G-Protein-Coupled Receptor-Membrane Interactions Depend on the Receptor Activation state

Abstract

G-protein-coupled receptors (GPCRs) are the largest family of human membrane proteins and serve as primary targets of ~1/3 of currently marketed drugs. In particular, adenosine A₁ receptor (A₁AR) is an important therapeutic target for treating cardiac ischemia-reperfusion injuries, neuropathic pain and renal diseases. As a prototypical GPCR, the A₁AR is located within a phospholipid membrane bilayer and transmits cellular signals by changing between different conformational states. It is important to elucidate the lipid-protein interactions in order to understand the functional mechanism of GPCRs. Here, all-atom simulations using a robust Gaussian accelerated molecular dynamics (GaMD) method were performed on both the inactive (antagonist bound) and active (agonist and G protein bound) A₁AR, which was embedded in a 1-palmitoyl-2-oleoyl-glycero-3-phosphocholine (POPC) lipid bilayer. In the GaMD simulations, the membrane lipids played a key role in stabilizing different conformational states of the A₁AR. Our simulations further identified important regions of the receptor that interacted distinctly with the lipids in highly correlated manner. Activation of the A₁AR led to differential dynamics in the upper and lower leaflets of the lipid bilayer. In summary, GaMD enhanced simulations have revealed strongly coupled dynamics of the GPCR and lipids that depend on the receptor activation state.

Graphic Abstract:

Using adenosine A₁ receptor (A₁AR) as a model G-protein-coupled receptor (GPCR), Gaussian accelerated molecular dynamics (GaMD) was applied to explore GPCR-membrane interactions. Membrane lipids played a key role in stabilizing different conformational states of the A₁AR. Activation of the A₁AR led to differential dynamics in the upper and lower leaflets of the lipid bilayer. The GaMD simulations revealed strongly coupled dynamics of the GPCR and lipids that depend on the receptor activation state.

Introduction

G-protein-coupled receptors (GPCRs) are primary cell surface receptors that account for vital physiological and pathological functions in the human body. About 1/3 of currently marketed drugs approved by the Food and Drug Administration (FDA) target GPCRs. Four subtypes of adenosine receptors (ARs), the A₁AR, A_2AAR, A_2BAR and A₃AR, mediate a broad range of physiological functions. Particularly, the Adenosine A₁ Receptor (A₁AR) has emerged as an important therapeutic target for treating cardiac ischemia-reperfusion injuries, neuropathic pain and renal diseases.(1) Being a GPCR, the A₁AR is embedded in cell membrane, maintaining close contact with lipid molecules. Lipids have been suggested to affect the receptor conformation and dynamics, which play an important role in transmitting cellular signals from extracellular environment to the cytoplasm. Similarly, lipid metabolites are also known to bind proteins and act as messengers(2). These include lysolipids, sphingo-1-phosphate (S1P), diacylglycerol and fatty acyl derivates. In addition, lipids help in the partitioning of membrane and receptors. Membrane proteins are affected by lipid compositions and function differently in healthy and diseased individuals.(3) Therefore, it is important to study GPCR-membrane interactions in order to elucidate functional mechanism of the membrane proteins.

Experimental techniques including fluorescence resonance energy transfer (FRET), fluorescence correlation spectroscopy (FCS), fluorescence recovery after photobleaching (FRAP) and fluorescence-based monitoring of solvent relaxations rates have been utilized to study protein-membrane biology(4–7). Experiments showed that cholesterol could affect the stability, oligomerization, and ligand binding properties of membrane proteins(8–21). X-ray crystal structures identified allosteric sites for cholesterol binding to GPCRs(13, 15, 22). Phospholipids were found to modulate dynamic processes of GPCRs such as G protein association and ligand binding(23, 24). Recently, Dawaliby et. al. showed experimentally that lipids with different head groups favor different activation states of the β₂-adrenergic receptors (β₂AR)(25). Lipids with phosphatidylglycerol (PG) headgroups preferred agonist binding and receptor activation, whereas lipids with phosphatidylethanolamine (PE) headgroups preferred antagonist binding and inactive state of the β₂AR. Despite these advances, there remains a knowledge gap in the understanding of protein-membrane interactions. From atomic motions of lipid molecules to curvature change across the cell membrane, protein-membrane interactions span a wide range of time scales(26). It is often difficult to directly examine protein-membrane interactions in experiments due to limited time resolution.

