Introduction

With the discharge substances and decomposition substances of pesticides, washing liquid, plastic industry, and so on, a large number of environmental endocrine-disrupting chemicals (EDCs) are produced in nature [1, 2]. EDCs possess hormonal and estrogen-like activity that are environmentally stable and not easily destroyed. So they can be enriched through the food chain in the ecological environment. Upon entering the human body, they can combine with the corresponding “receptors,” resulting in changes in the biochemical reaction of the body, and potentially causing abnormal changes in the body and reproductive system [3,4,5]. For example, 4-isopropylphenol is a compound commonly used in antimicrobial and antifungal products for disinfection, sterilization, and preservation. Additionally, it is utilized as a food additive to keep food fresh and extend its shelf life. In the pharmaceutical field, it is also employed in the production of certain drugs, exhibiting specific pharmacological effects [6]. Besides, bisphenol F (BPF) is commonly used in industry as a raw material for manufacturing epoxy resins and other plastic products. Additionally, BPF is also utilized in coatings, sealants, adhesives, and other industrial products [7]. Due to the fact that 4-isopropylphenol and BPF belong to the group of phenols, it may potentially cause some health and environmental issues. Therefore, it is of great significance to study these endocrine disruptors to find potential substitutes.

Human estrogen-related receptor γ (hERRγ) is an important nuclear receptor [8]. It is involved in regulating glucose and lipid metabolism pathways, influencing the differentiation and activation of brown adipocytes, as well as mitochondrial biogenesis and function. Abnormal activation or inhibition of hERRγ may lead to endocrine disruption, affecting the balance of the body’s endocrine system and causing metabolic diseases such as obesity and diabetes [9]. Besides, hERRγ can also regulate the expression of various genes, affecting cell cycle regulation and the direction of cell differentiation. By interacting with other transcription factors and regulatory factors, hERRγ participates in a complex network regulating cell proliferation and differentiation. Aberrant hERRγ activity may lead to uncontrolled cell proliferation or abnormal cell differentiation, closely associated with diseases such as obesity and cancer [10]. hERRγ mainly including ligand independent activation function 1 (AF-1), DNA binding domain (DBD), and ligand binding domain (LBD) [11]. LBD contains another ligand-dependent transcriptional activation region, which presents a different configuration binding to different estrogens and determines which co-activators and co-inhibitors bind to transcription target gene. LBD is composed of two double-stranded antiparallel β-sheets (S1 and S2) and 12 α-helices (H1-H12) with a hydrophobic binding pocket of the EDCs. Most EDCs in the external environment can be located in this pocket, thereby interfering with downstream signaling pathways [12]. When its normal function is affected by EDCs such as 4-isopropylphenol and BPF, it can severely affect normal function of the body [13,14,15,16]. Experimental research revealed that 4-isopropylphenol and BPF have low binding capacity with hERRγ, with EC50 values of about 300 and 645 nM, respectively [12]. However, the specific details are currently unclear. Furthermore, lignin, as a renewable raw material, is a natural polymer that exists in plant cell walls. For example, renewable bisphenols derived from lignin, BP(2,2)(Un) is considered as the potential alternative of commercial bisphenols [17]. With its abundance of aromatic hydroxyl and methoxy groups, this compound is utilized in the production of bioplastics and bio-based chemicals, contributing to a decrease in reliance on petroleum resources and minimizing environmental impact. In addition, reports by molecular docking methods have shown that BP(2,2)(Un) has weak binding ability to ERα [17]. Nevertheless, the dynamic interactions between these three EDCs with hERRγ at the micro level have not yet been came to light. Therefore, in-depth study on the interaction mechanisms of hERRγ by these EDCs is crucial for deciphering the endocrine disrupting mechanisms of hERRγ, which contributes to comprehensively evaluate latent diseases associated with EDCs exposure.

In recent years, in silico approaches such as molecular docking and molecular dynamics (MD) simulation [18,19,20,21,22,23] have emerged as reliable tools for exploring the three-dimensional structure of protein complexes with small compounds, identifying key binding residues, evaluating interaction strength, conducting energy analysis, elucidating interaction mechanisms, understanding ligands binding with proteins pockets, and estimating the stability of the complexes. Hence, in our work, molecular docking and multiple MD simulations were executed to obtain the conformation of hERRγ-EDCs and to describe the dynamic and detailed interactions between them, which are difficult for experiment studies. We also performed molecular mechanics/Poisson-Boltzmann surface area (MM-PBSA) [24] plus solvation interaction energy (SIE) methods [25, 26] to investigate thoroughly the recognition mechanisms of hERRγ by these EDCs, including 4-isopropylphenol, BPF, and BP(2,2)(Un) (Fig. 1), and important residues distributions were also evaluated. The results could provide novel insights into estrogenic disruption effects of EDCs.

