Direct effect of membrane environment on the activation of mGluR2 revealed by single-molecule FRET

Summary:

The microenvironment of membrane receptors controls their mobility, structure, interactions, and dynamics, but a systematic understanding of how it modulates receptor function is often lacking. Using single-molecule Förster resonance energy transfer (smFRET), we characterized how detergents and cholesterol modulate the conformational dynamics of metabotropic glutamate receptor 2 (mGluR2), a class C GPCR. We found that, within the resolution of our measurements, all tested detergents stabilize the same overall active and inactive structure of different domains of mGluR2. However, the degree of stabilization and the equilibrium between active and inactive conformations depended on the detergent. Detergents with a single hydrophobic tail increased the active state occupancy compared to those with long, branched tails. Adding cholesterol to micelles with branched hydrophobic tails shifted the equilibrium toward the inactive state. Mutagenesis identified residues potentially involved in cholesterol interaction with mGluR2. Targeting the cholesterol-binding site with synthetic molecules could be a viable therapeutic approach.

Graphical Abstract

Banerjee et al. used single-molecule FRET, and found that detergents and cholesterol shape the conformational dynamics of mGluR2, a class C GPCR, by shifting the active/inactive equilibrium. These findings highlight how the local environment controls GPCR activation and function. Targeting cholesterol-binding sites could provide new therapeutic strategies for modulating mGluR2 activity.

Introduction:

Membrane receptors are essential for transmitting external signals into the cell. In humans, G protein-coupled receptors (GPCRs) are the largest family of membrane receptors and play crucial roles in virtually all physiological functions.^1,2 The canonical model of GPCR activation suggests that receptors can adopt a variety of distinct conformational states, each with a distinct functional profile.^3,4 The relative distribution of these different receptor states is influenced by various factors, including ligand binding, interactions with signaling proteins, and the cellular environment, collectively determining the outcome of the signaling. In recent years and with the advances in structural biology the atomic structure of many GPCRs has been determined. These findings have provided structural insights into the mechanism of activation of GPCRs, and their conformational landscape.⁵ Additionally, new spectroscopic methods and progress in chemical biology tools and biosensors have elaborated different aspects of GPCR signaling pathways and regulation of GPCR signaling in cells.^6–8

Among recent advances is the growing evidence for the vital roles that lipids play in regulating ligand-induced GPCR signaling.⁹ Generally, the effect of lipids on a receptor structure and function is thought to be either through direct and specific interaction with the receptor^10–13 or via altering physical properties of membrane such as membrane curvature and thickness.^14–17 The impact of membrane composition on GPCR function and its mechanism of action varies depending on the receptor,^18,19 and for many receptors, the effect is unknown and uncharacterized. Elucidating lipid membrane protein interactions will enable a more holistic interpretation of protein structure-function, potentially resulting in the development of more effective therapeutics.

Metabotropic glutamate receptor 2 (mGluR2) is a member of class C GPCRs.^20,21 All the class C receptors function as obligate dimer and have a large extracellular domain (ECD) that contains the orthosteric ligand binding site. Recent structures of members of family C GPCRs have revealed the existence of specific lipids at the transmembrane domain (TMD),^22–24 suggesting possible functional roles for lipids in controlling function of class C GPCRs.²⁵ Biochemical and signaling assays have also shown that detergents can affect stability of purified mGluR2²⁶ and specific amino acids in the calcium-sensing receptor (CaSR) are shown to be important for its interaction with cholesterol.²³ Despite this evidence, a systematic characterization of the effect of the local environment on the conformation of class C GPCRs is absent.

In this study, we used single-molecule fluorescence resonance energy transfer (smFRET)^3,27 to characterize the conformational dynamics of mGluR2 in various nonionic detergents and in the presence or absence of cholesterol. We found that, in the presence of the orthosteric agonist glutamate and within the resolution of our measurement, different detergents stabilize a similar active conformation of mGluR2 but to varying degrees. Moreover, in some detergents cholesterol exhibited negative allosteric modulator-like behavior. Our findings directly show how the local environment of mGluR2 affects its activation, and targeting these specific cholesterol-binding sites could be a promising drug targeting opportunity.

Results:

Different detergents stabilize the same active conformation of mGluR2 to different levels.

