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 radio-labeled 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 , the binding of BPA to ERR-γ was specific and saturated. Specific binding of [3
H]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 [3
H]BPA was very low (). The high level of specific binding of [3
H]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 [3
H]BPA at all. It should be noted that the specific binding of [3
H]4-OHT was only about 50% (Takayanagi et al. 2006
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 (more ...)
The Scatchard plot analysis showed a distinct single binding mode (). 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 × 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 [3
H]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 [3
= 10.9 nM) (). 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 [3
H]BPA tracer were almost identical to those obtained by [3
H]4-OHT (Takayanagi et al. 2006
Receptor binding affinity (mean ± SE) of BPA and its analogs, and 4-OHT for ERR-γ.
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 (), we found the resulting bisphenol B to be approximately half as potent (IC50 = 26.3 nM) as BPA (). 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.
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 (more ...)
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] () exhibited slightly better binding activity (IC50 = 8.14 nM) than BPA (). Bisphenol E is indeed the most potent chemical to date for the NR ERR-γ (). 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 (). Bisphenol AP exhibited approximately 15-fold weaker binding affinity for ERR-γ than bisphenol E, with IC50 = 123 nM (, ). 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 (). 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-C6
-OH]. The CH3
substitution in BPA creates this compound (), 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 (), 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
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
; 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 (). When one of the phenol-hydroxyl groups (–OH) of BPA was eliminated, the resulting 4-α-cumylphenol (HO-C6H4-C(CH3)2-C6H5; ) still bound very strongly to ERR-γ . 4-α-Cumylphenol was as potent as BPA (), having an IC50 value of 10.6 nM (). 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).
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): (more ...)
The receptor binding affinity (mean ± SE) of BPA and its derivatives lacking of the phenol group for human ERR-γ .
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 (, ). 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 (). 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] () were considerably potent (), with values of 26.1 nM and 33.2 nM, respectively (). This reveals that alkyl groups can be substituted for the aromatic benzene ring without affecting the basal binding capability.
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 (more ...)
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] (), and this para-isopropyl phenol was fairly potent at displacing [3H]BPA (), with an IC50 value of 71.1 nM (). 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.
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 (more ...)
When one of the two methyl groups was eliminated from 4-isopropylphenol, the resulting 4-ethylphenol [HO-C6H4-CH2-CH3] () was found to be very weakly active (289 nM) (). 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 (). 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 (). 4-tert-Amylphenol [HO-C6H4-C(CH3)2-CH2CH3] is almost as active as 4-tert-butylphenol. However, 4-tert-octylphenol [HO-C 6 H 4 -C(CH 3 ) 2 -CH 2 -C(CH 3 ) 3 ] () was significantly weaker (approximately 10 times less potent) than 4-tert-butylphenol (). Thus, the activities of H O - C 6 H 4 - C ( C H 3 ) 2 - C H ( C H 3 ) 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-amyl-phenol 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), HOC6H4-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 (). 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 (), 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.
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 (more ...)
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 (). 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 (). 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 (). 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 (). In particular, bisphenol E was found to be exclusively selective and specific for ERR-γ because it was almost completely inactive for ER-α.
Receptor binding affinity (mean ± SE; n = 3) of BPA and its analogs for ER-α and their receptor selectivity for ERR-γ over ER-α .
para-Alkyl phenols in group B (IC50ERR-γ = of 26–71 nM) were also almost completely inactive for ER-α (). 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 (). 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 (). 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.
-Octylphenol is a well-known endocrine disruptor candidate, but it was only moderately potent for ERR-γ (IC50
= 238 nM; ). 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
-amylphenol, and 4-isopropylphenol. These 4-alkyl phenols are definitely novel candidates of the endocrine disruptor specific for ERR-γ.