Understanding the molecular mechanism of endothelin ET_A receptor selecting isopeptides endothelin-1 and -3

Abstract

Endothelin-1 and -3 (ET1 and ET3) are potent vasoconstricting isopeptides that are involved in the pathophysiology of various diseases, such as cardiovascular and renal diseases. The two ETs exert their effects by binding to two G-protein-coupled receptors (GPCRs), the endothelin ET_A receptor (ET_AR) and the endothelin ET_B receptor (ET_BR). ET_AR and ET_BR reveal different preferences in the recognition of the two ETs: The binding affinity of ET_AR with ET1 is much higher than that with ET3, while the binding affinities of ET_BR with the two ETs are similar. Recently, the structures of ET_BR with ET1 and ET3 were determined. These crystal structures provide detailed descriptions of how ET_BR interacts with ET1 and ET3. However, the knowledge of how ET_AR recognizes the two isopeptides is still lacking. Based on the structure of ET_BR in complex with ET1, the structures of ET_AR with ET1 and ET3 were modeled. Then, molecular dynamics simulations were performed to study how ET_AR discriminates the two ETs. Simulation results demonstrate that ET1 has greater binding free energy with ET_AR than ET3, which is in agreement with the binding affinity data of ET_AR. By interaction energy analysis, the key residues of ET_AR that discriminate between ET1 and ET3 are identified. Structural and dynamical analyses indicate that, compared with ET3, ET1 is more stable in binding with ET_AR. Furthermore, the orientation change of W319 in the conserved CWxP motif that plays an important role in the signaling function of GPCRs was observed. Through the orientation change of this residue, the difference in the orthosteric pocket volume resulting from the binding of ET1 and ET3 on the extracellular side of ET_AR leads to a conformational difference of TM6 on the intracellular side of ET_AR. This study elucidates the molecular mechanism of how ET_AR selects the isopeptides ET1 and ET3.

Graphical abstract

Significance

To date, how ET_AR discriminates ET1 from ET3 is still unknown, since no crystal, cryo-EM, or NMR structure of ET_AR is available. In this study, the structures of ET_AR in complex with ET1 and ET3 were set up by homology modeling. Through molecular simulations, we found that ET1 shows higher structural and dynamical stabilities compared with ET3, and has a preferable binding free energy with ET_AR. The key residues that play important roles in recognizing ET1 and ET3 have also been identified through interaction energy analysis. Our results provide insights into the molecular mechanism of how ET_AR discriminates ET1 from ET3, which may help to design novel endothelin receptor antagonists to treat the cardiovascular and renal diseases caused by endothelins.

Introduction

Endothelins (ETs) are a family of potent vasoconstricting isopeptides that have been implicated in the pathophysiology of cardiovascular diseases, pulmonary diseases, cancers, and renal diseases (1,2). To date, three isoforms of ETs have been found. All of them contain 21 amino acids in their mature and active forms (3). Endothelin-1 (ET1) is the most prevalent isoform, which is widely expressed in various cells, such as endothelial cells, vascular smooth muscle cells, and epithelial cells (4,5). ET1 is currently identified as the most potent endogenous vasoconstrictor. It can produce extremely powerful contraction of human arteries and veins, and the effect of the contraction is long-lasting and difficult to wash out (4). While ET2 is mainly expressed in the gastrointestinal tract (6), ET3 is detected in plasma, the heart, brain, and other tissues (4,7). Besides its vasoconstrictor role, ET3 can mediate the release of vasodilator molecules, such as nitric oxide, and induce a relatively more potent vasodilator action than the other two ETs (8).

ETs exert their effects by binding to two G-protein-coupled receptors (GPCRs), the endothelin ET_A receptor (ET_AR) and the endothelin ET_B receptor (ET_BR) (9). ET_AR is found in vascular smooth muscle cells and many human tissues including the heart, lung, and kidney, whereas ET_BR is located in vascular endothelial cells and has a high expression level in the brain (4). The two receptors are involved in a vast array of physiological processes, such as vasoconstriction/vasodilatation, bronchoconstriction, and cell proliferation (10). ET_AR and ET_BR reveal different preferences in the recognition of the three ET isopeptides: ET_AR binds ET1 and ET2 with similar affinities, but ET3 with 100-fold lower affinity (ET1 ≈ ET2 >> ET3) (11). In contrast, ET_BR recognizes all the three isoforms with similar affinities (ET1 ≈ ET2 ≈ ET3) (12).

Recently, the structures of the human ET_BR in complex with ET1 and ET3 have been determined through x-ray crystallography (13,14). The structural data show that the ET_BR adopts a typical GPCR architecture that contains seven transmembrane helices (TM1-7) and an intracellular helix (helix 8) in the C-terminal region. ET1 and ET3 fit in the orthosteric pocket of ET_BR, which is covered by a conserved β-hairpin motif in the extracellular loop 2 (ECL2). Both ETs adopt a bicyclic structure with two intrachain disulfide bond pairs (C1-C15 and C3-C11), and the two ETs can be divided into three regions (Fig. 1 A): the N-terminal region (residues 1–7), the α-helical region (residues 8–17), and the C-terminal region (residues 18–21), which deeply penetrates into the pocket of the receptor. In addition to determining the crystal structure of ET_BR with ET3, Shihoya et al. also investigated the binding affinities of ET1 and ET3 to ET_AR and ET_BR through TGF-α shedding and β-arrestin recruitment assays (14). In both assays, the binding affinities of ET3 to ET_BR are similar to those of ET1 to ET_BR (the EC₅₀ values for ET1 and ET3 are 0.11 and 0.13 nM, respectively, in the TGF-α shedding assay; and the EC₅₀ values for ET1 and ET3 are 2.7 and 3.3 nM, respectively, in the β-arrestin recruitment assay), while the binding affinities of ET1 to ET_AR was about fivefold lower than those of ET3 to ET_AR (the EC₅₀ values for ET1 and ET3 are 0.13 and 0.65 nM, respectively, in TGF-α shedding assay; and the EC₅₀ values for ET1 and ET3 are 2.3 and 10.8 nM, respectively, in the β-arrestin recruitment assay). These data further confirm the notion that ET_AR has a higher selectivity of ET1 than ET3.

Figure 1.

