
induced by compound 1 and 47 both in MDA-MB-436 and Capan-1 cells detected by
Western blotting.
2
.6. Compound 47 Induces G2/M Arrest and Apoptosis in HR-Deficient Cells
As a result of inducing DSBs accumulation, PARP inhibitors can cause G2/M
arrest [35, 49-50]. Our results showed that 47 induced typical G2/M arrest in a
concentration-dependent manner in BRCA2-deficient Capan-1 cells (Figure 5A).
Compound 47 enhanced the protein levels of phosphorylated Chk1 and p-ATM but

ACCEPTED MANUSCRIPT
did not change the levels of corresponding total proteins (Figure 5B). Compound 47
also increased the formation of phosphorylated ATM (p-ATM) foci in the cells (Figure
5
C). The results indicate that 47 activates the cell cycle checkpoint and induces G2/M
arrest.
On the other hand, DSBs accumulation caused by PARP inhibitors can trigger
the caspase cascade and induce apoptosis [35, 49-50]. The treatment of
BRCA1-deficient MDA-MB-436 cells with 47 caused apoptosis in
a
concentration-dependent manner, which was reversed by pretreatment with
Z-VAD-FMK, a pan-caspase inhibitor (Figure 6A). The treatment consistently
resulted in the activation of caspase 3 and caspase 7 (Figure 6B). Parent compound 1
caused similar but weaker effects than 47 (Figure 6). The data indicate that 47
induced caspase-dependent apoptosis.

ACCEPTED MANUSCRIPT
Figure 5. Compound 47 induces G2/M arrest (A) and activates the cell cycle
checkpoint (B and C) in BRCA2-deficient Capan-1 cells. A, G2/M arrest induced by
compound 1 and 47 in Capan-1 cells analyzed by PI-staining-based flow cytometry.
Upper, representative images; lower, quantitative results (mean ± SD) from
corresponding images with 3 independent experiments. B, cell cycle checkpoint
protein levels changed by compound 1 and 47 in Capan-1 cells determined by

ACCEPTED MANUSCRIPT
Western blotting. C, cell cycle checkpoint p-ATM foci induced by compound 1 and 47
in Capan-1 cells analyzed by confocal microscopy. Left, confocal images; right,
quantification of p-ATM positive cells with more than 5 bright foci. 50 cells were
counted on each slide.
Figure 6. Compound 47 induces apoptosis (A) and activates caspase 3/7 (B) in
BRCA1-deficient MDA-MB-436 cells. A, apoptosis induced by compound 1 and 47
alone or pretreated with 20 µM Z-VAD-FMK for 1 hour analyzed by Annexin V-FITC
and PI-staining-based flow cytometry. Upper, representative images; lower,
quantitative results (mean ± SD) from corresponding images. B, the changes in
caspase 3/7 activity in MDA-MB-436 cells exposed to compound 1 and 47 for 72 h
®
detected by Caspase-Glo 3/7 assays. Data were expressed as mean ± SD from 3
independent experiments.
2
.7. Compound 47 Selectively Kills HR-Deficient Cells

ACCEPTED MANUSCRIPT
As mentioned before, 47 strongly inhibited the proliferation of BRCA2- deficient
V-C8 cells with IC = 0.003 nM but practically did not affect the wild-type V79 cells
5
0
at all (IC50 > 10 µM). Using two other BRCA-deficient cells, including
BRCA1-deficient MDA-MB-436 and BRCA2-deficient Capan-1 cells, we further
confirmed that 47 has potent cell killing capacity in HR-deficient cells. Moreover, the
cell killing ability of 47 was over 50-fold greater than that of 1 in all tested
BRCA-deficient cells (Table 3).
Table 3. Cell proliferation inhibition activities of compounds 1, 5 and 47.
a
Cell activity IC50 (nM)
PARP1 enzyme
No.
V-C8
V79
Capan-1
MDA-MB-436
a
IC (nM)
5
0
-/-
+/+
-/-
-/-
(
BRCA2 )
(BRCA2 )
(BRCA2 )
(BRCA1 )
1
5
19.1±11.2
7.2±3.5
565.6±105.4 54205.0±3297.9 399.3±52.7
39.3±8.0
0.7±0.2
4.2±0.2
5011.4±823.4
1.8±1.4
7.1±0.1
4
7
1.3±0.3
0.003±0.001 10339.7±463.0
0.2±0.01
a
Average of more than two experiments.
2
.8. In Vivo Antitumor Activities
To assess the in vivo antitumor effects of compound 47 as a single agent, nude
mice bearing subcutaneous BRCA1-mutated human breast cancer MDA-MB-436
xenografts and BRCA2-mutated human pancreatic cancer Capan-1 xenografts were
both treated with 47 once each day for 21 consecutive days (Figure 7). In the
MDA-MB-436 xenograft model, the oral administration of 50 mg/kg 47 significantly
inhibited growth, as was evident from the excellent growth inhibition rate of 94.2% at

