Discovery of the cancer cell selective dual acting anti-cancer agent (Z)-2-(1H-indol-3-yl)-3-(isoquinolin-5-yl)acrylonitrile (A131)

Article abstract of DOI:10.1016/j.ejmech.2018.07.011

Selective targeting of cancer cells over normal cells is a key objective of targeted therapy. However few approaches achieve true mechanistic selectivity resulting in debilitating side effects and dose limitation. In this work we describe the discovery of A131 (4a), a new agent with an unprecedented dual mechanism of action targeting both mitosis and autophagy. Compound 4a was first identified in a phenotypic screen in which HeLa cells treated with 4a manifested mitotic arrest along with formation of multiple vesicles. Further investigations showed that 4a causes an increase in mitotic marker pH3 and autophagy marker LC3. Importantly 4a induces cell death in cancer cells while sparing normal cells which regrow after 4a is removed. Dual activities against pH3 and LC3 markers are required for cancer cell selectivity. An extensive SAR investigation confirmed 4a as the optimal dual inhibitor with potency against a panel of 30 cancer cell lines (average antiproliferative GI₅₀ 1.5 μM). In a mouse model of paclitaxel-resistant colon cancer, 4a showed 74% tumor growth inhibition when administered at a dose of 20 mg/kg IP twice a day.

Full text of DOI:10.1016/j.ejmech.2018.07.011

Accepted Manuscript
Discovery of the cancer cell selective dual acting anti-cancer agent (Z)-2-(1H-indol-3-
yl)-3-(isoquinolin-5-yl)acrylonitrile (A131)
Cheng Shang See, Mayumi Kitagawa, Pei-Ju Liao, Kyung Hee Lee, Jasmine Wong,
Sang Hyun Lee, Brian W. Dymock
PII:
S0223-5234(18)30567-1
10.1016/j.ejmech.2018.07.011
EJMECH 10546
DOI:
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 27 April 2018
Revised Date: 4 July 2018
Accepted Date: 6 July 2018
Please cite this article as: C.S. See, M. Kitagawa, P.-J. Liao, K.H. Lee, J. Wong, S.H. Lee, B.W.
Dymock, Discovery of the cancer cell selective dual acting anti-cancer agent (Z)-2-(1H-indol-3-yl)-3-
(isoquinolin-5-yl)acrylonitrile (A131), European Journal of Medicinal Chemistry (2018), doi: 10.1016/
j.ejmech.2018.07.011.
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 of the Cancer Cell Selective Dual Acting Anti-Cancer Agent (Z)-
2
-(1H-indol-3-yl)-3-(isoquinolin-5-yl)acrylonitrile (A131)
1
2
2
2
2
Cheng Shang See , Mayumi Kitagawa , Pei-Ju Liao , Kyung Hee Lee , Jasmine Wong , Sang
2
1,*
Hyun Lee and Brian W. Dymock
1
Department of Pharmacy, National University of Singapore, 117543, Singapore
2
Program in Cancer & Stem Cell Biology, Duke-NUS Medical School, 169857, Singapore
*
Correspondence: bwdnus@gmail.com
We dedicate this work to the memory of Professor Sang Hyun Lee.
Running Title: Discovery of the Cancer Cell Selective Dual Inhibitor A131
Keywords: Antitumor; Antimitotic; Antiproliferative; Tumor-selective

ACCEPTED MANUSCRIPT
Abstract
Selective targeting of cancer cells over normal cells is a key objective of targeted therapy.
However few approaches achieve true mechanistic selectivity resulting in debilitating side
effects and dose limitation. In this work we describe the discovery of A131 (4a), a new agent
with an unprecedented dual mechanism of action targeting both mitosis and autophagy.
Compound 4a was first identified in a phenotypic screen in which HeLa cells treated with 4a
manifested mitotic arrest along with formation of multiple vesicles. Further investigations
showed that 4a causes an increase in mitotic marker pH3 and autophagy marker LC3.
Importantly 4a induces cell death in cancer cells while sparing normal cells which regrow
after 4a is removed. Dual activities against pH3 and LC3 markers are required for cancer cell
selectivity. An extensive SAR investigation confirmed 4a as the optimal dual inhibitor with
potency against a panel of 30 cancer cell lines (average antiproliferative GI 1.5 µM). In a
5
0
mouse model of paclitaxel-resistant colon cancer, 4a showed 74% tumor growth inhibition
when administered at a dose of 20 mg/kg IP twice a day.

