Bisphenol A [BPA; 2,2-bis(4-hydroxyphenyl)propane]
has a symmetrical chemical structure of HO-C6H4-C(CH3)2-C6H4-OH.
BPA is used mainly in the production of polycarbonate plastics
and
epoxy resins. Its worldwide manufacture is approximately 3.2
million
metric tons/year. BPA has been acknowledged to be an estrogenic
chemical able to interact with human estrogen receptors (ER)
(Dodds and Lawson 1938; Krishnan et al. 1993; Olea et al.
1996), and many lines of evidence have revealed that BPA, at
even low doses, acts as an endocrine disruptor (Gupta 2000;
Nagel et al. 1997; vom Saal et al. 1998; Welshons et al. 2003).
However, its binding to and hormonal interaction with ER are
extremely weak, 2–3 orders of magnitude lower than those
of natural hormones, and thus the intrinsic significance of
these low-dose effects is rather intangible and obscure (Safe
et al. 2002). These facts led us to hypothesize that BPA may
interact with nuclear receptors (NRs) other than ER.
We have recently demonstrated that BPA
binds strongly to estrogen-related receptor-γ (ERR-γ)
with high constitutive activity (Takayanagi et al. 2006). ERR-γ
is a member of the human NR family and the estrogen-related
receptor
(ERR)
subfamily of orphan NRs, which are closely related to the ERs
ER-α and ER-β (Gigučre 2002; Horard and Vanacker
2003). The ERR family includes three members—ERR-α,
ERR-β, and ERR-γ—with
ERR-γ being the most
recently identified (Eudy et al. 1998; Hong et al. 1999). The
amino acid sequences are quite highly conserved among ERRs and
ERs, but 17β-estradiol (E2), a natural ligand of
ERs, does not bind to any of the ERR family members. Our discovery
that BPA binds
strongly to ERR-γ, but not to ERs, indicates that the effects
of the so-called endocrine disruptors should be examined for
all
NRs without delay.
ERR-γ is expressed in a tissue-restricted
manner—for example, very strongly in the mammalian brain
during development, and then in the brain, lung, and many other
tissues during adulthood (Eudy et al. 1998; Heard et al. 2000;
Lorke et al. 2000). Our preliminary results have shown that the
highest expression is brought about in the placenta (Takeda Y,
Sumiyoshi M, Liu X, Matsushima A, Shimohigashi M, Shimohigashi
Y, unpublished data). Strong binding of BPA to ERR-γ would
affect not only the physiologic functions but also the metabolism
of
this NR as a transcription-activating factor. Although the
intrinsic physiologic functions of ERR-γ have
not yet been clarified, it is crucial that a structure–function
study be performed to clarify the structural requirements for
the
binding of BPA to ERR-γ.
|
Figure 1. The
saturation binding analysis of BPA for ERR-γ. (A) Saturation binding curve
of [3H]BPA for the recombinant human ERR-γ–LBD showing
total, nonspecific, and specific binding. Determination of
nonspecific binding was carried out by excess unlabeled BPA (10
µM). (B) Binding data analyzed by Scatchard plot analysis to
estimate the dissociation constant (KD) and the
receptor density (Bmax). The
plot was linear, the KD value
was estimated to be 5.50 ± 0.87 nM, and Bmax was 14.4 nmol/mg protein. The saturation binding
analysis was performed in duplicate and repeated four times.
|
|
Figure 2. Chemical
structure of BPA and its derivatives and their
dose–response curves in the radioligand receptor binding
assay for ERR-γ. (A) Chemical structures of BPA (two methyl groups) and
its derivatives: bisphenol B (a methyl group and an ethyl
group) and bisphenol AP (a methyl group and a phenyl group). (B) Binding activities
of BPA, bisphenol B, and bisphenol AP examined by the
competitive binding assay using [3H]BPA and GST–ERR-γ–LBD.
(C) Chemical structures
of bisphenol E (one methyl group) and its derivatives,
bisphenol F and bisphenol AF [two trifluoromethyl groups (CF3)]. (D) Binding activities
of BPA, bisphenol E, bisphenol F, and bisphenol AP examined by
the competitive binding assay. (B) and (D) each show representative curves with the IC50 values
closest to the mean IC50 from at least five independent assays for each
compound. B/B0 is the relative inhibitory activity estimated
from the calculation of the percentage of displacement by the
chemical tested (B) against the specific binding (B0 = 100%)
of [3H]BPA.
