Diffusion and Oligomerization States of the Muscarinic M₁ Receptor in Live Cells—The Impact of Ligands and Membrane Disruptors

Abstract

G protein-coupled receptors (GPCRs) are a major gateway to cellular signaling, which respond to ligands binding at extracellular sites through allosteric conformational changes that modulate their interactions with G proteins and arrestins at intracellular sites. High-resolution structures in different ligand states, together with spectroscopic studies and molecular dynamics simulations, have revealed a rich conformational landscape of GPCRs. However, their supramolecular structure and spatiotemporal distribution is also thought to play a significant role in receptor activation and signaling bias within the native cell membrane environment. Here, we applied single-molecule fluorescence techniques, including single-particle tracking, single-molecule photobleaching, and fluorescence correlation spectroscopy, to characterize the diffusion and oligomerization behavior of the muscarinic M₁ receptor (M₁R) in live cells. Control samples included the monomeric protein CD86 and fixed cells, and experiments performed in the presence of different orthosteric M₁R ligands and of several compounds known to change the fluidity and organization of the lipid bilayer. M₁ receptors exhibit Brownian diffusion characterized by three diffusion constants: confined/immobile (∼0.01 μm²/s), slow (∼0.04 μm²/s), and fast (∼0.14 μm²/s), whose populations were found to be modulated by both orthosteric ligands and membrane disruptors. The lipid raft disruptor C6 ceramide led to significant changes for CD86, while the diffusion of M₁R remained unchanged, indicating that M₁ receptors do not partition in lipid rafts. The extent of receptor oligomerization was found to be promoted by increasing the level of expression and the binding of orthosteric ligands; in particular, the agonist carbachol elicited a large increase in the fraction of M₁R oligomers. This study provides new insights into the balance between conformational and environmental factors that define the movement and oligomerization states of GPCRs in live cells under close-to-native conditions.

Introduction

G protein-coupled receptors (GPCRs), a large family of seven transmembrane domain proteins, are essential components of signaling networks throughout the body which trigger complex cellular responses to subtle environmental clues.¹ Signaling occurs when a ligand (agonist), binding at the extracellular surface of a GPCR, induces long-range conformational changes in the receptor, which in turn promote interaction with and then subsequently activate the cognate G protein.² Localized at the cell surface, GPCRs are easily accessible to extracellular therapeutics that can activate or inhibit intracellular reaction cascades, thus making them ideal drug targets. Unsurprisingly, more than one-third of all approved drugs—treating cancer, cardiac dysfunction, diabetes, obesity, inflammation, asthma, pain, and neuropsychiatric disorders—act on members of this diverse class of proteins.³ However, major knowledge gaps remain with regard to identifying new drugs that are tissue and subtype specific while eliciting the desired downstream signaling pathways.

According to the classic textbook view, a monomeric GPCR couples to a single heterotrimeric G protein in a process promoted by agonist binding.⁴ In recent years, a different view has emerged that indicates many GPCRs form transient or stable, homo- or hetero-oligomers, and that those oligomers fulfill physiological roles.⁵ In previous studies, using single-molecule photobleaching (smPB) and fluorescence correlation spectroscopy (FCS), we found that the muscarinic M₂ receptor forms tetramers and that M₂ oligomers couple to G_i1 protein oligomers in a ligand-dependent manner in live cells.^6,7 Raicu et al. used spectral Förster resonance energy transfer (FRET) imaging to infer rhombic tetramers of M₂,⁸ and, more recently, fluorescence intensity fluctuation spectrometry to quantify the oligomeric sizes of the epidermal growth factor receptor tyrosine kinase and of the human secretin receptor (SecR).⁹ A recent study on opioid receptors using a combination of single-molecule colocalization and the split GFP assay showed that the kappa receptor forms dimers even at low densities (<5/μm²), while delta and mu remain monomeric even at high densities (>25/μm²).¹⁰

Single-particle tracking (SPT) and single-molecule FRET in live and fixed cells dissected the homodimerization of representative class A, B, and C GPCRs, i.e., the μ-opioid receptor (MOR), the SecR, and the metabotropic glutamate receptor 2 (mGluR2), respectively.¹¹ mGluR2 was found to be dimeric and MOR monomeric at all receptor densities explored, whereas SecR forms dimers only at a surface density high enough (>40 molecules/μm²) to establish relatively long-lived interactions (>100 ms). A fixed-cell colocalization study on the class A β₂-adrenergic receptor (β₂AR) found it to behave (almost) exclusively as a monomer,¹² whereas an earlier single-molecule study reported a high level of dimerization for β₂AR in live cells.¹³ Other studies aimed at estimating the size of GPCR oligomers in live cells identified a variety of species, including monomers, transient dimers, stable dimers, and stable tetramers.^14,15 However, the identity of receptors, varying or undefined expression levels, and the evaluation methods may be partly responsible for this lack of consensus. Additionally, the heterogeneous physical properties of the cellular plasma membrane (e.g., fluidity, lipid nanodomains) are likely a determining factor for the observed signaling heterogeneity (e.g., hotspots¹⁶). As such, a rigorous characterization of the transport properties of GPCRs in their native membrane environment is a much-needed foundation to understand the spatiotemporal determinants of their activation, signaling, and oligomerization.

The M₁ muscarinic acetylcholine receptor (M₁R) is primarily expressed in the cortex and the hippocampus regions of the central nervous system (CNS), has been shown to play an important role in memory and cognition, and is a therapeutic target for treatment of schizophrenia and Alzheimer’s disease.¹⁷ Previous fluorescence studies have shown that M₁R exists as a dynamic mixture of monomers and dimers at low expression levels compatible with SPT analysis.¹⁸ Subsequently, a spatial intensity distribution analysis (SpIDA) study showed that treatment with antagonists caused upregulation of the receptor and significantly increased the fraction of M₁R oligomers.¹⁹ A recent SpIDA study in mouse neuronal cell cultures also showed that M₁R exists as a mixture of monomers and higher order oligomers.²⁰ However, these latest measurements were conducted at high, although clearly physiologically relevant, receptor density levels (50–100 receptors/μm²) and the SpIDA technique produced relatively large error bars for the oligomer fractions.

Here, we describe single-molecule fluorescence measurements of membrane transport properties and quaternary organization of the M₁R in live cells under various conditions. The expression of the receptor was controlled in the Flp-In T-REx 293 cell system, corresponding to surface densities between ∼0.1 and ∼50 receptors/μm². Diffusion properties and oligomeric assembly were quantified using SPT and FCS at low and high receptor densities, respectively. M₁R exhibits different diffusion regimes, which are spatially heterogeneous across the surface of the plasma membrane and are impacted by orthosteric ligands and membrane modulators. In addition, M₁R is largely monomeric at low expression, and the oligomerization increases in an expression level-dependent manner. Finally, the fraction of receptor oligomers is dependent on the conformation/activation state, with binding of orthosteric M₁R ligands, especially the agonist, significantly promoting supramolecular complexes.

