Discovery of an Orally Selective Inhibitor of Signal Transducer and Activator of Transcription 3 Using Advanced Multiple Ligand Simultaneous Docking

Article abstract of DOI:10.1021/acs.jmedchem.6b01489

Targeting signal transducer and activator of transcription 3 (STAT3) is a potential anticancer strategy. However, STAT3 inhibitors with good selectivity and bioavailability are rare. The aim of this study was to discover selective direct STAT3 inhibitors with good druglikeness. By the advanced multiple ligand simultaneous docking (AMLSD) method, compound 9 was designed as an orally bioavailable STAT3 inhibitor that presented superior druggability and selectivity compared with other representative STAT3 inhibitors. 9 directly and selectively inhibited the pY705 site of STAT3 with an affinity (K_i) of 440 nM. The IC₅₀ of 9 for MDA-MB-231 breast cancer cells was 184-fold lower than its IC₅₀ for MCF-10A normal breast epithelial cells. 9 in vivo induced significant antitumor responses (better than gefitinib), and its therapeutic index should be over 100, indicating good safety of 9.

Full text of DOI:10.1021/acs.jmedchem.6b01489

Subscriber access provided by University of Newcastle, Australia
Article
Discovery of an Orally Selective Inhibitor of Signal
Transducer and Activator of Transcription 3 Using
Advanced Multiple Ligand Simultaneous Docking
Wenying Yu, Chenglong Li, Wenda Zhang, Yuanzheng Xia, Shanshan Li, Jiayuh Lin,
Keqin Yu, Mu Liu, Lei Yang, Jian-Guang Luo, Yijun Chen, Hongbin Sun, and Ling-Yi Kong
J. Med. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.jmedchem.6b01489 • Publication Date (Web): 28 Feb 2017
Downloaded from http://pubs.acs.org on February 28, 2017
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Journal of Medicinal Chemistry is published by the American Chemical Society.
1155 Sixteenth Street N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the
course of their duties.

Page 1 of 48
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Discovery of an Orally Selective Inhibitor of
Signal Transducer and Activator of
Transcription 3 Using Advanced Multiple
Ligand Simultaneous Docking
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
†‡
‡
‡
‡
§
Wenying Yu, Chenglong Li, Wenda Zhang, Yuanzheng Xia, Shanshan Li, Jia-yuh
Lin⊥,Keqin Yu#, Mu Liu,†‡ Lei Yang,‡ Jianguang Luo,‡ Yijun Chen,∥ Hongbin Sun@
and Lingyi Kong*,‡
†State Key Laboratory of Natural Medicines, ‡Department of Natural Medicinal
∥
Chemistry, Laboratory of Chemical Biology, @Center for Drug Discovery, China
Pharmaceutical University, 24 Tong Jia Xiang, Nanjing 210009, China.
§
Division of Medicinal Chemistry and Pharmacognosy, College of Pharmacy, The
Ohio State University, Columbus, Ohio 43210, United States
⊥Department of Biochemistry and Molecular Biology, School of Medicine, University
of Maryland, Baltimore, Maryland 21201, United States
#Water supply and Dramage, Nanchang Hangkong University, 696 Fenghe Ave. S.,
Nanchang 330046, China.
*Corresponding Author: Phone: +86-25-8327-1405; Fax: +86-25-8327-1405; E-mail:
lykong@cpu.edu.cn
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 2 of 48
1
2
3
4
5
6
7
8
9
ABSTRACT. Targeting signal transducer and activator of transcription 3 (STAT3) is
a potential anticancer strategy. However, STAT3 inhibitors with good selectivity and
bioavailability are rare. The aim of this study was to discover selective direct STAT3
inhibitors with good drug-likeness. By advanced multiple ligand simultaneous docking
(AMLSD) method, compound 9 was designed as an orally-bioavailable STAT3
inhibitor, which presented superior druggability and selectivity to other representative
STAT3 inhibitors. 9 directly and selectively inhibited the pY705 site of STAT3 with
an affinity (Ki) of 440 nM. The IC50 of 9 for breast cancer cell MDA-MB-231 was 184-
fold lower than its IC50 for normal breast epithelial cell MCF-10A. 9 in vivo induced
significant antitumor responses better than Gefitinib and its therapeutic index should be
over 100, indicating good safety of 9.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
INTRODUCTION. Constitutive activated Signal Transducers and Activators of
Transcription (STAT) 3 protein is frequently detected in many malignancies,1-3
representing one of the most promising anticancer targets.4, 5 The activation of JAK-
STAT signaling pathway can be initiated by the binding of cytokines, hormones, and
growth factors to the cell surface receptors. Activated receptors phosphorylate JAK
kinases which then phosphorylate STAT3 protein. Phosphorylated STAT3 (p-STAT3)
proteins form dimers which then translocate into nuclear and bind to DNA, causing cell
replication and cancers.6, 7 p-STAT3 can be dimerized by forming the salt bridges
between the R609 residue in the SH2 domain and the pY705 residue of the native
peptide. In the STAT3 SH2 domain, there are two “hot spots”, pY705 (also named
ACS Paragon Plus Environment

