Structure-based virtual screening and optimization of modulators targeting Hsp90-Cdc37 interaction

Article abstract of DOI:10.1016/j.ejmech.2017.04.074

Identification of novel Hsp90 inhibitors to disrupt Hsp90-Cdc37 protein-protein interaction (PPI) could be an alternative strategy to achieve Hsp90 inhibition. In this paper, a series of small molecules targeting Hsp90-Cdc37 complex are addressed and characterized. The molecules' key characters are determined by utilizing a structure-based virtual screening workflow, derivatives synthesis, and biological evaluation. Structural optimization and structure–activity relationship (SAR) analysis were then carried out on the virtual hit of VS-8 with potent activity, which resulted in the discovery of compound 10 as a more potent regulator of Hsp90-Cdc37 interaction with a promising inhibitory effect (IC₅₀?=?27?μM), a moderate binding capacity (K_D?=?40?μM) and a preferable antiproliferative activity against several cancer lines including MCF-7, SKBR3 and A549?cell lines (IC₅₀?=?26?μM, 15?μM and 38?μM respectively). All the data suggest that compound 10 exhibits moderate inhibitory effect on Hsp90-Cdc37 and could be regard as a first evidence of a non-natural compound targeting Hsp90-Cdc37 PPI.

Full text of DOI:10.1016/j.ejmech.2017.04.074

Accepted Manuscript
Structure-based virtual screening and optimization of modulators targeting Hsp90-
Cdc37 interaction
Lei Wang, Li Li, Zi-Han Zhou, Zheng-Yu Jiang, Qi-Dong You, Xiao-Li Xu
PII:
S0223-5234(17)30356-2
10.1016/j.ejmech.2017.04.074
EJMECH 9425
DOI:
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 17 February 2017
Revised Date: 26 April 2017
Accepted Date: 29 April 2017
Please cite this article as: L. Wang, L. Li, Z.-H. Zhou, Z.-Y. Jiang, Q.-D. You, X.-L. Xu, Structure-based
virtual screening and optimization of modulators targeting Hsp90-Cdc37 interaction, European Journal of
Medicinal Chemistry (2017), doi: 10.1016/j.ejmech.2017.04.074.
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

Structure-based virtual screening and optimization of modulators targeting
ACCEPTED MANUSCRIPT
Hsp90-Cdc37 interaction
Lei Wanga,b, Li Lia,b, Zi-Han Zhouc, Zheng-Yu Jianga,b,Qi-Dong Youa,b,* and Xiao-Li Xua,b*
aState Key Laboratory of Natural Medicines, and Jiang Su Key Laboratory of Drug Design and Optimization,
China Pharmaceutical University, Nanjing 210009, China
bDepartment of Medicinal Chemistry, School of Pharmacy, China Pharmaceutical University, Nanjing
210009, China
c Nanjing Foreign Language School, Nanjing 210008, China
*
Corresponding author: Ph.D. Xiao-Li Xu
Address: China Pharmaceutical University, Nanjing 210009, China.
Tel/Fax: +86 025 83271216.
E-mail address: xuxiao_li@126.com
*
Corresponding author: Prof. Qi-Dong You
Address: China Pharmaceutical University, Nanjing 210009, China.
Tel/Fax: +86 025 83271351.
E-mail address: youqd@163.com

Abstract
ACCEPTED MANUSCRIPT
Identification of novel Hsp90 inhibitors to disrupt Hsp90-Cdc37 protein-protein
interaction (PPI) could be an alternative strategy to achieve Hsp90 inhibition. In this paper, a
series of small molecules targeting Hsp90-Cdc37 complex are addressed and characterized.
The molecules’ key characters are determined by utilizing a structure-based virtual screening
workflow, derivatives synthesis, and biological evaluation. Structural optimization and
structure–activity relationship (SAR) analysis were then carried out on the virtual hit of VS-8
with potent activity, which resulted in the discovery of compound 10 as a more potent
regulator of Hsp90-Cdc37 interaction with a promising inhibitory effect (IC50 = 27 µM), a
moderate binding capacity (KD = 40 µM) and a preferable antiproliferative activity against
several cancer lines including MCF-7, SKBR3 and A549 cell lines (IC50 = 26 µM, 15 µM
and 38 µM respectively). All the data suggest that compound 10 exhibits moderate inhibitory
effect on Hsp90-Cdc37 and could be regard as a first evidence of a non-natural compound
targeting Hsp90-Cdc37 PPI.

1. Introduction
ACCEPTED MANUSCRIPT
Heat shock protein 90 (Hsp90) has been recognized as a crucial role for cell cycle
control, protein assembly, kinase folding, cell survival and many signaling pathways [1-3].
Targeting Hsp90 to inhibit its biological function is a therapeutic strategy to treat cancer, and
to date, over 20 distinct N-terminal Hsp90 inhibitors have entered into clinical trials [4].
These inhibitors are composed of geldanamycin (GA) analogues, resorcinaol analogues,
purine analogues and other synthetic inhibitors that all of them target the Hsp90 ATP pocket
[5]. However, several severe side-effects have been reported in some clinical trials and none
of them have been approved by the Food and Drug Administration (FDA) [6-8].
ATP-competitive inhibitors could induce heat shock responses by upregulating the
multifarious stress inducible heat shock proteins, such as Hsp70, which is inevitable and
ultimately contribute the insensitivity and resistance of Hsp90 inhibitors on tumor cells [8].
The Hsp90 chaperone machine works through a multi-cycle process [9]. Simply inhibiting
the ATP hydrolysis of Hsp90 will inevitably impair some regular client proteins including
tyrosine kinases, signal transduction proteins, cell-cycle regulatory proteins, anti-apoptotic
proteins and telomerases [2]. The toxicity and heat shock response induced by Hsp90 ATPase
inhibitors are possibly because of the partial damage of those proteins.
During the maturation of client proteins of Hsp90, ATP hydrolysis is only one step of
this complicated procedures. Multiple co-chaperones, such as Hsp70, Hsp40, HOP, Cdc37,
and Aha1, are indispensable of each other [10, 11]. Protein kinases are a significant part of
Hsp90 chaperone client proteins and play crucial roles in the occurrence and development of
tumor growth [12]. Among them, more than 50% of the client proteins need the
co-chaperone Cdc37 to regulate their correct folding, transformation and maturation [13].
Cdc37, known as protein kinases-specific Hsp90 co-chaperone, plays an essential role in