Initial simulation structure and structural deviations. (A) Endothelin-1 and -3 (ET1 and ET3) are modeled in the orthosteric pocket of endothelin type A receptor (ET_AR). The simulation systems of ET_AR with ET1 and ET3 are denoted as ET_AR-ET1 (I) and ET_AR-ET3 (II), respectively. The three regions of ET1/3, i.e., the N-terminal, α-helical, and C-terminal regions, are shown in cyan, orange, and purple, respectively. The black dashed lines indicate the surfaces of membrane. (B) Root mean-square deviations (RMSDs) for Cα atoms of ET_AR (black lines) and for that of ET1 or ET3 (red lines) as a function of simulation time. Three 1,000 ns molecular dynamics simulations were performed for each system. (C) Comparison of the binding of ET1 and ET3 in the orthosteric pocket of ET_AR. Representative structures of ET_AR-ET1 and ET_AR-ET3 were used for comparison. The residues different between ET1 and ET3 are labeled in cyan and blue, respectively. To see this figure in color, go online.

The crystal structures provide detailed descriptions of how ET_BR interacts with ET1 and ET3. However, the knowledge of how ET_AR recognizes the two isopeptides is still lacking since the structure of ET_AR is not available to date. To this end, the structural models of ET_AR in complex with ET1 and ET3 were built based on the structure of ET_BR with ET1. Then, three 1,000 ns molecular dynamics simulations were performed for each model. The simulation results demonstrate that ET_AR has a greater binding free energy with ET1 than ET3, which is consistent with the binding affinity data of ET_AR. By interaction energy analysis, the key residues of ET_AR that discriminate between ET1 and ET3 are identified. Structural and dynamical analyses indicate that, compared with ET3, ET1 is more stable in binding with ET_AR. Furthermore, the volume of the orthosteric pocket is increased in the binding of ET3 compared with that in the binding of ET1. This study elucidates the molecular mechanism of how ET_AR discriminates the isopeptides ET1 and ET3, which may help in designing novel ET receptor antagonists to treat cardiovascular and renal diseases.

Materials and methods

Homology modeling

The structural models of ET_AR in complex with ET1 and ET3 were built using the software MODELLER 9.13 (15). The crystal structure of ET_BR in complex with ET1 (PDB: 5GLH) (13) was used as a template for the modeling. Before modeling, the structure of bacteriophage T4 lysozyme (used to facilitate crystallization) was removed from the template. The sequence of ET_AR was aligned with that of ET_BR based on the work of Shihoya et al. (13). Totally, 100 homology models were generated by MODELLER, and the modeled structure with the lowest DOPE energy was chosen as the best model. The disulfide bonds (C69-C340 and C159-C239 in ET_AR, and C1-C15 and C3-C11 in ET1) were retained in the modeled structures. The residues of ET_AR corresponding to those missing in the template of ET_BR were added by MODELLER. The protonation states of the titratable residues were determined using the H++ program (16). Histidine was protonated at the epsilon nitrogen, and all the other titratable residues were assigned the protonation states corresponding to their default values at neutral pH. The modeled structure of ET_AR includes amino acids 67–384, and contains the secondary structures of TM1-7, ICL1-3, ECL1-3, and helix 8 (Figs. 1 A and S1). To validate our model, the quality of the modeled structures was evaluated using the QMEANDisCo global score (17). The score value is 0.60 ± 0.05, indicating that our models are reliable.

The amino acids of ET3 that are different from ET1 (Fig. 1 A) were substituted using the mutagenesis function of CHARMMI-GUI (18). The two simulation systems of ET_AR in complex with ET1 and ET3 are denoted ET_AR-ET1 and ET_AR-ET3, respectively. The membrane environment was modeled by a palmitoyl-oleoyl-phosphatidyl-choline bilayer composed of 127 lipid molecules using the CHARMMI-GUI membrane builder (18). Then TIP3P water molecules were added to the two sides of the lipid bilayer (Fig. S1), which resulted in a simulation box of 74 × 74 × 130 Å along the x, y, and z directions, respectively. The parameters of the FF99SB force field (19) were assigned to the protein, ions, and water molecules, while the FFlipid14 force field (20) was used for the palmitoyl-oleoyl-phosphatidyl-choline lipids.

Molecular dynamics simulations

Three independent simulations starting from different initial velocities were performed for ET_AR-ET1 and ET_AR-ET3. All the molecular dynamics simulations were performed using the AMBER16 simulation package (21). The simulation protocol is similar to our previous study (22). In brief, each system was first minimized using the steepest descent and conjugate gradient methods. After that, the whole system was gradually heated from 0 to 300 K under NVT (constant volume and temperature) conditions with the restraints of ET_AR and two ETs. Then the harmonic restraints were gradually switched off during the simulation in a stepwise manner. Finally, 1,000 ns production simulations were performed under NTP (constant temperature and pressure) conditions. The time step of the simulations was 2 fs and the trajectories were collected every 1 ps.

Data analyses

Simulation data analyses were mainly performed using the CPPTRAJ program of AMBER16 (21). The equilibration of the simulations was assessed by calculating the root mean-square deviations (RMSDs) for the Cα atoms of ET_AR and the two ETs. The α-helix secondary structures of ET_AR and the two ETs were evaluated using the DSSP method (23). The α-helix occupancy for each residue was calculated by the percentage of time that the residue existed in the α-helix during the equilibrated simulations. The dynamical flexibilities of ET_AR and the two ETs were assessed by computing root mean-square fluctuation (RMSF) on a residue-by-residue basis and then averaged over the equilibrated simulations. The volumes for the orthosteric pocket of ET_AR were calculated by POVME 2.0 (24).

The binding free energies between ET_AR and the two ETs were calculated using the molecular mechanics/generalized Born surface area approach (25). The calculation procedure is similar to our previous study (22). In brief, the binding free energy (ΔG_binding) was computed by summing the molecular mechanics energy including the electrostatic energy (ΔE_elec) and the van der Waals energy (ΔE_vdw), the solvation free energy including a polar part (ΔG_pol), and a nonpolar part (ΔG_nonpol), and the entropy contribution to the binding (-TΔS). The entropy calculations were performed using normal mode analysis in AMBER16. In the calculations, the distance-dependent dielectric constant was set to 4.0, and the energy gradient of minimization was 0.001 with 10,000 minimization cycles per frame. For each simulation, 1,000 frames were evenly extracted from the trajectory to obtain the time-dependent binding free energy. To compare the binding free energies between ET_AR-ET1 and ET_AR-ET3, the energy data from the last 400 ns were used for analysis. The data from the three independent simulations were averaged to get the mean values and the standard errors. To determine which residues of ET_AR and the two ETs play important roles in the binding, the interaction energy was calculated by decomposing the binding free energy based on each residue of the receptor and the two ETs.