To characterize the effect of the local environment on mGluR2 conformation, we first performed smFRET in various detergents (Figure 1) with a mGluR2 sensor that reports the conformation of the Venus flytrap domain (VFD).²⁸ The VFD was tagged with SNAP-tag after the signal peptide, and smFRET changes were measured using donor and acceptor fluorophores conjugated to benzyl guanine (Figure 1I, J). Structural studies and previous smFRET experiments in Dodecyl-β-D-maltoside (DDM) confirmed that the distance at the N-terminal domain of VFD increases upon receptor activation.^28,29 We observed two distinct FRET states at 0.45 and 0.28 in the presence of the antagonist LY341495 or 1 mM glutamate (Figure 2A), respectively, corresponding to the inactive and active conformations.²⁸ Next, we wondered whether different detergents stabilized the same active and inactive conformational states. To test this, we performed experiments in the presence of Tridecan-3-yloxy-β-D-maltoside (Mal(11.2)), Octyl Glucose Neopentyl Glycol (OGNG), Lauryl Maltose Neopentyl Glycol (LMNG), and Tandem malonate glucoside-A13 (TMG) or Glyco-diosgenin (GDN) (Figure 1). These detergents have different chain length and size and are commonly used in protein purification and structural studies of membrane proteins.³⁰

Figure 1. Structures of the detergents and mGluR2 used in this study.

(A) DDM, (B) Mal (11.2), (C) LMNG, (D) OGNG, (E) TMG, (F) GDN, (G) Cholesterol and (H) Cholesterol hemisuccinate (CHS). (I) Inactive (left) and active (right) structures of mGluR2, showing different receptor domains. (J) Cartoon representation of mGluR2, illustrating the various domains and the SNAP-tag used for attaching fluorophores in the smFRET experiment for VFD.

Figure 2: The variability in the active state percentage of mGluR2 across various nonionic detergents.

(A) smFRET population histograms of the VFD at 2 µM LY341495 (black), 10 µM (red), and 1 mM (blue) glutamate in DDM micelle. Similar plots for mGluR2 in other micelles are shown in (B) Mal, (C) OGNG, (D) LMNG, (E) TMG, and (F) GDN, respectively. (G) smFRET population histogram of VFD of mGluR2 at 10 μM glutamate in various detergents. Dashed lines showing the position of the active and inactive peaks. (H) The percentage of the active state population in different detergents at 10 μM glutamate. Significance was tested by non-parametric ANOVA, Kruskal-Wallis, *P < 0.05. (I) Cross-correlation of donor and acceptor intensities at 10 μM glutamate in different detergents. Data was acquired at 50 ms time resolution. The error bars for all panels represents the ± standard error of the mean (SEM, n=3).

We observed that mGluR2 exhibited the same FRET peaks in all tested detergents, suggesting that they all stabilize the same active and inactive conformations within the resolution of smFRET measurements (Figure 2A–F, S1). Interestingly, at the intermediate concentration of 10 µM glutamate, different detergents stabilized the active state to different degrees (Figure 2G, H). Specifically, while the occupancy of the active state at 10 µM glutamate was highest in DDM at 73%, it decreased to 61% and 43% for LMNG and TMG (Figure 2H). Finally, in the case of GDN, the active state population was only 28% at 10 µM glutamate (Figure 2H). To confirm that the observed effects are due to the detergents themselves, rather than the selective extraction of specific lipids during protein purification we performed detergent exchange experiments. We observed that receptors that were originally purified in DDM and then exchanged to LMNG showed a profile similar to the receptor purified in LMNG and upon exchanging back into DDM buffer they exhibited a profile similar to DDM sample (Figure S1D).

Next, to characterize the receptor dynamics we performed cross-correlation analysis between donor and acceptor signals across various detergents at intermediate glutamate concentration. This analysis showed similar amplitude for all detergents, except for GDN which exhibited a notably higher cross-correlation magnitude. This suggests that the VFD of mGluR2 is more dynamic within the GDN micelles compared to other detergents tested (Figure 2I). Taken together these data suggest that while different detergents stabilize the same conformational states, the degree of stabilization depends on the detergent with branched chains stabilizing the receptor in the active state to lesser degree.

Cholesterol modulates the conformation of the VFD in branched chain detergents.

Cholesterol is a major component of cell membranes and is essential for the maintenance of membrane fluidity and thickness and the formation of lipid rafts.³¹ Previous structures of a closely related receptor to mGluR2, the calcium-sensing receptor (CaSR), showed an association of cholesterol with the TMD of CaSR.²³ Therefore, we aimed to determine whether cholesterol could also associate with and influence mGluR2 dynamics in our system. To test this, we performed smFRET experiments in the presence of detergents and cholesterol hemisuccinate (CHS), a close analog of cholesterol. First, we tested DDM:CHS ratio of 10:1 and 10:2 (%w/w) and did not observe a significant difference in the occupancy of states in the presence or absence of CHS (Figure 3A, S2A). Similarly, we did not observe an effect for CHS in Mal or OGNG (Figure S2B–E) Interestingly, however, we observed that the addition of CHS in LMNG resulted in a significant reduction of the active state (Figure 3B, S3A). For example, at 10 µM glutamate the occupancy of active state reduced from 61% in LMNG to 21% in LMNG:CHS (10:2) micelles (Figure 3E). At 10 µM glutamate we also observed a reduction of active state occupancy in the presence of CHS for TMG and GDN as well. In TMG micelles, the changes were from 43% to 23%, and in GDN, from 28% to 9% (Figure 3C–E, S3B, C). Similarly, at 1 mM glutamate, we observed a decrease in active state population of mGluR2 in LMNG micelles in the presence of CHS (Figure 3F).