Fig. 1
figure 1

Cartoon structure of hERRγ-4-isopropylphenol complex (A, PDB ID: 6I65) and molecular structures of three EDCs (B)

Full size image

Methods

Molecular systems

Complexes coordinates of hERRγ-4-isopropylphenol were acquired from RCSB (PDB ID: 6I65) [12]. The binding site of 4-isopropylphenol to hERRγ was used as the active site for docking study. Initially, the EDC 4-isopropylphenol was redocked with the experimental protein hERRγ to verify the robustness of the docking procedure. Upon superimposing the experimental and docked complexes, it was observed that the conformation remained consistent across both structures (Fig. S1). Since there were no complexes of hERRγ with BPF and BP(2,2)(Un) were available, AutoDock software [27] was conducted to dock these two molecules to the binding site of hERRγ. During docking, the initial structure and coordinate of the complex were obtained by retaining crystal water. Docking between ligands, BPF and BP(2,2)(Un), and hERRγ was conducted by AutoDockTools graphical user interface. We applied the semi-flexible docking approach, maintaining the rigidity of hERRγ while allowing flexibility in these three EDCs. The grid box was defined according the 4-isopropylphenol binding pocket. The grid center coordinates on hERRγ were defined as − 15.28, Y − 5.15, and − 28.06 in the X, Y, and Z directions. The grid size was set as 60 Å × 60 Å × 60 Å, and spacing between the grid points was 0.375 Å. Docking was performed with Lamarckian genetic algorithm and default parameters. And then, the strongest binding mode was retained for further analysis according to the molecular docking mode. Lack hydrogen atoms were then appended applying tLeap program in Amber 18 [28]. Standard Amber ff14SB force filed [29] and general amber force filed [30] plus AM1-BCC charges were performed to form force field parameters of hERRγ and these EDCs, respectively. Antechamber program in Amber was used to assign appropriate force field atom types for these three EDCs. Truncated octahedral periodic box composing TIP3P water [31] was placed into each system. At the same time, distance of water box with outermost atoms was ensure to be no less than 12 Å. Overall, each system included over 12,000 water molecules, and the volumes of the solvated box are about 410,000 Å3. Besides, to stay neutrality, 12 Na+ was affiliated to each system using the tLeap procedure according to the Coulomb potential grid.

MD simulation

In order to do away with adverse interaction between atoms, firstly, the systems underwent steepest descent of 4000 steps and conjugate gradients of 8000 steps under 50 kcal mol−1 Å−2 constraint. Subsequently, they were further minimized in 8000 steps with all atoms unrestricted. After minimization, applying harmonic restraints with 10 kcal mol−1 Å−2 force constants on the solute atoms, the temperature was progressively increased to 310 K in 600 ps, then an equilibrium process of 1000 ps. As can be seen from the plots of energy, temperature, and pressure (Fig. S2), clearly, during equilibrium, all complexes showed high stability. Finally, three repeated 250 ns simulations for each system were eventually conducted under NPT ensemble using periodic boundary conditions. Randomly assign three seeds with different values to repeated system according to Maxwell’s distribution. The system temperature was kept at 310 K by coupling to a Langevin heatbath with a collision frequency of 1 ps−1, while maintaining a constant isotropic pressure at 1 bar using the Berendsen barostat. The time step was set to 2 fs. Particle mesh Ewald method [32] to handle remote electrostatic interactions was conducted. SHAKE algorithm [33] dealt with bonds associating hydrogen atoms. Coordinate trajectories were recorded every 2 ps for subsequent analysis. Hydrogen bonds are considered when the distance between the donor and acceptor is less than 3.5 Å and the angle formed is less than 120°. Electrostatic interactions are considered when the distance is below 4.5 Å. For the hydrophobic interactions, functional groups with a distance shorted than 5.0 Å are taken into account. Data analysis was mainly carried out by cpptraj program in Amber. The PyMOL software [34] was utilized to visualize the trajectories and create structural representations.