loading protein kinases to Hsp90 chaperone machine, leading to natural maturation [14]. In
ACCEPTED MANUSCRIPT
addition to classic ATP-competitive Hsp90 inhibitors, disrupting Hsp90-Cdc37 PPI will
affect certain protein kinases rather than all the client proteins of Hsp90 which definitely
exhibit higher specificity and lower potential toxicity and avoid the side effects and [15].
This is as an alternative and promising strategy to break through the barriers of classic Hsp90
ATPase inhibitors currently in clinical trials.
At present the rational development of Hsp90-Cdc37 have been challenging due to the
complexity of protein-protein interaction. Hsp90-Cdc37 PPI have been reported all exhibited
a low micromolar binding ability in vitro [16, 17]. Nevertheless, the X-ray crystal structure
and NMR spectroscopy of the mid-domain of Cdc37 and N-terminal of Hsp90 were
presented in 2004 and 2009 respectively, providing a great foundation for development of
small molecular inhibitors to disrupt Hsp90-Cdc37 PPI [17-19]. In 2008, one of the potent
ingredients from the extracts of Celastranceae plants, Celastrol, was reported to exhibit
potency of Hsp90-Cdc37 disruption by binding to a charged pocket of Hsp90 N-terminal
[20]. However, another result they put forward in 2009 demonstrated that celastrol prevented
Hsp90 C-terminal from trypsin digestion by a direct binding mechanism through a
fingerprinting assay [21]. Thus, celastrol might reveal its potency by multiple regulation
mechanisms. Compared to celastrol, other reported modulators including withaferin A (WA)
[22], FW-04-806 [23], apigenin [24], sulforaphane [25] and kongensin A (KA) showed
similar low activities. As shown in Fig. 1, all of these modulators are natural products
showing the absence of target specificity, certainty in modulation mechanism, specific
binding site and co-crystal structures, leading to no ideal lead compounds for future
modification and development. Thus, small molecules targeting the Hsp90-Cdc37 PPI
interface with high affinity and specificity are urgently required.

The most common method to obtain PPI modulators is initiated from the peptides
ACCEPTED MANUSCRIPT
directly derived from the PPI binding interface [26]. According to our previous study [27],
we utilized the potential peptides derived from Cdc37 to explore the detailed binding site of
Hsp90-Cdc37 for the discovery of small molecules. In this paper, we report our workflow of
discovery and identification of small molecules targeting Hsp90-Cdc37 PPI by a virtual
screening-based method combined with biological validation. Five hundred thousand
compounds in Specs and NCI databases were screened by pharmacophore and cross-docking
filtrations. Thirty-one compounds were obtained and purchased for further identification in
in vitro assays. Finally, VS-8 (specs ID: AN-329/40718894) revealed weak activity in vitro
(IC50 values: 77 µM) and exhibited anti-proliferative potency against several cancer lines
including MCF-7, SKBR3 and A549 cell lines. To enhance the potency and explore the
structure relationship of VS-8, 16 derivatives were designed and synthesized to reveal
compound 10, which exhibited more promising inhibitory effect (IC50 = 27 µM), better
binding capacity (Kd = 40 µM) and preferable antiproliferative activity against MCF-7,
SKBR3 and A549 cell lines (IC50 = 26 µM, 15 µM and 38 µM respectively) compared to
VS-8. All the data suggest that compound 10 exhibited moderate inhibitory effect on
Hsp90-Cdc37 and could be regard as a first evidence of a non-natural compound targeting
Hsp90-Cdc37 PPI.

ACCEPTED MANUSCRIPT
Figure 1. 2D structures of the reported modulators targeting Hsp90-Cdc37 PPI.
2. Materials and Methods
2.1 Compounds and Reagents
Positive compounds including celastrol (Selleck, TX, USA), GA-FITC (BML-E1361,
Enzo Life Sciences, USA, FITC labelled geldanamycin, used as a probe molecule in
fluorescence polarization competitive assay) and 17-DMAG (Selleck, TX, USA) were
o
dissolved in DMSO (sigma, St. Louis, MO, USA) and stored at -20 C in a final
concentration of 10 mM before use.
2.2 Computational methods
2.2.1 Pharmacophore generation
The basic pharmacophore model was generated from the crystal structure of

Hsp90-Cdc37 (PDB: 1US7). To obtain detailed information of the ligand-receptor binding
ACCEPTED MANUSCRIPT
mode, a Cdc37 derived peptides docking model was utilized to construct the final
pharmacophore [27]. Receptor-Ligand Pharmacophore Generation protocol in Accelrys
Discovery Studio (DS) 3.0 (Accelrys Inc., San Diego, CA, USA) was utilized to construct
the pharmacophore model. Finally, based on hot-spots and binding mode analysis, we
manually selected the indispensable pharmacophore elements that contain two hydrogen
bond donors (HBD), one hydrogen bond acceptor (HBA), one hydrophobic feature and one
positive ionizable center.
2.2.2 Pharmacophore and docking screening
The chemical databases (Specs and NCI database) with 500,000 molecules were
screened using the ensemble pharmacophore model by ligand pharmacophore mapping
module in DS 3.0. Considering the complexity of the pharmacophore model and harsh
criteria for eligible ligands, one feature can be omitted in the ligand-pharmacophore mapping
procedure. All the molecules which fitted at least four pharmacophores (1532 remained)
were aligned in DS 3.0 and processed the cross-docking procedure. These ligands were
docked into Hsp90 protein structure by Libdock, Flexdock, CDOCKER and Glod5.0 (DS
3.0). The root-mean-square deviation (RMSD) was utilized to compare the differences
between the docked poses and co-crystallized pose to measure docking reality. Finally, 31
compounds with preferable docking scores and binding models were picked out for
purchasing list and further biological validation.
2.3 Chemistry
All reagents in this paper were purchased from commercial sources. Rotary evaporator
(BüchiRotavapor) below 50 °C under reduced pressure was used to concentrate all organic
solutions. Melting points were determined using a Melt-Temp II apparatus. Bruker AV-300