Methods

DNA Constructs

The constructs Halo-M₁R and Halo-CD86 were initially assembled in vector pCEMS1-CLIP10 m. An mGluR5 signal sequence and HA-tag was made by annealing two primers, such that the cohesive ends for EcoR1 and BamH1 were formed and this was inserted into these sites in the plasmid. The CLIP sequence was excised and replaced with HaloTag which was PCR amplified with primers which added Cla1 and Sbf1 sites and this was then inserted into these sites in the plasmid. CD86 and M₁R were PCR amplified with primers designed to add Asc1 and Not1 sites, and these fragments were subcloned into the plasmid. Finally, the whole insert was cut out with BamH1 and Not1 and subcloned into pcDNA5-FRT-TO. All steps were confirmed by sequencing.

Cell Culture

All cells used in this study were maintained in a humidified incubator with 5% CO₂ at 37 °C prior to measurements. Parental Flp-In T-REx 293 cells (Invitrogen, R78007) were kept in high glucose Dulbecco’s modified Eagle’s medium (DMEM, Sigma-Aldrich, D5796) supplemented with 10% (v/v) fetal bovine serum (Invitrogen, 12484028), 100 units/mL penicillin and 0.1 mg/mL streptomycin (Gibco, 15070063), 0.1 mM nonessential amino acids (Gibco, 11140050), 5 μg/mL blasticidin (MilliporeSigma, 203350), and 100 μg/mL zeocin (Thermo Scientific Chemicals, J67140XF). Transfected cells were maintained in the complete culture medium, which is the same medium as above with 5 μg/mL hygromycin B (Thermo Scientific Chemicals, J60681-MC) replacing the zeocin.

Parental Flp-In T-REx 293 cells were transfected in a 5 cm Petri dish (Sarstedt, 83.3901) with an 8 μg mixture of the pcDNA5/FRT/TO vector (harboring Halo-M₁R or Halo-CD86) and the pOG44 plasmid in a 1:9 ratio, along with 10 μL Lipofectamine 2000 (Invitrogen, 11668027) in 1 mL reduced serum medium Opti-MEM (Gibco, 31985070). After 48 h, the transfection medium was changed to the complete culture medium to initiate the selection of stably transfected cells. Pools of cells were established, allowing 10–14 days for hygromycin-B-resistant colonies to form, then split into 35 mm glass bottom μ-dishes (Ibidi, 81158), where they were grown to 50–70% confluency. They were then incubated with 1–100 ng/mL doxycycline (Sigma-Aldrich, PHR1145) for 8 h to obtain controlled expression levels that are suitable for single-molecule fluorescence experiments.²¹

Fluorescence Labeling In Situ

For SPT experiments, cells with low expression levels of Halo-M₁ or Halo-CD86 (treated with 1–10 ng/mL doxycycline) were treated post induction with the HaloTag dye JF635i-HTL (Janelia Farm). As such, cells were incubated with 1 nM JF635i-HTL in the complete culture medium for 5 min at 37 °C to achieve efficient in situ fluorescence labeling of the membrane protein of interest, i.e., M₁R or CD86. The cells were then washed three times with 2 mL of complete culture medium to eliminate unbound dyes and then incubated with 1 mL of FluoroBrite DMEM medium (Gibco, A1896701) for fluorescence imaging. For FCS experiments, cells were induced for 24–48 h at a higher level of expression (0.1–1 μg/mL doxycycline) and the same labeling protocol was followed. For dcFCS experiments, cells were incubated with a mixture of two spectrally different dyes, 1 nM JF549i-HTL (Janelia Farm) and 1 nM JF635i-HTL, in complete culture medium for 5 min at 37 °C.

M₁R Ligands and Membrane Disruptors

To assess how receptor activation and membrane architecture impact the diffusion and oligomerization of GPCRs, we incubated the cells with saturating amounts of M₁R ligands and plasma membrane disrupters, respectively. For ligand studies, postlabeling, cells were incubated with 10 μM of the antagonist pirenzepine (Thermo Scientific Chemicals, J62252-MC) for 90 min at 37 °C or with 10 μM of the agonist carbachol (Millipore Sigma, 21238) for 30 min at 37 °C. To modify the organization of the membrane, postlabeling, cells were incubated with either 50 μM C6 ceramide (Cayman Chemical, 0658066-12), or with epigallocatechin gallate (EGCG) (Cayman Chemical, 0531242-85) for 2 h at 37 °C. To disrupt the cytoskeleton, postlabeling, cells were incubated with 2.5 μg/mL cytochalasin D (MilliporeSigma, 250255) for 3 h at 37 °C.

Cell Fixation

Before plating cells, 35 mm glass bottom μ-dishes were coated with 5 μg/mL fibronectin (Bachem Americas Inc., 4030597.0001) in Hank’s balanced salt solution (HBSS) (Cytiva, SH30268.01) for 1 h at room temperature. Cells were plated in these dishes to 50–70% confluency before the expression of proteins of interest was induced by adding doxycycline, as described above. Cells were then chilled on ice for 10 min and the proteins of interest were labeled with HaloTag dyes as described above. Post labeling, cells were washed in ice-cold HBSS before fixation with 4% paraformaldehyde (Thermo Scientific Chemicals, 047392.9 L) and 0.2% glutaraldehyde (Thermo Scientific Chemicals, A17876.AE) in HBSS on ice for 1 h. Post fixation, cells were washed 3–4 times with ice-cold HBSS before changing to FluoroBrite DMEM for fluorescence imaging experiments.

Confocal Imaging

Screening for expression levels and fluorescence in situ labeling conditions was performed on an X-Light V2 spinning disc confocal microscope with an LDI-7 laser engine (Quorum Technologies). The microscope features multiple laser excitation wavelengths (405, 470, 532, 555, and 640 nm) as well as a bright-field illumination source (X-Cite 110 LED, Lumen Dynamics). The system is based on an inverted microscope body (Leica Di8) with a motorized stage (Applied Scientific Instrumentation, MS2000), five switchable objectives (10×–63× magnification), multiple emission filter sets (435–740 nm), and a twin scientific complementary metal–oxide–semiconductor (sCMOS)/electron-multiplying charged-coupled-device (EMCCD) camera detection system. For our experiments, the cells were illuminated with the 640 nm laser set at a power of 1 mW and the images were acquired using the sCMOS camera at a rate of 10 frames-per-second (fps) in the bright-field mode and 2 fps in the confocal mode.

