Assessmen o he Binding Pa e ns o Endoc ine Dis up ing
Chemicals in Complex wi h Es ogen and And ogen Recep o s by
Le e aging he Asclepios Enalos KNIME Nodes
Ha alampos Tzoupis, Michail Papadou akis, Kons an inos D. Papa asileiou, Oli e Bu k,
Volke M. Lauschke, And eas Tsoumanis, Geo gia Melag aki, and An eas A an i is*
Ci e This: J. Chem. In . Model. 2025, 65, 11950−11964
Read Online
ACCESS Me ics & Mo e A icle Recommenda ions *
sı Suppo ing In o ma ion
ABSTRACT: Endoc ine dis up ing chemicals (EDCs) ha e been shown o
media e me abolic dis up ions in human cells and ha e been associa ed wi h
se e e ad e se heal h e ec s. By an agonizing he ho mones ha ac on nuclea
ho mone ecep o s, like he es ogen ecep o α(ERα) and he and ogen
ecep o (AR), hese chemicals dis up he egula ion o a ious biochemical
p ocesses, he eby ad e sely a ec ing me abolic homeos asis. The exp ession o
es ogen and and ogen ecep o s in he li e and panc eas, which play an
impo an ole in lipid and glucose homeos asis egula ion, has made hem
p ime a ge s a ec ed by EDCs. The di e en chemical s uc u es o EDCs
impose limi a ions on elucida ing hei binding mechanisms in nuclea
ecep o s. In his con ex , in silico ools a e able o highligh he po en ial
in e ac ions be ween he chemicals and he ecep o s. The aim o his s udy is o
apply molecula simula ion and expe imen al echniques o iden i y common
pa e ns in he binding p ocess o selec ed EDCs o ERαand AR and, hus,
pinpoin key elemen s ha could be cha ac e ized as molecula ini ia ing e en s (MIE). MM-GBSA and alchemical ela i e binding
ee ene gy (RBFE) calcula ions ha e e i ied he ends obse ed in he expe imen al assays ega ding he binding a ini y o
bisphenol compounds. The indings ha con i m he ag eemen be ween compu a ional and expe imen al me hods o e a
amewo k o u u e s udies on he beha io o EDCs wi h o he me abolically ele an ecep o s.
■INTRODUCTION
Endoc ine dis up ing chemicals (EDCs) a e de ined as
“exogenous subs ances ha induce ad e se heal h e ec s in
in ac o ganisms”.
1,2
EDCs bind o a ious p o ein a ge s
inside he cells and cause dis up ions in me abolic pa hways.
As hei name implies, EDCs exe hei ac ion mainly by
a ec ing ho mone syn hesis and he homeos asis o he
endoc ine sys em.
3
These chemicals ha e also been linked o
ep oduc i e sys em dys unc ion and de elopmen al diso de s
in humans.
4
The mos common pa hway o EDCs o exe
hei unc ion is h ough binding o nuclea ho mone ecep o s
such as es ogen and and ogen ecep o s.
5
The binding is
acili a ed by hei di e se chemical s uc u es and hei
molecula p ope ies, pa icula ly hei s uc u al simila i ies
wi h endogenous ligands such as s e oid ho mones.
5
Upon
binding o ho mone ecep o s, EDCs can cause al e a ions in
hei unc ionali y and inhibi o ac i a e ho monal esponses.
6
Al hough EDCs ha e ini ially been epo ed o ac mainly
h ough in e ac ions wi h he es ogen and and ogen ecep o s,
ecen esea ch has shown ha hey exe di e se ac ions
h ough in e ac ions wi h ecep o s like he pe oxisome
p oli e a o -ac i a ed ecep o γ(PPARγ)
7
and he p egnane
X ecep o (PXR).
8
The chemical di e ences obse ed o he
a ious EDCs mi o he a iabili y in hei ac ion and show
ha hey can ha e pleio opic e ec s. Fo ins ance, dichlo o-
diphenyl- ichlo oe hane (DDT) can ha e an agonis ic ac ion
agains and ogens h ough i s me aboli es.
9−12
The es ogen ecep o α(ERα) is p ima ily exp essed in he
ep oduc i e issues, whi e adipose issue, li e , and b eas ,
13
while he and ogen ecep o (AR) has been shown o ha e an
impo an ole in a ious sys ems like he ep oduc i e,
ca dio ascula , immune, o ne ous sys em.
14,15
Bo h ecep o s
belong o he same amily o p o eins and, hus, sha e
s uc u al cha ac e is ics ha e lec common unc ional and
egula o y aspec s. The common unc ional mo i o hese
ecep o s comp ises wo dis inc a eas, he ligand binding
domain (LBD) and he DNA-binding domain (DBD), ha a e
sepa a ed by a hinge egion.
16−18
Es ogens in he li e can
Recei ed: July 7, 2025
Re ised: Sep embe 12, 2025
Accep ed: Sep embe 25, 2025
Published: Oc obe 20, 2025
A iclepubs.acs.o g/jcim
© 2025 The Au ho s. Published by
Ame ican Chemical Socie y 11950
h ps://doi.o g/10.1021/acs.jcim.5c01437
J. Chem. In . Model. 2025, 65, 11950−11964
This a icle is licensed unde CC-BY 4.0
Downloaded ia 46.251.101.132 on No embe 24, 2025 a 10:27:59 (UTC).
See h ps://pubs.acs.o g/sha ingguidelines o op ions on how o legi ima ely sha e published a icles.
ha e dual e ec s ei he by p omo ing o inhibi ing li e
unc ions
19
and cance de elopmen .
20
Mo eo e , es ogens
ha e been associa ed wi h glucose me abolism and panc ea ic
egula ion o glycemic con ol.
21
Simila ly, he AR has been
associa ed wi h cance de elopmen ,
22
as well as o he li e
23
and panc ea ic diso de s.
