Structure‑Activity Relationships for Receptor Binding of Environmental Estrogens : Analogies with β‑Cyclodextrin Inclusion Complexes
著者名(英) Takahiro SUZUKI, Stuart SHAPIRO journal or
publication title
Journal of Toyo University. Natural science
number 53
page range 9‑32
year 2009‑03
URL http://doi.org/10.34428/00002543
Creative Commons : 表示 ‑ 非営利 ‑ 改変禁止
http://creativecommons.org/licenses/by‑nc‑nd/3.0/deed.ja
Journal of Toyo University, Natural Science, No.53.9-32(2009) 9
Structure-Activity Relationships for Receptor Binding of Environmental Estrogens:
Analogies withβ一Cyclodextrin Inclusion Complexes
Takahiro SUZUKI*and Stuart SHAPIRO**
Abstract
Agroup contribution model fOr calculating relative binding af五nities(RBAs)of struc-
turally diverse ligands to rat uterine estrogen receptor一α(ER一α)was developed. Least squares regression yielded an equation correlating RBA with the且rst-order molecular connectivity index iZ(a measure of molecular bulk)and weighted structural fragments
(measures of molecular fbrces between the receptor and ligand). This model showed a very good index of determination(r2=0.848)between experimental and calculated loglo RBAs fOr a training set comprised of 128 compounds. Silnilarities between ER一α:(xeno)
estrogen binding and fOrmation ofβ一cyclodextrin:(xeno)estrogen inclusion complexes were explored by comparing experimental free energies for(xeno)estrogens bound to ER一α with those for the same ligands complexed with natural and 2-hydroxpropyl-6-
cyclodextrin. Indices of determination in the range O.6-0.7 were obtained fbr the ER一α andβ一cyclodextrin systems. Development of a group contribution method for in silico screening of potential xenoestrogens represents a practical alternative to tedious, costly receptor-ligand binding assays and animal experimentation.
Keywords:environmenta正estrogen, group-contribution method, molecular connectivity,
QSAR, xenoextrogen
1. Introduction
Cyclodextrins(CDs)are cyclic polysaccharides usually comprised of six(α一CD),
seven(β一CD), or eight(γ一CD)α一d-glucosyl units connected by 1,4’-0-glycosidic bonds. CDs assume an open conical shape, with their hydroxyl moieties directed towards the outside, leaving a relatively non-polar interior which can accommodate a variety of organic and inorganic“ №浮?塔[molecules(Connors.1997). The natural CDs
* N atural Science Laboratory. Toyo University.5-28-20 Hakusan, Bunkyo-ku, Tokyo I I 2-8606.
Japan
**aasilea Pharmaceutica International AG. Grenzacherstrasse 487, Postfach, CH-4005 BaseL Switzerland
10
Takahiro SuzuKI and Stuart SHAPIROhave been modified by adding alkyl or hydroxyalkyl groups to the glucosyl hydroxyls.
The binding properties of natural and modified CDs have Ied to their being used in a wide range of practical applications, including solubilization/stabilization of pharmaceu-
ticals and stationary phases for chromatographic separation(Szejtli,1988:Szejtli,
1998;Hedges,1998). CDs also represent interesting models fOr interactions of ligands with enzyme and receptor binding sites(Szejtlit 1988).
Most steroids and many other chemicals that interact with endocrine systems
(Brunstrom et al.2003;Colborn et aL,1993:Hutchinson and PickfOrd,2002;Singleton and Khan,2003)fall within the size range of guest molecules forβ一CD cavities. Con-
siderable effort has been invested to understand the toxic activities of xenoestrogens C‘endocrine disruptors”), defined by the United States Environmental Protection Agency as substances which interfere with the synthesis, secretion, transport, binding,
action, or elimination of hormones responsible for the maintenance of homeostasis,
reproduction, development, and/or behavior(Kavlock,1996). To date the European Union has classified 553 synthetic substances as potential xenoestrogens(European Commission,2001). Jn vitro bioassays for endocrine disruptive activity using MCF-7 human breast adenocarcinoma cells and other estrogen-sensitive cel川ines are very costly, time-consuming, and labor-intensive procedures;therefbre, quantitative struc-
ture-activity relationship(QSAR)models, using traditional Hansch and molecular con-
nectivity methods as well as sophisticated 3D techniques, have been applied to analyz-
ing ligand binding to estrogen receptors(ERs)(Blair et al.,2000;Bradbury et aL,
1996;Galltchev et aL,1994;Gao et aL,1999a,b;Gao,2001;Hutchinson and Pickford,
2002:Kramer and Giesy,1999;Roncaglioni et aL.2004:Sadler et aL,1998;Shi et aL,
2001;Suzuki et al.,2001;Tong et aL,1997a,b;Tong et al.,1998;Waller et aL,1995,
1996;Waller, 2004:Zheng and Tropsha,2000).
Steroids(Wallimann et al.,1997)and other compounds(Brunstrom et aL,2003:Col-
born et al.,1993;Hutchinson and Pickford,2002;Singleton and Khan,2003)that inter-
act with endocrine systems also form inclusion complexes with natural or modified β一CDs in aqueous milieux. The moiecular shape and size of the guest molecule, desol-
vation of the guest molecule(release of“high energジwater from the“host”CD cavi-
ty),relief of conformational strain in the uncomplexed CD, van der Waals interactions,
and hydrogen bonds between moieties of the guest and host are believed to contribute in varying degrees to the formation and stability of CD inclusion complexes(Connors.