Molecular dynamics (MD) simulation has emerged as a powerful computational technique to bridge the gap of knowledge for studying membrane-protein interactions. Both atomistic and coarse-grained MD simulations have been applied to study the effects of lipids in protein dynamics and cellular signaling.(3) Bruzzese et al. confirmed the above mentioned experimental results obtained by Dawaliby et. al. that different charges of PG and PE lipid headgroups affected the GPCR activation and deactivation in MD simulations(27). The net negative charge in PG molecules favored interaction with positively-charged residues in the intracellular loop 3 (ICL3) and intracellular end of transmembrane helix 6 (TM6). This stabilized the outward movement of TM6 and hence the active state of the β₂AR. Neale et. al.(28) showed that the PG lipid blocked formation of the R131^3.50-E268^6.30 ionic lock by interacting with R131^3.50 in the β₂AR. Residue superscripts denote Ballesteros and Weinstein (BW) numbering of GPCRs(29). The R^3.50 and E^6.30 are highly conserved residues in GPCRs and often form an ionic lock in inactive receptors. In contrast, net-neutral zwitterionic lipids such as PE with strongly favored the inactive structure of β₂AR and destabilized the active structure(27). Salas-Estrada et. al(30) showed that activation in rhodopsin induced changes in the membrane structure, including increase in the local order and effective length of lipid acyl chains in the vicinity of the protein. Dror et. al. showed that gradual inactivation of the β₂AR occurred in the neutral lipid 1-palmitoyl-2-oleoyl-glycero-3-phosphocholine (POPC)(31, 32). This was consistent with another study favored partial deactivation of the β₂AR was found in the 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) lipids(27). In coarse-grained MD simulations, Song et. al. showed that PIP₂ (Phosphatidylinositol 4,5-bisphosphate) stabilized the outward movement of TM6 by binding in the crevice between TM6 and TM7 of adenosine A_2A receptor (A_2AAR)(33).

In addition, enhanced sampling techniques have been applied to investigate protein membrane interaction. Using steered MD(34) and umbrella sampling(35) techniques, Song et. al. suggested that the PIP₂ facilitated recruitment of the G protein by forming bridging interactions with basic residues of the G_α subunit and hence stabilizing the active A_2AAR(33). However, these enhanced sampling techniques require predefined collective variables, which are often difficult to identify in the context of protein-membrane interactions. In this context, Gaussian accelerated MD (GaMD) is a robust technique that provides unconstrained enhanced sampling without the need to set predefined collective variables(36, 37). GaMD simulations have been successfully applied to investigate GPCR activation(36, 38, 39), protein folding(36, 39), ligand binding and unbinding(36, 38, 39), protein-protein interactions(40–42) and protein-nucleic acid interactions(43, 44).

Here, we have applied GaMD to investigate lipid interactions with the A₁AR in two different conformational states, the cryo-EM structure of the active adenosine ADO-bound A₁AR coupled with the G_i protein (referred to as ADO-A₁AR-Gi)(45) and the X-ray structure(46, 47) of the inactive antagonist PSB36-bound A₁AR (referred to as PSB36-A₁AR). Our simulations showed that the protein-membrane interactions depended on different conformational states of the A₁AR. The membrane lipids played an important role in stabilizing different conformations of the A₁AR. GaMD simulations further identified important regions of the receptor that interacted distinctly with the lipids. Activation of the A₁AR led to differential dynamics in the upper and lower leaflets of the lipid bilayer.

Materials and Methods

Gaussian Accelerated Molecular Dynamics (GaMD)

GaMD is an enhanced sampling technique, in which a harmonic boost potential is added to reduce the system energy barriers(36). GaMD is able to accelerate biomolecular simulations by orders of magnitude(39, 48). GaMD does not need predefined collective variables. Moreover, because GaMD boost potential follows a gaussian distribution, biomolecular free energy profiles can be properly recovered through cumulant expansion to the second order(36). GaMD has successfully overcome the energetic reweighting problem in free energy calculations that was encountered in the previous aMD (accelerated molecular dynamics) method(49, 50) for free energy calculations. GaMD has been implemented in widely used software packages including AMBER (36, 51) and NAMD(52). A brief summary of GaMD is provided here.