24,25
The di e se unc ionali y epo ed o hese wo ecep o s
has made hem he p ima y ocus as bioma ke s o EDC isk
assessmen . Binding aspec s o di e en g oups o EDCs, such
as PFAS,
26,27
ph hala es,
28,29
o bisphenols,
30−32
ha e been
s udied ia he implemen a ion o compu a ional app oaches.
The esul s iden i ied possible aspec s o he binding
mechanism o hese chemicals, highligh ing a ious s uc u al
componen s o he ecep o s.
Se e al molecula dynamics (MD) s udies ha e been
p e iously pe o med in an e o o ad ance ou unde -
s anding o how small molecules in e ac wi h ERαand AR,
33
ocusing speci ically on hei LBDs. Fo ERα, ea ly wo k by
McGee e al.
34
demons a ed how emo al o an agonis s
a ec s ecep o con o ma ion, while la e s udies such as hose
by Ng
35
and Pu anik e al.
36
highligh ed ligand-speci ic
luc ua ions, pa icula ly in helix 12 (H12), which is c ucial
o ac i a ion.
37
Mo e ecen wo k by Sinyania e al.
38
in es iga ed la onoids o e eal how s uc u al dynamics and
binding ene ge ics go e n ecep o modula ion. On he o he
hand, AR- ocused s udies ha e shown ha agonis s and
an agonis s di e en ially in luence AR lexibili y and coac-
i a o binding. Wahl and Smiesko
39
add essed he challenge
Figu e 1. (A) Enalos Asclepios KNIME wo k low o he p epa a ion and execu ion o he molecula docking pe o med in his s udy wi h
Au odock Vina and molecula dynamics (MD) simula ions. (B) Schema ic ep esen a ion o he he modynamic cycle o he pe u ba ion o
ligand A (bisphenol A) o ligand B (bisphenol S). The ela i e binding ee ene gy (ΔΔGbind) o bisphenol S wi h espec o bisphenol A can be
calcula ed ei he di ec ly (ho izon al p ocesses, ΔGB−ΔGA) o ia an alchemical pa h ( e ical p ocesses, ΔG2−ΔG1).
Jou nal o Chemical In o ma ion and Modeling pubs.acs.o g/jcim A icle
h ps://doi.o g/10.1021/acs.jcim.5c01437
J. Chem. In . Model. 2025, 65, 11950−11964
11951
o lacking an agonis -bound c ys al s uc u es o he AR by
employing MD simula ions o gene a e an agonis con o ma-
ions. This app oach imp o ed he i ual sc eening accu acy
o AR an agonis s and highligh ed he ole o H12 dynamics
in dis inguishing be ween agonis and an agonis binding
modes. Gim e al.
40
and Liu e al.
41
used MD o elucida e how
ligand binding egula es AR ac i a ion and co egula o
in e ac ions. Addi ionally, e o s by Xu e al.
42
ex ended he
analysis o DNA-binding beha io and dime iza ion, emphasiz-
ing he s uc u al in icacies unde lying AR unc ion.
Wi h espec o EDCs, a comp ehensi e s udy by Tan e
al.
43
analyzed o e 4000 compounds using molecula docking
and MD simula ions o iden i y s uc u al agmen s
esponsible o binding and ac i a ion o ERα. They disco e ed
ha p ima y and seconda y EDC agmen s acili a e ecep o
binding, while e ia y agmen s de e mine he ac i i y ype
(agonis , an agonis , o mixed). Addi ionally, a QM/MM s udy
ocused on bisphenol A (BPA) demons a ed ha BPA binds
o ERα’s ac i e si e simila ly o 17β-es adiol, leading o
con o ma ional changes associa ed wi h es ogenic ac i i y.
44
Simila ly, Huang e al.
45
explo ed he po en ial binding
mechanism o bisphenol A o AR, while a ecen s udy by
Pa hak e al.
46
explo ed he binding mechanics o o he
bisphenol analogues.
The scope o he p esen s udy was o employ molecula
docking calcula ions, MD simula ions, molecula mechanics-
gene alized bo n su ace a ea (MM-GBSA) analysis, and
alchemical ela i e binding ee ene gy calcula ions me hods
o in es iga e and compa e he binding pa e ns o common
EDCs wi h di e se chemical s uc u es om di e en chemical
g oups (e.g., bisphenols, ph hala es, and PFAS). The
simula ions p o ide he necessa y in o ma ion o iden i ica-
ion o po en ial common binding pa e ns be ween he
di e en EDCs. They we e co obo a ed by expe imen al
assays employing ERαand AR epo e gene cons uc s in
ans ec ed HEK293 cells, allowing alida ion o bo h he
binding ac i i y and unc ional esponse. Consequen ly, his
in o ma ion can be employed o iden i y molecula ini ia ing
e en s (MIEs) and, in combina ion wi h in i o and in i o
s udies, assis in de eloping a eliable EDC isk assessmen
ool.
■MATERIALS AND METHODS
P o ein and Compound S uc u e P epa a ion. The
c ys al s uc u es o (i) he es ogen ecep o αin complex
wi h he d ug amoxi en (PDB ID: 3e )
17
and (ii) he
and ogen ecep o in complex wi h 5α-dihyd o es os e one
(PDB ID: 1 7 )
18
we e employed o he molecula docking
and molecula dynamics simula ions. The p epa a ion o he
p o ein s uc u es was pe o med wi h he AsclepiosPDBFixe
KNIME node (Figu e 1A). The A chain o he ecep o s was
e ained, and all he e oa oms we e emo ed, ollowed by he
addi ion o hyd ogen a oms a neu al pH (i.e., pH = 7.4). All
p o ein esidues in he iles ha e been enumbe ed s a ing
om he i s esidue.
The s uc u es o he EDCs s udied (Table 1) we e
downloaded om he PubChem da abase.
47
Open Babel
48
was employed o he con e sion o he chemical s uc u es o
he SDF o ma . The hyd ogen a oms we e added a pH 7.4,
using he AsclepiosAddHyd ogen node, and he 2D s uc u es
we e con e ed in o low-ene gy 3D con o ma ions by
implemen ing he AsclepiosGene a e3Dcoo dina es node
(Figu e 1A). All he nodes equi ed o ligand and p o ein
s uc u e p epa a ion a e inco po a ed in he Enalos Asclepios
KNIME pla o m (Figu e 1A).