MM-PBSA calculation

We extracted 3000 snapshots from the equilibrated portions at 200-ps intervals for energy calculations (ΔGbind) by MM-PBSA method [35,36,37,38,39] using single trajectory in amber. ΔGbind of these EDCs (Gligand) binding to hERRγ (Greceptor) to yield the complex (ΔGcomplex) is broken down into distinct energy items as follows:

$$\Delta {\text{G}}_{bind}={\text{G}}_{complex}{-{\text{G}}}_{receptor}-{\text{G}}_{ligand}$$
(1)
$$\Delta {\text{G}}_{bind}=\Delta \text{H} - \Delta {\text{S}}$$
(2)
$$\Delta {\text{H}}=\Delta {\text{E}}_{ele}+\Delta {\text{E}}_{vdW}+\Delta {\text{G}}_{PB}+\Delta {\text{G}}_{SA}$$
(3)
$${\Delta {\text{G}}}_{SA}\text{=}\gamma \Delta SASA+\beta$$
(4)

Enthalpy (ΔH) includes electrostatic interaction energy (ΔEele), van der Waals interaction energy (ΔEvdW), polar solvation energy (ΔGPB), and nonpolar solvation energy (ΔGSA). ΔGPB is estimated using PB model, and ΔGSA is calculated by solvent accessible surface area (ΔSASA). γ and β in the formula are 0.00542 kcal mol−1 Å−2 and 0.92 kcal mol−1, respectively. In addition, the ionic strength was 0.1 M. Dielectric constants of solute and solvent were 2.0 and 80.0, respectively [40]. Conformations entropy change (TΔS) is usually computed using Normal-mode analysis [41]. Considering computational cost, entropy is calculated based on the equilibrated trajectories by extracting 100 snapshots at 6-ns intervals.

Apart from conducting binding free energy calculations, the molecular mechanics/generalized born surface area (MM-GBSA) energy decomposition method was conducted by taking into account molecular mechanics and desolvation energies without accounting for entropy contributions. To scrutinize the impact of each residue, the total energy between hERRγ and EDCs was dissected into individual residues to pinpoint the key residues involved in the interaction with these EDCs.

SIE calculation

The energy between EDCs and hERRγ (ΔGbind) was also calculated with the sietraj program, which employed the SIE method for computation [25, 26, 42]. The same 3000 snapshots at 200-ps intervals as MM-PBSA methods were performed for SIE calculations. The SIE function utilizes the following physical parameter dependencies as outlined below:

$$\Delta {\text{G}}_{bind}(\rho ,{\text{D}}_{in},\alpha ,\gamma ,{\text{C}})=\alpha [{\text{E}}_{Coul}({\text{D}}_{in})+{\text{E}}_{vdw}+\Delta {\text{G}}_{\text{R}}(\rho ,{\text{D}}_{in})+{\text{G}}_{cav}(\rho )+{\text{C}}]$$
(5)
$${\text{G}}_{\text{cav}}=\gamma \Delta {\text{M}}_{SA}$$
(6)

ΔGbind includes intermolecular Coulomb energy (ECoul), van der Waals (EvdW), reaction energy (ΔGR), and nonpolar solvation energy (Gcav). ΔGR and Gcav were computed based on boundary element method and solvent-accessible surface area (ΔMSA), respectively.

The parameters of ρ (linear scaling factor of van der Waals radii), Din (internal dielectric invariable of solute), α, γ, and C are obtained according to energy values measured in experiment and were 1.1, 2.25, 0.1048, 0.0129 kcal mol−1 Å−2, and − 2.89 kcal mol−1, respectively.

Conformational dynamics analysis

To intuitively observe internal structure changes of hERRγ induced by different EDCs, cross-correlation matrix (Cij) was performed [43] as follows:

$${\text{C}}_{\text{ij}}=\frac{<\Delta {\text{r}}_{\text{i}}\Delta {\text{r}}_{\text{j}}>}{{(<\Delta {\text{r}}_{\text{i}}^{2}> <\Delta {\text{r}}_{\text{j}}^{2}>)}^{2}}$$
(7)

The sharp brackets and Δri represent the mean simulation time and displacement from mean position of ith atom, respectively. The Cij values are taken from − 1 to 1. If Cij is greater than zero, it means that atoms i and j are positively correlated; otherwise, it shows that atoms i and j are negatively correlated.

