Small Angle Neutron Scattering Study of Rhodopsin Oligomerization and G-Protein Coupling in a Physiologically Relevant Lipid Membrane

Abstract

Although the oligomeric states of G-protein-coupled receptors (GPCRs) and interactions with cognate G proteins are central to their signal transduction capabilities, they remain poorly defined. In this study, we used small-angle neutron scattering (SANS) and a neutron contrast matching approach to elucidate the oligomeric states of the archetypal GPCR, rhodopsin, and its interaction with the G protein transducin (G_t). At a rhodopsin/lipid molar ratio of 1/360, we found that dark-adapted rhodopsin exists as a monomer, a finding consistent with its high functional activity measured upon photoactivation by spectrophotometry and the rate of catalyzed [³⁵S]-GTP-γ-S exchange. Following light activation, we observed that rhodopsin forms a stable 1:1 stoichiometric complex with G_t, the structure of which is consistent with recent cryo-EM data. In contrast, activated rhodopsin in the absence of G_t showed a propensity to form higher order oligomers. This research underscores the concentration-dependent nature of rhodopsin oligomerization and establishes SANS and the ability to produce appropriately contrast-matched samples, as a robust strategy for characterizing integral membrane protein interactions under biologically relevant conditions.

Graphical Abstract

1. Introduction

G protein-coupled receptors (GPCRs) are cell surface receptors characterized by a seven alpha-helical transmembrane structure that interact with members of the heterotrimeric G protein family upon ligand binding. GPCRs are essential for cellular communication, transducing external signals from hormones, neurotransmitters, ions, photons, odorants, and other stimuli into intracellular signaling responses. Recent advances, including the availability of over 450 unique structures from X-ray crystallography and cryo-Electron Microscopy studies (cryo-EM) [1], have revealed GPCRs as being highly dynamic, allosteric proteins with multiple and complex conformational states. These distinct three-dimensional structures, which are influenced by ligands and effectors, are responsible for receptor activation, signaling pathways and ultimately, important physiological responses [2, 3]. Oligomerization has emerged as a crucial mechanism regulating GPCR biosynthesis, ligand binding, G protein coupling, and downstream signaling [1] and reference therein. Despite their importance, determining GPCR oligomeric states within a functional membrane environment remains a significant challenge.

Here, we used small angle neutron scattering (SANS) to measure GPCR oligomerization and to follow the activation of its associated G protein. We chose rhodopsin – the GPCR responsible for low-light photoreception in vertebrates – as a model GPCR, because it is by far the best understood GPCR both structurally and mechanistically. Rhodopsin consists of the apoprotein opsin and its light absorbing 11-cis retinal chromophore that is covalently bound by a protonated Schiff base to Lys 296, which is located in the transmembrane helix VII. In the dark, 11-cis retinal behaves as an inverse agonist that keeps rhodopsin in its least active state, ensuring negligible basal activity of the receptor. Photon absorption triggers the in situ isomerization of 11-cis retinal into its all-trans form, and following a cascade of intermediate conformations, the formation of a G protein dependent equilibrium between meta-rhodopsin I (M_I) and several distinct meta-rhodopsin II states (M_II). M_I formation, which takes place within microseconds, involves localized changes near the retinal binding pocket, maintaining a structure similar to dark-adapted rhodopsin; importantly, M_I cannot bind G protein. In contrast, M_II formation takes place on a timescale of milliseconds and is characterized by larger conformational changes taking place outside the protein’s photochemical core. These structural changes alter the cytoplasmic receptor surface to allow two orders of magnitude higher rates of catalytic nucleotide exchange after binding of the G protein transducin (G_t), triggering a biochemical amplification cascade, which in vivo, ultimately results in the signaling of second order neurons and a visual signal. In rod outer segments, the reaction is terminated by rhodopsin phosphorylation by rhodopsin kinases which allow arrestin binding and prevent further activation of G_t e.g. Gulati et al. [4] and references therein.