Consider a system with N atoms at positions $\stackrel{\rightharpoonup }{r}=\left\{{\stackrel{\rightharpoonup }{r}}_1,\cdots ,{\stackrel{\rightharpoonup }{r}}_N\right\}$. When the system potential $V\left(\stackrel{\rightharpoonup }{r}\right)$ is lower than a reference energy E, the modified potential $V^{*}(\stackrel{\rightharpoonup }{r})$ of the system is calculated as:

$$\begin{gathered} V^{*}(\stackrel{\rightharpoonup }{r})=V(\stackrel{\rightharpoonup }{r})+\Delta V(\stackrel{\rightharpoonup }{r}), \\ \Delta V(\stackrel{\rightharpoonup }{r})=\{\begin{array}{cc}\hfill \frac{1}{2}k{\left(E-V(\stackrel{\rightharpoonup }{r})\right)}^2,& V(\stackrel{\rightharpoonup }{r})<E\\ \hfill 0,& V(\stackrel{\rightharpoonup }{r})\ge E\end{array} \end{gathered}$$ (1)

where k is the harmonic force constant. The two adjustable parameters E and k are automatically determined based on three enhanced sampling principles(36). The reference energy needs to be set in the following range:

$$V_{\mathit{max}}\le E\le V_{\mathit{min}}+\frac{1}{k},$$ (2)

where V_max and V_min are the system minimum and maximum potential energies. To ensure that Eqn. (2) is valid, k has to satisfy: $k\le \frac{1}{V_{max}-V_{min}}$ Let us define $\equiv k_0\frac{1}{V_{max}-V_{min}}$, then 0 < k₀ ≤ 1. The standard deviation of ΔV needs to be small enough (i.e., narrow distribution) to ensure proper energetic reweighting(53): σ_ΔV = k(E − V_avg)σ_V ≤ σ₀ where V_avg and σ_V are the average and standard deviation of the system potential energies, σ_ΔV is the standard deviation of ΔV with σ₀ as a user-specified upper limit (e.g., 10k_BT) for proper reweighting. When E is set to the lower bound E=V_max, k₀ can be calculated as:

$$k_0=\min (1.0,k_0^{\prime })=\min (1.0,\frac{{\sigma }_0}{{\sigma }_V}\frac{V_{\mathit{max}}-V_{\mathit{min}}}{V_{\mathit{max}}-V_{\mathit{avg}}}).$$ (3)

Alternatively, when the threshold energy E is set to its upper bound $E=V_{min}+\frac{1}{k}$, k₀ is set to:

$$k_0=k_0^{″}\equiv (1-\frac{{\sigma }_0}{{\sigma }_V})\frac{V_{\mathit{max}}-V_{\mathit{min}}}{V_{\mathit{max}}-V_{\mathit{avg}}},$$ (4)

if $k_0^{″}$ is found to be between 0 and 1. Otherwise, k₀ is calculated using Eqn. (3).

Similar to aMD, GaMD provides schemes to add only the total potential boost ΔV_P, only dihedral potential boost ΔV_D, or the dual potential boost (both ΔV_P and ΔV_D). The dual-boost simulation generally provides higher acceleration than the other two types of simulations(54). The simulation parameters comprise of the threshold energy E for applying boost potential and the effective harmonic force constants, k_0P and k_0D for the total and dihedral potential boost, respectively.

Energetic Reweighting of GaMD Simulations

To calculate potential of mean force (PMF)(55) from GaMD simulations, the probability distribution along a reaction coordinate is written as p*(A). Given the boost potential $∆V\left(\stackrel{\rightharpoonup }{r}\right)$ of each frame, p*(A) can be reweighted to recover the canonical ensemble distribution, p(A), as:

$$p\left(A_j\right)=p^{*}\left(A_j\right)\frac{{\langle e^{\beta \Delta V(\stackrel{\rightharpoonup }{r})}\rangle }_j}{{\sum }_{i=1}^M{\langle p^{*}(A_i)e^{\beta \Delta V(\stackrel{\rightharpoonup }{r})}\rangle }_i},j=1,\dots ,M,$$ (5)

where M is the number of bins, β = k_BT and ${〈e^{\beta ∆V\left(\stackrel{\rightharpoonup }{r}\right)}〉}_j$ is the ensemble-averaged Boltzmann factor of $∆V\left(\stackrel{\rightharpoonup }{r}\right)$ for simulation frames found in the j^th bin. The ensemble-averaged reweighting factor can be approximated using cumulant expansion:

$$\langle e^{\beta \Delta V(\stackrel{\rightharpoonup }{r})}\rangle =\mathit{exp}\left\{{\sum }_{k=1}^{\infty }\frac{{\beta }^k}{k!}{C}_k\right\},$$ (6)

where the first two cumulants are given by

$$\begin{gathered} C_1=\langle \Delta V\rangle , \\ {C}_2=\langle \Delta V^2\rangle -{\langle \Delta V\rangle }^2={\sigma }_v^2. \end{gathered}$$ (7)

The boost potential obtained from GaMD simulations usually follows near-Gaussian distribution. Cumulant expansion to the second order thus provides a good approximation for computing the reweighting factor(36, 53). The reweighted free energy F(A) = −k_BT ln p(A) is calculated as:

$$F(A)=F^{*}(A)-{\sum }_{k=1}^2\frac{{\beta }^k}{k!}{C}_k+F_c,$$ (8)

where F*(A) = −k_BT ln p*(A) is the modified free energy obtained from GaMD simulation and F_c is a constant.