1
instrument was utilized to measure all H NMR spectra. Deuterated solvents were used to
ACCEPTED MANUSCRIPT
dissolve all compounds with tetramethylsilane (TMS) as internal standard. Shimadzu
GCMS-2010 instruments were used to obtain EI-MS and Water Q-Tofmicro mass
spectrometer was used to obtain ESI-mass and high resolution mass spectra (HRMS). HPLC
was performed on Agilent C18 column (4.6 mm × 150 mm, 3.5 µm) to determine the purity
of all compounds.
2.3.1 General procedure for preparation of 2a-e.
Malonyl chloride (2a). Malonic acid (1.4 g, 10 mmol) was directly dissolved in thionyl
chloride (10 mL) and then heated to reflux for 2 h. The excess thionyl chloride was removed
by rotary evaporator below 50 °C under reduced pressure. The crude product was obtained
and directly used in the next step. 2b-2e were obtain using the similar procedures of 2a
2.3.2 General method for the preparation of 5f-5o.
4-amino-N-(5-methylisoxazol-3-yl)benzenesulfonamide
(5f).
3-amino-5-methylisoxazole (1.18 g, 12 mmol) was added to the solution of THF (20 mL)
and pyridine (1 mL). The reaction mixture was stirred at room temperature for 6 h after
4-acetamidobenzenesulfonyl chloride (2.34 g, 10 mmol) was added to the reaction solution.
After the completion of the reaction monitored by TLC, the solvent was remove. Then, 30
mL of 0.5 M hydrochloric acid was added and a lot of precipitation was precipitated for
filtration. The crude solid was washed with water and directly used in the next step. All the
intermediates were dissolved in 20% NaOH solution and the reaction mixtures were heated
to reflux in an oil bath for 2 h. Then the reaction solution was quenched with 2 M
hydrochloric acid to pH 7.4, resulting a lot of white solid. The crude product was obtained
and washed with 5*10 mL water, then dried overnight in a vacuum desiccator, yielding 5f as
o
a white solid, yield 86%; mp 168-172 C.1H NMR (300 MHz, DMSO, δ) 7.55 (d, 2H, J =

7.08 Hz), 6.78-6.82 (d, 2H, J = 8.56 Hz), 6.35 (s, 1H), 6.05 (s, 2H), 2.36 (s, 3H). HRMS
ACCEPTED MANUSCRIPT
(ESI): found 239.2656 (C10H10N2O3S, [M + H]+, requires 239.2641). HPLC (80:20
methanol:water with 1‰ TFA): tR= 3.86 min, 98.1%.
5g to 5o were obtained using the similar procedures of 5f.
2.3.2 General synthesis procedure of 1-16.
N1,N3-bis(4-(N-(5-methylisoxazol-3-yl)sulfamoyl)phenyl)malonamide (1). 5f (253 mg,
1 mmol) was dissolved in the solution of THF (10 mL) and pyridine (1 mL) before 2a (240
mg, 2 mmol) was dropwise added. Then the reaction mixture was stirred at room temperature
overnight until the white solid was precipitated. The crude product was filtered and
recrystallized with ethanol to give a 1 as a light-yellow solid, yield 72%; mp 213-216 oC, 1H
NMR (300 MHz, DMSO, δ) 10.58 (s, 2H), 7.79 (d, 8H, J = 3.01Hz), 6.10 (s, 2H), 3.55 (s,
2H), 2.28 (s, 6H). HRMS (ESI): found 573.0863 (C23H22N6O8S2, [M-H]-, requires 573.0868).
HPLC (80:20 methanol:water with 1‰ TFA): tR= 4.64 min, 96.91%.
N1,N4-bis(4-(N-(5-methylisoxazol-3-yl)sulfamoyl)phenyl)succinamide (2). The same
o
1
procedure as 1, white solid, yield 83%; mp 263-265 C, H NMR (300 MHz, DMSO, δ)
11.30 (s, 2H), 10.46 (s, 2H), 7.79 (s, 8H), 6.13 (s, 2H), 2.72 (s, 4H), 2.31 (s, 6H). HRMS
(ESI): found 611.0262 (C24H24N6O8S2, [M+Na]+, requires 611.0091). HPLC (80:20
methanol:water with 1‰ TFA): tR= 4.12 min, 99.91%.
N1,N6-bis(4-(N-(5-methylisoxazol-3-yl)sulfamoyl)phenyl)adipamide (3). The same
o
procedure as 1, white solid, yield 89%; mp 266-267 C, 1H NMR (300 MHz, DMSO, δ)
11.31 (s, 2H), 10.32 (s, 2H), 7.76 (s, 8H), 6.11 (s, 2H), 2.36 (s, 4H), 2.28 (s, 6H), 1.61 (s, 4H).
HRMS (ESI): found 639.1452 (C26H28N6O8S2, [M+Na]+, requires 625.1426). HPLC (80:20
methanol:water with 1‰ TFA): tR= 4.25 min, 98.95%.
N1,N4-bis(4-(N-(5-methylisoxazol-3-yl)sulfamoyl)phenyl)terephthalamide (4). The