Since protein oligomerization is thought to be concentration dependent, rhodopsin was reconstituted into phosphatidylcholine (PC) bilayers over a wide range of concentrations, from physiologically relevant to very dilute. The M_II/M_I ratio was measured by spectrophotometry to establish a concentration threshold at which the M_II/M_I equilibrium was independent of rhodopsin concentration, suggesting a homogenous protein structure. SANS experiments were conducted on dark adapted rhodopsin as well as opsin in a lipid bilayer. Interaction between photoactivated rhodopsin and G_t was quantitatively measured by the rate of [³⁵S]-GTP-γ-S binding and SANS experimentation. Our findings indicate that dark-adapted rhodopsin, at a rhodopsin/lipid molar ratio of 1/360, exists as a monomer and has high functional activity. When exposed to light, rhodopsin forms a stable 1:1 complex with G_t, in agreement with recent cryo-EM structural data. In contrast, activated rhodopsin in the absence of G_t formed higher-order oligomers. Our study highlights the concentration-dependent nature of rhodopsin oligomerization and establishes SANS and contrast matching methodology as a powerful experimental approach for studying the interactions between integral and peripheral membrane proteins under biologically relevant conditions.

2. Material and methods

Sample Preparation. 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC-d₈₂) (Fig. SI1A) and 1-stearoyl-2-oleolyl-sn-glycero-3-phosphocholine (SOPC) were purchased from Avanti Polar Lipids Inc. (Alabaster, AL) and used as supplied. Rhodopsin was purified from bovine retinas using procedures adapted from the method of Khun by the Litman laboratory [5]. Rhodopsin fractions purified from bovine retinas were solubilized in 3 wt.% octylglucoside (OG), resulting in a UV-vis absorption ratio at 280 nm/500 nm of 1.8 or lower were used. G_t was extracted from a highly purified fraction of bovine retinae rod outer segments (ROS). G_t extracted from ROS was further purified using a 50 mL DEAE Sephacel (Sigma-Aldrich) column and eluted with a solution of 20 mM Tris, 0.5 mg MgCl₂, 30 μM GTP, 1 mM DTT, 500 mM NaCl, pH 7.5 with added protease inhibitor. The DEAE elution solution was exchanged to 10 mM MOPS, pH 7.5 using an Amicon concentrator (MW cutoff 30 kDa) and the protein concentrated to up to 12 mg/mL. For long-term storage, G_t was lyophilized and redissolved in H₂O to its initial concentration, as needed.

For the reconstitution of rhodopsin into membranes and for M_I-M_II equilibrium measurements, the rhodopsin solution in OG was mixed with a micellar lipid/OG solution to adjust the protein/lipid ratio. After equilibration, the protein/lipid/OG micellar solution was added dropwise at a rate of 400 μL/min to deoxygenated Piperazine-N,N′-bis(2-ethanesulfonic acid (PIPES) buffer (10 mM PIPES, 100 mM NaCl, 50 μM diethylenetriaminepentaacetic acid (DTPA), pH 7.0) under rapid stirring, resulting in the formation of proteoliposomes. Protein content in the final sample was determined spectrophotometrically by measuring the absorbance of dark-adapted rhodopsin at 500 nm before and after photobleaching, using an extinction coefficient of 40,600 M⁻¹·cm⁻¹ at its absorbance maximum to calculate concentration. Lipid concentration was measured by NMR.

To determine the M_II/M_I ratio after photoactivation, four absorption spectra were collected sequentially using an Agilent 8453 diode array spectrophotometer. The acquired spectra were as follows: (i) after the sample was equilibrated in the dark at 37°C; (ii) 3 s after 15–20% of the sample was bleached by a 520-nm flash; (iii) 10 min after addition of 30 mM hydroxylamine to convert bleached rhodopsin to opsin and retinal oxime; and (iv) after complete bleaching of the sample. Individual M_I and M_II spectra were deconvolved from spectra of their equilibrium mixture using extinction coefficients at their absorbance maxima of 44,000 and 38,000 cm⁻¹, for M_I and M_II, respectively.