Page 3 of 48
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
pY+0) site and a nearby pY+X site.8 The pY705 site is related to the disruption of
STAT3 phosphorylation and dimerization of inhibitors, while the nearby pY+X site is
associated to the selectivity of STAT3 inhibitors. Therefore, compounds targeting the
pY705 and pY+X sites of STAT3 can selectively inhibit both STAT3 phosphorylation
and dimerization. In addition, phosphorylation of Serine727 (pS727) residue of STAT3
enhances transcriptional activation and acetylation of Lys685 (Ac-K685) residue is
crucial for methylation of tumor-suppressor gene promoters.9, 10 Thus, proximal sites,
pS727 and Ac-K685 sites, are also functional in the tumor cell proliferation.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
STAT family includes seven members, among which, STAT3 and STAT5 are
oncoproteins, but STAT1 plays opposite roles of STAT3 in carcinomas by triggering
anti-proliferative effect in tumorigenisis.3 STAT5 plays a critical role in lobuloalveolar
proliferation, whereas STAT3 regulates lobuloalveolar apoptosis during pregnancy.11
In addition, STAT3 is structurally most similar to STAT1 and STAT5. Hence,
designing compounds with highly selective inhibitory activity of STAT3 over STAT1
and STAT5 is difficult and critical for precision anti-cancer therapy.
Strategies directly deactivating STAT3 mainly focus on the disruption of STAT3
dimerization and phosphorylation. Several series of direct STAT3 inhibitors have been
discovered via both computational and experimental methods, represented by
phosphor-peptide mimics,12, 13 non-peptide small molecule inhibitors,7, 8, 14-16 metal-
based inhibitors,6 and natural-product like inhibitors.17, 18 Some of them have entered
clinical trials.19-23 However, there are none marketed STAT3 inhibitors, due to poor
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 4 of 48
1
2
3
4
5
6
7
8
9
selectivity, undesirable PK/PD properties or other reasons. Therefore, the improvement
of selectivity and druggability will substantially advance the discovery of STAT3
inhibitors as precision anticancer agents.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Computer-aided drug design is widely utilized for the efficient identification and
optimization of lead compounds.24-27 Aiming to design druggable STAT3 inhibitors,
multiple ligand simultaneous docking method was advanced in two aspects, i.e. the
fragment library design and linker design (Figure 1). Using AMLSD, we designed
compounds directly inhibiting both phosphorylation and dimerization of STAT3
protein offering an orally-available potent and selective STAT3 inhibitor, 9, which
presented superior druggability or selectivity to other representative non-peptide small
STAT3 inhibitors (Figure 2 and Table 1).7, 8, 14, 15, 19, 28, 29 For example, 9 has better
druggability than 14,7 which violates the Lipinski’s rules.
Figure 1, 2 & Table 1
RESULTS.
Design by AMLSD method. 1) Fragment libraries design. Drug scaffolds (MW<300)
were categorized into six fragment libraries according to their physical properties,
including polarity, aromaticity, bulkiness, charge, and acidity/alkalinity. 2) Screening
of fragment combinations (FCs). Naphthalene-5,8-dione-1-sulfonamide (fragment 1)
was paired with each fragment from the polar fragment libraries forming 963 FCs,
which were simultaneously docked to the crystal structure of STAT3 SH2 domain
(PDB: 1BG1). The resultant multi-fragment virtual screening was ranked by their
ACS Paragon Plus Environment

Page 5 of 48
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
binding energies (Table S1, Supporting Information (SI)), and the binding modes of the
25 top ranked FCs were shown in Figure 3. 3) Linker design. Possible linear linkers for
the selected FCs were designed by the measured distances. Since the measured
distances between two fragments were in the range of 3~5 Å in Table S1, the length of
linkers should be within the range of 3~5 Å, thus all the proper linear linkers were listed
in Figure 4. Suitable linkers for FCs were then designed based on two criteria, the
feasibility of synthesis and the superposition of the docking modes between the linked
molecules and the separate FCs. Finally, linker a1 could connect two fragments through
a C-N coupling reaction while the linked compounds could reposition to their FCs, thus
we chose a1 as an ideal linker to connect the FCs. 4) Synergistic hit selection. FCs were
linked and docked to the STAT3 protein. Their docking modes were repositioned to
that of the pre-linked FCs. If linked compound could reposition to its FCs, what’s more,
the original contacts between FCs and protein could be maintained, and even more good
contacts were formed after linkage, then we consider the FCs were synergistically
linked. Finally, ten out of 25 compounds were chosen with synergistic effects (Figure
5 & Table S2, (SI)), for example, the docking modes of the linked compound 9 could
be repositioned to the docking modes of its FCs (No.175) (Figure 6). Moreover,
additional intermolecular hydrogen bonds were formed between the linker of 9 and
Ser636 residue which enhanced the synergistic effects. 5) ADME properties. Drug-like
parameters including the ADME and toxicity properties of the hits were computed,
predicating all compounds with decent drug-like properties.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 6 of 48
1
2
3
4
5
Figure 3-6
6
7
8
9
In order to investigate the in silico selectivity of the 10 compounds over STAT1/3/5.
Compounds were also docked to the crystal structures of STAT1 (PDB: 1BF5) and
STAT5 proteins (PDB: 1Y1U).30 Compound 1 was eliminated since it was also docked
to STAT1 SH2 domain with free energy, -7.8 kcal/mol. Compounds 3, 4, 6 and 8 were
excluded since they were weakly bound to STAT5 SH2 domain. Limited by the
availability of the starting materials, compound 10 was not synthesized. Finally,
compounds 2, 5, 7 and 9 were synthesized for further biological tests.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Chemistry. Synthesis of Designed Compounds. Fragment 1 and aromatic amines
were catalyzed by Cu(OAc)2·H 2O in an acidic environment. Reaction system was
heated to 65℃ and terminated when 1 was consumed completely. The atom numbering
of 1 is shown in Scheme 1. The temperature of reaction was a key reason to cause the
cyclized by-products since the sulfonamide group can attack the nearby 1-ketone and
deplete a water molecule to form the cyclized products. A higher temperature can cause
a lower yield of the target compounds. For the synthesis of compound 9, aiming to
selectively react to the aromatic amine group of the reactant, the piperazine amine was
protected by the Cbz group initially and then deprotected by TFA.
Scheme 1
Cell Assays. In a preliminary in vitro evaluation of the designed compounds, 9
exhibited best potency than other compounds against a human basal-like breast cancer
cell line, MDA-MB-231 (Figure 7A). However, its IC50 for normal breast epithelial cell
ACS Paragon Plus Environment

