Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Open Access

Peer-reviewed

Research Article

Molecular dynamics of the ERRγ ligand-binding domain bound with agonist and inverse agonist

Abstract

Estrogen-related receptor gamma (ERRγ), the latest member of the ERR family, does not have any known reported natural ligands. Although the crystal structures of the apo, agonist-bound, and inverse agonist-bound ligand-binding domain (LBD) of ERRγ have been solved previously, their dynamic behavior has not been studied. Hence, to explore the intrinsic dynamics of the apo and ligand-bound forms of ERRγ, we applied long-range molecular dynamics (MD) simulations to the crystal structures of the apo and ligand-bound forms of the LBD of ERRγ. Using the MD trajectories, we performed hydrogen bond and binding free energy analysis, which suggested that the agonist displayed more hydrogen bonds with ERRγ than the inverse agonist 4-OHT. However, the binding energy of 4-OHT was higher than that of the agonist GSK4716, indicating that hydrophobic interactions are crucial for the binding of the inverse agonist. From principal component analysis, we observed that the AF-2 helix conformation at the C-terminal domain was similar to the initial structures during simulations, indicating that the AF-2 helix conformation is crucial with respect to the agonist or inverse agonist for further functional activity of ERRγ. In addition, we performed residue network analysis to understand intramolecular signal transduction within the protein. The betweenness centrality suggested that few of the amino acids are important for residue signal transduction in apo and ligand-bound forms. The results from this study may assist in designing better therapeutic compounds against ERRγ associated diseases.

Citation: Sasidharan S, Radhakrishnan K, Lee J-Y, Saudagar P, Gosu V, Shin D (2023) Molecular dynamics of the ERRγ ligand-binding domain bound with agonist and inverse agonist. PLoS ONE 18(4): e0283364. https://doi.org/10.1371/journal.pone.0283364

Editor: Sandipan Chakraborty, Dr Reddy’s Institute of Life Sciences, INDIA

Received: October 6, 2022; Accepted: March 7, 2023; Published: April 6, 2023

Copyright: © 2023 Sasidharan et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: The raw simulations trajectory data of this study is publicly available on FigShare at DOI: 10.6084/m9.figshare.22306747.

Funding: V.G and D.S grant number: NRF-2018R1C1B6008141 (V.G) NRF-2022R1A2C4002510 (D.S) National Research Foundation of Korea The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests: The authors have declared that no competing interests exist.

Introduction

The nuclear receptor (NR) superfamily is a group of 48 ligand-regulated transcription factors in humans [1]. The amino acid sequence of the members of this superfamily has conserved regions, such as the DNA binding domain (DBD), which is highly conserved, the C-terminal ligand-binding domain (LBD), and the N-terminal domain, which is poorly conserved. The DNA-binding domain is responsible for the sequence-specific recognition of DNA in the regulatory regions of the target genes, while the C-terminal is capable of binding to small and lipophilic compounds that influence transcriptional regulation by interacting with cofactors. The LBD comprises 10–13 α-helices that are arranged like a three-layered sandwich, thereby regulating volume and function by specifically binding to molecules.

The estrogen-related receptor (ERR) subfamily consists of three orphan NRs, ERRα (NR3B1), ERRβ (NR3B2), and ERRγ (NR3B23), which are closely related to the estrogen receptors (ER) [2]. The regions of the DNA-binding domains of both classic ERs and orphan ERRs are conserved and therefore recognize common DNA-binding sites or response elements that are proximal to the target genes [3]. ERRs are capable of controlling classic ER target genes in the breast and bones [4, 5]. Although a certain degree of sequence identity exists between the ERR and ER LBD, there is a significant difference in their ligand binding capacities. ERRα and ERRβ were studied extensively before ERRγ was identified; therefore, not much is known about the role and dynamics of ERRγ [2, 6, 7].

ERRγ was first identified when it was linked to a region critical for Usher syndrome type IIa. Later, the receptor was discovered to interact functionally with the NR co-activator glucocorticoid receptor-interacting protein (GRIP) [3, 8]. This protein receptor is primarily expressed in metabolic tissues such as muscle, liver, brain, heart, and adipose tissues. The complete role and function of the receptor have not yet been elucidated, as ERRγ-null mice were found to be non-viable after birth [9]. However, tissue-specific ERRγ knock-out mice, ERRγ-specific ligands that can modulate the transcriptional activity of the receptor, and studies involving gain/loss of function have allowed researchers to understand the functions of this unique receptor [10, 11]. Indeed, studies have shown that ERRγ plays a vital role in the metabolic functions of the liver, such as the regulation of glucose, alcohol, and lipids, along with iron metabolism and modulation of specific gene expression in endocrine and metabolic processes [12, 13]. In addition, it is now well known that abnormal regulation of ERRγ results in hepatocellular carcinoma, as it regulates the expression of microRNA and DNA methyltransferase [14, 15].

As discussed earlier, ERRγ exhibits transcriptional activation of target genes in the absence of specific regulatory ligands; hence, the receptor is assigned as a constitutively active orphan NR. Structurally, ERRγ houses an N-terminal activation function (AF-1) domain, a central zinc finger DNA-binding domain (DBD) that binds to monomers or dimers to the core sequence TCAAGGTCA, referred to as the ERR response element (ERRE), and a C-terminal AF-2 domain that interacts with co-regulators (Fig 1). The crystal structure of the inverse agonist shows that the binding of the compounds to the small ligand pocket results in the displacement of the AF-2 helix and consecutively, the binding of the coactivator is blocked [16]. On the other hand, its agonist GSK4716, a micromolar acyl hydrazone ERRγ agonist, is capable of binding and activating receptors similar to PGC-1α, [17] but is 50 times more selective towards ERRγ. The X-ray crystal structure of the apo receptor has been shown to have an LBD in the active conformation with the AF-2 helix, which allows the binding of the coactivator [18]. The ligand-binding pocket of ERRγ has a total volume of 280 Å3, which is extremely small compared to that of ERα, which has a volume of 370 Å3 [16, 19]. To date, there are no endogenous small molecule ligands that have been discovered for ERRγ [20]. However, fasting-induced co-activator protein peroxisome proliferator-activated receptor γ coactivator 1-α (PGC-1α) has been known to bind to AF-2 and activate the receptor [21–23]. In addition to PGC-1α, GRIP-1, a co-activator, and co-repressors such as small heterodimer partner (SHP) and small heterodimer partner-interacting leucine zipper protein (SMILE), bind to AF-2 for functional activity [3, 24–26]. In addition to small proteins, studies have found that high-affinity ER inhibitors such as diethylstilbestrol [3] and 9-hydroxytamoxifen (4-OHT) [4] also have the capacity to be inverse agonists of ERRγ.

thumbnail
Download:
Fig 1.