Lipid -S_CD Order Parameter

The -S_CD order parameter measures orientational anisotropy of the C-H bond in sn-2 acyl chains of lipids that are usually obtained from NMR experiments. It is a function of the angle between the C-H bond and lipid bilayer normal. It is defined by following equation:

$${\mathrm{S}}_{\mathrm{CD}}=\frac{1}{2}\langle 3{\mathit{cos}}^2\theta -1\rangle ,$$ (9)

where θ is the angle between the bilayer normal and C-H bond and 〈…〉 denotes an ensemble average. Here, the -S_CD order parameter is averaged over all the lipids in the system and all the frames in the simulation trajectory. The -S_CD order parameter calculated from GaMD simulations is not reweighted due to complexity of the function. However, since GaMD maintains the overall shape of the original potential energy surface(36), the resulting order parameter is found to be close to the experimental values(56) (see Results). The -S_CD value usually ranges from −0.25 to 0.5, with 0.5 for the C-H bond being fully ordered along the bilayer normal and −0.25 being parallel to the bilayer plane. The -S_CD approximates the mobility of each C-H bond and hence estimates the membrane fluidity.

System Setup

The cryo-EM structure of the ADO-A₁AR-Gi complex (PDB: 6D9H(45)) and X-ray structure of PSB36-A₁AR complex (PDB: 5N2S(46)) were used to prepare the simulation systems. As helix 8 region was missing in the crystal structure of PSB36-A₁AR, atomic coordinates were added using another X-ray structure of the inactive A₁AR (PDB: 5UEN(47)) after aligning the receptor TM domain. All chain termini were capped with neutral groups, i.e. the acetyl group (ACE) for the N-terminus and methyl amide group (CT3) for C terminus. Protein residues were set to the standard CHARMM protonation states at neutral pH with the psfgen plugin in VMD(57). Then the receptor was inserted into a POPC bilayer with all overlapping lipid molecules removed using the Membrane plugin in VMD(57). The system charges were then neutralized at 0.15 M NaCl using the Solvate plugin in VMD(57). Periodic boundary conditions were applied on the simulation systems. The simulation systems of the active and inactive A₁AR systems are summarized in Table 1.

Table 1:

Summary of GaMD simulations performed on the adenosine A₁ receptor (A₁AR).

System	aNatoms	Dimension (Å3)	bMethod	Simulation	cΔVavg (kcal/mol)	dσΔv (kcal/mol)
PSB36-A1AR	77,809	93x100x101	GaMD_Dih	300 ns x 2	4.47	1.81
			GaMD_Dual	300 ns x 2	8.45	3.33
ADO-A1AR-Gi	180,394	93x111x167	GaMD_Dih	300 ns x 2	5.04	2.22
			GaMD_Dual	150 ns x 2	9.94	2.57

^aN_atoms is number of atoms in the simulation systems.

^bGaMD_Dih and GaMD_Dual represent the dihedral and dual boost GaMD simulations respectively.

^cΔV_avg and ^dσ_ΔV are the average and standard deviation of the GaMD boost potential.

Simulation Protocol

The CHARMM36 parameter set(58) was used for the protein and POPC lipids. For agonist ADO and antagonist PSB36, the force field parameters were obtained from the CHARMM ParamChem web server(59, 60). Initial energy minimization and thermalization of the A₁AR system follow the same protocol as used in the previous GPCR simulations(61). The simulation proceeded with equilibration of lipid tails. With all the other atom fixed, the lipid tails were energy minimized for 1000 steps using the conjugate gradient algorithm and melted with constant number, volume, and temperature (NVT) run for 0.5ns at 310 K. Each system was further equilibrated using constant number, pressure, and temperature (NPT) run at 1 atm and 310 K for 10 ns with 5 kcal (mol Ų)⁻¹ harmonic position restraints applied to the protein. Further equilibration of the systems was performed using an NPT run at 1 atm and 310 K for 0.5ns with all atoms unrestrained. Conventional MD simulation was performed on each system for 10 ns at 1atm pressure and 310 K with a constant ratio constraint applied on the lipid bilayer in the X-Y plane. The GaMD simulations were carried out using NAMD2.13(52, 62). Both dihedral and dual-boost GaMD simulations were then performed to study the protein-membrane interactions in the inactive and active A₁AR systems (Table 1). In the GaMD simulations, the threshold energy E for adding boost potential is set to the lower bound, i.e. E = V_max(36, 52). The simulations included 50ns equilibration after adding the boost potential and then multiple independent production runs lasting 150 – 300 ns with randomized initial atomic velocities. GaMD production simulation frames were saved every 0.2ps for analysis.