49−51
Molecula Docking Simula ions. All simula ions we e
pe o med using Au odock Vina
52
so wa e as implemen ed in
he Enalos Asclepios KNIME nodes (Figu e 1A). Fo ERα, he
coo dina es (x,y, and z) o he ac i e si e we e 30.2, −2.9, and
23.4 and he box size was se a 21 Å ×24.5 Å ×19.5 Å, while
o AR, he coo dina es we e −0.9, 3.1, and 36.9 and he box
size was 31 Å ×24 Å ×15 Å, espec i ely. De ailed
in o ma ion is a ailable in he Suppo ing In o ma ion. To
assess he alidi y o he docking p o ocol, we ha e isola ed he
c ys al s uc u es o amoxi en and 5α-dihyd o es os e one,
espec i ely, and edocked he compounds in ERαand AR
ollowing he same p o ocol.
Molecula Dynamics (MD) Simula ions. The p epa a-
ion o he p o ein-EDC complexes was pe o med using he
wo k low depic ed in Figu e 1A. The bes sco ing docking
con o ma ion o each compound was employed as he ini ial
con o ma ion o he MD simula ions. All simula ions we e
pe o med wi h he OpenMM 7.5 so wa e
53
as pa o he
Asclepios KNIME wo k low. The An echambe Plus and
AsclepiosAmbe Sys emsP ep nodes, implemen ed in he
Enalos Asclepios KNIME nodes, we e used o cons uc ing
he pa ame e s o each p o ein-compound complex. The
AMBER14SB
54
and he Gene alized Ambe Fo ce Field
(GAFF)
55
we e employed o he building o he pa ame e s
o he p o ein and espec i e ligands. All p o ein-EDC
complexes we e sol a ed in wa e , using he SPC/E
56,57
model and a unca ed oc ahed on sol en box wi h a 10 Å
bu e dis ance. The SPC/E model has been epo ed o be e
ep oduce bulk wa e p ope ies.
58,59
Pe iodic bounda y
condi ions we e applied in he sys em, and he o al cha ge
o he sys ems was neu alized by adding he espec i e
numbe o Na+/Cl−ions.
The sys ems we e subjec ed o ene gy minimiza ion wi h
posi ional es ain s on he p o eins and ligands by using a
ha monic o ce cons an ha was g adually educed un il ully
Table 1. Endoc ine Dis up ing Chemicals Employed in he
P esen S udy
Jou nal o Chemical In o ma ion and Modeling pubs.acs.o g/jcim A icle
h ps://doi.o g/10.1021/acs.jcim.5c01437
J. Chem. In . Model. 2025, 65, 11950−11964
11952
abolished. The hea ing o he sys em was pe o med o he
a ge empe a u e o 300 K o 200 ps, in he canonical
ensemble (NVT), using he Lange in he mos a ,
60
ollowed
by p essu e equilib a ion a 1 a m in isoba ic−iso he mal
condi ions (NPT ensemble). Finally, he p oduc ion un was
pe o med o 200 ns (3 independen uns o each complex),
wi h no posi ional es ain s imposed on he p o eins and
ligands in he NPT ensemble.
Molecula Mechanics-Gene alized Bo n Su ace A ea
Analysis (MM-GBSA). The es ima ion o he ela i e ligand
binding a ini ies was pe o med using he MM-GBSA
me hodology. MM-GBSA has been shown o pe o m well in
p edic ing ela i e binding a ini ies, despi e no always being
able o ep oduce he expe imen al esul s co ec ly.
61
The
calcula ions o he p o ein complexes we e conduc ed wi h he
AsclepiosAmbe MMPBGBSA node, using he Enalos Asclepios
implemen a ion o he Ambe Tools21 MM-GBSA.py sc ip
62
o he espec i e KNIME wo k low node o e he 1000 las
ajec o y ames. All igu es we e p epa ed using he PyMOL
Open-Sou ce so wa e.
63
Alchemical Rela i e Binding F ee Ene gy Calcula-
ions. A nonequilib ium ee ene gy wo k low
64−66
was
employed o calcula e he ela i e binding ee ene gies o
bisphenol F o ERαand AR, in compa ison wi h bisphenols S
and A. An example o he he modynamic cycle applied in his
s udy is shown in Figu e 1B. The pa ame e s o he p o eins
and he ligands we e buil using he same me hod desc ibed
abo e. Following ligand pa ame iza ion, hyb id s uc u es and
opologies o each ligand pai we e gene a ed using he pmx
67
module o GROMACS 2024.4.
68
Simula ion sys ems o bo h
he sol a ed ligands and he ligand−p o ein complexes we e
p epa ed by placing he molecules in cubic boxes, ensu ing a
minimum dis ance o 15 Å be ween he solu e and he box
bounda ies.
Fo each ligand pai , simula ions we e ca ied ou o bo h
physical end s a es�s a e A and s a e B�co esponding o
ligands 1 and 2, espec i ely. The sys ems we e i s subjec ed
o ene gy minimiza ion, ollowed by a 500 ps equilib a ion in
he NVT ensemble a 310 K. This was ollowed by a 10 ns
p oduc ion un in he NPT ensemble a 310 K and a p essu e
o 1 ba .
Tempe a u e con ol in he simula ions was achie ed using
Lange in dynamics
69
wi h a collision equency o 1 ps−1.
P essu e was main ained a 1 ba using he Pa inello−Rahman
ba os a ,
70
wi h a ime cons an o 2 ps and a comp essibili y o
4.5 ×10−5ba −1. All bonds in ol ing hyd ogen a oms we e
cons ained using he LINCS algo i hm.