Domain organization and structure of ERRγ (A) The full-length domain organization of ERRγ. (B) Ligand-binding domain (LBD) structure of ERRγ which comprises of 12 helices and a β-sheet (left). Agonist- (pale green color) and inverse agonist- (wheat color) bound ERRγ structures are overlaid (right). Ligands are shown in stick representation with their respective pockets shown in dots. AF2-helix conformational differences are shown in the black circle.

https://doi.org/10.1371/journal.pone.0283364.g001

In the current study, we performed all-atom molecular dynamics simulations to identify the intrinsic dynamics associated with the binding of the inverse agonist 4-OHT and the agonist GSK4716 to ERRγ. Previous studies have shown the effective use of molecular dynamics simulations to understand molecular interactions, allostery [27] and intrinsic dynamics of macromolecule complexes [28–32]. By utilizing the apo structure (ERRγapo), GSK4716-bound ERRγ (ERRγag), and 4-OHT-bound ERRγ (ERRγiag), the binding free energy between the ERRγ and ligand was studied. In addition, conformational changes with respect to the LBD and AF-2 helix were also analyzed. We further report the key amino acid residues from the betweenness centrality analysis, which may be important for the functional role of ERRγ.

Materials and methods

Computational resource and analysis of data

All computational works were performed on a Linux (Ubuntu) workstation. All-atom molecular dynamics (MD) simulations were performed using GROMACS v5.1.4 [33, 34] and the trajectories were analyzed using the in-built modules of the software. The residue network analysis was conducted using the NAPS web server [35, 36]. All graphical images were generated using PyMol [37]. Graphs were plotted using Microsoft Excel, and free-energy landscape (FEL) plots were generated using Mathematica.

Structure preparation of apo and ligand bound ERRγ

Previously, several structures have been solved for the ERRγ binding with agonist/ inverse agonist as well as coactivator molecules [16]. Since the aim of the study is to identify the crucial residues as well as binding free energies of agonist and inverse agonist, we considered only the ERRγapo and ERRγ bound with agonist or inverse agonist. At first, the crystal structures of ERRγapo, ERRγag, and ERRγiag were obtained from the Protein Data Bank (PDB) [16]. For ERRγapo, we retrieved the crystal structure of ERRγ ligand binding domain (PDB id: 2GP7) with a resolution of 2.45 Å. Whereas, for ERRγag, and ERRγiag, the X-ray crystal structure of ERRγ ligand binding domain with agonist (GSK4716) and inverse agonist (4-OHT) (PDB id: 2GPP and 2GPU, respectively) were retrieved with a resolution of 2.6 Å and 1.7Å, respectively. Water molecules and ions were removed from all the three structures. In ERRγag, the RIP140 protein bound to the crystal structure was removed, and the structure was saved using only GSK4716 as the ligand. The side chains of few residues in ERRγapo and ERRγiag had more than one conformation as well as missing in ERRγag as a result of low resolution from X-ray scattering in that region. Therefore, we applied mutation with the same residue to acquire better sidechain conformation with probable rotamers using mutation option in PyMol.

All-atom molecular dynamic (MD) simulation

Once the structures were prepared, the three systems were subjected to MD simulations using GROMACS v5.1.4 with the AMBER-ff99SB-ILDN force field [38]. Topology files of the Amber force field for the ligands were generated using the ACPYPE web server [39]. Some of the mismatching atom types of the corresponding atoms from the ACPYPE server were renamed to be compatible with the corresponding GROMACS atom types. Initially, the structures were placed in a dodecahedron box with a 1.2 nm distance between the protein and box wall. All three systems were solvated using the TIP3P water model and then neutralized by replacing the solvent molecules with either Na+ or Cl- ions. The systems were energy-minimized by the steepest descent algorithm with a maximum energy tolerance of 1,000 kJ/mol. The systems were then equilibrated using NVT and NPT ensembles for periods of 100 and 500 ps, respectively. Modified Berendsen thermostat (v-rescale) [40] was employed at 300 K for the NVT ensemble, while we employed Parrinello-Rahman barostat at 1 bar for NPT equilibration [41]. Following equilibration, the production run was performed for 500 ns, without considering any restraints. The long-range interactions were maintained with the help of the Particle Mesh Ewald algorithm, while the cut-off for short-range and long-range electrostatic interactions was set to 1.0 nm. The simulation grid space was set at 0.16 nm for FTT with fourth-order cubic interpolations. The simulations were performed with a time-step of 2 fs, and 2 ps coordinates were collected for the entire simulation trajectory. Subsequently, the trajectories were analyzed using in-built trajectory tools. The analysis (RMSD, Rg, SASA and H-bonds) were performed using 5000 frames (with 100 ps snapshot) from the entire trajectory. Binding free energy calculations were performed using the gmx mmpbsa tool (Molecular Mechanics Poisson-Boltzmann Surface Area) using the stable trajectory of 100ns (240 to 340 ns) of RMSD with 100 ps coordinates (1000 frames) [42].

Principal Component Analysis (PCA)

PCA can be employed to understand the dominant and large-scale motions of the three systems along with their collective modes. Initially, a covariance matrix was constructed for the three systems based on the fluctuations of the protein [43]. The coordinates for all three systems were extracted every 100 ps from the last 450 ns, the translational and rotational motions were eliminated, and the covariance matrix was diagonalized. After the sum of the eigenvalues for ERRγapo, ERRγag, and ERRγiag was calculated, the gmx anaeig was used to analyze the eigenvectors [44, 45]. The free energy landscapes (FEL) were constructed using the gmx sham tool to obtain the minimum-energy configurations and metastable states.