|
|
Figure 3. Chemical
structure of BPA and its derivatives lacking the hydroxyl
group(s) and their dose–response curves in the
radioligand receptor binding assay for ERR-γ. (A)
Chemical structure of BPA and its derivatives lacking the hydroxyl
group(s): 4-α-cumylphenol
(without one hydroxyl group from BPA), and 2,2-diphenylpropane
(without either hydroxyl groups from BPA). (B) Binding
activities of BPA, 4-α-cumylphenol, and 2,2-diphenylpropane
examined by the competitive binding assay using [3H]BPA
and GST-ERR-γ–LBD; representative curves indicate the
IC50 value
closest to the mean IC50 from at least five independent
assays for each compound.
|
|
Figure 4. Chemical
structure of BPA and its derivatives lacking the phenol group
and their dose–response curves in the radioligand
receptor binding assay for ERR-γ. (A) Chemical structure of BPA and its derivatives with
the alkyl group at the position of phenol group: 4-tert-butylphenol (a
methyl group); 4-tert-amylphenol (an ethyl group); and 4-tert-octylphenol (a tert-butyl methyl
group). (B) Binding activities of BPA, 4-tert-butylphenol, 4-tert-amylphenol, and
4-tert-octylphenol
examined by the competitive binding assay using [3H]BPA
and GST-ERR-γ–LBD; representative curves indicate the
IC50 value
closest to the mean IC50 from at least five independent
assays for each compound.
|
|
Figure 5. Chemical
structure of BPA and a series of alkyl phenols and their dose–response
curves in the radioligand receptor binding assay for ERR-γ.
(A) Chemical structure of BPA and its derivatives with
the alkyl group at the para position: 4-isopropylphenol (a 4-isopropyl
group); 4-ethylphenol (an ethyl group); p-cresol (a methyl group);
and phenol (a hydrogen atom). (B) Binding activities of BPA, 4-isopropylphenol, 4-ethylphenol,
p-cresol,
and phenol examined by the competitive binding assay using [3H]BPA
and GST-ERR-γ–LBD; representative curves indicate the
IC50 value
closest to the mean IC50 from at least five independent
assays for each compound.
|
|
Figure 6. Luciferase-reporter gene assay of BPA
and its derivatives for human ERR-γ. (A) Deactivation
of the fully activated human ERR-γ by the inverse
agonist 4-OHT and sustainment by BPA. (B)
Reversing activity of BPA, bisphenol E, and bisphenol AF against
the inverse agonist
activity of 1.0 µM 4-OHT; 1.0 µM 4-OHT exhibited
approximately 0.4-fold deactivation, and the inhibitory
activities are shown by the percentage of relative activity.
(C) Sustainment of the
fully activated human ERR-γ by bisphenol E and bisphenol
AF together with inverse agonist activity by 4-OHT. (D)
Reversing activity of BPA, 4-α-cumylphenol, and 4-tert-butylphenol;
the inverse agonist activity of 4-OHT was clearly reversed by
all
bisphenols tested in a dose-dependent manner. Data are from a
single experiment performed in triplicate; two additional
experiments gave similar results. High basal constitutive
activity of ERR-γ was evaluated with the luciferase-reporter
plasmid (pGL3/3 ξ ERRE), and the highest activity was estimated
in a cell preparation of 1.0 ξ 105 HeLa cells/well.
|
Table 1.
|
Table 2.
|
Table 3.
|
In a previous study (Takayanagi et al.
2006), we used tritium (3H)-labeled 4-hydroxytamoxifen
(4-OHT) as a tracer in a receptor binding assay for ERR-γ.
4-OHT binds strongly to ERR-γ and deactivates it as an inverse
agonist, decreasing the very high level of spontaneous constitutive
activity (Coward et al. 2001). As a substitute for [3H]4-OHT,
BPA was found to be as potent as 4-OHT in this binding assay.
Furthermore, BPA was found to retain or rescue ERR-γ's
high basal constitutive activity in the reporter gene assay for
ERR-γ using HeLa
cells. These results indicated that BPA and 4-OHT bind to ERR-γ
with equal strength, but have structural differences that affect
their
occupation of ERR-γ's ligand binding pocket. In the
complex formed between 4-OHT and the ERR-γ–ligand binding
domain (LBD), 4-OHT remained at the ligand binding pocket of
ERR-γ–LBD,
but the α-helix 12 of
the receptor was repositioned from the activation conformation
(Greschik et al. 2004; Wang et al. 2006). In contrast, BPA was
suggested to bind to the pocket without changing the
positioning of helix 12, and thus preserved the high receptor
constitutive activity of ERR-γ.
