Biased ligands of G protein-coupled receptors (GPCRs) may have improved therapeutic benefits and safety profiles. However, the molecular mechanism of GPCR biased signaling remains largely unknown. Using apelin receptor (APJ) as a model, we systematically investigated the potential effects of amino acid residues around the orthosteric binding site on biased signaling. We discovered that a single residue mutation I109A (I1093.32) in the transmembrane domain 3 (TM3) located in the deep ligand-binding pocket was sufficient to convert a balanced APJ into a G protein signaling biased receptor. APJ I109A mutant receptor retained full capabilities in ligand binding and G protein activation, but was defective in GRK recruitment, β-arrestin recruitment, and downstream receptor-mediated ERK activation. Based on molecular dynamics simulations, we proposed a molecular mechanism for biased signaling of I109A mutant receptor. We postulate that due to the extra space created by I109A mutation, the phenyl group of the last residue (Phe-13) of apelin rotates down and initiates a cascade of conformational changes in TM3. Phe-13 formed a new cluster of hydrophobic interactions with the sidechains of residues in TM3, including F1103.33 and M1133.36, which stabilizes the mutant receptor in a conformation favoring biased signaling. Interruption of these stabilizing interactions by double mutation F110A/I109A or M113A/I109A largely restored the β-arrestin-mediated signaling. Taken together, we describe herein the discovery of a biased APJ mutant receptor and provide detailed molecular insights into APJ signaling selectivity, facilitating the discovery of novel therapeutics targeting APJ.
G protein-coupled receptors (GPCRs) are the largest group of cell membrane proteins responsible for transducing signals from extracellular stimuli to intracellular responses through multiple downstream effectors, such as G proteins and β-arrestins. They are the most important group of target proteins for drug discovery, accounting for more than 30% of all currently marketed therapeutics [1,2]. Most GPCR agonists activate G protein and β-arrestin signaling pathways equally (i.e. balanced agonists). However, a few GPCR agonists preferentially activate either G protein or β-arrestin pathways (i.e. biased agonists) [3,4]. It is perceived that biased activation confers therapeutic benefits, for example, prolonged efficacy and/or reduced side effects [5–9]. Indeed, a few biased GPCR agonists have progressed to human clinical trials. Most notably, G protein biased µ-opioid receptor agonists are currently under development potentially as new generation of analgesics with significant improved therapeutic profile and reduced side effects such as respiratory suppression and constipation [7,9].
All GPCRs share a common architecture with seven transmembrane domains. Recent advances in crystallography, spectroscopy, and molecular modeling indicate that GPCRs could take on multiple conformational states, which have different implications for various downstream signaling pathways . The function of GPCRs relies essentially on their ability to change conformation upon ligand binding, to cascade such changes through the transmembrane domains and to switch between distinct conformational states [10,11]. However, how ligand-binding influences GPCR conformational states and how the conformational changes transit to downstream functional selectivity remains unclear. The limited understanding of structure–function relationships of biased agonism was mostly obtained from spectroscopic characterization of conformational changes induced by biased ligands or co-crystal structures of GPCR bound to biased ligands [9,12]. One significant challenge of such studies is the prerequisite of biased ligands. Alternatively, modifying the receptor structure through mutagenesis to generate functional bias offers a different perspective and complementary approach in understanding the relationship between receptor conformation and downstream signaling [13,14]. Herein, we describe the identification of G protein signaling biased apelin receptor (APJ) mutant as an example to elucidate how GPCR structural changes translate into G protein biased signaling.
APJ is a class A GPCR with high sequence similarity to the angiotensin receptor AT1R  and plays important roles in a variety of physiological processes such as heart contractility, energy metabolism, and fluid homeostasis. It is also known that APJ is involved in angiogenesis and can facilitate responses to immunotherapy [16,17]. Hence, it has emerged as an attractive therapeutic target for the treatment of cardiovascular and metabolic diseases and cancers [17–20]. Recently, we published the high-resolution crystal structure of APJ in complex with a peptide agonist, wherein we proposed a ‘two-site’ ligand-binding mode and identified critical residues involved in ligand interaction at the orthosteric site . In this report, we describe the discovery and comprehensive characterization of a biased mutant receptor APJ I109A and elucidate the molecular mechanism of signal transduction using molecular dynamics (MD) simulation. The results presented here provide a general understanding of GPCR biased signaling and also shed light on rational design for biased ligands as potential therapeutics [22,23]. Furthermore, biased mutant receptor APJ I109A could serve as a valuable pharmacological tool to investigate the contribution of β-arrestin signaling in APJ physiological functions.
