Discovery of orally active indirubin-3′-oxime derivatives as potent type 1 FLT3 inhibitors for acute myeloid leukemia

Article abstract of DOI:10.1016/j.ejmech.2020.112205

FMS-like receptor tyrosine kinase-3 (FLT3) is expressed on acute leukemia cells and is implicated in the survival, proliferation and differentiation of hematopoietic cells in most acute myeloid leukemia (AML) patients. Despite recent achievements in the development of FLT3-targeted small-molecule drugs, there are still unmet medical needs related to kinase selectivity and the progression of some mutant forms of FLT3. Herein, we describe the discovery of novel orally available type 1 FLT3 inhibitors from structure-activity relationship (SAR) studies for the optimization of indirubin derivatives with biological and pharmacokinetic profiles as potential therapeutic agents for AML. The SAR exploration provided important structural insights into the key substituents for potent inhibitory activities of FLT3 and in MV4-11 cells. The profile of the most optimized inhibitor (36) showed IC₅₀ values of 0.87 and 0.32 nM against FLT3 and FLT3/D835Y, respectively, along with potent inhibition against MV4-11 and FLT3/D835Y expressed MOLM14 cells with a GI₅₀ value of 1.0 and 1.87 nM, respectively. With the high oral bioavailability of 42.6%, compound 36 displayed significant in vivo antitumor activity by oral administration of 20 mg/kg once daily dosing schedule for 21 days in a mouse xenograft model. The molecular docking study of 36 in the homology model of the DFG-in conformation of FLT3 resulted in a reasonable binding mode in type 1 kinases similar to the reported type 1 FLT3 inhibitors Crenolanib and Gilteritinib.

Full text of DOI:10.1016/j.ejmech.2020.112205

Journal Pre-proof
Discovery of orally active indirubin-3′-oxime derivatives as potent type 1 FLT3
inhibitors for acute myeloid leukemia
Pyeonghwa Jeong, Yeongyu Moon, Je-Heon Lee, So-Deok Lee, Jiyeon Park,
Jungeun Lee, Jiheon Kim, Hyo Jeong Lee, Na Yoon Kim, Jungil Choi, Jeong Doo
Heo, Ji Eun Shin, Hyun Woo Park, Yoon-Gyoon Kim, Sun-Young Han, Yong-Chul
Kim
PII:
S0223-5234(20)30172-0
DOI:
https://doi.org/10.1016/j.ejmech.2020.112205
EJMECH 112205
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 9 September 2019
Revised Date: 3 March 2020
Accepted Date: 3 March 2020
Please cite this article as: P. Jeong, Y. Moon, J.-H. Lee, S.-D. Lee, J. Park, J. Lee, J. Kim, H.J. Lee, N.Y.
Kim, J. Choi, J.D. Heo, J.E. Shin, H.W. Park, Y.-G. Kim, S.-Y. Han, Y.-C. Kim, Discovery of orally active
indirubin-3′-oxime derivatives as potent type 1 FLT3 inhibitors for acute myeloid leukemia, European
Journal of Medicinal Chemistry (2020), doi: https://doi.org/10.1016/j.ejmech.2020.112205.
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 Masson SAS.

Discovery of Orally Active Indirubin-3′-oxime
Derivatives as Potent Type 1 FLT3 Inhibitors for
Acute Myeloid Leukemia
†
§
‡
‡
‡
Pyeonghwa Jeong, Yeongyu Moon, Je-Heon Lee, So-Deok Lee, Jiyeon Park, Jungeun
‡
‡
∥
⊥
§
§
Lee, Jiheon Kim, Hyo Jeong Lee, Na Yoon Kim, Jungil Choi, Jeong Doo Heo, Ji
∆
∆
⊥
*,∥
*,†,‡
Eun Shin, Hyun Woo Park, Yoon-Gyoon Kim, Sun-Young Han , Yong-Chul Kim
†
Biomedical Science and Engineering, Gwangju Institute of Science and Technology,
Gwangju, Republic of Korea
‡
School of Life Sciences, Gwangju Institute of Science and Technology, Gwangju, Republic
of Korea
⊥College of Pharmacy, Dankook University, Cheonan 330-714, Republic of Korea
§
Gyeongnam Department of Environmental Toxicology and Chemistry, Korea Institute of
Toxicology, Jinju, Gyeongsangnam-do 52834, Republic of Korea
∥College of Pharmacy and Research Institute of Pharmaceutical Sciences, Gyeongsang
National University, Jinju, Gyeongsangnam-do 52828
∆
Department of Biochemistry, College of Life Science and Biotechnology, Yonsei University,
Seoul 03722, Korea