1997).Schematic representations of the inclusion complex formed betweenβ一CD and 17fi-estradiol, and of the binding of 17β一estradiol to estrogen receptor一α(ER一α)
(Brzozowski et al.1997)are shown in Figure 1. Though there are differences in cavi-
ty size and shape between estrogen receptors andβ一cyclodextrins, similar steric and energetic interactions likely drive both complexation of 17fi-estradiol byβ一CD and binding of 17β一estradiol by ER一α. Therefore, CD binding of xenoestrogens may func一
QSARs of environmental estrogens
11
炉CD
A HO
OH
Analogy
ER一α
OH
Fig.1.β一CD:17β一estradiol complex(A)ER-a:17β一estradiol complex(B),
tion as surrogates fbr ER一αbmdj五g of xenoestrogells.
Diverse computational tools, including molecular modeling, neural networks, and 2D-and 3D-QSAR/QSPR, have been used to elucidate the factors governing host:
guest hlterac60ns in CD inclusion complexes and to predict their themodynamic sta-
bhities(Katritzky et al.,2004;Klein et aL,2000;Lipkowitz,1998;Liu and Guo,1999;
Matsui et al.,1985;Park孤d Na瓦1994;Suzuki et al.,2000). Nonl血ear group contri-
bution models(GCMs)for calculating binding constants or free energies of complex・
ation of guest molecules with naturalα一alldβ一CDs have been reported(Suzuki,
2001).正[ere we present a GCM, derived from expe血1ental binding data fbr a structur・
a皿ydiverse set of molecules, for calculating relative binding affinhies(RBAs)of ligands to ER一α, as wel as a correlation between free energies of complexation of molecules with native and modi丘edβ一CDs and RBAs fbr xenoestrogens with ER一α.
2.Methods
Relative bindin a伍nities
Binding a伍ni6es fOr rat uterine cytosoUc estrogen receptor ER一αwere determined by competitive binding assay with[3H]17β一estradiol(E2)(Blair et a1.,2000:Shi et a1,2001;Waner et aL,1996). The relative binding af丘nity(RBA)fbr a compound“X”
is a dimensionless number defilled as 100 times the molar ratio E2/X required to decrease E2 binding to ER一αby 50%. On the RBA scale,17fi-estradiol is assiglled a value of 100, with lower a面ty analogues having lower values and higher af丘11ity ana-
logues having higher values.
SAR delrivation
Atraining set of 128 molecules, ranging from 4-methylpheno1(molecular weight
108.14)to ICI-164384(molecular weight 525.82), is presented in Table 1, along with12
Takahiro SuzuKI and Stuart SHAp1Roexperimental loglo RBA values. The structures f()r 128 molecules are shown in Figure 2.
Derivation of a predictive model for(xeno)estrogen RBAs follows from the assump-
tion that loglo RBA is at least partially describable in terms of ligand fragment contri-
butions, dimensionless values associated with single atoms or multiatom assemblies.
Moreover, in an earlier study(Suzuki et aL,2000)naturalβ一CD:guest binding con-
stants were f6und to have a nonlinear dependence on molecular size. Therefore, as a first approximation loglo RBA was assumed to be a quadratic equation of general form:
loglo RBA=cl・D+c2・D2+Σ(ni・Gi)+co (1)
where
D=descriptor representing molecular bulk or size:
c],c2=regression coefficients;
G=value assigned to ligand structural fragment i;
ni=number of times molecular fragment i occurs in the ligand;and Co=regreSSIOn COnStant.
As possible descriptors of molecular bulk or size, zeroth-order(oZ)and丘rst-order
(IZ)molecular connectivity indices(Kier and Hall,1986),as well as molecular weights
(Mw), were calculated fOr each molecule in the data set(Table 1). Regression coe伍一 cients were determined by the method of least squares, using Excel Multivariate Anal-
ysis v3.0(Esumi Co., Ltd., Tokyo. Japan)on a microcomputer running the Windows XP operating system.
Derivation of fra ent constants
The quality of a GCM(i.e., the precision of its predictions and their reliability)
depends upon the definition of molecular fragments and their assigned weightings.
After considering the available number of data points, facile identification of groups contributing molecular structure, and implicit inclusion of the constitutive factors, a set of fraglnents similar to that used earlier by Lydersen(1955)for estimating critical properties was employed. Fragment constants Gi for 26 chemical groups are Iisted in Table 2. IoH-oH is an indicator variable. equεd to either O or l depending upon the pre-
sumptive absence or presence, respectively, of cooperative hydrogen bonding involving oxygen atoms between the ligand and ER一α. Three examples of the decomposition of molecules into defined fragments are shown in Figure 3.
