Structural basis of tethered agonism of the adhesion GPCRs ADGRD1 and ADGRF1

Abstract

Adhesion G protein-coupled receptors (aGPCRs) are essential for a variety of physiological processes such as immune responses, organ development, cellular communication, proliferation and homeostasis^1-7. An intrinsic manner of activation that involves a tethered agonist in the N-terminal region of the receptor has been proposed for the aGPCRs^8,9, but its molecular mechanism remains elusive. Here we report the G protein-bound structures of ADGRD1 and ADGRF1, which exhibit many unique features with regard to the tethered agonism. The stalk region that proceeds the first transmembrane helix acts as the tethered agonist by forming extensive interactions with the transmembrane domain; these interactions are mostly conserved in ADGRD1 and ADGRF1, suggesting that a common stalk-transmembrane domain interaction pattern is shared by members of the aGPCR family. A similar stalk binding mode is observed in the structure of autoproteolysis-deficient ADGRF1, supporting a cleavage-independent manner of receptor activation. The stalk-induced activation is facilitated by a cascade of inter-helix interaction cores that are conserved in positions but show sequence variability in these two aGPCRs. Furthermore, the intracellular region of ADGRF1 contains a specific lipid-binding site, which proves to be functionally important and may serve as the recognition site for the previously discovered endogenous ADGRF1 ligand synaptamide. These findings highlight the diversity and complexity of the signal transduction mechanisms of the aGPCRs.

Fig. 1. Overall structures of G protein-bound ADGRD1 and ADGRF1.

a, Cryo-EM maps of the ADGRD1–miniG_s, ADGRF1–miniG_s, ADGRF1–miniG_i1 and ADGRF1(H565A/T567A)–miniG_i1 complexes, coloured according to chains. The stalk and TMD of ADGRD1 are coloured orange and green, respectively; the stalk and TMD of ADGRF1 are coloured magenta and blue, respectively; the lipid LPC bound to ADGRF1 is coloured yellow; and Gα_s, Gα_i1, Gβ, Gγ and Nb35 are coloured cyan, gold, grey, pink and light gold, respectively. b, Structure of the ADGRD1–miniG_s complex. The structure is shown in cartoon representation. The binding cavities for the stalk and G protein are highlighted by two dashed boxes and are shown in detail on the left. c, Structure of the ADGRF1–miniG_s complex. The lipid LPC bound to the receptor intracellular region is shown as yellow sticks. The binding cavities for the stalk and G protein are highlighted by two dashed boxes and are shown in detail on the right.

Fig. 2. Interaction pattern between the stalk and the TMD.

a, b, The stalk binding cavities in the ADGRD1–miniG_s (a) and ADGRF1–miniG_s (b) structures. c, Schematic diagram of the tethered stalk-mediated activation of ADGRF1 with the autoproteolysis at the GPS. Upon activation, the stalk dissociates from the GAIN domain and then interacts with the TMD. The release of the stalk leads to a collapse of the original folding of the GAIN. d, Schematic diagram of the tethered stalk-mediated activation of ADGRF1 with the proteolysis-deficient mutations H565A and T567A introduced in the GPS. The proteolysis is not required for stalk exposure that results in receptor activation and unfolding of the GAIN. e, g, Interactions between the TMD and the stalk residues F^S3, L^S6 and M^S7 in ADGRD1 (e) and ADGRF1 (g). f, h, Interactions between the TMD and the stalk residues N/S^S2, A/S^S4 and I^S5 in ADGRD1 (f) and ADGRF1 (h). Polar interactions are displayed as red dashed lines. i, j, Basal activity of wild-type (WT) and mutant versions of ADGRD1 (i) and ADGRF1 (j), measured by cAMP accumulation assay. The mutants are divided into three groups by dashed lines: (i) mutations of the stalk residues F^S3, L^S6 and M^S7 (stalk-N inward) and the TMD residues that interact with these residues; (ii) mutations of the stalk residues N/S^S2, A/S^S4 and I^S5 (stalk-N outward) and the TMD residues that interact with these residues; and (iii) mutations of the stalk-C residues. Data are presented as a percentage of wild-type activity and are shown as mean ± s.e.m. (bars) from at least five independent experiments performed in technical triplicate with individual data points shown (dots). ***P < 0.0001 by one-way analysis of variance followed by Dunnett’s post-test compared to the response of wild type. Extended Data Table 2 provides detailed independent experiment numbers (n), P values and expression level.