Abstract
FMS-like receptor tyrosine kinase-3 (FLT3) is expressed on acute leukemia cells and is
implicated in the survival, proliferation and differentiation of hematopoietic cells in most
acute myeloid leukemia (AML) patients. Despite recent achievements in the development of
FLT3-targeted small-molecule drugs, there are still unmet medical needs related to kinase
selectivity and the progression of some mutant forms of FLT3. Herein, we describe the
discovery of novel orally available type 1 FLT3 inhibitors from structure-activity relationship
(SAR) studies for the optimization of indirubin derivatives with biological and
pharmacokinetic profiles as potential therapeutic agents for AML. The SAR exploration
provided important structural insights into the key substituents for potent inhibitory activities
of FLT3 and in MV4-11 cells. The profile of the most optimized inhibitor (36) showed IC50
values of 0.87 and 0.32 nM against FLT3 and FLT3/D835Y, respectively, along with potent
inhibition against MV4-11 and FLT3/D835Y expressed MOLM14 cells with a GI value of
5
0
1
.0 and 1.87 nM, respectively. With the high oral bioavailability of 42.6%, compound 36
displayed significant in vivo antitumor activity by oral administration of 20 mg/kg once daily
dosing schedule for 21 days in a mouse xenograft model. The molecular docking study of 36
in the homology model of the DFG-in conformation of FLT3 resulted in a reasonable binding
mode in type 1 kinases similar to the reported type 1 FLT3 inhibitors Crenolanib and
Gilteritinib.

Introduction
FMS-like receptor tyrosine kinase-3 (FLT3), a receptor tyrosine kinase (RTK) in the type
III RTK family, is expressed on most acute leukemia cells and normal hematopoietic stem
cells. The activation of FLT3 is initiated by the binding of the FLT3 ligand (FL), which is
expressed in stromal cells. Upon stimulation with FL, FLT3 activation induces dimerization
and conformational changes in the receptor as well as autophosphorylation of specific
tyrosine residues, consequently triggering cascade signaling, including Ras/Raf, PI3K, AKT,
STAT5, Grb2, SHP-2, SHIP, MAPK and ERK1/2. The downstream cascades from FLT3 can
be categorized into the PI3K/AKT, RAS/MAPK, and STAT5 signaling pathways. [1, 2]
Acute myeloid leukemia is a cancer of the myeloid line of blood cells, characterized by the
rapid growth of abnormal blood cells originating from bone marrow, interfering with normal
blood cells. [3] AML is the most common form of acute leukemia that generally occurs in the
elderly patient population and is responsible for half of pediatric leukemia mortality. [4, 5]
Recently, genomic sequencing analysis has identified that mutations commonly appearing in
the signaling genes led to FLT3, which has been consequently studied as an attractive
therapeutic target for AML. [6, 7] Two broad types of FLT3 mutations have been identified:
internal tandem duplication (ITD) mutations [8] and point mutations in the tyrosine kinase
domain (TKD), typically at the activation loop (AL) residue D835. [9]
In recent decades, a number of small molecule inhibitors have been reported for AML,
such as Sunitinib, [10] Midostaurin, [11] Sorafenib, [12] Tandutinib, [13] Quizartinib, [14]
Ponatinib, [15] Crenolanib, [16] and Gilteritinib [17] (Figure 1). However, most of these
inhibitors showed multikinase inhibitory activities that limit their potency and repositioning
for the treatment of AML with the FLT3 mutation.[11, 18, 19] Moreover, the occurrence of
drug resistance due to the FLT3/TKD mutation during chemotherapy represents a challenge

to successful FLT3 inhibition. [20] Therefore, the development of potent and selective FLT3
kinase inhibitors, especially those effective for resistant AML caused by mutations, is in high
demand to solve the unmet medical needs of AML monotherapy.
Indirubin core skeleton has been extensively studied for the development of effective
kinase inhibitors,[21] [22] such as CDK and GSK3b. Our group also previously reported
several indirubin-3`-oxime derivatives as potent FLT3[23] and CDK inhibitors[24]
depending on the substitutions at the 5 and 5` positions, which have indicated the versatile
utilities of the skeleton toward particular class of kinase inhibitors. Recently, we also reported
an indirubin derivative, compound 9 (Figure 1), which has potent inhibitory activity against
FLT3 (IC50 = 3 nM) and the proliferation of MV4-11 cells (GI50 = 1 nM) with limited
pharmacokinetic profiles. [25] Aiming the druggable physicochemical properties maintaining
FLT3 inhibitory activities, herein, we report more optimized 5,5`-substituted indirubin
analogs with additional pharmacophores at the 3`-oxime moiety starting from previously
reported derivatives. [24] The biological profiles of compounds along with the oral
bioavailability were evaluated, including FLT3 inhibition, anti-proliferative activity in the
AML cell line and in vivo anticancer efficacy in a xenograft animal model by oral
administration.