Figure 3. Addition of CHS in detergents with branched hydrophobic chains decreases the occupancy of active state.

(A) smFRET population histograms of the VFD in the presence (red) and absence (black) of CHS at 2 µM LY341495, 10 μM, and 1 mM glutamate in DDM, (B) LMNG, (C) TMG, and (D) GDN micelles. (E) Percentage of the active state population in different detergents in the absence and presence of CHS at 10 μM glutamate, or (F) 1 mM glutamate. Significance was tested by the t-test. The error bars for all panels represents the ± standard error of the mean (SEM, n=3). *P < 0.05, **P < 0.01, ***P < 0.001.

Together these results suggest that CHS in the presence of long chain branched detergents exhibit negative allosteric modulator (NAM) type behavior.

The conformational dynamics of cysteine-rich domain of mGluR2 is influenced by detergents.

During the activation process of mGluR2, ligand binding at the VFD induces a local conformational change that propagates downstream to the CRD and TMD to activate G proteins.^27,28 Considering the loose conformational coupling between the VFD and CRD,³ we wondered if the observed effect of detergents and CHS are also detectable at the CRD. To test this, we performed smFRET experiments with a CRD sensor³ to directly quantify the conformational changes of the CRD in LMNG, TMG, and GDN micelles in the absence and presence of CHS (Figure 4, S4).

Figure 4. CRD data shows the effect of CHS on the activation of mGluR2 and supporting VFD results.

(A) Schematic representation of mGluR2 showing the various domains and the location of the fluorophore to measure the dynamics of the CRD. (B) Single-molecule time traces of donor (green) and acceptor (red) and corresponding FRET value (blue) in LMNG (10)-CHS (2) micelle at 2 µM LY341495, 10 µM glutamate and 1 mM glutamate. Data was acquired at 50 ms time resolution. (C) smFRET population histograms of the CRD in the presence (red) and absence (black) of CHS at concentrations of 2 µM LY341495, 10 μM, and 1 mM glutamate in LMNG, (D) TMG, (E) GDN micelles. (F) The percentage of the active state population in the absence and presence of CHS in LMNG, (G) TMG or (H) GDN micelles. Significance was tested by the t-test. The error bars for all panels represents the ± standard error of the mean (SEM, n=3). *P < 0.05, **P < 0.01, ***P < 0.001.

First, we characterized mGluR2 dynamics in the LMNG micelles. We found that the receptor occupies multiple conformational states and state occupancy shifted towards higher FRET in the presence of glutamate, consistent with the structural studies and our previous work in DDM-CHS³ (Figure 4C). In LMNG and in the presence of antagonist, CRD showed a broad conformational distribution and a dominant low FRET state (Figure 4C). In the presence of 10 µM and 1 mM glutamate, the peak position shifted to a high FRET state, corresponding to the more compact active state of mGluR2 (Figure 4C). The behavior of the CRD domain in TMG detergent was different (Figure 4D). In the TMG micelles and in the presence of antagonist, we observed a relatively sharp peak at a low FRET state which shifted to higher FRET state in the presence of glutamate (Figure 4D). Finally, we characterize the behavior of the CRD domain in GDN micelles. In GDN, the CRD showed a sharp peak in the presence of antagonist as well as 10 µM glutamate, which even at 1 mM glutamate did not reach the high FRET levels as in LMNG and TMG (Figure 4E). Together these data suggest that the occupancy and stabilization of the CRD active state at a specific glutamate concentration depends on the detergent.

Next, we tested the effect of CHS in these three detergents. In all cases, the addition of CHS reduced the occupancy of the active state at the intermediate or saturating glutamate concentrations (Figure 4F–H). Taken together, these findings provide further validation for the observations made in the VFD that cholesterol exhibits NAM-like effect in some detergents.

Lastly, experiments using the VFD and the CRD sensor were repeated with the soluble cholesterol instead of CHS. In this case, methyl-β-cyclodextrin (MβCD) is used to directly solubilize and deliver cholesterol. The results showed that water-soluble cholesterol also exhibits a NAM-like effect similar to what we observed with CHS (Figure S5A–D). Importantly, we did not observe any effect in the presence of MβCD alone (Figure S5E), further establishing that the observed effect is due to the incorporation of cholesterol in micelles. Finally, as a control we performed experiments with another cholesterol analog, 27-hydroxycholesterol. We observed a similar trend where the presence of 27-hydroxycholesterol increased the inactive state population (Figure S5F).