Residue interaction network (RIN) and Betweenness centrality (CB)

To further understand the intramolecular signal transduction within the protein, a non-weighted residue-residue interaction network was constructed using the NAPS web server [36] for the low-energy structures of apo and ligand-bound forms of ERRγ. In the network, each amino acid residue represents a node and the contacts represent edges. For network construction, we used a Cα-Cα pair maximum threshold of 7 Å with a non-weighted edge. The effect of the side chains, that is, the Cα backbone and the heavy atoms that are far away from the Cα of the side chains were considered by taking two coarse grain centers per residue. Contact was assumed to be established only if the above-defined atoms between two residues in the protein were present at a distance of less than 7 Å [46]. Using the residue interaction network, we calculated the betweenness centrality (CB). CB(u) measures signal transduction within the protein and is used to identify the central node in a protein. where σst (u) is the shortest path that connects nodes s and t through node u, and σst is the shortest path that connects nodes s and t [35].

Results

Structural deviation and compactness of both apo and ligand-bound ERRγ

The ERRγapo and ligand-bound forms (ERRγag and ERRγiag) were analyzed for conformational deviations with respect to the initial structure during simulations. The root mean square deviation (RMSD) of backbone atoms of all three models was calculated and the measured average was 0.1724 ± 0.02 nm, 0.1691 ± 0.02 nm and 0.1489 ± 0.01 nm for ERRγapo, ERRγag and ERRγiag, respectively. As shown in Fig 2A, all three systems showed similar structural deviations during the simulations indicating that the systems were well equilibrated. In particular, ERRγiag was largely stable compared to ERRγapo and ERRγag. When the RMSD of the ligands was analyzed, the agonist and inverse agonist showed lesser deviations from the initial structure, indicating that they were stable inside the ERRγ-LBD binding pocket (Fig 2B). Furthermore, the compactness of the protein was measured by calculating the radius of gyration (Rg) of the three systems over the simulation period. ERRγapo, ERRγag and ERRγiag had relatively stable compactness for ligand bound forms compared to ERRγapo suggesting a structural alteration upon ligand binding (Fig 2C). Taken altogether from the RMSD and Rg of the models, we assume that the three models exhibited fewer fluctuations and only marginal changes in the values, which suggests that there might not be large conformational changes upon ligand binding to ERRγ compared to ERRγapo, except for few local structural rearrangements [16]. To further corroborate the conformational stability of the apo and complex models, we measured the solvent-accessible surface area (SASA). The SASA values of ERRγapo ranged from 108 to 125 nm2, whereas those of ERRγag and ERRγiag ranged between 113–131 nm2 (Fig 2D). The smaller SASA values of apo ERRγ compared to the ligand-bound forms suggest that ligand binding induced small structural alterations in ERRγ, which might have caused an increase in the SASA of the ligand-bound forms of ERRγ.

thumbnail
Download:
Fig 2.

Trajectory analysis of ERRγapo, ERRγag and ERRγiag: (A) RMSD of ERRγapo, ERRγag and ERRγiag backbone atoms. (B) RMSD of agonist and inverse agonist (C) Radius of gyration (Rg) of ERRγapo, ERRγag and ERRγiag backbone atoms (D) Solvent accessible surface area (SASA) of ERRγapo, ERRγag and ERRγiag.

https://doi.org/10.1371/journal.pone.0283364.g002

Residue flexibility of both apo and ligand bound ERRγ

To understand the intrinsic dynamics of ERRγapo, ERRγag, and ERRγiag upon binding to the ligands, we performed a root mean square fluctuation (RMSF) analysis for backbone atoms. The results are presented in Fig 3, where ERRγapo exhibited large fluctuations only in the loop regions of the simulated LBD structure, particularly, the loops connecting helix 3 to helix 4 and helix 8 to helix 9. When we compared the RMSF of ERRγag to that of ERRγapo, we observed high residue fluctuations for ERRγag in the loop region of 243–248, suggesting that the region is crucial for agonist binding as reported in the crystal structure [16]. Other loop regions were largely stable for ERRγag indicating that the binding of the agonist GSK4716 to ERRγ caused rigidification of the loop regions, i.e., residues 285–300 and 375–395 compared to ERRγapo (Fig 3). On the other hand, when the RMSF of ERRγiag was compared to ERRγapo, a trend similar to that of ERRγag was observed in the loop regions, except in the region of the AF-2 helix (the loop connecting helix 12). The results strongly suggests that the displacement of the AF-2 helix conformation was necessary to block the functional activity of ERRγ upon inverse agonist binding [16]. It is worth mentioning that both ligand-bound forms of ERRγ exhibited an overall reduction in fluctuations, indicating that both ligands stabilize ERRγ structure upon binding (Fig 3).

thumbnail
Download:
Fig 3. Root mean square fluctuation (RMSF) of ERRγapo, ERRγag and ERRγiag: RMSF of the backbone atoms of ERRγapo, ERRγag and ERRγiag.

https://doi.org/10.1371/journal.pone.0283364.g003

Hydrogen bonds and binding free energy of ligand bound ERRγ

The intra H-bonds were calculated for the trajectories of the three systems (i.e., ≤ 3.5 Å distance and ≤ 30° angle). ERRγapo had a range of 158 to 205 H-bonds, with an average of 182 ± 7 bonds. ERRγag had an average of 175 ± 7 intra H-bonds, whereas ERRγiag had 182 ± 7 H-bonds (Fig 4A). The intra H-bonds of ERRγag and ERRγiag ranged between 143–202 and 155–209, respectively. We further examined the hydrogen bonding between the ligands and ERRγ structure. The number of inter H-bonds in ERRγag ranged between 0–4 bonds while those in ERRγiag ranged between 0–2 bonds (Fig 4B). The LBD of ERRγ consists of two pockets, pockets 1 and 2. From the crystal structures obtained from PDB, the agonist GSK4716 bound to pocket 2 with a volume of 390 Å3 while 4-OHT bound to pocket 1 with a volume of 280 Å3 [16]. Therefore, the binding of inverse agonist 4-OHT to pocket one has a lower H-bond but causes the AF-2 helix to displace. In addition, we assessed the H-bond occupancy during simulations, which suggested that I249 and Y326 were key residues with occupancy of ~35% and ~85% in ERRγag, whereas, for ERRγiag, we observed only E275 (acts as both acceptor and donor) with occupancy of ~6%. (Fig 4C).

thumbnail
Download:
Fig 4.