Principal component analysis

To reveal motion changes of functional significance, principal component analysis (PCA) [44,45,46] was conducted. Based on ProDy software [47], according to the sampling from the joined equilibrated trajectories, covariance matrix of diagonal coordinate system is constructed by PCA, and the movement of hERRγ can be observed as follows: (a) project trajectories through direction depicted by corresponding eigenvector, (b) compute the first two eigenvalues, thus determining the highest directions of hERRγ. Three-dimensional structure snapshots of hERRγ were visualized by VMD [48] software with its plug-in NMWiz.

Results and discussion

MD simulation analysis

To insight into overall structural stability, root mean square deviations (RMSDs) of these complexes were measured (Fig. S3). Obviously, all complexes showed high stability fluctuation around 2.2 Å in the last 200 ns time of each simulation. The stability of the simulated trajectory can also be demonstrated from the evolution of the rotation radius, SASA, and hydrogen bond over time in Fig. S4. Thus, the following analysis was carried out for equilibrated 600 ns trajectories after stabilization. For complexes of hERRγ and 4-isopropylphenol, BPF, BP(2,2)(Un), and RMSD values are 2.35, 2.13, and 1.98 Å, respectively. Clearly, during simulations, hERRγ in three complexes did not experience primary structural transformation, which is compatible with the outcome of experimental by Thouennon et al. [12] and can also be observed in Figs. S5 and S6. All simulated hERRγ-EDC complexes converged and displayed stable trajectories, validating the stability of these hERRγ-EDC complexes and also confirming the applicability of simulation methods for them. Moreover, by selecting structures from the simulations and calculating enthalpy and entropy values, stability of simulated trajectory can be further inferred (Figs. S7 and S8). Although enthalpy and entropy reckoned for various snapshots are quite different, their cumulative mean became stable expeditiously within equilibrated 600 ns trajectory. It leads to conclude equilibrated 600 ns trajectory should be rational and trustworthy for subsequent analysis.

For investigating the reason for difference of structure and flexibility of hERRγ when binding with EDCs, root mean square fluctuations (RMSF) was calculated (Fig. S9). On the whole, RMSF values of hERRγ varied similarly across the studied systems, except for variations in individual regions. Most varied residues are in the N-terminal, C-terminal, and loops in hERRγ, and residues in β-sheets S1 and S2 varied as well. Whereas for these complexes, residues near the active pockets have lower values, indicating hydrophobic binding pockets formed in the above regions maintain strong stability during simulations. Interestingly, among all these complexes, loops H1-H2, S1-S2, H7-H9, and H11-H12 showed distinct flexibility. What’s more, to investigate movements of different regions for hERRγ when binding to distinct EDCs, correlated fluctuations of Cα were reckoned (Fig. S10). Among these complexes, internal motions in domains D1 among H2 with H3-H4, D2 among H2 with s1, s2, as well as H5, D3 among H3-H4 with H7-H8, D4 among H4-H5 with H5-H7, D5 among H5-H6 with H10-H12, and D6 among H7-H8 with H8-H9 were more obvious changes. It can be seen that the above deviations for internal dynamics of hERRγ corresponded to the relative position changes of residues when binding to various EDCs, thus leading to changes in their binding affinity.