ACCEPTED MANUSCRIPT
Introduction
Success in the treatment of cancer is increased if selectivity between cancer cells and normal
cells can be achieved. Agents which kill cancer cells while sparing normal cells can be given
at higher doses leading to improved efficacy while balancing side effects. However it has
been very challenging to find agents which have fundamentally differentiated mechanisms
between cancer and normal cells [1].
Several compounds bearing an arylacrylonitrile structure have been shown to exhibit a range
of anticancer biological activities (Fig. 1). A series of benzotriazole-acrylonitriles
demonstrated antituberculosis activity as well as antiproliferative activity in human tumor
cells [2-4]. Closely related triazolopyridinyl derivatives were subsequently found to be
antimicrotubule agents [5]. In recent years, indoleacrylonitriles such as Paprotrain (1) were
reported to be inhibitors of the kinesin MKlp2 [6-8]. Cyclization of paprotrain derivatives
using photochemistry led to fused tricyclic compounds that showed inhibitory activities on
kinases involved in neurodegenerative disorders [9].
We started our work studying 1, an inhibitor of the kinesin MKlp2 which induces a unique
cellular phenotype of binucleation [6]. We synthesised the novel analogue 4a which was
found to have unusual phenotypic properties in cancer cells but not in normal cells. Hence we
embarked upon a detailed study of the chemistry and biology of 4a. Herein we discuss the
discovery of 4a, subsequent SAR studies and characterisation in cells and in a mouse model
of drug resistant cancer.

ACCEPTED MANUSCRIPT
R
R
Br
N
R
N
N
H
N
H
N
N
N
N
N
Benzotriazole derviatives:
Antituberculosis activity,
antiproliferative activity in human tumor cells
Triazolopyridinyl derivatives:
Antiproliferative activity in human tumor cells,
inhibitor of tubulin polymerization
N
N
N
N
N
R
H
H
N
HN
R
N
H
X
N
H
R
Paprotrain (1):
MKlp2 inhibitor
Cyclized analogues of paprotrain:
A131 (4a)
Inhibitors of DYRKs and CLKs
Induces cell death in
cancer cells while
sparing normal cels
Fig. 1. Examples of arylacrylonitriles with biological activities.
Identification of compound 4a from phenotypic screening
We performed a phenotypic screen of a small in-house library of derivatives of 1 in HeLa
cells, monitoring for visible perturbations to mitosis. One compound, 4a (Scheme 1),
presented a unique and interesting phenotype. In HeLa cells treated with 4a, prolonged
mitotic arrest was observed. At the same time, distinct large vacuole-like structures that filled
out the cytoplasmic area were also visible using a high resolution microscope (Fig. 2).

ACCEPTED MANUSCRIPT
Fig. 2. HeLa cells stably expressing GFP-tubulin were treated with DMSO and 4a for 20 h.
White scale bar represents 5 ꢀm.
Treatment with chloroquine, an effective late-stage inhibitor of autophagy in vitro, has been
known to result in an accumulation of double-membraned structures called autophagosomes
which can be observed using microscopy [10]. These structures contain cytoplasmic material
that will be digested upon fusion with lysosomes [11]. To investigate if the vacuole-like
structures induced by treatment with 4a were related to autophagic vesicles, the levels of
microtubule-associated protein light chain 3 (LC3) in 4a-treated cells were evaluated by
immunoblot. LC3 is a widely used marker for autophagy and it exists as two forms: LC3-I
which is found in the cytosol and LC3-II which is conjugated with phosphatidylethanolamine
and localized in the inner and outer membranes of the autophagosome [12, 13]. The amount
of LC3-II is correlated with the number of autophagosomes [12].
As shown in Fig. 3, a concentration dependent increase in LC3-II levels without any
significant changes to LC3-I levels was seen when HeLa cells were treated with increasing
doses of 4a for 6 h. Such effects were identical to that seen with chloroquine, however,
compared to chloroquine, 4a appears to be 4 to 10 times more potent as an inhibitor of

ACCEPTED MANUSCRIPT
lysosome function, determined by the relative intensity ratio of LC3-II to LC-I compared to
DMSO-treated control. Consistent with these findings, immunofluorescence imaging using
antibodies against LC3 saw an increase in LC3 puncta formation following treatment with 4a,
but not with DMSO control (Fig. 4).
Fig. 3. Immunoblot of LC3 and pH3(Ser10) in HeLa cells following treatment with
chloroquine and 4a for 6 h. The relative band intensity ratios of LC3-II to LC3-I as compared
to DMSO control (lane 1, 0 ꢀM chloroquine and lane 7, 0 ꢀM 4a) are indicated.