G_t activation assays were based on a method outlined in [6]. The catalyzed dissociation of G_t into unoccupied G_α and G_βγ by photoactivated rhodopsin was measured using the [³⁵S]-GTP-γ-S binding. Briefly, the binding of [³⁵S]-GTP-γ-S to G_α was carried out in a solution containing 0.25 μM of G_t⦁GDP in 20 mM MOPS (pH 7.5), 3 mM MgSO4, 1mM Ethylenediaminetetraacetic acid (EDTA), 1 mM DTT, 3 mg/mL Bovine Serum Albumine (BSA), 100 mM NaCl, 0.1 % Na-cholate, and [³⁵S]-GTP-γ-S at a radiospecific activity 5000 cpm/pmol. For experiments measuring the time dependence of rhodopsin catalyzed GTP-γ-S binding to G_t, sonicated disks (or reconstituted vesicles) containing various concentrations of rhodopsin were added and kept in the dark for 30 minutes at 22° C prior to photoactivation for equilibration (t_1/2 3 min). Assays were initiated by a flash of light (520 nm, 1cm) and then placed in the dark. Under these conditions, it was determined that the concentration of activated protein was independent of membrane concentration. Assays were terminated at different times (between 5 and 1800 s) by diluting 10 μl aliquots of the reaction directly into 2 ml of an ice-cold solution containing 10 mM Tris-HC1 (pH 8), 100 mM NaC1, and 25 mM MgC1₂ (solution D). Samples were collected and held on ice prior to rapid filtration using 0.45-pm nitrocellulose membranes (25 mm, Millipore type HA) under a vacuum of 600 mm Hg. Unbound [³⁵S]-GTP-γ-S was removed by sequential washing of the membranes seven times with 2-ml samples of ice-cold solution D. Membranes were then dried, and the radioactivity associated with membranes was measured after addition of 10 ml of a toluene and 0.5 % 2,5-diphenyloxazole mixture. Nonspecific binding was determined by the absence of G_tα in the reaction mixture.

MII stability measurements were performed using vesicles containing dark-adapted rhodopsin (P/L = 1/4000) that were preincubated with or without G_t for 30 min (reaction solution 1). The assay was initiated by a flash of light (520-nm, 1 cm), which resulted in 65 ± 1% of the rhodopsin being activated (rho*), in agreement with the quantum yield of retinal [7, 8]. At selected time interval ranging between 0.08 and 120 min, 10 uL of the vesicle mixture (0.015 uM total rhodopsin) was added to 30 uL of a solution containing Gt at 0.225 μM in 20 mM 3-(N-morpholino) propanesulfonic acid (MOPS) (pH 7.5), 3 mM MgSO4, 1mM EDTA, 1 mM DTT, 3 mg/mL BSA, 100 mM NaCl, 0.1 % Na-cholate, and [³⁵S]-GTP-γ-S at a radiospecific activity 5000 cpm/pmol (reaction solution 2). Reaction solution 2 was equilibrated at 22° C in a water bath 2 min prior to use. After 1 minute, this secondary reaction was terminated (as described above) and the initial rate of G_t activation as a function of the time after rhodopsin activation was determined. The amount of G_t carried from reaction solution 1 to reaction solution 2 was negligible – between 0.8 and 6% of the total G_t concentration.

SANS. Experiments were conducted at the Bio-SANS beamline (CG-3) [9], located at the Oak Ridge National Laboratory High Flux Isotope Reactor with a neutron wavelength, λ, of 6 Å, having a wavelength distribution, Δλ/λ, of 0.132. SANS data were collected at ambient temperature and acquired over the angular range of 0.003 < q < 1 Å^-1. The magnitude of scattering vector q is defined as q = 4π sin (θ) / λ where θ is the scattering angle.

Sample preparation was carried out in a darkroom. Proteoliposomes were concentrated to a rhodopsin content of ~1 mg/mL using a spin concentrator (Amicon, 100 kDa cutoff, Millipore). The proteoliposome dispersion (1.5 mL) was then dialyzed against 1 L of MOPS buffered solution (Slide-A-Lyzer membrane, 10 kDa cutoff; Pierce, Rockford, IL). The solution was exchanged three times over a 16–24 h period. The proteoliposomes were then dialyzed in 0.1–0.5 mL cassettes (Slide-A-Lyzer, 10 kDa cutoff, Pierce, Rockford, IL) against 100 mL solution of the same composition dissolved with 99.9% D₂O. Samples containing rhodopsin in POPC-d₈₂ were extruded ten times prior to SANS experiments through a 0.1 μm pore size polycarbonate filter (Whatman). Samples were loaded into 280 μL quartz absorption cells (Hellma) with a 1 mm path length. For dark adapted rhodopsin experiments, cells were packed into light-tight pouches made of aluminum foil that were cleaned with acetone before use.