ACCEPTED MANUSCRIPT
the end of treatments, and it was more effective than 1 at 100 mg/kg dose (growth
inhibition rate: 65.0%). Meanwhile, 47 at 50 mg/kg was well tolerated, with no
animal lethality or significant weight loss observed during the 21 consecutive
treatment days (Figure 7A). In the Capan-1 xenograft model, we used 5 as the
reference. The oral administration of 40 mg/kg 47 significantly inhibited growth, as
was evident from the growth inhibition rate of 48.1% at the end of treatments,
although this was less effective than 5 at a 0.5 mg/kg dose (growth inhibition rate:
8
0.0%). However, 47 at 40 mg/kg was still well tolerated, with no animal lethality
and evidently less weight loss than that caused by 5 (Figure 7B).
Figure 7. In vivo antitumor activity in BRCA-deficient human breast (A) and
pancreatic (B) cancer xenograft models. The nude mice bearing subcutaneous
xenografts were derived from MDA-MB-436 (A) and Capan-1 (B) cell lines were
orally given Compound 47 once every day. Olaparib (100 mg/kg) was used as the
positive control for MDA-MB-436 xenografts and talazoparib (0.5 mg/kg) was used

ACCEPTED MANUSCRIPT
for Capan-1 xenografts. Relative tumor volume (RTV) and body weight were
separately plotted over the treatment time. Data were analyzed by Student t test. **P
<
0.01; ***P < 0.001.
2
.9. Pharmacokinetic Properties
Its excellent in vitro and in vivo biological properties demonstrated that 47 was a
good drug candidate for targeting PARP1. Therefore, we next assessed the
pharmacokinetic properties of 47. As shown in Table 4, 47, 1 and 5 demonstrated
significantly different oral bioavailability after oral exposure in Sprague-Dawley rats,
with oral bioavailability values of 4.7%, 45.4% and 103.5%, respectively. Analogue
4
7 had a better oral half-life than 1 (3.0 h vs 1.0 h) and displayed faster plasma
clearance (3.6 L/hr/kg vs 1.8 L/hr/kg) as well as bigger volume distribution (27.5 L/kg
vs 3.0 L/kg). Considering the significant difference in oral bioavailability between 5
and 47 (F = 103.5% vs 4.7%), we proceeded to evaluate other pharmacokinetic
properties. As shown in Table 5, 47 had a short terminal elimination half-life
compared to 5 (3.0 min vs >60 min) and displayed similar plasma stability (104.9% vs
1
05.7%) and plasma protein binding rate (98.8% vs 94.3%). The Caco-2
transmembrane transport results showed 47 had better transmembrane permeability
-6
-6
-6
than 5 (Papp(
A B)
→
= 6.8*10 cm/s vs 0.9*10 cm/s, Papp
(B→A)
= 34.1*10 cm/s vs
-6
5
1.8*10 cm/s) and that 47 and 5 were both P-gp substrates (Efflux Ratio 5.0 vs 58.8).
To understand the short terminal elimination half-life of 47 in liver microsome
metabolism, the metabolites of 47 were further explored. As shown in Scheme 2,
1
9.8% of 47 was oxidized into M5, 3.0% of 47 was metabolized into an alcohol (M6),
and 0.9% of 47 was metabolized into an acid (M8) (Table S6, Supporting
Information). Obviously, the difluorobenzylphthalazinone fragment was the main

ACCEPTED MANUSCRIPT
metabolic site (16.4%), and aromatic hydroxylation was a major metabolic pathway,
which might be a major cause of the short terminal elimination half-life.
Scheme 2. Quantitative analysis and metabolite identification of 47.