TIRF Imaging

Single-molecule imaging of fluorescently labeled cells were performed on a custom-built TIRF microscope described in detail previously.⁷ Briefly, surface-immobilized samples were illuminated in the evanescent mode through a high numerical-aperture oil-immersion objective (Olympus PlanApo N, 60×/1.45) using red laser excitation at 638 nm modulated by an acoustic optical tunable filter (Gooch & Housego, MSD040-150-0.2ADS2-A5H-8 × 1). Fluorescence from the sample was collected through the same objective, filtered using dichroic (Semrock, FF650-Di01), long-pass (Semrock, BLP01-647R-25), and bandpass (Semrock, FF01-698/70-25) filters, and detected using an EMCCD camera (ANDOR, Ultra 897). For TIRF imaging, cell dishes were mounted on a custom-built sample holder with focus stabilization. For all experiments, the laser excitation intensity in the sample was set to 185 W/cm². The area of detection was 42 μm × 42 μm, with 2–3 cells typically in the field of view. Fluorescence movies of the cells were acquired using the EMCCD camera at a frame rate of 10 fps for live cells and 2 fps for fixated cells. The total duration of the TIRF movies was typically 3–5 min.

Single Particle Tracking Analysis

Individual fluorescent particles were detected and tracked in time and space using the TrackMate Linear Assignment Problem (LAP) algorithm²² implemented in the Fiji plugin.^23,24 Briefly, the position and intensity of each particle in each frame of the TIRF movie were calculated by the difference of Gaussian (DoG) detector.²⁵ In each frame, two Gaussian filters with different standard deviations were produced according to an estimated particle diameter. Results of these two filters were then subtracted, yielding a smoothed image with sharp local maxima at particle locations, from which their (x,y) coordinates and brightness intensity (I) were extracted. 2D tracking was performed using a simplified version of the LAP tracker, which only accounts for gap-closing events in a trajectory, which are caused by the fluorophore’s transient dark states, while splitting and merging of different trajectories were ignored. A tracking radius of 0.5 μm and a maximum lag/dark time of 100 ms were used as the gap-closing parameters. For each step in the trajectory, the algorithm assigns a cost to every possible event (e.g., blinking, appearing, and disappearing), and the solution that minimizes the sum of all costs is selected. Diffusing fluorescent particles were tracked until they photobleached or merged with other particles, yielding a mean trajectory length of around 2 s.

Spatiotemporal analysis of SPT trajectories extracted from the raw data was performed using a software based on variational Bayesian analysis of Hidden Markov models (vbSPT).²⁶ Discrete diffusion states characterized by diffusion constants, fraction occupancies, dwell times, and interstate transition rates are inferred from the global analysis of individual traces without any prior information. In the vbSPT software, the number of iterations and of bootstrapping samples was set to 25 and 100, respectively. Out of all vbSPT output parameters, we retained only the diffusion coefficients and the fraction occupancies for each state for further analysis. Since other physical scenarios (e.g., splitting/merging, photobleaching, and spatial density difference) were not considered here, the dwell times and transition rates are less robust.

The effect of ligands and membrane disrupters were analyzed in terms of changes in these diffusion parameters, with a nonrelated monomeric membrane protein CD86 as negative control and as a reference for random colocalizations. The particle intensity distributions in the initial frames of TIRF movies were extracted from the detected intensity traces and further analyzed to inform on the oligomeric states of the protein of interest.

smPB Analysis

smPB analysis was performed using a custom-written MATLAB GUI program based on the change-point algorithm.²⁷ Details regarding the image processing, the extraction of intensity traces and subsequent statistical analysis are given in the Supporting Information. Briefly, the program conducts morphological opening on the last few frames of the TIRF movie to estimate and correct for the uneven TIRF laser excitation field across the imaging area. Upon correction, the program identifies fluorescent particles in a sequence of frames that are brighter than the background by at least a 3-sigma threshold and removes spots that are too close to each other (<0.8 μm) or too close to the edge of imaging area (<0.5 μm).

To account for background due to nonspecific attachment of fluorophores to fixated cells, we applied further local corrections. For each diffraction-limited spot, which has a size of 5 px × 5 px (0.8 μm × 0.8 μm), the local background in each frame is estimated by taking the average intensity of the 4 dimmest pixels from the 24 pixels within a 7 px × 7 px area surrounding the initial spot (each pixel is 167 nm). The corrected intensity in each frame is then calculated by subtracting the local background per pixel in the diffraction-limited spot.

Downward change-points in each intensity-time trace were identified based on the principles laid out by Watkins and Yang,²⁸ the results from all traces in a set of samples acquired under the same conditions were assembled as histograms depicting the initial molecular brightness I₀ and the time-to-photobleaching T_pb. The distribution of molecular brightness I₀ was then compared and used as prior to evaluating the particle intensity distribution in the initial frames of single particle tracking movies from live cells.

FCS Analysis

FCS measurements on live cells were performed on a custom-built confocal microscope using a hardware correlator as previously described.⁶ Prior to acquiring intensity correlation data, a confocal scan was performed to find cells that exhibit a fluorescence signal of 5–100 kHz, which is a count rate optimal for FCS, at a 638 nm laser excitation intensity of 50–250 W/cm. For dual-color FCS (dcFCS) experiments, both 532 and 638 nm lasers were used, at similar excitation intensities.

In a 50 μm × 50 μm area, typically 3–4 cells satisfied the signal requirement for FCS data acquisition and analysis. Multiple regions on either the bottom or the top cell membrane were selected for FCS measurements. Cells contained fluorescently labeled proteins at expression levels of 1–50 molecules/μm². Under these conditions, a single correlation curve was acquired in 20 s, and a measurement consisted of multiple repeats (∼10 or more at a single spot), to increase the signal-to-noise and estimate the standard deviation of the correlation curve for fitting purposes.⁶

Correlation curves from single- and dual-color FCS were analyzed in MATLAB using a custom-written program based on Marquart–Levenberg algorithm.⁶ For single-color FCS, the intensity fluctuations on both the bottom and top membranes are described by an autocorrelation function with a 2D diffusion component and a photophysical dark state²⁹

1

The variable τ in eq 1 is the lag time, τ_D is the average residence time of diffusing molecules in the detection volume, α is a factor for the anomalous diffusion,³⁰ and ⟨N⟩ is the average number of molecules in the detection area. The parameters τ_ds and f_ds are the lifetime and population fraction of the photophysical dark state of the fluorophore, respectively. The diffusion coefficient D was calculated from the fitted estimate of τ_D as D = w²/4τ_D, where w is the lateral radius of the confocal detection volume. The correlation curves were fitted in the interval from 100 μs to 10 s, to focus on the (slow) diffusion of labeled proteins in the cell membrane and ignore the (fast) submillisecond photophysical dynamics of the label.