Figure 1. Chemical structures of FLT3 kinase inhibitors

RESULT AND DISCUSSION
Chemistry
The synthesis of indirubin 3`-oximes 16a, 17a-e, and 18a-b was performed following
the synthetic routes represented in Scheme 1.
Scheme 1. General Synthetic Pathway of the Preparation of Indirubin 3′-oximes
Reagents and conditions: (a) Na CO , methanol or methanol:water(2:1), RT, 2-3 h, 42-55%;
2
3
(b) Hydroxylamine hydrochloride, pyridine, reflux, 3 h, 60-70%;
As our group reported in the previous study, we have included the process of
making the final product in HCl salt form to solve the solubility problem of the indirubin
derivative. To synthesize the 5,5’-substituted indirubin 3-oxime derivatives, the 5-substituted
indoxyl-N,O-diacetates 10a-b were prepared from their corresponding 5-substituted
anthranilic acids.[24] For the 5-hydroxy analog, the cyclized compound 10a was obtained as
a triacetate form because the phenolic hydroxyl group was also acetylated under cyclization
conditions. The 5,5’-substituted-3-indirubin oxime derivatives 16a, 17a-e, and 18a-b were
synthesized from conjugate reactions of the 5-substituted indoxyl-N,O-diacetates 10a-b or 11
with various isatins (12a-e) under basic conditions to yield the 5,5′-substituted indirubin

derivatives 13a, 14a-e, and 15a-b, followed by the reaction of each of the latter with
hydroxylamine to convert the ketone group to an N-oxime.
Scheme 2. Synthetic of Compounds 20 and 24-29
Reagents and conditions: (a) 2-(boc-amino)ethyl bromide, K CO , DMF, RT, overnight; (b)
2
3
4
N HCl in 1,4-dioxane, DCM, 0 ℃ , 3 h; (c) 1,2-dibromoethane, Et N, DMF, RT, overnight;
3
(d) Various amines, DMF, 50 ℃ , overnight; then 4N HCl in 1,4-dioxane, THF, 0 ℃ , 30 min;
Scheme 2 presents the synthetic routes for compounds 20 and 24-50. Reaction of 5-
hydroxy-5-nitro indirubin 3-oxime (16a) with 2-(Boc-amino)ethyl bromide gave intermediate

compound 19, which was subjected to a sequence of reactions, including deprotection of the
Boc protecting group followed by HCl salt formation to yield compound 20. Compounds
2
1a-e, 22, and 23a-b were obtained using the O-alkylation process between 1,2-
dibromoethane and N-oxime compounds 16a, 17a-e, and 18a-b, respectively. The terminal
bromides of 21a-e, 22 and 23a-b were coupled with various heterocyclic compounds
containing amino moieties under basic conditions to afford the corresponding intermediates,
which were then treated with 4 N HCl in dioxane to yield the desired final compounds 24-49
in excellent yield.
Structure-Activity Relationships with Pharmacokinetic Studies
The biological activities of the synthesized indirubin derivatives were evaluated for
their inhibitory activities against FLT3 and the proliferation of MV4-11 cells, an FLT3-ITD
positive AML cell line using the WST assay protocol. In our previous report, [24] potent
CDK inhibitors with the indirubin skeleton were discovered, including compound 17 (IC =
5
0
1
.9 nM for CDK2), which was further identified as an inhibitor of FLT3 (IC = 23.4 nM),
50
showing moderate antiproliferative activity against MV4-11 cells (GI = 133 nM, Table 1).
5
0
[
23] We have also reported that the introduction of heterocyclic moieties on the 3`-oxime
position (R3) led to a significant increase in the potency against both FLT3 and MV4-11
cells. [25] Also, the docking analysis with a nitro group at the R2 position such as the case of
compound 24 resulted in the sufficient numbers of interactions in the ATP binding site of
FLT3. (See supporting information) Therefore, to improve the inhibitory potency of
compound 17 against FLT3, we introduced several functional groups with ethyl linkers at the
R3 position of compound 17 (20-26, Table 1).

Table 1. Biological Activities of compound 17, 20, 22 and 24-35 against FLT3 and MV4-11
Cells
IC50 (nM)
GI50 (nM)
Compound
R
1
R
2
R
3
a
b
FLT3
MV4;11
1
2
2
2
2
2
2
2
2
3
3
3
3
3
3
7
0
2
4
5
6
7
8
9
0
1
2
3
4
5
H
OH
OH
OH
OH
OH
OH
NO
2
23.4 ± 1.06
8.6 ± 1.22
324 ± 42.4
12.2 ± 1.64
133 ± 19.9
95.3 ± 10.3
470 ± 86.1
26.5 ± 11.1
2 2 2
CH CH NH ·HCl
NO2
NO2
NO2
NO2
NO2
NO2
NO2
NO2
F
2 2
CH CH Br
18.9 ± 3.53
50 ± 16.6
345 ± 93.8
100 ± 10.9
27.6 ± 2.77
31.2 ± 5.06
13.2 ± 0.87
15.8 ± 1.46
27.0 ± 2.58
11.6 ± 1.05
98.3 ± 17.6
45.0 ± 1.19
116 ± 1.11
3
10.2 ± 2.33
118 ± 28.1
7.01 ± 1.90
8.3 ± 1.73
31.5 ± 8.8
8.4 ± 2.93
115.5 ± 16.5
105.3 ± 43.3
633.9 ± 173
33
F
F
H
F
F
F
F
F
F
F
F
F
F
F
Lestaurtinib
a
IC = 50% inhibitory concentrations of FLT3 were obtained from dose-dependent response curves.
5
0
b
GI50 = 50% inhibitory concentrations of cell proliferation were obtained from dose-dependent
response curves. All biological data were obtained in triplicate and presented as the mean mean ±
standard error of the mean (SEM)