It is evident that the binding ability of
BPA to ERR-γ should be examined by means of tritium-labeled
BPA. Fortunately, [3H]BPA is now commercially available; thus, in
the present study we performed the first saturation binding
assay for direct exploration of the binding characteristics of
BPA. We then established a competitive receptor binding assay
in which chemicals were assessed for their ability to displace
[3H]BPA from the receptor binding pocket. In particular,
industrial chemical products of BPA analogs were inspected
structurally in order to better understand the structural
elements of BPA that are required for binding to the ERR. Here
we describe the structural elements of BPA that are required
for the binding to ERR-γ–LBD and for maintaining the
receptor in an active conformation.
Chemicals. We
purchased 2,2-bis(4-hydroxyphenyl)propane and
4,4-isopropylidene-diphenol, both denoted as BPA, from Tokyo
Kasei Kogyo Co. (Tokyo, Japan), Nakarai Tesque (Kyoto, Japan),
Aldrich (Madison, WI, USA), Junsei Chemical (Tokyo, Japan),
Acros (Geel, Belgium), Lancaster Synthesis (Windham, NH, USA),
Merck (Darmstadt, Germany), and Fluka (Buchs, Switzerland). The
purity designated on the labels varied from 95 to 99%. We also
obtained the following analogs of BPA: bisphenol AF
[2,2-bis(4-hydroxyphenyl)hexafluoropropane; Tokyo Kasei],
bisphenol AP [4,4´-(1-phenylethylidene)bisphenol; Tokyo
Kasei], bisphenol B [2,2-bis(4-hydroxyphenyl)butane; Tokyo
Kasei], bisphenol E [2,2-bis(4-hydroxyphenyl)ethane; Aldrich],
and bisphenol F [bis(4-hydroxyphenyl)methane; Tokyo Kasei].
4-α-Cumylphenol [2-(4-hydroxyphenyl)-2-phenylpropane],
4-tert-amylphenol,
4-tert-butylphenol,
4-isopropylphenol, and 4-ethylphenol were obtained from Tokyo
Kasei. 2,2-Diphenyl propane, and 4-tert-octylphenol were obtained from Aldrich, and p-cresol and phenol
from Kishida Chemical (Osaka, Japan).
Preparation of receptor protein GST-fused
ERR-γ–LBD. ERR-γ–LBD
was amplified from a human kidney cDNA library (Clontech
Laboratories, Mountain View, CA, USA) by polymerase chain
reaction (PCR) using gene-specific primers and cloned into
pGEX6P-1 (Amersham Biosciences, Piscataway, NJ, USA).
Glutathione S-transferase (GST)–fused receptor protein
expressed in Escherichia coli BL21α was purified on
an affinity column of glutathione-sepharose 4B (GE Healthcare
Bio-Sciences Co.,
Piscataway, NJ, USA) to obtain GST-ERR-γ–LBD.
The glutathione used for elution of GST-ERR-γ–LBD
from the column was removed by gel filtration on a column of
Sephadex
G-10
(15 ξ 100
mm; GE Healthcare Bio-Sciences Co.) equilibrated with 50 mM
Tris-HCl (pH 8.0), and the protein content (506.24 µg/mL)
was estimated by the Bradford method using a Protein Assay CBB
Solution (Nakarai Tesque). Preparation of GST-fused ER-α–LBD
was carried out as described previously (Takayanagi et al. 2006).
Radioligand binding assays for saturation
binding. The saturation binding
assay for GST-ERR-γ–LBD was conducted at 4°C using
[3H]BPA
(5 Ci/mmol; Moravek Biochemicals, Brea, CA, USA) with or without
BPA (10 µM in the final solution). Purified protein (0.32
µg/mL) was incubated with increasing concentrations of
[3H]BPA
(2.1–24.3 nM) in a final volume of 100 µL of
binding buffer [10 mM HEPES (pH 7.5), 50 mM sodium chloride,
2 mM magnesium chloride, 1 mM EDTA, 2 mM CHAPS
{3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate},
and 2 mg/mL γ-globulin]. Nonspecific binding was determined
in a parallel set of incubations that included 10 µM
nonradiolabeled BPA. After incubation for 2 hr at 4°C, all
the fractions were filtered by the direct vacuum filtration
method (MultiScreenHTS HV, 0.45 µm pore size; Millipore,
Billerica, MA, USA) for the B/F separation (the separation of
receptor-bound ligand from free ligand) (Nakai et al. 1999).
Filtration was carried out on a multiscreen separation system
(Millipore). Before filtration, 100 µL of 1%
dextran-coated charcoal (DCC) (Sigma) in phosphate buffer (pH
7.4) was added to the assay vessels, and the mixture was incubated
for 10 min on ice. The radioactivity of the filtered solution
was counted on a liquid scintillation counter (LS6500; Beckman
Coulter, Fullerton, CA, USA). The saturation assay was
performed in triplicate. The specific binding of [3H]BPA
was calculated by subtracting the nonspecific binding from the
total binding.