Dulbecco's modified Eagles medium (DMEM), fetal bovine serum (FBS), Lipofectamine LTX, 0.25% Trypsin-EDTA and other cell culture reagents were purchased from Life Technologies. Coelenterazine was purchased from Cayman. Apelin peptides were synthesized by Chinese Peptide Ltd. Alexa Fluor 488 and Alexa Fluor 594 conjugated anti-HA mouse IgG were purchased from Life Technologies. Lance Ultra cAMP kit and p-ERK 1/2 (Thr202/Tyr204) Assay Kits were purchased from PerkinElmer. Nanoluc AssayTM kits were purchased from Promega.
Generation of plasmids and BacMam virus
All gene synthesis, cloning, and mutagenesis were done by GenScript and confirmed by DNA sequencing. Lyn-rGFP, FYVE-rGFP, and β-arrestin 2-rLuc sequences were synthesized, as described previously . GRK2-rLuc DNA was synthesized by in-frame fusing rLuc sequence to the 3-prime end of human GRK2 encoding sequence. Lyn-rGFP and FYVE-rGFP sequences were then cloned into the pIRES(puro+) vector, while β-arrestin 2-rLuc and GRK2-rLuc sequences were cloned into the pIRES(hygro+) vector. APJ mutant bacmids were generated by PCR-based site-directed mutagenesis using pJiF1.1/APJ as the template. The resulting bacmids were used to generate BacMam virus by infecting insect SF9 cells according to the standard protocol . Full-length sequences of arrestin 2-Lgbit and APJ-Smbit were synthesized as previously described  and cloned into pIRES(hygro+) or pIRES(puro+) vector, respectively.
Stable cell line generation and cell transfection
Unless otherwise specified, HEK293FT cells were used for all stable cell line generation. Briefly, HEK293FT cells in six-well plate were transfected with plasmids using Lipofectamine LTX according to the manufacturer's protocol. Forty-eight hours after transfection, cells were selected with DMEM medium supplemented with 10% FBS and either 500 µg/ml Hygromycin (for pIRES(hygro+) vector) or 1 µg/ml Puromycin (for pIRES (puro+) vector). Cells stably expressing BRET (bioluminescence resonance energy transfer) pair (rGFP and rLuc fusion proteins) were selected and maintained in DMEM medium supplemented with 500 µg/ml hygromycin and 1 µg/ml puromycin. For receptor transient expression, HEK293FT cells were infected with recombinant APJ or APJ mutant BacMam virus. Briefly, cells were cultured in FreeStyle™ 293 Expression medium in suspension and grown to a density of 1–2 × 106 cells/ml before transduction. The cells were diluted to a density of 0.4 × 106 cells/ml and infected with BacMam virus at a multiplicity of infection of 300. Infected cells were incubated at 37°C with slow shaking (130 rpm) for 24 h before harvested for functional assay or membrane preparation. Cell surface receptor expression was checked by flow cytometry using AlexaFluor 488 conjugated anti-HA antibody.
Activation of APJ was measured based on intracellular cAMP inhibition using Lance Ultra cAMP kit according to the manufacturer's protocol. Briefly, 10 µl of 2000 cells in assay buffer (HBSS buffer pH7.4 with 5 mM HEPES, 0.1% BSA, and 0.5 mM IBMX) was added to each well of 96-well half area assay plate, and stimulated with 10 µl of various concentrations of apelin 13 (AP-13) in the presence of 2.5 µM forskolin for 30 min at 37°C. Then 10 µl of 4× Eu-cAMP and 10 µl of 4× Ulight-Anti-cAMP working solutions were added to each well sequentially. The assay plate was incubated at room temperature for 60 min before reading with an Envision plate reader using configurations recommended by the manufacturer.
