Structural and functional determination of peptide versus small molecule ligand binding at the apelin receptor

Abstract

We describe a structural and functional study of the G protein-coupled apelin receptor, which binds two endogenous peptide ligands, apelin and Elabela/Toddler (ELA), to regulate cardiovascular development and function. Characterisation of naturally occurring apelin receptor variants from the UK Genomics England 100,000 Genomes Project, and AlphaFold2 modelling, identifies T89^2.64 as important in the ELA binding site, and R168^4.64 as forming extensive interactions with the C-termini of both peptides. Base editing to introduce an R/H168^4.64 variant into human stem cell-derived cardiomyocytes demonstrates that this residue is critical for receptor binding and function. Additionally, we present an apelin receptor crystal structure bound to the G protein-biased, small molecule agonist, CMF-019, which reveals a deeper binding mode versus the endogenous peptides at lipophilic pockets between transmembrane helices associated with GPCR activation. Overall, the data provide proof-of-principle for using genetic variation to identify key sites regulating receptor-ligand engagement.

Fig. 1. Overview of apelin receptor ligands that regulate the cardiovascular system in health and disease.

a Aligned amino acid sequences of the endogenous apelin receptor peptides, [Pyr¹]apelin-13 (upper) and ELA-11 (lower). Amino acids are provided as single letter codes and coloured according to classification, as indicated in figure. b Schematic showing the signalling cascades following ligand binding and activation of apelin receptor present on the cell surface. Canonical G_i protein signalling mediates physiological effects of the apelin receptor, including increasing cardiac contractility, inducing vasodilatation, and antithrombotic action. Prolonged binding of an agonist can facilitate phosphorylation of the receptor by G protein-coupled receptor kinase (GRK), and subsequent recruitment of β-arrestin protein. This typically desensitises and ‘switches off’ receptor signalling by sterically hindering engagement with G protein complexes and inducing internalisation of the receptor. Created in BioRender. Davenport, A. (2024) https://BioRender.com/f02c819. c Chemical structures of two apelin receptor small molecule agonists, CMF-019 (left) and cmpd644 (right). The structures show differences in chemotypes between the two ligands. d Snake plot of the apelin receptor showing the 380 amino acid sequence generated using GPCRdb (https://gpcrdb.org/). Colours indicate the residue locations for missense (V38, blue; T89, red; R168, magenta) or frameshift (G45, T227, both yellow) variants identified in individuals recruited to the 100,000 Genomes Project. Six other missense variants (green) were identified but were not further characterised in vitro, as they did not significantly alter cell surface expression or [¹²⁵I]-apelin-13 radioligand binding compared to wild-type.

Fig. 2. Apelin receptor variants show significant differences in expression and peptide ligand binding.

a Experimental pipeline schematic for in vitro characterisation of apelin receptor variants. CHO-K1 cells were transiently transfected with wild-type (WT) or variant apelin receptor constructs tagged C-terminally with an eGFP reporter. In high content screening studies (upper track), cells were plated and treated with fluorescently labelled apelin647 (300 nM) or ELA647 (1 µM) for 90 min before fixation and visualisation. In radioligand binding studies (lower track), cells were lysed and the membranes harvested before treatment with a concentration range (up to 1 nM) of [¹²⁵I]-apelin-13. Bound radioactivity was counted to determine receptor density and affinity. Created in BioRender. Davenport, A. (2024) https://BioRender.com/o44f254. b Representative confocal fluorescent images (n ≥ 3 independent experiments). Panels show a merge (eGFP reporter in green and 647 nm fluorescent ligand in red) or the 647 nm fluorescent peptide signal in isolation. Non-specific binding (NSB) was determined in WT cells treated with fluorescent peptide in the presence of 10 µM unlabelled [Pyr¹]apelin-13. Scale bars 50 µm. c Bar chart showing mean ± SD membrane eGFP fluorescence (expressed as % WT) from cells imaged over 12 regions, pooled from n = 4 independent experiments in triplicate. Untrans=cells untransfected with apelin receptor cDNA. d Bar chart showing mean ± SD count (n = 4 independent experiments in triplicate) of eGFP positive cells per region. e Grouped bar chart showing mean membrane apelin647 (solid bars) or ELA647 (slashed bars) fluorescence (expressed as %WT, n = 4 independent experiments in triplicate). For all data expressed in bar charts, statistical significance was determined using a one-way ANOVA, with Dunn’s correction for multiple comparisons. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. f Saturation binding curves showing specific binding of [¹²⁵I]-apelin-13 in transfected membrane preparations. Data are mean ± SD, n = 3 independent experiments. NSB was determined in the presence of 10 µM unlabelled [Pyr¹]apelin-13. g Competition radioligand binding curves showing ELA-11 peptide competing for binding (% specific) with a single 0.1 nM concentration of [¹²⁵I]-apelin-13 at WT or T/M89^2.64 variant apelin receptor in membrane preparations. Data expressed as mean ± SD, n = 3 independent experiments. Source data are provided as a Source Data file.

