Novel androgen receptor antagonist identified by structure-based virtual screening, structural optimization, and biological evaluation

Article abstract of DOI:10.1016/j.ejmech.2020.112156

Androgen receptor (AR) plays important roles in the development of prostate cancer (PCa), and therefore it has been regarded as the most important therapeutic target for both hormone-sensitive prostate cancer (HSPC) and advanced PCa. In this study, a nove

Full text of DOI:10.1016/j.ejmech.2020.112156

Journal Pre-proof
Novel androgen receptor antagonist identified by structure-based virtual screening,
structural optimization, and biological evaluation
Qin Tang, Weitao Fu, Minkui Zhang, Ercheng Wang, Lvhu Shan, Xin Chai, Jinping
Pang, Xuwen Wang, Xiaohong Xu, Lei Xu, Dan Li, Rong Sheng, Tingjun Hou
PII:
S0223-5234(20)30123-9
DOI:
https://doi.org/10.1016/j.ejmech.2020.112156
EJMECH 112156
Reference:
To appear in: European Journal of Medicinal Chemistry
Received Date: 23 December 2019
Revised Date: 7 February 2020
Accepted Date: 16 February 2020
Please cite this article as: Q. Tang, W. Fu, M. Zhang, E. Wang, L. Shan, X. Chai, J. Pang, X. Wang, X.
Xu, L. Xu, D. Li, R. Sheng, T. Hou, Novel androgen receptor antagonist identified by structure-based
virtual screening, structural optimization, and biological evaluation, European Journal of Medicinal
Chemistry (2020), doi: https://doi.org/10.1016/j.ejmech.2020.112156.
This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition
of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of
record. This version will undergo additional copyediting, typesetting and review before it is published
in its final form, but we are providing this version to give early visibility of the article. Please note that,
during the production process, errors may be discovered which could affect the content, and all legal
disclaimers that apply to the journal pertain.
© 2020 Published by Elsevier Masson SAS.

Novel androgen receptor antagonist identified by
structure-based virtual screening, structural optimization,
and biological evaluation
Qin Tang1,#, Weitao Fu1,#, Minkui Zhang1,#, Ercheng Wang1, Lvhu Shan2, Xin Chai1,
Jinping Pang1, Xuwen Wang1, Xiaohong Xu2, Lei Xu3, Dan Li1,*, Rong Sheng1,*,
Tingjun Hou1,*
1College of Pharmaceutical Sciences, Zhejiang University, Hangzhou 310058, Zhejiang, China
2Institute of Cancer Research and Basic Medical Sciences of Chinese Academy of Sciences, Cancer
Hospital of University of Chinese Academy of Sciences, Zhejiang Cancer Hospital, Hangzhou 310022,
Zhejiang, China
3Institute of Bioinformatics and Medical Engineering, School of Electrical and Information
Engineering, Jiangsu University of Technology, Changzhou 213001, Jiangsu, China
# These authors contributed equally to this work
*Corresponding authors:
Tingjun Hou
E-mail: tingjunhou@zju.edu.cn
Rong Sheng
Email: shengrong1973@163.com
Dan Li
Email: lidancps@zju.edu.cn
1

Abstract: Androgen receptor (AR) plays important roles in the development of
prostate cancer (PCa), and therefore it has been regarded as the most important
therapeutic target for both hormone-sensitive prostate cancer (HSPC) and
advanced PCa. In this study, a novel hit (C18) with IC50 of 2.4 µM against AR
transcriptional activity in LNCaP cell was identified through structure-based virtual
screening based on molecular docking and free energy calculations. The
structure-activity relationship analysis and structural optimization of C18 resulted in
the discovery of a structural analogue (AT2), a more potent AR antagonist with
16-fold improved anti-AR potency. Further assays indicated that AT2 was capable of
effectively inhibiting the transcriptional function of AR and blocking the nuclear
translocation of AR like the second generation of AR antagonists. The antagonists
discovered in this study may be served as the promising lead compounds for the
development of AR-driven PCa therapeutics.
Keywords: Prostate cancer; androgen receptor; structure-based virtual screening;
antagonist; molecular docking
2

1. Introduction
Prostate cancer (PCa) is the most diagnosed cancer among men worldwide [1]. The
androgen receptor (AR) is a member of the nuclear receptor superfamily of
ligand-activated transcription factor. Abnormal activation of the AR signaling
pathway plays a pivotal role in the development and progression of PCa [2, 3].
Currently, a vital approach to prevent the excessive activation of androgens is the
treatment with AR antagonists to block the androgens binding to AR [2, 3]. Androgen
deprivation therapy (ADT), the most important treatment for advanced PCa, reduces
the levels of androgen production by surgical or pharmacological castration.
Unfortunately, most patients usually develop into castration-resistant prostate cancer
(CRPC) after 2 years of ADT treatment. The possible mechanisms of CRPC include
AR mutations, AR amplifications, and active AR splice variants [4]. In general,
existing studies show both PCa and CRPC are closely related to AR.
Current clinically used first-generation or second-generation AR antagonists, such
as R-bicalutamide and Enzalutamide (Enz), have achieved great success against
androgen-dependent PCa and improved the survival rate of PCa patients [2, 8].
However, after the initially effective response, efficacy of these AR antagonists is
suffered from the rapid emergence of drug resistance [9-12]. One of the main reasons
for drug resistance is the acquired point mutations at the AR LBP. An explanation for
this reason is that some point mutations in the AR LBP, such as W741L, W741C, and
T877 A, would create a more spacious LBP, thus converting these antagonists to
agonists and inducing drug resistance [13-15]. To deal with this situation, a widely
used approach in drug design is to modify the original antagonists into larger
chemical structures [16-19]. Nevertheless, in general, larger molecules often confront
several problems associated with unfavorable permeability and physiological
distribution [20]. Therefore, it is still urgent to discover new AR antagonists with
novel scaffolds to improve clinical outcomes.
3