71
Long- ange elec o-
s a ic in e ac ions we e handled using he Pa icle Mesh Ewald
(PME) me hod,
72
wi h a eal-space cu o o 12 Å, a Fou ie
g id spacing o 1 Å, and a ela i e in e ac ion s eng h a he
cu o se o 10−6. Sho ange elec os a ic and an de Waals
in e ac ions we e calcula ed wi h a cu o o 12 Å and a
swi ching dis ance o 10 Å. Fo each p o ein−ligand complex,
he nonequilib ium wo k low was pe o med in iplica e, and
he epo ed binding ee ene gies (ΔGbind a e age) ep esen he
mean ac oss he eplica es. S a is ical unce ain ies a e
p o ided by p opaga ing he e o s o he espec i e
calcula ions.
The de ailed p o ocols employed o he MD simula ions,
MM-GBSA, and alchemical ee ene gy calcula ions a e
desc ibed ex ensi ely in he Suppo ing In o ma ion.
Chemicals and Reagen s. DMSO, bisphenol A (BPA),
bisphenol F (BPF), and bisphenol S (BPS) we e pu chased
om Sigma-Ald ich (Tau ki chen, Ge many). 11-Ke odihy-
d o es os e one (11-ke oDHT) was ob ained om MedChem
T onica (Sollen una, Sweden). Compounds o chemical
ea men o ans ec ed cells we e p epa ed as 1000×s ocks
in DMSO. Dulbecco’s modi ied Eagle’s medium (DMEM) was
p o ided by The mo Fishe Scien i ic (Wal ham, MA, USA). L-
Glu amine, sodium py u a e, and penicillin−s ep omycin
mix u e we e p o ided by Biozym (Hessisch Oldendo ,
Ge many). Fe al bo ine se um (FBS) was pu chased om
Sigma-Ald ich. PEI MAX ans ec ion-g ade linea polye hyle-
nimine hyd ochlo ide MW 40,000 (PEI MAX 40K) was
ob ained om Polysciences (Wa ing on, PA, USA).
Cell Cul u e. HEK293 cells (ACC305, DSMZ, B aunsch-
weig, Ge many) we e cul i a ed a 37 °C in 5% CO2in
DMEM supplemen ed wi h 10% FBS, 2 mM L-glu amine, 1
mM sodium py u a e, 100 U/mL penicillin, and 100 μg/mL
s ep omycin. Fo exposu e s udies, cells we e cul i a ed in
phenol ed- ee DMEM, supplemen ed wi h 1% dex an-coa ed
cha coal- ea ed FBS.
Plasmids. Pa o he open eading ame (NM_000125) o
human ERα(ESR1), encoding amino acids 251−595 (hinge
and LBD), was ampli ied om human b eas cDNA by PCR
using app op ia e p ime s, which in oduced EcoRI and BamHI
es ic ion si es, espec i ely. The PCR p oduc was cloned
in o EcoRI/BamHI diges ed ec o pM (Taka a Bio Eu ope,
Sain -Ge main-en-Laye, F ance). The esul ing plasmid (pM-
ESR1(251−595) encodes a usion p o ein o N- e minal
GAL4-DBD and he a o emen ioned egion o ERα. Pa s o
he open eading ame o human AR (NM_000044),
encoding amino acids 1−503 (N- e minal egion wi h AF1)
and 640−920 (pa o hinge and LBD), we e ampli ied by PCR
om human es is cDNA and cloned in o pVP16(AD) (Taka a
Bio Eu ope) and pM, espec i ely. The esul ing plasmids
pVP16-AR (1−503) and pM-AR (640−920) encode usion
p o eins o N- e minal VP16 ac i a ion domain and GAL4-
DBD and he espec i e egions o AR. The pGL4-G5 i e ly
luci e ase epo e gene plasmid was desc ibed p e iously.
73
The Renilla luci e ase exp ession plasmid pGL4.75 [hRluc/
CMV] was pu chased om P omega (Madison, WI, USA).
Cell Viabili y. HEK293 cells we e seeded a 40,000 cells
pe well in o 96-well pla es. On he ollowing day, cells we e
ea ed wi h 10, 30, o 100 μM chemicals o 24 h. Cell
iabili ies we e de e mined as desc ibed p e iously using he
CellTi e -Glo 2.0 assay (P omega).
73
Each o he h ee
independen expe imen s was pe o med in echnical ipli-
ca es.
Mammalian One-Hyb id and Two-Hyb id Assays. Pe
well o a 96-well pla e, a plasmid DNA mix u e consis ing o
0.26 μg o pGL4-G5, 0.04 μg o pM-ESR1 (251−495), and
0.005 μg o pGL4.75 [hRluc/CMV] was dilu ed wi h 150 mM
NaCl o a inal olume o 25 μL. In he case o he N−C-
e minal in e ac ion assay o AR, he ESR1 e ec o plasmid
was eplaced by 0.02 μg each o pVP16-AR (1−503) and pM-
AR (640−920). Aliquo (0.6 μL) o 1 mg/mL PEI MAX 40K
solu ion was dilu ed o 25 μL wi h 150 mM NaCl. The dilu ed
PEI MAX 40K was added o he dilu ed DNA mix u e, and he
mix u e was incuba ed a oom empe a u e o 15 min. In
pa allel, HEK 293 cells we e adjus ed o 40,000 cells in 200 μL
o cul u e medium pe well. The ans ec ion mix u e (50 μL)
was added o he cell suspension, gen ly mixed, and pipe ed
in o a well o a 96-well pla e. The nex day, he ans ec ed cells
we e ea ed wi h chemicals o 24 h be o e cell lysis wi h 50
μL pe well o passi e lysis bu e (P omega). Fi e ly and
Jou nal o Chemical In o ma ion and Modeling pubs.acs.o g/jcim A icle
h ps://doi.o g/10.1021/acs.jcim.5c01437
J. Chem. In . Model. 2025, 65, 11950−11964
11953
Renilla luci e ase ac i i ies we e measu ed as desc ibed
p e iously.