Data reduction was carried out using the Mantid software package [10], while determination of the radius of gyration, protein molecular weight and form factor were conducted using the SasView software suite. Background subtraction was performed using the BioXTAS RAW 2 suite [11] and calculation of SANS intensity from PDB files was carried out with the CRYSON program [12]. The PDB files used were PDB ID: 6OY9 (rhodopsin-Gt complex) and a snapshot extracted from a μsec all-atom simulation of rhodopsin in the MI state performed using 1-stearoyl-2-docohexaenoyl-sn-glycero-3-phosphatidylethanolamine (18:0–22:6 PE) lipids (files provided on demand). POPC-d₈₂ liposomes in 99.9% D₂O were used as the background sample in our proteoliposome experiments. We should note that perdeuterated lipids in 99.9% D₂O and the deuterated proteoliposomes in 99.9% D₂O are not perfectly equivalent. Specifically, the average neutron scattering length density of the proteoliposomes is closer to that of D₂O, resulting in an over subtraction of scattering signal at low q ranges in the processed scattering data (Figure SI1B, C and SI2) limiting the useful q-range at the low end to approximately 0.01 Å⁻¹.

3. Results

3.1. M_I-M_II equilibrium depends on rhodopsin membrane density.

Previous studies by Soubias et al. [13–15] and others [16] have observed a link between rhodopsin density, membrane lipid composition and fraction of M_II in reconstituted membranes. Because those studies were limited to protein/lipid ratios (P/L) of between 1/1000 and 1/70 - the latter corresponding to the physiological ratio observed in rod outer segment (ROS) disks - we felt it necessary to extend those studies to lower protein densities. Dark adapted rhodopsin was reconstituted into PC bilayers at P/L ratios ranging from 1/6,000 – 1/70. The fraction of M_II after photoactivation by a flash of light and the catalytic power (V_max/K_m) of G_t exchange were measured by spectrophotometry and [³⁵S]-GTP-γ-S exchange rate, respectively (Fig. 1A, B). While the amount of M_II was high and steady, ranging between 1/6,000 and 1/300, there was a continuous decrease in M_II content, from 1/300 – 1/70 (Fig. 1A). Similarly, the density of rhodopsin in the vesicles profoundly influenced its catalytic power (Fig. 1B), which notably decreased as the protein/lipid ratio in the reconstituted membranes approached the physiological value observed in ROS disks membranes. This was most likely due to emerging interactions between rhodopsin either through direct protein-protein interactions or - interactions mediated by the superposition of layers of protein-associated lipids [14, 15]. Below 1/300, the fraction of M_II and G_t catalytic power exchange were found to be independent of P/L. We therefore surmised that the oligomeric rhodopsin state at P/L less than 1/300 was homogeneous. As a result, all subsequent SANS experiments were conducted at a rhodopsin/PC molar ratio of 1/360 in order to achieve both a sufficiently high protein concentration and rhodopsin homogeneity.

Fig. 1. Rhodopsin concentration affects MII formation and G_t activation.

(A) Meta-rhodopsin II fraction as a function of lipid/rhodopsin ratio in 18:0–18:1-PC (SOPC) bilayers measured by UV-Vis spectroscopy at 37 °C and pH 7.4. (B) Catalytic power of Gt activation by rhodopsin as a function of lipid/rhodopsin ratio in SOPC bilayers. V_m/K_m is expressed as the fraction of the maximum. To simplify data analysis, a fixed concentration of G_t > K_m was used for all experiments ([Rho]=15 nM, [G_t]=250 nM).