Radioligand binding assays for competitive
binding. BPA
and the BPA-related chemicals were dissolved in a binding buffer
containing
0.3–1.0% N,N-dimethylsulfoxide (DMSO). These compounds were
examined for their ability to inhibit the binding of [3H]BPA
(3 nM in the final solution) to GST-ERR-γ–LBD (0.32 µg/mL
in the final solution). The reaction mixtures were incubated
for 2 hr at 4°C and free radioligand was removed with 1%
DCC by filtration as described above. Radioactivity was
determined on a liquid scintillation counter (TopCount NXT;
PerkinElmer Life Sciences Tokyo, Japan). The IC50 values
(the concentrations for the half-maximal inhibition) were
calculated from the dose–response curves obtained using
the nonlinear analysis program ALLFIT (De Lean et al. 1978).
Each assay was performed in duplicate and repeated at least
three times. The competitivebinding assay for GST-ER-α–LBD
was carried out as described above using [3H]E2 (5.74 TBq/mmol;
Amersham Biosciences, Buckinghamshire, UK).
Cell culture and transient transfection
assays. HeLa
cells were maintained in Eagle's MEM (EMEM; Nissui, Tokyo, Japan) in the
presence of 10% (vol/vol) fetal bovine serum at 37°C. For
luciferase assays, HeLa cells were seeded at 5 ξ 105 cells/6-cm
dish for 24 hr and then transfected with 4 µg of reporter
gene (pGL3/3ξERRE) and 3 µg of ERR-γ expression plasmids
(pcDNA3/ERR-γ) by Lipofectamine Plus reagent (Invitrogen
Japan, Tokyo, Japan) according to the manufacturer's protocol.
Approximately 24 hr after transfection, cells were harvested
and plated into 96-well plates at 5 ξ 104 cells/well.
The cells were then treated with varying doses of chemicals
diluted with 1% bovine serum albumin/phosphate-buffered saline
(BSA/PBS, vol/vol). To measure the antagonistic activity, a
fixed concentration of compounds (10–5 M to
10–10 M in the final solution) was added along
with 4-OHT. After 24 hr, luciferase activity was measured with
the
appropriate reagent using a Luciferase Assay System (Promega,
Madison, WI, USA) according to the manufacturer's
instructions. Light emission was measured using a Wallac 1420
ARVOsx multilabel counter (PerkinElmer). Cells treated with 1%
BSA/PBS
were used as a vehicle control. Each assay was performed in
triplicate and repeated at least three times.
Highly specific binding of BPA to ERR-γ. To
demonstrate the direct binding of BPA to ERR-γ, we first
attempted to establish a saturation receptor binding assay
using radiolabeled BPA. We analyzed the saturation binding of
[3H]BPA
against the recombinant ERR-γ–LBD protein, to which
GST was fused at the N-terminus. In the actual receptor binding
assay, we used
[3H]BPA
(2.0–24 nM) against purified protein at a concentration
of 0.32 µg/mL, which corresponds to a concentration of
6.3 nM. The removal of receptor-free [3H]BPA
was carried out with 1% DCC. In this procedure, DCC mixtures
were
transferred to a 96-well HV-plate with a filter (0.45-µm
pore size) for direct vacuum.
As shown in Figure 1A, the binding of BPA
to ERR-γ was specific and saturated. Specific binding of
[3H]BPA to
ERR-γ was estimated to be approximately 80%, which we
judged to be a very high value. In other words, the level of
nonspecific binding of [3H]BPA was very low (Figure 1A). The high level
of specific binding of [3H]BPA clearly demonstrated
that BPA has no structural elements for nonspecific binding to
the receptor
protein and exclusively occupies the binding pocket of ERR-γ–LBD.
GST did not bind [3H]BPA at all. It should be noted that the
specific binding of [3H]4-OHT was only about 50% (Takayanagi et al.
2006).
The Scatchard plot analysis showed a
distinct single binding mode (Figure 1B). From the slope, the
binding affinity constant (KD) was
calculated to be 5.50 nM. The receptor density (Bmax) was estimated to be 14.4 nmol/mg protein,
which is roughly compatible with the calculated value of 18.9 nmol/mg
protein. The Bmax value of [3H]4-OHT
is much smaller than that of [3H]BPA. These results
further demonstrate that ERR-γ binds [3H]BPA very specifically and exclusively.