For comparison, the mean values and standard errors for the binding free energy, pocket volume, and distance between TM1 and TM6 on the intracellular side were calculated from the three independent simulations. Based on the trajectory equilibration analysis, only the last 400 ns simulation data were used to compare the average data between ET_AR-ET1 and ET_AR-ET3. Significant differences for the studied variables were determined using Student's t-test (26) with 95% confidence. To compare the binding modes of ET1 and ET3 in ET_AR, the representative structures for ET_AR-ET1 and ET_AR-ET3 were obtained by clustering analysis using the MMTSB toolset (27). PDB structures for ET_AR-ET1 and ET_AR-ET3 were generated from the last 400 ns molecular dynamics trajectories with a 100 ps interval. A centroid structure was obtained by averaging the PDB structures. Clustering analysis was performed using the K-means algorithm (28) based on the RMSD similarity of the structures. The structure that has the lowest RMSD from the centroid structure was obtained as the representative structure. Visual molecular dynamics (29) was used for the visualization of the structures.

Results and discussion

In this study, the structures of ET_AR in complex with ET1 and ET3 were modeled based on the crystal structure of ET_BR in complex with ET1 using homology modeling. Homology modeling is a powerful tool to predict an unknown protein structure based on a structure of a homologous protein (30). Over the modeled region, the sequence of ET_AR (amino acids 67–384) has nearly 67% identity with that of ET_BR (amino acids 88–401). This is much higher than the 30% sequence identity that is considered reliable to model membrane proteins (31,32).

Totally, three independent molecular dynamics simulations were performed for the system of ET_AR in complex with ET1 (ET_AR-ET1) and that of ET_AR in complex with ET3 (ET_AR-ET3). Before data analyses, the RMSDs for the Cα atoms of ET_AR and the two ETs were calculated to evaluate the equilibrations of the simulations (Fig. 1 B). The RMSD values of ET_AR and the two ETs got equilibrated after 600 ns in all the simulations, thus the last 400 ns simulation trajectories were used to obtain average values and standard errors for the studied variables.

The RMSD values of ET_AR are around 4–5 Å (black lines in Fig. 1 B). These large structural deviations mainly resulted from the structural flexibility of the loop regions of ET_AR, and the RMSD values only reached around 2 Å when the loop regions of the receptor were excluded in the RMSD calculations (Fig. S2). To investigate the detailed structural deviation of ET_AR, the RMSD values for each helix of the receptor were calculated (Fig. S3). Most helices kept stable structures with the RMSD values smaller than 2 Å. However, the RMSD values of TM6 and TM7 were larger than other helices and reached around 3 Å at the end of most simulations. This indicates that TM6 and TM7 underwent large conformational changes during the simulations. Compared with those of ET_AR, the RMSD values of ETs are smaller (red lines in Fig. 1 B). Representative structures show that ET1 and ET3 share a similar binding mode with ET_AR, especially for the helical and C-terminal regions of ETs (Fig. 1 C). The RMSD difference between ET1 and ET3 is 1.09 Å, with large structural deviations in the N-terminal region of ETs. The detailed analyses for the interactions between ETs and ET_AR will discuss later in this study.

ET1 has greater binding free energy with ET_AR than ET3

Experimental data indicate that the binding affinity of ET_AR with ET1 is much higher than that with ET3 (11). To test whether it is the case in the simulations, binding free energies between ET_AR and the two ETs were calculated (Table 1 and Fig. S4). In addition, the components of binding free energy were computed to analyze which component plays an important role in the binding (Table 1). For the binding free energies, a negative value means the energy is favorable for the binding, whereas a positive value means the energy is unfavorable for the binding.

Table 1.

Comparison of the binding free energies (kcal/mol) between ET_AR-ET1 and ET_AR-ET3

Energy	ETAR-ET1	ETAR-ET3
ΔEelec	−625.47 ± 2.26	−478.03 ± 2.38
ΔEvdw	−140.68 ± 0.44	−121.21 ± 0.47
ΔGpol	697.17 ± 2.05	535.89 ± 2.29
ΔGnonpol	−21.97 ± 0.05	−20.00 ± 0.05
–TΔS	51.32 ± 0.80	51.17 ± 0.80
ΔGelec+pol	71.70 ± 3.05	57.86 ± 3.30∗
ΔGvdw+nonpol	−162.65 ± 0.44	−141.21 ± 0.47∗
ΔGbinding	−39.63 ± 1.16	−32.18 ± 1.04∗

ET1 has greater binding free energy with ET_AR than ET3. ΔE_elec, electrostatic energy in the gas phase; ΔE_vdw, van der Waals energy; ΔG_pol, polar solvation energy; ΔG_nonpol, nonpolar solvation energy; ΔG_binding = ΔE_ele + ΔE_vdw + ΔG_pol + ΔG_nonpol –TΔS. ^∗Indicates that the difference is significant.

Among the energy components, the electrostatic energy (ΔE_elec), van der Waals energy (ΔE_vdw), and nonpolar solvation energy (ΔG_nonpol) are favorable for binding, whereas the polar solvation energy (ΔG_pol) and entropy contribution (–TΔS) are unfavorable for binding. The entropy contribution is the same between the two systems (51.32 ± 0.80 kcal/mol in ET_AR-ET1 vs. 51.17 ± 0.80 kcal/mol in ET_AR-ET3), thus it does not contribute to the difference of the binding. The effect of electrostatic energy (ΔE_elec) is counteracted by that of polar solvation energy (ΔG_pol), and the sum of the two polar energy contributions (ΔG_elec+pol) is unfavorable for the binding. In contrast, the sum of the two nonpolar energy contributions (ΔG_vdw+nonpol) shows that the nonpolar and van der Waals interactions contribute favorably to the binding. By summing all the components, the binding free energy (ΔG_binding) of ET_AR-ET1 (−39.63 ± 1.16 kcal/mol) is much greater than that of ET_AR-ET3 (−32.18 ± 1.04 kcal/mol). To validate the simulation results, the binding affinities between ET_AR and ETs were also calculated using the PRODIGY web server (33). PRODIGY is a contact-based binding affinity predictor that uses structural properties of protein-protein interactions, the number of interfacial contacts, and noninteracting surfaces to predict the binding affinity between proteins (34). Results show that the binding affinity is −11.18 ± 0.98 kcal/mol for ET_AR-ET1, whereas it is −9.68 ± 0.80 kcal/mol for ET_AR-ET3. Thus, our simulation results are consistent with the experimental data that ET_AR has a higher binding affinity with ET1 than that with ET3.