Binding free energy analysis

To describe differences in the interactions of various EDCs with hERRγ, MM-PBSA was first applied to reckon different energy terms (Table 1). Binding free energies between 4-isopropylphenol, BPF and hERRγ are − 5.39 and − 5.16 kcal/mol, respectively. The order of energy calculation was in accordance with experimental consequence for hERRγ-4-isopropylphenol (− 8.89 kcal/mol) and hERRγ-BPF (− 8.44) complexes [12]. After calculating energy errors, the order of energy results was still the same as that of removing the errors. Moreover, binding free energy were − 4.84 kcal/mol for hERRγ-BP(2,2)(Un) complex, which has a similar trend to ERα [17]. The differences between the energy obtained by MM-PBSA method and the experimental results are mainly caused by the following aspects [35, 36, 39, 40]: (1) variations in the dielectric constant ε across protein internal residues are not considered in MM-PBSA calculations; (2) the discrepancy in water models used between MM-PBSA calculations (implicit solvent) and MD simulations (explicit solvent); (3) the limited sampling of conformational entropy due to computational constraints can hinder direct comparisons with experimental data; and (4) energy calculations based on initial structures instead of multiple MD snapshots reduce computational time but overlook dynamic effects, leading to predictions heavily reliant on initial structures and lacking statistical accuracy information. Nevertheless, the MM-PBSA method provides a reliable ranking of energies across various systems, demonstrating a strong correlation with experimental findings. In analysis of composition of these various complexes, nonpolar energy, particularly van der Waals, acts as primary driving force for EDCs interacting with hERRγ. The hERRγ-BP(2,2)(Un) complex showed the highest van der Waals energy (− 46.30 kcal/mol), whereas hERRγ-4-isopropylphenol and hERRγ-BPF complexes are 23.98 and 20.02 kcal/mol lower than that of hERRγ-BP(2,2)(Un). At the same time, electrostatic interaction energy of these complexes also afforded beneficial contributions. When interacting with hERRγ, BPF brought out superior electrostatic interaction energy (− 22.66 kcal/mol) than 4-isopropylphenol (− 14.81 kcal/mol) and BP(2,2)(Un) (− 5.49 kcal/mol). Among them, electrostatic interaction energies of BP(2,2)(Un) interacting with hERRγ w oppositely minor. The above results were consistent with the following analysis about hydrogen bonds. Electrostatic solvation interactions generated by various EDCs with hERRγ disrupt their binding, whereas nonpolar solvation interactions showed opposite trend. For entropy in hERRγ-4-isopropylphenol, hERRγ-BPF, and hERRγ-BP(2,2)(Un) complexes are 14.65, 14.86, and 22.19 kcal/mol, respectively. These discrepancies were chiefly caused by vibrational entropy, which was related to structural transformation of hERRγ interacting with various EDCs. For the hERRγ-BP(2,2)(Un) complex, the increase in conformational entropy is mainly due to the expansion of hydrophobic side chain size through the two methoxy substituents, and the increase in steric hindrance around the binding pocket leads to the structural instability of the complex. This should be one of the reasons for their weakened energy.

Table 1 Binding free energy and its terms (kcal/mol) of EDCs-hERRγ complexes by MM-PBSA
Full size table

To verify the above energy results, SIE approach was also performed to assess their binding energy (Table 2). Energies between hERRγ and 4-isopropylphenol, BPF, and BP(2,2)(Un) are − 6.30, − 6.23, and − 5.93 kcal/mol, respectively. The energy order was also consistent with experimental measurements [12, 17]. By analyzing the results obtained by both calculation methods, we believed that they are used trustworthy to further explore the interactional ways of EDCs with hERRγ. In analyzing various energy terms, nonpolar interactions (ΔGnonpol), including ΔEvdW and non-polar contributions (ΔGcav) play positive role in interactions between these EDCs and hERRγ. Therefore, hydrophobic interaction was the major driving force for these EDCs interacting with hERRγ. What’s more, ΔEvdW, ΔECoul, and ΔGcav played the active role in their binding. ΔGR (reaction energies) are between 5.26 and 9.30 kcal/mol, which are not conductive to interact between various EDCs and hERRγ. Despited ΔECoul also generated an active function on the binding of EDCs to hERRγ; this effect was offseted by ΔGR. To sum up, hydrophobic interactions consisting mainly of van der Waals energies play a crucial role in stabilization of the EDCs-hERRγ complexes. Results of the above analysis were consistent with our preceding MM-PBSA analysis and experimental measurements [12, 17].

Table 2 Binding free energy and its terms (kcal/mol) of EDCs-hERRγ complexes by SIE
Full size table

Key residues in stabilizing complexes

For investigating hotspots of hERRγ binding to EDCs, interactions between these three EDCs and residues were analyzed. The key residues, identified based on a contribution exceeding − 0.5 kcal/mol, are shown in Fig. 2 and listed in Tables S1–S3. In short, four regions near L271, L309, L345, as well as F435 played main roles in interacting with various EDCs. Hydrophobic residues are key residues that mainly affected the binding of various EDCs to hERRγ such as leucine. And they are almost verified by static structures from experiment [7]. For instance, with 4-isopropylphenol, residues E275, and R316 formed hydrogen bonds, while F435 formed C–H-π interactions. Almost all key residues binding with EDCs obviously possessed smaller polar interactions and somewhat larger non-polar interactions (Tables S1–S3 and Fig. 3). As bound to various EDCs, in addition to residue E275, a good many of residues that took part in polar interactions are involved in hydrogen bonds (Fig. 4 and Table S4), which were not listed when the probability is small.

