Allosteric modulators enhance agonist efficacy by increasing the residence time of a GPCR in the active state

Abstract

Much hope in drug development comes from the discovery of positive allosteric modulators (PAM) that display target subtype selectivity and act by increasing agonist potency and efficacy. How such compounds can allosterically influence agonist action remains unclear. Metabotropic glutamate receptors (mGlu) are G protein-coupled receptors that represent promising targets for brain diseases, and for which PAMs acting in the transmembrane domain have been developed. Here, we explore the effect of a PAM on the structural dynamics of mGlu2 in optimized detergent micelles using single molecule FRET at submillisecond timescales. We show that glutamate only partially stabilizes the extracellular domains in the active state. Full activation is only observed in the presence of a PAM or the G_i protein. Our results provide important insights on the role of allosteric modulators in mGlu activation, by stabilizing the active state of a receptor that is otherwise rapidly oscillating between active and inactive states.

Fig. 1. Structure and conformational rearrangements of mGlu receptor.

Structures of dimeric mGlu2 in resting and active conformations. The major structural elements of each subunit include the extracellular domain (ECD, comprising the Venus fly-trap domain (VFT) and the cysteine-rich domain (CRD)) and the seven-transmembrane domain (7TM). Orthosteric ligand binding sites are found in the cleft between the upper and lower lobes (black circles) of the VFT and the majority of allosteric modulators bind to sites in the 7TM (blue circles). Activation leads to a closure of the VFTs and a reorientation of the ECDs, the CRDs, and the 7TMs bringing the two subunits into closer proximity. In N-terminally SNAP-tag labeled receptor dimers this leads to a transition from a high FRET/resting to a low FRET/active state. G protein activation through interactions with the cytoplasmic side of the 7TM is reported to occur at >10 ms timescales. The shown structures were generated using PDB ID 7EPA and 7E9G.

Fig. 2. Evaluation of detergents for functional solubilization of full-length mGlu2 using LRET.

a SNAP-mGlu2 dimers were labeled with cell-impermeable lanthanide donor and “green” acceptor fluorophores on living HEK293T cells. After preparation of crude membrane fractions, LRET measurements were performed in microtiter plates either directly on membranes or after detergent solubilization. b–h The functional integrity of SNAP-labeled receptors was monitored over time at room temperature based on the dose-dependent intersubunit LRET changes in response to the orthosteric agonist Glu (- PAM) and in combination with 10 µM positive allosteric modulator BINA (+ PAM). b–g Dose-response curves at time 0 h (top) and time course of pEC₅₀ values (bottom) obtained on crudes membranes (b, n = 4), in IGEPAL (c, n = 3), DDM (d, n = 3), LMNG (e, n = 3), LMNG-CHS (f, n = 4) and LMNG-CHS-GDN (g, n = 4). Data represent the mean from different biological replicates ± SD. Statistical differences of pEC₅₀ values for Glu (black) and Glu + BINA (blue) compared to time 0 h were determined using two-sided unpaired t-tests and are given as: (b) p_Glu-24h = 0.92 (ns), p_Glu+BINA-24h = 0.46 (ns), n = 4 independent biological samples examined over 3 independent experiments; (c) p_Glu-2h = 0.65 (ns), p_Glu-4h = 0.49 (ns), p_Glu-6h = 0.14 (ns), *p_Glu-24h = 0.014, p_Glu+BINA-2h = 0.68 (ns), p_Glu+BINA-4h = 0.72 (ns), p_Glu+BINA-6h = 0.46 (ns), p_Glu+BINA-24h = 0.086 (ns), n = 3 independent biological samples examined over 3 independent experiments; (d) p_Glu-2h = 0.29 (ns), p_Glu-4h = 0.20 (ns), p_Glu-6h = 0.46 (ns), p_Glu-24h= 0.38 (ns), p_Glu+BINA-2h = 0.0995 (ns), p_Glu+BINA-4h = 0.068 (ns), p_Glu+BINA-6h = 0.25 (ns), p_Glu+BINA-24h = 0.086 (ns), n = 3 independent biological samples examined over 3 independent experiments; (e) *p_Glu-2h = 0.048, *p_Glu-4h = 0.011, **p_Glu-6h = 0.0057, **p_Glu-24h = 0.009, *p_Glu+BINA-2h = 0.014, ***p_Glu+BINA-4h = 0.00025, ***p_Glu+BINA-6h = 0.00024, ****p_Glu+BINA-24h = 0.000008, n = 3 independent biological samples examined over 3 independent experiments; (f) p_Glu-2h = 0.33 (ns), *p_Glu-4h = 0.027, **p_Glu-6h = 0.0063, ***p_Glu-24h = 0.00035, p_Glu+BINA-2h = 0.16 (ns), *p_Glu+BINA-4h = 0.039 (*), *p_Glu+BINA-6h = 0.018, ****p_Glu+BINA-24h = 0.00004; n = 4 independent biological samples examined over 1 independent experiment; g p_Glu-2h = 0.89 (ns), p_Glu-4h = 0.6 (ns), p_Glu-6h = 0.54 (ns), p_Glu-24h = 0.78 (ns), p_Glu+BINA-2h = 0.39 (ns), p_Glu+BINA-4h = 0.2 (ns), p_Glu+BINA-6h = 0.43 (ns), p_Glu+BINA-24h = 0.83 (ns), n = 4 independent biological samples examined over 3 independent experiments. h Scatter plot of ΔEC₅₀, i.e. the difference in EC₅₀ obtained in presence and absence of BINA, at time 24 h at RT (y-axis) vs. at time 0 h (x-axis), for membrane fractions and detergent mixtures. The conditions along the diagonal represent those experiencing the lowest changes over time. Source data of panels b–h are provided as a source data file.