3.2. Rhodopsin oligomeric state by SANS.

The observed intensity from a SANS experiment is sensitive to the square of the difference in NSLD of a protein and that of its surroundings. In the case of a membrane protein, the protein is surrounded by both the membrane and solvent. In order to observe scattering from just the protein, isotopic substitution of deuterium for natural abundance hydrogen may be used to tune the NSLD of the lipid membrane to match that of the solvent and enhance the difference from the protiated protein of interest. This minimizes the contribution of scattering from the liposomal particle and enhances scattering signal from the protein. The overall proton content of the sample is also reduced to minimize the large incoherent neutron scattering that contributes to the signal background and obscures the coherent scattering signal at higher angles of diffraction. This angle is expressed here as the scattering wave vector, q.

Optimal SANS experimental conditions are when a protiated protein is embedded into a perdeuterated lipid matrix, in ~ 100 % deuterated solution (see Fig. 2A, B). Fig. 3A shows a SANS profile (0.01 Å⁻¹ < q < 0.3 Å⁻¹) for dark adapted rhodopsin in a POPC-d82 membrane after subtraction of the residual SANS intensity due to an equivalent POPC-d82 liposome in the same D₂O solution (Fig. SI1B, C). Guinier analysis of the neutron scattering signal determined a protein radius of gyration (R_g) of 21.0 ± 0.3 Å, supporting a monomeric state of rhodopsin. This value is in close agreement with the R_g calculated from the X-ray structure of dark-adapted rhodopsin (PDB ID: 1F88, 21.72 Å) and falls within 10% of the R_g obtained from snapshots extracted from an all-atom molecular dynamics (MD) simulation of rhodopsin embedded in a lipid bilayer [17] (22.2 ± 0.25 Å). The scattering data reveal no detectable structure factor arising from protein-protein interactions, reinforcing the conclusion that rhodopsin remains monomeric within the bilayer at this concentration. Further computational analysis corroborates this finding, with CRYSON-predicted SANS intensity from the PDB structure closely mirroring the experimental SANS intensity. Moreover, form-factor fitting of the SANS data provides the radial distribution function, g(r), of scattering centers. As depicted in Fig. 3B, the g(r) profiles calculated from SANS data and MD simulation snapshots align almost perfectly. Finally, modeling the rhodopsin SANS data as an elliptical cylinder (Fig. 3C, left panel) yields dimensions consistent with a monomeric form of rhodopsin.

Fig. 2. Contrast matching to highlight membrane proteins in a SANS experiment.

(A) Calculated neutron scattering length densities (NSLD) of water, lipid bilayer (volume average of headgroup- and acyl chain layers; Lipid Avg), and protein (MW 40 kDa) at a protein/lipid molar ratio of 1/500 as a function of H₂O/D₂O ratio. The NSLD of the protein and lipid bilayer changes with increasing D₂O due the exchangeable hydrogens on the protein, as well as percent D₂O. (B) NSLDs for lipid and protein as a function of percent D₂O. The lipid and protein become “invisible” to neutrons at 8% and 42% D₂O, respectively. Fully deuterated lipids become almost invisible at 100% D₂0.

Fig. 3. Rhodopsin oligomeric state determined by SANS.

(A) Scattered intensity as a function of scattering wavefactor, q, for dark adapted rhodopsin in perdeuterated 16:0–18:1 PC (POPC-d82) bilayers at 295 K. (B). Experimental and molecular dynamics simulation radial distribution functions (g(r)) for dark adapted rhodopsin in PC/PE bilayers. (C) Superimposition of I (q) data for dark adapted rhodopsin (grey square) and opsin (blue square). SANS data for rhodopsin and opsin were fitted to an elliptical cylinder.

When dark adapted samples were exposed to intense white light for one minute (entirely photobleaching the rhodopsin), the spectra changed noticeably (Fig. 3C, left panel). Analysis of the SANS data indicated the formation of higher-order oligomers of opsin under those conditions (Fig. 3C). This is evident from an increase in R_g extracted from a Guinier analysis to 38.2 +/− 0.5 Å for the opsin sample. Further analysis using an elliptical cylinder model shows a small decrease in the cylinder thickness and a significant increase in both the overall volume of the object and the aspect ratio of the cylinder. The volume is compatible with a row of 4 to 5 opsin molecules (Fig. 3, right panel).