To confirm that the observed effects are due to CHS and not the selective extraction of specific lipids during protein purification, we first purified mGluR2 using LMNG-CHS and then exchanged it to LMNG. We observed that receptors that were originally purified in LMNG-CHS and then exchanged to LMNG showed a profile similar to the receptor purified in LMNG (Figure S5G).

Specific residues at the transmembrane dimeric interface of mGluR2 confer its functional sensitivity to cholesterol

Our results with the VFD and CRD sensors clearly showed that addition of cholesterol in some detergent micelles such as in LMNG and TMG can reduce the occupancy of the active state. As most allosteric modulators identified for class C GPCRs act within the seven-transmembrane (7TM) region,^32,33 we hypothesized that cholesterol could potentially interact with the transmembrane domain of mGluR2 in the similar manner that was observed with GABAB receptor²² and CaSR.^24,34 Previous molecular dynamics simulation using a homology model of mGluR2 structure based on the crystal structure of monomeric mGluR1 truncated TMD showed that cholesterol could interact with multiple positions on mGluR2.³⁵ Recent structure of CaSR²³ also revealed the presence of cholesterol at the dimer interface interacting with specific residues including F809, L812, I813 and mutations at these positions affected CaSR signaling.²³ We, therefore, focused the corresponding residues in mGluR2 as possible cholesterol-binding amino acids (Figure 5A, B). Specifically, we identified three probable binding sites for cholesterol, one in helix V (L744) and two in helix VI (F764 and Y781) at the dimer interface (Figure 5B). We mutated each of those to an alanine and performed the smFRET experiments in LMNG and LMNG-CHS micelles and compared the conformational distributions in wildtype and mutant receptors. First, we found that in the LMNG-CHS micelles, the wildtype and mutant receptors exhibit the same FRET distribution for the saturating glutamate (active) and saturating antagonist (inactive) conditions (Figure 5C–E). Therefore, these mutations do not alter the active and inactive conformations of mGluR2, within the resolution of FRET measurements. Interestingly at the intermediate glutamate conditions we found that while L744A mutation does not change the effect of CHS on the receptor (Figure 5C), in the F764 and Y781 mutations the NAM-like effect of CHS that we observed in the WT receptor was significantly reduced (Figure 5D–G). Control experiments in LMNG only micelles did not show any significant change in the occupancy of the active state between wildtype and mutant receptors further supporting the notion that the effect of mutants is on CHS binding (Figure 5H, S5H–S5J). Together these results are consistent with the idea that cholesterol in the detergent environment binds to the receptor at specific locations to exhibit its NAM-like effect.

Figure 5. Multiple residues contribute to interaction of CHS with mGluR2.

(A) The active state structure of mGluR2 (green, 7MTR), overlaid with a CaSR monomer (pink, 7SIM) bound with cholesterol (yellow). (B) Overlaid structure of the transmembrane domain of mGluR2 (green) and CaSR with cholesterol (pink and yellow), displaying mGluR2 residues that could interact with cholesterol. (C) smFRET population histograms of the VFD in LMNG (10)-CHS (2) micelles at 2 µM LY341495, 10 μM, and 1 mM glutamate for wild-type mGluR2 (red) and L744A mutant (black), (D) F764A mutant (black), or (E) Y781A mutant (black). (F) smFRET population histogram at 10 mM glutamate in LMNG (10)-CHS (2) micelle for wild-type mGluR2 (red), and the mutants, L744A (green), F764A (blue), and Y781A (black) and wild type mGluR2 (yellow) in LMNG alone micelle. The error bar represents the ± standard error of the mean (SEM, n=9, three biological replicates). (G) The percentage of active states for both wild-type and mutants at 10 μM glutamate in LMNG (10)-CHS (2) micelle. Significance was tested by non-parametric ANOVA, Kruskal-Wallis, *P < 0.05. The error bar represents the ± standard error of the mean (SEM, n=3). (H) The percentage of active states for both wild-type and mutants at an intermediate glutamate concentration of 10 μM in LMNG micelle.

Discussion:

Approximately 30% of the proteins in mammalian cells are transmembrane proteins, which play crucial roles in signaling, cellular adhesion and motility, metabolism, and sensing. The local environment surrounding membrane proteins is increasingly recognized as a crucial factor in regulating their function.^31,36 This regulation occurs either through direct interactions between specific lipids and particular amino acid residues within the protein or by altering the properties of the membrane itself. Therefore, understanding the mechanisms of this functional modulation is essential for describing how membrane proteins are activated, as well as for developing novel therapeutic strategies.