Substitutions to the R3 moiety with an amino ethyl moiety (20) resulted in more
potent inhibitory activity compared to compound 17 (IC = 8.60 nM vs 23.4 nM and GI =
5
0
50
9
5.3 nM vs 133 nM, respectively). In contrast, the corresponding bromo ethyl derivative (21)
of compound 20 showed a drastic loss in FLT3 inhibitory activity as well as MV4-11
antiproliferative efficacy (IC = 324 nM and GI = 470 nM, respectively), suggesting that
5
0
50
the basic properties at this position play an important role in optimizing the biological activity.
Thus, compounds 24 and 25 with piperazine and N-methyl piperazine moieties substituted for
the bromo moiety of compound 22, respectively, exhibited 10- to 20-fold enhanced potencies
for FLT3 and MV4-11 (IC = 12.2 nM, 18.9 nM and GI = 26.5 nM, 50 nM, respectively).
50
50
When the piperazine ring was replaced with a morpholine ring (26) at the R3 position, both
FLT3 inhibitory activity (IC = 345 nM) and MV4-11 antiproliferative activity (GI = 100
5
0
50
nM) significantly decreased, indicating that the terminal amine of the heterocyclic ring may
be beneficial.
Among the series of compounds 20-26, we evaluated the pharmacokinetic (PK)
properties of compounds 20 and 24 in mice by intravenous or oral administration. The
relevant pharmacokinetic parameters are described in Supporting Table 1. The results
demonstrated poor oral bioavailability (F = 0% and 0% for 20 and 24, respectively) and rapid
clearance (CLobv = 10,379 and 70.4 mL/min/kg, respectively) of compounds 20 and 24,
respectively. The poor PK profiles might be due to metabolically labile groups, such as
hydroxyl and nitro groups.
Next, to improve the FLT3 inhibitory activities and pharmacokinetic (PK) profiles,
the hydroxyl and nitro groups at the R1 and R2 positions was further modified to fluorine or
hydrogen groups based on our previous report, in which hydrogen and fluorine groups at the
R1 and R2 position showed beneficial effects for the FLT3 inhibition. [23] As the results,

compounds 27 and 28 displayed slightly decreased activity against FLT3 (IC = 27.6 nM and
50
3
1.2 nM, respectively). Compound 29, with a hydrogen at the R1 position, maintained its
inhibitory potency against FLT3 (IC = 13.2 nM). These results indicate that fluorine or
5
0
hydrogen the substituents at the R1 position along with nitro group at R2 position, may
minimally affect for the FLT3 inhibitory activity. Interestingly, compounds 27 and 29
displayed increased MV4-11 anti-proliferative efficacy.
In the evaluation of the PK profiles of compound 29 without the hydroxyl group at
the R1 position, disappointingly, poor oral bioavailability (F = 1.33%) and rapid clearance
(CLobv = 108 mL/min/kg) was observed. (Supporting Table 1) Therefore, it is expected that
the poor PK profiles might be due to the nitro group at R2 position.
To further improve the PK profile while maintaining the biological activity, we next
prepared 5,5`-difluoro indirubin derivatives 30-35 (Table 1). Among them, compounds 30
and 32 containing piperazine and 4-amino piperidine, respectively, at the R3 position showed
the best potency with a low nanomolar range of biological activities either in the FLT3 assay
or in MV4-11 cells (IC = 15.8 and 11.6 nM and GI = 8.30 and 8.47 nM, respectively).
5
0
50
Comparing this series of derivatives, the morpholine, piperidine, and pyrrolidine-containing
compounds (33, 34 and 35) displayed 2- to 7-fold less inhibitory activity against FLT3 than
the heterocyclic ring derivatives containing terminal nitrogens, such as piperazine, N-methyl
piperazine and 4-amino piperidine (30, 31, and 32, respectively).
Next, the PK profiles of 30 and 32 were determined in mice by intravenous or oral
administration (Supporting Table 2). Significant improvement, compared with 20, 24, and 29,
was achieved for oral bioavailability (F = 16.7 and 15.3%, respectively), with the suitable PK
parameters of half-life (T = 3.4 and 9.2 h, respectively) and clearance (CL = 28.1 and
1
/2
obv
2
31 mL/min/kg, respectively). Thus, fluoro substitutions for either the hydroxyl or nitro

groups at the R1 and R2 positions are tolerable for maintaining the biological activity, and
more importantly, granting improved oral PK profiles.