H-bond and free energy decomposition of ERRγapo, ERRγag and ERRγiag complex: (A) Number of intra H-bonds in ERRγapo, ERRγag and ERRγiag. (B) H-bonds formed between agonist/inverse agonist and ERRγ. (C) The percentage of H-bond occupancy between the key residues with the ligands. The two bars of E275 in ERRγiag represent two atoms that act as either an acceptor or donor. (D) Energy decomposition of the residues in ERRγag and ERRγiag.

https://doi.org/10.1371/journal.pone.0283364.g004

We further calculated the binding free energies of the agonist and the inverse agonist of ERRγ. The binding free energy of the inverse agonist was higher than that of the agonist bound to ERRγ. From the individual energy contributions, the van der Waals energy and non-polar energy contribute significantly to the higher binding energy of inverse agonists to ERRγ (Table 1), showing that hydrophobic interactions are crucial for inverse agonist binding compared to agonists. We performed energy decomposition studies to gain a better understanding of the residues contributing to the energies. Residues E245, K248, I249, D261, L268, L271, A272, E275, M306, E307, L309, I310, G312, V313, R316, E323, Y326, A327, D328, D329, D333, D344, K357, K430 and F435 has significant energy changes while for inverse agonist, energy changes were observed for D261, L265, L268, C269, D270, L271, A272, D273, E275, L276, W305, M306, E307, L309, I310, V313, R316, Y326, L342, D344, L345, N346, I349, K430, A431, V432, F435, I438, L440, K443 and L449 (Fig 4D).

thumbnail
Download:
Table 1. Binding free energy calculations of ERRγag and ERRγiag.

https://doi.org/10.1371/journal.pone.0283364.t001

To check the interactions between ligands and ERRγ, we extracted low-energy structures for both agonist and inverse agonist-bound forms. For ERRγag, the ligand formed hydrogen bonds with Y326 after simulation, whereas, for inverse agonist, no hydrogen bonds were observed. However, we observed more hydrophobic contacts for inverse agonist compared to agonist corroborating the binding free energy calculations (Table 2 and Fig 5).

thumbnail
Download:
Fig 5. Interaction analysis: The 2D interaction pattern between agonist/inverse agonist and ERRγ.

https://doi.org/10.1371/journal.pone.0283364.g005

thumbnail
Download:
Table 2. Comparison of interaction between the ligands and ERRγ structure.

https://doi.org/10.1371/journal.pone.0283364.t002

In order to analyze the mutational impact of the residues involved in hydrophobic interaction with ligands, we performed in-silico mutagenesis for a few hydrophobic residues common for both agonist and inverse agonist bound ERRγ after simulations (i.e., L268, M306, L309 and V313) (Table 2) and the residues important for AF2-helix displacement in ERRγiag (F435 and F450). We replaced the amino acids (hydrophobic to hydrophobic or aromatic to aromatic), e.g. L309 to V309, to check whether like-to-like amino acid mutations could maintain their role in ligand binding and protein stability. The protein design module from the molecular operating environment (Chemical Computing Group, Montreal, Canada) was used to calculate the relative binding affinity (dAffinity) and thermal stability (dStability) of the mutated ERRγ structures (Table 3) as reported in previous studies [47, 48]. A positive dAffinity and dStability value suggest lower binding affinity and reduced stability of the mutants, respectively. The observed dAffinity values for the mutations of both ERRγag and ERRγiag complex showed that like-to-like mutations may not largely affect the binding affinity or stability of the protein. However, the positive dAffinity and dStability values observed for M306, and L309 strongly reflect the importance of these residues for ligand binding and stability of ERRγ.

thumbnail
Download:
Table 3. Computation mutagenesis of hydrophobic residues involved in ligand binding.

dAffinity and dStability represent the relative binding affinity and relative thermostability of the mutated to wild-type ERRγ, respectively.

https://doi.org/10.1371/journal.pone.0283364.t003

Principal components (PC) and Free energy landscape (FEL) of both apo and ligand-bound ERRγ

To understand the distribution of large motions responsible for the functional dynamics of the proteins, we performed a principal component analysis. The eigenvalues for the projections were calculated, and the projections with the largest eigenvalues, PC1 and PC2, were plotted (Fig 6A). The spread of the distribution of points, where each point represents a conformation of the protein, will be greater if the conformational changes in the protein are large. As shown in Fig 6A, the spread of PC1 and PC2 of ERRγapo was larger than those of ERRγag and ERRγiag. This is possibly due to the absence of ligands, which allowed conformational freedom in ERRγapo. When ERRγag and ERRγiag were compared, ERRγag showed a slightly larger spread. The cumulative percentages of motion of the first 50 eigenvectors of ERRγapo, ERRγag, and ERRγiag were 69.3%, 65.9% and 61.2%, respectively (Fig 6B). This is also in line with the large fluctuations in the residues recorded in ERRγapo and ERRγag. The results suggested that the binding of ligands to ERRγ allowed ERRγag and ERRγiag to be highly stable in the trajectory analysis. We further extracted the extreme conformations for PC1 and PC2 to understand the regions that exhibited large conformational changes in all three systems. As expected, large conformational changes for apo ERRγ in the loop were seen between helix-8 and helix-9 similar to the RMSF results; however, no large structural variations were observed in the ligand-bound forms, strongly indicating that ligand binding may stabilize the whole structure compared to apo ERRγ (Fig 6C).

thumbnail
Download:
Fig 6.