Page 7 of 48
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
MCF-10A was 128.9μM, 184-fold higher than that for MDA-MB-231, 0.7μM. So 9
was highly selectively targeting the cancer cells. 14 was selected as the positive control
since it is a representative orally available STAT3 inhibitor. Compound 9 was more
potent than 14 to kill multiple types of cancer cells, i.e. UW426, UW288-1, BKPC3,
MDA-MB-231 and U2OS cell lines (Figure 7B).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 7
Compound 9 is a Direct STAT3 Inhibitor. 1) 9 was predicated as a direct STAT3
SH2 domain inhibitor by both docking and molecular dynamics (MD) simulations. The
intermolecular interactions were revealed between compound 9 and STAT3. Based on
the analysis of the simulations, the sulfamine group of fragment 1 formed hydrogen
bonds with residues Arg609, Ser611, Ser612, and Ser613 in the pY705 site. In detail,
the 8-ketone group of fragment 1 as a major pharmacophore interacted with residues
Arg609 and Lys591. The linker amine group contacted with residues Ser636 and
Glu638 in the loop. The piperazinyl fragment formed ionic bond with residue Glu594
and formed hydrogen bond with Gln635 in pY+X site (Figure 8). The RMSDs of the
snapshots during the MD simulaltions also reconfirmed the stable binding between
compound 9 and STAT3 (Figure S2, (SI)).
Figure 8
2) The direct binding between 9 and STAT3 protein were demonstrated by both
Fluorescent Polarization (FP)-based competitive binding assay and Microscale
Thermophoresis (MST) assay. 9 was proven to preferentially bind to the pY705 site of
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 8 of 48
1
2
3
4
5
6
7
8
9
STAT3, competitively disrupting the binding of the native high-affinity, fluorescence-
labeled phosphorylated peptide, 5-FAM-SpYLPQTV, with a Ki value of 0.44 μM
(Figure 9A). While, there was no direct binding between compound 9 and STAT1/5
proteins. Thus, 9 was determined as a selective STAT3 inhibitor by FP assays.
MST is a recently broadly used technology for the interaction analysis of
biomolecules, which was used to confirm that 9 directly bound to the STAT3 protein
with a Ki value of 2.42 μM±0.26 μM. The molar ratio of ligand/protein was determined
as 1: 1 (Figure 9B).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 9
Selectivity of Compound 9. 1) 9 in silico selectively docked to the pY site of
STAT3 protein, distant from the pY sites of STAT1/5 proteins (Figure 10). 2) 9
selectively inhibited p-STAT3 at pY705 site without altering at S727 and K685 sites.
The level of p-STAT3 (Y705) was reduced in a dose-dependent manner (Figure 11A),
while the levels of acetylated STAT3 (K685) and p-STAT3 (S727) were not dose-
dependently influenced by 9 and the expression of total STAT3 was not changed,
indicating that 9 selectively decreased the level of pY705-STAT3 which was not due
to a constitutional decrease of total STAT3 expression (Figure 11B). 3) In western blot
(WB) assays, 9 selectively inhibited p-STAT3 (Y705) (Figure 11A) and IL-6 induced
phosphorylation of STAT3 (Figure 12A), without any inhibition on p-STAT1 (Y701)
(Figure 11B) and IFN-γ induced phosphorylation of STAT1 (Figure 12B). 4) By
electrophoretic mobility shift assay (EMSA), 9 in vitro preferentially inhibited STAT3
ACS Paragon Plus Environment

Page 9 of 48
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
homodimer DNA-binding activity in a time-dependent manner, over STAT5-STAT5
DNA-binding activity, indicating 9 was a selective STAT3 inhibitor. 9 weakly
decreased the level of STAT1 dimer DNA-binding activity, while no observable
inhibition by compound 9 of the IFN-γ induced STAT1-STAT1 DNA-binding (Figure
12C). Quantitative real-time RT-PCR showed that compound 9 decreased the
expression of STAT3-specific oncogene, MMP9, while the expression of STAT1-
specific gene, CDK2, treated with 9, was not significantly influenced (Figure 12D). 5)
The stimulation of STAT3 upstream targets, such as EGFR, can not only induce the
level of p-STAT3, but also induce the levels of p-AKT and p-ERK1/2 kinases. To
investigate whether compound 9 is selective to the JAK-STAT signaling pathway, we
also evaluated the levels of p-AKT and p-ERK1/2 kinases in bypass pathways. The
phosphorylation levels of both kinases were not inhibited by 9 either (Figure 11C &
Figure S8, (SI)). 9 showed better selectivity than 14 to p-AKT, since the level of p-
AKT was decreased by 14 while it was not affected by 9 in MDA-MB-231 cancer cell
line (Figure 11C).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
In all, compound 9 selectively inhibited pY705-STAT3 protein.
Figures 10-12
Compound 9 Induced the Level of SHP-1. Interestingly, 9 was found to induce the
expression of STAT3 phosphatase, SHP-1. Mechanism is unknown yet, however, the
effect may partially contribute to the low IC50 values of 9 (Figure 11B&D).
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 10 of 48
1
2
3
4
5
6
7
8
9
Compound 9 Induced Apoptosis in Malignant Cancers. The inhibition of STAT3
phosphorylation by 9 seemed to be consistent with the induction of apoptosis as
evidenced by the increase of cleaved caspase-3 in the MDA-MB-231 cancer cell line
(Figure 11C).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Compound 9 Inhibited the Nuclear Translocation of p-STAT3 to Nucleus. The
IL-6 induced level of p-STAT3 in the cytoplasm and nucleus treated with 9 was lower
than that treated with 14 by immunofluorescence staining, suggesting 9 blocked STAT3
nuclear translocation with a better effect than 14 (Figure 13).
Figure 13
In vivo Study of Compound 9. To investigate the safety profile of compound 9, the
acute toxicity of 9 was tested in ICR mice at doses of 200, 400, 600, 800 and 1000
mg/kg (ip, n = 10 per group). All the mice were vital and lively even treated at the
highest dose (1000 mg/kg), thus the median lethal dose (LD50) was unable to be
obtained. Since 70% tumor weight was shrank at 10 mg/kg 9 (Figure S9), the
therapeutic index of 9 should be over 100 (1000/10=100), indicating good safety of 9.
1) 9 orally inhibited the growth of cancer xenografts with better therapeutic effects
than first-line anticancer drug, Gefitinib, and with less toxicity than doxorubicin. The
nude mice tumor xenografts were injected subcutaneously with human breast cancer
MDA-MB-231 cells. 9 orally shrank 53% tumor weight at 2.5 mg/kg, 60% tumor
weight at 5 mg/kg and 70% tumor weight at 10 mg/kg, meanwhile for Gefitinib, dosage
at 40 mg/kg were needed to achieve similar therapeutic effects as 9 at 10 mg/kg,
ACS Paragon Plus Environment