73
Resul s we e no malized o ans ec ion
e iciency by di iding i e ly luci e ase ac i i y by Renilla
luci e ase ac i i y, measu ed om he same well. Assays we e
pe o med in echnical iplica e in each independen expe i-
men . The numbe o independen expe imen s (i.e., biological
eplica es) is indica ed in he espec i e igu e legends.
Concen a ion esponse cu es we e i ed wi h nonlinea
eg ession using he o mula wi h a a iable slope ( ou
pa ame e s) o he o mula wi h a s anda d slope ( h ee
pa ame e s) o G aphPad P ism 10.4.1 (G aphPad So wa e,
Bos on, MA, USA), as indica ed in he espec i e igu e
legends.
■RESULTS
Molecula Docking Calcula ions. The calcula ed binding
a ini ies o he molecula docking calcula ions o he EDCs
conside ed in bo h ecep o s a e p esen ed in Table S1. The
esul s showed ha he di e en compounds do no p esen
any g ea a ia ions in he binding o AR and ERα(Figu e
S2A). The na i e ligands o bo h AR and ERα( es os e one
Figu e 2. Bes sco ing con o ma ions, de i ed om molecula docking calcula ions, o he: (A) bisphenols, (B) ph hala es, (C) PFAS, and (D)
myco oxin zea alenone, in complex wi h AR (le panel, pink) and ERα( igh panel, whi e). All s uc u es ha e been supe imposed wi h he c ys al
s uc u e o he AR in complex wi h dihyd o es os e one (PDB ID: 1 7 ) and he ERαin complex wi h he inhibi o d ug amoxi en (PDB ID:
3e ). The c ys al subs a es a e depic ed as pu ple s ick models, while all docked compounds a e depic ed as wi e ames.
Jou nal o Chemical In o ma ion and Modeling pubs.acs.o g/jcim A icle
h ps://doi.o g/10.1021/acs.jcim.5c01437
J. Chem. In . Model. 2025, 65, 11950−11964
11954
and 17β-es adiol, espec i ely) ha e simila s uc u al ea u es
and consequen ly p esen simila binding pa e ns. Fu he -
mo e, he s uc u al cha ac e is ics o he wo ecep o s (Figu e
S2B) sugges ha he s udied EDCs will bind wi h a pa e n
simila o hose o bo h ecep o s and po en ially p esen
simila sco es. This obse a ion is alida ed by he simila i y in
he epo ed docking sco es (Table S1 and Figu e S2A). The
la ges di e ence be ween AR and ERαdocking sco es is
obse ed only o diisononyl ph hala e, Hexamoll DINCH, and
zea alenone (ΔΔG−1.1, −1.1, and 1.4 kcal mol−1,
espec i ely). Zea alenone appea s o bind mo e igh ly o
ERα han AR, wi h an ΔΔGo 1.4 kcal mol−1(Figu e S2A,
ed). The end is e e sed o diisononyl ph hala e and
Hexamoll DINCH ha seem o bind igh e o AR han ERα
(Figu e S2, blue) wi h he same ΔΔGo −1.1 kcal mol−1. Ou
o he nine EDCs employed o he docking calcula ions,
die hyl ph hala e appea s o ha e he leas a o able docking
sco e o AR and ERαwi h docking sco es −6.2 and −5.8 kcal
mol−1, espec i ely.
Figu e 2 p esen s he bes docking con o ma ions o he
selec ed EDCs in he binding ca i ies o AR and ERα. As a
e e ence, in bo h cases, we selec ed he c ys al s uc u es o
he ecep o in complex wi h he na u al subs a e
(dihyd o es os e one) o he AR
18
and he known inhibi o
d ug amoxi en o ERα.
17
Bisphenols (Figu e 2A), ph hala es
(Figu e 2B), and PFAS (Figu e 2C) a e docked in a posi ion
like he espec i e c ys al s uc u e ecep o subs a e. Only
zea alenone appea s o bind in di e en posi ions in he AR
(Figu e 2D, le panel) and ERα(Figu e 2D, igh panel). In
ERα, zea alenone binds away om he si e o known inhibi o
amoxi en. This de ia ion may be explained by he di e en
s uc u e o he compound ha in ol es a la ge, uncommon
ing s uc u e (Table 1). Mo eo e , hese di e ences in he
binding loca ions p edic ed by docking may explain he
di e ence obse ed in he espec i e docking sco es (Figu e
S2A) be ween he wo ecep o s. The alidi y o he docking
p o ocol ollowed in his pape is u he suppo ed by he
edocking o he c ys al ligands. The p o ocol, employed,
duplica es he c ys al con o ma ions in bo h ecep o s (Figu e
S2C,D) wi h RMSD alues 1.6 Å o amoxi en in ERαand 0.5
Å o dihyd o es os e one in AR compa ed o hei
expe imen ally de i ed posi ion.
Expe imen al Resul s. In e es ingly, hese binding di e -
ences ansla ed also in o expe imen ally de e mined di e -
ences in po ency (EC50) and e icacy (Emax) among bisphenols
and be ween bisphenols and zea alenone in he ac i a ion o
ERα(Figu e 3). Zea alenone exhibi ed a po ency ha was
120-, 430-, and 1060- old highe han he po encies o BPA,
BPF, and BPS, espec i ely (Table 2). I also demons a ed
highe e icacy han he bisphenols. Among he bisphenols,
BPA showed highe po ency han BPF and BPS, o e all
aligning wi h he ela i e di e ences in docking sco es o ERα.
Wi hin he ele an concen a ion in e als, no appa en
cy o oxici y was obse ed (Figu e S3).
Bisphenols a e also known o bind o AR and ecognized as
AR an agonis s.
74
Zea alenone also was shown o exe
an agonis ac i i y on AR.
75
To analyze whe he molecula
docking sco es p edic he po ency and e icacy o bisphenols
and zea alenone as AR an agonis s, we made use o AR ligand-
dependen dime iza ion, which elies on he in e ac ion
be ween he LBD and he N- e minal egion o AR.