Fig. 3. AlphaFold2 models of apelin and ELA bound apelin receptor provide a structural rationale for altered binding at the T/M892.64 and R/H1684.64 variants.

a Cartoon representation of the overall model of human apelin receptor (wheat) bound to mini-Gi (blue). Amino acid positions for variants identified in individuals recruited to the 100,000 Genomes Project are represented as spheres: V38^1.42 (blue), T89^2.64 (red), R168^4.64 (magenta), G45^1.49 and T227^5.64 (yellow), and six others not experimentally characterised (green). The C-termini of apelin (cyan) and ELA (orange) peptides are shown as sticks. b Close-up views of apelin (left, cyan) and ELA (right, orange) peptide show C-termini binding in the orthosteric site of the human apelin receptor. Side chains of residues lining the peptide binding pocket, notably R1684.64 interacting with the C-terminal carboxyl group of the peptides, and the modelled T/M89^2.64 substitution impacting binding of ELA (steric clash indicated by red polygon) but not that of apelin, are represented as sticks. Source data are provided as a Source Data File for 3b.

Fig. 4. The crystal structure of the apelin receptor NxStaR in complex with CMF-019 reveals deep engagement with lipophilic sub-pockets at the bottom of the orthosteric site.

a Cartoon representation (rainbow colours) of human apelin stabilised receptor (NxStaR) crystal structure (ICL3 bRIL fusion was omitted for clarity), with the CMF-019 ligand represented as spheres, and oleic acid as sticks. b Close-up view of the CMF-019 binding pose at the bottom of the apelin receptor orthosteric site, with lining amino acids and oleic acid represented as sticks (left), and CMF-019 binding pocket represented as a light grey mesh, with lipophilic hotspots coloured in pale wheat (right). The extracellular vestibule is labelled as ECV, and the lipophilic sub-pockets are labelled as I, II and III. The small table below panel b lists the residues interacting with CMF-019 and the corresponding sub-pockets they line.

Fig. 5. Comparison of small molecule agonist apelin receptor complex crystal and cryoEM structures reveal that the CMF-019 complex exists in an intermediate active state.

a Views of an overlay of the apelin receptor NxStaR crystal structure (orange) in complex with CMF-019 (magenta), with the cmpd644 crystal structure and Gi bound (not shown) cryoEM structure (green) in cartoon representation. Small molecule agonists and relevant amino acid side chains are represented as sticks. Top panels show top-down views, lower left panel shows a side view, and lower right panel shows a close-up view on the TM6-TM7 interface. b (Upper) Close-up view of an overlay of the small molecule agonist bound crystal structures, in cartoon representation, with CMF-019 (dark purple), cmpd644 (blue), and relevant amino acid side chains represented as sticks. (Lower) Close-up view of the cmpd644 binding pose in the apelin receptor orthosteric site, with lining amino acid side chains represented as sticks and the cmpd644 binding pocket represented as a light grey mesh with lipophilic hotspots coloured in pale wheat. The extracellular vestibule is labelled as ECV and lipophilic sub-pockets are labelled as I, II and III. The small table lists the residues interacting with CMF-019 or cmpd644, and their corresponding sub-pockets.