Membrane preparation for ligand binding
HEK293FT suspension cells were pelleted and washed with PBS once, followed by low speed centrifugation (800×g) to remove PBS. Cell pellets were resuspended in cold cell lysis buffer (25 mM HEPES, 1 mM EDTA, 2 mM MgCl2, and 1× protease inhibitor) and incubated on ice for 10 min. The suspension was homogenized using a glass homogenizer with 30–40 strokes. The homogenates were then centrifuged at 650×g for 5 min, and the resulting supernatants were centrifuged again at 12 000×g for 30 min. The crude membrane pellets were resuspended in cell lysis buffer with 10% sucrose and passed through a 27-gauge needle three times. The membrane preparations were aliquoted and stored at −80°C freezer. The membrane protein concentration was determined with a BCA kit using BSA as the standard.
Radioligand binding assay
Radioligand binding assay was performed in a 96-well filtration plate. Wild-type (WT) or mutant APJ membranes (3 µg) were incubated with various concentrations of 125I-AP-13 in the ligand binding buffer (50 mM HEPES, pH 7.4, 5 mM MgCl2, 1 mM CaCl2, 0.2% BSA) in 120 µl. Following incubation at room temperature for 2 h with gentle shaking, the reaction mixtures were then transferred to UniFilter GF/B filtration Plate (PEI Coated, PE). The plate was washed three times immediately with ice-cold washing buffer (50 mM HEPES, pH 7.4, 500 mM NaCl, 0.1% BSA) using PerkinElmer FilterMate™ Universal Harvester. After the plates were dried at 37°C for 2 h, scintillation cocktail (50 µl) was added to each well, and radioactivity was measured by MicroBeta Trilux. Nonspecific binding was determined in the presence of 100 nM cold AP-13 in the reaction mixture. All data were fit to one-site binding model using GraphPad Prism 7 software, and receptor expression level was calculated from the best-fit Bmax value of the specific binding.
Stable cells expressing both rGFP and rLuc fusion proteins were seeded into 6-well plates at a density of 800 000 cells per well and transfected with APJ receptor plasmids or infected with APJ BacMam virus. After 24 h transfection, cells were suspended in DMEM supplemented with 10% FBS and split into poly-l-ornithine coated 96-well plates at a density of 20 000 cells per well for overnight growth. After the removal of growth medium, cells were then treated with 50 µl of agonists diluted in Opti-MEM for 30 min before 50 µl of diluted substrate Coelenterazine (10 µM in Opti-MEM) was added to each well. BRET signals were measured using a CLARIO Star microplate reader (BMG LABTECH) with a filter set of 400/80 nm (donor) and 510/30 nm (acceptor). Raw BRET ratio was determined by calculating the ratio of the light intensity emitted by the rGFP over the light intensity emitted by the rLuc .
HEK293FT/Lgbit-β-arrestin 2 stable cells were transiently transfected with APJ-Smbit or APJ I109A-Smbit plasmids. After 24 h transfection, the cells were suspended in Opti-MEM supplemented with 10% FBS and seeded into a 384-well plate for overnight growth before kinetic analysis with FLIPR Tetra (Molecular Device). Before the compound treatment, 6 µl nanoluc reagent (dilute the substrate for 20-fold with buffer according to the product manual) was added to each well. The basal luminescence signal was recorded for 60 sec by FLIPR before 3 µl of 10× compounds solution was added to each well. The signals were recorded every 15 sec for 15 min to generate a time-course curve .