ACCEPTED MANUSCRIPT
Fig. 4. Immunofluorescence analysis of HeLa cells following treatment with DMSO and 4a
for 6 h using antibodies against LC3 and counterstaining with DAPI to visualize the nucleus.
Increased puncta formation is indicative of increase in LC3-II levels. White scale bar
represent 5 ꢀm.
Histone H3 phosphorylated at serine 10 (pH3(Ser10)) levels were also evaluated by
immunoblot of 4a-treated HeLa cells (Fig. 3, lower panel). Accumulation of pH3 was seen in
a concentration-dependent manner with 4a but not with chloroquine, signifying that 4a has
the ability to impede mitotic progression in HeLa cells while chloroquine does not. Indeed,
examination of the mitotic spindles with immunofluorescence microscopy revealed that
spindle defects were not observed with chloroquine treatment, only HeLa cells treated with
4
a were arrested in mitosis with grossly misaligned chromosomes and disorganized mitotic
spindles (Fig. 5).

ACCEPTED MANUSCRIPT
Fig. 5. Immunofluorescence analysis of HeLa cells following treatment with DMSO,
chloroquine and 4a for 6 h using antibodies against β-tubulin and counterstaining with DAPI
to visualize the chromosomes. Only 4a-treated HeLa cells arrested in mitosis with
disorganized spindles and misaligned chromosomes. White scale bar represents 5 ꢀm.
The intriguing dual phenotypic outcomes upon treatment of HeLa cells with 4a led us to
initiate a program of study to systematically investigate the structure activity relationships
between analogues of 4a, and to further characterise the most potent compounds in cellular
models.

ACCEPTED MANUSCRIPT
Strategy for SAR exploration
In this study, we aimed to establish a preliminary structure-activity relationship (SAR)
around lead compound 4a to elucidate the key chemical and structural features that are
important for the dual activities of inhibiting mitotic progression and disrupting autophagy.
Medicinal chemistry efforts focused on modifying the indole ring, introducing different
substituents to the indole ring, altering the position of the nitrogen atom in the isoquinoline
ring, as well as replacing the acrylonitrile moiety with alternative linkers. Synthesized
compounds were evaluated using immunoblot analysis of the pH3 (Ser 10) and LC3 markers
following treatment in transformed BJ fibroblast cells for 24 h at 5 µM. Relative intensity
ratios of the bands for pH3 to DMSO treated control, as well as LC3-II to LC3-I compared to
DMSO treated control, were determined. For each biomarker, a less than 2-fold induction is
considered insignificant.

ACCEPTED MANUSCRIPT
Chemistry
Scheme 1. Preparation of isomers and analogues of the isoquinoline ring 1, 4a-g and 6.
N
3
a, 4a: R = 5-isoquinoline
3b, 4b: R = 8-isoquinoline
c, 4c: R = 5-quinoline
3d, 4d: R = 4-quinoline
e, 4e: R = 8-quinoline
3f, 4f: R = 4-isoquinoline
N
R
H
R
H
a
3
O
3
HN
N
H
3
3
g, 4g: R = 4-(2,7-naphthyridine)
h, 1: R = 3-pyridine
2
3a-h
4a-g, 1
N
N
N
a
3
a
N
H
HN
H
N
N
5
6
Reagents and Conditions: (a) CH ONa, MeOH, MW 95 °C, 8.5 min, 22-74%.
3