A common approach to design antagonists with novel scaffolds is to use virtual
screening (VS), which has also been used to identify novel hits of AR. For instance,
the
AR
antagonist
6-(3,4-dihydro-1H-isoquinolin-2-yl)-N-(6-methylpyridin-2-yl)nicotinamide (DIMN)
was screened out by structure-based VS (SBVS) based on the crystal structure of the
AR-metribolone
complex.
Another
AR
antagonist
5,5a,6,10b-tetrahydroindeno[2,1-b]indole (VPC 12060) was discovered by SBVS
and ligand-based virtual screening (LBVS). Wang et al. identified
pyrazolopyramidine analogs as novel potent AR antagonists also by combining SBVS
and LBVS [23]. Recently, we reported the identification of a series of novel AR
ligands, including AR agonist and AR antagonist, through an integrated strategy by
combining SBVS based on the crystal AR structures in complex with its agonists [24].
It has been proved that SBVS is potent to discover novel AR antagonists.
Molecular docking has been recognized as the most popular method for
structure-based drug design [25-27]. It can predict the binding conformations of
ligands to the target, and rank the ligands by scoring functions [28-30]. Compared
with most scoring functions in molecular docking, the Molecular
Mechanics/Generalized Born Surface Area (MM/GBSA) method can achieve a better
balance between identifying the binding poses and predicting the binding free
energies, and therefore it has been widely used in SBVS and lead optimization
[31-33]. In recent years, variable dielectric MM/GBSA (VD-MM/GBSA), a novel
modified MM/GBSA algorithm, has attracted increasing attention [34-37].
In this study, a multi-step SBVS strategy based on molecular docking and
VD-MM/GBSA rescoring was employed to screen the Specs database, and 32
compounds were finally purchased for bioassay verification. Three of them exhibited
strong bioactivities, and competitive ligand binding assay showed that 1 compound
(C18) was targeting the LBP of AR. Structural optimization based on molecular
dynamics (MD) simulation and structure-activity relationship (SAR) analysis was
then applied to the hit of C18, which resulted in the discovery of AT2, a more
4

promising AR antagonist with 16 folds improved anti-AR potency relative to C18.
Moreover, AT2 displayed much better anti-proliferative effects than Enz in three
androgen independent cell lines, including PC3, C4-2, and DU145. The qPCR and
immunofluorescence assays illustrated that AT2 can inhibit AR transcriptional
activity and block nuclear translocation of AR. Our study provided valuable clues for
the development of a novel class of AR therapeutic antagonists to combat PCa.
2. Results and discussion
2.1. Evaluation of candidate compounds and discovery of hit C18
The schematic workflow of the SBVS protocol used in this study is presented in
Figure 1. By applying molecular docking, VD-MM/GBSA rescoring and structural
clustering, a number of potential AR antagonists were identified. As a result, a total of
32 potential active candidates were purchased and submitted to bioassays. At first,
LNCaP-ARR2PB-eGFP-based transcriptional activity assay was performed to
determine the antagonistic activities of the screened compounds at the concentration
of 10 µM [24]. Compounds C10, C12, and C18 have achieved over 50% of
androgenic activity of Enz (Figure 2A and Table S1). Then, the 3 active compounds
were subjected to the competitive ligand binding assay to assess their binding
affinities to the LBP of the AR LBD. C18 exhibited satisfactory binding affinity at 10
µM (Figure 2B). Thereafter, the binding affinity of C18 at gradient concentrations
was evaluated. As shown in Figure 2C, C18 binds to the AR LBP in a dose-dependent
fashion (IC50 = 4.02 µM), indicating that C18 could directly target the LBP. Then the
AR antagonistic activity of C18 was evaluated. It showed that C18 inhibited AR
transcriptional activity with IC50 = 2.4 µM, while that for Enz was 0.08 µM (Figure
2D). To avoid potential false positive, the effect of C18 on prostate-specific antigen
(PSA), the most common used biomarker for PCa, was detected. It’s observed that
C18 decreased the PSA level also in a dose-dependent manner and the IC50 value was
1.40 µM, while the value for Enz was 0.13 µM (Figure 2E). The structure of C18 was
then compared with the known AR antagonists deposited in the BindingDB database
5

using the Similarity/Distance Screen module in Canvas software [38]. As indicated by
relatively low Tanimoto coefficient (Tanimoto coefficient < 0.3), C18 did not share
high structural similarity with any previously reported AR antagonists. Taken together,
C18 is a novel and potent AR antagonist, and can be served as a starting point for
further structural optimization to improve its activity.
2.2. Structural analysis of C18
To guide further structural optimization for C18, molecular dynamics (MD)
simulations were performed to predict the dynamic binding behavior between C18
and the AR LBD. The root-mean square deviations (RMSD) of the backbone atoms of
the AR LBD and the heavy atoms of C18 were calculated to monitor the stability of
the complex during the MD simulations. As shown in Figure 3A, the RMSD
evolutions for both the backbone atoms of the AR LBD and the heavy atoms of C18
tended to converge after ~0.2 µs with the fluctuations within 1 Å (Figure 3A),
suggesting the complex reached stability via 1 µs MD simulation. In search of the
essential residues responsible for the binding of C18 to the AR LBD, the per-residue
decomposition based on VD-MM/GBSA was carried out as shown in Figure 3B. It
can be found that the major contributors were Met745, Leu704, Thr877, Phe764,
Leu873, Met895, Leu707, Met742, Met749, and Asn705. Most of them were
hydrophobic residues around C18 (Figure 3C). In particular, the residue Leu704
formed a hydrogen bond with the nitrogen atom of C18. That is to say, the
antagonistic effect of C18 was predominantly dependent on hydrophobic interactions
and an important hydrogen bond with Leu704, which was highly consistent with the
extremely hydrophobic feature of C18. According to these observations, we
concluded that the alteration of different substituted groups on the phenyl ring of the
R1 moiety might lead to a rotation of the phenyl moiety or increase the probability of
the formation of hydrogen bonds with Asn705 and Thr877, so as to change the
antagonistic activity of the compound. The hydrophobic interactions between the AR
LBD and C18 were quite critical, and thus modification of the naphthalene ring to
6

alter hydrophobicity and introduce more hydrogen bond donors/acceptors may
reinforce hydrogen bonding interaction and improve binding affinity. The
3a,4,5,11c-tetrahydro-3H-benzo[f]cyclopenta[c]quinoline ring that offers a hydrogen
bond with Leu704 should be retained to maintain the anti-AR potency. Replacing the
A-ring may alter the hydrophobicity, which may be favorable to improve binding
affinity. Accordingly, we designed three series of C18 analogues (AT-CT series, Table
1) to verify the above hypotheses.
The general strategy that may be used to improve the AT, BT and CT series is to
alter different substituted groups R1 moiety, alter naphthalene ring to change
hydrophobicity and introduce hydrogen bond donors/acceptors, and replace A ring to
alter hydrophobicity. The AT and BT series were purchased from the commercial
chemical compound vendor and the CT series were synthesized by our lab (Table S2).
2.3. Biological evaluation of the C18 analogues
As shown in Table 1, the effect of the substituents on the phenyl ring of the R1 moiety
was firstly examined (AT1-AT8) by AR transcriptional activity assay. It was observed
that the monosubstituted compounds with smaller substituents had better activity, and
the activity of the chlorine substituted compound AT2 was the best. Subsequently, the
importance of the naphthalene ring was examined. The antagonistic activity was
totally lost when the naphthalene ring was replaced by different R2 groups (BT1-BT7).
In addition, the contribution of the A ring was also investigated (CT1-CT6). It was
found that the antagonistic activity decreased obviously when the ring A was replaced,
suggesting that the cyclopentene ring is an essential component for the maintenance
of bioactivity. The binding of three representative compounds (AT2, BT5, and CT5)
were analyzed by molecular docking to understand why the BT series completely lost
the antagonistic activity compared with the AT and CT series. As shown in Figure S1,
the binding modes of AT2, BT5, and CT5 were quite similar. The docking scores for
AT2, BT5 and CT5 were -11.05, -11.97, and 11.78 kcal/mol, respectively. Though the
docking scores of BT5 and CT5 were comparable to that of AT2, their anti-AR
7