For dcFCS experiments, whereas the two autocorrelation curves detected in the green (g) and red (r) channels were each fitted by eq 1, the cross-correlation curve was fitted by eq 2

2

This follows the assumption that the photophysical dark state dynamics of different fluorophores (JF549i-HTL and JF635i-HTL in this case) do not correlate with each other. Here, N_g and N_r are the average numbers of green and red fluorescent molecules, respectively, in the common detection volume, while N_x is the average number of codiffusing species in the same volume. Details of the fitting were as previously described,⁶ and the prevalence of M₁R oligomers were represented by the average fraction of each fluorescent species that codiffuses with the other (fcd), calculated according to eq 3

3

Here, G_x(0), G_r(0), and G_g(0) are the amplitudes of the three curves, respectively, with OVCF_g and OVCF_r being the overlap volume correction factors for green and red channel, respectively (see the Supporting Information). Since the labeling of M₁R with green or red fluorophore was stochastic, the lowest of the two fcd values is listed as the codiffusion/oligomeric fraction.

Estimation of Receptor Surface Density

For TIRF experiments, fluorescent particles were detected by finding local maxima in the background filtered image (Figure S2D). The expression level (surface density) of the receptors was calculated by dividing the number of detected particles in each cell by the surface area of the bottom cell membrane. The surface area was estimated by defining the cell contour using an open-source software Outfi,³¹ and was typically 600–700 μm². At low expression levels, typically 30–150 receptors were detected in a single cell, resulting in a surface density of 0.05–0.25 mol/μm². At the higher expression levels in FCS experiments, the surface density of receptors was calculated by dividing the average number of molecules in the detection volume by the horizontal cross-section of the detection volume

4

where ⟨N⟩ is a data fitting parameter using eq 1 and w is the width of the FCS detection volume obtained by calibration experiments (Figure S4). FCS data were acquired in regions of receptor densities around 10–100 mol/μm².

Results and Discussion

Imaging M₁R in Live Cells: Expression Control and In Situ Labeling

To express and label GPCRs in the membrane of living cells, we fused the HaloTag, to the N-terminus of the M₁R sequence (FRT/TO/Halo-M₁R). This plasmid was then cotransfected with the Flippase recombinase expression vector (pOG44) into Flp-In T-REx 293 cells. Populations of cells resistant to hygromycin B³² were collected as these are anticipated to stably incorporate Halo-M₁R, and able to express Halo-M₁R at levels controlled by the concentration of doxycycline (or tetracycline) added to the growth medium (Figure 1A). For single-molecule imaging experiments, we varied the doxycycline concentration between 1 and 100 ng/mL to modulate the surface density of receptors at the plasma membrane between 0.05 and 0.5 mol/μm². M₁R was labeled in situ by a cell impermeable and fluorogenic dye, JF635i-HTL, which binds specifically and covalently to the exposed extracellular HaloTag attached to the N-terminus of the receptor (Figure 1E–G).

Figure 1.

Schematic view of the live cell system and the experimental setup used for imaging the M₁ receptor. (A) Halo-M₁R was stably transfected in the Flp-In T-REx 293 cell system and expression controlled by doxycycline concentration (1–100 ng/mL), and labeled in situ with JF635i-HTL. (B) Confocal imaging shows that the fluorescence is primarily localized at the external cell membrane, e.g., the cross-section profile (C), confirming successful and specific labeling of Halo-M₁R in live cells. Cells with fluorescently labeled Halo-M₁R were imaged on a custom-built TIRF microscope (D). Examples of cells expressing the receptor at different surface densities are shown: low (<0.05 molecules/μm², E), intermediate (0.05–0.25 molecules/μm², F), and high (>0.25 molecules/μm², G).

Confocal imaging of a section of a JF635i-HTL-labeled Halo-M₁R cell at ∼5 μm above the dish surface, showed that the fluorescence signal was predominately located at the cell membrane (Figure 1B). The bright regions appearing inside the cell’s interior are likely labeled receptors that have been internalized, in an agonist-independent manner, to endosomal compartments.³³ The cross-section intensity profile (Figure 1C) further confirmed the specific fluorescence labeling of the muscarinic M₁ receptor in situ, at the external membrane of live cells. As control, TIRF images of untransfected parental Flp-In T-REx 293 cells subjected to the same labeling procedure showed very low fluorescence signals, comparable to the background/autofluorescence level (see the Supporting Information).

To investigate the oligomerization and the diffusion in the plasma membrane of M₁R expressed at relatively low levels, cells were imaged on a custom-built TIRF microscope (Figure 1D). The (x, y) positions and the emission intensities (I) for all detected single spots/particles in each image of the TIRF movies, which are typically acquired at 10 fps, were stored and analyzed.

Examples of static TIRF images of cells with increasing levels of M₁R expression are shown in Figure 1E–G: low (<0.05 mol/μm²), intermediate (0.05–0.25 mol/μm²), and high (>0.25 mol/μm²), respectively. An example of data acquisition with optimal receptor density for SPT experiments was included in the Supporting Information (Movie S1). Note that the categorization of M₁R expression levels into low, intermediate and high was made here in the context of the TIRF experiments, as even the high density level is well below physiological M₁R expression levels in cortico-hippocampal neurons of the mouse CNS cells (10–100 mol/μm²).²⁰ All TIRF experiments were conducted at low and intermediate receptor densities in the membrane, as the current spatial resolution and pixel size on this setup hinders resolving and tracking single emitters in crowded areas (>0.5 mol/μm²).

SPT Analysis of the Membrane Transport of M₁R

SPT is a powerful method to delineate the transport properties of M₁R in the cell membrane. Compared to other fluorescence-based approaches measuring molecular diffusion, such as fluorescence recovery after photobleaching and FCS, SPT provides a superior spatial resolution and, more importantly, state-dependent heterogeneity rather than just ensemble- and time-averaged information.³⁴ Diffraction-limited fluorescent spots, attributed to labeled receptors, were detected and tracked across multiple frames in a TIRF movie using the TrackMate Linear Assignment Problem (LAP) algorithm²³ (Figure 2A). Options such as gap closing were used, while others, such as splitting and merging, were not implemented in the current analysis (see Methods); the average duration of a trajectory was 1.6 ± 0.1 s, with 0.1 s steps.

Figure 2.

Tracking the diffusion of M₁ receptors in the plasma membrane of live cells. (A) Single receptor particles were detected in each frame of a TIRF movie and their 2D trajectories were built using TrackMate.^23,24 The trajectories were color-coded according to their mean-square displacement, with increasing values from blue to red. (B) The trajectories were analyzed using vbSPT,²⁶ yielding distinct diffusion states characterized by diffusion coefficients, population fractions and transition rates between each state. Three diffusion states were found for M₁R: confined/immobile (D₁ ≈ 0.01 μm²/s), slow (D₂ ≈ 0.04 μm²/s), and fast (D₃ ≈ 0.14 μm²/s). SPT analysis in the presence of lipid membrane disruptors (C) and of muscarinic ligands (E), revealed changes in the diffusion pattern of M₁R due to the lipid environment and the activation state of the receptor, respectively. As a control, the same conditions were applied to cells expressing the Halo-CD86 protein (D,F). *, p < 0.05. **, p < 0.01. ***, p < 0.001. NS, not significantly different. See Table 1 for the full list of the parameters extracted from the SPT analysis.