ACCEPTED MANUSCRIPT
Scheme 2. Preparation of indole analogues 10a-i, 14a-d.
H
N
N
O
N
4
7
5
a
b
c
2
R
6
R
H
N
N
HN
HN
H
H
7
7
7
a-b,
8a-b,
8d,
8e-i
R
R
d,
g-i
9a-i
10a-i
7
7
9
7
8
8
7
7
7
a, 8a, 9a, 10a: R = 2-CH3
b, 8b, 9b, 10b: R = 4-Br
c, 10c: R = 5-OCH3
d, 8d, 9d, 10d: R = 5-F
e, 9e, 10e: R = 5-Cl
f, 9f, 10f: R = 5-Br
g, 8g, 9g, 10g: R = 5-NHCOCH3
h, 8h, 9h, 10h: R = 6-OCH3
i, 8i, 9i, 10i: R = 6-Br
N
N
N
N
4
5
d
e
c
2
R
6
R
H
N
N
7
X
X
HN
HN
H
H
X
X
11a-e
12a-e
R
R
13a-d
14a-d
1
1
1
1
1
1a, 12a, 13a, 14a: X = CH, R = 5-NO2
1b, 12b, 13b, 14b: X = CH, R = 7-OCH3
1c, 12c, 13c, 14c: X = CH, R = 7-Br
1d, 12d, 13d, 14d: X = N, R = H (7-azaindole)
1e, 12e: X = CH, R = H
Reagents and Conditions: (a) POCl , DMF, 0 °C to RT, 1.5 h, 49-90%; (b) KCN, NaBH ,
3
4
MeOH/formamide, 55 °C, 16 h, 39-76%; (c) 3a, CH ONa, MeOH, MW 95 °C, 8.5 min or
3
sealed tube 75 °C, 16 h, 15-94%; (d) (CH ) NH, HCHO, AcOH, H O, RT, 16 h, 66-97%; (e)
3
2
2
KCN, CH I, EtOH/H O, RT, 16 h, 39-71%.
3
2

ACCEPTED MANUSCRIPT
Scheme 3. Preparation of intermediate 7g and indazole analogue 20.
H
N
NO2
NH2
a
b
O
N
N
N
H
H
H
15
16
7g
H
N
N
O
N
c
d
e
N
H
N
N
N
N
HN
HN
H
H
17
18
19
20
Reagents and Conditions: (a) H , Pd/C, EtOH, RT, 2.5 h, 53%; (b) CH COCl, Et N, DCM, rt,
2
3
3
3
h, 85%; (c) NaNO , HCl (aq), dioxane/H O, RT, 2 h, 7%; (d) KCN, NaBH ,
2 2 4
MeOH/formamide, 55 °C, 16 h, 60%; (e) 3a, CH ONa, MeOH, sealed tube 75 °C, 16 h, 15-
3
1
5%.

ACCEPTED MANUSCRIPT
Scheme 4. Preparation of analogues 21, 24-27.
H
N
a
12e
3a
H
HN
21
OH
N
O
N
O
NH
b,c
NH2
23
HN
N
H
22
24
4
a
d
e
f
N
O
N
N
H N
N
N
2
H
H
N
HN
HN
25
26
27
Reagents and Conditions: (a) PBu , MeCN, 90 °C, 22 h, 28%; (b) (COCl) , DCM, RT, 2 h; (c)
3
2
Et N, THF, RT, 16 h, 42% (2 steps); (d) Pulverized KOH, t-BuOH, reflux, 18 h, 39%; (e)
3
NaBH , THF/MeOH, 75 °C, 40 h, 20%; (f) CH I, NaH, THF, 0 °C to RT, 4 h, 88%.
4
3

ACCEPTED MANUSCRIPT
Scheme 5. Preparation of analogues 32a-c, 36a-c.
N
N
N
N
N
a
b
c
H
HN
HN
HN
HN
R
R
O
OH
O
O
28
29
30a: R = (CH ) OCH ,
2 2 3
32a: R = (CH
32b: R = (CH ) C(OCH ) ,
)
OCH ,
3
2
2
3
0b: R = (CH ) C(OCH ) ,
2
2
2 2
2 2 2 2
3
0c: R = (CH ) Cl
2
3
32c: R =
N
d
c
R =
N
31c
e
N
BF K
Br
BF K
3
3
33
34
N
N
N
N
f
c
H
HN
HN
HN
N
N
R
10e: R = 5-Cl,
10i: R = 6-Br,
14c: R = 7-Br
35a: 5-substituted,
35b: 6-substituted,
35c: 7-substituted
3
3
6a: 5-substituted,
6b: 6-substituted,
36c: 7-substituted
Reagents and Conditions: (a) H , Pd/C, THF, RT, 7 h, 84%; (b) Alkyl halide, K CO , DMF,
2
2
3
5
7
4
0 °C, 16 h, 45-62%; (c) 3a, CH ONa, MeOH, MW 95 °C, 8.5 min or sealed tube 75 °C, 16 h,
3
-76%; (d) Pyrrolidine, K CO , NaI, MeCN, reflux, 40 h, 78%; (e) Pyrrolidine, THF, RT,
2
3
0 h, 16%; (f) 33, Pd(OAc) XPhos, Cs CO THF/H O, 85 °C, 24 h, 53-59%.
2
,
2
3,
2