activities were significantly different. A possible explanation is that the cell membrane
penetrating ability of the BT series is quite low because the AR transcriptional
activities were determined at the cellular level. Another possible reason is that BT5
cannot correctly target to the AR LBD. Overall, these results suggested that alteration
of the naphthalene ring with hydrogen bond donors/receptors would lead to a
complete loss of the antagonistic activity, and appropriate steric hindrance and
hydrophobicity are also necessary for optimal antagonistic activity.
2.4. AT2 is a promising AR antagonist
The AT series (AT1, AT2, AT4, AT5, AT6, AT7, and AT8) were validated for their
targeting ability. As shown in Figure 4A, all of the tested compounds showed
relatively lower polarization values than the control of DMSO, demonstrating these
compounds bound to the AR LBP effectively. Considering that AT2 possesses optimal
anti-AR transcriptional activity, the production of endogenous PSA in AT2 treated
LNCaP cells was examined. It was observed that AT2 caused a dose-response
decrease of the PSA production with IC50 of 0.43 µM, which was consistent with its
anti-AR transcriptional activity (Figure 4B). Then, the transcriptional level of PSA
was evaluated by qPCR (Figure 4C), further confirming the effect of AT2 on PSA. In
addition, another AR downstream gene transmembrane protease serine 2 (TMPRSS2)
was likewise assessed. AT2 could also dose-dependently reduce the mRNA level of
TMPRSS2 (Figure 4D). Taken together, the above bioassays suggested that AT2
could efficiently target the AR LBP and antagonize AR transcription activity. To
confirm that the antagonistic effect was not induced by cytotoxicity, the viability of
murine embryonic fibroblast cells (NIH 3T3 cells) exposed to AT2 was investigated
using MTT assay. Similar to Enz and C18, AT2 showed no cytotoxicity against NIH
3T3 cells even at a high dose of 50 µM (Figure 4E). Thus, it could be concluded that
AT2 was nontoxic at its effective dose against AR, which confirmed the possibility of
AT2 as a candidate of AR antagonist.
8

2.5. Inhibition of nuclear translocation of AR by AT2
Different from the first generation of AR antagonist, Enz is capable of blocking the
translocation of AR into nucleus, which is also an important feature for the second
generation of AR antagonists including the recently proved apalutamide and
darolutamide [39]. To investigate the effect of AT2 on the subcellular localization of
AR, immunofluorescence was performed in LNCaP cells. As shown in Figure 5, AR
was primarily cytoplasmic in the absence of DHT (DMSO), and exposure to androgen
(DHT) markedly increased the nuclear-cytoplasmic ratio of AR immunofluorescence
intensity, demonstrating the translocation of AR from the cytoplasm to the nucleus,
while AT2 blocked the DHT-induced AR nuclear translocation, and the corresponding
results were also observed in the presence of Enz. This finding indicated that AT2
possessed similar feature to Enz and could antagonize AR with a second-generation
antagonist like mechanism.
2.6. In vitro anti-proliferative activity study
The antiproliferative potential of AT2 was evaluated in representative androgen
dependent and independent cells lines, including LNCaP, C4-2, PC3, and DU145
(Figure 6). The cell lines of LNCaP and C4-2 are known to express varying degree of
AR. PC3 and Du145 are AR independent cell lines isolated from bone and brain
metastatic models of human PCa, respectively. These cells were exposed to varying
concentrations of Enz, C18 and AT2 in serum-fed condition. As shown in Figure 6A,
the effect of AT2 was comparable with that of Enz on AR-dependent cell lines of
LNCaP. However, in androgen-independent C4-2 and AR-independent PC3 and
DU145, AT2 exhibited better antagonistic activity than Enz. Interestingly, the lead hit
C18 showed similar performance in C4-2 while totally different performances in
AR-independent cell lines. The lead hit C18 performed better than AT2 in PC3, but
much worse than AT2 in DU145, suggesting that C18 and AT2 might exert their
antiproliferative activities against PCa cell lines through a combination of
AR-dependent and -independent pathways.
9

3. Conclusion
In this study, the SBVS strategy based on molecular docking and VD-MM/GBSA
rescoring was employed to discover novel AR antagonists against PCa. A novel hit
(C18) with promising anti-AR activities in a group of bioassays was identified. The
subsequent structural optimization focusing on the change of the spatial
conformations and hydrophobicity leads to a more potent AR antagonist (AT2)
without toxicity at its effective concentration against AR. Besides, AT2 could
efficiently antagonize AR transcriptional activity, suppress downstream target gene of
AR, and block the DHT-induced AR nuclear translocation as the second generation of
AR antagonists. Moreover, the cytotoxicity results of AT2 towards the cells of LNCaP,
C4-2, PC3, and DU145 illustrated that C18 and AT2 might exert their
antiproliferative activities via a combination of interfering AR-dependent and
-independent pathways. Collectively, further studies on structural optimization and
action mechanism of AT2 are hopefully fruitful and will benefit the development of
novel AR antagonists.
4. Methods and materials
4.1 Chemistry
4.1.1 General synthetic procedure for compounds CT1~CT6
A mixture of 2-naphthylamine (1, 0.02 mol) and appropriate benzaldehyde (2a-c,
0.022 mol) in dry dichloromethane was stirred under reflux overnight. After
completion of the reaction, the mixture was concentrated to give appropriate Schiff
base 3a-c, which were used directly without further purification. Then a mixture of
3a-c (0.02 mol), 3,4-dihydro-2H-pyran or 2-norbornylene (0.022 mol) and indium
trichloride (0.002 mol) in dichloromethane/acetonitrile (1:1, 20 mL) was stirred at
room temperature overnight. After completion of the reaction, water (20 mL) was
added, and the aqueous layer was extracted with dichloromethane (20 mL×3). The
10