U2OS cells stably expressing β-arrestin2-GFP were seeded into a 6-well plate at a density of 80 000 cells/well and grown overnight. Attached cells were transduced with BacMam virus of APJ or APJ I109A mutants. After 6 h virus transduction, cells were suspended and seeded at a density of 30 000 cells/well into 8-well chamber dishes and grown overnight before experiment. Cell surface receptors were stained with 10 µg/ml Alex Fluor 594 conjugated anti-HA antibody at room temperature for 1 h. The cells were then stimulated with 500 nM apelin in freshly prepared medium. The distributions of APJ receptors and β-arrestin 2 were monitored using a Leica SP8 confocal microscope.
p-ERK 1/2 assay
HEK293FT cells were transduced with APJ BacMam virus in a 6-well plate, as described above. After 6 h virus transduction, cells were seeded into 96-well plates at 20 000 cells/well in DMEM medium supplemented with 10% FBS and grown overnight, then cells were starved for 4 h in starvation medium (DMEM with 1% FBS) before treated with various concentrations of AP-13 for the indicated time. At the end of the treatment, cellular phosphorylated ERK levels were detected using SureFire p-ERK1/2 (Thr202/Tyr204) assay kit according to the manufacturer's procedure (PerkinElmer).
Molecular modeling of APJ in complex with endogenous peptide AP-13
The initial model of the APJ (19–330) in complex with AP-13 was built by homology modeling in Discovery Studio , with the previously published APJ-AMG3054 structure  as the template. The ICL3 loop region was optimized with Prime Refine Loops in Schrödinger . The PPM server  was used to calculate rotational and translational positions of APJ in membranes. A two-step minimization was further performed with Prime in Schrödinger considering implicit membrane environment: only AP-13 and residues within 5 Å of it were initially minimized, then the whole complex was minimized. The minimized model was embedded in the POPC membrane with TIP3P explicit water model and counterions under the OPLS3 force field. Sodium and chloride ions were added at a concentration of 150 mM. MD simulation was performed on the system with Desmond in Schrödinger [29,30]. The system was relaxed before simulation with default protocol. Simulation of 60 ns was performed in the NPT ensemble, at a temperature of 300 K and a pressure of 1 bar. I1093.32, F1103.33 and M1133.36 on APJ WT structure model were mutated to Ala to generate the APJ I109A, I109A/F110A and I109A/M113A models in complex with AP-13. Then, they were minimized, and MD simulation of 60 ns was performed as described above. The distances between the centers of phenyl ring of AP-13 Phe-13 and phenyl ring of APJ F1103.33 and distances between the centers of phenyl ring of AP-13 Phe-13 and sidechain of APJ M1133.36 during MD simulations were calculated by VMD .
All samples were tested in duplicates and data were analyzed using GraphPad Prism v7.0 (GraphPad Software Inc., San Diego, CA), and presented as mean ± SEM of at least three independent experiments. Concentration-response curves for cAMP, BRET, and p-ERK1/2 assays were fit to a non-linear regression (four parameters) model to determine pEC50 with the Prism software. Radioligand binding data were fit to a one-site-specific binding model to determine Kd with the Prism software.
Identification of APJ I109A as a biased receptor
To study the structure–function relationships and to identify critical residues for APJ function, we performed systematic site-directed mutagenesis for residues at or near the orthosteric ligand-binding site of APJ guided by our previously published APJ co-crystal structure . We characterized these mutants using radioligand-binding assay with 125I-labeled AP-13 for binding affinity, cAMP assay for G protein signaling, and the newly established BRET assays for measuring β-arrestin-mediated endocytosis activity [24,32]. While most mutants exhibited comparable profiles to WT receptor in all these assays (Supplementary Table S1), mutations in the ligand-binding pocket, such as Y35A, R168A, D284A, and W85A that were previously reported to disrupt direct receptor-ligand interactions and were compromised in AP-13-induced cAMP response , obliterated β-arrestin-mediated receptor internalization (Figure 1). Interestingly, the isoleucine to alanine mutation at position 109, located near the bottom of ligand-binding pocket, dramatically changed the receptor signaling selectivity. Compared with WT receptor, APJ I109A mutant displayed slightly decreased binding affinity to 125I-labeled AP-13. The Kd values for I109A and WT APJ were 0.15 nM and 0.05 nM, respectively (Figure 2A). It also showed slightly decreased but comparable functional activity in the cAMP assay with an EC50 of 1.15 nM versus an EC50 of 0.57 nM for WT APJ (Figure 2B). However, this mutant receptor was significantly compromised in AP-13 stimulated β-arrestin recruitment to the plasma membrane (Figure 2C) and the subsequent β-arrestin-mediated receptor endocytosis (Figure 2D), indicating that APJ I109A behaves as a signaling biased receptor. In all these experiments, APJ R168A was used as a negative control due to its lack of agonist binding and functional activities.