All single-particle trajectories obtained under a certain condition were then analyzed using vbSPT,²⁶ a software based on variational Bayesian analysis of Hidden Markov models (HMM). Global analysis identified three diffusion states (D₁, D₂, and D₃) from raw tracking data of M₁R with no prior information, with characteristic parameters such as diffusion coefficients, fraction occupancies, average dwell times, and interstate transition rates (Figure 2B). In the absence of any ligands (the Apo state), most (∼3/4) of the receptors are in the D₃ state, ∼1/5 of them are in the D₂ state, and a minor fraction ∼5%, in the D₁ state (Table 1).

Table 1. Diffusion State Parameters Derived by SPT Analysis of M₁R and CD86 in Different Conditions^a.

	compounds	D1		D2		D3
	condition	fraction (%)	diffusion coefficient	fraction (%)	diffusion coefficient	fraction (%)	diffusion coefficient
M1R	Apo	5 ± 1	0.016 ± 0.003	22 ± 5	0.043 ± 0.002	73 ± 5	0.144 ± 0.004
	+C6 ceramide	4 ± 1	0.015 ± 0.001	23 ± 2	0.041 ± 0.003	73 ± 2	0.154 ± 0.002
	+EGCG	9 ± 2	0.014 ± 0.002	38 ± 4	0.036 ± 0.002	53 ± 5	0.136 ± 0.003
	+cytochalasin D	17 ± 3	0.009 ± 0.000	40 ± 1	0.031 ± 0.002	43 ± 3	0.128 ± 0.004
	+pirenzepine	10 ± 2	0.017 ± 0.001	36 ± 2	0.036 ± 0.002	54 ± 3	0.143 ± 0.002
	+carbachol	19 ± 4	0.008 ± 0.002	37 ± 5	0.029 ± 0.005	44 ± 8	0.117 ± 0.006
CD86	Apo	7 ± 2	0.015 ± 0.001	37 ± 2	0.037 ± 0.001	56 ± 4	0.150 ± 0.002
	+C6 ceramide	9 ± 3	0.011 ± 0.004	24 ± 5	0.037 ± 0.002	67 ± 6	0.152 ± 0.001
	+EGCG	18 ± 2	0.008 ± 0.002	46 ± 4	0.029 ± 0.004	36 ± 6	0.130 ± 0.007
	+cytochalasin D	13 ± 5	0.012 ± 0.001	39 ± 5	0.035 ± 0.002	48 ± 9	0.147 ± 0.008
	+pirenzepine	9 ± 2	0.014 ± 0.000	33 ± 2	0.040 ± 0.001	58 ± 3	0.153 ± 0.002
	+carbachol	8 ± 2	0.015 ± 0.002	31 ± 3	0.039 ± 0.004	61 ± 5	0.154 ± 0.003

^aUncertainties of the fractions and diffusion coefficients are the weighted standard deviation across results from all cells measured under each condition. Unit of the diffusion coefficient: μm²/s.

According to the diffusion coefficient for each state, D₃ (∼0.14 μm²/s) was classified as the fast diffusion, D₂ (∼0.04 μm²/s) as the slow diffusion, and D₁ (∼0.01 μm²/s) as the confined/immobile population. Note that for D₁, the mean displacement between subsequent frames is (τ = 0.1 s), similar to the precision of localization (σ) in SPT experiments³⁵ and well below the pixel size and the resolution limit. Furthermore, for an average trajectory length (τ = 1.6 s), the displacement is on the order of 200–250 nm, suggesting that the D₁ motion is spatially confined. Previous SPT studies on GPCRs also reported some fraction of receptors as being confined or immobile.^16,36 Confinement radii determined by SPT are on the order of ∼100 nm, in agreement with the size of lipid raft domains in the plasma membrane.³⁷ Lipid nanodomains are known to be important for signaling processes,³⁸ thus justifying the use of small molecule raft modulators to dissect the impact of membrane organization on the transport properties of M₁R.

Fluorescently labeled Halo-M₁R cells were treated with C6 ceramide and EGCG, potent raft modulators that were validated recently using a giant plasma membrane vesicle (GPMV) assay³⁹ (Figure 2C). That study confirmed that these compounds decrease (C6 ceramide) or increase (EGCG) the fraction of liquid-ordered (raft-like) domains and the extent of phase separation of the plasma membrane. We also treated the cells with cytochalasin D (CD),⁴⁰ a compound that disrupts cytoskeletal filaments. For comparison, we applied the same conditions to cells expressing Halo-CD86 and performed SPT experiments and analysis of them (Figure 2D).

In the absence of membrane disruptors, CD86 showed a similar fraction of immobile/confined diffusion (7 ± 2%) as M₁R and a higher fraction of slow diffusion (37 ± 2%) (Table 1). This indicates that M₁R does not partition in raft-like domains but may prefer more fluid-like regions of the plasma membrane. To confirm this hypothesis, cells with fluorescent M₁R or CD86 were treated with 50 μM C6 ceramide for 90 min to disrupt the lipid rafts on the cell membrane. Indeed, SPT results showed that the diffusion regimes of M₁R under these conditions did not experience significant changes, while the fast diffusion population of CD86 increased significantly (Figure 2C,D).

Upon treatment with 50 μM EGCG for 90 min, or with 2.5 μg/mL CD for 3 h, M₁R, and to a smaller extent CD86, exhibited significant increases in the fractions of immobile/confined and slow diffusion, accompanied by an overall decrease in the diffusion coefficients (Table 1). While EGCG was expected to increase the membrane phase separation and slow down diffusion of transmembrane proteins, CD unexpectedly produced a similar output, despite previous studies showing an opposite trend.⁴¹ One possible explanation is that as CD disrupts the actin filament network and its interaction with the plasma membrane, nanoscopic raft-like domains (50–200 nm) are prone to aggregate into larger domains (>300 nm), similar to those observed using the GPMV system,³⁹ and thus preserve a significant population of slow diffusion. Raft and nonraft phase fluorescent lipids, NBD-DSPE, and DiD, respectively, could be used to validate this hypothesis.⁴²

Next, we assessed how the diffusion pattern of the receptor is affected by its activation status. As such, Halo-M₁R and Halo-CD86 cells were incubated with the antagonist pirenzepine (10 μM, 90 min) or with the agonist carbachol (10 μM, 30 min). Binding of either ligand to M₁R caused larger occupancies of both the D₁ and D₂ diffusion states (Figure 2E). In addition, a significant overall decrease in diffusion coefficients was observed (Table 1) upon activation by carbachol. Notably, the fast diffusion coefficient in the presence of antagonist agrees well with a previous SPT study of M₁R diffusion in CHO cells using fluorescent ligands (D = 0.089 ± 0.019).¹⁸ As expected, the diffusion pattern of CD86 did not exhibit significant changes in the presence of muscarinic ligands (Figure 2F), since it is functionally unrelated to M₁R and unlikely to bind these ligands. This also indicates that, at the concentrations used, the muscarinic ligands do not produce a nonreceptor-mediated effect on membrane characteristics and fluidity.