combined organic layers were dried over Na2SO4, concentrated and purified by silica
gel chromatography to give CT1-3 or CT4-6 (Scheme 1).
Scheme 1. Synthesis of CT1~CT6
4.1.2 5-Phenyl-3,4,4a,5,6,12c-hexahydro-2H-benzo[f]pyrano[3,2-c]quinolone (CT1)
1
White solid, yield 65%. H NMR (500 MHz, DMSO-d6): δ 7.76 (d, J = 8.5 Hz, 1H),
7.62 (d, J = 8.0 Hz, 1H), 7.55 (d, J = 9.0 Hz, 1H), 7.50 (d, J = 7.5 Hz, 2H), 7.42 (t, J =
7.5 Hz, 2H), 7.38-7.34 (m, 2H), 7.12 (t, J = 7.5 Hz, 1H), 6.94 (d, J = 9 Hz, 1H), 6.48
(s, 1H), 4.88 (d, J = 2.5 Hz, 1H), 4.68 (d, J = 11.5 Hz, 1H), 3.98 (dd, J1 = 11 Hz, J2 =
3.5 Hz, 1H), 3.78 (t, J = 11 Hz, 1H), 1.99-1.97 (m, 1H), 1.87-1.78 (m, 1H), 1.74-1.67
13
(m, 1H), 1.33-1.27 (m, 2H). C NMR (125 MHz, DMSO-d6): δ143.8, 142.8, 134.0,
129.5, 128.9, 128.6, 128.5, 128.1, 127.1, 126.8, 121.8, 121.3, 118.5, 109.9, 71.0, 68.3,
54.0, 38.5, 24.0, 21.9. HRMS (ESI): Calcd for (M+H)+ (C22H21NO) 316.1701, found
316.1658.
4.1.3
5-(4-Chlorophenyl)-3,4,4a,5,6,12c-hexahydro-2H-benzo[f]pyrano[3,2c]-quinolone
(CT2)
1
White solid, yield 62%. H NMR (500 MHz, DMSO-d6): δ 7.77 (d, J = 8.5 Hz, 1H),
7.64 (d, J = 7.5 Hz, 1H), 7.56 (d, J = 9.0 Hz, 1H), 7.53 (d, J = 8.5 Hz, 2H), 7.47 (t, J =
8 Hz, 2H), 7.39-7.36 (m, 1H), 7.13 (t, J = 7 Hz, 1H), 6.92 (d, J = 9 Hz, 1H), 6.47 (s,
1H), 4.88 (d, J = 2.5 Hz, 1H), 4.69 (d, J = 11.5 Hz, 1H), 3.98 (dd, J1 = 9 Hz, J2 = 3.5
Hz, 1H), 3.79 (t, J = 11 Hz, 1H), 1.98-1.95 (m, 1H), 1.86-1.70 (m, 2H), 1.30-1.28 (m,
2H). 13C NMR (125 MHz, DMSO-d6): δ143.6, 141.8, 134.0, 132.5, 130.4, 129.6,
11

128.9, 128.5, 127.1, 126.8, 121.9, 121.4, 118.4, 110.1, 70.8, 68.3, 53.3, 38.4, 24.0,
21.8. HRMS (ESI): Calcd for (M+H)+ (C22H20ClNO) 350.1312, found 350.1316.
4.1.4
5-(2,4-Dichlorophenyl)-3,4,4a,5,6,12c-hexahydro-2H-benzo[f]pyrano[3,2-c]quinolone
(CT3)
1
White solid, yield 68%. H NMR (500 MHz, DMSO-d6): δ 7.77 (d, J = 8.5 Hz, 1H),
7.69-7.64 (m, 3H), 7.55 (d, J = 8.5 Hz, 1H), 7.51 (dd, J1 = 8.5 Hz, J2 = 2 Hz, 1H),
7.40-7.37 (m, 1H), 7.14 (t, J = 7 Hz, 1H), 6.88 (d, J = 9 Hz, 1H), 6.51 (s, 1H), 5.17 (d,
J = 11.5 Hz, 1H), 4.91 (d, J = 2.5 Hz, 1H), 3.97-3.95 (m, 1H), 3.78 (t, J = 9.5 Hz, 1H),
2.11-2.09 (m, 1H), 1.87-1.75 (m, 2H), 1.38-1.27 (m, 2H). 13C NMR (125 MHz,
DMSO-d6): δ143.5, 139.4, 134.9, 133.9, 133.2, 131.8, 129.6, 128.9, 128.5, 127.2,
126.9, 121.9, 121.5, 118.3, 110.0, 70.8, 68.1, 49.1, 38.3, 24.0, 22.6. HRMS (ESI):
Calcd for (M+H)+ (C22H19Cl2NO) 384.0922, found 384.0932.
4.1.5 5-Phenyl-1,2,3,4,4a,5,6,12c-octahydro-1,4-methanobenzo[a]phenanthridine
(CT4)
1
White solid, yield 75%. H NMR (500 MHz, DMSO-d6): δ 7.69 (d, J = 8.5 Hz, 1H),
7.65 (d, J = 7.5 Hz, 1H), 7.48 (d, J = 8.5 Hz, 1H), 7.41-7.37 (m, 1H), 7.29 (d, J = 7.5
Hz, 2H), 7.23 (t, J = 7.5 Hz, 2H), 7.15 (t, J = 9 Hz, 2H), 6.96 (d, J = 8.5 Hz, 1H), 6.11
(s, 1H), 4.24 (d, J = 4 Hz, 1H), 3.14 (d, J = 8.5 Hz, 1H), 2.40 (s, 1H), 2.27 (dd, J1 =
8.5 Hz, J2 = 4 Hz, 1H), 2.24 (s, 1H), 1.73 (d, J = 9.5 Hz, 1H), 1.62-1.54 (m, 3H) ,
1.39-1.34 (m, 1H), 0.97 (d, J = 9.5 Hz, 1H). 13C NMR (125 MHz, DMSO-d6): δ146.7,
143.6, 133.1, 128.9, 128.6, 128.2, 127.6, 127.3, 127.0, 126.3, 122.4, 121.4, 119.0,
115.7, 57.9, 50.1, 45.1, 43.1, 40.7, 34.7, 30.2, 29.4. HRMS (ESI): Calcd for (M+H)+
(C24H23N) 326.1909, found 326.1861.
4.1.6
5-(4-Chlorophenyl)-1,2,3,4,4a,5,6,12c-octahydro-1,4-methanobenzo[a]-phenanthridi
12