The binding affinities of ET1 and ET3 for ET_AR and ET_BR have been studied by TGF-α shedding and β-arrestin recruitment assays (14). To further investigate whether our simulation results are consistent with the data of the two assays, two additional simulations for systems of ET_BR with ET1 and ET3 (denoted as ET_BR-ET1 and ET_BR-ET3) were carried out. Then, molecular mechanics/generalized Born surface area calculations were performed for the two systems and the results are presented in Fig. S5 and Table S1. The binding free energy for ET_BR-ET1 (−37.41 ± 1.97 kcal/mol) is similar to that for ET_BR-ET3 (−37.96 ± 2.18 kcal/mol). Furthermore, the binding free energy for ET_AR-ET3 (−32.18 ± 1.04 kcal/mol) is much smaller than that for ET_AR-ET1 (−39.63 ± 1.16 kcal/mol), ET_BR-ET1 (−37.41 ± 1.97 kcal/mol), and ET_BR-ET3 (−37.96 ± 2.18 kcal/mol). Therefore, the simulation results are consistent with the data of TGF-α shedding and β-arrestin recruitment assays that ET1 binds ET_AR and ET_BR with similar affinities, while ET3 binds ET_AR with a much lower affinity. Since we focus on the selectivity of ET_AR with ET1 and ET3 in this study, we do not further discuss the interaction details of ET_BR with ET1 and ET3.

Residues of ET_AR that show different preferences in the binding of ET1 and ET3 are identified

To investigate which residues of ET_AR play important roles in the binding with ET1 and ET3, the interaction energy for each residue of ET_AR was calculated by decomposing the binding free energy (upper panel in Fig. 2 A). Interaction energy results indicate that residues R145^ECL1, W146^ECL1, F161^3.28, P162^3.29, K166^3.33, V169^3.36, F224^ECL2, V227^ECL2, F229^ECL2, Y231^ECL2, T238^ECL2, M240^ECL2, K255^5.38, L259^5.42, Y263^5.46, W319^6.48, L322^6.51, R326^6.55, R340^7.24, and I355^7.39 of ET_AR contribute positively to the binding of ET_AR with both ET1 and ET3; whereas residues D133^2.57, K140^2.64, E220^4.60, E234^ECL2, D256^5.39, K329^6.58, E335^ECL3, and D351^7.35 are unfavorable for the binding. (The superscript after each residue indicates Ballesteros-Weinstein numbering for the residues of GPCRs (35).) All the residues unfavorable for binding are charged residues, thus these residues may affect the binding through long-range electrostatic interactions. In addition to the residues that are favorable for the binding of ET_AR with the two ETs, several residues only favor the binding with one of the ETs: residues Y68^N-ter, L344^7.28, L348^7.32, and Y352^7.36 are favorable for the binding of ET_AR with ET1, whereas residues F148^ECL1, Q165^3.32, L241^ECL2, N242^ECL2, and A243^ECL2 are favorable for the binding of ET_AR with ET3.

Figure 2.

Identification of residues preferable for the binding of ET_AR with ET1 or ET3. (A) Upper panel: interaction energy for each residue of ET_AR in ET_AR-ET1 (black) and ET_AR-ET3 (red). Residues responsible for the binding of both ET1 and ET3 are labeled in blue, residues that favor ET1 are labeled in gray, residues that favor ET3 are labeled in purple, and residues that disfavor the binding are labeled in orange. Lower panel: difference of the interaction energy (green) between ET_AR-ET1 and ET_AR-ET3. Residues that favor ET1 are labeled in black, whereas residues that favor ET3 are labeled in red. Residues labeled in orange are those that disfavor the binding of both ET1 and ET3 in the upper panel. The regions for the transmembrane helices are shaded in gray. (B) Upper panel: interaction energy for each residue of ET1 and ET3. Lower panel: difference of the interaction energy between ET1 and ET3. The regions for the three regions of ET1/3 are shaded in blue, orange, and purple, respectively. The concentric circles show the positions of cysteine residues that form two intramolecular disulfide bonds. The residues that favor ET1 and ET3 binding to ET_AR are labeled using black and red dotted lines, respectively. To see this figure in color, go online.

Among the residues that are favorable for the binding of ET_AR with both ET1 and ET3, several of them display different preferences in the binding with the two ETs. To further study how these residues discriminate ET1 and ET3, the interaction energy difference between each residue of ET_AR in ET_AR-ET1 and ET_AR-ET3 was calculated (lower panel in Fig. 2 A). In the results of interaction energy difference, a residue with a negative value is preferred for the binding with ET1, whereas a residue with a positive value is preferred for the binding with ET3. It reveals that besides the four residues (i.e., Y68^N-ter, L344^7.28, L348^7.32, and Y352^7.36), residues R145^ECL1, F229^ECL2, Y231^ECL1, R232^ECL2, H236^ECL2, L259^5.42, W319^6.48, L322^6.51, R326^6.55, K329^6.58, M336^ECL3, and R340^7.24 are preferred for the binding with ET1; while in addition to the five residues (i.e., F148^ECL1, Q165^3.32, L241^ECL2, N242^ECL2, and A243^ECL2), residues F161^3.28, K166^3.33, F224^ECL2, Q252^5.35, and K255^5.38 are preferred for the binding with ET3.

Recently, the interaction features of GPCRs with peptide ligands have been studied based on the knowledge of GPCR-peptide structures and mutagenesis data (36). The study reveals that the residues at positions 3.29, 3.32, 3.33, 4.60, 4.64, 5.42, and 6.55 in GPCRs are highly conserved between the members of GPCRs and these residues are vital for the binding of peptide ligands (36). Consistent with the study, our interaction energy analysis demonstrates that these residues of ET_AR are essential for the binding of ET1 and ET3 (Fig. 2 A).