Simulation analysis

The VMD(57) and CPPTRAJ(63) tools were used for trajectory analysis. In particular, distance was calculated between the Cα atoms residues Arg^3.50 and Glu^6.30. Root-mean-square fluctuations (RMSFs) were calculated for the protein residues and ligands, averaged over two independent GaMD simulations and color coded for schematic representation of each complex system. MEMBPLUGIN, a plugin for the VMD package was used to calculate the -S_CD order parameter for POPC lipid tails(64). The -S_CD order parameters were averaged over all lipids and frames of the two independent GaMD simulations for each system. The CPPTRAJ tool was used to calculate the correlation matrices. The Cα atoms of the receptor and phosphorous atoms in the POPC lipid head groups were used for the calculations. In addition to the phosphorous atom, the C₈ and C₁₈ atoms representing different regions of the lipids were also used to calculate dynamic correlations with the receptor. The PyReweighting toolkit(53) was applied to reweight GaMD simulations for free energy calculations by combining independent trajectories for each system. A bin size of 1 Å was used for the Arg^3.50-Glu^6.30 distance and 1 for the number of lipids. The cutoff was set to 500 for calculating the 2D PMF profiles.

Results

Structural flexibility of the A₁AR depended on the receptor conformational state

All-atom GaMD simulations were performed on two different conformational states of the A₁AR, active (ADO-A₁AR-Gi) and inactive (PSB36-A₁AR) states (Table 1). For the inactive A₁AR system, the boost potential was 4.47±1.81 kcal/mol and 8.45±3.33 kcal/mol in dihedral and dual-boost GaMD simulations, respectively. For the active A₁AR system, the boost potential was 5.04±2.22 kcal/mol and 9.94±2.57 kcal/mol in dihedral and dual-boost GaMD simulations, respectively (Table 1). Thus, dual-boost GaMD provided higher acceleration in the simulations with greater boost potential. In the dihedral GaMD simulations of the inactive A₁AR, the TM helices of the receptor were rather rigid. Only the intracellular end of TM6, the terminus of helix 8 (H8), extracellular end of TM1 and extracellular loop 2 (ECL2) regions were flexible (Figure 1A). Similar results were obtained for the active A₁AR in the dihedral GaMD simulations (Figure 1B). However, the intracellular ends of TM6 and TM5 of the A₁AR exhibited more fluctuations in the active state compared to the inactive state. The ECL2 region was relatively more flexible in the active A₁AR than in the inactive A₁AR. In both systems, the ligands remained stably bound at the orthosteric site throughout the simulations. In comparison, the G protein coupled to the active A₁AR exhibited higher fluctuations. In particular, C terminus of the α5 helix, α4-β5 loop and α4-β6 loop of the Gα subunit and terminal ends of the G_βγ subunits exhibited high fluctuations up to 3 Å. Similar results were also found in the dual-boost GaMD simulations of the inactive and active A₁AR systems (Figure S1).

Figure 1:

Comparison of structural flexibility of the inactive and active A1AR systems obtained from dihedral GaMD simulations: (A) Root-mean-square fluctuations (RMSFs) of the inactive PSB36-A1AR complex. (B) RMSFs of the active ADO-A1AR-Gi protein complex. A color scale of 0 Å (blue) to 3 Å (red) was used.