Principal component analysis of ERRγapo, ERRγag and ERRγiag: (A) Principal component analysis of ERRγapo, ERRγag and ERRγiag projected by PC1 and PC2 onto the phase space. (B) The first 50 eigenvectors, corresponding to the best eigenvalue principal components, are represented in the plot along with their associated cumulative fluctuations. (C) The conformational changes, minimum (red) to maximum (blue) observed for ERRγapo, ERRγag, and ERRγiag using PC1 and PC2 with 30 conformations extracted using the extreme module. The black circle represents the large conformational changes in the loop between helix-8 and helix-9 in ERRγapo.

https://doi.org/10.1371/journal.pone.0283364.g006

Furthermore, we determined the global minimum energy conformation using free-energy landscape (FEL) analysis. The FEL map was derived by plotting PC1 and PC2 against Gibbs free energy. An FEL analysis of ERRγapo is shown in Fig 7. The plot shows a large distribution of conformations in the apo form with three minimal-energy clusters. On the other hand, when ERRγag and ERRγiag were analyzed, the distribution of conformations was smaller than those of the apo form. ERRγag showed two minimal energy clusters, whereas ERRγiag exhibited three strong clusters of minimum energy, with most of the structural conformations concentrated around the minimal cluster (Fig 7). This might be the reason for the reduced fluctuation of residues and the smaller distribution in the principal component analysis of ERRγiag. In addition, when we analyzed the conformations of the F435 and F450 residues, we observed a large difference in conformations between ERRγag and ERRγiag. In Fig 7, F435 and F450 were aligned towards each other in ERRγapo and ERRγag, whereas in ERRγiag, the inverse agonist positioned itself near F435 and deflected the F450 residue by steric hindrance. This steric hindrance caused the AF-2 helix to be displaced in the ERRγiag compared to the ERRγapo form.

thumbnail
Download:
Fig 7. Free energy landscape of ERRγapo, ERRγag and ERRγiag: 2D projection of free energy landscapes of ERRγapo, ERRγag and ERRγiag as a function of PC1 and PC2.

The lowest point of the free energy reservoir shown in blue and violet represents the energy minima. The representative structures from each cluster of FEL were extracted and superimposed to show the variations in the AF-2 helix region. Residues F435, F450 and ligands are represented in the stick model.

https://doi.org/10.1371/journal.pone.0283364.g007

Betweenness centrality (CB)

To investigate the key residues required for the protein function, we performed network centrality calculations [49]. Centrality calculations can be successfully applied to identify specific residues present in the active site or co-factor binding site, along with allosteric networks in proteins [46]. In this study, we employed low-energy structures to construct residue interaction networks and calculated the betweenness centrality (CB). The residues (nodes) with high CB values may have functional relationships and exhibit allosteric and protein-protein/ligand interactions [50].

CB value plots for residues in ERRγapo, ERRγag, and ERRγiag are shown in S1 Fig.

Residues with CB ≥ 0.5, constituting ~10% of the CB distribution, were selected for ERRγapo, ERRγag, and ERRγiag structures. The residues are shown in Table 4, where ERRγapo had 16 residues, whereas ERRγag and ERRγiag had 25 and 13 residues, respectively. Among them, few residues involved in hydrophobic interactions with the ligands (M306,V313 and V325 for agonist and L276, L309, V313 and L345 for inverse agonist). Moreover, V313 and L345 were previously reported to be involving in ligand interactions in ERRγ [51, 52].

thumbnail
Download:
Table 4. Critical residues in ERRγapo, ERRγag, and ERRγiag with betweenness centrality values CB ≥ 0.5.

Residues involved in hydrophobic interactions are indicated in bold.

https://doi.org/10.1371/journal.pone.0283364.t004

Further, we computed the difference in the CB values (CBd) of ERRγapo with ERRγag and ERRγapo with ERRγiag, and these residues were deemed to be critical if the CBd were larger (absolute value). The plot (Fig 8A) shows the CBd values of the residues in ERRγag and ERRγiag with |CBapo—CBag/iag| ≥ 0.3. From this result, we observed 18 residues for agonist and 15 residues for inverse agonist, respectively, which are shown in Table 5. Among these residues, M306 (for agonist), L276 and L309 (for inverse agonist) were involved in hydrophobic interactions with the ligands and few other residues located in the vicinity of binding pocket (Fig 8B). These results suggest that these residues might play an important role in ligand binding as well as protein function. Though these residues are suggested to be crucial, further experimental investigation would be required to unravel the functional and mechanistic importance.

thumbnail
Download:
Fig 8.

Network centrality analysis of ERRγapo, ERRγag and ERRγiag systems: (A) Plot showing the difference in CB values between ERRγapo and ERRγag / ERRγiag. The threshold for identification of the residues |CBapo—CBag/iag| was maintained greater than or equal to 0.3 (black line). (B) Critical residues identified by |CBapo—CBag/iag| ≥ 0.3 are represented in hot pink using ball and stick model. Agonist GSK4716 and inverse agonist 4-OHT are represented in the stick model.

https://doi.org/10.1371/journal.pone.0283364.g008

thumbnail
Download:
Table 5. Critical residues with betweenness centrality values |CBapo—CBag/iag| ≥ 0.3.

Bold residues represent residues common in both ERRγag and ERRγiag.

https://doi.org/10.1371/journal.pone.0283364.t005

Discussion

The current study was conducted to identify the dynamics of ERRγapo, ERRγag, and ERRγiag, and the key residues responsible for the interactions between ERRγ and its agonist or inverse agonist. The structures retrieved from the PDB database served as the initial points of study and were simulated for 500 ns each. The first phase of this study was to determine whether the retrieved structures displayed reduced fluctuations in the simulation trajectory during the simulation period. RMSD studies of the protein structure and ligands showed that all three models exhibited conformational stability by reducing fluctuations throughout the simulation period. The Rg and SASA values also corroborated the RMSD results that the apo and bound forms of ERRγ were compact-folded during the simulation period. ERRγapo remained in the canonical three-layered α-helical sandwich throughout the simulation as reported earlier for other nuclear receptors [16]. The ligands 4-OHT and GSK4716 were anchored in the active site of the ERRγ structure throughout the simulation period. Previous studies have shown that the 4-OHT inverse agonist binds to pocket 1 of ERRγ, while the GSK4716 agonist binds to the larger pocket 2 of ERRγ [16]. Similar to the previous report, we observed that the ligands were positioned comfortably in the respective pockets. It is worthwhile to note that the pockets are adjacent to each other as shown in Fig 1.