Table 2. Biological Activities of compound 36-49 against FLT3 and MV4-11 Cells
IC50 (nM)
GI50 (nM)
Compound
R
1
R
2
R
3
a
b
FLT3
MV4;11
3
3
3
3
4
4
4
4
4
4
4
4
4
4
6
7
8
9
0
1
2
3
4
5
6
7
8
9
H
H
H
H
H
H
H
H
F
F
0
.87 ± 0.16
1.0 ± 0.3
3.3 ± 1.27
9.47 ± 0.31
142 ± 7.07
257 ± 44.2
11.1 ± 0.37
F
11.9 ± 0.26
263.3 ± 43.1
212 ± 65.4
F
F
H
H
H
2
.7 ± 1.10
0.51 ± 0.016
54.8 ± 21.5
9.29 ± 2.77
20.9 ± 2.88
128 ± 22.5
41.3 ± 2.68
15.1 ± 3.18
138 ± 29.4
103.5 ± 18.0
22.0 ± 5.0
20.0 ± 8.3
17.0 ± 5.1
7.0 ± 3.0
H
H
H
H
H
H
H
H
OCH
3
3
12.4 ± 1.08
OCH
2.1 ± 1.14
OCF
OCF
3
3
40.2 ± 1.13
5.62 ± 1.05
Lestaurtinib
3
33
a
IC = 50% inhibitory concentrations of FLT3 were obtained from dose-dependent response curves.
5
0
b
GI50 = 50% inhibitory concentrations of cell proliferation were obtained from dose-dependent
response curves. All biological data were obtained in triplicate and presented as the mean mean ±
standard error of the mean (SEM)

Since the above results from compounds 30 and 32 provided clues to achieve oral
bioavailability, further optimization of the biological efficacy toward single-digit nanomolar
IC values was attempted by modification of the substituents at the R1 and R2 positions
5
0
(Table 2). Thus, the piperazine and N-methyl piperazine derivatives with no substitution at
the R1 position (36 and 37) showed a dramatic increase in the potency compared with the
corresponding fluoro analogs 30 and 31, with IC50 values of 0.87 and 3.3 nM for the
inhibitory activities of 36 and 37 against FLT3, respectively, along with an impressive
enhancement of MV4-11 anti-proliferation activity, with GI values of 1.0 and 11.1 nM,
5
0
respectively. In the case of compound 38 containing the ethyl(4-amino-piperidine) moiety at
the R3 position, the activities against FLT3 and MV4-11 compared to corresponding
compound 32 were similarly maintained (IC = 9.47 versus 11.6 nM and GI = 11.1 versus
5
0
50
8
.47 nM for 38 and 32, respectively). In contrast, the biological activities of compounds 39
and 40 with ethyl piperidine and ethyl pyrrolidine moieties at R3 position, respectively, were
significantly decreased for their inhibitory activities against FLT3 (IC = 142 nM and 257
5
0
nM, respectively) and anti-proliferative efficacies against MV4-11 (GI = 263 nM and 212
5
0
nM, respectively).
We further removed the fluoro groups at the R2 position of 36-40 resulting in
compounds 41-45, to compare the electronic effects between fluorine and hydrogen at the R2
position. Among them, compounds 41-43, containing piperazine, N-methyl piperazine and 4-
amino piperidine moieties at the R3 position, respectively, displayed generally decreased
inhibitory activities against FLT3 (IC = 2.7, 9.3 and 21 nM, respectively) and MV4-11 cells
5
0
(
GI = 0.51, 55 and 15 nM, respectively). Following similar activity profiles in the above
50
cases, piperidine and pyrrolidine derivatives 44 and 45 exhibited a large decrease in
inhibitory activity compared to 41.

Next, we examined the effect of another functional groups at the R2 position by
introducing methoxy or trifluoromethoxy moieties (46, 47, 48, and 49), which possess a
similar size but different electronic properties. In general, the methoxy derivatives 46 and 47
showed slightly more potent inhibitory activity against FLT3 compared to the corresponding
trifluoromethoxy analogs 48 and 49. Interestingly, unlike the SAR patterns of the previously
discussed compounds, the N-methyl piperazine-containing compounds 47 and 49 were more
potent than the corresponding piperazine derivatives 46 and 48.
Next, two of the most active compounds, 36 and 41, were subjected to PK profile
investigations, with an intravenous or oral administration of a 10 mg/kg dose of the
compounds in mice (Figure 2 and Supporting Table 3). The results showed that 36 had an
AUClast of 25.0 µg·min/mL, a Cmax of 36.5 ng/mL, and a remarkable increase in the oral
bioavailability of 42.6%. The PK study of compound 41 also achieved satisfactory results,
with an AUClast of 72.1 µg·min/mL, a Cmax of 69.7 ng/mL, and a bioavailability (F) of 51.0%.
As the consequences, we have successfully discovered orally administrable
compounds with more potent FLT3 inhibitory activities and MV4-11 antiproliferative effects
than the previously reported inhibitor with same skeleton, compound 9. [25] Especially, the
oral bioavailability (BA) of compounds 36 and 41 (BA = 42.6% and 51.0%, respectively)
was greatly improved compared with compound 9 (BA = 1.43%)