During the activation of mGluR2, the glutamate binding at the VFD results in a local conformational change that propagates downwards through the CRD domain to the TMD to activate the receptor. In this study we systematically characterized the impact of commonly used detergents on the global conformational dynamics of mGluR2 at two different domains, using smFRET. We found that while different detergents stabilize the same active and inactive conformations of the VFD and the CRD domains, the degree of stabilization of the active state depends on the detergent used. Our results support the model that the number of hydrophobic chains and their length determines the degree of stabilization of the active state. While our work does not provide a mechanism for the observed effects, previous work has shown that the degree of packing of detergent molecules within the receptor-detergent micelle is a key factor in the stabilization of the receptor dynamics.^37,38

The distinct effects of LMNG and GDN on receptor conformational states could be attributed to their unique structural properties. LMNG has been extensively used to purify GPCRs, and class C GPCRs specifically, and for structural studies. LMNG has two hydrophilic maltoside head groups and two hydrophobic chains to mimic the structure of lipids better than classical detergents with a single tail. This architectural feature likely enables LMNG to form a tightly packed micelle around the receptor and to potentially stabilize the receptor conformations.³⁷ LMNG form much larger micelles compared to DDM (hydrodynamic radius ~10 nm vs ~3 nm) with very low critical micelle concentrations (CMCs) mainly due to the presence of two alkyl chains. The low CMC makes LMNG desirable for cryo-EM studies because it can solubilize proteins at a lower concentration.³⁹ Previous work using NMR to quantify the conformational dynamics of β₂AR found that receptor dynamics is distinct in LMNG compared to DDM.⁴⁰ Specifically, receptors in LMNG were more stable and showed improved resolution of the dynamics and slower exchange between conformational states compared to DDM, likely due to significantly lower detergent off-rate for LMNG.⁴⁰ In contrast, GDN, which was introduced as a synthetic substitute for digitonin, is structurally distinct from other conventional detergents. It has a high hydrophobic density with a steroid based group attached to branched di-maltose hydrophilic head group.⁴¹ GDN forms spherical or ellipsoidal micelles, similar to DDM, which are larger than DDM micelles.⁴² High hydrophobic density and rigidity of GDN promotes enhanced interactions between detergent molecules. GDN is very effective at solubilizing membrane proteins and has been widely used for purification and structural studies of GPCRs, ion channels, transporters, and other membrane proteins. In contrast, LMNG has been mostly used to study GPCRs.⁴³ Understanding how detergents modulate the conformation of membrane proteins can provide insights into how lipids, and the local membrane environment in general, affect their structure and function.

We also explored the effect of cholesterol in the detergent context on the activation of mGluR2. Previous work has shown that cholesterol can affect signaling in many GPCRs through direct binding to the receptor or changing the local biophysical properties of the membranes.^14,44 For mGluR2 specifically, molecular dynamics simulations has suggested that cholesterol can interact with multiple sites on the receptor.³⁵ We discovered that in the presence of detergents with hydrophobic branched chains, such as LMNG and TMG, CHS exhibits an NAM-like effect. Through mutagenesis, we identified two residues, F764 and Y781, that are important for the effects of cholesterol. This suggests the presence of multiple cholesterol binding sites on mGluR2. The existence of such binding sites could motivate the development of novel pharmacological agents targeted at these sites to finely tune receptor signaling.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to the lead contact, Reza Vafabakhsh (reza.vafabakhsh@northwestern.edu)

Materials availability

All plasmids are available upon request, subject to a material transfer agreement (MTA), from the lead contact.

STAR * Methods

Experimental model and study participant details

Cell Culture

Mycoplasma-free HEK293T cells (Sigma) were used to express mGluR2. The cell culture was performed with DMEM (Corning) supplemented with 10% (v/v) fetal bovine serum (Fisher Scientific), 100 unit/mL penicillin-streptomycin (Gibco), and 15 mM HEPES (pH = 7.4, Gibco) at 37°C and 5% CO2. We used 0.05% trypsin-EDTA (Gibco) during the cell passage. To express UAA-containing protein, the cell culture media was supplemented with 0.6 mM 4-azido-L-phenylalanine (Chem-Impex International) for an hour before the transfection. Before use, all media were sterilized using a 0.2 µM PES filter (Fisher Scientific).

Bacterial strains

Escherichia coli DH5α cells were used for cloning and transformation of plasmids used. The bacteria were grown in Luria Broth (LB) media for large-scale production of the plasmid. All experiments were done with HEK293T cells.

Method details

Transfection and protein expression

HEK293T cells were cultured on poly-d-lysine-coated 18 mm glass coverslips (VWR) a day before the transfection. For SNAP-tag mGluR2, the cell were transfected using lipofectamine 3000 (Fisher Scientific). The total concentration of 1 µg per 18 mm coverslip was used. After 24 hours of transfection, media was replaced with a fresh media and grown for another 24 hours. The cells were washed with recording buffer solution containing (in mM): 128 NaCl, 2 KCl, 2.5 CaCl₂, 1.2 MgCl₂, 10 sucrose, 10 HEPES, pH = 7.4, and labeled with SNAP-tag dyes.

