Direct interrogation of context-dependent GPCR activity with a universal biosensor platform

Abstract

G protein-coupled receptors (GPCRs) are the largest family of druggable proteins in the human genome, but progress in understanding and targeting them is hindered by the lack of tools to reliably measure their nuanced behavior in physiologically-relevant contexts. Here, we developed a collection of compact ONE vector G-protein Optical (ONE-GO) biosensor constructs as a scalable platform that can be conveniently deployed to measure G-protein activation by virtually any GPCR with high fidelity even when expressed endogenously in primary cells. By characterizing dozens of GPCRs across many cell types like primary cardiovascular cells or neurons, we revealed new insights into the molecular basis for G-protein coupling selectivity of GPCRs, pharmacogenomic profiles of anti-psychotics on naturally-occurring GPCR variants, and G-protein subtype signaling bias by endogenous GPCRs depending on cell type or upon inducing disease-like states. In summary, this open-source platform makes the direct interrogation of context-dependent GPCR activity broadly accessible.

Figure 1.. Direct detection of active Gαs in cells with a BRET biosensor.

(A) Identification of Gαs-GTP peptide binders. Top, schematic of phage display screen. Bottom, purified GST-fused peptides immobilized on glutathione-agarose beads were incubated with either active, GDP-AlF₄⁻-loaded Gαs or Gαo, and bead-bound proteins detected by Ponceau S staining or immunoblotting. Results are representative of n ≥ 2 experiments. (B) Gαs-GTP biosensor using KB1691 and KB2123 peptides. Left, diagram of BRET-based detection of GPCR-mediated activation of Gαs. Center and left, BRET was measured in HEK293T cells expressing β2AR, Gαs-99V, Gβγ and either KB1691-Nluc or KB2123-Nluc. A representative kinetic trace from 4 independent experiments in shown in the center panels, and the mean ± S.E.M. of n=4 is shown for the dose-dependence curves on the right. (C) Sensors based on KB1691, but not on KB2123, specifically detect Gαs. BRET was measured in HEK293T cells expressing the indicated G protein / cognate GPCR pairs and either KB1691-Nluc or KB2123-Nluc. Cells were stimulated with either 10 μM isoproterenol (for Gαs), 5 μM brimonidine (for Gαi3), 100 μM carbachol (for Gαq), or 30 μM TRAP-6 (for Gα13). Results are mean ± S.E.M. of n=3–4. (D) KB1691 does not interfere with Gαs-mediated cAMP production. Luminescence was measured in HEK293T cells expressing the cAMP probe Glosensor and exactly the same components as in panel B, except that KB1691-Nluc was omitted in the control. Results are the mean ± S.E.M. of n=4. A representative immunoblotting result confirming expression of KB1691-Nluc is shown on the right.

Figure 2.. Mechanism of Gαs activation by oncogenic mutations and of G_s coupling to GPCRs.

(A) Activation properties of oncogenic Gαs mutants. Left, view of Gαs nucleotide-binding pocket with residues mutated in cancers (PDB: 1AZT). Right, BRET was measured in HEK293T cells expressing the same components as in Fig. 1B with KB1691-Nluc, except that Gαs-99V (WT) was replaced as indicated by Gαs-99V R201C or Q227L mutants. Bar graph represents BRET signal in unstimulated cells or cells stimulated with 1 μM isoproterenol relative to unstimulated Gαs-99V WT. Results are the mean ± S.E.M., n=4. Kinetic traces of BRET measurements with cells expressing the same components and treated as indicated. Results are expressed as the difference in BRET from their corresponding unstimulated baselines. Results are the mean ± S.E.M., n=3. The immunoblot is a representative result confirming equal expression of Gαs and KB1691-Nluc. (B) View of structural elements of Gαs involved in coupling to β2AR based on their complex structure (PDB: 3SN6). (C, D) Contribution of the C-terminal tail (CT) and the αN/β1 ‘hinge’ of Gα subunits to their coupling to GPCRs. BRET was measured in HEK293T cells expressing the indicated YFP-tagged Gα chimeras, GPCRs, and Gβγ with either a Gαs-GTP biosensor (KB1691-Nluc, C) or a Gαi-GTP biosensor (KB1753-Nluc, D). Heat maps correspond to the efficacy of the BRET responses detected relative to the maximal response observed with the cognate WT G protein. Results are the mean of n=3–4. Full dose-dependence curves are presented in Fig. S4).

Figure 3.. ONE vector G protein Optical (ONE-GO) biosensor designs display improved features.

(A) Schematic of the process to develop ONE-GO biosensors. (B-C) Gαs and Gαi3 ONE-GO biosensor designs provide increased responses and reduced component expression. BRET was measured in HEK293T cells transfected with the indicated single-plasmid ONE-GO biosensors or their multi-plasmid (M–P) counterparts. β2AR or α2_A-AR were co-expressed for Gαs (B) or Gαi3 (C), respectively. Results are the mean ± S.E.M. of n=3. A representative immunoblotting result comparing expression of sensor components is shown alongside the BRET traces. (D) ONE-GO biosensors do not interfere with GPCR-G protein signaling. Luminescence was measured in HEK293T cells expressing the cAMP probe Glosensor with or without the Gαs ONE-GO sensor (Left) or Gαi3 ONE-GO sensor (Right). Cells used on the right co-expressed exogenous α2_A-AR, whereas isoproterenol responses were elicited by endogenous β2AR. Results are the mean ± S.E.M. of n=3. A representative immunoblotting result confirming expression of sensors and lack of changes in endogenous G protein subunits is shown for both ONE-GO sensors.