Alanine scanning of residues located at the APJ orthosteric binding pocket.
Identification of I109A as a critical residue for APJ biased signaling.
Characterization of APJ I109A mutant receptor
To further confirm the effect of I109A mutation on β-arrestin recruitment, we employed a Nanoluc luciferase-based enzyme complementation assay (NanoBit assay) to monitor β-arrestin recruitment to receptor in a time-resolved manner. In this system, the small part of Nanoluc (Smbit) tag was fused to the C-terminal APJ, and the large part of Nanoluc (Lgbit) was fused to the N-terminal β-arrestin. Upon receptor activation, recruitment of β-arrestin to receptor will drive the two split parts of Nanoluc to form a functional enzyme, which can be measured by luminescence signals . In cells expressing APJ, agonist AP-13 stimulation induced a rapid and robust signal increase, which reached plateau ∼6 min, indicating a fast β-arrestin recruitment process. Under the same condition, APJ I109A mutant completely abolished AP-13 stimulated β-arrestin recruitment (Figure 3A).
Characterization of APJ I109A in β-arrestin, GRK2 recruitment, and receptor internalization.
G protein-coupled receptor kinases (GRKs) are considered to play important roles in receptor phosphorylation, which is critical for β-arrestin binding, receptor internalization, and desensitization . Among six members of GRK family, GRK2 is the most predominate isoform expressed in the heart and HEK293 cells . We postulated that deficiency in β-arrestin recruitment of APJ I109A mutant may result from the lack of receptor phosphorylation by GRKs. To test this hypothesis, we established a BRET assay to measure GRK2 recruitment after AP-13 stimulation . Indeed, GRK2 recruitment was almost completely abolished in the APJ I109A mutant (Figure 3B).
Next, we directly monitored agonist-induced β-arrestin recruitment and receptor internalization using confocal microscopy. U2OS cells stably expressing C-terminal GFP-tagged β-arrestin were transiently transfected with HA-tagged receptors using BacMam system. β-arrestin and receptor were visualized with a fluorescence microscope using GFP (green channel) and Alexa-594 conjugated anti-HA antibody (red channel), respectively. In WT APJ expressing cells, the treatment of 100 nM AP-13 caused a quick β-arrestin translocation from cytoplasm to plasma membrane. The APJ/β-arrestin complexes disappeared from the plasma membrane and were internalized into the cytosol to form large intracellular vesicles at 60 min (Figure 3C). However, agonist-stimulated β-arrestin recruitment and receptor internalization were not detected in APJ I109A mutant expressing cells even after agonist treatment for 60 min (Figure 3D), further confirming that APJ I109A mutant is a functionally biased receptor.
To test if signaling bias of APJ I109A is ligand dependent, we measured β-arrestin-mediated receptor endocytosis induced by endogenous ligands of different lengths, apelin (AP-13, AP-17, and AP-36) and elabela (Ela-11, Ela-21, and Ela-32) . While elabela has little sequence similarity to apelin, for all these ligands, receptor internalization in APJ I109A mutant was compromised in comparison with WT receptor (Figure 4A,B). But cAMP response of APJ endogenous ligands was not affected significantly by I109A mutation (Figure 4C,D), indicating that signaling bias of APJ I109A is ligand independent.
Effect of different endogenous ligands on β-arrestin and G protein signaling of APJ I109A mutant.
