Control of G protein-coupled receptor function via membrane-interacting intrinsically disordered C-terminal domains

Abstract

G protein-coupled receptors (GPCRs) control intracellular signaling cascades via agonist-dependent coupling to intracellular transducers including heterotrimeric G proteins, GPCR kinases (GRKs), and arrestins. In addition to their critical interactions with the transmembrane core of active GPCRs, all three classes of transducers have also been reported to interact with receptor C-terminal domains (CTDs). An underexplored aspect of GPCR CTDs is their possible role as lipid sensors given their proximity to the membrane. CTD-membrane interactions have the potential to control the accessibility of key regulatory CTD residues to downstream effectors and transducers. Here, we report that the CTDs of two closely related family C GPCRs, metabotropic glutamate receptor 2 (mGluR2) and mGluR3, bind to membranes and that this interaction can regulate receptor function. We first characterize CTD structure with NMR spectroscopy, revealing lipid composition-dependent modes of membrane binding. Using molecular dynamics simulations and structure-guided mutagenesis, we then identify key conserved residues and cancer-associated mutations that modulate CTD-membrane binding. Finally, we provide evidence that mGluR3 transducer coupling is controlled by CTD-membrane interactions in live cells, which may be subject to regulation by CTD phosphorylation and changes in membrane composition. This work reveals an additional mechanism of GPCR modulation, suggesting that CTD-membrane binding may be a general regulatory mode throughout the broad GPCR superfamily.

Keywords: GPCR; NMR; disordered protein; mGluR; β-arrestin.

Fig. 1.

An NMR-based assay reveals phospholipid membrane binding of the intrinsically disordered mGluR2 and mGluR3 CTDs. (A) Schematic of the structural organization of mGluR domains highlighting the location of the CTD compared to the ordered parts of the protein and the membrane. (B) Schematic of NMR-based CTD–membrane binding assay that takes advantage of changes in tumbling rates of the CTD due to interactions with large unilamellar vesicles (LUVs). (C) ¹H-¹⁵N HSQC spectra of isolated mGluR2* and (D) mGluR3* CTD in the presence (red) and absence (black) of 10 mM 100 nm diameter LUVs comprised DOPS at pH 6.8 at 10 °C, with zoomed insets of crowded regions, highlighting loss of signal of specific residues in the presence of LUVs.

Fig. 2.

N-terminal regions of mGluR2 and 3 CTDs interact with negatively charged lipids. NMR intensity ratios for (A) mGluR2 and (B) mGluR3 from spectra collected with and without LUVs of three different lipid compositions. Prolines, which do not give rise to signals in ¹H-¹⁵N HSQC spectra, are denoted by *, overlapping peaks for which values were not included by **, and residues not detected in the spectra by ***. (C) Averaged intensity ratios over the first ~20 residues (mGluR2 Q822-A842; mGluR3 Q831-T851) from (A) and (B) illustrate the lipid composition dependence of the interactions in this region (±SEM of this average). (D) Averaged intensity ratios over the last ~20 residues (mGluR2 Q853-L872; mGluR3 Y861-L879) from (A) and (B) illustrate the lack of lipid composition dependence of the interactions in this region (±SEM of this average).

Fig. 3.

An N-terminal cluster of basic residues is critical for CTD membrane binding. (A) sequence alignment of the first 15 residues of the mGluR2 and mGluR3 CTDs highlighting conserved (*) and positively charged (highlighted) residues. (B) Intensity ratio plots of mGluR2 CTD constructs containing alanine substitutions for each of the four basic residues from spectra collected with and without LUVs containing a 1:1 mixture of DOPS:DOPC lipids. (C) Averaged intensity ratios over the first ~20 residues (Q822-A842) from (B) illustrate the regional effect of each mutation for LUVs of different lipid composition (±SEM of this average). (D) Averaged intensity ratios over the last ~20 residues (Q853-L872) from (B) illustrate the regional effect of each mutation for LUVs of different lipid composition (±SEM of this average). (E) Intensity ratio plots of R843A mGluR3 CTD compared to WT from spectra collected with and without LUVs containing a 1:1 mixture of DOPS:DOPC lipids.

Fig. 4.

Molecular dynamics simulations reveal multimodal membrane interactions of the intrinsically disordered mGluR3-CTD. (A) Snapshots of mGluR3 TM7-CTD (comprising TM7 residues 796-821 shown in cartoon helix representation, and CTD residues 822–879) replica 6 trajectory, highlighting three conformations of the CTD: the beginning of the simulation with no membrane contacts (t = 0, Left), and two membrane-associated states [t = 174.4 ns (Middle) and t = 1,231 ns (Right)]. Protein backbone is in blue cartoon; R838 (brown) and R843 (red) sidechains shown as spheres. (B) Total number of hydrogen bonds between each side chain and lipid headgroups averaged over all simulations. (C) Distributions, in the form of violin plots, of the distance of each residue (side chain center of mass) from the lipid phosphate plane over the course of all simulations (mean and quartiles depicted by solid and dotted horizontal lines, respectively).