76
A
espec i e mammalian wo-hyb id assay shows ha 11-
ke oDHT induced he in e ac ion o bo h egions o AR
wi h an EC50 o 3.3 nM (95% CI 2.3−4.7 nM) (Figu e 4A).
BPA, BPF, and zea alenone an agonized AR dime iza ion wi h
simila po ency and e icacy, while BPS was less po en and
e icacious (Figu e 4B and Table 3).
Molecula Dynamics Simula ion Analysis. The docked
con o ma ions o he compounds se ed as he ini ial
s uc u es o a se ies o 18 p o ein−ligand MD sys ems (9
pe ecep o ), each un in iplica e (54 ajec o ies in o al).
The RMSD analysis o he backbone a oms o he wo
ecep o s (Figu es S4 and S5) highligh s he de ia ion o he
p o ein s uc u e om i s ini ial con o ma ion. The ERα
beha io in he complexes wi h he a ious EDCs (Figu e S4)
shows he ecep o s uc u e o be mo e mobile compa ed o
AR du ing he simula ions (Figu e S5). ERαp esen s highe
RMSD alues compa ed o i s ini ial con o ma ion, eaching up
o almos 4 Å in he complex wi h Hexamoll DINCH (Figu e
S4, middle panel, igh ). This a ia ion in he RMSD alues
could be a ibu ed o he loop egions o he ERα ha show
inc eased RMSF alues (Figu e 5A and Figu e S6). On he
o he hand, AR appea s less mobile du ing he MD
simula ions, wi h RMSD alues no exceeding 2 Å (Figu e
S5). The highes RMSD alues o AR a e obse ed in he
complex wi h pe luo ohexanesul onic acid (PFHxS, Figu e S5,
bo om panel le ), wi h a mean alue o 1.5 Å (±0.17). The
mos s able con o ma ion o ERαis obse ed in i s complex
wi h pe luo ooc anoic acid (PFOA, Figu e S4, bo om panel,
middle), while in he complex wi h zea alenone (Figu e S4,
bo om panel, igh ), he ecep o appea s o be mo e lexible
h oughou he simula ion. The lexibili y o he ERαis u he
Figu e 3. Concen a ion esponse cu es o ERαac i a ion by
bisphenols and zea alenone. ERαmammalian one-hyb id assays we e
execu ed in ansien ly ans ec ed HEK293 cells, which we e ea ed
a e wa d o 24 h wi h he indica ed concen a ions o chemicals.
Da a a e p esen ed as a mean ±SD old change (n≥3 independen
expe imen s) wi h espec o he no malized epo e gene ac i i y o
cells ea ed wi h 0.1% DMSO only. Cu es we e i ed wi h nonlinea
eg ession using he o mula wi h a a iable slope ( ou pa ame e s) o
G aphPad P ism e sion 10.4.1.
Table 2. Po ency and E icacy o Bisphenols and
Zea alenone in Ac i a ing ERα
EC50 [μM]
EDC mean 95% C.I. Emax [FC]
a
conc. [μM]
BPA 0.60 0.46−0.76 19.6 10
BPF 2.1 1.7−2.9 56.1 10
BPS 5.2 4.2−6.5 48.5 30
ZEA 0.0049 0.0036−0.0065 89.9 0.1
a
FC, old change.
Jou nal o Chemical In o ma ion and Modeling pubs.acs.o g/jcim A icle
h ps://doi.o g/10.1021/acs.jcim.5c01437
J. Chem. In . Model. 2025, 65, 11950−11964
11955
suppo ed by he RMSFs (Figu e 5A and Figu e S6). The
backbone p o ein a oms in ERαappea o be sligh ly mo e
lexible compa ed o he alues obse ed o he AR (Figu e 5B
and Figu e S7). In all AR complexes, he esidues show small
luc ua ions om hei ini ial con o ma ion wi h he highes
de ia ion obse ed in he a ea be ween esidues 55−65 and he
a ea 145−155 ha each alues o almos 2 Å. This di e ence
in lexibili y be ween he wo ecep o s may indica e ha EDCs
a ec hem di e en ly, leading o a ia ions in hei
in e ac ions. The mos impo an obse a ion is ha he
ecep o a eas ha a e epo ed in he li e a u e o in e ac
wi h na i e ligands and d ugs
17,18
p esen low RMSF alues.
This obse a ion poin s o a po en ial common binding
mechanism o he EDCs s udied he e.
The nex s ep in he analysis in ol ed he iden i ica ion o
po en ial hyd ogen bond in e ac ions (HBs) ha a e obse ed
du ing he MD simula ions. As expec ed, he hyd ogen bond
in e ac ions do no a y g ea ly among he di e en EDCs.
Based on he s uc u es o he compounds (Table 1), he e a e
ew g oups ha can c ea e hyd ogen bonds wi h he esidues o
he wo ecep o s. The esul s o he analysis a e p esen ed in
Tables S2 and S3 and Figu e 6. In bo h ecep o s, he EDCs
s udied es ablished ΗΒ in e ac ions wi h app oxima ely he
same esidues. In he case o ERα, esidues Glu48 and His219
ake pa in mos o he ΗΒ in e ac ions (Table S2 and Figu e
6A). Bo h esidues as well as he o he s in ol ed in HBs a e
epo ed o ake pa in ligand binding. Consequen ly, hese
esidues a e loca ed in egions o he p o ein ha ha e low
mobili y (Figu e 5A and Figu e S6). Zea alanone binds o an
al e na i e loca ion, dis inc om he o he EDCs (Figu e 2D)
and hus c ea es hyd ogen bonds wi h Th 42, Me 223, and
Asn227. In he complexes o he a ious EDCs wi h he AR, a
simila pa e n is obse ed. Mos o he HBs o med in ol e
esidues Asn37 and Gln43 (Table S3 and Figu e 6B). Al hough
hese wo esidues a e no epo ed o di ec ly pa icipa e in
he binding p ocess o dihyd o es os e one,
18
hey a e loca ed
on a helix ha in e ac s wi h he na i e ligand (Figu e 6B) and
keeps i in posi ion inside he binding si e. The analysis o he
HBs o med be ween he ecep o and he a ious EDCs
u he unde sco es ha he esidues pa icipa ing in hese
in e ac ions a e loca ed in egions wi h low RMSF alues
(Figu e 5 and Figu es S6 and S7). The only EDC ha does no
o m any HB in e ac ions, in bo h ecep o s, is Hexamoll
DINCH. The lack o hyd ogen bonding can be a ibu ed o i s
h ee-dimensional s uc u e ha may no assis in he co ec
posi ioning o i s oxygen a oms. Ou o all o he EDCs
s udied, die hyl and diisononyl ph hala es c ea e s ong
hyd ogen bond in e ac ions in AR (Table S3). In ERα, die hyl
ph hala e pa icipa es in HB wi h His219 (Table S2) ha is
p esen o only 10% o he simula ion ime, while diisononyl
ph hala e does no c ea e any HB in e ac ions (Table S2).