APJ I109A is defective in β-arrestin-dependent ERK activation
It is well established that APJ activation increases downstream ERK phosphorylation. But it was inconclusive which signaling pathway (Gi or β-arrestin signaling) dominantly mediates ERK activation . APJ I109A provided a pharmacology tool to potentially resolve the issue and to investigate the relationship between β-arrestin signaling and downstream ERK activation. Using the AlphaScreen pERK assay, we measured the time course of ERK phosphorylation upon AP-13 stimulation of WT and mutant receptors. In WT APJ expressing cells, the level of ERK phosphorylation peaked at 20 min. However, in APJ I109A expressing cells, ERK phosphorylation was barely detectable (Figure 5A), suggesting that ERK phosphorylation by WT receptor activation is mediated by the β-arrestin signaling pathway. We also conducted dose–response studies of AP-13 stimulated ERK activation in APJ and APJ I109A mutant expressing cells and observed significant right shift of EC50 values in APJ I109A mutant cells, confirming the critical role of β-arrestin signaling pathway in ERK activation (Figure 5B).
ERK activation defects in APJ I109A mutant.
Molecular modeling for understanding the biased agonism of APJ I109A mutant
To interpret the biased agonism of the APJ I109A mutant at the molecular level, we performed molecular modeling studies on WT APJ and mutant APJ I109A in complex with AP-13 (Figure 6A–E). MD simulations of the two complex models demonstrated that both WT and I109A mutant of APJ can undergo partial activation in the presence of AP-13 with outward movement of TM6 (Figure 6A), which is a common feature in agonist-stimulated GPCR activation. However, in the AP-13 and APJ I109A complex, the Ile to Ala mutation at position 109 creates more space at the bottom of the orthosteric binding site and allows the sidechain phenyl group of Phe-13 in AP-13 to rotate down during the simulation, a phenomenon that is not observable in the AP-13 and WT APJ complex (Figure 6B). This rotation may further induce the conformational change of adjacent residues in TM3 (transmembrane domain 3) of APJ, including F1103.33and M1133.36 (Figure 6B) by forming a cluster of hydrophobic interactions. The formation of such interactions was supported by the simulation results that the distances between sidechains of Phe-13 in AP-13 and F1103.33 or M1133.36 in APJ decreased from ∼9–10 to 5–6 Å at the early stage of the simulation and remained stable thereafter (Figure 6D,E). The adjacent Y2997.43 may also be involved in the cluster of hydrophobic interactions. These interactions may lead to subsequent movement of TM3 and the adjacent TM2 and influence the conformations and movements of TM1 and TM7. As shown in Figure 6C, the above changes with APJ 109A mutant cascade down to conformational arrangements that are different from those of WT receptor in ICL1, N-terminus of helix VIII, and the C-terminal region of TM7, which are known to be directly involved in β-arrestin binding . These conformational differences underpin the structural basis for biased signaling of the APJ I109A mutant receptor.
MD simulations of WT APJ and APJ I109A mutant in complex with AP-13.
To test if the cluster of hydrophobic interactions between Phe-13 of AP-13 and F1103.33, M1133.36 and Y2997.43 is critical for stabilizing I109A mutant in a conformation favoring biased signaling, we created additional mutant receptors containing F110A, M113A, or Y299A. First, we found that single mutation of F110, M113, or Y299 alone had no effect on APJ receptor functions, including cAMP response, β-arrestin recruitment, and endocytosis (Supplementary Table S1). We then generated receptors with double mutations (I109A/F110A, I109A/M113A, and I109A/Y299A) and examined their functions by cAMP and β-arrestin recruitment assays. The results indicated that the double mutations of two residues in TM3 but not Y299A in TM7 retained G protein-signaling capabilities (Figure 7A), and they also largely restored the function of agonist-stimulated β-arrestin recruitment (Figure 7B), confirming the observed involvement of hydrophobic interactions and its importance of F1103.33 and M1133.36 in maintaining conformational states favoring biased signaling for APJ I109A mutant receptor. MD simulations revealed that for the double mutants, the sidechain of Phe-13 in AP-13 can still rotate down due to I109A mutation at the early stage of simulation. However, it does not stay there as in I109A single mutant but assumes a conformation which resembles what was observed with WT APJ. This is in line with the hypothesis that the cluster of hydrophobic interactions between sidechains of Phe-13 with F1103.33 and M1133.36 plays a crucial role in maintaining a conformation for biased signaling of APJ I109A mutant receptor.
β-arrestin signaling deficiency in APJ I109A was largely recovered in F110A/I109A and M113A/I109A double mutants.