Further, decreased residue flexibility among the ligand bound systems compared to apo ERRγ indicated that ligand binding stabilizes the protein by residue interactions (Fig 3). When the low energy structures were retrieved and analyzed for interactions, GSK4716 interacted with L268, L309, V313, R316 and Y326 of ERRγ. These interactions were also recorded earlier in a study involving bisphenol A with ERRγ, and they were found to be highly energetically favorable interactions [51]. On the other hand, in ERRγiag, the hydrophobic interactions (with L268, C269, L309, V313, Y326, and L345) were similar to the interactions recorded for the agonists (Figs 4 and 5) [51]. The H-bonds observed for both agonist and inverse agonist, also suggested that, collectively, the binding of GSK4716 to ERRγ rigidified the entire protein and formed H-bonds with key residues of the protein (Fig 4). This is similar to a previous study where GSK4716 was identified as an agonist that was 50 times selective towards ERRγ [17]. In contrast, the inverse agonist 4-OHT formed fewer H-bonds (4-OHT forms H-bond with E275, however, less occupancy during simulations (Fig 4C)) with the residues in pocket 1. Moreover, the fewer hydrogen bonds, observed for the inverse agonist, suggested that hydrophobic interactions were crucial for the activity of the inverse agonist when compared to the agonist.

The PCA further showed the large dominant motions of apo ERRγ compared to ligand bound forms particularly on the loop regions (loop between helix 8 and 9) (Fig 6), which in turn reflected on the FEL with large metastable states for apo ERRγ. From RMSF studies, the larger fluctuations observed for the ERRγiag, compared to ERRγag in the region of the AF-2 helix showed that the inverse agonist binding displaces the AF-2 helix conformation [16]. The reason is that the binding of 4-OHT, which is a large molecule in a smaller ligand-binding pocket (280 Å3), causes displacement of the AF-2 helix. When snapshots from the minimum energy clusters of FEL were observed for ERRγag, both F435 and F450 aligned comfortably with few conformational changes. In contrast, the inverse agonist was positioned near the F435 residue, thereby deflecting the F450 residue. A previous study has shown that F435 is a crucial residue that prevents the common steroidal estrogens from binding to ERRγ and F450 displacement affects the volume available for ligand binding [16]. This steric hindrance between F435 and F450 caused the AF2-helix to be displaced, thereby, forcing it to adopt a conformation different from ERRγago (Fig 7). The large residue fluctuations observed (Fig 3) in the AF-2 helix and end of helix 10 of ERRγiag were due to the presence of pocket 1 in the vicinity. As discussed earlier, the AF-2 helix is important for co-activator binding, and binding of 4-OHT changes the conformation of the AF-2 helix. This conformational change blocks the binding of any NR-box co-activator peptides and represses the activity of ERRγ [16, 53, 54]. The observed betweenness centrality (CB) of the residues and the difference in the betweenness centrality values (CBd) further identified residues that were critical for intra-residue signal transduction and functioning of ERRγ [16, 51]. Few of the residues observed in CB and CBd (Tables 4 and 5), were involved in hydrophobic interactions as well as in the vicinity of ligand binding pocket (Fig 8). Hence, we assume these residues might play a crucial role in ligand binding as well as in the functional activity of the protein.

Conclusions

This study provides insight into the dynamic behavior of apo and ligand bound LBD of ERRγ and shed light into the residues involved in the binding of ligands to ERRγ. The results suggest the crucial residues for the interaction and functioning of ERRγ when bound to ligands. In addition, we observed that the conformation of the AF-2 helix upon binding to 4-OHT was maintained during simulations, and the residues F435 and F450 were involved in the conformational dynamics. Apart from agonist/inverse agonist studied here, several other compounds suggested in other studies may be capable of exhibiting similar effects. The major limitation of the study is the use of the only ligand bound LBD domain complex of the ERRγ. We assume that long-range molecular dynamic simulations on the ligand and coactivator bound LBD complex of ERRγ might provide a better understanding of conformational changes or allostery of ERRγ. Concluding from the results, we believe that the pharmacophore screening of ligands based on the results of this study may assist in identifying better drugs against ERRγ related diseases. Understanding the ERRγ dynamics as well as the binding of ligands to active pockets may also assist researchers design better therapeutic compounds for ERRγ.

Supporting information

S1 Fig. CB values of ERRγapo, ERRγag, and ERRγiag residues.

https://doi.org/10.1371/journal.pone.0283364.s001

(DOCX)

Acknowledgments

We would like to acknowledge Dr. Masaud Shah from Ajou University, Suwon, Korea for his kind help to perform computational mutagenesis using molecular operating environment (MOE).