Figure 2. The mean plasma concentration-time profiles of 36 and 41. (A) Oral administration
to ICR mouse of 36 at dose of 10 mg/kg (n=5) (B) Intravenous administration to ICR mouse
of 36 at dose of 10 mg/kg (n=4) (C) Oral administration to ICR mouse of 41 at dose of 10
mg/kg (n=5) (D) Intravenous administration to ICR mouse of 41 at dose of 10 mg/kg (n=5).
All data were obtained in triplicate and the error bars represent the standard error of the
mean.
In conclusion, from the SAR analysis and PK optimization of the series of
derivatives in Tables 1 and 2, the ethyl piperazine moiety at the R3 position and the fluoro
moiety at the R2 position, such as the case of compound 36, are the best combination for the
highly potent FLT3 inhibitory activity along with MV4-11 anti-proliferative efficacy. In

addition, the PK profiles, including oral bioavailability, were critically dependent on the
substituents at the 5 and 5`-positions (R1 and R2). For example, the oral bioavailability
dramatically decreased depending on the substituent at the R1 and R2 positions in the
following order: -H, -F < -F, -F < -H, -NO < -OH, -NO , respectively.
2
2
Kinase Selectivity Assay
To investigate kinase selectivity profiles, we screened a single concentration of 100
nM of compound 36 against a Eurofins broad oncology panel of 104 kinases (Figure 3).

Figure 3. Kinase selectivity profiles of 36. (A) Kinase tree illustration of compound 36 using KinMap.
Kinase tree indicates the % inhibition value at 100 nM of compound 36 with concentration of ATP
near to the K value for each individual target kinases. The red circle indicates kinases that were
m
inhibited > 50%. Size of red circle is proportional to the % inhibition value. (B) Biological activities
of 36 against other kinases.
The kinome tree is illustrated in Figure 3A using KinMap, showing major selective
inhibitory activities at the tyrosine kinase family with more than 90% inhibition such as
FLT3/D835Y, FLT4, cSRC, Fms, Mer, Ret, TrkA, TrkB. [26] Remarkably, the inhibition of
the mutant FLT3/D835Y (0.32 nM) turned out to be more potent than that of the wild type
FLT3 (IC50 = 0.87 nM). Moreover, compound 36 displayed strong binding affinity at FLT3-
ITD with a K value of < 0.25 nM. In contrast, kinases PDGFRα and c-KIT (Type 3 RTK),
d

which are typically inhibited by current FLT3 inhibitor drugs, [13, 27, 28] were weakly
inhibited at 5% and 22%, respectively. In addition, compound 36 demonstrated selective
inhibitory activities for FLT3 over the CDKs (21%, 42% and 54% inhibition against CDK1,
CDK2 and CDK5 at 100 nM, respectively), even though indirubin derivatives have been
reported as well-known CDK inhibitors. [22, 29-31] Recently, c-KIT inhibition of TKIs was
reported as an important toxicological index associated with the disruption of hematopoietic
progenitor cells related to myelosuppression.[28] In our case, an important, focused
selectivity profile of weak c-KIT inhibition by compound 36 was observed, indicating that
the possible hematopoietic toxicity associated with c-KIT could be avoided. The current
FLT3-targeted drugs Quizartinib, Tandutinib, Crenolanib and Gilteritinib [32] have been
reported with different kinase selectivity profiles. The analysis of the kinase selectivity
profile of compound 36 in Figure 3 demonstrated that the selective inhibitory activities are
focused on the class of the tyrosine kinase family.
Cell Activity Evaluation of Compound 36
Since the FLT3 and FLT3(D835Y) inhibitory activities and the binding affinity for FLT3-
ITD of compound 36 were determined as sub-nanomolar range of IC and K values (0.87
5
0
d
nM 0.32 nM and < 0.25 nM, respectively), the anti-proliferative activities of compound 36
were assessed against several FLT3-ITD positive human AML cell line including MOLM14
wild type, other clinically relevant mutant kinase expressed MOLM14 FLT3-ITD-site direct
mutagenesis (MOLM14-ITD), MOLM14 FLT3-ITD- D835Y (MOLM14-ITD-D835Y) and
MOLM14 FLT3-ITD-F691L (MOLM14-ITD-F691L) cells that are FLT3 growth dependent
(Figure 4).

Figure 4. Anti-proliferative activities of compound 36 against MOLM14 mutant cells.
(A) Anti-proliferative activities of compound 36 against MOLM14 wild type cell line. (B)
Anti-proliferative activities of compound 36 against MOLM14-ITD site direct mutagenesis
cell line. (C) Anti-proliferative activities of compound 36 against MOLM14-ITD-D835Y cell
line. (D) Anti-proliferative activities of compound 36 against MOLM14-ITD-F691L cell line.