Fig. 2
figure 2

Important residues contributions for different complexes. A hERRγ-4-isopropylphenol. B hERRγ-BPF. C hERRγ-BP(2,2)(Un)

Full size image
Fig. 3
figure 3

Nonpolar interaction and polar interaction energy for key residues. A hERRγ-4-isopropylphenol. B hERRγ-BPF. C hERRγ-BP(2,2)(Un)

Full size image
Fig. 4
figure 4

Interactions between various EDCs and hERRγ. A hERRγ-4-isopropylphenol. B hERRγ-BPF. C hERRγ-BP(2,2)(Un)

Full size image

In light of Fig. 2A and Table S1, interactions between 4-isopropylphenol and ten residues are greater than − 0.5 kcal/mol. Remarkably, resides L271, A272, E275, L309, and V313 contributed more than − 1 kcal/mol. In the study of nonpolar interactions in hERRγ-4-isopropylphenol complex (Fig. 3A), we found that five residues, including L268, L271, A272, L309, and Y326, had energies above − 1 kcal/mol. While studying polar interaction (Fig. 3A), there is only one critical residue, E275, whose energy is above − 1 kcal/mol, as well as their distributions were shown (Fig. 4A). Energies of residues L268, L271, A272, and V313 with 4-isopropylphenol are − 0.98, − 1.37, − 1.16, and − 1.06 kcal/mol, respectively. They are consistent with the interaction of their hydrophobic alkyls with benzene group of 4-isopropylphenol. The isobutyl-methyl group of L309 and benzene group of Y326 in CH-π and T-shaped π-π interactions with 4-isopropylphenol contributing energies with − 1.51 and − 0.87 kcal/mol, respectively. The alkyl of I310 and alkyl group of 4-isopropylphenol produces − 0.93 kcal/mol energy. And interaction of van der Waals (− 0.60 kcal/mol) between F435 and side chain of 4-isopropylphenol is one of the essential causes for changing in energy (Fig. 4A). Interaction between L309 and 4-isopropylphenol is about − 1.51 kcal/mol. It related with interaction of CH-π formed by hydrophobic alkyl group of L309 with phenol group of 4-isopropylphenol. Hydrophobic interaction formed by benzene ring of F435 with alkyl of 4-isopropylphenol is about − 0.53 kcal/mol. This results from the benzene group of F435 and alkyl group of 4-isopropylphenol. While residue E275 provides − 2.96 kcal/mol energy with 4-isopropylphenol, mainly because the hydrogen bond occupancy of E275-OE1-H…OAC-4-isopropylphenol is about 92.27% (Table S4).

Observing hERRγ-BPF complex, 14 residues contributed more than 0.5 kcal/mol (Fig. 2B). Significantly, residues L268, L271, A272, E275, L309, V313, Y326, together with N345 have a major contribution exceeding − 1 kcal/mol. In accordance with Fig. 3B, the absolute value of four residues L268, L309, Y326, and L345 are greater than − 1 kcal/mol for nonpolar interaction in hERRγ-BPF complex (Tables S2). Besides, polar interaction of residue E275 over − 1 kcal/mol plays an active role in their interaction (Fig. 4B). Residues L271, E275, L309, I310, and F435 have a total energy resembling to that of hERRγ-4-isopropylphenol complex and also play the prominent interaction with BPF. The reason for these residues is similar to the role of them in hERRγ-4-isopropylphenol complex. Compared with hERRγ-4-isopropylphenol complex, energy of L268 is enhanced about 0.61 kcal/mol resulted from enhancing CH-π interaction forming by side chain of L268 with benzene of BPF. Increased interaction of hydrophobic alkyl of V313 and benzene group of BPF results into increased their nonpolar interaction. By comparison to hERRγ-4-isopropylphenol, interaction between alkyl group of A272 and phenol of BPF is reduced approximately 0.15 kcal/mol. It is caused by increased distance between these two groups. Interaction forming by hydrophobic benzene group of Y326 with benzene group of BPF is enhanced about 0.45 kcal/mol. The above outcome is consistent with decreased distance between them. Although BPF is present O–H…O interaction with carbonyl oxygen of Y326, but electrostatic energy in gas phase of Y326 is − 1.48 kcal/mol, this profitable factor is completely shielded by electrostatic energy of liquid phase, thus weakening their interaction. Interaction between I349 and BPF is increased by 0.52 kcal/mol by comparison to hERRγ-4-isopropylphenol complex. This seemed to be consistent with enhanced interaction between hydrophobic alkyl of I349 and benzene group of BPF. In addition, the oxygen atom of Ala431 and phenolic hydroxyls of BPF also form the hydrogen bond with the occupancy about 67.96; the residue provides about − 0.88 kcal/mol electrostatic energy with BPF. But this profitable factor is completely shielded by electrostatic energy of liquid phase, thus weakening their interaction (Table S4).