To further test whether the interaction residues identified in our simulations are in line with experimental data, the mutagenesis data that analyze the interactions of ET_AR with ETs and antagonists were obtained from the GPCRdb database (37) and summarized in Table S2. Among the mutations, Q165D^3.32 and K140I^2.64 reduce the binding affinity of ET_AR with ET1. This is consistent with our interaction energy analysis that Q165 and K140 play important roles in the binding of ET1 and ET3 (Fig. 2 A). Of the mutations, L142M^2.66, A143K^2.67, G144M^ECL1, F156I^3.23, S167A^3.34, T172V^3.39, E220N^4.60, V225A/D^ECL2, G261A^5.44, F264L^5.47, F320A^6.49, H323F^6.52, K330I^6.59, and N361A^7.45 show no significant change in the binding of ET1. This is in agreement with our simulation results that the residues corresponding to the mutations are not involved in the binding of ETs (Fig. 2 A). Furthermore, the mutagenesis data show that the mutations including D126A^2.50, Y129F/W/K/S/A^2.53, D133N/A^2.57, K159Q^3.26, F260Y^5.43, Y263F^5.46, F315L^6.44, R326Q^6.55, and D351N^7.35 reduce the binding affinity of ET receptor antagonists (bosentan or BMS-182874) rather than that of ETs (Table S2). As the residues corresponding to these mutations do not affect the binding of ETs (Fig. 2 A), these residues may be used to differentiate the binding of ETs and antagonists.

To date, three nonpeptide ET receptor antagonists have been approved, i.e., bosentan, ambrisentan, and macitentan (38). While ambrisentan is an ET_AR-selective antagonist, bosentan and macitentan are nonselective antagonists that block both ET_AR and ET_BR. The crystal structure of ET_BR in complex with bosentan shows that K182^3.33, K273^5.38, and R343^6.55 of ET_BR are critical for the binding of bosentan (39). Computational study identifies that the corresponding residues in ET_AR (i.e., K166^3.33, K255^5.38, and R326^6.55) play important roles in the binding of bosentan and the potential natural inhibitors of ET_AR (40). Similar to the binding of the antagonists, our simulation results also indicate that the three residues are vital in the binding of ET1 and ET3 (Fig. 2 A).

Recent mutagenesis data demonstrate that residues L322^6.51, R326^6.55, and I355^7.39 of ET_AR showed pronounced and differential effects on the binding of the three approved antagonists (41): 1) the mutation L322A^6.51 reduces the antagonist activity of macitentan, whereas bosentan and ambrisentan are unaffected; 2) the mutation R326Q^6.55 does not affect the antagonist activity of macitentan, but reduces the activity of bosentan and ambrisentan; and 3) the mutation I355A^7.39 significantly reduced bosentan potency, but not ambrisentan and macitentan potencies. It is noted that L322^6.51 and R326^6.55 in our simulations are preferable for the binding of ET1, while I355^7.39 contributes equally to the binding of ET1 and ET3 (Fig. 2 A). These findings that residues L322^6.51, R326^6.55, and I355^7.39 differentiate the binding of ETs and the antagonists may help to design novel ET_AR inhibitors.

ET_AR discriminates ET1 and ET3 by the residue at position 14 of the two ETs

A comparison of the sequence between ET1 and ET3 shows that six residues (five in the N-terminal region and one in the α-helical region) are different between ET1 and ET3, while most of the residues in the α-helical and C-terminal regions are conserved between the two ETs (Fig. 1 A). To investigate which residues of ET1 and ET3 contribute positively to the binding of ET_AR, the interaction energy for each residue of ETs was calculated (upper panel in Fig. 2 B). Since the four cysteine residues of ETs form two intramolecular disulfide bonds (i.e., C1-C15 and C3-C11) that are mainly responsible for the structural stability of the two ETs, the effects of these four residues on the binding are not further discussed. Results show that residues D8, V12, Y13, residue 14 (F14 in ET1 and Y14 in ET3), H16, and L17 in the α-helical region and the four residues (¹⁸DIIW²¹) in the C-terminal region have large interaction energies, indicating that these residues play essential roles in the binding with ET_AR.

Although the residues in the N-terminal region play a less important role in the binding, the difference between these residues in ET1 and ET3 may affect the ligand selectivity of ET_AR. To evaluate which residues in ET1 and ET3 are responsible for the selectivity of ET_AR, the interaction energy difference for the two ETs was calculated (lower panel in Fig. 2 B). For the interaction energy difference, a negative value indicates that the residue is preferred for the binding of ET1 with ET_AR, whereas a positive value indicates that the residue is preferred for the binding of ET3 with ET_AR. Results reveal that residues S5, M7, D8, F14, and W21 in ET1 are preferred for the binding with ET_AR compared with the corresponding residues in ET3; while residues T2, F4, Y13, and I20 in ET3 are preferred for the binding with ET_AR compared with the corresponding residues in ET1.

It is noted that, among the six residues that are different between ET1 and ET3, the residue at position 14 (F14 in ET1 and Y14 in ET3) is a key residue that plays an important role in the binding between ET_AR and ETs (upper panel in Fig. 2 B). Moreover, compared with the other residues, this residue has the largest value for the interaction energy difference (lower panel in Fig. 2 B), demonstrating that F14 in ET1 is more preferable for the binding with ET_AR than Y14 in ET3. Thus, these data suggest that the residue at position 14 plays an important role in discriminating the selectivity of ET_AR with ET1 and ET3.

ET_AR reveals different interaction patterns with ET1 and ET3

The residues of ET_AR that discriminate ET1 and ET3 have been identified in this work (Fig. 2). To give a clear view of how these residues interact with the two ETs, the representative simulation structures showing the detailed interaction for each residue of ETs are presented in Figs. 3 and S6. The interactions for residues in positions 1, 3, 11, and 15 of ETs are not discussed since these four residues form two intramolecular disulfide bonds that are mainly responsible for the structural stability of the two ETs. By comparing the interactions between each residue of ET1 and that of ET3, the reasons why the residues of ET1 and ET3 are more preferable for the binding with ET_AR are discussed.

Figure 3.