A compa ison be ween he c ys al s uc u es o ERα
complexes wi h bisphenol A, es adiol, and amoxi en and
he ep esen a i e con o ma ion de i ed om MD simula ions
(Figu e 7) e eals a good ag eemen be ween he simula ed
esul s and hose obse ed in c ys allog aphic s udies.
Speci ically, supe posi ion o he di e en s uc u es shows
ha bisphenol A binds o he same egion as es adiol ( he
ecep o ’s na i e ligand, Figu e 7, g ay) and amoxi en (Figu e
6A, g ay). As p e iously epo ed by Del osse e al.,
31
he
esidues ha in e ac wi h bisphenol A a e mainly His219 and
Glu48 ha a e also implica ed in in e ac ions wi h es adiol.
Mo eo e , he speci ic esidue side chains ha e he same
o ien a ion as hose epo ed in he c ys allog aphic s uc u es
(Figu e 7, magen a). I is impo an o no e ha he s uc u es
sol ed by Del osse e al.
31
con ained he mu a ion Ty 232Se .
The only small disc epancy is obse ed wi h ega d o Me 116,
which is o ien ed di e en ly in he simula ions compa ed o
he c ys al s uc u e (Figu e 7, magen a and lime g een).
We compa ed ou indings wi h hose o o he compu a-
ional s udies o gain u he insigh s in o EDC in e ac ions. As
expec ed, mos s udies ocus on a ange o ep esen a i e
EDCs, o en exhibi ing s uc u al di e si y compa ed wi h he
Figu e 4. Bisphenols and zea alenone an agonize AR ac i a ion. AR N−C- e minal in e ac ion mammalian wo-hyb id assays we e un in
ansien ly ans ec ed HEK293 cells, which we e ea ed a e wa d o 24 h wi h he indica ed concen a ions o chemicals. (A) Concen a ion
esponse cu e o agonis 11-ke oDHT. Da a a e p esen ed as a mean ±SD old change (n= 4 independen expe imen s) wi h espec o he
no malized epo e gene ac i i y o cells ea ed wi h 0.1% DMSO only. (B) Inhibi ion o agonis -induced AR dime iza ion by concu en
ea men wi h he indica ed concen a ions o bisphenols and zea alenone. Da a a e p esen ed as mean ±SD (n= 3 independen expe imen s)
wi h espec o he no malized epo e gene ac i i y o cells ea ed wi h 3 nM 11-ke oDHT only, which was designa ed as 100%. Cu es we e
i ed wi h nonlinea eg ession using he o mula wi h a s anda d slope ( h ee pa ame e s) o G aphPad P ism 10.4.1.
Table 3. Po ency and E icacy o Bisphenols and
Zea alenone in Inhibi ing AR
a
IC50 [μM]
EDC mean 95% C.I. % esidual agonis ac i i y conc. [μM]
BPA 0.98 0.55−1.7 3.1 30
BPF 1.8 1.0−3.3 3.6 30
BPS 23.0 2.9−??? 47.6 30
ZEA 3.7 2.0−6.8 1.5 30
a
???, no compu able by G aphPad P ism.
Jou nal o Chemical In o ma ion and Modeling pubs.acs.o g/jcim A icle
h ps://doi.o g/10.1021/acs.jcim.5c01437
J. Chem. In . Model. 2025, 65, 11950−11964
11956
na i e ligands o d ugs a ge ing he ecep o s. This is e iden
in he PFAS g oup, whe e he compounds ha e linea chains
and no ing sca olds (Table 1). Ano he dis inc i e ea u e o
PFOA and PFHxS is he p esence o he −COO−and −SO3−
Figu e 5. A omic luc ua ions o he ecep o esidues in he di e en complexes o (A) ERα, e ical dashed lines indica e bounda ies o key
seconda y-s uc u e elemen s: Me 38−Glu48 (yellow), T p78−A g89 (o ange), Me 116−Ile119 (g een), Gly216−Leu220 ( ed), and Leu231−Leu239
(blue). (B) AR, e ical dashed lines indica e bounda ies o key seconda y-s uc u e elemen s: Leu36−Gly40 (yellow), Me 74−Me 77 (o ange), A g111
(g een), Leu205−Th 209 ( ed), and Ile230 (blue). Wi h ed le e s, he esidues we e epo ed o in e ac wi h na i e ligands and d ugs in he
li e a u e. The RMSF was calcula ed on Cαa oms ela i e o he ini ial docked con o ma ion s uc u e.
Jou nal o Chemical In o ma ion and Modeling pubs.acs.o g/jcim A icle
h ps://doi.o g/10.1021/acs.jcim.5c01437
J. Chem. In . Model. 2025, 65, 11950−11964
11957
moie ies ha a e pola , in con as o he neu al na u e o
es adiol and es os e one, he endogenous ligands o ERαand
AR. Based on he s udy by Cao e al.,
26
PFOA was epo ed o
c ea e a hyd ogen bond in e ac ion wi h A g89 o ERαas well
as Glu48, while His219 is conside ed impo an in posi ioning
he ligands in he binding ca i y. This in e ac ion wi h A g89 is
also epo ed o be impo an o PFOA ecogni ion.