Fig. 3. Signalling cascade of ADGRD1 and ADGRF1.

a, Overall view of the interaction cores that are important for receptor activation. The three key interaction cores (cores 1–3) are highlighted by blue, green and magenta dashed boxes, respectively. The stalk residues F^S3, L^S6 and M^S7 are shown as spheres and coloured orange; and the TMD residues in interaction cores 1, 2 and 3 are shown as spheres and coloured blue, green and magenta, respectively. b, d, f, Interactions within cores 1 (b), 2 (d) and 3 (f) in ADGRD1. The residues involved in interactions are shown as green sticks. Polar interactions are shown as red dashed lines. c, e, g, Interactions within cores 1 (c), 2 (e) and 3 (g) in ADGRF1. The residues involved in interactions are shown as blue sticks. h, i, Basal activity of wild-type (WT) and mutant versions of ADGRD1 (h) and ADGRF1 (i),measured by cAMP accumulation assay. Data are presented as a percentage of wild-type activity and are shown as mean ± s.e.m. (bars) from at least five independent experiments performed in technical triplicate with individual data points shown (dots). ***P < 0.0001 by one-way analysis of variance followed by Dunnett’s post-test compared to the response of wild type. ^#Low surface expression level (less than 40% of wild-type expression level). Extended Data Table 2 provides detailed independent experiment numbers (n), P values and expression level.

Fig. 4. Lipid molecule in ADGRF1.

a, Lipid-binding site in ADGRF1. The ADGRF1–miniG_s structure is shown in cartoon representation. The receptor is also shown as surface. The lipid LPC is shown as yellow sticks. The receptor residues involved in lipid binding are shown as blue sticks. b, High-resolution tandem mass spectrometry (MS/MS) spectra of two LPC molecules specifically associated with ADGRF1. Their experimental spectra matched with the reference spectra recorded in the Lipid-Blast database. c, Basal activity of wild-type (WT) and mutant versions of ADGRF1, measured by cAMP accumulation assay. Data are presented as a percentage of wild-type activity and are shown as mean ± s.e.m. (bars) from at least five independent experiments performed in technical triplicate with individual data points shown (dots). *P < 0.05, **P < 0.001, ***P < 0.0001 by one-way analysis of variance followed by Dunnett’s post-test compared to the response of wild type. ^#Low surface expression level (less than 40% of wild-type expression level). Extended Data Table 2 provides detailed independent experiment numbers (n), P values and expression level. d, A8-induced G_s and G_i activation of ADGRF1. Data are shown as mean ± s.e.m. from at least four independent experiments performed in technical duplicate. Extended Data Table 3 provides detailed independent experiment numbers (n), P values, statistical evaluation and expression level.

Extended Data Fig. 1. Protein optimization and ligand structures.

a, Schematic diagrams of ADGRD1 and ADGRF1 constructs used in this study. ADGRD1-construct was used to determine the ADGRD1–miniG_s structure. ADGRF1-construct 1 was used to determine the ADGRF1–miniG_s and ADGRF1–miniG_i1 structures. ADGRF1-construct 2 was used to determine the ADGRF1(H565A/T567A)–miniG_i1 structure. PTX, pentraxin domain; SEA, sperm protein/enterokinase/agrin domain. b, c, Receptor optimization of ADGRD1 and ADGRF1 for the structural studies. The curves of analytical size-exclusion chromatography (aSEC) of purified protein samples show higher yield and better homogeneity for the optimized receptors. d, e, G protein screening for ADGRD1 and ADGRF1. The aSEC curves of purified receptor–G protein complexes show higher yield and better homogeneity for the miniG protein-bound receptors. f, Schematic diagrams of the stalk peptides pD1 and pF1. g, Chemical structures of LPC 16:0, synaptamide and A8.

Extended Data Fig. 2. Cryo-EM processing and 3D reconstruction workflow.

a–f, Results of ADGRD1–miniG_s. a, Data processing workflow. b, Cryo-EM map coloured according to local resolution (in Å). c, Representative cryo-EM image from one independent experiment. d, Two-dimensional (2D) averages. e, Gold-standard Fourier shell correlation (FSC) curve showing an overall resolution of 2.8 Å. f, Cross-validation of model to cryo-EM density map. FSC curves for the final model versus the final map and half maps are shown in black, green and yellow, respectively. g–l, Results of ADGRF1–miniG_s. g, Data processing workflow. h, Cryo-EM map coloured according to local resolution (in Å). i, Representative cryo-EM image from two independent experiments with similar results. j, 2D averages. k, Gold-standard FSC curve showing an overall resolution of 3.1 Å. l, Cross-validation of model to cryo-EM density map. m–r, Results of ADGRF1–miniG_i1. m, Data processing workflow. n, Cryo-EM map coloured according to local resolution (in Å). o, Representative cryo-EM image from three independent experiments with similar results. p, 2D averages. q, Gold-standard FSC curve showing an overall resolution of 3.4 Å. r, Cross-validation of model to cryo-EM density map. s–x, Results of ADGRF1(H565A/T567A)–miniG_i1. s, Data processing workflow. t, Cryo-EM map coloured according to local resolution (in Å). u, Representative cryo-EM image from two independent experiments with similar results. v, 2D averages. w, Gold-standard FSC curve showing an overall resolution of 3.0 Å. x, Cross-validation of model to cryo-EM density map.