ACCEPTED MANUSCRIPT
The general approach for the synthesis of the acrylonitriles entails firstly the preparation of
various
indole-3-acetonitrile
precursors
followed
by
in
condensation
with
isoquinoline/quinoline/pyridine-5-carboxaldehydes
(3a-h)
a Knoevenagel-like
condensation reaction using sodium methoxide as the base to afford the final compounds
Scheme 1) [6]. The reaction was performed in anhydrous methanol in a sealed glass tube at
5 °C for 16 h in darkness. Microwave irradiation at a higher temperature of 95 °C allowed
(
7
the reaction time to be shortened to 8.5 min. A number of the final compounds exhibited
limited solubility in methanol. Reducing the solvent volume by evaporation and then chilling
the mixture in an ice-bath facilitated precipitation of the desired products. For the more polar
compounds that dissolved well in methanol, flash column chromatography and subsequent
recrystallization furnished the desired products in good purity. X-ray single crystal analysis
of 4a revealed that the Z-configuration (heterocycles on opposite sides of the double bond)
was obtained as the predominate isomer from the reaction. All compounds were obtained as
single isomers except for 10a and 10b, which were isolated as inseparable mixtures with Z:E
ratios of 88:12 and 75:25, respectively.
Two synthetic routes were employed for the preparation of the indole-3-acetonitrile
precursors (Scheme 2). Starting from a wide variety of commercially available substituted
indoles 7a-b, 7d, 7i (reduction of 5-nitroindole (15) to 5-aminoindole (16) followed by
acetylation allowed access to the acetamide-substituted indole 7g (Scheme 3)), a formyl
group was first introduced at the 3-position via the Vilsmeier-Haack reaction. This was
followed by a one-pot sodium borohydride reduction and cyanide displacement to obtain the
indole-3-acetonitrile intermediates 9a-i. Alternatively, the starting indoles were subjected to a
Mannich reaction to generate the gramines 12a-e, which underwent subsequent substitution
by the nucleophilic cyanide anion to afford the indole-3-acetonitrile intermediates 13a-d. The
key precursor to the synthesis of the indazole analogue 20 is the 3-formyl-indazole fragment

ACCEPTED MANUSCRIPT
1
8. By reacting indole 17 with sodium nitrite under acidic conditions, this intermediate was
successfully synthesized, but only in a meagre yield of 7%. Nevertheless, the subsequent
nitrile transformation and condensation steps proceeded smoothly in good yields, rendering a
straightforward overall route to indazole 20 (Scheme 3).
Following the procedure reported by Low and Magomedov [14], alkene-linked 21 was
prepared through the direct coupling of gramine 12e with isoquinoline-5-carboxaldehyde (3a)
in an operationally simple and efficient phosphine-mediated reaction (Scheme 4). Amide
bond formation between indole-3-carboxylic acid (22) and 5-amino-isoquinoline (23), using
oxalyl chloride to generate the acyl chloride, afforded the amide-linked analogue 24 in
moderate yield. Under reflux conditions in the presence of pulverized potassium hydroxide
and t-butanol, the nitrile group in 4a was transformed to a primary amide to furnish 25, which
was isolated as an inseparable 75:25 Z:E mixture. Reducing the double bond in 4a using
sodium borohydride provided 26, while sodium hydride-assisted methylation of the indole
nitrogen in 4a yielded 27 in good yield.
A series of compounds exploring different alkoxy substituents on the indole 5-position (32a-
c) were prepared according to the approach summarized in Scheme 5. The versatile hydroxy
building block 29 was prepared from commercially available 5-benzyloxyindole-3-
acetonitrile (28) in good yield using a chemoselective hydrogenation method described by
Maegawa et al [15]. In this reaction, the catalytic activity of palladium on carbon was
partially suppressed by the aproptic solvent THF to effect O-debenzylation without reducing
the aliphatic nitrile. Alkylation of 29 gave the intermediates 30a-c. The displacement of 30c
with pyrrolidine provided 31c which led to more soluble compound 32c.
To install an ionizable basic centre tethered to the indole ring by a short methylene linker,
Molander's procedure was utilized [16]. Potassium bromotrifluoroborate (33) was stirred with
pyrrolidine at room temperature for 40 h to afford potassium pyrrolidin-1-yl-methyl-