Comparison of the interactions of ET_AR with ET1 and ET3. (A) Comparison of the interactions of ET_AR with the residues in the N-terminal region of the two ETs. Compared with S2 and S4 of ET1, T2 and F4 of ET3 have more interactions with ET_AR. (B) Comparison of the interactions of ET_AR with the residue at position 14 of the two ETs. Dotted circles indicate that there are hydrophobic contacts between the residues in the circle. There are large numbers of hydrophobic contacts between F14 of ET1 and residues Y68, L344, and L348 of ET_AR; however, there are few contacts between Y14 of ET3 and the three residues of ET_AR. (C) Comparison of the interactions of ET_AR with I20 of the two ETs. Both F161 of ET_AR and I20 of ET3 undergo orientation changes, which allows F161 to move closer to I20 of ET3. (D) Comparison of the interactions of ET_AR with W21 of the two ETs. Compared with that in ET1, W319^6.48 in ET3 undergoes an orientation change. The green dotted lines indicate that there is a hydrogen bond between the two residues. The carbon atoms in residues of ET_AR are shown in tan, while the carbon atoms in residues of the N-terminal, α-helical, and C-terminal regions of the two ETs are shown in cyan, orange, and purple, respectively. The nitrogen atoms are shown in blue, oxygen atoms in red, and hydrogen atoms in white. For clarity, only the side chain of each residue is shown. To see this figure in color, go online.

The interactions of residues in the N-terminal region of ETs are shown in Fig. 3 A, which reveals that residue S2 and L6 of ET1 are surrounded by M240^ECL2, L241^ECL2, Q252^5.35, N242^ECL2, V225^ECL2, and V227^ECL2 of ET_AR; whereas T2 and Y6 of ET3 are surrounded by the same residues except that the interacting residues N242^ECL2 and V255^ECL2 in ET_AR-ET1 are replaced by A243^ECL2 and T244^ECL2 in ET_AR-ET3. In ET1, there is no interaction for residues S4, S5, and M7. Similarly, no interaction is found for T5 and K7 in ET3. However, residue F4 in ET3 interacts with E355^ECL3 and M336^ECL3 of ET_AR. In addition, the side chain of T2 in ET3 forms a hydrogen bond with the main chain oxygen atom of Q252^5.35. Due to these interactions, T2 and F4 in ET3 are more preferable for the binding with ET_AR than S2 and S4 in ET1 (lower panel in Fig. 2 B).

The interactions for residues in the α-helical region of ETs are shown in Figs. 3 B and S6, which reveals that residues D8, K9, E10, V12, and Y13 in the two ETs have similar interactions with ET_AR (Fig. S6 A). These residues in ET1 are surrounded by Y68^N-ter, F227^ECL2, F229^ECL2, Y231^ECL2, H236^ECL2, T238^ECL2, M240^ECL2, and R340^7.24 of ET_AR. Similarly, these residues in ET3 are surrounded by the same residues of ET_AR except that the interacting residue F277^ECL2 in ET_AR-ET1 is replaced by R145^ECL1 in ET_AR-ET3. Compared with that in ET1, residue Y13 in ET3 forms a cation-π interaction with R145^ECL1. The strong interaction formed by these two residues in ET_AR-ET3 may explain why Y13 is more preferable for the binding of ET_AR with ET3 (Fig. 2 B).

The comparison of the interactions between F14 in ET1 and Y14 in ET3 is displayed in Fig. 3 B, which reveals that F14 of ET1 is located in a hydrophobic environment formed by Y68^N-ter, L344^7.28, and L348^7.32 of ET_AR. However, Y14 of ET3 brings in a hydrophilic hydroxyl group that would impede the hydrophobic interactions with the three residues of ET_AR. To confirm whether this is the case, the hydrophobic contacts between F/Y14 of the two ETs and the three residues of ET_AR were calculated (Fig. S7). According to the work of Cheng et al. (42), a contact was counted if any two carbon atoms between the two residues were identified within 5.4 Å. Results show that there are large numbers of hydrophobic contacts between F14 of ET1 and the three residues of ET_AR; however, there are few or even no contact between Y14 of ET3 and the three residues of ET_AR (Fig. S7). Combining the interaction energy data that Y68^N-ter, L344^7.28, and L348^7.32 of ET_AR are favorable for the binding with ET1 but not ET3 (Fig. 2 A), these results demonstrate that ET_AR discriminates ET1 and ET3 by using the three residues Y68^N-ter, L344^7.28, and L348^7.32 to recognize the residue at position 14 of the two ETs.

In both ET1 and ET3, residues H16 and L17 in the α-helical region are surrounded by five residues of ET_AR, i.e., K140^2.64, R145^ECL1, W146^ECL1, H236^ECL2, and T238^ECL2 (Fig. S6 B). However, in ET1, another three residues of ET_AR (i.e., Y68^N-ter, L348^7.32, and Y352^7.36) have contacts with L17; whereas in ET3, residue T73^N-ter of ET_AR is also involved in the interaction with L17. Residues D18 and D19 in the C-terminal region are surrounded by five residues of ET_AR, i.e., K140^2.64, L322^6.51, R326^6.55, D351^7.35, and I355^7.39 (Fig. S6 C). However, in ET1, another three residues of ET_AR (i.e., L141^2.65, L348^7.32, and Y352^7.36) have contacts with D18; whereas in ET3, residues D133^2.57 and Q165^3.32 of ET_AR are also involved in the interaction with I19. The interaction energy differences for residues H16, L17, D18, and I19 are small between ET1 and ET3 (Fig. 2 B), thus these residues are less important for the ligand selectivity.

The interaction comparison for I20 in ET1 and ET3 is presented in Fig. 3 C, which shows that I20 in both ET1 and ET3 is surrounded by F161^3.28, P162^3.29, Q165^5.32, and F224^ECL2 of ET_AR. However, compared with that in ET1, I20 in ET3 undergoes an orientation change. This change is examined by calculating the dihedral angle for the side chain of I20. The dihedral angle is distributed around 65° in ET1, but it is also distributed around 207 and 300° in ET3 (Fig. S8 A). Accompanied with this change, F161^3.28 also undergoes an orientation change. The dihedral angle for the side chain of F161^3.28 is changed from ∼275° in ET1 to ∼190° in ET3 (Fig. S8 B), which allows F161^3.28 to move closer to I20 of ET3. Combining with the result that F161^3.28 is favorable for the interaction with ET3 (Fig. 2 B), the closed interaction between F161^3.28 of ET_AR and I20 of ET3 makes I20 preferable for the binding of ET_AR with ET3 but not ET1.