32,77
Ou
calcula ions highligh his pa e n, since PFOA is ound o
in e ac wi h His219, while he pola in e ac ion be ween ligand
and A g89 is subs i u ed by he ligand−Lys226 in e ac ion.
Simila ly, o ou knowledge, he e a e no in silico s udies ha
in es iga e he in e ac ions be ween he speci ic ph hala es and
he wo ecep o s. None heless, in he s udy by Dahbi e al.,
28
i was epo ed ha ph hala e de i a i es c ea e in e ac ions
wi h esidues A g84 and Gln43 in AR, while in ERα, in e ac ions
a e o med wi h esidues like Glu48. These obse a ions a e in
ag eemen wi h ou indings ha when ph hala es bind o AR,
hey in e ac mainly wi h A g84, while in he es ogen ecep o ,
he ph hala es s udied seem o in e ac mainly wi h His219.
MM-GBSA Analysis. To u he unde s and he binding
p ocess o he a ious EDCs, MM-GBSA calcula ions we e
pe o med on he MD ajec o ies o all o he complexes. The
esul s a e summa ized in Figu e 8 and Tables S4 and S5. In
addi ion o he calcula ions o he en halpic and en opic
con ibu ions, decomposi ion analysis was pe o med o all
complexes (Figu es S10−S19). The analysis o he esul s does
no e eal a common pa e n be ween he binding o EDCs o
he wo ecep o s. Hexamoll DINCH shows he highes
binding ene gy o bo h ERαand AR (−30.95 and −36.88 kcal
mol−1, espec i ely). This igh binding is en halpically d i en
(Figu e 8), especially in he AR. The hyd ophobic na u e o
Hexamoll DINCH may u he enhance he hyd ophobic
in e ac ions inside he binding ca i ies o he ecep o s; his is
mi o ed in he highly a o able an de Waals con ibu ions
(Tables S4 and S5). Thus, his p o ides a po en ial explana ion
o he igh binding despi e he absence o hyd ogen bonding
in e ac ions. A common end o mos o he EDCs is he
igh e binding o AR compa ed o ERα, wi h he excep ion o
diisononyl ph hala e and die hylph hala e (Figu e 8, o ange).
Ano he end de i ed om he MM-GBSA esul s is he
co ela ion be ween he binding a ini y and EDC size.
Compounds like ph hala es (Table 1), which con ain a ing
sca old and lexible ca bon chains, show inc eased binding
ene gy o bo h ecep o s (Figu e 8). On he o he hand,
compounds like bisphenols, PFOA, and PFHxS display less
a o able binding ene gies. This could possibly be a ibu ed o
he ac ha we employed only a single molecula en i y in ou
s udy. Thus, he smalle molecules may show an inc eased
lexibili y inside he binding ca i y o he ecep o . In ac , i
has been ecen ly epo ed ha small molecules, such as
PFOA, could simul aneously bind o he ecep o wi h wo
molecules, and hese molecules could in e ac wi h di e en
si es in he p o ein’s ligand binding pocke (LBP).
79
As
expec ed, en halpy con ibu ions a e d i ing he binding
p ocess in all complexes, highligh ing he impo ance o he
an de Waals o all he EDCs. Fo PFOA and PFHxS (Tables
S4 and S5), his obse a ion is in ag eemen wi h Cao e al.
26
ha ha e pe o med simila calcula ions in he es ogen
ecep o and PFOA complexes and ha e epo ed ha he
binding ene gy is lowe han he one epo ed o na i e
ligands.
Alchemical Rela i e Binding F ee Ene gy Calcula ion
Analysis. To compu e he ela i e binding ee ene gies o
bisphenol F o ERαand AR�compa ed o bisphenols S and
A�MD simula ions we e combined wi h ela i e ee ene gy
calcula ions using a nonequilib ium app oach o accele a e
con e gence. Ini ially, a 10 ns p oduc ion un in he NPT
ensemble was pe o med o each ligand pai . To assess he
con e gence o he calcula ions, he h ee ligands we e
connec ed h ough a closed he modynamic cycle, allowing
he e alua ion o hys e esis. In a well-con e ged alchemical ee
ene gy calcula ion, he sum o he ee ene gy changes a ound
he closed cycle should be ze o o nea ze o. Fo he AR−
ligand complexes, he cycle closu e e o was 0.68 ±2.25 kcal
mol−1, while o he ERα−ligand sys ems, i was −0.37 ±3.20
kcal mol−1. These alues indica e ha he o al de ia ions we e
no minimal, as bo h cycles showed high s anda d de ia ions.
To add ess his, we ex ended he equilib ium NPT calcula ions
o he bisphenol S →bisphenol F pe u ba ion, which had
exhibi ed he highes s anda d de ia ion among eplicas (2.74
kcal mol−1 o ERαand 2.01 kcal mol−1 o AR) om 10 o 20
ns. The addi ional sampling esul ed in a subs an ial educ ion
in he s anda d de ia ion o his pe u ba ion (0.62 kcal mol−1
o ERαand 0.60 kcal mol−1 o AR). Consequen ly, he
s anda d de ia ion o he cycle closu e e o s also dec eased
Figu e 6. Recep o esidues in ol ed in hyd ogen bond in e ac ions
wi h he a ious EDCs o e he du a ion o he MD simula ions in
ERα(A) and AR (B). In da k g ay a e he c ys al s uc u es o
amoxi en (PDB ID: 3e ) and dihyd o es os e one (PDB ID: 1 7 ).
Residues highligh ed in ed a e epo ed o in e ac wi h he ligands in
he ecep o s, while in blue is helix 12 o ERα ha plays a ole in d ug
binding.
17,18
Jou nal o Chemical In o ma ion and Modeling pubs.acs.o g/jcim A icle
h ps://doi.o g/10.1021/acs.jcim.5c01437
J. Chem. In . Model. 2025, 65, 11950−11964
11958