Determination of bi皿din constants for -CD: est inclusion com lexes
The thermodynamic stability of a 1:1CD:guest inclusion complex can be expressed as the dissociation constant of that complex, Kdis㍉、11,. Values of Kdissq、いfor inclusion complexes between naturalβ一CD or 2-hydroxypropyl⊃β一CD(2-HP-⊃β一CD)(Yoshida et aL,1988)and l l estrogens or xenoestrogens(kepone,4-nonylphenoL bisphenol A,
methoxychlor, estrone,1 7fi-estradiol,1 7)6 -estriol, ethynylestradiol, di-n-butyl phtha-
late、 butyl benzyl phthalate, endosu1fan)were determined by the solubility rnethod
QSARs of environmentaI estrogens
13
Table l. Experimental and calcしrlated loglo RBA values for the rat uterine estrogen receptor(ER一 α)for 128 molecules、
no,
compound
formula logie RBAobs’d calc’d
1δH-OH ref,**
123456789012345678901234567890123456789012345 111111111122222222223333333333444444
4-methylphenol 4-chloro-3-methylphenol 2てhloro-4-methylpheno1 4-chloro-2-rnethylphenol 3-ethylphenol4-ethylphenol
methyl-4-hydroxybenzoate ethy十4-hydroxybenzoate 4-sec-butylphenol 2-sec-butylphenol 4-tert-butylphenol kepone
propyl paraben 4-tert-amylphenol butyl-4-hydroxybenzoate 2.4Ldichlorobiphenyl 4-phenylphenol 3-phenylphenol
2,3.4.5-tetrachloro-4’-biphenylo1 2’.5’-dichloro-4-biphenylol 4-chloro-4’-biphenylol 2-chloro-4-biphenylol 4.4Lsulfbnyldiphenol 4-heptyloxyphenol 4-benzyloxyphenol
2.2’-methylenebis(4-chlorophenol)
bis(4-hydroxyphenyl)methane 4,4’-dihydroxybenzophenone 2,4-dihydroxybenzophenone 4-phenethyLphenol 4-tert-octylphenol 4-octylphenol 4、4’-dihydroxystilbene 4.4’-ethylene diphenol 2.2’.4.4Ltetrahydroxybenzil o,グーDDT
4.4’一(dichlorovinylidene)diphenol bisdemethylmethoxychlor benzy14-hydroxybenzoate heptyl 4-hydroxybenzoate coumestrol
chalcone 4-nonylphenol 4’-hydroxychalcone 4-hydroxychalcone
ワピ ぐつ ぐコ 。。・。・。・;急㍑品骨㍑㍑榴疏繰霊ぽ芯㍑霊慧翫ぽ㍊四曇輌 凪品且7氏日且-恥品旧旧沮に沮沮沮沮沮沮沮坦沮沮沮沮沮3R沮沮沮沮4且沮沮沮旧沮温4且沮沮沮沮5R沮沮
ドじ ヅ みご リイ お お ぽ リ ト エ エ エ ペ ハ ユ エ ヒ ヒ エ ミ し エ エ ペ エ エ エ エ ペ じ エ エ ユ エ エ エ
CCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCC 086777427419267144448778452619245485204952335536681423568220604641708440466835468465008545433334333331333333012232323222120102002202122 一 一一一一﹇ 一一一 一一一一一一一一 一一一﹁ 一一﹁ 一一一一 一一一一一一一﹇ 644477729910384400799497668113949953089852615733366439982159622116661444888556625146324398333333332222322233122233322222120001012203222 一一一 一一一一一一一一﹁ 一一一 一一一 一一一一﹇一一一一一一一一一一一 一一一一一 000000000000000000000000000000001110000000000 333333223333333333333333333333323333333333333
14
no.
67’44▲
89012345678901234567890123456789012345678904455555555556666666666777777777788888888889
Takahiro SuzuKI and Stuart SHApIRo
Table 1(continued)
compound
formula logi{]RBAobs’d calc’d
bisphenol A
4-[2,2-dichloro-1-(4-methoxyっhenyl)
ethenyl]phenol
monodemethylmethoxychlor
6-hydroxyf]avone4’-hydroxyfiavanone 3’-hydroxy且avanone 6-hydroxy且avanone 7-hydroxy且avanone equol
2-ethylhexyl 4-hydroxybenzoate 6.4’-dihydroxyflavone
4,2:4Ltrihydroxychalcone daidzein
73:4’-trihydroxyisoflavone 3.6.4’-trihydroxyf]avone genlsteln
aPlgenln baicalein narlngenln phloretin kaernpferol fisetin
morin myrlcetln formononetln
1.3-diphenyltetramethyldisiloxane dimethylstilbestrol
bisphenol B methoxychlor biochanin A
prumetm
dimethyl allenolic acid diphenolic acid 3-deoxyestrone
estra-1.3,5(10)-trien-3-oI 3-deoxyestradiol
17-deoxvestradiol 4-dodecylphenol dienestrol diethyLstiLbestrol estron
3-hydroxy-estra-1,3.5 q O)-trien-16-one meso-hexestrot
dl-hexestrol l7α一estradiol
Cl5Hl602 Cl5H]2Cl202
リゴ ワニ
0 °12 0 1333333334445声:。555566784S2213553」 2222322 COOOOOOOOOOOOOOOOOOOOOOOOOCOOOOOOOOOOOOOOOO
3 0 2 2 2 2 4 2 0 2 0 0 0 0 0 0 2 4 D .U O O 2 2 6 8 5 2 2 0 8 2 】 4 」 0 8 “ 2 2 2 2 」1 1 1 1 1 1 1 2 1 1 1 ﹈ 1 1 ﹇ l 1 1 1 1 1 ︸ 1 2 1 1 1 1 1 2 1 2 2 2 2 3 1 2 2 ヨ 2 2 a HHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHH 5555555555555555555555.666666、b778S88S8札88S.S8
ユ し ぺ ユ ヒ エ エ エ ト ヒ エ ヒ ユ ユ ぺ エ エ ヨ ト ユ ヒ ユ エ ユ ヒCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCC
一2ユ1 -1.59
-0.64 -0.37
915853242655565536155586670742306043706986984670787826333531163379110237012231756824540322330101100013211232231132203210111200200
一一一一一一一 一一一一一一一一一一一一一}一一 一一一一一 一 一 一 一 790323901176652183653546037208220400377811574222129281666221388380184056037707821116601333330111100022212221331122213001012200001 一一一 一一一一 一一一一一一「一一 一一一一一 一一 一一 一 一 一 一一
1&1.。}{
∩UO
0000001010111100000000101000000000001100111
ref.**
つ」う0
3333333333333333333333333313333003023322322
QSARs of environmental estrogens
Table l(continued)
no. cornpound formula logiO RBA
obs’d calc’d
1234567899999999
99 100 101 102 103 104 105 106
789012345670001111111111111111111
118 119 120 121 122 123 124 125 126 127 128
17!7-estradiol 17fi-estriol
6α一hydroxy-estradiol 4-hydroxy-estradiol 2-hydroxy-estradiol 3.3二dihydroxyl hexestrol nordihydroguariaretic acid
4-ethyl-7-OH-3-(p-methoxyphenyl)-
ditiy’dro-1-benzopyran-2-one α一zearalenol
β一zearalenol zearalanone α一zearalano1 β一zearalanol
hexestrol monomethyi ether 3-methylestriol
p-(α.β一diethyl-♪-methylphenethyl)-meso-
phenol mestilbo1
5α一androstane-3α.17β一diol doisyrloestrol
aurin phenol red
2.6-dimethyl hexestrol triphenylethylene ethynylestradiol
diethylstilbestrol dimethyl ether norethynodrel
16fi-hydroxy-16-methyl-17β一estradio1 3-methyl ether
phenolphthalein phenolphthalin mestranol moxestrol tamoxifen toremifene ctomiphene 4-hydroxytamoxifen drolox迂ene nafoxidine ICI 164384
C18H2402 ClgH2403 ClsH240:3 Cl8H2.103 CtsH2,403 C18H2204 ClsH2204
CisE[lsO.1
ClsH2405 C18H2405 ClsHz、10s Cl8H2605 C]6H2605 ClgH2402 ClgH2603 ClgH240
C]gHL?L・02
ClgH3002
ClgH2L」O:l ClgHl.103
Clf)Hl,IOt」S
C20H2602 C20Hi6
CL,oHL?40L・
C20H2402 C20H2602 C20H2803
C20H1.;04 C2し)HtfiO.9
C21H2502 C2[H2603
C26H2gNO
C26H28ClNO C26H28CINO C26H2gNO2 C211H2gNO2 C2gH3]NO2 C34H55NO32.00 1.05 0.99 -0.15
-0ユ5 -0.15 1.82 1.59 1.47 1.64 1.19 1.72
-1.51 0.14
-0.05 -0.28
1.63 0.32
-0.69 0.25 0.32 -0.33 1.48 090
-0,19 0.84 0.97 0.63
-1.65 -1.53 0.60 -091
1.31 0.34
-0.92 0,01
-2.74 -2.42
-1.50 -1.15
-3.25 -2.63 1.11 0.43
-2.78 -1.43 2.28 2.16
-L25 0.04
-0.65 0.35
-1.48 -1.82
一1.87
-3.67 0.35 1.14 0.21 0.14
-0.14 2.24 1.18
-0.14 1.16
一〇.33
-2.73 0.32 0.29 0.15 0.16 0.16 0.69 0.69
-O.14 2.96
*See Tab▲e 2.
** P)Waller et al.,1996:2)Blair et al.、2000:3)Shi et al.,2001.
Is,1.OH
11111100 11011100 01000001000 00000000001
15
ref.**
22233333 33333333 33333333333 33333333333
16
Takahiro SuzuKI and Stuart SHAplRoOH
1
m
HO
5
O9 H
O
’,
OH 2
O6 H
H・
0仏
10
Cl
H°
ト己
l
O O
\ ノ
OH
7
HO
ll 3
HO
Cl Cl
H°
Cl
/ 叉}
OH
8
CI O
\
4
Cl
0
Cl
12 Cl C1[
CICI Cl
HO
Cl Cl
16 13
Cl
OH
1グ
\ \ ノ C1
20
\ノ
/ー\ \ノ
17
OH
14
18
OH
15
OH
ヘ グ 1
「 ノ
\Cl グ グ
1 \
\Cl OH 21 22
Fig.2. Chemical structures of the compounds in Table l.
:〕〔).Cl
,,.
Cl
19
㎝め山◇曇
QSARs of environmental estrogens
HO
0
へ ∠
OH
28
/\
へ〃
24
OH
HO
。9
25
HO 2 O9 H
HO HO
グ \、∠
26 C1
HO
31
㎝人」
H H
dグ
\
OH
34
CI 1
、で㎝
37
\∠
OH
HO
Cl Cl
0
32
㎝⊥」 ベア皿額
CI
グ
\
OH
38
0
グ\
OH
OH
Fig.2 (continued)
Cl
30 、1 ∠
OH
27 へ「h
、r㎝ へ∠ ㎝人ー丁
|グ \ OH 33
36
0
グ 1
\
OH
39
へ∠
O
Cl
17
OH
’i。 ∵s”’XPIR°
℃し・-H°
O
OH 40 41
HO HO
O 43 44
0H
Cl
HO OH
/0 46 47
0H グ プ o \1 \l
HO O O
49 50
グ HO O HO
グ \
HO
O O 52 53
0H
HO ℃Y・一し一 H°
O HO
O 55 56 Fig.2 (continued)
HO
Cl Cl
O
42
45
Cl グ
\
「グ
\
/0 48
0
OH
51
4
A…
557
OH
H O
H
O
HO
HO
HO
HO
HO
0
61
80
﹇OOH
OH O
64
07 6
O
70
HO
73
OH
H O
O H
HH OO
0/
QSARs of environmental estrogens
HO O
59
HO
OH O 62
HO OH
グ \ OH O65
OH
HO
OH
OH
HO
HO
HO
Cl
OH O
68
。ミ℃}
r
71 ー0
/0
74
Fig.2 (continued)
OH OH
00 6
HO
HO
HO
OH O
6319
0H
OH O
66
OH
OH O
69O
㎝人」
1グ
\
OH
7275
OH
HH OO
O/
20
/O
0
OH O
76
O
OH
Takahiro SuzuKI and Stuart SHApIRo
HO
グ\
OH
77
H
HO
0
H H
H
78
HO OH
\/ グ\
﹇H 一H79 80 81
HO HO
82
84
OH
HOHO
85
OH HO
83
0
86
QH
HO
87
O HO
88,89
Fig.2 (continued〕
OH HO
90
H
QSARs of environmental estrogens
21
OH
HO
HO
00HH
HO
OH
91
OH
OH
94
97
0 0H
100
HHOO
OH HO
O
\O グ\
92
H H H
O,
H
O H
>
OH
95
0
80
9
OH OH
H O