Extended Data Fig. 3. Cryo-EM density maps of the G protein-bound ADGRD1 and ADGRF1 structures.

a, ADGRD1–miniG_s; b, ADGRF1–miniG_s; c, ADGRF1–miniG_i1; d, ADGRF1(H565A/T567A)–miniG_i1. Cryo-EM maps and models of the four structures are shown for all transmembrane helices, stalk, ECL2, LPC and Gα α5-helix. The models are shown as sticks. The maps are coloured grey.

Extended Data Fig. 4. Comparison of aGPCR structures.

a, Structural comparison of the CTFs in ADGRD1 and ADGRF1. The receptors in the structures of ADGRD1–miniG_s and ADGRF1–miniG_s are shown in cartoon representation, and coloured green and blue, respectively. The stalks in the two receptors are coloured orange and magenta, respectively. b–d, Structural comparison of the helical bundles in ADGRD1, ADGRF1 and ADGRG3. The transmembrane helical bundles in the structures of ADGRD1–miniG_s and ADGRF1–miniG_s and the beclomethasone (BCM)–ADGRG3–G_o structure (PDB ID: 7D76) are shown in cartoon representation. b, Extracellular view. The red arrows indicate the movements of helices I, VI and VII in ADGRD1 and ADGRF1 relative to those in ADGRG3. c, Comparison of helix VI conformation. The sharp kink of helix VI in ADGRD1 and ADGRF1 is highlighted by a red dashed box. The palmitoylation in the ADGRG3 structure is shown as grey sticks. d, Intracellular view. The red arrows indicate the movements of helices V and VI in ADGRD1 and ADGRF1 relative to those in ADGRG3. e, Comparison of the G protein-binding cavities in ADGRD1, ADGRF1 and ADGRG3. The receptors in the structures of ADGRD1–miniG_s, ADGRF1–miniG_s and BCM–ADGRG3–G_o are shown in cartoon and surface representations. The α5-helix in Gα is coloured cyan (Gα_s) and gold (Gα_o). f, Comparison of the Gα α5-helix binding poses in ADGRD1, ADGRF1 and ADGRG3. The structures of ADGRD1–miniG_s, ADGRF1–miniG_s, ADGRF1–miniG_i1 and BCM–ADGRG3–G_o are shown in an intracellular view. The red arrow indicates the movement of the C terminus of Gα α5-helix in the ADGRG3 structure relative to that in the ADGRD1 and ADGRF1 structures. g, Comparison of G_s and G_i binding in ADGRF1. The ADGRF1–miniG_s and ADGRF1–miniG_i1 structures are shown in cartoon representation. The red arrows indicate the movements of the intracellular tip of helix VI, ICL3 and the C terminus of Gα α5-helix in the ADGRF1–miniG_s structure relative to those in the ADGRF1–miniG_i structure. h, Comparison of the stalk conformation in the ADGRF1 structures. The structures of ADGRF1–miniG_s, ADGRF1–miniG_i1 and ADGRF1(H565A/T567A)–miniG_i1 are shown in an extracellular view. i, Comparison of the binding sites for the stalk in ADGRD1 and ADGRF1 and the ligand glucocorticoid in ADGRG3. The structures of ADGRD1–miniG_s, ADGRF1–miniG_s and BCM–ADGRG3–G_o are shown. The stalk residues F^S3, L^S6 and M^S7 are shown as sticks. The glucocorticoid BCM in the ADGRG3 structure is shown as grey sticks.

Extended Data Fig. 5. Synthetic stalk peptide-induced G protein activation of wild-type ADGRD1 and ADGRF1 and mutants using BRET assays.

Data are shown as mean ± s.e.m. from at least three independent experiments performed in technical duplicate. Extended Data Table 3 provides detailed numbers of independent experiments (n), statistical evaluation and expression level.

Extended Data Fig. 6. Sequence alignment of aGPCRs.

The stalk-N is highlighted by a red background. Some key positions in the TMD are highlighted by a green background. The alignment was generated using UniProt (http://www.uniprot.org/align/) and the graphic was prepared on the ESPript 3.0 server (http://espript.ibcp.fr/ESPript/cgi-bin/ESPript.cgi).

Extended Data Fig. 7. EIC peak ratios of identified phospholipids associated with ADGRF1 versus ADGRD1.

Representative phospholipids in different classes are shown, with their EIC peak ratios indicating the compound abundance in ADGRF1 versus ADGRD1. Two specific binders to ADGRF1, LPC 16:0 and LPC 16:1, were distinguished with a mean ratio > 2 and P < 0.05, and are highlighted in pink. Data are presented as mean ± s.e.m. (bars) from three independent experiments performed in technical triplicate with individual data points shown (dots).