Figure 4.. ONE-GO biosensors report activation across G protein families and for many GPCRs.

(A) Ten ONE-GO biosensors report activity across all G protein families. Top, dendrogram of Gα subunits with their corresponding detector modules used in ONE-GO designs, and expression of biosensor components in HEK293T cells. Results are representative of n ≥ 3 experiments. Bottom, BRET was measured in HEK293T cells expressing the indicated ONE-GO biosensors and GPCRs. Results are the mean ± S.E.M. of n=3–5. (B) ONE-GO biosensors report the activity of dozens of GPCRs. Dose-dependence BRET responses were measured in HEK293T cells for 75 GPCRs to determine EC₅₀ values. Results for one representative GPCR (β2AR) are shown (all curves are presented in Fig. S7). EC₅₀ values were plotted against curated pharmacological parameters (pK_d, pK_i, or pE/IC₅₀) available in the IUPHAR database.

Figure 5.. Large-scale parallel interrogation of GPCR activity with ONE-GO biosensors.

(A) Parallel profiling of atypical antipsychotics across a large set of receptors. Left, schematic of the assay and structure of the compounds investigated. Middle, BRET responses were measured in HEK293T cells expressing the indicated GPCRs along with a cognate ONE-GO biosensor upon stimulation with the indicated concentrations of brexipiprazole, iloperidone, or cariprazine alone (to measure agonist activity) or in presence of an agonist at its EC₈₀ concentration (to measure antagonist activity). Results are the mean of n=3–7. Right, principal component analysis (PCA) of the data presented in the heatmaps. See Supplemental Table 1 for agonists and EC₈₀ concentrations used. (B) Pharmacogenomic profiles of atypical antipsychotics. Left, schematic of the assay (coarse-grained curves shown in Fig. S8A). Middle, Snake plots for 5-HT1a and 5-HT1b showing the location of genetic variations investigated on the right. Right, BRET responses were measured in HEK293T cells expressing the indicated GPCRs and the Gαi1 ONE-GO sensor. The indicated compounds were tested by themselves (filled circles) or in presence of 5-HT (open circles). Results are the mean ± S.E.M. of n=3–4. (C) Profiling of G protein selectivity across short-chain fatty acid receptors. Left, schematic of variables investigated. Right, BRET responses were measured in HEK293T cells expressing the indicated GPCR / ONE-GO biosensor combinations upon stimulation with the indicated agonist. Results are maximal responses normalized by biosensor (mean, n=3–6). Full dose-dependence curves in Fig. S8B.

Figure 6.. Detection of endogenous GPCR activity across a wide palette of primary cells with ONE-GO biosensors.

Top, summary of responses triggered by endogenous GPCRs detected with ONE-GO biosensors for all G protein families in multiple human and mouse primary cells of different origins. Bottom, BRET responses were measured in the indicated primary cells transduced with ONE-GO biosensors. Results are the mean ± S.E.M. of n=3–6. Additional examples and controls measured in parallel in Fig. S10.

Figure 7.. ONE-GO biosensors reveal context-dependent activity of endogenous GPCRs.

(A) Cell type-dependent G protein selectivity profiles of protease-activated receptor 1. Left, spider plot summarizing PAR1 G protein activation profiles for each cell type. Middle, BRET responses were measured in each one of the 4 primary cell types transduced with the indicated ONE-GO biosensor. Right, maximal BRET responses normalized to Gαq ONE-GO and raw luminescence counts indicative of biosensor expression. Results are the mean ± S.E.M. of n=3–4. (B) Discrimination across G_i/o isoforms by neuroinhibitory GPCRs in primary neurons. Left, BRET responses were measured in primary mouse cortical neurons transduced with Gαi1 ONE-GO (top row, blue) or Gαo_A ONEGO (bottom row, brown) biosensors stimulated with the indicated neurotransmitters. Results are the mean ± S.E.M. of n=3–6. Right (box), BRET responses were measured in HEK293T cells expressing either the D2R or the A1R along with the indicated ONE-GO biosensor. Results are the mean ± S.E.M. of n=3. (C) Myofibroblast transformation remodels the G protein selectivity profile of protease activated receptor 1. Left (box), confirmation of TGFβ-induced myofibroblast transformation by RT-qPCR (mean ± S.E.M. of n=4), and spider plot summarizing PAR1 G protein activation profiles before and after myofibroblast transformation (mean, n=5). Middle, BRET responses were measured in human cardiac fibroblasts transduced with the indicated ONE-GO biosensors and treated (orange) or not (grey) with TGFβ. Bar graphs represent from left to right: maximal BRET responses normalized to Gαq ONE-GO, BRET signal in unstimulated cells, and raw luminescence counts indicative of biosensor expression. Results are the mean ± S.E.M. of n=5. Right, representative immunoblots showing no difference in expression of the biosensor components (Gα, Nluc) or PAR1 upon TGFβ treatment Results are representative of n=3 experiments.