Lipids in the lower leaflet of the active A₁AR system showed higher fluidity than in the inactive-A₁AR system

The lipid -S_CD order parameters were calculated for the upper (extracellular) and lower (cytoplasmic) leaflets from GaMD simulations of the inactive and active A₁AR systems (Figure 2). In both inactive and active A₁AR systems, lower leaflet was more fluid than the upper leaflet with smaller -S_CD order parameters. The lower leaflet exhibited significant differences between the inactive and active A₁AR systems. In particular, the -S_CD order parameter of the fifth carbon atom in POPC was ~0.20 in the lower leaflet in the inactive A₁AR system (Figure 2A, 2C), but decreased to ~0.17 in the active A₁AR system (Figure 2B, 2D). This indicated higher inclination of C-H bonds being ordered along the bilayer normal in the lower leaflet of the active A₁AR system. This appeared to correlate with the outward movement of TM6 as the A₁AR changed from the inactive to active state. In the inactive A₁AR, the R131^3.50-E268^6.30 distance at the free energy minimum was ~7 Å (Figure 3A, 3C). In comparison, this distance at the free energy minimum increased to ~17 Å in the active A₁AR (Figure 3B, 3D). The lateral movement of the TM6 could push the surrounding lipids. Higher flexibility of the receptor TM6 intracellular end was accompanied by increased fluidity of lipids in the lower leaflet of membrane. Similar results were found from the dihedral and dual-boost GaMD simulations (Figure 3). Therefore, the structural conformation and flexibility of the GPCR are strongly coupled with the surrounding membrane lipids.

Figure 2:

The -S_CD order parameters calculated for sn-2 acyl chains of POPC lipids in different simulation systems: (A) Inactive A1AR using dihedral-boost GaMD, (B) Active A1AR using dihedral-boost GaMD, (C) Inactive A1AR using dual-boost GaMD and (D) Active A1AR using dual-boost GaMD. Red diamond lines represent the average -S_CD order parameters for the cytoplasmic lower leaflet and blue diamond lines for the extracellular upper leaflet.

Figure 3:

Free energy profiles of the extracellular upper leaflet of membrane in different simulation systems regarding the number of lipids within 5 Å of the receptor TM6 and the receptor R3.50 – E6.30 distance: (A) Inactive A₁AR using dihedral-boost GaMD, (B) Active A₁AR using dihedral-boost GaMD, (C) Inactive A₁AR using dual-boost GaMD and (D) Active A₁AR using dual-boost GaMD. The R3.50 – E6.30 distance is ~7 Å in the inactive A1AR and increases to ~17 Å in the active A1AR due to outward movement of TM6.

The inactive A₁AR attracted more lipids in the upper leaflet to the TM6 than the active A₁AR

Considering significant conformational changes (especially in the TM6) during receptor activation, we hypothesized that the number of lipids interacting with the active and inactive receptor are different. In order to identify the low-energy states of the membrane-receptor interactions, potential of mean force (PMF) profiles were calculated by reweighting the GaMD simulations (Figure 3). The R131^3.50-E268^6.30 distance was chosen as one reaction coordinate to characterize activation of the GPCR. The number of POPC phosphate head groups within 5 Å of TM6 was calculated as the other reaction coordinate (Figures S2 and S3). In the upper leaflet, approximately one lipid molecule was found interacting with TM6 in the inactive A₁AR (Figure 3A, 3C). But no lipid in the upper leaflet was found within 5 Å of TM6 in the active A₁AR (Figure 3B, 3D). Further analysis revealed that one positively-charged residue was located in the receptor ECL3 (K265^ECL3), being close to the extracellular end of the TM6 (Figure 5A). In the inactive A₁AR, this lysine pointed towards the lipid membrane and thus attracted the negatively-charged phosphate head group of a POPC molecule. Instead, the positively-charged side chain of K265^ECL3 formed a stable salt-bridge with negatively-charged glutamate (E172^ECL2) of ECL2 in the active A₁AR (Figure S4). Residue K265^ECL3 did not interact with the lipid in the active A₁AR. Therefore, the inactive A₁AR interacted with more phospholipids in the upper leaflet compared to the active A₁AR.

Figure 5:

Minimum energy states of POPC lipid interacting with the positively-charged lysine residues in TM6 of the receptor obtained from dihedral GaMD simulations. (A) One POPC molecule in the upper leaflet interacts with one lysine residue (K265^ECL3) of the inactive A₁AR. (B) Three POPC molecules (POPC1, POPC2, POPC3) in the lower leaflet interact with four Lysine residues (K263^6.25, K267^6.29, K270^6.32 and K273^6.35) of the active A₁AR. The receptor TM6 is colored in gray.