Photobleaching Analysis of M₁R Oligomers at Low/Moderate Expression Levels

To characterize the oligomerization state of M₁R, we analyzed the emission intensity of both static (in fixed cells) and mobile (in live cells) particles in the TIRF data by combining smPB step counting and single particle tracking. Robust outcomes of this analysis critically depend on prior information on the fluorescent probe under the same conditions, such as the molecular brightness of the monomer (I_m) and the average time-to-photobleaching, (T_pb).¹⁴ These parameters were obtained using cells expressing the monomeric Halo-CD86 protein, which were labeled with the same fluorophore as the receptor and imaged under identical experimental conditions. The results were then used to determine the oligomerization states of M₁R at varying expression levels in the subphysiological range of 0.05–0.25 mol/μm².

For fixation, live cells expressing either Halo-M₁R or Halo-CD86 and labeled with 1 nM JF635i-HTL were treated with 4% para-formaldehyde and 0.2% glutaraldehyde for 60 min, before changing the buffer to Fluorobrite DMEM and imaging under the same condition as SPT in live cells. To improve the signal quality, 100 s long TIRF movies of fixed cells were recorded at a slower frame rate than live cells, i.e., 2 vs 10 fps, respectively (Figure 3A). Typically, in the last frame of the sequence >90% of the fluorescent particles from the first frame were photobleached; the unbleached particles were excluded from further analysis. Intensity traces exhibiting one- and two-step photobleaching transitions, as shown in Figure 3B, originate from M₁R monomers and dimers, respectively.

Figure 3.

smPB analysis on fixed cells. (A) TIRF image of a cell with low expression (∼0.05 molecules/μm²) of Halo-M₁R, labeled with JF635i-HTL prior to fixation to the glass coverslip surface. (B) Examples of intensity-time traces of individual spots showing one- (red) and two-step transitions to the background level, associated with monomeric and dimeric receptor particles, respectively. (C) 2D histogram of the initial intensity (I₀) against time-to-photobleaching (T_PB); the I₀ distribution was fitted to a Gaussian centered at 4.4 ± 0.2 kHz, and the T_pb distribution was fitted to an exponential with a time constant of 14.7 ± 0.5 s. (D) The I₀ – T_PB histogram for the monomeric CD86 protein in fixed cells, with an average initial intensity of 4.3 ± 0.2 kHz and an average photobleaching time of 18.0 ± 1.3 s.

Analysis of smPB data was performed using custom-written MATLAB software, GLIMPSE (see the Supporting Information). After local background and illumination corrections were applied, the initial brightness intensity (I₀) was calculated for each detected fluorescent particle. A 2D histogram of I₀ against T_pb is shown in Figure 3C, with the brightness distribution fitted to a Gaussian centered at 4.4 ± 0.2 kHz and the time-to-photobleaching distribution fitted to an exponential with a lifetime of 14.7 ± 0.5 s. The same data for CD86 have an average I₀ of 4.3 ± 0.2 kHz and T_pb of 18.0 ± 1.3 s (Figure 3D).

That the I₀ distribution for CD86 was well fitted by a single Gaussian was indeed expected, as CD86 is a monomeric protein.⁴³ As such, the average brightness for CD86 is an accurate measure of the molecular brightness (I_m) of a single JF635i-HTL fluorophore bound to a HaloTag. For the M₁R distribution (Figure 3C), the average I₀ is close to I_m, with only a small fraction (<10%) of receptor particles having intensities around 2I_m. As such, at low expression levels, M₁R is largely monomeric.

Notably, the average value of T_pb is about an order of magnitude larger than the average duration of SPT trajectories in live cells (1.6 ± 0.1 s). This suggests that the limiting factor of the length of diffusion trajectories in cells is not the photobleaching of the fluorophore but splitting/merging events or out-of-focus movement (e.g., receptor internalization). A more thorough SPT analysis including trajectory segmentation and diffusivity transitions⁴⁴ could provide further insights into these unaccounted effects in the present study.

To validate the results obtained from fixed cells, the oligomerization status of M₁R was also estimated from the SPT data in the live cells. In this case, only the initial frames, typically 20 frames (2 s), in TIRF movies were used to extract the intensity distribution of detected fluorescent particles using TrackMate.^23,24 The intensity distribution of diffusing M₁R particles at low expression levels in live cells (<0.05 mol/μm²) was overlaid with the monomer molecular brightness distribution obtained from CD86 in fixed cells (Figure 4A). The two distributions were very similar, confirming that the diffusing M₁R particles being tracked in live cells are largely monomeric.

Figure 4.

M₁R oligomerization in live cells at low and intermediate expression levels. (A) The intensity distribution of diffusing M₁R particles extracted from the initial frames (<2 s) of TIRF movies of live cells with low expression levels of receptor. For comparison, the initial intensity distribution of CD86 from fixed cells is also shown (dashed red curve). (B) Gaussian fit of the initial SPT intensity distributions for M₁R at low expression (<0.05 mol/μm², red) and intermediate expression (0.05–0.25 mol/μm², yellow). Two components were needed in each case, centered at 4.3 and 7.5 kHz for low expression, and at 4.6 and 8.0 kHz, for intermediate expression. We assigned them to monomeric and oligomeric species, respectively. (C) Monomer and oligomer fractions of M₁R upon treatment with the muscarinic antagonist (pirenzepine, red) and agonist (carbachol, yellow), calculated by integrating the area under each Gaussian. The oligomer fraction of M₁R increased significantly in the presence of antagonist and agonist (49 ± 6 and 47 ± 6%, respectively) compared to the Apo state (27 ± 4%).

When increasing the M₁R expression to the highest levels suitable for SPT analysis (∼0.25 mol/μm²), however, we have noticed significant differences between such distributions (Figure S3A,B), suggesting that a single Gaussian component is not sufficient to fit the M₁R intensity distribution. Indeed, at these intermediate expression levels of the receptor, a two-Gaussian fit with average intensities of 4.6 and 8.0 kHz is needed, which can be assigned to monomeric and dimeric species, respectively (Figure 4B). Using the area under each Gaussian component, at low expression M₁R appears to be ∼86% monomeric, while at intermediate levels, there are ∼27% M₁R dimers, indicating that receptor oligomerization occurs in an expression-dependent manner.