Modulation of GPCR function is one of the most common approaches in drug discovery, either with agonists to mimic endogenous receptor activation or with antagonists to block endogenous ligand binding [2,14]. Understanding the relationship between structural features of the receptor and downstream signaling could provide the basis for designing GPCR modulators with desirable therapeutic profiles. Owing to the potential therapeutic benefit of biased agonists, GPCR crystallography and other biophysical methods have been utilized in mechanistic studies of biased signaling for several GPCRs [8,13,23,36,37]. Most of these studies focus on ligand structure–activity relationship. How receptor structural changes translate into downstream signaling selectivity is still largely unclear. In the present study, we systematically mutated residues involved in ligand interaction based on our recently published APJ crystal structure , and identified a G protein biased mutant receptor, APJ I109A. Using MD simulation, we proposed a possible molecular mechanism of biased signaling through a cascade of conformational changes along the sidechains of amino acid residues in TM3 initiated by a rotation of the sidechain phenyl group of Phe-13 in AP-13. This is the first example demonstrating from the receptor perspective how a single mutation at the orthosteric site leads to receptor conformational changes and signaling selectivity upon agonist stimulation.
The nanoseconds of MD simulation conducted here are not long enough to study the full activation of the receptor. However, similar to previous studies [38–40], such timescale did allow us to investigate the early events during the activation process. The outward movement of TM6 was observed in simulations of both WT and I109A mutants of APJ in complex with AP-13, a signature of GPCR activation for G protein signaling. The simulation results from APJ I109A further shed light on our understanding of the structural basis of GPCR activation and downstream signaling. In comparison with WT receptor, Ile 109 to Ala mutation caused a rotation of TM3 in the presence of AP-13 and the subsequent conformational rearrangement of TM7, helix VIII, and ICL1 which have been implicated in arrestin biased signaling . We further confirmed the contribution of the hydrophobic interactions between the sidechains of Phe-13 of AP-13 and F1103.33, or M1133.36 in stabilizing biased conformation states. All these results, from our studies of the I109A mutant and from recent reports of different TM3 mutations in other GPCRs leading to receptor inactivation or constitutive activation [4,41,42], imply the important structural and functional role of TM3 in arrestin binding and signaling selection of GPCRs.
The biased signaling mechanism of APJ I109A mutant illustrated by our molecular modeling studies here may be apelin specific as the modeling was based on the complex structure of APJ and the apelin analog peptide, AMG3054. The similar APJ signaling bias caused by I109A mutation can have different mechanism for other ligands (e.g. elabela), considering the sequence differences compared with apelin. Further structural biology studies (e.g. APJ/elabela complex) will provide more clues to agonist-stimulated APJ activation mechanism and reveal the structural basis behind the ligand-independent biased signaling of APJ I109A mutant.
Receptor phosphorylation is a prerequisite for arrestin recruitment and binding. It was reported that removal of a phosphorylation site by mutation S348A in the C-terminal tail of APJ led to the elimination of both GRK and β-arrestin recruitment to receptor, while G protein activation and G protein-dependent intracellular signaling were not affected . With APJ I109A mutant receptor, we observed significant structural differences from the simulations above in the C-terminal tail when compared with the WT. We wondered if these structural differences are sufficient to diminish GRK recruitment and subsequent receptor phosphorylation and β-arrestin recruitment. Indeed, our data showed that recruitment of GRK2 to I109A mutant receptor was compromised in comparison with WT receptor (Figure 3B), thus providing experimental support that deficiency in β-arrestin recruitment and biased signaling by APJ I109A is likely due to the lack of agonist-induced GRK recruitment and receptor phosphorylation.