(
E) GI value of compound 36, Quizartinib and Gilteritinib against tested cell lines. All data
50
ware obtained by triplet testing. For A-D, the error bar = SD.
In the same experiments, positive control drugs, Quizartinib and Gilteritinib which
are representative type I and II inhibitor, respectively[33] were also tested in parallel. As the
results, compound 36 showed similar dose-dependent anti-proliferative activity against
MOLM14 wild type cell line with a GI values of 4.88 nM compared to Gilteritinib (GI =
5
0
50
4
.94 nM) (Figure 4A). In addition, compound 36 was more sensitive to MOLM14-ITD cells
(
GI = 1.85 nM) than MOLM14 wild type with similar trend of Gilteritinib (Figure 4B). In
5
0
the evaluation of the anti-proliferative effects against ITD-TKD dual mutant cell lines,
MOLM14(FLT3-ITD, D835Y and F691L), compound 36 and the representative type I
inhibitor, Gilteritinib, retained potent anti-proliferative effects against MOLM14 (FLT3-ITD,
D835Y) dual mutant cells (GI50 = 1.87 nM and 1.91 nM, respectively) whereas type II
inhibitor, Quizartinib, showed decreased inhibitory activity (GI = 6.72 nM) (Figure 4C).
5
0
For MOLM14(FLT3-ITD, F691L) cell line, compound 36 and Gilteritinib also exhibited
much higher anti-proliferative activities (GI = 3.27 nM and 5.24 nM, respectively) than that
5
0
of Quizartinib (GI = 13.58 nM) (Figure 4D).
5
0
Selectivity Profiles of Compound 36 in Various Human Cancer Cell Lines
To evaluate the selective anti-proliferative activities of the representative compound
3
6 in the FLT3-ITD expressing AML cells, MV4-11, we examined WST assay to measure
the cytotoxicity of compound 36 against several other cancer cell lines, including chronic
myeloid leukemia (K562), lung cancer (A549), liver cancer (HepG2), triple negative breast
cancer (MDA-MB-231), colon cancer (HCT-116), prostate cancer (PC3) and ovarian cancer

(SK-OV-3) compared with the positive control drug, Sorafenib and Gilteritinib. (Table 3) As
the results, compound 36 showed 430 times more sensitive against MV4-11 cells (GI = 1
5
0
nM) than K562 cell line, which is insensitive to FLT3 inhibitors similar profile to Gilteritinib.
Furthermore, compound 36 displayed good safety profiles against other cancer cell lines.
Table 3. Anti-proliferative Activities of Compound 36 Sorafenib and Gilteritinib against
Various Cancer Cell Lines
a
GI value (µM) or % inhibition at 1 µM
5
0
Cell line
3
6
Sorafenib
0.0009 ± 0.0001
0.612 ± 0.076
0 %
Gilteritinib
0.0009 ± 0.0001
1.006 ± 0.263
4.05 %
b
MV4-11
K562
0.001 ± 0.0003
0.437 ± 0.038
2.05 %
A549
HepG2
1.1 ± 0.2
8.95 %
2.5 ± 0.5
5.45 %
2.2 ± 0.6
MDA-MB-231
HCT-116
PC3
42.13 %
38.46 %
11.01 %
45.10 %
40.67 %
32.98 %
41.76 %
SK-OV-3
16.04 %
4.51 %
23.33 %
a
Compound 36, Sorafenib and Gilteritinib were tested at 8 concentration to measure their
anti-proliferative effects on various human cancer cell lines using WST assay. The GI50
b
values were obtained in triplicate and the error bars represent the standard error of the mean.
Reported GI value of Gilteritinib is 0.92 nM [34]
5
0

Compound 36 and 41 Inhibits the Phosphorylation of Proteins Downstream of FLT3
Compound 36, 41 and positive control Gilteritinib was further evaluated for its inhibitory
activity of the downstream signal transduction pathway of FLT3 activation in MV4-11 cells
at the concentrations of 1 µM, 100 nM and 10 nM (Figure 5).
Figure 5. Inhibition of STAT5 and Erk1/2 phosphorylation by compound 36, 41 and
Gilteritinib.
MV4-11 cells were treated with 10 nM, 100 nM and 1 µM of compound 36 for 4 hours.
MV4-11 lysates were subjected to western blot analysis with indicated antibodies. Actin was
used to probe for equal protein loading.
Compound 36 showed complete blockage of the phosphorylation of STAT5 even at the
lowest test concentration, 10 nM, indicating the comparable functional activity resulted from

the FLT3 inhibition. Similarly, the phosphorylation of Erk1/2 was also strongly suppressed
by compound 36. Based on these results, the inhibition of downstream signal pathway of
FLT3 autophosphorylation, occurring in FLT3-ITD expressing AML cell line, MV4-11, was
confirmed compared with Gilteritinib, of which activity was obtained as similar to the
reported data.[34]
Molecular Docking Study
To understand the binding mode of compound 36 in the kinase FLT3, we performed
homology modeling and molecular docking studies at the ATP binding site of FLT3 using the
Discovery Studio 3.5 program. Since the ligand-bound FLT3 cocrystal structures were
reported only for inactive conformations with the DFG-out motif of the protein, current type
1
FLT3 inhibitors and compound 36 were not able to fit in the binding site. Therefore, to
perform docking experiments for the type 1 inhibitors, homology protein models with DFG-
in motifs were constructed using a previously reported method by Yi-Yu Ke et al. [35]
Compound 36 along with the known FLT3 inhibitors Crenolanib and Gilteritinib were
docked into the above FLT3 homology model using the CDOCKER protocol in the
Discovery Studio 3.5 program.