The active A₁AR attracted more lipids in the lower leaflet to the TM6 than the inactive A₁AR

In contrast to the upper leaflet, the lower leaflet had more lipids within 5 Å of TM6 in the active A₁AR than in the inactive system (Figure 4). In the lowest energy state, the inactive A₁AR interacted with approximately two lipids within 5 Å of TM6 (Figure 4A, 4C). In comparison, the active A₁AR exhibited a relatively broader energy well. The TM6 intracellular domain interacted with ~2-4 lipid molecules (Figure 4B, 4D). Upon activation of the A₁AR, the TM6 moved outwards by ~10 Å and exposed its positively-charged residues to the membrane. Therefore, the lipids that diffused in the lower leaflet of the membrane interacted more frequently with the receptor. Fewer lipids interacted with the inactive A₁AR because the receptor has a narrower curvature in the TM6 intracellular region. Moreover, the lower leaflet had a larger number of lipids interacting with receptor TM6 than the upper leaflet. Out of five positively-charged residues in TM6, four were located in the intracellular region (K224^6.25, K228^6.29, K231^6.32 and K234^6.35) (Figure 5B). The negatively-charged phosphate head groups of POPC tended to interact with these positively-charged lysine residues to stabilize the active receptor conformation. Therefore, the lipid-GPCR interaction should play an important role in the conformational changes during activation of the A₁AR.

Figure 4:

Free energy profiles of the cytoplasmic lower leaflet of membrane in different simulation systems regarding the number of lipids within 5 Å of the receptor TM6 and the receptor R3.50 – E6.30 distance: (A) Inactive A₁AR using dihedral-boost GaMD, (B) Active A₁AR using dihedral-boost GaMD, (C) Inactive A₁AR using dual-boost GaMD and (D) Active A₁AR using dual-boost GaMD. The R3.50 – E6.30 distance is ~7 Å in the inactive A1AR and increases to ~17 Å in the active A1AR due to outward movement of TM6.

GaMD simulations revealed strongly coupled dynamics between the GPCR and membrane lipids

Dynamic correlations were identified between residues of the A₁AR and lipids in both the upper and lower leaflets. The Cα atoms in the receptor residues and the phosphorous atoms in the lipid head groups were used to calculate the correlation matrices (see details in Methods). Similar results were obtained using the C₈ and C₁₈ atoms in the lipid hydrophobic tails to calculate the dynamic correlation matrices (Figure S5). In all the simulation systems, motions of the receptor N-terminus, ECL1, ECL2 and ECL3 regions were positively correlated to those of lipids in upper leaflet (Figure 6). Similarly, motions of the receptor ICL1, ICL2 and ICL3 were positively correlated to those of lipids in the lower leaflet (Figure 7). For the inactive A₁AR, most TM helix residues exhibited negatively correlated motions with the lipids. In this regard, the TM helices appeared to move in the opposite direction relative to the lipids in the inactive A₁AR simulation system (Figure 6A, 6C, 7A and 7A). For the active A₁AR, correlations between TM helix residues and the lipids in both the upper and lower leaflets were very weak, being close to zero (Figure 6B, 6D, 7B and 7D). However, marked positive correlations were identified between the intracellular region of the receptor TM6 and lipids in the lower leaflet (Figures 7B and 7D). Therefore, the TM6 intracellular region of the active A₁AR appeared to move in the same direction with the surrounding lipids. This was highly consistent with the simulation finding that significantly more lipids were found within 5 Å of the TM6 intracellular domain in the active A₁AR system (Figure 4) and they formed remarkably stronger electrostatic interactions (Figure 5) compared with the inactive A₁AR system. In summary, the GaMD simulations revealed strongly coupled dynamics between the GPCR and membrane lipids.

Figure 6:

Dynamic correlation matrices calculated for lipids in the extracellular upper leaflet with residues in the A1AR in different simulation systems: (A) Inactive A₁AR using dihedral-boost GaMD, (B) Active A₁AR using dihedral-boost GaMD, (C) Inactive A₁AR using dual-boost GaMD and (D) Active A₁AR using dual-boost GaMD. The Cα atoms of the receptor and phosphorous atoms in the lipid head groups were used for calculating the correlation matrices here. Similar results were obtained using the C₈ and C₁₈ atoms in the lipid hydrophobic tails as shown in Figure S5. The receptor ICL1, ICL2 and ICL3 represent intracellular loops between TM helices 1-2, 3-4, and 5-6 respectively. Similarly, the receptor ECL1, ECL2 and ECL3 represent extracellular loops between TM helices 2-3, 4-5, and 6-7 respectively.