Page 11 of 48
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
indicating 9 was more active than the first-line anticancer Drug, Gefitinib. (Figure 14
and Figure S9). The body weight, post-treatment of either Gefitinib or 9, increased
normally (Figure 14D and Figure S9D) and no damage of the major organs (heart, liver,
spleen, lung and kidney) was detected by H&E staining (Figure S9B).
Figure 14
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The IC50 of 9 was firstly evaluated as 0.42 µM in S180 cancer cell line. In the mouse
sarcoma tumor xenografts, 73% tumor weight was significantly (P < 0.001) shrank by
9 after a 21 days therapy (Figure 15A, C&E), showing much less side effect by 9 at 5
mg/kg than by Doxorubicin at 1 mg/kg. Dramatic loss of body weight was observed
from the 13th day treated by Doxorubicin; whereas for 9, the increase of body weight
was normal (Figure 15D&F). Additionally, the size of liver, spleen and kidney organs
treated by Doxorubicin was notably smaller than that treated by 9 (Figure 15B).
Figure 15
2) 9 in vivo disabled STAT3 activity. The in vivo inhibitory activity of 9 against the
overexpression level of p-STAT3 was studied by both immunohistochemical (IHC)
analysis and WB. From the IHC results, 9 showed dose-dependent inhibitory effects on
p-STAT3 (Figure 16). The results were confirmed by WB results, 9 dose-dependently
inhibited the level of p-STAT3 while showing little influence on the level of total-
STAT3 (Figure 17).
Figure 16 & 17
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 12 of 48
1
2
3
4
5
6
7
8
9
3) The oral bioavailability of compound 9 was measured as 44.7%±11.9%,
comparable to Gefitinib, 47.4%±0.9% by PK study. The pharmacokinetic parameters
of 9 were well comparable to Gefitinib too (Table 2). The change of plasma
concentrations in rats through 2 min to 24 h was shown in Figure 18, treated by 9 (i.v.
and p.o.). These results indicated that 9 as an orally bioavailable candidate. The method
was fully validated according to the FDA guidelines for selectivity, linearity, lower
limits of quantification (LLOQ), accuracy, precision, recovery, matrix effect and
stability (Tables S8-S11).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Figure 18 & Table 2
PAINS analysis. All designed compounds were filtered by the online filter,
http://zinc15.docking.org/patterns/home/, for Pan Assay Interference Compounds
(PAINS). Quinone group was detected, which is a common structure in natural
products. It is also a substructure for endogenous molecules, such as Vitamin K, and
some approved drugs, such as Mitomycin, Menadione and Idebenone. To test the
potential reactivity of the α,β-unsaturated carbonyl group as Michael acceptor site in
the quinone group, compound 9 and cysteamine were mixed in DMSO or water
solutions at 37℃ overnight, no reaction was found, monitored by TLC (see SI),
indicating that 9 wasn’t very reactive.
Experimental results proved good activity, selectivity and safety of compound 9. The
STAT3 siRNA transfection experiment demonstrated that the major antitumor target
for compound 9 was STAT3 and compound 9 is selective to STAT3 protein (see SI).
ACS Paragon Plus Environment

Page 13 of 48
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
For the cell based assays, both normal breast epithelial cell line, MCF-10A, and breast
cancer cell line, MDA-MB-231, were treated with compound 9. The IC50 for normal
cells, 128.9 μM, was 184-fold higher than that for cancer cells, 0.7 μM, indicating
compound 9 is selectively targeting the tumor cells with good safety. For the acute
toxicity study, mice were vital and lively, orally treated by even 1 g/kg compound 9,
indicating its therapeutic index should be over 100. The growth of the body weight of
mice was also not affected by compound 9, however the body weight of mice was
significantly decreased by Doxorubicin. No damage of the major organs (heart, liver,
spleen, lung and kidney) was detected treated by 9 by H&E staining (Figure 14B). The
size of liver, spleen and kidney organs treated by Doxorubicin was notably smaller than
that treated by 9 (Figure 15B) too. All the in vivo results indicated that compound 9 had
better safety than the first-line anticancer drug Doxorubicin. In addition, for the protein
based assays, compound 9 has been proved as the direct inhibitor of STAT3 protein.
Therefore, both the in vitro and in vivo results pointed to compound 9 as a
nonpromiscuous selective STAT3 inhibitor.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
DISCUSSION.
By AMLSD method, multiple fragments are simultaneously docked to one target,
which can cover the chemical spaces of binding sites better than one ligand hopping.
MLSD was advanced in fragment library design and evolving process. The fragment
library design is crucial for a successful drug design. In our study, fragments were
derived from the drug scaffolds aiming to design more druggable STAT3 inhibitors.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 14 of 48
1
2
3
4
5
6
7
8
9
Old drugs might “repurpose” for new uses.27 Building blocks of the clinical drugs share
many interesting properties, thus there is a higher chance to find drug candidates by
designing compounds derived from drug scaffolds.27 In addition, AMLSD determines
linkers reminiscent as the BRIGE program in LUDI.31 For example, different xylenes
were determined by distance measurement in our case. Since the intermolecular
interactions between fragments have been considered by AMLSD, there is a higher
chance that the linked compounds could be repositioned to their fragment combinations.
For previous application of MLSD method, fragment library was designed according
to the similarity of fragments to the scaffolds of known inhibitors. Then the selected
fragments are grouped as FCs, resulting a very limited number of FCs before docking.
The number of final linked compounds are small and determined by the variety of
known inhibitors. If drug targets were lack of known inhibitors, then it would be hard
to apply MLSD method to design compounds. AMLSD can overcome the above
problems. The selection of fragments is no longer relied on the scaffolds of known
inhibitors. Thus, AMLSD is more useful to design candidates for a target not so well-
studied.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Compound 9 was found to selectively bind to the p-STAT3(705) protein rather than
STAT1/5 proteins, p-AKT and p-ERK kinases in bypass pathways, and other proximal
sites. 9 also selectively targeted the MDA-MB-231 cancer cells by majorly disabling p-
STAT3 level. The IC50 of 9 for normal cells was 184-fold higher than that for breast
cancer cells. For the animal study, the growth of the body weight of mice was not
ACS Paragon Plus Environment