Binding ability of BPA to ERR-γ. We performed the competitive receptor binding
assay using [3H]BPA (3 nM in the final solution) for
GST–ERR-γ–LBD (0.32 µg/mL in the final
solution). To confirm that BPA is a truly specific ligand for
ERR-γ, we tested all nonradiolabeled BPA compounds
available in Japan, which we obtained from seven different
reagent companies. Because the compounds all had different
levels of purity (95–99%), we adjusted their initial
concentration, 1.0 ξ 10–2 M, based on the purity indicated on the label.
We found that BPA displaces [3H]BPA in
a dose-dependent manner. Its binding curve was sigmoidal in a
single binding mode (slope = ~ 1), which afforded an average IC50 value
of 9.78 nM. We found all BPA compounds purchased to be equally
potent. These results clearly demonstrate that BPA binds very
strongly to the NR ERR-γ.
4-OHT as a potent displacer of BPA in ERR-γ. 4-OHT has been reported to potently displace [3H]4-OHT
in the binding to ERR-γ (Greschik et al. 2004; Takayanagi et
al. 2006). In the present study, 4-OHT very potently displaced
[3H]BPA
(IC50 = 10.9 nM) (Table 1). BPA and 4-OHT yielded sigmoidal
binding curves indistinguishable from each other (data not
shown), indicating that the two are almost equipotent. These
results obtained using the [3H]BPA tracer were almost identical to those
obtained by [3H]4-OHT (Takayanagi et al. 2006).
BPA and 4-OHT share only a phenol group,
and thus the phenol groups of these compounds are highly likely
to occupy the same binding site in the ERR-γ receptor. Because
the phenol group of 4-OHT is anchored by hydrogen bonds to
Glu275 and Arg316 of ERR-γ (Greschik et al. 2004), the phenol
group of BPA may also bind to these ERR-γ residues. Indeed,
this has been proven by our recent X-ray crystal structure analysis
of the complex between
BPA and human ERR-γ–LBD (Matsushima et al. 2007). Hereafter,
we designate the benzene ring of this phenol group of BPA as
the A-ring and the additional benzene ring as the B-ring.
BPA-methyl as a structural requirement for
binding to ERR-γ. We evaluated the
role of the two methyl (CH3) groups on the sp3-C
atom of BPA in binding to ERR-γ by a series of
analogs of BPA, HO-C6H4-C(CH3)2-C6H4-OH. First, we examined the effect of
incorporation of the methyl group on the binding affinity of
BPA. When CH3 was incorporated into the parent methyl group
to produce HO-C6H4-C(CH3)(CH2CH3)-C6H4-OH (Figure 2A), we found the resulting
bisphenol B to be approximately half as potent (IC50 = 26.3
nM) as BPA (Table 1). This result clearly indicates that a
bulky group on the central sp3-C atom is obviously
disadvantageous in terms of the binding of BPA to ERR-γ's
binding pocket.
On the other hand, an enhancement of
activity was observed when one of the methyl groups was
eliminated from BPA. The resulting bisphenol E [HO-C6H4-CH(CH3)-C6H4-OH]
(Figure 2C) exhibited slightly better binding activity (IC50 =
8.14 nM) than BPA (Table 1). Bisphenol E is indeed the most potent
chemical to date for the NR ERR-γ (Figure 2D). The maximal
activity was attained when one of the methyl groups was removed
from BPA. Apparently,
the concomitance of two methyl groups on the central sp3-C atom
of BPA is disadvantageous and unfavorable.
The fact that a single methyl group had
the best fit for ERR-γ was further demonstrated by the diminished
activity of bisphenol AP, which has a phenyl group in place of
the hydrogen atom that is found in bisphenol E (Figure 2A).
Bisphenol AP exhibited approximately 15-fold weaker binding
affinity for ERR-γ than bisphenol E, with IC50 = 123 nM
(Figure 2B, Table 1). Steric hindrance by the benzene ring, as
well as its electron-rich characteristics, might be responsible
for this drop in the receptor binding affinity of bisphenol AP.
The importance of the remaining methyl
group in bisphenol E became evident from the drastically
reduced activity of bisphenol F [HO-C6H4-CH2-C6H4-OH].
This compound was approximately 16-fold less potent than
bisphenol E, exhibiting an IC50 value of 131 nM (Table 1). All of these results
clearly indicate that one of the two methyl groups is involved
in the intermolecular interaction with the receptor residue(s).
The interaction involving the CH3 group is a kind of hydrophobic interaction,
such as CH3-alkyl and CH/π interactions.