Fig. 6. Comparison of the CMF-019 complex apelin receptor structure with peptide bound complexes emphasises the deeper reach of the small molecule agonist at the TM6-7 and TM7-2 interfaces.

Panels show overlay of the apelin receptor NxStaR crystal structure (rainbow colours) in complex with CMF-019 (magenta) with the previously reported AMG3054 (blue) bound crystal structure (left), the higher resolution cryoEM ELA (dark teal) G protein bound monomeric complex (7W0P, middle), and the apelin (brown) bound apelin receptor complex (right). Aromatic cage amino acids are represented as sticks. The small table lists the residues interacting with CMF-019, and corresponding sub-pockets as shown in Fig. 3, apelin, ELA, and AMG3054.

Fig. 7. Surface plasmon resonance (SPR) sensorgrams of the apelin receptor NxStaR show loss of binding of truncated peptides and that the R/H168^4.64 variant can still bind small molecule agonists.

a SPR sensorgrams of the apelin receptor NxStaR in the presence of endogenous apelin-13 peptide versus C-terminally truncated peptides (ΔC2-ΔC6). b SPR sensorgrams of apelin receptor NxStaR or the R/H168^4.64 variant NxStaR in the presence of endogenous apelin-13 peptide (first column), the chimeric peptide NXE’992 (second column), the G protein biased, small molecule agonist CMF-019 (third column), and NXE’870 (fourth column) at pH 7.5 or 6.0. Source data are provided as a Source Data file.

Fig. 8. Using base editing to introduce the R/H168^4.64 substitution into endogenously expressed apelin receptor reduces binding and induces phenotypic dysfunction in a human hESC-CM model.

a Schematic showing experimental pipeline. Cytosine base editing technology was used to introduce the R/H1684.64 substitution into endogenously expressed apelin receptor in human embryonic stem cells (hESCs) originally harvested from the inner cell mass of a human blastocyst, or into differentiated cardiomyocytes (hESC-CMs). Cells were characterised for R/H168^4.64 apelin receptor mRNA expression, protein expression, and radioligand binding. Functional consequences of the R/H168^4.64 substitution were also assessed. Created in BioRender. Davenport, A. (2024) https://BioRender.com/e87y890. b Relative apelin receptor gene (APLNR) expression in wild-type (WT) or R/H168^4.64 variant hESC (solid bars) and hESC-CM (slashed bars) lines. Data expressed as n = 2 for WT hESC, n = 4 (mean ± SD) for WT hESC-CMs, and n = 8 (mean ± SD), for R/H168^4.64 hESC and hESC-CMs. c Representative confocal fluorescent images (n = 4 independent experiments) of WT and R/H1684.64 hESC-CMs treated with anti-apelin receptor antibody (visualised in green) and nuclear marker (visualised in blue) (upper panels) or 300 nM apelin647 fluorescent ligand (visualised in red) (lower panels). Control WT hESC-CMs (upper) were treated with secondary antibody in the absence of the primary, or (lower) in the presence of 10 µM unlabelled [Pyr¹]apelin-13. Scale bars indicate 50 µm. d Saturation radioligand binding curves showing specific binding of [¹²⁵I]-apelin-13 in hESC (upper) or hESC-CM (lower) membrane preparations. Non-specific binding was determined in the presence of 10 µM [Pyr¹]apelin-13. Data expressed as mean ± SD, n = 3 independent experiments. e Functional consequences of the R/H1684.64 substitution in hESC-CMs. Bar charts show differentiation efficiency measured by % hESC-CMs staining positive for troponin T (TnT) (left), apelin peptide concentrations secreted from hESC-CMs as determined by ELISA (middle), and time-to-peak (TTP) voltage per unit time (ms) (right). For all panels here, data are expressed as mean ± SD, n ≥ 3 independent experiments. Statistical significance was determined using a one-way ANOVA, with Tukey’s correction for multiple comparisons looking for differences between wild-type (WT) and variant (R/H168^4.64) apelin receptor. *p < 0.05, **p < 0.01, ***p < 0.001. Source data are provided as a Source Data file.