The interaction energy results have demonstrated that the C-terminal W21 in both ET1 and ET3 plays an essential role in the binding with ET_AR (upper panel in Fig. 2 B). In both ET_AR-ET1 and ET_AR-ET3, W21 penetrates into the bottom of the orthosteric pocket and interacts with eight residues of ET_AR, i.e., Q165^3.32, K166^3.33, V169^3.36, L259^5.42, Y263^5.46, W319^6.48, H323^6.52, and R326^6.55 (Fig. 3 D). Taken together, these residues contribute greatly to the binding between ET_AR and the two ETs. Besides the eight residues, W21 in ET1 is also surrounded by E220^4.60, K255^5.38, and L322^6.51, whereas W21 in ET3 is also surrounded by D256^5.39. Notably, the orientation of W319^6.48 in ET_AR-ET3 is different from that in ET_AR-ET1, which may make the residue W319^6.48 less favorable for the binding of ET3. The residue W319^6.48 is located in the CWxP motif that is reported to be involved in the signaling function of GPCRs (43). The effect of how the orientation change alters the signaling function of ET_AR is discussed in the following section.

ET1 has higher structural and dynamical stabilities in binding with ET_AR

To study whether there are structural differences of ET_AR in the binding of ET1 and ET3, secondary structure analysis for ET_AR was performed (Fig. 4 A). A comparison of the occupancies for all the helices of ET_AR reveals that the occupancies for TM2, TM3, TM4, and TM5 are similar between ET_AR-ET1 and ET_AR-ET3 (Table S3). In contrast, the occupancies for TM1, TM6, and helix 8 in ET_AR-ET1 are smaller than those in ET_AR-ET3, whereas the occupancy for TM7 in ET_AR-ET1 is larger than that in ET_AR-ET3 (Table S3). By summing the occupancies for all the residues, the total helix occupancy of ET_AR in ET_AR-ET1 is similar to that in ET_AR-ET3 (63.8 vs. 64.3%). The secondary structure analysis was also performed for the α-helical region of ET1 and ET3 (Fig. 4 B). Results indicate that the occupancy for the α-helical region of ET1 is larger than that of ET3 (59.3 vs. 56.7%). Furthermore, the occupancy for F14 of ET1 is much larger than that for Y14 of ET3 (97.8 vs. 81.1%). From a structural point of view, this result supports the data that F14 of ET1 is more preferable for the binding of ET_AR than Y14 of ET3.

Figure 4.

Comparison of the secondary structure between ET_AR-ET1 and ET_AR-ET3. (A) Comparison of the helix occupancy of ET_AR in ET_AR-ET1 with that in ET_AR-ET3. The total helix occupancy for ET_AR is similar between the two systems. (B) Comparison of the helix occupancy for each residue in the α-helical region of ET1 and ET3. The total helix occupancy for the α-helical region of ET1 is larger than that of ET3, indicating that ET1 is structurally more stable in the binding with ET_AR. To see this figure in color, go online.

The RMSF for each residue of ET_AR and the two ETs was calculated to assess whether there are dynamical differences between ET_AR-ET1 and ET_AR-ET3 (Fig. 5). For ET_AR, the RMSF values for the loop regions and helix 8 are large; whereas the values for TM1-7 are small (<1 Å) in both ET_AR-ET1 and ET_AR-ET3 (Fig. 5 A). However, the RMSF values for TM2-7 in ET_AR-ET1 are smaller than those in ET_AR-ET3, indicating that these transmembrane helices of ET_AR are dynamically stable in the binding of ET1. For the two ETs, the RMSF values for the residues in the α-helical region are similar between ET_AR-ET1 and ET_AR-ET3 (Fig. 5 B). However, the RMSF values for the residues in the N-terminal and C-terminal regions of ET1 are smaller than those of ET3, indicating that ET1 is dynamically more stable in the binding with ET_AR than ET3.

Figure 5.

Comparison of the dynamic flexibility between ET_AR-ET1 and ET_AR-ET3. (A) Comparison of the RMSF value for each residue of ET_AR in ET_AR-ET1 with that in ET_AR-ET3. The RMSF values for TM2-7 in ET_AR-ET1 are smaller than those in ET_AR-ET3. (B) Comparison of the RMSF values for each residue of ET1 and ET3. The RMSF values for the residues in the N-terminal and C-terminal regions of ET1 are smaller than those of ET3, indicating that ET1 is dynamically more stable in the binding with ET_AR. To see this figure in color, go online.

Conformational difference of ET_AR in the binding of ET1 and ET3 is identified

We have found that, compared with that in ET_AR-ET1, W319^6.48 in ET_AR-ET3 undergoes an orientation change (Fig. 3 D). This change results in a vertical conformation of the side chain of W319^6.48 with respect to the side chain of W21, which leads to a less tight packing of W319^6.48 with W21 of ET3 (Fig. 3 D). This observation is consistent with the interaction energy analysis that the interaction energy value of W319^6.48 in ET_AR-ET3 is smaller than that in ET_AR-ET1 (−0.49 kcal/mol in ET_AR-ET1 vs. −0.27 kcal/mol in ET_AR-ET3 in the upper panel of Fig. 2 A). In both ET_AR-ET1 and ET_AR-ET3, the contribution of W319^6.48 to the interaction energy is small, thus it is less likely that W319^6.48 determines the binding affinities of ET_AR with ETs. However, since this residue is located in the CWxP motif that plays an important role in regulating the signaling function of GPCRs, the orientation change of W319^6.48 caused by the binding of ET3 may lead to the conformational rearrangement of ET_AR. This proposition is supported by the findings that, when ET1 or ET3 binds to ET_BR, W336^6.48 in ET_BR (residue corresponding to W319^6.48 in ET_AR) moves downward and induces a conformational change of its neighboring residue F332^6.44 (residue corresponding to F315^6.44 in ET_AR) in the middle part of TM6, which may ultimately result in an outward displacement of TM6 in the intracellular side of ET_BR (13,14).

To investigate whether the orientation change of W319^6.48 can affect the interactions with its neighboring residues, the representative structures showing the interactions of W319^6.48 in ET_AR-ET1 and ET_AR-ET3 are presented (Fig. 6 A). This shows that W319^6.48 is surrounded by four residues Y129^2.53, F315^6.44, I355^7.39, and N361^7.45 in ET_AR-ET1; however, due to the rotation of Y129^2.53, W319^6.48 lost its interaction with Y129^2.53 in ET_AR-ET3. To check whether the neighboring residues also undergo orientation changes, the dihedral angles for the side chain of W319^6.48 and its neighboring residues were computed (Figs. 6 B and S9). Results reveal that, besides W319^6.48, the neighboring residues Y129^2.53, F315^6.44, and I355^7.39 also undergo orientation changes. Since W319^6.48 is located in the CWxP motif that plays an important role in the signaling function of GPCRs, the changes of this residue and its neighboring residues may cause the conformational change of ET_AR.