The fundamental nature of this interaction
involving the CH3 group became rather apparent from the binding
result of bisphenol AF [HO-C6H4-C(CF3)2-C6H4-OH]. The CH3ĆCF3 substitution
in BPA creates this compound (Figure 2C), which has two
electron-rich trifluoromethyl CF3 groups instead of the rather electron-poor
methyl CH3 group. The molecular size of CF3 is
almost equal to that of CH3. A drastically reduced
activity of bisphenol AF, about 35-fold less potent (358 nM)
than BPA (Table 1), thus
demonstrates that the BPA's CH3 group is in
an electrostatic interaction with the electron-rich residue(s)
of the receptor. Replacement of CH3 with CF3 is definitely disadvantageous, because CF3 is
very electron-rich and thus brings about a strong repulsion with
such electron-rich residues of the receptor. One of the
electron-rich candidates of the receptor is the aromatic ring
of Phe, Tyr, His, and Trp. Based on the reported X-ray crystal
structure of ERR-γ, feasible candidates are Phe-435 and Phe-450
(Greschik et al. 2002, 2004; Matsushima et al. 2007; Wang et
al.
2006).
A single phenol-hydroxyl group is enough
for BPA to bind to ERR-γ. BPA has a very
simple symmetrical chemical structure of HO-C6H4-C(CH3)2-C6H4-OH
(Figure 2A). When one of the phenol-hydroxyl groups (–OH)
of BPA was eliminated, the resulting 4-α-cumylphenol (HO-C6H4-C(CH3)2-C6H5;
Figure 3A) still bound very strongly to ERR-γ. 4-α-Cumylphenol
was as potent as BPA (Figure 3B), having an IC50 value
of 10.6 nM (Table 2). Contrary to the expectation that both of
the phenol-hydroxyl groups of BPA would participate in the
hydrogen bonds, this result indicates that the second hydroxyl
group does not necessarily participate in the hydrogen bonding.
Given that this hydroxyl group forms a hydrogen bond with the
ERR-γ receptor residue, the bond would be considered
extremely weak, as suggested by the X-ray crystal analysis of
4-α-cumylphenol–ERR-γ complex
(Matsushima A, Teramoto T, Okada H, Liu X, Tokunaga T, Kakuta
Y, Shimohigashi Y, unpublished data).
When both of the phenol-hydroxyl groups
were eliminated from BPA, the resulting 2,2-diphenyl propane
[C6H5-C(CH3)2-C6H5] was
almost completely inactive (Figure 3B, Table 2). This compound
elicits only about 30% inhibition of the binding of [3H]BPA
at the 1-µM concentration, whereas BPA almost completely
inhibits the binding of [3H]BPA at this concentration
(Figure 3B). It is clear that one of the phenol-hydroxyl groups
of BPA is
indispensable for the interaction with a binding pocket of ERR-γ.
These results, together with the fact that 4-α-cumylphenol
and BPA are equipotent, emphasizes the significance of one of
the two
phenol groups in the interaction of BPA with ERR-γ. As described
above, this hydroxyl group should be attached to the benzene
A-ring.
It became apparent that the phenol-hydroxyl group attached to
another phenol-benzene ring (B-ring) is not necessarily
required for binding of BPA to ERR-γ.
BPA-phenol as a structural requirement for
binding to ERR-γ. As described
above, 4-α-cumylphenol is as active as BPA. The importance
of the benzene B-ring can be examined by replacing the B-ring
with
the alkyl groups. When the benzene B-ring of 4-α-cumylphenol
was substituted with either methyl or ethyl, the resulting 4-tert-butylphenol
[HO-C6H4-C(CH3)2-CH3] and 4-tert-amylphenol [HO-C6H4-C(CH3)2-CH2CH3] (Figure 4A) were considerably potent (Figure 4B),
with values of 26.1 nM and 33.2 nM, respectively (Table 2).
This reveals that alkyl groups can be substituted for the
aromatic benzene ring without affecting the basal binding
capability.
However, because both 4-tert-butylphenol and 4-tert-amylphenol
are still a few times less active than 4-α-cumylphenol, a
specific binding site of ERR-γ appears to prefer the aromatic
benzene ring to the alkyl groups. This suggests that BPA's
second phenol-phenyl group (benzene B-ring) is in the π interaction
with the receptor residue(s), that is, either a XH/π interaction
(X = N, O, and C) or a π/π interaction. The most plausible
candidate for the receptor residue in this interaction is the
Tyr residue at
position 326 of ERR-γ. Indeed, the phenol-hydroxyl group
of this Tyr-326 was found in the OH/π interaction with the B-ring
of BPA (Matsushima et al. 2007).
In a BPA molecule, two C6H4-OH
(phenol) groups are connected to the sp3 carbon atom
(sp3-C) together with two CH3 (methyl)
groups. The most simple structure–activity study is to
compare the activity of compounds lacking one of these groups.