Page 15 of 48
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
affected by compound 9, which had much less toxicity than the first-line anticancer
drug Doxorubicin. The therapeutic index of 9 should be over 100. In protein, cell and
animal levels, 9 was proved with good selectivity and safety.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Compound 9 was found as a direct small molecule inhibitor (MW<500) of STAT3
with a very low Ki value, 440 nM. 9 also showed very low IC50 values in different
cancer cell lines. 9 had excellent aqueous solubility. The oral bioavailability of 9 was
44.7% by PK study, comparable to the first-line drug, Gefitinib, 47.4%. Thus, 9 was an
orally-bioavailable STAT3 inhibitor. All of its physic-chemical properties follow the
Lipinski’s rules of five. Comprehensive assessing 9 and other representative STAT3
inhibitors, 9 showed outstanding features among all. Further experiments are needed to
evaluate the potential of 9 to undertake clinical trials.
CONCLUSION. AMLSD was used to design new STAT3 inhibitors by screening
FCs built from the library of drug scaffolds. Among these compounds, 9 was proved to
directly and selectively bind to the pY705 site of STAT3 protein, which significantly
decreased the level of STAT3 phosphorylation/dimerization as a dual functional
inhibitor. 9 also significantly shrank the tumor weight of both breast cancer and
sarcoma cancer xenographs in vivo with better activity than first-line drug, Gefitinib,
and less toxicity than Doxorubicin. The therapeutic index of 9 should be over 100,
indicating good safety of 9.
EXPERIMENTAL SECTION
Design
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 16 of 48
1
2
3
4
5
6
7
8
9
Software. The design part was mainly performed in Dr. Chenglong Li’s lab in the
Ohio State University. MLSD software was written by Dr. Li’s lab.32 Schrödinger
packages was purchased from its commercial supplier and installed in English version.
Autodock4 was downloaded from its official website http://autodock.scripps.edu/ for
free and installed in English.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Binding sites analysis. There are two “hot spots” of STAT3 SH2 domain, pY705
and pY+X sites. According to the crystal structure of the pY+X site of STAT3 (PDB:
1BG1), several key polar residues were embedded inside and formed a polar binding
region. Based on the structural analysis of the crystallized STAT1, STAT3 and STAT5,
the pY705 site is highly conserved, however the pY+X site varies within this family of
proteins, which plays an important role in the selectivity of STAT inhibitors. We
combined fragment 1 with each fragment of the polar fragment libraries to perform the
hopping of FCs in the two adjacent pY705 and pY+X sites, purposing to design
selective inhibitors of STAT3.
Preparation of Fragment Libraries. The fragment libraries in our study were
generated from 1554 FDA approved drugs, 4959 experimental drugs, 84 nutraceutical
drugs and 521 investigational drugs collected by Drugbank database
(https://www.drugbank.ca/). Complete database of all drugs were downloaded and
prepared by LigPrep of Schrödinger using the default parameters.33 Fragments were
extracted from the drugs using scaffold decomposition program by Canvas of
Schrödinger using default parameters.22 Fragments were calculated with MW and Polar
ACS Paragon Plus Environment

Page 17 of 48
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
using physicochemical descriptors by Canvas of Schrödinger.34 All the scaffolds were
exported and the scaffolds with molecular weight higher than 300 were deleted. The
total charge of the fragments was calculated by ligfilter descriptors. Other properties
were divided manually. Fragments were reorganized using property filters by Canvas
of Schrödinger as six fragment libraries, including polar_uncharged_nonaromatic
fragment library, polar_uncharged_aromatic fragment library, polar_charged_basic
fragment library, polar_charged_acidic fragment library, large and bulk fragment
library, and nonpolar fragment library.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Screening FCs and the Evolving of FCs to Final Hits. Two fragments as one
fragment combination were grouped to probe the two “hot spots” of STAT3 SH2
domain, pY705 and pY+X sites. Fragment 1, which was the privileged structure of the
two most potent STAT3 SH2 inhibitors, 16 and 17,8, 35 formed two dipole-ionic
interactions with residue R609 in the pY705 site analyzed by the MD simulations
(Figure S3, (SI)). Thus, fragment 1 was chosen for the inhibition of the pY705 site.
While, fragments fitting to the pY+X site were screened from the newly designed
fragment libraries. Since the pY+X site of STAT3 has a key polar binding region,
therefore, fragments fitting to pY+X site were screened from the four polar fragment
libraries, a total of 963 fragments, including 280 polar uncharged nonaromatic
fragments, 532 polar uncharged aromatic fragments, 61 polar charged basic fragments,
and 90 polar charged acidic fragments. The PDB file of each fragment from the four
polar fragment libraries was grouped together with the PDB file of fragment 1 and
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 18 of 48
1
2
3
4
5
6
7
finally formatted as one PDB file, resulting 963 fragment combinations. Each FC
includes two fragments.
8
9
The FCs best fit for pY705 and pY+X sites were docked to the crystal structure of
STAT3 SH2 domain and scored by the PSO method,32 which were then ranked by their
docking energies.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
The evolving from fragments to one compound was guided by the docking modes of
the FCs, which was no longer determined by designers’ chemical intuition or
experience. The shortest distances between C6 or C7 of fragment 1 and the heavy atoms
of the other fragment were around 3~5 Å (Table S1, (SI)), measured by Maestro of
Schrödinger. For example, for compound 9, the shortest distance between C6 of
fragment 1 and the other 2-(piperazin-1-yl) aniline fragment were 3.671 Å (Figure 5).
Thus, the length of possible linear linkers should be within the range, listed in Figure
4. Every linker was tried to link the FCs, and the linker a1 was finally decided on the
synthetic feasibility and its repositioning ability. Linked compounds were repositioned
to the SH2 domain scored by Lamarckian Genetic Algorithm (LGA) of Autodock4 and
ranked by binding energies. Final hits were selected based on both the docking energies
and their repositioned docking modes, which were further synthesized and tested by all
kinds of biological assays.
The docking procedure involved the preparation of the ligand and macromolecule
using the Schrödinger software.33, 36 AutoDockTools were used to assign Gasteiger
charges to the ligands. AutoGrid maps were then calculated for all atom types in the
ACS Paragon Plus Environment