o
same procedure as 1, yellow solid, yield 65%; mp 223-225 C, 1H NMR (300 MHz, DMSO,
ACCEPTED MANUSCRIPT
δ) 11.31 (s, 2H), 10.59 (s, 2H), 8.34 (s, 2H), 7.82-7.67 (m, 8H), 7.37 (d, 2H, J = 6.04Hz)
6.12 (s, 2H), 2.29 (s, 6H). HRMS (ESI): found 659.6504 (C28H24N6O8S2, [M+Na]+, requires
659.6511). HPLC (80:20 methanol:water with 1‰ TFA): tR= 4.21 min, 98.25%.
N1,N3-bis(4-(N-(4,5-dimethylisoxazol-3-yl)sulfamoyl)phenyl)malonamide (5). The
o
same procedure as 1, light yellow solid, yield 64%; mp 230-234 C, 1H NMR (300 MHz,
DMSO, δ) 10.94 (s, 2H), 10.64 (s, 2H), 7.79 (d, 4H, J = 8.86Hz), 7.71 (d, 4H, J = 8.82Hz),
3.58 (s, 2H), 2.08 (s, 12H). HRMS (ESI): found 625.6431 (C25H26N6O8S2, [M+Na]+, requires
625.6423). HPLC (80:20 methanol:water with 1‰ TFA): tR= 4.18 min, 99.82%.
N1,N3-bis(4-(N-(1-methyl-1H-pyrazol-5-yl)sulfamoyl)phenyl)malonamide (6). The
same procedure as 1, white solid, yield 75%; mp 277-279 oC, 1H NMR (300 MHz, DMSO, δ)
10.86 (s, 2H), 7.82 (d, 4H, J = 8.71Hz), 7.67 (d, 4H, J = 9.01Hz), 7.26 (d, 2H, J =1.82Hz),
5.60 (d, 2H, J =1.84Hz), 3.61 (s, 2H), 3.56 (s, 6H). HRMS (ESI): found 595.1326
(C23H24N8O6S2, [M+Na]+, requires 595.1321). HPLC (80:20 methanol:water with 1‰ TFA):
tR= 3.98 min, 99.91%.
N1,N3-bis(4-(N-phenylsulfamoyl)phenyl)malonamide (7). The same procedure as 1,
o
white solid, yield 89%; mp 202-204 C, 1H NMR (300 MHz, DMSO, δ) 10.54 (s, 2H), 7.69
(s, 8H), 7.21-7.18 (m, 4H), 7.07-7.04 (m, 4H), 7.00 (s, 2H), 3.50 (s, 2H). HRMS (ESI):
found 587.1136 (C27H24N4O8S2, [M+Na]+, requires 587.1129). HPLC (80:20 methanol:water
with 1‰ TFA): tR= 4.66 min, 99.85%.
N1,N4-bis(4-(N-(4,5-dimethylisoxazol-3-yl)sulfamoyl)phenyl)succinamide (8). The
same procedure as 1, white solid, yield 92%; mp 247-249 oC, 1H NMR (300 MHz, DMSO, δ)
10.91 (s, 2H) 10.48 (s, 2H), 7.78 (d, 4H, J = 8.98Hz), 7.68 (d, 4H, J = 8.91Hz) 2.72 (s, 4H),
2.07 (s, 6H), 1.61 (s, 6H). HRMS (ESI): found 639.1303 (C26H24N6O6S2, [M+Na]+, requires

639.1302). HPLC (80:20 methanol:water with 1‰ TFA): tR= 4.22 min, 99.14%.
ACCEPTED MANUSCRIPT
N1,N4-bis(4-(N-(thiazol-2-yl)sulfamoyl)phenyl)succinamide (9). The same procedure
as 1, pink solid, yield 86%; mp>300 oC, 1H NMR (300 MHz, DMSO, δ) 12.59 (s, 2H), 10.27
(s, 2H), 7.62 (s, 8H), 7.14 (d, 2H, J = 4.62Hz), 6.71 (d, 2H, J = 4.62Hz), 2.41-2.39 (m, 4H).
HRMS (ESI): found 615.6814 (C22H20N6O6S4, [M+Na]+, requires 615.6810). HPLC (80:20
methanol:water with 1‰ TFA): tR= 3.82 min, 99.85%.
N1,N4-bis(4-(N-(1-methyl-1H-pyrazol-5-yl)sulfamoyl)phenyl)succinamide (10). The
o
same procedure as 1, white solid, yield 88%; mp>300 C, 1H NMR (300 MHz, DMSO, δ)
10.48 (s, 2H), 10.18 (s, 2H), 7.80 (d, 4H, J = 8.40Hz), 7.66 (d, 4H, J = 8.76Hz), 7.28 (d, 2H,
J = 1.65Hz), 5.62 (d, 2H, J = 4.08Hz), 3.58 (d, 6H, J = 4.62Hz), 2.74 (s, 4H). HRMS (ESI):
found 609.6428 (C24H26N8O6S2, [M+Na]+, requires 609.6419). HPLC (80:20 methanol:water
with 1‰ TFA): tR= 4.72 min, 99.67%.
N1,N4-bis(4-(N-phenylsulfamoyl)phenyl)succinamide (11). The same procedure as 1,
white solid, yield 88%; mp 280-281 oC, 1H NMR (300 MHz, DMSO, δ) 10.29 (s, 2H), 10.05
(s, 2H), 7.58 (s, 8H), 7.11 (s, 5H), 6.97 (s, 5H), 2.56 (s, 2H), 2.41 (s, 2H). HRMS (ESI):
found 601.1205 (C28H26N4O6S2, [M+Na]+, requires 601.1186). HPLC (80:20 methanol:water
with 1‰ TFA): tR= 3.95 min, 99.96%.
N1,N4-bis(4-(N-(pyridin-2-yl)sulfamoyl)phenyl)succinamide (12). The same procedure
as 1, light yellow solid, yield 88%; mp 280-281 oC, 1H NMR (300 MHz, DMSO, δ) 10.35 (s,
2H), 8.00 (d, 2H, J = 6.84Hz), 7.78 (d, 4H, J = 8.88Hz), 7.70-7.65 (m, 6H), 7.10 (d, 2H, J =
8.58Hz), 6.87-6.83 (m, 2H), 2.66 (s, 4H). HRMS (ESI): found 603.1112 (C26H24N6O6S2,
[M+Na]+, requires 603.1091). HPLC (80:20 methanol:water with 1‰ TFA): tR= 4.08 min,
99.63%.
N1,N4-bis(4-(N-(4-methylpyrimidin-2-yl)sulfamoyl)phenyl)succinamide (13). The