Since ligand induced activation of the receptor alters its diffusion pattern (see above), it may also impact its oligomerization state. As such, the intensity distributions of Halo-M₁R cells treated with 10 μM of antagonist (pirenzepine) or agonist (carbachol) were fitted to a sum of Gaussians (Figure S3C,D). Compared to the Apo state at similar (intermediate) expression levels, binding of either pirenzepine or carbachol produced significantly larger oligomeric (dimeric) fractions, i.e., 49 and 47%, respectively (Figure 4C). Note that these values were obtained solely from intensity information, and the current approach does not exclude the nonspecific/transient colocalization events of diffusing receptors. As such, the above M₁R oligomer fractions should be viewed as upper limits under these conditions.

It is worth mentioning that even though both antagonist and agonist seem to have similar effects on the M₁R supramolecular organization, they may function in different ways. Spinning-disc confocal scanning of pirenzepine-treated Halo-M₁R cells showed a quasi-uniform fluorescence distribution at the cell membrane, while for carbachol-treated cells the distribution was very heterogeneous (see Supporting Information, Movies S2 and S3). Clusters of diffusing bright spots (∼500 nm in diameter) could be identified both on the membrane and inside the cell, suggesting significant relocation or internalization of M₁R, similar to previous studies.^45,46 Fluorescence studies showed significant internalization of M₁R happening on a time scale comparable to the duration of agonist treatment,15–30 min.⁴⁵ However, endocytic vesicles containing internalized receptors at the inner cell membrane can be distinguished from the oligomers of receptors in the membrane based on inherent differences in their lateral size and brightness. Granulated regions with high intensity were also avoided in our further FCS experiments on cells treated with agonist.

FCS Analysis of Diffusion and Oligomerization of M₁R at High Expression Levels

In order to dissect the expression/density-dependent oligomerization of M₁R in more detail, we sought to perform experiments in conditions similar to physiological expression levels (15–30 mol/μm²). Due to its optical resolution limit, TIRF-based single-molecule fluorescence experiments cannot identify and track individual molecules under such crowded conditions.⁴⁷ Although super-resolution methods like stochastic optical reconstruction microscopy⁴⁸ and single particle tracking photoactivated localization microscopy (sptPALM)⁴⁹ are available, they either lack the temporal resolution for dynamics studies or suffer from artifacts due to labeling/photoactivation efficiency. Therefore, we turned to the spectroscopic methods.

dcFCS is a powerful method to study coupling interactions between biological molecules in live cells,^6,50 and a useful complement for SPT and FRET⁵¹ experiments, with higher temporal resolution and lower false positives. We performed one- and two-color FCS on cells at physiological expression level, by inducing with higher concentrations of doxycycline (0.1–1 μg/mL, 24–48 h). Calibration experiments for FCS detection volume in the channel for each color (green and red) and overlapping volume correction factors (OVCFs) between two detection volumes were carried out using standard dyes (Rhodamine 6G and Atto655-maleimide) and fluorescent microspheres (see Supporting Information, Figure S4). FCS experiments were first conducted on the bottom membrane of Halo-M₁R and Halo-CD86 cells labeled with 1 nM JF635i-HTL. The autocorrelation (AC) curves were best fitted by a 2D anomalous diffusion model with one photophysics term (τ_ds) using eq 1 (Figure 5A,B).

Figure 5.

FCS analysis of Hao-M₁R in live cells at higher expression levels. (A) Experimental data (blue circle) and fitted curve to eq 1 (red line) for JF635i-M₁R at the bottom cell membrane (adhering to the glass surface), at an expression level of ∼1 mol/μm². (B) Experimental data (blue rhombus) and fitted curve to eq 1 (red line) from JF635i-M₁R at the bottom cell membrane, at an expression level of ∼50 mol/μm². Data (orange circle) and fit (purple line) are also shown for measurements at the top cell membrane. (C,D) dcFCS from cells expressing Halo-M₁R at levels of ∼10 mol/μm² which were labeled with JF635i-HTL (red) and JF549iHTL (green) simultaneously. The autocorrelation (AC) data were fitted to eq 1 and the cross-correlation (CC) data (black cross) were fitted to eq 2. The amplitude of the CC curve indicates a small fraction (∼20%) of codiffusing (oligomeric) species in the Apo state (C), while this increases significantly (∼50%) in the presence of the agonist carbachol (D). See Table 2 for the full list of the parameters extracted from the FCS analysis.

The results show that at subphysiological expression levels (∼4 mol/μm²), the diffusion coefficient of M₁R was 0.089 ± 0.012 μm²/s, which remains unchanged at higher densities (∼50 mol/μm²) and for the CD86 control (Table 2). This diffusion coefficient should be viewed as an average of D₂ and D₃ populations from SPT experiments as FCS provides ensemble- and time-averaged information and is not sensitive to confined diffusion in nanodomains or to immobile receptors. However, an advantage for the FCS method is that its confocal arrangement can probe molecular transport at various positions in the cell, as surface interactions at the bottom membrane may affect SPT results using TIRF imaging. We also performed FCS at the top membrane of Halo-M₁R cells, with autocorrelation functions generally showing faster decay (i.e., faster diffusion) compared to that at the bottom membrane (Figure 5B and Table 2).

Table 2. FCS Fitting Parameters of M₁R Diffusion in the Cell Membrane under Various Conditions^a.

	condition	location of measurement	density (mol/μm2)	D (μm2/s)	fraction of codiffusion (%)
M1	Apo	bottom	3.97 ± 0.32	0.089 ± 0.012	6.3 ± 0.3
	Apo	bottom	48.6 ± 4.2	0.083 ± 0.014	14.2 ± 0.6
	Apo	top	38.4 ± 4.3	0.095 ± 0.018	20.4 ± 1.6
	+pirenzepine	bottom	61.7 ± 5.2	0.074 ± 0.007	29 ± 6
	+carbachol	bottom	11.2 ± 0.4	0.066 ± 0.007	50 ± 7
CD86	Apo	bottom	49.8 ± 5.3	0.087 ± 0.008	8.6 ± 0.5

^aParameters were obtained by fitting the data to eqs 1 and 2; uncertainties were calculated across all cells measured for each experimental condition.