01 1
HO
HO HO
H
\ i 三
H H
/ OH
93
0H
96
H -0
O OH
HO
99
O OH OH
H O
2 0 1
OH
HO
O OH
103
OH HO
0/
104
Fig.2 (continued)
105
O\
OH
22
\
106
HO
109
OH
、,.../L,, COOH
”・・、t。/
112
〔)/
O H
e
Takahiro SuzじKI and Stuart SHAplRo
0
㎝⊥〕 CT、・㎝ 1
OH
110
グS
\ グ\ ク\
3 1 1
OH
...、.Ll“多
OH
HO
HO
OH
H\
8 0 1
O H
H H
/O
l11
OH
114
HO
OH
OH
OH
、1、多
OH
O
HO
HO
ll5
119
OH
116 ll7
OH
1、、、多
120
Fig.2 (continued)
HO
118
121
OH
QSARs of environmental estrogens
23
\N~O
l122
\N~0
1123
Cl
、
XvN- x/xo124
\N~O
]OH HO
\N〜o
一〇
O
125 126 127
OH
HO
O
N-(CHz)3
/
128
Fig.2 (continued)
(Higuchi et aL,1965). An excess of guest molecule was added to solutions containing increasing concentrations ofβ一CD or 2-HP二β一CD;after equilibration by shaking at 25
℃for 7-10 days. an aliquot was centrifuged the supernate filtered. and the filtrate analyzed spectrophotometrically at 220nm(25℃)after dilution with water. Kdiss,1、1‘
was calculated from the initial slope of phase solubility diagrams and aqueous solubility according to eq(2)(Uekama,1983).
Kdis㍉】;lI=slope/[intercept・(1-slope)] (2)
24
Takahiro SuzUKI and Stuart SHAPIROTable 2. Structural fragrnents and their contribution to log RBA for ER一α.
fragment or parameter no. of compds freq of occurrence contribution constant
lz
lZ2
八Jonring In(rre〃2ents
-CH3
-CH2-
>CH-
>C〈
=CH-
=C<
-C≡CH
ノ~ing∫ncre〃2ents
-CH2-
>CH-
>C<
=CH-
=C〈
Oxygen Incre〃le〃ts
-OH(alcoho1)
-OH(phenol)
-0-(nonring)
-O-(ring)
>C=O(nonring)
>C=O(ring)
-COOH(acid)
-COO- (ester)
OH-一一〇H*
ハJitrogen力tcアθ〃lents
>N-(nonring)
>N-(ring)
Other Incre〃lents
-Cl
-SO2-
>Si<
882211 7352n◎447411
1
り09α4ρ07 33ワρ22 11 383397456 ウ臼022 2 ーワ・ 1 ⊂」- 89匂1
1800
9白21 1
3213564 ∩δ22112 11 238Q42 7Q∨312 1 8⊂O 815307456 282212 12 1 ⊂」- 5ワ臼2
4一4.89 0.0415
-6.1×10-5
-0.09 0.17 0.73
L18
0,19 1.51 0.87
0.23 0,42 0.63
-0.05 0.71
-1.41
-O.23
-0.46
-0,94
-0.18
-0.66
-2.21 0.29 1.54
-O,42
-1.69
-0,27
-LO7
0,59
*The incidence of this parameter is shown in Table 1,
3.Results and Discussion
Many experimental and theoretical descriptors are related to molecular size or bulk,
but unlike parameters such as molecular volume, rnolecular surface area, or molar refractivity, molecular weight and topological indices can be calculated easily with high precision without any conditionaユassumptions. Therefore, regression analyses were per-
formed using the experimental RBAs Iisted in Table 1, assuming∬near and quadratic
QSARs of environmental estrogens
25
OH
HO
1
、ノ グ㍉016 lH91
、ノ
H
Cl
H
[-CH3コ+4[=CH-(Ring)]+2[=CH-(Ring)]+[OH(phenol)]
H O
、ミξ
8[=CH-(Ring)]+4[=CH-(Ring)]+2[-C1]
[-CH3]+6[-CH2-(Ring)]+3[=CH-(Ring)]+4[>CH-(Ring)]+
3[=C〈〔Ring)]+[>C<〔Ring)]+[-OH(alcohol)]+[-OH(phenol)コ
Fig.3. ExampIes of decomposition of molecules to molecular fragments.
terms for molecular weight, oZ, or iZ and summation of fragrnent constants 2(nドGi).
For this data set the best GCM was obtained with the丘rst-order Kier-Hall molecular connectivity index 1ノど:
loglo RBA=O.0415 iZ-6.1×10“s(IZ)2+Σ(ni ’ Gi)-4.89
(n=128,r2=O.848, q2=O.654, SD=0.345, absolute ME=0.538, F=19.8)
(3)
where
n=sample size,
r2ニindex of determination,
q2=cross-validated leave-one-out index of determination,
SD=standard deviation,
absolute ME=absolute mean error, and F=Fisher distribution statistic.
The statistical quality of eq(3)was slightly reduced if the]Z and(IZ)2terms were omitted. Good models also were obtained using〔’Z or Mw instead of LZ.
Aplot of experimental loglo RBA values vs. calculated logio RBA values obtained from eq(3)for the complete data set(nニ128)is shown in Figure 4. A reasonable且t was found for molecules of diverse chemical type comprising the data set. Except one parameter, viz. two oxygen atoms having cooperative hydrogen bonding with ER一α,
such as occurs in molecules like 1 7fi-estradiol and diethylstilbestrol(loH_oH), all struc一
26
Takahiro SuzuKI and Stuart SHAPIRO<口出c成。三£旦己罵O
4 3 2 1 0 一1 一2 一3 一4 一5
◆◆・
● .・’
.●.・
・...・δど 一ρ”
◆・■■・◆
・.