same procedure as 1, white solid, yield 81%; mp 254-255 oC, 1H NMR (300 MHz, DMSO, δ)
ACCEPTED MANUSCRIPT
11.49 (s, 2H), 10.30 (s, 2H), 8.21 (d, 2H, J = 5.10Hz), 7.80 (d, 2H, J = 8.85Hz), 7.63 (d, 2H,
J = 8.85Hz) 6.80-6.77 (m, 2H), 2.41-2.40 (m, 4H), 2.20 (s, 6H). HRMS (ESI): found
633.6648 (C26H26N8O6S2, [M+Na]+, requires 633.6639). HPLC (80:20 methanol:water with 1‰
TFA): tR= 3.82 min, 96.19%.
N1,N4-bis(4-(N-(pyrimidin-2-yl)sulfamoyl)phenyl)succinamide (14). The same
o
procedure as 1, white solid, yield 81%; mp 254-255 C, 1H NMR (300 MHz, DMSO, δ)
11.58 (s, 2H), 10.31 (s, 2H), 8.61 (d, 4H, J = 8.85Hz), 7.80 (d, 4H, J = 8.85Hz), 7.57-7.50 (m,
4H), 6.95-6.92 (m, 2H), 2.40 (s, 4H). HRMS (ESI): found 605.6122 (C24H22N8O6S2,
[M+Na]+, requires 605.6127). HPLC (80:20 methanol:water with 1‰ TFA): tR= 4.10 min,
99.92%.
N1,N4-bis(4-(N-(6-methoxypyridazin-3-yl)sulfamoyl)phenyl)succinamide (15). The
same procedure as 1, white solid, yield 78%; mp 270-271 oC, 1H NMR (300 MHz, DMSO, δ)
10.26 (s, 2H), 7.87-7.60 (m, 10H), 7.24 (d, 2H, J = 9.63Hz), 3.74 (s, 6H), 2.59 (s, 4H).
HRMS (ESI): found 665.1214 (C26H26N8O8S2, [M+Na]+, requires 665.1207). HPLC (80:20
methanol:water with 1‰ TFA): tR= 3.81 min, 99.89%.
N1,N4-bis(4-((2-(diaminomethylene)hydrazinyl)sulfonyl)phenyl)succinamide (16). The
o
same procedure as 1, yellow solid, yield 65%; mp 251-253 C, 1H NMR (300 MHz, DMSO,
δ) 8.79 (t, 1H, J = 1.44Hz), 7.88-7.83 (m, 1H), 7.36-7.31 (m, 3H), 6.61-6.51 (m, 8H), 2.41(t,
4H, J = 1.77Hz). HRMS (ESI): found 563.5724 (C18H24N10O6S2, [M+Na]+, requires
563.5729). HPLC (80:20 methanol:water with 1‰ TFA): tR= 4.10 min, 96.32%.
2.4 Biology
2.4.1 Expression and purification of Hsp90 and Cdc37.

The protein expression and purification assays were carried out as previously described
ACCEPTED MANUSCRIPT
[27].
2.4.2 Homogeneous Time-Resolved Fluorescence (HTRF)assay
For each HTRF assay, 4 µL of His-tagged full-length Hsp90 and 4 µL of GST-tagged
Cdc37 M were mixed into 384-well black low volume plate (Greiner#784076) in PBS buffer
(containing additional 400 mM KF, pH=7.4) with a final concentration of 250 nM,
respectively. Subsequently, 4 µL of a different gradient of diluted compounds (ranging from
300 µM to 1.23 µM with three times dilution) were added into the plate for 1 h incubation at
37 °C. Another 0.5 h incubation at 37 °C was performed after adding 4 µL of
anti-GST-Cryptate (61GSTKLA, CISBIO) and 4 µL of anti-6His-XL665 (61HISXLA,
CISBIO) detection reagent to reach a final 20 µL testing system. Molecular Device
(SpectraMax paradigm; λex = 320 nm, λem = 665 nm and 620 nm) was carried out to
measure the time-resolved fluorescence values. The final HTRF ratio was obtained by a
two-wavelength signal proportion calculated as following equation: (Signal 665 nm/Signal
620 nm) ×10000. Negative control (without addition of compounds) and blank control
(detection reagent only) were performed for each test. All data were analyzed by GraphPad
Prism 5.0 software.
2.4.3 ATPase inhibition assay
Hsp90 ATPase inhibition assay was performed using Discover RX ADP Hunter™ Plus
Assay Kit (Discovery, Fremont. CA). Following the instructions, constant concentration of
compounds (diluted to a final concentration of 100 µM with buffer containing 15 mM
HEPES, 20 mM NaCl, 1 mM EDTA, 10 mM MgCl2 and 0.02% Tween20, pH = 7.4) and
positive control (celastrol and 17-DMAG) were tested in 384-well black plates at 37 °C.
Each well was filled with 20 µL diluted compound, 20 µL Hsp90 protein (5 µM) and 20 µL

ATP (100 µM) for 1 h incubation. 10 µL detection reagent A and 20 µL detection reagent B
ACCEPTED MANUSCRIPT
was added subsequently and incubated at room temperature for 30 mins before final
determination. To examine the ADP generation, a Varioskan multimode microplate
spectrophotometer (Thermo Scientific Varioskan Flash, 540 nm excitation and 620 nm
emission) was used. Background values were measured in a condition without protein and
ligands, while negative control was determined in a solution without ligand and was
regarded as 100% protein activity.
2.4.4 Fluorescence polarization (FP) competitive assay
Generally, the FP experiments was performed as previously reported [28].The FP
experiments were carried out using the excitation and emission filters. 20 µL Hsp90 protein
and 20 µL GA-FITC probe mixture (100 mM Tris at pH 7.4, 20 mM KCl, 6 mM MgCl2, 5
mg/ml BSA, 100 nM Hsp90, 10 nM GA-FITC) were added to 384-well black plates
combined with 20 µL diluted compounds (constant concentration: 100 µM). After 30 min
incubation at room temperature in the dark, fluorescence polarization values were measured
on a SpectraMax multimode microplate reader (Molecular Devices SpectraMax Paradigm;
Excitation wavelength 485 nm, Emission wavelength 535 nm). The final inhibition ratio was
calculated using GraphPad Prism 5.0. The inhibition of the inhibitors were determined by
equation %inhibition = 1- (Pobs - Pmin)/(Pmax - Pmin). Pmax refers to the polarization values of
the wells containing protein Hsp90 and the GA-FITC probe. Pmin refers to the polarization
values of free GA-FITC probe and Pobs refers to the wells simply containing the inhibitors of
a certain concentration under the assay conditions.
2.4.5 Biolayer interferometry (BLI) assay
General experiment procedures were described previously [29]. The binding capacity
between the compounds and the Hsp90 protein was determined by biolayer interferometry