ACCEPTED MANUSCRIPT
trifluoroborate (34). The ensuing Suzuki-Miyaura cross coupling with bromo/chloro-
substituted indole-3-acetonitriles 10e, 10i, 14c using palladium acetate, XPhos and cesium
carbonate proceeded smoothly to afford intermediates 35a-c which underwent Knoevenagel-
like condensation with 3a to furnish the more soluble analogues 36a-c.
Results and Discussion
Structure-activity relationships
Investigation of the isoquinoline ring of 4a was carried out with a series of quinoline and
isoquinoline analogues exemplified in Table 1. Transposing the nitrogen atom to other
regioisomeric positions on the fused bicyclic ring had a striking impact on both mitosis and
LC3-II induction activity. Simply moving the nitrogen round the ring by one position to
create the isomeric isoquinoline 4b abolished activity against both pH3 and LC3-II
biomarkers. This dramatic SAR likely indicates a highly specific interaction, possibly a
hydrogen bond acceptor, being made by the nitrogen of the 5-isoquinoline ring with a cellular
target, as opposed to a change in physicochemical properties reducing cell penetration by a
large margin. Similarly, shifting the nitrogen one position further to give the 8-quinoline 4e
or to the other ring giving 4-isoquinoline 4f and 4-quinoline 4d also abolished activity for
both pH3 and LC3-II biomarkers. On the other hand, the 5-quinoline 4c exhibited a notable
pH3-selective profile with strong anti-mitotic potency (more than 10-fold increase in pH3
levels) without LC3-II activity. Interestingly, introducing another nitrogen atom into the 5-
isoquinoline ring to afford the 2,7-naphthyridin-4-yl analogue 4g resulted in the loss of both
pH3 and LC3-II activities. MKlp2 inhibitor Paprotrain (1), which bears a smaller monocyclic
3
-pyridine ring, was inactive for both pathways, suggesting that the dual activities of 4a on
both markers is not due to MKlp2 inhibition. Evidently, the precise placement of the nitrogen

ACCEPTED MANUSCRIPT
atom(s) in the bicyclic aromatic system is absolutely crucial to enable the dual mode of action
of 4a.

ACCEPTED MANUSCRIPT
Table 1. Fold induction for pH3 and LC3 for isomers and analogues of the isoquinoline ring
tested at a concentration of 5 µM.
N
R
H
HN
Fold induction (pH3
Fold induction
Compound
R
a
a
(Ser10))
(LC3-II / LC3-I)
4
a
7.3
5.0
0.7
0.7
0.4
N
4
b
1.5
10.5
1.4
N
N
4
c
N
4
d
4
e
1.3
1.3
0.6
0.7
N
N
N
4
f
4
g
1.4
1.2
1.0
1.1
N
N
1

ACCEPTED MANUSCRIPT
a
The band relative intensity ratios were determined for pH3 (Ser10) to DMSO treated
control, as well as LC3-II to LC3-I as compared to DMSO treated control.
Next, we set out to modify the acrylonitrile bridge connecting the isoquinoline and indole
rings (Table 2). Reducing the central double bond to allow free rotation of the two aromatic
rings (26) abolished activity for both pH3 and LC3-II markers. This suggests that the central
double bond could be responsible for constraining the aromatic rings in a configuration that is
critical for both anti-mitotic and LC3-II induction functions. Removing the nitrile group in 4a
to afford a disubstituted alkene (21) retained strong LC3-II potency without activity for pH3.
When an amide group was employed as the linkage between the aromatic rings (24), activity
for both biomarkers was lost. Converting the nitrile group in 4a to a larger primary amide
moiety afforded 25 which was also not active at the concentration tested. Together, these
observations suggest that the nitrile group is a key structural feature required for anti-mitotic
activity whereas LC3-II potency is preserved with a simple alkene linker.