Page 19 of 48
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
ligand set. Grid box was centered in the SH2 domain of STAT proteins. After 10 million
energy evaluations, all the resulting conformations of the ligands in the binding pocket
of the macromolecule were clustered into groups according to their conformations with
a root-mean-square deviation threshold of 1.5 Å. The most significant low energy
clusters were identified and binding energies were evaluated.37
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
ADME Prediction. Designed compounds were prepared by LigPrep wizard in the
Schrödinger software. The ADME properties of the compounds were calculated using
QikProp wizard in the Schrödinger software, including whether (I) less than 5
maximum possible metabolites would be formed in vivo; (II) the logP value for
octanol/water was in the range of -2 to 2 indicating excellent aqueous solubility profile;
(III) predicted logIC50 values for hERG K+ channel blocking was below -5; (IV)
predicted Caco-2 and MDCK cell permeability values were acceptable; (V) predicted
index of binding to human serum albumin was well within the recommended range of
-1.5 to 1.5; (VI) predicted human oral absorption percentage was above 50%.
Chemistry.
Chemicals and Reagents. The solvents and reagents were purchased from
commercial suppliers and were used as received. Silica gel was purchased from Sigma-
Aldrich Chemical Co. (Milwaukee,WI). All the synthetic derivatives and compounds
purchased were obtained with >96.0% purity by HPLC (UV detection at 254 and 280
nm). Positive control 14 was purchased from commercial suppliers.
Synthesis of Fragment 1. 1 was synthesized as previously described.8
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 20 of 48
1
2
3
4
5
6
7
8
9
Synthesis of Designed Compounds. Fragment 1 was reacted with aromatic amines
as previously described,8,38 which were catalyzed by Cu(OAc)2·H 2O in an acidic
environment. Reaction system was heated to 65 ℃ and terminated when 1 was
consumed completely. Compounds 2, 5, 7 and 11 were synthesized.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Compound 2, 5, 11. Reported38
5-((1,4-dioxo-5-sulfamoyl-1,4-dihydronaphthalen-2-yl)amino)nicotinic acid (7).
Yield: 71%. 1H NMR (500 MHz, DMSO) δ 8.31 (s, 1H), 8.04 (d, J = 7.6 Hz, 1H), 7.98
(d, J = 2.5 Hz, 1H), 7.94 (d, J = 7.5 Hz, 1H), 7.78 (t, J = 7.5 Hz, 1H), 7.44 (s, 1H), 7.40
(s, 2H), 5.62 (s, 1H), 5.34 (s, 2H). HRMS (ESI) of C16H11N3O6SNa [M + Na]+ calcd,
396.0266; found, 396.0270.
5,8-dioxo-6-((2-(piperazin-1-yl)phenyl)amino)-5,8-dihydronaphthalene-1-
sulfonamide (9). 11 (0.15 mM, 80 mg) was dissolved in 5 mL TFA, heated to 50°C for
8 hours. As detected by TLC, after 11 had completely reacted, all the volatiles were
removed under reduced pressure. The resulting crude product was applied to a column
of silica gel. The column was eluted with CH2Cl2/MeOH (20:1 v/v), then further
purified by recrystallization. The solvent was removed under reduced pressure to give
product 9. Yield: 92%. 1H NMR (300 MHz, DMSO) δ 9.15 (s, 1H), 8.85 (m, 2H), 8.43
(d, J = 7.1 Hz, 1H), 8.09 (m, 2H), 7.63 – 7.05 (m, 4H), 5.80 (s, 1H), 3.60 (m, 4H), 3.14
(s, 5H). HRMS (ESI) of C20H21N4O4S [M + H]+ calcd, 413.1284; found, 413.1290.
Biological Assays and Biochemical Assays.
ACS Paragon Plus Environment

Page 21 of 48
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
Cell Cultures. Human medulloblastoma cell lines (UW426, UW288-1 and DAOY)
and human pancreatic cancer cell line (BkPC-3), were used as stored in the Center for
Childhood Cancer, the Research Institute at Nationwide Children’s Hospital, in 2014.
Human breast cancer cell lines (basal-like subtype MDA-MB-231 and luminal A
subtype MCF-7), human osteosarcoma cell line (U2OS), human normal breast
epithelial cell line (MCF-10A) and mouse sarcoma cell line (S180) were purchased
from ATCC in 2015. All cell lines were tested without mycoplasma contamination.
Human cancer cell lines, UW426, UW288-1, DAOY, BkPC-3, MDA-MB-231 and
MCF-7 were maintained in Dulbecco’s Modified Eagle Medium (DMEM)
supplemented with 10% penicillin/streptomycin FBS and stored in a humidified 37 °C
incubator with 5% CO2. Human bone osteosarcoma (U2OS), human breast cell line
MCF-10A and mouse sarcoma cell line (S180) were maintained in RPMI-1640 medium
supplemented with 10% FBS and antibiotics (100 U/mL of penicillin G and 100 μg/mL
of streptomycin) and stored in a humidified 37 °C incubator with 5% CO2.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Chemicals. Compounds were dissolved in sterile DMSO to make 20 mM stock
solutions. Aliquots of the stock solutions were stored at −20 °C.
Cell Viability Assay. Studies were performed as previously described.8 UW426,
UW288-1, BKPC3, MDA-MB-231, U2OS, S180 and MCF-10A cell lines were seeded
in 96-well plates at a density of 3000 cells per well. The cells were incubated at 37 °C
for a period of 24 h. For the preliminary assay, 2, 5, 7 and 9 (5 μM) were added in
triplicate to the plates in the presence of 10% FBS, 24h. 3-(4,5-Dimethylthiazolyl)-2,5-
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 22 of 48
1
2
3
4
5
6
7
8
9
diphenyltetrazoliumbromide (MTT) was added to evaluate cell viability. For the IC50
measurements, different concentrations of 9 (0.1-10 μM for cancer cells and 3.125-
400μM for MCF-10A), 14 (0.1-20 μM), and other compounds were added in triplicate
to the plates in the presence of 10% FBS. For S180 cell line, cells were only treated
with compound 9 (0.1-10 μM). MTT was added to evaluate cell viability. The
absorbance was read at 595 nm. IC50 values were determined using SigmaPlot 9.0
Software (SySTAT Software, Inc., San Jose, CA).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Fluorescent Polarization. Studies were performed as previously described.8,39 FP
assay was used to analyze the ability of 9 to inhibit phosphopeptide binding to the STAT
SH2 domain. STAT3 protein and peptides were purchased from Genscripts. STAT1/5
proteins were purchased from Abcam and their peptides were purchased from
Taopushengwu, Inc., Shanghai, China.
Microscale Thermophoresis. A titration series of up to 16 dilutions of the
inhibitor 9 was prepared. Here the concentration of the fluorescently labeled STAT3
protein was kept constant at 10-20 nM and the concentration of the titrant was varied.
Therefore 2-20 μM of STAT3 protein were labeled with a fluorescent dye using
Monolith NT™ Protein Labeling Kits. The labeling procedure and the subsequent
removal of free dye were performed within 45 min. In the dilution series the highest
concentration was chosen to be 20-fold higher than the expected Ki. 10 μL of the serial
dilution of the non-labeled molecule was mixed with 10 μL of the diluted fluorescently
ACS Paragon Plus Environment