In summary, SANS experiments showed that rhodopsin at low protein density in membranes exists as a monomer before activation and as higher order oligomers after photoactivation as it decays to opsin.

3.3. Rhodopsin-Transducin interaction by SANS.

We constructed a test to probe the stoichiometry for the activated rhodopsin-G_t complex. Key characteristics of the complex were established by Kuhn, Chabre and Hofmann in the 1980s and 1990s using isolated rod segment disk membranes [18–20]. Those pioneering studies showed that: (i) G_t binds preferentially to the M_II photo intermediate and shifts the M_I-M_II equilibrium towards M_II; (ii) binding of G_t to M_II stabilizes a guanine nucleotide free state in the G_t α subunit; and (iii) binding of GTP dissociates this complex. In our test, we took advantage of the stable interaction formed between M_II and G_t, which was also used to determine the structure of the rhodopsin-G protein complex using cryo-EM [21, 22].

Prior to photobleaching by a 520 nm flash of light, vesicles containing dark-adapted rhodopsin (P/L = 1/4000) were preincubated with or without G_t for 30 min. Photobleaching resulted in 65 ± 1% of rhodopsin being activated (rho*), consistent with the retinal quantum yield [23]. The decay of the M_II photo intermediate was followed by measuring the initial rate of G_t activation as a function of the time after rhodopsin activation (Fig. 4A and material and methods for details). Fig. 4A shows the time dependence of the initial rate of G_t activation measured when vesicles were preincubated in the absence of G_t, in the presence of an excess of G_t and at a sub stoichiometric concentration of G_t. Without G_t, M_II decayed rapidly with a half time t_1/2 = 3 min. In contrast, when rhodopsin was activated in excess G_t, the decay of the M_II photo intermediate was much slower (t_1/2 ~ 180 min). This shows that the M_II-G_t nucleotide-free complex is stable, a result that is consistent with previous observations. Note that these results were obtained in the absence of apyrase, which explains the small, but observable decay of M_II over time due to potential trace amounts of nucleotide binding. Interestingly, when rhodopsin was activated in the presence of a sub stoichiometric G_t concentration, we observed two populations having different decay rates. For instance, at a ratio rho*/G_t = 0.25, half of the active rhodopsin decayed quickly with a t_1/2 similar to the t_1/2 decay of rhodopsin without G_t, while the other half decayed with a t_1/2 corresponding to that of rhodopsin stabilized by G_t. Experiments done at rho*/G_t = 0.5 showed that the G_t concentration leading to full stabilization of the active rhodopsin is ~ ½ of the total activated rhodopsin. Assuming random rhodopsin orientation in the vesicles, these results indicate that each rhodopsin molecule binds a single G_t molecule, forming a 1:1 complex with our experimental conditions.

Fig 4. Rhodopsin- G_t complex determined by SANS.

(A) Stoichiometry of the MII-G_t complex. (top) Schematic describing the experimental set up. Equilibration step: 60 nM of rhodopsin was reconstituted in lipid vesicles (Rho: 18:0–18:1 PC = 1:4000) and G_t. Different rhodopsin to G_t ratios were premixed prior to rhodopsin being activated (30 min, Temp. 22^◦C). Flash step: After loading into an automatic pipette dispenser, rhodopsin was photoactivated by a flash of light (520 nm, 1 cm). Initial rate Sampling: 10 ul of the sample was pipetted into 30 uL of G_t ( 225 nM, Buffer ), resulting in a final rhodopsin concentration of 15 nM. The fraction of rhodopsin-catalyzed [³⁵S]-GTPγS binding to G_t was measured at 1 min. (bottom) Decay rates of MII in the absence of G_t (red circle), in the presence of excess G_t (blue square), and a G_t at a MII-Gt ratio of 4 (orange triangle) or 2 (green square). (B) Scattered intensity as a function of q for rhodopsin in a complex with G_t in perdeuterated 16:0–18:1 PC (POPC-d82) bilayers at 295 K (grey square). The data were fitted (red line) using a linear combination of back-calculated SANS curves from rhodopsin (blue) and the rhodopsin-Gt complex (green, PDB ID: 6OY9).