References

  1. 1. Giguère V. Orphan nuclear receptors: from gene to function. Endocrine reviews. 1999;20(5):689–725. pmid:10529899
  2. 2. Giguère V. Transcriptional control of energy homeostasis by the estrogen-related receptors. Endocrine reviews. 2008;29(6):677–96. pmid:18664618
  3. 3. Hong H, Yang L, Stallcup MR. Hormone-independent transcriptional activation and coactivator binding by novel orphan nuclear receptor ERR3. Journal of Biological Chemistry. 1999;274(32):22618–26. pmid:10428842
  4. 4. Lu D, Kiriyama Y, Lee KY, Giguere V. Transcriptional regulation of the estrogen-inducible pS2 breast cancer marker gene by the ERR family of orphan nuclear receptors. Cancer research. 2001;61(18):6755–61. pmid:11559547
  5. 5. Vanacker J-M, Delmarre C, Guo X, Laudet V. Activation of the Osteopontin Promoter by the Orphan Nuclear Receptor Estrogen Receptor Related. Cell Growth and Differentiation. 1998;9:1007–14.
  6. 6. Huss JM, Garbacz WG, Xie W. Constitutive activities of estrogen-related receptors: Transcriptional regulation of metabolism by the ERR pathways in health and disease. Biochimica et Biophysica Acta (BBA)-Molecular Basis of Disease. 2015;1852(9):1912–27. pmid:26115970
  7. 7. Divekar SD, Tiek DM, Fernandez A, Riggins RB. Estrogen-related receptor β (ERRβ)–renaissance receptor or receptor renaissance? Nuclear receptor signaling. 2016;14(1):nrs. 14002.
  8. 8. Eudy JD, Yao S, Weston MD, Ma-Edmonds M, Talmadge CB, Cheng JJ, et al. Isolation of a gene encoding a novel member of the nuclear receptor superfamily from the critical region of Usher syndrome type IIa at 1q41. Genomics. 1998;50(3):382–4. pmid:9676434
  9. 9. Alaynick WA, Kondo RP, Xie W, He W, Dufour CR, Downes M, et al. ERRγ directs and maintains the transition to oxidative metabolism in the postnatal heart. Cell metabolism. 2007;6(1):13–24.
  10. 10. Radhakrishnan K, Kim YH, Jung YS, Kim DK, Na SY, Lim D, et al. Orphan nuclear receptor ERR-gamma regulates hepatic FGF23 production in acute kidney injury. Proc Natl Acad Sci U S A. 2021;118(16). pmid:33853949
  11. 11. Bhardwaj VK, Purohit R. A lesson for the maestro of the replication fork: Targeting the protein‐binding interface of proliferating cell nuclear antigen for anticancer therapy. Journal of Cellular Biochemistry. 2022. pmid:35486518
  12. 12. Misra J, Kim D-K, Choi H-S. ERRγ: a junior orphan with a senior role in metabolism. Trends in Endocrinology & Metabolism. 2017;28(4):261–72.
  13. 13. Radhakrishnan K, Kim Y-H, Jung YS, Kim J, Kim D-K, Cho SJ, et al. Orphan Nuclear Receptor ERRγ Is a Novel Transcriptional Regulator of IL-6 Mediated Hepatic BMP6 Gene Expression in Mice. International Journal of Molecular Sciences. 2020;21(19). pmid:32998264
  14. 14. Song G, Wang L. Transcriptional mechanism for the paired miR-433 and miR-127 genes by nuclear receptors SHP and ERRγ. Nucleic acids research. 2008;36(18):5727–35.
  15. 15. Zhang Y, Wang L. Nuclear receptor SHP inhibition of Dnmt1 expression via ERRγ. FEBS letters. 2011;585(9):1269–75.
  16. 16. Wang L, Zuercher WJ, Consler TG, Lambert MH, Miller AB, Orband-Miller LA, et al. X-ray crystal structures of the estrogen-related receptor-γ ligand binding domain in three functional states reveal the molecular basis of small molecule regulation. Journal of Biological Chemistry. 2006;281(49):37773–81.
  17. 17. Zuercher WJ, Gaillard S, Orband-Miller LA, Chao EY, Shearer BG, Jones DG, et al. Identification and structure− activity relationship of phenolic acyl hydrazones as selective agonists for the estrogen-related orphan nuclear receptors ERRβ and ERRγ. Journal of medicinal chemistry. 2005;48(9):3107–9.
  18. 18. Greschik H, Wurtz J-M, Sanglier S, Bourguet W, van Dorsselaer A, Moras D, et al. Structural and functional evidence for ligand-independent transcriptional activation by the estrogen-related receptor 3. Molecular cell. 2002;9(2):303–13. pmid:11864604
  19. 19. Brzozowski AM, Pike AC, Dauter Z, Hubbard RE, Bonn T, Engström O, et al. Molecular basis of agonism and antagonism in the oestrogen receptor. Nature. 1997;389(6652):753–8. pmid:9338790
  20. 20. Horard B, Vanacker J. Estrogen receptor-related receptors: orphan receptors desperately seeking a ligand. Journal of molecular endocrinology. 2003;31(3):349–57. pmid:14664699
  21. 21. Hentschke M, Süsens U, Borgmeyer U. Domains of ERRγ that mediate homodimerization and interaction with factors stimulating DNA binding. European journal of biochemistry. 2002;269(16):4086–97.
  22. 22. Huss JM, Kopp RP, Kelly DP. Peroxisome proliferator-activated receptor coactivator-1α (PGC-1α) coactivates the cardiac-enriched nuclear receptors estrogen-related receptor-α and-γ: identification of novel leucine-rich interaction motif within PGC-1α. Journal of Biological Chemistry. 2002;277(43):40265–74.
  23. 23. Kamei Y, Ohizumi H, Fujitani Y, Nemoto T, Tanaka T, Takahashi N, et al. PPARγ coactivator 1β/ERR ligand 1 is an ERR protein ligand, whose expression induces a high-energy expenditure and antagonizes obesity. Proceedings of the National Academy of Sciences. 2003;100(21):12378–83.
  24. 24. Hentschke M, Süsens U, Borgmeyer U. PGC-1 and PERC, coactivators of the estrogen receptor-related receptor γ. Biochemical and biophysical research communications. 2002;299(5):872–9.
  25. 25. Sanyal S, Kim J-Y, Kim H-J, Takeda J, Lee Y-K, Moore DD, et al. Differential regulation of the orphan nuclear receptorsmall heterodimer partner (SHP) gene promoter by orphan nuclear receptor ERR isoforms. Journal of Biological Chemistry. 2002;277(3):1739–48. pmid:11705994
  26. 26. Xie Y-B, Park J-H, Kim D-K, Hwang JH, Oh S, Park SB, et al. Transcriptional corepressor SMILE recruits SIRT1 to inhibit nuclear receptor estrogen receptor-related receptor γ transactivation. Journal of Biological Chemistry. 2009;284(42):28762–74.
  27. 27. Singh R, Bhardwaj VK, Purohit R. Computational targeting of allosteric site of MEK1 by quinoline‐based molecules. Cell Biochemistry and Function. 2022;40(5):481–90. pmid:35604288