using an Octet Red 96 instrument (ForteBio Inc.). All the assays were performed on Corning
ACCEPTED MANUSCRIPT
96-well black plate and data were collected at 25 °C. Before the protein was immobilized
onto the Aminopropylsilane (APS, ForteBio Inc., Menlo Park, CA) biosensors, all the tips
were placed in the buffer of 0.1 M PBS to establish a pre-baseline. Next, the experiments
were composed of following steps: (1) baseline 1, sensors transferred to the 0.1 M PBS
buffer for 120 s equilibration; (2) loading, sensor tips were moved to Hsp90 protein wells for
600 s immobilization; (3) baseline 2, sensors were moved to plates containing 0.1 M PBST
for another 120 s equilibration; (4) association, sensors moved to ligand buffer for 300 s to
measure Kon; (5) dissociation, sensors moved to 0.1 M PBST buffer for 300 s to measure Koff.
Four concentrations of each compound were utilized to obtain the final curve (ranging from
100 µM to 12.5 µM). All the data were analyzed by ForteBio data analysis software and the
equilibrium dissociation constant (KD) values were calculated using the equation (KD =
Koff/Kon).
2.4.6 GST pull-down assay
GST pull-down experiments were conducted using Pierce™ GST Protein Interaction
Pull-Down Kit (Thermo Scientific). Loading buffer consisted of 25 mM Tris, 0.15 M NaCl
(pH 7.2) and elution buffer contained an extra 10 mM reduced glutathione. 10 µg purified
GST-Hsp90N protein (N-terminal Hsp90 with GST-tag) was used as bait protein to
immobilize on the sepharose-contained resin in the Pierce Spin Columns for binding to the
substrates. Compounds and celastrol were added into the columns after incubation with
Cdc37 for 1 h at 37 °C. In addition, then centrifugation (1 min at 1,000 × g, 4 °C) was
performed to remove solution containing Cdc37 (prey protein) before elution buffer was
added. The columns were washed by loading buffer three times and the eluent was collected
via centrifugation (1 min at 1,000 × g, 4 °C) for analysis of the eluted protein by SDS-PAGE

gel.
ACCEPTED MANUSCRIPT
2.4.7 Cell culture
The human cancer cell lines MCF-7, SKBR-3 and A549 were purchased from American
Type Culture Collection (ATCC) or Cell Culture Center at the Institute of Basic Medical
Sciences, Chinese Academy of Medical Sciences. MCF-7 cells were cultured in MEM,
SKBR-3 cells were cultured in RPMI-1640 medium and A549 cells were cultured in F12K.
All of them were supplemented with 10% fetal calf serum and antibiotics. Cells were
maintained in a 37 oC incubator (Thermo, USA) with 95% humidity and 5% CO2.
2.4.8 Western Blot Analysis
The expression levels of client proteins were determined by Western blot assay.
SKBR-3 cells were treated with multiple diluted compound 10 (1 µM, 10 µM and 50 µM) for
24 h. Celastrol (1 µM and 10 µM) was chosen as the reference. Lysis buffer (50 mM
Tris-HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1% NP-40, 0.2 mM PMSF, 0.1 mM NaF and
1.0 mM DTT) was used to lyse the cells after they were rinsed with PBS and trypsinized.
Then, the lysates were centrifuged at 12000 rpm for 10 min at 4 oC and the supernatants were
collected. The concentration of the protein was determined by BCA assay (Thermo, Waltham,
MA). Equal amounts of the total protein extracts were determined by SDS-PAGE. The
primary antibodies were anti-Hsp90 (#4874, Cell Signaling Technology, Inc., USA), anti-Akt
(#9272, Cell Signaling Technology, Inc., USA), anti-Cdk4 (#12790, Cell Signaling
Technology, Inc., USA) and anti-β-actin (#66009-1-Ig, Proteintech Group, Inc., China).
Subsequently, the membranes were treated with a DyLight 800-labeled secondary antibody
o
(#072-07-15-06, KPL, USA) at 37 C for 1 h. Finally, the Odyssey infrared imaging system
(LI-COR, Lincoln, Nebraska, USA) were used for scanning.
2.4.9 Anti-proliferative assay

Cell-based MTT assay was utilized to evaluate the anti-proliferative activity in MCF-7,
ACCEPTED MANUSCRIPT
SKBR-3 and A549 cells. All the Cells were seeded in 96-well plates (10000/well) and
maintained for 24 h. Next, the compounds dissolved in DMSO in disparate concentrations
was added, and cells were incubated for 72 h. Additional incubation in 37 °C was conducted
after adding 20 µL MTT solution (5 mg/mL in PBS). Next, 150 µL DMSO was added in the
wells to dissolve the MTT-formazan crystal, which generated only in viable cells. The OD
values were determined by the Elx800 absorbance microplate reader (BioTek, Vermont, USA)
at 490 nm. According to the linear relationship between OD values and the number of viable
cells, the IC50 value was calculated from parallel experiments via Graphpad Prim 6.0
software and performed as the mean ± SD.
3. Results and Discussion
3.1 Pharmacophore model generation
Since numerous protein crystal structures have been determined, the receptor-ligand
complex based pharmacophore method turns to be more and more significant and efficient
[30]. To discover non-peptide modulators targeting Hsp90-Cdc37 PPI, the significant
chemical features in binding hot-spots were generated to perform multiple pharmacophore
elements based on the binding mode of active peptides and complex crystal structure [29].
As shown in Fig. 2, two polar regions are composed of a histidine and a terminal arginine
which have been demonstrated to be the determinants for the recognition and binding
between target and ligands. Phenylalanine and the terminal acetyl represent a hydrophobic
pocket which contribute to the main entropy for binding. The hydrophobic linker (providing
less binding contribution) occupies a superficial cavity on Hsp90 and exhibits a moderate
distance of approximately 15 Å. In addition, it has been reported that the side-chain of

Arg167 in Cdc37 forms an electrostatic interaction and a H-bond with the side-chain of
ACCEPTED MANUSCRIPT
Glu47 in Hsp90,which were regarded as the most decisive factor for Hsp90-Cdc37 PPI in the
crystal structure [17]. Thus, one positive feature at polar region 2 is indispensable for ligand
binding. All these results taken together resulted in five pharmacophore features: three
H-bond features, one hydrophobic feature and one ionizable feature.
Figure 2. Pharmacophore model generation based on active peptide which is represented in green sticks. The
surfaces represent Hsp90. (A) Ligand and receptor-based pharmacophore generation containing two polar
regions and one hydrophobic pocket. (B) The final pharmacophore used for screening. Pharmacophore
features are described as follows: hydrogen bond donor (HBD), violet; hydrogen bond acceptor (HBA),
green; hydrophobic center, azure; positive ionizable center, red. (C) Pharmacophore distance. The distance
between certain features is represented in Å.
3.2 Pharmacophore-based virtual screening
Based on the pharmacophore established above, Specs and NCI database was prepared
and filtered primarily. Depending on the critical features of Hsp90-Cdc37, the standard
charge of each compound in the database was calculated at neutral condition (pH = 7.4). All
compounds were prepared using ligand prepare protocol in Discovery Studio 3.0. However,