UUA-containing mGluR2 was used to monitor the CRD. HEK293T cells were cultured in a 60 mm flask, and 1 hour before the transfection, the media was replaced with fresh media containing 0.6 mM 4-azido-L-phenylalanine. Co-transfection of mGluR2 plasmids with an amber codon (azi-CRD) and pIRE4-Azi plasmid (gift from Irene Coin, Addgene plasmid #105829) was performed using transporter 5 at a total plasmid concentration of 9 µg per 60 mm dishes. After 24 hours, the media containing 0.6 mM 4-azido-L-phenylalanine was changed to a fresh media with 0.6 mM 4-azido-L-phenylalanine and further grew for 20 hours. On the day of the experiment, the media was replaced with DMEM media and incubated for 30 minutes before labeling. The cells were washed with recording buffer and labeled with alkyne dyes.

SNAP-tag labeling in VFD for FRET measurements

To label the SNAP-mGluR2, the cells were incubated with 2 µM of SNAP-Surface Alexa Fluor 549 (NEB) and 2 µM of SNAP-Surface Alexa Fluor 647 (NEB) in recording buffer at 37°C for 30 minutes. The cover slip was washed twice in the recording buffer to remove the excess unbound dyes before harvesting the cells.

UAA labeling in CRD by azide-alkyne click chemistry

The azide-alkyne click chemistry-based UAA labeling procedure described earlier. ^3,45 In Brief: First, stock solutions were made: Cy3 and Cy5 alkyne dyes (Click Chemistry Tools) were made at a concentration of 10 mM in DMSO, BTTES (Click Chemistry Tools) at 50 mM, copper (II) sulfate (Sigma) at 20 mM, aminoguanidine (Cayman Chemical) at 100 mM, and (+)-sodium L-ascorbate (Sigma) at 100 mM in ultrapure distilled water (Invitrogen). Cy3 and Cy5 alkyne dyes were added in 656 µL of recording buffer to achieve a concentration of 18 µM for each dye to create the labeling mixture. A freshly prepared mixture of copper (II) sulfate and BTTES, at a molar ratio of 1:5, was added to achieve final concentrations of 150 µM for copper (II) sulfate and 750 µM for BTTES. Subsequently, aminoguanidine was added to reach a final concentration of 1.25 mM, and (+)-sodium L-ascorbate was added to bring its concentration to 2.5 mM, resulting in a total mixture volume of 0.7 mL. This calculation is based on an 18 mm coverslip; however, for the 60 mm dishes, the ratios were adjusted accordingly.

The mixture was incubated for 8 minutes at 4°C, followed by a 2-minute incubation at room temperature. Before adding the mixture to the cells, the cells were rinsed with the recording buffer. Throughout the labeling process, the cells were kept in the dark inside the incubator operating at 37°C and 5% CO2. After labeling, the cells were washed with a recording buffer to remove any excess dye.

Cloning and Plasmids

The C-terminal FLAG-tagged mouse mGluR2 construct in the pcDNA3.1(+) vector was obtained from GenScript (ORF clone: OMu19627D) and its sequence confirmed by sequencing. The fulllength mGluR2 construct containing an amber codon (TAG) at position A548 (azi-CRD), along with an N-terminal SNAP-tag (SNAP-mGluR2), were created using site-directed mutagenesis.³ Alanine mutagenesis of N-terminal SNAP-tag (SNAP-mGluR2) at the positions L744, F764, and Y781 were done using the QuikChange site-directed mutagenesis kit from Agilent. The forward and reverse primers are tabulated in the key resource table.