The compound that lacks
the phenol group is 4-isopropylphenol [HO-C6H4-CH(CH3)2]
(Figure 5A), and this para-isopropyl phenol was fairly potent at
displacing [3H]BPA (Figure 5B), with an IC50 value
of 71.1 nM (Table 2). However, 4-isopropylphenol was still
approximately 7-fold less active than BPA, indicating that the
phenol backbone structure is an essential structural element
for the binding to ERR-γ.
When one of the two methyl groups was
eliminated from 4-isopropylphenol, the resulting 4-ethylphenol
[HO-C6H4-CH2-CH3] (Figure 5A) was found to be very weakly active
(289 nM) (Table 2). Elimination of another methyl group still
afforded a compound of inactive p-cresol [HO-C6H4-CH3], but with the IC50 value
being approximately 1.3 µM. Phenol
[HO-C6H5] tended to bind to ERR-γ (Figure
5B). These results clearly indicate that the phenol group is
a core
structure for the attachment of BPA to ERR-γ.
4-Alkyl phenols as putative potent binders
to ERR-γ. Attachment of the methyl
group to 4-isopropylphenol [HO-C6H4-CH(CH3)2] to create 4-tert-butylphenol [HO-C6H4-C(CH3)3]
considerably facilitates the binding of the phenol derivative
to ERR-γ (Table 2). 4-tert-Amylphenol [HO-C6H4-C(CH3)2-CH2CH3] is almost as active as 4-tert-
butylphenol.
However, 4-tert-octylphenol [HO-C6H4-C(CH3)2-CH2-C(CH3)3] (Figure 4B) was significantly weaker
(approximately 10 times less potent) than 4-tert-butylphenol (Table 2).
Thus, the activities of HO-C6H4-C(CH3)2-CH(CH3)2,
HO-C6H4-C(CH3)2-CH(CH3)3, HO-C6H4-C(CH3)2-CH2-CH2-CH3, and HO-C6H4-C(CH3)2-CH2-CH(CH3)2 are expected to be intermediate between those
of 4-tert-amylphenol and 4-tert-octylphenol, although these molecules are not
commercially available. It appears that, among the 4-alkylphenols
of HO-C6H4-C(CH3)2-CnH2n+1 (=R), 4-tert-butylphenol (R = CH3) and 4-tert-amylphenol (R = CH2-CH3)
show the maximum competitive activity with the binding of ERR-γ.
The structural comparison of HO-C6H4-C(CH3)2-CH3 (4-tert-butylphenol),
HO-C6H4-C(CH3)2-CH2CH3 (4-tert-amylphenol), and BPA HO-C6H4-C(CH3)2-C6H4-OH
clearly indicated that the R group should not be bulky for high
receptor binding activity. A plain π electron-rich benzene aromatic
ring is thus optimal for interaction with the receptor residue
of ERR-γ-Tyr326.
Inhibitory activity of BPA derivatives for
ERR-γ. We found that BPA
retained a high constitutive basal activity of ERR-γ in the
luciferase reporter gene assay (Figure 6A). ERR-γ is
in a full activation with no ligand; it is one of the self-activated
NRs
and is deactivated by the so-called "inverse
agonists" such as 4-OHT (Greschik et al. 2004; Takayanagi
et al. 2006). Although BPA shows no apparent effect on the high
basal activity of ERR-γ, BPA evidently antagonizes or inhibits
the deactivation activity of 4-OHT in a dose-dependent manner
(Figure 6B), as reported by Takayanagi et al. (2006). This
neutral antagonist is a distinct inhibitor or suppressor of the
inverse agonist, reversing the deactivation conformation to the
activation conformation.
All of the potent BPA derivatives (i.e.,
bisphenol E, bisphenol AF, 4-α-cumylphenol, and 4-tert-butylphenol)
were found, just like BPA, to retain a high constitutive basal
activity of ERR-γ in the same
luciferase reporter gene assay (Figure 6C). In addition, these
compounds inhibited the inverse agonist activity of 4-OHT and
thus were specific inhibitors against the inverse agonist
4-OHT. Their abilities to antagonize 4-OHT are approximately
one order lower than their binding potencies to ERR-γ (Figure
6B,D). This discrepancy is probably caused by the inclusion
of a
number of co-effecter proteins for eliciting a gene expression
in the luciferase reporter gene assay.
Receptor selectivity of BPA derivatives
for ERR-γ over ER-α. We
classified BPA and its derivatives into the four groups, depending
on their
receptor binding affinity for ERR-γ: that is, group A, BPA
and chemicals as potent as BPA; group B, chemicals considerably
potent; group C,
chemicals moderately potent; and group D, inactive chemicals.