considering the complexity of the pharmacophore model and harsh screening criteria for
ACCEPTED MANUSCRIPT
eligible ligands, one feature could be omitted in the ligand-pharmacophore mapping
procedure. During this process, shape constraints were also applied to maintain the proper
conformation of the ligands. Only 1532 compounds in the database, which mapped to the
pharmacophore model, were retained for further docking filtration. Libdock, Ligandfit and
CDDOCKER were sequentially applied to evaluate the binding capacity of the filtered
compounds. Finally, we purchased 31 compounds that satisfied all the demands mentioned
above. The whole screening process is shown in Fig. 3.
Figure 3. Multistep virtual screening and biology validation strategy for the discovery of modulators
targeting Hsp90-Cdc37 PPI.
3.3 In vitro validation of virtual hits
The virtual screened hit compounds were subsequently determined by a series of in
vitro biological assays including homogeneous time-resolved fluorescence (HTRF) assay,
ATPase inhibition assay, fluorescence polarization (FP) assay and direct binding assay
(biolayer interferometry, BLI). HTRF is a sensitive method to determine the effect of
protein-protein interaction based on a time-resolved fluorescence resonance energy transfer

(TR-FRET) mechanism. When Hsp90 bound to Cdc37, the conjugated donor and acceptor
ACCEPTED MANUSCRIPT
came into close proximity to exhibit high emission spectrum. Once the interaction between
two proteins is disrupted or affected by inhibitors, the emission spectrum will reduce
accordingly [31]. As shown in Table 1, HTRF assay was utilized to verify the potency of all
compounds to disrupt Hsp90-Cdc37 PPI in vitro. The detailed chemical structures are shown
in Fig.4. As shown in the results in Table 1, compound VS-2, VS-8, VS-22 and VS-25 exert
more than 30% inhibition at the concentration of 100 µM. Considering all hits might directly
bind to Hsp90 N-terminal, an FP competitive assay was performed to determine whether
these compounds could occupy the ATP binding site of Hsp90. The detailed method was
reported previously [32]. Based on FITC labeled Geldanamycin, the first identified Hsp90
ATPase inhibitor, all the compounds showed no binding at 100 µM ligand concentration
which demonstrated an obvious difference between these compounds and classic Hsp90
N-terminal inhibitors. To examine the effect of ATPase inhibition, an assay of the Discover
RX ADP HunterTMPlus Assay Kit was performed. The final data showed that compound
VS-2, VS-8 and VS-25 exhibited no influence on Hsp90 ATPase even when the
concentration raised up to 100 µM. However, VS-22 revealed a weak inhibition as 45.67%
which might be a result of resemble structure to purine analogues. Among all the assays,
celastrol was used as a positive control to show 79.34% inhibition in HTRF assay while little
inhibition effect in ATPase and FP assay at 100 µM was observed, which is consistent with
literature reports [20]. 17-DMAG, a classic Hsp90 ATPase inhibitor, exerted no
Hsp90-Cdc37 PPI disruption ability but high inhibitory effect in ATPase and FP assay
(shown in Table 2).
Subsequently, biolayer interferometry (BLI) assay, which is identified as one of the
most common ways to determine the direct binding affinity of ligand-protein interactions

[33], was applied to examine the direct binding capacity of these four compounds. A
ACCEPTED MANUSCRIPT
dose-dependent binding manner was observed according to the association and dissociation
curves. KD values were calculated by the equation: KD = Koff/Kon. Finally, VS-8 obtained a
measured KD value of 80 µM. Compound VS-2 and VS-25 exhibited no binding while
VS-22 exerted a quite feeble binding ability (＞100 µM). The data of the in vitro assays we
obtained showed that VS-8 could bind to Hsp90 with a moderate binding ability, exhibiting
an entirely different binding mechanism with Hsp90 ATPase inhibitors. VS-8 exerted no
effect on Hsp90 ATPase while disrupted Hsp90-Cdc37 PPI with micromolar inhibitory
activity in HTRF assay through a direct binding manner. Although VS-8 gave its potential to
disrupt Hsp90-Cdc37 PPI, the potency of VS-8 is feeble which needs a series of derivatives
for further investigations.
Table 1 100 µM inhibition and binding constants (KD) of 4 compounds c
Compound No.
HTRF
inhibition
(100 µM)
47.64%
ATPase
inhibition
(100 µM)
FP
KD
inhibition
(100 µM)
N/Aa
(BLI)
N/A
VS-2 (AP-263/41670332）
VS-8 (AN-329/40718894)
＜10%
59.33%
36.59%
30.67%
N/Aa
N/Aa
N/Aa
80.4µM
＜10%
45.67%
VS-22 (AP-906/41637264）
VS-25 (AM-900/14781022)
＞100µM
N/A
＜10%
Celastrol (10 µM)
79.34%
N/Aa
12.31%
96.45%
N/Aa
NDb
NDb
17-DMAG (10 µM)
99.82%
a No activity
b Not determined
c all values are the mean of three independent experiments.

ACCEPTED MANUSCRIPT
Figure 4. Chemical structures of four virtual screened hits, VS-2, VS-8, VS-22, VS-25.
3.4 Potential binding model analysis and Structure activity relationship study of VS-8

ACCEPTED MANUSCRIPT
Scheme 1: Reagents and conditions: a) SOCl2, reflux; b) THF, pyridine, rt; c) 20% NaOH solution, reflux; d)
THF or 1,4-dioxane, pyridine, rt.
To confirm the potency of VS-8, a series of derivatives were designed and synthesized.
VS-8 is a symmetrical structure containing a central benzamide with three alkyl carbons as a
linker to perform as two equal aromatic systems by connecting a sulfonamide. In order to
have a better understanding of VS-8 binding mode for further compounds design, we
modeled VS-8 into Hsp90-Cdc37 binding site using the crystal structure (PDB: 1US7). As
shown in Fig 5A and 5B, VS-8 could bind to Hsp90 by occupying both P1, P2 and P3
sub-pockets to exert potential disruption ability of Hsp90-Cdc37 PPI. One of the methyl
oxazole side-chain occupied P1 (including Leu93, Ala97, Phe124 and Tyr125) and the other
formed hydrogen bond with Gln119 in Hsp90 to occupy P3. The central linker chain with