Fig. 5.

Membrane interactions of the S/T-rich region of the mGluR3 CTD are modulated by mutation of a key residue and by cancer mutations. (A) mGluR3-CTD sequence annotated with the NMR- and MD-determined membrane-binding region and the overlapping Ser/Thr-rich region (* denotes residues conserved in mGluR2). (B) Snapshots of residue Y853 (shown as violet spheres with the hydroxyl group in red) in membrane-embedded and membrane-associated positions (from MD replica 6). Lipid phosphates are shown as transparent orange spheres. (C) Distance of Y853 sidechain to the lipid phosphate plane plotted as a function of time for MD replica 6 (first 1,000 ns). (D) Comparison of the averaged integrated NMR intensity ratios of WT (dotted blue) mGluR3-CTD (from Fig. 2B) with Y853A (purple) mGluR3-CTD (from SI Appendix, Fig. S9B) taken over the S/T-rich region (S845-T860) as a function of LUV lipid composition (±SEM of this average; Wilcoxon test; *P < 0.05, ***P < 0.001, n.s. P ≥ 0.05). (E) Snapshots of residues R843 (blue), G848 (orange), and E870 (red) at different time points during the time course of MD replica 6 showing prolonged membrane-association of R843 and G848 and fluctuating membrane-association of E870 (protein backbone is in gray cartoon; side chains shown as spheres colored as in (F) below; lipid phosphates are shown as transparent orange spheres). (F) Position of side chains R843, G848, and E870 relative to the phosphate plane of the membrane (Methods) throughout the time course of MD replica 6. (G) Comparison of the averaged NMR intensity ratios of WT (dotted blue) mGluR3-CTD (from Fig. 2B) with G848E (orange) mGluR3-CTD (from SI Appendix, Fig. S9B) taken over the S/T-rich region (S845-T860) as a function of LUV lipid composition (±SEM of this average; Wilcoxon test; **P < 0.01, ***P < 0.001). (H) Comparison of the averaged NMR intensity ratios of WT (dotted blue) mGluR3-CTD (from Fig. 2B) with R869Q (black) and E870K (red) mGluR3-CTD (from SI Appendix, Fig. S9B) taken over the last 19 residues (Y861-L879) as a function of LUV lipid composition (±SEM of this average; Wilcoxon test; ***P < 0.001).

Fig. 6.

CTD mutations that alter membrane binding affect mGluR3 internalization and function. (A) Schematics of mGluR3 CTD mutational positions and their effects on mGluR3-CTD free vs. membrane-bound equilibrium. Larger arrows show the direction in which each variant perturbs the equilibrium. (B) Quantification of the extent of receptor internalization for each mGluR3 variant (with dotted line denoting mGluR3 WT internalization) (averaged internalization per day, 10 to 12 images per condition/day and 4 to 9 d per condition; one-way ANOVA with multiple comparisons, *P < 0.05, ***P < 0.001). (C) Glutamate dose–response curves for mGluR3 variants in a patch-clamp experiment using GIRK currents as a reporter for mGluR3 G-protein activation (EC50: WT = 137 ± 27 nM, R843A = 51 ± 12 nM, G848E = 44 ± 9 nM, Y853A = 418 ± 65 nM, R869Q = 14 ± 4 nM, E870K = 102 ± 22 nM, CAAX = 157 ± 43 nM; F-test of EC50 shifts; **P < 0.01, ***P < 0.001). (D) Quantification of the extent of receptor internalization of WT mGluR3 vs. Y853D phospho-mimetic mutant vs. Y853F (averaged internalization of 10 images per condition/day across 3 d; t test; *P < 0.05). (E) Representative images of HEK293T cells expressing SNAP-tagged mGluR3 WT vs Y853A treated with 100 ng/mL EGF for 30 min (red arrows represent internalization). (Scale bar, 5 µm.) (F) Quantification of the extent of internalization for mGluR3 WT vs Y853A mutant in EGF or Glu+EGF incubated conditions (averaged internalization per day, 10 images condition/day and 3 to 4 d per condition; t test, **P < 0.01, n.s. P ≥ 0.05). (G) Working model of mGluR3-CTD free vs. membrane-bound equilibrium and changes that favor the less accessible membrane-bound (E870K, Y853F, anionic lipids) or the more accessible free (R843A, G848E, R869Q, Y853A/D, phosphorylation) state.