ACCEPTED MANUSCRIPT
Table 2. Fold induction for pH3 and LC3 for linker analogues of compound 4a tested at a
concentration of 5 µM.
N
X Y
HN
Fold induction (pH3
Fold induction
Compound
X-Y
a
a
(Ser10))
(LC3-II / LC3-I)
N
4
a
7.3
5.0
2
1
4
1.3
1.4
5.4
O
2
NH
0.4
O
H N
2
2
5
6
1.3
1.2
0.6
0.3
H
N
2
a
The band relative intensity ratios were determined for pH3 (Ser10) to DMSO treated
control, as well as LC3-II to LC3-I as compared to DMSO treated control.
Exploration of the effects of different substituents at all the available positions of the indole
ring was studied with a wide range of compounds (Table 3). Installing a methyl group on the
indole C-2 (10a) was unfavourable as both pH3 and LC3-II activities were eliminated.
Introducing a bromine atom at the 4-position also resulted in an inactive compound (10b). At

ACCEPTED MANUSCRIPT
the 5-position, addition of a nitro group (14a) afforded a very poorly soluble compound that
tested inactive for both pH3 and LC3-II markers. An acetamide moiety (10g) was not
tolerated either as it failed to increase the levels of the two biomarkers. Varying effects were
observed when different halogen atoms were introduced at the 5-position. Activity on both
fronts was abolished with fluoro substituted 10d. A small 2-fold increase in LC3-II/LC3-I
was seen with chloro (10e) while pH3 and LC3-II induction improved to 2-fold and 4-fold,
respectively, for 5-bromo compound 10f. However, when the bromo substituent was located
at positions 6 (10i) and 7 (14c), LC3-II activity was attenuated. Installing an electron
donating methoxy group at positions 5, 6 and 7 afforded analogues 10c, 10h and 14b,
respectively, with moderate dual activities in the range of 2.5 to 4.5-fold induction of both
biomarkers. Extending the 5-methoxy group into the longer methoxyethoxy side chain (32a)
led to the decline of pH3 activity while maintaining LC3-II activity. Bulkier dioxolane (32b)
suffered loss of activity against both pathways. Improving aqueous solubility was one of our
goals, hence we made significant efforts to install a basic centre where possible. Interestingly,
attaching a longer pyrrolidinopropyloxy side chain (32c), which we expected to be inactive
given the result from the similar sized 32b, demonstrated strong single activity for LC3-II.
The anti-mitotic pH3 activities were abrogated but excellent LC3-II potency (6-fold
induction) was achieved with this compound. Likewise, attaching a pendant basic
methylpyrrolidine moiety to the 5, 6 and 7-positions (36a, 36b and 36c, respectively), also
aimed at improving aqueous solubility, resulted in very similar LC3-II active but non-mitotic
profiles.

ACCEPTED MANUSCRIPT
Table 3. Fold induction for pH3 and LC3 for substitutions on the indole ring tested at a
concentration of 5 µM.
N
N
2
H
4
HN
R
5
7
6
Fold induction
Fold induction (LC3-
Compound
R
Position
a
a
(pH3 (Ser10))
II / LC3-I)
4
a
H
Me
Br
-
7.3
1.1
1.1
3.0
0.9
1.4
2.1
0.9
1.2
5.0
0.7
0.5
3.2
0.5
2.0
4.0
0.6
0.5
1
0a
0b
0c
0d
0e
0f
2
4
5
5
5
5
5
5
1
1
OMe
F
1
1
Cl
1
Br
1
4a
0g
NO
2
1
NHCOCH
3
O
O
3
2a
5
5
1.6
1.0
3.2
O
3
2b
0.7
O
O
O
N
3
2c
5
0.9
6.0
3
6a
5
6
1.0
3.3
2.6
N
1
0h
OMe
4.3