ne (CT5)
1
White solid, yield 72%. H NMR (500 MHz, DMSO-d6): δ 7.69 (d, J = 8.5 Hz, 1H),
7.66 (d, J = 7.5 Hz, 1H), 7.48 (d, J = 9 Hz, 1H), 7.40-7.38 (m, 1H), 7.31-7.27 (m, 4H),
7.18-7.15 (m, 1H), 6.95 (d, J = 9 Hz, 1H), 6.14 (s, 1H), 4.28 (dd, J1 = 3.5 Hz, J2 = 2
Hz, 1H), 3.16 (d, J = 9 Hz, 1H), 2.38 (s, 1H), 2.26-2.24 (m, 2H), 1.71 (d, J = 10 Hz,
13
1H), 1.63-1.51 (m, 3H) , 1.40-1.36 (m, 1H), 0.96 (d, J = 10 Hz, 1H). C NMR (125
MHz, DMSO-d6): δ145.6, 143.3, 133.0, 131.5, 129.4, 128.9, 128.5, 128.3, 127.4,
126.4, 122.4, 121.5, 118.9, 115.9, 57.1, 49.9, 45.3, 43.2, 40.6, 34.7, 30.3, 29.3. HRMS
(ESI): Calcd for (M+H)+ (C24H22ClN) 360.1519, found 360.1523.
4.1.7
5-(2,4-dichlorophenyl)-1,2,3,4,4a,5,6,12c-octahydro-1,4-methanobenzo[a]-phenanthr
idine (CT6)
1
White solid, yield 70%. H NMR (500 MHz, DMSO-d6): δ 7.69 (d, J = 2.5 Hz, 1H),
7.67 (d, J = 3 Hz, 1H), 7.58 (d, J = 2 Hz, 1H), 7.54 (d, J = 9 Hz, 1H), 7.40 (t, J = 7.5
Hz, 1H), 7.22-7.16 (m, 2H), 7.12 (d, J = 8.5 Hz, 1H), 7.00 (d, J = 8.5 Hz, 1H), 6.22 (s,
1H), 4.63 (s, 1H), 3.14 (d, J = 8.5 Hz, 1H), 2.35 (s, 1H), 2.29 (s, 1H), 2.21 (d, J = 8.5
Hz, 1H), 1.67 (d, J = 9.5 Hz, 1H), 1.60-1.54 (m, 3H) , 1.42-1.37 (m, 1H), 0.98 (d, J =
13
9.5 Hz, 1H). C NMR (125 MHz, DMSO-d6): δ143.1, 143.0, 133.6, 132.9, 132.1,
129.3, 129.2, 129.0, 128.3, 127.8, 127.7, 126.5, 122.3, 121.7, 118.5, 115.3, 54.1, 49.4,
45.9, 44.2, 40.6, 34.7, 30.5, 28.9. HRMS (ESI): Calcd for (M+H)+ (C24H22ClN)
394.1129, found 394.1123.
4.2 Virtual screening workflow
The crystal structure of the AR LBD domain (PDB entry: 2PNU) was assessed in our
previous study and also used as the initial structure for the virtual screening in this
study [24]. The grid box of the protein for SBVS was generated and centered on the
co-crystallized ligand (EM-5744) in the LBP. The scaling factors for van der Waals
interaction and the maximum partial atomic charges were set to 1.0 and 0.25,
13

respectively. Subsequently, the compounds (~200,000) in the Specs database were
docked into the crystal structure of 2PNU, and the binding energies were scored and
ranked by the Glide SP scoring mode [40]. After that, the top ranked 50,000
compounds were submitted to the Glide XP scoring, and the top ranked 10,000
compounds were rescored by the VD-MM/GBSA method [37]. The AM1-BCC
charges were calculated for the ligands using the sqm module in Amber 18 simulation
package [41]. The protein-ligand systems were constructed using the antechamber and
tleap modules in Amber18 [41]. FF14SB force field and General Amber force field
(GAFF) were assigned to the proteins and ligands, respectively [42, 43]. Each
protein-ligand complex was immersed into a water box with a distance of 12 Å
extended from any solute atom. Three phases of minimizations were performed to
optimize each prepared system. At first, the whole protein and ligand were restrained
by 5 kcal mol-1 Å-2 elastic constant for 10,000 cycles (5,000 steps of steepest descent
and 5,000 steps of conjugate gradient minimizations). Thereafter, the backbone atoms
of the protein were restrained by 5 kcal mol-1 Å-2 elastic constant for 10,000 cycles
(5,000 steps of steepest descent and 5,000 steps of conjugate gradient minimizations).
At last, the whole system was relaxed without any restraint for 10,000 cycles (5,000
steps of steepest descent and 5,000 steps of conjugate gradient minimizations).
Subsequently, the optimized structures were submitted for the binding free energy
calculations based on the VD-MM/GBSA method [37]. The top ranked 2,000
compounds were clustered based on the structural similarity using the Canvas module
in Schrödinger [38]. Then, the binding poses of the clustered compounds were
carefully checked and filtered. Finally, the compounds were carefully checked and
filtered, and 32 compounds were purchased for subsequent bioassays.
4.3 MD simulation and VD-MM/GBSA of C18
The system preparation of the AR LBD and AT2 were processed by the protocol
reported in our previous studies, including calculation of the partial charges for AT2,
assignment of appropriate force field, addition of water molecules and ions, system
14

minimization, and MD heating and equilibrium [24, 44, 45]. Then the system was
submitted to 1 µs MD simulation in the isothermal isobaric (NPT) ensemble. The
CPPTRAJ module in AmberTools package for data analysis [46]. The last 0.2 µs MD
simulation trajectory with 1000 snapshots was submitted to VD-MM/GBSA for free
energy decomposition [37]. The VD-MM/GBSA calculations were performed by the
MMPBSA.py module in AmberTools package by modifying the sander module to
identify a new block of atomic dielectric constants in the revised AMBER system
topology [37, 47]. The entropies were not considered because of the low prediction
accuracy and expensive computational demand. The polar component of desolvation
was estimated by the modified GB model (GBOBC1) reported by Onufriev et al [48].
The exterior dielectric constant for solvent was set as default. The non-polar
component of desolvation was computed using the LCPO algorithm [49].
4.4 AR transcriptional activity assay
LNCaP-ARR2PB-eGFP was cultured in RPMI-1640 media, and the cells of LNCap
were starved with 5% charcoal-stripped serum (CSS) for 5 days. Thereafter, the cells
were plated into a 96-well plate with 3.5×104 cells/well, and continued to incubate at
37 in a 5% CO2 atmosphere for 24 h. Then, 5 nM DHT and intended concentrations
(10 µM for 32 compounds, 0-50 µM for C18 and its analogues) of compounds were
added to the prepared cells. Finally, the fluorescence intensities were determined after
treatment for 3 days (Synergy H1, BioTek. Excitation, 485 nm; Emission, 535 nm).
4.5 Measurement of prostate-specific antigen (PSA)
After the AR transcriptional activity assay was finished, the media supernatant (300 µl)
for each sample was sent to Cancer Hospital of University of Chinese Academy of
Sciences, Zhejiang Cancer Hospital (Hangzhou, Zhejiang) to measure the secreted
PSA using IMMULITE® 2000 XPi Immunoassay System (Siemens Ltd., Erlangen,
Germany).
15