⑨・
◆●?◆
.◆
一5 一4 一3 一2 -1 0 1
0bserved logio RBA
2 3 4
Fig.4. Correlation between caiculated(eq(3))and experimental logit]RBAs for ER一αwith l28(xeno)estrogens.
tural fragments correspond to those described in a previous study on free energies of inclusion complexation of organic molecules(Suzuki.2001). .
The contributions of lZ,(IZ)2, and ligand fragments to the RBA fbr ER一αis reflect-
ed in the vaIues of the regression coef丘cients for the topological indices and weightings
of individual fragments(Table 2). Fragment constants for most carbon moieties
(except-CH3 and=CH-(ring), whose contributions are very close to O) were>0, i.e.
they enhanced ligand binding to ER一α, indicating a positive impact of van der Waals forces on the stability of receptor-ligand complexes. In contrast, with two exceptions
(-COO-(ester),>Si〈)fragments containing heteroatoms impacted negatively on the stability of receptor-1igand complexes。 The high fragment constant for OH-一一〇H,+
1.54is attributable to the importance of hydrogen bonding for ligand binding to ER一α.
However, the fragment constant for phenolic hydroxyl is only-0.23, implying that phenols per se modestly destabilize the receptor-ligand complex;whereas the frag-
ment constant of-L41 for non-aromatic hydroxyl groups indicates that this moiety has a pronouonced destabilizing effect on receptor-ligand complexes. A negative effect of aliphatic-OH groups on the stability ofβ一CD:guest inclusion complexes has been described previously(Suzuki,2001).
Correlation of bindin stabilities of rece tor-li and com lexes and -CDl est
Gibbs’free energies of complexation forノβ一CD:guest complexes with l l structurally diverse(xeno)estrogens(Figure 5), and Gibbs’free energies for these same com-
pounds bound to ER一αwere calculated from
QSARs of environmental estrogens
27
llCC 1、
C
CI Cl
Cl CI Cl 12.kepone
Cl Cl
\O 0
Cl
HO
o〆\
74,methoxychlor
0
\1/ ノ\
0
92.17β一estriol
O/
42.4-nonylphenol
HO
OH
\1/ /\
…川llOH
86,estrone
HO
㎝メ〉
0 HO
HO
45.bisphenol A
\1/ ノ\
91.17β一estradiol
OH
り’lltt≡
OH
OH
114.ethynyl estradiol
o-Xv,_
O
Oxx/ O!S _
0
C1
Cl
129.dト〃-butyl phthalate l30, butyl benzyl phthalate (BBP) 13Lα一benzoepin Fig.5. Xenoestrogens used to construct a correlation between their∠Gof binding to ER一αand A G of complexation to !7-CDs.
28
Takahiro SuzuKI and Stuart SHApIRo0
一5
ロー10
o E≧巴一15
∩
0 6-20 巨
o
弍 一25一30
β一CD
一35
- 30
4● H
一20
6 ●8
2●945 2竃▼
1一10 0 AG for ER一α[kJ/mo1]
12gd
10
● 131
20
0
一5
10@ 15 20 25
一 一
﹇る已\重﹈口O﹄﹂口O弍
一30
一35 -30 一20 一 10 0
AG for ER一α[kJ/mol]
10 20
Fig.6. Plots of experimental AG of binding of(xeno)extrogens to ER一αvs. experimental AG of complexation of(xeno)extrogens to !7-CDs.
∠G=-2.303RT loglo K, (4)
where K=Kdis㍉1、加for theβ一CD:guest complex and K=equilibriurn dissociation con-
stant(Uekama et aL,1983)for the receptor-ligand complex, Figure 6 compares experi-
mental free energies of complexation for(xeno)estrogens with /7-CDs vs. experimental free energies of binding of these(xeno)estrogens with ER一α. A slightly better correla-
tion was obtained for the free energies of 2-HP⊃∠ヲーCD:(xeno)estrogen vs. ER一α:
QSARs of environmental estrogens
29
(xeno)estrogen(r2=α661)than for the free energies of naturalβ一CD:
乙1∫.ER一α:(xeno)estrogen(r2=0.581).
(xeno)eStrogen
∠G(naturalβ一CD)=O.265 A G[ER一α:(xeno)estrogen]-19.46
(n=11,r2=0.581, SD =3.03, F=12.6)
(5)
∠G(2-HPうβ一CD)=0.315∠I G[ER一α:(xeno)estrogen]-20.31
(n=ILr2 = O、661,SD=3.04,F=17.5)
(6)
The slight discrepancy in statistical quality between eqs(5)and(6)may, to some extent, re且ect more precise Kdiss〔1、pvalues fbr the 2-HPっβ一CD complexes, since 2-HP一 β一CD is much more soluble in aqueous milieux than nativeβ℃D(Loftsson and Duch-
6ne,2007). Nonetheless, reasonably linear relationships exist for(xeno)estrogen bind-
ing toノβ一CDs vs.(xerlo)estrogen binding to ER一α, suggesting thatβ一CD:(xeno)
estrogen systems. and particularly the 2-HPっθ一CD:(xeno)estrogen system, can func-
tion as alternatives to receptor binding studies and animal experimentation Ibr project-
ing Potential xenoestrogenic activity expressed as relative binding affinity to rat uter-
1ne estrogen receptor一α.