All chemicals were then examined for their ability to bind to
ER-α, and the affinity measured was compared respectively
with that for ERR-γ (Table 3). As reported previously (Takayanagi
et al. 2006), BPA is highly selective for ERR-γ. It binds
to ER-α only weakly; we calculated BPA's receptor
selectivity to be 105, which suggests that BPA prefers ERR-γ
105 times more strongly than ER-α. Other group A compounds,
namely,
bisphenol E and 4-α-cumylphenol, were also greatly selective
for ERR-γ (Table 3). In
particular, bisphenol E was found to be exclusively selective
and specific for ERR-γ because it was almost completely inactive
for ER-α.
para-Alkyl
phenols in group B (IC50ERR-γ =
of 26–71 nM) were also almost
completely inactive for ER-α (Table 3). Those include 4-tert-butylphenol, 4-tert-amylphenol,
and 4-isopropylphenol, and they were fully selective and specific
for ERR-γ. In contrast, bisphenol B was very weakly active
(246 nM) for ER-α, although it was still selective (about
9.5 times) for ERR-γ.
Among group C chemicals (IC50ERR-γ =
120–350 nM), bisphenol F was almost completely inactive
for ER-α, making it fully selective for ERR-γ (Table
3). This was also true for 4-ethylphenol. Bisphenol AP showed
a weak binding
affinity (361 nM) for ER-α, but it was still selective (about
3 times) for ERR-γ. However, bisphenol AF emerged as a ligand
selective for ER-α with a selectivity ratio of 0.15 (Table
3). The reciprocal of 0.15 [i.e., ERR-γ (IC50)/ER-α
(IC50)
= 6.67] denotes a selectivity ratio of bisphenol AF for ER-α.
The results clearly indicate that the
alkyl groups on the central sp3-C atom of bisphenol
derivatives play a key role in selection of the NR ERR-γ
and ER-α. When we checked the receptor binding
activities of one series of bisphenol derivatives (i.e.,
bisphenol E, BPA, bisphenol B, bisphenol AP, and bisphenol AF),
we found this line-up to be the order of compounds with
increasing affinity to ER-α. At the same time, it was the
order of compounds with decreasing affinity to ERR-γ. ERR-γ
prefers the less bulky and less electrophilic alkyl groups, whereas
ER-α appears
to prefer the bulkier and more electrophilic alkyl groups.
4-tert-Octylphenol is a well-known endocrine
disruptor candidate, but it was only moderately potent for ERR-γ
(IC50 =
238 nM; Table 2). However, it was considerably weak for ER-α,
with an IC50 of 925
nM; thus, we judged 4-tert-octylphenol to be somewhat
selective (approximately 4 times) for ERR-γ. Another representative
endocrine disruptor candidate is 4-nonylphenol, which was moderately
active for ERR-γ (Takayanagi et
al. 2006). Thus, 4-nonylphenol was slightly more selective for
ERR-γ. However, some 4-alkyl phenols are distinctly more
potent for ERR-γ than 4-tert-octylphenol and 4-nonylphenol: 4-tert-butylphenol, 4-tert-amylphenol,
and 4-isopropylphenol. These 4-alkyl phenols are definitely
novel
candidates of the endocrine disruptor specific for ERR-γ.
In the present study
we have shown that all the structural elements of BPA—the
phenol and methyl groups and the phenyl group on the central
sp3-C
atom—are prerequisite for binding to the NR ERR-γ.
Furthermore, we have shown that the phenol derivatives are potent
candidates
for the endocrine disruptor that binds to ERR-γ. The binding
affinity of [3H]BPA to ERR-γ–LBD is extremely
high, with a KD value of 5.50 nM. Thus, it
appears to be important to evaluate whether the previously reported
effects
of BPA at low doses are mediated through ERR-γ and its specific
target gene(s).
At the same time, it is necessary to
clarify the physiologic roles of ERR-γ and to examine the
degree and ways in which BPA may influence these. This is
particularly important because ERR-γ is expressed in a tissue-restricted
manner—for example, it is expressed very strongly in the
mammalian fetal brain and placenta—at sites that could
have important outcomes for newborns. Recently, many lines of
evidence have indicated that low doses of BPA affects the
central nervous system (reviewed by vom Saal and Welshons 2005;
Welshons et al. 2003, 2006). The molecular mechanism for these
effects could involve, at least in part, the high affinity
binding of BPA to ERR-γ. A similar phenomenon may be observed
for other NRs, and the exploration of such chemical–receptor
interactions requires a specific assay system or concept
applicable to all the NRs.