4.6 Competitive ligand binding assay
The PolarScreen™ Androgen Receptor Competitor Assay Kit (Invitrogen Life
Technologies, Inc.) was utilized to determine the AR LBD binding affinity. Briefly, 10
µl 2× intended compounds (or DMSO, DHT, Enz) were dispersed in a low-volume
384-well plate. Then, 4× AR-LBD (GST) and 4× Fluormone AL (Flu-AL) Green
dissolved in complete AR Green Assay Buffer were added. Afterwards, aluminized
paper was used to protect the reagents from light, and after incubation for 4 h at room
temperature, a multi-function plate reader (Synergy H1, BioTek, Winooski, VT) was
used to measure the fluorescence polarization value (mP) for each well.
4.7 Cell proliferation assay
The cell proliferation for the cell lines of LNCAP, PC3, C4-2, and DU145 were
evaluated by 3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-2-H-tetrazolium bromide
(MTT) colorimetric assay. These cells were seeded in RPMI-1640 media (5%
charcoal-stripped serum) with 1-5×103 cells/well in 96-well plates (3×103, 1.5×103,
1.5×103, and 1.5×103 cells/well for LNCaP, PC3, C4-2 and DU145, respectively).
After incubation at 37 for 24 h, cells were then treated with 5nM DHT and serial
dilutions of Enz and AT2, followed by incubation of 72 h. Then, each well was added
with MTT (10 µl of 5 mg/ml) solutions and incubated for 4 h. After that, 100 µl of
triplex 10% SDS-0.1% HCl-PBS solutions were added to dissolve the formazan
crystals, and the plates were incubated overnight at 37 . Finally, the absorbance at
570 nm was measured with the reference wavelength at 650 nm using a
spectrophotometer (Eon, Bioteck, Winooski, VT).
4.8 Q-PCR
LNCaP cells were cultured in medium containing 5% charcoal-stripped serum (CSS)
in six-well plates and treated with Enz, AT2 (0.1, 1 or 10 µM), 5 nM DHT, or DMSO
for 24 h. After 48 h of the treatment, mRNA were extracted using EZ-10 DNA away
RNA Mini-Preps Kit (Sangon Biotech, Shanghai, China) and reversely transcribed
16

into cDNA using Hifair®
1st Strand cDNA Synthesis SuperMix for qPCR
(YEASEN, Shanghai, China), and finally detected by qPCR with qPCR SYBR Green
Master Mix (YEASEN, Shanghai, China). All the procedures followed the operation
manuals (QuantStudio 3, Applied Biosystems, Graphpad Prism 7.0).
4.9 Immunofluorescence
LNCaP cells were cultured in 12-well plates containing coverslips and incubated at
37
for 24 h. 10 µM of AT2 and Enz were individually added and then further
incubated for 12 h, and then 5 nM DHT was added and incubated for 90 mins.
Afterwards, the cells were fixed with 4 precooling 4% (vol/vol) paraformaldehyde,
and permeabilized with Triton X-100. Then the cells were incubated with AR
antibodies (#5153, Cell Signaling Technology) overnight after washing with PBS
three times. An Alexa-488 conjugated goat-anti rabbit lgG (#4412, Cell Signaling
Technology) diluted at 1:1000 was used as the secondary antibody. The counterstain
4',6-diamidino-2-phenylindole (DAPI) was utilized to visualize cell nucleus. The
images were taken at 60 magnification using Nikon AR fluorescence microscope,
followed by analysis with NIS-Elements Viewer (Northern Eclipse, Empix Imaging,
Inc.).
4.10 3T3 cytotoxicity assay
3T3 cells were seeded with RPMI-1640 medium at a density of 3000/well. After the
cells were attached, Enz, C18 and AT2 were seeded with various concentrations of
less than 10 µM. After incubation at 37 for 24 h, the medium was sucked out, 100
µl DMSO was added into each well, and the absorbance was measured after shaking
evenly for 5-10 minutes.
4.11 Statistical analysis
Data were analyzed statistically using the one-way analysis of variance test or
Student’s t-test via software of Graphpad prism 7.0 (GraphPad, San Diego, CA, USA).
17

Results were expressed as mean ± SEM with at least 3 replicates, and P < 0.05 was
considered as significant.
Acknowledgments
This study was supported by National Key R&D Program of China
(2016YFA0202900), Key R&D Program of Zhejiang Province (2020C03010),
National Natural Science Foundation of China (21575128, 81773632), and Zhejiang
Provincial Natural Science Foundation of China (LZ19H300001).
Reference
[1] F. Bray, J. Ferlay, I. Soerjomataram, R.L. Siegel, L.A. Torre, A. Jemal, Global cancer statistics 2018:
GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries,
CA Cancer J Clin, 68 (2018) 394-424.
[2] D. Li, W. Zhou, J. Pang, Q. Tang, B. Zhong, C. Shen, L. Xiao, T. Hou, A magic drug target:
androgen receptor, Med Res Rev, 39 (2019) 1485-1514.
[3] N. Elshan, M.B. Rettig, M.E. Jung, Molecules targeting the androgen receptor (AR) signaling axis
beyond the AR-Ligand binding domain, Med Res Rev, 39 (2019) 910-960.
[4] A.A. Shafi, A.E. Yen, N.L.J.P. Weigel, therapeutics, Androgen receptors in hormone-dependent and
castration-resistant prostate cancer, 140 (2013) 223-238.
[5] T.J. Maatman, M.K. Gupta, J.E. Montie, Effectiveness of castration versus intravenous estrogen
therapy in producing rapid endocrine control of metastatic cancer of the prostate, J Urol, 133
(1985) 620-621.
[6] C. Cai, S.P. Balk, Intratumoral androgen biosynthesis in prostate cancer pathogenesis and response
to therapy, Endocr Relat Cancer, 18 (2011) R175-182.
[7] M.H. Tan, J. Li, H.E. Xu, K. Melcher, E.L. Yong, Androgen receptor: structure, role in prostate
cancer and drug discovery, Acta Pharmacol Sin, 36 (2015) 3-23.
[8] D. Leveque, Enzalutamide in metastatic prostate cancer, N Engl J Med, 381 (2019) 1494.
[9] P.A. Watson, V.K. Arora, C.L. Sawyers, Emerging mechanisms of resistance to androgen receptor
inhibitors in prostate cancer, Nat Rev Cancer, 15 (2015) 701-711.
[10] J.M. van Dodewaard-de Jong, H.M.W. Verheul, H.J. Bloemendal, J.M.H. de Klerk, M.A. Carducci,
A.J.M. van den Eertwegh, New treatment options for patients with metastatic prostate cancer:
what is the optimal sequence?, Clin Genitourin Cancer, 13 (2015) 271-279.
[11] M.R. Smith, E.S. Antonarakis, C.J. Ryan, W.R. Berry, N.D. Shore, G. Liu, J.J. Alumkal, C.S.
Higano, E. Chow Maneval, R. Bandekar, C.J. de Boer, M.K. Yu, D.E. Rathkopf, Phase 2 study
of the safety and antitumor activity of apalutamide (ARN-509), a potent androgen receptor
antagonist, in the high-risk nonmetastatic castration-resistant prostate cancer cohort, Eur Urol,
70 (2016) 963-970.
[12] J.D. Joseph, N. Lu, J. Qian, J. Sensintaffar, G. Shao, D. Brigham, M. Moon, E.C. Maneval, I. Chen,
18