Page 23 of 48
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
labeled molecule. Mixed samples were loaded into glass capillaries and the MST-
analysis was performed using the Monolith.NT115.
8
9
Western Blot Analysis. Studies were performed as previously reported.8 MDA-MB-
231 cells were treated with 9 (2.5-10μM) or 14 (2.5-5μM) or DMSO at 60−80%
confluence in the presence of 10% FBS for 12 h. UW426 cells were treated with 9 (1-
2.5 μM) or 11 (1-2.5 μM) or DMSO at 60−80% confluence in the presence of 10% FBS
for 24 h.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
IL-6/IFN-γ Induction of STAT3/1 Phosphorylation. MDA-MB-231 cancer cells
were seeded in 10 cm plates and allowed to adhere overnight. The following day, the
cells were serum-starved. The cells were then left untreated or treated with 9 (2.5-5 μM)
or DMSO. After 2 h, the untreated and treated cells were stimulated by IL-6/ IFN-γ (50
ng/mL). The cells were harvested after 30 min and analyzed by Western blot.
Quantitative real-time RT-PCR. Levels of mRNA expression were analyzed with
the RT-PCR assay in MDA-MB-231 cells. Total RNA was isolated using an EASYspin
Plus tissue/cell RNA extraction kit (Aidlab Biotechnologies Co. Ltd), which was
quantified at 260 nm. 1μg RNA was reverse transcribed to cDNA using the Transcriptor
First Strand cDNA Synthesis Kit (Roche Diagnostics, Basel, Switzerland). Thermal
cycling conditions included 95 °C initial denaturation for 5 min, followed by 40 cycles
of denaturation (10 s at 95 °C ), annealing (15 s at 60 °C ) and extension (15 s at 72 °C
with a single fluorescence measurement), a melt curve program (60–95 °C with a 0.11
°C /s heat increase and continuous fluorescence measurement) and a cooling step to 40
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 24 of 48
1
2
3
4
5
6
7
8
9
°C . The Δ cycle threshold method was used for the calculation of relative differences
in mRNA abundance with a LightCycler 480 (Roche Molecular Biochemicals,
Mannheim, Germany). Data were normalized to the expression of GAPDH. The RT-
PCR primers that were used in this study are listed in Table S4, (SI).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Immunofluorescence Staining. Studies were performed as previously described.40
Nuclear Extract Preparation and Gel Shift Assays. Nuclear extract preparations
and EMSA were carried out as previously described.41 MDA-MB-231 cells were
pretreated with the 9 at 5μm at different time spots ranging from 0-24 h and then
harvested for nuclear extract preparation for EMSA analysis at room temperature prior
to incubation with the biotin-labeled probe for 30 min before subjecting to EMSA
analysis. Serum-starved MDA-MB-231 cells were preincubated with 9 and then
stimulated by IFN-γ, 50 ng/mL, and then harvested for nuclear extraction.
Wound-Healing Assay for Migration. Assays were performed as previously
reported.42
In Vivo Tumor Studies. The experiments were carried out in accordance with the
guidelines issued by the Ethical Committee of China Pharmaceutical University
(SYXK 2012-0035).
Mouse Xenograft Tumor Model. MDA-MB-231 human breast cancer cells (5 x
106) were injected subcutaneously into the flank area of 6-week-old female athymic
SPF BALB/c nude mice which were purchased from Canvens (Changzhou, Jiangsu,
China). After tumor volume reaching about 100 mm3, mice were divided into five
ACS Paragon Plus Environment

Page 25 of 48
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
8
9
treatment groups consisting of eight mice per group (each mouse generated one
xenograft tumor): saline vehicle control, oral injection of 9 (2.5, 5 and 10 mg/kg or 6,
12, and 24 μmol/kg) and Gefitinib (40 mg/kg or 90μmol/kg). The inhibitors were
directly dissolved in saline. Gefitinib was formulated with 0.05% tween80 to enhance
solubility. Tumor growth was monitored by measuring length (L) and width (W) of the
tumor every other day with a caliper. The tumor volume was calculated according to
the formula: Tumor volume=0.5236×L×W2. The treatment lasted for 32 days. The
detailed dosage regimen is listed in Table S6, (SI). For dose 5 mg/kg, a repeated study
was performed. Mice were divided into two groups, consisting of ten mice per group:
saline vehicle control and oral injection of 9 (5 mg/kg).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
S180 mouse sarcoma cancer cells (3 x 106) were injected subcutaneously into the
flank area of 4-week-old male mice which were purchased from Canvens (Changzhou,
Jiangsu, China). On the second day, mice were divided into two treatment groups
consisting of ten mice per group (each mouse generated one xenograft tumor): saline
vehicle control, oral injection of 9 (5 mg/kg) and i.p. injection of Doxorubicin (1 mg/kg
or 1.8μmol/kg). Both compounds were directly dissolved in saline. Tumor growth was
determined by measuring length (L) and width (W) of the tumor every other day with
a caliper. The tumor volume was calculated according to the formula: Tumor
volume=0.5236×L×W2. The treatment lasted for 21 days. The detailed dosage regimen
is listed in Table S7, (SI).
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 26 of 48
1
2
3
4
5
6
7
8
9
H&E Staining of Mice Organs. The five major organs (heart, liver, spleen, lung and
kidney) of all groups were anatomized from the mice, which were then fresh fixed in
10% neutral-buffered formaldehyde for 48 h, embedded in paraffin, and sliced at 5 μm
thickness. The sections were stained with haematoxylin and eosin (H&E), and
examined by light microscopy.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
IHC of Tumor Tissues. Immunohistochemical staining was carried out as previously
described.43 The fixed tumor sections were incubated with antibodies against p-STAT3.
The DAB staining Kit (NanJing KeyGEN Biotechnology Co., Ltd., Nanjing, China)
were used to visualize positive cells according to the manufacturers’ recommendations.
The Acute Toxicity Test. Both genders of ICR mice (7 weeks, SLACCAS) were
randomized and treated intraperitoneally (i.p.) with a single dose of 9 at 200, 400, 600,
800 and 1000 mg/kg, respectively. Each group contained 10 mice, and the animals were
observed for abnormal behavior and mortality up to 2 weeks post treatment.
ASSOCIATED CONTENT
SUPPORTING INFORMATION
Supporting Information on design, biological assays, PAINS assay and methods,
method validation of pharmacokinetic study and molecular formula strings. This
material is available free of charge via the Internet at http://pubs.acs.org.
ACS Paragon Plus Environment