Figure 6.

Orientation changes of W319 and its neighboring residues. (A) Comparison of the interactions of W319 with its neighboring residues Y129, F315, and I355. W319 undergoes orientation change and loses its contact with Y129 in ET_AR-ET3. (B) The distributions for the dihedral angle of the side chain of W319 and its neighboring residues. Accompanied with W319, residues F315, I355, and Y129 also undergo orientation changes. To see this figure in color, go online.

Similar to the role of W366^6.48 in regulating the conformation of ET_BR, the orientation changes of W319^6.48 and its neighboring residues may also cause the conformational change of ET_AR. The conformational change of ET_AR is first explored by calculating the volume of the orthosteric pocket for the binding of ET1 and ET3 (Figs. 7 and S10). The averaged pocket volume is 1701.02 ± 10.00 Å³ in ET_AR-ET1, but is 1932 ± 8.68 Å³ in ET_AR-ET3. Compared with that for ET1, the pocket volume for ET3 is significantly increased. A comparison of the final structures of ET_AR in ET_AR-ET1 and ET_AR-ET3 shows that the volume expansion for the orthosteric pocket of ET_AR in ET_AR-ET3 comes from the rearrangements of TM2, TM6, and TM7 on the extracellular side, which may result in the unstable binding of ET_AR with ET3.

Figure 7.

Conformational difference of ET_AR in binding of ET1 and ET3. Figures are shown from side and extracellular views. Compared with ET1, the orthosteric pocket volume for ET3 is significantly increased. The volume expansion for the orthosteric pocket of ET_AR in ET_AR-ET3 results from the rearrangements of TM2, TM6, and TM7 on the extracellular side. Compared with that in ET_AR-ET1, TM6 in ET_AR-ET3 also undergoes conformational change on the intracellular side. To see this figure in color, go online.

The conformational change in the orthosteric pocket of ET_AR may subsequently lead to the intracellular conformational change of ET_AR. To test this proposition, the distance between R103^1.55 and K199^6.28, which represents the distance between the intracellular ends of TM1 and TM6, was calculated (Figs. 7 and S11). Results show that the distance is 33.53 ± 0.05 Å in ET_AR-ET1, but it is 31.57 ± 0.07 Å in ET_AR-ET3. Compared with that in ET_AR-ET1, the distance in ET_AR-ET3 is significantly decreased. This indicates that the intracellular conformations of ET_AR are different between ET_AR-ET1 and ET_AR-ET3. Recently, a simulation study demonstrated that angiotensin type 1 receptor, a prototypical GPCR, adopts two active intracellular conformations when binding with angiotensin II and its peptide analogs (44). One of the conformations couples to both G-proteins and β-arrestins, whereas the other only couples to β-arrestins and shows more inactive-like positions of TM6 (44). In this study, a large outward displacement of TM6 that characterizes the activation of GPCRs was not observed in ET_AR-ET1 or ET_AR-ET3. To further check the activation state of ET_AR, the distance between L122^2.46 and V308^6.37 that is suggested as an indicator for the activation state of ET_AR (37) was calculated. The distance is 13.89 ± 0.07 Å in ET_AR-ET1 and 12.91 ± 0.13 Å in ET_AR-ET3. Thus, despite the conformational differences in the intracellular side of the receptor, ET_AR still adopts an inactive state in the simulations. This is consistent with the experimental and simulation findings that the binding of agonists only induces structural heterogeneity of GPCRs, whereas the binding of G-proteins or β-arrestins is necessary to cause the activation conformation of GPCRs (45,46). Similar to angiotensin type 1 receptor, the conformational difference of ET_AR in the intracellular side of TM6 caused by the binding of the two ETs may result in preferential signaling to recruit various G-proteins or β-arrestins. To date, the research about the G-protein or β-arrestin biased ET_AR signaling is still lacking. Further experimental and theoretical studies are needed to investigate whether the peptide agonists could regulate the biased signaling of ET_AR.

Conclusion

In this study, the models of ET_AR in complex with the two isopeptides, i.e., ET1 and ET3, were built. Then molecular dynamics simulations were performed for the two models to investigate how ET_AR recognizes ET1 and ET3. First, the binding free energy between ET_AR and the two ETs was calculated to test which ET is preferable for the binding. Results demonstrate that ET1 has greater binding free energy with ET_AR than ET3, which is consistent with the binding affinity data of ET_AR. Then the interaction energy for each residue of ET_AR and the two ETs was obtained by decomposing the binding free energy. It reveals that residues Y68, R145, F229, Y231, R232, H236, L259, W319, L322, R326, K329, M336, R340, L344, L348, and Y352 of ET_AR are preferred for the binding with ET1; while residues F148, F161, Q165, K166, F224, L241, N242, A243, Q252, and K255 of ET_AR are preferred for the binding with ET3. Detailed interaction analyses indicate that ET_AR can recognize ET1 and ET3 by using residues Y68, L344, and L348 to discriminate F14 of ET1 and Y14 of ET3. Secondary structure and dynamical flexibility analyses show that ET1 is more stable in the binding of ET_AR than ET3. In addition, the orientation change of W319 was observed. This residue is located in the conserved CWxP motif that plays an important role in the signaling function of GPCRs. Furthermore, the difference in the orthosteric pocket volume caused by the binding of ET1 and ET3 was found. This difference results from the rearrangement of TM2, TM6, and TM7 on the extracellular side of ET_AR. Accompanied with the change in the orthosteric pocket, a conformational difference of TM6 on the intracellular side of ET_AR was also observed. All the findings in this study provide a molecular mechanism of how ET_AR selects the isopeptides ET1 and ET3, which may help in designing novel ET receptor antagonists to treat cardiovascular and renal diseases.

Author contributions

L.W. and F.Y. conceived the work. L.W. and L.W. carried out all simulations. L.W., L.W., and F.Y. analyzed the data and wrote the manuscript.

Editor: Chris Chipot.