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
FLAG-tag antibody	Genscript	A01429
Chemicals, peptides, and recombinant proteins
Trolox	Sigma	53188-07
NeutrAvidin	Thermo Scientific	31000
Protocatechuic acid (PCA)	HWI group	99-50-3
SNAP-Surface 549	New England Biolabs	S9112S
SNAP-Surface Alexa Fluor 647	New England Biolabs	S9136S
DDM	Anatrace	Cas# 69227-93-6
Mal (11.2)	Avanti	Cas# 1219008-64-6
LMNG	Anatrace	Cas# 1257852-96-2
OGNG	Anatrace	Cas# 1257853-32-9
TMG	Avanti	SKU850558P
GDN	Anatrace	Cas#1402423-29-3
Water soluble Cholesterol	Sigma-Aldrich	C4951
Cholesterol hemisuccinate tris salt	Anatrace	Cas# 102601-49-0
27-hydroxycholesterol	Sigma-Aldrich	Cas# 20380-11-4
Glutamate	Sigma Aldrich	Cat # 6106-04-3
Cy3 Alkyne	Click Chemistry Tools	TA117-5
Cy5 Alkyne	Click Chemistry Tools	TA116-5
4-azido-L-phenylalanine	Chem-Impex International	Cat # 06162
Aminoguanidine (hydrochloride)	Cayman Chemical	81530
BTTES	Click Chemistry Tools	1237-500
Copper (II) sulfate	Sigma Aldrich	Cat # 451657-10G
(+)-Sodium L-Ascorbate	Sigma Aldrich	Cat # 11140-250G
Sodium Pyruvate	Gibco	11360-070
DMEM	Corning	10-013-CV
Defined Fetal Bovine Serum	Thermo Fisher Scientific	SH30070.03
Penicillin-Streptomycin	Gibco	15140-122
Lipofectamine 3000 Transfection Reagent	Thermo Fisher Scientific	L3000015
transporter 5 Transfection Reagent	Polysciences	26008
protocatechuate 3,4-dioxygenase (rPCO)	Oriental Yeast Co	N/A
Poly-D-lysine hydrobromide	Sigma Aldrich	27964-99-4
Experimental models: Cell lines
HEK 293T	Sigma Aldrich	Cat # 12022001
Oligonucleotides
M2SNAP_Y781A-F GCTGGCTTTCCTTCCCATCTTCGCCGTCACCTCCAGTGATTATCGG	This paper	N/A
M2SNAP -Y781A-R CCGATAATCACTGGAGGTGACGGCGAAGATGGGAAGGAAAGCCAGC	This paper	N/A
M2SNAP_L744A-F CATCGCTCTCTGCACGGCCTATGCCTTCAAGACCCGC	This paper	N/A
M2SNAP_L744A-R GCGGGTCTTGAAGGCATAGGCCGTGCAGAGAGCGATG	This paper	N/A
M2SNAP_F764A-F CGAGGCCAAGTTCATCGGCGCTACCATGTACACCACCTGTATC	This paper	N/A
M2SNAP_F764A-R GATACAGGTGGTGTACATGGTAGCGCCGATGAACTTGGCCTCG	This paper	N/A
Recombinant DNA
azi-CRD	Liauw et al.3 (Ref. 3)	N/A
pIRE4-Azi	Addgene	Plasmid # 105829
Software and algorithms
smCamera (Version 1.0)	N/A	https://github.com/Ha-SingleMoleculeLab/smCamera2
OriginPro (2020b)	OriginLab	https://www.originlab.com/
Matlab R2022b	Mathworks	https://se.mathworks.com/products/new_products/release2022b.html
Adobe Illustrator (2022)	Adobe	https://www.adobe.com/

smFRET measurements

The flow chamber to perform the smFRET experiments was made first by passivating the glass slide (Fisher Scientific) and cover slip (VWR). The passivation of coverslip and slides was done by coating with mPEG (Laysan Bio) and 1% (w/w) biotin-PEG. The flow chambers were made by sandwiching double-sided tape between the glass slides and the coverslip.⁴⁵ The flow cell was incubated with 500 nM NeutrAvidin (Thermo Scientific) for 2 minutes, followed by 20 μM biotinylated FLAG-tag antibody (A01429, GenScript) for 30 minutes. The flow cell was washed thoroughly using T50 buffer (50 mM NaCl, 10 mM Tris, pH 7.4) to remove unbound NeutrAvidin and antibodies.

After labeling with SNAP dyes, cells were detached from an 18 mm coverslip by gently pipetting with recording buffer. The cells were subsequently pelleted by centrifugation at 4°C for 10 minutes at 4000 g. The supernatant was discarded, and the cells were resuspended in 100 µL of lysis buffer. This buffer contained 200 mM NaCl, 50 mM HEPES, 1 mM EDTA, a protease inhibitor tablet (Thermo Scientific), and various detergents or detergents with CHS (at 50 times the critical micelle concentration of detergents) and pH 7.4. The cells were lysed while gently rotating at 4°C for 2 hours. Finally, the lysate was centrifuged at 20,000 g and 4°C for 20 minutes. DDM, LMNG, OGNG, GDN, and Cholesterol hemisuccinate were purchased from Anatrace, and TMG, Mal(11.2) was procured by Avanti Lipid.