4.Conclusions
AGMC for predicting the estrogenic activity of chemicals was developed Results obtained using a training set comprised of 128 compounds demonstrated that predic-
tion of RBAs of molecules to rat uterine ER一αcan be achieved with reasonable accu・
racy using the first-order molecular connectivity index IZ in combination with molecu-
lar fragments. To the best knowledge of the authors, this work represents the first application of GCMs or additive schemes to the quanti丘cation of ligand binding to ERs.
This method should prove especially useful for the prioritization or preliminary screen-
ing of chemicals for their potential to cause endocrine disruption.
Acknowledgement
The author acknowledges Mr. Teppei Ito for his help collecting the data for this study.
References
Blair, R.M.;Fang, H.;Branham. W, S.:Hass, B. S.;Dia1, S. L;Moland. C. L.;Tong, W.;Shi, L. l
Perkins, R l Sheehan, D. M,(2000)The estrogen receptor relative binding aMnities of 188 natural and xenochemicals:structural diversity of ligands. Toxicol. Sci,54,138-153,
Bradbury, S. P.;Mekenyan, O. G.;Ankley, G T.(1996)Quantitative structure-activity rela-
tionships for polychlorinated hydroxybiphenyl estrogen receptor binding affinity:An
30
Takahiro SじzじKI and Stuart SHAprRoassessment of conformer flexibility?Environ.7’oxicol. Chem.,15,1945-1954.
Brunstrom. B. l Axelsson, J.;Halldin, K.(2003)Effects of endocrine modulators on sex differen-
tiation in birds. Ecotoxicology l 2,287-295.
Brzozowski, A. M.;Pike, A, C. W.:Dauter. Z.;Hubbard, R. E.;Bonn, T.;Engstr6m, O.;
ぷ b
Ohman, L.;Geene. GL.:Gustafsson, J.-A,;Carlquist, M.(1997)Molecular basis of agonism and antagonism in the oestrogen receptor.∧「a tu re 389,753-758.
Colborn, T.;Saal, F.S.V.;Soto. AM.(1993)Developmental effects of endocrine disrupting chemicals in wildlife and humans. E〃v〃on. Ueαlth Perspect.101,378-384,
Connors, K. A.(1997)The stability of cyclodextrin complexes in solution. Chem.1~et,,97,
1325-1357.
European Commission(2001)Communication from the Commission to the Council and the European Parliament on the Implementation of the Community Strategy for Endocrine Dis-
ruptors-aRange of Substances Suspected of Interfering with the Hormone Systems of Humans and Wildlife, COM 262丘nal, Brussels.14 June 2001, http://europa.eu.int/comm/
environment/docum/01262 en.htm#bkh.
Gantchev, T., T. G.;Ali, H,;van Lier. JE.(1994)Quantitative structure-activity relatiorlships/
comparative molecular field analysis(QSAR/CoMFA)for receptor-binding properties of halogenated estradiol derivatives.ノニMed. Che〃1.37,4164-4176.
Gao, H.;Williams, C.;Labute, P.;Bajorath, J.(1999a)Binary quantitative structure-activity relationship(QSAR)analysis of estrogen receptor ligands.ノ. Chem. lnf. Compert. Sci.39,
164-168.
Gao, H.:Katzenellenbogen, J. A.;Garg, R.;Hansch, C,(1999b)Comparative QSAR analysis of estrogen receptor ligands. Che〃z..fi~ev.99.723-744.
Gao, H.(2001)AppIication of BCUT metrics and genetic algorithm in binary QSAR analysis. J.
C乃θ〃z.Inf Co〃z♪zat. Sci.41,402-407.
Hedges, AR.(1998)Industrial applications of cyclodextrins. Che〃2. Rev.98,2035-2044.
Higuchi, D.;Connors, K.A., in Reilley, C.N,(1965)∠4dvances in Analytical Che〃listry and Instrumentation, vol.4, Interscience, New York, p.132.
Hutchinson, T.H.;Pickford, D.B、(2002)Ecological risk assessment and testing for endocrine disruption in the aquatic environment. Toxicology 181,383-387.
Katritzky, A.R;Fara, D. C.:Yang, H.;Karelson, M.;Suzuki, T.:Solov’ev, V. P.;Varnek, A.
(2004)Quantitative structure-property relationship modeling ofβ一cyclodextrin complex-
ation free energies..み Chem. Inf Co〃1ノ)ut. Sci.44,529-541.
Kavlock, RJ.(1996)Research needs for risk assessment of health and environmental effects of endocrine disruptors:areview of the US. EPA-sponsored workshop. Environ. Health Per-
sl)ect.104.715-740.
Kier, L, B,:Hall, L. H,(1986)Molecular Connectivめ, in Structure-.4ctivめ,∠4〃α己v∫i∫;Research Studies Press:Letchworth, Herts., U.K.
Klein, C, T.;Polheim, D.;Viernstein, H.;Wolschann. P,(2000)Amethod for predicting the free energies of complexation betweenβ一cyclodextrin and guest molecules.ノ. Inclusion l)heno〃1.⊥Wiacro CJ,c〃c Che〃1.36,409-423,
Kramer, V. J,;Giesy, J. P.(1999)Specific binding of hydroxylated polychlorinated biphenyl rnetabolites and other substances to bovine calf uterine estrogen receptor:structure-bind-
ing relationships. Sci.?〔otαl Environ.233,141-161,
QSARs of environmental estrogens
31
Lipkowitz, K. B.(1998)Applications of computational chemistry to the study of cyclodextrins.
Chem.1~el.i.98,1829-1873.
Liu, L;Guo, Q.-X.(1999)Wavelet neural network and its application to the inclusion of β一cyclodextrin with benzene derivativesJ Che〃z. Inf Comput. Sci.39,133-138.
Loftsson, T,;Duch6ne. D.(2007)Cyclodextrins and their pharmaceutical applications. Int.ノ f)ノtaプm.329,1-11.