Fig. 4B shows a SANS profile (0.01 Å⁻¹ < q < 0.3 Å⁻¹) for rhodopsin photoactivated in the presence of G_t after subtraction of the residual SANS intensity due to the liposome particles and buffer (Fig. SI1). We used a sub stoichiometric amount of G_t (rho*:G_t 2:0.8) to optimize complex formation while reducing the amount of unbound transducin in the sample, considering the random orientation of insertion for rhodopsin in the membrane. This should result in a 1:1 ratio of monomeric rho* oriented with its C-terminal side to the interior of the proteoliposome and the rho*:G_t . We generated SANS predictions of the rho*:G_t complex using the CRYSON program and the published cryo-EM structure of rho*:G_t (PDB: 6OY9). Using this in combination with the monomeric calculated structure for monomeric rhodopsin allowed us to examine the SANS data in terms of a linear combination of these two contributions to the scattering. The relative intensity of each contribution is proportional to the product of the number of that type of molecule, the square of the volume of that individual scatterer, and the square of the NSLD contrast. Using the computed scattered intensity and molecular volumes provided by the CRYSON for these two structures, we constructed a combined scattering curve. The relative intensity of the two contributions was allowed to vary and was found to replicate the data closely when scaled at a 66.4 to 33.6 stoichiometry of rho* to rho*:G_t. This would be associated with a contribution from 33.6% of the number of rho*:G_t complexes, in good agreement with the expected 60:40 ratio.

4. Discussion

Neutron-based experiments have proven well-suited for the study of membranes and of membrane associated proteins [24]. For example, SANS has provided new insights into the conformational changes adopted by a range of large membrane protein systems [25]. However, smaller membrane proteins, such as rhodopsin, require a different “SANS approach” than that used for the study of larger proteins, because any amount of residual scattering from the lipids can overwhelm the signal from smaller proteins [26]. We used SANS in combination with NSLD contrast matching of the lipid matrix to probe rhodopsin oligomeric states and complexes with its G protein transducin. To increase the SANS signal from a membrane protein, the difference between the protein and lipid NSLDs needs to be maximized. Furthermore, the NSLD of the lipid matrix needs to be matched to the NSLD of the buffer, making only the protein “visible” to neutrons. Since protium is a large incoherent neutron scatterer and adds disproportionately to the background signal, especially at higher scattering angles, the sample’s proton content was also minimized. We conducted experiments on protiated protein embedded into a perdeuterated lipid matrix. The sample was hydrated using 99 % deuterated buffer.

This strategy is similar to the one developed by the Arleth group who pioneered a stealth nanodiscs strategy for small-angle neutron scattering (SANS) of protiated membrane proteins, leveraging selective deuteration to render the lipid carrier system invisible while preserving the neutron contrast of the embedded protein, enabling precise structural analysis [27–30]. Here, however, the use of lipid vesicles allows us to extend the investigation to longer-range protein-protein interactions.

Our SANS experiments revealed that rhodopsin exists predominantly as a monomer within POPC membranes prior to photoactivation, at a P/L ratio lower than 1/360. This is consistent with our prior NMR studies showing monomeric rhodopsin down to a P/L of ~1/250 and oligomerized rhodopsin at lower P/L when placed in almost perfectly hydrophobically matched membranes, such as POPC. The observation that dark adapted rhodopsin is monomeric under these experimental conditions is of functional significance. Measurements of activation rates of the native G protein transducin after photoactivation yielded the highest rates when dark-adapted rhodopsin was monomeric. The data suggest that monomeric M_II is activating G protein with high efficiency, consistent with results obtained for rhodopsin embedded into lipid nanodiscs [31, 32] and in detergents [33]. Given that both MII formation and catalytic activity decline as the protein/lipid ratio approaches that of ROS membranes, this suggests that densely packed rhodopsin in ROS disks may attenuate visual transduction by limiting the population of rhodopsin in the active MII conformation and hindering lateral diffusion of transducin.