The supernatant was collected and diluted in imaging buffer (in mM) 128 NaCl, 2 KCl, 2.5 CaCl₂, 1.2 MgCl₂, 40 HEPES, 4 mM Trolox, and different amounts of detergents (2 times of CMC) at pH 7.4. We employed a pull-down approach to attach the receptor to the surface of the coverslip. The diluted sample was introduced into the flow chamber for sparse surface immobilization of labeled receptors using their C-terminal FLAG tags (around 400 molecules in the field of view). After achieving optimal receptor coverage, the flow chamber was extensively washed (>20× chamber volume) with the imaging buffer. The labeled receptors were finally imaged in an imaging buffer containing detergents at 2 times the critical micelle concentration (CMC) and an oxygen scavenging system, which included protocatechuic acid (Sigma) and 1.6 U/mL bacterial protocatechuate 3,4-dioxygenase (rPCO) (Oriental Yeast Co.), at pH 7.35. All reagents were purchased from Sigma. The samples were imaged with a 100× objective (Olympus, 1.49 NA, oil immersion) on a custom-built TIR microscope with 50 ms time resolution. Donor and acceptor excitation were achieved using 532 nm and 638 nm lasers (RPMC Lasers).

smFRET data analysis

The single-molecule FRET data was analyzed using smCamera software (http://ha.med.jhmi.edu/resources/) along with custom-modified MATLAB code. The selection of particles was based on particles in the acceptor channel that were 10% brighter than the background signal upon excitation of the donor particles. The intensities of the donor and acceptor were monitored across all frames for all of the selected particles. Selection criteria included the identification of particles demonstrating a single donor and acceptor bleaching step within the acquisition period. Additionally, stable total intensity $\left(I_D+I_A\right)$, along with anticorrelated behavior between donor and acceptor intensities without blinking occurrences, and a minimum duration of 3 s were required for selection, comprising approximately 10%–15% of total molecules per movie. The acceptor intensity was corrected due to leakage of the donor intensity into the acceptor channel according to $I_{correct}=I_A-0.088\times I_D$, where $I_D$ and $I_A$ denote the raw donor and acceptor intensities, respectively. The apparent FRET efficiency was derived using the formula $\left(I_{correct}\right)/\left(I_D+I_{correct}\right)$. Each experiment was performed in triplicate unless specified otherwise. Before trace compilation, FRET histograms of individual particles were normalized to 1 to ensure equal contribution, irrespective of trace length. Error bars on histograms represent the standard error of the mean. OriginPro was used to perform peak fitting of smFRET histograms, utilizing 2 Gaussian distributions using: $y\left(x\right)={\sum }_{i=1}^n\frac{A_i}{w_i\sqrt{\frac{\pi }{2}}}{e}^{-2\frac{{\left(x-x{c}_i\right)}^2}{w_i^2}}$ where ‘n’ represents the number of Gaussians, ‘A’ denotes the peak area, ‘xc’ stands for the FRET peak center, and ‘w’ refers to the full-width half maximum for each peak. The Levenberg-Marquardt algorithm was employed for peak fitting in OriginPro, with a Chi-square tolerance of 1E-9 used to determine the best fit. The cross-correlation (CC) of donor and acceptor intensity traces at time $\text{τ}$ is defined as $\text{CC}\left(\text{τ}\right)=\text{δ}{I}_D\left(\text{t}\right)\text{δ}{I}_A\left(t+\tau \right)/(I_D〉+〈I_A)$, where $\text{δ}{I}_D\left(\text{t}\right)=I_D\left(\text{t}\right)-〈I_D〉$, and $\text{δ}{I}_A\left(\text{t}\right)=I_A\left(\text{t}\right)-〈I_A〉$. $〈I_D〉$ and $〈I_A〉$ represent the time-averaged donor and acceptor intensities, respectively. The cross-correlation calculations were conducted using the same traces employed in preparing the single-molecule population histograms. The cross-correlation data were fitted with a single exponential decay function $y\left(x\right)=y_0+A$. $e^{-x/\tau }$ in OriginPro (OriginLab).

Active state population

The percentage of active state for experiments with the VFD sensor was calculated by fitting the FRET histogram with the Gaussian function of two peaks. Values represent the area under each FRET peak from the smFRET histogram as a fraction of the total area. For the CRD sensor, the FRET peak for the active state is at 0.89, with a distribution ranging from ~0.8 to 1.³ For simplicity, we calculated the active state population by calculating the area under the FRET curve for FRET ≥ 0.8.

Quantification and statistical analysis

The statistical details of our analyses can be found in the figure legends, where applicable. All the significance was tested either by t-test or by non-parametric ANOVA, Kruskal-Wallis test using OriginPro. The values are shown as mean ± standard error of the mean (SEM) from 3 independent experiments (n=3). ns = not significant; *p < 0.05, **p < 0.01, ***p < 0.001.

Highlights:

Detergents influence the activation dynamics of mGluR2.

cholesterol allosterically modulates the activation of mGluR2 in some detergents.

Cholesterol directly interacts with mGluR2 at the dimer interface.

Data and code availability

All data reported in this paper will be shared by the lead contact upon request.

This paper does not report original code.

Any additional information, data, or code utilized in this paper will be made available from the lead contact upon request.