B. Darimont, J.H. Hager, A clinically relevant androgen receptor mutation confers resistance to
second-generation antiandrogens enzalutamide and ARN-509, Cancer Discov, 3 (2013)
1020-1029.
[13] T. Hara, J. Miyazaki, H. Araki, M. Yamaoka, N. Kanzaki, M. Kusaka, M. Miyamoto, Novel
mutations of androgen receptor: a possible mechanism of bicalutamide withdrawal syndrome,
Cancer Res, 63 (2003) 149-153.
[14] M.S. Ozers, B.D. Marks, K. Gowda, K.R. Kupcho, K.M. Ervin, T. De Rosier, N. Qadir, H.C.
Eliason, S.M. Riddle, M.S. Shekhani, The androgen receptor T877A mutant recruits LXXLL
and FXXLF peptides differently than wild-type androgen receptor in a time-resolved
fluorescence resonance energy transfer assay, Biochemistry, 46 (2007) 683-695.
[15] K. Steketee, L. Timmerman, A.C. Ziel-van der Made, P. Doesburg, A.O. Brinkmann, J. Trapman,
Broadened ligand responsiveness of androgen receptor mutants obtained by random amino acid
substitution of H874 and mutation hot spot T877 in prostate cancer, Int J Cancer, 100 (2002)
309-317.
[16] M. Bassetto, S. Ferla, F. Pertusati, S. Kandil, A.D. Westwell, A. Brancale, C. McGuigan, Design
and synthesis of novel bicalutamide and enzalutamide derivatives as antiproliferative agents for
the treatment of prostate cancer, Eur J Med Chem, 118 (2016) 230-243.
[17] A.V. Ivachtchenko, Y.A. Ivanenkov, O.D. Mitkin, A.A. Vorobiev, I.V. Kuznetsova, N.A. Shevkun,
A.G. Koryakova, R.N. Karapetian, A.S. Trifelenkov, D.V. Kravchenko, M.S. Veselov, N.V.
Chufarova, Design, synthesis and biological evaluation of novel 5-oxo-2-thioxoimidazolidine
derivatives as potent androgen receptor antagonists, Eur J Med Chem, 99 (2015) 51-66.
[18] M. Zuo, X. Xu, Z. Xie, R. Ge, Z. Zhang, Z. Li, J. Bian, Design and synthesis of indoline
thiohydantoin derivatives based on enzalutamide as antiproliferative agents against prostate
cancer, Eur J Med Chem, 125 (2017) 1002-1022.
[19] J. Zhou, G. Geng, Q. Shi, F. Sauriol, J.H. Wu, Design and synthesis of androgen receptor
antagonists with bulky side chains for overcoming antiandrogen resistance, J Med Chem, 52
(2009) 5546-5550.
[20] C.A. Lipinski, Lead- and drug-like compounds: the rule-of-five revolution, Drug Discov Today
Technol, 1 (2004) 337-341.
[21] C.H. Song, S.H. Yang, E. Park, S.H. Cho, E.Y. Gong, D.B. Khadka, W.J. Cho, K. Lee,
Structure-based virtual screening and identification of a novel androgen receptor antagonist, J
Biol Chem, 287 (2012) 30769-30780.
[22] H. Li, X. Ren, E. Leblanc, K. Frewin, P.S. Rennie, A. Cherkasov, Identification of novel androgen
receptor antagonists using structure- and ligand-based methods, J Chem Inf Model, 53 (2013)
123-130.
[23] L. Wang, T. Song, X. Wang, J. Li, Discovery and identification of pyrazolopyramidine analogs as
novel potent androgen receptor antagonists, Front Pharmacol, 9 (2018) 864.
[24] W. Zhou, M. Duan, W. Fu, J. Pang, Q. Tang, H. Sun, L. Xu, S. Chang, D. Li, T. Hou, Discovery of
novel androgen receptor ligands by structure-based virtual screening and bioassays, Genomics
Proteomics Bioinformatics, 16 (2018) 416-427.
[25] S. Forli, R. Huey, M.E. Pique, M.F. Sanner, D.S. Goodsell, A.J. Olson, Computational
protein-ligand docking and virtual drug screening with the AutoDock suite, Nat Protoc, 11
(2016) 905-919.
[26] M. Zheng, J. Zhao, C. Cui, Z. Fu, X. Li, X. Liu, X. Ding, X. Tan, F. Li, X. Luo, K. Chen, H. Jiang,
19