It has been well established that the beneficial effect of APJ on cardiac contractility and vasodilator activity is mediated by Gi signaling, while the stretch-induced cardiac hypertrophy is mediated by β-arrestin pathway in G protein-independent manner . Therefore, APJ biased agonists could provide clinical benefits following chronic use for the treatment of heart failure. There are a few APJ-specific biased agonists reported previously in the literature, such as MM07 and CMF-019. These compounds are useful tool compounds while they would require further development to become clinical candidates [44,45]. We believe that new insights into the molecular mechanism and conformational dynamics of biased signaling from APJ I109A receptor will facilitate the rational design of novel biased agonists with improved pharmacokinetics for the treatment of cardiovascular diseases. Generally, it is believed that ligand interaction with TM6 is required for G protein signaling, while interaction with TM7 is critical for arrestin singling pathway [11,35]. Our results reported here suggest a complementary and crucial role for TM3 involvement in biased signaling and ligand design. More specifically for the APJ receptor, preferential interactions with residues I109 and F110 in TM3 could provide a structural basis for biased APJ ligand design. Molecular docking studies of CMF-019, the first reported biased small molecule for the APJ, support this notion .
Roles of β-arrestin have been reported for many GPCRs, the physiological contribution of β-arrestin in GPCR is still not well defined. So far, definitions of β-arrestin function in GPCR activation and downstream signaling were based on knockdown of β-arrestin expression level by siRNAs, β-arrestin gene knock-out, or G protein-biased agonists [12,14,22]. These approaches provided useful information indicating that β-arrestins play important roles in different physiological processes, but there are caveats regarding data interpretation. First, β-arrestins were associated with numerous signaling pathways that are not limited to GPCRs, so knockdown of β-arrestin may yield many confounding effects. Second, limited biased properties and off-target profiles marginalize the application of biased small molecules or peptide analogs and may only provide small pharmacological outcomes. More importantly, under physiological conditions, there are abundant endogenous GPCR ligands available which could bind and compete in receptor activation in a balanced manner. As such, the responses from biased ligands could be masked significantly by such complicated background signals. This could be the reason why some biased compounds only showed mediocre effect in clinical studies. The APJ I109A mutant we report here provides a very useful pharmacology tool to investigate the functions of β-arrestin because of a clean biased signaling pathway this mutant creates. In such a study system expressing APJ I109A, any observed physiological changes compared with the WT system could be attributed to the deficiency in β-arrestin signaling. More prospectively, rewiring GPCR pathways using techniques such as clustered regularly interspaced short palindromic repeats (CRISPR) to cause biased signaling could have therapeutic potential that could not be achieved by conventional drug molecules . Of course, we should keep in mind that one key caveat in such studies using APJ I109A as model system is that all endogenous ligands for the I109A mutant have slightly lower EC50s as compared with the WT receptor.
In summary, we have identified a critical residue I109 in TM3 for APJ signaling selectivity and rationalized the molecular mechanism using MD simulations. We have shown that altering the APJ binding pocket by mutagenesis influences the balance of receptor conformation states. Whether similar structural alterations at or near the orthosteric binding site of other GPCRs could generate biased mutants remains a subject of future investigation. Such studies could potentially consolidate a completely new approach for GPCR pharmacological study and drug discovery.
bioluminescence resonance energy transfer
clustered regularly interspaced short palindromic repeats
Dulbecco's modified Eagles medium
fetal bovine serum
fluorescent imaging plate reader
G protein-coupled receptors
G protein-coupled receptor kinase
- NanoBit assay
nanoluc luciferase-based enzyme complementation assay
small interfering RNA
transmembrane domain 3
T.B. and L.A.H. designed the experiments. T.B. performed BRET, NonaBit, imaging and pERK assays, was responsible for data analysis and wrote the manuscript. X.L. performed the MD simulation analysis for WT APJ and mutants with AP-13 peptide and supported manuscript preparation. X.M. contributed to assay development and study design. M.S., Y.Song, Y.Su., and H.Y. contributed to radioligand binding and cAMP assays. N.L. and M.Y.Z. assisted in study design and data interpretation. Y.M. contributed to mutagenesis design and helped molecular modeling result interpretation. W.Z. supervised MD simulation work, assisted in experiments design, and data review and manuscript preparation. M.Q.Z. supervised the project, reviewed data, and assisted in manuscript preparation. L.A.H. conceived the project, supervised the research, and wrote the manuscript.
The work was funded by Amgen Inc. and the authors on the publication were employees of Amgen at the time the research was conducted.