The strong agreement between our experimentally obtained SANS data and the calculated SANS profile from the cryo-EM-determined rhodopsin-Gt complex highlights the robustness of our SANS methodology. While SANS does not provide molecular details on the interaction between the two proteins, the agreement between the experimental and calculated SANS profiles not only validates the cryo-EM structure but also reinforces the utility of SANS in probing macromolecular interactions in physiologically relevant lipid environments. However, it is important to note that the non-vectorial insertion of rhodopsin which results in a random orientation further complicated our analysis, suggesting that stealth nanoscale lipid domains (nanodiscs), such as the one developed by the Arleth group [30] could offer a more effective strategy in the future for applying SANS to protein complexes that involve both membrane and cytoplasmic components. In this context, the development of larger stealth nanodiscs would represent a significant advancement, enabling the structural interrogation of membrane-associated protein complexes with molecular weights exceeding 100 kDa using SANS.

Finally, our SANS data clearly shows a shift in rhodopsin’s oligomeric state upon photoactivation and its subsequent decay to opsin, leading to the formation of higher order oligomers. An obvious question is what is triggering this monomeric to oligomeric state transition? Besides its affinity to interact with specific sites on rhodopsin [34, 35], the lipid bilayer in which the protein resides appears to play an important role [36, 37]. For example, MD simulations [38] and experiments by Brown et al. [16, 39] and our group [40], strongly suggest that elastic deformation of the lipid bilayer surrounding rhodopsin not only influences its structure, but also its states of oligomerization. The increased tendency of opsin to oligomerize is likely attributable to the interplay between its inherently more open and flexible conformation and the surrounded lipid matrix. Opsin oligomerization could help safeguard photoreceptor function under intense illumination. In bright light, ROS disks are saturated and cease functioning as photoreceptors, yet must remain viable until ambient light levels fall. One protective mechanism could involve the clustering of bleached rhodopsins, which would facilitate efficient arrestin binding despite substoichiometric arrestin expression. As reported before [41–43], arrestin binding to rhodopsin oligomers could effectively quench excess signaling, thereby safeguarding ROS disks integrity during periods of intense illumination.

We should also point out that the current experimental methodology has certain limitations. The reconstitution procedure used produces rhodopsin with roughly equal amounts of N-, and C-termini of rhodopsin on the surface of liposomes. Therefore, we are not able to exclude that opsin oligomers are not vectorially oriented, a condition that is non-physiological. Furthermore, even after contrast matching, signal intensity at Q < 0.01 Å⁻¹ was dominated by scattering from the perdeuterated lipid matrix. Although the SANS profile is well-fit in the case of protein-free liposomes, we were never able to completely contrast-match the lipid matrix of proteoliposomes samples (Fig. SI1B, C). Overall, however, the ability of SANS to monitor changes in the oligomeric states of GPCRs within a membrane environment, under different functional conditions, highlights SANS’s sensitivity and utility to changes in protein-protein interactions within the lipid bilayer.

5. Conclusion

Our SANS study successfully implemented neutron contrast strategies to elucidate the oligomeric state of the prototypical GPCR, rhodopsin, in a relevant lipid vesicle environment and when complexed with its cognate G-protein transducin. Our data clearly show the protein transitioning to a multimeric state upon photoactivation and decaying to opsin, highlighting the dynamic nature of GPCR self-association and the ability of SANS to track these changes within the membrane. The efficacy of SANS in lipid vesicles provides a valuable tool for future investigations of membrane protein organization and interactions under physiologically relevant conditions. The procedures for conducting SANS experiments on integral membrane-incorporated proteins are applicable to the study of other membrane proteins. Although early experiments suffered from a lack of availability of perdeuterated lipids, this is less so today. We hope to make even better use of SANS to study protein oligomerization in the future, based on the current results, which underscore the complex interplay between GPCRs, their partners, and the lipid environment needed for initiating cellular signaling.

Highlights.

• Small Angle Neutron Scattering showed that dark-adapted rhodopsin is monomeric at low protein-membrane concentrations.

• Opsin forms higher order oligomers in membranes.

• Photoactivation of rhodopsin with its G protein results in a 1:1 complex, consistent with recent cryo-EM data.

• SANS with contrast matching in lipid vesicles reveals GPCR oligomerization in biologically relevant membranes.