Figure 7:

Dynamic correlation matrices calculated for lipids in the intracellular lower leaflet with residues in the A1AR in different simulation systems: (A) Inactive A₁AR using dihedral-boost GaMD, (B) Active A₁AR using dihedral-boost GaMD, (C) Inactive A₁AR using dual-boost GaMD and (D) Active A₁AR using dual-boost GaMD. The Cα atoms of the receptor and phosphorous atoms in the lipid head groups were used for calculating the correlation matrices here. Similar results were obtained using the C₈ and C₁₈ atoms in the lipid hydrophobic tails as shown in Figure S5. The receptor ICL1, ICL2 and ICL3 represent intracellular loops between TM helices 1-2, 3-4, and 5-6 respectively. Similarly, the receptor ECL1, ECL2 and ECL3 represent extracellular loops between TM helices 2-3, 4-5, and 6-7 respectively.

Discussion

In this study, we have applied all-atom GaMD simulations to investigate GPCR-membrane interactions, using the A₁AR as a model receptor. In the GaMD simulations, the inactive and active A₁AR showed different structural flexibility profiles. The ECL2 region, intracellular ends of TM6 and TM5 exhibited higher fluctuations in the active A₁AR compared to the inactive A₁AR. The receptor TM domain was rigid and the ligands remained tightly bound at the orthosteric site. However, the G protein coupled to the active A₁AR exhibited high flexibility during the simulations, especially in the α5 helix, α4-β5 loop and α4-β6 loop of the Gα subunit and terminal ends of the G_βγ subunits. These results were consistent to our earlier simulation findings of AR-G protein complexes(65).

The -S_CD order parameter values obtained from GaMD simulations were consistent with experimental data. In NMR experiments, the -S_CD order parameter for the fifth carbon C-H bond of POPC was observed to be at ~0.18-0.20(56). The -S_CD order parameter of POPC’s fifth carbon atom was ~0.20±0.02 in the lower leaflet in the inactive A₁AR system. It decreased to ~0.17±0.02 in the active A₁AR system. The -S_CD order parameter of the ninth carbon C-H bond in POPC was measured as ~0.10 in NMR experiments(56), for which the same value was obtained from GaMD simulations. Furthermore, the GaMD simulations showed that POPC lipids in the lower leaflet of the active A₁AR system were more fluid than in the inactive A₁AR system. The -S_CD order parameters for the lower leaflet in the active A₁AR system were smaller than those in the inactive A₁AR system. This finding correlated with the outward movement of TM6 in the active A₁AR, which caused higher inclination of the C-H bonds to be aligned along the bilayer normal. The smaller -S_CD order parameters suggested higher membrane fluidity in the lower leaflet of the active A₁AR system.

In the GaMD simulations, the inactive A₁AR attracted more lipids in the upper leaflet than the active A₁AR. The membrane facing positively-charged lysine residue (K265^ECL3) interacted with the negatively-charged phosphate head group of POPC. In contrast, this lysine pointed towards ECL2 in the active A₁AR. This was consistent with our previous study⁽⁶⁶⁾, in which the positive allosteric modulator (PAM) enhanced the agonist binding at the orthosteric site by forming a salt-bridge between E172^ECL2-K265^ECL3. Moreover, the active A₁AR attracted more lipids in the lower leaflet compared with the inactive A₁AR. When exposed to the membrane, the positively-charged residues in intracellular region of TM6 of the A₁AR interacted with negatively-charged head groups. This was further verified by the correlation matrix. The only positive correlation between the transmembrane helices and the lipids was observed between the intracellular region of TM6 and lipids in the lower leaflet of active A₁AR. Four lysine residues (K263^6.25, K267^6.29, K270^6.32 and K273^6.35) present in the intracellular end in the β₂AR of TM6 were also known to interact with negatively-charged headgroups of 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE) lipid molecules and thus stabilize the GPCR active state(27).

In summary, all-atom GaMD simulations have revealed strongly coupled dynamics between a GPCR and the membrane lipids that depend on the receptor activation state. The GaMD method has greatly enhanced sampling of the lipid-protein interactions, which would take significantly longer simulation time using cMD. Nonetheless, the activation or deactivation conformational transitions of the GPCR were not observed in the presented GaMD simulations. Longer GaMD simulations (e.g., microseconds) are expected to capture such conformational transitions (67) and the related effects of lipid-receptor interactions will be investigated in the future. Furthermore, the effects different lipid types (e.g., cholesterol, PIP₂, etc.) on GPCR-membrane interactions are subject to future studies. It is important to study specific lipid interactions with GPCRs during the receptor activation. Developments of enhanced sampling methodologies and computing power would aid to further address these challenges.