ACCEPTED MANUSCRIPT
Table 4. Pharmacokinetic parameters in rats for compounds 1, 5 and 47.
No.
Structure
t (h)
Cl (L/hr/kg)
1.8±0.3
V (L/kg)
Cmax (µg/L)
AUC (µg/L*hr)
5500.3±834.5
F (%)
1
/2
D
a
iv (10 mg/kg)
1.1±0.1
1.0±0.1
4.6±1.1
3.0±0.6
12287.3±809.9
1
a
po (20 mg/kg)
2594.6±408.2
2059.6±66.1
4989.5±479.8
6541.8±2066.0
45.4±4.4
a
iv (1 mg/kg)
0.2±0.1
3.6±0.1
1.1±0.3
5
a
po (2 mg/kg)
3.4±0.2
5.3±0.4
3.0±1.9
3104.1±282.7
3258.4±504.2
74.4±8.2
13173.3±537.4 103.5±4.2
1376.7±39.4
a
iv (5 mg/kg)
27.5±2.5
4
7
a
po (10 mg/kg)
129.5±20.7
4.7±0.8
a
Formulation: 5% DMSO, 95% HP-β-CD aqueous solution.

ACCEPTED MANUSCRIPT
Table 5. Pharmacokinetic parameters in vitro for compounds 5 and 47.
Liver microsome metabolism
Caco-2 transmembrane transport
Papp(A→B) Papp(B→A) Efflux ratio
(*10 cm/s) (Papp(B→A)/Papp(A→B))
Plasma protein
No.
Plasma stability (%)
Elimination
t1/2
binding rate (%)
-6
-6
rate k
(min)
(*10 cm/s)
a
a
5
105.7±5.2
94.3±0.3
0.3
>60
3.0
0.9
51.8
34.1
58.8
5.0
a
a
4
7
104.9±3.6
98.8±0.03
<0.01
6.8
a
Average of more than three experiments.

ACCEPTED MANUSCRIPT
.10. Molecular Docking of 1, 15 and 47 with PARP1
2
The molecular docking results of the binding of 1, 15 and 47 to the PARP1
catalytic domain provided insight into the different inhibitory potencies among the
three compounds. As shown in Figure 8, the phthalazine scaffold of the three
compounds all formed three hydrogen bonds with the Ser904 and Gly863 residues
+
surrounding the NAD site. In contrast to the binding pose of 1, the docking pose of
1
5 (Figure 8B) displayed two additional hydrogen bonds, which were formed between
the two fluorine atoms and Tyr889 (Figure 8A). Moreover, in contrast to analogue 15,
the binding pose of 47 (Figure 8C) demonstrated an additional hydrogen bond via the
interaction of the nitrile group on the terminal pyridine ring with Tyr889 (Figure 8B).
The hydrogen-bonding network changes predicted by molecular docking may be a
plausible factor that contributes to the difference in enzymatic inhibitory activity
between 1, 15 and 47 (IC = 19.1 nM, 11.2 nM and 1.3 nM, respectively).
5
0
Figure 8. Molecular docking studies on 1, 15 and 47. Docked poses of 1 (green sticks)
+
(
A), 15 (green sticks) (B) and 47 (green sticks) (C) in the NAD binding site (cyan
and pink cartoon). Hydrophilic interacting residues are represented with yellow
carbons. Hydrogen bonding distances (Å) are represented with black numbers.
Hydrogen bonds are shown with yellow dash lines.
3
. Conclusion