  28. 28. Kamaraj B, Al-Subaie AM, Ahmad F, Surapaneni KM, Alsamman K. Effect of novel leukemia mutations (K75E & E222K) on interferon regulatory factor 1 and its interaction with DNA: insights from molecular dynamics simulations and docking studies. Journal of Biomolecular Structure and Dynamics. 2021;39(14):5235–47.
  29. 29. Al-Subaie AM, Kamaraj B. The structural effect of FLT3 mutations at 835th position and their interaction with acute myeloid leukemia inhibitors: in silico approach. International journal of molecular sciences. 2021;22(14):7602. pmid:34299222
  30. 30. Singh R, Bhardwaj VK, Das P, Purohit R. Identification of 11β-HSD1 inhibitors through enhanced sampling methods. Chemical Communications. 2022;58(32):5005–8.
  31. 31. Bhardwaj VK, Oakley A, Purohit R. Mechanistic behavior and subtle key events during DNA clamp opening and closing in T4 bacteriophage. International Journal of Biological Macromolecules. 2022;208:11–9. pmid:35276295
  32. 32. Kumar S, Bhardwaj VK, Singh R, Das P, Purohit R. Identification of acridinedione scaffolds as potential inhibitor of DENV‐2 C protein: An in silico strategy to combat dengue. Journal of Cellular Biochemistry. 2022.
  33. 33. Berendsen HJ, van der Spoel D, van Drunen R. GROMACS: A message-passing parallel molecular dynamics implementation. Computer physics communications. 1995;91(1–3):43–56.
  34. 34. Abraham MJ, Murtola T, Schulz R, Páll S, Smith JC, Hess B, et al. GROMACS: High performance molecular simulations through multi-level parallelism from laptops to supercomputers. SoftwareX. 2015;1:19–25.
  35. 35. Chakrabarty B, Parekh N. NAPS: Network Analysis of Protein Structures. Nucleic Acids Res. 2016;44(W1):W375–82. Epub 2016/05/07. pmid:27151201
  36. 36. Chakrabarty B, Naganathan V, Garg K, Agarwal Y, Parekh N. NAPS update: network analysis of molecular dynamics data and protein–nucleic acid complexes. Nucleic Acids Research. 2019;47(W1):W462–W70. pmid:31106363
  37. 37. Schrödinger L. The PyMOL molecular graphics system, version 1.8. 2015.
    • 38. Lindorff-Larsen K, Piana S, Palmo K, Maragakis P, Klepeis JL, Dror RO, et al. Improved side-chain torsion potentials for the Amber ff99SB protein force field. Proteins. 2010;78(8):1950–8. pmid:20408171
    • 39. Sousa da Silva AW, Vranken WF. ACPYPE—AnteChamber PYthon Parser interfacE. BMC research notes. 2012;5:367. Epub 2012/07/25. pmid:22824207
    • 40. Bussi G, Donadio D, Parrinello M. Canonical sampling through velocity rescaling. The Journal of Chemical Physics. 2007;126(1):014101. pmid:17212484
    • 41. Sasidharan S, Gosu V, Shin D, Nath S, Tripathi T, Saudagar P. Therapeutic p28 peptide targets essential H1N1 influenza virus proteins: insights from docking and molecular dynamics simulations. Molecular Diversity. 2021;25(3):1929–43. pmid:33575983
    • 42. Kumari R, Kumar R, Lynn A. g_mmpbsa—A GROMACS Tool for High-Throughput MM-PBSA Calculations. Journal of Chemical Information and Modeling. 2014;54(7):1951–62. pmid:24850022
    • 43. Raj S, Sasidharan S, Dubey VK, Saudagar P. Identification of lead molecules against potential drug target protein MAPK4 from L. donovani: An in-silico approach using docking, molecular dynamics and binding free energy calculation. PloS one. 2019;14(8):e0221331. pmid:31425543
    • 44. Sasidharan S, Saudagar P. Biochemical and structural characterization of tyrosine aminotransferase suggests broad substrate specificity and a two‐state folding mechanism in Leishmania donovani. FEBS Open bio. 2019;9(10):1769–83. pmid:31393078
    • 45. Sasidharan S, Saudagar P. Flavones reversibly inhibit Leishmania donovani tyrosine aminotransferase by binding to the catalytic pocket: An integrated in silico-in vitro approach. International Journal of Biological Macromolecules. 2020;164:2987–3004. pmid:32798546
    • 46. Gosu V, Sasidharan S, Saudagar P, Lee H-K, Shin D. Computational Insights into the Structural Dynamics of MDA5 Variants Associated with Aicardi–Goutières Syndrome and Singleton–Merten Syndrome. Biomolecules. 2021;11(8):1251.
    • 47. Batool M, Shah M, Patra MC, Yesudhas D, Choi S. Structural insights into the Middle East respiratory syndrome coronavirus 4a protein and its dsRNA binding mechanism. Scientific Reports. 2017;7(1):11362. pmid:28900197
    • 48. Shah M, Anwar MA, Park S, Jafri SS, Choi S. In silico mechanistic analysis of IRF3 inactivation and high-risk HPV E6 species-dependent drug response. Scientific Reports. 2015;5(1):13446. pmid:26289783
    • 49. Jeong H, Mason SP, Barabási A-L, Oltvai ZN. Lethality and centrality in protein networks. Nature. 2001;411(6833):41–2. pmid:11333967
    • 50. Gosu V, Son S, Shin D, Song K-D. Insights into the dynamic nature of the dsRNA-bound TLR3 complex. Scientific reports. 2019;9(1):1–14.
    • 51. Li L, Wang Q, Zhang Y, Niu Y, Yao X, Liu H. The molecular mechanism of bisphenol A (BPA) as an endocrine disruptor by interacting with nuclear receptors: insights from molecular dynamics (MD) simulations. PloS one. 2015;10(3):e0120330. pmid:25799048
    • 52. Usha T, Middha SK, Shanmugarajan D, Babu D, Goyal AK, Yusufoglu HS, et al. Gas chromatography-mass spectrometry metabolic profiling, molecular simulation and dynamics of diverse phytochemicals of Punica granatum L. leaves against estrogen receptor. Frontiers in Bioscience-Landmark. 2021;26(9):423–41. pmid:34590457
    • 53. Yao B, Zhang S, Wei Y, Tian S, Lu Z, Jin L, et al. Structural Insights into the Specificity of Ligand Binding and Coactivator Assembly by Estrogen-Related Receptor β. Journal of Molecular Biology. 2020;432(19):5460–72. https://doi.org/10.1016/j.jmb.2020.08.007.
    • 54. Shang J, Mosure SA, Zheng J, Brust R, Bass J, Nichols A, et al. A molecular switch regulating transcriptional repression and activation of PPARγ. Nature Communications. 2020;11(1):956. pmid:32075969