ACCEPTED MANUSCRIPT
Fold induction
Fold induction (LC3-
Compound
R
Position
a
a
(pH3 (Ser10))
II / LC3-I)
1
0i
Br
6
6
1.6
0.7
3
6b
4b
1.1
5.7
N
1
OCH
Br
3
7
7
3.9
2.7
2.2
1
4c
6c
1.4
3
7
0.8
3.7
N
a
The band relative intensity ratios were determined for pH3 (Ser10) to DMSO treated
control, as well as LC3-II to LC3-I as compared to DMSO treated control.
Methylating the indole nitrogen (27) led to a conspicuous loss of both pH3 and LC3
activities, suggesting that the indole N-H could be making a key hydrogen bond interaction
with its molecular target(s) responsible for its dual mode of action (Table 4). Efforts were
also put into exploring potential bioisosteres of indole such as indazole, 7-azaindole and 2-
benzimidazole in the hope of engendering compounds with improved potency. However,
these endeavours ultimately proved disappointing. No activity was observed with the 2-
benzimidazole analogue (6) while the indazole compound (20) showed only feeble LC3-II
activity. 7-Azaindole 14d was found to exhibit exceptionally poor solubility (poorly soluble
even in DMSO). This is most likely because of its flat overall conformation facilitating tight
packing within the crystal lattice enabling the formation of hydrogen bonds between the N-H
of one azaindole molecule and the N-7 of another.

ACCEPTED MANUSCRIPT
Table 4. Fold induction for pH3 and LC3 for other analogues of the indole ring tested at a
concentration of 5 µM.
N
N
R
H
Fold induction (pH3
Fold induction
Compound
R
a
a
(Ser10))
(LC3-II / LC3-I)
4
2
2
a
7
0
7.3
5.0
1.6
2.0
1.4
HN
1.1
1.4
1.8
1.1
N
N
HN
HN
14d
N
HN
N
6
1.2
a
The band relative intensity ratios were determined for pH3 (Ser10) to DMSO treated
control, as well as LC3-II to LC3-I as compared to DMSO treated control.

ACCEPTED MANUSCRIPT
Through this SAR study, compounds with dual activities on pH3 and LC3-II biomarkers
(10c, 10f, 10h, 14b), as well as compounds with single activity selective for LC3-II (21, 32a,
3
2c, 36a, 36b, 36c) and a compound with single activity selective for pH3 (4c) have been
obtained. Lead molecule 4a exhibits a high degree of structural specificity for exerting the
dual effects of mitosis inhibition and LC3-II induction. All of the structural features of 4a are
critical for its dual actions including the acrylonitrile linker, specific position of the
isoquinoline nitrogen and the indole ring with optional 5-position substitution. The SAR
landscape around the indole ring suggests a cliff for pH3 activity as substitution on the indole
generally led to a considerable drop in pH3 potency. In contrast, LC3-II potency appears to
be more tolerant of substitutions at the indole positions 5, 6 and 7 (Fig. 6). From this work we
conclude that 4a remains the most potent dual-acting compound identified thus far.
Fig. 6. Summary of key structure-activity relationship (SAR) findings.

ACCEPTED MANUSCRIPT
Evaluation of cell death in isogenic normal and transformed BJ fibroblasts
Compounds were classified into four groups based on the SAR analysis: Group 1 compounds,
which are active against both pH3 and LC3 markers; Group 2 compounds, which are active
against LC3 only; Group 3 compounds, which are active against pH3 only; and Group 4
compounds, which are inactive against both pH3 and LC3 markers. Selected compounds
from the various groups were evaluated for their effects on cell death in isogenic normal BJ
cells and transformed BJ cells (Fig. 7). Following treatment with each compound at 5 µM for
4
8 h, the percentage of cells in the sub-G1 phase, an indicator of the extent of cell death, was
determined. Compound 4a, a group 1 compound with strong dual activities, exhibited a
strong cell-killing effect in transformed cells but not in normal cells. This transformed cell
selectivity was not seen in compounds from group 2 or group 3. Group 3 compound 4c killed
both transformed and normal cells with less selectivity than 4a while group 2 and group 4
compounds did not kill either normal cells or transformed cells. This suggests that dual
activity against both pH3 and LC3 markers are required for inducing selective cancer cell
death.

ACCEPTED MANUSCRIPT
Fig. 7. Percentage of isogenic normal BJ cells and transformed BJ cells in the sub-G1 phase
following treatment with the respective compounds at 5 µM for 48 h.