Computational chemical biology and drug design: Facilitating protein structure, function, and
modulation studies, Med Res Rev, 38 (2018) 914-950.
[27] Z. Wang, H. Sun, X. Yao, D. Li, L. Xu, Y. Li, S. Tian, T. Hou, Comprehensive evaluation of ten
docking programs on a diverse set of protein-ligand complexes: the prediction accuracy of
sampling power and scoring power, Phys Chem Chem Phys, 18 (2016) 12964-12975.
[28] C.N. Cavasotto, N.S. Adler, M.G. Aucar, Quantum chemical approaches in structure-based virtual
screening and lead optimization, Front Chem, 6 (2018) 188.
[29] I.A. Guedes, F.S.S. Pereira, L.E. Dardenne, Empirical scoring functions for structure-based virtual
screening: applications, critical aspects, and challenges, Front Pharmacol, 9 (2018) 1089.
[30] J. Li, A. Fu, L. Zhang, An overview of scoring functions used for protein-ligand interactions in
molecular docking, Interdiscip Sci, 11 (2019) 320-328.
[31] S. Genheden, U. Ryde, The MM/PBSA and MM/GBSA methods to estimate ligand-binding
affinities, Expert Opin Drug Discov, 10 (2015) 449-461.
[32] S.I. Virtanen, S.P. Niinivehmas, O.T. Pentikainen, Case-specific performance of MM-PBSA,
MM-GBSA, and SIE in virtual screening, J Mol Graph Model, 62 (2015) 303-318.
[33] E. Wang, H. Sun, J. Wang, Z. Wang, H. Liu, J.Z.H. Zhang, T. Hou, End-point binding free energy
calculation with mm/pbsa and mm/gbsa: Strategies and applications in drug design, Chem Rev,
119 (2019) 9478-9508.
[34] X. Liu, L. Peng, J.Z. Zhang, Accurate and efficient calculation of protein-protein binding free
energy-interaction entropy with residue type specific dielectric constants, J. Chem. Inf. Model.,
(2018) doi: 10.1021/acs.jcim.1028b00248.
[35] T. Venken, D. Krnavek, J. Munch, F. Kirchhoff, P. Henklein, M. De Maeyer, A. Voet, An optimized
MM/PBSA virtual screening approach applied to an HIV-1 gp41 fusion peptide inhibitor,
Proteins, 79 (2011) 3221-3235.
[36] I.S. Moreira, P.A. Fernandes, M.J. Ramos, Computational alanine scanning mutagenesis--an
improved methodological approach, J Comput Chem, 28 (2007) 644-654.
[37] E. Wang, G. Weng, H. Sun, H. Du, F. Zhu, F. Chen, Z. Wang, T. Hou, Assessing the performance
of the MM/PBSA and MM/GBSA methods. 10. Impacts of enhanced sampling and variable
dielectric model on protein-protein Interactions, Phys Chem Chem Phys, 21 (2019)
18958-18969.
[38] S.J.L.N.Y. Canvas, NY, USA, version 3.1. 011, (2017).
[39] H. Borgmann, N. Lallous, D. Ozistanbullu, E. Beraldi, N. Paul, K. Dalal, L. Fazli, A. Haferkamp,
P. Lejeune, A. Cherkasov, M.E. Gleave, Moving towards precision urologic oncology: targeting
enzalutamide-resistant prostate cancer and mutated forms of the androgen receptor using the
novel inhibitor darolutamide (ODM-201), Eur Urol, 73 (2018) 4-8.
[40] L.J.S. Schrödinger, LLC: New York, NY, USA, Glide, (2017).
[41] T.-S. Lee, D.S. Cerutti, D. Mermelstein, C. Lin, S. LeGrand, T.J. Giese, A. Roitberg, D.A. Case,
R.C. Walker, D.M.J.J.o.c.i. York, modeling, GPU-accelerated molecular dynamics and free
energy methods in Amber18: performance enhancements and new features, 58 (2018)
2043-2050.
[42] J. Wang, R.M. Wolf, J.W. Caldwell, P.A. Kollman, D.A. Case, Development and testing of a
general amber force field, Journal of computational chemistry, 25 (2004) 1157-1174.
[43] J.A. Maier, C. Martinez, K. Kasavajhala, L. Wickstrom, K.E. Hauser, C. Simmerling, ff14SB:
improving the accuracy of protein side chain and backbone parameters from ff99SB, J Chem
20

Theory Comput, 11 (2015) 3696-3713.
[44] M. Duan, N. Liu, W. Zhou, D. Li, M. Yang, T. Hou, Structural diversity of ligand-binding
androgen receptors revealed by microsecond long molecular dynamics simulations and
enhanced sampling, J Chem Theory Comput, 12 (2016) 4611-4619.
[45] N. Liu, W. Zhou, Y. Guo, J. Wang, W. Fu, H. Sun, D. Li, M. Duan, T. Hou, Molecular dynamics
simulations revealed the regulation of ligands to the interactions between androgen receptor and
its coactivator, J Chem Inf Model, 58 (2018) 1652-1661.
[46] D.R. Roe, T.E. Cheatham, 3rd, PTRAJ and CPPTRAJ: software for processing and analysis of
molecular dynamics trajectory data, J Chem Theory Comput, 9 (2013) 3084-3095.
[47] B.R. Miller, 3rd, T.D. McGee, Jr., J.M. Swails, N. Homeyer, H. Gohlke, A.E. Roitberg,
MMPBSA.py: an efficient program for end-state free energy calculations, J. Chem. Theory
Comput., 8 (2012) 3314-3321.
[48] A. Onufriev, D. Bashford, D.A. Case, Modification of the generalized Born model suitable for
macromolecules, J Phys Chem B, 104 (2000) 3712-3720.
[49] J. Weiser, P.S. Shenkin, W.C. Still, Approximate atomic surfaces from linear combinations of
pairwise overlaps (LCPO), Journal of Computational Chemistry, 20 (1999) 217-230.
Figure Legends
Figure 1. The schematic workflow of SBVS.
Figure 2. Biological evaluation of the compounds identified by VS. (A)
LNCaP-ARR2PB-eGFP-based transcriptional activity assay to determine the
antagonistic activities for the 32 tested compounds (10 µM); (B) PolarScreenTM AR
competitor assay to assess the AR binding affinities of the 3 identified active
compounds (10 µM); (C) The AR binding affinity of the hit C18; (D) Transcriptional
activity assay of C18; (E) C18 reduces the PSA expression in LNCap cells.
Figure 3. Rational design of the C18 analogues. (A) RMSDs of the backbone atoms
21

of the AR LBD and the heavy atoms of C18 as a function of MD simulation time; (B)
The 10 key residues for the binding of C18 predicted by VD-MM/GBSA. (C)
Structural analysis of the 10 key residues for the interactions between the AR LBD
and C18.
Figure 4. AT2 can target the AR-LBD and reduce the PSA expression. (A)
PolarScreenTM AR Competitor Assay to assess the C18 analogues (10 µM); (B) AT2
reduces the PSA expression in LNCap cells with a dose-dependent fashion; (C) The
relative mRNA expression of PSA; (D) The relative mRNA expression of TMPSS2;
(E) Cytotoxicity assay effects of Enz, C18 and AT2 against NIH-3T3 cells.
Figure 5. AT2 inhibits the AR nuclear translocation induced by DHT (scale bar = 50
µm).
Figure 6. Antiproliferative effects of Enz, C18, and AT2 against the cells of (A)
LNCap, (B) PC3, (C) C4-2, and (D) DU145 were determined by MTT assay. The
cells were treated by different concentrations of the tested compounds for 72 h.
22

Table 1. IC50 (µM) of the 21 analogues of C18 determined by transcriptional activity assay.
Comp.
AT1
R1
IC50 (µM) Comp.
R1
R2
IC50 (µM)
Comp.
CT1
R1
A
IC50 (µM)
N/A
N/A
BT1
BT2
BT3
BT4
BT5
0.22
0.15
N/A
N/A
N/A
N/A
N/A
N/A
20.35
5.60
AT2
AT3
AT4
AT5
CT2
CT3
CT4
CT5
0.82
1.92
7.45
N/A
N/A
AT6
0.15
BT6
CT6
23

N/A
2.40
AT7
AT8
BT7
C18
0.26
0.69
0.036
Enz
Note: N/A means not applicable
24

Figure 1
25

Figure 2
26

Figure 3
27

Figure 4
28