In hERRγ-BP(2,2)(Un) complex, 16 residues have energy values over − 0.5 kcal/mol (Fig. 2C). More important, 12 residues L268, L271, A272, L309, I310, V313, Y326, L342, L345, N346, I349, and F435 provide more than − 1 kcal/mol. When dissecting nonpolar interaction in hERRγ-BP(2,2)(Un) complex (Fig. 3C), absolute value of 11 residues, including L268, L271, L309, I310, V313, Y326, L342, L345, L346, I349, and F435, is over 1 kcal/mol (Tables S3 and Fig. 4C). Whereas, nearly all of key residues bound to BP(2,2)(Un) had minor polar interactions. Residues L268, L309, and I349 had a similar variation in total energy as other complexes, and also played important roles in hERRγ-BP(2,2)(Un). Carbonyl oxygen of L268 is nearby hydrophobic methoxyl of BP(2,2)(Un), which tended to form CO-OC interactions. The results showed that contribution of residue L268 to the interaction between BP(2,2)(Un) is − 3.16 kcal/mol. Hydrophobic benzene group of L342 and F435 located close methoxyl group of BP(2,2)(Un), and tend to form π-HC interactions, providing energy nearly − 1.54 and − 1.64 kcal/mol. Leu268, L271, Leu309, V313, L324, Y326, L342, L345, N346, and I349 showed enhanced interaction with BP(2,2)(Un) compared with hERRγ-4-isopropylphenol. It stems from reducing distance forming hydrophobic alkyls of them with methoxyl of BP(2,2)(Un). This suggested that these residues are of great importance in interaction with BP(2,2)(Un). CO-OC interactions is also found between arbonyl oxygen of N346 and hydrophobic methoxy group of BP(2,2)(Un), and thus electrostatic energy (− 0.68 kcal/mol) is the chief driving force. Overall, hydrophobic methoxide groups are the primary cause for the energy distinction of above pivotal residues. Two methoxy substituents ortho to the hydroxyl group enlarges the size of the hydrophobic side chain through the methoxy groups, and this effect may be due to an increase in steric hindrance around the binding pocket. This steric hindrance weakens the hydrogen bonds between BP(2,2)(Un) and several important residues such as Glu275, Tyr316, and Ala431, thereby reducing the binding of these phenolic hydroxyl groups to hERRγ, which should be one of the reasons for their weakened energy.

Based on the above analysis, we can infer these EDCs lead to local structure changes in loops H1-H2, H7-H8, together with H8-H9 mainly by influencing their interactions with various residues, which can be confirmed by the most representative structural superimposition in Figs. S11 and S12. Residues near L268, V313, L345, and F435 can stably bind to these EDCs, in which CH-π, π-π, and hydrogen bonds interactions are the key roles of these EDCs with important residues of hERRγ.

Conclusion

In this study, molecular docking, MD simulations with MM-PBSA, and SIE means were conducted to uncover conformational change and interaction of these three EDCs with hERRγ. When various EDCs bind to hERRγ, domains near residues L268, V313, L345, and F435 contribute significantly to the interaction of these hERRγ-EDCs complexes. And hydrophobic interactions consisting mainly of van der Waals energies play a crucial role in stabilization between them, which were essential for their interacting. What’s more, interactions of CH-π, π-π, and hydrogen bonds are essential in stabilization of EDCs with hERRγ. And most noticeable of all, the binding of the hydrophobic methoxide groups of BP(2,2)(Un) is useful for decreasing the binding ability between EDCs and hERRγ. The structural and energy information uncovered in our study can provide estimable knowledge for the search for further insights into the effects of EDCs on health problems.