Figure 6. Structural analysis of FLT3 inhibitors against the DFG-in FLT3 homology model.
The compounds 36 and 39 were protonated in docking study. (A) 2D diagram of docking
pose of 36. Hydrogen bonds are indicated by red-hatched lines. Ionic interaction is indicated
by brown-hatched line. Residues incorporating hydrophobic pocket in ATP binding site are
noted in green color. Residues incorporating polar surface are noted in brown color. (B) Best
docking pose of compound 36 into the homology model of DFG-in conformation of FLT3 (C)
Docking pose of compound 38. (D) Docking pose of compound 39. (E) Docking pose of
Crenolanib (E) Docking pose of Gilteritinib. In panel B to F, green and brown colors
represent classical hydrogen bonding and ionic interaction respectively.

The docking results showed that compound 36 successfully docked into the ATP binding
pocket, which was incorporated with hinge region and phosphate binding site of the FLT3.
The detailed docking pose analysis of compound 36 using a 2D diagram is displayed in
Figure 6A. There are two important hydrogen bonds formed between the backbone carbonyl
and amino groups of the Cys694 residue with the secondary amine and carbonyl groups of
the indirubin core scaffold of 36. In addition, the aromatic portions of 36 were tightly
surrounded by hydrophobic pockets consisting of the residues Leu616, Leu818, Cys828, and
Tyr693. Interestingly, no interaction of compound 36 with gatekeeper residue F691 was
observed, suggesting that 36 might be free from the concern of a mutation of F691. Another
important pharmacophore, the piperazine moiety of 36, may play important binding roles
with the hydrophilic surface of phosphate binding pocket consisting of residues Asn816 and
Asp829 of DFG motif. In particular, the tertiary amine of the piperazine moiety may form
ion-dipole or ionic interactions with the backbone carbonyl group of Asn816 and the side
chain of Asp698. In addition, the terminal amino group of piperazine may serve as a
hydrogen bond donor to interact with the side chain of Asn816 and side chain of Asp829.
(Figure 6B) In this binding model, the hydrogen bonding or ionic interactions of the nitrogen
atoms in the piperazine of compound 36 with Asp698, Asn816 and Asp829 residue would
play the critical roles for inhibitory activities from the analysis of the docking studies with
less active compounds 38 and 39. In the case of compound 38 (IC = 9.47 nM), the loss of
5
0
interaction with Asp698 while maintaining the hydrogen bonding and ionic interaction with
Asn816 and Asp829 residues, respectively, might result in the decreased inhibitory activity
compared with compound 36 (IC = 0.87 nM). The docking analysis with much weaker
5
0
derivative compound 39 (IC = 142 nM) suggested that the ethyl piperidine moiety of 39
5
0
headed to only Asp698 residue and solvent exposed area without the important interactions
with Asn816 and Asp 829 residues. (Figure 6D) Therefore, the prediction of important

binding mechanism in the regions accommodating R3 moieties could be explained with the
distinguished docking modes of the derivatives, 36, 38 and 39 with different biological
activities. The analysis and comparison of the docking poses of Crenolanib (Figure 6E) and
Gilteritinib (Figure 6F) with those of 36 (Figure 6B) resulted in the conformational
similarities of occupying the binding pocket. The piperazine moiety of compound 36, the 4-
amino piperidine moiety of Crenolanib, and the piperidinyl piperazine moiety of Gilteritinib
showed common interactions with the polar surface of the FLT3 protein. Protein
conformations, including the polar surface in several other kinases, are critical structural
features for determining type 1 inhibitors and type 2 inhibitors. It has been reported that the
ATP binding site containing the polar surface area is covered by the activation loop[35] in the
inactive kinase conformation. [36-38] Overall, compound 36 turned out to fit well in our
homology model of the active DFG-in conformation of FLT3, adopting the known structural
features as a type 1 inhibitor similar to Crenolanib and Gilteritinib.
In vivo Anti-tumor Activity of Compounds 36 and 41
Figure 7. Anticancer efficacy in MV4-11 xenograft mouse model of compound 36 and 41. (n
=
10 for control group, n = 6 for test group) BALB/c nude mice bearing MV4-11 tumor
xenograft were treated once daily with each compound at 20 mg/kg or vehicle. (A) Relative
tumor size (B) Body weight measurement of MV4-11 xenograft mice after oral
administration of compound 36 and 41. (***, p<0.001)