Page 27 of 48
Journal of Medicinal Chemistry
1
2
3
4
5
Accession Codes
6
7
8
9
STAT3β/DNA complex: Protein Data Bank ID 1BG1; STAT1/DNA complex:
Protein Data Bank ID 1BF1; STAT5A: Protein Data Bank ID 1Y1U.
Authors will release the atomic coordinates of STAT3-9-(FC of 9) complex, STAT1-9
complex, STAT5-9 complex and experimental data upon article publication.
AUTHOR INFORMATION
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Corresponding Author
*E-mail: cpu_lykong@126.com; Phone: +86-25-8327-1405.
Notes
The authors declare no competing financial interest.
ACKNOWLEDGMENT
We are grateful to Dr. Inhee Park, Dr. Sihui Long and Dr. Feng Zhang for paper
editing. This work was supported by National Natural Science Foundation of China
(Grant No. 81402791 and No. 81673298), Natural Science Foundation of the Jiangsu
Higher Education Institutions of China (Grant No. BK20140670), Program for
Changjiang Scholars and Innovative Research Team in University (IRT_15R63),
Project Funded by the Priority Academic Program Development of Jiangsu Higher
Education Institutions (PAPD), College Students Innovation Project for the R&D of
Novel Drugs (J1030830), Jiangsu Shuangchuang Talents for Oversea Ph.D.s.
ACS Paragon Plus Environment

Journal of Medicinal Chemistry
Page 28 of 48
1
2
3
4
5
ABBREVIATIONS USED
6
7
8
9
FBDD, fragment-based drug design; SH2 domain, Src Homology 2 domain; pY705,
phosphor-tyrosine 705; PK/PD, Pharmacokinetic/Pharmacodynamic; FAM, carboxy-
fluorescein; IL-6, Interleukin 6; AKT, Protein kinase B; H&E staining, Hematoxylin
and eosin staining; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; Ac-STAT3,
acetyl-STAT3; P-STAT3, phosphor-STAT3; T-STAT3, total-STAT3; FDA, Food and
Drug Administration.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
REFERENCES
1. Liu, C. Y.; Su, J. C.; Ni, M. H.; Tseng, L. M.; Chu, P. Y.; Wang, D. S.; Tai, W. T.;
Kao, Y. P.; Hung, M. H.; Shiau, C. W.; Chen, K. F. Obatoclax analog SC-2001 inhibits
STAT3 phosphorylation through enhancing SHP-1 expression and induces apoptosis in
human breast cancer cells. Breast Cancer Res. Treat. 2014, 146, 71-84.
2. Wei, C.; Ball, S.; Lin, L.; Liu, A.; Fuchs, J. R.; Li, P.; Li, C.; Lin, J. Two small
molecule compounds, LLL12 and FLLL32, exhibit potent inhibitory activity on STAT3
in human rhabdomyosarcoma cells. Int. J. Oncol. 2010, 38.
3. Chen, H.; Yang, Z.; Ding, C.; Xiong, A.; Wild, C.; Wang, L.; Ye, N.; Cai, G.;
Flores, R. M.; Ding, Y.; Shen, Q.; Zhou, J. Discovery of potent anticancer agent
HJC0416, an orally bioavailable small molecule inhibitor of signal transducer and
activator of transcription 3 (STAT3). Eur. J. Med. Chem. 2014, 82, 195-203.
ACS Paragon Plus Environment

Page 29 of 48
Journal of Medicinal Chemistry
1
2
3
4
5
6
7
4. Yue, P.; Turkson, J. Targeting STAT3 in cancer: how successful are we? Expert
Opin. Invest. Drugs 2009, 18, 45-56.
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
5. Bharadwaj, U.; Eckols, T. K.; Kolosov, M.; Kasembeli, M. M.; Adam, A.; Torres,
D.; Zhang, X.; Dobrolecki, L. E.; Wei, W.; Lewis, M. T.; Dave, B.; Chang, J. C.;
Landis, M. D.; Creighton, C. J.; Mancini, M. A.; Tweardy, D. J. Drug-repositioning
screening identified piperlongumine as a direct STAT3 inhibitor with potent activity
against breast cancer. Oncogene 2014, 1-14.
6. Ma, D. L.; Liu, L. J.; Leung, K. H.; Chen, Y. T.; Zhong, H. J.; Chan, D. S.; Wang,
H. M.; Leung, C. H. Antagonizing STAT3 dimerization with a rhodium(III) complex.
Angew. Chem. Int. Ed. Engl. 2014, 53, 9178-9182.
7. Zhang, X.; Yue, P.; Page, B. D.; Li, T.; Zhao, W.; Namanja, A. T.; Paladino, D.;
Zhao, J.; Chen, Y.; Gunning, P. T.; Turkson, J. Orally bioavailable small-molecule
inhibitor of transcription factor STAT3 regresses human breast and lung cancer
xenografts. Proc. Natl. Acad. Sci. U. S. A. 2012, 109, 9623-9628.
8. Yu, W.; Xiao, H.; Lin, J.; Li, C. Discovery of novel STAT3 small molecule
inhibitors via in silico site-directed fragment-based drug design. J. Med. Chem. 2013,
56, 4402-4412.
9. Schuringa, J. J.; Schepers, H.; Vellenga, E.; Kruijer, W. Ser727-dependent
transcriptional activation by association of p300 with STAT3 upon IL-6 stimulation.
FEBS Lett. 2001, 495, 71-76.
ACS Paragon Plus Environment