five carbon atoms bound to P2 and formed interactions with Asn37 and Gln119 through its
ACCEPTED MANUSCRIPT
amides. Based on this point of view, the major concerns for optimization of VS-8 might
include linker chain variations to obtain a better ligand efficiency in P2 and different
aromatic substitutions to form extra possible interactions with Hsp90, especially in P3 which
exert fewer interactions.
Subsequently, we first explored the main linker on the original structure by altering the
amount of middle alkyl carbons or directly replacing it with phenyl ring while the original
sulfonamide substitution was maintained. Different lengths of alkyl carbons (from one to
four) were utilized to obtain main linker variation derivatives 1-3 and phenyl ring substituted
compound 4 according to a two-step synthetic route, shown in Scheme 1. As shown in Table
2, besides compound 4, the derivatives (compounds 1-3) of main linker variations slightly
affected the inhibition in HTRF assay under the compound concentration of 100 µM which
indicated that phenyl ring was not favored in central linker modifications. Among them,
compound 1 and 2 exhibited the optimal inhibition of 66% and 70%, respectively. Based on
this result, the central linker length of one to two alkyl carbons might be the most ideal to
exert optimal occupation of P2. Further modifications of different substitutions of
sulfonamide were designed to rigidify the one/two carbon linker and explored the
significance of different heterocyclic ring of the negative charge/H-bond acceptors to form
more interactions in P1 and P3. Using the scaffold of several penta/hexa-heterocycles as the
sidechain amide modifications, compounds 5-16 were obtained as second series of
symmetrical compounds. The preparation pathway of compounds 5-16 from 5f-5o with
2a-2d was shown in Scheme 1. Unfortunately, the majority of these compounds 5-16
exhibited declined or lost activity (less than 50% inhibition under 100 µM), except
compound 10 with N-methylpyrazole substitution and the central linker of two alkyl carbons

exhibited slightly improved activity. Using the same model as VS-8, we modeled compound
ACCEPTED MANUSCRIPT
10 to the same binding site of Hsp90-Cdc37 PPI. As shown in Fig. 5C and 5D, compound 10
exerted similar binding mode as VS-8. As we expected, the central linker shrinkage made
this chain more unfolded which might result in a higher ligand efficiency in occupation of P2,
leading to the formation of significant interactions between amides and Arg32, Glu33 as well
as Ser36. N-methylpyrazole substitution in P1 formed similar interactions (interacted with
Ala97, Val122 and Tyr125) with VS-8. Additionally, N-methylpyrazole substitution in P3
could form more interactions with multiple residues including Ser198, Glu199 and Phe200
which might result in a better activity.
To further evaluating the detailed inhibitory activity and binding capacity between
compound 10 and VS-8, we calculated the IC50 values using HTRF assay and determined the
binding constants using biolayer interferometry assay. As shown in Fig. 6, compound 10
exhibited more promising inhibitory effect (IC50 = 27.40 µM) compared to VS-8 (IC50 =
76.85 µM).
Table 2.SAR exploration of central linker and sidechain substitution. a
Com.
1
x or Ar
R
Inhibition % (100 µM)
-CH2-
65.6
-CH2CH2-
-CH2CH2CH2CH2-
-Ph-
69.8
20.7
NA
2
3
4
5
6
7
8
-CH2-
＜10%
-CH2-
50.9
-CH2-
30.4
48.7
-CH2CH2-

-CH2CH2-
9
＜10
ACCEPTED MANUSCRIPT
-CH2CH2-
-CH2CH2-
-CH2CH2-
-CH2CH2-
-CH2CH2-
-CH2CH2-
-CH2CH2-
79.4
10
11
12
13
14
15
16
49.6
＜10
＜10
＜10
＜10
＜10
59.3
92.3
VS-8
Celastrol
a all values are the mean of three independent experiments.
Figure 5. Potential binding models of VS-8 and compound 10. The surface of Hsp90 was colored. (A,C)

Interaction modes of the VS-8 and compound 10 respectively. Compound 10 formed extra multiple
ACCEPTED MANUSCRIPT
interactions with Hsp90 compared to VS-8. (B,D) Binding surface depiction of the VS-8 and compound 10.
Both two compounds occupied the P1, P2 and P3 sub-pockets. Compound 10 performed better
conformational occupation in P1 and P2 with the central linker shrinkage and N-methylpyrazole substitution
of the side-chain in VS-8.
3.5 Compound 10 disrupts Hsp90-Cdc37 PPI in vitro, exhibits anti-proliferative
capacity and down-regulates the expression of client proteins
To confirm the PPI disruption ability of compound 10, we utilized a GST-pull down
assay kit for further validation. Detailed experiment procedure was described in the methods
and materials. After all the eluted liquids were analyzed by SDS-PAGE and western-blot
with anti-tag antibody (anti-GST and anti-His). Three concentrations of compound 10 were
set from 1 µM to 20 µM and two concentrations of celastrol were set as 1 µM and 20 µM
which including two same concentrations of these two compounds. As shown in Fig. 6C,
compound 10 exhibited little inhibition ability at the concentration of 1 µM and 5 µM, but
almost completely abrogated the interaction of Hsp90 with Cdc37 at the 20 µM
concentration. In addition, celastrol could also completely disrupted the interaction between
Hsp90-Cdc37 at 20 µM concentration. All these data demonstrated that compound 10 could
disrupt the interaction between Hsp90-Cdc37 in vitro through a concentration-dependent
manner and exhibit similar disruption ability compared with celastrol.

ACCEPTED MANUSCRIPT
Figure 6. (A) Dose-dependent inhibition of compound VS-2, VS-8, VS-22, VS-25 and compound 10 in the
HTRF assay. (B) Biolayer interferometry sensorgrams of the binding of multiple concentrations of VS-8 to
Hsp90. (C) compound 10 and VS-8 disrupt Hsp90-Cdc37 in vitro through a concentration-dependent
manner by GST-pull down assay.
It has been reported that Hsp90 and Cdc37 revealed high relevance with the occurrence
and development of cancer [14, 34]. The anti-proliferative capacities of 5 compounds and
celastrol were determined by MTT assay against the human breast cancer MCF-7 and
SKBR3 cell lines as well as non-small cell lung cancer cell line A549. Cells were treated
with various concentration (ranging from 200 µM to 0.823 µM with three times dilution) of
compounds for 72 h. The cell viability was assessed by mitochondrial dehydrogenase activity.
As shown in Table 3, among all the compounds, compound 10 exhibited the optimal
anti-proliferative activity with IC50s of 26 µM, 15 µM and 38 µM against MCF7, SKBR3
and A549, respectively, which indicated a high correlation with previously described in vitro