Lydersen, A, L.(1955)Estimation of critical propertles of organic compounds by group contri-
butions.乙%⑫. Wisconsin. Co〃. Eng. Expt. Stn.1~ep., no.3.
Matsui, Y.;Nishioka, T,;Fujita, T.(1985)Quantitative structure-reactivity analysis of the inclusion mechanism by cyclodextrins.7「⑳. Curr. Chem.128,61-89.
Park. J、 H.;Nah. T.H.(1994)Binding forces contributing to the complexation of orgarlic mole-
cules withノβ一cyclodextrin in aqueous solution.ノChem. Soc., Parkin T〆ans,2,1359-1362.
Roncaglioni, A.;Novi6. M.;Vra6ko, M.;Benfenati, E.(2004)Classification of potential endo-
crine disruptors on the basis of molecular structure using a nonlinear modeling method.1.
Che〃1. Inf Co〃zPL‘t. Sci.44.300-309.
Sadler, B.R. l Cho, S. J.;Ishaq, K.S.;Chae, K.;Korach, K.S.(1998)Three-dimensional quantita-
tive structure-activity relationship study of nonsteroidal estrogen receptor ligands using the
ワ
comparative molecular且eld analysis/cross-validated r-guided region selection approach.∫.
ルfε∂.Che〃t.41,2261-2267.
Shi, L. M.:Fang, H.;Tong, W.;Wu, J.;Perkins, R,;Blair, R. M.;Branham, W. S.;Dial, S. L;
Moland, C. L.;Sheehan. D, M.(2001)QSAR models using a large diverse set of estrogens./.
Che〃τ. Inf Co〃1ノ)ut. Sci.41,186-195.
Singleton, D.W.;Khan, S.A.(2003)Xenoestrogen exposure and mechanisms of endocrine dis-
ruption. Fro〃’. Biosci,8.110-118.
Suzuki, T.:Ishida, M.:Fabian, W.MF,(2000)Classical QSAR and comparative molecular field anaIyses of the host-guest interaction of organic molecules with cyclodextrins.ノComput.-
Aidedルtol.1)θ∫.14,669-678.
Suzuki, T.;Ide, K.;Ishida. M,;Shapiro, S.(2001)Classification of environmental estrogens by physicochemical properties using principal component anaiysis and hierarchical cluster anal-
ysis.ノChe〃z. Inf Co〃zput. Sci.41,718-726.
Suzuki, T.(2001)Anonlinear group contribution method for predicting the free energies of inclusion complexation of organic molecules withα一andβ一cyclodextrins.1. Chem./㎡
Co〃zノ)ut. Sci.41.1266-1273.
Szejtli, J.(1988)(]」・clodextrin Technology, Kluwer Academic Publishers:Dordrecht.
Szejtli, J.(1998)Introduction and general overview of cyclodextrin chemistry. Chem. Rev.98,
1743-1753、
Tong, W.;Perkins, R.:Xing, L.;Welsh, WJ.;Sheehan, D.M,(1997a)QSAR models for binding of estrogenic compounds to estrogen receptorαandβsubtypes, Endocrinol(]g),138,4022-
4025.
Tong, W.:Perkins. R.lStrelitz, R;Collantes, E.R. l Keenan, S.:Welsh, WJ.;Branham, W.S.l Sheehan, D.M,(1997b)Quantitative structure-activity relationships(QSARs)for estrogenic binding to the estrogen receptor:Predictions across species. Ent’iron、」Uealth Pε7巧りect.105.
1116-1124,
Tong, W.;Lowis, D.R:Perkins, R;Chen, Y.:Welsh, W.J.;Goddette, D.W.;Heritage, TW.;
32
Takahiro SuzuKI and Stuart SHAplRoSheehan, D.M.(1998)、Evaluation of quantitative structure-activity relationship methods for Iarge-scale prediction of chemicals binding to the estrogen receptor. 1. Chem. 1勿『Comput.
Sci.38,669-677.
Uekama, K.;Fujinaga, T.;Hirayama, F.;Otagiri, M.:Yamasaki, M;Seo, H;Hashimoto, T.;
Tsuruoka, M,(1983)Improvement of the oral bioavailability of digitalis glycosides by cyclo-
dextrin complexation..入f)harm. Sci,72,1338-1341.
Waller, C.L;Minor, D.L;McKinney, JD、(1995)Using three-dimensional quantitative struc-
ture-activity relationships to exarnine estrogen receptor binding affinities of polychlorinated hydroxybiphenyls. Environ.」Uealth Perspect.103,702-707.
Waller, C.L,:Oprea, T.1.;Chae, K.;Park, H.-K.;Korach, K.S.;Laws, S.C;Wiese, TE.;Kelce,
W.R.;Gray, LE., Jr.(1996)Ligand-based identi丘cation of environmental estrogens. Che〃1.
」1~es. Toxicol.9,1240-1248,
Waller, C.L.(2004)Acomparative QSAR study using CoMFA, HQSAR, and FRED/SKEYS paradigms for estrogen receptor binding affinities of structurally dlverse compoundsJ Che〃1. Jnf Co〃IPut. Sci.44,758-765.
Wallimann, P.;Marti, T.l FOrer, A.;Diederich, F.(1997)Steroids in molecular recognition.
Che〃t.1~ev.97.1567-1608.
Yoshida, A.;Yalnamoto, M.;Hirayama F.;Uekama, K.(1988)Improvement of chemical insta-
bility in aqueous solution by complexation withβ一cyclodextrin Derivatives. Chem. Pharm.
BulL 36,4075-4080.
Zheng、 W.;Tropsha, A.(2000)Anovel variable selection QSAR approach based on the k-nearest neighbor principle.ノ Che〃z. Inf Co〃tput. Sci.40,185-194.