To investigate the oligomer fractions at high expression levels, dcFCS was performed on Halo-M₁R cells labeled with 1 nM JF635i-HTL and JF549i-HTL simultaneously (Figure 5C). The AC curves were fitted to eq 1 and crosscorrelation (CC) curves to eq 2, with the fraction of codiffusion (fcd) given by eq 3 representing the fraction of receptor oligomers (see Materials and Methods). The fcd values estimated for the Apo state show a minimal fraction (<10%) of M₁R oligomers, with a slight increase with the expression level (Table 2). Upon treatment with ligands, the CC amplitude exhibited a significant increase, e.g., up to (∼50%) for the agonist carbachol (Figure 5D). Similar to SPT outcomes, FCS results show that binding of ligands to M₁R slows its diffusion in the plasma membrane (Table 2).

While FCS measurements of diffusion agree with SPT results, the estimated oligomer fractions are much lower. One possible explanation is the stochastic and competitive labeling, i.e., both fluorophores bind to the same HaloTag on the receptor, and as their binding kinetics may be different, the local densities of receptors labeled with the two fluorophores may also be different. This is consistent with different AC amplitudes observed in the two channels, green (JF549i) and red (JF635i). As such, we report the lowest value of the two fcd’s as the oligomer fraction, to be seen as the lower limit (compared to the upper limit from SPT experiments).

Notably, the red fluorophore (JF635i) shows significant photophysical activity on the millisecond scale (τ_ds = 2.7 ± 0.2 ms, f_ds = 0.37 ± 0.01), assigned to the fastest decay in the AC curve, while this is less prominent in the AC data of the green fluorophore (JF549i) (f_ds = 0.13 ± 0.01) (Figure 5C,D). We attribute this decay to a long-lived dark state of rhodamine dyes, rather than a second diffusion component, which would be unreasonably (>10 times) faster compared to reported values for membrane proteins.

Conclusions

Growing evidence has shown that class A CPCRs can form functional dimers and higher order oligomers,^{11−13,19,20} while the molecular mechanisms underlying their oligomerization and diffusion kinetics are not fully understood. This study focused on describing the spatial movement and the supramolecular organization of the muscarinic M₁ receptor at varying expression levels at the plasma membrane of live cells as well as delineating the impact of receptor activation and of membrane fluidity.

By tracking the trajectories of many individual particles, it was found that M₁R has three Brownian diffusion states (confined/immobile, slow, and fast). The fast diffusion regime is predominant (∼70%), unlike the control protein CD86 which has more than 50% in the confined/immobile and slow states, indicating that M₁R does not partition in raft-like domains. This was further supported by experiments in the presence of the lipid raft disruptor C6 ceramide, which led to significant changes for CD86 (increased fast diffusion to ∼70%), while the M₁R diffusion regimes were left unchanged. With the addition of lipid raft enhancer EGCG both M₁R and CD86 showed significantly increased confined/immobile and slow diffusion populations, with overall lower diffusion coefficients. A similar effect was observed in the presence of the cytoskeleton filament disruptor CD, which has been shown previously to lead to faster, less confined diffusion.⁴¹ The opposite trend seen here might be caused by aggregation of nanoscale lipid raft domains (50–200 nm) into larger domains (>300 nm) that retain the ordered lipid phase and the higher viscosity.

The diffusion of M₁R has been shown to be affected by its activation state, as previous studies showed GPCRs oligomerization states varied upon binding to ligands.^6,52 Both the antagonist (pirenzepine) and the agonist (carbachol) led to higher fractions of confined/immobile and slow diffusion of M₁R, while not significantly affecting the diffusion of CD86, as expected. For the antagonist, we suspect that slower diffusion lowers the probability of interactions between M₁R and its cognate G proteins (G_q/11), thus inhibiting signaling by means of a spatiotemporal barrier. On the other hand, agonist binding leads to a conformation change that favors coupling of the receptor to the G protein, thus leading to slower diffusion of the complex. Further spatial analysis of diffusion maps and how orthosteric ligands alter them, as well as future dual-color SPT studies tracking both the receptor and the G protein will clarify these proposed scenarios and help elucidate spatiotemporal aspects of the initial steps in cell signaling that received little attention so far. Overall, our results indicate that M₁R prefers nonraft domains, the membrane diffusion of M₁R is heterogeneous, and is prone to be slowed down by lipid raft rich cellular environments as well as by binding to muscarinic ligands.

At the relatively low expression levels required for single-molecule experiments (<0.25 mol/μm²), M₁R exists primarily as a monomer (>75%); however, the fraction of homodimers increased as the expression levels increased, even within this limited range. The formation of supramolecular complexes may be dependent on conformation/activation state of the receptor; indeed, we found that both pirenzepine and carbachol promoted oligomerization of M₁R, which is consistent with previous findings.¹⁸⁻²⁰

At higher quasiphysiological expression levels, we probed the diffusion and the oligomerization states of M₁R using FCS. The measured diffusion coefficient of 0.08–0.09 μm²/s represents an average of the slow and fast diffusions in SPT experiments. The diffusion at the top cell membrane appeared slightly faster than at the bottom membrane, suggesting a minor surface interaction effect in the latter case.

Dual-color experiments revealed the extent of the M₁R oligomers. In contrast to the trend observed using single-molecule methods, a smaller fraction of M₁R oligomers in the Apo state was inferred from the cross-correlation amplitude of dcFCS experiments, although it increased with the expression level. Binding of orthosteric ligands, antagonist (pirenzepine), and agonist (carbachol), led to higher fractions of M₁R oligomers, with the effect of carbachol being more significant (∼50%) even at lower expression levels. Because of the stochastic labeling, differences in binding kinetics of the two fluorescent probes may lead to different local densities, which was reflected in the different autocorrelation function amplitudes. As such, the oligomer fractions estimated by dcFCS should be seen as lower limits, as opposed to upper limits from SPT experiments. dcFCS experiments conducted with orthogonally labeled receptors (e.g., HaloTag and SNAP-Tag) will lead to more precise estimations of oligomer fractions.

This study makes use of controlled, stable expression of the M₁ receptor the Flp-In T-REx 293 cell system and applies several single-molecule fluorescence techniques to quantify the dynamic heterogeneous molecular transport and oligomerization of the receptor M₁R in live cells. The results obtained indicate that the motility patterns and macromolecular assembly of M₁R vary considerably depending on its activation state and on the membrane nanoenvironment. A better understanding of the role of oligomers in GPCR-meditated signaling has significant implications for dissecting the underlying molecular mechanisms and its malfunction in related diseases.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.jpcb.4c01035.

• Receptor labeling specificity, GLIMPSE program for photobleaching analysis, intensity distributions, and FCS controls and calibrations (PDF)

• Movie S1 (AVI)

• Movie S2 (AVI)

• Movie S3 (AVI)

Author Present Address

^∥ Centre for Cold Matter, Department of Physics, Imperial College London, London, England, SW7 2BX, U.K

The authors declare no competing financial interest.

Special Issue

Published as part of The Journal of Physical Chemistry B virtual special issue “Advances in Cellular Biophysics”.