Fig. 3. smFRET reveals the conformational landscape of full-length mGlu2 in LMNG-CHS-GDN micelles.

a SNAP-mGlu2 dimers were labeled with cell-impermeable Cy3B donor and d2 acceptor fluorophores on living HEK-293T cells. Then mGlu2 dimers were detergent-solubilized from crude membrane fractions and smFRET measurements were performed on freely diffusing molecules with confocal illumination. b–h Representative histograms displaying the number of doubly labeled molecules as a function of apparent FRET efficiency (E_PR). Distributions were obtained in the absence of ligand (Apo) or in the presence of Glu (10 mM), competitive antagonist LY341495 (1 mM), BINA (10 µM), and G protein (1 µM), as indicated. Colored lines represent Gaussian fitting, black lines correspond to the cumulative fitting (see text). All histograms revealed four major populations at very low FRET (VLF, purple), low FRET (LF, green), high FRET (HF, yellow), and very high FRET (VHF, red). i smFRET analysis of the effect of Glu without (Agonist) or with BINA (10 mM) or G_i (1 µM), as indicated. The fraction of the active state is defined as the fraction of molecules in the LF population over all molecules in the LF+HF populations. j) smFRET analysis of the reversibility of the PAM-induced full ECD activation (500 nM, 2 h) through competition with an excess of the NAM Ro64-5229 (10 µM, 4 h). The statistical difference was determined using a two-sided, paired t test and is given as *p = 0.017. n = 3 independent biological replicates examined over 3 independent experiments. i–j Data were obtained from three biological replicates and are given with mean ± SD. Source data of panels b–j are provided as a source data file.

Fig. 4. Structural dynamics analysis of mGlu2 dimers in response to orthosteric and allosteric ligands.

a–c Representative τ_DA vs. E histogram for mGlu2 dimers in the absence (Apo) or presence of Glu or Glu + BINA. For the Apo receptors, the major population deviates from the “Static FRET” line (yellow), indicating conformational dynamics at the submillisecond time scale. The addition of Glu stabilizes the VFT in an ensemble of low FRET conformations with less flexibility, an effect that is reinforced by the allosteric modulator BINA. d–f Time window analysis for different integration times (from 0.2–1 ms) reveals large conformational flexibility of the Apo VFT at 200–600 µs timescales, which is strongly restricted when bound to orthosteric agonist and allosteric modulator. The number of data points used varied from 2183-1640-901 (d–f respectively, 200 µs) to 14082-13132-13269 (1 ms) g–j Schematic representation of the major species observed in all cases, with the timescales of the transition between them. Black and blue dots represent Glu and BINA, respectively. Source data of panels a–f are provided upon request.

Fig. 5. Different efficacies of orthosteric ligands on mGlu2 ECD rearrangement.

a–h FRET distributions were obtained in the presence of partial agonists LCCG-I (a) and DCG-IV (b) or full synthetic agonist LY379268 (g) alone or in the presence of BINA (c, d, h respectively) or G_i (e and f, respectively). i Comparison of the fraction of the active state (LF/(LF+HF)) in response to different orthosteric and allosteric ligands. n = 3 independent biological replicates examined over 3 independent experiments for each condition. The scatter plot shows data together with the mean± SD. Statistical differences were determined using a one-way ANOVA with Sidak multiple comparisons test and are given as: p_LY35/Glu = 0.0068 (**), p_Glu/LY37 = 0.0006 (***), p_{DCG-IV+BINA/Glu+BINA} = 0.0028 (**), p_{Glu+BINA/LY37+BINA} = 0.61 (ns). j Comparison of the fraction of all states in response to different orthosteric and allosteric ligands. Data represent the stacked means ± SD from 3 independent biological replicates for each state with error bars centered around the mean for the VLF, LF, HF, and VHF from top to bottom, respectively. Source data of panels a–j are provided as a source data file.